‘ y ’ ‘ f'..\ , . f V .. ‘ . \bxflll .l ‘ 1 . I? fl“ ‘ y E I: , liraal -.I Ir - I I I I I r ‘ . w . \ . IHEFEv LIB R A R Y Michigan Stave University This is to certify that the thesis entitled LITHIUM : CRYPTAND 2.1.1. ELECTRIDES : A STUDY OF SOME MAGNETIC AND OPTICAL PROPERTIES presented by John Steven Landers has been accepted towards fulfillment of the requirements for Ph . D . degree in Chemistry Major professor 0-7639 OVLhDUE FINES. 25¢ per day per hen. RETURNING LIBRARY MATERIALS: Place in book return to remove charge from circulation records LITHIUM : CRYPTAND 2.1.1 ELECTRIDES A STUDY OF SOME MAGNETIC AND OPTICAL PROPERTIES By John Steven Landers A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1981 CP/725eécpz ABSTRACT LITHIUM : CRYPTAND 2.1.1 ELECTRIDES: A STUDY OF SOME MAGNETIC AND OPTICAL PROPERTIES John Steven Landers Dark blue microcrystalline powders produced from lithium metal and the cation complexing agent cryntand 2.1.1 (C211) were prepared from ammonia and from methylamine solutions. The mole ratio of lithium to cryptand, R, was varied from 0.60 to 2.0 but there was no evidence for Li-. The optical and magnetic properties indicate that Li+C2ll-e_ is an electride, a member of the class of materials in which the anions are electrons. Samples were studied by optical trans- mission spectroscopy, EPR spectroscopy, magnetic suscepti— bilitya and :microwave and D.C. conductivities. In many cases several methods were used to characterize samples prepared from the same solution; correlation among the various properties is high. Solvent—free films consisting only of lithium metal E::j——————————————————————————————————"————————--————————-—————————————————-n=z— John Steven Landers flecks deposited by ammonia or methylamine evaporation did not show any optical absorption. However, when the flecks are allowed to absorb solvent, the spectra show a plasma edge typical of concentrated metal—ammonia solutions. Solvent—free films from solutions containing both lithium and 0211 have properties which depend upon the lithium to cryptand mole ratio, R. Films from solutions with R < 1.15 have an absorption spectrum with low absorbance below U000 cm_l, peaks at 5000 cm‘1 and 7000 cm—1 and a high energy shoulder at ml2,000 cm—l indicating electron localization in several different non-equivalent environments. Anneal— ing the films increases the infrared absorbance below 5000 cm_l. On the other hand, spectra from films with R = 2 show a conduction electron plasma edge and high absorbance below 5000 cm_l. Two Li+C2ll-e' systems with R = 0.9M also showed metallic character, one throughout the range —35° to —70°C and the other above —45°C. This apparent metal— lnonmetal transition at ~A5°C was confirmed by EPR spectros— copy and microwave conductivity. Why these two systems showed metallic character while three other systems at the same mole ratio did not is uncertain. EPR Spectra of L1+C211-e' systems generally show the following: g—values at or near the free electron value, 2.00232; narrow spectra with AHp_p No.15 — 0.6 Gauss; A/B ratios less than 1.25; and no free lithium metal. Samples that showed multiple absorptions in the optical spectra fi—W John Steven Landers also showed multiple EPR peaks below about 30K. The number of unpaired spins in the samples was determined by compari— son with a ruby spin standard. "Metallic" samples have significantly less than 1% of their spins unpaired whereas 30% to 100% of the spins are unpaired in nonmetallic samples at 235K. At least two reversible temperature—dependent spin—pairing processes were observed in several samples with pairing energies of approximately 30 cal/mole and 100 cal/mole. The pairing is virtually complete by 3K. These results were verified by static magnetic sus— ceptibilities which showed maxima in their temperature dependence. Maximum susceptibilities varied from 5. x 10"3 esu/mole at 20K to 0.7 X 10—3 esu/mole at 70K, depending upon the lithium content. In all cases the susceptibilities dropped sharply at liquid helium temperatures. At high temperatures the susceptibility curves fit the Curie—Weiss law rather well and indicate the presence of about one unpaired electron per lithium. This confirms that essen— tially all of the electrons in the samples participate in the spin—pairing process. The optical spectra of solvent—free films of K+C222~e— also showed metallic character while films of Na+C222'e—, Rb+C222-e— and Cs+C322-e— generally showed localized (non— metallic) electron absorptions. Even with R < l, the Rb/C222 system showed an anion absorption peak. The mixed alkalide systems Li+C2ll-Na_ and Cs+C322-Na- showed strong Na- absorptions at about 13,800 cm-l. to a speedy reunion with Michael ii ACKNOWLEDGMENTS In the pursuit of a degree, one is indebted to many organizations and people for their assistance. My grati- tude extends to a multitude with some deserving specific mention. Thanks go to the United States Air Force and the Air Force Academy Department of Chemistry for sponsoring this educational opportunity. Professor James L. Dye's ability to blend encouragement and direction were vital to \ my completing the program. Of the colleagues in the re— 1 search group, Dr. Mike DaGue provided tutelage on low temperature, high vacuum techniques, while Brad Van Eck was extremely helpful as the backbone of the group. I also appreciated the aid and advice of Dr. Harlan Lewis, Dr. Long Dinh Le and Mike Yemen. The cheerful help of under- graduates Holly Lystad, Jeff Gulcher, Jim Anderson and Steve Hillson is acknowledged. I appreciated the warm hospitality of Cornell Univer- sity's Dr. Michael J. Sienko and his research group, especially Angy Stacy. Their aid with magnetic susceptibility measure— ments has proven very important in confirming what we believe about lithium—cryptand systems. Thanks also to Drs. William Pratt and Jerry Cowen of the M.S.U. Department of Physics who built a magnetic susceptometer for this study. iii MSU's technical and clerical staff deserve thanks also, especially glassblowers Keki Mistry, Jerry DeGroot and Andy Seer for their excellent and timely service, the very skilled machinists Dick Menke and Len Eisle, electronics designer Marty Rabb, and secretary Naomi Hack for her care and concern. Above, all Sherry's patience, help and understanding have been vital to me. Last, and definitely least, thanks to MSU for all its power outages, each one carefully timed for the most in— opportune moment, and for The Great Parking Hassle, ul— timately (?) won by Professor Horne. Research support from NSF under Grant Number POM—78— 15750 is gratefully acknowledged. iv TABLE OF CONTENTS Chapter Page LIST OF TABLES. . . . . . . . . . . . . . . . . . viii LIST OF FIGURES . . . . . . . . . . . . . . . . . x I. INTRODUCTION . . . . . . . . . . . . . . . . l A. Metal—Ammonia Solutions . . . . . . . . . 2 B. Metal—Amine Compounds . . . . . . . . . . 5 C. Electrons in Low Temperature Glasses . . . . . . . . 8 D. F-Centers . . . . . . . . . . . . . . . . 13 E. The Metal—Nonmetal Transition . . . . . . 15 F. Alkali Metal. — Cryptand Systems . . . . . . . . . . 17 II. EXPERIMENTAL . . . . . . . . . . . . . . . . 23 A. Reagents. . . . . . . . . . . . . . . . . 23 l. Complexing Agents . . . . . . . . . . 23 2. Metals. . . . . . . . . . . . . . . . 27 3. Solvents. . . . . . . . . . . . . . . 31 B. Glassware Cleaning. . . . . . . . . . . . 32 C. Solution and Sample Preparation . . . . . 33 1, Li+C2ll-e' Preparation. . . . . . . . 33 2. Sample Preparation and Instrumen— tal Description “0 a. Magnetic Susceptibility . . . . . Al b. EPR . . . . . . . . . A5 c. Optical Spectra . . . . . . A6 d. Microwave Conductivity. . . . . . 50 e. Pressed Powder Conductivity . . . 51 f. Crystal Growth Attempts . . . . . 52 D. Sample Analyses . . . . . . . . . . . . . 5A 1. Hydrogen Evolution. . . . . . . . . . 55 Chapter 2. pH Titration. 3. Flame Emission. A. Ammonia Content of Samples. III. LITHIUM CRYPTAND 2.1.1 ELECTRIDES . . . . . 59 A. Optical Spectroscopy. . . . . . . . . . . 60 1. Li with Ammonia and with Methylamine . . . . . . . . . . . . 61 2. Li/C2ll Films from Ammonia. . . . . . 6A 3. Li/C2ll Films from Methyl— amine . . . . . . . . . . . . . . . . 73 A. Summary and Discussion. . . . . . . . 78 B EPR . . . . . . . 87 1. Results . . . . . . . . . . . . . . . 90 2. DPPH Calibration. . . . . . . . . . . 106 3. Summary and Discussion. . . . . . . . 107 C. Magnetic Susceptibility . . . . . . . . . 112 1. Results . . . . . . . . . . . . . . . 118 2. Summary and Discussion. . . . . . . . 129 D. Conductivity. . . . . . . . . . . . . . . 132 Sample Analyses . . . . . . . . . . . . . 138 F. Li+C2ll-e" Summary and Conclusions. . . . 142 IV. OPTICAL SPECTRA OF OTHER SYSTEMS . . . . . . 146 A. Spectra in the Absence of Complexer . . . . . . . . . . . . . . 1A7 1. Na and K with Ammonia . . . . . . . . 1M7 2. Na/DABCO Films from Ammonia . . . . . . . . . . . . 150 B. Films from Ammonia. . . . . . . . . . . . 151 1. Na/C222 Systems . . . . . . . . . . . 151 2. Na/C221 System. . . . . . . . . . . 159 3. K/C222 and K/C2N22 Systems. . . . . . 161 A. Rb/C222 System. . . . . . . . . . . . 168 5. Cs/C322 Systems 168 vi Chapter Page C. Films from Methylamine. . . . . . . . . . 172 l. K/C2N22 System. . . . . . . . . . . . 17A 2. Cs/C322/Na System . . . . . . . . . . 175 3. Li/C211/Na System . . . . . . . . . . 176 D. Optical Spectra Summary . . . . . . . . . 178 V. AMMONIA ANALYSES. . . . . . . . . . . . . . . 18A VI. SUMMARY AND SUGGESTIONS FOR FUTURE STUDIES . . . . . . . . . . . 190 A. Summary . . . . . . . . . . . . . . . . . 190 B. Suggestions for Future Studies. . . . . . 191 APPENDIX. . . . . . . . . . . . . . . . . . . . . 19A BIBLIOGRAPHY. . . . . . . . . . . . . . . . . . . 208 Table LIST OF TABLES Li+C2ll-e'/NH3 vapor pressure study. Summary of optical spectra for Li/C211 systems. Li+C2ll'e_ spin—spin relaxation time extrema Results of EPR for Li+C211-e' (III) & (IV) Results of EPR for Li+C211°e- (IX), vapor pressure study Fermi temperatures of several systems. Spin pairing energies for Li+C2ll-e_ systems. Values for the parameters in the fit of the sample VIII static susceptibility results with the Wojciechowski equation for inter- acting spin 1/2 systems. Summary of Li+C2ll°e— spin and static susceptibility data Viii 71 80 89 98 106 110 130 Table 10 11 12 13 14 A—1 Results of the analyses magnetic susceptibility VII and VIII Results of the analyses magnetic susceptibility VII and VIII Summary of the data for systems. Swmayof positions; Li/C2ll. Summary of Film shape of samples of MSU samples Li+C211.e_ optical spectra peak systems other than NH3 content of samples. effect: max b = —a/2 Film shape effect: Amax b = a/2. Film shape effect: Amax a = 0, b > 0 Film shape effect: Amax b = -a Film shape effect: Amax b = 0; uniform thickness ix for for for for 1A0 143 180 185 200 201 201 Figure LIST OF FIGURES Page Reflectance spectra of Na—NH3 solutions from Reference 7 Cryptand molecular structures. Selectivity and stability of various cryptand complexes with alkali metal cations in 95% methanol 20 Liquid cryptand purification apparatus. 2“ Li+C2ll-e— multiple sample prepara- tion apparatus 26 Spectra of lithium metal films which contain methylamine: A — damp film; B - wet film. Lithium films contain— ing ammonia are virtually identical. 62 Spectra of a solvent—free Li+C2ll.e' (II) film from ammonia with R = 0.95: A —48°c; B -53°C; 0 -61°C. Elapsed time from A to C: 57 minutes. . . . . . . 66 Figure Page 8 Spectra of solvent—free Li+C2ll°e— (V) films from ammonia with R = 0.97: A —39°C, unannealed; B —70°C, un- annealed; C —36°C, annealed. . . . . . . . 68 9 Spectra of solvent—free Li+C2ll-e' films from ammonia: A — VII with R = 0.60; B — VIII with R = 1.57. Spectra of system VI with R = 1.15 were virtually identical to spectrum A. 70 10 Spectra of a single Li+C2ll-e' (X) film from ammonia with R = 0.96. The film was held at —65°C and the bulk solution temperature was varied: A —78°C, damp film; B —9l°C, film in transition; C —ll6°C and lower, dry film 72 ll Spectra of a solvent—free Li+C2ll-e_ film from methylamine with R = 0.97: A ~36°C, unannealed; B —36°C, an- nealed. The solution was made by replacing the ammonia of L1+C2ll°e' (V) with methylamine . . . . . . . . . . . 7A 12 Spectra of solvent—free Li+C2ll e— films with R = 2: A — unannealed xi Figure 13 1A 15 16 film from methylamine at —76°C; B — same film at -28°C; C — annealed film from ammonia, from Reference 61 EPR spectra of Li+C2ll-e' (II): upper spectrum at -59°C with A/B = 0.95; lower spectrum at —U5°C with A/B = 1.77 after transition. Semi—log plot of A/B vs. reciprocal temperature from Li+C2ll'e_ (II) EPR spectra. The three different kinds of symbols represent data collected on separate occasions during a three month period Linewidths from Li+C211‘e' (II) EPR spectra. The three different kinds of symbols represent the same three data collections as in Figure 1A Semi—log plot of the number of un- paired spins from Li+0211'e_ (II) EPR spectra. Data point in upper left corner represents 0.6% of the spins potentially present in the sample while the lower right data point represents .02% of the spins potentially present. xii Page 76 92 93 95 97 Figure 17 l8 19 2O 21 Page EPR spectra of Li+call.e’ (VI) with R = 1.15. The shift of the low field signals which is apparent from the lower to the upper spectrum continued until all signals appeared to have the same g—value at 65K. This merging of signals was gen— erally complete in other samples by 30K. . 100 A/B ratios from Li+c2ll.e' EPR spectra: A — VII with R = 0.60; B — VIII with R = 1.573 C — V with R = 0.97; D — VI with R = 1.15 . . . . . . 101 Lindwidths from Li+C211-e— EPR spectra: A — VII with R = 0.60; B - VIII with R = 1.57; C - V with R = 0.97; D - VI with R = 1.15. Solid symbols — data collected with the l—He cryostat; open symbols — z—NZ cryostat . . 102 Electronic g—Values from Li+C2ll-e_ EPR spectra: A — VII with R = 0.60; B — VIII with R = 1.57; C — V with R = 0.97; D — VI with R = 1.15; E - IV with R = 0.99; F — III with R = 0.98. . . . . . . . 103 Semi—log plot of the percent of un- paired spins vs. reciprocal tempera- ture from Li+C2ll-e_ EPR spectra: xiii Figure Page A — VII with R = 0.60; B — VIII with R = 1.57; C - V with R = 0.97 . . . . 105 22 Molar spin susceptibility of Li+c2ll-e‘ (VII): sample A with R = 0.60. The fraction of unpaired spins used to determine XM was from the straight lines of Figure 21, not from actual data points. 23 Molar spin susceptibility of Li+C2ll-e— (VIII): sample B with R = 1.57. The fraction of unpaired spins used to determine XM was from the straight lines of Figure 21, not from actual data points. 120 2A Molar spin susceptibility of Li+c2ll-e‘ (VI): sample D with R = 1.15. The fraction of unpaired spins used to determine XM was from the straight lines of Figure 21, not from actual data points. 121 25 Molar spin susceptibilities for Li+C2llve_ from Figures 22, 23 and 2A: A — VII with R = 0.60; B — VIII with R = 1.57; D — V1 with R = 1.15. The fraction of Figure 26 27 28 29 unpaired spins used to determine XM was from the straight lines of Figure 21, not from actual data points Molar static susceptibilities for Li+C2ll-e‘: A - VII with R = 0.60; B — VIII with R = 1.57; C — IV with R = 0.99. Note: The mole ratio of sample C is approximately the same as for samples labeled 0 in Figures 18 — 21, but this curve 0 represents a different preparation. Reciprocal of the molar susceptibilities displayed in Figure 26. Figures 26 and 27 courtesy of A. Stacy, Cornell University Microwave power transmitted by equal volume samples of Li+C2lloe_ (II) (circles) and palladium (squares) in a TE103 cavity. Temperature—dependent resistance of Li+C2ll-e_ (x) with R = 0.96. The region on the left above —33°C with BR/BT > 0 may either indicate the sample has decomposed slightly or has undergone a MNM transition XV Page 122 123 125 Figure 30 31 32 33 3A 35 36 Page Spectra of metal films which are damp with ammonia (concentrated M-NH3 solutions): A — Na; B — K . . . . . 1A9 Spectra of Na/C222 films with R = 1 from ammonia: A - dry; B — damp; C — wet. . . . . . . . . . . . . . . . . . 152 Spectra of Na/C222 films with R = 3 from ammonia: A — dry; B — wet. . . . . . 155 Spectra of Na/C222 films with R = A from ammonia: A — dry (fresh); B — dry (annealed); C — wet. . . . . . . . . . 156 Spectra of Na/C22l films with R = 2 from ammonia: A — dry; B — damp; C — wet. . . . . . . . . . . . . . . . . . 160 Spectra of solvent—free films of K/C222 with R = 0.95 from ammonia: A — fresh; B — annealed. . . . . . . . . . 162 Unnormalized spectra of a K/C222 solvent—free film with R = 0.94 from ammonia. Spectra were recorded at the following times after solvent removal: A — 2 min (fresh); B — 28 min (intermediate); C — 81 min (annealed) 16“ Figure Page 37 Unnormalized spectra of the same K/C222 film from which the spectra of Figure 36 were obtained: D — damp; E - intermediate; F — wet. . . . . . 165 38 Spectra of a solvent—free film of K/C2N22 with R = 0.91 from ammonia: A — fresh; B — annealed, 70 min after spectrum A. . . . . . . . . . 167 39 Spectra of Rb/C222 films with R = 1 from ammonia: A — dry (fresh); B — dry (annealed), 40 minutes after spectrum A; C — semi—wet . . . . . . . . . 169 A0 Spectra of Cs/C322 solvent-free films from ammonia: A — R = 2 dry (fresh) film at —A8°C; B — same film, dry (annealed) at —A8°C after cycling temperature to —58°C; 0 — R = 1 dry film at —US°C. . . . . . . . . . . . . . . 171 Al Spectra of solvent-free Cs/cryptand films with R = 2 from ammonia: A — with C322; B — with C222, from Ref— erence 61. . . . . . . . . . . . . . . . . 173 “2 Spectra of solvent-free equimolar Cs/C322/Na films from methylamine: A — fresh film at —30°C; B — annealed film at -0.7°C . . . . . . . . . . . . . . 