am

'1
5

‘fl'r'm? “”9-

a)

n,

.-.-L.

. z

 

5:.W'1?
,3'5: {IQ I’M" {“3

W .
:52! . j" '. 4
(3‘3” "3?. 1M?“ ‘ “fig”

 

.V I 3"

’ .4 . . .u, 4‘;
‘4 43‘} ‘
4 54-4047?»
42’ r .'

i‘DI‘

1

 

.I-

"Lit-W! -'
4km“p

'fir‘i’l‘Pum‘ "
s

'4‘

:h.‘

A
#7}

3.1 4; {ff
{2" 9“. 4:}:1‘};

.E‘ .

II
CZJAJ'E’E' -"
I',~.

{31.5%} r
' 9
'r'r' P443?

("1;

~~4 ' L '
4444444; 4 4+ 4444
A“ ‘f'qlnb‘l .- .4 I-v-r‘? mfifldf “t f'
4 l , I! ‘4, ”'1“ "‘ryf'; I
1 ,4 'f . ‘.- 4 .1

 

 

ll ‘14};

1 . " 1 -
“L m; ‘ ._ .' t. ‘ 4 ‘ .l‘ 4 . . . c. ' . - 4.
x (4157141414 ’13:; '- -4 ' .E 1’ S i'. 41.43, L 9’11"“: f VIA Afi" ‘2‘.§(;" "’- ‘2'“J‘ ’ ' ’
u . . -+ ~ ‘ '9 - ~ -‘."3’.« 4
4} ,‘ S‘- ’15.“ Jiisf},‘}§ 23;: 7p . , I r/ .
. 4p.

 

 

W- V
,I..4f;"r1,{"lrr
gt" .3

J?
',:.','- 2:“ J"):

‘,‘ ‘V‘ H I I
n 4 - 7‘ 4 ‘ I”. Y. ' 1:1“
.- ’l ' . ‘ (Jr W l
. . . -, z 4 . 4 '. ""~ ' ‘ ' “ ‘ I r“
,. (.;?.{’.-, .l .4 ._ . + J A, 1‘. 4'?“"" .- -I . “4"4 - '- . ’ “Hg/“rt
.0 I ' ’54-} 1:.

 

 

:11} .
”AER—5431.4

. I"
g. :12?

» 3x *Pfiw'” 5’
5‘;..,4+7L+’“P‘1~; 32.13335}

'3'
~ -.~+ 52554444343; was:
r‘ 0' " I ~
w 35—?!

IA.‘

.5.

 

Ear:

 

.. ..
J ”9"" ‘
in; ..n "’"'"'r' ' '
Mu 544.1. m: 4 2"‘4'4‘3'5’
'—
- =33" '

4 .
'L—f -

. .U’ a '- 3"l‘q' ' :- 4'

. 41:4... .44 '- - ' “’ ‘

{'J. 1‘4. ' 4! . Y4 a_ 4 . . m 9‘1: ""5"“1'.“ ' ‘ :1" ' '
. . ’ ,4. 4 , . inf; W42 4 ' 4- -’-"xxfu‘ ', ;"'",. $4.319“ 2""? " Ila-M? ,Zmngtw

 

 

 

 

n '
'11:. r5!" . ”13-9, _ . a. ‘ ~‘. 4 -- '73",
1'... 4. . .’ ' 4 ‘ . $22.57”:

' 3,15 4“", .
J'v' ’1‘ g 37;?5‘: :71: I???”
6.9.4 '1"4"‘I

’7 “mill”; LL69

 

sn‘v Ll IBRARIES

Mill?“ \1\\\2\\ “ll “Mlle \ “El“ l}; “W \\\\\|

 

 

This is to certify that the

dissertation entitled

ALKALI METAL NMR AND MAGNETIC SUSCEPTIBILITIES OF
ALKALIDES AND ELECTRIDES

presented by

Judith Lynne Eglin

has been accepted towards fulfillment
of the requirements for

Ph. D. degreein Chemistry

 

 

 

Major p essor

Date W0

M5 U is an Affirmatiw Action/Equal Opportunity Institution 0-12771

 

 

 

 

 

r ' “ T‘
LIBRARY
Michigan State
University
1 \ J

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

_o L
I: §

F—_

 

 

 

 

 

 

 

 

 

 

'1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

fil—fi

MSU In An Affirmative Action/Equal Opportunity Imitution
cmmhr

 

 

 

ALKALI METAL NMR AND MAGNETIC SUSCEP’I'IBILITIES OF
ALKALIDES AND EIECI'RIDES

by

Judith Lynne Eglin

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OFPHILOSOPHY
Department of Chemistry

1990

Q4S—Wooq

ABSTRACT

ALKALI METAL NMR AND MAGNETIC SUSCEPTIBILITIES OF
ALKALIDES AND EIECI‘RIDES

by

Judith Lynne Eglin

The electride Cs+(15CS)2e’ undergoes an antiferromagnetic

Neel transition at 4.6 K. At higher temperatures, the susceptibility
follows the Curie-Weiss law until the temperature of a first-order
phase transition at 266 K. This transition can be observed by both
magnetic susceptibility and 133Cs NMR studies. Below 240 K, the
1”CS chemical shift is linear in 1/T with an intercept of 180 1 10
ppm. The slope of the line and the magnetic susceptibility data yield
a Fermi contact density at Cs+ that corresponds to 0.063% atomic
character.

The physical properties of Cs+(HMHCY)Na' at two different

electron dopant levels were studied and the 133Cs MAS NMR spectra
were observed for the first time. The compound, Li2+(TMTCY)2Na'

-CH3NH‘ was characterized and the crystal structure obtained. Since

solvent is incorporated in the crystal structure, a method was
developed to test for its presence.

The complexant TMPAND, a fully methylated aza analog of a
cryptand was used to synthesize a sodide and an electride. The
temperature dependence of the quadrupole coupling constant for 7Li
in the sodide was studied above and below the observed phase

transition. The phase transition could also be detected by

conductivity measurements and 23Na NMR spectra, but not by
optical absorption studies.

' The systems Li+(CH3CH2NH2)4Na‘, Li+(CH3CH2NH2)4K', and
Li+(CH3CH2NH2)4CS‘ were characterized by magnetic susceptibility
measurements and 7Li NMR studies. While the sodide appears to be
a diamagnetic compound, the potasside and ceside are paramagnetic.
The 7Li chemical shift value is temperature dependent and increases
with temperature. The intriguing properties of these systems
suggest that further characterization might verify a new class of
“metallic” compounds.

The spin-echo NMR technique with phase cycling is a powerful
tool in the determination of the environment of 87Rb nuclei in
alkalides and elecuides. Computer simulations of the experimental
powder patterns yielded the quadrupole coupling constants for the
complexed rubidium cation as well as values of the asymmetry

parameter.

ACKNOWLEDGEMENTS

For his guidance and support throughout the course of this
work, I would like to express my appreciation to Dr. James L. Dye.
For their helpful advice, I would like to thank Dr. William Pratt, Dr.
Jerry Cowan, and Dr. Carl Foiles of the Physics Department as well as
Dr. Kim Dunbar of the Chemistry Department.

I appreciate all of the advice and support that I received from
all of the members of the Dye research group with whom I have
worked, in particular Rui Huang, Jineun Kim, Lauren Hill McMills,
Mark Kuchenmeister, Marc DeBacker, Sylvie Doueff, Mike Wagner,
Evy Jackson, Erik Hendrickson and especially Kevin Moeggenborg. I
would also like to thank the Master Glassblowers at Michigan State
University, Keki Mistry, Manfred Langcr, and Scott Bancroft, for the
numerous pieces of glassware that they have designed, built, and
repaired for me. I am grateful for Kermit Johnson’s help with the
NMR insuuments, Marty Rabb for his aid in the repair of the SQUID
and Dr. Don Ward for his help in my many attempts to obtain single
crystal X-ray data. Special thanks go to all of my professors in the
Department of Chemistry at The College of Wooster for all of their
encouragement and enthusiasm. Finally, I would like to thank “our
class” of fifth year students for all of their help and moral support.

I also appreciate the support I received from the National
Science Foundation Solid State Chemistry Grant DMR-87—l4751 and

the Michigan State University Center for Fundamental Research.

iv

II.

III.

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

LIST OF TABI FS - viii
LIST OF FIGURES ix
TABLE OF ABBREVIATIONS - - xv
INTRODUCTION _ l
COMPLEXED CESIUM CATION - -- ........... l7
ILA. Introduction ............................................................................ 17
ILB. Theory 20
KC. Experimental ........................................................................... 25
1. Synthesis ...................................................................... .25
2. NMR - _ - 26
3. Magnetic Susceptibility .......................................... 29
4 Differential Scanning Calorimetry ...................... 30
5 Optical Absorption .................................................... 30
RD. Results and Discussion ........................................................ 31
1. Cs+(15C5)2e'-... - _ ..... ..................... 31
2. Cs+(l 8C6)(15C5)Na' and

Cs+(18C6)(lSC5)CS' .................................................. 42
3. Cs+(18C6)(15C5)e‘ .................................................... 47
AZA COMPOUNDS 53
III.A. Introduction ............................................................................ 53
111.8. Theory - -56
l. Decomposition Studies ............................................ 56
2. Quadrupole Coupling Constants ........................... 5 8
lII.C. Experimental ........................................................................... 62
l. Complexant Purification ......................................... 6 2
2. Synthesis ....................................................................... 63
a. Cs+(HMI-ICY)Na' _ - 63

b. Li+2(TMTCY)2Na'-CH3NH' and
LIT2WCY)26'°CH3NH' ............................ 64
c. Li+(TMPAND)Na' and Li+(TMPAND)e'.65
3. Methylamine Test. .................................................... 66
4. NMR - - 68

 

a. Cs+(I-IMHCY)Na' .............................................. 68
b . Li+2(TMTCY)2Na’-CH3NH'

 

 

 

and Ll+2(TMTCY)26"CH3NH' ................... 68
c. Li+(TMPAND)Na' and Li+(TMPAND)e .69
5. Magnetic Susceptibility ................................................. 7O
6. Differential Scanning Calorimetry ............................. 71
7. Optical Absorption ........................................................... 72
8. Conductivity Methods .................................................... 72
9. Electron Paramagnetic Resonance ............................. 73
10. X-ray Crystallography ................................................. 73
III.D. Results and Discussion ........................................................ 75
I. Cs+(I-IMHCY)Na‘ - - ................. 75

2. Li+2(TMTCY)2Na' HCH3NH
and Li+2(TMTCY)2e CH3NH ............................... 81
3. Li+(TMPAND)Na' and Li+(TMPAND)e' ............. 89
IV. ETHYLAMINE SOLUTIONS ....................................................................... 106
IV.A. Introduction .......................................................................... 106
IV.B. Theory ..................................................................................... 108
NC. Experimental ........................................................................ 1 10
1. Sample Preparation ............................................... 110
2. NMR ............... -.-_- - .-.112
3. Magnetic Susceptibility ........................................ 113
ND. Results and Discussion ...................................................... 114
V SPIN-ECHO NMR STUDIES ........................................................................ 127
V.A. Introduction .......................................................................... 127
V.B. Theory ..................................................................................... 128
I. Quadrupole Coupling ............................................. 128
2. Nuclear Relaxation ................................................. 129
V.C. Experimental ........................................................................ 131
1. Synthesis .................................................................... 131
2. NMR----....-. ........ .......... . 131
V.D. Results and Discussion ...................................................... 135
l. 15C5 Complexes ...................................................... 135
2. 18C6 Complexes ...................................................... 138
3. HMHCY Complex ...................................................... 141
4 C222 Complexes ...................................................... 143
5 Single Crystal NMR ................................................. 147
VI. CONCLUSIONS AND SUGGESTIONS FOR RITURE WORK .................... 151

vi

APPENDIXA ....... -
APPENDIXB- - A---“
REFERENCES-.."- - -

vii

AAAA A A AA
v-vv vvvv-vvwvvv-w—vvvvvv

AAA AAA -_-A_‘

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

LIST OF TABLES.

Characteristic NMR Chemical shifts of alkali metal anions
and complexed cations. 31 ............................................................... l4

Characteristic 133Cs NMR Chemical shifts of the complexed

 

cation.75- -------------- - - ------- -- - --------- ........ 43
7Li Quadrupole Coupling Constants (QCC) of lithium
halides.103 ............................................................................................ 62

Magnetic Susceptibility conditions for the data sets of
compounds that were made with the complexants HMHCY,
TMTCY, and TMPAND ........................................................................ 71

Selected bond distances for the compound,

Ll+2('IMI‘CY)2Na'-CH3NH' ............................................................... 33

Table of thermal decomposition temperatures for

Li+(TMPAND)Na and the ramp rate at which they were

7 Li NMR chemical shift values obtained for
Li+(CH3CH2NH2)4Na' and Li+(CH3NH2)4Na-.65 ................. 107

7Li NMR chemical shift values obtained for
Li(NH3)4.55 ........................................................................................ 108

NMR parameters determined from the simulation of
powder patterns from 87Rb NMR spectra ............... 144

viii

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

LIST OF FIGURES

Representative Complexants ........................................................... 4

Magic angle sample spinning 23Na NMR spectrum of
Na+(C222)Na‘. Also shown is the nonspinning spectrum
of the same sample.52 ....................................................................... 8

23Na MAS-NMR spectra of Na+(C222)Na' at three
frequencies. Inset shows an expansion of a portion of the

bottom spectrum due to Na"'(C222) (solid line) and its
computer simulation (dashed line).55 ..................................... 10

133Cs (65.61 MHz) and 871111 (163.6 MHz) MAS-NMR

spectra of polycrystalline Cs+(18C6)2Rb'. 58 ....................... 12

Plot of chemical shift versus III for Cs+(18C6)2¢'.76 The

temperatures of points marked with an it were calibrated
with methanol .................................................................................... 19

A schematic diagram of the sapphire rotor and bolt .......... 28

Differential Scanning Calorimetry (DSC) trace of
Cs+(15C5)2e' at rarnping rates of 2 K/min (lower trace)

and 5 K/min (upper trace ofiset by one W/g) ..................... 32

The optical absorption spectrum of a thin film of
Cs+(15C5)2e' made from solvent evaporation of

methylamine at temperatures of 188 K (diamonds) and
253 K (solid line) ............................................................................... 34

Figure 9.

Figure 10.

Figure 11.

Figure 12.

Figure 13.

Figure 14.

Figure 15.

Figure 16.

Figure 17.

Figure 18.

l33Cs MAS NMR spectrum at a temperature of 276 K
with two peaks at chemical shifts of 493 and 298 ppm
obtained at a frequency of 23.6 MHz ....................................... 37

Plot of chemical shift versus III for Cs"'(15C5)2e'; 180

MHz MAS NMR data (open squares); and 400 MHz static
data with the isotropic chemical shift obtained from

simulations of the experimental powder patterns
(circles) .................................................................................................. 39

Magnetic susceptibility plot of l/xem versus T for

Cs+(15C5)2e‘ with an inset of the high temperature
data. ........................................................................................................ 41

The 133Cs MAS NMR spectrum of the mixed crown ether

complex, Cs+(l8C6)(15C5)CS', obtained at 245 K and a
frequency of 23.6 MHz ................................................................... 46

Magnetic susceptibility plot of llxem versus T for the

electride, Cs+(18C6)(15C5)e‘, obtained at four different
fields ....................................................................................................... 49

DSC trace of the mixed crown ether elecu'ide,
Cs+(18C6)(15C5)e', at a ramping rate of 5 K/min .............. 51

5,12,17-trimethyl-l,5,9,12,17-penta-
azabicyclo[7.5.5]nonadecane (TMPAND) ................................ 56

The idealized line-shapes for a polycrystalline sample
with quadrupolar coupling for I=3/2 when the
asymmetry parameter is zero ..................................................... 61

Magnetic susceptibility plot of l/xem versus T for the

heavily doped sodide, Cs+(HMHCY)Na', quenched and run
in a field of 5 kG ................................................................................ 78

The DSC traces of Cs+(HMHCY)Na' for the heavily doped
sample (bottom line) and the minimally doped sample
(top line offset by 1 Mg) obtained at a ramping rate of

5Klmin -- -- ---- ------------ - ........................................... 80

 

Figure 19.

Figure 20.

Figure 21.

Figure 22.

Figure 23.

Figure 24.

Figure 25.

Figure 26.

Figure 27.

A) The molecular structure and numbering of atoms in
Li+2(TMTCY)2Na'-CH3NH'; and B) Stereographic packing
diagram 0f LI+2(TMTCY)2N3'°CH3NH' ..................................... 82

Methylamine Test A) Li+2(TMTCY)2Na'-CH3NH'

quenched in D20 at cryogenic temperatures; and B)
Li+2(TMTCY)2Na'-CH3NH' quenched in D20 at cryogenic
temperatures to which a solution of CH3NH2 in D20 had
been added .......................................................................................... 85

DSC trace of the sodide, Li‘l'2(TMTCY)2Na'-CH3NH',
obtained at a ramping rate of 5.5 Klmin ................................ 87

Optical absorption spectra of Li+(TMPAND)Na' obtained
at 262 K (top line offset by +0.2 on the Relative
Absorption scale) and 183 K ........................................................ 90

The DSC trace of Li+(TMPAND)Na' obtained at a ramp rate
of 12 Klmin (top line ofl’set by 2 W/g) and 7 Klmin ........ 92

Plot of ln(¢/Tm2) versus 111‘ for Li+(TMPAND)Na' used
to determine the activation energy for decomposition....93

The variable temperature d.c. conductivity data for
Li+(TMPAND)Na' obtained in a 2-probe conductivity
apparatus.100 ..................................................................................... 95

23Na static NMR spectra of Li+(TMPAND)Na' obtained at
A) 189 Kand a) 213 K .................................................................... 97

Plot of the full width at half height versus temperature
for the 7Li static NMR spectra obtained for
Li+(TMPAND)Na‘. 'Ihe solid-solid phase transition occurs

at the temperature of the observed discontinuity in the
curve-.- - - ......... - ..................... 99

 

Figure 28. 7Li static NMR spectrum of Li+(TMPAND)Na’ obtained at
203K ..... 100

 

Figure 29. Plot of the Quadrupolar Coupling Constant (QCC)
calculated from 7Li static NMR spectra of

Li+(TMPAND)Na’ versus the temperature of the

sample. The solid-solid phase transition occurs

at the temperature of the observed discontinuity in the
curve .................................................................................................... 102

Figure 30. Magnetic susceptibility plot of llxem versus T for

Li+(TMPAND)e' obtained for two different syntheses.

The squares represent the sample with 29 percent
unpaired electrons run at 3, 5 and 7 k6. The plus
symbols represent the sample with 12 percent unpaired
electrons run at 5 and 7 kG ....................................................... 104

Figure 31. Schematic diagram of the quartz SQUID cell ...................... 111

Figure 32. Magnetic susceptibility plot of xcm versus T for

Li+(CH3CH2NI-I2)4Na' obtained at two different fields (3

and 7 k6) and two different times, the initial data
collection and the data collection after the sample had
“ag ” - 115

 

Figure 33. Magnetic susceptibility plot of 16‘“ versus T for

Li+(CH3CH2NH2)4K' obtained at two different fields (3

and 7 k6) and two different times, the initial data
collection and the data collection after the sample had
“aged”-- -- - -- 113

 

Figure 34. Plot of the 7Li NMR chemical shift value for the
compound Li+(CH3CH2NH2)4K' versus temperature ..... 120

Figure 35. Plot of the 7Li NMR chemical shift value for the
compound Li+(CH3CH2NH2)4CS' versus temperature....121

Figure 36.

Figure 37.

Figure 38.

Figure 39.

Figure 40.

Figure 41.

Figure 42.

Figure 43.

Figure 44.

Figure 45.

Plot of xcm versus T for Li+(CH3CH2NH2)4CS‘

obtained at 7 k6 for two different time periods, the initial
data collection and the data collection after the sample
had“aged”..------- -- - - ....................................... 123

 

Plot of xem versus T for Li+(CH3CH2NH2)4Cs' obtained

at 3 k6 for two different time periods, the initial data
collection and the data collection after the sample had
“aged”- -- -- _- 124

 

The spin-echo pulse sequence .................................................. 132

The 87Rb NMR spectra obtained for Rb+(18C6)c1- at
different pulse lengths to determine the optimum pulse
lengths for the spin-echo pulse sequence ........................... 133

The pulse sequence used to determine the

spin-spin relaxation time of Rb+(C222)Cl' at four
different orientations of the crystal ....................................... 134

The pulse sequence used to determine the spin-lattice

relaxation time of Rb+(C222)Cl’ at four different
orientations of the crystal .......................................................... 135

37Rb NMR spectra of A) Rb+(15C5)2Rb' and B)
Rb+(15C5)2e‘ obtained with the spin—echo pulse
seque ............................................................................................. 137

37Rb NMR spectra of A) Rb+(l8C6)Na' and B)
Rb+(l8C6)Rb' obtained with the spin-echo pulse
sequence. The solid lines are the observed spectra and
the dotted lines the simulated spectra ................................. 139

A cross section of the crystal structure of Rb+(l8C6)Rb'
showing the rubidium anion chains that are

present.131 ....................................................................................... 140

87Rb NMR spectra of Rb+(HMHCY)Na’ obtained with the
spin-echo pulse sequence. The solid line is the

observed spectrum and the dotted line is the simulated
spectrum. ........................................................................................... 142

Figure 46. 87Rb NMR spectra of the model salts A) Rb+(c222)SCN-

and Rb+(C222)I' obtained with the spin-echo pulse
sequence ....................... - ....... -- - ----------------- 145

 

Figure 47. 87Rb NMR spectrum of Rb+(c222)Rb- obtained with
the spin-echo pulse sequence. The solid line is the
observed spectrum and the dotted lines are the
simulated spectra. .......................................................................... 146

Figure 48. The single crystal 87Rb NMR orientation plot for the
compound Rb+(C222)C1‘ .............................................................. 148

Figure 49. A) The 87Rb NMR spectrum of Rb+(C222)Cl' obtained
with the spin-echo pulse sequence for a powdered
sample and B) Plot of the relaxation time (T1

spin-lattice and T2 spin-spin) versus chemical shift
of the rubidium cation for four orientations of the single
crystal Rb+(C222)Cl’ ...................................................................... 149

xiv

DIWICY

PMPCY

TMPAND

18C6
15C5
C222

TABLE OF ABBREVIATIONS
dimethyltricyclen
trimethyltricyclen
pentamethylpentacyclen
hexamethylhexacyclen

5,12,17-trimethyl-1,5,9,12,l7-penta-
azabicyclo[7.5.5]nonadecane

18-Crown-6

15 -Crown-5

Cryptand [2.2.2]

Nuclear Magnetic Resonance
Magic Angle Spinning
Quadrupole Coupling Constant

MegaHertz

Chemical shift referenced to M+(aq) at infinite
dilution

Spin-lattice relaxation time
Spin-spin relaxation time

Differential Scanning Calorimetry
kiloGauss
Electron Paramagnetic Resonance

Temperature

XV

I. INTRODUCTION

Since Weyl first discovered that alkali metals dissolve in liquid
ammonial, these solutions have been of interest to chemists. The
dilute blue metal ammonia solutions, which become bronze upon an
increase in mole percent metal, are intriguing. Not only do these
solutions provide information about the metal-nonmetal transition,
but they also allow the investigation of the solvation of a
fundamental particle, the electron.

The description “solvated electron” was first applied by Kraus2
and was later elaborated on by Jortner.3v4 Iortner described the
metal ammonia system with the electron density of the solvated
electron symmetric about the center of a cavity of radius 3.2 to 3.4
angstroms in the solvent. The liquid ammonia in the vicinity of the
cavity is polarized by the electron. The major species present in
dilute solutions are the solvated electron and the metal cation

formed according to the following equation:

W + NH3 —" M+(solv.) + °’(solv.) “-1)

Over the intermediate concentration range, I to 7 mole percent
metal, a metal-nonmetal transition takes place with changes in the
electronic, thermodynamic and mechanical properties of the metal
ammonia solutions.5 The transition to metallic behavior as the metal
concentration is increased is not unique to metal-ammonia systems.

The transition to metallic behavior has been observed in both

1

2

supercritical fluid mercury and cesium.5v7v3v9o10v11 In the case of
metal-ammonia systems the change is a first order Mott transition in
which one can consider the metal to non-metal transition as
essentially due to a competition between the kinetic and potential
energies of electrons. While the potential energy localizes electrons,
the kinetic energy delocalizes them until the metal concentration is
high enough to permit significant overlap of the electronic
wavefunction.12

In the lithium-ammonia system, a eutectic point occurs at 20
mole percent metal and 89 K.13 A great deal of evidence has
accumulated which suggests that a compound, Li(NH3)4, is formed at
or near this composition.14 Upon freezing solutions of stoichiometry

Li(NH3)4, a homogeneous, bronze-colored metallic solid results which

is, in fact, polycrystalline. In particular, the powder X-ray diffraction
patterns obtained for this solid are neither those of Li nor NH3, a

verification of the formation of the compound, Li(NH3 )4.15 The
original work in ammonia has been expanded to include other
solvents, in particular, methylaminc.

Lithium is soluble in methylamine up to 20 mole percent metal,
where the metallic liquid compound Li(CH3NH2)4 is fortned16vl7,

and freezes at 155 K. Li(CH3NH2)4 appears to be right at the metal-

nonmetal transition, and in fact goes from metallic to nonmetallic as
the temperature is lowered.18 Lithium methylamine solutions are
the only nonammonia metal solutions in which the metal solubilities
are large enough to permit a transition to the metallic state. The
increase in mole percent metal at the transition from 1 to 7 mole

percent of lithium in ammonia to 15 to 20 mole percent in

3

methylamine is in accord with the expectation that the larger size of
Li+(CH3NH2)4 compared with that of Li+(NH3)4 results in weaker

electron-electron overlap.

