MSU EURNING MATERIALS: Place in book drop to LIBRARIES remove this checkout from “ your record. FINES will be charged if book is returned after the date stamped below. 7 1 SYNTHESIS AND CHARACTERIZATION OF IRON OXYCHLORIDE INTERCALATED BY ORGANOSULFUR ELECTRON DONORS By Susan Mary Kauzlarich A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1985 gm. 3- "7.3. 5;“, ABSTRACT SYNTHESIS AND CHARACTERIZATION OF IRON OXYCHLORIDE INTERCALATED BY ORGANOSULFUR ELECTRON DONORS By Susan Mary Kauzlarich In an effort to develop a new type of low-dimensional conductor, the intercalation chemistry of FeOCl with tetrathiolene molecules was explored. New intercalation compounds of the organic electron donors, TTF (tetrathiafulvalene), TMTTF (tetramethyl-TTF), TTN (tetra- thianaphthalene), and TTT (tetrathiatetracene) with the inorganic host FeOCl were prepared by direct reaction of solutions of the organo- sulfur compound with solid FeOCl. The new phases obtained were FeOCl(TTF) FeOCl(TMTTF) FeOCl(TTN) (toluene) and 1/8-5’ 1/13’ 1/21 FeOCl(TTT)1,9(toluene)l,23. 1/9 The pressed powder electrical conductivity of FeOCl intercalated with these tetrathiolenes is about 103-10“ times that of the pristine material (aRT(FeOCl)~10'5(n-cm)‘1). The temperature dependence of the conductivity is consistent with FeOCl and its intercalates being semiconductors with apparent bandgaps of 0.6 eV and 22- 0.3-0.4 eV, respectively. The highest conductivity (oRT~10'2 (n-cm)'1) is observed for FeOCl(TMTTF)1,13. Infrared spectroscopy indicates that the tetrathiolene molecule exists as a radical cation within the layers. Variable temperature magnetic susceptibility data for FeOCl indicate: that there is strong short range magnetic ordering between Susan Mary Kauzlarich 90 and 300 K. Comparison of the susceptibility data for the inter- calates with that of FeOCl suggests that short range magnetic inter- actions are also important for the intercalates. The combination of X-ray powder diffraction and EXAFS spectros- copy shows that these intercalates are, in general, well-ordered, crystalline solids. Upon intercalation the b axis (the interlayer distance) expands and doubles. In addition, alternate layers move one-half unit cell along the ag_plane, giving rise to extinctions in the X-ray powder diffraction data consistent with a body-centered unit cell. The X-ray powder diffraction data are consistent with the tetrathiolene molecule being oriented perpendicular to the layers, with the exception of FeOCl(TMTTF)1,13 in which the TMTTF molecule is oriented parallel to the layers. Detailed neutron powder diffraction and wideline 1H and 2H NMR studies were carried out on FeOCl(TTF)1,8.5, the best characterized and the most crystalline intercalate. The neutron powder diffraction data are consistent with a space group of Immm or 1222 for the inter- calate. A “soft" sphere model was proposed to account for the data; it assumes that sulfur atoms of the TTF molecule have a nonspherical electronic distribution that can be described by a pseudo-Sp3 hybrid- ization. The neutron data provide evidence for long range ordering of the TTF molecule within FeOCl. 1H and 2H wideline NMR studies indi- cate that there is more than one type of ordering of the TTF molecules in FeOCl(TTF) .5, characterized by different average TTF environ- 1/8 ments and/or dynamics. To my family and Larry ii ACKNOWLEDGEMENTS I would like to thank Professor Bruce A. Averill for his support and encouragement throughout the course of this work. I an also indebted to Dr. Boon K. Teo and Dr. John Faber for their helpful and generous collaboration. I would like to thank Dr. Jeff Ellena for his work on the NMR studies and for many helpful discussions. I am grateful to Professor James L. Dye for serving as my advisor at Michigan State University and to his group, 0. Fussa, M. Tinkham, M. Faber, S. Dawes and R. Huang, for many enlightening discussions and for their friendship. In addition, I would like to thank Dr. M. R. Antonio, Dr. H. E. Cleland and M. Rogers for their advice and help during the course of this work. I thank C. Hulse and E. Ponzetto for being great company and S. Hefler for sharing a lab with me. I would like to express my sincere thanks to my family and Larry who have provided me with considerable support and continuous encour- agement throughout this work. TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES SUMMARY OF ABBREVIATIONS INTRODUCTION EXPERIMENTAL Materials and Methods Preparation of FeOCl General Preparation of Intercalates FeOCl(TTF)x(toluene)y FeOCl TTF or d,-TTF)X FeOCl TMTTF)x FeOCl(TTN or TTT)x(toluene)y FeOCl(Pyridine)x SYNTHESIS OF FeOCl AND ITS TETRATHIOLENE INTERCALATES MOX Intercalation Solvent effects Effects of FeOCl decomposition and FeCl3 upon intercalation Preliminary results using other guests Preliminary results using hosts other than FeOCl STRUCTURAL STUDIES OF THE INTERCALATES X-Ray Diffraction Results Extended X-Ray Absorption Fine Structure iv Page vi viii xiv 12 12 18 21 21 23 23 23 24 25 26 29 3O 33 37 38 38 49 Page Hard Sphere Model 67 Neutron Difraction Results 74 Soft Sphere Model 90 SOLID STATE NMR STUDIES 97 1H NMR 98 13C NMR 116 2H NMR 124 ELECTRONIC PROPERTIES OF THE INTERCALATES 132 Infrared Spectroscopy 132 Optical Spectra 144 Conductivity 148 X-Ray Absorption Near Edge Structure 149 Magnetic Studies of FeOCl and the Intercalates 153 Variable temperatuare magnetic susceptibility 153 Neutron diffraction 157 SUMMARY 162 LIST OF REFERENCES 169 Table 10 11 12 LIST OF TABLES Chemical analysis for FeOCl and the FeOCl(0)x(S)y compounds where 0 = organosulfur and S = solvent Chemical analysis for two samples of FeOCl whose variable temperature magnetic susceptibility is shown in Figure 7 Kinetic data for FeOCl + TTF(0.24 M in dimethoxyethane) Chemical analysis for FeOCl(TMTTF)x prepared under different reaction conditions Calculated and observed X-ray powdEr diffraction data for FeOCl (Cu Ka1 = 1.54056 A using silicon as an internal standard) X-ra powder diffraction data for FeOCl(TTF)l,9 (tol 1,21, Co Ka = 1.7902 A X—ray powder diffraction data for the intercalates showing the calculated and observed d-spacings for the cell parameters given The distances and coordination numbers obtained from the EXAFS analysis for FeOCl and the intercalates Debye-Naller factors (A2) obtained from the best fit to the experimental data using theoretical functions Cell parameters, final positional, and thermal parameters for FeOCl at 300 and 10 K Cell constants and figures of merit obtained for twenty reflections from neutron powder diffraction data for FeOCl(TTF)1,8.5 Observed and calculated d-spacings (A) of the neutron powder diffraction spectra for FeOCl(TTF) with 8. the proposed indexing scheme: a = 3.7836(4) A? g = 25.9629(3) A and _c_ = 3.3416(4) A vi Page 20 27 32 34 4O 41 43 65 66 77 82 83 Table Page 13 Atomic positions refined for FeOCl(TTF) / .5 including carbons, space group = Immm (daga set length 3.80 - 1.0 A, R = 7.2%) 89 14 Definitions of the abbreviations used in the parameter listing for the NMR spectra . 100 15 Linewidths and relative areas of the single pulse lH spectrum of TTF 103 16 Linewidths and relative areas of the oo-r-ego 1H spectrum of TTF 105 17 Linewidths and relative area of echo 1H spectra for FeOCl(TTF)1,8.5 111 18 Position of the a v3 band observed in the Raman data with the valaes for the charge transfer 135 19 Infrared spectral features of FeOCl(TTF)1 8. and the TTF infrared absorptions of TTF(CA), (ssi 137 20 Infrared spectral features of FeOCl(d -TTF) ,9 and the dk-TTF infrared absorptions of du-TTFICA). 1 $5) 138 21 Infrared spectral features for FeOCl(TMTTF)1, 3 and the TMTTF infrared absorptions of (TMTTF)2X (i = BFu', CTOL‘, and PFG‘) 140 22 Infrared spectral features for FeOCl(TTT)1,9(tol) ,z3 and the infrared absorptions of TTT(X), X = Cl', r , and SCN‘ and TTT’ 142 23 Infrared spectral features for FeOCl(TTN)1,9(tol)1,21 and the infrared absorptions of TTN‘ 143 24 Two-probe pressed powder conductivity measurements for FeOCl and the intercalates 151 25 Effective moments calculated for high Spin Fe+3, FeOCl, and the intercalates 300 K using spin-only formula 156 vii Figure 10 11 12 13 LIST OF FIGURES The structure of TTF-TCNQ showing the segregated stacking of the TTF (donor) and TCNQ (acceptor) molecules A representation of the structure of (TMTSF) X showing the electronic density of the anion (X = PF6') A view of two layers of the FeOCl structure An illustration of the expansion of the TaS2 layers upon intercalation of octadecylamine Schematic drawings of the organosulfur molecules intercalated into FeOCl An illustration of the reaction flask used for intercalation Molar susceptibility vs temperature (K) for two preparations of FeOCl-Tsee Table 2) A schematic of FeOCl showing four unit cells X-ray powder diffraction data plotted as intensity vs 29 for (a) FeOCl, (b) FeOCl(TTF) I (tol) 21, IE) FeOCl(TTN)1,9(tol)1/23 and (o) FeOCl(TT 51,7 (a) Fe K-edge transmission X-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of FeOCl (a) Fe K-edge transmission X-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of FeOCl(TTF)1,8.5 (a) Fe K-edge transmission X-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of FeOCl(TMTTF)l,13 (a) Fe K-edge transmission X-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of FeOCl(TTN)1,9(tol)1,23 viii Page 11 22 28 39 47 50 51 52 53 Figure 14 15 16 17 18 19 20 21 22 23 24 Page (a) Fe K-edge transmission X-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of FeOCl(TTT)“7 54 Fourier transforms of the Fe K-edge transmission EXAFS k3x(k) vs r for (a) FeOCl, (b) FeOCl(TTF)1,8.5 and (c) FeOCT(TMTTF)1,13 57 Fourier transforms of the Fe K-edge transmission EXAFS k3x(k) vs r for (a) FeOCl(TTT)1,7 and (b) FeOCl(py)1,3.6 0" 8 Comparison of the Fourier transforms of the Fe K-edge transmission EXAFS k3x(k) vs r for FeOCl (solid curve) and FeOCl(TTN)1,9(tol)1,22'Tdotted curve) with peak assignments 59 (a) AE vs Ar for Fe--Fe nonbonded near neighbors and (b)_B_v_s_ o for FeOCl, FeOCl(TTF) (tol)1,21, FeOCl(TTN)1,9(tol)l,22 and for FeOCl(TTT)1,7 62 (a) AE vs Ar for Fe—-Fe (Efaxis) and (b) B vs a for FeOCl, TEDCl(TTF)1,9(tol)1,21, Feoc1(TTN)l;;Ttoi)l,22 and FeOCl(TTT)1,7 63 (a AE vs Ar for Fe--Fe a-axis) and (b) B vs a for e Cl,‘FEOCl(TTF)1,9(tol j},1. FeOCl( TN),;;Ttol)1,22 and FeOCl(TTN“7 64 Schematic representation of the orientation of the (top) TTF (bottom) TTT, TTN molecules predicted by the hard sphere model 70 Geometrical basis for equations 6-12 for TTF orientation based on the hard Sphere model 71 Geometrical basis for equations 13-15 for the TTN and TTT orientation based on the hard sphere model 73 Nuclear scattering (barns) y§_atomic weight 75 ix Figure Page 25 Profile refinement of the 150' TOF detector bank data for FeOCl at 300 K showing the low d region. The observed data are indicated by points and the calculated results by a solid line. Marks directly beneath the pattern indicate the positions of reflections. A difference curve appears at the bottom 78 26 Profile refinement of the 150’ TOF detector bank data for FeOCl at 10 K showing the low d region. The observed data are indicated by points and the calculated results by a solid line. Marks directly beneath the pattern indicate the positions of reflections. A difference curve appears at the bottom 79 27 Graph of sigma vs d including error bars obtained from the fit of FeOCl(TTF)1/8. 5 and Si at 300 K 81 28 Profile refinement of the 90’ bank for FeOCl(TTF) ,8 5 using only Fe, 0, Cl in the refinement (low d re 1on The observed data are indicated by points, and t e calculated results by a solid line. Marks directly beneath the pattern indicate the positions of reflections. A difference curve appears at the bottom 85 29 Profile refinement of the 90° bank for FeOCl(TTF)1 including carbons (low d region). The observed data5 are indicated by points, and the calculated results by a solid line. Marks directly beneath the pattern indicate the positions of reflections. A difference curve appears at the bottom 86 30 Profile refinement of 90' bank for FeOCl(TTF) including data to 1. 0 A. The observed data are indicated by points, and the calculated results by a solid line. Marks directly beneath the pattern indicate the positions of reflections. A difference curVe appears at the bottom 88 31 Illustration of the lateral shift of the chloride ion layers of FeOCl upon intercalation 91 Figure Page 32 Schematic representation of the possible orientation of TTF taking into account the lateral shift of the layers for both a (a) soft and (b) hard Sphere model 92 33 Schematic representation of the orientation of TTN and TTT taking into account the lateral shift of the FeOCl layers 94 34 Illustration of the TTF molecule on the bg plane showing the number of cells the molecule spans 95 35 (a) FT 1H NMR of TTF obtained from a single pulse radio frequency, (b) Expanded scale (Table 14 defines the abbreviations used in the parameter listing) 99 36 The 1H FT NMR spectrum obtained for TTF (a) after removing the first 100 us of the F10, (b) by sub- traction of (a) from the full FT NMR spectrum of TTF 102 37 1H FT NMR spectrum for TTF obtained by employing a 90-1-990 pulse sequence 104 38 1H FT NMR spectrum obtained for TTF by a 90-1-90 pu1se 106 39 1H FT NMR spectrum obtained for FeOCl(TTF)1,8.5 obtained by a single radio frequency pulse 107 40 1H FT NMR spectrum for FeOCl(TIF)1,8.S obtained by a 00-1-990 pulse sequence 108 41 1H FT NMR spectrum for FeOCl(TTF)l,8.s obtained 90-1-90 pulse sequence 109 42 1H FT NMR using the same parameters as those shown in Figure 37, drawn to the same scale as Figures 43 and 44 113 43 (a and b) 1H FT NMR for FeOCl(TTF)1,‘§.i hole burning 4 experiments using a field strength 0 5 Hz. Arrows indicate location of preirradiation 114 xi Figure Page 44 (a and b) 1H FT NMR for FeOCl(TTF)l hole burning experiments using a field strength of 43Hz Arrows indicate location of preirradiation 115 45 (a; 1H-coupled-13C FT NMR spectrum of TTF in CDCl3 b The expanded scale showing the peaks corresponding to TTF 117 46 13C CPMAS spectrum for TTF 118 47 13C CPMAS spectrum for TTFaBrZ-HZO 120 43 13c CPMAS spectrum for TTF(CA) (ss) 121 49 13C CPMAS spectrum for TTF(CA) (ms) 122 50 A schematic representation of a quadrupolar powder pattern resulting from a sterically rigid C-D bond 125 51 2H FT NMR of dk-TTF obtained from a single 90' pulse 126 52 The 2H quadrupole echo spectrum for FeOCl(dh-TTF)1/9 128 53 The 2H quadrupolar-dipolar echo spectrum for FeOCl(d -TTF)1,9 129 54 Infrared spectra of FeOCl and FeOCl(TTF) ,8 5 with assignments. Values for TTF(CA) are in parenthesis 133 55 The infrared spectrum for FeOCl(d -TTF) with assignments. Values for d“ -TTF (CA) are in parenthesis 139 56 Optical spectra of (flurolube mulls between NaCl plates) for FeOCl(---), FeOCl(TTF)1/8’ s (———), and TTF(CA)(ss)(-----). The baseline 15 1ndicated also (------). Absorption is an arbitrary scale 145 57 Optical spectra (flurolube mulls between NaCl plates) for FeOCl(------), FeOCl(TMTTF)1 FeOCl(TTN)1,9(toi)u 21(um) anleeOCl(TTT)1,9(tol)1/23 (---). Absorption 1s an arbitrary scale 145 xii Figure Page 58 2nd x§_1/T for two probe pressed powder conduc- tivity measurements 150 59 The Fe K-edge X-ray absorption edge spectrum for (a) FeOCl and various materials (b) and FeOCl and FeOCl(TTF)1,8,S 152 60 x molar vs T for FeOCl, FeOCl(TMTTF)“1 and FeOCl(TT'l—F')"1,8,5 with lines drawn througa the points 154 61 x molar vs T for FeOCl, FeOCl(TTN)1,9(tol)1,21 and FeOCl(TTTTl,9(tol)1,23 with lines drawn through the points 155 62 Neutron powder diffraction for FeOCl 300 K (top) and 10 K (bottom) data 158 63 Magnetic lattice for FeOCl (8 unit cells) showing the direction of spins proposed by Adam and Buisson 159 64 Neutron powder diffraction for FeOCl(TTF) 300 K (top) and 10 K (bottom) data 1/8'5’ 161 xiii 4-AP: BFBT: CA: CHESS: CluTTN: DBTTF: DME: DMTSF/DETSF: DTT: MeCN: EXAFS: FABM: GPPD: ILL: IPNS: Dy: TMTSF: