LIBRARIES RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES wiII be charged if book is returned after the date stamped beIow. "--—‘M u L ‘1‘ SYNTHESIS, CHARACTERIZATION AND MAGNETIC PROPERTIES OF CRYSTALLINE ELECTRIDES AND ALKALIDES BY Ahmed Saad Ellaboudy A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1984 ABSTRACT SYNTHESIS, CHARACTERIZATION AND MAGNETIC PROPERTIES OF CRYSTALLINE ELECTRIDES AND ALKALIDES BY Ahmed Saad Ellaboudy Two new crystalline alkalides, Rb+18C6.Na- and Rb+18C6.Rb- and the first crystalline electrides, Cs+(18C6)2.e_, Cs+C222.e- and Rb+18C6.e- were synthesized, analyzed and characterized by optical, electrical and magnetic techniques. Rb+18C6.Na-, Rb+18C6.e- and Cs+(18C6)2.e- behave as semiconductors with apparent band gaps of 0.9, 0.4 and 0.9 eV respectively. 23Na-MASS-NMR spectra of MNaLn (M=K,Rb,Cs; L=18C6,C222; n=l,2) show only the Na- peak at ~61 ppm proving that they are all sodides. N§b222.Na- shows the Na- peak and that of Na+C222. Field dependence and proton decoupling showed that Na+C222 is quadrupole shifted and broadened (Eégg==l.2 MHz; n =0.l) while Na- is broadened by dipolar coupling to protons. I 133Cs-MASS-NMR of the compounds Cs+(18C6)2N- (N=I,SCN, Na,K,Rb) shows the Cs+ peak at ~-58 ppm, essentially inde- pendent of anion. CslBCG is the ceside, Cs+(18C6)2.Cs- since it shows peaks at -61 and -228 ppm. The electride Ellaboudy, Ahmed Saad Cs+(18C6)2.e- shows only one peak (+81 ppm at--20°C) that shifted upfield with increasing temperature indicating a 21 electrons-cm-B. contact electron density of 2.39 XIO CsC222 shows peaks at +138 and +238 ppm attributed to "exclusive" and "inclusive" Cs+ complexes respectively. 87Rb-MASS-NMR could not detect complexed Rb+ but Cs+(l8C6)2.Rb- yielded an Rb- peak at -187 ppm and KRbl8C6 gave a weak doublet (-185 and -194 ppm). Sodide salts generally showed a single EPR line at g =2.0023, often superimposed on two or three other lines. This was attributed to electron trapping at anionic vacancies. Rb+18C6.Na- showed a narrow line at g =2.0023 and hyperfine lines (g==l.9974) due to electron coupling to the cation with Aiso =58.8 and 196.6 G for 8512b and 87m: respectively (16% atomic character). CsC222 shows a very weak inhomogenously broadened axial symmetry powder pattern with gl =2.oozo and gII =1.9771. Rb+18ce.e" has a tempera- ture independent Lorentzian EPR line at g==2.0017 with AHp_p==4.7 G. Cs+(18C6)2.e- gave an asymmetric line (g =2.0023, AH :=o,5 G, both independent of temperature P'P from 3 to 250 K). The 9 ratio increased with increasing temperature indicating substantial microwave conductivity and an apparent band gap of 0.1 eV. Cs+(18C6)2.e- magnetic susceptibilities very nearly obeyed the Curie-Weiss Law} xfi==fC/(T-G) with f =0.74, o =—1.4 K, while Rb+18C6.e- and Cs+c222.e" had only about 1% unpaired spins. In the name 06 God the moat me/Lctfiut and the moat benefitcent. To my 601.60., Fadéa -11- ACKNOWLEDGMENTS The author wishes to express his sincere gratitude to Professor James L. Dye for his guidance, encouragement and whole-hearted support throughout this study. I would like to thank those with whom I worked most closely, Drs. Long Dinh Le, Dheeb Issa, Bradley Van Eck; my colleagues in the solid state group, Odette Fussa, Mary Tinkham, Rui Huang, Zheng Li, Francoise Tientega and especially Margie Faber and Steve Dawes for their aid, discussions and moral support. Thanks are also extended to the enzyme group, Zexia Barnes and Iraj Behbahani. Special thanks go to Dr. Pat Smith in the Analytical Laboratory of Dow Chemical Company and Professor Eric Oldfield's group in the Chemistry Department of the University of Illinois for their help in the magic angle NMR work. Financial support is gratefully acknowledge from the Chemistry Department, Michigan State University through a teaching assistantship, National Science Foundation (Grant DMR 79-21979) for a research assistant- ship, the Dow Chemical Company for a summer 1983 fellowship and the SOHIO Company for a 1983-84 fellowship. -iii- Thanks go to the glassblowers: Keki Mistry, Manfred Langer and Scott Bankroff for their excellent and timely service; for without their help this work would have been impossible. Thanks also go to Ms. Margy Lynch for doing an excellent job in typing this Dissertation. Thanks to my beloved University, Alexandria Univer- sity in Egypt, the clerical staff in the Embassy of Arab Republic of Egypt-—Cultural and Educational Bureau in Washington, D.C., especially to Mr. Hany El-Malty, and all my friends in the Egyptian Club in East Lansing, especially Mr. Mohsen Shabana for making me feel at home with their friendship and brotherhood. Special thanks go to my family and my wife's family for their unending encouragement, understanding and moral support. Above all, a special "thank you" goes to my wife, Fadia, whose understanding and extremely long patience was vital to the completion of this degree. To her I dedicate this thesis. _iv_ TABLE OF CONTENTS PAGE LIST OF TABLES. . . . . . . . . . . . . . . . . . . . .Viii LIST OF FIGURES. . . . . . . . . . . . . . . . . . . . ..ix CHAPTER I -- INTRODUCTION. . . . . . . . . . . . . . . . .1 A. Alkali Metal-Ammonia Solutions. . . . . . . . . .1 B. Alkali Metal-Amines, Ethers and Other Solvents. . . . . . . . . . . . . . . . . . . . .5 C. Metal-Ammonia Compounds. . . . . . . . . . . . .10 D. Alkali Metal Anions (M'). . . . . . . . . . . . 15 E. Trapped Electrons in Solids. . . . . . . . . . .16 F. Macrocyclic and Macrobicyclic Diamine- Polyethers. . . . . . . . . . . . . . . . . . . 22 G. Alkalides and Electrides. . . . . . . . . . . . 26 H. Objectives of the Present Work. . . . . . . . . 32 CHAPTER II -- EXPERIMENTAL. . . . . . . . . . . . . . . .34 A. Materials. . . . . . . . . . . . . . . . . . .34 1. Alkali metals. . . . . . . . . . . . . . . .34 2. Complexants. . . . . . . . . . . . . . . . .35 3. Solvents. . . . . . . . . . . . . . . . . . 37 B. Glassware Cleaning. . . . . . . . . . . . . . . 38 C. General Synthesis Procedure. . . . . . . . . . .38 D. Analysis. . . . . . . . . . . . . . . . . . . 41 1. H2 evolution. . . . . . . . . . . . . . . . 42 2. pH titration. . . . . . . . . . . . . . . . 44 3. Flame emission. . . . . . . . . . . . . 45 4. Quantitative 1H NMR. . . . . . . . . . . . .45 E. Sample Handling and Instrumental Techniques. . .46 1. Optical absorption spectra. . . . . . . . . 47 2. Pressed powder conductivity. . . . . . . . .48 3. Solid state NMR. . . . . . . . . . . . . . .49 4. EPR spectra. . . . . . . . . . . . . . . .50 5. Magnetic susceptibility. . . . . . . . . . .51 F. Model Salts. . . . . . . . . . . . . . . . . . .52 -V— CHAPTER III -- SYNTHESIS AND CHARACTERIZATION Introduction. . . . . . . . . . . . Methods of Synthesis from Solution. A. B. CHAPTER A. B. 1. Method 1. . . . . . . . . . . . . . 2. Method 2. . . . . . . . . . . . . . 3. Method 3. . . . . . . . . . . . . . . Results and Discussion. . . . . . . . . . . . 1. Synthesis. . . . . . . . . . . . . . . . 2. Analysis. . . . . . . . . . . . . . . . . 3. Optical spectra. . . . . . . . . . a. RbNa18C6 and RbNaC222. . . . . . . . b. Rbl8C6 and Rb218C6. . . . . . . . . . c. C318C6 and Cs(18C6)2. . . . . . . . . d. CsC222. . . . . . . . . . . . . . . . 4. D.C. pressed powder conductivity. . . . 5. Differential scanning calorometric (DSC) studies 0 O O O O O O C O O O O O O O O 0 IV -- SOLID STATE NMR. . . . . . . . . . . . . Introduction. . . . . . . . . . . . . . . . Magic Angle Sample Spinning (MASS)-NMR. . . . 1. 2. 3. Direct magnetic dipolar interactions. . . Chemical shift interactions. . . . . . . Electrical quadrupolar interactions. . . Results and Discussion. . . . . . . . . . . . 1. 23Na-MAss-NMR. . . . . . . . . . . . a. Identification of the species. . . . b. Frequency dependence and proton decoupling. . . . . . . . . . . . . . 133cS-MAss-NMR. . . . . . . . . . . . . . a. Species identification and frequency dependence. . . . . . b. The chemical shift of Cs+(18C6)2.e . c. Electron doping in the ceside compound. . . . . . . . . . . . . . . 37Rb-MAss—NMR. . . . . . . . . . . . . . The chemical shifts of alkali metal anions. . . . . . . . . . . . . . . . . . CHAPTER V -- MAGNETIC PROPERTIES OF ALKALIDES AND A. ELECTRIDES. . . . . . . . . . . . . . . Electron Paramagnetic Resonance (EPR). . . . l. 2. Introduction. . . . . . . . . . . . . . Results and discussion. . . . . . . . a. Trapped electrons in sodides "F-center alkalides". . . . . . . . . b. Electrons in electrides . . . . . . . i. EPR of singlet-ground state electrides. . . . . . . . . . ii. EPR of doublet- -ground state electrides. . . . . . . . . . . . -vi- PAGE 53 53 54 54 55 56 .58 .68 .69 . 69 . 74 . 77 .85 . 91 . 91 .97 100 .109 109 109 .117 .124 .124 138 .142 148 '154 '157 157 157 170 ~170 ~183 ~183 '186 PAGE B. Magnetic Susceptibility. . . . . . . . . . . . 198 1. Introduction. . . . . . . . . . . . . . . .198 2. Results and discussion. . . . . . . . . . .202 a. The electride Cs+(18C6)2.e’ . . . . . .202 b. The ceside, Cs+(18C6)2.Cs’ . . . . . . 204 c. The di-electrides. . . . . . . . . . . 206 CHAPTER VI -- CONCLUSIONS AND SUGGESTIONS FOR FUTURE WORK. . . . . . . . . . . . . . . .212 A. Conclusions. . . . . . . . . . . . . . . . . . 212 B. Suggestions for Future Work. . . . . . . . . . 216 BIBLIOGRAPHY. . . . . . . . . . . . . . . . . . . . . . 218 -vii- LIST OF TABLES TABLE PAGE I Synthesis and Description of Crystalline Electrides and Alkalides. . . . . . . . . . . . . 59 II Results of the Analysis. . . . . . . . . . . . . .64 III DSC Results for Alkalides and an Electride. . . . 89 IV Results of 23Na-MAss-NMR at 52.94 MHz. . . . . . 112 V Proton-Decoupled and Frequency Dependence of the Chemical Shift and Linewidth of 23Na NMR. . . . . . . . . . . . . . . . . . . . .118 VI 133Cs-NMR Results. . . . . . . . . . . . . . . . 125 VII 87Rb MASS-NMR.Results . . . . . . . . . . . . . .150 VIII Chemical Shifts of Alkali Metal Anions. . . . . 155 IX Resonance Field of Some Hyperfine Lines. . . . . 180 X Parameters of the Curie-Weiss Equation for Cs+(18C6)2.Cs'. . . . . . . . . . . . . . . .206 -viii- FIGURE 10 11 12 13 LIST OF FIGURES Classification of EPR spectra 0~296 K) of fluid metal solutions on the basis of the product (TM) Aiso: THF==tetrahydrofuran, EA = ethylamine , 1 , 2PDA = 1 , 2-propane- diamine, EDA==ethylenediamine, AM= ammonia [41].. . . . . . . . . . . . . . Packing of Na +C222 and Na (solid circles) in the crystalline sodide Na+C222. Na' [30]. Apparatus for distribution of alkali metals under vacuum. . . . . . . . . . . . . . . . Apparatus for the synthesis of crystalline electrides and alkalides. . . . . . . . . . Apparatus for the optical spectra of thin films (5a) and hydrogen evolution (5b) . . . Optical spectra of films of RbNa18C6 ( ----- ) and RbNaczzz (.....) o o o o o o 0 Optical spectra of films, made from methyl- amine solutions of Rb18C6 ( ----- ) and Rb218C6(---—°-—) thin films (-'-") thick film. 0 O O O O O O O O O O O O O O O O 0 Optical spectra of films of the compounds C318C6 (-—°—---) and Cs(18C6)2 ( ). Data for the compound C318C6 are taken from reference 98. . . . . . . . . . . . . . . Optical spectra of film of compound CsC222. Ohm' 3 Law plot of polycrystalline powder of Rb+18C6. Na" . . . . . . . . . . . . . . Ohm's Law _plot of polycrystalline powder of Rb+18C60 e- O O O O O O O O O O O O O I O 0 Ohm‘ 5 Law plot of polycrystalline powder of Cs+ (18C6)2.e . . . . . . . . . . . . . . . . A plot of log conductivity vs. +reciprocal temperature of the compound Cs+ (18C6)2. Cs' (open circles) and Cs+ (18C6)2. e (solid circles). . . . . . . . . . . . . . . -ix- PAGE .11 28 . 36 . 39 .43 70 71 .75 76 .83 FIGURE PAGE 14 A plot of log resistivity vs. reciprocal temperature for polycrystaIIine Rb+1BC6.Na‘. . . . . . . . . . . . . . . . . . . .86 15 A plot of log resistivity vs. reciprocal temperature for polycrystaIIine Rb+16C6.e . . . . . . . . . . . . . . . . . . . . 87 16 DSC for a polycrystalline sample of Cs+(18C6)2.Na’. . . . . . . . . . . . . . . . . . 88 17 DSC for polycrystalline samples of _ Rb+18C6.Na' ( ) and Cs+(18C6)2.e (°°°-°). . . . . . . . . . . . . . . . . . . . . .90 18 A diagram illustrating the motion of a typical inter-nuclear vector, ri-, when a solid is rotated with angular velocity wr about an axis inclined at angle 8 to Ho- . . . 96 19 (a) NMR powder pattern of axially symmetric crystals; (b) 2 Na-MASS-NMR spectrum of an equimolar mixture of NaCl and NaNO3. . . . . . . 106 20 Static (dotted line) and spinning (solid line) 23Na-NMR spectra of Na+18C6.SCN' at 132.35 MHz. . . . . . . . . . . . . . . . . . 110 21 Static and Spinning 23Na-NMR spectra of Na+c222.Na‘ at 52.94 MHz. . . . . . . . . . . . .113 22 23Na-MASS-NMR spectrum of one sample of K+18C6.Na' at 132.35 MHz. . . . . . . . . . . . .115 23 23Na-MASS-NMR spectra of Na+C222.Na' at three different frequencies. . . . . . . . . . . 120 24 ‘Observed and simulated 23Na-NMR signal of Na+C222 in Na C222.Na at different frequencies. . . . . . . . . . . . . . . . . . . 121 25 23Na-MASS-NMR of K+C222.Na- at different frequencies. . . . . . . . . . . . . . . . . . . 123 26 l33Cs-MASS-NMR spectra of the two crystalline compounds Cs+(l8C6)2.Cs- and Cs (18C6)2.e‘ at 42.24 MHz. . . . . . . . . . . .130 27 23Na and l33Cs-MASS-NMR spectra of the compound Cs+(18C6)2.Na’. . . . . . . . . . . . . 131 28 Static (top) and spinning (bottom) 133Cs- MASS—NMR spectra of the compound Cs+(18C6)2.e‘ at 65.61 MHZ. . . . . . . . . . . .132 29 133Cs-MASS-NMR spectra of the two crystalline compounds Cs+C222.SCN .HZO (top) and Cs+c222.e' (bottom) at 42.24 MHz. . . . . . . . .135 -X— FIGURE 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 133Cs—MASS-NMR spectra at 65.61 MHz of the compound Cs+C222.e' at different spinning frequencies: A: 1.8, B: 2.6, C: 3.8 and D: 4.0 KHz. . . . . . . . . . . 133Cs-MASS-NMR spectrum of one sample of Cs+(18C6)2.Cs' (method 2) at 65.61 MHz. . . . 133Cs-MASS-NMB spectrum of one sample of Cs+(18C6)2.Cs (method 3) at 42.24 MHz. . . 133Cs-MASS-NMR spectrum of one sample of Cs+(l8C6)2Cs' (method 3) at 65.61 MHz. . . . . 133cS (at 65.61 MHz) and 37Rb (at 163.6 MHz) MASS-NMR spectra of the compound Cs+(18C6)2.Rb'. . . . . . . . . . . . 87Rb-MASS-NMR spectrum of the compound K+18C6.Rb’ at 163.6 MHz. . . . . . . . . . . . Magnetic energy levels for I =3/2 and S =1/2. The solid lines are energy levels computed from Eq. (v.10); dotted lines are the correspondin first-order energy levels (Eq. (V.7)) [133%. . Variation of EPR signal amplitude with microwave power (PC) for a homogeneously broadened line (solid curve) and an inhomogeneously broadened line (dotted curve). Typical first derivative lineshape of the electron spin resonance absorption in thick metals. . . . . . . . . . . . . . . . EPR spectra of polycrystalline sample of Na C222.Na' at different microwave power levels . . . . . . . . . . . . . . . . EPR spectrum of a polycrystalline sample of C3 (18C6)2.Na . . . . . . . . . . . . . . . Variation of the relative intensity of the central line of the EPR spectrum Of Cs+(18C6)2.Na- with the square root of the microwave power level. . . . . . . . . . EPR spectra of a polycrystalline sample of Cs+(18C6)2.Na at different temperatures. EPR spectrum of a polycrystalline sample of Rb 18C6.Na at 2.8 K. . . . . . . . . . . Variation of the relative intensity of the central line of the EPR spectrum of Rb+18C6.Na' with the microwave power level. . . . . . . . . . . . . . . ’Xi" PAGE 137 .144 .145 146 152 153 161 165 .169 172 174 175 .177 178 179 FIGURE PAGE 45 EPR spectrum of a polycrystalline sample of Rb 18C6.e' at 123 K. . . . . . . . . . . . . .184 46 Variation of the relative intensity of the EPR line of Rb+18C6.e' with the square root of the microwave power level. . . . .185 47 EPR spectrum of a polycrystalline sample of Cs+c222.e’ at 2.8 K. . . . . . . . . . 187 48 Variation of the relative intensity of the EPR line of Cs+c222.e' with the square root of the microwave power level. . . . . . . . . . . . . . . . . . . . . . 188 49 EPR spectrum of a polycrystalline sample of Cs (18C6)2.e' at 174 K (sample in 2 mm tube) 0 o o o o o o o o o o o o o o o o o o o o o 189 50 Variation of the linewidth (AH -p) of the ‘ EPR line of Cs+(18C6)2.e' with temperature. . . . . . . . . . . . . . . . . . . 191 51 Variation of_A/B ratio of the EPR line of Cs+(18C6)2.e with temperature (sample in 3 mm tube). . . . . . . . . . . . . . . . . . 192 52 a/6 gs. A/B for Spherical metal particles. . . . 193 P 53 Plot of the relative A.C. resistivity (solid line) at X-band microwave frequency and D.C. resistivity (dashed line) vs. temperature. . . . . . . . . . T_. . . . . . . . 194 54 A plot of log relative microwave resitivity with l/T. . . . . . . . . . . . . . . . . . . . .196 55 A plot of l/Xfi vs. T for a polycrystalline sample of Cs+(1856)2.e . . . . . . . . . . . . . 203 56 A plot of xfi vs. T for a polycrystalline sample of Cs+TI8C6)2.Cs (method 3). . . . . . . 207 57 A plot of xe vs. T for a polycrystalline sample of Rg+TI8C6).e' . . . . . . . . . . . . . 209 58 A plot of x§+vs. T for a polycrystalline sample of Cs C222.e . . . . . . . . . . . . . . .211 -xii- CHAPTER I INTRODUCTION The discovery of two new classes of ionic solids, named electrides and alkalides, came as a result of the use of macrocyclic polyethers to enhance the solubilities of alkali metals in amine and ether solutions. The similarity between these two new types of ionic solids and metal-ammonia compounds is striking. The behavior of the two classes of macrocyclic polyethers and their complex compounds with simple alkali metal salts provides a good model far the nature of the complex cation. Trapped electrons in solids (F-centers) and frozen solution (glasses) may well provide examples of a "dilute electride". Accordingly, all of these subjects provide background for the present work and will be reviewed here. A. Alkali Metal-Ammonia Solutions It has been over a century since Weyl in 1864 [1] discovered that sodium and potassium metals form blue- colored solutions in liquid ammonia when dilute and bronze- colored solutions when concentrated. Seven years later, Seely [2] recovered sodium and potassium from ammonia -1- -2- solutions by solvent evaporation. He concluded that alkali metals do not react with liquid ammonia and that these solutions are similar to those of ordinary salts in water. Since that time, metal-ammonia solutions have attracted the attention of both chemists and physicists. This controversial field has already resulted in six conferences named "Colloque Weyl" [3-8] which have provided a great deal of information about the nature of these systems. However, the detailed nature of the species present is still the subject of much speculation and controversy. In general, liquid ammonia dissolves alkali, alkaline earth (except Be), and Yb and Eu metals. The alkali metal solutions have been studied more extensively than those of other metals and will be the subject of this section. The solubilities of alkali metals in liquid ammonia vary from 16 MPM (mole percent metal) for Na, K and Rb to 20 MPM for L1 to 65 MPM for Cs and do not vary greatly with temperature. The properties of alkali metal-ammonia solutions vary from electrolytic for dilute and moderate concentrations to metallic for concentrated solutions. In the range of 3 to 9 MPM alkali metal-ammonia solutions exhibit a nonmetal to metal transition and, except for cesium solutions, cooling the homogeneous solution gives rise to liquid-liquid phase separation. A low-density bronze metallic phase floats on top of the more dense, less concentrated, dark blue phase. Only the optical and -3- magnetic properties of alkali metal-ammonia solutions will be reviewed here, since in principle much information can be extracted from the knowledge of these two properties. For very dilute solutions (<10"3 MPM), the Optical absorption spectrum of alkali metal-ammonia solutions consists of a distinct band in the infrared which drops to low absorbance on the low energy side. The position of maximum absorption, A , is independent of the metal [9]. max Based upon pulse radiolysis studies [10L this band is assigned to the solvated electron. It is universally accepted that only solvated electrons and solvated cations are present in very dilute solutions (<10"3 MPM). In the 3 to 10-1 concentration range 10- MPM there is an observable red shift of the band position with solution composition. This was interpreted in terms of the presence of a second diamagnetic absorber and the results were confirmed by the change of the peak linewidth with temperature and pressure. For concentrated solutions only reflectance spectra were obtained and showed a plasma type absorption due to the conduction electrons [ll]. O'Reilly [12,13] measured the Knight shifts [14] 0f 7Li, 23Na, 87Rb, 133Cs and 14N nuclei in alkali metal- ammonia solutions at 298 K in the concentration range 0.1 to 1.0 mole-liter-l. Although there are some discrepancies between his results and those obtained by McConnell and Hollm [15], and Acrivos and Pitzer [16], the general behavior is clear. The Knight shift results -4- indicated very small electron spin density at the metal nucleus (1% of the free-atom value) and a relatively large spin density at the nitrogen nuclei. The 14N Knight shift appeared to be independent of the metal cation and increased with increasing metal concentration,whi1e those of the alkali metal cations were independent of metal concentration and varied by a factor of 100 on going from Li to Cs. In contrast, the proton Knight shift [17] showed a "negative" electron spin density. The calculated average total spin density at the nitrogen nuclei was found to be a constant on the order of 6.44 XlO24 cm.3 and was independent of metal concentration [18]. The static [19-22] and microwave (EPR) [23-25] spin only susceptibility showed similar results. At high dilution (~6 ><10-3 M) the free spin concentration was equal to the metal concentration and the susceptibility was independent of the metal cation up to 0.1 M. Increasing the metal concentration to 1.0 M gave an increase in the electronic susceptibility which was less than expected indicating that a spin-pairing process occurs in this region. Above 1.0 M the susceptibility increased, showing a metallic kind of behavior and indicating progres- sive electron delocalization. The EPR spectra of dilute alkali metal-ammonia solutions showed extremely narrow symmetrical lines of the order of a few millgauss in width [26,27]. The electron spin relaxation mechanism which gives rise to extremely narrow EPR lines is the modulated hyperfine interaction of the solvated electron with the nitrogen atoms -5- of the solvation shell. Also the relaxation times (Tl==T2) are essentially independent of metal concentration at dilute to moderate concentrations. Although no resolved hyperfine splitting could be observed, the measured alkali metal Knight shifts reflected the cation-electron interaction which was washed out by the very short electron-nuclear correlation time (TM). Increasing the metal concentration to the region of the non-metal to metal transition resulted in the appearance of highly asymmetrical EPR lines with A/B (ratio of the low-field amplitude to high-field ampli- tude of the first derivative EPR line) greater than 1, characteristic of metallic samples with thicknesses comparable to or greater than the skin depth. This is now known as the "Dysonian" lineshape [28]. In summary, one might speculate about the species present in alkali metal-ammonia solution as follows: in very dilute solutions alkali metals dissociate to solvated cations and the paramagnetic solvated electron. An increase in concentration leads to electron-cation interaction through the formation of "loose" aggregated complexes of the solvated electron such as Mg-e; (paramagnetic) and the triple ion eg-Mg-e; (diamagnetic). The very short electron correlation time TM gives the cation a bystander role in these solutions. B. Alkali Metal-Amines, Ethers and Other Solvents In low dielectric constant amines, ethers and amides, alkali metals dissolve to a lesser extent than they do in -5- ammonia. The dramatic decrease in the dielectric constant relative to ammonia solutions is of cardinal importance in determining the change in cation and electron solvation and hence electron-cation interactions. Then, the association of the solvated electron and solvated cation can no longer be described as weakly interacting ion-pair species and one can safely make the general comment that the majority of aggregate species are more "tightly bound" than in ammonia solutions. Therefore, metal-dependent properties are expected for these solutions rather than the "null results" of metal-ammonia solutions. Optical absorption studies [29] of alkali metals in amine and ether solutions showed metal-dependent absorption spectra. In addition to the solvated electron band there is at least one more peak at higher energy which depends upon the metal cation. Based on pulse radiolysis studies [10] of alkali metal cation solutions in amine and ether solvents, the correlation of the band positions with the ns-rnp transitions in gaseous alkali atoms [30] and the similarity to charge transfer to solvent (CTTS) bands, the metal dependent bands were attributed to alkali metal anions [30]. Solutions of alkali metals (except Li and Na) in ethylenediamine (EDA) [29] showed bands of both the alkali metal anion and the solvated electron with absorbance ratios that increased with increasing metal concentration Li-EDA