The compound, Li(CH3NH2)4, has been shown to assist the
solubilization of lithium and sodium metals in various amine and
ether solvents.19 Upon addition of diethylether (DEE),
trimethylamine (TMA), or dimethoxyethane (DME) to Li(CH3NH2)4,
dark blue solutions were initially formed which became light blue

over a period of minutes.
Li(CH3NH2)4, which is believed to form Li+(CH3NH2)4 and

c1801“) upon dilution, increases the solubilization of Na° in

methylamine. This is believed to follow the reaction:

61801“) + Na(s) —"'> Na' (1.2)

The solvation of alkali metals in tetrahydrofuran (THF) and
diethylether (DEE) can also be increased by the use of crown ethers
which complex the cation, as reported in 1970 using dicyclohexano-
18-crown-6.20 The extension of this work to other solvents, metals,
and macrocyclic or macrobicyclic polyethers and polyamines led to
the synthesis of alkalide and electride salts.21t22t23

Alkalides are ionic salts that consist of an alkali metal cation,
complexant, and alkali metal anion. Electrides have an alkali metal
cation, but the anion is a trapped electron. In both cases, the cation

is complexed by macrobicyclic or macrocyclic polyethers or

4

polyamines.24v25v26 The structural formulas of 18-crown-6,
cryptand [2.2.2] and pentamethylpentacyclen are shown in Figure 1.

/'_\ I
(0—0 0‘} “sci—fl“ :7—CH3
{.0qu C{_IN\_/l\l\

CH3

l8-Crown-6 Pentamethylpentacyclen

Cryptand [2.2.2]
Figure 1. Representative Complexants.

Alkalides and electrides are extremely air and temperature
sensitive and therefore must be handled at low temperatures
(T<253 K) and under inert atmosphere or vacuum.27123929
Syntheses of alkalides and electrides consist of the introduction of
stoichiometric amounts of alkali metals and complexant in a helium

glove box into a synthesis apparatus. The synthesis cell is then

5

placed on a vacuum line, evacuated, and the metals Cs, Rb, K, or Na
distilled.30 Lithium is manipulated directly in a helium glove box
and never distilled.31 A nonreacting solvent such as dimethylether
or methylamine is then added by vacuum distillation to dissolve
either the complexant, alkali metal or both. The solution is usually
kept at temperatures of 253 K or below throughout this process. The
powdered alkalide or electride sample is then obtained by removal
of solvent or the addition of a cosolvent such as trimethylamine or
diethylether to precipitate crystals. Once all solvent has been
removed, a washing solvent is added and excess complexant is
removed through several cycles of washing and filtrations.32v33v34

It has been known for some time that alkali anions are stable
in the gas phase, the electron affinity of the sodium atom is +0.54
eV.35’36 Stabilization of the alkali metal cation by complex
formation allowed the first sodide, Na+(C222)Na', a stable salt with a
sodium anion, to be synthesized.33

The characterization of alkalides and electrides includes
pressed powder d.c. conductivity28v37, EPR spectroscopy, magnetic
susceptibility”, X-ray crystallography39v40t41, and optical '
absorption spectra of thin films produced by rapid solvent
evaporation42943t44 or vapor deposition.“ The optical absorption
specuum has been used to identify the alkali metal anion in
heteronuclear alkalides since it has been shown42 that the peak
position in M+(C222)M' corresponds closely to that observed for the
respective anion in solution. Na', K‘, Rb', and CS‘ in M+(C222)M'
have absorption peaks at 15400, 11900, 11600 and 10500 cm'1

respectively, while in ethylenediamine Na‘, K', Rb', and CS' have

6

comparable values of 15400, 12000, 11200, and 9800 cm‘lf”6 Due
to the sensitivity of the peak position to the surroundings,
assignments are not completely unambiguous for the identification of
heteronuclear films of the stoichiometry MLNa where L is the
complexant.47

In contrast, alkali metal anions can be unequivocally identified
in the solid state by NMR methods due to their unique chemical
shifts. In the absence of unpaired electrons, the primary contributor
to the chemical shift of the alkali metal nuclei, M+ or M' (except
lithium), is one of the paramagnetic terms in the general Ramsey
shift expression.48 Values obtained are relative to the values for the

isolated gaseous ions. The paramagnetic term is:

‘I’,> (1.3)

in which AB is the average outer p electron excitation energy, tyo is

 

 

2 ll.
- -e kk
up 22 <W°|z° 3

chE 3.3 rk

the groundstate wave function, I]; and 1]" are orbital angular
momentum operators, and r]; is the distance of the kth electron from
the nucleus. The Ramsey shift is the result of the introduction of
orbital angular momentum due to the interaction of the filled outer p
shell with electron pairs from surrounding molecules or ions. Only
nearest neighbor electron-pair donors contribute significantly,

therefore the large radius of the alkali metal anion and its weak

7

interactions with surrounding electrons make the Ramsey shifts of
Na' and K“ negligible. While not negligible, the Ramsey shifts of Rb'
and C3' are much smaller than for the cations.

Static solid state NMR generally results in rather broad
featureless spectra for most powders. Absence of sufficient nuclear
motion in crystalline solids results in substantial broadening of the
NMR line relative to those in fluids. The line broadening is the result
of a distribution of crystalline orientations and is due to magnetic
dipolar and quadrupolar interactions as well as chemical shift
anisotropy.

A technique called magic angle sample spinning nuclear
magnetic resonance (MAS-NMR) was introduced by Andrew and
Lowe in 19594950951 In some cases when this technique is used,
one can obtain solid state spectra that rival those obtained in
solution. Line narrowing is achieved by rapidly spinning a sample
about an axis inclined at or near the “magic angle” of 54.740
to the Zeeman field. By rotating the sample about the magic angle,
the first order quadrupolar, dipolar and chemical shift interactions
can be removed from the center of the spectrum and appear as
spinning side bands.

In the static spectrum of Na"’(C222)Na', a broad peak of width

2600 Hz centered at -60 ppm relative to Na‘l' ) is observed with a

(aq.
barely discemable broad shoulder on the low-field side of this
peak.52 Figure 2 demonstrates the dramatic reduction in line width
and the resultant improvement of the signal to noise ratio for the
same sample of Na+(C222)Na' run on a Brukcr CXP 200 spectrometer

in the nonspinning and spinning modes.44 The chemical shift of

 

 

 

 

 

 

 

 

 

+ J 1 L LL 1 1

L
4400 O 40? -200
8 [ppm FROM uo’tqu]

 

Figure 2. Magic angle sample spinning 23Na NMR spectrum of
Na+(C222)Na'. Also shown is the nonspinning spectrum
of the same samples2

9

Na+(C222) at -23.7 ppm is clearly resolved. The substantial
paramagnetic shifts of Na+ in crystalline salts from that expected for
Na+® is caused by interactions with electron pairs on atoms in the
cryptand with the p orbitals of Na+.52953

Through a combination of the results of static and MAS studies
at several frequencies, it is possible to determine the sources of line
broadening. This may be achieved by lineshape analysis of the static
spectra and proton decoupling experiments. In the compound
Na+(C222)Na‘, as for all the quadrupolar alkali metal nuclei, only the
central transition (1/2 <—> -1l2) can be observed since the central
transition is unshifted to first order, unlike the other transitions.
With second-order perturbation theory, the shift of the central

transition with frequency can be represented by the equation54:

V0 2
v - VL - [I(I+1)- 3/21(1 +11 l3) (1.4)

30vL

where V1. is the is the isotropic Larmor frequency, VQ is the nuclear

quadrupolar frequency, I is the spin of the nucleus and 11 is the
asymmetry parameter. Figure 3 shows the proton coupled and
decoupled 23Na MAS-NMR spectrum of Na+(c222)Na- and the
variation with frequency.55 The inset is that of the proton-
decoupled peak of Na"'(C222) (solid line) overlayed with the
calculated lineshape for 2vQ = e2qQ/h = 1.2 + 0.1 MHz (dashed line).

The value of the quadrupole coupling constant is an indication of

10

 

 

 

 

 

52,94 Ill-Ix o'I-I-couuod

 

1 I 13235 mi: - 'H-eovmd
92.29 MHz-'H-ceupled

L1

 

52.94 MHz-'H-«eoupled

 

Figure 3. 23m MAS-NMR spectra of Na+(c222)Na- at three
frequencies. Inset shows an expansion of a portion of the

bottom spectrum due to Na+(C222) (solid line) and its
computer simulation (dashed line).55

11

why the observed chemical shift of Na+(C222) varies with frequency.
The complexed cation has an isotropic chemical shift of
-6.5 ‘1' 0.5 ppm with respect to aqueous Na‘l’ as zero.

The observed static line width of the complexed cation,
Na+(C222), is 3700 i 300 Hz. The quadrupolar contribution to the
linewidth is 320 Hz with the difference, 3400 i 300 Hz, due to
dipolar coupling between the -CH2- protons on the cryptand and Na+.
The second moment can be calculated from the Van Vleck expression

with atomic distances obtained from the crystal structure.55 The
Van Vleck equation, in which 7“ and 7m are the magnetogyric ratios

and rk is the distance of the kth hydrogen from the sodium cation, is

shown below:

v - — .5)
<A > 20112 N“ H kl It]

For Na+(C222), the calculated Gaussian dipolar contribution to the
linewidth is 3340 Hz, in excellent agreement with that observed.5 5
Studies have been extended to other alkali metal cations and
anions through the use of 87Rb, l33Cs, 7L1 and 39K magic angle
spinning and static NMR techniques.33-52:55r57a53 For the
compound Cs+(18C6)2Rb', Figure 4 demonstrates the ability of alkali
metal MAS-NMR to identify the alkali metal species present.58 In

the 133Cs NMR spectrum, only the peak for the complexed cesium
cation, Cs+(18C6)2, is present and only the Rb' peak is present in the

12

Cs’lIBCSI,

 

 

    

400 200 O - 200 400 -600
Chem Shift from M’taq), ppm

Figure 4. 133Cs (65.61 MHz) and ”Rh (163.6 MHz) MAS-NMR
spectra of polycrystalline Cs+(18C6)2Rb'. 58

13

87R!) spectrum. These chemical shifts, ~57 ppm for the “sandwich”
cation Cs"‘(18C6)259t60 and -187 ppm for the Rb' anion51o52, are
characteristic for these nuclei in the environment described. Table 1
gives values of characteristic NMR shifts of some alkali metal anions
and complexed cations. The conclusion drawn from NMR data, that
this salt of stoichiometry Cst(18C6)2 is a pure rubidide, is
supported by X-ray absorption studies.63 In contrast to the 23Na
NMR studies of alkalides where the quadrupolar and dipolar
interactions dominate, 133Cs NMR central transitions are dominated
by chemical shift anisou'opy and dipolar interactions. The 133Cs
nuclei possesses a small quadrupole moment.

The low sensitivity and large quadrupolar broadening of the
39K and 8'7Rb NMR linewidths have limited the detection of
potassium and rubidium cation signals in alkalides and
electrides”,56 Recently, the first observation of K+ in compounds
other than simple salts was published by Oldfield and coworkers.64
Through the use of spin echo techniques used by Oldfield et. al., it
has been possible to detect potassium and rubidium cation signals in
alkalides and e1ectrides.65.66 Data from the spin echo studies of
rubidium salts will be presented in Chapter 5.

Although NMR has proven to be an excellent tool in the
characterization of alkalides and electrides, the utility of these
compounds has been limited due to their extreme sensitivity to air
and temperature. While extensive use has been made of crown
ethers and cryptands as complexants, all of the compounds studied to

date undergo irreversible thermal decomposition at elevated

14

Table 1. Characteristic NMR Chemical shifts of alkali metal anions
and complexed cations.31

 

Species
Na‘

K'

Rb'

CS'

Na+C222
Cs+(18C6)2
CS+(15C5)2

Cs+(C222)

Cl . l 1.: C :a
~57 to ~63

~105

~185 to ~l98

~212 to ~240 (In Cs+(l8C6)2CS'
temperature dependent)

~7 (High field limit)
~40 to ~62

+24 to +29

+238 to +275 (Inclusive
complex)

 

aChemical shifts are referenced to M+(aq) at infinite dilution.

15

temperatures.67 Studies done with differential scanning calorimetry
(DSC) have indicated that the peak of the trace of the decomposition
exotherm increased with the heating rate68 and this phenomenon
can be used to derive kinetic parameters of decomposition.59v70

In order to overcome the thermal instability, research has been
directed towards the search for more stable complexing agents to be
used in the synthesis of alkalides and electrides. Although the
instability of the crown ether and cryptand complexants is believed
to result from reductive attack on the ether complexant by the
unbound or weakly bound electrons, the cleavage of ethers by alkali
metals is a complex phenomenon.71

Research has led to the investigation of complexants in which
nitrogen donor atoms have been substituted for oxygen in crown
ethers and cryptands. The first complexant of this type to be used
was hexamethylhexacyclen, HMHCY, the fully methylated aza analog
of 18-crown-6.72 The complexant appears to be extremely resistant
to decomposition to temperatures exceeding 393 K, even in the
presence of alkali metals. The compounds synthesized with this
complexant, Cs+(HMHCY)Na', Rb+(I-IMHCY)Na', and K+(HMHCY)Na' do
not undergo decomposition; however, irreversible decomplexation
occurs at temperatures as low as 310 K when the sample is ramped
in a differential scanning calorimeter at 5 Klmin. The
decomplexation correlates with studies done on the complexation of
K+. When nitrogen is substituted for oxygen in a crown ether, the
stability of the potassium complex in methanol decreases with an

increase in the number of nitrogen donor atoms.7 3

l6

Aside from the unique thermal properties of the compounds
synthesized with the complexant HMHCY, the crystal structure of
these" compounds provided the first evidence for contact ion pair
formation between an alkali metal cation and an alkali metal
anion.72 The promising results obtained with the first aza
complexant led to the use of other complexants with nitrogen donor
atoms. The smaller aza ring systems, pentamethylpentacyclen and
trimethyltricyclen, have been studied as complexing agents for the Li
cation.74a75 Since early studies of crown ether ligands reported that
stronger complexation occurs for those metal cations whose ionic
radii best match the radius of the cavity formed by the polyether
ring, lithium was chosen as the cation in these complexes based on
"fit".24 Data will be be presented on the compounds synthesized
with the complexant trimethyltricyclen in Chapter 3. The results of
work that has been initiated in the study of the aza analogs of
cryptands will be discussed as well.

II. COMPLEXED CESIUM CATION
ILA._Intr.c.dllcticn

The chemical shift of the cesium cation is extremely sensitive
to the electron density at the nucleus. Combined with the fact that
the nucleus has a relatively high sensitivity and small quadrupole
moment, this makes the NMR spectrum of 133CS a very useful tool in
the study of the influence of the surroundings on the chemical shift
and line shape of cesium in alkalides and electrides.

The chemical shifts of complexed cesium cations of both the
18C6 and 15C5 alkalide sandwich systems are nearly independent of
the anion and agree with values found for halides and
thiocynates.76-77 The resonance peak of the complex, Cs+(18C6)2
occurs at ~60 1 2 ppm for all salts except the ceside.'76

The relatively constant chemical shift of the cesium cation
coordinated by 12 crown ether oxygen atoms, implies that neither
the su'ucture of the complex nor the electron density at the cesium
cation differ significantly in the sodide, potasside and rubidide salts.
The chemical shift of Cs+, believed to be strongly influenced by the
degree of Cs-O overlap, is inversely related to the mean Cs-O

distance.76
Although the electride, Cs+(18C6)2e‘, is nearly isostructural to

the sodide, Cs+(l8C6)2Na' 41’73, the 133Cs MAS NMR spectra of the

electride consists of a single temperature dependent peak at +81

ppm, a significant paramagnetic chemical shift compared to that of

17

l 8
other Cs+(18C6)2 complexes.76 The origin of the paramagnetic
chemical shift in Cs+(18C6)2e' is the Knight or contact shift.76

' The fractional atomic character of the cesium ion in
Cs+(18C6)2e' may be calculated from the magnetic susceptibility and

the slope of the Knight shift versus l/T for this Curie-Weiss
paramagnet. The trapped electron in Cs+(18C6)2e' has a very low
cesium s~orbital character relative to other electron-cesium systems
such as Cs-ammonia solutions”:80 The fractional atomic character
of Cs+(18C6)2e‘ is 3.3 X 104, while Cs/ammonia solutions that
contain 2 to 20 mole percent metal have fractional atomic characters
that range from 0.04 to 0.60.

The chemical shift of Cs+(18C6)2e' extrapolated to infinite
temperatures is ~61 t. 10 ppm, which is within experimental error of
the chemical shift of Cs+(18C6)2 observed in alkalides. The plot of
chemical shift versus VT for Cs+(18C6)2e' is shown in Figure 5.

The study of Cs+(l8C6)2e’ has yielded evidence that the trapped

electrons in this compound show only weak overlap with the
complexed cation, as evidenced by the Knight shift, as well as the

absence of any spin pairing or magnetic ordering. In order to
compare the Cs+(l8C6)2e' data to other electrides, a study was

undertaken of the Cs+(15C5)2e‘ system in which the cesium is

coordinated to 10 crown ether oxygens. The crystal structure of this
compound is known and the magnetic susceptibility obeys the Curie-
Weiss law above approximately 15 K. In contrast to Cs+(18C6)2e'

with no magnetic ordering, annealed polycrystalline samples of
Cs+(15C5)2e‘ show an antiferromagnetic transition at 4.6 K.

19

240

 

g

A (1 A A j 1 a
I 2 " 3.0 3.8 4.0 4.3 5.0 53 6.0
In Iqu’I

Figure 5. Plot of chemical shift versus lfI' for Cs'*‘(18C6)2¢'.76 The

temperatures of points marked with an x were calibrated
with methanol.

20

Also of interest was a system intermediate between the pure
Cs+(l8C6)2e' and Cs+(15C5)2e' systems. Syntheses of the mixed
crown ether sandwich compounds in which the cesium cation is
complexed by a 15C5 molecule and an 18C6 molecule,
Cs+(l8C6)(15C5), were attempted and the resultant compounds
characterized. The investigation of mixed crown ether sandwich
compounds offered the potential to fine tune the characteristics,
such as magnetic susceptibility or alkali metal NMR chemical shift
values, of an alkalide or electride based on the appropriate choice of

crown ethers and the alkali metal cation.

Menu

The primary contributor to the chemical shift of the alkali
metal nuclei in the absence of unpaired electrons is one of the
paramagnetic terms in the Ramsey shift.48 This paramagnetic term
is given in equation (1.3). The origin of the extra paramagnetic
chemical shift in Cs+(15C5)2e' is a contact or Knight shift. The Knight
shift is the result of a strong local magnetic field generated at the
nucleus of interest by paramagnetic electron density. Chemical shifts
exist in the atoms themselves with conu'ibutions from the ion cores
and from surrounding atoms, but changes in chemical shifts from one
compound to another are usually much smaller than the Knight shift.
The cesium cation should be very sensitive to any overlap with the
paramagnetic trapped electron since the lowest energy unoccupied
orbital in Cs+ is the 6s orbital with a non-zero electron density at the

nucleus. The Knight shift is given by“:

21

2
K(T)- 611 <| \P(o)| > X(T) (2.1)
BNaV

where (I “’(0) l2) is the average electron density at the nucleus, and

 

x(T) is the magnetic susceptibility of the paramagnetic species. .In
the case of a Curie-Weiss paramagnet, K(T) is proportional to UT
since x(T) varies inversely with temperature.

The value of the chemical shift in ppm at low temperatures fits

the equation:

8(1) = 5((13) + K(T) (2.2)

Where 8(T) is the chemical shift at a specific temperature, T, and
5(00) is the chemical shift at infinite temperature. The fraction of
atomic s-orbital character of the trapped electron can be calculated
from information obtained from the slope of the Knight shift versus
UT and the magnetic susceptibility. The fractional atomic character

is defined as:

22

2
< “‘ol >
F ' (2.3)
2
< “‘ >
atom

o
where (I W0 I2) “om is the electron density at the nucleus for an

 

 

 

 

isolated gas atom and has been estimated to be
2.645 x 1025 e-cm-3.79

Although the overall magnetic behavior of Cs+(15C5)2e' is that
of a Curie-Weiss paramagnet, the magnetic moment of a free atom
has three principle sources. The spin that the electrons are“ endowed
with and their orbital angular momentum about the nucleus yield
two paramagnetic contributions to the magnetization while the
change in the orbital moment induced by an applied magnetic field
gives a single diamagnetic contribution. The magnetic susceptibility

per unit volume is defined as:

X - - (2.4)

where H is the macroscopic magnetic field intensity and M is defined
as the magnetic moment per unit volume. Substances with negative
magnetic susceptibility are called diamagnetic, the tendency of

electrical charges partially to shield the interior of a body from an

23

applied magnetic field. Substances with positive susceptibility are
paramagnetic. Paramagnetism is a property exhibited by substances
that contain unpaired electrons. Paramagnetic susceptibilities are
temperature dependent and to first order (high T) approximation,
the susceptibility x varies inversely with temperature. For a spin

1/2 system, the magnetic susceptibility is given by:

22
X.M..§_g_£'£.£ (25)
H 41st T

which is in the form of the Curie law where the Curie constant is

given by:

N 21,2
c ~ —§.—2 (2.6)
4k

where N is Avogadro’s number, k is the Boltzman constant, [.13 is the

Bohr magneton and g is a constant characteristic of each system. For
the case J=S=1/2, the value of the Curie constant is 0.376. This
applies to a system that consists of 100 percent independent
unpaired spins in which the orbital angular momentum is quenched.
Deviations from Curie law behavior occur because of magnetic

interactions that occur between paramagnetic species. In the

24

simplest approximation, this behavior is expressed by the Curie-

Weiss law:

X s (2.7)

 

T-G

where the correction term, 9, has units of temperature. The value of
9, the Weiss constant, can be used to indicate whether a material is
purely paramagnetic (9 =0), antiferromagnetic (6<0) or ferromagnetic
(9>0).

A ferromagnetic state is one in which the moments on a given
lattice are aligned spontaneously in the same direction. Even in zero
applied magnetic field, a ferromagnet has a nonvanishing magnetic
moment. Magnetization is at the maximum possible value at T=0. As
temperature is increased, magnetization decreases as the spins
become uncorrelated until spontaneous magnetization becomes zero
at a characteristic temperature, 8, the Curie temperature. The
susceptibility of a ferromagnet obeys the Curie-Weiss law at
temperatures where T >> 6.

A typical antiferromagnet consists of two interpenetrating
sublattices. Each sublattice is uniformly magnetized with its spins
aligned parallel, and the spins on one sublattice are aligned
antiparallel to the spins on the other sublattice. The transition from
antiparallel nonalignment to parallel alignment of neighboring spins

is a cooperative one that is accompanied by a characteristic long-

25

range ordering temperature called the Neel temperature, 9.
Characterized by a decrease in susceptibility with increasing
temperature, the spins are ordered below the Néel temperature. At
temperatures well above the Néel temperature, the susceptibility

follows the Curie-Weiss law.
W11
mm

The synthesis of alkalides and electrides has been described in
detai129t34, and these techniques were used in the synthesis of
Cs+(15C5)2e'. Approximately 20 percent excess crown ether was
used in the synthesis in order to avoid contamination of the
compound by ceside. The mixed crown ether compounds,
Cs+(18C6)(15C5)Na‘ , Cs+(18C6)(15C5)e’ and Cs+(18C6)(15C5)Cs'
were synthesized under more stringent conditions.

The synthesis of the mixed crown ether compounds required
stoichiometric amounts of each crown ether be added to the
synthesis apparatus, a K-cell. A very slight excess of the crown ether
15~crown~5 was added since this complexant is the more volatile of
the two complexants used and a small amount can be lost during the
flame seal-offs on the K-cell, which were made under active vacuum.
This problem was minimized by freezing the complexing agents in
liquid nitrogen or dry ice during the flame seal-offs on the cell.
'Dimethylether was then added to the metal(s) mirror and crown

ethers to allow the complexation of the alkali metal cation. When all

26

of the complexant had been dissolved and no more of the metal
mirror would visibly dissolve over a two to three hour period, the
cosolvent trimethylamine was added. No crystals were immediately
visible; however, the slow removal of the solvent and cosolvent over
a six hour period yielded an apparent homogeneous phase. Rapid
precipitation of crystals from the synthesis must be avoided since
this leads to multiple l33Cs MAS NMR lines, an indication that more
than one phase is present. The crystals, once precipitated, were
washed with trimethylamine then harvested.

Attempts were made to grow single crystals for X-ray structure
determination. Crystals of Cs+(18C6)(15C5)Na' and
Cs+(l8C6)(15C5)e’ were grown from saturated solutions of
dimethylether and trimethylamine by slowly cooling the
temperature of the solution from approximately 233 K to 203 K over
a three day period. Slow solvent evaporation of a solution of the
sodide in dimethylether and trimethylamine over a period of several
days was also attempted. Crystals of suitable quality to perform

X-ray structure determinations were not obtained.

LEMR

Cs+(15C5)2e'

. Magic Angle Sample Spinning (MAS) 133Cs NMR spectra were
obtained at 23.6 MHz on a Bruker WH 180 MHz (proton frequency)
NMR spectrometer operating at a field strength of 4.2277 T, with 4-5
psec pulse lengths and 0.5 to 2 second delay times. The temperature

27

was controlled by flowing the nitrogen spinning gas through a liquid
nitrogen heat exchanger and then warming the gas with an in-line
heater. After the samples had been loaded into the rotor in a
nitrogen-filled glove bag at liquid nitrogen temperatures, the
sapphire rotor that contained the sample was loaded into a precooled
NMR probe. Although the temperature was monitored with a
thermocouple placed in the gas stream at a point just ahead of where
the gas entered the Aspinner, it proved to be difficult to determine the
actual sample temperature. Unfortunately, there seem to be only a
few reliable methods for the determination of the sample
temperature, especially the temperature of the sample where the
NMR signal is detected.82

The temperature can be obtained by measuring the
temperature-dependent chemical shift of a standard sample such as
acidified methanol. At low temperatures, more hydrogen bonding
occurs and the 1H peak of the OH group in methanol shifts downfield.
The chemical shift, Av in Hz, between the CH3 and OH groups can be
fit with the following quadratic equation with an error (RMS) of 0.8 K
over the temperature range from 175 to 330 K:8 3

r=403.0 - 0.491 I Av I - 6.62 (10'3 Av )2

In order to incorporate the methanol sample into a MAS sapphire
sample rotor, the Kel-F bolt used to attach the end caps to the rotor
was modified. The Kel-F bolt was drilled lengthwise into its center

and the hollowed portion filled with a sample of methanol and sealed

28

with epoxy. A schematic diagram of the sapphire rotor and bolt is

shown in Figure 6.