TOF: TTF: TMTTF: TTN: TTT: TCNQ: tol: XANES SUMMARY OF ABBREVIATIONS 4-aminopyridine Best Fit Based on Theory chloranil Cornell High Energy Synchrotron Source tetrachloroTTN dibenonTF dimethoxyethane dimethyl/diethyltetraselenofulvalene dithiotetracene ' acetonitrile Extended X-ray Absorption Fine Structure Fine Adjustment Based on Models General Purpose Powder Diffractometer Institute Laue Langevin Intense Pulsed Neutron Source pyridine tetramethyltetraselenofulvalene Time of flight tetrathiafulvalene tetramethleTF tetrathianaphthalene tetrathiatetracene tetracyano-p-quinodimethane toluene X-ray Absorption Near Edge Structure xiv INTRODUCTION The traditional view that solid-state systems are an area of interest primarily to physicists has changed rapidly within the past twenty years. Chemists are actively investigating the synthesis and properties of solid-state low-dimensional conductive materials.1'5 The activity in this field was initially stimulated by the hypothesis that a one-dimensional organic compound could be designed that would exhibit room temperature superconductivity.5:7 Further stimulus was provided by the discovery of systems with high electrical conductivity (comparable to that of conductors such as copper) along one or more axes. These compounds have potential uses as antistatic coatings, as electronic and optical devices, as lightweight batteries, and as superconductors. There are basically three categories of well-studied low- dimensional conducting materials: (1) those containing linear poly- mers; (2) those containing linear chains of transition metals; and (3) those containing stacks of donor and acceptor molecules. Compounds such as K2[Pt(CN)“]O.3Br-3H20 (Krogmann's salt) fall into the first category; in these materials, it is the overlap of the transition metal g-orbitals that provides a pathway for electrical conduction.8 In the case of linear polymers such as (SN)x and (CH)ny (doped polyacetyiene), electrons can travel both along and between the chains of atoms.9’1° Doping of the polymeric metal phthalocyanines (Mch)n (where M = Al, Ga, Cr, Si, Ge, Sn and Co, Pc = phthalocyanine, and X = F, 0, S and CN)11'13 by iodine causes these materials to become highly conducting; the conductivity is attributed to the stacking arrangement of the phthalocyanine ligand as well as to partial charge transfer from the iodine to the macrocycle.1“ TTF-TCNQ (tetrathiafulvalene-tetracyanoquinodimethane) is the prototypical member of the last category. The high conductivity of this compound is due to the presence of segregated stacks of donor (TTF) and acceptor (TCNQ) molecules (Figure 1) and to the occurrence of partial charge transfer between the two.15 Compounds of this general type that crystallize as mixed stacks are usually insulators.16 Although a great deal of interest in organic metals has stemmed from their technological potential, their use in industry has been limited due to the fragility of the crystals as well as to the difficulties in synthesis. Synthetic strategies under investigation in other laboratories include the use of selenium and tellurium analogues of organic donors, more complex n systems, a variety of inorganic anions, and electro- chemical synthesis. All of these strategies have had some success. The use of the selenium analog of the organic electron donor molecule, TMTTF (tetramethyl-TTF), provided the first organic superconductor at anbient pressure, (TMTSF)2C10“ (TMTSF=tetranethyltetraselenofulvalene, see Figure 2).17 The critical temperature for this compound is 1.3 K.18 When the anion is replaced by other anions (TMTSF)2X (X = PFS', AsFG', SbFG', BF“' and N03'), the material undergoes a superconducting mmpzumpoe Acouqmuumv ozop use Acocouv mph we» do mcwxuopm umummmcmmm mzu mcpzozm ozupumph do mczuuzcum ash .H meam_d 6v. ,1 ’> \7 // OZUHlmPH [TMTSF 12X Figure 2. A representation of the structure of (TMTSF) X showing the electronic density of the anion (X = PFG') transition only under high pressures.19 The most recent addition to the class of organic superconductors is a triiodide salt of bis(ethylenedithiolo)tetrathiafulvalene (BEDT-TTF, referred to as "ET"), which is superconducting at 1.5 K at ambient pressure.20 The selenium-selenium or sulfur-sulfur interactions between the stacks of either the TMTSF or ET charge transfer salt are important in deter- mining whether the material will conduct and undergo a superconducting transition. The ET charge transfer compounds lack the columnar stacking that is the most prominent feature in (TMTSF)2X and other organic metals.20 It appears to form a two-dimensional network in which the interstack S-S contact distances are less than the sum of the van der Naals radii (3.6 A). Thus, the primary S-S interactions are between the loosely connected zig-zag stacks, not within a stack as in the case for (TMTSF)2X.20 In order to maximize the 5-5, or for TMTSF, the Se-Se overlap and thus superconductivity, fine tuning the structure of these compounds by varying the anions of these structural types is currently being pursued by a number of groups.2°'22 In addition to the structural constraints outlined above, a further criterion is that the electron donor (or acceptor) molecules achieve a nonintegral oxidation state in the material.2:“:23:2“ This is essential to provide a partially filled valence band and low energy for charge transport through the material.“v15’23'26 The combination of these two requirements, overlapping 5-5 or Se-Se inter- actions and partial charge transfer, significantly restricts the number of highly conductive materials. Although the (ET)I3 compound is superconducting, the reaction procedure results in more than one phase, of which only one is superconducting.20 A desire to enforce the formation of segregated stacks of cations led to the consideration of intercalation chemistry as a general route to new low-dimensional materials. Intercalation, in its simplest conception, can be viewed as the insertion of a species into a layered matrix by expanding one axis and preserving the identity of both the guest and host species.27 A variety of transition metal compounds are known to exhibit a lamellar structure in which layers containing metal atoms lie between parallel layers of chalcogen or halogen atoms.23'31 Examples are the Group IVB, VB and VIB transition metal dichal- cogenides (e.g., TaSz, ZrSz, NbSez, M082), the transition metal oxyhalides (e.g., FeOCl, VOCl), and the transition metal phosphorus trisulfides (e.g., MnPS3, FePS3); the structure of FeOCl is shown in Figure 3. In all cases the layers are held together by weak van der Naals interactions, which accounts for the fact that small molecules can be easily intercalated (or inserted) into these solids.27’29'32 The expansion of the layers of TaSZ by 51 A upon intercalation of octadecylanine is shown in Figure 4. This is rather an extreme example, but it illustrates the fact that preferred orientation of a guest molecule within the host layers can be achieved while preserving the identity of both.33 One of the important questions in intercalation chemistry is the orientation of the intercalant with respect to the host layers. Although in most cases direct structural evidence is lacking, for Nb523“ and TaSZ35 intercalated by deuteropyridine, neutron diffraction studies have shown that the pyridine is oriented perpendicular to the layers, with the lone pair electrons on the nitrogen atom directed Fe OCI orthorhombic , Pmnm o=3.7eoii, b=2917A, c= 3.3033 "" .‘IIIIF'.."--"" r n? 'r’ ‘1'“??{11 ‘0 . "r gigggfggé". “I ("'1 mfififif .11 ' L11 ' .111 t."t\\""fi\s$"g\‘f. . “Eiegfgitgfi 411? "if: P”%% .00 Figure 3. A view of two layers of the FeOCl structure 2.25 (3338 we :owum_oucmucw con: mcwamp «we» ms» mo :oPmccaxm ms» mo co_uccumsfip_ :< .e wczmwm parallel to the layers. A similar orientation has been proposed for pyridine intercalated into FeOCl.36 Studies on the cobaltocene intercalant of both TaSz,37:38 and FeOCl39 are consistent with an orientation similar to that of pyridine; the molecule appears to be oriented with the C5 axis parallel to the layers. Replacing the anions in (TMTSF)2X (Figure 2) with a layered material such as FeOCl provides visual evidence that a synthetic route to low-dimensional conductors might be successful. The use of a layered material can enforce the stacking of organic electron donors that have the potential to stack similarly to the pyridine or cobaltocene molecules. Since intercalation compounds generally adopt stoichiometries approximating close packing of the intercalant mole- cules, the resulting orbital overlap should allow facile charge transfer between adjacent molecules. Partial charge transfer from the guest to the host layers should result in a highly conducting mate- rial. Although the band structure for the metal oxyhalides has not been determined, a chemical analogy can be proposed. Thus, FeOCl, in which the Fe+3 is relatively easy to reduce, is expected to be most susceptible to intercalation by electron donors (such as TTF). Metal oxyhalides such as V001 and TiOCl, in which the metal is relatively easy to oxidize, may be most easily intercalated by oxidants (such as TCNQ). A wide variety of polar organic compounds"0 as well as organo- metallic species301‘*1"‘13 are known to intercalate into FeOCl, resulting in a significant expansion of the interlayer distance and increased conductivity (103-10“) over that of FeOCl (ORT = 10'7 (ohms-cm)'1). 10 As discussed herein, the intercalation chemistry of FeOCl has been expanded to include TTF and the other tetrathiolene molecules shown in Figure 5. The following new phases have been obtained: FeOCl(TTF)1,8,5, FeOCl(TMTTF)1,13, FeOCl(TTN)1,9(tol)1,22, (TTN = tetrathianaphthalene), and FeOCl(TTT)1,9(tol)1,23, (TTT = tetrathiatetracene). A number of physical techniques were used to obtain information on the properties of these materials. The tech- niques included X-ray powder diffraction, EXAFS (Extended X-ray Absorption Fine Structure) spectroscopy, infrared and optical spec- troscopy, variable temperature magnetic susceptibility and conduc- tivity studies on all the intercalates. In addition, detailed neutron powder diffraction and wideline 1H and 2H NMR studies were performed on the FeOCl(TTF)1,8.5 and FeOCl(d,.-TTF)1,9 compounds, which could be easily synthesized in large amounts and appeared to be highly crystalline. EXAFS and X-ray powder diffraction provide information on the short and long range order, respectively, of the host. Neutron powder diffraction and NMR provide information on the static and dynamic ordering, respectively, of the guest species. Electronic information was obtained by infrared and optical spectroscopy as well as from conductivity measurements. The magnetic studies help to ascertain the changes in the host lattice that occur upon intercala- tion. All these techniques, taken together, have led to a better understanding of this system and has provided some insight on the details of the reactions and the resulting properties of the inter- calates. The synthesis and characterization of these new materials will be discussed and evaluated as a route towards low dimensional conductors. 11 Puowu our; DwuwPMULmH—hw mopzuw—OE Lawpzmocmeo 05.... Lb mmcw3mgfl Uwuwcmzom Egg” mlm ukksh ,Prh I'm mlm kL1r .Ilm minim .m mcamwu EXPERIMENTAL Materials and Methods Preparation of all intercalates was performed under an inert atmOSphere of argon dried by passage through supported phosphorus pentoxide (Aquasorb). Solutions and solvents were degassed prior to use by repeated evacuation and flushing with dry argon. All other solvents were distilled once from appropriate drying agents, except acetone which was reagent or spectroscopic grade (Scientific Products). TTF, obtained from Parish or Aldrich Chemical Co., was recrystal- lized once from cyclohexane/hexanes.““ High purity TTF and TMTSF, obtained from Strem Chemical Co., were handled exclusively under inert atmOSphere. TMTTF,"s BDTTF,“‘5’“7 TTF(CA),"8 VOCl,"9’so and TTF38r2H2051 were prepared by published procedures. dg-TTF was pre- pared by a literature method52 substituting dl-EtOH for dl-MeOH. TTT, prepared by a published procedure,53:5“ was purified by Soxhlet extraction55 and subsequently by ten to fifteen sublimations at 210°C at 10"2 torr. DTT was obtained as a blue side product in the sublima- tion.5“ High purity 2H-TaS2 was a gift from F. J. DiSalvo of Bell Laboratories, Murray Hill, New Jersey. TTN and CluTTN were gifts from B. K. Teo of Bell Laboratories, Murray Hill, New Jersey. CluTTN was used without further purification, whereas TTN was purified by Soxhlet 12 13 extraction with benzene and recrystallized by addition of MeCN to the saturated benzene solution.56 CA was purchased from Eastman (practi- cal grade) and recrystallized once from toluene. 3-Bromo-2-butanone, stabilized with MgO, was purchased from Alfa Chemical Co. and dis- tilled prior to use in the preparation of TMTTF. Peracetic acid (35%), P“SIO(Gold Label), decalin, and perchloric acid (69-72%) were purchased from Aldrich Chemical Co., KSCOEt57 was prepared by a pub- lished procedure and these materials used in the preparation of TMTTF. All other reagents used in the preparation of published compounds were common reagent grade and used without further purification. The identity of the prepared tetrathiolenes was verified by melting point, infrared spectroscopy and elemental analysis. Solution electronic spectra were recorded on a Cary 219 spectro- photometer. Solid state spectra of Fluorolube mulls between NaCl or quartz plates were recorded on a Cary 1701 spectrophotometer at the University of Virginia. Infrared spectra were obtained on a Perkin- Elmer Model 1430 grating spectrophotometer at the University of Virginia, using KBr pellets (Harshaw Chemical Co., Solon, Ohio; stabilized infrared quality). Proton, deuterium and carbon-13 NMR spectra of TTF, dk-TTF, FeOCl(TTF)1,8.5, and FeOCl(du-TTF)1/9 were recorded by J. Ellena on a Nicolet NT-360 spectrometer at the University of Virginia. Digit- ization rates of up to 2 MHz were achieved with a Nicolet 2090 digital oscilloscope. All spectra were obtained at 20 1 2'C, chemical shifts are reported relative to TMS ((CH3)uSi). Proton spectra (361 MHz) were recorded with a standard Nicolet transmitter and probe. The 14 NT-360 was equipped with a Henry Radio Tempo 2006 amplifier and home- build probe for deuterium operation at 55.4 MHz. The 90' pulse width was 5 microseconds. Solution carbon-13 spectra were recorded at 90.8 MHz. Proton-carbon-13 cross polarization and magic angle spinning were used to obtain carbon-13 spectra of solids at 37.7 MHz on a Nicolet NT-150 at the Colorado State University Regional NMR Center by J. Frye. In all experiments the delay time between data acquisition and the following pulse was adjusted so that an increase in delay time had no effect on the Spectral lineshapes. Standard cyclopsseas9 phase cycling and data routing for echo experiments was the same as that described by Bloom et al.60 and Griffin.61 During echo experi- ments data acquisition was begun before the top of the echo and points collected prior to the echo peak were removed before Fourier trans- formation. The only spectra phase correction necessary for echo spectra was that required to correct for misalignment of the trans- mitter and receiver phases (no first order phase correction was used). Flipback pulses62 were used to accelerate data acquisition in CPMAS experiments. Temperature dependent magnetic susceptibilities were measured from 5-300 K on all samples at Michigan State University on a SHE Corp. SQUID susceptometer. The samples were run in a Delrin bucket able to hold approximately 50 mg of material. The empty bucket was run separately before each sample and the data were fit using KINFIT4, a nonlinear least-squares fitting program available at Michigan State. Selected samples were run and analyzed at the University of Virginia on a similar instrument by D. Holtman. 