Optical Expressions of Ion-Pair Interactions in Minerals

Citation

Mattson, Stephanie Margaret (1985) Optical Expressions of Ion-Pair Interactions in Minerals. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/acn3-nr54. https://resolver.caltech.edu/CaltechETD:etd-04082005-133814

Abstract

Clusters of transition element cations in neighboring sites frequently govern the optical properties of minerals. This is particularly true of Fe-bearing minerals which may exhibit several types of ion-pair transitions. In this thesis four different types of interactions were distinguished: intensified spin-forbidden transitions of Fe ³⁺ clusters, intensified Fe ²⁺ spin-allowed transitions of Fe ²⁺ -Fe ³⁺ clusters, heteronuclear charge transfer transitions of Fe ²⁺ -Ti ⁴⁺ and Mn ²⁺ -Ti ⁴⁺ clusters, and homonuclear charge transfer transitions of Fe ²⁺ -Fe ³⁺ clusters.

The optical characteristics of Fe ³⁺ in red Fe ³⁺ rich and black Fe ³⁺ ,Fe ²⁺ -rich tourmalines were examined by absorption spectroscopy in the visible and near-infrared, Mössbauer spectroscopy and magnetic susceptibility measurements. Intense optical absorption features at 485 and 540 nm were assigned to transitions of exchange-couple Fe ³⁺ pairs in two different site combinations. Absorption spectra at variable temperatures and of samples which were oxidized and reduced were used to establish these assignments. Site assignments were based on intensity ratios in different polarizations according to the polarization of these transitions along the vector between the interacting cations. The 485 nm band occurs at an unusually low energy for Fe ³⁺ in silicate minerals. Similar behavior was observed in the spectrum of coalingite, a Mg,Fe-hydroxy carbonate, and has been proposed to result from magnetic exchange in large, edge-shared octahedra. The antiferromagnetic exchange which is generally associated with intensity increases in Fe ³⁺ clusters was confirmed by variable temperature magnetic susceptibility measurements. The Mössbauer spectrum of a red tourmaline with 3.4% Fe exhibits an unusual decrease in width of peaks by ~30% from 298 K to 5 K which may be related to an unusual interaction between Fe ³⁺ and trace amounts of Fe ²⁺ .

Optical absorption and Mössbauer studies of Fe ²⁺ -bearing tourmalines with variable Fe ³⁺ contents were used to examine Fe ²⁺ transitions which are intensified through an interaction with Fe ³⁺ neighbors. The variation of molar absorptivity of Fe ²⁺ bands with the fraction of Fe ²⁺ in Fe ²⁺ -Fe ³⁺ pairs indicates that Fe ³⁺ ions increase the absorptivity of Fe ²⁺ bands to ~1200 M ^-1 cm ^-1 as compared to ~5 M ^-1 cm ^-1 for non-interacting Fe ²⁺ . Approximately equal degrees of intensification were observed for both components of the ⁵ T ₂ → ⁵ E Fe ²⁺ transition as well as for Fe ²⁺ in two different sites. Although the detailed behavior of non-interacting Fe ²⁺ ions differ in Mg-tourmalines and Li,Al-tourmalines, the characteristics of Fe ²⁺ -Fe ³⁺ absorption are constant. Intensity increases were restricted to the polarization which coincided with the vector between the Fe ²⁺ and Fe ³⁺ ions. The intensified Fe ²⁺ transitions are characterized by an unusual temperature response. The integrated intensity of these transitions increases by 10-50% at 83 K as compared to 296 K. The positions and widths of the intensified transitions maintain the values of the non-interacting Fe ²⁺ . Tourmalines with the lowest Fe ³⁺ contents were the gemmy Li,Al-tourmalines which generally form in pockets within pegmatites. Fe,Mg-tourmalines exhibited consistently higher Fe ³⁺ contents than any of the Li-bearing tourmalines examined. Oxidation of Fe ²⁺ which resulted from gamma irradiation of blue Li-tourmalines which contained several percent each of MnO and FeO could be monitored by increases in Fe ²⁺ intensity in one polarization.

Fe ²⁺ -Ti ⁴⁺ charge transfer transitions were examined in minerals which contain stoichiometric quantities of Fe and Ti -- taramellite, neptunite, and traskite -- and tourmaline. The wavelength of these transitions ranged between 400 and 500 nm, and the halfwidths ranged between 7000 and 9000 cm ^-1 . These characteristics can generally be used to assign Fe ²⁺ -Ti ⁴⁺ charge transfer transitions. The molar absorpitivities of these transitions, however, exhibit very large variations. The molar absorptivity of Fe ²⁺ -Ti ⁴⁺ charge transfer in neptunite is ~225 M ^-1 cm ^-1 in beta polarization, in taramellite it is ~1300 M ^-1 cm ^-1 and in tourmaline it is ~4000 M ^-1 cm ^-1 . Tentative assignments of Fe ²⁺ -Ti ⁴⁺ in more dilute minerals generally compare favorably with the energy and width stated above. However, sapphire and other Al-minerals such as kyanite have very different characteristics for bands assigned to Fe ²⁺ -Ti ⁴⁺ charge transfer. The Fe ²⁺ -Ti ⁴⁺ charge transfer transition in taramellite exhibits no change in integrated intensity with decreasing temperature but increases by 15% from 296 K to 83 K in tourmaline. Mn ²⁺ -Ti ⁴⁺ charge transfer was also assigned to a transition at 320 nm in two unusual yellow tourmalines.

The characteristics of Fe ²⁺ -Fe ³⁺ charge transfer transitions were reviewed in light of recent data and with regard to their utility as diagnostic criteria. A correlation between charge transfer energy and the separation of the interacting cations proposed by Smith and Strens (1976) could not be supported by the expanded data base. Temperature variations of charge transfer transition areas were also examined. The magnetic behavior of two minerals which exhibited different temperature responses were investigated. General agreement with the theories of Cox (1980) and Girered (1983) that suggest that ferromagnetic exchange should produce intensity increases at low temperature and that antiferromagnetic exchange produces intensity decreases was confirmed by these examples of Fe ²⁺ -Fe ³⁺ charge transfer but could not explain the temperature response of Fe ²⁺ -Ti ⁴⁺ charge transfer transitions. In any case, an increase in intensity with decreasing temperature, which is generally expected on the basis of experimental observations, cannot be used in a negative sense to eliminate a charge transfer assignment. The large width of charge transfer transitions is generally the most useful diagnostic criterion.

Cox, PA (1980) Electron transfer between exchange-coupled ions in a mixed-valency compound. Chem Phys Lett 69: 340-343

Girerd, J-J (1983) Electron transfer between magnetic ions in mixed valence binuclear systems. J Chem Phys 79: 1766-1775

Smith, G and Strens, RGJ (1976) Intervalence transfer absorption in some silicate, oxide and phosphate minerals. In: The Phyics and Chemistry of Minerals and Rocks, Strens, RGJ (ed.). New York: Wiley and Sons, pp. 583-612

Item Type:

Thesis (Dissertation (Ph.D.))

Subject Keywords:

Geochemistry

Degree Grantor:

California Institute of Technology

Division:

Geological and Planetary Sciences

Major Option:

Geochemistry

Thesis Availability:

Public (worldwide access)

Research Advisor(s):

Stolper, Edward M.

Thesis Committee:

Albee, Arden Leroy (chair)
Kamb, W. Barclay
Rossman, George Robert
Stolper, Edward M.

Defense Date:

11 January 1985

Funders:

Funding Agency	Grant Number
NSF	EAR 8212540
NSF	EAR 7904801
Gemmological Institute of America	UNSPECIFIED

Record Number:

CaltechETD:etd-04082005-133814

Persistent URL:

https://resolver.caltech.edu/CaltechETD:etd-04082005-133814

DOI:

10.7907/acn3-nr54

Related URLs:

URL	URL Type	Description
https://doi.org/10.1007/bf00308220	DOI	Article adapted for Chapter 2.

Default Usage Policy:

No commercial reproduction, distribution, display or performance rights in this work are provided.

ID Code:

1307

Collection:

CaltechTHESIS

Deposited By:

Imported from ETD-db

Deposited On:

08 Apr 2005

Last Modified:

16 Apr 2021 23:19

Thesis Files

Preview

PDF (Mattson_sm_1985.pdf) - Final Version
See Usage Policy.
5MB

Repository Staff Only: item control page

OPTICAL EXPRESSIONS OF ION-PAIR INTERACTIONS IN MINERALS Thesis by Stephanie Margaret Mattson In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosovhy California Institute of Technology Pasadena, California 1985 ii Acknowledgments My special thanks are due to George Rossman, my thesis advisor, fer the many enjoyable discussions we have had regarding this thesis and related topics and the freedom to follow my own direction. I have also appreciated the interest expressed by Edward Stolper in my progress and activities. I would like to gratefully acknowledge the influence of my interactions and collaborations with my fellow graduate students. In particular, the support, help and ideas of Roger Aines, Anne Hofmeister, Gerry Fine, Cleve Solomon and Greg Miller have been of great value. Samples for this research were kindly provided by Pete Dunn iSmithsanian Institution), Fete Flusser (Los Angeles), Robert Gaal and Peter Keller (Gemmological Institute of America), Julius Petsch (Idar—-Oberstein), and Roger Burns (MIT). Méssbauer spectra and helpful discussions were provided by Wayne Dollase (UCLA), Glen Waychunas (Stanford), Georg Amthauer (Marburg) and Roger Burns (MIT). My appreciation is also extended to members of Peter Rice’s group (USC) for assistance in obtaining the magnetic susceptibility data and Edward Stolper (Caltech) for the use of controlled atomosphere ovens. This study was funded in part by NSF grants EAR 8212540 and EAR 7904501 and a grant from the Gemmological Institute of America. iii Abstract c ht C uf rr th 4 uw 0 wh rr 4° r ransition element cations in neighbori 2 g Sites frequently govern the optical properties of minerals. This is) particularily true of Fe-bearing minerals which may exhibit several types of ion-pair transitions. In this thesis four different types of interactions were distinguished: intensified spin-forbidden transitions of Fet* clusters, intensified Fe=" spin-allowed transitions of Fe-*-Fe tt clusters, heteronuclear charge transter transitions of Fe**-Tit* and Mn=*-Tit~* clusters. and homonuclear charge transfer transitions of Fe?*-Fet> Clusters. The optical characteristics of Fe"" in red Fe *-rich and black Fes*,Fe=*-rich tourmalines were examined by absorption spectroscopy in the visible and near-infrared, Mossbauer spectroscopy and magnetic susceptibility measurements. Intense optical absorption features at 435 and 340 nm were assigned to transitions of exchange-coupled Fe ** pairs in two different site combinations. absorption spectra at variable temperatures and of samples which were oxidized and reduced were used to establish these assignments. Site assignments were based on intensity ratios in different polarizations according to the polarization ef these transitions along the vector between the interacting Hi tia a) tA cations. Th: mm band occurs at an unusually iow energy a ed ke re pou ilicate minerals. Similar behavior was bel At observed in the spectrum sof coalingite, a Mg,Fe-hydrox, iv carbonate, and has been proposed to resuit from magnetic exchange in large, edge-shared octahedra. The vantiferromagnetic exchange which is generally associated with intensity increases in Fe=* clusters was confirmed by variable temperature magnetic susceptibility measurements. The Méssbauer spectrum of a red tourmaline with 3.4% Fe exhibits an unusual decrease in width of peaks by “30% from 2598 K to 5S K which may be related to an unusual interaction between Fe=* and trace amounts of Fe=*. Optical absorption and Méssbauer studies of Fe=*- bearing tourmalines with variable Fe=* contents were used to examine Fe#* transitions which are intensified through an interaction with Fe** neighbors. The variation of molar absorptivity of Fe?* bands with the fraction of Fe?* in Fe?*-Fe** pairs indicates that Fe =* ions increase the absorptivity of Fe** bands to “1200 M—tem-? as compared to ~S Mctcm-*? for non-interacting Fe=*. Approximately equal degrees of intensification were observed for both components of the =Tz > 7E Fe=* transition as weil as for Fe?" in two different sites. Although the detailed behavior of non- interacting Fe=* ions differ in Mg-tourmalines and Li,Al-tourmalines, the characteristics of Fe*t-Fe>> absorption are constant. Intensity increases were restricted to the polarization which coincided with the vector between the Fe=* and Fe®* ions. The intensified Fe=* transitions are characterized by an unusual temperature response. The integrated intensity of these transitions . ; v increases by 10-30% at 83 K as compared to 294 K. The posi- tions and widths of the intensified transitions maintain the “values of the non-interacting Fe=*. Tourmalines with the lowest Fe =* contents were the gemmy Li,Al-tourmalines which generally form in pockets within pegmatites. Fe,Mg—-tourma— lines exhibited consistentiy higher Fe** contents than any of the Li-bearing tourmalines examined. Gxidation of Fe=* which resulted fram gamma irradiation of blue Li-tourmalines which contained several percent each af MnO and Fel could be monitored oy increases in Fe" intensity in one polarization. Fe=*-Ti+** charge transfer transitions were examined in minerals which contain stoichiometric quantities of Fe and Ti--taramellite, neptunite, and traskite--and tourmaline. The wavelength of these transitions ranged between 400 and 500 nm, and the halfwidths ranged between 7000 and 70006 cm—*., These characteristics can generally be used to assign Fe?*-Ti** charge transfer transitions. The molar absorptivities of these transitions, however, exhibit very large variations. The molar absorptivity of Fe?*-Tis~ charge transfer in neptunite is “225 M71*cm—+ in beta Polarization, in taramellite it is “1300 M7+cm7—? and in tourmaline it is ~4000 M7tcm—-?. Tentative assignments of Fe=*-Tit** in more dilute minerals generaily compare favorably with the eneroy and width stated above. However , sapphire and other Al-minerals such as kyanite have very different characteristics for bands assigned to Fe**~-Tit* . vi charge transfer. The Fe2+-Ti++ Charge transfer transition in taramellite exhibits no change in integrated intensity with decreasing temperature but increases by 15% from 296 K to 83 Kk in tourmaline. Mn2*-Ti** charge transfer was also assigned to a transition at 320 nm in two unusual yellow tourmalines. The characteristics of Fe**-Fe** charge transfer transitions were reviewed in light of recent data and with regard to their utility as diagnostic criteria. A correlation between charge transfer energy and the Separation of the interacting cations proposed by Smith and Strens (1976) could not be supported by the expanded data base. Temperature variations of charge transfer transition areas were also examined. The magnetic behavior of two minerals which exhibited different temperature responses were investigated. General agreement with the theories of Cox (19780) and Girerd (1983) that suggest that ferromagnetic exchange should produce intensity increases at low temperature and that antiferromagnetic exchange produces intensity decreases was confirmed by these examples of Fe?*—Fes* charge transfer but could not explain the temperature response of Fe**-Tit* charge transfer transitions. In any case, an increase in intensity with decreasing temperature, which is generally expected on the basis of experimental observations, cannot be used in a negative sense toe sliminate a charge transfer assignment. The large width of charge transfer transitions is vil d- generally the most useful diagnostic criterion. \ Cox, PA (1986) Electron transfer between exchange-coupled ions in a mixed-valency compound. Chem Phys Lett 69: 240-343 Girerd, J-J (1983) Electron transfer between magnetic ions in mixed valence binuclear systems. J Chem Phys 7921766-1775 Smith, G and Strens RGJ (1976) Intervalence transfer absarp- tion in some silicate, oxide and phosphate minerals. In: The Physics and Chemistry of Minerals and Rocks, Strens, RGJ (ed.). New York: Wiley and Sons, pp. 5S83- 4612 Vili Table of Contents Acknowledgments . ii Abstract! iii Extended Table of Contents ix Chapter 1 Introduction 1 Chapter 2 Fe **-Fe + Interactions in Tourmaline 9 Chapter 3 .Fe@*-Fe=* Interactions in Tourmaline 49 Chapter 4 Heteronuciear Charge Transfer in Minerals a9 Chapter 3 Characteristics of Intervalence Charge 126 Transfer in Minerals Chapter & Summary 146 Appendix i351 ix y Extended Table of Contents Chapter 1. Introduction References Chapter 2 Fe *-Fe** Interactions in Tourmaline , Introduction Experimental Results and Discussion Magnetic Moment Missbauer Spectroscopy Optical Spectroscopy Fe *-Fe >" pairs in the trimer Fe=*—Fe =" pairs in the helix Structural Effects Summary References Chapter 3 Fe**-Fet* Interactions in Tourmaline . Introduction Experimental Results and Discussion Intensity of Fe2+ absorption in Fe**—Fe "groups Diagnostic Attributes of Intensified Fe=* Transitions Fea =*Fe** Charge Transfer in Tourmaline Site Assignments of Fe=* Absorption Application to Radiation Effects Summary References Chapter 4 Heteronuclear Charge Transfer in Minerals Introduction Experimental Results Taramellite Neptunite Traskite Tourmaline Mn#*-Tit* Charge Transfer Discussion References Chapter S Characteristics of Intervaience Charge Transfer in Minerals introduction Experimental Results and Discussion Charge Transfer Eneroqies i 7 9 10 * ~~ i7 i? 20 24 26 35 38 41 43 48 49 52 57 63 69 74 7& 79 83 835 8° 90 92 95 99 101 163 i168 iii 116 120 i2ti 134 Effect of Temperature on Intensity izé x Widths of Charge Transfer ; Transitions Summary References | Chapter & Summary Appendix GH Vibrations as a Chemical Probe in Tourmaline is? 140 142 146 isi Chapter lL Introduction = » ‘Optical absorption bands of transition element ions are often used to determine the concentration of these ions in Solution. The extension of such analytical capabilities to solids, in particular to minerais, has met with limited success. Difficulties have arisen in part because transition elements are often partitioned into multiple sites in minerals which have different optical properties which complicate the analysis of optical data from which more information (i.e. site ecccupancy) is required. When the sites are very different, as in orthopyroxenes (Goldman and Rossman 1979), quantitative site occupancies may be determinable. More significant problems result from the complicated chemistries of natural materials and the possibility of interactions among cations. Dramatic color changes which attended the mixing of transition element cations of differing valence were the first indications of the properties associated with multiple cations. These color changes were later attributed to optical transitions which involved the transfer of electrons between neighboring cations. These intervalence charge transfer transitions were investigated extensively in chemical systems (Robin and Day i967; Allen and Hush 19473. Their importance to the fields of physics, biology and geology has been more gradually recognized (Brown i980). The initial emphasis on systems with highiy controlled chemistries produced a large amount of data on charge transfer between cations of the same element which differed = >: in their oxidation states. The most important mineralogical example of this group is Fe**-Fe** charge transfer. Charge . transfer between cations of different elements and differant valences received very little attention. Transitions which occurred in Ti-bearing minerals were often assigned to Ti**-Ti** charge transfer without consideration of more abundant Fe2+-Ti++ ion-pairs (e.g. in tourmaline, Faye et al. 19468). Although these errors have generally been rectified, there are very few unequivocal examples of Fe?=*—Ti** charge transfer. Burns (1981) has reviewed Current assignments of intervalence charge transfer transitions in Fe,Ti-minerals. ‘In a few cases, simple assignments of transitions to Fe?*-Fe®* vs Fe2+-Ti4+ charge transfer is not yet possibie. The amount of information available through the analysis ef intervalence charge transfer transitions is potentially greater than that available from crystal field transitions. Site occupancies of cations can possibly be determined by intensity ratios in various polarizations according to the polarization of charge transfer transitions along the vector between the interacting cations. The involvement of multiple cations suggests that the evaluation of preferential ordering of cations should also be possible. The large relative intensities of charge transfer transitions indicate their potential as sensitive probes of cations which are weakiy absorbing, such as Fe ®t, or which do not exhibit electronic transitions in the visible or 4 n@ar~infrared, such as Ti**. These applications, however, require quantitative measures of absorbance vs concentration ; which are generally known to orders of magnitude accuracy if at all in minerals. Quantitative studies have been hampered by the difficulty of producing synthetic crystals suitable for optical studies, the complex chemistries of many minerals, and experimental difficulties associated with preparing sections which transmit enough light far the most concentrated systems. The last of these was recently. surmounted, permitting the evaluation of quantitative Fe?*-Fet* charge transfer intensities in several minerals (Amthauer and Rossman 1984). In addition to producing intervaience charge transfer transitions which cannot be generated by the superpositisan of the characteristic absorptions of the individual cations, ion-pair interactions can also produce changes in the Characteristic absorption of individual cations. This effect was first noted for groups of Mn=* ions tLlohr and McClure 1958) and has since been found in Fe=* clusters (Murray 1974). The normally weak spin-forbidden bands of these ions can be intensified by two orders of magnitude through antiferromagnetic exchange. It is generally believed that this intensification occurs through a circumvention of the spin-selection rules via simuitaneous transitions in coupled icon-pairs. The most common mineralogical examples of this effect are found in iron sulfates (Rossman 1975). Smith (1978) proposed that the spin-allowed bands of Fe2* ions were intensified through an interaction with Fe>* ions. Aithough this intensification cannot be justified on theoretical grounds, the behavior of these transitions, particularly with regard to temperature response, is also inconsistent with the current theory of crystal field transitions. The magnitude of the intensity increase was not evaluated. The utility of intensified Fe2* transitions as probes of Fet* ions is greatly dependent on the magnitude of the intensity increase. However, the possible corruption of expected concentration vs intensity correlations of transitions in which this effect is not recognized are significant even at fairly low increases. The basic requirement of all of these transitions is that transition element ion-pairs involve next—-nearest neighbor cations which share one to three nearest neighbor anions. Transition elements in minerals often occupy edge-sharing octahedra in clusters from dimers up to chains and sheets of octahedra. Several common minerals combine these structures with complex chemistries to produce spectra that consist of multiple ion-pair transitions as well as transitions of single ions. Reliable characteristics which can be used to differentiate among these transitions are essential for the interpretation of such spectra but are not yet available for all three types of ion-pair transitions. The purposes of this study were twofold, to srovide reliable characteristics of ion-pair transitions in minerais é >» such that assignments of transitions can be more certain, and to evaluate quantitative absorption intensities such that concentrations and site Occupancies can be evaluated from. intensities of ion-pair transitions. The study of intensified Fe@* transitions was focused upon tourmalines (Chapter 3) because of their relative abundance and wide range of compositions among the minerais discussed by Smith (1978). However, the chemical complexity of tourmalines required a fairly complete understanding of the spectroscopy of Fe-bearing tourmalines. Chapter 2 presents the spectroscopy of Fe =* in tourmaline and Fe**-Ti** and Mn2*+-Ti4*+ charge transitions in tourmaline are included in Chapter 4. Stoichiometric Fe,Ti-minerals are emphasized in the study of Fe@*-Ti** charge transfer (Chapter 4) because of the possibility of quantitative as well as qualitative evaluations. The characteristics of Fe**-Fe** charge transfer are reexamined in Chapter 5 to evaluate theoretical and experimental discrepancies and to incorporate more recent experimental data. Chapter 46 concludes the main body of the thesis with a summary discussion. 4 short section an the interpretation of OH vibrational svertones in tourmalines is presented in the appendix. F REFERENCES Alien, BC and Hush, NS (1967) Intervalence transfer absorption. Part i. Qualitative evidence for intervalence-transfer absorption in inorganic systems in solution and in the solid state. Frog Inorg Chem 8:257- 389 Amthauer , G and Rossman, GR (1984) Mixed valence of iron in minerals with cation clusters. Fhys Chem Minerals Llis27-5Si Brown, DB, ed. (1980) Mixed Valence Compounds. Theory and Applications in Chemistry, Physics, Geology, and Biology. Holland: Reidel, Si? pp. Burns, RG. (1981) Intervalence transitions in mixed-valence minerals of iron and titanium. Ann Rev Earth Flanet Sci 93 345-385 Faye, GH, Manning, PG and Nickel, EH (1968) An interpretation of the polarized optical absorption spectra of tourmaline, cordierite, chloritocid and vivianite. Am Mineral 53:1174-i201 Goldman, DG and Rossman, GR (1979) Determination of quantitative cation distribution in orthopyroxenes from electronic absorption spectra. Phys Chem Minerais 4:43-53 Lohr, LL Jr and McClure, DS (1948) Optical spectra of divalent manganese salts. II. The effect of interionic coupling on absorption strength. dg Chem Phys 49:3514- soel 8 Murray, KS (1974) Binuclear oxo-bridged iron (IIT) complexes. Coord Chem Rev i2:1-35 “Robin, MB and Day, P (1947) Mixed vaience chemistry—~a_ survey and classification. Adv Inorg Chem Radiochem 10: 248-422 Rossman, GR (1975) Spectroscopic and magnetic studies of ferric iron hydroxy sulfates: intensification of color in ferric iron clusters bridged by Aa Single hvdroxide ion. Am Mineral 460:498-704 Smith, G (19798) Evidence for absorption by exchange-coupled Fe**-Fes* pairs in the near infra-red spectra of minerals. Phys Chem Minerals 3:375-383 Chapter 2 Fest - Fe>* Interactions in Tourmaline (Accepted for publication in Physics and Chemistry of Minerals George Rossman coauthor) 10 Introduction 3+ The presence of Fe~ in minerals is usually easily determined by '- its. characteristic optical absorption bands. The consistency of these features in many silicates has made quite puzzling their apparent absence in tourmaline and micas. Fe? has a large role in the optical spectroscopy of tourmaline through its interaction with Fe2* (Smith 1978; Faye et al. 1974; Wilkins et al. 1969). Some natural samples have 3+ been shown to contain substantial amounts of Fe by other methods (Hermon et al. 1973; Saegusa et al. 1979; Walenta and Dunn 1979). +, . . Fe? interactions in 3 However, in spite of the prevalence of Fe2t- tourmaline, consistent independent optical evidence of Fe + has not been 3 recognized. Even in buergerite, nearly the Fe *-tourmaline end-~member , only Fe2t features were recognized (Manning 1969). The discrimination 3 of Fe?* bands in tourmaline has been hampered both by a complex composition in which other chromophoric ions are generally present and 3+ by low levels of Fe in the gemmy samples which have been preferentially studied. Spectral features have been previously assigned 3+ in tourmaline by Manning (1969) and Smith (1978). However, to Fe these assignments refer to different features. Furthermore, bands at these positions are rarely encountered in tourmaline. Unusual red tourmalines from Kenya were reported to contain only trace amounts of transition elements other than iron (Bank 1974; Dunn et al. 1975). Our preliminary study of these samples indicated a very low Fe**/Fe3t ratio 3+ optical features. and the potential to isolate Fe In addition to features of individual cations, in minerals such as tourmaline, neighboring transition elements interact to produce optical features which are more intense than those of isolated ions. These 11 transitions frequently dominate the spectra of minerals in which the sites occupied by transition elements are linked directly to similar ‘sites. In tourmaline, interactions among cation pairs can occur in 3 possible site combinations (Buerger et al. 1962). They are highly favored by edge-shared connections and short next-nearest neighbor distances (295-305 pm). Transition elements are known to substitute into two 6-coordinate sites which form helices (Z-site) and trimers (Y-site) (Fig. 1). The linkage of these groups provides pairs within the trimer, within the helix and in combination. The site occupancy of Fe*t is potentially quite variable as evidenced by the existence of two 3+ Fe~ end-members: buergerite, NaFe3*Al,S$i,0, 9 (B03) (00H) (Donnay et al. 1966) and ferridravite, NaMg ,Fe?*Si,0,9(BO4) (0H), (Walenta and Dunn 1974). Fe?* site occupancy in tourmaline has been firmly established only in buergerite by X-ray structure refinement (Barton 1969). Structure refinements of more typical tourmalines have suffered from the 2 inability to differentiate Fe * from Fe* (Fortier and Donnay 1975). Méssbauer studies of Fe>t in tourmaline have suffered from the general 2 3+ dominance of Fe* peaks as well as the relative insensitivity of Fe peaks to site geometry. Optical transitions of multiple cations provide another. teasure of site occupancy which will be explored here. The spectroscopy and color of most tourmalines are dominated by an 2+ 3+ interaction between Fe~ and Fe~ which intensifies Fe2* absorption bands (Smith 1978). Although this interaction occurs in other minerals, in tourmaline it is particularly pronounced. The expected signature of : . + such interactions, Fe" -Fe?t intervalence charge transfer, has been assigned to a comparatively weak feature. These characteristics of ion- pair interactions can provide a sensitive probe of oxidation states and 12 “S878 osayy Suowe suozjoeuucs ayQ pue andv0 sjususya uoFIFsuery YOTYM UT saqys ayQ sq pouws0zy xTTey pue Jeu f3y 34} ButMoys SANGONAS SUFTewWINO}Z |yz JO uot Aod e jo uoTIefoad (100) °] *3Ta D site occupancies of Fe cations in minerals, but they must be understood in regard.to the factors which govern their intensity. Tourmaline 2+_y 3+ ‘provides an extreme version of Fe e~ interactions and furnishes the opportunity to study the factors involved in these effects. The 3+ recognition and understanding of the optical properties of Fe is essential to this study. Experimental Sample localities are presented in Table 1. Chemical analyses were obtained using a MAC5-SA5 electron microprobe. Data reduction was accomplished using the methods of Bence and Albee (1968). Partial sample compositions are shown in Table 2. Formula proportions were calculated using Mg + Fe + Ti + Mn + Al + Si = 15. Magnetic susceptibility measurements were taken on a S.H.E. model VTS magnetometer/susceptometer. The raw data were corrected for the Al sample container and fit to combined paramagnetic and antiferromagnetically-coupled components. ‘The exchange-coupled components were modelled as 'dimers' and ‘trimers' of equivalent ions interacting isotropically through Heisenberg intracluster exchange (O'Connor 1982); g=2.0 was assumed for all groups. The magnetic moment 3+ of Fe” was assumed to be the spin-only value of 5.92 B.M. which is consistent among similar high-spin compounds (Kénig 1966). Because the +. : . : : . 