/ Hollow central bolt

g'

CPD

 

Sapphire rotor

 

 

 

 

 

(:4;

Figure 6. A schematic diagram of the sapphire rotor and bolt.

This method allowed alternate 133Cs and 1H measurements to
be made without removing the sample rotor from the instrument.
This provided a reliable and reproducible measurement of the
sample temperature although the presence of temperature gradients
within the sample holder could not be ruled out.

Static NMR measurements were also carried out at 52.5 MHz in
a Varian VXR~400 spectrometer equipped with a 45 to 165 MHz
broadband probe by using single pulses of width 0.5 psec and a

delay time of 0.5 seconds. Cold nitrogen gas was used to control the

29

temperature to within 0.3 K after a 3 to 10 minute equilibration

time.

Cs+(l8C6)(15C5)X'

The MAS and static 23Na and 133Cs NMR spectra of the mixed
complexant alkalides and electride were obtained at 47.6 MHz and
23.6 MHz respectively, with a Bruker WH 180 MHz (proton
frequency) NMR spectrometer operating at a field of 4.2277 T with 3
to 5 11sec pulse lengths and 1 to 2 second delay times. A variable
temperature Doty dual—bearing MAS probe was used and samples
were loaded into the precooled probe once the sample had been
loaded into the rotor in a nitrogen filled glove bag at liquid nitrogen
temperatures. The temperature was controlled by flowing the
nitrogen spinning gas through a liquid nitrogen heat exchanger and
then warming to the set temperature with an in-line heater. The
temperature was monitored with a thermocouple placed in the gas

stream at a point just ahead of where the gas entered the spinner.

3 ll . S 'l'l'

Magnetic susceptibilities of Cs+(15C5)2e'.samples were

measured with an S.H.E. 800 Series SQUID magnetometer over a
temperature range of 2 to 283 K in applied fields of 0.1 to 7 k6. The
mixed crown ether alkalide and electride complexes were studied

over a more limited temperature range of 2 to 220 K in applied fields

30

of 0.25 to 7 k6 for the electride and 5 k6 for the alkalides. Samples
were loaded into a Kel-F SQUID bucket in a nitrogen filled glove bag
at liquid nitrogen temperature, then transported and loaded into the
susceptometer at cryogenic temperatures while under inert
atmosphere.

Once the susceptibility of the “live” sample had been measured,
the background susceptibility was determined and subtracted from
that of the live sample. The background susceptibility was obtained
by allowing the sample to decompose to a diamagnetic product and
rerunning the sample. This procedure allows one to obtain the
electronic contribution to the susceptibility. Since the decomposed
samples are diamagnetic and provide a direct measure of the atomic

susceptibility, the electronic contribution results only from M' or e'.

ID'EE 'lS . Cl'

Differential scanning calorimetry (DSC) was performed with a
DuPont 910 Differential Scanning Calorimeter. Samples were
hermetically sealed in aluminum pans under nitrogen gas at
cryogenic temperatures. The samples were transported and loaded
into the DSC instrument under inert atmosphere at cryogenic

temperatures.
5 Q . I ll .

Optical absorption spectra of thin solvent-free films of the
electride Cs+(15C5)2e' were measured from 700 to 2300 nm on a

31

Model 260 Guided Wave Spectrophotometer at temperatures from
164 to 273 K as recorded with a copper-constantan thermocouple.
Thin films were made by preparing a solution of the compound in

CH3NH2, pouring the solution over the quartz cell window and then

rapidly evaporating the solvent.
II D B l l D' .
W22“

The details of the crystal structure of this electride have been
reported elsewhere.84 It belongs to the triclinic space group PI with
one molecule per unit cell and cell parameters: a = 8.59(4), b =
8.886(8), c = 994(1) A; a = 102.91(8), B = 9006(6), y = 9.4(6)°; v =
733.1(11)A3, p = 1.30g cm'3. The structure was determined at a
temperature of 215 I. 2 K. Below about 260 K powdered samples of
the electride are polycrystalline and can be readily transferred from
one container to another by pouring. As the temperature in
increased, particularly above 266 K, the sample became ”sticky" but
remained solid. As shown in Figure 7, DSC reflects these changes in
behavior. A pronounced endotherm at 266 K corresponds to AH =
11.2 1' 0.6 U mole'1 and is reversible. When the sample that had
been warmed to 283 K in the DSC was cooled to 230 K the same
endothermic transition occurred during the second cycle. The
position of the exothermic peak, due to irreversible decomposition,
depends on the ramping rate. The temperature at which

decomposition occurs increases with an increase in ramping rate. As

HEAT FLOW (WI g)

32

 

 

 

 

 

 

 

Figure 7.

I I 7 I I

1
303 323

TEMPERATUREGQ

Differential Scanning Calorimetry (DSC) trace of
Cs+(15C5)2e' at remping rates of 2 Klmin (lower trace)
and 5 Klmin (upper trace ofset by one Wig). .

 

33

the ramping rate is increased, an endothermic peak becomes visible
at 295 K, immediately preceding irreversible decomposition. This
peak is believed to be due to the melting of Cs+(15C5)2e'. The
endothermic melting transition is masked by the exothermic
decomposition at slower ramping rates.

While the molar enthalpy change for the endothermic
transition at 266 K can be measured as 11.2 1 0.6 k] mole'l for all
ramp rates, glass formation prevents the direct determination of the
heat of fusion of 15C5. However, both 12C4 and 18C6 can be
crystallized and have heats of fusion of 20 1 l and 37 1 2 k] mole"1
respectively. The average, 28 kJ mole'1 is assumed to be
approximately the heat of fusion of 15C5. Thus, the observed
transition of Cs+(15C5)2e' has a value of AH that is about 20 percent
of that expected for complete melting of two moles of 15C5. Enough
energy is involved in the endothermic transition at 266 K for the
compound to undergo structural changes, but these changes are
smaller than those associated with melting.

The absorption spectra of thin films of Cs+(15C5)2e' are more
complex than those of other electrides which contain a single broad
absorption in the near IR region at 1200 to 1600 nm.43~44t35»35
Shown in Figure 8, the absorption spectra of thin films of
Cs+(15C5)2e’ contain three peaks at 900 to 950 nm, approximately
1200 nm and 1600 to 1700 nm. At lower temperatures, the three
absorption peaks are quite distinct. An increase in temperature
appears to broaden each of these peaks. No change in the absorption
spectra was observed at or above the phase transition at 266 K.

However, once the temperature of the film had been taken above

I‘M?“ norms-x"

Figure 8.

 

34

 

 

 

 

 

 

1.5 1.! I

u l 1.2 .
(Thousands
“"1116?! (an

The optical absorption spectrum of a thin film of
Cs+(15C5)2e‘ made from solvent evaporation of

methylamine at temperatures of 188 K (diamonds) and
253 K (solid line).

35

that of the phase transition, the narrowing of the three distinct peaks
was no longer reversible when the temperature was lowered. The
appearance of three absorption peaks instead of a single broad
absorption may be due to a degeneracy in the excited-state p
orbitals; an s to p transition would then produce three distinct peaks
as observed.

The magnetic susceptibility of Cs+(15 C5 )ze' was studied over
the temperature range 1.6 to 283 K and at field strengths from 0.1 to
7 kG.87 Samples that were quenched by sudden cooling to 5 K
showed Curie-Weiss behavior from 5 to 260 K with only small
deviations below 5 K. When a sample was ”annealed” at a
temperature of approximately 230 K for a few minutes, followed by
slow cooling, a maximum in the suséeptibility was observed at 4.5 K.
The susceptibility data show that an antiferromagnetic transition
occurs at about 4.5 K and that the anisotropy is relatively small. At
temperatures well above the Neel temperature, the susceptibility
followed the Curie-Weiss law. However, at about 260 to 265 K the
slope of a plot of I/x vs. T changed in a direction that corresponds to
less paramagnetism. This occurred just below the temperature of the
endothermic transition seen in the DSC experiments. Whether an
abrupt change in susceptibility accompanies the phase transition
cannot be ascertained from the susceptibility data since the sample
temperature in the SQUID susceptometer tends to overshoot before
coming to equilibrium. In fact, high temperature susceptibility data
needed to be corrected for partial decomposition of the sample due

'to the overshoot in temperatures in the instrument. Thus, partial

36

conversion to a less paramagnetic high temperature form could have
occurred.

' A large number of NMR experiments have been carried out on
Cs+(15C5)2e'. Both the early l33Cs MAS measurements carried out
by Steve Dawes that were hampered by difficulty in measuring the
sample temperature, and later MAS measurements with more
accurate temperature calibrations showed two peaks with chemical
shifts of 490 and 290 ppm. Both peaks were present over the
apparent temperature range 265 to 280 K, each at a constant value
of the chemical shift as shown in Figure 9. These measurements had
to be made rather rapidly to avoid decomposition, a process that
could be monitored by the growth of a peak of the decomposition
products at 61 ppm. Since the DSC results showed a reversible
endothermic transition at 266 K, the phase transition may not have
been complete in the NMR rotor at these temperatures during the
measurement time and the actual sample temperature may have
been at or near the transition temperature even though the
temperature of the spinning gas continued to increase. The heat
absorbed during the transition would tend to keep the temperature
constant. That this was indeed the case was shown by studying the
time-dependence of the static spectrum at 266 K. Over a 120 minute
period the static peak with an isotropic chemical shift of 488 ppm
decayed while that at 294 ppm grew in. Rapid cooling to 253 K
caused reversal of the process over a period of 3 to 5 minutes. These
results are in complete agreement with the DSC data and confirm the
presence of a first-order phase transition at 266 K. In the
temperature range of 280 to 290 K only one peak, corresponding to

37

 

 

 

 

 

 

44441 AJLI+LJALLJ AALJ kLA 114141 1 J l 1 LL 1

1000 500 0
PP'“

Figure 9. l”Cs MAS NMR spectrum at a temperature of 276 K
with two peaks at chemical shifts of 493 and 298 ppm
obtained at a frequency of 23.6 MHz.

38

the new high temperature peak observed in the 265 to 280 K range,
was present in both the MAS and static NMR data.

' The static 133Cs NMR spectra at 52.5 MHz were also obtained
as a function of temperature. Typical gaussian-broadened powder
patterns arising from chemical shift anisotropy (CSA) were obtained.
The isotropic chemical shift ranged from 604 ppm at 183 K to 496
ppm at 258 K, in excellent agreement with that obtained from MAS-
NMR measurements at 23.6 MHz. The anisotropy parameter, “CS A'
was temperature dependent, with values of 0.30 1 0.05 from 266 to
243 K. The chemical shift anisotropy parameter increased with
decreasing temperature with a value of 0.60 1 0.05 at 213 K and
0.65 1 0.05 at 183 K. The required gaussian broadening
corresponded to full widths at half maximum that ranged from 3.3 1
0.2 KHz at 183 K to 1.45 1 0.1 KHz at 243 K. Since the proton dipolar
broadening as calculated from the crystal structure with the Van
Vleck equation was 1.1 KHz38, much of the broadening can be
attributed to incompletely relaxed electron-nucleus dipolar coupling.
The rigid-lattice limit for such coupling was calculated to be 57.4
KHz. Electron spin-lattice relaxation averages out most, but not all of
the strongly temperature dependent electron-nucleus coupling.

Below about 240 K, the chemical shift is linear in III with an
intercept of 180 1 10 ppm and has a slope of 7.9 1 0.5 x 104 ppm K.
Figure 10 shows a plot of chemical shift vs l/T for Cs+(15C5)2e' for
both the MAS and static NMR data. This intercept is about 150 ppm
higher than the chemical shift obtained for model salts that contain
the Cs+(15C5)2 cation.76 The slope corresponds to a contact density

at Cs‘l’ that has 0.063% atomic character, nearly twice that found for

39

 

 

 

 

fl
0
U. O
.t D U
or moon? 0 a
Do 0
.t 58- 6°
I 8% 8
t
" and - 0 9.
1| “0...!
P:
i at-
3 at
I
I
I
o a.
"Mail’s
a I I I I I I I I I I I
mu 0“ I.“ m I.“ m m
mmumr

Figure 10. Plot of chemical shift versus 1/1‘ for Cs"'(15C5)2e'; 180

MHz MAS NMR data (open squares); and 400 MHz static
data with the isotropic chemical shift obtained from
simulations of the experimental powder patterns (circles).

40
Cs+(18C6)2¢’ 75 The rather large anisotropy in the Cs+ to e'

distances suggests that the contribution to the contact density is
different for different electrons.

The similarity in geometry and Cs-O distances in the electride
complex when compared to the sodide and the small electron spin
density at cesium strongly suggest that the excess electron density is
centered in the anionic sites with probable extension of the wave
function beyond the Van der Waals boundaries of the cavity.

The endothermic transition at 266 K, accompanied by a
decrease in paramagnetism that is sluggish, made it impossible to
distinguish between a discontinuity in the susceptibility and a
change in the s10pe of lfx vs. T as shown in Figure 11. One can
conclude, however, that the electronic susceptibility 10 K above the
transition is about 60 percent of the value 10 K below the transition.
The chemical shift of the 133C8 NMR signal decreases abruptly from
493 ppm below the transition temperature to 294 ppm above this
temperature. The most likely cause of the change is the decrease in
magnetic susceptibility since the chemical shift of a diamagnetic
model salt is directly proportional to the susceptibility for a constant
value of the unpaired spin density at Cs+. The decrease in the
chemical shift at the transition, 57 percent, is very close to the 60
percent decrease in susceptibility, an indication that the transition to
the higher temperature phase is accompanied by partial electron
spin-pairing.

A remaining puzzle is the high value, 180 ppm, of the
extrapolated chemical shift of the low temperature phase as well as

the curvature of a plot of the shift vs. l/T. This contrasts with the

41

.5... 4.59858 an... a... Co .35 a. a? -ouaoncto
8. L. seae aux: Co a... 3.228.... ocean-z .: seam

 

 

 

 

 

 

9:.
con ova on" concur amp o3. our 3.. an co 2. an a
_ _ _ _ _ _ _ _ . _ . _ _ o
luan
)
e...
I100
let...“
m
Imm
.. w
em.
a
In

 

42
chemical shift behavior of Cs+(18C6)2e’, whose intercept at infinite

temperature is the same as that of diamagnetic model compounds.7 5
The high intercept in the present case may reflect a gradual increase
in the contact density at CST that accompanies the increased motion
of the crown ethers. Of particular interest in the study of this
compound would be 133Cs MAS NMR at temperatures of 180 K and
below, especially at temperatures corresponding to the

antiferromagnetic transition.

2 93+118961115951Na' and 93+“ SEQ)“ 59:195-

Although several methods have been used in attempts to grow
single crystals of the compounds Cs+(18C6)(15C5)e' and
Cs+(l8C6)(15C5)Na', a crystal of suitable quality for the collection of
single crystal X-ray diffraction data could not be found. The first
synthesis of a cesium mixed crown ether sodide system was
characterized by 133Cs MAS NMR and contained multiple NMR
peaks. A more careful synthesis yielded a single 133Cs MAS NMR
peak at ~11 ppm. In Table 2, the values of the 133Cs chemical shift
in some pure Cs+(15C5)2 and Cs+(l8C6)2 complexes are listed along
with those of the Cs+(l8C6)(15C5) systems synthesized to date.

The chemical shift of approximately 90 ppm downfield of the
Cs+(15C5)2 salts from that of the Cs+(18C6)2 salts suggests a stronger
paramagnetic Ramsey interaction of the cesium cation with the
oxygens of the crown ether 15C5 versus those in 18C6. In the

su'uctures of the Cs+(15C5)2 and Cs+(18C6)2 complexes, the mean Cs-
0 distance in the 18C6 sandwich is longer than that in the 15C5

43

Table 2. Characteristic 133Cs NMR Chemical shifts of the
' complexed cation.”-

 

Camelot Chenrtu~.ttl_shift_crznmlal
Cs+(18C6)2Na' ~ 6 1
Cs+(18C6)(15C5)Na‘ ~8
Cs+(15C5)2Na' +24
Cs+(18C6)2K’ ~ 5 8
Cs+(15C5)2K' +24
Cs+(18C6)2Rb- -57
Cs+(15C5)2Rb' +29
Cs+(18C6)2C8' ~ 4 1
Cs+(18C6)(15C5)C8‘ ~ 1 1
Cs+(18C6)(15C5)e‘ ~6 to +31

 

21Chemical shifts are referenced to M+(aq) at infinite dilution.

44
complex.40v73»34 This reduction in bond length could result in a

greater overlap of the cesium cation with the lone pair electrons on
oxygen. If one were to predict that the chemical shift of the mixed
crown ether system were to fall half way between the chemical
shifts of the pure complexes, one would obtain a chemical shift value
of approximately ~16 ppm.

The experimental value of ~11 ppm is quite close to the
predicted value. This chemical shift was reproducible on several
different samples. The sodide chemical shift in the mixed crown
ether complex occurs at ~61 ppm, unshifted from the value
calculated for gaseous Na'.89 There is no Na‘l' NMR peak observed,
an indication of the formation of a true cesium sodide compound.
Upon decomposition of the sodide, two cesium cation peaks occur at
21 and 13 ppm, paramagnetically shifted from the 1”CS peak in the
live sample. The change in chemical shift could be due to the
shortening of the Cs-O bonds since the decomposed sample is a very
viscous liquid and the Cs—O distances may change once the ordered
crystalline solid is no longer present.

In order to confirm the presence of a single phase, the DSC
pattern of Cs+(18C6)(15C5)Na' was measured and the decomposition

temperatures of this compound were compared to those obtained at
the same ramping rates for Cs+(l8C6)2Na' and Cs+(15C5)2Na'. For

example, at a ramping rate of 2 Klmin, Cs+(18C6)2Na' decomposes at
367 K with a AH value of ~208 kJ/mole, while that of Cs+(15C5)2Na‘
decomposes at 340 K with a AH value of ~141 kJ/mole. The mixed
sodide at the same ramping rate decomposes at 343 K with a AH

value of ~215 kI/mole and an irreversible endothermic transition at

45
260 K with a AH value of 7.2 kJ/mole. Preceding the decomposition

exotherm, there is a small peak visible at slower ramping rates that
may be due to a small portion of the sample on the outer edges or in
another area in the sample pan so that the decomposition does not
take place homogeneously.

Originally, one of the reasons for pursuing the mixed
complexant systems was that, in molecular models, the “fit” of the
cesium cation was better for a Cs+(18C6)(15C5) system than for
either of the pure sandwich compounds. The cavity for Cs‘l’ in the
18C6 sandwich complex appeared to be slightly too large while that
of the 15C5 sandwich complex was slightly too small. Pedersen and
Izatt have shown that the cation radius and cavity size of the crown
are the most important factors in the determination of the cation-
crown complex stability.24v90t91 Although no study has been done
on the Stability of the complex, the DSC data on the sodide indicate
that there has been no significant increase in the stability of the
alkalides towards thermal decomposition through the mixing of two
different crown ethers in the formation of the sodide.

The other mixed crown ether alkalide that was prepared was
the compound Cs+(18C6)(15C5)Cs-. A single 133Cs MAS NMR peak
was observed at ~8 ppm. The slight paramagnetic shift of the 133Cs
NMR peak for this mixed crown ether ceside as compared to that of
the sodide may be due to the slight shortening of the Cs—O bond
distances to cause a greater overlap between the cesium cation and

the unpaired electrons on the oxygens. For example, the chemical
shift of the cesium cation in the analogous compound Cs+(18C6)2CS'

is more paramagnetic than that of Cs+(l8C6)2 in the sodide, rubidide

46

 

Figure 12. The 133C: MAS NMR spectrum of the mixed crown ether
complex, Cs+(18C6)(15C5)Cs', obtained at 245 K and a
frequency of 23.6 MHz.

47

and potasside.76 Figure 12 shows the 133C8 MAS NMR spectrum of
Cs+(18C6)(15C5)CS'.

' The peak for the cesium anion, however, was not observed. In
the compound Cs"'(C222)CS‘, two Cs‘l’ peaks in the 133Cs NMR can be
observed for the CS+(C222), presumably due to the inclusive and
exclusive complexes, respectively; however, none of the samples
showed the NMR signal of CS".76 Similarly samples known to contain
Rb“ have also shown no 87Rb Signal.58 The anion peaks in these
compounds may be broadened by strong interactions between
adjacent alkali metal anions to such an extent that the signal is no
longer detectable. This may be the case for the compound
Cs+(18C6)(15C5)C8', however, the CS' to CS' distances are not known
since no crystal structure for this compound has been obtained.

The magnetic susceptibility of this compound was obtained
and, as for the sodide with a dopant level of approximately 1
percent, displays a slightly paramagnetic behavior with an electron
dopant level of approximately 3 percent. .The 133Cs NMR of these
alkalides demonstrates the potential of mixed crown ether complexes

to allow one to fine tune the characteristics of a compound.

Several preparations of the mixed electride, Cs+(l8C6)(15C5)e',
have been synthesized and studied. The first synthesis yielded a
compound that, when studied by magnetic susceptibility, contained
only approximately 3% unpaired electrons and had two 133Cs NMR
peaks at 220 and 132 ppm respectively. Synthesis and handling

48

techniques for these compounds were improved and a set of more
reproducible results was obtained.

. 133Cs MAS NMR spectra were obtained from the mixed crown
ether electride with some interesting results. In one sample, a single
peak at 36 ppm was observed, and this peak did not appear to shift
with temperature. The chemical shift of the decomposed sample
appears at 15 ppm. However, for the most recent sample of the
compound, 133Cs MAS NMR data were obtained in the 180 MHz
Brukcr instrument. The sample has both a temperature and time
dependent chemical shift. The chemical shift of the 133Cs peak
became more paramagnetic with increasing temperature from
approximately 6 ppm at 180 K to 31 ppm at 300 K. Upon recooling
the sample to 180 K, the chemical shift was no longer at 6 ppm, but
9.5 ppm. When the sample was placed in the instrument at a set
temperature and data were collected as a function of time, the 133C 8
chemical shift became more paramagnetic.

The 133CS chemical shift is very sensitive to the environment
of the cesium cation, so that only small changes need to occur within
the compound to cause these changes in the chemical shift. The
increase in the paramagnetic chemical shift may be due to a
shortening of the Cs-O bond distances as one increases the
temperature. The data, however, will have to be reproduced on
another synthesis of the sample.

The magnetic susceptibility of the compound, as shown in
Figure 13, follows the Curie-Weiss law. The Weiss constants for two

different data sets are rather large and negative, 6 = ~45 and ~72, an

indication of the presence of antiferromagnetic interactions and

22o:
223% so. a Sense «23338320 82.6%

o... .8 A. 32? E082 he 3:— 31—3:33?» oceans: .2 2:3..—

05 gear—an...

 

49

 

 

8N own an 8 2 o

P — A A b L A I? P A a
\i «.0
3 N6

9Q
see. --..
a
on 66 .t +6
0 + 4

o o m o t a...

a n + i .
D D + 0 no
0 D + O .I 6.6

0
o 0 + + «v i a...
80 m0 0 w
+ t a...
+ a 03 nmd 3.3.3.... A.

++ + Ox A Eat. .33 _ A

.2. co Au o: n 8.38:. t 3
oo Q Can 33.3.2.

I N..—
. n4

 

I r' i- l .‘
-’|ll" J3.
. .' a’b‘

(epuuenoqm)
“111111480011. mom/r

50

synthesis-dependent properties. The data were fit at temperatures
of 40 K and above, Since the Curie-Weiss equation is a high
temperature approximation and deviations are expected to occur at
low temperatures. Annealing the sample did not appear to have an
effect on the susceptibility, nor did there appear to be any significant
field dependence. The smaller values of l/xm at every temperature
for the 0.25 kG data are probably due to residual field effects, which
are more important at lower fields. The average percent of unpaired
Spins was 60 1 1 percent for the two different sample data sets when
only temperatures greater than or equal to 40 K were used to fit the
data. The deviation of the percentage in unpaired spins from 100%
may be due to partial decomposition of the compound when the
sample was loaded into the instrument or contamination of the
elecu'ide by ceside, a diamagnetic compound.

The DSC trace of Cs+(18C6)(15C5)e' showed that the enthalpy

of decomposition is intermediate to those of the two “parent”
electrides. At a ramping rate of 2 Klmin, CS+(18C6)2e' decomposes

at 315 K with 8 AH value of ~71 kJ/mole and Cs+(15C5)2e'

decomposes at 299 K with a AH value of ~103 kJ/mole while the
compound Cs+(l8C6)(15C5)e' falls intermediate between the two
with a decomposition temperature of 308 K and a AH value of ~98
kJ/mole. As with Cs+(15C5)2e', the mixed crown ether electride
undergoes a transition at 284 K with a AH value of 7 kJ/mole before
decomposition. The transition, however, is not reversible. A DSC
trace of the mixed crown ether electride is shown in Figure 14.
While this and other mixed crown ether alkalides and

electrides offer great potential in the study of how structure affects

HEAT FLOW (WI g)

51

 

 

 

 

0.4 '-
02 -.

 

 

JV

 

 

4'2 I I I I j I

Figure 14. DSC. trace of the mixed crown ether electride,
Cs+(l8C6)(15C5)e', at a ramping rate of 5 Klmin.

 

52

the properties of these compounds, the main difficulty in these
systems is the synthesis of pure bulk materials for characterization.
This-is especially true in the case of electrides where a small
contamination level may cause large changes in the bulk properties.
The synthesis techniques for these compounds need to be improved
and the work presented on the electride reproduced in other
syntheses. Synthesis techniques should be developed to improve the
methods for the transfer of the crown ethers so that complexant is
not lost during the interval between weighing the complexant and
the addition of solvent to the synthesis apparatus. The X-ray
structure Should be obtained for these compounds; that of the

electride would be of particular interest.