15 Temperature dependent two-probe powder conductivity measurements were taken with an apparatus (described elsewhere53) at Michigan State University in Professor J. L. Dye's laboratory. Constant voltage was supplied by a Heathkit Regulated Low Voltage Power Supply and the current measured by a 610 BR Keithley electrometer. All samples measured obeyed Ohm's law. Typical data acquisition of points from 25 to -60'C, cycled once, took twelve to sixteen hours. The current was measured every two degrees, and the temperature was measured by a thermocouple situated inside the cryostat cavity. Additional four- and two-probe single crystal conductivity measurements were obtained with the help of John Papioannou, a research associate in Professor Dye's laboratory. A two-probe conductivity cell, modeled on the same design as that used in Professor Dye's laboratory, was constructed at the University of Virginia. All measurements were conducted at room temperature. The samples were 3-4 mm long and were held in a 2 mm diameter precision bore Pyrex cell. The sample could be pressed to a maximum pressure of 22- 800 Pa. The constant voltage source was a home-made D.C. power source capable of supplying 1.3 to 24 volts. The current was measured by a Keithley Model 600 electrometer or a Model 177 microvolt digital multimeter. X-ray diffraction data were recorded on a General Electric powder diffractometer with either Cu or C0 Ka radiation using Pt powder (Aldrich 99.999% pure) as an internal standard. High precision data were obtained on a Huber Model 621 powder camera by asymmetric trans- mission and back-reflection using Cu Kal radiation and silicon (Hipure, Spex Industries, Metuchen, New Jersey) as a standard at the 16 University of Virginia Materials Science Department X-Ray Laboratory. At Michigan State University, X-ray powder diffraction data were obtained on a small radius Guinier powder camera in Professor H. Eick's laboratory or on a Siemens powder diffractometer in Professor T. Pinnavia's laboratory using Cu Ka radiation. Film from the Huber powder camera was measured on a Huber Model 622 film reader capable of 10.01 degrees accuracy. Lattice parameters were verified by using an X-ray program based on the relationship of spacing to reciprocal cell edges,5“ written by D. Holtman for a HP-87 minicomputer (Department of Chemistry, University of Virginia). Single crystal diffraction photo- graphs were obtained at the University of Virginia on a precession camera. I EXAFS (Extended X-Ray Absorption Fine Structure) and XANES (X-Ray Absorption Near Edge Structure) spectra were obtained at CHESS (Cornell High Energy Synchrotron Source) using experimental stations C-1 and C-2. Spectra were obtained on four different occasions on the same and different sample preparations to ensure reproducibility. The samples were diluted with boron nitride in a glove bag under nitrogen atmosphere, pressed into homogeneous pellets in aluminum cells (3x19 mm), and sealed with I-mil Kapton tape. The sample concentration was adjusted to obtain a value of ux = 1 in order to minimize thickness effects. All the data were taken in the parasitic mode with the electron storage ring operating at an average current of 10 mA and at energies of 4.7 or 5.0 GeV. All spectra were taken with a beam width of 15 mm and resolution of 1-2 eV. The monochromator was detuned by 50% to avoid contributions from higher harmonics. Typical spectra 17 of the Fe K-edge took 15-20 minutes and were taken in four regions during the energy scan in order to obtain constant k steps (A‘l). The spectra of the compounds were taken by transmission where the incident and transmitted beam intensities were measured by ionization chambers at 8 and 30 cm in length, respectively, and filled with nitrogen (flow type). The spectra were analyzed at Bell Laboratories, Murray Hill, New Jersey in collaboration with B. K. Teo, using programs described elsewhere,65 which were adapted to run on a DEC PDP Model 11/45 minicomputer. Neutron powder diffraction data were collected and analyzed at Argonne National Laboratory at the Intense Pulsed Neutron Source (IPNS) in collaboration with J. Faber, Jr. The samples (FeOCl, FeOCl(TTF)1,8,5, and FeOCl(du-TTF)1,9), were run at room temperature and 10 K in a vanadium cell placed in a displex unit on the General Purpose Powder Diffractometer (GPPD). The structure was refined using Rietveld profile analysis program.66 Initial values of the unit cell parameters for the intercalates were determined from X-ray powder diffraction data. For FeOCl, initial atomic positions were taken from single crystal X-ray work.57 The FeOCl structure was refined satis- factorily using 21 parameters in the final refinement (29 zero, back- ground, half width parameter, lattice parameters, an asymmetry parameter, a scale factor, and an anisotropic parameter). The struc- tures of FeOCl(TTF)l,8.5 and FeOCl(dk-TTF)“9 were refined using only Fe, 0 and Cl atoms over a limited data range (1.359 - 3.797 A). Initial atomic positions were calculated from an idealized model39 using Immm or I222 as a space group. It was found that 1222 with 2 18 Fe, 2 O and 2 Cl atoms gave the best weighted profile R: R = 8.773% (Rexpected = 1.997%). A Fourier difference map showed signifi- cant density between the layers; it was necessary to include a repre- sentation of the intercalated TTF molecule. Upon inclusion of the carbon atoms of TTF a much larger data range could be fit (occupational parameters of the carbons were allowed to vary). Mass spectra were obtained as a service at either Michigan State University by E. Oliver or at the University of Virginia by E. K. Johnson on a Finnegan MAT, Model 4600 GC/MS. Elemental analyses were performed by Galbraith Laboratories, Inc., Knoxville, Tennessee. Preparation of FeOCl The overall reaction scheme is shown below. The two methods used a-Fe203 + 4/3 FeCl3 ___» 3 FeOCl are as follows. Method A. a-F8203 (Baker Chemicals) was dried overnight under vacuum over CaSOL at 120'C. The Pyrex reaction tube (1 = 14.5 cm, id = 1.5 cm and V = 25.62 cc) was cleaned with ethanolic KOH overnight, rinsed with 6N HCl, then 95% EtOH, and dried in an oven at 110°C overnight. In a drybox, the reaction tube was charged with 4.22 gm of a-Fe203 and 5.18 gm anhydrous FeCl3 (Aldrich). The tube was sealed under dynamic vacuum (10'2 torr) and the contents mixed well by a vortex stirrer. The tube was placed in a Lindberg one-zone furnace at 350‘C with the empty end of the tube outside the heating coils in order to produce a temperature gradient. After one week, the reaction tube was allowed to cool to room temperature inside the furnace. Large red-violet crystals were deposited at the cool end and 19 microcrystalline powder was deposited about half way down the tube. In addition, green FeCl3 crystals were formed at the cool end in some preparations. The reaction tube was opened in air, the crystals separated and washed on a coarse glass frit with reagent grade acetone to remove FeCl3. In addition, the frit allowed any unreacted a-F9203 to be removed by acting as a sieve. The solid typically was washed with approximately 1-2 liters of acetone and dried over CaSOu under vacuum overnight. The yield was typically 95-IOO%. Method B. a-Fe203 (52.70 gm) was dried overnight at 110°C and added to anhydrous FeCl3 (71.38 gm). The reactants were transferred into a quartz tube (1 = 60 cm, id = 5 cm, V = 1.178 x 10"3 cc, cleaned in the same manner as Method A) in an argon atmosphere glove bag. The tube was evacuated for two hours and sealed under dynamic vacuum (10'2 torr). The tube was placed in a Lindberg three zone furnace and the temperature was raised to 370°C at the reactant end and 350°C at the other over a period of an hour. After one week the temperature was raised to 390°C at the reactant end and reduced to 310°C at the other. After another week, the tube was allowed to cool to room temperature inside the furnace. The yield was 99-IOO%. From the cool end, large crystals were obtained, and from the hot end, microcrystalline powder. The powder was washed on a medium glass frit with about 5 liters of reagent grade acetone, then with absolute EtOH (1 liter). The solid was placed in a vacuum oven at 110°C overnight, and subsequently in Schlenk tubes under argon. The elemental analysis is listed in Table 1. 20 .cmmosaw: acmucmao mo.m mm.me ~m.e~ om.~ mm.mH eHo.m o~.me me.- eH.H H3.NH - - e.M\H sq .ooou ek.m ee.km mo.mm “m.o mm.eH ao.a He.~m mm.mm km.o mN.NH MN\H _oo m.m\H Fee _ooou o~.oH em.oe mm.e~ ee.o em.~H ~o.oH -.oe mm.m~ mm.o mH.~H ~N\H _oe NN.¢\H zee _uoou mm.“ mH.ee ee.- 3~.o A we.“ mm.me mm.~m mk.o eN.~ - - mH\H ueeze _uoou NH.m ee.~e ofi.A~ Ho.o Ne.“ mm.m em.me o~.em mm.o oH.m HN\H .oo HH\H ace _uoou ~3.HH mfi.me ko.e~ ce.o mm.e me.H~ mH.Ne ma.e~ em.o me.e - m.m\H uee .uoou mm.~m Ne.mm mo.Nm eo.mm _uoou ma oux _ux Ia ox Ameov » m x c AoPeov ucm>Pom u m new campsmocmmco u o oeoez meeaoaeoo zflmvxflov_uoou use new Poood toe mwuapeee _eo_eoeu .H o_eee 21 General Preparation of Intercalates FeOCl (100 mg) and the intercalant (100 mg) were weighed and transferred in a dry box to a Teflon stoppered reaction flask (Figure 6), which contained a teflon coated stir bar (10 x 4 mm). Approxi— mately 15-25 ml of an appropriate solvent was added via cannula or anaerobic syringe to the reaction flask. The reaction mixture was degassed; the flask was evacuated and sealed via the Teflon stopcock. The flask was wrapped with aluminum foil to ensure uniform heating as well as protection from light (TTF is photosensitive). It was placed in an oil bath whose temperature was controlled by a variable resistor and had typical fluctuations of t5'C. After a period of one to Six weeks, the reaction flask was removed and the mixture transferred to an anaerobic glass frit under argon. The black microcrystalline solid was washed with freshly distilled solvent and subsequently with degas- sed reagent grade acetone. The solid was dried at room temperature for twelve hours under vacuum. Intercalation was verified by weight gain, X-ray powder diffraction, mass spectrometry, and elemental analysis. Elemental analyses for the intercalates obtained are listed in Table 1. FeOCl(TTF)x(toluene)y 25-30 ml of toluene was added to the FeOCl and TTF solids (mole ratio (2:1)) to produce a bright yellow solution. This reaction takes a minimum of one week at 70°C. The absorption peaks in the optical spectra of the acetone eluant could be assigned to TTFT.68 In 22 l2.5 crn ’1 2.75crn ‘ 1 v Figure 6. An illustration of the reaction flask used for intercalation 23 addition, there was a strong peak at 360 nm which was assigned to FeCl3. 1/9 < x < 1/12, 1/13 < y < 1/25. FeOCl(TTF or dg-TTF)x 5-7 mM concentrations of intercalant in ONE solution were used in these reactions. The mole ratio of FeOCl to intercalant was 2:1. The reaction took 10 days at 50°C. The optical spectra of the acetone eluant consisted of absorption peaks due to TTF+ and a very small amount of FeCl3. No solvent molecule was detectable by mass spec- trometry. 1/8 < x < 1/9. FeOCl(TMTTF)x TMTTF is only slightly soluble in MeCN. Therefore, a saturated solution was prepared by dissolving TMTTF in a minimum amount of warm MeCN. This solution was transferred via cannula to the reaction flask. The mole ratio of FeOCl to TMTTF was 2:1. Once in contact with solid FeOCl, the solution gradually turned from yellow-orange to yellow-green color. The reaction took a minimum of 13 days at 55°C. Optical spectra of the MeCN eluant indicated the presence of TMTTFT.52 1/13 < x < 1/14. FeOCl(TTN or TTT)x(toluene)y Mole ratios were 6:1 and 3:1 for TTN and TTT intercalants, respectively. These compounds are sparingly soluble in only a few solvents such as DMF, hot toluene and methylene chloride. There was always undissolved intercalant in the reaction mixture, which was heated at 70°C. The TTN and TTT reactions took 28 and 45 days, 24 respectively. Both reaction mixtures were washed thoroughly (Soxhlet extraction with methylene chloride for 3 days or 8 to 10 hours of continuous washing with toluene). As with the other acetone eluants, the optical spectra could be assigned to the radical cation.69 1/7 < x < 1/9, 1/19 < y < 1/26. FeOCl(Pyridine)x The preparation of FeOCl(py)x where x = 1/3 or 1/4 is a published procedure.70 It was found that the pyridine solution turned yellow over a period of time; this depended on the FeOCl preparations. Even FeOCl samples of very high purity (measured by low temperature variable susceptibility) turned the pyridine solution yellow after one week at 70°C. The optical spectra could be assigned to FeCl3. x = 1/3.4. SYNTHESIS OF FeOCL AND ITS TETRATHIOLENE INTERCALATES Absolute criteria regarding an acceptable purity for "single phase“ products are not possible, since results depend greatly on the particular compound and the physical methods used to characterize them. For example, the Debye-Scherrer method for powder patterns may give inadequate resolution71 and can easily miss up to 10% of a second phase. Unfortunately, a significant nunber of articles explain at length the physical properties of certain materials, but provide little or no information regarding the preparation or the phase purity of the material studied. There are many articles on intercalation compounds of FeOCl, in which the elemental analysis show serious discrepancies between the calculated and observed percentages. In addition, often little or no X-ray powder diffraction data are available. This chapter discusses some of the problems encountered in the synthesis of the tetrathiolene intercalation compounds of FeOCl. These observations do not constitute a rigorous study but are derived from the results of many different preparations with varied time, solvent, temperature, etc. of the reactions. 25 26 MOX There are three procedures outlined in the literature for the preparation of FeOCl: (1) the reaction of Fe203 with HCl,72 (2) the reaction of FeCl3 with water or water-saturated oxygen73’7“ and (3) the reaction of a-F8203 with FeCl3. The last reaction has been reported with various preparation temperatures and ratios of reactants and reaction times.57:75:76 It has been called a vapor transport reaction,76 but there is little evidence for this type of mechanism. In the absence of water, the product does not deposit at the cooler end of the reaction tube. Absence of any volatile FerCly prod- ucts in addition to FeOCl indicates that this is not a vapor transport reaction. In this work, careful drying of the a-F6203 has been found to be important in order to obtain high purity FeOCl. Although a temperature gradient was generally used in the reaction, it was found that the microcrystalline product formed at the hot end of the reac- tion tube unless water was present. Very large FeOCl crystals (2 x 0.5 cm) deposited at the cool end in some preparations along with the excess FeCl3 that was used in the reaction when snall anounts of water were present. FeOCl is water sensitive; the extent of its hydrolysis has been studied by monitoring the Cl-Fe ratio as a function of time and water pressure.72 It was found that in a dry atmosphere FeOCl seems to be indefinitely stable, but it decomposes completely in 3 days in 100% humidity. The decomposition product is not detectable by X-ray powder diffraction; it has been suggested to be y-FeOOH (lepidocrite),77 which has a similar structure. A small amount of impurity could 27 easily go unnoticed in FeOCl. In solid state reactions such as that of Fe203 and FeCl3 it is sometimes difficult to ensure a product free from impurities. Small amounts of impurities in FeOCl can have a significant effect on rates of intercalation reactions and the physi- cal properties of the intercalate. The elemental analysis for two distinct samples of FeOCl prepared by the same method is listed in Table 2. These apparently identical preparations (by powder X-ray diffraction) of FeOCl also show quite different variable temperature magnetic susceptibility curves (Figure 7). Low temperature magnetic susceptibility is sensitive to small amounts of impurities, and thus is a good technique for ascertaining the purity of FeOCl."1 Table 2. Chemical analysis for two samples of FeOCl whose variable temperature magnetic susceptibility is shown in Figure 7 Fe% Cl% calculated 52.05 33.04 FeOCla observed 52.33 32.47 Feoc1b observed 49.19 29.18 aRepresented in Figure 7 by by A. bRepresented in Figure 7 0. 