177 xvii Figure 43 A-A Spectra of solvent—free equimolar Li/C2ll/Na films from methylamine: B — initial film; A — subsequent film which annealed and then re— mained unchanged from —7A° to —10°C. Effect on the peak amplitude of an optical film which is of non— uniform thickness (shape as indi- cated) and which fills various amounts of the optical beam. Effect on the peak amplitude of an optical film which is of non— uniform thickness (shape as indi- cated) and which fills various amounts of the optical beam. Effect on a Lorentzian absorption peak with a nominal 1.50 absorbance when the optical film is of non— uniform thickness (shape as indi— cated) and when it fills various amounts of the optical beam. Effect on a Lorentzian absorption peak with a nominal 2.00 ab— sorbance when the optical film xviii Page 202 205 Figure Page film is of non—uniform thickness (shape as indicated) and when it fills various amounts of the optical 206 beam xix INTRODUCTION The study of solutions and compounds prepared with alkali metal cations and cyclic polyethers is less than twelve years old, but it is based upon the nearly 120— year-old study of alkali metals in ammonia. From the original work of Weyl on the ammonia solutions of potassium and sodium (1) in the 1860s, the research has burgeoned, reaching the point in the past several decades that a number of international conferences have been held on the general subject. One series of five conferences whose proceedings have been published bears the title "Colloque Weyl" in honor of the original researcher. These and other conferences, review articles and important works in this field are listed in Reference 2. A branch of this research was propagated in the early 19708 by Dye and coworkers who demonstrated the tremendous solubility enhancement that cyclic polyethers provide for alkali metals in amine and ether solvents (3-5). This work led to further growth: the isolation of solvent-free com- pounds of alkali metals and cyclic polyethers, including the single crystal x-ray structure of one such compound (6). The research presented herein is an offshoot of this rapidly rowing and interesting branch concerning the effects of cyclic polyethers on alkali metals. A review of concentrated metal-ammonia solutions as well as metal-ammonia and metal-methylamine compounds pro— vides some background for the current work. Because the subject compound, lithium : cryptand 2.1.1 electride, has electrons in its structure which are independent of the lithium cation core, the compound is somewhat similar to low temperature glasses containing trapped electrons. These glasses will be discussed, followed by a description of F—centers which consist of electrons trapped in anionic vacancies in crystalline salts. Lithium cryptand 2.1.1 electride may be thought of as an F-center material in which all anions are replaced by electrons. In some cases the electron density becomes sufficient to cause the com- pound to undergo a transition to metallic character, but it is not clear which factor(s) control the metal-nonmetal transition observed in this study. Several theories de— SCTibingsnuflia transition will be very briefly mentioned. I.A. Metal-Ammonia Solutions Metal—ammonia (M—NH3) solutions can be divided, some- what arbitrarily, into three classifications according to the concentration of the dissolved metal. Generally solu- tions of one mole percent metal (MPM) or less are electro- lytic in nature, those between 1 and 8 MPM show variable character which places them in the metal-nonmetal (MNM) transition region, and those above 8 MPM are metallic. The concentration region of the DND transition may vary depend— ing upon the property being observed, but the crossover is generally complete by 8 MPM. Optical spectra of concen— trated M—NH3 solutions have generally been obtained by reflectance techniques. Figure 1 shows the reflectance spectra of Na—NH3 solutions in the intermediate and con— centrated regions (7). As the concentration of metal rises above 5.6 MPM, the reflectance spectra show a very sharp drop at the plasma frequency which is a collective resonance of the conduction electrons. Concentrated Li—NH3 solutions show a similar response (8). The electrical con— ductivity Of Li-NH3 solutions in this region increases ap— proximately as the cube of the metal concentration (9). At 20 MPM it reaches nearly 1.5 x 10” (ohm—cm)-l, but not until this very concentrated region does 80/3T become nega— tive as expected for a true metal. Electron paramagnetic resonance (EPR) spectroscopy of M—NH3 solutions shows an onset of metallic character at relatively low concentrations. In 1963 Vos noted a distinct change in the A/B ratio to that characteristic of metallic systems (10). Two years later Catterall determined that the A/B change occurs at 0°C in a 0.86 MPM solution of Cs in NH3 and at higher temperatures in solutions of even lower concentration (11). (A and B are the respective amplitudes of the low and high field lobes of first derivative EPR . r,—.. . .v he oocosomom Eogm mQOApsaom mEZImz mo mahooom oocmpooamom .H ogzmwm BI 2:2. 5:3 umcooEoo 8:28am; Economy 1 O O m m e m 2 q — J _ . — 1 — . , .. m \A I 3 5. 5 2 8 I. l 6 a M W W m oz 29.: one 2 £12 8.9: «o 2:: M a o 6. 7. 5 l H 2 5 3 2 .22: .523 20E _ _ p _ L L . — P spectra which will be discussed in greater detail in Sec— tion I.B.) Lelieur showed that the linewidth, AHp_p (where "p—p" indicates the width in Gauss between the first derivative peaks) for Na—NH3 increases from 30 mG in dilute solutions to about 5 G at 10 MPM (12). This rapid increase in AHp—p may be due to the increase in electrical conduc— tivity associated with the more concentrated solutions (13). Static susceptibilities are diamagnetic but become less so as the M—NH3 solutions become metallic, although they never lose their weak temperature dependence (1A—16). The effect of the concentrated metal is never sufficient to overcome ammonia's diamagnetism. I.B. Metal—Amine Compounds Of the alkali metal—ammonia systems which Marshall studied, only the saturated Li-NH3 system showed a dramatic decrease in the solution vapor pressure as the lithium mole fraction was increased (17). This strongly implies the existence of a compound Li(NH3)u since the mole frac— tion of lithium in the saturated NH3 solution is 0.20. In fact a metallic compound with stoichiometry Li(NH3),_l is formed when a 20 MPM solution is frozen at ~89K. Compounds M(NH3)6 have also been identified where M is Ca, Sr, Ba, Eu and Yb (12). The Li(NH3)u compound is face centered cubic between 82 and 88K, with a transition to a hexagonal structure below 82K. Magnetic susceptibility measurements show that the hexagonal phase obeys the Curie—Weiss law to 15K. The optical spectra are similar to those of the liquid (12). Evidence suggests that, unlike the hexagonal phase, the cubic phase exhibits nearly-free-electron (NFE) behavior (18). By freezing a Li—NH3 solution more dilute than 0.2 MPM, the cubic phase can be stabilized well below its normal 82-88K range. Studies over the expanded temperature range confirm the metallic character of the cubic phase (18). In their studies of cubic Li(NH3)u by EPR, Glaunsinger and Sienko used a theoretical lineshape equation derived by Dyson and extended by Webb to analyze the data for their metallic samples. In 1955 Dyson solved the theoretical equations accounting for the diffusion of conduction elec- trons into and out of the region penetrated by the EPR radio-frequency (rf) field (19). For a metallic sample which is thick compared to this rf skin depth, 6, the EPR lineshape assumes a characteristic asymmetric shape now commonly known as Dysonian in which 2.7 < A/B < 20. The theoretical shape, which also depends upon the electronic diffusion (TD) and relaxation (T2) times, was confirmed experimentally by Feher and Kip, also in 1955 (20). Two years later Webb extended Dyson's derivation for use with spherical particles in the region of the normal skin ef- fect (21). Several conclusions can be drawn from this extension: metallic particles whose radii, a, are large compared to 5 will still have a nearly symmetric EPR signal (A/B W 1) if TD >> Tl where T1 = T2 in a metal; and of more practical importance in the current study, even though the sample may be metallic, if the sample particle sizes are small compared to the skin depth (a/d < 1), the EPR signal will be very nearly symmetric. The Dyson theory was ex— tended to the region of the anomalous skin effect by Kittel (20). In this case the mean free path of the conduction electrons is long compared to the skin depth, 6. Assuming TD< 0.25 H where n is the electron density and aH is the Bohr radius. Empirically Edwards and Sienko (A8) showed that this rela— tionship is applicable to a large number of systems when * using an effective Bohr radius aH and a critical density n I C A system with an effective Bohr radius of 2.6A would be at the MNM transition when its electron density reaches lOzl/cm3. Hubbard (A9) viewed the problem based upon electron repulsion at a particular site. The bandwidth of energies from the overlap of atomic wavefunctions is given by W = 221 where z is the number of nearest neighbors and I is the overlap integral of wavefunctions for adjacent electron centers. If U is the single site Coulombic repul— sion energy, then when W/U 3 1.15, the electrons will be itinerant. Mott concluded through another approach that W/U W l was indeed appropriate for the prediction of the MNM 17 transition, hence the theory has advanced as the Mott— Hubbard model. Anderson's approach is one based on disorder (50). Suppose that the lattice is composed of potential energy wells of variable depth where V0 is the spread in energy. If W is the bandwidth as described in the Hubbard model, then the Anderson MNM transition occurs at Vo/W = 5 for z = 6. More recently others have placed the ratio at 2 - A (A7). Essentially, an electron without phonon assistance is unable to find another site with the same energy within a given distance and it therefore remains localized. Possibly the MNM transition can also occur through a spatial disorder as well as or in addition to a potential energy disorder (A7). Some have used a combina— tion of Mott-Hubbard and Anderson models to explain the MNM transition in Li-CH3NH2 solutions (2A). I.F. Alkali Metal - Cryptand Systems One class of macrocyclic polyether complexing agents, the macrobicyclic diamines, was synthesized by Lehn in 1969 (51). Because they are three—stranded, the molecules create their own cavity, hence the name cryptands. They are depicted in Figure 2. The numbers refer to the quantity of ether oxygens in each strand and the trival names are based upon these numbers. For example, the polyether in Figure 2 with two oxygens in each strand would is I. m = 0; n = 1 ‘ (C211) II m = 1; n = 0 '(C221) III. m = n = 1 (c222) IV. m = 1; n = 2 (C322) m (\o o/h N’\,o’\,o\/N ‘\/N N\/' \_J Tetraza C222 or C2N22 Figure 2. Cryptand molecular structures. 19 be cryptand 2.2.2 or more simply C222. IUPAC names are listed in Section II.A.1. The large number of nitrogen and oxygen atoms lining the cryptand cavity provides a prime environment for com— plexing a cation. The number of ether linkages in each strand dictates the cavity size which in turn dictates the size of cation which can be readily complexed in the cavity. Figure 3 shows this selectivity as well as the high com— plexation constants when a cation is of optimum size for a particular cryptand cavity. The lack of selectivity among large cations by C322 (and other large cryptands) seems to result from the flexibility of the larger molecules (52). By proper choice of alkali metal, cryptand and sol— vent, a solution of controlled stoichiometry can be prepared which may contain either alkali metal anions or simply electrons as the counterions for the encrypted cations. Dye has discussed the considerations and strategies for these preparations in numerous articles (53—57). When a cryptand contains a cation, the complex will be referred to as a "cryptate". When a cryptate in a solution or solid has an alkali metal counterion, the complex, M+C°N-, is called an ”alkalide" where N may be the same as M or different (5A). An example is the alkalide Cs+C322oNa_ 9 cesium C322 sodide. When a cryptate has a counterion which is an electron, the complex, M+C-e’, is called an 20 Log K Li‘ Na" K‘ Rb" Cs‘ 0 ll . l1 . 1 .l I L l l 0.8 LO l.2 L4 L6 Cation Radius (A) Figure 3. Selectivity and stability of various cryptand complexes with alkali metal cations in 95% methanol. 21 "electride" (58). The electride of importance in this study is Li+c2ll-e‘, lithium 0211 electride. Crystals of Na+C222~Na‘ were first reported in 197A (6). An x-ray structural determination of these thin, gold—colored plates showed cubic closest—packing (59). Essentially the crystal contains closest—packed cryptate complexes with anions in the octahedral holes. The Na- - Na‘ intraplane distance is 8.8A while the interplane Na- — Na- distance is 11.0A, leading to a probable an- isotropy of properties dependent upon the anion—anion distance (5A). No crystal structures of electrides have been determined although paramagnetic blue powders of probable stoichiometry M+C-e' have been prepared (58,60,61). If a single crystal of Na+C222-e' could be isolated, it is possible that its structure would also be hexagonal but with electrons occupying the octahedral holes. The effect that an unsymmetrical cryptand such as C211, 0221 or C322 might have on hexagonal packing is uncertain. However, the structure of L1+C2ll.I' is tetragonal, space group PA1212, with four molecular species in a unit cell of dimensions a = b = 8.72A and c = 2A.36% (62). The iodide ion with a radius of 2.16A would be too large to occupy the octahedral hole if the L1+C2ll cryptates were to pack hexagonally like hard spheres, but an electron could fit, of course. The deep blue paramagnetic solid Li+C211-e_ is the primary subject of this dissertation. It is probably 22 more ordered than a molecular glass and it certainly has a greater density of trapped electrons. On the other hand the electride, a fully substituted F-center, is more dis— ordered than an alkali halide F—center single crystal, but once again its trapped electron density is much higher. Because of the size of the cryptate, the electron density of the electrides is about an order of magnitude lower than that of simple metals. In this respect the electride is similar to Li(NH3)A or MH Amv ooo.mfie ooose .ooome and :IMWIIIIIIIIIIIIIIIII mm.o mmz HHH o no wsmm HQ hammoc lull A©08909aopv map A Anv ooo Nae soon .ooom uses: v .sno aaaaaaaaa willulu ooom eofimu v .sso 0w 0 a U mammad III: oomzl A mho mm.o mmz HH popompmgo mEmde comm hos omoo mEmmHQ Ill: game If, owoo mammad Ill: she mm.o mmz H gopomnmco mammad ommm p03 1% owoo osmde run: game wwwm Gasman—HQ It‘ll. NACHU N mmz O||I| sopomnmno mammad case? p03 (In: owoo mammfio III: demo null mmmeo Aaamo osv hopomsmso mammad comma p03 :11: owoo mEmmHQ III: damp IIII mmz “Hamo osv mnogpo p mcoflpflocoo m pso>aom Eopmsm : Sang A Eov mCOApHmom smog an n .m canoe .mEopmmm Hflmo\flq pom mgpoodm amoepdo no mLmEEsm :lllltlilllllllillflli smoocsmso mEmch cosme Assoc mac .smso cEmcHo aAmvocm.mae comm Ahmv mac sm.c mmmeo > Illlllllltlllitll mmwm memHQ hfihmmc OOON Dowbl\omml .pwzo mammac wsfimmoposfl omom coHoI\omol cosmfiu\omoa ocm m x Ano oom.mfie ooose .oomme Anna a sec ago no.0 mz sopowgmzo mammao occm ooosl .Assmv ago owco mEmmHo III: cosml .Acsmv ago m Locomnwso csmwao comma Anmo ago sm.H m2 HHH> 1 Ir. . m H> Ra gopomncco mEmeo cozse .comme ago co c mz H . m H> sopownmno mammac cccsa .Asooomme ago ma H oz omcm msmmfio manmo: cos: Ascmoznc .ncmo . mmz > wsmmac .Amv coc.mae ooose hoccme Ancvmsc so c .ll o> o Eopmmm wsocpc mo whoapflccoo m as fl m EHflm AHIEoV omQOHpHmom zoom .ooosaodoo .m oaooe 82 Ease one as phone .oLSpmsoQEop soapsaom xfiso one we ccooom who ccc one .Uopmfla ohm monopmthEop 03p when; mcoamocsm .Ho oocopomom Eonm mpcmo .hocfizogm n w .xmog LOnmE I E o aohdpmpoQEop u :sm .Qmosc n so m owco memHQ III: oowml owco wEwwHo hashes comma coon: ACCGV NPHU m NEmeO llll whospo mo doompflccoc m pso>fiom Eopmzm Efiflm AHIEoV oncompflmom xmom .ooscaosoo .m oases optical spectra while the relative amount of metal was varied. Grain boundary effects or sample thickness are dismissed as a major factor because of the consistency of results from thin optical films to thicker films in EPR samples to bulk samples for magnetic susceptibility, typi— fied by systems VII and VIII to be discussed later in this chapter. One fairly obvious rationale for the discrepancy among the R = 1 preparations is the possible presence of solvent. The ammonia analyses of Chapter V and the vapor pressure study of systems IX and X were attempts to determine the possible effects of solvent. The results of Chapter V indi- cate that a "dry" electride optical film may still contain approximately one mole of NH3 for every ten moles of cryp- tand. Chapter V results also argue that the mole ratio C211 : NH3 increases (more NH3 is removed) for longer static pumping times. However the optical spectra do not show annealing effects unless the bulk solution is warmed con— siderably above liquid nitrogen temperature. Therefore it is concluded that, for optical films in this research ex— posed to a bulk solution at -196°C, the films were es- sentially dry within several minutes of freezing the bulk solution and that any subsequent decrease of the film's NH3 content had a negligible effect on the film, its spectra and its metallic or nonmetallic character. This conclusion is reinforced by the vapor pressure study wherein the 8A NH3 vapor pressure was raised approximately thirteen orders of magnitude before any effect