III. AZA COMPOUNDS
W

Alkalides and electrides are two unique classes of compounds
with unusual properties. Unfortunately, due to their extreme
sensitivity, these compounds must be handled under inert
atmosphere at cryogenic temperatures. Since the compounds
undergo irreversible decomposition du to their thermal instability,
the utility of alkalides and electrides, such as in reduction
processes92t93 has been limited.

To date, most of the alkalides and electrides synthesized have
utilized two classes of complexing agents, crown ethers or cryptands.
In the pursuit of more robust alkalides and electrides, the use of the
fully methylated aza analogs of crown ethers was pursued.72 Pez
and co~workers began to work with solutions of the complexant
hexamethylhexacyclen (HMHCY), the fully methylated aza analog of
18C6, in the search for more chemically resistant cation-complexing
agents in organic syntheses.92 While benzene solutions of the
cryptand C222 in the presence of NalK alloy rapidly degrade at room
temperature, HMHCY develops a deep blue color which persists for
approximately 30 minutes at room temperature. When used in the
selective hydrogen reduction of benzene to cyclohexane, these
reactions were always accompanied by extensive degradation of the
HMHCY complexant.

53

54

Based on the potential for more thermally stable alkalides and
electrides, HMHCY was used in the synthesis of several sodides.
When monitored by DSC, the thermal properties indicate that the
compounds CS+(HMHCY)Na' and K+(HMHCY)Na' melt at 281 K and
315 K respectively, but without thermal decomposition. However, an
overall increase in stability of the alkalides is not achieved since
decomplexation is observed at 310 K and 347 K respectively.

Further heating yields no sign of decomposition of the complexant at
temperatures up to 393 K.

The use of these more robust aza complexants has potential for
an increase in the thermal stability of alkalides and electrides, but
the substitution of nitrogens for ether oxygens in a ring reduces the
affinity of the ligand for the alkali metal ion.94 This factor may be
part of the reason that Cs+(HMHCY)Na' undergoes decomplexation.
Attempts to synthesize Cs+(HMHCY)e' were unsuccessful. In the
pursuit of 133Cs NMR data for the cesium cation in the compound
Cs+(HMHCY)Na', an investigation of electron doping in this compound
was undertaken to determine how several properties such as
magnetic susceptibility and NMR spectra, are affected by the dopant
level.

Further studies have been done on the use of other aza-crown
ethers as complexants in the synthesis of alkalides and electrides.
Preliminary work has been done on compounds synthesized with the
fully methylated aza analog of 15C5, pentamethylpentacyclen
(PMPCY). For the first time in the use of aza complexants an
electride was formed, Li+(PMPCY)e', although the crystal Structure of

this compound is not known.74 Unlike the compound

5 5
CS+(HMHCY)Na', this electride does not decomplex before it

undergoes irreversible decomposition at 353 K when monitored by
DSC. at a ramping rate of 5 Klmin. The magnetic susceptibility
measurements indicate that antiferromagnetic interactions are
present in Li+(PMPCY)e', making the properties of this compound
and the crystal structure of particular interest.

The trimethyltricyclen (TMTCY) complex with lithium has been
studied, but the resulting compound had not been not fully
characterized.74 A mixture of the complexants trimethyltricyclen
and dimethyltricyclen (DMTCY), however, produced a compound that
contained a mixed unit of Li+(TMTCY) and Li+(DMTCY')-CH3NH2 with

a sodide anion to yield the overall stoichiometry of
Li+(TMTCY)[Li+(DMTCY')-CH3NH2]Na'.95 The lithium sodide
synthesized with the pure TMTCY complexant was characterized by
several methods and the crystal structure was obtained once
problems with complexant purity were overcome. Once again,
however a mixed system, Li+2(TMTCY)2Na'-CH3NH' was formed.
While fully methylated aza analogs of the crown ethers have
been previously studied, no work had been done with the analogs of
cryptands. Since three dimensional cryptands form considerably
more stable metal complexes than the corresponding two
dimensional crown ethers, alkalides and electrides synthesized from
the fully methylated aza analogs of cryptands might be more stable
to decomplexation than the crown ether analogs.96»97t98 These
complexants offer the additional advantages of the ability to
selectively encapsulate small ions, as well as pre-orientation of the

ligand to yield relatively low entropies of formation for the

56

complexes.99 The use of 5,12,17-trimethyl—1,5,9,l2,17~penta~
azabicyclo[7.5.5]nonadecane (TMPAND) shown in Figure 15

 

Figure 15. 5,12,17-trimethyl-l ,5,9,l2,17~penta~aza
bicyclo[7.5.5]nonadecane (TMPAND).

in the synthesis of a lithium sodide and electride has shown

promising results.

W

1D "51'

In the search for more thermally stable alkalides and
electrides, the ability to more quantitatively determine the relative
thermal stability of a compound was needed. Until now only
qualitative determinations of the thermal stability of a compound
were made, either through Differential Scanning Calorimetry (DSC)
data collected at a single ramp rate or visible observations.74 The

first quantitative determination of the activation energy of the

5 7

decomposition of the electrides, K+(C222)e' and Li+(PMPCY)e', was
undertaken with DSC methods.100

. It has been found that the peak temperature of a DSC trace
increases with an increase in the rate of heating or ramp rate.6 8
This observation can be used to determine decomposition parameters
with the methods derived by Kissinger7o and Ozawa.69r70 For the
Kissinger method, it has been shown by Murray and White that first

order chemical reactions follow the equation:101

Aexpnmrm- E 31- (3.1)

2
RTm dt

where R is the gas constant, T the absolute temperature, dT/dt the
heating rate, Tm the peak temperature, E the activation energy for

decomposition and A the pre-exponential factor. Equation (3.1) can

be written as:
4) RA E
m _ . 1n — - — .
(131) (E) mm (32)

where III = dT/dt and all other terms are as previously defined. In
plots of ln(¢/Tm2) versus ll’I‘m for a number of materials, Kissinger
obtained Straight lines from which activation energies were
calculated.70 In a subsequent paper, it was shown that this equation

applies to nth as well as first order reactions.102

58

Analysis shows that for a generalized chemical reaction, lle

varies linearly with ln(¢) with the slope of the line equal to
(0.4567*l=.)/R in the Ozawa method.69 Activation energies
determined by each method are in good agreement“:100 If a
reaction monitored possesses an activation energy, the position of the
peak will vary with the ramp rate. Since all alkalides and electrides
studied to date possess an activation energy, decomposition
temperatures obtained at a Single ramp rate for a compound do not a

provide a quantitative basis for a comparison of relative stability.
Wars

An electric quadrupole moment can be described as a series of

mutipoles of the form:

where r is the distance to the charge ei from the electrical center of

gravity. A quadrupole does not interact with a homogeneous electric
field, however, there is an energy of interaction of a quadrupole with
an inhomogeneous electric field, an electric field gradient. Nuclei
with spin quantum numbers greater then 1/2 possess electric
quadrupole moments usually expressed in terms of eQ, where e is I

the charge on the proton and Q is the quantity referred to as the

59

electric quadrupole moment of the nucleus. Since there are two
types of distributions of charge, roughly cigar-shape (+) or disc-shape
(~), nuclear quadrupole moments also possess signs.

If a nucleus has a quadrupole moment or is magnetically
coupled to another nucleus with a quadrupole moment in a magnetic
field, the quadrupolar nucleus has two energy terms that need to be
considered, Zeeman and quadrupole terms where Q is the nuclear
quadrupole moment and q is the strength of the field-gradient. In
the high field case the Zeeman term predominates and the energy

levels are essentially those of the Zeeman term perturbed by
quadrupole effects. Therefore, the effect of J'IQ is usually treated by

perturbation theory and to first order each level Emo is shifted by an

1 .
amount Em .

l

En1

. .33. [360826 ~1][3nt12 - 1(1 + 1)] (3.4)

where 6 is the angle between the principal axis of the field gradient
tensor and the direction of the magnetic field Ho for the case in
which the asymmetry parameter is equal to zero and A is defined as
e2Qq/(41(21-1)). Since the shift depends on m with Am = .1. 1, the
original line is split into 21 lines corresponding to different values of

m, as shown in equation (3.5).

60
v° + 3A

”memst' m —h-(m+1/2)(3cosze-1) (3.5)

To first order, the central line (m = ~ll2 —> m = 1/2) is unshifted
for nuclei with half-integral spin. The second order shift of the
central line is given in equation (1.4).

For a polycrystalline sample, the effect of quadrupole coupling

is averaged over all orientations. Figure 16 illustrates the idealized
line-shapes expected for I=3/2 when the asymmetry parameter, nQ,

is zero. The distance between the two satellite peaks iS equal to vq.

V 392qQ' ( 6)
q (1121(21 -1)) 3'

 

2
v . eqo
q 2n

 

For 1-3/2 (3.7)

From equations 3.6 and 3.7, the quadrupole coupling constant for
I=3/2 can be calculated as 2vq = e2qQ/h = Quadrupole Coupling
Constant (QCC). The value of QCC reflects different values of the field
gradient for the same nucleus in different compounds. Examples of
the value of QCC for several ion pairs of lithium halides in the gas
phase are given in Table 3. Since the splitting by quadrupole effects

is symmetric about the unperturbed resonance line, the position of

61

Figure 16. The idealized line-shape for a polycrystalline sample
with quadrupolar coupling for I=3/2 when the asymmetry
parameter is zero.

62

the quadrupole satellites does not reveal the Sign of the coupling

0011818111.

Table 3. 7Li Quadrupole Coupling Constants (QCC) of lithium
halides for the gas phase ion-pairs.103

 

 

Species QCC
LiCl 0.192 MHz
LiBr 0.184 MHz
Lil 0.172 MHz
WW
I C l E .fi .

The fully methylated aza analogs of the crown ethers were
obtained from Dr. Farnum at Michigan State University. The
complexant HMHCY was placed over Na° and evacuated to 2 x 10‘5
Torr to remove water and volatile solvents. The complexant HMHCY
was then further purified by vacuum distillation at 398 K. The
purified HMHCY is a viscous clear liquid. The complexant TMTCY was
purified by vacuum distillation at 298 K to yield a clear liquid.
Contamination of this aza complexant with dimethyltricyclen
(DMTCY) resulted in the need for several distillation cycles of the
complexant over Li° to remove the impurity. '

The complexant TMPAND was obtained from Dr. M. Micheloni

at the University of Florence, Italy. The complexant was obtained as

63

the hydrochloride salt which was stirred in l N NaOH for 1 hour, then
extracted into an organic solvent and dried over MgSO4. Following
removal of the drying agent and evaporation of the solvent, the
freebase of the complexant was purified by vacuum sublimation at
423 K at a pressure of 2 x 10‘5 Torr or below. The free base was
obtained from the hydrochloride salt by Dr. Evelyn Jackson. A white
crystalline compound with a trace of yellow tinges was obtained
after several days of sublimation. Due to the extremely long times
needed for sublimation, the second time the complexant was purified
by simply placing the complexant under vacuum at 2 x 10"5 Torr for
a 48 hour period while being gently heated. Syntheses done with
complexant purified in this manner were not visibly different from
those done with complexant purified by vacuum sublimation, yet the

loss of complexant due to the purification process was minimized.
Am
a. Cs+(HMI-ICY)Na'

The general synthesis of alkalides and electrides has been
described in detail elsewhere.29t34 The synthesis of Cs+(HMHCY)Na'
was performed under two different set of conditions, (P) and (E). For
the first synthesis (P), a 60 percent excess of sodium metal was
added to the complexant HMHCY and cesium metal. Extra care was
taken to maximize the amount of sodium dissolved through the use
of relatively warm synthesis temperatures and longer times allowed

to dissolve the metal. For the second synthesis (E), a 2:1 ratio of the

64

alkali metals cesium:sodium was used and no extra care was taken to
allow all of the sodium metal to dissolve.

. When the synthesis of Cs+(HMHCY)e' was attempted, light blue
solutions were visible in dimethylether. No deep blue solutions
formed. Attempts to dissolve additional metal by distillation of the
solvent onto the metal mirror lead to decomplexation, since cesium
metal could be observed on both sides of the frit in the synthesis

apparatus. Removal of the solvent led to the Starting materials.

b. Li+2(TMTCY)2Na'-CH3NH' and
Li+2(TMTCY)2e'-CH3NH'

Since lithium metal is relatively insoluble in dimethylether,
methylamine was added in order to dissolve the alkali metal(s) in
the synthesis of these compounds. Once the metal(s) had dissolved,
all of the solvent was removed and the cell was pumped on the line
overnight to remove any residual solvent. Dimethylether was then
added to wash the compound through a frit, away from any excess
lithium metal. Crystals formed after slow evaporation of the solvent
and were then washed with n~pentane. During the attempts to
synthesize the electride, no solid was observed until n~pentane was
added. If all of the dimethylether was removed before the n-
pentane was added, then a viscous deep blue solution formed and
lithium metal began to collect on the bottom of the synthesis cell as
decomplexation occurred. Once the crystals were washed, solvent
was removed and the cell was evacuated to 2 x 10‘5 Torr before the

crystals were harvested.

6 5
c. Li+(TMPAND)Na' and I.i"'(TMPAND)e'

‘ The attempted synthesis of the compound Na+(rMFAND)Na-
failed. A blue solution, an indication of the presence of 6' and/or Na‘
in solution, was not visible even after 16 hours with the sodium
metal and complexant in a solution of dimethylether. Once all of the
solvent had been removed from the failed synthesis, the cell was
pumped out and placed in the dry box to introduce another metal.
The complexant remained a pale yellow color throughout this
process.

Lithium metal was placed in the synthesis apparatus and
methylamine was added to dissolve the sodium and lithium alkali
metals. The solution bubbled as the metals dissolved and a deep
blue solution resulted. The solution was packed in dry ice overnight
to allow the lithium time to complex. After approximately 12 hours
the methylamine solution was deep purple in color with bronze films
that formed if the solution was swirled on the sides of the glass
synthesis apparatus. All of the methylamine was removed, since
methylamine will complex alkali metals by itself, and the excess
metal present in the cell became apparent as the cell took on an
overall grey color. There were copper colored crystals present in the
bulk metal.

Dimethylether was added to the synthesis bulb and a deep blue
solution was observed. Since dimethylether will not dissolve sodium
or lithium metal, this solution was poured through a frit to another
' glass bulb on the synthesis apparatus to separate the complexed

cation from any excess metal present. Deep bronze films were

66

observed on the sides of the bulb. Trimethylamine was then added
to the solution as a cosolvent to induce crystallization. Bronze
colored crystals could be observed on the bottom of the bulb and no
metal was apparent. Once the dimethylether/trimethylamine
solvent mixture was removed, pure trimethylamine was added to
wash the compound and remove any excess complexant. The yield of
the compound was relatively large, although the need to keep the
sample cold did not permit an actual weight to be obtained.
Throughout the entire synthesis the solution was kept below 248 K
in dry ice/isopropanol baths or dry ice alone.

The electride Li+(TMPAND)e' was prepared in the same
manner in a quartz synthesis apparatus. A solution of lithium metal,
complexant, and methylamine was made to dissolve the metal and
aid complexation. Methylamine was then removed and
dimethylether added to wash the complex from the excess metal.
Trimethylamine was added to the dimethylether to crystallize the
electride. Finally pure trimethylamine was added to solvent-free
crystals to wash the compound. The synthesis was done at a

temperature of 228 K or below with good yield.

Mus—Ital

When the crystal structure of the compound
Li+(TMTCY)[Li+(DMTCY')-CH3NH2]Na' was solved, methylamine was
found to be present in the structure.95 This was the first time that
solvent had been observed in the crystal structure of an alkalide or

electride. All other analyses done on this compound and that of

6 7
Li+2(TMTCY)2Na'-CH3NH' gave no indication that methylamine was
present, therefore a simple method was designed to test for the
solvent.

To test for methylamine, harvested crystals of the compound to
be tested were cooled to liquid nitrogen temperatures in a glove bag.
Approximately 10 to 25 mg of sample was then placed in a beaker
that had been cooled to the same temperature. The beaker was
removed from the cooling source, then placed on the bottom of the

glove bag. While still cold, approximately 1 ml of D20 was pipetted

over the sample.
The D20 immediately froze and then was allowed to slowly

warm to room temperature. Since the volatile methylamine is
soluble in D20, upon decomposition of the sample the solvent was

dissolved in D20 rather than escaping into the glove bag. Proton

NMR spectra were then obtained for the decomposed sample as well
as the same decomposed sample with methylamine added. If one of
the peaks present in the decomposed sample grows upon addition of
methylamine, then solvent is present in the sample. Integration of
the peak heights yields the relative ratios of complexant to
methylamine. This test gives no indication of whether the
methylamine is present as CH3NH2 or CH3NH' in the crystals.
Attempts were made to decompose the sample directly in the
synthesis apparatus and use this solution for the test in order to

avoid decomposition of additional sample. This Should be avoided
since residual CH3NH2 present in the synthesis apparatus will give a

CH3NH2 peak in the proton NMR spectrum, although the solvent may

not be a integral part of the crystal structure.

68
LNMR

a. Cs+(HMI-ICY)Na'

Variable temperature 133CS and 23Na MAS NMR spectra were
obtained at 23.6 and 47.6 MHz respectively on a Bruker WH 180 MHz
(proton frequency) NMR spectrometer operating at a field strength of
4.2277 T, with 4 usec pulse lengths and 0.5 to 1 second delay times.
The temperature was controlled by the flow of nitrogen spinning gas
through a liquid nitrogen heat exchanger followed by increasing the
gas temperature with an in-line heater. After the samples had been
loaded into the rotor in a nitrogen filled glove bag at liquid nitrogen
temperatures, the sapphire rotor that contained the sample was
loaded into a precooled NMR probe. The sample temperature was
monitored with a thermocouple placed in the gas Stream at a point

just ahead of where the gas entered the spinner.

b. Li+2(erCY)2Na--CH3NH-
and Li+2(rMTCY)2e--CH3NH-

The variable temperature Static and MAS 7Li and 23Na NMR
spectra were obtained at 69.95 and 47.6 MHz respectively on a
Bruker WH 180 MHz (proton frequency) NMR spectrometer operating
at a field Strength of 4.2277 T, with 4 usec pulse lengths and 2
second delay times. Sample preparation and temperature control of

the Brukcr instrument were as described above.

69

Variable temperature solution 7Li and 23Na NMR spectra were
obtained at 69.95 and 47.6 MHz respectively on a Bruker WH 180
MHz. (proton frequency) NMR spectrometer, as the solid samples
were. Samples were prepared in a nitrogen filled glove bag.
Crystalline samples of Li+2(TMTCY)2Na'-CH3NH’ were loaded into a
cooled quartz NMR tube. The tube was then temporarily sealed with
a stop-cock and placed on a vacuum line and evacuated until a
pressure of 2 x 10‘5 Torr was obtained. To dissolve the sample,
dimethylether was distilled into the quartz NMR tube until a level of
2 to 4 cm was reached. The solution was then frozen and a quartz
seal-off done to permanently seal the tube. Samples were kept at
temperatures below 193 K until placed in the NMR instrument.

c. 1.1+(TMPAND)Na' and I.i+(TMPAND)e'

Variable temperature MAS 7L1 NMR spectra for
Li+(TMPAND)e' were obtained at 69.95 MHz on a Bruker WH 180
MHz (proton frequency) NMR spectrometer operating at a field
strength of 4.2277 T, with 4 usec pulse lengths and 1 second delay
times. Sample preparation and temperature control of the Brukcr
instrument were as described above.

7L1 and 23 Na static NMR measurements were carried out for
Li+(TMPAND)Na' at 155.454 and 105.8 MHz respectively on a Varian
VXR~400 spectrometer equipped with a 45 to 165 MHz broadband
probe by using single pulses of width 2 p.sec and a delay of 1 second.
Cold nitrogen gas was used to control the temperature to within 0.3 K

after 3 to 10 minutes equilibration time.

7O

5 II . S 'l'l'

Magnetic susceptibilities were measured with an S.H.E. 800
Series SQUID magnetometer. Samples were loaded into a Kel-F SQUID
bucket in a nitrogen filled glove bag at liquid nitrogen temperature,
then transported and loaded into the susceptometer at cryogenic
temperatures while under inert atmosphere. The samples were
loaded into the instrument by either zero field cooling the sample or
cooling the sample to 5 K in an applied field. Table 4 contains the
Specific conditions for each set of data.

Once the susceptibility of the “live” sample had been measured,
the background susceptibility was determined and subtracted from
that of the live sample. The background susceptibility was obtained
by allowing the live sample to decompose to a diamagnetic product
and rerunning the sample. This procedure allows one to obtain the
electronic contribution to the susceptibility. Since the decomposed
samples are diamagnetic and provide a direct measure of the atomic

susceptibility, the electronic contribution results only from M' or e’.

71

Table 4. Magnetic Susceptibility conditions for the data sets of
compounds that were made with the complexants HMHCY,

 

TMTCY, and TMPAND.
Species Imus-trains Eisld
Cs+(HMHCY)Na-(F) 1.8 to 180 K 1 and 5 k6
Zero field cooled
Cs+(1mnCY)Na-(I>) 1.8 to 204 K 5 k6
Quenched
Cs+(HMHCY)Na-(E) 1.8 to 223 K 5 k6
Quenched
Li+2(TMTCY)2e'-CH3NH' 1.8 to 193 K 1 and 5 RG
Quenched
Li+(rMPAND)e- 1.8 to 220 K 3, 5 and 7 kG

Zero field cooled

Li+(TMPAND)e' 2.5 to 200 K 0.5, 5 and 7 k6
Zero field cooled

 

ED'EE 'lS . Cl'

Differential scanning calorimetry (DSC) was performed with a
DuPont 910 Differential Scanning Calorimeter. Samples were
hermetically sealed in aluminum pans under nitrogen atmosphere at
cryogenic temperatures. The samples were transported and loaded
into the DSC instrument under inert atmosphere at cryogenic

temperatures.

72
ZQ|°IEI .

1 Optical absorption spectra of thin solvent-free films of the
sodide were measured from 300 to 1500 nm on a Model 260 Guided
Wave Spectrophotometer at temperatures from 262 to 283 K as
recorded with a copper-constantan thermocouple. Thin films were
made by first preparing a solution of the compound in
dimethylether, then pouring the solution over the quartz cell window

and then rapidly evaporating the solvent.
8 C l . . 11 l l

A 2~probe conductivity cell for d.c. measurements was
designed by M. R. Yemen and J. L. Dye.104 The 5 to 15 mg sample to
be studied was placed in a 4 mm glass tube which fit tightly over the
bottom stainless steel electrode. The top electrode was spring loaded
to ensure proper contact against the bottom electrode. To shield the
cell, the bottom electrode was electrically insulated from the body of
the cell by a teflon spacer and the cell body was connected to the
shield of the coaxial leads used in conductivity measurement.loo

All d.c. conductivity experiments were performed with a
Keithley 617 Programmable Electrometer used as both a voltage
source and an ammeter. Variable temperature data were collected
with a computer controlled Keithley 617 electrometer to record
sample measurements and a Keithley 580 micro-ohmmeter to record
resistance data from a four-probe carbon glass thermometer.

Rectangular voltage pulses 10 seconds in length and 100 mV in

73

amplitude of alternate polarity were applied to the sample to
eliminate polarization effects. Temperature data were collected

immediately following each elecuical measurement.

Wars

Samples for electron paramagnetic resonance (EPR)
measurements were prepared in a nitrogen filled glove bag. A
quartz EPR tube was filled to a length of approximately 1 cm with
sample while the tube and sample were kept at liquid nitrogen
temperatures. The stopcock attached to the EPR tube was closed and
then the EPR tube was removed from the glove bag and placed on a
vacuum line. When the EPR tube had been evacuated to 2 x 10-5
Torr, a quartz sealoff was done to permanently seal the sample tube.
Samples were kept at temperatures below 193 K until EPR
measurements were made.

EPR measurements were performed with an X-band ER200D
Series Brukcr Spectrometer by Dr. Michael Atarnian. The
spectrometer is controlled by a Nicolet 1180 data system and a slow

cryogenic system was used for low temperature work.

W911!

The single crystal X-ray diffraction study of this compound was

done in collaboration with Dr. Rui Huang. Single crystals of
Li+2(TMTCY)2Na‘-CH3NH' were obtained by slow solvent evaporation

from a solution of dimethylether at 193 K. Once grown, single

74

crystals of the compound were placed under a microscope on a
microscope slide at approximately 223 K under a layer of octane.
The Crystal selected for the study had the approximate dimensions
0.3 x 0.3 x 0.7 m3. The handling and mounting procedures have
been described in detail elsewhere.105 The diffractometer was a
Nicolet P3F that used graphite monochromatized Mo Ka radiation
with a locally modified Nicolet LT-l low-temperature system. Unit
cell parameters were determined by least squares from the setting
angles of 19 reflections in the range 15° < 29 < 20°. Intensity data
were collected with (t) scans at 2 deg/min to Sine/3. < 0.595 A4;
reflection indices: 0 1 h 1 10, 0 1 k 1 19, and 0 111 24. A linear
decay correction was based on the intensities of three monitored
reflections measured every 93 reflections. A numerical absorption
correction was obtained from the program DIFABS with a maximum
correction of 1.191; minimum of 0.258 and average of 0.818. Taking
a data cut-off of I > 36(1), there are 996 observed and 2114
unobserved data. The structure was solved by direct methods with
the SI-IELXS-86 program. Full least squares refinement was on F with
all atoms except protons refined anisotropically. H atoms were
constrained to ride on their bonded C-atoms with fixed isotropic
thermal parameters. There were 263 total parameters, W = 1.78,

and the function minimized was Zw(lFo| ~ chl)2. The maximum shift

error ratio (bid) was equal to 0.08, with R = 0.073 and wR = 0.067. R
is defined as Z IFOI ~ IFCIIZ IFOI. The final difference map peak

heights ranged from ~0.l7 to 0.19 e/A3. Scattering factors were
obtained from Cromer and Waber106 and f’ and f” from Cromer1 07

except that of the sodium anion which was obtained from Dr. David

7 5
A. Liberman.4o All computations were carried out on a VAXIIflSO

computer.
The cell parameters for the compound were: A = 8.896, B =
16.704 and C = 21.051 A with or = [5 =7 = 900. The cell volume is

equal to 3092.99 with 4 molecules per unit cell. The crystal belongs
to the Orthorhombic 222 space group P212121.