2 is somewhat more variable in oxide coordination, magnetic moment of Fe this parameter was varied between 4.9 and 5.4 B.M.. The Méssbauer spectra of sample 1 were taken by Georg Amthauer using conventional techniques described in Amthauer et al. (1976). The mirror symmetric, simultaneously recorded spectra (512 channels each) 14 Table 1. Sample Descriptions Sample # | Mineral Color? Locality Archival Code Name 1 Dravite red Osarara, Narok District NMNH # 126030 , Kenya 2 Dravite red Kenya CIT # 13764 3 Uvite orange- Brumado, Bahia, Brazil CIT # 13765 red 4 Dravite red Diamantina area NMNH # 137101 Minas Gerais, Brazil 5 Dravite red Chipata, Zambia CIT # 15053 6 Schorl yellow? Schindler Claim, Cahuilla Mt. CIT # 13763 Southern California 7 Uvite green® Pierrepont, New York NMNH # 81511 8 Schorl green® Santa Cruz, Sonora, Mexico NMNH # R15504 9 Dravite brown® Madagascar Harvard # 108796 10 Dravite brown’ Syndenham, Ontario CIT # 2884 11 Buergerite brown Mexquitic, Mexico CIT # 8207 12 Coalingite yellow Near Dallas Gem Mine CIT # 7298 San Benito Co., California NMNH: U.S. National Museum of Natural History CIT: Calfornia Institute of Technology a Macroscopic color unless otherwise indicated b color of oxidized sample ¢ macroscopically black, color quoted is Ele color in thin section 15 Table 2. Electron Microprobe Partial Analyses (wt.Z%) Sample 1 2 35 4 5 6 7 8 9 10 Na,0 2.72 1.98 0.80 1.77 2233 1.34 1.44 1.86 1.67 1.58 Cad 0.06 0.84 3.69 1.64 0.14 0.05 2.92 1.98 2.42 2.61 K,0 0.07 0.07 0.03 0.11 0.02 0.05 0.05 0.05 0.08 0.04 MgO 9.45 9.46 12.8 10.1 10.0 0.48 10.8 6.48 8.58 10.9 Feo? 4.32 2.10 1.93 4.46 2.06 13.3 8.23 18.1 13.4 7.84 MnO tr 0.12 tr 0.22 - 0.36 - tr tr tr Tid, 0.27 0.20 0.58 0.22 0.18 0.05 0.55 1.18 1.69 0.72 A193 32.0 33.9 29.3 31.0 33.9 35.3 26.3 22.2 23.6 26.5 Sid, 36.7 37.1 36.7 36.1 36.7 35.5 35.2 34.0 34.4 35.3 F tr tr 0.1 0.1 - - 1.6 0.5 1.1 1.6 Formula Proportions Na 0.88 0.62 0.24 0.56 0.72 0.44 0.47 0.63 0.56 0.51 Ca 0.01 0.14 0.64 0.28 0.03 0.01 0.52 0.37 0.44 0.46 K 0.02 0.01 0.01 0.02 0 0.01 0.01 0.01 0.02 0.01 Mg 2.27 2.27 3.09 2.46 2.40 0.12 2.70 1.69 2.20 2.72 Fe 0.58 0.28 0.26 0.60 0.28 1.87 1.15 2.64 1.92 1.09 (Fet*ye (0.53) (0.25) (0.23) (0.53) (0.26) (1.87) (0.55) (0.93)4 (1.19) (0.59) Mn 0 0.02 0 0.03 0 0.05 0 0.01 0 0 Ti ; 0.03 0.02 0.07 0.03 0.02 0.01 0.07 0.15 0.22 0.09 Al 6.12 6.44 5.60 5.97 6.40 7.00 5.18 4.56 4.76 5422 Si 5.99 5.97 5.96 5.90 5.89 5.96 5-91 5.94 5.90 5.89 F 0 ie) 0.1 0 0 QO. 0.8 0.3 0.6 0.8 2 total Fe > contains trace amounts of V or Cr ¢ calculated using calibrated optical absorption except where noted minimum value based on charge balance in the absence of H-defects 16 were folded and: fit using a modified version of 'MOSFIT' (Kulscar, Nagy, and Verbiest--Leuven, Belgium). The fitting was constrained to ‘symmetric doublets having the same isomer shift. These constraints were employed because the outer doublet is poorly constrained by the spectra. No significant improvement or difference in the fits was obtained by independent isomer shifts. One-doublet fits required unusually large quadratic backgrounds and were 1.5-2 times poorer fits as measured by chi-squared. The final parameters are shown in Table 4. The isomer shift is quoted relative to sodium nitroprusside. A Méssbauer spectrum of sample 3 was obtained by W. Dollase at UCLA and was similar to that of sample 1 at 298 K but with a lower signal to noise ratio. Independent Méssbauer data on sample 7 indicates ~60% Fe-t (RG Burns personal communication). Optical spectra were obtained in a manner described in Amthauer and Rossman (1984). In addition to Méssbauer data for samples 1, 3, and 7, calibrated Fe? absorption (Chapter 3) was used to determine the relative amounts of ferrous and ferric iron. The assignment of various transitions to ferric vs. ferrous iron was based upon spectra of oxidized and reduced samples. Sample 1 was oxidized by heating in air at 1073 K for 15 and 40 hours and sample 6 for 7 hours. Oxidation via this treatment has been confirmed by Méssbauer studies (Gorelikova et al. 1976). Reduction of sample 1 was accomplished by heating at 1073 K for 8 hours in a H,/CO. mixture set at the iron/wiistite buffer. Molar absorptivities are given in units of liters/mole/cm (wtem!). ‘17 Results and Discussion Magnetic Moment The magnetic moment of solids and its variation with temperature provide information regarding oxidation state and the extent of interaction among paramagnetic ions. In materials such as the tourmalines .of this study the degree of ordering can also be indicated. The temperature dependence of the magnetic moment of sample 1 is shown in Fig. 2 together with the calculated values. The decreasing moment with decreasing temperature indicates a small antiferromagnetic interaction. The solid line represents a calculation for a combination of the most probable components based on stoichiometry and the simplest model: non-interacting ferrous and ferric iron, two populations of ferric iron 'dimers' and a population of ferric iron 'trimers'. The parameters of this calculation are presented in Table 3. This distribution is in rough agreement with a random distribution 3+ of Fe within the trimer -- 67% Fe in Fe?*X,, 30% in Fe3*x, and 3% in Fe3* (X: other cations in the trimer). A small proportion of iron in the helix would decrease the percentage of 'dimer' and 'trimer' leading to somewhat better agreement with the data. Preferential association of Feo* is. inconsistent with the observations. The strength of the 3+ interaction between adjacent Fe ions is indicated by the coupling constant, J. Values of -14 and ~14.5 em! 1 -are slightly larger than edge-shared dimers (J = -8 to -ll cm“, Schugar et al. 1969; Wu et al. 1972) and roughly one half of that for the trimeric iron(III1) carboxylates (Long et al. 1973; Earnshaw et al. 1966). The coupling constants are thus consistent with any of the 3 edge-shared pairs, and the agreement of the calculated distribution within the trimer with the 18 OSE "E aTqey uy pequasaad szoqaweaed Syd aojy woOPReTNI{eI e& st SUIT PFTOS ayy, ‘e2ep psaraasqo ayy yNaseadas saqzucqoay +1 otdues JO JUoMIOW DzyIoUTEW aATIDazZJa ayy JO aouspuadap aanjeirsdway +z -Ryg CoD eUnzouedue |, OBE SS2 8O2 BSI ee! as oO ' v Li t v 1 a: 9 a: i 9 CW’ GD FUSWOL > 1zSUBLDLY, d- 19 Table 3. Calculated Magnetic Parameters for Sample 1 Percentage upe > (BeM.) = (em) P 3+ | c Non-interacting Fe 55 5.92 Non-interacting Fe2t 7.5 5.2 Binuclear Fe-* cluster #1 20 -14.5 Binuclear Fe°* cluster #2 15 -1.25 Trinuclear Fe?* cluster 2.5 14 a magnetic moment in Bohr magnetons (assumed standard values 298 K) coupling constant for magnetic exchange Table 4. Méssbauer Parameters for Sample 1 Temperature I.S.4 g.s.? Halfwidth Area® Xy2 (K) (mm/sec) (mm/sec) (mm/sec) 298 0.62 0.76, 0.56 1.45 1.04 0.62 1.67 1.59 0.84 80 0.73 0.74 0.44 256 1.27 ‘ 0.73 1.66 1.70 1.20 5 0.74 0.76 0.37 1.63 1.07 0.74 2.01 0.73 * isomer shift (I.8.) is quoted relative to NajFe(CN) NOs 2H,0 b quadrupole splitting 2 Xy (chi-squared)/(number of degrees of freedom) ¢c area is normalized to background 20 fit parameters supports a majority of Y-Y pairs. Previous studies of Fe-rich tourmalines indicated J ~ -6 emt (Tsang et al. 1971). The “weakly interacting population (J = -1.25 cm, Table 3) was required to fit the low temperature region. It probably corresponds to interaction among Fest ions with one intervening diamagnetic ion (e.g. in the helix). The non-interacting ions contribute a constant background to the observations which does not strongly constrain the relative percentages of non-interacting ferrous and ferric iron. However, the proportions of coupled versus non-interacting ions are well constrained. Méssbauer Spectroscopy The most striking feature of the Méssbauer spectrum of sample 1 is its unusual temperature dependence (Fig. 3). At 298 K the halfwidth of the doublet is similar to those of glasses (Kurkjian and Buchanan 1964), while at 5 K it is within the normal range for crystalline material. The area of this doublet is roughly constant with temperature. The overall broad, non-Lorentzian peak shape is similar to partially relaxed hyperfine structure (Wignall 1966) and complex compositional-structural intermediates (Dollase and Gustafson 1982). The major cause of these peak shapes can be discerned from the observed temperature dependence. Since the quadrupole splitting of ferric iron is normally independent of temperature, in contrast to Fe (Greenwood and Gibb 1971), the coalescence of doublets from multiple environments is unlikely to produce the decreasing peak widths. The major cause of the broad non- Lorentzian shape, therefore, cannot be ascribed to compositional complexity. Relaxational processes can also affect the peak shape of 3+ a3t Méssbauer absorption. Fe~ may relax by spin-spin relaxation among Fe 24 PERCENT ABSORPTION *una330ds ¥ g6zZ BY} OJ uMOYS sz syUeUCdMOD pue wnijoeds pajetnoteo ¥TqNOp 7 ‘37J-3s8aq y “ap Tssnido1zy yu unypos 03 BATIETSI uMOYsS st AYPIOTIA PeIeoOTpuy saanjersdway snopaea ye T 2Tdwes Jo e1aqoeds araneqssoy *tinajoeds yoes Jo 3y3TaA reMoOT ay} 09 "€ °3ha 00 0 & bo tite 33 fom Me oo a6 200 o o@ © w oe woe oo ng? 4 88 ¥ ¥ ¥ ¥ T A 862 | I- 2 ¢- — (98 7HH) ALTIOTAA © (998 /at) ALTIOTIA e ~- _ 28/88) ALTIOTIA ¢<- £- NOTLdYOS¥ LN30u3d 22 + : . 2 which in turn relaxes more ' groups or by. cross. relaxation with Fe rapidly by spin-lattice relaxation. The width of broad bands generated by spin-spin relaxation alone is independent of temperature (Wignall 1966). Therefore, we propose that the large decrease in width can be explained by cross relaxation enhanced by greater interaction among ferric and ferrous ions at lower temperatures. Supporting evidence for 2+ this is provided by increased intensity of Fe optical absorption bands at low temperatures which are intensified by interaction with Fe?t (Chapter 3). These spectra were fit to symmetric lorentzian doublets to compare to previous data and to determine general absorption characteristics. The best fit parameters are presented in Table 4 and the calculated peaks are indicated for the 298 K spectrum in Fig. 3. The increase in isomer shift with temperature is in agreement with that expected from the second-order Doppler shift (Greenwood and Gibb 1971). These calculated doublets represent overlapping contributions from non- lorentzian peaks with broad tails due to relaxation effects, as well as 2+ Fe>t in multiple sites and a small amount of Fe” . The isomer shift (IS) and quadrupole splitting (QS) of both doublets are consistent with ferric iron in an octahedral site. The inner doublet accounts for most of the absorption. Its halfwidth at 5 K corresponds to that of a single, well-defined site and points to the preferential occupancy of either the Y or Z site. Its IS and QS are in reasonable agreement with an approximate fit to an oxidized Fe?*-tourmaline (Gorelikova et al. 1976) and the general 3+ 2+ characteristics of Fe~ in Fe“ -rich tourmalines (Saegusa et al. 1979; Hermon et al. 1973; Gorelikova et al. 1976). The apparent discrepancy 23 of the IS with the. major doublet in buergerite (Hermon et al. 1973) was not substantiated by an examination of the published spectrum, nor by ‘independent data (IS = 0.61, QS = 1.12 mm/sec, RG Burns, personal communication). The outer doublet represents the non-lorentzian character of the inner doublet as well as the other components described above. This. complexity -is indicated by a halfwidth which is ~3.5 times greater than that generally observed for single components in 2+ erystalline phases. The maximum amount of Fe estimated from the intensity in the wings of the 298 K spectrum at a position 2+ characteristic of Fe in tourmaline and a typical halfwidth is 13 3+ percent. The symmetry of these spectra about the Fe isomer shift would argue against even this much Fe*t, This compares to ~9% indicated by optical data. A comparison of sample 1 to buergerite will be used to establish which site corresponds to the inner doublet. X-ray structure refinement has established that ~90% of the Fe? in buergerite is in the Y-site (Barton 1969). The isomer shift of the major doublet agrees with sample 1 within experimental error. The quadrupole splitting, however, is ~0.3 mm/sec larger in buergerite. Larger quadrupole splittings in est compounds correspond to greater distortion and/or smaller sites. Since a structure refinement is not available for this sample, the buergerite structure will be compared to the Mg-tourmaline structure of 3+ Buerger et al. (1962). While the Fe~ environment in sample 1 will not be precisely that of Mg, it should be intermediate between the Mg-site and the Feo+ ~site in buergerite. The Y-site in Mg-tourmaline is not only larger than in buergerite, it is also less distorted as measured by the range in Y-O bond lengths. Although the Z-site in Mg-tourmaline is aly also less distorted, it is smailer than the Y-site in buergerite. These 3+ prefers the Y-site in this sample. The 2+ considerations suggest that Fe ‘maxi mum possible proportion of Fe in the Z-site can be estimated from a doublet area corresponding to the height of the outer doublet at 5 K and a typical halfwidth. This calculation leads to a maximum Fe>* in the Z~site of ~10% or 0.05 per formula unit. Optical Spectroscopy The optical spectrum of sample 1 is illustrated in Fig. 4 and is typical of samples 1-5. The broad bands at ~700 nm and ~1100 nm are produced by electronic transitions of ferrous ions and indicate ~0.30 wt% FeO for this sample (Chapter 3). Vibrational overtones of hydroxyl groups generate the weak, sharp bands between 1350 and 1500 nm. These features will not be discussed further. The other features of this spectrum have not been recognized in previous studies of tourmaline and cannot be assigned by analogy to other silicate minerals. The 425 nm 3+ band Manning (1969) assigned to Fe~’ is not evident and is more likely 2+_ pat charge transfer (Chapter 4). The 485 mm (ELc) 3+ associated with Fe and 540 nm (Elle) bands increase in intensity with increasing Fe~" and are assigned to Fe", The unusual red color of samples 1-5 results from 3+ the dominance of these Fe transitions. Unlike most tourmalines, the iron in these samples is at least 904 ferric iron. These samples thus provide an ideal opportunity to observe the optical spectroscopy of Fe>* in tourmaline. Transitions among the electronic levels of Feot generally produce 3 sets of bands in the visible and near infrared. Two of these are broad bands (halfwidth ~2500 em |) in the regions 750-1000 nm and 550-700 nm 25 ; nae Se ae ee a a a a ~ = . . Hf 7 a | <j Wu TIE°QO wssauyTYI aTdues JH r9UuTT PETOS 3TH 2BUTT ueyorg *] atTdwes jo eaqoeds uotydiosqe pazzaejod ainqjersdway wooy QGBSI Bbc} wu “HLONATSAVM 81! 626 B@2 "y °3Ta QoE v i A. ae ee ee Ss Ss Ss Se ee Oe Gs Se SO Ge GO OS SO Ge Re ND ! BOOZL |—- WD [J 8eeel “YAGWNNAAVM Q: ra SINVEedOSEv 26 for octahedral coordination. The most distinctive Feo* feature is a sharp band or bands in the region 410-450 nm with a halfwidth of ~500 em), This typical pattern is not observed in any of these samples or in any published spectrum of tourmaline. In addition to unusual band positions and widths, the transitions observed are also much more 3 intense than‘’simple Fe + electronic transitions. Also, the intensities of the bands do not increase linearly with increasing Fe, but with the 3+ proportion of Fe-* having Fe~ neighbors (Fig. 5). Thus, all of the optical features associated with Fe>t 3+ in these tourmalines are assigned to simultaneous transitions of Fe clusters which are antiferro- magnetically coupled. Such transitions were first described for Mn2* compounds which are isoelectronic with Feot (Lohr and McClure 1968). These transitions occur at the same wavelength as those of isolated ions but with much higher intensities. The intensification occurs in the direction corresponding to the vector between the interacting cations. 'Ion-pair' transitions involving Fe?* have been documented in sapphire (Krebs and Maisch 1971) and various sulfates (Rossman 1975, 1976a). Based upon the polarization of the bands observed in samples 1-5, ion- pairs within the trimer and those within the helix can be distinguished. 3+ Fe? Fe + pairs in the trimer 3+ Ion-pair transitions among Fe~ groups within the trimer are represented by the absorption bands at 485 nm and ~530 nm in the 3+ is supported by direction Elc. The assignment of these bands to Fe decreasing intensity upon partial reduction of the iron in sample 1. No significant change in intensity was observed upon oxidation which removed the Fe** features at ~700 and 1100 nm. The assignment of these 27 Absorbance/cm Concentration Fig. 5. Absorbance/cm of the 485 mm band (Elc) vs concentration in 3+ and +: Fes? in Fe3* X and Fes", A moles/liter of [[: total Fe Beer's law best fit is shown for interacting Fe?*(+). The solid line associated with total Feo* approximates the deviation from _ Beer's law. 28 3+ bands to Y-Y¥ Fe pairs was based upon their polarization along Feot-pe>t vectors in Elc and the dominance of Y-site occupancy indicated by the Méssbauer spectrum of sample 1. The 485 and 530 nm bands were also generated in Sample 6 upon complete oxidation. Sample 6, which is 2+ an Al-rich tourmaline originally containing predominantly Fe” , undoubtedly contains most of its iron in the trimer and shows only these features. The minimum molar absorptivity of the 485 nm band based on 3+ which ranges from 14 to 31 M lem! is greater than an order 3+ total Fe of magnitude more intense than the analogous bands of isolated Fe and points to the influence of exchange coupled groups. Molar 3+ absorptivities for samples 1-5 based on the amount of Fe~ in ‘'dimers' and 'trimers' calculated assuming a random distribution within the 1 3+ Y-site yield a value of ~90 Mom! This fraction of Fe°* adjacent to in Y-Y pairs based on optically determined Fe ratios is in close agreement with the distribution indicated by the magnetic data for 1 sample 1. The molar absorptivity of these pairs (90 M em!) is in agreement with molar absorptivities found for antiferromagnetically- coupled Fe-*, ~30-100 M!cem! (Rossman 1975, 1976a, 1976b). Samples 1-5 exhibit a red color which is similar to the color of 2+ _ py At minerals in which Fe charge transfer (CT) is the dominant optical feature, such as fassaite from ADOR (Mao et al. 1977). In tourmaline, Fe2*-t44* CT has been assigned to a broad band in the region 400-460 nm ELc (Smith 1978; Faye et al. 1974). An assignment of Fe2t-74 4+ CT to the 485 nm band was: rejected on the basis of 3 factors. The intensity At ‘ 9 does not correlate with Ti and Fest contents. The halfwidth (~1600 2+_pp4t cm) is much smaller than that of Fe CT in tourmaline ~ (~8000 em}, Chapter 4). The calculated maximum possible intensity 29 2+ ms 4+ 1 -Ti 2 . associated with Fe content is ~l cm or ~0.3 in Fig. 3 (Chapter 4). Thus, although Fe t+ipy4t CT may contribute to the spectrum in this region, it will only raise the apparent background. Additdonal electronic transitions occur at 360, 385, 400, and 440 nm at 296 K. Sample 4, which was also examined in this region, showed similar bands. These transitions can be distinguished from ligand to metal charge transfer bands because of their relatively low _ energy (Lehmann 1970). The differing temperature responses of these bands indicate different types of simultaneous transitions within Fe? groups. Simultaneous transitions in which electronic transitions occur in both ions have greater intensities at low temperatures and are known as 'cold' bands. These 2-exciton bands occur at the combined frequency of both electronic transitions. The 360 nm band can be assigned to a ‘cold! band. It increases in intensity by a factor of ~3 at 80 K and is ep ) + at an energy appropriate for the combination Ca, > Ca, > “I, The energy of these individual transitions are not precisely known, but the ~530 nm band Pa, > "t,) would require that the wavelength of the 8, > “r, band be higher than 1000 nm. A weak, broad band at ~1000 nm is indicated in the spectra of the oxidized sample 6. The increase in intensity of the 360 nm band together with a shift of band maximum to lower energy and the shift of the 485 nm band to higher energy constitute the major changes illustrated in Fig. 6. Simultaneous transitions which involve an electronic transition for one atom and a spin-flip for another are termed exciton-magnon transitions. These bands are known as ‘hot! bands because of their greater intensity at high temperatures (Murray 1974; Fujiwara et al. 1972). They occur at “the same energy as the electronic transitions of isolated ions. WAVENUMBER, cm | 3008 «= 25000 20008 «=s«1 7000 | T 7 7 T | 7 T —t- ABSORBANCE 356 400 458-508 550800 | WAVELENGTH, nm Fig. 6. Elec polarized absorption spectra of sample l. Broken line: 296 K Solid line: 83 K Sample thickness: 0.021 mm 3h Although such considerations are complicated by energy shifts and splittings with temperature and contributions from Feo* pairs in the “helix, the bands at 385, 400, 440 and 485 nm appear to be thot! bands. The band at 440 nm may also be a 'hot' band or it may be the Pa, > (ta, .°8) band of isolated Fe-* . Two factors which deserve further attention are noted here. First, the energy shift of the 360 nm and 485 nm bands are not typical of the limited data on other Fe?t systems. Second, the general behavior of the Feot pair transitions which is that the "A, > 4ep and rN > (4a 5) 1 l have the highest intensity of the three lowest energy bands is not observed. The a > “A transition Y-Y pairs in this case has a very low intensity and may not be intensified at all. An evaluation of the degree of noncompliance of these pairs requires a larger data base particularly in silicates and a more thorough theoretical description of environmental factors. These considerations require the assignment of the 485 nm band to the °A + Ca "BD transition. This is an unusually low energy for this transition. It also exhibits a mich broader width than normal. Low energies have been observed in epidote (Burns and Strens 1967) and appear to result from extreme distortion. The Y-site of tourmaline is only moderately distorted, however. There are a few other compounds which also exhibit such low energies. A relevant comparison is provided by the mineral coalingite, Mg pFe7*, (0H) 5,004°2H50, in which the cations occupy edge-sharing octahedral sheets (Pastor-Rodriguez and Taylor 1971). The major Feet transition is observed as a broad band at ~475 nm with a shoulder at ~409 nm (Fig. 7). Exchange interaction between large sites with edge-shared connections may produce the low energies observed ABSORBANCE 32 | WAVENUMBER, cn | 25000 ~«©=—ss«(2000 (5000 | | 358 459 558 658 WAVELENGTH, nm Fig. 7. Unpolarized room temperature absorption spectrum of coalingite (sample 12). The feature at 490 nm has a molar absorptivity of ~75 Leg t 3+ “ M™*em™*+ which is much higher than those of isolated Fe ions and can thus be associated with ion-pair transitions. Sample thickness: 7.8 um 33 in tourmaline and coalingite. This is also supported by a lowering by ~30 nm of the °8 > Ca "8 transition for edge-shared units in the sulfate. amarantite and the phosphates leucophosphite (Rossman 1976b) and pharmacosiderite (446.nm, Rossman unpublished results) as compared to 3+ > isolated Fe?* (Lehmann 1970). A contribution from tetrahedral Fe which also has bands in this region, can be discounted because of the lack of a Si-deficiency and the mismatch of the Méssbauer features with tetrahedral’ Fe?* (Waychunas and Rossman 1983). 3+ 3+ Fe~ -Fe pairs in the helix ITon~-pair transitions among Fe?* groups within the helix are represented by the bands at 540 and 425 nm. These transitions are illustrated in Fig. 8 for the polarization Ellc at 296 K and 83 K. The assignment of these features to Fe?t is based upon their response to partial oxidation and reduction which was similar to the 485 nm (Elc) band. This behavior eliminates Fe2t/past charge transfer from the assignment possibilities. A significant contribution from vn3* at 540 nm can be dismissed because of the temperature response of these bands. Since the iron content of the helix and the degree of ordering is not known, the molar absorptivities of these bands cannot be evaluated. However, an upper limit of ~10Z Fe>* in the Z-site can be estimated from the 5 K Méssbauer spectrum assuming a halfwidth of 0.5 mm/sec. This leads to an estimate of the minimum value for the molar absorptivity of ~50 whem! for the 540 nm band. This value is 2 orders of magnitude greater than that for Feot bands of isolated ions and indicates the intensification associated with exchange coupled pairs. The intensity of these bands vary independently of the 485 nm band in Ele associated with the trimer groups. These bands can be attributed to 34 WAVENUMBER, cn”! 