Iridium and Rhodium Analogues of the Shilov Cycle Catalyst; and The Investigation and Applications of the Reduction-Coupled Oxo Activation (ROA) Mechanistic Motif towards Alkane Upgrading Citation Fu, Ross (2014) Iridium and Rhodium Analogues of the Shilov Cycle Catalyst; and The Investigation and Applications of the Reduction-Coupled Oxo Activation (ROA) Mechanistic Motif towards Alkane Upgrading. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/WY3F-DZ94. https://resolver.caltech.edu/CaltechTHESIS:05212014-155104252 Abstract This dissertation will cover several disparate topics, with the overarching theme centering on the investigation of organometallic C-H activation and hydrocarbon transformation and upgrading. Chapters 2 and 3 discuss iridium and rhodium analogues of the Shilov cycle catalyst for methane to methanol oxidation, and Chapter 4 on the recently discovered ROA mechanistic motif in catalysts for various alkane partial oxidation reactions. In addition, Chapter 5 discusses the mechanism of nickel pyridine bisoxazoline Negishi catalysts for asymmetric and stereoconvergent C-C coupling, and the appendices discuss smaller projects on rhodium H/D exchange catalysts and DFT method benchmarking. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: Rhodium, iridium, nickel, Shilov Cycle, C-H activation, DFT screening, organometallic compounds, Negishi coupling Degree Grantor: California Institute of Technology Division: Chemistry and Chemical Engineering Major Option: Chemistry Thesis Availability: Public (worldwide access) Research Advisor(s): Goddard, William A., III Thesis Committee: Agapie, Theodor (chair) Bercaw, John E. Miller, Thomas F. Goddard, William A., III Defense Date: 9 May 2014 Record Number: CaltechTHESIS:05212014-155104252 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:05212014-155104252 DOI: 10.7907/WY3F-DZ94 Related URLs: URL URL Type Description http://dx.doi.org/10.1021/om201069k DOI Article adapted for ch. 2 http://dx.doi.org/10.1021/cs5005322 DOI Article 1 adapted for ch. 3 http://dx.doi.org/10.1002/chem.201405460 DOI Article 2 adapted for ch. 3 http://dx.doi.org/10.1039/C4CY00972J DOI Article adapted for Appendix A Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 8255 Collection: CaltechTHESIS Deposited By: Ross Fu Deposited On: 29 May 2014 21:28 Last Modified: 04 Oct 2019 00:04 Thesis Files Preview PDF (Ross Fu Thesis) - Final Version See Usage Policy. 36MB Preview PDF (Ross Fu Thesis TOC and Chapter 1) - Final Version See Usage Policy. 537kB Preview PDF (Ross Fu Thesis Chapter 2) - Final Version See Usage Policy. 9MB Preview PDF (Ross Fu Thesis Chapter 3) - Final Version See Usage Policy. 12MB Preview PDF (Ross Fu Thesis Chapter 4) - Final Version See Usage Policy. 9MB Preview PDF (Ross Fu Thesis Chapter 5) - Final Version See Usage Policy. 3MB Preview PDF (Ross Fu Thesis Appendix A) - Final Version See Usage Policy. 1MB Preview PDF (Ross Fu Thesis Appendix B) - Final Version See Usage Policy. 364kB Repository Staff Only: item control page

Iridium and rhodium analogues of the Shilov cycle catalyst; and the investigation and applications of the Reduction-Coupled Oxo Activation (ROA) mechanistic motif towards alkane upgrading Thesis by Ross Fu In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy California Institute of Technology Pasadena, California 2014 (Defended May 9th, 2014) ii c 2014 Ross Fu All Rights Reserved iii To Annie, Bruce, Mom, Dad, and Harold. iv Acknowledgments There are many people who have helped me throughout the course of my graduate studies, and my education in general. Therefore, I cannot guarantee that this list will be comprehensive. First and foremost I want to thank my lovely wife, Annie Tigranyan, who through good times and bad has been a steady and ongoing source of support, encouragement, and motivation; and my parents, Prof. Xiang-Dong Fu and Dr. Mingjie Jin, who have shaped me to be the person I am and who have been the role models for the person I aspire to be. I also want to mention my son Bruce, who has been a continuing source of inspiration for me to be the best father I can; and Toshi, who has been my faithful and unwavering companion, guarding my side for many hours in the office as I put together this dissertation. Finally, I want to thank my parents and parents-in-law, Ruzanna Parvizi and Ashot Tigranyan, for their invaluable support in helping me raise my young family; and in particular my mother-in-law, who has devoted countless hours and effort taking care of Bruce so that I could concentrate on my work. I am greatly indebted to my advisor, Prof. William Goddard, for accepting me into his research group despite my (at the time) purely experimental background, for his extensive help at the beginning of my computational training, and for his continuing mentorship and scientific insights. I also want to thank my previous advisor at Caltech, Prof. John Bercaw, for his mentorship in synthetic chemistry, for helping me initiate my graduate-level work in chemistry, and for his continued work on my thesis committee. In addition, I want to thank Prof. Theo Agapie for chairing my committee, and Prof. Thomas Miller for serving in my committee. Finally I also want to thank Dr. Jay Labinger and Prof. Robert Grubbs for serving in my previous committee during my candidacy exam. In the Goddard group, a.k.a. the Materials Simulation Center, I want to especially thank the director for catalysis, Dr. Robert “Smith” Nielsen, for his collaboration, teaching, advice, patience, and incredible depth of knowledge. I also want to thank my various collaborators, including Prof. Brent Gunnoe, Dr. George Fortman, Dr. Matthew O’Reilly, and Michael Webster-Gardiner at the University of Virginia, and Prof. Greg Fu, Dr. Nathan Schley, and Trixia Buscagan at Caltech. In addition, I want to thank many other group members for their advice, discussions, and friendship; including but not limited to Dr. Mu-Jeng Cheng, Dr. Wei-Guang Liu, Dr. José Mendoza-Cortés, Dr. Julius Su, Samantha Johnson, Yanchoi Lam, Fan Liu, and Caitlin Scott. v In the Bercaw group, I want to thank Dr. Jay Labinger for his invaluable advice in the preparation of my Organometallics publication. I also want to thank the help and advice of my fellow students and postdocs Dr. Steve Baldwin, Dr. George Chen, Prof. Nilay Hazari, Dr. Rachel Klet, Dr. Valerie Kristof, Dr. Taylor Lenton, Prof. Alex Miller, Prof. Ian Tonks, Prof. Dave Weinberg, Prof. Nathan West, Dr. Matt Winston, Yanchoi Lam, Josef Meier, and Tonia Ahmed. I also want to thank Larry Henling and Dr. Michael Day for help in X-ray crystallography, and Naseem Tourian and Dr. Mona Shahgholi for their help in mass spectrometry. Finally, I want to thank the many friends who have made my years here at Pasadena so enjoyable. In particular, Elena Bibikova, Jerry Koo and Fiona Chiang, Tracey Liu, Hemanth Siriki, Jerzy Szablowski, Chloe Wang, Arlene Santos and the Salsa Club and of course my dance partner Annie Tigranyan; and some of my oldest friends from high school, Michael Liao, Jason Lin, Erik Tsou, and David Wang. vi Abstract This dissertation will cover several disparate topics, with the overarching theme centering on the investigation of organometallic C−H activation and hydrocarbon transformation and upgrading. Chapters 2 and 3 discuss iridium and rhodium analogues of the Shilov cycle catalyst for methane to methanol oxidation, and Chapter 4 on the recently discovered ROA mechanistic motif in catalysts for various alkane partial oxidation reactions. In addition, Chapter 5 discusses the mechanism of nickel pyridine bisoxazoline Negishi catalysts for asymmetric and stereoconvergent C−C coupling, and the appendices discuss smaller projects on rhodium H/D exchange catalysts and DFT method benchmarking. Chapter 2: The iridium complex (ONO)Ir(PPh3 )2 Me, where (ONO) is a bis(phenolate)pyridine pincer ligand, undergoes C−H activation of benzene, related arenes, and acetonitrile. Labeling and kinetic studies indicate a unique and heretofore undescribed mechanism involve intramolecular C−H activation followed by intermolecular C−H activation. Chapter 3: The Shilov catalytic cycle for methane to methanol oxidation via a homogeneous Pt catalyst has long been a target of investigation, with work focusing on improving catalyst stability, yield, and robustness. A Rh analogue to the Shilov cycle was explored due to its increased stability at high oxidation states. A density functional theory-driven virtual screening of ligand F frameworks revealed that the fluorine-substituted amidinate (NN ) ligand provides the lowest transition state energies for methane C−H activation and methyl group functionalization, and stands most promising as a target for experimental investigation. Chapter 4: The vanadium phosphorus oxide (VPO) catalyst has been used in industry for the remarkably selective conversion of butane to maleic anhydride. Investigation of the reaction mechanism revealed the surprising conclusion that, although redox action occurs in the vanadium centers as expected, initial C−H activation takes place in the phosphorus oxide moieties. Further work has revealed that the same phenomenon can be extended to other transition metal/nonmetal oxo pairs. This phenomenon, which has been named the Reduction-Coupled Oxo Activation (ROA) mechanistic motif, is attributed to the concomitant formation of a strong oxo-hydrogen bond on the nonmetal and a one-electron reduction on the transition vii metal. Finally, a homogeneous vanadium phosphorus oxo complex is proposed to incorporate the novel ROA mechanistic pathway as a potential propane oxidation catalyst. Several oxidation pathways are proposed, leading to end products such as propylene, isopropanol, and propylene oxide. These assertions are supported with density functional theory calculations on the potential reaction pathways. Chapter 5: The in situ generated Ni(iPr−pybox) complex catalyzes enantioselectively the crosscoupling of secondary sp3 -C substrates. This system is very notable for its property of stereoconvergence: both enantiomers of a racemic substrate may be converted into a single enantiomer of a product. This chapter will computationally explore the mechanism responsible for both the remarkable activity and selectivity of the Ni((S, S)−iPr−pybox system; such a mechanism is believed to deviate substantially from the canonical “textbook” mechanism of Negishi coupling. Furthermore, a generalized calculation scheme is presented that allows for the rapid enantioselectivity prediction of many related pybox ligands as well, in order to help predict the next generation of asymmetric Negishi coupling catalysts. viii Contents Acknowledgments iv Abstract vi 1 Introduction 1 1.1 Iridium and Rhodium analogues of the Shilov cycle catalyst . . . . . . . . . . . . . . 1.2 Investigation and applications of the Reduction-Coupled Oxo Activation (ROA) mech- 1 anistic motif towards alkane upgrading. . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Other projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.4 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2 Intra- and intermolecular C−H bond activation by bis(phenolate)pyridineiridium(III) complexes 8 2.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Synthesis and Characterization of IrI complexes . . . . . . . . . . . . . . . . . 9 2.3.1 III 2.3.2 Synthesis and Characterization of Ir Complexes . . . . . . . . . . . . . . . 13 2.3.3 C−H Activation by IrIII (ONO ) Complexes . . . . . . . . . . . . . . . . . . 21 2.3.4 Kinetics of C−H Activation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 2.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.5 Supporting Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.5.1 Experimental Section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.5.2 Kinetics Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.5.3 Crystallographic Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.6 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 tBu ix 3 DFT virtual screening identifies rhodium-amidinate complexes as active homogeneous catalysts for methane to methanol oxidation 56 3.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 3.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 3.4.1 63 3.4.2 The Rh(NN) complexes in TFAH . . . . . . . . . . . . . . . . . . . . . . . . . F 67 F The Rh(NN ) complexes in TFAH . . . . . . . . . . . . . . . . . . . . . . . . 3.4.3 The Rh(NN ) complexes in water . . . . . . . . . . . . . . . . . . . . . . . . 67 3.4.4 The Rh(ONNx ) family of complexes in TFAH . . . . . . . . . . . . . . . . . . 73 3.4.5 The Rh(DPMS) and Rh(DPES) complexes in TFAH . . . . . . . . . . . . . . 78 3.4.6 The Rh(PN) complexes in TFAH . . . . . . . . . . . . . . . . . . . . . . . . . 78 x 3.4.7 The Rh(bisq ) family of complexes in TFAH . . . . . . . . . . . . . . . . . . 81 3.4.8 Product protection and C−H activation . . . . . . . . . . . . . . . . . . . . . 92 3.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 3.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 3.7 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 3.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4 The Reduction-Coupled Oxo Activation mechanistic motif for C−H bond activation and oxidation 110 4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 4.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 4.2.1 The VPO catalyst for n-butane oxidation . . . . . . . . . . . . . . . . . . . . 111 4.2.2 The ROA mechanistic motif . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 4.2.3 Potential homogeneous catalysts utilizing the ROA principle . . . . . . . . . 115 4.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 4.4 Validation of ROA on small oxide clusters . . . . . . . . . . . . . . . . . . . . . . . . 118 4.5 Reactions of a homogeneous oxidation catalyst utilizing the ROA mechanistic motif 128 4.5.1 Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 4.5.2 The resting VV state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 4.5.3 Propane activation transition states . . . . . . . . . . . . . . . . . . . . . . . 132 4.5.4 Monohydrogenated VIV states . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 4.5.5 Catalytic cycle involving H-atom abstraction only . . . . . . . . . . . . . . . 136 4.5.6 Reduced VIII states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 4.5.7 Catalytic cycle involving H-atom abstraction followed by isopropyl addition . 143 x 4.5.8 Comparison of the catalytic cycles . . . . . . . . . . . . . . . . . . . . . . . . 144 4.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 4.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 5 The mechanism of Ni(iPr−pybox) complexes as asymmetric Negishi C−C coupling catalysts 150 5.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 5.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 5.2.1 The Ni(iPr−pybox) system . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 5.2.2 Existing computational work . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 5.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 5.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 5.5 5.4.1 The overall general mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . 157 5.4.2 Nature of the organozinc reagent . . . . . . . . . . . . . . . . . . . . . . . . . 158 5.4.3 Mechanism for NiI oxidation and methylation . . . . . . . . . . . . . . . . . . 161 5.4.4 Mechanism for reductive elimination . . . . . . . . . . . . . . . . . . . . . . . 166 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 5.5.1 Overall catalytic cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 5.5.2 Predictions for new ligands . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 5.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 5.7 Supporting figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 5.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 A C−H activation using RhI complexes supported by bis-nitrogen chelating ligands180 A.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 A.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 A.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 A.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 A.4.1 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 A.4.2 Computational results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 A.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 A.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 B Energy levels of group 10 transition metal atoms and ions 189 B.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 B.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 B.3 The neutral atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 xi B.3.1 Platinum (0) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 B.3.2 Palladium (0) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 B.3.3 Nickel (0) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 B.4 The singly-charged cation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 B.5 Angular momentum-dependent artifacts in DFT calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 B.5.1 Angular momentum of the electron/hole in the d1 /d9 configuration of the cation196 B.5.2 Angular momentum of the holes in the d8 configuration of the dication . . . . 199 B.6 Energies of d2 and d8 triplet states . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 B.6.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 B.6.2 The d2 and d8 triplet states . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 B.6.3 Pure and mixed complex orbital configurations . . . . . . . . . . . . . . . . . 204 B.6.4 Real orbital configuration energies . . . . . . . . . . . . . . . . . . . . . . . . 205 B.6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 B.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 B.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 xii List of Figures 1.1 C−H activation by a phosphorus oxo leads to a one electron reduction on the neighboring vanadium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.1 The ONO and ONOtBu ligands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2 Crystal structure of pyridine-activated complex 2-3. . . . . . . . . . . . . . . . . . . . 11 2.3 2.4 2.5 2.6 2.7 tBu Crystal structure of [NaIr(cod)(ONO tBu Crystal structure of [Ir(cod)]2 (ONO )]2 , 2-4. . . . . . . . . . . . . . . . . . . . . . 13 ), 2-5. . . . . . . . . . . . . . . . . . . . . . . . 14 Crystal structure of [(ONO tBu )Ir(cod)(MeCN)][PF6 ], 2-6. . . . . . . . . . . . . . . . . 14 Crystal structure of [(ONO tBu )Ir(cod)OTf, 2-7. . . . . . . . . . . . . . . . . . . . . . . 15 tBu )Ir(PPh3 )2 OTf, 2-8. . . . . . . . . . . . . . . . . . 16 tBu )Ir(PPh3 )2 OH, 2-9. . . . . . . . . . . . . . . . . . 17 Crystal structure of trans-(ONO 2.8 Crystal structure of trans-(ONO 2.9 Crystal structure of (ONO tBu )Ir(PPh3 )2 H, 2-11. . . . . . . . . . . . . . . . . . . . . . 17 2.10 Crystal structure of (ONO tBu )Ir(PPh3 )2 Me, 2-12. . . . . . . . . . . . . . . . . . . . . 19 2.11 Crystal structure of (ONO tBu )Ir(PPh3 )2 I, 2-13. . . . . . . . . . . . . . . . . . . . . . 19 2.12 Crystal structure of (ONO tBu )Ir(PEt3 )2 Me, 2-14. . . . . . . . . . . . . . . . . . . . . 20 2.13 Crystal structure of (ONO tBu )Ir(PEt3 )2 I, 2-15. . . . . . . . . . . . . . . . . . . . . . . 21 2.14 Cyclic voltammetry of 2-14. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.15 EPR spectrum of [(ONO )Ir(PEt3 )2 Me]PF6 , 2-16. . . . . . . . . . . . . . . . . . . . 22 2.16 Crystal structure of [(ONO tBu )Ir(PEt3 )2 Me]PF6 , 2-16. . . . . . . . . . . . . . . . . . 23 2.17 Crystal structure of (ONO tBu )Ir(PPh3 )2 Ph, 2-17. . . . . . . . . . . . . . . . . . . . . 24 2.18 Crystal structure of (ONO tBu 2.19 Crystal structure of [(ONO )Ir(PPh3 )(µ−CH2 CN)]2 , 2-19. . . . . . . . . . . . . . . 26 2.20 Kinetics of conversion of 2-12 to 2-17 and their deuterated isotopologues. . . . . . . . 30 2.21 Dependence of the reciprocal of the rate of conversion of 2-12 to 2-17 on the amount tBu 2 )Ir(PPh3 )[κ -PPh2 (o-C6 H4 )], 2-18. . . . . . . . . . . . . tBu 24 of added triphenylphosphine. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.22 Kinetics for conversion of 2-17 to 2-18 in C6 D6 and p-xylene-d10 . . . . . . . . . . . . 31 2.23 Conversion of 2-12 to 2-17-d6 in C6 D6 with varying amounts of extra PPh3 added. . 47 3.1 The rhodium-ligand complexes screened. . . . . . . . . . . . . . . . . . . . . . . . . . . 63 xiii 3.2 Rh(NN) transition state structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Rh(NN ) transition state structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 3.5 F F Rh(NN ) in water transition state structures. . . . . . . . . . . . . . . . . . . . . . . . 87 88 The various Rh (bisq )(Me) III-II SR 2 functionalization transition states. . . . . . . 3.7 Graph of the activation and functionalization transition state energies of the various Rh(bisqx ) complexes at 298 K and 498 K. . . . . . . . . . . . . . . . . . . . . . . . . . 91 Transition states for the functionalization of methane, Me−TFA, and methanol using Rh(NN) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9 72 x 3.6 3.8 69 x The various RhIII (bisq )(Me) III-I SN 2 functionalization transition states. . . . . . . . III 66 93 Transition states for the functionalization of methane, Me−TFA, and methanol using F Rh(NN ) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 3.10 Summary of the data presented in Figures 3.8, 3.9, and 3.11. . . . . . . . . . . . . . . 95 3.11 Comparison of activation energies for methane and for methanol in the same structure F for Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12 Comparison of the activation states of methane and methyl trifluoroacetate by Rh(DPMS) complexes, showing the product protection afforded. . . . . . . . . . . . . . . 3.13 98 Comparison of the activation states of methane and methyl trifluoroacetate by the x Rh(bisq ) family of complexes, showing the product protection afforded. . . . . . . . . 3.15 97 Comparison of the activation states of methane and methyl trifluoroacetate by Rh(DPES) complexes, showing the product protection afforded. . . . . . . . . . . . . . . . . . . . 3.14 96 99 Graph of the methane activation energy of the various rhodium complexes vs. the 4s orbital energy of the rhodium in their resting states. . . . . . . . . . . . . . . . . . . . 101 3.16 Graph of the methane activation energy of the various rhodium complexes vs. the Rh−C bond energy of the lowest-energy Rh−Me complex. . . . . . . . . . . . . . . . . 103 3.17 A graph of the Rh−TFA to Rh−Me change in the 4s orbital energy, a measure of the Rh atom’s softness, versus the adjusted TS barrier. . . . . . . . . . . . . . . . . . . . . 104 3.18 The calculated free energy of the (NNC)Ir(TFA)2 system, via the same internal substitution mechanism applied to RhIII complexes. . . . . . . . . . . . . . . . . . . . . . 105 IV 4.1 O−H bond strength for oxygen atoms on the (V O)2 (P2 O7 ) surface. . . . . . . . . . 111 4.2 Oxygen adsorption energies on the (V O)2 (P2 O7 ) surface and the O−H bond strength IV (DH ) to each of these O atoms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 4.3 O−H bond strengths for the various oxygen atoms on the X1-VV OPO4 surface. 4.4 C−H activation by a phosphorus oxo leads to a one electron reduction on the neigh- . . . 113 boring vanadium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 xiv 4.5 Use of finite cluster models to investigate how the O−H bond strength of P−O moieties varies based on the number of V coupled to it via O linkages. . . . . . . . . . . . . . . 114 4.6 Known complexes featuring a phosphinite ligand bound κ−P to a transition metal, and a proposed vanadium κ−P phosphinite complex. . . . . . . . . . . . . . . . . . . . . . 116 4.7 General models for small oxide clusters. . . . . . . . . . . . . . . . . . . . . . . . . . . 118 4.8 Optimized structures of the Px V4−x O10 species (x ∈ {0, 1, 2, 3}) and their monohydrogenated equivalents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 4.9 Optimized structures of the dihydrogenated P2 V2 O10 H2 and trihydrogenated P2 V2 O10 H3 species. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 4.10 Optimized structures of the XV3 O10 species (X ∈ {P, As, Sb, Bi}) and their monohydrogenated XV3 O10 H counterparts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 4.11 Optimized structures of the Z2 V2 O11 species (Z ∈ {S, Se, Te} and their monohydrogenated Z2 V2 O11 H counterparts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 4.12 Optimized structures of the hydrated species Te2 V2 O12 H2 and Te2 V2 O12 H3 . . . . . . 122 4.13 Optimized structures of the fully hydrated species Te2 V2 O14 H6 and Te2 V2 O14 H7 . . . 123 4.14 Optimized structures of Se2 V2 O9 and Te2 V2 O9 , as well as their monohydrogenated counterparts Se2 V2 O9 H and Te2 V2 O9 H. . . . . . . . . . . . . . . . . . . . . . . . . . . 124 4.15 Optimized structures of the group 14-vanadium oxide species Z2 V2 O10 H2 , where Z ∈ {Si, Ge, Sn, Pb}; as well as that of their corresponding monohydrogenated species. . . 126 4.16 Optimized structures of the molybdenum-containing species P2 Mo2 O11 , Te2 Mo2 O10 , and Te2 Mo2 O12 ; as well as that of their corresponding monohydrogenated species. . . 127 4.17 Template for the nomenclature of investigated (OPO)V complexes. . . . . . . . . . . . 129 4.18 The geometry-optimized structures of (OPO)fH V Oc (OH)s (H2 O)t , (OPO)f V Oc (H2 O)st 2 , V V V 2– and [(OPO)f V Oc (OH)st 2 ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 V 4.19 ‡ The geometry-optimized structure of [(OPO)fHiPr V Oc (H2 O)st 2 ] . . . . . . . . . . . . . 135 4.20 f c st The geometry-optimized structures of (OPO)fH V Oc (H2 O)st 2 , (OPO) V (OH) (H2 O)2 , IV IV IV – and [(OPO)f V Os (H2 O)ct 2 ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 4.21 III III 3 f cst The geometry-optimized structures of 3 (OPO)f V (HOiPr)s (H2 O)ct 2 , (OPO) V (H2 O)3 , and 3 (OPO)H VIII (OH)s (H2 O)a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 5.1 The computationally studied reaction of the Ni(iPr−pybox) Negishi coupling catalyst as reported by Lin et al. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 5.2 The various (MeZnI)m (dma)n species, where m, n ∈ {0, 1, 2}. . . . . . . . . . . . . . . 159 5.3 The various (ZnBrI)m (dma)n species, where m, n ∈ {0, 1, 2}. . . . . . . . . . . . . . . 160 5.4 Transition states for the reductive elimination of 1-methylindane from various NiIII ((S, S)−iPr−pybox)(Me)ind species. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 xv 5.5 Reaction coordinate diagram of the entire catalytic cycle for the coupling of ind−Br and MeZnI to form ind−Me. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 5.6 Modified ligands which may allow for different enantioselective properties as compared to the original (iPr−pybox) ligand. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 A.1 Temperature dependence vs. turnover numbers for complexes 1 and 2. . . . . . . . . . 182 A.2 Concentration of Rh dependence vs. TON for complexes 1 and 2. . . . . . . . . . . . 183 xvi List of Schemes 1.1 The original PtII /PtIV cycle of the Shilov catalyst, as well as IrIII /IrV and RhI /RhIII analogues. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 The oxidation of n-butane to maleic anhydride using vanadium phosphorus oxide as a catalyst. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 2.1 2.2 2.3 2 3 The generalized mechanism of the Ni((iPr)−pybox) Negishi coupling catalyst as it proceeds through a NiI -NiIII cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Synthesis of Na[Ir(cod)(ONO)], 2-1. . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 tBu Synthesis of the ligand H2 (ONO ). . . . . . . . . . . . . . . . . . . . . . . . . . . tBu Synthesis and reactivity of [NaIr(cod)(ONO )]2 , 2-4. . . . . . . . . . . . . . . . . . 12 III species. . . . . . . . . . . . . . . . . 13 )Ir(PPh3 )2 H, 2-11. . . . . . . . . . . . . . . . . . . . . 18 2.4 Oxidation of [NaIr(cod)(ONO 2.5 Synthetic routes to (ONO 2.6 tBu Methylation to (ONO I2 to (ONO tBu tBu tBu )]2 , 2-4, to Ir )Ir(PPh3 )2 Me, 2-12, and subsequent functionalization with )Ir(PPh3 )2 I, 2-13. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . tBu 2.7 Synthesis of (ONO 2.8 Proposed mechanism for the transformation of 2-12 in benzene to the kinetic product tBu (ONO 12 Ir(PEt3 )2 complexes. . . . . . . . . . . . . . . . . . . . . . . . tBu )Ir(PPh3 )2 Ph, 2-17 and the thermodynamic product (ONO 18 20 )Ir(PPh3 )- [κ2 -PPh2 (o-C6 H4 )], 2-18. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.9 Simplified version of Scheme 2.8, used for kinetic analysis. . . . . . . . . . . . . . . 29 3.1 Hypothetical catalytic cycle for the activation and functionalization of methane. . . 62 3.2 Catalytic cycle for the activation and functionalization of methane using Rh(NN) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 F Catalytic cycle for the activation and functionalization of methane using Rh(NN ) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 68 F Catalytic cycle for the activation and functionalization of methane using Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 64 F Transition states for the activation of methane using Rh(NN ) complexes in water. . 70 71 xvii 3.6 F Transition states for the III-I functionalization pathway using Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 73 F Transition states for the III-II functionalization pathway using Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8 74 F Transition states for the III-IV-II functionalization pathway using Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9 74 Catalytic cycle for the activation and functionalization of methane using Rh(ONN) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 F 3.10 Catalytic cycle for the activation and functionalization of methane using Rh(ONN ) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11 Catalytic cycle for the activation and functionalization of methane using Rh(ONN NMe2 76 ) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.12 Catalytic cycle for the activation and functionalization of methane using Rh(DPMS) complexes in TFAH, via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. . . . 79 3.13 Catalytic cycle for the activation and functionalization of methane using Rh(DPES) complexes in TFAH, via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. . . . 80 3.14 Catalytic cycle for the activation and functionalization of methane using Rh(DPMS) complexes in TFAH, via the Rh(I-III) pathway. . . . . . . . . . . . . . . . . . . . . . 81 3.15 Catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, via the Rh(III-I) and Rh(III-II) pathways. . . . . . . . . . . . . 82 3.16 Catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, via the Rh(I-III) pathway. . . . . . . . . . . . . . . . . . . . . . x III 83 x 3.17 The inorganic RhIII (bisq )(TFA)3 species and its protonated analogues [Rh (bisq )+ III x + (TFA)2 (TFAH)] and [Rh (bisq )(TFA)2 )] . . . . . . . . . . . . . . . . . . . . . . III x +‡ 3.18 The various methane activation transition states [Rh (bisq )(TFA)(Me··H··TFA)] III 84 . 85 x 3.19 Comparison of the functionalization of Rh (bisq )(Meax )(TFA)2 via the III-I SN 2 and III-II SR 2 pathways and the completion of the catalytic cycle. . . . . . . . . . . 89 x 3.20 Catalytic cycle for the activation and functionalization of methane using Rh(bisq ) complexes in TFAH, via the Rh(I-III) pathway. . . . . . . . . . . . . . . . . . . . . . 90 3.21 A schematic diagram showing the metal activation of an R−H bond, for example, methane. 4.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 The oxidation of n-butane to maleic anhydride using vanadium phosphorus oxide as a catalyst. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 4.2 Potential catalytic cycle involving H-atom abstraction only. . . . . . . . . . . . . . . 138 4.3 Potential catalytic cycle assuming isopropyl addition and VIII formation. . . . . . . . 145 xviii 5.1 The generalized cross-coupling reaction and its mechanism. . . . . . . . . . . . . . . 151 5.2 The general reaction of the Ni(iPr−pybox) Negishi coupling catalyst. . . . . . . . . . 152 5.3 The generalized mechanism of the Ni(iPr−pybox) Negishi coupling catalyst as it proceeds through a NiI -NiIII cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 5.4 Complexation of the various MeZnX and ZnX2 species with X– , where X ∈ {Br, I}. . 159 5.5 Three potential pathways for the methylation and oxidation of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br. . . . . . . . . . . . . . . . . . . . . . . . . . 162 5.6 The dissociative pathway for the methylation of NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)−iPr−pybox)Me. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162 5.7 NiI ((S, S)−iPr−pybox)Br by ind−Br to NiII ((S, S)−iPr−pybox)Br2 and ind · . . . . 163 5.8 NiII ((S, S)−iPr−pybox)Br2 and its subsequent methylation or recombination with ind · as part of the NiII or NiIII pathways. . . . . . . . . . . . . . . . . . . . . . . . . 165 5.9 The methylation of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) to produce NiIII ((S, S)−iPrpybox)(Br)(ind)Me. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 5.10 A generalized catalytic mechanism that Ni(R−pybox) complexes undergo in Negishi coupling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 5.11 Expanded diagram of NiII ((S, S)−iPr−pybox)Br2 methylation as part of the NiII pathway. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 5.12 Expanded diagram of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) methylation as part of the NiIII pathway. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 5.13 The reactions of the various NiIII ((S, S)−iPr−pybox)(Me)ind species that lead to the reductive elimination of ind−Me. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 F A.1 Synthesis of rhodium H/D exchange catalysts (DAB )Rh(COE)(TFA) A-1 and (BOZO)Rh(COE)(TFA) A-2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 F A.2 Oxidative addition of benzene by Rh(DAB ) and Rh(BOZO) complexes A-3 and A-4. 184 F A.3 Various possible pathways for the H/D exchange of benzene catalyzed by a Rh(DAB ) complex. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 A.4 Various possible pathways for the H/D exchange of benzene catalyzed by a Rh(BOZO) complex. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186 F A.5 Various possible pathways for the H/D exchange of methane catalyzed by a Rh(DAB ) complex. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 xix List of Tables 2.1 Summary of the results of Figure 2.23, used to create Figure 2.21. . . . . . . . . . . . 46 2.2 Crystallographic data for the complexes 2-4, 2-5, 2-6, and 2-7. . . . . . . . . . . . . . 48 2.3 Crystallographic data for the complexes 2-8, 2-9, 2-10, and 2-11. . . . . . . . . . . . . 49 2.4 Crystallographic data for the complexes 2-12, 2-13, 2-14, and 2-15. . . . . . . . . . . 50 2.5 Crystallographic data for the complexes 2-16, 2-17, 2-18, and 2-19. . . . . . . . . . . 51 tBu 2.6 Crystallographic data for the complex (ONO )Ir(PEt3 )2 Cl. . . . . . . . . . . . . . . 52 3.1 Antoine equation parameters for trifluoroacetic acid and water. . . . . . . . . . . . . . 60 3.2 Lowest activation and functionalization energies for each series of Rh-ligand complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 65 Comparison of the various Rh parameters with the rhodium-ligands complexes’ associated transition state barriers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 V 4.1 Comprehensive compilation of all (OPO)V species studied. . . . . . . . . . . . . . . 132 4.2 Comprehensive compilation of all the transition states for propane activation by (OPO)VV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 IV species studied. . . . . . . . . . . . . . . 137 IV species studied. . . . . . . . . . . . . . . 142 IV species studied. . . . . . . . . . . . . . . 143 4.3 Comprehensive compilation of all (OPO)V 4.4 Comprehensive compilation of all (OPO)V 4.5 Comprehensive compilation of all (OPO)V 5.1 Examples of various C−C cross-coupling reactions. . . . . . . . . . . . . . . . . . . . . 152 5.2 Some examples of substrates coupled by the Ni(iPr−pybox) catalyst. . . . . . . . . . 153 5.3 Comparison of activation energies for various Ni(R−pybox) catalysts. . . . . . . . . . 171 5.4 Relative energies of the neutral A, D, and G species for various Ni(R−pybox) complexes.176 5.5 Relative energies of the cationic B, E, and H species for various Ni(R−pybox) complexes.176 5.6 Relative energies of the cationic C, F, and J species for various Ni(R−pybox) complexes.177 B.1 Relativistic vs. nonrelativistic Hartree-Fock excitation energies . . . . . . . . . . . . . 190 B.2 Relativistic vs. nonrelativistic Hartree-Fock excitation energies for the M+ ions. B.3 Excitation energies of Pt atom as calculated by various DFTs. . . . . . . . . . . . . . 191 . . . 191 xx B.4 Excitation energies of Pt atom calculated without f functions. . . . . . . . . . . . . . 192 B.5 Excitation energies of Pt atom calculated using electron-correlation methods. . . . . . 193 B.6 d10 − d9 s1 Excitation energies of Pd atom as calculated by Jaguar. . . . . . . . . . . . 194 B.7 Excitation energies of Pd atom as calculated by Q-Chem. . . . . . . . . . . . . . . . . 194 B.8 Excitation energies of Ni atom as calculated by Q-Chem; electron-correlation methods. 194 B.9 Excitation energies of Ni atom as calculated by Q-Chem; DFT methods. . . . . . . . . 195 B.10 Excitation energies of the group 10 atoms from d9 to d8 s1 , as calculated using various DFTs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 B.11 Excitation energies of the group 10 atoms from d9 to d8 s1 , as calculated using various electron-correlation methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 B.12 “Degenerate” energies of the Pt+ d9 ground state. . . . . . . . . . . . . . . . . . . . . 197 B.13 “Degenerate” energies of the Pt9+ d1 ground state. . . . . . . . . . . . . . . . . . . . . 197 B.14 “Degenerate” energies of the Ni+ d9 ground state as calculated by DFT methods. . . 198 B.15 Degenerate energies of the Ni+ d9 ground state as calculated by electron-correlation methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198 B.16 Energies of the Pt2+ d8 configuration, calculated by B3LYP/LACV3P**++. . . . . . 199 B.17 Energies of the Pd2+ d8 configuration, calculated by B3LYP/LACVP. . . . . . . . . . 200 B.18 Energies of the Ni2+ d8 configuration, calculated on Jaguar using the 6-31G(tm)**++ basis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 B.19 Energies of the Ni2+ d8 configuration, experimental and calculated on Q-Chem using the G3LARGE basis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 1 Chapter 1 Introduction 1.1 Iridium and Rhodium analogues of the Shilov cycle catalyst The facile, selective, and direct conversion of methane into methanol has long been a goal of industrial chemists [1]. Methane, the chief component of natural gas, is typically found in abundant quantities wherever petroleum deposits are found. However, due to its difficulty in liquefaction (its boiling point being only 110 K) and its relative paucity of demand as compared to other hydrocarbons, it is often not economical to harvest this gas, which is essentially a byproduct or co-product of petroleum recovery, and to transport it to market. Since methane is also a potent greenhouse gas, with a global warming potential 72 times more powerful than that of carbon dioxide [2], it cannot simply be released either. Hence it is often flared, a process that is both intrinsically wasteful and which results in a large amount of heat pollution. One potential solution is to convert these vast quantities of methane into methanol. As methanol is liquid at room temperature and pressure, it is much more easily transported and stored. Methanol can be used as a fuel in flex fuel vehicles, blended with gasoline, or converted to gasoline or diesel fuel. In addition, it has many applications as an industrial solvent and chemical feedstock. As methanol is both more easily processed and in greater demand than methane, the efficient conversion from the former to the latter has the potential to be both environmentally friendly and economically favorable. Although the reaction CH4 + 21 O2 −−→ CH3 OH is exoenergetic by 30 kcal/mol, efficient catalysis is hampered by the fact that the C−H bond dissociation energy (BDE) of methane is 104 kcal/mol, whereas the C−H BDE of methanol is only 95 kcal/mol. Hence, overoxidation is a major problem. In fact, potential methods that involve radicals have a theoretical maximum yield of only 5%, even with perfect conversion [3]. Hence, the currently favored industry-scale conversion of methane to methanol involves an indirect pathway via syngas [1]: 2 Ni CH4 + H2 O −→ CO + 3 H2 (1.1) CO + 2 H2 −−→ CH3 OH (1.2) However, this pathway suffers from a difficult first step that is highly endergonic, requiring very high temperatures and pressures (700-1100◦ C, 40 atm). Hence, this state-of-the-art pathway requires large inputs of energy and very expensive production facilities despite the significantly exothermic nature of the overall reaction. An alternative mechanism that does not require such extreme conditions would thus be highly desirable. Molecular compounds in solution have long been investigated as potential catalysts for this transformation, as they are typically well-defined, easy to characterize and model, and operate under relatively mild conditions. Much of the focus has been on routes involving C−H activation at a transition metal center followed by oxidative cleavage of the resulting C−M bond, which offer greater possibilities for good selectivity by avoiding radical-based pathways. Platinum complexes 2– have received a great deal of attention [4], starting with Shilov’s seminal work on the PtCl2– 4 /PtCl6 system, which can oxidize methane to methanol with some selectivity [5], and including the platinum bipyrimidine system, which has achieved the most impressive performance to date, converting methane to methyl bisulfate in up to 70% yield [6]. However, none of this platinum-based chemistry has yet been shown to lead to a practical process, owing in part to low reactivity, and a good deal of research activity has turned to other metals. Chapters 2 and 3 will address certain Ir and Rh analogues, respectively. The original Pt Shilov cycle, as well as modifications incorporating Ir and Rh, is given in Scheme 1.1. Scheme 1.1. Left: The original PtII /PtIV cycle of the Shilov catalyst. Center: A potential analogous IrIII /IrV cycle. Right: A potential analogous RhI /RhIII cycle. Chapter 2 focuses on bis(phenolate)pyridineiridium(III) ((ONO)Ir) complexes as an Ir analogue to the Shilov catalyst. Iridium as an alkane oxidation catalyst has been a popular choice; many 3 (primarily low-valent) Ir complexes have been reported to show good activity for C−H activation [7], and examples of oxidative functionalization with Ir systems are known as well [8]. Interconversions between IrIII and IrV might be important in this chemistry, by analogy to the Shilov system, which involves PtII and PtIV . In Chapter 2, the experimental synthesis and characterization of a multitude of ((ONO)Ir) complexes in the +III and +IV oxidation state are described. In particular, the complex (ONO)Ir(PPh3 )2 Me has been found to undergo oxidation with iodine to release MeI, analogous to the Shilov system’s production of MeCl. Furthermore, (ONO)Ir(PPh3 )2 Me has been found to undergo an unusual C−H activation route with benzene that initially proceeds by intramolecular C−H activation, followed by intermolecular C−H activation. Chapter 3 focuses on rhodium due to its well-documented nature as an effective C−H activating metal [9] and whose lower electronegativity may allow it to avoid poisoning by coordinating media. In this chapter, a computational screening of a variety of ligands was undertaken in order to find a Rh-ligand complex with the predicted optimal combination of low methane activation energy, low methyl group functionalization energy, and overoxidation protection of the MeX product. Our best F results were the RhIII bis(N -pentafluorophenyl)pentafluorobenzylamidinate (RhIII (NN )) complex and the RhIII bis(quinolinyl)benzene (RhIII (bisq)) complex. In addition, we also report a correlation between Rh−Me bond energies and methane activation barriers that would allow us to easily predict the suitability of new complexes for methane activation. 1.2 Investigation and applications of the Reduction-Coupled Oxo Activation (ROA) mechanistic motif towards alkane upgrading. Vanadium phosphorus oxide (VPO) is an inorganic complex that catalyzes the oxidation of n-butane to maleic anhydride with a surprisingly high selectivity of 60-70% [10] (Scheme 1.2). Although the overall yield is only ca. 50%, its ease of synthesis, use of common elements, and selectivity have allowed this catalyst to be commercialized, producing ca. 500 kilotons of maleic anhydride annually [11]. Scheme 1.2. The oxidation of n-butane to maleic anhydride using vanadium phosphorus oxide as a catalyst. Surprisingly, the mechanism has been shown to initiate with C−H activation by a P−O moiety [12]. Whereas the many vanadium oxos and peroxos investigated bound with hydrogen with a 4 maximum energy of only 70.0 kcal/mol, too low for initial C−H activation of n-butane to occur, the phosphorus oxo has a much larger bond strength with hydrogen, at 84.3 kcal/mol. This implies a first step that is endothermic by only 5.0 kcal/mol, with an activation energy later calculated to be 13.6 kcal/mol, which is within the experimental range. The stability of the resulting monohydrogenated species was attributed to the fact that spin density appears to be localized on the neighboring vanadium atoms (Figure 1.1). Figure 1.1. C−H activation by a phosphorus oxo leads to a one electron reduction on the neighboring vanadium. Hence the oxidation state of the phosphorus atom does not change. Chapter 4 focuses on deeper investigations into the ROA mechanistic motif. Specifically, questions such as the generalizability of the mechanism to other metal-main group pairs, as well as the applications of the ROA motif to homogeneous catalysts, are addressed. We have designed the homogeneous complex κ-P bis(2-phenoxyl)phosphinite vanadium ((OPO)V) to be a catalyst that may activate propane using the ROA mechanistic motif. We then continued to follow the reaction forward, tracing pathways for the catalytic oxidation of propane in aqueous solution using O2 as the terminal oxidant. We found that the ((OPO)V) system has the potential to be a viable catalyst in acidic aqueous solution, converting propane to either propylene or isopropanol. 1.3 Other projects Chapter 5 is a computational mechanistic study of the 2,6-bis(4-isopropyl-2-oxazolin-2-yl)pyridine nickel (Ni((iPr)−pybox)) system, a Negishi C−C coupling catalyst. In our computational work, we investigated the coupling of 1-bromoindane with methylzinc iodide catalyzed by the Ni((S, S−iPr)pybox) system to form 1-methylindane. The chiral diastereomers of this catalyst are most notable for their stereoconvergence: Using Ni((S, S−iPr)−pybox) as a catalyst, both enantiomers of racemic 1bromoindane will couple with methylzinc iodide to form (S)-1-methylindane with 90% enantiomeric excess [?]. We confirm that the mechanism proceeds through a NiI -NiIII couple, in contrast to the classic Ni0 -NiII cycle that Negishi catalysts are typically assumed to undergo [?] (Scheme 1.3). I Appendix A is an experimental and computational study of the (L)Rh (COE)(TFA) complexes in which L is DABF or BOZO (DABF = N, N -bis(pentafluorophenyl)-2,3-dimethyl-1,4-diaza-1,3butadiene, COE = cyclooctene, TFA = trifluoroacetate, BOZO = bis(2-oxazolin-2-yl)). These com- 5 Scheme 1.3. The generalized mechanism of the Ni((iPr)−pybox) Negishi coupling catalyst as it proceeds through a NiI -NiIII cycle. The overall equation is ind−Br + MeZnI −−→ ind−Me + ZnBrI. plexes have been shown to catalyze the H/D exchange of benzene in trifluoroacetic acid. Computationally, the mechanism was found to occur by the reversible oxidative addition and reductive I elimination of benzene to and from the (L)Rh (TFAH)(TFA) species. Appendix B is a comparison of various DFT and electron-correlation methods for calculating the group 10 transition metals Ni, Pd, and Pt. These elements are notable in that, despite being in the same periodic group, they each have a different electronic configuration (d8 s2 , d10 , and d9 s1 , respectively). The accuracy of DFT and electron-correlation methods in reproducing the transition energies between the three configurations for the three elements is an important consideration that has deeper implications in the reactivity of more complex systems including these elements. 6 1.4 References [1] Cheng, W-H.; Kung, H. H. Methanol Production and Use (Chemical Industries vol. 57), 1994. [2] IPCC Fourth Assessment Report, Working Group 1, Chapter 2. [3] (a) Labinger, J. A.; Bercaw, J. E. Nature 2002, 417, 507–514; (b) Crabtree, R. H. J. Chem. Soc., Dalton Trans. 2001, 2437–2450; (c) Labinger, J. A. Fuel Proc. Technol. 1995, 42, 325–338; (d) Labinger, J. A. J. Mol. Catal. A: Chem. 2004, 220, 27–35. [4] (a) Fekl, U.; Goldberg, K. I. Adv. Inorg. Chem. 2003, 54, 259–320; (b) Lersch, M.; Tilset, M. Chem. Rev. 2005, 105, 2471–2526. [5] (a) Goldshlegger, N. F.; Tyabin, M. B.; Shilov, A. E.; Shteinman, A. A. Zh. Fiz. Khim. 1969, 43, 2174–2175; (b) Goldshlegger, N. F.; Eskova, V. V.; Shilov, A. E.; Shteinman, A. A. Zh. Fiz. Khim. 1972, 46, 1353–1354; (c) Shilov, A. E.; Shul’pin, G. B. Activation and Catalytic Reactions of Saturated Hydrocarbons in the Presence of Metal Complexes; Kluwer Academic Publishers: Dordrecht, 2000; pp. 259– 317. [6] Periana, R. A.; Taube, D. J.; Evitt, E. R.; Loffler, D. G.; Wentrcek, P. R.; Voss, G.; Masuda, T. Science 1993, 259, 340–343. [7] Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154–162. [8] (a) Wong-Foy, A. G.; Bhalla, G.; Liu, X. Y.; Periana, R. A. J. Am. Chem. Soc. 2003, 125, 14292–14293; 7 (b) Oxgaard, J.; Muller, R. P.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2004, 126, 352–363; (c) Bhalla, G.; Liu, X. Y.; Oxgaard, J.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 11372–11389; (d) Tenn, W. J.; Young, K. J. H.; Bhalla, G.; Oxgaard, J.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 14172–14173; (e) Bhalla, G.; Periana, R. A. Angew. Chem., Int. Ed. 2005, 44, 1540–1543; (f) Young, K. J. H.; Mironov, O. A; Periana, R. A. Organometallics 2007, 26, 2137–2140; (g) Young, K. J. H.; Oxgaard, J.; Ess, D. H.; Meier, S. K.; Stewart, T.; Goddard, W. A.; Periana, R. A. Chem. Commun. 2009, 3270–3272. [9] (a) Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154–162; (b) Hanson, S. K.; Heinekey, D. M.; Goldberg, K. I. Organometallics 2008, 27, 1454–1463; (c) Evans, M. E.; Jones, W. D. Organometallics 2011, 30, 3371–3377; (d) Zakzeski, J. J.; Bell, A. T. J. Mol. Cat. A: Chem. 2007, 276, 8–16; (e) Zhang, X. X.; Wayland, B. B. J. Am. Chem. Soc. 1994, 116, 7897–7898; (f) Tenn, W. J. III; Conley, B. L.; Bischof, S. M.; Periana, R. A. J. Organomet. Chem. 2011, 696, 551–558; (g) Rhinehart, J. L.; Manbeck, K. A.; Buzak, S. K.; Lippa, G. M.; Brennessel, W. W.; Goldberg, K. I.; Jones, W. D. Organometallics 2012, 31, 1943–1952; (h) Li, L.; Brennessel, W. W.; Jones, W. D. Organometallics 2009, 28, 3492–3500; (i) Kloek, S. M.; Heinekey, D. M.; Goldberg, K. I. Angew. Chem. Int. Ed. 2007, 46, 4736–4738. [10] Centi, G.; Trifiro, F.; Ebner, J. R.; Franchetti, V. M. Chem. Rev. 1988, 88, 55–80. [11] Hodnett, B. K. Heterogeneous Catalytic Oxidation. Wiley: New York, 2000. [12] Cheng, M.-J. and Goddard, W. A. J. Am. Chem. Soc. 2013, 135, 4600–4603. 8 Chapter 2 Intra- and intermolecular C−H bond activation by bis(phenolate)pyridineiridium(III) complexes Reproduced with permission from Fu, R.; Bercaw, J. E.; Labinger, J. A. Organometallics 2011, 30, 6751–6765. c 2011 American Chemical Society; available at http://dx.doi.org/10.1021/om201069k. 2.1 Abstract A bis(phenolate)pyridine pincer ligand (henceforth abbreviated as ONO) has been employed to support a variety of iridium complexes in oxidation states I, III, and IV. Complexes (ONO)IrL2 Me (L = PPh3 , PEt3 ) react with I2 to cleave the Ir−C bond and liberate MeI, apparently via a mechanism beginning with electron transfer to generate an intermediate IrIV complex, which can be isolated and characterized for the case L = PEt3 . The PPh3 complex is transformed in benzene at 65◦ C to the corresponding phenyl complex, with loss of methane, and subsequently to a species resulting from metalation of a PPh3 ligand. Labeling and kinetics studies indicate that PPh3 is the initial site of C−H activation, even though the first observed product is that resulting from intermolecular benzene activation. C−H activation of acetonitrile has also been observed. 9 2.2 Introduction Functionalization of alkanes, such as the selective oxidation of methane to methanol, has long been a desired goal of both academic and industrial research [1]. Much of the focus has been on routes involving C−H activation at a transition metal center followed by oxidative cleavage of the resulting C−M bond, which offer greater possibilities for good selectivity by avoiding radical-based pathways. (Yields of methanol from the direct reaction of methane with dioxygen, where radical pathways predominate, appear to be limited to about 5%, because methanol is more rapidly oxidized than methane, a consequence of its weaker C-H bond [1]d.) Platinum complexes have received a great 2– deal of attention [2], starting with Shilov’s seminal work on the PtCl2– 4 /PtCl6 system, which can oxidize methane to methanol with some selectivity [3], and including the platinum bipyrimidine system, which has achieved the most impressive performance to date, converting methane to methyl bisulfate in up to 70% yield [4]. However, none of this platinum-based chemistry has yet been shown to lead to a practical process, owing in part to low reactivity, and a good deal of research activity has turned to other metals. Iridium in particular has been a popular choice; many (primarily low-valent) Ir complexes have been reported to show good activity for C−H activation [5], and examples of oxidative functionalization with Ir systems are known as well [6]. Interconversions between IrIII and IrV might be important in this chemistry, by analogy to the Shilov system, which involves PtII and PtIV . A separate program in our group has examined dianionic tricoordinate (“LX2 ”) pincer ligands for reactions involving highvalent early transition metal centers [7], and it occurred to us that such strongly electron donating, hard, and robust ligands might also be useful here, to provide more facile access to the relatively high IrV oxidation state. We previously reported some chemistry of a bis(phenolate)-N -heterocyclic carbene ligand (OCO) that supports complexes in the oxidation states IrI , IrIII , and IrIV , but no C−H activation chemistry was demonstrated [8]. Here we report on two related bis(phenolate)pyridine ligands (ONO = pyridine-2,6-bis[2-(4,6-di-tert-butylphenolate)]; ONOtBu = 4-tert-butylpyridine-2,6-bis[2-(4,6-ditert-butylphenolate)] 2.1), which also enable synthesis of a wide variety of IrI , IrIII , and IrIV comtBu plexes. In addition, the IrIII complex (ONO )Ir(PPh3 )2 Me readily activates C−H bonds of benzene and acetonitrile, as well as of coordinated triphenylphosphine. 2.3 Results and Discussion 2.3.1 Synthesis and Characterization of IrI complexes Because IrIII is usually highly substitutionally inert, we decided to enter the Ir(ONO) system via IrI . H2 (ONO) [9] was doubly deprotonated with NaH, followed by metalation with [Ir(cod)Cl]2 , giving 10 Figure 2.1. The ONO and ONOtBu ligands. Na[Ir(cod)(ONO)], 1, which was characterized by NMR and mass spectrometry. The symmetry of the 1 H NMR (only two tert-butyl peaks, for example) is consistent with either a five-coordinate structure or a fluxional four-coordinate structure, with the phenolates coordinated to Na+ and Ir rapidly interchanging (1a and 1b, respectively, in Scheme 2.1). The latter appears more likely on two grounds: First, the related OCO complex of IrI was crystallized (as the [K(18-crown-6)] + salt 2) and shown to have a structure analogous to that of 1b [8]. Second, attempted crystallization of 1 in the presence of 15-crown-5 (attempts to crystallize 1 without 15-crown-5 were unsuccessful) instead yielded 3, the result of C−H activation at the 3-position of the central pyridine ring (Figure 2.2). Scheme 2.1. Synthesis of Na[Ir(cod)(ONO)], 1. C−H activation probably results from complexation of the sodium ion by 15-crown-5, making it too sterically bulky to coordinate to the pendant phenolate in structure 1b, and thus labilizing the pyridine, which can flip over to allow the oxidative addition of the 3-C−H bond; the resulting 11 Figure 2.2. Crystal structure of 3. The nearest Na(15-crown-5) is located above the plane of the ligand and has been omitted for clarity. Because of crystal sensitivity, the structure obtained was only of sufficient quality to establish connectivity. iridium (III) hydride is deprotonated by the pendant phenolate to regenerate square-planar IrI . This reactivity appears to require a solution of 1 to contain the isomer 1b, either as the sole species present or, at least, in equilibrium with 1a. C−H activation in preference to N -coordination of a pyridine ring has been observed previously for an NHC-pyridine ligand [10]; in that case (and probably here as well) N -coordination is disfavored sterically. tBu In order to prevent this unwanted ligand C−H activation, we prepared H2 (ONO ), with an additional tert-butyl group to inhibit C−H activation of the pyridine linker. The synthesis of the ligand, which was straightforward and analogous to that of H2 (ONO), is shown in Scheme 2.2. DetBu protonation of H2 (ONO ) with NaH, followed by metalation with half an equivalent (i.e. one Ir per ligand) of [Ir(cod)Cl]2 , gave deep red [NaIr(cod)(ONO tBu )]2 (4, Scheme 2.3), shown crystallo- graphically to have a dimeric structure (Figure 2.3). Unlike ONO complex 1, 4 is readily soluble in pentane, presumably because the hydrophilic Na2 O2 core is surrounded by a lipophilic shell of tertbutyl-substituted phenyl rings. As with 1, the 1 H NMR spectrum indicates either a more symmetric structure in solution or a fluxional process; on cooling to −100◦ C, the spectrum exhibited some line broadening, but no decoalescence of peaks. 4 reacts with CO to give Ir4 (CO)12 and is reduced by H2 to metallic Ir. tBu If instead a full equivalent of [Ir(cod)Cl]2 is added to one equivalent of Na2 (ONO tBu diiridium complex [Ir(cod)]2 (ONO ), the orange ) (5) is obtained. 5 may also be prepared from 4 plus half an tBu equivalent of [Ir(cod)Cl]2 , but 4 is not formed from 5 and additional Na2 (ONO ) (Scheme 2.3). The two iridium atoms in 5 are inequivalent: Ir1 is five-coordinate, bound to both oxygens and the 12 Scheme 2.2. Synthesis of the ligand H2 (ONOtBu ). Scheme 2.3. Synthesis and reactivity of [NaIr(cod)(ONOtBu )]2 , 4. 13 Figure 2.3. Crystal structure of 4. The tert-butyl groups and a disordered pentane solvent molecule have been removed for clarity. Selected bond lengths (Å) and angles (◦ ): Ir-O1, 2.0990(1); Ir-N1, 2.1274(1); O1-Na1, 2.2145(1); O2-Na1, 2.1926(1); O20 -Na1, 2.2841(1); O1-Ir-N1, 81.268(3); O1-Na1-O2, 104.799(3); O1-Na1-O20 , 106.989(3); O2-Na1-O20 , 100.484(2). nitrogen of the ligand, while Ir2 is square planar and coordinated only to the two phenolate oxygens (Figure 2.4). (The Ir1-N and Ir2-N distances are 2.07 and 3.00 Å, respectively.) 2.3.2 Synthesis and Characterization of IrIII Complexes tBu Complex 4 is oxidized by AgPF6 or [FeCp2 ][PF6 ] in a mixture of THF and acetonitrile to [(ONO )- Ir(cod)(MeCN)][PF6 ] (6, Scheme 2.4), which was isolated as a crystalline product, although it decomposes over time in the presence of excess acetonitrile. Oxidation of 4 with AgOTf in benzene gave tBu the neutral triflate complex (ONO )Ir(cod)OTf (7). Both complexes exhibit approximately octa- hedral geometry (Figures 2.5 and 2.6); in solution they behave as strong electrophiles, polymerizing THF and slowly decomposing in other coordinating solvents. Scheme 2.4. Oxidation of [NaIr(cod)(ONOtBu )]2 , 4, to IrIII species. Treatment of 7 with two equivalents of triphenylphosphine yielded yellow trans-(ONO tBu )Ir- (PPh3 )2 OTf (8, Equation 2.1). Attempts to displace cyclooctadiene from 7 with triethylphosphine or 14 Figure 2.4. Crystal structure of 5. The tert-butyl groups have been removed for clarity. Selected bond lengths (Å) and angles (◦ ): Ir1-O1, 2.2701(1); Ir1-O2, 2.2962(1); Ir1-N1, 2.0818(1); Ir2-O1, 2.0808(1); Ir2-O2, 2.0771(1); Ir2-N1, 2.9953(1); O1-Ir1-O2, 68.864(1); N1-Ir1-O1, 76.957(2); N1-Ir1-O2, 75.747(1); O1-Ir2-O2, 76.775(2); Ir1-O1-Ir2, 103.134(2); Ir1-O2-Ir2, 102.367(2). Figure 2.5. Crystal structure of 6. A benzene solvent molecule and the PF6 counterion have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.0736(8); Ir1-N2, 2.0456(8); Ir1-O1, 2.0540(7); Ir1-O2, 2.0849(7); N2-C46, 1.1371(12); O1-Ir1-N1, 85.86(3); O1-Ir1-N2, 81.05(3); O2-Ir1-O1, 89.17(3); O2Ir1-N2, 85.09(3); O1-Ir1-O2, 165.29(3); N1-Ir1-N2, 89.69(3); Ir1-N2-C46, 173.51(8). 15 Figure 2.6. Crystal structure of 7. The two tert-butyl groups ortho to the phenoxy moieties have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O1, 2.0431(6); Ir1-N1, 2.0738(6); Ir1-O2, 2.0836(6); Ir1-O3, 2.1247(6); Ir1-C39, 2.1725(8); Ir1-C40, 2.1865(8); Ir1-C44, 2.2592(8); Ir1-C43, 2.2656(8); O1-Ir1-N1, 84.59(2); O1-Ir1-O2, 163.22(2); N1-Ir1-O2, 88.90(2); O1-Ir1-O3, 80.91(2); N1-Ir1-O3, 87.90(2); O2-Ir1-O3, 83.42(2); O1-Ir1-C39, 121.97(3); N1-Ir1-C39, 95.20(3); O2-Ir1-C39, 73.96(3); O3-Ir1C39, 157.08(3); O1-Ir1-C40, 84.60(3); N1-Ir1-C40, 90.09(3); O2-Ir1-C40, 110.91(3); O3-Ir1-C40, 165.50(3); C39-Ir1-C40, 37.42(3); O1-Ir1-C44, 110.92(3); N1-Ir1-C44, 164.19(3); O2-Ir1-C44, 75.34(3); O3-Ir1-C44, 91.51(3); C39-Ir1-C44, 79.40(3); C40-Ir1-C44, 94.28(3); O1-Ir1-C43, 77.60(3); N1-Ir1-C43, 159.93(3); O2-Ir1C43, 110.72(3); O3-Ir1-C43, 98.24(3); C39-Ir1-C43, 86.58(3); C40-Ir1-C43, 79.18(3); C44-Ir1-C43, 35.50(3). tricyclohexylphosphine led only to decomposition, while the reaction of 7 with tris(o-tolyl)phosphine resulted in protonated ligand, perhaps by benzylic C−H activation followed by reductive elimination of phenol. The crystal structure of 8 is show in Figure 2.7. (2.1) Exposure of 8 to water, followed by addition of proton sponge at room temperature gave transtBu (ONO )Ir(PPh3 )2 OH (9, Equation 2.2). Both 8 and 9 were converted to the corresponding chloride (10) by heating at 90◦ C in dichloromethane for several hours or several days, respectively (Equation 2.3). Formation of 10 from 8 or 9 is much faster in the presence of added tetrabutylammonium chloride. The crystal structure of 9 is show in Figure 2.8. 16 Figure 2.7. Crystal structure of 8. Two benzene solvent molecules, a tert-butyl group’s methyls, and the phosphine ligands’ phenyl groups (except ipso carbons) have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.001(3); Ir1-O1, 2.049(2); Ir1-O2, 2.051(2); Ir1-O3, 2.166(2); Ir1-P1, 2.4142(8); Ir1-P2, 2.4165(8); N1-Ir1-O1, 90.68(9); N1-Ir1-O2, 91.55(9); O1-Ir1-O2, 177.70(8); N1-Ir1-O3, 176.28(9); O1-Ir1-O3, 87.33(8); O2-Ir1-O3, 90.41(8); N1-Ir1-P1, 90.90(7); O1-Ir1-P1, 96.69(6); O2-Ir1-P1, 82.71(6); O3-Ir1-P1, 86.22(6); N1-Ir1-P2, 88.06(7); O1-Ir1-P2, 81.27(6); O2-Ir1-P2, 99.37(6); O3-Ir1-P2, 94.74(6); P1-Ir1-P2, 177.70(3). (2.2) (2.3) All three complexes (8, 9, and 10) react with proton sponge to give the hydride (ONO tBu )Ir- (PPh3 )2 H (11). The reaction of 8 takes place over several days at room temperature or in an hour at 90◦ C, while 9 and 10 react more slowly at 90◦ C. The crystal structure of 11 is show in Figure 2.9. The proposed mechanism for this transformation is single-electron transfer followed by H-atom abstraction (Scheme 2.5); the oxidized proton sponge byproduct shown was observed (quantitatively) 17 Figure 2.8. Crystal structure of 9. The tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.0395(14); Ir1-O3, 2.0561(13); Ir1-O1, 2.0624(11); Ir1-O2, 2.0686(11); Ir1-P1, 2.3654(5); Ir1-P2, 2.3920(5); N1-Ir1-O3, 179.30(6); N1-Ir1-O1, 89.78(5); O3-Ir1-O1, 90.03(5); N1-Ir1O2, 89.99(5); O3-Ir1-O2, 90.22(5); O1-Ir1-O2, 178.69(5); N1-Ir1-P1, 97.67(4); O3-Ir1-P1, 83.02(4); O1-Ir1P1, 95.33(4); O2-Ir1-P1, 83.43(4); N1-Ir1-P2, 90.84(4); O3-Ir1-P2, 88.46(4); O1-Ir1-P2, 81.46(4); O2-Ir1-P2, 99.83(4); P1-Ir1-P2, 170.911(15). Figure 2.9. Crystal structure of 11. A pentane solvent molecule and the tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O1, 2.0663(11); Ir1-O2, 2.0764(11); Ir1-N1, 2.1020(16); Ir1-P2, 2.3229(5); Ir1-P1, 2.3320(5); Ir1-H1, 1.70(2); O1-Ir1-O2, 176.19(5); O1-Ir1-N1, 87.97(5); O2-Ir1-N1, 88.25(5); O1-Ir1-P2, 86.17(4); O2-Ir1-P2, 94.68(3); N1-Ir1-P2, 95.82(4); O1-Ir1-P1, 96.13(4); O2-Ir1-P1, 83.91(3); N1-Ir1-P1, 97.67(4); P2-Ir1-P1, 166.384(19); O1-Ir1-H1, 95.6(6); O2-Ir1-H1, 88.2(6); N1-Ir1-H1, 176.3(7); P2-Ir1-H1, 85.7(7); P1-Ir1-H1, 80.7(7). 18 by NMR. A related reaction of proton sponge with an IrIII complex has previously been reported [11]. In further support of this proposal, 11 was also obtained by treatment of 8 with cobaltocene (a oneelectron donor) and the H atom donor 9,10-dihydroanthracene (DHA), or by heating 8 with sodium hydride. 11 can also be obtained from the reaction of 8 with lithium methoxide or diethylzinc, presumably via β-hydride elimination from an iridium-methoxy [12] or iridium-ethyl intermediate, respectively. The two latter routes give 11 most cleanly; in the reactions with proton sponge, we were unable to separate 11 from the byproducts. Scheme 2.5. Synthetic routes to (ONOtBu )Ir(PPh3 )2 H, 11. Treatment of 8 with dimethylzinc yielded the water-stable methyl complex (ONO (12). Oxidation of 12 with I2 yields the iodide (ONO tBu tBu )Ir(PPh3 )2 Me )Ir(PPh3 )2 I (13) and methyl iodide (Scheme 2.6). Scheme 2.6. Methylation to (ONOtBu )Ir(PPh3 )2 Me, 12, and subsequent functionalization with I2 to (ONOtBu )Ir(PPh3 )2 I, 13. Crystal structures for complexes 8, 9, 11, 12, and 13 all exhibit closely analogous octahedral geometries. The structures of 12 and 13 are shown in Figures 2.10 and 2.11. Unlike 8, 12 reacts cleanly with triethylphosphine to give the analogous PEt3 complex (14, 19 Figure 2.10. Crystal structure of 12. Only ipso carbon atoms of the phenyl groups on phosphorus are shown. Selected atom distances (Å) and angles (◦ ): Ir-O1, 2.0901(1); Ir-O2, 2.0700(1); Ir-C74, 2.0952(1); Ir-N, 2.1042(1); Ir-P1, 2.3449(1); Ir-P2, 2.3895(1); O1-Ir-O2, 176.495(3); P1-Ir-P2, 172.872(2); N-Ir-C74, 175.671(3); C74-Ir-O1, 95.588(2); C74-Ir-O2, 87.799(2); C74-Ir-P1, 85.584(2); C74-Ir-P2, 87.309(2); O1-IrP1, 82.692(2); O1-Ir-P2, 98.559(2); O2-Ir-P1, 96.698(2); O2-Ir-P2, 82.473(2); N-Ir-O1, 88.429(2); N-Ir-O2, 88.212(2); N-Ir-P1, 96.553(2); N-Ir-P2, 90.503(2). Figure 2.11. Crystal structure of 13. Two pentane solvent molecules and the tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.040(3); Ir1-O2, 2.066(2); Ir1-O1, 2.076(2); Ir1-P2, 2.4233(10); Ir1-P1, 2.4247(11); Ir1-I2, 2.7013(3); N1-Ir1-O2, 89.52(10); N1-Ir1-O1, 88.88(10); O2-Ir1-O1, 178.36(9); N1-Ir1-P2, 89.69(9); O2-Ir1-P2, 78.59(8); O1-Ir1-P2, 101.03(8); N1-Ir1-P1, 89.04(9); O2-Ir1-P1, 102.10(8); O1-Ir1-P1, 78.25(8); P2-Ir1-P1, 178.55(3); N1-Ir1-I2, 179.19(9); O2-Ir1-I2, 90.56(6); O1-Ir1-I2, 91.05(7); P2-Ir1-I2, 91.12(2); P1-Ir1-I2, 90.15(2). 20 Scheme 2.7). 14 appears to be more thermally stable than 12, only slowly decomposing at tempertBu atures above 120◦ C. It reacts similarly with I2 , giving methyl iodide and (ONO )Ir(PEt3 )2 I (15). Both 14 (Figure 2.12) and 15 (Figure 2.13) were characterized crystallographically. Iodide complex 15 is unstable in methylene chloride, converting to the analogous chloride over several days at room temperature. Such behavior was not observed with the PPh3 analogue 13. Scheme 2.7. Synthesis of (ONOtBu Ir(PEt3 )2 complexes. Figure 2.12. Crystal structure of 14. The ethyl groups of one phosphine ligand have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O2, 2.0674(11); Ir1-O1, 2.0702(11); Ir1-N1, 2.1050(14); Ir1-C50, 2.1154(17); Ir1-P2, 2.3479(5); Ir1-P1, 2.3514(5); O2-Ir1-O1, 178.26(5); O2-Ir1-N1, 88.65(5); O1Ir1-N1, 89.83(5); O2-Ir1-C50, 91.12(6); O1-Ir1-C50, 90.42(6); N1-Ir1-C50, 178.40(6); O2-Ir1-P2, 84.03(4); O1-Ir1-P2, 95.23(4); N1-Ir1-P2, 93.93(4); C50-Ir1-P2, 87.61(5); O2-Ir1-P1, 97.78(4); O1-Ir1-P1, 83.10(4); N1-Ir1-P1, 91.40(4); C50-Ir1-P1, 87.07(5); P2-Ir1-P1, 174.416(16). In both I2 oxidations a dark blue color is observed over much of the course of the reaction, even though the starting complexes and the products are all yellow. Cyclic voltammetry experiments show quasi-reversible oxidations at 20 and 780 mV for 12 and reversible oxidations at −100 and 21 Figure 2.13. Crystal structure of 15. The ethyl groups of the phosphine ligands have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.0619(13); Ir1-O2, 2.0648(11); Ir1-O1, 2.0661(11); Ir1-C(50), 2.148(7); Ir1-P1, 2.3711(4); Ir1-P2, 2.3722(4); Ir1-I2, 2.6625(3); N1-Ir1-O2, 89.86(5); N1-Ir1-O1, 88.67(5); O2-Ir1-O1, 178.36(5); N1-Ir1-C(50), 177.12(15); O2-Ir1-C(50), 89.65(17); O1-Ir1-C(50), 91.86(17); N1-Ir1-P1, 93.81(4); O2-Ir1-P1, 95.64(3); O1-Ir1-P1, 83.74(3); C(50)-Ir1-P1, 89.06(15); N1-Ir1-P2, 91.30(4); O2-Ir1-P2, 82.36(3); O1-Ir1-P2, 98.40(3); C(5)-Ir1-P2, 85.82(15); P1-Ir1-P2, 174.510(14); N1-Ir1-I2, 178.40(4); O2-Ir1-I2, 91.65(3); O1-Ir1-I2, 89.83(3); C(50)-Ir1-I2, 3.05(16); P1-Ir1-I2, 86.571(11); P2-Ir1-I2, 88.374(11). 840 mV for 14 both vs. ferrocene (Figure 2.14). The reaction of 14 with AgPF6 gave dark blue tBu [(ONO )Ir(PEt3 )2 Me]PF6 (16, Scheme 2.7), characterized by a crystal structure and an EPR spectrum (Figure 2.15) that are very similar to those of the (relatively few) organometallic IrIV species reported [13]. 16 could be reduced back to 14 with either cobaltocene or methyllithium; furthermore, treatment of 16 with tetrabutylammonium iodide gave MeI and some 15. All of this suggests that the reaction of 14 (or 12) with I2 begins with a single electron transfer to generate an intermediate cationic IrIV -Me species, which undergoes SN 2 attack by I– at the Ir−C bond. The CV data suggest that further oxidization of 16 might be possible, perhaps even to an IrV -methyl complex, but no such species has yet been isolated or otherwise identified. (The close similarity of the second potentials, in contrast to the first, may indicate that the second wave corresponds to oxidation of the ONOtBu ligand rather than at Ir.) The crystal structure of 16 is show in Figure 2.16). 2.3.3 C−H Activation by IrIII (ONOtBu ) Complexes In benzene solution, (ONO tBu )Ir(PPh3 )2 Me converts over the course of several hours at 65◦ C to the analogous phenyl complex (ONO tBu )Ir(PPh3 )2 Ph (17), with evolution of methane. The benzene solution of 17 reacts further over the course of several weeks at 65◦ C to give a cyclometalated tBu product, (ONO )Ir(PPh3 )[κ2 -PPh2 (o-C6 H4 )] (18, Equation 2.4). 22 Figure 2.14. Cyclic voltammetry of 14, carried out in the air at room temperature. The solvent was dichloromethane with 0.1 M tetrabutylammonium tetrafluoroborate added as the electrolyte. Multiple runs were made with scan rates from 50 to 400 mV/s. Quasi-reversibility is seen in the form of a following chemical step. The dip at 0.3 V is an artifact. Figure 2.15. EPR spectrum of [(ONOtBu )Ir(PEt3 )2 Me]PF6 , 16, in dichloromethane at 20 K. 23 Figure 2.16. Crystal structure of one of the molecules of 16 in the unit cell. The unit cell has four iridiumcontaining molecules total; this figure only shows one of them. In addition, ethyl groups on the phosphine ligands, as well as the six benzene molecules per unit cell, have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O2A, 1.994(5); Ir1-O1A, 2.011(5); Ir1-C50A, 2.101(8); Ir1-N1A, 2.129(7); Ir1P2A, 2.400(2); Ir1-P1A, 2.418(2); O2A-Ir1-O1A, 173.5(2); O2A-Ir1-C50A, 94.2(3); O1A-Ir1-C50A, 92.3(3); O2A-Ir1-N1A, 87.4(2); O1A-Ir1-N1A, 86.1(2); C50A-Ir1-N1A, 177.4(3); O2A-Ir1-P2A, 82.62(17); O1A-Ir1P2A, 98.06(18); C50A-Ir1-P2A, 84.4(2); N1A-Ir1-P2A, 93.78(18); O2A-Ir1-P1A, 98.16(17); O1A-Ir1-P1A, 81.81(18); C50A-Ir1-P1A, 89.8(2); N1A-Ir1-P1A, 91.96(18); P2A-Ir1-P1A, 174.24(8). (2.4) When 17 is isolated, redissolved in a solvent other than benzene (such as p-xylene), and heated to 65◦ C, conversion to 18 is complete within a few days. Both 17 (Figure 2.17 and 18 (Figure 2.18) were characterized crystallographically. Thermolysis of 12 in C6 D6 yields CH4 as the only detectable methane isotopologue, and mass spectrometry of the resulting isolated phenyl product 17 shows incorporation of six D atoms (EquatBu tion 2.5). In contrast, heating (ONO )Ir[P(C6 D5 )3 ]2 (CH3 ) (12-d30 ) in C6 D6 yields only CH3 D; this reaction is slower than that of all-protio-12, exhibiting a KIE of 4 2.20. The PEt3 analogue 14 tBu reacts very slowly in C6 H6 or C6 D6 , requiring several weeks at 120◦ C to give (ONO and CH4 or CH3 D, respectively. )Ir(PEt3 )2 Ph 24 Figure 2.17. Crystal structure of 17. A benzene and a pentane solvent molecule, as well as the phosphine phenyl groups, have been removed for clarity. The molecule is symmetric about the iridium center. Selected atom distances (Å) and angles (◦ ): Ir1-O1, 2.0744(12); Ir1-C40, 2.074(3); Ir1-N1, 2.103(2); Ir1-P1, 2.3921(6); O1-Ir1-C40, 90.05(4); O1-Ir1-O10 , 179.91(8); C40-Ir1-O10 , 90.04(4); O1-Ir1-N1, 89.95(4); C40-Ir1N1, 180.0; O10 -Ir1-N1, 89.96(4); O1-Ir1-P10 , 97.70(4); C40-Ir1-P10 , 88.831(13); O10 -Ir1-P10 , 82.30(4); N1-Ir1P10 , 91.169(13); O1-Ir1-P1, 82.30(4); C40-Ir1-P1, 88.829(13); O10 -Ir1-P1, 97.70(4); N1-Ir1-P1, 91.171(13); P10 -Ir1-P1; 177.66(3). Figure 2.18. Crystal structure of 18. A dodecane solvent molecule and parts of several phenyl and tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir-O1, 2.0696(1); Ir-O2, 2.0612(1); Ir-C46, 2.0669(1); Ir-N, 2.1156(1); Ir-P4, 2.3462(1); Ir-P5, 2.3483(1); O1-Ir-O2, 174.874(3); P4Ir-P5, 164.008(2); N-Ir-C46, 167.055(4); C46-Ir-O1, 100.779(3); C46-Ir-O2, 80.655(3); C46-Ir-P4, 95.708(2); C46-Ir-P5, 68.499(2); O1-Ir-P4, 85.316(3); O1-Ir-P5, 94.944(2); O2-Ir-P4, 99.475(2); O2-Ir-P5, 80.959(3); N-Ir-O1, 88.656(3); N-Ir-O2, 89.198(3); N-Ir-P4, 93.867(3); N-Ir-P5, 102.125(3). 25 (2.5) Thermolysis of 12 at 65◦ C in methylene chloride, THF, ether, or p-xylene leads directly to 18 (Equation 2.6). These transformations require less than one day for complete conversion (in methylene chloride or p-xylene), comparable to the time needed for conversion of 12 to 17 and much faster than the conversion of 17 to 18 in benzene. Reaction of 12 in toluene at 65◦ C proceeds similarly to that in benzene; although a single pure compound could not be isolated, the mass spectrum of the mixture is consistent with formation of an IrIII tolyl compound, (ONO The 31 tBu )Ir(PPh3 )2 (C6 H4 CH3 ). P NMR shows two strong signals of roughly equal intensity with shifts close to that of phenyl compound 17, suggesting the formation of two isomers, presumably meta and para (since p-xylene is unreactive); a much weaker nearby signal may indicate a small amount of the ortho isomer. (2.6) Although 12 is insoluble in acetonitrile, thermolysis in methylene chloride containing a small amount of acetonitrile affords a mixture of 18 and a product resulting from C−H activation of acetBu tonitrile, [(ONO )Ir(PPh3 )(CH2 CN)]2 (19); with CD3 CN, CH4 is the only methane isotopologue produced (Equation 2.7). 19 has the dimeric structure shown in Figure 2.19, where coordination of N from the (CH2 CN) group bonded to each Ir has displaced a PPh3 ligand from the other. Although two 1 H NMR signals would be expected for the (CH2 CN) methylene protons, which are diastereotopic in the solid-state structure, only one is observed; this could result from either accidental degeneracy or dissociation of the dimeric structure in solution. 26 (2.7) Figure 2.19. Crystal structure of 19 [14]. A benzene solvent molecule, phosphine phenyl, and (ONO) tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O1, 2.0190(2); Ir1-O2, 2.0883(2); Ir1-C104, 2.1472(2); Ir1-N1, 2.0758(2); Ir1-N4, 2.0882(2); Ir1-P1, 2.2654(2); N3-C103, 1.1271(1); C103-C104, 1.4380(1); N4-C102, 1.1278(1); C102-C101, 1.4482(2); Ir2-O3, 2.0912(2); Ir2-O4, 2.0396(2); Ir2-C101, 2.1740(3); Ir2-N2, 2.0854(3); Ir2-N3, 2.1290(2); Ir2-P2, 2.2574(2); O1-Ir1O2, 173.203(10); N4-Ir1-P1, 173.670(9); N1-Ir1-C104, 170.591(10); C104-Ir1-O1, 81.941(8); C104-Ir1-O2, 98.236(8); C104-Ir1-N4, 86.476(8); C104-Ir1-P1, 90.036(7); O1-Ir1-N4, 88.426(8); O1-Ir1-P1, 96.321(8); O2Ir1-N4, 84.807(8); O2-Ir1-P1, 90.475(8); N1-Ir1-O1, 89.793(8); N1-Ir1-O2, 89.439(8); N1-Ir1-N4, 88.823(8); N1-Ir1-P1, 95.360(8); Ir1-N4-C102, 156.881(14); N4-C102-C101, 171.704(16); C102-C101-Ir2, 104.11(1); Ir2N3-C103, 156.797(13); N3-C103-C104, 172.115(16); C103-C104-Ir1, 105.731(10); O3-Ir2-O4, 174.306(10); N3Ir2-P2, 171.891(9); N2-Ir2-C101, 172.156(10); C101-Ir2-O3, 96.853(8); C101-Ir2-O4, 84.460(8); C101-Ir2-N3, 85.543(7); C101-Ir2-P2, 88.633(7); O3-Ir2-N3, 85.175(8); O3-Ir2-P2, 89.930(7); O4-Ir2-N3, 89.412(8); O4Ir2-P2, 95.646(8); N2-Ir2-O3, 89.263(8); N2-Ir2-O4, 88.981(8); N2-Ir2-N3, 90.101(8); N2-Ir2-P2, 96.309(8). 27 A plausible mechanism for the chemistry of 12 in benzene is shown in Scheme 2.8. The sequence of product formation (17 and 18 being preferred kinetically and thermodynamically, respectively [15]) would seem to suggest that the first C−H activation is the intermolecular reaction with benzene. But that is excluded by the isotopic labeling results. The H atom that departs with methane clearly comes from a PPh3 ligand, not solvent benzene, so PPh3 must be the site of the first C−H activation, even though the corresponding product 18 does not appear at all until much later. We propose therefore that dissociation of one PPh3 ligand gives coordinatively unsaturated A, followed by intramolecular C−H activation and loss of methane to generate cyclometalated intermediate C. In benzene solution coordination of PPh3 to C is not competitive; instead C reacts with benzene to give 17 as initial observed product. Subsequent conversion to the thermodynamic product 18 is much slower. Scheme 2.8. Proposed mechanism for the transformation of 12 in benzene to the kinetic product 17 and the thermodynamic product 18. 28 C−H activation is shown in Scheme 2.8 as proceeding via oxidative addition to give an IrV intermediate, a mechanism that has been supported by both experiment [16] and calculations [17] for the above-mentioned [15] Cp* system. It seems reasonable that similar considerations would apply here as well; but an alternative pathway, such as σ-bond metathesis, cannot be excluded. In any case, conversion of A to C must be irreversible, as no multiple H/D exchange is observed, whereas conversion of C to F must be reversible, since 17 eventually reverts to 18. The fact that products are obtained from solvent activation with toluene but not p-xylene suggests that C−H activation ortho to a methyl substituent is sterically inhibited. As a consequence, in the latter solvent C (or a π-arene adduct analogous to D) has no alternative but to bind PPh3 and 18 forms directly, as it does in solvents (dichloromethane, ether, THF) that are not susceptible to C−H activation at all. In acetonitrile the two pathways are competitive. The reaction of 14 is much slower, presumably because PEt3 dissociates much less readily, and leads only to a phenyl product, indicating that cyclometalation of PEt3 does not take place. 2.3.4 Kinetics of C−H Activation The transformation of 12 to 17 and/or 18 is readily followed by 31 P NMR, as each of the species has a distinct spectrum. In early experiments results were not always highly reproducible; most probably this is due to oxidation of small amounts of PPh3 by adventitious impurities, which could effect significant acceleration; indeed, deliberate addition of oxidants such as chloranil and dioxygen did speed up conversion of 12 considerably. However, reasonably reproducible rates (±15%) could be achieved by careful attention to purification of starting complex and solvent as well as exclusion of air. The somewhat simplified mechanism shown in Scheme 2.9 may be used for analysis; it leaves out the (unobserved) intermediates B (which will rapidly convert to C) and D and E (which are presumed to be in rapid equilibrium with C and F); none of these should have any kinetic consequences. Three transformations will be considered in turn: (1) conversion of 12 to 17 in benzene; (2) conversion of 12 to 18 in p-xylene; and (3) conversion of 17 to 18 in benzene or p-xylene. The first of these may be analyzed in terms of reversible dissociation of PPh3 (henceforth abbreviated as L) followed by rate-determining C−H activation (the k2 step). Since the conversion of A to C is irreversible (vide supra), no buildup of any intermediate is observed, and formation of 17 is essentially complete before any 18 is observed. In this simplified framework, we can treat the dissociation of L from 12 as a fast pre-equilibrium and the intramolecular C−H activation of L as rate-determining, giving us K1 = [A][L]/[12]. Thus the rate law would be −d[12]/dt = d[17]/dt = k2[A] = (k2 K1 [12])/[L], and the rate should be cleanly first order in 12 and inverse-first-order in L. There is a potential complication if no extra L is added: [L] will be governed by both K1 and K4 , so if these two equilibrium constants were significantly different, [L] would vary 29 Scheme 2.9. Simplified version of Scheme 2.8, used for kinetic analysis. 30 in a complex manner as conversion of 12 to 17 proceeds, and the disappearance of 12 or appearance of 17 might not follow clean first-order kinetics. Nonetheless, it does: a good exponential fit is achieved (Figure 2.20, left), with kobs = 1.8(3)×10−4 s−1 (equivalent to a half-life of about an hour), so any difference in K values is small enough to neglect. The 12-d30 ((ONO isotopologue behaves similarly, yielding kobs = 4.4(6) × 10 −5 s −1 tBu )Ir[P(C6 D5 )3 ]2 Me) and hence KIE = 4(1) (Figure 2.20, right), consistent with the assignment of C−H activation of L as rate-determining. Figure 2.20. Left: kinetics of conversion of 12 to 17 in C6 D6 at 65◦ C. Right: kinetics of conversion of 12-d30 ((ONOtBu )Ir[P(C6 D5 )3 ]2 Me) to 17-d36 ((ONOtBu )Ir[P(C6 D5 )3 ]2 C6 D5 ) in C6 D6 at 65◦ C. As expected, the addition of extra L slows the reaction significantly; a plot of 1/kobs is linear in [L] (Figure 2.21). In principle, at sufficiently high [L] some 18 should form directly in competition with 17; in practice, such conditions are probably not reachable (and conversion of 12 would become extremely slow anyway). Figure 2.21. Dependence of the reciprocal of the rate of conversion of 12 to 17 on the amount of added triphenylphosphine. See Section 2.5.2 for details on how this plot was constructed. The rate law for direct conversion of 12 to 18 in p-xylene should be identical to the above, and the reaction does exhibit clean first-order kinetics, at a rate (kobs = 8(1) × 10−5 s−1 ) about a factor of 2 slower than that for 12 to 17. There are several possible explanations for the discrepancy. First, the values of K1 and k2 need not be identical to those in benzene (though they would not be expected to differ very much). Second, intermediate C may be complexed by p-xylene to give 31 an analogue of intermediate D, which in this case (in contrast to benzene) would not lead to any productive reaction, but would reduce the concentration of C and hence the rate of formation of 18. However, any substantial (i.e. sufficient to account for the entire factor of 2) concentration of a new intermediate should have been detectable by NMR. Finally (and perhaps most likely), it may be that agreement within a factor of 2 (especially on changing solvent) is as good as can be expected, given the reproducibility considerations discussed earlier. As noted above, the conversion of 17 to 18 is much slower in benzene than in p-xylene, taking place over weeks or days, respectively (Figure 2.22). This observation is readily explained by Scheme 2.9: the concentration of C, the precursor to 18, will be inversely related to the concentration of benzene. The rate law for that mechanism can be derived by using the steady-state approximation for both C and F, both of which should be present in low concentration and in rapid equilibrium with one another. This is most simply accomplished by defining a dummy variable [X] = [C] + [F]. For steady-state in [X]: 0 = d[X]/dt = k−4 [17] − k4 [F][L] − k5 [C][L]. Substituting [F] = K3 [C][PhH] gives k−4 [17] − k4 [F][L] − k5 [C][L] = k−4 [17] − K3 k4 [C][PhH][L] − k5 [C][L] = 0; hence [C] = (k−4 [17])/(k5 [L] + K3 k4 [PhH][L]). Hence the rate law for the conversion is −d[17]/dt = d[18]/dt = k5 [C][L] = (k−4 k5 [17])/(k5 + K3 k4 [PhH]). In benzene solution the reaction should be first-order in 17 and independent of [L]; both were observed. In p-xylene, on the other hand, [PhH] is continuously increasing as 17 is consumed, so first-order kinetics (which are observed) would not be expected unless k5 is large compared to k4 K3 [PhH] over the course of the reaction. Figure 2.22. Kinetics for conversion of 17 to 18 in C6 D6 (left) and p-xylene-d10 (right) at 65◦ C. We can test this for self-consistency with the observed rate constants, again assuming that there is little difference in parameters between benzene and p-xylene reactions. For conversion of 17 to 18 in benzene, kobs = 4.2(6) × 10−7 s−1 . Since no direct formation of 18 from 12 was observed, k5 must be small compared to k4 K3 [PhH], so that value of kobs is approximately equal to k−4 k5 /k4 K3 [PhH]. For the reaction in p-xylene, kobs = 2.9(4) × 10−6 s−1 ; if our assumption that k5 is large compared to k4 K3 [PhH] is correct, that value of kobs = k−4 . Substituting this value into 32 the above expression, we get k5 /k4 K3 [PhH] = 0.13, so k5 is not completely negligible compared to k4 K3 [PhH], but small enough (within the limits of reproducibility as well as the assumption of solvent-independent parameters) to validate the approximation for kobs in benzene. Furthermore, since for neat benzene [PhH] ≈ 11 M, k5 /k4 K3 = 1.4 M. For the reaction in p-xylene, the starting [17] = the maximum [PhH] = 8.8 mM and the minimum value of k5 /k4 K3 [PhH] = 160, so the assumption required to account for first-order kinetics is valid as well. Accordingly, the kinetics appear completely consistent with the proposed mechanism. 2.4 Conclusions The ONO ligand provides a versatile framework for synthesis of a wide variety of IrI and IrIII complexes. Using this framework, two key steps in a scheme for functionalization of hydrocarbons have been demonstrated. Conversion of the Ir−Me bond in complexes 12 and 14 to MeI is effected by addition of I2 ; there is evidence that the reaction proceeds via one-electron oxidation to an (isolable) IrIV methyl intermediate. Triphenylphosphine complex 12 is able to activate C−H bonds in benzene (and toluene) under quite mild conditions; the mechanism follows an unusual pathway in which the first C−H activation is intramolecular, but the product of that process, although thermodynamically preferred, is kinetically disfavored with respect to the intermolecular benzene activation product. Unfortunately the (ONO)Ir III system is not sufficiently reactive to activate nonaromatic C−H bonds (except for acetonitrile). 2.5 Supporting Details 2.5.1 Experimental Section 2.5.1.1 General Methods All compounds were prepared in a nitrogen-filled glovebox unless otherwise specified. Reactions taking place in dichloromethane at 90◦ C were conducted in either sealed glass bombs or J-Young NMR tubes. Solvents in the glovebox were dried using the Grubbs method [18]. All NMR solvents were purchased from Cambridge Isotope Laboratories, Inc., (except for CDCl3 ) filtered through alumina, and stored over 4 Å molecular sieves. POCl3 was purchased from Acros Organics. 2,6-dibromopyridine, nBuLi, tBuLi, 4-tert-butylpyridine, [FeCp2 ][PF6 ], indene, N,N -dimethylaminoethanol, PPh3 , propylene, proton sponge, AgPF6 , AgOTf, 15-crown-5, and NaH were purchased from Aldrich. H2 O2 was purchased from EMD. [Ir(cod)Cl]2 and Pd(PPh3 )4 were purchased from Strem. The synthetic route towards the ligand H2 (ONO) closely follows a previously described procedure [9]. 1 H, 13 C, 19 F, and 31 P NMR spectra were recorded on Varian Mercury 300 MHz NMR spectrometers. X-ray data were 33 collected using a Bruker KAPPA APEX II diffractometer. 2.5.1.2 Synthesis of 4-tert-Butylpyridine-N -oxide Following a published procedure [19], 4-tert-butylpyridine (18.0 mL, 16.6 g, 123 mmol) was mixed with 135 mL of glacial acetic acid and 100 mL of hydrogen peroxide (30% in water) in a 1 L roundbottom flask under air. After refluxing for 4 h, 100 mL of additional hydrogen peroxide (30% in water) was added, and the mixture was refluxed overnight. Then 200 mL of the solvent was removed by distillation, and 100 mL water was added and then removed by distillation as well. The remaining solution was neutralized with sodium carbonate. The organic layer was extracted with methylene chloride, filtered, and evaporated to dryness to yield white crystals (which gave a yellow solution in CH2 Cl2 ). Isolated yield: 15.3 g, 101 mmol, 82%. 1 H NMR (300 MHz, CDCl3 ): δ 1.31 (s, 9H, C(CH3 )3 ), 7.26 (d, 2H, 3 J = 7.2 Hz, aryl-H), 8.15 (d, 2H, 3 J = 7.5 Hz, aryl-H). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 30.5 (C(CH3 )3 ), 34.6 (C(CH3 )3 ), 123.1, 138.6, 151.2 (aryl). 2.5.1.3 Synthesis of 2-Chloro-4-tert-butylpyridine In a 250 mL round-bottom flask open to the air, 4-tert-butylpyridine-N -oxide (15.3 g, 101 mmol) was dissolved in 61 mL of neat POCl3 . The mixture was allowed to reflux under argon for 12 to 16 h (refluxing can also be done under air, but the yield is somewhat lower). Excess POCl3 was removed by distillation, and the remaining liquid was washed with a saturated aqueous solution of sodium carbonate. The organic layer was extracted with ether, filtered, and evaporated to dryness, yielding a dark red oil (13.5 g, 79.3 mmol, 79% yield). 1 H NMR (300 MHz, CDCl3 ): δ 1.26 (s, 9H, C(CH3 )3 ), 7.16 (dd, 1H, 3 J = 5.4 Hz, 4 J = 1.8 Hz, aryl-H), 7.25 (d, 1H, 4 J = 1.8 Hz, aryl-H), 8.23 (dd, 1H, 3 J = 5.7 Hz, 5 J = 0.6 Hz, aryl-H). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 30.3 (C(CH3 )3 ), 35.0 (C(CH3 )3 ), 119.7, 121.3, 149.4, 151.7, 163.5 (aryl). 2.5.1.4 Synthesis of 2-Chloro-4-tert-butylpyridine-N -oxide In a 1 L round-bottom flask, 2-chloro-4-tert-butylpyridine (13.5 g, 79.3 mmol) was mixed with 100 mL of glacial acetic acid and 80 mL of hydrogen peroxide (30% in water). After refluxing for 4 h under air, 80 mL of additional hydrogen peroxide (30% in water) was added and the mixture was refluxed overnight. Then 200 mL of the solvent was removed by distillation, and 100 mL of water was added and then distilled off as well. The remaining solution was neutralized with sodium carbonate. The organic layer was extracted with methylene chloride, filtered, and evaporated to dryness to yield yellow crystals (10.3 g, 55.3 mmol, 70% yield). 1 H NMR (300 MHz, CDCl3 ): δ 1.31 (s, 9H, C(CH3 )3 ), 7.19 (dd, 1H, 3 J = 6.6 Hz, 4 J = 3.0 Hz, aryl-H), 7.44 (d, 1H, 4 J = 2.7 Hz, aryl-H), 8.27 (dd, 1H, 3 J = 6.6Hz, 5 J = 0.6 Hz, aryl-H). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 30.4 (C(CH3 )3 ), 34.7 (C(CH3 )3 ), 121.3, 124.1, 139.8, 151.3 (aryl). 34 2.5.1.5 Synthesis of 2,6-Dichloro-4-tert-butylpyridine In a 250 mL round-bottom flask open to the air, 2-chloro-4-tert-butylpyridine-N -oxide (15.3 g, 101 mmol) was dissolved in 40 mL of neat POCl3 . The mixture was allowed to reflux under argon for 12 to 16 h. Excess POCl3 was removed by distillation, and the remaining liquid was washed with a saturated aqueous solution of sodium carbonate. The organic layer was extracted with ether, filtered, and evaporated to dryness, yielding a thick black oil. This oil was subjected to column chromatography over silica (95:5 hexanes/ethyl acetate), affording brown crystals after evaporation of solvent. The crystals were washed with hexanes, giving white 2,6-dichloro-4-tert-butylpyridine (5.57 g, 27.3 mmol, 49% yield; yield over four steps from 4-tert-butylpyridine: 22%). 1 H NMR (300 MHz, CDCl3 ): δ 1.30 (s, 9H, C(CH3 )3 ), 7.23 (s, 2H, aryl-H). 13 C{1 H} NMR (75.5 MHz, CDCl3 ): δ 30.3 (C(CH3 )3 ), 35.4 (C(CH3 )3 ), 120.2, 150.6, 166.3 (aryl). 2.5.1.6 Synthesis of O-Methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide Bromine (3.7 mL, 12 g, 73 mmol) was added via syringe to a solution of 2,4-di-tert-butylphenol (15 g, 73 mmol) in CH2 Cl2 (200 mL). The brown color of Br2 disappeared upon addition. GC-MS analysis after 5 min shows only the presence of the desired brominated product(M+ = 284 and 286). The organic mixture was washed with water, then dried over MgSO4 , and filtered. Removal of volatile material by rotary evaporation gave a golden oil which solidified after placing under high vacuum (< 1 mTorr). This material (4,6-di-tert-butyl-2-bromophenol) was dissolved in dry THF (200 mL) under argon, and was deprotonated with NaH (1.92 g, 80 mmol). After the addition of NaH the reaction mixture was stirred for 1 h at room temperature; then MOMCl (6.1 mL, 6.5 g, 80 mmol) was added via syringe. The reaction mixture was stirred at room temperature for 9 h. Water was added and the mixture was concentrated under vacuum. The desired product was extracted with CH2 Cl2 (three times); the combined organic fractions were dried over MgSO4 , filtered, and dried under high vacuum overnight to give 23.5 g (98% yield over two steps) of O-methoxymethyl-2-bromo4,6-di-tert-butylphenoxide as a golden oil. 1 H NMR (300 MHz, CDCl3 ): δ 1.30 (s, 9H, C(CH3 )3 ), 1.44 (s, 9H, C(CH3 )3 ), 3.70 (s, 3H, OCH3 ), 5.23 (s, 2H, OCH2 O), 7.32 (d, 2H, aryl-H), 7.41 (d, 2H, aryl-H). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 31.0 (C(CH3 )3 ), 31.5 (C(CH3 )3 ), 34.8 (C(CH3 )3 ), 36.1 (C(CH3 )3 ), 57.9 (OCH3 ), 99.5 (OCH2 O), 117.7, 124.1, 128.9, 144.6, 147.8, 150.7 (aryl). MS FAB+: 328 (M+ ). 2.5.1.7 Synthesis of MOM2 (ONO) A mixture of O-Methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide (5.19 g, 15.8 mmol) and THF (25 mL) in a Schlenk tube fitted with a screw-in Teflon stopper was frozen in a cold well, along with a solution of tBuLi (2.13 g, 33.3 mmol, 2.1 eq) in hexanes (10 mL), in an inert atmosphere 35 glove box. Both solutions were taken out of the box, and the tBuLi solution was added to the solution of O-Methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide while allowing the latter to thaw. The mixture was stirred for 1 h while allowing it to reach room temperature. ZnCl2 (1.51 g, 7.12 mmol, 0.70 eq) was added with the aid of THF. After stirring the reaction mixture for 30 minutes, 1,3-dibromobenzene (1.69 g, 7.12 mmol, 0.45 eq) and Pd(PPh3 )4 (188.4 mg, 0.163 mmol, 0.01 eq) were added with the aid of some THF. The reaction vessel was placed in an oil bath preheated to 75◦ C and stirred for 16 h, then allowed to cool to room temperature and quenched with water. Volatile materials were removed under vacuum and water was added. This mixture was extracted with Et2 O (three times). The combined organics were dried over MgSO4 , filtered, and concentrated by rotary evaporation. The resulting residue was suspended in MeOH cooled to −25◦ C. The resulting white precipitate was collected by filtration through a sintered glass funnel, washed with cold MeOH, and immediately used for the synthesis of H2 (ONO). 1 H NMR (300 MHz, CDCl3 ): δ 1.37 (s, 18H, C(CH3 )3 ), 1.51 (s, 18H, C(CH3 )3 ), 3.41 (s, 6H, OCH3 ), 4.64 (s, 4H, OCH2 O), 7.45 (d, 2H, arylH), 7.61 (d, 2H, aryl-H), 7.68 − 7.80 (m, 3H, NC5 H3 ). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 31.1 (C(CH3 )3 ), 31.6 (C(CH3 )3 ), 34.8 (C(CH3 )3 ), 35.6 (C(CH3 )3 ), 57.6 (OCH3 ), 99.7 (OCH2 O), 123.2, 125.2, 126.7, 134.1, 136.1, 142.5, 146.1, 151.5, 158.4 (aryl). Synthesis of H2 (ONO) Compound MOM2 (ONO) was suspended in 25 mL of MeOH, and 25 mL of concentrated HCl was added. This mixture was heated at 80◦ C for 5 h. After cooling, volatile materials were removed under vacuum yielding 3.45 g (0.70 mmol, 90% based on O-Methoxymethyl-2-bromo-4,6di-tert-butylphenoxide) of H2 (ONO) as a white powder. 1 H NMR (300 MHz, CDCl3 ): δ 1.39 (s, 18H, C(CH3 )3 ), 1.48 (s, 18H, C(CH3 )3 ), 7.46 (d, 2H, aryl-H), 7.51 (d, 2H, aryl-H), 7.67 (d, 2H, 3 5 −NC5 H−H 2 ), 8.01 (t, 1H, 4- NC5 H2 −H), 10.59 (s, 2H, OH). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 29.8 (C(CH3 )3 ), 31.8 (C(CH3 )3 ), 34.6 (C(CH3 )3 ), 35.6 (C(CH3 )3 ), 120.5, 121.3, 123.0, 126.4, 137.5, 140.0, 141.5, 153.3, 157.6 (aryl). HR-MS: C33 H45 O2 N: Calculated: 487.3450. Measured: 487.3446. 2.5.1.8 Synthesis of MOM2 (ONO tBu ) O-Methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide (6.15 g, 18.7 mmol), was dissolved in 60 mL of THF. tert-butyllithium (2.52 g, 39.4 mmol) was dissolved in 27 mL of pentane and added to the solution of O-methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide at −78◦ C. The mixture was allowed to warm to room temperature and stirred for 1 h. A suspension of ZnCl2 (1.80 g, 13.2 mmol) in THF was then added, and the mixture stirred for another 30 min. A solution of 2,6-dichloro-4tert-butylpyridine (1.72 g, 8.42 mmol) and Pd(PPh3 )4 (218 mg, 189 µmol) in THF was added, the reaction vessel was transferred to a preheated 75◦ C oil bath, and the mixture was stirred for 16 h. The 36 mixture was quenched with 10 mL of water, and the volatiles were removed by rotary evaporation. Another 70 mL of water was then added, and the organic layer was extracted three times with ether, dried with MgSO4 , and evaporated to dryness, giving a white powder, which was purified by washing with cold methanol (2.67 g, 4.23 mmol, 50.3% yield). 1 H NMR (300 MHz, CDCl3 ): δ 1.35 (s, 18H, C(CH3 )3 ), 1.37 (s, 9H, C(CH3 )3 ), 1.49 (s, 18H, C(CH3 )3 ), 3.38 (s, 6H, OCH3 ) 4.60 (s, 4H, OCH2 O), 7.42 (d, 2H, 4 J = 2.6 Hz, aryl-H), 7.56 (d, 2H, 4 J = 2.6 Hz, aryl-H), 7.66 (s, 2H, NC5 H2 ). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 30.6, 30.9, 31.4, 34.6, 34.9, 35.4 (C(CH3 )3 ), 57.3 (OCH3 ), 99.3 (OCH2 O), 120.4, 124.8, 126.6, 134.3, 142.2, 145.8, 151.3, 158.0, 159.7 (aryl). 2.5.1.9 tBu Synthesis of H2 (ONO tBu MOM2 (ONO ) ) (2.67 g, 4.23 mmol) was suspended in a mixture of methanol (40 mL) and con- centrated HCl (40 mL) and heated at 80◦ C for 6 h. The suspension was then cooled to 0◦ C and tBu filtered, and the precipitate, a white powder, was washed with cold MeOH to afford H2 (ONO ) 1 (1.71 g, 3.14 mmol, 74% yield). H NMR (300 MHz, d8 -THF): δ 1.36 (s, 18H, C(CH3 )3 ), 1.46 (br, 27H, C(CH3 )3 ), 7.42 (d, 2H, 4 J = 2.4 Hz, aryl-H), 7.53 (d, 2H, 4 J = 2.7 Hz, aryl-H), 7.81 (s, 2H, NC5 H2 ), 10 (br s, concentration dependent, 2H, OH). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 29.6, 30.6, 31.6, 34.4, 35.3, 35.5 (C(CH3 )3 ), 117.7, 121.7, 122.6, 126.0, 137.2, 141.2, 153.0, 157.2, 164.1 (aryl). MS FAB+: 543.4052 (M+ ). Analysis calculated for C37 H53 NO2 : C, 81.72; H, 9.82; N, 2.58. Found: C, 81.57; H, 10.06; N, 2.55. 2.5.1.10 Synthesis of Na[Ir(cod)ONO], 1 H2 (ONO) (91.2 mg, 187 µmol) was dissolved in 10 mL of THF, and an excess of NaH (24.6 mg, 1.03 mmol) was added. The yellow suspension first turned orange and then yellow again, with vigorous evolution of H2 . The suspension was allowed to stir for one hour, after which it was filtered through Celite and added to a stirring solution of [Ir(cod)Cl]2 (60.0 mg, 89.3 µmol) in 5 mL of THF. After 30 min of stirring the solution was filtered and pumped down, giving 1 as a red powder. Residual solvent was removed by trituration with benzene followed by overnight evacuation under high vacuum. Isolated yield was 148.3 mg (183.3 µmol, 98.0%); the yield was quantitative by NMR. 1 H NMR (300 MHz, d8 -THF): δ 0.84 (m, 4H, sp3 on cod), 1.32 (s, 18H, 2 tBu on ligand phenolate), 1.49 (s, 18H, 2 tBu on ligand phenolate), 1.68 (br m, 4H, sp3 on cod), 2.99 (br d, 4H, J = 2.7 Hz, sp2 on cod), 7.19 (d, 2H, 4 J = 2.7 Hz, ligand aryl), 7.24 (d, 2H, 4 J = 2.6 Hz, ligand aryl), 7.26 (br s, 2H, ligand aryl), 7.66 (t, 1H, 3 J = 8.0 Hz, ligand pyridine 4-position). 13 C{1 H} NMR (75.4 MHz, d8 -THF): δ 14.6, 23.4, 31.5, 32.5, 32.8, 34.7, 36.2, 49.7, 121.3, 123.8, 125.0, 131.9, 135.7, 137.2, 139.2, 159.4, 166.9. MS FAB+: 786.3856 (M+ −Na). 37 2.5.1.11 Synthesis of Pyridine-Activated Complex 3 Na[Ir(cod)(ONO)] (1) (85.1 mg, 105 µmol) was dissolved in 1 mL of C6 H6 . 15-Crown-5 (21 µL, 106 µmol) was mixed in, and the solution was allowed to sit undisturbed for several days. Orange crystals formed and were shown crystallographically to consist of 3. A pure (by NMR) sample was obtained by washing with C6 H6 , redissolving in THF, and drying under vacuum. Isolated yield: 82.7 mg (79.8 µmol, 75.9%). 1 H NMR (500 MHz, d8 -THF): δ 1.35 (s, 9H, tBu on ligand phenolate), 1.42 (s, 9H, tBu on ligand phenolate), 1.51 (br s, 18H, 2 tBu on ligand phenolate), 1.77 (br, 4H, sp3 on cod), 2.18 (br, 4H, sp3 on cod), 3.29 (br, 2H, sp2 on cod), 3.37 (s, 20H, crown ether), 4.46 (br, 2H, sp2 on cod), 7.24 (d, 1H, 4 J = 2.6 Hz, ligand aryl), 7.25 (d, 1H, 4 J = 2.4 Hz, ligand aryl), 7.48 (d, 1H, 3 J = 8.4 Hz, ligand aryl), 7.78 (d, 1H, 4 J = 2.4 Hz, ligand aryl), 7.91 (d, 1H, 3 J = 8.2 Hz, ligand aryl), 8.45 (d, 1H, 4 J = 2.6 Hz, ligand aryl), 18.14 (s, 1H, ligand phenol). 13 C{1 H} NMR (126 MHz, d8 -THF): δ 30.4, 31.1, 31.6, 32.4, 32.9, 33.9, 35.1, 35.4, 36.1, 36.4, 45.1, 70.2, 74.4, 112.5, 120.2, 120.8, 123.0, 124.3, 124.4, 126.9, 135.0, 137.4, 138.5, 138.8, 146.5, 149.7, 152.6, 153.2, 159.7, 162.7. tBu 2.5.1.12 Synthesis of [NaIr(cod)(ONO tBu H2 (ONO )]2 , 4 ) (246.8 mg, 453.8 µmol) was dissolved in 10 mL of THF, and an excess of NaH (64.8 mg, 2.70 mmol) was added. The light yellow solution first turned orange and then yellow again, with vigorous evolution of H2 . The suspension was allowed to stir for an hour, after which it was filtered through Celite. [Ir(cod)Cl]2 (145.2 mg, 216.2 µmol) was dissolved in THF and slowly added to the tBu stirring solution of Na2 (ONO tBu ) in order to prevent formation of overmetalated [Ir(cod)]2 (ONO ), 5. After an hour of stirring the solvent was removed in vacuo, redissolved in hexanes, and filtered, and the solvent removed in vacuo again, giving 4·THF as a red solid with a glassy appearance. The residual THF was removed by adding more hexanes and evacuating to give 4. 4·THF is very soluble in hexanes, whereas 4 is much less so; consequently impurities in 4 can be removed with a hexanes wash. Crystals were obtained by allowing a pentane solution of 4·THF or 4 to evaporate. Isolated yield was 326.4 mg (188.6 µmol, 87.2%). 1 H NMR (300 MHz, d8 -THF): δ 0.83 (br d, 16H, 2 J = 7.5 Hz, sp3 on 2 cod), 1.33 (s, 36H, 4 tBu on ligand phenolate), 1.37 (s, 18H, 2 tBu on ligand pyridine), 1.49 (s, 36H, 4 tBu on ligand phenolate), 2.97 (br s, 8H, sp2 on 2 cod), 7.19 (d, 4H, 4 J = 2.7 Hz, 2 ligand phenolate aryl), 7.23 (d, 4H, 4 J = 3.0 Hz, 2 ligand phenolate aryl), 7.25 (s, 4H, 2 ligand pyridine aryl). 13 C{1 H} NMR (75.4 MHz, d8 -THF): δ 31.6, 32.1, 33.1, 33.5, 35.4, 36.2, 36.9, 50.3, 119.2, 124.4, 125.4, 133.0, 136.2, 139.6, 159.6, 161.2, 167.7. MS FAB+: 842.4478 ([Ir(cod)(ONO tBu (M+ −Na). Analysis calculated for C90 H126 Ir2 N2 Na2 O4 : C, 62.47; H, 7.34; N, 1.62. Found: C, 62.26; H, 7.52; N, 1.35. 38 2.5.1.13 Reaction of 4 with CO tBu [NaIr(cod)(ONO )]2 (179.4 mg, 103.7 µmol) was dissolved in THF in a Schlenk flask. The solution was cooled to −78◦ C, and an atmosphere of CO (about 5 mmol) was introduced into the headspace. The solution was then brought back to room temperature and allowed to stir overnight. Afterward, the solution was pumped down, yielding a yellow-brown solid with IR stretching frequencies at 2057, 2030, and 1977 cm−1 . This solid was reintroduced into the glovebox, redissolved in pentane, and filtered. Allowing the pentane solution to slowly evaporate yielded brown crystals. X-ray diffraction of the crystals revealed their identity to be Ir4 (CO)12 , consistent with the aforementioned IR stretching frequencies. 2.5.1.14 Reaction of 4 with H2 tBu [NaIr(cod)(ONO )]2 (10.2 mg, 5.89 µmol) was dissolved in 0.7 mL of d8 -THF in a J-Young NMR tube. Two drops of benzene were added as an internal standard. The headspace was then charged with 2.9 atm H2 , and the NMR tube continuously inverted for mixing. After 11 h, the solution color had lightened considerably, and an iridium mirror had deposited. 1 H NMR showed quantitative tBu conversion to the protonated ligand H2 (ONO tBu 2.5.1.15 Synthesis of [Ir(cod)]2 (ONO tBu H2 (ONO ). ), 5 ) (57.6 mg, 106 µmol) was dissolved in 5 mL of THF, and an excess of NaH (30.2 mg, 1.26 mmol) was added. The yellow suspension first turned orange and then yellow again, with vigorous evolution of H2 . The suspension was allowed to stir for an hour, after which it was filtered through Celite. A 5 mL THF solution of [Ir(cod)Cl]2 (71.1 mg, 107 µmol) was then added while stirring. After 1 h of stirring, the solvent was removed in vacuo; and the orange compound was redissolved in pentane and filtered. The solvent was then removed in vacuo again, giving 5 as an orange powder. Crystals were obtained by allowing a pentane solution to evaporate. Yield was 119.3 mg (104 µmol, 98.6%). 1 H NMR (300 MHz, d8 -THF): δ 0.81 (br d, 16H, 2 J = 6.9 Hz, sp3 on 2 cod), 1.39 (s, 18H, 2 tBu on ligand phenolate), 1.52 (s, 9H, tBu on ligand pyridine), 1.58 (s, 18H, 2 tBu on ligand phenolate), 3.08 (br s, 8H, sp2 on 2 cod), 7.51 (br d, 4H, J = 5.4 Hz, 2 ligand phenolate aryl), 7.82 (s, 2H, ligand pyridine aryl). 13 C{1 H} NMR (75.4 MHz, d8 -THF): 15.2, 24.1, 31.0, 31.6, 32.65, 32.75, 33.4, 35.1, 35.8, 35.9, 36.9, 38.3, 56.1, 121.7, 124.9, 130.0, 134.9, 139.2, 141.8, 159.3, 163.1, tBu 163.6. MS FAB+: 1141.5018 (M+ with 191Ir), 843.4935 ([Ir(cod)(ONO + )] ). Analysis calculated for C53 H75 Ir2 NO2 : C, 55.71; H, 6.62; N, 1.23. Found: C, 54.81; H, 6.77; N, 1.12. tBu Alternatively, [NaIr(cod)(ONO )]2 (15 mg, 9 µmol) was dissolved in 1 mL of THF. [Ir(cod)Cl]2 (6 mg, 9 µmol) was dissolved in 2 mL of THF and added. After two hours the solution was the solvent was removed in vacuo, redissolved in pentane, and filtered, and the solvent removed in vacuo again, 39 giving 5 (20 mg, 18 µmol, 100% yield) as an orange powder. Characterization data were as reported above. 2.5.1.16 tBu Synthesis of [(ONO tBu [NaIr(cod)(ONO )Ir(cod)(MeCN)][PF6 ], 6 )]2 (106.3 mg, 61.4 µmol) was dissolved in 10 mL of THF. A solution of AgPF6 (59.4 mg, 235 µmol) in acetonitrile was added, resulting in the mixture immediately turning black. The solution was allowed to stir overnight. Afterward, the solvent was removed in vacuo, and the black solid redissolved in benzene, filtered, and lyophilized to give 6 as a yellow powder. This complex was recrystallized by layering pentane onto a benzene solution. Yield after recrystallization was 93.0 mg (90.4 µmol, 73.6%). 1 H NMR (300 MHz, CD3 CN): δ 1.37 (s, 18H, 2 tBu on ligand phenolate), 1.45 (s, 9H, tBu on ligand pyridine), 1.53 (s, 18H, 2 tBu on ligand phenolate), 2.26 (m, 4H, sp3 on cod), 2.42 (s, 3H, MeCN ligand), 2.75 (br s, 2H, sp3 on cod), 4.31 (br s, 2H, sp3 on cod), 5.45 (m, 2H, sp2 on cod), 6.65 (m, 2H, sp2 on cod), 7.41 (two overlapping peaks, s, 4H, 2 ligand phenolate aryl), 7.70 (s, 2H, ligand pyridine aryl). 13 C{1 H} NMR (75.4 MHz, C6 D6 ): δ 1.3, 28.0, 29.3, 29.4, 30.3, 31.2, 33.9, 34.4, 35.5, 100.2, 113.9, 119.9, 120.6, 123.2, 125.8, 126.4, 139.4, 140.7, 154.3, 161.8, 162.1. 19 F{13 C} NMR (282 MHz, C6 D6 ): δ −71.1 (d, 1 JPF = 711 Hz). 31 P{1 H} NMR (122 MHz, C6 D6 ): tBu δ −143.2 (septet, 1 JPF = 711 Hz). MS FAB+: 842.4472 ([Ir(cod)(ONO + )] ). Analysis calculated for C47 H66 F6 IrN2 O2 P · C6 H6 : C, 57.54; H, 6.56; N, 2.53. Found: C, 57.06; H, 6.48; N, 2.42. Alternatively, [NaIr(cod)(ONO tBu )]2 (103.9 mg, 60.0 µmol) was dissolved in 10 mL of THF. A suspension of [FeCp2 ][PF6 ] (75.6 mg, 228 µmol) in acetonitrile was added, the solution was allowed to stir overnight, the solvent was removed in vacuo, and the solid redissolved in benzene, filtered, and lyophilized to give 6. Yield after recrystallization (as above) was 98.7 mg (96.0 µmol, 79.9%). Characterization data were as reported above. 2.5.1.17 tBu Synthesis of (ONO tBu [NaIr(cod)(ONO )Ir(cod)OTf, 7 )]2 (310.0 mg, 179.1 µmol) and AgOTf (175.7 mg, 683.8 µmol) were each dis- solved in about 10 mL of benzene. The AgOTf solution was then added dropwise to the stirring solution of 4. The mixture immediately turned black and was allowed to stir overnight. It was then filtered, pumped down, and washed with pentane, yielding (ONO tBu )Ir(cod)OTf, 7 (259 mg, 261 µmol, 76% yield). Since 7 has limited solubility in benzene (about 10 mg/mL), it can be purified by washing with benzene. Crystals were obtained by layering a saturated benzene solution with pentane. 1 H NMR (300 MHz, C6 D6 ): δ 0.93 (s, 9H, tBu on ligand pyridine), 1.19 (br s, 8H, sp3 on cod), 1.42 (s, 18H, 2 tBu on ligand phenolate), 1.74 (s, 18H, 2 tBu on ligand phenolate), 2.42 (br s, 2H, sp2 on cod), 4.62 (m, 2H, sp2 on cod), 7.35 (s, 2H, ligand pyridine aryl), 7.45 (d, 2H, 4 J = 2.7 Hz, ligand phenolate aryl), 7.58 (d, 2H, 4 J = 2.7 Hz, ligand phenolate aryl). 13 C{1 H} NMR (75.4 MHz, C6 D6 ): δ 27.4, 29.5, 30.0, 30.7, 31.4, 34.0, 34.2, 35.7, 90.7, 112.8, 120.6, 123.3, 125.9, 126.2, 40 139.1, 139.6, 153.2, 160.9, 161.2. 19 F{13 C} NMR (282 MHz, C6 D6 ): δ −77.6. MS FAB+: 991.5552 (M+ ), 858.5524 ([(ONO tBu + )Ir(cod)] ). Analysis calculated for C46 H63 F3 IrNO5 PS: C, 55.74; H, 6.41; N, 1.41. Found: C, 55.51; H, 6.36; N, 1.38. 2.5.1.18 tBu Synthesis of (ONO )Ir(PPh3 )2 OTf, 8 7 (89.4 mg, 90.2 µmol) and PPh3 (47.3 mg, 180 µmol) were each dissolved in benzene and mixed together. The mixture was stirred at 90◦ C for at least 3 h, filtered, and then lyophilized. 8 (117.6 mg, 83.5 µmol, 92.6% yield) was formed as a yellow powder, soluble in benzene and slightly soluble in pentane (about 1 mg/mL). 8 is stable for several days at 90◦ C with little decomposition. 1 H NMR (300 MHz, C6 D6 ): δ 0.94 (s, 9H, tBu on ligand pyridine), 1.25 (s, 18H, 2 tBu on ligand phenolate), 1.35 (s, 18H, 2 tBu on ligand phenolate), 6.52 (s, 2H, ligand pyridine aryl), 6.68 (d, 2H, 4 J = 2.4 Hz, ligand phenolate aryl), 6.87 (br s, 12H, triphenylphosphines), 7.16 (br s, 6H, triphenylphosphines), 7.18 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.63 (br s, 12H, triphenylphosphines). 13 C{1 H} NMR (75.4 MHz, C6 D6 ): δ 29.6, 29.7, 29.9, 31.2, 33.6, 35.7, 119.4, 123.2, 125.0, 125.6, 126−128 (may be obscured by C6 D6 ), 129.6, 134.0, 135.1 (br), 136.7, 140.7, 141.1. 19 F{13 C} NMR (282 MHz, C6 D6 ): tBu δ −76.4. 31 P{1 H} NMR (122 MHz, C6 D6 ): δ −17.2. MS FAB+: 1258.5423 ([Ir(PPh3 )2 (ONO tBu 1145.3984 ([Ir(PPh3 )(ONO + tBu )(OTf)] ), 996.4537 ([Ir(PPh3 )(ONO + )] ), + )] ). Analysis calculated for C74 H81 F3 IrNO5 P2 S: C, 63.14; H, 5.80; N, 1.00. Found: C, 63.24; H, 5.95; N, 0.87. 2.5.1.19 tBu Synthesis of (ONO )Ir(PPh3 )2 OH, 9 8 (20.8 mg, 14.7 µmol) was dissolved in benzene, and 2 µL of H2 O was added via a microliter syringe. The mixture was stirred for one hour, after which proton sponge (3.6 mg, 17 µmol) dissolved in benzene was added. A precipitate was immediately observed, and the mixture was stirred for four hours, pumped down, redissolved in benzene, filtered, and lyophilized, giving 9 (18.8 mg, 14.7 µmol, quantitative yield) as a yellow powder. 9 is slightly soluble in pentane (about 3 mg/mL) and was crystallized by letting a pentane solution evaporate. 1 H NMR (300 MHz, C6 D6 ): δ −1.63 (s, 1 H, IrOH), 1.05 (br, 27H, 2 tBu on ligand phenolate, tBu on ligand pyridine), 1.34 (s, 18H, 2 tBu on ligand phenolate), 6.75 (br, 4H, 2 ligand phenolate aryl), 6.8 − 7.2 (many broad overlapping peaks, 30H, triphenylphosphines), 7.38 (br s, 2H, ligand pyridine aryl). 31 P{1 H} NMR (122 MHz, C6 D6 ): tBu δ −23.2. MS FAB+: 1275.5450 ([Ir(PPh3 )2 (ONO tBu 1013.4492 ([Ir(PPh3 )(ONO + tBu )OH] ), 1258.5423 ([Ir(PPh3 )2 (ONO + )] ), + )OH] ). Analysis calculated for C73 H82 IrNO3 P2 : C, 68.73; H, 6.48; N, 1.10. Found: C, 68.71; H, 6.66; N, 1.10. 2.5.1.20 tBu Synthesis of (ONO )Ir(PPh3 )2 Cl, 10 8 (39.0 mg, 27.7 µmol) was dissolved in 3 mL of dichloromethane and heated to 90◦ C for five hours. The solution was then pumped down, redissolved in benzene, filtered, and lyophilized, giving 10 (35.4 41 mg, 27.4 µmol, 98.7% yield) as a yellow powder. 1 H NMR (300 MHz, CD2 Cl2 ): δ 0.79 (s, 18H, 2 tBu on ligand phenolate), 1.12 (s, 9H, tBu on ligand pyridine), 1.27 (s, 18H, 2 tBu on ligand phenolate), 6.50 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 6.65 (s, 2H, ligand pyridine aryl), 6.93 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 6.9 − 7.3 (many broad overlapping peaks, 30H, triphenylphosphines). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 29.6, 30.4, 31.9, 34.3, 34.9, 35.7, 120.2, 124.5, 125.1, 126.1, 127.4 (br), 127.8 (br), 128.3 (br), 129.7 (br), 133.7 (br), 136.3 (br), 136.8, 141.0, 155.7, 159.1, 170.1. 31 tBu P{1 H} NMR (121 MHz, CD2 Cl2 ): δ −25.1. MS FAB+: 1294.5082 ([Ir(PPh3 )2 (ONO tBu 1257.5300 ([Ir(PPh3 )2 (ONO + tBu )] ), 1031.3262 ([Ir(PPh3 )(ONO + )Cl] ), + )Cl] ). Analysis calculated for C73 H81 ClIrNO2 P2 : C, 67.75; H, 6.31; N, 1.08. Found: C, 68.60; H, 6.84; N, 1.07. 2.5.1.21 tBu Synthesis of (ONO )Ir(PPh3 )2 H, 11 LiOMe (3.9 mg, 0.10 mmol) was dissolved in methanol, and 8 (67.2 mg, 47.7 µmol) was added while stirring. A light yellow filtrate formed overnight and was filtered out. When benzene was added to this filtrate, a black precipitate immediately formed, which was also filtered out. The remaining yellow solution was lyophilized, giving (ONO tBu )Ir(PPh3 )2 H, 11 (37.6 mg, 29.9 µmol, 62.5% yield), as a 1 yellow powder. H NMR (300 MHz, C6 D6 ): δ −15.4 (t, 1H, 2 JPH = 17.8 Hz, IrH), 0.79 (s, 18H, 2 tBu on ligand phenolate), 1.17 (s, 9H, tBu on ligand pyridine), 1.41 (s, 18H, 2 tBu on ligand phenolate), 6.8 − 7.4 (many broad overlapping peaks, 36H, ligand aryl, triphenylphosphine aryl). 13 C{1 H} NMR (75.4 MHz, C6 D6 ): δ 28.4, 30.0, 31.7, 33.7, 34.2, 34.9, 118.6, 123.5, 124.9, 126-128 (may be obscured by C6 D6 ), 128.9 (br), 134.3 (br), 135.2, 154.1, 158.0, 170.3. 31 P{1 H} NMR (122 MHz, C6 D6 ): δ 9.8. tBu MS FAB+: 1260.5502 ([Ir(PPh3 )2 (ONO + tBu )H2 ] ), 997.4542 ([Ir(PPh3 )(ONO + )H] ). 11 was also obtained from the following reactions, although in most cases it was observed only by NMR and could not be isolated cleanly: • 8 (25.7 mg, 18.3 µmol) and proton sponge (3.8 mg, 17.7 µmol) were dissolved in benzene and mixed. The solution was heated to 90◦ C for 24 h. It was then pumped down, redissolved in pentane, filtered, and pumped down again. The yield of 11 was nearly quantitative (by NMR). • 8 (11.3 mg, 7.2 µmol) and dihydroanthracene (1.3 mg, 7.2 µmol) were dissolved in C6 D6 and mixed. CoCp2 (1.4 mg, 7.4 µmol) dissolved in C6 D6 was then added, and the entire mixture transferred to a J-Young NMR tube. After stirring for one hour, NMR revealed the formation of 11 as the major product. • 8 (16.7 mg, 10.7 µmol) was dissolved in approximately 1 mL of C6 D6 in a J-Young NMR tube, and an excess of solid NaH (30 mg, 1 mmol) was added. The NMR tube was heated to 90◦ C, and the reaction was followed by NMR; after 26 h the reaction was complete. The solution was then removed, filtered, and pumped down. Isolated yield was 11.8 mg, 9.4 µmol, 88%. 42 • 8 (38.4 mg, 27.3 µmol) was dissolved in approximately 5 mL of benzene, and ZnEt2 (1.5 µL, 14 µmol) was added via a microliter syringe. The stirring solution first turned red and then yellow and was allowed to stir overnight. The solution was then pumped down, redissolved in pentane, filtered, and pumped down again to form 11 as a clean solid. 2.5.1.22 tBu Synthesis of (ONO )Ir(PPh3 )2 Me, 12 8 (203.8 mg, 144.8 µmol) was dissolved in benzene, and ZnMe2 (6.2 µL, 90 µmol) was added via a microliter syringe. The solution was stirred at room temperature overnight, filtered, and then lyophilized. 12 (182.6 mg, 143.4 µmol, 99.0% yield) was formed as a yellow powder, soluble in benzene and slightly soluble in pentane, and could be recrystallized by layering pentane on a benzene solution. 1 H NMR (300 MHz, C6 D6 ): δ 1.13 (s, 9H, tBu on ligand pyridine), 1.18 (s, 18H, 2 tBu on ligand phenolate), 1.40 (s, 18H, 2 tBu on ligand phenolate), 3.03 (t, 3H, 3 JPH = 6.1 Hz, IrMe), 6.82 (d, 2H, 4 J = 2.4 Hz, ligand phenolate aryl), 6.89 (br s, 12H, triphenylphosphines), 6.92 (s, 2H, ligand pyridine aryl), 7.19 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl), 7.16 (br s, 6H, triphenylphosphines), 7.46 (br s, 12H, triphenylphosphines). 13 C{1 H} NMR (75.4 MHz, CD2 Cl2 ): δ −31.4 (t, 2 JPC = 8.6 Hz), 29.3, 29.9, 31.3, 33.6, 34.3, 35.1, 119.4, 123.3, 125.9, 126.1, 127.1, 128.7, 134.7 (br), 135.1, 140.3, 153.9, 157.8, 170.4. 31 P{1 H} NMR (122 MHz, C6 D6 ): δ −14.6. MS FAB+: 1273.4640 (M+ ), tBu 1011.4714 ([Ir(PPh3 )(ONO + )(Me)] ), 995.3394 ([Ir(PPh3 )(ONO tBu + )] ). Analysis calculated for C74 H84 IrNO2 P2 : C, 69.78; H, 6.65; N, 1.10. Found: C, 69.81; H, 6.53; N, 1.05. 2.5.1.23 tBu Synthesis of (ONO )Ir(PPh3 )2 I, 13 8 (10.5 mg, 8.24 µmol) and I2 (2.1 mg, 8.3 µmol) were each dissolved in benzene and mixed together. The mixture was stirred at room temperature for at least 3 h, filtered, and then lyophilized. 13 (10.8 mg, 7.80 µmol, 94.6% yield) was formed as a yellow, pentane-soluble powder, although impurities can give it a green color. Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (300 MHz, CD2 Cl2 ): δ 0.90 (s, 18H, 2 tBu on ligand phenolate), 1.15 (s, 9H, tBu on ligand pyridine), 1.28 (s, 18H, 2 tBu on ligand phenolate), 6.52 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl), 6.70 (s, 2H, ligand pyridine aryl), 6.96 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.08 (br s, 12H, triphenylphosphines), 7.21 (br s, 12H, triphenylphosphines), 8.16 (br s, 6H, triphenylphosphines). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 30.3, 30.4, 31.8, 34.3, 34.9, 35.7, 120.5, 124.8, 125.1, 126.6, 127.4 (br), 129.8 (br), 133.7 (br), 136.9, 137.3 (br), 141.2, 154.8, 159.2, 170.4. 31 P{1 H} NMR (122 tBu MHz, C6 D6 ): δ −29.1. MS FAB+: 1386.5627 (M+ ), 1123.3712 ([Ir(PPh3 )(ONO 2.5.1.24 tBu Synthesis of (ONO + )(I)] ). )Ir(PEt3 )2 Me, 14 12 (79.4 mg, 62.3 µmol) was dissolved in 5 mL of benzene, and PEt3 (20 µL, 136 µmol) was added. The solution was allowed to stir for seven hours, after which the benzene and excess PEt3 were 43 removed by lyophilization. The yellow powder obtained was redissolved in 5 mL of dichloromethane, and [Rh(cod)2 ]OTf (30.5 mg, 62.7 µmol) dissolved in 3 mL of acetonitrile was added to complex residual PEt3 . The solution was allowed to stir for three hours. Afterward, it was pumped down, and the resulting 14 was redissolved in hexanes, filtered to remove the byproducts and impurities, and pumped down again. Isolated yield was 51.4 mg (52.2 µmol, 83.7%). Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (300 MHz, C6 D6 ): δ 0.75 (m, 18H, 6 P−CH2 −CH 3 ), 1.16 (s, 9H, tBu on ligand pyridine), 1.46 (s, 18H, 2 tBu on ligand phenolate), 1.65 (m, 33H, 2 tBu on ligand phenolate, 6 P−CH 2 −CH3 , IrMe), 7.34 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl), 7.43 (s, 2H, ligand pyridine aryl), 7.53 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl). 13 C{1 H} NMR (126 MHz, C6 D6 ): δ −33.4 (t, JCP = 8.2 Hz), 8.1 (t, JCP = 1.5 Hz), 13.3 (t, JCP = 14.3 Hz), 30.6, 31.3, 32.4, 34.4, 35.3, 36.3, 118.7, 124.0, 125.6, 128.1, 136.2, 140.6, 156.3, 159.3, 172.9 (t, J = 2.2 Hz). 31 P{1 H} tBu NMR (122 MHz, C6 D6 ): δ −19.2. MS FAB+: 985.5609 (M+ ), 867.4800 ([Ir(PEt3 )(ONO + )(Me)] ). Analysis calculated for C50 H84 IrNO2 P2 : C, 60.95; H, 8.59; N, 1.42. Found: C, 61.18; H, 8.32; N, 1.45. 2.5.1.25 tBu Synthesis of (ONO )Ir(PEt3 )2 I, 15 14 (11.8 mg, 12.0 µmol) and I2 (3.0 mg, 12.0 µmol) were each dissolved in benzene and mixed together. The mixture was stirred at room temperature for 5 h, filtered, and then lyophilized. 15 was formed as a yellow, pentane-soluble powder. Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (300 MHz, C6 D6 ): δ 0.74 (m, 18H, 6 P−CH2 −CH 3 ), 1.09 (s, 9H, tBu on ligand pyridine), 1.42 (s, 18H, 2 tBu on ligand phenolate), 1.71 (s, 18H, 2 tBu on ligand phenolate), 1.96 (m, 12H, 6 P−CH 2 −CH3 ), 7.36 (s, 2H, ligand pyridine aryl), 7.37 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.46 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl). 13 C{1 H} NMR (126 MHz, C6 D6 ): δ 8.4 (t, JCP = 1.6 Hz), 15.0 (t, JCP = 15.3 Hz), 30.0, 31.0, 31.9, 34.1, 34.9, 35.8, 118.8, 124.5, 124.7, 126.5, 137.2, 141.3, 156.3 (t, JCP = 1.4 Hz), 159.9, 171.7 (t, JCP = 2.2 Hz). 31 P{1 H} NMR (122 MHz, C6 D6 ): δ −26.3. The reaction of 14 and I2 in dichloromethane gives a new species along with 15 and MeI; over the course of four days all of the 15 is converted to the new species. Crystals were obtained after the contents of the NMR tube were allowed to evaporate. X-ray crystallography, NMR, and MS tBu support identification of this compound as the analogous chloride, (ONO )Ir(PEt3 )2 Cl. 1 H NMR (300 MHz, CD2 Cl2 ): δ 0.74 (m, 18H, 6 P−CH2 −CH 3 ), 1.01 (m, 12H, 6 P−CH 2 −CH3 ), 1.33 (s, 18H, 2 tBu on ligand phenolate), 1.37 (s, 18H, 2 tBu on ligand phenolate), 1.38 (s, 9H, tBu on ligand pyridine), 7.11 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.26 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.31 (s, 2H, ligand pyridine aryl). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 7.7, 13.0 (t, JCP = 15.1 Hz), 30.5, 30.6, 31.9, 34.4, 35.5, 35.9, 119.4, 124.6, 124.8, 126.6, 137.6, 140.3, 157.0, 160.5, tBu 171.7. 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −20.7. MS FAB+ 970.5417 ([Ir(PEt3 )2 (ONO tBu 887.3802 ([Ir(PEt3 )(ONO + tBu )(Cl)] ), 852.4370 ([Ir(PEt3 )(ONO + )] ). + )] ), 44 2.5.1.26 tBu Synthesis of [(ONO )Ir(PEt3 )2 Me]PF6 , 16 14 (31.8 mg, 32.3 µmol) was dissolved in dichloromethane, and a dichloromethane suspension of AgPF6 (8.1 mg, 32 µmol) was added. The mixture was allowed to stir for one hour before being pumped down. It was then redissolved in a small amount of benzene and filtered. The solution was then layered with pentane and allowed to sit overnight. Dark blue needle-like crystals of 16·2 C6 H6 (39.1 mg, 30.4 µmol, 94.1% yield) were recovered. 1 H NMR (300 MHz, CD2 Cl2 ): δ −3.1 (very broad), −2.0 (broad), 0.73, 3.99, 12.0 (very broad). 19 F{13 C} NMR (282 MHz, CD2 Cl2 ): δ −73.8 (d, 1 JPF = 710 Hz). 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −144.7 (septet, 1 JPF = 710 Hz). MS tBu FAB+: 985.5561 (M+ ), 867.4697 ([Ir(PEt3 )(ONO 2.5.1.27 tBu Synthesis of (ONO + )(Me)] ). )Ir(PPh3 )2 Ph, 17 12 (20.4 mg, 16.0 µmol) was dissolved in 10 mL of benzene and stirred in a 65◦ C oil bath for 24 h. The solution was reintroduced into the glovebox, and the solvent was removed. The residue was redissolved in pentane and filtered, and the solution pumped down to obtain 17 (20.9 mg, 15.6 µmol, 98% yield). Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (300 MHz, CD2 Cl2 ): δ 0.70 (s, 18H, 2 tBu on ligand phenolate), 1.20 (s, 9H, tBu on ligand pyridine), 1.43 (s, 18H, 2 tBu on ligand phenolate), 5.83 (br s, 2H, IrPh), 6.54 (br s, 1H, IrPh), 6.62 (d, 2H, 4 J = 2.7 Hz, ligand phenolate aryl), 6.78 (s, 2H, ligand pyridine aryl), 6.8 − 7.1 (multiple broad peaks, 30H, triphenylphosphines), 7.16 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl), 7.29 (br s, 2H, IrPh). 13 C{1 H} NMR (126 MHz, C6 D6 ): δ 30.3, 30.6, 32.2, 34.2, 34.4, 35.9, 120.4, 123.6, 124.1, 125.6, 125.9, 126.9, 127.0, 129.6, 129.7, 133.3 (br), 135.7, 136.5 (br), 141.2, 145.9, 155.3, 157.7, 171.5. 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −13.2. 2.5.1.28 Reaction of 12 with Toluene 12 (7.8 mg, 6.1 µmol) was dissolved in 700 µL of toluene-d8 in a J-Young NMR tube. The tube was heated at 65◦ C and periodically monitored by 1 H NMR until all starting material signals had disappeared, which took six days; the final 1 H NMR spectrum was complex. The mass spectrum of the solid obtained by evaporation exhibited a signal indicating the presence of the analogous tolyl complex (1094.5436, [Ir(PPh3 )(ONO tBu + + )(C6 D4 CD3 ] , [M−PPh3 ] ), while the 31 P{1 H} NMR of the dissolved solid showed two approximately equal signals at δ −12.2 and −12.3, tentatively assigned to meta and para isomers. A much weaker signal at −12.8 may indicate formation of a very small amount of ortho isomer. Isolation of a pure product was not achieved. 45 2.5.1.29 tBu Synthesis of (ONO )Ir(PPh3 )[κ2 -(o-C6 H4 )PPh2 ], 18 12 (8.2 mg, 6.4 µmol) was dissolved in 1 mL of CD2 Cl2 in a J-Young NMR tube and submerged in a 65◦ C oil bath for 21 h. After NMR confirmation that the reaction was complete, the solvent was removed to obtain 18 (7.7 mg, 6.1 µmol, 95% isolated yield). Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (500 MHz, CD2 Cl2 ): δ 0.47 (s, 18H, 2 tBu on ligand phenolate), 1.27 (s, 18H, 2 tBu on ligand phenolate), 1.34 (s, 9H, tBu on ligand pyridine), 6.64 (m, 4H, triphenylphosphines), 6.9 − 7.4 (many multiplets, 23H, triphenylphosphines), 7.50 (dd, 3 J = 4.4 Hz, 3 J = 6.6 Hz, 2H, triphenylphosphines). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 29.58, 30.76, 32.11, 34.24, 35.27, 35.32, 119.51, 119.69, 122.48, 122.55, 123.83, 124.00, 125.60, 125.77, 128.32, 128.39, 129.03, 129.54, 133.76, 133.83, 136.03, 139.93, 140.51, 140.61, 152.80, 153.23, 154.77, 159.36, 170.44. 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −69.5 (d, 2 JPP = 448 Hz), −2.5 (d, 2 JPP = 448 Hz). tBu MS FAB+: 1257.5292 (M+ ), 995.4405 ([Ir(PPh3 )(ONO 2.5.1.30 tBu Synthesis of [(ONO + )] ). )Ir(PPh3 )(CH2 CN)]2 , 19 12 (13.1 mg, 10.3 µmol) was dissolved in 1 mL of CD2 Cl2 in a J-Young NMR tube, and 40 µL of acetonitrile was added. The tube was then submerged in a 65◦ C oil bath for 72 h. After NMR confirmation that the reaction had completed, [Rh(cod)2 ]OTf (2.6 mg, 5.3 µmol) dissolved in 1 mL of acetonitrile was added. The solution was allowed to stir for three hours. Afterward, it was pumped down, and the resulting 19 was redissolved in hexanes, filtered to remove the byproducts and impurities, and pumped down again. Crystals were obtained by allowing a pentane solution to evaporate. Isolated yield was 91% (9.7 mg, 4.7 µmol). 1 H NMR (500 MHz, CD2 Cl2 ): δ 1.23 (s, 9H, 2 tBu on ligand), 1.25 (s, 9H, 2 tBu on ligand), 1.43 (s, 9H, 2 tBu on ligand), 1.44 (s, 9H, 2 tBu on ligand), 1.55 (s, 9H, 2 tBu on ligand), 1.87 (s, 2H, Ir−CH2 −CN), 6.4-6.5 (m, 3H, triphenylphosphine), 6.7-6.8 (m, 3H, triphenylphosphine), 6.9-7.1 (m, 6H, triphenylphosphine),7.2-7.5 (m, 6H, triphenylphosphine), 7.6-7.8 (m, 2H, triphenylphosphine), 7.9-8.1 (m, 1 H, triphenylphosphine). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 3.4, 30.1, 30.3, 30.5, 31.9, 32.2, 34.2, 34.6, 35.2, 35.7, 35.9, 116.0, 118.0, 120.2, 122.5, 123.5, 125.6, 126.0, 128 − 130 (br), 135.0, 138.0, 138.7, 141.1 (d, JCP = 3.8 Hz), 153.9, 154.4, 158.2, 159.8 (d, JCP = 2.7 Hz), 161.1, 167.6 (d, JCP = 3.4 Hz), 170.8 (d, JCP = 1.0 Hz). 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −88.3. 2.5.1.31 Kinetic Experiments Kinetic experiments were carried out with 8.8 mM solutions of 12 or 14. In a typical example, 7.8 mg (6.1 µmol) 12 was dissolved in 700 µL of C6 D6 inside a nitrogen-filled glovebox, and the solution was transferred to a J-Young NMR tube. The NMR tube was then removed from the glovebox and fully submerged in a 65◦ C ethylene glycol bath. The reactions were periodically halted by transferring the 46 Entrya a b c d e f g h i j k l m eq PPh3 b 0.000 0.125 0.250 0.375 0.500 0.500 0.625 0.750 0.750 0.875 1.000 1.125 1.250 [PPh3 ]c 0 1.1 2.2 3.3 4.4 4.4 5.4 6.5 6.5 7.6 8.7 9.8 10.9 d kobs 2.18 × 10−4 1.54 × 10−5 1.19 × 10−5 6.27 × 10−6 5.28 × 10−6 5.77 × 10−6 4.14 × 10−6 2.58 × 10−6 3.14 × 10−6 2.32 × 10−6 1.88 × 10−6 1.90 × 10−6 1.70 × 10−6 R2 0.9937 0.9965 0.9842 0.9964 0.9910 0.9921 0.9877 0.9768 0.9912 0.9904 0.9821 0.9558 0.9777 Table 2.1. Summary of the results of Figure 2.23, used to create Figure 2.21. a) The particular run is detailed in the corresponding section in Figure 2.23; b) equivalents of PPh3 added; c) concentration of added PPh3 in mM. d) the kobs value, in s−1 , calculated by fitting the curve to the first-order rate equation [12] = [12]0 exp(−kobs t). NMR tube into an ice bath before NMR measurements. The progress of the reactions were measured by 31 P NMR integration, with NOE turned off and the d1 time set to 22.895 s (12 has the longest relaxation time, 4.579 s, of all the relevant iridium compounds). For PPh3 inhibition experiments, the Ir complex was mixed with various amounts of a 20 mg/mL stock solution of PPh3 in C6 D6 , and additional C6 D6 was added to bring the volume up to 700 µL. 2.5.2 Kinetics Details Figure 2.21 was constructed from the data in table 2.1. The entries in this table were obtained from individual runs detailed in Figure 2.23. 2.5.3 Crystallographic Data Crystallographic data for the complexes 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, and tBu (ONO )Ir(PEt3 )2 Cl are found in the following tables 2.2, 2.3, 2.4, 2.5, and 2.6. 2.6 Acknowledgments We thank Dr. Michael W. Day and Lawrence M. Henling (Caltech) for X-ray crystallographic studies. The Bruker KAPPA APEXII X-ray diffractometer was purchased via an NSF CRIF:MU award to the California Institute of Technology, CHE-0639094. This work was supported by BP in the MC2 and XC2 programs. 47 Figure 2.23. Conversion of 12 to 17-d6 in C6 D6 with varying amounts of extra PPh3 added. [12]0 = 8.7 mM. a) [PPh3 ]0 = 0 mM, b) [PPh3 ]0 = 1.1 mM, c) [PPh3 ]0 = 2.2 mM, d) [PPh3 ]0 = 3.3 mM, e) [PPh3 ]0 = 4.4 mM, f) [PPh3 ]0 = 4.4 mM, g) [PPh3 ]0 = 5.4 mM, h) [PPh3 ]0 = 6.5 mM, i) [PPh3 ]0 = 6.5 mM, j) [PPh3 ]0 = 7.6 mM, k) [PPh3 ]0 = 8.7 mM, l) [PPh3 ]0 = 9.8 mM, m) [PPh3 ]0 = 10.9 mM. a b c d e f g h i j k l m 48 Empirical formula Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.2. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari 4 C90 H126 N2 O4 Ir2 ·2 Na · C4 H10 O 1804.43 Monoclinic 5 C53 H75 NO2 Ir2 1142.54 Monoclinic 6 [C47 H66 N2 O2 Ir] [PF6 ] · C6 H6 1106.30 Triclinic 7 C46 H63 F3 NO5 SIr ·1.5(C6 H6 ) 1108.40 Triclinic 29.8716(11) 14.4205(5) 24.4163(15) 90 123.7310(10) 90 8747.0(7) C2/c 4 1.370 3.100 100(2) 9.5497(4) 22.3182(9) 22.9384(10) 90 92.086(2) 90 4885.7(4) P 21 /n 4 1.553 5.481 100(2) 11.0592(5) 11.8291(5) 20.4215(9) 105.154(2) 97.156(2) 95.370(2) 2536.27(19) P1 2 1.449 2.727 100(2) 11.7099(4) 13.1880(5) 17.2153(6) 86.059(2) 72.263(2) 87.034(2) 2524.87(16) P1 2 1.458 2.744 100(2) 1.66 to 35.19 15255 1.82 to 35.82 19264 1.80 to 40.84 26070 1.55 to 43.59 34153 485 3.747 0.003 533 1.885 0.007 874 1.957 0.016 883 1.536 0.015 0.0314; 0.0719 0.0355; 0.0535 0.0206; 0.0392 0.0221; 0.0381 Semi-empirical from equivalents 0.7469; 0.5672 2.370; −1.617 Semi-empirical from equivalents 0.7470; 0.5076 2.894; −2.135 Semi-empirical from equivalents 0.7480; 0.5945 2.557; −1.524 Semi-empirical from equivalents 0.7486; 0.6139 2.541; −1.556 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 49 Empirical mula for- Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.3. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari 8 C74 H81 F3 NO5 P2 SIr · 0.53(C5 H12 ) ·1.47(C6 H6 ) 1560.64 Monoclinic 9 C73 H82 NO3 P2 Ir 10 C73 H81 NO2 P2 Cl Ir · 5 (CH2 Cl2 ) 11 C73 H82 NO2 P2 Ir ·C5 H12 1275.54 Monoclinic 1718.61 Monoclinic 1331.68 Monoclinic 12.9942(5) 22.9825(9) 25.8969(9) 90 103.513(2) 90 7519.7(5) P 21 /n 4 1.379 1.906 100(2) 13.0223(4) 21.0636(7) 22.7792(8) 90 99.1960(10) 90 6167.9(4) P 21 /n 4 1.374 2.265 100(2) 19.3956(9) 13.1877(6) 31.5557(15) 90 106.611(2) 90 7734.6(6) P 21 /c 4 1.476 2.194 100(2) 16.8289(8) 23.8694(10) 17.4749(8) 90 102.373(2) 90 6856.5(5) P 21 /n 4 1.290 2.040 100(2) 1.84 to 31.76 18398 2.41 to 29.67 14018 1.35 to 37.99 32658 1.47 to 33.55 19762 966 2.750 0.002 1049 1.713 0.004 871 2.152 0.011 1086 2.140 0.004 0.0463; 0.0958 0.0238; 0.0373 0.0340; 0.0589 0.0310; 0.0519 Semi-empirical from equivalents 0.7463; 0.6010 3.915; −2.838 Semi-empirical from equivalents 0.7459; 0.6381 2.894; −0.696 Semi-empirical from equivalents 0.7474; 0.6430 3.136; −2.303 Semi-empirical from equivalents 0.7466; 0.6219 2.907; −2.314 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 50 Empirical formula Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.4. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari 12 C74 H84 NO2 P2 Ir 14 C50 H84 NO2 P2 Ir 1273.56 Monoclinic 13 C73 H81 INO2 P2 Ir ·2 (C5 H12 ) 1529.72 Monoclinic 985.32 Monoclinic 15 0.67(C49 H80 INO2 P2 Ir); 0.33(C50 H83 NO2 P2 Ir) 1097.19 Monoclinic 13.1981(7) 20.9991(9) 22.8801(12) 90 100.388(3) 90 6237.2(5) P 21 /n 4 1.356 2.239 100(2) 14.4250(5) 19.4180(6) 26.3735(9) 90 99.362(2) 90 7288.9(4) P 21 /c 4 1.394 2.346 100(2) 12.2897(7) 35.3664(19) 11.4244(6) 90 93.830(2) 90 4954.4(5) P 21 /c 4 1.321 2.796 100(2) 12.2460(6) 35.7358(18) 11.4544(6) 90 93.490(2) 90 6856.5(5) P 21 /c 4 1.457 3.386 100(2) 1.66 to 29.33 10617 1.57 to 28.67 15318 1.76 to 27.89 10841 1.87 to 40.01 25951 737 1.268 0.003 848 2.742 0.002 527 1.824 0.003 537 1.448 0.010 0.0387; 0.0478 0.0482; 0.0750 0.0188; 0.0366 0.0344; 0.0620 Semi-empirical from equivalents 0.7459; 0.6362 1.724; −1.487 Semi-empirical from equivalents 0.4314; 0.3230 1.880; −2.150 Semi-empirical from equivalents 0.7456; 0.6727 0.633; −0.778 Semi-empirical from equivalents 0.7479; 0.6295 2.448; −2.844 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 51 Empirical formula Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.5. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari 16 [C50 H84 NO2 P2 Ir] [PF6 ] · 1.5(C6 H6 ) 1249.08 Triclinic 17 C79 H86 NO2 P2 Ir ·2 (C6 H6 ) 745.92 Orthorhombic 18 C73 H80 NO2 P2 Ir 1257.52 Monoclinic 19 C114 H136 Li2 N4 O4 P2 Ir2 2086.49 Triclinic 21.3366(12) 21.4842(13) 26.6266(15) 81.836(4) 85.844(3) 89.459(4) 12050.2(12) P1 8 1.377 2.354 100(2) 16.0771(8) 20.5290(10 23.1672(11) 90 90 90 7646.3(6) P bcn 8 1.296 1.837 100(2) 41.3489(19) 18.8871(9) 20.6067(9) 90 111.731(2) 90 14949.3(12) C2/c 8 1.117 1.867 100(2) 16.4557(8) 17.2824(8) 21.7560(10) 93.951(2) 96.184(3) 101.190(3) 6008.7(5) P1 2 1.153 2.284 100(2) 1.66 to 27.99 37498 1.76 to 37.38 13314 1.47 to 32.01 20406 2.34 to 34.09 30107 2386 1.780 0.003 417 3.326 0.001 758 2.133 0.008 1176 2.866 0.007 0.0739; 0.1072 0.0443; 0.0804 0.0311; 0.0569 0.0335; 0.0589 Semi-empirical from equivalents 0.8340; 0.6254 2.889; −2.188 Semi-empirical from equivalents 0.7473; 0.6479 4.097; −2.794 Semi-empirical from equivalents 0.7463; 0.6478 1.457; −0.873 Semi-empirical from equivalents 0.7461; 0.6684 1.933; −1.882 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 52 tBu Empirical formula Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.6. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari (ONO )Ir(PEt3 )2 Cl C49 H81 ClNO2 P2 Ir 1005.74 Monoclinic 12.2623(6) 35.4340(17) 11.3773(6) 90 93.368(2) 90 4934.9(4) P 21 /c 4 1.354 2.861 100(2) 1.88 to 39.87 26158 526 2.079 0.003 0.0371; 0.0602 Semi-empirical from equivalents 0.7478; 0.5515 2.623; −2.981 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 53 2.7 References [1] (a) Labinger, J. A.; Bercaw, J. E. Nature 2002, 417, 507–514; (b) Crabtree, R. H. J. Chem. Soc., Dalton Trans. 2001, 2437–2450; (c) Labinger, J. A. Fuel Proc. Technol. 1995, 42, 325–338; (d) Labinger, J. A. J. Mol. Catal. 2004, 220, 27–35. [2] (a) Fekl, U.; Goldberg, K. I. Adv. Inorg. Chem. 2003, 54, 259–320; (b) Lersch, M.; Tilset, M. Chem. Rev. 2005, 105, 2471–2526. [3] (a) Goldshlegger, N. F.; Tyabin, M. B.; Shilov, A. E.; Shteinman, A. A. Zh. Fiz. Khim. 1969, 43, 2174–2175; (b) Goldshlegger, N. F.; Eskova, V. V.; Shilov, A. E.; Shteinman, A. A. Zh. Fiz. Khim. 1972, 46, 1353–1354; (c) Shilov, A. E.; Shul’pin, G. B. Activation and Catalytic Reactions of Saturated Hydrocarbons in the Presence of Metal Complexes; Kluwer Academic Publishers: Dordrecht, 2000; pp. 259– 317. [4] Periana, R. A.; Taube, D. J.; Evitt, E. R.; Loffler, D. G.; Wentrcek, P. R.; Voss, G.; Masuda, T. Science 1993, 259, 340–343. [5] Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154–162. [6] (a) Wong-Foy, A. G.; Bhalla, G.; Liu, X. Y.; Periana, R. A. J. Am. Chem. Soc. 2003, 125, 14292–14293; (b) Oxgaard, J.; Muller, R. P.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2004, 126, 352–363; 54 (c) Bhalla, G.; Liu, X. Y.; Oxgaard, J.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 11372–11389; (d) Tenn, W. J.; Young, K. J. H.; Bhalla, G.; Oxgaard, J.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 14172–14173; (e) Bhalla, G.; Periana, R. A. Angew. Chem., Int. Ed. 2005, 44, 1540–1543; (f) Young, K. J. H.; Mironov, O. A; Periana, R. A. Organometallics 2007, 26, 2137–2140; (g) Young, K. J. H.; Oxgaard, J.; Ess, D. H.; Meier, S. K.; Stewart, T.; Goddard, W. A.; Periana, R. A. Chem. Commun. 2009, 3270–3272. [7] (a) Agapie, T.; Day, M. W.; Bercaw, J. E. Organometallics 2008, 27, 6123–6142; (b) Agapie, T.; Henling, L. H.; DiPasquale, A. G.; Rheingold, A. L.; Bercaw, J. E. Organometallics 2008, 27, 6245–6256. [8] Weinberg, D. R.; Hazari, N.; Labinger, J. A.; Bercaw, J. E. Organometallics 2010, 29, 89–100. [9] The synthesis of H2 (ONO) was first reported in: Agapie, T. Ph.D. Thesis, California Institute of Technology, Pasadena, CA, 2007, pp. 156–159, 214–215. [10] Danopoulos, A. A.; Winston, S.; Hursthouse, M. B. J. Chem. Soc., Dalton Trans. 2002, 3090– 3091. [11] Hughes, R. P.; Kovacik, I.; Lindner, D. C.; Smith, J. M.; Willemsen, S.; Zhang, D.; Guzei, I. A.; Rheingold, A. L. Organometallics 2001, 20, 3190–3197. [12] Recent calculations have indicated that β-hydride elimination from some methoxide complexes may be energetically disfavored: Theophanis, P. L.; Goddard, W. A. Organometallics 2011, 30, 4941–4948, although there are many examples in the literature. [13] Lu, Z; Jun, C-H; de Gala, S. R.; Sigalas, M. P.; Eisenstein, O; Crabtree, R. H. Organometallics 1995, 14, 1168–1175. [14] The crystal structure included a complex solvent region that could not be satisfactorily modeled and also showed the presence of Li+ ions, presumably carried along as impurities from an earlier step in the overall synthetic sequence; the required charge-balancing counteranions could not be located, probably because they lie within the solvent region. However, determination of the molecular structure of the Ir2 complex 19 was unproblematic. [15] A somewhat related situation has been described previously: Luecke, H. F.; Bergman, R. G. J. Am. Chem. Soc. 1997, 119, 11538–11539. The cyclometalated IrIII complex Cp*(OTf)Ir(κ2 -PPh2 (o-C6 H4 )), obtained from the room-temperature reaction of 55 Cp*(OTf)Ir(PPh3 )Me in CH2 Cl2 , reacts with benzene to form Cp*(OTf)Ir(PPh3 )Ph. Here however the intermolecular activation product is thermodynamically preferred; probably the difference between the two systems is primarily of steric origin, as both include ligands (tertbutyl substituents on ONO; the large Cp* group) that can lead to severe crowding. [16] Golden, J. T.; Andersen, R. A.; Bergman, R. G. J. Am. Chem. Soc. 2001, 123, 5837–5838. [17] (a) Stour, D. L.; Zaric, S.; Niu, S. Q.; Hall, M. B. J. Am. Chem. Soc. 1996, 118, 6068–6069; (b) Su, M. D.; Chu, S. Y. J. Am. Chem. Soc. 1997, 119, 5373–5383. [18] Pangborn, A. B.; Giardello, M. A.; Grubbs, R. H.; Rosen, R. K.; Timmers, F. J. Organometallics 1996, 15, 1518–1520. [19] Ko, P.-H.; Chen, T.-Y.; Zhu, J.; Cheng, K.-F.; Peng, S.-M.; Che, C.-M. J. Chem. Soc., Dalton Trans. 1995, 2215–2220. 56 Chapter 3 DFT virtual screening identifies rhodium-amidinate complexes as active homogeneous catalysts for methane to methanol oxidation Reproduced in part with permission from Fu, R.; Nielsen, R. J.; Goddard, W. A. III; Fortman, G. C.; Gunnoe, T. B. ACS Catalysis, submitted for publication. Unpublished work c 2014 American Chemical Society. Also includes original work. 3.1 Abstract In the search for new organometallic catalysts for the low temperature selective conversion of CH4 to CH3 OH, we apply quantum mechanical virtual screening to select the optimum combination of ligand and solvent on rhodium to achieve low barriers for CH4 activation and functionalization to recommend for experimental validation. We report quantum mechanical predictions (including implicit and explicit solvation) of the mechanisms for various bidentate and tridentate RhIII complexes to catalytically activate and functionalize methane, using trifluoroacetic acid (TFAH) or F water as a solvent. Our most notable results include the design of a RhIII (NN ) ((NNF = bis(N pentafluorophenyl) pentafluorobenzylamidinate) complex, with a methane activation transition state barrier of ∆G‡ = 27.6 kcal/mol in TFAH, and a RhIII (bisq) ((bisq) = bis(quinolinyl)benzene) complex, with a methane activation transition state barrier of ∆G‡ = 33.4 kcal/mol in TFAH. To close the catalytic cycle, the functionalization of methylrhodium intermediates was also investigated, involving carbon-oxygen bond formation via SN 2 attack by solvent, or SR 2 attack by a vanadium oxo. In addition, we have found a correlation between CH4 activation barriers and rhodium-methyl bond energies that allow us to predict the activation transition state energies for future ligands as well. 57 3.2 Introduction The facile, selective, and direct conversion of methane into methanol has long been a goal of industrial chemists [1]. Molecular compounds in solution have long been investigated as potential catalysts for this transformation, as they are typically well-defined, easy to characterize and model, and operate under relatively mild conditions. Whereas much previous work has been done on Pt and Pd catalysts [2, 3], in this chapter we focus on rhodium due to its well-documented nature as an effective C−H activating metal [4] and whose lower electronegativity may allow it to avoid poisoning by coordinating media. Our initial calculations found that using neutral tridentate pincer ligands (L3 ) resulted in neutral L3 Rh(TFA)3 resting states. In order for such complexes to activate methane, a TFA ligand must be both protonated and removed to create an open coordination site. The frequent result was that the overall methane activation energy (i.e. energy of the transition state minus energy of the resting state) was too high to be feasible. Thus, to provide the best candidates for experiments, we shifted our attention to bidentate and monoanionic ligands, which are expected to favor RhIII states that would incorporate a labile, protonated TFAH solvent molecule, thus removing the extra energy penalty for protonation. Bidentate and tridentate ligands were chosen to facilitate redox processes which interconvert square-planar and octahedral coordination environments. Thus we initially examined four classes of ligands that we thought might be prove effective while likely not too hard to synthesize: • bis(N -phenyl)benzylamidinate (NN); • (N -phenyl)acetaldiminyl quinolate (ONN); • bis(pyridyl)(m)ethanesulfonate (DP M E S); • bis(pyrrolyl)quinolinyl phosphine (PN); and two solvents: water and trifluoroacetic acid, due to their pH range and oxidative stability. We then embarked on quantum mechanical (QM) virtual screening to select the optimum combinations of ligand and solvent to recommend for experimental validation. Here we used density functional theory (DFT) at the B3LYP and M06 levels, including both implicit and explicit solvation, for systematic searches over possible reactions mechanisms for CH4 activation and for functionalization. We found that rhodium complexed with the bidentate bis(N -phenyl) benzylamidinate (NN) ligand was the most promising in our initial screen. We then designed a new ligand, bis(N -pentafluorophenyl) pentafluorobenzylamidinate (NNF ), essentially an electron-poor version of (NN), that we found to be even more promising, with lower transition state barriers for both methane activation and Rh−Me functionalization. In addition, we also designed more and less electron-donating variations of the (ONN) ligand; and broadened our investigation to a fifth ligand class, bis(quinolinyl)benzene (bisq), that was expected to have further reduced functionalization barriers. 58 Our efforts culminated in two complexes, F • RhIII (bis(N -pentafluorophenyl) pentafluorobenzylamidinate), denoted as RhIII (NN ); and • RhIII (bis(quinolinyl) benzene), denoted as RhIII (bisq); that can catalytically activate and functionalize methane, using TFAH or water as a solvent. The F first case, RhIII (NN ), leads to a transition state barrier of ∆G‡ = 27.6 kcal/mol at 298 K for methane activation in TFAH (35.0 kcal/mol in water), the lowest we have found using RhIII . The barrier for functionalization is ∆G‡ = 36.8 kcal/mol at 298 K for TFAH (29.7 kcal/mol in water). The second case, RhIII (bisq), leads to a transition state barrier of ∆G‡ = 33.4 kcal/mol at 298 K for methane activation in TFAH, and ∆G‡ = 32.0 kcal/mol at 298 K for Rh−Me functionalization in TFAH, also among the lowest we have found using RhIII . For RhIII complexes with neutral transition states, we find that increasing the electron-withdrawing nature of the ligands decreases barriers for both activation and functionalization. Thus, the only condition opposing even less donating ligands (i.e. lower barriers) is catalyst stability. This chapter should be of great interest to experimentalists who can now focus on these two ligands in TFAH and water to validate and further optimize these systems. 3.3 Materials and Methods All quantum mechanical calculations were carried out using the Jaguar software version 7.6 developed by Schrödinger Inc. [5]. Geometry optimizations were carried out on initial guess structures, and vibrational frequencies were calculated to confirm the optimized geometries as intermediates or transition states and to construct a free energy profile. Solvation energies were calculated using the PBF Poisson-Boltzmann implicit continuum solvation model [6] in Jaguar, with a dielectric constant of 8.55 and a probe radius of 2.451 Å for TFAH and 80.37 and 1.40 Å for water. Explicit waters F were added into the calculations of aqueous Rh(NN ) (two explicit waters per aqua ligand) for more accurate solvation modeling. Geometry optimization and vibrational data were calculated using the B3LYP density functional [7] with a smaller basis set, whereas single point gas-phase and solvated energies were calculated using the M06 functional [8] and a larger basis set. Here the “smaller basis set” consists of a modified double-ζ Los Alamos basis set and pseudopotential [9] that includes f functions for rhodium [10], and the 6-31G** basis set [11] for the other atoms; whereas the “larger basis set” consists of the triple-ζ Los Alamos basis set and pseudopotential (LACV3P**++) modified to include f functions and diffuse functions for rhodium, and the 6-311G**++ basis set [12] for the other atoms. For orbital analysis, the Pipek-Mezey localization procedure was used [13]. 59 Rather than specify a particular chemical oxidant, we adopted a consistent electrostatic potential for electrons to determine the free energy changes of redox reactions. A value of 1.23 V vs. SHE, the standard potential for the reduction of oxygen at pH = 0 and 25◦ C (O2 (g) + 4 H+ (aq) + 4 e– (aq) −−→ 2 H2 O(`)), was adopted for models in trifluoroacetic acid. For water, the pH was taken as 7 and the potential thus reduced to 0.817 V; this was calculated using the equation E = E ◦ − ∆G/nF where F is Faraday’s constant and ∆G = kT ln Q = kT · pH · ln 10. The free energy of the electron was then calculated using the equation G = −F(E + 4.28 V) where 4.28 V represents the absolute potential of the SHE reference, yielding electron free energies of −127.1 and −117.5 kcal/mol, respectively [14]. The free energy of the proton was taken as −260 kcal/mol in TFAH and −279.80 kcal/mol in pH = 7 water [15]. The free energy for each molecular species in solution was calculated using the formula G = Egas + ∆Gsolv + ZPE + Hvib + 6kT − T [Svib + 0.54(Strans + Srot − 14.3 e.u.) + 7.98 e.u.] where the last term is an empirical approximation for the change in the translational and rotational entropy of the molecule between the gas phase and the solution phase (due to the finite librational frequencies) derived from Wertz [16]. For gas phase molecules (methane, methanol at 498.15 K, and for the purposes of this study methyl trifluoroacetate), we assumed that equilibration between the dissolved gas and the headspace occurred at a much faster timescale than the reactions in question; leading to ∆Ggas→solv = 0. Thus, the free energy of such gas molecules can be simply calculated using the formula G = Ggas = Egas + ZPE + Htot − T Stot . We simulated methanol in the gas phase at 498.15 K and the solution phase at room temperature by adding the empirical hydration free energy of −3.2 kcal/mol [17] to the gas phase free energy. For pure liquids (e.g., trifluoroacetic acid and water), the Gibbs free energy was calculated using the formula Gliquid = Egas + ZPE + Htot − T Stot + ∆Ggas→liquid where ∆Ggas→liquid = Gliquid − Ggas (1 atm) is the free energy of condensation to liquid from 1 atm gas. We can solve for this by noting that ∆Ggas→liquid = ∆Gexp + ∆Ggas→solv where ∆Gexp = Ggas (P ) − Ggas (1 atm) is the expansion of the gas from 1 atm to the vapor pressure P , and ∆Ggas→solv is the condensation of gas to liquid. Since a liquid is by definition at equilibrium 60 Temperature 298.15 K 323.15 K 348.15 K 373.15 K 398.15 K 423.15 K 448.15 K 473.15 K 498.15 K Trifluoroacetic acid A B C 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 A 5.40221 5.20389 5.07680 5.08354 3.55959 3.55959 3.55959 3.55959 3.55959 Water B 1838.675 1733.926 1659.793 1663.125 643.748 643.748 643.748 643.748 643.748 C -31.737 -39.485 -45.854 -45.622 -198.043 -198.043 -198.043 -198.043 -198.043 Table 3.1. Antoine equation parameters used for trifluoroacetic acid and water at each temperature point investigated. Numbers taken from [18] and are set such that P will be measured in bar. with its vapor pressure, ∆Ggas→solv = 0, and we thus have  ∆Ggas→liquid = Ggas (P ) − Ggas (1 atm) = RT ln P 1 atm  . We can find the vapor pressure P at a given temperature using the Antoine Equation: log10 P = A − B , C +T where the empirical parameters A, B, and C vary with the solvent and temperature range and were taken from table 3.1. Calculations were performed at nine temperature levels spaced by 25 K from 298.15 K to 498.15 K. The SR 2 attack involving OVV Cl3 converts two singlets to two doublets, and hence the transition states feature spin contamination (S 2 > 0) while representing an overall singlet. Structures that did not feature such spin contamination were considered to be transition states for an alternative SN 2 attack that forms MeOVIII Cl–3 and a RhI species, all of which were higher in energy. 3.4 Results The most common oxidation state of rhodium is +III, but oxidation states from 0 to +VI have been observed [19]. While previous work has focused on the C−H activation ability of RhI [3]b, our current results center on a RhIII resting state. This is due to the requirement that the desired process operate above the CH4 /CH3 OH redox couple (0.59 V vs. SHE at pH = 0) and below the reduction potential of O2 in the reaction conditions, which favor higher oxidation states. Scheme 3.1 shows hypothetical catalytic cycles for the activation and functionalization of methane. Starting from an inorganic RhIII resting state (1), we investigated the C−H activation of methane 61 following the displacement of a TFAH solvent molecule (2‡ ) to form a RhIII -methyl organometallic species (3). Starting with (3), there are several pathways to functionalization: 1. III-I: SN 2 attack (4‡ ) by the conjugate base of the solvent to form methyl trifluoroacetate and a RhI species (5), which is reoxidized to the inorganic RhIII resting state (1). In all cases investigated we found that this SN 2 attack occurs on a five-coordinate Rh complex, with the neutral solvent trans to the methyl group dissociated in the transition state. This is consistent with previous work by Goldberg et al. implicating five-coordinate platinum intermediates [20]; 2. III-II: SR 2 attack (6‡ ) by a metal-oxo species to form a metal-methoxy species and a RhII species (7) via methyl radical transfer. Both of these are then reoxidized by one electron, to the metal-oxo species and methanol and to the inorganic RhIII resting state (1), respectively. Here, OVCl3 was used as a model metal-oxo capable of one-electron reduction, even though it would likely hydrolyze in these solvents. We are developing stable metal oxo reagents optimized for this reaction mechanism [21] and use OVCl3 here as a computationally simple model. The SR 2 F transition state for Rh(NN) involves a five-coordinate Rh complex as well. For the Rh(NN ) case, the transition states retain weak interactions with the neutral trans ligand, with an elongated Rh−O bond distance of 2.472 Å for TFAH and 2.228 Å for water. This is presumably F due to the increased electrophilicity of the metal center in the Rh(NN ) complexes; 3. III-IV-II: Oxidation to a RhIV species (8) followed by SN 2 attack (9‡ ) generating a RhII species (7), which is then further oxidized back to the RhIII resting state (1). This route was x only investigated for the Rh(NN ) and Rh(DP M E S) complexes, which had both kinetically accessible activation transition states and thermodynamically accessible RhIV intermediates. In all cases investigated we found that this SN 2 attack occurs on a six-coordinate Rh complex, presumably due to the increased electrophilicity of the RhIV center. The six-coordinate Rh complexes have elongated Rh−O bond distances to the neutral trans ligand (e.g., 2.377 Å for F F Rh(NN), 2.379 Å for Rh(NN ), and 2.255 Å for Rh(NN ) in water). In all cases, the energy of reoxidation was calculated assuming a potential of 1.23 V vs. SHE, i.e. with O2 as the terminal oxidant. However, in practice it may be necessary to use intermediate oxidants as well, although that is beyond the scope of this study. A system which can proceed around any of these cycles with a global activation barrier below approximately 36 kcal/mol at 200◦ C is of academic interest, as transition state theory gives a turnover frequency (TOF) of roughly 1 hr−1 at 1 atm CH4 for such a process. Achieving an industrially relevant TOF on the order of 1 s−1 requires a global barrier of 29 kcal/mol, although greater pressures of CH4 would increase this limit [22]. We screened complexes of rhodium with the following ligands, as depicted in Figure 3.1: 62 Scheme 3.1. Hypothetical catalytic cycle for the activation and functionalization of methane. This shows several potential routes, including III-I (red), III-II (blue), and III-IV-II (green), which are further described in the text. In addition, a fourth possibility, I-III (purple), involves the direct C−H activation (10‡ ) by a RhI species (5) to a RhIII hydride methyl (11). However, given the oxidizing nature of the reaction conditions, we do not anticipate this to be a fruitful pathway, as any catalytic cycle involving a RhI -(III) cycle would likely suffer from the fact that oxidation of 5 to 1 would act as a thermodynamic sink. Hence, this last option was not pursued to the same extent in this investigation. 63 F • the bis(N -phenyl)benzylamidinate (NN) ligand and its fluorinated analogue (NN ); F NMe2 • the (N -phenyl)acetaldiminyl quinolate (ONN) ligand and its related (ONN ) and (ONN ) ligands; • the bis(pyridyl)methanesulfonate (DPMS) and bis(pyridyl)ethanesulfonate (DPES) ligands; • the bis(pyrrolyl)quinolinyl phosphine (PN) ligand; and • the bis(quinolinyl)benzene (bisq) ligand, its di- and tetrafluorinated analogues (bisqF2 ) and (bisqF4 ), and its di- and tetramethylated analogues (bisqMe2 ) and (bisqMe4 ). Figure 3.1. The rhodium-ligand complexes screened. For each complex, the RhI form in TFAH is shown here, but full catalytic cycles were computed. The best kinetics were predicted for the Rh(NNF ) and Rh(bisq) complexes. For each of these ligands complexed with rhodium, we generated catalytic cycles in the template of Scheme 3.1. For the (DPMS), (PN), and (ONNx ) family of ligands, we found that either the methane activation energy or the methyl functionalization energy (or both) were too high for effective catalysis (Table 3.2). However, we found lower numbers for the (NNx ) family of ligands, on which we have decided to focus our discussion. 3.4.1 The Rh(NN) complexes in TFAH Scheme 3.2 shows the calculated free energies of Rh(NN) complexes in trifluoroacetic acid solvent. III 2– + Note that all free energies are referenced to the resting state, [(NN)Rh (TFA)4 ] (Hax )2 , which is highlighted by a purple box. Hence, although the resting complex is regenerated at the end of the catalytic cycle, the new free energy is −13.7 kcal/mol at 298 K, because this is the energy change of the overall equation CH4 + TFAH −−→ TFA−Me + 2 H+ + 2 e– ; and thus all energies at the second cycle would be −13.7 kcal/mol lower. Oxidation of the inorganic RhI species to inorganic RhIII species is highly favorable, from 40 to 50 kcal/mol downhill, depending on the temperature and the charge of the species. We conclude that 64 a. b. Scheme 3.2. (a) Activation of methane using Rh(NN) complexes in TFAH; and (b) the subsequent functionalization. Red denotes the III-I functionalization pathway; blue the III-II pathway, and green the III-IV-II pathway. Free energies (kcal/mol) are referenced to the resting state, [(NN)RhIII (TFA)4 ]2– (H+ ax )2 . The resting state is highlighted by a purple box, and the key methyl inter+ mediate, [(NN)RhIII (Meax )(TFA)3 ]2– (H+ ax )(Heq ), is highlighted by a yellow box. 65 TS type Activation Functionalization TS type TS Meax Meeq (III-I) (III-II) (III-IV-II) TS Meax Activation Meeq Functiona- (III-I) lization (III-II) Rh(NN) 32.6/34.8 38.3/40.4 54.8/60.3 43.2/44.2 42.2/48.6 Rh(NN ) 27.6/29.6 33.7/36.3 42.7/48.2 38.1/40.0 36.8/43.6 F Rh(ONN) 37.2/39.8 37.9/40.7 44.2/49.3 40.4/42.5 N/A Rh(ONN ) 35.8/37.8 38.1/40.1 41.7/45.8 39.1/40.8 N/A F 2 Rh(ONN ) 38.7/41.2 40.0/42.9 46.7/52.6 41.4/43.4 N/A NMe Rh(DPMS) 38.0/40.2 33.9/36.2 46.6/51.3 41.0/41.6 49.3/56.3 Rh(PN) 40.1/43.0 35.0/37.7 38.3/43.8 47.7/56.9 Rh(bisq) 38.3/40.7 33.4/35.8 31.8/37.8 32.0/33.7 Rh(bisqF2 ) 39.3/42.4 36.4/39.4 32.3/38.4 31.8/33.6 Rh(bisqF4 ) 40.4/43.0 38.2/40.9 30.2/35.0 33.7/35.3 Rh(bisqMe2 ) 38.8/41.9 32.8/35.9 35.0/41.9 35.7/38.3 Rh(bisqMe4 ) 40.7/44.4 34.3/37.5 38.0/45.5 38.9/42.0 Rh(DPES) 41.0/44.0 35.6/38.4 53.8/59.3 45.5/47.4 49.9/57.1 Table 3.2. Lowest activation and functionalization energies for each series of Rh-ligand complexes in TFAH. Detailed cycles for the (NNx ) family of ligands are found in Schemes 3.2 and 3.3. Detailed cycles for the (ONNx ) family of ligands are found in Schemes 3.9 through 3.11. Detailed cycles for the (DP M S) family of E ligands are found in Schemes 3.12 through 3.14. Detailed cycles for the (PN) ligand are found in Schemes 3.15 and 3.16. Detailed cycles for the (bisqx ) family of ligands are found in Schemes 3.17 through 3.20. For each entry, the first number is at 298 K and the second at 498 K. Not all III-IV-II transition states were calculated. All free energies in kcal/mol. the III-I SN 2 attack pathway is not likely. Indeed, the activation barrier for this is a prohibitively high 54.8 kcal/mol at room temperature (Figure 3.2b). In addition, we also conclude that a I-III pathway involving methane activation by RhI is not likely due to the oxidation potential of the RhI species. The remaining viable pathways all involve methane activation by the inorganic RhIII (NN) complex to form a RhIII (NN)−Me species. We were initially gratified to see that thermodynamically the methyl species is not very endergonic. We were even more intrigued to find that the methane activation transition state barrier is only 32.6 kcal/mol at 298 K (Figure 3.2a), making it the most accessible barrier that we had found for RhIII complexes in TFAH (Table 3.2). In contrast, neither the III-II (Figure 3.2c) nor the III-IV-II (Figure 3.2d) functionalization pathways seemed as promising. Although further oxidation to a RhIV (NN)−Me species was roughly thermoneutral, the barrier to reductive functionalization was 42.2 kcal/mol. The oxidation from RhIII to RhIV was effective in lowering the global SN 2 barrier from 54.8 to 42.2 kcal/mol. To explain the high functionalization barrier for SN 2 attack, we note that this may be due to the high energy or low chemical activity of trifluoroacetate ion in solution. Since TFA– is necessary as the attacking nucleophile, a higher pH may be employed to lower this barrier before a deprotonated intermediate becomes the new resting state. Due to the unusually low barrier for methane activation, we did not want to dismiss this ligand framework. Rather, our efforts focused on how potential modifications on the (NN) ligand might lower the functionalization barrier. Reasoning that a more electronegative ligand may inductively increase the electrophilicity of the rhodium-methyl bond and make nucleophilic attack more likely, we decided to investigate Rh complexes with the fluorinated analogue bis(N -pentafluorophenyl) F pentafluorobenzylamidinate (NN ) ligand. 66 Figure 3.2. Rh(NN) transition state structures. (a) methane activation: Rh1-C36 2.273 Å, C36-H19 1.365 Å, H19-O10 1.321 Å, Rh1-C36-H19 64.624◦ , C36-H19-O10 155.492◦ . (b) Rh(III-I) SN 2 attack: Rh1-C60 2.431 Å, C60-O65 1.936 Å, Rh1-C60-O65 176.680◦ , Rh1-O53 3.473 Å. (c) Rh(III-II) SR 2 attack: Rh1-C60 2.368 ÅC60-O65 1.995 Å, Rh1-C60-O65 175.541◦ , Rh1-O53 3.171 Å. (d) Rh(III-IV-II) SN 2 attack: Rh1-C60 2.138 Å, C60-O64 2.263 Å, Rh1-O53 2.377 Å, Rh1-C60-O64 168.802◦ , C60-Rh1-O53 164.984◦ . 67 The Rh(NNF ) complexes in TFAH 3.4.2 F Compared to the analogous Rh(NN) complexes, the fluorinated Rh(NN ) complexes (Scheme 3.3) show some marked differences in free energies. Many of these differences can be attributed to the F much higher electron-withdrawing character of the (NNF ) ligand. For instance, the RhI (NN ) species is 13.3 kcal/mol more stable to oxidation than the corresponding RhI (NN) species. In the same way, the SN 2 functionalization barrier on the III-I path is 12.1 kcal/mol lower. This is due to the highly F electron-withdrawing (NN ) ligand, which stabilizes lower oxidation states. Unfortunately, this effect is not sufficiently large for us to consider the III-I or I-III pathways. Other thermodynamic effects of fluorine atom substitution include increased acidity of all the complexes (i.e. stabilization of the anionic species by 4-8 kcal/mol), an increased stabilization of the RhIII −Me species (by about 8 kcal/mol), and a decreased stabilization of the RhIV −Me species (by about 20 kcal/mol relative to RhIII −Me.) The barrier for methane activation, already relatively low for the Rh(NN) case, was even lower F for the Rh(NN ) case, at only 28.7 kcal/mol at 298 K (Figure 3.3a). However, the real test of the utility of the (NNF ) ligand is its effectiveness in aiding the functionalization step. It lowers the global III-I SN 2 barrier from 54.8 to 42.7 kcal/mol and the isolated SN 2 barrier (from the RhIII −Me intermediate) to 40.8 from 45.1 kcal/mol (Figure 3.3b). It raises the neutral III-IV-II SN 2 barrier by raising the RhIV intermediate energy. Hence, we put our hopes into the (NNF ) ligand lowering the barrier for a net anionic III-IV-II SN 2 or a III-II SR 2 attack (Figure 3.3c,d). And indeed it does: the new global SN 2 and SR 2 barriers of 36.8 and 38.1 kcal/mol, respectively, approach the target values. The Rh(NNF ) complexes in water 3.4.3 Another method that may lower transition state barriers is switching the solvent, and therefore the nucleophile, to water. This has the practical advantage that any commercialized process will have greater simplicity if carried out in water as opposed to a strong acid. Hence, we investigated the F same Rh(NN ) complexes in water. F The thermodynamic profile for Rh(NN ) complexes in water (Scheme 3.4) is very promising: F F F the RhIII (NN )-methyl, RhIV (NN )-methyl, and RhII (NN ) species are all stabilized relative to the F inorganic RhIII (NN ) reference state. Hence, whereas for the TFAH solvent case a thermodynamic barrier must be climbed in each catalytic step, in water each step is either thermodynamically downhill or approximately neutral. F III In Scheme 3.4, all energies are referenced to the lowest inorganic RhIII state, [(NN )Rh 2– + + (OH)4 ] (Hax )(Heq ), which is highlighted by a purple box. This is to keep consistent with the practices in Schemes 3.2 and 3.3, where the lowest inorganic RhIII state is also the resting state. F However, in this case the organometallic RhIII (NN )-methyl complexes are stabilized to such an 68 a. b. Scheme 3.3. (a) Activation of methane using Rh(NNF ) complexes in TFAH; and (b) the subsequent functionalization. Red denotes the III-I functionalization pathway; blue the III-II pathway, and green the III-IV-II pathway. Free energies (kcal/mol) are referenced to the resting state, [(NNF )RhIII (TFA)4 ]2– (H+ ax )2 . The resting state is highlighted by a purple box, and the key methyl intermediate, [(NNF )RhIII (Meax )(TFA)3 ]2– (H+ ax ), is highlighted by a yellow box. 69 Figure 3.3. Rh(NNF ) transition state structures. (a) methane activation: Rh1-C61 2.267 Å, C61-H19 1.365 Å, H19-O10 1.321 Å, Rh1-C61-H19 63.875◦ , C61-H19-O10 157.549◦ . (b) Rh(III-I) SN 2 attack: Rh1-C60 2.368 Å, C60-O65 2.021 Å, Rh1-C60-O65 176.330◦ , Rh1-O53 3.265 Å. (c) Rh(III-II) SR 2 attack: Rh1-C60 2.399 ÅC60-O65 1.935 Å, Rh1-O53 2.472 Å, Rh1-C60-O65 172.164◦ , C60-Rh1-O53 168.496◦ . (d) Rh(III-IV-II) SN 2 attack: Rh1-C60 2.259 Å, C60-O64 2.156 Å, Rh1-O53 2.379 Å, Rh1-C60-O64 159.539◦ , C60-Rh1-O53 169.220◦ . 70 Scheme 3.4. Thermodynamic profile for the activation and functionalization of methane using Rh(NNF ) complexes in water. This chart is meant to be read from left to right; the first row contains the pathway through neutral species and the second row through anionic species. Red denotes the III-I functionalization pathway; blue the III-II, and green the III-IV-II pathway. All free energies (kcal/mol) are at pH = 7 and + referenced to the lowest inorganic RhIII state, [(NNF )RhIII (OH)4 ]2– (H+ ax )(Heq ), highlighted by a purple box; F III whereas the key methyl intermediate and true resting state, [(NN )Rh (Meeq )(OH)3 ]2– (H+ ax )2 , is highlighted by a yellow box. 71 F III 2– + extent that the true resting state is now [(NN )Rh (Meeq )(OH)3 ] (Hax )2 , which is highlighted by a yellow box and is analogous to the key methyl intermediate referenced in Schemes 3.2 and 3.3. The practical consequence is that the barrier to CH4 activation is properly referenced to the lowest inorganic RhIII state, whereas the barrier to methyl group functionalization ought to be referenced to the true resting state. F Having verified that the Rh(NN ) catalytic system in water was thermodynamically feasible, we sought accessible transition states for both the methane activation and functionalization steps. For methane activation, we began by assuming that the activation would take place via electrophilic substitution with a cis hydroxo ligand acting as a proton acceptor (Scheme 3.5). We then found transition states, starting from both neutral and anionic species, with the methane and the interactF ing hydroxo in every possible configuration with respect to the (NN ) ligand. Our lowest transition state barrier of 35.0 kcal/mol at 298 K occurs with the methyl group forming in the axial position, donating its extra proton to an equatorial hydroxo ligand and trans to an axial hydroxo ligand (Figure 3.4a). This is significantly higher than in the TFAH solvent case, but still within the limits of feasibility. We examined the possible role of a concerted reaction with a bridging water molecule, but we did not see a decrease in the transition state energy. Scheme 3.5. Transition states for the activation of methane using Rh(NNF ) complexes in water. Chart is meant to be read from top to bottom; the left side contains the pathway through neutral species and the right side through anionic species. All free energies are in kcal/mol and referenced to the lowest inorganic RhIII state. Following the activation of methane, the other transition state that needed to be calculated was for the functionalization of the rhodium-methyl bond. We investigated the functionalization through 72 Figure 3.4. Rh(NNF ) in water transition state structures. (a) methane activation: Rh1-C66 2.296 Å, C66H65 1.383 Å, H65-O5 1.271 Å, Rh1-C66-H65 55.587◦ , C66-H65-O5 161.845◦ . (b) Rh(III-I) SN 2 attack: Rh1-C64 2.698 Å, C64-O67 1.803 Å, O67-H68 1.043 Å, H68-O47 1.516 Å, O47-H53 1.001 Å, H53-O5 1.718 Å, Rh1-C60-O65 148.135◦ , Rh1-O54 3.265 Å. (c) Rh(III-II) SR 2 attack: Rh1-C65 2.371 ÅC65-O68 2.010 Å, Rh1-O55 2.228 Å, Rh1-C65-O68 164.320◦ , C65-Rh1-O55 169.736◦ . (d) Rh(III-IV-II) SN 2 attack: Rh1-C64 2.526 Å, C64-O67 1.930 Å, O67-H68 1.033 Å, H68-O46 1.503 Å, O46-H51 1.008 Å, H51-O47 1.594 Å, O47-H3 0.981 Å, H3-O5 1.901 Å, Rh1-O54 2.255 Å, Rh1-C64-O67 171.136◦ , C64-Rh1-O54 167.491◦ . 73 the III-I, III-II, and III-IV-II routes (Schemes 3.6, 3.7, and 3.8, respectively). For each route, we restricted our investigation to activations from rhodium species with axial methyl groups, since (as in the TFAH case) the axial isomer is preferentially formed during the activation step. Scheme 3.6. Transition states for the III-I functionalization pathway via SN 2 attack in water. The left side is the pathway through neutral species and the right side through anionic species. All free energies are in kcal/mol and referenced to the resting state, [(NNF )RhIII (Meeq )(OH)3 ]2– (H+ ax )2 . For the III-I route, we required an aqua ligand to be the leaving group trans to the methyl, limiting us to only two possible transition states: a neutral transition complex (Scheme 3.6, left) and an anionic complex (Scheme 3.6, right). Although the neutral transition state (Figure 3.4b) is somewhat lower in energy, we found that in both cases the activation energy for this pathway is prohibitively high. For the III-II route (Scheme 3.7), we no longer needed require the leaving group trans to the RhIII methyl to be an aqua ligand. Hence we performed calculations on four different potential transition F III 2– + states, starting with two isomers of neutral [(NN )Rh (Meax )(OH)3 ] (H )2 (Scheme 3.7, upper F III 2– + left) and two isomers of anionic [(NN )Rh (Meax )(OH)3 ] (H ) (Scheme 3.7, upper right). We found that this SR 2 pathway was even more facile, with all barriers in the 30s (Scheme 3.7, middle) and the lowest barrier being 31.7 kcal/mol for OVCl3 attack on a neutral species with an axial methyl ligand, which is shown in detail in Figure 3.4c. For the III-IV-II route (Scheme 3.8), a good leaving group trans to the RhIII -methyl once again became important. Hence, we were limited to only investigating one possible transition state, shown in detail in Figure 3.4d. We found the activation energy for this process to be approximately equal to that of the methane activation step. F Hence, we conclude that for the Rh(NN ) catalytic system in water, both III-IV-II SN 2 and III-II SR 2 are viable pathways for functionalization, whereas the III-I SN 2 pathway is still too high in energy. This is similar to the case in TFAH, except that the relative ease of activation vis-à-vis functionalization has now been reversed. 3.4.4 The Rh(ONNx ) family of complexes in TFAH Schemes 3.9, 3.10, and 3.11 depict the catalytic cycle for the activation and functionalization of F NMe2 methane using Rh(ONN), Rh(ONN ), and Rh(ONN ) complexes in TFAH, respectively. In all 74 Scheme 3.7. Transition states for the III-II functionalization pathway via SR 2 attack in water. This chart is meant to be read from top to bottom; the left side contains the pathway through neutral species and the right side through anionic species. All free energies are in kcal/mol and referenced to the resting state, [(NNF )RhIII (Meeq )(OH)3 ]2– (H+ ax )2 . Scheme 3.8. Transition state for the III-IV-II functionalization pathway via SN 2 attack in water. All free energies are in kcal/mol and referenced to the resting state, [(NNF )RhIII (Meeq )(OH)3 ]2– (H+ ax )2 . 75 cases, the resting state contains an axial TFAH ligand and equatorial TFA ligands, but the activation transition state which involves an axial methyl group is slightly lower in energy than the transition state which involves an equatorial methyl group. The methane activation energy for Rh(ONN) is F 37.2 kcal/mol at 298 K, which is somewhat too high for promising catalysis. Switching to Rh(ONN ) decreases this activation energy slightly to 35.8 kcal/mol, whereas using Rh(ONN NMe2 ) raises it to 38.7 kcal/mol. Scheme 3.9. Catalytic cycle for the activation and functionalization of methane using Rh(ONN) complexes in TFAH. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. The RhIII (ONNx )−Me species investigated have a methyl group axial to the (ONNx ) plane; any isomer with an equatorial methyl group would likely be less stable due to the trans effect of the (ONNx )’s quinoline nitrogen. The RhIII (ONNx )−Me species are slightly endergonic relative to their respective resting states. In all cases, the III-II SR 2 pathway is favored over the III-I SN 2 pathway. For Rh(ONN) the F SR 2 barrier is 40.3 kcal/mol at 298 K; this decreases to 39.2 kcal/mol for Rh(ONN ) but increases 76 Scheme 3.10. Catalytic cycle for the activation and functionalization of methane using Rh(ONNF ) complexes in TFAH. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 77 Scheme 3.11. Catalytic cycle for the activation and functionalization of methane using Rh(ONNNMe2 ) complexes in TFAH. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 78 to 41.4 kcal/mol for Rh(ONN NMe2 ). The resultant RhIII (ONNx ) species were all approximately the same (20.5, 20.3, and 20.7 kcal/mol, respectively) in energy relative to their resting states. 3.4.5 The Rh(DPMS) and Rh(DPES) complexes in TFAH Scheme 3.12 shows the catalytic cycle for the activation and functionalization of methane using Rh(DPMS) complexes in TFAH, with the latter via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. While there is a relatively accessible activation barrier of 33.9 kcal/mol at 298 K, it differs from the activation transition states of the NNx and ONNx families in that it forms a methyl complex with the methyl group equatorial to the N−Rh−N plane. Interconvertibility with the lowerenergy axial methyl complex is assumed given the lability of the sulfonate moiety and the TFAH ligands. Unfortunately, the lowest functionalization barrier found was 41.0 kcal/mol at 298 K for the III-II pathway, which is too high. The analogous catalytic cycle for Rh(DPES) complexes in TFAH is shown in Scheme 3.13. The energy profile is very similar to that of the Rh(DPMS) complexes; however, the transition states found are typically several kcal/mol higher than their Rh(DPMS) counterparts. This may be due to the extra methyl group rendering the overall ligand more electron donating, destabilizing the formation of a Rh−Me bond and decreasing its electrophilicity for functionalization. We also investigated RhI (DPMS) species, the results of which are shown in Scheme 3.14. The energy profile appears to be much more favorable, with the highest barrier being the oxidative addition of methane at 26.4 kcal/mol at 298 K. However, under reaction conditions one might expect the oxidative addition of methane by RhI (DPMS) to be outcompeted by RhI (DPMS)’s oxidation to a RhIII (DPMS) species, which according to Scheme 3.12 is downhill by 33.0 kcal/mol at 298 K. Therefore, the I-III pathway was not considered to be viable. 3.4.6 The Rh(PN) complexes in TFAH Scheme 3.15 shows the catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, with the latter via the Rh(III-I) and Rh(III-II) pathways. Unfortunately, it appears that the functionalization barriers are too high for effective catalysis with this ligand: the lowest III-I SN 2 barrier is 38.0 kcal/mol at 298 K and the lowest III-II SR 2 barrier is 47.7 kcal/mol. Taking into account the additional fact that the activation transition state to produce an axial methyl species is 40.1 kcal/mol high, we conclude that RhIII (PN) is not likely to be an effective catalyst for methane oxidation. We also investigated RhI (PN) species, the results of which are shown in Scheme 3.16. As in the case of the Rh(DPMS) complexes, oxidative addition of methane by RhI (PN) appears to be more facile, with a barrier of 28.1 kcal/mol at 298 K. Unfortunately, the resultant RhIII (PN)(H)(Me) 79 Scheme 3.12. The activation and functionalization of methane using Rh(DPMS) complexes in TFAH, via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 80 Scheme 3.13. The activation and functionalization of methane using Rh(DPES) complexes in TFAH, via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 81 Scheme 3.14. Catalytic cycle for the activation and functionalization of methane using Rh(DPMS) complexes in TFAH, via the Rh(I-III) pathway. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. species are all significantly thermodynamically uphill (at least 22.0 kcal/mol at 298 K). Furthermore, as in the case of the Rh(DPMS) complexes, under reaction conditions one might expect the oxidative addition of methane by RhI (PN) to be outcompeted by RhI (PN)’s oxidation to a RhIII (PN) species, which according to Scheme 3.15 is downhill by 23.7 kcal/mol at 298 K. Therefore, the I-III pathway was not further pursued. 3.4.7 The Rh(bisqx ) family of complexes in TFAH Due to the large in-TFAH Rh−Me functionalization barriers encountered in the other ligands so far (Table 3.2), we designed the bis(quinolinyl) benzene (bisqx ) family of ligands (Figure 3.1) in the hopes of finding more facile SN 2 and SR 2 pathways. These ligands would be expected to be coordinated to rhodium in a fac-L3 manner, with a weak η 2 -benzene interaction axial to the N−Rh−N plane. A methyl group would then be expected to be coordinated axial to the N−Rh−N plane as well, due to the weaker trans effect of the benzene ring as opposed to the nitrogen donors. The η 2 benzene interaction is also expected to be a better leaving group than TFA/TFAH, thereby lowering both the barrier and thermodynamics of functionalization. Beyond the base (bisq) ligand itself, we also investigated the di- and tetrafluorinated analogues (bisqF2 ) and (bisqF4 ), and the di- and tetramethylated analogues (bisqMe2 ) and (bisqMe4 ), to see how changing the electronics of the ligand 82 Scheme 3.15. Catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, via the Rh(III-I) and Rh(III-II) pathways. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 83 Scheme 3.16. Catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, via the Rh(I-III) pathway. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. might change its associated transition state barriers. x x Scheme 3.17 shows the ground states of the various RhIII (bisq ) species: the neutral RhIII (bisq )III x + (TFA)3 , the protonated [Rh (bisq )(TFA)2 (TFAH)] III x + and [Rh (bisq )(TFA)2 )] , and their relax tive free energies. The first point to note is that, since the (bisq ) ligands are L3 , activation of methane requires first the protonation of a TFA ligand followed by its removal as TFAH to form an open x coordination site. As Scheme 3.17 shows, as we increase the electron-withdrawing groups on (bisq ), replacing Me with H and H with F, the resultant decrease in electron donation to the central Rh decreases the basicity of the attached TFA ligands and increases the energy of their protonated forms. x The second and related point to note is that the true resting state is not consistent for all (bisq ) species: the more electron-rich ligands (bisqMe2 ) and (bisqMe4 ) have positively-charged resting states, the base ligand (bisq) has both the neutral and a protonated species in even equilibrium, while the less electron-rich ligands (bisqF2 ) and (bisqF4 ) prefer to be neutral. However, the energies of all species x have been referenced to neutral RhIII (bisq )(TFA)3 for consistency. The final point to note is that the specific protonated resting state for the more electron-rich ligands (bisqMe2 ) and (bisqMe4 ) depends III Mex on temperature: at 298 K [Rh (bisq + )(TFAax )(TFAHeq · · · TFAeq )] is favored, but at 498 K it is III Mex more advantageous to dissociate the TFAH and have [Rh (bisq + )(κ1 −TFAeq )(κ2 −TFAax, eq )] instead. Scheme 3.18 shows the three potential methane activation transition states that may result 84 Scheme 3.17. The inorganic RhIII (bisqx )(TFA)3 species and its protonated analogues III x [Rh (bisq )(TFA)2 (TFAH)]+ and [RhIII (bisqx )(TFA)2 )]+ , along with their relative energies in TFAH. Both protonation and TFAH removal are necessary before methane activation can take place. All free energies are in kcal/mol and referenced to the corresponding RhIII (bisqx )(TFA)3 species. III x + depending on the CH4 molecule’s approach towards the [Rh (bisq )(TFA)2 )] ’s open coordination site. The top transition state shows the result of a methane approach to an axial open coordination site, whereas the other two show the result of an equatorial approach, with either the axial or III equatorial TFA ligand gaining the methane’s proton. The axial approach results in a [Rh −Meax ] III species, whereas the equatorial approaches result in [Rh −Meeq ] + + species, all of which may be IV deprotonated to form their neutral analogues. Further oxidation to Rh species is quite uphill; hence the III-IV-II functionalization pathway was not investigated for these rhodium-ligand complexes. x It should be noted that regardless of the specific (bisq ) ligand, the lowest energy methane actiIII x +‡ vation transition state is [Rh (bisq )(TFAax )(Meeq · · · H· · · TFAeq )] III , which involves an equatorial x + approach for the methane and results in the methyl complex [Rh (bisq )(TFAax )(Meeq )(TFAHeq )] . However, both the neutral and protonated RhIII −Meax species are lower in energy than their equatorial counterparts. We assume that interconversion between the RhIII −Meeq and RhIII −Meax is facile due to the lability of the TFAH and η 2 -benzene ligands. Upon examination of the methane activation transition state energies, it appears that the general trend is that increasing the electron-donating nature of groups on the (bisq) ligand appears to favor lower barriers. This is opposite to that observed in the other ligand families, in which less III x electron-donating ligands are preferred. However, a comparison of the energies of [Rh (bisq )(TFA)+‡ (Me· · · H· · · TFA)] III x + transition states with that of their protonated [Rh (bisq )(TFA)2 (TFAH)] 85 Scheme 3.18. The various methane activation transition states [RhIII (bisqx )(TFA)(Me· · · H· · · TFA)]+‡ and their resultant [RhIII (bisqx )(TFA)(Me)(TFAH)]+ products. Ensuing deprotation (in preparation for III-I and III-II functionalization) and oxidation (in preparation for III-IV-II functionalization) are also depicted. Blue denotes the most favored Rh−Me intermediate and red the most favored activation transition state. All free energies are in kcal/mol and referenced to the corresponding RhIII (bisqx )(TFA)3 species. 86 precursors (Scheme 3.17) shows that the decreased activation barriers for more electron-donating x ligands is simply due to the decreased energy of protonation of the inorganic RhIII (bisq )(TFA)3 precursors. It should also be noted that for (bisqMe2 ) and (bisqMe4 ), the energies shown in Scheme 3.18 are not the full transition barriers, since they are not referenced to the ground states of those rhodium-ligand complexes. The true barriers are recorded in Table 3.2. x An examination of the energies of the lowest Rh−Me intermediate for each (bisq ) ligand (Scheme 3.18, blue) shows that the relative energy decreases as the electron-withdrawing nature of groups on the (bisq) ligand is increased. This is easily explained by noting that a decrease in the electron-donating power of the (bisq) ligand is expected to increase the rhodium atom’s elecF trophilicity, and thus increase the strength of its bond with the methyl group. Indeed, for (bisq 4 ) x the Rh−Me species is slightly lower (by 0.3 kcal/mol) in energy than its RhIII (bisq )(TFA)3 precursor at 298 K, and is the true resting state. However, this does not change the overall methane activation barrier. The III-I SN 2 and III-II SR 2 functionalization pathways were investigated by searching for transition states resulting from TFA– /TFAH or OVCl3 attack on each Rh−Me species, axial and equax torial, neutral and positively charged; for all five (bisq ) ligands. A total of 65 transition state structures were analyzed. Figure 3.5 shows the functionalization transition states for SN 2 attack by TFA– /TFAH on various Rh−Me intermediates. Both H+ pre-protonation followed by TFA– attack and concerted TFAH attack/deprotonation transition states were investigated. The lowest energy SN 2 functionalization barrier was found to arise from the attack of a TFA– ion on a protonated Rh−Meax species. Not x surprisingly, lower barriers were found for RhIII (bisq ) complexes with less electron-donating ligands. The energies are overall significantly lower than those found with other ligand families, thus confirming our initial hypothesis that a weak η 2 -benzene coordinated trans to the methyl would facilitate SN 2 attack. Indeed, barriers of 34.1 kcal/mol and lower at 298 K are found for (bisq) and x its fluorinated analogues, thereby establishing the viability of the RhIII (bisq ) family of complexes as potentially effective catalysts. Figure 3.6 shows the functionalization transition states for SR 2 attack by OVCl3 on both neutral x III x + RhIII (bisq )(Me)(TFA)2 and protonated [Rh (bisq )(Me)(TFA)(TFAH)] species, with both axial and equatorial methyl conformations. In most cases the lowest energy SR 2 functionalization barrier was found to arise from the attack of a OVCl3 molecule on a neutral Rh−Meax species (the lone exception is (bisq Me4 ), which prefers a protonated transition state). As in the III-I SN 2 case, lower x barriers were found for RhIII (bisq ) complexes with less electron-donating ligands, and overall the barriers were reduced compared to those of other ligand families; although the effect is not as F4 pronounced as for the SN 2 case, and the tetrafluorinated (bisq ) version is an outlier. Barriers of 33.4 kcal/mol and lower at 298 K are found for (bisq) and its fluorinated analogues, thereby establishing 87 Figure 3.5. The various RhIII (bisqx )(Me) functionalization transition states for the III-I SN 2 pathway. All transition states resulting from TFA– /TFAH attack on each Rh−Me species, axial and equatorial, neutral and positively charged, are shown. The lowest transition state for each (bisqx ) ligand is shown in red. All free energies are in kcal/mol and referenced to the corresponding RhIII (bisqx )(TFA)3 species. 88 the viability of the III-II pathway as well as the III-I pathway for methyl functionalization of the x RhIII (bisq ) family of complexes. Figure 3.6. The various RhIII (bisqx )(Me) functionalization transition states for the III-II SR 2 pathway. All transition states resulting from OVCl3 attack on each Rh−Me species, axial and equatorial, neutral and positively charged, are shown. The lowest transition state for each (bisqx ) ligand is shown in red. All free energies are in kcal/mol and referenced to the corresponding RhIII (bisqx )(TFA)3 species. Scheme 3.19 shows the remainder of the catalytic cycle after either III-I or III-II functionalization, and compares the two functionalization processes. For the III-I pathway, SN 2 attack by TFA– /TFAH x results in the formation of the Me−TFA product and RhI (bisq )(TFA)(TFAH), which is reoxidized x to the starting complex RhIII (bisq )(TFA)3 . For the III-II pathway, SR 2 attack by OVCl3 results in x the formation of Me−OVCl3 and RhII (bisq )(TFA)(TFAH), which are also reoxidized to the starting x complex RhIII (bisq )(TFA)3 as well as the product Me−TFA and regenerated OVCl3 . Either way, x the final relative energy of the RhIII (bisq )(TFA)3 complex is −13.7 kcal/mol at 298 K, which represents a completed cycle of the overall reaction CH4 + TFAH −−→ TFA−Me + 2 H+ + 2 e– . x Finally, the I-III pathway was also investigated, starting from RhI (bisq )(TFA)(TFAH) as an 89 Scheme 3.19. Comparison of the functionalization of RhIII (bisqx )(Meax )(TFA)2 , the lowest Rh−Me species, via the III-I SN 2 and III-II SR 2 pathways; and the completion of the catalytic cycle. Only the lowest functionalization transition state is shown for each pathway. 90 alternative ground state. The results are shown in Scheme 3.20. Although the oxidative addition of x methane by RhI (bisq ) species was feasible with minimum transition states of 27.9-30.6 depending x on the particular ligand, the resulting RhI (bisq )(H)(Me)(TFA) species are very uphill in energy. Deprotonation to form a RhI −Me species was prohibitively uphill both kinetically and thermodyx x namically. Along with the fact that RhI (bisq ) species would be less stable than their RhIII (bisq ) analogues at reaction conditions, we did not pursue this route further. Scheme 3.20. Catalytic cycle for the activation and functionalization of methane using Rh(bisqx ) complexes in TFAH, via the Rh(I-III) pathway. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. By plotting the activation and transition state energies of the various RhIII (bisqx ) ligands, we can determine the particular ligand with the optimal activity. The results are shown in Figure 3.7, and show convincingly that at both 298 K and 498 K, whether using the III-I or III-II pathway, the base ligand RhIII (bisq) is the best choice. In comparing the relative merits of the III-I and III-II pathways, the energies of the two transition x states at 298 K are very close, within 1 kcal/mol for each (bisq ) ligand (although, as mentioned F4 before, (bisq ) appears to be an outlier). However, the energies are expected to increase at 498 K, due to the increased entropy penalty of bringing an extra TFAH or OVCl3 to the system. For the III-I SN 2 pathway, this is a significant penalty of 6.0-6.2 kcal/mol, and may be enough to render functionalization inaccessible. However, for the III-II SR 2 pathway the entropy penalty is much less F2 at 1.7-2.0 kcal/mol, and both the base (bisq) and fluorinated (bisq ) and (bisq F4 ) have barriers at 91 Figure 3.7. Top: graph of the activation and III-I functionalization transition state energies of the various Rh(bisqx ) complexes. The best choice at both 298 K and 498 K is the base Rh(bisq) complex. This complex has a maximum barrier of 33.4 kcal/mol at 298 K and 37.8 kcal/mol at 498 K. The rate-determining step is activation at 298 K and functionalization at 498 K. Bottom: graph of the activation and III-II functionalization transition state energies of the various Rh(bisqx ) complexes. Again, the best choice at both 298 K and 498 K is the base Rh(bisq) complex. This complex has a maximum barrier of 33.4 kcal/mol at 298 K and 35.8 kcal/mol at 498 K. The rate-determining step is methane activation at both temperature points. Both: Blue diamond markers and lines denote transition state energies at 298 K, and red square markers and lines at 498 K. Outline marker shapes denote activation transition states and filled marker shapes denote functionalization transition states. The overall barrier for each particular complex is the greater of the activation and functionalization barriers. 