showed only the solvated electron band while Na-EDA showed only the Na- band. The presence of one or two peaks -7- in the optical absorption spectrum may be understood in terms of the dissociation reaction _ + .. M g‘M -+2e (1.1) s s EPR spectra of alkali metal-amine solutions showed different behavior upon changing solvent polarity. Lithium- methylamine solutions are well characterized in terms of "ammonia-like" aggregate species [31,32]. A narrow EPR singlet, which is similar to those of dilute metal-ammonia solutions, is observed but in contrast to metal-ammonia solutions, a recent spin echo study [33] showed that the electron relaxation times are not equal. This suggested that the relaxation mechanism for dilute ammonia solutions (motional-modulation of the electron-nitrogen hyperfine interaction) cannot alone account for the inequality. Upon changing the solvent to ethylamine, lithium solutions showed resolved EPR signals due to coupling to the nitrogen nuclei. Sodium and potassium solutions in methylamine [34] showed no resolved hyperfine coupling but rather broad lines having g-values close to that of the free electron. On the other hand, rubidium and cesium solutions in methylamine and potassium, rubidium and cesium in ethylamine and n-propyl- amine [35] showed a resolved hyperfine pattern due to coupling to the metal nuclei, in addition to a central line of the solvated electron. A recent NMR study [36] -3- of methylamine solutions of sodium through cesium showed no evidence of an NMR signal. Solutions of alkali metals in ethers received compara- tively little attention compared with amine solutions. Catterall [37] studied the EPR spectrum of saturated potassium solutions in tetrahydrofuran (THF). The spectrum consisted of two quartets and a central singlet due to coupling with the two naturally occurring isotopes of potassium and the isolated solvated electron respectively. A very recent investigation of alkali metal amide solutions by NMR [36] showed a sharp (8-15 Hz) signal in the case of sodium solutions; thus far no signals have been reported from other alkali metals. The sodium signal was attributed to the Na- anion based on the chemical shift results and no Na+ signal was observed. Solutions of alkali metals in hexamethylphosphoramide (HMPA) are very interesting and worth mentioning. Although the high dielectric constant of HMPA (29.6 at 298 K) insured that only a single EPR line was found [38], similar to those in metal-ammonia solutions, the presence of a metal- dependent absorption band in the visible region [39] established a strong link with amine and ether solutions. The 23Na NMR spectrum of Na-HMPA solution showed a single peak corresponding to Na- and again no Na+ peak could be detected [40]. Attempts to observe other alkali metal anions have been unsuccessful in alkali metal-HMPA solu- tions [36]. When alkali metal-HMPA solutions were quenched -9- to liquid nitrogen temperature, glassy blue solids were formed. EPR studies [41] of these quenched solutions showed different behavior from that of the unfrozen solutions. The existence of several discrete localized excess electron states similar to those in amines and ethers was observed. In summary, the results discussed above are under- standable in terms of the quantitative picture first proposed by Dye [35] and used by Catterall [41]. In alkali metal solutions a simple ion-pairing theory predicts the equilibrium + _ + - $3 0 Ms-i-es Ms es(Ms) (1.2) Two important factors govern the lifetime of the monomeric species Mg-eg. They are the electron-nuclear correlation time, TM, which is a function of the electron relaxation rate and the dielectric constant of the solution and the hyperfine splitting (hfs) constant A, which is a function of the electron spin density at the nucleus. These two factors dictate whether or not a resolved metal hyperfine coupling is observed in EPR. In solvents of high dielec- tric constant (NH3,HMPA), the average lifetime of Ms is short compared to the inverse of the hfs constant, i.e. TM°Aw<>1.and signals from both -10- paramagnetic centers are observed. In addition, the presence of electronic paramagnetism will give rise to an extremely efficient electron-nuclear relaxation mechanism, hence the nuclear spin relaxation will be so rapid in amine and ether solvents that the NMR lines are broadened beyond detection. Figure 1 shows the classification of EPR results in alkali metal solutions on the basis of the product of T -A [41]. M C. Metal-Ammonia Compounds The phase diagrams of metal-ammonia systems [1] have a very remarkable feature, that is, the presence of a deep eutectic which reflects the appearance of crystalline compounds upon freezing saturated solutions of Li, Ca, Ba, Sr, Eu and Yb. These compounds have the composition Li(NH3)4 and M(NH3)6, where M==Ca, Ba, Sr, Eu and Yb [42]. The metallic nature of these compounds arises from the loss of one, in case of Li(NH3)4, and two, in case of M(NH3)6, electrons to the conduction band [43]. The ammonia molecules act as space-filling diluents to increase the metal-metal separation relative to that of the pure metal. Therefore, these compounds have large ionic radii and hence low electron density. Thus, they were classified as low- electron density metals or "expanded metals". The alkaline earth hexaammines,M(NH3)6,are characterized by highly unusual structures [44]. They crystallize in a body-centered cubic (bcc) structure with the hexaammine -11.. Time - averaged TM) A '04 _— Na-HMPA spectra 10'3- —-Cs-AM l0“ _‘— Na-AM IO" __ k-EDA Strong __ k-ME A resonance Critical 1,—- Rb- EDA broadening “‘3 """ “*— Cs- EDA """"" “39”" ‘ :K- :2- FDA 1 '0' 4"— \"CS MEA ——-tK DG '02 _ K- EA -— Cs-EA . IO3- Resolved metal h.f.s. IO4J—_ K‘THF Figure 1: Classification of EPR spectra (~296 K) of fluid metal solutions on the basis of the product (TM)Aiso EA = ethylamine, THF = tetrahydrofuran, l , 2PDA = l , 2-propanediamine , EDA = ethylenediamine, AM = ammonia [41] . -12- complex located at each lattice site. The metal center is octahedrally coordinated to six nitrogen atoms of the six ammonia molecules. Sr(NH3)6 and Ba(NH3)6 undergo phase transitions near 40 K to rhombohedral and trigonal structures respectively, while Ca(NH3)6 showed no structural change down ‘to 4 K. Powder neutron diffraction studies [45] highlight the unique structural aspects of these compounds. The fully deuterated calcium compound, Ca(ND3)6 showed highly distorted, nearly planar ammonia geometry in which there are two inequivalent sets of deuterons. One of the three N-D distances is ~0.l A shorter than the normal distance in pure ND3 (1.00 A). The other two N-D distances are extremely long (~l.4 A). Also, the "pseudo-trigonal" axis of the ND3 molecule is not coincident with the metal-nitrogen bond but rather deflected by ~13°. Proton NMR studies of hexaammines [43] showed two linewidth narrowings in the range of 20-150 K. The unusual weak narrowing in the proton linewidths of Ca(NH3)6 in the range of 60-80 K and 70-100 K in the case of Ba(NH3)6 was observed previously for pure solid ammonia and was attributed to the transition from quantum tunnelling to thermally activated rotation of ammonia molecules [45]. Another strong- narrowing transition was observed for Ca(NI—I3)6 in the temperature range 100-130 K and, for Ba(NH3)6, above 140 K. This transition was interpreted in terms of rapid diffusion of thermally activated ammonia molecules. EPR studies of hexaammines [46,47] showed that a Dysonian line shape was -13- observed for Ca(NH3)6 and Sr(NH3)6, similar to concentrated alkali metal ammonia solutions, thus indicating metallic behavior. Although Ca(NH3)6 retained its line shape as the temperature was lowered, Sr(NH3)6 showed a transforma— tion to a Lorentzian line shape as the temperature decreased to 8.5 K. This reflects electron localization in the Sr(NH3)6 compound where lowering the temperature tends to freeze out the conduction electrons and creates a small concentration of localized moments. One the other hand, the Ba(NH3)6 compound showed no EPR signal. Magnetic susceptibility measurements of hexaammines [48] showed strong temperature dependent susceptibilities. The susceptibility of Ca(NH3)6 went through a maximum at about 10 K, a broad minimum at about 120 K and increased rapidly above 120 K. In all the temperature ranges the compound showed more paramagnetism than expected on the basis of the free electron model. In contast, Sr(NH3)6 and Ba(NH3)6 were essentially diamagnetic showing rapid reduction in the susceptibility as the temperature was lowered. The compound Lithium tetraammine, Li(NH3)4 crystallizes from saturated lithium—ammonia solution. It exists in at least two crystalline forms. The low temperature structure was first thought to be hexagonal. However, neutron diffrac- tion studies of Li(ND3)4 [49] showed that the compound has a body-centered cubic structure with a =14.80 A below 30 K, increasing to 15.03 A at 85 K. The results were the same for the compound Li(NH3)4 except for the presence of an additional -14- phase above 82 K that was not observed for Li(ND3)4. This phase is cubic with a==9.55 A. However, Sienko and Stacy [50] re-examined the structure of Li(NH3)4 and suggested that the phase which forms between 82 and 25 K is also cubic with =a_=14.93 A while below 25 K a superstructure with period 23 is formed. Magnetic susceptibility studies [51] of Li(NH3)4 supported the presence of three solid phases. The susceptibility at high temperature (above 82 K) showed a temperature independent paramagnetism, characteristic of a metallic system. Between 82 and 25 K, the susceptibility was lower than that of the phase forming above 82 K but increased with decreasing temperature as expected for Curie-Weiss type of paramagnetism, indicating electron localization. The maximum susceptibility was achieved at about 25 K and then decreased with the formation of another phase below 25 K indicating an antiferromagnetic interaction. The Li(ND3)4 compound showed similar behavior except that the transition at 82 K was absent. Clearly, metal ammonia compounds are very interesting systems which provide rich information about the nature of the non-metal to metal transition. An understanding of the compounds might help to understand and solve the "dilemma" of metal-ammonia solutions, and hence more studies are required. -15- D. Alkali Metal Anions (M’) Our knowledge of the chemistry of alkali metals tells us that they are exceptionally good electron donors and form the +1 oxidation state with an inert gas configuration. However, the existence of the -1 oxidation state in the gas phase was known for over 35 years [52]. Golden £5 £1. [53] were the first to propose that M- is a major species in alkali metal-ammonia solutions. However, there was no evidence for its existence, probably because of the fact that ammonia is "too good" a solvent so that it greatly enhances the dissociation of M- to the solvated cation and solvated electrons. In contrast, the presence of the metal- dependent band in the optical spectra of alkali metal-amine and ether solutions established the existence of M- in these solutions. However, the complete proof was provided when Dye gt al.[54,55] reported the fingerprint NMR spectra of Na-, Rb- and Cs' in solutions in amines and tetrahydro- furan. The 23Na NMR spectra of sodium solutions containing the added complexing agent cryptand 222 showed the presence of two sodium peaks. The chemical shifts of both peaks were remarkably solvent independent which was in contrast to that of the solvated sodium cation. The chemical shift of the low field (high frequency-less shielded) peak was the same as for sodium in Na+C222.I-. The other peak had a nearly identical chemical shift to that of the gaseous sodium anion and an extremely narrow linewidth (<3 Hz in THF),which -16- attested to the highly spherical and well shielded nature of the species. These two peaks were assigned to the sodium complex cation and the sodium anion. The absence of any solvent-induced paramagnetic shift in the case of the Na— peak indicated that the 2p orbital was well shielded from the interaction with the solvent by the completely filled 33 orbital. The Rb- and Cs- spectra in these three component systems were similar to that of Na-. Recently, it has been shown [56] that Na- can be formed in the presence of a reducible group such as carbonyl in sodium-diethylacetamide solution. The isolation of the crystalline compound Na+C222.Na- by Dye st 31. [57,58] provided the final proof of the existence of the alkali metal anion as a stable, long lived anion in the gas, fluid and solid phases. E. Trapped Electrons in Solids The phenomenon of "excess" electrons or "electron trapping" in solids is known to occur in both ordered (F-centers) and disordered (aqueous and organic glasses) systems. In crystalline materials [59]such as alkali halides and hydrides "excess" electrons can be brought into the system by irradiation of the crystal with high energy photons or particles, by electrolysis at high temperature, or by heating the salt in an alkali metal atmosphere. The electrons are trapped at anion vacancies; hence, a low -17- concentration of trapped electrons is expected. It was found that only 0.1% of the radiation-produced electrons are trapped in crystalline alkali halides. Optical spectra of F-centers [60] in alkali halide crystals showed broad bell shaped absorption bands. The positions of the bands were dependent on temperature, pressure and alkali halide used. Ivey [61] correlated the transition energy of the trapped electron in F-centers in salts with the NaCl structure to the interionic distancea_by means of a simple emperical equation known as the MBllow-Ivey relation: v = 17.6 a-l'84 m _ (1.3) where Vm is the frequency of the absorption maximum in eV. Equation (1.3) predicts a shift to higher energy (lower wavelength) with decreasing interionic distance. Indeed, the values of vm vary from 250 nm for LiF (smallest a) to 785 nm for CsI (largest a). The EPR of F—centers [59] showed two kinds of spectral lines mainly due to electron interaction with the nearby nuclei. The first type, which applies to most of the alkali halide F-centers, consists of a single, broad, Gaussian shaped line without any further structure. The linewidth varies from 47 to several hundred Gauss. The second type showed a resolved hfs with a discernible number of components on the order of 10 to 100. The g-values are slightly decreased from that of the free -18.. electron indicating a weak interaction of the electron with its surroundings and hence very small spin-orbit coupling to the lattice. The important feature of the EPR spectra was that they are independent of temperature, method of preparation and the concentration of the centers. However, further irradiation of F-centers leads to the formation of aggregate centers for which the EPR spectra were different. For example, irradiation of F-centers in KCl crystals with light in the F-band gave an EPR line which was narrower and less strongly saturated than the undisturbed F-centers. ENDOR studies of F-centers [62,63] gives a great deal of information about the electron nuclear interaction through the dependence of the spectral patterns upon the crystal orientation and the further quadrupole splitting of the lines. Therefore, the hfs constants to nuclei in the first and higher order coordination shells, the elements and axes of the anistropic hfs and the quadrupolar interaction tensor could be determined. In contrast to the F-center, where the electrons are trapped in pre-existing traps, aqueous and organic glasses provide very rich media for trapping high concentrations of electrons at low temperatures [64]. The discovery of electron trapping in water as a transient optical spectrum in the visible by pulse radiolysis [65] opened the door to a new field of study of electron trapping in polar and non-polar matrix glasses. In a similar manner to F-centers, electrons can be trapped in aqueous and organic glasses by exposing -19- the latter to high energy ionizing radiation at low tempera- tures [66]. The change in the Optical and magnetic resonance properties with time suggests that the process of electron trapping in matrix glasses can occur in two stages [67]. The first stage involves the localization of the "pre- solvated" electron which depends upon the relative energies of the conduction electron level of the medium (V0) and the total ground state energy of the localized electron relative to vacuum (Et)' Kevan [68] suggested that the criterion of electron localization in glassy matrices is that Et'HO mo mammnuoam map How msumummmd “v musmflm -40- the Teflon tube and broken under dynamic vacuum. The broken metal tube(s) was then moved to the constriction E and a vacuum seal-Off was made at F. The meta1(s) was distilled into chamber D and the metal side-arm A was sealed off at E. Since lithium metal cannot be distilled, the apparatus was slightly modified to include another side-arm attached to the chamber D. The side-arm had a coarse frit to separate lithium metal from the other metal in chamber D. At a pressure of less than 10-5 torr., an amount of solvent (reaction solvent) [(210 mls/mmole complexant)] was distilled into the chamber C to dissolve the complexant. The apparatus was removed from the vacuum line and immersed in a dry ice-isopropanol bath at~-40°C. The complexant solution was transferred through a coarse frit G into the metal-containing compartment D. After metal dissolution, a partial evaporation of the reaction solvent was followed by the distillation into D of another solvent(s) (crystallization solvent(s)). The solution was filtered through the frit G into C and cooled to dry ice (-78°C) or acetone-liquid nitrogen (~-92°C) temperatures to precipitate the crystals. When the precipitate had settled, the supernatant liquid was poured into D, frozen with liquid nitrogen and removed by making a vacuum seal- 'off at I. A washing solvent was then distilled into C and the slurry of liquid and the crystals was poured into the side-arm J. The supernatant liquid was repeatedly poured -41- into C and distilled back to J to wash the crystals free of excess complexant. Finally, the solution at C was frozen with liquid nitrogen, the crystals were pumped to ~10-5 torr _and a seal-Off was made at K. The dry crystals were poured into the sample tubes L that were immersed in liquid nitrogen so that flame-seal-offs could be made. The sample tubes were kept in a freezer at -78°C until needed. D. Analysis Because of the tendency of crystalline electrides and alkalides to decompose thermally, elemental analysis was not practical. Instead, a "home-built" analytical scheme was constructed based upon the decomposition reaction Of the crystalline compounds with water in a closed vacuum system [100]. The decomposition reactions of electrides and alkalides can be written as follows: M+Ln.N' +2H20 —+ M+ +N+ +nL +20H" +H2+ ---+ (11.1) alkalide O ———+ M+-+nL-+OH--+LII + ---+ (11.2) + - M Ln.e -+H2 2 2 electride The amount of hydrogen evolved, pH titration of the solution, flame emission to detect the metal content -42.. and quantitative proton NMR combine to give the stoichio- metry of the compound. The experimental techniques used with this analytical scheme will be discussed below. 1. H2 evolution A scribed sample tube was first cleaned by acetone and weighed quickly enough, at room temperature, to prevent sample thermal decomposition. In a glove bag that had been purged several times with dry nitrogen gas, the sample tube was carefully broken and dropped into a preevacuated H2 evolution apparatus. The apparatus used is shown in Figure 5b. It has a large surface area to allow the crystals to spread out. The apparatus was removed while cold from the glove bag and connected to the vacuum line of the hydrogen collection apparatus [100]. The entire system was evacuated to~lO-5 torr and conductance water, which had been degassed three times through freeze-pump-thaw cycles, was condensed onto the sample at low temperatures. The sample was then allowed to react with water very slowly while it was cold to prevent any thermal decomposition. The hydrogen evolved was pumped manually by a Toepler pump through two liquid nitrogen traps and collected in a known volume pipet. The cycles with a mercury leveling bulb were repeated a number of times until two consecutive readings of the mercury height were the same. The atmospheric pressure -43- Cajun Ultra-Torr Union Graded Seal Figure 5: Apparatuses for the optical spectra of thin films (5a) and hydrogen evolution (5b). -44- and temperature at the pipet were measured and the number of moles of hydrogen was calculated using the ideal gas laws. 2.ng Titration During hydrogen gas collection water was condensed into the liquid nitrogen traps which left a white-dry residue in the analysis vessel. Under an inert atmosphere, the apparatus was opened and a known volume of standardized HCl solution was added in excess. The solution was then transferred into another container and divided into three portions for pH titration, flame emission and 1H NMR. A known volume of the first portion was titrated with a standarized NaOH solution by using a pH electrode (Corning, catalog number 476050) and a digital pH meter (Orion Research model 701A) which had been calibrated with buffer solutions. The titration buret, which contained NaOH solution, had a glass sheath which allowed a continuous flow of dry nitrogen gas to prevent CO2 absorption by the base. The number of gram equivalents Of NaOH at the end point is equivalent to that of unreacted HCl from.which the number of moles of base in the residue could be evaluated. The broken sample tube was cleaned, dried and weighed so that the mass of the sample could be determined. -45- 3. Flame emission A known volume of the second portion of solution was diluted to a predetermined volume in order to adjust the concentration on the basis of the expected concentration (ppm) Of the metal ion(s) in the sample. A series of standard solutions of the appropriate metal(s) were prepared by dilution of 1000 ppm standard solutions (Aldrich Chemical Company, Inc.). The emission intensity for the alkali metal atoms was read from a digital averager that was connected to the flame emission instru- ment (Jarrel Ash). The instrument was adjusted for maximum emission at the wavelength of the corresponding element. The emission from conductance water was measured between every reading to give the background or noise level. A calibration curve was constructed by plotting the emission values for the standard solutions against the concentration in ppm. The amount of the metal(s) in the sample could then be determined from its emission value and the calibration curve. 4. Quantitative 1H NMR: 1 The number of moles of the complexant was determined by quantitative proton NMR. Samples for NMR measurements were treated in two different ways. First, the third portion of the acidic solution was neutralized with NaOH and allowed to evaporate to dryness in a partially -46- evacuated desiccator with "Drierite" as a drying agent. The white residue was pumped to insure complete dryness and dissolved in D20. Second, a new weighed sample tube was used and the sample was allowed to react with 020. This method was also used to determine the solvent content, if any, in the crystals. For both samples, a known weight of sodium acetate was added as an internal standard to give approximately a 1:1 mole ratio between the complexant and sodium acetate protons. 