I _ - _ .- ! - . ~51 vrxfio . 11.44.10...) 2... ago. .. .. .411 i 1. . .b? 6" .0. l t . to... £73.... .12.... t. 5». 11:2.iif;cwn|¥h.vYflmmUI.§ ”was; ’- }A .1 114... D: . 1...... . .1. rrfxa+tirtlliwfbir fi .39 -111: «erbium. ro‘r . '4 L. 1t.ff.a;tflatvv{firrcttévto...vr«'[~vh .vc.‘ . v . w? vr v S075... .o. ‘91.. v... I .t‘uczif V {£75 {gig :é’é -. £55.? i i": If: FE; 3:: f“ .f a. :1 E? J. I r? w ‘11 fl .7 a . .1» 5.1991. .5... 3. Q v I. t sakflflinv-Wfiuiétrfiur \ o A} .vsoc.‘rlt:ttn..to.vt: . s... . (19.1 . It...) :23... 1912-..}.roflflihflmflovrb ’Lflv‘oo,bf?. :- . .eva ‘13.? .4390} .h #1:; [’11.- §3 .45 —.v 3.1.10: 01?. . .31 I. . bu . . 3.9.. u .. vflorc.!v.u.th.pvbvrro.h.bnav1r¢ol..rvr¢ cu . 1 2...). $0" .1 inhihv... v! '7 f.e¢.qr.1vo.m'rulr v1.5 . Evf. I I . Afr.» . fr . I Vflftkc-Irrfifvifyfivdwm.’ ! :91 '3'; VI. . I. 1 . Jaw: .: .- . m . , .167bafbe1hbuunt «pk-wivsvoflfia (191.4%! u Qingév’d: H . . o. 6...... IX: . Vt . ONSAT. . . 1' 1.115.951; “$0.“ ‘I 9'72. # rift,— nlinggxu! r. 6.. f '9’... if! rtvvefioltfvaa¢$vluv+u , It}. .ovaiJ... ‘10? a" Y‘VtOu b... o 9. 1- r- 1 I . it“; £n.‘.'.r' . . v0 . . I L1 17 r c if .v . otltvof .. v .6. r so o>asvcravvhnflqrhrroetvnrz . f: reef ta? 1 t .o. . v.16... OCR-Marivrlri (.5..th - a» CI .0 f. favvbélbtr g . n 1 t. ._ . .4 gifiauubehv..¢rul.1§?vu t95%.... .Pv I . I I. .f. . e .. .tiicgdav NWfNWViaim «:1 f , Lfétii rcw¢.a§r 5.1.53; .1 . firi‘?r¢s€§i v . 94.9.31 {14.2.wa or. $ Inlgflflvil v to». 61".9afigloVVQ-frncflkfiNiib. :1 .74.. . . Nil-B.) o. at. T! ‘0‘... I... 7...! Qt naoxrfavrv‘or1 Orvvrr c irittfhw . . 15:03.95. .i Ozm {chi . rt Nun“. E r00. . I t. J! ‘r : !Vr r «Jr: wt arid???.bflrflgavfiirofz..vgirflmutfltn‘wpauf.f alto“: ryti‘rffva “'15wa “AU"! hgfirv Vt! (phat! r z t . ff ‘9'??? I w . f... r 0 INT?! .r. 3.9?er ‘16.: T. 1'? vVnrftflw it.) 3% .JV. (ii-.0. . it... . .1 .L . Efevf..vo%.wvrr’vtoh...flul Hr! éohfllrrg 91 1,719.5: .. . ovinmv.”rv‘..-.vtorv¢.“if... 70:325.... [42¢ . v 1.? t). [flute 1‘6. .25. 6.}. nu”: o . rW-Rflvaff r“ .iffrtvfllra..vvo¢§ll.rvfi J1;rrfrfpr'nvtviofltv駧;Onnrifiiflfiflzgfi , 1H1)! . , 4.6.900... crib: « x V r. (.316... 3‘ .rtft: F 3‘1; ivy: $7.... IF; $29k...) . , t...» it: 131.. hi... finmflr 193' .33.. .. . A .Nlr'lrffrv £§Ir.x§. I firituo.‘ £0.4'7qivf1s‘v601irt‘i £315: . g v 8‘ 35" V . 9. O. U .Q .Q t 91!}! v . . r... firtvf u .rtterfWIXI 21.....3-1. ”inlittovfcwrr .1! 7A}... v . 0700 . ‘ c. griffin“; .41 . .2. r a fi. .991. form: $.19: vscnot.h”‘fc.§.!\{Jr‘rvfffiahhflhlaflg nit...“ .yrnlfrotl.£;§e:§fto «A 9003.”; w .. . W o c.}.l...fl.ltn.‘éfnttvi¥. 1......oo'vrrer...ithirrnfixérvt.fir}uvbbwiuflnr.rtv o. IEH¢‘.QH% . .6 up, get”; .1111! L M gitiqgwinter...influfiurnhmulfvfiornulfrleitfdfinfiolugmfi.§..Hn....t.i v2.82. J. lifts...‘ Us lug? .. . c 1 I!!! . '1'. . otf.’1}0....£cl¢ flu}. [fifth-KI. '32:... of. re; .1 a. 7 ... fiftirii. . if. L:t~.9!fo§.u¢ vfi 0 . . 4““. vi... {of {ltgllicfzrflotiiltvrrvfirrvghte 5.0.9.. .2.‘ . .L, .gtlttxlfvoourervt “fig . Yr. tfl:riref.(rll.l..vv$>;rlf‘ervc1.In$¢V. gift?! , gpegtrfirlfifflnfldofitfhflw}; I b :3 0'99... t r! . 1-0.!!! Var-6 ng‘ffo‘m. fvul if! QfIIOeltftoND ,fo-i: . . riff: hfih.flup..wp...7rrw pat vflltélxv: It vrvvvhnvovlvfiw. .89... tflgv exhatve.rv§v.¢vv : “Irvifflrllflfh . glint! if.) u .. P.... ellirfvraxilur v£v¥i1.£vto , 0...! 7 ‘0‘ . Lr’ flintl C C . 5.0;”?! ’I 5)”! ¢ o. 0.: v 3. \ r . .1 .izkmmm » \Irvvo'!‘ .3 . . . ii .13 r v . .01. \u. . liter». fi 3. ‘15.; . . L. . g . , .s a... iglziflflwz . Rut: . . 5...... . r»... . 21.38... s. - g?! f.- ' . . . r. K - "l . if Fill? val.’ . . . r hilt. . .. 1...}... a? a! $1.... t.£lri..!t..lt¢!ol.£rfgptl¥£x .. . .15. , . is {I}!!! .5. . . 1’ . . 9. . . “grifl 2...; . . . Hug . a.» . u v. . 2.3 . X . I “I fig! flfigtu It‘lIII-l. u I. .—‘~n-. zit - . .. . . git-3].. fifrrtrrrflrbuinfif. lbw. turf. .. .rtrrif {I’ll-gt»; A a. o If; sift-Alb. K 01...... Eli‘s. __ w . ll. ’29:... I}... . . . . & fi‘cf 51;. P. if. fitter- . lg ) g r . 5..ngng , V we...“ fi.flfif1.§2§foflffilifu trfbrfwttgg . If!“ fififulbf... {1"}!“(39 If. . ilfiibfrif. 5‘ . . . . In}... $0.? i . . . . , . r .. .3. .7. it)... 3E1... _ . ftk‘. \If figktittdg iai. L, Ehmiétgk .000th . ., ....... ................tr... 1.. fizfifi. it? . . v1.1.3... in”??? E; . . H .I‘Otslf, r)? . .u . {robin .. .lr fiftluortfl 1.} 4:50}... , . if!!! It}?! All... “(Egg-f . . . gi£§;.fiffi.. .. . . . “If? . lg. . . . 33...... -15)-;511314 i . IE; Is.;.....F.+:...+ .3 , sizififiwfz. . W?...$:flnri!..hfrruféuwtclrfihfluflixggffl. J . ..I.£J........l.iur.¥t.3t.f “Hr“..vqnqnphuwowwhdw? .7-.- . . , 5-..; : , V 41.1.13. . . .. .‘fl .vVL ....s......_ ) . arcs) .' i}1i&—"5‘S w 'r 1.. "3’4".“ .1. -.,_:, yarns-r: ;; 1 “it. :’ . 7-,- ’; . ~-.- 1. so *1 F :« . A ., .-/ This is to certify that the dissertation entitled STRUCTURAL AND MAGNETIC PROPERTIES OF THE ELECTRIDES OF CESIUM COMPLEXED BY lS—CROWN—S AND 18-CROWN-6 presented by Steven B. Dawes has been accepted towards fulfillment of the requirements for Ph . D . degree in Chemistry Major professor fl Date SM 9, WSL MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 MSU LIBRARIES RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. “70005120053 ‘ ._ ' ‘ . d it." 15?“: ,1" u . ; ‘2‘ r . , .. “‘6. . 2. , . t . ”9 A“, 1 ‘|.‘ ‘ .. I .- "are M j “New; 2‘ 1 "l‘ \. * '11.“- l' a I u l '1 A -‘ ‘ .L "A/ b L/ I,“ f/ [,4/ E/ .SWNWICWIBSOFTHBBWBS OFCBSIIHWBYlfi-CWMDlfl-Cm-G By Stem 3. Due. A DISSERTATION Submitted to Michigan State University in partial fulfill-ant of the militant- for the degree of mwmnosorm Depart-eat of Chemistry 1% ABSTRACT The nature of the trapped electron in two compounds, the newly synthesized Cs+(1505)2-e- and Cs+(18C6)2-e_, were explored by structural, magnetic, optical, and thermal techniques. The first crystal structure of.an electride, Cs+(1866)2-e_, was determined. The unit cell is nono- clinic, CZ/c, with a = 13.075 A, g = 15.840 A, g = 17.359 A, B = 92.30', and g = 4. The electride is iso— structural to Cs+(1806)2-Na- except that the anionic sites contain less than 0.07 e_ A'3. The lonoclinic (CZ/c) cell of c3*(18cs)2-Na" has g = 13.5811A, 1; = 15.684 A, _<_:_ = 17.429 A, B = 93.16', and g = 4. The structure of Cs+(l505)2-I- was solved. The tetragonal (I4) unit cell has a = 13.173 A, g = 16.645 A, and g = 4. Both electrides are paramagnetic and obey the Curie-Weiss law. Cs+(1505)2-e_, when slowly cooled from a phase transition at 238 K, displays antiferronagnetisn (TN = 4.2 K). The interacting electrons appear to couple via direct exchange in spite of a long interelectronic distance (~8.7 A). The spin lattices are weakly affected by crystalline aniso- tropy; at 1.6 K a spin—flop occurs at 1.5 RC. The cation in both electrides displays a temperature dependent Knight 133 shift in the . Cs MAS-NMR spectrum, that is proportional to the paramagnetic electron density at the nucleus, -- 5-—3 <{(¢(0))2}> = 6.2 x 1024 e cm 3 and 2.1 x 102 e CI for Dawes, Steven B. Cs+(1806)2-e— and Cs+(1505)2-e_, respectively. The para— magnetic Ramsey shift of "inclusive" complexed Cs+ in dis- nagnetic salts from the value of 08:8) is larger for complexes with shorter Cs-O distances, and is largely independent of the counter—ion. The static powder and single crystal NMR spectra of Cs+(1806)2-I‘ and Cs+(15C5)2-I— are broadened by first order quadrupole, and chemical shift anisotropy effects, equ/h = 89.6 KHz and 42? K32, and a = 32 ppm and 52 ppm respectively. Thermal notion of the conplexants in soae salts is rapid enough to partially average the 18 contribution to the dipolar line— width in cations and anions.- The chesical shift of Cs_ in Cs+(18C6)2-Cs- is downfield from the calculated value of 082g) by 140 ppm, and is sensitive to changes in the local environment. A compound of mixed anionic stoichionetry, Cs+(18C6)2-Na 5, gave a five line NMR spectrum. o.125°o.87 The data fit a model in which each successive dianagnetic peak corresponds to a sequential substitution of Na_ for electrons in the nearest anion shell about a conplexed cation. TO MY PARENTS ii ACKNOWLEDGEMENTS I am grateful to many people, who over the course of this thesis work have contributed their time and friendship to my efforts. In particular, I am indebted to Professor James L. Dye, who taught, guided and supported me throughout this research. I appreciate the efforts and advice of my guidance committee, Dr. R.D. Schwendeman, Dr. H.A. Rick, and especially Dr. J.A. Cowen. I am grateful to Dr. Donald L. Ward whose efforts were vital to the crystal structure determinations in this work. I thank the technical staff for their excellent work, electronics designer Marty Rabb glassblowers, Keki Mistry, Manfred Langer and Scott Bankroff machinists Deak Watters, Russ Geyer, and Dick Menke and NMR specialist Kermit Johnson. I appreciate the efforts and the patience of Margy Lynch, Carol Zink, Naomi Hack and Bill Draper who helped in the production of this manuscript. Finally, I acknowledge the tremendous support, help and friendship of the solid state research group, Drs. D. Issa, B. Van Eck, L.D. Le, A. Ellaboudy, M. Faber, and M. Tinkham, and 0. Fussé, J. Papaioannou, H. Huang, Z. Li, F. Tientega, J. Skowyra, L. Hill, J. Kim, M. Kuchenmeister, and again Dr. J. Dye. iii I acknowledge research support from the National Science Foundation Grants DMR 84—14154 and DMR 79-21979. iv TABLE OF CONTENTS CHAPTER PAGE TABLE OF CONTENTS. . . v LIST OF TABLES. . . . . . . . . .x LIST OF FIGURES. . . . . . . . . xiii I. Introduction. . . . . . . . . . . . l I.A Trapped Electrons in the Solid State. . . . . . . . . . . . . . . .2 I.A.1 Color Centers. . . . . . . . . . .2 I.A.Z Electrons Trapped in Glassy Matrices. . . 6 I.A.3 Lithium Tetraamine - Expanded Metals. . . . . . . . . . . .15 I.D Solutions of Alkali Metals. . . . .20 1.3.1 Metal Ammonia Solutions. . . . . . 20 1.3.2 Alkali Metal Complexation. . . . . 27 I.C Alkalides and Electrides. . . . .35 1.0.1 Alkalides. . . . . . . . .35 1.0.2 E1ectrides.. . . .42 I.D Objectives of this Work. . 45 II. Experimental Details. . . . . . . . . . . . . .48 II.A Synthetic and General Handling Techniques. . . . . .48 II.A.l Alkali Metals. . . . . . . . .‘. .48 II.A.2 Complexants. . . . . . . . . . . . 48 II.A.3 Solvents. . . . . .49 II.A.4 Synthesis of Alkalides and Electrides. . . . . . . . .50 11.3 General Handling Techniques. . 52 CHAPTER III. 11.0 11.0 II.D.1 II.D.1. II.D.1. H H 5555 £31:wa II.D.6 Recrystallization Methods for Single Crystal X— ray Diffraction Studies. Instrumental and Experimental Description. . . . . X- ray Crystallography Mounting of Crystals of Cs‘(1806)2-Na' . . X- ray Data Collection for Cs’(1806)2-Na . . . . X— ray Data Collection for Cs'(1806)2-e' . . . . . . . Mounting of a crystal of Cs‘(1505)2-I . X- ray Data Collection for Cs (l5C5)2-I' Magnetic Susceptibility. Optical Spectroscopy. Electron Paramagnetic Resonance. Differential Scanning Calorimetry. Nuclear Magnetic Resonance. Crystal Structures of Cs‘(1806)2-Na', Cs’(1806)2-e' III.A. III.A.l III.A.2 III.B. III.B.1 III.B.2 III.B.3 and Cs‘(l5C5)2-I‘. Background. . . . . . . . . . . Previous Work. . Fundamental Methods for Solving Structures. . . . . . . . Structure Solution and Refine- ment for Cs’(1806)2-Na' and Cs (18CS)2-e' . . . General Approach. Results of the Structure Determi— nation of Cs (1806)2-Na at 213 K. Results of the Structure Determi— nation of Cs (18C6)2-e' at 216 K. . . . . . vi PAGE .54 .58 .58 .60 63 66 66 68 .70 .70 71 .72 73 .73 73 .76 83 .83 84 .100 CHAPTER FAGB III.C. Structure Solution and Refine- ment of Cs*(1505)2-I‘. . . . . . .116 III.C.l General Approach. . . . . 116 III.C.2 Results of the Structure Determi- nation of Cs‘(1505)2-I I‘. . . . . .117 IV. Thermal Characteristics of the Iodides and Electrides of Cs'(1806)2 and Cs (15C5)2 Complexed Cations. . . . . . . .126 IV.A Differential Scanning Calorimetry. . . . . . . . .127 IV.A.1 DSC of Cs (15C5)2-I' and Cs (1866)2- I'. . . . . . . . . .127 IV.A.2 DSC of Cs (15C5)2-e and Cs (1806)2-e . . . . . . . . . . .129 IV.B Optical Spectroscopy of Cs (1505)2-e' . . . . . . . . .134 V. The Nuclear Magnetic Resonance of 13305 in Salts that Contain Complexed Cesium Cations and/or Cesium Anions. . . . . . . . . . . . . 139 V.A Cs —— 133 Nuclear Magnetic Resonance Studies of Crystalline Salts With and Without Complexe— tion by Crown Ethers and a Cryptand. . . . . . . . . . . . . 140 V.A.l Introduction. . . . . . . . . . . 140 V.A.1.a 133Cs NMR Spectra. . . . .140 V.A.2 Lineshape Functions for Static NMR Spectra of Powders and Single Crystals. . . . . . .143 V.A.2.a Dipolar Broadening. . . . . 143 V.A.2.b Chemical Shift Anisotropy. . . . .143 V.A.2.c Quadrupolar Broadening. . . . . . 146 V.A.3 Experimental. . . . . . . . . 148 V.A.4 Results and Discussion. . . . . . 149 V.A.4.a MAS— —NMR of Simple Salts and Complexed Salts. . . .149 V A.4.b Static Spectra of Simple Salts and Complexed Salts. . . . .154 V.A.4.b.i 052804. . . . . . . . . . . 154 CHAPTER VI. VII. V.A.4.b.ii Static Spectra of Cs (1806)2-I' . and Cs (1505)2-I'. Conclusions. . Acknowledgement. References. <<< >>> Qmm V.B Cesium-133 Nuclear Magnetic Resonance Spectroscopy of Alkalides and Electrides. V.B.1 Introduction. V.B.2 Experimental. V.B.3 Results and Discussion. . V.B.3.a 133Cs MAS- NMR of Alkalides. V.B.3.b 13305 MAS-NMR of Cs‘C222-e' V.B.3.c 1330s MAS~NMR of Cs’(18C6)2-e' V.B.3.d 13305 MAS- NMR of Mixed Alkalide- Electride Salts. V.B.4 Conclusions. . V.B.5 Acknowledgement. V 8.6 References. V. C MAS— NMR of Cs (1505)2-e Magnetic Susceptibility of Cs (1806)2-e Cs‘(1505)2-e . VI.A Background. VI.A.l Metallic Paramagnetism. VI.A.2 Diamagnetism. VI.A.3 Curie Law Paramagnetism. VI.A.4 Magnetic Ordering and the Curie—Weiss Law. . . . . . . VI.B The Magnetic Susceptibility of Electrides. VI.B.1 Magnetic Susceptibility of Cs (18C6)2-e . VI.B.2 Magnetic Susceptibility of Cs (1505)2-e . VI.C Electron Paramagnetic Resonance of Cs'(1505)2-e' . . . . . Conclusions and Suggestions for Future Work. . . . . . . . . . . . . . . VII.A Conclusions. viii PAGE .156 .168 .169 169 171 171 173 174 174 .179 181 188 .197 .198 199 201 .207 208 208 213 .216 .219 .224 .226 .229 246 257 257 CHAPTER PAGE VII.B Suggestions for Future Work. . . .263 REFERENCES. . . . . . . . . . . . . . . . . . . . . . .267 ix LIST OF TABLES TABLE PAGE 1 Values of log Ke eq for the Complexation of Alkali Metal Cations with Crown Ethers and Cryptands in Methanol [45]. . . . .30 2 Results of Analyses for Cs (15C5)2-e' and Cs (1806)2-Na x-e1_ . . . 53 -x 3 Table of Experimental Details for the Structure Determination of Cs’(18C6)2-Na' . . . .62 4 Table of Experimental Details for the Structure Determination of Cs‘(18C6)2-e‘ .65 5 Table of Experimental Details for Structure Solution and Refinement of Cs (18C6)2-Na' and Cs (18C6)2-e . . . . . . . . . . . . . 85 6 Table of Positional Parameters in Fractional Unit Cell Coordinates and Their Estimated Standard Deviations for Cesium (18-Crown—6)2 Sodide at 213 K. 86 7 Table of General Displacement Parameter Expressions -- U’s (in A2) for Cesium (18—Crown-6)2 Sodide. . . . . . . .88 8 Table of Least Squares Planes for Cesium‘ (18—Crown-6)2 Sodide. . . . . . . . . .93 9 Table of Bond Distances (in Angstroms) for Cesium (lB-Crown-6)2 Sodide at 213 K and Cesium (18—Crown—6)2 Electride at 216 K. 95 TABLE 11 19 20 Table of Selected Torsional Angles (in Degrees) for Cesium (18- Crown- 6)2 Sodide at 213 K and Cesium (18- Crown— 6)2 Electride at 216 K. Table of Closest Anionic Contact Distances for Cesium (lB—Crown-6)2 Sodide of 213 K, Cesium (18—Crown-6)2 Electride at 216 K, and Cesium (15-Crown—5)2 Iodide at 291 K. Table of Positional Parameters in Fractional Unit Cell Coordinates and Their Estimated Standard Deviations for Cesium (18—Crown-6)2 Electride at 216 K. . . . . . . Table of General Displacement Parameter Expressions —— U’s (in A2) for Cesium (18-Crown—6)2 Electride. . . . . A Comparison of Lattice Constants and Interanionic Distances in Cesium (1806)2 Sodide, Cesium (18C6)2 Electride, and Cesium (1505)2 Iodide. . . . . Table of Least-Squares Planes for Cesium (lB-Crown—6)2 Electride. Table of Positional Parameters and Their Estimated Standard Deviations in Fractional Coordinates for Cesium (15C5)2 Iodide at 291 K. . . . . . . . . . Table of General Displacement Parameters Expressions —— U’s (in A2) for Cesium (15~Crown-5)2 Iodide at 291 K. Table of Bond Distances (in Angstroms) for Cs’(15—Crown—5)2-I' at 291 K. . . . . Table of Least Squares Planes for Cesium (l5‘Crown-5)2 Iodide at 291 K. . . . Summary of DSC Results for the Iodides and Electrides of Cs’(1806)2 and Cs’(1505)2. xi PAGE .97 .99 101 .103 .106 .109 .118 119 121 .123 .131 TABLE 21 22 23 24 25 26 27 Chemical Shifts and Linewidths of Simple Cesium Salts With and Without Complexed Cations. . . . . . . . . . . . . . . Parameters of Anisotropic Interactions in the NMR of Single Crystals of Cs (15C5)2-I at 23. 62 MHz. . . . . Chemical Shifts and Linewidths of 13303 MAS—NMR Lines in Alkalides and Electrides. Summary of Calculated and Representative Observed Chemical Shifts of Alkali Metal Anions. . . Fractional Atomic Character of Some Cs----e‘ Solutions and Solids. Table of Least-Squares Best—Fit Parameters for the Temperature Dependence of the Knight Shift 9f Peaks in Cs‘(1806)2-Nax-e1_x. Table of Parameters and Their Estimated Standard Deviations for the Modified Curie- Weiss Law for Cs'(1505)2-e'. . . . . . . PAGE .150 166 .175 178 .187 .196 .232 IGURE LIST OF FIGURES PAGE The geometric structure about the trapped electron in glasses of a) 10 M NaOH, and b) ethanol [9—11].. . . The semicontinuum model for the trapped electron. rv is the void radius, rs is the effective spherical radius of a first shell solvent molecule, the dipolar interac- tion radius is rd and R is the distance from the center of the cavity to the dielectric continuum [12]. . . . . . . . . . . . . . Molecular structures of a) 18—Crown—6, b) 15—Crown—5, and c) Cryptand—222. . . . . . . .28 Ortep stereoviews of the complexed cations in a) K*(1806)-SCN' [51], b) Cs’(18C6)-SCN' [51], and c) Na’(1204)2-0H'-8H20 [52]. . . . . . 34 A schematic packing diagram of the cations and anions in Na*(0222)-Na' [60] . . . . . . . . 38 The 23Na NMR spectra of Na‘CZZZ-Na‘ in a) methylamine, and b) in the solid state [56,63]" . . . . . . . . . . . . . . . . .40 Apparatus for the recrystallization of alka- lides and electrides. . . . . . . . . . .55 Packing diagram of the cesium cations, represented by solid circles, and the sodium anions, represented by open circles, in Cs’(18C6)2-Na‘. Dotted lines represent cation-cation distances, dashed lines repre— sent anion—anion distances, and the weak xiii [GURE PAGE solid lines represent cation anion distances about one cation. . . . . . . . . . . . The structure of the complexed cation, Cs*(1806)2 in Cs*(1806)2-Na'. . . . . . . . . . .94 A stereoview of a unit cell in Cs’(1806)2-e’. The center of the anionic cavity is noted by the symbol 0. . . . . . . . . . . . . . . . . 104 A surface contour representation of an anionic cavity in Cs*(1806)2-e‘, as viewed down the 9 axis at a = 1/2, g = 0, g = 1/2. The cavity is shown at its maximum size in the g-g plane; channels to neighé boring cavities at l, 1/2, 1/2, and 0, —1/2, 1/2 are indicated by arrows. . . . . . . . .112 A surface contour representation of the packing of complexed cations and anionic cavities in an 339 plane at g = 0 in Cs*(1806)2-e'.‘ The oblong shape cavities and the channel structure in the 5 direction are apparent. . . . . . . . . . . . . . . . . 113 A surface contour representation of the packing of complexed cations in Cs*(1806)2-e‘ as viewed down the 1,1,0 direction. The interstitial open spaces are channels to cavities behind the "wall" of complexed cations. . . . . . . . . . . . . 114 DSC trace of a polycrystalline sample of Cs*(1806)2-I‘. . . . . . . . . . . . . . . . . .128 DSC trace of a polycrystalline sample of Cs*(15C5)2-e‘. . . . . . . . . . . . . . . . . .130, DSC trace of a polycrystalline sample of Cs’(1806)2-e‘. . . . . . . . . . . . . . . . . .133 Normalized optical absorption spectrum of Cs*(15C5)2-e‘ at -100 C. . . . . . . . . . . . .135 Normalized optical absorption spectrum of Cs‘(1505)2-e' at —35'C and —1°C. . . . . . . . .137 xiv FIGURE 19 20a 20b 21 22 23 13308 static NMR spectra, comparing line— widths of 032304 at 52.48 MHz and 23.62 MHz. Static 133Cs'NMR spectra of Cs’(1806)2-I’ at 52.48 MHz. The dashed lines are calculated spectra from equation 16 for = 6.5 KHz, 8 = 32 ppm and n - 0.. ”0 Static 133C3 NMR spectra of Cs’(18C6)2-I' at 23.62 MHz. The dashed lines are calculated spectra from equation 16 for v0 = 6.5 KHz, 6 = 32 ppm and n = 0.. . Static 13305 NMR spectrum of Cs'(1505)2-I‘ at 23.62 MHz. The dashed line is the calculated spectrum from equation 16 for = 30.5 KHz, 6 = 52 ppm and n = 0. ”0 Static 23Na NMR of Cs‘(1866)2-Na' at two temperatures, T g 210 and 250 K. . . . . Single crystal 1330s NMR spectra of Cs’(15C5)2-I' at two random orientations. Lines labelled A and B are central —1/2 <—-> 1/2 transitions for two Cs'(l5C5)2 orientations. Lines labelled a and b represent corresponding non-central quadrupolar transitions. . . . . . . . . . Single crystal 133Cs NMR spectrum of Cs (1806)2-I' at a random orientation. The sharp spike is apparently due to residual solution. . . 133Cs MAS-NMR spectrum of Cs‘CZZZ-e‘, showing inclusive and exclusive complexed cations, prepared from solutions that contained excess quantities of a) cesium and b) 0222. Other lines are spinning sidebands. . . . . . . . . . . . . The chemical shift versus T'1 for Cs*(1806)2-e'. The temperatures of points marked by x were calibrated by comparison with the dipolar splitting of protons in methanol. Other points were constrained to lie on the best—fit line through XV PAGE .155 .157 .157 .160 .163 .164 .167 .180 FIGURE PAGE the four calibrated points. 184 27 13303 MAS-NMR spectrum of the product obtained from initial stoichiometry Cs’(18C6)a. Peaks at ~50 ppm and —230 ppm are Cs‘(18C6)2 resonances, the peak at 77 ppm is a Cs'(1806)2-e‘ resonance. Other features are spinning sidebands. . . . . . . . . . . 190 28 13305 MAS—NMR spectra of compounds formed from initial stoichiometry Cs’(1806)2-Nax-e1_x for a) x = 0.2 and b) x = 0.8. Peaks at 73, 57, 42 and 25 ppm are from a mixed compound with x = 0.1-0.2; the peak at —61 ppm is from excess Cs'(1806)2-Na‘. Other features are spinning sidebands. . . . . . . . . . . . . . . . .191 29 An anionic packing_diagram showing a postu- lated mixed sodide/electride structure that could account for the observed mixed electrideesodide 133Cs NMR spectra. Open circles are trapped electron sites, solid circles are sodium anions. . . . . . .194 30 The chemical shift versus T‘1 for Cs’(15C5)2-e‘. The temperatures of points marked by x were calibrated by comparison with the dipolar splitting of methanol. The temperatures of other points were calibrated by simultaneous measurement of the Knight shift of Cs'(18C6)2-e’ . . . . . . . 202 31 The density of states with spin up (T) and spin down (1) in a metal as a function of energy, E, a) at equilibrium in the absence of an external magnetic field, b) prior to equilibrium upon application of a field, H, and c) at equilibrium in the presence of the applied field [94]. . . . . . . . . 210 32 The susceptibility of an antiferromagnet as a function of temperature, normalized to the values at T ; a) for a single crystal with easy axis u to the applied field, H, b) for a single crystal with easy axis 1 to H and the powder average. . . . . . . . . .223 FIGURE 33 34 35 36 37 38 39 408 40b 41 The reciprocal molar electronic susceptibility of Cs (1806)2-e' as a function of temperature [71] The reciprocal molar electronic susceptibility O of a quenched sample of Cs (l5C5)2-e function of temperature. . The reciprocal molar electronic susceptibility of an annealed sample of Cs*(15C5)2-e' function of temperature. The temperature dependence of the molar electronic susceptibility an annealed sample of Cs’(l505)2-e‘ as a function of the applied field. 8.88 888 o The molar electronic susceptibility of an annealed sample of Cs’(1505)2-e' of the applied field at 1.6 K. EPR spectra of Cs‘(l5C5)2-e', sample 1, at 2.0 K, 48 K, 200 K, 80 K, 14 K, and 5 K. . EPR spectra of Cs‘(15C5)2-e‘, sample 2, at 2.8 K, 19 K, 10 K and 2.5 K . . . . EPR spectra of Cs’(l5C5)2-e', sample 3, at 2.5 K, 6 K, 158 K, and 2.6 K . . . . . EPR spectra of Cs’(15C5)2-e‘, sample 3, at 7.5 K, 19 K, 50 K, and 2.6 K . . AH as a function of temperature for sample 1 of Cs’(15C5)2-e‘ xvii as a function PAGE 227 .231 .233 .235 .236 .249 250 251 .252 253 CHAPTER 1 INTRODUCTION Alkalides and electrides are compounds that consist of complexed alkali metal cations countered by alkali metal anions and trapped electrons respectively. This thesis concerns the physical characterization of the unique anions in these materials, with an emphasis on the trapped electron in the electrides. The primary techniques used to characterize the compounds are structural analysis, magnetic susceptibility and magnetic resonance. This chapter will explore the background information about trapped electrons in the solid state, first in the low electron density regime of F-centers and localized electrons in glasses, and then in the high electron density regime of metal ammonia compounds. The description of these systems will emphasize their structure and the magnetic properties of the electrons. Studies of trapped electrons originated with investigations of metal solutions in ammonia, amines, and ethers. The electron density can be varied easily in these olvents and the association of the solvated electron with ther ions in the high concentration limit (and as a function of the dielectric constant of the solvent) to form various paramagnetic and diamagnetic species will be discussed. Indeed, electrons show similar types of aggregate ion formation in the solid state, which may strongly affect the physical properties of the material. The complexation chemistry of alkali metals with crown ethers and cryptands allowed metal solutions to be studied at higher concentrations in less polar media; these complexes permitted more complete characterization of the associated ionic species. It was this development which led directly to the isolation of alkalide and electride salts by removal of the solvent. The fundamental properties of the alkalides and of some previously synthesized but incompletely characterized electrides will be reviewed. Finally the issues involved in the charac- terization of electrides will be presented in order to put the present work in perspective. I.A Trapped Electrons in the Solid State I.A.1 Color Centers The first example of electrons trapped in solids was reported 70 years ago when it was noticed that alkali halide salts that had been exposed to high energy radiation became colored [l]. The radiation interacts with the salt to form excitons, mobile excited state species which consist of an electron bound to a hole. The exciton can dissociate in the vicinity of an anionic vacancy to give a localized electron and a localized hole species. The Fe center is an electron trapped at an anionic vacancy. Several variables are pertinent to the stable fermation of the localized electron state including the energy of the radiation, the pressure, and the temperature, each of which can affect the mechanism of the excitation and thus the nature of the excitonic species [2]. For example, color centers may be produced in alkali halides by irradiation with ultraviolet light which strips the valence electron from the anion. The electron is initially loosely bound to the halogen atom at an energy just below the conduction band. The electron may be trapped at a pre-existing vacancy to form the F-center. Harder radiation is capable of creating the anionic vacancy by ejecting an anion from its lattice site; the electron may be trapped at the vacant site while the hole forms some other defect center. Regardless of the mechanism of its formation, the nature of the color center remains constant. Color centers can be formed at concentrations up to~1018 electrons/cma. At high concentrations there is a tendency for F-centers to aggregate and form more complex crystalline defects. The optical absorption spectrum of these colored salts was found to have a bell—shaped band with an energy at the maximum absorbsnce that was inversely related to the nearest neighbor separation g in Angstroms. An empirical relationship due to Mollwo and Ivey gives 8 = 17.7 a_1'84 [3] for the energy in electron volts of the absorption maximum. The breadth of the band increases with increasing temperature, while the oscillator strength remains constant. The Mollwo-Ivey relation roughly accounts for this phenomenon since the interionic separation, 5, will have a larger range of values for increased thermal vibration in the lattice [4]. The nature of the wave function of the electron at the ionic defect has been treated at various levels of sophistication. At its simplest, it can be approximated as a particle in a cubical box of edge-length 2 5. More complex potential wells which vary periodically similar to the Madelung Energy, with an average decrease proportional to l/r have been used. The idea that the electron might interact with more than one shell of nearest neighbors is substantiated by EPR and ENDOR studies. Hyperfine coupling due to ions as far away as the eighth nearest neighbor shell has been detected for color centers in KBr [5]. These experiments indicate that, while the electron has long range interactions with the ionic lattice, the bulk of the electron density (63%) lies inside the vacancy, and 99X of the electron density is found within the first two shells of ionic neighbors. Electrons that behave much like the color centers can be formed in the molten alkali halides by addition of pure alkali metal to the melt. The concentration of electrons can be varied by varying the amount of metal, which dissociates in the fluid to M+ and e“. It has been shown that molten salt color centers follow a Mollwo—Ivey type relation [6]. Although the concentration of metal can be raised to beyond the non-metal to metal transition, the concentration of color centers does not get large. Trapped 18 electrons/cm3) tend electrons in high concentrations (>10 to form aggregates with other ions, resulting in metal clusters of low charge. Even in this simplest of trapped electron systems some themes that will recur in similar trapped electron materials arise. The localized electron is highly polarizable; it can have a significant density away from the electron trap. The trapped electron tends to associate with other ionic species at high concentrations to form more complex ionic aggregates. The local structure of the color center is known since the electron is trapped at an anionic vacancy in a simple salt structure. Even so, it is _____A apparent that to fully understand the localized electron its interaction with the local environment must be probed. I.A.Z Electrons Trapped in Glassy Matrices Ionizing radiation has been used to generate electrons in non—polar systems such as amines and ethers. and glasses of alcohols and H20. Electrons trapped in fluids will be discussed in section B of this chapter. The study of electrons trapped in glasses has been especially fruitful for describing the general solvated electron structure. ‘In contrast to the electrons trapped in ionic vacancies in a crystalline lattice, which were surrounded by attractive and repulsive charges that stablize the trapped electron, electrons localized in non-polar media often are not stable with respect to reduction of surrounding molecules. An electron injected into water has a half—life of about 1 millisecond with respect to reduction of the solvent [7]. The localization itself is not an obvious process. There are no vacant lattice sites, and no coulombic attractions to charged species with which to interact; instead, the electron interacts with the dipole soments of the solvent. The work that will be discussed here involves injection f electrons into glasses at temperatures between 4 and 7 K. The lifetime of trapped electrons is greatly enhanced over that in fluid systems due to the reduced thermal motion of the medium. The mechanism for solvation of the electron has been studied both in fluids and in glasses, primarily by rapid spectroscopic methods. After complete solvation, the trapped electron is assumed to have an average equilibrium structure which can be determined in glasses by electron spin echo modulation spectroscopy. The geometric structure of the cavity about an electron in a glassy matrix might be expected to be similar to the average structure in solutions.