92 35.3 kcal/mol or below. Thus it appears that the III-II SR 2 pathway is the best choice for methyl functionalization. Hence, we conclude that the base Rh(bisq) rhodium-ligand complex is the best choice among the x entire Rh(bisq ) family, with an overall reaction barrier of 33.4 kcal/mol at 298 K and 35.8 kcal/mol at 498 K, when the III-II SR 2 functionalization pathway is used. 3.4.8 Product protection and C−H activation As mentioned in the introduction, a major hurdle that any putative methane to methanol catalytic scheme must overcome arises because the C−H bond dissociation energy of methanol is 9 kcal/mol weaker than that of methane. Preventing the overoxidation of the product is thus a challenge and is the reason we elected to investigate electrophilic metal centers in acidic solvent. Since the transition state of an electrophilic activation involves donation of electron density from the methane C−H σ bond to the metal, a highly electron-withdrawing trifluoroacetate substituent that decreases this σ bond electron density is expected to increase the transition state energy. Thus the highly electronegative trifluoroacetate group of methyl trifluoroacetate withdraws electron density from the methyl C−H bonds, thereby decreasing their ability to donate into the rhodium center and raising activation barriers. Indeed, we see this effect in explicit calculations with both the (NN) and (NNF ) ligand sets. Figures 3.8 and 3.9 show the various isomeric transition states for the activation of both methane and methyl trifluoroacetate in TFAH. At 298 K, each transition state for the activation of methane was 3 to 10 kcal/mol lower than the corresponding transition state for the activation of methyl trifluoroacetate (1 atm reference for both). The first two rows of Figure 3.10 show only the lowest energy isomer of each activation, and thus an overall ∆G‡ (lowest Me−TFA transition state minus lowest CH4 transition state) of 2.7 kcal/mol is found for (NN) and 5.5 kcal/mol for (NNF ). The numbers significantly improve to 6.4 and 9.8 kcal/mol at 498 K. This is due to the volatility of Me−TFA, which greatly raises its free energy at 498 K. F In contrast, activation of methanol (1 M reference state) by Rh(NN ), shown in Figure 3.11 and summarized in the third row of Figure 3.10, was even lower than that of methane: by 0.3 or 1.8 kcal/mol at 298 K when comparing the lowest CH3 OH transition state and the lowest CH4 transition state in TFAH (Figure 3.10, first and second rows); for (NNF ) in water this gap is larger at 2 to 3 kcal/mol (Figure 3.10, third row). We can explain this by noting that the hydroxyl group in methanol has electron donating properties and hence the donation effect of the C−H σ orbital into the electrophilic metal center is increased, leading to a decreased transition state barrier; however these numbers are lower than the 9 kcal/mol simple BDE difference between the C−H bonds in methane and methanol and comparable to work done by Owen et al. [23]. Product protection studies were also carried out for the Rh(DPMS) and Rh(DPES) complexes, 93 Figure 3.8. Transition states for the functionalization of methane, Me−TFA, and methanol using Rh(NN) complexes in TFAH. All free energies are in kcal/mol and relative to the lowest energy inorganic state (i.e. right before methane activation). 94 Figure 3.9. Transition states for the functionalization of methane, Me−TFA, and methanol using Rh(NNF ) complexes in TFAH. All free energies are in kcal/mol and relative to the lowest energy inorganic state (i.e. right before methane activation). 95 Figure 3.10. Summary of the data presented in Figures 3.8, 3.9, and 3.11. In each row, the leftmost structure is the lowest-energy activation transition state for methane (1 atm); the center structure is the lowest-energy transition state for the activation of MeTFA (1 atm, not applicable for the H2 O case); and the rightmost structure is the lowest-energy transition state for the activation of MeOH (1 atm). All free energies are in kcal/mol and relative to the lowest energy inorganic state (i.e. right before methane activation). 96 Figure 3.11. Comparison of activation energies for methane (1 atm) and for methanol (1 M at 298 K, 1 atm at 498 K) in the same structure for Rh(NNF ) complexes in water. In this case, methanol was easier to activate by 2 to 3 kcal/mol. All free energies are in kcal/mol and relative to the lowest energy inorganic state (i.e. right before methane activation). 97 the results of which are shown in Figures 3.12 and 3.13, respectively. Modest product protection ∆∆G‡ ’s of 1.4 kcal/mol and 2.2 kcal/mol at 298 K were found, but these improve to 5.3 kcal/mol and 6.6 kcal/mol at 498 K. Figure 3.12. Comparison of the activation states of methane and methyl trifluoroacetate by Rh(DPMS) complexes, showing the product protection afforded. Red denotes the most favored isomer of each type of activation. All free energies are in kcal/mol and relative to the resting state. x Finally, product protection studies were carried out for the Rh(bisq ) family of rhodium-ligand complexes. The results are summarized in Figure 3.14. In general the product protection is comparable to the Rh(NNx ) family at 298 K, but improves significantly by 3.8-4.2 kcal/mol at 498 K. As in the other cases, this is due to the high volatility of Me−TFA. Since the ease of activation decreases CH3 OH > CH4 > Me−TFA, it is important to consider the relative chemical potential of CH3 OH and Me−TFA in solution. As we have found that the − * equilibrium Me−TFA(g) + H2 O(solv) − ) − − CH3 OH(g) + TFAH(`) is exergonic by 1.3 kcal/mol at 298 K, it is predicted that these catalysts would be selective for methane oxidation only in the absence of water. In comparison with previous work, we note that Periana et al. [24] showed that the reason why the Catalytica-Periana bipyrimidine Pt catalyst [25] was able to achieve high selectivity is that the barrier to activate the Me−OSO3 H product is 14 kcal/mol higher than for CH4 or CH3 OH and the acidity of the medium drives the protection of methyl products via esterification. Although 98 Figure 3.13. Comparison of the activation states of methane and methyl trifluoroacetate by Rh(DPES) complexes, showing the product protection afforded. Red denotes the most favored isomer of each type of activation. All free energies are in kcal/mol and relative to the resting state. 99 Figure 3.14. Comparison of the activation states of methane and methyl trifluoroacetate by the Rh(bisqx ) family of complexes, showing the product protection afforded. Red denotes the most favored isomer of each type of activation. The overall product protection ∆∆G‡ energies are shown in blue. All free energies are in kcal/mol and relative to the resting state. 100 the Catalytica-Periana system distinguishes between methane and the methyl ester more than the rhodium complexes, we note that a key advantage to our system is the volatility of CH3 OH and Me−TFA (boiling points at 64.7◦ C and 43◦ C, respectively) relative to the nonvolatile Me−OSO3 H. Increasing selectivity beyond what is provided by the electrophilicity of the metal and esterification is still an important area of research. The addition of nonpolar ligand side-chains may decrease the propensity of methanol coordination to the metal relative to methane coordination. 3.5 Discussion Since our ligands have poor electron-donating ability and RhIII -centered cycles, we expect the metal center to be electrophilic with reactions proceeding along an electrophilic route [3]h. We can rule out oxidative addition due to the nature of our ligands and the general instability of RhV species; hence we expected that the most likely C−H activation pathway is through a base-mediated electrophilic mechanism (Scheme 3.21). Scheme 3.21. A schematic diagram showing the metal activation of an R−H bond, for example, methane. Activation occurs via ligand donation to the electrophilic metal center (a) and basic abstraction of the hydrogen (b). Because we propose an electrophilic pathway for the methane activation step [3]h, we expect there to be a relationship between the activation transition state energy and some aspect of the metal’s electropositivity. To find this relationship, we examined more deeply the calculated constituent x x energies of rhodium complexes with the (NN ), (ONN ), (DP M E S), and (PN) ligands and ligand families. We did not include the L3 (bisqx ) family because they require protonation before methane activation can take place; as a result a direct comparison of these ligands with the neutral transition states of our other ligands is not possible. To estimate the electropositivity, we initially used the calculated energy of the rhodium 4s core orbital in the RhIII resting state, which we compared to the corresponding methane activation energies (Figure 3.15). We found that there is approximately a positive correlation between increasing 101 electropositivity (as measured by negative numbers closer to zero) and increase overall transition state barriers. We explain this by noting that decreasing the metal’s electropositivity increases its electrophilicity and thus encourages σ donation of the Me−H bond. Figure 3.15. Graph of the methane activation energy of the various rhodium complexes vs. the 4s orbital energy of the rhodium in their resting states, based on data from Table 3.3. This calculated value is a proxy for the relative overall electropositivity of the specific rhodium-ligand complexes. In all cases the lowest transition state was used; this involves an axial Rh−Me being formed in the cases of the (NNx ) and (ONNx ) ligand families, and an equatorial Rh−Me being formed in the (DP M S) and (PN) cases. E In the hopes that a more directional aspect of a given rhodium-ligand’s electron density might give us a better correlation with the transition state barrier, we then plotted the methane activation energies with the energies of the Rh−C bond in the resultant RhIII −Me intermediate (Figure 3.16). These Rh−C bond energies were obtained by performing a Pipek-Mezey orbital localization procedure on the RhIII −Me complex’s electronic wavefunction [13]. We found that the correlation is much better, with increasingly negative Rh−C bond energies being correlated with decreasing transition state barriers. This may be due to the increasing favorability of forming the rhodium-methyl bond. We give the caveat that again the only outlier is the Rh(NN) ligand complex, but we note that our trend represents only an overestimation, and therefore does not include false positives. In addition, F F there is a definite positive correlation within a ligand family (i.e. Rh(NN) vs. Rh(NN ); Rh(ONN ) NMe2 vs. Rh(ONN) vs. Rh(ONN )). We conclude from this relationship that (a) given a rhodium-ligand complex that already shows some promise, we can further fine-tune its properties with additional modifications on the ligand to further lower the transition state barriers of interest; and (b) given a new rhodium-ligand complex, by calculating the RhIII −Me intermediates and extracting the Rh−C bond energy, we can get an estimate of the expected methane activation transition state barrier for that ligand set and thus gauge its worthiness for further investigation. 102 Ligand (NN) F (NN ) (ONN) F (ONN ) NMe2 (ONN ) (DPMS) (DPES) (PN) TSa 32.6 27.6 37.2 35.8 38.7 33.9 35.6 35.0 ∆Gbprecursor 1.3 2.2 3.9 4.0k 3.7 0.0 0.0 N/Am Precursorc 2– + + [Rh (NN)(TFA)4 ] (Hax )(Heq ) III F 2– + + [Rh (NN )(TFA)4 ] (Hax )(Heq )j III – + [Rh (ONN)(TFA)3 ] (Hax ) III F – + [Rh (ONN )(TFA)3 ] (Hax )l III NMe2 – + [Rh (ONN )(TFA)3 ] (Hax ) III – + [Rh (DPMS)(TFA)3 ] (Heq ) III – + [Rh (DPES)(TFA)3 ] (Heq ) III – + l [Rh (PN)(TFA)3 ] (HN ) Ligand (NN) F (NN ) (ONN) F (ONN ) NMe2 ) (ONN (DPMS) (DPES) (PN) Adj. TSd 31.3 25.4 33.3 31.8 35.0 33.9 35.6 N/Am Rh−C σ e −0.38096 −0.39557 −0.37870 −0.38498 −0.37739 −0.38700 −0.38539 −0.38515 Rh−Mef III 2– + + [Rh (NN)(Meax )(TFA)3 ] (Hax )(Heq ) III F 2– + + [Rh (NN )(Meax )(TFA)3 ] (Hax )(Heq )j III – + [Rh (ONN)(Meax )(TFA)2 ] (Hax ) III F – + [Rh (ONN )(Meax )(TFA)2 ] (Hax ) – + III NMe2 )(Meax )(TFA)2 ] (Hax ) [Rh (ONN III – + [Rh (DPMS)(Meeq )(TFA)2 ] (Heq ) III – + [Rh (DPES)(Meeq )(TFA)2 ] (Heq ) III – [Rh (PN)(MeN )(TFA)3 ] (HP ) III Rh−TFA 4sg −3.1858 −3.21571 −3.19023 −3.19724 −3.18448 −3.20873 −3.20332 −3.19499 Rh−Me 4sh −3.13265 −3.15195 −3.13949 −3.14661 −3.13750 −3.15703 −3.15567 −3.15474 ∆4si 0.05315 0.06376 0.05074 0.05063 0.04698 0.05170 0.04765 N/Am Table 3.3. Comparison of the various Rh parameters with the rhodium-ligands complexes’ associated transition state barriers. These numbers are plotted in Figures 3.15, 3.16, and 3.17. An “ax” subscript indicates that the moiety is axial with respect to the rhodium-ligand plane; An “eq” subscript indicates that the moiety is equatorial with respect to the rhodium-ligand plane. For (DP M S) the rhodium-ligand plane is E defined as the plane spanned by the N−Rh−N moiety. For (PN) complexes, An “N” subscript indicates that the moiety is equatorial with respect to the P−Rh−N plane and trans to the N; A “P” subscript indicates that the moiety is equatorial with respect to the P−Rh−N plane and trans to the P. (a) The overall transition state barrier, as reproduced from Table 3.2, in kcal/mol. (b) The free energy of the immediate precursor before the lowest methane activation barrier, in kcal/mol. This typically differs from the resting state only in the geometric placement of TFAH and TFA ligands. (c) The identity of the immediate precursor. (d) The “adjusted” TS barrier, i.e. the free energy change of the actual transition state itself, in kcal/mol. Equal to the overall TS − Gprecursor . (e) The energy of the Pipek-Mezey localized Rh−C bond of the RhIII −Me species formed as a result of the lowest transition state, in hartrees. (f) The identity of the RhIII −Me species formed as a result of the lowest transition state. (g) The 4s orbital energy of the Rh atom in the resting state, in hartrees. (h) The 4s orbital energy of the Rh atom in the RhIII −Me species in column f, in hartrees. (i) The change in Rh 4s energies, equal to the difference between the preceding two columns, in hartrees. (j) These species are the lowest neutral species, done for the sake of consistency with the other ligands. The anionic resting states are very similar in energy. (k) An estimated value based on the Gprecursor values of related species Rh(ONN) and Rh(ONNNMe2 ). (l) Stable conformations of these complexes could not be found due to the ease of isomerization of the TFAH/TFA ligands into axial/equatorial positions. (m) The value is unknown and an estimation was not attempted. 103 Figure 3.16. Graph of the methane activation energy of the various rhodium complexes vs. the Rh−C bond energy as localized by the Pipek-Mezey method [13] of the lowest-energy Rh−Me complex. This calculated value is a proxy for the relative directional electropositivity of the specific rhodium-ligand complexes towards the methyl bond being formed. Based on data from Table 3.3. In all cases the lowest transition state was used; this involves an axial Rh−Me being formed in the cases of the (NNx ) and (ONNx ) ligand families, and S) and (PN) cases. an equatorial Rh−Me being formed in the (DP M E We also wished to see if there was a correlation between the transition state barrier and the rhodium center’s hardness as modified by its ligand. Since the hardness of an atom or molecule is defined as the polarizability of its electron cloud [26], we measured the difference in the Rh 4s orbital energy going from the inorganic resting state to the RhIII −Me as a proxy for hardness. Although we did not find a correlation with the overall transition state barrier of activation, we did find a negative correlation with the “adjusted” TS barrier, i.e. the free energy change of the actual transition state itself (Figure 3.17). In other words, the activation transition state [Rh· · · CH4 ]‡ is derived from the replacement of a TFAH ligand with CH4 , but the immediate Rh−TFAH precursor is not the resting state, but differs slightly in its axial/equatorial arrangement of TFAH/TFA ligands and is a few kcal/mol higher in energy. Taking this out gives us our correlation. Hence, we see that increased polarizability or softness, as measured by the amount of energy increase in the rhodium-ligand system when a TFA ligand is exchanged for a methyl, correlates with a lower activation barrier. The result of our fine tuning has yielded, among those rhodium-ligand complexes with neutral F activation transition states, the (NN ) ligand set. With a methane activation energy of 27.6 kcal/mol at 298 K, it compares favorably with other homogeneous systems. Indeed, an effective activation barrier of 33.5 kcal/mol is derived from the turnover frequency of the Catalytica-Periana catalyst (bpym)PtCl2 in H2 SO4 [27]. Specifically, C−H activation by (η 3 -6-phenyl-2,20 -bipyridine)IrIII (TFA)(C2 H4 )(C2 H5 ) is shown to catalyze H/D exchange between CH4 and TFAD with a turnover frequency 104 Figure 3.17. A graph of the Rh−TFA to Rh−Me change in the 4s orbital energy, a measure of the Rh atom’s softness, versus the adjusted TS barrier (see text). Note that the (ONNF ) data point is an estimated value, which may explain its deviation from the monotonicity (see Table 3.3). In all cases the lowest transition state was used; this involves an axial Rh−Me being formed in the cases of the (NNx ) and (ONNx ) ligand families, and an equatorial Rh−Me being formed in the (DP M S) and (PN) cases. This graph is based on E data from Table 3.3. of 2.12 × 10−2 s−1 at 180◦ C (Scheme 3.18). From the turnover frequency a free energy barrier of 33.5 kcal/mol can be calculated using transition state theory [28]. In the course of these studies we found that it was more difficult to find effective routes of Rh-methyl functionalization than to find low methane activation barriers. We found that the III-II pathway (SR 2 attack with OVCl3 ) typically gives the lowest barriers, whereas the III-I and III-IVII pathway (SN 2 attack on RhIII or RhIV , respectively) have some use as well; the I-III pathway (oxidative addition of methane by RhI ) was much less likely. F We found that Rh(NN ) complexes in water lead to a flatter thermodynamic profile than in TFAH, increased transition state energies for methane activation, and decreased transition state energies for functionalization. We can understand these changes in the thermodynamics by noting that hydroxo ligands’ lone pairs stabilize higher oxidation states of rhodium. Overall these changes are favorable for activity, since the functionalization energy is lowered to 31.7 kcal/mol while the methane activation energy of 35.0 kcal/mol remains acceptable. Unfortunately, an aqueous solution is predicted to offer no product protection, a result consistent with the reactivity of platinum diimine catalysts [23]. We can understand the decreased barrier for SN 2 methane functionalization (III-I, III-IV-II) in terms of several factors: the increased stability of the RhIV starting material in water, the greater thermodynamic activity of water in aqueous solution (at 55 M) as opposed to the TFA anion in 105 Figure 3.18. The calculated free energy (at 298 K, 1 atm CH4 ) of the (NNC)Ir(TFA)2 system, via the same internal substitution mechanism applied to RhIII complexes. TFAH, and the fact that the transition state involves a concerted reaction featuring simultaneous formation of a C−O bond, deprotonation to form neutral methanol, and protonation of a hydroxo ligand on the Rh complex. However, we remain unclear about the reason behind the decreased SR 2 functionalization barrier for III-II. We speculate that, since some transition states for SN 2 attack by OVCl3 (to form Me−OVCl–3 ) were found and appear to be about 10 kcal/mol higher than their SR 2 equivalents, there may be nucleophilic character on the oxygen atom in OVCl3 SR 2 attack as well. The most important hurdle that must be cleared experimentally in order for us to have a viable catalytic system is catalyst stability. The trend we have observed, that RhIII complexes with less electron-donating ligands tend to have lower activation and functionalization barriers, must be balanced with the consideration that a ligand that is too electron-poor may not have sufficient binding strength for a stable complex with rhodium to be made. The next logical step of our investigation is F the experimental synthesis and stability study of the RhIII (NN ) and RhIII (bisq) complexes. However, even if we fail to see good robustness for this particular complex, our more valuable contribution is that we have undertaken the most detailed ab initio study of the requirements of a Rh methane activation catalyst to date. 3.6 Conclusions F In our QM virtual screening of potential methane to methanol catalysts, we identified the Rh(NN ) F and Rh(bisq) complexes as highly promising candidates. Features of the Rh(NN ) complex include: • Transition state barriers for methane activation at 298 K of 27.6 (TFAH) and 35.0 kcal/mol (water), and • Transition state barriers for functionalization at 298 K of 36.8 (TFAH) or 31.7 kcal/mol (water), 106 and features of the Rh(bisq) complex include: • Transition state barriers for methane activation at 298 K of 33.4 (TFAH), and • Transition state barriers for functionalization at 298 K of 32.0 (TFAH). These results are better than any other RhIII complex we have so far investigated. Moreover, we provide a rational accounting for the role that ligand design plays in the effectiveness of this promising catalyst. In particular for RhIII complexes with neutral methane activation transition states, increased electron-withdrawing ligands leads to lower barriers for both activation and functionalization. We can estimate the overall transition state barrier by simply calculating the RhIII −Me species and extracting the Rh−C bond energies. Thus, the only condition opposing x even less donating ligands (i.e. lower barriers) is catalyst stability. For the RhIII (bisq ) family of complexes, requiring one coordination site to be a weak η 2 -benzene interaction greatly facilitates functionalization on a methyl group trans to it. These results from QM virtual screening are now ready for experimental testing, validation, and improvement. It can also be anticipated that future design of addition ligands may incorporate both concepts, with a weak interaction as well as an anionic or bidentate ligand, so that both activation barriers can be easily prediction and functionalization barriers minimized. 3.7 Acknowledgments This work was supported as part of the Center for Catalytic Hydrocarbon Functionalization, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DOE DE-SC0001298. 107 3.8 References [1] Cheng, W-H.; Kung, H. H. Methanol Production and Use (Chemical Industries vol. 57), 1994. [2] (a) Labinger, J. A.; Bercaw, J. E. Nature 2002, 417, 507–514; (b) Crabtree, R. H. J. Chem. Soc., Dalton Trans. 2001, 2437–2450; (c) Labinger, J. A. Fuel Proc. Technol. 1995, 42, 325–338; (d) Labinger, J. A. J. Mol. Catal. A: Chem. 2004, 220, 27–35. [3] (a) Muehlhofer, M.; Strassner, T.; Hermann, W. A. Angew. Chem. Int. Ed. 2002, 41, 1745–1747; (b) Sen, A.; Benvenuto M. A.; Lin M. R.; Hutson A. C.; Basickes N. J. Am. Chem. Soc. 1994, 116, 998–1003. [4] (a) Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154–162; (b) Hanson, S. K.; Heinekey, D. M.; Goldberg, K. I. Organometallics 2008, 27, 1454–1463; (c) Evans, M. E.; Jones, W. D. Organometallics 2011, 30, 3371–3377; (d) Zakzeski, J. J.; Bell, A. T. J. Mol. Cat. A: Chem. 2007, 276, 8–16; (e) Zhang, X. X.; Wayland, B. B. J. Am. Chem. Soc. 1994, 116, 7897–7898; (f) Tenn, W. J. III; Conley, B. L.; Bischof, S. M.; Periana, R. A. J. Organomet. Chem. 2011, 696, 551–558; (g) Rhinehart, J. L.; Manbeck, K. A.; Buzak, S. K.; Lippa, G. M.; Brennessel, W. W.; Goldberg, K. I.; Jones, W. D. Organometallics 2012, 31, 1943–1952; (h) Li, L.; Brennessel, W. W.; Jones, W. D. Organometallics 2009, 28, 3492–3500; (i) Kloek, S. M.; Heinekey, D. M.; Goldberg, K. I. Angew. Chem. Int. Ed. 2007, 46, 4736–4738. [5] Jaguar, version 7.6; Schrödinger, LLC: New York, NY, 2007. 108 [6] (a) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.; Sitkoff, D.; Nicholls, A.; Ringnalda, M.; Goddard, W. A.; Honig, B. J. Am. Chem. Soc. 1994, 116, 11875–11882; (b) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.; Ringnalda, M. N.; Sitkoff, D.; Honig, B. J. Phys. Chem. 1996, 100, 11775–11788. [7] (a) Becke, A. D. Phys. Rev. A 1998, 38, 3098–3100; (b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652; (c) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B 1988, 37, 785–789. [8] (a) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215–241; (b) Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157–167. [9] Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310. [10] Martin, J. M. L.; Sundermann, A. J. Chem. Phys. 2001, 114, 3408–3420. [11] (a) Hehre, W. J.; Ditchfield, R; Pople, J. A. J. Chem. Phys. 1972, 56, 2257–2261; (b) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; Defrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654–3665. [12] (a) Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. V. J. Comput. Chem. 1983, 4, 294–301; (b) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72, 650–654. [13] Pipek, J.; Mezey, P. G. J. Chem. Phys. 1989, 90, 4916–4926. [14] Kelly, C. P.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2006, 110, 16066–16081. [15] Tissandier, M. D.; Cowen, K. A.; Feng, W. Y.; Gundlach, E.; Cohen, M. H.; Earhart, A. D.; Coe, J. V.; Tuttle, T. R. J. Phys. Chem. A 1998, 102, 7787–7794. Calculated at standard conditions. [16] Wertz, D. H. J. Am. Chem. Soc. 1980, 102, 5316–5322. [17] Wagman, D. D.; Evans, W. H.; Parker, V. B.; Schumm, R. H.; Halow, I.; Bailey, S. M.; Churney, K. L.; Nuttall, R. L. J. Phys. Chem. Ref. Data 1982, 11, 1–14. [18] For trifluoroacetic acid: Kreglewski, A. Bull. Acad. Pol. Sci. Ser. Sci. Chim. 1962, 10, 11-12, 629–633. For water: (a) Bridgeman, O.C.; Aldrich, E.W. J. Heat Transfer 1964, 86, 279–286; (b) Liu, C.-T.; Lindsay, W.T., Jr. J. Chem. Eng. Data 1970, 15, 510–513. 109 [19] Holleman, A. F.; Wiberg, E.; Wiberg, N. Lehrbuch der Anorganischen Chemie, (91–100 ed.;) Walter de Gruyter, 1985; pp. 1056–1057. [20] Goldberg, K. I.; Yan, J. Y.; Breitung, E. M. J. Am. Chem. Soc. 1995, 117, 6889–6896. [21] Cheng, M. J., Nielsen, R. J., Goddard, W.A. Manuscript under preparation.   ‡ . [22] Using k = kBhT · exp −∆G kB T [23] Owen, J. S.; Labinger, J. A.; Bercaw, J. E. J. Am. Chem. Soc. 2006, 128, 2005–2016. [24] Periana, R. A.; Taube, D. J.; Evitt, E. R.; Loffler, D. G.; Wentrcek, P. R.; Voss, G.; Masuda, T. Science 1993, 259, 340–343. [25] Periana, R. A.; Taube, D. J.; Gamble, S.; Taube, H.; Satoh, T.; Fujii, H. Science 1998, 280, 560–564. [26] Anslyn, E.; Dougherty, D. Modern Physical Organic Chemistry; University Science Books, 2006. [27] Young, K. J. H.; Oxgaard, J.; Ess, D. H.; Meier, S. K.; Stewart, T.; Goddard, W. A.; Periana, R. A. Chem. Commun. 2009, 3270–3272. [28] According to transition state theory, the activation energy ∆G‡ may be derived from the   ‡ . turnover frequency TOF using the formula TOF = kBhT · exp −∆G kB T 110 Chapter 4 The Reduction-Coupled Oxo Activation mechanistic motif for C−H bond activation and oxidation Reproduced in part with permission from: 1. Cheng, M.-J.; Fu, R.; Goddard, W. A. III. Chem. Commun. 2014, 50, 1748–1750. c 2014 The Royal Society of Chemistry; available at http://dx.doi.org/10.1039/C3CC47502F; 2. Cheng, M.-J.; Goddard, W. A. III.; Fu, R.; Nielsen, R. J. Top. Catal., in press. Unpublished work c 2014 Springer. Also includes original work. 4.1 Abstract The vanadium phosphorus oxide (VPO) catalyst is used commercially for the selective oxidation of n-butane to maleic anhydride. Surprisingly, the mechanism has been shown to initiate with C−H activation by a P−O moiety. In this chapter, we call this phenomenon the Reduction-Coupled Oxo Activation (ROA) mechanistic motif, and attribute it to the concomitant formation of a strong oxo-hydrogen bond on the phosphorus atom and a one-electron reduction on the vanadium atom. In addition, we show that the same ROA phenomenon can be extended beyond phosphorus and vanadium to additional main-group elements and transition metals, respectively, and may lead to a new strategy in designing selective catalysts for alkane activation. Finally, we propose the existence of a homogeneous vanadium phosphorus oxo complex that can incorporate the novel ROA mechanistic pathway as a potential propane oxidation catalyst. We considered several oxidation pathways leading to end products such as propylene, isopropanol, and propylene oxide. We support our assertions with density functional theory calculations on potential reaction pathways. 111 4.2 Introduction 4.2.1 The VPO catalyst for n-butane oxidation Vanadium phosphorus oxide (VPO) is an inorganic complex that catalyzes the oxidation of nbutane to maleic anhydride with a surprisingly high selectivity of 60-70% [1] (Scheme 4.1). Although the overall yield is only ca. 50%, its ease of synthesis, use of common elements, and selectivity have allowed this catalyst to be commercialized, producing ca. 500 kilotons of maleic anhydride annually [2]. Hence, there was a strong impetus to investigate the mechanism of this reaction in the hopes of rationally designing a modified catalyst with improved yield. Scheme 4.1. The oxidation of n-butane to maleic anhydride using vanadium phosphorus oxide as a catalyst. A useful parameter for the C−H activating ability of each oxo moiety in the catalyst is the H-atom affinity DH , which is defined as the enthalpy change when the O−H bond in ·M−O−H is homolytically cleaved to form the M−O oxo. Mechanistic work initially centered on vanadyl pyIV rophosphate ((V O)2 (P2 O7 ), abbreviated VOPO), the major component of VPO [3]. However, it was quickly found that no vanadium oxo on the VOPO surface has sufficient H-atom affinity to activate n-butane (Figure 4.1). The most active V−O in VOPO has a DH value of 58.2 kcal/mol, whereas the weakest C−H bond in n-butane has a dissociation energy of 89.3 kcal/mol, implying an extremely endothermic first step that is categorically inconsistent with the experimentally determined activation energy of 12.9-23.6 kcal/mol [4]. Hence, it was concluded that the reduced vanadyl IV pyrophosphate (V O)2 (P2 O7 ) alone is unable to initiate the catalytic process. Figure 4.1. O−H bond strength for oxygen atoms on the (VIV O)2 (P2 O7 ) surface. The energy released when an H · is bound to the V−O (in red) is only 58.2 kcal/mol, not nearly high enough for alkane activation to occur. The energy released when an H · is bound to the bridging V−O−P oxide (in green) is even lower, at 17.0 kcal/mol. The energy released when an H · is bound to the P−O was not determined because geometry optimization caused it to be transferred to the V−O−P moiety (in green). It was then supposed that catalysis requires the presence of adsorbed O2 on the VIV O(P2 O7 ) 112 surface. Adsorption of O2 may be molecular, in which the O−O bond is retained, two VIV centers are oxidized to VV , and a vanadium peroxo moiety is formed; or it may be dissociative, in which the O−O bond is broken and four VIV centers are oxidized to VV . The first case is uphill by 2.8 kcal/mol, whereas the second is downhill by 38.8 kcal/mol. Unfortunately, the DH values calculated for the resulting vanadium peroxo and vanadium oxo moieties were only 70.0 and 58.9 kcal/mol, respectively. Hence, it was concluded that the reduced VPO complex is not the active catalyst. Figure 4.2. Oxygen adsorption energies on the (VIV O)2 (P2 O7 ) surface and the O−H bond strength (DH ) to each of these O atoms. Left: molecular O2 adsorption in which the O−O bond is preserved. Right: dissociative adsorption in which the O−O bond is broken. All energies are in kcal/mol. Finally, it was proposed that catalysis takes place on a metastable VV OPO4 surface, known as the X1 phase, which has been experimentally determined to be produced in the high oxygen environment of the reaction conditions [5]. However, the largest DH value found for a vanadium oxo in this phase is only 62.8 kcal/mol (Figure 4.3). Surprisingly though, it was discovered that the phosphorus oxo had a much larger DH value of 84.3 kcal/mol! This implies a first step that is endothermic by only 5.0 kcal/mol, with an activation energy later calculated to be 13.6 kcal/mol, which is within the experimental range. The stability of the resulting monohydrogenated species was attributed to the fact that spin density appears to be localized on the neighboring vanadium atoms (Figure 4.4). 4.2.2 The ROA mechanistic motif The surprising discovery of the C−H activation ability of PV −O moieties in the VV OPO4 catalyst led to the creation a new term, Reduction-Coupled Oxo Activation (ROA), to describe this phenomenon. Its usefulness is due to the observation that the basicity of simple transition metal oxos increases with decreasing electron affinity of the metal, due to increasing electron density; whereas for C−H activation it is preferred that both basicity and electron affinity are maximized, since H-atom transfer 113 Figure 4.3. O−H bond strengths for the various oxygen atoms on the X1-VV OPO4 surface. The O−H bond strength of 84.3 kcal/mol to the P−O moiety leads to a first step that is endothermic by only 5.0 kcal/mol, with an activation energy later calculated to be 13.6 kcal/mol. Hence, the P−O moiety is the only candidate site whose C−H activation energy is compatible with experimental data. Figure 4.4. C−H activation by a phosphorus oxo leads to a one electron reduction on the neighboring vanadium. Hence the oxidation state of the phosphorus atom does not change. (a) Vanadium-phosphate archetype: conceptual arrow-pushing scheme for P−V electron transfer with an intervening µ-oxo. (b) Vanadium-κ−Pphosphinite archetype: conceptual arrow-pushing scheme for P−V electron transfer in which they are directly bound to each other. 114 involves both proton and electron transfer. The ROA mechanistic motif, by keeping the sites of basicity and reduction on separate moieties, is able to decouple this process and hence allow the best of both worlds with high basicity and high electron affinity [6]. The validity of the ROA mechanistic motif was tested using quantum mechanical calculations on small X4 O10 clusters, where each X is a P or V. The results, shown in Figure 4.5, indicate that a single linkage to V makes the P−O bond 32.3 kcal/mol more active, while coupling to a second V makes it an additional 23.8 kcal/mol more active and coupling to a third V (as in VPO) makes it an additional 4.9 kcal/mol more active. Compared to a pure vanadium (V) oxide case, coupling to one or two P atoms makes it 13.4 and 8.5 kcal/mol more active, respectively. Figure 4.5. Use of finite cluster models to investigate how the O−H bond strength of P−O moieties varies based on the number of V coupled to it via O linkages. Data is summarized from Figure 4.8. All numbers are in kcal/mol. a. b. c. d. e. Species P4 O10 P3 VO10 P2 V2 O10 PV3 O10 V4 O10 DPO(1)−H 28.2 60.5 84.3 89.2 75.8 The data presented in Figure 4.5 shows the validity of the small cluster approach in gauging the activity of the ROA mechanism. In hopes of expanding the chemical space in which this ROA mechanism is available, we have undertaken an investigation of mixed oxos with vanadium and the heavier group 15 (pnictogen) and 16 (chalcogen) elements in the first part of this chapter. (We do not consider group 14 or 17 elements due to the instability of terminal oxos for such elements, against hydration in the group 14 case and explosion in the group 17 case.) For each of these complexes, we have calculated DH , the oxo-hydrogen bond energy. We use DH as a proxy for hydrogen-abstraction potential, which is typically the most difficult step in alkane oxidation/functionalization catalysis. We have found that the ROA mechanism is general to group 15 and 16 (pnictogen and chalcogen) vanadium mixed oxos, with phosphorus, selenium, and tellurium being the most active. We have also taken these most active cases, with molybdenum (VI) substituted for vanadium (V), and found similarly promising results. We hence conclude that the reduction-coupled oxo activation mechanistic motif is a viable and interesting new lead for the development of new oxidation catalysts. 115 4.2.3 Potential homogeneous catalysts utilizing the ROA principle Although the ROA phenomenon was originally discovered in the VPO system, a heterogeneous catalyst, the principle should hold for homogeneous catalysts as well. Homogeneous catalysts have the advantages of being well-defined, easier to study, and thus more readily able to contribute to basic scientific research. However, homogeneous complexes containing a P(O)−O−V(O) moiety (as shown in figure 4.4a) are very rare in the literature, owing to their difficulty in synthesis [7]. The only known examples are complexes produced by Herron et al. [8] and Cummins et al. [9], both of which were simply precursors to the production of VOPO or related network complexes; and the very recent work by Doyle et al. [7] that features the divanadium pyrophosphate complex IV {[(V O)bipy(H2 O)]2 (µ−P2 O7 )} · 3 H2 O, which exhibits some catalytic oxidative ability with benzyl alcohol. Hence, potential catalysts incorporating an alternative metal core featuring a direct V−P bond (as in figure 4.4b) were explored. Complexes featuring such κ−P phosphinite coordination to metals including Co [10], Ni [11]a, Pd [11], and Pt [12]a have been described. In addition, many other complexes containing a phosphido ligand bonded to metals such as Co [10], Rh [12]b, Pd [12]bc, Ir [13], and Pt [12]abd have also been described. These phosphido complexes are a single monooxygenation step away from the corresponding phosphine oxide complexes, a transformation which is expected to be facile due to the large formation energy of phosphine-oxo bonds and which has in fact been reported for a few of the complexes [10, 12]a. However, since in all these complexes the metal center has been from the late row transition metals, they are not predicted to be effective alkane activation catalysts under the ROA mechanism due to their large number of d electrons. In order to design a homogeneous oxidation catalyst that utilizes the ROA mechanistic motif, a natural choice for the metal center of a phosphine-oxide-type complex would once again be vanadium (V), due to its activity in the VPO catalyst and its status as an effect d0 one-electron acceptor. The existing metal κ−P phosphinite and phosphido complexes all feature shared traits such as a late transition metal in a low oxidation state and a phosphorus moiety incorporated as the center of a tricoordinate pincer ligand with flanking phosphines and aryl linkers. Such ligands, which were designed for late transition metals, would likely not form very stable complexes with an early transition metal such as vanadium. However, replacement of the ligand’s side phosphines with harder oxygens would be expected to greatly improve stability. Hence, we propose the ligand bis(2-phenoxyl)phosphinite κ−P coordinated on vanadium, abbreviated (OPO)V, as our candidate catalyst, in analogy to the existing Pd complex [11] (Figure 4.6). The robustness of this complex is predicted based on factors such as the chelation effect, the stability of O−V bonds, and the rigidity of the ligand, which guards against insertion of small activating molecules into the V−P bond. We complete the first coordination sphere around V by noting that it prefers to be six-coordinate and to contain a strong vanadium oxo bond [8], and we fill in the last open coordination sites with aqua ligands. 116 Figure 4.6. Left: known complexes featuring a phosphinite ligand bound κ−P to a transition metal. Right: A proposed vanadium κ−P phosphinite complex (OPO)VOL2 . In aqueous solution or when water is present in the system, L is H2 O. Note that both fac and mer stereoisomers are possible. We thus propose, in the second part of this chapter, that the hypothetical complex diaqua[bis(2V phenoxy)phosphinite-κ−P]oxidovanadium (V) ((OPO)V O(H2 O)2 ) utilizes the reduction-coupled oxo activation (ROA) mechanistic motif as a catalyst for the oxidation of alkanes in air. In this proposition, we support our assertion with density functional theory (DFT) calculations of the various reaction paths starting from propane, our representative alkane. We describe the most plausible transition state barriers for the conversion of propane and dioxygen to oxidized products such as propylene and isopropanol. Through our theoretical findings we are confident that a realized molecule will hold catalytic ability in a manner heretofore unseen in homogeneous catalysis. 4.3 Materials and methods All quantum mechanical calculations were carried out using the Jaguar software version 7.9 developed by Schrödinger Inc. [14]. Geometry optimizations were carried out on initial guess structures, and vibrational frequencies were gathered to confirm the optimized geometries as intermediates or transition states and to construct a free energy profile. The temperature was set to be 298.15 K. Solvation energies were calculated in water using the PBF Poisson-Boltzmann implicit continuum solvation model [15] in Jaguar, with a dielectric constant of 80.37 [16] and a probe radius of 1.40 Å [17] for water. Two explicit solvating waters per aqua, hydroxo, or oxo ligand were added for more accurate solvation modeling. For our proposed molecular catalysts, geometry optimization and vibrational data were calculated using the B3LYP density functional [18] with a smaller basis set, whereas single point gas-phase and solvated energies were calculated using the same functional and a larger basis set. Here the smaller basis set consists of a modified double-ζ Los Alamos basis set and pseudopotential (LACVP**) [19] for transition metals, the uncontracted Los Alamos valence-only basis set and pseudopotential (LAV3P) [20] for heavy main-group elements (e.g., As, Se, Sb, Te, Bi), and the 117 6-31G** basis set [21] for the other atoms; whereas the larger basis set consists of the triple-ζ Los Alamos basis set and pseudopotential (LACV3P**++) for transition metals, the same LAV3P basis set and pseudopotential for heavy main-group elements, and the 6-311G**++ basis set [21] for the other atoms. Proposed intermediates and transition states were found to have zero and one negative frequencies, confirming their status as local energy minima and saddle points, respectively. The pH of the aqueous solution was set at 0 or 7; this corresponds to the free energy of a solvated proton being set as −270.28 or −279.80 kcal/mol, respectively [22]. The free energy for each molecular species in solution was calculated using the formula G = Egas + ∆Gsolv + ZPE + Hvib + 6kT − T [Svib + γ(Strans + Srot − α) + β] where the last term (with α = 14.3 e.u., β = 7.98 e.u., and γ = 0.54) is an empirical approximation for the change in the translational and rotational entropy of the molecule between the gas phase and the solution phase (due to the finite librational frequencies), derived from Wertz [23]. For gas phase molecules (H2 , O2 , propylene, propane, and for our purposes the isopropyl radical), we assumed that equilibration between the dissolved gas and the headspace occurs at a much faster timescale than the reactions in question; leading to ∆Ggas→solv = 0. Thus, the free energy of such gas molecules can be simply calculated using the formula G = Ggas = Egas + ZPE + Htot − T Stot . For water, the Gibbs free energy was calculated using the formula Gliquid = Egas + ZPE + Htot − T Stot + ∆Ggas→liquid where ∆Ggas→liquid = Gliquid − Ggas (1atm) is the free energy of condensation to liquid from 1 atm gas. We can solve for this by noting that ∆Ggas→liquid = ∆Gexpansion + Ggas→solvation , where ∆Gexpansion = Ggas (P ) − Ggas (1 atm) is the expansion of the gas from 1 atm to the vapor pressure P , and Ggas→solvation is the condensation of gas at vapor pressure P to liquid. Since a liquid is by definition at equilibrium with its vapor pressure Ggas→solvation = 0, and we thus have  ∆Ggas→liquid = Ggas (P ) − Ggas (1 atm) = RT ln P 1 atm  . The vapor pressure of water at 298.15 K is 23.75 mmHg [24]. For a given species X and its monohydrogenated analogue XH · , the DH value is defined as 118 DH = H(X) + H(H · ) − H(XH · ), and the GH value is defined analogously. For our small cluster calculations, only gas phase energies and frequencies were calculated using the smaller basis set. 4.4 Validation of ROA on small oxide clusters We performed our calculations on small oxide clusters (Figure 4.7), which we used as models for large inorganic network systems. Restricting our investigation to small discrete clusters allowed us to perform highly accurate ab initio quantum mechanical computations. Figure 4.7. General models for small oxide clusters. (a) X is a pnictogen in the +V oxidation state; (b) Z is a chalcogen in the +VI oxidation state; (c) Z is a chalcogen in the +IV oxidation state. The oxidation state of each vanadium is +V before hydrogen atom abstraction and either +V or +IV after a hydrogen atom has been adducted. We begin with X = V, an all-transition metal cluster that will serve as a control. The first column of Figure 4.8 shows the structure of both the non- and monohydrogenated forms. Note that the DH value of 75.8 kcal/mol is already significantly high, owing to the naturally effective oxidizing potential of vanadium in the +V oxidation state. Replacing one vanadium atom with phosphorus (Figure 4.8, second column), we find that the DH energy has strengthened by 13.4 kcal/mol, to 89.2 kcal/mol. This is the canonical example of our reduction-coupled oxo activation mechanism. Replacing a second vanadium with phosphorus (Figure 4.8, third column) gives a slightly reduced, but similar DH value of -84.3 kcal/mol. Compared to the experimental results in [3], in which the phorphoryl site has a DH value of 84.3 kcal/mol, 119 Figure 4.8. Optimized structures of the Px V4−x O10 species (first row) (x ∈ {0, 1, 2, 3}) and their schematic representations (second row); optimized structures of the monohydrogenated Px V4−x O10 H species (third row) and their schematic representations (fourth row). Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. 120 these numbers are a good validation for our cluster model. We also examined the P2 V2 O10 H2 cluster, i.e. the dehydrogenated version. As the first row of Figure 4.9 indicates, regardless of the location of the hydrogen atoms, it is always the two vanadiums that are reduced to +IV. However, it is preferable by about 12 kcal/mol to attach the hydrogens onto the phosphoryl groups, due to the special stability of the vanadyl unit. The unpaired spins on each VIV are essentially independent, as there is almost no difference in energy or geometry between the singlet and triplet in both cases. We found that addition of a third hydrogen to the system always reduces one vanadium atom to +III, leading to either an overall doublet (doublet VIV and singlet VIII ) or an overall quartet (doublet VIV and triplet VIII ). The phosphorus atoms, in a sign of their resistance to reduction, stay in the +V oxidation state in all cases. The lack of VV centers compared to the case in Figure 4.8 accounts for the significant reduction in DH values. Figure 4.9. Optimized structures of the dihydrogenated P2 V2 O10 H2 species (first row), their schematic representations, and their energy relative to the lowest isomer; optimized structures of the trihydrogenated P2 V2 O10 H3 species (second row). Only the triplet and quartet structures are shown because the singlet and doublet structures are almost identical to them. Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. Having established the validity of our model, we now endeavored to survey the span of the group 15 (pnictogen) vanadium (V) oxides. We found that when going down the periodic table, from phosphorus to arsenic to antimony, the DH energy decreases (Figure 4.10). Bismuth, the last 121 element in the series, has an DH that is slightly higher than antimony above it, but it is still not as effective as a pure vanadium-oxygen cluster and furthermore is too susceptible to reduction to the +III state. Figure 4.10. Optimized structures of the XV3 O10 species (first row) (X ∈ {P, As, Sb, Bi} and their monohydrogenated XV3 O10 H counterparts (second row). Schematic structures are as in Figure 4.7a. Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. We also examined the potential group 16 (chalcogen) vanadium (V) oxides, in clusters as depicted in Figure 4.7bc. We found that for sulfur, selenium, and tellurium in the maximum +VI oxidation state, all structures had DH values of 90 kcal/mol or greater (Figure 4.11). We view this as a very promising potential direction for new catalysts. It is well documented that tellurium prefers higher coordination numbers; for example, unlike its lighter cogeners, the hydrated H6 TeO6 is stable whereas the four-coordinate H2 TeO4 is unknown. Examining several potential structures for the hydrated complex Te2 V2 O12 H2 , we found that the lowest energy conformation had a hydration energy of 6.3 kcal/mol downhill. The resultant structure (Figure 4.12) has a bridging µ-oxo that renders the two tellurium atoms five coordinate, and a proton on an oxygen atom on each tellurium. The lowest conformation of the corresponding monohydrated complex Te2 V2 O12 H3 gave a DH value of 92.0 kcal/mol, very comparable to the dehydrated complex. Upon hydrogen atom abstraction the µ-oxo breaks, resulting in only one of the tellurium atoms remaining five-coordinate. Unfortunately, it appears that there is a strong drive to fully hydrate the complex until both 122 Figure 4.11. Optimized structures of the Z2 V2 O11 species (first row) (Z ∈ {S, Se, Te}) and their monohydrogenated Z2 V2 O11 H counterparts (second row). Schematic structures are as in Figure 4.7b. Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. Figure 4.12. Optimized structures of the hydrated species Te2 V2 O12 H2 and Te2 V2 O12 H3 , along with their schematic representations. Mulliken spin density was used to assign radical character. DH = 92.0 kcal/mol. 123 tellurium atoms are six-coordinate: Te2 V2 O11 + 3 H2 O −−→ Te2 V2 O14 H6 , ∆E = −49.2 kcal/mol. The six-coordinate complex lacks terminal tellurium oxos that can easily abstract hydrogen atoms. Hence, the DH values are much less, diminished to about 60 kcal/mol. Attachment of a hydrogen atom onto a tellurium hydroxide results in water loss, and is not energetically more favorable than attachment onto a bridging oxo (Figure 4.13). The implication of this is that any putative catalyst incorporating tellurium in the +VI oxidation state will be poisoned by water and will need to be kept in dry conditions. Figure 4.13. Optimized structures of the fully hydrated species Te2 V2 O14 H6 and Te2 V2 O14 H7 (first row), along with their schematic representations (second row). Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. As SeVI and TeVI are both easily reduced to the +IV oxidation state, we also investigated whether vanadium oxide clusters incorporating SeIV and TeIV might also have good hydrogen abstraction capability. Using the structures in Figure 4.7c to build our model, we found that the complexes Se2 V2 O9 and Te2 V2 O9 are also very strong hydrogen atom abstractors (Figure 4.14). In fact, Te2 V2 O9 has a DH value of 100.0 kcal/mol, the strongest value we have seen. We note that abstraction of one hydrogen atom on a tellurium oxo causes the other tellurium oxo to convert to a bridging oxo, rendering the first tellurium five-coordinate. This is not seen in the selenium case. We also investigated the activity of vanadium oxide clusters with the group 14 elements Si, Ge, Sn, and Pb in the +IV oxidation state (Figure 4.15). Construction of a cluster model analogous to the templates found in Figure 4.7 results in the complexes Z2 V2 O10 H2 , where Z = Si, Ge, Sn, or Pb, which are shown in the first row of Figure 4.7. Note that unlike the group 15 and 16 cases, the base Z2 V2 O10 H2 clusters do not have any terminal oxo functionalities. This is reasonable and consistent 124 Figure 4.14. Optimized structures of Se2 V2 O9 and Te2 V2 O9 and their schematic representations (first row); optimized structures of the monohydrogenated Se2 V2 O9 H and Te2 V2 O9 H (second row). Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. 125 with the common knowledge that group 14 elements other than carbon greatly prefer two single bonds with oxygen over one double bond. Unfortunately, the lack of such terminal oxos results in the inapplicability of the ROA mechanism towards these types of compounds. Hence, the location on each Z2 V2 O10 H2 species that results in the greatest amount of energy released when H · is bound is a V−O moiety. The DH are similar to that of V4 O10 (Figure 4.8). Placement of the H · atom on a Z−OH moiety or bridging oxygen resulted in decomposition of the cluster for the cases of Z = Ge, Sn, and Pb; and led to much smaller DH values for the case of Z = Si. Hence we conclude that the ROA mechanistic motif does not likely extend to the group 14 elements. Although our work presented up to now have all involved vanadium as the transition metal that absorbs the spin density, we were also desirous of expanding the breadth of our ROA principle to other transition metals. Hence, we decided to take our best cases, involving phosphorous, selenium, and tellurium, and substitute in MoVI for VV . Unfortunately, we were not able to find a stable structure for monohydrogenated selenium complexes. Our data for phosphorous and tellurium are presented in Figure 4.16. It can be seen from the data in Figure 4.16 that the activity of phosphorus and tellurium (VI) species are comparable to those with vanadium, but the tellurium (IV) species is ineffectual. This may be due to the two tellurium atoms being too far apart for an oxygen atom to bridge in the same manner as the vanadium case. Nevertheless, we have demonstrated that the ROA pathway which characterizes vanadium (V) chemistry can also be extended to molybdenum (VI). Whereas calculations on the periodic system appear to show that the spin is delocalized [3], for our cluster calculations, in all cases the spin is not shared, but localized on one atom. This discrepancy may be explained by noting that those prior calculations were made using the PBE density functional, which has been documented to have a bias towards delocalized systems. Regardless, we were able to see the same increased DH effect without spin delocalization, so that cannot the physical reason behind our ROA mechanism. Given the observation that main group-vanadium oxide clusters with only certain elements have an increased hydrogen abstraction energy relative to a vanadium oxide cluster alone, it would appear that increases in hydrogen abstraction energy via the ROA motif is element-specific and not due to any general chemical principle. For pnictogen vanadium oxide clusters, the only viable candidate was phosphorus, whereas all the chalcogens exhibited similar high activity, especially selenium and tellurium in the +IV oxidation state. These observations cannot be explained by simple periodic trends. Similar complexes with group 14 elements were not included in this chapter due to the lack of notable positive results. This may be for the same reason as why a fully hydrated tellurium (VI) complex was also poor, due to the lack of an oxo ligand. 126 Figure 4.15. Optimized structures of the group 14-vanadium oxide species Z2 V2 O10 H2 , where Z ∈ {Si, Ge, Sn, Pb} (first row); along with their schematic representations (second row). Optimized structures of the corresponding monohydrogenated species (third row) and their schematic representations (fourth row). Alternative locations for H · addition, as well as the corresponding DH energies, are shown in the fifth row for Z = Si. Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. 127 Figure 4.16. Optimized structures of the molybdenum-containing species P2 Mo2 O11 , Te2 Mo2 O10 , and Te2 Mo2 O12 (first row), along with their schematic representations (second row); optimized structures of the corresponding monohydrogenated species (third row) and their schematic representations (fourth row). Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. 128 4.5 Reactions of a homogeneous oxidation catalyst utilizing the ROA mechanistic motif Potential catalysts incorporating κ−P phosphinite coordination to metal, i.e. a direct V−P bond as in Figure 4.4b, were explored. Although the inspiration for our proposed catalyst structure was heuristic in nature, our confidence in its predicted activity is to a large degree bolstered by comprehensive theoretical investigations in the system using density functional theory. In this section we will discuss our results on various potential reactivity pathways, and suggest additional theoretical pathways that ought to be investigated before experimental work commences. One potential pitfall in the installation of the bis(2-phenoxyl)phosphinite (OPO) ligand onto V would be the possibility of the P−O moiety on the ligand coordinating via the oxygen atom to the vanadium. However, at no time in our DFT geometry optimizations, aside from some high energy dianionic five-coordinate cases, did we see this sort of isomerization happen, nor did we see negative frequencies in our calculated Hessian matrices. This implies that the homogeneous vanadium complex with a V−P−O unit is a stable coordination isomer that may be realistically synthesized. 4.5.1 Nomenclature Due to the number of complexes investigated, a systematic shorthand naming convention is desirable. As all of our complexes share the common motif of a vanadium atom bound κ−P to the O−P(−O)−O pincer ligand bis(2-phenoxyl)phosphinite (OPO); we denote this commonality as (OPO)V. If there is an additional R group (such as a hydrogen atom) on the PO moiety of the (OPO), we label it as a subscript: (OPO)R V. If the V atom is five-coordinate, then there are two remaining coordination sites; these are labeled with the s or a superscripts depending on whether they are synperiplanar (syn) or antiperiplanar (anti ) to the bis(2-phenoxyl)phosphinite’s PO moiety, respectively (Figure 4.17, left). If the V atom is six-coordinate, then the (OPO) ligand may be in either a facial (fac) or meridional (mer ) configuration (Figure 4.17, center and right, respectively); and there are three remaining coordination sites. In the fac case, the three sites are labeled with the c, s, and t superscripts, depending on whether they are clinal or syn to the PO moiety of the (OPO), or trans to the P, respectively. In the mer case, the three coordination sites are labeled with the s, t, or a superscripts, depending on whether they are syn to the PO, trans to the P, or anti to the PO, respectively. Under this system, the two specific conformational species shown in the right side of V V m s ta Figure 4.6 could be unambiguously expressed as (OPO)f V Os Lct 2 and (OPO) V O L2 . 129 Figure 4.17. Template for the nomenclature of investigated complexes. For the (OPO) ligand, an f superscript denotes that the chelation is facial whereas an m superscript denotes that it is meridional. For the V, n is the oxidation state. For the ligands Ln , the superscript a denotes that the ligand is anti to the PO moiety; s denotes that it is syn to the PO moiety; c denotes that it is clinal to the PO moiety; and t denotes that it is trans to the P atom. 4.5.2 The resting VV state The first consideration that needs to be investigated is the precise nature of the resting VV state: whether it is five or six coordinated, charged or uncharged, with the (OPO) ligand in a fac or mer conformation, and with what optimal combination of oxo, hydroxo, and aqua ligands. The pH of the environments investigated was set to either 0 or 7. Changing the ambient pH does not affect the relative energetics of isomeric species with the same charge, but does change which charge cohort of species will be most stable. V Table 4.1 is a comprehensive compilation of all (OPO)V −O species studied. The most stable V species at pH 0 is the neutral (OPO)fH V Oc (OH)s (H2 O)t , in which the P−O moiety on the ligand V is already protonated. However, the isomer (OPO)f V Oc (H2 O)st 2 , which does have a P−O moiety available, is the second-lowest energy species at only 1.6 kcal/mol higher. At pH 7, the most stable V 2– species is the doubly deprotonated [(OPO)f V Oc (OH)st 2 ] . The optimized structures of these three species are shown in Figure 4.18. Note that in some fac cases, the PO moiety is twisted enough such that the clinal and syn positions on the V both become gauche. Since this phenomenon was not universal, no relabeling of clinal and syn markers was done. Species Egas Gaq V −1481.9660 (OPO)V Os (H2 O)a V V (OPO)V −O (OPO)V Oa (H2 O)s Grel (pH = 0) Grel (pH = 7) −1481.7587 20.9 29.8 −1711.4245 −1711.1621 2.8 11.7 −1711.4238 −1711.1506 10.0 18.9 130 Species Egas Gaq (OPO)V (OH)2 −1711.4121 −1711.1479 (OPO)H VV Os (OH)a geometry unstable (OPO)H VV Oa (OH)s geometry unstable V + −1711.8126 + geometry unstable [(OPO)H VV Os (H2 O)a ] [(OPO)H VV Oa (H2 O)s ] −1711.5923 Grel (pH = 0) Grel (pH = 7) 8.6 20.5 3.1 21.5 [(OPO)V Os (OH)a ] V – −1710.9247 −1710.7246 7.0 6.4 V – −1710.9181 −1710.7148 13.1 12.5 −1710.3205 −1710.2869 11.4 1.2 (OPO)f V Oc (H2 O)st 2 V −1940.8700 −1940.5384 1.6 10.5 V (OPO)f V Os (H2 O)ct 2 −1940.8658 −1940.5334 4.8 13.7 V −1940.8678 −1940.5327 5.2 14.1 −1940.8696 −1940.5410 0.0 8.9 −1940.8648 −1940.5338 4.5 13.4 −1940.8690 −1940.5363 2.9 11.8 −1940.8598 −1940.5254 9.8 18.7 −1940.8744 −1940.5375 2.2 11.1 −1940.8719 −1940.5372 2.4 11.3 [(OPO)V Oa (OH)s ] V [(OPO)V O2 ] 2– (OPO)f V Ot (H2 O)cs 2 V (OPO)fH V Oc (OH)s (H2 O)t V (OPO)fH V Oc (OH)t (H2 O)s V (OPO)fH V Os (OH)c (H2 O)t V (OPO)fH V Os (OH)t (H2 O)c V (OPO)fH V Ot (OH)c (H2 O)s V (OPO)fH V Ot (OH)s (H2 O)c V – −1940.3696 −1940.0919 11.5 10.9 V – −1940.3858 −1940.0992 6.9 6.3 V – −1940.3795 −1940.0966 8.6 8.0 [(OPO)fH V Oc (OH)st 2 ] [(OPO)fH V Os (OH)ct 2 ] [(OPO)fH V Ot (OH)cs 2 ] V – −1940.3743 −1940.0988 7.2 6.6 V – −1940.3722 −1940.0978 7.8 7.2 V – −1940.3677 −1940.0910 12.1 11.5 [(OPO)f V Oc (OH)s (H2 O)t ] [(OPO)f V Oc (OH)t (H2 O)s ] [(OPO)f V Os (OH)c (H2 O)t ] 131 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) V – −1940.3600 −1940.0874 14.3 13.7 V – −1940.3649 −1940.0887 13.5 12.9 V – −1940.3726 −1940.0916 11.8 11.1 [(OPO)f V Oc (OH)st 2 ] V 2– −1939.7811 −1939.6634 10.1 0.0 [(OPO)f V Os (OH)ct 2 ] V 2– −1939.7864 −1939.6622 10.9 0.7 V 2– −1939.7640 −1939.6517 17.5 7.3 [(OPO)f V Os (OH)t (H2 O)c ] [(OPO)f V Ot (OH)c (H2 O)s ] [(OPO)f V Ot (OH)s (H2 O)c ] [(OPO)f V Ot (OH)cs 2 ] V 2– −1939.7706 −1939.6550 15.4 5.3 V 2– −1939.7651 −1939.6465 20.7 10.6 V 2– −1939.7836 −1939.6595 12.6 2.4 V 2– −1939.7795 −1939.6556 15.0 4.9 V 2– −1939.7662 −1939.6425 23.3 13.1 [(OPO)fH V (OH)t (O)cs 2 ] V 2– −1939.7788 −1939.6546 15.7 5.5 (OPO)m V Os (H2 O)ta 2 V −1940.8672 −1940.5346 4.0 12.9 (OPO)m V Ot (H2 O)sa 2 V −1940.8513 −1940.5180 14.5 23.4 V −1940.8592 −1940.5289 7.6 16.5 V geometry unstable V geometry unstable V geometry unstable V −1940.8548 −1940.5201 13.1 22.0 V −1940.8449 −1940.5093 19.9 28.8 V −1940.8578 −1940.5270 8.8 17.7 −1940.0966 8.6 7.9 [(OPO)f V (H2 O)c (O)st 2 ] [(OPO)f V (H2 O)s (O)ct 2 ] [(OPO)f V (H2 O)t (O)cs 2 ] [(OPO)fH V (OH)c (O)st 2 ] [(OPO)fH V (OH)s (O)ct 2 ] (OPO)m V Oa (H2 O)st 2 s t a (OPO)m H V O (OH) (H2 O) s a t (OPO)m H V O (OH) (H2 O) t s a (OPO)m H V O (OH) (H2 O) t a s (OPO)m H V O (OH) (H2 O) a s t (OPO)m H V O (OH) (H2 O) a t s (OPO)m H V O (OH) (H2 O) V – −1940.3660 V – geometry unstable [(OPO)m V Os (OH)t (H2 O)a ] [(OPO)m V Os (OH)a (H2 O)t ] 132 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) 16.2 15.6 V – geometry unstable V – −1940.3614 V – geometry unstable V – −1940.3724 −1940.0932 10.7 10.1 s ta [(OPO)m H V O (OH)2 ] V – −1940.3355 −1940.0626 29.9 29.3 t sa [(OPO)m H V O (OH)2 ] V – −1940.3573 −1940.0785 20.0 19.3 V – −1940.3447 −1940.0654 28.1 27.5 V 2– −1939.7563 −1939.6389 25.5 15.3 V 2– −1939.7430 −1939.6271 32.9 22.7 V 2– −1939.7472 −1939.6324 29.6 19.4 [(OPO)m V Ot (OH)s (H2 O)a ] [(OPO)m V Ot (OH)a (H2 O)s ] [(OPO)m V Oa (OH)s (H2 O)t ] [(OPO)m V Oa (OH)t (H2 O)s ] a st [(OPO)m H V O (OH)2 ] [(OPO)m V Os (OH)ta 2 ] [(OPO)m V Ot (OH)sa 2 ] [(OPO)m V Oa (OH)st 2 ] −1940.0845 V 2– geometry unstable V 2– geometry unstable V 2– geometry unstable s ta [(OPO)m H V (OH) (O)2 ] V 2– −1939.7499 −1939.6332 29.1 18.9 t sa [(OPO)m H V (OH) (O)2 ] V 2– −1939.7799 −1939.6580 13.5 3.4 V 2– −1939.7503 −1939.6286 32.0 21.8 [(OPO)m V (H2 O)s (O)ta 2 ] [(OPO)m V (H2 O)t (O)sa 2 ] [(OPO)m V (H2 O)a (O)st 2 ] a st [(OPO)m H V (OH) (O)2 ] Table 4.1. Comprehensive compilation of all (OPO)VV species studied, along with their gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous free energies relative to each other, in kcal/mol, at pH 0 and 7, respectively. For some entries, geometry optimization led to the formation of a lower-energy isomer. The energies for these entries are not reported. 4.5.3 Propane activation transition states Propane consists of two types of C−H bonds: primary and secondary, and the secondary C−H bond in the middle carbon is more easily activated due to the increased stability of the resulting secondary V radical. Those (OPO)V species listed in Table 4.1 which contain unadorned intact P−O moieties were further investigated for their ability to activate the secondary C−H bond of propane. The transition states were located and their free energies at pH 0 and 7 relative to the ground (OPO)V V V 2– V states of (OPO)fH V Oc (OH)s (H2 O)t and [(OPO)f V Oc (OH)st 2 ] , respectively, were calculated. The results are shown in Table 4.2. 133 V V Figure 4.18. The geometry-optimized structures of (OPO)fH V Oc (OH)s (H2 O)t and (OPO)f V Oc (H2 O)st 2 , V 2– the most stable (OPO)VV isomers at pH 0; and of the doubly deprotonated [(OPO)f V Oc (OH)st 2 ] , the most stable isomer at pH 7. Species Egas Gaq Grel (pH = 0) Grel (pH = 7) [(OPO)HiPr VV Os (H2 O)a ]‡ −1830.5749 −1830.2107 39.6 48.5 [(OPO)HiPr VV Oa (H2 O)s ]‡ −1830.5628 −1830.1985 47.2 56.1 [(OPO)HiPr VV Os (OH)a ]−‡ geometry unstable [(OPO)HiPr VV Oa (OH)s ]−‡ geometry unstable 2−‡ [(OPO)HiPr VV Osa 2 ] geometry unstable s ta ‡ [(OPO)m HiPr V O (H2 O)2 ] V −2060.0127 −2059.5889 37.2 46.1 t sa ‡ [(OPO)m HiPr V O (H2 O)2 ] V −2060.0129 −2059.5909 36.0 44.9 a st ‡ [(OPO)m HiPr V O (H2 O)2 ] V −2060.0036 −2059.5752 45.9 54.8 V −2059.5055 −2059.1441 46.1 45.5 V −2059.4829 −2059.1232 59.2 58.6 s t a −‡ [(OPO)m HiPr V O (OH) (H2 O) ] s a t −‡ [(OPO)m HiPr V O (OH) (H2 O) ] 134 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) V −2059.4979 −2059.1286 55.8 55.2 V −2059.5033 −2059.1384 49.7 49.1 V −2059.4868 −2059.1247 58.3 57.7 V −2059.4888 −2059.1282 56.1 55.4 s ta 2−‡ [(OPO)m HiPr V O (OH)2 ] V −2058.9071 −2058.6901 60.7 50.6 t sa 2−‡ [(OPO)m HiPr V O (OH)2 ] V −2058.8825 −2058.6799 67.1 57.0 a st 2−‡ [(OPO)m HiPr V O (OH)2 ] V −2058.8811 −2058.6783 68.1 57.9 V −2058.8702 −2058.6684 74.4 64.2 t sa 2−‡ [(OPO)m HiPr V (H2 O) O2 ] V geometry unstable V geometry unstable t s a −‡ [(OPO)m HiPr V O (OH) (H2 O) ] t a s −‡ [(OPO)m HiPr V O (OH) (H2 O) ] a s t −‡ [(OPO)m HiPr V O (OH) (H2 O) ] a t s −‡ [(OPO)m HiPr V O (OH) (H2 O) ] s ta 2−‡ [(OPO)m HiPr V (H2 O) O2 ] a st 2−‡ [(OPO)m HiPr V (H2 O) O2 ] V ‡ [(OPO)fHiPr V Oc (H2 O)st 2 ] V ‡ [(OPO)fHiPr V Os (H2 O)ct 2 ] V ‡ [(OPO)fHiPr V Ot (H2 O)cs 2 ] V [(OPO)fHiPr V Oc (OH)s (H2 O)t ]−‡ V [(OPO)fHiPr V Oc (OH)t (H2 O)s ]−‡ V [(OPO)fHiPr V Os (OH)c (H2 O)t ]−‡ V [(OPO)fHiPr V Os (OH)t (H2 O)c ]−‡ V [(OPO)fHiPr V Ot (OH)c (H2 O)s ]−‡ V [(OPO)fHiPr V Ot (OH)s (H2 O)c ]−‡ V 2−‡ [(OPO)fHiPr V Oc (OH)st 2 ] V 2−‡ [(OPO)fHiPr V Os (OH)ct 2 ] V 2−‡ [(OPO)fHiPr V Ot (OH)cs 2 ] V 2−‡ [(OPO)fHiPr V (H2 O)c Ost 2 ] −2060.0212 −2059.6000 30.3 39.2 −2060.0208 −2059.5954 33.2 42.1 −2060.0229 −2059.5991 30.8 39.7 −2059.5101 −2059.1478 43.8 43.1 −2059.5120 −2059.1514 41.5 40.9 −2059.5096 −2059.1432 46.7 46.1 −2059.5118 −2059.1455 45.2 44.6 −2059.5186 −2059.1506 42.0 41.4 −2059.5170 −2059.1512 41.6 41.0 −2058.9002 −2058.6969 56.4 46.3 −2058.9152 −2058.7010 53.9 43.7 −2058.9121 −2058.6971 56.3 46.2 −2058.8851 −2058.6791 67.6 57.4 135 Species V 2−‡ [(OPO)fHiPr V (H2 O)s Oct 2 ] V 2−‡ [(OPO)fHiPr V (H2 O)t Ocs 2 ] Egas Gaq Grel (pH = 0) Grel (pH = 7) −2058.8771 −2058.6841 64.5 54.4 −2058.8941 −2058.6763 69.4 59.3 Table 4.2. Comprehensive compilation of all the transition states for propane activation by (OPO)VV , along with their gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous free energies, in kcal/mol, relative to the ground states at pH 0 and 7 V V 2– ((OPO)fH V Oc (OH)s (H2 O)t and [(OPO)f V Oc (OH)st 2 ] , respectively). For some entries, geometry optimization led to the formation of a lower-energy isomer. The energies for these entries are not reported. According to Table 4.2, the most favorable transition state at both pH 0 and 7 is the neutral V ‡ species [(OPO)fHiPr V Oc (H2 O)st 2 ] , whose structure is shown in detail in Figure 4.19. At pH 0, V this species is 30.3 kcal/mol above the neutral ground state (OPO)fH V Oc (OH)s (H2 O)t , and with V ∆G‡ = 28.7 kcal/mol from its immediate precursor (OPO)f V Oc (H2 O)st 2 ; we therefore consider it to be accessible for C−H activation. At pH 7, however, the species is 39.2 kcal/mol above the V 2– ground state [(OPO)f V Oc (OH)st 2 ] , as this ground state must be the doubly protonated before C−H activation can occur. Doubly protonating the ground state before C−H activation is still preferable to the dianion directly activating propane; this is due to the fact that the negative charge stabilizes the VV −O moiety and thus decreases its tendency to be reduced to VIV . We conclude from these results that catalysis with the (OPO)V system is much more likely at pH 0. Analysis of the Mulliken spin density supports our assignment of radical character: Whereas the overall spin is 0, the spin density on the vanadium (1.0882), propane’s secondary carbon (-0.5389), and the attacking PO oxygen (-0.2084), but not the phosphorus itself (-0.0350), supporting our assertion of attack by the ROA mechanism. V ‡ Figure 4.19. The geometry-optimized structure of [(OPO)fHiPr V Oc (H2 O)st 2 ] , the most stable transition state for C−H activation of propane at either pH 0 or 7. 136 Monohydrogenated VIV states 4.5.4 The result of the C−H activation transition states as described in the previous subsection is an isopropyl radical and a monohydrogenated (OPO)H VIV complex, which may then isomerize into other IV (OPO)V species. The various (OPO)V IV complexes are listed in Table 4.3. Only six-coordinate complexes with fac-(OPO) were exhaustively investigated, as Table 4.1 shows that they are lower in energy than their five-coordinate and six-coordinate mer counterparts. At pH 0, the lowest energy IV species is (OPO)fH V Oc (H2 O)st 2 at 13.4 kcal/mol above the ground state. However, isomerization IV to (OPO)f V (OH)c (H2 O)st 2 , at 13.6 kcal/mol above the ground state, is nearly thermoneutral. At IV – pH 7, the lowest energy species is [(OPO)f V Os (H2 O)ct 2 ] at 16.8 kcal/mol above the ground state. These species are shown in detail in Figure 4.20. All species are doublets, and Mulliken spin density analysis reveals that the spins are localized on the vanadium atoms, hence supporting the assignment of V in the +IV oxidation state. IV IV f c st Figure 4.20. The geometry-optimized structures of (OPO)fH V Oc (H2 O)st 2 and (OPO) V (OH) (H2 O)2 , the most stable (OPO)V stable isomer at pH 7. 4.5.5 IV f IV s – isomers at pH 0; and of the deprotonated [(OPO) V O (H2 O)ct 2 ] , the most Catalytic cycle involving H-atom abstraction only The simplest catalytic cycle that can now be envisioned is one that alternates only between the V resting state and its corresponding monohydrogenated species, e.g., (OPO)fH V Oc (OH)s (H2 O)t and IV (OPO)fH V Oc (H2 O)st 2 (We will only consider the pH 0 case since it is kinetically feasible with overall ∆G‡ = 30.3 kcal/mol; see Subsection 4.5.3). After monohydrogenation, the starting VV species is regenerated by sequential removal of H+ and e– (Scheme 4.2). First deprotonating the 137 Species (OPO)H VIV (O)s (H2 O)a (OPO)H VIV (O)a (H2 O)s IV – [(OPO)V (O)s (H2 O)a ] IV – [(OPO)V (O)a (H2 O)s ] IV s ta (OPO)m H V (O) (H2 O)2 IV t sa (OPO)m H V (O) (H2 O)2 m IV a (OPO)H V (O) (H2 O)st 2 – IV [(OPO)m V (O)s (H2 O)ta 2 ] IV – [(OPO)m V (O)t (H2 O)sa 2 ] – IV [(OPO)m V (O)a (H2 O)st 2 ] Egas −1712.0701 −1712.0474 −1711.5717 −1711.5440 −1941.5038 −1941.4939 −1941.4894 −1940.9974 −1940.9955 −1941.0011 Gaq −1711.7906 −1711.7653 −1711.3486 −1711.3323 −1941.1599 −1941.1627 −1941.1451 −1940.7170 −1940.7103 −1940.7064 Grel (pH = 0) 16.2 32.1 23.3 33.5 19.5 17.7 28.8 27.1 31.3 33.8 Grel (pH = 7) 25.1 41.0 22.7 32.9 28.4 26.6 37.7 26.5 30.7 33.1 (OPO)f V (OH)c (H2 O)st 2 IV −1941.5108 −1941.1692 13.6 22.5 (OPO)f V (OH)s (H2 O)ct 2 IV −1941.5065 −1941.1636 17.2 26.1 −1941.4999 −1941.1580 20.7 29.6 −1941.4953 −1941.1503 25.5 34.4 −1941.4931 −1941.1497 25.9 34.7 −1941.5068 −1941.1598 19.5 28.4 −1941.5092 −1941.1695 13.4 22.3 −1941.5180 −1941.1663 15.4 24.3 −1941.5093 −1941.1667 15.2 24.1 −1941.0201 −1940.7264 21.2 20.6 −1941.0201 −1940.7325 17.4 16.8 −1941.0125 −1940.7266 21.1 20.5 −1940.9999 −1940.7101 31.4 30.8 −1941.0011 −1940.7205 24.9 24.3 −1941.0136 −1940.7218 24.1 23.5 −1941.0143 −1940.7228 23.5 22.9 −1941.0161 −1940.7247 22.3 21.6 −1941.0095 −1940.7226 23.6 23.0 −1940.4101 −1940.2769 33.0 22.8 −1940.4110 −1940.2804 30.8 20.6 −1940.4079 −1940.2745 34.5 24.4 −1940.4018 −1940.2721 36.0 25.8 −1940.4013 −1940.2790 31.7 21.5 −1940.3952 −1940.2736 35.1 24.9 −1940.3899 −1940.2599 43.6 33.5 −1940.4093 −1940.2757 33.8 23.6 −1940.4031 −1940.2828 29.3 19.2 IV f t (OPO) V (OH) (H2 O)cs 2 IV (OPO)fH V (H2 O)c (OH)st 2 IV (OPO)fH V (H2 O)s (OH)ct 2 IV (OPO)fH V (H2 O)t (OH)cs 2 IV (OPO)fH V (O)c (H2 O)st 2 IV (OPO)fH V (O)s (H2 O)ct 2 IV (OPO)fH V (O)t (H2 O)cs 2 IV – [(OPO)f V (O)c (H2 O)st 2 ] IV – [(OPO)f V (O)s (H2 O)ct 2 ] IV – [(OPO)f V (O)t (H2 O)cs 2 ] – f IV c s [(OPO)H V (O) (OH) (H2 O)t ] IV – [(OPO)fH V (O)c (OH)t (H2 O)s ] IV – [(OPO)fH V (O)s (OH)c (H2 O)t ] IV – [(OPO)fH V (O)s (OH)t (H2 O)c ] IV – [(OPO)fH V (O)t (OH)c (H2 O)s ] IV – [(OPO)fH V (O)t (OH)s (H2 O)c ] IV 2– [(OPO)f V (O)c (OH)s (H2 O)t ] IV 2– [(OPO)f V (O)c (OH)t (H2 O)s ] IV 2– [(OPO)f V (O)s (OH)c (H2 O)t ] IV 2– [(OPO)f V (O)s (OH)t (H2 O)c ] IV 2– [(OPO)f V (O)t (OH)c (H2 O)s ] IV 2– [(OPO)f V (O)t (OH)s (H2 O)c ] IV 2– [(OPO)fH V (O)c (OH)st 2 ] IV 2– [(OPO)fH V (O)s (OH)st 2 ] IV 2– [(OPO)fH V (O)t (OH)st 2 ] Table 4.3. Comprehensive compilation of all (OPO)VIV species studied, along with their gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous V free energies, in kcal/mol, relative to the ground states at pH 0 and 7 ((OPO)fH V Oc (OH)s (H2 O)t and V 2– [(OPO)f V Oc (OH)st 2 ] , respectively). 138 IV IV – monohydrogenated (OPO)fH V Oc (H2 O)st 2 forms the anionic intermediates [(OPO)V O(H2 O)2 ] , IV – the lowest isomer of which ([(OPO)f V (O)s (H2 O)ct 2 ] ) is 17.4 kcal/mol above the starting complex. The deprotonation energy gain of 4.0 kcal/mol leads to a calculated pKa value of 2.9 for the neutral IV monohydrogenated complex (OPO)fH V Oc (H2 O)st 2 . However, the reduction potential of the neutral V V f c st (OPO)f V Os (H2 O)ct 2 is calculated to be 1.07 V versus SHE, whereas that of (OPO) V O (H2 O)2 is IV calculated to be 0.77 V versus SHE. Hence, it is difficult to reoxidize the deprotonated [(OPO)f V Oc – V (H2 O)st 2 ] back to V , but not impossible; since the standard reduction potential of O2 is 1.23 V vs. SHE. Scheme 4.2. Potential catalytic cycle involving H-atom abstraction only: The one-electron oxidation and catalyst regeneration that forms the core of this catalytic cycle. All energies are in kcal/mol. Note that the overall reaction in this cycle is C3 H8 + 21 O2 −−→ C3 H6 + H2 O, with ∆G = −32.3 kcal/mol. Hence, whereas V V f s ct the (OPO)f V Oc (H2 O)st 2 and (OPO) V O (H2 O)2 complexes are initially 1.6 and 4.8 kcal/mol above the starting state, after one complete cycle they are −30.7 and −27.4 kcal/mol below the starting state, respectively. Since this would only be a one-electron oxidative process, two complete cycles are necessary for propane oxidation: first to abstract a hydrogen atom from the starting alkane, and then to abstract a second hydrogen atom from the resulting isopropyl radical. As hydrogen abstraction from an alkyl radical to a closed-shell species is expected to be facile, the rate-limiting step in this case will be the V initial propane activation step. From the (OPO)f V Oc (H2 O)st 2 precursor (1.6 kcal/mol above the ground state), initial H-atom abstraction from propane is uphill by 11.8 kcal/mol. As the weakest C−H bond in propane (i.e. the iPr−H bond) has a dissociation energy of 103 kcal, this implies a very strong hydrogen abstracting power of 91.2 kcal/mol. This number is very close to the hydrogenabstraction power of the P−O moiety in VV OPO4 , the oxidized surface of the VPO catalyst [3], hence confirming the validity of our ROA mechanism as modified in Figure 4.4b. The overall kinetic 139 barrier for H-atom abstraction is 30.3 kcal/mol, a value that is low enough to be accessible at elevated reaction temperatures. As indicated by Scheme 4.2 and its associated discussion, deprotonation of the monohydrogenated IV V (OPO)fH V Oc (H2 O)st 2 precedes oxidation back to a V species. In fact, these steps are both slightly uphill, by 4.0 and 1.6 kcal/mol, respectively, but they are low enough to remain feasible. The second H-atom abstraction, from the isopropyl radical, is quite facile as expected, creating propylene as the IV oxidation product. The resulting (OPO)fH V Oc (H2 O)st 2 then goes through one more deprotonation, oxidation, and isomerization cycle to regenerate the starting catalyst. Overall, this reaction is quite feasible thermodynamically, and the reaction barriers are not expected to be large, either. Hence we can expect this to be a complete potential catalytic cycle for the dehydrogenation of propane to propylene. 4.5.6 Reduced VIII states Since reactive isopropyl radicals are generated upon H-atom abstraction, it is necessary to also consider the possibility that they may be trapped by the VIV intermediate instead of being converted to propylene by another equivalent of a VV species. Once again taking inspiration from Cheng and Goddard [3], in which the produced isopropyl radical is adsorbed onto a V−O moiety, we analogously IV considered the possibility that the isopropyl radical and monohydrogenated (OPO)V (O)(H2 O)2 III species react with each other to form (OPO)V (OiPr) species. The isopropoxy ligand may then be protonated and exchanged with an aqua ligand, resulting in isopropanol being produced. Table 4.4 is a comprehensive list of (OPO)V III species with only oxo/hydroxo/aqua ligands, and neutral or −1 charge. As in the case of the VV and VIV species, the (OPO)V III species are most stable with the (OPO) ligand in the fac configuration. Furthermore, the triplet states are universally more stable than their singlet counterparts, supporting the +III assignment for the vanadium atom’s III oxidation state. Hence, Table 4.5, which lists the (OPO)V (OiPr) species, only contains the triplet six coordinate species with the (OPO) ligand in the fac configuration. Species Egas Gaq Grel (pH = 0) Grel (pH = 7) 1 (OPO)V (H2 O)sa 2 III −1712.6358 −1712.3473 25.0 33.9 3 (OPO)V (H2 O)sa 2 III −1712.6811 −1712.3911 −2.5 6.4 1 (OPO)H VIII (OH)s (H2 O)a geometry unstable 3 (OPO)H VIII (OH)s (H2 O)a −1712.6895 −7.9 1.0 −1712.3998 140 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) 1 (OPO)H VIII (OH)a (H2 O)s −1712.6609 −1712.3705 10.5 19.4 3 (OPO)H VIII (OH)a (H2 O)s −1712.6784 −1712.3916 −2.7 6.2 1 [(OPO)V (OH)s (H2 O)a ] 3 [(OPO)V (OH)s (H2 O)a ] 7.7 7.1 1 [(OPO)V (OH)a (H2 O)s ] 3 [(OPO)V (OH)a (H2 O)s ] 10.9 10.3 1 [(OPO)H VIII (OH)sa 2 ] – geometry unstable 3 [(OPO)H VIII (OH)sa 2 ] – geometry unstable 1 (OPO)m V (H2 O)sta 3 III −1942.0707 −1941.7179 27.5 36.4 3 (OPO)m V (H2 O)sta 3 III −1942.1201 −1941.7641 −1.5 7.4 1 s ta (OPO)m H V (OH) (H2 O)2 III −1942.0625 −1941.7111 31.8 40.7 3 s ta (OPO)m H V (OH) (H2 O)2 III −1942.1196 −1941.7628 −0.7 8.2 1 t sa (OPO)m H V (OH) (H2 O)2 III −1942.0710 −1941.7216 25.2 34.1 3 t sa (OPO)m H V (OH) (H2 O)2 III −1942.1162 −1941.7627 −0.6 8.3 1 a st (OPO)m H V (OH) (H2 O)2 III −1942.0597 −1941.6995 39.1 48.0 3 a st (OPO)m H V (OH) (H2 O)2 III −1942.1097 −1941.7567 3.2 12.0 1 [(OPO)m V (OH)s (H2 O)ta 2 ] III – −1941.5661 −1941.2673 40.0 39.3 3 [(OPO)m V (OH)s (H2 O)ta 2 ] III – −1941.6145 −1941.3211 6.2 5.6 1 [(OPO)m V (OH)t (H2 O)sa 2 ] III – −1941.5611 −1941.2660 40.8 40.2 3 [(OPO)m V (OH)t (H2 O)sa 2 ] III – −1941.6143 −1941.3217 5.8 5.2 1 [(OPO)m V (OH)a (H2 O)st 2 ] III – −1941.5539 −1941.2535 48.6 48.0 3 [(OPO)m V (OH)a (H2 O)st 2 ] III – −1941.6047 −1941.3114 12.3 11.7 1 s ta [(OPO)m H V (H2 O) (OH)2 ] III – −1941.5762 −1941.2783 33.1 32.5 III – geometry unstable III – −1712.1787 III – geometry unstable III – −1712.1727 −1711.9442 −1711.9391 141 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) 3 s ta [(OPO)m H V (H2 O) (OH)2 ] III – −1941.5886 −1941.3002 19.3 18.7 1 t sa [(OPO)m H V (H2 O) (OH)2 ] III – −1941.5562 −1941.2561 47.0 46.4 3 t sa [(OPO)m H V (H2 O) (OH)2 ] III – −1941.6073 −1941.3093 13.6 13.0 1 a st [(OPO)m H V (H2 O) (OH)2 ] III – −1941.5988 −1941.3017 18.4 17.7 3 a st [(OPO)m H V (H2 O) (OH)2 ] III – −1941.6188 −1941.3235 4.7 4.1 1 (OPO)f V (H2 O)cst 3 III −1942.0823 −1941.7202 26.0 34.9 3 (OPO)f V (H2 O)cst 3 III −1942.1432 −1941.7900 −17.7 −8.8 −1942.0873 −1941.7262 22.3 31.2 −1942.1429 −1941.7811 −12.2 −3.3 −1942.0885 −1941.7307 19.5 28.4 −1942.1386 −1941.7818 −12.6 −3.7 −1942.0732 −1941.7139 30.0 38.9 −1942.1288 −1941.7765 −9.3 −0.4 III 1 (OPO)fH V (OH)c (H2 O)st 2 3 (OPO)fH V (OH)c (H2 O)st 2 III III 1 (OPO)fH V (OH)s (H2 O)ct 2 3 (OPO)fH V (OH)s (H2 O)ct 2 1 (OPO)fH V (OH)t (H2 O)cs 2 III III III 3 (OPO)fH V (OH)t (H2 O)cs 2 1 [(OPO)f V (OH)c (H2 O)st 2 ] 3 [(OPO)f V (OH)c (H2 O)st 2 ] 1 [(OPO)f V (OH)s (H2 O)ct 2 ] 3 [(OPO)f V (OH)s (H2 O)ct 2 ] III – −1941.5871 −1941.2738 35.9 35.3 III – −1941.6398 −1941.3377 −4.2 −4.8 III – −1941.5861 −1941.2299 63.4 62.8 III – −1941.6343 −1941.3371 −3.8 −4.5 III – −1941.5637 −1941.2725 36.7 36.1 III – −1941.6283 −1941.3282 1.8 1.2 III – −1941.6034 −1941.3092 13.7 13.1 III – −1941.6221 −1941.3289 1.3 0.7 III – −1941.5798 −1941.2804 31.8 31.1 III – −1941.6323 −1941.3331 −1.3 −2.0 1 [(OPO)f V (OH)t (H2 O)cs 2 ] 3 [(OPO)f V (OH)t (H2 O)cs 2 ] 1 [(OPO)fH V (H2 O)c (OH)st 2 ] 3 [(OPO)fH V (H2 O)c (OH)st 2 ] 1 [(OPO)fH V (H2 O)s (OH)ct 2 ] 3 [(OPO)fH V (H2 O)s (OH)ct 2 ] 142 Species 1 3 Egas Gaq Grel (pH = 0) Grel (pH = 7) III – −1941.6015 −1941.3001 19.4 18.8 III – −1941.6254 −1941.3234 4.8 4.1 [(OPO)fH V (H2 O)t (OH)cs 2 ] [(OPO)fH V (H2 O)t (OH)cs 2 ] Table 4.4. Comprehensive compilation of all (OPO)VIII species containing only oxo/hydroxo/aqua ligands studied, along with their gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous free energies, in kcal/mol, relative to the ground states at pH 0 V V 2– and 7 ((OPO)fH V Oc (OH)s (H2 O)t and [(OPO)f V Oc (OH)st 2 ] , respectively). For some entries, geometry optimization led to the formation of a lower-energy isomer. The energies for these entries are not reported. Species Egas Gaq III −1907.1454 III III 3 (OPO)f V (HOiPr)c (H2 O)st 2 3 (OPO)f V (HOiPr)s (H2 O)ct 2 3 (OPO)f V (HOiPr)t (H2 O)cs 2 III 3 (OPO)fH V (HOiPr)c (OH)s (H2 O)t 3 (OPO)fH V (HOiPr)c (OH)t (H2 O)s III III 3 (OPO)fH V (HOiPr)s (OH)c (H2 O)t 3 (OPO)fH V (HOiPr)s (OH)t (H2 O)c 3 (OPO)fH V (HOiPr)t (OH)c (H2 O)s III III III 3 (OPO)fH V (HOiPr)t (OH)s (H2 O)c 3 (OPO)fH V (OiPr)c (H2 O)st 2 3 (OPO)fH V (OiPr)s (H2 O)ct 2 III III III 3 (OPO)fH V (OiPr)t (H2 O)cs 2 3 [(OPO)f V (HOiPr)c (OH)s (H2 O)t ] Grel (pH = 0) Grel (pH = 7) −1906.7524 −12.8 −3.9 −1907.1518 −1906.7610 −18.2 −9.3 −1907.1489 −1906.7546 −14.2 −5.3 −1907.1417 −1906.7463 −9.0 −0.1 −1907.1398 −1906.7456 −8.6 0.3 −1907.1520 −1906.7574 −16.0 −7.1 −1907.1339 −1906.7422 −6.5 2.4 −1907.1472 −1906.7494 −10.9 −2.0 −1907.1439 −1906.7503 −11.5 −2.6 −1907.1448 −1906.7528 −13.1 −4.2 −1907.1544 −1906.7567 −15.5 −6.6 −1907.1454 −1906.7458 −8.7 0.2 III – −1906.6425 −1906.3086 −4.6 −5.2 III – −1906.6463 −1906.3113 −6.3 −6.9 III – −1906.6475 −1906.3125 −7.1 −7.7 III – −1906.6369 −1906.3039 −1.7 −2.3 III – −1906.6436 −1906.3069 −3.6 −4.2 3 [(OPO)f V (HOiPr)c (OH)t (H2 O)s ] 3 [(OPO)f V (HOiPr)s (OH)c (H2 O)t ] 3 [(OPO)f V (HOiPr)s (OH)t (H2 O)c ] 3 [(OPO)f V (HOiPr)t (OH)c (H2 O)s ] 143 Species 3 III – [(OPO)f V (HOiPr)t (OH)s (H2 O)c ] Gaq Grel (pH = 0) Grel (pH = 7) −1906.6461 −1906.3043 −1.9 −2.6 III – −1906.6476 −1906.3131 −7.5 −8.1 III – −1906.6425 −1906.3043 −1.9 −2.5 III – −1906.6420 −1906.2957 3.4 2.8 3 [(OPO)f V (OiPr)c (H2 O)st 2 ] 3 [(OPO)f V (OiPr)s (H2 O)ct 2 ] 3 [(OPO)f V (OiPr)t (H2 O)cs 2 ] III – geometry unstable III – −1906.6284 −1906.2897 7.2 6.6 III – −1906.6261 −1906.2951 3.8 3.2 3 [(OPO)fH V (HOiPr)c (OH)st 2 ] 3 [(OPO)fH V (HOiPr)s (OH)ct 2 ] 3 [(OPO)fH V (HOiPr)t (OH)cs 2 ] III – −1906.6346 −1906.2971 2.6 1.9 III – −1906.6346 −1906.3005 0.5 −0.2 III – −1906.6482 −1906.3055 −2.7 −3.3 III – −1906.6398 −1906.2991 1.4 0.7 III – geometry unstable III – geometry unstable 3 [(OPO)fH V (OiPr)c (OH)s (H2 O)t ] 3 [(OPO)fH V (OiPr)s (OH)c (H2 O)t ] 3 [(OPO)fH V (OiPr)s (OH)c (H2 O)t ] 3 [(OPO)fH V (OiPr)s (OH)t (H2 O)c ] 3 [(OPO)fH V (OiPr)t (OH)c (H2 O)s ] 3 Egas [(OPO)fH V (OiPr)t (OH)s (H2 O)c ] Table 4.5. Compilation of all (OPO)VIII (OiPr) species studied, with only six-coordinate fac-(OPO) triplet complexes considered. The second and third columns show the gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous free energies, in kcal/mol, relative V V 2– to the ground states at pH 0 and 7 ((OPO)fH V Oc (OH)s (H2 O)t and [(OPO)f V Oc (OH)st 2 ] , respectively). For some entries, geometry optimization led to the formation of a lower-energy isomer. The energies for these entries are not reported. III III At pH 0, the lowest energy (OPO)V (OiPr) species is 3 (OPO)f V (HOiPr)s (H2 O)ct 2 at −18.2 V III kcal/mol relative to the starting (OPO)fH V Oc (OH)s (H2 O)t , whereas the lowest energy (OPO)V III f species without any isopropoxy groups is 3 (OPO)f V (H2 O)cst 3 at −17.7 kcal/mol relative to (OPO)H III VV Oc (OH)s (H2 O)t . Furthermore, the lowest energy five-coordinate (OPO)V species without any V isopropoxy groups is 3 (OPO)H V (OH)s (H2 O)a at −7.9 kcal/mol relative to (OPO)fH V Oc (OH)s III III III 3 f cst (H2 O)t , indicating that it is possible for 3 (OPO)f V (HOiPr)s (H2 O)ct 2 and (OPO) V (H2 O)3 to interconvert via a labile isopropanol ligand. These three species are shown in detail in Figure 4.21. Note that at pH 7 the same three species are still the most stable, but with energies of −9.3, −8.8, V 2– and 1.0 kcal/mol with respect to [(OPO)f V Oc (OH)st 2 ] , respectively. 4.5.7 Catalytic cycle involving H-atom abstraction followed by isopropyl addition By considering the lowest energy configurations of the various VV , VIV , and VIII intermediates and transition states, we can now construct a catalytic cycle that features both H-atom abstraction 144 III III 3 f cst Figure 4.21. The geometry-optimized structures of 3 (OPO)f V (HOiPr)s (H2 O)ct 2 , (OPO) V (H2 O)3 , III 3 III s a and (OPO)H V (OH) (H2 O) . These are the most stable (OPO)V (OiPr) species, and six and five coordinate (OPO)VIII species, respectively, at both pH 0 and 7. from propane and isopropyl radical trapping before reoxidation. This cycle is shown in Scheme 4.3. V V The first part, involving isomerization of the starting (OPO)fH V Oc (OH)s (H2 O)t to (OPO)f V Oc IV f c st (H2 O)st 2 , followed by C−H activation to (OPO)H V O (H2 O)2 , is identical to that of Scheme 4.2. However, instead of being immediately reoxidized as in Scheme 4.2, the VIV species traps the generated isopropyl radical to form a VIII (OiPr) species. One example of such a pathway is shown in IV Scheme 4.3, in which the (OPO)fH V Oc (H2 O)st 2 species first isomerizes to the slightly higher (at IV 15.4 kcal/mol relative to the starting VV complex) (OPO)fH V Os (H2 O)ct 2 , which then combines III with the isopropyl radical to form 3 (OPO)fH V (OiPr)s (H2 O)ct 2 at −15.5 kcal/mol, which readily III III isomerizes to 3 (OPO)f V (HOiPr)s (H2 O)ct 2 , the lowest energy V (OiPr) species at −18.2 kcal/mol V relative to the starting (OPO)fH V Oc (OH)s (H2 O)t . Loss of the HOiPr product to form the five coIII ordinate species 3 (OPO)V (H2 O)sa 2 (at −1.5 kcal/mol) is uphill but not insurmountably so, and III recoordination of an aqua ligand to the sixth coordination site to form 3 (OPO)f V (H2 O)cst 3 recovers (at −17.7 kcal/mol) most of this energy. As the overall reaction C3 H8 + 1 /2 O2 −−→ HOiPr has III V f c st ∆G = −31.2 kcal/mol, reoxidation of 3 (OPO)V (H2 O)sa 2 to (OPO) V O (H2 O)2 (assuming that O2 is the terminal oxidation) is downhill by 11.9 kcal/mol. 4.5.8 Comparison of the catalytic cycles The H-atom abstraction-only pathway results in the dehydrogenation of propane to form propylene. However, as this is significantly uphill (C3 H8 −−→ C3 H6 + H2 , ∆G = 20.3 kcal/mol), it is unlikely unless some oxidant is coupled to the reaction. Coupling dehydrogenation with dioxygen greatly 145 Scheme 4.3. A potential catalytic cycle that assumes that the isopropyl radical adds to the vanadium oxo, reducing it to VIII . This assumption derives from the behavior of the heterogeneous VPO catalyst. All energies are in kcal/mol. Note that the overall reaction in this cycle is C3 H8 + 1 /2 O2 −−→ HOiPr, with V ∆G = −31.2 kcal/mol. Hence, whereas the (OPO)f V Oc (H2 O)st 2 complex is initially 1.6 kcal/mol above the starting state, after one complete cycle it is −29.6 kcal/mol below the starting state. increases its favorability, for a ∆G of −32.3 kcal/mol. We have not found any large barriers in the Hatom abstraction-only pathway that might preclude the production of propylene and water as in the overall equation C3 H8 + 21 O2 −−→ C3 H6 + H2 O; however, we note that our results do not include the IV IV – f s ct details of how the one-electron oxidation of [(OPO)f V (O)c (H2 O)st 2 ] or [(OPO) V (O) (H2 O)2 ] – takes place and how the released electron is transferred to dioxygen to ultimately form water. Such a mechanism may involve an inner sphere coordination of dioxygen to form a VV superoxo intermediate and may involve further C−H activation by this superoxide. Given the reduction potential of IV – IV – f s ct dioxygen versus that of [(OPO)f V (O)c (H2 O)st 2 ] or [(OPO) V (O) (H2 O)2 ] (1.23 V versus 0.77 V or 1.07 V), we believe that the H-atom abstraction-only pathway will be viable and productive in the absence of lower-energy alternative pathways. Comparing the H-atom abstraction-only versus the H-atom abstraction and isopropyl addition IV pathways, we note that the branching point is the monohydrogenated complexes (OPO)fH V (O)(H2 O)2 , which may either undergo reoxidation as in the abstraction-only mechanism (Scheme 4.2), or add the isopropyl radical (Scheme 4.3). The former overall is thermodynamically downhill by 0.8 kcal/mol in the lowest isomers; whereas the latter is thermodynamically downhill by 30.9 kcal/mol in one simple elementary step with an intuitively minimal barrier. Hence, the catalytic cycle involving isopropyl addition appears to be more likely to occur than the cycle that only features H-atom 146 abstraction. As in the H-atom abstraction-only case, we have not considered the precise nature of the reoxidation step. However, we anticipate that it may also involve inner sphere coordination of dioxygen, in which case a VV η 2 -peroxo intermediate is formed. This would be analogous to the heterogeneous VPO system, and we would also need to anticipate C−H activation by these peroxide species as well. 4.6 Conclusions We have identified a new mechanistic motif, which we call the reductive-coupled oxo activation mechanism, in which a normally unreactive main-group oxo becomes a potent hydrogen atom abstractor when coupled to a transition metal (in our case VV or MoVI ). The new activity of this oxo, which we quantify using the O−H dissociation parameter DH , is modified from that of the transition metal oxo itself and may be significantly higher for certain elements, specifically phosphorous in the +V oxidation state and selenium and tellurium in the +IV and +VI oxidation states. Our results are consistent with existing experimental and theoretical mechanistic work on the n-butane to maleic anhydride VOPO catalyst. We believe that knowledge of this new mechanistic motif will allow for the de novo rational design of future oxidation catalysts. As a showcase of the power of the ROA mechanistic motif, we have designed a potential inorganic V complex (OPO)fH V Oc (OH)s (H2 O)t that may serve as a homogeneous oxidation catalyst for alkanes. We have identified catalytic cycles for the oxidation of propane to propylene or isopropanol. Our likely catalytic cycle involves H-atom abstraction from propane, followed by recombination of the produced isopropyl radical and VIV species to form a VIII isopropoxy species. Ligand exchange results in the formation of free isopropanol and reoxidation regenerates the starting VV species. Our theoretical results are encouraging for us to continue our computational work in investigating reoxidation pathways, and to commence experimental work. 147 4.7 References [1] Centi, G.; Trifiro, F.; Ebner, J. R.; Franchetti, V. M. Chem. Rev. 1988, 88, 55–80. [2] Hodnett, B. K. Heterogeneous Catalytic Oxidation. Wiley: New York, 2000. [3] Cheng, M.-J. and Goddard, W. A. J. Am. Chem. Soc. 2013, 135, 4600–4603. [4] Schuurman, Y.; Gleaves, J. T. Ind. Eng. Chem. Res. 1994, 33, 2935–2941. [5] Hutchings, G. J.; Desmartinchomel, A.; Olier, R.; Volta, J. C. Nature 1994, 368, 41–45. [6] Cheng, M.-J.; Goddard, W. A.; Fu, R. Methods for providing bond activation catalysts and related catalysts, systems, and methods. U.S. Patent application 14/192,856, filed February 27, 2014. Patent pending. [7] Marino, N.; Hanson, S. K.; Müller, P.; Doyle, R. P. Inorg. Chem. 2012, 51, 10077–10079. [8] Herron, N.; Thorn, D. L.; Harlow, R. L.; Coulston, G. W. J. Am. Chem. Soc. 1997, 119, 7149–7150. [9] Solis-Ibarra, D.; Silvia, J. S.; Jancik, V.; Cummins, C. C. Inorg. Chem. 2011, 50, 9980–9984. [10] Pan, B. F.; Bezpalko, M. W.; Foxman, B. M.; Thomas, C. M. Organometallics 2011, 30, 5560– 5563. [11] (a) Derrah, E. J.; Martin, C.; Mallet-Ladeira, S.; Miqueu, K.; Bouhadir, G.; Bourissou, D. Organometallics 2013, 32, 1121–1128. (b) Derrah, E. J.; Ladeira, S.; Bouhadir, G.; Miqueu, K.; Bourissou, D. Chem. Commun. 2011, 47, 8611–8613. [12] (a) Pan, B. F.; Bezpalko, M. W.; Foxman, B. M.; Thomas, C. M. Dalton Trans. 2012, 41, 9083–9090. 148 (b) Bauer, R. C.; Gloaguen, Y.; Lutz, M.; Reek, J. N. H.; Bruin, B. de; Vlugt, J. I van der. Dalton Trans. 2011, 40, 8822–8829. (c) Mazzeo, M. Lamberti, M.; Massa, A.; Scettri, A.; Pellecchia, C.; Peters, J. C. Organometallics 2008, 27, 5741–5743. (d) Mazzeo, M.; Strianese, M.; Kuhl, O.; Peters, J. C. Dalton Trans. 