1H NMR spectra were recorded on a Bruker 250 MHz Fourier transform NMR instrument. In the case of 18C6, the solution gave two NMR peaks corresponding to the CHZ-protons Of the crown ether and to CH3-protons of the acetate. The areas under the peaks were determined by fitting the line to a Lorentzian line shape with a line fitting program provided by Bruker that gives the linewidth at the half height and the peak amplitude [101]. In the case of cryptand 222, which has multiple peaks for the cryptand protons for which the line fitting routine cannot be used, the areas were determined by the integration procedure. E. Sample Handling and Instrumental Techniques Due to the sample sensitivity to temperature, moisture and oxygen all sample handling was carried out in an inert atmosphere glove-bag at low temperatures. -47- 1. Optical absorption spectra The apparatus used in the Optical absorption studies is shown in Figure Sat It was made of fused silica glass to avoid sodium contamination from Pyrex and consisted of a quartz Optical cell, a side-arm reservoir and Kontes high vacuum valve. The clean apparatus was first evacuated to insure complete dryness and brought into a glove-bag that had been flushed several times with dry nitrogen gas. The sample tube was broken in the glove- bag while it was cold and a few crystals were dropped into the bottom of the optical cell. The apparatus was removed from the glove-bag and connected to the vacuum line while the optical cell was immersed in a dry ice-isopropanol bath at'v-40°C. After attaining a pressure of less than 10.5 torr , methylamine or dimethylether was distilled into the Optical cell. The apparatus was removed from the vacuum line and the crytals were shaken until completely dissolved. A portion of the solution was transferred into the side-arm reservoir and the solution in the Optical cell was diluted by distilling some solvent into the cell from the reservoir. A thin solvent-free film was made by rapidly shaking the solution in the Optical cell at ~-40°C while the bulk frozen solution in the reservoir was kept at liquid nitrogen temperatures. This flash solvent evaporation caused film formation by the splashing of the solution onto the cell walls. Because of the nature of this method the film prepared was often not -43- uniform and it was necessary sometimes to repeat the process to get a film of the proper thickness. The Optical spectra were recorded on a double beam spectrophotometer (Beckman DK-2) which was modified to permit temperature control of the sample compartment. The temperature of the sample compartment was measured with a copper-constantan thermocouple placed near the cell and connected to a digital temperature readout (Doric model 08-350). The spectra were recorded from 2500 nm (4,000 cm-l) to 400 nm (25,000 cm-l) and normalized to unit amplitude after subtracting the base line. The base line was obtained by determining the spectrum of an empty cell. 2. Pressed powder conductivity Conductivity measurements Of electrides and alkalides were made by using the apparatus designed and described by Michael R. Yemen [102]. A glove-bag around the apparatus was purged several times with dry nitrogen. The sample tube was broken inside the glove-bag and the crystals were loaded between two stainless steel electrodes into a heavy walled 2 mm I.D. fused silica tube. The sample was compressed by means Of a steel spring whose force constant had been measured. Ohm's law behavior was first checked by measuring the current at various voltages, then the current through the sample was measured at constant voltage as a function of temperature. -49- 3. Solid state NMR High resolution alkali metal NMR spectra for crystal- line electrides and alkalides were obtained by using the magic angle sample spinning (MASS) technique. The measurements were done in the analytical laboratory of Dow Chemical Company (Midland, Michigan) and in the National Science Foundation Regional NMR Center (University of Illinois, School of Chemistry). At Dow, a Bruker CXP high power pulse spectrometer was used. It is equipped with an Aspect 2000 computer system. Oxford Instruments superconducting magnets of 4.6975 Tesla (200 MHz) and 8.4558 Tesla (360 MHz) field strength were used to Obtain the spectra. At the University of Illinois, spectra were recorded on a "home-built" Fourier transform NMR spectro- meter [103] with a 11.7440 Tesla (500 MHz), 3.5-inch bore solenoid (Oxford) and a Nicolet 1280/2933 data system with a control Data Corporation disk. In both measurements a standard multinuclear Andrew-Beam MASS-NMR probe was used. The spinners were made of either Delrin or d8 polymethylmethacrylate and were spun at the "magic angle" with a spinning rate of 2-4 KHz. The "magic angle" was set using 79Br in a sample of KBr. The spinning gas was cooled by passing it through a coil immersed in an acetone-dry ice bath. This gave a sample temperature between -12 and -l6°C. The spinners were loosely filled with the crystals in an inert atmosphere glove-bag. The -50- chemical shifts were measured with respect to the infinitely diluted aqueous alkali metal cations. Upfield (diamagnetic) shifts were negative. 4. EPR spectra Samples for EPR measurements were loaded in 2-4 mm O.D. "spectrosil" fused silica glass tubes connected to Kontes high vacuum valves. The sample tubes were broken in an inert atmosphere glove-bag and a few crystals were loaded into the spinner while they were cold. The tubes were evacuated to less than 10-5 torr and a vacuum seal-Off was made while the sample was immersed in liquid nitrogen. X-band EPR spectra were recorded on a Bruker model 200 EPR spectrometer. Above 100 K a Varian model 4341 variable temperature controller was used and the temperature at the sample was calibrated with a copper-Constantan thermocouple and a Doric model DS-350 digital readout system. Temperatures below 100 K were provided by a continuous flow-helium system (Oxford Instruments Model ESR9) and measured with a thermocouple (Au +0.03% Fe/ chromel) placed just below the sample. EPR modulation frequencies of 12.5 and 100 KHz were used with amplitudes low enough to avoid distortion of the lineshapes. Power studies were made to insure the absence of saturation effects. The g-values were evaluated by measuring the magnetic field at the microwave frequency. An NMR Gauss meter (Model ER035M) with l mG resolution was used to -51- measure the magnetic field while the microwave frequency was measured to 10.01 MHz with a frequency counter (Hewlett-Packard Model 5245L). The accuracy of this system was checked with a diphenylpicrylhydrazyl (DPPH) sample. 5. Magnetic susceptiblity Magnetic susceptibilities were measured with an S.H.E. computer-controlled variable temperature superconducting quantum interference device (SQUID) spectrometer capable Of measurements at temperatures between 1.7 and 400 K. The crystalline samples were loaded in an inert atmosphere glove-bag into small cylindrical containers (inside dimensions 8.5 mm by 6.5 mm diameter) made Of either an aluminum silicon alloy or Delrin. Prior to loading the sample, a thread 15-20 cm long was attached to the bucket through four holes. The sample filled bucket was removed from the glove-bag and transferred under liquid nitrogen to another glove-bag placed around the SQUID's airlock. The bucket was loaded into the airlock by attaching the thread to the hook of the driving tape. The airlock was evacuated and pressurized with helium three times, then the sample was loaded into the SQUID. The sample was centered between the two measuring coils by examining the analog voltage from the SQUID output. At every tempera- ture, the SQUID was allowed to take ten readings and an average value was printed out. Readings of 10% difference -52- at the same temperature were rejected and the measurements were repeated. The computer printout, P, is the total magnetic moment in e.m.u. divided by the field in Gauss and by the total scale factor (which is the scale of the SQUID control multiplied by the scale of the sample measurement control). To correct for the diamagnetism of the sample and the bucket, the sample was ejected from the SQUID and allowed to decompose at room temperature in the airlock. The electronic contribution to the susceptibility was calculated from the equation e = P(sample+bucket)'-P(decomposed sample+bucket) XM number of molés In the sampIe (II'B) F. Model Salts Conventional salts in which the alkali metal cations are complexed by crown ethers or cryptands were prepared by procedures similar to those of Pedersen [74,77] and Weiss [76]. The salts, together with the stoichiometric amounts of complexant (mole ratio 1:1 or 1:2), were dissolved in either hot methanol or l-propanol, the solutions were filtered, concentrated by evaporation and allowed to cool until crystals formed. Conventional salts without complexants were reagent grade and were used without further purification. CHAPTER III SYNTHESIS AND CHARACTERIZATION A. Introduction The most difficult problem encountered in the synthe- sis Of crystalline alkalides and powdered electrides was solution decomposition. In general, alkali metal solutions in amine and ether solvents are thermodynamically unstable with respect to solvent reduction. This results in the disappearance of the characteristic blue color Of the solution even in the presence of macrocyclic- polyether complexing agent. The decomposition reaction in an amine solvent can be described as, e; +RNII2 -——+ RNH' +8H2+ (111.1) The nature of the decomposition reaction in an ethereal solvent is unknown. Kinetically stable metal-complexant solutions can be prepared if the solvent, metal, complexant and glassware are appropriately treated to remove all easily reducible impurities. Also, the rate Of such a decomposition reaction can be greatly reduced to a negligible value by lowering the temperature. -53.. -54- In general, three procedures can be followed to synthesize solid electrides and alkalides [30]. These are solvent evaporation, direct vapor deposition and crystal- lization from solution. Only the last procedure was studied in the present work. Three methods of crystalliza- tion were used and will be discussed here. The classifica- tion of these three methods is based upon the kind of solvents used and/or the added stabilizing agent. B. Methods of Synthesis from Solution 1. Method 1: Prior to the present work, crystalline alkalides were prepared [94] by either using ammonia as a reaction solvent, with its great ability to dissolve the alkali metals and the complexant or methylamine to dissolve the complexant and then the metals with agitation at reduced temperatures. In some cases, the metal and complexant were dissolved simultaneously. The crystal- lization solvent(s) was 2-aminopropane and diethylether or n-pentane. This method has been slightly modified as follows: ammonia was eliminated as a reaction solvent to avoid the formation of bronze-colored metallic drops which form during solvent evaporation and which are hard to dissolve and often resulted in metal precipitation. Instead, methylamine was used, since the solubility of 18-crown-6 in it is high enough to give a clear solution -55- before pouring onto the metal film. In the case of cryptands 222 and 211, simultaneous dissolution of metal(s) and complexant was achieved by moving the solution back and forth between the metal and complexant compartments. The crystallization solvent always included some methylamine, i.e. the reaction solvent was never distilled to dryness or near dryness. Finally, the washing solvent was diethylether since crown ethers and cryptands have high solubilities in it. This method was found to be very useful for the preparation of crystalline alkalides but not for electrides since rapid decomposition occurred upon adding 2-aminopropane. 2. Method 2: Dheeb Issa [98] attempted to prepare the crystalline compound, C518C6Li, which might have been analogous to the crystalline sodide Cs+18C6.Na-, by dissolving equi- molar amounts Of Cs, Li and 18-crown-6 in methylamine. He found that the solution was much more stable than one containing only cesium and dark blue crystals were formed upon adding a mixture of 2-aminopropane and diethylether to a concentrated solution at reduced temperatures (~-78°C). Surprisingly, the crystals contained very little lithium as indicated from the analysis [104] which showed a 1:1 stoichiometry of Cs to 18-crown-6. This method was generalized to involve adding lithium metal in small amounts as a stabilizing -56- agent when using methylamine as a reaction solvent, a mixture of methylamine, 2-aminopropane and diethylether as a crystallization solvent and diethylether as a washing solvent. Solutions of lithium metal in amine and ether solvents contain only the solvated lithium cation and the solvated electron since there is no evidence for the formation of lithide ion (Li-) in these solutions. Also, the cation- electron interaction is very weak as indicated from Optical absorption and EPR studies [31,32]. Since the lithium cation size is too small to fit the lB-crown-6 or cryptand 222 cavity, complexation of lithium by 18-crown-6 or cryptand 222 is unlikely to occur. Also, Dye pp 31. [30] by using a modified Born-Haber cycle, estimated that the lithide compounds, M+Ln.Li- are thermodynamically unfavorable compared to the corresponding electride M+Ln.e- and lithium metal. This was confirmed by the absence of lithium in all of the systems studied. The simple explanation for solution stabilization in the presence of dissolved lithium metal is that lithium is acting as a scavenger to inhibit autocatalytic decomposi- tion processes, probably by reaction with free radicals or anion radicals which take part in a chain decomposition reaction. Perhaps lithium reacts with R' to form lithium alkyl [30] which is less reactive than lithium metal. -57- 3. Method 3: Professor J.L. Dye [30], during his sabbatical year (1983-84) at A.T.T. Bell Laboratories, examined the use of non B-hydrogen solvents, such as dimethylether and tri- methylamine, for the synthesis of alkalides and electrides. He found that such solvents greatly stabilized solutions of alkali metals with complexants and that much more stable crystalline alkalides could be prepared relative to those prepared by methods 1 or 2. Accordingly, this method has been tested and generalized to be the "best" method to date for the preparation of crystalline electrides and alkalides from solution. In this method, dimethylether is used as a reaction solvent while a mixture of dimethyl- ether and trimethylamine is used as the crystallization solvent and trimethylamine as a washing solvent. The ' metal-complexant solutions in dimethylether were found to be stable for days at temperatues as high as 0 to -10°C and for more than six months at -78°C. Some of the crystals formed were found to be much more stable than those prepared by methods 1 or 2. However, the rigorous purification procedures for metal, complexants and solvents are still required since the electride and alkalide salts are intrinsically unstable even in the absence of solvent. -53- C. Results and Discussion 1. Synthesis Table I describes the synthesis of crystalline elec- trides and alkalides by the three methods mentioned before. For completeness, the last column is added to give the final assignments since some crystals precipitated from solutions Of different stoichiometry tend to be the same. Each of the crystals described here has a metallic appearance and the fact that some of the crystals (e.g. Rb+18C6.Na-) showed different colors when precipi- tated from different solutions can probably be explained in terms Of different illumination conditions and/or different degrees of roughness of the crystal surfaces. Three categories of the crystals have been synthesized. during the course of study of the present work. First, five new crystalline compounds were synthesized for the first time. These are RbNa18C6, Rb218C6, Rb18C6, Cs(18C6)2 and CsC222. Second, the attempted syntheses of six compounds were unsuccessful either because they don't exist or because of the precipitation of another more thermodynamically stable compound from the solution. These systems are Na218C6, NaC222, Rb(18C6)2, C5218C6, CsZC222 and CsNaC222. Third, crystalline compounds that had been previously synthesized and characterized in our laboratories, were prepared for further investigation. -59- .mooowucoo H OHQMB -o.-mo+no Io.-~o+uo -oz.-~o+oz mUmH+.HMUOE mz noz.HH~O+eq noz.~mmo+oz no.-~o+uo noz.bOm-+sm manuscmwnmd Hosea A .Hmume m0 m0 meHO+.mHoummuo oomanm wapomc .OOH xnoo .OHOuMHOQEOu EOOH up waucoaofl> OmOdEOOOOII Doom: on do maooumllmaoumwuo common macaw: .pou xuao .mzop HON ousuoummfiou Eoou no o-bmumllmamumwuo Hmcomoxmn cop-ow .musuxfle muwn3 poo ammo .Hson ca uoooa How ouounuOQEOu Eoou um anmumllmamummuo beam ODHQIHO>Hflm .mwao MOM OHouauOdEOu Soon as magnumllmaoummuo coo-ow .Uoomn on do manaumllmoapwo: OOHOHOOIONcOHo xnao .mcflunos com: OmomEOOOU cons peso-H moan O>flm Op u-oEIImuson aouo>on “Om monououOdEOu Soon on manaumllmmaoooc pone-OOIONcOHm :Oaudfluommo HoumNuo mauvesm m0 NNNUHAmU NNNUHAMZ N Domfimz IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII N mUmH oz HHNUOZHA NNNUosz N NNNU m0 oUmHmznm H QOmBmE IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII whomeOAoOfloum cowuoaom .mOpHHoxH< poo mopwquOHm mowaaoummuo mo coflumwuonmo poo mwmonuoxm «H manna -60- noz.-~o+nm Id2.oOmH+s noz.-bOmH-+mo -oz.-oOmH-+uo mom-am oUmHnm -uON-GOmH-+no -o.-GOmHo+no (mum-bumao+no mucoacmflmmd Hosea .mnooo Haum>mm MOM mus» IoHOQEOu Soon on manoumilmamummno uaam OOROHOOIHOQQOU .muson Hono>om you ousuonwmfimu Soon no Oanoumllmaoumano OOH xuma .maop How OHoHOHOQEOv EOOH um manmumllmamumwuo umaw OOHOHOOIHOQQOO xuoo .m>co How nmuouoHOQEou EOOH up magnumllmaoumhuo moan pouc-oouuomdoo xnmo .uco>aom mzflnmoz may ca O>H0mmwp wads-puma not» maoum>uo oufl531.moaomoc OOHOHOOIONcoum .Hoos on moons How ououoummsou Soon on magnumllmoapmoc OOHOHOOIONooum .Hauoa mul.maaum>HO moan uoaofl> Ou moan xnmo .nuoon Hauo>mm MOM ououaH0daou Soon on maomumllmaounauo oomanmloou xoaam .nson mannimco uoooo MOM mnououOmEOu Econ no Odouumllmaaumhuo uoam uOHOa> Ou OOHQ xumo conmqumOo asymmmu m QOSBMZ IIIIIIIIIIIIIIIIIIIIIIII .moocwucou H canoe Namoozom mam-oz- N-oOmH-ozuo bOmHoZuo m-oOs-eesom wUmeanm oOmHaqmno m-bOmHvesno mUmHHQmU huncEOA:Ofioum oofluoomm .oosssusoo H manna -51- nom.bOmH+sm oOmH1.oOmsnm oUmHnm -o.-oOmH-+no (e.m-oOmH-+no + -nON-bOmH-+no uo.-~o+uo noz.-~o+x mucoeomammd Hosea .mNMp manna How ououmuomEOu Econ um magnumllmaaumano OOHOHOOIHOQQOO :mwcmmuw .HOOBOQ muwn3+.maoummuo contamIOHpooo mo-m .Hso: manslmoo uoono How onsumummfiou Soon on oaonumllmaoumauo padonmioapooo moan .mnsos HnHO>Om How ouou unnamed» Econ us manounllmaoumwuo oomonnloou xooam .musoo H0HO>Om How ousuouomEOu Eoou um magnumllmaaumzno Rom-o pom can no Ousuxwz .OnsuouomEOu EOOH um wanna-Oa> mmOmEOOOOIIUoOHI on do manoumllmaoumxuo pmmannlmapooc OOH xumo .musoo Hauo>om How ousuauoofiou Econ an maooumllnoaoooc OOHOHOOIONCOHQ nmwcoouu :Ofiumwnomoo Hoummuu N wUmH am m-GOm-Vom messes m-oOmHVuo mOmHmU NNNUmU NNNUOZM quoeofi£0aoum seaboaom .oossauooo H wands -52- These compounds are KNa18C6, CsNa18C6, KNaC222, RbNaC222, LiNaCle, Rb2C222 [94] and C518C6 [104]. In general, method 3 proved to be the "best" method for synthesizing crystalline electride and alkalide salts from solution. This method has many advantages over the other two methods that can be summarized as follows: 1) The high solubilities of l8-crown-6 and cryptand 222 in dimethylether make solution preparation easy; 2) The insolubilities of alkali metals in dimethylether ensure that the only way that solubility can be achieved is through complexation; it avoids the presence of excess uncomplexed metal cations in solutions; 3) The solutions have excellent stability which allows us to use somewhat elevated temperatures (~0° to +10°C) when dissolving the metals; 4) The high vapor pressure Of dimethylether allows it to be condensed into or removed from the solution very easily; and 5) The low dielectric constant of trimethyl- amine tends to increase the yield of crystals. 2 . Analysis The results Of the analysis Of the five new crystal- line compounds and the crystals of the system CsNa18C6 are listed in Table II). It is clear that for the five new compounds the stoichiometries of the crystals are the same as the solution stoichiometry except for crystals prepared in the presence of lithium metal, in which case, the crystals were almost free of lithium. -63- It is worthwhile mentioning the reasons for the synthesis of some of the new compounds and the unsuccessful attempts to synthesize the six other compounds since this may shed some light on problems faced in the past. The dark blue crystals of stoichiometry C518C6 [104] could be either of the two compounds, the electride Cs+18C6.e_, or the alkalide Cs+(l8C6)2Cs-. If this substance is the electride, i.e., if the Cs+ cation tends to form a 1:1 complex with lB-crown-6, then perhaps the corresponding ceside Cs+18C6.Cs-, analogous to Cs+18C6.Na-, should exist. On the other hand, if it is an alkalide, i.e., if the Cs+ cation tends to form a 1:2 complex with 18-crown-6, then the corresponding electride Cs+(18C6)2.e- should exist. Indeed, black shiny crystals were precipitated from a 2-aminopropane-diethylether solution with the solution stoichiometry of Cs(18C6)2 in the presence of dissolved lithium metal. Also the same crystals could be obtained from dimethylether-trimethyl- amine solutions in the absence of any lithium. However, solutions Of stoichiometry C3218C6 in the presence Of lithium metal yield only dark blue crystals and cesium metal when precipitated. The results indicated that the Cs+ cation is more likely to form 1:2 complex compounds with 18-crown-6 when precipitated out of solution. If this is true, then the compound C318C6 should be a ceside Cs+(18C6)2.Cs- rather than the electride Cs+1806.e- as initially assigned. This conclusion can be confirmed by -64- .Omom momonHOm co HH manna ou mouoz -H.m+- -oo Hhmé mama.- m -m.me+o -o- oz -oOmH-mo III Hem.m III. III III III osn.m ozoomsmo -m.o+- -o.ou- -o. .l... 935 Ill ..II II. mooJ MNON HANNNOmO -n.HHu- -o.H+-v -o.m+- -m.m+- -v.o+- -o- . mmmOmO Nmm.~ I vo.ov NmmN «2.4.. Sum..- moo.~ flqmmmumu m Ao.mn- -o.mn- -v.mu- -m.ouo -o- m Sow-emu II mm-é l oHHN mHHd mmoé N>H.N Qua-mo m -m.oHu- -n.mn- -o- m Some-mu ..II hmoN II II. III mhoé m>N.N «A Some-mo m -o.H+V -v.ou- -o.mu- AH.man -o- m -oomaeno II: (I: oo.o mos.m mmo.m mmo.e mem.~ as -oomH-no Ao.vu- An.H- -o.o+- -m.o- .o- mom-“nu ll. Nmo.m I..| Hafim mmm.m mamé mmm.m bum-em m -m.mu- -v.ou- -.mnv -m.on- -o- m oom- nm ll. mmaé III Hmmé ommo Hmm.m 2N5 @Uma nm N 3.3.; 3.07.- RNIV 3.7.- 8- N bum.- nm II mvoN II Nvmé Nmbé Nva thN bum: em -m.o+- xo.m+- -.m+- -m.mu- Am.m+- -o- woman-3m I Hmm . H mam . .- vmm .H mom . m 0mm .H Omm . .- ozoumanm muOsOaso-Moum :Ofluouufla «:2 m- z 2 noun-95.3. ceauoozoo v OTmOHO-s 2.:42 Houmsno as at so on so so we owem awuosoasoeoum . n ucmxoa Sou ._.. «Em O H.“- Odds-om m.u:muomwm mumwnhaon< may mo muHOmOm "HH manna -55- .ozN-mUmH-mo auuofiownomoum ocoOQEOO ooh com: canon :OHuoHsOHaUn .wupOEOH:O«Oun poEfimoum 0:» Boom :Ofluow>noo ucoonom ucOnOHdOH womanhoouom «0 Hanan: mafia: ooax mOHOE no Hones: on» out co>wm mosaa> .N can H.HH mcowuooon com: panama .HH manna 0» nouoz -55- some other results which have been obtained in our labora- tory. The compounds Cs+(l8C6)2.K- [105] and Cs+(18C6)2Rb- [106] were precipitated from amine and ether solutions. In summary, solutions Of cesium metal and other alkali metals precipitated the compounds Cs+(18C6)é.e-, Cs+(18C6)2.K-, Cs+(18C6)2.Rb- and Cs+(18C6)2.Cs-. The compound which is not consistent with this result is CsNa18C6. Accordingly the analysis Of this compound was repeated and it was found that the 18C6 content is more consistent with the stoichiometry Cs(18C6)2.Na rather than C318C6Na. For the rubidium compounds, the same kind of procedure was followed after crystals of stoichiometry Rbl8C6 had been synthesized but the results were quite different. Three trials to prepare Rb(18C6)2 resulted only in the precipitation of white crystals of 18C6 together with shiny, bronze-colored needles of Rb18C6. On the other hand, solutions of reactant stoichiometry Rb218C yield greenish-coppery crystals which have the same stoichiometry as the solution. Adding to this, crystals of stoichiometry RbNa18C6 are more likely to be a sodide than a rubidide, since solutions of stochio- metry Na218C6 yield crystals only in the presence of solvent, but form a white-gray powder, probably of sodium metal and 18-crown-6, when the solvent is removed. This indicates that compounds which have the Na+ cation complexed with 18-crown-6 cannot be formed in the absence of solvent. These results suggest that, in -67- contrast to the Cs+ cation, the Rb+ cation forms 1:1 complexes with 18-crown-6, when precipitated from amine and ether solutions. Although the 1:1 rubidium and cesium thiocyanate complexes with lB-crown-6 have similar crystal structures [86], the 1:1 complex for rubidium and the 1:2 complex for cesium appear to be the most thermodynamically stable forms regardless of the solution stoichiometry. Dye g5 31. [94] reported the.formation of two crystalline compounds from 2-aminopropane solutions with reactant stoichiometries C32C222 and CsNaC222 when they used ammonia as the reaction solvent. The analysis of these two systems was poor, which made it difficult to assign the compound stoichiometries. It was of great interest to compare the properties of these two compounds to the new compound of stoichiometry CsC222. Out of 6 attempts to reproduce the synthesis of CsZC222 by the previous method only one was successful in yielding crystals. However, when lithium metal was used to stabilize the solution (method 2) the same kinds of crystals were easily prepared but they were mixed with the cesium metal. When a solution of stoichiometry CsNaC222 (method 1) was allowed to cool, golden colored crystals of Na+C222.Na- were formed. On the other hand, when crystals of stoichiometry CsC222 were dissolved in dimethylether and were allowed to react with a sodium film followed by adding trimethylamine and then cooling -53- to dry ice temperature, mixture of two crystals (CsC222 and Na+C222.Na-) were formed as indicated from solid state 23Na and 133Cs NMR. These results suggest that the packing of the complex cation Cs+C222 does not leave holes large enough to be occupied by alkali metal anions or else the compound Cs+C222.e- is thermodynamically more stable than the corresponding alkalide CS+C222.M-. It should be mentioned at this point that the formula Cs+(C222)2.Cs- for crystals with stoichiometry CsC222 is rejected since it is known that only 1:1 complexes are formed between alkali metals and cryptand 222 [82]. Accordingly, the stoichiometry CsC222 implies that the compound should be an electride CS+C222.e-. 