~ The theoretical ideas concerning the localization of electrons should be applicable to both glasses and solutions. A brief outline of the most universally accepted model for the trapped electron will be presented. An injected electron is trapped in a glass or solvent by a two stage mechanism, in which the first step accounts for the localization of the itinerant electron at a pre—‘ existing trap, and the second step is the equilibration of the trap into the IOWest energy configuration. The first tep is rapid, typically on the order of 10 picoseconds in 20 and 40 nanoseconds in glasses. The amorphous structure f the glass has a large number of sites arising from ortuitous random nolecular orientation of the dipoles hich serve as initial traps for the electron. The time- cales of localization into pre-existing traps were determined by picosecond pulse radiolysis experiments [8]; the optical absorption maximum tends to occur at ~1500 nm in most solvent systems. As the trap relaxes to an equilibrium structure, the low energy peak decays and an absorption at ~600 nm arises which is stable for the life- time of the species. The structure of the equilibrium salvation shell of an electron trapped in a glassy natrix has been deduced by electron spin resonance techniques. The hyperfine interaction of magnetic nuclei surrounding a paramagnetic electron is very informative in terms of the strength (distance) of the coupling and the nuclear species involved. Traditional methods for determining hyperfine coupling failed in the glasses because broad resonance lines obscured the weak couplings. Kevan [9] employed a multipulse electron resonance technque known as electron spin echo modulation to probe the hyperfine interactions of the trapped electron. Several pulse sequences may be used to measure the modulation function. The most common experiment was a two pulse EPR sequence in an applied field, E, oriented along the z axis. The pulse sequence must be short relative to the relaxation time of the electron spin. The physical description of the pulse sequence is as follows. A 90' pulse is applied perpendicu- lar to the applied field, which flips the electron spins into the x-y plane. The spins are allowed to precess for a time, 7, during which they dephase in the x-y plane due to various hyperfine contacts. A second pulse is applied which flips the spins 180' and keeps them in the x-y plane. The dephasing which had occurred during time 1 rephases after another period of w, releasing a burst of microwave energy whose intensity is measured as a function of 1. Two factors influence the strength of the microwave emission; first, the relaxation of the spins causes an exponential decay of the power with time, and second a modulation function arises from the dephasing influence of the dipolar hyperfine contact with neighboring nuclei [10]. The modulation function can be fit with an equation which has three parameters: the number of neighbors, the hyperfine coupling constant, and the distance between the electron and the interacting nucleus. In order to get unique neighbor information, selective deuteration of equivalent protons was used. Since the magnetic moment of deuterium is much smaller than that of the proton, the hyperfine contact with only the substituted deuterons could be measured. hus, added information about the orientation of solvent olecules around the trapped electron could be obtained. The average geometric structure of the solvated lectron in glasses made from 10 M NaOH in water and in are ethanol glasses is shown in Figure l [11.12]. The 7 ._ ____ 10 0) b) gure 1. The geometric structure about the trapped electron glasses of a) 10 M NaOH, and b) ethanol [9-11]. 11 structure of the aqueous glass was found to be independent of the ions in solution; thus, the only hyperfine interactions of the trapped electrons were with protons on 320. The best fit to the modulation function gave two proton distances at 2.1 A and 3.5 A for six nearest neighbor molecules. Studies with enriched H2170 verified the coordination number. A geometry with an octahedral arrangement of the six water molecules oriented with an O-H bond directed toward the center of the electron cavity agreed with all of the data. The electronic character at the first shell molecules from hyperfine constants was only IX, so that 96* of the trapped electron density resides in In s-type wave function centered at the cavity. The geometric structure of a trapped electron in ethanol required selective deuteration methods to determine the orientation of the solvent molecules. The results ndicated that four ethanol molecules arranged etrahedrally form the electron cavity. The distances to eighboring protons are 2.2 A to the hydroxyl proton, 3.3 A o the B-proton and 3.8 A to the a-proton. The only eometry that coincides with these data is one which rients the molecular dipole moment of the ethanol towards he center of the cavity, rather than the O-H bond as was Ben in H O. The trap has remarkably similar dimensions to 2 he cavity formed in H20. Dielectric .Concinuum Figure 2. The semicontinuum model for the trapped electron. rV is the void radius, rs is the effective spherical radius of a first shell solvent molecule, the dipolar interac— tion radius is rd and R is the distance from the center of the cavity to the dieletric continuum [12]. ~— 13 The results of electron spin echo modulation experi~ nts concurred with a theoretical model known as the semi- ntinuum model, which was independently preposed in 1970 two groups [13,14]. The semi-continuum model gives only mi—quantitative agreement to observed electrical, Optical d magnetic data, but is still regarded as the best model r the trapped electron. The semi—continuum model, depicted in Figure g, assumes t the excess electron is trapped in a spherical cavity radius rv created by a finite number, 2, of solvent ecules. The 2 first shell molecules are treated as al point dipoles in their electrostatic interactions th the electron, but are assigned a real spatial radius Solvent beyond the spherical volume of radius R = -r of the cavity and first solvent shell is treated as a electric continuum. The energy of the trapped electron determined by long range interactions with the continuum . screened coulombic polarization, and by short range erections with the first salvation shell. The model n has three parameters, the coordination number 2, the ity radius rv and the free (itinerant) electron energy. agreement with the structures implied from electron n echo experiments is good. Typical values for the ameters are 2 = 4—8 and rv = 1.7—2.2 A. The semi- inuum model is an extension of Jortner’s [15] earlier 14 tinuum model which treated the electron cavity immersed the dielectric continuum without accounting for any licit short range forces. The major improvements of the l—continuum model are that the energy required to orient [first salvation sphere is accounted for and the r »ngest dipole attractions of the electron with the first 'atian shell are explicitly included. The ground state he electron is a Is type wavefunction, and various odsethods have been used to calculate the energy of the ped electron. The agreement with experimental data s not to be quantitative; however, the model does ict trends with variation of temperature, pressure, and ents. It reproduces the optical absorption maxima, but not properly account for the shape of the band. Two types of solids which are able to localize trons into stable traps at cavities in the medium in :oncentrations have been reviewed. The predominant a that stabilizes the F-center is clearly the reduction 1e coulombic repulsions of the cations bordering the toy. It is remarkable that non—ionic systems are able ’calize an electron at energies that are 1—2 electron below the conduction band by primarily long and short : polarization of dipolar molecules. l5 A.3. Lithium Tetraamine —- Expanded Metals Lithium (as well as other alkali metals) dissociates in quid ammonia to form the solvated cation and electronic ecies which vary in nature depending on the concentration ‘the metal. At a concentration of one lithium to four mania molecules there is a deep eutectic in the phase sgram. The solution has a low vapor pressure and a large Jthermic heat of solution at the eutectic, which is lque to lithium in the alkali series; this indicates that :NHa)4 forms a compound rather than being an immiscible Item. The conductivity of the solid phase was measured; ! results were not those of a eutectic mixture of Li and Ionia, but of a compound [16]. Several of the alkaline ‘ths form similar compounds of stoichiometry "(NH3)6 for Ca, Sr, or Ba. The properties of only the lithium raamine will be discussed here, as it has the lowest ctron density and thus most nearly approximates ctrides. The solution freezes at 89 K to form a gold id that retains many of the nearly free electron parties of the solution. The electron density in the 21 electrons/ems, much greater id is approximately 5 x 10 n the electron densities in either the color centers or irradiated glasses. 16 X-ray and neutron powder diffraction patterns have been analyzed as a function of temperature for this compound [[17,18]. The compound Li(NHa)4 crystallizes into a body centered cubic lattice with 16 formula units per unit cell. The lattice parameter a is 14.98 A between 89 and 82 K; at 82 K there is a phase transition below which the lattice parameter is only 14.93 A. Interestingly, the fully deuterated ammonia compound does not have a different high :emperature crystalline phase; it forms only the low :emperature phase upon crystallization at 89 K. As the :emperature is lowered still further to 25 K neutron liffraction shows a transition to an antiferromagnetic Ihase as indicated by a doubling of the magnetic lattice »arameter. The magnetic susceptibility of this material has been easured over the entire temperature range of 4 to 300 K 19,20]. The magnetic susceptibility is, in this case, a ensitive probe of electronic interactions with other aramagnetic electrons. Above 82 K in Li(NH3)4 the usceptibility is Pauli paramagnetic, indicative of a early free electron metal. The phase transition at 82 K auses the system, which is still metallic by conductivity 1d Hall measurements, to show an inverse temperature apendence in susceptibility following the Curie Weiss law. to strongly correlated electron gas arises from the larger 17 -Li distances (or the low electron density) such that the ectron density is found to partially occupy cavity ates, so that the electron gas is not fully degenerate. The picture that is generally drawn for the lithium traammine system is that of an "expanded metal" [21]; at is, a Li(NH3)Z ion enclosed in a spherical Li 3s ectron orbital. The Li 2s and 2p orbitals are hybridized a set of op3 orbitals which interact with the lone pairs the ammonia. There is no evidence that anionic electron ips away from the nuclear species exist in this system; more strongly corresponds to'a metallic structure lilar to that of pure lithium, a closest packed lattice cationic cores bound by the electron gas. Compounds which have stoichiometry Li (methylamine)4 e also been isolated. Preliminary studies of the netic susceptibility, ESR and conductivity of the erial indicate that this system is also a highly related metal [22]. The solid state materials that have been discussed ve show trends and characteristics that must be sidered when discussing electrides. The low electron Iity glasses or alkali halides have electrons that are oped in otherwise void space. The electron has a iency to "leak" into its surroundings, spreading its lity over its nearest neighbors. Highly polar ionic 18 Its enable electron density to spread into distant ionic 211s more easily than do the relatively non—polar lutions. The trapped electrons in alkali halides and asses are present at such low concentrations that there :no interaction with other trapped electrons, and, in the ie of the glasses, with any ionic species. There is also dence in these systems that, as the electron density :reases, there is a tendency for much more complex cies to form, which involve more than one electron /or cation in direct association. Certainly, in the metal ammonia compounds the electron- ctron interaction is strong enough to form the band type rgy levels characteristic of metals. Mott described the dency of materials to go through a transition from a illic state to a non—metallic state in terms of two Imeters; the effective Bohr radius (a‘) of the outermost :tron and the electron density p. The Mott criterion :es that a material should go metallic when pl/aa‘ )0.25 . The idea is that in a given medium a valence :tron will have same volume which can be represented in us of a spherical Bohr radius a“. The electron is 'acted to a positive charge with a potential energy ortional to -l/r. As the dielectric constant of the ‘um is increased there is a decrease in the attraction he electron to neighboring charges, thus increasing a‘. 19 s the concentration of valence electrons is increased, the verlap of electrons increases the screening between an lectron and positive charges. The combined effect of ncreasing both a‘ and the electron density, p, is to educe the energy required to remove an electron from one ositive charge and place it an another. When the electron s free to move it is considered to be metallic [24]. The lectron density in Li(N83)4 can be calculated from the 21 3 tructure and is 4.8 x 10 e-/cm . The effective Bohr adius is a'II = 2.76 A. The Mott criterion for metallic 1 aracter is satisfied, p /°a‘ = 0.47 which is greater than .25. If it is assumed that the electron density in 1(NHZCH3)4 is one-half that seen in the tetraammine, then he Mott criterion is satisfied for a“ >1.87 A. The ielectric constant in methylamine is smaller than that of Imonia so the radius a‘ would be reduced, and the material I expected to be just at the metallic- nonmetallic »rderline. The metal ammonia compounds are very near the metal— tnmetal transition. It is unfortunate that other mpounds such as Na(NH3)6 or Li(NHZCZH5)4 do not exist, r the reduction of electron density or dielectric nstant could push these compounds to the nonmetallic ime. It would be interesting in these cases to see ther the nonmetallic compounds would have the electron 20 localized in excited s—orbitals or whether trap states would have a lower energy. It might be expected that electrides are within an order of magnitude in electron density of the expanded metals. Since dielectric constant and electron densities of electrides might be expected to vary from one compound to another the study of their electron—electron interactions might provide valuable insights into the mature of the metal-nonsetal transition. I.B Solutions of Alkali Metals .B.1 Metal Ammonia Solutions Most of the scientific attention to the properties of rapped electrons has been in the area of netal solutions 0 ammonia, amines, and ethers, rather than in the solid tate materials described above. Indeed, metal ammonia impounds, electrons in glasses, and alkalides and lectrides have evolved from research initiated in the atal solution area. There are advantages and disadvan— Lges to studying solutions rather than the solid state. Ie greatest advantage is the ability to increase yntinuously the electron concentration in solution by .creasing the metal concentration and/or to change the 21 ectric constant of the medium by varying the solvent. major disadvantage is that ions in solution are able to ate easily and rapidly; this makes the structures of ionic species and their interactions with other ions icult to measure. The published observations of Weyl [25] in the 1860’s he physical nature of solutions of alkali metals in id ammonia were the starting point for what was to »me an enormously popular field of research in the ical sciences. Davy’s notebooks of 1808 [26] indicate he had observed ffine blue solution films" upon mixing ssium and ammonia. It wasn’t until 100 years later Kraus began to make progress in the understanding of l ammonia solutions by measuring physical properties as the vapor pressure above the solution [27] and the ctivity as a function of metal composition and concen— on [28]. The experimental and theoretical work in field is comprehensively reviewed in Thompson’s book, trons in Liquid Ammonia” [29], and in the Preceding: uoque yeyj’n [30], V [31], and VI [32]. The ssion here will focus on those aspects of metal ions that concern the structure of the solvated ions heir interionic interactions. This may aid in the pretation of results obtained in solid state studies slides and electrides. 22 The alkali metals (Li, Na, K, Rb, and Cs), some aline earths (Ca, Sr, Ba), and some rare earths (Ru and dissociate in liquid ammonia according to the ilibrium + _. M + N83 w MB + e. (la) ++ - M + NH3 3 Ms + 2es 1 (lb) re the nature of the ionic species can be complex 3nding on the metal concentration. Similar solutions of 11s in solvents with lower dielectric constant may form in general the total metal concentration in solution is 4 LI. Very dilute metal solutions in ammonia [below 10' percent metal (MPM)] consist of essentially non- racting solvated ions. The magnetic susceptibility has general limiting result that the number of paramagnetic ies is equal to the metal concentration [33,34]. The ty in which the electron is trapped is considered to be lar to that described by the semi-continuum model [l2]. optical absorption, electrical conductance, excess me of solution of the metal and magnetic properties elate well with the model of a solvated electron. As oncentration of the metal is increased (or as the ctric constant of the solution is decreased) the ionic 23 cies tend to interact more strongly with one another ]. Electron-cation interactions result from the ractive coulomb potential between the ions, drawing them 0 an associated ion pair species of stoichiometry Msolv' structure of the ion paired species can vary between extremes; the solvent-shared ion pair and the solvated a1 atom. The ions in the solvent-shared species retain .ir respective traps, but one or more of the solvent .ecules may form part of each ion’s salvation sphere. . solvent-shared ion pair seems to be the most accurate cription of this species in metal-ammonia solutions, in ch optical absorption and NMR Knight shift data indicate t the electron trap retains its integrity. The lectric constant and donor number of the medium are ortant in determining the extent and type of ionic erection. Higher dielectric media favor less associa— n of ionic species. The contact ion-pair- solvated ion- ? equilibrium in this case has been driven to the left he media’s inability to solvate the ions. The ngest evidence of atom—like associated species comes magnetic resonance detection of hyperfine contact of excess electron to the cationic nucleus [36]. KPH riments in the less polar solvents show hyperfine tting patterns from which the extent of atomic acter can be calculated. In more polar solvents, rapid 24 ange between the associated species M and ionic species ows the EPR line so that hyperfine information is lost. studies of the cations permit detection of a magnetic shift due to the hyperfine contact in all ents [37]. The general trend in these experiments is the ion-pair is of smaller atomic character as the rity of the solvent is increased, and the solvent— ed species is favored. As the dielectric constant of medium is decreased, the associated ion pair tends rds the ”solvated metal monomer". The hyperfine tent was measured for K solutions in NH3, MeNHz, EtNHz, THF, and the density of the trapped electron at the eus corresponded to 0.5, 3.0, 12, and 36% atomic icter respectively. Edwards and Catterall [38] found Ca in frozen HMPA solutions gave solvated species with Itomic character. In addition to the paramagnetic ion pairs, even more .ex diamagnetic species form as the metal concentration creased in ammonia or as the dielectric constant is, ased. Magnetic susceptibility results indicate that e metal concentration in NH3 approaches 1 MPH, the tration of paramagnetic species is an order of magni- maller, so that most of the electrons are interacting ach other strongly enough to couple their spin r momenta. to pat are am 303 hi ve co in ti :3 to te P! r—r——i 25 Ehe nature of the spin-paired species is more difficult ptermine than that of the paramagnetic species because Fagnetic probes such as EPR and magnetic susceptibility tot useful. NMR techniques are also unable to rtain the nature of the associated species in polar ants, because rapid exchange broadens the resonance to the information, while non-polar solvents support only ‘low solubilities of the ionic species. Theoretical .derations of a doubly occupied electron cavity :ate that e2 could be stable with respect to dissocia- into two solvated electrons [39], but if this were a species the volume of the solution would be expected crease with spin-pairing, and the equilibrium would to shift to the spin-paired state with increased ure; neither effect is seen in any solvent. Other gnetic species which could form are the dimer M2, the riple e_-M+-e-, and the alkali metal anion M- [35]. irst species could consist of dimers of any of the aired paramagnetic species discussed previously. The ture of the ion triple may vary between the solvent- 1 species and the contact ion species. The three ions solvent shared ion triple share one or more solvent iles, but retain their respective traps. The contact riple includes direct interaction of the three :ed ions. Ammonia solutions almost surely have so th an in ca '1: t1 8! ti 26 :nt shared types of electron-electron coupling. Beyond .iamagnetism of the solution there is no evidence of hanges in the properties of the solution that would ate a trap greatly different from that seen at low ntration. The limiting case of the contact ion-triple at in which the two electrons occupy the s orbital of lkali metal, giving the third species M;. This as has not been observed in ammonia solutions although adynamic arguments have been constructed that predict l_ should exist in ammonia solutions [40]. Optical ra were obtained in solvents such as ethylenediamine, show a metal dependent peak attributable to the M“. :s [41], which followed the trends predicted by charge ?er to solvent theory [42]. There was, however, no I evidence that pointed to the presence of alkali anions with filled s-orbitals as the proper descrip— f this species. Other spin-paired species could also t for the absorption. Again the diamagnetism of the s and the low solubilities of the alkali metals in w dielectric solvents (where stronger ionic ation is favored) prevented direct determination of ructure of the spin-paired species. hancement of the metal solubility in non-polar ts became possible in the late 1960’s with the pment of a new alkali metal complexation chemistry. .V .m: 27 avelopment was crucial in the determination of the of the anionic species in low dielectric media. Alkali Metal Complexation ~ chemistry of alkali metals prior to the 1960’s had nfined to the hard sphere ionic nature shown in the ace state. ,With growing recognition of the nce of Na+ and K+ in biological systems, researchers interested in the chemistry of ionic transport. a and Moore [43] discovered that the antibiotic vcin could act by specifically binding alkali this became a challenge to organic syntheticists iduce such complexation by smaller molecules. I [44] in 1967 succeeded by synthesizing a class of ls known as crown ethers. generic crown ether m-Crown-n (or an) consists of ethoxy groups forming a carbon and oxygen ring atoms, and n oxygens. Figures 33 and 3b show the 1505 molecules. Side groups such as benzo- or. ano groups can be attached to the ring. 'These crown ethers interact with alkali cations via a barge attraction in a very selective way. The ring of the crown ether formed by the oxygens. is fized by a hole radius rH which is on the order of 28 a) ‘ E0 03 b) (O O} 1c) ‘<,fu hfl‘>' Quit? O \__/ Figure 3. Molecular structures of a) 18-Crown-6, b) IS—Crown-S, and c) Cryptand—ZZZ. 29 a ionic radii of the alkali cations. A 1:1 complex tends form when the cation interacts strongly with the oxygen Ie pairs that line the hole. The stability of the Iplex depends on the "fit" of the ion into the hole; if 1 ion is too small the mean contact distance lengthens, it is too large the ion cannot fit into the center of : ring. Lehn and coworkers in 1969 synthesized yet another [ortant class of compounds known as cryptands [46]. 'ptands are three dimensional bicyclic polyethers sisting of two nitrogens tying down 3 strands of m, n, o ethoxy groups, respectively, for CmnO. Cryptand 222 22) is drawn in Figure 3c. The three dimensional nature the polyether cavity puts a more stringent constraint on size selectivity of cryptands than the crown ethers. general result is that cryptands tend to have larger. ilibrium constants for the complexation reaction than do wn ethers. In addition cryptands are more selective for ions based on size effects [47]. Tab‘ COI 30 Values of log [eq for the Complexation of Alkali Metal Cations with Crown Rthers and Cryptands in Methanol [45]. D‘-------‘_—'--------”—-—_--_—-——-—_---——--—--~_--— l‘—-----—---------—-———’----‘--—-----—_---—-—------ :xant 1505 1806 2107 0222 ndius 0 85 A 1.38 A 1 70 A Radius 0.74 ~1.7 1.02 3.3 4.4 1.7 7.2 1.38 3.7 6.1 4.2 9.6 1.49 3.3 5.3 4.5 8.3 1.70 2.0 4.6 5.0 3.4 --_‘---------‘--—------¢_--------_‘--------——“---— --———---——-.-‘--‘-__‘---_--‘-—.-——-------—--.¢—-—- e 1 shows the equilibrium constant for 1:1 tion reactions in methanol. The equilibrium here M+ + c a n+0 (2) :n be seen that several trends arise that are related,to the size effects of the ion and the The series of 1806 with the various cations shows 06, which is a perfect fit, is more stable than nd Cs+1806 (each of which have ionic radii ~o.3 A than K+) by nearly two orders of magnitude. Even 1 is only 0.1 A larger than K+, has a K89 nearly ten coup equa cons the cryp its fit ire: val! equi 180( acc< 9382 the Par' 3th of 2.0 Gin div 31 : smaller. 1505 does not form such strong I with any of the cations, because its hole radius pr than the ionic radius of Na+. 1505 is nearly +, Rh+g the 05+ equilibrium lble to complex Na+, K is an order of magnitude smaller. The trend in r dimensional cryptands is even more dramatic. The molecule is quite flexible and is able to change rmation to better complex small cations. The best 222 is K+, the equilibrium constant is ~1000 hen the for complexation of K+ with 1806. The Keq for Na+ and Rb+ are similarly increased. The um constant for 0s+ is smaller for 0222 than for cavity in the cryptand is too small to readily te 0s+, whereas the two-dimensional 1806 can Iplex the large cation. >mp1exation equilibrium has a strong dependence on It as well as on ionic radii. The more able any ’ solvent is to solvate the bare cation, the more t is able to compete with the ligand. The value 9) for complexation of K+ with 1806 in water is orders of magnitude smaller than the same I in ethanol [45]. :ryptands were very important in the historical : of alkalides and electrides, and will be .ater in this context, the present work is almost 011' vi :1 32 vely based on crown ethers and further discussion : limited to their pr0perties. The complexation of metals by crown ethers is not just as simple as the »mistry that has been discussed so far'[48]. The ing is quite flexible so that many conformations of may exist. For cations which fit into the plane of gens, 1:1 complexes are the most stable [49]. cations that are too large for the hole most sit on and may be Complexed in a 1:2 ratio (the cation bed” by two crown ethers) or in a 2:3 ratio (a ndwich” structure). Large crown ethers which have times larger than the size of the cation may assume rmation in which the-cation is surrounded by the ther in much the same way that the cork center of a L is surrounded by it’s seam (i.e. Na+ and 2408) Itructures have been determined for several crown >mp1exes with alkali metal cations including the 1:1 + :s of ”+1806-SCN_ for M+ = Na , K+, Rb+, and 0s+ the sandwiched 1:2 complex of Na+(1zc4)2-OH’-aa o 2 base structures indicate how the conformation of tively flexible crown ether changes to best complex on. For example, in the thiocyanate series, K+1806 irly planar arrangement of the six ether oxygens : equator of the cation, whereas the smaller Na+ is l by a plane of 5 of the 6 oxygens, the sixth is 10 p1 It di ea at at 1« 33 above the plane to make closer contact to the small Rb_ sits 0.94 A (and Cs+ sits 1.44 A) above the mean of the ether oxygens [51]. The Rb+ and Cs+ ares are also characterized by the formation of ; two cations sit above their respective rings facing ther. The structure is stabilized by two anions, are each coordinated to the two cations. Figure 4 structures of the complexed cations in the K+1806 :he 0s+1806-SON_ and the Na+(12c4)2-ou‘-83 o salts to 2 ‘ate the variety of complexes that can form. It is Lhat the structure of the complexed cation could .y affect the interactions of the cation with its rnvironment. enhancement of metal solubility by alkali complexa- low dielectric media allowed the study of ionic to be carried out at higher concentrations by magnetic resonance methods [53]. The nature of line and metal-ether solutions was explored by Dye workers in the early 1970’s. Itions of sodium and Cryptand 222 in ethylenediamine Ldied by 23 Na NMR [54], and the two-peak spectrum ulted provided the first positive identification of li metal anion. The first peak was shifted by relative to the chemical shift of Na+ at infinite in water, the same peak position was observed in a 41. Ortep stereoviews of the complexed cations in 18C6)-SCN’ [51], b) Cs+(18C6)-SCN' [51], and c) C4)2-0H‘-8H20 [52]. 