was noted in the optical spectra, at —1l6°C < Temp < -91°C for the bulk solution with the film at —65°C. To summarize the above discussion, the reasons for the differences in behavior of Li+C211-e- samples I and II from those of later samples remain unknown. Several general trends are noted among the other lithium electride systems. As film temperatures are raised, the infrared absorbance increases, probably reflecting an increase in the conduction band population that is typical of thermal promotion in a small band-gap semiconductor. As the mole fraction of Li is raised from 1.15 to 1.57 to 2, the optical spectra appear to become more homogeneous. Only three systems with R values of 1.57 or 2 were studied so there is a possibility that these spectra are not representative. However if they are, then the increase in uncomplexed lithium cations (no evidence for Li° or Li') must be causing or helping create a more homogeneous en— vironment for the electrons. Possible explanations follow. It appears from optical spectra that somewhere between mole ratios 1.15 and 1.57 the high energy e; peak and shoulder begin to disappear. It is possible, but it has not been demonstrated, that Li and 0211 may form ins elusive-exclusive complexes as in the case of Cs and C222 (81). Then the electrons would feel at least two 85 different environments (the cryptand asymmetry may create more), one in which the electron is in the vicinity of an inclusive Li/C211 complex and the other in which the elec— tron may be forming some variety of loose ion pair (EPR shows no hyperfine splitting) with a protruding or exclusive cryptated lithium cation. It is possible that the higher energy peak of the two typical e; absorption peaks is due to some form of ion pair, analogous to the 8A75 cm-1 (1180 nm) peak attributed to the ion pair (Li+, esolv) in tetrahydrofuran (82). Then as the mole ratio is in- creased, free lithium cations begin to fill enough inter- stices to homogenize the environment for the electrons, progressively decreasing the effect of some quantity of exclusive Li+ A second and equally speculative possibility is that NH3 is such a powerful solvating agent that the equilibrium Li° + C211 : Li+C2ll + e" does not lie far to the right as with many other solvents (52). Perhaps competing equilibria Li° + ANH3 : Li(NH3)u OI’ + o O _>, 0 Li + xNH3 a Li (NH3)X + esolv are so energetically favorable that the first equilibrium above is shifted somewhat to the left. Then when the sol- vent is removed by flash evaporation during optical film formation, some lithium is trapped outside the cryptand. The increase in the mole ratio, as in the previous hypothe- sis, homogenizes the solid structure and the electronic environment. It should be noted, however, that at R = 2 there are many more octahedral and tetrahedral holes than there are tiny lithium cations to occupy them. But before the structure is homogenized physically, it is homogenized by promotion of the electrons into the conduction band. Some critical minimum of Li+ concentration is passed after which the electrons are metallic, with the transport process possibly being site—to—site hopping on Li+ centers in the minimally metallic regime. The spectra of damp and wet films show that electride films have a fairly high ammonia affinity, but considerably lower than that of lithium metal. With the bulk solution at -ll6°C there should have been enough NH3 present to saturate a —65°C Li° film, yielding a typically metallic spectrum. However, this metallic spectrum did not appear for the electride until the bulk solution was at m—78°C where the vapor pressure was a factor of W90 higher than at —ll6°C. 87 III.B. EPR Electron paramagnetic resonance (EPR) probes the change of the local environment of an unpaired spin, as opposed to methods which monitor bulk properties such as magnetic susceptibility and conductivity which will be discussed later. In the EPR experiment, degenerate spin energy levels are split by application of a magnetic field. Transitions between the Zeeman levels can be induced by radiation of the appropriate frequency. Since the Zeeman splitting is proportional to the applied magnetic field, field strength and/or radiation frequency can be varied to induce the resonance according to the equation AB = hv = gBH where h is Planck's constant, 0 is the frequency of the electromagnetic radiation, g is the Landé g factor, B is the Bohr magneton and H is the magnetic field strength. In practice, microwave radiation of a fixed frequency is generated and the magnetic field is swept through the reson- ance value. The Landé g factor is -3 Sm mo mpcflx pgohoocwc posse one .mspooom mam AHHV o.HHmo+Hq Eosc oncomgoCEou Hmoogpfloot .m> m\< co pOHc aoHIHEom to: 8.180 t. ONO 05 no.0 No.0 5.0 mOOO . _ — — - - c u — - u — _ - u q q — - one v.00. xOON V _ _ mm 1 in < Jth o no 0 o my . 4 us v3 v5. van col .4. O I .. l _ . . Ioo._ ... w i 4 4 4‘ 1 I w l 4 low; 3 :1 Au u no I . S . Ion. ... w I n. no ‘ . nu . XNNN P .— —- n p n — p . - —.- p - p — . .aH thorns 9A at about 100K. The plot of AH against temperature in p-p Figure 15 shows definite changes in the same temperature regions. Hysteresis is evident in both figures. The sig- nificant change in AHp_p near 5K was not an artifact. As the temperature was raised above m3.5K a broader signal diminished and a narrower one increased. During an in— crease of 3 — AK the low temperature signal became neg— ligibly small compared to that of the high temperature species. This same phenomenon was noted in many other Li+C2ll~e_ systems. From —160°C up to the transition temperature the g—value was a constant 2.00200, but from —A5°C to —A1°C it was 2.00167. It is of interest to determine the number of paramag— netic centers giving rise to the observed signals. This can be done absolutely or by comparison with a spin standard. The area under the absorption curve is propor- tional to the number of spins, so tedious double integra— tion of the first derivative curves of the sample and the standard would give the desired information. However, the intensity of a Lorentzian or gaussian signal is proportional to Ymax°(A to—peak amplitude (69). So in this study, the sample and 3+ Hp_p)2 where 2Ymax is the derivative curve peak- a National Bureau of Standards crystalline A1203;Cr (ruby) rod were placed in the EPR cavity adjacent to one another. The intensity of each curve at a given temperature 2 was approximated by 2Y' .(AHp—p) and the absolute number max 95 - .aH onaaaa CH mm wooepooflaoo some conga oemm one pcomogcog mHooE>m no mpcflx pcogoomflp oopgp one .cppooaw mom AHHV Io.HHNc+Hq Eogm mgpcflsoCHq .mH opswfim Av: aim... CON ONN 0m. 03 . 00. om ON _ _ _ _ _ _ . do. 441 No 27 ‘1 ‘1 1 «1'4:r l C) a at». I. I. CI 4'. III I GD (3 (ssnoo) cl"de 96 of sample spins was then calculated from the following equation (85): O‘oBo 2 NOgOU (Intensity) (M.A.)O (R.G.)O g (Intensity)O(M.A.) (R.G.) N: a B where N is the number of spins, g is the g—value, U 0 O is the integrated intensity of the ruby signal which varies with ruby crystal orientation in the EA EPR cavity, M.A. is modulation amplitude and R.G. is receiver gain. The subscript "0" refers to the ruby standard. Figure 16 shows the results of this spin calibration for Li+C2ll-e' (II). The absolute number of spins participating in the EPR signal is very low, but undergoes an order of magnitude increase at the transition temperature. The EPR samples for Li+C2ll-e' (III) and (IV) each contained approximately 1.5 x 10 spins, a faCtOP Of three less than for sample II. Yet both of these samples were far too large for EA AFC stability, hence the results will not be reported here. The effects noted by Catterall (70) which were discussed in Section II.C.2.b. were most pronounced for sample III in the range 30 — 160K where rapid signal crossover caused many apparent linewidths to be <0.10 G. One observation which was independent of these effects was the same signal change at m5K as noted for system II. From 3.9 to A.8 and 6.2K the gain was increased by a factor of 500 to obtain a comparably sized signal and 97 40—' _ § 30— 0_ _ CD CD I 20‘ _ E 0) £2 :2 '0 IO— _ : : - m _ - .EI \\.I - (D 5 .°\ ‘ H— . — CD __ \\\\\\ L. ‘C -_ 8 3 E _ :3 Z 2— _ I I I I I’\ 4.0 6.0 8.0 l/T (K"oIo3) Figure 16. Semi—log plot of the number of unpaired spins from Li C2ll'e— (II) EPR spectra. Data point in upper left corner represents 0.6% of the spins potentially present in the sample while the lower right data point represents .02% of the spins potentiallv present. 98 at 8.3K it was reduced back to near the 3.9K settings. Later both samples 111 and IV were reduced in size and data were collected in the liquid nitrogen temperature range only. Electronic g—values for both samples are in— cluded in Figure 20, and in Table A. Because both samples were reduced in size by an unknown amount, no calibration of the number of spins was attempted. Table A. Results of EPR for Li+C2ll°e_ (III) and (IV). Sample III, R = 0.98 Sanple IV, R = 0.99 Temp (K) 115 233 118 233 A/B Ratio 1.21 1.23 1.08 1.29 AHp—p (G) 0.37 0.31 0.28 0.2A g—Value 2.00200 2.00229 2.00230 2.00229 EPR samples for systems V (R = 0.97), VI (R = 1.15), VII (R = 0.60) and VIII (R = 1.57) were prepared with W2 x 1017 spins and were apparently just small enough to avoid signal distortion. All samples except VIII had multiple signals up to approximately 30K, as depicted in Figure 17. The asymmetry ratios A/B are presented in Figure 18, the linewidths in Figure 19 and the g-values in Figure 20. The percentages of unpaired spins in the samples studied with the liquid nitrogen cryostat are shown on a Figure 17. 99 EPR spectra of Li+C211-e‘ (VI) with R = 1.15. The shift of the low field signals which is apparent from the lower to the upper spectrwn continued until all signals appeared to have the same g—value at 65K. This merging of signals was generally complete in other samples by 30K. l89 K R6. = 25 MA. = .050 G 5.2 K R.G. = 40 MA = .080 G k353i Figure 17. 101 .mH.H u m fipfls H> I Q mom.o u m Sufi: > I o mom.a u m Spa: HflH> I m mcoo n m Spas HH> I < umppomgw mmm Im.HHNo+flA Eogm mOflpmh m\< Av: 0.2m... OmN OON . On. 00. _ _ _ _ _ _ _ I «4 d 4 .1 o_ a «V 4 < I . I. u u ‘ ‘ * ‘ u t I o I I I a 0 mm" o 4 I o I I o I Q I 4 I 4 I 4 Q 4 4 4 T Q 4 I d .wH ogsmflm OIIVH S/V .pmumomho mzlul mHODEmm Como I wHonEmm pflaom m Sofia H> I m HHH> I m moo.o .mH.H n m Sufi; HH> I < ”no.0 "wgpoomm Mom I m one: s I o mpmpmomgo ozIa one goes cocooaflo msm.H n Io.HfiNc+Hq Eopm mgppfizogfiq 0 some m noes 102 .xooSmp alum cam. os oo. oo 8 . _ a _ .D .p .9 av b. or b. s. “v .9 b. h. min. QM s Q o r t p o I d.nq .v a a a $3.. $30 mo 1% ...: date... I.» o o s a « s.» o o < A o o m II ‘1 O I O “ m. I. .ma 0: e: mw_ NAN ohsmflm (ssnog) d-de .wm.o u m spas HHH I m mma.o u m hens >H I m mmH.H n m can: H> I c msm.o n m sea: > I o msm.a u m son: HHHs I m mcoo I m . amen gums HH> I < ”caveman mom Im.HHmo+Hq Eoso modam>Im oesopuooam cm on .L C: dim... . one. com on. _ ow. _ . _ . _ . 4 km I 808 4 4 4 4 _ . Q . m. a 4 m I «NOON .0 cu. £ Ico - n no I mu. I on men. U. Iummoom W a 4. 4. .Iamgunzw mw . 4 4 < 4 4 4 I 88m 4 4 . < c 4 . o 4 q 4 4 I mmoow semi-log plot in Figure 21. Neither the ruby nor DPPH standards would fit into the liquid helium cryostat along with the sample, hence g—values and the calibrated number of unpaired spins are not shown below 95K. However, by direct comparison of spectrometer settings the percentages of unpaired spins were approximated down to 3.5K and these values are also included in Figure 21. Only sample V had a signal which could have been from free lithium metal. The approximately 7 Gauss wide signal appeared between 3 and AK. Calculations from three different spectra placed the quantity of unpaired spins in this signal at 0.3 i 0.1% of those potentially present. EPR samples from the vapor pressure study Li+C211°e' (IX) were evaluated in the nitrogen—cooled range only. Because the ruby standard would not fit into the cryostat with these sample tubes, no information is available for the numbers of spins contributing to the signals. The data for the EPR samples which were sealed off at —116/—65°C and at -91/—65°C are in Table 5. Each of these data values remained fairly constant or changed smoothly over the tem— perature range; no transitions were obvious. One item of considerable interest was the extremely high receiver gain settings required for Sample B. They were 25 — 50 times those for Sample A even though the sample sizes were comparable. a .,\. 9' '0 a \ U) a; - 8 5 C) U] E 9‘ a 2 " | - .5 - O O .2 — \ l . I .ll 11 .J 30 90 ISO 2|0 270 I/T (K" x Io') Figure 21. Semi-log plot of the percent of un aired spins vs. reciprocal temperature from Li C2ll°e‘ EPR spectra: A - VII with R = 0.60; B — VIII with R = 1.57; C — V with R = 0.97. 106 Table 5. Results of EPR for Li+CBll-e‘ (IX), vapor pres- sure study. Sample Aa Sample Bb Temp. (K) 118 2A3 118 2A3 A/B ratio 1.27 1.26 ml.l 1.11 AHp—p (G) 0.32 0.2A 0.55 0.22 g—value 2.0026 2.0026A 2.0023 2.0022 aEPR sample sealed with sample at —65°C and bulk solution at -116°C. bEPR sample at —6500, bulk solution —91°C when sealed. III.B.2. DPPH Calibration DPPH, d,d'-diphenyl—B—picrylhydrazyl, is often used as a standard for g—value determinations. While its g-value is normally reported as 2.0037 i .0002 (69), presumably any given DPPH sample could be calibrated to greater ac— curacy by using a standard whose g factor is accurately known. In this study the DPPH sample used for all lithium electride g—value studies was calibrated against the benzene radical anion. Benzene” in a 2:1 solvent of THFzDME is reported to have a g—value of 2.00285A i .000007 at —101°C with a splitting constant of 3.781 Gauss (86). The seven—peaked benzene— signal showed a splitting of 3.80 Gauss in this study, and four determinations of the DPPH 107 g-value after repositioning the adjacent DPPH/benzene— sample tubes yielded a g—value of 2.00370 (3) at —102°C (the value in parentheses is the standard deviation of the last digit). The DPPH g-value at —73°C determined in the same way was 2.00351 (1). Segal, Kaplan and Fraenkel stated that their g—value for benzene- did not vary with temperature over the range from room temperature to —100°C, so presumably the variation observed here is due to DPPH. This DPPH sample was not calibrated at other temperatures, hence all electride g-values in this study are determined by comparison with DPPH as though it were at ~102°C with a g—value of 2.00370. III.B.3. Summary and Discussion Li+C2ll°e', regardless of R value, does not appear to contain any free lithium metal with the possible exception of a very minor amount in sample V. The unpaired electrons in the signals are all at or near the free electron g—value, 2.00232, indicating that there is virtually no spin-orbit coupling present. The A/B ratios in some cases (Figures 1A and 18) are less than expected. There is no theoretical basis for A/B < 1.0 (19,21) so systems 11 and VII, whose optical spectra show electrons in several trapping sites at 200 — 220K, may have interference from multiple electron— site signals which skew the EPR spectra. System 11 shows a dramatic jump in the A/B ratio near 225K which is 108 presumably associated with its transition to metallic character. Metallic samples thick compared to the micro— wave penetration depths normally show much higher A/B ratios. Only when the skin depth is comparable to the sample thickness will the ratio be low and this is prob- ably the case for Li+C2ll-e_(II) in its metallic region. The AHp_p values of sample II increase with increasing temperature in the metallic regime (Figure 15) correspond— ing to a decrease in the spin—lattice relaxation time. This decrease is characteristic of delocalized electrons (2A). All systems have signals which are quite narrow, but the widest lines are those of systems 11 and VIII. The former underwent a MNM transition and the latter has the highest R value (1.57) of those thoroughly studied by EPR. These two samples have the most efficient relaxation mechanisms. There seems to be a significant difference in the num— ber of unpaired spins in samples which appear metallic and in those which are non-metallic. Li+C2ll-€— (II), by cali- bration with a ruby standard, was approximately 0.02% un— paired at 110K and 0.6% unpaired at 228K after the MNM transition. The optical spectra of K+C222-e' show high metallic character and the EPR results by DaGue on a sample from the samg solution indicated only 0.02A% of the spins were unpaired at 157K (6A). For metallic systems only the fraction of electrons (T/TF), where TF is the Fermi 109 temperature, would be in the conduction band and presum— ably unpaired. Table 6 contains a comparison of these systems. One other result which indicates that metallic samples have few unpaired spins is the high receiver gain required for sample B of system IX. While nothing in Table 5 sug— gests it has peculiar behavior, the sample was approxi— mately the same size as sample A, yet it required receiver gain settings 25 — 50 times those for sample A. This may indicate that spin pairing was occurring in the sample as it acquired NH3 on its way to metallic character. The optical spectrum (curve B, Figure 10) appears to be in transition from a dry, standard spectrum to a damp, metallic one. It is probable that the EPR sample is indicating the same transition, hence the onset of spin pairing and an abnormally low EPR signal. On the other hand, the fact that the non—metallic and/or near—metallic systems V — VIII have a considerable percentage of their spins unpaired is clear from Figure 21. There is a systematic alignment of spins in each system as the temperature is decreased and in several samples the rate of spin pairing changes in the region of 50K. The energy of this antiferromagnetic spin coupling can be readily calculated from the equation —Ea/RT N = N e co 110 Table 6. Fermi temperatures of several systems. System Temp (K) Fermi Temp (K) Li+C2ll-e_ (II) 228 3.8 x 10Ll Li+c2ll.e“ (II) 110 .5 x 105 K+C222.e- 157 6.5 x 105 (a) K+C222~e_ -—- 5.6 x 103 (b) Li+02ll-e‘ (VIII) A 8.0 x 102 Li+c211-e‘ (VIII) 225 1.1 x 103 Free Metals L1 78 5.5 x 10LI (c) K 78 2.5 x 10L1 (c) aReference 6A. bReference 6A, based upon estimated conduction electron density. CReference A2. 