I. Molecular Structure Investigations. II. A Theory of Molecular Compounds Citation Stosick, Arthur James (1939) I. Molecular Structure Investigations. II. A Theory of Molecular Compounds. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/g401-yz91. https://resolver.caltech.edu/CaltechTHESIS:01032025-215224287 Abstract No abstract. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: (Chemistry) Degree Grantor: California Institute of Technology Division: Chemistry and Chemical Engineering Major Option: Chemistry Thesis Availability: Public (worldwide access) Research Advisor(s): Unknown, Unknown Thesis Committee: Unknown, Unknown Defense Date: 1 January 1939 Record Number: CaltechTHESIS:01032025-215224287 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:01032025-215224287 DOI: 10.7907/g401-yz91 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 16943 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 07 Jan 2025 22:34 Last Modified: 07 Jan 2025 22:38 Thesis Files PDF - Final Version See Usage Policy. 21MB Repository Staff Only: item control page

I. MOLECULAR STRUCTURE UJVE''. TIGATIONS II. A. THEORY OF MOtmUtAR COMPOUNDS Arthur James Stosick In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy California. Institut e of Technology Pasadena, California 1939 Table of Contents Page I. Molecular Structure Investigations: l. 2. The electron diffraction investigation of phosphorus trioxide, P4 06 ; phosphorus pentoxide, P4 010 ; arsenic trioxide, A.8 4 0 8 ; and hexamethylenetetra.mine, (CH 2 ) 6 N4 • l The electron diffraction investigation of tetra.nitromethane, C(N03 ) 4 • 10 3• The electron diffraction investigation of 4. phosphorus sulfoxide, P..,0 6 S4 • 19 The space group a.nd unit cell of the :potassium salt of nitroform, KO(N02 ) 3 • 26 5. The space group and unit cell of phenylprop iol ic acid. 6. 31 The space grO'li:p and unit cell of the molecular COl14)0und: naphthalene: p-dinitrobenzene. II. A Theory of organic Molecular Compounds. 41 I II. Summary 57 IV. Ref' erences 59 v. Propositions 60 Introduction In part I of this thesis there are presented the results of investi... gat ions of molecular structures by means of electron diffraction by gases, and by x-ray experiments with crystals. The electron diffraction studies were made on a group of related molecules of the fifth group elements. The molecules of this class are the trioxides of phosphorus a.nd arsenic, the pentoxide and sulfoxide of phosphorus, and hexamethylene tetra.mine. The observed interatomic dis- tances are discussed in terms of single-double and double-triple bend resonance, and the valence angles in terms of bond orbital hybridization. The remaining substance investigated by the electron diffraction method is tetranitromethane. In order to obtain a satisfactory inter- pretation of the data. it was necessary to ass'Ullle a rotatory osoillation of nitre groups about C-N bonds. In addition to this it was found necessary to include in the approximate intensity f'orill\lla the usually omitted temperature factor. The x-ra¥ inve~tiga.tions included the determination of the space group and unit cell for three crystals; these are the potassium salt ef nitroform, KC (N02 ) 3 , phenylpropiolic acid, C8 H5 C'=C-COOH, and the molecular compound between naphthalene and p-dinitrobenzene. In part II an attenpt has been made to provide a theory of molecular compounds. The class of complexes discussed is that for which the well- understood hydrogen bond or coordination type of bond cannot possibly be postulated. It is suggested that van der Waals forces are the source of complex stabilization. and arguments in support of this view are given. Tl,.ere follovl$ next a 'brief discussion of the color of the complexes with a. suggestion as to the origin of the color. Part II concludes with a summary of the available x-ra3 data for several crystalline complexes. Part I MOLECULAR STiUCTUQ lNVESf IGA.TlONS [Reprinted from the Journal of the American Chemical Society, 60, 1814 (1938).) [CONTRIBUTlON FROM THE GATES AND CRELLIN LABORATORIES OF CHEMISTRY, CALIFORNIA INSTITUTE OF TECHNOLOGY, No. 644] The Molecular Structure of Arsenious Oxide, As 0 Phosphorus Trioxide, P 0 Phosphorus Pentoxide, P 0 and Hexamethylenetetramine, (CH N by Electron Diffraction , 4 , 6 4 , 4 ) 10 2 BY G. C. HAMPSON AND A. In a recent paper, 1 which was published while this work was in progress, there were reported the results of an electron diffraction investigation of the structures of phosphorus trioxide, phosphorus pentoxide and arsenious oxide. The values which we have obtained for P406 and As406 are in good agreement with those of Maxwell, Hendricks and Deming. On the other hand, the latter authors were unable to deduce a structure for P40 1o and came to the conclusion that the molecule probably has lower symmetry than that of the point group T d· The reason for their failure probably lies in the fact that the molecule has an abnormally shqrt P-0 distance and we were led to our final structure, which gives an excellent fit with the photographs, only after very many models had been shown to be wrong. (1) Maxwell, Hendricks and Deming. J . Chem . Phys., 5, 626 {1937) . 6 , 6 4 J. STOSICK The visual method of measurement was used, the results being compared in the usual way with the approximate scattering formula I_ ~- - z.z. sin sr,; 1 - '' 1 Sri; in which r,j is the distance between the ith and jth atoms, Z,Zi their atomic numbers and s = (411" sin 0/2)/-,._, where 0 is the scattering angle artd A.the wave length of the electrons. Radial distribution curves 2 were also calculated and interatomic distances deduced from them. When a molecule contains several approximately equal distances, the radial distribution method fails to resolve the closely spaced maxima and very little information can be obtained from the curve. A modification of the method, suggested by Dr. V. Schomaker of these Laboratories, in wWch the estimated intensities _are multiplied by a factor ·,2) Pauling and Brockway, THIS JOURNAL, 57, 2684 (1935). 2.. Aug.,-1938 MOLECULAR STRUCTURE 'OF SOME ARSENIC, PHOSPHORUS AND NITROGEN COMPOUNDS 181.5 TABLE I Max. Min. 20 3 30 20 6 12 10 50 4 25 4 35 4 45 1 12 2 25 2 25 1.602 2.365 3.273 4.239 5.434 6.240 7..055 8.121 9 213 10.278 11. 261 12.353 13.292 14 .218 15.347 16.309 17.298 18.212 19 .246 20 .224 1 10 22.087 l 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 •(VIIJ ,t(Vlll) svu/so svn1/so 1.41 2,33 3.19 4.25 5.39 6.24 6.95 7.95 9.20 10.42 11.00 11.84 13.12 14 .33 15.37 16.15 17.05 18.13 19.17 20.08 1.45 2.34 3.20 4.26 5.37 6.23 6.94 7.99 9.18 10.35 10.90 11.89 13.11 14 .36 15 .41 16.18 17.09 18.17 19.22 20.03 (0. 881)) .986 .975 1.002 0.992 1.000 0.986 .979 999 1.014 0.977 .959 .987 1.007 1.002 0 .991 . 986 .996 .996 .993 (0.905) .989 .978 1 .005 21.84 21.89 0.988 .998 .; .984 .984 .996 1.007 0 .968 .963 .986 1.010 1.004 0.992 .988 .998 .999 .990 11 11 Mean .990 .9909 Mean .991 .9909 s 3e-a s', with a chosen such that the exponential share of this free phosphorus was removed by irfactor is equal to one-tenth for the last measured radiating the impure oxide for two days with a ring, gave a satisfactory resolution of most of the mercury vapor lamp. This treatment largely distances, and was invaluable in fixing the model . converts the yellow form into the much less volafor P4O10. • tile red form permitting separation by a subseAs4O6, P4O6, P4O10, and (CH2) 6N4 were measured quent distillation in vacuQ. The P 4O10 used was commercial c. P. phosphorus in Oxford using the apparatus described by de Laszlo ;3 for his assistance in operating the elec- pentoxide which was sublimed in a stream of oxytron diffraction camera we are greatly indebted to gen to remove lower oxides. This treatment is Dr. A. H. Gregg. The measurements on As4O6, necessary since the lower oxides are all more volaP4O6, and P4O10 were then repeated in Pasadena tile. The hexamethylenetetramine was a commercial using the apparatus described by Brockway, 4 and the two sets of results were found to agree to sample purified by vacuum sublimation. within 1%- With the latter apparatus two or three extra outer rings were obtained, but the longer jet-to-camera distance employed in the de Laszlo apparatus enabled some inner fine structure to be resolved which was of great help in the P4O10 investigation. Experimental The As4O6 used was the c. P. arsenious oxide of commerce which was not further treated. • The P4O6 was prepared by the method of Wolf 11nd Schmager, 5 a modification of the older method of Thorpe and Tatton. 6 The trioxide so prepared contains 1 to 2% of free yellow phosphorus even aftt!!" repeated vacu1;1111 distillation. The greater (3) De l.f!lzlo, Proc. Roy. Soc. (London), A146, 672 (1934). (4) Brockway, Rev. Modern Ph1·s., 8,231 (1936). (5) Wolf a,nd Schmager, Ber. , 62, 771 (1929). (fi) Thorpe and Tutton, J. Chem . Soc., 61, 54.5 (1890) . 0 Fig. 1.- Photograph of model o~ As,O,, P.06, or (CH2).N.. The black balls represent A~, P, or N ; the silver balls represent O ~r ·CH'i. 3 1816 Vol. 60 Arsenious Oxide.-The models tried for As 4O6 consist of four As atoms at the positions (v,v,v), (v,v,v), ·(v,v,v) and (v,v,v) and six O atoms at the positions ( ±u,0,0), (0, ±u,0) and (0,0, ±u) (Fig. I). The photographs of As4O6 show thirteen maxima, eleven of which were measurable. The first, second and fourth maxima are strong. Preceding the fifth maximum and following the sixth maximum there are deep minima of about equal depth. The fifth and sixth maxima are of equal intensity. The values of so for the maxima and minima and the estimated intensities of the maxima are given in 'ta!:ile I together with the weighted intensity values, c, used in the modified radi~ distribution method. The corresponding values of s for model VII and VIII are also given. VI is not satisfactory because the sixth maximum has become weaker than the fifth. TABLE II Model I II III V VI VII VIII u As-As V As-O O- As-O As- O- A, 2.310 1.155 3.26A. 2.ooA. 109°28 1 109°28' 2.057 1 .308 3.70 2.00 1.890 1.160 3.27 1. 79 1.890 1.142 3.23 l. 78 97°20' 130°14' 2.038 1.142 3.23 1.84 102°34 1 121 °58' 1.981 1.142 3.23 1.82 100°39' 125 °6 1 1 .959 1.142 3 .23 1.81 99°54 1 126 °20• Models VII and VIII lead to the following values of the interatomic distances and valence angles Model VII As- As = 3.20 A. As- 0 = 1.80 A. 0-As-O = 100°39 1 As-0- As = 125 °6 1 Model VIII As- As = 3.20 A. As-0 = 1.79 A. • 0 - As- O = 99 °54 1 As- 0-As = 126 °20 1 The radial distribution method applied to tlte As4O6 photographs gives a curve with maxima -at 1.82 and 3.21 A. corresponding to the values given above for the distances in the molecule. The curve for this calculation is given in Fig. 2. In Fig. 3 the theoretical intensity curves for As4O 6are 2 3 4 5 A. Inleratomic dista nce. Fig. 2.- Radial distribution curves. In Table II the parameters of all the models for whieh intensity curves were calculated are given. Models VII and VIII are both satisfactory since both are in qualitative agreement with the photographs. Model V ~ not satisfactory ~use for it the eighth. roaximum~m~~ ~,-Jl, and model 5 10 15 20 s. Fig. 3.