2011, 40, 9026–9033. [13] Gloaguen, Y.; Jacobs, W.; Bruin, B. de; Lutz, M.; Vlugt, J. I. van der; Inorg. Chem. 2013, 52, 1682–1684. [14] Jaguar, version 7.9 ; Schrödinger, LLC: 2012, New York, NY. [15] (a) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.; Sitkoff, D.; Nicholls, A.; Ringnalda, M.; Goddard, W. A.; Honig, B. J. Am. Chem. Soc. 1994, 116, 11875–11882. (b) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.; Ringnalda, M. N.; Sitkoff, D.; Honig, B. J. Phys. Chem. 1996, 100, 11775–11788. [16] Sears, F. W.; Zemansky, M. W.; Young, H. D. University Physics, 6th Ed.; Addison-Wesley, 1982. p [17] The probe radius r is automatically calculated by Jaguar using the formula r = 0.5833 3 MW/ρ where MW is the molecular weight and ρ is the density of the solvent. [18] (a) Becke, A. D. Phys. Rev. A. 1998, 38, 3098–3100. (b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652. (c) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B. 1988, 37, 785–789. [19] Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310. [20] (a) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270–283; (b) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 284–298. [21] (a) Hehre, W. J.; Ditchfield, R; Pople, J. A. J. Chem. Phys. 1972, 56, 2257–2261. (b) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; Defrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654–3665. [22] Tissandier, M. D.; Cowen, K. A.; Feng, W. Y.; Gundlach, E.; Cohen, M. H.; Earhart, A. D.; Coe, J. V.; Tuttle, T. R. J. Phys. Chem. A. 1998, 102, 7787–7794. Calculated at standard conditions. [23] Wertz, D. H. J. Am. Chem. Soc. 1980, 102, 5316–5322. 149 [24] Bridgeman, O.C.; Aldrich, E.W. J. Heat Transfer 1964, 86, 279–286. [25] Pangborn, A. B.; Giardello, M. A.; Grubbs, R. H.; Rosen, R. K.; Timmers, F. J. Organometallics 1996, 15, 1518–1520. [26] Bartlett, P. D. Rec. Chem. Prog. 1950, 11, 47–51. [27] (a) Wife, R. L.; Oort, A. B. van; Doorn, J. A. van; Leeuwen, P. W. N. M. van Synthesis 1983, 71. (b) Baccolini, G.; Bazzocchi, M.; Carla, B. Eur. J. Org. Chem. 2001, 2229–2233. [28] Some recent publications that show similar chemistry of using VOCl3 to install hard multidentate anionic ligands onto vanadium include: (a) Kurmaev, D. A.; Kolosov, N. A.; Gagieva, S. C.; Borissova, A. O.; Tuskaev, V. A.; Bravaya, N. M.; Bulychev, B. M. Inorg. Chim. Acta 2013, 396, 136–143. (b) Villanneau, R.; ben Djamâa, A.; Chamoreau, L.-M.; Gontard, G.; Proust, A. Eur. J. Inorg. Chem. 2013, 1815–1820. (c) Tong, L. H.; Wong, Y.-L.; Lee, H. K.; Dilworth, J. R. Inorg. Chim. Acta 2012, 383, 91–97. (d) North, M.; Stewart, E. L.; Young, C. Tetrahedron: Asymmetry 2012, 23, 1218–1225. [29] Wolff, F.; Choukroun, R.; Lorber, C.; Donnadieu, B. Eur. J. Inorg. Chem. 2003, 628–632. [30] Romanowski, G. J. Mol. Cat. A: Chemical 2013, 368, 137–144. [31] Busacca, C. A.; Lorenz, J. C.; Grinberg, N.; Haddad, N.; Hrapchak, M.; Latli, B.; Lee, H.; Sabila, P.; Saha, A.; Sarvestani, M.; Shen, S.; Varsolona, R.; Wei, X.; Senanayake, C. H. Org. Lett. 2005, 7, 4277–4280. [32] Bickelhaupt, F.; van der Does, T. Biphosphine production. U.S. Patent US4874897 A, 1989. [33] Doorn, J. A. van; Frijns, J. H. G.; Meijboom, N. Recl. Trav. Chim. Pays-Bas 1991, 110, 441– 449. 150 Chapter 5 The mechanism of Ni(iPr−pybox) complexes as asymmetric Negishi C−C coupling catalysts 5.1 Abstract The 2,6-bis(4-isopropyl-2-oxazolin-2-yl)pyridine (iPr-pybox) ligand, when combined with a nickel salt in N ,N -dimethylacetamide (DMA), generates an in situ Ni(iPr−pybox) complex that catalyzes enantioselectively the cross-coupling of secondary sp3 -C substrates. This system is very notable for its property of stereoconvergence: both enantiomers of a racemic substrate may be converted into a single enantiomer of a product. This chapter will computationally explore the mechanism responsible for both the remarkable activity and selectivity of the Ni(iPr−pybox) system; such a mechanism is believed to deviate substantially from the canonical “textbook” mechanism of Negishi coupling. Furthermore, a generalized calculation scheme is presented that allows for the rapid enantioselectivity prediction of many related pybox ligands as well, in order to help predict the next generation of asymmetric Negishi coupling catalysts. 151 5.2 Introduction The controlled formation of C−C bonds is one of the central challenges of synthetic organic chemistry. Generally, the cross-coupling of two distinct organic moieties R and R0 is done by selecting precursors with opposite polarity, for example, an electrophilic R−X (where X is a leaving group) along with a nucleophilic R0 −[M0 ] (where M0 is a less electronegative element, often a main-group metal or metalloid). Late transition metals M are typically added as well in order to catalyze activity and encourage the formation of R−R0 as opposed to undesired side products. Ligands may also be added in order to further enhance the activity of the transition metal catalyst by the generation of in situ complexes. Scheme 5.1 shows the overall cross-coupling reaction, as well as a general scheme for the mechanism. This mechanism [1] begins with the oxidative addition of R−X by a low-valent M (hence the necessity of using a late transition metal). The X ligand is then exchanged for an R0 in the transmetalation step, and finally reductive elimination regenerates the starting M and releases the coupled product R−R0 . Scheme 5.1. Left: the overall C−C bond-forming cross-coupling reaction. Right: the general mechanism for the cross-coupling reaction. Table 5.1 shows a few examples of cross-coupling reactions. As already mentioned, the electrophile is typically R−X where X is a leaving group such as a halide or sulfonate. Substrates in which the R group is an aryl or alkenyl, or a resonance-stabilized sp3 radical such as an allyl or propargyl group, tend to be more active. The metal catalyst M is typically Cu, Ni, or Pd, with Pd being the most versatile element but also the rarest and most expensive. The nucleophile is R0 −[M0 ] where M0 may well be any element that forms a weaker bond to R0 than M, and/or a stronger bond to X. 152 Coupling Kumada Heck Sonogashira Negishi Stille Suzuki Hiyama Electrophile R−X R−X R−X R−X R−X R−X R−X Nucleophile R0 −MgBr alkene R0 −C− −CH R0 −ZnX R0 −SnR03 R0 −B(OR00 )2 R0 −SiR03 Catalyst Pd, Ni Pd Pd, Cu Pd, Ni Pd Pd Pd Reference [2] [3] [4] [5] [6] [7] [8] Table 5.1. Examples of various C−C cross-coupling reactions. This list is not exhaustive. 5.2.1 The Ni(iPr−pybox) system Another central challenge of synthetic organic chemistry is the enantioselective synthesis of chiral compounds. For cross-coupling reactions in which C−C bonds are formed, the carbon atoms being bonded are often chiral centers, necessitating techniques for asymmetric synthesis. For cross-coupling reactions, this may be done by adding a chiral ligand to complex the catalyst metal, thus preferentially coupling only one enantiomer of a racemic starting material. Alternatively, one may also simply begin with an enantiopure substrate and chose a coupling reaction that preserves chirality. The Ni(iPr−pybox) system, developed by Gregory Fu and coworkers, is a Negishi coupling catalyst that utilizes nickel and the 2,6-bis(4-isopropyl-2-oxazolin-2-yl)pyridine ligand, abbreviated as (iPr-pybox), and couples organic halides to organozinc complexes [9]. The (iPr-pybox) ligand contains two chiral centers and can exist in three diastereomers: the C2 enantiomers (((R, R)-iPr-pybox) and ((S, S)-iPr-pybox)), and the Cs meso diastereomer ((R, S)-iPr-pybox). If a chiral C2 ligand is used, asymmetric catalysis results. The complex and a summary of its catalytic scope is shown in Scheme 5.2 and Table 5.2 [9, 10]. Scheme 5.2. The general reaction of the Ni(iPr−pybox) Negishi coupling catalyst with benzylic secondary alkyl halides. α-Bromo amides and propargylic halides are also viable substrates that have been studied. This Ni(iPr−pybox) system is advantageous for several reasons [10]: 153 R0 −ZnBr Yield (%) ee (%) C6 H13 −ZnBr 89 96 N− −C−(CH2 )3 −ZnBr 82 64 91 91 69 94 47 72 91 98 7 39 96 8 41 99 Entry 1 2 3 4 5 6 R−X Cl−(CH2 )6 −ZnBr nBu−ZnBr 9 Ph−(CH2 )3 −ZnBr 76 98 10 BnO−(CH2 )3 −ZnBr 63 75 Table 5.2. Some examples of benzylic secondary alkyl halide substrates coupled by the Ni(iPr−pybox) catalyst. All reactions were performed under conditions as in Scheme 5.2, except for entries 7 and 8 which were performed at room temperature. The data in this table was taken from Arp and Fu [10]. 154 1. It is a Negishi coupling catalyst, utilizing organozinc reagents as the nucleophile. Organozinc compounds are relatively tractable and accessible; 2. The ligand is known and is not difficult to synthesize from commercially-available starting materials; 3. The system is not particularly oxygen or water sensitive, and the reaction may be run under air; 4. Coupling reactions with alkyl substrates are typically more difficult than with their aryl counterparts, and unwanted side reactions such as β-hydride elimination may be present. Secondary alkyls are particularly problematic due to their added steric bulk. The pybox system has been demonstrated to work with various secondary alkyl electrophiles such as benzylic halides, αbromo amides, and propargylic halides [11]. However, one of the Ni(iPr−pybox) system’s most interesting properties may be its stereoconvergence. A racemic starting material may be used to generate a nearly enantiopure product with high yield (See Scheme 5.2). This implies that the mechanism requires both the destruction of the starting material’s stereochemistry as well as the formation of the product’s stereochemistry, and indicates that the mechanism may be more complex than the existing “textbook” mechanism as shown in Scheme 5.1. 5.2.2 Existing computational work The mechanism for the Ni(iPr−pybox) system’s remarkable reactivity was computationally investigated by Lin et al., who modeled the coupling of 1-bromoindane (the electrophile in Table 5.2, entry 3) with methylzinc bromide as catalyzed by NiI ((S, S)−iPr−pybox)Br to form (R)- and (S)-1methylindane [12]. Their results, which are summarized in 5.1, claim that NiI ((S, S)−iPr−pybox)Br is first methylated by methylzinc bromide to form NiI ((S, S)−iPr−pybox)Me. Oxidation by 1bromoindane results in NiII ((S, S)−iPr−pybox)(Me)Br and the indanyl radical, which then recombine to form NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind) (ind = 1-indanyl), which undergoes reductive elimination to form the 1-methylindane product and to regenerate the NiI ((S, S)−iPr−pybox)Br starting catalyst. The mechanism put forth by Lin et al. supports the hypothesis of Fu et al. that a NiI -NiIII cycle is involved, as opposed to the Ni0 -NiII cycle of a classical coupling reaction. In addition, their conclusions show a decisive preference for the formation of the (S)-1-methylindane enantiomer, in line with experimental evidence. However, their methodology and analysis have several issues: 1. The density functional theory used, B3LYP, neglects van der Waals interactions, which typically increase in importance for large molecules; 155 Figure 5.1. The computationally studied mechanism by Lin et al. of Ni((S, S)−iPr−pybox) catalysis of methyl-indane coupling. The steps are: methylation and oxidation of NiI ((S, S)−iPr−pybox)Br to NiII ((S, S)−iPr−pybox)(Me)Br, recombination with indanyl radical to form NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind), and reductive elimination to form 1-methylindane product. All energies in kcal/mol. Note that for the first two steps (i.e. methylation and oxidation to NiII ), the energy differences are only estimates, as the actual calculations were only performed on the parent Ni(pybox) complexes. 2. The basis set used, 6-31G* (a Pople-type double-ζ Gaussian basis set with polarization functions on non-hydrogen atoms) is rather small and limited, and may lead to inaccurate energies [13]; 3. Whereas the authors claim to investigate the enantioselectivity of the Ni(iPr−pybox) system, in many reactions the simplified pybox ligand (without the isopropyl groups) was used in lieu of (iPr)-pybox to simplify calculations; 4. The authors model the methylzinc bromide nucleophile as a simple two-coordinate complex; however, the true nature of the zincate in solution is likely to be much more complex [14]; 5. Although the reaction conditions and all experimental investigations have used DMA as the solvent, the authors modeled the reaction in DMSO instead; 6. Only one radical attack pathway (i.e. that of the indanyl radical approaching axially to the Ni) was considered. Therefore, it is not surprising that enantioselectivity of the Ni(iPr−pybox) system, which can be measured using the metric ∆∆G‡ := ∆G‡R − ∆G‡S , was greatly overestimated at 8.7 kcal/mol, whereas the experimentally determined enantiomeric excesses correspond to a ∆∆G‡ value of 2–3 kcal/mol. Due to these deficiencies, we have undertaken our own computational study of the Ni(iPr−pybox) system’s asymmetric catalysis reaction, using larger basis sets, more accurate DFTs and molecular models, and more potential reaction pathways considered. We have found a substantially different mechanism from that of Lin et al., with an overall reaction barrier of 15.9 kcal/mol for the formation of S-methylindane and a ∆∆G‡ value of 2.7 kcal/mol, in line with the experimentally determined values [15]. In addition, we have found a useful protocol for the rapid determination of the ∆∆G‡ enantioselectivity values of other related Ni(pybox) systems. We present some new results 156 and predictions using this new method. 5.3 Materials and methods For our calculations, the specific reaction we investigated was the coupling of racemic 1-bromoindane with methylzinc iodide to form (S)-1-methylindane, using NiI ((S, S)−iPr−pybox)Br as a catalyst or precatalyst. The solvent was taken to be DMA and the temperature 258.15 K. All quantum mechanical calculations were carried out using the Jaguar software version 7.9 developed by Schrödinger Inc. [16]. Geometry optimizations were carried out on initial guess structures, and vibrational frequencies were gathered to confirm the optimized geometries as intermediates or transition states and to construct a free energy profile. Solvation energies were calculated using the PBF Poisson-Boltzmann implicit continuum solvation model [17] in Jaguar, with a dielectric constant of 37.78 [18] and a probe radius of 2.642 Å [19] for DMA. Geometry optimization and vibrational data were calculated using the B3LYP-D3 density functional [20] with a smaller basis set, whereas single point gas-phase and solvated energies were calculated using the same functional and a larger basis set. The “-D3” suffix refers to the post-SCF correction developed by Grimme et al. that accounts for van der Waals forces [21]. Here the smaller basis set consists of a modified double-ζ Los Alamos basis set and pseudopotential (LACVP**) [22] for transition metals, and the 6-31G** basis set [13] for the other atoms; whereas the larger basis set consists of the triple-ζ Los Alamos basis set and pseudopotential (LACV3P**++) for transition metals, the same LAV3P basis set and pseudopotential for heavy main-group elements, and the 6-311G**++ basis set [23] for the other atoms. Proposed intermediates and transition states were found to have zero and one negative frequencies, confirming their status as local energy minima and saddle points, respectively. The free energy for each molecular species in solution was calculated using the formula G = Egas + ∆Gsolv + ZPE + Hvib + 6kT − T [Svib + γ(Strans + Srot − α) + β] where the last term (with α = 14.3 e.u., β = 7.98 e.u., and γ = 0.54) is an empirical approximation for the change in the translational and rotational entropy of the molecule between the gas phase and the solution phase (due to the finite librational frequencies), derived from Wertz [24]. For the monatomic anions Br– and I– , the Wertz approximation was not necessary. Hence we used the equation 5 G = Egas + ∆Gsolv + kT − T S + 1.9 2 to construct the free energy. 157 For DMA, the Gibbs free energy was calculated using the formula Gliquid = Egas + ZPE + Htot − T Stot + ∆Ggas→liquid where ∆Ggas→liquid = Gliquid − Ggas (1atm) is the free energy of condensation to liquid from 1 atm gas. We can solve for this by noting that ∆Ggas→liquid = ∆Gexpansion + Ggas→solvation , where ∆Gexpansion = Ggas (P ) − Ggas (1 atm) is the expansion of the gas from 1 atm to the vapor pressure P , and Ggas→solvation is the condensation of gas at vapor pressure P to liquid. Since a liquid is by definition at equilibrium with its vapor pressure Ggas→solvation = 0, and we thus have  ∆Ggas→liquid = Ggas (P ) − Ggas (1 atm) = RT ln P 1 atm  . We can find the vapor pressure P at a given temperature using the Antoine Equation: log10 P = A − B , C +T where the empirical parameters A, B, and C vary with the solvent and temperature. For DMA at 258.15 K, A = 6.0888, B = 2725.96, C = 28.209, so the vapor pressure is 3.71 × 10−4 atm [25]. 5.4 Results In this section, we present the results of our calculations and the likely mechanism of C−C coupling that the Ni(iPr−pybox) system undergoes. We begin by first outlining the general overarching mechanism, and then examine in depth the details of each step. 5.4.1 The overall general mechanism In the previous work of Lin et al., the authors claimed early in their report that the Ni0 -NiII cycle that a typical Negishi catalyst might undergo had an overall barrier of 57.3 kcal/mol, and consequently was not feasible [12]. Despite our concerns about their methodology (see Section 5.2.2), we do not believe that addressing these concerns would have any potential to lower the barrier enough to make the Ni0 -NiII cycle accessible. Hence, we agree with their conclusion that the Ni(iPr−pybox) system most likely goes through a NiI -NiIII cycle. We therefore restrict our mechanistic investigations to the NiI -NiIII cycle. Scheme 5.3 shows the NiI -NiIII cycle in its most general form. The catalyst alternates between a 158 NiI and a NiIII state (for example, NiI ((S, S)−iPr−pybox)Br and NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind), respectively). Conversion of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind) should proceed via methylation and addition of indanyl and bromine radicals, in any order. Conversion of NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind) back to NiI ((S, S)−iPr−pybox)Br should proceed via reductive elimination of 1-methylindane. The rest of this section will examine the detailed pathways of this generalized scheme. Scheme 5.3. The generalized mechanism of the Ni(iPr−pybox) Negishi coupling catalyst as it proceeds through a NiI -NiIII cycle. The overall equation is ind−Br + MeZnI −−→ ind−Me + ZnBrI. 5.4.2 Nature of the organozinc reagent In order to address how alkyl transfer of the methylzinc iodide to the Ni(iPr−pybox) complex occurs, it is necessary to first identify the exact nature of the methyl transferring species. Lin et al. depict the complexation of Ni(pybox) with a simple two-coordinate linear MeZnBr species in their calculations; however, it is unclear if such an electron-deficient species is at all present in a strongly polar solvent such as DMA. Indeed, the active methylation species has often been assumed to be a three coordinate MeZnX–2 [26]. Recent titration work by Organ et al. suggests, however, that the true methylating species is in fact four-coordinate dianionic MeZnX–3 with a Pd catalyst and a THF or THF/DMI 2:1 solvent (THF = tetrahydrofuran; DMI = N ,N 0 -dimethyl-2-imidazolidinone) [14]. For our mechanistic investigation, we decided to undertake a comprehensive computational investigation of the methylzinc and inorganic zinc species that may be present in the DMA solution under reaction conditions. Scheme 5.4 shows our calculated free energies of complexation of MeZnX and ZnX2 with one and two equivalents of X– , where X may be either bromine or iodine. We found that, contrary to the results of Organ et al., three-coordinate MeZnX–2 species are preferred; this discrepancy may be due to our choice of DMA as solvent, which has a much larger dielectric constant 159 of 37.78 compared to the THF as used by Organ et al. ( = 7.2). Scheme 5.4. Complexation of the various MeZnX (left) and ZnX2 (right) species with X– , where X ∈ {Br, I}. Numbers next to each arrow are the free energy change for the specific reaction. All numbers in kcal/mol. We also considered the possibility that the zinc species might form adducts with DMA, or cluster into polynuclear complexes. Given the strong polarity of the DMA solvent, we expected the former possibility to be more likely, but both pathways were explored. Figure 5.2 shows the various adducts and cluster complexes of MeZnX, whereas Figure 5.3 shows the various adducts and cluster complexes of ZnX2 . Figure 5.2. The various (MeZnI)m (dma)n species, where m may equal 0 (top row, left), 1 (bottom row), or 2 (top row, right); and n may equal 0, 1, or 2 as well. All numbers are free energies in kcal/mol and relative to MeZnI. The lowest energy species, MeZnI(dma)2 , is highlighted in red. Our results show that the most stable methylzinc species are MeZnBrI– and MeZnI(dma)2 , at −3.1 kcal/mol relative to the base MeZnI; and that the most stable inorganic zinc species is ZnBrI(dma)2 , at −15.2 kcal/mol relative to the base ZnBrI. For consistency, we have set MeZnI(dma)2 to be our active methylating agent and ZnBrI(dma)2 to be our ultimate product. Therefore, the overall reaction that we are investigating can now be written as ind−Br + MeZnI(dma)2 −−→ ind−Me + ZnBrI(dma)2 . 160 Figure 5.3. Left: The various (ZnBrI)m (dma)n species, where m may equal 0 (first row) or 1 (second to fifth rows); and n may equal 0 (first column), 1 (second column), or 2 (third and fourth columns). Right: The various (ZnBrI)4 tetrameric clusters. All numbers are free energies in kcal/mol and relative to ZnBrI. The lowest energy species, ZnBrI(dma)2 , is highlighted in red. 161 5.4.3 Mechanism for NiI oxidation and methylation We now address the mechanistic steps that must be taken for the top half of the cycle in Scheme 5.3, i.e. the conversion of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br. This conversion is expected to proceed through methylation with MeZnI(dma)2 and oxidation with ind−Br. Due to the stereoconvergent nature of the catalysis, in which the prior stereochemistry of the starting ind−Br is destroyed, the oxidation is not expected to be a typical one-step oxidative addition reaction. Instead, there is believed to be an initial one-electron oxidation in which NiII and a benzylicstabilized indanyl radical are produced, followed by recombination to form the NiIII species. These assertions are supported by experimental evidence (i.e. the formation of coupling products [15]), and the calculations of Lin et al. [12] support their feasibility. As a result, we have three possibilities for the conversion of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br: The NiI pathway: Methylation occurs first, converting NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)iPr−pybox)Me, and subsequent oxidation by ind−Br produces NiIII ((S, S)−iPr−pybox)(Me)(ind)Br; The NiII pathway: Oxidation by ind−Br to NiII ((S, S)−iPr−pybox)Br2 and free indanyl radical occurs first. Methylation to NiII ((S, S)−iPr−pybox)(Me)Br follows, and recombination with ind · produces NiIII ((S, S)−iPr−pybox)(Me)(ind)Br; The NiIII pathway: Oxidation by ind−Br proceeds to completion with NiIII ((S, S)−iPr−pybox)(ind)Br2 being formed; subsequent methylation produces NiIII ((S, S)−iPr−pybox)(Me)(ind)Br. These three pathways are shown in Scheme 5.5. The first branching point is the first step: NiI ((S, S)−iPr−pybox)Br may either undergo methylation as in the NiI pathway, or one-electron oxidation by ind−Br as in the NiII and NiIII pathways. Methylation of NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)−iPr−pybox)Me is thermodynamically uphill by 15.2 kcal/mol. Unfortunately, we were unable to find any transition states for the concerted transfer of the methyl group from MeZnI(dma)2 to any NiI species. However, we can consider a dissociative mechanism whereby the bromine on NiI ((S, S)−iPr−pybox)Br is exchanged for a methyl + group by the successive steps of Br– dissociation, MeZnI(dma)2 association, and ZnI(dma)2 dissociation (Scheme 5.6). Since each ligand association/disassociation step is expected to have minimal barriers if any, we can therefore gain an upper bound for the NiI methylation barrier by simply calculating the relative energies of each intermediate in Scheme 5.6. The results of our analysis shows that the overall barrier for the methylation of NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)−iPr−pybox)Me is between 15.2 and 20.7 kcal/mol. The NiII and NiIII pathways, by contrast, begin with the oxidation of NiI ((S, S)−iPr−pybox)Br 162 Scheme 5.5. Three potential pathways for the methylation and oxidation of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br. Red denotes the NiI pathway, green the NiII pathway, and blue the NiIII pathway. Open coordination sites may be occupied by dma solvent. Scheme 5.6. The dissociative pathway for the methylation of NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)−iPr−pybox)Me, as part of the NiI pathway in Scheme 5.5. The transition state barriers for each associative/dissociative step is expected to be minimal, if existent. Hence, the overall barrier to methylation by this pathway is 20.7 kcal/mol. All numbers are in kcal/mol and relative to NiI ((S, S)−iPr−pybox)Br. 163 by ind−Br to NiII ((S, S)−iPr−pybox)Br2 and ind · . This oxidation most likely proceeds through an inner sphere route that begins with coordination of ind−Br to the Ni center. Our results are shown in Scheme 5.7. Initial coordination of (R)- and (S)-1-bromoindane with NiI ((S, S)−iPr−pybox)Br produces an adduct NiI ((S, S)−iPr−pybox)(Br)(Br−ind) that is favorable at −4.6 and −5.2 kcal/mol relative to the initial NiI species, respectively. Loss of ind · then results in NiII ((S, S)−iPr−pybox)Br2 , whose triplet energy is −9.3 kcal/mol relative to the initial NiI species. The coordination of a DMA solvent molecule further lowers the triplet energy to −11.9 kcal/mol relative to the initial NiI species. The transition state barriers to ind−Br coordination and ind · dissociation are indeed quite minimal at less than 1.7 kcal/mol, supporting our assumptions about the NiI methylation pathway. Since NiI methylation as per the NiI pathway is uphill thermodynamically by 15.2 kcal/mol, and NiI oxidation as per the NiII and NiIII pathways is downhill by 11.9 kcal/mol with a maximum barrier of 1.2 kcal/mol, we conclude that the NiII and NiIII pathways for NiI ((S, S)−iPr−pybox)Br methylation and oxidation are much more likely than the NiI pathway, which was the only pathway investigated by Lin et al. Scheme 5.7. The oxidation of NiI ((S, S)−iPr−pybox)Br by ind−Br to NiII ((S, S)−iPr−pybox)Br2 and ind · , as part of the NiII and NiIII pathways in Scheme 5.5. The transition state barrier for the coordination of DMA to NiII ((S, S)−iPr−pybox)Br2 is expected to be minimal, if existent. Hence, the overall barrier to oxidation by this pathway is 1.2 kcal/mol for (R)-1-bromoindane and 0.4 kcal/mol for (S)-1-bromoindane. All numbers are in kcal/mol and relative to NiI ((S, S)−iPr−pybox)Br. The second branching point is the subsequent reaction of NiII ((S, S)−iPr−pybox)Br2 (dma), which can either undergo methylation to NiII ((S, S)−iPr−pybox)(Me)Br as in the NiII pathway, or recombine with ind · to form NiIII ((S, S)−iPr−pybox)Br2 (ind) as in the NiIII pathway. Scheme 5.8 shows a comparison of the essential features of the two pathways, whereas Scheme 5.11 is an unabridged diagram of the NiII pathway. Methylation of NiII ((S, S)−iPr−pybox)Br2 (dma) to NiII ((S, S)−iPr−pybox)(Me)Br as in the NiII pathway is uphill by 5.4 kcal/mol. By contrast, recombination with ind · to form NiIII ((S, S)−iPr−pybox)Br2 (ind) as in the NiIII pathway is downhill by 164 1.2 or 5.8 kcal/mol, depending on whether the indanyl ligand attaches in an (R) or (S) configuration, respectively. Since the dissociation of DMA from NiII ((S, S)−iPr−pybox)Br2 (dma) is uphill by 2.6 kcal/mol, and the addition of ind · is expected to have minimal barrier, we can conclude that the overall barrier of the reaction of NiII ((S, S)−iPr−pybox)Br2 (dma) and ind · is likely lower than that of NiII ((S, S)−iPr−pybox)Br2 (dma) and MeZnI(dma)2 , and hence that the NiIII pathway is preferred over the NiII pathway. The NiI , NiII , and NiIII pathways all converge to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br in the final step. For the NiIII pathway, this involves the methylation of the NiIII ((S, S)−iPr−pybox)Br2 (ind) formed as described in the previous Scheme 5.8. As in the NiI methylation case, we were unable to find any transition states for the concerted transfer of the methyl group from MeZnI(dma)2 to any NiIII species. Hence we again considered a dissociative mechanism whereby a bromine on NiIII ((S, S)−iPr−pybox)Br2 (ind) is exchanged for a methyl group by the successive steps of Br– + dissociation, MeZnI(dma)2 association, and ZnI(dma)2 dissociation, assuming that our barriers are minimal and thus estimating our overall methylation barriers using thermodynamic energies only. A comprehensive accounting of all species studied for this NiIII methylation reaction is given in Scheme 5.12 in Section 5.7. Scheme 5.9 shows the essential features of the mechanism of NiIII methylation. The initial NiIII ((S, S)−iPr−pybox)Br2 (ind) may lose a bromo ligand, resulting in a five-coordinate cationic III + species [Ni ((S, S)−iPr−pybox)Br(ind)] that may exist in three separate but interchangeable geometric conformations. Each conformation may then trap an equivalent of MeZnI(dma)2 , with III eq ax the species [Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] + the most favorable at III −3.1 kcal/mol relative to the starting Ni ((S, S)−iPr−pybox)Br2 ((S)−ind). However, it is the III ax ax + isomer [Ni ((S, S)−iPr−pybox)Br (ind) −Meeq −ZnI(dma)2 ] III ax III that eventually leads to the low- ax est Ni (Me)(ind)Br species, Ni ((S, S)−iPr−pybox)Br (ind) Meeq . Dissociation of ZnBrI(dma) ax ax III from NiIII ((S, S)−iPr−pybox)Br (ind) −Meeq −ZnBrI(dma) leads to [Ni ((S, S)−iPr−pybox)ax Brax (ind) Meeq and is 6.8 or 4.5 kcal/mol uphill from the starting NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind), but coordination of ZnBrI(dma) with solvent back to ZnBrI(dma)2 drives this reaction forward. As we have assumed that association and dissociation barriers are minimal for these reactions, the overall barrier for the methylation of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) is apIII ax ax proximately the difference between the highest point ([Ni ((S, S)−iPr−pybox)Br (ind) Meeq + III eq ax + ZnBrI(dma)) and the lowest point ([Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] ), ax ax equal to 9.9 and 7.6 kcal/mol for the formation of NiIII ((S, S)−iPr−pybox)Br ((R)−ind) Meeq ax and NiIII ((S, S)−iPr−pybox)Brax ((S)−ind) Meeq , respectively. 165 Scheme 5.8. NiII ((S, S)−iPr−pybox)Br2 and its subsequent methylation as part of the NiII pathway, or recombination with ind · as part of the NiIII pathway. The lowest energy NiII ((S, S)−iPr−pybox)MeBr complex is 5.4 kcal/mol higher than the starting NiII ((S, S)−iPr−pybox)Br2 , so the overall barrier for methylation is at least this much. By contrast, the transition state barrier for the coordination of ind · to NiII ((S, S)−iPr−pybox)Br2 is expected to be minimal, if existent. Hence, the overall barrier to ind · recombination is 2.6 kcal/mol, and the NiIII pathway is thus more likely to occur. All numbers are in kcal/mol and relative to NiII ((S, S)−iPr−pybox)Br2 . Red denotes the starting NiII ((S, S)−iPr−pybox)Br2 (dma) complex, green the NiII pathway, and blue the NiIII pathway. Note that indanyl recombination is reversible, so that NiIII ((R)−iPr−pybox)Br2 ((S)−ind) and NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) can interconvert. 166 Scheme 5.9. The methylation of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) to produce NiIII ((S, S)−iPr−pybox)(Br)(ind)Me. The reaction initially proceeds to [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ , a thermodynamic sink, and must increase in energy in order to reach the NiIII ((S, S)−iPr−pybox)Brax (ind)ax Meeq product. All numbers are in kcal/mol and relative to NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind). Red denotes the lowest point and thermodynamic sink, [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ ; and green the highest point, NiIII ((S, S)−iPr−pybox)Brax (ind)ax Meeq + ZnBrI(dma). 5.4.4 Mechanism for reductive elimination After the formation of NiIII ((S, S)−iPr−pybox)(Me)(ind)Br, the final step to closing the catalytic cycle is the reductive elimination of ind−Me and the reformation of NiI ((S, S)−iPr−pybox)Br (Scheme 5.3, bottom half of cycle). In the computational work of Lin et al., only the reductive eq ax elimination of ind−Me from NiIII ((S, S)−iPr−pybox)(Me) (ind) Brax was considered. However, we believe that this is not the only possible pathway for the reductive elimination of ind−Me. Specifically: ax eq 1. Reductive elimination may proceed from either NiIII ((S, S)−iPr−pybox)(Me) (ind) Brax or eq ax NiIII ((S, S)−iPr−pybox)(Me) (ind) Brax ; III 2. It is possible for the bromo ligand to be lost, resulting in five-coordinate cationic [Ni + ((S, S)−iPr−pybox)(Me)(ind)] species that may also reductively eliminate; 3. It is possible for the five coordinate cationic species to be stabilized by a solvent ligand, III + resulting in the six-coordinate cationic [Ni ((S, S)−iPr−pybox)(Me)(ind)dma] species that may also reductively eliminate; 4. It is also possible for the five coordinate cationic species to coordinate to zinc, forming an adduct NiIII ((S, S)−iPr−pybox)(Me)(ind)−Br−[Zn] that may also reductively eliminate. 167 For the last possibility, the only transition states found that contained the Ni−Br−Zn moiety III – were the geometric isomers of [Ni ((S, S)−iPr−pybox)(Me)(ind)−Br−ZnBr2 I] . In all other cases, the Br−[Zn] moiety dissociated from the Ni during geometry optimization. Figure 5.4 shows all the transition states that have been found, and their energies relative III eq ax + to [Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] , the lowest energy state of the catalyst. A reaction scheme showing the pathways to all transition states is given in Section 5.7 III as Scheme 5.13. The lowest energy transition state that produces (R)-1-methylindane is [Ni ax eq – ((S, S)−iPr−pybox)(Me) ((R)−ind) −Brax −ZnBr2 I] , with an overall barrier of 18.6 kcal/mol, III and the lowest energy transition state that produces (S)-1-methylindane is [Ni ((S, S)−iPr−pybox)ax eq + (Me) ((S)−ind) ] , with an overall barrier of 15.9 kcal/mol. Hence, the preference for (S)-1methylindane is ∆∆G‡ = 2.7 kcal/mol, in line with experimentally determined enantiomeric excesses [15]. 5.5 Discussion 5.5.1 Overall catalytic cycle Figure 5.5 is a reaction coordinate diagram that summarizes our results for the investigation of the oxidation and methylation of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br, and the subsequent reductive elimination of ind−Me and reformation of NiI ((S, S)−iPr−pybox)Br. All three pathways are shown and the NiIII pathway is clearly the easiest. The lowest point and therIII eq ax + modynamic sink, [Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] , is now considered the resting state, and all other free energies given are relative to it. The overall barrier is 18.6 and 15.9 kcal/mol for the formation of (R)-1-methylindane and (S)-1-methylindane, respectively. 5.5.2 Predictions for new ligands Now that the mechanism of Ni((S, S)−iPr−pybox) asymmetric Negishi coupling catalysis has been ascertained, it is useful to consider how modification of the ((S, S)-iPr-pybox) ligand may affect enantioselectivity. The most obvious point of modification is the R groups on the oxazoline moieties. However, the synthesis and mechanistic analysis of a multitude of R−pybox ligands would be expected to be costly and time-consuming. Hence, it is better to computationally examine a generalized mechanism of R−pybox catalysis that may be systematically and rapidly adapted for different R groups, for high-throughput screening of new R−pybox catalysts for asymmetric Negishi coupling. A generalized catalytic mechanism for the Ni(R−pybox) complexes is given in Scheme 5.10. The starting complex NiI (R−pybox)Br, G, is oxidized and methylated to NiIII (R−pybox)(Me)(ind)Br (not shown). The NiIII (R−pybox)(Me)(ind)Br species, which can exist in three geometric isomers 168 Figure 5.4. Transition states for the reductive elimination of 1-methylindane from various NiIII ((S, S)−iPr−pybox)(Me)ind species. All numbers are in kcal/mol and relative to [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ . The lowest transition states for the formation of (R)- and (S)-1methylindane are highlighted in green and blue, respectively. 169 Figure 5.5. Reaction coordinate diagram of the entire catalytic cycle for the coupling of ind−Br and MeZnI to form ind−Me. All numbers are in kcal/mol and relative to [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ . III + A1, A2, and A3, may lose Br– to form the five coordinate cationic [Ni (R−pybox)(Me)ind] species B1, B2, and B3, respectively. A DMA solvent molecule may then coordinate to the open III + coordination site to form cationic [Ni (R−pybox)(Me)(ind)dma] species C1, C2, and C3, re- spectively. Neutral A1 and A2 (but not A3) may also undergo reductive elimination through the transition states D1 and D2, respectively, to release methylindane and reform the starting complex G. Likewise, B1, B2, and B3 may undergo reductive elimination through E1, E2, and E3, I + respectively, to release methylindane and form cationic [Ni (R−pybox)] , H, which can regain the bromide to reform the starting complex G. Finally, cationic C1 and C2 (but not C3) may undergo reductive elimination through the transition states F1 and F2, respectively, to release methylindane I + and cationic [Ni (R−pybox)dma] , J, which can replace its dma ligand with a bromide to reform the starting complex G. In order to rapidly calculate the enantioselectivity of Ni(R−pybox) catalysts containing different R groups, we simply need to find the transition states D, E, and F. (In our future analyses, we will also consider Zn adducts as well, but that is not included in this study.) For our demonstration, we have selected potential R groups to be adamantyl (Ad), cyclohexyl (Cy), methyl (Me), phenyl (Ph), (R)-sec-butyl ((R)−sBu), (S)-sec-butyl ((S)−sBu), tert-butyl (tBu), and 2-phenylethyl (C2 H4 Ph). In addition, we also considered the a variant pyridine bis(indenoxazoline) ligand with fused indane rings, which we term pybindox. These are shown in Figure 5.6. Our results are given in Table 5.3. A complete collection of all calculated intermediate and transition state energies is given in Tables 5.4 through 5.6 in Section 5.7. Despite the variety of different R groups investigated, several universal characteristics may be observed from the data in Tables 5.3 through 5.6. In every case, the lowest energy intermediate was a neutral A species, usually the A2 isomer. However, the lowest transition states are of type E, 170 Scheme 5.10. A generalized catalytic mechanism that Ni(R−pybox) complexes undergo in Negishi coupling. This scheme is used to rapidly calculate the predicted enantioselectivity of a variety of complexes related to the original Ni(iPr−pybox) system. Figure 5.6. Modified ligands studied which, when complexed to Ni, may allow for different enantioselective properties as compared to the original (iPr−pybox) ligand. 171 III R iPr Ad Cy Me Ph (R)−sBu (S)−sBu tBu C2 H4 Ph pybindox [Ni (L)(Me −−− ind)Br]‡ (D) minimum ∆G‡ ∆∆G‡ (R) (S) [Ni (L)(Me −−− ind)] ‡ (E) minimum ∆G‡ ∆∆G‡ (R) (S) + [Ni (L)(Me −−− ind)dma] ‡ (F) minimum ∆G‡ ∆∆G‡ (R) (S) 18.4 19.1 28.5 20.3 17.7 18.3 18.1 18.9 17.5 20.5 15.7 14.9 20.3 15.2 14.9 12.5 15.4 16.1 12.8 13.7 3.3 −1.5 1.5 1.4 2.3 0.5 1.9 −0.6 2.1 0.3 16.4 19.3 N/A N/A 17.1 N/A N/A 19.5 N/A 17.3 17.3 19.6 28.9 18.9 16.2 15.7 17.0 18.9 19.4 18.9 1.1 −0.5 −0.3 1.4 1.6 2.6 1.2 0.0 −1.9 1.7 III 12.4 16.4 18.8 13.8 12.6 12.1 13.5 16.7 10.7 13.3 III + 15.3 17.2 30.9 19.2 15.8 N/A 16.5 18.3 14.7 15.0 1.1 2.1 N/A N/A 1.3 N/A N/A 1.2 N/A 2.3 Table 5.3. Comparison of activation energies for transition states of type D, E, and F for various Ni(R−pybox) catalysts. The Ni(iPr−pybox) data is also given here for comparison. For each transition state type, only the energy of the lowest geometric isomer is reported. For all R groups, cationic five-coordinate transition states (E) were most favored. The (R)-(S) ∆∆G‡ value is also calculated for each transition state type and (R−pybox) ligand. All numbers are in kcal/mol. Energies given are relative to the lowest calculated type A neutral NiIII (R−pybox)(Me)(ind)Br complex. Numbers in blue are the lowest energy transition states for the production of (R)-1-methylindane, whereas numbers in red are the lowest energy transition states for the production of (S)-1-methylindane. Numbers in green are the final ∆∆G‡ values for each R. (L) is ((S, S)−R−pybox). although the particular isomer (i.e. E1, E2, or E3) may vary. The enantioselectivity as measured by ∆∆G‡ varies from −1.5 to 2.3 kcal/mol, in all cases inferior to the original Ni(iPr−pybox)’s ∆∆G‡ value of 2.7 kcal/mol. In our future studies, we will seek to gain more accurate ∆∆G‡ values for our existing R groups by also considering Zn adduct formation, as we have done for Ni(iPr−pybox); and broaden our search to examine more R groups. 5.6 Conclusions In this chapter, we have computationally studied the mechanism of the Ni(iPr−pybox) system, which is a enantioselective and stereoconvergent Negishi C−C coupling catalyst. Our results support existing experimental and theoretical evidence that the mechanism proceeds through a NiI -NiIII cycle, with the stereochemistry of the initial ind−Br destroyed by homolytic cleavage of the C−Br bond. We found that NiI ((S, S)−iPr−pybox)Br is initially oxidized by ind−Br, and then methylated to form NiIII ((S, S)−iPr−pybox)(Me)(ind)Br, which can form an adduct with Zn to result in III eq ax + the resting state [Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] . Reductive elimiIII ax eq – nation then proceeds through either [Ni ((S, S)−iPr−pybox)(Me) ((R)−ind) −Brax −ZnBr2 I] III at an overall barrier of 18.6 kcal/mol to form (R)-1-methylindane), or [Ni ((S, S)−iPr−pybox)ax eq + (Me) ((S)−ind) ] at an overall barrier of 15.9 kcal/mol to form (S)-1-methylindane, with an enantioselectivity of ∆∆G‡ = 2.7 kcal/mol preference for forming the (S) enantiomer. These results 172 are also in line with experimental evidence. In addition, we formulated a systematic approach that allows us to rapidly study related Ni(R−pybox) systems, where R is Ad, Cy, Me, Ph, (R)−sBu, (S)−sBu, tBu, or C2 H4 Ph, or the III + ligand is pybindox. For these cases, we have found [Ni (R−pybox)(Me)(ind)] (E) to be universally the lowest energy transition states for the production of either (R)- or (S)-1-methylindane. Our ∆∆G‡ values measured range from −1.5 to 2.3 kcal/mol, which are not as good as the original Ni(iPr−pybox). We are continuing to improve our screening system and broadening our scope to find new catalysts for efficient asymmetric and stereoconvergent catalysis for C−C coupling. 5.7 Supporting figures Scheme 5.11 is an expanded diagram of the NiII pathway, whose essential features were presented in Scheme 5.8. Although methylation of NiII ((S, S)−iPr−pybox)Br2 to NiII ((S, S)−iPr−pybox)(Me)Br is uphill by 5.4 kcal/mol, further methylation to NiII ((S, S)−iPr−pybox)Me2 is uphill by an additional 16.7 kcal/mol. Reductive elimination of ethane is thermodynamically favorable, but the loss of methyl radical from either NiII ((S, S)−iPr−pybox)(Me)Br or NiII ((S, S)−iPr−pybox)Me2 is very uphill. Hence, the overall barrier to ethane production is at least 22.1 kcal/mol, the energy of the lowest NiII ((S, S)−iPr−pybox)Me2 species. Scheme 5.12 is an expanded version of Scheme 5.9, showing the methylation of NiIII ((S, S)−iPrIII pybox)Br2 ((S)−ind) to [Ni ((S, S)−iPr−pybox)Br(ind)Me, as well as all side reactions and isomers investigated. Scheme 5.13 shows the reactions of NiIII ((S, S)−iPr−pybox)(Br)(ind)Me that lead to ind−Me reductive elimination. (R)- and (S)-1-methylindane are produced in separate pathways, highlighted in red and blue, respectively. The overall barriers are 18.6 kcal/mol for the production of (R)-1methylindane and 15.9 kcal/mol for production of (S)-1-methylindane, leading to a ∆∆G‡ preference of 2.7 kcal/mol for the (S) isomer. Tables 5.4 through 5.6 show the calculated energies of all transition states and intermediates A through J for the high-throughput screening of Ni(R−pybox) catalysts, where R is Ad, Cy, Me, Ph, (R)−sBu, (S)−sBu, tBu, or C2 H4 Ph, or the ligand is pybindox. Specifically, Table 5.4 shows the neutral sequence A −−→ D −−→ G; Table 5.5 shows the cationic sequence B −−→ E −−→ H; and Table 5.6 shows the cationic sequence C −−→ F −−→ J (see Scheme 5.10). A future study will also include Ni−Br−Zn adducts as well. 173 Scheme 5.11. Expanded diagram of NiII ((S, S)−iPr−pybox)Br2 methylation as part of the NiII pathway. The lowest energy NiII ((S, S)−iPr−pybox)MeBr complex is 5.4 kcal/mol higher than the starting NiII ((S, S)−iPr−pybox)Br2 , and the lowest energy NiII ((S, S)−iPr−pybox)Me2 complex is 22.1 kcal/mol higher than the starting NiII ((S, S)−iPr−pybox)Br2 . All numbers are in kcal/mol and relative to NiII ((S, S)−iPr−pybox)Br2 . The lowest NiII ((S, S)−iPr−pybox)Br2 , NiII ((S, S)−iPr−pybox)(Me)Br, and NiII ((S, S)−iPr−pybox)Me2 species are highlighted in red. 174 Scheme 5.12. Expanded diagram of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) methylation as part of the NiIII pathway. All numbers are in kcal/mol and relative to NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind). Red denotes the lowest point and thermodynamic sink, [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ , as well as an alternative thermodynamic sink NiIII ((S, S)−iPr−pybox)Brax ((S)−ind)eq −Brax −ZnR MeI(dma); and green the highest point, NiIII ((S, S)−iPr−pybox)Brax (ind)ax Meeq + ZnBrI(dma), in the production of [NiIII ((S, S)−iPr−pybox)(Br)(ind)Me. 175 Scheme 5.13. The reactions of the various NiIII ((S, S)−iPr−pybox)(Me)ind species that lead to the reductive elimination of ind−Me. All numbers are in kcal/mol and relative to [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ . The starting point is NiIII ((S, S)−iPr−pybox)(Br)ax (ind)ax Meeq , which was produced at the end of Schemes 5.9 or 5.12; the end point is NiI ((S, S)−iPr−pybox)Br. Both the starting and end points are highlighted in red. The lowest energy pathways leading to the formation of (R)- and (S)-1-methylindane are highlighted in green and blue, respectively, when separate and in purple when concurrent. 176 R Ad Cy Me Ph (R)−sBu (S)−sBu tBu C2 H4 Ph pybindox NiIII ((S, S)−R−pybox)(Me)(ind)Br A1 A2 A3 (R) (S) (R) (S) (R) (S) 4.9 9.8b 5.8 0.0 3.8 3.0 5.0 2.5 5.5 2.8 8.1 2.1 1.7 0.7 1.0 3.1 0.9 2.5 2.1 2.1 1.4 1.2 2.7 2.3 2.1 2.6 1.3 0.0 0.0 0.0 1.2 0.0 0.0 0.0 0.0 0.0 5.0a 6.7a 7.7 9.4 3.7a 8.2 4.8a 9.5 5.8 10.3 11.3 6.6 8.4 6.8 6.1 9.8 6.2 5.0 III [Ni (L)(Me −−− ind)Br]‡ D1 D2 (R) (S) (R) (S) NiI (L)Br 19.1b 28.5b 20.3 17.7 18.3 18.1 18.9b 17.6 20.5 −23.9 −24.1 −27.2 −21.7 −32.1 −27.5 −25.3 −27.3 −28.0 19.6b 28.9b 18.9 16.2 15.7 17.0 18.9b 19.4 18.9 21.3 32.5 21.0 19.0 20.5 21.3 21.3 17.5 21.0 20.0 32.5 19.9 19.0 19.8 22.1 19.9 20.0 20.5 G Table 5.4. Relative energies of the neutral A, D, and G species for various Ni(R−pybox) complexes. All numbers are in kcal/mol. Energies given are relative to the lowest calculated type A neutral NiIII (R−pybox)(Me)(ind)Br complex. Numbers in blue are the lowest energy (R)−ind geometric isomers for A and D, whereas numbers in red are the lowest energy (S)−ind geometric isomer for A and D. (L) is ((S, S)−R−pybox). a) The Ni−ind bond is very long and likely to be very labile. b) An oxazoline ligand is dissociated from the Ni center in the geometry optimized structure. III R Ad Cy Me Ph (R)−sBu (S)−sBu tBu C2 H4 Ph pybindox + [Ni ((S, S)−R−pybox)(Me)(ind)] B1 B2 B3 (R) (S) (R) (S) (R) (S) 13.6 8.8 11.1 10.4 7.9 9.4 5.8 8.1 10.8 9.5 7.5 6.5 6.5 5.1 6.7 10.1 0.6 6.2 6.6 5.6 3.3 4.3 2.7 4.6 5.9 2.6 2.5 5.3 5.0 2.9 2.5 0.9 3.3 4.9 1.9 1.5 16.3a 11.7a 14.9a 15.2a 16.1a 16.0a N/Aa c 15.7a N/Aa c N/Aa c 10.6a 13.2 13.6 12.4a 13.1 12.4a 12.7 12.6 III + [Ni (L)(Me −−− ind)] ‡ E1 E2 E3 (R) (S) (R) (S) (R) (S) 21.2 N/Ac N/Ad 14.9 12.5 15.4 16.1 14.6 17.3 19.1 N/Ac N/Ad 12.6 12.1 13.5 19.7 7.0 16.2 14.9 20.3 15.2 15.1 13.8 15.8 16.9 9.1 13.7 16.4 19.4 14.9 14.8 14.0 16.0 16.7 9.6 13.3 N/Ac 26.6 17.1 N/Ac N/Ac N/Ac N/Ac 12.1 N/Ac N/Ac 18.8 13.8 N/Ac N/Ab N/Ab 20.7 10.3 15.2 I + [Ni (L)] H −13.0 −15.1 −5.4 −9.7 −18.4 −13.4 −16.2 −16.3 −11.6 Table 5.5. Relative energies of the cationic B, E, and H species for various Ni(R−pybox) complexes. All numbers are in kcal/mol. Energies given are relative to the lowest calculated type A neutral NiIII (R−pybox)(Me)(ind)Br complex. Numbers in blue are the lowest energy (R)−ind geometric isomer for B and E, whereas numbers in red are the lowest energy (S)−ind geometric isomer for B and E. (L) is ((S, S)−R−pybox). a) The Ni−ind bond is very long and likely to be very labile. b) Geometry optimized to a Meax indeq (B1/E1) configuration. c) Geometry optimized to a Meeq indax (B2/E2) configuration. d) Geometry optimized to a Meax indax (B3/E3) configuration. 177 III R Ad Cy Me Ph (R)−sBu (S)−sBu tBu C2 H4 Ph pybindox + [Ni ((S, S)−R−pybox)(Me)(ind)dma] C1 C2 C3 (R) (S) (R) (S) (R) (S) 4.1 7.4 3.8 0.4 1.7 1.8 5.8 4.3 3.7 1.5 6.9 3.5 2.3 −0.2 1.9 3.1 2.7 −0.1 0.9 1.3 3.1 2.6 3.8 3.3 2.8 1.2 −0.7 −0.3 0.6 2.5 1.7 1.6 1.4 0.8 −0.6 −2.6 4.9a 4.8a 7.0a 3.9a 7.9a 7.1a 5.1a 4.5a 4.4a 2.6a 5.4a 6.4a 2.8a 3.2a 4.3a 5.6a 4.5a 3.5a III + [Ni (L)(Me −−− ind)dma] ‡ F1 F2 (R) (S) (R) (S) 19.3 N/Ab N/Ab 17.1 N/Ab N/Ab 19.5 N/Ab 17.3 17.2 30.9 19.2 15.8 N/Ab 16.5 18.3 14.7 15.0 N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab I + [Ni (L)dma] J −25.9 −27.2 −24.6 −25.7 −30.0 −27.3 −26.5 −28.2 −25.2 Table 5.6. Relative energies of the cationic C, F, and J species for various Ni(R−pybox) complexes. All numbers are in kcal/mol. Energies given are relative to the lowest calculated type A neutral NiIII (R−pybox)(Me)(ind)Br complex. Numbers in blue are the lowest energy (R)−ind geometric isomer for C, whereas numbers in red are the lowest energy (S)−ind geometric isomer for C. (L) is ((S, S)−R−pybox). a) The Ni−ind bond is very long and likely to be very labile. b) A species containing a Ni−dma bond could not be found. 178 5.8 References [1] de Meijere, A.; Diederich, F. Metal-Catalyzed Cross-Coupling Reactions, 2nd ed.; Wiley-VCH: Weinheim, Germany; 2004. [2] Tamao, K.; Sumitani, K.; Kumada, M. J. Am. Chem. Soc. 1972, 94, 4374–4376. [3] Heck, R. F.; Nolley, Jr., J. P. J. Org. Chem. 1972, 37, 2320–2322. [4] Sonogashira, K. J. Organomet. Chem. 2002, 653, 46–49. [5] King, A. O.; Okukado, N.; Negishi, E. J. Chem. Soc. Chem. Comm. 1977, 19, 683–684. [6] Stille, J. K. Angew. Chem. Int. Ed. Engl. 1986, 25, 508–524. [7] Miyaura, N.; Yamada, K.; Suzuki, A. Tet. Lett. 1979, 20, 3437-3440. [8] Hatanaka, Y.; Hiyama, T. J. Org. Chem. 1988, 53, 918–920. [9] Fischer, C.; Fu, G. C. J. Am. Chem. Soc. 2005, 127, 4594–4595. [10] Arp, F. O.; Fu, G. C. J. Am. Chem. Soc. 2005, 127, 10482–10483. [11] Smith, S. W.; Fu, G. C. J. Am. Chem. Soc. 2008, 130, 12645–12647. [12] Lin, X; Sun, J.; Xi, Y.; Lin, D. Organometallics 2011, 30, 3284–3292. [13] (a) Hehre, W. J.; Ditchfield, R; Pople, J. A. J. Chem. Phys. 1972, 56, 2257–2261. (b) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; Defrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654–3665. [14] Angew. Chem. Int. Ed. 2012, 51, 7024–7027. [15] Buscagan, T. M.; Schley, N.; Fu, G. C. Manuscript under preparation. [16] Jaguar, version 7.9 ; Schrödinger, LLC: 2012, New York, NY. 179 [17] (a) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.; Sitkoff, D.; Nicholls, A.; Ringnalda, M.; Goddard, W. A.; Honig, B. J. Am. Chem. Soc. 1994, 116, 11875–11882. (b) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.; Ringnalda, M. N.; Sitkoff, D.; Honig, B. J. Phys. Chem. 1996, 100, 11775–11788. [18] Sears, F. W.; Zemansky, M. W.; Young, H. D. University Physics, 6th Ed.; Addison-Wesley, 1982. p [19] The probe radius r is automatically calculated by Jaguar using the formula r = 0.5833 3 MW/ρ where MW is the molecular weight and ρ is the density of the solvent. [20] (a) Becke, A. D. Phys. Rev. A. 1998, 38, 3098–3100. (b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652. (c) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B. 1988, 37, 785–789. [21] Grimme, S.; Antony, J.; Erlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104. [22] Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310. [23] (a) Hehre, W. J.; Ditchfield, R; Pople, J. A. J. Chem. Phys. 1972, 56, 2257–2261. (b) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; Defrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654–3665. [24] Wertz, D. H. J. Am. Chem. Soc. 1980, 102, 5316–5322. [25] Gopal, R.; Rizvi, S.A. J. Indian Chem. Soc. 1968, 45, 13–16. [26] Ochiai, H.; Jang, M.; Hirano, K.; Yorimitsu, H.; Oshima, K. Org. Lett. 2008, 10, 2681–2683. 180 Appendix A C−H activation using RhI complexes supported by bis-nitrogen chelating ligands Reproduced in part with permission from Webster-Gardiner, M. S.; Fu, R.; Fortman, G. C.; Gunnoe, T. B.; Nielsen, R. J.; Goddard, W. A. III. Catal. Sci. Technol., submitted for publication. Unpublished work c 2014 Royal Society of Chemistry. A.1 Abstract F F Rhodium (I) complexes (DAB )Rh(COE)(TFA) 1 and (BOZO)Rh(COE)(TFA) 2 ((DAB ) = N, N bis(pentafluorophenyl)-2,3-dimethyl-1,4-diaza-1,3-butadiene, (COE) = cyclooctene, (TFA) = trifluoroacetate, (BOZO) = bis(2-oxazolin-2-yl)) have been synthesized. Complexes 1 and 2 are catalyst precursors for the efficient C−H bond activation of arenes with TFA−D. DFT calculations determined that H/D exchange is possible through reversible oxidative addition of benzene onto the RhI complex. Catalyst 1 is reported to have a TOF of 227.5 h−1 and is the most active H/D exchange catalyst reported to date. 181 A.2 Introduction In the search for a rhodium analogue to the Shilov cycle catalyst, putative rhodium complexes are often first experimentally tested for their C−H activation ability by investigating their ability to catalyze the exchange of hydrogen and deuterium atoms on various aliphatic and aromatic substrates. towards this end, Kloek, Heinekey, and Goldberg have reported an example of a rhodium complex containing a (PNP) pincer ligand that catalyzes the H/D exchange of benzene with D2 O with a turnover frequency (TOF) of 0.1 h−1 [1]. Goldberg and Jones et al. have also reported H/D exchange with deuterated trifluoroacetic acid (TFA−D) and benzene with a pendant acetate rhodium complex, which had a TOF of 1.6 h−1 [2]. Despite the success of these rhodium catalysts for the C−H activation of arenes, the rates of reaction as determined by turnover frequency have been significantly slower, e.g., by an order of magnitude, than comparable results for Pd and Pt based catalysts [3]. Bidentate nitrogen chelating ligands have been used successfully for H/D exchange with these Pd and Pt systems. In this chapter we report the reactivity and computational study of two new rhodium complexes with bidentate nitrogen chelating ligands which have rates of arene H/D exchange comparable to platinum/palladium systems. A.3 Materials and methods F F Experimentally, complexes (DAB )Rh(COE)(TFA) 1 and (BOZO)Rh(COE)(TFA) 2 ((DAB ) = N, N -bis(pentafluorophenyl)-2,3-dimethyl-1,4-diaza-1,3-butadiene, (COE) = cyclooctene, (TFA) = trifluoroacetate, (BOZO) = bis(2-oxazolin-2-yl)) (Scheme A.1) were synthesized by reaction of [(COE)2 Rh(TFA)]2 dimer with 1 equivalent of ligand in THF at room temperature for 1 hour to afford 1 and 2 as purple solids in 58% and 73% yield, respectively. Both of these complexes were characterized by 1 H, 13 C, and 19 F NMR spectroscopy, and elemental analysis. Computational investigations were made as described in Section 3.3. A.4 Results and discussion A.4.1 Experimental results We screened these rhodium complexes for C−H activation by monitoring H/D exchange between d1 -trifluoroacetic acid and benzene. Under our conditions (1.6 mol % Rh, 1 equiv. C6 H6 in 17.5 equiv. of d1 -trifluoroacetic acid, 2 hours, 800 psi Ar, at 130◦ C), both complexes 1 and 2 showed rapid H/D exchange with turnover numbers of 89±13 and 90±11, respectively, resulting in turnover frequencies (TOFs) of 0.01 s−1 over the 2 hour period for both catalysts, as determined by GC-MS. Minimal turnovers were detected for the reaction of just [(COE)2 Rh(TFA)]2 in trifluoroacetic acid 182 Scheme A.1. Synthesis of rhodium H/D exchange catalysts (DABF )Rh(COE)(TFA) 1 and (BOZO)Rh(COE)(TFA) 2. under our conditions. Measuring the effect of temperature on catalysis (Figure A.1) shows that highest turnovers at 2 hours occur at 150◦ C. TON decreases at temperatures above 150◦ C, most likely due to catalyst decomposition. Complex 2 has the highest TON when the temperature is 130◦ C, with a decrease in turnovers at higher temperatures. Figure A.1. Temperature dependence vs. turnover numbers for complexes 1 and 2. Reactions were performed using 1.6 mol % Rh, 1 equiv. C6 H6 in 17.5 equiv. d1 -trifluoroacetic acid, for 2 hours, under 800 psi Ar. To determine the optimal solvent for catalysis, H/D exchange of benzene was then examined with other deuterated solvents at the optimized temperature. However, no significant measurable H/D exchange of benzene above the background reactions was observed for D2 O, CD3 OD, D3 PO4 , D2 SO4 , or CD3 COOD for either 1 or 2. 183 We then sought to determine the dependence on catalyst concentration. By lowering the catalyst loading to 0.4% relative to benzene, a large increase in TON was observed (Figure A.2). After 2 hours at 150◦ C, complex 1 showed 456 TON to give a TOF of 228 h−1 . Figure A.2. Concentration of Rh dependence vs. TON for complexes 1 and 2. Reactions were performed at 150◦ C for 1 and 130◦ C for 2, with 1 equiv. C6 H6 in 17.5 equiv. d1 -trifluoroacetic acid, for 2 hours, under 800 psi Ar. The selectivity of the reaction mechanism was determined by examining the H/D exchange of toluene in trifluoroacetic acid at each catalyst’s respective optimal temperature. The ortho:meta:para selectivity is 6.87 : 1 : 6.39 and 5.68 : 1 : 5.24 for 1 and 2, respectively. No clear deuteration of the methyl fragment of toluene was seen. A crucial aspect of catalysis is the recyclability of the catalyst. To study this, each complex was run and sampled, and then the contents of the reaction were evaporated off, leaving behind the catalyst. Fresh trifluoroacetic acid and benzene were added to the reaction vessel and the experiment was run again. For complex 1, this was successfully done over four times for a period of over 72 hours, with H/D exchange observed each time. An interesting question is the role of cyclooctene in the mechanism. The addition of cyclooctene to the experiment showed a clear decrease in turnover number for complex 2. We hypothesized that this is due to COE binding to the rhodium metal competitively to slow down catalysis. However, upon careful consideration of the 1 H and 13 C NMR spectra, clear peaks for the trifluoroacetate adduct of cyclooctene can be seen. Indeed, a report from Nordlander indicates that cyclooctene in trifluoroacetic acid funnels to a cyclooctyl trifluoroacetate complex [4]. Attempts to synthesize either complex with a trifluoroacetic acid ligand instead of cyclooctene were unsuccessful. Attempts to extend this reactivity to substrates other than aromatics were investigated. H/D exchange was attempted with methane under a variety of pressures and temperatures; however, no 184 deuterated methane species was observed by GC-MS. In addition, H/D exchange of benzene sp2 C−H bonds with cyclohexane was also attempted, but no deuteration was observed after 2 hours at the optimal conditions for each catalyst. A.4.2 Computational results We have also undertaken a computational investigation for the mechanism of H/D exchange by F complexes 1 and 2. We set our reference complex to (L)Rh(TFA)(TFAH) (3 and 4 for L = (DAB ) or (BOZO), respectively); these being the presumed species after the (COE) ligand is hydrotrifluoroacetylated off from the base (L)Rh(TFA)(COE) species. We hypothesized that C−H activation of benzene could proceed through one of two routes. The first route is by direct oxidative addition I III to (L)Rh (TFA)(TFAH) to form (L)Rh (TFA)(TFAH)(Ph)(H). H/D exchange could then occur beI − * cause the RhIII (H) bond can be reversibly reductively deprotonated: RhIII (TFA)(H) − ) − − Rh (TFAH). I III The second is by an internal protonation (L)Rh (TFA)(TFAH) −−→ (L)Rh (TFA)2 (H), followed by either benzene coordination and deprotonation by TFA via a six-membered ring transition state, F or direct hydrogen exchange with the RhIII (H) hydride. We found that, in both the (DAB ) and (BOZO) ligands, the first scenario (direct oxidative addition) was more likely, with lower transition states. Direct oxidative addition of benzene by 3 or 4 is shown in Scheme A.2. The lowest barrier found for 3 is 22.2 kcal/mol at 298 K and 23.7 kcal/mol at 498 K. In contrast, the lowest barrier found for 4 is 18.5 kcal/mol at 298 K and 19.5 kcal/mol at 498 K. All of these values imply accessible benzene oxidative addition at the reaction temperatures investigated. The increased activity of 3 compared to 4 is most likely simply due to the increased temperature that H/D exchange with catalyst 1 can be run without decomposition. A comparison of direct oxidative addition and benzene coordination/deprotonation is shown in Scheme A.3 for complex 3 and Scheme A.4 for complex 4. Scheme A.2. Oxidative addition of benzene by Rh(DABF ) and Rh(BOZO) complexes 3 and 4. H/D exchange most likely proceeds through these transition states. All free energies in kcal/mol. Methane activation by complex 3, in a manner analogous to benzene activation, was also in- 185 Scheme A.3. The Rh(DABF ) complex 3 activates benzene with a barrier of 22.2 kcal/mol at 298 K and 23.7 kcal/mol at 498 K. Top row: direct oxidative addition of benzene by RhI . Elsewhere: internal oxidative addition of 3 followed by benzene coordination. All free energies in kcal/mol. 186 Scheme A.4. The Rh(BOZO) complex 4 activates benzene with a barrier of 18.5 kcal/mol at 298 K and 19.5 kcal/mol at 498 K. Bottom row, first and second TSs: Direct oxidative addition of benzene by RhI . Elsewhere: internal oxidative addition followed by benzene coordination. All free energies in kcal/mol. 187 vestigated, with the results shown in Scheme A.5. The Rh(Me) intermediates and transition states were approximately 7-19 kcal/mol higher in energy than their Rh(Ph) counterparts. In particular, the direct oxidative addition of methane has in its lowest transition state a barrier of 32.4 kcal/mol at 298 K and 37.6 kcal/mol at 498 K, which is prohibitively high under reaction conditions. This is consistent with the lack of experimentally observed methane activation by complex 1. Scheme A.5. Activation of methane by the Rh(DABF ) complex 3 in a manner analogous to the oxidative addition of benzene (see Scheme A.3). The lowest transition state found has a barrier of 32.4 kcal/mol at 298 K and 37.6 kcal/mol at 498 K, rendering methane activation inaccessible. All free energies in kcal/mol. A.5 Conclusions This work has demonstrated two efficient rhodium catalysts for H/D exchange between benzene and TFAD. To the best of our knowledge, catalyst 1 is the most active rhodium catalyst yet discovered for benzene H/D exchange. These results demonstrate TOFs similar to platinum and palladium systems. In addition, these rhodium complexes were studied under catalytically relevant conditions for industrial C−H activation and functionalization. 188 A.6 References [1] Kloek, S. M.; Heinekey, D. M.; Goldberg, K. I. Angew. Chem. Int. Ed., 2007, 46, 4736–4738. [2] Rhinehart, J. L.; Manbeck, K. A.; Buzak, S. K.; Lippa, G. M.; Brennessel, W. W.; Goldberg, K. I.; Jones, W. D. Organometallics 2012, 31, 1943–1952. [3] (a) Muehlhofer, M.; Strassner, T.; Hermann, W. A. Angew. Chem. Int. Ed. 2002, 41, 1745–1747; (b) Sen, A.; Benvenuto M. A.; Lin M. R.; Hutson A. C.; Basickes N. J. Am. Chem. Soc. 1994, 116, 998–1003. (c) Hickman, A. J.; Cismesia, M. A.; Sanford, M. S. Organometallics 2012, 31, 1761-1766. [4] Nordlander, J. E.; D. Kotian, K. D.; Raff, D. E. II; Njoroge, G. F.; Winemiller, J. J. J. Am. Chem. Soc. 1984, 106, 1427–1432. 189 Appendix B Energy levels of group 10 transition metal atoms and ions B.1 Abstract The energies of the group 10 transition metals (Ni, Pd, and Pt) in different configurations (d8 s2 , d9 s1 , and d10 ) are calculated using various DFT methods. Furthermore, energies of the corresponding 1+ and 2+ ions are also calculated. The results of the various DFT methods are compared with the experimental energies [1] and previous calculations [2], with the purpose of determining the best DFT method for each atom. Finally, inconsistent energies in supposedly degenerate orbitals of the ionic states are observed and a potential solution is proposed. 190 Table B.1. Relativistic vs. nonrelativistic Hartree-Fock excitation energies, taken from Martin and Hay [2]. Atom Nickel Palladium Platinum Config. Hartree Fock Relativistic HF Experimental 8 2 0.00 1.27 5.47 0.00 −3.01 −3.76 0.00 −3.28 −4.69 0.00 1.63 6.04 0.00 −2.09 −2.19 0.00 −0.40 0.50 0.00 −0.03 1.71 0.00 −2.43 −3.38 0.00 −0.64 −0.16 d s d9 s1 d10 d8 s2 d9 s1 d10 d8 s2 d9 s1 d10 Relative energies in eV. The experimental energies are averages over spin-orbit components [1]. B.2 Introduction Density function theory has found widespread application in computational chemistry as a fast alternative to ab initio calculations with correlation. However, the accuracy of DFT calculations is sometimes called into question in many transition metal-containing systems, due to the fact that transition metals often have several closely-spaced low-lying energy levels and that DFT is not always able to predict the correct ground state. In this report we present the results of calculations of the levels of Ni, Pd, and Pt with the following DFTs: B3LYP, M06, M06-2X, M06-HF, M06-L, PBE, PBE0, and XYGJ-OS (Ni only); and we contrast these with a few electron-correlation calculations, including HF, MP2, and CCSD, and with previous calculations done by Martin and Hay [2]. B.3 The neutral atom The group 10 transition metals are notable in that each atom in the column has a different ground state electron configuration, viz. d8 s2 for Ni, d10 for Pd, and d9 s1 for Pt. While Ni conforms to the Aufbau principle, Pd and Pt do not. This is due to a complex interplay of electron correlation and relativistic effects. The ordering of the ground and excited states of Pd and Pt have long been known (Table B.1), but the identity of the ground state of Ni has been somewhat more contentious [3]. There are a few observations to note in Table B.1. First, the experimentally determined ground state of Ni is d8 s2 3 F4 , but the energies in Table B.1 are averages over all J = 2, 3, 4, which in the case of Ni results in an energy that is slightly higher than the 3 P average. Secondly, we find that relativistic corrections, though unnecessary for Ni, become important for Pd and absolutely vital for Pt. Finally, we find that the Hartree-Fock calculations do a poor job of ordering the configurations 191 Table B.2. Relativistic vs. nonrelativistic Hartree-Fock excitation energies for the M+ ions [2]. Atom Config. Hartree Fock Relativistic HF Experimental Nickel 8 2 0.00 7.61 6.98 0.00 2.91 6.36 0.00 2.68 6.26 0.00 8.12 7.09 0.00 4.25 6.66 0.00 7.26 7.51 0.00 7.59 8.67 0.00 5.13 8.32 0.00 8.46 9.22 Ni+ Palladium Pd+ Platinum Pt+ d s d9 d8 s1 d8 s2 d9 d8 s1 d8 s2 d9 d8 s1 Relative energies in eV. Table B.3. Excitation energies of Pt atom as calculated by various DFTs. Config. d9 s1 d10 d8 s2 Exp. 11.07 14.76 HF −118.23 25.41 5.95 B3LYP −119.08 14.67 6.84 M06 −119.03 0.40 19.17 M06-2X −118.95 11.72 15.24 M06-HF −118.91 24.95 21.07 M06-L −119.08 14.01 0.86 PBE −119.07 14.39 −0.05 PBE0 −119.04 15.03 9.08 The d9 s1 row contains the absolute energies in hartrees. The other rows contain the energies relative to the d9 s1 ground state, in kcal/mol. Experimental energies derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 1 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. for the Pt atom. Hence, the first part of this report shall many focus on DFT calculations of the Pt atom, with additional data for Ni and Pd for purposes of comparison. We also examine the previously-calculated excitation energies for the singly-charged cations Ni+ , Pd+ , and Pt+ . These numbers are taken from the same reference [2] and are reproduced in Table B.2. Looking at the values in Table B.2, two things are apparent: First, the numbers for Ni+ , while close, are in the wrong order, putting the d8 s1 configuration below the d9 configuration in energy. And secondly, while the HF calculations for Pd+ and Pt+ give the correct ground state configuration, the relative energies are significantly off. This report will also give updated excitation energy calculations for the singly-charged cations. B.3.1 Platinum (0) Calculations with Pt were performed on Jaguar version 7.5 [4]. The basis set used was a variant of LACV3P**++, modified to include f functions [5]. The results are presented in Table B.3. 192 Table B.4. Excitation energies of Pt atom calculated without f functions. Config. d9 s1 d10 d8 s2 δδ (a) ππ (b) σδ (c) Exp. 11.07 14.76 14.76 HF −118.23 27.96 1.41 30.24 8.50 1.25 B3LYP −119.08 14.67 5.10 28.38 12.77 7.57 M06 −119.03 0.40 12.22 26.43 19.62 17.36 M06-2X −118.95 11.72 15.04 31.96 19.46 15.30 M06-HF −118.91 24.95 21.11 39.30 30.11 27.04 M06-L −119.08 14.01 6.99 32.84 16.22 10.68 PBE −119.07 14.43 0.02 29.52 15.22 10.50 PBE0 −119.04 15.03 8.24 30.94 15.26 10.03 9 1 The d s row contains the absolute energies in hartrees. The other rows contain the energies relative to the d9 s1 ground state, in kcal/mol. Experimental energies derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 1 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. (a) Energy of the d8 s2 configuration with singly occupied dxy and dx2 −y2 orbitals. (b) Energy of the d8 s2 configuration with singly occupied dxz and dyz orbitals. (c) Extrapolated energy of the ground d8 s2 configuration according to the formula Eσδ = 34 Eππ − 13 Eδδ . An examination of the data presented in Table B.3 shows that while most DFT methods are very good at predicting the d9 s1 − d10 separation, they are considerably poorer at correctly ordering the d8 s2 state, often putting it near the d9 s1 state instead of above the d10 state. An examination of the molecular orbitals in the d8 s2 output files in Molden revealed considerable mixing of the d and s orbitals in some of the cases. Hence, the true configuration, d9 s1 or d8 s2 , could not be ascertained, and hence the values could not be trusted. In order to prevent this catastrophic mixing of orbitals, the input files for the d8 s2 cases were rerun with the symmetry turned on (Table B.4, row 3). This prevented the basis set from using the added f functions. Fortunately, an examination of the rerun numbers for the d9 s1 and d10 cases (Table B.4, rows 1-2) showed that the f functions play a very minor role in the total energy and hence can be safely ignored. Unfortunately, examination of the molecular orbitals in the d8 s2 output files in Molden revealed that there was still some mixing between the filled s orbital and the empty dz2 orbital among the β electrons. This is due to both orbitals being totally symmetric in the point group used for the calculations (D4h or its largest Abelian subgroup D2h ). Hence, instead of calculating the lowest d8 s2 state directly, it may be more accurate to calculate higher d8 s2 states and extrapolate the ground d8 s2 (see Section B.6). The revised energies are given in Table B.4, rows 4-6. Electron-correlation calculations were also performed on the Pt atom, on Q-Chem version 4.0 using the MP2 and CCSD methods on the LACVP basis set with ECP. A guess wavefunction with the correct orbital population was first generated by DFT methods, and then the electron-correlation values were calculated with the orbital overlap maximized. The results, given in Table B.5, are in decent agreement with the experimental values, but are not an improvement over DFT methods. 193 Table B.5. Excitation energies of Pt atom calculated using electron-correlation methods. Config. d9 s1 d10 d8 s2 δδ (a) ππ (b) σδ (c) Exp. 11.07 14.76 14.76 MP2 −118.284 13.41 5.43 30.16 11.37 5.10 CCSD −118.289 16.83 0.92 12.41 3.29 0.24 9 1 The d s row contains the absolute energies in hartrees. The other rows contain the energies relative to the d9 s1 ground state, in kcal/mol. Experimental energies derived from Moore [1]. Color coding represents deviation from experimental value. Blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. (a) Energy of the d8 s2 configuration with singly occupied dxy and dx2 −y2 orbitals. (b) Energy of the d8 s2 configuration with singly occupied dxz and dyz orbitals. (c) Extrapolated energy of the ground d8 s2 configuration according to the formula Eσδ = 34 Eππ − 13 Eδδ . Unfortunately, neither method gave the correct ordering of states; both methods put the singlet state at the highest energy. Furthermore, CCSD performed more poorly than MP2. This observation is probably due to the fact that CCSD is designed for calculating ground state energies. We can draw several conclusions from the data presented for the Pt atom. First, DFT values are closer to the relativistic HF calculations than to the nonrelativistic calculations (see Table B.1). This is due to the effective core potential taking relativistic effects into account. Secondly, for the d9 s1 − d10 transition, most DFTs gave values that were a little high, while the HF calculation gave a value that was much too high. Thirdly, most DFTs ordered the relative energy levels of the different configurations in the same way as Martin and Hay’s relativistic HF values, differently from the experimental result. And finally, M06-2X gave the best values. B.3.2 Palladium (0) Calculations with Pd were performed on Jaguar version 7.5 using the basis set LACV3P**++ [4], and on Q-Chem version 4.0 using the basis set LACVP [6]. The results from Jaguar are presented in Table B.6; however, it was difficult to calculate the d8 s2 configuration properly due to the wide separation from d9 s1 and the aforementioned symmetry issues. Hence, all DFT values were recalculated in Q-Chem; the results are presented in Table B.7. The d10 − d9 s1 energy gap is very similar to the numbers in Jaguar; hence we can expect reasonably accurate values for the d8 s2 configuration as well. In addition, the use of Q-Chem allows us to take advantage of the electron-correlation MP2 and CCSD methods. It can be concluded from the tables for palladium that the best overall method was MP2, whereas 194 Table B.6. d10 − d9 s1 Excitation energies of Pd atom as calculated by Jaguar. Config. d10 d9 s1 Exp. 21.91 HF −125.87 1.97 B3LYP −126.72 20.79 M06 −126.72 38.15 M06-2X −126.61 27.53 M06-HF −126.52 10.89 M06-L −126.75 28.16 PBE −126.68 22.28 PBE0 −126.67 20.72 The d10 row contains the absolute energies in hartrees. The d9 s1 row contains the energies relative to the d10 ground state, in kcal/mol. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. Table B.7. Excitation energies of Pd atom as calculated by Q-Chem. Config. d10 d9 s1 d8 s2 Exp. 21.91 77.94 HF −125.87 1.72 55.17 MP2 −125.97 22.42 80.54 CCSD −125.96 15.95 67.28 B3LYP −126.71 19.98 84.79 M06 −126.71 38.63 114.45 M06-2X −126.61 26.07 97.61 M06-HF −126.51 10.09 89.50 M06-L −126.74 30.66 97.18 PBE −125.90 9.43 67.17 PBE0 −126.67 19.99 88.40 The d10 row contains the absolute energies in hartrees. The other rows contain the energies relative to the d10 ground state, in kcal/mol. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. the best DFT calculation was B3LYP. B.3.3 Nickel (0) Unlike the cases for Pt and Pd, calculations with Ni were performed without any effective core potential, using the basis set G3LARGE with the program Q-Chem version 4.0. The results are summarized in Table B.8 for the electron-correlation methods and Table B.9 for the DFT methods. It can be concluded from the tables for nickel that the best overall method was CCSD, whereas Table B.8. Excitation energies of Ni atom as calculated by Q-Chem; electron-correlation methods. Config. d8 s2 d9 s1 d10 Exp. −0.69* 39.43 HF −1506.83 1.72 55.17 MP2 −1507.83 −30.92 −44.60 CCSD −1507.78 0.49 47.07 The d10 row contains the absolute energies in hartrees. The other rows contain the energies relative to the d10 ground state, in kcal/mol. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; red: within 25 kcal/mol. * See the discussion for Table B.1. 195 Table B.9. Excitation energies of Ni atom as calculated by Q-Chem; DFT methods. Config. d8 s2 d9 s1 d10 Exp. −0.69* 39.43 XYGJ-OS −1507.36 −5.66 19.57 B3LYP −1508.31 −9.11 22.65 M06 −1508.19 −10.33 19.43 M06-2X −1508.24 −3.59 48.36 M06-HF −1508.35 −14.01 49.11 M06-L −1508.03 5.26 50.00 PBE −1506.75 −12.30 29.37 PBE0 −1508.02 −9.18 85.06 The d10 row contains the absolute energies in hartrees. The other rows contain the energies relative to the d10 ground state, in kcal/mol. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol; red: within 25 kcal/mol. * See the discussion for Table B.1. Table B.10. Excitation energies of the group 10 atoms from d9 to d8 s1 , as calculated using various DFTs. + 9 Ni d d8 s1 ∆E Pd+ d9 d8 s1 ∆E Pt+ d9 d8 s1 ∆E Exp. 175.03 199.93 24.90 197.30 271.61 74.31 195.09 212.62 17.53 XYGJ-OS 390.57 392.65 2.08 N/A N/A N/A N/A N/A N/A B3LYP 173.62 201.51 27.89 198.16 287.29 89.14 217.38 242.80 25.42 M06 166.76 209.30 42.54 209.32 323.64 114.33 207.69 253.62 45.92 M06-2X 167.85 199.48 31.63 196.01 298.99 102.98 206.89 242.30 35.41 M06-HF 156.59 187.66 31.07 182.01 280.43 98.42 212.36 238.70 26.34 M06-L 75.48 105.56 30.08 196.53 297.30 100.77 203.26 237.12 33.86 PBE 164.53 183.01 18.48 177.85 250.10 72.25 208.54 233.96 25.42 PBE0 167.67 193.56 25.90 193.77 282.91 89.14 211.33 235.47 24.14 All numbers are given in kcal/mol relative to d8 s2 for Ni, d10 for Pd, and d9 s1 for Pt. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol; red: within 25 kcal/mol. XYGJ-OS not implemented for ECP; hence only Ni was calculated. the best DFT calculation was M06-2X. B.4 The singly-charged cation Whereas the neutral group 10 metals each have different ground state configurations, the singlycharged cations Ni+ , Pd+ , and Pt+ are uniformly d9 in the ground state. The excitation energy from d9 to d8 s1 for each metal has been calculated by various DFTs and electron-correlation methods. A summary of the results, along with the experimental values [1], is given in tables B.10 and B.11. All calculations were done on Q-Chem version 4.0. The G3LARGE basis set was used for Ni+ while the LACVP basis set with ECP was used for Pd+ and Pt+ . An examination of the data presented in tables B.10 and B.11 shows that CCSD is unequivocally the best method for Ni+ , while B3LYP is the best DFT method. M06-2X and PBE0 also gave good 196 Table B.11. Excitation energies of the group 10 atoms from d9 to d8 s1 , as calculated using various electroncorrelation methods. Ni+ d9 d8 s1 ∆E Pd+ d9 d8 s1 ∆E Pt+ d9 d8 s1 ∆E Exp. 175.03 199.93 24.90 197.30 271.61 74.31 195.09 212.62 17.53 HF 252.85 162.26 −90.59 147.93 217.87 69.94 178.87 188.39 9.52 MP2 140.27 192.18 51.91 176.65 283.37 106.72 188.91 205.44 16.54 CCSD 167.93 192.82 24.90 171.10 251.33 80.22 188.25 205.81 17.57 All numbers are given in kcal/mol relative to d8 s2 for Ni, d10 for Pd, and d9 s1 for Pt. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; red: within 25 kcal/mol. results. As for Pt+ , both CCSD and MP2 gave excellent results. Unfortunately, there was no single method for Pd+ that gave good results for all energies: most methods were good at estimating the d9 state; only M06-HF and PBE0 came close to estimating the d8 s1 state; and HF, CCSD, and PBE were able to approximate the energy difference between the ion states well. Overall, however, these numbers are all nonetheless in better agreement with experimental data than the previous calculations as presented in Table B.2. B.5 Angular momentum-dependent artifacts in DFT calculations During the course of calculating these energies it became apparent that certain artifacts were being introduced in the calculations of the cations. These errors have been documented before [7] and the reason is innate to the nature of the exchange functional of DFT [8]. Potential solutions have been proposed [9] [10] but not yet widely implemented. B.5.1 Angular momentum of the electron/hole in the d1 /d9 configuration of the cation The various microstates of a d1 or d9 configuration give rise to only one permissible spin state: 2 D. Hence, it should not matter which real d orbital contains the electron or hole, respectively. However, we find that for DFT methods the calculated energy does vary: For example, according to B3LYP the 197 Table B.12. “Degenerate” energies of the Pt+ d9 ground state. d9 hole z2 xy x2 − y 2 xz yz LACV3P**++ Energy Erel −118.739 0.000 −118.741 −1.123 −118.741 −1.123 −118.737 1.141 −118.737 1.141 LACV3P Energy Erel −118.738 0.000 −118.740 −1.157 −118.740 −1.154 −118.736 1.079 −118.736 1.078 LAV1S Energy Erel −26.961 0.000 −26.960 0.791 −26.960 0.870 −26.958 1.591 −26.958 1.591 Calculations done in Jaguar. The “Energy” columns contain values in hartrees, whereas the “Erel ” columns contain values in kcal/mol. Table B.13. “Degenerate” energies of the Pt9+ d1 ground state. d1 z2 xy 2 x − y2 xz* yz* Energy −98.151343 −98.151819 −98.151822 −98.151822 −98.151822 Erel 0.000 −0.299 −0.301 −0.301 −0.301 M06-2X/LACV3P**++ done in Jaguar. The “Energy” column contains values in hartrees, whereas the “Erel ” column contains values in kcal/mol. * Jaguar could only run these calculations using Abelian (D2h ) symmetry. Therefore, when the electron was localized in a π symmetry orbital, the dx2 −y2 and dz2 orbitals mixed and thus the dxz or dyz orbital was reanalyzed into a δ symmetry orbital. Hence the energy is not expected to differ from the energy of the dxy or dx2 −y2 hole. energy of the ion Pt+ is different whether the hole is of σ symmetry (dz2 , L = 0), π symmetry (dxz or dyz , L = 1), or δ symmetry (dx2 −y2 or dxy , L = 2). The specific numbers are given in Table B.12. Similarly, the energy of Pt9+ as calculated by M06-2X depends (completely unphysically) on which orbital the lone d electron is placed (Table B.13). A more extensive study was done on the Ni+ ion, using Q-Chem version 4.0. The results of calculations using DFT methods are given in Table B.14 and the results of calculations using electroncorrelation methods are given in Table B.15. It is clear that, whereas the electron-correlation methods correctly predict the degeneracy of the five d holes in 2 D Ni+ , the DFT methods mostly do not. The only exception among the DFTs is XYGJ-OS, a double-hybrid DFT. 198 Table B.14. “Degenerate” energies of the Ni+ d9 ground state as calculated by DFT methods. d9 z2 xy 2 x − y2 xz* yz* B3LYP 0.000 0.293 0.289 0.290 0.290 M06 0.000 0.407 0.414 0.416 0.430 M06-2X 0.000 −0.589 −0.560 −0.583 −0.568 M06-HF 0.000 −1.681 −1.603 −1.675 −1.647 M06-L 0.000 −0.285 −0.329 −0.276 −0.287 PBE 0.000 −0.111 −0.109 −0.111 −0.109 PBE0 0.000 0.372 0.369 0.358 0.359 XYGJ-OS 0.000 0.000 −0.001 0.018 0.016 Energies are relative to the dz2 hole and in kcal/mol. * Because Q-Chem runs the calculations using Abelian (D2h ) symmetry, when the electron was localized in a π symmetry orbital, the dx2 −y2 and dz2 orbitals mixed and thus the dxz or dyz orbital was reanalyzed into a δ symmetry orbital. Hence the energy is not expected to differ from the energy of the dxy or dx2 −y2 hole. Table B.15. Degenerate energies of the Ni+ d9 ground state as calculated by electron-correlation methods. d9 z2 xy 2 x − y2 xz* yz* HF 0.000 −0.012 −0.008 0.000 −0.001 MP2 0.000 0.007 0.007 0.023 0.009 CCSD 0.000 −0.002 −0.002 −0.002 −0.002 Energies are relative to the dz2 hole and in kcal/mol. * Because Q-Chem runs the calculations using Abelian (D2h ) symmetry, when the electron was localized in a π symmetry orbital, the dx2 −y2 and dz2 orbitals mixed and thus the dxz or dyz orbital was reanalyzed into a δ symmetry orbital. Hence the energy is not expected to differ from the energy of the dxy or dx2 −y2 hole. 199 Table B.16. Energies of the Pt2+ d8 configuration, calculated by B3LYP/LACV3P**++. Hole type z , x2 − y 2 σδ xy, z 2 σδ xz, yz ππ xz, x2 − y 2 πδ yz, x2 − y 2 πδ xz, xy πδ yz, xy πδ 2 xz, z σπ σπ yz, z 2 2 2 xy, x − y δδ 2 Energy −118.0461 −118.0461 −118.0372 −118.0366 −118.0366 −118.0359 −118.0359 −118.0201 −118.0201 −118.0145 Erel 0.00 0.02 5.59 6.00 6.00 6.43 6.43 16.33 16.33 19.84 Energies are in hartrees; Erel values are in kcal/mol. B.5.2 Angular momentum of the holes in the d8 configuration of the dication When there are two holes, there are two triplet states: 3 F and 3 P . In terms of the real orbital holes, there are several intermediate energy levels: the σδ holes are at 3 F ; the ππ and πδ holes are at 13 23 33 13 43 43 5 F + 5 P , the σπ holes are at 5 F + 5 P , and the δδ holes are at 5 F + 5 P . These values are derived in the Section B.6. Table B.16 has the values for Pt2+ calculated in B3LYP with the LACV3P**++ basis set in Jaguar. While the numbers are close to the theoretically predicted values, the ππ and πδ holes are split into three different sets of closely spaced energy levels. As in the case of the d9 M+ ions, this is an artifact of the DFT calculation and is not physical. Similar calculations were also performed for Pd2+ . The results are shown in Table B.17. However, due to difficulties with Jaguar, Q-Chem was used instead. The reduction of the atom to an Abelian subgroup prevented the calculation of πδ and σπ holes, however. Fortunately, the calculated σδ, ππ, and δδ holes are in good agreement with experiment. The most extensive calculations were done on nickel. Values calculated with Jaguar are given in Table B.18 while values calculated with Q-Chem and experimental values are given in Table B.19. Again, note that it was not possible to get πδ and σπ hole energies with Q-Chem. B.6 Energies of d2 and d8 triplet states Given a d2 or d8 atomic configuration, we can easily derive that the various microstates give rise to the 3 F and 3 P triplet states and the 1 G, 1 D, and 1 S singlet states. Furthermore, we know from 200 Table B.17. Energies of the Pd2+ d8 configuration, calculated by B3LYP/LACVP. Hole type z 2 , x2 − y 2 σδ xy, z 2 σδ xz, yz ππ xz, x2 − y 2 πδ yz, x2 − y 2 πδ xz, xy πδ yz, xy πδ xz, z 2 σπ yz, z 2 σπ xy, x2 − y 2 δδ Energy −125.663 −125.663 −125.651 −125.663 −125.663 −125.651 −125.651 −125.663 −125.663 −125.619 Erel 0.00 0.01 7.45 0.01 (c) 0.01 (c) 7.45 (d) 7.45 (d) 0.01 (c) 0.01 (c) 27.93 Exp/J (a) 0.00 0.00 7.83 7.83 7.83 7.83 7.83 23.50 23.50 31.33 Exp/avg (b) 0.00 0.00 6.46 6.46 6.46 6.46 6.46 19.38 19.38 25.84 Energies are in hartrees; Erel and experimental values are in kcal/mol. Experimental values from [1]. (a) Experimental value using the lowest J energy. (b) Experimental value derived from the average of spin-orbit coupling. (c) Collapsed to σδ. (d) Collapsed to ππ. Table B.18. Energies of the Ni2+ d8 configuration, calculated on Jaguar using the 6-31G(tm)**++ basis. Hole type z , x2 − y 2 σδ xy, z 2 σδ ππ xz, yz 2 2 xz, x − y πδ yz, x2 − y 2 πδ xz, xy πδ yz, xy πδ xz, z 2 σπ yz, z 2 σπ 2 2 xy, x − y δδ 2 HF 0.00 0.00 11.29 11.62 11.62 11.62 11.62 34.34 34.34 45.42 B3LYP 0.00 0.05 10.25 9.80 9.80 10.42 10.42 28.33 28.33 38.44 M06 0.00 −0.27 10.35 7.50 7.50 10.49 10.49 18.12 18.12 23.84 M06-2X 0.00 −0.83 4.13 6.05 6.05 4.29 4.29 19.90 19.90 26.92 M06-HF 0.00 −0.53 −7.93 0.52 0.52 −7.70 −7.70 10.28 10.28 14.84 Energies are in kcal/mol. (a) Experimental value using the lowest J energy. (b) Experimental value derived from the average of spin-orbit coupling. (c) Collapsed to σδ. (d) Collapsed to ππ. M06-L 0.00 −0.42 16.22 13.17 13.17 16.38 16.38 35.85 35.85 48.36 PBE 0.00 0.08 10.23 9.39 9.39 10.38 10.38 26.60 26.60 36.66 PBE0 0.00 0.06 10.37 9.79 9.79 10.53 10.53 28.15 28.15 38.33 201 Table B.19. Energies of the Ni2+ d8 configuration, experimental and calculated on Q-Chem using the G3LARGE basis. Hole type z 2 , x2 − y 2 σδ σδ xy, z 2 xz, yz ππ xz, x2 − y 2 πδ yz, x2 − y 2 πδ πδ xz, xy yz, xy πδ xz, z 2 σπ yz, z 2 σπ 2 2 xy, x − y δδ Exp/J (a) 0.00 0.00 9.53 9.53 9.53 9.53 9.53 28.58 28.58 38.11 Exp/avg (b) 0.00 0.00 9.00 9.00 9.00 9.00 9.00 27.01 27.01 36.02 XYGJ-OS 0.00 −0.03 9.39 0.00 (c) 0.00 (c) 9.39 (d) 9.39 (d) 0.00 (c) 0.00 (c) 37.82 MP2 0.00 0.00 8.54 −0.01 (c) 0.00 (c) 8.54 (d) 8.54 (d) 35.08 (e) 0.00 (c) 35.08 CCSD 0.00 0.00 7.58 0.00 (c) 0.00 (c) 7.58 (d) 7.58 (d) 35.09 (e) 0.00 (c) 35.09 Energies are in kcal/mol. (a) Experimental value using the lowest J energy. (b) Experimental value derived from the average of spin-orbit coupling. (c) Collapsed to σδ. (d) Collapsed to ππ. (e) Collapsed to δδ. Hund’s rules that the 3 F state is the ground state. However, sometimes it is useful to know the energy given the two electrons or holes in specific, real orbitals (e.g., dz2 or dxy ). In this case we must calculate the energies of intermediate states, i.e. between that of 3 F and 3 P . B.6.1 Background The d orbital electrons, which have ` = 2, can be represented by either the complex spherical harmonics Y2m or the real spherical harmonics Z2m . Below is a list of the complex spherical harmonics: r 15 (x − iy)2 32π r 15 2 r Y21 = z(x − iy) 8π r 5 r2 Y20 = (2z 2 − x2 − y 2 ) 16π r 15 2 r Y21 = z(x + iy) 8π r 15 (x + iy)2 r2 Y22 = 32π 2 r Y22 = 202 And of the real spherical harmonics: r 15 xy 4π r 15 r2 Z21 = yz 4π r 5 r2 Z20 = (2z 2 − x2 − y 2 ) 16π r 15 2 r Z21 = xz 4π r 15 2 r2 Z22 = (x − y 2 ) 16π 2 r Z22 = It is easy to see that the complex and the real spherical harmonics are closely related: 1 Y22 = √ (Z22 − iZ22 ) 2 1 Y21 = √ (Z21 − iZ21 ) 2 1 Y21 = √ (Z21 + iZ21 ) 2 1 Y22 = √ (Z22 + iZ22 ) 2 Y20 = Z20 1 Z22 = √ (Y22 − Y22 ) 2i 1 Z21 = √ (Y21 − Y21 ) 2i 1 Z21 = √ (Y21 + Y21 ) 2 1 Z22 = √ (Y22 + Y22 ) 2 B.6.2 The d2 and d8 triplet states The triplet states are 3 F and 3 P . We shall now examine the complex spherical harmonic wavefunctions associated with the ML values of each state. To do so we will start with the highest possible ML value and successively apply the lowering operator L̂− : L̂− |`mi = p (` + m)(` − m + 1)|`m − 1i. 203 Also, we will use A to represent antisymmetric functions, e.g., A[U V ] = U V − V U . B.6.2.1 The 3 F state The 3 F state has ML values that range from −3 to 3. For ML = 3 the only possible configuration is A[Y22 Y21 ]. Similarly, for ML = 2 the only possible configuration is A[Y22 Y20 ]. For ML = 1, however, we note that L̂− can act on either Y22 or Y20 ; consequently we get a mix: A[2Y21 Y20 + which is normalized to r A[ √ 2 Y21 Y20 + 5 6Y22 Y21 ]; r 3 Y22 Y21 ]. 5 For ML = 0, applying L̂− once again yields r A[ 2 √ ( 6Y21 Y21 ) + 5 r 3 (2Y21 Y21 + 2Y22 Y22 )], 5 which simplifies and normalizes to 2 1 A[ √ Y21 Y21 + √ Y22 Y22 ]. 5 5 By symmetry arguments we can see that ML = −1 corresponds to r A[ 2 Y20 Y21 + 5 r 3 Y21 Y22 ]; 5 We also see that ML = −2 corresponds to A[Y20 Y22 ] and ML = −3 corresponds to A[Y22 Y22 ]. B.6.2.2 The 3 P state The 3 P state has ML values that range from −1 to 1. For ML = 1 the wavefunction must be orthogonal to the 3 F ML = 1 case; it is easy to see that the wavefunction r A[ 3 Y21 Y20 − 5 r 2 Y22 Y21 ] 5 fulfills this condition. Likewise, for ML = 0 it is easy to see that 1 2 A[ √ Y21 Y21 − √ Y22 Y22 ] 5 5 is orthogonal to the 3 F ML = 0 configuration. 204 By symmetry we can add that r A[ 3 Y20 Y21 − 5 r 2 Y21 Y22 ] 5 corresponds to the 3 P ML = −1 case. B.6.3 Pure and mixed complex orbital configurations In the previous section, we saw that the complex orbital configurations A[Y22 Y21 ], A[Y22 Y20 ], A[Y20 Y22 ], and A[Y21 Y22 ] are pure 3 F states, while the other combinations are mixed. In this section we shall deconvolute the mixed states. B.6.3.1 ML = 1 We have r 3 Ψ1F = A[ and 3 r Ψ1P = A[ 2 Y21 Y20 + 5 r 3 Y21 Y20 − 5 r Hence and r A[Y22 Y21 ] = B.6.3.2 2 Y22 Y21 ]; 5 2 3F Ψ + 5 1 r 3 3F Ψ − 5 1 r r A[Y21 Y20 ] = 3 Y22 Y21 ] 5 3 3P Ψ 5 1 2 3P Ψ . 5 1 ML = 0 We have 3 2 1 Ψ0F = A[ √ Y21 Y21 + √ Y22 Y22 ] 5 5 and 3 1 2 Ψ0P = A[ √ Y21 Y21 − √ Y22 Y22 ]; 5 5 Hence 2 3 1 3 A[Y21 Y21 ] = √ Ψ0F + √ Ψ0P 5 5 and 1 3 2 3 A[Y22 Y22 ] = √ Ψ0F − √ Ψ0P . 5 5 205 B.6.3.3 ML = −1 From symmetry with the ML = 1 case we have 2 3F Ψ + 5 −1 r 3 3F Ψ − 5 −1 r r A[Y20 Y21 ] = and r A[Y21 Y22 ] = B.6.4 3 3P Ψ 5 −1 2 3P Ψ . 5 −1 Real orbital configuration energies We are now ready to calculate the energies of the configuration of electrons or holes in real orbitals. We should first note that the dz2 orbital (Z20 ) has σ symmetry, the dxz and dyz orbitals (Z21 and Z21 ) have π symmetry, and the dx2 −y2 and dxy orbitals (Z22 and Z22 ) have δ symmetry. Also, let us define  = E(3 P ) − E(3 F ) as the excitation energy to 3 P . B.6.4.1 Energy of σπ and σπ Y20 σπ ≡ A[Z21 Z20 ] = A[ √ (Y21 + Y21 )] 2 1 (E(Y21 Y20 ) + E(Y21 Y20 )) 2 3 2 3 1 2 = [( E(3 F ) + E(3 P )) + ( E(3 F ) + E(3 P ))] 2 5 5 5 5 2 3 3 3 3 3 = E( F ) + E( P ) = E( F ) +  5 5 5 E(Z21 Z20 ) = A very similar calculation for the σπ case yields the same energy, 53  above the ground state. B.6.4.2 Energy of σδ and σδ Y20 σδ ≡ A[Z22 Z20 ] = A[ √ (Y22 + Y22 )] 2 E(Z22 Z20 ) = 1 1 (E(Y22 Y20 ) + E(Y22 Y20 )) = [E(3 F ) + E(3 F )] = E(3 F ) 2 2 A very similar calculation for the σδ case yields the same energy, i.e. the energy of the ground state. B.6.4.3 Energy of ππ ππ ≡ A[Z21 Z21 ] = A[ E(Z21 Z21 ) = 1 1 (Y21 + Y21 )(Y21 − Y21 )] = √ (Y21 Y21 − Y21 Y21 ) 2i 2i 1 4 1 1 (E(Y21 Y21 ) + E(Y21 Y21 )) = E(Y21 Y21 ) = E(3 F ) + E(3 P ) = E(3 F ) +  2 5 5 5 206 B.6.4.4 Energy of πδ, πδ, πδ, and πδ 1 1 πδ ≡ A[Z22 Z21 ] = A[ (Y22 + Y22 )(Y21 + Y21 )] = A[ (Y22 Y21 + Y22 Y21 + Y22 Y21 + Y22 Y21 )] 2 2 1 [E(Y22 Y21 ) + E(Y22 Y21 ) + E(Y22 Y21 ) + E(Y22 Y21 )] 4 1 3 2 3 2 = [E(3 F ) + ( E(3 F ) + E(3 P )) + ( E(3 F ) + E(3 P )) + E(3 F )] 4 5 5 5 5 1 3 1 4 3 = E( F ) + E( P ) = E(3 F ) +  5 5 5 E(Z22 Z21 ) = Very similar calculations for the πδ, πδ, and πδ cases yield the same energy, i.e. 51  above the ground state. B.6.4.5 Energy of δδ δδ ≡ A[Z22 Z22 ] = A[ E(Z22 Z22 ) = B.6.5 1 1 (Y22 + Y22 )(Y22 − Y22 )] = √ (Y22 Y22 − Y22 Y22 ) 2i 2i 1 1 4 4 (E(Y22 Y22 ) + E(Y22 Y22 )) = E(Y22 Y22 ) = E(3 F ) + E(3 P ) = E(3 F ) +  2 5 5 5 Summary A triplet d2 or d8 atomic configuration gives rise to a mix of 3 F and 3 P states. The σδ configurations are pure 3 F , the ππ and πδ configurations are one-fifth of the way towards 3 P , the σπ configurations are three-fifths of the way towards 3 P , and the δδ configuration is four-fifths of the way towards the 3 P energy level. Hence we see that the energy is the lowest when the electrons or holes are allowed to freely roam about the atom (the σδ case) and the highest when they are restricted to one plane (the δδ case). B.7 Conclusions For the neutral atoms, M06-2X gave the best DFT results for Pt and Ni, while B3LYP gave the best DFT results for Pd. This may be due to the contraction of the d functions of the basis set in Pd. Considering only electron-correlation methods, MP2 gave the best results for Pt and Pd, while CCSD gave the best results for Ni. This may be due to the use of an ECP on Pt and Pd. In comparing DFT versus electron-correlation methods, the electron-correlation method (CCSD or MP2) was better for Pd and Ni, whereas the DFT M06-2X was better than any electron-correlation method for Pt. For the singly-charged cation, B3LYP gave the best results for Pt and Ni whereas there was no method found that gave good results for Pd. The singly-charged cation in the d9 configuration should have a hole in one of five degenerate orbitals; this is not borne out by DFT calculations and the discrepancy is not physical. The use of 207 electron-correlation calculations removes this artifact. Similar issues are observed with the doublycharged cation in the d8 configuration. This problem is innate to DFT and its resolution is still an active area of research. 