111 where N00 is the fraction of spins N unpaired at infinite temperature, R is the gas constant and Ea is the energy of coupling/decoupling. While system V1 with R = 1.15 is not included in Figure 21 because its unsymmetric EPR lineshapes yielded very scattered data, it is still included below. The spin pairing energies for systems V — VIII are listed in Table 7. Table 7. Spin pairing energies for Li+C2ll-e- systems. Spin Pairing Energy (cal/mole) System R Low Temp High Temp V 0.97 35 7A VI 1.15 28 177 VII 0.60 32 117 VIII 1.57 26 26 Note that the energies are in calories: the spin pairing in these lithium electrides is very weak. The percentages of unpaired spins for System VIII appear to lie along a single straight line in Figure 21. Hence it only has one pairing energy in Table 7. This is also the system which showed the most homogeneous optical spectra (Section III.A.2). It is perhaps quite significant that the same systems exhibiting inhomogeneous, multi—peaked optical 112 spectra are the systems in Figure 21 and Table 7 which have two different spin pairing energies. It is conceivable that the two major electron trap sites indicated in the optical spectra are the same sites where spin pairing is occurring and that each site has its own pairing energy. Sample V1 with R = 1.15 showed EPR properties which did not seem to follow any pattern set by the systems with other mole ratios. Sample VI had the highest A/B ratio by far, the highest g—value, the narrowest spectra and the great— est number of unpaired spins of Systems V — VIII. It is not clear whether this EPR sample or the whole preparation might have been anomalous or whether the pattern which this sample fits is just obscured. 111.0. Magnetic Susceptibility Magnetic susceptibility is the bulk response of a material when it is placed in a static magnetic field. All matter has a diamagnetic, or negative, contribution to its susceptibility associated with the orbital motion of its electrons. When the material is placed in a magnetic field, the electronic motion is altered such that the induced cur— rent creates a magnetic moment which opposes the applied field (87). For materials with no unpaired electrons, this diamagnetism may provide the major contribution to the susceptibility. For other systems with spin angular II A. 113 momentum only or spin and orbital angular momentum, there is a positive (paramagnetic) contribution to the suscep- tibility which may or may not be larger than the diamag— netic contribution. If it is, the magnetic susceptibility is positive and the bulk material is labeled paramagnetic. There are a number of sources for this positive contribu- tion to the susceptibility, several of which will be dis— cussed here. When an isotropic substance is placed in a magnetic 7]: h v u / fi- -—- v; + -> field H, the ratio of the magnetization M, the magnetic moment per unit volume, to the magnetic field is defined as the magnetic susceptibility x, or xV = M/H (3) in units of cm3. The susceptibility is also commonly ex— pressed as gram susceptibility Xg (cmB/g) and molar sus- ceptibility XM (cm3/mole). For a system with independent unpaired electrons, the molar susceptibility may be de— scribed by the Curie Law (A2): M _ _______ =__ u R ' XM ‘ 3kB T T ( ) where N is Avogadro's number, “B is the Bohr magneton, kB is the Boltzmann constant and p is the effective number of 11A Bohr magnetons defined by p s sIJ 0, TC is the Curie temperature of a ferromagnet above which the spontaneous magnetization (i.e., in the absence of an applied field) vanishes. For 0 = TN > 0, 116 TN is the Neel temperature above which the ordered anti— parallel arrangement of dipoles disappears. Weiss also expressed the susceptibility above the Néel temperature as C x=______ T + CNW where NW = l/2(Nii + NAB) in a body—centered cubic lattice (87). If an atom on site A has nearest neighbors on sites B and next nearest neighbors on sites A, then NAB = NBA g is the molecular field constant for nearest neighbors and for next nearest neighbors NAA = NBB = N since the same ii types of atoms occupy the A sublattice and B sublattice sites. The interaction between nearest neighbors is anti- ferromagnetic, hence NAB is positive and it is generally greater than Nii' So, ° (8) where 0 = 1/2C(NAB + Nii)° Below the Néel temperature the theory predicts that the polycrystalline susceptibility X can be expressed by p xp = l/3x||+ 2/3Xl where XII and XI refer to susceptibilities parallel and 117 perpendicular to the z-axis or easy direction. At T = OK the parallel component vanishes and K Xp(zero ) p For a metallic system with conduction electrons, Pauli explained the paramagnetism in terms of the Fermi—Dirac distribution of electron energies. The net magnetization of this conduction electron system is given by Nu2H/k T, but only those electrons within approximately kBT of the Fermi energy are likely to change spin in the presence of an applied field (A2). Only this fraction of the elec— trons T/TF contributes to the susceptibility, hence B (10) which is the Pauli susceptibility after Landau's diamag- netic correction. If the electrides may be thought of as expanded (low electron density) metals, then they would presumably follow this temperature—independent Pauli sus— ceptibility for conduction electrons. 7m 3 —-7h-r «rm- - " ' 118 111.C.l. Results For both the magnetic susceptibility and EPR experi— ments, degenerate spin levels are split by an applied mag— netic field. If it were possible to calculate the number of spins contributing to the EPR spin susceptibility, it might then be possible to estimate the bulk magnetic susceptibility for samples made from the same solutions. In this study, the number of unpaired spins in various samples was determined from spin susceptibilities by com— §_5 parison with a ruby standard. The molar susceptibility was then calculated from Equation 6: _ 0.3760A n XM_ T using the temperature-dependent fraction of unpaired spins in each sample as depicted in Figure 21. It should be noted that the fraction of unpaired spins was taken from the straight lines of Figure 21, not from actual data points. The spin susceptibility results for samples VII (sample A, R = 0.60), VIII (sample B, R = 1.57) and V1 1.15) are shown in Figures 22, 23 and 2A, (sample D, R respectively. In Figure 25 they are combined for compari- son. Results of the bulk static magnetic susceptibilities are displayed in Figure 26 for samples V11 (R = 0.60), curve A; VIII (R = 1.57), curve B; and IV (R — 0.99), 119 SPIN SUSCEPTIBILITY x I0“ (MOLAR) I I I I l I Figure 22. 40 l20 200 TEMP (K) Molar spin susceptibility of Li+C2lloe— (V11): sample A with R = 0.60. The fraction of unpaired spins used to determine XM was from the straight lines of Figure 21, not from actual data points. 2I l8 I5 SPIN SUSCEPTIBILITY x IO‘ (MOLAR) 120 I2 5 - 9 —I 6 - 3 c— I I I I I . 40 l20 200 TEMP ( K) Figure 23. Molar spin susceptibility of Li+C2ll-e‘ (VIII): sample B with R = 1.57. The fraction of unpaired spins used to determine XM was from the straight lines of Figure 21, not from actual data points. 121 7' ..I 2 51 30 - _ E '0 , ; if D _ ‘3 ). t :9 , Q.) In: 20 -» _ “J k) U) I) U) a .I. — Q. U) '0 IA _. I . J I I I 40 l20 200 TEMP (K) Figure 2A. Molar spin susceptibility of Li+C211-e- (VI): sample D with R = 1.15. The fraction of un- paired spins used to determine XM was from the straight lines of Figure 21, not from actual data points. 122 4.0 ' 3.0 r SPIN SUSCEPTIBILITY x Io’ (MOLAR) 1‘ 4 l I l l l L 20 60 l00 I40 I80 Figure 25. Molar spin susceptibilities for Li+C2ll-e" from Figures 22, 23 and 2A: A - V11 with R = 0.60; B - VIII with R = 1.57; D - V1 with R = 1.15. The fraction of unpaired spins used to determine XM was from the straight lines of Figure 21, not from actual data points. 123 u. 0 o 5£)- I _ 0 I o ' c I—- . —I 4.0 B I I ”9 . X 3()" a g c 2.0 -' 0 - I I ’ ‘ .'.. I II I ' . I' ' I l l' I . A . ' " LO ‘1 A‘A i A A“A“ ‘ ‘ ‘ A L J l I J J 1‘ , 20 60 l00 I40 I80 TEMP (K) Figure 26. Molar static susceptibilities for Li+Cle-e‘: A — VII with R = 0.60; B — VIII with R = 1.57; C — IV with R = 0.99. Note: the mole ratio of sample C is approximately the same as for samples labeled 0 in Figures 18—21, but this curve 0 represents a different preparation. 12A curve C, for which no spin susceptibility is available. Sample VI did not yield any data. It probably decomposed while being inserted into the Faraday balance Dewar. The molar static susceptibilities were calculated by Ms. Angelica Stacy at Cornell University based upon the number of moles of lithium in the samples determined during post-analysis for samples VII and VIII (Section III.E.) and by an estimate from the solution stoichiometry and sample size for IV. The bucket correction for sample C was initially determined by reconstructing and rehanging the empty bucket (Section II.C.2.a.). This was judged unsatisfactory because of the apparent presence of field—dependent paramagnetic impuri- ties on the empty bucket. Ms. Stacy then scaled bucket cor- rections based upon the combined sample/bucket mass from those for VII and VIII, and this method, though not exact, appears quite satisfactory. Figure 27 displays the reciprocal molar susceptibilities for the systems of Figure 26. Each has a negative Weiss constant: V11 (curve A): -l30K; V111 (curve B): —ll.8K; and IV (curve C): -102K. In addition the high temperature regions (> m80K) are fit by the Curie—Weiss Law rather well with approximately one unpaired electron per lithium: VII, m.9A unpaired electrons/Li; V111, .70/Li; and IV, m.9A/Li. The susceptibilities of the samples observed in the SQUID susceptometer in the MSU Physics Department were ..... 125 .‘ I .hpflwgo>fi23 Haocgoc .zompm .< mo amopQSOO mm pcm om mohswflm .om madman CH pozmadmflc mofipflfifloflpooomSm gmHoE one mo Hmoosofioom .sm ogswflm C: n=2mk ONN 0m. 03 00. cm ON _ _ _ _ _ _ . _ _ _ I ...: ....) Imo m I I I. I o. . coo I O I O C .l I . O I I I I I I I I I J @O O I I I I U II I VA... 1 I I I I #QAV W I .u I x 1| I I l O._ m. 4 I 4‘ «..d 14 4 I.N_ C ‘ A I 4 44 I v.- . .... . 4‘ 126 inconclusive. Certainly there was no strong paramagnetism in any of the samples and some appeared diamagnetic at AK. Larger samples would have been necessary for any definitive results with that system. However, after the success with magnetic susceptibility at Cornell University, no large sample preparations were attempted. There have been several attempts to fit the experimental magnetic susceptibility results for systems VII and VIII to theoretical expressions. The generalized weighted non— linear least—squares KINFIT program (88) was used for the analysis. Equations based upon Fermi—Dirac and upon Bose- Einstein statistics involvingsinglet-tripletequilibria have not reproduced the experimental shapes well, even when two different types of electron-trapping sites (evi— dent in Optical spectra and spin susceptibility studies) were employed. To date the best reproduction of the experi— mental susceptibility shape has been with an equation by Wojciechowski (89). He derived expressions for the magnetic susceptibility of a metal—metal interaction in coordination compounds with a dn — dn electronic configuration. His 1 _ d1 curves for d ,or S1 = 1/2 and S2 = 1/2 with an ex— change integral J < 0,1ook similar to both Figure 23 and curve B of Figure 26. His expression is 2 2 XM = 2kB T 3 + EXP(:J/RBT) 127 where N is the number of pgipg of interacting spins and the other symbols have their usual meaning. This expression did not fit the experimental susceptibility data for system VIII until two exchange integrals were employed, correspond— ing to two different electron trapping sites in the sample. The fractional populations ml and n2, based upon total lithium content, and the exchange integrals J1 and J2 for sites 1 and 2 were the adjustable parameters, as well as a Curie term 0 to account for the paramagnetic signal evident in the EPR spectra below approximately 6K. Table 8 lists the computer—fit values of these parameters. Table 8. Values for the parameters in the fit of the sample V111 static susceptibility results with the Wojciechowski equation for interacting spin 1/2 systems. Parameter Value i Standard Deviation 01 (%> 17.2 .i 1.1 Jl (cal) 33-5 i 1.2 n2 (%) A6.3 i 1.1 J2 (cal) 107. i 3.3 c (75) 0.20 i 0.01 Populations 01 and n2 were not constrained by making their sum equal 100%, nor were n1, U2 and C forced to equal 100%. 128 The KINFIT program determined the populations necessary to fit the experimental data. Wojciechowski's equation did not fit the data for sample VII where two sites were evident in the optical spectra and in the spin susceptibility. Using the same five parameters as for system VIII, the fit was poor. Even then, the population n1 + n2 was 8A.8%. The search for theoretical expressions of antiferromagnetically interact— ing spin systems continues. The general theory of Oguchi (90) as extended by Ohya—Nishiguchi (91) holds some promise. Basically they include longer range interactions than that of spin 1 and spin j with exchange integral J: the exchange integral J' is for spin i with all nearest neighbors ex— cluding j and also for spin j with all nearest neighbors excluding i. Ohya-Nishiguchi defines a parameter K: K = zIJ'l/lJI in which z is the number of nearest neighbors and K indicates the relative strength of the long—range molecular field. Theoretical susceptibility curves 0.0 < K < 1.5 show a broad maximum and indistinct Néel temperature (there is no Neel temperature for K < 1) similar to the curves of Figures 22 and 2A and the lower two curves of Figure 26. This similarity raises the possibility that these experimental data may eventually be fit with the Ohya- Nishiguchi expression or a similar one which incorporates multiple nearest—neighbor interactions. Without a crystal structure or prior knowledge of the number of types of -_.'.....- 129 sites available for the electron, it is difficult to apply any theoretical analysis. 111.0.2. Summary and Discussion The spin pairing phenomena detected by EPR were also evident in the static magnetic susceptibility studies. Table 9 contains the values of maximum susceptibility for the spin and static susceptibility samples and the tempera- ture at which each maximum occurred. None of the sus— ceptibility curves has a sharp transition point (Néel tem— perature) at which antiferromagnetic alignment occurs, al— though sample VIII is close to that appearance, and none of the curves show x(0)/x(TN) = 2/3, as eXpected for a poly- crystalline antiferromagnetic powder. The static sus- ceptibility maxima are broad as might be expected where spin pairing is occurring gradually instead of at a specific transition temperature. The static susceptibility curves, then, are the result of the interplay of two factors: the susceptibility of the spins present is enhanced as the temperature decreases (x a l/T), but there are progressively fewer free or unpaired spins to contribute to the suscep- tibility. There is definitely an antiferromagnetic inter— action, but not in the Néel sense. The similarity of the general temperature dependence of spin and static susceptibilities for sample VII and for sample VIII is reassuring. To be sure, the results are not 130 Table 9. Summary of Li+C211°e- spin and static suscepti— bility data. Maximum XM Temperature (K) Sample R (esu/mole x 103) of Maximum Spin Susceptibilities VI 1.15 3.71 ' 13 2.AA m85 VII 0.60 1.A2 17 0.93 W70 VIII 1.57 2.11 13 Static Susceptibilities IV 0.99 1.63 6A VII 0.60 W0.89 “70 VIII 1.57 5.A2 21 identical, nor are they expected to be. Since EPR probes the local environment while susceptibility studies the bulk property, and since the EPR samples were essentially thin films while the susceptibility samples were bulk powders with approximately twenty times the mass of the EPR samples, differences are expected and it is gratifying that the results are so close. Optical spectra and spin sus— ceptibility indicated that system V111 with R = 1.57 had primarily one type of electron site in which spin pairing occurred. However static magnetic susceptibility indicated 131 two sites. It is interesting that EPR spin pairing showed an interaction energy of 26 cal/mole while Wojciechowski's equation for antiferromagnetic interaction yielded an energy of 33.5 cal/mole at low temperatures. On the other hand both the Optical spectra and the spin susceptibility results for system V11 with R = 0.60 showed at least two sites for electron localization. However the static mag— netic susceptibility showed only one broad maximum at m7OK for sample VII. EPR spin pairing indicated an inter- action energy of 117 cal/mole while the (poor) fit with Wojciechowski's equation yielded an approximate interaction energy of 237 cal/mole. It is likely that more than one site is involved in both cases. The presence or absence of the various spin pairing sites cannot be explicitly explained. Rather the dif— ferences are not unexpected for samples of different sizes, different shapes, with possibly different grain boundary effects and probed with different experimental techniques. The spin and static susceptibilities of samples IV, VII and VIII show that the temperature of maximum sus— ceptibility decreases with increasing lithium to cryptand ratio and the susceptibility at the maximum increases. The spin susceptibility for V1 with R = 1.15 indicates that it might be an exception to both of these trends, though there is no obvious reason why this is so. From spin susceptibility and the Wojciechowski equation fit, the 132 higher the sample lithium content,the weaker the inter— action energy, again with the exception of system V1. System V1 with R = 1.15 showed unexplained behavior as noted in Section III.B. and no explanation is evident after the spin susceptibility determination. 