-Calculated intensity cwves ior..4-f\9••· Aug., 1938 MOLECULAR STRUCTURE OF SoME ARSENIC, PHOSPHORUS AND NITROGEN COMPOUNDS 1817 .... TABLE III Ma:s. Min . I 10 8 1 2 5 20 1 6 2 16 2 18 16 19 90 4 4 5 5 6 6 7 7 1. 62 2.47 3.39 4 . 56 5 91 6 .30 7.02 8.37 9 . 68 10 . 71 11.19 1242 14 .05 15 . 62 16.84 17.58 18.44 19.63 2 3 3 1.60 2.49 3 .40 4 . 57 5.86 6.30 7 08 8.39 9.70 10 .72 11 .23 12.49 14.09 15 .69 16.95 17.70 18.52 19 . 71 12 2 2 s(VI) 1.803 2.659 3.581 4.591 5.623 6.641 7.542 8.514 9 .634 10 . 687 11 .820 12.863 14.319 15.548 1 1 So s(VII I) 8 8 1 9 9 2 sv1/so sv,u/s, (0.887) ( . 936) .949 . 995 (1042) 0.949 . 939 . 985 1 .007 1 .003 0.950 .971 . 984 1.009 (0. 899) ( . 929) .947 . 993 (1. 051) 0 . 949 .931 . 983 1 . 005 1.002 0.947 .966 . 981 1.005 Mean given. As final values of the distances the follow ing are given : As-As = 3.20 ± 0.03 A. As-O- As = 126 ± 3° As-O = 1.80 ± 0.02 A. 0-As-O = 100 ± 1.5° Maxwell, Hendricks and Deming 1 give As- As = 3.20 ± 0.05 A. They were unable to fix the oxygen parameter because of the much greater scatte.ring power of the As atoms. Fair agreement wa:s found with the oxygen valence angle As-O-As equal to 120°, 127.5° or 140°. Phosphorus Trioxide.-The models used for P,Os are of the same type as for As4O 6 . The four p atoms are at positions (v,v,v), (v,v,v), (v,v,v), and ( ,v,v), and the six O atoms at ( ±u,0,0), (0, ::1::12,0), and (0,0, ± u). The photographs of P 40 6 show nine maxima, eight of which are measurable. The first, second and fourth maxima are strong; the second maximum is followed by a weak, almost shelf-like maximum; the sixth and seventh maxima are sir~. of nearly equal intensity, separated by a broad and deep minimum; and the minimum following the weak eighth maximum is slightly deeper than that which precedes. The values of s 0 for the maxima and minima and the estimated intensities of the maxima are given in Table III. Values of the parameters of models for which theoretical intensity curves were calculated are listed in Table IV. . Models VI an<i Vlll<we,;e found to be in satisfactory qualitativo ~ • with the photographs. Model-VII is unsatisfaetQry becau&~ the minimum following the eighth maximum lra:s 0.990 .9802 Mean 0 . 9866 . 9769 become weaker than that which precedes. Also any model with the value of the parameter u greater than that of model VIII would make this same minimum too deep in comparison with the preceding one. TABLE IV Model u IV I V VII VI VIII III II 1.600 1.664 1. 750 1.780 1.800 1.820 1. 900 2 . 128 • 1.064 1.064 1.064 1.064 1.064 1 . 064 1.064 1.064 P- P P- 0 0 - P- O P--0- P 3.010 3.010 3.010 3.010 3 .010 3.010 3.010 3.010 1.597 1 . 620 1.654 1.666 1.675 1 . 684 1. 721 1.843 90 °11' 93°10' 96 °53' 98 °6 1 98°50' 99°41' 102°36' 109°28' 140 °48' 136°31' 130°59' 129°6' 127 °52' 126°39' 121 °53' 109°28' Models VI and VIII lead to the following values of the interatomic distances: Model VI P- P 2.95 P.. P- 0 = 1.64 A. 0 - P- O = 98 °50' P-O-:P = 127°52' Model VIII P- P 2.94 A. P- 0 1.65 A. 0 - P- O 99°41' P- 0-P 126°39' A final choice of model must lie between models VI and VIII, possibly favoring model VI. This leads to the final values of the distances: P- P = 2.95 ± 0.03 A. P- O = 1.65 ± 0.02 A. O-P- 0 = 99 ° ± 1 ° P- O- P = 127.5 ± 1 ° The results given by Maxwell, Hendricks· and Deming 1 are: • P- P ·= 3.00 % 0.05 k = " P- O "' 1.67 If.~ . ' P- 0 - P = 128.5 = 1.5° 5 G. C. HAMPSON AND A. J. STOSICK 1818 Radial distribution calculations result in a CUC-Ve (Fig. 2) with maxima corresponding to in- t~mic distances of 1.66 and 3.03 A. The tli&>i'etical intensity curves for P 4O6 are given in Fig; 4. Vol. 60 resolved. Following the eighth measured maximum there is another doublet of which only ,the outer ring is measi'trable. The second ring of this doublet is followed by a fairly pronounced minimum and a much stronger maximum. The last measured maximum is broad, and is preceded by a well marked minimum. The minima immediately preceding and following the weak seventh maximum are both weak, the outermost beb1g possibly a little deeper. The measured values of s0 of the maxima and minima and tbe estimated intensities of the maxima are listed in Table. V. In Table VI the models for which theoretical intensity curves were calculated are listed. TABLE V Max. Min. 1 I C so s (XIV) ax1v/so 10 2 2.823 3.965 5 .015 2.82 3.75 4.91 (0.999) ( . 946) .979 6.20 7.05 8 .50 .984 .977 1.019 11.00 11. 72 12.76 13.72 14.25 14.84 15.86 0.99~ 18 .46 19.24 20.23 21.38 23.00 .990 1 2 3 4 5 6 7 8 9 10 11 15 17 2 Absent 6.300 5 10 7.215 3 8.345 8 35 9.86 4 2 16 11.102 12.097 5 3 31 12.922 13.730 6 1 12 14.491 15.351 7 2 27 16.043 8 Not measurable 1 14 18.645 19.662 9 2 27 20.470 21. 525 10 1 12 22.943 Mean .009 .988 .999 .983 961· ~ .979 .988 .993 1.003 0.9873 TABLE VI Model 10 15 20 s. Fig. 4.-Calculated intensity curves for P,Oa. Phosphorus Pentoxide.-The photographs taken show eleven measurable maxima. The second maximum is the strongest and is followed by a shelf with no Qbservable minimum interposed; and the third- maximum is moderately strong am,l ~ followed by a broad shelf-like region. On p ~ p h s whkh ~ere taken µsjng a longer jet-to--caaiera distance (28.0 cm.) ~ she~. appeara as two very ea.k ~ a With poorly marked minima interposed. <;y/Jth ~ shorter camera distance (!C>.85 cm.) the doublet is not JI I IV V VI VII VIII IX X ·xr XII XIII XIV a b 1.064 1.820 2.011 0.910 1.820 1.926 .903 U!67 1.919 .923 1. 785 1.943 1.030 1.785 1.830 1.018 1.805 1.844 0 . 992 1.720 2.008 .993 1.840 1. 795 1.045 1.750 1.830 1.022 1. 795 1.829 1.040 1.795 1.829 1.02().. 1.800 1.836 1.018 · 1.i~ 1.827 P- P P- O p :._o• OPO 3.01 2 . 57 2.56 2.61 2.91 2.88 2.81 2.81 1.68 1.58 1.60 1.57 1.64 1.64 1.58 1.64 1.64 POP 1.64 99°41' 126°39' 1. 76 109°28' 109°28' 1. 76 111°12' 106°4' 1. 77 107°24' 112°50' 1.39 100°38' 125°24' 1.43 102°18' 122°40' 1. 76 100°36' 125~15 1.39 104° O' l't'r'46' 1.36 98° 0' 1211.,°l.4' 1.40 101 °30' 1~•42• 1 . 40 100°18' ~2lr~9' 1.41 101 °49' _l~~ I 1.40 101 °39'. 1~ All of t:1,,e. models.. are fo • four P ato]tlS at positidns (a,a,a), (a,a,a), (i~i}, and (a.iMl.mJC (0,±bd»J' ,md 0 a.ttJips at positions (~G. J#, {01 b), and the rema;,mtvftff·..O .a t~:>plS'itions (c,c,c), {;c~,e). ~. -,,. artd i; - ,o). Models XI, XIII and XIV are all satisfactory in 6 Aug.; 1938 MOLECULAR STRUCTURE OF SOME ARSENIC, PHOSPHORUS AND NITROGEN COMPOUNDS 18H) their- 'qualitative agreement with the photographs. Mol'l.el XIV is roughly the mean of models xt and XIII; and it has been used for the final calculati0ns of the interatomic distances of the molecule. The following distances result from these calculations: P- P = 2.84 ± 0 .03 ,A. P - O = 1.62 ± 0.02 A. P- O' = 1.39 ± 0.02 A. • The theoretical intensity curves for the models of Table VI are shown in Fig. 6. OPO = 101.5 ± 1 ° POP = 123.5 "= 1 ° OPO' = 116.5 ± 1 ° The angle designated by OPO' is the angle formed by the P atom at (a,a,a) and the O atoms at (b,0,0) and (c,c,c) (Fig. 5) . It may be seen from an inspection of the models listed that the valence angles are very closely limited. Model XII differs from model XI only in the phosphorus coordinates, causing a change in the P valence angle of only about one degree, but it is not in qualitative agreement with the photographs in that for it the shelf-like region following the fourth maximum has become a real maximum with a preceding minimum. The ,lfort P- O' distance is also fixed closely by the disagreement of model VII. This model dijie,rs from the satisfactory models only in that the coordinates of the outer four O atoms are changed by a very small amount. 5 10 15 20 s. Fig. 6.- Calculated intensity curves for P•O,o. Fig. 5. -Photograph of model of P .O10. The black balls represent P; t he silver balls represent 0. The modified radial distribution method suggested by Dr. Schomaker gave a remarkably good resolution of the various peaks, nearly all the interatomic distances in the molecule appearing as separate maxima. The curve is shown in Fig. 2. From the data whid:i it prov~ded, the three parameters could be determined within narrow limits, these parameters ~gr~eing well wit4,; ~~ chosen for the final model. Hexamethylenetetramine.-The photographs of this substance were taken using a long jet-tocamera distance and show six maxima. The first two are strong, the third broad and diffuse, the fourth very weak, the fifth fairly strong and the sixth very weak. The hexamethylenetetramine model is essentially the same as that for P4O 6 and As4O5. Four N atoms are in positions (v,v,v) , (v,v,v), (v,v,v) and (v,v,v), six C atoms in positions ( ±u,0,0), (0, ± u,0) and (0,0, ± u) and the twelve H atoms in positions (x,x,z), (x,x,z), etc. The hydrogen parameter could not be deduced from the photographs and an assumed value of 1.09 A. for the C-H distance was used in computing the theoretical intensity curves. A model in whicli all t\le angles had the regular tetrahedral valu~.91 109.5° and witi} a C-H distahce of 1,4.8 A. (~~Nor C- C = 2.42 A.) was found to agrtt well with·the photographs. The curve for )hi~ moddl'_1s. shown in Fig. 7. Table VII gives ttre~measured vatues of 7 'f 1820 G. C,. H11;~0N mo A. J ..51'0$.tCK . Vol. 60 valence angles ~ tetrahedral in this solid, the As".""all distance 3.28 ~- ·end th~ As-O distance 2.01 i.. There are two str~ng bopds for each 0 atom to, two As atoms in the same molecule, and two weak bonds to two As atoms of a neighboring molecule. This attraction between neighboring molecules apparently draws ~he O atom out, increasing the As-O distance and decreasing the oxygen valence angle from the values observed 15 10 5 for the vapor molecule to those observed in the s. Fig. 7.-Calculated intensity curve for (CH 2) 6N,. solid. To a smaller extent this applies to the A$ atoms too, since each forms three strong bonds t~ so for the maxima and minima together with the oxygens in the same molecule and three weak estimated intensities of the maxima; the s values bonds to oxygens of adjacent molecules. A re: for the above model are also listed, and the averdetermination of the crystal parameters was madc;l age value of s/s0 is seen to be practically unity. by Harker and Eskijian 12 in these Laboratorits TABLE VII confirming the older values but fixing them more Max. Min. so s/so I closely (to within about 0.03 A.). 0.988 2 2.966 2.93 1 10 Arsenic apparently has a tendency to form 4.261 4.06 .953 bonds at angles smaller than the tetrahedni.i"angle, 5.56 .992 2 15 5.606 8 7.303 7.15 .979 as As4Oa is not unique in this respect. In this 2 1.073 compound where the atoms form closed rings it 8.774 9.41 a 4 5 1.020 might be thought that the As valences are strained 10.75 10.54 3 2 11.47 11.40 0.995 2 ~ 12.49 12.35 0 .989 into taking up this angle because of the tendency 4 13 .68 1.012 of the oxygen angle to expand beyond the tetra4 10 13 .52 5 16 .25 6 4 15 . 