28 .1 a 23 83 Fuomu wo mcowuocwawea or» cow Axv meanesmaEmu m> »u_ppnwuamum=m capo: .N mc=m_d 3...; wmnhdmwazw... on oc on ON 0— . . q . _ A _ _ _ T I imwm 1 d 4 d d 4 d d a G o Jvd _# WA s-Ol xX 29 Apparently FeOBr does not exist.23:78 Attempts using a-Fe203 with anhydrous FeBr379 or with FeBr3 and A520380 in a sealed tube at reaction temperatures of 500 and 400°C, respectively, proved unsuc- cessful. This is due to the decomposition of FeBr3 at elevated tem- peratures. FeOCl(4-AP) (4-AP = 4-aminopyridine) has been reported 1/4 to undergo deintercalation ahd chloride-methoxide exchange using MeOH.81 FeOCl is reported to be indefinitely stable to MeOH at 80°C,82 but when FeOCl(4-AP)l,k is soaked in MeOH at 80°C it forms FeOOCH3, which has the FeOCl type structure with the chlorides replaced by methoxides. It was suggested that the expansion of the interlayer distance on intercalation of FeOCl makes it easy for methoxide to replace the chlorides.82 The 4-aminopyridine intercalate is the only one reported to undergo this type of reaction with FeOCl, although there is recent evidence that MeOH cointercalates as well when used as a solvent.36 FeOCl also undergoes exchange with NH3 to form FeONHz, but little work has been done with this compound.83 This type of approach, deintercalation and exchange, may be more promising for the synthesis of FeOBr. Intercalation Intercalation reactions may be represented chemically by the addition reaction L + xG-———+ LGx. L represents the contents of one unit cell of the layered material and x represents the nunber of guest species (usually a fraction) associated with that unit cell. In gen- eral, for intercalation reactions, the number of guest molecules in a unit cell is expressed as 1/x where x is an integer. This convention will be followed throughout the text except when clear identification 30 of particular preparations of an intercalate phase is necessary in order to aid in the discussion of the synthesis. Solvent effects Solvent effects in the case of the intercalation of FeOCl have been noted.3":85 Investigation of the intercalation of Lewis bases indicates that the rate of intercalation depends upon the acid strength of the solvent used.85 In the case of metallocenes, solvent and steric effects are more important than differences in ionization potentials of the guest species.3“ It was found that materials which did not intercalate with toluene as the solvent were easily inter- calated with DME as the solvent. It was suggested that DME facili- tates electron transfer from the guest to the host by stabilization of the radical cation. DME cointercalates with some of the metallocenes, but there is no evidence (MS or IR) that it cointercalates with TTF. All intercalation reactions of tetrathiolenes and related com- pounds were run with various solvents such as toluene, MeCN and DME. Intercalation with TTF using toluene was first performed in our laboratory.86 Since then, various synthetic procedures have been examined using DME and MeCN as solvents, as well as no solvent. The intercalation reactions of TTN and TTT required longer reaction times (22-45 days) at higher temperature (65-70°C) than the other tetra- thiolenes. This is due to their extremely low solubility in most solvents. Solvents clearly play a role in the intercalation process: total intercalation of TTF is achieved in 10 days at 55°C using DME as a solvent whereas only 40% intercalation is observed using MeCN under 31 the same conditions.87 Even without solvent TTF will intercalate Slowly: about 20% intercalation was observed by reacting TTF and FeOCl in a sealed tube at 135°C for a 24 hour period. A study of the kinetics of the FeOCl-TTF intercalation reaction suggests that impuri- ties in the guest species aid in the rate of reaction (Table 3). The TTF used in this study was either a high purity Strem (99.9%) or Aldrich (97% pure) product. The final intercalation products were identical by X-ray powder diffraction and variable temperature mag- netic susceptibility. Further studies are necessary with FeOCl of known particle size in order to obtain a quantitative measure of the effect of impurities on the rate of intercalation. TMTTF is soluble in toluene, methylene chloride and hot MeCN. Due to the initial success of toluene as a solvent for TTF, it was used as a solvent for the intercalation reaction of FeOCl with TMTTF. Repeated attempts at a reaction temperature of 85°C always produced a two phase system in addition to 40-80% unreacted FeOCl. The use of lower temperatures produced similar results, but required longer reac- tion times. The powder X-ray diffraction showed two low angle peaks, corresponding to d = 13.59 and 11.94 A, in addition to that of the FeOCl (010) reflection, suggestive of the presence of two b_axis parameters for the intercalated phase. The observed d spacings are consistent with the TMTTF in two orientations: one lying perpendicular and the other parallel to the FeOCl layers. The same phenomenon occurs with FeOCl(p-phenylenediamine)1,9.85 Its b_axis parameters are 13.61 and 11.42 A, consistent with the orientation of the plane of the phenyl ring perpendicular and parallel to the host layers. In the 32 Table 3. Kinetic data for FeOCl + TTF (0.24 M in dimethoxyethane) time (days) h k 2 99.9% TTF 97% TTF (%)a (%)a 3 o 2 0b 78 60 o 1 0C 100 100 6 o 2 0b 100 100 o 1 0c 33 6 10 o 2 0b 100 100 0 0° 4 - 14 o 2 0b 100 100 0 0c 2 1 18 o 2 0b 100 100 o 1 oc - - aPercentages were calculated from the measured intensity of the 0 2 0 reflection of the intercalate and the O I 0 reflection of FeOCl and reported as 1/1 (strongest peak). bThe 0 2 0 reflection for the intercalate. cThe 0 1 O reflection for FeOCl. 33 case of p-phenylenediamine no evidence for unintercalated FeOCl was observed. Further efforts with MeCN provided a one phase system for FeOCl(TMTTF)1,x, in which the plane of the TMTTF molecule is parallel to the layers. Stabilization of the radical cation by DME has been cited as the reason for facile intercalation of metallocenes into FeOCl.3“ For TTF, DME appears to induce a faster rate of intercalation than toluene or MeCN. TMTTF is insoluble in DME. The fact that only MeCN results in complete intercalation of TMTTF into FeOCl suggests a similar expianation to that proposed for the solvent effects for the TTF and the metallocene reactions. MeCN, a more polar solvent than toluene, may stabilize the radical cation of TMTTF. Although this stabiliza- tion of the cation may in general be important for intercalation, the use of iodine to generate the radical cation of TTF jg_§jtg_had no effect on the rate of intercalation, which was monitored by X-ray powder diffraction. Effects of FeOCl decomposition and FeCleupon intercalation Elemental analysis of several TMTTF intercalation reactions has afforded some insight on the effects of FeOCl decomposition upon intercalation. Table 4 shows the chemical analysis for the products of the reactions discussed below. FeOCl was made in large quantity and several reactions were prepared with TMTTF in MeCN. One of the reactions showed immediate decomposition upon addition of the hot TMTTF MeCN solution. The decomposition could be observed by the change in color of a small amount of FeOCl from purple to orange-red. The chemical analysis of 34 Table 4. Chemical analysis for FeOCl(TMTTF)x PFEPGFEG under different reaction conditions (calc) Conditions x (obs) C H S Fe Cl (%) 21 days 0.068 6.52 0.66 6.98 44.59 28.33 T = 65°C 6.61 0.59 6.74 44.83 28.58 55 days 0.077 7.24 0.73 7.73 43.87 27.85 T = 65°C 7.32 0.79 7.89 44.13 27.44 21 days 0.074 7.01 0.71 7.50 44.05 28.03 decomposed FeOCl 6.69 0.56 7.82 42.61 28.04 T = 65°C 23 days 0.083 7.76 0.78 8.28 43.29 27.48 T.Z 70°C 8.66 0.98 7.93 43.80 24.04 the resulting intercalate is significantly low in iron content compared to the calculated value. The stoichiometry (ignoring this discrepancy) is FeOCl(TMTTF)O.o7k. This is a slightly higher anount of TMTTF than that in a sample which showed no visible signs of FeOCl decomposition (FeOCl(TMTTF)o.068). When reactions proceeded at temperatures greater than 70°C, decomposition occurred and the resulting chloride analysis was low. The decomposition of FeOCl in intercalation reactions has been noted with the intercalation reac- tions involving metallocenes at temperatures greater than 80°C."2 If 35 the discrepancy between the calculated and observed percentage of chloride is disregarded, the stoichiometry obtained is FeOCl(TMTTF)0.083. These differences are greater than can be accounted for by errors in the analysis (see Table 4). It appears as if the decomposition of the host increases the rate of reaction. This effect has been noted for the M52 intercalates, where excess sulfur in the solution (obtained from small amounts of decomposition of the host) increases the rate of reaction.88 Subtle changes in the stoi- chiometry are also observed (based on analytical data) using longer reaction times. A reaction time of 55 days increased the TMTTF content in FeOCl from 0.068 to 0.077 per FeOCl. This constitutes about a 1% increase in the intercalant analysis which is significant. The changes which involve only the TMTTF content in FeOCl appear to have little effect on the resulting physical properties. The reaction of FeOCl with pyridine at 70°C for 7 days forms the compound FeOCl(Py)l,3.89 Analytical and mass spectrometry studies indicated the formation of bipyridine and chloropyridine.32 It was suggested that redox mechanism similar to that proposed for TaSZ(py)l,2 might be valid for the FeOCl(py)l,3 reaction.90 It has also been suggested that small amounts of water (<100 ppm) aid the intercalation reaction by forming pyH"'.32’91 In the preparation of FeOCl(py)1/x a slightly different phase was obtained than that reported in the literature. The pyridine used in this reaction was reagent grade, dried 24 hours over activated 4 A molecular sieves, in contrast to the published procedure in which the pyridine was dis- tilled from sodium hydroxide. The stoichiometry obtained from this 36 preparation was FeOCl(py) The infrared spectrum obtained for 1/3.4’ the FeOCl(py)1,3.k sample is identical to that reported for FeOCl(py),”92 in which peaks corresponding to bipyridine were iden- tified.36 It was noted in different preparations that sometimes the reaction mixture turned yellow immediately, indicative of the presence of FeCl3 as an impurity in FeOCl. Even with very pure FeOCT samples . (as determined from low temperature magnetic susceptibility), it was noted that the pyridine solution turned faintly yellow by the end of 7 days at 70°C. Although this discoloration has not been reported, at least one other scientist working in this area has made the same observation.93 The optical spectra corresponded to that of FeCl3-6H20. The presence of 90811 amounts of FeCl3 may be important to this intercalation reaction. Preliminary results using other guests DTT could be successfully intercalated into FeOCl using DME as a solvent, but the few Bragg peaks obtained in the X-ray powder diffrac- tion pattern were very broad, and no further work was done with this compound. DBTTF appears to intercalate into FeOCl very slowly (50% after 6 months at 65°C) using toluene as a solvent. ClkTTN, TMTSF and DM/DETSF in toluene at 70°C or in methylene chloride at 30°C have not been intercalated to date. These three intercalants are fairly insoluble in most solvents and further studies are necessary in order to ascertain whether intercalation of these materials is possible. 37 Preliminary results using hosts other than FeOCl Alternative hosts whose intercalation chemistry was investigated were VOCl and 2H-TaSz. Although TaS2 is easily intercalated by pyridine,9“ intercalation with TTF, TMTTF, DBTTF, and TMTSF was unsuc- cessful with tolune as a solvent at 100°C for 9 days. Attempts with TaS2 and TTF under more strenuous reaction conditions (sealed tube at 180°C, 24 hours) were also unsuccessful; the X-ray diffraction pattern was that of pristine TaSz. The intercalation of VOCl with TTF or CA (chloranil) in toluene at 70°C, and in acetone, methylene chloride or MeCN at 28°C for two weeks was unsuccessful. The failure of these reactions may be attributed to the low temperature used; 28°C was initially used both because of the low boiling point of methylene chloride and to minimize decomposition of the intercalant. Attempts to intercalate pyridine into VOCl were also unsuccessful despite published procedures.95:96 The literature preparation uses a reac- tion temperature of 80°C whereas only 70°C was used to attempt the synthesis. Temperature and/or impurities can have significant effects upon intercalation, as has been shown in the proceeding paragraphs for FeOCl. A study of these effects upon the intercalation of VOCl is in progress in this laboratory and will provide complementary data to this work on FeOCl. STRUCTURAL STUDIES OF THE INTERCALATES X-Ray Diffraction Results The FeOCl structure consists of stacked neutral layers formed from distorted cj§f(FeCl20,)7' octahedra, which share half their edges to produce a central sheet of (Fe0)n+ with Cl‘ layers outermost on either side of the sheet.67 The iron atoms share a single oxygen atom along the 3 axis, and both a chloride and an oxygen atom along the 5 axis (Figures 3 and 8). FeOCl crystallizes in an orthorhombic space group, Pmnm, with_a = 3.780 A,.b = 7.917 A, and_g = 3.303 A. These cell dimensions agree very well with the Bragg reflections observed in the X-ray powder diffraction (Table 5). X-ray powder diffraction data of FeOCl(TTF)1/9(tol)1,21 could be simulated by a monoclinic cell with the_g axis corresponding to the interlayer distance.97 Although the positions of the calculated reflections were in close agreement to those of the observed reflec- tions (Table 6), extinctions in the experimental data could not be rationalized. A model originally proposed for FeOCl(metallocene) 39 1/x was considered for FeOCl(tetrathiolene) In this model, the b 1/X° axis expands and doubles upon intercalation, producing symmetric vacancies for the guest molecules. Alternate layers move one-half unit cell along the gg_plane, giving rise to extinctions consistent with a body centered unit cell of symmetry Immm or 1222. The X-ray 38 39 mppmu “we: Lao» mcpzosm ”comm we ovumewsum < .m weaned Tabie 5. Calculated and observed X-ray powder diffraction data for FeOCl (Cu Ka1 = 1.54056 A using silicon as an internal standard) Ia dgbs h k 1 dcalc 5 7.9158 0 1 0 7.9170 5 3.4093 1 1 0 3.4111 m 2.5364 0 1 2 2.5361 m 2.3735 1 1 1 2.3729 m 2.1787 1 3 0 2.1638 w 1.9785 0 4 0 1.9792 m 1.8895 2 0 0 1.8900 w 1.8350 2 0 1 1.8383 w 1.8082 1 1 3 1.8100 w 1.6503 0 0 2 1.6515 m 1.6167 0 1 2 1.6167 vw 1.5482 1 4 1 1.5487 m 1.5132 2 2 1 1.5155 mw 1.4855 1 1 2 1.4865 w 1.3916 2 3 1 1.3932 m 1.2420 2 0 2 1.2436 vvw 1.0602 0 2 3 1.0607 3 = 3.780 A, P. = 7.917 A, g = 3.303 A aRelative intensities. bd-spacings in A. 41 Table 6. X-ray powder diffraction data for FeOCl(TTF)1,9 (tol)1,21, Co Ka = 1.7902 A ' Ia dgbs h k ‘ dEalc 100 13.08 0 0 1 13.205 20 6.60 o 0 2 6.602 10 3.475 2 o 0 3.480 1 3.079 4 o 5 3.075 1 2.742 2 1 5 2.739 1 2.636 0 0 5 2.641 6 2.493 4 0 3 2.491 1 1.338 0 1 4 2.337 1 1.880 0 0 7 1.886 1 1.811 0 1 6 1.833 1 1.644 0 o 8 1.651 2 1.505 2 2 0 1.450 a = 3.84 x 2 A, b_= 3.31 A, £_= 14.43 A, a = 115° aRelative intensities. bd-spacings in A. 42 powder diffraction data and lattice cell parameters for the FeOCl(tetrathiolene)“x compounds are given in Table 7. In all cases, the data are consistent with the model proposed by Halbert39 described above. Early in this work, extinctions were not considered; it was thought that intercalation would decrease the number of observ- able X-ray peaks due to disorder of the lattice structure.36 The decrease in the number of Bragg peaks of the intercalate compared to FeOCl can be attributed to the space group; body-centering increases the number of extinctions by including the requirement of 2n = h+k+z. X-ray powder diffraction peaks for the intercalates are slightly broader than those for pristine FeOCl. TTT is the largest of the organosulfur molecules, and the TTT intercalate has the broadest X-ray peaks (Figure 9). FeOCl(TTT)1,9(tol)1,21 may contain unintercalated FeOCl: the broad peak observed in the X-ray powder diffraction at about 8 A is always present. The peak at 8 A can be attributed to either the (0 1 0) reflection of FeOCl (b_= 7.917 A) or to the (0 4 0) reflection of the intercalate (b_= 30.89 A). Very few reflections are observed for this compound and only the b axis could be obtained from the X-ray powder diffraction data. FeOCl(TMTTF)“13 shows the smallest expansion along the b axis; presumably the intercalant is oriented parallel to the layers. For some samples, containing a higher content of TMTTF, the X-ray powder diffraction data showed additional diffraction peaks to those reported in Table 7, which could not adequately be accounted for. A model was proposed97 which was based on the sulfur atoms of the intercalant and the chlorine atoms of FeOCl interacting as hard Spheres of fixed van der Naals radii. 