11112 R l ”1' .
W‘

Although the crystal structure of CS+(HMHCY)Na' is known and
the compound has been well characterized, previous attempts to
obtain 133CS NMR spectra on the compound have failed.72 The
separation of the cation-anion Van der Waals surfaces is between
013 and +0.07 A if the effective radius of the cesium cation is taken
to be 1.69 A and that of the sodium anion to be between 2.50 and
2.70 A. This is the first evidence for the presence of contact ion pairs
in alkalides. It was believed that the absence of the 133Cs NMR
Signal may have been due to a strong perturbation of the p electrons
in C81 as a result of their close proximity to Na'. There is evidence,
however, that the cesium cation signal also is sensitive to the level of
electron doping in the compound. The solutions from which crystals
of alkalides are grown contain solvated electrons as well as alkali
metal anions that cause defect electrons in alkalide salts.108 Work
was carried out in collaboration with Sylvie Doueff on samples that

contained two different electron dopant levels, those that were

76

relatively pure with approximately 2 percent or less electron doping
(P) and one that had a relatively large level of electron doping (E).

' In the Cs+(HMHCY)Na' samples with minimal electron doping, 3
133CS NMR Signal was observed at 249 1 15 ppm for two different
samples. Reasonable signal to noise ratios were obtained with
approximately 3000 scans. The electron dopant level measured by
magnetic susceptibility on one of these samples was approximately 2
percent. The NMR lines were relatively narrow, with a full width at
half height (AVIIZ) value of 722 Hz. When the temperature was
slowly raised to approximately 300 K, the 133CS Signal disappeared.
Since the compound decomplexes at 310 K, the loss of signal was
probably due to the decomplexation of CS'*'(HMHCY)Na'. No Signal
could be observed in the Bruker 180 MHz NMR instrument for Cs° or
the CsNa alloy. The sodide signal was observed before the variable
temperature 133CS data were collected. The 23Na peak with a
Av 1,2 value of 233 Hz, occurred at ~59 ppm a chemical shift
characteristic of all sodide salts.47.55

The EPR spectrum of CS+(HMHCY)Na' (P) indicates that there is
a clear hyperfine coupling to the cesium metal cation, I=7I2. While it
is clear that electrons are trapped in the crystal, the nature of the
trapping site is Still under investigation.109 When no attempt was
made to minimize the electron dopant level in the compound, the EPR
signal contains the same hyperfine interactions with the cesium
cation overlayed on a broad single featureless peak. The intensity of
this line was not as large as one would expect for an alkalide so

heavily doped with electrons. The loss in intensity may be due in

77

part to a poorly packed sample and/or partial decomposition or
decomplexation of the sample.

' While the sodide peak in the 23Na MAS NMR spectra could be
observed at ~61 ppm (E), the value of AV1/2 was relatively large,
2250 Hz. The line width in the heavily doped sample was
approximately 10 times broader than in the minimally doped
sample. This may be due in part to an increase in dipolar broadening
from the dipole interactions between Na’ and e'. The 133Cs MAS
NMR Spectra of the heavily doped sample changed as well. A very
poor signal with a signal to noise ratio of approximately 2 was
present after 17,000 scans at 186 ppm. No signal was observed if a
smaller number of scans was used. Therefore, this signal may have
been due to noise in the probe, but even if it is not, the need to
obtain 17,000 scans to observe the signal indicates that it was not
from a major fraction of the 133Cs nuclei in the sample.

The electron dopant level in this sample was determined to be
33 1 1 percent. The sample followed the Curie-Weiss law with a
Weiss constant of ~27 1 4 K, an indication of antiferromagnetic
interactions. The magnetic susceptibility data for the heavily doped
sample are shown in Figure 17. With 33 1 1 percent of the sodium
anions replaced by electrons, the dipolar and quadrupolar
broadening would be extremely large and could cause the loss of the
133Cs cation NMR signal.

Although a change in the NMR, EPR and magnetic susceptibility
data occurred when the electron dopant level was increased, the
transition temperature for melting and decomplexation in the DSC

trace of the compound did not change. The DSC trace for a lightly

78

 

l/xcm (emu/mole)
(Thousands)
8
D

 

TEMPERATURE (K)

Figure 17. Magnetic susceptibility plot of 11x em versus T for the

 

heavily doped sodide, Cs+(HMHCY)Na', quenched and run

in a field of 5 k6.

7 9
'doped and heavily doped compound are Shown in Figure 18. The

same transition temperatures for melting and decomplexation were
observed in both the samples (P) and (E), although the AH value of
the transitions had changed. The melting transition at 281 K for (P)
and (E) had AH values of 12.9 1 1.0 and 38 1 15 kJ/mole
respectively. One possible reason for the increase in the AH value of
the heavily doped sample in this transition is that the melting of the
sample may have initiated decomplexation. The electron may have
been able to recombine with the cesium cation more easily than the
sodium anion was able to. This would also explain the broad
featureless decomplexation peak in the heavily doped sample, since
partial decomplexation would already have occurred. This was an
indication that major Structural changes have not occurred, since one
would expect a change in the transition temperatures if radical
Structural changes had occurred.

The compound Cs+(HMHCY)Na' is the most heavily doped
alkalide synthesized to date. This is unusual in that the electride of
this compound could not be synthesized. Perhaps it is the inability of
the compound to undergo phase separation that allows such large
dopant levels to be obtained. It has been shown that it is possible to
obtain 133Cs NMR data for this compound in a minimally doped
sample. The large paramagnetic Shift of the cation was not
unexpected since the chemical shift of the cesium cation increases
with an increase in polarizability of the anion3'7 as well as with
overlap of the cesium cation with the unpaired electrons on the

nitrogen.

80

 

11
1.4 ‘I'

1.2 "

11
”-1

0.4 '1

HEAT FLOW (WI 3)
I;

 

 

 

 

MATURE (K)

Figure 18. The DSC traces of Cs+(HMHCY)Na' for the heavily daped

sample (bottom line) and the minimally doped sample
(top line ofl‘set by I W/g) obtained at a ramping rate of
5 Klmin.

 

81

A Study of the electron dopant levels in the intermediate range
of 5 to 15 percent would be of interest to Study the transition from
minimal dopant levels to maximum dopant levels. This Study has
also shown the need for care in the synthesis of alkalides and
electrides, since the dopant level of unpaired electrons can change

the properties of a compound under study.

Z—Lflzmzmlflfl'
M+21Im22£fl3ufl’

In the original attempts to synthesize Li+(TMTCY)2Na’, the

complexant was contaminated with DMTCY. The crystal structure
obtained with the mixture of complexants contained DMTCY', TMTCY
and CH3NH2 as integral parts of the crystal structure.95 In order to
overcome this problem, the complexant was purified extensively to
obtain pure TMTCY.

The molecular structure of the compound synthesized with
pure TMTCY is shown in Figure 19. A sandwich compound in which
the lithium cation is complexed between two molecules of the
complexant was expected, however, methylamine is also present in
the crystal structure as an anion. Both lithium cations have

coordination numbers of four. Each nitrogen is coordinated by the
three nitrogens on the TMTCY molecule and by the N' of the CH3NH'

anion. The N' of the CH3NH' forms a bridge between the two lithium

cations and the bond distances between N' and the two lithium
cations are shorter than the six Li+~N bonds. Table 5 lists selected

bond distances. The refined atomic coordinates and isotropic

82

 

 

Figure 19. A) The molecular Structure and numbering of atoms in
Li+2(TMTCY)2Na'-CH3NH‘; and B) Stereographic packing

diagram of Li+2(TMTCY)2Na‘-CH3NH'.

83

temperature factors as well as bond distances and bond angles are
presented in Appendix A.

. The N-Li+~N bond angles for nitrogens on the TMTCY molecule
range from 83.80 to 88.80. The lithium cation, while four coordinate,
is not in a tetrahedral coordination Site. The small bond angles
indicate that the lithium cation is well above the plane of the three
nitrogen atoms of the complexant. Although molecular models
indicate that Li+ would fit well between two TMTCY molecules, this
does not appear to be the case. The reason that a sandwiCh
compound does not form may be due to the lithium cation not fitting
well in the cavity of the complexant and the plane of the nitrogens or

the preference of lithium for a four coordinate environment.110

Table 5. Selected bond distances for the compound
Li+2(TMTCY)2Na‘~CH3NH'.

 

Bond LenstLiAI

Lil ~ N 2.135, 2.091, 2.074
Lil ~ N' 1.903

L12 ~ N 2.116, 2.088, 2.055
L12 ~ N' 1.928

C~N 1.444 to 1.521

C~C 1.369 to 1.457

 

84

This was only the second crystal structure where methylamine
is observed and incorporated as a stoichiometric part of the
su'ucture. In order to more easily determine if this solvent is
present in crystals without the determination of a crystal structure, a
test was developed to determine the presence of methylamine.

Figure 20 shows the proton NMR spectra from the test for this
solvent when performed on Li+2(TMTCY)2Na’-CH3NH' . The ratio of

methyl protons: CH3NH2/TMTCY is approximately 7:1, close to the

6:1 ratio expected for one methylamine molecule to two TMTCY
molecules, the ratio in the compound. 7Li and 23Na MAS and static
NMR spectra were performed on this compound as well as several
other analysis techniques, but no other methods except the X-Ray
diffraction data and the test specifically for CH3NH2 indicated the
presence of the solvent.

7Li Static NMR data were recorded and a Single rather broad
line was recorded at approximately 0.1 ppm with AV1/2 = 5036 Hz.
The chemical Shift value is reasonable for a lithium cation
coordinated by four nitrogen atoms in a sodide.65 The dipolar
broadening for the 7L1 cation was calculated with the Van Vleck
equation88 from the coordinates obtained from the crystal structure.
The calculated value of 5691 Hz is in good agreement with that of the
actual sample. Variable temperature 7Li MAS NMR spectra indicated
the linewidth narrowed so that AV ”2 = 3700 Hz at 220 K, however,
the chemical shift does not change. All attempts to perform 7Li and
23N8 solution NMR failed.

85

 

 

f T r v I v T I V V

 

 

V I V T v v T f *1 v

2.20 2.15 2.10 2.05 2.00 1.95
(919m) I

 

 

 

 

f—r f v f

2.3 2:2 I

f f

(99m)
Figure 20. Methylamine Test A) 1.112(TMTCY)2Na'°CH3NH' quenched
in D20 at cryogenic temperatures; and B)
Li+2(TMTCY)2Na‘-CH3NH' quenched in D20 at cryogenic
temperatures to which a solution of CH3NH2 in D20 has
been added.

2.1

86

The 23Na NMR peak for the solid compound occurred at
~59 ppm, the chemical shift characteristic of all sodide salts, even
though the crystals contained both Na‘ and CH3NH'. The optical
absorption spectrum with a single peak at 700 nm obtained by M.
Kuchenmeister for the sodide also gave no indication that the crystals
studied were not those of a pure sodide.74 Magnetic susceptibility
studies of this compound indicate a diamagnetic sample as expected,
with less than a 0.1 percent electron dopant level.

The DSC trace of the compound was originally done by M.
Kuchenmeister, however, these data were obtained before the
complexant was properly purified. A decomposition temperature of
353 K at a ramp rate of 5 Klmin was obtained by M.
Kuchenmeister.74 In contrast, a decomposition temperature of 330 K
with a AH value of ~41.9 kJ/mole was obtained when the
experiments were repeated at a ramp rate of 5.5 Klmin. Figure 21
shows the DSC trace of the more recently obtained data. A low
temperature transition was observed in some of the samples;
however, it was not reproducible in all of the samples. Not all of the
samples loaded in the DSC sample pans had reproducible thermal
behavior. The samples, which contain a lithium cation, may have
been extremely sensitive to storage in a liquid nitrogen filled dewar
or have reacted with the aluminum sample pans.100

Several attempts were made to synthesize an electride with the
TMTCY complexant. Although deep blue solutions form, severe
decomplexation problems occurred when solvent was removed.
Attempts to collect magnetic susceptibility data on the compound
failed. 7Li MAS NMR spectra of the compound, however, had a peak

87

 

3.5-:

I

254

 

 

 

 

 

 

Figure 21. DSC trace of the sodide, Li+2(TMTCY)2Na'-CH3NH‘,
obtained at a ramping rate of 5.5 Klmin.

 

88

at approximately 36 ppm at a temperature of 180 K when data were
collected on the Brukcr 180 MHz instrument. As the temperature of
the Sample was raised, no change occurred in the spectra until 220 K.
At this temperature the peak at 36 ppm disappeared and a peak
appeared at approximately 1 ppm, the region where one would
expect the chemical shift value of the decomposed sample to

occur.1 11 When the sample temperature was lowered to 180 K, the
peak at 36 ppm did not return.

If an electride forms with the TMTCY complexant it appears to
be very unstable. Good crystalline material of this compound has not
yet been obtained due to the severe decomplexation problems in the
synthesis of the compound. Problems with decomplexation may be
due to the volatile nature of the complexant. As one pumps off the
solvent, the complexant is removed as well which leads to the
formation of lithium metal particles.

Although the crystal structures of compounds formed with
TMTCY as a complexant are of interest, there are several drawbacks
to the use of TMTCY in the synthesis of alkalides and electrides; the
incorporation of solvent molecules in the crystal Structure and the
inability to synthesize a crystalline electride for characterization.
There is evidence that the pure TMTCY sodide may, in fact, have
another crystal structure with the same molecular Structure but
another packing arrangement.112 One possible solution to the
problems with the TMTCY complexant may be the alteration of the
complexant through the use of propyl instead of ethyl linkages
between the nitrogens to open up the complexant ring so that the

lithium cation may more easily sit in the plane of the nitrogen atoms.

3 I'+[IIIE!IIDJII , lI'flIlIEEIID] _

The first fully methylated aza analog of a cryptand was used in
the synthesis of a lithium sodide and electride. The crystal structure
of the TMPAND compounds has not been determined; however, the
methylamine test was performed on harvested crystals of the sodide.
No peak associated with methylamine was observed in the proton
NMR of this sample, a strong indication that this solvent is not an
integral part of the crystal structure as it is in the TMTCY compound.

Optical absorption spectra of this compound were obtained at
several temperatures. Figure 22 shows the optical absorption
spectra of Li+(TMPAND)Na' at two different temperatures, 183 and
262 K, with the maximum at 698 and 717 nm (14330 and 13950
cm‘l) respectively. The major peak at approximately 700 nm in the
spectra of these thin films produced by solvent evaporation is
attributed to the 382 to 3s3p transition of the anion, Na".31 The
transition was slightly temperature dependent with a red shift at
higher temperatures. No sharp break at any temperature occurred
but a gradual change with duldt of approximately ~5 cm'lK'l . The
peak also became broader at higher temperatures. Both of these
properties are observed in Na"’(C222)Na' films.42

Although no sharp break occurs in the optical absorption
spectra, a phase transition occurs at low temperatures in
Li+(TMPAND)Na‘. At an onset temperature of 200 1 2 K with a AH

90

 

0.9‘

0.8 -

0.1 ‘1

0.6-1

N"

RHATIVE ABSORBANCE

 

 

 

0.1 I

I l
400 600 800 1000

WAVELENGTH (11111)

Figure 22. Optical absorption spectra of Li+(TMPAND)Na‘ obtained
at 262 K (top line ofset by +0.2 on the Relative
Absorption scale) and 183 K.

91

value of 3.3 1 1 kJ/mole, the compound undergoes a solid to solid
transition. The phase transition could not be observed visibly Since
the Crystals neither change color not appear to soften. The kinetics
of decomposition for Li+(TMPAND)Na' were studied by DSC methods
with variable ramp rates. Figure 23 illustrates that the
decomposition temperature of the compound is dependent on the
ramp rate and shows the DSC trace of Li+(TMPAND)Na' at two
different ramp rates. Specific data for the decomposition
temperature at different ramp rates are given in Table 6. The AH
value for decomposition was determined to be ~52 1 17 kJ/mole.
With the Kissinger method70, the activation energy for the
decomposition of Li+(TMPAND)Na‘ was determined. The graph of
ln(¢/Tm2) versus ll’I‘m used to determine the activation energy is
shown in Figure 24. At 101 1 5 kJ/mole, the activation energy of the
sodide Li+(TMPAND)Na' is greater than the two electrides studied
with this method. Li+(PMPCY)e' and K+(C222)e' have activation
energies of 87 kJ/mole and 66 kJ/mole respectively.100 Although
there is not a large basis for comparison, the compound
Li+(TMPAND)Na' appears to be relatively stable to thermal
decomposition. Activation energies, however, have not been

determined for any other alkalides.

HEAT FLOW (w, 3)

92

 

 

 

 

 

 

 

0.5
IIIIIIIIIIIITIFIIII

I” 210 230 260 270 2” 310 M 30 370 3”
TEMPERATURE (K)

Figure 23. The DSC trace of Li+(TMPAND)Na' obtained at a ramp rate
of 12 Klmin (top line ofset by 2 Wig) and 7 Klmin.

93

 

ln(¢l'l‘m2)

 

 

 

Figure 24. Plot of ln(¢lT'm2) versus 171' for Li+(rMPANn)Na- used
to determine the activation energy for decomposition.

9 4
Table 6. Table of thermal decomposition temperatures for
Li+(TMPAND)Na’ and the ramp rate at which they were

 

obtained.
D . . I [K] B B [Kl . 1

349.5 1
357.8 2
369.7 7
371.5 10
376.1 12
378.3 15

The d.c. conductivity of the sodide was measured by K.
Moeggenborg.100 Figure 25 shows a plot of the the d.c. conductivity
results for this compound. At temperatures below the solid-solid
phase transition the sample resistance was beyond the measurement
range of the instrument. The sample behaves as a semiconductor
with a band gap of approximately 1.2 eV above a point with
corresponds to the temperature of the phase transition. The
possibility of ionic conductivity in Li+(TMPAND)Na' was tested
through the use of a sodium electrode. The Ohm’s law plot for the
cell was symmetrical and passed through the origin of the I~V plane.
This fact, along with the lack of electrochemical cell behavior,

suggests that ionic conductivity is not present. Therefore

95

 

 

 

 

 

 

l/I'EMPERATURE (100

Figure 25. The variable temperature d.c. conductivity data for
Li+(TMPAND)Na' obtained in a 2~probe conductivity
apparatus. 1 00

0.007

96

Li+(TMPAND)Na', an insulator below the phase transition, is a
semiconductor with a band gap of 1.2 eV above approximately 200 K.

i The high temperature static 23Na NMR peak of
Li+(TMPAND)Na' occurs at a chemical shift value of ~60.7 ppm with a
AV1/2 value of approximately 1500 Hz. The chemical Shift value,
very characteristic of sodides, indicates that the sodium anion is in a
highly symmetric environment with minimal overlap of its p orbitals
with the surroundings.31 When the temperature of the sample was
lowered below the temperature of the phase transition, a second
peak could be observed. Figure 26 shows the 23Na static NMR
spectra above and below the phase transition. The second rather
broad peak occurred at approximately ~10 ppm, a paramagnetic Shift
from the typical sodide peak at ~60.7 ppm. If the temperature was
raised above the phase transition, the second 23Na peak at ~10 ppm
disappeared. It appears that only a portion of the sodium in the
compound is involved in the process in the compound that causes the
~10 ppm peak to occur since the peak at ~60.7 ppm remains. To
date, no transition of this type has been observed in any other sodide
studied.

A possible explanation of the peak at ~10 ppm is that the p
orbitals in the sodium anion may interact with the lithium cation or
are deformed by the crystal packing below the phase transition.
Perhaps there are two different sodide sites in the lattice, the reason
only some of the sodide anions have a paramagnetic Shift.

The 7Li static NMR spectra of Li+('rMI>AND)Na- have a central
transition at approximately ~9 ppm. Unlike the sodide, the chemical

shift is not temperature dependent; however, the value of the full

97

 

 

 

 

 

A B
m MW
100 so 20 ~20 ~60 ~1oo one 20 0 ~20 ~60 ~loo po-

Figure 26. 231% static NMR spectra of u+(rMFAND)Na- obtained at
A) 189 K and a) 213 K.

9 8
width at half height (Av 112) does depend on temperature. Due to

the absence of nuclear motion in crystalline solids, quadrupolar
interactions, dipolar interactions and chemical shift anisotropy are
not averaged and cause line broadening. Therefore, it is not totally
unexpected that an increase in temperature would cause line
narrowing due to an increase in molecular motion. At the phase

transition, there was also a Sharp break in the plot of the value of
AV1/2 versus temperature as Shown in Figure 27. The Sharp change

in the value of Av 1,2 may be due in part to an large increase in

molecular motion above the transition.

Although the crystal structure of Li+(TMPAND)Na' is not
known, that of an analogous crystal Li+(TMPAND)BPh4‘ has been
determined.113 The dipolar interaction was calculated with the Van
Vleck equation with the lithium-nitrogen and lithium-proton
distances obtained from the crystal structure of the model salt.88 A
calculated value of 8200 Hz was obtained. This compares quite well
with the experimental value of 7700 Hz for Li+(TMPAND)Na'
obtained at 173 K where molecular motion is minimized in the
sodide.

Not only could the central transition of the 7Li static NMR
spectra be observed, but also the satellite transitions. When the
asymmetry parameter, nq, is assumed to be zero, the distance
between the satellite transitions can be used to calculate the
quadrupolar coupling constant for this compound. In Figure 28, the
central transition observed in 7L1 static NMR as well as the satellite
transitions are shown. The quadrupolar coupling constant (QCC)

simply reflects different values of the electric field-gradient for the

99

 

FULL WIDTH AT HALF HEIGHT (Hz)
(Thousands)
‘I

 

 

 

Figure 27. Plot of the full width at half height versus temperature
for the 7Li static NMR spectra obtained for
Li+(TMPAND)Na'. The solid-solid phase transition occurs
at the temperature of the observed discontinuity in the
curve.

100

 

 

 

1A-

300 100 -100 -300 ppm

Figure 28. 7L1 static NMR spectrum of Li+('rMFAND)Na- obtained at
203 K.

101

7Li nucleus in different compounds. The value for QCC of .188 MHz
at 173 K for Li+(TMPAND)Na' is quite close to the value obtained for
LiCl'of .192 MHz. This value of QCC for LiCI in the gas phase is for a
contact ion pair.103 This may also be the interaction that occurs
between the lithium cation and sodium anion in Li+(TMPAND)Na‘.
The value of QCC was temperature dependent and the plot of the
value of QCC versus temperatures is shown in Figure 29. The plot
has a distinct break at the phase transition, however, the calculated
contribution to the value of Av] ,2 due to quadrupolar interactions is
much less than 80 H2114

Through a combination of 7Li and 23Na static NMR spectral
data, a possible explanation of the observed behavior can be
proposed. At temperatures below the phase transition, the crystal
structure of the compound may be locked in a conformation that
increases the interaction between the lithium cation and sodium
anion, perhaps through the formation of a contact ion pair. This
would explain the large QCC value obtained from the 7Li NMR data as
well as the second paramagnetic chemical shift in the 23Na NMR at
low temperatures. According to this model, once the temperature of
the compound was increased above the temperature of the phase
transition, an increase in molecular motion or a conformational
change in the crystal would cause the interactions between the cation
and anion to decrease. This would be reflected in a sharp decrease in
the value of QCC and the disappearance of the second more
paramagnetic chemical shift in the 23Na static NMR spectra. Single
crystal X-ray data, especially if performed at temperatures both

102

 

 

 

 

 

 

i a 3T
0107
g tll~
1% sin 0
% tit-f
0 014-1
5
é uh °
at o
g tll~
5 1.1-
:: m~
a
“a
0.07 1 I I I I T
170 no 211 no
mmm

Figure 29. Plot of the Quadrupolar Coupling Constant (QCC)
calculated from 7L1 static N MR spectra of
Li*(TMPAND)Na' versus the temperature of the
sample. The solid-solid phase transition occurs
at the temperature of the observed discontinuity in the
curve.

103

above and below the phase transition, would be required to
substantiate this explanation.

The lithium electride was also synthesized with the complexant
TMPAND. When the sample was tested for methylamine, however,
solvent was detected in the crystals. It is not known whether the
methylamine is a regular part of the crystal lattice or simply due to
residual solvent. Since solvent was present in the crystals, some of
the general properties reported here, may or may not change when a
pure electride is synthesized and characterized.

Although DSC data were collected, no reproducible exothermic
peak could be observed, therefore the decomposition temperature is
not known. The magnetic susceptibility of the electride was recorded
for two different syntheses. The plots of l/xfim (emu/mole) versus
temperature are shown in Figure 30. The samples follow the Curie~
Weiss law. The more paramagnetic sample contained 29 1 1 percent
unpaired electrons with a Weiss constant of ~24 1 9 K for data above
40 K. The less paramagnetic sample shown by the plus symbols in
Figure 30, contained only 12 1 0.3 percent unpaired electrons with a
Weiss constant of ~18 1 5 K when determined for data obtained
above 40 K. The samples did not appear to be field dependent and
the Weiss constants indicate the presence of antiferromagnetic
interactions.

Since 100 percent unpaired electrons are expected for a pure
electride, the Li+(TMPAND)e' may contain CH3NH' as an anion as
with the lithium sodide synthesized with the TMTCY complexant.
Part of the difficulty in the use of TMPAND as a complexant is the

llxe'“ (emu/mole)

(Thousands)

104

 

 

 

6
+
5- + +
+
1
4‘ +
++
8" *
*1
+ + a
2“ 011% GB an
I: I ° 03
it a In
* * no DI
' 000
#0000000
0 i I I I I I I I I r I
0 40 U m 1m 2”

Figure 30. Magnetic susceptibility plot of ”er“ versus T for

 

Li+(TMPAND)e‘ obtained for two different syntheses.
The squares represent the sample with 29 percent
unpaired electrons run at 3, 5 and 7 k6. The plus

symbols represent the sample with 12 percent unpaired

electrons run at 5 and 7 k6.