111.D. Conductivity The conductivity of several lithium electride samples was investigated, some by microwave conductivity and one by D.C. conductivity. One of the first indications of the metallic character of Li+C2ll-e- was the microwave conductivity of sample 1. At —50°C it possessed a conductivity between that of zinc and insulators. Later liquid nitrogen providing the cool— ing boiled away, causing the sample to warm. The over- sight was discovered and the sample was rechilled, but not before the microwave power absorption at —l5°C increased to that of zinc or slightly higher. The sample did not appear to decompose and later the —50°C relative conduc— tivity was repeated. Sample 11 showed the MNM transition evident in optical and EPR studies. At -A5°C the conductivity increased significantly, showing microwave power absorption equal to or greater than that of an equal volume of palladium powder, as shown in Figure 28. This rapid change in conductivity was repeated many times both as the sample temperature was 133 2007— I80— l60 — ' Power Transmitted (,IW) m CD 6 I?) 3 (D C) (D C) (D I I I I I II 43 C) I II ‘zis‘ 3' O I I I 1- J II I L. I L |80 I90 200 2l0 220 230 240 Temperature (°K) N C) | Figure 28. Microwave power transmitted by equal volume samples of Li+C2ll-e‘ (II) (circles) and palla— dium (squares) in a TE103 cavity, 13A raised and lowered. There was a slight hysteresis in the transition, but in general the change was very reproducible in that sample. It is possible to calculate the conductivity, or at least the limits of conductivity, for the Li+c2ll-e‘ (II) EPR sample from its response in the EA spectrometer micro— wave cavity. Penetration of a radio—frequency field into a metal is limited by the material's skin depth, 0 (l9). Classically the skin depth is d = (c2/2'IIOIII)1/2 (11) where o is the conductivity and w is the angular frequency of the r-f radiation. Limits can be placed on the Li+Cle°e- conductivity by comparing it with the response of a well characterized metal such as silver. Then Equation 11 becomes fl; : (G2w2)l/2 2 °1°l If the skin depth of silver is taken at the same frequency as that in the EPR cavity (m9.2 GHz), then = (:1)2 (12) O2 °l 52 At 9.2 GHz, the skin depth of silver is .7p (83) and at 135 295K the conductivity is 6.2 x 105 (ohm—cm)_l (A2). In the metallic region, the A/B ratio for Li+C2lloe- is mlo5 which yields W1.25 for the ratio a/d in Webb's theory for the EPR spectra of conduction electrons where a is the spherical particle radius (21). The limits on the con- ductivity then arise from the estimates for a. At the lower limit, a may be mlp as observed under a microscope or at the other extreme, the whole sample may act as one large conducting particle. In that case a = 9A5u for the EPR sample. With 9A5u 3 a 3 10, then 756p 3 0 3 .80. Therefore 2 54.1.. __ "1 o 3 (6.2 x 10 )(756) (ohm cm) so —1 0.3 0.5 (Ohm—cm) and o _<_ 5. x 105 (ohm—cm)-1 assuming the conductivity of silver is essentially un— changed at 228K. EPR samples are normally thin films on the tubing walls. This appeared to be true for sample 11 also. With the assumption that the sample was spread out uniformly on the EPR tube walls and using the measured tube I.D. and the sample mass and height in the tube, the sample thickness was approximately 76p. With these 136 assumptions the lower limit of the conductivity becomes 8. x 101 (ohm—cm)'1. Li+C2ll°e_ sample 111 was definitely an insulator in the region —68° to —AA°C according to its microwave con— ductivity. System X was also non—metallic as determined from the D.C. conductivity of a packed powder sample. The current at various temperatures was read and converted to resistance. Figure 29 is a plot of log R against recipro— cal temperature. An HP—65 calculator fit of the right- hand line yields log R = —5.3A + 2.A7 x 10+3 (1/T), giving a band gap of 0.99 eV. From 0 = %% or (%)% where L is the length of the conductivity sample and A is the area of the electrode, 0 = 5.1 x 10—5 (ohm-cm)_l at —A5°C for sample x. For 298K, o = 1.8 x 10"3 (ohm—cm)‘l and at infinite temperature Go = 3.6 x 10+5 (ohm—cm)—l. The band gap for the other straight line is 1.28 eV. The high temperature segment of Figure 29 (> -32°C) shows an increasing resistance with increasing temperature. That could be because the sample underwent a MNM transition at —33°C or because it began decomposing there. When the sample was removed from the conductivity apparatus at -10°C, it still appeared to be blue, though that would not rule out the presence of sufficient decomposition to raise the resistance less than an order of magnitude. In summary, Li+C211-e_ samples I and 11 showed metallic character by conductivity, though 11 was metallic only 6.7 - _ ' IV 6.5 - '/ . ' _ 6.3 -- '/ ./ .. 0: I 8" 5.9 — I / J 'I .J ' . /- / I I 5.7 " '. /' / - I I I '5 / . 5.5 — 1. . . / .. "N .-' «IV: 5.3 - ‘h/ . - J l 11 J l J, l 3.70 4J0 4.50 4.90 I/T (K")' I0’ Figure 29. Temperature—dependent resistance of Li+C2ll°e— (X) with R = 0.96. The region on the left above -33°C with 3R/3T > 0 may either indicate the sample has decomposed slightly or has under— gone a MNM transition. 138 after a transition at —U5°C. The conductivity limits at -45°C of EPR sample II, 8 X 101 g o 5 5 x 105 (ohm-Cm)-l, places it between a near—metal and an excellent metallic conductor. Sample III was an insulator by microwave con— ductivity and sample X was a low band-gap (0.99 eV) semi— conductor by D.C. conductivity. As previously described, there is no obvious reason why the character of these R 2 1 samples should have varied so widely. However, the conductivity measurements correlate well with the optical spectra. High conductivity occurs with samples whose optical spectra indicate plasma character. III.B. Sample Analyses A series of tests was performed on the decomposed magnetic susceptibility samples VII and VIII to precisely determine the amount of material present in each. Then the same tests plus an additional one were conducted on (presumably) undecomposed magnetic susceptibility samples VII and VIII used in the MSU Physics Department. This additional analysis shows the amount of reducing power still present in samples handled similarly to the sus— ceptibility samples VII and VIII. When samples VII and VIII were removed from the Faraday balance at Cornell University, each appeared white. But after three weeks of room temperature storage, the material seemed to be a yellow—brown viscous liquid with a reddish 139 tint, as though it were primarily impure cryptand. Table 10 lists the results for these samples. Table 10. Results of the analyses of magnetic suscepti- bility samples VII and VIII. VII VIII Moles Li from flame emission 5.25 X 10—6 4.93 X lo—6 Moles C211 from titration 9.85 X 10-6 3.11 x 10—6 Total sample mass (mg) 2.8“ 0.93 Estimated sample mass (mg) 3.“ 1.1 Error in mass estimate + 20% + 18% Mole ratio Li:C2ll, R 0.53 1-59 Estimated R 0.60 1.57 After the flame emission, each solution was dried and then the samples were dissolved in D20 and t~butanol. Proton NMR graciously accomplished by B. Van Eck was not of sufficient quality for an accurate integration because of very low cryptand concentrations. However qualitatively both spectra showed that the C211 was substantially in- tact. Signals that could have been from decomposed cryptand were very minor. This indicates that thermal decomposition 1A0 in the absence of any external oxidizing agent may occur with very little rupture of the cryptand. Table 11 lists the results for samples handled similarly to those of Table 10,but which were not intentionally decomposed before the hydrogen evolution was performed. Table 11. Results of the analyses of MSU magnetic sus— ceptibility samples VII and VIII. VII VIII Moles Li from H2 evolution 7.9 x 10‘6 A.3 x 10’6 Moles Li from flame emission 5.03 x 10"6 7.29 x 10—6 Moles C211 from titration 8.55 x 10_6 5.05 x 10—6 Total sample mass (mg) 2.50 1.51 Estimated sample mass (mg) 2.9 1.8 Error in mass estimate + 16% + 19% Mole ratio Li:C2ll, R g 0.58 1.uu Estimated R 0.60 1.57 Tables 10 and 11 show some interesting results. In Table 11 the number of moles of reducing species, presumably electrons equal in quantity to the moles of Li, varied considerably. The measurement of the amount of H2 evolved contained the greatest error of all the results. The gas was collected in a 10.00 m1 volume, creating only a pres— sure of several torr because the samples were so small. lUl This H2 pressure was the difference between two large num- bers, yielding a large percentage error. It should be noted that the sample sizes were an order of magnitude smaller than those usually analyzed in this laboratory. The results are not inconsistent with the assumption that both samples were substantially undecomposed before the analyses were begun. The inference is, then, that the magnetic susceptibility samples at Cornell University were substantially undecomposed during the measurements. There- fore most of the electrons potentially present were par- ticipating in the spin—pairing process in those samples. The estimated sample masses appear to be systematically W18% high. The reason is unknown but the effect is not substantial. If the same error existed in the EPR sample sizes, then the actual number of unpaired spins for the samples described by Figure 21 would be increased 2 - 5% at high temperatures. Finally, the estimated sample mole ratios are close to those from the analyses. It is not surprising that three of the four actual ratios are a bit lower than the estimates. If decomposed, uncomplexed C211 were present in the sample, it would still gain protons during the titration and be included in the moles of 0211. But if the Li had been contaminated prior to the preparation or if the reducing power had decreased during the sample lifetime and if either situation had created NH3- or H20- insoluble products, the Li flame emission result would have 1M2 been low. Hence the R value would be low. The conclusion, then, is that the Li was nearly free of contamination. III.F. Li+C2ll-e_ Summary and Conclusions Some of the results for lithium electride systems from ammonia are briefly summarized in Table 12. One of the major thrusts of this study was to determine the consistency of results from samples prepared from the same solution but analyzed by different methods. This correspondence is quite good as typified by the internal consistency of samples VII and VIII. Also, the consistency between dif— ferent preparations with the same mole ratio seems good, as in the optical Spectra of III, IV and V. So, in the future, individual preparations for EPR and magnetic susceptibility can be made to optimize the sample size with an assurance that the results will be fairly repre- sentative for a given mole ratio. The second major objective of this study was to charac— terize Li+C2ll°e_ and the results are quite interesting. The material apparently consists of large cryptated cations of about 83 diameter and interstitial electrons when R i 1. At higher values of R both Li+ and e— are in the interstices, but they do not form lithium metal. The system has a low electron density near that required for a metal-nonmetal transition. The metallic region is ap— proached when the mole fraction of lithium is raised to 1A3 mpHSmoL mmm cogs woman COfipmepmo>cfi gocppsm mpcmmnmz OHPwp mHoE wasp oocwfimp hwompmm SH oomom IEoooo mwpmo oc mgpoodw umosoggmg mosfim>lm Ugo ofloon m\< pnonwag mpopflmac: mcflam no & pnogman mH.H H> .QoLQ CH oomoppmop H 8 m Low Upmocmpm wm.o Mzm me .owdm mzaso opflm mac» n: owhma 00p oHQEmm mm.o >H HHonoo pm oomOQ IEoooo mmpwo oc omgma 00p oHQEmm oxHHImEmeQ oLoEIpoEpmz moppomam QESQ IHpHSE phmpcmpm mm.o HHH Oozzl pm COHpHmehp 222 .poSUgoo o>mzogofiz mcflmm Umpflmmgs R 30H Oomzl #6 COflpflwGQQP ZZZ Domzl Pm coflpflnconp 222 mm.o HH omega: non: oflfiaoooz mzz w E Cmmzpmn .pUSUQOO ®PNSOQOHE osm. n o1om< o.H8 m\< mooomfin on gfico oflaampoe mm.o H mangEoo mopwm gospo sonaflofloooonsm aflposwmz onoooom mmm mppoodm Hmoflpao oSHw> m w % Empmmm .mEopmmm no.HHmo+flq Log mpmo ozp mo mhmEESm .mH mfint HIAEoIEQOV oocmnhomnm mica x m n VHmmmo mH 30H I mponpo Ham wa Ucmn MCOflpHmcwhp omml\oaml >m mm.o .posocoo owoo mm.o poUZOQ ooxowd llllllllllllllllllll mEmeQ oomm:\owwl x u. x w NH "gonna onoonoon goomou\ooafil No.0 .m onSnnono nooo> lllll sanflom oomm1\ofimn uuuuuuuuuuuuuuu xH mEopmmm Meme xme apes opfiw oco meEop omoo mammaa H a m soap psopow .omdm mopmcHEooopm 30H pm wgpooaw I .ggm «Ego: mm.a Imflp thQQOngngm open damp 30H msooComoEos pmoE mmsooCoonog HHH> Mia m Sufi: mEopwzm moss xme omzm oppoodm H a m @559 00.0 oxfla zapcopmflmcoo maco opflm damp H: oxfia nose IHpHSE ogmocmpm HH> 1w mpCoEEoo hpflaflpwpaoomzm oppoodm mmm appoodm Hmofiudo osfim> m w mopom nonoo oflposmoz % aonngm .oozcflpcoo .mH ofinmE 1A5 ml.5 and the system is metallic at R = 2. One system at R = l was also metallic and another existed so near to the metallic state that warming the samples (optical, EPR and microwave) above -A5OC caused the MNM transition. It is still unclear what caused these two systems to be metallic while so many others with the same mole ratio were not. In the nonmetallic systems the large inter— electron distances result in weak electron—electron inter— actions which lead to spin pairing as the temperature is lowered. In the high temperature region there is approxi— mately one unpaired electron per lithium atom, indicating that essentially all of the electrons in the sample are participating in this spin-pairing process. This is perhaps the most encouraging aspect of this study since it proves that the properties attributed to electrides are not just those of a minor constituent. CHAPTER IV OPTICAL SPECTRA OF OTHER SYSTEMS Many systems other than Li/C2ll were studied, though none as thoroughly as this metal/complexer combination. Optical transmission spectroscopy of thin films was the method for study, chosen because of its relative ease and because of its apparent ability to accurately depict, al— beit qualitatively, the nature of the solution or solid being investigated. As noted in this chapter and in the previous one, the initial indications of transmission spectroscopy were corroborated by any other method used. Thin film optical transmission spectroscopy as used in this study was not without its problems, however. The splashing, rapid solvent evaporation method of film forma— tion produced heterogeneous films of probably varying thicknesses, geometries and degrees of coverage of the Optical beam path. As detailed in the Appendix, peak positions should be unchanged by these effects, but ampli- tudes are decreased and peak widths are broadened, the extent of which is dependent upon the film's deviation from uniform thickness over the optical beam area on both cell walls. Particularly in K/C222 systems, but in other 1A6 1A7 systems from ammonia as well, apparently solvent—free dry films showed changes with time, and required up to forty minutes to reach stable time-independent values. In several cases excessive light scattering from isopropanol on the outside of the optical cell was noted, especially when the sample cavity was —60°C and lower. Also, it was possible that differential rates of precipitation changed the stoichiometry between the initial portions of the film and the regions from which final bulk solvent evaporation oc- curred. To guard against such artifacts, only reproducible spectra are reported. Throughout this chapter when the mole the system is slightly cryptand ratio is given as R — 1, rich with R g 1.0. IV.A. Spectra in the Absence of Complexer IV.A.l. Na and K with Ammonia Before obtaining spectra of alkali metals in the pres— ence of complexing agents, it seemed worthwhile to ascer— tain the effect of alkali metals alone and then in the presence of a non—complexing diluent. To determine if the alkali metals might be responsible for the observed absorp— tion spectra of concentrated metal—ammonia solutions (MAS), sodium and potassium films were prepared by evaporating the MAS to dryness on the optical cell walls. When the bulk solution in the apparatus sidearm was frozen, the MAS film 1A8 in the optical cell lost its characteristic blue color in 2 - 3 seconds, resulting in an inhomogeneous "film" which appeared to contain lusterless gray-white flecks of metal. Spectra of each metal were virtually flat without any absorption bands. There was a pronounced absorption, however, when these dry metal "films" were damp with am— monia as shown in Figure 30 for sodium (spectrum A) and potassium (spectrum B). The bulk solution temperature was within 10°C of the film temperature before the absorption was noticeable and within 3 - 5°C before the absorption was sig— nificant. This indicates that sodium and potassium have low ammonia affinities compared to lithium and that Na(NH3)X and K(NH3)X compounds are not readily formed, if they are at all. The two spectra in Figure 30 are the first transmission spectra of concentrated MAS. The general shape is quite similar to reflectance spectra of sodium in ammonia in Figure l by Beckman and Pitzer (7). They found an ex— tremely sharp dr0p in reflectance (a plasma edge) at ap- proximately 7700 cm—1 (1300 nm) for solutions between 15.A and 7.0 mole percent metal (MPM). In addition the general shape is similar to that calculated from reflectance spectra of 1A MPM (Li) solutions in ammonia by Thompson (92). Figure 30 spectra have a more gentle descent which occurs at shorter wavelengths than in either reflectance study. This more gradual lepe could be the result of film 1A9 A(pm) l 0 2.0 l3 l0 0.7 0.6 0.5 0.4 . 