3. Optical Spectra The Optical absorption spectra of thin, solvent-free films have been obtained by dissolving the crystals in methylamine or dimethylether followed by fast solvent evaporation into a Side-arm that was kept in liquid nitrogen to avoid any solvent effects on the spectra. Because of the nature of the film preparation the spectra were Often artificially broadened. The optical Spectra of films made from crystalline compounds of stoichiometries RbNa18C6, RbNaC222, Rbl8C6, Rb 18C6, Cs(18C6)2 and CsC222 2 will be discussed below. -59- a. RbNa18C6 and RbNaC222 The Optical absorption Spectra of films made by evaporating methylamine from solutions of the two compounds are shown in Figure 6. Both compounds Show essentially the same spectrum with an absorption maximum at 13,800 cm.1 and a pronounced shoulder on the high energy side, at 19,500 cm"1 for RbNa18C6 and 19,900 cm"1 for RbNaC222. These features are similar to those observed with films produced from solutions of Na+C222.Na- except that in the latter case the band position is at 15,400 cm-l. If the major band is due to Na- then Rb+18C6 and Rb+C222 cause a red 1 shift of 1600 cm- from the position of the Na- band in Na+C222.Na-. If, on the other hand, the absorptions are due to Rb‘, then a blue shift of 2200 cm"1 from the absorption Of Rb- in Rb+C222.Rb- has occurred. Because of this uncertainty, the optical spectra do not permit us to determine whether these two compounds are sodides, rubidides or mixed systems. b. Rb18C6 and Rb218C6 Figure 7 shows the optical absorption spectra Of films of the two crystalline compounds of stoichiometries Rb18C6 and Rb218C6. Films of Rb18C6 made from methyl- amine solutions showed two peaks at 11,600 and 8,900 cm'l, presumably due to Rb- and trapped electrons (eg) respectively. Films of Rb218C6 showed different behavior depending upon the solvent used to make the film and the ‘70- .-.....V mmmooznm coo - nnnnn o mumeozom mo useem mo ouuooam Hooeuao ON 0. n9 . -..Eo V m.- "b doomed II nnnnnnnnnn . .6. ...... OI .0 226.4... 31.4)... 0.0 XDUJv/V 1 00.. mes new 0- AVN -71- N .EHHM Roan» -.....- .senm secs -II.II.||- meme om oco - ..... - oceans mo mcowusaom Tomas-anuoa Scum TOME madam mo muuoomm amoeumo 70. 37.53% ON 0. O_ m . . - _ - . w ....... .. . .... bee-m -o .M. I . \ TIP/bill. \x ”.I/ ... \\\ L No 1%... . t .//..z../ ‘- s /...z. e - (v - . 4..., 1 .- .. so .7..//.x -\ ......- .. /J ...\ ”.1 \ ...x. //.z s 1 mo l \ .....\ .. 12,! e \ .. xH/ \ x I, ./. \ x 4 m C x / l. \ .\\ I /.I.l.nl ...uv /. \m.\ .lf... \\H..H.L.Hflu..u.\....\ . ....S... w.\ 1 oo- [ b p _ mdv NO 0. AVN "h musmflm XDUJv/v -72- concentration of the solution from.which films were made. Films made by dissolving the crystals in dimethylether showed a spectrum with two peaks at 11,050 and 6,900 cm-l, corresponding to Rb- and e; methylamine was used instead of dimethylether, metal respectively. However, when precipitation occurred and the spectrum showed a peak at 9,000 and a shoulder at 11,600 cm-1. When the initial film was redissolved, and the solution concentrated by solvent evaporation, the spectrum of the resulting film had a peak at 11,600 and a shoulder at 8,500 cm-l. Although it is difficult to assign the chemical formula of the two compounds based upon the Optical spectra, some conclusions can be drawn. If we assume that a 1:1 complex forms between the Rb+ cation and l8-crown-6, then the compound Rb18C6 Should be an electride Rb+18C6.e-. The formation of the Rb- band in the optical spectra can be explained in one of two ways. First, complexation of Rb+ with 18-crown-6 in solution may not be strong enough to prevent decomplexation; i.e. the following equilibria may occur: + - solution + - Rb 18C6.e(s) M Rb 18C6 +eSOlV. (III.2) + - - Rb 18C6 +2esolv. le=é§ Rb -+18C6 (III.3) Second, a solid state reaction may occur in the film of the type -73- 2Rb+18C6.e' e==é Rb+18C6.Rb--+18C6 (111.4) which would mean that the rubidide, Rb+18C6.Rb', is thermodynamically more stable than the electride, Rb+18C6.e-. However, metal precipitation from the methylamine solution of Rb218C suggests that in the presence of methylamine the electride Rb+l8C6.e- is thermodynamically favored over Rb+18C6.Rb-. In dimethylether solution no metal precipitation occurred and yet the spectrum still shows bands of both Rb- and e;. In general, the results do not provide clear-cut information about the species in the crystals. c. C518C6 and Cs(18C6)2_ The crystalline compound of stoichiometry C518C6 was synthesized by Dheeb Issa [104]. The Optical spectra Of thin solvent-free films of this compound exhibit time- dependence. Initially the spectra Show the absorption band attributed to both Cs- and to trapped electrons (e2), however, with time the absorption due to Cs- decreases and that due to e; increases. This suggests that the electride is the thermodynamically stable form. Accordingly this compound was assigned to be the electride, Cs+18C6.e-, although the sandwich ceside, CS+(18C6)2.Cs-, could not be ruled out. The stoichiometry Cs(18C6)2, immediately suggested that this compound is the electride Cs+(18C6)2.e-. The Optical spectrum of a thin film obtained by dissolving the crystals in methylamine -74- shows only a single narrow peak at 6,700 cm-l, independent of time, as expected for an electride. Figure 8 shows a comparison of the optical spectra of the two compounds. These results suggest that either the two compounds are electrides, Cs+18C6.e- and CS+(18C6)2.e- or that the former is a ceside, Cs+(18C6)2.Cs-. The latter assignment was later confirmed so that the change in the Optical spectrum of a film of C518C6 with time can best be explained in terms of the solid state reaction in the film Cs+(18C6)2.Cs‘ ———» Cs+(l8C6)2.e- +Cs (111.5) (5) i.e. the electride is thermodynamically favored over the ceside. d. CsC222 Figure 9 shows the Optical absorption spectrum of the compound CsC222. Although the stoichiometry of the compound implies that it is the electride CS+C222.e-, l which could be the spectrum shows a band at 10,500 cm- due to Cs- and a shoulder at.z8,000 cm.l attributed to the trapped electron. The formation of Cs- could occur in solution through the decomplexation of the compound. That this could happen was confirmed by the formation of Na+C222.Na- when the compound, CsC222, was allowed to react with sodium metal after dissolving it in -75- .mm mocoummou Eoum coxmu who mom-mo posoafioo may now mama .- - N-mUmH-mu pom .II.II- mom-mo mpooomeoo may mo madam mo muuoomm Hoveumo "m ousmem 9.0— ATEOK om m._ o_ m 00.0 ./ ./. . /. .1 .mo mow-moms? I .... . . 1 ..oa-momsmo II /. .\ . /. . .\ .1 - _ ._ ._ 0.. no no 0 _ o N -75- .NNNUmU pcsomeoo on» mo Eaem mo mnuoomm Hoveumo "m musmflm nIO_ .A-IEUvm ON m. 0. n - 4 H - e - 1.....1 o o O o o.. no Wm .0...o.. ...- .9. .LMWO o. a so ... v o . / o 1 v o v .... . m . a o m x fix a» lmWO ..o... .o....o p» oh IMHO ....o. ...o: o ...... 1 o.- - - p - DAV NAV O- AHN -77- dimethylether. The formation of the Na+C222 cation, as 23Na solid state NMR, requires the release indicated from of the cryptand 222 from the original compound. Also solutions of cesium salts and cryptand 222 in non-polar solvents contain two kinds of Cs+C222 complexes as indicated by the 133 Cs NMR [107] measurements. These two cation complexes are the inclusive complex, in which the cesium cation is Situated in the center of the cryptand cavity, and the exclusive complex, in which the cation is partially enclosed in the cavity of the ligand but is still partially solvated. If this is the case in methylamine solutions of CsC222, then the exclusive complex could provide an easy way for the formation of the Cs- anion during fast evaporation of the solvent. In any event, no definite conclusion can be reached about this system and more discussion will be deferred to the next chapter. 4. D.C. pressedppowder conductivity For an intrinsic semiconductor, at a given tempera- ture, there is a non-vanishing probability that some electrons will be thermally excited across the band gap, Eg. Accordingly, the number of holes in the valence bands and the number of electrons in the conduction band will be equal. The conductivity is then given by 0 = o exp[-Eg/2kT] (III.6) CD -73- where o0° is the limiting specific conductance at infinite temperature. However, intrinsic semiconductivity exists only in extremely pure substances and more commonly the resistance is characteristic of an extrinsic semiconductor or "doped" semiconductor. Two types of extrinsic semiconductors are known depending upon the position of the impurity energy level relative to those of the conduction and valence bands. In the P-type semiconductor the impurity energy level (Ea) lies close to the valence band (Ev)‘ If this level is empty then electrons can be excited from the valence band to the acceptor impurity level. An example of the P-type semiconductor is boron doped germanium which has an apparent band gap of 0.0104 eV [108] corresponding to the difference between the acceptor and valence band levels (Ea-Ev). A value of 0.67 eV is reported for pure germanium corresponding to the difference in energy between conduction and valence bands (Ec-Ev) [108]. On the other hand, if a filled or partially filled impurity level lies close to the conduction band, then electrons can be donated to the conduction band; this is called N-type semiconductivity. An example of an N-type semiconductor is the F-center in alkali metal halide crystals where electrons occupy anion Site vacancies due to crystal defects [109]. The apparent band gap in this case, will correspond to the energy difference between the conduction and donor energy level (Ec-Ed). -79- In general, conductivity in semiconductors arises from electron-hole formation in the conduction and valence bands (or in impurity bands). The behavior of both intrinsic and extrinsic semiconductors with temperature is exactly the same since, in principle, conductivity originates from electron-hole formation where the electrons are thermally excited to the conduction band from the valence band (intrinsic) or donor impurity band (N-type—-extrinsic) or to the acceptor impurity band from the nearby valence band (P-type-—extrinsic). The only notable difference is the limiting conductance at infinite temperature. For an intrinsic semiconductor, one expects to have equal populations of electrons in the valence and conduction bands at infinite temperature leading to a metallic conductivity value. On the other hand, an extrinsic semiconductor should yield a much smaller value for the limiting specific conductance at infinite temperature Simply because the number of states available to donate or accept electrons is not enough to yield a metallic value and the conductivity due to impurities should approach a limiting value at high temperatures that is smaller than that of a metal. Although the parent sodide compound, Na+C222.Na- has a bright golden metallic appearance, the D.C. powder conductivity measurements showed that it is a semiconductor as indicated from the linar relationship of lnfiR versus l/T. The slope of the line gave an apparent band-gap of -30- 2.4 eV [102]. Extrapolation of the straight line to infinite temperature gave an intercept which corresponds 6 ohm"1 cm-l). The to metallic conductivity (0 >10 results show that Na+C222.Na- is an intrinsic semi- conductor. On the other hand, the compound Cs+(18C6)2.Na- shows an apparent band-gap ranging from ~0.6 to ~l.5 eV 1 1 [981. with am in the range of 10.2 to 10 Ohm- cm— indicating that the compound might be an extrinsic semiconductor. In the present research, D.C. pressed powder conducti- vity determinations were made on three crystalline compounds, RbNa18C6, Rb18C6 and Cs(18C6)2. This was first done by checking Ohm's Law (Figures 10-12) by reading the current (I) versus the voltage (V) at a given temperature. Then the current was measured at a constant voltage (~5 Volt) as a function of temperature and the resistivity (p) or the conductivity (0 =1/o) was calculated for each sample. Figure 13 shows the plot of IhIo versus l/T for the compound Cs(18C6)2. The data for the compound C518C6 [98] have been included for comparison. Both compounds gave straight lines with slopes that correspond to apparent band-gaps of 0.9 and 0.8 eV for Cs(l8C6)2 and C518C6 respectively. However, the values for the limiting specific conductance at infinite temperature were about 102 and 1 ohm.1 cm“1 for Cs(18C6)2 and C318C6 respectively. This may indicate that Cs(18C6)2 is an intrinsic semiconductor while C518C6 is an extrinsic -81- .Imz.o0ma+om mo HooBOm OGHHHmummuoxaom mo.uoHd 3mg m.Eno "o- whomem to): N. : O. m m - w 0 ¢ m N _ O — - q u - d a q u q - la 1N ..m .... m. m 1? ..IC\ (D -82- .Io.m0ma+nm Mo HOOBOQ Toe-Houmwuowaom mo uon 3mg m.Eno "Ha choose Azo>v> m s m b c m m _ o - . - - - A - - mu 0 ..o. .8 m 09 m (on m -33- 2C>r ,- l5)- 0. E 3 I» O ..- '0 .— F4 5 .- l l l l l L L l L l 2. ll (5 EB K) v(volI) Figure 12: Ohm's Law plot of polycrystalline powder of Cs+(18C6)2.e'. log.o O’ '8!) Figure 13: -34- 36 3.8 4 .O 4.2 4.4 4.6 4.8 5.0 —-.(03 Plot of log conductivity vs. reciprocal temperature of the compound Cs+ (18C6) .Cs- (open circles) and Cs+ (18C6)2. e (solid circles). -35- semiconductor. Although the value of 0.8 eV for the band- gap of the compound C318C6 is proably too small for a pure ceside, the value for o0° indicates that the compound is most likely to be a dOped ceside. Figures 14 and 15 Show the plots of an versus 1/T for the two compounds RbNa18C6 and Rb18C6. It has been found that the samples Show higher conductance at a given temperature when the temperature is first increased and then decreased, but that the slopes of the lines are almost the same. A similar behavior had been found to be observed for samples of KNa18C6 and CsNa18C6 except that the samples showed more resistance when the temperature was first increased and then decreased [98]. These two types of behavior can be explained in terms of changes in the resistance with sample packing in the quartz cell of the apparatus or the presence of decomposition products which affect the resistance but not the band-gap. The calculated band-gaps for the two compounds RbNa18C6 and Rb18C6 were 0.9 and 0.4 eV respectively indicating that they are probably extrinsic semiconductors. 5. Differential Scanning_Calorometric (DSC) studies: The melting and decomposition temperatures of some of the alkalides and the electride Cs(18C6)2 were measured by Professor J.L. Dye during his stay at the A.T.T. Bell Laboratories. The results are summarized in Table 3. A typical DSC trace is shown in Figure 16 for -86- 24.0 23.0 22.0 0- C .J 2|.0 20.0 ISO 1 l l ' I 3.5 4.0 4.5 5.0 +- no3 . Figure 14: Plot of log resistivity vs. reciprocal temperature for polycrysEElline Rb+18C6.Na-. -37- I0.0 - LnP——>- 9.0 *- 35 4.0 4.5 ('/T )° (0° ——> Figure 15: Plot of log resistivity XE: reciprocal _ temperature for polycrystalline Rb+18C6.e . -88- o z. AOUGHV _ m m 8mm mg . “UmxfiHUN on m M um N H QC Q "0_ mHsmH .m 86- 2323th . . . q q u u — a q a q d a d d d d d u q , d u u u L O... L on o N (Mm) MOIs :oaH -39- the compound CS(18C6)2Na. The compound starts to melt at about 38.5°C (endothermic process) and the temperature for maximum decomposition, which corresponds to the position of the large exothermic peak, is 107°C. Figure 17 Shows the DSC trace for the two compounds Rbl8C6Na and Cs(l8C6)2. These compounds showed behavior similar to that of Cs(18C6)2Na but other processes apparently occur during decomposition. Table III: DSC Results for Alkalides and an Electride Melting Point Decomposition Compound (°C)~ Temp.(°C) RbNaC222 ~50a 63 CsNa(l8C6)2 38.5b 107 RbNa18C6 ~66 92 Cs18C6 60’”C 95 Cs(18C6)2 36 ~60 aInflection during decomposition. bEndothermic signal superimposed on exothermic peak. cSmall endothermic Signal at ~36-38°C. -90- .-.....v (o.m-oOmee+mo new A - (oz.moma+nm mo monEmm mom-Hmumwuowaom HON 0mm and mesmem 8..- mmnkdmmm-zwk on om- o- oo- om on 2. 8 on ow on cm 1 - a — q d d d d 1 a - a H q d a d a - q a ..N... - m- 10... I I \ Tm- Tm: ....... (MW) MO'Id lVEl-l CHAPTER IV SOLID STATE NMR A. Introduction The results obtained in Chapter III Show that two main problems remain concerning the assignments of species present in crystalline salts of electrides and alkalides. First, the analysis gives only the stoichio- metries and cannot differentiate between two different species that have the same stoichiometry. An example is the compound of stoichiometry Rb18C6. The analysis cannot differentiate between the electride Rb+18C6.e- and the rubidide Rb+(18C6)2.Rb-. However, in some cases, e.g. Cs(l8C6)2 and Rb218C6, the analysis tells us immediately that the first must be an electride, Cs+(18C6)2.e-, while the second is a rubidide Rb+18C6.Rb-. Second, a problem lies in the measurements of the Optical spectra of films. The crystals are first dissolved in a given solvent; the solvent is then rapidly evaporated to produce the film. Therefore, the absorbance of the film might reflect the Species present in solution which may or may not be the same as in the crystals. Therefore -91- -92- decomplexation, equilibria between species with different molecular formulae and dissociation of alkali metal anions into cations and solvated electrons as well as solid state reactions in the film should be considered in interpretating the optical spectra. Our assignment of the optical bands is based on the results obtained from studies of alkali metal anions in alkali metal- ethylenediamine solutions [29] and in films produced by solvent evaporation from solutions of stoichiometry M2C222 [95]. Such band assignments will only be valid if we assume that the nature of the absorber is independent of the environment. In fact, this is not strictly true since it was found that the band position of alkali metal anions in solution and in films as well as that of the trapped electron are sensitive to the solvent or crystalline environment. For example, the compound RbNa18C6, which is likely to be a sodide, Shows an absorption band that lies between those corresponding to Na- and Rb- in the,two compounds Na+C222.Na- and Rb+C222.Rb-. Films produced from solutions of stoichiometry LiC21l [96] showed several different absorption bands that must correspond to different sites for the trapped electrons since no metal anions are formed. The band positions of some of these trapped electrons lie in the range of the absorption maxima of Rb- and Cs- in both solutions and films. Consequently, the unequivocal assignment of the species present in -93- these salts from Optical spectra alone is not possible and another technique is needed to determine the nature of the species present in electride and alkalide compounds. B. Magic Angle Sample Spinning (MASS)-NMR Nuclear magnetic resonance spectroscopy is a powerful way to probe the microscopic environment of a particular constituent in a given sample. It is well- known that NMR spectral lines of solutions are much narrower than those of solids. This very substantial difference in behavior arises from the static anisotropic interactions to which the nuclei are subjected in the solid state. A good example to illustrate this is the 1H-NMR linewidth of water and ice. The resonance line of ice is a million times broader than that of liquid water. Line broadenings in solids result from magnetic dipolar, chemical shift, electron-coupled nuclear spin (J-coupling) and quadrupolar interactions. In fluids, the rapid random thermal motion averages most of the anistrOpic interactions (except second-order quadrupolar contributions) and . effectively removes them from the spectrum. Building upon this simple idea, in a rapidly rotating solid, if the speed of rotation is high enough and the angle of rotation is properly chosen, only the isotropic interactions should contribute to the spectrum. This is the basic principle of the MASS-technique, first developed by Andrew [110,111] and Lowe [112] in 1959. -94- Three kinds of interactions are of great interest to the present work and will be discussed in detail. These are the magnetic dipolar, chemical shift anisotropy and electron-quadrupolar interactions. 1. Direct magnetic dipolar interactions: The origin of the magnetic dipolar interaction arises from the modification of the magnetic environment of a nucleus by the field due to the magnetic moments of neighboring nuclei. The magnitude of such a field can vary between iZu/rB, where u is the nuclear magnetic moment and r is the distance between the two nuclei [113]. Accordingly, nuclei of the same kind will experience various fields depending on their position and orientation in the solid, and magnetic resonance will occur over a range of frequencies so that the spectral line will be broadened. The truncated dipolar interaction Hamiltonian, for all nuclear pairs i,j is given by [114] 2 = 2 A..fl3cos G..-l) (IV.1) J(l) i%-possess an electric quadrupole moment that can be attributed to a non-spherically symmetric nuclear charge distribution. The electric- nuclear quadrupole interactions in nuclear magnetic resonance can be divided roughly into two regions according to the magnitude of these interactions relative to Zeeman interactions. The first is called the "high field" case in which the nuclear electric quadrupole interaction energy is assumed to be small as compared to the interaction energy of the nuclear magnetic moment with the external magnetic field. In this case, the quadru- polar effects manifest themselves only as more or less significant perturbations of the purely magnetic interactions. The perturbation can split the resonance -101- line into several components whose number depends upon the nuclear spin. When the electrostatic field gradient at the nuclear position can be calculated the separation in frequency between these components yields the nuclear quadrupole moment. Accordingly, the quadrupolar interac- tions can result in a fine structure of the resonance line, because of overlapping lines, and may lead instead to a broadening or apparent loss of intensity of the resonance line. Furthermore, these interactions often reduce significantly the relaxation time necessary for the spin system to attain thermal equilibrium. The second region is known as the "low field" case or the pure quadrupolar resonance. In such a case the quadrupole interaction can be so large that it becomes responsible for almost the entire dependence of the energy of a nucleus on its spin orientation. Only the first case will be considered here and will be discussed below. The Hamiltonian for the interaction between the nuclear quadrupole moment and the electric field gradient is given by [114] e01 . £0 = i EIiIZIIIIT (Ii'Vi‘Ii) (1v.13) in which eQi is the nuclear electric quadrupole moment of nucleus i and Vi is the electric field gradient tensor -102- at its site. For simplicity a single crystal will be considered and we will assume that the electric field gradient has cylindrical symmetry so that the asymmetry xx-VYY)/VZZ vanishes [114], where VXX' are the principal axes values of the electric parameter n = (V VYY and sz field gradient tensors. The Hamiltonian can then be rewritten as $43 cos2 o-l) (31:-I(1+1) :K- 5‘9 +3'o o'1(1+1)+(1+1)1 (1v14) Q - 4I I-l 2 Sin cos L z + - 1 - z] ‘ 3 . 2 2 2 :2- 31n 0 (I++I_) .1 where O is the angle between the principal axis of the electric field gradient tensor and the laboratory axis, and eq==V The quadrupolar Hamiltonian can be treated ZZ' as a perturbation of the Zeeman Hamiltonian, Mg, and the calculation is extended to include the second-order terms, so that _ (0) (l) (2) Em - Em -+Em -+Em (IV.15) and the frequencies that correspond to the spin transition m+——+m-l will be given by E Vm = JE:%—JE = vL-+vél)-+vé2) (IV.16) -103- where VL is the Larmor frequency of the nucleus and vél) is the first-order splitting of the energy levels and is given by 2 0(1) = v (m--]2-'-)[%(3 cos G-l)] (1v.17) m Q where 00 is the nuclear-quadrupole frequency and is given by 2 -3eQ ”Q ‘ h21(21-l (IV°18) and ezQ/h is the electric quadrupole coupling constant of the nucleus. The second-order frequency shift, véz), is given by [117] 2 0r: ) = [AZ—181‘] |f1| [24m(m-1)-4a+9] -§-If§l x [12m(m-l)-4a+6] (IV.19) where a=I(I+1), lf’fl =/372’sino coso and (13%| =1/2/37—2 sin‘2 0. For quadrupolar nuclei of half-integer spin, such as alkali metals (except 6Li), the frequency of the central transition -%«——+% is not shifted in first-order by the (l) quadrupole interaction because the first-order shift um vanishes for m==% as indicated by Equation (IV.17). The frequencies of the other transitions will be shifted and satellite lines will appear on each side of the central -104- line. The second-order frequency shift of the central transition (-%-—+%) can be obtained by replacing m in Equation (IV.9) by i and is given by 2 0,22) = 1.67% (a “2) (1 - c0320) (9 cos2 o-l) (1v.20) Figure 19a Shows the NMR powder pattern of a nucleus of Spin I==% and the effect of first- and second-order quadrupolar interactions on the three transitions [118]. Frequently, the satellites are spread out over such a large frequency range that their wings become unobservable and it is well accepted that only the central transition is observable for the half-integer spin nuclei in non- cubic solids. This has been demonstrated by E. Oldfield pp 31. [118] who showed that the 23Na-MASS-NMR spectrum of a mixture of equimolar amounts of NaCl (equ/h;:0) and NaNO3 (eZqQ/h;:0.3MHz) (Figure 19b) consists of two 23Na lines of different intensities. The two lines correspond to the three spin transitions in 23Na in NaCl and only the central spin transition (%u-—+-%-in NaNO3. The effect of high-speed sample-rotation on the quadrupolar interaction in solids has been theoretically studied by Cunningham and Day [119]. They found that when the sample is rotating with an angular velocity mEaround an axis included at an angle 8 to the magnetic field, the first-order time-averaged diagonal terms of Figure l9a(Top): Figure l9b(Bottom): -105- Spin I==3/2 NMR powder pattern of an axially symmetric crystal Showing the effect of first-order quadrupolar interaction (broken line) and second-order quadrupolar, dipolar, chemical shift (CSA) interactions (solid line) [114]. 