 

105

extremely slow rate at which this complexant is able to complex the
lithium cation, even at room temperature.113

7Li MAS NMR data obtained on the more paramagnetic
electride sample contained two NMR peaks at 46 and 33 ppm with a
rather rolling baseline when originally placed in the Brukcr WH
instrument at 180 K. When the sample was warmed to 220 K, only
two peaks were observed, one at 36 ppm due to the compound
Li+(TMPAND)e' and one at 260 ppm the observed chemical shift of
lithium metal. The rather large paramagnetic Shift of the sample
indicates the presence of unpaired electrons. Other than the lithium
metal peak, only one peak was observed for the sample. Therefore,
any other lithium species present must have been broadened into
the baseline, Since DSC and magnetic susceptibility results indicated
that this was not a homogeneous sample.

Data collected for both compounds are promising, and the
Li+(TMPAND)Na' compound exhibits some interesting properties.
Since the value of QCC for 7Li has been determined, 6Li static NMR
Spectra could yield the ratios of the quadrupole moment for 7Li/6Li.
The determination of spin-lattice relaxation times for 7Li would
allow the correlation of relaxation time and the value of the QCC for
the compound Li+(TMPAND)Na'. In order to synthesize a pure
electride, the kinetics of the complexation of the lithium cation by
TMPAND at low temperatures in methylamine should be
investigated. Properties of other fully methylated aza analogs of
cryptand should be pursued, not only for their unique properties, but
also in the attempt to form more thermally stable alkalides and

electrides.

IV. ETHYLAMINE SOLUTIONS
~ ILAAInmdnetinn

It has been observed that addition of lithium metal to a
solution of starting materials in methylamine inhibited the
decomposition process in the synthesis of other alkalides and
electrides. These Li/methylamine solutions appeared to be more
stable than equivalent solutions that did not contain lithium
metal.86r115 M. Faber also observed that methylamine enhances the
solubility of lithium in certain solvents.19 For example, lithium is
insoluble in 2-aminopropane but a solution of Li+(CH3NH2)4 is
soluble in this solvent. Li/methylamine solutions also appeared to
increase the solubility of other alkali metals in solution.

In the absence of complexing agents, sodium metal is only
slightly soluble in methylamine.19 However, concentrated solutions
of sodium metal form in the presence of Li+(CH3NH2)4. Similar
results were found when sodium was replaced by rubidium. It was
presumed that the concentrated solutions had the stoichiometry
Li+(CH3NH2)4M', where M' is Na‘ or Rb’. Optical absorption spectra
of thin films of Li/CH3NH2/Na solutions contained a peak at 660 nm
as expected for a sodide. However, attempts by O. Fussa to isolate
the compound Li+(CH3NH2)4Na' were unsuccessful.116

L. E. H. McMills extended the methylamine assisted
solubilization of O. Fussa and M. Faber to other solvents.19 The
studies of the Li/methylamine systems were extended to include a

more extensive study of the magnetic properties of these systems,

106

107
Specifically 7Li and 23Na solution NMR spectra. The Li+(amine)xNa'

systems were Studied from x = 4 to 16 by J. Kim and L. E. H. McMills.
Table 7 contains the lithium chemical shifts for x = 4 at several
temperatures in methylamine and ethylamine.“ The value in
ethylamine is consistent with that expected for a diamagnetic
compound that contained lithium as a cation while the temperature-
dependent shift in. methylamine suggests the presence of some

paramagnetic species. Table 8 gives the lithium chemical Shifts at
different temperatures for the known compound Li(NH3)4.18.65

Table 7. 7L1 NMR chemical shift values obtained for
Li+(CH3CH2NH2)4Na' and Li+(CH3NH2)4Na'.65

 

gunman WWW

Li+(CH3CH2NH2)4Na' 253 2.1
2 3 3 2.1
21 3 2.0
2 3 3 7 .4
21 3 5 .6

 

The Li(NH3 )4 system is of particular interest since the liquid

phase of this compound is a nearly free electron metal and the solid
is a highly correlated metal below 80 K.18 When ammonia is
replaced by methylamine, the liquid is metallic but never reaches

the nearly free electron behavior found in concentrated

108
lithium/ammonia systems.18v ”7 The behavior of Li(CH3NH2)4

remains that of a highly correlated metal.

Table 8. 7L1 NMR chemical shift values obtained for Li(NH3)4.55

 

WK). ChemicaLsmmm All/2 mm
200 19.9 1.4
213 20.3 1.4
233 20.8 1.5
203 20.0 1.7

 

While the lithium sodide systems Studied with the solvent
ethylamine are diamagnetic compounds, the systems
Li+(CH3CH2NH2)4K' and Li+(CH3CH2NH2)4CS' studied in
collaboration with M. DeBacker display some unexpected properties.
These compounds may lie on the border of the metal to non-metal

transition.

W

In his view of the transition between metallic and insulating
states, Mott considered an array of hydrogen atoms.118 At high
densities, the electron gas screens the positive ions and the electrons
are itinerant. As the density decreases, the electrons are captured

by the ionic cores and the material becomes nonconducting. The

109

Mott criterion predicts whether a metallic or nonmetallic State exists

and is given in equation 4.1.

nCI/3a" .~.-. 0.26 (4.1)

where n is the critical carrier concentration and a“ is the value of

c
the effective Bohr radius.118 Classical percolation, in which small
metallic regions permeate an insulating medium and provide
conducting paths, as well as non-interacting scaling theory suggested
by Abrahams are other models proposed to explain the metal to
insulator transition.l 19,120

In the metallic state, itinerant electrons produce the electrical,
magnetic and optical properties associated with metals. The
transition from metallic to insulating behavior is accompanied by a
marked change in electrical conductivity. In a semiconductor or
insulator, the electrical conductivity increases with temperature
while it decreases with temperature in a normal metal.

The itinerant electrons in a metal also exhibit a reflectance
spectrum that has a Sharp cutoff at the plasma frequency.121 It is
this reflectance that is responsible for the metallic glitter of metals.
However, this Single observation cannot be used to unambiguously
judge metallic character.

The magnetic properties of localized unpaired electrons are also
significantly different from itinerant electrons in a metal. In contrast
to the Curie law behavior of localized, non-interacting electrons, the

Pauli susceptibility of itinerant electrons is essentially independent

110

of temperature. Itinerant electrons are not spatially localized, nor do
they respond independently to the magnetic field. The Pauli
susceptibility is considerably smaller than the corresponding Curie

value, even at room temperature.
Wm
LJamMthuaralinn

Samples of the compounds studied were prepared in the quartz
SQUID cell shown in Figure 31. Since the solutions are quite volatile,
a Kel-F squid bucket could not be used since loss of sample would
occur upon loading and during decomposition of the sample.
Therefore, a cell was designed that allowed the sample to be
manipulated on a vacuum line during preparation while permitting
one to permanently seal the sample for magnetic susceptibility
measurements. Before the specific amount of metal to be used was
calculated, the amount of solvent was determined. A small glass
solvent bottle or break seal ampule was weighed, then
approximately 0.2 ml of purified ethylamine was distilled into the
container. The difference in weight between the full and empty

container was used to determine the weight of ethylamine.

l 1 1
E2) Fischer Porter

 

 

 

A... ‘v‘ -.~.~ Seal ~01! point.

‘6

Figure 31. Schematic diagram of the quartz SQUID cell.

 

 

Once the amount of ethylamine had been determined, the
amount of alkali metal necessary to obtain the correct stoichiometry
in the solutions, Li:4 CH3CH2NH2:X where X = Na, K, and Cs, was
calculated. After the quartz SQUID cell was evacuated to a pressure
of 2 x 10-5 Torr, the cell was placed in a helium glove box and the
calculated amount of alkali metal was added to the cell. Due to the
“sticky” nature of alkali metals and the necessity to have exact
stoichiometric amounts, the metals were chilled for several minutes
before they were placed in the SQUID cell. Cooling the metals
seemed to keep them from sticking to the sides of the cell and the
loss of small amounts of metal was minimized. Once the metal had
been added, the SQUID cell was sealed with a stop-cock and removed
from the dry box. The metals were kept under helium atmosphere
until evacuated to a pressure of 2 x 10‘5 Torr. The metals were then

frozen at 77 K as the cooled ethylamine was slowly distilled over the

112

metals. Care must be taken in the distillation since these solutions
are extremely sensitive and tend to ”bump”. If the lithium metal
”bumped” above the line of the seal-off, the quartz SQUID cell
cracked when the lithium metal was heated during the seal-off. Once
all of the solvent had been distilled over the metal, then a quartz
seal-off was made approximately 2 cm from the bottom of the hook
on the SQUID cell. The sample was kept at cryogenic temperatures
until loaded in the NMR instrument or SQUID susceptometer.

2.18MB

The NMR spectra and the magnetic susceptibility data for each
compound were obtained on the same sample. In order to obtain
NMR Spectra on the sample contained in the SQUID cell, a 10 mm
NMR tube was used. Kimwipes were packed in the bottom of the
cooled NMR tube in order to place the sample on the same level as
the NMR coil in the Brukcr 180 MHz instrument. A small amount of
Kimwipe was also packed on top of the SQUID cell to prevent sample
movement in the cooled NMR tube that might cause breakage when
the sample was loaded in the instrument.

Variable temperature 7Li and 133Cs solution NMR spectra
were obtained at 69.95 and 23.6 MHz respectively on a Bruker WH
180 MHz (proton frequency) NMR spectrometer operating at a field
strength of 4.2277 T, with 4 11sec pulse lengths and 0.25 to 1 second
delay times. The temperature was controlled by the flow of nitrogen
- Spinning gas through a liquid nitrogen heat exchanger and then

warming the gas with an in-line heater. After the samples had been

1 13
loaded into the NMR tube as described above, the sample was loaded

into a precooled NMR probe. The sample temperature was monitored
with i a thermocouple placed in the gas Stream at a point just ahead of

where the gas entered the spinner.

3 II . S 'l'l'

Magnetic susceptibilities were measured with an S.H.E. 800
Series SQUID magnetometer. Samples were loaded into a quartz
SQUID cell as described in the sample preparation, then transported
and loaded at cryogenic temperatures into the susceptometer by a
string tied to the hook on the SQUID cell. The samples were loaded
into the instrument by zero field cooling the sample then applying a
field of 3 k6. A field Study was obtained at 3 and 7 k6 over the
temperature range 2 to 220 K. The sample was then placed in a
constant temperature bath at approximately 223 K for several days.
Data were then obtained at the same fields with the same loading
procedure to determine if “aging” of the sample changed magnetic
properties. The NMR data obtained by L. E. H. McMills for the
compound Li+(CH3CH2NH2)4Na' were dependent on the “age” of
some samples.65

Once the susceptibility of the “live” sample had been measured,
the background susceptibility was determined and subtracted from
that of the live sample. The background susceptibility was obtained
by allowing the live sample to decompose to a diamagnetic product
and rerunning the sample. Since the decomposed samples are

diamagnetic and provide a direct measure of the atomic

114

susceptibility, this procedure allows one to obtain the electronic
contribution to the susceptibility. Since volatile gases are generated
upon. decomposition and contained in the volume of the small sealed-
off SQUID cell, care must be taken in handling the decomposed
sample. The sample was decomposed and loaded into the SQUID
susceptometer behind a blast shield. The SQUID cell of the
decomposed sample was also destroyed behind a shield.

HIDE l 111' .

Although the NMR chemical shifts for the 7Li and 23Na nuclei
in Li+(CH3CH2NH2)4Na’ have been investigated extensively}55 the
magnetic susceptibility of this sodide was not studied. The 7Li NMR
chemical shift at 2 ppm indicates the lithium cation is well-shielded
by the ethylamine molecules.65 The chemical shift of the sodium
anion at -59 ppm is consistent with the value obtained for other
sodides.47t55

The magnetic susceptibility data for the compound
Li+(CH3CH2NH2)4Na' are shown in Figure 32. The diamagnetic
susceptibility of an atom is proportional to the ionic radius. The
halide anions were used to predict the value of the diamagnetic
susceptibility of the sodide anion. The known ratios of the ionic radii
squared and the corresponding diamagnetic susceptibilities of the
halide ions were compared. When this comparison is extended to the
sodide anion with a radius of 2.5 A,” a value of approximately -8 x
10'5 emu/mole is obtained for the diamagnetic susceptibility of Na'

based on the Languin formula.122 This value is very close to the

llS

 

-31 squares 3 k6 initial
plus signs 7 k6 initial

diamonds 3 k6 “aged”
triangles 7 k6 “aged”

 

xem( x105 emu/mole)
J.

 

 

D
’7 0 0+
0 Dang
B D +++ +
no a ++
4 DD 0 u+++++a+g
D o '
" 3899000022 go, A 5° 2
A§°o A
Hi
'10! r 1 l l I I l l I 1 l l
0 4| I m I“ w m

Figure 32. Magnetic susceptibility plot of 16'“ versus T for
Li+(CH3CH2NH2)4Na' obtained at two different fields (3

and 7 k6) and two different times, the initial data

collection and the data collection after the sample had
“aged”.

116

experimental value obtained for the sodide anion in
Li+(CH3CH2NH2)4Na- of -9 2: 15 x 10"5 emu/mole. This compares

well with the value obtained by D. Issa of -10.5 x 10'5 emu/mole for
Na' in Cs+(18C6)2Na'.123

One of the problems with the magnetic susceptibility data for
the sample is the unexpected field dependence of the magnetic
susceptibility. The residual field in the SQUID magnometer would
have to be on the order of 500 G to obtain the magnitude of field
dependence observed in the sample. In order to determine the
origin of the field dependence, the quartz tube of the SQUID cell was
run separately. Although field dependence was observed, this could
be attributed to a residual field of 50 G. It is therefore unlikely that
the field dependence of the sample is due to the SQUID cell or
residual field. M. Kuchenmeister obtained field dependent data for a
sample of pure lithium metal. The residual field in the SQUID
magnometer would have to be on the order of 600 G to obtain the
magnitude of field dependence observed for the lithium metal
sample.124 Therefore, the field dependence of the
Li+(CH3CH2NH2)4Na' sample appears to be due to a field dependent
impurity in the lithium metal. The impurity does not contribute to
the magnetic susceptibility of the decomposed sample or it would be
subtracted out with the diamagnetic background of the decomposed
sample. The impurity might be destroyed during the decomposition
of the sample. The field dependence is likely to be the result of a
small amount of ferromagnetic material such as iron.

In contrast to the diamagnetism of the sodide, the compounds
Li+(CH3CI-12NH2)4K' and Li+(CH3CH2NH2)4Cs' are paramagnetic.

117

Their magnetic susceptibility is similar to the susceptibility data of
liquid Li(NH3 )4 which has a temperature independent value of

60 x. 10'6 emu/mole, consistent with a Pauli metal.18 Figure 33
shows a plot of xmolar versus temperature obtained at two different
fields. The data taken at a field of 7 kG at low temperatures appear
temperature independent and paramagnetic. The value of the molar
susceptibility, approximately 65 x 10'6 emu/mole, is quite close to
the value of the known metal Li(NH3)4.18 For the data obtained at a
field of 7 k6, the sample became more paramagnetic above a
temperature of 140 K. A small amount of K" might dissociate to form
potassium metal and electrons at high temperatures. This is one
possible explanation of the increase in paramagnetic behavior in the

sample.
Unlike the Li+(CH3CH2NI-12)4Na' sample, the field dependence

in the Li+(CII3CI-12NI-I2)4K' sample cannot be attributed completely to

a paramagnetic contaminant. A residual field on the order of 1.6 kG
would be necessary to account for the field dependence of the
susceptibility for the sample at 3 and 7 k6 in the “aged" sample. The
potassium metal used to synthesize the compound was never tested
for field dependent behavior, therefore, a new source of field
dependent contamination may be due to the potassium metal as well
as the lithium metal. If the sample is field dependent, the NMR
chemical shift may also be field dependent. Variable field NMR data
for the 7Li chemical shift would be of interest to determine if the .
sample is field dependent or the field dependence is due to the

impurity.

118

 

 

 

9
+6
I t
A A
+ ++A
H + + + + A A
1"? ‘I' .1. A
A has 1 1 + + a +
33’ a ”stil‘ataxa“ A t A A
a ‘-
3
g Squares 3kGinitial
5.. plus signs 7 k6 initial
'51»
x diamonds 3 k6 “aged”
v 4- triangles 7 k6 “aged”
E
o . o o o
N o 0
Sivoo°e°o°o°o°o°°°o o o o o o o
81
an n n an
n a n n n n n a n n n n
a D a D a U D U U U
‘HITIIIFITIIIIITIIII
1| 4| 0 I II t! m 160 m an 220

TEMPERATURE (K)

Figure 33. Magnetic susceptibility plot of xem versus T for
Li"'(CH3CI-12NH2)4K‘ obtained at two different fields (3

and 7 k0) and two different times, the initial data
collection and the data collection after the sample had

“aged”.

 

119

The temperature dependent increase in susceptibility at higher
temperatures is reflected in the chemical shift value of the 7Li peak
for Li+(CH3CH2NH2)4K’. As with the susceptibility, the chemical
shift value of the 7Li peak increases at higher temperatures with a
value of approximately 18 ppm at 210 K. This value is more
paramagnetic than the chemical shift value of 2 ppm for the
diamagnetic compound Li+(CH3CH2NH2)4Na' in the same
temperature region”.65 The value of the 7Li chemical shift of the
known metallic compound Li(NH3)4 of 20.3 ppm at 213 K is quite
close to the value obtained for Li+(CI-I3CH2NH2)4K'. A plot of the
chemical shift value of Li+(CH3CH2NH2)4K' versus temperatures is

shown in Figure 34. In the temperature range of 253 to 193 K for
very concentrated Cs/ammonia solutions, the variation in chemical
shift versus temperature is linear.80 In these Cs/ammonia solutions,
the electrons are delocalized and the solution has metallic
character.125

The same temperature dependence is also observed in the 7Li
chemical shift of the compound Li+(CH3CH2NH2)4Cs' and is shown in
Figure 3S~ The 7Li NMR data for this compound were obtained in a
temperature region where the magnetic susceptibility of the

compound became more paramagnetic with increasing temperature.
The 7Li peak was sharp and relatively narrow, with a value of Avl /2

of less than 350 Hz. 133Cs NMR data for Li+(CH3CH2NH2)4Cs-

consisted of a broad rolling baseline with no distinguishable peak for
data collected over 15,000 scans.
The magnetic susceptibility data for the “aged” sample have a

temperature independent value of approximately 105 x 10‘6

120

 

”-1

CHEMICAL SHIFT (ppm)

1.1

 

 

 

Figure 34. Plot of the 7Li NMR chemical shift value for the compound
Li+(CI-I3CH2NH2)4K' versus temperature.

_121

 

 

 

 

at
a.
11" i-
+
E "
5 t! +
E
g 1.0-1 I
S u- +
U 1" 1.
u.
+
u.
13 T I l l l l y
‘- u u. ‘
mmmm

Figure 35. Plot of the 7Li NMR chemical shift value for the compound
Li+(CH3CH2NI-12)4Cs' versus temperature.

122
emu/mole below 120 K at a field of 7 k6. Between 120 K and 140 K,

the magnetic susceptibility contains a discontinuity as shown in
Figure 36 for the 7 kG data. This change in susceptibility might be
due to a phase change in the compound, however no DSC data have
been obtained for these compounds. At higher temperatures (180 K
and above for the 7 k6 as well as the 3 kG data that are shown in
Figure 37) the susceptibility becomes more paramagnetic with

increasing temperature. As with the compound Li+(CH3CH2NH2)4K',

this increase towards more paramagnetic behavior may be due to CS"
dissociating to cesium metal and e'. The Li+(CH3CH2NH2)4Cs'
magnetic susceptibility data in Figure 37 for the 3 kG field study
show two different curves at low temperatures. The top curve in
both the initial data collected and the data obtained for the “aged”
sample had a more paramagnetic susceptibility for the sample when
it was rapidly cooled to 5 K at zero field followed by the application
of a field and data collection. The lower more diamagnetic curve in
each of the data sets shown was obtained by slowly cooling the
sample in a field of 3 k6; this effect disappeared at higher fields. The
field dependence of the susceptibility data for Li+(CH3CH2NH2)4CS'
was of the same magnitude as that of the lithium metal. The effect
of the lithium impurity might be eliminated by purification of the
metal and should be investigated.

If the individual atoms are sufficiently large or their density is
sufficiently high, the valence electrons which had been bound to
their respective atoms or ions are set free via mutual polarization
and the solid becomes metallic.126 K. F. Herzfeld recognized this link
between atomic properties, density and the metallic state.127 The

123

 

xcm( x105 emu/mole)

 

 

 

a.
IZJL- +
squares 7 k6 initial
plus signs 7 k0 “aged” 4'
' - a
a.
11.0% n
it + + +
- +*++++++++++ + +
+ a
lo.o~ *
n
D
i. . .
a cannon n 0 n n n a D
u D a O
9.0- a
D
' I ' T l T l l I l
I I 1. II 1. as
TEMPERATUREOQ

Figure 36. Plot of lgm versus T for Li+(CH3CH2NII2)4Cs’

obtained at 7 k6 for two different time periods, the initial
data collection and the data collection after the sample
had “aged”.

124

 

 

 

F
' n
. +
n 01 squares 3 k6 initial *
plus signs 3 k6 “aged” 0
,3 +
185 10.0-l + + + D
3 + + ‘I'
.5, ‘+ + + + + + i’ *1 + n
+ +
'3 , 1
— + D
at 9.01 + D D D D
"’ 0 n a
5 . + + n n a n
)3 n U D D D D
9 a a a
8.0- a
D
a
.4
7'0: I I I I I r r1 I T I I I I 1 I I I I I
I 40 I C II 1’ “I II I” an
TEMPERATURE“)
Figure 37. Plot of 1:10 versus T for Li+(CH3CH2NH2)4Cs' obtained

at 3 k6 for two different time periods, the initial data
collection and the data collection after the sample had
“aged”.

 

125

two properties that are of importance are the force that holds an
ionizable electron in place versus the density of valence electrons in
the solid. The gaseous alkali metal atoms have positive electron
affinities that range from 0.62 eV for lithium to 0.47 eV for
cesium.128 Therefore if the density of the ethylamine solutions
presented here are equivalent, the force that hold the ionizable
electron decreases from sodium to cesium. This decrease in force
could cause a non-metal to metal transition in these ethylamine
systems as one progresses from sodium metal to cesium metal.
Although the data that have been presented are very
preliminary, the Li+(CI-I3CH2NH2)4K’ and Li+(CH3CH2NH2)4Cs'
systems do not have the diamagnetic susceptibility nor the 7Li
chemical shifts expected for an alkalide. As with the Li(NH3)4
systems, the behavior of these new systems appears more metallic in

nature.18 For example, the Fermi temperature of the compound
Li(NH3)4 is 6,300 K while the Fermi temperature of the compounds

Li+(CH3CH2NH2)4K' and Li+(CH3CH2NH2)4Cs' are 5,800 K and 3,600
K respectively.18 The Fermi temperature is related to the energy
per electron in the ground state and a simple way to compare the
Pauli paramagnetism of a metallic compound to the Curie law form is
to find the temperature through the equation x = CIT. For
comparison, the Fermi temperature of sodium metal is 37,700 K and
decreases to 18,400 K for cesium metal.129

The data from the studies presented here must be reproduced
on other samples. The nature of the difference in magnetic
susceptibiltiy data collected from “fresh” and “aged” sample must be

investigated. Conductivity measurements could determine if these

126

compounds exhibit metallic behavior, but the conductivity for the
compounds would be very difficult to obtain. If the ethylamine
solvent were to partially evaporate or “bump”, the lithium and
cesium or potassium metal that was deposited would have metallic
behavior. This would make it difficult to differentiate the
conductivity of the ethylamine compounds from that of deposited
metal. The optical absorption spectra of these systems could be
obtained to look for the existence of a plasma edge.

The difficulty with these samples is the extreme sensitivity of
the ethylamine solutions to temperature gradients that cause the
solvent to “bump”, which destroys the stoichiometry of the solution.
Another difficulty is that metal precipitation cannot be easily
detected. The intriguing properties of these systems suggest that the
further characterization might verify a new class of “metallic”

compounds.

V SPIN-ECHO NMR STUDIES
W

Solid state NMR has been used extensively as a probe of the
alkali metal environment in alkalides. Alkali metal NMR of the alkali
metal anions has been observed for all of the anions synthesized,
Na', K', Rb', and Cs' in crystalline alkalides.47r57r53r59r60 As shown
in Chapters 2 and 4, both the cesium and lithium cation have also
been studied extensively with single pulse solid state NMR
techniques.

Due to the low sensitivity of the 39K nucleus, single pulse
techniques could not be used to obtain the NMR spectra of the
potassium cation in solids. An added problem with this nuclei is the
low magnetogyric ratio that leads to probe “ringing” and either
distortion or severe broadening of the line. Oldfield et. al. reported
the use of a two-pulse spin-echo sequence with appropriate phase
cycling to obtain relatively undistorted second order powder pattern
spectra of the nuclei 39K in solids.64 A study was undertaken to
obtain 39K and 87Rb NMR spectra for alkalides or electrides that
contain the complexed cations Rb+ and K+. Rubidium nuclei were
included in the study since previous attempts to obtain NMR spectra
for the rubidium cation in solid state NMR experiments had been
unsuccessful due to the large quadrupole moment of this nucleus.

A collaboration with L. E. H. McMills, J. Kim, and A. Ellaboudy to

obtain data on both the potassium and rubidium cations was

127

128

successful.55v66 However, the lineshapes of the rubidium spectra
were extremely difficult to phase. The reason for this difficulty in
phasing the spectra appeared to be due to the extremely broad
linewidths of the rubidium spectra. An NMR instrument with a fast
digitizer and therefore a larger sweepwidth was used to repeat the
data collection on the alkalides and electrides that contain rubidium
cations. This work has been continued at the new NMR facility at

Michigan State University in collaboration with J. Kim.

