INVESTIGATlON OF SOLUTIONS AND
CRYSTALLINE SOLIDS WHICH CONTAIN
ALKAU CRYPTATE CATIONS
AND ALKALI ANIONS

Dissertafion for the Degree at m. D.
MICHIGAN STATE UNWERSITY
FREDERECK JOHN TEHAN
1973

 

 

This is to certify that the

thesis entitled
Investigation of Solutions and crystalline Solids

Which Contain Alkali Cryptate Cations and
Alkali Anions
presented by

Frederick John Tehan

has been accepted towards fulfillment
of the requirements for

_Eli._]l.__ degree in ihemisirx.

   

Major professor

Date Dec. 1+, 1975

0-7639

    

:5.— v
BINDlNG BY :

HOAG & SUNS'
800K BINDERY INC.

le. rm! amozas
' ulcumn

l

l *

 

ABSTRACT

INVESTIGATION OF SOLUTIONS AND CRYSTALLINE SOLIDS
WHICH CONTAIN ALKALI CRYPTATE CATIONS AND
ALKALI ANIONS

BY

Frederick John Tehan

By the use of cyclic polyethers of the "crown" and
"cryptate" classes to complex the alkali cations, alkali
metals were dissolved in secondary amines and in straight
and branched chain ethers. In the absence of the complexing
polyether, the metals are insoluble in these solvents. Pro-
nounced solubility enhancement occurred in primary mono-
and di-amines, tetrahydrofuran and di-ethers in which the
metals are only very slightly soluble without the complexing

agent. The optical spectra of Na-, K-, and, in most cases,

e solv' were measured for solutions in ethylamine, diethyl-

amine, di-nfpropylamine, 1,2—propanediamine, hexamethyl-
phosphoric triamide, diethyl ether, di-isopropyl ether,
tetrahydrofuran, dimethoxyethane and diglyme. In most cases
the peak positions were measured as functions of temperature.
For each of the three species the wavenumber of maximum
absorbance decreased linearly with an increase in tempera-
ture. Linear correlations were found among peak positions

of e solv Na , and K ~in various solvents at a
0

Frederick John Tehan
given temperature. The spectra of Na- and K- obey the
criterion for a ctts transition that the position of the
peak be a linear function of the temperature coefficient.
However, no satisfactory correlation between the peak posi-
tions of Na- and K_ and solvent properties has been found.

The spectrum of e- has now been determined by

solv
metal solution studies, as well as by flash photolysis and
pulse radiolysis techniques for the solvents ammonia,

ethylenediamine, diethylamine, diethyl ether, tetrahydro-
furan, dimethoxyethane and diglyme. The agreement between
the peak positions in all cases is good. However, in some
cases the band is substantially broader in metal solutions
than in those studied by radiolysis. This is probably due

to the formation of an alkali metal "monomer" which has

been shown in certain instances to absorb between the e solv

band and the M— bands.

The relative intensities of the single ESR line of the
solvated electron in Na, K, Rb, and Cs-ammonia solutions
were measured as a function of temperature, in order to
determine the effect of the cation on the spin-pairing
phenomenon. All samples were concentrically surrounded by
a DPPH spin standard solution. If the nature of the cation
were important, one would expect a variation of the amount
of spin-pairing in solution with the various cations.
However, results show that the spin-pairing process is
relatively insensitive to the nature of the cation in the

range studied (0.02 to 0.1M and -60 to 0°).

Frederick John Tehan

A crystalline salt which contains the alkali anion has
been made. When a saturated solution of sodium in ethyl-
amine in the presence of a cation complexing ether (2,2,2—
crypt) is slowly cooled, thin hexagonal plates of stoichio—
metry Na2C18H3606N2 are formed. The crystals exhibit a
bright gold color at 770K which changes reversibly to
bronze color at room temperature. The X-ray structure was
also determined and verified that one sodium is trapped
inside the complexing agent while the outside sodium is far
removed from any atoms. Thermal factors are much larger
for the sodium outside the crypt which may indicate that the
charge density around this sodium is more diffuse. The struc-
ture was compared with the X-ray structure of

+ - 1
[Na C18H3606N2]I o

The analogy of the iodide ion with a
sodium anion is appropriate in view of the similarity be-
tween the structures. The interatomic distances also favor
the assignment of the outside sodium to be a sodium anion.

The sodium species outside the crypt is over 4.8 X
from the nearest atom (carbon) and is well over 5 X from
the closest oxygen and nitrogen atoms.

The crystallographic R factor, or residual index, was
refined to a value of 0.0850, based on the Scattering factor
tables of atomic sodium.

An empirical scattering factor curve for the sodium
anion, and the known function for Na+, when used in place

of the atomic scattering tables resulted in a lowering of

the R value.

Frederick John Tehan

In View of the results of the structure, there is no
doubt that the sodium species inside the crypt is a sodium
cation and that the sodium species outside the crypt is a

sodium anion.

REFERENCE

1. D. Moras and R. Weiss, Acta. Cryst., 829, 396 (1973).

 

INVESTIGATION OF SOLUTIONS AND CRYSTALLINE SOLIDS
WHICH CONTAIN ALKALI CRYPTATE CATIONS AND
ALKALI ANIONS

BY

Frederick John Tehan

A DISSERTATION

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

(DOCTOR OF PHILOSOPHY

Department of Chemistry

1973

To My Parents

ii

ACKNOWLEDGMENTS

The author wishes to express his deepest appreciation
to Professor James L. Dye for his encouragement, assistance
and guidance throughout the course of this work.

He would also like to thank Dr. M. T. Lok, a special
friend, for the many valuable suggestions given, especially
in the collaborative work. Special thanks go to Dr. B. L.
Barnett for his cooperation in the determination of an X-
ray structure. Thanks go to Dr. G. Leroi, for serving as
second reader, and to Drs. C. Brubaker and A. Tulinsky for
their suggestions and for the use of their equipment.

The author also wishes to acknowledge the help of J.
M. Ceraso, R. B. Coolen, N. Papadakis, L. D. Long, E. Mei,
and M. DeBacker. The cooperation of the MSU glass shop, and
financial support from the Atomic Energy Commission is also
acknowledged.

Last, but not least, the author wishes to thank his
parents, Mr. and Mrs. Louis B. Tehan, for their unending

encouragement.

iii

TABLE OF CONTENTS

Page
1. INTRODUCTION . . . . . . . . . . . . . . . . . 1

1.1. Electrons in Fluids . . . . . . . . . . .
1.1.1. Quasi-free State . . . . . . . . .

1.1.2. Localization State . . . . . . . .

CDNNH

1.2. PrOperties of Metal-Ammonia Solutions . .
1.3. Models . . . . . . . . . . . . . . . . . 10
1.4. Properties of Amine and Ether Solutions 13
1.5. General Summary . . . . . . . . . . . . . 21
1.6. Solid-State Studies . . . . . . . . . . . 22

1.7. Objectives of Dissertation Research . . . 23
II. EXPERIMENTAL . . . . . . . . . . . . . . . . . 26

11.1. General Techniques . . . . . . . . . . . 26
11.1.1. Glassware Cleaning . . . . . . . . 26
11.1.2. Vacuum Techniques . . . . . . . . 26

11.2. Metal Purification . . . . . . . . . . . 27

11.2.1. Storage of Alkali Metals in Small
Quantities . . . . . . . . . . . . 27

11.2.2. Premeasured Amounts of the Alkali
Metals . . . . . . . . . . . . . . 29

11.3. Premeasured Amounts of Materials . . . . 32
110301. General 0 O o o o o o o o o o o o 32

11.3.2. Sub-milligram Amounts of Alkali
Metals . . . . . . . . . . . . . . 33

11.4. Purification of Solvents . . . . . . . . 34

11.5. Purification of the Complexing Agents . . 35

iv

TABLE OF CONTENTS (Cont.)

Page

11.6. Synthesis of 2,2,2—Crypt . . . . . . . 36
11.6.1. Synthesis of 1,8-Diamino-3,6-
dioxaoctate (hereafter called

diamine) . . . . . . . . . . . . 38

11.6.2. Preparation of Triglycolic Chloride
(hereafter called diacidchloride) 40

11.6.3. First Cyclization (preparation of

the dilactam) . . . . . . . . . 42

11.6.4. First Reduction (preparation of
the monocyclic diamine) . . . . 43

11.6.5. Second Cyclization (preparation
of the bicyclic dilactam) . . . 44

11.6.6. Second Reduction (preparation of
bicyclic diamine -- 2,2,2—crypt) 45
III. SPECTRA . . . . . . . . . . . . . . . . . . 47
111.1. Introduction . . . . . . . . . . . . . 47
111.2. Experimental . . . . . . . . . . . . . 50
111.3. Instrumentation . . . . . . . . . . . . 50
111.4. Data Handling . . . . . . . . . . . . . 51
111.5. Results and Discussion . . . . . . . . 51
IV. ESR STUDIES OF METAL-AMMONIA SOLUTIONS . . . 81
1V.1. Introduction . . . . . . . . . . . . . 81
1V.1.1. Historical . . . . . . . . . . . 81
1V.1.2. Preliminary Studies . . . . . . 84
1V.1.3. Statement of the Problem . . . . 84
1V.2. Sample Preparation . . . . . . . . . . 85
1V.2.1. Preparation of the Spin Standard 85
1V.2.2. Sample Preparation . . . . . . . 86
1V.3. Instrumentation . . . . . . . . . . . . 86
1V.4. Results and Discussion . . . . . . . . 89

1V.4.1. Line-shape and Saturation Problems 89
1V.4.1.1. Line-shape Problems . . . 89
1V.4.1.2. Saturation . . . . . . . . 89

TABLE OF CONTENTS (Cont.)

1V.4.2. Data Treatment . . . . . .
IV.4.2.1. Line—shape Analysis
IV.4.2.2. Saturation Studies .

1V.4.3. Results y§_Cation and Temperature

V. SOLID-STATE STUDIES . . . . . . . . .

v.1.

v.2.
V03 0

VI. SUMMARY AND SUGGESTIONS FOR FURTHER WORK

V1.1.
V1.2.
V1.3.
V1.4.
V1.5.

Introduction . . . . . . . . . .
V.1.1. Historical . . . . . . . .

V.1.2. Use of Complexing Agents .

V.1.3. Preparation of Na2C18H3303N2

V.1.4. Appearance and Stability .
V.1.5. Stoichiometry . . . . . .
V.1.6. Conductivity . . . . . . .

Single Crystal Growing . . . . .

X-ray StIUCture o o o o o o o o o

V.3.1. Isolation of a Single Crystal
V.3.2. Background to X-ray Diffraction

V.3.3. Space Group Determination
V.3.4. Data Collection . . . . .

V.3.5. Fundamentals of Crystal Structure

Analysis . . . . . . . . .

V.3.6. Solution of the Structure
V.3.7. Refinement of the Structure

V.3.8. Results . . . . . . . . .

V.3.9. Discussion . . . . . . . .

V.3.9.1. General . . . . . . .

V.3.9.2. Proof for the Sodium Anion

Amine and Ether Solvents . . . .
Metal-Ammonia Solutions . . . . .
Solid-State Studies . . . . . . .
Proposed Model . . . . . . . . .
Suggestions for Further Work . .

vi

Page
90
90
94
98

111

111
111
111
112
113
113
114
114
115
115
118
119
120

122
123
124
126
137
137
141

146

146
146
147
147
148

TABLE OF CONTENTS (Cont.)

APPENDICES

A.

THE ISOLATION OF SMALL AMOUNTS OF MATERIALS IN

THIN-WALLIED VESSEIS o o o o o o o o o o o o

1.1. Introduction . . . . . . . . . . . . .
1.2. Heat-Shrink Tubing and Bulbs . . . . .
1.3. Bulb Preparation .. . . . . . . . . . .
1.4. Bulb Filling . . . . . . . . . . . . .
1.4.1. Solids . . . . . . . . . . . . .
1.4.2. Pure Liquids . . . . . . . . . .
1.4.3. Solutions and Solutes . . . . .
1.5. Utilization . . . . . . . . . . . . . .
1.6. Summary . . . . . . . . . . . . . . . .
THE ISOLATION OF SUB-MILLIGRAM QUANTITIES OF
ALKALI METAL . . . . . . . . . . . . . . . .
2.1. Introduction . . . . . . . . . . . . .
2.2. Technique . . . . . . . . . . . . . . .
2.2.1. Capillary Diameter Measurement .
2.2.2. Bulb Preparation . . . . . . . .
2.2.3. Sample Filling . . . . . . . .
2.2.4. Sample Length Measurement . . .
2.2.5. Sealing the Bulb . . . . . . . .
2.3. Calibration Results . . . . . . . . . .
2.4. Conclusion . . . . . . . . . . . . . .

NON-LINEAR LEAST SQUARES PROGRAM AND MODIFI-
CAT IONS . C O O O C C O O O O C . O C O I C

3.1. Introduction . . . . . . . . . . . . .

3.2. Partial Pairwise Correlation Coefficients

3.3. More Flexible Use of Subroutine EQN . .
3.4. Generation of Data . . . . . . . . . .
3.5. Curvefit Procedure . . . . . . . . . .

3.6. Flow Diagram . . . . . . . . . . . . .

LIST OF REFERENCES . O O C O O O O O O C O C O O O

Page

150

150
151
154
156
158
159
160
160
162

163

163
164
164
165
165
168
168
170
170

173

173
174
175
176
176
176

179

10.

11.
12.
13.

14.

LIST OF TABLES

Stoichiometry of species in metal-ammonia
solutions according to various models . . .

Tabulation of data of K- in diglyme . . . .
Solvents in which K- has been observed . . .

Peak positions and temperature coefficients
for Na , K , and esolv in various solvents .
Peak positions and temperature coefficients
for I' in various solvents . . . . . . . . .

Corrected area ratio of the solvated electron
ESR signal in 0.02M Na, K. Rb, and Cs-ammonia
solutions to that of the center DPPH radical
peak 0 0 O O O O O O O O O O O O O O O O O O

Corrected area ratio of the solvated electron
ESR signal in 0.04M Na, K, Rb, and Cs-ammonia
solutions to that of the center DPPH radical
peak 0 O O O O O O O O O O O O O O O O O O O

Corrected area ratio of the solvated electron
ESR signal in 0.06M Na, K, Rb, and Cs-ammonia
solutions to that of the center DPPH radical
peak 0 O O I O C O O O O O O O O O O O O O O

Corrected area ratio of the solvated electron
ESR signal in 0.10M Na, K, Rb, and Cs-ammonia
solutions to that of the center DPPH radical
pe ak O O O O O I I O C O O I O O O O O O O 0
Area ratio of the solvated electron in liquid
ammonia for Na, Rb, and Cs to that of K at
the same concentrations and temperatures . .
Crystal lattice parameters . . . . . . . .
List of structure factors . . . . . . . . .

List of structure factors (cont.) . . . . .

Positions and thermal parameters of the
independent atoms in the unit cell . . . . .

viii

Page

14
52

53

59

63

99

100

101

102

107
127
128
129

130

LIST OF TABLES (Cont.)

TABLE

15.

16.

17.
18.
19.

20.

21.

22.

23.

24.

Centers, symmetry operations and the positions
of some atoms in the lattice . . . . . . . . .

Bond distances and estimated standard devia-
tions 0 O O O O O O O O O O O O O O C O O O 0

Bond angles and estimated standard deviations.
Some interesting interatomic distances . . . .
Some interesting interatomic angles . . . . .

Comparison of some interatomic distances with
the+interatomic distances obtained from
[Na C18H3606N211 o o o o o o o o o o u o o o

Conformational angles of this structure com—
pared with the conformational angles in the
iodide structure

Scattering factor curves for Na+, Nao, and

"Na-" 0 O O O O O O O O O O O O O O O O O O 0

Summary of crystallographic significance
tests 0 O O O O O I O O O I O O I O O O O O 0

Comparison of observed and calculated
weights of potassium metal . . . . . . . . . .

ix

Page

131

134
134
135

135

136

136

145

145a

172

LIST OF FIGURES

FIGURE
1. Qualitative potential well diagrams for polar
and non-polar solvents . . . . . . . . . . .
2. Comparison of the solvated electron spectrum
in ethylenediamine by metal solutions and
by pulse radiolysis . . . . . . . . . . . .
3. Dicyclohexyl-lB-crown-6 and 2,2,2-crypt . .
4. Apparatus for the preparation of storage tubes
for sodium or potassium metal . . . . . . .
5. Vacuum tight glass vessel for the storage of
cesium and rubidium metal . . . . . . . . .
6. Apparatus for the isolation of known amounts
Of metal 0 O O O O O C O O O I O O O O O O O
7. Spectra at 25° of e- , Cs-, K-, and Na_ in
solv
tetrahydrofuran in the presence of crown or
2'2'2-Crypt o o o o o o o o o o o o o o o o
8. Temperature dependence of the position of the
band maximum for Na" in various solvents . .
9. Temperature dependence of the position of the
band maximum for K' in various solvents . .
10. Temperature dependence of the position of the
band maximum for e' in various solvents .
solv
11. The relation between the Na- peak position at
25° and its temperature coefficient . . . .
12. Temperature dependence of the position of the
band maximum for 1' in various solvents . .
13. The relation between the peak position of 1-

and of Na' in various solvents . . . . . . .

Page

16

24

28

30

31

49

56

57

58

60

62

64

LIST OF FIGURES (Cont.)

FIGURE
14. The composite of gaussian curves which fit the
experimental absorption of I in various
SOlventS O O C O O O O O O C C O I O O O O O
15. Relation between the peak position of Na- and
of K“ in various solvents at 25° . . . . . .
16. The spectra of the solvated electron at 25°
in various solvents . . . . . . . . . . . .
17. The spectrum of the solvated electron in 1,2-
propanediamine at various temperatures . . .
18. The relation between the peak position of K-
and of e‘ in various solvents at 25° . .
solv
19. The relation between the peak position of eSolv
obtained by dissolving metals and by pulse
radio lysis O C O O O O O O O O O O O I O O O
20. Comparison of the band shape of e-solv at 25°
in diethyl ether and tetrahydrofuran obtained
by dissolving metals and by pulse radiolysis
21. Pulse radiolysis studies of pure tetrahydro-
furan, tetrahydrofuran with Na+, and tetra-
hydrofuran with Na+ and 2,2,2-crypt . . . .
22. Comparison of the absorption spectrum of the
solvated electron produced by metal solutions
in ND3 and by pulse radiolysis . . . . . . .
23. The spectrum of the solvated electron in 1,2-
propanediamine at various temperatures obtained
by pulse radiolysis and by metal solution .
24. Concentration of paramagnetic species in
Na-NHs SOlUtiOI‘lS at -33 o o o o o o o o o o
25. Reference Dewar with DPPH standard and metal
sample for ESR study . . . . . . . . . . . .
26. Changes in the solvated electron spectrum from
lorentzian to gaussian shape . . . . . . . .
27. Percent lorentzian character in the solvated

electron band shape in Na, K, Rb, and Cs
SOlutionS O O O O O O I O O O I O O O O O 0

xi

Page

65

67

68

7O

71

73

74

76

78

80

83

87

92

93

LIST or FIGURES (Cont.)

FIGURE Page

28. Variation of the linewidth of DPPH and of the
solvated electron in cesium solutions with the
square root of the microwave power . . . . . . 95

29. Plot of the peak-to-peak height for DPPH and
for the solvated electron in cesium solutions
versus the square root of the microwave power. 96

30. Plot of the square of the peak-to—peak width of
the solvated electron signal versus the square
of the amplitude of the DPPH radical signal . 97

31. Plot of the logarithm of the area ratio of the
solvated electron signal in cesium solution to
that of the DPPH spin standard absorption
versus l/T . . . . . . . . . . . . . . . . . . 104

32. Plot of the logarithm of the area ratio of the
solvated electron signal in sodium, potassium,
and rubidium solutions to that of the DPPH spin
standard absorption versus 1/T . . . . . . . . 105

33. Plot of area ratio of sodium, rubidium and cesium
to that of potassium versus temperature . . . 106

34. Comparison of the relative area of our potassium
data with the data of Hutchison and Pastor as
a function of l/T . . . . . . . . . . . . . . 108

35. Plot of the spin concentration versus the concen-
tration of sodium and potassium solutions . . 110

36. Photomicrograph of Na2C18H3606N2 . . . . . . . 116

37. Unit cells in the rhombohedral and hexagonal
lattices O O O O O O O O O O O C O O O O I O O 121

38. Labelling scheme of atoms in the hexagonal
lattice O O O O C I O O O O C O O O O O O O O 132

39. Model showing the sodium-cryptate and the six
nearest outside sodiums . . . . . . . . . . . 138

40. Stereoscopic view of the complexing agent with
the inner sodium . . . . . . . . . . . . . . . 139

41. Stereoscopic view of the sodium-cryptate with
the six nearest outside sodiums . . . . . . . 140

xii

LIST OF FIGURES (Cont.)

FIGURE Page

42. Fourier map of the electron density in the
unit Cell 0 O O C O O I O O O O O O O O O O 142

43. Use of heat-shrink tubing to introduce metal
samples into vessels without exposure to the
atmosphere 0 O O O O O O O C O O O O O O O O 152

44. Manifold used to seal small amounts of
materials into breakable bulbs . . . . . . . 153

45. First steps for the preparation of the
breakable bulbs O C C O O O C O O C O O O O 155

46. Recommended storage stages in the preparation
Of bu1bs O I O O O O O O C O O O O O C O O O 157

47. Vessel containing several bulbs for the
introduction of materials . . . . . . . . . 161

48. Preparation of bulbs for the isolation of
small known amounts of alkali metal . . . . 166

49. Various stages of driving the metal into the
bulbs O O O O O O O I O O O O O O O O O O O 167

50. Measurement of the alkali metal in the glass
vessel with a vernier . . . . . . . . . . . 169

51. Calibration curve for the atomic absorption
results . . . . . . . . . . . . . . . . . . 171

52. Flow diagram for K1NFIT2 . . . . . . . . . 177

xiii

I. INTRODUCTION

Alkali metals were first noted to dissolve in liquid
ammonia in 1863.1 For over a century, these solutions have
been the object of numerous scientific investigations. A
great number of models have been proposed to account for
the properties of such solutions. However, to this date,
no one model can explain all the characteristics which have
been observed. As new measurements and observations were
made, models were revised accordingly. Historical surveys
have been given elsewherezu3 It is not our purpose to
describe the historical development, but rather to first
discuss the general pr0perties of electrons in fluids, then
to incorporate the prOperties of alkali metal solutions,
and finally to briefly summarize the models which have been

proposed to account for these phenomena.

I.1. Electrons in Fluids

 

When an excess electron is introduced into a polar or
non-polar fluid, the nature of the energetically stable
state of the electron is not obvious because of the complex-
ity of the system. The general description given here
follows that of Jortner.4 The ground state energy may be

expressed as the sum of two contributions:

1

2
(1) Electronic Energy: Be:
(2) Medium Rearrangement Energy: EM;

such that the total energy, B may be expressed as

tl

There are two limiting cases which can occur in solution.

I.1.1. Quasi-free State

i

The quasi-free electron state is characterized by high
electron mobility. The electron may be described by a
plane wave which is scattered by the molecules of the
fluid. Under these circumstances, EM = 0 since the
liquid structure is not perturbed by thegxesence of the ex-
cess electron. The electronic energy is denoted by Vo(=Ee)
and is determined by a balance of, (1) short range repul-
sions, and (2) long-range polarization interactions. The
total ground state energy of the quasi-free electron state,

Et(=V0) corresponds to the bottom of the conduction band in

the liquid relative to the vacuum level.

I.1.2. Localization Case

 

The localized excess electron state is characterized
by low mobility and requires that the wavefunction approach
zero at large distances from the localization center. In

this case, the medium rearrangement energy, E must be

Ml
greater than zero since it requires energy to form a center.

Often, both E8 and EM are considered to be functions of

3

a distance parameter, R, sometimes Specifically called a

cavity radius. Then,

Et(R)(localized) = Ee(R) + EM(R).

BE (R)
The optimal radius, R0 is obtained by setting -7§%——-

equal to zero. R0 is then the value of R which corre-
sponds to the lowest total energy. Since the electronic
energy, Ee can be either positive or negative and EM(R0),
the medium rearrangement energy, is positive, the resulting
total energy can be either positive or negative in sign.

One must then compare the total energy, Et(Ro) with
V0 (the energy of the quasi-free state) to find out which is
the energetically stable state of the system. If Et(Ro)
is less than V0, then localization will occur.

For a localized state, the electronic energy is the sum
of (1) kinetic energy, (2) long-range attractive polariza-
tions and, (3) short-range attractive and repulsive inter-
actions. The medium rearrangement energy is dependent on
the type of fluid involved.

For a non-polar fluid, EM consists of:

(1) Surface work required to form the cavity. This
is sometimes estimated from the surface tension
and the surface area of the cavity.

(2) The pressure-volume work required to form the
cavity.

In the case of a polar fluid, the situation becomes

much more complicated. In addition to the work terms

4
noted above, one must also consider the following:

(1) The long-range polarization energy--the energy
required to form the potential well by alignment
of the permanent molecular dipoles.

(2) The dipolar repulsions--the energy required to
overcome the repulsions of the aligned dipoles.

(3) The short-range repulsion energy--the energy re-
quired to overcome repulsions between the re-
oriented solvent molecules on the boundary of the
localization site.

(4) The energy required to break hydrogen bonds.

The qualitative features of the potential wells which are
formed in localized electron systems are shown in Figure 1,
for a non-polar and a polar solvent respectively. In non-
polar systems, because of the absence of long-range inter-
actions and the resulting "square well“ potential, the
ground state energy usually lies high in the well, while in
polar solvents, the ground state energy can lie low in the
well and thus permit several bound states to lie below the
continuum.

It is difficult to predict the nature of the electron
in different solvents. For example, it is now known that
although argon, krypton, xenon, and solid helium form the
quasi-free state, the localized state exists in liquid
helium, as well as in hydrogen, deuterated hydrogen, and
some hydrocarbon solvents. Because of large polarization

effects, the localized state seems to be the rule in polar

solvents.