79 1.029. hedral value 13 but even in compounds where the Mean 1.003 groups attached to the As atom form no other less than The radial distribution curve shown in Fig. 2 bonds, the As bond angle is considerably 14 16 109°28'. Examples are AsC!s 103°, • 101 ± gives C-N = 1.47 A. and C-C or N- N = 2.43 A. 16 AsB't"a 100 ± 2°, 16 As(CH ) 96 ± 5°. 17 4°, 3 3 agreeing well, with the above model. The remarkable constancy of this angle of Discussion of Resu~ around 100° already has been remarked tipon. 16 Of the substances reported in this paper, two, Steric effect and electrostatic repulsions are so hexamethylenetetramine and arsenious oxide, different in this series of compounds that they have·oe~n studied in the solid form. Hexamethyl- cannot be the deciding factors in fixing the angle. 18 enetetfamine crystallizes in a body-centered cubic According to the theory of directed valency the lattice, Dickinson and Raymond 7 reporting a regu- utilization of p orbitals alone leads tcr t-I.e formalar tetrahedral arrangement of valences with C- tion of bonds which are mutually perpendicular. N == 1.44 A. Gonell and Mark 8 gave C-N = Hybridization with the s orbital gives rise to 1.48 A. and C-C = 2.58 A., and this same type of stronger bonds which, if hybridization is comstructa~e has also been confirmed by Wyckoff and plete, as with carbon compounds, are at an angle Corey,9 whO' give C- N = 1.42 ± 0.08 A. The of 109°28'. If, however, there are unshared elecsum of the covalent single-bond radii 10 of carbon trons, as in the case we are considering, hybridizaand nitrogen is 1.47 A. in good agreement with the tion may not be complete, for there are two opposing tendencies. On the one hand, hybridizavalue which we find here. tion tends to stabilize the bonds but at the same Arsenolite, cubic As4O 6, was reported by Bo(12) Unpublished work. zorth 11 to be a lattice of As 4O6 molecules in a dia(13) S!W;on and Hampson, Trans. Faraday Soc., st, 945 (1935). mond-type arrangement. Both the As and the 0 . (14) Br'/flr;wa;v and Wall , THIS JOURNAL, 56, 2373 (1934) . (7) Dickinscm and Raymond, THIS JOURNAL, 45, 22 (1923). (8) Gooell aid. Mark, Z . physik. Chem ., 107, 181 (192::1) . . (9) Wyckoff and Corey, Z . .Krist., 89, 462, (1934~. ( fO) Pauling and Huggins, ibid., 81, 205 (tll34). (11)_.~rth, THIS JQpRNAL', 4,1 0 .l62J ('lf23), . (15) ~id.inf &bd Brockway, J. Chem . Phys ., 2, 867 (1934). (16) ~::IIP11P$Gll, Jenkins, Jones and Sutton, Trans. Faraday Soe. .,aa, 852 0937) , , _ U7) Springall add 'Brockway, 1'Ht!I JOURNAL, 60, 996 (1938). (18) Pauling, ibid. , 53, 1367 (1931). 8 Aug., 1938 MOLECULAR STRUCTURE OF SOME ARSENIC, PHOSPHORUS AND NITROGEN COMPOUNDS time this means that the unshared electrons also ten~ to become hybridized between the s and p orbitals, whereas the unshared pairs of electrons a,re ·aU(V'ays most stable when they occupy the s orbitals. As usual the actual configuration taken up is that which gives the lowest energy. The sum of the covalent single-bond radii 10 of arsenic and oxygen is 1.87 A. The decrease from the sum of the radii to the observed value of 1.80 A. may be explained as the result of double ·bond ~acter caused by unshared pairs from the oxygen atoms forming double bonds •with the As atoms. 14- 15 The value 126 ± 3° which is observed for the oxygen angle is close to 125°16', the angle between a double and a single bond on the regular tetrahedral model. Phosphorus, like arsenic, shows a tendency to form bond angles less than the tetrahedral angle. In P4Os if is found to be 99°. In other substances where the groups attached form no other bonds the same behavior is noted. Examples are PF3 99 ± 4°, 14 PCla 100 ± 2°, 14 PBr3 100 ± 2°, 16 Pia 98 ± 4°, 16 POCla 104°, 19 P(CH 3) 3 100 ± 4°.17 The sum of the covalent radii 10 for phosphorus and oxygen is 1. 76 A. · Again the observed value is lower presumably because of double-bond character, as with As4Os. In P4O10 the phosphorus angles are 101.5 and lHi.5°. 'Fhe latter angle is that between an outer oxygen atom and one within the P 4O6 "kernel." Addition of this outer oxygen atom has had very little effect on the other bond angles •and this might have been anticipated from the values given for PCla and POCia. The most startling feature of the P 4O10 molecule is the unusually short distance of 1.39 A. between each P atom and the "extra" 0 atom. The value found is about 79% of the sum of the single bond radii (1.76 A.) and is approximately that expected for a triple bond between the two atoms. A similar anomaly has been observed in thiophosphoryl chloride, PSCla, 20 where the P- S distance is found to be 1.94 A. instead of the normal single bond distance of 2.14 A. It is clear that double / 0- bond character in a structure such as O=P~ &=• I, is not sufficient to explain such a large shortening as is observed. The addition of four extra oxygen atoms to P40& has very little effect on the dimensions of this P 4O6 "kernel," and since we have ( l9) Brockway and Beach, THIS JoURN.AL, 60, 1836 (1938) . (20) Beach and Stevenson, J . Chem . Phys., 6, 75 (1938) . 1821 postulated single-double bond resonance in P 4O6 , it seems reasonable to conclude that it also occurs in P4O10, The important systems are probably II and III in resonance with I. III , , ',/V - There is another piece of evidence which favo~ this conclusion. In P4Os the contribution of a - /o+- double-bonded structure IV P°"' 0 - 0- was postu. lated to explain the shortening of the P-O bond below the single bond value. In this structure the phosphorus and oxygen atoms carry opposite charges and so there is no "formal charge effect" 21 influencing the bond length. On the other hand, in structures II and III of P4O 10 the positive charge on the oxygen or phosphorus atom is not compensated by a negative charge within the P 4O8 "kernel" and hence because of the increased effective nuclear charge we should expect the bond length to be diminished. The P- O distance in P4Os is 1.65 A. and the corresponding distance in P4O10 is indeed shorter, being 1.62 A. The difference may be due to experimental error, although in view of the above argument it may be of some significance. As regards the other P- O distance having the extremely low value of 1.39 A., the "formal charge effect" in structure II (negative charge on the oxygen, no charge on the phosphorus) would lead one to expect an increase rather than a decrease. One is forced to conclude that the predominant factor here must be the polar character of the bond. In discussirtg PSCia, Beach and Stevenson 20 ruled out any effect of ionic character and formal charges on the length of the bond, but concluded that the shortening of the P- S distance must be due to a con. - /4 c1+ s1derable (about one half) contribution of s-p-f'Cl °"'c1 without, however, giving any reasons. The short P- O distance in P4O10 is in accord with the chemical properties of this molecule. Such short bonds are presumably extremely stable, and the thermal stability and resistance to reduction of P 4O10 are well known. We wish to express our thanks to Professor Brockway, Professor Pauling and Professor Sidgwick for their help and interest in this work, (21) Elliott, THIS JOURNAL , 69, 1380 (1937) . 1822 G. C. HAMPSON AND A. J. STOSICK and the Commonwealth Fundfor a-Fellowship to one of us (G. C.H.). Summary Electron diffraction measurements on arsenious oxide, phosphorus trioxide and hexamethylenetetramine show that the molecules consist of four phosphorus or nitrogen atoms in positions (vvv) (vvv) (vvv) and (vvv) and six oxygen atoms or methylene groups in the positions (. ±uOO) (O±uO) and (OO±u). In phosphorus pentoxide there are four additional oxygen atoms in the positions (www) (www) (www) and (www). The interatomic distances and angles are: for As40e, As-0 = 1.80 ± 0.02 A., 0 - As- O = 100 Vol. 60 ~ i.5°, .As-O-As = 126 ± 3°, for P 40 6 , P- 0 .= 1~65 ·= 0.02 A:, 0 - P-0 = · 99 •:r;1; 1°, P-0-P = 12'1-.a ± 1 °, for N/CH2)d, <>N ,,;:: -;I.:47 =11> 0,02 A., C-N-C = N-C- N = 109.5°, for P40ro, ·P-O = 1.62 ± 0.02 A., P- 0' = 1.39 ± 0.02.A., 0--P-O = 101.5 ± 1°, P- 0 - P = 123.5 ± 1° andO'--P-0' = 116.5 ± 1°. The shortening of the bond distances in As 40 6, P406 and P4010 below the theoretical single-bond values is attributed to single-bond double-bond resonance. It is concluded that the abnormally low value of 1.39 A. for the P- 0' bond in P 40 10 is due to the polar character of the bond. PASADENA, CALIF. RECEIVED MAY 31, 1938 10 THE D:FEERMINA.TION OF TliE MOLECULAR STRUCTURE OF 'l'El.'RANITROJJ.m'HA.llTE BY THE EUX:TRON D!Fl'RAOTIO?l METHOD Because of the unusual properties of tetra.nitromethane and the con:flicU.ng views with regard to . its structure, it was felt that an electron diffraction investigation of it would be desirable. 'l.1he sa.rr!Ples of the compound used in this investigation came from two sources; one sample was purchased from Ka.hl baum, and the other sarrple was made acc0rding to the method of Chattaway. l Both samples were subjected to successive vacuum fractionations until the first, last. and middle :fractions had the same vapor pressure, in agreement with the values found by Menz 1es. 2 The electron diffract ion photographs· showed seven measurable rings. The measured values of s 0 = T47f sin 2O for the maxima. and minima. the estimated visual intensities, and the coefficients of the modified radial distribution function are listed in Table 1. s O , In the above expression for A is the de :Broglie wa.ve length of the electrons, and ~ c0 is the scattering angle. The coefficients ,,ere obtained from the estimated visual intensities 10 is the value of S 0 for the k 1 th ring, the value of "a." being chosen by multiplying the latter by 2 -ask2 8ke , where sk such that the exponential factor is equal to 0.1 for the last ring. In all of the assumed models :for which intensity curves were calculated, there a.re four nitro groups attached to the central carbon atom at tetrahedral a.ngles, O C-N = 1.46 A, N-0 i-Ei. L N3.Cl1 2 0 = 109°28 = 1.21 A, Md the 1 • LO-N-0 The assumed values of = 127° found by Brockway, Beach a.nd Pauling3 for nitrometha.ne. a.re those \\ Table I Max. Min. l l I Co g 4 10 24 3 4 5 1.76 (0.920) (O. 903) 1.000 1.033 5.83 7.02 3 4 5 6 3 12 3 15 4 25 7.Sl 1.027 1.001 9.43 1.oos 1.027 s.63 10.45 0.996 11.71 0.997 15 17.20 6 21.70 0.994 1.021 1.018 Mot accurately mea.swable 6 7 BVIxf so 2.79 4.50 2 2 s. 2 1 1 19.10 Mean 1.011 After assuming the previous dimensions and configuration of the nitro groups a.nd the disposition of the C-N bonds, there remains the possibility of twisting the nit:ro groups about their respective C-N bond directions. In all models the nitro groups were twisted in phase, i.e. in such a ma.oner ,9.S to keep the oxygen atoms of adjacent nitro groups as far apart as possible. all groups being given the same angle of twist 'f is a sketch of the molecule for 'f = 0° • • Figure I lHgure II shows the equilibrium configuration of the molecule which is finally accepted as the result of this investigation. Models I to III are static models, the nitro groups remaining at a fixed angle 'f In models IV and V the nitio gr0ups were assumed to be in rotatory oecill3.t1on a.bout the position a.n:q,litudes of 300 and 400, respectively. was assumed to be of the forro 'f = 0° with oscillaticjn The character of the oscillation 'P = 'f O sin 27(t. an obviously idealized ·; :formulation, which however is probably as good as any other simple "Fl& . .I 13 approximation that can be made. The range of oscillation was divided into three fl time equal II sub-intervals: 'f 0 /2 to + <f O • + - 'f O to ... 