00 di at Na SP e1 fr 3C 8! R] T1 ._—.._. -.'|-_. 35 222 salt solution; thus, the peak is due to a ed Na+-0222. The NMR of complexed cations will be ad in Chapter V. The second peak was a narrow line ppm, at exactly the position calculated for gaseous I. Thus in low dielectric constant media the M- exists as a true alkali metal anion with two Is filling the 3s orbital, essentially unperturbed [as-like state. As the dielectric constant of the is reduced, the tendency to form the alkali metal increased, the chemical shift of Na- in MeNHz, nd THF were correspondingly found at —61 ppm [56]. major diamagnetic anionic constituent in solvents olarity has been determined, but the problem still in ammonia; there is no direct identification of -paired species. Alkalides and Electrides Alkalides tolutions of alkali metals with 0222 were found to :resting properties when the solvent was removed. rf the recombination of the ionic species to form ing materials as is seen in the pure metal , the residue was a gold colored crystalline 901i 222 ch: can uni let the yie up: Na so? so bo at no an at tl 36 The material that was isolated from sodium cryptate ;tions was found to be the first example of a new ' compounds now known as alkalides [57]. Alkalides ascribed as ionic solids that contain alkali metal The -1 valence state for a member of the alkali as very unusual, but a full characterization proved gnment. optical spectrum of a thin film of Na+0222-Na- a broad absorption with a maximum at 650 nm. This was nearly identical to that seen for solutions of hylenediamine [58]. This absorption band in had been attributed to a charge-transfer-to- band for the alkali metal anion. A similar bound- .e. 3s*3p) transition is postulated for the solid ion. In solution however the nature of M— could scribed necessarily to the pure centrosymmetric 3s2 .nce various ion pairs could account for the 650 nm ln. Likewise the optical absorption of a film of le was not definite proof of the alkali anion. e Na+0222cNa- appears to be metallic with a gold lnductivity measurements indicate that it is not a :uctor. The material behaves as an intrinsic semi- ‘ with a band gap of ~2.4 eV. Recent measurements that the conduction that occurs in this material nificant ionic contribution. It appears that a I‘E 37 .ble migration of Na+ to the cathode occurs [59]. r proof that the material was a true sodide was ed when Tehan, Barnett and Dye [60] solved the structure of the material. Figure 5 shows the f of anions about s complexed cation of Na+0222-Na-. ’ucture consists of closest packed cryptated cations e sodium anions filling the octahedral holes. The ed cation consists of a sodium cation in a nearly ral arrangement of cryptand oxygens, with the Na—O the proper sums of the van der Waals radii of Na+ ten. The anionic radius, estimated from van der ontact with the nearest cryptate protons to the is ~2.2-2.4 A. The Na_ radius is slightly larger 3 I- radius (2.2 A), and corresponds well to the >proximation that for monovalent ions the sum of the Idii equals the sum of the atomic radii. :r the determination of the crystal structure of Na* it became a prototype for the development of niques used to characterize other alkalides and es. The thermodynamics of its formation from the d complexant were experimentally determined by high precision B alumina electrode [61]. The Gibbs rgy of formation was found to be —7.1 +/4 0.6 KJ at 298 K. The photoemission spectrum of ’a— has been measured [62]. The spectrum consists 38 No+ ' \§: - hkfl' \\, 'e 5. A schematic packing diagram of the ns and anions in Na+(C222)-Na‘ [60]. 01 31 tl ——— 39 ask at 370 as due to photoemission from ha— and one nm due to photoemission from electrons trapped in :tice. .id state Magic Angle Sample Spinning (MAS) NMR [ues have become very important in the characteriza- F alkalide salts. Figure 6 shows the 23Na MAS NMR 1m of Na+0222-Na_ along with the spectrum of the same ll dissolved in methylamine [56,63]. The similarity two spectra strengthens the assertion that the in low dielectric solutions are alkali metal anions. plexed cation resonance occurs downfield from the amagnetic anion resonance, and can usually be ced in other complexed salts such as Na+0222-I~. e, theNa- resonance occurs at the value (-61 ppm) ed for Na; by theoretical calculations [55]. MAS capable of detecting complexed 7Li, 23Na and 133Cs as well as 23Na, 39K, 87Rb and 133Cs anions, and lme a powerful technique for identifying the lmetry of various alkalide salts. e the synthesis of the first alkalide in 1973, more others have been isolated, including cations of the stable alkali metals, and anions of each except a complexants include the cryptands 0222 and 0211 the crown ethers 1204 [65], 1505 [66], and 1806 40 L -60 -100 ppm The 23Na NMR spectra of Na+C222'Na' in a) methylamine, ;he solid state [56,63]. te en 10 pa 1‘! 41 optical spectroscopy of thin films of alkalides a follow the charge-transfer-to-solvent order of the of maximum absorbsnce: larger anions absorb at nergies [68,69]. Red or blue shifts may occur for anion, depending on the band structure for that lar salt. Conductivities tend to be those of lc semiconductors, with band gaps typically in the F 0.6 to 1.5 eV [70]. These low apparent band gaps :cause, in general, significant populations of electrons tend to reside in the crystal. The -e1ectron population allows an EPR signal to be I in most compounds. Hyperfine structure arising rapped electron interacting with the cation can be in nh*(18cs)-Na' [71] and x*(13cs)-Na‘ [72], the BPR resonances in other salts are typically featureless. The hyperfine structure consists of ines for 87Rb and 4 lines each for 85Rb and 39K dicates that the paramagnetic electron trapped in 3 seems to be associated with only one cation ian with several equivalent cations. The bulk susceptibility is diamagnetic, showing that the concentrations are very low. LC all re] It 0p' 9e 18 or ab 8F in 01 42 Electrides the same time that optical spectra of thin films of as were being determined, the effect of varying the 3 concentration of metal-to—complexant was studied. Found that complexant rich solutions left films with absorptions further into the infrared than had been :h alkalides [69]. In particular, films of 0s and .:1 or 1:2), Na and 0222 (1:1), and Cs and 0222 (1:1 made from N83 solutions gave peaks which had won maxima between 1300 nm and 2000 nm. These are similar to the solvated electron spectra seen e metal ammonia solutions, including the presence sorption tail at high energies. Films of K0222 d Li0211 (1:1) had spectra which were similar to ma absorption seen in more concentrated (metallic) monia solutions, an increasing absorbance from ith no maxima to 2500 nm. The analogy to the electrons in metal ammonia solutions led Dye to se films electrides [68], the first salts that trapped electrons as the stoichiometric anionic ent. ers of Li+c211-e’ [73] and K+0222-e_ [74] were and preliminary characterization showed them to different from alkalides. The dark blue to black coI co op' bi «3 pa EP be 43 if the powders contrasted with the gold/bronze/copper seen in alkalide crystals. In addition to the . spectroscopy referred to above, magnetic suscepti— rs, microwave conductivities, and RPR spectra were 2d. Li0211 showed a temperature dependent spin t both in magnetic susceptibility measurements and ctra. At high temperatures (>60 K) the material as a Curie Weiss paramagnet with the susceptibility off as l/T, indicative of a weakly interacting set magnetic (unpaired) electrons. A powder with = 1.57 showed a maximum in the susceptibility at th the susceptibility falling to zero as the ture went to zero. As the lithium concentration was , the temperature at which the susceptibility a maximum increased. The general behavior was ent with a material which has a lattice composed of packed Li+021l cations with trapped electron holes, dependence of the magnetic behavior on composition . and is still not, well understood [75]. A 3 explanation for the spin-pairing is that the undergoes a phase transition which results in [tion so that electrons in two adjacent cavities are .red. Since excess lithium tends to decrease the .ring transition temperatures it is possible that .tial lithium may affect the coupling of the two 44 The nature of the spin-pairing and the role that ithium plays in the physical properties of this are still unclear at this point. on K+0222-e_ indicated a different type of [74]. This salt, with a plasma optical absorption e of electron delocalization was very temperature and therefore difficult to handle. The microwave ity of this material was high, similar to that of ic sample. The magnetic susceptibility did not he Curie Weiss law, but could be attributed to ramagnetism. Again, all of the results correlated idel of an expanded metal electride, perhaps in nature to the metal ammonia compounds. :akthrough came in 1983 when improved synthetic as [76] were used toisolate a crystalline Cs+(1806)2-e— {77]. This material was ized by a localized optical absorption peak at . The magnetic susceptibility followed the Curie ; and the BPR spectrum gave a narrow asymmetric = 2.0023 indicative of a good microwave conductor e MAS 133NMR spectrum showed a one line spectrum, paramagnetically shifted by 140 ppm from typical 2 chemical shifts. Again all of the results agree picture of a salt which has a trapped electron as 45 Objectives of This Work h effort-has been devoted to defining the properties trons trapped in cavities. The electron trapped at alline vacancy (the F—center) has been quite hly explained, in part because the structure of the l defines the structure of the cavity. In dipolar I, solid or solution, there is inevitably some I which hampers the full characterization of the l electron. The low concentrations of electrons in ited glasses and in metal solutions in non-polar .s preclude techniques such as NMR from probing the nteractions of the electron. The average “ical structure of the electron was determined in ; however, these are electrons which are at very low rations and are not able to interact freely with one . The complexity of ionic associations in metal solutions has limited experimenters’ abilities to ully the structure of the species responsible for ical properties of the solution. class of compounds known as electrides has s in high concentrations in the crystalline state, ve and below the electron density normally required llic character. The possibility of defining the 46 re of the electron trap, as well as measuring the nteractions of the trapped electron with other ns and with cations in the lattice was the impetus s thesis research. Each new synthesis of another de material has generated a new set of properties, ndicates that these crystalline compounds do have a range of fundamental interactions. objectives of this thesis may be summarized as: 1) 1 characterization of the electride 0s+(1806)2-e-, ad been synthesized and partially characterized 0 this work [71]; and 2) the synthesis and charac- ion of a new crystalline electride, 0s+(1505)2-e’. er compounds were studied to model the complexed or to measure the effect of doping an electride diamagnetic anion; however, the focus of this work e properties of the electrides. Several techniques d to characterize electrides. jor goal of the electride group has been to obtain 1 structure of an electride salt. In much the same the proof of the stoichiometry of the first sodide 1y established by determination of the structure of Na-, the structure of an electride is required to he structural implications of their properties. rts to determine the structure of the electrides is d in Chapter III. de st nu wi At a] 47 ic Angle Sample Spinning NMR has been used to as the valence state of alkali ions in the solid it is also an excellent probe of the electron- interaction by means of the Knight shift. Chapter ibes 13303 NMR studies of complexed cations, first amagnetic anions and then with trapped electrons. s to dope the electride with alkali metal anions are ?scribed. sensitivity of the paramagnetic trapped electrons hboring trapped electrons can be measured in certain y magnetic susceptibility methods. The previously erized electrides have shown great diversity in agnetic susceptibilities, ranging from apparent 3 Pauli paramagnetism to diamagnetic complete spin (0s+0222-e—). The susceptibility results of the ject compounds are carefully compared and contrasted er VI. implications of these results and suggestions for ork are presented in Chapter VII. II Su CC ti I] CHAPTER II EXPERIMENTAL DETAILS Synthetic and General Handling Techniques sful syntheses required high purity metals and xants and dried solvents free of reducible impuri- Alkali Metals dium metal (99.95:) was obtained from Alpha Ventron ts in 5 gram argon filled glass ampoules. Cesium (99.95%) was donated by Dow Chemical Company, and was d in vacuum sealed Pyrex ampoules. Redistribution K amounts of the metals into small diameter (2-5 mm) was done under high vacuum (~10_5 torr) as described | [79]. Quantified amounts of the metals were d by sealing off a measured length of metal in a ube of known diameter. Complexants rown-5 (15C5, 1,4,7,10,l3-pentaoxa-cyclopenta- and lB-Crown-S (1806, 1,4,7,10,13,16,hexaoxa-cyclo- 0C Ch BC 49 :ane) were obtained from P.C.R. Inc. or Parrish ll Co. The solid 1806 was recrystallized from .trile. Both complexants were vacuum distilled to impurities [72]. Solvents ethyl ether (Me20, Matheson, Anhydrous), methylamine theson, Anhydrous), trimethylamine (TMA, Matheson, us) and diethyl ether (DEE, E.M. Science, 99.953, us) were purified by first distilling the solvent scuum over calcium hydride to remove water. The was transferred repeatedly into bottles containing 3 or NaK3 alloy, with benzophenone for Me 0, THA, 2 until the blue solution, which is indicative of the e of solvated electrons or free radicals, was The solvents were repeatedly frozen in liquid l and pumped to 10-5 torr to remove soluble gases. IA, and MA were transferred to clean stainless steel »r storage; DEE was stored over NaK3 in glass [71]. II. for clt r11 001 119 so ev ha 8! 5O 4 Synthesis of Alkalides and Electrides he synthesis of alkalides and electrides is straight- rd but rigorous. Glassware (Pyrex or quartz) is ed by an HF/detergent cleaner followed by a thorough ng and then is soaked in aqua regia overnight. The usly rinsed glassware is oven dried at 180°C prior to ll handling of alkalides and electrides and their Ions is done at temperatures below -20’C after sting the vessel to «'10—5 torr. The synthetic methods )een described in detail elsewhere [71,76]. The I1 procedure is first to load an appropriate apparatus ;he proper molar ratios of the metal(s) and complexant .ly 1—2 millimoles of the complexant). The metal is .led under vacuum to form a film on the wall of the .. An appropriate solvent is added (MeZO or MA) to .ve the complexant and then to dissolve the metal. the materials are dissolved, the solution is concen— t to saturation by removing the volatile solvent. .ls are formed upon addition of a less polar solvent 'r DEE) and cooling; further removal of solvent allows sired compound to completely fall out of solution. crystals are washed with TMA or D88, and vacuum dried being sealed off into storage fingers. The samples 51 a stored at -75'C indefinitely in the case of ace)2-Ne' and Cs+(1806)2-e— and for weeks in the case *(1505)2-e'. ll compounds have been analyzed for metal and crown content. The analysis scheme consists of 5 over- ig experiments. The reducing power of a known mass of Lal is measured by slowly decomposing the alkalide or ‘ide with water to generate Hz(g). The net reaction I alkalide is I_ + 2H20 ---> M+ + N+ + 20 + 20H“ + 32(8) (2.1) »r an electride ‘ + H20 -—-> M+ + 20 + on’ + 1/2H2(g) . (2.2) drogen gas was collected to quantify the number of. of water that had been reduced. The hydroxide that rmed was titrated for the total number of moles of Individual metal quantities are found by flame on spectroscopy of a solution of the residue relative st of standards of known concentration. Crown ethers antified by 1H NMR. The unknown crown is dissolved with a known amount of sodium acetate. The NMR peak ities are measured for relative proton content. Full 3 of this analysis scheme are given by Van Bck et a1. [54] Go ( Cs+( est! flu x, 1 f on 11.] MS cal or Rac be: ale to Va: 111‘ he ta‘ 01 ex 52 The results of the analysis for two samples of :5)2-e- are given in Table 2. The stoichiometry of :6)2-e‘ and Cs+(1806)2-Na_ had been previously .shed by Ellaboudy [71]. Also tabulated are the rmission results and the per cent sodide character, three mixed alkalide-electride salts of general e- where 0 it: 0 w . MLQIO Mb o MLQI Amvmm o . D no.0 b mm.c HN.H mm.~ Obo.o mmo.o mmo.c mNc.o mmd _ mom m.xmscuxma nos: eons all cases nuuonssminoz Has ll: 0 O : «Amavu\mas in: O O E s: a.-. as: ..-.. kHHGOAQOhaomH cusses: can nuances canon: consensus nonssomtannog xfiuasz dash assaulmhsn Hahsuosmsnomn ""Il'l||-" m.Aoaano has: moon N «a . q I'll-I'll no.cxccmae+uo -nz.mamomse+nc l'll‘"i'III‘I"'II-"'IIII I. I ‘l'--"ll"'|"l-l|' | I'l'- I- "5||'|"I'll"‘-|ll"|ll||||||I|ll||l|'l""|"|"|-'l|| owszamom sousmloo mussvusm musseloo as: anemones: sense as suns can: shame umvmhsa .oosomhe>:oo games: was: noflas>nonno as we mom a 3 m censuses snouuasnsm povsaonu anewuooamom codenamed: n=o~0Ios< assume: noussvmlunooa newnessm sewuswwlfimwz cacao mousse»: assesses .Iu.mxmomsc+no sen .nz.uxmoa~v+no ho umumusfiwomxmeIWOMasHom ousaoshum mom masses: Hoasulfimomxm me wanna .m ofinss 86 ABLE 6. Table of Positional Parameters in Fractional Unit Cell Coordinates and Their Estimated Standard Deviations for Cesium (18-Crown-6)2 Sodide at 213 K Atom x y z B(A2) ‘C81 0 000 0.02094(3) 0.250 3.238(7) Nal 0.000 0.500 0.000 8.5(1) 01 -0.0980(2) 0.1741(2) 0.3653(2) 4.42(7) O4 0.1135(2) 0 1611(2) 0.3643(2) 4.48(8) O7 0.1361(2) -0 0171(2) 0.4081(2) 4.29(7) 010 -0.0224(2) -0 1364(2) 0.3877(2) 4.48(7) 013 -0.1819(3) -0 1166(2) 0.2736(2) 5.72(9) 016 -0.2334(2) 0 0605(2) 0.2861(2) 5.23(8) C2 -0.0318(4) 0.2434(3) 0.3716(3) 5.1(1) C3 0.0614(4) 0.2169(4) 0.4118(3) 5.2(1) C5 0.1979(4) 0.1252(4) 0.4032(3) 5.1(1) C6 0.1766(4) 0.0506(4) 0.4539(3) 5.0(1) C8 0.1279(4) -0.0932(4) 0.4518(3) 5.1(1) C9 0.0765(4) -O.1599(4) 0.4054(3) 5.4(1) C11 -0.0823(5) -0.2028(3) 0.3564(3) 6.2(1) C12 -0.l829(4) -0.1706(4) 0.3369(3) 6.6(1) C14 -0.2753(4) -0.0765(5) 0.2481(4) 7.9(2) C153 -0.306(1) 0.004(1) 0.275(1) 10.4(4)* C15b -0.301l(7) -0.0103(6) 0.2987(5) 4.2(2)* C17 -0.2616(4) 0.1299(4) 0.3319(3) 5.8(1) C18 -0.1878(4) 0.1981(4) 0.3285(3) 5.6(1) Continued 87 TABLE 6. (CONTINUED) Table of Positional Parameters in Fractional Unit Cell Coordinates and Their EStimated Standard Deviations for Cesium (18—Crown-6)2 Sodide at 213 K Atom x y z 3(A2) H28 -0.060(3) 0.296(3) 0.397(3) 6(1)* H2b -0.024(4) 0.264(3) 0.327(3) 7(1)* H38 0.054(3) 0.186(3) 0.450(3) 6(1)* H3b 0.107(4) 0.268(3) 0.426(3) 7(1)* H58 0.230(4) 0.176(3) 0.435(3) 6(1)* H5b 0.238(3) 0.102(3) 0.368(2) 5(1)* H68 0.130(4) 0.068(3) 0.487(3) 8(1)* H6b 0.241(3) 0.028(3) 0.478(3) 6(1)* H88 0.093(3) -0.080(3) 0.493(3) 5(1)* H8b 0.198(4) -0.111(3) 0.468(3) 7(1)* H98 0.079(3) -0.215(3) 0.426(2) 5(1)* H9b 0.105(3) -0.166(3) 0.364(2) 4(1)* H118 -0.082(3) -0.250(3) 0.386(2) 5(1)* Hllb -0.062(4) -0.223(3) 0.319(3) 6(1)* H128 -0.211(4) -0.134(4) 0.368(3) 8(2)* H12b -0.226(4) -0.217(4) 0.327(3) 8(2)* H148 -0.326 -0.117 0.252 8.86 H14b -0.271 -0.061 0.196 8.86 H1588 -0.354 0.028 0.239 11.45 H158b -0.334 -0.004 0.323 11.45 H15b8 -0.368 0.006 0.289 5.15 H15bb —O.292 -0.030 0.350 5.15 H178 -0.267(4) 0.111(3) 0.384(3) 7(1)* H17b -0.330(4) 0.148(3) 0.310(3) 7(1)* H188 -0.216(4) 0.249(3) 0.351(3) 6(1)* H18b -0.180(4) 0.215(3) 0.278(3) 6(1)* Starred atoms were refined isotropically. \nisotropically refined atoms are given in the form of the isotropic equivalent thermal parameter defined as: <4/3) * [a2*B<1,1) + ab(cos gamma)*B(1 + bc(cos a1pha)*B(2 + b2*B(2,2) + c2*B(3,3) , + ac(cos beta)*B(1,3) I 2) 3)] Table 7. 88 Table of General Displacement Parameter Expressions U’s in (A2) for Cesium (18-Crovn-6)2 Sodide _——_-——-——---——‘_-——--—-—-——----—---—_—-—-——-—-——————-—-— ame U(1.1) U(3.3) U(1,2) U(1,3) U(2,3) sl 0.0453(2) 0.0448(2) 0.0331(2) 0 0.0045(2) 0 i1 0.077(2) 0.118(3) 0.129(3) -0.003(2) 0.023(2) 0.024(2) 1 0.064(2) 0.058(2) 0.046(2) 0.006(2) 0.003(2) 0.003(2) 4 0.068(2) 0.064(2) 0.040(2) -o.009(2) 0.012(2) -0.005(2) 7 0.054(2) 0.073(2) 0.036(1) -0.008(2) 0.005(1) 0.014(2) [0 0.069(2) 0.050(2) 0.053(2) -0.009(2) 0.015(2) -0.003(2) 13 0.075(2) 0.070(2) 0.074(2) —0.013(2) 0.012(2) -0.005(2) [6 0.051(2) 0.068(2) 0.080(2) -o.009(2) 0.005(2) —0.002(2) 2 0.099(4) 0.042(3) 0.054(3) -o.003(3) 0.017(3) -0.004(2) 3 0.085(4) 0.065(3) 0.049(3) -0.016(3) 0.016(3) -o.011(3) 5 0.059(3) 0.081(4) 0.053(3) —0.016(3) 0.011(2) -0.006(3) 5 0.051(3) 0.096(4) 0.043(3) —0.010(3) —0.000(2) 0.004(3) 3 0.060(3) 0.084(4) 0.051(3) 0.005(3) 0.008(2) 0.026(3) 9 0.086(3) 0.062(3) 0.059(3) 0.013(3) 0.025(3) 0.023(3) ll 0.144(5) 0.051(3) 0.040(3) —0.029(3) 0.016(3) -0.003(2) 12 0.090(4) 0.107(4) 0.055(3) —0.050(3) 0.017(3) -0.028(3) 13 0.062(4) 0.122(5) 0.117(5) —0.027(4) 0.004(4) -0.034(4) [7 0.063(3) 0.100(4) 0.058(3) 0.026(3) 0.014(3) —0.013(3) 18 0.082(3) 0.071(3) 0.058(3) 0.031(3) 0.001(3) -0.007(3) EEEJSE'EQIQE$335223};gleamt parameter 2 2 exp[-2n2{h2a20(1,1) + k2b20(2,2) + 1 c U(3,3) 4 2hkabU(1,2) + 2hlacU(1,3) + 2klch(2,3)}] where a, b, and c are reciprocal lattice constants. 89 The ionic packing in os+(18cs)2-Na“ (and os+(1806)2-e") is schematically drawn in Figure 8. Only the cations, 03+, with solid circles, and the anions, Na- (or e-), with open circles, are included, and distances are not to scale. Nearest anion-anion distances are represented by dotted lines; and nearest cation-cation distances are represented by dashed lines. Each ion is surrounded by eight counter ions; this is represented (for only one cation) by fine solid lines. Anions and cations lie in alternating planes perpendicular to the c-axis, anions at z = 0, 1/2, etc. and cations at z = 1/4, 3/5, etc. Anions at each level in z- are at fractional coordinates g = 0, 1 and y = 1/2; and x = 1/2 and y = 0, l. The cations at z = 1/4 are at fractional 0.5 coordinates: x = 0, 1 and y = 0.0209, 1.0209; and g and 1 = 0.5209. The cations at z = 3/4 are at fractional coordinates: x = 0, l and 1 = —0.0209, 0.9791; and x = 0.5 and 1 = 0.4791. The cations are alternately distorted by + and -0.33 A from 1 = 0, 1/2 and 1 positions. This is shown in Figure 8 by an exaggerated zig-zag stack for cations along the grdirection. The anions form linear stacks along the g-direction. The distance between Cs+ ions along the 5 axis is 8.74 A, which is 0.03 A longer than the separation of Na- ions along the g axis. In planes perpendicular to the 9 axis both cations and anions are separated by 10.37 A from 90 .sofipmo one scene moocmsmfic soflcm :ofiumo pcomosmma mesefi cflHOm xmoz one was .meocmumac sowcmnco«:m pcomoemos mocflH eocmmp .moosmumfie soHumUIschmo psemosmes mocflfi coupon .-mz.mmouwav+mu ca .mofioawo some so copcomosmos .mcoflcm ESAeOm on» was .moaosfio efiaom an voucomosmea .m:0flumo Esflmoo one we Emammflm mcfixoma .w osswfim c a c». _ 8d; .9 o 8.12 4 3.0.0. , \nw/hxmv ... a \I 1 \O‘ 1. .0.0.N\: ““4. . . .. ................... fl. .3. .880. : i \u\'..2mm.... can demos... : OI+CZ 'l D) 91 four nearest like-ions. Each Na- is coordinated by eight complexed cations. The distance of each inversion related pair of cations from Na_ is different than the others because of the distortion of Cs+ ions from linear packing along the g axis, and because the Cs+-Na_ coordination involves a component in the a direction in the 3-9 plane so that the monoclinic angle, 8 = 93.16. affects the distances. The four unique distances to neighboring Cs+ ions from an Na_ ion are: 7.87 A, 8.28 A, 8.69 A and 9.26 A. The two shorter. distances are coordinations to Cs+ ions that lie just off the peg plane that contains the central Na—. The two longer distances are coordinations to four Cs+ ions that lie in the same g-g plane as the Na-. Thus the coordina- tion of Na- by cations has only six of the eight first shell anion—cation distances which are shorter than the shortest anion—anion distance. The crown ether molecules complex the Cs+ ion by lssuming a conformation with the oxygens directed toward :he cation. The outer surface of the complexed cation :onsists almost entirely of the CH2 groups of the crown ather. Only one of the two crown ethers is unique; the lecond is generated by two-fold symmetry about the b axis :hrough Cs+. A least squares plane was fit to the posi- :ions of the six oxygens on one crown ether; the results of 92 :his fit are summarized in Table 8. The best fit plane can is described by the equation (in A units) as 3.284; - 0.1251 - 16.4042 + 5.956 = 0 'he plane is very nearly parallel to the b axis, and forms in angle when projected onto an gag plane of ~25' from the L axis toward the 9 axis. The packing of the crown ethers n the lattice is such that the hole of the crown ether is 'illed by electron density of carbons and hydrogens of the earest crown ethers associated with adjacent cations along he g axis. The structure of one Cs+(1806)2 ion is shown in Figure , The distances between the cation and the six crown ther oxygens, as well as bond distances in the crown ether olecule are summarized in Table 9. The mean Cs-O nteraction distance is 3.357 A, with individual contacts anging from 3.29 A to 3.47 A. This mean distance is omewhat longer than that observed in a 1:1 complex of s+(18C6)-SCN* [51], and is similar to that seen in a Cs+ dibenzo 18C6) sandwich structure [87] which had Cs-O istances ranging from 3.17 A to 3.43 A. The sum of the onic radius of Cs+ and the van der Waals radius of oxygen s 3.2 A. The long distance for the Cs-O interaction in he Cs+(18C6)2 sandwich indicates that the size of the Table 8. 93 Table of Least-Squares Planes for Cesium (18-Crown-6)2 Sodide Crystallographic Equation of Plane in Angstroms -0.1249 Y + —16.4041 Z - —5.9555 a 0 0.0117 0.0252 0.0018 Y(A) 2(8) Distance Esd (Angstroms) — ———— Atoms in Plane -—-- 2.7290 6.3567 -0.576 +- 0.003 2.5243 6.3404 0.558 +- 0.003 —0.2704 7.1018 -0.018 +- 0.003 —2.1416 6.7469 -0.505 +- 0.003 -1.8307 4.7612 0.521 +- 0.004 0.9478 4.9796 0.021 +- 0.004 ————— Other Atoms -—-—— 0.3271 4.3506 1.852 +- 0.000 7.8420 0.0000 5.893 +- 0.000 3.8152 6.4662 -0.338 +- 0.005 3.3997 7.1667 -0.503 +- 0.005' 1.9616 7.0175 0.371 +— 0.005 0.7908 7.8997 -0.564 +- 0.005 -1.4645 7.8619 -0.768 +- 0.005 -2.5101 7.0552 —0.271 +- 0.005 -3.1821 6.2021 —0.300 +— 0.005 -2.6769 5.8628 -0.516 +— 0.005 —1.2017 4.3175 0.440 +— 0.007 0.0691 4.7793 -0.167 +— 0.017 -0.1635 5.1987 —0.535 +— 0.009 2.0358 5.7766 -0.888 +— 0.005 3.1053 5.7163 -0.450 +- 0.005 5.2844 X + 0.0090 Atom X(A) 01 -1.6807 04 l 1926 O7 1 4561 010 -0 6771 013 —2.7348 016 -3 4443 C51 -0.2399 Nal 0.0045 C2 -0.7869 C3 0.4399 C5 2.3013 C6 1.9631 C8 1.3027 C9 0.6482 C11 —1.4612 C12 -2.8092 C14 -3.9780 ClSa -4.4l90 C15b —4.3764 C17 -3.8706 C18 -2.8641 95 TABLE 9 . Table of Bond Distances (in Angstroms) for Cesium (18—Crown-6)2 Sodide at 213 K and Cesium (18-Crown-6)2 Electride at 216 K Sodide Electride Atoml Atom2 Distances Distances C51 01 3.445(3) 3.453(8) C51 04 3.292(3) 3.296(8) C51 07 3.287(3) 3.305(7) C51 010 3.468(3) 3.457(8) C31 013 3.324(4) 3.296(9) C51 016 3.324(3) 3.304(8) 01 C2 1.411(6) 1.42(2) 01 C18 1.397(6) l.44(2) 04 C3 1.420(6) 1.42(2) 04 C5 1.416(6) 1.42(2) 07 C6 1.421(6) l.44(1) 07 C8 1.424(6) 1.39(2) 010 C9 1.410(6) 1.41(2) 010 C11 1.412(7) 1.43(2) 013 C12 1.391(7) 1.39(2) 013 C14 1.462(7) 1.43(2) 016 C15a 1.33(2) 1.39(2) 016 C15b 1.467(10 016 C17 1.414(7) 1.42(2) C2 C3 1.472(8) 1.47(2) C5 C6 1.505(8) 1.47(2) C8 C9 1.474(7) 1.52(2) C11 C12 1.479(9) 1.50(3) C14 C15a 1.42(2) 1.51(2) C14 C15b 1.419(12) C15a C15b 0.48(2) C17 C18 1.470(8) 1 47(2) Numbers in parentheses are estimated standard deviations in the least significant digits. 96 cationic hole in the complex is not the limiting factor in how closely the two crown ethers may approach each other. The two crown ethers make van der Waals contact with each other before they are fully in contact with the cation. The Cs+ ion is in a highly enclosed environment; there is very little exposure to species beyond the two complexant molecules. The disorder on the crown ether is not a surprising result considering the many conformations that are avail- able to the flexible 18 member ring. A summary of selected torsion angles around the backbone of the crown ether is iven in Table lg. This conformation has been observed previously in crystalline salts and has no particular internal symmetry [88]. A theoretical consideration of the conformation energy suggests that this structure is not a particularly low energy conformation of the crown ether, but even major conformational changes correspond to only about a 5 Kcal/mole change in the energy [88]. This :onformation in Cs+(18C6)2-Na~ is probably stabilized by improved van der Waals attraction of 03+ to the ether :xygens, and perhaps by improved lattice packing. The anion occupies sites in the lattice bounded by right complexed cations. The cavity in which the sodide on resides is lined with protons of the crown ethers. The onic radius of Na” may be calculated on the assumption 97 .Table 10. Table of Selected Torsional Angles (in Degrees) for Cesium (18-Crown-6)2 Sodide at 213 K and for Cesium (18-Crown—6)2 Electride at 216 K ----.