.mucw>H0m HmHomIcoc
paw HmHom How mEmHmmwo HHmB HMflucmuom m>HumufiHm50 .H ousmam

 

 

 

 

 

 

 

 

\ /, ILL

hwaom . pmaom soz

 

6
It is possible to use semi-empirical methods and continu-
um models to estimate the energies involved and several such
calculations have been performed.5’7 The most successful cav-

ity calculations involve the Landau8 potential which gives:

2
V(r): —f—e{-—(bL-El) for r<R
' op 3
while
2 _
V(r) = - §-(53F- 5L) for r > R;
op S
where Ds = static dielectric constant, DOp = optical dielec-

tric constant, R = cavity radius, and r = distance. Several
midifications have also been made“!9 to account for terms

not included in Landau's formulation.
1.2. Properties of MetaleAmmonia Solutions

Experimentally, excess electrons can be introduced into
solvents by two methods:

(1) Pulse radiolysis and flash photolysis techniques,

(2) Alkali metal dissolution.
The former has only been recently deve10ped while the latter
has been studied for over one hundred years.1 Nevertheless,
both techniques are used today and each yields valuable in-
formation. Of all the solvents which dissolve metals,
ammonia has been the most extensively studied. This is
probably because alkali metals dissolve to a much greater
extent in ammonia than in any other known solvent. PrOper-
ties of concentrated metal-ammonia solutions are vastly

different from the dilute solution properties. The following

7
discussion is limited to the properties and models of dilute
solutions.
The blue color produced when alkali metals are dis-
solved in ammonia was first noted by Weyl.1 Based upon
previous work, Weyl called the species responsible for this

color "metal ammoniums". Kraus 10 was the first to propose

 

+ -. .
the equilibrium, M > M + e in such solutions.

<

It is known that amide formation is thermodynamically favored

in these solutions according to:

 

1
> W2 + — H2 0

M +NH3 2

<

However, this reaction can be effectively eliminated by
avoiding all catalyzing agents. To prevent decomposition,
extensive glassware cleaning procedures have been utilized.
If great care is taken, solutions are stable for weeks at
-80°, without visible decomposition. This reaction has
also been shown to be reversible in solutions under high
pressures of hydrogen gas.11r12

In a recent review article, Dye13 has conveniently
classified the properties of metal—ammonia solutions as
follows:

(1) Null Results:

These are the properties which do not change over
the entire dilute solution (< 0.05M) concentration
range. The most important result in this classifica-
tion is that provided by the optical Spectrum. Both

the band shape and the extinction coefficient are

8
independent of concentration while the absorption maxi-
mum is only slightly concentration dependent.“-16
The absorption band is a broad, asymmetric, structure-
less band with a long tail on the high energy side.

The results of Gunn and Green17‘19 indicate that
the large partial molar volume of the solute is inde-
pendent of concentration.

Electron relaxation, which is dominated by nitro—
gen nuclei coupling, is also independent of concentra-
tion as shown by solvent n.m.r.2° and e.p.r. linewidths

21'23 studies.
(2) Cation-Electron Interactions

E.P.R. studies on solutions of europium24r25 in
liquid ammonia indicate that the interaction between
the solvated electron and the cation is weak, since
separate patterns are observed for the electrons on
europium and the solvated electron.

Transference number26 and conductancez":28 measure-
ments demonstrate a decrease in the cationic conduc—
tance with increasing concentration. This is typical
of any salt in ammonia and can be explained by ion-
pair formation.

Cation n.m.r. Spectram’"31 Show that the inter-
action between a cation and an electron is extremely

-11

Shonrlived (correlation time ~o10 sec).31 Other

studies24 have shown that the electron apparently has

9
the capability of tunnelling from one cation to another.
This also supports a short-lived cation-electron inter-

action.
(3) Electron-Electron Interactions

One of the most amazing properties of metal-
ammonia solutions is the decrease of magnetic sus-
ceptibilityz’z“34 with increasing concentration. At
least in the case of potassium solutions,34 the sus-
ceptibility is also strongly dependent on temperature.

Heat-of-solution datalsr35 indicate similar char-

acteristics for sodium solutions.
(4) Non-Specific Concentration Effects

The properties which correSpond to the average
behavior of species are classified in this section.

Activity coefficient data yield, with increasing
concentration, a decrease of the mean activity coef—
ficient. However, this decrease is much smaller than
expected for the number of species required to de-
scribe other properties.

Measurements of the heat of dilution correlate
well with susceptibility data. Like the susceptibility,
the heat of solution is both temperature and concentra-
tion dependent.

Transport numbers demonstrate that most of the

current is carried by negatively charged species.

10
Furthermore, the fraction of current carried by
negatively charged Species increases as the concentra-
tion increases. Simultaneously, a drOp followed by an

increase is noted in the total equivalent conductance.
1.3. Models

Kraus36 was the first to introduce the cavity model
for metal-ammonia solutions. He came to this conclusion
because of the large volume expansion which occurs when
metals are dissolved in ammonia. In dilute solutions, he
proposed that the electron is in a cavity and that the cation
is removed from its vicinity.

Ogg5 performed calculations on this model by assuming
an infinite potential at the spherical boundaries of the
cavity. This simple model of a particle in a box was used
to calculate both the cavity radius and the energy of the
lowest state. 099 attempted to explain the optical absorp-
tion band but his predicted value of 3000 cm.1 for the
peak position is less than half the observed value of 6300
cm-1. Nevertheless, it was a major advance and set the
stage for a number of later calculations.

The model of Fueki, Fang, Kevan and Christoffersen37
is based upon a semi-continuum medium. The variation princi-
ple is used to optimize the cavity radius. Long—range
interactions are calculated from a polarization potential
and short-range attractive interactions are represented by

a point dipole approximation. This model is in fair

11
agreement with experimental data for glasses of water,
ethanol and methanol, but no attempts have yet been re—
ported for the ammonia case.

The most "successful" calculations were performed by
Jortner,4 who utilized the Landau8 potential. A continuous
dielectric medium was assumed.

It was found that the Optimum cavity radius required
to fit the optical Spectrum is 3.2 R which is only slightly
lower than the experimental value of 3.4 X. Jortner also
noted that the optical absorption could be considered as a
bound4bound transition in the well. A 15 ——+»2p type
transition was proposed. Further work was done by Jortner,
§£_gl.38 using SCF calculations. However, the agreement
with experimental data worsened.

The latest treatment in the "Jortner series" by COpe-
land, Kestner and Jortner”!4o has the following features:

(1) In the first coordination shell around the elec-

tron, there is assumed to be a small number, N, of

oriented solvent molecules. Calculations were per-

formed for N = 4, 6, 8, and 12.

(2) Solvent molecules in the first layer interact with

the electron through both permanent and induced dipole

moments. Furthermore, the solvent molecules in this
layer interact with one another through dipolar and
hydrogen-hydrogen repulsions.

(3) The electron-continuum interactions are contained

in the V0 term.

12
(4) In contrast to the previous theories, the elec-
tronic energy was calculated by assuming a free elec-
tron-neutral atom pseudopotential for short-range
attractions and a Wigner-Seitz potential for short-
range repulsive interactions. The Landau type inter-
actions as developed in the previous theories were
again utilized to account for long—range interactions
of the electron with the continuum.
(5) V0, the energy of the quasi-free electron state
is assumed in this model to be approximately in the
range of -0.5 ev to + 0.5 eV. V0 has been experiment-
ally determined to be in this vicinity.4o
From the results of these calculations, the authors
conclude:
(1) The major cause of the shift of the absorption
maximum with temperature is the effect of temperature
on the dipole orientation in the first layer.
(2) Line shapes are asymmetric, but the "tail" is
calculated to be on the low energy side rather than
the high energy side as experimentally determined.
(3) The calculated linewidths are relatively large
(0.1-0.13 eV) but are still only about one—fourth

of the experimental value.

Although the calculations of Copeland, Kestner, and
Jortner39r4° are not quantitatively correct, they seem to

provide a reasonable model which is complete enough to be

13
physically realistic, yet simple enough to be tractable.
Indeed, such calculations have also been extended to a
number of other solvents.“0

Recently, Newton“1 has performed extensive ab-initio
calculations on the (H20);- system. The configuration was
assumed to be the same as that in ice. The calculations
show that for this isolated species, thermodynamic stability
is unlikely. Upon the addition of a continuous dielectric
medium outside the primary layer, however, a strongly bound
localized electron system is predicted, with a cavity
radius of 2.68 8. Even though such calculations have not
been completed for ammonia, the results for the hydrated
electron are very encouraging.

While the calculations at least qualitatively account
for the properties of dilute solutions, controversy has
arisen over the Species which are responsible for the
changes in the observed properties as the concentration
increases. A variety of models have been proposed and the
stoichiometry of the species which are presumed to be pres—
ent is given in Table 1. It should be noted that the
structure assumed by different models for Species of the

same stoichiometry may be very different.
I.4. Pr0perties of Amine and Ether Solutions

The dissolution of alkali metals in "inert" solvents
has been extended over the years to include amines, ethers

and hexamethyl phosphoric triamide.43'55 Although the

14

Table 1. Stoichiometry of species in metal-ammonia
solutions according to various models.

 

 

 

Reference
+ _
Kraus M e 36
+ ..
Huster M e M2 32
Bingel M+ e' M’ 42
Deigen M+ e_ M M2 M” 43
BLA M+ e- M M2 44
Arnold and + _ _
Patterson M e M M2 M 45
Golden,Guttman + _ _
and Tuttle M e M M2 M 46
Pitzer, Gold, + _ _

and Jolly M e M M 47

 

15

solubility of the metals in most of these solvents is
orders of magnitude lower than in ammonia, studies of metal
solutions in these solvents have continued with the hope
that a more complete picture of the species responsible for
the properties of metal solutions might be obtained.

Definite proof of the existence of a monomeric species,
M, was first obtained by Vos and Dye for solutions of ru-
bidium and cesium in methylamine56 and independently by
Bar-Eli and Tuttle57 for solutions of potassium in ethyl-
amine. The hyperfine pattern observed in the e.s.r.
spectra of these solutions shows typical splittings by the
metal nucleus. For example, potassium (nuclear spin = g)
gives a four—line pattern, cesium (nuclear spin = %) ex-
hibits an eight-line pattern, etc.

The optical spectrum reveals that there is more than
one absorbing species present.

(1) The Infrared Band

The band in the near infrared region is similar to
the absorption band in ammonia. Only when this Optical
band is present are the solutions strongly paramagnetic.
Furthermore, pulse radiolysis and flash photolysis stud-
ies show the presence of this optical band even in the
absence of alkali metal. Figure 2 shows the comparison
of the spectrum of cesium in ethylenediamine with the
spectrum produced by pulse radiolysis of the same sol-

vent in the absence of alkali metal. From the

16

.mwmmHOHomu
wmasm an can mcofluaaom Hmuoe an megawavmcmamsum
cw Eduuummm conuumam owum>H0m on» no GOmHHmmEoo .N mnnmam

 

 

 

AWIOd x HIEUV umflfiucm>m3
m... we m4. 3 0H m . m w
_ I _ _ I a a 0.0
l
0 ll
lim.o
O
I xmza
o <
O l
O \
O \I
. \\\.
mummHoHomu wmanm n o O .0 .1
o .010
coHusHOm Enammo n Ill 0\0 1.0.4

 

17
similarity of these spectra it is reasonable to assume
that this band arises from the solvated electron,
again presumably trapped in a cavity. Conductance

measurements also agree with this assignment.
(2) The Visible Bands

In addition to the infrared band, metal-dependent
bands appear at higher energies. Sodium solutions in
ethylenediamine give rise to a band with an absorption
maximum at 650 nm while potassium solutions have a
peak at 850 nm.58 Since solutions which do not exhibit
an infrared band are either diamagnetic or only weakly
paramagnetic, the absorbing Species must contain
paired electron spins. Oscillator strength measure-
ments by DeBacker and Dye59 have shown that the
species responsible for these bands have two electrons
per absorbing unit. In addition, kinetics studies
by Dye, g£_§l,§° which utilized pulse radiolysis
techniques, have shown that both the growth of the
sodium metal-dependent band and the simultaneous decay
of the infrared band are second order with respect to
solvated electrons.

Prior to 1968, the species responsible for Optical
bands in metal amine and ether solutions were in dis-
pute because results in one laboratory were not re-
producible elsewhere. Finally, when Hurley, Tuttle,

and Golden,61 showed that potassium metal can rapidly

18
exchange with sodium from Pyrex glass, the results
began to fall into place and it became clear that a

given metal yielded at most two major absorption bands.

After this work and the proof that the optical band
could not be attributed to the monomer,56 it was generally
felt that the metal-dependent Species had either the stoi-
chiometry M- or M2. In 1969, Matalon, Golden, and
Ottolenghi62 compared the metal-dependent bands with the
iodide spectrum in the ultraviolet region. Iodides give
rise to the so-called charge-transfer—to-solvent (CTTS)
bands in this region.63 These authors showed that the
band of sodium in ethylamine-methylamine mixtures under-
goes shifts, similar to those of the iodides, with reSpect
to both solvent and temperature.62 Although the iodide CTTS
bands are commonly observed in liquids,63 they have no
counterpart in the gas phase. The isolated iodide ion has
no available low lying orbitals for this transition and
therefore the CTTS bands were postulated.

The CTTS theory can be summarized as follows: The
theory was first develOped by Frank and Platzman.64 Later,
Smith and Symons,65 as well as Stein and Treinen66 refined
the treatment to yield more accurate descriptions.

Briefly, the Smith and Symons treatment utilizes a
spherical infinitedwell calculation. Because the potential
outside the well is not known, it is set equal to infinity

such that,

and

where, r0 is the radius Of the spherical well.
Stein and Treinen invoke Landau's potential and the
general treatment of an electron in a cavity. Both theories

use cycles as shown below:

 

 

I- + { ’ EXi> I + e- + { }
9 ’ 9 g
t E1 1E2
- hvmax -
I { } > I i J
and
thaX = E1 + EX + E2:

-E1 represents the solvation energy of I-.
E represents the ionization energy of 1-.
E2 represents the energy required to attach an e-
to an excited state and to introduce the radical
into the organized medium. The brackets repre-

sent the solvent configuration apprOpriate to a

solvated iodide ion.

Although different approaches to the calculation of the
various energy terms for this cycle were used, both theories

predict the following equation:

20

Emax z Ea _ Lx + §%'(%'+ 2DL_'- 3L.)
op s
2 2 4 2
-§g<1-5§;>-(”2§e wig-514
where,
Ea = electron affinity of the ion
LX = heat of solvation of the ion
Re = radius of the excited state
DS 2 static dielectric constant
DOp = Optical dielectric constant
ro : radius of the cavity in which I- is contained.

This relation could, of course, be solved for the
radius by substituting the experimentally determined absorp-
tion maximum.

Both theories predict the band position to be solvent
dependent. Similarly, both treatments also conclude that
the absorption maximum will shift linearly to higher energies
at lower temperatures. Smith and Symons go one step further

and predict that a plot of Em y§_dEmaX/dT will have a

ax
slope equal to the absolute temperature.

A major problem in comparing alkali metal solutions
with iodides has been the compatibility of the solvents
used. Matalon, Golden, and Ottolenghi62 made the comparison
With an extremely limited solvent range.

A major breakthrough occurred in 1970 when Dye, DeBacker,

and Nicely67 reported the solubilization of alkali metals

21
in diethyl ether-~a solvent in which the alkali metals are
normally totally insoluble. Furthermore, solubility of the
metals was shown to be greatly enhanced in such solvents
as primary amines, tetrahydrofuran, etc. These solubility
enhancements were accomplished by the use of dicyclohexyl-
18-crown-6.68 This cyclic polyether complexes the alkali

cation and thus forces the equilibrium

+ —
> + e
<—-——'— M SOlV

 

M(s)

to the right. Even better complexing agents for the alkali
cation, develOped by Lehn, gt 31.,69 have been utilized in
this laboratory and are a major concern of this disserta-

tion.

1.5. General Summary

 

It would, of course, be feasible to treat ammonia as an—
other "typical solvent? such as the amines and ethers. How-
ever, the pIOperties of metal ammonia solutions differ great-
ly from those of the other systems and no comparison has
been validly argued.

In ammonia, the infrared band is assigned to the iso-
lated electron. At higher concentrations, properties of
such solutions reflect an interaction of the cation (M+)

With the solvated electron. Although an interaction is
present, it is very short-lived. At still higher concentra-
tions, spin-pairing occurs but the nature of the Species

involved is again arguable. Both weak inter-ionic

22
interactions such as "e- --- M+ --- e-" and strong inter-
actions such as the spherically symmetric alkali anion,

"M-", have been proposed.

The amines and ethers definitely show "strong" inter-
actions. It has been shown that the solvated electron
interaction with the alkali cation is long lived enough to
be observable by e.p.r. in ethylamine and methylamine solu-
tions. Yet, the exact nature of the monomeric species is
undetermined. For example, the interactions with 4-equiva-
lent 14N nuclei in lithium solutions in ethylamine have
been "explained" by both an "expanded monomer" of lithium
and by relatively weak interactions between the lithium
cation-ethylamine and a solvated electron.7°v71 The dia-
magnetic Species in amine and ether solvents is spectro-
scopically and kinetically identifiable and has the stoichi-
ometry, M-. However, as with the monomer, the detailed

structure of this Species is unknown.

1.6. Solid-State Studies

 

One major problem which occurs when metal amine solu-

tions are studied is that Species such as M-, and esolv

are present in the liquid phase and are therefore difficult
to identify unequivocally. Although studies of excess

electrons in solids have been made for F and F' centers
in crystals and for excess electrons in glasses, pure solid

salts of M_ have not been isolated. Metallic solids

23

which contain ammonia have been studied for some time. For
example, the work of Sienko and coworkers at Cornell has
identified compounds such as Ba(NH3)6 and Li(NH3)4. Although
the pure solids have been isolated and studied, they are
stable only at low temperatures or in the presence of
ammonia vapor. X-ray studies72 have also been performed
showing that the systems do not merely trap ammonia but are
true solid compounds.

However, no solids containing an alkali anion or a
solvated electron "anion" have been isolated. The evapora-
tion of solvent merely yields precipitated finely divided

metal powder.
1.7. Objectives of Dissertation Research

The purpose of this work was to:
(1) Expand the solvent range by the use of complexing
agents. The two agents used, dicyclohexyl-lS-crown-6,
or simply "crown" and 4,7,13,16,21,24-hexaoxa-1,10-
diaza-bicyclic(8,8,8)hexacosane, or "2,2,2-crypt"
are shown in Figure 3.

The Spectra of "M-" and the solvated electron in
a variety of solvents and solvent types were studied.
Simultaneously, the Spectra of the iodides in these
same solvents were investigated.

A comparison was made not only between species

but also with the general CTTS theory.

24

Figure 3. Dicyclohexyl-lB-crown-6 (I) and
2,2,2-crypt (II).

25
(2) Determine the metal dependence of the spin-pairing
effect in ammonia solutions. It is reasonable that
if discrete species such as M_ exist in metal-ammonia
solutions, spin-pairing should be strongly affected
by a change in the metal used. A comparison of the
number of unpaired Spins in solutions containing dif—
ferent metals was made. The effect on spin-pairing
of complexation of the cations by cyclic polyethers

was also tested.

(3) Attempt to isolate a salt of the alkali anion or
a solid in which the electron is the "anion". If the
cation is forced to stay inside the cage of a complex-
ing agent, it is reasonable to assume that upon solvent

evaporation, precipitation of metal might be prevented.

(4) Unify assignments for the models of such solutions.
It was hoped that with the results of this work, enough
knowledge might be gained in ammonia, amines and ethers
to permit a unified scheme to be developed which de—
scribes the species and the interactions which exist

in these solutions.

II. EXPERIMENTAL

II.1. General Techniques

 

II.1.1. Glassware Cleaning

 

All glassware was first cleaned with an HF cleaner
(33% HNO3, 5% HF, 2% detergent, 60% water) followed by a
thorough rinsing with distilled water. The vessel was then
filled with boiling aqua regia which was allowed to remain
in the vessel for at least ten hours. After rinsing six
times with distilled water and six times with conductance

water, the vessel was dried in an oven (110°) overnight.

11.1.2. Vacuum Techniques

 

Standard high vacuum techniques were utilized in this
work. Dual stage mechanical pumps and two stage oil dif-
fusion pumps allowed us to reach pressures of 10_7 mm Hg in
Pyrex vacuum lines which were fitted with liquid nitrogen
cold traps. Pressure measurements were made with a Veeco
RG75P glass ionization tube and an RGLL6 control unit. Ex-
tensive use was made of greaseless Teflon valves (Fischer-
Porter, Delmar, and Kontes types) some of which were speci-

ally modified for high vacuum work.73

26

27

11.2. Metal Purification

Alkali metals were purchased as follows:

Na : J. T. Baker Co. (99.99%)

K : J. T. Baker CO. (99.99%)
Rb : Fairmount Chemical Co.

Cs A gift from the Dow Chemical Co.

The metals were distilled several times prior to

actual use.

II.2.1. Storage of Alkali Metals in Small gpantities

 

a) Sodium and Potassium:

Small pieces of the metal were cut and most of the
oxide was scraped off the surface. AS shown in Figure 4,
these pieces of metal were then introduced into a 40 mm
diameter tube containing 20 to 30 lengths of one to 4 mm
diameter tubing which had been previously sealed at one
end. The system was evacuated and then gradually heated.
After the evolution of gas ceased and the metal had been
melted to form a pool of liquid at the bottom of the tube,
helium gas was introduced (P = 1/2 to 1 1/2 atm) to force
the molten metal into the small glass tubes. In this way,
small quantities of relatively pure alkali metal could be

isolated by simply cutting a length of glass tubing.

28

T High vacuum

 

 

 

~$-' Helium

J\ supply
)
W TS 40/50

I
WW“) mm OD

{fl//Z//’” 4 mm OD

 

 

 

 

90 cm

 

 

<9 ’fl/x—Sodium or Potassium

L
I t) metal
V w

 

 

 

 

 

 

 

 

 

 

Figure 4. Apparatus for the preparation of storage
tubes for sodium or potassium metal.

 

29

b) Rubidium and Cesium

Both rubidium and cesium are spontaneously flammable
upon exposure to the atmOSphere and thus special techniques
for the isolation Of small quantities of these metals are
required. The desired quantities of metal were Obtained by
the method of Dewald74 followed by the method of DeBacker?
The metals were then stored in a vessel as shown in Figure 5.
The metal could easily be melted down the length of tube so
that small quantities of oxide-free metal could be obtained

by sealing off the desired length of tubing.
11.2.2. Premeasured Amounts of the Alkali Metals

Known quantities of sodium, potassium, rubidium and
cesium were obtained in the following manner.

First, several lengths of approximately 1 mm I.D.
tubing were obtained and their inside diameters were accur-
ately determined at both ends with the aid of a microscope.
The microscope which had internal moveable cross hairs and
a vernier scale was easily calibrated with a vernier caliper.
In addition, to insure that the tube was relatively uniform,
the O.D. was also determined at several points along its
length. To check this technique, several such tubes were
cut at various points along their lengths and the diameters
were again measured. The tube was then attached at one end
to a 10 mm O.D. tube as shown in Figure 6. The appropriate

alkali metal, after being isolated as described in Section

3O

.HmumE EOHOflQsH
Usm Enammo mo mmmuoum OLD How Hmmmm> mmmam unmflu Essom> .m musmam

 

 

 

 

E §\\ Q\\ \ s\\ \\ J51
. S J? \\\\\\\\S

31

 

 

 

 

 

 

 

r V
K E

 

 

Figure 6. Apparatus for the isolation of known amounts
of metal.

32
II.2.1, was then intrOduced. The glass was sealed behind
the metal and the vessel was then evacuated. In the case
of rubidium and cesium, the small ampoules were cooled in
liquid nitrogen before exposure to the atmOSphere. The
metal was then melted and distilled. After the 1 mm I.D.
tube had been half filled, it was sealed near the top.
Small cylinders of metal could be heated down the tube, and
sealed off. In this way, small cylinders of metal were
isolated. Since the diameter of the tube was known, the
length of a cylinder of metal could be measured and the_
volume and hence the weight determined. In the case of
sodium, quartz tubes were used to avoid any melting of the
small tube which might occur because of the higher tempera-

ture required.

11.3. Premeasured Amounts of Materials

 

11.3.1. General

Handling of isolated amounts of materials, eSpecially
for use in vacuum systems has been a serious problem over
the years. Some authors have stated that the isolation of
material in thin-walled vessels with the total exclusion of
air and water is impossible.75 In 1947, however, a new
technique was utilized by COOps, £5 21,75 The method in-
volved the use of fragile glass ampoules. The disadvantages
were numerous, however. A total mass of approximately 2.5

grams had to be weighed on the balance, even though the

33

bulbs were developed for 400 milligram samples. Both before
and after filling, all grease had to be removed from the
ground glass joint, to which the bulb was attached. Errors
were introduced by any excess grease left on the ground
glass joint. Furthermore, small quantities of material
(2 100 mg) could not be accurately determined because the
total weight of the bulb and joint was so large. The
general method was revised for the isolation of alkali
metal in small quantities (approximately 50-100 mg) by Watt
and Sowards77 and also by Shriver.78 These authors were
able to eliminate the ground glass joint and therefore
reduce the bulb weight to 100-400 mg. However, this method
could not be used for sealing-off small amounts of salts or
liquids in the bulb without introducing air into the
ampoules.

We have extended the general bulb technique to permit
one to seal-Off very small known amounts of salts, solvents,

and solutions. The procedure is outlined in Appendix A.
11.3.2. Sub-milligram Amounts of Alkali Metals

Premeasured sub-milligram quantities of alkali metals
were also isolated in small, fragile, easily breakable
glass ampoules. This method however, differs slightly from

that for normal substances and is outlined in Appendix B.