0,. I l ' f I T I A AMAx 0.5 r 0.0 L ------ L .. 5 l0 IS 20 25 '9' (cm") ° IO" Figure 30. Spectra of metal films which are damp with ammonia (concentrated M—NH3 solutions): A — Na; B — K. 150 inhomogeneity in this study. The continued rise into the infrared, also noted by Beckman and Pitzer, contrasts sharply with the localized absorption peaks attributed to trapped electrons in later spectra. The spectra in Figure 30 for concentrated MAS were reproducible and will be considered typical of conduction electron plasma absorp— tion. IV.A.2. Na/DABCO Films from Ammonia Optical films of sodium and the non—complexing bi— cyclic diamine DABCO, N(CH20H2)3N, were also studied. Am— monia was evaporated from a film of approximately R = 2 (ratio of moles metal to moles complexer, or in this case diluent) and no absorption bands were evident. Even a wet film of this system had little character. Presumably the diluent dispersed the metal sufficiently to affect the plasma absorption, though the general shape was as de— picted in Figure 30. It is apparent, then, that neither the presence of an alkali metal nor a bicyclic diamine with an alkali metal can account for dry film absorption bands. 151 IV.B. Films from Ammonia IV.B.l. Na/C222 Systems An ammonia solution containing sodium and C222 in the ratio R = 1 gives the dry film Spectrum A in Figure 31 at -U9°C. The predominant peak at 8500 cm-1 (1175 nm) is a locally trapped electron (eg) band while the other peak at 16,100 cm-1 (620 nm) is due to Na-. A dry film re-formed by removing NH3 from a wet film yielded broader peaks which were slightly red Shifted compared to those of spectrum A. Two different dry films initially showed a small, separate e; peak at 5000 cm'1 (2000 nm) which decayed within fifteen minutes. As it did so, both e; and Na- peaks increased in amplitude by approximately 40%, reflecting a probable re- distribution of electrons from low energy traps into deeper, more stable traps. That both e; and Na” peaks exist in these spectra is somewhat surprising for an R = 1 system and is indicative of incomplete complexation of Na+ by 0222. The solution was prepared with a nominal 11% excess of 0222 over the amount of Na, thus increasing the probability that this equilibrium was shifted to the right: —-> + — Nao + C222 + Na C222 + esOlV However there were still uncomplexed sodium cations present 152 X(ym) IO 2.0 1.3 LO 0.7 0.6 0.5 0.4 . l/f‘\E- l I l I l ,5 ‘3 a, . \ :1 . ‘3 L .- \‘z IKNUUK g \é ° \1 0.5 \. I A O..\\ O. . 5 l 1 l0 l5 7 (cm")°l0" Fi ure 31. g A — dry; B — damp; C — wet. 20 25 Spectra of Na/C222 films with R = 1 from ammonia: 153 in solution which acted as electron scavengers during flash solvent evaporation, yielding the Na— absorption peak. Possibly the sodium cations may have become uncomplexed during the 10 — 15 second solvent evaporation process, but this seems unlikely in light of thermodynamic evidence from NMR on alkali metal decomplexation rates in numerous sol— vents (93). For example extrapolation of aqueous solution data to —u5°c yields a complexation half—life for Na+ into 0222 of 6 x 10"5 sec and a decomplexation half-life of 27 sec when using an equilibrium constant obtained from a solution of 5% water in methanol (52). Although the current study is in ammonia, complexation should still be favored over decomplexation by a wide margin. Spectral behavior during decomposition supports the sodium scavenger idea. As the solution aged, decomposition most likely destroyed some of the cryptand, thereby releas— ing more Na+. The ratio of absorbances ANa_/Ae_ increased t steadily from 0.6 to 2.0 during this time, as expected for a system in which metal concentration was increasing relative to cryptand. Damp and wet films for Na/C222 with R = l are depicted as spectra B and C respectively in Figure 31. The damp -l spectrum has a single localized electron peak at 7800 cm (1280 nm) with a slight low energy shoulder, while the wet —I . film shows an electron peak at 6200 cm (1615 nm) With very high absorbance extending into the infrared. These 15A films which contain ammonia will be discussed collectively With those 0f1k10222 for R = 3 and u later in this section. Solutions containing Na/C222 with R = 2 have been reported elsewhere (61) and will only be described here in comparison with solutions of different R values. To ascer— tain the effect of amounts of Na in excess of R = 2, experi— ments were conducted with R = 3 and R = A. The dry film spectra are depicted as curves A in Figures 32 and 33 for R = 3 and A, respectively. Both curves have a typical peak attributable to Na— which is almost identical to that found in methylamine (58) and in NH3 (61) with R = 2: a major peak at l5,A00 cm.1 (650 nm) with a pronounced high energy shoulder and a small low energy shoulder and a distinct small peak at 25,000 cm—l (A00 nm). A surprising feature of the R = 3 and A spectra is the continued existence of e; peaks. Perhaps sodium cations and monomers are not as effective at trapping electrons as first thought. Even though there are nominally twice as many sodium atoms present outside the cryptand in the R = 3 films as there are electrons, there still must be non—anionic traps which are more stable for the electrons. As a result the initial dry R = 3 films have a substantial e; peak at 8100 cm’1 (1235 nm) and a small peak at 5500 cm‘1 (1820 nm). With more sodium present in the R = H films, however, the relative intensity of et peaks is decreased, as the excess sodium perhaps traps more of the available electrons. 155 X(pm) 2.0 L3 LO . 0,7 05 IO r , . I AMAx B l 1 1 00 5 K) IS '17 (cm")°l0" Figure 32. Spectra of Na/C222 films with R = 3 from am— monia: A — dry; B — wet. 156 2.0 1.3 LO 0.7 0.5 0.5 0.4 l.0 'ii ( cm") ° IO" Figure 33. Spectra of Na/C222 films with R = A from am— monia: A - dry (fresh); B — dry (annealed); C - wet. 157 In the R = A initial dry films, the e; peaks are at 7800 cm—1 (1280 nm) and 5100 cm—1 (1960 nm). In both R = 3 and A films there are reproducible, sig— nificant changes with time. A typical annealed R = A spectrum is shown in Figure 33 as curve B. In this case the Na_ peak is slightly narrower but otherwise unchanged and the high energy e; peak has shifted to 6800 cm-1 t cm—1 (2600 nm) though it is somewhat obscured by instru- (1A70 mm). The low energy e peak is at approximately 3850 mental peaks at that location (not depicted). The R = 3 annealed films, while not pictured in Figure 32, show similar movement of the e; peaks: 6900 cm_1 (lASO nm) with a shoulder at A500 cm—1 (2220 nm). These peak positions for both R = 3 and A films represent the extremes of observed e; peak movement. Virtually all positions between these extremes were observed while the Na- peak position and absolute magnitude were constant. The excess sodium perhaps provides an inhomogeneous environment in the dry films such that multiple electron trapping sites exist which are of nearly the same energy. Electron movement between these sites is quite easy, though the cause of such travel is unknown. Temperature changes of the film had no ap— parent effect on the direction or rate of peak movement, nor did photolysis. It should be noted, however, that the Beckman DK—2A tungsten lamp was the photolytic light source and that it took 30 — 60 sec to reset the DK-2A 158 from the particular bleaching wavelength and scan the region of interest. Spectra of films of all three R values are similar when the films contain ammonia. As the temperature of the bulk solution in the sidearm is raised increasing the NH3 vapor pressure, the immediate consequence is the disappearance of the Na_ peak followed by a significant,broad increase in the infrared absorption. A film of R = 3 which contains nearly enough NH3 to wash off the optical cell walls (wet) is shown as spectrum B in Figure 32. Wet films for R = l and A (spectra 0 in Figures 31 and 33) appear to have con— siderable plasma absorption. While the absorptions do not continue to rise into the infrared as in wet metal films (Figure 30), they are characteristic of wet alkali metal/ complexing agent films and will hereafter be referred to as having plasma character. These spectra are quite similar to those Of Na in NH3 between 5.6 and 2.5 MPM by Beckman and Pitzer in Figure l (7). From the plasma edge of higher MPM spectra, the reflectance between 5000 and 10,000 cm-1 (2000 to 1000 nm) drops significantly in the lower energy region until a peak is formed, although there is still considerable reflectance on the infrared side compared to the higher energy side. Possibly the conduction electrons now have a shorter mean free path due to the presence of excess NH3 and are no longer in the conduction band. This change is gradual, though, on the macroscopic scale as 159 depicted in Figure 1. It may be that the height of the 1 in Figures infrared absorbance of wet spectra at A000 cm— 31 and 33 depicts a similar decrease in the conduction character due to the increased presence of NH3 in the film's structure. IV.B.2. Na/C22l System Because C221 is the optimum cryptand for complexation of sodium, rather than C222 (Figure 3), a solution of Na/ C221 in NH3 with R = 2 was prepared to determine the effect cryptand cavity size might have. Spectrum A (Figure 3A) indicates that the effect may be very significant. Absent are the high and low energy shoulders on the Na" peak as well as the small peak at 25,000 cm‘l (uoo nm). Additionally, there is absolutely no indication of any peak attributable to et nealing effects were noted, no peaks shifted position with in any of the solvent—free Na/C22l spectra. No an— film temperature changes or with time and no other peaks were observed from 220 - 3200 nm. The lack of these phenomena indicate that no other traps compete favorably with sodium cations for electrons when solvent is removed from the films. The dry film structure may be fairly homo- geneous, implying that the Na+ encryptation by C221 is sub- The Na" peak at lA,A00 cm.1 (695 nm) stantially complete. —1 is quite broad and is red shifted by 1000 cm or more from the similar peak in Na/C222 spectra. This broadness may 160 X(pm) 2.0 L3 l.0 , ID I -./...\ I 07 0i6 01.5 01.4 1’ '-.\ I =3 / ’2 / \ I ‘2 A [I \‘z A AMAx I ‘I. l \‘a . \ \\ 3. 0.5 " \ \ . \ . B \ \ -. \ . \ \\ . C \\ \\ ° \ 1 J 0.0 5 l0 l5 7 (cm")°lO" Figure 3A. Spectra of Na/C22l films with R = 2 from am- monia: A - dry; B - damp; C — wet. 161 be due to the unsymmetrical nature of C221, resulting in a range of similar but not identical locations for the sodium anions. Spectra of Na/C22l films with R = 2 which contain am- monia are shown as spectra B and C of Figure 3A. They are typical films damp and wet with NH3, All three spectra have an underlying rising absorbance which may have been caused by light scattering from the sample cell but which were more likely the result of poor spectrometer adjustment. This rise is an artifact and should be disregarded though it is evident only in spectrum A. IV.B.3. K/C222 and K/C2N22 Systems A previous study in NH3 showed that K/C222 with R = l annealed from a predominant K— peak at 11,200 cm"1 (890 nm) to strong plasma character (61). This experiment was repeated to verify the plasma character: the annealed film does indeed show nearly identical plasma character (Figure 35, spectrum B), but the fresh dry film is sig— nificantly different than that previously reported. The major feature of the fresh dry film (spectrum A) is a localized electron peak at 6500 cm"1 (15A0 nm) with only a very slight shoulder at 10,500 cm_1 (950 nm) WhiCh is prob— ably due to K-. The difference is most certainly caused by the mole ratios of K and C222 in the two studies. Though both nominally had a ratio of R = l, the prior study actually 162 X(ym) 2.0 l3 :0 0.7 0.6 0.5 04 ID If I I , I I I A AMAx 0.5- 00/1 I 1 1"-....1 ° 5 IO I5 20 25 “:7 (cm") - IO" Figure 35. Spectra of solvent—free films of K/C222 with R = 0.95 from ammonia: A - fresh; B — annealed. 163 had R = 1.05 (9A) while the current ratio is R = 0.95. Thus it is understandable that the previous fresh dry spec— trum would show a larger peak due to K_. It is interest— ing, however, that both systems anneal to yield spectra which are virtually indistinguishable despite the mole ratio difference. The study of K/C222 in NH3 with R = l was repeated at R = 0.9A. The spectra of Figure 36, though presented on an absolute absorbance scale, are virtually identical to those of the R = 0.95 system. Figure 36 depicts the changes occurring in a film at —A9°C over an 80 minute period after subtraction of the quartz cell background from all spectra. The same film, dampened with NH3 vapor, changed immediately back to a localized electron system (spectrum D, Figure 37), followed by the appearance of typical plasma character in spectra E and F. A limited study of potassium electride was undertaken to confirm by other means the possible metallic character indicated by dry annealed thin films. A K+0222oe— micro— wave sample at —710 and -50°C appeared to be strongly con— ducting. Another microwave sample, from the same solution as the films whose spectra are shown in Figure 35, was an insulator at —80°C. Unfortunately the tube was broken before the temperature could be raised to m—A5°C where annealing may have occurred. Details of the response of an EPR sample from this same solution are reported by Figure 36. Absorbance 16A A (pm) 20 [0 07 0.5 0.4 ° 0~o.~.-—f’ ‘5 :0 :5 20 25 '37 (cm"): IO" Unnormalized spectra of a K/C222 solvent—free film with R = 0.9A from ammonia. Spectra were recorded at the following times after solvent removal: A — 2 min (fresh); B - 28 min (inter— mediate); C — 81 min (annealed). 165 A (pm) 2.0 [P 0.17 0.5 0.4 I "000. LC Absorbance 5 l0 I5 20 25 ’ 7 (cm") ° IO" Figure 37. Unnormalized spectra of the same K/C222 film from which the spectra of Figure 36 were ob- tained: D - damp; E — intermediate; F — wet. 166 DaGue (6A). The sample did not appear to be metallic but an unusually low percentage of electron spins seemed to be unpaired. An EPR sample from the same solution as the film whose spectra are in Figures 36 and 37 was too large for the EA AFC. Since it nominally contained fewer spins than the previous K+C222-e" sample, this may indicate its conductivity was somewhat higher. The data collected to date on K+C222-e— systems concerning their metallic charac- ter are inconclusive. Professor J. M. Lehn gave the Dye research group a quantity of C2N22, the tetraza analog of C222 in which both oxygens of one strand are replaced by nitrogens, depicted in Figure 2. Curve A of Figure 38 shows the spectrum of an ammonia—free fresh film of K/C2N22 in which R = 0.91. At 1 t (1515 nm). Considerable potassium must be uncomplexed to —39°C a broad peak due to e is centered at 6600 cm— give the additional peak at 11,700 cm‘1 (855 nm) which is due to K’. Over a 70 minute period the fresh film annealed, yielding spectrum B at -35°C where the K_ presence is even more pronounced. Peak positions are nearly unchanged at 11,800 cm"l (8A5 nm) for K“ and 6300 cm‘1 (1590 nm) for eg. The change in relative peak magnitudes apparently occurred independently of numerous temperature cycles between —30° and —60°C during the 70 minutes. Because the tetraza C2N22 gave a mixed alkalide/electride spectrum rather than a pure electride absorption when there was excess cryptand, its properties in ammonia were not investigated further. 167 A‘pm) 20 |3I|0 07 06 05 (M4 L0 I {3 I I AMAx B 0.5 -:.:"° I L 1 -nm" v 00 5 l0 BI 83 25 i7 (cm")-IO" Figure 38. Spectra of a solvent—free film of K/C2N22 with R = 0.91 from ammonia: A — fresh: B — annealed, 70 min after spectrum A. 168 IV.B.A. Rb/C222 System Curve A of Figure 39 is the initial spectrum of a dry Rb/C222 film with R = l. The major feature is an e; peak at 6750 cm.1 (1A85 nm) while the Rb“ absorption is just a shoulder at 12,050 cm_1 (830 nm). Over a period of A0 min, the spectral shape gradually changed to that shown by curve B in Figure 39. With the optical cell maintained at —50°i:2°C and the bulk solution in the sidearm at liquid t creased by 25% and the peak shifted to 5600 cm"1 (1785 nm), 1 nitrogen temperature, the magnitude of the e peak de— a red shift of 1100 cm- Meanwhile, the absolute ab— sorbance of the Rb” peak remained nearly constant. This appears to be another C222 system in which complexation of the alkali metal cation is not quite complete. Spectrum C of Figure 39 is that of a semi-wet film which shows a significant, broad infrared absorption that appears to result from the superposition of solvated electron bands with a plasma absorption. IV.B.5. Cs/C322 Systems This is the first report of the use of 3,2,2-cryptand as the alkali metal complexing agent in optical spectra. The previously reported dry film spectrum of Cs/C222 with R = 2 had four peaks of nearly the same amplitude: one attributable to e— and three in the region of Cs— (61). t 169 AIpm) '0 2.0. L3 |.0 0.7 0.6 0.5 0.4 - ’I’“\\I I I I I .5’. \ ° \ I’ A \C A : B \ AMAX. \ °. \\ \ 0.5 a, \ 0.. \ ....... \ ° \ \ \ \ .. \ \ o. \ \ \ .\ OXV. l I I 1 1' ~r 1-1.. 