—m--——---_------———-------“--—-—--—-------. Sodide Electride Atom 1 Atom 2 Atom 3 Atom 4 Angle Angle C18 01 C2 C3 178.9 -178.15 C2 01 C18 C17 .-171.6 -170.30 C5 04 C3 C2 -172.8 -175.94 C3 04 C5 C6 80.5 75.59 C8 07 C6 C5 171.6 172.90 C6 07 C8 C9 174.2 178.24 C11 010 C9 C8 -168.4 -170.70 C9 010 C11 C12 -177.6 —178.61 C14 013 C12 C11 -177.9 179.96 C12 013 C14 C15a 91.5 78.18 C12 013 C14 C15b 74.0 C17 016 C15a C14 -163.4 175.45 C17 016 C15b C14 176.0 C15a 016 C17 C18 -167.5 179.65 C15b 016 C17 C18 172.7 01 C2 C3 04 70.0 72.84 04 C5 C6 07 63.1 63.25 07 C8 C9 010 -65.9 —66.20 010 C11 C12 013 70.9 73.46 013 C14 C15a 016 36.5 62.37 013 C14 C15b 016 72.2 016 C17 C18 01 —68.3 -67.19 98 that the anion is in van der Waals contact with the nearest hydrogens. Table ;l_1ists the nearest atomic neighbors to 23 Na . Na- can be assumed to be a spherical anion. The Na NMR chemical shift of the anion in this and other sodide salts shows no Ramsey shift from the gaseous state anion [63] which indicates that the anion has 3s2 configuration. The van der Waals radius of aliphatic hydrogens has been estimated to be 1.2 A [89], so the Na- radius can be calcu— lated to be between 2.47 and 2.65 A. The smaller radius corresponds to the contact with the nearest protons, the larger estimate is from the average contact distance over the six nearest protons. The latter figure is probably a better value because the closest hydrogen is likely to be quite acidic, and thus have a smaller effective radius. This range for the ionic radius of Na_ is in good agreement with the results from the structure determination of Na+0222-Na— [60]. Issa’s attempts to isolate crystals of Cs+(1806)-Na- for structure determination yielded lattice constants but the structure could not be solved. The unit cell para— meters for cells containing 4 formula units deduced by Issa [79] were 3 = 13.895(ll), b = 15.501(24) and g = 8.939(7) with a monoclinic angle 8 = 93.25“ and cell volume = 1920.7 A3. NMR methods later revealed that the compound was in fact Cs+(1806)2-Na‘. The unit cell parameters are 99 TABLE 11. Table of closest anionic contact distances for Cesium (18- Crown- 6) Sodide at 213 K, for Cesium (18- Crown- 6) Elgctride at 216 K, and for Cesium (15— Crowa- 5)2 Iodide at 291 K Contact Pair Van der Waals Anion Atom 1 Atom 2 Contact Distance Radius Na_ H6b 3.67 A 2.47 A Na_ H2a 3.68 A 2.48 A Na_ H3b 4.208 A . 3.01 A Na HSb 4.509 A 3.30 A Mean Anionic Radius From 6 Nearest Proton Contacts: rNa-- 2.65 A Cesium (lB-Crown—6)2 Electride e: 36a 3.292 A 2.09 A e_ H15b 3.635 A 2.24 A e_ H2a 3.762 A 2.56 A e H17a 4.085 A 2.88 A Mean Anionic Radius from 6 nearest proton contacts: re—= 2.30 A Cesium (lS-Crown—S)2 Iodide For I1 1: H9 , 3.403 A 2.20 A I H3: 3.456 A 2.26 A 1‘ H11 3.615 A 2.42 A For 12 a 1‘36 3.139 A 2.14 A 1:315b 3.479 A 2.28 A IH11b 3.795 A 2.59 A Mean Anionic Radius from 12 Nearest Proton Contacts: {11" $.33 ‘ r -- . 100 very close to those obtained in this study if the 9 axis length is doubled. III.B.3 Results of the Structure Determination of Cs+(18C6)2-e- at 216 K The refined positional parameters of all of the atoms in Cs+(1806)2-e_ with their estimated standard deviations and general isotropic thermal displacement parameters are given in Table 1;. The general anisotropic thermal displacement parameters of the non-hydrogen atoms are given in Table lg. The positions of the hydrogens were not refined independently, but were constrained to ride on the carbons of the crown ether with a fixed bond length of 0.95 A and tetrahedral bond angles. There was no indica— tion of disorder in the crown ether ring. The main features of the crystal structure of Cs+(18C6)2-e- are nearly identical to those of the Cs+(18C6)2-Na— structure; the ionic packing shown in Figure Q is essentially unchanged except for minor changes in distances. The stereographic ORTEP [90] packing diagram.is shown in Figure 19. The positions noted with the symbol 0 are the positions that were occupied by the anion in the sodide salt. In this case the final Fourier difference map indicated that only noise level electron density existed there. The noise level of the difference map was 0.059 101 Table 12. Table of Positional Parameters in Fractional Unit Cell Coordinates and Their Estimated Standard Deviations for Cesium (18-Crown—6)2 Electride at 216 K Atom x y z 8(A2) CSl 0.000 0.02650(9) 0.250 2.74(2) 01 -0.0982(7) 0 1783(5) 0.3693(5) 3.6(2) O4 0.1224(7) 0 1626(6) 0.3639(5) 3.8(2) O7 0.1425(6) -0 0137(6) 0.4073(4) 3.9(2) 010 -0.0235(6) -0 1289(5) 0.3879(5) 3.8(2) 013 -0.1870(7) -0.1097(6) 0.2728(5) 4.2(2) 016 -0.2393(6) 0.0642(5) 0.2937(5) 3.8(2) C2 -0.027(l) 0.2462(8) 0.3765(7) 4.3(3) C3 0.068(1) 0.2147(9) 0.4150(7) 4.5(4) C5 0.211(1) 0.1252(9) 0.3984(8) 4.2(3) C6 0.191(1) 0.0541(8) 0.4499(7) 3.9(3) C8 0.133(1) -0.0856(9) 0.4522(8) 4.1(3) C9 0.078(1) -0.1543(9) 0.4054(8) 4.6(3) C11 -0.083(1) -0.1967(8) 0.3547(8) 4.6(4) C12 -0.188(1) —0.1628(9) 0.3367(7) 4.5(4) C14 -0.283(1) -0.073(1) 0.2493(8) 4.9(4) C15 -0.311(1) 0.000(1) 0.2993(9) 6.4(5) C17 -0.268(1) 0.1366(9) 0.3356(8) 4.5(3) C18 -0.192(1) 0.2038(9) 0.3299(8) 5.2(4) Continued. 102 Table 12 (continued). Table of Positional Parameters in Fractional Unit Cell Coordinates and Their Estimated Standard Deviations for Cesium (l8-Crown-6)2 Electride at 216 K Atom x y z 3(A2) H2a -0.056 0.290 0.406 5.4* H2b -0.012 0.267 0.327 5.4* H3a 0.110 0.261 0.431 6.1* H3b 0.051 0.182 0 459 6 1* HSa 0.248 0.167 0 427 5.5* H5b 0.253 0.105 0 359 5.5* H6a 0.253 0.034 0 474 5 0* H6b 0.146 0.073. 0 488 5 0* H8a 0.094 —0.073 0 496 5 4* H8b 0.199 —0.105 0 469 5 4* H9a 0.113 ‘ —0.162 0 359 5.9* H9b 0.078 —0.206 0 433 5.9* Hlla -0.084 -0.242 0.390 5.9* Hllb —0.053 -0.215 0.309 5.9* H12a -0.233 —0.209 0 325 6 0* H12b -0.211 -0.133 0.380 6.0* H14a —0.335 -0.115 0.251 6.3* Hl4b —0.279 -0.053 0.198 6.3* H15a —0.377 0.021 0.283 7.5* H15b —0.313 -0.018 0.351 7.5* H17a -0.274 0.122 0.388 6.0* H17b -0.333 0.156 0.316 6.0* H18a —0.180 0.214 0.277 6.8* H18b -0.216 0.254 0.352 6.8* Starred atoms were refined isotropically. Anisotropically refined atoms are given 1n the form of the isotropic equivalent displacement parameter defined as: (4/3) * [a2*B(1,l) + b2*B(2,2) + c2*B(3,3) + ab(cos gamma)*B(1,2) + ac(cos beta)*B(1,3) + bc(cos alpha)*B(2,3)] where a, b, and c are reciprocal lattice vectors. 103 Table 13. Table of General Displacement Parameter Expressions U's (in A2) for Cesium (18-Crown-6)2 Electride Name U(1,1) U(3,3) U(1,2) U(1,3) U(2,3) Csl 0.0331(5) 0.0441(6) 0.0274(5) 0 0.0065(4) 0 01 0.059(6) 0.040(5) 0.039(5) 0.004(5) 0.005(5) 0.004(4) 04 0.050(5) 0.065(6) 0.029(4) -0.000(5) 0.011(4) 0.001(5) 07 0.056(5) 0.064(6) 0.028(4) —0.005(5) —0.000(4) 0.021(4) 010 0.057(5) 0.044(5) 0.044(5) —0.004(5) 0.014(4) —0.002(4) 013 0.057(5) 0.069(6) 0.035(5) -0.010(5) 0.013(4) 0.008(5) 016 0.046(5) 0.048(5) 0.051(5) 0.000(4) 0.019(4) —0.006(4) C2 0.09(1) 0.038(8) 0.039(7) 0.015(8) 0.003(8) —0.019(6) 03 0.066(9) 0.07(1) 0.035(7) —0.008(8) 0.009(7) -0.004(7) 05 0.060(8) 0.056(9) 0.046(8) -0.013(8) 0.017(7) —0.005(7) C6 0.038(7) 0.07(1) 0.042(7) —0.012(7) —0.003(6) -0.009(7) C8 0.041(8) 0.058(9) 0.057(8) 0.009(7) 0.003(7) 0.015(8) C9 0.050(8) 0.054(9) 0.071(9) 0.027(7) 0.026(7) 0.021(8) C11 0.10(1) 0.034(8) 0.046(8) —0.023(8) 0.012(8) -0.002(7) C12 0.08(1) 0.054(9) 0.039(8) -0.024(8) 0.012(8) -0.005(7) C14 0.052(9) 0.08(1) 0 056(9) -0.025(8) 0.002(8) -0.014(8) 015 0.059(9) 0.11(2) 0.08(1) 0.002(9) 0.020(8) -0.029(9) C17 0.046(8) 0.064(9) 0 062(9) 0.009(8) 0.004(7) —0.024(7) 018 0.060(9) 0.07(1) 0.063(9) 0.038(8) -0.016(8) 0.000(8) The form of the anisotropic displacement parameter is: 2 2 exp[-2n2{h2a2U(1,1) + k2b20(2,2) + 1 c 0(3,3) + 2hkabU(1,2) + 2hlacU(1,3) + 2klch(2,3)}] reciprocal lattice constants. where a, b, and c are 104 A stereoview of a unit cell in Cs (18C6)2°e‘. Figure 10. The center of the anionic cavity is noted by the symbol 0. 105 electrons/A3; a hydrogen atom with a van der Waals radius of 1.2 A would have an average electron density of 0.14 e/A3 were it localized in the cavity. An electron trapped at this site would not be expected to appear on the final difference Fourier map. While the gross features of the sodide and electride structures are similar, several different details are notable. Table 14 summarizes the lattice constants and interionic distances of the two compounds. The volume of a unit cell of the electride is 114 A3 smaller than that of the sodide, but the cell length in the brdirection is 0.15 A larger in the electride.. The distortion of Cs+ from linear stacks along the 9 direction is stronger in the electride than in the sodide. In the sodide, cations lying in planes parallel to the 3-9 plane at g = 1/4 lie +0.328 A from x = 0 and z = 1/2 positions; those cations lying in planes perpendicular to g = 3/4 are displaced by -0.328 A from y = 0 and 1 = 1/2. The electride has larger displace— ments, of + and -0.42 A from x = 0 and y = 1/2 respec— tively. This displacement does not affect the distances between like ions in planes perpendicular to g, Cs+-Cs+ distances and e--e- distances are only slightly shorter than those seen in the sodide. The interionic distance in these planes are 10.27 A in the electride (that is 0.1 A shorter than in the sodide). Along the g axis the inter— Table 14. 106 A Comparison of Lattice Constants and Interionic Distances in Cesium (18C6)2 Sodide, Cesium (18C6)2 Electride, and Cesium (15C5)2 Iodide Characteristic Cell Dimensions a (A)- b (A) c (A) 5 (°) v (A3) Z Anion-anion Distances (A) along z axis in an x-y plane Cation—cation Distances (A) Along z axis In an x-y plane Anion-cation Distances (A) Cesium-oxygen Distances (A) Range Mean distance Cs+(18C6)2'Na_ Cs+(1806)2'e 13.581 15.684 17.429 93.16 3706.8 Two at 8.71 Four at 10.37 Two at 8.74 Four at 10.37 Two each at 7.87, 8.28 8.69, 9.26 3.287— 3.468 3.357 13.075 15.840 17.359 92.30 3592.2 Two at 8.68 Four at 10.27 Two at 8.72 Four at 10.27 Two each at 7.71, 8.00 8.67, 9.40 3.297- 3.457 3.352 Cs*(1505)2'1‘ 13.172 16.645 2888 8.04 and 8.60 Four at 9.32 Two at 8.32 Four at 9.32 Four at 7.73 Four at 7.86 2.960- 3.219 3.108 107 ionic distance is 8.68 A for anions and for cations 8.72 A (only 0.03-0.02 A shorter than equivalent distances in the sodide). The spread of distances from an anion to its first coordination shell cations is greater than that of the sodide. The electride has Cs+ ions both 0.15 A closer than the closest Cs+ in the sodide, and 0.15 A longer than the farthest first shell Cs+ in the sodide. The structure of the complexed cation is strikingly similar to that of the complexed cation in Cs+(1806)2-Na—. Table g includes the cesium-oxygen distances and bond lengths in the crown ether molecules for the electride. Cs+(1806)2-e- has Cs-O distances that range from 3.296 A to 3.457 A with a mean distance of 3.352 A. The sodide, for comparison, has a mean Cs-O interaction distance of 3.357 A. The crown ether has essentially the same conformation in both structures; the bond distances are generally more nearly ideal and there is no sign of disorder in the crown ethers in the electride structure. The torsion angles are listed with those of the crown ether of the sodide in Table 19, any differences are small. A plane was fit by least-squares methods to the oxygens of one crown ether and has the crystallographic equation, in Angstroms. 4.8285 + 0.06651 + 16.3765 = 5.944 A The results are listed in Table 15. The plane is nearly parallel to the x axis, and when projected onto the x—g plane the mean oxygen plane makes an angle of ~20' from the 5 axis toward the 5 axis. The Cs+ ion sits 1.852 A off of the mean plane, exactly as in the sodide. The environment of the trapped electron in the electride salt is of crucial importance to the interactions of the electron with other electrons, and with the cations in the crystal. The electron cavity, an interstitial void left by the packing of the complexed cations, is bounded essentially entirely by protons. If is is assumed that the hydrogens maintain a van der Waals radius of 1.2 A then the effective size of a "hard sphere" electron may be esti- mated. Table 11 includes the closest contacts to the inversion center at the center of the electron cavity. As in the sodide there is one pair of hydrogens that is much closer than the next nearest hydrogens. The estimated "hard-sphere" radius ranges from 2.09 A for contact with the nearest hydrogens to 2.30 A for the average contact with the nearest Cs+(1806)2-e- hydrogens. Intuitively it is clear that the trapped electron would not be hard sphere-like, but would be polarizable similar to the behavior of electrons in other trapped electron systems. The thermodynamically most stable trap geometry in the electride is not a spherical cavity created by a vacancy in 109 Table 15. Table of Least—Squares Planes for Cesium (18-Crown-6)2 Electride Crystallographic Equation of Plane 1 in Angstroms 4.8278 X + 0.0665 Y + —16.3764 2 - -5.9440 s 0 0.0216 0.0301 0.0305 0.0042 Atom X(A) Y(A) Z(A) Distance(A) Esd -—- Atoms in Plane -—-- 01 -1.5406 2.8235 6.4055 —0.566 +— 0.008 04 1.3469 2.5761 _6.3115 0.587 +- 0.008 07 1.5792 -0.2171 7.0649 —0.039 +- 0.007 010 -0.5771 —2.0422 6.7287 -0.531 +- 0.008 013 -2.6348 -1.7373 4.7318 0.566 +- 0.008 016 —3.3337 1.0170 5.0943 —0.017 +— 0.008 ————— Other Atoms ---—- Csl -0.1742 0.4198 4.3363 1.852 +- 0.000 C2 -0.6149 3.8993 6.5304 -0.336 +- 0.013 C3 0.6003 3.4014 7.1990 -0.510 +~ 0.013 C5 2.4872 1.9828 6.9111 0.448 +— 0.013 C6 2.1820 0.8575 7.8032 —0.498 +- 0.012 C8 1.4223 -1.3551 7.8431 —0.825 +- 0.013 C9 0.7432 —2.4440 7.0309 -0.326 +- 0.014 C11 -1.3260 —3.1153 6.1529 -0.277 +- 0.014 C12 -2.6936 -2.5781 5.8392 —0.488 +— 0.013 C14 -3.8738 -1.1561 4.3235 0.491 +- 0.014 C15 —4.2765 0.0064 5.1920 —0.460 +— 0.016 C17 —3.7427 2.1639 5.8217 —0.839 +- 0.014 C18 —2.7366 3.2276 5.7216 —0.370 +- 0.014 110 a cubic lattice, as in an F-center, or a self oriented co- ordination, as in glasses or solutions, but appears to be simply an interstitial lattice site in the closest-packed cationic system. The radius of the trapped electron cavity is slightly smaller than the radius of Na—. The effect of putting the larger Na_ anion into the cavity in the Cs+(18C6)2 system is to push the ions in 3:5 planes apart; there is little elongation of the distance between anions in the 9 direction in going from the electride to the sodide. The magnetic properties of these compounds, which will be presented in Chapters V and VI, are used to probe the interactions of the trapped electron with its ionic environment. To better understand the nature of the anionic cavity the structure of one cavity was studied on an Evans and Sutherland P8300 Graphics system. The surface contour representation of the van der Waals radii of all of the atoms of eight nearest complexed cations to a cavity was constructed on the screen. No thermal displacements of the atoms are accounted for in the van der Waals surface. It was possible to rotate the structure to any crystallo- graphic orientation and to cut into the map to get 3 dimen- sional cross sections of variable thickness of the ' molecules and the cavity. The cavity is football shaped, nearly circular in cross sections perpendicular to the g lll axis, and is elongated along the 9 direction. The packing of complexed cations, as shown by the van der Waals radii is very efficient, but does not completely fill the space between the electron cavities so that channels of low electron density connect each cavity to the six nearest cavities in the lattice. Electrons are most likely to interact with other electrons through these channels, rather than through space occupied by high electron density of the complexed cations. The channels are most constricted in the 3-9 plane that includes four of the six nearest cavities. A view of a cavity from the inversion center along the 9 axis is shown in Figure 11. The cavity is nearly square shaped at the inversion center with an edge length of ~4.5 A. Channels that connect the cavity to other cavities in the same 355 planes begin at the corners of the cavity and travel through the packed cations to the next electron traps. In the background of the central trap is an hourglass shaped constriction, which is the interface with the next cavity along the 9 direction. Cavities along this direction are separated by 8.68 A. The approximate lateral dimensions of the constriction are 5.0 A along the 6 axis by 2.2 A and 1.0 A at the widest and narrowest points in the 3 direction. The stacked structure of the ions along the 9 direction is shown in Figures 1; and 1;, which were constructed 112 ...:..‘c..'.s‘...‘ C ... . .§. .0" . . . . m s . “ANNEL To A ~-.....o... one-:2.- 1.. 4 A .. o. ‘ . ’ . .... .O.‘.. .‘1‘ s .g . . O .0 CAVITY AT 9‘: ° - ~' ° .. ' 0 L0 + b) 3...... 0......\ . . s 1 ... 2 ... ‘ .. ‘0. . . . . . .. .. at. '.-°\.o.:..-O'.. ...-'1". 9 .O.m~.... ‘0 z... 0.. . 0 0.... . o °.0.°'°.: ...... °°°.'..'o O '0. ‘ O o 0. O s" ‘ ‘ :’ v‘of‘lsfh‘ °'~"-‘.'.'°.v.1; ° .:° \ -..-,...~..-.. :....'.- .a. ..-. . . .. .. .0...::u‘... . :g‘n;s.' . ~ '. ‘ \ . m . . a . 0 s o'.. ..' . 1: . C .. . a .. . . . .... ... ‘ o... '0 . 0 . Q . . m .0 s O. . O . m 7... O 0.. ' s... . 0. .0 “.O :. .0 . U 0 ° - O s g o ...“ g . .... ‘7‘: 9 0....0 ... x. .0. ....‘tjr o ..ggo... . .0: . .0... ': 6... 9V . -'°‘ . 0. o C ‘0 0.. o .. 0‘. . . ‘ . 0 ...~.... . Q ..‘.. . . .:a‘....-.1......O..::O..’.‘ . ..1‘29'.. . .. . .‘ ‘o..‘. o '0 . .....- . .' ° 6. 9 ‘32-... ‘ I o "‘7‘. 0‘0 . O ’ . a n ~'.-.' - - °. ., '~ :1- .-' "9m" .~.:.~--:‘ -.:~°. . - . 2.". ... . 0",...0 .00 o “‘0. ... o...a~ ~ 9 o . a. .0 .0 ‘:‘... . 0..... . . ... . . . O . . . s. 0 ~ 0 O 0 Q 0 0 . O . ‘ 9 O. 0 . g 0 o a . o . O ' s . o o , 0.. 0. . g ' s .... .0 s O... ...... .o o ......... .. .... . .. .. '°' ' ‘ ...°0 ."8. '0". ..0. \‘ 0 .. . .0 o~ 0.. in I. g 0 o s m -'. ' ."'. 5, . '.5 .~ .... '0... o . .0...-: .. '.. ... ..‘ .0 . '. . 0.. . . .... 0. ....‘. ' .‘o. . 0. .‘ 0 °-u. o \ ... n. o ... ‘0 or. 3:80. . O s 0 . s o 'a'.-".-.. 0’... O ‘ 0......’ O.‘ 90"..9' 0. ..'o. . ...00......°.";.. ....O..‘ O.." ..‘ 9 '. . o ‘ . ‘ .0 .. . .o‘.. -. . .. . O . 9 . 0 02“: O ‘0 O . . 0 ... . .. .. . .... I. ..’ . . o . . .0 .. ... .I . . ' .‘ .. '5 s 2.... . . ‘g . . . ’ o . ... 0 .\\-. .0. °..°:":: . . ......23' . . .... 5'.- . . . "0..“ .. ‘ . ..., "o.... .",' 3 ‘..o~. o. ..;O.':s . . - . 00... U. 'o "."-,' ... .- -'9 :‘A '.,." . ' .'o-° ..'.::,' 0..-o” ..‘1.".’-."-v.g.". 3....- '. \. '. - .0 .. .... O O. ‘Q’. . m0. .. .' ‘ ’...o ,. . .. s .._... O... .‘q‘ ' o o '. "O . I 60“.. . ‘ . ‘ . ' 83$. 0 00“ ' . .0 ‘0 O O o‘ ' o ' .5-5 “a“. '. ' ." "‘ s-m ‘ '.' .9 .. ..... .. . .. . .0..'.. . .0 .... . ... ... . o O . . z. s so _ 0. m .. '0 “‘.m ... .. .... 9.. o . 0.. . . ‘ ‘2' {285%. O..:. .. ~.. .0. O o. . . . . ... . O . Q. ‘9‘... C . . . . 0.. O . ... U. . O: . 0 °. ' ..t.. .0 0 ‘0 § I ' 0. ‘ ..s.’:.f‘. '...' .' :°.~.‘;0.o‘"..‘ . .... .. o . O. . o ‘ ... O 0.. O . ..-. . ... .zu‘ . o \. ... ... . . . o . . 0.. . . . .0 .. .o s . o' , . - - 6,-0 .‘ \ '. b '0. . .0 ' . '- ‘m .. 0.. '0. ... 0. o \"-. o '. :o"° ". r- " . . .. . . ........ . m'. “o a. ‘ ~ ‘ O .‘o‘ ..‘I.. . a .. . 0. 0‘0. 0.. o .' .... s c 0' ' 0 a g It ' ' “:HV'-33:' a Figure 11 A surf ‘_'_‘-T-'—— ace Conto 1n CS (18C6 “T representation 0 The Cavity 1% :h6 as viewed down the c axis at : aU/anion:c cavity neighboring cavitjggn;:ti m?Ximum size in the a— —b p13; 3 O C :1/2 by arrows ’ '2.1/2, and o e Channels to '1/2,1/2 are indicated 113 ' ' 5° . 0 ' v ' . $5. - ... " 4 ' . on” ' ". . .30. . . . -. . . \ .... .‘,: ‘3‘}. . . '.. _ 1*}?2 .... . ... .. .... 1}.": I‘." a ’. 0- s. 0 . ~‘09‘u u - . '5 F. 3.. I s . : s . em. . 0 o 0 _ s .. .. . ... ' . . a . . . " J": *0". 0'... .:.. '0 .' . .‘9‘ .. -- '_ :0‘70‘ . : ' . ob. .... l~$$ 03.....‘0 . “s '. Joe... . 0 . .\ ~ .. . . . - ‘ ‘ I :3"; {"7 -.‘3 c“ :0: fl ”(.6 .334: fiz‘f... .- ° '~ 0‘: §§5“° ~ 17- '-* 3 ~14» ~03 . .~- .:‘u‘. \‘ ..~.- - ’ 35‘" w' ' .. 0' ... - , .0: ... := (3: 0"“. . ‘-.- ...: X‘s-'0‘ ‘ y ._ .... . .....- M. .0 st .. ‘0. a. ' . u ‘4: at}. 1'. 23‘. " o a ’o t 4% 7‘ ‘- "‘33-...“- ".4 \ .:-.= m . s . v . '.. . . - - ‘Q “a. .‘ . :9:- . °-~.-...'.-.- h‘q‘t an 9.8.". ‘K .. 6‘- ' ~~ -° ' ‘~‘ . I.“ ° . 9'. . ..~ '&’.".'.“ 09" ' ' ‘ .'\o.’l . '0“ :7”. a3. ' u‘ ' ' . s. ...},3 .0 . ..‘ . I _ '9‘ ' ‘Q o ' . g . . . . -. ‘Q t..." .0 . . § 0 . .h'..r . at. 0.0 s ...... . o :00. 11;; i 2 .1 (0 '3 'i ‘1' Ar h :4 .0 ~10 - nu $3.53. '9‘ .. ,‘ '- 24w»: 5 .... é a ‘ O _ .:.: -. o {‘v.. ".- ~ ° . .. .:.. . ~1. . ~'~“§, 5% " . -‘ .9. .- ‘ ..=':-‘ ...-... ‘ - ~ .-.-.-.- .- . ~00 , .0» . . '.... on, . . fi ‘0‘” o O o. ‘ b o o - 9‘ ‘ ‘ s ‘ ’ ‘ “0 t.'. 2’. .‘ ' I. ... n O c I 1 . . x z." . . or. . ..c o; .5: ._- .. " ' . 31333375" "‘ ' . ~§‘-‘-" . . m. 0 . "- .... 0 o m. '..k .:. ¢ '. m ' - ‘ O . o .5. . .623. IRE“ ; :no . .47.‘t ...,. -. ‘1 - ~ ' ‘lh " ~v-‘ * E ‘n . 'h 9 ‘ ' 'u . - o s - .. .0 ' 00s ‘ s. .3. '. .1“ \”.-.-‘ o . . I J’. x “0' on. ' ‘ . ' ‘06.. Q. :‘ .:.. °. ‘ ' ‘0“ 423.]. .00.... . --§-. . ‘ \ Q. .0 s :s‘ ': a... 1" , so... ‘z. a. ' ° '3 J og'.‘ 33.“ '7 0 on. '3. .m‘ I. $.30. 0". :.: \ D . w\‘ “.. ' v ..o‘? “ ‘g. . ' .9 : ‘,‘ . . .~ J00: -- ~."* .~.- . v. -. sac. -- .. ~-~ \* ~.-- . . ~ ~ '°- 90.... -- g: .6:- 3': w\ ... w o -..". \‘ ‘ “". '00 I... §": oce- - ‘9‘. -. $“‘ ‘5‘. m w ' o a: z..- 5 ? " «v ~ 06': Figure 12. A surface contour representation of the packing of complexed cations and anionic cavities in an 3-3 plane at 11:0 in Cs (18C6)2'e . The oblong shape cavities and the channel structure in the g_direct10n are apparent. 114 Aim-u": . Figure 13. A surface contour representation of the packing of complexed cations in Cs+[l8C6)2-e' as viewed down the 1,1,0 direction. The interstitial open spaces are channels to cavities behind the “wall" of complexed cations. 115 by superposition of several identical complexed cation images according to the proper symmetry of the space group. Figure 1; is a view down the g axis of a slice of van der Waals electron density in an g-g plane that includes the inversion centers at the electron cavities. Complexed Cs+ ions lie only 0.42 A from this plane and are visible as the boundary in the 3 direction of the stack of electron traps. The oblong structure of the cavity with the interconnecting channels between the adjacent cavities are notable. Figure 13 is a view of a (l 1 0) plane of complexed cations with all of the atoms included. The constricted point of the channels between cavities in the g—b plane are apparent in the figure as small gaps in the efficiently packed complexed cation "wall". Two types of channels exist, a bow tie shaped channel of dimension 2.5 A by 1.2 A, and a diamond shaped constriction of dimension 2.0.A by 1.2 A. Electrons in the same g—b plane might be expected to have a greater barrier to interaction with each other than the electrons along the 9 direction not only because of the smaller cross sectional area of the channel but also because the constricted channel is 4 A long between the cavities (see the channels marked by arrows in Figure 11) whereas in the g direction the channel is severely constricted for only ~l.5 A. Thus an electron trapped in a cavity has a structurally anisotropic environ- 116 ment in terms of the void space that connects it to other nearby trapped electrons. III.C. Structure Solution and Refinement for 03+(150532-I' III.C.l General Approach The crystal structure of Cs+(l505)2-I- was solved by the Patterson method. Refinement of 146 parameters required restraint of the positional parameters of the crown ether atoms to maintain a chemically reasonable structure. Least-squares refinement of an unrestrained crown ether resulted in unstable adjustments of the parameters and poor agreement factors. Both Cs+ and I— occupy special positions on rotation axes in the lattice; several positional and thermal parameters for these ions were fixed. The final agreement factors obtained in the restrained full matrix least-squares refinement were R - 8.6, and R = 11.6. w 117 III.C.2 Results of the Structure Determination of 06+(1505)2-I’ The positional and anisotropic thermal parameters are given in Tables 16 and 11, respectively, for all non- hydrogen atoms in Cs+(1505)2-I_. The positional parameters of the hydrogen atoms were constrained to ride on the carbons of the crown ether, these are included in Table 16. All hydrogens were assigned a constant value of the thermal parameter, U, of 0.216 A2; no refinement of the hydrogen thermal parameter was attempted“ The ionic packing in the tetragonal Cs+(1505)2-I_ is similar to the ionic packing in the 1806 salts described above. Cations and anions lie in alternating planes perpendicular to the g axis. The cations lie in planes perpendicular to g at g = 0, 1/2, etc.; the positions of the Cs+ ions on each plane are at the midpoints of the cell edges (i.e. 1/2, 0, z; 0, 1/2, 2; etc.). Cations are separated by 8.323 A along the g direction and by 9.316 A from four nearest cationic neighbors in the g—y plane. The anions have a small distortion from 5 coordinates of g = 1/4, 3/4, etc., and lie on four planes perpendicular to the g axis. Two planes consist of anions at the corners of the intersection of the plane with the unit cell at g = 0.24047, and 0.75729. The other half of the anions occupy sites at general positions 1/2, 1/2, a, for g = 0.24729 118 Table 16. Table of Positional Parameters and Their Estimated Standard Deviations in Fractional Coordinates for Cesium (15C5)2 Iodide at 291 K Atom x y 2 C81 0.000 0.500 0.000000 I1 0.000 0.000 0.2405(3) 12 0.000 0.000 0.7573(3) 01 0.234(1) 0.496(2) 0.012(2) C2 0.282(2) 0.428(3) -0.042(2) C3 0.243(2) 0.447(2) -0.125(2) 04 0.141(1) 0.414(2) -0.130(2) C5 0.128(2) 0.308(2) -0.119(3) C6 0.022(2) 0.280(3) -0.140(2) 07 -0.042(2) 0.300(2) -0.074(1) C8 —0.062(2) 0.212(2) —0.027(1) 010 —0.067(2) 0.301(3) 0.096(2) C11 —0.028(2) 0.313(2) 0.175(1) C12 ‘ 0.084(2) 0.302(2) 0.172(2) 013 0.126(2) 0.397(2) 0.143(3) C14 0.231(2) 0.386(2) 0.124(2) C15 0.268(2) 0.484(2) 0.092(2) C9 —0.009(2) 0.227(2) 0.051(1) C9' —0.017(2) 0.222(2) 0.054(1) H2A 0.352(2) 0.453(3) —0.044(2) H28 0.282(2) 0.356(3) -0.027(2) H3A 0.245(2) 0.519(2) -0.140(2) H38 0 289(2) 0.408(2) -0.l60(2) HSA 0.164(2) 0.284(2) -0.167(3) H58 0.158(2) 0.277(2) —0.071(3) H6A 0.017(2) 0.207(3) -0.151(2) H68 -0.007(2) 0.318(3) —0.186(2) H8A —0.027(2) 0.153(2) —0.049(1) H8B -0.136(2) 0.202(2) -0.032(1) HllA —0.056(2) 0.267(2) 0.215(1) H118 —0.043(2) 0.384(2) 0.192(1) H12A 0.108(2) 0.285(2) 0.226(2) H128 0.101(2) 0.248(2) 0.135(2) H14A 0.265(2) 0.370(2) 0.175(2) H148 0.240(2) 0.330(2) 0.086(2) H15A 0.248(2) 0.540(2) 0.127(2) H158 0.342(2) 0.480(2) 0.091(2) HQA 0.042(2) 0.203(2) 0.090(1) H98 0.024(2) 0.270(2) 0.011(1) ngA —0.021(2) 0.154(2) 0.077(1) H9'B -0.039(2) 0.220(2) -0.002(1) 119 Table 17. Table of General Displacement Parameter Expressions 0's in (A2) for Cesium (lS-Crown-S)2 Iodide at 291 K _———————_---————-—————————-——_———-—_-—-—————————--— Name U(1,1) U(2,2) U(3,3) U(1,2) U(1,3) U(2,3) CSl 0.0599(14) 0.0438(12) 0.092(2) -0.003(l6) 0.0000 0 0000 I1 0.0523(13) 0.0523(13) 0.0630(36) 0.0000 0.0000 0 0000 I2 0.092(2) 0.092(2) 0 083(6) 0.0000 0.0000 0.0000 01 0.108(11) 0.191(14) 0.147(13) -0.032(14) 0.008(14) -0.053(11) C2 0.130(15) 0.261(16) 0.231(16) 0.04(2) -0.046(15) -0.038(16) C3 0.169(15) 0.283(16) 0.119(15) 0.069(15) 0.083(15) 0.059(15) 04 0.149(13) 0.180(13) 0.072(10) —0.014(11) 0.042(11) 0 032(12) C5 0.221(16) 0.238(16) 0.266(16) 0.064(16) -0.078(16) O 106(15) C6 0.163(16) 0.138(15) 0.255(16) —0.057(16) 0.041(16) 0 051(15) 07 0.293(16) 0.101(13) 0.183(14) -0.069(12) —0.027(14) 0 024(14) 08 0.056(12) 0.076(12) 0.204(16) -0.017(12) ~0.046(12) 0.054(10) 010 0.216(15) 0.265(16) 0.349(17) —0.004(16) 0.046(16) 0.016(15) C11 0.290(16) 0.212(16) 0.131(15) 0.089(15) -0.047(15) 0 016(16) C12 0.127(14) 0.099(13) 0.078(12) 0.032(11) 0.042(12) O 048(11) 013 0.226(15) 0.255(15) 0.156(15) -0.052(15) 0.048(14) 0 013(15) C14 0.194(15) 0.175(15) 0.184(15) -0.015(15) -0.131(14) 0 028(15) C15 0.115(14) 0.115(14) 0.198(16) 0.019(14) —0.076(14) -0 022(13) C9 0.099(9) (Refined Isotropically) C9' 0.145(13) (Refined Isotropically) Hydrogen U's were not refined, and were set to U(Iso)=0.216 A2. The form of the anisotropic displacement parameter is: exp[-2n2{h2a20(1,l) + k2b20(2,2) + 12C2U(3,3) + 2hkabU(1,2) + 2hlacU(1,3) + 2klch(2,3)}] where a, b, and c are reciprocal lattice constants. 120 and 0.74047. The distortion of 1‘ positions from g = 1/. and 3/4, etc. results in two Ifi-I— distances along the 2 direction; a short distance of 8.043 A and a longer distance of 8.602 A. The next nearest inter—ani6nic distances are 9.318 A between iodides which are nearly planar perpendicular to the 5 axis. Bach ion is coordi- nated by eight counterions, four separated by 7.726 A and four separated by 7.855 A. Both the anions and the cations occupy special posi- tions on rotation axes. The anions are on four-fold rotation axes parallel to the 5 axis, this symmetry generates a square environment about the anion in planes perpendicular to the 9 axis.‘ The cations lie on 42 axes aligned parallel to the g direction. This symmetry element combines a 90' rotation with a translation of +1/2 g, which generates a local two-fold rotation axis. The 42 axis, in combination with the four-fold axes at the anions, requires each complexed cation to be rotated by 90' about the g axis relative to each of the six nearest neighbor complexed cations. The local two-fold axis at the Cs+ relates one crown ether to the other, so only one crown ether of each complexed cation is unique. The crown ether molecule was refined with restrained bond lengths and angles, but was allowed conformational freedom. Table 18 lists cesiumeoxygen distances and 121 Table 18. Table of 80nd Distances (in Angstroms) for Cs (lS—Crown-5)2’I— at 291 K, Atom 1 Atom 2 Distance (Angstroms) C51 01 3.092(17) C51 04 3.075(19) C81 07 2.960(18) C81 010 3.194 CS1 013 3.219 01 C2 1.418(2) 01 C15 1.416(2) C2 C3 1.480(2) C3 04 1.418(2) 04 C5 1.417(3) C5 C6 1.480(2) C6 07 1.417(2) 07 C8 1.421(2) C8 C9 1.494(3) C8 C9' 1.481(2) 010 C11 1.428(6) 010 C9 1.445(9) 010 C9' 1.419(4) C11 C12 1.485(2) C12 013 1.427(6) 013 C14 1.417(3) C14 C15 1.480(2) C9 C9' 0.128 All Carbon—Oxygen Bonds Were Restrained During Refinement. 