'f of 2. - f 0 / 2 to + 'f of 2, and In accordance with the assumed character of the oscillation the group spends equal times in each of these ranges. average values of 'f Time were calculated for each of these intervals; and intera.tomic distances were calculated for each range using the average f for the range. It is to be noted that the distances so calculated are not time average distances, but distances calculated. from a time aver~e angle. I In making the intei:1s1ty curve calculation by the approximate formula. sin s1 1 ~ = ~ ~ z. 1zj slij 111 • the coefficients for the <f dependent distances were divided equally rut0ng the three ranges of 4f • It became apparent that models I to V all sho,,ed mere maxima and minima in the region beyond s0 = 10 than the photographs did, and tha.t no obvious change in the above t;rpe of model would alteJ," this. It was suggested by Dr~ Schomaker of these laboratories that the usually discarded te~erature :fa.etor e-A. in the intensity fornro.la ought to be retained for moleo\lles with oh~ing distances. fhe intensity formula 4 including the temperature :fa.et or is : I in which A1 • 1 = =EZ i. J 1/ 2 (<S r ij ) 8 s 2 • and va.ria.t ion of the distance r ij . ( 6 r ij ) 2 is the mean square The ef:f ect of the temperature factor is to malce the widely varying distances increasingly less important as the scattering angle increases. In order to test this suggestion an intensity calcu1ation was made in which only the 'f invariant distances were included. If thermal .motion of the nitro groups 1s great. au.c h a calculation shculd show better agreement with the photographs in the region of large 1 0 values 14 since the temperatur e factor makes all important for lai·ge valnes of f dependent distances u.u- • The curve r~sulting from this " calculation ( Model VI) is s hown in Figure III. and is in excellent 8 qualitative ~reement with the photographs beyond s0 = 10. Since Model III showed the best qualitative agreement in the region out as :fa.r as s0 = 10 , a model was assumed in which the n itro groups o s cill,,t.ed with a;..r1 am::,litude of 20c about the pos ition of 30-0 twist; i.e. <f = of ·(or . . ) 3 30° + 20° sin 27ft. J.J For the calcul :<1.t ion of approximate values the oscill ation ra.nE·e ,,i;;>;,s divided into three time equivalent r :m g es e.s before. A time 9,ve:r age cp was calcu.1.q,ted for ea.ch range as b efore, a.nd inter at omic distances calcul?..t.ed. f r om these &.verae~ angle values. The deviations from the .s,..nd t h e meaii taken. <f = 30c distan ces were then $qua.red 'I'he appr oxinn tion involved in t h is calcu.h.'~ion of ;-r.e,..-.1. s quare devia tions is p::i.• obably no more obJ ectionr:1.ble th.1.."1 is the use of a terrperature factor derived for small displ'.:1.cements. intensity calcula tion the lij In rt1,-,::tking the values (interatomic clistances) used were t hose for the equ.ilib:dum configura tion of <f' = 30°, ,"tnd. the coefficients including the tem:per ':!.t'1re f actor were calcu.l .,;i.ted for e a ch integral value of s0 • No det ect able discontinuities in the curves 2.re caused by these chang es of coefficients. rrhe moclel just descr ibed appears as Model VU in Figure I I I• ln a simila.r manner Models VIII and IX were c a lcul ated for angles of t wist of 0° and 2:j 0 and oscillat ion a!f4)li tudes of 30° and 200 respective1y. Table II i s a list of t he various assumed reodeh. 15 l'= \ (r . :m. 5 10 s ... 15 16 Table II Model Type Amplitude I Static oo II Static 90° oo oo III Static 300 Qo IV Oscillating; no temperature f actor oo 300 V Oscill a.tingj no temperature factor Qo 40o VI Independent distances only VII Oscilla ting; temperature factor VIII Oscillating; tempera ture f actor IX Oscillating; terr.pe r ature f actor 20° = 1.46 A. N-0 = l. 21 A, 0 In all models C-N LO-N-0 'i'he curve for Model •!II best reproduces the observed data. with On the photographs the minimum follow- regard to qualitative features. i ng t h e fourth m9.Xinrum appeared slightly deeper tha.n the curve for Model VII indicates, but the over-all agreement is very satisfactory. A quantitative comparison was next made; the values of s for Model VII are, listed in Table I. ca1c /s 0 After making the small change of scale the values of the interatomic distances a.nd valence angles are: N-0 0 = 1.22 .± 0.02 A Discussion: 0 The two oxyg en a.toms bonded to the sa,me nitrogen a tom are 2.18 A apart. whereas the closest approach of oxyg en atoms of different nitro 0 Q groups varies from 2. 48 A to 2. 73 A. a s the angle of twist varies, the 0 equilibrhllll distance being 2. 57 A. 0 If it is reme1;:-:b ered that 2. 5 A \ '7 is approximately the distance of closest approach of oJcygen a.toms not bonded to the same third a.tom as is indicated. by numerous crystal structure investigations the reason for the nitro groups eacilla.ting rather closely in phase about the apparent equilibri'Ulll position is evident. Mother model proposed by Mark and Noethling5 on the basis of admittedly incomplete x-r03 data. had three nitro groups and a linear nitrite group, the molecule exhibiting point group symmetry ~V, the three ... fold ax.is coinciding with the linear a.xis of the nitrite group. The molecule which this investigation indicates as the correct one belongs to the tetx-agonal point group S4 • It is possible that the S4, molecule is in a condition of almost free rotation in the orystal lattice at temperatures close to the melting point, thereby asswning a statiet ica.l syrr.metry which will meet the synmetry requirements of the spa-ce gro"u"P determined by Mark and Iroethling. Moreover. a. linear nitrite group is very unlikely in the light o:f other structure determinations of nitrites, all of which indicate non-linear groups. Another piece of evidence in favor of the symmetrical strueture proposed in this pa.per is the zero dipole moment ef' the molec'Ule. 6 and of Weissberger and Sangewald7 The investigations of Williams, both indicate a zero moment to within the a.ocu:ra.cy of the experimental measurement. The rr~dified radial distribution method gave very little information, 0 The only peak which was of much value was that at 1.21 A. w"n.ich indicated the N-0 distance quite accurately. The c-urve is shown in Figure IV. 'l1'he heaV'J vertical lines indicate 'P -independent distances. the thinner 1 in es indicate <f -dependent distances. The 1 engths of the 1 ines 18 indicate their relative importa...rice at s 0 = O, the importance of the Cf-dependent distances falling off rapidly with increasing s0 • In all cases the position is that of the equilibrium position for Model VII. The observed C-N distance is just the sum of the covalent radii of carbon and nitrogen atoms, that is, 0 Q 1.47 A· The value 1.22 A found for the N-0 distance is considerably less than the sum of the single 0 bond radii for nitrogen and oxygen mich is equal to 1. 36 .A.. 0 This O value found in this invest ig,'9.tion is within O. 01 A of 1. 23 A, the value expected for 5of double bond character for the bond on the basis of resonance between the structures I and II below. There is apparently no pronounced additional shortening of the U-0 bond caused by the form-::l.l change of. the nitrogen a.tom. 0 -t-/ -N ~o :r.. Su.mrr;;,1.ry: p,_ structure for tetranitrometha:ne is proposed which is in agreement with electron diffr action data. In the model there are fou.r nitro groups attached to the carbon atom a t tetrahedral angles, the 0-N distance is o a 1. l..1.7 + O. 02 A., the l{-0 dista..'1ce is 1. 22 .± O. Q2 A, and. the angle O-N-0 in the nitro group is 127°. In order to obtain satisf actory intensity agree- ment :l.t wa,s necessary to assume the n1 tro groups to be undergoing rotatory oscilla tion 9.bout the C-U bonds in such a. Wa;J as to keep the oxygen-oxygen repulsions essentially minirrd.zed, and to incl'O.de the usually omitted temperature factor in the approximate intensity formula.. Acknowledgment The ,':t\u thor wishes to express his appreciation to Dr. Buchman for sn;ggestin.g t h e problem, to Dr. Schomaker for SU€1£8Sting the temperature factor treatment used, and to Professor Pauling for helpful diSC'U$Siona. \9 THE ELECTRON DIF.FllA.OTION INVESTIGATION OF Phosphorus sulfoxide, P 4 0 8 S4 , wa s first reported by Thorpe and g '.l..1u:tton in a paper giving the method of prepaJ·ation of the col'J'4}ound and the results of vapor density measurements. Its mode of formation by direct reaction of P4 0 6 and the stoichiometric equivalent of free sv.lfur (a reaction paralleling the reaction of P4 0 6 and 0~) suggests used in this investigation wa s prepared by the method. referred to above, and was purified by a ya.cu-um distillation, crystallization from OS 2 • and '.'l. second vacuum d.istill9.tion~ Electron diffra ct ion photo.g raphs of the com.pound were made in the usual ws;;, and were found to be excellent, showing thirteen measurable ringso The measured values of minima (in which C so = 4 1r sin jz A 2 for the ma,.'tirna and A is the de Broglie wave length of the electrons (about o. o6 A.). and G the scattering angle). the visual intensities I,,, and the coefficients are listed in Table I. c0 of the modified. adial distribution function These coefficients are the visually estim"ted i ntensities multiplied by k •th ring. 1md 11 a 11 is chosen such that l ,;;.st ring . e -ae 2 is equal to 0.1 for the The resnl ting numbers are norm.3.,lized to one hundred. for ease of computation. The radial distribution curve is presented in Figure I, the vert iaal lines indica.t ing the distances found in the final model and. their relative importance. 0 C The large peaks a t l. 61 A. &nd 2. 87 A were ir!J!J:lediately recognized as due to the bonded P-0 distance and the nonbond.ed B-P d.ist,ance resp ectively, since these are within a few hvndredths of an Angstrom of' the va lues found in a previous investigation of P•oH?. I • 2 3 10 5 - 4A 15 Fl& . JI: 5 20 6 25 Z.1 Table I Min. Max. lo co s. 2.43 (O. 827) 10 l 15 (0.929) (O. 948) 4 3.10 3.83 1 l 2 2 3 3 5 5 7 7 g g 9 9 5 4 10 14 11 13 13 0.994 0.986 S.87 11.26 5 16 13.14 13.87 2 7 20 3 14 4 19 5 24 1 ? 14 14 7.09 9 12 12 0.960 8.01 4 10 11 6.45 0.991 0.979 0.995 10.21 :;i:· 10 0.974 Shelf; not measured 6 6 4.64 Absent 4 4 vso ~ c 15 8 12.28 0.998 o.9s7 0.934 14.1+9 15.12 15.s1 0.977 0.979 0.992 16.55 17.26 0.995 o.9s9 17.90 18.43 19.24 20.49 .2 1. 