208 B.8 References [1] C. E. Moore, Atomic Energy Levels (Natl. Bur. Stand., Washington, D.C., 1952) Circular 467, Vols. 2 and 3, 1959. [2] R. L. Martin and P. Jeffrey Hay, J. Chem. Phys. 1981, 75, 4539–4545. [3] Eric R. Scerri, The periodic table: its story and its significance; Oxford University Press, 2007; pp. 239–240. [4] Jaguar, version 7.5; Schrödinger, LLC: New York, NY, 2007. [5] (a) Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. V. J. Comput. Chem. 1983, 4, 294–301; (b) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72, 650–654. [6] Q-Chem, version 4.0; Q-Chem, Inc: Pittsburgh, PA, 2010. [7] Frank Jensen, Introduction to Computational Chemistry, 2nd ed.; John Wiley and Sons, 2007; p. 260. [8] Wales, David J. Molecular Physics 2005, 103, 269–271. [9] Axel D. Becke, J. Chem. Phys. 2002, 117, 6935–6938. [10] Sergey N. Maximoff, Matthias Ernzerhof, and Gustavo E. Scuseria, J. Chem. Phys. 2004, 120, 2105–2109.  Iridium and rhodium analogues of the Shilov cycle catalyst; and the investigation and applications of the Reduction-Coupled Oxo Activation (ROA) mechanistic motif towards alkane upgrading Thesis by Ross Fu In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy California Institute of Technology Pasadena, California 2014 (Defended May 9th, 2014) ii c 2014 Ross Fu All Rights Reserved iii To Annie, Bruce, Mom, Dad, and Harold. iv Acknowledgments There are many people who have helped me throughout the course of my graduate studies, and my education in general. Therefore, I cannot guarantee that this list will be comprehensive. First and foremost I want to thank my lovely wife, Annie Tigranyan, who through good times and bad has been a steady and ongoing source of support, encouragement, and motivation; and my parents, Prof. Xiang-Dong Fu and Dr. Mingjie Jin, who have shaped me to be the person I am and who have been the role models for the person I aspire to be. I also want to mention my son Bruce, who has been a continuing source of inspiration for me to be the best father I can; and Toshi, who has been my faithful and unwavering companion, guarding my side for many hours in the office as I put together this dissertation. Finally, I want to thank my parents and parents-in-law, Ruzanna Parvizi and Ashot Tigranyan, for their invaluable support in helping me raise my young family; and in particular my mother-in-law, who has devoted countless hours and effort taking care of Bruce so that I could concentrate on my work. I am greatly indebted to my advisor, Prof. William Goddard, for accepting me into his research group despite my (at the time) purely experimental background, for his extensive help at the beginning of my computational training, and for his continuing mentorship and scientific insights. I also want to thank my previous advisor at Caltech, Prof. John Bercaw, for his mentorship in synthetic chemistry, for helping me initiate my graduate-level work in chemistry, and for his continued work on my thesis committee. In addition, I want to thank Prof. Theo Agapie for chairing my committee, and Prof. Thomas Miller for serving in my committee. Finally I also want to thank Dr. Jay Labinger and Prof. Robert Grubbs for serving in my previous committee during my candidacy exam. In the Goddard group, a.k.a. the Materials Simulation Center, I want to especially thank the director for catalysis, Dr. Robert “Smith” Nielsen, for his collaboration, teaching, advice, patience, and incredible depth of knowledge. I also want to thank my various collaborators, including Prof. Brent Gunnoe, Dr. George Fortman, Dr. Matthew O’Reilly, and Michael Webster-Gardiner at the University of Virginia, and Prof. Greg Fu, Dr. Nathan Schley, and Trixia Buscagan at Caltech. In addition, I want to thank many other group members for their advice, discussions, and friendship; including but not limited to Dr. Mu-Jeng Cheng, Dr. Wei-Guang Liu, Dr. José Mendoza-Cortés, Dr. Julius Su, Samantha Johnson, Yanchoi Lam, Fan Liu, and Caitlin Scott. v In the Bercaw group, I want to thank Dr. Jay Labinger for his invaluable advice in the preparation of my Organometallics publication. I also want to thank the help and advice of my fellow students and postdocs Dr. Steve Baldwin, Dr. George Chen, Prof. Nilay Hazari, Dr. Rachel Klet, Dr. Valerie Kristof, Dr. Taylor Lenton, Prof. Alex Miller, Prof. Ian Tonks, Prof. Dave Weinberg, Prof. Nathan West, Dr. Matt Winston, Yanchoi Lam, Josef Meier, and Tonia Ahmed. I also want to thank Larry Henling and Dr. Michael Day for help in X-ray crystallography, and Naseem Tourian and Dr. Mona Shahgholi for their help in mass spectrometry. Finally, I want to thank the many friends who have made my years here at Pasadena so enjoyable. In particular, Elena Bibikova, Jerry Koo and Fiona Chiang, Tracey Liu, Hemanth Siriki, Jerzy Szablowski, Chloe Wang, Arlene Santos and the Salsa Club and of course my dance partner Annie Tigranyan; and some of my oldest friends from high school, Michael Liao, Jason Lin, Erik Tsou, and David Wang. vi Abstract This dissertation will cover several disparate topics, with the overarching theme centering on the investigation of organometallic C−H activation and hydrocarbon transformation and upgrading. Chapters 2 and 3 discuss iridium and rhodium analogues of the Shilov cycle catalyst for methane to methanol oxidation, and Chapter 4 on the recently discovered ROA mechanistic motif in catalysts for various alkane partial oxidation reactions. In addition, Chapter 5 discusses the mechanism of nickel pyridine bisoxazoline Negishi catalysts for asymmetric and stereoconvergent C−C coupling, and the appendices discuss smaller projects on rhodium H/D exchange catalysts and DFT method benchmarking. Chapter 2: The iridium complex (ONO)Ir(PPh3 )2 Me, where (ONO) is a bis(phenolate)pyridine pincer ligand, undergoes C−H activation of benzene, related arenes, and acetonitrile. Labeling and kinetic studies indicate a unique and heretofore undescribed mechanism involve intramolecular C−H activation followed by intermolecular C−H activation. Chapter 3: The Shilov catalytic cycle for methane to methanol oxidation via a homogeneous Pt catalyst has long been a target of investigation, with work focusing on improving catalyst stability, yield, and robustness. A Rh analogue to the Shilov cycle was explored due to its increased stability at high oxidation states. A density functional theory-driven virtual screening of ligand F frameworks revealed that the fluorine-substituted amidinate (NN ) ligand provides the lowest transition state energies for methane C−H activation and methyl group functionalization, and stands most promising as a target for experimental investigation. Chapter 4: The vanadium phosphorus oxide (VPO) catalyst has been used in industry for the remarkably selective conversion of butane to maleic anhydride. Investigation of the reaction mechanism revealed the surprising conclusion that, although redox action occurs in the vanadium centers as expected, initial C−H activation takes place in the phosphorus oxide moieties. Further work has revealed that the same phenomenon can be extended to other transition metal/nonmetal oxo pairs. This phenomenon, which has been named the Reduction-Coupled Oxo Activation (ROA) mechanistic motif, is attributed to the concomitant formation of a strong oxo-hydrogen bond on the nonmetal and a one-electron reduction on the transition vii metal. Finally, a homogeneous vanadium phosphorus oxo complex is proposed to incorporate the novel ROA mechanistic pathway as a potential propane oxidation catalyst. Several oxidation pathways are proposed, leading to end products such as propylene, isopropanol, and propylene oxide. These assertions are supported with density functional theory calculations on the potential reaction pathways. Chapter 5: The in situ generated Ni(iPr−pybox) complex catalyzes enantioselectively the crosscoupling of secondary sp3 -C substrates. This system is very notable for its property of stereoconvergence: both enantiomers of a racemic substrate may be converted into a single enantiomer of a product. This chapter will computationally explore the mechanism responsible for both the remarkable activity and selectivity of the Ni((S, S)−iPr−pybox system; such a mechanism is believed to deviate substantially from the canonical “textbook” mechanism of Negishi coupling. Furthermore, a generalized calculation scheme is presented that allows for the rapid enantioselectivity prediction of many related pybox ligands as well, in order to help predict the next generation of asymmetric Negishi coupling catalysts. viii Contents Acknowledgments iv Abstract vi 1 Introduction 1 1.1 Iridium and Rhodium analogues of the Shilov cycle catalyst . . . . . . . . . . . . . . 1.2 Investigation and applications of the Reduction-Coupled Oxo Activation (ROA) mech- 1 anistic motif towards alkane upgrading. . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Other projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.4 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2 Intra- and intermolecular C−H bond activation by bis(phenolate)pyridineiridium(III) complexes 8 2.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Synthesis and Characterization of IrI complexes . . . . . . . . . . . . . . . . . 9 2.3.1 III 2.3.2 Synthesis and Characterization of Ir Complexes . . . . . . . . . . . . . . . 13 2.3.3 C−H Activation by IrIII (ONO ) Complexes . . . . . . . . . . . . . . . . . . 21 2.3.4 Kinetics of C−H Activation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 2.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.5 Supporting Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.5.1 Experimental Section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.5.2 Kinetics Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.5.3 Crystallographic Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.6 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 tBu ix 3 DFT virtual screening identifies rhodium-amidinate complexes as active homogeneous catalysts for methane to methanol oxidation 56 3.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 3.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 3.4.1 63 3.4.2 The Rh(NN) complexes in TFAH . . . . . . . . . . . . . . . . . . . . . . . . . F 67 F The Rh(NN ) complexes in TFAH . . . . . . . . . . . . . . . . . . . . . . . . 3.4.3 The Rh(NN ) complexes in water . . . . . . . . . . . . . . . . . . . . . . . . 67 3.4.4 The Rh(ONNx ) family of complexes in TFAH . . . . . . . . . . . . . . . . . . 73 3.4.5 The Rh(DPMS) and Rh(DPES) complexes in TFAH . . . . . . . . . . . . . . 78 3.4.6 The Rh(PN) complexes in TFAH . . . . . . . . . . . . . . . . . . . . . . . . . 78 x 3.4.7 The Rh(bisq ) family of complexes in TFAH . . . . . . . . . . . . . . . . . . 81 3.4.8 Product protection and C−H activation . . . . . . . . . . . . . . . . . . . . . 92 3.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 3.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 3.7 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 3.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4 The Reduction-Coupled Oxo Activation mechanistic motif for C−H bond activation and oxidation 110 4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 4.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 4.2.1 The VPO catalyst for n-butane oxidation . . . . . . . . . . . . . . . . . . . . 111 4.2.2 The ROA mechanistic motif . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 4.2.3 Potential homogeneous catalysts utilizing the ROA principle . . . . . . . . . 115 4.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 4.4 Validation of ROA on small oxide clusters . . . . . . . . . . . . . . . . . . . . . . . . 118 4.5 Reactions of a homogeneous oxidation catalyst utilizing the ROA mechanistic motif 128 4.5.1 Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 4.5.2 The resting VV state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 4.5.3 Propane activation transition states . . . . . . . . . . . . . . . . . . . . . . . 132 4.5.4 Monohydrogenated VIV states . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 4.5.5 Catalytic cycle involving H-atom abstraction only . . . . . . . . . . . . . . . 136 4.5.6 Reduced VIII states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 4.5.7 Catalytic cycle involving H-atom abstraction followed by isopropyl addition . 143 x 4.5.8 Comparison of the catalytic cycles . . . . . . . . . . . . . . . . . . . . . . . . 144 4.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 4.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 5 The mechanism of Ni(iPr−pybox) complexes as asymmetric Negishi C−C coupling catalysts 150 5.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 5.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 5.2.1 The Ni(iPr−pybox) system . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 5.2.2 Existing computational work . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 5.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 5.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 5.5 5.4.1 The overall general mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . 157 5.4.2 Nature of the organozinc reagent . . . . . . . . . . . . . . . . . . . . . . . . . 158 5.4.3 Mechanism for NiI oxidation and methylation . . . . . . . . . . . . . . . . . . 161 5.4.4 Mechanism for reductive elimination . . . . . . . . . . . . . . . . . . . . . . . 166 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 5.5.1 Overall catalytic cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 5.5.2 Predictions for new ligands . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 5.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 5.7 Supporting figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 5.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 A C−H activation using RhI complexes supported by bis-nitrogen chelating ligands180 A.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 A.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 A.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 A.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 A.4.1 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 A.4.2 Computational results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 A.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 A.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 B Energy levels of group 10 transition metal atoms and ions 189 B.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 B.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 B.3 The neutral atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 xi B.3.1 Platinum (0) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 B.3.2 Palladium (0) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 B.3.3 Nickel (0) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 B.4 The singly-charged cation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 B.5 Angular momentum-dependent artifacts in DFT calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 B.5.1 Angular momentum of the electron/hole in the d1 /d9 configuration of the cation196 B.5.2 Angular momentum of the holes in the d8 configuration of the dication . . . . 199 B.6 Energies of d2 and d8 triplet states . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 B.6.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 B.6.2 The d2 and d8 triplet states . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 B.6.3 Pure and mixed complex orbital configurations . . . . . . . . . . . . . . . . . 204 B.6.4 Real orbital configuration energies . . . . . . . . . . . . . . . . . . . . . . . . 205 B.6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 B.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 B.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 xii List of Figures 1.1 C−H activation by a phosphorus oxo leads to a one electron reduction on the neighboring vanadium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.1 The ONO and ONOtBu ligands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2 Crystal structure of pyridine-activated complex 2-3. . . . . . . . . . . . . . . . . . . . 11 2.3 2.4 2.5 2.6 2.7 tBu Crystal structure of [NaIr(cod)(ONO tBu Crystal structure of [Ir(cod)]2 (ONO )]2 , 2-4. . . . . . . . . . . . . . . . . . . . . . 13 ), 2-5. . . . . . . . . . . . . . . . . . . . . . . . 14 Crystal structure of [(ONO tBu )Ir(cod)(MeCN)][PF6 ], 2-6. . . . . . . . . . . . . . . . . 14 Crystal structure of [(ONO tBu )Ir(cod)OTf, 2-7. . . . . . . . . . . . . . . . . . . . . . . 15 tBu )Ir(PPh3 )2 OTf, 2-8. . . . . . . . . . . . . . . . . . 16 tBu )Ir(PPh3 )2 OH, 2-9. . . . . . . . . . . . . . . . . . 17 Crystal structure of trans-(ONO 2.8 Crystal structure of trans-(ONO 2.9 Crystal structure of (ONO tBu )Ir(PPh3 )2 H, 2-11. . . . . . . . . . . . . . . . . . . . . . 17 2.10 Crystal structure of (ONO tBu )Ir(PPh3 )2 Me, 2-12. . . . . . . . . . . . . . . . . . . . . 19 2.11 Crystal structure of (ONO tBu )Ir(PPh3 )2 I, 2-13. . . . . . . . . . . . . . . . . . . . . . 19 2.12 Crystal structure of (ONO tBu )Ir(PEt3 )2 Me, 2-14. . . . . . . . . . . . . . . . . . . . . 20 2.13 Crystal structure of (ONO tBu )Ir(PEt3 )2 I, 2-15. . . . . . . . . . . . . . . . . . . . . . . 21 2.14 Cyclic voltammetry of 2-14. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.15 EPR spectrum of [(ONO )Ir(PEt3 )2 Me]PF6 , 2-16. . . . . . . . . . . . . . . . . . . . 22 2.16 Crystal structure of [(ONO tBu )Ir(PEt3 )2 Me]PF6 , 2-16. . . . . . . . . . . . . . . . . . 23 2.17 Crystal structure of (ONO tBu )Ir(PPh3 )2 Ph, 2-17. . . . . . . . . . . . . . . . . . . . . 24 2.18 Crystal structure of (ONO tBu 2.19 Crystal structure of [(ONO )Ir(PPh3 )(µ−CH2 CN)]2 , 2-19. . . . . . . . . . . . . . . 26 2.20 Kinetics of conversion of 2-12 to 2-17 and their deuterated isotopologues. . . . . . . . 30 2.21 Dependence of the reciprocal of the rate of conversion of 2-12 to 2-17 on the amount tBu 2 )Ir(PPh3 )[κ -PPh2 (o-C6 H4 )], 2-18. . . . . . . . . . . . . tBu 24 of added triphenylphosphine. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.22 Kinetics for conversion of 2-17 to 2-18 in C6 D6 and p-xylene-d10 . . . . . . . . . . . . 31 2.23 Conversion of 2-12 to 2-17-d6 in C6 D6 with varying amounts of extra PPh3 added. . 47 3.1 The rhodium-ligand complexes screened. . . . . . . . . . . . . . . . . . . . . . . . . . . 63 xiii 3.2 Rh(NN) transition state structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Rh(NN ) transition state structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 3.5 F F Rh(NN ) in water transition state structures. . . . . . . . . . . . . . . . . . . . . . . . 87 88 The various Rh (bisq )(Me) III-II SR 2 functionalization transition states. . . . . . . 3.7 Graph of the activation and functionalization transition state energies of the various Rh(bisqx ) complexes at 298 K and 498 K. . . . . . . . . . . . . . . . . . . . . . . . . . 91 Transition states for the functionalization of methane, Me−TFA, and methanol using Rh(NN) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9 72 x 3.6 3.8 69 x The various RhIII (bisq )(Me) III-I SN 2 functionalization transition states. . . . . . . . III 66 93 Transition states for the functionalization of methane, Me−TFA, and methanol using F Rh(NN ) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 3.10 Summary of the data presented in Figures 3.8, 3.9, and 3.11. . . . . . . . . . . . . . . 95 3.11 Comparison of activation energies for methane and for methanol in the same structure F for Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12 Comparison of the activation states of methane and methyl trifluoroacetate by Rh(DPMS) complexes, showing the product protection afforded. . . . . . . . . . . . . . . 3.13 98 Comparison of the activation states of methane and methyl trifluoroacetate by the x Rh(bisq ) family of complexes, showing the product protection afforded. . . . . . . . . 3.15 97 Comparison of the activation states of methane and methyl trifluoroacetate by Rh(DPES) complexes, showing the product protection afforded. . . . . . . . . . . . . . . . . . . . 3.14 96 99 Graph of the methane activation energy of the various rhodium complexes vs. the 4s orbital energy of the rhodium in their resting states. . . . . . . . . . . . . . . . . . . . 101 3.16 Graph of the methane activation energy of the various rhodium complexes vs. the Rh−C bond energy of the lowest-energy Rh−Me complex. . . . . . . . . . . . . . . . . 103 3.17 A graph of the Rh−TFA to Rh−Me change in the 4s orbital energy, a measure of the Rh atom’s softness, versus the adjusted TS barrier. . . . . . . . . . . . . . . . . . . . . 104 3.18 The calculated free energy of the (NNC)Ir(TFA)2 system, via the same internal substitution mechanism applied to RhIII complexes. . . . . . . . . . . . . . . . . . . . . . 105 IV 4.1 O−H bond strength for oxygen atoms on the (V O)2 (P2 O7 ) surface. . . . . . . . . . 111 4.2 Oxygen adsorption energies on the (V O)2 (P2 O7 ) surface and the O−H bond strength IV (DH ) to each of these O atoms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 4.3 O−H bond strengths for the various oxygen atoms on the X1-VV OPO4 surface. 4.4 C−H activation by a phosphorus oxo leads to a one electron reduction on the neigh- . . . 113 boring vanadium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 xiv 4.5 Use of finite cluster models to investigate how the O−H bond strength of P−O moieties varies based on the number of V coupled to it via O linkages. . . . . . . . . . . . . . . 114 4.6 Known complexes featuring a phosphinite ligand bound κ−P to a transition metal, and a proposed vanadium κ−P phosphinite complex. . . . . . . . . . . . . . . . . . . . . . 116 4.7 General models for small oxide clusters. . . . . . . . . . . . . . . . . . . . . . . . . . . 118 4.8 Optimized structures of the Px V4−x O10 species (x ∈ {0, 1, 2, 3}) and their monohydrogenated equivalents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 4.9 Optimized structures of the dihydrogenated P2 V2 O10 H2 and trihydrogenated P2 V2 O10 H3 species. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 4.10 Optimized structures of the XV3 O10 species (X ∈ {P, As, Sb, Bi}) and their monohydrogenated XV3 O10 H counterparts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 4.11 Optimized structures of the Z2 V2 O11 species (Z ∈ {S, Se, Te} and their monohydrogenated Z2 V2 O11 H counterparts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 4.12 Optimized structures of the hydrated species Te2 V2 O12 H2 and Te2 V2 O12 H3 . . . . . . 122 4.13 Optimized structures of the fully hydrated species Te2 V2 O14 H6 and Te2 V2 O14 H7 . . . 123 4.14 Optimized structures of Se2 V2 O9 and Te2 V2 O9 , as well as their monohydrogenated counterparts Se2 V2 O9 H and Te2 V2 O9 H. . . . . . . . . . . . . . . . . . . . . . . . . . . 124 4.15 Optimized structures of the group 14-vanadium oxide species Z2 V2 O10 H2 , where Z ∈ {Si, Ge, Sn, Pb}; as well as that of their corresponding monohydrogenated species. . . 126 4.16 Optimized structures of the molybdenum-containing species P2 Mo2 O11 , Te2 Mo2 O10 , and Te2 Mo2 O12 ; as well as that of their corresponding monohydrogenated species. . . 127 4.17 Template for the nomenclature of investigated (OPO)V complexes. . . . . . . . . . . . 129 4.18 The geometry-optimized structures of (OPO)fH V Oc (OH)s (H2 O)t , (OPO)f V Oc (H2 O)st 2 , V V V 2– and [(OPO)f V Oc (OH)st 2 ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 V 4.19 ‡ The geometry-optimized structure of [(OPO)fHiPr V Oc (H2 O)st 2 ] . . . . . . . . . . . . . 135 4.20 f c st The geometry-optimized structures of (OPO)fH V Oc (H2 O)st 2 , (OPO) V (OH) (H2 O)2 , IV IV IV – and [(OPO)f V Os (H2 O)ct 2 ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 4.21 III III 3 f cst The geometry-optimized structures of 3 (OPO)f V (HOiPr)s (H2 O)ct 2 , (OPO) V (H2 O)3 , and 3 (OPO)H VIII (OH)s (H2 O)a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 5.1 The computationally studied reaction of the Ni(iPr−pybox) Negishi coupling catalyst as reported by Lin et al. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 5.2 The various (MeZnI)m (dma)n species, where m, n ∈ {0, 1, 2}. . . . . . . . . . . . . . . 159 5.3 The various (ZnBrI)m (dma)n species, where m, n ∈ {0, 1, 2}. . . . . . . . . . . . . . . 160 5.4 Transition states for the reductive elimination of 1-methylindane from various NiIII ((S, S)−iPr−pybox)(Me)ind species. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 xv 5.5 Reaction coordinate diagram of the entire catalytic cycle for the coupling of ind−Br and MeZnI to form ind−Me. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 5.6 Modified ligands which may allow for different enantioselective properties as compared to the original (iPr−pybox) ligand. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 A.1 Temperature dependence vs. turnover numbers for complexes 1 and 2. . . . . . . . . . 182 A.2 Concentration of Rh dependence vs. TON for complexes 1 and 2. . . . . . . . . . . . 183 xvi List of Schemes 1.1 The original PtII /PtIV cycle of the Shilov catalyst, as well as IrIII /IrV and RhI /RhIII analogues. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 The oxidation of n-butane to maleic anhydride using vanadium phosphorus oxide as a catalyst. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 2.1 2.2 2.3 2 3 The generalized mechanism of the Ni((iPr)−pybox) Negishi coupling catalyst as it proceeds through a NiI -NiIII cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Synthesis of Na[Ir(cod)(ONO)], 2-1. . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 tBu Synthesis of the ligand H2 (ONO ). . . . . . . . . . . . . . . . . . . . . . . . . . . tBu Synthesis and reactivity of [NaIr(cod)(ONO )]2 , 2-4. . . . . . . . . . . . . . . . . . 12 III species. . . . . . . . . . . . . . . . . 13 )Ir(PPh3 )2 H, 2-11. . . . . . . . . . . . . . . . . . . . . 18 2.4 Oxidation of [NaIr(cod)(ONO 2.5 Synthetic routes to (ONO 2.6 tBu Methylation to (ONO I2 to (ONO tBu tBu tBu )]2 , 2-4, to Ir )Ir(PPh3 )2 Me, 2-12, and subsequent functionalization with )Ir(PPh3 )2 I, 2-13. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . tBu 2.7 Synthesis of (ONO 2.8 Proposed mechanism for the transformation of 2-12 in benzene to the kinetic product tBu (ONO 12 Ir(PEt3 )2 complexes. . . . . . . . . . . . . . . . . . . . . . . . tBu )Ir(PPh3 )2 Ph, 2-17 and the thermodynamic product (ONO 18 20 )Ir(PPh3 )- [κ2 -PPh2 (o-C6 H4 )], 2-18. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.9 Simplified version of Scheme 2.8, used for kinetic analysis. . . . . . . . . . . . . . . 29 3.1 Hypothetical catalytic cycle for the activation and functionalization of methane. . . 62 3.2 Catalytic cycle for the activation and functionalization of methane using Rh(NN) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 F Catalytic cycle for the activation and functionalization of methane using Rh(NN ) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 68 F Catalytic cycle for the activation and functionalization of methane using Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 64 F Transition states for the activation of methane using Rh(NN ) complexes in water. . 70 71 xvii 3.6 F Transition states for the III-I functionalization pathway using Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 73 F Transition states for the III-II functionalization pathway using Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8 74 F Transition states for the III-IV-II functionalization pathway using Rh(NN ) complexes in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9 74 Catalytic cycle for the activation and functionalization of methane using Rh(ONN) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 F 3.10 Catalytic cycle for the activation and functionalization of methane using Rh(ONN ) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11 Catalytic cycle for the activation and functionalization of methane using Rh(ONN NMe2 76 ) complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.12 Catalytic cycle for the activation and functionalization of methane using Rh(DPMS) complexes in TFAH, via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. . . . 79 3.13 Catalytic cycle for the activation and functionalization of methane using Rh(DPES) complexes in TFAH, via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. . . . 80 3.14 Catalytic cycle for the activation and functionalization of methane using Rh(DPMS) complexes in TFAH, via the Rh(I-III) pathway. . . . . . . . . . . . . . . . . . . . . . 81 3.15 Catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, via the Rh(III-I) and Rh(III-II) pathways. . . . . . . . . . . . . 82 3.16 Catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, via the Rh(I-III) pathway. . . . . . . . . . . . . . . . . . . . . . x III 83 x 3.17 The inorganic RhIII (bisq )(TFA)3 species and its protonated analogues [Rh (bisq )+ III x + (TFA)2 (TFAH)] and [Rh (bisq )(TFA)2 )] . . . . . . . . . . . . . . . . . . . . . . III x +‡ 3.18 The various methane activation transition states [Rh (bisq )(TFA)(Me··H··TFA)] III 84 . 85 x 3.19 Comparison of the functionalization of Rh (bisq )(Meax )(TFA)2 via the III-I SN 2 and III-II SR 2 pathways and the completion of the catalytic cycle. . . . . . . . . . . 89 x 3.20 Catalytic cycle for the activation and functionalization of methane using Rh(bisq ) complexes in TFAH, via the Rh(I-III) pathway. . . . . . . . . . . . . . . . . . . . . . 90 3.21 A schematic diagram showing the metal activation of an R−H bond, for example, methane. 4.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 The oxidation of n-butane to maleic anhydride using vanadium phosphorus oxide as a catalyst. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 4.2 Potential catalytic cycle involving H-atom abstraction only. . . . . . . . . . . . . . . 138 4.3 Potential catalytic cycle assuming isopropyl addition and VIII formation. . . . . . . . 145 xviii 5.1 The generalized cross-coupling reaction and its mechanism. . . . . . . . . . . . . . . 151 5.2 The general reaction of the Ni(iPr−pybox) Negishi coupling catalyst. . . . . . . . . . 152 5.3 The generalized mechanism of the Ni(iPr−pybox) Negishi coupling catalyst as it proceeds through a NiI -NiIII cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 5.4 Complexation of the various MeZnX and ZnX2 species with X– , where X ∈ {Br, I}. . 159 5.5 Three potential pathways for the methylation and oxidation of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br. . . . . . . . . . . . . . . . . . . . . . . . . . 162 5.6 The dissociative pathway for the methylation of NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)−iPr−pybox)Me. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162 5.7 NiI ((S, S)−iPr−pybox)Br by ind−Br to NiII ((S, S)−iPr−pybox)Br2 and ind · . . . . 163 5.8 NiII ((S, S)−iPr−pybox)Br2 and its subsequent methylation or recombination with ind · as part of the NiII or NiIII pathways. . . . . . . . . . . . . . . . . . . . . . . . . 165 5.9 The methylation of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) to produce NiIII ((S, S)−iPrpybox)(Br)(ind)Me. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 5.10 A generalized catalytic mechanism that Ni(R−pybox) complexes undergo in Negishi coupling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 5.11 Expanded diagram of NiII ((S, S)−iPr−pybox)Br2 methylation as part of the NiII pathway. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 5.12 Expanded diagram of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) methylation as part of the NiIII pathway. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 5.13 The reactions of the various NiIII ((S, S)−iPr−pybox)(Me)ind species that lead to the reductive elimination of ind−Me. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 F A.1 Synthesis of rhodium H/D exchange catalysts (DAB )Rh(COE)(TFA) A-1 and (BOZO)Rh(COE)(TFA) A-2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 F A.2 Oxidative addition of benzene by Rh(DAB ) and Rh(BOZO) complexes A-3 and A-4. 184 F A.3 Various possible pathways for the H/D exchange of benzene catalyzed by a Rh(DAB ) complex. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 A.4 Various possible pathways for the H/D exchange of benzene catalyzed by a Rh(BOZO) complex. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186 F A.5 Various possible pathways for the H/D exchange of methane catalyzed by a Rh(DAB ) complex. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 xix List of Tables 2.1 Summary of the results of Figure 2.23, used to create Figure 2.21. . . . . . . . . . . . 46 2.2 Crystallographic data for the complexes 2-4, 2-5, 2-6, and 2-7. . . . . . . . . . . . . . 48 2.3 Crystallographic data for the complexes 2-8, 2-9, 2-10, and 2-11. . . . . . . . . . . . . 49 2.4 Crystallographic data for the complexes 2-12, 2-13, 2-14, and 2-15. . . . . . . . . . . 50 2.5 Crystallographic data for the complexes 2-16, 2-17, 2-18, and 2-19. . . . . . . . . . . 51 tBu 2.6 Crystallographic data for the complex (ONO )Ir(PEt3 )2 Cl. . . . . . . . . . . . . . . 52 3.1 Antoine equation parameters for trifluoroacetic acid and water. . . . . . . . . . . . . . 60 3.2 Lowest activation and functionalization energies for each series of Rh-ligand complexes in TFAH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 65 Comparison of the various Rh parameters with the rhodium-ligands complexes’ associated transition state barriers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 V 4.1 Comprehensive compilation of all (OPO)V species studied. . . . . . . . . . . . . . . 132 4.2 Comprehensive compilation of all the transition states for propane activation by (OPO)VV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 IV species studied. . . . . . . . . . . . . . . 137 IV species studied. . . . . . . . . . . . . . . 142 IV species studied. . . . . . . . . . . . . . . 143 4.3 Comprehensive compilation of all (OPO)V 4.4 Comprehensive compilation of all (OPO)V 4.5 Comprehensive compilation of all (OPO)V 5.1 Examples of various C−C cross-coupling reactions. . . . . . . . . . . . . . . . . . . . . 152 5.2 Some examples of substrates coupled by the Ni(iPr−pybox) catalyst. . . . . . . . . . 153 5.3 Comparison of activation energies for various Ni(R−pybox) catalysts. . . . . . . . . . 171 5.4 Relative energies of the neutral A, D, and G species for various Ni(R−pybox) complexes.176 5.5 Relative energies of the cationic B, E, and H species for various Ni(R−pybox) complexes.176 5.6 Relative energies of the cationic C, F, and J species for various Ni(R−pybox) complexes.177 B.1 Relativistic vs. nonrelativistic Hartree-Fock excitation energies . . . . . . . . . . . . . 190 B.2 Relativistic vs. nonrelativistic Hartree-Fock excitation energies for the M+ ions. B.3 Excitation energies of Pt atom as calculated by various DFTs. . . . . . . . . . . . . . 191 . . . 191 xx B.4 Excitation energies of Pt atom calculated without f functions. . . . . . . . . . . . . . 192 B.5 Excitation energies of Pt atom calculated using electron-correlation methods. . . . . . 193 B.6 d10 − d9 s1 Excitation energies of Pd atom as calculated by Jaguar. . . . . . . . . . . . 194 B.7 Excitation energies of Pd atom as calculated by Q-Chem. . . . . . . . . . . . . . . . . 194 B.8 Excitation energies of Ni atom as calculated by Q-Chem; electron-correlation methods. 194 B.9 Excitation energies of Ni atom as calculated by Q-Chem; DFT methods. . . . . . . . . 195 B.10 Excitation energies of the group 10 atoms from d9 to d8 s1 , as calculated using various DFTs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 B.11 Excitation energies of the group 10 atoms from d9 to d8 s1 , as calculated using various electron-correlation methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 B.12 “Degenerate” energies of the Pt+ d9 ground state. . . . . . . . . . . . . . . . . . . . . 197 B.13 “Degenerate” energies of the Pt9+ d1 ground state. . . . . . . . . . . . . . . . . . . . . 197 B.14 “Degenerate” energies of the Ni+ d9 ground state as calculated by DFT methods. . . 198 B.15 Degenerate energies of the Ni+ d9 ground state as calculated by electron-correlation methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198 B.16 Energies of the Pt2+ d8 configuration, calculated by B3LYP/LACV3P**++. . . . . . 199 B.17 Energies of the Pd2+ d8 configuration, calculated by B3LYP/LACVP. . . . . . . . . . 200 B.18 Energies of the Ni2+ d8 configuration, calculated on Jaguar using the 6-31G(tm)**++ basis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 B.19 Energies of the Ni2+ d8 configuration, experimental and calculated on Q-Chem using the G3LARGE basis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 1 Chapter 1 Introduction 1.1 Iridium and Rhodium analogues of the Shilov cycle catalyst The facile, selective, and direct conversion of methane into methanol has long been a goal of industrial chemists [1]. Methane, the chief component of natural gas, is typically found in abundant quantities wherever petroleum deposits are found. However, due to its difficulty in liquefaction (its boiling point being only 110 K) and its relative paucity of demand as compared to other hydrocarbons, it is often not economical to harvest this gas, which is essentially a byproduct or co-product of petroleum recovery, and to transport it to market. Since methane is also a potent greenhouse gas, with a global warming potential 72 times more powerful than that of carbon dioxide [2], it cannot simply be released either. Hence it is often flared, a process that is both intrinsically wasteful and which results in a large amount of heat pollution. One potential solution is to convert these vast quantities of methane into methanol. As methanol is liquid at room temperature and pressure, it is much more easily transported and stored. Methanol can be used as a fuel in flex fuel vehicles, blended with gasoline, or converted to gasoline or diesel fuel. In addition, it has many applications as an industrial solvent and chemical feedstock. As methanol is both more easily processed and in greater demand than methane, the efficient conversion from the former to the latter has the potential to be both environmentally friendly and economically favorable. Although the reaction CH4 + 21 O2 −−→ CH3 OH is exoenergetic by 30 kcal/mol, efficient catalysis is hampered by the fact that the C−H bond dissociation energy (BDE) of methane is 104 kcal/mol, whereas the C−H BDE of methanol is only 95 kcal/mol. Hence, overoxidation is a major problem. In fact, potential methods that involve radicals have a theoretical maximum yield of only 5%, even with perfect conversion [3]. Hence, the currently favored industry-scale conversion of methane to methanol involves an indirect pathway via syngas [1]: 2 Ni CH4 + H2 O −→ CO + 3 H2 (1.1) CO + 2 H2 −−→ CH3 OH (1.2) However, this pathway suffers from a difficult first step that is highly endergonic, requiring very high temperatures and pressures (700-1100◦ C, 40 atm). Hence, this state-of-the-art pathway requires large inputs of energy and very expensive production facilities despite the significantly exothermic nature of the overall reaction. An alternative mechanism that does not require such extreme conditions would thus be highly desirable. Molecular compounds in solution have long been investigated as potential catalysts for this transformation, as they are typically well-defined, easy to characterize and model, and operate under relatively mild conditions. Much of the focus has been on routes involving C−H activation at a transition metal center followed by oxidative cleavage of the resulting C−M bond, which offer greater possibilities for good selectivity by avoiding radical-based pathways. Platinum complexes 2– have received a great deal of attention [4], starting with Shilov’s seminal work on the PtCl2– 4 /PtCl6 system, which can oxidize methane to methanol with some selectivity [5], and including the platinum bipyrimidine system, which has achieved the most impressive performance to date, converting methane to methyl bisulfate in up to 70% yield [6]. However, none of this platinum-based chemistry has yet been shown to lead to a practical process, owing in part to low reactivity, and a good deal of research activity has turned to other metals. Chapters 2 and 3 will address certain Ir and Rh analogues, respectively. The original Pt Shilov cycle, as well as modifications incorporating Ir and Rh, is given in Scheme 1.1. Scheme 1.1. Left: The original PtII /PtIV cycle of the Shilov catalyst. Center: A potential analogous IrIII /IrV cycle. Right: A potential analogous RhI /RhIII cycle. Chapter 2 focuses on bis(phenolate)pyridineiridium(III) ((ONO)Ir) complexes as an Ir analogue to the Shilov catalyst. Iridium as an alkane oxidation catalyst has been a popular choice; many 3 (primarily low-valent) Ir complexes have been reported to show good activity for C−H activation [7], and examples of oxidative functionalization with Ir systems are known as well [8]. Interconversions between IrIII and IrV might be important in this chemistry, by analogy to the Shilov system, which involves PtII and PtIV . In Chapter 2, the experimental synthesis and characterization of a multitude of ((ONO)Ir) complexes in the +III and +IV oxidation state are described. In particular, the complex (ONO)Ir(PPh3 )2 Me has been found to undergo oxidation with iodine to release MeI, analogous to the Shilov system’s production of MeCl. Furthermore, (ONO)Ir(PPh3 )2 Me has been found to undergo an unusual C−H activation route with benzene that initially proceeds by intramolecular C−H activation, followed by intermolecular C−H activation. Chapter 3 focuses on rhodium due to its well-documented nature as an effective C−H activating metal [9] and whose lower electronegativity may allow it to avoid poisoning by coordinating media. In this chapter, a computational screening of a variety of ligands was undertaken in order to find a Rh-ligand complex with the predicted optimal combination of low methane activation energy, low methyl group functionalization energy, and overoxidation protection of the MeX product. Our best F results were the RhIII bis(N -pentafluorophenyl)pentafluorobenzylamidinate (RhIII (NN )) complex and the RhIII bis(quinolinyl)benzene (RhIII (bisq)) complex. In addition, we also report a correlation between Rh−Me bond energies and methane activation barriers that would allow us to easily predict the suitability of new complexes for methane activation. 1.2 Investigation and applications of the Reduction-Coupled Oxo Activation (ROA) mechanistic motif towards alkane upgrading. Vanadium phosphorus oxide (VPO) is an inorganic complex that catalyzes the oxidation of n-butane to maleic anhydride with a surprisingly high selectivity of 60-70% [10] (Scheme 1.2). Although the overall yield is only ca. 50%, its ease of synthesis, use of common elements, and selectivity have allowed this catalyst to be commercialized, producing ca. 500 kilotons of maleic anhydride annually [11]. Scheme 1.2. The oxidation of n-butane to maleic anhydride using vanadium phosphorus oxide as a catalyst. Surprisingly, the mechanism has been shown to initiate with C−H activation by a P−O moiety [12]. Whereas the many vanadium oxos and peroxos investigated bound with hydrogen with a 4 maximum energy of only 70.0 kcal/mol, too low for initial C−H activation of n-butane to occur, the phosphorus oxo has a much larger bond strength with hydrogen, at 84.3 kcal/mol. This implies a first step that is endothermic by only 5.0 kcal/mol, with an activation energy later calculated to be 13.6 kcal/mol, which is within the experimental range. The stability of the resulting monohydrogenated species was attributed to the fact that spin density appears to be localized on the neighboring vanadium atoms (Figure 1.1). Figure 1.1. C−H activation by a phosphorus oxo leads to a one electron reduction on the neighboring vanadium. Hence the oxidation state of the phosphorus atom does not change. Chapter 4 focuses on deeper investigations into the ROA mechanistic motif. Specifically, questions such as the generalizability of the mechanism to other metal-main group pairs, as well as the applications of the ROA motif to homogeneous catalysts, are addressed. We have designed the homogeneous complex κ-P bis(2-phenoxyl)phosphinite vanadium ((OPO)V) to be a catalyst that may activate propane using the ROA mechanistic motif. We then continued to follow the reaction forward, tracing pathways for the catalytic oxidation of propane in aqueous solution using O2 as the terminal oxidant. We found that the ((OPO)V) system has the potential to be a viable catalyst in acidic aqueous solution, converting propane to either propylene or isopropanol. 1.3 Other projects Chapter 5 is a computational mechanistic study of the 2,6-bis(4-isopropyl-2-oxazolin-2-yl)pyridine nickel (Ni((iPr)−pybox)) system, a Negishi C−C coupling catalyst. In our computational work, we investigated the coupling of 1-bromoindane with methylzinc iodide catalyzed by the Ni((S, S−iPr)pybox) system to form 1-methylindane. The chiral diastereomers of this catalyst are most notable for their stereoconvergence: Using Ni((S, S−iPr)−pybox) as a catalyst, both enantiomers of racemic 1bromoindane will couple with methylzinc iodide to form (S)-1-methylindane with 90% enantiomeric excess [?]. We confirm that the mechanism proceeds through a NiI -NiIII couple, in contrast to the classic Ni0 -NiII cycle that Negishi catalysts are typically assumed to undergo [?] (Scheme 1.3). I Appendix A is an experimental and computational study of the (L)Rh (COE)(TFA) complexes in which L is DABF or BOZO (DABF = N, N -bis(pentafluorophenyl)-2,3-dimethyl-1,4-diaza-1,3butadiene, COE = cyclooctene, TFA = trifluoroacetate, BOZO = bis(2-oxazolin-2-yl)). These com- 5 Scheme 1.3. The generalized mechanism of the Ni((iPr)−pybox) Negishi coupling catalyst as it proceeds through a NiI -NiIII cycle. The overall equation is ind−Br + MeZnI −−→ ind−Me + ZnBrI. plexes have been shown to catalyze the H/D exchange of benzene in trifluoroacetic acid. Computationally, the mechanism was found to occur by the reversible oxidative addition and reductive I elimination of benzene to and from the (L)Rh (TFAH)(TFA) species. Appendix B is a comparison of various DFT and electron-correlation methods for calculating the group 10 transition metals Ni, Pd, and Pt. These elements are notable in that, despite being in the same periodic group, they each have a different electronic configuration (d8 s2 , d10 , and d9 s1 , respectively). The accuracy of DFT and electron-correlation methods in reproducing the transition energies between the three configurations for the three elements is an important consideration that has deeper implications in the reactivity of more complex systems including these elements. 6 1.4 References [1] Cheng, W-H.; Kung, H. H. Methanol Production and Use (Chemical Industries vol. 57), 1994. [2] IPCC Fourth Assessment Report, Working Group 1, Chapter 2. [3] (a) Labinger, J. A.; Bercaw, J. E. Nature 2002, 417, 507–514; (b) Crabtree, R. H. J. Chem. Soc., Dalton Trans. 2001, 2437–2450; (c) Labinger, J. A. Fuel Proc. Technol. 1995, 42, 325–338; (d) Labinger, J. A. J. Mol. Catal. A: Chem. 2004, 220, 27–35. [4] (a) Fekl, U.; Goldberg, K. I. Adv. Inorg. Chem. 2003, 54, 259–320; (b) Lersch, M.; Tilset, M. Chem. Rev. 2005, 105, 2471–2526. [5] (a) Goldshlegger, N. F.; Tyabin, M. B.; Shilov, A. E.; Shteinman, A. A. Zh. Fiz. Khim. 1969, 43, 2174–2175; (b) Goldshlegger, N. F.; Eskova, V. V.; Shilov, A. E.; Shteinman, A. A. Zh. Fiz. Khim. 1972, 46, 1353–1354; (c) Shilov, A. E.; Shul’pin, G. B. Activation and Catalytic Reactions of Saturated Hydrocarbons in the Presence of Metal Complexes; Kluwer Academic Publishers: Dordrecht, 2000; pp. 259– 317. [6] Periana, R. A.; Taube, D. J.; Evitt, E. R.; Loffler, D. G.; Wentrcek, P. R.; Voss, G.; Masuda, T. Science 1993, 259, 340–343. [7] Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154–162. [8] (a) Wong-Foy, A. G.; Bhalla, G.; Liu, X. Y.; Periana, R. A. J. Am. Chem. Soc. 2003, 125, 14292–14293; 7 (b) Oxgaard, J.; Muller, R. P.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2004, 126, 352–363; (c) Bhalla, G.; Liu, X. Y.; Oxgaard, J.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 11372–11389; (d) Tenn, W. J.; Young, K. J. H.; Bhalla, G.; Oxgaard, J.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 14172–14173; (e) Bhalla, G.; Periana, R. A. Angew. Chem., Int. Ed. 2005, 44, 1540–1543; (f) Young, K. J. H.; Mironov, O. A; Periana, R. A. Organometallics 2007, 26, 2137–2140; (g) Young, K. J. H.; Oxgaard, J.; Ess, D. H.; Meier, S. K.; Stewart, T.; Goddard, W. A.; Periana, R. A. Chem. Commun. 2009, 3270–3272. [9] (a) Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154–162; (b) Hanson, S. K.; Heinekey, D. M.; Goldberg, K. I. Organometallics 2008, 27, 1454–1463; (c) Evans, M. E.; Jones, W. D. Organometallics 2011, 30, 3371–3377; (d) Zakzeski, J. J.; Bell, A. T. J. Mol. Cat. A: Chem. 2007, 276, 8–16; (e) Zhang, X. X.; Wayland, B. B. J. Am. Chem. Soc. 1994, 116, 7897–7898; (f) Tenn, W. J. III; Conley, B. L.; Bischof, S. M.; Periana, R. A. J. Organomet. Chem. 2011, 696, 551–558; (g) Rhinehart, J. L.; Manbeck, K. A.; Buzak, S. K.; Lippa, G. M.; Brennessel, W. W.; Goldberg, K. I.; Jones, W. D. Organometallics 2012, 31, 1943–1952; (h) Li, L.; Brennessel, W. W.; Jones, W. D. Organometallics 2009, 28, 3492–3500; (i) Kloek, S. M.; Heinekey, D. M.; Goldberg, K. I. Angew. Chem. Int. Ed. 2007, 46, 4736–4738. [10] Centi, G.; Trifiro, F.; Ebner, J. R.; Franchetti, V. M. Chem. Rev. 1988, 88, 55–80. [11] Hodnett, B. K. Heterogeneous Catalytic Oxidation. Wiley: New York, 2000. [12] Cheng, M.-J. and Goddard, W. A. J. Am. Chem. Soc. 2013, 135, 4600–4603.  8 Chapter 2 Intra- and intermolecular C−H bond activation by bis(phenolate)pyridineiridium(III) complexes Reproduced with permission from Fu, R.; Bercaw, J. E.; Labinger, J. A. Organometallics 2011, 30, 6751–6765. c 2011 American Chemical Society; available at http://dx.doi.org/10.1021/om201069k. 2.1 Abstract A bis(phenolate)pyridine pincer ligand (henceforth abbreviated as ONO) has been employed to support a variety of iridium complexes in oxidation states I, III, and IV. Complexes (ONO)IrL2 Me (L = PPh3 , PEt3 ) react with I2 to cleave the Ir−C bond and liberate MeI, apparently via a mechanism beginning with electron transfer to generate an intermediate IrIV complex, which can be isolated and characterized for the case L = PEt3 . The PPh3 complex is transformed in benzene at 65◦ C to the corresponding phenyl complex, with loss of methane, and subsequently to a species resulting from metalation of a PPh3 ligand. Labeling and kinetics studies indicate that PPh3 is the initial site of C−H activation, even though the first observed product is that resulting from intermolecular benzene activation. C−H activation of acetonitrile has also been observed. 9 2.2 Introduction Functionalization of alkanes, such as the selective oxidation of methane to methanol, has long been a desired goal of both academic and industrial research [1]. Much of the focus has been on routes involving C−H activation at a transition metal center followed by oxidative cleavage of the resulting C−M bond, which offer greater possibilities for good selectivity by avoiding radical-based pathways. (Yields of methanol from the direct reaction of methane with dioxygen, where radical pathways predominate, appear to be limited to about 5%, because methanol is more rapidly oxidized than methane, a consequence of its weaker C-H bond [1]d.) Platinum complexes have received a great 2– deal of attention [2], starting with Shilov’s seminal work on the PtCl2– 4 /PtCl6 system, which can oxidize methane to methanol with some selectivity [3], and including the platinum bipyrimidine system, which has achieved the most impressive performance to date, converting methane to methyl bisulfate in up to 70% yield [4]. However, none of this platinum-based chemistry has yet been shown to lead to a practical process, owing in part to low reactivity, and a good deal of research activity has turned to other metals. Iridium in particular has been a popular choice; many (primarily low-valent) Ir complexes have been reported to show good activity for C−H activation [5], and examples of oxidative functionalization with Ir systems are known as well [6]. Interconversions between IrIII and IrV might be important in this chemistry, by analogy to the Shilov system, which involves PtII and PtIV . A separate program in our group has examined dianionic tricoordinate (“LX2 ”) pincer ligands for reactions involving highvalent early transition metal centers [7], and it occurred to us that such strongly electron donating, hard, and robust ligands might also be useful here, to provide more facile access to the relatively high IrV oxidation state. We previously reported some chemistry of a bis(phenolate)-N -heterocyclic carbene ligand (OCO) that supports complexes in the oxidation states IrI , IrIII , and IrIV , but no C−H activation chemistry was demonstrated [8]. Here we report on two related bis(phenolate)pyridine ligands (ONO = pyridine-2,6-bis[2-(4,6-di-tert-butylphenolate)]; ONOtBu = 4-tert-butylpyridine-2,6-bis[2-(4,6-ditert-butylphenolate)] 2.1), which also enable synthesis of a wide variety of IrI , IrIII , and IrIV comtBu plexes. In addition, the IrIII complex (ONO )Ir(PPh3 )2 Me readily activates C−H bonds of benzene and acetonitrile, as well as of coordinated triphenylphosphine. 2.3 Results and Discussion 2.3.1 Synthesis and Characterization of IrI complexes Because IrIII is usually highly substitutionally inert, we decided to enter the Ir(ONO) system via IrI . H2 (ONO) [9] was doubly deprotonated with NaH, followed by metalation with [Ir(cod)Cl]2 , giving 10 Figure 2.1. The ONO and ONOtBu ligands. Na[Ir(cod)(ONO)], 1, which was characterized by NMR and mass spectrometry. The symmetry of the 1 H NMR (only two tert-butyl peaks, for example) is consistent with either a five-coordinate structure or a fluxional four-coordinate structure, with the phenolates coordinated to Na+ and Ir rapidly interchanging (1a and 1b, respectively, in Scheme 2.1). The latter appears more likely on two grounds: First, the related OCO complex of IrI was crystallized (as the [K(18-crown-6)] + salt 2) and shown to have a structure analogous to that of 1b [8]. Second, attempted crystallization of 1 in the presence of 15-crown-5 (attempts to crystallize 1 without 15-crown-5 were unsuccessful) instead yielded 3, the result of C−H activation at the 3-position of the central pyridine ring (Figure 2.2). Scheme 2.1. Synthesis of Na[Ir(cod)(ONO)], 1. C−H activation probably results from complexation of the sodium ion by 15-crown-5, making it too sterically bulky to coordinate to the pendant phenolate in structure 1b, and thus labilizing the pyridine, which can flip over to allow the oxidative addition of the 3-C−H bond; the resulting 11 Figure 2.2. Crystal structure of 3. The nearest Na(15-crown-5) is located above the plane of the ligand and has been omitted for clarity. Because of crystal sensitivity, the structure obtained was only of sufficient quality to establish connectivity. iridium (III) hydride is deprotonated by the pendant phenolate to regenerate square-planar IrI . This reactivity appears to require a solution of 1 to contain the isomer 1b, either as the sole species present or, at least, in equilibrium with 1a. C−H activation in preference to N -coordination of a pyridine ring has been observed previously for an NHC-pyridine ligand [10]; in that case (and probably here as well) N -coordination is disfavored sterically. tBu In order to prevent this unwanted ligand C−H activation, we prepared H2 (ONO ), with an additional tert-butyl group to inhibit C−H activation of the pyridine linker. The synthesis of the ligand, which was straightforward and analogous to that of H2 (ONO), is shown in Scheme 2.2. DetBu protonation of H2 (ONO ) with NaH, followed by metalation with half an equivalent (i.e. one Ir per ligand) of [Ir(cod)Cl]2 , gave deep red [NaIr(cod)(ONO tBu )]2 (4, Scheme 2.3), shown crystallo- graphically to have a dimeric structure (Figure 2.3). Unlike ONO complex 1, 4 is readily soluble in pentane, presumably because the hydrophilic Na2 O2 core is surrounded by a lipophilic shell of tertbutyl-substituted phenyl rings. As with 1, the 1 H NMR spectrum indicates either a more symmetric structure in solution or a fluxional process; on cooling to −100◦ C, the spectrum exhibited some line broadening, but no decoalescence of peaks. 4 reacts with CO to give Ir4 (CO)12 and is reduced by H2 to metallic Ir. tBu If instead a full equivalent of [Ir(cod)Cl]2 is added to one equivalent of Na2 (ONO tBu diiridium complex [Ir(cod)]2 (ONO ), the orange ) (5) is obtained. 5 may also be prepared from 4 plus half an tBu equivalent of [Ir(cod)Cl]2 , but 4 is not formed from 5 and additional Na2 (ONO ) (Scheme 2.3). The two iridium atoms in 5 are inequivalent: Ir1 is five-coordinate, bound to both oxygens and the 12 Scheme 2.2. Synthesis of the ligand H2 (ONOtBu ). Scheme 2.3. Synthesis and reactivity of [NaIr(cod)(ONOtBu )]2 , 4. 13 Figure 2.3. Crystal structure of 4. The tert-butyl groups and a disordered pentane solvent molecule have been removed for clarity. Selected bond lengths (Å) and angles (◦ ): Ir-O1, 2.0990(1); Ir-N1, 2.1274(1); O1-Na1, 2.2145(1); O2-Na1, 2.1926(1); O20 -Na1, 2.2841(1); O1-Ir-N1, 81.268(3); O1-Na1-O2, 104.799(3); O1-Na1-O20 , 106.989(3); O2-Na1-O20 , 100.484(2). nitrogen of the ligand, while Ir2 is square planar and coordinated only to the two phenolate oxygens (Figure 2.4). (The Ir1-N and Ir2-N distances are 2.07 and 3.00 Å, respectively.) 2.3.2 Synthesis and Characterization of IrIII Complexes tBu Complex 4 is oxidized by AgPF6 or [FeCp2 ][PF6 ] in a mixture of THF and acetonitrile to [(ONO )- Ir(cod)(MeCN)][PF6 ] (6, Scheme 2.4), which was isolated as a crystalline product, although it decomposes over time in the presence of excess acetonitrile. Oxidation of 4 with AgOTf in benzene gave tBu the neutral triflate complex (ONO )Ir(cod)OTf (7). Both complexes exhibit approximately octa- hedral geometry (Figures 2.5 and 2.6); in solution they behave as strong electrophiles, polymerizing THF and slowly decomposing in other coordinating solvents. Scheme 2.4. Oxidation of [NaIr(cod)(ONOtBu )]2 , 4, to IrIII species. Treatment of 7 with two equivalents of triphenylphosphine yielded yellow trans-(ONO tBu )Ir- (PPh3 )2 OTf (8, Equation 2.1). Attempts to displace cyclooctadiene from 7 with triethylphosphine or 14 Figure 2.4. Crystal structure of 5. The tert-butyl groups have been removed for clarity. Selected bond lengths (Å) and angles (◦ ): Ir1-O1, 2.2701(1); Ir1-O2, 2.2962(1); Ir1-N1, 2.0818(1); Ir2-O1, 2.0808(1); Ir2-O2, 2.0771(1); Ir2-N1, 2.9953(1); O1-Ir1-O2, 68.864(1); N1-Ir1-O1, 76.957(2); N1-Ir1-O2, 75.747(1); O1-Ir2-O2, 76.775(2); Ir1-O1-Ir2, 103.134(2); Ir1-O2-Ir2, 102.367(2). Figure 2.5. Crystal structure of 6. A benzene solvent molecule and the PF6 counterion have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.0736(8); Ir1-N2, 2.0456(8); Ir1-O1, 2.0540(7); Ir1-O2, 2.0849(7); N2-C46, 1.1371(12); O1-Ir1-N1, 85.86(3); O1-Ir1-N2, 81.05(3); O2-Ir1-O1, 89.17(3); O2Ir1-N2, 85.09(3); O1-Ir1-O2, 165.29(3); N1-Ir1-N2, 89.69(3); Ir1-N2-C46, 173.51(8). 15 Figure 2.6. Crystal structure of 7. The two tert-butyl groups ortho to the phenoxy moieties have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O1, 2.0431(6); Ir1-N1, 2.0738(6); Ir1-O2, 2.0836(6); Ir1-O3, 2.1247(6); Ir1-C39, 2.1725(8); Ir1-C40, 2.1865(8); Ir1-C44, 2.2592(8); Ir1-C43, 2.2656(8); O1-Ir1-N1, 84.59(2); O1-Ir1-O2, 163.22(2); N1-Ir1-O2, 88.90(2); O1-Ir1-O3, 80.91(2); N1-Ir1-O3, 87.90(2); O2-Ir1-O3, 83.42(2); O1-Ir1-C39, 121.97(3); N1-Ir1-C39, 95.20(3); O2-Ir1-C39, 73.96(3); O3-Ir1C39, 157.08(3); O1-Ir1-C40, 84.60(3); N1-Ir1-C40, 90.09(3); O2-Ir1-C40, 110.91(3); O3-Ir1-C40, 165.50(3); C39-Ir1-C40, 37.42(3); O1-Ir1-C44, 110.92(3); N1-Ir1-C44, 164.19(3); O2-Ir1-C44, 75.34(3); O3-Ir1-C44, 91.51(3); C39-Ir1-C44, 79.40(3); C40-Ir1-C44, 94.28(3); O1-Ir1-C43, 77.60(3); N1-Ir1-C43, 159.93(3); O2-Ir1C43, 110.72(3); O3-Ir1-C43, 98.24(3); C39-Ir1-C43, 86.58(3); C40-Ir1-C43, 79.18(3); C44-Ir1-C43, 35.50(3). tricyclohexylphosphine led only to decomposition, while the reaction of 7 with tris(o-tolyl)phosphine resulted in protonated ligand, perhaps by benzylic C−H activation followed by reductive elimination of phenol. The crystal structure of 8 is show in Figure 2.7. (2.1) Exposure of 8 to water, followed by addition of proton sponge at room temperature gave transtBu (ONO )Ir(PPh3 )2 OH (9, Equation 2.2). Both 8 and 9 were converted to the corresponding chloride (10) by heating at 90◦ C in dichloromethane for several hours or several days, respectively (Equation 2.3). Formation of 10 from 8 or 9 is much faster in the presence of added tetrabutylammonium chloride. The crystal structure of 9 is show in Figure 2.8. 16 Figure 2.7. Crystal structure of 8. Two benzene solvent molecules, a tert-butyl group’s methyls, and the phosphine ligands’ phenyl groups (except ipso carbons) have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.001(3); Ir1-O1, 2.049(2); Ir1-O2, 2.051(2); Ir1-O3, 2.166(2); Ir1-P1, 2.4142(8); Ir1-P2, 2.4165(8); N1-Ir1-O1, 90.68(9); N1-Ir1-O2, 91.55(9); O1-Ir1-O2, 177.70(8); N1-Ir1-O3, 176.28(9); O1-Ir1-O3, 87.33(8); O2-Ir1-O3, 90.41(8); N1-Ir1-P1, 90.90(7); O1-Ir1-P1, 96.69(6); O2-Ir1-P1, 82.71(6); O3-Ir1-P1, 86.22(6); N1-Ir1-P2, 88.06(7); O1-Ir1-P2, 81.27(6); O2-Ir1-P2, 99.37(6); O3-Ir1-P2, 94.74(6); P1-Ir1-P2, 177.70(3). (2.2) (2.3) All three complexes (8, 9, and 10) react with proton sponge to give the hydride (ONO tBu )Ir- (PPh3 )2 H (11). The reaction of 8 takes place over several days at room temperature or in an hour at 90◦ C, while 9 and 10 react more slowly at 90◦ C. The crystal structure of 11 is show in Figure 2.9. The proposed mechanism for this transformation is single-electron transfer followed by H-atom abstraction (Scheme 2.5); the oxidized proton sponge byproduct shown was observed (quantitatively) 17 Figure 2.8. Crystal structure of 9. The tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.0395(14); Ir1-O3, 2.0561(13); Ir1-O1, 2.0624(11); Ir1-O2, 2.0686(11); Ir1-P1, 2.3654(5); Ir1-P2, 2.3920(5); N1-Ir1-O3, 179.30(6); N1-Ir1-O1, 89.78(5); O3-Ir1-O1, 90.03(5); N1-Ir1O2, 89.99(5); O3-Ir1-O2, 90.22(5); O1-Ir1-O2, 178.69(5); N1-Ir1-P1, 97.67(4); O3-Ir1-P1, 83.02(4); O1-Ir1P1, 95.33(4); O2-Ir1-P1, 83.43(4); N1-Ir1-P2, 90.84(4); O3-Ir1-P2, 88.46(4); O1-Ir1-P2, 81.46(4); O2-Ir1-P2, 99.83(4); P1-Ir1-P2, 170.911(15). Figure 2.9. Crystal structure of 11. A pentane solvent molecule and the tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O1, 2.0663(11); Ir1-O2, 2.0764(11); Ir1-N1, 2.1020(16); Ir1-P2, 2.3229(5); Ir1-P1, 2.3320(5); Ir1-H1, 1.70(2); O1-Ir1-O2, 176.19(5); O1-Ir1-N1, 87.97(5); O2-Ir1-N1, 88.25(5); O1-Ir1-P2, 86.17(4); O2-Ir1-P2, 94.68(3); N1-Ir1-P2, 95.82(4); O1-Ir1-P1, 96.13(4); O2-Ir1-P1, 83.91(3); N1-Ir1-P1, 97.67(4); P2-Ir1-P1, 166.384(19); O1-Ir1-H1, 95.6(6); O2-Ir1-H1, 88.2(6); N1-Ir1-H1, 176.3(7); P2-Ir1-H1, 85.7(7); P1-Ir1-H1, 80.7(7). 18 by NMR. A related reaction of proton sponge with an IrIII complex has previously been reported [11]. In further support of this proposal, 11 was also obtained by treatment of 8 with cobaltocene (a oneelectron donor) and the H atom donor 9,10-dihydroanthracene (DHA), or by heating 8 with sodium hydride. 11 can also be obtained from the reaction of 8 with lithium methoxide or diethylzinc, presumably via β-hydride elimination from an iridium-methoxy [12] or iridium-ethyl intermediate, respectively. The two latter routes give 11 most cleanly; in the reactions with proton sponge, we were unable to separate 11 from the byproducts. Scheme 2.5. Synthetic routes to (ONOtBu )Ir(PPh3 )2 H, 11. Treatment of 8 with dimethylzinc yielded the water-stable methyl complex (ONO (12). Oxidation of 12 with I2 yields the iodide (ONO tBu tBu )Ir(PPh3 )2 Me )Ir(PPh3 )2 I (13) and methyl iodide (Scheme 2.6). Scheme 2.6. Methylation to (ONOtBu )Ir(PPh3 )2 Me, 12, and subsequent functionalization with I2 to (ONOtBu )Ir(PPh3 )2 I, 13. Crystal structures for complexes 8, 9, 11, 12, and 13 all exhibit closely analogous octahedral geometries. The structures of 12 and 13 are shown in Figures 2.10 and 2.11. Unlike 8, 12 reacts cleanly with triethylphosphine to give the analogous PEt3 complex (14, 19 Figure 2.10. Crystal structure of 12. Only ipso carbon atoms of the phenyl groups on phosphorus are shown. Selected atom distances (Å) and angles (◦ ): Ir-O1, 2.0901(1); Ir-O2, 2.0700(1); Ir-C74, 2.0952(1); Ir-N, 2.1042(1); Ir-P1, 2.3449(1); Ir-P2, 2.3895(1); O1-Ir-O2, 176.495(3); P1-Ir-P2, 172.872(2); N-Ir-C74, 175.671(3); C74-Ir-O1, 95.588(2); C74-Ir-O2, 87.799(2); C74-Ir-P1, 85.584(2); C74-Ir-P2, 87.309(2); O1-IrP1, 82.692(2); O1-Ir-P2, 98.559(2); O2-Ir-P1, 96.698(2); O2-Ir-P2, 82.473(2); N-Ir-O1, 88.429(2); N-Ir-O2, 88.212(2); N-Ir-P1, 96.553(2); N-Ir-P2, 90.503(2). Figure 2.11. Crystal structure of 13. Two pentane solvent molecules and the tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.040(3); Ir1-O2, 2.066(2); Ir1-O1, 2.076(2); Ir1-P2, 2.4233(10); Ir1-P1, 2.4247(11); Ir1-I2, 2.7013(3); N1-Ir1-O2, 89.52(10); N1-Ir1-O1, 88.88(10); O2-Ir1-O1, 178.36(9); N1-Ir1-P2, 89.69(9); O2-Ir1-P2, 78.59(8); O1-Ir1-P2, 101.03(8); N1-Ir1-P1, 89.04(9); O2-Ir1-P1, 102.10(8); O1-Ir1-P1, 78.25(8); P2-Ir1-P1, 178.55(3); N1-Ir1-I2, 179.19(9); O2-Ir1-I2, 90.56(6); O1-Ir1-I2, 91.05(7); P2-Ir1-I2, 91.12(2); P1-Ir1-I2, 90.15(2). 20 Scheme 2.7). 14 appears to be more thermally stable than 12, only slowly decomposing at tempertBu atures above 120◦ C. It reacts similarly with I2 , giving methyl iodide and (ONO )Ir(PEt3 )2 I (15). Both 14 (Figure 2.12) and 15 (Figure 2.13) were characterized crystallographically. Iodide complex 15 is unstable in methylene chloride, converting to the analogous chloride over several days at room temperature. Such behavior was not observed with the PPh3 analogue 13. Scheme 2.7. Synthesis of (ONOtBu Ir(PEt3 )2 complexes. Figure 2.12. Crystal structure of 14. The ethyl groups of one phosphine ligand have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O2, 2.0674(11); Ir1-O1, 2.0702(11); Ir1-N1, 2.1050(14); Ir1-C50, 2.1154(17); Ir1-P2, 2.3479(5); Ir1-P1, 2.3514(5); O2-Ir1-O1, 178.26(5); O2-Ir1-N1, 88.65(5); O1Ir1-N1, 89.83(5); O2-Ir1-C50, 91.12(6); O1-Ir1-C50, 90.42(6); N1-Ir1-C50, 178.40(6); O2-Ir1-P2, 84.03(4); O1-Ir1-P2, 95.23(4); N1-Ir1-P2, 93.93(4); C50-Ir1-P2, 87.61(5); O2-Ir1-P1, 97.78(4); O1-Ir1-P1, 83.10(4); N1-Ir1-P1, 91.40(4); C50-Ir1-P1, 87.07(5); P2-Ir1-P1, 174.416(16). In both I2 oxidations a dark blue color is observed over much of the course of the reaction, even though the starting complexes and the products are all yellow. Cyclic voltammetry experiments show quasi-reversible oxidations at 20 and 780 mV for 12 and reversible oxidations at −100 and 21 Figure 2.13. Crystal structure of 15. The ethyl groups of the phosphine ligands have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-N1, 2.0619(13); Ir1-O2, 2.0648(11); Ir1-O1, 2.0661(11); Ir1-C(50), 2.148(7); Ir1-P1, 2.3711(4); Ir1-P2, 2.3722(4); Ir1-I2, 2.6625(3); N1-Ir1-O2, 89.86(5); N1-Ir1-O1, 88.67(5); O2-Ir1-O1, 178.36(5); N1-Ir1-C(50), 177.12(15); O2-Ir1-C(50), 89.65(17); O1-Ir1-C(50), 91.86(17); N1-Ir1-P1, 93.81(4); O2-Ir1-P1, 95.64(3); O1-Ir1-P1, 83.74(3); C(50)-Ir1-P1, 89.06(15); N1-Ir1-P2, 91.30(4); O2-Ir1-P2, 82.36(3); O1-Ir1-P2, 98.40(3); C(5)-Ir1-P2, 85.82(15); P1-Ir1-P2, 174.510(14); N1-Ir1-I2, 178.40(4); O2-Ir1-I2, 91.65(3); O1-Ir1-I2, 89.83(3); C(50)-Ir1-I2, 3.05(16); P1-Ir1-I2, 86.571(11); P2-Ir1-I2, 88.374(11). 840 mV for 14 both vs. ferrocene (Figure 2.14). The reaction of 14 with AgPF6 gave dark blue tBu [(ONO )Ir(PEt3 )2 Me]PF6 (16, Scheme 2.7), characterized by a crystal structure and an EPR spectrum (Figure 2.15) that are very similar to those of the (relatively few) organometallic IrIV species reported [13]. 16 could be reduced back to 14 with either cobaltocene or methyllithium; furthermore, treatment of 16 with tetrabutylammonium iodide gave MeI and some 15. All of this suggests that the reaction of 14 (or 12) with I2 begins with a single electron transfer to generate an intermediate cationic IrIV -Me species, which undergoes SN 2 attack by I– at the Ir−C bond. The CV data suggest that further oxidization of 16 might be possible, perhaps even to an IrV -methyl complex, but no such species has yet been isolated or otherwise identified. (The close similarity of the second potentials, in contrast to the first, may indicate that the second wave corresponds to oxidation of the ONOtBu ligand rather than at Ir.) The crystal structure of 16 is show in Figure 2.16). 2.3.3 C−H Activation by IrIII (ONOtBu ) Complexes In benzene solution, (ONO tBu )Ir(PPh3 )2 Me converts over the course of several hours at 65◦ C to the analogous phenyl complex (ONO tBu )Ir(PPh3 )2 Ph (17), with evolution of methane. The benzene solution of 17 reacts further over the course of several weeks at 65◦ C to give a cyclometalated tBu product, (ONO )Ir(PPh3 )[κ2 -PPh2 (o-C6 H4 )] (18, Equation 2.4). 22 Figure 2.14. Cyclic voltammetry of 14, carried out in the air at room temperature. The solvent was dichloromethane with 0.1 M tetrabutylammonium tetrafluoroborate added as the electrolyte. Multiple runs were made with scan rates from 50 to 400 mV/s. Quasi-reversibility is seen in the form of a following chemical step. The dip at 0.3 V is an artifact. Figure 2.15. EPR spectrum of [(ONOtBu )Ir(PEt3 )2 Me]PF6 , 16, in dichloromethane at 20 K. 23 Figure 2.16. Crystal structure of one of the molecules of 16 in the unit cell. The unit cell has four iridiumcontaining molecules total; this figure only shows one of them. In addition, ethyl groups on the phosphine ligands, as well as the six benzene molecules per unit cell, have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O2A, 1.994(5); Ir1-O1A, 2.011(5); Ir1-C50A, 2.101(8); Ir1-N1A, 2.129(7); Ir1P2A, 2.400(2); Ir1-P1A, 2.418(2); O2A-Ir1-O1A, 173.5(2); O2A-Ir1-C50A, 94.2(3); O1A-Ir1-C50A, 92.3(3); O2A-Ir1-N1A, 87.4(2); O1A-Ir1-N1A, 86.1(2); C50A-Ir1-N1A, 177.4(3); O2A-Ir1-P2A, 82.62(17); O1A-Ir1P2A, 98.06(18); C50A-Ir1-P2A, 84.4(2); N1A-Ir1-P2A, 93.78(18); O2A-Ir1-P1A, 98.16(17); O1A-Ir1-P1A, 81.81(18); C50A-Ir1-P1A, 89.8(2); N1A-Ir1-P1A, 91.96(18); P2A-Ir1-P1A, 174.24(8). (2.4) When 17 is isolated, redissolved in a solvent other than benzene (such as p-xylene), and heated to 65◦ C, conversion to 18 is complete within a few days. Both 17 (Figure 2.17 and 18 (Figure 2.18) were characterized crystallographically. Thermolysis of 12 in C6 D6 yields CH4 as the only detectable methane isotopologue, and mass spectrometry of the resulting isolated phenyl product 17 shows incorporation of six D atoms (EquatBu tion 2.5). In contrast, heating (ONO )Ir[P(C6 D5 )3 ]2 (CH3 ) (12-d30 ) in C6 D6 yields only CH3 D; this reaction is slower than that of all-protio-12, exhibiting a KIE of 4 2.20. The PEt3 analogue 14 tBu reacts very slowly in C6 H6 or C6 D6 , requiring several weeks at 120◦ C to give (ONO and CH4 or CH3 D, respectively. )Ir(PEt3 )2 Ph 24 Figure 2.17. Crystal structure of 17. A benzene and a pentane solvent molecule, as well as the phosphine phenyl groups, have been removed for clarity. The molecule is symmetric about the iridium center. Selected atom distances (Å) and angles (◦ ): Ir1-O1, 2.0744(12); Ir1-C40, 2.074(3); Ir1-N1, 2.103(2); Ir1-P1, 2.3921(6); O1-Ir1-C40, 90.05(4); O1-Ir1-O10 , 179.91(8); C40-Ir1-O10 , 90.04(4); O1-Ir1-N1, 89.95(4); C40-Ir1N1, 180.0; O10 -Ir1-N1, 89.96(4); O1-Ir1-P10 , 97.70(4); C40-Ir1-P10 , 88.831(13); O10 -Ir1-P10 , 82.30(4); N1-Ir1P10 , 91.169(13); O1-Ir1-P1, 82.30(4); C40-Ir1-P1, 88.829(13); O10 -Ir1-P1, 97.70(4); N1-Ir1-P1, 91.171(13); P10 -Ir1-P1; 177.66(3). Figure 2.18. Crystal structure of 18. A dodecane solvent molecule and parts of several phenyl and tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir-O1, 2.0696(1); Ir-O2, 2.0612(1); Ir-C46, 2.0669(1); Ir-N, 2.1156(1); Ir-P4, 2.3462(1); Ir-P5, 2.3483(1); O1-Ir-O2, 174.874(3); P4Ir-P5, 164.008(2); N-Ir-C46, 167.055(4); C46-Ir-O1, 100.779(3); C46-Ir-O2, 80.655(3); C46-Ir-P4, 95.708(2); C46-Ir-P5, 68.499(2); O1-Ir-P4, 85.316(3); O1-Ir-P5, 94.944(2); O2-Ir-P4, 99.475(2); O2-Ir-P5, 80.959(3); N-Ir-O1, 88.656(3); N-Ir-O2, 89.198(3); N-Ir-P4, 93.867(3); N-Ir-P5, 102.125(3). 25 (2.5) Thermolysis of 12 at 65◦ C in methylene chloride, THF, ether, or p-xylene leads directly to 18 (Equation 2.6). These transformations require less than one day for complete conversion (in methylene chloride or p-xylene), comparable to the time needed for conversion of 12 to 17 and much faster than the conversion of 17 to 18 in benzene. Reaction of 12 in toluene at 65◦ C proceeds similarly to that in benzene; although a single pure compound could not be isolated, the mass spectrum of the mixture is consistent with formation of an IrIII tolyl compound, (ONO The 31 tBu )Ir(PPh3 )2 (C6 H4 CH3 ). P NMR shows two strong signals of roughly equal intensity with shifts close to that of phenyl compound 17, suggesting the formation of two isomers, presumably meta and para (since p-xylene is unreactive); a much weaker nearby signal may indicate a small amount of the ortho isomer. (2.6) Although 12 is insoluble in acetonitrile, thermolysis in methylene chloride containing a small amount of acetonitrile affords a mixture of 18 and a product resulting from C−H activation of acetBu tonitrile, [(ONO )Ir(PPh3 )(CH2 CN)]2 (19); with CD3 CN, CH4 is the only methane isotopologue produced (Equation 2.7). 19 has the dimeric structure shown in Figure 2.19, where coordination of N from the (CH2 CN) group bonded to each Ir has displaced a PPh3 ligand from the other. Although two 1 H NMR signals would be expected for the (CH2 CN) methylene protons, which are diastereotopic in the solid-state structure, only one is observed; this could result from either accidental degeneracy or dissociation of the dimeric structure in solution. 26 (2.7) Figure 2.19. Crystal structure of 19 [14]. A benzene solvent molecule, phosphine phenyl, and (ONO) tert-butyl groups have been removed for clarity. Selected atom distances (Å) and angles (◦ ): Ir1-O1, 2.0190(2); Ir1-O2, 2.0883(2); Ir1-C104, 2.1472(2); Ir1-N1, 2.0758(2); Ir1-N4, 2.0882(2); Ir1-P1, 2.2654(2); N3-C103, 1.1271(1); C103-C104, 1.4380(1); N4-C102, 1.1278(1); C102-C101, 1.4482(2); Ir2-O3, 2.0912(2); Ir2-O4, 2.0396(2); Ir2-C101, 2.1740(3); Ir2-N2, 2.0854(3); Ir2-N3, 2.1290(2); Ir2-P2, 2.2574(2); O1-Ir1O2, 173.203(10); N4-Ir1-P1, 173.670(9); N1-Ir1-C104, 170.591(10); C104-Ir1-O1, 81.941(8); C104-Ir1-O2, 98.236(8); C104-Ir1-N4, 86.476(8); C104-Ir1-P1, 90.036(7); O1-Ir1-N4, 88.426(8); O1-Ir1-P1, 96.321(8); O2Ir1-N4, 84.807(8); O2-Ir1-P1, 90.475(8); N1-Ir1-O1, 89.793(8); N1-Ir1-O2, 89.439(8); N1-Ir1-N4, 88.823(8); N1-Ir1-P1, 95.360(8); Ir1-N4-C102, 156.881(14); N4-C102-C101, 171.704(16); C102-C101-Ir2, 104.11(1); Ir2N3-C103, 156.797(13); N3-C103-C104, 172.115(16); C103-C104-Ir1, 105.731(10); O3-Ir2-O4, 174.306(10); N3Ir2-P2, 171.891(9); N2-Ir2-C101, 172.156(10); C101-Ir2-O3, 96.853(8); C101-Ir2-O4, 84.460(8); C101-Ir2-N3, 85.543(7); C101-Ir2-P2, 88.633(7); O3-Ir2-N3, 85.175(8); O3-Ir2-P2, 89.930(7); O4-Ir2-N3, 89.412(8); O4Ir2-P2, 95.646(8); N2-Ir2-O3, 89.263(8); N2-Ir2-O4, 88.981(8); N2-Ir2-N3, 90.101(8); N2-Ir2-P2, 96.309(8). 27 A plausible mechanism for the chemistry of 12 in benzene is shown in Scheme 2.8. The sequence of product formation (17 and 18 being preferred kinetically and thermodynamically, respectively [15]) would seem to suggest that the first C−H activation is the intermolecular reaction with benzene. But that is excluded by the isotopic labeling results. The H atom that departs with methane clearly comes from a PPh3 ligand, not solvent benzene, so PPh3 must be the site of the first C−H activation, even though the corresponding product 18 does not appear at all until much later. We propose therefore that dissociation of one PPh3 ligand gives coordinatively unsaturated A, followed by intramolecular C−H activation and loss of methane to generate cyclometalated intermediate C. In benzene solution coordination of PPh3 to C is not competitive; instead C reacts with benzene to give 17 as initial observed product. Subsequent conversion to the thermodynamic product 18 is much slower. Scheme 2.8. Proposed mechanism for the transformation of 12 in benzene to the kinetic product 17 and the thermodynamic product 18. 28 C−H activation is shown in Scheme 2.8 as proceeding via oxidative addition to give an IrV intermediate, a mechanism that has been supported by both experiment [16] and calculations [17] for the above-mentioned [15] Cp* system. It seems reasonable that similar considerations would apply here as well; but an alternative pathway, such as σ-bond metathesis, cannot be excluded. In any case, conversion of A to C must be irreversible, as no multiple H/D exchange is observed, whereas conversion of C to F must be reversible, since 17 eventually reverts to 18. The fact that products are obtained from solvent activation with toluene but not p-xylene suggests that C−H activation ortho to a methyl substituent is sterically inhibited. As a consequence, in the latter solvent C (or a π-arene adduct analogous to D) has no alternative but to bind PPh3 and 18 forms directly, as it does in solvents (dichloromethane, ether, THF) that are not susceptible to C−H activation at all. In acetonitrile the two pathways are competitive. The reaction of 14 is much slower, presumably because PEt3 dissociates much less readily, and leads only to a phenyl product, indicating that cyclometalation of PEt3 does not take place. 2.3.4 Kinetics of C−H Activation The transformation of 12 to 17 and/or 18 is readily followed by 31 P NMR, as each of the species has a distinct spectrum. In early experiments results were not always highly reproducible; most probably this is due to oxidation of small amounts of PPh3 by adventitious impurities, which could effect significant acceleration; indeed, deliberate addition of oxidants such as chloranil and dioxygen did speed up conversion of 12 considerably. However, reasonably reproducible rates (±15%) could be achieved by careful attention to purification of starting complex and solvent as well as exclusion of air. The somewhat simplified mechanism shown in Scheme 2.9 may be used for analysis; it leaves out the (unobserved) intermediates B (which will rapidly convert to C) and D and E (which are presumed to be in rapid equilibrium with C and F); none of these should have any kinetic consequences. Three transformations will be considered in turn: (1) conversion of 12 to 17 in benzene; (2) conversion of 12 to 18 in p-xylene; and (3) conversion of 17 to 18 in benzene or p-xylene. The first of these may be analyzed in terms of reversible dissociation of PPh3 (henceforth abbreviated as L) followed by rate-determining C−H activation (the k2 step). Since the conversion of A to C is irreversible (vide supra), no buildup of any intermediate is observed, and formation of 17 is essentially complete before any 18 is observed. In this simplified framework, we can treat the dissociation of L from 12 as a fast pre-equilibrium and the intramolecular C−H activation of L as rate-determining, giving us K1 = [A][L]/[12]. Thus the rate law would be −d[12]/dt = d[17]/dt = k2[A] = (k2 K1 [12])/[L], and the rate should be cleanly first order in 12 and inverse-first-order in L. There is a potential complication if no extra L is added: [L] will be governed by both K1 and K4 , so if these two equilibrium constants were significantly different, [L] would vary 29 Scheme 2.9. Simplified version of Scheme 2.8, used for kinetic analysis. 30 in a complex manner as conversion of 12 to 17 proceeds, and the disappearance of 12 or appearance of 17 might not follow clean first-order kinetics. Nonetheless, it does: a good exponential fit is achieved (Figure 2.20, left), with kobs = 1.8(3)×10−4 s−1 (equivalent to a half-life of about an hour), so any difference in K values is small enough to neglect. The 12-d30 ((ONO isotopologue behaves similarly, yielding kobs = 4.4(6) × 10 −5 s −1 tBu )Ir[P(C6 D5 )3 ]2 Me) and hence KIE = 4(1) (Figure 2.20, right), consistent with the assignment of C−H activation of L as rate-determining. Figure 2.20. Left: kinetics of conversion of 12 to 17 in C6 D6 at 65◦ C. Right: kinetics of conversion of 12-d30 ((ONOtBu )Ir[P(C6 D5 )3 ]2 Me) to 17-d36 ((ONOtBu )Ir[P(C6 D5 )3 ]2 C6 D5 ) in C6 D6 at 65◦ C. As expected, the addition of extra L slows the reaction significantly; a plot of 1/kobs is linear in [L] (Figure 2.21). In principle, at sufficiently high [L] some 18 should form directly in competition with 17; in practice, such conditions are probably not reachable (and conversion of 12 would become extremely slow anyway). Figure 2.21. Dependence of the reciprocal of the rate of conversion of 12 to 17 on the amount of added triphenylphosphine. See Section 2.5.2 for details on how this plot was constructed. The rate law for direct conversion of 12 to 18 in p-xylene should be identical to the above, and the reaction does exhibit clean first-order kinetics, at a rate (kobs = 8(1) × 10−5 s−1 ) about a factor of 2 slower than that for 12 to 17. There are several possible explanations for the discrepancy. First, the values of K1 and k2 need not be identical to those in benzene (though they would not be expected to differ very much). Second, intermediate C may be complexed by p-xylene to give 31 an analogue of intermediate D, which in this case (in contrast to benzene) would not lead to any productive reaction, but would reduce the concentration of C and hence the rate of formation of 18. However, any substantial (i.e. sufficient to account for the entire factor of 2) concentration of a new intermediate should have been detectable by NMR. Finally (and perhaps most likely), it may be that agreement within a factor of 2 (especially on changing solvent) is as good as can be expected, given the reproducibility considerations discussed earlier. As noted above, the conversion of 17 to 18 is much slower in benzene than in p-xylene, taking place over weeks or days, respectively (Figure 2.22). This observation is readily explained by Scheme 2.9: the concentration of C, the precursor to 18, will be inversely related to the concentration of benzene. The rate law for that mechanism can be derived by using the steady-state approximation for both C and F, both of which should be present in low concentration and in rapid equilibrium with one another. This is most simply accomplished by defining a dummy variable [X] = [C] + [F]. For steady-state in [X]: 0 = d[X]/dt = k−4 [17] − k4 [F][L] − k5 [C][L]. Substituting [F] = K3 [C][PhH] gives k−4 [17] − k4 [F][L] − k5 [C][L] = k−4 [17] − K3 k4 [C][PhH][L] − k5 [C][L] = 0; hence [C] = (k−4 [17])/(k5 [L] + K3 k4 [PhH][L]). Hence the rate law for the conversion is −d[17]/dt = d[18]/dt = k5 [C][L] = (k−4 k5 [17])/(k5 + K3 k4 [PhH]). In benzene solution the reaction should be first-order in 17 and independent of [L]; both were observed. In p-xylene, on the other hand, [PhH] is continuously increasing as 17 is consumed, so first-order kinetics (which are observed) would not be expected unless k5 is large compared to k4 K3 [PhH] over the course of the reaction. Figure 2.22. Kinetics for conversion of 17 to 18 in C6 D6 (left) and p-xylene-d10 (right) at 65◦ C. We can test this for self-consistency with the observed rate constants, again assuming that there is little difference in parameters between benzene and p-xylene reactions. For conversion of 17 to 18 in benzene, kobs = 4.2(6) × 10−7 s−1 . Since no direct formation of 18 from 12 was observed, k5 must be small compared to k4 K3 [PhH], so that value of kobs is approximately equal to k−4 k5 /k4 K3 [PhH]. For the reaction in p-xylene, kobs = 2.9(4) × 10−6 s−1 ; if our assumption that k5 is large compared to k4 K3 [PhH] is correct, that value of kobs = k−4 . Substituting this value into 32 the above expression, we get k5 /k4 K3 [PhH] = 0.13, so k5 is not completely negligible compared to k4 K3 [PhH], but small enough (within the limits of reproducibility as well as the assumption of solvent-independent parameters) to validate the approximation for kobs in benzene. Furthermore, since for neat benzene [PhH] ≈ 11 M, k5 /k4 K3 = 1.4 M. For the reaction in p-xylene, the starting [17] = the maximum [PhH] = 8.8 mM and the minimum value of k5 /k4 K3 [PhH] = 160, so the assumption required to account for first-order kinetics is valid as well. Accordingly, the kinetics appear completely consistent with the proposed mechanism. 2.4 Conclusions The ONO ligand provides a versatile framework for synthesis of a wide variety of IrI and IrIII complexes. Using this framework, two key steps in a scheme for functionalization of hydrocarbons have been demonstrated. Conversion of the Ir−Me bond in complexes 12 and 14 to MeI is effected by addition of I2 ; there is evidence that the reaction proceeds via one-electron oxidation to an (isolable) IrIV methyl intermediate. Triphenylphosphine complex 12 is able to activate C−H bonds in benzene (and toluene) under quite mild conditions; the mechanism follows an unusual pathway in which the first C−H activation is intramolecular, but the product of that process, although thermodynamically preferred, is kinetically disfavored with respect to the intermolecular benzene activation product. Unfortunately the (ONO)Ir III system is not sufficiently reactive to activate nonaromatic C−H bonds (except for acetonitrile). 2.5 Supporting Details 2.5.1 Experimental Section 2.5.1.1 General Methods All compounds were prepared in a nitrogen-filled glovebox unless otherwise specified. Reactions taking place in dichloromethane at 90◦ C were conducted in either sealed glass bombs or J-Young NMR tubes. Solvents in the glovebox were dried using the Grubbs method [18]. All NMR solvents were purchased from Cambridge Isotope Laboratories, Inc., (except for CDCl3 ) filtered through alumina, and stored over 4 Å molecular sieves. POCl3 was purchased from Acros Organics. 2,6-dibromopyridine, nBuLi, tBuLi, 4-tert-butylpyridine, [FeCp2 ][PF6 ], indene, N,N -dimethylaminoethanol, PPh3 , propylene, proton sponge, AgPF6 , AgOTf, 15-crown-5, and NaH were purchased from Aldrich. H2 O2 was purchased from EMD. [Ir(cod)Cl]2 and Pd(PPh3 )4 were purchased from Strem. The synthetic route towards the ligand H2 (ONO) closely follows a previously described procedure [9]. 1 H, 13 C, 19 F, and 31 P NMR spectra were recorded on Varian Mercury 300 MHz NMR spectrometers. X-ray data were 33 collected using a Bruker KAPPA APEX II diffractometer. 2.5.1.2 Synthesis of 4-tert-Butylpyridine-N -oxide Following a published procedure [19], 4-tert-butylpyridine (18.0 mL, 16.6 g, 123 mmol) was mixed with 135 mL of glacial acetic acid and 100 mL of hydrogen peroxide (30% in water) in a 1 L roundbottom flask under air. After refluxing for 4 h, 100 mL of additional hydrogen peroxide (30% in water) was added, and the mixture was refluxed overnight. Then 200 mL of the solvent was removed by distillation, and 100 mL water was added and then removed by distillation as well. The remaining solution was neutralized with sodium carbonate. The organic layer was extracted with methylene chloride, filtered, and evaporated to dryness to yield white crystals (which gave a yellow solution in CH2 Cl2 ). Isolated yield: 15.3 g, 101 mmol, 82%. 1 H NMR (300 MHz, CDCl3 ): δ 1.31 (s, 9H, C(CH3 )3 ), 7.26 (d, 2H, 3 J = 7.2 Hz, aryl-H), 8.15 (d, 2H, 3 J = 7.5 Hz, aryl-H). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 30.5 (C(CH3 )3 ), 34.6 (C(CH3 )3 ), 123.1, 138.6, 151.2 (aryl). 2.5.1.3 Synthesis of 2-Chloro-4-tert-butylpyridine In a 250 mL round-bottom flask open to the air, 4-tert-butylpyridine-N -oxide (15.3 g, 101 mmol) was dissolved in 61 mL of neat POCl3 . The mixture was allowed to reflux under argon for 12 to 16 h (refluxing can also be done under air, but the yield is somewhat lower). Excess POCl3 was removed by distillation, and the remaining liquid was washed with a saturated aqueous solution of sodium carbonate. The organic layer was extracted with ether, filtered, and evaporated to dryness, yielding a dark red oil (13.5 g, 79.3 mmol, 79% yield). 1 H NMR (300 MHz, CDCl3 ): δ 1.26 (s, 9H, C(CH3 )3 ), 7.16 (dd, 1H, 3 J = 5.4 Hz, 4 J = 1.8 Hz, aryl-H), 7.25 (d, 1H, 4 J = 1.8 Hz, aryl-H), 8.23 (dd, 1H, 3 J = 5.7 Hz, 5 J = 0.6 Hz, aryl-H). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 30.3 (C(CH3 )3 ), 35.0 (C(CH3 )3 ), 119.7, 121.3, 149.4, 151.7, 163.5 (aryl). 2.5.1.4 Synthesis of 2-Chloro-4-tert-butylpyridine-N -oxide In a 1 L round-bottom flask, 2-chloro-4-tert-butylpyridine (13.5 g, 79.3 mmol) was mixed with 100 mL of glacial acetic acid and 80 mL of hydrogen peroxide (30% in water). After refluxing for 4 h under air, 80 mL of additional hydrogen peroxide (30% in water) was added and the mixture was refluxed overnight. Then 200 mL of the solvent was removed by distillation, and 100 mL of water was added and then distilled off as well. The remaining solution was neutralized with sodium carbonate. The organic layer was extracted with methylene chloride, filtered, and evaporated to dryness to yield yellow crystals (10.3 g, 55.3 mmol, 70% yield). 1 H NMR (300 MHz, CDCl3 ): δ 1.31 (s, 9H, C(CH3 )3 ), 7.19 (dd, 1H, 3 J = 6.6 Hz, 4 J = 3.0 Hz, aryl-H), 7.44 (d, 1H, 4 J = 2.7 Hz, aryl-H), 8.27 (dd, 1H, 3 J = 6.6Hz, 5 J = 0.6 Hz, aryl-H). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 30.4 (C(CH3 )3 ), 34.7 (C(CH3 )3 ), 121.3, 124.1, 139.8, 151.3 (aryl). 34 2.5.1.5 Synthesis of 2,6-Dichloro-4-tert-butylpyridine In a 250 mL round-bottom flask open to the air, 2-chloro-4-tert-butylpyridine-N -oxide (15.3 g, 101 mmol) was dissolved in 40 mL of neat POCl3 . The mixture was allowed to reflux under argon for 12 to 16 h. Excess POCl3 was removed by distillation, and the remaining liquid was washed with a saturated aqueous solution of sodium carbonate. The organic layer was extracted with ether, filtered, and evaporated to dryness, yielding a thick black oil. This oil was subjected to column chromatography over silica (95:5 hexanes/ethyl acetate), affording brown crystals after evaporation of solvent. The crystals were washed with hexanes, giving white 2,6-dichloro-4-tert-butylpyridine (5.57 g, 27.3 mmol, 49% yield; yield over four steps from 4-tert-butylpyridine: 22%). 1 H NMR (300 MHz, CDCl3 ): δ 1.30 (s, 9H, C(CH3 )3 ), 7.23 (s, 2H, aryl-H). 13 C{1 H} NMR (75.5 MHz, CDCl3 ): δ 30.3 (C(CH3 )3 ), 35.4 (C(CH3 )3 ), 120.2, 150.6, 166.3 (aryl). 2.5.1.6 Synthesis of O-Methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide Bromine (3.7 mL, 12 g, 73 mmol) was added via syringe to a solution of 2,4-di-tert-butylphenol (15 g, 73 mmol) in CH2 Cl2 (200 mL). The brown color of Br2 disappeared upon addition. GC-MS analysis after 5 min shows only the presence of the desired brominated product(M+ = 284 and 286). The organic mixture was washed with water, then dried over MgSO4 , and filtered. Removal of volatile material by rotary evaporation gave a golden oil which solidified after placing under high vacuum (< 1 mTorr). This material (4,6-di-tert-butyl-2-bromophenol) was dissolved in dry THF (200 mL) under argon, and was deprotonated with NaH (1.92 g, 80 mmol). After the addition of NaH the reaction mixture was stirred for 1 h at room temperature; then MOMCl (6.1 mL, 6.5 g, 80 mmol) was added via syringe. The reaction mixture was stirred at room temperature for 9 h. Water was added and the mixture was concentrated under vacuum. The desired product was extracted with CH2 Cl2 (three times); the combined organic fractions were dried over MgSO4 , filtered, and dried under high vacuum overnight to give 23.5 g (98% yield over two steps) of O-methoxymethyl-2-bromo4,6-di-tert-butylphenoxide as a golden oil. 1 H NMR (300 MHz, CDCl3 ): δ 1.30 (s, 9H, C(CH3 )3 ), 1.44 (s, 9H, C(CH3 )3 ), 3.70 (s, 3H, OCH3 ), 5.23 (s, 2H, OCH2 O), 7.32 (d, 2H, aryl-H), 7.41 (d, 2H, aryl-H). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 31.0 (C(CH3 )3 ), 31.5 (C(CH3 )3 ), 34.8 (C(CH3 )3 ), 36.1 (C(CH3 )3 ), 57.9 (OCH3 ), 99.5 (OCH2 O), 117.7, 124.1, 128.9, 144.6, 147.8, 150.7 (aryl). MS FAB+: 328 (M+ ). 2.5.1.7 Synthesis of MOM2 (ONO) A mixture of O-Methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide (5.19 g, 15.8 mmol) and THF (25 mL) in a Schlenk tube fitted with a screw-in Teflon stopper was frozen in a cold well, along with a solution of tBuLi (2.13 g, 33.3 mmol, 2.1 eq) in hexanes (10 mL), in an inert atmosphere 35 glove box. Both solutions were taken out of the box, and the tBuLi solution was added to the solution of O-Methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide while allowing the latter to thaw. The mixture was stirred for 1 h while allowing it to reach room temperature. ZnCl2 (1.51 g, 7.12 mmol, 0.70 eq) was added with the aid of THF. After stirring the reaction mixture for 30 minutes, 1,3-dibromobenzene (1.69 g, 7.12 mmol, 0.45 eq) and Pd(PPh3 )4 (188.4 mg, 0.163 mmol, 0.01 eq) were added with the aid of some THF. The reaction vessel was placed in an oil bath preheated to 75◦ C and stirred for 16 h, then allowed to cool to room temperature and quenched with water. Volatile materials were removed under vacuum and water was added. This mixture was extracted with Et2 O (three times). The combined organics were dried over MgSO4 , filtered, and concentrated by rotary evaporation. The resulting residue was suspended in MeOH cooled to −25◦ C. The resulting white precipitate was collected by filtration through a sintered glass funnel, washed with cold MeOH, and immediately used for the synthesis of H2 (ONO). 1 H NMR (300 MHz, CDCl3 ): δ 1.37 (s, 18H, C(CH3 )3 ), 1.51 (s, 18H, C(CH3 )3 ), 3.41 (s, 6H, OCH3 ), 4.64 (s, 4H, OCH2 O), 7.45 (d, 2H, arylH), 7.61 (d, 2H, aryl-H), 7.68 − 7.80 (m, 3H, NC5 H3 ). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 31.1 (C(CH3 )3 ), 31.6 (C(CH3 )3 ), 34.8 (C(CH3 )3 ), 35.6 (C(CH3 )3 ), 57.6 (OCH3 ), 99.7 (OCH2 O), 123.2, 125.2, 126.7, 134.1, 136.1, 142.5, 146.1, 151.5, 158.4 (aryl). Synthesis of H2 (ONO) Compound MOM2 (ONO) was suspended in 25 mL of MeOH, and 25 mL of concentrated HCl was added. This mixture was heated at 80◦ C for 5 h. After cooling, volatile materials were removed under vacuum yielding 3.45 g (0.70 mmol, 90% based on O-Methoxymethyl-2-bromo-4,6di-tert-butylphenoxide) of H2 (ONO) as a white powder. 1 H NMR (300 MHz, CDCl3 ): δ 1.39 (s, 18H, C(CH3 )3 ), 1.48 (s, 18H, C(CH3 )3 ), 7.46 (d, 2H, aryl-H), 7.51 (d, 2H, aryl-H), 7.67 (d, 2H, 3 5 −NC5 H−H 2 ), 8.01 (t, 1H, 4- NC5 H2 −H), 10.59 (s, 2H, OH). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 29.8 (C(CH3 )3 ), 31.8 (C(CH3 )3 ), 34.6 (C(CH3 )3 ), 35.6 (C(CH3 )3 ), 120.5, 121.3, 123.0, 126.4, 137.5, 140.0, 141.5, 153.3, 157.6 (aryl). HR-MS: C33 H45 O2 N: Calculated: 487.3450. Measured: 487.3446. 2.5.1.8 Synthesis of MOM2 (ONO tBu ) O-Methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide (6.15 g, 18.7 mmol), was dissolved in 60 mL of THF. tert-butyllithium (2.52 g, 39.4 mmol) was dissolved in 27 mL of pentane and added to the solution of O-methoxymethyl-2-bromo-4,6-di-tert-butylphenoxide at −78◦ C. The mixture was allowed to warm to room temperature and stirred for 1 h. A suspension of ZnCl2 (1.80 g, 13.2 mmol) in THF was then added, and the mixture stirred for another 30 min. A solution of 2,6-dichloro-4tert-butylpyridine (1.72 g, 8.42 mmol) and Pd(PPh3 )4 (218 mg, 189 µmol) in THF was added, the reaction vessel was transferred to a preheated 75◦ C oil bath, and the mixture was stirred for 16 h. The 36 mixture was quenched with 10 mL of water, and the volatiles were removed by rotary evaporation. Another 70 mL of water was then added, and the organic layer was extracted three times with ether, dried with MgSO4 , and evaporated to dryness, giving a white powder, which was purified by washing with cold methanol (2.67 g, 4.23 mmol, 50.3% yield). 