34

II.4. Purification of Solvents

 

All solvents were first dried with either barium oxide
(BaO) or calcium hydride (CaHz) for at least 24 hours, with
the exception of hexamethylphosphoric triamide (HMPA),
paraffin oil, and dimethyl sulfoxide (DMSO). After initial
drying, ethylenediamine (EDA), ethylamine (EA), and diglyme
were distilled onto Na-K(1:3) alloy which yielded blue
solutions. To avoid decomposition these solvents were
distilled from the Na-K alloy into storage bottles after a
few days.

Ammonia and deuterated ammonia were also distilled onto
Na-K alloy which yielded blue solutions. After several days
of freeze-pump-thaw cycles, these solvents were stored on
the Na-K alloy at liquid nitrogen temperatures.

The solvents tetrahydrofuran (THF), dimethoxyethane
(DME), diethyl ether (DEE), tetramethyl guanidine (TMG),
nfhexane, and gfheptane were distilled onto a mixture of
benzophenone and an excess of Na—K alloy. The resulting‘
blue to purple color of the ketyl served as an indicator of
dryness. If the color faded, the solvent was redistilled
onto a fresh drying mixture.

Paraffin oil was degassed by repeated freeze-pump-thaw
cycles. The oil was then stored over Na-K alloy. The
purification of HMPA was performed by passage through a
column of activated alumina (80—100 mesh). DMSO was distilled

under vacuum after it had been freeze-purified. After this

35
solvent had been degassed by repeated freeze—pump-thaw
cycles, a potassium mirror was distilled onto the surface.
However, in this case, reaction occurred upon warming and

no solution could be obtained.

II.5. Purification of the Complexing Agents

The crown was purchased from E. I. Dupont de Nemours
and Co., Inc. and was further purified by passage through
an alumina column. Small samples of 2,2,2-crypt were gener-
ously supplied by two outside sources79 and in the prelimin-
ary work, this compound was used without further purifica-
tion. As the great potential of this complexing agent in
our work became apparent it was necessary to synthesize addi-
tional quantities of 2,2,2-crypt, since this compound was
commercially unavailable at the time. Efince the synthesis
procedure had been only briefly outlined in the literature,69
many steps in the synthesis had to be approached on a trial-
and-error basis. Several major improvements in the pro—
cedure were develOped by us in the process of synthesizing
the material. The synthesis was carried out in collaboration
with J. Ceraso and M. T. Lok and led to the following pro-
cedure. Although this may also appear in their theses, it
is included here for completeness. It is based upon the

synthesis of Lehn and co-workers.69

36

11.6. Synthesis of 2,2,2-Crypt

The general approach taken is outlined in the schematic
description. Many of the starting materials for the 2,2,2-
crypt had to be synthesized as shown below.

1) Preparation of 1,8—diamino-3,6-dioxaoctane (diamine)

CO\\ - +
ClCHZCHZOCHZCHZOCH2CH2Cl + o /N K
co

1,2-Bis(2-chloroethoxy) Potassium phthalimide
ethane
O O
I. II
C C
2fl§_., \\NCH2CH20CH2CH20CH2CH2N’//
C// \\\c
O O

Triethylene glycol diphthalimide

NH2 NH2 C0\NH
———> H2 NCH2 CH2 OCH2 CH2 OCHz CH2 NH2 +
EtOH CO//NH

1,8-Diamino-3,6-dioxaoctane Phthalhydrazide

HCl _ + + _ coc1
> ClNH3CH2 CH2 OCH2CH2 OCH2CH2 NH3C1 +
COCl

NaOH COONa
m> NHzCHch20C32CH20CH2CH2NH2 +

COONa

 

1,8-Diamino-3,6-dioxaoctane

2) Preparation of triglycolyl chloride

37

NH4V03
HOCHz CH2 OCHz CH2 OCHZ CH2 OH + HNO 3 —-—-——>

Triethylene glycol
OClz

 

S
Hococnaocuzcnzocnzocon > c1cocnzocnzcnzocnzocc1

Triglycolic acid Triglycolyl chloride

3) Preparation of 5,12-dioxo-l,7,10,16—tetraoxa-

4,13-diazacyclooctadecane. (lst cyclization)

Hancnzcnzocnzcnzocnzcnznnz + clcocnzocnzcnzocnzocc1

 

benzene ‘ NH//CH2CH20CH2CH20CH2CH2\\NH
hlgh d11Utl°n \\‘C-CH20CH2CH20CH2-C’/’

II II

o o

5,12-Dioxo-1,7,10,16-tetraoxa-4,13-

diazacyclooctadecane

4) Preparation of 1,7,10,16-tetraoxa-4,13—diazacyclo-

octadecane. (lst reduction)

//CH2CH20CH2CH20CH2CH2\\

HN\\\ NH + AlLiH4
c-cnzocnzcnzocnz-c”'
o o
cnzcnzocnzcnzocnzcn2
yfle>rm/' \NH

\\CH2CH20CH2CH20CH2CH5”
1,7,10,16-Tetraoxa-4,13-diazacyclo-
octadecane
5) Preparation of 2,9-dioxo-4,7,13,16,21,24-hexaoxa-

1,10-diazabicyclo(8,8,8)hexacosane. (2nd cyclization)

38

CH2 CH2 ocnz CH2 OCHz CH2\
\\ //,NH + clcocnzocnzcnzocnzocc1
cazcnzocnzcszocnzcnz

HN//

 

///CH2CH20CH2CH20CH2CH2
benzene
high dilution) N "CH2€H20CH2CHzoCH2cng/;,N
C-CH20CH2CH20CH2-C
0 o

2,9-Dioxo-4,7,13,16,21,24-hexaoxa-1,10-

diazabicyclo(8,8,8)hexacosane
6) Preparation of 4,7,13,16,21,24-hexaoxa-1,10-diaza-
bicyclo(8,8,8)hexacosane. (2nd reduction)
CHZCHzOCH2CH20CH2CH2
N — CH2CH20CH2CH20CH2CH2 —-N + BH3 in THF

c-cnzocnzcnzocnz-c’///
o o

 

_+ +-
> H3BN(CHZCHZOCHZCH20CH2CH2)3NBH3
Diborane of cryptate

- + +-
> HClN(CHZCHZOCHZCHZOCHZCHZ)3NClH
Cryptate-Hcl

HCl

 

CHZCHZOCH2CH20CH2CH2
’o e h
1 n XC ange ~ N -CH2CH20CH2CH20CH2CH2 ‘N

Dowexl-x8 \\\ ///

CH2CH20CH2CH20CH2CH2

 

4,7,13,16,21,24-hexaoxa-1,10-diaza-
bicyclic(8,8,8)hexacosane
(2,2,2-crypt)

11.6.1. Synthesis of 1,8-Diamino-3,6-dioxaoctane (here-

after called”diamine")

One mole of potassium phthalimide (185 grams), one-
half mole of 1,2[bis-(2-chloroethoxy)]ethane (93.5 grams)

and one liter of dimethyl formamide (DMF) are added to a

39
3-liter flask equipped with a mechanical stirrer. The
mixture is heated overnight to 95-100o with an oil bath.
The solution is then cooled to room temperature and poured
into 2 liters of stirred ice water. A white precipitate
forms. The whole solution is filtered with a fritted fun-
nel by use of an aSpirator. The white precipitate is re-
crystallized from glacial acetic acid. The recrystallized
product is then first washed with 5% Na2C03 solution and
then distilled water until no strong smell of acetic acid
remains. (The yield at this stage should be «'200 grams.)
The white solid so obtained is suSpended in «2000 ml of 95%
ethanol. The mechanically stirred ethanol solution is
heated to boiling under reflux. Just after the solution
reaches boiling, 102 ml of 85% hydrazine hydrate is added
and refluxing is continued for 2 hours. (After the addition
of hydrazine hydrate, a clear yellow solution is first formed
and then in about 10 minutes the whole solution solidifies
at once. At this point the stirrer is stopped.) At this
point 225 ml of 10N HCl is slowly added and the solution
is heated to reflux for an additional one-half hour. (Note:
HCl should be added in small portions with care.) Most of
the solvent (~'80%) should be distilled off.

Any precipitate which remains should be filtered off.
Solid NaOH and H20 are then added to the filtrate to make
the liquid strongly basic. This basic solution is extracted
with diethyl ether in a continuous liquid-liquid extractor

for 1-2 days. The ether solution is mzipped off in a

40
flash evaporator, and the condensed solution,which contains
a large amount of diamine,is vacuum distilled twice. The
final yield is ~v50 grams. (The reaction scale can be pro—
portionally increased if so desired.)

This method for the preparation of diamine is different
from that described by Lehn.69 With commercially available
1,2[bis-(2-chloroethoxy)]ethane, purchased from J. T. Baker
Chemical Co. rather than the corresponding bromine compound
(not commercially available) as used by Lehn, the synthesis
appeared to be much simpler and faster. The yield and

quality are about equivalent in both methods.

II.6.2. Preparation of Triglycolic Chloride (hereafter
called diacidchloride) ‘

 

In a five-liter flask, 5 grams of triethylene glycol
and 3 grams of ammonium metavanadate (NH4VO3) are added
into 3150 grams of 60% nitric acid. The solution is heated
to 68-730 and stirred with a mechanical stirrer. When
brown fumes form, 745 grams of triethylene glycol is drop-
ped into the flask by means of a dropping funnel over a
period of about four hours. The temperature should be main-
tained at 68-730. When the addition is completed, the
solution is stirred for another hour, after which 80% of
the nitric acid should be distilled off under vacuum. A
green syrup is obtained after most of the nitric acid has
been distilled off. Further evaporation is carried out in

an evaporating dish. The color of the solution changes

41

from green to brown to dark brown to purple and finally to
sky-blue at about 140°. When it is cooled it becomes a
hard, sky-blue solid. An ether Soxhlet extraction is then
done on the solid. A white solid (triglycolic acid) is re-
covered from the ether solution. (Because the diacid
chloride is unstable, large amounts of triglycolic acid can
be used for long storage.)

Thionyl chloride instead of oxalyl chloride is used
in the preparation of triglycolyl chloride from triglycolic
acid. In this way, the time needed for this procedure is
shortened from 24 hours as mentioned by Lehn69 to 4 hours.

Into 200 ml of diethyl ether in a one liter flask are
dissolved 60 grams of triglycolic acid and 180 ml of SOC12
(redistilled from commercially available SOClz). The
solution is heated to reflux for 4 hours at moderate tempera—
tures. Then ether is removed with a flash evaporator. The
yellow residue is washed twice with diethyl ether and then
recrystallized in an ether-petroleum ether mixture at
.~ -50°. The recrystallized white solid product is pumped
to dryness at 15° and stored in the freezer. The product
can also be stored in the freezer before stripping off the
ether. No distillation can be performed since the product
decomposes at elevated temperatures.

M.P. = 20-250

N.M.R.: 3.76 ppm (singlet); 4.48 ppm (singlet).

‘ t

42

11.6.3. First Cyclization (preparation of the di-lactam)

A solution of 29.6 grams of diamine in 1000 ml of
dry benzene and a solution of 22 grams of the diacid
chloride in 1000 ml of dry benzene are slowly allowed to
drop into one liter of dry benzene in a 5-liter flask. The
solution is stirred vigorously with a mechanical stirrer
and the system is purged continually with a dry nitrogen
stream. The entire addition is completed in 12 hours.
After the addition is completed, the benzene solution is
filtered and stripped to dryness. A white crystalline
solid is obtained with a yield of 75-80%. This white
solid is further purified by passing it through an alumina
column with benzene as an eluent.

TWice the amount of the diamine is used for cycliza-
tion instead of equal portions of triethylamine and diamine
as originally recommended by Park.3° We found that by
using double the stoichiometric amount of diamine, not only
is a purer product obtained, but also we attain a much
higher yield (especially in the second cyclization).

Half the amount of the amine used here reacts with the
HCl released from cyclization. This half can be recovered
in the following way:

The solid from the cyclization (amine-HCl-salt) is dis-
solved in 200 ml of concentrated NaOH solution. This
strongly basic solution is poured into a continuous liquid-

liquid extractor and extracted with diethyl ether for 2 days.

43
Finally the ether is evaporated off and amine is recovered.
The recovered yield is 85-90%.
Properties of the dilactam: M.P. - 110°
N.M.R. (CDC13): -CO-CH2-O : 4.00 ppm (singlet)
-CH2-N and CHz-O; 3.6 ppm (multiplet)

Mass Spectrum: 290 parent peak

II.6.4. First Reduction (preparation of the monocyclic

diamine)

Twenty-HUB grams of purified first cyclization product
is added with care in small portions to a mixture of 600 ml
of dry THF and 7.6 grams of LiAlH4. (LiAlH4 is first added
to the THF in small portions.) At the time of the addition,
the reaction mixture is purged with a slow stream of nitro-
gen. When the addition has been completed, a drying tube
is placed at the top of the reflux condenser and the solu-
tion is refluxed for 24 hours. The solution is then cooled
to 5° in an ice bath and 7.6 ml of H20 are added (with
caution) followed by 7.6 ml of 15% NaOH solution and finally
25 ml of H20. The solution is stirred for an additional
hour at room temperature. The white precipitate is filtered
and washed 3 times with THF and 3 times with ether. The
combined filtrate is evaporated to dryness. A white solid
forms which is purified by passing it through an alumina
column with benzene as an eluant.

The yield of monocyclic diamine is 70%.

44
Properties of the monocyclic diamine:
M.P. = 115°
N.M.R.: -CH2-O 3.60 ppm (singlet, triplet)
-CH2-N 2.80 ppm (triplet)
-N-H 2.25 ppm (singlet)(with a wet

sample, this peak may shift to
as high as 3.00 ppm)

I.R.: No absorption band of c=o can be detected.

Mass Spectrum: No 262 parent peak was found.

11.6.5. Second Cyclization (preparation of the bigyclic
di-lactam) '

 

A solution of 26.2 grams Of lst reduction product in
500 ml of dry benzene and a solution of 11.0 grams of di-
acid chloride in 500 ml of dry benzene are added into one
liter of dry benzene in a 5-liter flask. The solution is
stirred vigorously with a mechanical stirrer and is flushed
with a stream of dry nitrogen. The entire dropping time is
8 hours. At the end of the addition, the solution is
filtered and the filtrate is evaporated to dryness. A
white (slightly yellowish) product is Obtained with a yield
of 80-85%. The product is purified by passing it through
an alumina column with benzene as an eluant.

Again, a 2 to 1 mole ratio of diamine to diacid chlor-
ide is used. It not only increases the purity of the pro—
duct, but the excess first reduction product used is re-
covered at the end of the cyclization in the same manner as

for the first cyclization.

45
Properties of the bicyclic di—lactam:
M.P. = 1140
N.M.R. (CDC13): Several peaks between 3.3 and
‘ 3.6 ppm

Mass Spectrum: a parent peak at 404.

11.6.6. Second Reduction (preparation of the bicyclic
diamine-- 2,2,2-crypt)81

 

Ten grams of the second cyclization product is dis-
solved in 200 ml of THF. One hundred fifty ml of a 1M
solution of borane in THF is added to the flask slowly at
0°. After the addition is completed, the solution is
stirred for a half hour at this temperature and then refluxed
for an additional hour. A white precipitate is formed
during the process. The solution is cooled to room tempera-
ture and excess reagent is decomposed by adding 50 ml of
H20. (The solution becomes clear after the H20 addition.)
Solvents are evaporated. The diborane adduct compound is
changed to "cryptate-Hcl" salt by the addition of 200 ml of
6N HCl to the compound. The mixture is held at reflux
temperatures for one hour. The solution is evaporated to
dryness. The white crystalline solid is dissolved in 100 ml
of conductance water and the solution is passed through an
anion exchange column OXxex 1-x8, 20—80 mesh in the hydrox-
ide form). The column is washed continuously with conduc-
tance water until the solution coming out of the column is

neutral. The water is evaporated and further drying is

46
accomplished by using the azeotropic mixture evaporation
technique. The white solid 2,2,2-crypt so obtained is
pumped for a day before any further purification. The
yield is 90%.

Purification of the 2,2,2-crypt is accomplished by the
vacuum sublimation method. The snow-white product is col-
lected on a cold finger. The product is pumped under high
vacuum to remove the trace amounts of water remaining in
the sample. Zone melting purification was also performed.

Properties of 2,2,2-crypt:

M.P. = 68°

N.M.R.: -CH2-N 2.65 ppm (triplet)
~O-CH2CH2-O 3.78 ppm (singlet)
-CH2-O 3.65 ppm (triplet).

In addition, a new technique was developed to replace
the slow high-dilution cyclization step.69 This utilized
the principle of rapid, efficient mixing of reagents. The
yield was found to be comparable to the previous method.
Both the required time and the amount of solvent needed for
the reaction were drastically reduced. The reader is

referred to the thesis of M. T. Lok82 for details.

III. SPECTRA

III.1. Introduction

There is no direct Spectroscopic evidence for any
species other than the solvated electron in metal-ammonia
solutions. However, in metal—amine solutions, the optical

and ESR spectra can only be explained by invoking species

solv' In

of stoichiometry "M", and "M-" in addition to e
addition, the metal solubility is much lower in amines and
ethers than in ammonia.

The equilibria involved may be represented by the

following scheme:

2M(s) 5!.» M+ + M- (3-1)

H

M + e

 

solv (3'2)

+ ._

M + esolv (3.3)

Although the equilibrium constants are unknown, they
are dependent upon the solvating power and dielectric con-
stant of the solvent. Metal solutions in ethers and mono-
amines are described mainly by the first equilibrium. In

is

these solvents, the concentrations of rd and egolv

47

48
generally small compared with that of M-. In contrast,
the only spectroscopically observable species in ammonia is
eSolv' although magnetic and electrochemical data require
some aggregation. In some solvents, such as ethylenedi-
amine or hexamethyl phosphoric triamide, both M- and eSolv

are present.

Both the M- and e optical absorption bands are

solv
strongly influenced by both solvent and temperature. The
location of the M- peak position is also strongly influ-

enced by the metal used, whereas the absorption band of

esolv is independent of the metal used. For example,

Figure 7 shows the optical Spectra of esolv’ Cs , K , and
Na- in tetrahydrofuran (THF).

The spectra of eSOlv in metal-ammonia and metal-
ethylenediamine (EDA) solutions have been shown to be com-
parable with those obtained by pulse radiolysis.83 Recently,
spectral data have been obtained by pulse radiolysis techni-
ques for a number of other solvents.84 In the present work
we have:

(1) extended the solvent range in which the alkali

metals can be made to dissolve;

(2) measured the peak position and its temperature

dependence for the species Na-, K- and in most cases

and (3) compared the results for esolv with

solv;
those obtained by pulse radiolysis and flash photolysis

e

techniques.

49

.ummHUIN.N.N
Ho GBOHO mo mocmmmum on» EH Ammfiv amusmoupwnmuumu
CH Avvumz cam Amv & .Amvnmo .Afiv >HOmm mo com um mHfiwmm .h wusmflm

Amiga TEOV smgfiacw>a>>
ON 9 2 S S 9 w o e.

 

_ a _ _ _ 1 a _ _
[N.O
de
V m —
N V.
100W
W
nu
lm.ox

 

 

 

50

III.2. Experimental

Potassium iodide was recrystallized from triply dis-
tilled water and pumped under high vacuum before it was
introduced into the cell.

The metals were vacuum-distilled into the cell through
a sidearm. The appropriate purified solvent was then
vacuum-distilled into the cell. The cells consisted of 2
parts, (1) a quartz Optical cell, (pathlengths of 10.0, 2.0,
0.5, and 0.1 mm were used), and (2) a quartz or Pyrex side
vessel for both solvent and metal distillation. If the
formation of Na— by extraction of sodium from Pyrex is to
be prevented, it is necessary to use quartz vessels. For
the complete experimental details about the system used for

pulse radiolysis, the reader is referred to the thesis of

M. T. Lok.82
III.3. Instrumentation

Optical Spectra were obtained with either a Beckman
DK-Z Spectrophotometer or with a scanning system which was
designed for stopped flow measurements.85 The uv Spectra of
the iodides were determined with a Cary 14 Spectrophotometer.
The cell compartments of the DK—2 and the Cary 14 were
modified to permit variable temperature measurements over
the range of -100° to +70°. The temperature was controlled
by means of a Varian V-4540 variable temperature control

unit and measured with a thermocouple. For the pulse

51
radiolysis studies, the linear accelerator at the Ohio State

University was utilized.
III.4. Data Handling

All data were treated by non-linear least squares
analysis (see Appendix C). The data collected on the
scanning system were stored on FM tape and later averaged
through a Varian Computer of Average Transients (CAT). The
description of this system has been given elsewhere2 and
will not be repeated here. Peak positions observed with
the DK-Z and Cary 14 were tabulated with the estimated
absolute variances, as shown for example, in Table 2, for
solutions of potassium in diglyme. These data were then
punched by hand directly into Program KINFIT for non-linear
least squares analysis. The spectral data from the CAT were
fed into program SPFIT for calibration of wavelength and
absorbance before punching the data for the least squares

treatment.
III.5. Results and Discussion

With the aid of crown and 2,2,2-crypt, the Optical
spectra of Na-, K-, and eSOlv have been Obtained in a
variety of solvents. The range of solvents for the study
of these Species has now been extended to include primary
mono- and diamines, secondary amines, and straight, branched

chain, and cyclic ethers. The solvents in which the

Optical band of K- has been Observed are listed in Table 3.

52

Table 2. Tabulation of data of K- in diglyme; the tempera-
ture and variance of temperature are listed with
the peak position and the variance of the peak
position.

 

 

 

T (0C) OT2 cmax(cm-1) 052
max

-47 4 12,005 5041
-68.5 4 12,255 5476
-48 4 12,063 5184
-29 4 11,976 5041
-25 4 11,904 4900
—19 4 11,792 4761
-13 4 11,737 4624
-10 4 11,723 4624
- 4 4 11,669 4624
0 4 11,655 4624
10 4 11,403 4225
20 4 11,261 3969
-80 4 12,300 5625

 

53

.mmmwzucwumm CH cm>flm ma musumumewu Eoou Ham: ucmumcoo UHHDOOHOHQ

**

 

. Ow>ummno Omam >Homm*

Av.> V cmuswoupazmnume

AN.h V mcmnummxozumafln

Am.mv unflEm meoumtmlfla mamamwn
Av.mv Hmnum Hmmoumhmlwo OCHEMHOmcmmOHmIN.a
Ao.vv stum HmQOHQOmflIHQ Am wv mcHEmamsum
Am.nv Ocean Hmnumaa Aw. av *mcdfimamsumz
Am.vv Hmsum Hmzumao Am. NH v - *mcflsmapmcmahsum
:Ao. omv OOHEMHHD tOHuonmmonm Hmnummemm

 

wumummuu no :3ouo mmnfisvmm

Omumflmmmco mm>H0mmHQ

 

 

.©m>ummno comp mm; 1% LOHSB cfl mucm>aom .m magma

54
It was not possible to dissolve K in triethylamine (D =
2.42) even in the presence of 2,2,2-crypt. Because of the
solubility enhancement and solvent range extension it has
now become possible to better understand the equilibria
existing among different species in solution. The overall
equilibrium scheme given previously must simply be modified
to include the effect on reactions 3.1 and 3.3 Of the com-

plexation reaction

 

<

 

M+ + c > MC+ . (3.4)

When a metal such as Na or K dissolves in liquid ammonia,

the equilibria lie far to the right and only e- can

solv
be detected Optically or by ESR techniques. On the other

hand, with EDA. MA, EA, etc., mixtures of M'. M. and egolv

result, and the solubility is drastically reduced. In all
cases, M appears to be a minor species and only for a few
solvents is e-
solv
as DEE, DEA, etc., the metals are normally insoluble and a

Optically detectable. In solvents such

cation complexing agent is required to dissolve the alkali
metal. In all cases tested, the addition of crown or 2,2,2-

crypt increased the solubility and tended to give M- and/or

e .
solv

The spectrum of the latter Species was always ob
served (frequently along with the spectrum of M-) when crypt-
ate was added (in excess of the amount of metal dissolved)

to those solvents listed in the second column of Table 3,

and when either crown or 2,2,2-crypt was added to solu-

tions of potassium in the solvents listed in the first

55
column. These results are adequately described by equilibria
3.1-3.4.
At 25°, the positions of the absorption maxima of Na_,
K—, and e-

solv
15,200, 10,700-12,2oo, and 4,880-7,810 cm"1

in these solvents vary over the range 12,700-
, respectively.
In all the solvents tested, the peak positions proved to be
strongly temperature dependent. A linear blue shift with
decreasing temperature was Observed in all cases as Shown

in Figures 8-10 for Na-, K-, and e-

solv' The temperature

coefficient (determined by least-squares analysis) together
with the corresponding positions of the maxima at 25° are
reported in Table 4. Although the variation of the tempera-
ture coefficient with solvent is only a few times larger
than the error in measuring it, the results are in agreement
with the prediction of charge transfer to solvent (CTTS)
theory°5o66 that the position of the maximum be a linear
function of the temperature coefficient. According to one
theory65 the slope of a plot of Umax y§_deaX/dT should be
T. The results for Na- are shown in Figure 11. The solid
line has a SIOpe equal to the absolute temperature.

Based upon these results, the shift of the Na— and K-
absorption bands with both solvent and temperature is pre-
sumed to be characteristic of a CTTS band as first suggested
by Matalon, Golden, and Ottolenghi.62 As a further test of
the applicability of CTTS theory, a linear correlation of
the absorption maxima of two anions measured in the same

set of solvents is expected.86 Matalon, Golden, and

56

 

  

 

 

 

 

1.5 *-
\ououme
o
1.4 -- \ °
2
0"“ 9 QNA
: lo
.. _ F
8.): m ’0 EH.
:1 L; g
i b as
a. QB
n. z ‘ \
In , a
Z I“ \ A
a
, 12PDA
la -
\
a
1.4 *
\DME
l l l l l J 1
«100 ~60 '—20 20
f ( ° C ) _

Figure 8. Temperature dependegce of the position of the
band maximum for Na in various solvents.