122 selected interatomic distances along the crown ether back- bone in the Cs+(l505)2-I- structure. The mean Cs-O distance in this salt is 3.15 A, 0.2 A less than the mean Cs-O distance in the Cs+(1806)2 salts. A least—Equares plane was fit to the oxygens of one crown ether to describe the orientation of the crown in the structure. The plane fits the crystallographic equation, in Angstroms, -7.107§ + 11.0901 — 0.081; - 3.668 = 0. The results are given in Table 1g. The least-squares plane is nearly parallel to the g axis so that the least-squares planes of adjacent complexed cations along a 5 axis are perpenduclar. The anionic radius of I- has been estimated to be 2.15- 2.20 A [91]. Table ll_lists the nearest proton contact distances to each of the unique iodides. The nearest protons are centered 3.34-3.40 A away from the I- centers, which implies a van der Waals radius of 2.14-2.20 A. The twelve nearest protons give an average radius of 2.3 A for each of the two I- ions. The good agreement of the estimated van der Waals radius of I— from anionic contact distances with established estimates strengthens the validity of the method used to estimate the values of the Na_ and e- radii. Since the anionic radius of I- is similar to that of the trapped electron in Cs+(1806)2-e-, it might be expected that the structure of Cs+(1505)2-e— would be similar to 123 Table 19. Table of Least-Squares Planes for Ce51um (15-Crown-5)2 Iodide at 291 K Crystallographic Equation of Plane in Angstroms -7.1070 X + 11.0900 Y + -0.0808 Z - 3.6683 - 0 ----- Atoms in Plane ----- 01 3.0855 6.5336 0.1949 0.1668 04 1.8612 5.4493 -2.1673 -0.0741 07 -0.5530 3.9503 -1.2291 —0.0381 010 -0.8834 3.9662 1.5998 0.1399 013 1.6659 5.2074 2.3842 —0.1944 ----- Other Atoms ----- CSl 0.0000 6.5860 0 0000 1 8767 I1 0.0000 0.0000 4.0026 -3.6878 I2 0.0000 0.0000 12.6051 —3.7295 C2 3.7162 5.6389 -0.7057 -0.9224 C3 3.2069 5.8855 -2.0738 —0.4333 C5 1.6863 4.0558 -1.9774 -1.1539 C6 0.2961 3.6926 -2.3340 —0.7078 C8 —0.8219 2.7942 —0.4487 -0.8702 C11 -0.3670 4.1377 2.9199 —0.0008 C12 1.1090 3.9845 2.8644 —0.9259 C14 3.0425 5.0908 2.0673 —1.0338 C15 3.5288 6.3814 1.5313 —0.2070 C9 —0.1197 2.9918 0.8551 -l.0890 C9' -0.2204 2.9262 0.8982 —1 0902 124 that of Cs+(15C5)2'I_. The surface contour representation of the eight complexed cations surrounding an anionic cavity was generated on the Evans and Sutherland P8300 graphics station as a possible model for the electron cavity of Cs+(15C5)2-e-. The void space left in the crystal structure of Cs+(15C5)z-I- when the anion, I-, is removed resembles the cavity and channel structure of Cs+(1806)2-e-. There is higher symmetry in the 1505 salt; the four-fold rotation axis along the 5 axis through the anionic site gives the iodide cavity a more circular shape in planes perpendicular to the g axis. In addition there is less elongation along the g axis in the iodide, so that the three dimensional shape of the cavity is very nearly spherical with a radius of ~2.5 A. Adjacent cavities are joined by channels through the complexed cations. The symmetry in the crystal requires the four channels to neighboring anions in a plane perpen— dicular to the 5 axis to be identical, thus only two of the six channels are unique. The channel along the 9 axis is most constricted in those planes perpendicular to 9 that contain the Cs+ ions at g = 1/4, 3/4, etc. The constric- tion in the channel is square shaped, with an edge length 'of 1.75 A. The second type of channel connects the set of four neighboring cavities which are nearly in the same g—b 125 plane as the central anion. The channel is ~4.5 A long along the 9 direction and 1-2 A wide. This channel has a 5 component since the four anions surrounding the central anion are displaced by 0.559 A along the g direction; this results in an effective channel size through the lattice that is reduced to only 1 by 3 A. I If Cs+(15C5)2-e— crystallizes into a similar structure, based on close packing of the complexed cations, these channels would probably be important to electron-electron interactions. In addition to the more spherical anionic cavity in the Cs+(l505)2-I_ relative to Cs+(1806)2-e_, the anion-anion distances are both shorter and more isotropic (see Table 12). The channel structure in the Cs+(1505)z lattice is most notably different from the Cs+(1806)2-e- structure in the gr! directions, the channel is significantly larger and shorter. The channels along the IO direction are more subtle in their differences; Cs+(1806)2-e— has a larger cross-section and a shorter constriction, but Cs+(15C5)2-e— has a more open channel, the relative effect of these structural features on interactions between anions cannot be easily predicted. Magnetic proPerties will be used in Chapter VI to probe the electron-electron interaction in the electrides Cs+(1565)2-e— and os+(18cs)2-e'. CHAPTER IV THERMAL CHARACTERISTICS or THE IODIDES AND.ELECTRIDES 0F Cs+(18C6)2 AND Cs+(1505)2 COMPLEXED CATIONS Many of the prOperties of Cs+(15C5)2-e_ and Cs+(18C6)2-e- are measured as a function of temperature, and thus may be sensitive to phase transitions, or more subtle changes in the interionic interactions. Differen- tial Scanning Calorimetry (DSC) allows the endothermic and exothermic transitions of a material to be defined in terms of the heat of the transition and the temperature of the transition. Optical spectroscopy, in the case of Cs+(l5C5)2-e-, offers information about the temperature dependence of the extent of electronic localization in the cavities of the solid. Information from both of these experiments will be used to interpret the temperature dependence of magnetic data in the following chapters. 127 IV.A Differential Scanning Calorimetry IV.A.1 use of Cs+(1505)2-I- and Cs+(1806)2-1‘ The DSC traces of Cs+(1505)2-I- and Cs+(1806)2-I- were measured over the temperature range of ~120'C (153 K) to 250°C (523 K). The DSC trace of Cs+(1806)2-I— is shown in Figure 13, Three features are evident in the trace at -100'C, -35.5'C, and at 185.7'0. The low temperature feature is not a characteristic of the sample, but rather a surge that occurs whenever the instrument begins a sub- ambient temperature scan. The melting point of the Cs+(1806)2-I- occurs at 185.7'C and has an approximate AB melt = 97.1 J/g. A second endothermic transition occurs at -35.5'C, with a much weaker heat of transition, Ant = 4.7 J/g. The trace of Cs+(1505)2-I‘ is nearly identical to that of Cs+(18C6)2-I-. Endothermic transitions at -36.3'C, with AB = 5.4 J/g, and at the melting point, 180.2'C, with t AB = 23.6 J/g are the only transitions in the range of melt -100'Cov m\n «.dm u\H o.m o ®.mN U o.ON o m.ml o H.0MI Anxsoa uawmaodho>ov u\fi H.mm m\H o m.mN o o.pH o m.ml o o mm.wm o hm.mm o mm.¢ m\H N.bm 0 o.mm~ e\e m.mm o ~.ome Ame Ame Ame Ame e e e a .aaoooe ado: m sossasssne N scenessdne '|"‘l"'ll"‘- ls'l’l'"|l|'ll|lll-'l IIIII‘II'III'II'I.I'||I| I'llIIl-'||"l'-l‘lll'l-llo‘"'l||'|-'III' . 1‘ lll"|"'lll'l||ll'lll'll'lllll'l"-l-'l‘.e'""l"'l'l‘llI'lls|ll|llllll'1'll|l|‘|lll' Assess o e.emu m Axsmxv e o m.emn N to. Amomev+no -o.afimomsv+so m\e em.e o m.mm: m 1H. Amomev+no w\e mm.m o m.omu m 1H. Amomev+no Ame a vasomaoo e consencnns .NAeoch+so ass . . NAmoQHv+mo ho movwhaoodm was nwvwvoH 04¢ mom magnum: own «0 khaalsm .cN Canoe 132 at the melting temperature and continues to 50°C. The approximate heats of each transition are tabulated in Table 29. Approximate values are listed because some sample oozed from the sample pan prior to mass determination. Visual observation of Cs+(l505)2-e- indicates that the sample melts with decomposition near room temperature; the DSC trace corroborates this observation. Two other transi- tions are characteristic of Cs+(1505)2-e-; the weak pair of endotherms at ~—35'C are similar to the low temperature endotherm in Cs+(1505)2-I-, the stronger transition at ~5'C is unique to the electride. The DSC traces of three samples of Cs+(1806)2-e_ were measured with inconsistent results. The melting and decomposition characteristics were consistent for all samples and were in agreement with Dye’s results [71]. A typical trace is shown in Figure 16. Cs+(1806)2-e— melts at ~39°C, and decomposition starts at immediately higher temperatures with an exothermic peak at 54-58'C. No signi- ficance may be attached to the lower temperature transi- tions as none were reproduced. All three samples were prepared from the same synthesis, and the samples were large enough to expect homogeneous results. 133 .-o.mnouwfiv+mu mo mHsEmm ocflaampmxsoxaom m mo muses umo .Uev munumcmaemb cod cm ow ov ON C ow: p . D D I D! m b m I -m b m ovn . cm: 901 . .ce answee coal \r, .r» \r m.s- tog: (B/n) hora 393B 134 IV.B Optical Spectroscopy of Cs+(15C5)2-e_ The optical spectrum of Cs+(15C512-e- has been measured several times with different samples, with only partially reproducible results. The main features of the spectrum are, however, reproducible. There are three maxima in the absorption spectrum which can appear, although any one spectrum may or may not display all of the features. One film was remarkable in its temperature dependence. The spectrum changed reversibly as a function of temperature, showing different spectral features in different tempera- ture regions. Similar results were never reproduced. The method of making films involves several variables which are not controlled, such as the exact temperature of the solution from which the film is made, the rate of film formation, and the thickness and uniformity of the film. These may affect the microcrystallinity of the film and subsequently give variable results. In all spectra the optical absorption tends to be quite broad, rising from 700 nm to maxima at 900-1000 nm and/or 1250-1350 possibly with a shoulder at 1550 nm and a gradual decay of absorbance to 2500 nm. A typical spectrum, of a film displaying all three features, is shown in Figure 11. The absorption maximum in localized electrides is generally considered to be the excitation energy required to promote \III' 135 .UoooHu pm no.NmmumHv+mu mo Edspoomm :oflumsomam Hwoflpmo wouflamssoz nmwwmmmwflm :eozmq m> = 7 (Y Y h)2 s)?/2 Hz. (5.2) 1/2 , , s . . 2O . MAS-NMR reduces dipolar broadening to zero if the frequency of rotation is greater than Av Dipolar broadening may also be 1/2'. eliminated by using decoupling techniQues in static spectroscopy, and (in MAS experiments when the dipolar coupling can not be completely removed by spinning. V.A.2.b. Chemical Shift Anisotropy. Inhomogeneous broadening of an NMR transition may arise from an anisotropic chemical shielding environment about a nucleus.21 The orientation of the principal axes of a diagonalized chemical shift interaction tensor with respect to the applied field will shift the resonance position from the isotropic its value of the chemical shift. The chemical shift Hamiltonian for a nucleus 1 is H = Y in which 022 is related to the contributions of the principal components Ci of chemical shift tensor thrOugh o = A o + A20 + 120 (5.“) where A1 is the direction cosine of principal axis 1. The isotropic chemical shift occurs at if = 1/3 for all i. That is, 0. = (0 + 0 130 1 + 03) / 3 7 (5i5) 2 Transforming the Eulerian angles, 0;, to the angles 0 and o in spherical polar coordinates according to 0% = G, 0030; = sinesin¢ and 00305 = sinecos¢ gives 0 = 0 + l-0nsinzecos2q} + l-0‘(3cosZL-)-1) (5.6) ZZ 130 2 2 _ ’ where the anisotropy parameter, 6, is 0 E 0 - 0. , and the . 1 o asymmetry parameter, n 18 3 s + The shielding of the Cs ion in the sandwiched cesium crown ether complexes studied in this work is generated by the Coordinating oxygen lone pairs. There is a nearly cylindrical symmetry of the oxygens about the cation, so that two components of the chemical shift tensor, l__‘ its 01 and 02, are equal. This axial symmetry further simplifies the anisotropic chemical shift. For an axially symmetric single crystal with random orientation of the principal axis 0 with respect to the 3 applied field, the shift from 0iso is 1 l 2 022 - 0180 + 2 0(3cos 0 1) . (5.7) or re-expressed in frequency units, the shift from the isotropic Larmor frequency is N|—* (n-VL) = vL6(300826-1) . (5.8) The high field limit of the frequency shift occurs at 0 = 0, 180° where v-vL = 0L5, and the low field limit of the chemical shift at « “de O = i90° is v r v = 2 The powder pattern for an axially symmetric material may be given as g(v) = dcose/dv. (v-v ) *1/2 1 r 1 2 L " - 8(V) ‘ (30Lo) L 3 + 3 0L5 1 (5'9) for -v 0 2 < (v‘vL) < 0L0 Rotating the sample at an angle 8 from the applied field has been shown to introduce a factor (3cosZB~1) into the time averaged chemical shift expression2O '“rr 2 *1 , 2 1 2 . OZZ - Oiso + 6/2 (3005 B 1) L2 nain BCOSZo + 2 (3003 O 1)). (5.10) lh6 When 8 is the magic angle, 54.7H°, 022 = 0150’ and the effect of chemical shift anisotropy broadening is removed. V.A.2.c. Quadrupolar broadening. The NMR spectrum of a quadrupolar nucleus (I>1/2) in an axially symmetric anisotropic environment will show a powder pattern that reflects removal of the degeneracy of the 2I+1 spin transitions. The first order quadrupole effect, in the high field limit where the Zeeman splitting is large relative to the quadrupolar splitting, causes the resonance line to split into ZI+1 22,23 - components. To first order, the shift of an energy level in a spin state m is given by V . AB; = 59-[ (300520-1) 1 1 3m2 - I(I+1) J . (5.11) Nl-J O is the angle between the principal axis of the electric field gradient tensor and the applied magnetic field, and 0Q is the nuclear quadrupole freQuency obtained from the product of the electric field gradient, eq, and the quadrupole moment of the nucleus, eQ, as 2 e 90 V = 3 2hI(21-1) 5.12 0 (._) Allowed NMR transitions have Am = :1, so the freQuency shift of the transition m + m-1 is 1 2 , 1 ‘. Avm = v é-(3cos 0 1) (m 2) 1 (5.13) Q L To first order, there is no shift from the isotropic value in the case of the transition m = 1/2 9 -1/2; a single crystal will show a spectrum consisting of 21+1 lines, equally spaced from the central 1A7 line at positions given by \) Q §—-(3cosze-1) (m-1/2) The pattern that arises for a powder that contains crystallites at random orientations may be calculated as Av _ dcose _ 1 . 2_ m --1/2 g(v) - dv - 3vQ(m-1/2) L 3 vQ(m-1/2) (5114) where Av lies between v (m - l) (at O = 0 180°) to :29-(m - l) at m Q 2 ’. 2 2 e - i90°. Magic angle sample spinning is capable of averaging the firSt order quadrupole effect to zero,214 but only in the limit that the spinning frequency is greater than the nuclear quadrupole frequency. In general this condition is seldom satisfied. For nuclei with large quadrupole moments in strong electric field gradients, even the second order quadrupole effect is large. The second order effect shifts the center of gravity of the spectrum, broadens the central line, and can only be partially reduced by MAS. 13303 has a very small quadrupole moment and does not show significant second order effects. The combined effects for a nucleus having both CSA and first order quadrupole effects can be calculated for the single crystal and powder cases. The simplest case is that which has both asymmetry parameters (n) eQual to zero with colinear principal axes. The line that is shifted and broadened by CSA (eduation 8) will be split into 21 lines by the first order quadrupole effect. The resonance position for each transition m + m ~ 1 will then be 1h8 59-(300320-1) so that V v = v + (m ~ 1/2) L 2 j ( 3003 0 1 ) V: 2 vL + L vLé + (m — 1/2) vQ (5.15) The limits on the chemical shift are 1 1 . . “ E'leé + (m r 5’) vQ|S|v - VLISIvLC + (m - —) v The shape function of the powder pattern for each m can be derived as 1/2 ' 8(v) = d0°39 = 3- _ I(5.16) dv 2v v ,. (v 0 + (m - 19v E L + 1 )1/2 L 2 Q ‘ (de + (m— %) 0Q) The effect of CSA on the first order quadrupolar powder pattern is to extend both the high frequency and low frequency limits of nuclear transitions with m > 1/2 and to compress the breadth of the powder pattern for nuclear quadrupolar transitions with m s - 1/2. The central line assumes the profile of a CSA broadened line in the absence of quadrupolar interactions. V.A.3 EXPERIMENTAL The synthesis of salts with complexed cesium cations used in this study followed the procedures of Metz e£.§l.9 for salts complexed with C222, and those of Pedersen?“ for salts complexed with crown ethers. A stoichiometric methanol solution containing reagent grade salt and vacuum~sublimed complexant was warmed gently until all components were dissolved. Butanol was added until the solution was saturated. The solution was placed in a flat crystallization dish in 1&9 a dust free area. Evaporation of the methanol left crystalline material in a butanol slurry. The crystals were first filter-dried and then oven~dried at ~70°C. 133Cs NMR spectra were obtained at three frequencies: at 65.61 MHz with a home-built 11.75T spectrometer25 (pulse length 4.5 ms and 10 s delay time); at 47.2A MHz with a Bruker 8.45T,spectrometer (pulse length 6.0 ms and 10 s delay time); and at 23.62 MHz with a Bruker H.23T spectrometer (pulse length A + 6 ms and 3 s delay time). Magic angle sample spinning with the two high frequency instruments used an AndrewsrBeam spinning system. The rotors were made of either Delrin or d8—polymethyl methacrylate. Spinning rates were between 2 and 4 KHz. Measurements at 23 MHz used a Doty dual bearing spinning system. The rotors were made of either sapphire or Delrin. Spinning rates were between 1.5 and 3 KHz. All measurements were made at room temperature. Static spectra were obtained with the same probes but with no spinning gas. Single crystal studies were done in the Doty probe by fixing a large single crystal (~2mm3) onto a rotor wall with Apiezon N grease. All chemical shifts were measured with respect to a + value of 0 ppm for Cs (aq) extrapolated to infinite dilution. V.A.4. RESULTS AND DISCUSSION V.A.4.a. MAS-NMR of Simple Salts and Complexed Salts The corrected chemical shifts and linewidths of several simple salts and complexed simple salts as determined by MAS-NMR are Summarized in Table 21. The simple cesium salts show a paramagnetic shift from the 150 Table 21: Chemical Shifts and Linewidths of Simple Cesium Salts With ‘ and WithOut Complexed Cations. “ Compound 6(ppm)a Av1/2(Hertz)b + .— Cs SCN +190 . 190 Cs+Cl~ +232 125 Cs+Br- +26“ Cs+I- +28u 420 Cs+TPB- undetected --- Cs+(18C6)-SCN_ + 73 100 Cs+(1806)°I- +171 280 Cs+(1806)2oSCN" - 59 70 Cs+(1806)2+1‘ — 59 ‘ 47 Cs+(18C6)2°TPB- , - 03 80 Cs+(1505)2-I‘ + 29 90 Cs+(15C5)2°SCN_ + 32 108 Cs+c222-SCN”oH20 +50, +275 210, 020 + ‘ . Cs C222-SCN +238 140 + _ _ Cs C222-I +232, 25A 63. 135 Cs+c222-01” +167 860 + aChemical shift relative to a value of 0 ppm for Cs (a ) at infinite dilution. q bFull width at half height. 151 gaseous Cs° resonance position of ~550 ppm.26 The chemical shift increases with increasing polarizability of the anion, ranging from +190 ppm relative to the Chemical shift of Cs+(aq) for the thiocyanate salt to +284 ppm for the iodide. Ramsey27 described the paramagnetic shift in terms of the change in the total angular momentum about the nucleus in one state relative to another. For the ionic cesium halides, for example, the magnitude of the shift is proportional to the energy gap between the ground state and an excited state that has electron density provided by an electron donor in Cs p or d orbitals. An anion which is a better electron donor provides stronger overlap and hence a larger chemical shift. Of the alkali halides Cs+ is expected to show the greatest shifts, since the unoccupied p, d and f orbitals provide lower lying excited states than are available in the lighter alkali metals. 0ne-to~one complexes of simple salts with 18C6 were isolated for the iodide and thiocyanate. The large Cst cation is not fully coordinated by the complexant molecule; thus, it still has a significant ability to interact with coordinating anions. The chemical shifts of these salts are upfield from the corresponding simple salts by some 110 ppm. The effect of interaction of the cation with the lone pairs on the crown ether oxygens is not as strong as that of any of the anions Studied. There is, however, still a strong dependence on the counter-ion. Compounds with a 2:1 mole ratio of complexant to cation were prepared from several simple salts with both 15C5 and 18C6. The 1806 sandwich systems show a remarkably constant Chemical shift as a function of the anion, 0 = -59 ppm for the iodide and thiocyanate, and 152 6 = ~M3 ppm for the tetraphenyl borate. These shifts are close to the value of the chemical shift seen at high complexant concentrations in various solvents.8 This invariance to the anionic environment indicates that the Cs+ ion is well shielded by the crown ether sandwich from strong interactions with the anion. The paramagnetic Ramsey shift is thus due only to the interaction of Cs+ with the twelve ether oxygens. The chemical shifts of Cs+(1505)2ox- salts are also upfield from the pure simple salts, and show only a weak dependence on the anion. The chemical shift range is +28iu ppm for the iodide, bromide and thiocyanate salts; however, the previously established order for paramagnetic shift based on the anion donicity does not hold. Further, the chemical shift is about 90 ppm downfield from the Cs+(1806); resonance. This suggests a stronger paramagnetic Ramsey interaction with the oxygens of 15C5 than with the oxygens of 18C6, and that the anion does not directly interact with Cs+ in either case. In the structures of the two compounds, Cs+(1505)2-I 28 and CS+(18C6)2-Na‘ 29, the mean Cs-O distance in the 18C6 sandwich is significantly longer than that in the 15C5 sandwich ( Cer18c6 = 3.35A; Cs-O = 3.15A ) 15C5 This reduction in interatomic distances results in stronger electronic overlap of oxygen lone pairs with Cs orbitals in the case of the smaller complexing agent. The sensitivity to the Cs~0 distance also implies that the small variations in the chemical shifts of the sandwiched cations from one salt to another may result from small changes in the mean Cs+0 distances. For example, the +16 ppm shift 153 + , —- _ + -— seen in Cs (18C6)2-TPB relative to Cs (18C6)2-l may be the result of shortened Cs-O distances in the TPB salt because of the conformational changes in the crown ether required to accommodate the bulky anion. The three dimensional cryptand 222 (C222) is also able to form + complexes with Cs , but only in a 1:1 ratio. Mei et_al.8 and 9 studied the complexation of Cs+0222°X- in several Kauffmann.gg‘al, solvents for X- = TPB- and SCN“. The temperature dependence of the resonance line of Cs+C222 implied an equilibrium between an exclusive complex, stable at high temperatures, and an inclusive complex stable at low temperatures. with X” = octanoate in a THF-HZO mixture, a two line spectrum indicative of slow inclusive-exclusive exchange was observed.9 The crystal structure of Cs+C22208CN-~H20 was solved?5.and consists of only the inclusive complex, in which the cation is completely encapsulated by the three polyether strands of the cryptand. In the exclusive complex, the cation interacts strongly with only two of the polyether strands of the cryptand, and not as strongly with the third. This leaves the cation exposed to its anionic environment on one face. The chemical shifts of inclusive complexes of Cs+ with C222 in solution occur at ~240 ppm and are independent of the solution environment, whereas the exclusive complex displays a range of chemical shifts that are strongly dependent on the solvent, counter ion and temperature of the solution. Solid state MAS~NMR results indicate that the inclusive complex may be seen either alone or in addition to the signal of an exclusive complex. The inclusive complexes of Cs+C222 salts show a range of chemical shifts near 240 ppm. Cs+C222-SCN—0H20 was the only sample to show only the signal of inclusive complex at 238 ppm, very close to isu the value seen in solution. The mean Cs-O distance in this salt is 2.96 A,T5 so the Ramsey shift is due to a strong interaction of Cs+ with oxygen lone pairs. In those samples that show both inclusive and exclusive peaks, the range of chemical shifts is large for both peaks, indicating that ionic packing may place constraints on the available conformations of the 0222 molecule. The linewidths of the MAS-NMR spectra are, for the most part, between 50 and M00 Hz. Two trends are apparent in the comparison of linewidths. First, lines are broader at high frequency than at low frequency and, second, at any given frequency the more paramagnetic the chemical shift, the broader is the MAS-NMR line. These observations indicate that the broadening mechanisms are probably chemical shift anisotropy and dipolar broadening. V.A.4.b. Static Spectra of Simple Salts and Complexed Salts V.A.H.b.i. §§g304° The importance of anisotropic chemical shift interactions in Cs salts can be seen in the C5280“ spectra, at two resonance frequencies. The primary broadening mechanism in this salt was originally attributed to first order quadrupolar effects.30 Figure l9 shows the static spectrum of C3280“ at 52.n8 MHz and at 23.62 MHz. The line is broadened by a factor of about 2.2 by increasing the applied magnetic field by the same factor. Chemical shift anisotropy broadening is prOportional to the applied field, ‘whereas quadrupolar effects are constant on a frequency scale and are a less significant effect at high fields. Therefore, the broadening mechanism in this non—cubic quadrupolar solid is due to chemical shift anisotrOpy rather than quadrupolar effects as was originally 155 52.48 Megahertz 1 u' : 23.62 Megahertz : 1 .15?" 5 L.3 : 940 Hz W : 2100 Hz 41 1 i J 4 1__JL_ i_ L l 1 4— L. n. 1_ J J 1 ‘[_ 13 10 7 I l -2 -5 KILO H ERTZ F' l 133 . . . . igure 9. Cs static NMR spectra, comparing 11new1dths of C52804 at 52.48 MHz and 23.62 MHz. 156 O assumed.3 + - _— _- V.A.H.b.ii. Static spectra of Cs (18C6)q-I and Cs+(1SCS)Q;I . The static T33Cs NMR spectra of Cs+(18C6)2-I_ and Cs+(15C5)201~ were studied to deduce the nature of the chemical shift and quadrupolar interactions. The parameters of these interactions give some insight into the local structure about the Cs cation. The static spectrum at 52.n8 MHz for Cs+(18C6)2-I‘ is shown in Figure 20a. The spectrum consists of a broadened central component near the isotropic chemical shift and unsymmetrically spaced quadrupolar satellites. The same compound was studied at 23.62 MHz and the spectrum is shown in Figure 20b. The spectrum has an unstructured, broadened central line with symmetrically spaced quadrupolar satellites. This simple, low field spectrum reveals a nuclear quadrupolar coupling frequency, vQ = 6.4 KHz that corresponds to a quadrupolar coupling constant, e2qQ/h = 89.6 KHz. The central line in the low field spectrum has a width at half height of 1200 Hz which is due to CSA broadening, and dipolar broadening. The strength of the chemical shift interaction is proportional to the field; the high field spectrum reflects the addition of a significant CSA contribution to the resonance. The central transition has a peak width at half height more than two times greater than the low field case, and has structure that reflects axial symmetry of the CSA tensor. The shape of the central peak is consistent with an asymmetry parameter (n) of zero. The width of the peak gives the anisotropy parameter 6332 ppm. The anisotropic (me1/2) Quadrupolar spin transitions are not equally spaced because of the CSA interaction. Equation 16 was used to generate the static powder 157 .ou c was Ema mm" 0 £5. m.on 09 How 3 ccwumsvm 509m «.30QO woumasoamo 0.3 3:: cognac 9:. .sz fitmm um. 1H.mmouw$+mu mo «50QO $22 83. 033m .mom 92%: $15 A «Ts. 0.? m..+ N... m+ ..1 o v: m- NT 0.- ON- P p li'll“l. fi-‘l ‘ —--=-——--- - --- --‘--- o-- - -.---~~ - . 158 .H .oucpcm Ema Sum .26. ménc.» ow 0H :OMymswo Echm mhuocmm wouwfisoflmo mew mocha wonmmw 6:5 N22 No.mm um 1H.Nflouwfiv+wu mo whuommm mzz wommH oeumpm .nom oHDMflm Ev. .:-s m... N... 9 .1 0 Le- m- N... 9.. fl -..----.“ m-—“'"" ---‘--‘ “w; " '21:-..— 159 lineshape for the two Larmor frequencies with estimated values of vQ = 6.4 KHz, 6 = 32 ppm and n = O. The results are shown as dotted lines along with the actual spectra in Figure 20a & and 20b. The agreement is good in both cases although no refinement of the parameters was performed. It has been assumed thrOughout this discussion that second order quadrupole effects are negligible in these salts. The second order quadrupole interaction shifts the center of gravity of the spectrum, the shift being given by3? VL * v = VQZ/(vL'BO) { I(I+1) ‘ 3/4} (1 + n2/3) . (5.17) At VL = 23.62 MHz the caICulated center of gravity is shifted by only 3.5 Hz and at v = 52.u8 MHz by only 1.56 Hz. Thus, second-order L quadrupole effects can be neglected. The static powder spectrum of Cs+(15C5)2~I‘ was measured only at 23.62 MHz. The spectrum is shown in Figure 21. The central line displays an asymmetric profile similar to that at high fields for Cs+(1806)2-I-, but without any of the non-central quadrupolar transitions apparent. A simulation of the central line yielded the chemical shift parameters 6 = 52 ppm, n = O. The ca10ulated lineshape is shown as the dotted line in Figure 21. The absence of observable quadrupolar satellites indicates that the quadrupole coupling constant is large. The sweepwidth of this spectrum was 30 KHz. The effect of the dipolar interaction of crown ether hydrogens 133 with the Cs nucleus on the static line of Cs+(15C5)2-I- may be calculated by equation (5.1). The cesium~hydrogen distances are + .. obtained from the crystal structure of Cs (15C5)2°I .28 Interactions 160 .on C mcm Ema mm" m. .NE m.omno> how 0H :oflpmzcm Eoem Edeuommm woumHSUHmc ogp mfl mafia eosmmv omh .sz No.mm um +H.Nmmumfi.+mu mo espuuoam mzz momm. ouuaum ..N otsmum NIVZ 4:13 O .9 ¢ 3 9 cu § - 1/2 transitions fbr two Cs+(15C5)2 orientations. Lines labelled a and b represent corresponding non-central quadrupolar transitions. 