20088 20088 15888 | | I a 4 a cf t T ABSORBANCE cS ow FF fF ttt 350 459 550 658 is | WAVELENGTH, nm Fig. 8. Elle polarized absorption spectrum of sample 1. Broken line: 296 K Solid line: 83 K Sample thickness: 1.11 om 35 Fe** groups in the helix on the basis of intensity ratios among Elle and Ele because the intensity is increased along the vector between the interacting cations. Vectors between adjacent cations in the helix lie 35° off the c-axis and thus have components in both polarizations. Specifically, the intensity ratio Elc/Elle should be 0.49. This cannot be rigorously tested because of the interfering trimer feature in the Same general wavelength. Using sample 6 to remove the trimer contribution in Elc¢, Elc/Elle ratios of 0.4-0.6 were obtained for samples 1-4; in approximate agreement with the observed ratio within the uncertainty of this calculation. Sample 5 has an unusually low ratio of 0.1, but this could be related to low levels of other chromophores. The contribution in Ele is most easily seen by the increase in intensity at 540 nm at 83 K as shown in Fig. 64. The temperature response of the 540 and 425 nm bands is not as simply explained as the trimer bands. The 540 nm band increases in integrated intensity by 60% from 296 K to 83 K. This behavior is typical of 'cold' bands and would indicate an assignment of 2 x Ca, > “r,). The 425 nm band maintains an intensity 0.2 times that of the 540 mm band. However, its energy does not correspond to a sum or combination of reasonable band positions. In addition, a weak band at ~800 nm is evident in the fully oxidized sample 1. This appears to be the Oy > “r, transition and would contradict the assignment above. An 1 alternate interpretation based on exciton-magnon bands is possible. The 3+ 425 and 540 nm bands have energies typical of Fe transitions of the great majority of high-spin Fest compounds. The typical effects of exchange coupling are large increases in intensity of the sharp band(s) between 410 and 450 nm and lesser increases for the broad band in the 36 region 750-1000 nm in exciton-magnon transitions. In this case, the band near 500 nm, which is generally little if at all affected, has the greatest intensity. The strength of magnetic exchange is affected by the geometry of the connection (Murray 1974). The unusual temperature response, therefore, may result from anisotropic contraction in the helix with decreasing temperature. This proposal must be tested by low temperature structure refinements. Tron-rich tourmalines which are Al-deficient exhibit a Fe?* feature at 410 nm (Fig. 9), as well as the transitions associated with pairs in Fe2t-pest the helix and trimer. interactions (Smith 1978; Chapter 3) 3+ groups within the produce the dominant bands at ~700 and ~1100 nm Fe trimer produce the characteristic absorption band at 485 nm in El¢ as seen previously. Fe ot groups within the helix are evident from the small band at ~540 nm in Elle. A possible assignment of the 410 nm feature to the 8, > Ca 0°B) band of Mn2* which occurs at 414 nm in Li,Al-tourmalines (Reinitz and Rossman, in prep.) can be ruled out by the extremely low Mn content of these samples. The band at 410 nm in Elc has an intensity nearly equal to the 485 nm band and can be assigned to ion-pairs between the trimer and helix. This assignment is supported by the polarization of this band. In addition, it is restricted to tourmalines in which substantial amounts of Feo* are incorporated into the helix due to Al-deficiencies. 410 nm is a very unusual wavelength 6 4 4 - = 3+ ta for the 4 > ¢ a E) band of Fe in silicates. The spectroscopy of non-interacting Feot ions in tourmaline would ordinarily indicate the energy of pair transitions. An attempt to observe dilute ferric iron spectroscopically by heating Mg-tourmalines with <0.5% FeO at g50°C for : : : 2+ ~69 hours were unsuccessful despite an obvious decrease in Fe” . 37 aut, PTTOS 2540 BUTT vayoaq :5 7a suOoz;eryp vopavzypaeypog wn Qz0°O :ssauyxITYyI aTdmeg ‘QI]-g seTdwes jo aap wyuosoadas S} STUL ‘4 oTdwes Jo e1390ds uoT3daosqe aanjeraduay wooy °6 *3y4 | wu “HLONATAAVM @O6! BOLI BBS! BEI BBIl BB6 BBL BBS BLE 7 ll - << v v ¥ 7 T ¥ ¥ v v v ool — ww “\ ~ bee \ o = ‘ ow . \ Z bee \ «J A ’ . | oe | orn © © GO TFT WN Bal eee! 92082 wo “ASEWNNSAVM e 8 8 8 ~SONVaeOSsv 38 However, the transitions associated with pairs in the helix and trimer can provide similar information given correct assignments. — While it is generally believed that intensified pair transitions occur at the same wavelength as the corresponding transitions of isolated ions, these views are based on studies of pairs which are symmetric or nearly so. These Y-Z pairs are asymmetric and do not simply exhibit the transitions of the individual ions. An optical investigation of ferridravite, which is extremely Al deficient (Walenta and Dunn 1979), would help to fully establish this assignment Structural Effects Although the transitions related to Fe?t discussed above are observed in tourmalines which span a large compositional range, their 3+ . properties cannot be simply extended to the highest Fe concentrations. Buergerite (Mason et al. 1964) has ~4 times the Fest 3+ content of sample 1. Approximately 90% of this Fe~ is located in the trimer (Hermon et al. 1973), but the optical spectrum is not dominated by a band at 485 nm. It consists of a shoulder at ~490 nm on a background of what appears to be Fe-t-o CT (Fig. 10). The distance between cations within the trimer in buergerite is 317 pm (Barton 1969) as opposed to 304 pm in dravite (Buerger et al. 1962). This increased separation results in a decreased antiferromagnetic interaction among 3+ the Fe ions and is evidenced by a coupling constant which is ~1/3 that found for sample 1 (Tsang et al. 1971). A correlation between molar absorptivity and coupling constant has been observed in a survey of 3+ antiferromagnetically coupled Fe compounds (Rossman unpublished data). The weaker interaction is correspondingly seen as a lower 1 absorptivity, ~7 wlem!. A lower absorptivity at higher Fe°t 39 WAVENUMBER, cm! 25000 20000 | 45000 1.4 1.27 1.0; 0.8; ABSORBANCE 0.6 Q.4+ 8.2t — 400 088 680 708 WAVELENGTH, nm Fig. 10. Room temperature absorption spectra of buergerite (sample 11). Polarization directions Ele: broken line Elle: solid line Sample thickness: ~16 um ho concentrations is also indicated by the iron-rich tourmaline, sample 6, 3+ content of this sample is ~802 which was completely oxidized. The Fe of that in buergerite. The molar absorptivity of the 485 nm band based on calculated proportions of iron in ‘dimers' and 'trimers' is 20 wlem7! as compared to 90 wlem! for samples 1-5. The Fe ot transitions discussed above represent the optical features of Fe?* in a wide range of tourmaline chemistries, but they are not apparent in oxidized tourmalines studied by Smith (1978). Smith observed bands at 524 and 463 nm in oxidized Fe-bearing tourmalines and attributed them to Fe?t-pair transitions. These tourmalines were in the 2+_tourmalines, whereas all compositional series between Li,Al- and Fe tourmalines of this study were Mg,Fe-tourmalines. Small differences in the optical spectroscopy might thus be expected due to minor structural differences accompanying compositional differences. The 524 nm band was observed to occur in both polarizations and increase in intensity at low temperatures. It may correspond to the 540 nm band assigned to Fe? pairs in the helix in Mg ,Fe?*-t ourmalines. In Smith's Mn-rich . + tourmalines, however, Mn? which absorbs at 520 nm (Manning 1969) may also contribute to this band. The 463 nm band does not correlate with the 524 nm band and is not consistently present in the oxidized samples. Since iron is known to prefer the trimer, the 524 nm band should not dominate the Fe?* features in these samples. These assignments therefore, cannot be confirmed by this study and await the 3 synthesis of Fe *—bearing Li,Al-tourmalines. ha Summary Interactions among various cations are very pronounced in tourmaline. In the ferric-rich tourmalines of this study these interactions are evident in magnetic and spectroscopic properties. 3+ Antiferromagnetic exchange among Fe ions is indicated by decreasing magnetic moment at lower temperatures and by intense optical 3+ transitions. Three types of Fe~ -pairs involving different site combinations can be distinguished by their optical characteristics. These characteristics have been established in ferric-rich tourmalines which contain low levels of other chromophoric ions. These observations 3+ have permitted the distinction of Fe transitions in more complex members of the tourmaline group. In the absence of a large Fe2t Fest content, bands at 485 nm (Elc) and 540 nm (Elle) dominate the visible region and give rise to a red or yellow color. The high 3+ intensities of these Fe transitions are one indication that they result from ion-pair interactions. The transitions of isolated Fe?* which were tentatively identified in sample 1 are not generally present in Fe?*~bearing tourmalines. This is somewhat puzzling and suggests that the interactions with Fe2+ 3+ which are so prevalent in tourmaline may cause Fe to lose its single ion identity. The transitions of Fe?t-pe st pairs in the trimer of the tourmaline structure (485 nm Elc) occur at unusually low energies. These low energies were proposed to result from interaction in the highly constrained, edge-shared sites of 3+ the trimer. Optical evidence of Fe is particularly useful in minerals which are zoned or otherwise unamenable to bulk techniques such as 3 Méssbauer spectroscopy. Because Fe + Méssbauer features are often 2+ -obscured by Fe in tourmaline, the Méssbauer spectra of the Kenyan 42 tourmaline provides the first clear standard for tourmalines of typical compositions. The temperature dependence of the Méssbauer spectrum of this dravite has not been observed in other tourmalines, or in other 2+_5 3+ minerals, and may be related to Fe” -Fe~ interactions. 43 REFERENCES Amthauer, G, Annersten, H, and Hafner, SS (1976) ‘The Méssbauer spectrum of 37 Fe in silicate garnets. Z Kristallogr 143:14-55 Amthauer, G and Rossman, GR (1984) Mixed valence of iron in minerals with cation clusters. Phys Chem Minerals, in press Bank, H (1974) Rote Turmaline mit hoher Licht- und Doppelbrechung aus Kenya. Z Dtsch Gemmol Ges 23:89-92 ‘Barton, R Jr. (1969) Refinement of the crystal structure of buergerite and the absolute orientation of tourmalines. Acta Crystallogr B25:1524-1533 Bence, AE and Albee, AL (1968) Empirical correction factors for electron microanalysis of silicates and oxides. J Geol 76:382-403 Buerger, MJ, Burnham, CW, and Peacor, DR (1962) Assessment of the several structures proposed for tourmaline. Acta Crystallogr 15:583-590 Burns, RG and Strens, RGJ (1967) Structural interpretation of polarized absorption spectra of the Al-Fe-Mn-Cr epidotes. Mineral Mag 36:204- 226 | Dollase, WA and Gustafson, WI (1982) 27 Fe Méssbauer spectral analysis of the sodic clinopyroxenes. Am Mineral 67:311-327 Donnay, G., Ingamells, CO and Moser, B (1966) Buergerite, a new species of tourmaline. Am Mineral 51:198-199 Dunn, PJ, Arem, JE and Saul, J (1975) Red dravite from Kenya. J Gemmol 14:386-387 bby - Earnshaw, A, Figgis, BN, and Lewis, J (1966) Chemistry of polynuclear | compounds. Part VI. Magnetic properties of trimeric chromium and iron carboxylates. J Chem Soc A 1966:1656-1663 Faye, GH, Manning, PG, Gosselin, JR and Tremblay, RJ (1974) The optical absorption spectra of tourmaline: Importance of charge transfer processes. Can Mineral 12:370-380 Fortier, S and Donnay, G (1975) Schorl refinement showing the composition dependence of the tourmaline structure. Can Mineral 13:173-177. Fujiwara, T, Gebhardt, W, Petanides, K, and Tanabe, Y (1972) Temperature dependent oscillator strengths of optical absorptions in MnF 5 and RbMnF3. J Phys Soc Japan 33:39-48 Gorelikova, NV, Perfil'tyev, Yu D, and Bubeshkin, A M (1976) Méssbauer data on distribution of Fe ions in tourmaline. Zap Vses Mineral O- va 4: 418-427 (transl. Int Geol Rev 20:982-990, 1978) Greenwood, NN and Gibb, TC (1971) Méssbauer spectroscopy. Chapman and Hall Ltd, London. Hermon, E, Simkin, DJ, Donnay, G, and Muir, WB (1973) The distribution of Fe? and Feo? in iron-bearing tourmalines: a Médssbauer study. - Tschermaks Mineral Petrogr Mitt 19:124-132 Kénig, E (1966) Magnetic properties of coordination and organometallic transition metal compounds. In KH Hellwege and AM Hellwege, Ed., Landolt-Bérnstein Numerical Data and Functional Relationships in Science and Technology, New Series II, 2:2-114 Krebs, JJ and Maisch, WG (1971) Exchange effects in the optical- absorption spectrum of Feo* in Al503.- Phys Rev B: Solid State 4757-769 45 : Kurkjian, CR and Buchanan, DNE (1964) Méssbauer absorption of 57 Re in inorganic glasses. Phys Chem Glasses 5:63-70 Lehmann, ¢ (1970) Ligand field and charge transfer spectra of Fe(II1I)-0o complexes. Z Phys Chem (Wiesbaden) 72:279-297 Lohr, LL Jr. and McClure, DS (1968) Optical spectra of divalent manganese salts. II. The effect of interionic coupling on absorption strength. J Chem Phys 49:3516-3521 . Long, GJ, Robinson, WT, Tappmeyer, WP, and Bridges, DL (1973) The magnetic, electronic, and Méssbauer spectral properties of several trinuclear iron(III) carboxylate complexes. J Chem Soc, Dalton Trans 1973:573-579 Mao, HK, Bell, PM and Virgo, D (1977) Crystal-field spectra of fassaite from the Angra dos Reis meteorite. Earth Planet Sci Lett’ 35:352-356 Manning, PG (1969) Optical absorption spectra of chromium-bearing tourmaline, black tourmaline, and buergerite. Can Mineral 10:57-70 Mason, B, Donnay, G and Hardie, LA (1964) Ferric tourmaline from Mexico. Science (Washington DC) 144:71-73 Murray, KS (1974) Binuclear oxo-bridged iron(II1) complexes. Coord Chem Rev 12:1-35 O'Connor, CJ (1982) Magnetochemistry--Advances in theory and experimentation. Prog Inorg Chem 29:203-283 Pastor-Rodriguez, J and Taylor, HFW (1971) Crystal structure of coalingite. Mineral Mag 38:286-294 Rossman, GR (1975) Spectroscopic and magnetic studies of ferric iron hydroxy sulfates: ‘intensification of color in ferric iron clusters bridged by a single hydroxide ion. Am Mineral 60:698-704 46 . Rossman, GR (1976a) Spectroscopic and magnetic studies of ferric iron hydroxy sulfates: the series Fe(OH)SO,*nH,0 and the jarosites. Am Mineral 61:398-404 _ Rossman, GR (1976b). The optical spectroscopic comparison of ferric iron tetrameric clusters in amarantite and leucophosphite. Am Mineral 61:933-938 Saegusa, N, Price, DC, and Smith, G (1979) Analysis of the Méssbauer spectra of several iron-rich tourmalines (schorls). J Phys (Paris) 40:€02,456-C2,459 Schugar, HJ, Rossman, GR, and Gray, HB (1969) A dihydroxo-bridged ferric dimer. J Am Chem Soc 91:4564-4566 Smith, G (1978) A reassessment of the role of iron in the 5,000- 30, 000 om”! region of the electronic absorption spectra of tourmaline. Phys Chem Minerals 3:343-373 Tsang, T, Thorpe, AN, Donnay, G, and Senftle, FE (1971) Magnetic susceptibility and triangular exchange coupling in the tourmaline mineral group. J Phys Chem Solids 32:1441-1448 Walenta, K and Dunn, PJ (1979) Ferridravite, a new member of the tourmaline group from Bolivia. Am Mineral 64:945~948 Waychunas, GA and Rossman, GR (1983) Spectroscopic standard for tetrahedrally coordinated ferric iron: YLiA10,:Fe?*. Phys Chem Minerals 9:212-215 Wignall, JWG (1966) Méssbauer line broadening in trivalent iron compounds. J Chem Phys 44: 2462-2467 Wilkins, RWT, Farrell, EF, and Naiman, CS (1969) The crystal field spectra and dichroism of tourmaline. J Phys Chem Solids 30:43-56 4? Wu, CS, Rossman, GR, Gray, HB, Hammond, GS, and Schugar, HJ (1972) Chelates of 8-diketones. VI. Synthesis and characterization of dimeric dialkoxo-bridged iron(III1) complexes with acetylacetone and 2,2,6,6-tetramethylheptane-3,5-dione (HDPM). Inorg Chem 11:990-994 Chapter 3 Fe2+ - Fest Interactions in Tourmaline 49> Introduction A review of tourmaline chemistries (e.g. Deer.set ail. 1962) illustrates the lack ef correlation between intensity of color and iron content in green, blue and black tourmalines. This behavior is similar to many compounds in which transition metals of multiple and variable oxidation states occupy adjacent sites. The intense colors of mixed-valence systems have been ascribed to optically induced charge transfer between adjacent cations. However, the optical spectroscopy of iron-bearing tourmalines is not typical of mixed valence systems. A study of green and blue tourmalines (Smith 1978a) revealed a previously unknown effect of the interaction among mixed valence cations -- an intensification of the characteristic absorption of individual cations. The intensification of Fe=* transitions by neighboring Fe **, first demonstrated in tourmaline and biotite (Smith 1978b), has since been suggested for MgO:Fe (Smith 19380) and the protein hemerythrin (Loehr et al. i980). Quantitative absorption intensities are of interest because they can be used to determine the concentration and site occupancy of Fe=" in minerals such as olivine (Burns 1970) and pyroxene (Goldman and Rossman 1979). These determinations depend on a linear correlation between absorption intensity and concentration and would be corrupted by a strong interaction among adjacent cations. Although previous studies have demonstrated the existence of intensified Fe** absorption in Fe=*-Fe=+ compounds, the 30 ‘quantitative magnitude of the intensification of Fe2+ by “Fe ** neighbors has not been determined. This paper presents a systematic study of Fe=*-Fe =" interactions ina broad | range of tourmaline chemistries directed toward a quanti- tative characterization of intensified Fe2> transitions. The wide range of chemistries and colors of the tourmaline group have prompted many optical investigations. These studies have generally encompassed a small range of chemistries from gem—-bearing pegmatites and produced many conflicting assignments. Smith (1978a) reviewed these assignments for iron-bearing tourmalines. Missbauer data on a few of these Li,Al—tourmalines have shown very low or undetectable Fes" (Burns 1972; Simon i973; Faye et al. 1974). Most Mossbauer studies have been focused on black Fe, Mg-tourmalines and have detected much larger and quite variable Fe ** contents (Hermon et al. 1973; Gorelikova et al. 19763 Saegusa et al. 1979). Kecause of the greater abundance of Fae F* in black tourmalines, they will be emphasized in this study of Fe2*-Fes* interactions. Optical spectra of a few black tourmalines have been qualitatively examined by Manning (1969, 19723), Faye et al. (1974) and Wilkins et al. (1969) but without regard to intensified Fe=* transitions. Tourmalines can generally be grouped into 2 solid solution series. The detailed spectroscopic behaviors of these groups are distinct and will be discussed in addition to the larger variations indicated above. These series will Si ‘be differentiated using the sample nomenclature described below. The major chemical variations in tourmaline occur in ‘the K-, Y- and 7-sites as defined by the general formula XV¥s2.5i 2019 (BOs) = (0H,0,F) 5. The Z-site is predominantly Al in common compositions. In addition ta extensive substi- tutions in the Y-site, transition elements generally occur at lower concentrations in the Z-site. Previous optical studies have emphasized the elbaite-schorl series in which X = Na and Y = Li, Al-for the elbaite end member and kK = Na and Y = Fe?* for the schorl end member. This group is dominated by chemistries closer to elbaite than schorl. Greater chemical variation is found in the dravite-uvite- schorl series. The dravite end member chemistry is X = Na and ¥ = Mg. The uvite end member differs in X = Ca and Z = Al, Mg. The spectroscopic properties of the transition elements are not affected by changes in composition in the uvite-dravite series. The major substitutions of interest include the Mg, Fe** variations between these end members and schorl and the incorporation of Fet*. Substantial amounts of Fe=* can occur in both the Y- and Z-sites but do not commoniy approach the level of Fe * in the end members discussed in chapter 2. The dravite-uvite-schorl series will be emphasized in this paper and will be designated Fe, Mg-tourmalines. a2 Experimental Sample localities are presented in Table i. Chemical analyses were obtained using a MACS-SAS electron microprobe and are presented in Tables 2 and =. Data reduction was accomplished using the methods of Bence and Albee (1968). Semi quantitative (420%) analyses of samples Mi and 731 were obtained using a Kevex 0700 EDS XRF system. The fluores— cence of polished crystal slabs was calibrated against visually homogeneous tourmalines which were analyzed by microprobe. Differentiation of the elibaite-schorl series from Fe,Mg-tourmalines was based on Mg content and characteristic OH vibrations (Appendix). Optical spectra were obtained in a manner similar to that described in Rossman (1975). The spectra of thin samples (0.005-0.630 mm) were generally measured while attached to a glass slide with. epoxy. The thickness of these samples was determined using a microscope micrometer on polished, perpendicular sections. Many of the samples were extremely inhomogeneous in color. Zones of uniform color were selected, and the same area was analyzed by electron microprobe. Low temperature spectra were obtained by placing the sample on a coldfinger immersed in liquid nitrogen in a quartz dewar. Temperatures were monitored using a copper—constantan thermocouple placed sightiy above the sampie. Temperature variation daring a spectrum was limited to BO to 84 K. Roam temperature spectra were measured in the same apparatus without moving the sampie to “Table i aS Sample Descriptions Sample Colar Locality Archival Code 52 Dbluex S3 ' b=brownk g=-greenk S4 b=bluek ckg=greenx tf ui 54 brown k $7 greenk Sa greenk S9 brawn x 510 a=biluex b=bluex c=greenk Sit bluex Siz brown * Di brown D2 brown DBD? red T22 bluek Ei blue ERS light biue ERS blue White Queen Mine, Faia, CA Newry, ME Cecil Co., MD Anza Borrege, CA Fortland, CT Santa Cruz, Sonera Mexico Fierrepont, NY Madagascar Brazil Minas Gerais, Cahuilla Mt., Riverside C(c., CA Syndenham, Gntaria Yinnietharra, Australia Bryant Lake, NY Minas Gerais, Erazil Blue Lady Mine, San Diego Ca., CA Afghanistan ‘Afghanistan hy ghanistan GRR# 3/isS/78 CIT# 1372 CiT# 3575 GRRE 3/14/83 CIT# 10387 NMMHH RiSsso4 NMNH# Si51i Harvard# i087%96 CIT# 12763 CiIT# 2504 CiT# 126685 CIT# 12684 NMNH# 137101 GRR# 8/10/35 GRR# 1/27/80 GRR# 1/27/80 GRRE 1/27/50 “Fable i cont. o4 Sample. Coaior Locality Archival Code EGS . green EG4 dark green EG1#4 bluex EGS light green Tis green Zi blue— green mi blue Afghanistan Afghanistan Stewart Lithia Mine Pala, CA Himalaya Mine, Mesa Grande, CA zambia Zambia Moz ambi que GRR# 1/27/80 GRR# 1/27/80 GRRE 3/27/77 GRRE 46/7/54 GRR# 2/11/53 #16 GRR 2/11/85 #31 GRR# 9/74/32B x Elec color in thin section so “Table 2 lectron Microprobe Analyses of Fe, Mo-—tourmalines Sampie Nao Cat F200 MoO Feo Mind Ales 5ids Tide = S2 2.03 0.03 0.04 0.09 13.5 0.95 0.08 35.2 34.7 0.4 S3b 1.67 0.43 0.02 4.97 8.45 0.07 0.35 34.7 35.3 BDL 33g 1.88 0.85 0.04 6.25 6.98 0.03 1.01 34.2 34.7 BDL S4b 2.26 0.72 0.02 98.01 7.79 0.06 6.28 30.9 35.6 6.4 S4g 2.39 0.41 0.01 7.78 6.93 0.05 0.53 31.9 35.6 0.3 S4c 2.43 0.51 0.02 7.91 8.38 0.04 0.36 36.5 34.1 G.2 Sb 1.68 0.34 0.04 3.11 9.16 0.28 6.34 36.6 34.8 0.2 S5g 1.78 0.41 6.04 5.04 6.82 6.15 0.31 36.1 35.4 0.4 86 2.31 0.435 0.06 3.56 11.6 0.14 1.03 32.5 34.7 1.2 S7 1.86 1.99 0.05 6.48 18.1 0.05 1.18 22.2 34.0 O.5 sa 1.44 2.92 0.05 10.8 9.23 BDL 0.55 26.3 35.2 1.4 Ss? 1.67 2.42 0.08 93.58 13.4 G.01 1.49 23.6 24.4 1.1 S1Ga- 1.67 0.06 0.05 2.90 11.0 0.15 0.15 34.8 35.4 o.2 $10b 1.72 0.07 0.01 2.94 10.8 0.14 0.23 34.9 35.4 0.2 S10c 1.94 0.11 0.05 2.99 10.8 0.16 0.40 34.5 35.2 0.2 Sil 1.34 9.06 6.04 6.48 13.3 6.28 6.05 35.4 35.4 BDL S12 1.58 2.61 0.04 10.9 7.84 BDL 0.72 26.5 35.3 1.4 Di «2.93 0.37 6.02 11.46 0.46 BDL 1.69 32.0 37.2 O.1 D2 0.77 4.25 0.10 14.3 0.38 BDL 0.58 28.7 36.5 6.3 D7 1.77 1.44 O.11 10.1 4.46 G.22 0.22 31.0 FS.1 O.1 BDL ——- below detection limit 36 “Table 3 Electron Microprobe Analyses of Elbaites Sample NazQO CaO MgO Fel MnO Tide AleOs ZnO Side TiS 2.45 G.47 0.21 2.16 1.94 0.05 39.4 NA 37.35 T22 2.01 0.06 BDL i2.2 0.89 6.04 35.6 G.59 35.2 EGS 2.50 0.26 BDL 2.49 1.51 09.032 39.8 0.09 37.5 EG4 2.7/4 0.18 BDL 4.32.1.47 0.04 37.98 9.41 37.4 EGS 1.79 1.20 BDL 6.42 0.51 06.02 42.4 36.06 38.6 EBL 2.70 G.17 BDL S.34 1.10 BDL SY.7 6.14 34.4 ERS 2.45 9.45 BDL O17] 3.17 BDL 29.5 6.04 37.1 ERS 2.57 0.25 BDL 1.62 1.71 BDL 40.9 6.04 34.9 EG1I#4 2.546 0.14 0.08 4.41 2.74 BDL 38.4 NA 36.9 BDL -- below detection limit NA -—- not analyzed a7 ‘insure. quantitative comparisons. = Samples EBS, Mi and 231 were subject to radiation from a *57Cs source which emits 6.442, 1.01, and 1.434 MeV gamma rays and was producing doses of “1.3 Mrads/day. Missbauer spectra (fitted and raw data) were provided for samples S7, S38 and S9 by Roger Burns (MIT) and by Glen Waychunas (Stanford) for samples 52 and S46. Three doublet fits were used to estimate Fe=" concentrations in the Y- and 2-Sites and Fa =" concentrations. Z-site occupancy of Fett was estimated by charge balance requirements in the absence of H-defects. With the exception of D2, these results were used to calculate the fraction of Fe** with Fet* neighbors assuming a random distribution. The same calculation for D2 was accomplished using correlated ESR and optical data. ESRF spectra were obtained on powdered samples of De before and after partial oxidation using a Varian E-line X-band spectrometer at *9.2 GHz. Results and Discussion The Dasic characteristics of Fe="* absorption in tourmaline are illustrated in Fig. 1. At very low iran contents and low Fe **/Fe=" ratios the spectroscopy of non- interacting Fe2+ can be seen (Fig. ia}. The broad bands at 720 and 12590 mm can be assigned to components of the =T2 > SE Fe=* transition split by the non-cubic crystal fieid h 197S8a}. Unlike the previously published spectra and rt mi zs z *. tf most of my tourmaline spectra, the intensity of these 58 ABSORBANCE yubta ayers aauequosqe ‘peygy ardwes (q “aw [9P°Z FO SSAUYITYR © O] pazi{ewsoU Sazreg(a 40 esjjads vorydiosqe aunzesadway vooy oy “Ory wu “HLONTTIAVA Vay] apeas azuequosqe ‘7qy aydwes (y Wu "HLONTTIAVA TS 0001 009 008} =a) Sts« 009 OTS cod | | cg) CVD 9001 0000! 