57

 

 

 

 

 

 

 

1??

1.14
2f 1.06
0'1
E (3
23' ,5 -
m .c
x a 3’ 53
f .2
'c- i?
I): 15".:

\\

1.22 '-

1.14

1.06 w 1 1 1 1 _.._1 1 -__J

---100 ~60 —-2O 20

H°c)

Figure 9. Temperature dependence 9f the position of
the band maximum for K in various solvents.

 

9
E!
f”)
910
y. '-
; >< 8
o.—
o. 115
‘t u
d -_
W .
a. ()1
.- 1.“
‘61A
Inf 7
6
5
~10
Figure 10.

58

 

 

 

 

 

LQPDA
A
a
A
A
X
\\\I A
X
\\ EA
0\
Qn\o\ x
000 [DEE
0 X
\u\\
0
--_._... J I l l I 1 l
0 —60 —20 20

1 (0c )

Temperature dependepce of the position of the

band maximum for

e in various solvents.

solv

59

.Amhmfiv mphmuw>flgo mnmnm cmmflsoflz .mflmmse .o.:m

m.xoq .B .2 Eoum cmxmu puma.

. OOI
\l\l.. Cll MU
.Awmmfiv chow .ms .smro .mth .s .psmzmn .m .m tam mxoohm .s .so
.COHumH>wO pumpcmumfl
.Omuoc mmHBumnuo mmmacs 0mm um Godufimomm

 

 

 

 

 

 

 

 

III III III NN.H III III DQ>HOIN.N.N OCHEM ammoumImIHQ*
III III III mo.H III nN.HDQ>HUIN.N.N Hmzum HMQOHQOmflIHo*
III Hm.h III 4H.az pm.o 4m.mH mm.H “msgqu.m.m mtgsmsmhumhg.
no.0 mm.mfi mH.m mo.H wfi.mfi so.H mh.H NH.SH om.H uQSHUIN.m.m Hmsum assumho*
III III III ho.H III Hm.fi CBOHO Hmnuw Hunumflo.
III os.h sm.o mm.s mfl.fi 4m.o sm.ofi mm.H nacho ocmnummxonumshn
III mw.v III HH.H mm.o mm.mH om.H ugsHoIm.m.m awhamoupmsmhums*
III III mv.o vu.m NH.H III mm.H CBOHO cmusmouchMHumH
III pflm.m mm.o mp.mH mH.H SN.H «n.mfi ov.H czoho mssHmHa
hm.m mo.mm om.m sv.o mm.oH pH.H mm.H om.mH m¢.H g3ouo mchemngmgmmoumIm.fi*
mm.o mm.mm vh.h vfi.o mm.s m~.H sm.fi om.oH Hv.H :zouu mcgsmassumI
mh.H mo.mH om.s mm.o ms.s ma.” ms.o mm.mH mh.H III mcgsmgtmtmastum
III . . III . III . III OOHEMHHD oeuonmmonm
new 4 so a Hm fl Iamtumsmxmm
Q XMEDUI meP Q XMEPUI XMEP Q XmEPUI XMED UCHXOH HCO>HOm
addWWO IM Imz IQEOU
> .mucm>aom

HOW . . m

msoflum> CH IT can x Imz How mucwfloamwmoo musumummfimu paw mcofluflmom xmmm .v OHQMH

60

 

 

   
   

 

 

 

 

EDII
If) 1..
I2PDA
H—o—o--—4
75."
o .’ EA
.5: 2 : o .
G X DIGLYME
0— I4 —- o—A
c. 1'.
V u DEA
:33 .., DME *--—0---—*
a. :1 "_O“—"
'o 1;:
Z
THF
.__.o_q.,
DEE
, n
13 — - V
l I “_ 1-
~10 -14 —18 4??

diam/d I

Figure 11. The relation between the Na- peak position at
25° and its temperature coefficient. The
lepe of the straight line has a magnitude
of 2980K.

61

(Dttolenghi62 found a linear dependence of the peak position
of the Na- band with the 1‘ band at the same temperature in
ethylamine—ammonia mixtures.62 we attempted to extend this
test from mixtures of two solvents to a variety of solvents.
To do this, the spectrum of KI was also studied by us in
these solvents. (To enhance the solubility of KI, crown was
added to the solvents, EA, DME, THF, and DEE, while 2,2,2-
crypt was used with DIPE. No complexing agent was used for
the KI-HMPA system.

The iodide shifts with temperature in the various sol-
vents are shown in Figure 12 and tabulated in Table 5.

A plot Of the energy maximum for I_ y§_ the energy
:maximum for Na- is shown in Figure 13.

The lack of correlation might be associated with the
iodide absorption spectrum for the following reasons:

(1) The iodide CTTS transition is strongly dependent

upon the cation in solvents of low and intermediate di-

electric constant, presumably because of ion-pairing

effects.35.87

(2) Determination of the position of the first CTTS
transition may be complicated by overlap with other
transitions,3"I°9 as shown in Figure 14 from the results
Hayon and Fox.89 In addition, the solvent absorption
prohibits the determination of such bands. These bands

may shift differently with temperature.

62

 

 

 

 

 

451—
.. 5 n
44 -
'3’
f‘
E
.2
g _
8
In
42 —-
l I l I l L 1
-80 -40 0 40

T (DC)

Figure 12. Temperature dependence Of the position of the
band maximum for I- in DEE with crown (1), THF
(2), THF with crown (3), DME with crown (4),
and DIPE with 2,2,2-crypt (5).

63

Table 5. Peak positions and temperature coefficients for

 

 

 

I in various solvents.
Solvent ngpiixing vmax -dvmax/dT o
(cm-1) (cm’l/deg)
THF --- 41,550 18 0.453
THF crown 41,580 24 0.915
DEE crown 41,520 16.8 0.392
DME crown 42,320 12 0.262
DIPE 2,2,2-crypt 43,520 10 0.558
EDA --- >44,000 -- ---

HMPA --- 38,491 -- ---

 

64

m2

.omm um muam>HOm msoaum> ca

 

 

mo pcm IH mo coauflmom xmwm may :mw3umn coaumamu use .mH musmflm
. O
A In TEX—Bu... eh .
Ow . ov mm.
A _ _ A N—
0
was
0 0 I1 HP
who <e2:
4_
m
o o
are x
o )
L . 3
min 0 VP w._..
(a x
m.
.7
N
3
J n—(
IIo
«on

 

 

IFigure 14.

 

 

 

 

 

 

 

 

65

/

/

l- -

F

O '3
9

O

,—

 

 

45 40 35

x.(!n‘I.IIu'I

Absorption spectra of 5 x 10-3M iodide in
hexafluorO-Z-propanol(top, tetramethylammonium
salt and in acetonitrile (bottom, potassium
salt , at 274°K. Heavy lines represent the
spectra Obtained experimentally, The continu-
ous narrow lines represent the gaussian analogue
analysis of the experimental curve, and the
broken lines the necessary continuation of the
gaussian analysis beyond the limit of measure-
ment.

66

These difficulties may also account for the poor cor-
relation of gmax with dUmaX/dT tor 1-, observed by several
authors 87,83 with solvents of low dielectric constant.

Although the spectrum of Na- fails to correlate with
that of 1-, the Spectra of Na- and K- are well corre-
lated. The peak position of Na- y§_that of K- at 25° is
Shown in Figure 15. The linear correlation is excellent for
all solvents in which both the Na- and K- bands were
Observed. It is clear that whatever property (or properties)
affects the absorption maximum of Na- has a similar effect
on the absorption maximum of K-. Interestingly, the ratio
of the variation with solvent of the peak position of Na-
to that of K- is about the same as the ratio of their
temperature coefficients. Correlation of the peak position
with various solvent properties such as the static (DS) and
Optical (Dop) dielectric constants, dipole moment, polariza-
bility, Kosower's Z values,9°I91 and (1/DOp - 1/DS)2 have
been tried without success. I

Compared with the M- absorption, the esolv band is
often much broader and has a larger solvent dependence and
a greater variability of the temperature dependence.
Figure 16 shows the solvated electron absorption Spectrum
at room temperature in a number of different solvents. As
is readily seen from this figure, there are marked variations
from one solvent to another in both the band shape and the

peak position. For example, while ethylamine and diethyl

ether give rise to nearly the same position Of the

13 ~-

L5
2
O ,A-
I': If
g 9 M
a ?'
M T
4 55
E: u
.0 --..-’
Z
L?
1
Figure 15.

67

 

“~- -.--.-

EDAJ//

 

 

 

12PDA°
EAP
ODIGLYME
ODME
OTHF
Imuwo
DLE
our
0
J l l
.0 IJ L2 L3
—I —3
(cm on )
K” PEAK POSITION
Relagion between the peak position of Na_ and

of K in various solvents at 25°.

68

.mucm>aom msoflum> ca ohm um couuomaw Omum>HOm mgu mo muuommm was .OH musmam

 

.01

Op 0 m h o n v

A A a _ In A .4 _ _ _ _ o

o I

h
N.l
vi
91
mfI a

 

 

 

 

69

absorption maximum there is a very pronouncel Iifference in

the width of the band. The spectrum of e-

solv in ethyl-

amine is much narrower than it is in diethyl ether. On the
basis of similar solvent properties one might expect simi-
lar band widths in EDA and 1,2-PDA. However, EDA yields a
much broader band for eSOlv than does 1,2-PDA.

The solvated electron band width and position are not
only affected by the solvent, but are also strongly in-
fluenced by temperature. It was Shown that the solvated
electron band shape in EDA shifts to higher energies and
becomes broader as the temperature is decreased. A more
pronounced effect was demonstrated by Lok82 in 1,2-PDA and
this is shown in Figure 17. This same phenomenon was also
observed in the case of ethylamine. In Spite of the greater
sensitivity of the solvated electron spectrum to solvent and
temperature, the position of the absorption maximum corre-
lates well with that of Na- and K- as shown in Figure 18.
Therefore, the predominant effect of solvent on the Spectra
appears to be the same for all three cases. The band width
of M- is also dependent upon temperature, but it decreases
with a decrease in temperature in contrast to the band width

J

of e

solv' For example, the width at half-height of the K-

1

band in ethylamine decreases from 3320 cm- at 39° to 2480

-1 1

cm at -24°. Similarly a decrease from 4150 to 3190 cm-
occurs between -9 and -71° for the Na- band in diethyl
ether. A similar but smaller effect has been noted for the

absorption of I- in water.65

70

.mmusumnwmemu mzoflum>
um mcflemflowcmmoumIN.H ca couuooam Omum>HOm ms» mo Esuuommm one .SH mnsmam

A 7.23 $3.32 m><>>

AA 0000 coon 000m

 

 

--—..--—.

// \NK.
OVA.\./\\m ..

(JO
1\
I

<3
N(’)

1
-/I$<(D

U
U

N'—

C

 

 

71

> Om .omm um mucm>HOm msoaum> cH
H w mo cam x m0 coauflmom xmmm mzu cmmBqu COwumamH One

20:30.. 53.. Is.
A: 9 x . 5.3 .85..
(I r1

NA _..—

 

“ _ _ a

 

 

non-n...

.wH madman
CA
N
a
Iv ml.
..u_A
mu
HIV
3 x
W_d
xtm
I00_m
00
(N
Im

 

 

72
It is important to determine whether the change in shape
of the spectrum of esolv with temperature is an intrinsic
property of the solvent or whether the influence of solute
concentration is important. The formation of ion pairs

+
between M and e l or of a Spin-paired species32134
so v
could result in temperature-dependent equilibria among
species with slightly different Spectra. Several authors92.94

have reported the presence of optical bands which cannot

be attributed to either M- or e .
solv

Such bands could
arise from the interaction of esolv with the cation91o93
and might be responsible for the variations in band shape
which we observe.

Although the Spectrum of e could be studied as a

Solv
function of concentration to determine whether the band
shape changes, intense solvent absorptions limit the path
length which can be conveniently used. Therefore perhaps
the best method to separate concentration effects from sol-
vent effects is to compare these results with those deter-
mined by pulse radiolysis. In the latter case, the concen-
tration of ions can be kept very small so that ion—pairing
effects can be minimized.

As shown in Figure 19 the position of the absorption
maxima found in this work correlates well with the results
of pulse radiolysiss4 and flash photolysis.91 However, the

Spectrum of e- obtained by dissolving metals is, in

solv

some cases, broader than that observed by pulse radiolysis.

This is shown in Figure 20, in which the Spectra obtained

l

\

XICM~1 X 10—3 “2
O

u

PULSE RADIOLYSIS

VT 1"
CI
in

 

Figure 19.

 

73

 

  
   

 

 

5.5 I 0.53 7.5
7mm ( CM— x10" )
METAL SOLUTION

The relation between the peak position of

esolv obtained by dissolVing metals and by

pulse radiolysis. The straight line has a
unit slope. Comparison in the case of diglyme
is with flash photolysis rather than pulse
radiolysis.

74

CuHB mme CH 2 .
LuHB mmo CH M .000

.mme mo mHmmHOHOmu mesm . IIIIII num>HOIN.N.N
“mam mo mHmmHoHomu mmHsm .I.I.I. “DQAHOIN.N.N
o umHmmHOHOmH mmHsm an UCm mHmumE mCH>HommHO m9

 

OOCHmqu mmfi OCm mmo CH omm um >Homm mo mmmnm OCmQ OCH mo ComHummEoo .om musmHm
AwIOWKTEuv nmnwruncmeeusfi
o w A o m v
. I_ . I. e ”I _ I I. A I .8

NOII

 

 

 

75
in this work for the solvents THF and DEE are compared with
the corresponding spectra measured by pulse radiolysis.34
The broader band Observed in the present study may also ac—
count for the fact that the peak positions Obtained from
metal solutions tend to be at slightly higher energies than
those found by pulse radiolysis (Figure 19).

The discrepancies in peak position and band width
prompted us to perform some pulse radiolysis experi-
ments.

Reports of new bands in ethylamine by Fletcher95 and
in THF by Dorfman,96 caused us to investigate some of these
solutions.

We pulsed pure THF solutions and obtained only the
solvated electron band. After the addition of sodiumions to
the solution, in the form of sodium tetraphenylborate, we
observed only the band at xmax = 890 nm upon pulsing, in
accordance with the results of Dorfman, 35.31.96 Further-
more, after the addition of stoichiometric amounts of 2,2,2-
crypt into this solution, pulsing again yielded only the
solvated electron. This is quite reasonable since the com-
plexing agent would be expected to effectively isolate the
cation from any strong interaction with the electron. These
results are shown in Figure 21.

More recent measurements by Fletcher, ep‘gl.97 in
ethylamine also support the formation of a monomer Species,
"M". Fletcher and coworkers97 have shown that pulsing vari—

ous M- solutions in ethylamine gives rise to the growth

76

.mumo DQAMOIN.N.N OCm +mz UCm mmB H50 .4 “camp +mz
UCM mma H50 .x “camp hue mudm H50 .0 “camp m.CMEMHOQ OCm £umuxoom . D
“mumo m.CmeuOQ UCm >0b .o umcsum CoHusHOm HmumE .4 “um>HOIN.N.N pcm

 

 

+mz CuH3 mus OCm .+mz CuH3 mma .mma wusm mo mmHosum mHthOHOmu mmHsm .HN wusmHm
A mIoH x NICO V hmmfizC mbwz
4H mH NH HH 0H m m I m m
H J
n u — — _ — X\ _
Ux I m.o
/
O/O \ /
I.
o 1
.IOIIIII [JV/Ill // l
QXLV 1\U O/ ‘
\o\0 0. 0]“ O O /
. 1
O / / N
Q Q 0 ‘ UV

,IUIxU

 

 

77

of the infrared band (due to e ). This band then decays,

solv
giving rise to another band which is intermediate in wave-

length between the M- and e-

solv bands, thus also support-

ing a monomer band. Van Voorst and coworkers92 obtained
similar results upon flashing metal solutions in diglyme.
Kinetics studies performed by Dorfman 2.13 _a__]_._.96 also support
a species of stoichiometry "M" in THF.

Because of the proof for the existence of the species
"M" in THF, we also investigated other solvents by pulse
radiolysis techniques.

We pulsed deuterated ammonia, containing no alkali metal
and mapped the complete spectrum. Earlier work98 in this
area yielded only the high energy side of the solvated elec-
tron absorption maximum. Since the complete spectrum was
available from metal solution work, we could compare our
pulse radiolysis results. If the metal solution band were
broader on the low energy side, then, perhaps, a monomer
band was also being formed in ammonia. However, as shown
in Figure 22, the spectra matched very well, which indicates
that ip_ammonia, there is pg formation of a monomer band.

We also mapped the spectrum of 1,2-propane diamine at
different temperatures, by pulse radiolysis to determine
whether the Spectral results from metal solutions (which
indicated a broader band at lower temperature) were caused
by the formation of a monomer band at lower temperatures.

It might be possible that the monomer band lies at higher

energy than the electron band and is favored at lower

78

.mump memHo

IHpmn mmHsm HOHHHmm .q Q Q "much mHmeOHOmH mmHsm mac .0 O o “COHDCHOm

Hmuoa umHthOHUmH OmHsm an OCm «Oz CH mCOHquOm HmumE an OOOCO
IOHQ COHDOOHO Omum>HOm may no Esuuommm COHDQHOmnm may no COmHHmmEOU .NN musmHm

 

 

 

 

A mIOH x NICO V smMEsC m>w3
2 m m .1
. H 4 H 4 a _ _ o
o o \I
Q 0
o O Ox
0
a \ I
Q o
o
Imam,
IL WV
00 X
0 I1
I \
A O In
I
I I
V O
. O I ,OH
0 II
o
r o

 

 

79

temperature. The band broadening at lower temperature could
be, therefore, caused by the overlap of the two bands. How-
ever, as shown in Figure 23, the spectra Obtained from the
pulse radiolysis of 1,2-propane diamine, in the absence of
alkali metal, showed similarly broader bands at lower tem—
peratures. Thus, at least in the case of 1,2-propane di-
amine, the broadening of the solvated electron band with
lower temperatures is an intrinsic property of the solvent.

Attempts to dissolve a hydride in ammonia and in di-
ethylether failed. Potassium hydride was used in the
presence of crown. However, rapid bubbling occurred, which
was presumably, the liberation of hydrogen gas.

Several attempts were made to check the theory that the
solvated electron band in ammonia is composed of two bands,

one from e and one from M-. For this purpose, diethyl-

solv
ether was distilled into an ammonia solution containing a
moderate concentration of potassium. It was reasoned that
if the two-band theory is correct, the solvated electron
band should shift to longer wavelengths while the second
band, presumably "M-", shifts to shorter wavelengths. How-
ever, no such shifts were Observed in solutions containing
up to 10 mole percent diethyl ether. Above this concentra-
tion, the solution forms two phases. Similar phase separa-
tions were also observed in solutions containing potassium

iodide. In the absence of any salt, these two solvents are

miscible.

80

.mpsum mHmaHOHOmu mmHsm .4 OCm o “mosum COHusHom Hmume
“COHusHom Hmuma wn OCm mHmeOHOwH mmHsm ma UmCHmuno mwusumummEmu
msoHHm> um OCHEMHOOCMQOHQIN.H CH COHDOOHO Owum>Hom can no Esnuommm use

AmIOH N I no V MOMHSC ohms

r-

 

 

 

 

 

 

.mm musmHm

m.o

v/v

XBUI

IV. ESR STUDIES OF METAL-AMMONIA SOLUTIONS"

IV.1. Introduction

 

IV.1.1. Historical

 

In view of the existence Of the Species "M" and "M-"
in metal-amine and metal-ether solutions, it has been pro-
posed that such species might also exist in metal-ammonia
solutions. As stated earlier, these prOposals range from

weak interactions among the Species M+ and e to the

Solv
strong interactions characteristic of the monomer or the
alkali anion.

To check for the existence of Species, "M" and "M-" in
metal-ammonia solutions both conductance and magnetic sus-
ceptibility measurements have been performed.

Conductivity measurementsz":28 with sodium-ammonia
solutions reveal the formation of ion-pairs. This effect
is also observed with normal salts in ammonia and thus,
comes as no surprise.

Magnetic susceptibility data were obtained over 20 years
ago by Hutchison and Pastor.34 Results from potassium-

ammonia solutions indicate the formation of a spin-paired

species. Furthermore, this diamagnetic Species must be

 

*This work was carried out in colloboration with M. T. Lok.

81

82
energetically more stable by at least several times kT
than the unpaired electron spin.

Preliminary studies of sodium-ammonia solutions by
Hutchison and Pastor34 and by O'Reilly31 gave similar re-
sults.

The dilemma arises in attempting to fit a valid model
to the system. Electrochemical data, activity coefficients,
and transference numbers can all be adequately explained in
terms of free-ions and ion-pairs.13 However, upon the intro-
duction of a spin-paired Species such as M2 or M- to
explain susceptibility data, the conductivity results can no
longer be satisfactorily rationalized. For the details Of
these calculations, the reader is referred to the review by
Dye.13

Recently, DeMortier, pp 31.99 investigated the suscepti-
bility of sodium-ammonia solutions and concluded that no
detectible spin-pairing occurs in solutions containing up
to 0.01M sodium. The authors further concluded that the
previous results in the 0.01 to 0.1M region for sodium solu-
tions could not possibly fit the extrapolation of their
curves and thus were incorrect. This is demonstrated in
Figure 24.

DeMortier, ep'gl,99 concluded that the dilemma pointed
out by Dye13 had been resolved, since if there were little
or no spin-pairing in sodium-ammonia solutions in this re-

gion, the conductivity data could be adequately understood.

83

 

mm W" Tor-qutl I 1 II'WW.T

mole . '

10"

r V‘jvr‘r

V

I0‘2

I V ‘lvvvi—_‘

12
'0
U r *‘Yr—I-“fi
. \\
k..- :..:.».:;.z;.._.._i__o__i.i.'.&.u!.__._J.—;_.I_L.Lu.a

IO'3

 

' ‘ V—I'YVII— if v

 

 

l I lLljlll A 2 1.111111 1 l 24 inntl A 444,;4“,

Io'3 Io'2 Io c

-1 molo‘

Figure 24. Concentration of paramagentic species in
Na-NH3 solutions at -33°: 0, results of
Hutchison and Pastor; 0, results of DeMortier.

84

1V.1.2. Preliminary Studies

 

To check the conclusion of DeMortier and coworkers,
several preliminary experiments were performed in this
laboratory in which the complexing agent "crown" was added
to a potassium-ammonia solution. If the cation strongly
affects the extent of the spin-pairing process (as would be
anticipated if DeMortier, DeBacker, and LePoutre are
correct), an increase in signal strength would be expected.
However, no noticeable increase in the number of free spins
occurred upon the addition of an excess of "crown" to a
potassium solution in which the Spins were already approxi-
mately 95 percent spinepaired. This suggested that the Spin-
pairing process was not affected by complexation of the
cation and that perhaps DeMortier, DeBacker, and LePoutre

had erred in their measurements or conclusions.
IV.1.3. Statement of the Problem

To perform a rigorous and detailed study of this prob-
lem is a very difficult task, and was beyond the scope of
this project. The problems which arise in trying to
measure the absolute number of Spins are enormous. In an
X-band ESR Spectrometer, temperature, concentration, cavity
tuning, and sample position, to name just few, all pose
serious problems which must be overcome in order to measure
the absolute number of Spins.

In addition to the problems noted above for normal

solutions, metal-ammonia solutions impose extra restrictions.

85
The electrical conductivity of these samples is extremely
high while the linewidths are among the narrowest known.
For these reasons, previous work in this area utilized radio
frequency (megahertz region) ESR Spectrometer.

BecauSe of these limitations, we attempted to obtain
only the area ratio of the metal-solution spectrum to that
of a Spin standard. While we did not attempt to measure the
absolute number of Spins, it was felt that a valid compari-
son Of the metals could be easily made by the use of Spin
standards. In this way, we could check the sodium to potas-
sium ratio, as a function of both temperature and concentra-
tion. Also, by normalizing our potassium results at one
temperature and concentration to the accepted potassium re—
sults of Hutchison and Pastor34 under the same conditions,
we could (1) check our potassium results to see if they are
in agreement with those of Hutchison and Pastor, and (2)
compare our sodium results with those of DeMortier, e£_§l,?9

Hutchison and Pastor,34 and O'Reilly.31
1V.2. Sample_Preparation
IV.2.1. Preparation of the Spin-Standard

a.a' Diphenyl-fi-picryl hydrazyl (DPPH) crystals were
dissolved in carbon disulfide to a concentration of ~0.02M.
The solution was made under vacuum by distillation of carbon
disulfide onto the weighed DPPH crystals. The solution was

mixed and later transferred to the reference holder. The

86
alkali metal sample fit inside the reference cell so that
the sample was concentrically surrounded by the reference
solution, as shown in Figure 25. In this way, the field at

the sample should be nearly the same as that at the reference.
1V.2.2. Sample Preparation

Pre-determined amounts of alkali metal were isolated
by the method outlined in Chapter II. The metals were intro-
duced into the cell sidearm with the aid of heat-Shrink
tubing (see Appendix A). After distillation of the metal
through the sidearm, a predweighed amount of ammonia was
distilled into the cell. The ammonia was first distilled
into an intermediate storage vessel capable of withstanding
up to 25 atmospheres pressure. This vessel was weighed both
before and after the distillation of ammonia into the cell.
After thorough mixing, the solution was poured into the
quartz section to rinse the walls of the sample cell. After
the entire vessel has been thoroughly rinsed with the solu—
tion, a small portion of the solution was placed in the ESR
cell while most of the solution was poured away. The cell
was then sealed. Although only a small amount of solution
was used, the larger volume of solution allowed us to better

control the concentration.