23Na-MASS-NMR spectrum of equimolar mixture of NaCl and NaNO3 lines labeled x and y re resent the spinning side bands [118]. . Resonance -106- CSA 2nd order quadrupolar ( -) 2’ 2 {dipolar _-—-- Frequency r / N aCl NaN03 (equ/h z 0-3 MHz) NaNO3 x x x y Y X l l I ' I --100 O 100 Frequency, ppm -107- the quadrupolar Hamiltonian are given by eZQ Mo(t) = 61 21- (312-12) (1+n) (1-3 cos2 8) (1-3 c0320) (IV.21) Clearly, at the magic angle (cos-luCE73) the first-order quadrupolar broadening can be eliminated provided that there is a sufficiently high rotation speed compared to the lifetime of a spin state, so that a given nucleus has ample time to "see" all orientations. Although the first-order quadrupolar interaction vanishes for the central transition of half-integer spin nuclei, the second-order effect is not averaged to zero by the magic angle experiment. This led to a general belief that the spectra of quadrupolar nuclei would be intractable. Contrary to that belief, it has been shown [120] that relatively high resolution spectra may be obtained from quadrupolar systems with half-integer Spin by using the MASS-technique. Equation (IV.20) predicts a residual linebroadening of ~(e2qQ/h)zvL which is of the order of ~60 KHz for a quadrupolar coupling constant of 1 MHz and a Larmor frequency of 15 MHz. However, the detailed expression for the static second-order effect [(Equation (IV.20)] suggests that Operation at high field (high 0L) and preferably large I can make the experimental spectra tractable. In addition, when the numerical factors in -108- (2) the full expression of vL5 are taken into account, there is a dramatic decrease in the residual linewidths [118] (véz).z10 ppm) upon fast spinning of the sample. In summary, fast specimen rotation about an axis oriented at the angle 8 introduces the factor %43IOOSZB-l) into the time-average Hamiltonians which represent magnetic dipolar, chemical shift and first- order quadrupolar interactions. In fact, Andrew [115, 116] showed that the same factor also appears in the time-average Hamiltonian representing the indirect- electron-coupled spin-spin interaction. At the "magic angle" 54.74° the anisotrOpic parts of these interactions are eliminated from the center of the Spectrum and appear as spinning sidebands at each Side of the spectrum at multiples of wr/2n i.e. increasing the rate of rotation will move the sidebands further out. The intensities of these sidebands fall rapidly with increasing or as m;2n for the mpg sideband. Finally, it is worth mentioning some of the limita- tions and factors that affect the resolution in the MASS technique. Although a relatively narrow NMR-line can be obtained from the MASS experiment, this is achieved at the expense of the information contained in the interactions described before. Three important factors greatly affect the resolution in the MASS experiment. These factors are, imperfect adjustments of the magic angle, angular instability and insufficiently ~109- fast spinning. If the angle 8 deviates from the "magic" value by a small angle 8, 0.l°, the residual linewidth will be /2E of the original width [116]. Accordingly, the spinner axis must remain in adjustment and must not wander, wobble, or precess Significantly from its correct orientation. In the MASS-experiment all of the interactions described above can be removed from the spectrum only if the spinning speed is high enough to average these interactions. For example, for the magnetic dipolar interaction, the spinning frequency should be equal to (or greater than) yH where H is the field produced by the other magnetic nuclei at the position of the nuclei in question. Accordingly, proton-MASS-NMR is totally intractable unless the protons in the system are very dilute or are largely separated from each other. C. Results and Discussion 1. 23Na-MASS-NMR: g, Ideptifigation of the species The sodium nucleus 23Na, which has 100 percent abundance and spin 3/2, has an electric quadrupole 24 cm2 [121]. Consequently, moment on the order of 0.14 Xlo- the quadrupolar broadening in sodium salts may vary from 0 to 30 KHz [120]. Figure 20 shows a comparison between the MASS and static NMR spectra of Na+18C6.SCN- measured -110- . .nms mm.~ o H(zom mowe+oz mo owuooom mZZIcZMN mwcwe o.eon- caccean can -ocee oouuocc oecomm "om mucosa o - 2.2 so: can E-zm .222er Zan- Om- 0w. 0?. ON- 0 ON 0? b—-Fh_~PF_T.L «22 mass- 58 .28 .89 .oz -111- at 132.35 MHz (500 MHz proton frequency). The linewidth of the Na+ signal was reduced by a factor of 4 upon spinning the sample at the "magic angle" (8 = cos-1m) with a Spinning frequency of 3.7 KHz. Table IV giVes the chemical shifts (referred to Naiaq) at infinite dilution) and the linewidths of some conventional salts both with and without complexing agents and for a number of crystalline sodides. The listed chemical shifts are not corrected for second- Order quadrupolar shifts of the resonance. The last column in Table IV contains the position of the Optical absorption band of thin films prepared from methyl- amine solutions. The only homonuclear compound studied was Na+C222.Na- and its static and Spinning spectra are Shown in Figure 21. The MASS-spectrum contains two 23Na signals that correspond to the sodium-cryptate cation and the sodium anion (Na-). For heteronuclear sodides 23Na peak was Observed. The chemical shifts only a single of these Single peaks are in the range of -53 to -63 ppm which is quite different from the value obtained for the Na+ cation in model salts (:20 ppm). The corresponding chemical shift of the gaseous sodium atom is -61 ppm [124] while the calculated chemical shift for the gaseous sodium anion is -64 ppm [125]. Clearly, all of these alkalides are Sodides and the concentration of Na+, either free or complexed is too low to give a detectable NMR signal. Dye g; 31. [55] reported an average value -112- Table IV: Results of 23Na-MASS-NMR at 52.94 MHz Compound 6 (ppm)a Avg (Hz) Optical peak (cm-l) Nafaq)f o 25 -—— NaCl + 8 140 -—— NaBr + 6 200 -- NaI - 3 160 -- Na+18C6.SCN' -22 700 -—- Na+C222.SCN- -21 1120 -—- Na+c222.1‘ -21 1300 ——— Ngfc222.Na' -24 1200 -—— Na+c222.§3f -61 290 15,400° K+c222.Na‘ -61 400 15,100c Rb+c222.Na“ -61 360 13,800 LiC211.Na' -63 850 -—— K+18C6.Na‘ -56 120 14,000° Rb+1BC6.Na' -60 90 13,800 Cs+(18cs)2.Na’ -62 75 14,600d K+(12C4)2.Na- -61 51 13,1008 aChemical shifts of separate samples were generally reproducible to :1 ppm. bReference 95. cReference 122. dReference 98. eReference 123. quueous, infinite dilution. -113- NO+ C222 - NO- r——-NO- , (NON-SPINNING) NO+ Ic2222 I (SPINNING) 1 l l I L4, 1 _I I 1 +|OO 0 -I00 -200 8 [ppm FROM No+(oq.)] Figure 21: Static and spinning 23Na-NMR spectra of Na+C222.Na‘ at 52.94 MHz. ~114- of -62 ppm for the chemical shift of Na- in different solvents, nearly independent of the solvent. Out of the eight sodides studied, six have anionic chemical shifts that range from -61 to -63 ppm and the other two, K+18C6.Na_ and Rb+18C6.Na-, have chemical shifts of -58 and -60 ppm respectively. These two alkalides also have a substantial concentration of trapped electrons in the crystals as indicated by EPR spectroscopy (see next chapter). The interaction of such trapped electrons with Na- might cause this paramagnetic (downfield) Shift. However MASS-NMR spectra of some samples of Rb+18C6.Na- and K+18C6.Na- showed Egg Na- peaks. For example, Rb18C6Na (at 52.94 MHz) showed two peaks at -62 and -53 ppm with linewidths of 40 and 530 Hz respectively while K18C6Na (at 132.35 MHz) Showed two peaks at -60 and -57 ppm with linewidths of 180 and 223 Hz respectively. This behavior is not reproducible from one sample to another but was only observed for one sample of each while most of the samples studied showed only one Na- peak. Figure 22 shows the MASS-NMR Spectrum of K+18C6.Na- at 132.35 MHz. In general, the results Obtained here, together with those from solution [55] suggest that the chemical shift of Na- in the solid state as well as in solution, is remarkably insensitive to the nature of the sodide or the solvent. Also, the absence of any significant paramagnetic shift relative to the gaseous sodium anion -115- rill—T—IVU—I'YIIWYllTUfrrTT'TfiITITIj - 30 -40 -50 -60 -7O -80 -90 3 [ppm from N°+(aq.)] Figure 22: 23Na-MASS-NMR spectrum of one sample of K+18C6.Na‘ at 132.35 MHz. -116- suggests that the sodium anion exists in solid or solution as a "gas-like" anion with its 2p-orbita1 well shielded from the environment by the completely filled BS-orbital. On the other hand, the position of the optical absorption bands are very sensitive to the environment. This is not unexpected since large solvent- dependent shifts have also been Observed for the position of the optical absorption band of Na- in solution [126]. This suggests that the excited state of Na- (3s 3p) is very sensitive to the nature of the complex cation in the solid and the solvent in solution. The variation of the linewidths among different compounds is rather interesting. The substantially narrower peaks for the Na_ in the crown ether salts compared to the cryptate case suggests greater interaction in the cryptate sodides and/or enhanced relaxation via interaction with the nitrogen atoms of the cryptand. The narrow lines of the crown ether sodides suggest a rather symmetric environment and isolation from the neighboring cations. Similar line narrowing is observed in the model salts but in this case for the Na+-complex cation. The Na+-18-crown-6 NMR line is narrower than that of the Na+-cryptate. This suggests a greater asymmetry of the electric field gradients at the Na+-cryptate. These electric field gradients are determined largely by the complexant environment and not by the nature of the anion. -117- b. Freguency dependence and proton decoupling The results obtained for the linewidth of both the Na+ complex cation and Na" suggest that there are significant interactions between the sodium nucleus and its environment. In general, the MASS-NMR line contains the residual magnetic dipolar, chemical shift interactions and the second-order quadrupolar effect which is not averaged to zero in the MASS-experiment. The residual chemical shift anisotropy is dependent on the magnetic field since the total spectral breadth is directly proportional to the applied field. The second-order quadrupolar frequency shift is inversely proportional to the Larmor frequency, i.e. to the applied magnetic field, and the spectrum gets narrower and the resonance position is shifted downfield upon increasing the magnetic field [see Equation (IV.20)]. In sodides formed with crown ethers and cryptands any magnetic dipolar interactions must arise mainly from the interactions of the sodium nucleus with the complexant protons and the nitrogen atoms of the cryptands. Sodium-homonuclear magnetic coupling is probably weak because of the large separation between the nuclei and the shielding of the Na+ by the complexant. Interaction with other magnetic nuclei such as 13C and 17O is negligible because their Spins are very dilute. Table V summarizes the chemical shifts and linewidths of the sodide salts studied at three Larmor frequencies, 52.94, 95.29 and 132.35 MHz which correspond -118- mm Ho- we so- an ac. mm so- (oz.-aONH-+s I: (I so Ho- om Ho- mo so- (oz.-mOmH-+s (I I: me No. me mo- os mo- (oz.-oomeo+mo doe so- om om- om oo- oo oo- (oz.oome+om mam am- mme em- owe om- ms mm- (oz.oome+x (I (I ohm Ho- mam mo- oo mo- (oz.~m~o+om am so- cam so- one so- om No- (oz.mmmo+s omm so- one so- can so- om Ho- me.-~o+oz cap a. one He- come am- com om- (oz.-~oemm cam ea. In In some em- omo om- -H.~N~o+oz oem me- (I (I cos mm- oeo mm- (zom.oome+oz um . >4 Edd .0 am .xe< Edd .0 a: .m9< Edd .0 an .w?< Edd .0 as: mm.~me as: o~.mm as: cm.~m as: am.mm IIIIIIIIIIIIIIIIIIIIIIIII OOHdSOOImH IIIIIIIIIIIIIIIIIIIIIII II OOHdSOOOOIma II II 0:50dsou II mzz oz 00 mm nuowzocflq poo umwcm HaowEOcU moo mo mocOUGOdOo accosooud poo OOHdSOOOOIcouOHd u> mecca -ll9- to 200, 360 and 500 MHz proton frequencies. The effect of proton decoupling at 52.94 MHz is also included in the table. The Na+ signal in Na+C222.Na- and the model salts seems to be only partly broadened by the proton- Na— dipolar interactions since the linewidths are still of the order of several hundred Hertz. The most interesting point is the quadrupolar powder pattern observed for the Na+ peak of Na+C222.Na- and Na+C222.I- but not for Na+18C6.SCN-. This indicates that quadrupolar interaction is the main source of line- broadening in the Na+ cryptate signal. As expected for a quadrupolar system, the Na+- cryptate chemical shift moves downfield and approaches its isotropic value, -9 ppm, at high magnetic field. This chemical shift value is in agreement with the value obtained for Na+-cryptate in solution with various solvents (-9 :1 ppm) [55]. Figure 23 shows the MASS-NMR spectra of the compound Na+C222.Na- at three frequencies and at 52.94 MHz with proton decoupling. A computer lineshape simulation [127] for the Na+-cryptate signal with proton decoupling showed the same lineshape with an asymmetry parameter equal to 0.1 and a quadrupolar coupling constant of 1.2 MHz. These parameters were used to predict the lineshape and isotropic chemical shift at Larmor frequencies of 95.29 and 132.35 MHz. Figure 24 shows a comparison between the observed and simulated spectra. -120- 52.94 MHz -'H-coupled I32.35 MHz - IH-coupled 92.29 MHz-'H-coupled I 52.94 MHz-'H-decoupled #L 40 20 0 '20 ‘40 '60 '80 'lOO ‘l20 "l40 ppm 23Na-MASS-NMR spectra of Na+C222.Na- at Figure 23: three different frequencies. -121- IDEHBEFTVEHD EflhflLfl-Afliflb VI.” 52.94MHZ .... h 11:95-29 MHz Figure 24: Observed and simulated 23Na-NMR signal of Na+0222 in Na+0222.Na‘ at different frequencies. -122- The chemical shifts of the Na- in all of the sodide salts remained constant when the field was increased indicating that the quadrupolar effect is minimal for this species. However, some of the Na- lines showed a line broadening at higher fields which might be due to field inhomogeniety and/or residual chemical shift anistropy. On the other hand, the Na- linewidths are greatly affected by proton decoupling. For example the Na- signal in Na+C222.Na- is narrowed by a factor of 10 upon decoupling the protons. This indicates that the electric field at the Na- is nearly symmetrical and implies near centrosymmetry of the Na- anion. In summary, line broadening is mainly due to quadrupolar interactions for the Na+-complex cation. This interaction dominates all other interactions present at low fields. At high magnetic fields the second-order quadrupolar effect is greatly reduced and the chemical shift approaches its isotrOpic value. The quadrupolar interactions are absent for Na- because of its symmetry and magnetic dipolar coupling with the nearby protons of the complexant is the main source of line broadening. Figure 25 Shows the MASS-23Na NMR spectra of K+C222.Na- at 52.94 MHz with and without proton decoupling and at 95.29 MHz. -123- K‘C222-No‘ 21"No - MASS NMR AV'IZ ' '70 Hz Alfi/Z ‘2‘0 HZ Au”, =30 Hz l) 'H-coupIed at 52.94 MHz 2) 'H-coupled a: 95.29 MHz 3) 'H-decoupled at 52.94 MHz l l l 1 -40 -50 -60 -70 -80 -90 CHEMICAL SHIFT ppm from Not“, Figure 25: 23Na-MASS-NMR of K+C222.Na- at different frequencies. -124- 2. 133 Cs MASS-NMR a. Species identification and frequency dependence 133 The cesium nucleus CS, which has 100 percent abundance and Spin 7/2 has an unusually small electric 27 cm2 [121]. In general in quadrupole moment, -3 X10- solids the quadrupolar broadening in cesium salts is of the order of 103 Hz and first-order theory suffices to describe it [115]. Accordingly, relatively narrow lines would be expected for cesium salts since, in principle, first-order frequency splitting is averaged to zero by Spinning the sample at the magic angle [119]. Table VI summarizes the chemical shifts (referred to infinitely dilute CSIaq)) and the linewidths at 47.24 and 65.61 MHz (which correspond to 360 and 500 MHz proton frequencies) for cesium salts both with and without cation complexants and for a number of cesium-containing electrides and alkalides. Among the simple cesium salts only CsTPB shows no 133Cs-NMR signal probably because it is too broad to be detectable. The model compounds of simple cesium salts with 18-crown-6 show narrower peaks compared to the corresponding salts without complexing agents, which indicates higher symmetry at the Cs+ cation upon complexation. The chemical shifts of 1:1 complex cations of CS+ with 18-crown-6 depend on the anion and complexation of the Cs+ cation by lB-crown-6 causes upfield shifts relative to those of simple cesium salts. .moocwucoo H> wanna -125- one om~+ ova om~+ ..(zom.-~o+no II: (II one .oem ms~+ .om+ o Izom.-~o+no oom as- oo mo- (mesa-opoH-+no oo am. so am. (Hm-oOoao+mo III (In on om- (zom.-o0oeo+no oos ooe+ oom ease+ (H.oOoe+no (I: (I: ooe ms+ (zom.oOoH+mo (II (II III (II chemo (II (In oms ao~+ Hno .II. (II (In oo~+ Ammo (I: III has ~m~+ Homo (II (I: ooe ooa+ zomno oo mm- on ma: -oo-Hao z s.o u: .xac Edd .m0 N: .x>< Edd .m0 um: Ho.mo mm: s~.~o museums mzzummazuuo "H> canoe MMH -126- .cofluSaOn HocccHOE a Eoum cowuauOdc>o uco>aom do ooucdond,OHdEam .mmo- .wm mm mamas mo mEOHHOOHAO as» ou mcwouoooo oouodond OHdEam no ouseao Seouecemcs on cousowom .chd uoadflue IUvQU'U omm .oom oem+ .vma+ mmv omml oom .mmH ova .meu mm mm: mma mm: 00H moi mMN mm+ um .x>< Edd .m0 um: H0.m0 .-co- + oea .omm mm~+ .mma+ omm mmmn mod H0: mHH mm: mm mm: ov Hon mma Hm+ one .mo om~+ .~m~+ com hwa+ .x um 90 Edd .m0 um: (o.-~oao .mm.-oooa-+ao (no.-oOoH-+mm (om.-o0oeo+no (e.m-GOoHo+no (oz.-oOoeo+no (o.-oOoH-+no -.(H.-~o+no u (Ho.-~o+no .poscflucou H> mange -127- This indicates less donation of electron density to the cesium ion in the complex than in simple salts. Salts that contain the 1:2 "sandwich" complex cation of Cs+ with 18- crown-6 Show narrower peaks than those with the 1:1 complex. The chemical shift of the 1:2 complex is almost anion independent and the resonance position is shifted upfield relative to the 1:1 complex. This indicates that even more shielding from the counter-ion is achieved by adding a second ligand molecule to the 1:1 complex to form the "sandwich" complex. The results obtained here for solids are similar to those obtained by Dye and Popov [128,129] for solutions. They found that the 1:1 complex shows both solvent and anion dependent chemical shifts while the "sandwich” complex shows neither solvent nor anion dependency. The NMR results for both solid and solution show that simple cesium salts can form both 1:1 and 1:2 complexes with 18-crown-6 in solution or in the crystalline state. The formation of 1:1 or 1:2 complexes in the solid state depends, however, on the method of preparation. For example, solutions of stoichiometries CSSCN.18C6 and CSSCN.(18C6)2 formed the crystalline compounds Cs+18C6.SCN- and Cs+(18C6)2.SCN-, respectively, when the crystals were precipitated from l-propanol solutions, but a mixture of 1:1 and 1:2 complexes were formed when a methanol solution of stoichiometry CsSCN.l8C6 was evaporated to near dryness and allowed to cool to 0°C. -128- For cesium-containing electride and alkalide salts, 133Cs-MASS-NMR provides an easy way to identify the species present in these compounds. Figure 26 shows a comparison between the 133Cs-NMR spectra of the two compounds of stoichiometries C318C6 and Cs(18C6)2 measured at 47.24 MHz. The compound C518C6 shows two 133Cs-NMR lines at -61 and -228 ppm. The -61 ppm chemical shift value is in agreement with the chemical shift of the Cs+ cation in the "sandwich" complex Cs+(l8C6)2; the other peak is diamagnetically shifted to -228 ppm, which is close to the chemical shift value obtained for the CS- anion in THF solutions (-292 ppm) [55]. These results clearly Show that the compound C318C6 is a ceside CS+(18C6)2.CS- and not an electride, Cs+18C6.e- as initially presumed [104]. On the other hand the compound Cs(18C6)2 shows only a single 133Cs-NMR peak as expected for the electride Cs+(18C6)2.e-. The resonance position of the electride peak is shifted about 140 ppm downfield (paramagnetic shift) from the expected position of the Cs+ cation in the "sandwich" CS+(18C5)2 complex. The origin of this shift will be discussed later. The other alkalide compounds, CsNa(18C6)2, CsK(18C6)2 and Cst(18C6)2 Show only a single 133Cs-NMR peak that corresponds to the Cs+ cation in the Cs+(18C6)2 complex. This indicates that these compounds are Cs+(l8C6)2.Na-, Cs+(l8C6)2.K- and Cs+(18C6)2.Rb- respectively. The results confirm that in contrast to model salts that -129- can form both 1:1 and 1:2 complexes of the Cs+ cation and 18-crown-6, 9311 the 1:2 complex forms when alkalides or electrides are allowed to precipitate from amine and ether solutions, regardless of the solution stoichiometry. Figure 27 shows both 23Na and 133Cs-NMR spectra of the compound Cs+(18C6)2.Na_ which was initially believed to be Cs+18C6.Na‘ [94]. Increasing the magnetic field (i.e. increasing Larmor frequency) does not affect the resonance position, but the NMR lines at 65.61 MHz are broader than those at 47.24 MHz. The line broadening is observed for the central transition (-%-—++%) as well as for the total Spectrum. The results indicate that quadrupolar interactions are minimal and that chemical shift anisotropy and magnetic dipole interactions are the main sources of line broadening in these compounds. This could be verified by examining the spinning spectrum with and without proton-decoupling and by studying the powder pattern of the static spectrum at different magnetic fields. The former was unavailable because of instrument limitations and the latter has not yet been done. However the static spectra of all cesium compounds which contain the "sandwich" Cs+(18C6)2 complex show nearly the same powder pattern at 65.61 MHz. Figure 28 shows the static and spinning spectra of the compound Cs+(18C6)2.e-. The static spectrum shows the asymmetric profile characteristic of an axially symmetric chemical shift tensor similar to that observed -130- N .Nmz vN. No on o.-oOoH-+no can no. -oOoeo +no monsoosoo poem-manhuo 03» wow mo muuomdm mZZImmdzImO-mmH “0N ousmed Emerge-+3 E9; ream 6269-0 00m- oom- oo- - o oo- OON 00m - a - a _ . - . - 1 - . - East 88...: OZOomcmO mNN-um -mo _o-.m mo _®+uw lmu -13l- .Imz.-oOmeo+mo OESOdEOO onu mo mHDOOdm dzzlmmdzImO on m u Thoma mma 0 2mm mm .d macaroz zoo.- E8- .2 com- 00-. o co... m 4 - - - W 32262 woo-.8 sod-.1 Eda-b . 1 . . 1 1 .r l . coo - 0?... oo- .. o 00- + 0% + 4 _ _ ”:22 no. m .5 /' /+mU Toz .mom- .3 -132- 4 I l I I j I 400 300 200 IOO 0 #00 “200 Chemical Shift from Cs+(o.q.)(ppm) Figure 28: Static (top) and Spinning (bottom) l33Cs- MASS-NMR spectra of the compound Cs+(18C6)2.e‘ at 65.61 MHz. -133- by Andrew 32 £1. for 111Cd metal [130]. The isotropic chemical shift that is obtained by spinning the sample at the "magic" angle does not coincide with the peak position of the central transition of the static spectrum. This is expected Since the anisotropic part of the chemical Shift interactions is removed by spinning the sample at the magic angle leaving only the istropic part of it and appearing as spinning side bands. The structure that appears on both sides of the central transition (-%-—++§) in the static spectrum must be due to other spin transitions of 133 Cs which collapse to a single line by spinning the sample at the magic angle. The fact that all cesium cations in the "sandwich" complex CS+(18C6)2 produce similar static NMR spectra indicates that local structure around Cs+ cations in these compounds is determined mainly by the complex cation and not by the anion. The cesium-cryptate system shows rather interesting behavior. It is known that the CS+ cation forms only 1:1 complexes with the cryptand 222. Dye and Popov [107] found that two types of complexed cations can be formed between Cs+ and C222 in solutions of simple cesium salts and C222 in non-polar solvents. These are the "inclusive" and "exclusive" complexes in which the Cs+ cation is totally inside the cavity ("inclusive") or partially complexed by the ligand ("exclusive"). However, X-ray diffraction studies of the compound Cs+C222. SCN-.H20 [82] -134- show that there is only one cesium cation complex in this compound, the "inclusive" complex. The 133Cs-MASS-NMR spectrum of a polycrystalline sample of this compound, prepared according to the direction of Wiess 23 El. [82], 133Cs-NMR lines at showed, however,the presence of £32. +50 and +275 ppm, while samples prepared by dissolving equimolar amounts of CSSCN and C222 in methanol followed by solvent evaporation to dryness showed only 933 peak at +238 ppm. This demonstrates that cryptates with the Cs+ cation in either "inclusive" or "exclusive" sites (or both) can be formed depending upon the conditions used to prepare the compound. The electride of stoichiometry CsC222 prepared by method 2 (see Chapter III) and the crystals formed from solutions of reactant stoichiometries Cs2C222 (method 1), C82C222Li (method 2) and CsC222 (method 3) all have two 133Cs-NMR lines at +138 and +238 ppm. The relative peak intensities vary from one preparation to another. However, the results indicate that regardless of the solution composition only one type of compound is formed with the stoichiometry CsC222. The peak positions and the variation of the peak intensity with the sample preparation method indicate that neither of these two peaks can be due to the Cs- anion. The compound must, therefore, be an electride, CS+C222.e-, with either two different sites for the CS+ cation or two different kinds of crystals. Figure 29 shows a comparison between the -l35- l I I I 300 200 lOO 0 Chemical Shift from Cs+(a.q.)