Mm
Was

As with the 7Li nucleus with a spin I=3/2, the 87m: nucleus
posseses a quadrupole moment. Nuclei with spins greater than 1/2
have quadrupole moments that can interact with electric field
gradients and cause the 21 + 1 nuclear energy levels to shift and

broaden. The Hamiltonian can be written as the sum of the Zeeman
Hamiltonian, “Z, and the quadrupole Hamiltonian, HQ.

The effect of flQ is usually treated as a perturbation on the Zeeman

terms. The quadrupolar Hamiltonian is given by:

9qu l 2 z 2
. __ 31 ~10 l) 1/21’I(I +1-)l (5.2)
”0 41(21-1) z + + ”

129

The axes X, Y, and Z are the principle axes of the tensor that
describes the electric field gradient. The case "Q = 0 corresponds to

axial symmetry of the electric field gradient of the nuclei when vxx
vyy' The asymmetry parameter, nQ, is a measure of the deviation

from axial symmetry and defined by:

There is no first order shift for the central transition for 37kb,
a half-integer spin. The second order effects are inversely
proportional to the external magnetic field as defined by equation
1.4. In order to interpret the environment of the rubidium cation in
the compounds studied by the spin-echo NMR technique it is
necessary to determine the values of the asymmetry parameters and

the quadrupole coupling constants (QCC) = equ/h.
Wanna

Nuclear spins that have been excited to higher energy levels
return to lower energy states through the transfer of energy. This
process is called relaxation. When the transfer of energy is to the
surroundings of a particular nucleus, the variation of the population

difference between the energy levels as a function of time follows an
exponential decay curve with a time constant T1, the spin-lattice

130

relaxation time. The value of the spin-lattice relaxation time can be

obtained from the following equation:
Mt " MO I 1 - OXP(-t/T1)1 (5.4)

where M0 is the magnetization at infinite time and Mt is the

magnetization or intensity of the signal at time t.

Nuclei exchange their magnetic energy with their surroundings
indirectly through magnetic dipolar or electric quadrupole
interactions. There are several principal types of magnetic
interactions that contribute to spin-lattice relaxation in quadrupolar
nuclei. For quadrupolar nuclei, there is an effective coupling of the
nuclei to the surroundings. The non-spherical nuclear charge
distribution can interact strongly with electric field gradients.
Induced fluctuations in the electric field gradient at a particular
quadrupolar nucleus due to lattice vibrations give rise to quadrupole
relaxation.

Dipole-dipole interactions contribute to spin-lattice relaxation
and occur when the nucleus experiences a fluctuating field due to
motion of neighboring dipoles. In non-quadrupolar nuclei, this is an
important magnetic interaction that contributes to spin-lattice
relaxation. Another type of interaction is chemical shift anisotropy
that arises due to chemical shielding of the nucleus and depends on
the orientation of the molecule with respect to the magnetic field.
Fluctuations in the weak magnetic secondary field due to the
electrons surrounding the nucleus precessing in the magnetic field

occur due to movement in the molecule. In contrast to spin-lattice

131

relaxation, spin-spin relaxation is the decay of the transverse
magnetization or spin coherence. Mechanisms of spin-lattice

relaxation are also mechanisms of spin-spin relaxation.
mm
mm

The synthesis of alkalides and electrides has been described in
detai129r34, and these techniques were used in the synthesis of the
compounds used in this study. The methods used to synthesize the
model salts used in the study, as well as the methods to grow single
crystals of models salts have also been described in detail

elsewhere.6 5

am

Static 87Rb spectra were obtained at 130.9 MHz on a Varian
VXR-400 spectrometer equipped with a 45 to 165 MHz broad band
probe. A solution of RbCl was used to determine the 90 degree
solution pulse for this nucleus and half of this value, 1.7 ltsec, was
used for the 90 degree solid pulse in experiments. The spin-echo
pulse sequence is shown in Figure 38. Phase cycling was used to
cause destructive interference of the free induction decay tails, but
coaddition of the spin-echo signals.64 The program for the spin-echo

pulse sequence is given in Appexdix B. The spin-echo sequence
allowed both pulses, P1 and P2, and tau values, as well as the delay

132

time to be varied independently. With constant tau values and delay

times, the pulse lengths were varied to determine the relative signal

Figure 38. The spin-echo pulse sequence.

to noise ratios for several combinations of pulse lengths. As shown
in Figure 39 for the compound Rb+(18C6)Cl', the largest signal to
noise ratio was obtained for the 90° - 90° or 1.7 - 1.7 usec spin-echo
pulse sequence. Optimum tau values were determined through the
systematic variation of the tau values until the probe “ringing” was
absent without significant loss of signal due to the decay of the FID.
The optimized tau values determined by this method were taul = 50
usec and tau2 = 44 11sec. The delay time, D1, between each pulse
sequence was varied from 0.3 to 2.0 seconds, with most spectra
obtained with a delay time of 0.5 seconds. The spectra of model salts
were obtained at ambient temperature. The spectra of alkalides and
electrides were obtained at 223 K. For low temperature data
collection, cold nitrogen gas was used to control the temperature to
within 0.3 K after 3 to 10 minutes equilibration time.

Single crystal NMR spectra were obtained with the spin-echo
' pulse sequence for Rb+(C222)Cl' on a home built single crystal NMR

133

1800-1800

90°-180°

900-900 ‘
W-

450-900

 

.0

 

vvvvvvvvvvvvvvvvvvv 'vvvv'vvvw'vaVIVV‘V'I'VV'IVVI'IVVv'vVvv'vvVI'VVVVIV- v vvu'uuvv'v-vv'vvvv'vvvv'vvvv'v'vv'

mmmm-e-m-m-aoo-aoo DI.

Figure 39. The 87Rb NMR spectra obtained for Rb+(l8C6)Cl' at
different pulse lengths to determine the optimum pulse

lengths for the spin-echo pulse sequence.

134

probe tuned for 8711b. The experiments were performed on a Varian
VXR-400 instrument at 130.9 MHz at ambient temperature. The
angle of the sample in the single crystal probe was controlled by a
one dimensional goniometer that can be rotated through 180
degrees. The single crystal of Rb+(C222)Cl' used in these
experiments was mounted in a cubic box and data were collected
along the a or b crystalline axes; these axes are equivalent.

The spin-spin relaxation time of the model salt was determined
at four different orientations of the crystal with the pulse sequence

shown in Figure 40. Tau values were varied from 20 to 230 usec
with a delay time, D1, of 0.5 or 1 second for this experiment.

90° 180°

Figure 40. The pulse sequence used to determine the spin-spin

relaxation time of Rb+(C222)Cl' at four different
orientations of the crystal.

Data were fit with a simple exponential with KINFIT, a nonlinear
curve fitting program.”0 The spin-lattice relaxation times were
determined at the same four orientations used to determine the
spin-spin relaxation time. The variable delay of the pulse sequence
shown in Figure 41 was incremented from 0.005 to 1.0 seconds and
the data fit with an exponential by KINFIT.

13 5
90° 90° 180°

m \Dl - 0.5 or 1.0 second

Variable 20 usec lsnsec
Delay

Figure 41. The pulse sequence used to determine the spin-lattice

relaxation time of Rb+(C222)Cl' at four different
orientations of the crystal.

11128 1 ID' .
1.1532132111121932:

The peak for the rubidium cation was observed in both
Rb+(15C5)2Rb' and Rb+(15C5)2e‘. The lineshape for the compounds
was simulated with VMASS, a program written by J. Kim.°° This
program can be used to simulate either static or spinning NMR
spectra lineshapes and accounts for chemical shift anisotropy, dipolar
and quadrupolar interactions. For the rubidium NMR spectra
simulations in this study, the chemical shift anisotropy interactions
were assumed to be negligible.

The dominant feature in the NMR spectra of Rb+(15C5)2Rb' is
the narrow rubidium anion peak at -l95 ppm. The quadrupolar
interaction of the anion is estimated to be less than 2 MHz. From the
small quadrupolar interaction observed and the chemical shift value
of the rubidium anion, one can infer that the spherical symmetry of
the 58 orbital for the anion is nearly unperturbed by surrounding

charges and dipoles.

136
When the spectra of Rb+(15C5)2Rb' were expanded along the

vertical axis, an underlying peak was observed and is believed to be
due to the complexed rubidium cation, Rb+(15C5)2. The estimated

value of the quadrupolar coupling constant (QCC) of the cation in
Rb+(15C5)2Rb' is 11 to 12 MHz, however; even after 23,000 scans

the signal to noise ratio was poor and the asymmetry parameter
could not be determined. The spectra of Rb+(15C5)2Rb' and
Rb+(15C5)2e' are shown in Figure 42.

An NMR peak at a chemical shift value of -193 ppm in the
compound Rb+(15C5)2e', shows that there was a small amount of

Rb+(15C5)2Rb' in the electride. The signal to noise ratio in this

spectrum was poor, even after 43,000 scans. The value of the QCC
for this compound was determined to be approximately 8 MHz. The

difference in the values of the QCC constant indicates that the
environment of the complexed rubidium cation in Rb+(15C5)2Rb' is

different from that of the rubidium cation in Rb+(15C5)2e‘; however,

the crystal structure of the electride is not known and therefore no
direct comparison can be made. Although relatively narrow with a
linewidth of approximately 650 ppm, the NMR peak for the rubidium
cation in Rb+(15C5)2Na' had an extremely poor signal to noise ratio,
even after 24,000 scans. No simulations of this spectrum were
performed.

Although the rubidium anion has been observed in single pulse
and in spin-echo experiments, the rubidium cation could be observed
in the spin-echo experiments in the 15C5 complexes only after
several thousand scans and even then only with a poor signal to

noise ratio. The poor signal obtained for these complexed cations

137

Jr .

 

 

 

 

 

 

V v v v v V ' ‘
V T Y—r v v v v j T ‘t v v ‘v V 1
IL ass 0 0. ‘1”

(PP!!!)

Figure 42. 87kt NMR spectra of A) Rb+(15C5)2Rb' and B)
Rb+(15C5)2e' obtained with the spin-echo pulse
sequence.

138

might be due to an extremely short relaxation time for the cation,
therefore most of the FID could have decayed before acquisition.
When the tau values in the pulse sequence are shortened in attempts
to correct this problem, baseline distortion becomes a problem. If
the temperature at which the spectra were obtained were lowered,
the spin-lattice relaxation time would increase and a better signal to
noise ratio might be obtained for the complexed cation in these

compounds.

Wes

NMR spectra were obtained for the alkalides Rb+(18C6)Rb' and
Rb+(18C6)Na' with the spin-echo pulse sequence and the spectra are
shown in Figure 43. Although fewer transients were observed, the
relative signal to noise ratios for these compounds are better than
the three 15C5 complexes studied. One possible explanation is that
the relaxation time of the complexed cation has increased.

In contrast to the rubidium anion in Rb+(15C5)2Rb', the anion
in Rb+(18C6)Rb' has a non-axially symmetric powder pattern. The
rubidium anions in Rb+(l8C6)Rb' form chains throughout the
structure as shown in Figure 44 and the rubidium cation is in contact
with the anion.131 The non-axially symmetric lineshapes and
second order quadrupolar broadened NMR spectra for the rubidium
anion with a QCC value of 5 MHz are consistent with the highly
anisotropic crystal structure. The large paramagnetic shift of the

rubidium anion of approximately 70 ppm also suggests that there is

139

 

 

Figure 43. 87m: NMR spectra of A) Rb+(l8C6)Na’ and a)
Rb+(18C6)Rb' obtained with the spin-echo pulse
sequence. The solid lines are the observed spectra and
the dotted lines the simulated spectra.

140

 

Figure 44. A cross section of the crystal structure of Rb+(l8C6)Rb‘
showing the rubidium anion chains that are presentl 31

141

significant overlap of the p and d orbitals on the anion with its
surroundings.

' The rubidium cation in Rb+(l8C6)Rb' has an 11 value of 0.16,
therefore, neither the cation nor anion is in an axially symmetric
environment. This is consistent with the crystal structure and the
NMR data observed for the rubidium anion. The QCC value for the
rubidium cation, asymmetry parameter, and chemical shift value in
Rb+(18C6)Rb' of 11.2 MHz, 0.16 and 105 ppm respectively are quite
similar to the values determined for Rb+(18C6)Na’ of 10.8 MHz, 0.15
and 110 ppm respectively. Although the crystal structure of
Rb+(18C6)Na' is not known, the similarities in the NMR data obtained
for the rubidium cation in these two compounds suggest that the
crystal structure of Rb+(18C6)Na' is similar to the crystal structure of
Rb+(l8C6)Rb‘. '

Wax

The crystal structure of Rb+(I-IMHCY)Na' is known.74 The
separation between the Van der Waals surfaces of the cation and
anion is between 0.02 and 0.22 A. The asymmetry parameter of 0.8
as well as the large QCC value of 13 MHz reflects the nonsymmetric
environment of the rubidium cation in the crystal structure. The
spectra of the compound and the simulation are shown in Figure 45.
This was the first time that the rubidium cation had been observed
in a fully methylated aza analog of a crown ether by solution or solid

state NMR techniques.

142

 

 

 

Frame 45. 87ab NMR spectra of Rb+(HMHCY)Na' obtained with the
spin-echo pulse sequence. The solid line is the observed
spectrum and the dotted line is the simulated spectrum.

143
W

- The spin-echo NMR technique was used to obtain NMR spectra
for several model salts and the alkalide Rb+(C222)Rb'. Table 9
contains the values of the QCC, asymmetry parameter and chemical
shift values obtained for these compounds as well as the values
Obtained for the other compounds discussed in this chapter for
Comparison. The rubidium cation in Rb+(C222)I' and Rb+(C222)SCN'
model salts is not in an axially symmetric environment since both
Compounds have rather large asymmetry parameters determined
from lineshape simulations of the powder patterns. The spectra of
these two compounds are shown in Figure 46.

In contrast, the rubidium cation in the compounds Rb+(C222)Cl'
and Rb+(C222)Rb‘ is in an axially symmetric environment. The
Spectrum of Rb+(C222)Rb' is shown in Figure 47. These differences
in axial symmetry may be due to changes in the crystal packing in
these compounds that are dependent on the anion. All of the
l"lbidium cations complexed by the cryptand C222 have large
Quadrupole coupling constants. The large electric field gradients
Present in these compounds could be due to the close proximity of
the unpaired electrons on the oxygens to the rubidium cation in the
Cryptand. The average Rb‘l' -0 bond distance in the compound
Rb+(C222)Rb' is 296(6) 11.131 '

The rubidium anion in Rb"'(C222)Rb' is in an axially symmetric
ellvironment; however, the line is broadened due to quadrupolar
interactions. The crystal structure of this compound indicates that

the rubidium anions form dimers.131 This is similar to the anion

144

Table 9. NMR parameters determined from ,the simulation of

powder patterns from 87Rb NMR spectra.

 

Skunnnnnd
Rb+(15C5)2Rb'*
Rb+(15C5)2Rb'*
Rb+(15C5)2e'*
Rb+(18C6)Na‘
Rb+(l8C6)Rb'*
Rb+(18C6)Rb'
IHfiKEDdHKfiOMhr
Rh+(C222)C1-
Rb+(C222)I-
Rb+(C222)SCN-
Rb+(C222)Rb-

Rb+(C222)Rb'

QCQMHZ)

~11 to 12

<2

11.2
13.0
16.4
18.2
17.8
6.25

16.6

11

0.15
>0
0.16

0.8

0.82

0.7

mm
~2oo

-195

105
80

70

27

70

 

QCC = quadrupole coupling constant

1] = asymmetry parameter

5 = chemical shift
"' = estimated values

145

 

rTnnTTnnTTrnnTTrnq11TTrnnrpnTTrnnTTTnanTTrpnw-
4000 2000 O -3000

(PP!!!)

Figure 45. 87ab NMR spectra of the model salts A) Rb+(C222)SCN'
and Rb+(<:222)r- obtained with the spin-echo pulse
sequence.

146

 

 

 

 

Figure 47. 37Rb NMR spectrum of nb+(c222)Rb- obtained with
the spin-echo pulse sequence. The solid line is the
observed spectrum and the dotted lines are the
simulated spectra.

147

chains in the compound Rb+(l8C6)Rb’. Both compounds have
relatively large QCC values for the rubidium anion combined with a
paramagnetic chemical shift from the expected value for a rubidium
anion in a spherically symmetric 5s orbital. The NMR lineshape of
the rubidium anion is quite sensitive to the environment of the anion

in the crystal structure.
W

Although the lineshapes of the experimental data could be
simulated, the relative intensities of the powder pattern of the
simulated spectra did not always correspond to the experimental
data. The transmitter offset was changed to center it on different
portions of the powder pattern but this did not correct the difficulty
with relative intensities in the powder pattern. Therefore the
problem does not appear to be the inability of the pulse to excite the
entire linewidth of the powder pattern. Single crystal NMR data for
the model salt Rb+(C222)Cl' were obtained by J. Kim to determine
the reason for the loss of relative intensity in some of the powder
patterns. The single crystal NMR orientation plot for Rb+(C222)Cl' is
shown in Figure 48. The single crystal data were simulated with
KINFIT and XTAL, a program obtained by J. Kim for the simulation
of single crystal NMR data.66

The spin-lattice and spin-spin relaxation times were
determined at four different orientation of the single crystal. As
shown in Figure 49, the spin-lattice relaxation time does not appear

to change significantly at the different orientations studied.

148

1000
3 e Calculated .

A 0 Observed
E 500‘ ' '
o. '2
a. .
v .
«u ‘ 0 O 0
.3 o': . e
m 1
8 "5m"; ' : . ‘
s r .
.C ‘40001 e . e
0 j e e

-1500 . . . . . , . . . . .

 

 

0 so so 9'0 130 'léof'l'ao
Angle of Rotation (Degree)

Figure 48. The single crystal 87kt NMR orientation plot for the
compound Rb+(C222)Cl'.

149

 

 

 

 

A
—r e I l I I I I I I I I I '1 I I I I I I I I l T I I' I T I» ‘I
1000 500 o -5oo -looo -1500
(PP!!!)
2001
J + T1(ms)
. e T20»)
0 J
E 150- °
F a
c I a
~3- ram
0 d
e * °
‘8 1 °
a: 50':
B o‘ ’ ’ ’
1000 500 6 -§oo -1600 4500

Chemical Shift (ppm)

Figure 49. A) The "Rt: NMR spectrum of Rb+(c222)c1- obtained
with the spin-echo pulse sequence for a powdered
sample and B) Plot of the relaxation time (T1'

spin-lattice and T2 spin-spin) versus chemical shift

of the rubidium cation for four orientations of the single
crystal Rb+(C222)Cl'.

150

However, the spin-spin relaxation time decreased from 154 usec at a
chemical shift value of 800 ppm to 63 psec at a chemical shift value
of -600 ppm for a different orientation of the single crystal.

This indicates that loss of spin coherence in the transverse axis
' for certain crystal orientations may be the cause of the relative
intensity loss in portions of the powder pattern. Although the shape
of the powder pattern can be simulated to determine the value of
QCC and n , the relative intensities of the powder pattern between
simulated and experimental data cannot be compared directly.

This study has shown that the spin-echo technique with phase
cycling is a powerful tool in the determination of the environment of
rubidium in alkalides and electrides. The fast digitizer on the Varian
400 MHz instrument at Michigan State University has allowed the
observation of rubidium cation linewidths of over 2,000 ppm. The
pulse program and optimal delay times that have been determined
will allow routine determination of rubidium NMR spectra for new
alkalides or electrides. The further study of spin-lattice and spin-
spin relaxation time in these compounds would be of interest. This
study may also be extended to 85Rb for the comparison of the value
of the QCC for the observed powder patterns obtained by NMR
methods as well as the comparison of the relative relaxation times of
87 Rb/85Rb in alkalides and electrides.

vr. CONCLUSIONS AND SUGGESTIONS FOR FUTURE WORK 7

The chemical shift of the cesium cation is extremely sensitive

' to the unpaired electron density at the nucleus. 133Cs NMR has
proven to be a useful tool in the study of the electrides Cs'*’(15C5)2e'

and Cs*'(18C6)2e'.76 The fractional atomic 3 character of the electron

in these compounds is quite small, an indication that the electrons
have only weak overlap with the complexed cation. This is further
evidence that the electron density in electrides is centered in the
empty anionic cavities present in the crystals. If a high-quality
single crystal X-ray study of an electride could be done in which the
background noise is below the electron density of the trapped
electron, the unpaired electron in the cavities might be observed.
This would offer further proof of the location of the electron in
electrides.

The study of Cs+(15C5)(l8C6)Na', a mixed crown ether system,
has shown the potential to “fine-tune” the characteristics of alkalides
and electrides based on the appropriate choice of crown ethers and
alkali metal cations. The problem with these systems arises from the
difficulty in obtaining pure compounds for the study of bulk
properties. However, the synthesis of a pure compound
Cs+(15C5)( 18C6)e’ would allow the study of a system intermediate
between the two electrides Cs+(15C5)2e‘ and Cs+(18C6)2e', and
should yield information about how structure affects properties.
Therefore, the X-ray structures and the further characterization of

the properties should be pursued for these compounds.

151

152

The study of alkalides and electrides that contain fully
methylated aza analogs of crown ethers and cryptands has resulted
in the synthesis of several new and reasonably stable compounds.
Studies of the compound Cs+(HMHCY)Na' have demonstrated that
'defect electrons can cause significant changes in the properties
observed. The electron dopant level in this compound was 33
percent, but attempts to dope Rb+(HMHCY)Na' or K“'(HMHCY)Na' to
large levels have not been made. Because trapped electrons can
affect the properties and may be present at high levels, the electron
dopant levels in alkalides should be more systematically studied
when a compound is synthesized. This could be achieved by EPR or
magnetic susceptibility measurements. A study to determine to
what levels these systems can be doped would be of interest as well
as the effect of dopant level on the physical properties. The ability
to achieve extremely high dopant levels in Cs+(HMHCY)Na’ may be
due to a slow rate of decomplexation. The electride could not be
synthesized as the system decomplexed upon solvent removal. Other
systems such as Rb+(l8C6)Rb' and Rb+(C222)Rb', for which the pure
electride is not known to exist, should also be doped with excess
electrons and the resultant dopant levels and properties studied.

The compounds synthesized with TMTCY have allowed the
observation of solvent in the crystal structure of alkalides for the
first time. The sodide anion is not the only anion present in these
systems (as an amide ion was also present) and a pure electride
could not be synthesized. However, if a better “fit” were achieved
between the cation and complexant such as in the case of

tetramethyltetracyclen, an electride and pure sodide might be

153

synthesized. The work done in the TMTCY system has also
demonstrated the need, especially for lithium compounds __to test for
the presence of solvent in the crystals and to isolate crystals that
have not been subject to high vacuum. '
' In contrast, the TMPAND complexant forms a pure sodide with
no methylamine contamination. The use of DSC techniques in the
investigation of the decomposition kinetics of Li+(TMPAND)Na' has
shown the usefulness of these methods in the determination of the
activation energy of decomposition. Further research on the
decomposition kinetics and mechanisms would provide information
that might be useful in the choice of complexants for the synthesis of
more thermally stable alkalides and electrides. The QCC measured in
Li+(TMPAND)Na' for ha is the first time the value of QCC has been
obtained for the complexed lithium cation in an alkalide or electride.
Further studies of this type would allow the correlation of
quadrupole coupling constants with 7Li relaxation times. It would
also permit comparison of the ratio of the values of QCC for °Lil7Li.
The crystal structure of this compound should be obtained if
possible, both above and below the phase transition. Once the
kinetics of the complexation of the lithium cation by TMPAND at low
temperatures in methylamine have been determined, the electride
should be synthesized and its properties studied. The fully
methylated aza analogs of cryptands show great potential as
complexing agents and similar complexants should be used to

synthesize additional alkalides and electrides.
The simplest amine systems presented, Li+(CH3CH2NH2)4Na',

L1+(CH3CH2NH2)4K', and L1+(CH3CH2NH2)4CS', showed a progression

154

from the sodide, a diamagnetic material to the potasside and ceside
which are paramagnetic. The characterization of these systems by
magnetic susceptibility, EPR, NMR, and optical absorption . should be
pursued since further study may verify the supposition that these

' form a new class of “metallic” compounds. The electride
Li(CH3CH2NH2)4, as well as the rubidide, Li+(CH3CH2NH2)4Rb‘,
should also be studied to complete the series. The characterization of
more dilute solutions may also yield some insight into the properties
of the more concentrated solutions presented in this study.

The 87Rb NMR data collected with the spin-echo pulse
sequence have proven useful in the characterization of alkalides and
electrides that contain a complexed rubidium cation. This technique
can now be used in the routine study of new compounds. It could
also be used to obtain the 87Rb NMR spectra for
Li+(CH3CH2NH2)4Rb'. Single crystal NMR data should be obtained
for alkalides and electrides for comparison with the data already
obtained for the model salt Rb+(C222)Cl'. The spin-lattice and spin-
spin relaxation times might also be determined for the complexed
rubidium cation. The mechanisms of relaxation could then be
correlated with the values obtained for QCC from the simulation of
the experimental NMR lineshape and/or the variation of these
parameters with crystal orientation. To make the best use of such

data, the crystal structures must be determined by X-ray diffraction.