73 0.987 0,989 0.986 0.991 1.002 o.9ss 22.68 23.57 24.46 0.977 o.9s4 0.999 25.77 Mean 0.9867 LJk ing use of the information obt 8.ined from the radial distribution curve, a. series of models were calculated using the values o:f' the P and 0 pax ,_u neters for the P4 08 S parameter. 11 kerne1 11 as found in ? 4 0 10 and varying only the These models have four :P atoms a.t (a.a,a), (a,i,a.L .(a.,a,a), and (a.a..a); six O atoms at (,±b,0,0), (O,_±b,O),and (0,0,,±-b); and f~r S at oms at (c,c.c), {c,c!_c), (c,e.c). and (c.c,c). In P 4 0 6 S4, the four S -8.toms are substituted for fcror O a toms in similar :positions in 1?4 0 10 , the molecule exhibiting SJC.mmetry Td. A list of' the mod.els for which intensity calcub.tions were made is presented in Table II. In making Table II Model a b C P-S l'-0 P-P I II III 1.018 1.792 1.792 1.792 1. 792 1. 792 2.040 1.77 1.97 1.93 1.63 2.88 2.88 l.018 1.015 1.013 1.018 IV V 2.155 2.132 2.115 2.09g In all models L O-l?-0 1.63 1.63 1 .. 63 1.90 1.s7 2.ss 2.88 2.85 1.63 = 101. 5°, L P-0-P = 123. ~o • the in tensity c alculations the app:roxima.te intensity for nru.l a l = • wa.s used. The e1:u·ves r esulting from these calcul ations are shovm in .i:' i gure II, the vertical arrows indicating t h e measured. posit ions of the maxime, and minima. c11~u-acteris,t ic fea t'IJ,res of t he photographs a.re: a weak shelf fo!lowing the second ma.xinromt a marked ·broadness and a.symmetry of the third, fifth, eleventh, and thirteenth maxima., a. relatively closely s pac ed 11 triplet 11 betweel'.1 s0 = 15 and 19, with the middle ring slightly weaker than the outer two; and a. very weak but measurable t welfth ring. It may be seen from the curves t hat model V best re- produces t hes e featu.res. The quantitative comparison of sea.le. and. was nBde for model V, the values of sc/s 0 being listed in Table I. Af ter chang ing the scale o:t' the mod.el as indica ted by the s/s 0 ratio the following distances ~e found for i?4,0sS 4 : P-0 0 :::: 1.61 + 0.02 l\. 0 P-P = 2.s5 ..::t 0.03 A 0 P-S = 1-8:i .::t 0.02 A L O-l'-0 = 101.50 + l"" L P-0-P = 123.50 ..t 10 L 0-P-S = 116. 5u + 10 s0 Discussion: 0 0 The bonded distances P-0 and P-S at 1. 61 A and 1. $5 A are both less than the respective . covalent single bond values 0 10 0 of 1.76 A and 0 2.14 A. The shortening of the P-0 distance to 1.61 A is to be ascribed to partial double bond character of the bond. Since the octet rule need not rigorously apply to elements be:-.rond the first row of t he periodic t able. t he following structures are pernu.ss ibl e. It m~ be seen that none of them is in disagreement with the ,9..djacent charg e rule 11 • Since the P-S dist ance is considerable l ess than the double bond value + + -0~ -o~ - - 0 -:P-S -0-P=S -0- -0/ !V It -o'-.. + -O-P::.S -0- ~o "+ - 0 -P=S -0- v VI - - ~ "'++ 0 P=S -0 VII 0 of 1.95 I\., .etnd sinc e the P-0 distance is also considerably shorter than 0 the s i X)l; le bond v~lue of 1.73 A 10 • it is probable that structure 111 rep:resen ts to the first approximation the bonding arra ngements, with consider able contributions of structures such as IV and v. Strv.ctures such as I and II :probably do n ot contribute appreciably since these would. tend to increase the P-S distance to the single bond value., It ~1ould a:ppe;?tr that structure V must contribute to the r esonance system if the shortening of the :P-S bond below the do't1.ble bond value is to be simple expb.in ed. Structures 'Tl and VII coul d a.ls.o result in a shortening of t he P-S distance by a formr::.l change effect. A. comp a.rison of' the P-::::0 and l)=S distances in vari ous other phosphorus co~ounds is instructive. Compound P=O Compound P=S P4010 1.399 P 4 0 8 S4 1.85 POC1 3 1.5gl2 PSC1 6 POFC1 3 1.5412 POF 2 01 1.5512 POJ!i 3 1. ~612 1. 9413 It is evident from the above table that the P=O and P=S bond type in P 4 0 10 and P4 0 6 S4 respectively is different from that in the phosphoryl and thiophosphoryl halides, ,' :lllcl the inclusion of structure V m.~ be cons iderad .justifiable. The O-:P-0 vale-nee angle of 101. , 0 is close to the value of the phosphorus ?Jlgle in several other compounds• the d.ecraase below the tetrahedral value of 109°28 1 being explainable by a greater degree of orbital character than in normal p ep3 tetrahedral bonds. The change of the O-P-0 angle from 99° in P,.,o,l to 101.5° in P4 0 8 S4 a."ld P 4 0io indiea t es that increasing the coordin'lt ion of the phosphorus atom from three to four tends to make the valence angles become more like those of the tetrahedral P04 ---. Sumrnaryt The structure of P 4 0 6 S4 has been determined by electron diffr.f!l.etion in the g,,1.s phase. The molecule consists of four P atoms at (a,a.,a ), {a,a,a), (a,a, a ), and (a,a,a); six O a.toms at (±b,O,O). (O,.::tb,0), and (0,0,_±b); and four S atoms at (c,c,c), (c,c;c), (c:c,c), and (c,c,c). The valence :mgles and the principa.l interatomic distances are,: P-0 :: /' 0 l.ol .:!: 0.02 A, P-S 0 = 2. 35 .! O. 03 A, Single-dou.ble bond a.nd double-triple bond resonance is considered to 2.S be t h e cause of t h e s hortening of the P-0 and P-$ bond distances below t h e single and dou.ble bond values respectively. Acknowledgment The author wishes to ac-.k:nowledge t h e helpful criticisms of Professor Linus Pauling. THE SPACE GROtP AND UNIT CELL OF THE POTASSIUM SALT OF NITROFORM Potassium aci-nitroform, ICC(N0 2 ) 3 • was prepared from tetranitromethane and potassium ferrocyanide by the method of Chattawa.y and Harrison 14 • The yellow crystals obt a ined from water solutions are tetragonal bipyramids showing four prism faces. Laue photographs taken with the x-ray beam parallel to ~ and .!i indicated that the crystals belong to the D4h Laue class. Rotation and oscillation photographs were taken with copper Ka radiation (filter ed thro'tlgh a 27)-(. (100) , and nickel filter) with [no] vertical and normal to the x-ray beam. [oo]J , Layer line measurements on these photographs furnished a.pproxim..:i,te lengths for the axes. These values were refined by more precise measurements of equatorial reflections on the various photographs. The following unit cell was thereby determined: z :::: g • rhe density of the crystals measured by flotation in a. mixture of 1 methylene iodide and benzene was found to be 2.23 .± 0.01. This density requires that there be 8. 01 niole~ules in the unit found above. -.• The following reflection forms appeared on rotation, oscillation, or Lane photographs only under the following conditions hkt h + k + £, Ok.t k + I, = 2n = 2n hkO h +k = 2n t = 2n (b) 2h + I, = 4n. hht (a) 21 LD.ue photograph of KC(N0 3 ) 3 t 1cen with the x-r~ beam parallel to ~ and with ~ 1 vert ioal. Flt.or azure :tn ... tensifying screens were use-l. Z8 For the.form hht h + k + t = 2n • the condition that t = 2n is included in The greater restrictions of condition (b) appear to exist inasmuch as no refleXions disobeying this rule were found, al though m1merous planes not obeying it were in position to reflect on one or more o:f the photographs. :Probable space groups t h en are 13 Dad ... I 4 2d 21 and C4v - I 4 md. In order to distinguish between these space groups the crystal was tested for pyroelectricity with none (or at best very little) being observed. The methods used to test for pyroelectriei ty were: (1) suspending a crystal in liquid air in a silvered Dewar flask to test for image forces; (2) bringing a suspended crystal just cooled in liquid air nl!lar a massive metal ob.1ect to test for image forces; (3) attempting to observe ice "fibers" forming on a fi-eshly cooled crystal suspended in air on a silk fiber. From the smallness {or absence) of pyroelectricity it was concluded the space group is probably 12 Dad - I 4 2d. Further experiments with the compound were discontinued since the crystals exhibited the undesirable property of exploding after several h ours in the x-ray beam. It was hoped. when this investigation was begun1 that some information about the rescna.nce structures of the nitroform ion could be obtained. It is possible that this information ha.s been obtained without a complete structure (letermina tion. An examina tion cf the point symmetry of the various a vailable · posit ions in which to place the eigh\ e(NOa ) 3 - ions shows that the only sat isf' clCtory positions a.re the eight:-!ol~. posi:; ions for which the point symmetry is C~ .. 2. The four-fold positions h t-i.ve the point symmetry S4 - 'ij. and these can be dismissed at once slnce the ion could not possibly exhibit this symmetry except by being in a state vertic J.. Filtered copper r' d1a ion. vr..ms-ra r di s : O cillation ph tograph of KC( 02 ) 3 • Filtered copper radiation. ~1 ;,.,•O cm. vertio 1. cam~r r diu : ;.o cm. 30 of rotation. There appears to be no evidence :f'or rotation inasmeh as tb.e x-ray photographs do not show an abnormally 1.art?:e "temperature factor" decline in intensities which ion ret a tion would cause. It follows therefore that the ion must possess the symmetry C2 - 2, that is that the skeleton oi carbon a.nd the three nitrogen atoms are coplanar. Since this is just what would be expected from resonance of the structures: JI_ L with resonance of the in each of the single bonded ni tro groups. we can conclude that such resonance does indeed occur. The ion is probably not completely planar, since this would require oxygen atoms of adjacent nitro groups to be closer than the van der Waals radii would permit. The two forces, the repulsion of oxygen atoms, which tend$ to twist the C-N bonds a.nd destroy ooplana.rity. and the stabilizing effect of resonance, which requires coplana.rity; are in competition. !he effe.c t of resenance has presumably made the CN 3 skeleton planar. This investigation indicates that $1mila.r resonance may take place in other a.symnetric nitro ions• probably to a lesser extent. This accounts for the observed racemization of optically active nitro ions. 31 THE SPACE GROUP A.MD UNIT CELL OF PHF.NYLPROPIOLIC ACID The phenylpropiolic acid used in this study was crystallized from water in the form of long needle-like prisms showing in general only one prism form. A cursory optical examina tion of the crystal$ showed that the crystals were birefringent with parallel extinction. A goniometric examination gave the following angles between normals to the prism faces; 67°4• • 112°49 1 , 67°13•, and 112°54•. 'l'llese data place the crystal in either the orthorhombic or the monoclinic system. Atteirpts to take Laue photographs were unsuccessful because of the smallness of the crystals available and because of the hardness of the A series of' oscillation and rot ation photographs were taken with copper Ka radiation (filtered through a 27 A nickel filter). In taking these photographs the crystal was mounted in the following ways: (1) the needle a.xis vertical. (2) the needle ax.is horizcn.tal and the biBect:rix of the a.cute prism. angle vertical, and (3) the needle axis horizontal and the bisectrix of the obtuse prism angle vertical. La¥er line meas".lrements established a:n orthorhombic unit cell, 'f,nich w3s refined by measurement of the equatorial reflections on several photographs. The unit cell so determined ist - 0 a1 = 9.61 + 0-01 A 8.3 = 15.os + 0.01 A ~ : z = 4. Q 0 5.12 :t 0.01 A (needle axis) For the det er ;,,ination of the space group a. series of Weissenberg equi-inclination photographs were made. The reflections observed 3.c. Rot tion photogr~.ph of phsnylpropiolic acid taken . 