1 H NMR (300 MHz, CDCl3 ): δ 1.35 (s, 18H, C(CH3 )3 ), 1.37 (s, 9H, C(CH3 )3 ), 1.49 (s, 18H, C(CH3 )3 ), 3.38 (s, 6H, OCH3 ) 4.60 (s, 4H, OCH2 O), 7.42 (d, 2H, 4 J = 2.6 Hz, aryl-H), 7.56 (d, 2H, 4 J = 2.6 Hz, aryl-H), 7.66 (s, 2H, NC5 H2 ). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 30.6, 30.9, 31.4, 34.6, 34.9, 35.4 (C(CH3 )3 ), 57.3 (OCH3 ), 99.3 (OCH2 O), 120.4, 124.8, 126.6, 134.3, 142.2, 145.8, 151.3, 158.0, 159.7 (aryl). 2.5.1.9 tBu Synthesis of H2 (ONO tBu MOM2 (ONO ) ) (2.67 g, 4.23 mmol) was suspended in a mixture of methanol (40 mL) and con- centrated HCl (40 mL) and heated at 80◦ C for 6 h. The suspension was then cooled to 0◦ C and tBu filtered, and the precipitate, a white powder, was washed with cold MeOH to afford H2 (ONO ) 1 (1.71 g, 3.14 mmol, 74% yield). H NMR (300 MHz, d8 -THF): δ 1.36 (s, 18H, C(CH3 )3 ), 1.46 (br, 27H, C(CH3 )3 ), 7.42 (d, 2H, 4 J = 2.4 Hz, aryl-H), 7.53 (d, 2H, 4 J = 2.7 Hz, aryl-H), 7.81 (s, 2H, NC5 H2 ), 10 (br s, concentration dependent, 2H, OH). 13 C{1 H} NMR (75.4 MHz, CDCl3 ): δ 29.6, 30.6, 31.6, 34.4, 35.3, 35.5 (C(CH3 )3 ), 117.7, 121.7, 122.6, 126.0, 137.2, 141.2, 153.0, 157.2, 164.1 (aryl). MS FAB+: 543.4052 (M+ ). Analysis calculated for C37 H53 NO2 : C, 81.72; H, 9.82; N, 2.58. Found: C, 81.57; H, 10.06; N, 2.55. 2.5.1.10 Synthesis of Na[Ir(cod)ONO], 1 H2 (ONO) (91.2 mg, 187 µmol) was dissolved in 10 mL of THF, and an excess of NaH (24.6 mg, 1.03 mmol) was added. The yellow suspension first turned orange and then yellow again, with vigorous evolution of H2 . The suspension was allowed to stir for one hour, after which it was filtered through Celite and added to a stirring solution of [Ir(cod)Cl]2 (60.0 mg, 89.3 µmol) in 5 mL of THF. After 30 min of stirring the solution was filtered and pumped down, giving 1 as a red powder. Residual solvent was removed by trituration with benzene followed by overnight evacuation under high vacuum. Isolated yield was 148.3 mg (183.3 µmol, 98.0%); the yield was quantitative by NMR. 1 H NMR (300 MHz, d8 -THF): δ 0.84 (m, 4H, sp3 on cod), 1.32 (s, 18H, 2 tBu on ligand phenolate), 1.49 (s, 18H, 2 tBu on ligand phenolate), 1.68 (br m, 4H, sp3 on cod), 2.99 (br d, 4H, J = 2.7 Hz, sp2 on cod), 7.19 (d, 2H, 4 J = 2.7 Hz, ligand aryl), 7.24 (d, 2H, 4 J = 2.6 Hz, ligand aryl), 7.26 (br s, 2H, ligand aryl), 7.66 (t, 1H, 3 J = 8.0 Hz, ligand pyridine 4-position). 13 C{1 H} NMR (75.4 MHz, d8 -THF): δ 14.6, 23.4, 31.5, 32.5, 32.8, 34.7, 36.2, 49.7, 121.3, 123.8, 125.0, 131.9, 135.7, 137.2, 139.2, 159.4, 166.9. MS FAB+: 786.3856 (M+ −Na). 37 2.5.1.11 Synthesis of Pyridine-Activated Complex 3 Na[Ir(cod)(ONO)] (1) (85.1 mg, 105 µmol) was dissolved in 1 mL of C6 H6 . 15-Crown-5 (21 µL, 106 µmol) was mixed in, and the solution was allowed to sit undisturbed for several days. Orange crystals formed and were shown crystallographically to consist of 3. A pure (by NMR) sample was obtained by washing with C6 H6 , redissolving in THF, and drying under vacuum. Isolated yield: 82.7 mg (79.8 µmol, 75.9%). 1 H NMR (500 MHz, d8 -THF): δ 1.35 (s, 9H, tBu on ligand phenolate), 1.42 (s, 9H, tBu on ligand phenolate), 1.51 (br s, 18H, 2 tBu on ligand phenolate), 1.77 (br, 4H, sp3 on cod), 2.18 (br, 4H, sp3 on cod), 3.29 (br, 2H, sp2 on cod), 3.37 (s, 20H, crown ether), 4.46 (br, 2H, sp2 on cod), 7.24 (d, 1H, 4 J = 2.6 Hz, ligand aryl), 7.25 (d, 1H, 4 J = 2.4 Hz, ligand aryl), 7.48 (d, 1H, 3 J = 8.4 Hz, ligand aryl), 7.78 (d, 1H, 4 J = 2.4 Hz, ligand aryl), 7.91 (d, 1H, 3 J = 8.2 Hz, ligand aryl), 8.45 (d, 1H, 4 J = 2.6 Hz, ligand aryl), 18.14 (s, 1H, ligand phenol). 13 C{1 H} NMR (126 MHz, d8 -THF): δ 30.4, 31.1, 31.6, 32.4, 32.9, 33.9, 35.1, 35.4, 36.1, 36.4, 45.1, 70.2, 74.4, 112.5, 120.2, 120.8, 123.0, 124.3, 124.4, 126.9, 135.0, 137.4, 138.5, 138.8, 146.5, 149.7, 152.6, 153.2, 159.7, 162.7. tBu 2.5.1.12 Synthesis of [NaIr(cod)(ONO tBu H2 (ONO )]2 , 4 ) (246.8 mg, 453.8 µmol) was dissolved in 10 mL of THF, and an excess of NaH (64.8 mg, 2.70 mmol) was added. The light yellow solution first turned orange and then yellow again, with vigorous evolution of H2 . The suspension was allowed to stir for an hour, after which it was filtered through Celite. [Ir(cod)Cl]2 (145.2 mg, 216.2 µmol) was dissolved in THF and slowly added to the tBu stirring solution of Na2 (ONO tBu ) in order to prevent formation of overmetalated [Ir(cod)]2 (ONO ), 5. After an hour of stirring the solvent was removed in vacuo, redissolved in hexanes, and filtered, and the solvent removed in vacuo again, giving 4·THF as a red solid with a glassy appearance. The residual THF was removed by adding more hexanes and evacuating to give 4. 4·THF is very soluble in hexanes, whereas 4 is much less so; consequently impurities in 4 can be removed with a hexanes wash. Crystals were obtained by allowing a pentane solution of 4·THF or 4 to evaporate. Isolated yield was 326.4 mg (188.6 µmol, 87.2%). 1 H NMR (300 MHz, d8 -THF): δ 0.83 (br d, 16H, 2 J = 7.5 Hz, sp3 on 2 cod), 1.33 (s, 36H, 4 tBu on ligand phenolate), 1.37 (s, 18H, 2 tBu on ligand pyridine), 1.49 (s, 36H, 4 tBu on ligand phenolate), 2.97 (br s, 8H, sp2 on 2 cod), 7.19 (d, 4H, 4 J = 2.7 Hz, 2 ligand phenolate aryl), 7.23 (d, 4H, 4 J = 3.0 Hz, 2 ligand phenolate aryl), 7.25 (s, 4H, 2 ligand pyridine aryl). 13 C{1 H} NMR (75.4 MHz, d8 -THF): δ 31.6, 32.1, 33.1, 33.5, 35.4, 36.2, 36.9, 50.3, 119.2, 124.4, 125.4, 133.0, 136.2, 139.6, 159.6, 161.2, 167.7. MS FAB+: 842.4478 ([Ir(cod)(ONO tBu (M+ −Na). Analysis calculated for C90 H126 Ir2 N2 Na2 O4 : C, 62.47; H, 7.34; N, 1.62. Found: C, 62.26; H, 7.52; N, 1.35. 38 2.5.1.13 Reaction of 4 with CO tBu [NaIr(cod)(ONO )]2 (179.4 mg, 103.7 µmol) was dissolved in THF in a Schlenk flask. The solution was cooled to −78◦ C, and an atmosphere of CO (about 5 mmol) was introduced into the headspace. The solution was then brought back to room temperature and allowed to stir overnight. Afterward, the solution was pumped down, yielding a yellow-brown solid with IR stretching frequencies at 2057, 2030, and 1977 cm−1 . This solid was reintroduced into the glovebox, redissolved in pentane, and filtered. Allowing the pentane solution to slowly evaporate yielded brown crystals. X-ray diffraction of the crystals revealed their identity to be Ir4 (CO)12 , consistent with the aforementioned IR stretching frequencies. 2.5.1.14 Reaction of 4 with H2 tBu [NaIr(cod)(ONO )]2 (10.2 mg, 5.89 µmol) was dissolved in 0.7 mL of d8 -THF in a J-Young NMR tube. Two drops of benzene were added as an internal standard. The headspace was then charged with 2.9 atm H2 , and the NMR tube continuously inverted for mixing. After 11 h, the solution color had lightened considerably, and an iridium mirror had deposited. 1 H NMR showed quantitative tBu conversion to the protonated ligand H2 (ONO tBu 2.5.1.15 Synthesis of [Ir(cod)]2 (ONO tBu H2 (ONO ). ), 5 ) (57.6 mg, 106 µmol) was dissolved in 5 mL of THF, and an excess of NaH (30.2 mg, 1.26 mmol) was added. The yellow suspension first turned orange and then yellow again, with vigorous evolution of H2 . The suspension was allowed to stir for an hour, after which it was filtered through Celite. A 5 mL THF solution of [Ir(cod)Cl]2 (71.1 mg, 107 µmol) was then added while stirring. After 1 h of stirring, the solvent was removed in vacuo; and the orange compound was redissolved in pentane and filtered. The solvent was then removed in vacuo again, giving 5 as an orange powder. Crystals were obtained by allowing a pentane solution to evaporate. Yield was 119.3 mg (104 µmol, 98.6%). 1 H NMR (300 MHz, d8 -THF): δ 0.81 (br d, 16H, 2 J = 6.9 Hz, sp3 on 2 cod), 1.39 (s, 18H, 2 tBu on ligand phenolate), 1.52 (s, 9H, tBu on ligand pyridine), 1.58 (s, 18H, 2 tBu on ligand phenolate), 3.08 (br s, 8H, sp2 on 2 cod), 7.51 (br d, 4H, J = 5.4 Hz, 2 ligand phenolate aryl), 7.82 (s, 2H, ligand pyridine aryl). 13 C{1 H} NMR (75.4 MHz, d8 -THF): 15.2, 24.1, 31.0, 31.6, 32.65, 32.75, 33.4, 35.1, 35.8, 35.9, 36.9, 38.3, 56.1, 121.7, 124.9, 130.0, 134.9, 139.2, 141.8, 159.3, 163.1, tBu 163.6. MS FAB+: 1141.5018 (M+ with 191Ir), 843.4935 ([Ir(cod)(ONO + )] ). Analysis calculated for C53 H75 Ir2 NO2 : C, 55.71; H, 6.62; N, 1.23. Found: C, 54.81; H, 6.77; N, 1.12. tBu Alternatively, [NaIr(cod)(ONO )]2 (15 mg, 9 µmol) was dissolved in 1 mL of THF. [Ir(cod)Cl]2 (6 mg, 9 µmol) was dissolved in 2 mL of THF and added. After two hours the solution was the solvent was removed in vacuo, redissolved in pentane, and filtered, and the solvent removed in vacuo again, 39 giving 5 (20 mg, 18 µmol, 100% yield) as an orange powder. Characterization data were as reported above. 2.5.1.16 tBu Synthesis of [(ONO tBu [NaIr(cod)(ONO )Ir(cod)(MeCN)][PF6 ], 6 )]2 (106.3 mg, 61.4 µmol) was dissolved in 10 mL of THF. A solution of AgPF6 (59.4 mg, 235 µmol) in acetonitrile was added, resulting in the mixture immediately turning black. The solution was allowed to stir overnight. Afterward, the solvent was removed in vacuo, and the black solid redissolved in benzene, filtered, and lyophilized to give 6 as a yellow powder. This complex was recrystallized by layering pentane onto a benzene solution. Yield after recrystallization was 93.0 mg (90.4 µmol, 73.6%). 1 H NMR (300 MHz, CD3 CN): δ 1.37 (s, 18H, 2 tBu on ligand phenolate), 1.45 (s, 9H, tBu on ligand pyridine), 1.53 (s, 18H, 2 tBu on ligand phenolate), 2.26 (m, 4H, sp3 on cod), 2.42 (s, 3H, MeCN ligand), 2.75 (br s, 2H, sp3 on cod), 4.31 (br s, 2H, sp3 on cod), 5.45 (m, 2H, sp2 on cod), 6.65 (m, 2H, sp2 on cod), 7.41 (two overlapping peaks, s, 4H, 2 ligand phenolate aryl), 7.70 (s, 2H, ligand pyridine aryl). 13 C{1 H} NMR (75.4 MHz, C6 D6 ): δ 1.3, 28.0, 29.3, 29.4, 30.3, 31.2, 33.9, 34.4, 35.5, 100.2, 113.9, 119.9, 120.6, 123.2, 125.8, 126.4, 139.4, 140.7, 154.3, 161.8, 162.1. 19 F{13 C} NMR (282 MHz, C6 D6 ): δ −71.1 (d, 1 JPF = 711 Hz). 31 P{1 H} NMR (122 MHz, C6 D6 ): tBu δ −143.2 (septet, 1 JPF = 711 Hz). MS FAB+: 842.4472 ([Ir(cod)(ONO + )] ). Analysis calculated for C47 H66 F6 IrN2 O2 P · C6 H6 : C, 57.54; H, 6.56; N, 2.53. Found: C, 57.06; H, 6.48; N, 2.42. Alternatively, [NaIr(cod)(ONO tBu )]2 (103.9 mg, 60.0 µmol) was dissolved in 10 mL of THF. A suspension of [FeCp2 ][PF6 ] (75.6 mg, 228 µmol) in acetonitrile was added, the solution was allowed to stir overnight, the solvent was removed in vacuo, and the solid redissolved in benzene, filtered, and lyophilized to give 6. Yield after recrystallization (as above) was 98.7 mg (96.0 µmol, 79.9%). Characterization data were as reported above. 2.5.1.17 tBu Synthesis of (ONO tBu [NaIr(cod)(ONO )Ir(cod)OTf, 7 )]2 (310.0 mg, 179.1 µmol) and AgOTf (175.7 mg, 683.8 µmol) were each dis- solved in about 10 mL of benzene. The AgOTf solution was then added dropwise to the stirring solution of 4. The mixture immediately turned black and was allowed to stir overnight. It was then filtered, pumped down, and washed with pentane, yielding (ONO tBu )Ir(cod)OTf, 7 (259 mg, 261 µmol, 76% yield). Since 7 has limited solubility in benzene (about 10 mg/mL), it can be purified by washing with benzene. Crystals were obtained by layering a saturated benzene solution with pentane. 1 H NMR (300 MHz, C6 D6 ): δ 0.93 (s, 9H, tBu on ligand pyridine), 1.19 (br s, 8H, sp3 on cod), 1.42 (s, 18H, 2 tBu on ligand phenolate), 1.74 (s, 18H, 2 tBu on ligand phenolate), 2.42 (br s, 2H, sp2 on cod), 4.62 (m, 2H, sp2 on cod), 7.35 (s, 2H, ligand pyridine aryl), 7.45 (d, 2H, 4 J = 2.7 Hz, ligand phenolate aryl), 7.58 (d, 2H, 4 J = 2.7 Hz, ligand phenolate aryl). 13 C{1 H} NMR (75.4 MHz, C6 D6 ): δ 27.4, 29.5, 30.0, 30.7, 31.4, 34.0, 34.2, 35.7, 90.7, 112.8, 120.6, 123.3, 125.9, 126.2, 40 139.1, 139.6, 153.2, 160.9, 161.2. 19 F{13 C} NMR (282 MHz, C6 D6 ): δ −77.6. MS FAB+: 991.5552 (M+ ), 858.5524 ([(ONO tBu + )Ir(cod)] ). Analysis calculated for C46 H63 F3 IrNO5 PS: C, 55.74; H, 6.41; N, 1.41. Found: C, 55.51; H, 6.36; N, 1.38. 2.5.1.18 tBu Synthesis of (ONO )Ir(PPh3 )2 OTf, 8 7 (89.4 mg, 90.2 µmol) and PPh3 (47.3 mg, 180 µmol) were each dissolved in benzene and mixed together. The mixture was stirred at 90◦ C for at least 3 h, filtered, and then lyophilized. 8 (117.6 mg, 83.5 µmol, 92.6% yield) was formed as a yellow powder, soluble in benzene and slightly soluble in pentane (about 1 mg/mL). 8 is stable for several days at 90◦ C with little decomposition. 1 H NMR (300 MHz, C6 D6 ): δ 0.94 (s, 9H, tBu on ligand pyridine), 1.25 (s, 18H, 2 tBu on ligand phenolate), 1.35 (s, 18H, 2 tBu on ligand phenolate), 6.52 (s, 2H, ligand pyridine aryl), 6.68 (d, 2H, 4 J = 2.4 Hz, ligand phenolate aryl), 6.87 (br s, 12H, triphenylphosphines), 7.16 (br s, 6H, triphenylphosphines), 7.18 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.63 (br s, 12H, triphenylphosphines). 13 C{1 H} NMR (75.4 MHz, C6 D6 ): δ 29.6, 29.7, 29.9, 31.2, 33.6, 35.7, 119.4, 123.2, 125.0, 125.6, 126−128 (may be obscured by C6 D6 ), 129.6, 134.0, 135.1 (br), 136.7, 140.7, 141.1. 19 F{13 C} NMR (282 MHz, C6 D6 ): tBu δ −76.4. 31 P{1 H} NMR (122 MHz, C6 D6 ): δ −17.2. MS FAB+: 1258.5423 ([Ir(PPh3 )2 (ONO tBu 1145.3984 ([Ir(PPh3 )(ONO + tBu )(OTf)] ), 996.4537 ([Ir(PPh3 )(ONO + )] ), + )] ). Analysis calculated for C74 H81 F3 IrNO5 P2 S: C, 63.14; H, 5.80; N, 1.00. Found: C, 63.24; H, 5.95; N, 0.87. 2.5.1.19 tBu Synthesis of (ONO )Ir(PPh3 )2 OH, 9 8 (20.8 mg, 14.7 µmol) was dissolved in benzene, and 2 µL of H2 O was added via a microliter syringe. The mixture was stirred for one hour, after which proton sponge (3.6 mg, 17 µmol) dissolved in benzene was added. A precipitate was immediately observed, and the mixture was stirred for four hours, pumped down, redissolved in benzene, filtered, and lyophilized, giving 9 (18.8 mg, 14.7 µmol, quantitative yield) as a yellow powder. 9 is slightly soluble in pentane (about 3 mg/mL) and was crystallized by letting a pentane solution evaporate. 1 H NMR (300 MHz, C6 D6 ): δ −1.63 (s, 1 H, IrOH), 1.05 (br, 27H, 2 tBu on ligand phenolate, tBu on ligand pyridine), 1.34 (s, 18H, 2 tBu on ligand phenolate), 6.75 (br, 4H, 2 ligand phenolate aryl), 6.8 − 7.2 (many broad overlapping peaks, 30H, triphenylphosphines), 7.38 (br s, 2H, ligand pyridine aryl). 31 P{1 H} NMR (122 MHz, C6 D6 ): tBu δ −23.2. MS FAB+: 1275.5450 ([Ir(PPh3 )2 (ONO tBu 1013.4492 ([Ir(PPh3 )(ONO + tBu )OH] ), 1258.5423 ([Ir(PPh3 )2 (ONO + )] ), + )OH] ). Analysis calculated for C73 H82 IrNO3 P2 : C, 68.73; H, 6.48; N, 1.10. Found: C, 68.71; H, 6.66; N, 1.10. 2.5.1.20 tBu Synthesis of (ONO )Ir(PPh3 )2 Cl, 10 8 (39.0 mg, 27.7 µmol) was dissolved in 3 mL of dichloromethane and heated to 90◦ C for five hours. The solution was then pumped down, redissolved in benzene, filtered, and lyophilized, giving 10 (35.4 41 mg, 27.4 µmol, 98.7% yield) as a yellow powder. 1 H NMR (300 MHz, CD2 Cl2 ): δ 0.79 (s, 18H, 2 tBu on ligand phenolate), 1.12 (s, 9H, tBu on ligand pyridine), 1.27 (s, 18H, 2 tBu on ligand phenolate), 6.50 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 6.65 (s, 2H, ligand pyridine aryl), 6.93 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 6.9 − 7.3 (many broad overlapping peaks, 30H, triphenylphosphines). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 29.6, 30.4, 31.9, 34.3, 34.9, 35.7, 120.2, 124.5, 125.1, 126.1, 127.4 (br), 127.8 (br), 128.3 (br), 129.7 (br), 133.7 (br), 136.3 (br), 136.8, 141.0, 155.7, 159.1, 170.1. 31 tBu P{1 H} NMR (121 MHz, CD2 Cl2 ): δ −25.1. MS FAB+: 1294.5082 ([Ir(PPh3 )2 (ONO tBu 1257.5300 ([Ir(PPh3 )2 (ONO + tBu )] ), 1031.3262 ([Ir(PPh3 )(ONO + )Cl] ), + )Cl] ). Analysis calculated for C73 H81 ClIrNO2 P2 : C, 67.75; H, 6.31; N, 1.08. Found: C, 68.60; H, 6.84; N, 1.07. 2.5.1.21 tBu Synthesis of (ONO )Ir(PPh3 )2 H, 11 LiOMe (3.9 mg, 0.10 mmol) was dissolved in methanol, and 8 (67.2 mg, 47.7 µmol) was added while stirring. A light yellow filtrate formed overnight and was filtered out. When benzene was added to this filtrate, a black precipitate immediately formed, which was also filtered out. The remaining yellow solution was lyophilized, giving (ONO tBu )Ir(PPh3 )2 H, 11 (37.6 mg, 29.9 µmol, 62.5% yield), as a 1 yellow powder. H NMR (300 MHz, C6 D6 ): δ −15.4 (t, 1H, 2 JPH = 17.8 Hz, IrH), 0.79 (s, 18H, 2 tBu on ligand phenolate), 1.17 (s, 9H, tBu on ligand pyridine), 1.41 (s, 18H, 2 tBu on ligand phenolate), 6.8 − 7.4 (many broad overlapping peaks, 36H, ligand aryl, triphenylphosphine aryl). 13 C{1 H} NMR (75.4 MHz, C6 D6 ): δ 28.4, 30.0, 31.7, 33.7, 34.2, 34.9, 118.6, 123.5, 124.9, 126-128 (may be obscured by C6 D6 ), 128.9 (br), 134.3 (br), 135.2, 154.1, 158.0, 170.3. 31 P{1 H} NMR (122 MHz, C6 D6 ): δ 9.8. tBu MS FAB+: 1260.5502 ([Ir(PPh3 )2 (ONO + tBu )H2 ] ), 997.4542 ([Ir(PPh3 )(ONO + )H] ). 11 was also obtained from the following reactions, although in most cases it was observed only by NMR and could not be isolated cleanly: • 8 (25.7 mg, 18.3 µmol) and proton sponge (3.8 mg, 17.7 µmol) were dissolved in benzene and mixed. The solution was heated to 90◦ C for 24 h. It was then pumped down, redissolved in pentane, filtered, and pumped down again. The yield of 11 was nearly quantitative (by NMR). • 8 (11.3 mg, 7.2 µmol) and dihydroanthracene (1.3 mg, 7.2 µmol) were dissolved in C6 D6 and mixed. CoCp2 (1.4 mg, 7.4 µmol) dissolved in C6 D6 was then added, and the entire mixture transferred to a J-Young NMR tube. After stirring for one hour, NMR revealed the formation of 11 as the major product. • 8 (16.7 mg, 10.7 µmol) was dissolved in approximately 1 mL of C6 D6 in a J-Young NMR tube, and an excess of solid NaH (30 mg, 1 mmol) was added. The NMR tube was heated to 90◦ C, and the reaction was followed by NMR; after 26 h the reaction was complete. The solution was then removed, filtered, and pumped down. Isolated yield was 11.8 mg, 9.4 µmol, 88%. 42 • 8 (38.4 mg, 27.3 µmol) was dissolved in approximately 5 mL of benzene, and ZnEt2 (1.5 µL, 14 µmol) was added via a microliter syringe. The stirring solution first turned red and then yellow and was allowed to stir overnight. The solution was then pumped down, redissolved in pentane, filtered, and pumped down again to form 11 as a clean solid. 2.5.1.22 tBu Synthesis of (ONO )Ir(PPh3 )2 Me, 12 8 (203.8 mg, 144.8 µmol) was dissolved in benzene, and ZnMe2 (6.2 µL, 90 µmol) was added via a microliter syringe. The solution was stirred at room temperature overnight, filtered, and then lyophilized. 12 (182.6 mg, 143.4 µmol, 99.0% yield) was formed as a yellow powder, soluble in benzene and slightly soluble in pentane, and could be recrystallized by layering pentane on a benzene solution. 1 H NMR (300 MHz, C6 D6 ): δ 1.13 (s, 9H, tBu on ligand pyridine), 1.18 (s, 18H, 2 tBu on ligand phenolate), 1.40 (s, 18H, 2 tBu on ligand phenolate), 3.03 (t, 3H, 3 JPH = 6.1 Hz, IrMe), 6.82 (d, 2H, 4 J = 2.4 Hz, ligand phenolate aryl), 6.89 (br s, 12H, triphenylphosphines), 6.92 (s, 2H, ligand pyridine aryl), 7.19 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl), 7.16 (br s, 6H, triphenylphosphines), 7.46 (br s, 12H, triphenylphosphines). 13 C{1 H} NMR (75.4 MHz, CD2 Cl2 ): δ −31.4 (t, 2 JPC = 8.6 Hz), 29.3, 29.9, 31.3, 33.6, 34.3, 35.1, 119.4, 123.3, 125.9, 126.1, 127.1, 128.7, 134.7 (br), 135.1, 140.3, 153.9, 157.8, 170.4. 31 P{1 H} NMR (122 MHz, C6 D6 ): δ −14.6. MS FAB+: 1273.4640 (M+ ), tBu 1011.4714 ([Ir(PPh3 )(ONO + )(Me)] ), 995.3394 ([Ir(PPh3 )(ONO tBu + )] ). Analysis calculated for C74 H84 IrNO2 P2 : C, 69.78; H, 6.65; N, 1.10. Found: C, 69.81; H, 6.53; N, 1.05. 2.5.1.23 tBu Synthesis of (ONO )Ir(PPh3 )2 I, 13 8 (10.5 mg, 8.24 µmol) and I2 (2.1 mg, 8.3 µmol) were each dissolved in benzene and mixed together. The mixture was stirred at room temperature for at least 3 h, filtered, and then lyophilized. 13 (10.8 mg, 7.80 µmol, 94.6% yield) was formed as a yellow, pentane-soluble powder, although impurities can give it a green color. Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (300 MHz, CD2 Cl2 ): δ 0.90 (s, 18H, 2 tBu on ligand phenolate), 1.15 (s, 9H, tBu on ligand pyridine), 1.28 (s, 18H, 2 tBu on ligand phenolate), 6.52 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl), 6.70 (s, 2H, ligand pyridine aryl), 6.96 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.08 (br s, 12H, triphenylphosphines), 7.21 (br s, 12H, triphenylphosphines), 8.16 (br s, 6H, triphenylphosphines). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 30.3, 30.4, 31.8, 34.3, 34.9, 35.7, 120.5, 124.8, 125.1, 126.6, 127.4 (br), 129.8 (br), 133.7 (br), 136.9, 137.3 (br), 141.2, 154.8, 159.2, 170.4. 31 P{1 H} NMR (122 tBu MHz, C6 D6 ): δ −29.1. MS FAB+: 1386.5627 (M+ ), 1123.3712 ([Ir(PPh3 )(ONO 2.5.1.24 tBu Synthesis of (ONO + )(I)] ). )Ir(PEt3 )2 Me, 14 12 (79.4 mg, 62.3 µmol) was dissolved in 5 mL of benzene, and PEt3 (20 µL, 136 µmol) was added. The solution was allowed to stir for seven hours, after which the benzene and excess PEt3 were 43 removed by lyophilization. The yellow powder obtained was redissolved in 5 mL of dichloromethane, and [Rh(cod)2 ]OTf (30.5 mg, 62.7 µmol) dissolved in 3 mL of acetonitrile was added to complex residual PEt3 . The solution was allowed to stir for three hours. Afterward, it was pumped down, and the resulting 14 was redissolved in hexanes, filtered to remove the byproducts and impurities, and pumped down again. Isolated yield was 51.4 mg (52.2 µmol, 83.7%). Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (300 MHz, C6 D6 ): δ 0.75 (m, 18H, 6 P−CH2 −CH 3 ), 1.16 (s, 9H, tBu on ligand pyridine), 1.46 (s, 18H, 2 tBu on ligand phenolate), 1.65 (m, 33H, 2 tBu on ligand phenolate, 6 P−CH 2 −CH3 , IrMe), 7.34 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl), 7.43 (s, 2H, ligand pyridine aryl), 7.53 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl). 13 C{1 H} NMR (126 MHz, C6 D6 ): δ −33.4 (t, JCP = 8.2 Hz), 8.1 (t, JCP = 1.5 Hz), 13.3 (t, JCP = 14.3 Hz), 30.6, 31.3, 32.4, 34.4, 35.3, 36.3, 118.7, 124.0, 125.6, 128.1, 136.2, 140.6, 156.3, 159.3, 172.9 (t, J = 2.2 Hz). 31 P{1 H} tBu NMR (122 MHz, C6 D6 ): δ −19.2. MS FAB+: 985.5609 (M+ ), 867.4800 ([Ir(PEt3 )(ONO + )(Me)] ). Analysis calculated for C50 H84 IrNO2 P2 : C, 60.95; H, 8.59; N, 1.42. Found: C, 61.18; H, 8.32; N, 1.45. 2.5.1.25 tBu Synthesis of (ONO )Ir(PEt3 )2 I, 15 14 (11.8 mg, 12.0 µmol) and I2 (3.0 mg, 12.0 µmol) were each dissolved in benzene and mixed together. The mixture was stirred at room temperature for 5 h, filtered, and then lyophilized. 15 was formed as a yellow, pentane-soluble powder. Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (300 MHz, C6 D6 ): δ 0.74 (m, 18H, 6 P−CH2 −CH 3 ), 1.09 (s, 9H, tBu on ligand pyridine), 1.42 (s, 18H, 2 tBu on ligand phenolate), 1.71 (s, 18H, 2 tBu on ligand phenolate), 1.96 (m, 12H, 6 P−CH 2 −CH3 ), 7.36 (s, 2H, ligand pyridine aryl), 7.37 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.46 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl). 13 C{1 H} NMR (126 MHz, C6 D6 ): δ 8.4 (t, JCP = 1.6 Hz), 15.0 (t, JCP = 15.3 Hz), 30.0, 31.0, 31.9, 34.1, 34.9, 35.8, 118.8, 124.5, 124.7, 126.5, 137.2, 141.3, 156.3 (t, JCP = 1.4 Hz), 159.9, 171.7 (t, JCP = 2.2 Hz). 31 P{1 H} NMR (122 MHz, C6 D6 ): δ −26.3. The reaction of 14 and I2 in dichloromethane gives a new species along with 15 and MeI; over the course of four days all of the 15 is converted to the new species. Crystals were obtained after the contents of the NMR tube were allowed to evaporate. X-ray crystallography, NMR, and MS tBu support identification of this compound as the analogous chloride, (ONO )Ir(PEt3 )2 Cl. 1 H NMR (300 MHz, CD2 Cl2 ): δ 0.74 (m, 18H, 6 P−CH2 −CH 3 ), 1.01 (m, 12H, 6 P−CH 2 −CH3 ), 1.33 (s, 18H, 2 tBu on ligand phenolate), 1.37 (s, 18H, 2 tBu on ligand phenolate), 1.38 (s, 9H, tBu on ligand pyridine), 7.11 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.26 (d, 2H, 4 J = 2.5 Hz, ligand phenolate aryl), 7.31 (s, 2H, ligand pyridine aryl). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 7.7, 13.0 (t, JCP = 15.1 Hz), 30.5, 30.6, 31.9, 34.4, 35.5, 35.9, 119.4, 124.6, 124.8, 126.6, 137.6, 140.3, 157.0, 160.5, tBu 171.7. 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −20.7. MS FAB+ 970.5417 ([Ir(PEt3 )2 (ONO tBu 887.3802 ([Ir(PEt3 )(ONO + tBu )(Cl)] ), 852.4370 ([Ir(PEt3 )(ONO + )] ). + )] ), 44 2.5.1.26 tBu Synthesis of [(ONO )Ir(PEt3 )2 Me]PF6 , 16 14 (31.8 mg, 32.3 µmol) was dissolved in dichloromethane, and a dichloromethane suspension of AgPF6 (8.1 mg, 32 µmol) was added. The mixture was allowed to stir for one hour before being pumped down. It was then redissolved in a small amount of benzene and filtered. The solution was then layered with pentane and allowed to sit overnight. Dark blue needle-like crystals of 16·2 C6 H6 (39.1 mg, 30.4 µmol, 94.1% yield) were recovered. 1 H NMR (300 MHz, CD2 Cl2 ): δ −3.1 (very broad), −2.0 (broad), 0.73, 3.99, 12.0 (very broad). 19 F{13 C} NMR (282 MHz, CD2 Cl2 ): δ −73.8 (d, 1 JPF = 710 Hz). 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −144.7 (septet, 1 JPF = 710 Hz). MS tBu FAB+: 985.5561 (M+ ), 867.4697 ([Ir(PEt3 )(ONO 2.5.1.27 tBu Synthesis of (ONO + )(Me)] ). )Ir(PPh3 )2 Ph, 17 12 (20.4 mg, 16.0 µmol) was dissolved in 10 mL of benzene and stirred in a 65◦ C oil bath for 24 h. The solution was reintroduced into the glovebox, and the solvent was removed. The residue was redissolved in pentane and filtered, and the solution pumped down to obtain 17 (20.9 mg, 15.6 µmol, 98% yield). Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (300 MHz, CD2 Cl2 ): δ 0.70 (s, 18H, 2 tBu on ligand phenolate), 1.20 (s, 9H, tBu on ligand pyridine), 1.43 (s, 18H, 2 tBu on ligand phenolate), 5.83 (br s, 2H, IrPh), 6.54 (br s, 1H, IrPh), 6.62 (d, 2H, 4 J = 2.7 Hz, ligand phenolate aryl), 6.78 (s, 2H, ligand pyridine aryl), 6.8 − 7.1 (multiple broad peaks, 30H, triphenylphosphines), 7.16 (d, 2H, 4 J = 2.6 Hz, ligand phenolate aryl), 7.29 (br s, 2H, IrPh). 13 C{1 H} NMR (126 MHz, C6 D6 ): δ 30.3, 30.6, 32.2, 34.2, 34.4, 35.9, 120.4, 123.6, 124.1, 125.6, 125.9, 126.9, 127.0, 129.6, 129.7, 133.3 (br), 135.7, 136.5 (br), 141.2, 145.9, 155.3, 157.7, 171.5. 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −13.2. 2.5.1.28 Reaction of 12 with Toluene 12 (7.8 mg, 6.1 µmol) was dissolved in 700 µL of toluene-d8 in a J-Young NMR tube. The tube was heated at 65◦ C and periodically monitored by 1 H NMR until all starting material signals had disappeared, which took six days; the final 1 H NMR spectrum was complex. The mass spectrum of the solid obtained by evaporation exhibited a signal indicating the presence of the analogous tolyl complex (1094.5436, [Ir(PPh3 )(ONO tBu + + )(C6 D4 CD3 ] , [M−PPh3 ] ), while the 31 P{1 H} NMR of the dissolved solid showed two approximately equal signals at δ −12.2 and −12.3, tentatively assigned to meta and para isomers. A much weaker signal at −12.8 may indicate formation of a very small amount of ortho isomer. Isolation of a pure product was not achieved. 45 2.5.1.29 tBu Synthesis of (ONO )Ir(PPh3 )[κ2 -(o-C6 H4 )PPh2 ], 18 12 (8.2 mg, 6.4 µmol) was dissolved in 1 mL of CD2 Cl2 in a J-Young NMR tube and submerged in a 65◦ C oil bath for 21 h. After NMR confirmation that the reaction was complete, the solvent was removed to obtain 18 (7.7 mg, 6.1 µmol, 95% isolated yield). Crystals were obtained by allowing a pentane solution to evaporate. 1 H NMR (500 MHz, CD2 Cl2 ): δ 0.47 (s, 18H, 2 tBu on ligand phenolate), 1.27 (s, 18H, 2 tBu on ligand phenolate), 1.34 (s, 9H, tBu on ligand pyridine), 6.64 (m, 4H, triphenylphosphines), 6.9 − 7.4 (many multiplets, 23H, triphenylphosphines), 7.50 (dd, 3 J = 4.4 Hz, 3 J = 6.6 Hz, 2H, triphenylphosphines). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 29.58, 30.76, 32.11, 34.24, 35.27, 35.32, 119.51, 119.69, 122.48, 122.55, 123.83, 124.00, 125.60, 125.77, 128.32, 128.39, 129.03, 129.54, 133.76, 133.83, 136.03, 139.93, 140.51, 140.61, 152.80, 153.23, 154.77, 159.36, 170.44. 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −69.5 (d, 2 JPP = 448 Hz), −2.5 (d, 2 JPP = 448 Hz). tBu MS FAB+: 1257.5292 (M+ ), 995.4405 ([Ir(PPh3 )(ONO 2.5.1.30 tBu Synthesis of [(ONO + )] ). )Ir(PPh3 )(CH2 CN)]2 , 19 12 (13.1 mg, 10.3 µmol) was dissolved in 1 mL of CD2 Cl2 in a J-Young NMR tube, and 40 µL of acetonitrile was added. The tube was then submerged in a 65◦ C oil bath for 72 h. After NMR confirmation that the reaction had completed, [Rh(cod)2 ]OTf (2.6 mg, 5.3 µmol) dissolved in 1 mL of acetonitrile was added. The solution was allowed to stir for three hours. Afterward, it was pumped down, and the resulting 19 was redissolved in hexanes, filtered to remove the byproducts and impurities, and pumped down again. Crystals were obtained by allowing a pentane solution to evaporate. Isolated yield was 91% (9.7 mg, 4.7 µmol). 1 H NMR (500 MHz, CD2 Cl2 ): δ 1.23 (s, 9H, 2 tBu on ligand), 1.25 (s, 9H, 2 tBu on ligand), 1.43 (s, 9H, 2 tBu on ligand), 1.44 (s, 9H, 2 tBu on ligand), 1.55 (s, 9H, 2 tBu on ligand), 1.87 (s, 2H, Ir−CH2 −CN), 6.4-6.5 (m, 3H, triphenylphosphine), 6.7-6.8 (m, 3H, triphenylphosphine), 6.9-7.1 (m, 6H, triphenylphosphine),7.2-7.5 (m, 6H, triphenylphosphine), 7.6-7.8 (m, 2H, triphenylphosphine), 7.9-8.1 (m, 1 H, triphenylphosphine). 13 C{1 H} NMR (126 MHz, CD2 Cl2 ): δ 3.4, 30.1, 30.3, 30.5, 31.9, 32.2, 34.2, 34.6, 35.2, 35.7, 35.9, 116.0, 118.0, 120.2, 122.5, 123.5, 125.6, 126.0, 128 − 130 (br), 135.0, 138.0, 138.7, 141.1 (d, JCP = 3.8 Hz), 153.9, 154.4, 158.2, 159.8 (d, JCP = 2.7 Hz), 161.1, 167.6 (d, JCP = 3.4 Hz), 170.8 (d, JCP = 1.0 Hz). 31 P{1 H} NMR (122 MHz, CD2 Cl2 ): δ −88.3. 2.5.1.31 Kinetic Experiments Kinetic experiments were carried out with 8.8 mM solutions of 12 or 14. In a typical example, 7.8 mg (6.1 µmol) 12 was dissolved in 700 µL of C6 D6 inside a nitrogen-filled glovebox, and the solution was transferred to a J-Young NMR tube. The NMR tube was then removed from the glovebox and fully submerged in a 65◦ C ethylene glycol bath. The reactions were periodically halted by transferring the 46 Entrya a b c d e f g h i j k l m eq PPh3 b 0.000 0.125 0.250 0.375 0.500 0.500 0.625 0.750 0.750 0.875 1.000 1.125 1.250 [PPh3 ]c 0 1.1 2.2 3.3 4.4 4.4 5.4 6.5 6.5 7.6 8.7 9.8 10.9 d kobs 2.18 × 10−4 1.54 × 10−5 1.19 × 10−5 6.27 × 10−6 5.28 × 10−6 5.77 × 10−6 4.14 × 10−6 2.58 × 10−6 3.14 × 10−6 2.32 × 10−6 1.88 × 10−6 1.90 × 10−6 1.70 × 10−6 R2 0.9937 0.9965 0.9842 0.9964 0.9910 0.9921 0.9877 0.9768 0.9912 0.9904 0.9821 0.9558 0.9777 Table 2.1. Summary of the results of Figure 2.23, used to create Figure 2.21. a) The particular run is detailed in the corresponding section in Figure 2.23; b) equivalents of PPh3 added; c) concentration of added PPh3 in mM. d) the kobs value, in s−1 , calculated by fitting the curve to the first-order rate equation [12] = [12]0 exp(−kobs t). NMR tube into an ice bath before NMR measurements. The progress of the reactions were measured by 31 P NMR integration, with NOE turned off and the d1 time set to 22.895 s (12 has the longest relaxation time, 4.579 s, of all the relevant iridium compounds). For PPh3 inhibition experiments, the Ir complex was mixed with various amounts of a 20 mg/mL stock solution of PPh3 in C6 D6 , and additional C6 D6 was added to bring the volume up to 700 µL. 2.5.2 Kinetics Details Figure 2.21 was constructed from the data in table 2.1. The entries in this table were obtained from individual runs detailed in Figure 2.23. 2.5.3 Crystallographic Data Crystallographic data for the complexes 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, and tBu (ONO )Ir(PEt3 )2 Cl are found in the following tables 2.2, 2.3, 2.4, 2.5, and 2.6. 2.6 Acknowledgments We thank Dr. Michael W. Day and Lawrence M. Henling (Caltech) for X-ray crystallographic studies. The Bruker KAPPA APEXII X-ray diffractometer was purchased via an NSF CRIF:MU award to the California Institute of Technology, CHE-0639094. This work was supported by BP in the MC2 and XC2 programs. 47 Figure 2.23. Conversion of 12 to 17-d6 in C6 D6 with varying amounts of extra PPh3 added. [12]0 = 8.7 mM. a) [PPh3 ]0 = 0 mM, b) [PPh3 ]0 = 1.1 mM, c) [PPh3 ]0 = 2.2 mM, d) [PPh3 ]0 = 3.3 mM, e) [PPh3 ]0 = 4.4 mM, f) [PPh3 ]0 = 4.4 mM, g) [PPh3 ]0 = 5.4 mM, h) [PPh3 ]0 = 6.5 mM, i) [PPh3 ]0 = 6.5 mM, j) [PPh3 ]0 = 7.6 mM, k) [PPh3 ]0 = 8.7 mM, l) [PPh3 ]0 = 9.8 mM, m) [PPh3 ]0 = 10.9 mM. a b c d e f g h i j k l m 48 Empirical formula Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.2. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari 4 C90 H126 N2 O4 Ir2 ·2 Na · C4 H10 O 1804.43 Monoclinic 5 C53 H75 NO2 Ir2 1142.54 Monoclinic 6 [C47 H66 N2 O2 Ir] [PF6 ] · C6 H6 1106.30 Triclinic 7 C46 H63 F3 NO5 SIr ·1.5(C6 H6 ) 1108.40 Triclinic 29.8716(11) 14.4205(5) 24.4163(15) 90 123.7310(10) 90 8747.0(7) C2/c 4 1.370 3.100 100(2) 9.5497(4) 22.3182(9) 22.9384(10) 90 92.086(2) 90 4885.7(4) P 21 /n 4 1.553 5.481 100(2) 11.0592(5) 11.8291(5) 20.4215(9) 105.154(2) 97.156(2) 95.370(2) 2536.27(19) P1 2 1.449 2.727 100(2) 11.7099(4) 13.1880(5) 17.2153(6) 86.059(2) 72.263(2) 87.034(2) 2524.87(16) P1 2 1.458 2.744 100(2) 1.66 to 35.19 15255 1.82 to 35.82 19264 1.80 to 40.84 26070 1.55 to 43.59 34153 485 3.747 0.003 533 1.885 0.007 874 1.957 0.016 883 1.536 0.015 0.0314; 0.0719 0.0355; 0.0535 0.0206; 0.0392 0.0221; 0.0381 Semi-empirical from equivalents 0.7469; 0.5672 2.370; −1.617 Semi-empirical from equivalents 0.7470; 0.5076 2.894; −2.135 Semi-empirical from equivalents 0.7480; 0.5945 2.557; −1.524 Semi-empirical from equivalents 0.7486; 0.6139 2.541; −1.556 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 49 Empirical mula for- Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.3. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari 8 C74 H81 F3 NO5 P2 SIr · 0.53(C5 H12 ) ·1.47(C6 H6 ) 1560.64 Monoclinic 9 C73 H82 NO3 P2 Ir 10 C73 H81 NO2 P2 Cl Ir · 5 (CH2 Cl2 ) 11 C73 H82 NO2 P2 Ir ·C5 H12 1275.54 Monoclinic 1718.61 Monoclinic 1331.68 Monoclinic 12.9942(5) 22.9825(9) 25.8969(9) 90 103.513(2) 90 7519.7(5) P 21 /n 4 1.379 1.906 100(2) 13.0223(4) 21.0636(7) 22.7792(8) 90 99.1960(10) 90 6167.9(4) P 21 /n 4 1.374 2.265 100(2) 19.3956(9) 13.1877(6) 31.5557(15) 90 106.611(2) 90 7734.6(6) P 21 /c 4 1.476 2.194 100(2) 16.8289(8) 23.8694(10) 17.4749(8) 90 102.373(2) 90 6856.5(5) P 21 /n 4 1.290 2.040 100(2) 1.84 to 31.76 18398 2.41 to 29.67 14018 1.35 to 37.99 32658 1.47 to 33.55 19762 966 2.750 0.002 1049 1.713 0.004 871 2.152 0.011 1086 2.140 0.004 0.0463; 0.0958 0.0238; 0.0373 0.0340; 0.0589 0.0310; 0.0519 Semi-empirical from equivalents 0.7463; 0.6010 3.915; −2.838 Semi-empirical from equivalents 0.7459; 0.6381 2.894; −0.696 Semi-empirical from equivalents 0.7474; 0.6430 3.136; −2.303 Semi-empirical from equivalents 0.7466; 0.6219 2.907; −2.314 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 50 Empirical formula Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.4. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari 12 C74 H84 NO2 P2 Ir 14 C50 H84 NO2 P2 Ir 1273.56 Monoclinic 13 C73 H81 INO2 P2 Ir ·2 (C5 H12 ) 1529.72 Monoclinic 985.32 Monoclinic 15 0.67(C49 H80 INO2 P2 Ir); 0.33(C50 H83 NO2 P2 Ir) 1097.19 Monoclinic 13.1981(7) 20.9991(9) 22.8801(12) 90 100.388(3) 90 6237.2(5) P 21 /n 4 1.356 2.239 100(2) 14.4250(5) 19.4180(6) 26.3735(9) 90 99.362(2) 90 7288.9(4) P 21 /c 4 1.394 2.346 100(2) 12.2897(7) 35.3664(19) 11.4244(6) 90 93.830(2) 90 4954.4(5) P 21 /c 4 1.321 2.796 100(2) 12.2460(6) 35.7358(18) 11.4544(6) 90 93.490(2) 90 6856.5(5) P 21 /c 4 1.457 3.386 100(2) 1.66 to 29.33 10617 1.57 to 28.67 15318 1.76 to 27.89 10841 1.87 to 40.01 25951 737 1.268 0.003 848 2.742 0.002 527 1.824 0.003 537 1.448 0.010 0.0387; 0.0478 0.0482; 0.0750 0.0188; 0.0366 0.0344; 0.0620 Semi-empirical from equivalents 0.7459; 0.6362 1.724; −1.487 Semi-empirical from equivalents 0.4314; 0.3230 1.880; −2.150 Semi-empirical from equivalents 0.7456; 0.6727 0.633; −0.778 Semi-empirical from equivalents 0.7479; 0.6295 2.448; −2.844 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 51 Empirical formula Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.5. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari 16 [C50 H84 NO2 P2 Ir] [PF6 ] · 1.5(C6 H6 ) 1249.08 Triclinic 17 C79 H86 NO2 P2 Ir ·2 (C6 H6 ) 745.92 Orthorhombic 18 C73 H80 NO2 P2 Ir 1257.52 Monoclinic 19 C114 H136 Li2 N4 O4 P2 Ir2 2086.49 Triclinic 21.3366(12) 21.4842(13) 26.6266(15) 81.836(4) 85.844(3) 89.459(4) 12050.2(12) P1 8 1.377 2.354 100(2) 16.0771(8) 20.5290(10 23.1672(11) 90 90 90 7646.3(6) P bcn 8 1.296 1.837 100(2) 41.3489(19) 18.8871(9) 20.6067(9) 90 111.731(2) 90 14949.3(12) C2/c 8 1.117 1.867 100(2) 16.4557(8) 17.2824(8) 21.7560(10) 93.951(2) 96.184(3) 101.190(3) 6008.7(5) P1 2 1.153 2.284 100(2) 1.66 to 27.99 37498 1.76 to 37.38 13314 1.47 to 32.01 20406 2.34 to 34.09 30107 2386 1.780 0.003 417 3.326 0.001 758 2.133 0.008 1176 2.866 0.007 0.0739; 0.1072 0.0443; 0.0804 0.0311; 0.0569 0.0335; 0.0589 Semi-empirical from equivalents 0.8340; 0.6254 2.889; −2.188 Semi-empirical from equivalents 0.7473; 0.6479 4.097; −2.794 Semi-empirical from equivalents 0.7463; 0.6478 1.457; −0.873 Semi-empirical from equivalents 0.7461; 0.6684 1.933; −1.882 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 52 tBu Empirical formula Formula weight Crystal structure Lattice parameters a (Å) b (Å) c (Å) α (◦ ) β (◦ ) γ (◦ ) Volume (Å3 ) Space group Z value ρcalc (g/cm3 ) µ (mm−1 ) Temperature (K) θ range (◦ ) No. obs. (I > 2σ(I)) No. parameters Goodness of fit* Max. shift in cycle Residuals*: R, Rw Absorption correction max/min Largest peak & hole (e– /Å3 ) Table 2.6. * R = qP 2 hkl w(|Fobs |−|Fcalc |) ndata −nvari (ONO )Ir(PEt3 )2 Cl C49 H81 ClNO2 P2 Ir 1005.74 Monoclinic 12.2623(6) 35.4340(17) 11.3773(6) 90 93.368(2) 90 4934.9(4) P 21 /c 4 1.354 2.861 100(2) 1.88 to 39.87 26158 526 2.079 0.003 0.0371; 0.0602 Semi-empirical from equivalents 0.7478; 0.5515 2.623; −2.981 P ||Fobs |−|Fcalc || hklP hkl |Fobs | . ; Rw = rP |−|F |)2 hkl w(|F P obs 2 calc hkl wFobs , w = 1/σ 2 (Fobs ); GOF = 53 2.7 References [1] (a) Labinger, J. A.; Bercaw, J. E. Nature 2002, 417, 507–514; (b) Crabtree, R. H. J. Chem. Soc., Dalton Trans. 2001, 2437–2450; (c) Labinger, J. A. Fuel Proc. Technol. 1995, 42, 325–338; (d) Labinger, J. A. J. Mol. Catal. 2004, 220, 27–35. [2] (a) Fekl, U.; Goldberg, K. I. Adv. Inorg. Chem. 2003, 54, 259–320; (b) Lersch, M.; Tilset, M. Chem. Rev. 2005, 105, 2471–2526. [3] (a) Goldshlegger, N. F.; Tyabin, M. B.; Shilov, A. E.; Shteinman, A. A. Zh. Fiz. Khim. 1969, 43, 2174–2175; (b) Goldshlegger, N. F.; Eskova, V. V.; Shilov, A. E.; Shteinman, A. A. Zh. Fiz. Khim. 1972, 46, 1353–1354; (c) Shilov, A. E.; Shul’pin, G. B. Activation and Catalytic Reactions of Saturated Hydrocarbons in the Presence of Metal Complexes; Kluwer Academic Publishers: Dordrecht, 2000; pp. 259– 317. [4] Periana, R. A.; Taube, D. J.; Evitt, E. R.; Loffler, D. G.; Wentrcek, P. R.; Voss, G.; Masuda, T. Science 1993, 259, 340–343. [5] Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154–162. [6] (a) Wong-Foy, A. G.; Bhalla, G.; Liu, X. Y.; Periana, R. A. J. Am. Chem. Soc. 2003, 125, 14292–14293; (b) Oxgaard, J.; Muller, R. P.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2004, 126, 352–363; 54 (c) Bhalla, G.; Liu, X. Y.; Oxgaard, J.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 11372–11389; (d) Tenn, W. J.; Young, K. J. H.; Bhalla, G.; Oxgaard, J.; Goddard, W. A.; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 14172–14173; (e) Bhalla, G.; Periana, R. A. Angew. Chem., Int. Ed. 2005, 44, 1540–1543; (f) Young, K. J. H.; Mironov, O. A; Periana, R. A. Organometallics 2007, 26, 2137–2140; (g) Young, K. J. H.; Oxgaard, J.; Ess, D. H.; Meier, S. K.; Stewart, T.; Goddard, W. A.; Periana, R. A. Chem. Commun. 2009, 3270–3272. [7] (a) Agapie, T.; Day, M. W.; Bercaw, J. E. Organometallics 2008, 27, 6123–6142; (b) Agapie, T.; Henling, L. H.; DiPasquale, A. G.; Rheingold, A. L.; Bercaw, J. E. Organometallics 2008, 27, 6245–6256. [8] Weinberg, D. R.; Hazari, N.; Labinger, J. A.; Bercaw, J. E. Organometallics 2010, 29, 89–100. [9] The synthesis of H2 (ONO) was first reported in: Agapie, T. Ph.D. Thesis, California Institute of Technology, Pasadena, CA, 2007, pp. 156–159, 214–215. [10] Danopoulos, A. A.; Winston, S.; Hursthouse, M. B. J. Chem. Soc., Dalton Trans. 2002, 3090– 3091. [11] Hughes, R. P.; Kovacik, I.; Lindner, D. C.; Smith, J. M.; Willemsen, S.; Zhang, D.; Guzei, I. A.; Rheingold, A. L. Organometallics 2001, 20, 3190–3197. [12] Recent calculations have indicated that β-hydride elimination from some methoxide complexes may be energetically disfavored: Theophanis, P. L.; Goddard, W. A. Organometallics 2011, 30, 4941–4948, although there are many examples in the literature. [13] Lu, Z; Jun, C-H; de Gala, S. R.; Sigalas, M. P.; Eisenstein, O; Crabtree, R. H. Organometallics 1995, 14, 1168–1175. [14] The crystal structure included a complex solvent region that could not be satisfactorily modeled and also showed the presence of Li+ ions, presumably carried along as impurities from an earlier step in the overall synthetic sequence; the required charge-balancing counteranions could not be located, probably because they lie within the solvent region. However, determination of the molecular structure of the Ir2 complex 19 was unproblematic. [15] A somewhat related situation has been described previously: Luecke, H. F.; Bergman, R. G. J. Am. Chem. Soc. 1997, 119, 11538–11539. The cyclometalated IrIII complex Cp*(OTf)Ir(κ2 -PPh2 (o-C6 H4 )), obtained from the room-temperature reaction of 55 Cp*(OTf)Ir(PPh3 )Me in CH2 Cl2 , reacts with benzene to form Cp*(OTf)Ir(PPh3 )Ph. Here however the intermolecular activation product is thermodynamically preferred; probably the difference between the two systems is primarily of steric origin, as both include ligands (tertbutyl substituents on ONO; the large Cp* group) that can lead to severe crowding. [16] Golden, J. T.; Andersen, R. A.; Bergman, R. G. J. Am. Chem. Soc. 2001, 123, 5837–5838. [17] (a) Stour, D. L.; Zaric, S.; Niu, S. Q.; Hall, M. B. J. Am. Chem. Soc. 1996, 118, 6068–6069; (b) Su, M. D.; Chu, S. Y. J. Am. Chem. Soc. 1997, 119, 5373–5383. [18] Pangborn, A. B.; Giardello, M. A.; Grubbs, R. H.; Rosen, R. K.; Timmers, F. J. Organometallics 1996, 15, 1518–1520. [19] Ko, P.-H.; Chen, T.-Y.; Zhu, J.; Cheng, K.-F.; Peng, S.-M.; Che, C.-M. J. Chem. Soc., Dalton Trans. 1995, 2215–2220.  56 Chapter 3 DFT virtual screening identifies rhodium-amidinate complexes as active homogeneous catalysts for methane to methanol oxidation Reproduced in part with permission from Fu, R.; Nielsen, R. J.; Goddard, W. A. III; Fortman, G. C.; Gunnoe, T. B. ACS Catalysis, submitted for publication. Unpublished work c 2014 American Chemical Society. Also includes original work. 3.1 Abstract In the search for new organometallic catalysts for the low temperature selective conversion of CH4 to CH3 OH, we apply quantum mechanical virtual screening to select the optimum combination of ligand and solvent on rhodium to achieve low barriers for CH4 activation and functionalization to recommend for experimental validation. We report quantum mechanical predictions (including implicit and explicit solvation) of the mechanisms for various bidentate and tridentate RhIII complexes to catalytically activate and functionalize methane, using trifluoroacetic acid (TFAH) or F water as a solvent. Our most notable results include the design of a RhIII (NN ) ((NNF = bis(N pentafluorophenyl) pentafluorobenzylamidinate) complex, with a methane activation transition state barrier of ∆G‡ = 27.6 kcal/mol in TFAH, and a RhIII (bisq) ((bisq) = bis(quinolinyl)benzene) complex, with a methane activation transition state barrier of ∆G‡ = 33.4 kcal/mol in TFAH. To close the catalytic cycle, the functionalization of methylrhodium intermediates was also investigated, involving carbon-oxygen bond formation via SN 2 attack by solvent, or SR 2 attack by a vanadium oxo. In addition, we have found a correlation between CH4 activation barriers and rhodium-methyl bond energies that allow us to predict the activation transition state energies for future ligands as well. 57 3.2 Introduction The facile, selective, and direct conversion of methane into methanol has long been a goal of industrial chemists [1]. Molecular compounds in solution have long been investigated as potential catalysts for this transformation, as they are typically well-defined, easy to characterize and model, and operate under relatively mild conditions. Whereas much previous work has been done on Pt and Pd catalysts [2, 3], in this chapter we focus on rhodium due to its well-documented nature as an effective C−H activating metal [4] and whose lower electronegativity may allow it to avoid poisoning by coordinating media. Our initial calculations found that using neutral tridentate pincer ligands (L3 ) resulted in neutral L3 Rh(TFA)3 resting states. In order for such complexes to activate methane, a TFA ligand must be both protonated and removed to create an open coordination site. The frequent result was that the overall methane activation energy (i.e. energy of the transition state minus energy of the resting state) was too high to be feasible. Thus, to provide the best candidates for experiments, we shifted our attention to bidentate and monoanionic ligands, which are expected to favor RhIII states that would incorporate a labile, protonated TFAH solvent molecule, thus removing the extra energy penalty for protonation. Bidentate and tridentate ligands were chosen to facilitate redox processes which interconvert square-planar and octahedral coordination environments. Thus we initially examined four classes of ligands that we thought might be prove effective while likely not too hard to synthesize: • bis(N -phenyl)benzylamidinate (NN); • (N -phenyl)acetaldiminyl quinolate (ONN); • bis(pyridyl)(m)ethanesulfonate (DP M E S); • bis(pyrrolyl)quinolinyl phosphine (PN); and two solvents: water and trifluoroacetic acid, due to their pH range and oxidative stability. We then embarked on quantum mechanical (QM) virtual screening to select the optimum combinations of ligand and solvent to recommend for experimental validation. Here we used density functional theory (DFT) at the B3LYP and M06 levels, including both implicit and explicit solvation, for systematic searches over possible reactions mechanisms for CH4 activation and for functionalization. We found that rhodium complexed with the bidentate bis(N -phenyl) benzylamidinate (NN) ligand was the most promising in our initial screen. We then designed a new ligand, bis(N -pentafluorophenyl) pentafluorobenzylamidinate (NNF ), essentially an electron-poor version of (NN), that we found to be even more promising, with lower transition state barriers for both methane activation and Rh−Me functionalization. In addition, we also designed more and less electron-donating variations of the (ONN) ligand; and broadened our investigation to a fifth ligand class, bis(quinolinyl)benzene (bisq), that was expected to have further reduced functionalization barriers. 58 Our efforts culminated in two complexes, F • RhIII (bis(N -pentafluorophenyl) pentafluorobenzylamidinate), denoted as RhIII (NN ); and • RhIII (bis(quinolinyl) benzene), denoted as RhIII (bisq); that can catalytically activate and functionalize methane, using TFAH or water as a solvent. The F first case, RhIII (NN ), leads to a transition state barrier of ∆G‡ = 27.6 kcal/mol at 298 K for methane activation in TFAH (35.0 kcal/mol in water), the lowest we have found using RhIII . The barrier for functionalization is ∆G‡ = 36.8 kcal/mol at 298 K for TFAH (29.7 kcal/mol in water). The second case, RhIII (bisq), leads to a transition state barrier of ∆G‡ = 33.4 kcal/mol at 298 K for methane activation in TFAH, and ∆G‡ = 32.0 kcal/mol at 298 K for Rh−Me functionalization in TFAH, also among the lowest we have found using RhIII . For RhIII complexes with neutral transition states, we find that increasing the electron-withdrawing nature of the ligands decreases barriers for both activation and functionalization. Thus, the only condition opposing even less donating ligands (i.e. lower barriers) is catalyst stability. This chapter should be of great interest to experimentalists who can now focus on these two ligands in TFAH and water to validate and further optimize these systems. 3.3 Materials and Methods All quantum mechanical calculations were carried out using the Jaguar software version 7.6 developed by Schrödinger Inc. [5]. Geometry optimizations were carried out on initial guess structures, and vibrational frequencies were calculated to confirm the optimized geometries as intermediates or transition states and to construct a free energy profile. Solvation energies were calculated using the PBF Poisson-Boltzmann implicit continuum solvation model [6] in Jaguar, with a dielectric constant of 8.55 and a probe radius of 2.451 Å for TFAH and 80.37 and 1.40 Å for water. Explicit waters F were added into the calculations of aqueous Rh(NN ) (two explicit waters per aqua ligand) for more accurate solvation modeling. Geometry optimization and vibrational data were calculated using the B3LYP density functional [7] with a smaller basis set, whereas single point gas-phase and solvated energies were calculated using the M06 functional [8] and a larger basis set. Here the “smaller basis set” consists of a modified double-ζ Los Alamos basis set and pseudopotential [9] that includes f functions for rhodium [10], and the 6-31G** basis set [11] for the other atoms; whereas the “larger basis set” consists of the triple-ζ Los Alamos basis set and pseudopotential (LACV3P**++) modified to include f functions and diffuse functions for rhodium, and the 6-311G**++ basis set [12] for the other atoms. For orbital analysis, the Pipek-Mezey localization procedure was used [13]. 59 Rather than specify a particular chemical oxidant, we adopted a consistent electrostatic potential for electrons to determine the free energy changes of redox reactions. A value of 1.23 V vs. SHE, the standard potential for the reduction of oxygen at pH = 0 and 25◦ C (O2 (g) + 4 H+ (aq) + 4 e– (aq) −−→ 2 H2 O(`)), was adopted for models in trifluoroacetic acid. For water, the pH was taken as 7 and the potential thus reduced to 0.817 V; this was calculated using the equation E = E ◦ − ∆G/nF where F is Faraday’s constant and ∆G = kT ln Q = kT · pH · ln 10. The free energy of the electron was then calculated using the equation G = −F(E + 4.28 V) where 4.28 V represents the absolute potential of the SHE reference, yielding electron free energies of −127.1 and −117.5 kcal/mol, respectively [14]. The free energy of the proton was taken as −260 kcal/mol in TFAH and −279.80 kcal/mol in pH = 7 water [15]. The free energy for each molecular species in solution was calculated using the formula G = Egas + ∆Gsolv + ZPE + Hvib + 6kT − T [Svib + 0.54(Strans + Srot − 14.3 e.u.) + 7.98 e.u.] where the last term is an empirical approximation for the change in the translational and rotational entropy of the molecule between the gas phase and the solution phase (due to the finite librational frequencies) derived from Wertz [16]. For gas phase molecules (methane, methanol at 498.15 K, and for the purposes of this study methyl trifluoroacetate), we assumed that equilibration between the dissolved gas and the headspace occurred at a much faster timescale than the reactions in question; leading to ∆Ggas→solv = 0. Thus, the free energy of such gas molecules can be simply calculated using the formula G = Ggas = Egas + ZPE + Htot − T Stot . We simulated methanol in the gas phase at 498.15 K and the solution phase at room temperature by adding the empirical hydration free energy of −3.2 kcal/mol [17] to the gas phase free energy. For pure liquids (e.g., trifluoroacetic acid and water), the Gibbs free energy was calculated using the formula Gliquid = Egas + ZPE + Htot − T Stot + ∆Ggas→liquid where ∆Ggas→liquid = Gliquid − Ggas (1 atm) is the free energy of condensation to liquid from 1 atm gas. We can solve for this by noting that ∆Ggas→liquid = ∆Gexp + ∆Ggas→solv where ∆Gexp = Ggas (P ) − Ggas (1 atm) is the expansion of the gas from 1 atm to the vapor pressure P , and ∆Ggas→solv is the condensation of gas to liquid. Since a liquid is by definition at equilibrium 60 Temperature 298.15 K 323.15 K 348.15 K 373.15 K 398.15 K 423.15 K 448.15 K 473.15 K 498.15 K Trifluoroacetic acid A B C 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 3.33963 1267.252 -52.958 A 5.40221 5.20389 5.07680 5.08354 3.55959 3.55959 3.55959 3.55959 3.55959 Water B 1838.675 1733.926 1659.793 1663.125 643.748 643.748 643.748 643.748 643.748 C -31.737 -39.485 -45.854 -45.622 -198.043 -198.043 -198.043 -198.043 -198.043 Table 3.1. Antoine equation parameters used for trifluoroacetic acid and water at each temperature point investigated. Numbers taken from [18] and are set such that P will be measured in bar. with its vapor pressure, ∆Ggas→solv = 0, and we thus have  ∆Ggas→liquid = Ggas (P ) − Ggas (1 atm) = RT ln P 1 atm  . We can find the vapor pressure P at a given temperature using the Antoine Equation: log10 P = A − B , C +T where the empirical parameters A, B, and C vary with the solvent and temperature range and were taken from table 3.1. Calculations were performed at nine temperature levels spaced by 25 K from 298.15 K to 498.15 K. The SR 2 attack involving OVV Cl3 converts two singlets to two doublets, and hence the transition states feature spin contamination (S 2 > 0) while representing an overall singlet. Structures that did not feature such spin contamination were considered to be transition states for an alternative SN 2 attack that forms MeOVIII Cl–3 and a RhI species, all of which were higher in energy. 3.4 Results The most common oxidation state of rhodium is +III, but oxidation states from 0 to +VI have been observed [19]. While previous work has focused on the C−H activation ability of RhI [3]b, our current results center on a RhIII resting state. This is due to the requirement that the desired process operate above the CH4 /CH3 OH redox couple (0.59 V vs. SHE at pH = 0) and below the reduction potential of O2 in the reaction conditions, which favor higher oxidation states. Scheme 3.1 shows hypothetical catalytic cycles for the activation and functionalization of methane. Starting from an inorganic RhIII resting state (1), we investigated the C−H activation of methane 61 following the displacement of a TFAH solvent molecule (2‡ ) to form a RhIII -methyl organometallic species (3). Starting with (3), there are several pathways to functionalization: 1. III-I: SN 2 attack (4‡ ) by the conjugate base of the solvent to form methyl trifluoroacetate and a RhI species (5), which is reoxidized to the inorganic RhIII resting state (1). In all cases investigated we found that this SN 2 attack occurs on a five-coordinate Rh complex, with the neutral solvent trans to the methyl group dissociated in the transition state. This is consistent with previous work by Goldberg et al. implicating five-coordinate platinum intermediates [20]; 2. III-II: SR 2 attack (6‡ ) by a metal-oxo species to form a metal-methoxy species and a RhII species (7) via methyl radical transfer. Both of these are then reoxidized by one electron, to the metal-oxo species and methanol and to the inorganic RhIII resting state (1), respectively. Here, OVCl3 was used as a model metal-oxo capable of one-electron reduction, even though it would likely hydrolyze in these solvents. We are developing stable metal oxo reagents optimized for this reaction mechanism [21] and use OVCl3 here as a computationally simple model. The SR 2 F transition state for Rh(NN) involves a five-coordinate Rh complex as well. For the Rh(NN ) case, the transition states retain weak interactions with the neutral trans ligand, with an elongated Rh−O bond distance of 2.472 Å for TFAH and 2.228 Å for water. This is presumably F due to the increased electrophilicity of the metal center in the Rh(NN ) complexes; 3. III-IV-II: Oxidation to a RhIV species (8) followed by SN 2 attack (9‡ ) generating a RhII species (7), which is then further oxidized back to the RhIII resting state (1). This route was x only investigated for the Rh(NN ) and Rh(DP M E S) complexes, which had both kinetically accessible activation transition states and thermodynamically accessible RhIV intermediates. In all cases investigated we found that this SN 2 attack occurs on a six-coordinate Rh complex, presumably due to the increased electrophilicity of the RhIV center. The six-coordinate Rh complexes have elongated Rh−O bond distances to the neutral trans ligand (e.g., 2.377 Å for F F Rh(NN), 2.379 Å for Rh(NN ), and 2.255 Å for Rh(NN ) in water). In all cases, the energy of reoxidation was calculated assuming a potential of 1.23 V vs. SHE, i.e. with O2 as the terminal oxidant. However, in practice it may be necessary to use intermediate oxidants as well, although that is beyond the scope of this study. A system which can proceed around any of these cycles with a global activation barrier below approximately 36 kcal/mol at 200◦ C is of academic interest, as transition state theory gives a turnover frequency (TOF) of roughly 1 hr−1 at 1 atm CH4 for such a process. Achieving an industrially relevant TOF on the order of 1 s−1 requires a global barrier of 29 kcal/mol, although greater pressures of CH4 would increase this limit [22]. We screened complexes of rhodium with the following ligands, as depicted in Figure 3.1: 62 Scheme 3.1. Hypothetical catalytic cycle for the activation and functionalization of methane. This shows several potential routes, including III-I (red), III-II (blue), and III-IV-II (green), which are further described in the text. In addition, a fourth possibility, I-III (purple), involves the direct C−H activation (10‡ ) by a RhI species (5) to a RhIII hydride methyl (11). However, given the oxidizing nature of the reaction conditions, we do not anticipate this to be a fruitful pathway, as any catalytic cycle involving a RhI -(III) cycle would likely suffer from the fact that oxidation of 5 to 1 would act as a thermodynamic sink. Hence, this last option was not pursued to the same extent in this investigation. 63 F • the bis(N -phenyl)benzylamidinate (NN) ligand and its fluorinated analogue (NN ); F NMe2 • the (N -phenyl)acetaldiminyl quinolate (ONN) ligand and its related (ONN ) and (ONN ) ligands; • the bis(pyridyl)methanesulfonate (DPMS) and bis(pyridyl)ethanesulfonate (DPES) ligands; • the bis(pyrrolyl)quinolinyl phosphine (PN) ligand; and • the bis(quinolinyl)benzene (bisq) ligand, its di- and tetrafluorinated analogues (bisqF2 ) and (bisqF4 ), and its di- and tetramethylated analogues (bisqMe2 ) and (bisqMe4 ). Figure 3.1. The rhodium-ligand complexes screened. For each complex, the RhI form in TFAH is shown here, but full catalytic cycles were computed. The best kinetics were predicted for the Rh(NNF ) and Rh(bisq) complexes. For each of these ligands complexed with rhodium, we generated catalytic cycles in the template of Scheme 3.1. For the (DPMS), (PN), and (ONNx ) family of ligands, we found that either the methane activation energy or the methyl functionalization energy (or both) were too high for effective catalysis (Table 3.2). However, we found lower numbers for the (NNx ) family of ligands, on which we have decided to focus our discussion. 3.4.1 The Rh(NN) complexes in TFAH Scheme 3.2 shows the calculated free energies of Rh(NN) complexes in trifluoroacetic acid solvent. III 2– + Note that all free energies are referenced to the resting state, [(NN)Rh (TFA)4 ] (Hax )2 , which is highlighted by a purple box. Hence, although the resting complex is regenerated at the end of the catalytic cycle, the new free energy is −13.7 kcal/mol at 298 K, because this is the energy change of the overall equation CH4 + TFAH −−→ TFA−Me + 2 H+ + 2 e– ; and thus all energies at the second cycle would be −13.7 kcal/mol lower. Oxidation of the inorganic RhI species to inorganic RhIII species is highly favorable, from 40 to 50 kcal/mol downhill, depending on the temperature and the charge of the species. We conclude that 64 a. b. Scheme 3.2. (a) Activation of methane using Rh(NN) complexes in TFAH; and (b) the subsequent functionalization. Red denotes the III-I functionalization pathway; blue the III-II pathway, and green the III-IV-II pathway. Free energies (kcal/mol) are referenced to the resting state, [(NN)RhIII (TFA)4 ]2– (H+ ax )2 . The resting state is highlighted by a purple box, and the key methyl inter+ mediate, [(NN)RhIII (Meax )(TFA)3 ]2– (H+ ax )(Heq ), is highlighted by a yellow box. 65 TS type Activation Functionalization TS type TS Meax Meeq (III-I) (III-II) (III-IV-II) TS Meax Activation Meeq Functiona- (III-I) lization (III-II) Rh(NN) 32.6/34.8 38.3/40.4 54.8/60.3 43.2/44.2 42.2/48.6 Rh(NN ) 27.6/29.6 33.7/36.3 42.7/48.2 38.1/40.0 36.8/43.6 F Rh(ONN) 37.2/39.8 37.9/40.7 44.2/49.3 40.4/42.5 N/A Rh(ONN ) 35.8/37.8 38.1/40.1 41.7/45.8 39.1/40.8 N/A F 2 Rh(ONN ) 38.7/41.2 40.0/42.9 46.7/52.6 41.4/43.4 N/A NMe Rh(DPMS) 38.0/40.2 33.9/36.2 46.6/51.3 41.0/41.6 49.3/56.3 Rh(PN) 40.1/43.0 35.0/37.7 38.3/43.8 47.7/56.9 Rh(bisq) 38.3/40.7 33.4/35.8 31.8/37.8 32.0/33.7 Rh(bisqF2 ) 39.3/42.4 36.4/39.4 32.3/38.4 31.8/33.6 Rh(bisqF4 ) 40.4/43.0 38.2/40.9 30.2/35.0 33.7/35.3 Rh(bisqMe2 ) 38.8/41.9 32.8/35.9 35.0/41.9 35.7/38.3 Rh(bisqMe4 ) 40.7/44.4 34.3/37.5 38.0/45.5 38.9/42.0 Rh(DPES) 41.0/44.0 35.6/38.4 53.8/59.3 45.5/47.4 49.9/57.1 Table 3.2. Lowest activation and functionalization energies for each series of Rh-ligand complexes in TFAH. Detailed cycles for the (NNx ) family of ligands are found in Schemes 3.2 and 3.3. Detailed cycles for the (ONNx ) family of ligands are found in Schemes 3.9 through 3.11. Detailed cycles for the (DP M S) family of E ligands are found in Schemes 3.12 through 3.14. Detailed cycles for the (PN) ligand are found in Schemes 3.15 and 3.16. Detailed cycles for the (bisqx ) family of ligands are found in Schemes 3.17 through 3.20. For each entry, the first number is at 298 K and the second at 498 K. Not all III-IV-II transition states were calculated. All free energies in kcal/mol. the III-I SN 2 attack pathway is not likely. Indeed, the activation barrier for this is a prohibitively high 54.8 kcal/mol at room temperature (Figure 3.2b). In addition, we also conclude that a I-III pathway involving methane activation by RhI is not likely due to the oxidation potential of the RhI species. The remaining viable pathways all involve methane activation by the inorganic RhIII (NN) complex to form a RhIII (NN)−Me species. We were initially gratified to see that thermodynamically the methyl species is not very endergonic. We were even more intrigued to find that the methane activation transition state barrier is only 32.6 kcal/mol at 298 K (Figure 3.2a), making it the most accessible barrier that we had found for RhIII complexes in TFAH (Table 3.2). In contrast, neither the III-II (Figure 3.2c) nor the III-IV-II (Figure 3.2d) functionalization pathways seemed as promising. Although further oxidation to a RhIV (NN)−Me species was roughly thermoneutral, the barrier to reductive functionalization was 42.2 kcal/mol. The oxidation from RhIII to RhIV was effective in lowering the global SN 2 barrier from 54.8 to 42.2 kcal/mol. To explain the high functionalization barrier for SN 2 attack, we note that this may be due to the high energy or low chemical activity of trifluoroacetate ion in solution. Since TFA– is necessary as the attacking nucleophile, a higher pH may be employed to lower this barrier before a deprotonated intermediate becomes the new resting state. Due to the unusually low barrier for methane activation, we did not want to dismiss this ligand framework. Rather, our efforts focused on how potential modifications on the (NN) ligand might lower the functionalization barrier. Reasoning that a more electronegative ligand may inductively increase the electrophilicity of the rhodium-methyl bond and make nucleophilic attack more likely, we decided to investigate Rh complexes with the fluorinated analogue bis(N -pentafluorophenyl) F pentafluorobenzylamidinate (NN ) ligand. 66 Figure 3.2. Rh(NN) transition state structures. (a) methane activation: Rh1-C36 2.273 Å, C36-H19 1.365 Å, H19-O10 1.321 Å, Rh1-C36-H19 64.624◦ , C36-H19-O10 155.492◦ . (b) Rh(III-I) SN 2 attack: Rh1-C60 2.431 Å, C60-O65 1.936 Å, Rh1-C60-O65 176.680◦ , Rh1-O53 3.473 Å. (c) Rh(III-II) SR 2 attack: Rh1-C60 2.368 ÅC60-O65 1.995 Å, Rh1-C60-O65 175.541◦ , Rh1-O53 3.171 Å. (d) Rh(III-IV-II) SN 2 attack: Rh1-C60 2.138 Å, C60-O64 2.263 Å, Rh1-O53 2.377 Å, Rh1-C60-O64 168.802◦ , C60-Rh1-O53 164.984◦ . 67 The Rh(NNF ) complexes in TFAH 3.4.2 F Compared to the analogous Rh(NN) complexes, the fluorinated Rh(NN ) complexes (Scheme 3.3) show some marked differences in free energies. Many of these differences can be attributed to the F much higher electron-withdrawing character of the (NNF ) ligand. For instance, the RhI (NN ) species is 13.3 kcal/mol more stable to oxidation than the corresponding RhI (NN) species. In the same way, the SN 2 functionalization barrier on the III-I path is 12.1 kcal/mol lower. This is due to the highly F electron-withdrawing (NN ) ligand, which stabilizes lower oxidation states. Unfortunately, this effect is not sufficiently large for us to consider the III-I or I-III pathways. Other thermodynamic effects of fluorine atom substitution include increased acidity of all the complexes (i.e. stabilization of the anionic species by 4-8 kcal/mol), an increased stabilization of the RhIII −Me species (by about 8 kcal/mol), and a decreased stabilization of the RhIV −Me species (by about 20 kcal/mol relative to RhIII −Me.) The barrier for methane activation, already relatively low for the Rh(NN) case, was even lower F for the Rh(NN ) case, at only 28.7 kcal/mol at 298 K (Figure 3.3a). However, the real test of the utility of the (NNF ) ligand is its effectiveness in aiding the functionalization step. It lowers the global III-I SN 2 barrier from 54.8 to 42.7 kcal/mol and the isolated SN 2 barrier (from the RhIII −Me intermediate) to 40.8 from 45.1 kcal/mol (Figure 3.3b). It raises the neutral III-IV-II SN 2 barrier by raising the RhIV intermediate energy. Hence, we put our hopes into the (NNF ) ligand lowering the barrier for a net anionic III-IV-II SN 2 or a III-II SR 2 attack (Figure 3.3c,d). And indeed it does: the new global SN 2 and SR 2 barriers of 36.8 and 38.1 kcal/mol, respectively, approach the target values. The Rh(NNF ) complexes in water 3.4.3 Another method that may lower transition state barriers is switching the solvent, and therefore the nucleophile, to water. This has the practical advantage that any commercialized process will have greater simplicity if carried out in water as opposed to a strong acid. Hence, we investigated the F same Rh(NN ) complexes in water. F The thermodynamic profile for Rh(NN ) complexes in water (Scheme 3.4) is very promising: F F F the RhIII (NN )-methyl, RhIV (NN )-methyl, and RhII (NN ) species are all stabilized relative to the F inorganic RhIII (NN ) reference state. Hence, whereas for the TFAH solvent case a thermodynamic barrier must be climbed in each catalytic step, in water each step is either thermodynamically downhill or approximately neutral. F III In Scheme 3.4, all energies are referenced to the lowest inorganic RhIII state, [(NN )Rh 2– + + (OH)4 ] (Hax )(Heq ), which is highlighted by a purple box. This is to keep consistent with the practices in Schemes 3.2 and 3.3, where the lowest inorganic RhIII state is also the resting state. F However, in this case the organometallic RhIII (NN )-methyl complexes are stabilized to such an 68 a. b. Scheme 3.3. (a) Activation of methane using Rh(NNF ) complexes in TFAH; and (b) the subsequent functionalization. Red denotes the III-I functionalization pathway; blue the III-II pathway, and green the III-IV-II pathway. Free energies (kcal/mol) are referenced to the resting state, [(NNF )RhIII (TFA)4 ]2– (H+ ax )2 . The resting state is highlighted by a purple box, and the key methyl intermediate, [(NNF )RhIII (Meax )(TFA)3 ]2– (H+ ax ), is highlighted by a yellow box. 69 Figure 3.3. Rh(NNF ) transition state structures. (a) methane activation: Rh1-C61 2.267 Å, C61-H19 1.365 Å, H19-O10 1.321 Å, Rh1-C61-H19 63.875◦ , C61-H19-O10 157.549◦ . (b) Rh(III-I) SN 2 attack: Rh1-C60 2.368 Å, C60-O65 2.021 Å, Rh1-C60-O65 176.330◦ , Rh1-O53 3.265 Å. (c) Rh(III-II) SR 2 attack: Rh1-C60 2.399 ÅC60-O65 1.935 Å, Rh1-O53 2.472 Å, Rh1-C60-O65 172.164◦ , C60-Rh1-O53 168.496◦ . (d) Rh(III-IV-II) SN 2 attack: Rh1-C60 2.259 Å, C60-O64 2.156 Å, Rh1-O53 2.379 Å, Rh1-C60-O64 159.539◦ , C60-Rh1-O53 169.220◦ . 70 Scheme 3.4. Thermodynamic profile for the activation and functionalization of methane using Rh(NNF ) complexes in water. This chart is meant to be read from left to right; the first row contains the pathway through neutral species and the second row through anionic species. Red denotes the III-I functionalization pathway; blue the III-II, and green the III-IV-II pathway. All free energies (kcal/mol) are at pH = 7 and + referenced to the lowest inorganic RhIII state, [(NNF )RhIII (OH)4 ]2– (H+ ax )(Heq ), highlighted by a purple box; F III whereas the key methyl intermediate and true resting state, [(NN )Rh (Meeq )(OH)3 ]2– (H+ ax )2 , is highlighted by a yellow box. 71 F III 2– + extent that the true resting state is now [(NN )Rh (Meeq )(OH)3 ] (Hax )2 , which is highlighted by a yellow box and is analogous to the key methyl intermediate referenced in Schemes 3.2 and 3.3. The practical consequence is that the barrier to CH4 activation is properly referenced to the lowest inorganic RhIII state, whereas the barrier to methyl group functionalization ought to be referenced to the true resting state. F Having verified that the Rh(NN ) catalytic system in water was thermodynamically feasible, we sought accessible transition states for both the methane activation and functionalization steps. For methane activation, we began by assuming that the activation would take place via electrophilic substitution with a cis hydroxo ligand acting as a proton acceptor (Scheme 3.5). We then found transition states, starting from both neutral and anionic species, with the methane and the interactF ing hydroxo in every possible configuration with respect to the (NN ) ligand. Our lowest transition state barrier of 35.0 kcal/mol at 298 K occurs with the methyl group forming in the axial position, donating its extra proton to an equatorial hydroxo ligand and trans to an axial hydroxo ligand (Figure 3.4a). This is significantly higher than in the TFAH solvent case, but still within the limits of feasibility. We examined the possible role of a concerted reaction with a bridging water molecule, but we did not see a decrease in the transition state energy. Scheme 3.5. Transition states for the activation of methane using Rh(NNF ) complexes in water. Chart is meant to be read from top to bottom; the left side contains the pathway through neutral species and the right side through anionic species. All free energies are in kcal/mol and referenced to the lowest inorganic RhIII state. Following the activation of methane, the other transition state that needed to be calculated was for the functionalization of the rhodium-methyl bond. We investigated the functionalization through 72 Figure 3.4. Rh(NNF ) in water transition state structures. (a) methane activation: Rh1-C66 2.296 Å, C66H65 1.383 Å, H65-O5 1.271 Å, Rh1-C66-H65 55.587◦ , C66-H65-O5 161.845◦ . (b) Rh(III-I) SN 2 attack: Rh1-C64 2.698 Å, C64-O67 1.803 Å, O67-H68 1.043 Å, H68-O47 1.516 Å, O47-H53 1.001 Å, H53-O5 1.718 Å, Rh1-C60-O65 148.135◦ , Rh1-O54 3.265 Å. (c) Rh(III-II) SR 2 attack: Rh1-C65 2.371 ÅC65-O68 2.010 Å, Rh1-O55 2.228 Å, Rh1-C65-O68 164.320◦ , C65-Rh1-O55 169.736◦ . (d) Rh(III-IV-II) SN 2 attack: Rh1-C64 2.526 Å, C64-O67 1.930 Å, O67-H68 1.033 Å, H68-O46 1.503 Å, O46-H51 1.008 Å, H51-O47 1.594 Å, O47-H3 0.981 Å, H3-O5 1.901 Å, Rh1-O54 2.255 Å, Rh1-C64-O67 171.136◦ , C64-Rh1-O54 167.491◦ . 73 the III-I, III-II, and III-IV-II routes (Schemes 3.6, 3.7, and 3.8, respectively). For each route, we restricted our investigation to activations from rhodium species with axial methyl groups, since (as in the TFAH case) the axial isomer is preferentially formed during the activation step. Scheme 3.6. Transition states for the III-I functionalization pathway via SN 2 attack in water. The left side is the pathway through neutral species and the right side through anionic species. All free energies are in kcal/mol and referenced to the resting state, [(NNF )RhIII (Meeq )(OH)3 ]2– (H+ ax )2 . For the III-I route, we required an aqua ligand to be the leaving group trans to the methyl, limiting us to only two possible transition states: a neutral transition complex (Scheme 3.6, left) and an anionic complex (Scheme 3.6, right). Although the neutral transition state (Figure 3.4b) is somewhat lower in energy, we found that in both cases the activation energy for this pathway is prohibitively high. For the III-II route (Scheme 3.7), we no longer needed require the leaving group trans to the RhIII methyl to be an aqua ligand. Hence we performed calculations on four different potential transition F III 2– + states, starting with two isomers of neutral [(NN )Rh (Meax )(OH)3 ] (H )2 (Scheme 3.7, upper F III 2– + left) and two isomers of anionic [(NN )Rh (Meax )(OH)3 ] (H ) (Scheme 3.7, upper right). We found that this SR 2 pathway was even more facile, with all barriers in the 30s (Scheme 3.7, middle) and the lowest barrier being 31.7 kcal/mol for OVCl3 attack on a neutral species with an axial methyl ligand, which is shown in detail in Figure 3.4c. For the III-IV-II route (Scheme 3.8), a good leaving group trans to the RhIII -methyl once again became important. Hence, we were limited to only investigating one possible transition state, shown in detail in Figure 3.4d. We found the activation energy for this process to be approximately equal to that of the methane activation step. F Hence, we conclude that for the Rh(NN ) catalytic system in water, both III-IV-II SN 2 and III-II SR 2 are viable pathways for functionalization, whereas the III-I SN 2 pathway is still too high in energy. This is similar to the case in TFAH, except that the relative ease of activation vis-à-vis functionalization has now been reversed. 3.4.4 The Rh(ONNx ) family of complexes in TFAH Schemes 3.9, 3.10, and 3.11 depict the catalytic cycle for the activation and functionalization of F NMe2 methane using Rh(ONN), Rh(ONN ), and Rh(ONN ) complexes in TFAH, respectively. In all 74 Scheme 3.7. Transition states for the III-II functionalization pathway via SR 2 attack in water. This chart is meant to be read from top to bottom; the left side contains the pathway through neutral species and the right side through anionic species. All free energies are in kcal/mol and referenced to the resting state, [(NNF )RhIII (Meeq )(OH)3 ]2– (H+ ax )2 . Scheme 3.8. Transition state for the III-IV-II functionalization pathway via SN 2 attack in water. All free energies are in kcal/mol and referenced to the resting state, [(NNF )RhIII (Meeq )(OH)3 ]2– (H+ ax )2 . 75 cases, the resting state contains an axial TFAH ligand and equatorial TFA ligands, but the activation transition state which involves an axial methyl group is slightly lower in energy than the transition state which involves an equatorial methyl group. The methane activation energy for Rh(ONN) is F 37.2 kcal/mol at 298 K, which is somewhat too high for promising catalysis. Switching to Rh(ONN ) decreases this activation energy slightly to 35.8 kcal/mol, whereas using Rh(ONN NMe2 ) raises it to 38.7 kcal/mol. Scheme 3.9. Catalytic cycle for the activation and functionalization of methane using Rh(ONN) complexes in TFAH. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. The RhIII (ONNx )−Me species investigated have a methyl group axial to the (ONNx ) plane; any isomer with an equatorial methyl group would likely be less stable due to the trans effect of the (ONNx )’s quinoline nitrogen. The RhIII (ONNx )−Me species are slightly endergonic relative to their respective resting states. In all cases, the III-II SR 2 pathway is favored over the III-I SN 2 pathway. For Rh(ONN) the F SR 2 barrier is 40.3 kcal/mol at 298 K; this decreases to 39.2 kcal/mol for Rh(ONN ) but increases 76 Scheme 3.10. Catalytic cycle for the activation and functionalization of methane using Rh(ONNF ) complexes in TFAH. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 77 Scheme 3.11. Catalytic cycle for the activation and functionalization of methane using Rh(ONNNMe2 ) complexes in TFAH. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 78 to 41.4 kcal/mol for Rh(ONN NMe2 ). The resultant RhIII (ONNx ) species were all approximately the same (20.5, 20.3, and 20.7 kcal/mol, respectively) in energy relative to their resting states. 3.4.5 The Rh(DPMS) and Rh(DPES) complexes in TFAH Scheme 3.12 shows the catalytic cycle for the activation and functionalization of methane using Rh(DPMS) complexes in TFAH, with the latter via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. While there is a relatively accessible activation barrier of 33.9 kcal/mol at 298 K, it differs from the activation transition states of the NNx and ONNx families in that it forms a methyl complex with the methyl group equatorial to the N−Rh−N plane. Interconvertibility with the lowerenergy axial methyl complex is assumed given the lability of the sulfonate moiety and the TFAH ligands. Unfortunately, the lowest functionalization barrier found was 41.0 kcal/mol at 298 K for the III-II pathway, which is too high. The analogous catalytic cycle for Rh(DPES) complexes in TFAH is shown in Scheme 3.13. The energy profile is very similar to that of the Rh(DPMS) complexes; however, the transition states found are typically several kcal/mol higher than their Rh(DPMS) counterparts. This may be due to the extra methyl group rendering the overall ligand more electron donating, destabilizing the formation of a Rh−Me bond and decreasing its electrophilicity for functionalization. We also investigated RhI (DPMS) species, the results of which are shown in Scheme 3.14. The energy profile appears to be much more favorable, with the highest barrier being the oxidative addition of methane at 26.4 kcal/mol at 298 K. However, under reaction conditions one might expect the oxidative addition of methane by RhI (DPMS) to be outcompeted by RhI (DPMS)’s oxidation to a RhIII (DPMS) species, which according to Scheme 3.12 is downhill by 33.0 kcal/mol at 298 K. Therefore, the I-III pathway was not considered to be viable. 3.4.6 The Rh(PN) complexes in TFAH Scheme 3.15 shows the catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, with the latter via the Rh(III-I) and Rh(III-II) pathways. Unfortunately, it appears that the functionalization barriers are too high for effective catalysis with this ligand: the lowest III-I SN 2 barrier is 38.0 kcal/mol at 298 K and the lowest III-II SR 2 barrier is 47.7 kcal/mol. Taking into account the additional fact that the activation transition state to produce an axial methyl species is 40.1 kcal/mol high, we conclude that RhIII (PN) is not likely to be an effective catalyst for methane oxidation. We also investigated RhI (PN) species, the results of which are shown in Scheme 3.16. As in the case of the Rh(DPMS) complexes, oxidative addition of methane by RhI (PN) appears to be more facile, with a barrier of 28.1 kcal/mol at 298 K. Unfortunately, the resultant RhIII (PN)(H)(Me) 79 Scheme 3.12. The activation and functionalization of methane using Rh(DPMS) complexes in TFAH, via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 80 Scheme 3.13. The activation and functionalization of methane using Rh(DPES) complexes in TFAH, via the Rh(III-I), Rh(III-II), and Rh(III-IV-II) pathways. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 81 Scheme 3.14. Catalytic cycle for the activation and functionalization of methane using Rh(DPMS) complexes in TFAH, via the Rh(I-III) pathway. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. species are all significantly thermodynamically uphill (at least 22.0 kcal/mol at 298 K). Furthermore, as in the case of the Rh(DPMS) complexes, under reaction conditions one might expect the oxidative addition of methane by RhI (PN) to be outcompeted by RhI (PN)’s oxidation to a RhIII (PN) species, which according to Scheme 3.15 is downhill by 23.7 kcal/mol at 298 K. Therefore, the I-III pathway was not further pursued. 3.4.7 The Rh(bisqx ) family of complexes in TFAH Due to the large in-TFAH Rh−Me functionalization barriers encountered in the other ligands so far (Table 3.2), we designed the bis(quinolinyl) benzene (bisqx ) family of ligands (Figure 3.1) in the hopes of finding more facile SN 2 and SR 2 pathways. These ligands would be expected to be coordinated to rhodium in a fac-L3 manner, with a weak η 2 -benzene interaction axial to the N−Rh−N plane. A methyl group would then be expected to be coordinated axial to the N−Rh−N plane as well, due to the weaker trans effect of the benzene ring as opposed to the nitrogen donors. The η 2 benzene interaction is also expected to be a better leaving group than TFA/TFAH, thereby lowering both the barrier and thermodynamics of functionalization. Beyond the base (bisq) ligand itself, we also investigated the di- and tetrafluorinated analogues (bisqF2 ) and (bisqF4 ), and the di- and tetramethylated analogues (bisqMe2 ) and (bisqMe4 ), to see how changing the electronics of the ligand 82 Scheme 3.15. Catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, via the Rh(III-I) and Rh(III-II) pathways. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. 83 Scheme 3.16. Catalytic cycle for the activation and functionalization of methane using Rh(PN) complexes in TFAH, via the Rh(I-III) pathway. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. might change its associated transition state barriers. x x Scheme 3.17 shows the ground states of the various RhIII (bisq ) species: the neutral RhIII (bisq )III x + (TFA)3 , the protonated [Rh (bisq )(TFA)2 (TFAH)] III x + and [Rh (bisq )(TFA)2 )] , and their relax tive free energies. The first point to note is that, since the (bisq ) ligands are L3 , activation of methane requires first the protonation of a TFA ligand followed by its removal as TFAH to form an open x coordination site. As Scheme 3.17 shows, as we increase the electron-withdrawing groups on (bisq ), replacing Me with H and H with F, the resultant decrease in electron donation to the central Rh decreases the basicity of the attached TFA ligands and increases the energy of their protonated forms. x The second and related point to note is that the true resting state is not consistent for all (bisq ) species: the more electron-rich ligands (bisqMe2 ) and (bisqMe4 ) have positively-charged resting states, the base ligand (bisq) has both the neutral and a protonated species in even equilibrium, while the less electron-rich ligands (bisqF2 ) and (bisqF4 ) prefer to be neutral. However, the energies of all species x have been referenced to neutral RhIII (bisq )(TFA)3 for consistency. The final point to note is that the specific protonated resting state for the more electron-rich ligands (bisqMe2 ) and (bisqMe4 ) depends III Mex on temperature: at 298 K [Rh (bisq + )(TFAax )(TFAHeq · · · TFAeq )] is favored, but at 498 K it is III Mex more advantageous to dissociate the TFAH and have [Rh (bisq + )(κ1 −TFAeq )(κ2 −TFAax, eq )] instead. Scheme 3.18 shows the three potential methane activation transition states that may result 84 Scheme 3.17. The inorganic RhIII (bisqx )(TFA)3 species and its protonated analogues III x [Rh (bisq )(TFA)2 (TFAH)]+ and [RhIII (bisqx )(TFA)2 )]+ , along with their relative energies in TFAH. Both protonation and TFAH removal are necessary before methane activation can take place. All free energies are in kcal/mol and referenced to the corresponding RhIII (bisqx )(TFA)3 species. III x + depending on the CH4 molecule’s approach towards the [Rh (bisq )(TFA)2 )] ’s open coordination site. The top transition state shows the result of a methane approach to an axial open coordination site, whereas the other two show the result of an equatorial approach, with either the axial or III equatorial TFA ligand gaining the methane’s proton. The axial approach results in a [Rh −Meax ] III species, whereas the equatorial approaches result in [Rh −Meeq ] + + species, all of which may be IV deprotonated to form their neutral analogues. Further oxidation to Rh species is quite uphill; hence the III-IV-II functionalization pathway was not investigated for these rhodium-ligand complexes. x It should be noted that regardless of the specific (bisq ) ligand, the lowest energy methane actiIII x +‡ vation transition state is [Rh (bisq )(TFAax )(Meeq · · · H· · · TFAeq )] III , which involves an equatorial x + approach for the methane and results in the methyl complex [Rh (bisq )(TFAax )(Meeq )(TFAHeq )] . However, both the neutral and protonated RhIII −Meax species are lower in energy than their equatorial counterparts. We assume that interconversion between the RhIII −Meeq and RhIII −Meax is facile due to the lability of the TFAH and η 2 -benzene ligands. Upon examination of the methane activation transition state energies, it appears that the general trend is that increasing the electron-donating nature of groups on the (bisq) ligand appears to favor lower barriers. This is opposite to that observed in the other ligand families, in which less III x electron-donating ligands are preferred. However, a comparison of the energies of [Rh (bisq )(TFA)+‡ (Me· · · H· · · TFA)] III x + transition states with that of their protonated [Rh (bisq )(TFA)2 (TFAH)] 85 Scheme 3.18. The various methane activation transition states [RhIII (bisqx )(TFA)(Me· · · H· · · TFA)]+‡ and their resultant [RhIII (bisqx )(TFA)(Me)(TFAH)]+ products. Ensuing deprotation (in preparation for III-I and III-II functionalization) and oxidation (in preparation for III-IV-II functionalization) are also depicted. Blue denotes the most favored Rh−Me intermediate and red the most favored activation transition state. All free energies are in kcal/mol and referenced to the corresponding RhIII (bisqx )(TFA)3 species. 86 precursors (Scheme 3.17) shows that the decreased activation barriers for more electron-donating x ligands is simply due to the decreased energy of protonation of the inorganic RhIII (bisq )(TFA)3 precursors. It should also be noted that for (bisqMe2 ) and (bisqMe4 ), the energies shown in Scheme 3.18 are not the full transition barriers, since they are not referenced to the ground states of those rhodium-ligand complexes. The true barriers are recorded in Table 3.2. x An examination of the energies of the lowest Rh−Me intermediate for each (bisq ) ligand (Scheme 3.18, blue) shows that the relative energy decreases as the electron-withdrawing nature of groups on the (bisq) ligand is increased. This is easily explained by noting that a decrease in the electron-donating power of the (bisq) ligand is expected to increase the rhodium atom’s elecF trophilicity, and thus increase the strength of its bond with the methyl group. Indeed, for (bisq 4 ) x the Rh−Me species is slightly lower (by 0.3 kcal/mol) in energy than its RhIII (bisq )(TFA)3 precursor at 298 K, and is the true resting state. However, this does not change the overall methane activation barrier. The III-I SN 2 and III-II SR 2 functionalization pathways were investigated by searching for transition states resulting from TFA– /TFAH or OVCl3 attack on each Rh−Me species, axial and equax torial, neutral and positively charged; for all five (bisq ) ligands. A total of 65 transition state structures were analyzed. Figure 3.5 shows the functionalization transition states for SN 2 attack by TFA– /TFAH on various Rh−Me intermediates. Both H+ pre-protonation followed by TFA– attack and concerted TFAH attack/deprotonation transition states were investigated. The lowest energy SN 2 functionalization barrier was found to arise from the attack of a TFA– ion on a protonated Rh−Meax species. Not x surprisingly, lower barriers were found for RhIII (bisq ) complexes with less electron-donating ligands. The energies are overall significantly lower than those found with other ligand families, thus confirming our initial hypothesis that a weak η 2 -benzene coordinated trans to the methyl would facilitate SN 2 attack. Indeed, barriers of 34.1 kcal/mol and lower at 298 K are found for (bisq) and x its fluorinated analogues, thereby establishing the viability of the RhIII (bisq ) family of complexes as potentially effective catalysts. Figure 3.6 shows the functionalization transition states for SR 2 attack by OVCl3 on both neutral x III x + RhIII (bisq )(Me)(TFA)2 and protonated [Rh (bisq )(Me)(TFA)(TFAH)] species, with both axial and equatorial methyl conformations. In most cases the lowest energy SR 2 functionalization barrier was found to arise from the attack of a OVCl3 molecule on a neutral Rh−Meax species (the lone exception is (bisq Me4 ), which prefers a protonated transition state). As in the III-I SN 2 case, lower x barriers were found for RhIII (bisq ) complexes with less electron-donating ligands, and overall the barriers were reduced compared to those of other ligand families; although the effect is not as F4 pronounced as for the SN 2 case, and the tetrafluorinated (bisq ) version is an outlier. Barriers of 33.4 kcal/mol and lower at 298 K are found for (bisq) and its fluorinated analogues, thereby establishing 87 Figure 3.5. The various RhIII (bisqx )(Me) functionalization transition states for the III-I SN 2 pathway. All transition states resulting from TFA– /TFAH attack on each Rh−Me species, axial and equatorial, neutral and positively charged, are shown. The lowest transition state for each (bisqx ) ligand is shown in red. All free energies are in kcal/mol and referenced to the corresponding RhIII (bisqx )(TFA)3 species. 88 the viability of the III-II pathway as well as the III-I pathway for methyl functionalization of the x RhIII (bisq ) family of complexes. Figure 3.6. The various RhIII (bisqx )(Me) functionalization transition states for the III-II SR 2 pathway. All transition states resulting from OVCl3 attack on each Rh−Me species, axial and equatorial, neutral and positively charged, are shown. The lowest transition state for each (bisqx ) ligand is shown in red. All free energies are in kcal/mol and referenced to the corresponding RhIII (bisqx )(TFA)3 species. Scheme 3.19 shows the remainder of the catalytic cycle after either III-I or III-II functionalization, and compares the two functionalization processes. For the III-I pathway, SN 2 attack by TFA– /TFAH x results in the formation of the Me−TFA product and RhI (bisq )(TFA)(TFAH), which is reoxidized x to the starting complex RhIII (bisq )(TFA)3 . For the III-II pathway, SR 2 attack by OVCl3 results in x the formation of Me−OVCl3 and RhII (bisq )(TFA)(TFAH), which are also reoxidized to the starting x complex RhIII (bisq )(TFA)3 as well as the product Me−TFA and regenerated OVCl3 . Either way, x the final relative energy of the RhIII (bisq )(TFA)3 complex is −13.7 kcal/mol at 298 K, which represents a completed cycle of the overall reaction CH4 + TFAH −−→ TFA−Me + 2 H+ + 2 e– . x Finally, the I-III pathway was also investigated, starting from RhI (bisq )(TFA)(TFAH) as an 89 Scheme 3.19. Comparison of the functionalization of RhIII (bisqx )(Meax )(TFA)2 , the lowest Rh−Me species, via the III-I SN 2 and III-II SR 2 pathways; and the completion of the catalytic cycle. Only the lowest functionalization transition state is shown for each pathway. 90 alternative ground state. The results are shown in Scheme 3.20. Although the oxidative addition of x methane by RhI (bisq ) species was feasible with minimum transition states of 27.9-30.6 depending x on the particular ligand, the resulting RhI (bisq )(H)(Me)(TFA) species are very uphill in energy. Deprotonation to form a RhI −Me species was prohibitively uphill both kinetically and thermodyx x namically. Along with the fact that RhI (bisq ) species would be less stable than their RhIII (bisq ) analogues at reaction conditions, we did not pursue this route further. Scheme 3.20. Catalytic cycle for the activation and functionalization of methane using Rh(bisqx ) complexes in TFAH, via the Rh(I-III) pathway. Blue denotes the resting state and red the most favored pathway. All free energies are in kcal/mol. By plotting the activation and transition state energies of the various RhIII (bisqx ) ligands, we can determine the particular ligand with the optimal activity. The results are shown in Figure 3.7, and show convincingly that at both 298 K and 498 K, whether using the III-I or III-II pathway, the base ligand RhIII (bisq) is the best choice. In comparing the relative merits of the III-I and III-II pathways, the energies of the two transition x states at 298 K are very close, within 1 kcal/mol for each (bisq ) ligand (although, as mentioned F4 before, (bisq ) appears to be an outlier). However, the energies are expected to increase at 498 K, due to the increased entropy penalty of bringing an extra TFAH or OVCl3 to the system. For the III-I SN 2 pathway, this is a significant penalty of 6.0-6.2 kcal/mol, and may be enough to render functionalization inaccessible. However, for the III-II SR 2 pathway the entropy penalty is much less F2 at 1.7-2.0 kcal/mol, and both the base (bisq) and fluorinated (bisq ) and (bisq F4 ) have barriers at 91 Figure 3.7. Top: graph of the activation and III-I functionalization transition state energies of the various Rh(bisqx ) complexes. The best choice at both 298 K and 498 K is the base Rh(bisq) complex. This complex has a maximum barrier of 33.4 kcal/mol at 298 K and 37.8 kcal/mol at 498 K. The rate-determining step is activation at 298 K and functionalization at 498 K. Bottom: graph of the activation and III-II functionalization transition state energies of the various Rh(bisqx ) complexes. Again, the best choice at both 298 K and 498 K is the base Rh(bisq) complex. This complex has a maximum barrier of 33.4 kcal/mol at 298 K and 35.8 kcal/mol at 498 K. The rate-determining step is methane activation at both temperature points. Both: Blue diamond markers and lines denote transition state energies at 298 K, and red square markers and lines at 498 K. Outline marker shapes denote activation transition states and filled marker shapes denote functionalization transition states. The overall barrier for each particular complex is the greater of the activation and functionalization barriers. 