0.0 5 I0 I5 20 25 '17 (cm") - IO" Figure 39. Spectra of Rb/C222 films with R = 1 from am- monia: A — dry (fresh); B - dry (annealed), A0 minutes after spectrum A; C - semi-wet. 170 NMR studies have shown that C222 forms both inclusive and exclusive complexes with Cs+ in solution (81,95),so the films from NH3 may have had a variety of sites for e; and Cs”. On the other hand NMR has shown that complexes of 0s+ with C322 are only inclusive (96). The 0322 cavity size is more compatible with the Cs+ 3.33 radius (Figure 3). An experiment was conducted to determine if Cs/C322 films from NH3 with R = 2 might be more homogeneous than cor— responding C222 films. The improvement appears to be only marginal. Curve A of Figure A0 shows a very broad peak in a fresh dry film. The major peak at 7800 cm—1 (1280 nm) and the shoulder at 6500 cm’1 (l5A0 nm) are probably due to e; species while the shoulders at 9800 cm‘1 (1020 nm) and 13,000 cm‘1 (770 nm) are thought to be caused by Cs’ species. An interest— ing change occurred in this spectrum when it annealed: a — l t (1615 nm) while a Cs- peak at 10,000 cm‘1 (1000 nm) and a single e peak became predominant and shifted to 6200 cm— shoulder at 12,500 cm_1 (800 nm) maintained their positions but became more pronounced (spectrum B, Figure A0). This irreversible change only occurred after a warm film was chilled to approximately -57°C and rewarmed slightly. Be— cause the phenomenon was repeated in three films over a one week period, it is not regarded as an artifact though the change was sluggish in the final film when decomposition was more likely. -._-. 171 A(pm) IO 20 L3 |.0 0.7 0.6 0.5 0.4 . III-r“ ' l I T I I : '2‘ II °-\\ A ... \ : ... I '-..\' B l \ I \ A I: \\ AMAX ' C \\ \ \ \\ 0.5 \ \ \ \ \ \ \ \ \ \ \ \Ns .\ \‘-" \~___ (3‘) l I I l i 5 l0 I5 20 25 ‘27 (cm")°lO" Figure A0. Spectra of Cs/C322 solvent—free films from am- A — R = 2 dry (fresh) film at -A8°C; y (annealed) at —A8°C after monia: 1 dry B — same film, dr cycling temperature to —580C; 0 - R film at —A5°C. 172 Regardless of the interesting annealing, the spectrum is not as homogeneous as might be expected if the Cs+ com— plex of 0322 were entirely inclusive. The e; and low energy Cs— annealed peak positions are nearly unchanged from those found for the Cs/C222 system (Figure Al, spectrum B). The two lowest energy cs‘ peaks in Cs/C222 spectra may have collapsed into a single peak in Cs/C322 spectra (spectrum A, Figure Al) located midway between their Cs/C222 positions. If so, then that represents the only move toward homogeneity that is evident in the Cs/C322 system in NH3 with R = 2. The effect that the unsymmetrical 3,2,2—cryp— tand may have on the homogeneity of the solid state film structure is undetermined. Damp and wet spectra of these films Show typical plasma character. By contrast, curve C of Figure A0 depicts the slightly asymmetric but smooth spectrum of a dry Cs/C322 film with R = 1. No annealing was noted and no Cs— shoulder or peak is visible. The peak broadness indicates that the e; environment in the solid film is probably not particularly homogeneous, while the lack of Cs' structure indicates that Cs+ complexation in C322 is fairly complete. IV.C. Films from Methylamine Spectra of films from methylamine (MA) often lack shoulders and extra peaks found in the spectra of films from ammonia with comparable stoichiometry (61). Bands in 173 X(pm) I0 210—" L3 |.0 01.7 0i6 0.5 0.4 I I l 0.... . .o' 5 I0 I5 20 25 i? ( cm") ° IO" Figure Al. Spectra of solvent-free Cs/cryptand films with R = 2 from ammonia: A — with C322; B - With C222, from Reference 61. 17A solids from MA are the same ones present in solution while solids from NH3 exhibit bands not seen in the solutions (6A). In an attempt to produce more homogeneous films, then, several systems were prepared with MA as the solvent. IV.C.l. K/C2N22 System Two attempts were made to study K/C2N22 with R = 1 from methylamine solutions. However, in both cases the only peak observed was at 15,500 cm—:L (6A5 nm). A simple flame test on the decomposed second solution confirmed the presence of a large quantity of sodium. A quantitative flame emission study of the "potassium" sample taken im— mediately adjacent to the metal sample used in the second cryptate solution showed a sodium contamination of approxi— mately 20%. A check of potassium sample tubing of various sizes showed that the smaller the tubing, the greater the sodium contamination. The conclusion is that there was significant exchange of the sodium from the borosilicate glass with the alkali metal (97). The more the metal was distilled (into smaller tubing), the greater was the con— tamination. Despite the presence of Na— instead of K— or e; in the spectra, several observations should be noted. The spectra usually contained only one species at a time: e— in the early films and then the contaminant, Na-. t With the exception of one film, no spectra showed any 175 annealing. Both of these items would indicate that solids from MA are less complicated and more homogeneous than the multipeak, annealing films of K/C2N22 from NH3. Films containing MA showed only the same Na' peak as when dry and they continued to show it until they washed from the optical cell walls. This is consistent with the observa— tion that solids from MA exhibit the same bands as the solutions from which they are formed. IV.C.2. Cs/C322/Na System Conservative thermodynamic estimates indicate that many alkalides and electrides should be stable in the crystalline state (5A). For the reaction M + M+CM—( ) C(s) + X(s) + My(s) x y s AGO for Cs+C322'Cs_ is +30 kJ/mole. From a similar calcula— tion for Na+c222-Na‘, AGO = +28 kJ/mole although the latter compound has been crystallized and its structure determined by single—crystal x—ray diffraction studies (6). So CS+C322°CS- might actually be stable. However, sodide salts tend to be the most stable alkalides. For example AGO for Li+c2ll-Li‘ = +2 kJ/mole while AGO for Li+c2ll-Na‘ = —19 kJ/mole. So by comparison the cesium sodide should be more stable than the corresponding ceside, and a methyl— + _ amine solution of the stoichiometry Cs C322-Na was 176 produced. Curve A of Figure A2 shows the optical spectrum of a fresh dry film which includes a fairly broad peak for Na‘ at 13,800 cm‘1 (725 nm) and the hint of a shoulder at approximately 19,000 cm”:L (525 nm). Once the film was warmed above m—33OC, it converted irreversibly to a more homogeneous film whose typical spectrum is depicted by curve B in Figure A2. This is the first indication of a definite peak protruding from the asymmetric, high energy side of the Na’ peak. With only minor shifts but no shape changes, this spectrum remained constant from —71°C to —0.7°C, indicating that it should be stable to even higher temperatures. The most promising aspect of a Cs+0322~Na— MA solution is its bronze color by reflectance. The viscous solution appeared ready to crystallize. After changing the solvent to diethylether, B. Van Eck forced deep red dendritic crystals from the solution in the original optical appara- tus. He is taking steps to produce a single crystal of sufficient quality for x—ray structural determination and if he is successful, the origin of the high energy shoulder/ peak on the Na’ peak may become evident. IV.C.3. Li/C2ll/Na System After the success of the Cs+C322Na- system, the thermo— dynamically favorable L1+C2lloNa' system was attempted. The viscous, bronze-colored solution gave the initial dry l77 A (pm) ,0 2.0 L3 I.O 0.7 0.6 0.5 0.4 . I T T .: ' I *7 A Max 0.5 0.0 I I | . I5 20 25 :7 (cm")-IO" Figure A2. Spectra of solvent—free equimolar Cs/C322/Na films from methylamine: A — fresh film at —30°C; B — annealed film at —0.7°C. 178 film spectrum B in Figure A3 with a Na— peak at 13,200 cm—1 (760 nm) and high energy shoulders at 15,000 cm—1 (665 nm) and 21,000 cm-l (A75 nm). Moderate temperature changes did not make this film anneal, but a subsequent film yielded the spectrum depicted by Curve A in which the Na- peak at 13,900 cm"1 (720 nm) is accompanied by one high energy peak at 18,800 cm_1 (530 nm). This spectrum retained the same character, though it varied in amplitude, from —7A0 to —10°C. Dr. Long Dinh Le was able to precipi— tate crystals from this solution. He is attempting to produce more of these silver-colored rectangular crystals for x—ray structural determination. IV.D. Optical Spectra Summary Thin film optical transmission spectra are dependent upon the nature of the complexing agent, the metal, the ratio, R, of metal to complexer and the solvent used to make the film, as well as the solvent content of the film in the case of ammonia. Table 13 summarizes the peak positions for systems other than Li/C2ll. Four general classes of compounds previously reported (61) are evident + . . in this study: (1) alkalides Mx C-My in which the anion . + — . . is an alkali metal; (2) electrides M C-e in which the "anion" is a localized, trapped electron; (3) expanded metals M+C-e— in which the electron is in the conduction band; and (A) a combination in which localized electrons 179 A (pm) . ,0 2.0 13 IO 0.7 0.6 0.5 0.4 ' I T I ...". I A AMAx J l I IS 20 25 7 (cm")°IO" Figure A3. Spectra of solvent—free eguimpiiiigi/giiiéNa films from methylamine: A - subsequent film which annealed and then remained unchanged from —7A° to ~10°C. empowhmgo mammad cccc IIIIII p03 180 IIIIIIIIIII 00mm IIIIII QEQU Lopomhmco mammad AEVcczm Amvcccama Accmvmhp IIIIIIIIIII Aevoomc Anvoom.oa Asevane A Name ompo mammaa IIII IIIIII p03 I ego: mmz M Lopomhmso memHQ ccmc IIIIII p03 Ammvcmwme Admvcccc AEVcc:.mH Aggmcmhp Admvooam Aomcooma Aecooe.ma Aaeveso : Adnvoom.mm .Ancoom.aa oosc IIIIII coalesce fiancooo.mm .Aemvoom: comm Ascoom.ma Assocaso Admcoom.mm .Asncoomm ooaw Ascoom.ma “pecans m IIIIIIIIIIII Ascooms oom.ma coasoos hopomgmno mammad ccmc IIIIII p03 IIIIIIIIIII cccs IIIIII meow IIIIIIIIIII Ascoomw ooa.ce ass a mmmo gopomhmno mammfid ccmm IIIIII p03 hopompmno mEmeQ cccc IIIIII QEmc IIIIIIIIIII IIII ooa.aa ago m ammo owco mammam IIII IIIIII hos m ocmqm omco wEmmHQ IIII IIIIII p03 I 0:0: mmz m2 mponpo we I2 Coflpflpgoc m pcmmfld pco>flom 2 n Eaam m AHIEov omCOHpHmom xwom .HHmc\Hq swap gozpo mEopmzm meOHpHmOQ xmod oppoodm Hwoflpdo mo mngESm .MH oHQmB 181 .xwoa HHmEm u Qm fl:oUHsoSw n m .xon HOnmE u Eo .cono::m u ::m .:m0:: u :mn .cm.clmm.c HHHwE:o: n0:0 :m:p mmoH sz:wHHm 0:8 H mm cmpmHH moprm .Hmpos :omo co 0H0: 0:0 mopmoHc:H HHH MmeoHQEoo mo oHoE 0:0 on proe mo moHoE ho OHHm: n ma Hancomm.mH III- ome.mH coe.ou . AdmvccmacH IIII ccc.:H coHNI m m A::mv%:c 32 mo Hmvooomee I--- oom.mH Heavens HHH mmmo oz can no mdmvcccwcH IIII cchmH A::mv%:© mmZMmc Anvooo.Hm Hocooo.mH I--- oom mH Asevaao HHH HHmo dz one HH Ancoom.mH Hecoomc ooo.oH Hcsovasc Hmcooo.mH .Hmvoomc Hecoows Amcoomm Asevaao m IIIIIIIIIII OOONB IIIIII P03 m IIIIIIIII II comma IIIIII ago H mme :2 no Hopompm:o memHQ cmmwe IIIIII pozIHEom IIIIIIIIII I ooom Homvoow.HH Ascsvste IIII IIIIIII omsc Hmvomo.mH Atacaso H mmmo mmz om IIIIIIIIIII oomc Hecoom.HH Assovaso m IIIIIIIIIII Hecoocc oos.HH Asevaso H mmZmo m2 g mHmQPO pm 2 l I QCOHpflUCOO mm Uflwwfid PCO>HOW 2 AHIEoV om:OHpHmom xmom EHHm .oossHosoo .mH oHooe 182 and alkalides exist in the same film. This combination most commonly exists in films from ammonia. Perhaps the solvation ability of ammonia is so pronounced that metal cation complexation by the cryptands is significantly re- duced, as postulated for Li/C2ll in Chapter 111. When the solvent is removed by flash evaporation, the cations are trapped outside the cryptands. Spectra of these films are complex and the peaks are broadened reflecting multiple environments for the e; and M“ species. These complex spectra often have time-dependent behavior as an equilibrium is established among the variety of available traps. The nature of films containing solvent if; dependent upon the solvent used. Films damp with methylamine do not change appreciably; the species in solution appear to be the same species present in dry films. Films damp with ammonia, however, show distinct changes. There is no specific evidence that the centrosymmetric alkali anion in the dry film exists in a damp or wet film. The M‘ peak disappears rapidly as the film acquires ammonia and is replaced by a series of solvated electron peaks. As a concentrated MAS is formed most films assume plasma charac— ter, followed by a decrease in plasma absorption as the film becomes more dilute just before washing from the optical cell walls. This sequence is fairly consistent among the systems studied, however film response to the initial ammonia vapor (initial damp spectrum) varies. 183 Some dry films showing high plasma absorption immediately become localized before joining the sequence above. Others which show localized electron character when dry immediately assume plasma character when ammonia vapor is first intro— duced. This variety of responses in initially dampened films is currently without explanation. CHAPTER V AMMONIA ANALYSES The presence of ammonia in liquid alkali metal solutions (7,11) or in solid alkali metal compounds (18) has a sub- stantial impact on solution or solid properties. Alkali metal/cryptand systems have not been as thoroughly studied, yet there can be little doubt that ammonia, depending upon the relative amount present, can drastically alter some of the properties of these systems. This study is an attempt to determine the approximate relative amounts of cryptand and ammonia present. Properties of alkali metal/cryptand systems with variable amounts of ammonia were discussed briefly in Section III.A.2. The indophenol blue formation test for the presence of ammonia (Section II.D.A.) was chosen because of its ease compared to spectrophotometric (quantitative) methods. Yet the test was still semi—quantitative because sample colors were compared with standards of known NH“+ concentration. In most instances after an aliquot of the original solution was removed for testing, the remaining original solution was diluted and retested. The results for each sample in Table 1A are normally the average for two or three tests. 18A 185 .sHse Eom %:0> HEHHH Honpao 00: 0qumm H H cH oHpmpm cIcH x : HHH EmmH H n c: oHpmpm mIcH x : NIB< e we H u m oHuwpm cIcH x m cIB< Ecc .E5500> :wH: w:H:5© :0:p m n H 0Hp0um IcH x c mIB< :030Hw HH90QOHQ 00000:: w:H%Hc c ammo H000 w:HH50 :oHpHmOQEoo0c Ho EHc 0msmo0o :00: 05p0:0gdm :H ESSom> m H H oHpmpm c cH x : :IEH EHH m u H 0Hpmpm cIcH x m mle< Em .w:HQESQ 0Hpmpm m n H oHpmpm IcH x : NIB< :Hm0: op HH0>H0m0H :o c0ode c mzna :0:3 mHU no: 0:03 m0Hdsmm Em p:0omhcm H0:0>0m mmpw0p mmz :00: a u H 0Hpmpm cIcH x c HIB< m A:HEV AHHmc 0Hdemm mp:0EEoc m2 “ HHmc . . moHsmom oHfimem nwwmzv .m oprpm w:HQE:m .m0HQE00 Ho p:0p:oo mmz Ho HHmEESm .HH 0H903 186 Ecm .0HQ80m mmm H u e 0Hp0pm mIcH : H HH .emo H00m :000 :0pm0 Emc ::ou mIcH x m on ©0p0500>0 mconI H H m.H 0Hp0pm N.IcH x m NIXH p0 :Ho>:0m0: :sz comcI p0 meH mcocHHI H0 :Ho>:0m0: 0:0 comcI Emc pm HINH HHUSHm 0H5mm0HQ HOQ0> *H ” cHA oHpmpm NIcH x m HIXH EmzA H n m 0H80:%© IcH x m HHH> .mEHHm c :0:p :0:p0: 3H5: 0:03 00H050m EmmA :po: mm0HQE0m .0050 0Hp0:wme mm: H u c 0H80:mc cIcH x c 0HH> Emm u 0H 0 m IIIIIIII mlcm .HH0o :H HHmc m0HoE IcH x a H cm H p :mso:p mm ©0HSQEoo OHp0: mmmz mo Ecw :OHudHomcw 300:0 0p mHH0o mpdEm *H u ccmA 0H80:mc IIIIIIII chm mosoEEoo mmz “ HHmo HsHec HHHmo oHdsom mpH5m0m 0H80:%Q m0HoSV Ho 0NHm oHpmpm m:HQE:m .possHosoo .HH oHome .:0Hpo0p0c Ho HHEHH 0:0 :0:p mm0H m0: mmz mo mpHp:05® * H “ m.mH oHpmpm :zo:x:5 .:0Hpme:ow EHHM mo 0mo:p .QOH: I0.mmmc+m on HwHHEHm m:oHpH0:oo :0c:5 mm 00:0Q0HQ 0:003oa mac .H0m EOHH H n ccmA oHpmpm :Bo:x:5 1 mad as anHHoso 0sHeoHaseoz .oosd noz.mmmo+mz scm .HHo>:0m0: EH00UH0 0HQEwm H ” cHH oHem:%© mIcH x m QHH> nesoEEoo mmz " HHmo AsHec HHHmo oHQEom mpHSm0m 0H80:%Q m0Hon :o 0NHm oHpmpm w:HQESm .oossHesoo .HH oHcoe 188 Precision with these dilution steps was generally i50%. Ammonia tests 1 — 6 were specifically to determine the effect of static pumping time. A large apparatus was built with seven 8 mm 0.0. tubes on the sample sidearm. Li+c2ll.e’ solution was poured into the tubes and evaporated to near dryness. The bulk solution in a reservoir was frozen with liquid nitrogen and the NH3 was supposedly drawn from the samples to this trap while the apparatus was under static vacuum. Two problems were evident: first, samples adjacent to AT 1 — 3 were wet with NH3 when R-N2 was placed on the reservoir, thus invalidating the results; and secondly, there was cryptate residue in the vicinity of the AT-2 sample seal-off. When this region was heated the cryptate decomposed and raised the pressure certainly higher than 10—2 torr, though the actual pressure was not measured. Thus the removal of NH3 from the remaining samples was probably much slower than in a higher vacuum. However, samples AT A-—6 still show a progression of less NH3 as static pumping time increases. Sample shape also has an apparent effect on NH3 reten— tion. As expected, a thin optical film was more ammonia— free than either of two magnetic susceptibility samples in which the Li+C2ll-e_ powder was in a lump, even though the - -5 susceptibility samples were dynamlcally pumped to 010 torr. Predictably dynamic vacuum pumping decreases the NH3 content more effectively than static pumping. The empty 189 cell (EC) tests show this clearly, as do samples II and VIIb. It is interesting to note the apparently similar dry— ness of the Li+C2ll°e- (III) and the K+C222°e_ samples. The methylamine content of the potassium electride sample was determined by proton magnetic resonance (6A) and was much more quantitative than the NH3 analysis. Both samples were pumped statically, and although the two solvents have different vapor pressures at a given temperature, the result may indicate that alkali metal electrides have roughly similar solvent affinities. The results in Table 1A show that sample solvent content decreases with increased drying time and that dynamic pump— ing is more effective at solvent removal than is static pumping. Regardless of conditions, thin samples lose their solvent more readily. Most importantly the tests indicate