43 Table 7. X-ray powder diffraction data for the intercalates showing the calculated and observed d-spacings for the cell parameters given Ia dgbs h k 2 dEalc FeOCl(TTF)1,9(tol)1,21 s 13.1694 0 2 0 13.078 s 6.5378 0 4 0 6.5390 s 3.4776 1 3 0 3.4773 w 3.0770 1 5 0 3.0702 m 2.7664 0 5 1 2.8077 vw 2.5021 1 O 1 2.5011 m 2.4838 0 7 1 2.4860 w 2.4649 1 2 1 2.4666 m 2.3365 1 4 1 2.3361 m 1.8960 2 o 0 1.8960 w 1.8757 2 2 0 1.8764 mw 1.8077 1 10 1 1.8077 w 1.6638 0 0 2 1.6638 vw 1.6511 0 2 2 1.6505 vw 1.5085 2 7 1 1.5074 vw 1.2507 3 3 0 1.2509 _g = 3.792 A,_b = 26.1560 A,.E = 3.3276 A Table 7 (cont'd) Ia dobs h k 2 dgalc FeOCl(TTF) s 13.099 0 2 0 13.009 m 6.511 0 4 o 6.505 s 3.474 1 3 0 3.471 m 3.065 1 5 0 3.062 m 2.8212 0 5 1 2.8116 s 2.4842 0 7 1 2.4849 w 2.4604 1 2 1 2.4603 m 2.3419 1 4 1 2.3381 w 2.1880 0 9 1 2.1862 m 1.8937 2 o 0 1.8937 mv 1.8748 2 2 0 1.8740 mw 1.8048 1 10 1 1.8048 mw 1.6706 0 0 2 1.6706 w 1.6568 0 2 2 1.6570 vw 1.5699 2 5 1 1.5705 m 1.5056 1 3 2 1.5053 .3 = 3.7874 A,.b = 26.0193 A,.g = 3.3412 A 45 Table 7 (cont'd) Ia dgbs h k 1 dEalc FeOCl(TMTTF)1,13 S 11.588 0 2 0 11.672 m 5.834 O 4 0 5.836 m 3.394 1 3 0 3.4034 5 2.7082 0 5 1 2.7043 m 2.3586 0 7 1 2.3519 vvw 2.0457 0 9 1 2.0434 m 1.8923 2 O 0 1.8923 mw 1.8679 2 2 0 1.8679 m 1.6587 0 0 2 1.6587 w 1.6421 0 2 2 1.6422 w 1.5514 2 5 1 1.5504 vw 1.5288 1 12 1 1.5341 mw 1.5082 1 3 2 1.4910 vw 1.4758 2 7 1 1.4742 vw 1.4711 2 10 0 1.4698 mw 1.2478 3 3 0 1.2453 mw 1.2406 2 2 2 1.2403 3 = 3.7846 A, 9. = 23.3444 A, 3 = 3.3174 A 46 Table 7 (cont'd) Ia d5bs h k 1 dgalc FeOCl(TTN)l/9(tol)1,22 S 15.479 0 2 0 15.443 m 7.7014 0 4 0 7.722 m 3.5559 1 3 0 3.5583 w 3.2338 1 5 0 3.2317 w 2.9300 0 5 1 2.9332 w 2.6595 0 7 1 2.5595 w 2.3903 0 9 1 2.3909 w 2.2501 1 6 1 2.2513 w 2.1449 0 11 1 2.1474 m 1.8958 2 0 0 1.8960 m 1.8858 2 2 0 1.8819 m 1.8085 1 12 1 1.7945 m 1.5019 1 3 2 1.5091 mw 1.4841 2 9 1 1.4856 .3 = 3.7920 A, b_= 30.8860 A, £_= 3.3329 A aRelative intensities. bd-spacings in A. 47 (a) 11 (b) LJV 11 .11 1 1 1111141111414114J#11 42 36 30 2‘ I '2 § 42 36 3O 24 '0 ‘2 5 (c) (d) M11111! I'll 1111111141114] 42 36 30 24 IO l2 6 42 36 30 24 IO 12 6 Figure 9. X-ray powder diffraction data plotted as intensity !§_29 for (a) FeOCl, (b) FeOCl(TTF)1,9(tol)1 2 , (c) FeOCl(TTN)1,9(tol)1,23 and (d) FeOCl(TTT11,7 48 Reevaluation of the X-ray and neutron powder diffraction data has led to another model in which the sulfur and the chloride atoms can be closer together than the sum of their van der Naals radii, due to a nonspherical electron distribution around the sulfur atom of TTF. This is referred to as the soft sphere model for lack of a better descriptive title. In this model, the FeOCl layers are "locked" into place by the TTF molecule. This type of locking also occurs with metallocenes intercalated into TaSz.98 Both models will be described in detail later in this chapter. Single crystal studies on FeOCl(TTF)l,x have been carried out. A longer reaction time (4-5 months) is required for the intercalation of FeOCl crystals than for powder. Microscopic examination showed that the surface of many of the FeOCl(TTF)“x crystals was puckered, and usually the "crystals" were aggregates, consisting of several sheets stacked one upon the other. In general, the smaller crystals appeared to give better X-ray diffraction patterns. “Single crystals" of three different reactions were mounted and precession photographs obtained. One preparation was vacuum dried, and the other two were dried by passing dry argon over the crystals. All the photographs for the larger crystals showed broad streaks or spots that were not well defined. The smaller crys- tals provided better diffraction, but the photographs were interpreted as resulting from a twinned or otherwise imperfect crystal. Intercalation structures have been determined by calculated elec- tron density maps based on X-ray powder diffraction data,99:1°° EXAFS,1°1‘1°“ neutron powder diffraction,3'*:3S and more recently by 49 solid state NMR.37’1°5:1°6 The latter two techniques were applied to the FeOCl-TTF system, the best characterized and most crystalline of the FeOCl-tetrathiolene intercalates. Further structural informa- tion on all the intercalates was obtained by EXAFS spectroscopy. Extended X-Ray Absorption Fine Structure Results In recent years, there have been several excellent theoretical derivations and reviews of EXAFS (Extended X-Ray Absorption Fine Structure), and the reader is referred to these works for a formal discussion of EXAFS theory and its application.1°7"111 The following section will discuss experimental results that were analyzed by accepted procedures described in greater detail elsewhere.‘55'112'115 Theoretical background for EXAFS will onIy be included where necessary to aid the reader in understanding the results described herein. The Fe K-edge transmission X-ray absorption spectra of FeOCl and the intercalates are shown in Figures 10-14(a). The absorption spec- trum exhibits modulations of the transmitted X-ray intensity above the absorption edge: these modulations are referred to as Extended X-ray Absorption Fine Structure (EXAFS). The Spectrum is typically plotted as the £n(Io/I) x§_E, where I = intensity and E = energy in units of eV. If the absorbing atom is near other atoms, the final state wave- function of the photoelectron may be modified by scattering due to neighboring atom(s). This results in an interference effect in the final state of the photoelectron and modifies the absorption cross section. 50 2.5— 1010 I 1.5- l a A 1 1 l 1 n L 1 l 1 I l 1 l n l L 4 l a L A 1_l 7000 7200 7400 7600 7800 eooo EieV) ’ (b) P 2.1;. A k’xik) c> I -2.5'— N . . . 1 . . . .4 o 5 10 '5 MA") Figure 10. (a) Fe K-edge transmission x-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of FeOCl 51 E (a) 2 — 1.75 - u—TIu-o ' c 1.5 '— — )- \J .1. F l j J l J J j A l l A l l j I A l l J I l l l A l 7000 15’ (b) 7200 7400 7600 7800 E (eV) 2}»- fl k’Xik) I I r T l I -2.5 [Irlvl C J 'l isikiA ) <3 u 25 Figure 11. (a) Fe K-edge transmission X-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of FeOCl(TTF)l/8.s 52 I (0) 2h— C 1.75- c . _ . \) r r 1.25— l- b__l._141‘In]!1441141JgI_1_.|—.|——A—J—l.—l—l—l—lu—| 7 7000 7200 7400 7600 7800 E(eV) ' (b) 5 n 2.5 ‘ 3 X 13‘ 0 Figure 12. (3) Fe K-edge transmission X-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of 53 7’ (a) N 'VITUTV 1T \1 1.25 fIvTVI‘IIV J A L L J l A l A j J A J A 1 l 1 AJ—L—J—l—d—l—l—l 7000 7200 7400 7600 7800 8000 E (eV) 7-5'.‘ (b) r i 0 IV 2.5 VIII 3 k3 X (k) '1' - 2.5 1 ‘1T ‘1 P J a A a J 1 1 j J l j 1 A L _J 0 5 10 I5 k (‘4) Figure 13. (a) Fe K-edge transmission x-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of FeOCl(TTN)1,9(tol)1,23 V V 1' 1.5 litr'I' I I U U .31... 5 0.5— -0.5 r 6 k’Xik) li'ftl'! '1' V I I Figure 14. 7000 54 (o) J (b) 7200 7400 7600 7800 8000 E (9v) I 15k 1A") (a) Fe K-edge transmission X-ray absorption spectrum and (b) the background subtracted Fe EXAFS data of FeOCl(TTT)1/7 55 The background absorption for all compounds was removed in k-space using a cubic spline function of four sections, Ak = 3.48 A'1 (equation 1). k(A-1) = Havens-EEO“) (1) The normalized oscillatory part of the absorption coefficient in the form k3x(k)_v§ k, was obtained by dividing by the edge jump Auo. and by correcting for the fall-off of u(k) with Victoreen's true absorption coefficient equation.115:117 The edge jump, used to normalize the data, was taken as the step in the absorption cross section across the K absorption edge. The edge jump was measured as the difference between the EXAFS and pre-near-edge region absorption. For all these spectra, the edge position energy, E0, was determined as the photon energy at half-height of the absorption edge jump plus 13 eV (7124 eV). This provided a Fourier Transform in which the magnitude of the first peak was the largest. Since E0 values cannot be determined exactly from the features in the absorption spectrum, 50 was treated as an adjustable parameter in the final curve fitting of the data. Aon is defined as the threshold energy difference and was refined for each type of scattering atom j (equation 2). = th _ exp Aon on E0 (2) This allows for discrepancies between empirically chosen energy thresholds, (ESXP), and theoretical Eo's (EBh).107 In the analysis, it was found that the amplitude of the oxygen peak is sensitive to the number of sections of the spline in addition to the choice of E0. 56 The k3x(k) of the Fe K-edge for FeOCl and the intercalates are presented in Figures 10-14(b). A Fourier transformation relates the EXAFS function k3x(k) of the photoelectron wavevector k(A‘1) to its corresponding function 93(r') of distance r'(A) (equation 3). 43(r') = (211)-1/2 [kmax k3x(k)e12k"dk (3) m1n The Fourier transform provides a modified radial distribution map of the environment around the X-ray absorbing atom. The Fourier transform peaks are shifted from the true distances due to the effect of a phase shift, which amounts to approximately 0.2 to 0.5 A, depending principally upon the absorbing and backscattering wave functions. The Fourier transforms of FeOCl and the intercalates are shown in Figures 15 and 16. The first five peaks of the Fourier transforms are assigned as follows: Fe-O; Fe-Cl; Fe--Fe (nearest neighbors); Fe--Fe (£_axis); and Fe-Fe (a_axis). From the comparison (tol) data it of the Fourier transforms of FeOCl and FeOCl(TTN) 1,21 1/9 is apparent that the FeOCl lattice is largely unperturbed upon intercalation (Figure 17). A Fourier filtering technique was employed to isolate peaks due to the absorbing atom's near neighbors. The resulting Fourier filtered EXAFS were subsequently truncated at 3.0 A"1 and 13.0 A"1 to eliminate distortion.”9 The Fourier filtered curves were fit using the standard formulation shown in equation 4. Figure 15. FOURIER TRANSFORM 57 FeOCl 150 - 100 - n (b) FQOCI (TTF-h".g 125 100 75 5O 25 D 1 R IN ANGSTROMS Fourier transforms of the Fe K-edge transmission EXAFS k3x(k) y§_r for (a) FeOCl, (b) FeOCl(TTF) and 1/8-5 (c) FeOCl(TMTTF)1,13 58 {(0) FEOCI(TTT)V7 1 1 V 100'- 75* 50- 2(0) FeOCl my)”, 125; n U 100:- FOURIER TRANSFORM c: n: 75:— 25— 1 1 14 1 JJ 1 1 11 1 0 2 4 6 3 10 R IN ANGSTROMS Figure 16. Fourier transforms of the Fe K-edge transmission EXAFS k3x(k) 22 r for (a) FeOCl(TTT)1,7 and (b) FeOCl(py)1,3,6 59 vrrr'Irrvltrrtlrtiirtvvv 150 - “'0 an“ EXAFS ‘ I- n '2 F0 K-Odw d _ 5g ‘-'FOCKN 1 -'= ------- FeOCl (TTN),,, . Eng/Fo-CI .1 ,/§ 100— A J _ é 1//:;—-F0--Fea ‘ 50— M l 4 .0 I. p- n’ .- o' I: . o: "o .- b 'u s 4 .. 0 ~ 0 Q o O o I- 1 : '1 :'°.. ' '. °. : °. -. '. - : - , l o ' I : ". D :0. '. .I '. 0. o .. . 0 . .0 .0 .0 g .0: o ‘. .: 0' o. I 0 . a. " o. ‘ O-‘r41141111312111 o 2 4 6 “8 1 1o r(x) ¢31FI Figure 17. Comparison of the Fourier transforms of the Fe K-edge transmission EXAFS k3x(k).!§ r for FeOCl (solid curve) and FeOCl(TTN)1,9(tol)l,22 (dotted curve) with peak assignments 6O k3x(k) = g Bij(kj)k3exp(-2ogkj) - sin(2kjrj+ 63(k3))/r§ (4) A nonlinear least squares program for iterative estimations of parameters was used to refine the scale factors (independent of the photoelectron wavevector, kj) for the scattering atoms of the jth type Bj, at a distance rj from the absorber, and the root-mean square relative displacement aj (Debye-Naller factors) along rj. Theoretical backscattering amplitude Fj(kj) and phase aj(kj) functions, calculated by Tea and Lee,11“ were used in the curve fitting formulation. The scale factor, Bj, is related to the number of nearest neighbors, N3, of the jth type of atom and to the amplitude reduction factor, 83, as follows in equation 5. The amplitude reduction factors Sj combine into one parameter many effects including the energy resolution of the monochromator117 and thickness effects)”.120 The first two peaks, due to Fe-O and Fe-Cl distances, were filtered using a window of r' = 0.9 - 2.9 A and fit using Fe-O and Fe-Cl phase and amplitude functions. The next two partially resolved peaks Fe--Fe (near neighbors) and Fe--Fe (g_axis interactions) were filtered, r' = 2.0 - 3.6 A and fit with two terms using Fe phase and amplitude functions while holding the Debye-Naller factors equal. The fifth peak, assigned to Fe--Fe (3 axis) was filtered, r' = 3.2 - 3.8 A, 61 and fit using one term. Multiple scattering was not a problem for Fe--Fe £_interactions due to the small Fe-O-Fe and the Fe-Cl-Fe angles (103.71° and 88.35°, respectively). The Fe-O-Fe angle along the_a axis is 149.29‘, and the effects of multiple scattering are negligible.121:122 The best parameters based on theoretical functions (BFBT) obtained from the curve fitting were adjusted using the fine adjust- ment based on models (FABM) procedure outlined in the litera- . ture.112’113 Plots of the values obtained for AEo.!§ br and for B !§_O are shown in Figures 18-20 for FeOCl and the intercalates. In these plots or = r - rbf (bf = best fit) and 5E0, B and a were defined previously. Pristine FeOCl was used as the model compound for the intercalates. The EXAFS distances and coordination numbers were obtained by fine tuning the parameters obtained from the best fit to the data using theoretical functions as shown in Table 8. Results for different preparations of FeOCl and FeOCl(TTN)1,9(tol)1,21 are shown in order to illustrate the reproducibility of the results. Notable trends in the data are (1) the Fe-Cl distances for the intercalate are consistently longer than those for the pristine material; (2) Fe--Fe nonbonded distances are consistently shorter; and (3) in all cases the .5 axis is elongated compared to that of FeOCl. The c axis has expanded by about 1%, a significant amount for a well-defined extended lattice structure. Debye-waller factors are given in Table 9 and are consistent with little or no disorder of the FeOCl intercalate lat- tice, in contrast to EXAFS results on MnPS3 intercalates.103 62 SERGE“. 6.. eee 3530325881. .33_o:.£mt:uooe .Fuowu com o m> m any can mcongmwmc cum: umucoacoc mdiimm cow ca m> mc va .mH wc=m_m 0.. 00. 00. co. N0. v0. No. 0 N0... «0... 00.- 00... - u d d 4 1 J l I 3 d d 3 q 1 9 d d — d d d I... o NN\ O\ J . so: .22.. Go on. a . — ”NA O\_ IJ o O—l . =0: ELL. .000“. x N .N\.A O\. 4 O“. .2 o m. 0 th .00 .... D 1 0.0.. ¢. _UOou O 4 0N1 n. 1 0.0- m. 40.0 i b. J 0*. o. 1on- m. . 90M- 0.. 2: e." .. 