165 interactions if it is assumed that 0 = 0 Q CSA’ and n = 0 by equation (5.15). For any one magnetically unique cation the unknown 0 and v0 may be calculated as follows. The shift of the central line from the isotropic chemical shift has no quadrupolar component, so 0 can be calculated with the value of 6 obtained from the powder spectrum. VQ may then be caICulated from the position of any of the satellites. The validity of the assumptions, and the value of 6, may be checked by comparison of the results for several central line positions. The results for the four lines of Figures 23a and 23b are given in Table 22, A quadrupole coupling frequency v = 30,500 Hz, which corresponds Q to a quadrupole coupling constant of 427 KHz, is obtained for Cs+(15C5)2°I-, from each of the three observable lines. The fourth orientation predicts that the nearest quadrupolar satellites should occur beyond the sweep width of the spectrum. The results are consistent with the previously measured value of 6, and the assumption that 0Q = eCSA appears to be a good one. A single crystal of Cs+(18C6)2-I- was studied in the same fasnion as the 15C5 salt. The static single crystal spectrum (Figure 24) consists of only one line with quadrupolar satellites. While the structure of this compOund is not known, crystal structures of the Q electride and sodide of Cs+(18C6)2 have been determined. The packing in these salts indicates that only a single orientation of the + two crown ethers abOut Cs with respect to some axis in the crystal occurs. The NMR spectrum indicates that similar packing exists in the iodide. 166 Table 22. Parameters of Anisotropic Interactions in the NMR of Single ' Crystals of Cs (1505)2°I at 23.62 MHz. ' 85E?8B v-vL(Hz)b 00 300820-1 Av(Hz)d vaHze A —415 710 -O.689 21,000 30,500 B 255 46.5 0.424 13,010 30,679 A 1204 00 2.00 unobservedf -~— B ~550 80.2 —0.913 27,694 30,333 aRefers to spectra shown in Figure 23. bThe frequency shift due to CSA interactions of the central line from the isotropic chemical shift. 00 is calculated from v-vL by equation 5.8. dAv is twice the measured frequency difference between peaks corresponding to successive quadrupolar transitions; Av is the same as v-vL in equation 5.15. er is calculated from equation 5.15. fCalculated to appear outside of the spectral width used. 167 .COHHDHOm Heapwmow 0p one saucmsmmmw ma oxfimm macaw 0:9 .COfiHmpCOfiHO Eowcmh m #6 H.Nm©Uva+mU mo Sappoomm £22 MUMMH prmXHo ®Hm¢wm .VN oHDMHm 3.3.: $13 0.5+ 0+ m.N+ O 0N1 m1 0&1 _ . u d - 168 V.A.S. CONCLUSIONS It is clear that the local structure about a 133 Cs cation has a profound effect on its nuclear magnetic properties. Solid state NMR is sensitive to various anisotropic interactions; dipolar, CSA and quadrupolar interactions are usually significant. The isotropic chemical shift is an important property of a complexed Cs salt; it serves as a fingerprint for the nature of the complexation, inclusive or exclusive, and is even characteristic of the complexant. This information is typically accessible by MAS techniQues. Solid state spectra allow the parameters that characterize the strength and anisotropy of the line broadening mechanisms to be deduced; these are in turn strongly dependent on the local structure about the nucleus. The broadening characteristics of the iodides, Cs+(15C5)2-I~ and Cs+(18C6)2'I~, were determined by powder and single crystal static NMR methods. The spectra of other Cs+(18C6)2 salts have been measured, and have identical powder patterns, so that the anionic environment is relatively unimportant, just as was the case for the isotrOpic chemical shift. The iodide salts may serve as models for other salts with the complexed cesium cation, including the 32 alkalides and electrides. As new compounds are synthesized from more exotic complexants, NMR methods will remain as important probes, not only of the stoichiometry, but of local structural characteristics as well. Obviously, NMR does not add to the structural analysis of those compounds that can be studied by single crystal diffraction methods. There are, however, many compounds that contain complexed + Cs , but which are not easily crystallized, so that a solid basis for structural information cannot be obtained. An understanding of how 169 known structural characteristics affect the isotropic and anisotropic magnetic interactions will provide a powerful tool for studies of 133 - Cs containing compounds. V.A.6. ACKNOWLEDGMENT This research was supported by National Science Foundation - Solid State Chemistry Grants DMR 79-21979 and DMR 8414154. We are grateful to Dr. P. B. Smith of the Dow Chemical Company, and to Dr. E. Oldfield and Mr. B. Montez at the University of Illinois at Urbana- Champaign and Dr. Peter Edwards of Cambridge University, England for assistance in obtaining and interpreting NMR spectra. V.A.7. REFERENCES: 1) Ellaboudy, A.; Tinkham, M. L.; VanEck, B.; Dye, J. L.; Smith, P. ‘ B.; J. Phys. Chem., 1984, 88, 3852. ' - . . 2) Ellaboudy, A.; Dye, J. L.; J. Am. Chem. Soc., 1983, 105, 6490. 3) Dye, J. L.; Ellaboudy, A.; Chem. Br., 1984, 20, 21. 4) Tinkham, M. L.; Dye, J. L.; J. Am. Chem. Soc., 1985, 107, 6129. 5) Tinkham, M. L.; Ellaboudy, A.; Dye, J. L.; Smith, P. B.; J. Phys. Chem., 1986, 90, 14. . .- 6) Ellaboudy, A. Dye, J. L.; J. Hag. Res., 1986, 66, 491. 7) IUPAC names: 18-Crown-6; 1,4,7,10,13,16 hexaoxa-cyclooctadecane, abbreviation 18C6. 15"CPOWH‘5; 11”}7’10113 pentaoxa~ ' cyclopentadecane, abbreviation 1505. Cryptand (2-2-2); 4,7,13,15,21,24 hexaoxa-1,10 diazarbicyclic (8.8.8) hexacosane. abbreviation C222. 8) Mei, E.; Popov. A. I.; Dye, J. L.; J. Phys. Chem., 1977. 81, 1677. ' ' ' ' ' ' ‘ 9) Kauffman, E.; Dye, J. L.; Lehn, J.-M.; Popov, A. 1.; J. Am. Chem. Soc.,1980, 102, 2274. ' - - - - . 10) Gutowsky, H. S.; McGarvey, B. R.; J. Chem. Phys., 1953. 21, 1423. 11) Andrew, E. R.; Arch. Sci. (Geneva), 1959. 12, 103. 12) 13) 14) 15) 16) 17) 18) 19) 20) 21) 22) 23) 24) 25) 26) 27) 28) 29) 30) 31) 32) 170 Andrew, E. R.; Bradbury, A.; Eades, R. 0.; Nature, 1959, 183 1802. ' ‘ ‘ ~ ' Lowe, I. M.; Phys. Rev. Lett., 1222, 12, 285. Pedersen, C. J.; J. Am. Chem. Soc., 1291, 89, 7017. Moras, D.; Metz, B.; Weiss, R.; Acta. Cryst. 1212,1329, 388. Poonia. N. S.; Bajaj, A. V.; Chem. Rev., 1212, 79, 389. . Dobler, M.; Phizackerly, R. P.; Acta Cryst., 1212, B30, 2748, DaweS. s. B.; Huang, R. H.; Ward. D. L.; Dye, J. L.; unpublished results, this laboratory.” ' ' - . - Van Vleck, J. H.; Phys. Rev., 1948, 74, 1168. Andrew, E. R.; Intl. Rev. Phys. Chem., 1981, 1, 195. Bloembergen. N.; Rowland, T. J.; Acta Metal., 1953. 1, 733. Abragam, A.; "The Principles of Nuclear Magnetism", Oxford University Press (London) 1961. Cohen, M. H.; Reif, F.; Solid State Physics, 1957, 5, 321. Cunningham, A. C.; Day, S. M.; Phys. Rev., 1966, 152, 287. Meadows, M. D.; Smith, K. A.; Kinsey, R. A.; Rothgreb, T. M.; Skarjune, S. P.; Oldfield, B.; Proc. Nat'l. Acad. Sci., U.S.A., 1982. 79. 1351. Beckmann, A.; Boklen, K. D.; Elke, D.; Z. Phys. Chem., 1975. 79. 173. ‘ ' ' ' ' ' 7—__' Ramsey, N. F.; Phys. Rev., 1950. 77. 567. Dawes, S. B.; Ward, D. L.; Dye, J. L., unpublished results, this laboratory.‘ ' ' ‘ ' Dawes. S. B.; Fussa, 0.; Ward, D. L.; Dye, J. L., unpublished results, this laboratory.‘ ' ' . Tzalmona, A.; Andrew, E. R.; Proceedings of the Congress of Magnetic Resonance and Related Phenomena, 1974, Ampere 18th, 241. Behrens, H.2J.; Schnabel, B.; Physica, 1982, 1148, 185. Dawes, S. B.; Ellaboudy, A.; Dye, J. L.; J. Am. Chem. Soc. (companion paper). - - . 171 V.B. Cesium-133 Solid State Nuclear Magnetic Resonance Spectrosc0py of Alkalides and Electrides .V.B.1. INTRODUCTION I Complexed cations, formed from simple alkali metal cations and crown ethers or cryptands, are sufficiently stable to reduction that salts with alkali metal anions (M+C-N‘, known as alkalides) or with electrons at the anionic sites (M+Coe-, known as electrides) can be formed. The first alkalide, Na+C222°Na-, (C222 = cryptand[2.2.2]) was synthesized in 19741 and has been characterized by single crystal x- ray diffraction,2 optical spectroscopy,3’u powder conductivity,5 and more recently by photoelectron emission, and fluorescence spectroscopy.6 The nature of the sodium anion was best characterized by NMR, first in solutions of Na+C222-Na~ in ethylamine,7 THF and methylamine8 and later in the solid state with polycrystalline Na+C2220Na..9 The chemical shift of the anion in each case was identical to the chemical shift calculated for Na- in the gase0us state}0 Thus, the sodium anion was shown to be a genuine spherical anion with a filled 3s orbital that is not strongly affected by its environment. To date more than 40 alkalide and electride compounds have been synthesized, and Magic Angle Sample Spinning NMR (MAS~NMR) has become a primary technique for their characterization. The oxidation state if an ion or element in a compound is easily identified by the chemical shift of the resonance. The cationic and anionic species in an alkalide, which always contains two alkali metal nuclei, can often be properly identified by NMR. Signals have been detected for l72 + Li+-C , Na .0 9’11, N 23 a- 9-11 - 12,13 - 14 , K Rb _ , 0s+-0 and C87 15717 in the solid state. In these expressions, C represents a cryptand such as 0222 or a crown ether, such as 18-crown—6 (18C6) or 15-crown-5 13 (15C5). No signal was detected for either K+C . or Rb+°C 14 due to extreme quadrupolar line broadening that could not be reduced by MAS. In addition to the identification of alkali metal cations and anions in alkalides and electrides by measurement of the chemical shifts, analysis of the line shape, both in static and MAS narrowed spectra, can provide information about the interaction of the ions with their environments. The quadrupole coupling constants have been 23 determined for Na-CZZZTT (e20q\h= 1.2 MHz n = 0.1), Cs+(1806)21.7 (e20q\h= 89.6 KHz n = 0), and Cs+(15C5)212 (e2Qq\h= 427 KHz n = 0), and upper limits for the quadrupole coupling constants have been established for Na- (e20q\h s 0.23 MHz).11 K-(e20q\h s 0.1 MHz1a, and Rb-(e20q\h s 1.3 MHz).1u The relatively small value of the quadrupolar interaction in the anions testifies to their large, diffuse, spherical nature. Some properties of T33CS NMR in alkalides and electrides have 15,16 been previously reported.. The isolation of the electride, Cs+(18C6)2-e~, was verified by 13303 NMR when the compound with a 1:2 ratio of Cs to 1806 gave an NMR spectrum that consisted of a single paramagnetically shifted peak. The compound of stoichiometry Cs(18C6) was shown to be the ceside, Cs+(18C6)2-Cs-, by virtue of a two peak spectrum: Cs+(1806)2 resonated near the value seen for other diamagnetic Cs+(18C6)2 salts, and Cs” was shifted upfield by about 160 ppm. One other cesium comp0und, Cs+(18C6)2-Na‘, has been 133 9 characterized by Cs NMR. 173 V.B.2. EXPERIMENTAL Alkalides and electrides are synthesized by anaerobically introducing stoichiometric amounts of the metal(s) and complexant into an evaCuable vessel. The materials are dissolved in polar amines and/or ethers, and crystals are formed by first lowering the dielectric constant of the solution by adding less polar amines and/or ethers and then cooling the solution. The compounds are very reactive and must be handled under vacuum or in inert atmosphere at all times. Further, most compOunds must be handled at temperatures below ~20°C to avoid thermal decomposition. Details of the purification of materials and of the synthetic methods have been described elsewhere.18’19 NMR spectra were obtained at 47.24 MHz (at Dow Chemical Co., 8.4 Tesla, pulse length = 6.0 us, 105 repetition time) and at 65.61 MHz (at the University of Illinois, 11.7 Tesla, pulse length = 4.5 us, 103 repetition time) with Andrews Beam type MAS probes. The spinning gas was cooled by passing it through a copper coil immersed in an acetone/dry ice bath. Spinning rates were 2 to 4 KHz. The Andrews spinners were made of Delrin or d8-polymethyl methacrylate. The spinners were loosely filled with the polycrystalline sample in an inert (dry N2) atmosphere glove bag. Spectra were also obtained at 23.62 MHz (4.2 Tesla, pulse length = 4.5 us, 2.5 s. repetition time) with a variable temperature Doty probe. Doty rotors were made of ~sapphire and were equipped with low temperature end caps. The rotors were firmly packed with the polycrystalline samples in inert atmosphere bags. Spinning rates were typically 1.5 to 3 KHz. The spinning gas was passed through a liquid nitrogen heat exchanger and 174 warmed to a set temperature with a feedback-controlled heater. Temperatures of the spinning gas were measured with a copper constantan thermocouple inserted in the gas line just before the stator. Further calibration of the temperature is described later. Chemical shifts were measured with respect to a value of 0 ppm for + Cs (aq) at infinite dilution. Observed linewidths are reported as Av1/2, the full width at half height. V.B.3. RESULTS AND DISCUSSION V.B.3.a. 13st MAS~NMR of Alkalides The chemical shifts and linewidths of 13305 ions in alkalide salts, determined by MAS’NMR are summarized in Table 23. The chemical shifts of complexed cations in both the 18C6 and 1505 sandwiched systems are nearly independent of the anion, and agree with the values found for halides and thiocyanatesj7 Cs+(18C6)2 resonates at ~60:2 ppm for all salts except the ceside. The cesium cation is coordinated by 12 crown ether oxygens. The constant chemical shift implies that the structure of the complex does not differ significantly in the sodide, potasside and rubidide salts. The spectrum of Cs+(1806)2-Cs~ is discussed in detail below. Cs+(1505)2 resonates at +25 :1 ppm for both the sodide and potasside, and at 29 ppm20 for the rubidide. Cs+(1505)2oCs- has not been isolated to date. The width of an MAS narrowed line may be due either to second order anisotr0pic effects (i.e., 2nd order quadrupolar broadening) or to incompletely narrowed first order effects. Insufficient spinning speeds, and/or slight deviations of the spinning angle could contribute to residual first order linewidths.21 A trend in the MAS 175 Table 23. Chemical Shifts and Linewidths of T33Cs MAS—NMR Lines in Alkalides and Electrides. Compound 6(ppm)a Av1/2(Hz)b SpectrometerC ‘Cs+(18C6)2~Na~ -61 40 B ~62 160 Cs+(18C6)2-K- -58 85 B ~58 188 Cs+(18C6)2-Rb- -57 115 B ~57 65 g§f(1806)2oc§ ~41, ~49 ~51 175 ~40, ~50 0s+(1806)2-g§f ~212 -238 350 ~210 435 Cs+(15c5)2oNa” +24 130 A Cs+(15C5)2~K- +24 270 - A 0s+(1505)2oRb’ +29 250 A 030222-6’ 138,238 320,140 B 134,240 500,356 c + a)Chemical shift relative to a value of 0 ppm for Cs (aq) at infinite dilution. bAV1/2 is the full width at half height. 0)The spectrum was observed at applied fields of A) 4.2 T B) 8.4 T C) 11.7 T 176 linewidths of Cs complexes is apparent: cesium cations have a broader line for larger counter ions. Second order quadrupolar effects have been shown to be negligible in complexed Cs+ salts, so a syStematic residual linewidth must account for the trend in peakwidth. The dipolar linewidth of Cst from coupling with surrounding hydrogen atoms has been shown to be temperature dependent in Cs+ (crown ether)2 saltsJ7 Rapid motion of hydrogens on the crown ethers averages some of the dipolar anisotropy, and reduces the static linewidth. Larger anions would hinder rapid motion and increase the width of both the static spectrum and the imperfectly narrowed spectrum. No systematic decoupling experiments have been attempted. The only ceside that has been synthesized and fully characterized is Cs+(18C6)2-Cs..22 The structure of this compound has been determined,23 and shows that Cs- is an extremely large anion with a radius of 3.3-3.5 A. C57 is a unique alkali metal anion because low energy d and f orbitals are available. Both the cation and anion are paramagnetically shifted from the pure ground state character of the gaseous ions, and both have temperature dependent values of the chemical shift. At low temperatures (T < 225 K) the resonance position of the cation is ~40 :1 ppm and the anion has a chemical shift of -213 :2 ppm. Between 230 K and 285 K the cationic peak splits into two lines, one at -40 ppm, the other at “50 ppm. The relative intensities of the two lines are approximately 1:1 thr0ugh0ut this temperature range. The anion shifts downfield (from ~212 to -205 ppm) with increasing temperature between 230 K and 285 K. Above 285 K the anion peak is at —240 ppm, and only the -50 ppm peak appears for the cation 9 177 The chemical shift of Cs+ in inclusive crown ether and cryptand complexes is due to the Cs-O overlap, and is inversely related to the mean Cs-O distance}7 The change in paramagnetic shift of Cs+(1806)2oCs- from other Cs+(18C6)2 salts results from a different conformation of the 1806 molecules in the ceside which shortens the 23 The mean C340 distance at T = 225 K from 3.35 A to 3.30A. temperature dependence of the cationic chemical shift indicates that two different environments exist for the complexed cation; between 230 and 285 K they co-exist, perhaps as dimers, and above 285 K only the 50 ppm environment is seen. No structural information exists to describe the ~50 ppm resonance, but the complexed cation probably has a mean Cs-O contact distance shorter than those seen in salts of other anions, and longer than that of the low temperature phase. The chemical shift for Cs- is notable for the strong paramagnetic shift from that calculated for Cs-(g). The chemical shift of other alkali metal anions have been measured in solution and in the solid state; Table 24 shows these resonance positions and compares them to the calculated values for M—(g). It is clear that the paramagnetic deviation from the calculated values of the chemical shift as well as the range of observed shifts increases with increasing atomic weight. The paramagnetic Ramsey shift is inversely pr0portional to the energy required to promote an electron to excited states. Clearly Cs- and R07 have low lying d and f orbitals available, whereas Na- and K- retain more nearly pure 52 configurations. The Cs- shift with temperature suggests that the admixture of excited state character is very sensitive to the local structure about 178 Table 24. Summary of Calculated and Representative Observed Chemical Shifts of Alkali Metal Anions. 0(M ) 0(M ) observed Compound ' ' T(K) Ref. Calculated Solution Solid 13305 ~346.4 25 Cs/C222/THF ~292 ‘ 202 8 + r This Cs (18C6)2-Cs 213 220 Work + - This Cs (18C6)2;Cs 240 285 Work 87Rb -213.6 25 Rb/12/C4 ~191 26 Rb/C222/EA ~185 233 8 Rb/C222/THF ~197 227 8 Cs+(15C5)2{§§ ~189(2) 14 Cs+(18C6g ogpf ~194 ~260 14 39K ~103.4 25 K/Cs/1204 ~94.2 13 K/15C5/Me20 ~101 12. +. - . . . K (15C5)20K ~105(1) 12 Cs+(18C6)2-K7 ~115(10) 12 23Na ~63.4 25 Na/0222/THF ~62.8 269 8 Na/1ZC4 -61.8 26 Na+0222-Na' ~61 ~260 9 Cs+(1806)2-Na” -61 ~260 9 179 the anion. The anion peak shifts gradually downfield by +10 ppm as the temperature is increased from 225 to 285 K. The small paramagnetic shift may be due to increased thermal motion of the crown ethers which reduces the effective radius of the anionic cavity. A smaller cavity favors the more confined electron density of 5d or 4f orbitals over the diffuse 6s orbital. The chemical shift changes by ‘35 ppm above 285 K, probably because of a phase change which increases the effective radius of the anionic cavity. V.B.3.b. MAS-NMR of 03+0222-e' In contrast to the rich variety of alkalide and electride salts that contain sandwich complexes of Cs+ with crown ethers. only two such compounds have been isolated with the Cs+C222 cation. Indeed, until recently, when the structure of a crystal of the ceside, Cs+C222-Cs-, was determined by X-ray diffraction,23 only the 1:1 compound of stoichiometry CsC222 had been isolated, regardless of the solution stoichiometry. Its 133Cs MAS-NMR spectrum, shown in Figure 25, consists of two lines, whose relative intensities depend on the method of preparation of the sample. The two peak spectrum is similar to the inclusive/exclusive complex spectrum seen for 17 Cs+0222~SCN~-H 0.. 2 The absence of the Cs- NMR peak was taken as evidence that the 1:1 salt is an electride. Samples that were prepared from Cs rich mixtures of Cs and 0222 tended to have both peaks, with nearly the same intensities (Figure 25a), while cryptand~ rich starting mixtures and slowly grown crystals gave primarily the diamagnetic peak (Figure 25b). The two peaks are attributed to the presence of both inclusive and exclusive complexes of the cation. It is not clear whether a single crystallite is able 180 a) 1 T f 300 200 100 O 5 (PPM) b) I I T l 400 ZOO O -200 6 (PPM) Figure 25. 133Cs MAS-NMR spectrum of Cs+C222-e‘, showing inclusive and exclusive complexed cations, prepared from solutions that contained excess quantities of a) cesium and b) C222. Other lines are spinning sidebands. 181 to support both types of complexed cations, or whether the polycrystalline sample contains a mixture of crystallites of pure inclusive complexed cations and of pure exclusive complexed cations. Electron-electron interactions in the "electrides" are very strong. A sample taken from a batch with 1:1 stoichiometry was diamagnetic at all temperatures. Spin pairing is not uncommon in electrides; especially in compounds that contain cryptands as complexants. For example, Li0211 displays temperature dependent spin-pairing27 below 50 K. K+C222~e7 and Rb+C222-e- appear to have metallic character and the electronic susceptibility is dominated by a Pauli-type temperature-independent paramagnetism.28 The recent identification of ceside crystals in a preparation that was made in the same manner as the electrides23 indicates the need for caution in interpreting the behavior of polycrystalline samples. The lack of an NMR signal due to Cs- is not proof of its absence. For example, samples known to contain Rb- showed no 87Rb signal.1u Thus, although the 1:1 stoichiometry and absence of a Cs- NMR peak in C50222 strongly suggest that it is an electride, final proof awaits determination of its crystal structure. V.B.3.c. 1330s. MAS-NMR of Cs(18C6).,_-_<_e_- The first reports of the synthesis and characterization of + .- Cs (18C6)2-e .15 emphasized the eleCtronic nature of the compound. + .. Cs (18C6)2-e is a Curie-Weiss paramagnet which shows no magnetic ordering or spin pairing over the temperature range 1.5 to 250 K.22 The d.c. conductivity indicates that the electride is a semiconductor with an intrinsic band-gap of 0.9 eV; however, the conductivity at X- band microwave frequencies gives an apparent band~gap of only ~0.1 182 eV.22 The 133Cs MAS-NMR was reported to consist of only one peak at +81 ppm. This paramagnetic shift from that of other Cs+(1806)2 compounds could result from either a tighter coordination of the Cs+ by the crown ethers, or from contact with the paramagnetic electron. NMR can be used to probe the interactions of the trapped electron with the complexed cations in the solid. Structural information now exists for Cs+(18C6)2-e.;29 30 the electride is nearly isostructural with Cs+(1806)2-Na-; the mean Cs-O distance is 3.35 A in each compound. The electride features a cavity at anionic sites in the lattice with a very low electron density. The origin of the paramagnetic chemical shift in Cs+(18C6)2.e~ is the Knight (or contact) shift. The Knight shift results from the strong local magnetic field generated at the nucleus of interest by paramagnetic electron density. The lowest energy unoccupied orbital in Cs+ is the 6s orbital, which has a non~zero electron density at the nucleus, so that the cation should be very sensitive to any overlap with the paramagnetic trapped electron. The Knight shift is given by31 = 8n 3NAV K(T) < 11(0112> x(T) <1) where <1w(0)|2> is the average electron density at the nucleus, and x(T) is the magnetic susceptibility of the paramagnetic species. For a Curie-Weiss paramagnet, K(T) is proportional to 1/T, since X(T) varies inversely with temperature. Preliminary attempts to measure the chemical shift as a function of temperature revealed a strong variation of the chemical shift with temperature that was approximately proportional to 1/T. Instrumental temperature control 183 and measurement were poor, however; resulting in inconsistent values from one determination to another. Measurement of the temperature of the spinning gas prior to its entry into the stator suffers from poor contact of the thermocouple with the gas so that poorly controlled parameters (i.e. flow rate, heat leaks) affect the temperature reading. A different method, devised by English,32 was therefore used. This method takes advantage of the known temperature dependence of the splitting of the chemical shifts of hydroxyl and methyl protons in methanol.33 Over the range 220 K to 330 K the splitting, Av, of the two peaks.at 180 MHz may be used to calculate the temperature, T, by T = 435.5 * 0.398 Av - 3.25 x 10‘“ Av2. The method used to calculate the temperature of an electride sample was to alternately measure the proton splitting in methanol and the chemical shift of the electride at constant gas flow rates, and gas heater power. A value of the chemical shift of the electride was assumed to be at a known temperature if two alternate measurements of the proton splitting and the chemical shift remained constant. Four points measured in this way are shown in Figure 26. The chemical shift (in ppm) fits the equation 4.06 x10)-l 6(T) = 6(w) + K(T) = ~61 + T The value of the chemical shift at infinite temperature [6(a) = ~61 - + _ppmJ is nearly identical to the chemical shift of Cs (18C6)2 seen in alkalides. This reSult correlates well with the structural data, since the complexed cations in Cs+(18C6)2°e— and Cs+(18C6)2-Na- have nearly identical structures. 184 1 oaamhpmeoc 0903 mucaom 90:90 .mueuom ouaunuflau 9:60 6:0 sweeps» on“. gum 0666 6:0 so 6.. on e . . .Hocmnpwa CH mcou0Hm mo mcflpuflamm Hmfiomfip 0:0 no“: commemmEoo an nopmnaflfimo 0903 x xn poxpme mpcaom mo.m0H3umH0m50u one .10.NmoUmfiv+mu Mom Hue mamao> amasm Hmcflaono 0:9 .cm cpswflm $03 A: o... no 6.... m... 6.... ohm on» «w m u o? on- or. o. .2» l LKLL L 12 8 .225 On: om. Lo... LL Lom. om. com l__L -o- movu l85 The fraction of atomic s~orbital character of the trapped electron may be calculated from the magnetic susceptibility and the slope of the Knight shift verSus 1/T. The electronic part of the + . n- magnetic susceptibility of Cs (18C6)2°e at 250°K is x: = 1.37 x 10 5 emu/mole. The fractional atomic character is defined as 2 (1001 > F = 2 9 5019 <11’01 >atom . 2 where (lwol >atom’ the electron density at the nucleus for an isolated gas atom, has been estimated to be 2.645 x 1025 e-cm73.3u The data given above yield F = 3.3 x 10—14 for Cs+(18C6)2-e-. Table 25 compares this value with that of other paramagnetic cesium-containing systems. 35 Lelieur and Rigny measured both the magnetic susceptibilities and the Knight3b shifts of ammonia solutions and f0und that solutions containing 2 to 20 mole percent Cs (MPM) had fractional atomic characters ranging from 0.04 to 0.60. The overall electron density in the electride is 1.1 x 1021 e./cm3 at 216 K which is approximately the same as that of a 1.5 MPM solution of Cs in ammonia. Yet the contact density at the metal nucleus in the electride is much smaller than in cesium solutions in ammonia, amines and other solvents. Metal ions and solvated electrons in ammonia form loosely bound ion-aggregates in the intermediate range of metal concentrations betwen 10“2 MPM and 10..3 MPM, the majority of which are spin~paired and undetectable by magnetic resonance methods.37 The paramagnetic species that do exist are probably 'solvent-shared' ion-pairs, which consist of a solvated electron and a cation, each with their complete solvation shells.37 The spin denSity at the nucleus in metal ammonia solutions between 37 3 ..2 .. 10 and 1 MPM is nearly constant with F 2 2 x 10 , nearly an order 186 of magnitude higher than in the crystalline electride. The effect of reducing the dielectric constant of the medium is to create stronger interionic interactions, mediated by weaker solvent interactions so that paramagnetic ion pairs have more excess electron density at the nucleus as indicated in Table 25 for solutions of cesium in methylamine (MA), ethylamine (EA) and isopropyl amine (iPA). The contact interaction in frozen solutions of Cs in the highly polar solvent HMPA indicates a different type of excess electron character.39 Two EPR signals are detected in these solids, one with high contact density at the Cs nucleus (F = 0.73) attributed to solvated cesium atom and the other with a much lower contact density (F = 0.01).3u The electron in the low atomic character species was presumed to occupy a large hydrogenic wavefunction centered on the screened solvated Cs cation, but having very little Cs orbital character.314 The trapped electron in Cs+(18C6)2-e- has exceptionally low cesium s~orbital character relative to other electron-cesium systems. The cesium cationic charge is apparently very well screened by its interaction with the crown ether oxygen lone pairs. A monomeric species, in a state similar to that described as the low s~orbital species of Cs/HMPA solutions,3u might be invoked for the electride. The monomer in this case would have one electron in an extended orbital centered on each complexed cation. A major problem with this description of the crystal as a collection of "expanded atoms" is the observed weak electron-electron interactions. An alternative and better explanation of the low value of F is that the paramagnetic electron density is largely located in the anionic cavities, localized 187 Table 25. Fractional Atomic Character of Some Cs-o-oe‘ Solutions and Solids. Compound electron density 030(3) 2.645 x 1025 Cs (metal) 1.56 x 1025 CS:NH 3 15 MPM 2.12 X 102“ 5 MPM 7.94 x 1023 CszflMPA 25 Solvated Atom 1.93 x 10 . 22 Expanded orbital 6.08 x 10 a 24 CszMA 273 K 7.42 x 10 Cs:EAa273 K 5.03 x 102” Cs:iPAa273 K 7.14 x 102” Cs(1806)2-e- 8.75 x 1021 1.00 0.59 0.08 0.03 0.27 3.3 x 10’” aAbbreviations used: HMPA ~ hexamethyl phosphoramide MA ~ methylamine, EA ~ ethylamine, iPA ~ isopropylamine bChemical shift from Reference 36. Cs metal were calculated from X metal‘ ref. 34 36 36 34 34 34 34 34 this work Paramagnetic susceptibilities of , values from Reference 40. 188 by the coulombic attraction to eight nearest complexed cesium cations. The electron density at each cesium nucleus would then be the sum of the weak interactions with the eight nearest trapped electrons. Studies of electrides doped with sodium anions, described below, favor the latter interpretation of the electronic state in Cs+(1806)2°e- The chemical shift data shown in Figure 26 include a number of points at temperatures that were not calibrated as described above. Rather, the temperature corresponding to each data point was estimated from the peak position so that the plot of 6(T) vs 1/T was constrained to be linear. At temperatures below 225 K the spectrum became more complex, in that a second peak, paramagnetically shifted from the original, appeared as a shoulder at ~225 K and increased in relative intensity as the temperature was decreased. The two peaks are separated by about 20 ppm, and show a parallel temperature dependence. The increased paramagnetism does not necessarily signal increased electron density at the nucleus, but may instead be a meaSure of decreased Cs-O distances in the complexed cation. Extrapolation of the paramagnetic peak to infinite temperature gives a chemical shift f ~40 ppm, the same as that seen in low temperature Cs+(1806)2°Cs- salts. Apparently, at low temperatures, the complexed cation gradually shifts to a second conformation with a reduced mean Cs~0 distance. The transition does not appear to be first order, and does not go to completion to the lowest temperatures measured. 