1V.3. Instrumentation

 

A Varian V-4500 ESR spectrometer was utilized in this

work. The temperature was controlled by means of a V-4540

87

.hvsum mmm Mom mHmEmm HmumE UCm
Unmwcmum mama CDHB HMBOQ OOCmHmwmm .mm musmHm

ISL

 

 

 

 

m.O

 

 

 

 

\K \‘\ \7 ‘t\\ \ x\
///////3 ¥////
\\\,\ \ \x; n\ i\\

\

 

 

 

 

@

 

 

 

 

[J

 

 

 

 

 

 

 

 

 

./

 

 

<,._--___..-._.I'

88
Varian Variable Temperature Controller which forced a cold
stream of nitrogen both around the sample and through the
cavity. The temperature was measured with a thermocouple.
Unfortunately, the megahertz frequency range was not avail-
able and it was necessary to work in the X-band region (9.3
GHz). A 100 kHz modulation was used. A low power arm was
also not available at the time and the system had to be
Operated at microwave powers greater than 30 dB attenuation.
This resulted in serious problems with saturation.

Since sample positioning was important, we observed
the following procedure:

First, the reference vessel was placed in the cavity
and taped until it was stationary.

Second, a metal sample was placed inside the reference
cell. The Spectra were collected at different microwave
power settings, before changing the temperature. After a
complete set of spectra had been run (y§_both power and
temperature), the original temperature and power were re-set
to duplicate the original conditions. In this way, the
extent of decomposition could be checked.

The sample was removed from the cavity without disturb-
ing the position of the reference vessel. Another metal-
ammonia sample was then placed in the reference cell and the
above procedure was repeated. Fortunately, different metal-
ammonia samples, at the same concentration, had nearly the

same effect on the cavity, "Q".

89
The metal-ammonia samples were stored at liquid nitro-
gen temperatures before and between measurements to prevent

decomposition.

IV.4. Results and Discussion

 

IV.4.1. Line-Shape and Saturation Problems

 

Because of the very narrow linewidths and high elec-
trical conductivity of metal-ammonia solutions, two major
problems arise when an X-band ESR spectrometer is used to

measure the Spectra of these solutions.

IV.4.1.1. Line-shape Problems

 

The first problem is concerned with the line-shape and
linewidth of the spectrum. Spectrometers of this type have
modulation side-bands superimposed on the spectrum. Usually,
these side-bands are not a problem because they are buried
under the spectral peak. However, in the case of metal-
ammonia solutions, the true peak width is very narrow and
these modulation side-bans can significantly distort the
true linewidth and spectral shape. The peak shape can change
from the normal characteristic Lorentzian to a gaussian shape.

This can, of course, be easily checked.

IV.4.1.2. Saturation

 

As one increases the intensity of the incident micro-

wave power, the signal intensity of a peak will also increase.

90
However, at some point, further increase Of the incident
power will not increase the signal intensity proportionately.
Furthermore, if the power is turned even higher, the signal
will broaden and decrease in intensity. This phenomenon is
called saturation. Complete saturation occurs when the
population of the higher Spin state becomes equal to the
population of the lower Spin state. When saturation is a
problem, then, the population ratio has been shifted from
its natural equilibrium value. The theory of saturation has
been thoroughly discussed elsewhere1°° and will not be re-
peated here.

Saturation studies are very valuable since they are
capable of yielding the true linewidth at "zero" power, the
spin lattice relaxation time, T1, and, in certain Special
cases, the Spin-Spin relaxation time, T2. The recommended
procedures for obtaining such information are adequately

outlined by Poole.1°°

IV.4.2. Data Treatment

 

IV.4.2.1. Line-shape Analysis

 

The line-Shapes of sodium, potassium, rubidium and
cesium-ammonia samples were determined. The true Spectra
Should be Lorentzian in shape and should exhibit no gaus—
sian character. A tendency towards gaussian-shaped absorp—
tion peaks indicates that artificial broadening is present.

The comparison of gaussian and Lorentzian peaks is shown in

91
Figure 26. The peaks shown here are normalized to the same
amplitude and linewidth. The reader Should be aware that
the measured Spectrum will have different widths, heights
and areas, dependent upon the Shape. As can be seen, the
solvated electron Spectra are sometimes distorted from the
normal Lorentzian character.

As demonstrated in Figure 27, only cesium-ammonia
samples retain 100 percent Lorentzian character as a func-
tion of microwave power. The Spectra of the samples
which contain sodium, potassium, and rubidium change their
shape as the power is changed. It is fortunate, however,
that the three solutions behave Similarly as is demonstrated
in Figure 27. While absolute comparisons with the cesium
spectra cannot be made, the spectra of the three metal
samples may be compared with each other, since the shape-
change is the same for the three metal solutions. Further,
the ratio of the spectral area of sodium, rubidium, and
cesium absorption peaks to that of potassium may be studied
as a function of temperature and concentration, even though
absolute Spin concentrations may be invalid.

Because of adherence to the Lorentzian lineshape func-
tion,cesium solutions may be analyzed separately. The ESR
Spectra of cesium-ammonia solutions exhibit a larger line-
width than do the other metal samples and thus, were not

affected by the presence of the modulation Side-bands.

92

 

V CMHmmswm

.A

OD OIIIV CMHNDCOHOA Eoum Esuuowmm COHDOOHO Ompm>HOm on» CH mmmCmno .wm wustm

c N O N1 #1

_I i I l . A _ _._ J

 

 

 

 

 

 

 

 

 

 

93

 

 

 

 

4’
100 -
T ¢
¢ «b CS
50‘
0 .
.I a:
o\° I00—
3
A n
0
Q A R1)
50— 0N0
a
B K 9.
0»~—- l _L_.__,mwu 1 IV
16 20 24 28
P (dB)

Figure 27. Percent Lorentzian character in the solvated
electron band Shape in Na, K. Rb, and Cs
solutions.

94

IV.4.2.2. Saturation Studies

 

Saturation studies were performed on the various metal-
ammonia samples. AS is demonstrated for the cesium-ammonia
samples, the peak-to-peak linewidth (AHPP) was plotted ya
the spare root of the microwave power. Figure 28 shows that
the Spin standard, DPPH, is not saturated while the cesium-
ammonia sample undergoes typical saturation broadening.

This fact is also independently verified by a plot of the
peak height y§_the square root of the microwave power. Again,
as demonstrated in Figure 29, the Spectra of the cesium-
ammonia solution exhibits typical saturation behavior.

In order to determine the extent of saturation of the
sodium, potassium, rubidium and cesium-ammonia samples, a
plot of (Apr)2 y§_the square of the peak height (Y$)2, of
the reference absorption was made. This could have been
plotted as (AHPP)2 y§_the microwave power (P). However,
the microwave power incident on the sample was not accurately
determined. The square of the peak height of the reference
solution is directly porportional to the incident power (P)
and serves the purpose more accurately than does the reading
of the power meter. A typical plot is Shown in Figure 30.
The intercept of the least-squares fit is related to the
saturation factor. The intercept Of the line is (Ang)2
where the zero superscript denotes the linewidth at "zero"
microwave power. The saturation factor, 5, at each power is

iven b AHO AH 2.
g y ( pp/ pp)

95

 

 

 

 

A
DPPH
'I *'---X-><-x-*--"-"-"-—--x————-x————-x-—--
7 -
Cs
‘ /
O
O/
.3 /
O
o
0 0
a 7
i?‘ o,/°
4 //
O’O’O
./
.I
I I I __1 1 J l '
° -' -2 .3 .4 .5 .o .7 .3
P” (WWII IO)
Figure 28.

Variation of the linewidth of DPPH and of
the solvated electron in cesium solutions
with the square root of the microwave power.

96

 

n
20 - DPPH

15 a

 

 

 

I I -1--. .- l

.02 .04 .06
I/2 _ I/2
P ( w )
Figure 29. Plot of the peak-to-peak height for DPPH and

for the solvated electron in cesium solutions
ya the square root Of the microwave power.

97

 

 

 

 

 

III.
Cs 0.1M
IO - ////
O
oI 8
o
"" O
X
a -
0 —59°
0
a i 6
32
d
- O
4 O
/
00
I-.°/
/O
2 -_
- I I _,. _ _ |._.--__._..I
o I 2 3
)) —
(Y;D””)2(cmleos)

Figure 30. Plot of the square of the peak-to-peak width
of the solvated electron Signal in 0.1M cesium

solution yg the square of the amplitude of the
DPPH radical signal.

98

The saturation factor was determined at the various

power settings for each solution and temperature.

The ratio

of the spectral intensity of the metal-ammonia sample to

the spectral intensity of the center line of the DPPH

standard is plotted against the square root of the saturation

factor, S.

from:

where

and

C o

It is known100 that

AHZ

PP

(AH2

where the slope is the area ratio at "zero" power.

PP

O Ym

' Ym)DPPH

AH

2 0
PP

constant =

Ym

Spectral intensity

 

3w/f3 .
r- -I
2 .
= (A399 Ym)
2 C
(Apr Ym)DPPHJ

 

 

The spectral intensity or area was evaluated

(4.1)

81/2 (4.2)

0

An

intercept was allowed to float in the least squares fitting

but it was always within 2 standard deviations of zero.

The area ratios of metal sample to DPPH are tabulated

in Tables 6-9.

1V.4.3.

 

Results yglCation and Temperature

Since the cesium ammonia Spectra were not affected by

modulation Side-bands,

the calculations used to determine

the Spin susceptibility should be completely valid and,

therefore, this system warrants Special consideration.

99

 

 

 

Table 6. Corrected area ratio of the solvated electron ESR
signal in 0.02M Na, K, Rb, and Cs-ammonia solutions
‘to that of the center DPPH radical peak.
T(°C) AH 0 Area Ratio 0 Corrected
. pp Area Ratio
Na (0.02M)
-59 0.0840 0.0024 0.00606 0.00022 0.00292
-48 0.0692 0.0021 0.00885 0.00031 0.00427
-27 0.0486 0.0026 0.00888 0.00043 0.00428
- 8 0.0405 0.0072 0.01680 0.0054 0.00810
K (0.02M)
-59 0.0754 0.0054 0.00820 0.00093 0.00374
-48 0.0584 0.0038 0.0124 0.0013 0.00566
-27 0.0340 0.0036 0.0146 '0.0014 0.00667
- 8 0.0157 0.0067 0.0152 0.00046 0.00694
Rb
-59 0.0894 0.0012 0.00410 0.00052 0.00285
~48 0.0756 0.0013 0.00512 0.00075 0.00356
-27 0.0564 0.0052 0.00629 0.00068 0.00437
- 8 0.0492 0.0030 0.00896 0.0014 0.00622
CS
-59 0.109 0.0047 0.00522 0.00064 0.00464
-48 0.101 0.0021 0.00766 0.00048 0.00682
-27 0.0776 0.0073 0.00940 0.00084 0.00836
- 8 0.0719 0.0059 0.00878 0.0018 0.00781

 

100

 

 

 

Table 7. Corrected area ratio of the solvated electron ESR
si al in 0.04M Na, K, Rb, and Cs-ammonia solu-
'tions to that of the center DPPH radical peak.
T(°C) :AHPP 0 Area Ratio 0 Corrected
Area Ratio
Na (0.04m)
~59 0.0983 0.0050 0.00746 0.0013 0.00717
~48 0.0922 0.0049 0.00451 0.0023 0.00434
~27 0.0933 0.0062 0.0126 0.0049 0.0121
~ 8 0.0670 0.0058 0.00669 0.00044 0.00643
K
~59 0.0990 0.0052 0.00649 0.00075 0.00919
~27 0.0723 0.011 0.0101 3 ° 0.0017 0.0143
~ 8 0.0470 0.0070 0.0127 0.0041 0.0180
. Rb '
~59 0.0916 0.0048 0.00586 0.0012 0.00542
~48 0.0600 0.014 0.0104 0.0021 0.00961
~27 0.0632 0.0048 0.0111 0.0023 0.0103
- 8 0.0678 0.0085 0.0166 0.0054 0.0153
' CS
~59 0.137 0.0051 0.00516 0.00031 0.0105
~48 0.127 0.0029 0.00642 0.00072 0.0131
~27 0.122 0.0047 0.00878 0.00079 0.0179
- 8 0.130 0.0048 0.0428 0.0063 0.0873

 

101

 

 

 

Table it Corrected area ratio of the solvated electron ESR
signal in 0.06M Na, K, Rb, and CS~ammonia solu-
tions to that of the center DPPH radical peak.

0 . Corrected

T( C) Apr 0 Area Ratio 0 Area Ratio

Na (0.06M)
~59 0.0831 0.0025 0.00541 0.00027 0.00564
~48 0.0706 0.0032 0.00689 0.00081 0.00718
~27 0.0488 0.0039 0.00837 0.0010 0.00872
~ 8 0.0395 0.0037 0.00917 0.00067 0.00955
K
~59 0.0737 0.0069 0.00743 0.00056 0.00839
~48 0.0666 0.0058 0.00917 0.00067 0.0103
~27 0.0436 0.0088 0.110 0.0050 0.0124
Rb
~59 0.0927 0.0030 0.00653 0.000361 0.00653
~48 0.0710 0.010 0.00926 0.00102 0.00926
~27 0.0456 0.0042 0.0128 0.00152 0.0128
~ 8 0.0221 0.0075 0.0179 0.0053 0.0179
Cs
~59 0.139 0.0012 0.0102 0.00068 0.0115
~48 0.129 0.0027 0.0160 0.0012 0.0181
~27 0.109 0.0027 0.0258 0.0034 0.0291
~ 8 0.120 0.0028 0.0274 0.00362 0.0309

 

102

Table 59. Corrected area ratio of he solvated electron ESR
Signal in 0.10M Na, K, Rb, and Cs—ammonia solu-

tions to that of the center DPPH radical peak.

 

 

 

0 . Corrected
T( C) Apr 0 Area Ratio 0 Area Ratio
Na (0.1M)
-59 0.104 0.0036 0.0157 0.0040 0.0130
~48 0.0935 0.0047 0.0136 0.0015 0.0112
-27 0.0924 0.010 0.0194 0.0067 0.0160
- 8 0.0724 0.0094 0.0320 0.012 0.0264
K
-59 0.116 0.0044 0.0113 0.0027 0.00933
~48 0.0959 0.0075 0.00887 0.0010 0.00732
-27 0.0590 0.016 0.0345 0.0036 0.0285
- 8 0.0598 0.0067 0.0189 0.0039 0.0156
Rb
—59 0.125 0.0059 0.0142 0.0032 0.00685
~48 0.100 0.0055 0.0209 0.0044 0.0101
-40 0.0840 0.0063 0.0247 0.0036 0.0119
~28 0.107 0.011 0.0276 0.011 0.0133
- 8 0.0952 0.011 0.401 0.0092 0.0193
CS
-59 0.158 0.0036 0.0130 0.0010 0.0111
~48 0.154 0.0023 0.0232 0.0037 0.0199
~28 0.228 0.0063 0.0752 0.0081 0.0645
- 8 0.149 0.0042 0.0386 0.0037 0.0331

 

103

A logarithmic plot of the area ratio at "zero" power XE
1/T is shown in Figure 31 for cesium-ammonia solutions at
0.02, 0.04, 0.06, and 0.10M. The graph shows that Spin-
pairing increases at lower temperatures. Furthermore, there
is not much change in the Slope of the line which indicates
that in the concentration region of 0.02 to 0.10M, the
temperature dependence of the spin-pairing process is rela—
tively independent of concentration.

A logarithmic plot of the area ratio XE 1/T for sodium,
potassium, and rubidium-ammonia solutions is shown in Figure
32. As is readily seen, the three solutions behave nearly
identically. Because potassium solutions are known to have
paired spins in this concentration region, it is necessary
that sodium and rubidium solutions also Show the same
phenomenon and to about the same extent. Figure 33 Shows
that the metal-to-potassium ratio is independent of tempera-
ture. Again, it necessarily follows that sodium, rubidium,
and cesium Show the same temperature dependence of spin-
pairing as does potassium. These results are also tabulated
in Table 10.

If we normalize our relative potassium data to the
absolute data of Hutchison and Pastor34 at one temperature
and concentration, we find that our results corrolate well
with theirs. This is demonstrated by Figure 34. The
"scatter" at higher temperatures is due to the concentration
dependence of the Spin-pairing process and is systematic

rather than random.

104

 

 

 

 

 

I P.
A
c:
’6
5"
+3
(0
3..
m
(I)
:3
O!)
O
I»?!
-2
I 1
OJ; 0J6
l/T ( °K'l x 103 )
Figure 31. Plot of the logarithm of the area ratio of the

solvated electron signal in cesium solution to
that of the DPPH spin standard absorption versus
1 T.

105

 

 

 

 

 

#-
.-,’ 1) r—
-;".’ 1..
4:
.4,8 _
if r-
6
'1'!
.p
a! o __
‘4 -/ u
a, v
2: 4—
(U
4, r
DU A
3 _]_.() _ \g
A A x
' o
o
-2_3 _ A\
d: o
-2.1 -
X
A x
_ x
D O
—?.9 _ A X
0
. L l l l
3.5 14.0 24.5 5.0
l/T ( °K‘1x 103 )
Figure 32. Plot of the logarithm of the area ratio of the

solvated electron signal in sodium, potassium,
and rubidium solutions to that of the DPPH
spin standard absorption versus l/T.

106

 

.musumummamu msmum> Edammmuom

 

 

 

 

 

mo umnu on Edamoo cam .ESaUflQsH .Esflvom mo oflumu mmum mo poam .mm musmfim
A x. V a
0mm 0mm 05m 0mm oom
_ . _ _ To
m
a pm
0 oz
0
I o; N
X
0
O
0 0 mo

 

 

107

Table 10. Area ratio of the solvated electron ESR signal
in liquid ammonia for-Na, Rb, and Cs to that of
K at the same concentrations and temperatures.

 

 

Area Ratio

 

 

w 0

Concentration T( C) Na/K Rb/k CS7K
0.02M -59 0.7807 0.7620 1.241
-48 0.7544 0.6290 1.205
—27 0.6417 0.6552 1.253
- 8 1.167 0.8960 1.125

0.04M -59 0.7801 0.590 1.14

-43 __- __- _-_

~27 0.846 0.720 1.25

- 8 0.357 0.850 4.85

0.06M -59 —-— --- ---
-48 0.672 04778 1.370
~27 0.697 0.899 1.757

- 8 0.703 1.03 2.35
0.10M -59 1.391 0.734 1.190
-48 1.530 1.626 2.719
-27 0.514 0.467 2.263
- 8 1.692 1.237 2.122

 

108

.mump Esfimmmuom pmuwamEHOS H50 .2 “Hoummm 6cm comanousm
MO 0080 Edflmmmuom .8 "B\H mo coauossm m mm Hoummm cam couscousm mo
mumo may SuHB mumo Edwmmmuom Hso mo mwum m>flumamn 0:» mo cowaummeoo .fim musmflm

$2.; :—

 

mw. mm Wm
_ l _ _ _ _
1m.
2 -
a . 1
x
g l
/ v
u w m
2 o
x 8
w 8..
a 8 a
8
8 / , w:
8 /
8.L v

 

 

 

109

If we combine our potassium data with those of De
Mortier99 and of Hutchison and Pastor,34 we find that our
results correspond well with those of Hutchison and Pastor,
as shown in Figure 35.

In summary, we find 22_evidence that the extent of
Spin-pairing is strongly dependent upon the cation, in
agreement with the results of Hutchison and Pastor, but in
disagreement with the results of DeMortier, DeBacker and
LePoutre.

Finally, it is interesting to note that in this concen-
'tration region, the extent of Spin-pairing seems to change
very little with concentration. Perhaps the most logical
eXplanation is that this results from experimental error.
However, it is possible that an onset concentration is re-
quired before spin-pairing occurs. This could, if true,
reduce the discrepancy between the data of DeMortier and
coworkers and those obtained by us and by Hutchison and
Pastor. It would be very worthwhile to perform a careful
study of the concentration dependence of the spin suscepti-
bility in the concentration range of 0.001 to 0.1M, with
equipment that eliminates the problems which we had with

X-band ESR measurements in this region.

110

.mme 02
“mumo um H50 .0 “Hoummm cam somw£055m MO 5550 52 .x “Hoummm 05m

50mH£ousm mo mumo M .< “mumo E5HUOm m.H0HuH0200 .o umCOHu5Hom E5Hmmmuom
Ucm E5HUOm mo COHuMHucmocou 0:5 m5mH0> coaumHucwocoo cflmm 055 m0 uOHm

As: my
Op
fill

H50 .4

7m: 0m:

 

 

 

l .H .m
0
0
0
\ I 0..
.6 ml
00 N
4400\0 d
om
0.\0
4v .4 0 0
0 XXX. «0*. \
4d 0000 :1 0..
4 CG O “I.
. Kg «0
d \\
i \

 

 

 

.mm 0H5mflm

V. SOLID-STATE STUDIES

V.1. Introduction

 

V.1.1. Historical

 

Metallic solids which contain ammonia have been studied
for some time. Compounds such as Ba(NH3)3 and Li(NH3)4 have
been prepared by Sienko and coworkers.72 Pure solid salts
of M-, however, have not been isolated. The evaporation
of solvent from metal—solutions yields only precipitated

metal.

V.1.2. Use of ComplexingyAgents

 

It was thought that macrobicyclic complexing agents
could be used to prepare a salt of M-. It is known that
the cation is trapped inside the cage of the complexing
agent in solution. If the rate of release of the cation
from the ligand is slow enough, it might be possible to
precipitate a salt of the stoichiometry [M+ C]M- where C
represents the complexing agent.

Various attempts were made by Lok82 with the use of
crown. Several conclusions were drawn on the blue-colored
substances which were obtained, but for the most part, these

solids need yet to be characterized.

111

112
Similar attempts with 2,2,2-crypt were also made. A
gold-colored powder was observed to precipitate from sodium

solutions in ethylamine in the presence of 2,2,2-crypt.

V.1.3. Preparation of Na2C18H3603N2

 

The extremely low solubility of the alkali metals in
ethylamine at low temperatures has been known for some time.56
The solubilities of metals in the presence of complexing
agents in this solvent were checked by dissolving both the
complexing agent and the metal. The sample was then decom-
posed and the amount of metal hydroxide was titrated with
standard hydrochloric acid solutions. Concentrations of
metal as high as 0.4M were obtained in this way.

The new gold-colored substance was prepared by allowing
a solution of 2,2,2-crypt in ethylamine to contact a sodium
mirror. In the presence of excess sodium, a deep blue solu~
tion forms. If the solution is thoroughly mixed, removed
from the sodium mirror, and then cooled, the new substance
precipitates out of solution. The new material is easily
dissolved by allowing the temperature to rise. It seems
that the solution is saturated at high temperatures, and
upon cooling, crystallization occurs. The formation of the
material is also reversible, i.e., it may be recrystalliZed
as many times as desired in ethylamine.

Oddly enough, Slow evaporation of ethylamine apparently

gives only co-precipitation of sodium metal with 2,2,2—crypt.

113

V.1.4. Appearance and Stability

The color of the precipitated powder changes reversibly
from gold at ~1900 to bronze at 40°. The powder, obtained
by rapid cooling shows an X~ray diffraction powder pattern.
Thus, the substance is actually micro-crystalline. The pre-
cipitate,after it was washed with nfhexane,was stable in_

vacuo for days at room temperature and below.

V.1.5. Stoichiometry

 

Several tests were performed to check the composition
of the sample.

Flame emission tests were performed and these indicated
that there is more than one sodium atom but less than two
sodium atoms per molecule, (1.67 Na/molecule were obtained
if one assumes one complexing agent molecule per molecule
of substance). Although the results favor two sodium atoms
per molecule, the test was inconclusive in that only a 3.6
mg sample was used. Further tests by Spang Microanalytical
Laboratories101 Showed that the compound has the composition
of two sodium atoms per molecule of 2,2,2-crypt. The per-
centages obtained by elemental analysis were within experi-
mental error of the values predicted based on a stoichiometry
of N32C1833606N2- The 2,2,2-crypt was Shown to be intact
by mass Spectral studies. The mass spectrum of the gold-
colored compound Shows a parent peak at m/e = 376, which

indicates that the complexing agent is not altered, since

114
the identical Spectrum is observed in the case of the free

2,2,2-crypt.
V.1.6. Conductivity

Because of the highly charcteristic metallic Shine,
conductivity measurements were performed on the compound.

A wire press was constructed for the purpose of extruding
wires or rods of the material. It was possible to extrude
rods,which indicates that the material is either soft and
malleable or else can be readily packed into compressed form.
The extruded rods were approximately 2.5 to 5 cm long and
~o0.2 mm in diameter.

The press was specially built to be vacuum tight. The
rod was extruded into an evacuated tube, after the press
had been filled in vacuum through a sidearm.

The rods were placed under a protective layer of pf
hexane, and conductivity measurements were made. However,
the resistance of the extruded rod was much greater than
10,000 ohms. Similar results were also obtained with the
extruded rod placed under liquid nitrogen. Recent conduc-
tivity measurements by Lok82 with packed powders Show that
the resistance is very high (~109 ohms) and that this new

compound is probably a semi-conductor.
v.2. Single Crystal Growing

It is well known among crystallographers that the pro-

cess of growing Single crystals involves more mystery and

115

art than scientific technique. Various methods of crystal
growth were tried. The well-known technique of Slow evapora-
tion of solvent was attempted but we observed only co-precipi-
tation of sodium with the complexing agent.