90ea2 0001 00001 eoaez - “USEHANSAVA _ "ATSHNNSAVA ¢ 6 r6 ~ TNVSUOSEY a9 ‘transitions in Ele does not xceed that in Effie. The sharp bands between 1200 and i500 nm are vibrational transitions which will not be discussed in this chapter. A comparison of this spectrum to that of a more typical tourmaline {Fig. ib) shows the basic features cf Fe®*-Fe** interactions in tourmaline. Absorption intensity in Elle has increased ‘linearly with iron content. This behavior is typical of transition element absorption in most environments. The game transitions in Elec have increased “10 times over the increase in Fe content. This added intensity in Elec can be attributed to Fe** as has been demonstrated by partial oxidation of natural samples (Smith 1978a; Wilkins et al. 1969). Fe2*-Fe=* charge transfer, the expected signature of Fe2*-Fe>* interactions, has been assigned to the | comparatively weak feature at S50 nm (Smith 1978a; Faye et al. 1974; Townsend 1970). These spectra are representative of the gemmy, low Fe **/Fe?* eibaites which have been preferentially studied but which do not exhibit the greatest Fe“*-Fe** affects. Figure 2 illustrates the spectrum of a Fe,Mg-tourmaline particularly rich in Fe** as evidenced by the Fe** band at 485 nm (Chapter 2). In most Fe?*-Fe = compounds this large change in the relative amount of Fe=* would be accompanied by a large change in the optical spectrum. In particular, a greater contribution from Fe**-Fe** charge transfer ‘would be expected. However, excluding the Fes* feature at 485 nm and Fe**-Ti** charge transfer at 420 nm (Chapter 4), 60 ‘ op9 aTdwes - ee — ww PTO MSSBuUxITYR a Tdweg ZO PuQoads uotzdsosqe aun}yewaduaz wooy "gZ "Ora wu “HLONATSAVM @eS| e001 =" ry 4 2. r 5. 4. | SO REE SS GONE SUE SOOO SE OU, RS RO AOS SUE ED 1 |- t Boo B8a0 | wo “YSAGWNNsAVM t BBBG~e Ss: 4) SINVEsOSeV éi ‘this spectrum shows the same features as those of Fig. i. ‘The major difference lies in the much larger Elc/vEllc ratia of the Fe2* bands. Althcugh Fe=*-Fet+ charge transfer may contribute to absorption in Eic, it is not the dominant expression of Fe=*-Fe** interactions in tourmaline as it is in most Fe**-Fe=* compounds. Fe®*-Fe™* tourmalines thus provide a rare opportunity to fully characterize a second effect of Fe®*-Fe ®* interactions——-intensified Fe=* transitions. The qualitative relationship of enhanced intensity of Fe=* transitions to Fe ** can be simply demonstrated by partial oxidation of tourmaline as illustrated in Fig. 3. Oxidation of Fe** in tourmaline by heating in air has been established by Mossbauer studies (Gorelikova et al. 1976; Faye et al. i974). This is accurately reflected in Elle (Fig. 3b) by a decrease in intensity of the 1100 nm Fe=+ transition in an Fe,Mg-tourmaline. However, the intensity of rr his feature in Elec has increased, despite a decrease in Fe=", In addition a Fe** band at 485 nm has appeared. With the exception of the Fe =* band at 485 nm, similar changes have been observed in elbaites (Smith i97S8a; Wilkins et al. i969). These results indicate a higher intensity for Fe=> in Fe*"*-Fe>* pairs as compared to non-interacting Fe=*,. The restriction of anomalous intensity to Elec is further evidence that this effect can be attributed ta ion-pair interactions. |The polarization of ion—-pair transitions along intercation vectors is a familiar aspect of charge 62 WAVENUMBER, cm! 208289 - $9288 7288 }______, ee Pee ; CAD J 2.0F : lu r : z 1.55 rea) C ; yy t ; QO 1.ef 1 rea) r << ' « @.5f : Lu () Zz <f roa) rw O YY) M .~ 508 138g 1502 2889 . WAVELENGTH, nm Fig. 3. Roog teaperature absorption spectra of untreated {broken line) and oxidized (solid line) sample Sil. Sample thickness: 0.045 a= A} Ele polarization 8) Ec polarization 63 transfer transitions and the intensified spin-forbidden transitions of Chapter 2. The dominant distribution of Fee*-Fes* pairs in tourmaline are in site combinations with vectors perpendicular to the c-axis. 45 was indicated in the intraduction, transition elements occur in the ¥- and Z-sites in tourmaline but generally prefer the Y-site. lon-pairs within the Y-site trimer and those between the. trimer and the Z-site helix are oriented perpendicular te the canis (Huerger et al. i962). Either of these ion-pairs are consistent with the observed polarization. Ilon-pairs can also occur within the Zi-site helix but are oriented at ~S35° to the c-axis. intensity of Fe=* absorption in Fe®*-Fe=* groups Extreme variations in Fe@* molar absorptivities from compound to compound have made it difficult to differentiate intensified from unperturbed Fe" transitions without systematic studies of a range of compositions. Goldman and Roseman {in prep.) have shown, however, that absorption intensity can be estimated from structural parameters. The Variation of molar absorptivity with iron-oxygen distance fin the absence of Fe **) is shown in Fig. 4. The average distances for the Y- and Z-sites of elibaite (Donnay and Barton i972), dravite (Buerger et al. 1942), and schorl t(Fortier and Donnay 1973) are included in the crosshatcned regions. ThRis approximats correlation indicates that Fe?* molar absornptivities for tourmalines should be *4-15 64 6-Coordinate, High Spin Ferrous Iron im Minerals 7 —T 2 ' —y" os T 7 ry r {SO} Ww 3 > « «1ae > | _— QO. © © a wd <x 56+ a @ wd 2) = Q _®, yu 2.8 2.1 2.2 2.3 2.4 AVE. METAL-LIGAND BOND DISTANCE | CAD Fig. 4. Variation of Fe®* aolar absorptivity with average Fe-0 distance (taken from Goldman and Rossman, in prep.). The average Y-Q and 2-0 distances of tourmaline end eeabers are included within the labelled, hetchured regions. 65 Motem—?. These values can be contrasted with minimum molar absornptivities for the 1100 nm band based aon total Fe content (Fig. 3). Feet absorptivities in Elle (indicated by the solid symbols) vary from 3-3 M-4cm—-? with a few exceptions. Higher absorptivities are characteristic of samples with large Al deficiencies which permit a smail number of Fe#*-Fes* pairs in Z-Z combinations. Most of the apparent variation in Elle absorptivity can be related ta errors introduced by normalizing absorbance to Fecotar instead of Fe=*. Molar absorptivities for samples 54 and ha ify based on Mossbauer measurements of Fe=* are 4.4 and 5.23 M-*cm—*?, This discrepancy may reflect a small difference in the Y- and Z-site absorptivities. The molar absorptivity of the 720 nm band in elbaites (Elle) is somewhat smaller (3.2 M-tcm—2, Smith 1978a Fig. 4a) in accord with smaller sites. -In contrast to Elle, absorptivities in Elc approach the predicted values only in elbaites and at low iron contents in Fe,Mg-tourmalines and vary by more than an order of magnitude. These variations indicate two or more Fe=r species with very different absorptivities. Predicted absorptivities for the Y- and Z-sites (Fig. 4) are not large enough or different enough to produce the observed variations. AS was indicated by partial oxidation in Fig. Za and will be demonstrated below, Fe?*-Fet* sairs are the intensely absorbing Species. The magnitude of the variations in the minimum Ele absorptivities indicate a potential of at least an order of 66 1188 nm Transition in Tourmaline Absorbance/cm/(Fe conc.) 258 ———_ 1 200 + 4 a ah 1SO- q 4 oO oO oO 190+ O 4. 5 oo SO- 5 2 - a } © a ry re: ry" ave" 4 rs) 2 4 6 8 12 Fe Concentration Cmoles/liter)D Fig. 5. Minimum molar absorptivities of the 1100 nm Fe=* band of tourmalines. oO: Elc Fe,Mg-tourmalines © : Elc elbaite-schorls @&: = Elle Fe,Mg-tourmalines 57 ‘Magnitude increase in Fe2+ absorption intensity in the presence of Fe rr. In order to refine the magnitude of the intensity increase and demonstrate its relationship to Fes+ independent measures of Fe" and Fe** were acquired. The intensity of Fe=* absorption in Fe**-Fe** pairs can be seen by relating molar absorptivity to the fraction of Fe** with -Fe =" neighbors (Fig. 6). ° The fraction of Fe** with Fe =r neighbors was calculated with Fe@* and Fe ™* site occupancies assuming a random distribution. Details of this calculation are given in the experimental section. A direct correlatian of Fe =' with increasing intensity is shown in Fig. 4. Several complicating factors make it difficult to judge the exact form of the relationship between absorptivity and the fraction of Fe?*-Fe **, However, the average absorptivity of Fe" in Fe**-Fe =" pairs can be estimated by a linear fit to these points extrapolated to 100% Fe** with Fe =", This value, “1200 Mutcm~* is 2 orders of magnitude greater than the absorptivity of non-interacting Fer (*E!lc molar absorptivity} and *190 times greater than the large absorptivities associated with extremely distorted sites (Fig. 4). Both components of the =T > °E transition exhibit Similar intensifications as indicated by an approximately constant intensity ratio at *700 and 11900 nm (Ele in Fe,Mg- tourmalines). The magnitude of this effect points to the ‘extreme caution which must be excercised when interpreting optical intensities quantitatively. Tegether with another intense ionm-pair transition (Fe**-Tit* charge transfer) this 68 {188 nm Transition in Ele Polarization See + > 4004 — me 4 > = QO | & 320} He "7 ve = L 5 200b 4 al Oc + “Ny Hb @. %,) j J 8.2 | 8.2 8.3 P+ Fraction Fe” with Fe°* Neighbors Fig. & £Variation of Fe** molar absorptivity with the ratio of Fe=*-Fe =" groups to total Fe=*. Data represented by open symbols differ from the other data because of uncertainty arising from poor Méssbauer data. Included samples: a = DS, b = S46, c = S32, d = S9, @ = SF. &9 effect produces one of the most familiar properties of ‘tourmaline, its visibie pleochrcism. Errors in this assessment of the molar absorptivity of Fe2+-fe=+ pairs couid be introduced by 3 non-experimental uncertainties which could not be reduced. Two of these are related to the effect of different Fe=*-Fe>* clusters, various numbers of Fe ** neighbors as well as different site involvements. Deviations from a random distribution are aiso possible. Experimental uncertainties were reduced by selecting visually homogeneous samples to insure the correlation of Missbauer and optical data. The slightly concave shape of the correlation of Fig. 6 is undoubtedly related to deviations from the calculational assumptions. Both an increase in absorptivity with greater numbers of Fe** neighbors and the preferential avoidance of Fe=* and Fe=* could produce a similar correlation. A preferential avoidance of Fe** and Fe®* is indicated by several Fe ="-rich tourmalines (B4, DY and T2i) which have the lowest observed Eic/Ellc ratios of Fe,Mg-tourmalines. The effect of different Fe=*-Fe =" pairs cannot be assessed without a more extensive study using correlated Méssbauer and optical data. Biagnostic Attributes of Intensified Fe=* Transitions The characteristics of intensified transitions which are shared oy their parent unperturbed transitions, namely energy. and width, have made this effect of ion-pair 7O ‘interactions difficult to distinguish in individual cases. “Intensified Fe="* transitions, however, do possess. characteristics which can be used diagnostically. The restriction of intensity increases to the vector between the interacting cations can be useful with known structures. In these cases it can also be used to determine the sites involved. A more generally useful characteristic is the response of these transitions to temperature variation. Unperturbed d-d transitions maintain a constant integrated intensity with varying temperature or decrease in intensity with decreasing temperature. This behavior can be seen in Elle in tourmalines (Fig. 73; Smith i1978a) and for non-interacting Fe=* in Ele (Fig. 8). Fe** bands in Elc of most tourmalines exhibit large intensity increases with decreasing temperature. Smith*s study of elbaites found increases in inteorated intensity of 20-S0% from 293 K to 10 Ko at 1100 na. Both Fe=* bands in elbaites (725 and 1100 nm) in Elec show this behavior. Fe,Mgq—-tourmalines exhibit 2 bands at 745 and 5790 nm which maintain similar intensities as well as a band at 1100 nm (Fig. 9%). Increases in integrated intensity for Fe,Mg-tourmalines with temperature decreasing from 27946 Ko to 77 K ranged from 0 to 350% at 1100 nm. No change in intensity was observed only in samples with a Elc/Elic intensity ratio less than 4. However, above lox’ change in intensity there is no correlation of the | amount or intensity increase tao the degrees of intensifi- cation as indicated by the Elc/Elle ratia. These variations 71 ¥ OB = UaOIq = ¥ 947 = Pros adAz auty Aq pazedipur ase saunjesadeay "OH OFO'Q = SSauxdTY apdeeg ‘uotzeztuepod Sstyz ur suoTzIsueJ} a4 $0 JNTABYaG Juaysisuod ayy Gurqzeszsnett fq apdees jo esjzazads vuorzdsosqe paztuejod Tyq yz ‘bry wu “HLONATSAVM OOr | BGC | Bee | aIGbs! 8e9 4. A A. .% 4. 2. rt £. 4. A. 3. ri dn my 4. rt 1 78°9 1 80°90 BBL gee0| eaes WS SASWANSAVM - JINVEeOsgav 72 YCA = aUTT PELOS 4 94Z = AULT waYosG ssasnqzesadway “we 979°9 = S8auyTIyy apdweg ‘sautpewsnoz-by‘ay ut ,,a4 burzresazUT -uou 4O JOTABYaG ay} Hurzeszsnyys gq ardwes yo eayzlads worzdsosqe paziuejod 37y ‘gq “Bry wu “HLONSATAAVM geez Besl aegl Berl Bez Bee! Beg Bog 3 S. A A. A é. 4. A. Seme 4. G@GL 99001 ,-> “SS9EWNNSAVM AINVSaosay 73 WAVENUMBER, m7! 28889 18886 7688 ooo ABSORBANCE Nb OooON bd OD | eee eee ees eee Bee ee cee ee eee eed ee Gee es eee ee ee 2 1.44 1 uJ {.2 e “ > al al < I. baw 4 Om 2a.8t : D : 1 A 0.57 1 t 2.4} 1 @.2+ , 3e8 588 788 9389 1199 {1396 1580 1780 1988 WAVELENGTH, nm. Fig. 9. Els polarized absorption spectra of Fe, Mg-tourmalines illustrating the response of intensified Fe** transitions to temperature variation. Temperatures: broken line = 3945 E sGlid line = S32 E (Ad Sample Sa, thickness = 0.0098 mm (Bh) Sample 5&7, thickness = 0.0074 mm 74 bring into doubt the reliability of intensity increases at ‘low temperatures to distinguish intensified transitions in ‘other compounds. Experimental investigations of other Fe=*—-Fe>* compounds as well as a theoretical treatment of intensified transitions are required to assess the certainty of low temperature intensity increases. Fe=*—-Fet* Charge Transfer in Tourmaline Fe=+—Fet* charge transfer has previausly been assigned to a transition between 330 and 370 mm (Smith 197B8a; Smith and Strens i976; Faye et al. 1974; Townsend 1970). These assignments were based predominantiy on studies of the response of a few samples to partial oxidation. The applica- tian of this assignment to untreated tourmalines is complicated by the fact that there are several transitions which occur in this region. Ferric-rich Fe,Mg-tourmalines exhibit a transition at 335-545 nmin both Ele and Eltc which can be attributed to a Fe **-Fe** ion-pair transition (Chapter 3). Mns+ in elbaites has a transition at 320 nmin Eic (Manning 1969b). I have also found, in agreement with Faye et al. (197459), weak, spin-forbidden transitions of Fer in @lbaites at 497 (sharp) and S30 nm. The comparatively large width of charge transfer transitions (Smith and Strens 1974) should permit the differentiation of Fe=*-Fe*" charge transfer from d-d transitions in the same wavelength regian. In addition, Fe2+—pes+ charge transfer should be more intense in those tourmalines with highly intensified Fae-t 7a ‘transitions. 4 transition at S7S nmin Ele is clearly shown “in black Fe,Mg-tourmalines with Little Ti @Fig. Ya}. This is a4 relatively minor feature, however, and is frequently obscured by more intense Fe#*-Tit’ charge transfer (Chapter 4}. The general presence of the 375 nm band was indicated by fits of gaussian bands to Ti-containing samples. The “4500 cm—* halfwidth of this transition and its ‘“SOZ% increase in integrated intensity at low temperature are consistent with Fe@**-Fe** charge transfer. The polarization of the S75 nm band (Elc?3Elc) agrees with the major Y-site occupancy of iron. Amthauer and Rossman (1984) found molar absorptivities of Fe®**-Fe s+ charge transfer transitions to vary from 60 to 210 M7tcm7?. The lack of a prominent role for Fe2*-Fes+ charge transfer in tourmaline is consistent with a comparison of these values to the estimate of the absorptivity of Fe** in Fe**-Fe =" pairs (*1200 M7tcm7*). The presence of an additional band at ~700 nm in Fe,Mg-tourmalines as compared to elbaites might indicate Fe=*-Faes* charge transfer in another of the = site combinations which produce Elec polarization. However, the 80 EK halfwidths of the 470 and 7455 nm bands are similiar (“2100 cm71) and comparable to the *2500 cm7? haifwidth af the 1100 nm band. These values are distinctly lower than the observed range of halfwidths of Fe=*-Fe** charge transfer, 3500-6000 cm? (Chapter 3). I have therefore assigned the transitions at 479, 745 and 1100 nm in Ele toa intensified Fe=* d-d transitions. 7& i Ry re B Assignments of Fe=" Abscrnmtion The most significant difference illustrated thus far between elbaites and Fe,Mg-tourmalines is the number of intensified transitions. The 4679, 765 and 1160 nm bands of Fe,Mg-tourmalines are present at ali Fe concentration levels. Because charge transfer can be ruled out and 6&-coordinate Fe** axhibits at most 2 transitions in this region, these bands must refiect Fe=* in different sites. This is supported by the lack of a correlation between the proportion of the 670 and 765 nm bands and Fe=* concentration or the amount of Fe*=*. The 2 tourmalines presented in Fig. 9 show the extremes of variation in the ratio of the 470 and 745 nm bands. Assignment of the 765 nm band to Fe** in the Y site and the 4670 nm band to Fe" in the Z site with both sites absorbing at 1100 mm is based on the correlation of. Fe-O distance and the average eneray of the °T2 3 6 transition (Faye 1972). In the absence of independent data on the site distribution of Fe", approximate Y- and Z-site occupancies can be obtained from compositional data. A rough correlation between the 670 nm band and Z-site occupancy of Fe=* as indicated by the substitution Na* + Al =* = Ca=* + Mg@* (Fe=*) supports this assignment. In addition, comparison of Fe,Mg-tourmalines and elbaites provides further support of this assignment. Using the Correlation cresented by Faye (1972), Fe=t in the larger sites of Fe,Mg-tourmalines would be expected tos 77 absorb at lower eneray than in elbaites. The 720 nm band of eBlbaites can thus be associated with the 745 nm band of Fe,Mg-tourmalines. A ¥Y-site assignment is indicated by the general restriction of Fe=* to the Y-site in e@lbaites (Burns 1972; Faye et al. 19743. The assignments presented above conflict with Smith’s assignments in elbaites. Three Fe=" transitions are also present in elbaites, but it is the low energy band which is degenerate. At low temperatures the Fe** band at *1i200 nm in Elle is resolved into = components at 1680 and 1300 nm (Fig. 106). Smith (1978a) proposed that Fe=* in the Y site absorbed at 720 and 1300 nm and Fe=* in the Z site at 720 and 1080 mm. This would require a much larger absorptivity for Fe** in the Z site as compared to the Y site because of the predominance of Y-site occupancy in elbaites. Such a difference is inconsistent with the sizes of the sites (Fig. 4). In addition, the relative proportion of the 16080 and 1200 nm bands show an approximate correlation with Fe content. An increasing proportion of the i680 band with increasing Fe content is indicated in Fig. 1. Above i.5 Fe atoms per formula, Li-bearing tourmalines show a spectroscopy similar the Fe,Mg-tourmalines (Fig. 11). Fig. ida illustrates another difference between elbaites and Fe, Mg-tourmalines. At very iow Fe =* contents, both 1080 and 1200 om bands can be seen in Elc.' However, only the i080 na band is intensified. At higher Fe s* contents there is nso indication of the 1200 nm bands in Elc (Smith i?78ai}. if Fig. 78 WAVENUMBER, om! 29008 19889 7900 } I il CA> Ly 2@.3+ | O 2 << oO 28.2t 1 ro a a a.it 1 — CBD @.3+ 1 ABSORBANCE © N 520 182e 1588 2808 WAVELENGTH, nm 10 Absorption spectra of sample Tit at 296 F (broken line) and 82 KE (solid line). Sample thickness = G87 mm. (4) Ele polarization (B) Ellc polarization 79 the i0Bo and 1300 nm bands represent different sites, these results Suggest that Fe=* can be intensified in only one site. “This directly contradicts the behavior of | Fe, Mg-tourmalines. These large differences between the optical properties of Fe,Mg-tourmalines and elbaites are not expected from the slight structural differences between the end members of the tourmaline group. My preferred assignments to the i080 and 100 nim bands are that they both represent Fe, in the ¥ site but with differing degrees of distortion. At low iron contents, the Lioand Al neighbors of Fe=" in elbaites are smaller and may permit minor enlarging of the site for Fe producing a lowering of the energy of the optical transition. Neighbors such as Fe=* and Mg are similar in size and would not permit such adjustments. 4 shoulder on the 720 nm band is evident at high Fe contents in Li-bearing tourmalines (Fig. 11) and can be assigned to Fe** in the Z site. Application to Radiation Effects The extremely large absorption intensities af Fe=" in association with Fe * suggests that optical spectra of tourmalines should provide a sensitive probe of Fes". This is illustrated below for radiation induced changes. The effects of lonizing radiation are diverse and often difficuit to specify. Y-airradiation of Mn-bearins elbaites oxidizes Mn=* to Mn=* (Peinitz and Rossman, in prep.). & large variability in the accompanying color 80 Spy = AULT wayoIg Tz = aULy pLps ssuotjzezrseyoy ww AZO = S8aU4IT4] aydueg *(9a149S [40y9S-aztegra ayy UL 4oyIs ep ZZ1 afdwes yo esyrads uoryduosqe ¥ 9672 TT "bry WU SL LONATSAVM QB0Z ees} eee! BS oe a A D ae ee ee ee Se SO ee A ee Oe eG Ge See gees Be0L 9200! e000 »—e “ASSWNN3aAVvM GO O02 Sc JONYENOSaV Bi change (Nassau i?7sSi). however, indicates the complexity of ‘this process. In order to assess the influence of iron, several blue elbaites containing 32-42% Fel and MnO were subject to ‘Y-irradiation. These samples showed no evidence of the conversion of Mn=" to Mn=" at doses as high as 52 Mrads. Instead, small changes in Fe" absorption were noted {Fig. 12). The oxidation of Fe=* to Fe** is indicated by increasing absorption at 720 and i1100nm in Elec. These bands exhibit no discernibie change in Elle. Similar behavior was also observed in samples ERS and Mi which have similar Fe contents. The conversion of Mn=* to Mn=* was seen in sample EB? with FeO “1%. These observations suggest that neighboring cations can strongly affect the response of ions to irradiation. Voskresenskaya et al. (i979) alsa observed changes in the oxidation state of Fe upon Y-irradiatian using Mossbauer spectroscopy but abserved reduction of Fes below 0.35 Grads and oxidation above G.35 Grads. The changes iliustrated in Fig. 12 correspond to very smail increases in Fe ** which would not be resolved in Méssbauer spectra. The utility of Fe" absorption in tourmaline to monitor radiation dose has also been noted for neutron irradiation. Vargas and Tupinamba (1942) observed increasing absorption at *700 and “1100 mm with neutron irradiation and noted a iinear relation of neutron dose ta the log of the change in absorbance. 