92 35.3 kcal/mol or below. Thus it appears that the III-II SR 2 pathway is the best choice for methyl functionalization. Hence, we conclude that the base Rh(bisq) rhodium-ligand complex is the best choice among the x entire Rh(bisq ) family, with an overall reaction barrier of 33.4 kcal/mol at 298 K and 35.8 kcal/mol at 498 K, when the III-II SR 2 functionalization pathway is used. 3.4.8 Product protection and C−H activation As mentioned in the introduction, a major hurdle that any putative methane to methanol catalytic scheme must overcome arises because the C−H bond dissociation energy of methanol is 9 kcal/mol weaker than that of methane. Preventing the overoxidation of the product is thus a challenge and is the reason we elected to investigate electrophilic metal centers in acidic solvent. Since the transition state of an electrophilic activation involves donation of electron density from the methane C−H σ bond to the metal, a highly electron-withdrawing trifluoroacetate substituent that decreases this σ bond electron density is expected to increase the transition state energy. Thus the highly electronegative trifluoroacetate group of methyl trifluoroacetate withdraws electron density from the methyl C−H bonds, thereby decreasing their ability to donate into the rhodium center and raising activation barriers. Indeed, we see this effect in explicit calculations with both the (NN) and (NNF ) ligand sets. Figures 3.8 and 3.9 show the various isomeric transition states for the activation of both methane and methyl trifluoroacetate in TFAH. At 298 K, each transition state for the activation of methane was 3 to 10 kcal/mol lower than the corresponding transition state for the activation of methyl trifluoroacetate (1 atm reference for both). The first two rows of Figure 3.10 show only the lowest energy isomer of each activation, and thus an overall ∆G‡ (lowest Me−TFA transition state minus lowest CH4 transition state) of 2.7 kcal/mol is found for (NN) and 5.5 kcal/mol for (NNF ). The numbers significantly improve to 6.4 and 9.8 kcal/mol at 498 K. This is due to the volatility of Me−TFA, which greatly raises its free energy at 498 K. F In contrast, activation of methanol (1 M reference state) by Rh(NN ), shown in Figure 3.11 and summarized in the third row of Figure 3.10, was even lower than that of methane: by 0.3 or 1.8 kcal/mol at 298 K when comparing the lowest CH3 OH transition state and the lowest CH4 transition state in TFAH (Figure 3.10, first and second rows); for (NNF ) in water this gap is larger at 2 to 3 kcal/mol (Figure 3.10, third row). We can explain this by noting that the hydroxyl group in methanol has electron donating properties and hence the donation effect of the C−H σ orbital into the electrophilic metal center is increased, leading to a decreased transition state barrier; however these numbers are lower than the 9 kcal/mol simple BDE difference between the C−H bonds in methane and methanol and comparable to work done by Owen et al. [23]. Product protection studies were also carried out for the Rh(DPMS) and Rh(DPES) complexes, 93 Figure 3.8. Transition states for the functionalization of methane, Me−TFA, and methanol using Rh(NN) complexes in TFAH. All free energies are in kcal/mol and relative to the lowest energy inorganic state (i.e. right before methane activation). 94 Figure 3.9. Transition states for the functionalization of methane, Me−TFA, and methanol using Rh(NNF ) complexes in TFAH. All free energies are in kcal/mol and relative to the lowest energy inorganic state (i.e. right before methane activation). 95 Figure 3.10. Summary of the data presented in Figures 3.8, 3.9, and 3.11. In each row, the leftmost structure is the lowest-energy activation transition state for methane (1 atm); the center structure is the lowest-energy transition state for the activation of MeTFA (1 atm, not applicable for the H2 O case); and the rightmost structure is the lowest-energy transition state for the activation of MeOH (1 atm). All free energies are in kcal/mol and relative to the lowest energy inorganic state (i.e. right before methane activation). 96 Figure 3.11. Comparison of activation energies for methane (1 atm) and for methanol (1 M at 298 K, 1 atm at 498 K) in the same structure for Rh(NNF ) complexes in water. In this case, methanol was easier to activate by 2 to 3 kcal/mol. All free energies are in kcal/mol and relative to the lowest energy inorganic state (i.e. right before methane activation). 97 the results of which are shown in Figures 3.12 and 3.13, respectively. Modest product protection ∆∆G‡ ’s of 1.4 kcal/mol and 2.2 kcal/mol at 298 K were found, but these improve to 5.3 kcal/mol and 6.6 kcal/mol at 498 K. Figure 3.12. Comparison of the activation states of methane and methyl trifluoroacetate by Rh(DPMS) complexes, showing the product protection afforded. Red denotes the most favored isomer of each type of activation. All free energies are in kcal/mol and relative to the resting state. x Finally, product protection studies were carried out for the Rh(bisq ) family of rhodium-ligand complexes. The results are summarized in Figure 3.14. In general the product protection is comparable to the Rh(NNx ) family at 298 K, but improves significantly by 3.8-4.2 kcal/mol at 498 K. As in the other cases, this is due to the high volatility of Me−TFA. Since the ease of activation decreases CH3 OH > CH4 > Me−TFA, it is important to consider the relative chemical potential of CH3 OH and Me−TFA in solution. As we have found that the − * equilibrium Me−TFA(g) + H2 O(solv) − ) − − CH3 OH(g) + TFAH(`) is exergonic by 1.3 kcal/mol at 298 K, it is predicted that these catalysts would be selective for methane oxidation only in the absence of water. In comparison with previous work, we note that Periana et al. [24] showed that the reason why the Catalytica-Periana bipyrimidine Pt catalyst [25] was able to achieve high selectivity is that the barrier to activate the Me−OSO3 H product is 14 kcal/mol higher than for CH4 or CH3 OH and the acidity of the medium drives the protection of methyl products via esterification. Although 98 Figure 3.13. Comparison of the activation states of methane and methyl trifluoroacetate by Rh(DPES) complexes, showing the product protection afforded. Red denotes the most favored isomer of each type of activation. All free energies are in kcal/mol and relative to the resting state. 99 Figure 3.14. Comparison of the activation states of methane and methyl trifluoroacetate by the Rh(bisqx ) family of complexes, showing the product protection afforded. Red denotes the most favored isomer of each type of activation. The overall product protection ∆∆G‡ energies are shown in blue. All free energies are in kcal/mol and relative to the resting state. 100 the Catalytica-Periana system distinguishes between methane and the methyl ester more than the rhodium complexes, we note that a key advantage to our system is the volatility of CH3 OH and Me−TFA (boiling points at 64.7◦ C and 43◦ C, respectively) relative to the nonvolatile Me−OSO3 H. Increasing selectivity beyond what is provided by the electrophilicity of the metal and esterification is still an important area of research. The addition of nonpolar ligand side-chains may decrease the propensity of methanol coordination to the metal relative to methane coordination. 3.5 Discussion Since our ligands have poor electron-donating ability and RhIII -centered cycles, we expect the metal center to be electrophilic with reactions proceeding along an electrophilic route [3]h. We can rule out oxidative addition due to the nature of our ligands and the general instability of RhV species; hence we expected that the most likely C−H activation pathway is through a base-mediated electrophilic mechanism (Scheme 3.21). Scheme 3.21. A schematic diagram showing the metal activation of an R−H bond, for example, methane. Activation occurs via ligand donation to the electrophilic metal center (a) and basic abstraction of the hydrogen (b). Because we propose an electrophilic pathway for the methane activation step [3]h, we expect there to be a relationship between the activation transition state energy and some aspect of the metal’s electropositivity. To find this relationship, we examined more deeply the calculated constituent x x energies of rhodium complexes with the (NN ), (ONN ), (DP M E S), and (PN) ligands and ligand families. We did not include the L3 (bisqx ) family because they require protonation before methane activation can take place; as a result a direct comparison of these ligands with the neutral transition states of our other ligands is not possible. To estimate the electropositivity, we initially used the calculated energy of the rhodium 4s core orbital in the RhIII resting state, which we compared to the corresponding methane activation energies (Figure 3.15). We found that there is approximately a positive correlation between increasing 101 electropositivity (as measured by negative numbers closer to zero) and increase overall transition state barriers. We explain this by noting that decreasing the metal’s electropositivity increases its electrophilicity and thus encourages σ donation of the Me−H bond. Figure 3.15. Graph of the methane activation energy of the various rhodium complexes vs. the 4s orbital energy of the rhodium in their resting states, based on data from Table 3.3. This calculated value is a proxy for the relative overall electropositivity of the specific rhodium-ligand complexes. In all cases the lowest transition state was used; this involves an axial Rh−Me being formed in the cases of the (NNx ) and (ONNx ) ligand families, and an equatorial Rh−Me being formed in the (DP M S) and (PN) cases. E In the hopes that a more directional aspect of a given rhodium-ligand’s electron density might give us a better correlation with the transition state barrier, we then plotted the methane activation energies with the energies of the Rh−C bond in the resultant RhIII −Me intermediate (Figure 3.16). These Rh−C bond energies were obtained by performing a Pipek-Mezey orbital localization procedure on the RhIII −Me complex’s electronic wavefunction [13]. We found that the correlation is much better, with increasingly negative Rh−C bond energies being correlated with decreasing transition state barriers. This may be due to the increasing favorability of forming the rhodium-methyl bond. We give the caveat that again the only outlier is the Rh(NN) ligand complex, but we note that our trend represents only an overestimation, and therefore does not include false positives. In addition, F F there is a definite positive correlation within a ligand family (i.e. Rh(NN) vs. Rh(NN ); Rh(ONN ) NMe2 vs. Rh(ONN) vs. Rh(ONN )). We conclude from this relationship that (a) given a rhodium-ligand complex that already shows some promise, we can further fine-tune its properties with additional modifications on the ligand to further lower the transition state barriers of interest; and (b) given a new rhodium-ligand complex, by calculating the RhIII −Me intermediates and extracting the Rh−C bond energy, we can get an estimate of the expected methane activation transition state barrier for that ligand set and thus gauge its worthiness for further investigation. 102 Ligand (NN) F (NN ) (ONN) F (ONN ) NMe2 (ONN ) (DPMS) (DPES) (PN) TSa 32.6 27.6 37.2 35.8 38.7 33.9 35.6 35.0 ∆Gbprecursor 1.3 2.2 3.9 4.0k 3.7 0.0 0.0 N/Am Precursorc 2– + + [Rh (NN)(TFA)4 ] (Hax )(Heq ) III F 2– + + [Rh (NN )(TFA)4 ] (Hax )(Heq )j III – + [Rh (ONN)(TFA)3 ] (Hax ) III F – + [Rh (ONN )(TFA)3 ] (Hax )l III NMe2 – + [Rh (ONN )(TFA)3 ] (Hax ) III – + [Rh (DPMS)(TFA)3 ] (Heq ) III – + [Rh (DPES)(TFA)3 ] (Heq ) III – + l [Rh (PN)(TFA)3 ] (HN ) Ligand (NN) F (NN ) (ONN) F (ONN ) NMe2 ) (ONN (DPMS) (DPES) (PN) Adj. TSd 31.3 25.4 33.3 31.8 35.0 33.9 35.6 N/Am Rh−C σ e −0.38096 −0.39557 −0.37870 −0.38498 −0.37739 −0.38700 −0.38539 −0.38515 Rh−Mef III 2– + + [Rh (NN)(Meax )(TFA)3 ] (Hax )(Heq ) III F 2– + + [Rh (NN )(Meax )(TFA)3 ] (Hax )(Heq )j III – + [Rh (ONN)(Meax )(TFA)2 ] (Hax ) III F – + [Rh (ONN )(Meax )(TFA)2 ] (Hax ) – + III NMe2 )(Meax )(TFA)2 ] (Hax ) [Rh (ONN III – + [Rh (DPMS)(Meeq )(TFA)2 ] (Heq ) III – + [Rh (DPES)(Meeq )(TFA)2 ] (Heq ) III – [Rh (PN)(MeN )(TFA)3 ] (HP ) III Rh−TFA 4sg −3.1858 −3.21571 −3.19023 −3.19724 −3.18448 −3.20873 −3.20332 −3.19499 Rh−Me 4sh −3.13265 −3.15195 −3.13949 −3.14661 −3.13750 −3.15703 −3.15567 −3.15474 ∆4si 0.05315 0.06376 0.05074 0.05063 0.04698 0.05170 0.04765 N/Am Table 3.3. Comparison of the various Rh parameters with the rhodium-ligands complexes’ associated transition state barriers. These numbers are plotted in Figures 3.15, 3.16, and 3.17. An “ax” subscript indicates that the moiety is axial with respect to the rhodium-ligand plane; An “eq” subscript indicates that the moiety is equatorial with respect to the rhodium-ligand plane. For (DP M S) the rhodium-ligand plane is E defined as the plane spanned by the N−Rh−N moiety. For (PN) complexes, An “N” subscript indicates that the moiety is equatorial with respect to the P−Rh−N plane and trans to the N; A “P” subscript indicates that the moiety is equatorial with respect to the P−Rh−N plane and trans to the P. (a) The overall transition state barrier, as reproduced from Table 3.2, in kcal/mol. (b) The free energy of the immediate precursor before the lowest methane activation barrier, in kcal/mol. This typically differs from the resting state only in the geometric placement of TFAH and TFA ligands. (c) The identity of the immediate precursor. (d) The “adjusted” TS barrier, i.e. the free energy change of the actual transition state itself, in kcal/mol. Equal to the overall TS − Gprecursor . (e) The energy of the Pipek-Mezey localized Rh−C bond of the RhIII −Me species formed as a result of the lowest transition state, in hartrees. (f) The identity of the RhIII −Me species formed as a result of the lowest transition state. (g) The 4s orbital energy of the Rh atom in the resting state, in hartrees. (h) The 4s orbital energy of the Rh atom in the RhIII −Me species in column f, in hartrees. (i) The change in Rh 4s energies, equal to the difference between the preceding two columns, in hartrees. (j) These species are the lowest neutral species, done for the sake of consistency with the other ligands. The anionic resting states are very similar in energy. (k) An estimated value based on the Gprecursor values of related species Rh(ONN) and Rh(ONNNMe2 ). (l) Stable conformations of these complexes could not be found due to the ease of isomerization of the TFAH/TFA ligands into axial/equatorial positions. (m) The value is unknown and an estimation was not attempted. 103 Figure 3.16. Graph of the methane activation energy of the various rhodium complexes vs. the Rh−C bond energy as localized by the Pipek-Mezey method [13] of the lowest-energy Rh−Me complex. This calculated value is a proxy for the relative directional electropositivity of the specific rhodium-ligand complexes towards the methyl bond being formed. Based on data from Table 3.3. In all cases the lowest transition state was used; this involves an axial Rh−Me being formed in the cases of the (NNx ) and (ONNx ) ligand families, and S) and (PN) cases. an equatorial Rh−Me being formed in the (DP M E We also wished to see if there was a correlation between the transition state barrier and the rhodium center’s hardness as modified by its ligand. Since the hardness of an atom or molecule is defined as the polarizability of its electron cloud [26], we measured the difference in the Rh 4s orbital energy going from the inorganic resting state to the RhIII −Me as a proxy for hardness. Although we did not find a correlation with the overall transition state barrier of activation, we did find a negative correlation with the “adjusted” TS barrier, i.e. the free energy change of the actual transition state itself (Figure 3.17). In other words, the activation transition state [Rh· · · CH4 ]‡ is derived from the replacement of a TFAH ligand with CH4 , but the immediate Rh−TFAH precursor is not the resting state, but differs slightly in its axial/equatorial arrangement of TFAH/TFA ligands and is a few kcal/mol higher in energy. Taking this out gives us our correlation. Hence, we see that increased polarizability or softness, as measured by the amount of energy increase in the rhodium-ligand system when a TFA ligand is exchanged for a methyl, correlates with a lower activation barrier. The result of our fine tuning has yielded, among those rhodium-ligand complexes with neutral F activation transition states, the (NN ) ligand set. With a methane activation energy of 27.6 kcal/mol at 298 K, it compares favorably with other homogeneous systems. Indeed, an effective activation barrier of 33.5 kcal/mol is derived from the turnover frequency of the Catalytica-Periana catalyst (bpym)PtCl2 in H2 SO4 [27]. Specifically, C−H activation by (η 3 -6-phenyl-2,20 -bipyridine)IrIII (TFA)(C2 H4 )(C2 H5 ) is shown to catalyze H/D exchange between CH4 and TFAD with a turnover frequency 104 Figure 3.17. A graph of the Rh−TFA to Rh−Me change in the 4s orbital energy, a measure of the Rh atom’s softness, versus the adjusted TS barrier (see text). Note that the (ONNF ) data point is an estimated value, which may explain its deviation from the monotonicity (see Table 3.3). In all cases the lowest transition state was used; this involves an axial Rh−Me being formed in the cases of the (NNx ) and (ONNx ) ligand families, and an equatorial Rh−Me being formed in the (DP M S) and (PN) cases. This graph is based on E data from Table 3.3. of 2.12 × 10−2 s−1 at 180◦ C (Scheme 3.18). From the turnover frequency a free energy barrier of 33.5 kcal/mol can be calculated using transition state theory [28]. In the course of these studies we found that it was more difficult to find effective routes of Rh-methyl functionalization than to find low methane activation barriers. We found that the III-II pathway (SR 2 attack with OVCl3 ) typically gives the lowest barriers, whereas the III-I and III-IVII pathway (SN 2 attack on RhIII or RhIV , respectively) have some use as well; the I-III pathway (oxidative addition of methane by RhI ) was much less likely. F We found that Rh(NN ) complexes in water lead to a flatter thermodynamic profile than in TFAH, increased transition state energies for methane activation, and decreased transition state energies for functionalization. We can understand these changes in the thermodynamics by noting that hydroxo ligands’ lone pairs stabilize higher oxidation states of rhodium. Overall these changes are favorable for activity, since the functionalization energy is lowered to 31.7 kcal/mol while the methane activation energy of 35.0 kcal/mol remains acceptable. Unfortunately, an aqueous solution is predicted to offer no product protection, a result consistent with the reactivity of platinum diimine catalysts [23]. We can understand the decreased barrier for SN 2 methane functionalization (III-I, III-IV-II) in terms of several factors: the increased stability of the RhIV starting material in water, the greater thermodynamic activity of water in aqueous solution (at 55 M) as opposed to the TFA anion in 105 Figure 3.18. The calculated free energy (at 298 K, 1 atm CH4 ) of the (NNC)Ir(TFA)2 system, via the same internal substitution mechanism applied to RhIII complexes. TFAH, and the fact that the transition state involves a concerted reaction featuring simultaneous formation of a C−O bond, deprotonation to form neutral methanol, and protonation of a hydroxo ligand on the Rh complex. However, we remain unclear about the reason behind the decreased SR 2 functionalization barrier for III-II. We speculate that, since some transition states for SN 2 attack by OVCl3 (to form Me−OVCl–3 ) were found and appear to be about 10 kcal/mol higher than their SR 2 equivalents, there may be nucleophilic character on the oxygen atom in OVCl3 SR 2 attack as well. The most important hurdle that must be cleared experimentally in order for us to have a viable catalytic system is catalyst stability. The trend we have observed, that RhIII complexes with less electron-donating ligands tend to have lower activation and functionalization barriers, must be balanced with the consideration that a ligand that is too electron-poor may not have sufficient binding strength for a stable complex with rhodium to be made. The next logical step of our investigation is F the experimental synthesis and stability study of the RhIII (NN ) and RhIII (bisq) complexes. However, even if we fail to see good robustness for this particular complex, our more valuable contribution is that we have undertaken the most detailed ab initio study of the requirements of a Rh methane activation catalyst to date. 3.6 Conclusions F In our QM virtual screening of potential methane to methanol catalysts, we identified the Rh(NN ) F and Rh(bisq) complexes as highly promising candidates. Features of the Rh(NN ) complex include: • Transition state barriers for methane activation at 298 K of 27.6 (TFAH) and 35.0 kcal/mol (water), and • Transition state barriers for functionalization at 298 K of 36.8 (TFAH) or 31.7 kcal/mol (water), 106 and features of the Rh(bisq) complex include: • Transition state barriers for methane activation at 298 K of 33.4 (TFAH), and • Transition state barriers for functionalization at 298 K of 32.0 (TFAH). These results are better than any other RhIII complex we have so far investigated. Moreover, we provide a rational accounting for the role that ligand design plays in the effectiveness of this promising catalyst. In particular for RhIII complexes with neutral methane activation transition states, increased electron-withdrawing ligands leads to lower barriers for both activation and functionalization. We can estimate the overall transition state barrier by simply calculating the RhIII −Me species and extracting the Rh−C bond energies. Thus, the only condition opposing x even less donating ligands (i.e. lower barriers) is catalyst stability. For the RhIII (bisq ) family of complexes, requiring one coordination site to be a weak η 2 -benzene interaction greatly facilitates functionalization on a methyl group trans to it. These results from QM virtual screening are now ready for experimental testing, validation, and improvement. It can also be anticipated that future design of addition ligands may incorporate both concepts, with a weak interaction as well as an anionic or bidentate ligand, so that both activation barriers can be easily prediction and functionalization barriers minimized. 3.7 Acknowledgments This work was supported as part of the Center for Catalytic Hydrocarbon Functionalization, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DOE DE-SC0001298. 107 3.8 References [1] Cheng, W-H.; Kung, H. H. Methanol Production and Use (Chemical Industries vol. 57), 1994. [2] (a) Labinger, J. A.; Bercaw, J. E. Nature 2002, 417, 507–514; (b) Crabtree, R. H. J. Chem. Soc., Dalton Trans. 2001, 2437–2450; (c) Labinger, J. A. Fuel Proc. Technol. 1995, 42, 325–338; (d) Labinger, J. A. J. Mol. Catal. A: Chem. 2004, 220, 27–35. [3] (a) Muehlhofer, M.; Strassner, T.; Hermann, W. A. Angew. Chem. Int. Ed. 2002, 41, 1745–1747; (b) Sen, A.; Benvenuto M. A.; Lin M. R.; Hutson A. C.; Basickes N. J. Am. Chem. Soc. 1994, 116, 998–1003. [4] (a) Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154–162; (b) Hanson, S. K.; Heinekey, D. M.; Goldberg, K. I. Organometallics 2008, 27, 1454–1463; (c) Evans, M. E.; Jones, W. D. Organometallics 2011, 30, 3371–3377; (d) Zakzeski, J. J.; Bell, A. T. J. Mol. Cat. A: Chem. 2007, 276, 8–16; (e) Zhang, X. X.; Wayland, B. B. J. Am. Chem. Soc. 1994, 116, 7897–7898; (f) Tenn, W. J. III; Conley, B. L.; Bischof, S. M.; Periana, R. A. J. Organomet. Chem. 2011, 696, 551–558; (g) Rhinehart, J. L.; Manbeck, K. A.; Buzak, S. K.; Lippa, G. M.; Brennessel, W. W.; Goldberg, K. I.; Jones, W. D. Organometallics 2012, 31, 1943–1952; (h) Li, L.; Brennessel, W. W.; Jones, W. D. Organometallics 2009, 28, 3492–3500; (i) Kloek, S. M.; Heinekey, D. M.; Goldberg, K. I. Angew. Chem. Int. Ed. 2007, 46, 4736–4738. [5] Jaguar, version 7.6; Schrödinger, LLC: New York, NY, 2007. 108 [6] (a) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.; Sitkoff, D.; Nicholls, A.; Ringnalda, M.; Goddard, W. A.; Honig, B. J. Am. Chem. Soc. 1994, 116, 11875–11882; (b) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.; Ringnalda, M. N.; Sitkoff, D.; Honig, B. J. Phys. Chem. 1996, 100, 11775–11788. [7] (a) Becke, A. D. Phys. Rev. A 1998, 38, 3098–3100; (b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652; (c) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B 1988, 37, 785–789. [8] (a) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215–241; (b) Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157–167. [9] Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310. [10] Martin, J. M. L.; Sundermann, A. J. Chem. Phys. 2001, 114, 3408–3420. [11] (a) Hehre, W. J.; Ditchfield, R; Pople, J. A. J. Chem. Phys. 1972, 56, 2257–2261; (b) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; Defrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654–3665. [12] (a) Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. V. J. Comput. Chem. 1983, 4, 294–301; (b) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72, 650–654. [13] Pipek, J.; Mezey, P. G. J. Chem. Phys. 1989, 90, 4916–4926. [14] Kelly, C. P.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2006, 110, 16066–16081. [15] Tissandier, M. D.; Cowen, K. A.; Feng, W. Y.; Gundlach, E.; Cohen, M. H.; Earhart, A. D.; Coe, J. V.; Tuttle, T. R. J. Phys. Chem. A 1998, 102, 7787–7794. Calculated at standard conditions. [16] Wertz, D. H. J. Am. Chem. Soc. 1980, 102, 5316–5322. [17] Wagman, D. D.; Evans, W. H.; Parker, V. B.; Schumm, R. H.; Halow, I.; Bailey, S. M.; Churney, K. L.; Nuttall, R. L. J. Phys. Chem. Ref. Data 1982, 11, 1–14. [18] For trifluoroacetic acid: Kreglewski, A. Bull. Acad. Pol. Sci. Ser. Sci. Chim. 1962, 10, 11-12, 629–633. For water: (a) Bridgeman, O.C.; Aldrich, E.W. J. Heat Transfer 1964, 86, 279–286; (b) Liu, C.-T.; Lindsay, W.T., Jr. J. Chem. Eng. Data 1970, 15, 510–513. 109 [19] Holleman, A. F.; Wiberg, E.; Wiberg, N. Lehrbuch der Anorganischen Chemie, (91–100 ed.;) Walter de Gruyter, 1985; pp. 1056–1057. [20] Goldberg, K. I.; Yan, J. Y.; Breitung, E. M. J. Am. Chem. Soc. 1995, 117, 6889–6896. [21] Cheng, M. J., Nielsen, R. J., Goddard, W.A. Manuscript under preparation.   ‡ . [22] Using k = kBhT · exp −∆G kB T [23] Owen, J. S.; Labinger, J. A.; Bercaw, J. E. J. Am. Chem. Soc. 2006, 128, 2005–2016. [24] Periana, R. A.; Taube, D. J.; Evitt, E. R.; Loffler, D. G.; Wentrcek, P. R.; Voss, G.; Masuda, T. Science 1993, 259, 340–343. [25] Periana, R. A.; Taube, D. J.; Gamble, S.; Taube, H.; Satoh, T.; Fujii, H. Science 1998, 280, 560–564. [26] Anslyn, E.; Dougherty, D. Modern Physical Organic Chemistry; University Science Books, 2006. [27] Young, K. J. H.; Oxgaard, J.; Ess, D. H.; Meier, S. K.; Stewart, T.; Goddard, W. A.; Periana, R. A. Chem. Commun. 2009, 3270–3272. [28] According to transition state theory, the activation energy ∆G‡ may be derived from the   ‡ . turnover frequency TOF using the formula TOF = kBhT · exp −∆G kB T  110 Chapter 4 The Reduction-Coupled Oxo Activation mechanistic motif for C−H bond activation and oxidation Reproduced in part with permission from: 1. Cheng, M.-J.; Fu, R.; Goddard, W. A. III. Chem. Commun. 2014, 50, 1748–1750. c 2014 The Royal Society of Chemistry; available at http://dx.doi.org/10.1039/C3CC47502F; 2. Cheng, M.-J.; Goddard, W. A. III.; Fu, R.; Nielsen, R. J. Top. Catal., in press. Unpublished work c 2014 Springer. Also includes original work. 4.1 Abstract The vanadium phosphorus oxide (VPO) catalyst is used commercially for the selective oxidation of n-butane to maleic anhydride. Surprisingly, the mechanism has been shown to initiate with C−H activation by a P−O moiety. In this chapter, we call this phenomenon the Reduction-Coupled Oxo Activation (ROA) mechanistic motif, and attribute it to the concomitant formation of a strong oxo-hydrogen bond on the phosphorus atom and a one-electron reduction on the vanadium atom. In addition, we show that the same ROA phenomenon can be extended beyond phosphorus and vanadium to additional main-group elements and transition metals, respectively, and may lead to a new strategy in designing selective catalysts for alkane activation. Finally, we propose the existence of a homogeneous vanadium phosphorus oxo complex that can incorporate the novel ROA mechanistic pathway as a potential propane oxidation catalyst. We considered several oxidation pathways leading to end products such as propylene, isopropanol, and propylene oxide. We support our assertions with density functional theory calculations on potential reaction pathways. 111 4.2 Introduction 4.2.1 The VPO catalyst for n-butane oxidation Vanadium phosphorus oxide (VPO) is an inorganic complex that catalyzes the oxidation of nbutane to maleic anhydride with a surprisingly high selectivity of 60-70% [1] (Scheme 4.1). Although the overall yield is only ca. 50%, its ease of synthesis, use of common elements, and selectivity have allowed this catalyst to be commercialized, producing ca. 500 kilotons of maleic anhydride annually [2]. Hence, there was a strong impetus to investigate the mechanism of this reaction in the hopes of rationally designing a modified catalyst with improved yield. Scheme 4.1. The oxidation of n-butane to maleic anhydride using vanadium phosphorus oxide as a catalyst. A useful parameter for the C−H activating ability of each oxo moiety in the catalyst is the H-atom affinity DH , which is defined as the enthalpy change when the O−H bond in ·M−O−H is homolytically cleaved to form the M−O oxo. Mechanistic work initially centered on vanadyl pyIV rophosphate ((V O)2 (P2 O7 ), abbreviated VOPO), the major component of VPO [3]. However, it was quickly found that no vanadium oxo on the VOPO surface has sufficient H-atom affinity to activate n-butane (Figure 4.1). The most active V−O in VOPO has a DH value of 58.2 kcal/mol, whereas the weakest C−H bond in n-butane has a dissociation energy of 89.3 kcal/mol, implying an extremely endothermic first step that is categorically inconsistent with the experimentally determined activation energy of 12.9-23.6 kcal/mol [4]. Hence, it was concluded that the reduced vanadyl IV pyrophosphate (V O)2 (P2 O7 ) alone is unable to initiate the catalytic process. Figure 4.1. O−H bond strength for oxygen atoms on the (VIV O)2 (P2 O7 ) surface. The energy released when an H · is bound to the V−O (in red) is only 58.2 kcal/mol, not nearly high enough for alkane activation to occur. The energy released when an H · is bound to the bridging V−O−P oxide (in green) is even lower, at 17.0 kcal/mol. The energy released when an H · is bound to the P−O was not determined because geometry optimization caused it to be transferred to the V−O−P moiety (in green). It was then supposed that catalysis requires the presence of adsorbed O2 on the VIV O(P2 O7 ) 112 surface. Adsorption of O2 may be molecular, in which the O−O bond is retained, two VIV centers are oxidized to VV , and a vanadium peroxo moiety is formed; or it may be dissociative, in which the O−O bond is broken and four VIV centers are oxidized to VV . The first case is uphill by 2.8 kcal/mol, whereas the second is downhill by 38.8 kcal/mol. Unfortunately, the DH values calculated for the resulting vanadium peroxo and vanadium oxo moieties were only 70.0 and 58.9 kcal/mol, respectively. Hence, it was concluded that the reduced VPO complex is not the active catalyst. Figure 4.2. Oxygen adsorption energies on the (VIV O)2 (P2 O7 ) surface and the O−H bond strength (DH ) to each of these O atoms. Left: molecular O2 adsorption in which the O−O bond is preserved. Right: dissociative adsorption in which the O−O bond is broken. All energies are in kcal/mol. Finally, it was proposed that catalysis takes place on a metastable VV OPO4 surface, known as the X1 phase, which has been experimentally determined to be produced in the high oxygen environment of the reaction conditions [5]. However, the largest DH value found for a vanadium oxo in this phase is only 62.8 kcal/mol (Figure 4.3). Surprisingly though, it was discovered that the phosphorus oxo had a much larger DH value of 84.3 kcal/mol! This implies a first step that is endothermic by only 5.0 kcal/mol, with an activation energy later calculated to be 13.6 kcal/mol, which is within the experimental range. The stability of the resulting monohydrogenated species was attributed to the fact that spin density appears to be localized on the neighboring vanadium atoms (Figure 4.4). 4.2.2 The ROA mechanistic motif The surprising discovery of the C−H activation ability of PV −O moieties in the VV OPO4 catalyst led to the creation a new term, Reduction-Coupled Oxo Activation (ROA), to describe this phenomenon. Its usefulness is due to the observation that the basicity of simple transition metal oxos increases with decreasing electron affinity of the metal, due to increasing electron density; whereas for C−H activation it is preferred that both basicity and electron affinity are maximized, since H-atom transfer 113 Figure 4.3. O−H bond strengths for the various oxygen atoms on the X1-VV OPO4 surface. The O−H bond strength of 84.3 kcal/mol to the P−O moiety leads to a first step that is endothermic by only 5.0 kcal/mol, with an activation energy later calculated to be 13.6 kcal/mol. Hence, the P−O moiety is the only candidate site whose C−H activation energy is compatible with experimental data. Figure 4.4. C−H activation by a phosphorus oxo leads to a one electron reduction on the neighboring vanadium. Hence the oxidation state of the phosphorus atom does not change. (a) Vanadium-phosphate archetype: conceptual arrow-pushing scheme for P−V electron transfer with an intervening µ-oxo. (b) Vanadium-κ−Pphosphinite archetype: conceptual arrow-pushing scheme for P−V electron transfer in which they are directly bound to each other. 114 involves both proton and electron transfer. The ROA mechanistic motif, by keeping the sites of basicity and reduction on separate moieties, is able to decouple this process and hence allow the best of both worlds with high basicity and high electron affinity [6]. The validity of the ROA mechanistic motif was tested using quantum mechanical calculations on small X4 O10 clusters, where each X is a P or V. The results, shown in Figure 4.5, indicate that a single linkage to V makes the P−O bond 32.3 kcal/mol more active, while coupling to a second V makes it an additional 23.8 kcal/mol more active and coupling to a third V (as in VPO) makes it an additional 4.9 kcal/mol more active. Compared to a pure vanadium (V) oxide case, coupling to one or two P atoms makes it 13.4 and 8.5 kcal/mol more active, respectively. Figure 4.5. Use of finite cluster models to investigate how the O−H bond strength of P−O moieties varies based on the number of V coupled to it via O linkages. Data is summarized from Figure 4.8. All numbers are in kcal/mol. a. b. c. d. e. Species P4 O10 P3 VO10 P2 V2 O10 PV3 O10 V4 O10 DPO(1)−H 28.2 60.5 84.3 89.2 75.8 The data presented in Figure 4.5 shows the validity of the small cluster approach in gauging the activity of the ROA mechanism. In hopes of expanding the chemical space in which this ROA mechanism is available, we have undertaken an investigation of mixed oxos with vanadium and the heavier group 15 (pnictogen) and 16 (chalcogen) elements in the first part of this chapter. (We do not consider group 14 or 17 elements due to the instability of terminal oxos for such elements, against hydration in the group 14 case and explosion in the group 17 case.) For each of these complexes, we have calculated DH , the oxo-hydrogen bond energy. We use DH as a proxy for hydrogen-abstraction potential, which is typically the most difficult step in alkane oxidation/functionalization catalysis. We have found that the ROA mechanism is general to group 15 and 16 (pnictogen and chalcogen) vanadium mixed oxos, with phosphorus, selenium, and tellurium being the most active. We have also taken these most active cases, with molybdenum (VI) substituted for vanadium (V), and found similarly promising results. We hence conclude that the reduction-coupled oxo activation mechanistic motif is a viable and interesting new lead for the development of new oxidation catalysts. 115 4.2.3 Potential homogeneous catalysts utilizing the ROA principle Although the ROA phenomenon was originally discovered in the VPO system, a heterogeneous catalyst, the principle should hold for homogeneous catalysts as well. Homogeneous catalysts have the advantages of being well-defined, easier to study, and thus more readily able to contribute to basic scientific research. However, homogeneous complexes containing a P(O)−O−V(O) moiety (as shown in figure 4.4a) are very rare in the literature, owing to their difficulty in synthesis [7]. The only known examples are complexes produced by Herron et al. [8] and Cummins et al. [9], both of which were simply precursors to the production of VOPO or related network complexes; and the very recent work by Doyle et al. [7] that features the divanadium pyrophosphate complex IV {[(V O)bipy(H2 O)]2 (µ−P2 O7 )} · 3 H2 O, which exhibits some catalytic oxidative ability with benzyl alcohol. Hence, potential catalysts incorporating an alternative metal core featuring a direct V−P bond (as in figure 4.4b) were explored. Complexes featuring such κ−P phosphinite coordination to metals including Co [10], Ni [11]a, Pd [11], and Pt [12]a have been described. In addition, many other complexes containing a phosphido ligand bonded to metals such as Co [10], Rh [12]b, Pd [12]bc, Ir [13], and Pt [12]abd have also been described. These phosphido complexes are a single monooxygenation step away from the corresponding phosphine oxide complexes, a transformation which is expected to be facile due to the large formation energy of phosphine-oxo bonds and which has in fact been reported for a few of the complexes [10, 12]a. However, since in all these complexes the metal center has been from the late row transition metals, they are not predicted to be effective alkane activation catalysts under the ROA mechanism due to their large number of d electrons. In order to design a homogeneous oxidation catalyst that utilizes the ROA mechanistic motif, a natural choice for the metal center of a phosphine-oxide-type complex would once again be vanadium (V), due to its activity in the VPO catalyst and its status as an effect d0 one-electron acceptor. The existing metal κ−P phosphinite and phosphido complexes all feature shared traits such as a late transition metal in a low oxidation state and a phosphorus moiety incorporated as the center of a tricoordinate pincer ligand with flanking phosphines and aryl linkers. Such ligands, which were designed for late transition metals, would likely not form very stable complexes with an early transition metal such as vanadium. However, replacement of the ligand’s side phosphines with harder oxygens would be expected to greatly improve stability. Hence, we propose the ligand bis(2-phenoxyl)phosphinite κ−P coordinated on vanadium, abbreviated (OPO)V, as our candidate catalyst, in analogy to the existing Pd complex [11] (Figure 4.6). The robustness of this complex is predicted based on factors such as the chelation effect, the stability of O−V bonds, and the rigidity of the ligand, which guards against insertion of small activating molecules into the V−P bond. We complete the first coordination sphere around V by noting that it prefers to be six-coordinate and to contain a strong vanadium oxo bond [8], and we fill in the last open coordination sites with aqua ligands. 116 Figure 4.6. Left: known complexes featuring a phosphinite ligand bound κ−P to a transition metal. Right: A proposed vanadium κ−P phosphinite complex (OPO)VOL2 . In aqueous solution or when water is present in the system, L is H2 O. Note that both fac and mer stereoisomers are possible. We thus propose, in the second part of this chapter, that the hypothetical complex diaqua[bis(2V phenoxy)phosphinite-κ−P]oxidovanadium (V) ((OPO)V O(H2 O)2 ) utilizes the reduction-coupled oxo activation (ROA) mechanistic motif as a catalyst for the oxidation of alkanes in air. In this proposition, we support our assertion with density functional theory (DFT) calculations of the various reaction paths starting from propane, our representative alkane. We describe the most plausible transition state barriers for the conversion of propane and dioxygen to oxidized products such as propylene and isopropanol. Through our theoretical findings we are confident that a realized molecule will hold catalytic ability in a manner heretofore unseen in homogeneous catalysis. 4.3 Materials and methods All quantum mechanical calculations were carried out using the Jaguar software version 7.9 developed by Schrödinger Inc. [14]. Geometry optimizations were carried out on initial guess structures, and vibrational frequencies were gathered to confirm the optimized geometries as intermediates or transition states and to construct a free energy profile. The temperature was set to be 298.15 K. Solvation energies were calculated in water using the PBF Poisson-Boltzmann implicit continuum solvation model [15] in Jaguar, with a dielectric constant of 80.37 [16] and a probe radius of 1.40 Å [17] for water. Two explicit solvating waters per aqua, hydroxo, or oxo ligand were added for more accurate solvation modeling. For our proposed molecular catalysts, geometry optimization and vibrational data were calculated using the B3LYP density functional [18] with a smaller basis set, whereas single point gas-phase and solvated energies were calculated using the same functional and a larger basis set. Here the smaller basis set consists of a modified double-ζ Los Alamos basis set and pseudopotential (LACVP**) [19] for transition metals, the uncontracted Los Alamos valence-only basis set and pseudopotential (LAV3P) [20] for heavy main-group elements (e.g., As, Se, Sb, Te, Bi), and the 117 6-31G** basis set [21] for the other atoms; whereas the larger basis set consists of the triple-ζ Los Alamos basis set and pseudopotential (LACV3P**++) for transition metals, the same LAV3P basis set and pseudopotential for heavy main-group elements, and the 6-311G**++ basis set [21] for the other atoms. Proposed intermediates and transition states were found to have zero and one negative frequencies, confirming their status as local energy minima and saddle points, respectively. The pH of the aqueous solution was set at 0 or 7; this corresponds to the free energy of a solvated proton being set as −270.28 or −279.80 kcal/mol, respectively [22]. The free energy for each molecular species in solution was calculated using the formula G = Egas + ∆Gsolv + ZPE + Hvib + 6kT − T [Svib + γ(Strans + Srot − α) + β] where the last term (with α = 14.3 e.u., β = 7.98 e.u., and γ = 0.54) is an empirical approximation for the change in the translational and rotational entropy of the molecule between the gas phase and the solution phase (due to the finite librational frequencies), derived from Wertz [23]. For gas phase molecules (H2 , O2 , propylene, propane, and for our purposes the isopropyl radical), we assumed that equilibration between the dissolved gas and the headspace occurs at a much faster timescale than the reactions in question; leading to ∆Ggas→solv = 0. Thus, the free energy of such gas molecules can be simply calculated using the formula G = Ggas = Egas + ZPE + Htot − T Stot . For water, the Gibbs free energy was calculated using the formula Gliquid = Egas + ZPE + Htot − T Stot + ∆Ggas→liquid where ∆Ggas→liquid = Gliquid − Ggas (1atm) is the free energy of condensation to liquid from 1 atm gas. We can solve for this by noting that ∆Ggas→liquid = ∆Gexpansion + Ggas→solvation , where ∆Gexpansion = Ggas (P ) − Ggas (1 atm) is the expansion of the gas from 1 atm to the vapor pressure P , and Ggas→solvation is the condensation of gas at vapor pressure P to liquid. Since a liquid is by definition at equilibrium with its vapor pressure Ggas→solvation = 0, and we thus have  ∆Ggas→liquid = Ggas (P ) − Ggas (1 atm) = RT ln P 1 atm  . The vapor pressure of water at 298.15 K is 23.75 mmHg [24]. For a given species X and its monohydrogenated analogue XH · , the DH value is defined as 118 DH = H(X) + H(H · ) − H(XH · ), and the GH value is defined analogously. For our small cluster calculations, only gas phase energies and frequencies were calculated using the smaller basis set. 4.4 Validation of ROA on small oxide clusters We performed our calculations on small oxide clusters (Figure 4.7), which we used as models for large inorganic network systems. Restricting our investigation to small discrete clusters allowed us to perform highly accurate ab initio quantum mechanical computations. Figure 4.7. General models for small oxide clusters. (a) X is a pnictogen in the +V oxidation state; (b) Z is a chalcogen in the +VI oxidation state; (c) Z is a chalcogen in the +IV oxidation state. The oxidation state of each vanadium is +V before hydrogen atom abstraction and either +V or +IV after a hydrogen atom has been adducted. We begin with X = V, an all-transition metal cluster that will serve as a control. The first column of Figure 4.8 shows the structure of both the non- and monohydrogenated forms. Note that the DH value of 75.8 kcal/mol is already significantly high, owing to the naturally effective oxidizing potential of vanadium in the +V oxidation state. Replacing one vanadium atom with phosphorus (Figure 4.8, second column), we find that the DH energy has strengthened by 13.4 kcal/mol, to 89.2 kcal/mol. This is the canonical example of our reduction-coupled oxo activation mechanism. Replacing a second vanadium with phosphorus (Figure 4.8, third column) gives a slightly reduced, but similar DH value of -84.3 kcal/mol. Compared to the experimental results in [3], in which the phorphoryl site has a DH value of 84.3 kcal/mol, 119 Figure 4.8. Optimized structures of the Px V4−x O10 species (first row) (x ∈ {0, 1, 2, 3}) and their schematic representations (second row); optimized structures of the monohydrogenated Px V4−x O10 H species (third row) and their schematic representations (fourth row). Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. 120 these numbers are a good validation for our cluster model. We also examined the P2 V2 O10 H2 cluster, i.e. the dehydrogenated version. As the first row of Figure 4.9 indicates, regardless of the location of the hydrogen atoms, it is always the two vanadiums that are reduced to +IV. However, it is preferable by about 12 kcal/mol to attach the hydrogens onto the phosphoryl groups, due to the special stability of the vanadyl unit. The unpaired spins on each VIV are essentially independent, as there is almost no difference in energy or geometry between the singlet and triplet in both cases. We found that addition of a third hydrogen to the system always reduces one vanadium atom to +III, leading to either an overall doublet (doublet VIV and singlet VIII ) or an overall quartet (doublet VIV and triplet VIII ). The phosphorus atoms, in a sign of their resistance to reduction, stay in the +V oxidation state in all cases. The lack of VV centers compared to the case in Figure 4.8 accounts for the significant reduction in DH values. Figure 4.9. Optimized structures of the dihydrogenated P2 V2 O10 H2 species (first row), their schematic representations, and their energy relative to the lowest isomer; optimized structures of the trihydrogenated P2 V2 O10 H3 species (second row). Only the triplet and quartet structures are shown because the singlet and doublet structures are almost identical to them. Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. Having established the validity of our model, we now endeavored to survey the span of the group 15 (pnictogen) vanadium (V) oxides. We found that when going down the periodic table, from phosphorus to arsenic to antimony, the DH energy decreases (Figure 4.10). Bismuth, the last 121 element in the series, has an DH that is slightly higher than antimony above it, but it is still not as effective as a pure vanadium-oxygen cluster and furthermore is too susceptible to reduction to the +III state. Figure 4.10. Optimized structures of the XV3 O10 species (first row) (X ∈ {P, As, Sb, Bi} and their monohydrogenated XV3 O10 H counterparts (second row). Schematic structures are as in Figure 4.7a. Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. We also examined the potential group 16 (chalcogen) vanadium (V) oxides, in clusters as depicted in Figure 4.7bc. We found that for sulfur, selenium, and tellurium in the maximum +VI oxidation state, all structures had DH values of 90 kcal/mol or greater (Figure 4.11). We view this as a very promising potential direction for new catalysts. It is well documented that tellurium prefers higher coordination numbers; for example, unlike its lighter cogeners, the hydrated H6 TeO6 is stable whereas the four-coordinate H2 TeO4 is unknown. Examining several potential structures for the hydrated complex Te2 V2 O12 H2 , we found that the lowest energy conformation had a hydration energy of 6.3 kcal/mol downhill. The resultant structure (Figure 4.12) has a bridging µ-oxo that renders the two tellurium atoms five coordinate, and a proton on an oxygen atom on each tellurium. The lowest conformation of the corresponding monohydrated complex Te2 V2 O12 H3 gave a DH value of 92.0 kcal/mol, very comparable to the dehydrated complex. Upon hydrogen atom abstraction the µ-oxo breaks, resulting in only one of the tellurium atoms remaining five-coordinate. Unfortunately, it appears that there is a strong drive to fully hydrate the complex until both 122 Figure 4.11. Optimized structures of the Z2 V2 O11 species (first row) (Z ∈ {S, Se, Te}) and their monohydrogenated Z2 V2 O11 H counterparts (second row). Schematic structures are as in Figure 4.7b. Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. Figure 4.12. Optimized structures of the hydrated species Te2 V2 O12 H2 and Te2 V2 O12 H3 , along with their schematic representations. Mulliken spin density was used to assign radical character. DH = 92.0 kcal/mol. 123 tellurium atoms are six-coordinate: Te2 V2 O11 + 3 H2 O −−→ Te2 V2 O14 H6 , ∆E = −49.2 kcal/mol. The six-coordinate complex lacks terminal tellurium oxos that can easily abstract hydrogen atoms. Hence, the DH values are much less, diminished to about 60 kcal/mol. Attachment of a hydrogen atom onto a tellurium hydroxide results in water loss, and is not energetically more favorable than attachment onto a bridging oxo (Figure 4.13). The implication of this is that any putative catalyst incorporating tellurium in the +VI oxidation state will be poisoned by water and will need to be kept in dry conditions. Figure 4.13. Optimized structures of the fully hydrated species Te2 V2 O14 H6 and Te2 V2 O14 H7 (first row), along with their schematic representations (second row). Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. As SeVI and TeVI are both easily reduced to the +IV oxidation state, we also investigated whether vanadium oxide clusters incorporating SeIV and TeIV might also have good hydrogen abstraction capability. Using the structures in Figure 4.7c to build our model, we found that the complexes Se2 V2 O9 and Te2 V2 O9 are also very strong hydrogen atom abstractors (Figure 4.14). In fact, Te2 V2 O9 has a DH value of 100.0 kcal/mol, the strongest value we have seen. We note that abstraction of one hydrogen atom on a tellurium oxo causes the other tellurium oxo to convert to a bridging oxo, rendering the first tellurium five-coordinate. This is not seen in the selenium case. We also investigated the activity of vanadium oxide clusters with the group 14 elements Si, Ge, Sn, and Pb in the +IV oxidation state (Figure 4.15). Construction of a cluster model analogous to the templates found in Figure 4.7 results in the complexes Z2 V2 O10 H2 , where Z = Si, Ge, Sn, or Pb, which are shown in the first row of Figure 4.7. Note that unlike the group 15 and 16 cases, the base Z2 V2 O10 H2 clusters do not have any terminal oxo functionalities. This is reasonable and consistent 124 Figure 4.14. Optimized structures of Se2 V2 O9 and Te2 V2 O9 and their schematic representations (first row); optimized structures of the monohydrogenated Se2 V2 O9 H and Te2 V2 O9 H (second row). Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. 125 with the common knowledge that group 14 elements other than carbon greatly prefer two single bonds with oxygen over one double bond. Unfortunately, the lack of such terminal oxos results in the inapplicability of the ROA mechanism towards these types of compounds. Hence, the location on each Z2 V2 O10 H2 species that results in the greatest amount of energy released when H · is bound is a V−O moiety. The DH are similar to that of V4 O10 (Figure 4.8). Placement of the H · atom on a Z−OH moiety or bridging oxygen resulted in decomposition of the cluster for the cases of Z = Ge, Sn, and Pb; and led to much smaller DH values for the case of Z = Si. Hence we conclude that the ROA mechanistic motif does not likely extend to the group 14 elements. Although our work presented up to now have all involved vanadium as the transition metal that absorbs the spin density, we were also desirous of expanding the breadth of our ROA principle to other transition metals. Hence, we decided to take our best cases, involving phosphorous, selenium, and tellurium, and substitute in MoVI for VV . Unfortunately, we were not able to find a stable structure for monohydrogenated selenium complexes. Our data for phosphorous and tellurium are presented in Figure 4.16. It can be seen from the data in Figure 4.16 that the activity of phosphorus and tellurium (VI) species are comparable to those with vanadium, but the tellurium (IV) species is ineffectual. This may be due to the two tellurium atoms being too far apart for an oxygen atom to bridge in the same manner as the vanadium case. Nevertheless, we have demonstrated that the ROA pathway which characterizes vanadium (V) chemistry can also be extended to molybdenum (VI). Whereas calculations on the periodic system appear to show that the spin is delocalized [3], for our cluster calculations, in all cases the spin is not shared, but localized on one atom. This discrepancy may be explained by noting that those prior calculations were made using the PBE density functional, which has been documented to have a bias towards delocalized systems. Regardless, we were able to see the same increased DH effect without spin delocalization, so that cannot the physical reason behind our ROA mechanism. Given the observation that main group-vanadium oxide clusters with only certain elements have an increased hydrogen abstraction energy relative to a vanadium oxide cluster alone, it would appear that increases in hydrogen abstraction energy via the ROA motif is element-specific and not due to any general chemical principle. For pnictogen vanadium oxide clusters, the only viable candidate was phosphorus, whereas all the chalcogens exhibited similar high activity, especially selenium and tellurium in the +IV oxidation state. These observations cannot be explained by simple periodic trends. Similar complexes with group 14 elements were not included in this chapter due to the lack of notable positive results. This may be for the same reason as why a fully hydrated tellurium (VI) complex was also poor, due to the lack of an oxo ligand. 126 Figure 4.15. Optimized structures of the group 14-vanadium oxide species Z2 V2 O10 H2 , where Z ∈ {Si, Ge, Sn, Pb} (first row); along with their schematic representations (second row). Optimized structures of the corresponding monohydrogenated species (third row) and their schematic representations (fourth row). Alternative locations for H · addition, as well as the corresponding DH energies, are shown in the fifth row for Z = Si. Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. 127 Figure 4.16. Optimized structures of the molybdenum-containing species P2 Mo2 O11 , Te2 Mo2 O10 , and Te2 Mo2 O12 (first row), along with their schematic representations (second row); optimized structures of the corresponding monohydrogenated species (third row) and their schematic representations (fourth row). Mulliken spin density was used to assign radical character. All numbers are in kcal/mol. 128 4.5 Reactions of a homogeneous oxidation catalyst utilizing the ROA mechanistic motif Potential catalysts incorporating κ−P phosphinite coordination to metal, i.e. a direct V−P bond as in Figure 4.4b, were explored. Although the inspiration for our proposed catalyst structure was heuristic in nature, our confidence in its predicted activity is to a large degree bolstered by comprehensive theoretical investigations in the system using density functional theory. In this section we will discuss our results on various potential reactivity pathways, and suggest additional theoretical pathways that ought to be investigated before experimental work commences. One potential pitfall in the installation of the bis(2-phenoxyl)phosphinite (OPO) ligand onto V would be the possibility of the P−O moiety on the ligand coordinating via the oxygen atom to the vanadium. However, at no time in our DFT geometry optimizations, aside from some high energy dianionic five-coordinate cases, did we see this sort of isomerization happen, nor did we see negative frequencies in our calculated Hessian matrices. This implies that the homogeneous vanadium complex with a V−P−O unit is a stable coordination isomer that may be realistically synthesized. 4.5.1 Nomenclature Due to the number of complexes investigated, a systematic shorthand naming convention is desirable. As all of our complexes share the common motif of a vanadium atom bound κ−P to the O−P(−O)−O pincer ligand bis(2-phenoxyl)phosphinite (OPO); we denote this commonality as (OPO)V. If there is an additional R group (such as a hydrogen atom) on the PO moiety of the (OPO), we label it as a subscript: (OPO)R V. If the V atom is five-coordinate, then there are two remaining coordination sites; these are labeled with the s or a superscripts depending on whether they are synperiplanar (syn) or antiperiplanar (anti ) to the bis(2-phenoxyl)phosphinite’s PO moiety, respectively (Figure 4.17, left). If the V atom is six-coordinate, then the (OPO) ligand may be in either a facial (fac) or meridional (mer ) configuration (Figure 4.17, center and right, respectively); and there are three remaining coordination sites. In the fac case, the three sites are labeled with the c, s, and t superscripts, depending on whether they are clinal or syn to the PO moiety of the (OPO), or trans to the P, respectively. In the mer case, the three coordination sites are labeled with the s, t, or a superscripts, depending on whether they are syn to the PO, trans to the P, or anti to the PO, respectively. Under this system, the two specific conformational species shown in the right side of V V m s ta Figure 4.6 could be unambiguously expressed as (OPO)f V Os Lct 2 and (OPO) V O L2 . 129 Figure 4.17. Template for the nomenclature of investigated complexes. For the (OPO) ligand, an f superscript denotes that the chelation is facial whereas an m superscript denotes that it is meridional. For the V, n is the oxidation state. For the ligands Ln , the superscript a denotes that the ligand is anti to the PO moiety; s denotes that it is syn to the PO moiety; c denotes that it is clinal to the PO moiety; and t denotes that it is trans to the P atom. 4.5.2 The resting VV state The first consideration that needs to be investigated is the precise nature of the resting VV state: whether it is five or six coordinated, charged or uncharged, with the (OPO) ligand in a fac or mer conformation, and with what optimal combination of oxo, hydroxo, and aqua ligands. The pH of the environments investigated was set to either 0 or 7. Changing the ambient pH does not affect the relative energetics of isomeric species with the same charge, but does change which charge cohort of species will be most stable. V Table 4.1 is a comprehensive compilation of all (OPO)V −O species studied. The most stable V species at pH 0 is the neutral (OPO)fH V Oc (OH)s (H2 O)t , in which the P−O moiety on the ligand V is already protonated. However, the isomer (OPO)f V Oc (H2 O)st 2 , which does have a P−O moiety available, is the second-lowest energy species at only 1.6 kcal/mol higher. At pH 7, the most stable V 2– species is the doubly deprotonated [(OPO)f V Oc (OH)st 2 ] . The optimized structures of these three species are shown in Figure 4.18. Note that in some fac cases, the PO moiety is twisted enough such that the clinal and syn positions on the V both become gauche. Since this phenomenon was not universal, no relabeling of clinal and syn markers was done. Species Egas Gaq V −1481.9660 (OPO)V Os (H2 O)a V V (OPO)V −O (OPO)V Oa (H2 O)s Grel (pH = 0) Grel (pH = 7) −1481.7587 20.9 29.8 −1711.4245 −1711.1621 2.8 11.7 −1711.4238 −1711.1506 10.0 18.9 130 Species Egas Gaq (OPO)V (OH)2 −1711.4121 −1711.1479 (OPO)H VV Os (OH)a geometry unstable (OPO)H VV Oa (OH)s geometry unstable V + −1711.8126 + geometry unstable [(OPO)H VV Os (H2 O)a ] [(OPO)H VV Oa (H2 O)s ] −1711.5923 Grel (pH = 0) Grel (pH = 7) 8.6 20.5 3.1 21.5 [(OPO)V Os (OH)a ] V – −1710.9247 −1710.7246 7.0 6.4 V – −1710.9181 −1710.7148 13.1 12.5 −1710.3205 −1710.2869 11.4 1.2 (OPO)f V Oc (H2 O)st 2 V −1940.8700 −1940.5384 1.6 10.5 V (OPO)f V Os (H2 O)ct 2 −1940.8658 −1940.5334 4.8 13.7 V −1940.8678 −1940.5327 5.2 14.1 −1940.8696 −1940.5410 0.0 8.9 −1940.8648 −1940.5338 4.5 13.4 −1940.8690 −1940.5363 2.9 11.8 −1940.8598 −1940.5254 9.8 18.7 −1940.8744 −1940.5375 2.2 11.1 −1940.8719 −1940.5372 2.4 11.3 [(OPO)V Oa (OH)s ] V [(OPO)V O2 ] 2– (OPO)f V Ot (H2 O)cs 2 V (OPO)fH V Oc (OH)s (H2 O)t V (OPO)fH V Oc (OH)t (H2 O)s V (OPO)fH V Os (OH)c (H2 O)t V (OPO)fH V Os (OH)t (H2 O)c V (OPO)fH V Ot (OH)c (H2 O)s V (OPO)fH V Ot (OH)s (H2 O)c V – −1940.3696 −1940.0919 11.5 10.9 V – −1940.3858 −1940.0992 6.9 6.3 V – −1940.3795 −1940.0966 8.6 8.0 [(OPO)fH V Oc (OH)st 2 ] [(OPO)fH V Os (OH)ct 2 ] [(OPO)fH V Ot (OH)cs 2 ] V – −1940.3743 −1940.0988 7.2 6.6 V – −1940.3722 −1940.0978 7.8 7.2 V – −1940.3677 −1940.0910 12.1 11.5 [(OPO)f V Oc (OH)s (H2 O)t ] [(OPO)f V Oc (OH)t (H2 O)s ] [(OPO)f V Os (OH)c (H2 O)t ] 131 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) V – −1940.3600 −1940.0874 14.3 13.7 V – −1940.3649 −1940.0887 13.5 12.9 V – −1940.3726 −1940.0916 11.8 11.1 [(OPO)f V Oc (OH)st 2 ] V 2– −1939.7811 −1939.6634 10.1 0.0 [(OPO)f V Os (OH)ct 2 ] V 2– −1939.7864 −1939.6622 10.9 0.7 V 2– −1939.7640 −1939.6517 17.5 7.3 [(OPO)f V Os (OH)t (H2 O)c ] [(OPO)f V Ot (OH)c (H2 O)s ] [(OPO)f V Ot (OH)s (H2 O)c ] [(OPO)f V Ot (OH)cs 2 ] V 2– −1939.7706 −1939.6550 15.4 5.3 V 2– −1939.7651 −1939.6465 20.7 10.6 V 2– −1939.7836 −1939.6595 12.6 2.4 V 2– −1939.7795 −1939.6556 15.0 4.9 V 2– −1939.7662 −1939.6425 23.3 13.1 [(OPO)fH V (OH)t (O)cs 2 ] V 2– −1939.7788 −1939.6546 15.7 5.5 (OPO)m V Os (H2 O)ta 2 V −1940.8672 −1940.5346 4.0 12.9 (OPO)m V Ot (H2 O)sa 2 V −1940.8513 −1940.5180 14.5 23.4 V −1940.8592 −1940.5289 7.6 16.5 V geometry unstable V geometry unstable V geometry unstable V −1940.8548 −1940.5201 13.1 22.0 V −1940.8449 −1940.5093 19.9 28.8 V −1940.8578 −1940.5270 8.8 17.7 −1940.0966 8.6 7.9 [(OPO)f V (H2 O)c (O)st 2 ] [(OPO)f V (H2 O)s (O)ct 2 ] [(OPO)f V (H2 O)t (O)cs 2 ] [(OPO)fH V (OH)c (O)st 2 ] [(OPO)fH V (OH)s (O)ct 2 ] (OPO)m V Oa (H2 O)st 2 s t a (OPO)m H V O (OH) (H2 O) s a t (OPO)m H V O (OH) (H2 O) t s a (OPO)m H V O (OH) (H2 O) t a s (OPO)m H V O (OH) (H2 O) a s t (OPO)m H V O (OH) (H2 O) a t s (OPO)m H V O (OH) (H2 O) V – −1940.3660 V – geometry unstable [(OPO)m V Os (OH)t (H2 O)a ] [(OPO)m V Os (OH)a (H2 O)t ] 132 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) 16.2 15.6 V – geometry unstable V – −1940.3614 V – geometry unstable V – −1940.3724 −1940.0932 10.7 10.1 s ta [(OPO)m H V O (OH)2 ] V – −1940.3355 −1940.0626 29.9 29.3 t sa [(OPO)m H V O (OH)2 ] V – −1940.3573 −1940.0785 20.0 19.3 V – −1940.3447 −1940.0654 28.1 27.5 V 2– −1939.7563 −1939.6389 25.5 15.3 V 2– −1939.7430 −1939.6271 32.9 22.7 V 2– −1939.7472 −1939.6324 29.6 19.4 [(OPO)m V Ot (OH)s (H2 O)a ] [(OPO)m V Ot (OH)a (H2 O)s ] [(OPO)m V Oa (OH)s (H2 O)t ] [(OPO)m V Oa (OH)t (H2 O)s ] a st [(OPO)m H V O (OH)2 ] [(OPO)m V Os (OH)ta 2 ] [(OPO)m V Ot (OH)sa 2 ] [(OPO)m V Oa (OH)st 2 ] −1940.0845 V 2– geometry unstable V 2– geometry unstable V 2– geometry unstable s ta [(OPO)m H V (OH) (O)2 ] V 2– −1939.7499 −1939.6332 29.1 18.9 t sa [(OPO)m H V (OH) (O)2 ] V 2– −1939.7799 −1939.6580 13.5 3.4 V 2– −1939.7503 −1939.6286 32.0 21.8 [(OPO)m V (H2 O)s (O)ta 2 ] [(OPO)m V (H2 O)t (O)sa 2 ] [(OPO)m V (H2 O)a (O)st 2 ] a st [(OPO)m H V (OH) (O)2 ] Table 4.1. Comprehensive compilation of all (OPO)VV species studied, along with their gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous free energies relative to each other, in kcal/mol, at pH 0 and 7, respectively. For some entries, geometry optimization led to the formation of a lower-energy isomer. The energies for these entries are not reported. 4.5.3 Propane activation transition states Propane consists of two types of C−H bonds: primary and secondary, and the secondary C−H bond in the middle carbon is more easily activated due to the increased stability of the resulting secondary V radical. Those (OPO)V species listed in Table 4.1 which contain unadorned intact P−O moieties were further investigated for their ability to activate the secondary C−H bond of propane. The transition states were located and their free energies at pH 0 and 7 relative to the ground (OPO)V V V 2– V states of (OPO)fH V Oc (OH)s (H2 O)t and [(OPO)f V Oc (OH)st 2 ] , respectively, were calculated. The results are shown in Table 4.2. 133 V V Figure 4.18. The geometry-optimized structures of (OPO)fH V Oc (OH)s (H2 O)t and (OPO)f V Oc (H2 O)st 2 , V 2– the most stable (OPO)VV isomers at pH 0; and of the doubly deprotonated [(OPO)f V Oc (OH)st 2 ] , the most stable isomer at pH 7. Species Egas Gaq Grel (pH = 0) Grel (pH = 7) [(OPO)HiPr VV Os (H2 O)a ]‡ −1830.5749 −1830.2107 39.6 48.5 [(OPO)HiPr VV Oa (H2 O)s ]‡ −1830.5628 −1830.1985 47.2 56.1 [(OPO)HiPr VV Os (OH)a ]−‡ geometry unstable [(OPO)HiPr VV Oa (OH)s ]−‡ geometry unstable 2−‡ [(OPO)HiPr VV Osa 2 ] geometry unstable s ta ‡ [(OPO)m HiPr V O (H2 O)2 ] V −2060.0127 −2059.5889 37.2 46.1 t sa ‡ [(OPO)m HiPr V O (H2 O)2 ] V −2060.0129 −2059.5909 36.0 44.9 a st ‡ [(OPO)m HiPr V O (H2 O)2 ] V −2060.0036 −2059.5752 45.9 54.8 V −2059.5055 −2059.1441 46.1 45.5 V −2059.4829 −2059.1232 59.2 58.6 s t a −‡ [(OPO)m HiPr V O (OH) (H2 O) ] s a t −‡ [(OPO)m HiPr V O (OH) (H2 O) ] 134 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) V −2059.4979 −2059.1286 55.8 55.2 V −2059.5033 −2059.1384 49.7 49.1 V −2059.4868 −2059.1247 58.3 57.7 V −2059.4888 −2059.1282 56.1 55.4 s ta 2−‡ [(OPO)m HiPr V O (OH)2 ] V −2058.9071 −2058.6901 60.7 50.6 t sa 2−‡ [(OPO)m HiPr V O (OH)2 ] V −2058.8825 −2058.6799 67.1 57.0 a st 2−‡ [(OPO)m HiPr V O (OH)2 ] V −2058.8811 −2058.6783 68.1 57.9 V −2058.8702 −2058.6684 74.4 64.2 t sa 2−‡ [(OPO)m HiPr V (H2 O) O2 ] V geometry unstable V geometry unstable t s a −‡ [(OPO)m HiPr V O (OH) (H2 O) ] t a s −‡ [(OPO)m HiPr V O (OH) (H2 O) ] a s t −‡ [(OPO)m HiPr V O (OH) (H2 O) ] a t s −‡ [(OPO)m HiPr V O (OH) (H2 O) ] s ta 2−‡ [(OPO)m HiPr V (H2 O) O2 ] a st 2−‡ [(OPO)m HiPr V (H2 O) O2 ] V ‡ [(OPO)fHiPr V Oc (H2 O)st 2 ] V ‡ [(OPO)fHiPr V Os (H2 O)ct 2 ] V ‡ [(OPO)fHiPr V Ot (H2 O)cs 2 ] V [(OPO)fHiPr V Oc (OH)s (H2 O)t ]−‡ V [(OPO)fHiPr V Oc (OH)t (H2 O)s ]−‡ V [(OPO)fHiPr V Os (OH)c (H2 O)t ]−‡ V [(OPO)fHiPr V Os (OH)t (H2 O)c ]−‡ V [(OPO)fHiPr V Ot (OH)c (H2 O)s ]−‡ V [(OPO)fHiPr V Ot (OH)s (H2 O)c ]−‡ V 2−‡ [(OPO)fHiPr V Oc (OH)st 2 ] V 2−‡ [(OPO)fHiPr V Os (OH)ct 2 ] V 2−‡ [(OPO)fHiPr V Ot (OH)cs 2 ] V 2−‡ [(OPO)fHiPr V (H2 O)c Ost 2 ] −2060.0212 −2059.6000 30.3 39.2 −2060.0208 −2059.5954 33.2 42.1 −2060.0229 −2059.5991 30.8 39.7 −2059.5101 −2059.1478 43.8 43.1 −2059.5120 −2059.1514 41.5 40.9 −2059.5096 −2059.1432 46.7 46.1 −2059.5118 −2059.1455 45.2 44.6 −2059.5186 −2059.1506 42.0 41.4 −2059.5170 −2059.1512 41.6 41.0 −2058.9002 −2058.6969 56.4 46.3 −2058.9152 −2058.7010 53.9 43.7 −2058.9121 −2058.6971 56.3 46.2 −2058.8851 −2058.6791 67.6 57.4 135 Species V 2−‡ [(OPO)fHiPr V (H2 O)s Oct 2 ] V 2−‡ [(OPO)fHiPr V (H2 O)t Ocs 2 ] Egas Gaq Grel (pH = 0) Grel (pH = 7) −2058.8771 −2058.6841 64.5 54.4 −2058.8941 −2058.6763 69.4 59.3 Table 4.2. Comprehensive compilation of all the transition states for propane activation by (OPO)VV , along with their gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous free energies, in kcal/mol, relative to the ground states at pH 0 and 7 V V 2– ((OPO)fH V Oc (OH)s (H2 O)t and [(OPO)f V Oc (OH)st 2 ] , respectively). For some entries, geometry optimization led to the formation of a lower-energy isomer. The energies for these entries are not reported. According to Table 4.2, the most favorable transition state at both pH 0 and 7 is the neutral V ‡ species [(OPO)fHiPr V Oc (H2 O)st 2 ] , whose structure is shown in detail in Figure 4.19. At pH 0, V this species is 30.3 kcal/mol above the neutral ground state (OPO)fH V Oc (OH)s (H2 O)t , and with V ∆G‡ = 28.7 kcal/mol from its immediate precursor (OPO)f V Oc (H2 O)st 2 ; we therefore consider it to be accessible for C−H activation. At pH 7, however, the species is 39.2 kcal/mol above the V 2– ground state [(OPO)f V Oc (OH)st 2 ] , as this ground state must be the doubly protonated before C−H activation can occur. Doubly protonating the ground state before C−H activation is still preferable to the dianion directly activating propane; this is due to the fact that the negative charge stabilizes the VV −O moiety and thus decreases its tendency to be reduced to VIV . We conclude from these results that catalysis with the (OPO)V system is much more likely at pH 0. Analysis of the Mulliken spin density supports our assignment of radical character: Whereas the overall spin is 0, the spin density on the vanadium (1.0882), propane’s secondary carbon (-0.5389), and the attacking PO oxygen (-0.2084), but not the phosphorus itself (-0.0350), supporting our assertion of attack by the ROA mechanism. V ‡ Figure 4.19. The geometry-optimized structure of [(OPO)fHiPr V Oc (H2 O)st 2 ] , the most stable transition state for C−H activation of propane at either pH 0 or 7. 136 Monohydrogenated VIV states 4.5.4 The result of the C−H activation transition states as described in the previous subsection is an isopropyl radical and a monohydrogenated (OPO)H VIV complex, which may then isomerize into other IV (OPO)V species. The various (OPO)V IV complexes are listed in Table 4.3. Only six-coordinate complexes with fac-(OPO) were exhaustively investigated, as Table 4.1 shows that they are lower in energy than their five-coordinate and six-coordinate mer counterparts. At pH 0, the lowest energy IV species is (OPO)fH V Oc (H2 O)st 2 at 13.4 kcal/mol above the ground state. However, isomerization IV to (OPO)f V (OH)c (H2 O)st 2 , at 13.6 kcal/mol above the ground state, is nearly thermoneutral. At IV – pH 7, the lowest energy species is [(OPO)f V Os (H2 O)ct 2 ] at 16.8 kcal/mol above the ground state. These species are shown in detail in Figure 4.20. All species are doublets, and Mulliken spin density analysis reveals that the spins are localized on the vanadium atoms, hence supporting the assignment of V in the +IV oxidation state. IV IV f c st Figure 4.20. The geometry-optimized structures of (OPO)fH V Oc (H2 O)st 2 and (OPO) V (OH) (H2 O)2 , the most stable (OPO)V stable isomer at pH 7. 4.5.5 IV f IV s – isomers at pH 0; and of the deprotonated [(OPO) V O (H2 O)ct 2 ] , the most Catalytic cycle involving H-atom abstraction only The simplest catalytic cycle that can now be envisioned is one that alternates only between the V resting state and its corresponding monohydrogenated species, e.g., (OPO)fH V Oc (OH)s (H2 O)t and IV (OPO)fH V Oc (H2 O)st 2 (We will only consider the pH 0 case since it is kinetically feasible with overall ∆G‡ = 30.3 kcal/mol; see Subsection 4.5.3). After monohydrogenation, the starting VV species is regenerated by sequential removal of H+ and e– (Scheme 4.2). First deprotonating the 137 Species (OPO)H VIV (O)s (H2 O)a (OPO)H VIV (O)a (H2 O)s IV – [(OPO)V (O)s (H2 O)a ] IV – [(OPO)V (O)a (H2 O)s ] IV s ta (OPO)m H V (O) (H2 O)2 IV t sa (OPO)m H V (O) (H2 O)2 m IV a (OPO)H V (O) (H2 O)st 2 – IV [(OPO)m V (O)s (H2 O)ta 2 ] IV – [(OPO)m V (O)t (H2 O)sa 2 ] – IV [(OPO)m V (O)a (H2 O)st 2 ] Egas −1712.0701 −1712.0474 −1711.5717 −1711.5440 −1941.5038 −1941.4939 −1941.4894 −1940.9974 −1940.9955 −1941.0011 Gaq −1711.7906 −1711.7653 −1711.3486 −1711.3323 −1941.1599 −1941.1627 −1941.1451 −1940.7170 −1940.7103 −1940.7064 Grel (pH = 0) 16.2 32.1 23.3 33.5 19.5 17.7 28.8 27.1 31.3 33.8 Grel (pH = 7) 25.1 41.0 22.7 32.9 28.4 26.6 37.7 26.5 30.7 33.1 (OPO)f V (OH)c (H2 O)st 2 IV −1941.5108 −1941.1692 13.6 22.5 (OPO)f V (OH)s (H2 O)ct 2 IV −1941.5065 −1941.1636 17.2 26.1 −1941.4999 −1941.1580 20.7 29.6 −1941.4953 −1941.1503 25.5 34.4 −1941.4931 −1941.1497 25.9 34.7 −1941.5068 −1941.1598 19.5 28.4 −1941.5092 −1941.1695 13.4 22.3 −1941.5180 −1941.1663 15.4 24.3 −1941.5093 −1941.1667 15.2 24.1 −1941.0201 −1940.7264 21.2 20.6 −1941.0201 −1940.7325 17.4 16.8 −1941.0125 −1940.7266 21.1 20.5 −1940.9999 −1940.7101 31.4 30.8 −1941.0011 −1940.7205 24.9 24.3 −1941.0136 −1940.7218 24.1 23.5 −1941.0143 −1940.7228 23.5 22.9 −1941.0161 −1940.7247 22.3 21.6 −1941.0095 −1940.7226 23.6 23.0 −1940.4101 −1940.2769 33.0 22.8 −1940.4110 −1940.2804 30.8 20.6 −1940.4079 −1940.2745 34.5 24.4 −1940.4018 −1940.2721 36.0 25.8 −1940.4013 −1940.2790 31.7 21.5 −1940.3952 −1940.2736 35.1 24.9 −1940.3899 −1940.2599 43.6 33.5 −1940.4093 −1940.2757 33.8 23.6 −1940.4031 −1940.2828 29.3 19.2 IV f t (OPO) V (OH) (H2 O)cs 2 IV (OPO)fH V (H2 O)c (OH)st 2 IV (OPO)fH V (H2 O)s (OH)ct 2 IV (OPO)fH V (H2 O)t (OH)cs 2 IV (OPO)fH V (O)c (H2 O)st 2 IV (OPO)fH V (O)s (H2 O)ct 2 IV (OPO)fH V (O)t (H2 O)cs 2 IV – [(OPO)f V (O)c (H2 O)st 2 ] IV – [(OPO)f V (O)s (H2 O)ct 2 ] IV – [(OPO)f V (O)t (H2 O)cs 2 ] – f IV c s [(OPO)H V (O) (OH) (H2 O)t ] IV – [(OPO)fH V (O)c (OH)t (H2 O)s ] IV – [(OPO)fH V (O)s (OH)c (H2 O)t ] IV – [(OPO)fH V (O)s (OH)t (H2 O)c ] IV – [(OPO)fH V (O)t (OH)c (H2 O)s ] IV – [(OPO)fH V (O)t (OH)s (H2 O)c ] IV 2– [(OPO)f V (O)c (OH)s (H2 O)t ] IV 2– [(OPO)f V (O)c (OH)t (H2 O)s ] IV 2– [(OPO)f V (O)s (OH)c (H2 O)t ] IV 2– [(OPO)f V (O)s (OH)t (H2 O)c ] IV 2– [(OPO)f V (O)t (OH)c (H2 O)s ] IV 2– [(OPO)f V (O)t (OH)s (H2 O)c ] IV 2– [(OPO)fH V (O)c (OH)st 2 ] IV 2– [(OPO)fH V (O)s (OH)st 2 ] IV 2– [(OPO)fH V (O)t (OH)st 2 ] Table 4.3. Comprehensive compilation of all (OPO)VIV species studied, along with their gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous V free energies, in kcal/mol, relative to the ground states at pH 0 and 7 ((OPO)fH V Oc (OH)s (H2 O)t and V 2– [(OPO)f V Oc (OH)st 2 ] , respectively). 138 IV IV – monohydrogenated (OPO)fH V Oc (H2 O)st 2 forms the anionic intermediates [(OPO)V O(H2 O)2 ] , IV – the lowest isomer of which ([(OPO)f V (O)s (H2 O)ct 2 ] ) is 17.4 kcal/mol above the starting complex. The deprotonation energy gain of 4.0 kcal/mol leads to a calculated pKa value of 2.9 for the neutral IV monohydrogenated complex (OPO)fH V Oc (H2 O)st 2 . However, the reduction potential of the neutral V V f c st (OPO)f V Os (H2 O)ct 2 is calculated to be 1.07 V versus SHE, whereas that of (OPO) V O (H2 O)2 is IV calculated to be 0.77 V versus SHE. Hence, it is difficult to reoxidize the deprotonated [(OPO)f V Oc – V (H2 O)st 2 ] back to V , but not impossible; since the standard reduction potential of O2 is 1.23 V vs. SHE. Scheme 4.2. Potential catalytic cycle involving H-atom abstraction only: The one-electron oxidation and catalyst regeneration that forms the core of this catalytic cycle. All energies are in kcal/mol. Note that the overall reaction in this cycle is C3 H8 + 21 O2 −−→ C3 H6 + H2 O, with ∆G = −32.3 kcal/mol. Hence, whereas V V f s ct the (OPO)f V Oc (H2 O)st 2 and (OPO) V O (H2 O)2 complexes are initially 1.6 and 4.8 kcal/mol above the starting state, after one complete cycle they are −30.7 and −27.4 kcal/mol below the starting state, respectively. Since this would only be a one-electron oxidative process, two complete cycles are necessary for propane oxidation: first to abstract a hydrogen atom from the starting alkane, and then to abstract a second hydrogen atom from the resulting isopropyl radical. As hydrogen abstraction from an alkyl radical to a closed-shell species is expected to be facile, the rate-limiting step in this case will be the V initial propane activation step. From the (OPO)f V Oc (H2 O)st 2 precursor (1.6 kcal/mol above the ground state), initial H-atom abstraction from propane is uphill by 11.8 kcal/mol. As the weakest C−H bond in propane (i.e. the iPr−H bond) has a dissociation energy of 103 kcal, this implies a very strong hydrogen abstracting power of 91.2 kcal/mol. This number is very close to the hydrogenabstraction power of the P−O moiety in VV OPO4 , the oxidized surface of the VPO catalyst [3], hence confirming the validity of our ROA mechanism as modified in Figure 4.4b. The overall kinetic 139 barrier for H-atom abstraction is 30.3 kcal/mol, a value that is low enough to be accessible at elevated reaction temperatures. As indicated by Scheme 4.2 and its associated discussion, deprotonation of the monohydrogenated IV V (OPO)fH V Oc (H2 O)st 2 precedes oxidation back to a V species. In fact, these steps are both slightly uphill, by 4.0 and 1.6 kcal/mol, respectively, but they are low enough to remain feasible. The second H-atom abstraction, from the isopropyl radical, is quite facile as expected, creating propylene as the IV oxidation product. The resulting (OPO)fH V Oc (H2 O)st 2 then goes through one more deprotonation, oxidation, and isomerization cycle to regenerate the starting catalyst. Overall, this reaction is quite feasible thermodynamically, and the reaction barriers are not expected to be large, either. Hence we can expect this to be a complete potential catalytic cycle for the dehydrogenation of propane to propylene. 4.5.6 Reduced VIII states Since reactive isopropyl radicals are generated upon H-atom abstraction, it is necessary to also consider the possibility that they may be trapped by the VIV intermediate instead of being converted to propylene by another equivalent of a VV species. Once again taking inspiration from Cheng and Goddard [3], in which the produced isopropyl radical is adsorbed onto a V−O moiety, we analogously IV considered the possibility that the isopropyl radical and monohydrogenated (OPO)V (O)(H2 O)2 III species react with each other to form (OPO)V (OiPr) species. The isopropoxy ligand may then be protonated and exchanged with an aqua ligand, resulting in isopropanol being produced. Table 4.4 is a comprehensive list of (OPO)V III species with only oxo/hydroxo/aqua ligands, and neutral or −1 charge. As in the case of the VV and VIV species, the (OPO)V III species are most stable with the (OPO) ligand in the fac configuration. Furthermore, the triplet states are universally more stable than their singlet counterparts, supporting the +III assignment for the vanadium atom’s III oxidation state. Hence, Table 4.5, which lists the (OPO)V (OiPr) species, only contains the triplet six coordinate species with the (OPO) ligand in the fac configuration. Species Egas Gaq Grel (pH = 0) Grel (pH = 7) 1 (OPO)V (H2 O)sa 2 III −1712.6358 −1712.3473 25.0 33.9 3 (OPO)V (H2 O)sa 2 III −1712.6811 −1712.3911 −2.5 6.4 1 (OPO)H VIII (OH)s (H2 O)a geometry unstable 3 (OPO)H VIII (OH)s (H2 O)a −1712.6895 −7.9 1.0 −1712.3998 140 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) 1 (OPO)H VIII (OH)a (H2 O)s −1712.6609 −1712.3705 10.5 19.4 3 (OPO)H VIII (OH)a (H2 O)s −1712.6784 −1712.3916 −2.7 6.2 1 [(OPO)V (OH)s (H2 O)a ] 3 [(OPO)V (OH)s (H2 O)a ] 7.7 7.1 1 [(OPO)V (OH)a (H2 O)s ] 3 [(OPO)V (OH)a (H2 O)s ] 10.9 10.3 1 [(OPO)H VIII (OH)sa 2 ] – geometry unstable 3 [(OPO)H VIII (OH)sa 2 ] – geometry unstable 1 (OPO)m V (H2 O)sta 3 III −1942.0707 −1941.7179 27.5 36.4 3 (OPO)m V (H2 O)sta 3 III −1942.1201 −1941.7641 −1.5 7.4 1 s ta (OPO)m H V (OH) (H2 O)2 III −1942.0625 −1941.7111 31.8 40.7 3 s ta (OPO)m H V (OH) (H2 O)2 III −1942.1196 −1941.7628 −0.7 8.2 1 t sa (OPO)m H V (OH) (H2 O)2 III −1942.0710 −1941.7216 25.2 34.1 3 t sa (OPO)m H V (OH) (H2 O)2 III −1942.1162 −1941.7627 −0.6 8.3 1 a st (OPO)m H V (OH) (H2 O)2 III −1942.0597 −1941.6995 39.1 48.0 3 a st (OPO)m H V (OH) (H2 O)2 III −1942.1097 −1941.7567 3.2 12.0 1 [(OPO)m V (OH)s (H2 O)ta 2 ] III – −1941.5661 −1941.2673 40.0 39.3 3 [(OPO)m V (OH)s (H2 O)ta 2 ] III – −1941.6145 −1941.3211 6.2 5.6 1 [(OPO)m V (OH)t (H2 O)sa 2 ] III – −1941.5611 −1941.2660 40.8 40.2 3 [(OPO)m V (OH)t (H2 O)sa 2 ] III – −1941.6143 −1941.3217 5.8 5.2 1 [(OPO)m V (OH)a (H2 O)st 2 ] III – −1941.5539 −1941.2535 48.6 48.0 3 [(OPO)m V (OH)a (H2 O)st 2 ] III – −1941.6047 −1941.3114 12.3 11.7 1 s ta [(OPO)m H V (H2 O) (OH)2 ] III – −1941.5762 −1941.2783 33.1 32.5 III – geometry unstable III – −1712.1787 III – geometry unstable III – −1712.1727 −1711.9442 −1711.9391 141 Species Egas Gaq Grel (pH = 0) Grel (pH = 7) 3 s ta [(OPO)m H V (H2 O) (OH)2 ] III – −1941.5886 −1941.3002 19.3 18.7 1 t sa [(OPO)m H V (H2 O) (OH)2 ] III – −1941.5562 −1941.2561 47.0 46.4 3 t sa [(OPO)m H V (H2 O) (OH)2 ] III – −1941.6073 −1941.3093 13.6 13.0 1 a st [(OPO)m H V (H2 O) (OH)2 ] III – −1941.5988 −1941.3017 18.4 17.7 3 a st [(OPO)m H V (H2 O) (OH)2 ] III – −1941.6188 −1941.3235 4.7 4.1 1 (OPO)f V (H2 O)cst 3 III −1942.0823 −1941.7202 26.0 34.9 3 (OPO)f V (H2 O)cst 3 III −1942.1432 −1941.7900 −17.7 −8.8 −1942.0873 −1941.7262 22.3 31.2 −1942.1429 −1941.7811 −12.2 −3.3 −1942.0885 −1941.7307 19.5 28.4 −1942.1386 −1941.7818 −12.6 −3.7 −1942.0732 −1941.7139 30.0 38.9 −1942.1288 −1941.7765 −9.3 −0.4 III 1 (OPO)fH V (OH)c (H2 O)st 2 3 (OPO)fH V (OH)c (H2 O)st 2 III III 1 (OPO)fH V (OH)s (H2 O)ct 2 3 (OPO)fH V (OH)s (H2 O)ct 2 1 (OPO)fH V (OH)t (H2 O)cs 2 III III III 3 (OPO)fH V (OH)t (H2 O)cs 2 1 [(OPO)f V (OH)c (H2 O)st 2 ] 3 [(OPO)f V (OH)c (H2 O)st 2 ] 1 [(OPO)f V (OH)s (H2 O)ct 2 ] 3 [(OPO)f V (OH)s (H2 O)ct 2 ] III – −1941.5871 −1941.2738 35.9 35.3 III – −1941.6398 −1941.3377 −4.2 −4.8 III – −1941.5861 −1941.2299 63.4 62.8 III – −1941.6343 −1941.3371 −3.8 −4.5 III – −1941.5637 −1941.2725 36.7 36.1 III – −1941.6283 −1941.3282 1.8 1.2 III – −1941.6034 −1941.3092 13.7 13.1 III – −1941.6221 −1941.3289 1.3 0.7 III – −1941.5798 −1941.2804 31.8 31.1 III – −1941.6323 −1941.3331 −1.3 −2.0 1 [(OPO)f V (OH)t (H2 O)cs 2 ] 3 [(OPO)f V (OH)t (H2 O)cs 2 ] 1 [(OPO)fH V (H2 O)c (OH)st 2 ] 3 [(OPO)fH V (H2 O)c (OH)st 2 ] 1 [(OPO)fH V (H2 O)s (OH)ct 2 ] 3 [(OPO)fH V (H2 O)s (OH)ct 2 ] 142 Species 1 3 Egas Gaq Grel (pH = 0) Grel (pH = 7) III – −1941.6015 −1941.3001 19.4 18.8 III – −1941.6254 −1941.3234 4.8 4.1 [(OPO)fH V (H2 O)t (OH)cs 2 ] [(OPO)fH V (H2 O)t (OH)cs 2 ] Table 4.4. Comprehensive compilation of all (OPO)VIII species containing only oxo/hydroxo/aqua ligands studied, along with their gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous free energies, in kcal/mol, relative to the ground states at pH 0 V V 2– and 7 ((OPO)fH V Oc (OH)s (H2 O)t and [(OPO)f V Oc (OH)st 2 ] , respectively). For some entries, geometry optimization led to the formation of a lower-energy isomer. The energies for these entries are not reported. Species Egas Gaq III −1907.1454 III III 3 (OPO)f V (HOiPr)c (H2 O)st 2 3 (OPO)f V (HOiPr)s (H2 O)ct 2 3 (OPO)f V (HOiPr)t (H2 O)cs 2 III 3 (OPO)fH V (HOiPr)c (OH)s (H2 O)t 3 (OPO)fH V (HOiPr)c (OH)t (H2 O)s III III 3 (OPO)fH V (HOiPr)s (OH)c (H2 O)t 3 (OPO)fH V (HOiPr)s (OH)t (H2 O)c 3 (OPO)fH V (HOiPr)t (OH)c (H2 O)s III III III 3 (OPO)fH V (HOiPr)t (OH)s (H2 O)c 3 (OPO)fH V (OiPr)c (H2 O)st 2 3 (OPO)fH V (OiPr)s (H2 O)ct 2 III III III 3 (OPO)fH V (OiPr)t (H2 O)cs 2 3 [(OPO)f V (HOiPr)c (OH)s (H2 O)t ] Grel (pH = 0) Grel (pH = 7) −1906.7524 −12.8 −3.9 −1907.1518 −1906.7610 −18.2 −9.3 −1907.1489 −1906.7546 −14.2 −5.3 −1907.1417 −1906.7463 −9.0 −0.1 −1907.1398 −1906.7456 −8.6 0.3 −1907.1520 −1906.7574 −16.0 −7.1 −1907.1339 −1906.7422 −6.5 2.4 −1907.1472 −1906.7494 −10.9 −2.0 −1907.1439 −1906.7503 −11.5 −2.6 −1907.1448 −1906.7528 −13.1 −4.2 −1907.1544 −1906.7567 −15.5 −6.6 −1907.1454 −1906.7458 −8.7 0.2 III – −1906.6425 −1906.3086 −4.6 −5.2 III – −1906.6463 −1906.3113 −6.3 −6.9 III – −1906.6475 −1906.3125 −7.1 −7.7 III – −1906.6369 −1906.3039 −1.7 −2.3 III – −1906.6436 −1906.3069 −3.6 −4.2 3 [(OPO)f V (HOiPr)c (OH)t (H2 O)s ] 3 [(OPO)f V (HOiPr)s (OH)c (H2 O)t ] 3 [(OPO)f V (HOiPr)s (OH)t (H2 O)c ] 3 [(OPO)f V (HOiPr)t (OH)c (H2 O)s ] 143 Species 3 III – [(OPO)f V (HOiPr)t (OH)s (H2 O)c ] Gaq Grel (pH = 0) Grel (pH = 7) −1906.6461 −1906.3043 −1.9 −2.6 III – −1906.6476 −1906.3131 −7.5 −8.1 III – −1906.6425 −1906.3043 −1.9 −2.5 III – −1906.6420 −1906.2957 3.4 2.8 3 [(OPO)f V (OiPr)c (H2 O)st 2 ] 3 [(OPO)f V (OiPr)s (H2 O)ct 2 ] 3 [(OPO)f V (OiPr)t (H2 O)cs 2 ] III – geometry unstable III – −1906.6284 −1906.2897 7.2 6.6 III – −1906.6261 −1906.2951 3.8 3.2 3 [(OPO)fH V (HOiPr)c (OH)st 2 ] 3 [(OPO)fH V (HOiPr)s (OH)ct 2 ] 3 [(OPO)fH V (HOiPr)t (OH)cs 2 ] III – −1906.6346 −1906.2971 2.6 1.9 III – −1906.6346 −1906.3005 0.5 −0.2 III – −1906.6482 −1906.3055 −2.7 −3.3 III – −1906.6398 −1906.2991 1.4 0.7 III – geometry unstable III – geometry unstable 3 [(OPO)fH V (OiPr)c (OH)s (H2 O)t ] 3 [(OPO)fH V (OiPr)s (OH)c (H2 O)t ] 3 [(OPO)fH V (OiPr)s (OH)c (H2 O)t ] 3 [(OPO)fH V (OiPr)s (OH)t (H2 O)c ] 3 [(OPO)fH V (OiPr)t (OH)c (H2 O)s ] 3 Egas [(OPO)fH V (OiPr)t (OH)s (H2 O)c ] Table 4.5. Compilation of all (OPO)VIII (OiPr) species studied, with only six-coordinate fac-(OPO) triplet complexes considered. The second and third columns show the gas-phase energies (Egas ) and aqueous free energies (Gaq ), both in hartrees. The last two columns show their aqueous free energies, in kcal/mol, relative V V 2– to the ground states at pH 0 and 7 ((OPO)fH V Oc (OH)s (H2 O)t and [(OPO)f V Oc (OH)st 2 ] , respectively). For some entries, geometry optimization led to the formation of a lower-energy isomer. The energies for these entries are not reported. III III At pH 0, the lowest energy (OPO)V (OiPr) species is 3 (OPO)f V (HOiPr)s (H2 O)ct 2 at −18.2 V III kcal/mol relative to the starting (OPO)fH V Oc (OH)s (H2 O)t , whereas the lowest energy (OPO)V III f species without any isopropoxy groups is 3 (OPO)f V (H2 O)cst 3 at −17.7 kcal/mol relative to (OPO)H III VV Oc (OH)s (H2 O)t . Furthermore, the lowest energy five-coordinate (OPO)V species without any V isopropoxy groups is 3 (OPO)H V (OH)s (H2 O)a at −7.9 kcal/mol relative to (OPO)fH V Oc (OH)s III III III 3 f cst (H2 O)t , indicating that it is possible for 3 (OPO)f V (HOiPr)s (H2 O)ct 2 and (OPO) V (H2 O)3 to interconvert via a labile isopropanol ligand. These three species are shown in detail in Figure 4.21. Note that at pH 7 the same three species are still the most stable, but with energies of −9.3, −8.8, V 2– and 1.0 kcal/mol with respect to [(OPO)f V Oc (OH)st 2 ] , respectively. 4.5.7 Catalytic cycle involving H-atom abstraction followed by isopropyl addition By considering the lowest energy configurations of the various VV , VIV , and VIII intermediates and transition states, we can now construct a catalytic cycle that features both H-atom abstraction 144 III III 3 f cst Figure 4.21. The geometry-optimized structures of 3 (OPO)f V (HOiPr)s (H2 O)ct 2 , (OPO) V (H2 O)3 , III 3 III s a and (OPO)H V (OH) (H2 O) . These are the most stable (OPO)V (OiPr) species, and six and five coordinate (OPO)VIII species, respectively, at both pH 0 and 7. from propane and isopropyl radical trapping before reoxidation. This cycle is shown in Scheme 4.3. V V The first part, involving isomerization of the starting (OPO)fH V Oc (OH)s (H2 O)t to (OPO)f V Oc IV f c st (H2 O)st 2 , followed by C−H activation to (OPO)H V O (H2 O)2 , is identical to that of Scheme 4.2. However, instead of being immediately reoxidized as in Scheme 4.2, the VIV species traps the generated isopropyl radical to form a VIII (OiPr) species. One example of such a pathway is shown in IV Scheme 4.3, in which the (OPO)fH V Oc (H2 O)st 2 species first isomerizes to the slightly higher (at IV 15.4 kcal/mol relative to the starting VV complex) (OPO)fH V Os (H2 O)ct 2 , which then combines III with the isopropyl radical to form 3 (OPO)fH V (OiPr)s (H2 O)ct 2 at −15.5 kcal/mol, which readily III III isomerizes to 3 (OPO)f V (HOiPr)s (H2 O)ct 2 , the lowest energy V (OiPr) species at −18.2 kcal/mol V relative to the starting (OPO)fH V Oc (OH)s (H2 O)t . Loss of the HOiPr product to form the five coIII ordinate species 3 (OPO)V (H2 O)sa 2 (at −1.5 kcal/mol) is uphill but not insurmountably so, and III recoordination of an aqua ligand to the sixth coordination site to form 3 (OPO)f V (H2 O)cst 3 recovers (at −17.7 kcal/mol) most of this energy. As the overall reaction C3 H8 + 1 /2 O2 −−→ HOiPr has III V f c st ∆G = −31.2 kcal/mol, reoxidation of 3 (OPO)V (H2 O)sa 2 to (OPO) V O (H2 O)2 (assuming that O2 is the terminal oxidation) is downhill by 11.9 kcal/mol. 4.5.8 Comparison of the catalytic cycles The H-atom abstraction-only pathway results in the dehydrogenation of propane to form propylene. However, as this is significantly uphill (C3 H8 −−→ C3 H6 + H2 , ∆G = 20.3 kcal/mol), it is unlikely unless some oxidant is coupled to the reaction. Coupling dehydrogenation with dioxygen greatly 145 Scheme 4.3. A potential catalytic cycle that assumes that the isopropyl radical adds to the vanadium oxo, reducing it to VIII . This assumption derives from the behavior of the heterogeneous VPO catalyst. All energies are in kcal/mol. Note that the overall reaction in this cycle is C3 H8 + 1 /2 O2 −−→ HOiPr, with V ∆G = −31.2 kcal/mol. Hence, whereas the (OPO)f V Oc (H2 O)st 2 complex is initially 1.6 kcal/mol above the starting state, after one complete cycle it is −29.6 kcal/mol below the starting state. increases its favorability, for a ∆G of −32.3 kcal/mol. We have not found any large barriers in the Hatom abstraction-only pathway that might preclude the production of propylene and water as in the overall equation C3 H8 + 21 O2 −−→ C3 H6 + H2 O; however, we note that our results do not include the IV IV – f s ct details of how the one-electron oxidation of [(OPO)f V (O)c (H2 O)st 2 ] or [(OPO) V (O) (H2 O)2 ] – takes place and how the released electron is transferred to dioxygen to ultimately form water. Such a mechanism may involve an inner sphere coordination of dioxygen to form a VV superoxo intermediate and may involve further C−H activation by this superoxide. Given the reduction potential of IV – IV – f s ct dioxygen versus that of [(OPO)f V (O)c (H2 O)st 2 ] or [(OPO) V (O) (H2 O)2 ] (1.23 V versus 0.77 V or 1.07 V), we believe that the H-atom abstraction-only pathway will be viable and productive in the absence of lower-energy alternative pathways. Comparing the H-atom abstraction-only versus the H-atom abstraction and isopropyl addition IV pathways, we note that the branching point is the monohydrogenated complexes (OPO)fH V (O)(H2 O)2 , which may either undergo reoxidation as in the abstraction-only mechanism (Scheme 4.2), or add the isopropyl radical (Scheme 4.3). The former overall is thermodynamically downhill by 0.8 kcal/mol in the lowest isomers; whereas the latter is thermodynamically downhill by 30.9 kcal/mol in one simple elementary step with an intuitively minimal barrier. Hence, the catalytic cycle involving isopropyl addition appears to be more likely to occur than the cycle that only features H-atom 146 abstraction. As in the H-atom abstraction-only case, we have not considered the precise nature of the reoxidation step. However, we anticipate that it may also involve inner sphere coordination of dioxygen, in which case a VV η 2 -peroxo intermediate is formed. This would be analogous to the heterogeneous VPO system, and we would also need to anticipate C−H activation by these peroxide species as well. 4.6 Conclusions We have identified a new mechanistic motif, which we call the reductive-coupled oxo activation mechanism, in which a normally unreactive main-group oxo becomes a potent hydrogen atom abstractor when coupled to a transition metal (in our case VV or MoVI ). The new activity of this oxo, which we quantify using the O−H dissociation parameter DH , is modified from that of the transition metal oxo itself and may be significantly higher for certain elements, specifically phosphorous in the +V oxidation state and selenium and tellurium in the +IV and +VI oxidation states. Our results are consistent with existing experimental and theoretical mechanistic work on the n-butane to maleic anhydride VOPO catalyst. We believe that knowledge of this new mechanistic motif will allow for the de novo rational design of future oxidation catalysts. As a showcase of the power of the ROA mechanistic motif, we have designed a potential inorganic V complex (OPO)fH V Oc (OH)s (H2 O)t that may serve as a homogeneous oxidation catalyst for alkanes. We have identified catalytic cycles for the oxidation of propane to propylene or isopropanol. Our likely catalytic cycle involves H-atom abstraction from propane, followed by recombination of the produced isopropyl radical and VIV species to form a VIII isopropoxy species. Ligand exchange results in the formation of free isopropanol and reoxidation regenerates the starting VV species. Our theoretical results are encouraging for us to continue our computational work in investigating reoxidation pathways, and to commence experimental work. 147 4.7 References [1] Centi, G.; Trifiro, F.; Ebner, J. R.; Franchetti, V. M. Chem. Rev. 1988, 88, 55–80. [2] Hodnett, B. K. Heterogeneous Catalytic Oxidation. Wiley: New York, 2000. [3] Cheng, M.-J. and Goddard, W. A. J. Am. Chem. Soc. 2013, 135, 4600–4603. [4] Schuurman, Y.; Gleaves, J. T. Ind. Eng. Chem. Res. 1994, 33, 2935–2941. [5] Hutchings, G. J.; Desmartinchomel, A.; Olier, R.; Volta, J. C. Nature 1994, 368, 41–45. [6] Cheng, M.-J.; Goddard, W. A.; Fu, R. Methods for providing bond activation catalysts and related catalysts, systems, and methods. U.S. Patent application 14/192,856, filed February 27, 2014. Patent pending. [7] Marino, N.; Hanson, S. K.; Müller, P.; Doyle, R. P. Inorg. Chem. 2012, 51, 10077–10079. [8] Herron, N.; Thorn, D. L.; Harlow, R. L.; Coulston, G. W. J. Am. Chem. Soc. 1997, 119, 7149–7150. [9] Solis-Ibarra, D.; Silvia, J. S.; Jancik, V.; Cummins, C. C. Inorg. Chem. 2011, 50, 9980–9984. [10] Pan, B. F.; Bezpalko, M. W.; Foxman, B. M.; Thomas, C. M. Organometallics 2011, 30, 5560– 5563. [11] (a) Derrah, E. J.; Martin, C.; Mallet-Ladeira, S.; Miqueu, K.; Bouhadir, G.; Bourissou, D. Organometallics 2013, 32, 1121–1128. (b) Derrah, E. J.; Ladeira, S.; Bouhadir, G.; Miqueu, K.; Bourissou, D. Chem. Commun. 2011, 47, 8611–8613. [12] (a) Pan, B. F.; Bezpalko, M. W.; Foxman, B. M.; Thomas, C. M. Dalton Trans. 2012, 41, 9083–9090. 148 (b) Bauer, R. C.; Gloaguen, Y.; Lutz, M.; Reek, J. N. H.; Bruin, B. de; Vlugt, J. I van der. Dalton Trans. 2011, 40, 8822–8829. (c) Mazzeo, M. Lamberti, M.; Massa, A.; Scettri, A.; Pellecchia, C.; Peters, J. C. Organometallics 2008, 27, 5741–5743. (d) Mazzeo, M.; Strianese, M.; Kuhl, O.; Peters, J. C. Dalton Trans. 2011, 40, 9026–9033. [13] Gloaguen, Y.; Jacobs, W.; Bruin, B. de; Lutz, M.; Vlugt, J. I. van der; Inorg. Chem. 2013, 52, 1682–1684. [14] Jaguar, version 7.9 ; Schrödinger, LLC: 2012, New York, NY. [15] (a) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.; Sitkoff, D.; Nicholls, A.; Ringnalda, M.; Goddard, W. A.; Honig, B. J. Am. Chem. Soc. 1994, 116, 11875–11882. (b) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.; Ringnalda, M. N.; Sitkoff, D.; Honig, B. J. Phys. Chem. 1996, 100, 11775–11788. [16] Sears, F. W.; Zemansky, M. W.; Young, H. D. University Physics, 6th Ed.; Addison-Wesley, 1982. p [17] The probe radius r is automatically calculated by Jaguar using the formula r = 0.5833 3 MW/ρ where MW is the molecular weight and ρ is the density of the solvent. [18] (a) Becke, A. D. Phys. Rev. A. 1998, 38, 3098–3100. (b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652. (c) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B. 1988, 37, 785–789. [19] Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310. [20] (a) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270–283; (b) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 284–298. [21] (a) Hehre, W. J.; Ditchfield, R; Pople, J. A. J. Chem. Phys. 1972, 56, 2257–2261. (b) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; Defrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654–3665. [22] Tissandier, M. D.; Cowen, K. A.; Feng, W. Y.; Gundlach, E.; Cohen, M. H.; Earhart, A. D.; Coe, J. V.; Tuttle, T. R. J. Phys. Chem. A. 1998, 102, 7787–7794. Calculated at standard conditions. [23] Wertz, D. H. J. Am. Chem. Soc. 1980, 102, 5316–5322. 