APPENDICES

APPENDIX A

APPENDIX A

Table of Positional Parameters and Their Estimated Standard

 

155

Deviations

_ Atom x y z ’B(A2)
Na 0.0010(7) 0.6646(3) 0.8155(3) 8.9(1)
N1 0.052(1) 0.4876(5) 0.5756(4) 6.3(2)
N4 -0.248(1) 0.4447(5) 0.5452(4) 5.9(2)
N7 0.006(1) 0.3834(5) 0.4700(4) 6.5(3)
N13 0.062(1) 0.5943(6) 0.2772(4) 6.5(3)
N16 0.014(1) 0.7461(5) 0.3347(5) 7.5(3)
N19 0.269(1) 0.6468(6) 0.3748(4) 7.7(3)
N101 0.100(1) 0.5884(5) 0.4354(4) 7.0(3)
C2 0.048(2) 0.4636(9) 0.6244(6) 9.9(5)
C3 0.205(2) 0.454(1) 0.6111(6) 10.3(5)
C5 0.245(1) 3615(9) 0.5293(8) 12.0(5)
C6 0.144(2) 0.3359(8) 0.4833(7) 1 1.6(5)
C8 0.115(2) 0.3675(8) 0.5137(6) 10.9(5)
C9 0.155(2) 0.4226(8) 0.5588(8) 1 1 . 1(5)
C10 0.145(2) 0.5590(7) 0.5941(6) 9.8(4)
C11 0.032(2) 0.3746(7) 0.4028(5) 8.1(4)
C12 0.397(1) 0.4810(9) 0.5345(7) 10.1(5)

156

 

 

C14 -0.010(2) 0.6603(7) 0.2423(6) 9.0(4)
C15 0.023(2) 0.7383(7) 0.2652(6) _: 8.8(4)
C17 ' 0.151(2) 0.7830(7) 0.3606(7) 111.1(6)
_ C18 0.289(2) 0.7348(8) 0.3596(8) 11.1(5)
C20 0.317(1) 0.5929(9) 0.3269(6) 9.3(4)
C21 0.233(2) 0.5935(9) 0.2673(6) 9.5(4)
C22 —0.007(2) 0.5179(8) 0.2607(7) 15 .4(6)
C23 -0.115(2) 0.791(1) 0.3545(8) 17 .9(7)*
C24 0.344(2) 0.6255(9) 0.4336(6) 14.3(5)
C102 -0. 150(2) 0.6456(9) 0.4748(8) 1 5 . 1(7)
Li 0.035(3) 0.631(1) 0.374(1) 6.3(5)
Li2 -0.076(2) 0.498(1) 0.4910(9) 5.6(5)
Starred atoms were refined isotropically.
Anisotropically refined atoms are given in the form of the
isotropic equivalent displacement parameter defined as:
(4/3) * [82*B(1,1) + 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)]
Table of Bond Distances in Angstroms
Atom 1 Atom 2 Distance Atom 1 Atom 2 Distance
N1 C2 141(2) N16 C17 1.47(2)
N1 C9 1.46(2) N16 C23 1.43(2)
' N1 C10 1.49(2) N16 Li 209(2)

157

N1 Li2 212(2) N19 C18 151(2)
N4 C3 l.44(2) N19 C20 1.42(2)
N4 1 C5 1.43(2) N19 C24 1.45(2)
N4 C12 1.46(2) N19 Li '2.07(3)
N4 Li2 209(2) N101 C102 134(2)
N7 C6 l.48(2) N101 Li l.90(2)
N7 C8 1.43(2) N101 Li2 1.93(2)
N7 C11 1.47(1) C2 (:3 l.41(2)
N7 Li2 2.06(2) C5 C6 l.38(2)
N13 C14 l.48(2) C8 C9 137(2)
N13 C21 152(2) C14 C15 1.43(2)
N13 C22 1.46(2) C17 C18 1.46(2)
N13 Li 214(2) C20 C21 1.46(2)
N16 C15 1.47(2) C102 Li2 258(3)

 

Numbers in parentheses are estimated standard deviations in
the least significant digits.

Table of Bond Angles in Degrees

 

Atom 1 Atom 2 Atom 3 Angle
C2 N1 C9 111.(1)
C21 N13 C22 13.0)
C2 N1 C10 112(1)
C21 N13 Li 104.1(9)
C2 N1 Li2 107.4(9)

C22 N13 Li 116(1)

C15

- C15
C10
C15
C3

C17
C3

C17

C5
C18

C18
C12
C18
C6

C6
C6

C8
C102
C8
C102
C11
Li
C14
N1
C14
N4
C14
N4

N16
N7
N16
N1
N19
N13

N1
N16
N1
N16
N1
N16
N4
N16
N4
N16
N4
N16
N4
N19
N4
N19
N4
N19
N7
N19
N7
N19
N7
N19
N7
N101
N7
N101

N101
N13

N13
C3
N13
C5
C6

C8
Li
C9
Li
C14

158

C10
C17

Li2
Li
C5

C12
Li
L12
Li
C12

Li2
C24
Li2
Li
C8

C11

C11

N19

N101
C8
N101
C15

108.9(9)
111.(1)

(101(1)
{112(1)
116.5(9)

307.7(9)

110.(1)
109.(1)
110(1)
100(1)
106.9(9)
116.(1)
112(1)
116.(1)
105.8(9)
112.(1)
112.8(9)
103.(1)
113(1)
108.(1)
109(1)
102.2(9)
102.1(9)
116.(1)
116.(1)
111.(1)
104.9(9)
103.(1)
111.8(9)
13l.(l)
111.(l)
120.(1)
111.(1)
118.(1)
102.1(9)
118.(1)
120.(l)
88.8(9)
121.(1)
124.(l)
121.(1)
131.(1)
115.(1)

N1
N16
N1
N16
N1
N19
N1
N19
N4
N13
N4
N101
N4
N13
N7
N13
N7
N13
N101

L12
C 15
L12
C 17
L12
C 1 8
L12
C20
L12
C21
L12
C 102
L12
L1
L12
L1
L12
L1
L12

159

N4
C14
N7
C18
N101
C17
C102
C21
N7
C20
N101
L12
C102
N16
N101
N19
C102
N101
C102

83.9(8)
115.( 1)
87.0(8)
117.( 1)
129.(1)
116.(1)
108.8(9)
118.(1)
86.7(8)
113.(1)
126.(1)
46.9(8)
107.3(9)
84.5(8)
129.(1)
86.4(9)
160.( l)
128.(1)
30.4(6)

 

Numbers in parentheses are estimated standard deviations in
the least significant digits.

 

APPENDIX B

APPENDIX B
#endif

/* ssecho - solid state echo sequence modified Varian program

' 180 1- -
d1 p d2 " x taul pwy tau2 acquire
tittttt tittt *ttttt
*tttt Mtttuttttt MttMMttMtt tittfltttttttfltttiw
parameters:
p180: optional 180 pulse for t1 studies. if p180 = 0, the
delay d2 is ignored
pl: first pulse, should be 90 degrees
if p1=0, the first pulse width will be pw instead
pw: final pulse, usually 90 degrees but sometimes less

then 90 for spin i>3l2 nuclei

compul: if y, each 90 pulse. is replaced by the composite
pulse 135(x) - 90(-x) - 45(x)
if n, conventional pulses are used

Modification history

110188 potentially negative delays removed -ehw
010389 Rx gating modified for 2.1 -ehw

*/

#include <standard.h>

pulsesequence ()
{
double taul,
tau2,
p180;
char compulIMAXST'R];
I * initialize parameters *1
taul = getval(“tau1”)
tau2 = getval(“tau2”)

160

 

16 1
getstr(“compul”, compul);
p180 = getval(“p180”0;

if (p1 = 0.0)
- p = pw /* insure a value for p1 pulse */
if (p180 = 0.0)
d2 = 0.0; I“ don’t do d2 delay if no 180 pulse“!
status(A);
delay(dl);
status(B);

rgpulse(p180, zero, rofl, 0.0);
if (p180 > 0.0) I‘if we have inversion-recovery */

{

if (d2<= rofl) /* and d2 is smaller than amp tumon time*/

{
rcvroff();

delay(d2);
1

else /"' when d2 is long */

{
delay(dZ - rofl);

rcvroff();
delay(rofl);
l

1

else /" no inversion-recovery */

{
rcvroff();
delay(rof 1);

}

if (compullol == ‘3’)

{ /* composite pulse experiment 1"I
fprintf(stdout, “ssecho - Using composite pulses/n”)’
hlv(ct,v1); l“vl=0011223344556677*/
hlv(v1,v1); /*v1=0000111122223333*/
dbl(v1,v3)l I‘v3=0000222244446666*/
add(one,oph,v2); /* v2=1 2 3 4 1 2 3 4 l 2 3 4 1 2 3 4 */
add(v1,v2,v2); l*v2=1234234534564567l
mod4(v2,v2); l“v2=1230230130120123*/
add(oph,v3,v1); I“ v1=012 3 2 3 4 5 4 5 6 7 6 7 8 9 */
mod4(vl,v1); /*v1=0123230101230301*/
add(two,v2,v3); /* v3=3 4 5 2 4 5 2 3 5 2 3 4 2 3 4 5. */
mod4(v3,v3); /*v3=3012012312302301*/

162

add(two,v1,v4); I'l' v4=2 3
mod4(v4,v4); /* v4=2 3
rgpulse(1.5 "' pl, v4, rofl, 0.0);
rgpulse(pl, v1, 0.0, 0.0);
rgpulse(0.5 * pl, v4, 0.0, 0.0);
rgpulse(1.5 " pw, v3, taul, 0.0); ,
rgpulse(pw, v2, 0.0, 0.0);
rgpulse(0.5 * pw, VB, 0.0, rof2);
rcvron(); /"' new for 2.1 */

delay (tau2 - rof2 + 1.25 * pw); /" 1.25 corrects for pulse“!

l’widths‘l

Oak

else

VJ
\

=1
3
V

fprintf(stdout, “ssecho - Using sin
hlv(ct,v1); /"' v1=0 0
hlv(vl,v1); /* vl=0 0
dbl(v1,v3); /* v3=0 0
add(one,oph,v2); /* v2=1 2
add(vl,v2,v2); /" v2=1 2
2
1
1

00
{*1}

g.
\

mod4(v2,v2); /* v2=1
add(oph,v2,vl); I“ v1=0
mod4(vl,v1); I“ v1=0
rgpulse(pl, v1, rofl, 0.0);
rgpulse(pw, v2, taul, rof2);
rcvron();
delay(tau2 - rof2 + pw);
1
status(C);
delay(1.0 I (beta * fb)); /" wait out group delay “I
acquire(np, 1.0 I sw); /* acquire data “I

}

oaoAuN—u?
\“é

wuuuNN—N
Hue-lushnh‘ueu
cranium—4:10.55
s—uctstotsNe-R
NamMU¥NM
uQNaeeuu
Nascrho-‘Otwaa"
mquuuaua
coopaiwowq
...vou~lrbotu~1
\e

112
001
002
341
342
302
232
232

l:
\

LIST OF WCES

p—a
O

10.

11.

12.

13.

14.

REFEIENCB
W. Weyl, Ann. Physik.1_21_, 601 (1863).
C. A. Kraus, J. Am. Chem. Soc. 36, 864 (1914).
J. Jortner, J. Chem. Phys. 10, 839 (1959).
J. Jortner, S. A. Rice and E. G. Wilson, “Metal-Ammonia
Solutions”, (G. Lepoutre and M. I. Sienko, eds.), Benjamin, New
York, 1964.

J. C. Thompson, “Electrons in Liquid Ammonia”, Clarendon
Press, Oxford, 1976.

J. R. Buntaine, M. J. Sienko and P. P. Edwards, J. Phys. Chem. 84,
1230 (1980).

M. H. Cohen and I. Jortner, J. Phys. Chem. 12, 2900 (1975).

U. Even and I. Jortner, ”Electrons in Fluids", (J. Jortner and N. R.
Kestner, eds.), Springer-Verlag, Berlin, 1973.

R. Hensel, "Electrons in Fluids”, (1. Jortner and N. R. Kestner,
eds.), Springer-Verlag, Berlin, 1973.

F. Henscl and E. V. Franck, Ber. Bunsenges. Phys. Chem. 10,
1154 (1966).

F. Hensel, Ber. Bunsenges. Phys. Chem. 12, 619 (1971).

T. V. Ramakrishnan, “The Metallic and the Non—Metallic States
of Matter”, (P. P. Edwards and C. N. R. Rao, eds.), Taylor and
Francis, London, 1985.

N. Mammano and L. V. Coulter, J. Chem. Phys. 41, 1564 (1967).

A. M. Stacy and M. J. Sienko, Inorg. Chem. 2_l, 2294 (1982).

15.

16.

17.

18.

19.

21.

22.

25 .

26.

27.

28.

29.

164

N. Mammano and M. J. Sienko, J. Am. Chem. Soc. 2Q, 6322
(1968).

A. M. Stacy, D. C. Johnson and M. J. Sienko, J. Chem Phys. 16,
4248 (1982).

P. P. Edwards, A. R. Lusis and M. J. Sienko, J. Chem. Phys. 12,,
3103 (1980).

A. M. Stacy, P. P. Edwards and M. J. Sienko, J. Solid State Chem.
41, 63 (1982).

M. K. Faber, O. Fussa’-Rydel, J. B. Skowyra, L. E. H. McMills and J.
L. Dye, J. Am. Chem. Soc. LLL, 5957 (1989). r

J. L. Dye, M. G. DeBacker and V. A. Nicely, J. Am. Chem. Soc. 22,
5226 (1970).

J. L. Dye, M. T. Lok, F. J. Tehan, R. B. Coolen, N. Papadakis, J. M.
Ceraso and M. G. DeBacker, Ber. Bunsenges. Phys. Chem. 15, 659
(197 l ).

M. T. Lok, F. J. Tehan and I. L. Dye, J. Phys. Chem. 16,, 2975
(1972).

J. L. Dye, “Electrons in Fluids”, (1. Jortner and N. R. Kestner, eds.),
Springer-Verlag, Berlin-Heidelberg-NewYork, 1973.

C. J. Pederson, J. Am. Chem. Soc. 82, 7017 (1967).

B. Dietrich, I. M. Lehn and J. P. Sauvage, Tetrahedron Lett. 34,
2885 (1969).

G. P. Pez, I. L. Mador, J. E. Galla, R. K. Crissey and C. E. Forbes, J.
Am. Chem. Soc. 191, 4098 (1985).

J. L. Dye. C. W. Andrews and S. E. Mathews, J. Phys. Chem. 12,
3065 (1975).

J. L. Dye, J. Phys. Chem. 84, 1084 (1980).

I. L. Dye, J. Phys. Chem. 88, 3842 (1984).

30.

31

_ 32.

33.

34.

35.
36.
37.

38.

39.

41.

42.

43.

165

A. L. Wayda and I. L. Dye, J. Chem. Educ. 62, 356 (1985).

. l. L. Dye and M. o. DeBacker, Ann. Rev. Phys. Chem. 38, 271

(1987).

J. L. Dye, J. M. Ceraso, M. T. Lok, B. L. Barnett and F. I. Tehan, J.
Am. Chem. Soc. 26, 608 (1974).

F. J. Tehan, B. L. Barnett and I. L. Dye, J. Am. Chem. Soc. 26,
7203 (1974).

B. VanEck, L. D. Le, D. Issa and J. L. Dye, Inorg. Chem. 21, 1966
(1982).

A. W. Weiss, Phys. Rev. 1_§_§_, 70 (1968).
W. H. E. Schwarz, Chem. Phys. Lett.1_0_, 478 (1971).
J. L. Dye and J. Papaioannou, Rev. Sci. Instrum. 12, 496 (1988).

D. Issa, A. Ellaboudy, R. Janakiraman and I. L. Dye, J. Phys.
Chem. 28,, 3847 (1984).

S. B. Dawes, D. L. Ward, R. H. Huang and J. L. Dye, J. Am. Chem.
Soc. 1.0.8.. 3534 (1986).

R. H. Huang, D. L. Ward, M. E. Kuchenmeister and J. L. Dye, J.
Am. Chem. Soc. 192, 5561 (1987).

R. H. Huang, M. K. Faber, K. I. Moeggenborg. D. L. Ward and I. L.
Dye, Nature 31L, 599 (1988).

J. L. Dye, M. R. Yemen, M. G. DaGue and J. M. Lehn, J. Chem.
Phys. 63, 1665 (1978).

M. G. DaGue, I. S. Landers, H. L. Lewis and I. L. Dye, Chem. Phys.
Lett. 6,6, 169 (1979).

I. L. Dye, M. G. DaGue, M. R. Yemen, I. S. Landers and H. L. Lewis,
J. Phys. Chem. 84, 1096 (1980).

45.

47.

48.

49.

50.
51.
52.
53.
54.
55.
56.

57.

58.

59.

61.

166

L. D. Le, D. Issa, B. VanEck and J. L. Dye, J. Phys. Chem. 85,, 7
(1982).

. . R. R. Dewald and I. L. Dye, J. Phys. Chem. 63, 121 (1964).

A. Ellaboudy, M. L. Tinkham, B. VanEck, I. L. Dye and P. B.
Smith, J. Phys. Chem. 88, 3852 (1984).

N. F. Ramsey, Phys. Rev. 11, 567 ( 1950).

E. R. Andrew, A. Bradbury and R. G. Eades, Nature (London)
1.83, 1802 (1959).

E. R. Andrew, Arch. Sci. (Geneva)1_2_, 103 (1959).

I. J. Lowe, Phys. Rev. Lett. 2, 285 (1959).

D. Ikenberry and T. P. Das, Phys. Rev. A 118, 822 (1965).

C. Deverell, Progr. Nucl. Magn. Reson. Spectrosc. 4, 278 ( 1969).
H. I. Behrens and B Schnabel, Physica 8114, 185 (1984).

A. Ellaboudy and I. L. Dye, J. Mag. Res. 66, 491 (1986).

R. Concepcion and J. L. Dye, J. Am. Chem. Soc. 102, 7203 (1987).

M. L. Tinkham and J. L. Dye, J. Am. Chem. Soc. 19], 6129
(1985).

M. L. Tinkham, A. Ellaboudy, I. L. Dye and P. B. Smith, J. Phys.
Chem. 20, 14 (1986).

A. Ellaboudy, 1. L. Dye and P. 13. Smith, J. Am. Chem. Soc. 105.
6490 (1983)

J. L. Dye and A. Ellaboudy, Chem. Br. 20, 210 (1984).

J. L. Dye, C. W. Andrews and I. M. Ceraso, J. Phys. Chem. 12,
3076 (1975).

62.

63.

65.

67.

68.

69.
70.
71.

72.

73.

74.

75.
76.

77.

78.

167

D. M. Holton, P. P. Edwards, D. C. Johnson, C. I. Page, W.
McFarlane and B. Woods, J. Chem. Soc., Chem. Commun., 740

(1984).

O. Fussa’, S. M. Kauzlarich, I. L. Dye and B. K. Teo, J. Am. Chem.
Soc.1_0_1: 3727 (1985).

A. C. Kunwar, G. L. Turner and E. Oldfield, J. Mag. Res. Q2, 124
(1986).

L. E. H. McMills, Ph.D. Dissertation, Michigan State University,
1989.

J. Kim, Ph.D. Dissertation, Michigan State University, 1989.
K. J. Moeggenborg, J. L. Eglin and I. L. Dye, unpublished results.

R. G. Patel and M. M. Caudhri, Thermochimica Acta 25,, 247
(1978).

T. Ozawa, Bull. Chem. Soc. Jpn. 38, 1881 (1965).
H. E. Kissinger, J. Res. Nat. Bur. Stand. 11, 217 (1956).
A. Maercker, Angew. Chem. Int. Ed. Engl. 2.6,, 972 (1987).

M. E. Kuchenmeister and I. L. Dye, J. Am. Chem. Soc 111, 935
(1989).

H. K. Frensdorff, J. Am. Chem. Soc. 23, 600 (1971).

M. E. Kuchenmeister, Ph.D. Dissertation, Michigan State
University, 1989.

R. Huang, J. L. Eglin, D. Ward and I. L. Dye, unpublished results.

8. B. Dawes, A. S. Ellaboudy and I. L. Dye, J. Am. Chem. Soc. 1_Q2,
3508 (1987).

S. B. Dawes, A. Ellaboudy and I. L. Dye, unpublished results.

8. B. Dawes, D. L. Ward, 0. Fussa’-Rydel, R. Huang and J. L. Dye,
Inorg. Chem. 23,, 2232 (1989).

79.

81.
82.
83.

85.

86.
87.
88.
89.

91.

93.

94.

95.

168

P. P. Edwards, Adv. Inorg. Chem. Radiochem. 25, 135 (1982).

. - J. P. Lelieur and P. J. Rigny, J. Chem. Phys. 22, 1148 (1973).

I

W. D. Knight, Phys. Rev. 16, 1259 (1944).
O. Yamamotot and M. Yanagisawa, Anal. Chem. 42, 1463 (1970).
A. VanGeet, Anal. Chem. 42, 679 (1970).

D. L. Ward, Rui H. Huang, M. E. Kuchenmeister and I. L. Dye,
unpublished results.

I. S. Landers, I. L. Dye, A. Stacy and M. J. Sienko, J. Phys. Chem.
85, 1096 (1981).

D. Issa and I. L. Dye, J. Am. Chem. Soc. 194, 3781 (1982).

S. B. Dawes, Ph.D. Dissertation, Michigan State University, 1986.
I. H. Van Vleck, Phys. Rev. B, 1168 (1948).

A Beckmann, K. D. Boken and D. Elke, Z. Phys. 212, 173 (1974).
C. J. Pederson, J. Am. Chem. Soc. 20, 3299 (1968).

J. D. Lamb, R. M. Izatt, C. S. Swain and J. J. Christensen, J. Am.
Chem. Soc. 1112, 475 (1980).

G. P. Pez, I. L. Mador, J. E. Galle, R. K. Crissey, and C. E. Forbes, J.
Am. Chem. Soc. 101, 4098 (1985).

A. G. Barrett, C. R. Godfrey, D. M. Hollinshead, P. A. Prokopiou, D.
H. Barton, R. B. Boar, L. Joukhadar, J. F. McGhie and S. C. Misra,
JCS. Perkin I, 1501 (1981).

R. M. Izatt, I. S. Bradshaw, S. A. Nielsen, I. D. Lamb, 1. J.
Christensen and D. Sen, Chem. Rev. 8.5, 271 (1985).

R. H. Huang, D. Ward, and I. L. Dye, unpublished results.

95.

96.

97.

98.
99.

100.

101.
102.
103.

104.

105.

106.

107.

108.

109.

169

R. H. Huang, D. Ward, and J. L. Dye, unpublished results.

J. D. Lamb, R. M. Izatt, J. J. Christensen and D. J. Eatough,
“Coordination Chemistry of Macrocyclic Compounds”, (G. A.
Melson, ed.), Plenum Press, New York, 1979.

J. J. Christensen, D. J. Eatough and R. M. Izatt, Chem. Rev. 14,
351 (1974).

J. M. Lehn and J. P. Sauvage, J. Am. Chem. Soc. 21, 6700 (1975).
M. Micheloni, J. Coord. Chem. 18, 3 (1988).

K. J. Moeggenborg, Ph.D. Dissertation, Michigan State University,

1990.

P. Murray and J. White, J. Trans. Br. Ceram. Soc. 54, 204 (1955).
H. E. Kissinger, Anal. Chem. 22, 217 (1957).
R. A. Logan, R. E. Cote and P. Kusch, Phys. Rev. 85, 280 (1952).

M. R. Yemen, Ph.D. Dissertation, Michigan State University,
1982.

D. L. Ward, R. H. Huang and J. L. Dye, Acta Cryst. 5144, 1374
(1988).

D. T. Cromer and J. T. Waber, “International Tables for X-Ray
Crystallography”, Vol. IV, Table 2.2B, Kluwer Academic
Publisher, Dordrecht, 1974.

D. T. Cromer, “International Tables for X-Ray Crystallography”,
Vol. IV, Table 2.3.1, Kluwer Academic Publisher, Dordrecht,
1974.

J. L. Dye, “Valency and Charge Distribution in Alkalidc and
Electride Salts”, The Robert A. Welch Foundation Conference on
Chemical Research, VALENCY XXXII, 1988.

M. Kuchenmeister, A. Ellaboudy, C. Bender, J. Babcock, J. Kim,
Dae—Ho Shin and J. L. Dye, unpublished results.

110.

111.

112.

113.

114.

115.
116.

117.

118.

119.

120.

121.

122.

123.

124.

170

F. A. Cotton and G. Wilkinson, “Advanced Inorganic Chemistry A
Comprehensive Text”, 4th Edition, John Wiley and Sons, New
York, 1980.

L. E. H. McMills and J. L. Dye, unpublished results.

R. H. Huang, D. Ward and J. L. Dye, unpublished results.

A. Bencini, A. Bianchi, A. Borselli, M. Ciampolini, E. Garcia-
Espana, P. Dapporto, M. Micheloni, P. Paoli, J. Ramirez, and B.
Valtancoli, Inorg. Chem. 28, 4279 (1989).

D. Freude, T. A. Carpenter, J. Haase, J. Klinhowski and G.
Ronikier, Chem. Phys. Lett. 1_1_2, 365 (1985).

J. L. Dye, Prog. in Inorg. Chem. 32, 327 (1984).
O. Fussa’, Ph.D. Dissertation, Michigan State University, 1986.

C. N. R. Rao and P. P. Edwards, Proc. Indian Acad. Sci. 26, 473
(1986).

N. F. Mott, “Metal-Insulator Transitions”, Taylor & Francis,
London, 1974.

I. Weberman, J. Jortner and M. H. Cohen, Phys. Rev. B. 1_1, 2885
(1975).

E. Abrahams, P. W. Anderson, D. C. Licciardell and T. V.
Ramakrishman, Phys. Rev. Lett. 42, 673 (1979).

C. Kittel, “Introdution to Solid State Physics”, 2nd Edition, Wiley,
New York, 1963.

D. H. Martin, “Magnetism in Solids”, The M.I.T. Press, Cambridge,
1967.

D. Issa, Ph.D. Dissertation, Michigan State University, 1982.

M. Kuchenmeister and J. L. Dye, unpublished results.

126
127

128

'129

130.

131

171
. P. P. Edwards and M. J. Sienko, J. Chem. Edlm. 691 (1983).

. K. F. Herzfeld, Phys. Rev. 22, 701 (1927).

.- T. A. Patterson, H. Hotop, A. Kasdan, D. W. Norcross and W. C.
Lineberger, Phys. Rev. Lett. 22, 189 (1974).

I

. N. Aschcroft and N. Mermin, “Solid State Physics”, Holt, Rinehart
and Winston, New York, 1976.

J. L. Dye and V. A. Nicely, J. Chem. Ed. 48, 443 (1971).

. R. H. Huang, D. L. Ward and J. L. Dye, J. Am. Chem. Soc. 111.
5707 (1989)

"‘11111111111“