63 esuAeeevpuooe nee NutufipoovasdAzeev_uooa 9. mo. 00. eo. NO. 0 d I 3 d u d d I 1 I Any 0 1 l 4 0.0 N.— 1 10; ON 1 o~N\AAPOHV ax” Adeevpoooe .puomm com o m> m Any new Ampxm-uv mm-0m so; ca m> mc Amy .mH mcamwu v0. «0 0 N0? .0.- u u 1 q d — 1| 1 u d «ax ox .00- .:o: .25: so: a v0 nu.» as» 1 so: 3.5.0.009“. u . fix. a) JON: so: 0:15.000“. D a 1 .00.“. o 10.0.. 1 10.n- a 10..- 40.. 1 A3 101» 64 55:88 eee AmnfotaEzanmu .zzrotgfibsoom .Puomu go; o m> m Any use Am_xe-uv mm-mm so» L< m> ma Amy .om me=m_m 0.. 00. 00. co. N0. 00. V0. 0 V0.- 00.- 4 u 1 q 1 — a q q q o — q 1 q q _ d a d J 4 NN O\— # 5=2. 25:86.. a . ._ ¢. nuA a) 1 0.0.. :03 8.55.000... x 1. .N\. 0\. 1 . . so: Auk-5.006... a 1 m . 1 06- nxUmu o J 1 . N. 1o .10; 1Av¢ .4AVN 1520 2: 1 A 0v L 1.xu 65 .sessflpoo~asufiseev_soous . esssseev_uoouo .mwmapese mumu useuseum use Amse.fi Aesm Afivm AHVN Afive “evoe.m Advem.m saveo.m sevem.~ Amvmo.~ seeze «five Amve.s Amvm.m Asvm.fi “ave.” Aevoe.m sevem.m Afivso.m Amvam.e seveo.~ pee AHVN Amvm.s seve.m Rave Aese Amvss.m Aevmm.m Adveo.m Aevoe.e Anymo.~ zee Amvo.e “five Afive Ame.H Aevfl.m Amvoe.m sevem.m Afiveo.m Aevfie.~ seveo.~ ezes AHVN AHVN Aevm.q AHVN Aese seves.m Aevsm.m Aevmo.m Amvoe.~ seveo.m euee Asve Aevm Aeve save save “eves.m Aevmm.m seveo.m Amvee.e Aevmo.~ ouee save Aevo.~ AeANve AHVN save Afises.m Aevom.m Amvsfi.m Amvem.~ Aevmo.~ .uoou Aeso.e Afive save Asso.~ A003 “eves.m Amvoe.e AN0HA.M Amvsm.e Aevmo.~ e_esou .- oo.~ .- oo.~ 60.3 oo.N 60.4 s oes.m mos--61 e Nom.m mos-ou s soH.m eos-ou e eee.~ e_e-ou e mo.~ eo-ou eeeeossoe mmuepeusm»s_ use use Poems som m_mxpese ms mamwm we gamew .um mesa“; gamin— Od ON 0.— D I Ly .- D D I (F P b (P D W H e . E H o i... a 98m... , -591..- 1a 1 c O 'J 0 0 "1 1" .2! 1"; '5 D Q I I ON”? (933W 82 are discussed in the literature.131 Twenty experimental reflections whose peak position had been determined from the peak fitting program were used. There were three solutions found by the indexing program (Table 11) for an orthorhombic cell. Table 11. Cell constants and figures of merit obtained for twenty reflections from neutron powder diffraction data for FeOCl(TTF)1,3.5 (a(A) D(A) C(A) Figure of merit Unit cell volume 3.784 25.961 3.341 1614.9 328.17 10.016 12.986 3.854 14.2 501.31 6.688 25.959 4.585 11.5 796.02 A large figure of merit indicates a good fit: figures of merit less than 4 were disregarded. Although the parameters a = 3.784 A,-b = 25.961 and £_= 3.341 A adequately describe the unit cell (Table 12), it may not be the primitive cell. These parameters are based on the modification of the original host lattice dimensions, and thus provide a reasonable place to start fitting the data. The model used in the refinement for the FeOCl layers was dis- cussed previously: the alternate layers of FeOCl translate along the 83 Table 12. Observed and calculated d-spacings (A) of the neutron powder diffraction spectra for FeOCl(TTF)1,8.5 with the proposed indexing scheme: 2_= 3.7836(4) A, b_= 25.9629(3) A and g = 3.3410(4) A dobs(A) h k ‘ dcalc(A) 3.4676 1 3 0 3.4667 3.3128 0 1 1 3.3137 3.2452 0 8 0 3.2452 2.8099 0 5 1 2.8096 2.4827 0 7 1 2.4823 2.3366 1 4 1 2.3365 2.1834 0 9 1 2.1834 2.0029 1 11 0 2.0025 1.9828 1 8 1 1.9826 1.9280 0 11 1 1.9277 1.8918 2 0 0 1.8918 1.8720 2 2 0 1.8720 1.8551 0 14 0 1.8544 1.8157 2 4 0 1.8162 1.8026 1 10 1 1.8024 1.6705 0 0 2 1.6705 1.6566 0 2 2 1.6568 1.6377 1 12 1 1.6372 1.6220 0 16 0 1.6226 1.5701 2 5 1 1.5692 1.5049 1 3 2 1.5049 1.4301 2 9 1 1.4298 1.4164 1 17 0 1.4161 1.4055 0 10 2 1.4048 1.3893 0 12 1 1.3889 1.3509 1 9 2 1.3504 1.3241 0 18 1 1.3242 84 gE_diagonal ((101) direction) by one-half unit cell to accommodate the intercalant.39 The atomic positions were calculated for the hypo- thetical unit cell, and the atomic position parameters corresponding to the z coordinate were modified from the crystal structure of FeOCl67 by multiplying the z parameter of FeOCl by the b_axis of FeOCl, dividing by.b of the intercalate, FeOCl(TTF)1,8.5, and rede- fining the origin. A fit could be obtained with only Fe, 0, and Cl atoms, resulting in a weighted profile of R = 8.9% (Figure 28). Similar fits could be obtained from using either space group Immm or 1222 by refining 14 parameters (overall scale factor, cell constants (9), background parameters (2), and half-width paraneters (2)). Obvious errors in intensities could not be accounted for by either space group. A Fourier synthesis was performed on the observed and calculated structure factor amplitudes. It was found in the Fourier difference map that there was significant nuclear density within the layers, which corresponds to the TTF molecule. Upon inclusion of carbons representative of TTF (C has a scat- tering length of 0.665 barns (10‘1“ m) vs 0.285 barns for S), the R factor was reduced by one-third (R = 5.9%; structure factor reduced from 26.12% to 15.11%). Figure 29 shows the low d region for the observed and the calculated data as well as the difference curve for the Rietveld fit and the neutron data. Large isotropic thermal parameters for the carbon atoms were used to take into account the entire TTF molecule. Negative isotropic thermal parameters were obtained for the oxygen atoms. This could be due to an inadequate data set length or perhaps an error in the position of the oxygen atom. 85 eouuon on» an mcmmaam m>c=o oucocme$wu < .m:o_yompwmg mo meowupmoa mg» mumuwu=_ :gmuuua mg» :uwwcma x_uomg_u mxgm: .o:_— cwpom a an mupammc umum_=upmo on» on» new .m»:_oa An umuoommmv men came um>eomno mg» .Acowmog u :opv acmeocpwmg =2 .0 .o .01 spec m=_m= ”Acepv_uomc to. xcan .oa we“ co “cmeacpcme a__coea 32.30% 523) . - .ooma . . p . 3s... - - . .noma ohm... 3w.— coma t r: ____ __ _ __ ___=__ ______.___________ 11 41/ n d J . ..... . b. . e. u . ‘ + w. . w . . a V L. T 0'0 snwnoo 0°10 .mm menace ;n. 86 souuoa age an memmaao m>g=o mocmcmmmpu < .mcowuompwmc yo meowuwmoa mg» mumopucw cgmuumn on» gummcma apuomgwu mxcmz .mc_p uPFom a ma mapamme umumpaupmu m:u.ucm.mw~:voa An uwamopnc_ men mean um>cmmno mg» .Acomec e zo_v meaneeu neweapucp Aeeev_ooau to. xcea .om 0;“ co “caea=_.ae 0,3.oca 3 629..qu ”and L) b P P D I ’ b P b D r b bl h by D 1' h ’ 2 D D __ ____ __ _ __ :=.. ____=_____.::_., .mm w.=m_a 87 The data set length was enlarged to about 1 A. The occupational parameters for the carbon atoms were allowed to vary. The data could be refined to a weighted R = 7.2%, which was good considering the incomplete description of the TTF molecule (Figure 30). Atomic posi- tions used in the refinement are listed in Table 13. The data show some discrepancies in fitting the background, which can be attributed either to short range ordering or to incoherent scattering of the hydrogens. The low temperature data were not analyzed. The FeOCl(du-TTF) sample was run to eliminate the possibility 1/9 of incoherent scattering, which may be causing the slow oscillation in the hydrogen data background. In the Rietveld analysis, the goodness of fit remained about the same as for FeOCl(TTF)1,8.5. The lattice paraneters obtained were _a_ = 3.7857 (3) A, b = 26.016 (3) A and g = 3.43375 (4) A. The background deviations are similar to those found in the hydrogen data and indicate that there may be short range order- ing contributing to the diffraction. No further analysis was done on this compound due to the presence of 5-10% pristine FeOCl, which affected the peak positions and intensities. IPNS has a program available for multiphase systems, which could be used once an adequate description of the intercalate is obtained. From these data, it appears as if TTF is aligned along the c axis, but the complexity of including the full TTF molecule in the refinement has hindered a more detailed analysis. Once a complete refinement that includes the TTF molecule can be undertaken, the exact structure will be known. At the present stage, the structure of the FeOCl lattice of the intercalate could belong to either a Immm or an Figure 30. 88 31 , :i ' ._ llllllllllll I ll lllllllllllllllll Illlll II | II I Iv-v w v vfi ''''''''''''''''''''''' L171 m. 1.820 a” LOCI u-sPAcme (A) '9 3 ' 1 4 .1 d4 1 :1 J 1 4 3.. I ' ‘ 14 O . ) It I I II n: l l I II ‘A A A ._ * ' ww— ‘w V ~— ,— fififififififififififififififififififififififififi L.‘ 3.031 2.047 2.3% 2.511 Profile refinement of 90' bank for FeOCl(TTF) . including data to 1.0 A. The observed data are indicated by points, and the calculated results by a solid line. Marks directly beneath the pattern indicate the positions of reflections. A difference curve appears at the bottom 89 1222 space group. The neutron data provide direct evidence for long range ordering of TTF within FeOCl. Table 13. Atomic positions refined for FeOCl(TTF)1,8.5 including carbons, space group = Immm (data set length 3.80 - 1.0 A, R = 7.2%) Atom x y 2 Fe 0 0 0.2827 0 -0.5 0 0.2655 Cl 0.5 0 0.1509 C 0 0.5 0.0678 90 Soft Sphere Model The results of the neutron powder diffraction data indicate that the conclusions of the hard sphere model are incorrect. Careful evaluation shows that doubling the b_axis reduces the discrepancies between the observed and calculated d spacings of the mixed reflec- tions as was shown in Table 7. The model proposed for the metallocene intercalates, in which the FeOCl layers expand and move laterally along the 3£_plane, also appears to be valid for the organosulfur intercalates. Tilting the TTF molecule is no longer a valid assump- tion, since the layers are oriented as shown in Figure 31. Figure 32 shows the ab plane of FeOCl: with the chloride layers eclipsed, tilting of the TTF molecule cannot explain the observed interlayer distance and the relative orientation of the layers. Both the X-ray powder data and the EXAFS data indicate that the intercalates (other than FeOCl(TTT)1/9(tol)l,23) are well defined solids with layers that are apparently undistorted. This supports the hypothesis that the layers are "locked" by the intercalant. The TTF molecule is perpendicular to the layers as shown in Figure 32, locking the layers in place. This places the sulfur atoms of TTF 3.15 A away from a chloride ion, much further than a typical sulfur-chlorine bond (1.99 A for SCl2 and 2.06 A for $2Cl2)132:133 but well within the sum of the van der Haals radii (3.6 A). It has been determined from electron density mapsl3“ and from theoretical calculations135 that the selenium atoms in TMTSF have sp3 hybridized orbitals arranging the electronic density in a tetrahedral orientation with respect to the selenium. Since sulfur and selenium are similar, 91 cowucpmuemucp cog: Poem; mo mLoAnp cow muweopzu on» $0 uwwsm Fogmgmp on» we copumeumappm .Hm 8.5m: 20.._.<._<0mm._.z_ mwhu< ZOE.<..<0mw._.z. mmOuwm aOa---- 7\/‘ 7“ r‘“‘ ‘! 9!, P .-. (a) (b) \I 9 CH Fe e(. c .‘5 .;:4 C) Cl Figure 32. Schematic representation of the possible orientation of TTF taking into account the lateral shift of the layers for both a (a) soft and (b) hard sphere model 93 the electronic distribution for the sulfur molecule in TTF might be similar to that of TMTSF. Thus, the electronic density of the sulfur atom of TTF is directed between the two chlorides that are along the 2 axis and further minimizes any type of repulsive interaction between the sulfur of TTF and the chloride ions of the FeOCl layers. This is the first time a model other than the hard sphere model has been con- sidered for any intercalation compound. For TMTTF, the van der Naals radius of the molecule is estimated to be 2.0 A. The molecule must be parallel to the layers to account for the observed 3.7 A interlayer distance. The electronic density on sulfur, if considered to be located in sp3 hybridized orbitals, again gives minimal interactions between the chloride ions and the sulfur atom in the TMTTF molecule. The orientations of TTN and TTT are more difficult to predict. As in the hard sphere model, TTN and TTT still must tilt at an angle a (defined in the hard Sphere model) to accommodate the short S-S bond (2.10 A). The molecule must be aligned along the g_axis and be tilted at an angle 0 = 30' as shown in Figure 33. TTT, with its bulky naphthacene core, introduces disorder within the layers which may give rise to fewer and broader Bragg peaks. The maximum stoichiometry is predicted to be FeOCl(TTF)“8 by the hard sphere model; and for the soft sphere model, it is also Fe0Cl(TTF)1,8. The stoichiometry is obtained by simple geometric considerations (Figure 34); the soft sphere model predicts that the TTF molecule occupies a minimum of 4 unit cells along the c axis and Tu TS Tr. e dy. "a a] N] TIC To 9 fr... 0 e "h 0t .sl tf 30 t nt 8...! .1.1 r." as E] ha tr 3 ft 0.6 1 n 09 01h tt .4 tt nn EU so ec mic pa 6 r0 t n .1 g n .1 k a t Figure 33. Schematic 95 Figure 34. Illustration of the TTF molecule on the bc plane showing the number of cells the molecule spans 96 one unit cell dimension along the 3 axis. There are 2 FeOCl units per unit cell. The maximum experimental stoichiometry obtained to date is FeOCl(TTF)1,8.5. In the soft sphere model the next TTF molecule is 3.78 A away (one unit cell along a), in the hard sphere model it was assumed the distance between TTF molecules was the sum of the van der Naals radii (3.6 A). Both models indicate that the TTF molecule is approximately close packed within the layers, but only the soft sphere model takes into account the lateral movement of the FeOCl layers. SOLID STATE NMR STUDIES Solid state NMR has been used to infer the relative orientations of guest species within a host lattice.37:100:959105:135:137 Wideline NMR studies on Ta52(cobaltocene)1,u provided additional evidence in support of the proposed model based on steric arguments, in which the C5 Symmetry axis of the cobaltocene molecule lies parallel to the TaS2 layers.37 NMR spectroscopy provides complementary information to the neutron diffraction studies on FeOCl(TTF)1/8.5 and FeOCl(dH-TTF)1/9, thus aiding our understanding of the long and short range ordering of TTF within FeOCl. In addition, NMR can provide information on the molecular dynamics of the intercalant, as well as information on inter- and intramolecular interactions between nuclei in the TTF mole- cules. Single pulse Fourier transform 1H NMR spectra were obtained for TTF, FeOCl(TTF)1,8.5 and FeOCl(dH-TTF)1/9. 130-1H cross polari- zation, magic angle spinning (CPMAS) spectra were obtained for the samples already mentioned, as well as for the TTF(CA) salts (both the segregated and mixed stack structures) and for TTFaBrz-ZHZO. The major obstacle in obtaining undistorted wideline spectra of solids is the slow recovery of the receiver, which produces dead-time. This limitation does not allow one to collect the initial part of the free induction decay (FID), which for very broad line spectra causes 97 98 substantial distortion of the baseline.138 Also, more rapid digitiza- tion than is commonly used in solution FT NMR is needed. Use of an echo technique overcomes the dead-time problem. Both single pulse and echo experiments will be discussed and compared. All contributions to an NMR line can be classified as homogeneous or inhomogeneous. High resolution spectra of solutions are homo- geneous in general; those of solids may be either. An inhomogeneous line is one in which the separate contributions to the line can be identified. These contributions result from anisotropies in quadru- polar interactions or chemical shifts: crystallites with different orientations with respect to the static magnetic field contribute to different parts of the line. One method to determine whether the line is broadened homoge- neously or inhomogeneously is to irradiate and saturate a specific spot in the line with a large, coherent H1 field. Immediately after the irradiation, the line is detected either by constant wave or, in this case, pulse FT NMR; an inhomogeneous line will have a "hole" burned into it.138 This method was used to determine the homogeneity of the spectral line for the FeOCl(TTF) sample. 