1th filtered copper radiation in a ca er 'the a.xi o ,.o c~. radius. ~ 1 1s vertical {needle hor zonto..l). oscillation photofra.ph of p .er.ylpl·opiolic aci ·11th filtered copper radiation in e. c ra i a. ~he t ~en era o:f 5.0 cm. xis ~n is vertical (needle horizontal). 33 Oscill t, i o 1 !ilt ... red co· 'l'h "'>hotoc • • .3>h of 1, h e n y l rcpiolic qc d t"' :en err diation ir1 n c rern. of 5. 0 cu. r a.xis ,1ee with is. e !' i ) is ver ical. • eis ,en'b rg }':lhoitogr '.1-"h of the hkO refl ct ions of ph eylpropiolic acid ta.~en with unfiltered copper radiation in a camera of 2. 36 c r -i.dius . eissenbere equi-inclin tion photograph of the hkl (£ = l) .reflection of henylprooiolic acid ~ t 'en with unfiltered co~ per rad.ir1tion in a crunera of • 86 c • radius. b rg photogr ph of pheny propiolic a -op er radiation in f the aKt reflections .cid taken with unfiltered c era. of 2. 86 c .. r dius . 35 ( exp ressed in terms G)f the above axes) app eared under the following cond itions: hkt present for h + k + i., even or odd; hkt present for h + k, k + t, h + I, even or odd; hO.t :pr esent on ly for h + t even; 0kt present fork+ t even or odd; hk:0 present only for h • k even. On the basis ef this information the unit is simple, with two ! type g lide planes. There are only two orthorhombic space grot1ps which require only the obser,red absences {after a. suitable interchange of axes). These are 10 G2 v - P n n and 12 Dah - P n nm. Since the crysta ls were not sufficiently well developed to distinguish morphologically between the Oav and the Dah point symmetries, the crystals were tested for pyroeleet:d.city by the n~thods enumerated in the previous discussion of KC(N02 ) 3 • A rather weak e:f:f ect was observed, which indicated that the space group to be 10 C:av - P n n; this conclusion is however not completely certain, and the 18 space group D2 h .. P n n m must not be dismissed. The interchange o! axes ~e:ferred to in the l a st paragraph is the following : a.1 changed to aa ti II a3 It ti Aa Ila A1 .• Further study of the erystal with the hope of obtaining a complete structure determination is 'being made. The p'tlrpose of the in vest igat ion is to obtain information concerning the resonance conjugation of the benzene ring, the triple bond, and the carbo:x;yl group. The a.mount of such r esonance ca.n ·be estimated from the values observed for interat0mie distances in the molecules. 36 rh"E SPACE GROUP A1rD UNIT CELL OF TFJ.: MOLECULAR COMPOTJND 1 NAPHT1IALEHE; p-:DINITROEEMZENE Crystals of t he molecular COIT-4)ound naphthaleneip-dini t:robenzene (1:1) were obta ined a s yellow prismatic needles by slow evap oration of the solvent irem zolutions containing the t wo oompcmeiats. Solvents best suited for preparing; the co~lex are those in wh ich the t wo comp onents have cornpc>. .rable $Olubil1ties. T'he solvent which seemed best for t h is part icv.la.r pair of substances ia ethyl acetate. The crystals melted a t 11S 0 c., in ~reement. with t h e va lue given by J?feif:fer 15. .a.'l optioa.l et:amina tion showed that the crystals were 'birefring ent. Goniometric measurements around the needle a.xis gave the fol10wing was t here:f ore m0noclinic or orthorhombic. A Laue photograph was taken with the x.--ray beam perpendicular to the needle axis and bis ecting the obtuse prism angle. showed a t wo-fold sy mmetry axis. This photograph Another Laue photograph with the x-ray beam parallel to the needle axis a..'1d with t h e tvm-fold axis vertical sho wed a horizon tal symmetry plane. These observatio·ns indic '.1"lied that t he cry stal belonged to the Ceh Laue class, a.nd was therefore monoclinic. The gnon.'Onic pro.iection made fr om the photograph with the t wo-fold a.xis indicated a monoclinic a.ngle of approxina.t ely 70°. 0scilVa.tion and rotation :;:,hotogra.phs using coppe r Ka. radiation (filtered through a 27 )-,{ nickel filter) were next m..-:ide with !:3 (the needle axis) or ~ 2 x.-r~ beam. ( the two-fold axis) vert ica.l and :perpendicular to the Th ese photographs were successfully indexe d on the bas is of t he following unit cell: 0 9.s5 + 0.02 A al = as = 10.78 .:± 0.02 A. 8.3 ::: 0 j3 z 0 6.97 .± 0. 02 A = 69o57• .., = ,:;.. 37 0 r.a,,e photo _,r a-ph oi the mol eeu.l co bew ia par allel t o _ 2 and with ound. IJ,'he x.- ray vert1c l. 38 0 1 e hotobr pl of the oolecul~ compo1..md. beam i ':'he x-r,...y parallel to ~:l v,ith ~~ vertical. 39 110 aoo 1 llo 0110 Rot.a,tion pr.otogra;)h o:f the molecular compo\;lld tak •n with filtered copper radiation usin cm. radius. po ~der c ,er of ~.O ':he nee,ile axil:. {a ) is vertical. ~ttern 1 ca.u"'ed b;/ the wax to wnic:h the crystal waa attached. The 40 The above unit cell was tested for correctness by means of a series of Laue photographs a.symmetric with respect to the previou.sly mentioned orientations of the crystal. None of the first order spots, which numbered about one hundred, necessitated a change in the axes. The density of the cry st a.ls was next determined by flot a.t ion in calcium chloride solutions, and was f9und to be 1.39 .! 0.02. This density corresponds to 1.9s molecules of the complex in the unit cell listed above. No systematic absences were observed except for the planes 0k0 which a appeared only fork= 2n. Hence the probable space groups are C2 - p 2 1 or 2 Cm - p 2. iii'" . The morphology of the crystal specimens did not permit 2 a choice to be made between the point symmetries C~ - 2 or Oah - - • m and therefore the crystals were tested :for pyroelectricity by the methods already described. It was not possible to make a definite decision as to whether the effect was present or not). and mt11 a more fei'ined technique for observing pyroelectricity is available both space groups are to be regarded as possible ones. Inasmuch ~s the only practical method of attacking a crystal of this degree of complexity is by the Fo'UI'ier :pro,jection technique further investigation of the complex was stopped. direction is down the tv-10 ..:fold axis. The only available projection Since this axis is over ten () Angstroms long it was felt at the time tha.t the results would not justify the effort involved. The goal of the investigation was to obtain information as to the nature of molecular compounds (see pa.rt II of this thesis). This particular complex was chosen because the structures of the component molecules were thought to be sufficiently well known to be of a.id in determining the struct\lre of the complex. PART II A THEORY Ol!" OHGAlHC MOLECULAR COMPOt'll DS 4( Introduction By a molecular compewid one means a stoichiometric complex of molecules, each of which exists separ ately and m~ be recovered from the cornplex unchanged. In a strict sense one should speak of a compo'Ull.d only if it can be isolated as a. pure subs ta.nee with definite phys ica.l :pra- pert ies; however, there a.re numerous unisola.ted complexes wh ich give characteristic evidence of their existence and hence m~ also be considered. '.:'he existence of complexes has been inf erred. in a variety of wa;,s. The s:lmplest and most definite evidence is the isolation of a substance of definite properties the stoichiometry of wh ich can be represented by a formula m.\•n:B • • • , in whiah A, . ih • • • a.re the molecular formulae of the component molecules and m, n • • • a.re integers. In other cases less defi nite evidence has been used as a.n indication of the existence of a col'.fll)lex. Max.im,-a, minima, or non-additivity in a plot of some physical property of a. mixture as a function of composition ha'lte. been used. Such physical properties are the melting point, viscosity, mole refractions, soh1bility, etc. Still another type-·of evidence is the appearance of a. strong color on mixing colorless or weakly colored solutiens of the components. This discussion will restrict itself to the compounds in which at lea.st one con:q:,onent is either an arom,1,t ic or an olefinic rnolee1;.le. In certain of these complexes only one of the components is organic, the other being an inorganic halide; while in others both components are organic. l!,ola tion of crxstalline molecul ar compounds ..... Early in the history of organic chemistry molecular compounds were known and crune to be used as a. means of identifying substances. One need only mention the hmdred.s of compounds in which one component is picric acid as a single e-xample of this use of the s e complexes. For the preparation of such complexes the si4'1plest of procedures was used, namely, the picric acid and the second substance were placed in a solvent (not in excess), heated; cooled, and the crystals filtered off and identified. others wez·e tried. This simple If one sol vent failed method in the hands of able experimenters added hundreds of ne w complexes to those already knoYm. In 1924 Dimrath and Bamberger 16 showed that the following expresSion was "1.0. lid for the molecular species in solu.tions from which complexes could be f' ormed: sA sl3 = G In this expression A and :S are the components of a complex C; SA, SB, and s 0 a.re the satur a tion quotients (stt.ttigungsgraden) of species A. :B, and C; and G is a constant independent of the solvent. stable relative to its components, and for G :Slor G ) 1 the complex is < 1 it is U11Stable and will completely decompose (if prepared in some other w~ such as by fusion of its components in proper proportions) when placed in any solvent. The rel ation of t his expression to the association constant is the following K = tcJ etc. [ti.1 i.13] in which square bra ckets denote ordina..ry con centrations and curly brackets the solubilities, each in mols per liter. Hence the only way in wh ich t h e solvent enters is by way of the solubill•s of the various species. These au.thors prop o-u.nded a set of 43 working rules for the choice of a sol vent to be used in preparing a co~lax. Most important of' the rules for stability of the complex in a solvent is: Splubilitl Qf more soh,ble component Solubility of less soluble component < G. In si~ler terms it me<IDS that suitable solvents are those in which the two solubilities a.re nearly the same, the restriction ·b eing less rigorot1.s with incre::i.sing (}. the "stability constant. 11 I n the course of preparation of numerous corrplexes of aromatic hydroc arbons with a.rom;,,,t ic nitro compotmds I have found the following solvents useful: methyl and ethyl a.lcohel; acetic acid; ethyl, propyl, butyl-. and ."l.myl .~ etates; and acetone. If well defined crystals suitable for x..