'V.B.3.d. 13305 MAS-NMR of Mixed Alkalide-Electride Salts. The similar nature of the complexed cation in various electrides and alkalides suggests that crystals of mixed stoichiometry might be prepared. Reduction of the trapped electron density by 189 partial substitution of diamagnetic anions for anionic trapped electrons could significantly alter electron-electron interactions in these materials. The preparation of mixed alkalide/electride materials was attempted by using two different alkali metal anions, Cs- and Na‘, in the presence of Cs+(1806)2 complexed cations, for several alkalide 133 mole fractions, x. Figure 27 shows the Cs MAS-NMR spectrum for a sample of overall stoichiometry Cs+(18C6)2oC The $1.5) ' 9(5): spectrum consists of three lines together with their spinning sidebands. The resonance positions are clearly those of the pure ceside (6 = ~50 and ~230 ppm) and the pure electride (6 = 77 ppm at 294 K). Several other samples resulted in similar spectra with only the relative intensities of the lines varying. These results demonstrate that doping electrons into the ceside lattice, or cesides into the electride lattice is clearly unfavorable. The anionic radius of each of these species has been determined,29 the radius of the anionic cavity of the trapped electron in Cs+(18C6)2oe- is 2.2 A whereas 037 has a radius of ~3.3 A in Cs+(18C6)2-Cs-. Lattices which are able to support each anion separately are, in this case, unable to co~exist. Mixtures of pure electrides and pure alkalides were useful for establishing the temperature characteristics of the alkalide based on the known temperature dependence of the electride peak. The 133Cs MAS-NMR spectra of samples with overall stoichiometry 1~x are shown in Figure 28, for x = 0.2 and 0.8 + .- Cs (18C6)2-Naxe respectively. These spectra consist of five lines at ~61, +26, +42, - + +57 and +73 ppm. The extreme peaks are due to Cs (18C6)2 in the pure sodide (~61 ppm) and in the pure electride (73 ppm). The three 190 . . . anamom Ho;uo .oocm:0mos 1o.mmoumfiv+mu 0 ma Ema um um xmom 02p .moocmCOmoe Nnoowfiv+mu one 5mm 0mm- paw 5mm om- us 62662 .mmouwmv mu knuoaoHAUfiOpm Hmfipficfi scam pocfimuao uosvoym esp mo sapwoomm «221m<2 mummH .nm osswfim .338 52. 52. saw 855 com. 08 - . 00.1 on... o _ a A On 00. OON - — q 191 G l 1 L l 1 4L 1 I 1 100 50 0 -50 100 PPM b ___I. I I l l I I I I ll L. .1 100 50 0 ~ 50 -lOO PPM Figure 28. 133C5 MAS—NMR spectra of compounds formed from initial stoichiometry Cs+(18C6)2'NaX-el_X for a) x =0.2 and b) x =0.8. Peaks at 73, 57, 42 and 25 ppm are from a mixed compound with X =O.l~0.2; the peak at -61 ppm is from excess Cs+(l8C6)2-Na'. Other features are spinning sidebands. 192 intermediate peaks are Successively spaced by roughly 1/8 of the spectral breadth between the electride peak and the sodide positions. + _ Structural determinations of Cs (1806)2oe 29 30 and Cs+(18C6)2~Na- indicate that each complexed cation has eight nearest neighbor anions. The four paramagnetically shifted peaks may thus be attributed to a complexed cation surrounded by eight, seven, six and five electrons in order of increasing field; the other anionic sites are occupied by diamagnetic Na- ions. There does not appear to be any tendency for appreciable doping of the sodide lattice with electrons. The relative intensities of the four paramagnetic peaks remains nearly constant as the overall mole fraction of Na- is increased; only the pure sodide peak at ~61 ppm increases markedly. The relative. intensities of the four peaks are roughly 1:1:0.5:0.25. The constancy of the relative intensities Suggests that a compound with an ordered structure may be formed, although random substitution of Na‘ into the electride lattice up to saturation cannot be ruled out. A superlattice that includes ordered substitution of sodium anions into anionic sites in Cs+(18C6)20eg lattice can be postulated that is consistent with the known crystal structures of the pure compounds and gives approximate agreement with the mixed sodide/electride NMR spectra. The crystal structures of Cs+(18C6)2-e7 and Cs+(18C6)2-Na- are nearly identical.29’3O In the space group C2/c, the parameters of the unit cell of the electride are 2 = 13.075 A,.§ = 15.840 A,_g = 17.359 A, and B = 92.30°. The sodide has unit cell parameters _2: 13.581, 2_= 15.684, and 2.: 17.429 A, B = 93.16°. The anionic cavity in the sooide is larger in the 2:2 plane than is the anionic cavity in the electride, resulting in interanionic 193 distances that are 0.1 A longer in §j2_planes, (10.37 A in the sodide, 10.27 A in the electride). The cavities in each structure are separated by ~8.7 A in the 9 direction. The effect of inserting a sodium anion into the cavity in the electride lattice might be to distort the eight nearest complexed cations away from their equilibrium positions. The cavity sizes suggest that the distortion would have primarily an 2:2 component so that anionic sites along the .g~directionfrom the first impurity center should be better able to include a sodium anion, leading to an infinite chain of Na- ions along the gfdirection. A third sodium anion may be substituted into only one of the two sets of equivalent neighboring anionic sites, either along the gfaxis or the b~axis of the original unit cell, say at {3. In order to have cesium cations that are surrounded by three Na- ions, neighboring cavities in the 3 direction must be empty, allowing occupation of only every other 2:2 plane. Then a Na- ion will occupy a site at ~a in those planes which have no Na- sodide ion at £3. Additional substitution may be limited by strain in the crystal. The anions in one such superlattice unit cell are pictured in Figure 29. The volume of one mixed anion unit cell is equal to the volume of eight original unit cells, and requires that 4 of the 32 anionic sites be filled with sodium anions. In agreement with this, the intensity of the peaks in the sample with x = 0.2 showed that the sodium mole fraction in the mixed compound is less than 0.2. The model predicts 'relative ratios of complexed cations with 0, 1, 2, and 3 of the eight nearest anionic sites filled by Na” to be 3:3:1:1, in good agreement with the observed line intensities. Regardless of the detailed structure of the mixed system, these results provide strong evidence 194 m .mcowcm Esfleom ope moaoswo eflfiom .mopflm :oappofic eommmap ohm modoafioommmm .mhpoomm mzz mummH opfleom1oefiauooao.eoxwe eo>aomno can you pcsooom fiasco umnu.0esu smH mewspooao\0eflpom poxwe poowfisumom m mnwzosm Empmwfiw mcflxomm ofi=0ficm :< mm 09 .m 195 + that each complexed Cs interacts with eight nearest neighbor electrons, not just with one of them. The temperature dependence of one sample of Cs+(18C6)2-NaX-e1_x -was meaSured over a temperature range of 190 K to 255 K. The spectrum was very noisy so that uncertainties in the chemical shift are as large as 15 ppm. The most paramagnetic peak was assumed to be the resonance of the complexed cation surrOunded by eight trapped electrons (or the 'pure' electride). The temperature of the sample was calculated from the chemical shift of the pure electride peak by using the established Knight shift parameters. The temperature .1. .. .. dependence of the peaks of Cs (1806)2-NaX-e .for x = 1/8, 2/8 and 1~x . 3/8 appeared to be nearly linear, and were fit to a Knight shift equation 6(T) = 6(a) + m/T for two cases: 1) 6(W) = -61ppm, since the chemical shift of both the pure sodide and the pure electride at infinite temperature is ~61 ppm, and 2) for a variable 6(«). The reSults are tabulated in Table 26. The slope is a measure of the paramagnetic electron density at the cesium nucleus. The peak that corresponds to the doped electride with x = 0.125 has a slope ~ 85% of that of the pure electride; the next peak, with x = 0.25 continues the trend with a slope ~70% of that of the pure electride. In the case of the least diamagnetic peak, with x = 0.375, it was difficult to determine acourate values of the chemical shift. This gave the least consistent values of the parameters of the temperature dependence of the chemical shift; the slope indicates that the electron density at 196 Table 26. Table of Least-Squares Best-Fit Parameters for the ' Temperature Dependence of the Knight Shift of Peaks in Cs (18C6)2-Nax°e 0.000 0.125 0.250 0.375C 1-x. 6(°°)(ppm)a 259 ~61 ~66 ~61 ‘71 -61 ~118 ~61 i 1 (Fixed) :7 (Fixed) i 11 (Fixed) 1 13 (Fixed) m(10uppm K)b 0.003 4.05 i 0.02 4.091 3.56 i 0.16 3.44 i 0.01 2.85 i 0.02 2.63 i 0.02 3.05 1 0.30 1.74 i 0.06 % electride character 99 100 87 84 70 64 77 43 a6(°°) is the infinite temperature extrapolation of 6(T). i.e. where K(T) 0. b 0The peak corresponding to x 0.375 was very weak; m is the slope of the best fit line for 6 versus 1/T. the large spread in 6(a) and m is probably a result of large errors in measuring the chemical shift of the peak. 197 4. CS for x = 0.375 is between 45% and 75% of that of the pure electride. V.B.4. ' CONCLUSIONS i33Cs MAS-NMR is a powerful tool for the elucidation of cationic and anionic species in alkalide salts. The use of multinuclear metal NMR techniques can generally be used to fully characterize the stoichiometry of an electride or an alkalide. Moreover, the resonance position in 13303 NMR may be used as a sensitive probe of local structure about the cesium ion, and for the study of interionic interactions. ‘Changes in the Cs-O distances in cesium—crown ether or cryptand complexes brought about by phase transitions or by changes in the anionic constituent, or the temperature, can be monitored by the empirical observation that the paramagnetic Ramsey shift has an inverse relationship to rCs-Of The Cs- ion is also sensitive to the size of its anionic cavity. The presence of low lying excited states apparently causes the resonance position of Cs- to be paramagnetically shifted from the pure s-state anion. This Ramsey shift is stronger for cavities which are smaller, or which favor a non-spherical anion. Finally, the NMR data, in combination with structural data on Cs+(18C6)2~e~, may be used to define the nature of the trapped electron. The structure clearly indicates that anionic cavities exist in the lattice, in a distribution that is nearly identical to the distribution of Na- sites in the salt Cs+(1806)2-Na-, in which the electron could be localized by three dimensional coulombic attractions to the cations. The trapped electrons interact only very weakly with 198 each other, as evidenced by the absence of any spin pairing, or magnetic ordering. They also show only weak overlap with the complexed cations, as evidenced by the very small Knight shift. Further, the interaction of each electron with nearby cations is spread over all of the cations, and not just one of them, as shown by the mixed alkalide electride work. The electride described here is only one of a growing class of compounds which display rather diverse properties. The compound Cs+C222-e“, for example, is a spin~paired system that shows no paramagnetism at any temperature. The strength of electron-electron and electron cation interactions in other electride systems is of great interest. Undoubtedly solid state NMR will play a major role in measuring the nature of these interactions. V.B.5. ACKNOWLEDGMENT This research was supported by National Science Foundation-—Solid State Chemistry Grants DMR 79-21979 and DMR 84~ 14154. We thank Dr. P. B. Smith of the Dow Chemical Company, and Dr. E. Oldfield and Mr. B. Montez at the University of Illinois at Urbana~ Champaign for their assistance in obtaining some of the NMR spectra. We are grateful to Dr. M. Tinkham for help in preparing samples and measuring spectra. V.B.6. REFERENCES 10. 110 12. 13. 14. 15. 16. 17. 18. 19. 20. 199 Dye, J. L.; Ceraso, J. M.; Lok, M. T.; Barnett, B. L.; Tehan, F. J.; J. Am; Chem. Soc. 1974. 96. 608-' ‘ Tehan, F. J., Barnett, B. L.; Dye, J. L.; J. Am. Chem. Soc. 1974, 96. 72034 ' ' " ' " Dye, J. L.; Yemen, M. R.; DaGue, M. G.; Lehn, J.~M.; J. Chem. Phys. 1978, 68, 1665. ‘ ' ' ' ’ Le, L. D.; Issa, D.; Van Eck, B.; Dye, J. L.; J. Phys. Chem. 1982, 86, 7. ' ' ‘ ' ‘ ' Dye, J. L.; Angewante Chem., Intl. Eng. Ed., 1979. 18. 587. Jaenicke, S.; Dye, J. L.; J. Solid State Chem. 1984, 54, 320. Ceraso, J. M.; Dye, J. L.; J. Chem. Phys. 1974, 61, 1985. Dye, J. L.; Andrews, 0. W.; Ceraso, J. M.; J. Phys. Chem. 1975, 97. 3076.‘ ' ' ~ ' ' ‘ ‘ Ellaboudy, A.; Tinkham, M. L.; Van Eck, B.; Dye, J. L.; Smith, P. 3:;Phys. Chem. 1222, 88, 3852. ’ Dye, J. L.; Prog. Inorg. Chem. 1222, 32, 327. EllabOudy, A.; Dye, J. L.; J. Mag. Res. 1222, Tinkham, M. L.; Dye, J. L.; J. Am. Chem. Soc. 1222, 107, 6129. Edwards, P. P.; EllabOudy, A. S.; Holton, D. M.; Nature (London) 1122;, 317, 242. ‘ - Tinkham, M. L.; Ellaboudy, A.; Dye, J. L.; Smith, P. B.; J. Phys. Chem. 1222, 90, 14. ' ‘ - ~ . . Ellaboudy, A.; Dye, J. L.; Smith, P. B.; J. Am. Chem. Soc. 1983, 105. 6490. ‘ ' ' ' ' ' Dye, J. L.; Ellaboudy, A.; Chem. Brit. 1984, 20, 210. Dawes, S. B., Ellaboudy, A., Dye, J. L.; This issue. Van Eck, B.; Le, L. D.; Issa, D.; Dye, J. L.; Inorg. Chem. 1982, 21, 1966. ' ' ‘ ' ‘ ' ' Dye, J. L.; J. Phys. Chem. 1984, 88, 3842. Tinkham, M. L.; Ph. D. Dissertation, Michigan State University, 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33- 34. 35. 36. 37. 38. 39. 40. 200 Andrew, E. R.; Intl. Rev. Phys. Chem. 1981, 1, 195. Issa, D.; Ellaboudy, A.; Janakiraman, R.; Dye, J. L.; J. Phys. Chem. 1984, 88, 3847. ' ' Huang, R. H.; Ward, D. L.; Dye, J. L.; unpublished results, this laboratory.' ' ' ’ ' Beckmann, A.; Boklen, K. D.; Elbe, D.; Z. Phys. 1974, 270, 173. Pyper, N. C., Edwards, P. P.; J. Am. Chem. Soc. 1986, 108, 78. Holton, D. M.; Edwards, P. P.; Johnson, D. C.; Page, C. S.; McFarlane, W.; Wood, B.; J. Chem. Soc., Chem. Commun. 1985, 740. Landers, J. S.; Dye, J. L.; Stacy, A.; Sienko, M. J.; J. PhyS. Chem. 1981, 85, 1096. ’ - . . . Faber, M. K.; Dye, J. L.; unpublished results, this laboratory. Dawes, S. 8.: Ward, D. L.; Huang, R. H.;.Dye, J. L.; J. Am. Chem. Soc., ’ ’ ' ‘ ' ' ' ' Dawes, Fussa, Ward, Dye, unpublished results, this laboratory. Knight, W. D.; Phys. Rev. 1222, 76, 1259. English, A. D.; J. Mag. Res. 1222, 57, 491. Van Geet, A. L.; Anal. Chem. 1222, 40, 2227. Edwards, P. P.; Adv. Inorg. Chem. Rad. 1982, 25, 135. Lelieur, J. P.; Rigny, P.; J. Chem. Phys. 1973. 59. 1142. Lelieur, J. P.; Rigny, P.; J. Chem. Phys. 1212, 59, 1148. Dye, J. L.; Pure Appl. Chem. 1211, 49, 3. O'Reilly, D. E.; J. Chem. Phys. 1222, 41, 3729. Catterall, R.; Edwards, P. P.; Adv. Md. Relaxation Processes, 1978. 13. 123. CRC Handbook of Chemistry and Physics, 60th edition, R. C. Weast ed. , CRC Press, Boca Raton, Florida (1980). 201 v.0 MAS-NMR of Cs+(1505)2-e- 133Cs NMR spectra were measured for Cs+(1505)2-e_ over a temperature range of 170 K to 280 K. The chemical shift is temperature dependent and paramagnetically shifted from the chenical shifts of diamagnetic salts of Cs+(1505)2. The temperature dependence of the chemical shift is a linear function of the reciprocal temperature below 230 K, but deviates from a linear Knight shift near the tempera- tures corresponding to phase transitions (238 K and 268 K). The sample temperatures in the NMR experiment were cali- brated by two methods. The splitting of the chemical shifts of the hydroxyl and methyl protons in methanol, which has a known temperature dependence [91,92], was measured both prior to and after measurement of the cesiun chemical shift at four different values of cooled gas flow rates and gas heater power settings. The chemical shifts of Cs+ in Cs+(1505), plotted as x’s in Figure g9, were not entirely reliable however. The rotor that contained the electride was repeatedly renoved fron the stator and was subjected to temperature extremes that resulted in some decomposition. The 13303 spectra were very broad, even with good spinning rates, so that the error in the chemical shift was quite large. For example, the spectrum at 230 K 202 .uo.NmoumHV-mu mo uwfism pamflsx one mo unoEoHSmon msoocmuaseflm kn consanflamo oaoz.mu:flom Hocpo mo magnumaomEou 0:5 .Hocwnpoa mo wcflppwamm amfiomww esp Ape: comflammEoo xn woumanfifimo ohms x an noxsma munflom mo moazpmaomEou oak .:o.mmmUmav+mu How Hue mamao> pmflnm Hmofiaoso och .om oaswflm .-x $.03 .L Ow mm On m¢ O? 0% . Omfim OQN OOn ONn OVM own EQQ®N+ «AOOquw iOQn .OOV LON? 40¢? 100? 2:35 4000 ONO .Ocn lawn Own 000 Cum 10¢0 . --————---— L I J l L J_ g J -----—-------—--- L 203 had a broad peak with a maximum that could lie between 524 ppm and 610 ppm; the shift was chosen to be 524 ppm. In addition, the two high temperature peaks displayed chemical shifts that have not otherwise been observed in Cs+(15C5)2'e_. The spectra at 270 K (8 = 441 ppm) and at 242 K (8 = 487 ppm) had unique paramagnetic chemical shifts. In addition these two spectra had rather large peaks at +25 ppm, probably from decomposed electride. The chemical shifts of the four points that had consistent temperatures throughout two In acquisitions and one 1330s acquisition are shown in Figure g9 as a function of l/T, with their estimated errors. The best fit line gave an extrapolation for the chemical shift at infinite tempera- ture at the y-intercept of 5(a) = 26.2 +/—65.4 ppm and a slope of 1.11 +/—0.15 x 105 ppm K. The temperature dependence of the 133Ge NMR line of Cs+(l505)2-e‘ was also measured by a second method. The spectrum of a sample that contained both Cs+(1808)2-e- snd Cs+(1505)2-e~ was measured at several gas flow and heater settings and the correct temperature was determined from the position of the Cs+(1806)2-e- peak. The points measured in this way are represented by solid circles in Figure 29. The Cs+(1505)2-e- chemical shifts between 190 K and 230 K agree well with the values of chemical shift calibrated by the methanol method. The parameters of the 204 best-fit lines include various combinations of the data calibrated by each of the two methods. A satisfactory fit 5 was obtained with-8(a) = 26 ppm and slope = l.l x 10 . Between 230 K and 270 K the shift deviates paramagnetically from the line. In addition, a peak that is shifted upfield by about 190 ppm appeared at 265 K; this peak rapidly became the major feature in the spectrum. It appears that the phase transitions at 238 K and 268 K affect the electron density at the nucleus and therefore affect the temperature dependence of the chemical shift. Data presented in Chapter VI, however, indicate that there is no significant change in the magnetic susceptibility of Cs+(l565)2-e— in this range of temperatures. The phase transition at 238 K is weak, and may be similar in nature to the transition in Cs+(1806)2-Na- that resulted in motionally averaged dipolar interactions in the static Na— NMR spectrum. Hydrogen motion in the complexed cation may gradually allow more electron density to reside on the cesium nucleus. The phase transition at 268 K reduces the electron density at the nucleus dramatically. The transi- tion from the downfield peak to the upfield peak is complete by about 276 K. This phase transition is also responsible for the shift of the maximum of the optical absorption spectrum to higher energies. Thus the interpre- tation of this phase transition is that the lattice softens 205 and the electron is able to reorient its environment to increase its localization energy. The interpretation is consistent with the observed NMR data; an electron that is more strongly localized at the cavity will have a smaller density on surrounding ions. Unfortunately, the sample also begins to decompose near these temperatures so that a complete temperature dependence of the spectrum of Cs+(1505)2-e- above 270 K could not be obtained. The deviation of the chemical shifts above 230 K in the two temperature calibration methods may be caused by the partial decomposition of the sample that was calibrated by the methanol method. If the decomposition destroyed the crystallinity of the remaining electride it is possible that the phase transitions of the crystalline material might not be present. The cations that are observed at shifts greater than 400 ppm in the NMR spectrum do not appear to have lost any nearest neighbor trapped electrons to decomposition since there is no abrupt Change in the paramagnetic shift. The slope of the chemical shift versus l/T plot below 230 K may be used to calculate the paramagnetic electron density at the nucleus via Equation 5.16. The magnetic ' susceptibility was measured as described in Chapter 6. The fractional atomic character is about 8.4 x 10-4, which corresponds to an electron density at the cesium nucleus, 206 22 e_/cm3. An approximate fractional <|¢(o)|2> = 2.2 x 10 atomic character of the Cs+ in the high temperature phase, F may be calculated if it is assumed that 6(-) = 25 ppm HT’ for this peak. The Knight shift is 275 ppm at 270 K in the high temperature phase, compared to the 430 ppm shift calculated for the low temperature phase (calibration with methanol). If the magnetic susceptibility is linear in T—1 then FHT = (275/430)-8.4 x 10’4 = 5.4 x 10'4. This is still nearly two times greater than the value of F for Cs+(1806)2-e_. The electron density at the cesium nucleus is approximately three times greater in Cs+(1505)2-e_ than in Cs+(1806)20e_, which may be attributed to the structure of the complexed cations. Whereas the crown ethers are in van der Waals contact with each other in Cs+(1806)2-e_, thus shielding the cesium cation from interactions with its environment, the cesium cation in Cs+(1505)2-e— is less enclosed by the smaller hole of the two 15C5 molecules. The result is that the Cs+(15C5)2-e_ exposes a "harder" charge to the trapped electrons and thus may attract electron density to the cation. The electronic wavefunc- tion in Cs+(15C5)2-e- is probably more extended than is the trapped electron wavefunction in Cs+(18C6)2-e_ CHAPTER VI MAGNETIC SUSCBPTIBILITY or c.(1sce)2-e‘ and os+(1505)2-e‘ The trapped electron in electride salts has a spin angular momentum which responds to an applied magnetic field in various detectable ways depending on its interac~ tions with nearby electrons. The extent of electron— electron interactions in an electride salt can be measured by static magnetic susceptibility. Examples of metallic (K+0222-e— [93]) electrides, localized spin paired electrides (Cs+0222'e- [71]), and paramagnetic electrides have already been observed. This work will explore the electron-electron interaction of the two closely related electrides Cs+(1806)2-e_ and Cs+(1505)2-e-. The subtle structural differences in these two salts result in quite dramatic magnetic differences between them. 208 VI.A Background for Magnetic Susceptibility VI.A.l Metallic Paramagnetism Since magnetic susceptibility is an important probe in the study of electrides, a brief description of the theory of magnetic susceptibility will be given. The discussion follows several sources, particularly the solid state physics texts by Kittel [94], and Ashcroft and Merlin [95], and the magnetism texts by Morrish [96], Smart [97], and Hattie [98]. If we neglect the contribution of nuclear angular momentum to the bulk magnetic susceptibility, the behavior of the material in question in an applied magnetic field, H, will depend on the quantum mechanical state of the electrons. (Neglect of the nuclear contribution is a good assumption in almost all cases since the nuclear magneto- gyric ratio is nearly 2000 times smaller than that of the electron.) As is the case in almost any physical measure- ment, electrons which lie in a conduction band; that is, free or nearly free electrons behave much differently than do localized electrons. The magnetic susceptibility displayed by conduction electrons is a feeble temperature independent paramagnetism which requires simple band theory to explain. In a macro- 209 scopic metal, those electrons that lie in a conduction band can be shown to have a density of electronic states g(R), at an energy E that is proportional to Bl/a.' The Pauli exclusion principle states that each spatial energy level may be doubly occupied by electrons having opposing spin angular momentum. The population of these states as a function of the temperature is given by 13(3) = F(B)g(E)dE (6.1) where F(E) is the Fermi Dirac distribution function which may be derived from simple statistics of fermions and is given by +[(B-E )/kT)] -1 mg) = [e f +1] (6.2) The Fermi energy Bf is defined as the value of B for which F(R) ‘ 1/2. At T = 0 K the Fermi function has a value of unity for all energies below Bf, and zero for all energies above Rf as depicted in Figure 31a. Increased temperature effects the distribution of states only slightly as Bf is typically on the order of several eVs and room temperature corresponds to a thermal energy of only 1/40 eV. Because each state is occupied by two electrons of opposing spins, 210 a) N(E) 227w b) N(E) 7<2€ 21;” I e C) ME) / - ‘QQL ZLLH Figure 31. The density of states with Spin up (i) and spin down (1) in a metal as a function of energy, E, a) at equilibrium in the absence of an external magnetic field, b) prior to equilibrium upon applica- tion of a field, H, and c) at equilibrium in the presence of the applied field [94]. 211 it is possible to construct a density of states for separate spin states, arbitrarily assigned as parallel and antiparallel. The density of each spin state is one-half the full density of states in the absence of an applied magnetic field. Application of a magnetic field, H, lowers the energy of the distribution of electrons assigned as parallel to the field by -ngmaH where g is the electronic g factor, m8 is the electronic spin angular momentum quantum number, and ”a is the electronic Bohr magneton. Typically g = 2 and -s = 1/2 so that 43 = ~pBH. Likewise the energies of the states antiparallel to the field direc— tion are raised in energy by AB = pBH, so that before any equilibration is allowed the distribution of states might look like Figure 212, The high energy states in the anti- parallel states spill over (Figure 21g) to unoccupied parallel states to give a final distribution of spin states which has net spin angular momentum in the direction of the field. The magnetic moment of the metal is proportional to twice the number of electrons which spilled over from anti- parallel states to parallel states, N, and is calculated by M = 2pB(AN). afl is estimated by the number of electrons within pBH of the Fermi Energy at 0 K, and can be calcu- lated to be an = {2(2m)3/2/a3)}p3n(af<0)>1(2 (6.3) ...1, The magnetic susceptibility is given as x = dM/dH = 3(N/V)p:/2kTF (6.4) where Ef(0) has been expressed in fundamental constants. The important points to notice are that there is virtually no temperature dependence, and that for typical conduction electron densities near lOzze-h/cm3 the value of X = ~10-6/cm3. In addition to the paramagnetism that arises from the orientation of electronic spin angular momentum, there is a diamagnetic contribution that arises from the induced translational or orbital motion of the free electron gas. The oscillations of the electron gas have a very complex dependence on the applied field, so that the Fermi Dirac statistics become very difficult to analyze for the theoretical value of conduction electron diamagnetism. The calculation will not be done here, only the remarkable result will be stated that the diamagnetic susceptibility is -1/3 of the Pauli spin susceptibility. Thus the total contribution to the bulk susceptibility of an electron gas is 2/a of the total spin susceptibility and is relatively temperature independent throughout the range of solid temperatures. 