82 ALLA FE paysearg ‘spesy aatzIazya Q AUT] paysep ysoyg 4 6/8 ye buryseayg sazye spesy ppg saury paysep buoq UOTJEIPEIIE TEINJEU 4 SPEUY AHP FUT] Pl[oS ew gys"g = S5aUxITyY aTdweg “GaSOP WOTJL IPPs SNOTseA ye [EZ afdwes yo esziads uorydsosge paztuejod ITZ zy “614 wu “HLONSTSAVM BB8~ 88S I Bae l BeS “ 12 0 1¢7 9 19°89 Bool BERB I BBaAb~ 42 SASWNNAAVM AINVEeOSdv 83 Summary Tourmalines provide one of the most pronounced examples 0 “eh a recently discoverad aspect of Fe=*-Fes* interaction-—- intensified Fe=* transitions. Although the more familiar aspect of Fe@+-Fes+ interactions, Fe=*-Fe =" charge transfer, is also present, intensified Fe=* transitions are the dominant effect at all Fe =* concentrations. Fett? next to Fe=* was cboserved to increase the absorptivity of Fe** bands to ~1200 Mvtem7? from *S M7tcm7? for non-interacting Fe“. Approximately equal degrees of intensification were observed for both components of the ST2 3 SE Fe=* transition as weli as for Fe" in two different crystallographic sites. As with other ion-pair transitions, this effect is restricted ta the vector between the interacting cations. Significant differences between the optical spectroscopy of Fe,Mg tourmalines and elbaites have been related to structural and compositional factors. The basic characteristics of Fe=*—Pet* interactions, however, are constant within the tourmaline group. This extremely large change in absorption intensity implies that optical measurements of Fe** cannot be relied upon. However, because this effect is polarized, there may be useful unaffected polarizations such as Ellc in tourmaline. In addition, increases in intensity with decreasing temperature which occur in tourmaline may prove to be a consistent characteristic which can be used to detect intensified absorption bands. In itself, this effect B84 is a very sensitive probe of Fe** in tourmaline. Its general utility Must be assessed by similar studies in other materials. The predictability of intensfied Fe=+ transitions will require the development of a theoretical understanding. 85 REFERENCES “Amthauer, G and Rossman, GR (1984) Mixed valence of iron in minerals with cation clusters. Phys Chem Minerals 11:37-51 Bence, AE and Albee, AL (1968) Empirical correction factors for electron microanalysis of silicates and oxides. J Geol 746:382-403 Buerger, MJ, Burnham, CW and FPeacor, DR (1942) Assessment ef the several structures proposed for tourmaline. Acta Crystallogr 15:583-5970 Burns, RG (1970) Crystal field spectra and evidence of cation ordering in oGlivine minerals. Am Mineral 55: 1408- i632 Burns, RG (1972) Mixed valencies and site occupancies oF iron in silicate minerals from Mossbauer spectroscopy. Can J Spectr 17:51-5? Deer, WA, Howie, RA, and Zussman, J (1942) Rock—-forming minerals Vol. 1 OQrtho- and ring silicates, J Wiley & Sons, New York Donnay, G and Barton, FR Jr (1972) Refinement of the crystal structure of elbaite and the mechanism of tourmaline solid solution. Tschermaks Mineral Petrogr Mitt Faye , GH (1972) Relationship between crystai-field splitting parameter, " OS o2", and Myome-0 bond distance aS an aid in the interpretation of absorption spectra of Fe=*—-bearing materiais. Can Mineral 11:473-487 84 Faye, GH, Manning, PG and Nickel, EH (1948) The polarized opticai absorption spectra of tourmaline, cordierite, chioritoid and vivianite: Ferrous-ferric electronic interaction as a source of pleochroism. Am Mineral 53: 1174-1201 | Faye, GH, Manning, PG, Gosselin, JR and Tremblay, RJ (1974) The optical absorption spectra of tourmaline: Importance of charge-transfer processes. Can Mineral i232: 370-380 Fortier, S and Donnay, G (1975) Schorl refinement showing composition dependence of the tourmaline structure. Can Mineral 13:173-177 Goldman, DG and Rossman, GR (1979) Determination of quantitative cation distribution in orthopyroxenes from electronic absorption spectra. Phys Chem Minerals 4343-33 Gorelikova, NV, Perfil’yev, YuD and Bubeshkin, AM (1974) Mossbauer data on distribution of Fe ions in tourmaline. Zap Vses Mineral O-va 4:418-427 (transl. Int Geol Rev 20: 982-990, 1978) Hermon, E, Simkin, DJ, Donnay, G and Muir, WE (1973) The distribution of Fe** and Fe ** in iron-bearing tourmalines: A Massbauer study. Tschermaks Mineral Petrogr Mitt 197:124-i32 Loehr, JS, Loehr, ™, Mauk, AG and Gray, HE (1980) An electronic. spectroscopic study of iron coordination in hemerythrin. J Am Chem Soc 102:6992-6996 87 ‘Manning, PG (1969b) An optical absorption study of the origin of colour and pileochroism in pink and brown tourmalines. Can Mineral 9:478-490 Manning, PG (1969) Optical absorption spectra of chromium—bearing tourmaline, black tourmaline and buergerite. Can Mineral 16:57-70 Manning, FPG (1973) Effect of second—nearest—nei ghbour interaction on Mn =* absorption in pink and black tourmalines. Can Mineral 11:971-977 Nassau, K (1975) Gamma ray irradiation induced changes in the color of tourmalines. Am Mineral 46:710-713 Rossman, GR (1975) Spectroscopic and magnetic studies of ferric iron hydroxy sulfates: intensification of color in ferric iron clusters bridged by a single hydroxide ion. Am Mineral 60: 4598-704 Saequsa, N, Price, DC and Smith, G (1979) Méssbauer spectra of several iron-rich tourmalines (schoris). J Phys (Paris) 40:02/456-C2/459 Simon, HF (19723) Near-infrared and Méssbauer study of cation site occupancies in tourmalines. Unpublished MS Thesis MIT Smith, G (197Ga) A reassessment of the role af iron in the 3000-30000 cm-1i regicn of the electronic absorption spectra of tourmaline. Phys Chem Minerals 3:243-2723 Smith, & (1978b) Evidence for absorption by exchange-coupled Fe=*-Fe=* pairs in the near infra-red spectra of minerais. Fhys Chem Minerais 2:2375-383 a8 Smith, G (1980) Evidence for optical absorption by Fe2*-FeFr interactions in MgQ:Fe. Fhys Stat Sol A 61:1 91-Ki9S | Smith, G and Strens, RGIJ (1976) Intervalence-transfer absorption in some silicate, oxide, and phosphate minerals. In RGJ Strens, ed., The Physics and Chemistry of Minerals and FKocks, J Wiley &® Sons, New York, pp 3B3-612 Townsend, MG (1970) On the dichroism of tourmaline. J Phys Chem Solids 31:2481-2488 Vargas, JI and Tupinamba, GAC (1942) Tourmaline as a new thermal neutron dosimeter. Inter-Am. Symp. Peaceful Appl. Nucl. Energy 4th Mexico City 1, 6&7. Voskresenskaya, IE, Korovushkin, VV, Moiseev, EM and Shipko, PIN (1979) MaGssbauer investigation of Y-irradiated ferriferrous elbaites. Kristallogr 24:955-837 (transl. Sov Phys Crystallogr 244:480-481, (1979) Wilkins, RWT, Farrell, EF and Naiman, CS (1949) The crystal field spectra and dichroism of tourmaline. J Fhys Chem Solids 30:42-56 8&9 Chapter 4 Heteronuciear Charge Transfer in Minerals 90 Introduction Mixed valence ion-gairs of transition eiements are often present in minerals. Homonuclear ion-pairs which consist of @ single element in different oxidation states have been studied extensively, particularly with regard to the intense optical transitions which correspond to charge transfer between the cations (Robin and Day 1967; Creutz 19583). The complex chemistries of natural materials suggests that ion-pairs consisting of dissimilar eiements should be as common in minerals as homonuclear ion-pairs. Heteronuclear ion-pairs, such as Pd=*-Pt4** and Agt-Au=* ion-pairs (Allen and Hush 1947), have been studied in chemical systems to a much lesser extent and have consisted of mineralogically rare combinations. The observation of charge transfer transitions in minerals has been hampered by multiple, overlapping transitions and complex compositions in which several charge transfer transitions are possible. Fe<*-—Tit~ charge transfer should be a frequent component of mineral spectra, but the spectra of minerals such as biotite (Robbins and Strens 1972) and Ti-pyroxenes (Burns et al. 1974) exemplify these difficulties. Most assignments of more distinct transitions, such as those of blue sapphire (Burns i981), are complicated by low concentrations of the Cations involved and the presence of other transition element cations at Similar cancentrations. Transitions assigned to Fe**-Tit* charge transfer have encompassed large ut Variations in wavelength (400-900 nm) and bandwidth (Burns Fi i961). Increasing intensity with decreasing temperature has been utilized as the major criterion for identification. Very few unambiguous examples of Fe2@*-Tit* charge transfer are available to aid in the determination of these assignments. Charge transfer transitions are potentially useful as sensitive probes not only of specific ions, but also of their site occupancies and the extent of cation ordering. These applications require the correct identification of apecific transitions and a guantitative understanding of the relationship between intensity and concentration. Minerals containing the cations of interest in stoichiometric concen-— trations and edge- or face-sharing pair sites should provide reliable qualitative and quantitative characteristics of charge transfer transitions. These phases have not been studied previously because of experimental difficulties associated with the large intensities of charge transfer transitions. Similar data, however, can only be obtained from more dilute examples which span a large compositional range. This study will emphasize both concentrated Fe,Ti-minerals and a series of tourmalines containing large variations in Fe and Ti content to determine the characteristics of Fe?*-Tit* charge transfer transitions. Mn?*-Tit* charge transfer in tourmaline will also he characterized. 92 xperimental The stoichiometric Fe,Ti-mineralis which were selected include taramellite {(CiT# 68458) and traskite iCiTs# 7302) from Rush Creek, Fresno Co., CA and neptunite (‘GRR# 7/4/74) from San Benito Co., CA. The tourmaline samples are described in Table 3. Labels specifying zones of uniform color are indicated in Table = with the color descriptors. Differen-— tiation of the elbaite-schorl series from Fe,Mg—-tourmalines was -based on Mg content and characteristic GH vibrations (Appendix). Chemical analyses were obtained using a MACS-SAS electron microprobe and are oresented in Tables i and 2. Data reduction was accomplished using the methods of Bence and Albee (1968). Optical spectra were obtained in a manner similar to that described in Rossman (1975). The spectra of thin samples (0.005-0.030 mm) were generally measured while attached to a glass slide with epoxy. The thickness of these samples was determined using a microscope micrometer on polished, perpendicular sections. Low temperature spectra were obtained by placing the sample on a colidfinger immersed — in liquid nitrogen in a quartz dewar. Temperatures were monitored using a copper-constantan thermocouple placed sightly above the sample. Temperature variation during a spectrum was limited to 50 ta 86 KK. Room temperature spectra were measured in the same apparatus without moving the sample to insure quantitative comparisons. ae te] Tapbie i Electron Microprobe Analyses of Fe, Ti-minerals Sample weight 42 axides Ma2d Cad BaQ Fel M90 MnO TiGe Sit] Cl F Taramellite re) Oo 39.5 6.70 1.13 6.406 11.4 23.9 1.95 0 Traskite 0.14 0,50 48.2 5.77 0.146 1.28 7.12 245.1 4.23 0.5 Na2G K-20 Fel MgQ MnO Tite Side Neptunite 7.58 4.74 11.0 23.608 1.35 17.5 53.9 light zone Tabl e 2? Electron 74 Microprobe Analyses of Tourmalines ‘Samp le Na.o Cad Tio ~, or 2.61 0.26 2.24 1.22 2.51 0.81 1.98 6.85 2.24 0.72 2.59 G41 1.68 0.24 1.79 0.41 2e.ol 0.435 1.94 G.1i1 2275 0.237 a.77 4.25 1.977 1.106 MoO) Fel. MnO Tide 77 BBL BDL 6.85 0,33 2 BDL 0.13 53.74 0.46 2 BDL 1.46 5.10 9.24 3 BDL 4.79 3.558 0.78 3 BDL 3.49 1.51 0.62 35 BDL 4.32 1.47 09.04 3 f 0.09 13.5 4.95 0.098 3 es 4.87 98.45 O.07 0.35 23 6.235 6.398 0.03 i.01 8.01 7.79 0.04 0.28 7.73 6.73 0.05 0.53 3.11 9.16 6.28 0.34 3.94 6.82 0.15 0.31 3.54 11.4 G.14 1.03 2.79 10.8 0.14 0.40 11.4 0.46 BDL i.0° 14.3 0.38 BDL 9.58 19.1 0.08 0.01 0.49 So.2 NA 34.7 34.7 NA Soa 30.7 NA so.6 > NA 35.6 26.4 NA 34.8 36.1 NA sate 4 32.0 NA 24.7 24.5 NA Soe s2.0 NA afd 28.7 NA 36.5 35.2 NA 326.35 BBL BDL 0.4 —- below detection limits NA -- not analyzed 95 Results Tarameilite High Fe and Ti concentrations in tarameilite, Ba,(Fe,TijlaBsSieteell,, and its structure indicate the potential to unambiguously assign Fe=*-Ti** charge transfer. Fe and Ti occur in edge-sharing octahedral chains which lie along £160] and in which the nearest neighbor cations are 239 and 3460 om apart (Mazzi and Rossi i980). The polarization of charge transfer transitions along the vector between the interacting cations would thus predict that Fe?*—-Ti** charge transfer should have intensity ‘along this direction only. Figure i illustrates the optical spectrum of taramellite. There are several transitions which are predominantiy polarized along Cioej. The chemistry of this sample (Table 2) indicates that transitions due to Fe and Ti shouid be the major contributors to this spectrum. An approximate oxidation state distribution of 72% Fe?* and 28% Fe=* is indicated by analyses by Alfors and Pabst (1984) on taramellites from this locality. Thus Fe@*-Fes* charge transfer and intensified Fe@* transitions could also contribute to this spectrum. The transitions at 700 and 11356 nm have energies typical of crystal field transitions of Fe=*, A possible assignment to Fe@* intensified transitions (Chapter 3) is implied by their great intensity im [iool. Although it is not possible to predict precisely the energy of Fe**-Fe * charge transfer, the normal range of Fe?*-Fe>* charge transfer energies encompasses the 700 nm 96 aUET Uayosq -- (ANIL) KH © UIT prpas -- 743 He 400°) = SSAU¥ITY apdweg ‘azippawesez yo eszdads uotydsosqe paztuejod y 947 f ‘hry WU “HLONATAAVM PSUGT}EZ IE TOg G61 BBLI BBS! BBEl BBIl BOB BBL BBS VKWE —_— r t T =F po} = LE a a a es ee ee ee es ee ee ee eet meee ee pS CR RS CR A ON ES OO SO AS A UN SN SOU SS ML MS GO LV “as5No HSNa FAULT MI aWvav A 4. A. re nh r 4s. 4 1. t 2. 4, A. A t A BVOL Bee! J—WS “AIGWANSAYM QQB0¢ c: Q: 0 Cc FONVENOsay a7 , band and excludes the 4650 nm band (Chapter 3). Fe2+.Tj4+ charge transfer can thus be assigned to the 440 nm band. In addition to energy. the temperature response and width of transitions can be used as diagnostic criteria. Figure 2 illustrates the [100] spectrum of taramellite at 296 K and 82 K. Charge transfer transitions are generally expected to increase in intensity with decreasing temperature (Smith and Strens 1975). The 440 nm band, however, shows littie if any increase at 33 KF. In contrast, the FOO nm band, which has been assigned to Fe?*-Fae** charge transfer, shows a large increase in intensity. Smith (1977) also noted larger intensity increases in Fe@*-Fe>* charge transfer than Fe?*-Tit* charge transfer in more ambiguous examples of charge transfer. A potentially more useful characteristic of Fe?*-Ti** charge transfer is the large width of the 456 nm band (*9600 cm71}). This is distinctiy larger than the width generally observed for Fe@+-Fes+ | charge transfer (Chapter 5) and much larger than that of crystal field transitions (Smith and Strens 1976). A quantitative characterization of this transition reveals a large relative intensity. A molar absorptivity of “1300 Mv-tem 7+ was calculated. assuming a random distribution within the octahedral chains. 4 molar absorptivity of ‘450 M-*cm7* would be valid if each Fe** were in contact with two Tit’, However, this would require an ordered distribution oh of cations within the octahedral chain. No distinction between individual octahedra was determined by the structure 98 BULY ptyos -- 4 T8 AUT] UaxOug -- ¥ 9bZ rsaunqzesadual WH CONT) = S5aU4ITYy ajdwes ‘api [pawese} 40 pujaads vorzduosge paztsejod 73 Z ‘614 wu “HLONA TAAVM Ge8c BO8!l AG9!I Berl OGzcl Beal BES BES BBY rm A. A. i. A. 4. A ry re 5. A. 5. rl A. 4. 4. Geol | 8820} | BaB0C 42 “ASSWNNAAVM SINVEsOSEV 99 refinement (Mazzi and Rossi i960). The absorptivity of Fe?*-Tit* charge transfer in taramellite can be compared with absorotivities of Fe?'-Fet* charge transfer in minerais which range between 60 and 210 M-tem—? (Amthauer and Rossman i764). Neptunite Fe2*-Tit* charge transfer is also an expected feature of the optical spectrum of neptunite, KNasliFeaTieSieOe,a. Fea and Ti occur in two edge-sharing octahedral chains along nm $10] and €iiO] (Canilio et al. 1944; Borisov et al. 1765). These chains are interconnected through shared corners. The nearest cations within the chains are 214-328 pm apart, and the intercation vectors have substantial components in all optical directions. An ordered arrangement within the chains such that Fe?* and Mn=* alternate with Ti** was proposed to account for the piezoeslectricity of neptunite. iCanillo et al. 1946). Figure 3 illustrates the spectrum of neptunite in the visible region. The 415 nm transition can be unequivocally assigned to Fe?*-Ti** charge transfer. The simplicity of this spectrum aS opposed to that of taramellite can be related to the absence of Fe=* (Bancroft et al. i967). Relatively weak Fe=* transitions at 860 and 1270 nm (Rossman unpublished results) provide the only other features in the visible-near infrared region. The width and intensity of this feature confirm this assignment. The width at half—-height of the 415 nm transition is ~?000 cm7* in agreement with taramellite. The molar absorptivity is 100 WAVENUMBER, cn! + 30088 20000 (5208 | Ef + 4 T LU 1 ras’ / \ > 1.97 <— . — co r 4 ox 4 = Lo) . oA <r, 8 \ OS Ne ~ ~ ~ mw ee =o 308 488 568 688 788 WAVELENGTH, nm Fig. 3 296 k polarized absorption spectra of neptunite (a light brown zone). Ssatie thickness = 6,019 ae Polarizations: ERX -- shert dash ERY -- long dash idi “225 MTtem7? (calculated assuming each Fe has 2 Ti neighbors). The observed intensity distribution provides the only inconsistency of this assignment. Bue to experimental difficulties, gamma polarization is not shown. However, roughly equal intensities in gamma and beta polarizations were indicated. This intensity distribution is inconsistent with the structure of neptunite and the polarization of Charges transfer transition along intercation vectors. The average orientation of the intercation vectors within the octahedral chain in alpha:beta:gamma is 1i.2:1i:i.2. The only intercation vector which agrees with the observed intensity ratios is that between the chains in the corner-shared pairs which are 3°93 pm apart and have the ratio 1:16:14. Although there i585 insufficient data on corner-sharing pairs to exclude the possibility of charge transfer from this ion-pair, charge transfer from the larger number of ; edge-sharing pairs should predominate. This intensity distribution is confirmed by the pleochroic behavior of neptunite (Ford i909). Traskite The pleochroism and chemistry of traskite, BaoFeaTi2i(SiOs)12(0H,C1,F).-SH20, indicated another potential example of Fe*"*-Ti** charge transfer. The Elec spectrum Gf traskite is shown in Figure 4. The Elle spectrum was not available, but this direction is nearly colorless. The transitions at S30 and 950 can be assigned tea Fer ABSORBANCE 102. WAVENUMBER, cm! 20008 {5008 18808 358 45@ 558 65@ 750 858 958 1050 115@ 1250 WAVELENGTH, nm Sample thickness = O.O75 mm Temperature = 2976 K 193 transitions. The remaining feature at 440 nm can therefore -be assianed ta Fe2+_qia+ charge transfer. The “7000 cm width of this band is comparable to Fe**-Tit* in taramellite and neptunite. Malinovskiil et al. (1974) have proposed a structure in which Fe and Ti occupy octahedra 3/4 of which are in corner-sharing pairs which are oriented perpendicular to the c-axis. They have suggested the formula BazaAeBoCD i 2Si 12056 (Si20>).4(0,0H) solle-i4He0 where 4, &, C, and D are positions occupied by Ti, Fe, Mn, Ca, Sr, Mg, and Al atoms. The determination of this structure was heavily biased by the Ba sites. It is difficult to reconcile a previous analysis af traskite (Alfors and Putman 19745) and the microprobe analysis of this sample with this formula. This structure has therefore not been used to calculate the molar absorptivity. A minimum value of *75 Mc-4cm—? is based on the assumptions that all Fe is Fe=* and that every Fe is in contact with one Ti. Tourmaline Faye et al. (1974) and Smith (1778) have suggested that a broad transition in the region 400-450 nm (Fig. 3) in tourmaline could be assigned to Fe**-Tit* charge transfer. This assignment could not be confirmed by a correlation between Fe and Ti content and intensity. Tourmaline has the general formula A¥s2.5i1.01.8(805)-.(0H,0,F)4 where X is Na or fa, Y is Mo, Fe or Li-Al, and Z is generally Al. Unlike the phases discussed above, it does not contain staichicometric amounts Gf Ti, and in most gemmy tourmalines Ti concentra— 104 Le a a ee ee ee ee ee ee ee Se he ee ee ee ee WW OW" = SSAUyHITYR a Tdweg wu “HLONATSAVM eal} eOS! eee! Bail ee6 ee aes "6t4 Boe a3 bn rt BL. Bn A. S. A. rt 4. 4. a A. b. & 5. S°@ "I az oS SO Se ee ee ee ee ee ee ee ee ee ee Oe ee ee GOL 80a | BB00~c , 42 “YSEWNNAAVI | SONvasaosgv 1905 tions do not exceed 9.19 4% Tides. Such low Ti concentra-— tions are near the detecticn limits of electron microprobe analyses and are subject to large uncertainties. “Fes Mg- tourmalines generally possess higher Ti concentrations. The relationship between intensity of the 446 nm band and the concentration of Fe,Ti-pairs in a series of Fe,Mg— tourmalines was examined and is shown in Fig. 6. The lecalities and compositions of these samples are shown in Table 3 and Table 2. This correlation justifies the assignment te Fe?+-Tis+ charge transfer. A molar absorptivity of *4000 M7—tcm74 is indicated by the illustrated least squares fit to these data. The concentrations of Fe“*-Tit* pairs were calculated assuming a random distribution of Fe and Ti within the Y-trimer (Buerger et ali. 1762). Although transition elements aisa accur in the Z site, Y-site occupancy predominates. This Site preference is evident in the restriction of charge transfer intensity to the Elec polarization. Y-Y¥ and Y-Z ion pairs are criented perpendicular to the c-axis, and i-Z ion-pairs are oriented at 35° to the c-axis tiBuerger et al i942). Minor differences characterize Fe**-Tit* charge transfer in the two basic compositional series of tourmaline. Elbaites (Li,Al-tourmalines) may have a lower molar absorptivity (possibly as low as 40% of that stated above) as indicated by one sample with sufficient Fe and Ti toa compare to Fe,Mgo-tourmalines (solid symbol Fig. 4). 106 Fe/T i Charge Transfer in Tourmaline 1088 T 15 888 + = E LY . 14 0 L 0 682 Oo - C ej mee 5 400Fb 4 ) Me) 11 < ‘oO 13 Bo o 208 + Le Z a gi2 : 69,10 g { 1 L i @.002 8.05 @.18 2.15 8.28 9.25 Feet Tilt Pair Concentration Fig. & Intensity of the 440 nm band in Fe,Mg-tourmalines (open squares) and the 400 mm band in @elbaites (filled square) vs Fe,7i pair concentration im moles/liter. co pot we ro iT Table = far sample ksy.