-rey studies are desired slow evaporation of the solvent (mixed aol vents are generally not suitable for this t eohnique) at a nearly constant temperature is necessary. Exanmles of typic3.l cornplexEts. - At this point it may be ,,ell to list a few of the many hvndreds of complexes which have been isol3,ted in crystalline form: '1:oluene ! antimony trichloride l:l 'J:',.)1t1.ene t .antimony trichloride 1:2 antimony tribromide 1:2 Benzene i antimony tr1chlor1de 1;2 Durene : carbon tetrabromide 1:1 Naphthalene- 1 !)-d.ini trobenzene 1:1 Mthracene 1 2-trinitrobenzene 1:1 Hexamethylbenzene : s-trini trobenzene 111 Benzene : triphenylmethane 1:1 AJ1thra.cene : pic:dc acid ltl Benzene: platin-um tetra.methyl ltl (?) Benzsne 'f 1./Although the 1:1 complexes a.re by far t he most numerous they are not the only type found. As t ~o exampl es of other r a tios one might list: Fluorene: s-trinitrobenzene 2:3 s-trini t robenzene lt2 Stilbene Next in importance a.re t he complexes which h ave been postulated on the basis of color reactions. In numerous ca s es the admixture of two substances each dissolved in ;:m inert solvent produces a mo r e or leas intense a.nd characteristic color without permanent chemical changes appearing in t he molecules of either substance. A few exampl es will suffice as illustrations of t his t;ype of comp lex: Substances Solvent Color Benzene+ Tetranitromethane yellow red-violet l,1 1 di•p-tolyl ethylene+ SnC1 4 CsE's yellow 1,l'di-p-tolyl ethylene+ H2 S04 H2 S04 {cone.) yellow l,l 1 di-p-rnethoxyphenyl ethylene red-orange + SnC1 4 !Taph t hal en e + p- d i n it r obenzene yellow Triphenylethylen.e + SnC1 4 orange-y ellow Triph enyl etrwlene + Sn}3r 4 orange Ethyl en ic hy drocarbons + O(N02 ) 4 CCl4, y ellow to orange In numerous ca ses the only informa tion avail able concerning t hese complexes is the existence of a color rea ction. In other cases more thorough investigµ, tions of the nature of the colI!l)lex and its stability h a ve been ma.de. 45 Stabilit;y and energy effects in con52lex formation. - With regard to the question of stability of complex>?. s the existent data. are rather meager and disconnect ed. The early invest igato:cs of complex format ion were primarily interested in preparing new complexes and usually started with substances of great complexity. A.Ccordingly the great er part of the data do not perwit an easy analysis of what is relevant and essential for the existence of a complex. In marw eases the data are taken from long-forgotten doctorate theses. and equilibrium constants are given with no mention of wh:it the equilibrium measured is. 1 'l'he following table taken :from an article by Hamilton and Ramrnick 7 con ,.a,ins the rnost recent ~Ttd reliable data. available. rhe comiJlex 1 stability was determined by colorimetric determinations of the a.ssoci -" ltion constants in CC1 4 at several temperatures, The substances a.re s-trinitro- benzene and the substance in the first column of the t able. : -~ able I Substan ce o-'l' oluidine -.c:.:a: Substance -AH 2.0 o-Chloroaniline 2.1 kcal./mole k c al/mole m-Toluidine 2.0 m-Chloroaniline 2.0 :p-~:oluidine 1.7 p-Chloroaniline 1.9 Aniline 1.9 Methylan i l ine 2.3 o-:Bromoanil ine ,... 5 c. Dimethyla.niline 2.7 m-Bromoaniline 2.0 Dipheeylamine 1.8 p- Brornoaniline 2.1 Triphenyl.s.m.ine 1.1 d., Hapht~h.mine 3.8 In e ri c h the complex is assumed to be l: 1. The following t a ble taken from Davies and Ha.mrnick 18 is from a similar study of complexes between tetranitromethane and various aromatic hydro- carbons in CC14 solution. Table II Substance Subst ance -AH 1.4 kca.1./mol Triphenylmethane Du.rene o.66 kcal./mol Pentamethylbenzene Ethy 1 benzene 1.os Hexa.methyl benzene i-Propy l benzene 1.15 Dipheny l methane t-Butylbenzene In a.cldit ion to these may be added the r esnlt.s of Hammick and Yo,mg 19 f or tetranitromethane and various other substances in CC1 4 • Table III Subs ta.nee ----------- Su"t,s ta.nee -AH p- Xylene o. 72 k ca l/mol Di phe:oyl ether 1. J5 /3 Naphthol 0. 86 Naphthalene O. 60 a,..Ni tronn.ph t halene 1. 69 (3 l,ietbyl n aphthalene O. 71 .1inthracene 2 .4-Dimethylnaphthalene -AH o. :>8 k:ca.1/ mol Probably t h e only completely tenable con c lus ion that c an be reached from these data is tha:t in -~11 ca ses the heat effect is small ,and of the sa.me magn itude. Minor fluctuations among isomers may or ffi3Y not be of significance because of' the method used for the determina tion of the equilibrium c onstants • . BX:periments performed by rrvself wi t h tetranitro- me thane and naphthalen e in CC1 4 show tha t n ot only does t he color intensity in crease with increase in the concentra tion of the components . but the col or als o changes in char a cter from a orang e. 11 chromate 11 yellow to a 11 dichrom-'.lte 11 Th is chang e may be sui'ficient to invalidate col orin,eti·ic methods for high concentrat, ions. . Further da ta in the for m of d is s ociation cons t ants of c o~)lexes in benzene solution are gi ven by the following table from Pf'eiffer 1 5 . Table IV SnC1 4 Sn:Br 4 6.74 23.s orange-yellow orange 2.75 12.8 yellow orange-yellow 0.0251 0.114 orange red-orange ~C=()_,,,R db o._C=C/H 0 '-oH3 CH30'(\ H /.....____ C=C/ if "-cH3 ~atv.re o:f the CO!S)lexes ... From the smallness of the heats of formation of the complexes it can be st a ted with defin Heness that the bonds formed cann ot be covalent bonds. therefore be of a van der Waals type. The stabilization must Since complexes csn form between molecule pairs only one of which contains a group to which a permanent dipole m':'JY be attributed, dipole-dipole inter:1ctions between permanent dipoles do not provide the answer. In all cases in which a. complex does form one of the molecules must be an aroma.tic or ethylenic substance and t h e second molecule must possess one or more dipole groups of large or moderately large strength. Hence one form of sta.bilization may be the polai•iza.t ion of the aromatic ar ethylenic substru1ce by the second molecule with subsequent d ipole-dipole interaction. The fact that in general only olefins and aromatics, in which the average carbon atom polarizability is some 35% gre '.a ter than in aliphatic compounds, f orm complexes lends support to this vie\11. 1+8 In a ddition to t h is and p ossibly of gr eat er importance axe t h e London disp ,orsion forces wh ich a lso dep end on the a tomic pol a.riz abilities. Uni'ortuna.tely a direct c e.lcrul a tion of t hese effects is too difficult to make even with only approxill'lt}te reliability. To ha.ve significance such a calcul ation would have as a necessary prerequisite t h e kn owledg e of the configuration of the complex. Since none of tbe complexes has received such thorough study this avenue of a t;tack is a,t least temporarily closed. Nevertheles s pertinent argv.men ts in f avor of t he e ssenti a l correctn ess of t he van der Wa.a.h s t abilization c an be g iven . A.s a starting point we may examin e t he exper i mentally de termined arrMgewent of mol ecu l es in t he crys t als of' p- d ichlor obenz en e bromobenzen e 20 20 , p-chloro- ?0 21 22 , p-d ibr om.obenz ene .. , quinone , and p- di n it1·obenz ene • I n all of t hese substanc es t he subst ituent groups of one molecule are loc at ed a t a p oint near t h e per pEn,dJ. cul ar t o t he cent er of t h e ring of a second molecule. In p- d i n itroben~ene and ½.uinone the molecules are z.r:rfmged i n string s of th e t yp e .: -.~>?~o. . • • C/ () 0 1/ It will be no ticed t he,t t hese are j ust the positions which woul d be expected to lead to a larg e van der Wa.a.ls stabilization, with the p ol ar izing group close to the c en t er of maximum polariza.bility ~ Obviously in an app roach of such intimacy t he ide a lization of a dipole can no long er be ma intained and t h e dipole must now be t h ought of as an asymmetrical charge distribution. 'It is not surprising that the "dipole" of the nitro group is not normal to the ring. Indeed it is in a position of maximum effectiveness since the benzene ring must ce:t't:;;,inly have a higher pola.riza.bility pa..rallel to its plane than normal to it. G'n the basis of the preceding discussion we m.::w postulate that in a benzenoid or olefinic complex the molecules will be in similar configurations. Ji'or the complex of' naphthalene and p-dinitrobenzene one might expect some arrangement as: Further we IDIJIY postulate that the stability of the complex will decrease if for steric reasons such an approach is not p ermitted. The existence of only a feYT crystalline complexes of aliphatic nitro compounds with arom::i.tic molecules mt"\Y' be explained in this w,'33. By the geometry of the aliphatic molecule S'UCh as C(lli0a) 4 the nitro groups cannot be both close and par allel to the ring simultaneously and hence a. decrease in ata.bility results. Simil,'.ll" remarks will apply to complexes involving other tetrahedral molecules, such as SnC1 4 , Sn1lr4 , etc. Another type of steric e.ff act which can be oper11.t i ve in the molecule containing the nitro group is one affecting the nitro groups of a !+!Olecu.le su.ch as trini tromesitylene as comp3.l'ed with s-trinitrobenzene. In the former compound the ni tro gx·oups will be pushed out of coplanarity with the rest of the ring by the bitervening methyl groups. 50 This will have two effects: first there will be a decrease in the nitro moment (J3irtles-Harf!PSOn effect), and second the nitro group will not be able to achieve as close an approach to a second molecule as in the methyl-free compound. In support of these arguments there a.re the observations that all attempts to p1•epare a complex containing tri- nitromesitylene have failed, although the corresponding s-trinitrobenzene complexes exist. W61 m;;ey next consider the effect of substituents in the other molecule of the conplex, i.e. the aroma tic or olefin. ln the discussion only aromatic molecules will be mentioned with the assumption that similar arguments will apply to olefins. An examin~t ion of the hei?,ts of format ion of the complexes of C(!J0~) 4 with various benzene derivatives may be used as the ba.s is of the arguments {cf Tables I, I!, I II). ,\'bout all that can be said is that substituents exercise no gres.t influence (in addition to steric effects) on the stab:ility of the complex. In g eneral addition of methyl or amino groups prorr<0tes st.ability• the effect being more pronounced for a.'ilino groups either simple or substituted. i~his w,ay be attribut ad to slight increases in the polarizability of the ring by resonance of such structures as and -• • 0 /1-i =C.