1/8.5 1H NMR The 1H NMR spectrum of solid TTF obtained by Fourier transforming the free induction decay (FID) following a single radio frequency pulse is shown in Figure 35(a,b) (Table 14 defines the abbreviations used in the parameter listing). It consists of a relatively sharp component at about 4 ppm (relative to TMS in deuterated chloroform, 99 (a) TTFP.003 W/ LS=O T 1 i I i 1 j T Ifi T T I T l 7' I ‘ I 1 1 Ifi 1 Y I 1 1000 500 o —500 - lOOO PPM ONE-PULSE SEQUENCE TTFP.OO3 W/ LS—O P2: 3.50 USEC 05- 120.00 SEC (b) NA -= 8 SIZE = 16384 AT = 2.05 MSEC OPD ON I 1 ABC ON BESSEL FILTER ON 08 ATT.= 3 ADC = 16 BITS AI = 0 sw = +/— sooooo. DW c 263 RC 8 TO USEC DE= 263 USEC TL HIGH POWER ON F2= 361 064992 88 MODULATION OFF OF= 1715.58 SF= 361.064993 EM= 50.00 PA= 107.5 PB= -6524.l SCALE soooo.oo HZ/CM = 138 4792 PPM/CM YTjTTWVYTIi‘TTIIYTT1jYITITij1TYY"1111 )0 150 100 50 0 -50 —lOO —150 -200 PPM Figure 35. (a) FT 1H NMR of TTF obtained from a single pulse radio frequency, (b) Expanded scale (Table 14 defines the abbreviations used in the parameter listing) Table 14. Definitions listing for 100 of the abbreviations used in the parameter the NMR spectra P2 05 NA Size AT QDP 0N ABC Bessel filter 08 ATT. ADC AI sw ow RG TL F2 88 0F SF EM PA PB tipping pulse time delay between scans number of scans nunber of points in spectrum time for one Scan pulse phase cycling and data routing noise filter computer word Size scaling factor sweep width (Hz) time between points receiver gate (not used) transmitter level decoupler frequency decoupler mode frequency assigned to center of Spectrum radio frequency (rf) of tipping pulse apodization function value phase correction phase correction 101 used as an external standard) on top of a broader doublet. The Spectrum of only the sharp component can be obtained by removing the first 100 us of the F10 and Fourier transforming the remainder (Figure 36(a)). The Sharp component may then be removed by subtracting its spectrum from the initial single pulse spectrum (Figure 35); a doublet is obtained, which has a splitting of 15 kHz (Figure 36(b)). The line widths and relative intensities of the two components are shown in Table 15. The doublet obtained iS referred to as a Pake doublet, which arises when the dipolar interaction between a pair of spin 1/2 nuclei (1H) is the dominant magnetic interaction.139 This is most likely the case between the adjacent protons of TTF. This doublet is centered at about 12 ppm (relative to TMS in deuterated chloroform), in contrast to the 1H spectrum of TTF in solution, which consists of a Singlet at 6.2 ppm. The reason for this chemical Shift difference between the solid and solution spectra may be the presence of a large ring current effect due to TTF stacking in the solid. Ring current shifts of this magnitude and direction have been seen in large aro- matic rings in solution.”0 Removing the first 100 us of the F10 results in isolating the sharp component, indicating that the decay of the free induction Signal associated with this component is much Slower than that associ- ated with the broad component. The sharp spectral component may be due to a relatively mobile population of TTF molecules, possibly located at crystal defects, or to absorbed cyclohexane (obtained from the recrystallization), or to water. Assignments based on the solu- tion spectra of cyclohexane or water cannot be applied to the solid, 102 PHASE CYCLED QUAD ECHO FOR EXPLORER P2- 3.50 usec 03- 30.00 ussc o4— 1o.oo usec 05- 120.00 sec - . NA - 16 TTF SIZE = 16384 AT - 2.05 MSEC QPD ON - 4 (0) ABC OFF BESSEL FILTER ON 00 ATT.- 3 ADC - 15 BITS AI - 0 SW - +/- sooooo. ow - 263 no a 10 ussc DE- 263 USEC TL HIGH POWER ON F2- 361.064992 BB MODULATION OFF OF- .1715.58 SF- 361.064993 EM- 50.00 PA- 75.1 PB- -.0 SCALE soooo.oo HZ/CM = 133.4792 PPM/CM fiWUITTTTTT‘jI‘IIV‘TITIITITITIITIfTTTYW‘ 10 150 100 50 O —50 -100 —lSO -200 PPM (b) rT—111‘Tiv1lfitvilfifi11l1111l3111111111—1‘1—1—T— 00 150 100 50 0 -50 -100 —150 —200 PPM Figure 36. The 1H FT MR Spectrun obtained for TTF (a) after removing the first 100 us of the F10, (b) by subtraction of (a) from the full FT NMR Spectrum of TTF 103 Table 15. Linewidths and relative areas of the single pulse 1H Spectrum of TTF Spectral Full linewidth at % of total component half height (Hz) spectral area Sharp 1285 7 broad 35 x 103 93 due to experimental uncertainty and the lack of consideration of bulk susceptibility effects. In an attempt to obtain less distorted Spectra, echo experiments were performed. The most common pulse sequence used to produce echoes in solids is 00-1-090, where 0 is the pulse angle (the subscripts indicate the relative phases of the pulse and r is the time between pulses).133:1“1 Since the Nicolet 360 MHz Spectrometer at the University of Virginia is not capable of providing 90"pulses that uniformly excite the entire spectral frequency range of interest, pulse times (and therefore angles) were decreased until distortion was minimized. Figure 37 shows the spectrum of solid TTF obtained with a 00-1-090 sequence. When compared with the single pulse spectrum, the broad component obtained by the echo experiment is narrower and has reduced intensity relative to the Sharp component. The linewidths and relative intensities of the two components are shown in Table 16. The 104 8533 3:3 25.75 a 33295 .3 8538 at.» 28 .5383 ”.22 E :3 23¢ ooo.1 oom1 o oom coo, p.-—.ppprPplP|_»1Wbp—pb~FP».5 :o\:aa ~mn¢.onp 1 zo\Nz oo.oooom u3u umo mm1 broad 61 97 oo-r-oo low field sharp 4 4 high field sharp >1 broad 60 96 1.2 2 2 2 6 (AH) = (3/5)gNBN(1/n)I(I+1) Z1/r3,k + ’ (18) (4/15)g'§8§(1/n)1'(I'+1) z 1/rg’f J,f In equation 18, 9N is the proton 9 factor, 3N is the nuclear Bohr magneton, n is the number of protons in the unit cell, I is the nuclear spin quantum number, and r is the distance between two nuclei j,k, which is summed over all the intra- and intermolecular inter- actions (j,k). The sum f runs over all nonresonating nuclei. To calculate second moment by equation 18, the chloride ion of FeOCl was assumed to be 100% Cl35 and the shortest H--Cl distance was 2.3 A. The calculated second moment can then be used in equation 19 to obtain the width of the resonance line at half maximum height in units of Gauss. 112 1H1 = 2.35(AH2)1/2 (19) Equation 19 is valid assuming that the second moment is not dominated by a single nuclear pair interaction. This appears to be the case for the broad component in the FeOCl(TTF)1,8.5 1H echo spectrum. The experimental linewidth (60 kHz) is considerably greater than that calculated (AHl = 29.9 kHz) considering only H--H and H--Cl inter- actions, indicating that other strong magnetic interactions are impor- tant. It has been determined in the 1H NMR study of pyridine and picolines intercalated into VOCl that dipolar interactions between the unpaired spin density on the chloride ions and the protons provide a large contribution to the line widths at high magnetic field (4.7 T).95 A similar situation is expected for FeOCl(TTF)1,8.5. Line broadening due to the presence of unpaired spin density on the protons (hyperfine exchange broadening) of the radical cation of TTF would be small compared to the dipolar mechanism.1““ Broadening by the dipolar mechanism will contribute to inhomogeneous lines, while the hyperfine mechanism will contribute to homogeneous lines. In order to learn more about the interactions causing the line broadening in the FeOCl(TTF)1,8.5 Spectra, a series of hole-burning experiments were performed. In these experiments, a portion of the spectrum was irradiated with a continuous radio frequency field and then the entire spectrum was recorded (Figures 42-44). Field strengths (YBz/Zn) of 145 and 43 Hz were applied during the delay between the end of the acquisition and the echo pulse sequence for the spectra shown in Figures 41 and 42, respectively. Irradiation of a 113 00 000 00 0000000 00 0.000 0500 0:0 00 03000 .00 000000 00 03000 00000 00 0000020000 0200 000 00.00 022 00 :0 .00 0000.0 3&1 OO‘I OOQI OONI O OON 00v 000 p—p——F£brPPLrh—prhfbbhth~r :0\000 .000.00 1 20\~: 00.0000. 00<00 0. “mm «.mmw ”(a oo.om Izw nooew0.0mn I0m em.n0hp uuo 000 ZOHFo um>l where A0 = (3/8)eZQq/h ~ 4 x 105 S’l.1“7 This indicates that du-TTF undergoes isotropic reorienta- tion with a correlation time that is less than 30 us (which is the pulse separation, r). If, in fact, the reorientation is anisotropic, then the longest correlation time must be 30 us and the shortest must be greater than 2.5 us (l/Ao). The quadrupole echo spectrum of FeOCl(du-TTF)1,9 (obtained with a eo‘T‘eso sequence) is a broad, featureless line (Figure 52). The width at half height is 100 kHz. When the strength of dipolar inter- actions becomes a significant fraction of the quadrupolar interaction and wDT_Z 1 (wD = vaD, where V0 is the line broadening due to dipolar interactions and 1 is the time between pulses), it has been shown that the spectra are no longer symmetric about their center. If the spin dephasing due to dipolar interactions is not refocussed, the echo spectra are distorted.150 The 1H spectra indicate that the dipolar interactions between the unpaired electron density on the chloride and the 1H nucleus is a significant fraction of the elec— tronic quadrupolar interaction; it is, therefore, expected to be present in the FeOCl(dk-TTF)1,9 spectrum. The pulse sequence 90-1/2-18090-1/2-990-r/2-18090-r/2-echo has been shown to refocus both quadrupolar and dipolar interactions,”o reducing echo distortion due to spin dephasing resulting from dipolar interaction. The echo spec- trum obtained from this pulse sequence is shown in Figure 53. This 128 mZCCBESOmi c8 .5583 9.8 £83.62... :N 9; 2mm coat! OOONI o OOON coat. _ .1 Fl P Pl P :o\:aa .oan.mou n zo\N: ne.mon+. u40 um a cum: oo.on umo gum: column uno cum: oo.on :40 0mm: oo.on Ina cum: oo.m nun omm: oo.~_ nza OIUu «(gonna «<40a3m0<30 lP 009* (Dr P P L JUOuL uhhvo 130 echo spectrum has considerably better spectral resolution than in the quadrupole echo spectrum shown in Figure 52; four splittings appear to be present. Although the 2H NMR studies of Fe0(3l(d,.-TTF)1,9 are still at a preliminary stage, the large difference between the simple quadrupolar echo spectrum (Figure 52) and the quadrupolar-dipolar echo spectrun (Figure 53) is reproducible, and multiple splittings are always present in the latter spectrum. A central peak and several quadrupole splittings appear to be present. The central peak appears at the same frequency as the du-TTF line (Figure 51), and the splittings are 42, 84, 128, and 169 kHz for peaks A, B, C, and 0, respectively. The quadrupole coupling constant eZqQ/h and C-0 bonds in aromatic rings is about 185 kHz.”9 If no detectable motion took place (i.e., Tc > 1 x 10'3 s) then the C-0 bonds oriented at 0' and 90° with respect to the magnetic field would give resonance components that had splittings of 278 and 139 kHz, respectively. Orientations between 0' and 90' would give intermediate Splittings. The relatively sharp peaks often seen in powder 2H NMR specvtra are usually due to C-0 bonds oriented at 90’ with respect to the field. The peaks in Figure 53, which have a splitting of 169 kHz, cannot be due to 90° orientations because of the magnitude of the splitting. In order to obtain an understanding of complex 2H NMR spectra such as these, detailed simulation will be necessary. The work to date indicates that a strong dipolar interaction, probably between the unpaired electronic density on the chloride ion and the 2H nuclei, is present and must be refocused in order to avoid 131 Spectral distortion. Initial comparison of the quadrupolar-dipolar echo spectrum of FeOCl(dL+-TTF)1,9 with well characterized 2H echo spectra of other solids indicates that more than one component is present in the FeOCl(du-TTF)1,9 spectrum. The presence of more than one spectral component suggests that du-TTF domains in FeOCl(dL,-TTF)1,9 have different degrees of motional freedom and/or an average orientation with respect to their axes of motional averaging. The fact that an echo spectrum is not observed for FeOCl(du-TTF)1,9 indicates that the du-TTF reorientation is anisotropic and/or has a correlation time which is between 2.5 us and 30 us. Factors such as pulse spacing, spectrometer frequency and pulse power have a large effect on the spectra, and so details such as spectral asymmetry and exact splittings should not be considered final. More complete 2H NMR studies, which include variation of the parameters mentioned above, spectral simulations, relaxation measurements and spectra of oriented samples, will provide information on the dynamics of dk-TTF in FeOCl(d“--TTF)1,9 and aid the neutron data analysis by providing com- plementary information on the relative orientation of dn-TTF within the FeOCl layers. ELECTRONIC PROPERTIES OF THE INTERCALATES A combination of spectroscopic and physical methods has been applied to characterize FeOCl and the tetrathiolene intercalates. Differences in intercalant content (as much as 1%) appear to have little effect upon the data obtained from the techniques described below. Infrared Spectroscopy Extensive molecular vibration studies of charge transfer com- pounds with TTF,152 TMTTF,153 and TTT15“ as the electron donor molecule have been reported. There has been very little work on any TTN salt, presumably due to the difficulties in the synthesis of TTN.155’156 The infrared spectrun of pristine FeOCl shows only a single strong vibrational band at 485 cm‘l, which is assigned to the Fe-O stretch (Figure 54). Consequently, direct information on the guest species can be obtained by infrared spectroscopy. 0n the time scale of vibrational spectroscopy (10'13-10'1“ sec), electronic distribution in charge transfer compounds can be studied. For one-dimensional organic metals, vibrational spectroscopy provides evidence for partial charge transfer if the electronic transition does not introduce too much background absorption. The infrared spectrum for a compound having localized electronic charge distribution would exhibit frequencies of both neutral and ionized molecules, whereas 132 133 mwmwzpcmcma cw mew A .mucm5:mpmmo new: m.m\HAmPPVFooom can paced mo mguuwam umgmcmcH .vm mg=m_m TIEO 00' 000 OON- 80— OOON 80m COO”v d d d d :1 u d - d d d .Noon. .oson. a 3N A I .mpeme=mwmmm ;a_z Adhe-:uv_uomd Lee 55238888 cocacc=_ use .mm weaned TEo 00v 000 GON— ooc- OOON coon ooooc . d d d d d 1 q 1 a q d 4 ¢ 931:. 2 I 8?». n ON as a... 85: 888 6.8. \oNo sh on ”33.. aka... onn. omNN/J—nu T 1 9' . 9.0 ¢ .3. EC- 3 608 I s h I out. .. ow 83.. fine. r 833.. 25 2.. OS. Ono I 00 p p p p _ p p . p p p - 140 Table 21. Infrared spectral features for FeOCl(TMTTF)l, and the TMTTF infrared absorptions of (TMTTF)2X (x = 89,-, 010,-, and PFB') FeOCl(TMTTF)“13 (TMTTF)2xa Assignmenta 690(m) 940(sh,m) 935(ms) CH3 1332(m) 1340(vs) agv“ 1361(5) agv3 1470(m,br) 1438(mw,br) CH3 asymm 1560(m) 1560(sh) bluvzs (agv3) 2920(vw) 2923 CH3 asymm 2960(vw) 2973 Frequencies in wave numbers aReference 162. 