- 107 Tabie 3 Tourmaline Sample Descriptions Sampis Colar Locality Fig. 6 Archival Code Code $2 bliuek White Queen ma GRR 2/13/78 Mine, Fala, CA b=brown* Newry, ME 14 CiT# i372 $3 g=greenk 8 54 b=bluex Cecil Coa., MD 7 CIT# 3575 g=greenk ii 5S b=biues Anza Borrego, CAé 5 GRRH 2/14/33 g=greenk 4 54 orowns Fortland, CT is CiT# ios?y Sioc greenk Minas Gerais, is Brazil Di brown Yinnietharra, = CIT# 12483 Australia DBS brown Eryant Lake, NY i CiT# 12454 DS yellow East Africa io GRRE S/2/52 EGS green Afghanistan & GRRS 1/27/80 EG4 dark green Afghanistan >? GRRS 1/27/86 Tig yellow- Zambia GRR# Li/eR/ee green NF l=yellow Nepal GREHF 2/72/80 Slight brown sedark brown iz & Elec color in thin section 168 Fe**-Ti**? charge transfer also occurs at a higher energy 415 nm). in elbaites. Fe=*-Tia~ charge transfer in both groups increases in integrated intensity by iSk on going from 275 KF to B82 K iFig. 7}. Smith (1777) observed an increase in intensity of “25% at 5 K as compared to 293 K. The halfwidth of this transition is 8000-9000 cm-1 for both elbaites and Fe,Maq~-tourmalines. Mn=*-Tit* Charce Transfer Two unusual yellow eibaites with relatively high concentrations of Mn and Ti and extremely low concentrations of Fe afforded the opportunity to examine Mn=*-Tit’ charge transfer. Figure 8 illustrates the spectra of an eibaite which is zoned in Mn and Fe. An Fe rich zone exhibits Fee*-Tiae charge transfer at 415 nm. A second transition is evident in a zone with an Fe:Mn ratio of 1:4. Mn**-Tit+ charge transfer at 325 nm is clearly shown in a zone with trace amounts of Fe. Charge transfer energies can be related to the ionization potential of the donor ion and the electron affinity of the acceptor ion. Although such considerations are difficult to appiy to individual examples, the higher energy of Mn=*-Tit* charge transfer as compared to Fe?*-Ti** charge transfer can be correlated with a. higher third ionization potential for Mn than Fe. This relationship was alsa observed in a reflectance study of Fe and Mn doped MgTiOs ‘Blasse i981). The molar absorotivity of the 325 nm band, “450 M-*em-?, is much lower than that af Fe=*—Tit~ charge transfer in tourmaline. The width and 109 O86 | ee mee one ae a a na a RS an NO a SA aut, PryOS -- ¥ oR BUT] WaYOIg -- ¥ 947 saanzesadway we OF = SSauyITY] apdwes “[q aurpewuno, fo eszIads UoTydsosqe paztseyod Ipy ¢ “hry wu “HLONATSAVM GQL1 BOS!I BBE! BBIl BBE BAL BAS BBE Fe a Ss De SO SM Se DO Ue Se ee oe ce ee g00L = a0 gaae2 }-WO “AFEWNNSAAVM SONVEeosav ABSORBANCE 110 WAVENUMBER, cm: 30088 20008 | | a a3 4 a ' mi 3.0t 2.9 re ee eee 2.8 | rr or eee ee ee eee oe meee | 1.9 200 300 ti(it«CAR 502 600 WAVELENGTH nm Fig. & Ele polarized absorption spectra of toursaline NP. Saapie thickness = 0.150 an yellow zone (HPi} -- solid line © light brown zone iNP2) -- long dash line dark prawn zone {P23} -- long & short dash line iii temperature dependence of Fe=*-Ti** charge transfer and Mn?*-Tit* charge transfer, however, are similar. A small pda mcrease in intensity of Mn=*-Tit* at low temperature is ‘shown in Fig. 9: The halfwidth of the 325 nm band is “7000 cm7?. Both charge transfer transitions are aleo polarized in Elc. Discussicn with the exception Of intensity, Fe**-Tit* charge transfer transitions in taramellite, traskite, neptunite, and tourmaline exhibit similar characteristics. The energy of these transitions fall within a fairly narrow range, 400- 4650 mm, which 15 distinctly separated from the energy range of Fe@"-Fes* charge transfer (Chapter 5). The 7600-9000 cm—*? fhalfwidths are aiso distinctly larger than Fe*"*-Fet charge transfer as well as crystal field transitions which May occur in the same energy region. Minor or imperceptible increases in intensity with decreasing temperature were observed. Energy and haifwidth thus appear to be the most useful diagnostic characteristics for Fe**-Tit* charge transfer. Molar absorptivities varied between “1600 and ™~AIOO M-tem-?. These values are generally larger than or equal to_ those of Fe =*-Fe>* charge transfer in minerals (Amthauesr and Rossman i984) and larger than those of crystai field transitions. Thus the presence of Fe and Ti should generally have a large influence on color. The large Variation in molar absorptivity among different phases 112 WAVENUMBER, cm! 30088 20088 §=- 1/888 of Li Lj 6 € ms ad ABSORBANCE eel 258 358 458 558 658 | WAVELENGTH, nm Fig. # Ele polarized absorption spectra of touraaline Tid. Saagle thickness = 0.159 as Teaperature: 294 b -~ broken line 83 & -- solid line iis _ suggests that the quantitative use of these transitions will be limited to phases in which intensity can be calibrated against concentration. There are few unambiguous examples of Fe=*-Tit* charge transfer toa which the above data can be compared. Optical spectra of fassaite from the Angra dos Reis meteorite have been studied extensively (Mao et al. i977). Composition, polarization and pressure variations were used to assign a transition at SOG nm to Fe?*-Tit* charge transfer. This wavelength and a halfwidth of *9000 cm7?* are consistent wath the characteristics stated above. The most frequently cited mineral containing Fe?*-Tit* charge transfer is corundum. Blue or green corundums have been prepared Solely by simul— taneously doping Fe and Ti into Al2Os. Transitions at 540-590 nm, 7id nm (Ferguson and Fielding 19771) and 9700 nam {Gmith and Strens i976) have been assigned to Fe**-TitT charge transfer. The assignment of these transitions is tomplicated by a large number of potential transitions due to the presence of Fe t*, Tit’, Fe#=* and Tit*. The transition at 560-590 nm is at an energy close to the energies of Fe2*-Tit* charge transfer as well as the broad range of energies of Fe?*-Fet* charge transfer. The “S200 cm—? halfwidth (Smith and Strens 1974) is more similar to halfwidths observed for Fe?*-Fe=* charge transfer (Chapter 5); Gf the transitions which have been assigned ta Fe2*—-Titr charge transfer, this transition is the cniy ane which could be supported by my data. This comparisan is 114 limited by the presence of face-sharing ion-pairs in corundum as opposed to edge-sharing ion-pairs above. Another limiting fracter to the validity of previous assignments is the lack of consideration of intensified Fe=+ transitions (Chapter 33. Further work is required to assess the appropriate assignments in blue and green corundums. Similar dilemmas are faced in the interpretation of the optical spectra of minerals such as kyanite (Smith and Strens i976) and sillimanite (Rossman et al. i782), polymorphs of Al25ids, and dumortierite, AlzBO.(SiO04)<0. {Rossman, unpublished data}. Transition slements substitute far AL=* in these minerals. Trivalent cations generally predominate, and divalent and tetravalent cations must occur together for charge compensation. Thus Fe**-Ti**. pairs are generally accompanied by Fe=*-Fes* pairs, and multiple charge transfer transitions as well as intensified Fe=r transitions must be differentiated. Blue varieties of these minerals possess similar optical spectra with a band at ‘400 nm fhalfwidth ‘*S000 cm~?) that has been assigned ta Fe=*—-Tit* charge transfer. A comparison of these bands with the examples of Fe=*-Tit’ charge transfer presented in this paper would suggest that such assignments are unlikely. Further support for this contention is obtained from another colymorpn of A1l=2]5ids. A band at 480 nm with a halfwidth of 6500 cm7? (Smith and Strens 1974) has been assigned to charge transfer im andalusite. These Characteristics are similar toa those scutlined above and 115 / should be directiy comparable to these aluminum minerals. 116 REFERENCES Alfors, JT and Pabst. A (19784) Titanian taramellites in western North America. Am Mineral 69: 358-275 “Alfors, JT and Putman, BW (19465) Revised chemical analyses of traskite, verplankite and muirite from Fresno County, California. Am Mineral 30: 1500-1503 Alien, GC and Hush, NS (1967) Intervalence transfer absorption. Part 1. Qualitative evidence for intervaliance-transfer absorption in inorganic systems in solution and in the solid state. Frog Inorg Chem 8: 2357- ‘389 Bancroft, GM, Burns, RG, and Maddock, AG (1967) Uxidaticn state of iron in neptunite from Méssbauer spectroscopy. Acta Crystallogr 22:9234-925 Bence, AE and Albee, AL (1968) Empirical cearrection factors for electron microanalysis of silicates and oxides. J Geol 76:382-403 Blasse, G (1981) Charge transfer in mixed metal oxides: their consequences and influence on physical properties. Comments Inorg Chem 1: 245-2546 Borisov, SV, Elevtsova, RF, Bakakin, VV and Belov, NV (1945) The crystal structure of neptunite. Kristallografiya 1O0:S815-H21 (transi. Sov Phys Crystallogr 10:684-68%, Brown, DB, ed. (19806) Mixed Valence Comoounds. Theory and Applications in Chemistry, FRysics, Geology, ana Biology. Holland: Reidel, Bi? pp. “1i7 - Buerger, Md, Burnham, CW and Peacor, DR (19462) Assessment Of the several structures proposed for tourmaline. Acta gr 15:58s-s90 frat po if HO} re ory: a on i Burns, RG 991) Intervalence transitions in mixed-valence minerals Of iron and titanium. Ann Rev Earth Planet Sci Fi 345-285 Burns, RG, Parkin, KM, Loeffler, EM, Abu-Eid, RM and Leung, 1S (1974) Visible-region spectra of the moon: progress toward characterizing the cations in Fe-Ti bearing minerals. Proc 7th Lunar Sci Conf, Suppi 7, Geochim Casmochim Acta 3:2561-2578 Canilla, EF, Mazzi, F and Rossi, G (197466) The crystal structure of neptunite. Acta Crystallogr 21:200-208 Creutz, © (1963) Mixed valence complexes af d®-d* metal centers. Frag Inarg Chem 30:1-73 Faye, GH, Manning, FG, Gosselin, JR and Tremblay, Rd (1974) The optical absorption spectra af tourmaline: ‘importance of charge-transfer processes. Can Mineral 12: 370-380 Ferguson, J and Fielding, FE (1971) The origins of the ‘colours of yellow, green and blue sapphires. Chem Phys Lett 16:242-245 Ford, WE @1909) Meptunite crystals from San Benito County, California. Am J Sci 27:235-240 Malinovskii, YuA, Pobedimskaya, EA and Belov, NV (1974) Crystal structure of tras skite. Gokil Akad Nauk SSSR 229:1101-1104 itransl. Sev Phys Doki 21:424-425, i976 118 Maa, HE, Bell, PM and Virgo D (1977) Crystal-field spectra of fassaite from the Angra dos Feis meteorite. Earth “tl fot aanet Sci Lett 35:5352-356. ‘Mazzi, F and Rossi, G (i780) The crystal structure of taramellite. Am Mineral 45:122-128 Robin, MB and Day, F (1947) Mixed valence chemistry-—a survey and classification. Adv Inorg Chem Radiachem 16: 246-422 Robbins, DH and Strens, RGJ (1975S) Charge-transfer in ferromagnesian silicates: the polarized electranic spectra of trioctahedral micas. Mineral Mag 28:551-5435 Rossman, GR (1975) Spectroscopic and magnetic studies of ferric iran hydroxy sulfates: intensification of color in ferric iron clusters bridged by a single hydroxide ion. Am Mineral 60: 499-704 Rossman, GR, Grew, ES and Dollase, WA (1982) The colors of Sillimanite. Am Mineral 67:749-7641 Smith, & (1977) Low-temperature aptical studies of metal— metal charge-transfer transitions in various minerals. Can Mineral 15:500-507 Smith, G (1978) A reassessment of the role of iron in the 3,000-36,000 cm7* region of the electronic absorption te spectra of tourmaline. Phys Chem Minerals 3:243-37 119? Smith, G and Strens RGGI (19746) Intervalence transfer absorp- tion in some Silicate, oxide and phosphate minerals. in: The Physics and Chemistry of Minerals and Rocks, Strens, RGJ fed.}. New York: Wiley and Sons, pp. 5a83- 612 4120 Chapter 5 Characteristics of Intervalence Charge Transfer in Minerals iZi Introduction The intense color of compounds which contain both Fer and Fes* ions prompted the recognition of intervalence ‘charge transfer (IVCT) transitions more than 30 years ago (Allen and Hush 1967). In the case af Fe**-Fe =" compounds, these optical transitions correspond to the transfer of an electron fram Fe=" to Fe=*. Studies of a large number of mixed-valence ion-pairs in chemical compounds (Robin and Day i947: Alien and Hush i747; Hush 1747) firmly established the experimental and theoretical basis of IVCT transitions. These studies provided a basis for the recognition of Fe?*-Fe =" charge transfer transitions in mineral spectra. Smith and Strens (1976) and Burns (1981) have reviewed IVCT assignments in minerals. In spite of the predominance af Fe**-Fe=* ion-pairs, the certain recognition of Fe2+-Fes+ charge transfer transitions in mineral spectra is often not possible. The characteris- tics af transitions which have been assigned to Fe“*-Fet charge transfer encompass wide variations. in addition, another effect of Fe@*-Fe=" interaction, intensified Fe=* transitions, share many of the most frequently used characteristics of Fe**-Fet* charge transfer transitions iChapter 2). The polarization of these transitions along the vector between the interacting cations is a well established characteristic which can be used, with a uitable structure, ta differentiate ion-pair transitions fi rom transitions within the d leveis of a transition element by i2Z2 . ferystal field transitions}. The energy of Fe=*-Fe s+ charge ‘transfer transitions (Smith and Strens i974) and temperature (response iSmith 1977) have also been used as diagnostic criteria. These characteristics were established by empirical observations of limited examples of IVCT transitions in minerals. Most theoretical treatments of IVCT transitions which were developed for chemical systems da net support these characteristics (Creutz 19683; Day 1981).. The width of Fe**-Fet* charge transfer transitions has also been proposed as a diagnostic characteristic (Smith and Strens 19776) but is aften not considered. Many of the uncertainties regarding the characteristics of IVCT transitions in minerals are due to an inadequate data base of unambiguous examples and the lack of a suitable theoretical framework. Amthauer and Rossman (19784) have examined stoichiometric Fe**,Fe=*-minerais which should provide definite examples of Fe**-Fe>* charge transfer. The characteristics of these samples and previous, well-defined examples are examined in this paper ta evaluate the reliability of diagnostic characteristics. The characteristics af these examples will be compared to theoretical predictions. Characteristics which could be used to differentiate Fe**-Fe =" charge transfer from intensified Fe" transitions will also be discussed. i235 Experimental The temperature dependence of Fe=*-Fe** charge transfer was studied using samples previcusiy examined by Amthausr and Rossman (1984) and euclase ‘(GRR 1/15/79) from Rhodesia (Miami area). The samples provided by Amthauer and Rossman include rockbridgeite (LAH 7603) from Hagendorf, Germany, babingtonite (CIT# 9883) from Bombay, India, and lazulite {LA# 44508) from Graves Mt., Georgia. The compositions of these samples are given in Table = of Amthauer and Rossman .£1984). Chemical analysis af the esuclase was obtained using a MACS-SA5 eslectron microprobe. Blue zones contained 9.0424 FeO, 35.1% AleOs and 42.1% Side. Be and H20 could not be analyzed in this way. Data reduction was accomplished using the methods of Bence and Albee (1760). Optical spectra were obtained in a manner similar ta that described in Rossman (1973). Low temperature spectra were obtained by placing the sample on a coldfinger immersed in liquid nitrogen in a quartz dewar. Temperatures were monitored using Aa cCopper-constantan thermocouple placed sightly above the sample. Temperature variation during a spectrum was limited to 80 to 86 K. Reaom temperature spectra were measured in the same apparatus without moving the sample to insure quantitative comparisons. Magnetic susceptibility measurements of powdered samples of rockbridgsite and babingtonite were taken on a S.H.E. modei VTS magnetometer /susceptometer. Data for the empty Al sample container were used to correct the raw data. iZ4 Results and Discussion Charcse Transfer Energies Fe2+—FPe=+ charge transfer has been assigned to transitions which range in energy from 11500 cm7? (Ferguson and Fielding 1971) to 17500 cm-* (Rossman i974). The energy of a charge transfer transition is related to the ionization potential of the donor cation iFe=* in this case) and the electron affinity of the acceptor cation. Although theoretical Studies have made advances in the interpretation of the band area and the width of IVOCT transitions, the energy has not proven amenable to quantitative interpretation nor a priori calculation (Day 19681; Creutz 1983). Smith and Strens (19745), however, found a linear correlation between Fe**-Fet* and Fe**-Tit* charge transfer energies and the separation of the interacting cations in Silicate, oxide and phosphate minerals. These correlations have been used to assign transitions in minerals with known structures. Although the similarity of the nearest neighbor environment among these minerals may have left internuclear distance as the most important factor determining energy, the recent chemical studies referred ta above would suggest that factors other than nearest neighbor anions and catian Separation affect charge transfer energies. The correlation proposed by Smith and Strens can be reexamined in light of a large increase in the number of unambiguous examples of Fet*—-Pas* charge transfer in minerals. Figure 1 shows the relationship between the energy of Fe**-Fet* charge transfer 125 2+ _ 3+ | Fe /Fe Charge Transfer © 20 ——— ; I I8t — Q eo! oO I6t +e — _ * e° 1 _ g/ @ °° es & 4b +-@- 0 _ RS) | oe — 4 oD eh. 2 12F 4 Lid 2 IOF = 8 i | 250 275 300 325 350 Fé Fe’ "Distance (pm) Fig. i Fe=+-pes+ charge transfer transition energy vs internuclear distance. Squares: face-shared octahedra Circles: edge-shared octahedra Sample key: Table i ‘and the distance between the cations for this expanded data set. The sources of the data included in this figure are presented in Table i. No correlation between energy and distance is indicated. Although the specific energy of Fe?=*-Fe>* charge transfer cannot be used diagnostically, Fe?*—Fes* charge transfer transitions can be expected to fall between 106000 and 17000 cm7?, Effect of Temperature on Intensity Many other transitions, particularly crystal fieid transitions, occur in the large energy range of Fe=+-Fa=~+ charge transfer transitions. The variation of intensity with temperature has been used extensively to differentiate charge transfer transitions in mineral spectra. An increase in intensity with decreasing temperature is taken to be characteristic of charge transfer transitions in general (Smith and Strens 1976; Smith 1977). This belief is based on empirical studies of Fe?*-Fe=* systems. Confidence in the reliability of this behavior is diminished by a previous lack of recognition of intensitied Fe?" transitions which appear to exhibit the same temperature response (Chapter 3). in addition, the general theoretical treatment of charge, transfer transitions does not predict this behavior. The most familiar chemical example of Fe**-Fe ** charge transfer in Prussian Blue exhibits very iittie change in intensity {ierdi i780}. In accord with the FES theory (Wong and Schatz EPaensition shows no detectable increase in i ivyei; i 2 “P integrated intensity with decreasing temperature. However, iZ7 Sample Key to Fig. i Euclase Rockbridgeite Vivianite Babingtonite Lazulites Glaucophane Orthopyroxenes Biotite Enloritoid ha tl ate i Sources of ‘Optical Data Source oF Structural Data this paper pty Amthauer % Rossman (1°84) Burns (1970) a Smith (i?77) Halenius et al. (i781) Mrose & Appleman (1962) Moore (1970) Mori &® Ita (1970) Clark et al. (1949) Beran & Bittner i1974) fraki & Zoltai C1772) Moore (1970) Papike & Clark (1968) Burnham et ai. Ci?7i} Hazen & Burnham 1973) Hanscom ii?7) 128 the Fe?" ions in Frussian Blue are distinctly different from those in most minerals in one important sense, they are diamagnetic ions which cannot interact magnetically with their Fes neighbors. Cox (197580) and Girerd (1953) have suggested that magnetic exchange will alter the temperature response of charge transfer transitions. A larger data set of unambiguous examples of Fe**-Fe=* charge transfer is required to evaluate the reliability of an inverse temperature response as a diagnostic characteristic. Amthauer and Rossman (1934) have coupled Missbauer and optical studies of Fe=*-Fe=* minerals to quantitatively characterize Fe**-Fe=* charge transfer. The temperature response of Fe**-Fe** charge transfer in these samples is presented here. Lazulite, MgAl2(FO4)2(0His, contains Fe=* and Fe ?* ions in face-sharing Mg and Al octahedra (Moore 1970). A transition at 4665 nm has been assigned to Fe@*—Fe e* charge transfer. Figure 2 illustrates the gamma-polarized spectrum of lazulite at 276 K and B83 kK. The area of the 44558 nm band increases by 94% at 82 KE. This behavior contrasts sharply with the temperature response of crystal field transitions. The area of a crystal fieid transition sither decreases with decreasing temperature or remains constant. Figure = illustrates the beta-pclarized spectrum of suclase, BeAlSiOg0H. fron cations occur in edge-shared octahedral chains predominantiy occupied by aluminum (‘Mrose and Appleman i942). These spectra point out the difficuity 129 AUT] PIFOS -- YEH aut uayoug -- ¥ 94Z ssasnjeuadeay #0 740°9 = Ssauyatyy atdwes caqrpnzey yo eazrads uotydsosge paztseyod 7yy Z "bry wu “HLONATSAVM @SE1 eaSil @s6 = @SL Ss QSE Ls . 4LTNZv7 | | 80001 ee80c }-W2 “NIGWANSAVM | SINVasdOSdV 130 AULT PE[OS -- ¥ CB aut, waxO4g -- ¥ 947 ssauNnzRsadway mH GO") = S59u4I1y) aydweg casepIna yo e4syzIads uOTydsosqe paztaejod Tyz ¢ “bry wu “HLONATSAVM GQ61 QGL1 BBS! BBE! Ball Be6 BBL BES BOE bd | sepoyuy ASV TONG 4. lh. L | BBOL | —wo e000! 00002 “SSOWANSAYM AINVeaosev i3i of distinguishing Fe?*-Fe** charge transfer from intensified Fe@* transitions. At most Fe=*-Fet and Fe2+*-Ti4+ charge transfer transitions would be expected to contribute to this spectrum. However, three transitions exhibit intensity increases at low temperature. The two transitions at 860 and i250 nm have wavelengths typical of Fe** transitions and have slightly different intensity ratios in different polarizations than does the 470 nm band. The restricticon of Fes ions to one site and the width of the 4670 mm band (~4000 cm71) permits an assignment to Fe*t-Fe=* charge transfer. The area of the 670 nm transition increases by 20% at G2 kK. Similar increases characterize the 8460 and 12560 nm bands. The inhomogeneity of the lazulite and euclase samples did not permit a comparison of their temperature responses with magnetic behavior as measured by magnetic suscepti- bility. However, the magnetic and optical properties oF rockbridgeite, (Fe*",Mnd Fe *4,(PO4,}5(GHis, and babingtonite, Cas (Fe=", fin) Fe =*Sis0..0H, can be correlated and provide a Significant test of the theories of Cox (1980) and Girerd (1983). Figure 4 illustrates the variation of the intensity of Fe=*-Fet* charge transfer (980 nm in rockbridgeite with temeperature. A 10% increase in area is observed at 83 EK. The temperature response of Fe=*-Fe** charge transfer (480 mm} ain babingtonite is shown in Fig. 5S. This transition Maintains an approximately constant area at £976 and 53 &. Cox and Girerd both agree that ferromagnetic exchange should 132 BUTT PIOS -- ¥CH aT, VaxOIg ~~ ¥ 947 88 99070 = SSauyIIY} ardweg ‘azrabptaqyzoy yo euzrads uotyduosge paztaejod 7yz 4 "Oty | Wu “HLONATSAVM O86! BOLI BUSI BBE! BBll BBB BBL BBS BBE rsainjesadeal Ls rs es ee | t Ls re amen eee t be T 4LTAIGIYsAIOY “Tt T T T T Tv QB0L | —Wd -eeea| “SSSGWANSAAVM SONvaxosay 133 aU, PTTOS -- YeR auLy Wayosg ~- 4 742 WY (107) = SSAUXITYY afdues ‘aztuozburgeq yo esz3ads uorydsosqe paztuepad xyz ¢ “bry WU “HLONATAAVM rsaunjesaduay BOL! oOBSIlI BBE! BBII 686 OGL 86S Be es _—— v T v T “T “r s Om | . 412° .. 4 ° - 4 . - 9 ° 4 5 «4 g ° . 410 . F 4 . 