213 The situation is complicated when band structure arises from d or f orbitals and/or there is increased electron localization. Conduction electrons can interact with the metallic ions and mediate magnetic ordering. In addition, as the conduction electron density decreases, the electrons will follow Boltzmann statistics rather than the high density-low temperature Fermi Dirac statistics and magnetism of a stronger nature will occur. The predicted complexity of the magnetic susceptibility near the transi- tion from the nearly free electron to non-metallic regimes has been seen in metal amine and methylamine compounds. For example Li(MeNHz)4 shows the behavior of a highly correlated metal with a magnetic moment which varies inversely with temperature [22]. As new electrides are synthesized with more nearly expanded metallic character, the discussion of conduction electron paramagnetism will become more important. VI.A.Z Diamagnetism For materials which have no conduction band electrons, the magnetic character of the material depends on how strongly the wave functions of the electrons overlap. The vast majority of electrons, those that are involved in covalent bonds, and those that fill nonvalence core 214 electronic shells do not display a spin angular momentum paramagnetism. Only orbitals that have degenerate levels that are not filled will show a paramagnetic response to an applied magnetic field. The Pauli exclusion principle favors filling the electronic spatial wave function with two electrons possessing antisymmetric spins. These states have no easily accessible excited states so the total spin and orbital angular momentum is zero. In order for electrons of this sort to show a net spin angular momentum as a result of an applied magnetic field, there would have to be a large population of electronic excited states; this excited state population is highly unstable at normal temperatures as typical excited state energies are of the order of eVs, much greater than normal thermal energies. These electrons do contribute to the bulk magnetic suscep— tibility however by induced orbital motion of the electrons by the applied field. Classically, the Langevin diamagnetism is analogous to the effect that a magnetic field has on a current loop; a current is induced that generates a magnetic field opposing the applied field. If the motion of an atom’s electrons is considered to be a current loop, then the electronic response and resultant field can be calculated. The motion that the electron follows is a precession about the axis of the applied magnetic field vector. Then the magnetic 215 moment is a—a— pe (e/ZDC)°0Lp2 (6.5) where mL is the frequency of precession and is proportional to the applied field H, p2 is the mean square radius of the precession in a plane perpendicular to the applied field. Substituting 2r2/3 for p2, where r is the mean square radius of the precession in general spherical coordinates, and differentiating with respect to H the magnetic suscep- tibility corrected for volume is ” I 14'” x = (Ne2/6mc2) (6.6) i inpgN for a system having Z paired electrons. The result predicts no explicit temperature dependence; only if increased thermal motion affects the mean electronic radius will the Langevin magnetic susceptibility vary with temperature. All electrons will precess and thus show weak diamagnetism with xdia = -lO-'6 per mole of diamagnetic electrons. The major problem in calculating the diamag- netic susceptibility is in evaluating r2. xdia for atoms and ions may be evaluated from electronic distributions calculated with appropriate wave functions. The diamag- 216 netic susceptibility for molecules is somewhat more difficult to calculate from basic principles, but a set of values for various molecular fragments, known as Pascal’s constants, have been generated by differential measurement of the susceptibilities of related materials and can be used to predict values of xdia with fair accuracy. VI.A.3 Curie Law Paramagnetism In the opposite limit of electronic states, unpaired non-interacting localized electrons give an entirely different form of magnetism in the presence of applied fields. Electrons of this sort are found in organic radicals, transition metal salts, and, of particular importance to this study, electride salts. Application of the magnetic field lifts the degeneracy of the two elec- tronic spin states of the isolated electron in the S = 1/2 case. Electrons with spin parallel to the applied field have lower energies than electrons of antiparallel spin by an enerSY AB = -2uBH. This energy difference is comparable to thermal energies, so that there will be an appreciable population of electrons in the higher energy state; thus Boltzmann statistics apply. In the general case of an electron in an ionic orbital having both orbital and spin angular momentum (L and S respectively), with total angular 217 momentum (J) determined by Hund’s rules, the problem may be set up as -F/kT J -pBgHJz/RT e = 2 e (5.7) J -J z where F is the Helmholtz free energy and the 2J+1 states that are thermally accessible arise from Zeeman splitting of the angular momenta. Now the magnetic moment is given by aF/aH which may be calculated from the Boltzmann distri—, bution. The summation and differentiation of eqn. 6.7 yields for the magnetization of N ions in a volume V M = % ‘”BJBJ[ kT ] (6‘8) where the Brillouin function, BJ(x), is BJ(x) =2331 coth[g%ilx] « %3 coth[§3 ] (6.9) The magnetic susceptibility is given by M/H. In the limit that kT>>ngH (i.e. x (00 -‘U H 'U'U 043 .:n 0 a o 049 :In 00 232 4.76 .3 94 0.355 0.052 Quenched 5—260 K std. dev. est. —§ 165 x 10:3 x 10 0.314 x 10 1.201 x 10 0.61 96 0.361 0.014 Annealed 5-260 K std. dev. est. 1. .7 0.53 96 All data above 5 K 0.361 std. dev. 0.006 est. 233 .osswmwomsou mo sofipossm m we no.mflmumav_mo mo onEwm wofimoccm cm mo xpfiaflnflpmoomsm ofisoapoofio stoe Hmoosmfloms och .mm osswfia .xsve omm ov~ c r - MN s Oan b chad s ONfi s OWH m om— p CODA - Ohm p OPQ b 0.0 s ObN b o L .m rm r v r/; X a w x TL 0. r m 7. w W 3 fitfifiéé ... x em.vun¢ maom.cuu r m msoHoEmsmm mmsezumnsau mus 234 susceptibility follows Curie Weiss behavior with a fractional Curie constant of 97% and e = -4.5 K, in accordance with the parameters obtained from the quenched sample. Below 5 K the behavior is much different. A plot of x: vs. T as a function of the applied field is shown in Figure gg. A maximum in the susceptibility occurs at ~4.2 K for all fields. The behavior below 4.2 K is field dependent, and indicative of antiferromagnetic coupling.- At low fields (100 Gauss) the susceptibility falls off with decreasing temperature with an extrapolated value at 0 K of 69% of Xe at 4.2 K. The susceptibility of a powdered anti- ferromagnet theoretically is expected to fall to 2/a of the value of the Neel point. Therefore at low fields the material acts as a classic antiferromagnet. As the field is increased to 7 k6 the susceptibility falls with decreasing temperature to 3 K where it begins to rise with further decreases in temperature. Intermediate fields give intermediate behavior. For example the behavior at 1 k0 falls with decreasing temperature to 2 K where it levels off at about 80 percent of the maximum susceptibility. A plot of x: vs. H at a constant temperature of 1.6 K is shown in Figure 31, The susceptibility increases rapidly with the applied field until the field is ~4 kG. 235 . q . ’ - O 4 6 Iq’o’x -x\\\ 4.2- ’/ 4s0‘ / I m Xe(x10‘) . , 3.8 q , / 3.6--I / H'O . 1.“? 3.2 - 3.0 — In “-1 r(°x) Figure 36. The temperature dependence of the molar electronic susceptibility an annealed sample of Cs+(15C5)2°e‘ as a function of the applied field. 236 Ei(Kilogauss) Figure 37. The molar electronic susceptibility of an annealed sample of Cs+(lSC5)2'e‘ as a function of the applied field at 1.6 K. 237 At higher fields the susceptibility is approximately constant. An inflection in the curVe occurs at ~1.5 he. The magnetic susceptibility of a non—metallic para— magnet is a probe of the electron-electron interactions. In the case of the electride the paramagnetic electrons are localized in cavities which are separated by 8.68 and 10.27 A in the 1806 salt and probably by a slightly shorter distance in the 1505 salt. The Cs+(1505)2-I_ structure has interanionic distances of 8.32 and 9.31 A. Since the iodide, sodide, and electride anions are nearly the same size a rough approximation of 8.5 and 9.5 A from the iodide structure will be used for the nearest interelectron distance for comparative purposes. The shorter electron— electron distances in Cs+(1505)2-e— with respect to Cs+(18C6)2-e_ affects their respective interactions. The Weiss constant is ~-4.5 K for both quenched and annealed samples of Cs+(l5C5)2-e— and is only —l.4 K for Cs+(18C6)2-e_. The network of anionic nearest neighbors in each system gives an octahedral anionic nearest neighbor shell with four equivalent long interactions and two short axial distances. The short distances are very similar in the two electrides, 8.5 A for the 1505 salt, and 8.68 A for the 1806 salt, but the longer distances differ by ~1 A. The 1806 salt probably generates a mean field along the 9 direction with little interaction in the planes N to the g7 238 h planes. The shorter distances in these planes in the 1505 salt could allow a three dimensional magnetic lattice to exist. The Neél temperature of the annealed Cs+(1505)2-e— is only ~4.2 K. Relative to other antiferro- magnetic materials, Cs+(1505)2-e— has several unique features. First the orbital angular momentum of the ground state of the trapped electron is zero, it is probably an a— state electron; typical antiferromagnets are composed of transition metal or rare earth ions where the magnetic electrons partially filled d or f orbitals. The distances between magnetic centers in Cs+(1505)2-e_ are very large on the scale of most antiferromagnets. Transition metal ions that are greater than 3 A apart tend to have antiferro— magnetic coupling that is mediated by a non-magnetic ion or conduction electrons. MnO [99] for example has Mn+2 separations of 4.3 A -- too distant to have direct exchange. The material is antiferromagnetic (TN = 122 K) by virtue of a super-exchange mechanism involving exchange interaction with filled oxygen orbitals. The closest magnetic ions in the electride salt are ~8.6 A apart. It is possible but unlikely that a super-exchange mechanism is mediated by the Cs+ cation. While NMR has shown electron density at the cesium nucleus due to interactions with eight nearest electrons, the orientation of cationic orbitals to mediate exchange for eight interacting 239 electrons would be difficult to arrange. Further, the interactions of the crown ether oxygen lone pairs with the Cs+ ion would surely be a stronger influence on the p and d character of the cation than the diffuse electrons. Indeed, the paramagnetic Ramsey shift of the complexed cation is substantial and constant regardless of the anion, indicating that the excited state character of Cs+ is a result of the Cs+-0 attraction. The extrapolations to infinite temperature of the paramagnetic Knight shift due to s orbital character of the trapped electron is equal to the Ramsey shift due to the complexed cation; there is no additional p or d character added by the trapped electron. The electron density at the nucleus might be better thought of as a measure of the diffuseness of the electron cloud, strengthening the possibility that the exchange is direct. The coupling is very weak although several antiferromagnets have lower Neél temperatures. Finally the field dependence below the Neél temperature is unusual, especially at such low fields. This field dependence is due to very weak coupling of the spin lattices to the easy axis in the crystal. The phenomenon is known as a ’spin flop’ and is the result of weak crystalline anisotropy forces. In a single antiferromagnetic crystal at temperatures below the Neél point, the magnetic moments of the two spin lattices in the absence of an applied field are opposed to 240 each other and parallel to the easy axis. As a weak field is applied, the crystal displays a susceptibility depending on the orientation of the easy axis with respect to the direction of the field. The energy of the crystal is lowered as the easy axis is rotated towards a 90' angle with the field. The perpendicular orientation of the easy axis is the lowest energy orientation because the magnetic moments are able to equilibrate with the externally applied magnetic field whereas there is no torque exerted on the spins that are aligned parallel to the applied field. The spin flop occurs when the applied field creates a difference in the energy of )(‘l and x“ that is larger than the anisotropic force that is orienting the spins along the easy axis. Above this critical field all magnetic moments align perpendicular to the field regardless of the orienta- tion of the easy axis, and a powder sample should have a. similar susceptibility to that of the perpendicular single crystal as shown in Figure 322. The critical field for the spin flop, Hf, is related to the anisotropy energy by K = ‘/2(xl-x")fli (6.15) Other antiferromagnets, with what are considered to be weak crystalline anisotropy energies, are CuClz-Zfizo, with Hf = 7-8 ha [100], and Cr203, with Hf = 59 k6 [101]. 241 The crystalline anisotropy in Cs+(1505)2-e— is very weak, but this is not a surprising result considering the origin of anisotropy fields. In general the anisotropy fields are due to crystal field quenching of the orbital angular momentum, anisotropic exchange mechanisms, and anisotropic dipole-dipole interactions. Since the postulated ground state of the electron has no angular momentum, and the exchange is expected to be direct, the only source of magnetic anisotropy in this system is dipole-dipole coupling. The trapped electron is expected, at least in the ground state, to occupy a spherical s- orbital in the trap, so that even dipole-dipole interactions should be weak. The nearly octahedral arrangement of the neighboring magnetic ions would be expected to generate only a small anisotropic field. The nature of Cs+(15C5)2-e_ includes long range magnetic ordering if the compound is cooled slowly from temperatures between 220 and 250 K, and does not show anti- ferromagnetism if it is rapidly cooled from above these temperatures. The DSC results indicate that there is a weak endothermic transition at 238 K. The exact nature of the transition cannot be deduced without structural information; however, a guess as to its nature might be made. 242 The electron-electron interaction in the Cs+(l5C5)2-e- salt is probably similar to that of Cs+(1806)z-e- where channels through the closest packed complexed cations serve as the least impeded means of communication. If none of the trapped electron density is considered to reside on proton, carbon, or oxygen orbitals then these saturated bonded atoms in the molecule would repel the electron density should they be near a channel between two adjacent electron traps. (This is probably a simplistic view since the hydrogen atoms nearest the electron center probably have some acidic character, hydrogens farther from the electron center are not as likely to have excess electron density however.) The channels in the Cs+(1806)2-e- were described in Chapter III as being small in cross section with a significant amount of curvature at the most constricted point of the channel. The electrons in Cs+(1806)2-e— do not communicate well through the lattice and thus do not show magnetic ordering. To discuss the situation in Cs+(15C5)2-e_ it will be assumed that the electride and iodide of Cs+(1505)2 are isostructural. The structural features in Cs+(1505)2-e_l that would encourage stronger electron-electron overlap would be larger channels and shorter distances between cavities. The distances ggg on the average shorter in the 1505 salt with a more isotropic arrangement of traps about 243 each cavity. The distance between electron centers is 8.68 A along the géaxis and 9.6 A in the x—y plane. The anionic cavity of Cs+(1505)2-I_ was studied on the Evans and Sutherland P8300 graphics system. The cavity in the 1505 complex is considerably more spherical than the electron cavity in Cs+(1806)2-e- (Figures ii and 12), since the anion in the former compound sits on an inversion center and a four-fold rotation axis. There are only two unique channels between anionic cavities; anions along the g—axis are separated by a channel that is both longer and narrower than the interanionic channel in Cs+(1806)2-e-; however, the axial near neighbor anions in the iodide are separated by shorter, more open channels. The endothermic transition at 238 K strongly affects the extent of the electron-electron interaction; it may also be related to the transitions seen in the static NMR of Cs+(1806)2-Na— at similar temperatures. The linewidth of the static 23Na resonance in this salt is primarily due to dipolar coupling of the protons to the sodium anion. At temperatures below 240 K the linewidth is ~2700 Hz in agreement with the linewidth calculated from the Van Vleck formula (eqn. 5.1) and the known proton-Na_ distances. Above 240 K the peak narrows to ~1300 Hz, a change that is far too dramatic to be the result of a phase transition between two static crystalline phases. The explanation for 244 the narrowing is that near 240 K there is an onset of rapid motion of the protons that line the anionic cavity such that the anisotropic dipolar coupling to 23Na is partially averaged. It is clear that if this dynamic system were rapidly quenched the conformation of the crown ether would not necessarily be that of the lowest energy state. The electron-electron coupling in this compound is very weak so that a slight distortion, such as a hydrogen moved to further constrict a channel, could destroy the long range order necessary for magnetic ordering. The 1330s NMR in this temperature region does not show any splitting; however, the density at the cesium nucleus does deviate towards higher contact density at this point. It is unclear why there is a tendency for the electron to localize more strongly on the cesium with increased motion of the crown when at 270 K a second endothermic transition causes an abrupt decrease in the contact density.‘ The increased density at the 1330s nucleus at temperatures above 240 K with the decrease in the magnetic order does indicate that a superexchange mechanism is not responsible for mediating the magnetic ordering in this salt. The results of the magnetic susceptibility studies indicate that Cs+(1505)2-e- and Cs+(1806)2-e— have a very . sensitive and unique form of electron interaction. The structures of the two electrides are quite similar, but 245 subtle changes in the trapped electron network are responsible for quite different behavior in the two compounds. Cs+(18C6)2-e- has essentially non-interacting trapped electrons, Cs+(1505)2-e- shows a fragile magnetic ordering. Of the electrides isolated so far, only those with the complexed cations consisting of crown ether sandwiches show pure Curie-Weiss paramagnetism with or without magnetic ordering. In addition to the cesium compounds described here K+(1505)2-e-, and Rb+(l5C5)z-e— have been isolated [72] and behave as Curie-Weiss para- magnets. Electrides with cryptand complexants tend to have much more strongly interacting electrons. K+0222-e- and Rb+0222-e- are metallic Pauli paramagnets [93]; Cs+0222-e- is diamagnetically spin paired [93]; and Li+0211-e- shows a temperature dependent spin pairing [74] which seems to be best described in terms of a dimerization of electrons at low temperatures. To date, no strong characterization of electrides containing 1:1 complexes with crown ethers exists. Only Rb+1806-e- has been reported [71] and the stoichiometry of this compound remains a puzzle. If it is an electride it too will fall into the category of electrides which show extensive spin pairing. 246 VI.C. Electron Paramagnetic Resonance of os+(lscs)2-s’ The EPR spectra of Cs+(l5C5)2-e_ were measured over the temperature range 2-250 K for three different samples. The EPR line of the trapped electron of Cs+(1505)2¢e- might be expected to display hyperfine coupling to the eight cesium cations, since the NMR results show that there is a small but finite electron density at each cesium nucleus. The hyperfine constant, A, gives the splitting of each electron] state due to interactions with the magnetic moment of the cesium nuC1ei. The strength of the hyperfine interaction is related to <[¢(o)[2>, the electron density at the nucleus, by - 2 A130 - (8/3)wsuBsNuN<|¢(0)l > (6.16) where g and gN are the electronic and nuclear g-values, and p8 and p" are the electronic and nuclear magnetons. For Cs+(1806)2-e_ the density of one electron at each of eight nuclei is one-eighth of the total electron density at that nucleus if it is assumed that each anion-cation interaction is equal. The expected hyperfine spectrum for 22 3 Cs+(15C5)2-e- at 200 K where = 2.2 x 10 e_/cm from Knight shift measurements, if it is assumed that eight equivalent nuclei interact with each trapped electron, 247 would consist of 57 lines arising from 2(7/2-8) + 1 different summations of the nuclear magnetic moments of the eight nearest cations. Each line would be separated by 0.10 6 according to the value of <|¢(o)[2> from NMR measurements. Interactions with hydrogens on the crown ethers could further complicate the spectrum. In the absence of broadening interactions with the crown ether hydrogens, the spectral breadth would be about 6 G. The observed spectrum at 200 K is a narrow single line. The peak to peak width of the line, AHp-p’ is 0.6 G, and the g-value of the line is 2.0022 +/- l x 10“, which is nearly identical to the free electron g value. Deviations in the g value are indicative of spin-orbit coupling to orbitals of non-zero angular momentum. The linewidth is about one order of magnitude smaller than the expected width of the hyperfine broadened spectrum. Therefore, it must be assumed that the line is narrowed by rapid exchange [94], in which electrons are able to exchange their spin angular momenta via a first order process. The large concentration of electrons in Cs+(l5C5)2-e- and significant electron—electron interactions allow the exchange to proceed at a rate more rapid than the relaxation rate due to the hyperfine interaction so that the hyperfine coupling of electrons is averaged to a small value. Similar + - . exchange narrowing occurs in Cs (1806)2-e , and in other :ver—va'—*>— --.. - -— - 248 high electron density radicals, including DPPB. The temperature dependence of the EPR spectrum of three samples of Cs+(1565)2-e- was measured over a temperature range of 2.0 K to 240 K. Selected spectra at various temperatures for each sample are displayed in Figures 38, ;g and 40. In all cases the sample was rapidly quenched from 270 K to 2.5 K. The intensities of the spectra in general follow the l/T dependence expected for Curie-Weiss paramagnets, and there is no significant g-shift. an P"'P does vary with temperature, however. The temperature dependence of afi of sample 1, which maintained what P‘P appeared to be a single component peak at all temperatures, is shown in Figure 4;. The increase in AHp_p with decreasing temperature below 40 K empirically follows a T‘z/3 dependence. The low temperature spectra of the "quenched" sample in Figure gg displays a partially struc- tured, broadened line, the high field lobe of the first derivative of the resonance absorption is split into two peaks, while the low field absorbance shows no structure. Sample 3 displays the most complex behavior as a function of temperature. The spectra at 2.5 K and 6 K of the but ”quenched" sample are very narrow, AHp-p = 1.1 gauss, show some structure on the line. This sample was then 249 213K 48.K Sl( Figure 38. EPR spectra of Cs+(15C5)2-e-, sample 1 at 2.0 K, 48 K, 200 K, 80 K, 14 K, and S K. 250 215K f / WW / 19 K I}, Mr A"! / KJK k" \ ZJSK \ fl’ffl w“ / + ‘ K Figure 39. EPR spectra of C5 (15C5)2-e , sample 2, at 2.8 K, 19 , 10 K and 2.5 K. 251 2.5K 6K A 158K 2.6K [Lil Figure 40a. EPR spectra of Cs+(lSCS)2'e-, sample 3, at 2.5 K, 6 K, 158 K, and 2.6 K. 7d5K P9K 50|( 245K LljiJ Figure 40b. EPR spectra of Cs+(1SCS)2~e“, sample 3, at 7.5 K, 19 K, 50 K, and 2.6 K. 253 .uo.mnmumflv+mu me A ofimssm sow oesumsomaou mo coflpocsm s we mlmz< .He ossmAm .Z_>._mv: manzzmaimh com on. co. on — _ _l A _ q a . d O l Oil: OI. L m... l. /. L cg. O Ammm<0v adId L O n .ma 254 warmed to at least 160 K and then gradually recooled. The EPR spectrum of the "annealed” sample at 2.6 K, shows a remarkably broad, well—resolved six line structure. The splitting between peaks is roughly 1 G. As the sample was warmed to 7.5 K and then to 19 K the structure of the low field lobe of the line rapidly disappeared, while the structured high field absorption gradually narrowed. By 50 K the EPR line was indistinguishable from that of the other samples. The sample was then warmed to 220 K and rapidly cooled. The resulting spectrum at 2.6 K more closely resembled the low temperature spectra of Figure 39 with less resolved multi-line structure. The increased breadth and onset of complicated struc~ ture of the EPR line of Cs+(1505)2-e- at low temperatures appears to have a dependence on the thermal history of the sample, and, while the effect of thermal cycling is more profound in the temperature range of 5 K to 20 K, the similarity to the onset of quench/anneal behavior in the magnetic susceptibility is striking. The response of anti~ ferromagnetically coupled spins to the EPR conditions requires consideration of crystalline anisotropy forces, and crystalline shape anisotropies in addition to the exchange and thermal energies. All measurements were made at x-band microwave frequencies, v = 9.6 GHz, and at fields near 3400 gauss, well above the spin-flop critical field. 255 In order to investigate systematically the relative importance of the various influences on the EPR spectrum, single crystal experiments are required [96]. Single crystal EPR and zero-field or low—field antiferromagnetic resonance experiments [96] are capable of probing the strength of the sublattice magnetizations and the orienta- tion and strength of the crystalline anisotropy forces. While the low temperature spectra observed in Cs+(l5C5)2-e- are almost certainly affected by high temperature phase transitions that, in turn, affect bulk electron—electron interactions, the observed line broadening at low temperatures is not necessarily due to those electron-electron interactions, but may rather be the result of coincident changes in electron-nuclear interac~ tions. The onset of low temperature broadening occurs at nearly 50 K, and the structure in the EPR lines appears at temperatures as high as 20 K, whereas the onset of magnetic ordering comes at only 4.2 K. Increased isotropic hyperfine coupling may result in a line that is not completely averaged by rapid exchange narrowing. Again the results from the three samples, while sharing general characteristics, are too varied to justify speculation about their interpretation. J. Kim, in Profesor James L. Dye’s laboratory, is presently engaged in the design of an 256 EPR goniometer that could be used to further study the low temperature single crystal EPR spectra of Cs+(l5C5)2-e—. CHAPTER VII CONCLUSIONS AND SUGGESTIONS FOR FUTURE WORK VII.A. Conclusions A new electride, Cs+(15C5)2-e_, was synthesized and characterized by DSC, solid state NMR, optical spectro- scopy, and magnetic susceptibility. Analysis of the sample verified the stoichiometry. A new method for mounting single crystals of the air and temperature sensitive alkalides and electrides for x— ray crystallography enabled the determination of the struc- ture of Cs+(1806)2-Na_, the first crystal structure of an alkalide since 1974. The space group is the monoclinic CZ/c, with lattice parameters 3 = 13.581 A, b = 15.684 A, c = 17.429 A, B = 93.16. The first crystal structure of an electride, Cs+(1805)2-e-, was also determined. The elec- tride is nearly isostructural to the sodide, with lattice parameters 5 = 13.075 A, g = 15.840 A, g = 17.359 A, and B = 92.30 in CZ/c. The structure consists of a lattice of tightly packed complexed cations with large (5 A diameter) interstitial anionic cavities at the same location as the Na_ ions in the sodide. These cavities contain sub—noise 257 ¥ 258 level electron density, and they do not contain any atomic species. The structure provides the first direct evidence of a site for the.stoichiometric localized electrons. The crystal structure of Cs+(1505)2-I- was determined. The iodide anion has a similar anionic radius to the trapped electron and is expected to be similar in its packing to that of Cs+(l5C5)2-e—, whose structure has not been determined. Thespace group of the iodide is tetragonal I4, with lattice parameters g = 13.172 A, and g = 16.645 A. All three of the structures have similar packing of the ions, and similar anionic environments. Solid state 1330s NMR was performed on a variety of simple and complexed Cs+ salts, and on the one known compound that contains Cs+(1806)2-Cs- The chemical shift of simple cesium salts is dominated by a paramagnetic Ramsey shift that ranges from 284 ppm for CsI to 190 ppm for CsSCN; the shift is stronger for anions of greater polarizability. Exclusive complexes of Cs+ with 0222 as well as complexes of Cs+ with 1806 in a 1:1 ratio have a weaker, but anion dependent paramagnetic Ramsey shift. The chemical shift of Cs+(1806)2 sandwiched complexed cations is -60 +/-2 ppm for all anions except the ceside and the tetraphenylborate. The chemical shift of Cs+(1505)2 salts is 28 +/—6 ppm, and the chemical shift of 259 inclusive Cs+0222 salts is ~240 ppm. The chemical shift of inclusively complexed cations in the solid state, whether by 0222 or by sandwiched crown ethers, is very sensitive to the overlap of oxygen lone pairs with empty Cs+ orbitals, and is thus sensitive to the mean Cs-O distance in the complexed cation. Large anions in Cs+(1806)2 salts apparently require a different conformation of the crown ethers about the cation in which the mean Cs—O distance is decreased. Static and single crystal NMR studies revealed the anisotropic chemical shift and quadrupolar interaction parameters for Cs+(1806)2-I— (6 = 32 ppm, n = 0, eZQq/h = 89.6 KHz) and Cs+(1505)2-I_ (6 = 52 ppm, n = 0, equ/h = 427 KHz). Dipolar linewidths were measured for 1330s and 23Na in Cs+(1806)2-Na— at two temperatures, 200-210 K and 250 K. The low temperature linewidths correspond to the values calculated from structural parameters. The high tempera~ ture static spectra are substantially narrowed due to rapid motion of hydrogens on the crown ethers relative to Cs+ and Na- which averages their dipolar fields. Cs- in Cs+(18C6)2-Cs— is paramagnetically shifted from ' the chemical shift of the gaseous anions by ~135 ppm at 220 K. The anionic chemical shift approaches the gaseous anion value with increased temperature; a phase transition at 285 K shifts the resonance to -240 ppm. The phase 260 transitions in this material act to increase the anionic cavity size with increased temperature. The chemical shift of electrides is strongly affected by a contact interaction of the paramagnetic trapped electrons. The temperature dependence of the chemical shift was carefully measured for both Cs+(1806)2-e- and Cs+(l505)20e_; both follow a l/T dependence characteristic of the Curie Weiss paramagnetism of the electrons. The electron density at the Cs+ nucleus was calculated from the magnitude of the slope of the chemical shift versus l/T. The fractional atomic character of Cs+ in Cs+(1806)2-e- is 3.3 x 10’4 , and of Cs+ in 0s+(15c5)2.e‘ is 8.4 x 10’4. A phase transition occurs at ~220 K for Cs+(1806)2-e-. The NMR spectrum indicates that a conforma— tion of the complexed cation with shorter Cs-O distance is stable at low temperatures. Two phase transitions at 238 K and 268 K affect the temperature dependence of the NMR spectrum of Cs+(l505)2-e~. The low temperature transition appears to increase the electron density at the nucleus as the temperature increases. The phase transition at 268 K appears to localize the electron in the cavity more strongly, in that the Cs+(1505)2-e- resonance shifts abruptly upfield by 190 ppm. 261 The greater electron density at the cesium nucleus in Cs+(l5C5)2'e- compared to Cs+(1806)2-e- indicates that the cationic charge in the 1505 salt is less effectively shielded by the crown ethers than is CS+ in the 1806 complex. The poorly shielded cations in Cs+(l505)2-e— cause the electron density to be drawn into the lattice, resulting in relatively strong electron-cation interactions. NMR spectra were measured for several samples of compounds of mixed anionic stoichiometry. Solid samples prepared from solutions of stoichiometry Cs+(18C6)2-Cs;ezl_x) are separated into two phases, the pure electride and pure ceside, for all values of x between 0 and 1. Solid samples prepared from solutions of stoi- chiometry 68+(1806)Z'Na;e(1-x)’ for 0