that Li+C2ll'e— samples are substantially free of solvent. It is apparent that the MNM transition observed in the Li+C2ll-e' (II) system was not solely due to the presence of some NH3, although the precise effect has not been determined. CHAPTER VI SUMMARY AND SUGGESTIONS FOR FUTURE STUDIES VI.A. Summary The system Li+C2ll°e_ lies very nearly at the metal— nonmetal transition. This transition can be accomplished by changing the mole fraction of lithium: clearly the preparations with lithium to cryptand mole ratios of 0.60 and 2 lie on opposite sides of the transition. It appears that the system gradually changes to metallic character between R W 1.5 and 2. Accomplishing the MNM transition with the change of mole fraction of lithium indicates that the Li+C2ll'e‘ MNM transition may be described by the Mott— Hubbard model. As the electron density increases, the increased screening allows the electrons to overcome the Coulombic potential and become itinerant. On the other hand sample II with R = 0.95 completed the transition rapidly as its temperature was raised or lowered through —A5°C. Also, sample I with R = 0.93 showed only metallic character while three others with the same approximate mole ratio were predominantly nonmetallic. It is possible that disorder in these microcrystalline samples plays a role in 190 191 their approach to the MNM transition. If so, then perhaps the Anderson model provides a better description for the MNM transition in Li+c2ll-e‘ or it may be that the system is best represented by a combination of Mott—Hubbard and Anderson models. A great deal of additional study must be accomplished on Li+C2lloe_ before a conclusion may be drawn. In samples which are nonmetallic a significant percentage of the potentially available spins are unpaired at 2A5K. As the temperature is lowered, pairing occurs with very weak interaction energies until almost all spins are paired at liquid helium temperatures. Many systems other than Li+C211oe' were studied by optical transmission spectroscopy of solvent—free thin films. In some cases preparations with mole ratios slightly less than one show both electride and alkalide absorption peaks. This is particularly true when the films are made from ammonia solutions and it probably indicates that cation complexation by the cryptands is significantly reduced in ammonia. The mixed alkalides, Li+c2ll.Na’ and Cs+C322oNa-, showed very strong Na— absorptions and the films were stable to nearly 0°C. VI.B. Suggestions for Future Studies The study of the Li+C2lloe" system should be continued in order to further characterize it and to determine, if possible, the factor(s) other than the mole ratio of 192 lithium influencing the MNM transition. Sample VI with R = 1.15 showed EPR results which did not seem to fit the pattern of the results from samples with other mole ratios. The slightly-metal-rich region should be investigated more thoroughly. A study could be accomplished by NMR to deter— mine the rate of complexation of the lithium cation by C211. In either ammonia or methylamine the complexation at ~A5° or —50°C should be slow enough to observe a progressively smaller signal for the free cation and an increasingly larger signal for the complexed cation as equilibrium is approached. The result would give valuable kinetics and thermodynamic data. Finally, the study of Li+0211-e_ would be invaluably enhanced by the production of single crystals. This is not a trivial task because of the high chemical potential of the electrons. It may be necessary to include in the crystal structure some species upon which the electrons can localize so they will not destroy the cryptand. Perhaps then this difficult problem will become more tractable. Optical spectra of systems other than Li+C2ll.e“ point towards some interesting possibilities. Most of the elec— trides appear to contain localized electrons. However K+C222-e“ seems to be metallic, depending upon annealing conditions. This may be another system which is near the MNM transition and it should be explored more thoroughly, possibly with samples from methylamine. The mixed alkalides, 193 Li+C2ll-Na- and Cs+C322-Na_, are quite stable in their bronze-colored methylamine solutions. Both systems should be good candidates for producing single crystals. APPENDIX APPENDIX EFFECT OF OPTICAL FILM NON—UNIFORMITY A study was conducted to determine the effect on spectral band shape of films of variable thickness and of films partially covering the beam cross section as it passes through the optical cell. Professor J. L. Dye derived the spectral effect of non—uniform film thickness and Mr. Jim Anderson provided much of the computational effort. Suppose there is a circular light beam of radius rO with total intensity 10(1) and local intensity iO(A): 2wrdr @ rO Then 10(1) = iO(A)-S where S is the beam cross section and r0 . 2 10(1) = O io(l)2wrdr = lOTTI’O (A-l) Let I(A) be the total light intensity passing through the sample, such that 1”O -> I(1) = i(r,A)2nrdr (A-2) 0 19A 195 Bouguer—Lambert or Beer's Law states that, for a sample of uniform thickness x and a light beam of uniform cross sec— tion, where a : absorptivity (cm—1) is the molar extinction coefficient and c is the molar concentration. Then 100) 2.303 log W = A = OIX .+ However, for a non—uniform thickness, x(r), i(A) = 10(A)e-0(A)X(r) and I” O + 1(l) = 211i (12f re—a(>‘)x(r)dF O O = 2.303-e(k)°c where €(A) (A-3) (A-A) For simplicity, assume initially that the non—uniform film fills the light beam cross section and x(r) — a + b(r/ro) 196 Using this model, Equation A—A becomes I“ 0 1(x) = 2fiio(A)JA re‘a(A)[a+(P/Po)b]d? 0 After performing the integration with a table of integrals and using Equation A—l for 10(1): 1(1) _ —0(A)°a l — e-a(x)b(1 + d(i)b) 1 (l) ‘ 29 E 2. 1 (A—B) o (G(A)b) If the film were of uniform thickness <2> then fl%%% = 1 1 0 e'“<2> and d = 2.303logj? . Call logj? the nominal absorbance A . Therefore nom 2.303Anom = a<£> (A—6) But the film is not of uniform thickness and the "area weighted" or average film thickness <2> is calculated by frOdrx(r)2flrdr <£> = —9 r = {Dodr2firdr r 2 1”Q _ —§ I) I°x(r)dr (A 7) O For the model chosen, x(r) — a + b(r/ro), so 197 2 r <2.) = *2- JC.) 0 r [a + b(_r_)]dr+ r r0 0 r r <2> = 3% f O rdr + 32 f O I.2dr l” r3 O o o r2 r3 2a 0 2b 0 2 r2(2) +3r3(3 ) a + 3b (A 8) O O The relationship of a to b can assume any value. If b = — g is arbitrarily chosen, then Equation A—6 becomes 2 2 2.303Anom = d(a + gb) = aa(§) Then da = %(2.303)Anom (A_9) and ab = —%(2.303>An0m (A—lO) These values can now be used in Equation A-5 to compute %; . But first make this specific case more general by 0 allowing the film to cover a variable portion of the optical beam cross section. If 2 T = (area of beam covered)/(total beam area) = (EL) 0 198 then I(A) = TIM)Spot + (l—T)IO(A) (A-ll) Thus Equation A—5 becomes -0I(>\)b(1 + (db>2 1(A) z T28—aoa[1 — e 9b>3 + (1-0) (A—l2) Now, apply the following conditions: 2 (5;) = T = 1.00 I°o Anom = 1.50 as well as the previously selected b = —a/2. This calcula- tion is for a film which covers the whole beam cross section. If the film were of uniform thickness, the nominal ab— sorbance would be 1.5. However the film is conical: b o A nom .500 .750 1.000 1.250 1.500 1.750 2.000 .50 .119 .122 .123 .12A .12A .12A .125 .75 .272 .312 .331 .3A1 .3A6 .3A9 .352 i; .80 .308 .365 .395 .ull .u21 .427 .u3l o .90 .383 .A93 .565 .610 .6A0 .660 .673 .A62 .662 .8A2 1.001 1.1A3 1.269 1.381 .00 201 T bl A-A. F'l . a e l m shape effect. Amax for b —a .r. b<0 A -° .1. nom o I, r0 .500 .750 1.000 1.250 1.500 1.750 2.000 .50 .106 .112 .115 .117 .118 .119 .120 r .75 .228 .263 .283 .296 .305 .312 .317 P; .80 .257 .302 .330 .3A9 .362 .372 .380 .90 .317 .39A .AA7 .A85 .513 .536 .55A 1.00 .380 .506 .606 .689 .759 .819 .872 Table A—5. Film shape effect: Amax for b = 0; uniform thickness. T a .L Anom .500 .750 1.000 1.250 1.500 1.750 2.000 .50 .123 .125 .125 .125 .125 .125 .125 I, .75 .292 3311 .350 .356 .358 .359 .359 ro .80 .332 .395 .u23 .u35 .uuo .uu2 .AA3 .90 .AlA .59A .625 .671 .696 .709 .715 1.00 .500 .750 1.000 1.250 1.500 1.750 2.000 202 LOO _. I, .80 - _ AMAX _ r0 = Opgcal Beam - .60 — adIus _ ANOM _ 0,9 .40 " 939* — _. 6_\.O0 Cfi) ‘ ._ I . ‘$j"\ :IQL’ I .2“) F» lxflflwfi -+ ‘ 0 L I I I I 60 .80 I00 r/ro Figure A-l. Effect on the peak amplitude of an optical film which is of non-uniform thickness (shape as indicated) and which fills various amounts of the optical beam. 203 .E00Q HmoHpoo 0:p mo mp::oE0 mSOHH0> mHHHc :oH:3 0:0 Ac0pmoHc:H 00 0:0:mv 000:x0H:p EHOMH::I:0: mo 0H :OH:3 EHHm HmoHon :0 Ho 0059HHQ80 x00: 0:p :o po0mmm .mI< 0:5me o..\.. 3.8: III scum .850 ... 6.. x424. 20A It is informative to observe the effect of a film of non—uniform thickness on a complete lineshape, not just upon the maximum peak amplitude as in the previous tables and figures. An absorption curve of Lorentzian shape can be described by: T2/n 1 + (w—wo)2T s Figure A-A. Effect on a Lorentzian absorption peak with a nominal 2.00 absorbance when the optical film is of non-uniform thickness (shape as in— dicated) and when it fills various amounts of the optical beam.’ 207 However, most films in this study covered much of the optical cell walls, and with the beam passing through two walls, the probability of only 81% or 6A% coverage as depicted in Figures A—3 and A-A seems remote. It is very difficult to judge the uniformity of film thicknesses, so no one geometry in this study should be considered more likely than any other. It should be noted, however, that as films become very thin (the center of the disk for Table A—3 or the edges for Table A—A), the deviations from nominal absorbance become quite large. BIBLIOGRAPHY 10. 11. 12. 13. 1A. BIBLIOGRAPHY W. Weyl, Ann. Phys., 197, 601 (1863) or Poggendorffs Annln., 121, 601 (186A). ‘ J. L. Dye in "Progress in Macrocyclic Chemistry," Vol. 1, J. J. Christensen and R. M. Izatt (eds.), Wiley-Interscience (New York) 1979, Chap. 2. J. L. Dye, M. G. DeBacker and V. A. Nicely, J. Am. Chem. Soc., 22, 5226 (1970). J. L. Dye, M. T. Lok, F. J. Tehan, R. B. Coolen, N. Papadakis, J. M. Ceraso, and M. G. DeBacker, Ber. Bunsenges_ Phys. Chem., 15, 659 (1971). J. L. Dye in "Electrons in Fluids, Colloque Weyl III," J. Jortner and N. R. Kestner (eds.), Springer—Verlag (Berlin) 1973, p. 77- J. L. Dye, J. M. Ceraso, M. T. Lok, B. L. Barnett and F. J. Tehan, J. Am. Chem. Soc., 26, 608 (197A). T. A. Beckman and K. S. Pitzer, J:_Phys. Chem., 65, 1527 (1961). J. A. Vanderhoff, E. W. LeMaster, W. H. McKnight, J. C. Thompson and P. R. Antoniewicz, Phys. Rev. A, I. 427 (1971). R. L. Schroeder, J. C. Thompson and P. L. Oertel, Phys. Rev., 178, 298 (1969). K. D. Vos and J. L. Dye, J. Chem. Phys., _8, 2033 (1963). R. Catterall, J. Chem. Phys., A3, 2262 (1965). J. C. Thompson, "Electrons in Liquid Ammonia," Claren- don Press (Oxford) 1976. C. A. Hutchison, Jr. and R. C. Pastor, J. Chem. Phys., 2;; 1959 (1953)- E. Huster, Ann. Phys., 3, A77 (1938). 208 15. 16. 17. 18. 19. 20. 21. 22. 23. 2A. 25. 26. 27. 28. 29. 30. 31. 32. 33. 3A. 35. 209 S. Freed and N. Sugarman, J. Chem. Phys., 11, 35A (19A3). "— J. P. Lelieur and P. Rigny, J. Chem. Phys., 59, 11A2 (1973). __ P. R. Marshall, 6; Chem. Eng. Data, 7, 399 (1962). W. S. Glaunsinger and M. J. Sienko, J. Chem. Phys., 62, 1883 (1975). II— F. J. Dyson, Phys. Rev., 96, 3A9 (1955). G. Feher and A. F. Kip, Phys. Rev., 96, 337 (1955). R. H. Webb, Phys. Rev., £68, 225 (1967). P. P. Edwards, A. R. Lusis and M. J. Sienko, J. Chem. Phys., 12, 3103 (1980). - J. R. Buntaine, M. J. Sienko and P. P. Edwards, J. Phys. Chem., 6A, 1230 (1980). — P. P. Edwards, J. R. Buntaine and M. J. Sienko, Phys. M3 L9.) 5835 (1979). ‘——"'— R. B. Von Dreele, W. S. Glaunsinger, A. L. Bowman and J. L. Yarnell, J. Phys. Chem., 72, 2992 (1975). W. S. Glaunsinger, £6 Phys. Chem., 6A, 1163 (1980). In discussion following R. J. Peck and W. S. Glaunsinger, J. Phys. Chem., 65, 1176 (1980). Reference 12, page 265. L. Kevan, J. Phys. Chem., 6A, 1232 (1980). G. Dolivo and L. Kevan, J. Chem. Phys., 76, 2599 (1979). L. Kevan, J. Chem. Phys., 66, 838 (1972). L. Kevan, iLPhys. Chem., 79, 28A6 (1975). —-——— J. Paraszczak and J. E. Willard, J. Chem. Phys., 76, 5823 (1979). J. E. Willard, J. Phys. Chem., 79, 2966 (1975). c—— E. E. Budzinski, W. R. Potter, G. Potienko and H. C. Box, J. Chem. Phys., 16, 50A0 (1979). 36. 37- 38. 39. A0. A1. A2. A3. AA. A5. A6. A7. A8. A9. 50. 51. 52. 210 H. C. Box and H. G. Freund, Appl. Spect., _A, 293 (1980). D. F. Feng and L. Kevan, Chem. Rev., 66, l (1980). T. Shida, S. Iwata and T. Watanabe, J. Phys. Chem., 19, 3683 (1972). T. Shida, S. Iwata and T. Watanabe, J. Phys. Chem., 19, 3691 (1972). A. Banerjee and J. Simons, J. Chem. Phys., 66, A15 (1978)- S. A. Rice, J. Phys. Chem., 63, 1280 (1980). C. Kittel, "Introduction to Solid State Physics," 5th ed., John Wiley & Sons, Inc. (New York) 1976. J. J. Markham, "F—Centers in Alkali Halides," Academic Press (New York) 1966. 0. K. Ong and J. M. Vail, Phys. Rev. B, I2: 3898 (1977). R. S. Alger, "Electron Paramagnetic Resonance: Tech— niques and Applications," John Wiley & Sons, Inc. (New York) 1968. W. Schmitt and U. Schindewolf, Ber. Bunsenges. Phyg. Chem., 6;, 584 (1977). See, for example, N. F. Mott, Proc. Phys. Soc., A62, A16 (19A9); Phil. Mag., 6, 287 (19617; Phil. Magh, 19, 835 (196973 and—"Metél-Insulator Transitions," Taylor and Francis (London) 197A. P. P. Edwards and M. J. Sienko, Phys. Rev. B, 11, 2575 (1978). J. Hubbard, Proc. R. Soc., A276, 238 (1963); A277, 237 (196A); and A281, A01 (196A). P. W. Anderson, Phys. Rev., 109, 1A92 (1958) and Comments 801. St. Phys., 1, 190 (1970). B. Dietrich, J. M. Lehn and J. P. Sauvage, Tetrahedron Lett., 33, 2885 (1969). J. M. Lehn and J. P. Sauvage, J. Am. Chem. Soc., _7, 6700 (1975)- 53. 5A. 55- 56. 57. 58. 59. 60. 61. 62. 63. 6A. 65. 66. 67. 68. 69. 70. 211 J. L. Dye, J. Phys. Chem., 66, 108A (1980). J. L. Dye, Angew, Chem., Int. Ed. Engl., 16, 587 (1979). J. L. Dye, C. W. Andrews and S. E. Mathews, J. Phys. Chem., 12, 3065 (1975). ““““ L. Dye, J. Chem. Educ., 66, 332 (1977). L. Dye, Pure Appl. Chem., A9, 3 (1977). J J J. L. Dye, M. R. Yemen, M. G. DaGue and J. M. Lehn, J. Chem. Phys., 66, 1665 (1978). F. B. Tehan, B. L. Barnett and J. L. Dye, J. Am. Chem. oco, 99, 7203 (197A). M. G. DaGue, J. S. Landers, H. L. Lewis and J. L. Dye, Chem. Phys. Lett., 66, 169 (1979). J. L. Dye, M. G. DaGue, M. R. Yemen, J. S. Landers and H. L. Lewis, £;_Phys. Chem., 66, 1096 (1980). D. Moras and R. Weiss, Acta Cryst., B29, A00 (1973). J. M. Lehn and F. Montavon, Tetrahedron Lett., 66, A557 (1972). M. G. DaGue, Ph.D. Dissertation, Michigan State Uni— versity, 1979. I. Hurley, T. R. Tuttle and S. Golden, J. Chem. Phys., A , 2818 (1968). Y. M. Cahen, J. L. Dye and A. I. Popov, J. Phys. Chem., 12. 1292 (1975). J. E. Young, Jr., Ph.D. Dissertation, Cornell Uni— versity, 1970. B. S. Deaver, Jr., C. M. Falco, J. H. Harris and S. A. Wolf (eds.), "Future Trends in Superconductive Elec- tronics (Charlottesville, 1978)," AIP Conference Proceedings No. AA, American Institute of Physics (New York) 1978, p- 59- J. E. Wertz and J. R. Bolton, "Electron Spin Resonance: Elementary Theory and Practical Applications," McGraw— Hill (New York) 1972. Comments by R. Catterall following Reference 61. 71. 72. 73. 7A. 75- 76. 77. 78. 79. 80. 81. 82. 83. 8A. 85. 212 M. T. Lok, Ph.D. Dissertation, Michigan State Uni- versity, 1973. M. R. Yemen, private communication. A similar device is depicted in Reference 71, page 179. This apparatus was constructed by Van Eck and Lewis as noted in Reference 7A. The hydrogen evolution device was assembled by Mr. Bradley Van Eck and Dr. H. L. Lewis following the method of Dewald and Dye: R. R. Dewald and J. L. Dye, J. Phys. Chem., 66, 128 (196A). F. Feigl and V. Anger (eds.%,"Spot Tests in Inorganic Analysis," 6th English Ed., Elsevier Publishing (New York) 1972. H. Aulich, L. Nemec and P. Delahay, J. Chem. Phys., 91. A235 (197A). D. F. Burow and J. J. Lagowski, J. Phys. Chem., 72, 169 (1968). '“‘ R. E. Honig and H. 0. Hook, RCA Review, 2;, 360 (1960): the data for ammonia was based primarily upon R. Over- street and W. F. Giauque, J. Am. Chem. Soc., 59, 25A 1937 . ‘— R. E. Rondeau, J. Chem. Eng. Data, 6;, 12A (1966). H—' K. Bar—Eli and G. Gabor, 6; Phys. Chem., 77, 323 (1973). E. Mei, A. I. Popov and J. L. Dye, J. Am. Chem. Soc., 29, 6532 (1977). _— B. Bockrath, J. F. Gavias and L. M. Dorfman, J. Phys. Chem., 19, 306A (1975). C. P. Poole, Jr., "Electron Spin Resonance — A Com— prehensive Treatise on Experimental Techniques," Interscience Publishers, Inc. (New York) 1967. N. S. VanderVen, PEys. Rev., 166, 787 (1968). T. Chang and A. H. Kahn, National Bureau of Standards Special Publication 260—59: "Standard Reference Materials: Electron Paramagnetic Resonance Intensity Standard: SRM—260l; Description and Use," U.S. Depart- ment of Commerce (Washington, D.C.) 1978. 86. 87. 88. 89. 91. 92. 93- 9A. 95- 96. 97. 213 B. G. Segal, M. Kaplan and G. K. Fraenkel, J. Chem. Phys., ii. A191 (1965). "‘“"‘ A. H. Morrish, "The Physical Principles of Magnetism," John Wiley & Sons, Inc. (New York) 1965. J. L. Dye and V. A. Nicely, J. Chem. Educ., 66, AA3 (1971). W. Wojciechowski, Igorg. Chim. Acta., 1, 319 (1967). T. Oguchi, 630g. Theor. Phys., 13, 1A8 (1955). H. Ohya—Nishiguchi, Bull. Chem. Soc. Jpp., 66, 3A80 (1979). J. C. Thompson, private communication based on re— flectance spectra in Reference 8. J. M. Ceraso, P. B. Smith, J. S. Landers and J. L. Dye, QLPhys. Chem., 61, 760 (1977). H. L. Lewis, private communication. E. Mei, L. Liu, J. L. Dye and A. I. Popov, J. Solution 9290:. 9. 771 (1977). -—t- E. Kaufmann, J. L. Dye, J. M. Lehn and A. I. Popov, J. Am. Chem. Soc., 102, 227A (1980). M. G. DeBacker and J. L. Dye, J. Phys. Chem., 16, 3092 (1971). I TITIIITIWIMIWWWMAINTAIN“ . 1293 03085 5872 3