(ppm) 133Cs-MASS-NMR spectra of the two crystalline compounds Cs+C222.SCN'.H20 (top) and Cs C222.e' (bottom) at 42.24 MHz. Figure 29: -136- 133Cs-MASS-NMR spectra of the two compounds CS+C222.SCN-. H20 and Cs+C222.e- measured at 47.24 MHz. It is difficult, based upon the NMR results, to determine whether these two cesium sites are present in a Single crystal (mixed crystal) or if the compound CsC222 consists of two crystalline forms (mixture of crystals). However, the +238 ppm peak in the CsC222 NMR spectrum is close to the +244 ppm limiting value obtained for a 1:1 mixture of CSSCN and C222 in methanol solution at low temperatures [107]. This resonance was attributed [107] to the "inclusive" complex between the Cs+ cation and cryptand 222. The large paramagnetic shift occurs because the cryptand cavity has been deformed or stretched to permit inclusive complexation of the large Cs+ cation. This tightly fitting cesium ion inside the stretched cavity results in strong overlap of the lone-pair orbitals of the ligand and the outer p-orbital of the Cs+ cation, which causes large downfield (paramagnetic) shifts. The other peak at +138 ppm might be due to the "exclusive" complex, in which the Cs+ cation is partially enclosed in the cryptand cavity to give weaker overlap and less paramagnetic shift. The intensities of the two 133Cs-NMR peaks depend on the method of preparation of the compound CsC222. The crystals of CsC222 prepared by methods 1 and 2 Show comparable intensities of the two signals while the compound prepared by method 3 shows mainly the signal of the "exclusive" complex. Figure 30 -137- B 400 200',m -200 400 200 o -200 PPm .av—flaA—AMALJVJ C: 400 200 ' -200 400 200 3 360 PPm PPm Figure 30: 133Cs-MASS- -NMR spectra_ at 65. 61 MHz of the compound Cs+C222. e at different spinning frequencies: A: 1.8, B: 2.6, C: 3.8 and D: 4.0 KHz. ~138- 133 shows the Cs-MASS-NMR spectra at different spinning speeds of the compound CS+C222.e- prepared by method 3. b. The chemical shift of Cs+(1806)2.e' The resonance position of the Cs+ cation in the compound Cs+(18C6)2.e- is shifted about 140 ppm downfield from the expected position for a Cs+ cation in the "sandwich" complex Cs+(l8C6)2. There are two possible sources of this shift. These are the structural and paramagnetic effects. The former presumes that the structures are different and Should be temperature independent. If the shift is due to the paramagnetism of the sample, however, one expects a strong dependence of the shift on temperature that would approach the "normal" chemical shift value as the temperature goes to infinity. The chemical shift of the compound Cs+(18C6)2.e- decreased from +89 to +74 ppm upon increasing the temperature from ~-20°C to sr+15°C. This rules out structural difference between the electride and model compounds or alkalides, Cs+(18C6)2.N—, where N- is the alkali metal anion, as the origin of the 140 ppm shift. However, if we consider the +61 ppm Obtained for the chemical shift of the Cs+ cation in the compounds Cs+(18C6)2.CS- and Cs+(18C6)2.Na' as the "normal" chemical shift of the compound Cs+(18C6)2.e- at infinite temperature then the plot of 6 yg. % should yield a straight line if the 140 ppm shift is due to the paramagnetism of the -139- sample. Indeed, the plot of the chemical shift (6) of the compound Cs+(18C6)2.e- versus % yield a straight line which is compatible with Curie-Law behavior of this compound. For a system with unpaired spins the paramagnetic effect can occur through the presence of actual electron density in an S-state at the nucleus, the so-called "contact” or "Fermi" Shift, or through space by dipolar effects which is called the "pseudo-contact" shift [121]. The "contact" shift, Ave in a paramagnetic substance of isotropic g-value is given by C AH a -8‘IT 0 II NgeggNBN 3N where AH is the local magnetic field produced by the unpaired electron at the nucleus, x is the spin-only susceptibility, a is the isotropic hyperfine splitting constant, N is Avogadro's number, ge and gN are the electron and nuclear g-factors, B and 8N are the electron and nuclear magnetons respectively and p(N) is the electron density at the nucleus. The "pseudo-contact" shift occurs whenever there is an unpaired electron in a molecule. This dipolar effect, transmitted through space and felt by the nucleus, causes a paramagnetic shift. In solution with an isotropic g-value, this dipolar effect is averaged to zero by rapid rotation of the molecules in the field. In -140- solids and in solutions with anisotropic g-tensors, the magnitude of the dipolar contribution to the magnetic field at the nucleus of interest from the unpaired electron density will depend on the orientation of the molecule with respect to the field. The "pseudo-contact" contribution to the shift, Avi(pc), at the nucleus is given by [121], 2 Av.(p) 3cos O.-1 1 _ l _1. 1 v - 3N£Xz 2(Xx+xy)]< r3 >av. i 1 sin2 Oi cos2¢i - 531—(XX-Xy) ( r3— )av. (IV.23) i where Xx’ x , x2 are the susceptibility components, and Y the angles 0 and ¢ are the polar angles of the vector ri. The last equation should average to a very small value under the condition of fast spinning of the sample at the magic angle because of the term 3 cos2 Oi-l. For an axially symmetric system, Xx is equal to Xy and the second term of Equation (IV.23) vanishes. Therefore, the "pseudo-contact" contribution to the shift should vanish for an axially symmetric system in the MASS experiment. It should be mentioned that both the contact and pseudo-contact shifts depend on the susceptibility of the sample. However, Garroway gp_§l. [131] and Stoll 23 31. [132] reported that magic angle sample spinning should remove bulk susceptibility broadening in a powder of randomly shaped particles, provided that the -141- susceptibilities of the particles are small and isotropic. On the other hand, the effects of anisotropic susceptibili- ties are 225 removed by magic angle spinning because they are analogous to chemical shift screening by nearby currents [131], since both interactions are electron- nuclear dipolar in origin. However, the anisotrOpic magnetic susceptibility tensors should be averaged to % of the trace by spinning the sample at the magic angle. The linewidth of the central transition as well as the total spectral breadth in the MASS-NMR Spectrum of the compound Cs+(18C6)2.e- are comparable to those of the model salts and the alkalide compounds with the general formula Cs+(18C6)2.X-. This indicates that line broadening due to any anisotropic magnetic susceptibility of the sample has been removed by spinning the sample at the magic angle. Since the powder pattern of the static spectrum of the compound Cs+(18C6)2.e- Shows the asymmetric profile characteristic of an axially symmetric system, the "pseudo-contact" term cannot contribute appreciably to the shift. Therefore, the 140 ppm shift in the resonance position of the compound Cs+(18C6)2.e- is mainly due to the contact shift. The presence of electron density at the nucleus requires, however, a hyperfine coupling which normally gives splitting in the EPR spectrum. In most of the cases it also is associated with an electronic g-shift. Neither hyperfine Splitting nor a g-shift is observed in the EPR spectrum -142- of the compound Cs+(18C6)2.e-. The absence of hyperfine splitting could, of course, result from fast spin exchange among the electron spins which would "wash out" the hyperfine pattern and result in a very narrow EPR line. In fact the linewidth of the EPR line of the compound Cs+(1806)2.e' is very narrow (0.5 G) and it is temperature independent. The absence of an electronic g-shift associated with spin-orbit coupling could be cancelled because of symmetry. For example, the EPR lines of gaseous alkali metal atoms reflect the presence of a large contact density at the nucleus but there is not an appreciable g-shift [133]. c. Electron doping in the ceside compound: The 133Cs-MASS-NMR at 47.24 MHz of the "pure" ceside compound, Cs+(18C6)2.Cs- (Figure 26), which was 133Cs-NMR lines. One prepared by method 2, Shows two line, which corresponds to the Cs- anion, is diamagnetical- ly shifted to -228 ppm relative to infinitely dilute CSIaq) and is broader than the line that corresponds to Cs+ in the Cs+(18C6)2 complex. At 65.61 MHz another sample of the compound Cs+(18C6)2.Cs- had an NMR spectrum that contained three lines at chemical shifts of -210, -50 and -40 ppm. These correspond to the presence of one site for the Cs- anion that gives a chemical shift of -210 ppm and two sites for the complex Cs+(18C6)2 -143- (at -50 and -40 ppm). Figure 31 Shows the MASS-NMR spectrum of the compound Cs+(18C6)2.Cs‘ at 65.61 MHz. The "pure" electride compound Cs+(18C6)2.e- shows a single 133Cs-NMR peak whose resonance position depends on the temperature and is paramagnetically shifted by the presence of a large concentration of the unpaired Spins. For the compound corresponding to reactant stoichiometry C318C6, prepared by method 3, the MASS-NMR spectrum at 47.24 MHz shows the presence of at least three peaks at +74, -52 and -242 ppm respectively (Figure 32). The peak at -242 ppm also has a shoulder at -210 ppm. At 65.61 MHz another sample (also prepared by method 3) had an NMR spectrum which contained five NMR lines at -228, -210, -52, -50 and +93 ppm (Figure 33). The three lines obtained for one sample at 47.24 MHz and five lines obtained from another preparation are consistent with the presence of a mixture of two compounds: Cs+(18C6)2.e- and Cs+(18C6)2.Cs-. Only one line (at +74 and +93 ppm) can be attributed to the compound Cs+(l8C6)2.e' while either one or two lines are present for Cs+(18C6)2 and for the Cs- anion in the compound CS+(18C6)2.Cs-. Magnetic susceptibility measure- ments of this mixed system showed that the sample contained zl7% unpaired spins. However, further MASS-NMR and magnetic susceptibility studies [134] showed that samples of this mixed system with different susceptibilities have essentially the same chemical shift values and the same number of NMR lines. In other words, the chemical -144- -—50 -2|O ~40 I I I I l 200 O '200 ‘400 “600 Chemical Shift from Cs+ (a.q.)(ppm) Figure 31: 133Cs-MASS-NMR spectrum of one sample of Cs+(l8C6)2.Cs‘ (method 2) at 65.61 MHz. -145- .umz 4N.N¢ on Am oocuoE- ImON-momH-+mu mo OHdEmm oco mo EOHuOOdm mEZImm-mzlwummH AanV-oofmu E9: :Em _ooE-mcu COM I DON I 00— I O OO— _ _ _ _ _ vn+ NNNI NVNI «nu "Nm OnsoHd -l46- ..50 I -228 [52 n - 2(0 .6. ] l T T I T 200 o -200 -400 Chemical Shift from Cs+(a.q.)(ppm) 133Cs-MASS-NMR spectrum on one sample Figure 33: of Cs+(1806)ZCs' (method 3) at 65.61 MHz. -l47- shifts are independent of the bulk susceptibility of the sample. This indicates that a mixture of the two compounds is present in the polycrystalline samples rather than mixed crystals of one type. The results are complicated, however, by the chemical shift values obtained for the Cs- line in the "pure" ceside and in the mixed system. Three values of the Cs— chemical shift were observed in the samples studied. These values are -242, -228 and -210 ppm. It is difficult to tell which one of the three values represents the chemical shift of the "pure" Cs- anion because some electron dOping was observed for every sample by EPR and magnetic susceptibility measurements. The presence of three resonance lines for the Cs- anion suggests different degrees of electron doping in the various preparations of the ceside compound. The presence of electron doping could also explain the presence of two Sites for the complex cation Cs+ (18C6)2. It would be a mistake, however, to consider the most diamagnetic shift, -242 ppm, as the chemical shift of the least doped ceside, because electron doping could cause either an upfield or a downfield shift if it Operates via a spin polarization process. In general, the electride can always be produced along with the ceside compound because of the method of preparation. Also, the ceside crystals could be doped by electrons to different extents depending on the -148- method of preparation. Until some structural information is available, the origin of the multiple lines for the Cs+(18C6)2 and Cs- will remain unclear. 3. 87Rb--MASS-NMR The rubidium nucleus has two isotOpes, 85Rb with a natural abundance of 72.8% and spin 5/2, and 87Rb with 27.2% natural abundance and 3/2 spin. The magnetic properties of these two isotopes are quite different. The 85Rb nucleus has the largest electric quadrupolar moment among the alkali metal nuclei. A typical value of 24 0.31 XlO- cm2 [121] has been reported for the electric quadrupolar moment of 85Rb. Although 87 24 Rb has an electric moment of only 0.15 X10- cm2 [121], which is very close to that of the 23Na nucleus, the natural linewidth of 87Rb is much larger than that of 23Na in solution. This difference in linewidth is due to the large Steinheimer antishielding factor [135] of the Rb+ cation compared to that of the Na+ cation. The Steinheimer antishielding factor, l-+ym, is a measure of the effective field gradient, sz, produced at an atomic nucleus as a result of polarization effects in the core electron distribution induced when the atom as a whole is exposed to an electric field gradient [135]. It is given by the equation _ O vzz — sz(1+ym) (1v.24) -l49- in which Viz is the field gradient caused by charge distributions outside of the atom. The Steinheimer antishielding factor for the Na+ cation is 6.18 while that of the Rb+ cation is 56.0 Accordingly, the 87Rb NMR line should be nearly ten times broader than the 23Na NMR line for equal asymmetries of the field at the nucleus. Equation (IV.19) for the second-order quadrupolar broadening tells us that nuclei with higher Spin and larger Larmor frequencies will give narrower NMR lines. However, the ratio of the product 0L.21(21-1) for the two Rb isotOpes is almost unity. Therefore, there is no preference of one isotope over the other, as far as linewidths are concerned. The relative sensitivity at constant field for 87Rb is, however, 12 times larger than that for 85Rb. Accordingly, all studies of rubidium NMR solutions have 87Rb rather than 85Rb. Table VII summarizes the 87Rb-MASS-NMR results for used the isotope the chemical shifts (referred to as infinitely dilute Rbtaq))and the linewidths obtained for simple rubidium salts and two alkalide compounds at 65.5 and 163.6 MHz (which corresponds to 200 and 500 MHz proton frequencies respectively). The RbCl salt shows a relatively narrow line compared to RbSCN. The latter Shows the quadrupolar splitting characteristic of the NMR powder pattern for nuclei in which the quadrupolar interaction dominates the NMR line broadening. Among the model salts and all -150- vmm Haomem 0o bmHI mmm.mmm vaI.mmHI accmem o: daemon on Hoodoo oc Hoodoo oc Hoodoo o: ooam owl III >NH+ .III 0.ol um .m>< Edd .0 an: e.mma Ill acumen 0c Ill Hoodoo 0c Ill Hoodoo 0: III Hoodoo 0c con mmal Ill Hoodoo 0: III Hoodoo 0c III Hoodoo 0: III Haomwm 0c Ill Hoodoo oc vow oma+ III 0.0T um .x>< Edd .0 mm: om.m0 -oz.-~o+nm -cm.-~o+nm uo.-~o+om (o.oOoH+sm (com-oOoeoao (cm.o0o-+s (cm.o0oe+cm (oz.o0oe+om (sm.o0oe+om -Ho.ooos+sm zom.o0oe+om zoned doom 8 - o-Hosm z H.o UESOdEOU madammm mZZImm¢Z am u 0 M 5m HH> an 8 ~151- electride and alkalide compounds, only two alkalides Show rubidium-NMR Signals. These are Cs+(18C6)2.Rb- and K+18C6.Rb-. The former shows a single 87Rb-NMR line at -l93 ppm (at 65.50 MHz) while the latter shows two peaks at -185 and -194 ppm (at 163.6 MHz). The positions of these peaks are very close to the Rb- signals obtained at -185 and -197 ppm for 0.1 M Rb+C222.Rb- solutions in ethylenediamine and tetrahydrofuran respectively [55]. 133 7 Figure 34 shows the Cs and 8 Rb MASS-NMR Spectra of the compound Cs+(l8C6)2.Rb- and Figure 35 shows the 87Rb-MAss-NMR spectrum of the compound K18C6Rb at 163.6 MHz. It is clear from the two figures that for the same number of scans the 87Rb NMR signals for the compound K18C6Rb are much weaker than the signal obtained for the compound Cs+(l8C6)2.Rb-. This indicates that the compound K18C6Rb may be a mixture of K+18C6.Rb- and Rb+1806.K', although other possible electride compounds such as 'K+18C6.e- and Rb+18C6.e- may be present also. The presence of two 87Rb signals in the NMR spectrum of the compound K18C6Rb is not surprising since it has been seen before for some Na- and Cs- compounds. However, the absence of an Rb- signal in the two compounds Rb+18C6.Rb- and Rb+C222.Rb- lg surprising since a recent EXAFS study [106] has shown that the former is indeed a rubidide. This suggests that the presence of Rb+ with a large quadrupolar coupling constant somehow "washes out" the Rb- signal or that the cation-anion interaction in these -152- 87Rb-NMR L ‘ AA '- - I ‘A-_ V VT“ 7 T—w' v—v— v “- T I I I I T 200 I00 0 -I00 -200 ~300 -400 Chemical Shift from RI)" (a.q.)(ppm) “ace-NMR A; ~ ,...._-__ 1 -- 400 200 0 -200 -400 -600 Chemical Shift from CS"’{a.q.)(ppm) l33Cs (at 65.61 MHz) and 87Rb (at 163.6 MHz) MASS-NMR spectra of the compound Cs+(18C6)2.Rb . Figure 34: -153- l I I I I IOO 0 'I00 '200 ‘300 Chemical Shift from Rb+ (a.q.) (ppm) . 87 Figure 35: Rb-MASS-NMR spectrum of the compound K+18C6.-Rb' at 163.6 MHz. -154- compounds plays an important role since an NMR signal from Rb- has recently been observed in the compound Rb+(15CS)2.Rb- [136] (but no Rb+ signal was observed). This recent observation of the Rb- in the compound Rb+(15C5)2.Rb- also provides strong evidence that the compound of stoichiometry Rbl8C6 is not the rubidide, Rb+(18C6)2.Rb-, otherwise, a signal from Rb_ would be expected. No signal could be detected from the Rb+ cation complexed by 18-crown-6 or cryptand 222. This is probably because the line broadening decreases the signal below the detection limit in these compounds is a result of quadrupolar interaction.' Dye pp 31. [55] reported that a signal of 4,000 Hz linewidth could be detected from 0.4 M Rb+C222.I solution in methanol. Since the linewidth of Na+C222 in the solid is nearly 40 times broader than that in solution, we expect the linewidths of Rb+18C6 and Rb+C222 lines to be of the order 100 to 160 KHz which is beyond the limit of detection. In summary, 87Rb-MASS-NMR does not provide species identification in electride and alkalide compounds that contain the complexed Rb+ cation but, in some cases, the Rb- peak can be seen very easily. 4. The chemical shifts of alkali metal anions Except for Na- the NMR peaks of alkali metal anions in solution [55] as well as in the solid state are -155- paramagnetically shifted from that of the gaseous alkali metal atom [124]. Although the comparison should really be made with respect to the gaseous alkali metal anion, there are no data available for the calculated chemical shifts of Rb- and Cs“. Table VIII summarizes the average chemical shifts of alkali metal anions in solution and in solids referred to the gaseous alkali metal atoms. Table VIII: Chemical Shifts of Alkali Metal Anions Alkali metal anion Solutiona Solid Na— :-2 NO Rb' :+20 :+21 Cs' :+52 ::+ll7 aData taken from reference 55. Paramagnetic shifts usually indicate that there is an interaction between molecules or ions in the vicinity of the ion and the outer p-Orbitals of the alkali metal species. A shift for M- is somewhat unexpected since the alkali metal anion has the n52 configuration that should provide shielding of the outer (n-l)p-orbital. An explanation for the observed shift is as follows: The chemical shift of the alkali metal anion increases with increasing atomic number. This suggests that, except for Na-, the ground state of the alkali metal anion is not purely an s-State and that mixing of states to involve p- and d-states is possible. The interaction of the -156- np-orbital of the ground state with the environment will result in a paramagnetic shift that will depend on the degree of mixing. CHAPTER V MAGNETIC PROPERTIES OF ALKALIDES AND ELECTRIDES A. Electron Paramagnetic Resonances (EPR) 1. Introduction In the electron paramagnetic resonance experiment, the degeneracy of the spin (Zeeman) energy levels is removed because of the coupling of the external magnetic field, H, to the magnetic moments of the unpaired spins. Transitions between the Zeeman levels can be induced by radiation Of the apprOpriate frequency, v. For non- interacting (isolated) spins the spin Hamiltonian, Hg, is given by JCS = -gBH-S (v.1) where S is the electron spin operator and g is the electronic g-factor of the system. Since the electron has a spin of 1/2, this leads to the transition energy AB = hv = gBH (v.2) -157- -158- which gives rise to a single EPR line that occurs at a magnetic field (resonance field) HO equal to hv/gB. When the electrons interact with nearby nuclei, such that there is an actual electron density at the nucleus in an S-state because of this interaction, the total Hamiltonian of the system is given by [137] 16 = -gBH-§+EgN8NH-Ij +£71.35— (V.3) J J J 3 where Ij and Aj are the spin operator and the hyperfine coupling constant for the jEp_nucleus respectively. The first two terms in Equation (V.3) represent the electronic and nuclear Zeeman interactions while the third term is the "Fermi" contact term which represents the nuclear- electron interaction. The magnitude of the isotropic hyperfine coupling constant in Gauss is given by _ 2 Aiso(G) - (aw/3IgN8NIw(0)I (v.4) where [47(0)]2 is the normalized time average probability that the electron and its associated magnetic moments are at the nucleus. A more accurate equation for the Hamiltonian [Equation (V.3)] should replace the Scalar quantities Ais and g by tensor quantities and should 0 include an additional term which describes the quadrupolar interaction which will affect the electron-nuclear energy levels. Two cases can be considered to obtain the eigenvalues of the Hamiltonian described by Equation (V.3). First, in the high field limit where -159- the hyperfine coupling constant is only a small fraction of the applied magnetic field, i.e. Aiso <H (low field case) and Aiso SH (intermediate field case), the electron and nuclear moments are tightly coupled to give a resultant total spin angular momentum F(F==§+I) in which its -160- z-prOJectlon, mF, ls glven by: mF==ms+mI. Brelt and Rabi [138] gave the eigenvalues of the Hamiltonian in Equation (V.3) in the low and intermediate case as: — 4m 8 _ _ AW AW F 2 E(F"“F) ‘ 2(21+l) +gNBNHomF i 2 j +21+l “(J (v.8) where AW is the zero-field splitting and is given by (H=O)-EI_S(H=0) ’ EI+s AisogB(2I+l) = 2 (v.9) and x==H0(gB+gNBN)/AW. The positive Sign in Equation (v.8) goes with F==I+S and the negative sign goes with F =I-S. Figure 36 shows the energy level diagram (taken from reference [133]) for the system with I==§ and S==% in the high and low field cases. The selection rule for absorption is AmF==l, so that the transition energy is given by AW ”is?" 2 5 mg" 2 ’5 AB = gNBNHO +7 [1 +m+x ] + [l +m+XJ (v.10) where m: and m; refer to the upper and lower states respectively. There are two important EPR parameters that determine the resonance position and in some cases the shape of the EPR line. These are the electronic g-factor and the -161- 1.4 -‘V%KQV T' he 5° -4 \ I \ - \- \ \ K \ \ \ \ -\- \ \ V \ \ ‘V‘ \ \ \--\ \ \ \ \ c: I {I ’\ .2 --‘----\- l\ I\ I \ I \I \ 1" I \ I \ I I I I I I I I--I- I L / - -’— I I‘("" -'---.--....- -].4 - l l 0 (15 1.0 III 241 X: C 9.