We Slowly cooled a solution of sodium-cryptate in
ethylamine overnight. After the solution was cooled from
15 to ~28°, thin, hexagonal gold-colored platelets were
formed on the side of the vessel. Better results were ob-
tained by using a tube which had flat faces, rather than
a circular cross-section. Several crystals were photographed
through a microscope and one such crystal is Shown in Figure
36. It is interesting to note the different colors which

reflect from the surface of the crystal.
v.3. X-ray Structure

Although we had shown the stoichiometry to be
Na2C18H3606N2, we did not know the structure of the com~
pound. We believed the compound to be a salt of Na-.
However, it was entirely possible that the two sodium
Species were both outside the complexing agent, and as such
both could be neutral atoms. It was for this reason that

X-ray crystallographic measurements were performed.
V.3.1. Isolation of a Single Crystal

The isolation of a Single crystal seemed to be a very
Simple task, but proved to be a most tedious and frustrating

job. In addition to the difficulties inherent in working

116

.Hw30m ow u coaumOHMHcmmE
k«zoOonmmHonmz mo gm5Hm0H0HE05onm

on 0H50Hm

 

 

 

 

 

 

 

 

 

 

117

with such a reactive compound, the crystals for X-ray mea-
surements need to be thick in all three dimensions. Poor
diffraction is obtained from crystals which are large in
two dimensions and very thin in the third dimension.

Special X-ray capillary tubes were purchased from the
Uni~Mex Corp.102 and were cleaned by the usual technique.
The Special soft (lead free) glass was only 0.01 mm in wall
thickness to permit easy penetration by the X-ray beam.

The crystals had to be transferred to the capillaries inside
of an inert atmosphere box103 in order to prevent decompo-
sition. In addition, the crystal had to be mounted rigidly
on the side of the capillary wall so that it would not slide
as the capillary tube was rotated in various directions.

These restrictions were especially difficult to over~
come. However, through a long process of trial and error,
success was finally achieved.

The crystals were transferred into the X-ray capillary
tubes in an inert atmosphere glove box. Purified paraffin
oil was used to hold the crystal onto the wall by surface
tension forces. The tubes were first stoppered with the
inert gas inside the tube and then were removed from the
glove box. The tube was then evacuated and sealed. The.
crystals were stored at dry ice temperature (~78°) prior
to their use. The crystals were then checked for Size with
a microscope. It was very difficult to judge the dimensions
of the crystals in the inert atmosphere box, but the micro-

scope readily revealed their dimensions. After the

118
isolation of a suitably Sized crystal, it was placed on the
manual diffractometer to check its singularity. If the
crystal proved to be Singular, it was then checked for dif-
fraction quality. If the quality of the crystal was accept-
able, it was used for data collection. Needless to say,

very few crystals met all of the above requirements.

V.3.2. Background to X-ray Diffraction

 

W. L. Bragg stated the necessary conditions for diffrac-

tion in his famous equation,

n1 = 2d Sin 6 (5.1)
in which,
A = wavelength of incident radiation
d a distance between planes
9 = angle of incidence
and n = an integer.

There is, however, an alternate description of diffrac—
tion which is in terms of the reciprocal lattice.
If 57 ST and E> represent the direct Space unit cell

. ~> - ->
axes, then conSider a*, 5*, and c* where

'5 "5* =5. 5* =7? - "5* = 1 (5.2)

and
--5* =—5 . c* '75 - =5 - c* = 5 -—5* z-E .'fi* = 0.
(5.3)

A vector in the reciprocal lattice is then defined by,

119
-§*[hkfl] = h‘S* + k‘5* + z‘E* (5.4)

where h, k, and z are integers.

It can be shown104 that each point [hkz] in the recipro-
cal lattice represents the corresponding [hkz] planes in
the real space lattice.

Each point in the reciprocal lattice also corresponds
to a possible reflection from the crystal, since the integer
values of h, k, and i can be Shown104 to impose diffrac-
tion conditions. The reciprocal lattice is also consider—
ably simpler to visualize in that a set of points replaces

sets of planes.

V.3.3. gpace Group Determination*

 

There are 230 possible space groups in which a material
may crystallize. The determination of the space group for
a particular system aids in the determination of the struc-
ture. Each of the 230 Space groups can be classified into
one of the eleven Laue groups based on symmetry and equiva-
lent positions. Usually one assumes the lowest symmetry
system (triclinic) and works up to higher symmetry elements
if present.

A weighted intensity plot of the reciprocal lattice re—
vealed the existence of a C3 axis, and 3 C2 axes. No
systematic absences were seen in the system. Further studies

on the assignment of indices led us to the Laue group, R 3m.

 

* ,
The author is grateful to Dr. B. L. Barnett for his aid and
advice in the determination of the space group and the crys—
tal structure.

120
From this Laue system, three space groups were consistent
with our observations, namely the R 3m, R 3m and R 32.
These 3 space groups are not distinguishable by diffraction
symmetry.105 Solution of the structure later showed the
space group to be R32.

The three space groups all belong to the rhombohedral
system. However, the rhombohedral lattice is one of the
most difficult ones to visualize. Consequently a hexagonal
lattice iS often used to define the system. The relation-
ship of a rhombohedral cell to a hexagonal cell is Shown in
Figure 37. It should be noted that there are two possible
orientations. As is also demonstrated, the axes defined in
the hexagonal system are not the same as those in the rhombo-
hedral system. The introduction of the hexagonal lattice,
also requires a different reflection indexing scheme. Only

reflections of the type
-h + k + z = 3n (5.5)

where n = 0, 1, 2, .... etc,

are observed in the new indexing scheme.

V.3.4. Data Collection

 

Intensities were collected on a Picker Automatic 4-
circle diffractometer.106 Special data collection routines
allowed us to collect all the data in less than seven hours.
The moving crystal—stationary counter method was used in

which the crystal is set in diffracting conditions and ¢,

121

 

 

 

 

 

\0 so
.. lllllllllllllllllll «r...
.m
a m
w:--------. ....... a
.1, (0 1
—~ 0 / .—
64.; &\/ ..»
”Vol / .7 i-
.11 .. (5
I l I I I zl/Il/l I "Vo\|\\\\. .
__ , /\.\(V):.. _..
.u // \ano .:_
.. ,1)\: .fi
— 1 .
.m . .-
. ...
o~0ll|0llllllll|llllkpim
4

(0

(m

 

 

 

 

 

 

 

I.) 0
O
o
.70 n
O
o
0
'0. .’ o '0. 0’
O O O O O O
o 8 I 5 o
O I
u 0... ‘ a
O O 0
II”
0
0.. 0 o» O 8 0.. o
C O O O
o 0' /—'.AV
0 o O A
I
/
I
z .
c. n -0 o
I Oll ....... .I C I
\m.
\
x
0 x / n.
0 Cl 0 O
' - 0 / ‘0
.’ I: .l 0 D I. D 0
0 0
o 8 ' - o
O 0

 

 

gonal

Unit cells in the rhombohedral and hexa
1attices.1°5

Figure 37.

122
26, and X are held constant. The intensity is then counted
at the various w angles to which the crystal is moved.

COpper radiation was utilized and data were collected
for 26 values up to 120°. Because of the high symmetry
demanded by the space group, only 413 independent reflections
were counted. Absorption, Lorentz, and polarization cor—
rections as well as lack of balance corrections were applied
to obtain more accurate data. Three reflections were moni-
tored throughout the course of the data collection. These
showed that an isotrOpic decay of 6 percent had occurred
during data collection. Consequently, a time-dependent decay
correction was applied as a function of the reflection

sequence number.

4

V.3.5. Fundamentals of Crystal Structure Analysis

 

It is known that the intensity is related to the struc-

ture factor by
I(h k z) a [F(h k 2)]2 . (5.6)

Furthermore, the structure factors can be related to the

diffracting atoms in the unit cell by

N
F(hk£) = 2 f. exp[25i(hx. + ky. + 22.)}. (5.7)
j:1 J _ 3 3 J _
The summation is taken over all the atoms (j = 1 to N) in

the unit cell. The term, fj’ is the scattering factor of
atom j.

If the atom is spherically symmetric and is at rest,

123

the scattering factor may be defined as

(1)

f0 = ID 4wr2 U(r) sin14vr(sin 9)]1]

4wr(sin 6)/A

 

dr (5.8)

U(r) represents the radial distribution function for the
electrons in the atom.

It is well known,however, that atoms are not at rest.
Consequently, the thermal motion of the atom spreads the
electron distribution over a larger sphere. And, as a
result, a temperature dependent scattering factor may be
used.

It can be shown107 that the electron density [p(xyz)]

v

is the Fourier transform of the structure factors, i.e.,

<hd

p(xyz) = Z Z Z (F(hk£))exp{~27i(hx + ky + 32)] (5.9)
h k z _ .
Once the structure factors F(hk£) are known, the elec-
tron density can be easily calculated. However, the struc~
ture factors are complex [see eq. (5.7)]. While the mag-
nitude of the structure factors are known, the phases are
not. Once the phases are resolved, solution of the crystal
structure is straightforward. The deduction of the phases

is called the phase problem of x~ray crystallography.
V.3.6. Solution of the Structure

The Patterson function was used since no phase informa-
tion is required to construct this function. The Patterson

function is given as

124

<:|H

P(uvw) = z i Z |F(hk£)|2exp[~27i(hu + kv + £w)) (5.10)
h Z .

The peaks in this function correspond to vectors between
atoms. The largest peaks in the Patterson map usually cor~
respond to the vectors between the heaviest atoms. [It is
possible, however, for several interatomic vectors to add
up to give large peaks.]

The Patterson function is difficult to interpret in
that many peaks are obtained. For example, if there are
30 atoms in the unit cell, then 30 x 29 = 870 peaks are ob-
tained. Needless to say, the assignment of these peaks to
the correct atomic vectors is not trivial.

The structure of Na2C13H3506N2 was solved by examination
of the Patterson map. Because of the high symmetry (required
by the Space group) only seven independent atoms had to be
located. Careful examination of the Patterson map revealed
the positions of the seven atoms. The positions were then

assigned to the appropriate atoms.

V.3.7. Refinement of the Structure

 

The R value is the criterion used for testing the

completeness of a structure. The R factor is defined as

R = g S: = 2| 1:380} chi i. (5.11)

where F0 and PC represent the observed and calculated

structure factors, respectively. The R value, or residual

125
index, is not a perfect guide, but it has been very widely
used to determine the correctness of fit.

In this case, the R factor was reduced to 0.1463 by
isotropic refinement of the parameters. The isotropic
refinement uses four parameters for each atom. In addition
to the three coordinates (x,y,z), there is the thermal
motion factor, which tends to spread the charge. Thermal
factors are important in that they can reveal certain char-
acteristics of the nuclei. The thermal factor depends not
only on the motion of the atom, but is also strongly af-
fected by the charge distribution about the atom.

The R factor was further reduced to 0.1070 by
anisotrOpic refinement of the parameters. Anisotropic re-
finement assumes Six parameters to replace the one thermal
factor. In anisotrOpic refinement, the atomic motion is
not limited to a simple Sphere and indeed, the six param-
eters define a three dimensional ellipsoid.

All refinement was performed by the use of the block
diagonal approximation to least squares.

After the addition of the hydrogens, another cycle of
anisotropic refinement was allowed. The final residual
index was R = 0.0850 or 8.5%. This value of R is based
on the scattering factor curves of atomic sodium. There
is no known scattering factor curve for the sodium anion.

The entire refinement was based on 306 independent, non—

zero observations (out of 411 possible observations). The

126
zero level was set at two standard deviations above the
background count.
Table 11 lists the cell eimensions and appropriate

lattice information.
V.3.8. Results

The structure factors, both observed and calculated,
are tabulated in Tables 12 and 13.

Table 14 lists the relative coordinates of the independ-
ent atoms in the cell. The thermal parameters are also in-
cluded. Because of the special positions of several atoms
in the cell, limitations were placed on some coordinates
and thermal parameters as has been previously outlined by
Peterse and Palm.“8 The deviations also noted in Table 14
are the ideal standard deviations.

Data were actually collected from two different crystals.
The first crystal Showed very poor crystal quality. The w
scan of several low order peaks revealed some peak widths
to be nearly 2.20, from background to background. The second
crystal revealed better mosaic crystal quality Since w
scans Showed peaks to be ~a1.3° wide. The second crystal
also revealed more countable reflections at higher values of
Sin 6/1. This is also indicative of higher crystal quality

Even though there was a marked improvement in crystal
quality, the refinement of both data sets led to nearly
identical positions and thermal motion parameters.

It seems, therefore, that the large thermal vibrations

126a
may be the result of increased temperature, since the crys-
tals were formed at low temperature, and the decomposition
point (where the trapped sodium Species is released from the
crypt) is only 83°. There are, of course, a number of other
effects which could also account for the large thermal param-
eters (e.g. crystal shape, absorption, capillary, etc.).

Furthermore, full matrix least squares have recently
been performed on the data. However, the results are nearly
identical to those obtained from the block diagonal approxi-
mation.

Whatever the cause of the large thermal motion is, it
is also reflected to some degree in the iodide structure1°9
where some of the thermal motions, especially for the C3
atoms are very high.

The positions of the other atoms in the cell are dic-
tated by the symmetry of the space group. The symmetry
operations and positions of all the atoms are listed in
Table 15. Figure 38 shows the apprOpriate numbering scheme

of the atoms. The values of interatomic bond distances and

127

 

 

(Table 11. Crystal lattice parameters.
a z b = 8.831 0 = 0.001
(hexagonal axes)
c = 29.258 0 - 0.004
a = fi = 90° y = 120°
Crystal system = Rhombohedral
Diffraction symmetry 2 R 8 m
Space group = R 3 2 (no. 155)
pcalc = 1.064 g/cc : Z = 3
M.W. = 422
v = 1976.17 8
1 = 1.5418 R

 

128

Table 12. List of structure factors; * denotes unobserved
reflections; unobserved reflections are those
less than 14 counts or are within 2 standard
deviations of the background count.

 

 

K L FOR5 FCAL K L F089 FCAL K L F089 FCAL K L F085 FCAL
9’8 H = 0 00° 4 26 15 7' 2 2 P64 226 8 P 14 146
o 1 14P1 1641 4 29 14 110 P 9 178 118 404 H s P 9.4
o a 174 187 5 1 78 P7 2 8 P64 286 0 2 1177 1240
0 9 746 760 5 4 186 155 2 11 184 218 0 5 909 911
0 1? 122 P82 5 7 181 176 2 14 P64 P64 0 8 012 371
o 18 898 865 5 10 P70 P82 2 17 44 76 0 11 584 58?
o 18 P56 P10 5 13 71 76 2 PO 170 186 0 14 611 602
o 71 74 10 5 16 56 78 2 P1 130 144 0 17 1P4 11?
0 P4 111 103 5 19 15 12' 2 P6 51 57 0 20 17 36
O 27 157 118 5 PP 15 24' 2 P9 17 P19 0 P3 117 124
0 10 71 99 5 25 14 0° 3 1 180 188 0 26 .64 60
1 2 748 647 6 1 65 82 3 4 49? 480 0 29 18 210
1 5 1997 1450 6 6 15 90 3 7 134 108 1 1 441 371
1 8 491 518 6 ° 15 23* 3 10 190 199 1 4 878 831
1 11 168 113 6 12 47 65 3 13 121 114 1 7 52 55
1 14 ’1? P11 6 15 68 66 3 16 109 124 1 10 461 467
1 17 108 1P3 6 18 15 89 3 19 78 8? 1 13 P60 267
1 P0 297 272 6 21 11 40 1 P2 59 63 1 16 83 77
1 P3 70 6S 7 2 159 165 3 P5 56 69 1 19 84 91
1 P6 91 81 7 5 16 17' 3 P8 16 150 1 2? 87 78
1 29 4? 45 7 8 16 9* 4 0 P07 168 1 25 18 329
1 32 15 109 7 11 15 160 4 3 104 96 1 28 17 50
2 1 1076 1015 7 14 15 7O 4 6 148 311 1 31 36 36
P 4 984 942 7 17 14 328 4 9 181 167 2 0 1077 1011
2 7 P30 P15 8 1 15 7O 4 12 703 209 2 3 4P4 418
2 10 115 166 8 4 14 28* 4 15 84 89 2 6 65 81
2 13 145 165 8 7 14 9* 4 18 77 82 P 9 POO 229
2 16 210 PPI 8 10 PB 40 4 PI 17 19° 2 12 176 198
2 19 177 146 900 H = 1 444 4 P4 15 25 P 15 140 161
2 P? P78 785 o 1 1407 1513 4 P7 15 13* 2 18 142 150
2‘ 25° 17 14 0 4 04 112 S P 148 146 2 21 36 16
2 28. 16 P99 0 7 113 84 S 5 125 1P4 2 24 143 158
2 31 15 79 0 10 917 916 5 8 110 109 P 27 33 39
3 3 P82 255 0 13 44 19 5 11 88 9? 2 30 13 3?
3 6 11? 305 0 16 12 74 5 14 61 64 3 2 522 510
3 9 _101 46 0 19 61 55 S 17 17 15° 3 5 106 100
3 12 231 PPR 0 22 74 70 5 P0 68 86 3 8 188 175
3 15 111 105 0 25 18 6' 5 P3 40 48 1 11 15 31
3 18 120 142 0 28 93 89 5 1 147 140 1 14 qn 7g
3 21 54 67 0 31 PS 23 6 4 76 89 1 17 82 86
3 24 104 115 1 0 961 970 6 7 78 71 1 20 38 41
3 27 16 14° 1 3 1068 10P6 6 10 81 78 3 P3 57 64
3 10 15 16° 1 6 155 146 6 11 17 279 3 26 11 40
4 2 80 71 1 9 301 283 6 16 16 179 4 l 95 81
4 5 P51 251 1 12 4P0 411 6 19 15 13a 4 4 131 121
4 8 178 417 1 15 154 1?] 7 ‘0 47 49 4 7 81 87
4 11 91 94 1 18 318 131 7 3 51 62 4 10 135 157
4 14 256 758 1 21 147 140 7 6 51 57 4 11 P5 41
4 17 4° 52 1 24 41 31 7 9 39 33 4 16 66 75
4 P0 15 46° 1 27 18 14° 7 IP 45 61 4 19 87 98
4 23 16 18° 1 30 17 28' 7 15 14 16° 4 22 9? 9P

129

Table 13. List of structure factors (cont'd); * denotes
unobserved reflections; unobserved reflections

are those less than 14 counts or are within 2

standard deviations of the background count.

 

 

K L F085 FCAL K L F085 FCAL K L F085 FChL K L F089 FCAL
4 P5 15 7° 1 3 61 50 2 11 16P 17? 1 5 P7 14
5 o 14? 115 1 6 88 48 2 14 117 118 1 8 53 69
5 3 74 87 1 9 105 118 2 17 45 40 3 11 30 41
5 6 81 77 3 12 1P3 145 2 20 61 59 3 14 16 140
5 9 1? 19 1 15 50 51 2 P1 16 P80 1 17 15 8°
5 12 87 93 1 18 74 8P 3 1 82 82 4 1 43 70
5 15 17 10° 3 21 49 s4 3 4 75 61 4 4 17 P99
5 18 18 57 1 P4 16 100 1 7 62 65 4 7 16 189
5 P1 41 62 4 2 57 79 3 10 152 189 4 10 15 29°
6 2 127 127 4 S 159 167 3 I3 56 79 4 11 14 P09
6 5 3‘3 46 4 R 38 5| 3 16 63 79 one H 8 6 999
6 8 68 73 4 11 49 49 1 19 P9 44 0 0 144 111
6 11 66 72 4 14 50 47 3 P2 15 11» o 1 120 107
6 14 P9 43 4 17 44 52 4 0 P14 208 o 6 60 ‘ 41
6 17 14 7o 4 20 72 87 4 1 84 71 0 9 P7 10
7 1 15 4o 4 21 P4 P4 4 6 51 40 0 12 49 35
7 4 15 16» 5 1 118 118 4 9 14 37 0 15 46 47
7 7 15 PRO 5 4 92 75 4 1? 11 44 0 18 17 15°
7 10 40 52 5 7 PS P4 4 15 16 P49 0 21 16 89
98° H = 3 99° 5 10 74 78 4 18 15 11° 1 2 116 129
o o 785 428 5 11 P5 17 5 P 78 101 1 5 16 37
0 3 585 568 5 16 15 3° 5 5 31 16 1 8 76 76
O 6 423 374 5 19 14 11“ 5 8 16 40° 1 11 50 42
0 9 P41 197 6 0 19 46 5 11 15 P19 1 14 P6 P5
O 1? 662 610 6 3 43 55 404 H = 5 99° 1 17 16 1°9
0 15 64 51 6 6 40 54 o P 161 162 1 20 15 17°
0 18 24 22 6 9 P7 40 0 5 131 151 2 1 18 180
o 21 41 56 6 12 17 49 0 8 77 78 P 4 17 21°
0 P4 31 28 966 H = 4 008 0 11 06 101 P 7 17 198
0 P7 61 63 0 1 276 3P0 0 14 71 87 P 10 35 SI
0 10 15 12 0 4 283 104 0 17 17 429 P 11 16 198
1 P 551 510 0 7 473 417 0 P0 111 14? P .16 15 180
1 5 1P5 115 0 10 314 101 0 P1 17 339 1 0 19 46
1 8 171 159 0 13 179 180 0 P6 16 14» 1 1 16 30°
1 11 160 148 0 16 68 77 1 1 201 195 3 6 16 119
l 14 118 123 0 19 195 188 1 4 185 116 3 9 15 14°
1 17 180 192 0 2? 275 P7? 1 '7 65 56 1 1? 14 21°
1 20 111 129 0 P5 18 4“ 1 10 148 166 “9" 9 = 7 898
1 P1 156 147 0 28 17 27° 1 11 56 51 0 1 53 71
1 P6 17 10° 1 0 P16 168 1 16 62 I? n 4 11 46
1 79 16 In” 1 3 59 86 1 1° 47 51 0 7 30 34
2 1 47? 415 1 6 61 41 1 PP 16 18° 0 10 65 69
2 4 170 150 1 9 97 85 1 P5 15 106 0 11 16 1o
2 7 94 113 1 12 P42 P40 P 0 14? 115 0 16 106 7
2 10 171 172 1 15 P5 31 2 1 17 190 0 19 15 7o
2 13 15 26° 1 18 118 112 P 6 73 106 1 0 48 49
2 16 76 89 1 21 149 148 2 9 84 119 1 3 17 229
2 19 15 55 1 24 45 40 2 12 97 128 l 6 16 47°
2 PP 97 110 1 27 24 P9 2 15 17 144 1 9 16 41°
2 P5 16 22° P 2 208 196 P 18 27 41 1 1? 15 390
P 28 15 110 P 5 96 101 P P1 15 I7» 1 15 14 19°
3 0 190 P]? 2 “ P48 P69 3 P 06 101 P P 16 8°

130

 

 

 

 

 

 

 

 

0H00.0 0000.0 0000.0 $000.0 H000.0 0000.0 $000.0 0000.0 0H00.0 MUD
0000.0 0$00.01 0000.0 0$00.0 0000.0 0000.0 00H0.0 0000.0 $500.0 «0
$H00.0 0000.0 0000.0 H000.0 0000.0 0000.0 0000.0 5000.0 $H00.0 «00
0000.0 0H00.0| 0Hb0.0 0000.0 5000.0 0000.0 0000.0 0HBH.0 0$H0.0 NU
0H00.0 0000.0 $000.0 H000.0 $000.0 5000.0 0000.0 0$00.0 $H00.0 HUD
0000.0 $H00.01 00H0.0 0H00.0 H550.0 0000.0 b00H.0 H050.0 005H.0 HD
111 H000.0 0H00.0 H000.0 0000.0 $H00.0 0000.0 HH00.0 0000.0 >XOD
0000.0 0000.0 0HH0.0 $000.0 0000.0 0h00.0 0$00.0 0HFH.0 0000.0 0X0
111 Ill 0H00.0 0000.0 b000.0 5000.0 $000.0 III III “H20
0000.0 0000.0 00H0.0 0H00.0 0H00.0 0H00.0 0H00.0 0000.0 0000.0 UflZ
III III 0H00.0 0000.0 $000.0 $000.0 III III III 0 MZU
0000.0 0000.0 00H0.0 $000.0 0000.0 0000.0 0000.0 0000.0 0000.0 0 m2
111 Ill h000.0 H000.0 0H00.0 0H00.0 III III 111 H mZO
0000.0 0000.0 00H0.0 0H00.0 0$00.0 0$00.0 0000.0 0000.0 0000.0 H mZ
00® 0H® NHM 00m 00W HHM. N W x EOHAN

 

 

“flame 00:: man c0

 

.Umumasnmu omam mum mCOHumH>mU Unmvcwum mnu

mEOum ucmwcmmmvcfl msu mo mumuwfimumm HmEHmnu Ucm mcofluflmom

.44 64066

131

 

 

 

 

 

 

 

 

 

Table 15. Centers, symmetry Operations and the positions of

some atoms in the lattice.

Centers: ('0 0 O; %, g, g; g. é. é‘)

A = x,y,z; C = )7, x-y, z; E = y-x, i, z

B = y,x,§; D a: i, y-x, E: F = x—y, 37, 2
Atom x y 2
Na 1 0.0000 0.0000 0.0000
ClA 0.1796 0.0781 0.1067
ClB 0.0781 0.1796 -0.1067
C1C -0.0781 0.1015 0.1067
C1D -0.1796 -0.1015 —0.1067
ClE —0.1015 -0.1796 0.1067
C1F 0.1015 -0.0781 -0.1067
C2A 0.3143 0.1715 0.0806
C2B 0.1715 0.3143 -0.0806
C2C -0.1715 0.1428 0.0806
C2D -0.3143 -0.1428 -0.0806
C2E -0.1428 -0.3143 0.0806
C2F 0.1428 -0.1715 -0.0806
C3A 0.3874 0.3398 0.0190
C3B 0.3398 0.3874 -0.0190
C3C -0.3398 0.0486 0.0190
C3D -0.3874 -0.0476 -0.0190
C3E -0.0476 -0.3874 0.0190
C3F 0.0476 -0.3398 -0.0190
0 A 0.3055 0.1713 0.0342
0 B 0.1713 0.3055 —0.0342
0 C -0.1713 0.1342 0.0342
0 D -0.3055 -0.1342 -0.0342
0 E -0.1342 -0.3055 0.0342
0 F 0.1342 -O.1713 -0.0342
N A 0.0000 0.0000 0.0918
N B 0.0000 0.0000 -0.0918
Na2P 0.3333 0.6667 0.1667
NaZQ 0.6667 0.3333 -0.1667
NaZR -0.6667 -0.3333 0.1667
NaZS -0.3333 0.3333 -0.1667
Na2T 0.3333 -0.3333 0.1667
Na2U -0.3333 -0.6667 -0.1667

 

132

 

 

NA2T

 

NA2U

Figure 38. Labelling scheme of atoms in the hexagonal
lattice-

133
angles are listed in Tables 16 and 17 respectively. Since
the hydrogens were not allowed to completely float in the
refinement, only the average value of the hydrogen bond
lengths is given.