149 [24] Bridgeman, O.C.; Aldrich, E.W. J. Heat Transfer 1964, 86, 279–286. [25] Pangborn, A. B.; Giardello, M. A.; Grubbs, R. H.; Rosen, R. K.; Timmers, F. J. Organometallics 1996, 15, 1518–1520. [26] Bartlett, P. D. Rec. Chem. Prog. 1950, 11, 47–51. [27] (a) Wife, R. L.; Oort, A. B. van; Doorn, J. A. van; Leeuwen, P. W. N. M. van Synthesis 1983, 71. (b) Baccolini, G.; Bazzocchi, M.; Carla, B. Eur. J. Org. Chem. 2001, 2229–2233. [28] Some recent publications that show similar chemistry of using VOCl3 to install hard multidentate anionic ligands onto vanadium include: (a) Kurmaev, D. A.; Kolosov, N. A.; Gagieva, S. C.; Borissova, A. O.; Tuskaev, V. A.; Bravaya, N. M.; Bulychev, B. M. Inorg. Chim. Acta 2013, 396, 136–143. (b) Villanneau, R.; ben Djamâa, A.; Chamoreau, L.-M.; Gontard, G.; Proust, A. Eur. J. Inorg. Chem. 2013, 1815–1820. (c) Tong, L. H.; Wong, Y.-L.; Lee, H. K.; Dilworth, J. R. Inorg. Chim. Acta 2012, 383, 91–97. (d) North, M.; Stewart, E. L.; Young, C. Tetrahedron: Asymmetry 2012, 23, 1218–1225. [29] Wolff, F.; Choukroun, R.; Lorber, C.; Donnadieu, B. Eur. J. Inorg. Chem. 2003, 628–632. [30] Romanowski, G. J. Mol. Cat. A: Chemical 2013, 368, 137–144. [31] Busacca, C. A.; Lorenz, J. C.; Grinberg, N.; Haddad, N.; Hrapchak, M.; Latli, B.; Lee, H.; Sabila, P.; Saha, A.; Sarvestani, M.; Shen, S.; Varsolona, R.; Wei, X.; Senanayake, C. H. Org. Lett. 2005, 7, 4277–4280. [32] Bickelhaupt, F.; van der Does, T. Biphosphine production. U.S. Patent US4874897 A, 1989. [33] Doorn, J. A. van; Frijns, J. H. G.; Meijboom, N. Recl. Trav. Chim. Pays-Bas 1991, 110, 441– 449.  150 Chapter 5 The mechanism of Ni(iPr−pybox) complexes as asymmetric Negishi C−C coupling catalysts 5.1 Abstract The 2,6-bis(4-isopropyl-2-oxazolin-2-yl)pyridine (iPr-pybox) ligand, when combined with a nickel salt in N ,N -dimethylacetamide (DMA), generates an in situ Ni(iPr−pybox) complex that catalyzes enantioselectively the cross-coupling of secondary sp3 -C substrates. This system is very notable for its property of stereoconvergence: both enantiomers of a racemic substrate may be converted into a single enantiomer of a product. This chapter will computationally explore the mechanism responsible for both the remarkable activity and selectivity of the Ni(iPr−pybox) system; such a mechanism is believed to deviate substantially from the canonical “textbook” mechanism of Negishi coupling. Furthermore, a generalized calculation scheme is presented that allows for the rapid enantioselectivity prediction of many related pybox ligands as well, in order to help predict the next generation of asymmetric Negishi coupling catalysts. 151 5.2 Introduction The controlled formation of C−C bonds is one of the central challenges of synthetic organic chemistry. Generally, the cross-coupling of two distinct organic moieties R and R0 is done by selecting precursors with opposite polarity, for example, an electrophilic R−X (where X is a leaving group) along with a nucleophilic R0 −[M0 ] (where M0 is a less electronegative element, often a main-group metal or metalloid). Late transition metals M are typically added as well in order to catalyze activity and encourage the formation of R−R0 as opposed to undesired side products. Ligands may also be added in order to further enhance the activity of the transition metal catalyst by the generation of in situ complexes. Scheme 5.1 shows the overall cross-coupling reaction, as well as a general scheme for the mechanism. This mechanism [1] begins with the oxidative addition of R−X by a low-valent M (hence the necessity of using a late transition metal). The X ligand is then exchanged for an R0 in the transmetalation step, and finally reductive elimination regenerates the starting M and releases the coupled product R−R0 . Scheme 5.1. Left: the overall C−C bond-forming cross-coupling reaction. Right: the general mechanism for the cross-coupling reaction. Table 5.1 shows a few examples of cross-coupling reactions. As already mentioned, the electrophile is typically R−X where X is a leaving group such as a halide or sulfonate. Substrates in which the R group is an aryl or alkenyl, or a resonance-stabilized sp3 radical such as an allyl or propargyl group, tend to be more active. The metal catalyst M is typically Cu, Ni, or Pd, with Pd being the most versatile element but also the rarest and most expensive. The nucleophile is R0 −[M0 ] where M0 may well be any element that forms a weaker bond to R0 than M, and/or a stronger bond to X. 152 Coupling Kumada Heck Sonogashira Negishi Stille Suzuki Hiyama Electrophile R−X R−X R−X R−X R−X R−X R−X Nucleophile R0 −MgBr alkene R0 −C− −CH R0 −ZnX R0 −SnR03 R0 −B(OR00 )2 R0 −SiR03 Catalyst Pd, Ni Pd Pd, Cu Pd, Ni Pd Pd Pd Reference [2] [3] [4] [5] [6] [7] [8] Table 5.1. Examples of various C−C cross-coupling reactions. This list is not exhaustive. 5.2.1 The Ni(iPr−pybox) system Another central challenge of synthetic organic chemistry is the enantioselective synthesis of chiral compounds. For cross-coupling reactions in which C−C bonds are formed, the carbon atoms being bonded are often chiral centers, necessitating techniques for asymmetric synthesis. For cross-coupling reactions, this may be done by adding a chiral ligand to complex the catalyst metal, thus preferentially coupling only one enantiomer of a racemic starting material. Alternatively, one may also simply begin with an enantiopure substrate and chose a coupling reaction that preserves chirality. The Ni(iPr−pybox) system, developed by Gregory Fu and coworkers, is a Negishi coupling catalyst that utilizes nickel and the 2,6-bis(4-isopropyl-2-oxazolin-2-yl)pyridine ligand, abbreviated as (iPr-pybox), and couples organic halides to organozinc complexes [9]. The (iPr-pybox) ligand contains two chiral centers and can exist in three diastereomers: the C2 enantiomers (((R, R)-iPr-pybox) and ((S, S)-iPr-pybox)), and the Cs meso diastereomer ((R, S)-iPr-pybox). If a chiral C2 ligand is used, asymmetric catalysis results. The complex and a summary of its catalytic scope is shown in Scheme 5.2 and Table 5.2 [9, 10]. Scheme 5.2. The general reaction of the Ni(iPr−pybox) Negishi coupling catalyst with benzylic secondary alkyl halides. α-Bromo amides and propargylic halides are also viable substrates that have been studied. This Ni(iPr−pybox) system is advantageous for several reasons [10]: 153 R0 −ZnBr Yield (%) ee (%) C6 H13 −ZnBr 89 96 N− −C−(CH2 )3 −ZnBr 82 64 91 91 69 94 47 72 91 98 7 39 96 8 41 99 Entry 1 2 3 4 5 6 R−X Cl−(CH2 )6 −ZnBr nBu−ZnBr 9 Ph−(CH2 )3 −ZnBr 76 98 10 BnO−(CH2 )3 −ZnBr 63 75 Table 5.2. Some examples of benzylic secondary alkyl halide substrates coupled by the Ni(iPr−pybox) catalyst. All reactions were performed under conditions as in Scheme 5.2, except for entries 7 and 8 which were performed at room temperature. The data in this table was taken from Arp and Fu [10]. 154 1. It is a Negishi coupling catalyst, utilizing organozinc reagents as the nucleophile. Organozinc compounds are relatively tractable and accessible; 2. The ligand is known and is not difficult to synthesize from commercially-available starting materials; 3. The system is not particularly oxygen or water sensitive, and the reaction may be run under air; 4. Coupling reactions with alkyl substrates are typically more difficult than with their aryl counterparts, and unwanted side reactions such as β-hydride elimination may be present. Secondary alkyls are particularly problematic due to their added steric bulk. The pybox system has been demonstrated to work with various secondary alkyl electrophiles such as benzylic halides, αbromo amides, and propargylic halides [11]. However, one of the Ni(iPr−pybox) system’s most interesting properties may be its stereoconvergence. A racemic starting material may be used to generate a nearly enantiopure product with high yield (See Scheme 5.2). This implies that the mechanism requires both the destruction of the starting material’s stereochemistry as well as the formation of the product’s stereochemistry, and indicates that the mechanism may be more complex than the existing “textbook” mechanism as shown in Scheme 5.1. 5.2.2 Existing computational work The mechanism for the Ni(iPr−pybox) system’s remarkable reactivity was computationally investigated by Lin et al., who modeled the coupling of 1-bromoindane (the electrophile in Table 5.2, entry 3) with methylzinc bromide as catalyzed by NiI ((S, S)−iPr−pybox)Br to form (R)- and (S)-1methylindane [12]. Their results, which are summarized in 5.1, claim that NiI ((S, S)−iPr−pybox)Br is first methylated by methylzinc bromide to form NiI ((S, S)−iPr−pybox)Me. Oxidation by 1bromoindane results in NiII ((S, S)−iPr−pybox)(Me)Br and the indanyl radical, which then recombine to form NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind) (ind = 1-indanyl), which undergoes reductive elimination to form the 1-methylindane product and to regenerate the NiI ((S, S)−iPr−pybox)Br starting catalyst. The mechanism put forth by Lin et al. supports the hypothesis of Fu et al. that a NiI -NiIII cycle is involved, as opposed to the Ni0 -NiII cycle of a classical coupling reaction. In addition, their conclusions show a decisive preference for the formation of the (S)-1-methylindane enantiomer, in line with experimental evidence. However, their methodology and analysis have several issues: 1. The density functional theory used, B3LYP, neglects van der Waals interactions, which typically increase in importance for large molecules; 155 Figure 5.1. The computationally studied mechanism by Lin et al. of Ni((S, S)−iPr−pybox) catalysis of methyl-indane coupling. The steps are: methylation and oxidation of NiI ((S, S)−iPr−pybox)Br to NiII ((S, S)−iPr−pybox)(Me)Br, recombination with indanyl radical to form NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind), and reductive elimination to form 1-methylindane product. All energies in kcal/mol. Note that for the first two steps (i.e. methylation and oxidation to NiII ), the energy differences are only estimates, as the actual calculations were only performed on the parent Ni(pybox) complexes. 2. The basis set used, 6-31G* (a Pople-type double-ζ Gaussian basis set with polarization functions on non-hydrogen atoms) is rather small and limited, and may lead to inaccurate energies [13]; 3. Whereas the authors claim to investigate the enantioselectivity of the Ni(iPr−pybox) system, in many reactions the simplified pybox ligand (without the isopropyl groups) was used in lieu of (iPr)-pybox to simplify calculations; 4. The authors model the methylzinc bromide nucleophile as a simple two-coordinate complex; however, the true nature of the zincate in solution is likely to be much more complex [14]; 5. Although the reaction conditions and all experimental investigations have used DMA as the solvent, the authors modeled the reaction in DMSO instead; 6. Only one radical attack pathway (i.e. that of the indanyl radical approaching axially to the Ni) was considered. Therefore, it is not surprising that enantioselectivity of the Ni(iPr−pybox) system, which can be measured using the metric ∆∆G‡ := ∆G‡R − ∆G‡S , was greatly overestimated at 8.7 kcal/mol, whereas the experimentally determined enantiomeric excesses correspond to a ∆∆G‡ value of 2–3 kcal/mol. Due to these deficiencies, we have undertaken our own computational study of the Ni(iPr−pybox) system’s asymmetric catalysis reaction, using larger basis sets, more accurate DFTs and molecular models, and more potential reaction pathways considered. We have found a substantially different mechanism from that of Lin et al., with an overall reaction barrier of 15.9 kcal/mol for the formation of S-methylindane and a ∆∆G‡ value of 2.7 kcal/mol, in line with the experimentally determined values [15]. In addition, we have found a useful protocol for the rapid determination of the ∆∆G‡ enantioselectivity values of other related Ni(pybox) systems. We present some new results 156 and predictions using this new method. 5.3 Materials and methods For our calculations, the specific reaction we investigated was the coupling of racemic 1-bromoindane with methylzinc iodide to form (S)-1-methylindane, using NiI ((S, S)−iPr−pybox)Br as a catalyst or precatalyst. The solvent was taken to be DMA and the temperature 258.15 K. All quantum mechanical calculations were carried out using the Jaguar software version 7.9 developed by Schrödinger Inc. [16]. Geometry optimizations were carried out on initial guess structures, and vibrational frequencies were gathered to confirm the optimized geometries as intermediates or transition states and to construct a free energy profile. Solvation energies were calculated using the PBF Poisson-Boltzmann implicit continuum solvation model [17] in Jaguar, with a dielectric constant of 37.78 [18] and a probe radius of 2.642 Å [19] for DMA. Geometry optimization and vibrational data were calculated using the B3LYP-D3 density functional [20] with a smaller basis set, whereas single point gas-phase and solvated energies were calculated using the same functional and a larger basis set. The “-D3” suffix refers to the post-SCF correction developed by Grimme et al. that accounts for van der Waals forces [21]. Here the smaller basis set consists of a modified double-ζ Los Alamos basis set and pseudopotential (LACVP**) [22] for transition metals, and the 6-31G** basis set [13] for the other atoms; whereas the larger basis set consists of the triple-ζ Los Alamos basis set and pseudopotential (LACV3P**++) for transition metals, the same LAV3P basis set and pseudopotential for heavy main-group elements, and the 6-311G**++ basis set [23] for the other atoms. Proposed intermediates and transition states were found to have zero and one negative frequencies, confirming their status as local energy minima and saddle points, respectively. The free energy for each molecular species in solution was calculated using the formula G = Egas + ∆Gsolv + ZPE + Hvib + 6kT − T [Svib + γ(Strans + Srot − α) + β] where the last term (with α = 14.3 e.u., β = 7.98 e.u., and γ = 0.54) is an empirical approximation for the change in the translational and rotational entropy of the molecule between the gas phase and the solution phase (due to the finite librational frequencies), derived from Wertz [24]. For the monatomic anions Br– and I– , the Wertz approximation was not necessary. Hence we used the equation 5 G = Egas + ∆Gsolv + kT − T S + 1.9 2 to construct the free energy. 157 For DMA, the Gibbs free energy was calculated using the formula Gliquid = Egas + ZPE + Htot − T Stot + ∆Ggas→liquid where ∆Ggas→liquid = Gliquid − Ggas (1atm) is the free energy of condensation to liquid from 1 atm gas. We can solve for this by noting that ∆Ggas→liquid = ∆Gexpansion + Ggas→solvation , where ∆Gexpansion = Ggas (P ) − Ggas (1 atm) is the expansion of the gas from 1 atm to the vapor pressure P , and Ggas→solvation is the condensation of gas at vapor pressure P to liquid. Since a liquid is by definition at equilibrium with its vapor pressure Ggas→solvation = 0, and we thus have  ∆Ggas→liquid = Ggas (P ) − Ggas (1 atm) = RT ln P 1 atm  . We can find the vapor pressure P at a given temperature using the Antoine Equation: log10 P = A − B , C +T where the empirical parameters A, B, and C vary with the solvent and temperature. For DMA at 258.15 K, A = 6.0888, B = 2725.96, C = 28.209, so the vapor pressure is 3.71 × 10−4 atm [25]. 5.4 Results In this section, we present the results of our calculations and the likely mechanism of C−C coupling that the Ni(iPr−pybox) system undergoes. We begin by first outlining the general overarching mechanism, and then examine in depth the details of each step. 5.4.1 The overall general mechanism In the previous work of Lin et al., the authors claimed early in their report that the Ni0 -NiII cycle that a typical Negishi catalyst might undergo had an overall barrier of 57.3 kcal/mol, and consequently was not feasible [12]. Despite our concerns about their methodology (see Section 5.2.2), we do not believe that addressing these concerns would have any potential to lower the barrier enough to make the Ni0 -NiII cycle accessible. Hence, we agree with their conclusion that the Ni(iPr−pybox) system most likely goes through a NiI -NiIII cycle. We therefore restrict our mechanistic investigations to the NiI -NiIII cycle. Scheme 5.3 shows the NiI -NiIII cycle in its most general form. The catalyst alternates between a 158 NiI and a NiIII state (for example, NiI ((S, S)−iPr−pybox)Br and NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind), respectively). Conversion of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind) should proceed via methylation and addition of indanyl and bromine radicals, in any order. Conversion of NiIII ((S, S)−iPr−pybox)(Me)(Br)(ind) back to NiI ((S, S)−iPr−pybox)Br should proceed via reductive elimination of 1-methylindane. The rest of this section will examine the detailed pathways of this generalized scheme. Scheme 5.3. The generalized mechanism of the Ni(iPr−pybox) Negishi coupling catalyst as it proceeds through a NiI -NiIII cycle. The overall equation is ind−Br + MeZnI −−→ ind−Me + ZnBrI. 5.4.2 Nature of the organozinc reagent In order to address how alkyl transfer of the methylzinc iodide to the Ni(iPr−pybox) complex occurs, it is necessary to first identify the exact nature of the methyl transferring species. Lin et al. depict the complexation of Ni(pybox) with a simple two-coordinate linear MeZnBr species in their calculations; however, it is unclear if such an electron-deficient species is at all present in a strongly polar solvent such as DMA. Indeed, the active methylation species has often been assumed to be a three coordinate MeZnX–2 [26]. Recent titration work by Organ et al. suggests, however, that the true methylating species is in fact four-coordinate dianionic MeZnX–3 with a Pd catalyst and a THF or THF/DMI 2:1 solvent (THF = tetrahydrofuran; DMI = N ,N 0 -dimethyl-2-imidazolidinone) [14]. For our mechanistic investigation, we decided to undertake a comprehensive computational investigation of the methylzinc and inorganic zinc species that may be present in the DMA solution under reaction conditions. Scheme 5.4 shows our calculated free energies of complexation of MeZnX and ZnX2 with one and two equivalents of X– , where X may be either bromine or iodine. We found that, contrary to the results of Organ et al., three-coordinate MeZnX–2 species are preferred; this discrepancy may be due to our choice of DMA as solvent, which has a much larger dielectric constant 159 of 37.78 compared to the THF as used by Organ et al. ( = 7.2). Scheme 5.4. Complexation of the various MeZnX (left) and ZnX2 (right) species with X– , where X ∈ {Br, I}. Numbers next to each arrow are the free energy change for the specific reaction. All numbers in kcal/mol. We also considered the possibility that the zinc species might form adducts with DMA, or cluster into polynuclear complexes. Given the strong polarity of the DMA solvent, we expected the former possibility to be more likely, but both pathways were explored. Figure 5.2 shows the various adducts and cluster complexes of MeZnX, whereas Figure 5.3 shows the various adducts and cluster complexes of ZnX2 . Figure 5.2. The various (MeZnI)m (dma)n species, where m may equal 0 (top row, left), 1 (bottom row), or 2 (top row, right); and n may equal 0, 1, or 2 as well. All numbers are free energies in kcal/mol and relative to MeZnI. The lowest energy species, MeZnI(dma)2 , is highlighted in red. Our results show that the most stable methylzinc species are MeZnBrI– and MeZnI(dma)2 , at −3.1 kcal/mol relative to the base MeZnI; and that the most stable inorganic zinc species is ZnBrI(dma)2 , at −15.2 kcal/mol relative to the base ZnBrI. For consistency, we have set MeZnI(dma)2 to be our active methylating agent and ZnBrI(dma)2 to be our ultimate product. Therefore, the overall reaction that we are investigating can now be written as ind−Br + MeZnI(dma)2 −−→ ind−Me + ZnBrI(dma)2 . 160 Figure 5.3. Left: The various (ZnBrI)m (dma)n species, where m may equal 0 (first row) or 1 (second to fifth rows); and n may equal 0 (first column), 1 (second column), or 2 (third and fourth columns). Right: The various (ZnBrI)4 tetrameric clusters. All numbers are free energies in kcal/mol and relative to ZnBrI. The lowest energy species, ZnBrI(dma)2 , is highlighted in red. 161 5.4.3 Mechanism for NiI oxidation and methylation We now address the mechanistic steps that must be taken for the top half of the cycle in Scheme 5.3, i.e. the conversion of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br. This conversion is expected to proceed through methylation with MeZnI(dma)2 and oxidation with ind−Br. Due to the stereoconvergent nature of the catalysis, in which the prior stereochemistry of the starting ind−Br is destroyed, the oxidation is not expected to be a typical one-step oxidative addition reaction. Instead, there is believed to be an initial one-electron oxidation in which NiII and a benzylicstabilized indanyl radical are produced, followed by recombination to form the NiIII species. These assertions are supported by experimental evidence (i.e. the formation of coupling products [15]), and the calculations of Lin et al. [12] support their feasibility. As a result, we have three possibilities for the conversion of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br: The NiI pathway: Methylation occurs first, converting NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)iPr−pybox)Me, and subsequent oxidation by ind−Br produces NiIII ((S, S)−iPr−pybox)(Me)(ind)Br; The NiII pathway: Oxidation by ind−Br to NiII ((S, S)−iPr−pybox)Br2 and free indanyl radical occurs first. Methylation to NiII ((S, S)−iPr−pybox)(Me)Br follows, and recombination with ind · produces NiIII ((S, S)−iPr−pybox)(Me)(ind)Br; The NiIII pathway: Oxidation by ind−Br proceeds to completion with NiIII ((S, S)−iPr−pybox)(ind)Br2 being formed; subsequent methylation produces NiIII ((S, S)−iPr−pybox)(Me)(ind)Br. These three pathways are shown in Scheme 5.5. The first branching point is the first step: NiI ((S, S)−iPr−pybox)Br may either undergo methylation as in the NiI pathway, or one-electron oxidation by ind−Br as in the NiII and NiIII pathways. Methylation of NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)−iPr−pybox)Me is thermodynamically uphill by 15.2 kcal/mol. Unfortunately, we were unable to find any transition states for the concerted transfer of the methyl group from MeZnI(dma)2 to any NiI species. However, we can consider a dissociative mechanism whereby the bromine on NiI ((S, S)−iPr−pybox)Br is exchanged for a methyl + group by the successive steps of Br– dissociation, MeZnI(dma)2 association, and ZnI(dma)2 dissociation (Scheme 5.6). Since each ligand association/disassociation step is expected to have minimal barriers if any, we can therefore gain an upper bound for the NiI methylation barrier by simply calculating the relative energies of each intermediate in Scheme 5.6. The results of our analysis shows that the overall barrier for the methylation of NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)−iPr−pybox)Me is between 15.2 and 20.7 kcal/mol. The NiII and NiIII pathways, by contrast, begin with the oxidation of NiI ((S, S)−iPr−pybox)Br 162 Scheme 5.5. Three potential pathways for the methylation and oxidation of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br. Red denotes the NiI pathway, green the NiII pathway, and blue the NiIII pathway. Open coordination sites may be occupied by dma solvent. Scheme 5.6. The dissociative pathway for the methylation of NiI ((S, S)−iPr−pybox)Br to NiI ((S, S)−iPr−pybox)Me, as part of the NiI pathway in Scheme 5.5. The transition state barriers for each associative/dissociative step is expected to be minimal, if existent. Hence, the overall barrier to methylation by this pathway is 20.7 kcal/mol. All numbers are in kcal/mol and relative to NiI ((S, S)−iPr−pybox)Br. 163 by ind−Br to NiII ((S, S)−iPr−pybox)Br2 and ind · . This oxidation most likely proceeds through an inner sphere route that begins with coordination of ind−Br to the Ni center. Our results are shown in Scheme 5.7. Initial coordination of (R)- and (S)-1-bromoindane with NiI ((S, S)−iPr−pybox)Br produces an adduct NiI ((S, S)−iPr−pybox)(Br)(Br−ind) that is favorable at −4.6 and −5.2 kcal/mol relative to the initial NiI species, respectively. Loss of ind · then results in NiII ((S, S)−iPr−pybox)Br2 , whose triplet energy is −9.3 kcal/mol relative to the initial NiI species. The coordination of a DMA solvent molecule further lowers the triplet energy to −11.9 kcal/mol relative to the initial NiI species. The transition state barriers to ind−Br coordination and ind · dissociation are indeed quite minimal at less than 1.7 kcal/mol, supporting our assumptions about the NiI methylation pathway. Since NiI methylation as per the NiI pathway is uphill thermodynamically by 15.2 kcal/mol, and NiI oxidation as per the NiII and NiIII pathways is downhill by 11.9 kcal/mol with a maximum barrier of 1.2 kcal/mol, we conclude that the NiII and NiIII pathways for NiI ((S, S)−iPr−pybox)Br methylation and oxidation are much more likely than the NiI pathway, which was the only pathway investigated by Lin et al. Scheme 5.7. The oxidation of NiI ((S, S)−iPr−pybox)Br by ind−Br to NiII ((S, S)−iPr−pybox)Br2 and ind · , as part of the NiII and NiIII pathways in Scheme 5.5. The transition state barrier for the coordination of DMA to NiII ((S, S)−iPr−pybox)Br2 is expected to be minimal, if existent. Hence, the overall barrier to oxidation by this pathway is 1.2 kcal/mol for (R)-1-bromoindane and 0.4 kcal/mol for (S)-1-bromoindane. All numbers are in kcal/mol and relative to NiI ((S, S)−iPr−pybox)Br. The second branching point is the subsequent reaction of NiII ((S, S)−iPr−pybox)Br2 (dma), which can either undergo methylation to NiII ((S, S)−iPr−pybox)(Me)Br as in the NiII pathway, or recombine with ind · to form NiIII ((S, S)−iPr−pybox)Br2 (ind) as in the NiIII pathway. Scheme 5.8 shows a comparison of the essential features of the two pathways, whereas Scheme 5.11 is an unabridged diagram of the NiII pathway. Methylation of NiII ((S, S)−iPr−pybox)Br2 (dma) to NiII ((S, S)−iPr−pybox)(Me)Br as in the NiII pathway is uphill by 5.4 kcal/mol. By contrast, recombination with ind · to form NiIII ((S, S)−iPr−pybox)Br2 (ind) as in the NiIII pathway is downhill by 164 1.2 or 5.8 kcal/mol, depending on whether the indanyl ligand attaches in an (R) or (S) configuration, respectively. Since the dissociation of DMA from NiII ((S, S)−iPr−pybox)Br2 (dma) is uphill by 2.6 kcal/mol, and the addition of ind · is expected to have minimal barrier, we can conclude that the overall barrier of the reaction of NiII ((S, S)−iPr−pybox)Br2 (dma) and ind · is likely lower than that of NiII ((S, S)−iPr−pybox)Br2 (dma) and MeZnI(dma)2 , and hence that the NiIII pathway is preferred over the NiII pathway. The NiI , NiII , and NiIII pathways all converge to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br in the final step. For the NiIII pathway, this involves the methylation of the NiIII ((S, S)−iPr−pybox)Br2 (ind) formed as described in the previous Scheme 5.8. As in the NiI methylation case, we were unable to find any transition states for the concerted transfer of the methyl group from MeZnI(dma)2 to any NiIII species. Hence we again considered a dissociative mechanism whereby a bromine on NiIII ((S, S)−iPr−pybox)Br2 (ind) is exchanged for a methyl group by the successive steps of Br– + dissociation, MeZnI(dma)2 association, and ZnI(dma)2 dissociation, assuming that our barriers are minimal and thus estimating our overall methylation barriers using thermodynamic energies only. A comprehensive accounting of all species studied for this NiIII methylation reaction is given in Scheme 5.12 in Section 5.7. Scheme 5.9 shows the essential features of the mechanism of NiIII methylation. The initial NiIII ((S, S)−iPr−pybox)Br2 (ind) may lose a bromo ligand, resulting in a five-coordinate cationic III + species [Ni ((S, S)−iPr−pybox)Br(ind)] that may exist in three separate but interchangeable geometric conformations. Each conformation may then trap an equivalent of MeZnI(dma)2 , with III eq ax the species [Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] + the most favorable at III −3.1 kcal/mol relative to the starting Ni ((S, S)−iPr−pybox)Br2 ((S)−ind). However, it is the III ax ax + isomer [Ni ((S, S)−iPr−pybox)Br (ind) −Meeq −ZnI(dma)2 ] III ax III that eventually leads to the low- ax est Ni (Me)(ind)Br species, Ni ((S, S)−iPr−pybox)Br (ind) Meeq . Dissociation of ZnBrI(dma) ax ax III from NiIII ((S, S)−iPr−pybox)Br (ind) −Meeq −ZnBrI(dma) leads to [Ni ((S, S)−iPr−pybox)ax Brax (ind) Meeq and is 6.8 or 4.5 kcal/mol uphill from the starting NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind), but coordination of ZnBrI(dma) with solvent back to ZnBrI(dma)2 drives this reaction forward. As we have assumed that association and dissociation barriers are minimal for these reactions, the overall barrier for the methylation of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) is apIII ax ax proximately the difference between the highest point ([Ni ((S, S)−iPr−pybox)Br (ind) Meeq + III eq ax + ZnBrI(dma)) and the lowest point ([Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] ), ax ax equal to 9.9 and 7.6 kcal/mol for the formation of NiIII ((S, S)−iPr−pybox)Br ((R)−ind) Meeq ax and NiIII ((S, S)−iPr−pybox)Brax ((S)−ind) Meeq , respectively. 165 Scheme 5.8. NiII ((S, S)−iPr−pybox)Br2 and its subsequent methylation as part of the NiII pathway, or recombination with ind · as part of the NiIII pathway. The lowest energy NiII ((S, S)−iPr−pybox)MeBr complex is 5.4 kcal/mol higher than the starting NiII ((S, S)−iPr−pybox)Br2 , so the overall barrier for methylation is at least this much. By contrast, the transition state barrier for the coordination of ind · to NiII ((S, S)−iPr−pybox)Br2 is expected to be minimal, if existent. Hence, the overall barrier to ind · recombination is 2.6 kcal/mol, and the NiIII pathway is thus more likely to occur. All numbers are in kcal/mol and relative to NiII ((S, S)−iPr−pybox)Br2 . Red denotes the starting NiII ((S, S)−iPr−pybox)Br2 (dma) complex, green the NiII pathway, and blue the NiIII pathway. Note that indanyl recombination is reversible, so that NiIII ((R)−iPr−pybox)Br2 ((S)−ind) and NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) can interconvert. 166 Scheme 5.9. The methylation of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) to produce NiIII ((S, S)−iPr−pybox)(Br)(ind)Me. The reaction initially proceeds to [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ , a thermodynamic sink, and must increase in energy in order to reach the NiIII ((S, S)−iPr−pybox)Brax (ind)ax Meeq product. All numbers are in kcal/mol and relative to NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind). Red denotes the lowest point and thermodynamic sink, [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ ; and green the highest point, NiIII ((S, S)−iPr−pybox)Brax (ind)ax Meeq + ZnBrI(dma). 5.4.4 Mechanism for reductive elimination After the formation of NiIII ((S, S)−iPr−pybox)(Me)(ind)Br, the final step to closing the catalytic cycle is the reductive elimination of ind−Me and the reformation of NiI ((S, S)−iPr−pybox)Br (Scheme 5.3, bottom half of cycle). In the computational work of Lin et al., only the reductive eq ax elimination of ind−Me from NiIII ((S, S)−iPr−pybox)(Me) (ind) Brax was considered. However, we believe that this is not the only possible pathway for the reductive elimination of ind−Me. Specifically: ax eq 1. Reductive elimination may proceed from either NiIII ((S, S)−iPr−pybox)(Me) (ind) Brax or eq ax NiIII ((S, S)−iPr−pybox)(Me) (ind) Brax ; III 2. It is possible for the bromo ligand to be lost, resulting in five-coordinate cationic [Ni + ((S, S)−iPr−pybox)(Me)(ind)] species that may also reductively eliminate; 3. It is possible for the five coordinate cationic species to be stabilized by a solvent ligand, III + resulting in the six-coordinate cationic [Ni ((S, S)−iPr−pybox)(Me)(ind)dma] species that may also reductively eliminate; 4. It is also possible for the five coordinate cationic species to coordinate to zinc, forming an adduct NiIII ((S, S)−iPr−pybox)(Me)(ind)−Br−[Zn] that may also reductively eliminate. 167 For the last possibility, the only transition states found that contained the Ni−Br−Zn moiety III – were the geometric isomers of [Ni ((S, S)−iPr−pybox)(Me)(ind)−Br−ZnBr2 I] . In all other cases, the Br−[Zn] moiety dissociated from the Ni during geometry optimization. Figure 5.4 shows all the transition states that have been found, and their energies relative III eq ax + to [Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] , the lowest energy state of the catalyst. A reaction scheme showing the pathways to all transition states is given in Section 5.7 III as Scheme 5.13. The lowest energy transition state that produces (R)-1-methylindane is [Ni ax eq – ((S, S)−iPr−pybox)(Me) ((R)−ind) −Brax −ZnBr2 I] , with an overall barrier of 18.6 kcal/mol, III and the lowest energy transition state that produces (S)-1-methylindane is [Ni ((S, S)−iPr−pybox)ax eq + (Me) ((S)−ind) ] , with an overall barrier of 15.9 kcal/mol. Hence, the preference for (S)-1methylindane is ∆∆G‡ = 2.7 kcal/mol, in line with experimentally determined enantiomeric excesses [15]. 5.5 Discussion 5.5.1 Overall catalytic cycle Figure 5.5 is a reaction coordinate diagram that summarizes our results for the investigation of the oxidation and methylation of NiI ((S, S)−iPr−pybox)Br to NiIII ((S, S)−iPr−pybox)(Me)(ind)Br, and the subsequent reductive elimination of ind−Me and reformation of NiI ((S, S)−iPr−pybox)Br. All three pathways are shown and the NiIII pathway is clearly the easiest. The lowest point and therIII eq ax + modynamic sink, [Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] , is now considered the resting state, and all other free energies given are relative to it. The overall barrier is 18.6 and 15.9 kcal/mol for the formation of (R)-1-methylindane and (S)-1-methylindane, respectively. 5.5.2 Predictions for new ligands Now that the mechanism of Ni((S, S)−iPr−pybox) asymmetric Negishi coupling catalysis has been ascertained, it is useful to consider how modification of the ((S, S)-iPr-pybox) ligand may affect enantioselectivity. The most obvious point of modification is the R groups on the oxazoline moieties. However, the synthesis and mechanistic analysis of a multitude of R−pybox ligands would be expected to be costly and time-consuming. Hence, it is better to computationally examine a generalized mechanism of R−pybox catalysis that may be systematically and rapidly adapted for different R groups, for high-throughput screening of new R−pybox catalysts for asymmetric Negishi coupling. A generalized catalytic mechanism for the Ni(R−pybox) complexes is given in Scheme 5.10. The starting complex NiI (R−pybox)Br, G, is oxidized and methylated to NiIII (R−pybox)(Me)(ind)Br (not shown). The NiIII (R−pybox)(Me)(ind)Br species, which can exist in three geometric isomers 168 Figure 5.4. Transition states for the reductive elimination of 1-methylindane from various NiIII ((S, S)−iPr−pybox)(Me)ind species. All numbers are in kcal/mol and relative to [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ . The lowest transition states for the formation of (R)- and (S)-1methylindane are highlighted in green and blue, respectively. 169 Figure 5.5. Reaction coordinate diagram of the entire catalytic cycle for the coupling of ind−Br and MeZnI to form ind−Me. All numbers are in kcal/mol and relative to [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ . III + A1, A2, and A3, may lose Br– to form the five coordinate cationic [Ni (R−pybox)(Me)ind] species B1, B2, and B3, respectively. A DMA solvent molecule may then coordinate to the open III + coordination site to form cationic [Ni (R−pybox)(Me)(ind)dma] species C1, C2, and C3, re- spectively. Neutral A1 and A2 (but not A3) may also undergo reductive elimination through the transition states D1 and D2, respectively, to release methylindane and reform the starting complex G. Likewise, B1, B2, and B3 may undergo reductive elimination through E1, E2, and E3, I + respectively, to release methylindane and form cationic [Ni (R−pybox)] , H, which can regain the bromide to reform the starting complex G. Finally, cationic C1 and C2 (but not C3) may undergo reductive elimination through the transition states F1 and F2, respectively, to release methylindane I + and cationic [Ni (R−pybox)dma] , J, which can replace its dma ligand with a bromide to reform the starting complex G. In order to rapidly calculate the enantioselectivity of Ni(R−pybox) catalysts containing different R groups, we simply need to find the transition states D, E, and F. (In our future analyses, we will also consider Zn adducts as well, but that is not included in this study.) For our demonstration, we have selected potential R groups to be adamantyl (Ad), cyclohexyl (Cy), methyl (Me), phenyl (Ph), (R)-sec-butyl ((R)−sBu), (S)-sec-butyl ((S)−sBu), tert-butyl (tBu), and 2-phenylethyl (C2 H4 Ph). In addition, we also considered the a variant pyridine bis(indenoxazoline) ligand with fused indane rings, which we term pybindox. These are shown in Figure 5.6. Our results are given in Table 5.3. A complete collection of all calculated intermediate and transition state energies is given in Tables 5.4 through 5.6 in Section 5.7. Despite the variety of different R groups investigated, several universal characteristics may be observed from the data in Tables 5.3 through 5.6. In every case, the lowest energy intermediate was a neutral A species, usually the A2 isomer. However, the lowest transition states are of type E, 170 Scheme 5.10. A generalized catalytic mechanism that Ni(R−pybox) complexes undergo in Negishi coupling. This scheme is used to rapidly calculate the predicted enantioselectivity of a variety of complexes related to the original Ni(iPr−pybox) system. Figure 5.6. Modified ligands studied which, when complexed to Ni, may allow for different enantioselective properties as compared to the original (iPr−pybox) ligand. 171 III R iPr Ad Cy Me Ph (R)−sBu (S)−sBu tBu C2 H4 Ph pybindox [Ni (L)(Me −−− ind)Br]‡ (D) minimum ∆G‡ ∆∆G‡ (R) (S) [Ni (L)(Me −−− ind)] ‡ (E) minimum ∆G‡ ∆∆G‡ (R) (S) + [Ni (L)(Me −−− ind)dma] ‡ (F) minimum ∆G‡ ∆∆G‡ (R) (S) 18.4 19.1 28.5 20.3 17.7 18.3 18.1 18.9 17.5 20.5 15.7 14.9 20.3 15.2 14.9 12.5 15.4 16.1 12.8 13.7 3.3 −1.5 1.5 1.4 2.3 0.5 1.9 −0.6 2.1 0.3 16.4 19.3 N/A N/A 17.1 N/A N/A 19.5 N/A 17.3 17.3 19.6 28.9 18.9 16.2 15.7 17.0 18.9 19.4 18.9 1.1 −0.5 −0.3 1.4 1.6 2.6 1.2 0.0 −1.9 1.7 III 12.4 16.4 18.8 13.8 12.6 12.1 13.5 16.7 10.7 13.3 III + 15.3 17.2 30.9 19.2 15.8 N/A 16.5 18.3 14.7 15.0 1.1 2.1 N/A N/A 1.3 N/A N/A 1.2 N/A 2.3 Table 5.3. Comparison of activation energies for transition states of type D, E, and F for various Ni(R−pybox) catalysts. The Ni(iPr−pybox) data is also given here for comparison. For each transition state type, only the energy of the lowest geometric isomer is reported. For all R groups, cationic five-coordinate transition states (E) were most favored. The (R)-(S) ∆∆G‡ value is also calculated for each transition state type and (R−pybox) ligand. All numbers are in kcal/mol. Energies given are relative to the lowest calculated type A neutral NiIII (R−pybox)(Me)(ind)Br complex. Numbers in blue are the lowest energy transition states for the production of (R)-1-methylindane, whereas numbers in red are the lowest energy transition states for the production of (S)-1-methylindane. Numbers in green are the final ∆∆G‡ values for each R. (L) is ((S, S)−R−pybox). although the particular isomer (i.e. E1, E2, or E3) may vary. The enantioselectivity as measured by ∆∆G‡ varies from −1.5 to 2.3 kcal/mol, in all cases inferior to the original Ni(iPr−pybox)’s ∆∆G‡ value of 2.7 kcal/mol. In our future studies, we will seek to gain more accurate ∆∆G‡ values for our existing R groups by also considering Zn adduct formation, as we have done for Ni(iPr−pybox); and broaden our search to examine more R groups. 5.6 Conclusions In this chapter, we have computationally studied the mechanism of the Ni(iPr−pybox) system, which is a enantioselective and stereoconvergent Negishi C−C coupling catalyst. Our results support existing experimental and theoretical evidence that the mechanism proceeds through a NiI -NiIII cycle, with the stereochemistry of the initial ind−Br destroyed by homolytic cleavage of the C−Br bond. We found that NiI ((S, S)−iPr−pybox)Br is initially oxidized by ind−Br, and then methylated to form NiIII ((S, S)−iPr−pybox)(Me)(ind)Br, which can form an adduct with Zn to result in III eq ax + the resting state [Ni ((S, S)−iPr−pybox)Br ((S)−ind) −Meax −ZnI(dma)2 ] . Reductive elimiIII ax eq – nation then proceeds through either [Ni ((S, S)−iPr−pybox)(Me) ((R)−ind) −Brax −ZnBr2 I] III at an overall barrier of 18.6 kcal/mol to form (R)-1-methylindane), or [Ni ((S, S)−iPr−pybox)ax eq + (Me) ((S)−ind) ] at an overall barrier of 15.9 kcal/mol to form (S)-1-methylindane, with an enantioselectivity of ∆∆G‡ = 2.7 kcal/mol preference for forming the (S) enantiomer. These results 172 are also in line with experimental evidence. In addition, we formulated a systematic approach that allows us to rapidly study related Ni(R−pybox) systems, where R is Ad, Cy, Me, Ph, (R)−sBu, (S)−sBu, tBu, or C2 H4 Ph, or the III + ligand is pybindox. For these cases, we have found [Ni (R−pybox)(Me)(ind)] (E) to be universally the lowest energy transition states for the production of either (R)- or (S)-1-methylindane. Our ∆∆G‡ values measured range from −1.5 to 2.3 kcal/mol, which are not as good as the original Ni(iPr−pybox). We are continuing to improve our screening system and broadening our scope to find new catalysts for efficient asymmetric and stereoconvergent catalysis for C−C coupling. 5.7 Supporting figures Scheme 5.11 is an expanded diagram of the NiII pathway, whose essential features were presented in Scheme 5.8. Although methylation of NiII ((S, S)−iPr−pybox)Br2 to NiII ((S, S)−iPr−pybox)(Me)Br is uphill by 5.4 kcal/mol, further methylation to NiII ((S, S)−iPr−pybox)Me2 is uphill by an additional 16.7 kcal/mol. Reductive elimination of ethane is thermodynamically favorable, but the loss of methyl radical from either NiII ((S, S)−iPr−pybox)(Me)Br or NiII ((S, S)−iPr−pybox)Me2 is very uphill. Hence, the overall barrier to ethane production is at least 22.1 kcal/mol, the energy of the lowest NiII ((S, S)−iPr−pybox)Me2 species. Scheme 5.12 is an expanded version of Scheme 5.9, showing the methylation of NiIII ((S, S)−iPrIII pybox)Br2 ((S)−ind) to [Ni ((S, S)−iPr−pybox)Br(ind)Me, as well as all side reactions and isomers investigated. Scheme 5.13 shows the reactions of NiIII ((S, S)−iPr−pybox)(Br)(ind)Me that lead to ind−Me reductive elimination. (R)- and (S)-1-methylindane are produced in separate pathways, highlighted in red and blue, respectively. The overall barriers are 18.6 kcal/mol for the production of (R)-1methylindane and 15.9 kcal/mol for production of (S)-1-methylindane, leading to a ∆∆G‡ preference of 2.7 kcal/mol for the (S) isomer. Tables 5.4 through 5.6 show the calculated energies of all transition states and intermediates A through J for the high-throughput screening of Ni(R−pybox) catalysts, where R is Ad, Cy, Me, Ph, (R)−sBu, (S)−sBu, tBu, or C2 H4 Ph, or the ligand is pybindox. Specifically, Table 5.4 shows the neutral sequence A −−→ D −−→ G; Table 5.5 shows the cationic sequence B −−→ E −−→ H; and Table 5.6 shows the cationic sequence C −−→ F −−→ J (see Scheme 5.10). A future study will also include Ni−Br−Zn adducts as well. 173 Scheme 5.11. Expanded diagram of NiII ((S, S)−iPr−pybox)Br2 methylation as part of the NiII pathway. The lowest energy NiII ((S, S)−iPr−pybox)MeBr complex is 5.4 kcal/mol higher than the starting NiII ((S, S)−iPr−pybox)Br2 , and the lowest energy NiII ((S, S)−iPr−pybox)Me2 complex is 22.1 kcal/mol higher than the starting NiII ((S, S)−iPr−pybox)Br2 . All numbers are in kcal/mol and relative to NiII ((S, S)−iPr−pybox)Br2 . The lowest NiII ((S, S)−iPr−pybox)Br2 , NiII ((S, S)−iPr−pybox)(Me)Br, and NiII ((S, S)−iPr−pybox)Me2 species are highlighted in red. 174 Scheme 5.12. Expanded diagram of NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind) methylation as part of the NiIII pathway. All numbers are in kcal/mol and relative to NiIII ((S, S)−iPr−pybox)Br2 ((S)−ind). Red denotes the lowest point and thermodynamic sink, [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ , as well as an alternative thermodynamic sink NiIII ((S, S)−iPr−pybox)Brax ((S)−ind)eq −Brax −ZnR MeI(dma); and green the highest point, NiIII ((S, S)−iPr−pybox)Brax (ind)ax Meeq + ZnBrI(dma), in the production of [NiIII ((S, S)−iPr−pybox)(Br)(ind)Me. 175 Scheme 5.13. The reactions of the various NiIII ((S, S)−iPr−pybox)(Me)ind species that lead to the reductive elimination of ind−Me. All numbers are in kcal/mol and relative to [NiIII ((S, S)−iPr−pybox)Breq ((S)−ind)ax −Meax −ZnI(dma)2 ]+ . The starting point is NiIII ((S, S)−iPr−pybox)(Br)ax (ind)ax Meeq , which was produced at the end of Schemes 5.9 or 5.12; the end point is NiI ((S, S)−iPr−pybox)Br. Both the starting and end points are highlighted in red. The lowest energy pathways leading to the formation of (R)- and (S)-1-methylindane are highlighted in green and blue, respectively, when separate and in purple when concurrent. 176 R Ad Cy Me Ph (R)−sBu (S)−sBu tBu C2 H4 Ph pybindox NiIII ((S, S)−R−pybox)(Me)(ind)Br A1 A2 A3 (R) (S) (R) (S) (R) (S) 4.9 9.8b 5.8 0.0 3.8 3.0 5.0 2.5 5.5 2.8 8.1 2.1 1.7 0.7 1.0 3.1 0.9 2.5 2.1 2.1 1.4 1.2 2.7 2.3 2.1 2.6 1.3 0.0 0.0 0.0 1.2 0.0 0.0 0.0 0.0 0.0 5.0a 6.7a 7.7 9.4 3.7a 8.2 4.8a 9.5 5.8 10.3 11.3 6.6 8.4 6.8 6.1 9.8 6.2 5.0 III [Ni (L)(Me −−− ind)Br]‡ D1 D2 (R) (S) (R) (S) NiI (L)Br 19.1b 28.5b 20.3 17.7 18.3 18.1 18.9b 17.6 20.5 −23.9 −24.1 −27.2 −21.7 −32.1 −27.5 −25.3 −27.3 −28.0 19.6b 28.9b 18.9 16.2 15.7 17.0 18.9b 19.4 18.9 21.3 32.5 21.0 19.0 20.5 21.3 21.3 17.5 21.0 20.0 32.5 19.9 19.0 19.8 22.1 19.9 20.0 20.5 G Table 5.4. Relative energies of the neutral A, D, and G species for various Ni(R−pybox) complexes. All numbers are in kcal/mol. Energies given are relative to the lowest calculated type A neutral NiIII (R−pybox)(Me)(ind)Br complex. Numbers in blue are the lowest energy (R)−ind geometric isomers for A and D, whereas numbers in red are the lowest energy (S)−ind geometric isomer for A and D. (L) is ((S, S)−R−pybox). a) The Ni−ind bond is very long and likely to be very labile. b) An oxazoline ligand is dissociated from the Ni center in the geometry optimized structure. III R Ad Cy Me Ph (R)−sBu (S)−sBu tBu C2 H4 Ph pybindox + [Ni ((S, S)−R−pybox)(Me)(ind)] B1 B2 B3 (R) (S) (R) (S) (R) (S) 13.6 8.8 11.1 10.4 7.9 9.4 5.8 8.1 10.8 9.5 7.5 6.5 6.5 5.1 6.7 10.1 0.6 6.2 6.6 5.6 3.3 4.3 2.7 4.6 5.9 2.6 2.5 5.3 5.0 2.9 2.5 0.9 3.3 4.9 1.9 1.5 16.3a 11.7a 14.9a 15.2a 16.1a 16.0a N/Aa c 15.7a N/Aa c N/Aa c 10.6a 13.2 13.6 12.4a 13.1 12.4a 12.7 12.6 III + [Ni (L)(Me −−− ind)] ‡ E1 E2 E3 (R) (S) (R) (S) (R) (S) 21.2 N/Ac N/Ad 14.9 12.5 15.4 16.1 14.6 17.3 19.1 N/Ac N/Ad 12.6 12.1 13.5 19.7 7.0 16.2 14.9 20.3 15.2 15.1 13.8 15.8 16.9 9.1 13.7 16.4 19.4 14.9 14.8 14.0 16.0 16.7 9.6 13.3 N/Ac 26.6 17.1 N/Ac N/Ac N/Ac N/Ac 12.1 N/Ac N/Ac 18.8 13.8 N/Ac N/Ab N/Ab 20.7 10.3 15.2 I + [Ni (L)] H −13.0 −15.1 −5.4 −9.7 −18.4 −13.4 −16.2 −16.3 −11.6 Table 5.5. Relative energies of the cationic B, E, and H species for various Ni(R−pybox) complexes. All numbers are in kcal/mol. Energies given are relative to the lowest calculated type A neutral NiIII (R−pybox)(Me)(ind)Br complex. Numbers in blue are the lowest energy (R)−ind geometric isomer for B and E, whereas numbers in red are the lowest energy (S)−ind geometric isomer for B and E. (L) is ((S, S)−R−pybox). a) The Ni−ind bond is very long and likely to be very labile. b) Geometry optimized to a Meax indeq (B1/E1) configuration. c) Geometry optimized to a Meeq indax (B2/E2) configuration. d) Geometry optimized to a Meax indax (B3/E3) configuration. 177 III R Ad Cy Me Ph (R)−sBu (S)−sBu tBu C2 H4 Ph pybindox + [Ni ((S, S)−R−pybox)(Me)(ind)dma] C1 C2 C3 (R) (S) (R) (S) (R) (S) 4.1 7.4 3.8 0.4 1.7 1.8 5.8 4.3 3.7 1.5 6.9 3.5 2.3 −0.2 1.9 3.1 2.7 −0.1 0.9 1.3 3.1 2.6 3.8 3.3 2.8 1.2 −0.7 −0.3 0.6 2.5 1.7 1.6 1.4 0.8 −0.6 −2.6 4.9a 4.8a 7.0a 3.9a 7.9a 7.1a 5.1a 4.5a 4.4a 2.6a 5.4a 6.4a 2.8a 3.2a 4.3a 5.6a 4.5a 3.5a III + [Ni (L)(Me −−− ind)dma] ‡ F1 F2 (R) (S) (R) (S) 19.3 N/Ab N/Ab 17.1 N/Ab N/Ab 19.5 N/Ab 17.3 17.2 30.9 19.2 15.8 N/Ab 16.5 18.3 14.7 15.0 N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab N/Ab I + [Ni (L)dma] J −25.9 −27.2 −24.6 −25.7 −30.0 −27.3 −26.5 −28.2 −25.2 Table 5.6. Relative energies of the cationic C, F, and J species for various Ni(R−pybox) complexes. All numbers are in kcal/mol. Energies given are relative to the lowest calculated type A neutral NiIII (R−pybox)(Me)(ind)Br complex. Numbers in blue are the lowest energy (R)−ind geometric isomer for C, whereas numbers in red are the lowest energy (S)−ind geometric isomer for C. (L) is ((S, S)−R−pybox). a) The Ni−ind bond is very long and likely to be very labile. b) A species containing a Ni−dma bond could not be found. 178 5.8 References [1] de Meijere, A.; Diederich, F. Metal-Catalyzed Cross-Coupling Reactions, 2nd ed.; Wiley-VCH: Weinheim, Germany; 2004. [2] Tamao, K.; Sumitani, K.; Kumada, M. J. Am. Chem. Soc. 1972, 94, 4374–4376. [3] Heck, R. F.; Nolley, Jr., J. P. J. Org. Chem. 1972, 37, 2320–2322. [4] Sonogashira, K. J. Organomet. Chem. 2002, 653, 46–49. [5] King, A. O.; Okukado, N.; Negishi, E. J. Chem. Soc. Chem. Comm. 1977, 19, 683–684. [6] Stille, J. K. Angew. Chem. Int. Ed. Engl. 1986, 25, 508–524. [7] Miyaura, N.; Yamada, K.; Suzuki, A. Tet. Lett. 1979, 20, 3437-3440. [8] Hatanaka, Y.; Hiyama, T. J. Org. Chem. 1988, 53, 918–920. [9] Fischer, C.; Fu, G. C. J. Am. Chem. Soc. 2005, 127, 4594–4595. [10] Arp, F. O.; Fu, G. C. J. Am. Chem. Soc. 2005, 127, 10482–10483. [11] Smith, S. W.; Fu, G. C. J. Am. Chem. Soc. 2008, 130, 12645–12647. [12] Lin, X; Sun, J.; Xi, Y.; Lin, D. Organometallics 2011, 30, 3284–3292. [13] (a) Hehre, W. J.; Ditchfield, R; Pople, J. A. J. Chem. Phys. 1972, 56, 2257–2261. (b) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; Defrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654–3665. [14] Angew. Chem. Int. Ed. 2012, 51, 7024–7027. [15] Buscagan, T. M.; Schley, N.; Fu, G. C. Manuscript under preparation. [16] Jaguar, version 7.9 ; Schrödinger, LLC: 2012, New York, NY. 179 [17] (a) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.; Sitkoff, D.; Nicholls, A.; Ringnalda, M.; Goddard, W. A.; Honig, B. J. Am. Chem. Soc. 1994, 116, 11875–11882. (b) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.; Ringnalda, M. N.; Sitkoff, D.; Honig, B. J. Phys. Chem. 1996, 100, 11775–11788. [18] Sears, F. W.; Zemansky, M. W.; Young, H. D. University Physics, 6th Ed.; Addison-Wesley, 1982. p [19] The probe radius r is automatically calculated by Jaguar using the formula r = 0.5833 3 MW/ρ where MW is the molecular weight and ρ is the density of the solvent. [20] (a) Becke, A. D. Phys. Rev. A. 1998, 38, 3098–3100. (b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652. (c) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B. 1988, 37, 785–789. [21] Grimme, S.; Antony, J.; Erlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104. [22] Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310. [23] (a) Hehre, W. J.; Ditchfield, R; Pople, J. A. J. Chem. Phys. 1972, 56, 2257–2261. (b) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; Defrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654–3665. [24] Wertz, D. H. J. Am. Chem. Soc. 1980, 102, 5316–5322. [25] Gopal, R.; Rizvi, S.A. J. Indian Chem. Soc. 1968, 45, 13–16. [26] Ochiai, H.; Jang, M.; Hirano, K.; Yorimitsu, H.; Oshima, K. Org. Lett. 2008, 10, 2681–2683.  180 Appendix A C−H activation using RhI complexes supported by bis-nitrogen chelating ligands Reproduced in part with permission from Webster-Gardiner, M. S.; Fu, R.; Fortman, G. C.; Gunnoe, T. B.; Nielsen, R. J.; Goddard, W. A. III. Catal. Sci. Technol., submitted for publication. Unpublished work c 2014 Royal Society of Chemistry. A.1 Abstract F F Rhodium (I) complexes (DAB )Rh(COE)(TFA) 1 and (BOZO)Rh(COE)(TFA) 2 ((DAB ) = N, N bis(pentafluorophenyl)-2,3-dimethyl-1,4-diaza-1,3-butadiene, (COE) = cyclooctene, (TFA) = trifluoroacetate, (BOZO) = bis(2-oxazolin-2-yl)) have been synthesized. Complexes 1 and 2 are catalyst precursors for the efficient C−H bond activation of arenes with TFA−D. DFT calculations determined that H/D exchange is possible through reversible oxidative addition of benzene onto the RhI complex. Catalyst 1 is reported to have a TOF of 227.5 h−1 and is the most active H/D exchange catalyst reported to date. 181 A.2 Introduction In the search for a rhodium analogue to the Shilov cycle catalyst, putative rhodium complexes are often first experimentally tested for their C−H activation ability by investigating their ability to catalyze the exchange of hydrogen and deuterium atoms on various aliphatic and aromatic substrates. towards this end, Kloek, Heinekey, and Goldberg have reported an example of a rhodium complex containing a (PNP) pincer ligand that catalyzes the H/D exchange of benzene with D2 O with a turnover frequency (TOF) of 0.1 h−1 [1]. Goldberg and Jones et al. have also reported H/D exchange with deuterated trifluoroacetic acid (TFA−D) and benzene with a pendant acetate rhodium complex, which had a TOF of 1.6 h−1 [2]. Despite the success of these rhodium catalysts for the C−H activation of arenes, the rates of reaction as determined by turnover frequency have been significantly slower, e.g., by an order of magnitude, than comparable results for Pd and Pt based catalysts [3]. Bidentate nitrogen chelating ligands have been used successfully for H/D exchange with these Pd and Pt systems. In this chapter we report the reactivity and computational study of two new rhodium complexes with bidentate nitrogen chelating ligands which have rates of arene H/D exchange comparable to platinum/palladium systems. A.3 Materials and methods F F Experimentally, complexes (DAB )Rh(COE)(TFA) 1 and (BOZO)Rh(COE)(TFA) 2 ((DAB ) = N, N -bis(pentafluorophenyl)-2,3-dimethyl-1,4-diaza-1,3-butadiene, (COE) = cyclooctene, (TFA) = trifluoroacetate, (BOZO) = bis(2-oxazolin-2-yl)) (Scheme A.1) were synthesized by reaction of [(COE)2 Rh(TFA)]2 dimer with 1 equivalent of ligand in THF at room temperature for 1 hour to afford 1 and 2 as purple solids in 58% and 73% yield, respectively. Both of these complexes were characterized by 1 H, 13 C, and 19 F NMR spectroscopy, and elemental analysis. Computational investigations were made as described in Section 3.3. A.4 Results and discussion A.4.1 Experimental results We screened these rhodium complexes for C−H activation by monitoring H/D exchange between d1 -trifluoroacetic acid and benzene. Under our conditions (1.6 mol % Rh, 1 equiv. C6 H6 in 17.5 equiv. of d1 -trifluoroacetic acid, 2 hours, 800 psi Ar, at 130◦ C), both complexes 1 and 2 showed rapid H/D exchange with turnover numbers of 89±13 and 90±11, respectively, resulting in turnover frequencies (TOFs) of 0.01 s−1 over the 2 hour period for both catalysts, as determined by GC-MS. Minimal turnovers were detected for the reaction of just [(COE)2 Rh(TFA)]2 in trifluoroacetic acid 182 Scheme A.1. Synthesis of rhodium H/D exchange catalysts (DABF )Rh(COE)(TFA) 1 and (BOZO)Rh(COE)(TFA) 2. under our conditions. Measuring the effect of temperature on catalysis (Figure A.1) shows that highest turnovers at 2 hours occur at 150◦ C. TON decreases at temperatures above 150◦ C, most likely due to catalyst decomposition. Complex 2 has the highest TON when the temperature is 130◦ C, with a decrease in turnovers at higher temperatures. Figure A.1. Temperature dependence vs. turnover numbers for complexes 1 and 2. Reactions were performed using 1.6 mol % Rh, 1 equiv. C6 H6 in 17.5 equiv. d1 -trifluoroacetic acid, for 2 hours, under 800 psi Ar. To determine the optimal solvent for catalysis, H/D exchange of benzene was then examined with other deuterated solvents at the optimized temperature. However, no significant measurable H/D exchange of benzene above the background reactions was observed for D2 O, CD3 OD, D3 PO4 , D2 SO4 , or CD3 COOD for either 1 or 2. 183 We then sought to determine the dependence on catalyst concentration. By lowering the catalyst loading to 0.4% relative to benzene, a large increase in TON was observed (Figure A.2). After 2 hours at 150◦ C, complex 1 showed 456 TON to give a TOF of 228 h−1 . Figure A.2. Concentration of Rh dependence vs. TON for complexes 1 and 2. Reactions were performed at 150◦ C for 1 and 130◦ C for 2, with 1 equiv. C6 H6 in 17.5 equiv. d1 -trifluoroacetic acid, for 2 hours, under 800 psi Ar. The selectivity of the reaction mechanism was determined by examining the H/D exchange of toluene in trifluoroacetic acid at each catalyst’s respective optimal temperature. The ortho:meta:para selectivity is 6.87 : 1 : 6.39 and 5.68 : 1 : 5.24 for 1 and 2, respectively. No clear deuteration of the methyl fragment of toluene was seen. A crucial aspect of catalysis is the recyclability of the catalyst. To study this, each complex was run and sampled, and then the contents of the reaction were evaporated off, leaving behind the catalyst. Fresh trifluoroacetic acid and benzene were added to the reaction vessel and the experiment was run again. For complex 1, this was successfully done over four times for a period of over 72 hours, with H/D exchange observed each time. An interesting question is the role of cyclooctene in the mechanism. The addition of cyclooctene to the experiment showed a clear decrease in turnover number for complex 2. We hypothesized that this is due to COE binding to the rhodium metal competitively to slow down catalysis. However, upon careful consideration of the 1 H and 13 C NMR spectra, clear peaks for the trifluoroacetate adduct of cyclooctene can be seen. Indeed, a report from Nordlander indicates that cyclooctene in trifluoroacetic acid funnels to a cyclooctyl trifluoroacetate complex [4]. Attempts to synthesize either complex with a trifluoroacetic acid ligand instead of cyclooctene were unsuccessful. Attempts to extend this reactivity to substrates other than aromatics were investigated. H/D exchange was attempted with methane under a variety of pressures and temperatures; however, no 184 deuterated methane species was observed by GC-MS. In addition, H/D exchange of benzene sp2 C−H bonds with cyclohexane was also attempted, but no deuteration was observed after 2 hours at the optimal conditions for each catalyst. A.4.2 Computational results We have also undertaken a computational investigation for the mechanism of H/D exchange by F complexes 1 and 2. We set our reference complex to (L)Rh(TFA)(TFAH) (3 and 4 for L = (DAB ) or (BOZO), respectively); these being the presumed species after the (COE) ligand is hydrotrifluoroacetylated off from the base (L)Rh(TFA)(COE) species. We hypothesized that C−H activation of benzene could proceed through one of two routes. The first route is by direct oxidative addition I III to (L)Rh (TFA)(TFAH) to form (L)Rh (TFA)(TFAH)(Ph)(H). H/D exchange could then occur beI − * cause the RhIII (H) bond can be reversibly reductively deprotonated: RhIII (TFA)(H) − ) − − Rh (TFAH). I III The second is by an internal protonation (L)Rh (TFA)(TFAH) −−→ (L)Rh (TFA)2 (H), followed by either benzene coordination and deprotonation by TFA via a six-membered ring transition state, F or direct hydrogen exchange with the RhIII (H) hydride. We found that, in both the (DAB ) and (BOZO) ligands, the first scenario (direct oxidative addition) was more likely, with lower transition states. Direct oxidative addition of benzene by 3 or 4 is shown in Scheme A.2. The lowest barrier found for 3 is 22.2 kcal/mol at 298 K and 23.7 kcal/mol at 498 K. In contrast, the lowest barrier found for 4 is 18.5 kcal/mol at 298 K and 19.5 kcal/mol at 498 K. All of these values imply accessible benzene oxidative addition at the reaction temperatures investigated. The increased activity of 3 compared to 4 is most likely simply due to the increased temperature that H/D exchange with catalyst 1 can be run without decomposition. A comparison of direct oxidative addition and benzene coordination/deprotonation is shown in Scheme A.3 for complex 3 and Scheme A.4 for complex 4. Scheme A.2. Oxidative addition of benzene by Rh(DABF ) and Rh(BOZO) complexes 3 and 4. H/D exchange most likely proceeds through these transition states. All free energies in kcal/mol. Methane activation by complex 3, in a manner analogous to benzene activation, was also in- 185 Scheme A.3. The Rh(DABF ) complex 3 activates benzene with a barrier of 22.2 kcal/mol at 298 K and 23.7 kcal/mol at 498 K. Top row: direct oxidative addition of benzene by RhI . Elsewhere: internal oxidative addition of 3 followed by benzene coordination. All free energies in kcal/mol. 186 Scheme A.4. The Rh(BOZO) complex 4 activates benzene with a barrier of 18.5 kcal/mol at 298 K and 19.5 kcal/mol at 498 K. Bottom row, first and second TSs: Direct oxidative addition of benzene by RhI . Elsewhere: internal oxidative addition followed by benzene coordination. All free energies in kcal/mol. 187 vestigated, with the results shown in Scheme A.5. The Rh(Me) intermediates and transition states were approximately 7-19 kcal/mol higher in energy than their Rh(Ph) counterparts. In particular, the direct oxidative addition of methane has in its lowest transition state a barrier of 32.4 kcal/mol at 298 K and 37.6 kcal/mol at 498 K, which is prohibitively high under reaction conditions. This is consistent with the lack of experimentally observed methane activation by complex 1. Scheme A.5. Activation of methane by the Rh(DABF ) complex 3 in a manner analogous to the oxidative addition of benzene (see Scheme A.3). The lowest transition state found has a barrier of 32.4 kcal/mol at 298 K and 37.6 kcal/mol at 498 K, rendering methane activation inaccessible. All free energies in kcal/mol. A.5 Conclusions This work has demonstrated two efficient rhodium catalysts for H/D exchange between benzene and TFAD. To the best of our knowledge, catalyst 1 is the most active rhodium catalyst yet discovered for benzene H/D exchange. These results demonstrate TOFs similar to platinum and palladium systems. In addition, these rhodium complexes were studied under catalytically relevant conditions for industrial C−H activation and functionalization. 188 A.6 References [1] Kloek, S. M.; Heinekey, D. M.; Goldberg, K. I. Angew. Chem. Int. Ed., 2007, 46, 4736–4738. [2] Rhinehart, J. L.; Manbeck, K. A.; Buzak, S. K.; Lippa, G. M.; Brennessel, W. W.; Goldberg, K. I.; Jones, W. D. Organometallics 2012, 31, 1943–1952. [3] (a) Muehlhofer, M.; Strassner, T.; Hermann, W. A. Angew. Chem. Int. Ed. 2002, 41, 1745–1747; (b) Sen, A.; Benvenuto M. A.; Lin M. R.; Hutson A. C.; Basickes N. J. Am. Chem. Soc. 1994, 116, 998–1003. (c) Hickman, A. J.; Cismesia, M. A.; Sanford, M. S. Organometallics 2012, 31, 1761-1766. [4] Nordlander, J. E.; D. Kotian, K. D.; Raff, D. E. II; Njoroge, G. F.; Winemiller, J. J. J. Am. Chem. Soc. 1984, 106, 1427–1432.  189 Appendix B Energy levels of group 10 transition metal atoms and ions B.1 Abstract The energies of the group 10 transition metals (Ni, Pd, and Pt) in different configurations (d8 s2 , d9 s1 , and d10 ) are calculated using various DFT methods. Furthermore, energies of the corresponding 1+ and 2+ ions are also calculated. The results of the various DFT methods are compared with the experimental energies [1] and previous calculations [2], with the purpose of determining the best DFT method for each atom. Finally, inconsistent energies in supposedly degenerate orbitals of the ionic states are observed and a potential solution is proposed. 190 Table B.1. Relativistic vs. nonrelativistic Hartree-Fock excitation energies, taken from Martin and Hay [2]. Atom Nickel Palladium Platinum Config. Hartree Fock Relativistic HF Experimental 8 2 0.00 1.27 5.47 0.00 −3.01 −3.76 0.00 −3.28 −4.69 0.00 1.63 6.04 0.00 −2.09 −2.19 0.00 −0.40 0.50 0.00 −0.03 1.71 0.00 −2.43 −3.38 0.00 −0.64 −0.16 d s d9 s1 d10 d8 s2 d9 s1 d10 d8 s2 d9 s1 d10 Relative energies in eV. The experimental energies are averages over spin-orbit components [1]. B.2 Introduction Density function theory has found widespread application in computational chemistry as a fast alternative to ab initio calculations with correlation. However, the accuracy of DFT calculations is sometimes called into question in many transition metal-containing systems, due to the fact that transition metals often have several closely-spaced low-lying energy levels and that DFT is not always able to predict the correct ground state. In this report we present the results of calculations of the levels of Ni, Pd, and Pt with the following DFTs: B3LYP, M06, M06-2X, M06-HF, M06-L, PBE, PBE0, and XYGJ-OS (Ni only); and we contrast these with a few electron-correlation calculations, including HF, MP2, and CCSD, and with previous calculations done by Martin and Hay [2]. B.3 The neutral atom The group 10 transition metals are notable in that each atom in the column has a different ground state electron configuration, viz. d8 s2 for Ni, d10 for Pd, and d9 s1 for Pt. While Ni conforms to the Aufbau principle, Pd and Pt do not. This is due to a complex interplay of electron correlation and relativistic effects. The ordering of the ground and excited states of Pd and Pt have long been known (Table B.1), but the identity of the ground state of Ni has been somewhat more contentious [3]. There are a few observations to note in Table B.1. First, the experimentally determined ground state of Ni is d8 s2 3 F4 , but the energies in Table B.1 are averages over all J = 2, 3, 4, which in the case of Ni results in an energy that is slightly higher than the 3 P average. Secondly, we find that relativistic corrections, though unnecessary for Ni, become important for Pd and absolutely vital for Pt. Finally, we find that the Hartree-Fock calculations do a poor job of ordering the configurations 191 Table B.2. Relativistic vs. nonrelativistic Hartree-Fock excitation energies for the M+ ions [2]. Atom Config. Hartree Fock Relativistic HF Experimental Nickel 8 2 0.00 7.61 6.98 0.00 2.91 6.36 0.00 2.68 6.26 0.00 8.12 7.09 0.00 4.25 6.66 0.00 7.26 7.51 0.00 7.59 8.67 0.00 5.13 8.32 0.00 8.46 9.22 Ni+ Palladium Pd+ Platinum Pt+ d s d9 d8 s1 d8 s2 d9 d8 s1 d8 s2 d9 d8 s1 Relative energies in eV. Table B.3. Excitation energies of Pt atom as calculated by various DFTs. Config. d9 s1 d10 d8 s2 Exp. 11.07 14.76 HF −118.23 25.41 5.95 B3LYP −119.08 14.67 6.84 M06 −119.03 0.40 19.17 M06-2X −118.95 11.72 15.24 M06-HF −118.91 24.95 21.07 M06-L −119.08 14.01 0.86 PBE −119.07 14.39 −0.05 PBE0 −119.04 15.03 9.08 The d9 s1 row contains the absolute energies in hartrees. The other rows contain the energies relative to the d9 s1 ground state, in kcal/mol. Experimental energies derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 1 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. for the Pt atom. Hence, the first part of this report shall many focus on DFT calculations of the Pt atom, with additional data for Ni and Pd for purposes of comparison. We also examine the previously-calculated excitation energies for the singly-charged cations Ni+ , Pd+ , and Pt+ . These numbers are taken from the same reference [2] and are reproduced in Table B.2. Looking at the values in Table B.2, two things are apparent: First, the numbers for Ni+ , while close, are in the wrong order, putting the d8 s1 configuration below the d9 configuration in energy. And secondly, while the HF calculations for Pd+ and Pt+ give the correct ground state configuration, the relative energies are significantly off. This report will also give updated excitation energy calculations for the singly-charged cations. B.3.1 Platinum (0) Calculations with Pt were performed on Jaguar version 7.5 [4]. The basis set used was a variant of LACV3P**++, modified to include f functions [5]. The results are presented in Table B.3. 192 Table B.4. Excitation energies of Pt atom calculated without f functions. Config. d9 s1 d10 d8 s2 δδ (a) ππ (b) σδ (c) Exp. 11.07 14.76 14.76 HF −118.23 27.96 1.41 30.24 8.50 1.25 B3LYP −119.08 14.67 5.10 28.38 12.77 7.57 M06 −119.03 0.40 12.22 26.43 19.62 17.36 M06-2X −118.95 11.72 15.04 31.96 19.46 15.30 M06-HF −118.91 24.95 21.11 39.30 30.11 27.04 M06-L −119.08 14.01 6.99 32.84 16.22 10.68 PBE −119.07 14.43 0.02 29.52 15.22 10.50 PBE0 −119.04 15.03 8.24 30.94 15.26 10.03 9 1 The d s row contains the absolute energies in hartrees. The other rows contain the energies relative to the d9 s1 ground state, in kcal/mol. Experimental energies derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 1 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. (a) Energy of the d8 s2 configuration with singly occupied dxy and dx2 −y2 orbitals. (b) Energy of the d8 s2 configuration with singly occupied dxz and dyz orbitals. (c) Extrapolated energy of the ground d8 s2 configuration according to the formula Eσδ = 34 Eππ − 13 Eδδ . An examination of the data presented in Table B.3 shows that while most DFT methods are very good at predicting the d9 s1 − d10 separation, they are considerably poorer at correctly ordering the d8 s2 state, often putting it near the d9 s1 state instead of above the d10 state. An examination of the molecular orbitals in the d8 s2 output files in Molden revealed considerable mixing of the d and s orbitals in some of the cases. Hence, the true configuration, d9 s1 or d8 s2 , could not be ascertained, and hence the values could not be trusted. In order to prevent this catastrophic mixing of orbitals, the input files for the d8 s2 cases were rerun with the symmetry turned on (Table B.4, row 3). This prevented the basis set from using the added f functions. Fortunately, an examination of the rerun numbers for the d9 s1 and d10 cases (Table B.4, rows 1-2) showed that the f functions play a very minor role in the total energy and hence can be safely ignored. Unfortunately, examination of the molecular orbitals in the d8 s2 output files in Molden revealed that there was still some mixing between the filled s orbital and the empty dz2 orbital among the β electrons. This is due to both orbitals being totally symmetric in the point group used for the calculations (D4h or its largest Abelian subgroup D2h ). Hence, instead of calculating the lowest d8 s2 state directly, it may be more accurate to calculate higher d8 s2 states and extrapolate the ground d8 s2 (see Section B.6). The revised energies are given in Table B.4, rows 4-6. Electron-correlation calculations were also performed on the Pt atom, on Q-Chem version 4.0 using the MP2 and CCSD methods on the LACVP basis set with ECP. A guess wavefunction with the correct orbital population was first generated by DFT methods, and then the electron-correlation values were calculated with the orbital overlap maximized. The results, given in Table B.5, are in decent agreement with the experimental values, but are not an improvement over DFT methods. 193 Table B.5. Excitation energies of Pt atom calculated using electron-correlation methods. Config. d9 s1 d10 d8 s2 δδ (a) ππ (b) σδ (c) Exp. 11.07 14.76 14.76 MP2 −118.284 13.41 5.43 30.16 11.37 5.10 CCSD −118.289 16.83 0.92 12.41 3.29 0.24 9 1 The d s row contains the absolute energies in hartrees. The other rows contain the energies relative to the d9 s1 ground state, in kcal/mol. Experimental energies derived from Moore [1]. Color coding represents deviation from experimental value. Blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. (a) Energy of the d8 s2 configuration with singly occupied dxy and dx2 −y2 orbitals. (b) Energy of the d8 s2 configuration with singly occupied dxz and dyz orbitals. (c) Extrapolated energy of the ground d8 s2 configuration according to the formula Eσδ = 34 Eππ − 13 Eδδ . Unfortunately, neither method gave the correct ordering of states; both methods put the singlet state at the highest energy. Furthermore, CCSD performed more poorly than MP2. This observation is probably due to the fact that CCSD is designed for calculating ground state energies. We can draw several conclusions from the data presented for the Pt atom. First, DFT values are closer to the relativistic HF calculations than to the nonrelativistic calculations (see Table B.1). This is due to the effective core potential taking relativistic effects into account. Secondly, for the d9 s1 − d10 transition, most DFTs gave values that were a little high, while the HF calculation gave a value that was much too high. Thirdly, most DFTs ordered the relative energy levels of the different configurations in the same way as Martin and Hay’s relativistic HF values, differently from the experimental result. And finally, M06-2X gave the best values. B.3.2 Palladium (0) Calculations with Pd were performed on Jaguar version 7.5 using the basis set LACV3P**++ [4], and on Q-Chem version 4.0 using the basis set LACVP [6]. The results from Jaguar are presented in Table B.6; however, it was difficult to calculate the d8 s2 configuration properly due to the wide separation from d9 s1 and the aforementioned symmetry issues. Hence, all DFT values were recalculated in Q-Chem; the results are presented in Table B.7. The d10 − d9 s1 energy gap is very similar to the numbers in Jaguar; hence we can expect reasonably accurate values for the d8 s2 configuration as well. In addition, the use of Q-Chem allows us to take advantage of the electron-correlation MP2 and CCSD methods. It can be concluded from the tables for palladium that the best overall method was MP2, whereas 194 Table B.6. d10 − d9 s1 Excitation energies of Pd atom as calculated by Jaguar. Config. d10 d9 s1 Exp. 21.91 HF −125.87 1.97 B3LYP −126.72 20.79 M06 −126.72 38.15 M06-2X −126.61 27.53 M06-HF −126.52 10.89 M06-L −126.75 28.16 PBE −126.68 22.28 PBE0 −126.67 20.72 The d10 row contains the absolute energies in hartrees. The d9 s1 row contains the energies relative to the d10 ground state, in kcal/mol. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. Table B.7. Excitation energies of Pd atom as calculated by Q-Chem. Config. d10 d9 s1 d8 s2 Exp. 21.91 77.94 HF −125.87 1.72 55.17 MP2 −125.97 22.42 80.54 CCSD −125.96 15.95 67.28 B3LYP −126.71 19.98 84.79 M06 −126.71 38.63 114.45 M06-2X −126.61 26.07 97.61 M06-HF −126.51 10.09 89.50 M06-L −126.74 30.66 97.18 PBE −125.90 9.43 67.17 PBE0 −126.67 19.99 88.40 The d10 row contains the absolute energies in hartrees. The other rows contain the energies relative to the d10 ground state, in kcal/mol. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol. the best DFT calculation was B3LYP. B.3.3 Nickel (0) Unlike the cases for Pt and Pd, calculations with Ni were performed without any effective core potential, using the basis set G3LARGE with the program Q-Chem version 4.0. The results are summarized in Table B.8 for the electron-correlation methods and Table B.9 for the DFT methods. It can be concluded from the tables for nickel that the best overall method was CCSD, whereas Table B.8. Excitation energies of Ni atom as calculated by Q-Chem; electron-correlation methods. Config. d8 s2 d9 s1 d10 Exp. −0.69* 39.43 HF −1506.83 1.72 55.17 MP2 −1507.83 −30.92 −44.60 CCSD −1507.78 0.49 47.07 The d10 row contains the absolute energies in hartrees. The other rows contain the energies relative to the d10 ground state, in kcal/mol. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; red: within 25 kcal/mol. * See the discussion for Table B.1. 195 Table B.9. Excitation energies of Ni atom as calculated by Q-Chem; DFT methods. Config. d8 s2 d9 s1 d10 Exp. −0.69* 39.43 XYGJ-OS −1507.36 −5.66 19.57 B3LYP −1508.31 −9.11 22.65 M06 −1508.19 −10.33 19.43 M06-2X −1508.24 −3.59 48.36 M06-HF −1508.35 −14.01 49.11 M06-L −1508.03 5.26 50.00 PBE −1506.75 −12.30 29.37 PBE0 −1508.02 −9.18 85.06 The d10 row contains the absolute energies in hartrees. The other rows contain the energies relative to the d10 ground state, in kcal/mol. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol; red: within 25 kcal/mol. * See the discussion for Table B.1. Table B.10. Excitation energies of the group 10 atoms from d9 to d8 s1 , as calculated using various DFTs. + 9 Ni d d8 s1 ∆E Pd+ d9 d8 s1 ∆E Pt+ d9 d8 s1 ∆E Exp. 175.03 199.93 24.90 197.30 271.61 74.31 195.09 212.62 17.53 XYGJ-OS 390.57 392.65 2.08 N/A N/A N/A N/A N/A N/A B3LYP 173.62 201.51 27.89 198.16 287.29 89.14 217.38 242.80 25.42 M06 166.76 209.30 42.54 209.32 323.64 114.33 207.69 253.62 45.92 M06-2X 167.85 199.48 31.63 196.01 298.99 102.98 206.89 242.30 35.41 M06-HF 156.59 187.66 31.07 182.01 280.43 98.42 212.36 238.70 26.34 M06-L 75.48 105.56 30.08 196.53 297.30 100.77 203.26 237.12 33.86 PBE 164.53 183.01 18.48 177.85 250.10 72.25 208.54 233.96 25.42 PBE0 167.67 193.56 25.90 193.77 282.91 89.14 211.33 235.47 24.14 All numbers are given in kcal/mol relative to d8 s2 for Ni, d10 for Pd, and d9 s1 for Pt. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; yellow: within 15 kcal/mol; red: within 25 kcal/mol. XYGJ-OS not implemented for ECP; hence only Ni was calculated. the best DFT calculation was M06-2X. B.4 The singly-charged cation Whereas the neutral group 10 metals each have different ground state configurations, the singlycharged cations Ni+ , Pd+ , and Pt+ are uniformly d9 in the ground state. The excitation energy from d9 to d8 s1 for each metal has been calculated by various DFTs and electron-correlation methods. A summary of the results, along with the experimental values [1], is given in tables B.10 and B.11. All calculations were done on Q-Chem version 4.0. The G3LARGE basis set was used for Ni+ while the LACVP basis set with ECP was used for Pd+ and Pt+ . An examination of the data presented in tables B.10 and B.11 shows that CCSD is unequivocally the best method for Ni+ , while B3LYP is the best DFT method. M06-2X and PBE0 also gave good 196 Table B.11. Excitation energies of the group 10 atoms from d9 to d8 s1 , as calculated using various electroncorrelation methods. Ni+ d9 d8 s1 ∆E Pd+ d9 d8 s1 ∆E Pt+ d9 d8 s1 ∆E Exp. 175.03 199.93 24.90 197.30 271.61 74.31 195.09 212.62 17.53 HF 252.85 162.26 −90.59 147.93 217.87 69.94 178.87 188.39 9.52 MP2 140.27 192.18 51.91 176.65 283.37 106.72 188.91 205.44 16.54 CCSD 167.93 192.82 24.90 171.10 251.33 80.22 188.25 205.81 17.57 All numbers are given in kcal/mol relative to d8 s2 for Ni, d10 for Pd, and d9 s1 for Pt. Experimental energy derived from Moore [1]. Color coding represents deviation from experimental value. Gray: within 2 kcal/mol; blue: within 5 kcal/mol; green: within 10 kcal/mol; red: within 25 kcal/mol. results. As for Pt+ , both CCSD and MP2 gave excellent results. Unfortunately, there was no single method for Pd+ that gave good results for all energies: most methods were good at estimating the d9 state; only M06-HF and PBE0 came close to estimating the d8 s1 state; and HF, CCSD, and PBE were able to approximate the energy difference between the ion states well. Overall, however, these numbers are all nonetheless in better agreement with experimental data than the previous calculations as presented in Table B.2. B.5 Angular momentum-dependent artifacts in DFT calculations During the course of calculating these energies it became apparent that certain artifacts were being introduced in the calculations of the cations. These errors have been documented before [7] and the reason is innate to the nature of the exchange functional of DFT [8]. Potential solutions have been proposed [9] [10] but not yet widely implemented. B.5.1 Angular momentum of the electron/hole in the d1 /d9 configuration of the cation The various microstates of a d1 or d9 configuration give rise to only one permissible spin state: 2 D. Hence, it should not matter which real d orbital contains the electron or hole, respectively. However, we find that for DFT methods the calculated energy does vary: For example, according to B3LYP the 197 Table B.12. “Degenerate” energies of the Pt+ d9 ground state. d9 hole z2 xy x2 − y 2 xz yz LACV3P**++ Energy Erel −118.739 0.000 −118.741 −1.123 −118.741 −1.123 −118.737 1.141 −118.737 1.141 LACV3P Energy Erel −118.738 0.000 −118.740 −1.157 −118.740 −1.154 −118.736 1.079 −118.736 1.078 LAV1S Energy Erel −26.961 0.000 −26.960 0.791 −26.960 0.870 −26.958 1.591 −26.958 1.591 Calculations done in Jaguar. The “Energy” columns contain values in hartrees, whereas the “Erel ” columns contain values in kcal/mol. Table B.13. “Degenerate” energies of the Pt9+ d1 ground state. d1 z2 xy 2 x − y2 xz* yz* Energy −98.151343 −98.151819 −98.151822 −98.151822 −98.151822 Erel 0.000 −0.299 −0.301 −0.301 −0.301 M06-2X/LACV3P**++ done in Jaguar. The “Energy” column contains values in hartrees, whereas the “Erel ” column contains values in kcal/mol. * Jaguar could only run these calculations using Abelian (D2h ) symmetry. Therefore, when the electron was localized in a π symmetry orbital, the dx2 −y2 and dz2 orbitals mixed and thus the dxz or dyz orbital was reanalyzed into a δ symmetry orbital. Hence the energy is not expected to differ from the energy of the dxy or dx2 −y2 hole. energy of the ion Pt+ is different whether the hole is of σ symmetry (dz2 , L = 0), π symmetry (dxz or dyz , L = 1), or δ symmetry (dx2 −y2 or dxy , L = 2). The specific numbers are given in Table B.12. Similarly, the energy of Pt9+ as calculated by M06-2X depends (completely unphysically) on which orbital the lone d electron is placed (Table B.13). A more extensive study was done on the Ni+ ion, using Q-Chem version 4.0. The results of calculations using DFT methods are given in Table B.14 and the results of calculations using electroncorrelation methods are given in Table B.15. It is clear that, whereas the electron-correlation methods correctly predict the degeneracy of the five d holes in 2 D Ni+ , the DFT methods mostly do not. The only exception among the DFTs is XYGJ-OS, a double-hybrid DFT. 198 Table B.14. “Degenerate” energies of the Ni+ d9 ground state as calculated by DFT methods. d9 z2 xy 2 x − y2 xz* yz* B3LYP 0.000 0.293 0.289 0.290 0.290 M06 0.000 0.407 0.414 0.416 0.430 M06-2X 0.000 −0.589 −0.560 −0.583 −0.568 M06-HF 0.000 −1.681 −1.603 −1.675 −1.647 M06-L 0.000 −0.285 −0.329 −0.276 −0.287 PBE 0.000 −0.111 −0.109 −0.111 −0.109 PBE0 0.000 0.372 0.369 0.358 0.359 XYGJ-OS 0.000 0.000 −0.001 0.018 0.016 Energies are relative to the dz2 hole and in kcal/mol. * Because Q-Chem runs the calculations using Abelian (D2h ) symmetry, when the electron was localized in a π symmetry orbital, the dx2 −y2 and dz2 orbitals mixed and thus the dxz or dyz orbital was reanalyzed into a δ symmetry orbital. Hence the energy is not expected to differ from the energy of the dxy or dx2 −y2 hole. Table B.15. Degenerate energies of the Ni+ d9 ground state as calculated by electron-correlation methods. d9 z2 xy 2 x − y2 xz* yz* HF 0.000 −0.012 −0.008 0.000 −0.001 MP2 0.000 0.007 0.007 0.023 0.009 CCSD 0.000 −0.002 −0.002 −0.002 −0.002 Energies are relative to the dz2 hole and in kcal/mol. * Because Q-Chem runs the calculations using Abelian (D2h ) symmetry, when the electron was localized in a π symmetry orbital, the dx2 −y2 and dz2 orbitals mixed and thus the dxz or dyz orbital was reanalyzed into a δ symmetry orbital. Hence the energy is not expected to differ from the energy of the dxy or dx2 −y2 hole. 199 Table B.16. Energies of the Pt2+ d8 configuration, calculated by B3LYP/LACV3P**++. Hole type z , x2 − y 2 σδ xy, z 2 σδ xz, yz ππ xz, x2 − y 2 πδ yz, x2 − y 2 πδ xz, xy πδ yz, xy πδ 2 xz, z σπ σπ yz, z 2 2 2 xy, x − y δδ 2 Energy −118.0461 −118.0461 −118.0372 −118.0366 −118.0366 −118.0359 −118.0359 −118.0201 −118.0201 −118.0145 Erel 0.00 0.02 5.59 6.00 6.00 6.43 6.43 16.33 16.33 19.84 Energies are in hartrees; Erel values are in kcal/mol. B.5.2 Angular momentum of the holes in the d8 configuration of the dication When there are two holes, there are two triplet states: 3 F and 3 P . In terms of the real orbital holes, there are several intermediate energy levels: the σδ holes are at 3 F ; the ππ and πδ holes are at 13 23 33 13 43 43 5 F + 5 P , the σπ holes are at 5 F + 5 P , and the δδ holes are at 5 F + 5 P . These values are derived in the Section B.6. Table B.16 has the values for Pt2+ calculated in B3LYP with the LACV3P**++ basis set in Jaguar. While the numbers are close to the theoretically predicted values, the ππ and πδ holes are split into three different sets of closely spaced energy levels. As in the case of the d9 M+ ions, this is an artifact of the DFT calculation and is not physical. Similar calculations were also performed for Pd2+ . The results are shown in Table B.17. However, due to difficulties with Jaguar, Q-Chem was used instead. The reduction of the atom to an Abelian subgroup prevented the calculation of πδ and σπ holes, however. Fortunately, the calculated σδ, ππ, and δδ holes are in good agreement with experiment. The most extensive calculations were done on nickel. Values calculated with Jaguar are given in Table B.18 while values calculated with Q-Chem and experimental values are given in Table B.19. Again, note that it was not possible to get πδ and σπ hole energies with Q-Chem. B.6 Energies of d2 and d8 triplet states Given a d2 or d8 atomic configuration, we can easily derive that the various microstates give rise to the 3 F and 3 P triplet states and the 1 G, 1 D, and 1 S singlet states. Furthermore, we know from 200 Table B.17. Energies of the Pd2+ d8 configuration, calculated by B3LYP/LACVP. Hole type z 2 , x2 − y 2 σδ xy, z 2 σδ xz, yz ππ xz, x2 − y 2 πδ yz, x2 − y 2 πδ xz, xy πδ yz, xy πδ xz, z 2 σπ yz, z 2 σπ xy, x2 − y 2 δδ Energy −125.663 −125.663 −125.651 −125.663 −125.663 −125.651 −125.651 −125.663 −125.663 −125.619 Erel 0.00 0.01 7.45 0.01 (c) 0.01 (c) 7.45 (d) 7.45 (d) 0.01 (c) 0.01 (c) 27.93 Exp/J (a) 0.00 0.00 7.83 7.83 7.83 7.83 7.83 23.50 23.50 31.33 Exp/avg (b) 0.00 0.00 6.46 6.46 6.46 6.46 6.46 19.38 19.38 25.84 Energies are in hartrees; Erel and experimental values are in kcal/mol. Experimental values from [1]. (a) Experimental value using the lowest J energy. (b) Experimental value derived from the average of spin-orbit coupling. (c) Collapsed to σδ. (d) Collapsed to ππ. Table B.18. Energies of the Ni2+ d8 configuration, calculated on Jaguar using the 6-31G(tm)**++ basis. Hole type z , x2 − y 2 σδ xy, z 2 σδ ππ xz, yz 2 2 xz, x − y πδ yz, x2 − y 2 πδ xz, xy πδ yz, xy πδ xz, z 2 σπ yz, z 2 σπ 2 2 xy, x − y δδ 2 HF 0.00 0.00 11.29 11.62 11.62 11.62 11.62 34.34 34.34 45.42 B3LYP 0.00 0.05 10.25 9.80 9.80 10.42 10.42 28.33 28.33 38.44 M06 0.00 −0.27 10.35 7.50 7.50 10.49 10.49 18.12 18.12 23.84 M06-2X 0.00 −0.83 4.13 6.05 6.05 4.29 4.29 19.90 19.90 26.92 M06-HF 0.00 −0.53 −7.93 0.52 0.52 −7.70 −7.70 10.28 10.28 14.84 Energies are in kcal/mol. (a) Experimental value using the lowest J energy. (b) Experimental value derived from the average of spin-orbit coupling. (c) Collapsed to σδ. (d) Collapsed to ππ. M06-L 0.00 −0.42 16.22 13.17 13.17 16.38 16.38 35.85 35.85 48.36 PBE 0.00 0.08 10.23 9.39 9.39 10.38 10.38 26.60 26.60 36.66 PBE0 0.00 0.06 10.37 9.79 9.79 10.53 10.53 28.15 28.15 38.33 201 Table B.19. Energies of the Ni2+ d8 configuration, experimental and calculated on Q-Chem using the G3LARGE basis. Hole type z 2 , x2 − y 2 σδ σδ xy, z 2 xz, yz ππ xz, x2 − y 2 πδ yz, x2 − y 2 πδ πδ xz, xy yz, xy πδ xz, z 2 σπ yz, z 2 σπ 2 2 xy, x − y δδ Exp/J (a) 0.00 0.00 9.53 9.53 9.53 9.53 9.53 28.58 28.58 38.11 Exp/avg (b) 0.00 0.00 9.00 9.00 9.00 9.00 9.00 27.01 27.01 36.02 XYGJ-OS 0.00 −0.03 9.39 0.00 (c) 0.00 (c) 9.39 (d) 9.39 (d) 0.00 (c) 0.00 (c) 37.82 MP2 0.00 0.00 8.54 −0.01 (c) 0.00 (c) 8.54 (d) 8.54 (d) 35.08 (e) 0.00 (c) 35.08 CCSD 0.00 0.00 7.58 0.00 (c) 0.00 (c) 7.58 (d) 7.58 (d) 35.09 (e) 0.00 (c) 35.09 Energies are in kcal/mol. (a) Experimental value using the lowest J energy. (b) Experimental value derived from the average of spin-orbit coupling. (c) Collapsed to σδ. (d) Collapsed to ππ. (e) Collapsed to δδ. Hund’s rules that the 3 F state is the ground state. However, sometimes it is useful to know the energy given the two electrons or holes in specific, real orbitals (e.g., dz2 or dxy ). In this case we must calculate the energies of intermediate states, i.e. between that of 3 F and 3 P . B.6.1 Background The d orbital electrons, which have ` = 2, can be represented by either the complex spherical harmonics Y2m or the real spherical harmonics Z2m . Below is a list of the complex spherical harmonics: r 15 (x − iy)2 32π r 15 2 r Y21 = z(x − iy) 8π r 5 r2 Y20 = (2z 2 − x2 − y 2 ) 16π r 15 2 r Y21 = z(x + iy) 8π r 15 (x + iy)2 r2 Y22 = 32π 2 r Y22 = 202 And of the real spherical harmonics: r 15 xy 4π r 15 r2 Z21 = yz 4π r 5 r2 Z20 = (2z 2 − x2 − y 2 ) 16π r 15 2 r Z21 = xz 4π r 15 2 r2 Z22 = (x − y 2 ) 16π 2 r Z22 = It is easy to see that the complex and the real spherical harmonics are closely related: 1 Y22 = √ (Z22 − iZ22 ) 2 1 Y21 = √ (Z21 − iZ21 ) 2 1 Y21 = √ (Z21 + iZ21 ) 2 1 Y22 = √ (Z22 + iZ22 ) 2 Y20 = Z20 1 Z22 = √ (Y22 − Y22 ) 2i 1 Z21 = √ (Y21 − Y21 ) 2i 1 Z21 = √ (Y21 + Y21 ) 2 1 Z22 = √ (Y22 + Y22 ) 2 B.6.2 The d2 and d8 triplet states The triplet states are 3 F and 3 P . We shall now examine the complex spherical harmonic wavefunctions associated with the ML values of each state. To do so we will start with the highest possible ML value and successively apply the lowering operator L̂− : L̂− |`mi = p (` + m)(` − m + 1)|`m − 1i. 203 Also, we will use A to represent antisymmetric functions, e.g., A[U V ] = U V − V U . B.6.2.1 The 3 F state The 3 F state has ML values that range from −3 to 3. For ML = 3 the only possible configuration is A[Y22 Y21 ]. Similarly, for ML = 2 the only possible configuration is A[Y22 Y20 ]. For ML = 1, however, we note that L̂− can act on either Y22 or Y20 ; consequently we get a mix: A[2Y21 Y20 + which is normalized to r A[ √ 2 Y21 Y20 + 5 6Y22 Y21 ]; r 3 Y22 Y21 ]. 5 For ML = 0, applying L̂− once again yields r A[ 2 √ ( 6Y21 Y21 ) + 5 r 3 (2Y21 Y21 + 2Y22 Y22 )], 5 which simplifies and normalizes to 2 1 A[ √ Y21 Y21 + √ Y22 Y22 ]. 5 5 By symmetry arguments we can see that ML = −1 corresponds to r A[ 2 Y20 Y21 + 5 r 3 Y21 Y22 ]; 5 We also see that ML = −2 corresponds to A[Y20 Y22 ] and ML = −3 corresponds to A[Y22 Y22 ]. B.6.2.2 The 3 P state The 3 P state has ML values that range from −1 to 1. For ML = 1 the wavefunction must be orthogonal to the 3 F ML = 1 case; it is easy to see that the wavefunction r A[ 3 Y21 Y20 − 5 r 2 Y22 Y21 ] 5 fulfills this condition. Likewise, for ML = 0 it is easy to see that 1 2 A[ √ Y21 Y21 − √ Y22 Y22 ] 5 5 is orthogonal to the 3 F ML = 0 configuration. 204 By symmetry we can add that r A[ 3 Y20 Y21 − 5 r 2 Y21 Y22 ] 5 corresponds to the 3 P ML = −1 case. B.6.3 Pure and mixed complex orbital configurations In the previous section, we saw that the complex orbital configurations A[Y22 Y21 ], A[Y22 Y20 ], A[Y20 Y22 ], and A[Y21 Y22 ] are pure 3 F states, while the other combinations are mixed. In this section we shall deconvolute the mixed states. B.6.3.1 ML = 1 We have r 3 Ψ1F = A[ and 3 r Ψ1P = A[ 2 Y21 Y20 + 5 r 3 Y21 Y20 − 5 r Hence and r A[Y22 Y21 ] = B.6.3.2 2 Y22 Y21 ]; 5 2 3F Ψ + 5 1 r 3 3F Ψ − 5 1 r r A[Y21 Y20 ] = 3 Y22 Y21 ] 5 3 3P Ψ 5 1 2 3P Ψ . 5 1 ML = 0 We have 3 2 1 Ψ0F = A[ √ Y21 Y21 + √ Y22 Y22 ] 5 5 and 3 1 2 Ψ0P = A[ √ Y21 Y21 − √ Y22 Y22 ]; 5 5 Hence 2 3 1 3 A[Y21 Y21 ] = √ Ψ0F + √ Ψ0P 5 5 and 1 3 2 3 A[Y22 Y22 ] = √ Ψ0F − √ Ψ0P . 5 5 205 B.6.3.3 ML = −1 From symmetry with the ML = 1 case we have 2 3F Ψ + 5 −1 r 3 3F Ψ − 5 −1 r r A[Y20 Y21 ] = and r A[Y21 Y22 ] = B.6.4 3 3P Ψ 5 −1 2 3P Ψ . 5 −1 Real orbital configuration energies We are now ready to calculate the energies of the configuration of electrons or holes in real orbitals. We should first note that the dz2 orbital (Z20 ) has σ symmetry, the dxz and dyz orbitals (Z21 and Z21 ) have π symmetry, and the dx2 −y2 and dxy orbitals (Z22 and Z22 ) have δ symmetry. Also, let us define  = E(3 P ) − E(3 F ) as the excitation energy to 3 P . B.6.4.1 Energy of σπ and σπ Y20 σπ ≡ A[Z21 Z20 ] = A[ √ (Y21 + Y21 )] 2 1 (E(Y21 Y20 ) + E(Y21 Y20 )) 2 3 2 3 1 2 = [( E(3 F ) + E(3 P )) + ( E(3 F ) + E(3 P ))] 2 5 5 5 5 2 3 3 3 3 3 = E( F ) + E( P ) = E( F ) +  5 5 5 E(Z21 Z20 ) = A very similar calculation for the σπ case yields the same energy, 53  above the ground state. B.6.4.2 Energy of σδ and σδ Y20 σδ ≡ A[Z22 Z20 ] = A[ √ (Y22 + Y22 )] 2 E(Z22 Z20 ) = 1 1 (E(Y22 Y20 ) + E(Y22 Y20 )) = [E(3 F ) + E(3 F )] = E(3 F ) 2 2 A very similar calculation for the σδ case yields the same energy, i.e. the energy of the ground state. B.6.4.3 Energy of ππ ππ ≡ A[Z21 Z21 ] = A[ E(Z21 Z21 ) = 1 1 (Y21 + Y21 )(Y21 − Y21 )] = √ (Y21 Y21 − Y21 Y21 ) 2i 2i 1 4 1 1 (E(Y21 Y21 ) + E(Y21 Y21 )) = E(Y21 Y21 ) = E(3 F ) + E(3 P ) = E(3 F ) +  2 5 5 5 206 B.6.4.4 Energy of πδ, πδ, πδ, and πδ 1 1 πδ ≡ A[Z22 Z21 ] = A[ (Y22 + Y22 )(Y21 + Y21 )] = A[ (Y22 Y21 + Y22 Y21 + Y22 Y21 + Y22 Y21 )] 2 2 1 [E(Y22 Y21 ) + E(Y22 Y21 ) + E(Y22 Y21 ) + E(Y22 Y21 )] 4 1 3 2 3 2 = [E(3 F ) + ( E(3 F ) + E(3 P )) + ( E(3 F ) + E(3 P )) + E(3 F )] 4 5 5 5 5 1 3 1 4 3 = E( F ) + E( P ) = E(3 F ) +  5 5 5 E(Z22 Z21 ) = Very similar calculations for the πδ, πδ, and πδ cases yield the same energy, i.e. 51  above the ground state. B.6.4.5 Energy of δδ δδ ≡ A[Z22 Z22 ] = A[ E(Z22 Z22 ) = B.6.5 1 1 (Y22 + Y22 )(Y22 − Y22 )] = √ (Y22 Y22 − Y22 Y22 ) 2i 2i 1 1 4 4 (E(Y22 Y22 ) + E(Y22 Y22 )) = E(Y22 Y22 ) = E(3 F ) + E(3 P ) = E(3 F ) +  2 5 5 5 Summary A triplet d2 or d8 atomic configuration gives rise to a mix of 3 F and 3 P states. The σδ configurations are pure 3 F , the ππ and πδ configurations are one-fifth of the way towards 3 P , the σπ configurations are three-fifths of the way towards 3 P , and the δδ configuration is four-fifths of the way towards the 3 P energy level. Hence we see that the energy is the lowest when the electrons or holes are allowed to freely roam about the atom (the σδ case) and the highest when they are restricted to one plane (the δδ case). B.7 Conclusions For the neutral atoms, M06-2X gave the best DFT results for Pt and Ni, while B3LYP gave the best DFT results for Pd. This may be due to the contraction of the d functions of the basis set in Pd. Considering only electron-correlation methods, MP2 gave the best results for Pt and Pd, while CCSD gave the best results for Ni. This may be due to the use of an ECP on Pt and Pd. In comparing DFT versus electron-correlation methods, the electron-correlation method (CCSD or MP2) was better for Pd and Ni, whereas the DFT M06-2X was better than any electron-correlation method for Pt. For the singly-charged cation, B3LYP gave the best results for Pt and Ni whereas there was no method found that gave good results for Pd. The singly-charged cation in the d9 configuration should have a hole in one of five degenerate orbitals; this is not borne out by DFT calculations and the discrepancy is not physical. The use of 207 electron-correlation calculations removes this artifact. Similar issues are observed with the doublycharged cation in the d8 configuration. This problem is innate to DFT and its resolution is still an active area of research. 208 B.8 References [1] C. E. Moore, Atomic Energy Levels (Natl. Bur. Stand., Washington, D.C., 1952) Circular 467, Vols. 2 and 3, 1959. [2] R. L. Martin and P. Jeffrey Hay, J. Chem. Phys. 1981, 75, 4539–4545. [3] Eric R. Scerri, The periodic table: its story and its significance; Oxford University Press, 2007; pp. 239–240. [4] Jaguar, version 7.5; Schrödinger, LLC: New York, NY, 2007. [5] (a) Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. V. J. Comput. Chem. 1983, 4, 294–301; (b) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72, 650–654. [6] Q-Chem, version 4.0; Q-Chem, Inc: Pittsburgh, PA, 2010. [7] Frank Jensen, Introduction to Computational Chemistry, 2nd ed.; John Wiley and Sons, 2007; p. 260. [8] Wales, David J. Molecular Physics 2005, 103, 269–271. [9] Axel D. Becke, J. Chem. Phys. 2002, 117, 6935–6938. [10] Sergey N. Maximoff, Matthias Ernzerhof, and Gustavo E. Scuseria, J. Chem. Phys. 2004, 120, 2105–2109.