- 9')#H/AW Figure 36: Magnetic energy levels for I =3/2 and S =l/2. The solid lines are energy levels computed from Eq. (v.10); the dotted lines are the corresponding first order energy levels (Eq. (V.7)) [133]. -162- hyperfine splitting constant, A. The electronic g-factor describes the interaction of the electron with its environment and is given by, = 3H+S(S+l)-L(L+l) 9 2 2J(J+1) (V.ll) where S, L and J are the spin, orbital and total angular momentum quantum number respectively. Therefore, the g-factor is a measure of the mixing of spin and orbital angular momentum for the electron giving the EPR signal. For a system where the orbital angular momentum has no contribution to the total moment, i.e. spin only angular momentum, S is equal to J, therefore 9 is equal to 2. A more accurate value for the free electron g-value, ge, is 2.0023193 [139]. Deviation from the free electron g-value is usually interpreted in terms of spin-orbit coupling because of the interaction of the electron with its environment. In dilute solution of low viscosity the g-factor is a scalar but in a single crystal or powder the g-factor depends on the orientation of molecules with respect to the field and then the g-factor is a tensor. In general, anisotropy in the g-factor arises from coupling of the spin angular momentum with the orbital angular momentum. The spin angular momentum is oriented with the field, but the orbital angular momentum, which is associated with the electrons moving in molecular orbitals, is locked to the molecular wavefunction. -163- Anisotropy in the hyperfine coupling constant, A, arises from the dipolar interaction between the electron and nuclear moments which depends on the electron- nucleus separation and the orientation of the electron- nucleus vector with respect to the magnetic field. Anisotropy in the hyperfine interaction results in line broadening of the hyperfine components and hence mI dependence of the hyperfine lines. Line broadening in EPR arises mainly from the interac- tion of the unpaired electrons with the thermal vibrations of the lattice (spin-lattice relaxation). In addition, there are sources of line broadening that can be divided into two main groups. These are inhomogenous and homogenous broadening. In the first case, the unpaired electrons in the sample are subjected to slightly different effective magnetic fields. The observed line is then a superposition of a large number of individual components, each slightly shifted from the others. In the solid state, the anisotropic interactions in a randomly oriented system result from the anisotropic g and A and give rise to inhomogeneity. In addition to these sources of line broadening, the presence of unresolved hyperfine interaction because of a large number of hyperfine components or because of weak electron—nuclear coupling will result in inhomogeneous broadening. The lineshape for an inhomogeneously broadened EPR line is Gaussian. -l64- Homogeneous broadening arises from magnetic dipole interactions between the unpaired Spins which depend on the relative orientations of the dipoles with respect to the magnetic field as well as the separation between the interacting dipoles. Another source of homogeneous broadening is electron-spin exchange which is viewed as a bimolecular reaction in which two unpaired electrons exchange their spin states. If the electron spin exchange occurs at a rate comparable to or faster than the hyperfine frequency then a time averaged broad signal will be observed. For systems in which the electrons exchange their spins very rapidly, (normally systems of high spin concentration), the time averaged hyperfine field as well as the field produced because of magnetic dipole interactions are averaged to zero and a sharp, narrow EPR line is observed. This case of electron spin-exchange is called "exchange narrowing" and is observed for pure solid free radicals [139]. The line- shape of a homogeneously broadened EPR line is Lorenzian. One characteristic difference between inhomogeneously and homogeneously broadened EPR lines is the saturation behavior of the EPR line as a function of the microwave power, Po' which is shown in Figure 37. For a homogeneously broadened line, the EPR signal amplitude increases with increasing PO until it reaches a maximum and then decreases. On the other hand, the inhomogeneously broadened line's -165- Q g — — — — -_ _ —I- C“ .2 ”" ' E // lnhomogeneously broadened line (U '3 C .9 a: m: 3 Homogeneously broadened line j/Po Figure 37: Variation of EPR signal amplitude with microwave power (PC) for a homogeneously broadened line (solid curve) and an inhomogeneously broadened line (dotted curve). ~166- amplitude first increases with increasing PO and then levels off. One final point is worthy of mention concerning the factors that affect lineshapes and the positions of EPR lines. This is the interaction of the electron with a nucleus that has a large quadrupolar moment (I 11). This has the effect of broadening the EPR lines as well as affecting the separation between the hyperfine lines in two ways: (1) a displacement of the energy level, and hence the resonance field, by a constant amount and (2) a change in the separation of the energy levels that causes the spacing between adjacent EPR lines to be greater at the ends of the spectrum than in the middle. However, in some cases, the quadrupolar interaction also causes the appearance of additional lines that are normally forbidden by the selection rule AmI==0 [121]. The paramagnetic resonance observed in metals is called conduction electron spin resonance (CESR) and was first observed by Griswald, Kip and Kittel [140]. Dyson [141] and Kip [142] treated the problem of CESR theoretically by using a highly idealized model of electrons in the metal. The electrons are assumed to diffuse as free particles, and the electron magnetic moments are treated as free particle moments. The lineshape of the CESR signal is highly unsymmetrical and is determined by the skin depth, 6, the diffusion time, TD, which is the time it takes an electron to diffuse -167- through the skin depth, the electron Spin relaxation time, Tr,the spin-lattice relaxation time, T2, (for metals T1==T2) and TT’ which is the time it takes an electron to traverse the sample. Two main cases were considered in the lineshape calculation. These are the "classical skin effect" and "anomalous skin effect". In the first case, the skin depth, 6, is large compared with the electron mean free path, A. In this case, the "classical" skin depth, 6, is given by [141], 2 8 5 = [ifiaa] (v.12) where o is the microwave conductivity of the metal at frequency w and c is the speed of light. If 6 is not large compared to A we have the "anomalous skin effect" and the penetration of the microwaves is controlled by A. Dyson [141] treated the case for which TTI>I>TD and TTI>>T2 which can be applied to a thick plate metal sample whose thickness is large compared to the skin depth. In this case the EPR is highly unsymmetrical with 2.7 ma H0309 m>mzouoflE may mo uoou mumsqm map nufiB Imz.NAmomHV m0 m0 Eduuommm mmm mnu mo mafia Hmuucmo mnu mo wuwmcmucfl m>aum~mu map mo cOHuMHHm> “av musmflm sci It. 2:. 8m 8w 2: _ _ d a... ( mm 5mm”) A“ BNILNI a O — BAILV'IBH -l76- the central line and very weak structure on both sides of the line. Increasing the temperature gives a decrease in the relative intensity of the central line and a more resolved structure appears. Figure 42 shows the EPR spectra of a polycrystalline sample of Cs+(l8C6)2.Na- at different temperatures. The crystalline sodide, Rb+18C6.Na- shows a rather interesting powder EPR spectrum as shown in Figure 43. The spectrum consists of a central line, six hyperfine lines that correspond to a strong coupling to 85Rb and four hyperfine lines that correspond to a coupling to 87Rb. The central line has a g-value of 2.0023 i0.0001 and a linewidth, AHPP, of 4.63 $0.01 G. Both the g-value and the linewidth are temperature independent over the temperature range of 2.9 K to 248 K. The constant value of the linewidth of the central line over the whole temperature range and the saturation behavior of this line (Figure 44) indicates the absence of any inhomo- geneous broadening so that this line represents essentially a single site of electron trapping in Rb+18C6.Na-. Some of the ten hyperfine lines overlap but the dependence of the linewidth on mI is clear. The separation between adjacent hyperfine lines of a given rubidium isotope increases with increasing magnetic field: i.e. the separation is less at low field that that at high field. -177- T=143 K Gain = 32 x105 T: 212 K 5 Gain: 3.2 x 10 T: 253 K Gain: 5 x10 Figure 42: EPR spectra of a polycrystalline sample of Cs+(18C6)2.Na’ at different temperatures. -l78- .x m.m um utz.eomfl+hm mo oHQEmm mcflaamumwnowaom M NO Esnuommm mmm "me musmam a as... 9.; En... - £5» can...” = 3.... a 81mm _ r i .2; 9.3 u 2.5 “:3 :5 I name — I “New a~=v 08—. l‘ha 9.8 _ . 2:... at: -179- .Hm>oa Hm3om m>m30uowa on» :Ufl3 Imz.moma+nm Mo Eshuommm mmm mnu mo mafia Hmuucoo on» no wuflmcwucfl o>Humaou mcu mo coflumwum> As; It. 2— "vv Guzman an. an Aumq: v) (swan .KLIS.‘J3.'.NI BAILV'IEU -180- The extraction of the EPR parameters from this spectrum is very difficult because of the overlap of the hyperfine lines of the two rubidium isotopes and the further Splitting observed for the hyperfine lines at low field. However, if we use isotropic g-values and hyperfine interactions and compare the resonance fields of some of the hyperfine lines with those obtained by complete Breit-Rabi analysis using Equation (V.10) at a given 9 and Aiso’ some of the EPR parameters may be obtained. Table IX summarizes the observed and calculated resonance fields of some hyperfine lines (labeled by their m1) of 85Rb and 87Rb. The observed g-value, ghf’ Table IX: Resonance Field of Some Hyperfine Line 8 85 ghf SAiSO Rb(-§-) 85Rb(-%-) 85Rb(-%-) observed 1.9974 58.8 3553.7 3489.9 3431.7 calculated 1.9980 60.0 3559.8 3495.2 3431.8 87 8 8 7 3 ghf Aiso Rb(-2) observed 1.9974 199.6 3711.1 3489.9 3083.1 calculated 1.9980 200.0 3706.4 3486.0 3083.1 8 7 l 7 l Rb(‘§0 Rb(§) is calculated from the Spectrum at the field corresponding to the center of the hyperfine pattern, while the observed Aiso values for both isotopes are calculated from the average value of the estimated separations between two adjacent hyperfine lines of each isotope. AS shown from Table IX, the agreement between the observed and calculated -181- resonance fields of the hyperfine lines is not very good. This indicates that a Breit-Rabi analysis is not sufficient to explain the Rb+l8C6.Na- Spectrum and that anisotropy in both 9 and A should be considered. However, the 87 85A ratio between the observed A. and . va u is 150 150 1 es 3.40 which is in agreement with the expected ratio of 3.38, based upon the isotropic hyperfine Splitting constants of the neutral atoms of each isotope [121]. The electron density IMO)!2 at the rubidium nucleus which gives rise to the hyperfine splitting can be calculated by using Equation (v.4). This calculation 24 yields a value of 2.5767 xlo electrons-om‘3. Comparison Of this value with the Ith(0)|2 value obtained for the neutral rubidium atom in the gas phase, 15.8225 XlO24 electrons-cm.3 [133], gives 16.3% atomic character at the rubidium nucleus in the compound Rb+l8C6.Na-. This value suggests that electron trapping occurs at a Site that allows substantial interaction with the Rb+ cation, perhaps near a defect Rb+l8C6 cation. On the other hand, the single narrow line at g = 2.0023 indicates substantial electron localization in a site at which the electrons are very weakly interacting with the surroundings, perhaps at anionic site vacancies. One sample of Rb+18C6.Na- did not show the EPR hyperfine pattern but rather an EPR spectrum that was similar to that of CS+(18C6)2.Na- which was discussed before. -182- In summary, electron trapping (or doping) in alkalide salts is unavoidable because of the nature of preparation of these compounds from amine and ether solutions in the presence of the complexing agent. These solutions are rich in the solvated electron which can be trapped during the crystallization process. In general, electron trapping in cryptated sodide salts is less extensive than in 18-crown—6 sodides. This may be related to differences in the structure as well as the cation-complexant size in these compounds. One might consider two kinds of electron trapping that can occur in the alkalide salts. These are trapping at anionic vacancies (F-centers) and trapping at interstitial Sites near a defect cation or missing crown ether. The first kind may be responsible for the narrow EPR lines obtained for all the sodides at g==2.0023 as well as the broader lines superimposed on the narrow line. The difference between the narrow and broad lines obtained in most of the sodides may then be due to trapping at anionic vacancies which provide different broadening mechanisms. The second kind of electron trapping is responsible for the hyperfine pattern obtained for the Rb+18C6.Na- and may be referred to as a Frenkel impurity state [133]. -183- Electrons in electrides: 9.. Three crystalline electrides were studied by EPR spectroscopy. These are Rb+18C6.e_, CS+C222.e— and Cs+(18C6)2.e-. The first two electrides are essentially diamagnetic (singlet-ground state) while the third is paramagnetic (doublet-ground state). Accordingly, the EPR spectra of these compounds are classified according to the multiplicity of the ground state of the compound. i: EPR of singlet-ground state electrides: Since the two crystalline electrides, Rb+18C6.e- and CS+C222.e- are diamagnetic (see Section 82), their EPR Spectra result from single electron trapping or anion defects (since in this case the anion is the trapped electron pair). -Figure 45 shows the EPR spectrum of a polycrystalline sample of Rb+18C6.e-. The spectrum consists of a single Lorentzian line as indicated from the saturation behavior of the line (Figure 46). The EPR line has a g-value of 2.0017 £0.0001 and a linewidth of 4.7 $0.3 G. Both the g-value and the linewidth are temperature independent over the temperature range of 110 to 243 K. The % ratio is also temperature independent over the same temperature range with a value of 1.1 $0.1 indicating high symmetry of the EPR line. However, the intensity of the EPR decreases with increasing temperature. A plot of the relative intensity of the EPR line (arbitrary units) versus i is linear, indicating Curie-behavior of T the unpaired Spin in this system. However, the small -l84- 50 ~ Figure 45: EPR spectrum of a polycrystalline sample of Rh 18C6.e‘ at 123 K. -185- .Ho>ma Ho3om o>m30H0flE may no uoou mumsvm ocu nuaz o.oUmH am mo mafia mmm on» no xuwmcoucfl m>flumHmu on» m0I20aumem> A2; LC. S— 23 2: es 2: 2: "0v musmwm . . _ _ . Mummy) BAILVWSH (svun AllSNBLNI -l86- difference between the observed g-value and that of the free electron suggests that the electrons are only weakly interacting with their surroundings. The compound Cs+C222.e- shows an EPR powder pattern that corresponds to an axially symmetric system as shown in Figure 47. The EPR signal is very weak and could only be detected at liquid helium temperatures (2.8-20 K). Above 20 K the signal-to—noise ratio was very poor and no signal could be detected. The value of 91 is 2.0020 and that of gll is 1.9771, both of which are less than that of the free electron. This suggests that there is some interaction between the electron and CS+C222, probably present as the "exclusive complex cation". Although, it appears from the spectrum that only one site of the trapped electron exists since only a single line is observed, the saturation behavior of the sample Shows a behavior characteristic of an inhomogenously broadened line as shown in Figure 48. This indicates that there is a narrow distribution of trapping sites for electrons in this compound. ii: EPR of doublet-ground state electride: The electride, Cs+(18C6)2.e-, shows a conduction electron spin resonance (CESR) spectrum similar to that observed for a spherical metal particle whose radius, a, is comparable to the skin depth, 6. Figure 49 shows the EPR spectrum of a polycrystalline sample of Cs+(18C6)2.e-. The EPR Signal is very intense so that the Signal could be detected at very low microwave powers (2200 nW). The -187- 91 206 9“ Figure 47: EPR Spectrum of a polycrystalline sample of Cs C222.e' at 2.8 K. INTINOITV (Atbmuy Unlu ) IILATIVI -188- 80- m c: I m o l .5 o T 30.. 20- 10_ l L I 10 20 30 1/2 {7’ (uw) Figure 48: Variation of the relative intensity of the EPR line of CS+C222.e’ with the square root of the microwave power level. -189- Figure 49: EPR spectrum_of a polycrystalline sample of Cs+(18C6)2.e at 174 K (sample in 2 mm tube). -l90- spectrum consists of a single-asymmetric line with 1<<§~<2.' The temperature dependence of the linewidth Apr, of the EPR line is shown in Figure 50. The line- width is remarkably independent of temperature over the range 2.9 to 254 K and has a value of 0.45 10.05 G. The g-value (corresponding to the field at which the first derivative crosses zero) is 2.0023, also independent of temperature. The g-ratio increases with increasing temperature, Showing a greater asymmetry at higher temperatures. However, it was noticed that at a given temperature the % ratio increased as the radius of the sample tube decreased. This indicates that the lineshape of the EPR line of Cs+(18C6)2.e- may depend on the microwave resistivity of the sample, which decreases in the smaller EPR tube because the sample is more tightly packed. Figure 51 shows the variation of the % ratio with temperature for a sample in a 3 mm tube. The relative magnitude of the A.C. microwave resistivity as a function of temperature was estimated from the data shown in Figure 51 by using the graph of % versus 9 (shown in Figure 52) given by Glaunsinger and Sienko [144]. The value of the particle radius, a, was assumed to be independent of temperature. The resulting relative resistivities, p/p , are plotted ref. versus temperature in Figure 53. The dashed line shows the D.C. resistivity over the same temperature range (203-260 K). The resistivity at T==203 was taken as -l9l- .musumuwmaou nuflz no.mioomaa+mo so does mom as» so Aouosaa shoasosaa one mo soasmasm> “om musmflm (All. ._. 8N OWIN. CNN DAWN om. 09 ON.— ONW. Omv. 0% CW 0“? O.N .. ..N.O . i do m. 9.314 .. o o o o o o o o o o o. v0 0 m 0 O O O O O . 000009000000000 000000000000 400 -l92- N .Amnsu SE M GA meEmmv musumummfimu £Ufl3 no. AoomHv+mU mo mafia mmm mnu mo Oflumu m\< mo coflumfium> “Hm ousmflm (I... O®N OWN ONN OON Om_ Om. 01w. O.N_ 00. -ON._ a . oi L Om._ .. ON. 1 cm. 61/6 10.0 -l93- [1 2.0 1 1.0 I I I I T] 0.5 l 0.2 ._ .. 0.1 l l I L J 1 l l 0.2 0.6 1.0 1.4 1.8 A / B Figure 52: a/6 XE' A/B for spherical metal particles. |.O 0.9 0.8 0.2 O.l Figure 53: -194- r I _I I I bl I I "I I I 'I I I - I I I ‘- I I I — \ \ \ \ " \ \ \\N I' \\ ‘~.~~. 1 ll 1 I, l l, 1 -T-'T‘"i--;...u_¢_.4__4 220 240 260 W) Plot of the relative A.C. resistivity (solid line) at x-band frequency, microwave frequency and D.C. restitivity (dashed line) gs. tempera- ture. -l95- the reference value. The plot of IUIp/p versus l/T is ref. shown in Figure 54. It is linear and yields an apparent band-gap of only 0.1 eV compared with 0.9 eV from D.C. conductivity. It would be interesting to make corresponding measurements with single crystals in order to determine whether this difference in band—gaps arises from intergrain resistance and/or anisotropy of the electrical conductivity or whether it is an intrinsic frequency dependence of the specific impedance of the compound. It would also be of great interest to obtain the diffusion time, T as a function of temperature by D' fitting the spectra to the theoretical expression given by Webb [143], Equation (v.14), and also to obtain the relaxation times (T1,T2) of the electrons in this system. The extremely narrow, temperature independent EPR line obtained for this system suggests that the electrons exchange their spins very rapidly even at liquid helium temperatures. Since in the case of "exchange narrowing" T1 is equal to T2, it can be calculated from the relation T1 = 1.15/yeApr (v.16) in which Ye is the magnetogyric ratio of the electron. The calculated value of the relaxation time is 1.45 ><10.7 sec which is only an order of magnitude larger than the relaxation time obtained for spherical lithium particles [144]. -l96- J l l i J l l I 4.3 4.4 4.5 4.6 4.7 4.8 4.9 5.0 ..L.. 3 Figure 54: A plot of log relative microwave resistivity with l/T. -l97- The temperature independent-free electron g-value obtained for the compound CS+(18C6)2.e- is surprising because most systems which Show CESR yield a negative g-shift. This was discussed before in Chapter IV- Section C-Z-b and the argument was used to explain the absence of an apparent g-Shift in the presence of high symmetry. Another possibility that may be discussed here is a small electron density at the cesium nucleus and hence a very small atomic character as the reason for the absence of a g-shift. The electron density at the cesium nucleus can be calculated by using Equation (IV.22). By replacing‘%% by the 140 ppm obtained for the shift in 133 the resonance position of the Cs-NMR line of the Cs+ in CS+(18C6)2.e-, the calculated electron density at the 21 3 cesium nucleus is 2.39 X10 electrons-cm- . The value of the electron density at the cesium nucleus of the free cesium atom is 2.65 X10+25. This yields.:0.01% atomic character at the cesium nucleus in the compound Cs+(18C6)2.e'. If we compare the g-value of the compound Cs+(18C6)2.e- with that of the free cesium atom, 2.00258 4 [139], then a negative g-shift of 2.8 XlO- is obtained. This Shift is very small compared to the 104.5 ><10a4 obtained for frozen solution of cesium in HMPA [41] (80% atomic character). Accordingly, the small atomic character and hence small electron density may be the reason for the absence of g-shift (relative to the free electron value) in Cs+(18C6)2.e-. -l98- B. Magnetic Susceptibility 1. Introduction The most natural way to classify the magnetic proper- ties of a material is by its response to an applied magnetic field. This response is characterized by the susceptibility x, through the relation [145] M = xHO (v.17) where M is the magnetization, or magnetic moment per unit volume, and Ho is the applied field. In general, x is a function of both HO and the temperature, T. If the material is magnetically isotropic, M and Ho are parallel and x is a scalar: for anisotropic materials, x is a tensor. The magnetic moment of a free atom arises from three principle causes [146], the spin with which electrons are endowed, their orbital angular momentum about the nucleus, and the change in the orbital moment induced by an applied magnetic field. The first two effects give paramagnetic contributions to the magnetization, and the third gives a diamagnetic contribution. Accordingly, diamagnetism arises from field-induced electron circulation of electrons, which generate a field opposed to the applied magnetic field. Thus, all molecules have contributions from diamagnetic effects. The diamagnetic susceptibility -l99- of an atom, XA' is proportional to the number of electrons, and the sum of the squared values of the average orbital radius of the ith electron, F; [121]: 2 n Die 2 X = —7 Z r. A 6mc i 1 n = -2.83 x1010 2:2- (V.18) Large atoms with more electrons have greater diamagnetic susceptibilities than small atoms with fewer electrons. The molar diamagnetic susceptibility, Xd' for a molecule can be obtained by summing the diamagnetic contributions from all of the atoms, XA' and from all of the bonds in the functional groups, XB: Xd = prn.+ ng. (V.l9) 1 1 3 3 Values for XA and XB are referred to as Pascal's constants. In general, diamagnetic materials have small negative temperature-independent molar susceptibilities with magnitudes of the order of 10-5. Paramagnetism occurs only in materials in which individual atoms or molecules have permanent magnetic moments. The molar susceptibility is positive and temperature-dependent; it is of the order of lO-Z‘at room temperature and varies approximately as l/T. For a system with isolated non-interacting spins, the molar -200- electronic susceptibility may be described by the Curie Law [145]: e C XM = W " f ‘V'ZO’ B where N is AvogadroIs number, 8 is the Bohr magneton, k is the Boltzmann constant, C is the Curie constant and B P is the effective number of Bohr magnetons P s g[J(J+l)]% (v.21) in which J is the total angular momentum quantum number. For a system with a "quenched" orbital angular momentum, P becomes equal to g[S(S+l)]is where S is the spin quantum 3 1 number and the Curie-constant becomes 0.37604 cm K mole- for a mole of free spins. This kind of magnetic behavior can be explained as a consequence of two opposing effects [146]: first, the tendency of the applied field to orient the moments in the direction of the field, and second, the tendency of thermal agitation to preserve a random orientation of magnetic moments. For systems in which the electrons are weakly interacting so that there is no direct coupling between the spins, the internal interaction in the material tends to align the Spins. This interaction is electrostatic. and is related to the overlap of charge distribution [147]. Such interaction was first postulated by Weiss and it was -201- shown later by Heisenberg that it can be described as a result of quantum mechanical exchange interactions [147]. The exchange energy of two electrons can be written as -2J12(Sl-SZ), where J is the exchange integral and is positive for a ferromagnetic system and negative for an antiferromagnetic system. The electronic susceptibility at sufficiently high temperatures (T >>I0]) can be expressed by Curie-Weiss Law x; = TEE (v.22) where O is the Weiss-constant and is a characteristic of the material. For ferromagnetic behavior 0 is positive and usually, but not always, negative for antiferromagnetic behavior. Paramagnetism in metals results in a small, positive temperature-independent susceptibility. Pauli [146] explained this phenomenon by the application of the Fermi- Dirac distribution. In this case, only a fraction T/TF (where T is the Fermi temperature) of the total number F of the electrons contribute to the susceptibility. The net magnetic susceptibility for conduction electrons after making correction for diamagnetism is given by k—Tr' (v.23) -202- 2. Results and discussion a. The electride, Cs+(18C6)2.e-: Magnetic susceptibility measurements were carried out on a polycrystalline sample of Cs+(18C6)2.e- in the tempera- ture range of 1.7 K to 250 K. Figure 55 Shows a plot of the reciprocal of the molar electronic susceptibility, x; versus temperature. The electronic susceptibility is the measured susceptibility of the fresh sample minus that of the decomposed sample [see Equation (11.3)]. The relationship is very nearly linear and shows temperature dependent paramagnetism. The data shown in Figure 55 were fit by the equation f c e = —§— (v.24) XM Te where fl is the fraction of the electrons contributing to the susceptibilty and is given by fl==n/Nav, where n is the number of unpaired electrons in one mole of the compound. The data nearly fit the Curie-Weiss Law Equation (v.24) with fl = 0.736 i0.009 and O =-l.44 £0.05. However, deviations from this equation are systematic. An excellent fit to the total molar susceptibility (susceptibility minus that of the container) can be obtained by adding a simple Curie-Law term and diamagnetic term to give -203- m0 oHQEmm mcHHHmumzuowaom o How 9 .mM MWQH mo uoH OmN .41 . e.mxoomae no 4 "mm musmflh .08 .8... Wulom .om _ oo boom ~204- flC fZC XM = W+T+B 07-25) The new fitting parameters and their standard deviations are f1=0.595:0.005, 0=-4.5i0.2, f2 and B==-(310 :40) Xl0-6. The value of B is the same as =0.262 $0.006 the (temperature-independent) molar susceptibility of the decomposed sample, indicating that there is no diamagnetic electronic contribution to the molar susceptibility. The sum fl+f2 is 86% of the value expected for a mole of unpaired spins, suggesting that the sample probably contained some decomposition product. Alternative— 1y, the Curie-Weiss Law might not be applicable over the entire temperature range. The small and negative value of 0 suggests that the electrons are very weakly interacting and the sample exhibits antiferromagnetic behavior. b. The ceside, Cs+(l8C6)2.CS-: Dheeb Issa [104] prepared the crystalline compound of stoichiometry C518C6 by using method 2 and carrying out magnetic susceptibility measurements on polycrystalline samples of this compound. In the present work the magnetic susceptibility measurement of the compound Cs+(18C6)2.Cs-, prepared by method 2, was repeated and compared with that obtained for the same compound prepared by method 3 (see Chapter III). The molar electronic susceptibility data were fit by the equation -205- x3 = ——+B (v.26) The fitting parameters of the Curie-Weiss Law [Equation (V.26)] of three independently prepared samples precipitated from 2-aminopropane-diethylether mixtures (method 2) and another sample precipitated from dimethyl- ether-trimethylamine mixtures (method 3) are summarized in Table X. The results of the first three samples of the compound Cs+(18C6)2.Cs- confirmed the presence of trapped electrons which interact only weakly with one another since small-negative 0 values are obtained for these three samples. As expected for electron trapping at anion vacancies during crystallization, the value of f varies somewhat from one preparation to another. By contrast, sample 4 showed that 17% of the electrons are unpaired. The results obtained here are consistent with the 133 Cs-NMR results obtained for sample 4, which confirmed the presence of mixtures of Cs+(18C6)2Cs- and Cs+(18C6)2.e- rather than crystals of Cs+(18C6)2Cs- with all electrons trapped at anion vacancies. The diamagnetic contribution to the electronic susceptibility of Cs+(18C6)2.Cs-(B) is interesting. Since this is obtained from the difference between the susceptibilities of fresh sample and that of the decomposed sample, it is difficult to determine accurately, particu- larly if any of the sample is lost during decomposition. -206- Table X: Parameters of the Curie-Weiss Equation for x; of Cs+(l8C6)2.Cs- Method of 2 6 Sample Preparation 10 f 0,K 10 B 13 2 1.4 ro.1c -2.0 l -46 :4 2a 2 2.01 r0.02 -1.1 0.1 —36 :1 3b 2 2.13 to.02 -1.2 0.1 -73 :2 4b 3 17.0 10.6 -1.5 0.2 -4o :5 aData were taken by Dr. Ramanurthi Janakiraman. bThiS work. cStandard deviation estimates from the fit of each data set independently. 