Some other interatomic distances and angles which are
of interest are listed in Tables 18 and 19, respectively.

It is interesting to note that the sodium (Nal) inside the
complex is only 2.585 R from the oxygen atoms of the crypt
while the sodium outside the complex (Na2) is over twice
that distance from the oxygen atoms. A similar effect is
noted for the nitrogen atoms.

The structure compares well with the structure of
[Na+C18H3606N2]I-, which was determined by Moras and WeissE°9
The sodium iodide complex crystallizes in the P31c space
group. Although there are no two-fold rotation axes in the
iodide case, 3-fold symmetry is still present. Some of the
interatomic distances, for the two structures are listed in
Table 20. As can be readily seen, the sodium-oxygen and
sodium-nitrogen distances are very similar for the two
structures.

The analogy of the outside sodium with an iodide anion
is very appropriate in View of Table 20. For example, the
Na(2)-Na(2) closest distance is 8.831 2 while the I- - I—
closest approach is 8.630 R.

The conformational bond angles were determined and are

tabulated in Table 21. The results of the sodium-cryptate

134

Table 16. Bond distances and estimated standard deviations.

 

 

 

Dist. 0

Atom 1 Atom 2 (R) (8)
N CI 1.445 0.013
C1 C2 1.303 0.016
C2 0 1.359 0.010
0 C3 1.364 0.020
C3(A) C3(B, 1.330 0.019
C (Average) H 1.063 0.015

 

Table 17. Bond angles and estimated standard deviations.

 

 

 

Atom(l) - Atom(2) - Atom(3) Angle o
C1(A) - N(A) - c1(c) 111.3 .7
N(A) - C(lA) - C(2A) 124.2 .7
C(lA) - C(2A) - 0(A) 123.4 1.1
C(2A) - 0(A) - C(3A) 109.0 1.8
o(A) - C(3A) - C(3B) 122.6 1.4

 

135

Table 18. Some interesting interatomic distances.

M A.

 

Atom 1 - Atom 2 Distance 0
(3) (R)

0(A) - Na(l) 2.548 0.009

N(A) - Na(l) 2.685 0.010

N(A) - o(A) 2.885 0.006

0(A) - 0(8) 4.058 0.010

Na(l) - Na(l) 8.831 0.001
Na(2) - Na(2) 8.831 0.001
Na(l) - Na(2) 7.058 0.001
N(A) - Na(2) 5.549 0.001

0 - Na(2) 5.759 0.006

Na(2) - c(1) 4.801 0.011
Na(2) - c(2) 4.984 0.013
Na(2) - c(3) 5.355 0.014
Na(l) — c(1) 3.412 0.008
Na(l) - c(2) 3.370 0.012
Na(l) - c(3) 3.280 0.018

 

Table 19. Some interesting interatomic angles.

 

 

 

Atom(l) - Atom(2) - Atom(3) Angle O
N(A) - Na(l) - o(A) 66.9 0.1
0(A) - Na(1) - 0(8) 68.5 0.1
o(A) - Na(1) - o(c) 105.6 0.2
o(c) - Na(l) - 0(F) 172.6 0.3
Na(2) - Na(1) - Na(2) 77.4 --
Na(2) - Na(l) - Na(2) 102.6 --
N - Na(l) - Na(2) 46.2 0.1
Na(2) - N - Na(l) 113.3 0.1

 

136

Table 20. Comparison of some interatomic distances with

the interatomic distances obtained from
[Na+C13H3505N2]I-.

 

 

 

Atom(l) - Atom(2) This work Iodide work1°9
‘ (3) [Na2 = I’l (2)
Na(2) - Na(l) 7.058 7.400
Na(2) - N 5.549 5.090
Na(2) - o 5.759 5.186
Na(2) - Na(2) 8.831 8.630
Na(l) - Na(l) 8.831 8.630
Na(2) - c(1) 4.801 4.267
Na(l) - O 2.548 2.582: 2.566
Na(l) - N ‘ 2.685 2.72 ; 2.78

Table 21. Conformational angles of this structure compared

with the conformational angles in the iodide
structure.

 

 

 

This Structure Iodide Structure109
N - C1 - C2 - O - 17.1 - 58.1 - 40.6
C1 - C2 - O - C3A 131.3 163.1 148.3
C2 - O - C3A - C33 -149.5 -153.2 -162.2
OA-

C3A - C3B - 0(8) 31.4 45.2 45.2

137
iodide structure are also listed for comparison. Because
of the lower symmetry in the iodide lattice there are twice

as many independent angles.

V.3.9. Discussion

 

V.3.9.1. General

The ability of the complexing agent, 2,2,2-crypt, to
twist was noted by Moras and Weiss since the [K+C18H3606N2)I-
structure has been determined.110 The analysis of the
structure described in this thesis further confirms the
ability of the strands of the complexing agent to twist.

A very remarkable feature of this structure is the manner in
which the oxygen atoms are staggered away from the outside
sodiums. Figure 39 shows the atomic packing in the crystal
lattice.

Another amazing feature of the compound is the twist in
the strands of the crypt. Due to the twist of nearly 60°,
all atoms in the strand show a center of inversion through
the central sodium, with the exception of the center carbon
atoms. This may be visualized by examining Figure 39.

A stereo plot of the complexing agent is shown in

Figure 40. Figure 41 also includes the closest neighboring

outside sodium atoms.

138

.mEoum EsfloOm wofimuso ummnmmc xfim
map cam oumufimuquSHUOm 0:0 mGHBOSm Hobos

 

.mm musmfim

 

139

 

Stereoscopic view of the complexing agent with the

inner sodium.

Figure 40.

140

.meswpom woflmuso ummummc
xflm on» nuHB mumumauoladfloom 0:» mo 30H> aflmoomomnmum

 

.Hv musmflm

141

V.3.9.2. Proof for the Sodium Anion

In addition to the large thermal factors for the out-
side sodium whith may be an indication that its electronic
cloud is more diffuse, there is repulsion between the out-
side sodium and the oxygen atoms of the crypt. The repul-
sion is even greater than the repulsion between the oxygen
and the iodide ion in the sodium cryptate-iodide structure
(see Table 20). Similar effects with other atoms are also
observed as shown in Table 20, by analogy of the sodium
outside the complex (Na2) with the iodide anion. The sodium
species which are inside the crypt in the two structures,
have very similar interatomic distances to the oxygen and
nitrogen atoms. This is significant because in the iodide
structure, the trapped species is the sodium cation. One
would certainly expect different Na-O and Na-N distances
if the trapped species were Nao.

The outer sodium is also far removed from other atoms.
The closest atoms, whicn are the carbons, are over 4.8 A
away. Thus, it seems that the outer sodium is surrounded
by a large volume of empty space. This also supports its
assignment as the sodium anion, in that the extra electronic
charge would be expected to Spread over a large volume.

The radius of the sodium anion has been calculated62
from Slater atomic radii to be slightly larger than that of
an iodide ion (r1- = 2.16; rNa- s 2.15-2.35 R).

We have performed further calculations by the use of

Clementi's Self-consistent Field (SCF) wavefunctions117 for

141a
Na0 and Na-. A plot of the radial distribution function

for the 3s electrons reveals that the size of the sodium

anion will be at least as large as the iodide ion (rNa-:3

2.2-2.4572).
Indeed this ionic radius is consistent with some of
the interatomic distances. The Na--C(1) closest approach

is 4.801 8 while the 1’-c(2) distance109 is only 4.291 R.

The Fourier map of the electron density clearly shows

the empty space which surrounds the sodium anion. There is
absolutely no other electron density near the sodium anion
for several Angstroms. Figure 42 demonstrates the empti-
ness surrounding the sodium anion. In this figure, two
sections of the z direction are shown. In the first

section, 2 = 14/120. As can be seen readily,

142

I In \oIv‘ v In III": I - I6.I?OIN$
o I l I s s I I a Q I. II It I) I0 I! I. II II I! to {I I! I) (I I‘ 80 2' II II

I Is I! .9 I? I! Q‘ :6 Q I‘ I! so It 0 I0
I I! I9 O -0 I, II S O 9 I ' I
I '0 s l .0

I I) II 79 It 0 .4
n )9 no A! a) o. o It
s u n "HQ III II I -0 I1
o Is ‘9 In so 08 KI Is s

O -I II I! II In A
o I h S I

II .0 I I

II I o. I I

I? I I I 2 I

I‘ I I h 0 , O

u .0 I s I I I I-0

II I -o o. I I I I 4 -0 on

H -o I

I' -o -I

II -I O .. 0

I9 00 O I 8 O '00 .0

n -o -o 8 o. I -I I

II 0 I O I I I S )

n I I I I o I I I I

n I I 6 o I _ I I I I I -l I II .0
nuuaIIoIII: oasI-Ialllnun o
)5 I. II ‘1 I. $ O a. 1 I 1 .0 .0 I 3 f O O 6 I. 0S $9 99 I’ 9
n MI I- u. u II o o. I I :I I --I -I -I I -o .. I I I I 9 n n «I III I. .0 I

n I'- "@oo u u I ' .0 on .0 I III so @110 I:

I In )QIDQ V I! JOIW$ I I I. Iiiifl‘
0 I ' ’ 9 ‘ Q ’ . . I. II I, II I6 I’ I. I, I. I. I. II II I) I. '5 (Q I, I. '0

I o. o.

' In M I

. I. IIh IAI II. (I

° )0 III 200 :01 III )9
I0 I II. ,0) )‘5 '06 II‘ 3
II )0 IfiI '06 '9‘ IQI )0
I? I. II! IAI II. ,I
II 3 '0 )0 I
I9 -0 -0

II 0.

I) ..

IQ 0.

Figure 42. Fourier map of the electron density in the
unit cell: above, 2 I 14/120; below,
2 = 20/120.

143

there is a very large, vacant area in a large part of the
cell. Furthermore, the three-fold axis can be seen in this
particular slice of z. The peaks [value = 103] circled in
the figure show the three-fold rotation about a point
x = 2/3, y = 1/3. As one proceeds in the z direction,
the electron density of the complexing agent (due to the C1
atom) decreases while the electron density of the sodium
anion gets larger. At 2 = 20/120, or 1/6, the sodium
anion electron density peak is the strongest. One can
easily note the emptiness surrounding the outer sodium. At
higher z values, the electron density of the crypt grows
again, while the density around the sodium anion falls. At
z = 26/120, the electron density is almost the same as that
obtained from the Fourier map at z = 14/120.

Further evidence for the existence of the sodium anion
was obtained as follows:

Because the scattering factor curve for the sodium
anion is not known, an empirical approximation was made by

assuming,

Although this approximation is very inexact, it was substi-
tuted for the scattering factor table of the outside sodium
species. It is interesting to note that even though the

Na-empirical scattering factor table is based on the scat-

tering factor curve of the hydride ion, fNa‘ - fNao s:

f - f . For the sodium inside the crypt, the scat-

NaO Na+

tering factor table for the sodium cation was introduced.

144
The results yielded a new R value of 0.0822. As noted

by Hamilton,111 this represents a significant lowering of

 

the R value at the 99 percent confidence limit.

Other studies assumed the scattering factor table for
a sodium cation inside the crypt, and the scattering factor
table for a neutral sodium atom outside the crypt. In this
way, the R value was reduced to 0.0835. This is also
significant at the 99 percent confidence limit.

Based on the small difference between the scattering
factor curves for the sodium anion and neutral sodium, as
shown in Table 22, we feel that the significant lowering of
the residual index may be important. These results are
tabulated in Table 23.

It should be remembered, however, that there are very
few reflections that are effected by the different scatter-
ing factor tables. Furthermore, the empirical scattering
factor table for "Na-" makes these results incondlusive.
But, the trend to better R values may be an indication
that there is extra electronic charge on the outer sodium
species.

In view of the results of this structure, there can be
little doubt that the sodium species in the crypt is the
sodium cation, and that the sodium Species outside the

crypt is the sodium anion.

145

Table 22. Scattering factor curves for Na+, Nao, and "Na-".

 

 

 

sin e/x fNa+ fNao "fNa-"
0.00 10.00 11.00 12.00
0.05 9.89 10.60 11.32
0.10 9.57 9.82 10.18
0.15 9.07 9.11 9.27
0.20 8.45 8.44 8.50
0.25 7.75 7.75 7.79
0.30 7.02 7.04 7.07
0.35 6.31 6.34 6.36
0.40 5.64 5.67 5.68
0.50 4.46 4.48 4.49
0.60 3.55 3.56 3.57
0.70 2.88 2.89 2.90
0.80 2.40 2.41 2.42
0.90 2.07 2.07 2.07
1.00 1.84 1.84 1.84
1.10 1.67 1.67 1.67
1.20 1.55 1.55 1.55
1.30 1.45 1.45 1.45
1.40 1.38 1.38 1.38
1.50 1.31 1.31 1.31
1.60 1.25 1.25 1.25

 

145a

Table 23. Summary of crystallographic significance tests.

 

 

 

Inside Crypt Outside Crypt R
fNao fNaO 0.0850
fNa+ fNao 0.0835
fNa+ fNa-' 0.0822
f f + > 0.090

Na’ Na

 

VI. SUMMARY AND SUGGESTIONS FOR FURTHER WORK

V1.1. Amine and Ether Solvents

 

Because of the extension of the solvent range to in-
clude a number of amines and ethers, further studies have
been made on the species, "M-“, "M", and ”egolvu' Be—
cause of the characteristics of these species, we can
classify "M-" to be an alkali anion while "M" represents
either the very tight ion pair, "M+- e-" or an alkali metal
monomer. Comparison of the absorption maximum of the
solvated electron with the peak position obtained by pulse
radiolysis provides substantial proof that the IR band seen
in metal solutions is due to the solvated electron. The
slightly broader band observed in metal solutions is attribw

uted to the overlap of the monomer band with the solvated

electron band.

V1.2. Metal-Ammonia Solutions

 

The results obtained from the spin susceptibility
measurements indicate that the spin-pairing process is
relatively independent of the cation. This is best under-

stood if, in metal-ammonia solutions, there are only weak

. . . +
interactions among the Spec1es "M " and "esolv .

146

147

V1.3. Solid-State Studies

 

The solid-state studies indicate that species such as
"Na ” can exist in solids. Solid-state work is one method

for the unequivocal identification of such Species.

V1.4. Proposed Model

 

Because the Species which exist in these solutions de-
pend on temperature, concentration, and solvent, one needs
to consider what actually causes the formation of such
species. It is entirely possible that the degree of inter-

action between the species "M+", and "e " determines what

Solv

other species will also be found in the solution. It is

known that the degree of interaction among Species may change

as the temperature, concentration, or solvent is altered.
Thus, the proposed interaction is a function of solvent

type. Solvents such as ethylamine permit strong interactions

resulting in long-lived spectroscopically observable species

such as "M" and "M_". On the other hand, ammonia lies at

the other end of the spectrum since it permits only relatively

weak interactions between "M+“ and "e- ". The spin-paired

solv

species in ammonia may actually be considered to have the

. . +
st01chiometry of e -M -e where the aggregate is a weak

ion pair. Alternatively, it could arise from the dielectron,

e2 0

148

V1.5. Suggestions for Further Work

 

There are several areas of interest in which work needs
to be done:

(1) Although the proposed interaction scheme is con-
sistent with all the available information to date, it would
be desirable to obtain a solvent in which the interactions
are similar to those obtained in ammonia. Some of the di-
amines used to prepare the complexing agent, 2,2,2-crypt
might be suitable for this purpose.

(2) A detailed study of the spin-paired species in
metal-ammonia Solutions Should be performed to check whether
the apparent plateau in the region of 0.01 to 0.1M is real.
In order to avoid the problems described in Chapter IV, the
study needs to be performed on either a radio frequency ESR
spectrometer or with an X-band spectrometer equipped with a
low power arm. 1n the latter case, low frequency modulation
should be used to prevent the appearance of modulation side-
bands.

(3) Further studies need to be performed on the gold-
colored compound, Na2C18H3606N2, in order to fully charac-
terize this new solid. Based on its packing, single crystal
anisotropic conductivity Should be very different from the
conductivity observed with packed powders. Sodium NMR should
also be performed on the new solid compound.

It is very likely that other solids of this type might

also be prepared. A compound might be made with a potassium

149
cation trapped inside the crypt and a sodium anion outside
the crypt. Various combinations of the complexing agents

and the metals can be attempted.

APPENDICES

APPENDIX A

THE ISOLATION OF SMALL AMOUNTS OF MATERIALS IN
THIN‘WALLED VESSELS

1.1. Introduction

Before advocating a new method of sample introduction,
one ought to first consider the objectives which should be
met and the difficulties in meeting these objectives by
conventional methods. An ideal sample-handling system
should meet the following objectives:

(1) Complete isolation of the sample from air and

moisture. Both vacuum and inert atmosphere capabilities

are desirable.

(2) The ability to handle a wide range of sample
weights. The method should be capable of delivering

small as well as large amounts of material.

(3) The ability to introduce samples in the form of

liquids, solids, or solutions.

(4) The technique should be relatively rapid and simple

to perform.

(5) The method should allow quick handling and easy
manipulation of the samples.

150

f I ‘IIIIII lllllc It!

 

151

1.2. Heatjghrink Tubing and Bulbs

To meet the objectives listed above, the general bulb
technique was extended by the utilization of heat-shrink
tubing. Heat-shrinkable Teflon tubing is commercially avail-
able in a variety of sizes. A very common use for such
tubing is in vacuum systems where pressures as low as
2 x 10"6 torr can readily be obtained. For example, in this
laboratory a sealed tube containing cesium metal is scratched
with a glass cutter and is placed inside of a larger glass
tube equipped with heat-Shrink tubing as shown in Figure 43A.
After evacuation of the vessel, the Teflon tubing can be bent
to break the inner tube, which can then be allowed to slide
into the main vessel. The outer tube can be sealed off as
shown in Figures 43B and C.

Heat-shrink tubing has the advantage that it can be
easily removed by cutting it with a razor blade and, most
importantly, it leaves behind no residue (in contrast to
ground glass joints which may leave a residue of grease).

The use of heat-Shrink tubing allows us to weigh light
samples accurately since the total empty weight of the ampoule
is only 100-400 mg.

After weighing, the bulbs are attached onto a manifold
with either Teflon or irradiated polyolefin heat-shrink
tubing as shown in Figure 44. The bulbs can be sealed just
below the heat-shrink tubing after evacuation and filling.

After seal—off, the plastic tubing is cut and the sealed

152

.mumnmmoeum mnu ou wusmomxm uso

 

 

 

 

Inufl3 mammmm> oucfi mmamfimm amuse mosbouucfl ou Uneasy xGHHSqummn mo mmD
r H u I!
m Ofiéua I
b *Ib
|J—|l H

 

 

 

 

 

 

 

 

 

 

 

 

n
1‘ II< 11“
b D b
<
\ 0‘ di 0‘
SEND-MU
(I I9. \LDV Ilh”?

 

 

.mv seamen

153

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 44. Manifold used to seal small amounts
of materials into breakable bulbs.

154
bulb and stem are weighed again. A buoyancy correction for
the weight of air can be made my measuring the diameter of
the bulb and calculating its volume.
The manifold used is generally circular in shape so
that either a large Dewar flask can be used to simultaneously
cool all the bulbs, or a small Dewar flask can be used on

any particular bulb.

1.3. Bulb Preparation

 

Since the bulbs are rather difficult to clean, the
washing is usually performed before blowing the ampoules.
Ten millimeter tubing is cleaned in the normal fashion (see
Chapter 11). An area approximately two centimeters long is
heated uniformly and then pulled to a length of 25-40 cm.

It is very important to pull the tubing slowly at first in
order to allow the tubing to cool somewhat as it is stretched.
Otherwise, a non—uniform capillary will be formed and the
diameter will be too small.

As shown in step 4 of Figure 45, the capillary is cut
between the heavy sections such that each piece has capil—
lary at either end and a thick portion in the center. A
portion of the center (approximately 1/2 cm) is then heated
uniformly and the diameter at the end of the tube which was
just heated and closed should be approximately 1—2 mm. The
amount of glass left at the end of the tube is critical. It
determines the size and strength of the ampoule. For bulbs

10-13 mm in diameter, the length of solid glass should be

‘ luv" Il.‘|1|.

155

.mQHSQ manmxmonn may mo coflumummmnm on» How mmoum umuflm .nv musmflh

 

 

 

 

 

 

 

 

v

 

 

V

 

 

 

 

 

 

156
approximately 2-3 mm. The tube is then rotated with the
solid glass in the flame until it emits a soft red color.
To keep impurities to a minimum, we use a blow-bag to blow
the bulb. If the bulb is too small or too large, it may
sometimes be melted again down to solid glass and reblown.

The bulbs can easily be tested. First, for a bulb 10—
13 mm in diameter with a 5 to 8 cm Stem, the weight should
not be less than 100 mg nor greater than 400 mg. Second,
typical sample bulbs can be tested to see if they will
break under appropriate conditions. The bulbs can be sealed
with air inside and can then be placed in a test tube approx-
imately 15 cm high. One should be able to shake the bulb
and move it around without damage and yet it should break
easily when a 1-2 inch stir bar (Teflon covered magnet) is
dropped from a height of approximately 4 inches.

It might be helpful to point out that each of the vari-
ous stages of the preparation should be stockpiled before
proceeding to the next stage. The various recommended
stages are shown in Figure 46. One will have greater suc-
cess by blowing 100 bulbs in succession than by repeating
all the steps in succession the same number of times. Once
one "captures" the technique at a certain stage, it can be

reproducibly repeated with a large success ratio.
1.4. BulbLFilling

The bulbs can be easily filled. Different methods are

available depending upon the type of material to be isolated.

157

.mnasn mo coflumummmum on» Ca mommum ommuoum ombcweeoowm

m

 

.ov musmflm

 

 

 

 

 

 

 

 

158
In all cases, the empty bulbs are first weighed on an analy-

tical balance.

1.4.1. Solids (weighable amounts)

 

For solids that are unreactive to air and moisture, the
bulb can be filled by simply funneling the powder down the
stem and gently Shaking the ampoule. The only requirement
is that the substance be dry. If the material is wet, it is
desirable to later rinse the stem with an inert solvent
which is volatile. To get an approximate idea of the weight
of material, the filled bulbs can be weighed on an analytical
balance. The bulbs can then be loaded onto the manifold
shown in Figure 44. A heat gun is used to shrink the Teflon
tubing around the glass.

If the solid is reactive to the atmosphere, the entire
process may be performed in an inert atmosphere box. Elec-
trical outlets are usually available in the box so that not
only can the material be weighed but also the heat gun may
be used. A Teflon needle valve stopcock which is closed im-
mediately prior to the removal of the manifold from the box
keeps the compounds under the inert glove-box gas.

The entire manifold is placed on.a vacuum line and
evacuated. In this way, the material never contacts the
atmosphere. The bulbs can easily be sealed by bringing a
hot flame near the stem until it just starts to collapse.
Then, this procedure is done on another side of the tube.

By rotating around the stem in this manner, the glass

159
collapses nearly uniformly. This prevents the formation of
weak spots which are prone to form since the glass wall of
the stem is fairly thin.

After it is sealed, the stem (which is easily removed
by cutting away the Teflon tubing with a razor blade) and
the bulb are weighed again on the balance. A buoyancy cor-
rection can then be performed by measurement of the diameter
of the bulb and calculation of the volume of the sphere.

The weight of the solid can accurately be determined after

correction for the weight of air.
1.4.2. Pure Liquids

Liquids can be introduced into the bulbs by two general
procedures.

First the liquids can be injected with a hypodermic
syringe and needle. The hypodermic needle should be long
enough to allow the tip of the needle to enter the bulb area.
Otherwise, the surface tension of the liquid will cause
filling of the stem only. The bulbs can then be placed on
the manifold and the same method is followed as for solids,
except that the bulbs must be kept cold to prevent evapora-
tion of the liquid. This is done by placing a large Dewar
flask containing liquid nitrogen around all of the bulbs.

It is usually necessary to de-gas the liquid by repeated
freeze-pump-thaw cycles.

Second, it may be desirable to distill the liquids into

the bulbs under vacuum. This can be done by attaching the

160
empty bulb to the manifold and cooling the bulbs individually
or collectively with a Dewar flask. Again, the bulbs can be

sealed as described earlier and weighed with the stems.
1.4.3. Solutions and Solutes

A hypodermic syringe may also be used to inject a
solution and the same technique described for liquids is
used for sealing the bulb. Furthermore, by appropriate
dilutions and final evaporation of the solvent, very small
amounts of non-volatile solutes can be isolated in the bulb.
This method is extremely valuable when one wishes to pre-

pare small volumes of dilute solutions in a closed system.
1.5. Utilization

The bulbs can be piled into a sidearm on the reaction
vessel and then held in place by a Teflon bar magnet as
shown in Figure 47. The Teflon coated bar magnet can be
moved by means of an outer magnet. The bulbs can be shaken
one at a time into arm A(Figure 47) and the magnet can then
be dropped breaking the bulb. If desired, the bulbs can
also be rearranged by manipulation into the side arms A and
B. Furthermore, a coarse frit can be introduced between
the bulb—breaking section and the main body of the apparatus
in order to prevent the passage of glass fragments into the

system.