412° r =S3LINOLONTSVa 7 be 7 1 2 r 4 rn t 2 1 a 4 t BBO e600 1 BB80¢ WS “YAGWNNSAVM | -BZONVaeOSSV 1iz4 ‘produce an intensity increase at low temperature and that vantiferromagnetic exchange produces an intensity decrease at iow temperature. Figures 6 and 7 illustrate the effective magnetic moment per iron atom at various temperatures for rockbridgeite and babingtonite, respectively. The nature of the magnetic interaction can be determined through a qualitative examination cf these data. An antiferromagnetic interaction below 360 FE is indicated in babingtonite. Between 275 and 83 EK, however, no magnetic interaction is evident. The lack of an intensity increase at low temperature is consistent with the theoretical expectation. Rockbridgeite exhibits multiple magnetic interactions which may be related toa interactions within and among different Clusters. A transition to a ferromagnetic region occurs at “100 K. The proposed effect af this behavior is confirmed by the observed increase in charge transfer intensity at low temperature. Although these results confirm the theories of Cox and Girerd, a consideration of Fe**-Tit* charge transfer indicates that other factors may also produce a temperature variation. Fe=*—-Ti*~ charge transfer generally exhibits Smaller changes in intensity with temperature than Fe**-Fae => charge transfer (Smith 1977), but these theories would not predict any change because Tit’ cannot interact magneticaily. A i5% increase in the integrated intensity of Fe®*-—Ti4+ charge transfer in tourmaline occurs for a temperature change of 296 to S82 K (Chapter 4). The failure 135 ROCKBRIDGEITE 7 1 Gg 5 o Sr o a _ o Lod “To 3 = Sg o 8 O o q oo 4 = 4k, 5ooo” o) o ke o Fr 3. Lil ag & < 2r = 4b Q : l { l i} {2g 202 308 TEMPERATURE Fig. & Temperature (F) dependence of the effective magnetic mament (B.M.) of roackbridageite 136 BABINGTONITE S j { fe hooosss 2 8 8 ooosd oa g Sk mS ' oO Don b g MAGNETIC MOMENT G} j j j ) 18B 220 322 TEMPERATURE Fig. 7 Temperature (h) dependence of the effective magnetic mament (CE.M.) of babingtonite 137 “to account for intensity increase of Fe=*-Ti** charge transfer transitions may be related to a possibile limitation of the general theories for intervalence charge transfer which are focused upon symmetric, homonuclear systems. Both experimental and theoretical studies indicate that IVCT transitions may not increase in intensity with decreasing temperature. Thus a decrease or lack of change in intensity: does not eliminate a charge transfer assignment. In addition, an increase in intensity does. not necessarily indicate a charge transfer assignment because intensified crystal field transitions exhibit the same behavior. Widths of Charge Transfer Transitions In a popular theoretical formulation by Hush (1967) the ZOO E bandwidth at half intensity (7 -- in cm7?) of an IVCT transition is related to its energy (&) by the following formula: rP2o= 2310 * (E — Eo) where Eo is the difference in zera point energy between the two cations. IVCT transitions between cations in identical sites {Eo = 0) which have energies between 10000 and 17000 cm-2 would thus be expected to have halfwidths of 4800-6500 cm7?. Theories which incorporate vibronic effects (Wong and Schatz i981) suggest that these values are applicable to nmsitions of systems in xt] strongly lacalized systems. TVET bre which discrete valences are not distinguishable have much smaller halfwidths (Creutz i987). The haifwidths of some -well established mineralogical examples of Fe?*-—-Fet+ charge transfer are shown in Table 2. The halfwidths of crystal field transitions may be as low as 160 cm7? and generally do not exceed 3500 cmt. Intensified Fe=r transitions maintain the energy and width of the related transitions of non- interacting Fe=*. Thus the evaluation of transition widths should permit the differentiation of IVCT transitions from crystal field transitions and intensified Fe=* transitions. &s can be seen in Table 2, the majority of Fe@t-Fes* charge transfer transitions have widths which agrees with that predicted by Hush. However, a few of the transitions assigned to Fe**-Fe** charge transfer have widths which are low enough to overlap with the widths of crystal field transitions. Delocalized valences could produce a lower width as indicated above, but Missbauer spectra of biotite (Annersten 1974) indicate very little delocaiization. Very different local environments of Fe2* and Fe=* could also produce a small decrease in width but would not be expected to be more significant in biotite than the other minerals presented in Table 2. It is possible that a better assignment of the transitions in biotite and chiorite would be intensified Fe=* transitions, but such an assignment eliminates Fe?*-Fe >" charge transfer from the spectrum. The relative influence of these two different effects of Fa**—Fat* interaction is not known. Thus, whiie a iarce halfwidth undoubtedly indicates a charge transfer transition, further studies are required to differentiate Table 2 i Widths of Fe=t-Fe=* Charge Transfer Transitions Mineral Energy lalfwi Reference {cm} tcm Lazulite 14900 Seoo Aamthauer & Ressman (1984) Euclase 14900 “400 this paper Rockbridgeite 10200 wooo " Babingtonite 14700 “S000 " Voltaite 14506 MEO Beyeridge &® Day tif7F:) Gl aucophane idi3a 4400 Smith & Strens (i974) Biotite 14300 Bis0 " Chiorite 141006 S575 " 140 Broad crystal field transitions from IVCT transitions. As was indicated in Chapter 4, Fe®*-Ti4* charge transfer transitions have halfwidths which range between 6500 and 7000 cm, Halfwidth iss definitely the most reliable criterion to identify these transitions. Summary Charge transfer transitions occur in mineral spectra together with crystal field transitions and intensified ui foal crystal field transitions. Many of the characteristics ascribed to charge transfer transitions have not proven accurate or unique. Transition energies can rarely be used to differentiate crystal field and charge transfer transitions. An inverse temperature response of intensity is shared by intensified Fe=+ transitions and is mot necessarily expected of charge transfer transitions. Polarization along the vector between the interacting cations is also characteristic of intensified Fe=-* transitions. Large halfwidths are generally the most reiiable characteristic of charge transfer transitions. However, some overlap with crystal field transitions cannot be dismissed. The response of crystal field and charge transfer transitions to pressure variations (Abu-Eid 19764; Fiaao 1976) Rave been examined and may prove to be a usefui diagnostic tool. As yet, however, intensified Fe=* Ta Ey transitions Rave not been studied at variable oressures. if) 5 “hi a = as rr combina Width, temperatures response and polarization behaviar should permit the differentiation of single ion 141 ‘transitions and ion-pair transitions. The differentiation Of intensified Fe** transitions and Fe=*-Fet* charge transfer transitions is limited primarily by the uncertainty regarding widths of charge transfer transitions. 142 REFERENCES Te Abu-Eid, RM 61975) Absorption spectra of transitions metal bearing minerals at high pressures. in: The FPiysics and Chemistry of Minerals and Rocks, Strens RGJ (ted.}). London: Wiley and Sons, op. 641-875 Allen, GC and Hush, NS (1947) Intervaience transfer absorption. Fart i. Gualitative evidence for intervalence-transfer absorptiscn in inorganic systems in solution and in the solid state. Frog Inorg Chem §:257- 289 Amthauer, G and Rossman, GR (1984) Mixed valence of iron in minerals with cation clusters. Fhys Chem Minerals 11:237-S1 Annersten, H (1974) Mosshauer studies of natural biotites. Am Mineral 239:143-151 Araki, T and Zoitai, T (1972) Crystal structure of babingtonite. Z Eristallogr i55:555-3753 Bence, AE and Albee, Al (1958) Empirical correction factors for electron microanalysis of silicates and oxides. d Geol 76:352-403 rl eran, ® and Bittner, H (1974) Untersuchungen zur Eristalichemie des Ilvaits. Tschermaks Mineral Petragr Mitt 2isii-2? - Beveridgs, D and Day, FP (1979) Charge transfer in mixed— valence solids. Part >. Freparation, characterization, and aptical spectroscopy of the mixed-valence mineral voltaite Caluminum pentairon(Ii} tri-iron(Iit} i4Z dipotassium dodecasulphate 18—-hydrate? and its solid solutions with cadmiuméII}. J Chem Soc, Dalton Tran i 1979: 448-453 Burnham, CW, Ohashi, ¥Y, Hafner, S5 and Virgo, D (i971) Cation distribution and atomic thermal vibrations in an iron-rich orthopyroxene. Am Mineral 546:850-876 Burns, RG (1970) Mineralogical Applications of Crystal Field Theory. Cambridge: Cambridge Univ. Press. 224 pp. Burns, RG (19813 Intervalence transitions in mixed-valence minerals of iron and titanium. Ann Rev Earth Flanet Sci 9: 345-385 Clark, JR, Appleman, DE and FPapike, JJ (1969) Crystal— chemical characterization of clinopyroxenes based on eight new structure refinements. Mineral Soc Am Special Faper 2331-30 Cox, PA (1980) Electron transfer between exchange—-coupied ions in a mixed-valency compound. Chem Phys Lett 497: 340-242 Creutz, C (i985) Mixed valence complexes of d=-d® metal centers. Prog Inorg Chem 36:1-73 Day, P (i981) Theory and experiments on valence delocalization in mixed-valence compounds. Comments Inorg Chem 1:155-167 Ferguson, J and Fielding, PE (1971) The origins of the Letit io: 144 Girerd, J-3J (1985) Electron transfer between magnetic ions in mixed valence Dinuclear systems. dd Chem Phys iF ‘ond “ 21746-1775 Halenius, i, annersten, H and Langer, & (197681) Spectroscopic studies on natural chloritsoids. Phys Chem Minerals 72417-1253 Hanscom, RH (1975) Refinement of the crystal structure of monoclinic chloritoid. Acta Crystallogr B21:780-784 £ Hazen, RM and Burnham, CW (i972) The crystal structures ar one-layer phlogopite and annite. Am Mineral 58: 689-900 Hush, NS (1967) Intervalence-transfer absorption. Part 2. Theoretical considerations and spectroscopic data. Prog Inerg Chem 8:371-444 Ludi, A (1980) Descriptive chemistry of mixed-valence compounds. In: Mixed-Valence Compounds. Theory and Applications in Chemistry, Fhysics, Geology and Biolcoey, Brown DB (ed.}. Holland: Reidei, pp. 25-47 Mao, HE (1976) Charge transfer processes at high pressure. In: The Physics and Chemistry of Minerals and Rocks, Strens RGJ (ed.). ‘London: Wiley and Sons, pp. 372-c81 Moore, PR (1970) Crystal chemistry of the basic iron phosphates. Am Mineral 55:125-169 Mori, H and Ito, T (1970) The structure of vivianite and symplesite. Acta Crystallogr 5:1-4 and euclase, BeAlSiO.0H. %% Kristallogr 117:14-36 145 Fapike, Jd and Clark,J (1968) The crystal structure and cation distribution of glaucophane. Am Mineral S3:1156-1173 Robin, MB and Day, P (1967) Mixed valence chemistry——a survey and classification. Adv Inorg Chem Radiochem 10: 248-422 Rossman, GR (1974) Lavender jade. The optical spectrum of Fe =~ and Fe?@+ — Fe"*+ intervalence charge transfer in jadeite from Burma. Am Mineral 59:548-876 Rossman, GR (1975) Spectroscopic and magnetic studies of ferric iron hydroxy sulfates: intensification of color ‘in ferric iron clusters bridged by a single hydroxide ion. Am Mineral 60: 499-764 Smith, G (1977) Low-temperature optical studies of metai- metal charge-transfer transitions in various minerals. Can Mineral 15:500-S307 Smith, G and Strens RGJI (1976) Intervalence transfer absorp— tion in some silicate, oxide and phosphate minerais. In: The Physics and Chemistry of Minerals and Rocks, Strens, RGJ (ed.3. New York: Wiley and Sons, pp. S8s- &i2s Wong, KY and Schatz, PN (1981) A dynamic model for mixed— valence compounds. Prog Inorg Chem 29:349-449 146 Chapter 6 Summary 147 The prevalence of ion-pair interactions in minerals i‘ which contain edge- or face-sharing octahedra is indicated by the number of distinct ion-pair transitions whic were recognized in tourmalines. These assignments were mads possible by the great compositional diversity of the tourmaline group. In general, however, the qualitative interpretation of mineral spectra relies upon the oe rn characteristic it of specific transitions which are relatively invariant. Minerals with stoichiometric concentrations of i) fi “is the desired cations can be used to obtain guantitative absorption intensities as well as reliable qualitative characteristics. These minerals were generally emphasized in this study. The pleochroism of tourmalines, particularly macroscopically black samples, is frequently related to what appear to be extremely anisotropic Fe=* transitions to the casual observer. An indication that they are not ordinary Fe= transitions is evident in intensity increases of these transitions in the most intense polarization upon partiai oxidation of Fe=* in the sampie. The quantitative relationship of intensity to Fe=*-Fe =" pairs was determined through combined. aptical and Missbauer studies. The intensity of Fe=* absorption in Fe**-Fe** pairs is ‘1200 Meftem—?., This is much larger than the molar absorotivities associated with the most distorted sites anc compares to a Value of “~S MU4*cem7? for non-interacting Fe=* in tourma ine. fed This large change in intensity should provide a sensitive 148 ‘probe of Fe * ions. Such considerations have been employed amma irradiation. The most rt ul oF or the eftfec ri ‘ho moni ra] useful characteristi which can be used to differentiate ni yi these transitions from unintensified bands are the polarization characteristics and the general increase in intensity with decreasing temperature. These distinctions are eesential if crystal field transitions are to be used Quantitatively. The differentiation of Fe**-Fet* charge transfer transitions from intensified Fe=* transitions is somewhat more difficult. In principis, the characteristic width of Fe" crystal field transitions which is maintained by the intensified transitions should be distinctly lower than Fe**-Fe** charge transfer transitians. However, a few transitions which have been assigned to Fe?*—-Fes* charge transfer have unusually low widths. The concentrated Fe, Ti-minerals which have been examined have been characterized by consistent features which should be easily differentiated from both crystai field and Fe**-FeS* charge transfer transitions. The wavelengths of Fe?*-—Tit* charge transfer in the samples examined in this study and those that have been previously assigned wit certainty lie between 400 and S00 nm They are also Characterized by extremely large halfwidths which range between S500 and 7000 cm71+. Some uncertainty regarding the assignment af transitions between S00 and 600 nm is produced when sapphire is considered. The difference in face-sharing and edeeae-sharing systems cannot be evaluated 9 g ¥ 149 presently. In contrast to the consistency of the width and eneray, absarptivities vary between £00 and 4000 Mvtcm7?. Thus although Fe=+-Tis+ charge transfer should be recognizable in mineral spectra, quantitative interpretations will have to be based on calibrations of specific systems. It is somewhat ironic that these transitions appear to have consistent features and that the characteristics of well established transitions were brought into question by this study. Intensified FeS* transitions have been studied axtensively in chemical systems and the characteristics of these transitions compare well with most mineralogical examples. The transitions of Fet* in tourmaline, however, bear little resemblance to them. The transition energy of one Fe t+ jion-pair is much lower than that of both unintensified and intensified examples of Fet* transitions and has a much larger width. The second possesses very unusual intensity ratios of components of the three Fe-* transitions. Other characteristics such as polarization and magnitude of the intensity increase are consistent among ali groups. These assignments were based upon optical studies of samples which were partially reduced and oxidized and comparisons within a group of Fa **-rich samples. In addition, a very similar spectrum is exhibited by coalingite. @n empirical correlation between cation separation and an) charge transfer energy has heen used diagnostically as weil i506 ‘aS an expected increase in intensity with decreasing ro emperature. & review af the characteristics of Fe?+-Fes+ charge transfer transitions with emphasis oan the relationship of intensity to temperature did not permit these characteristics and left halfwidth as the only generally reliable characteristic. The correlation of temperature response to magnetic exchange in 2 Fe**-Fet* minerals was found to support recent theoretical proposals by Cox and Girerd. The temperature response of Fe=*-Tit* charge transfer, however, could mot be accommodated by theory. 151 Appendix CH Vibrations as a Chemical Frobe in Tourmaline 1i5d2 Investigations of the infrared spectra of micas and other shept silicates (Farmer i974) and amphiboles {(Streans 1974) have demonstrated the variation of GH vibration frequency with the cations which are bonded to the OH group. Frequency shifts can be correlated to changes in the electronegativity and the charge of the cations. Increases in electronegativity and charge shift the GH stretching vibration to lower eneroy. These variations can be used to differentiate among the various tourmaline end members and the two basic tourmaline compositional series. Hydroxyl groups in tourmaline occur in two different crystallographic environments which are illustrated in Fig. 1. The labels in this figure are related to the formula K¥osZeSieQie (BOs) s(GH},4. One (6(137H) out of four OH ions per formula unit is coordinated to three Y-site cations. The remaining three (0(23)H) are bonded to one Y-site cation and two Z-site cations. Hecause the Y-site cations are generaliy divalent or have an average charge of +2 and the Z-site cations are generally trivalent, these two OH groups are separated in frequency. In addition, the 0(13H vibration is shifted to higher energy because the OH group points toward the X-site cation, further ifAcreasing this separation. Variations in the X-, Y-, and Z-site cations a5 well as hydrogen and X-site vacancies give rise to 4-7 distinct bands. “1 poco arized spectra of sali ry A Mi hed ac sections were obtainesd using a Cary iff spectrophotometer and Glan-Thomson calcite 153 TOURMALINE O:z @:x Fig. 1 A portion of the tourmaline structure selected ta show the environment of the OH groups. The thin wertical line is the c-axis. (Taken from Gebert and Zemann ¢i96S53 Table i i54 Sample Descriptions i le Calc Sample Color “Lacality Archival Code brown D2 ' brown Tié ‘colorless EC colorless ERB4 biue T22 Black Yinnietharra, Australia Bryant Lake, NY Madagascar Afghanistan Afghanistan Blue Lady Mine, San Diego Co., CA CiT# 12685 CIiT# 12484 CIT# 13657 GRRE 1/27/86 GRRH# 1/27/80 GRRE 3/10/85 Table 2 Electron Microprobe Analyses of Tourmalines Sample NazG CaO MgO FeO MnG Tide AlsO. ZnO Sid. F EC 1.39 6.28 BDL 6.08 1.970 BBL 41.58 NA 37.5 O19 EB4 2.75 0.30 BDL 3.67 2.17 0.02 37.6 0.32 37.1 1.8 T22 2.01 0.04 BDL 12.2 0.89 0.04 35.0 0.59 35.2 1.4 Tié O.75 3.72 BDL O.168 G.21 0.032 39.1 NA 37.4 1.5 Bi 207973 0.37 11.6 0.46 BDL 1.09 32.0 NA 37.2 O.1 B2 0.77 4.25 14.3 6.233 BDL G.58 28.7 NA 36.5 0.5 BDL -- below detection limits NA -- not analyzed iSs pGlarizers. Descriptions and compositions of the tourma— “lines discussed here are presented in Tables i and 2 —_— a Samples which were used for optical studies were too thick for examination of the fundamental vibrations Detween 2B00 and 3450 cm74. The first overtone vibrations between 7400 and 4800 cm—* (1350-1500 nm) which are much less intense were utilized. The E"“c polarization of these transitions indicates an crientation of the OH vector along the c-axis iGebert and Zemann 1745). Variations in the charge of the X-, Y- and Z-site cations produce the major OH bands in Fe,Mg-tourmalines. Figure 2 illustrates the spectrum of a dravite and Figure 3 the spectrum of a uvite. Both samples deviate from the end member compositions by containing “1% TiOz. The largest band (14230 nm} can be assigned to the most abundant group in both samples -—— Mg(YIAI(ZI2O(S)H. The Of1)3H bands are the lowest wavelength bands fexcluding the non GH band at 1316 nm) and can be assigned by comparison of these two spectra. Dravites and uvites differ.in the X-site cation toward which this hydroxyl group is directed. Dravites contain predominantly Nat in this site and uvites contain Cat. Their comparison indicates the assignment of CatXMgi¥)s0(1)H to the 1350 nm band and NaiX)Mg(Y¥)sQ(1)H to the 1370 nm band. Uvites aiso differ from dravites in the Zz Site. Excess Mg which enters the Z-site tnormally occupied (Dunn et al. i977a). Mg CYIMg i ZIAL(ZI0QCS3H groups are Fig. 156 HAVENUMBER, on” 7588 71988 L_ ——! 6582 | 8.6 ay bo] f= v ABSORBANCE cS NO 1308 149 1508 WAVELENGTH, nm ry Sample thickness = 0.680 ma mm 1628 2 Elle polarized absorption spectrum of tourmaline Di. 157 WAVENUMBER, cn 1588 7882 6580 | | 7 t ‘ 0.8 Q.47 ABSORBANCE —— es 1300 «——ti(i«* 1500 1600 ~ WAVELENGTH, nm Fig. 3 No polarized absorstica F Q. 3S Ello polarized absorption spectrum of tourmaline D2 Sample thickness = 0.728 mm iS8 expected to absorb at a lower wavelength than the 1420 nm band. _The band at 1410 nmin Fig. 2 can be assigned to this group. Na vs Cais also indicated by a slight shift of the 1430 nm band to lower wavelength in uvites. QOH bands which absorb in the region 1370-1416 nm are related to charge differences in the ¥ site and cannot be assigned by simple comparitcon. These spectra are distinctly different from these of the elbaite-schorl compositional series. Figure 4 illustrates the spectrum of a sample which is representative of the elbaite end member (X=Na Y=Li,f1 Z=Al). The most common groups in this composition are Li (Y¥)A1(Z)201(S)H and ALCYIAL(Z)20(S)H. The two largest bands can be assigned to these groups. The charge difference between these groups would indicate the assignment of the 1470 nm band to Al (¥)A1(ZI20(3)H and the 1425 nm band to Li (¥)A1(Z) 2015)H. This can be confirmed by comparison to liddicoatite (x=Ca Y=Li,Al Z=Al). Charge compensation for Ca is attained by excess Li and H-defects (Dunn et al. 19775). Excess Li is evident by the increase in the intensity ratio of the 1420 nm band and the 1465 nm band (Fig. 5). In addition to this difference, the effect of Ca can aiso be seen as shifts of the two major bands to lower wavelength. The higher wavelength shoulders Can be related to a small amount of Na. The influence oF La is much more difficult to see in the Q(i3H bands. Large amounts of 4n and Fe can substitute ints both elbaite-liddicoatites and dravite-uvites but do not 159 “WAVENUMBER, cm 1588 1888 6588 a 4 1 i T ! 4 ABSORBANCE 1300 1400 1500 (60 WAVELENGTH, nm Fig. 4 Ellc polarized absorption spectrum of tourmaline EC. Sample thickness = 0.254 mm 160 WAVENUMBER, on™ 7588 1808 6522 i i { my ’ i] ’ c |. @.6t ABSORBANCE B47 8.2F ~~ do. 1388 | 1400 {588 1688 WAVELENGTH, nm Fig. S Ele oolarized absorstion spectrum af tourmaline Tié Sample tnickness = 1,005 am igi Produce large changes in the OH spectra of the latter because they are generally divalent and substitute into the o. - A new transition develops in elbaites, however, at th Yeosit approximately the same position as the Mg(¥)A1l(Z)20(3)H band in dravites and uvites. Figure 46 illustrates the spectrum of an elbaite with intermediate Fe, Mn contents. The i437 nm band can be assigned to (Fe,Mn) (¥)Al(Z7}-20(3)H groups. Figure 7 illustrates the spectrum of a schorl in the elbaite-scherl series. Although the Fei¥)41(Z)s.0{(25)H band is the most intense feature, the 1425 and 1470 nm bands indicate the correct compositional group. The use of OH vibrations to differentiate Fe,Mg-tourmalines from elhaite-schori tourmalines depends upon the 1425 nm band specifically and will not be possible at compositions closer to the schoril end member. A hydroxyl band at 1470 nm can also be found in Fe,Mg-tourmalines if Fe =* is present or there is excess Al. The O(1)H groups in elbaites can be assigned to the remaining OH transitions between i350 and 1410 nm. The complexity of this region indicates a potential for more detailed chemical analysis. However, the large number of possible groups is too large for simple interpretations. More complete analyses including Li and H would also be re essential to interpretation. ABSORBANCE Fig. 162 WAVENUMBER, cn! 7500 7000 6500 q i Rd t i 8.6 8.4 a 8.27 doe 1388 1408 [28g 168g WAVELENGTH, nm 6& Elic polarized abeorption spectrum of tourmaline ER4. Sample thickness = 1,022 mm 163 WAVENUMBER, cn 158 7888 6588 v ? 4 Ef L at Q.2+ ABSORBANCE a Ce O.1F i. (300 =—ti(té‘«‘N 1520 1600 WAVELENGTH, nm Fig. Y Elle sclarized absorption se ctrum of tourmaline TEE. Sample thickness = O.252 mm 144 REFERENCES Dunn, Fo, Aocieman, D, Nelen. J and Norberg, J (18 nal ~ ca ai Wi Uvite, 2 newloid) nember of the tourmaline group and it implications for collectors. Mineral Record &:100-108 Dunn, PJ, Appleman, D and Nelen, JA (1977b) Liddiceoatite, a new calcium end-member of the tourmaline group. Am Mineral 52:1121-i124 Farmer, VCO (1974) The layer silicates. in: The infrared Spectra of Minerals, Farmer, VO (ed.}. (London: The Mineralogical Society, pp. 331-265 Gebert, Wand Zemann, J (1965) Messung des Ultrarcat- Pleochroismus von Mineralen Ii. Der Pleochroismus der OH-Streckfrequenz in Turmalin. Neues Jb Mineral Mh 1945: 232-235 Strens, RGI (1974) The common chain, ribbon, and ring Silicates. In: The Infrared Spectra of Minerals, Bi Farmer, VC ted.). London: The Mineralogical Society z 9 5