6 6 The values obtained, ranging from -36 x10’ to -73 x10' per mole, presumably represent the diamagnetism of the 6s electron pair of Cs-. For all four samples the value of 0 ranges from -1 to -2 K. This small, negative value shows that antiferromagnetic interactions of the trapped electrons (if any) are very weak. Figure 56 Shows the electronic molar susceptibility of Cs+(l8C6)2.CS-, prepared by method 3, as a function of temperature. c. The di-electrides The prefix, di-, used here to describe the electride does not necessarily imply dimer electrides, rather it describes the ground state of the electride as a diamagnetic state. Two crystalline electrides fulfill this criterion. -207- Imo.maoomav+mo mo onEmm ocHHHmumwuomaom o How xL 00. I . cm oosuoé B .m> Wx no uon d On OON On. A q 0 C 'l-|'l'l'l'l'llr-l'l 'l' 'l .l'l'l'lll'l'l-l 0?. 0. ON 0? 00 GO ON Om "om ousmflm “I -208- These are Rb+18C6.e- and Cs+C222.e-. Figure 57 shows the variation of the molar electronic susceptibility of a polycrystalline sample of Rb+18C6.e- as a function of temperature. In the temperature range of 1.7 K to 120 K the electronic susceptibility shows a temperature dependent paramagnetism. Above 120 K the susceptibility is negative and nearly temperature independent. The data shown in Figure 57 were fit by Equation (v.26). The fitting parameters of the Curie-Weiss equation for Rb+18C6.e- are: fl: (0.85 ro.01) xlo’z, 0=+o.15 20.05 and B =(-18.7 $0.7) X10-6. Only 0.85% of the expected number of the unpaired spins are present which indicates that a substantial temperature-independent spin pairing is present in this system. The results indicate that the compound Rb+18C6.e- is essentially diamagnetic with some electron doping as an impurity. The crystalline compound CS+C222.e- Shows a negative molar susceptibility in the temperature range 1.7 to 225 K. However, the molar susceptibility becomes more negative upon increasing the temperature. This temperature- dependent diamagnetism can be explained as follows: the molar susceptibility of the compound CS+C222.e‘ is the sum of a very small positive paramagnetic contribution which is temperature-dependent and a large contribution from the negative temperature-independent diamagnetism. The net magnetization results in a temperature dependent negative susceptibility. Figure 58 shows the variation of -209- 150 III» 100 x2103 so; 300 ‘00 200 T(K) Figure 27: A plot of xfi KE’ T for a polycrystalline sample of Rb18C6.e‘. -210- x; of a polycrystalline sample of CS+C222.e- as a function of temperature. However, because of the violent thermal decomposition of the sample at room temperature no data could be obtained for the decomposed sample so that the electronic susceptibility could not be calculated. Also, no accurate measurements of the mass of the sample could be obtained because of the loss of part of the sample during the thermal decomposition. Accordingly no fitting of the data could be done. It is of interest to know why a system such as Cs+(18C6)2.e- is paramagnetic while systems such as CS+C222.e- and Rb+18C6.e- are diamagnetic. Spin-pairing processes are of course common for electrons in alkali metal-ammonia solutions [1] as indicated from static and microwave (EPR) susceptibility measurements [1]. Electrides prepared by solvent evaporation [96-98] often Show substantial Spin-pairing as indicated from magnetic susceptibility measurements. The reason for such spin pairing in these systems is not clear but Cyrot [148] and others pointed out that when electron correlations are taken into account, the singlet ground state is favored over the paramagnetic state. ~211- -4o -60 O -80 . ”S -1003 u°z 0 O '0 -I20 I- O 0 OK 01. O (D O .140 - ' " ° ' ° I 1 I I I 50 100 ISO 200 250 T(K) Figure 58: A plot of x3 vs. T for a polycrystalline sample of C315222.e‘. CHAPTER VI CONCLUSIONS AND SUGGESTIONS FOR FUTURE WORK A. Conclusions Crystalline electride and alkalide salts can be prepared as stable compounds in 33339 at low temperatures. The alkalide salts have greater stability than the electrides. Solutions of the alkali metal and complexant in the solvent dimethylether were shown to have much greater stability than those in ammonia and methylamine and therefore provide more stable and clean crystals. The stabilization of the solution by adding lithium metal to methylamine solutions of alkali metal and complexant is a general phenomenon and all the crystals prepared were essentially free of lithium. The new alkalides, Rb+18C6.Na- and Rb+18C6.Rb- are stable at room temperature for days. Analyses and properties of these compounds confirmed the proposed formulae. The first crystalline electride, Cs+(18C6)2.e-, was synthesized, and its optical, electrical and magnetic properties fulfill all the requirements for the elusive -212- -213- electride, with the trapped electron present as a genuine localized anion. Two other crystalline electrides, Rbl8C6 and CsC222, were synthesized and their properties were investigated. There is no direct evidence for the molecular formula of the compound Rb18C6, but indirect evidence from syntheses and NMR studies suggests that this compound is the electride, Rb+18C6.e-. The analysis proves that the compound CsC222 is the electride, Cs+C222.e-, which forms either as a mixture of two crystals or as single mixed crystals. Either the packing of the Cs+C222 complex cation does not allow enough room for other alkali metal anions, or the compound CS+C222.e- is thermodynamically favored over the alkalides CS+C222.N-, where N- is the alkali metal anion, since the compounds Cs+C222.Na- and Cs+C222.Cs- could not be prepared. The "pure" ceside compound, Cs+(18C6)2.Cs-, can be prepared only in the presence of dissolved lithium metal (method 2); otherwise, a mixture of ceside and electride crystals forms. The complexation of rubidium and cesium cations by 18-crown-6 in electride and alkalide crystals is different than in crystals formed from simple rubidium and cesium salts with l8-crown-6. In the latter case, both 1:1 and 1:2 complexes between the alkali metal cation and the crown ether could be formed. In alkalide and electride -214- salts only the 1:1 complex forms with rubidium while only the 1:2 "sandwich" complex forms with cesium. Analyses and optical spectra cannot be used as definitive tools to identify the species present. Analysis gives only the compound stoichiometry, while the optical spectra are complicated by the sensitivity of the excited state to the environment, and by decomplexation in solution, dissociation of alkali metal anions and solid state reactions in the film. The magic angle sample spinning (MASS) NMR technique provides a way to identify the species present in alkalides and electrides. The conclusions drawn from NMR studies can be summarized as follows: -—The heteronuclear alkalide compounds that contain sodium are sodides and no complexation between the sodium cation and 18-crown-6 in the crystalline alkalides occurs. ——The sodium anion (Na-) exists in solution and in the solid state as a stable, long-lived genuine, spherically-symmetric anion. Its ground state is largely unperturbed by the environment and its chemical shift can be used as a diagnostic tool for its identification. 23Na-NMR signal of Na’ is -—In sodide salts, the broadened by magnetic dipole interaction with the protons of the complexant. The Na+C222 Signal is mainly broadened by quadrupolar interactions whose effects can be minimized by working at high magnetic fields. -215- —-The two crystalline compounds CsNa(18C6)2 and Csl8C6 are CS+(18C6)2.Na- and Cs+(18C6)2.Cs- and not Cs+18C6.Na- and Cs+18C6.e- as previously believed. -—The central line of 133Cs-NMR signals in cesium- containing salts is mainly broadened by magnetic dipolar and chemical shift interactions. -—The chemical shift of the Cs+ cation in the elec- tride, Cs+(l8C6)2.e-, is shifted 140 ppm (at‘~-20°C) from the expected resonance position because of the presence of a small contact (Fermi) density at the cesium nucleus (~0.01% of the free atom density). This is also in accord with the paramagnetism of the sample. -87Rb-NMR is not a good way to identify the rubidium cations in alkalides and electrides because of the large quadrupolar interactions, but it may (or may not) be used to identify the rubidium anion. -The ground states of alkali metal anions other than Na- have some contribution to their wavefunctions from orbitals other than the outer s orbital. -The two crystalline electrides, Rb+18C6.e- and CS+C222.e-, are essentially diamagnetic while the electride, Cs+(18C6)2.e- is paramagnetic with the electrons showing only a very weak antiferromagnetic interaction. __The EPR lineshape of the electride, Cs+(18C6)2.e- depends on the microwave conductivity of the sample. It shows Dysonian asymmetry that increases with increasing -216- temperature, indicative of activated conductivity. The electrons in this system exchange their spins very rapidly, so the EPR Signal is exchange narrowed with the result that a very small, temperature independent linewidth was observed. Electron trapping (or doping) can occur in alkalides to different extents, which affects the chemical shift of the alkali metal cation and anion as well as the powder EPR pattern. Two kinds of electron trapping can occur in sodide salts. Electron trapping at anionic vacancies (F-centers) produces an EPR Spectrum that consists of superimposed lines at the free electron 9- value, which indicates trapping at different non- interacting sites with different broadening mechanisms. Electron trapping near a cation defect sometimes produces a hyperfine interaction that indicates a rather large atomic character for the affected alkali metal cation. B. Suggestions for Future Work. (1) After the success in preparing the first crystalline electride, Cs+(18C6)2.e-, further investiga- tion of its properties such as quantitative chemical shift studies as a function of temperature, quantitative single crystal EPR and single crystal D.C. and A.C. conductivity studies should be made. Systematic procedures for growing high-quality single crystals for BIBLIOGRAPHY -217- X-ray structure determination are absolutely essential in order to better understand the properties of electrides. (2) The data already obtained for X—ray structure determination of the compound CsNa18C6 should be reevaluated in terms of the more likely molecular formula, Cs+(18C6)2.Na-. (3) X-ray powder diffraction and differential scanning techniques should be used as analytical tools to investigate the electride and alkalide salts. (4) Synthesis of crystalline electrides and alkalides with other macrocyclic polyethers such as 12-crown-4, lS-crown-S, 21-crown-7, "Lariat" ethers and cryptands 211 and 221 should be explored in order to establish trends in the prOperties of these compounds. (5) A systematic procedure should be established to find a way to control electron doping in alkalides, M+Ln.N;.eI_x. (6) The crystalline electrides, Cs+(l8C6)2.e-, Rb+18C6.e— and Cs+C222.e- might be good candidates for quantum mechanical ab initio calculations in order to understand electron-electron correlations in these systems. 10. ll. 12. 13. 14. 15. 16. BIBLIOGRAPHY J.C. Thompson, "Electrons in Liquid Ammonia", Oxford University Press, Oxford, 1976. C.A. Seely, Chem. News 23, 169 (1871). G. LePoutre and M.J. Sienko (eds.), "Metal-Ammonia Solutions, Colloque Weyl l”, W.A. Benjamin (New York) 1964. J.J. Lagowski and M.J. Sienko (eds.), "Metal- Ammonia Solutions", Butterworth (London) 1970. J. Jortner and N.R. Kestner (eds.), "Electrons in Fluids", Springer-Verlag (New York) 1973. J. Phys. Chem. 19(26), (1975). J. Phys. Chem. 84(10), (1980). J. Phys. Chem., Colloque Weyl issue, 1984, in press. T.R. Tuttle, Jr. and S. Golden, J. Chem. Soc. Faraday Trans 2 11, 889 (1981). J.W. Fletcher and W.A. Seddon, J. Phys. Chem. 84(10), 3055 (1980). E. Beckman and K.S. Pitzer, J. Phys. Chem. 65, 1527 (1961). D.E. O'Reilly, J. Chem. Phys. 41, 3736 (1964). D.E. O'Reilly, J. Chem. Phys. 41, 3729 (1964). W.D. Knight in "Solid State Physics", Vol. 2, F. Seitz and D. Turnball (eds.) Academic Press (New York) 1956. H.M. McConnell and C.H. Holm, J. Chem. Phys. 26, 1517 (1957). J.V. Acrivos and K.S. Pitzer, J. Phys. Chem. 66 1693 (1962 ). -218- 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. -219- T.R. Hughes, Jr., J. Chem. Phys. 38, 202 (1963). R. Catterall, Phil. Mag. 22, 779 (1970). A. Depriester, J. Fackeure and J.P. LeLieur, J. Phys. Chem. 85, 272 (1981). S. Freed and N. Sugerman, J. Chem. Phys. 11, 354 (1943). J.P. LeLieur and R. Rigny, J. Chem. Phys. 52, 1142 (1973). J.P. LeLieur and R. Rigny, J. Chem. Phys. 22, 1148 (1973). A Demortier, M. DeBacker and G. LePoutre, J. Chim. Phys., 380 (1972). R.L. Harris and J.J. Lagouski, J. Phys. Chem. 82, 896 (1981). C.A. Hutchison, Jr. and R.C. Paster, Rev. Mod. Phys. 25, 285 (1953). D.E. O'Reilly, J. Chem. Phys. 35, 1856 (1961). V.L. Pollak, J. Chem. Phys. 34, 864 (1961). R.J. Peck and W.S. Glaunsinger, J. Phys. Chem.' 84(10), 1176 (1980). R.R. Dewald and J.L. Dye, J. Phys. Chem. 68, 121 (1964). J.L. Dye, in "Progress in Inorganic Chemistry", Vol. 32, Stephen J. Lippard (ed.), John Wiley and Sons Inc., (New York), in press. D.M. Holter, P.P. Edwards, W. McFarlane and B. Wood, J. Am. Chem. Soc. 105, 2104 (1983). Y. Nakamura, T. Toma and M. Shimoji, Phys. Lett A60, 373 (1977). C.J. Page, G.M. Millhauser, P.P. Edwards, J.M. Freed and M.J. Sienko, J. Phys. Chem., in press. J.L. Dye and L.R. Dalton, J. Phys. Chem. 11, 184 (1967). J.L. Dye, Pure Appl, Chem. 49, 3 (1977). 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. -220- D.C. Johnson, C. Page, D.M. Holten, B. Wood, W. McFarlane and P.P. Edwards, to be submitted. R. Catterall, J. Slater and M.C. Symons, J. Chem. Phys. 52, 1003 (1970). R. Catterall, L.P. Stodulski and M.C. Symons, J. Chem. Soc. (A), 437 (1968). H. Normat, Angew. Chem. Int. Edit. Engl. Z, 1046 (1967). P.P. Edwards, S.C. Grey and D.M. Holton, J.C.S. Chem. Comm., 1185 (1981). R. Catterall and P.P. Edwards, J. Phys. Chem. 19, 3010 (1975). N. Mammano, "Metal-Ammonia Solutions", J.J. Lagowski and M.J. Sienko (eds.), Butterworth (London) 1970, p. 367. R.F. Marzke and W.S. Glaunsinger, J. Phys. Chem. 19, 2976 (1975). W.S. Glaunsinger, T.R. White, R.B. Von Dreele, D.A. Gorden, R.F. Marzke, A.L. Bowman and J.L. Yarnell, Nature (London) 271, 414 (1978). R.B. Von Dreele, W.S. Glaunsinger, A.L. Bowman and J.L. Yarnell, J. Phys. Chem. 12' 2992 (1975). R.B. Von Dreele, W.S. Glaunsinger, P. Chieux and P. Damay, J. Phys. Chem. 84, 1172 (1980). F.Y. Robb and W.S. Glaunsinger, J. Mag. Res. 46, 98 (1982). M.J. Mobley and W.S. Glaunsinger, Solid State Commun. 40, 357 (1981). N. Mammano and M.J. Sienko, J. Am. Chem. Soc. 90, 6322 (1968). A.M. Stacy and M.J. Sienko, Inorg. Chem. 21, 2294 (1982). A.M. Stacy, D.C. Johnson and M.J. Sienko, J. Chem. Phys. 16, 4248 (1982). V.M. Dukel'skii, E.Ya. Zandberg and N.I. Ionov, Akad. Nauk SSSR 62, 232 (1948). 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. - 221- 8. Golden, G. Guttman and T.R. Tuttle, Jr., J. Am. Chem. Soc. 66, 135 (1965). J.M. Ceraso and J.L. Dye, J. Chem. Phys. 66, 1585 (1974). J.L. Dye, C.W. Andrew and J.M. Ceraso, J. Phys. Chem. 16, 3076 (1975). P.P. Edwards, S.C. Gray, D.M. Holten, D.C. Johnson, M.J. Sienko, W. McFarlane and B. Wood, J. Phys. Chem. 61, 4362 (1983). See also, C.A. Young, R.R. Dewald, J. Chem. Soc. Chem. Commun., 188 (1977). J.L. Dye, J.M. Ceraso, M.T. Lok, B.L. Barnett and F.J. Tehan, J. Am. Chem. Soc. 66, 608 (1974). F.J. Tehan, B.L. Barnett and J.L. Dye, J. Am. Chem. So. 66, 7203 (1974). W. Bell Flower, "PhySics of Color Centers", Academic Press (New York) 1968. J.J. Markham, "F-Centers in Alkali Halides", Academic Press (New York) 1966. H.F. Ivey, Phys. Rev. 16, 341 (1947). W.C. Holten and H. Blum, Phys. Rev. 616, 89 (1962). W.T. Doyle, Phys. Rev. 666, 1421 (1962). L. Kevan and B.C. Webser (eds.), "Electron-Solvent and Anion-Solvent Interactions", Elsevier,(Amsterdam) 1976. J.E. Hart and W.J. Boag, J. Am. Chem. Soc. 66, 4090 (1962). J. Jortner, S.A. Rice and E.G. Wilson, "Metal- Ammonia Solutions, Colloque Weyl 1", G. LePoutre and M.J. Sienko (eds.), W.A. Benjamin (New York) 1964. L. Kevan, J. Phys. Chem. 66, 1232 (1980). L. Kevan, J. Phys. Chem. 66, 1144 (1978). L. Kevan, Adv. Radiat. Chem. 6, 181 (1974). H. Hase, R.Q.H. Ngo and L. Kevan, J. Chem. Phys. 66, 985 (1975). 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. -222- \ D.P. Lin and L. Kevan, Chem. Phys. Lett. 66, 517 (1976). L. Kevan, J. Phys. Chem. 66, 1628 (1981). N. Narayama and L. Kevan, J. Am. Chem. Soc. 103, 1622 (1981). C.J. Pedersen, J. Am. Chem. Soc. 66, 7017 (1967); 66, 3299 (1968). B. Dietrich, J.M. Lehn and J.P. Sauvage, Tetrahedron Lett., 2885, 2889 (1969). B. Metz, D. Moras and R. Weiss, J. Chem. Soc. Perkin II, 423 (1976). C.J. Pedersen and H.K. Frensdorff, Andew. Chem. Int. Ed. Engl. 6, 16 (1972). J.J. Christensen, D.J. Eatough and R.M. Izatt, Chem. Rev. 66, 351 (1974). R.M. Izatt, R.B. Terry, B.L. Haymore, L.D. Hansen, N.K. Dally, A.G. Avonded and J.J. Christensen, J. Am. Chem. Soc. 66, 7620 (1976). J.D. Lamb, R.M. Izatt, C.S. Swain and J.J. Christensen, J. Am. Chem. Soc. 102, 475 (1980). N.S. Poonea and A.V. Bajaj, Chem. Rev. 66, 389 (1979). R.D. Moras and R. Weiss, Acta Cryst. 329, 383, 388, 396, 400 (1973). P. Seiler, M. Dobler and J.D. Dunitz, Acta Cryst. 830, 2744 (1974). M. Dobler and R.P. Phizackerley, Acta Cryst. B30, 2746 (1974). M. Dobler and R.P. Phizackerley, Acta Cryst. B30, 2748 (1974). J.D. Dunitz, M. Dobler, P. Seiler and R.P. Phizaker- 1ey, Acta Cryst. B30, 2733 (1974). M. Dobler, J.D. Dunitz and P. Seiler, Acta Cryst. B30, 2741 (1974). D.M. Dishong, C.J. Diamond, M.I. Cinoman and G.W. Gokel, J. Am. Chem. Soc. 105, 587 (1983). 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. -223- Y. Nakatsuji, T. Nakamura, M. Okahara, D.M. Dishong and G.W. Gokel, J. Org. Chem. 66, 1238 (1983). R.A. Schultz, D.M. Dishong and G.W. Gokel, J. Am. Chem. Soc. 104, 625 (1982). D.M. Goli, D.M. Dishong, C.J. Diamond and G.W. Gokel, Tetrahedron Lett. 66, 5243 (1982). J.L. Dye, C.W. Andrew and S.E. Mathews, J. Phys. Chem. 66, 3065 (1975). F.J. Tehan, B.L. Barnett and J.L. Dye, J. Am. Chem. Soc. 66, 7203 (1974). B. Van Eck, L.D. Le, D. Issa and J.L. Dye, J. Inorg. Chem. 66, 1966 (1982). J.L. Dye, M.R. Yemen and M.G. DaGue, J. Chem. Phys. 66, 1665 (1978). J.S. Landers, Ph.D. Dissertation, Michigan State University, 1981. M.G. DaGue, Ph.D. Dissertation, Michigan State University, 1979. D. Issa, Ph.D. Dissertation, Michigan State University, 1982. J.S. Landers, J.L. Dye, A. Stacy and M.J. Sienko, J. Phys. Chem. 66, 1096 (1981). B. Van Eck, Ph.D. Dissertation, Michigan State University, 1983. M.L. Tinkham, M.S. Thesis, Michigan State University, 1982. M.R. Yemen, Ph.D. Dissertation, Michigan State University, 1983. M.D. Meadows, K.A. Smith, R.A. Kinsey, T.M. Rothgeb, R.P. Skarjune and E. Oldfield, Proc. Nat'l. Acad. Sci. USA 66, 1351 (1982). D. Issa and J.L. Dye, J. Am. Chem. Soc. 104, 3781 (1982). Z. Li, unpublished results, this laboratory. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. -224- O. Fussa, unpublished results, this laboratory. E. Kanffmann, J.L. Dye, J. Lehn and A.1. Popov, J. Am. Chem. Soc. 102, 2274 (1980). N.W. Ashcroft and N.D. Mermiu, "Solid State Physics", D.G. Crane (ed.), Saunders College, (Philadelphia) 1976. P.D. Townsend and J.C. Kelly, "Colour Centers and Imperfections in Insulators and Semiconductors”, Crane Russak and Co. (New York) 1973. E.R. Andrew, Arch. Sci. (Geneva) 66, 103 (1959). E.R. Andrew, A. Bradbury and R.G. Eades, Nature 183, 1802 (1959). I.J. Lowe, Phys. Rev. Lett 6, 285 (1959). J.A. POple, W.G. Schneider and J.H. Bernstein, "High-Resolution Nuclear Magnetic Resonance", McGraw-Hill (New York) 1959. Abragam, ”The Principles of Nuclear Magnetism", Oxford University Press (London) 1961. E.R. Andrew, Prog. Nucl. Magn. Reson. Spectrosc. 6, l (1971). E.R. Andrew, Int. Rev. Phys. Chem. 6, 195 (1981). M.H. Cohen and F. Reif, Solid State Physics 6, 321 (1957). M.D. Meadows, K.A. Smith, R.A. Kinsey, T.M. Rothgeb, R.P. Skarjune and E. Oldfield, Proc. Nat'l. Acad. Sci. USA 66, 1351 (1982). A.C. Cunningham and S.M. Day, Phys. Rev. 152, 287 (1966). E. Oldfield, S. Schramm, M.D. Meadows, K.A. Smith, R.A. Kimsey and J. Ackerman, J. Am. Chem. Soc. 104, 919 (1982). R.S. Drago, "Physical Methods in Chemistry", W.B. Saunders Co. (Philadelphia) 1977. J.L. Dye, M.G. DaGrue, M.R. Yemen, J.S. Landers, H.L. Lewis, J. Phys. Chem. 66, 1096 (1980). N. Tientega, unpublished observation, this laboratory. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. -225- A. Beckmann, K.D. Baklen and D. Elke, Z. Phys. 270, 173 (1974). G. Malli and S. Fraga, Theor. Chim. Acta 6, 275 (1966). M.T. Lok, F.J. Tehan and J.L. Dye, J. Phys. Chem. 26, 2975 (1972). S. Ganapathy, S. Schramm and E. Oldfield, J. Chem. Phys. 66, 4360 (1982). E. Mei, A.I. Popov and J.L. Dye, J. Phys. Chem. 66, 1677 (1977). S. Khazaell, A.I. Popov and J.L. Dye, J. Phys. Chem. 66, 4238 (1982). E.R. Andrew, W.S. Hinshaw and R.S. Tiffen, J. Mag. Res. 66, 191 (1974). A. Garroway, D. Vanderhart and W. Earl, Phil. Trans. Roy. Soc. Lond. A299, 609 (1981). M. Stoll and T. Majors, Phys. Rev. B66, 2859 (1981). R. Catterall and P. Edwards, Advances in Molecular Relaxation and Interaction Processes 66, 123 (1978). S. Dawes, unpublished observations, this laboratory. C.P. Slichter, "Principles of Magnetic Resonance", Springer-Verlag (New York) 1978. M. Tinkham, unpublished observations, this laboratory. G. Pake, "Paramagnetic Resonance", W.A. Benjamin, Inc. (New York) 1962. G. Breit and I. Rabi, Phys. Rev. 66, 2082 (1931). J.E. Wertz and J.R. Bolton, "Electron Spin Resonance: Elementary Theory and Practical Applications", McGraw-Hill (New York) 1972. T.W. Griswold, A.F. Kip and C.H. Kittel, Phys. Rev. 66, 951 (1952). F.J. Dyson, Phys. Rev. 66, 349 (1955). G. Feher and A.F. Kip, Phys. Rev. 66, 337 (1955). -226- 143. R.H. Webb, Phys. Rev. 158, 225 (1967). 144. W.S. Glaunsinger and M.J. Sienko, J. Chem. Phys. 66, 1883 (1975). 145. J. Smart, "Effective Field Theories of Magnetism", W.B. Saunders Co. (Philadelphia) 1966. 146. C. Kittle, "Introduction to Solid State Physics", 5th ed., John Wiley & Sons, Inc. (New York) 1976. 147. A.H. Morrish, "The Physical Principles of Magnetism", John Wiley & Sons, Inc. (New York) 1965. 148. M. Cyrot, J. de Physique (Paris) 66, 125 (1972). "'111111111IIIIIIIII“