-H t\ + Sv.ch. resonance would be grets.test for amino g ro1;p s, and hence the correlation with stability increa ses IDc-.Y be regarded r,.s satisfactory. Another w9:1 in which a substituent can o:per~te is by the inductive ef fect, 'Whereby 51 electrons are ~emoved from the r ~ with a loss of polarizability and hence of complex stability. This can be correla ted with the existence of only a few Coll'\Plexes between d.i:ff e.r ent nitro compounds . AS a rough rule one might say that the ortho-para directing substituent;s stabilize a co~plex while the meta directing groups de... stabilize, the effect in either case being not very great . .c\noL er piece of evidence for the van der Waals stabiliz a tion can be obtained fro m Table IV. I n a ll ca ses a COTuJ)lex of SnBr4, is less stable (as s hown by t h e equilibrium constant) t ha.n the SnC1 4 coll4)lex with the s ame substance. This may ne expl a ined by the s maller va.n der Waals radius of Cl as compared to Dr and t he r t::sul t an t n:.o re i ntimate approach of SnC1 4 as compared to Sn:Br 4 • in the group 11 Also there is p robably a small decrease dipole 11 moment of Sn- Br as co mpared to Sn-Cl wh ich would f urther destabilize t h e compl ex. Another complex which can be explained only on a van der Via.a.ls basis is that from benzene and platinun:-tetramethyl observed by Mr . R1.mdle and nzyself of these laboratories. It was observed tha t vihen platinum- t etramethyl crystallizes out of benzene, birefring ent needle-like crystals a.re for med 1 wh ich disintegrate to a:n. i s otropic powde1· when t h e solvent is completely evapor ated. The powder gave a cu bic x-ray pattern. just a.s did sing le crystals or powders from other solvents, such as carbon disulfide and dibutyl ether. The anisotropic crystals st a.ble in the presence o:f ben:tene were undoubtedly those of a complex. '!'. he color of molecular compou.nd_s. - In the cas e of most molecular compounds. p articular the nitro con,-plexes, chara.ct ,:::dstic color changes oc:cur when the components a.re mixed. In general these colors are yellow, orange, red. and in some cases (arom:-1.tic -9,mines with nitre> compounds) violet, brown, er black. Any satisfa ctory theory of molecular compounds must indicate at least qualit~tively how such celors arise. That the color depends on both components may ·be seen from the following: Ben:i.o:ene + tetranitromethane yellow Naphthalene + tetranitromethane yellow to ora.nge Anthracene + tetranitromethane red-orange Naphthalene + s-trinitrobenzene pa.le yellow JITaphtha.lene + 2, l~, 6-trinitrotoluene colorless Naphthalene + 2.,4,6-trini tro an iline orang e-y ellow ~1aphthalene + 2 , 4, 6-trinitro-chlorobenzene bright y ellow It will be well to repeat at this point that only r.ll'orrn.tic and olei'inic cornpovnds show color re:?,Ct ions and complex formation. provid.es the clue to a po ssible origin of the color. This All of the olefinic and ax·o m3.tic compov.nds exhibit characteristic e.bsorpt ion in the near ultra violet. the absorption in complicated molecules extending 1nte> the visible (anthracene. polyenes, etc.) with the substances showing a. yellow color or tint. It is now suggested that when .9. color s;opears in sol'lAt ions of co1:4>lexes this color 1s to l)e x·el a ted to the absorption bands in the nee..r ultraviolet, the S!1ift of the band into the visible being caused by some non\J.niform polarization of the arom,:lt ic or olefin molecule by t h e second. molecule. The essential cor·rectness of this explanation can ·be seen .from the colors of' the tetr?...nitromethane complexes with the series ben~ene, 53 naphthalene, .and anthracene; in these r...ydrocarbons the 11 unperturbed 11 a bsorption moves in toward the viaible and the col0r of the complexes is accordingly deeper in the order named. The effect of changing the second component of the complex can be seen from examining the sequence Benzene + carbon tetra.bromide (solution) colorless ) colorless Benzene+ carbon tetrachloride ( II Benzene + chlorpicrin {CC1 3 1W2 ) very pale yellow J.1eniene + tetra.nitromethane yellow It is evident that an incre,.,se in t he group an i n c:s.·e ,1.se in t he color of t h e complex. 11 dipole 11 momen ts results in I n gene r a l the an,01Jnt of the s h ift in the absorption rr:~.xirm.1m par a llels t h e stability of t h e complex, al tho~:h not in a pracise way . i,f:;ain it is unfo:rtv.nate tha t no more pr ecise treat ment of the color of conrplex2s cw be g i ven as yet. complete kri.owledge of ti.1e To make such calcula tions a more 11 unperturbec1. 11 a bsoj:p tion and of t h e electronic wave functions would be nece~sa.::y. The sugg r,s tions ma.de here seem to be at l eas t qualitatively correct. Surmr:1,.r'l of E:."<-isting Structural Dat a for Complex Compouncls. A.s has been mentioned before, no complete stn1ct·cr· e investig~:'l.tion of 8, molecul a r con~, ound h~ts been ma de to da te. A nu.mber of x-r8JY investi- g. , tions of crysta.18 ha.ve teen ca rried out as f ar as the deterr,,ina.tion of 0 t !·,e unit cell and space group. In '2£tb le V these are su.Irm,~...rized. Of those list ed , t h e first (naph t halene : p-dinitrobenzene) is pro bably the one l)est suited to :f-urthe!' study bec ause of t he s1;1allness of t h e u nit cell and the simplicity of the components. .A nother comp lex , as yet n ot in vestigated, bet ween h e:x:amethylbenzene a"l d s-trinitrobenzene m,":iy prove still simpler. .~ pr 3,ctical obst a cle in this case is the p repara.t ion of good single crystals. rt'he substance shows a r:1ar k ed tendency to ward t winning , out this dif:ficulty mi ght be overcome. Cert a inly 11 complete inve,;tlgaticn of on e of t h e conxplexe s would be worth t h e effort involved, for then t he views with reg£>..rd to the structure of comp lexes given in the previous sect ions could be g ive~ a direct test. 26 1:1 red :form yellow form 4-Erom-1-naphthylamine : 2.6-d.initrophenol"'"7 1:2 Stilbene: S-trinitrcbenzene 3:4 Fluorene : s-trinitrobenzeu/5 1:1 ' 24 Aniline -: picric acid 1:1 2 Naphthalene : p-dinitrobenzane 3 Complex 0 7• 7 (3 = 1050 a= 0 9.5 A b = 13.5 C = 13.8 b = s.o C::: l4o5 /3 = 102°6• 0 a= 14.0 A /3 = s,o 3o, rt = 37°35 f a.= 102016• = .,, = F.4 b C a :::: 12. 7 A 0 4 ).f 2 1 = 2$.2 (3 = 97° C 0 7.64 ~ b = 7.3l+ a= 4 0 2 Eolecul es of comp lex in unit = 13. 2 A. b = 7.4 c = lj.2 /3 = 93° a /3 = 69°57' = 9• 85 + 0. 02 A. b =10.78 + 0.02 C = 6. 97 + 0.02 a Ur:i t Cell Table V T,') -"''-1 ]> Cah - P ~ C 6 5 C211 - p .=:..1 C "I £:l C a C2 - P2 1 6 0 2h - or S;2a.ce Gro"0J? .., C:2h- p £:.1. m :e C2 - 2 u, V'\ '"l l'\ 28 1:1 A.cenaphthene : styp11nie acid30 1:1 Acenaphthene : Ll-.6-dinitro-l,3-x;rlene29 ltl Fluorene: 2,7- dinitroa.nthraquinone 1:1 Anthra.cene : s-•rrinitrobenzene'- ..- C 0 C ::: 6.8 (3 = 9:;0 b = 1 1~-.8 - 3. 0:> A = 1 030 i:, ........ /3 = 7.25 (I a = 13.::;, i\ b = 1 1i.2 rt= 32• = e2° a. = 730 f3 0 0 a = 8.2 A b = 7.4 C = 19.0 \ a = 13. it 1\. b = 16.3 C = 9.9 13 = 122° 2 4 2 4 4, ,, 08 l C2 1 1 - - Pm. P2; or 2 Cah-P-; . .tn C3h- p .sJ. a 5 Cs - I !'l 0) <.rt S7 SU1mnary The follov,ing ta.ble is a. summ-ll'y of the interatomic distances and valence angles for the phosphorus and arsenic compounds discussed in part I of this thesis. P=O or P=S X-0 LX-0..X L 0-X-0 0 A.s 4 06 1.80 A 126° 1000 l?40e 1. 6:, 127.5° 99° P4:010 1.62 " 1°39A 123.5• 101.50 P4 06 S4 1. 61 1.s5 123.5° 101.50 For hexamethylenetetrandne the interatomic distances and valence angles a.re: 0 C-H 1.09 A (.assumed) L N-C-N 0-N 1.47 · L-0-N-C 109°28' L H-C-H 109°28 1 (assumed) The investigation of tetranitromethane resulted in the following interatomic distances and valence angles : 0 0-N 1.47 A L 0-N-O 127° (assumed) N-0 1.22 L N-0-N 10902g • (assumed). In order to obtain a satisfactory interpretation of the electron diffraction data it was necessary to assume a rotatory oscillation of the nitro groups about their C-N bonds, and to include the usually omitted ten;perature factor. The remaining pages of part I are devoted to a. presentation of the results of x-r~ investigations of three crystals. The unit cells and space groups are the following: 0 a1 = 11.31 A. a.3 = Space groupt z = g s. 77 l2 - ¾o_ - l 4 2d 58 Phenylpropiolic acid z = 4 as = 5.12 10 Space gro'tl:_p: C2 v - P n n ~i:olecular compound: naphthalene ; p-dinitrobenze:ne Z = 2 = 69•57• 2 Space group: 0 2 - P2 1 In part II of this thesis a theory of organ ic molecular compounds based on. van der Waals stabilization of complexes is proposed, and arguments in support of the theory a.re presented. 59 Ref erenees: :)l, 2◊99 (1910) 1. Cha.ttaway, J. Chem. Soc. 2. Menzies, J. Am. Chem. Soc. !il_, 1336 (1919) hi, -~691 , (1. 9 ,;7·.,,~) 3• n.oro ck way, n.oea.c·n , ana'~..,au l'ing, ib1'd. ...t.L. l.~. Brockway, Rev. Mod. Phys. 5• Mark and Noethling, !, 231 (1936) z. f-tir Krist. .§;J., 435 (1927) 6. Williams, Physik. zt. ~. 271 (1928) 7. Weiss berger and S&ngewa.ld, Ber. .§2_, 701 (1932) 8. Thorpe and Tutton~ J. Chem. Soc. .22, 1023 (1$91) 9. Ra.nrpson and Stosick, J. Am. Chem. Soc. 10. Pauling and Huggins, Zt. f-6.r Krist. 11. Pauling and Erockwa;y, J. Am. Chem. Soc . 12. Bz•ockwa;r and Ben.ch, ibid. 13. Berzch and Stevenson, J. Chem. Phys. li_~. Chattawa,y and Harrison, J. Chem. Soc. lj. Pfeiffer, Organische Molekfilverbindungen, FerdinEU1d I.1.ii;:e, Stuttgart (1927) 16. Dimrath and Bamberger, Ann. 17. Hamil ton rmo. Hrurirnick, J. Chem. Soc. 13. Davies a.nd llamdcl+:, ibid. 764, 1933 19. Hamruick and Young, ibid. 1463, lil§, 20. Hendricks. Zt. f"ttr Krist. .§.!±. S5 (1932) 21. .Robertson, Nature 22. Jamers, Kin,,;;, and Horrocks, Free. Roy. Soc. (,\) lS3, 225 (1935) ,_"3 . Part 2l+. .!2,2, 13 (1937) 60, 1836 (1933) ~' 7') (1933) 109, 172 (1916) 4]!_, 67 (1924) !, this t hesis. Hertel a.nd Schneider, z. l~hy s ik. Chern. (:S) jg, 109 (1931) 11, 77 (1930) Hertel and Schneider, ibid. Hertel and Cleef, Eer . 12,. 79 {1931) .§1, 1~45 (1928) Hez·tel and lH!mer, z. Physik. Chem. (:B) 11, 90 (1930) He:rtel a.nd !Heu, ibid. 30. fil, 205 (1934) 134, 143 (1934) Hertel and Rtimer, ibid. 2S. .§Q, 1814 (l;~o) Schneider, ibid. ll, 59 (1930) 16. 460 (1932) 60 Propositions l. Organic molecular con:pounds of the olefin or aromatic series can be divided. into two classes depending on the function of t he olefin or a.rom<-:i.tic molecule. 2. 'xhe h e,ats oi' :f'onnat ion of molecular compounds can be determined with greater precision and relb.bility than they have been to date. 3. The known structure of the tetranitromethane gas molecule is compatible with existing data on the c~Jstal if 1:-ota.tion of the &iolecu.les in the c.rystal is ass\lilled. 4. ~he electron di:t'i'raotion investiga tion of tetranitroreethane provides a.nether tYPe of situation for which the temper ,1.ture f actor must be considered. 5. t.qJ})l'oxiwate predict:i.ons of boncl lengths in niolecu.les can be made from thermochemica l data. 6. Simple valence i)ond ca.lculr:1.t ions pr edict the followin~ order of st 9.bility for the banzoquinones: 7. ortho > para /' meta. A. va.lue for the molecuh.r orbit al Coulomb integral suit able for predicting ionization p otentials of aromatic or ,msa.tu:n,,ted moleeul es is: q = 3. 6 electron vol ts . A probable structuxe foi· N4 S4 is that of_ EUl eight membered. ring. The use of a lecture syll :;.bus would mBtterially improve the freshman chemistry cou:cse. 10. In the interests of a more complete training i' ir s t yeax graduate students O'IJ.ght to be allowed more freedom from the i nordina te time demands of r es(~a.rch.