141 The infrared data for FeOCl(TTT)1,9(tol)1,23 are listed in Table 22 and compared with values for TTT+ in TTT(X), where X = Cl‘, Br’, I‘ and SCN'.59 With some samples of the intercalate, certain peaks could be assigned to neutral TTT: these may be due to the presence of small amounts of unintercalated TTT on the surface of the intercalate. Otherwise, the data correspond to that of the TTT cation. Infrared data for TTN salts are nonexistent. The only salt whose synthesis has been published to date (TTN-TCNQ) is metallic and thus has no discernable infrared spectrum.155’163 The infrared data for FeOCl(TTN)l,9(tol)1,21 are listed in Table 23 and compared with those for those of the neutral compound. Most of the peaks are red shifted with respect to neutral TTN, as are those of the other intercalants, and may be assigned to the radical cation. The broad absorption (4000-1600 cm'l) observed in the infrared spectra (Figures 54 and 55) for FeOCl(TTF)l,8.5 and for FeOCl(d,,--TTF)1,9 is present in all the spectra of the intercalates and appears to be due to an electronic transition. Sample preparations with various grinding times and suspension of the solid in mineral oil as well as KBr had little effect upon the presence of this absorption. Organic charge-transfer compounds with conductivities on the order of 10‘“ to 10'2 (ohms-cm)“1 exhibit a similar broad absorbence in this region of the spectrum (4000-1600 cm'l), which has been attributed to the continuum of electronic transitions.15“ Infrared spectroscopy of the tetrathiolenes is consistent with the guest species being fully ionic. The presence of the broad absorption at about 0.6 - 0.3 eV is similar to that observed in all 142 Table 22. Infrared spectral features for Fe0Cl(TTT)1,9(tol) ,23 and the infrared absorptions of TTT(X), X = Cl', r', and SCN' and TTT' Frequencies in wave nunbers FeOCl(TTT)l,9 TTT(X)a TTT'a (t01)1/23 683(s,br) 685(w) 738(m) 733(vs) 752(m,sh) 751(m) 742(5) 761(m) 949(w) 973(m) 972(m) 968(m) 998(m) 998(VS) 1008(m) 1010(W) 1059(m) 1058(5) 1167(m) 1163(m) 1148(w) 1273(w,sh) 1274(vs) 1248(w) 1284(5) 1286(vs) 1311(5) 1313(5) 1305(vs) 1361(m) 1361(5) 1319(5) 1429(m) 1432(m) 1452(m) 1463(5) 1463(5) 1504(w) 1504(w) 1522(w) 1552(m) 1550(m) 1598(m) 1599(m) 1612(m) aReference 69. 143 Table 23. Infrared spectral features for FeOCl(TTl‘l)1,9(tol)1,21 and the infrared absorptions of TTN' FeOCl(TTN)1,9(tol)l,21 TTN'a 623(m) 623(5) 695(m) 670(m) 740(5) 725(w) 810(5) 797(5) 962(w) 815(w,sh) 975(w) 965(w) 1100(w,br) 1070(m) 1185(m) 1185(5) 1200(w) 1260(m,sh) 1260(s,sh) 1280(m) 1280(5) 1388(5) 1362(5) 1370(5) 1495(w) 1455(s,sh) 1570(w) 1540(5) Frequencies in wave numbers aReference 156. 144 the mixed valence compounds of TTF which are conductors155 and suggests that the intercalates should exhibit an increase in conduc- tivity over that of FeOCl. From the infrared and Raman data of FeOCl(TTF) FeOCl(du-TTF)1/9, and FeOCl(TTF) (tol) the 1/8-5’ 1/9 1/21’ TTF+ molecules within the host layers appear to exist as dimerized units. Although the steric interactions of TMTTF are different from the TTF intercalant, the electronic structure deduced from the infra- red data is consistent with TMTTF also being a fully ionic cation. Similar conclusions can be drawn for the TTN and the TTT intercalates. Charge transfer has occurred, presumably from the guest to the host lattice. Optical Spectra The Optical spectra of the intercalates are shown in Figures 56 and 57. The band gap reported for FeOCl is 1.9 eV.99 The value of the band gap corresponds to a "break" in the absorption curve.165 Both Figures 54 and 55 both show the spectrum for FeOCl, but at different concentrations. The intercalates have a greater absorption at low energy (0.5 eV) and more structure in their spectra at high energy (3.5 - 1.5 eV) compared to FeOCl. FeOCl(TTF)1/8.S and FeOCl(TTF)U9(tol)1,21 exhibit a broad absorption centered at about 1200 nm (1.03 eV) (Figure 56). A pos- sibly related transition is observed in the spectra of salts such as TTF(CA) (55) at 833 nm (see Figure 56). This corresponds to charge transfer between the (TTFT)2 ions as illustrated below.155:152 (TTFT) (TTFT) ..___» TTF°TTF2+ (a) 145 500 900 1300 1700 2100 2500lun I I I j I 3.5 2.5 1.50V 10V .5 9V Optical spectra of (flurolube mulls between NaCl plates) for FeOCl(---), FeOCl(TTF)1,3_5 (———), and TTF(CA)(ss)(-----). The baseline is indicated also (------). Absorption is an arbitrary scale Figure 56. 146 l I I I I I I I I I I I I J 500 900 1300 1700 2100 2500Iun I l l r l 3.5 2.5 1.50V 10V .5 9V Optical spectra (flurolube mulls between NaCl plates) for FeOCl(---o--), FeOCl(TMTTF)1,13(-——J, FeOCl(TTN)1,9(tol)l,21(-----) and FeOCl(TTT)1,9(t01)1/23 (---). Absorption is an arbitrary scale Figure 57. 147 For mixed valence salts, there is an additional, stronger, charge transfer band, which is located at 1110 nm for TTF(Br)0.71 and 1250 nm for TTF(SCN)O.S7,166 corresponding to the transition shown below. (TTF)+(TTF)° ______+ (TTF)0(TTF)+ (b) Although the absorption in FeOCl(TTF) is centered at 1250 nm, 1/8-5 the mechanism of charge transfer shown in (b) is unlikely due to the absence of evidence for partial charge transfer in the infrared data. The probable explanation is that the charge transfer observed for the fully ionic TTF(CA) (55) compound has shifted to lower energy in the intercalate. Alternatively, the absorption at 1250 nm could be attributed to an intervalence charge transfer band. The existence of intervalence charge transfer (Fe3+ + Fe2+) has been proposed based on the evidence from Mbssbauer spectroscopy for charge transfer between the guest and host.36 Most intervalence bands are broad, intense absorp- tions.167 Typically they obey the relation A = (vITx 2310)“2 (20) at 300 K, where “IT is the frequency of the absorption maximum in wavenumbers and A is the width at half height of the absorption. Using this formulation the half width value obtained (4299 cm’l) is about four times larger than an estimation of the experimental value (1288 cm‘l) of the absorption peak at 1250 nm for FeOCl(TTF)1,3.5. Because of the underlying absorption present in the spectrum, the peak 148 is not well resolved and could correspond to an intervalence charge transfer band, but due to the absence of this peak in any of the other intercalates this explanation seems highly unlikely. The visible region exhibits bands that have been attributed to TTF+.86 The poor intensity and resolution of these spectra (Figures 56 and 57) are similar to that observed for the layered compound FePS3 and its intercalates.168 Although infrared data for the FeOCl inter- calates indicate that charge transfer has occurred, no intervalence charge transfer band could be identified. The TTF intercalates have a band centered at 1250 nm which is attributed to charge transfer between the dimeric units of TTF. Conductivity The conductivity obtained from "single crystals" was of the same order of magnitude as the powder samples,169 indicating that the "crystals" were probably aggregates. The anisotropy of the "crystal" could be measured by using a standard four-probe technique for the ac plane and a two probe conductivity apparatus for the b axis.53’7° Conductivity values of 4.4 x 10‘“ and 4.1 x 10'7 (ohms-cm)'1, for the ‘35 plane and the b axis, respectively, of FeOCl(TTF)1,9(tol)1,21, show the anisotropy to be 22: 103. While single crystals are desirable, the problems in obtaining X-ray diffraction quality single crystals by the reaction procedure described have been insurmountable thus far. This has made it neces- sary to measure two probe pressed powder conductivities. Using an apparatus described elsewhere,53’7o temperature dependent conduc- tivity data for FeOCl and the tetrathiolene intercalates have been 149 obtained (Figure 58). Multiple samples were examined and data from samples which did not obey Ohm's law were disregarded. The variation of conductivity over the temperature range of -60 to 80'C indicates that both FeOCl and its intercalates behave as semiconductors. Band gaps listed in Table 24 were obtained from the least squares fit to the relationship,171 In a = -Eg/2k3T, where Eg corresponds to the band gap and k3 is Boltzmann's constant. It should be noted that although a band gap could not be identified in either the infra- red or the near-infrared region, there was a significant background in the optical and vibrational spectra of all the intercalates compared to FeOCl, which suggests an electronic transition. The pressed powder conductivity increases from £3. 10'7 for pristine FeOCl to 5.3. 10-2 for reocumm)“13 (Table 24). The. similarity in the slopes for the variable temperature measurements suggests that the eletronic conductivity is due to the host lattice.172 FeOCl(TMTTF)“13 exhibits the highest room temperature conductivity, further corroborating this hypothesis. It has been shown that to obtain high conductivity of the type seen in the organic metals the electron donor molecules must be stacked one on top of the other. This requirement is obviously not met by the TMTTF intercalate in which the plane of the tetrathiolene is parallel to the FeOCl layers, rather than perpendicular. X-Ray Absorption Near Edge Structure As a result of the physical methods employed thus far, an inter- esting question arises as to whether the electron from the intercalant is in the host layers, and if so, where. Interpretation of 150 mucmewsamowe auw>wuo=ucou canyon uwmmmcq mecca age so; F\H.mm aca nku.x.¢a AV? 1 nun. Ayn l l J lg¥l “ my” ngm ngw. * gum; 1:. .mm wcsmpd 151 Table 24. TWO-probe pressed powder conductivity measurements for FeOCl and the intercalates Compound Eg(eV) aRT(n-cm)'1 FeOCl 0.61 4.4 x 10'6 2.0 x 10'”7 FeOCl(TTF)1,9(tol)1,21 0.36 3.5 x 10-3 FeOCl(TTF)1,8.5 0.45 7.8 x 10'“ FeOCl(TMTTF)1,13 1.6 x 10-2 FeOCl(TTF)1,9(tol)1,22 0.35 2.1 x 10-3 FeOCl(TTF)1,9(tol)1,23 0.38 2.3 x 10-3 Mbssbauer data has led to various charge transfer models,173'17S including an electron hopping model proposed for Lewis base intercalates.36 The Mossbauer spectra of FeOCl(py)“3 and related amine intercalates, analyzed using the electron hopping model, are consistent with the presence of 10-13% ferrous iron sites within the FeOCl lattice.176 Without access to Mbssbauer Spectroscopy (until our very recent collaboration), X-ray Absorption Near Edge Structure (XANES), spec- troscopy provided an alternate method to probe the electronic state of the iron. XANES comprises the pre-edge, edge, and about 10-20 eV of the post-edge region of the X-ray absorption spectrum. Figure 59 m.a\sAahhv_uome 8:8 F8861 new Anv 22532: 3922, use 8cm“. A3 .80 .550on mmuo 539.3% 13.3. 8383. a“. 9: .mm 953“. 152 $3 >985 on: 3: ca: 3:. 3: on: 3: on: on: 3: 1 — _ a d 4 q d u A W|_l - q q - _ d _ - q — d u 1 q n J — d d 4 q u 4 1 q q u d q q 1 q q q q I I I II I ... hi i .I I1! I I l T 1.! 1 1| I. w L. . fl 1'! J a l . H H. . n ma 4. g I 3.50.4.5. 6:7... L n . H 9.»: 60.....2 I V; chubby _000k .2... II GONOEIII L I 60.... ll l .00... II I — _ p _ p — p . _ p _ FF _ p r p F» - — - l (P p b p — p L P p h p p n n P p p p p — p b b 153 shows the edge shifts for various iron +3 and +2 compounds. The spectra of the materials containing ferrous iron are shifted to lower energy with respect to those of materials containing ferric iron. The Fe K-edge XANES spectra for FeOCl and for FeOCl(TTF)1/8.5 are shown in Figure 5. For all the intercalates the edge is shifted to lower energy, indicative of the presence of Fe+2. Small shifts such as seen in Figure 57 have also been observed for MTSZ (M = first row transi- tion metal, T = Ti, Zr, V, Nb).177 Unfortunately, XANES is not a quantitative technique, and although the data indicate that charge transfer to FeOCl has occurred, the edge shift cannot provide a measure of the percentage of ferrous sites within the lattice. Pre- liminary Mussbauer data on the tetrathiolene intercalates verify the presence of 8-10% ferrous sites, but further work is necessary before these results can be considered conclusive. Magnetic Studies of FeOCl and the Intercalates Variable temperature magnetic susceptibility The molar susceptibility of powdered FeOCl and the intercalants between 5 and 300 K is shown in Figures 60 and 61. The data are interpreted as due to strong antiferromagnetic coupling. The room temperature magnetic moments were calculated using the spin—only relationship (neff = 2.828 JXT) and are low compared to the spin only value for high-spin iron +3 (see Table 25). A considerable amount of information on the magnetic interactions has been obtained from Mbssbauer spectroscopy,35:'*°a7°a1°°’178 and very little by variable temperature susceptibility.“1a179 It has been suggested that the X(T) data below about 350 K reflect short 154 exact 85:2_ £323 m.m\bfidhpv_uoae uca m5\.AupPzpv_Qoaa mu:_on www.5maocsu .Puomm cow h m> capes x .08 ac=m_u n.0\— r._\ $5.806... o guppszoOad x Goo“. a. aw . 57 . mo .N l. wgimm .N l 80in .m 801mm .m liwochxupv cosmm .v ;3_; m~\_fl_oava\sAehev_uoaa use z~\sA_ouvm\sAzphv_uoaa ._uoad 280 » ms aa_os x mucwoa ms» smzocgu gamma mmcwp .Ho mc=m_m 155 EN _ . hhh;000u.* 2bh;000m O .OOau X 3: . aw . IOOImo.N wooiwm.m mooimo.m moonwm.m $8-8 .... l oo:wm.v 156 Table 25. Effective moments calculated for high spin Fe+3, FeOCl, and the intercalates at 300 K using spin-only formula Compound "eff Fe+3 5.9 FeOCl 2.76 FeOCl(TTF)1,8.5 2.8 FeOCl(TMTTF)1,13 2.8 FeOCl(TTF)1,9(tol)1,22 2.8 FeOCl(TTF)1/9(tol)1/23 2.9 range antiferromagnetic ordering of FeOCl."1 The Mossbauer data show no evidence for magnetic ordering at 300 K for this sample.180 After investigating various models employing linear chain theories,181 two-dimensional theories and other models which considered extended chains,132’183 it was found that the model that best describes the data so far is based on Short range interactions.13“ The model consists of a system of three spins interacting in the form of a linear chain. It has been proposed that the short range magnetic interactions are fluctuating on a time scale faster than can be resolved by MOSSbauer spectroscopy.“1 The short range interaction model provides a reasonable fit to the data, indicating that the hypothesis of short range magnetic order is basically correct. 157 The variable temperature susceptibility data for the intercalates (shown in Figures 60 and 61) are very similar to those of FeOCl. There is a low temperature Curie tail in the intercalates, which may be due to impurities in the intercalates, otherwise the curves are almost temperature independent between 90 and 300 K. The similar slopes for the susceptibility curves between 90 and 300 K for FeOCl and the intercalates suggest that the short range magnetic inter- actions observed for FeOCl are also present in the intercalates. Neutron diffraction Mossbauer spectra of both single crystal and powder show the onset of a magnetic hyperfine interaction at 92 i 3 K for FeOCl.“0:76:173 Analysis of the spectra shows that more than one set of crystallographically equivalent Fe+3 sites is present and that all the Fe+3 spins cannot be parallel to either the_a, b, or.c axis.177 This is in agreement with preliminary Mossbauer data in which the two different lattice sites can be resolved for pristine FeOCl.18“ AT 10 K the neutron diffraction data show the presence of new peaks (Figure 62), compared to the 300 K data, which are attributed to the magnetic lattice of FeOCl.185 The magnetic structure which was proposed for FeOCl by Adam and Buisson from the analysis of the neutron data is shown in Figure 63. This structure has a doubled crystallographic lattice along both the a and b_axis and the Spins turned by 99' from one unit cell to the next in the c direction. 158 8388 Asouponv x OH new Agapv g com P8061 soc =o_»uaccc_u amazon accuse: gaddmlo an 3 -3 “flitf. 34 ‘ . . own-”.moowcouttmu‘. . - . It”, x0. ”1., . xoom nu. m 0% CIIIDS'eavxiuwnoorkmaunaN nZHHfi 520850060 .No «same; 159 , . o Z=O - I § % @ Z=O.5 : I : l o Z=I.O C‘ .. _-.J,.._ __ ‘>-——‘-——’b—#-—— ’ I / 1‘1?“"':z>" 4-0 \ \ \ \ \ \ 0*, ------- ‘I l I .1 I I l \ \ O" k»—_--_-d-_- I \ : “ I I I L P: | \ | \ l l l I Figure 63. Magnetic Iattice for FeOCl (8 unit cells) showing the direction of Spins proposed by Adan and Buisson 160 The low temperature neutron data for FeOCl(TTF) (Figure 1/8-5 64) show no new peaks, thus indicating that the long range magnetic order has been abolished upon intercalation. Considering the anti- ferromagnetic coupling across the b_axis (Figure 63), it is not surprising that intercalation abolishes the long range order. Variable magnetic susceptibility measures bulk effects and no Neel temperature is observed for either FeOCl or for the intercalates over the available temperature range. Fitting the magnetic suscepti- bility of the intercalants will provide information on the changes in the short range magnetic interactions. More direct information can be determined from Mbssbauer spectroscopy, and those experiments are currently underway. 161 name Asouuonv x 0“ new Aaouv g com .m.m\cauhhv,uoau to. cowuumaccwu Lanzoa ectpaaz .eo ac=m_a §6