161

    

 

-. . .
\ — '
o
\_ ‘-—— —. .- -- O
— on..--— . - [7" . ...
- . o- |. .

in

5"
‘ n

‘_ ‘.‘ .
r... -:::.:-

i}-
_ 2CD ._

. ‘
-
-.—. 'a
' -

f"

'A\
‘6“-
Q

fi’

O ..._ — .——‘

.332:
p -
:‘.— .

f

-
‘ '.
' >1.
. . ‘
q...
\‘0-

Figure 47. Vessel containing several bulbs for the
introduction of materials; this cell was
used for pulse radiolysis studies.

162

1.6. Summary

This technique is not only rapid and efficient but also
can be adapted to many different problems. As many bulbs
as required may be stacked in a reaction vessel for use one
at a time. The technique also allows the introduction of
the materials during an experiment without a large time loss
and without opening the system. Another advantage is that
solvent "blanks" can be run on the same sample of solvent

which is later used to prepare the solutions.

APPENDIX B

THE ISOLATION OF SUB-MILLIGRAM QUANTITIES
OF ALKALI METAL

2.1. Introduction

 

The isolation of alkali metals in small amounts without
oxide has been a serious problem which is not easily over-
come. While several methods have been develOped for quanti-
ties greater than about 50 mg?7:73 no technique has been
successful for the easy preparation of known samples of
milligram and sub-milligram amounts of the alkali metals.
The main problems with the modifications of the bulb tech-
nique of Appendix A for metal samples are:

(1) The buoyancy correction (~.1 mg) will often be

greater than the quantity desired.

(2) The procedure of filling the bulbs involves the
general method of forcing molten alkali metal through
the stem and into the bulb. After filling the bulbs,
the remaining metal in the stem isolates the bulb from
the atmOSphere, and can be removed after sealing the
stem. However, it is almost impossible to obtain very

small quantities of the metal in this fashion, since

163

164
the metal in the stem near the seal will weigh much more

than 1 mgi

(3) The method demands that large quantities of metal
solidify in the stem. This is necessary Usinauetmattbe

metal in the bulb is free from exposure to the atmosphere.

Furthermore, the amount of metal isolated in a small
capillary tube as outlined in Chapter 11 cannot be accurately
determined in the sub-milligram region. Further problems
with this method often arise because the metal in the small

capillary cannot be readily exposed to the solvent.

2.2. Technique

 

A new method has been develOped for the purpose of
handling such small quantities of metal. The procedure is

an extension of the method outlined in Appendix A.

2.2.1. Capillary Diameter Measurement

 

Capillary tubing of «I0.25 mm I.D. and 4-6 mm O.D. is
cut into 10-15 cm lengths and cleaned in the normal fashion.
Both of the exposed cross sectional inside diameters of
the tubing are measured with a microscope. The microscope
can be easily calibrated with the millimeter scale on a
vernier caliper. The measured diameter should be nearly the

same at both ends.

165

2.2.2. Bulb Preparation

 

After appropriate tubes have been found, one end is
sealed as shown in Figure 48B. With a length of approxi—
mately 3-4 mm of sealed glass at the end, a bulb is blown
with the aid of a blow bag. In this way, bulbs of approxi-

mately 10 to 13 mm in diameter can be constructed.

2.2.3. Sample Filling

 

The bulbs are sealed onto a manifold. The alkali metal
is distilled through the manifold and seals are made at the
various constrictions behind the metal after distillation to
prevent back distillation. After the metal is heated,a pool
of alkali metal is formed on top of the capillary tube
(Figure 49A). Upon continued heating,tfle metal begins to
move down into the capillary. If the capillary is gently
warmed the metal will continue down the capillary tubing
(Figure 49B). After the approximate desired length has been
obtained, the capillary tubing is allowed to cool and the
remainder of the metal pool is heated to force the metal
out of the arm and back into the manifold (Figure 49C). The
capillary is again gently heated and the metal cylinder will
move down (Figure 49D). The tubing is then sealed (Figure
49E). Care should be taken not to heat the capillary too
strongly or else metal will distill from the metal cylinder
onto the walls of the capillary. The film can be easily
distinguished from the metal cylinder by shining light

through the capillary. The metals, sodium, potassium,

.Hmume Hamxam mo mDCSOEm
s3ocx HHmEm mo COHDMHOmw on» How mQHsn m0 cowumummmum .wv musmfim

 

166

 

 

 

167

 

L'flmrmm
__A_szm_mizfi' ”I )

 

 

 

 

 

 

 

 

 

IIHJ'Ivi’V

 

I
i
:

I
L ”M "b "I V
[ if m 17310

Various stages of driving the metal into the bulbs.

Figure 49.

168
rubidium and cesium can be easily isolated in this manner.
In the case of sodium, a slight pressure of helium gas is
desirable to force the metal pool down into the capillary

tubing.
2.2.4. Sample Length Measurement

Since the diameter of the metal cylinder is known,
only the length needs to be determined in order to calculate
the volume of metal and hence its mass. The length can be
found in a variety of ways:

(1) The length of the metal cylinder can be directly

measured by eye with a vernier caliper.
(2) The same method may be used with a magnifier.

(3) Photographic slides may be taken of the metal
cylinder with a vernier next to the metal cylinder as
shown in Figure 50. The slide can then be projected,
and magnified so that the length of metal can be easily

determined.

2.2.5. Sealing the Bulb

 

After measurement of its length, the metal cylinder is
heated so that it moves down into the bulb. When all of
the metal is in the bulb, the capillary stem is sealed off

under vacuum approximately one centimeter from the bulb.

169

   
 
  

F)

IITHIIH II IIIIIIIIII (Hf IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIHHIIHHIHHIHIIIHIII km

 

'“WI
. a. ”‘I' 2 ’ ‘ urn-noun .

II II. I Ia- IIIIII' IIII III IIIISIII' .IIIIIIIII IIII'II IIIII I II II I: II' ‘II IIIIIIIII I 6III' ,an'I' 'II7III II! III '
I.4.)“; ~‘ "SAW. I WM : ' I l W'flfli“ ‘r'finmam'

 

Figure 50. Measurement of the alkali metal in
the glass vessel with a vernier; a
slide is generally taken and enlarged.

170

2.3. Calibration Results

To check the accuracy of this technique, the amounts
of alkali metal inside of several ampoules were determined
by atomic absorption. A Jarrell-Ash AA unit, equipped with
a ten~centimeter path length burner was used. The potas~
sium blue line (A = 4042.8 R) was studied. Standard solu-
tions of potassium thiocyanate in a water-isopropanol mix-
ture were used for calibration.of the instrument. The
bulbs were broken inside a vessel containing a known volume
of solvent. Standards were prepared which contained 50 to
150 ppm potassium while all of the metal samples were pre-
pared to contain approximately 100 ppm (based on the calcu-
lated weight of the metal present in the bulb).

In addition, all solutions contained 1000 ppm of sodium
from the salt sodium thiocyanate, in order to reduce ioniza—
tion in the flame. The calibration results are shown in
Figure 51. Some samples are also marked on this figure.

The results of the observed XE the calculated values are

shown in Table 23.

2.4. Conclusion

 

Not only can accurately known sub-milligram quantities
of metal be isolated, but also, larger or smaller amounts
may be prepared depending on the diameter of the capillary

tube chosen.

171

.mucwom mamemm .Q "mucflom newumunwdmo
monmummmu .x “nuanmmn :owumuomnm Uwfioum mzu How m>uso Gawuwunflamo .Hn musmwm

ov— can 06— co co Oi ON
_ IA _ _ _ _ _

 

 

 

 

172

 

 

 

Table 24. Comparison of observed and calculated weights
of potassium metal.
Sample # Oggiggid giéghgigiigé Diff. Difif.
(m9) (m9)

1 2.443 2.472 -0.029 —1.17
2 1.287 1.293 —0.006 -0.49
3 2.017 1.998 0.019 0.92
4 2.385 2.408 ~0.023 —0.98
5 2.589 2.643 -0.054 -2.08

 

APPENDIX C

NON-LINEAR LEAST SQUARES PROGRAM AND MODIFICATIONS

3.1. Introduction

 

The general non-linear curve-fitting and equation-
solving program, named KINFIT, has been utilized extensively
in this work. The program is not only powerful but versa-
tile as well. It can be used for many different types of
equations and is therefore ngt_limited to particular prob-
lems. In addition, once the program has been studied and
used for one problem, it is a simple matter to adjust the
program to an entirely different type of calculation.

The program as presently dimensioned can handle up to
100 data points, 4 variables and 2 unknowns. It is, however,
a simple matter to redimension the program if necessary.

The methods of solution used are those of Pitha-Jones112
and of Powell.113 The potential of non-linear least squares
weighting has been extensively discussed by Wentworth.114
Furthermore, the mechanics of the program have been dis-
cussed elsewhere115 and will not be repeated here.

Besides using the program in some straightforward calc—

ulations, modifications were made to print additional

173

174
information. The new output was also calibrated with known
examples.

As with any complex instrument, it is necessary to
know if the program is functioning normally. It is desir—
able therefore to determine whether the program and the
measured data are capable of yielding the requested informa-
tion. Too often, one tries to obtain more information than
the data are capable of giving. It was for this purpose,
therefore, that the program was modified to generate "data"
from known parameters and constants. In this way, the
known generated data, after being "sprinkled" with random
deviations, can be fed back into the program to see whether
the program can determine the "best" parameters.

By introducing various known standard deviations on
the generated "data", it is possible to determine the ac-
curacy required of the real data in order to obtain the
parameters within given standard deviations. Obviously,
such capability in a program could be useful in the design

of the experiment itself.

3.2. Partial Pairwise Correlation Coefficients

 

While the KINFIT program prints the multiple correla-
tion coefficients, it did not previously output the partial
pairwise correlation coefficients. The multiple correla-
tion coefficient gives a measure of parameter coupling. It
is a measure of how well the parameter in question could be

replaced by a linear combination of all the other parameters.

175
The partial pairwise correlation coefficients relate to the
multiple correlation coefficients, but only two parameters
are allowed to float simultaneously. This indicates the
degree of mathematical coupling between any two parameters.
The correctness of the calculated partial pairwise
correlation coefficients was checked by using the test

cases given by Hamilton.116
3.3. More Flexible Use of Subroutine Egg.

The subroutine EQN was made more flexible by permit-
ting alteration of the number of unknowns and variables as
well as equation type from one data set to the next. The
modifications also permit one to re-order the input data
and to calculate the variances.

In addition, new plot routines were added. Besides
the original capability of plotting variable 2'zg variable 1
or the residual 25 any variable, the following options were
added:

(1) Plot the variable KVAR (Specified by the control

cards) and/or variable (KVAR +1) and/or variable

(KVAR + 2) v§_variable 1.

(2) Plot any combination of variables, constants and
parameters desired by definition of axes in subroutine

EQN .

176

3.4. Generation of Data

Modifications were made to calculate test “data" from
the equations, constants and parameters specified. In addi-
tion, normally distributed random noise of any desired

standard deviation, can be added to the generated “data".
3.5. Curvefit Procedure

Some protective statements were added to subroutine
CURFT (the main curvefitting routine) to avoid division by
zero or other "catastrophes“ which could previously occur

on occasion in the fitting procedure.
3.6. Flow Diagram

Flow diagrams are shown in Figure 52.

177

FLOW DIAGRAM FOR KINFIT2

I_sTART

I

 

 

 

READ (NOPT.IMETH.ITMAX.IWT.IRX.ISMIN.IPLT,NCST.TEST,KVAR
815. F10.0. I5

 

I
IS NOPT :.0 —13§—» END
I NO

IREAD ONE COMMENT CARD
I

IS NCST > O

 

 

 

 

 

 

 

 

I NO

 

READ CONST(I). I=1, NCST
8E10.0

I <

 

 

 

 

 

 

READ 0(1). 1 = 1, NOUNK ‘
8E10.0

 

Continued on next page

Figure 52.

 

178

.Jeou. .3 .862

 

 

 

 

 

 

 

S. :n .9 3. :u E
nx< .u. n.u has. .Inllnu. on
.aoupuoa4 a. unsung
.
.u.._x mu ...n.x .u...x mm
g 3.5” has .~.H.R §s

 

 

 

 

 

   
  

v. as. .usoa ..on..¢<>oa

. .
_Hv no a . c<xmma _wm‘- ¢<>oag

3.1.5.3 d4>.2..2x:
.

  

”hug

 

—v.°qu. ..§. .qln.::¢ ohm!) Alll. _Q8<U 55350 Ukufil—

 

 

gazes
.

 

 

 

 

 

 

. II nfllngfi §§ 805
r .
g as...

H: sh
.

. _
_. ouspu: uaa__w« aospu: any: « coauu: mmmg .nlllfi
.

 

. .

 

 

_ 92».
...§ 2 =5 93..

 

 

 

. .
0.0—u.
3L0: Jun. 4:55: Jan
:1 u— g). :1 n. I. v 2022
an? ¢
at I

 

 

sutuuuzn 2 can: “10.x

 

 

 

 

 

 

 

.z 9.9.». u

¢§ .- I

o. onuc . cozy»: an: .... u.¢<>. .u. ~.u.n4u¢
.to... _ a. .36... :— oupsazuo 3.2.823

..:. :53. :2 :5 93.. .52. :3 um: 32. 5:3 an: .

 

314333 :- :.v.xo: 2x

 

0.0uu. 7:03 .nofi
.EBZJI :h.u.x<>.xo:
:2. 225;... 2538‘

 

 

 

 

. .

FEE FEE 7!..th

 

 

 

 

 

 

 

l nuo .«.~u.—.3I~J_
IHI

LIST OF REFERENCES

13.

14.

15.

16.

17.

LIST OF REFERENCES

 

W. Weyl, Ann. Phys., 197, 601 (1863).

M. G. DeBacker, Ph.D. Thesis, Michigan State University,
1970.

E. M. Hansen. Ph.D. Thesis, Michigan State University,
1970. '

J. Jortner, J. Chem. Phys., g9, 839 (1959).

 

R. A. 099. Phxsic. Rev., ng 668 (1946).

 

W. N. Lipscomb, J. Chem. Phx§., El, 52 (1953).

 

R. A. Stairs, J. Chem. Phys., 21! 1431 (1957).

 

L. Landau, thsik Z. Soviet Union, 3, 664 (1933).

 

J. Jortner, Ber., 22, 696 (1971).

C. A. Kraus, J. Am. Chem. Soc., 32, 1197 (1908).

 

E. J. Kirschke, W. L. Jolly. Inorg. Chem., Q. 855 (1967).

 

U. Schindewolf, R. Vogelsgesang, and K. W. Boddeker,
Ang. Chem., Int. Ed., g, 1076 (1967).

 

Metal-Ammonia Solutions, Colloque Weyl I. p.1., G.
LePoutre and M. J. Sienko, eds. New York: W. A.
Benjamin, 1964.

 

R. C. Douthit and J. L. Dye, J. Am. Chem. Soc., §Zx
4472 (1960).

 

M. Gold and W. L. Jolly. Inorg. Chem., 1, 818 (1962).

 

 

D. F. Burow, and J. J. La owski, Advances in Chemistrx
Series, No. 69, 125 (1965?.

S. R. Gunn and L. G. Green, J. Chem. Phys., 36, 363
(1962); J. Am. Chem. Soc., 82, 358 (1963).

 

 

179

18.
19.
20.

21.

22.
23.

24.

25.

26.

27.

28.

29.

30.

31.
32.
33.

34.

35.

36.

37.

38.

Soc., gg, 4797 (1960).

180

S. R. Gunn, Ref. 13, p. 76.

 

S. R. Gunn, J. Chem. Phys., 41, 1174 (1967).
R. Catterall, Ref. 13, p. 105.

C. A. Hutchison, Jr., and D. E. O'Reilly, J. Chem. Phys.,
32, 1279 (1961).

 

V. L. Pollak. J. Chem. Phys., 34, 1856 (1961).

 

 

D. E. O'Reilly. J. Chem. Phys” g3, 864 (1961).

R. Catterall, and M. C. R. Symons, J. Chem. Phys., 42,
1466 (1965). ’””

 

D. S. Thompson, E. E. Hazen, Jr., and J. S. Waugh,
J. Chem. Phys., 44, 2954 (1966).

 

J. L. Dye, R. F. Sankuer, and G. E. Smith, J. Am. Chem.

 

C. A. Kraus, J. Am. Chem. Soc., 43, 749 (1921).

 

R. R. Dewald and J. H. Roberts, J. Phys. Chem., Z3.
4224 (1968).

 

H. M. McConnell and C. H. Holm, J. Chem. Phys., 26,
1517 (1957).

 

J. Acrivos and K. S. Pitzer, J. Phys. Chem., 66, 1693
(1962).

 

D. E. O'Reilly. J. Chem. Phys., 41, 3729 (1964).

 

E. Huster, Ann. Physik, 33, 477 (1938).

 

S. Freed and N. Sugarman, J. Chem. Phys., 11, 354 (1943).

 

C. A. Hutchison and R. C. Pastor, Rev. Mod. Phys.,
25, 285 (1953); J. Chem. Phys., 21, 7959(1953).

 

 

S. R. Gunn and L. G. Green, J. chem. Phys., 36, 368
(1962).

 

C. A. Kraus, J. Franklin, Inst., 212, 537 (1931).

K. Fueki, D. Fang. L. Kevan, and R. Christoffersen,
J. Phys. Chem., 26, 2297 (1971).

 

J. Jortner, S. Rice, and E. G. Wilson, Ref. 13, p. 222.

39.

40.

41.
42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

181

D. Copeland, N. Kestner, and J. Jortner, J. Chem. Phys.,
52, 1189 (1970).

 

N. Kestner and J. Jortner. J. Phys. Chem., 77, 1040
(1973). 'V”

 

M. Newton. J. Phys. Chem., in Print.

 

W. Bingel, Ann. Physik, 12, 57 (1953).

 

M. F. Deigen, Trudyilnst. Fiz, Akad. Nauk. Ukr. S.S.R..
5, 119 (1954); Zh. Eksp. Teor. Fiz., 26, 300 (1954).

 

E. Becker, R. H. Lindquist, and B. J. Alder, J. Chem.
Phys., 25, 971 (1956).

E. Arnold, and A. Patterson, Jr., J. Chem. Phys., 41,
3089 and 3098 (1964): Ref. 13, p. 160.

 

8. Golden, C. Guttman, and T. R. Tuttle, Jr., J. Am.
Chem. Soc., 21, 135 (1965). J. Chem. Phys., 44, 3791
(1966). '””

 

 

M. Gold, W. L. Jolly, and K. S. Pitzer, J. Am. Chem.
Soc., 22, 2264 (1962).

G. E. Gibson and T. E. Phipps, J. Amer. Chem. Soc., $8
312 (1926).

 

J. L. Down, J. Lewis, B. Moore, and G. Wilkinson, J,
Chem. Soc., 3767 (1959).

 

F.Cafasso and B. R. Sundheim, J. Chem. Phys., 2;, 809
(1959).

 

F. S. Dainton, D. M. Wiles, and A. N. Wright. J. Chem.
Soc., 4283 (1960).

T. Cavighy. J. Normant, and H. Normant, C. R. Acad.
Sci., Ser. C.. 258, 3503 (1964).

 

G. Fraenkel, S. H. Ellis, and D. T. Dix, J. Amer. Chem._
Soc., 21, 1406 (1965).

H. Normant, T. Cavigny, J. Normant, and B. Angelo, Bull.
Soc. Chim. Fr., 1561 (1965).

 

R. Catterall, L. P. Stodulski, and M. C. R. Symons,
J. Chem. Soc., A. 437 (1968).

 

K. D. Vos and J. L. Dye, J. Chem. Phys., §§, 2033 (1963).

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

182

K. Bar-Eli and T. R Tuttle, Jr., Bull. Am. Phys. Soc.,
2, 352 (1963).

R. R. Dewald and J. L. Dye. J. Phys. chem., 68, 121
(1964). V

 

 

M. G. DeBacker and J. L. Dye, J. Phys. Chem., 22, 3092
(1971).

J. L. Dye, M. G. DeBacker. J. A. Eyre, and L. M.
Dorfman. J. Phys. Chem., 26, 839 (1972).

 

I. Hurley. T. R. Tuttle. Jr., and S. Golden. J. Phys.
Chem., 42, 2818 (1968).

S. Matalon, S. Golden, and M. Ottolenghi, J. Phys.
Chem., 22, 3098 (1969).

G. Scheibe, Z. Elektrochem, 25, 701 (1929).

 

J. Frank, and R. L. Platzman, Z. Phys. (Leipzig), 122,
411 (1954). “

M. Smith, and M. C. R. Symons, Trans. Farad. Soc., 54.
347 (1958). V”

 

G. Stein, and A. Treinen, Trans. Farad. Soc., 55, 1086
(1959).

 

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

 

C. J. Pedersen, J. Amer. Chem. Soc., 82, 7017 (1967);
9g, 386 (1970).

 

B. Dietrich, J. M. Lehn, and J. P. Sauvage, Tetrahedron

 

Lett., 2885 (1969); 2889 (1969); J. Amer. chem. Soc.,
22,—2916 (1970).

 

R. Catterall, M. C. R. Symons, and J. W. Tipping, Ref.
13, p. 317.

K. Bar—Eli and T. R. Tuttle, Jr., J. Chem. Phys., 42,
2508 (1964).

 

N. Mammano and M. J. Sienko, J. Amer. Chem. Soc., 92,
6322 (1968).

 

V. A. Nicely. Ph.D. Thesis, Michigan State University,
East Lansing, Michigan. 1969.

74.

75.

76.

77.

78.

79.

80.
81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

183

R. Dewald, Ph.D. Thesis, Michigan State University,
East Lansing, Michigan, 1969.

E. J. Prosen and F. D. Rossini, J. Research Natl. Bur.
Standards, 22, 255 (1945); 22’(1), 59,65 (I945); 34 (3),
255 (1945). ‘””

 

J. Coops, D. Mulder, J. W. Dienske and J. Smittenberg,
Rec. Trav. Chim., fig, 153 (1947).

 

G. W. Watt, and D. M. Sowards, J. Amer. Chem. Soc.,
22, 4742 (1954).

 

D. F. Shriver, "The Manipulation of Air-Sensitive Com-
pounds," New York: McGraw-Hill Book Co., 1969.

We are grateful to Dr. C. H. Park of E. I. duPont de
Nemours Co. and to Professor J. M. Lehn of the Institut
de Chemie, Strasbourg, for providing samples of 2,2,2-
crypt.

C. H. Park, duPont Laboratories, private communication.

This section was performed by Dr. M. T. Lok.

M. T. Lok, Ph.D. Thesis, Michigan tate University,
East Lansing, Michigan, 1973.

J. L. Dye. M. G. DeBacker, and L. M. Dorfman, J. Chem.
Phys., 52, 6251 (1970).

L. M. Dorfman, F. Y. Jou, and R. Wagerman, Ber. Bunsenges,
Phys. Chem., 22, 681 (1971).

 

 

 

J. L. Dye, and L. H. Feldman, Rev. Sci. Instrum., 21,
154 (1966).

I. Burak and A. Treinen, Trans. Faraday Soc., 52, 1490
(1963).

 

M. Blandamer, T. Gough, and M. C. R. Symons, ibid..
62, 286 (1966).

T. Griffiths and R. Wijayanayake, Molec. Spectrosc..
1562 (1970).

 

M. Fox and E. Hayon. Chem. Phys. Lett., 14, 442 (1972).

E. M. Kosower, "An Introduction to Physical Organic
Chemistry," New York: Wiley, 1968.

184

91. E. M. Kosower, J. Amer. Chem. Soc., 88, 3253 (1958).

 

92. L. J. Giling, J. G. Kloosterboer, P. H. Rettschnick.
and J. D. W. VanVoorst, Chem. Phys. Lett., 8, 457
(1971); 8, 462 (1971).

 

93. G. Gabor and K. Bar-Eli, J. Chem. Phys., 28, 286 (1971).

 

94. M. Fisher. G. Ramme, S. Claesson, and M. Szwarc, Chem.
Phys. Lett., 9. 309 (1971).

 

95. J. W. Fletcher, W. A. Seddon, J. Jevcak, and F. C.
Sopchyshyn. Chem. Phys. Lett., 18, 592 (1973).

96. B. Bockrath, and L. M. Dorfman, J. Phys. Chem., in
print.

 

97. J. W. Fletcher, W. A. Seddon, and F. C. Sopchyshyn,
Can. J. Chem., 81, 2975 (1973).

 

98. J. L. Dye, M. G. DeBacker, and L. M. Dorfman, J. Chem.
Phys., 82, 6251 (1970).

99. A. DeMortier, G. LePoutre, and M. G. DeBacker, J. de
Chim. Physiq., 8, 380 (1972).

 

100. C. P. Poole, Jr., "Electron Spin Resonance," New York:
John Wiley and Sons, Co., 1967.

101. Spang Microanalytical Labs, Ann Arbor, Michigan.
102. UnisMex Corp., Griffith, Indiana.

103. We are indebted to Dr. C. H. Brubaker for the use of
equipment. The aid of Mr. E. Vitz is also acknowledged.

104. C. Kittel, "Introduction to Solid State Physics," New
York: John Wiley and Sons, Co., 1968.

105. "International Tables for X-ray Crystallography,"
Vol. I, Birmingham: Kynoch Press, 1965.

106. The author is indebted to Dr. A. Tulinsky and his group
of their help with the data collection and for many
valuable suggestions.

107. G. H. Stout, L. H. Jensen, "X-Ray Structure Determina-
tion." London: The Macmillan Co., 1968.

108. W. J. A. M. Peterse and J.H. Palm, Acta. Cryst., 28,
147 (1966). '

 

185
109. D. Moras and R. Weiss, ibid.. B29, 396 (1973).

110. D. Moras, B. Metz, and R. Weiss, ibid.. B29, 383
(1973).

111. w. c. Hamilton, ibid.. 18, 502 (1965).

112. J. Pitha, and R. N. Jones, Can. J. Chem., $4, 3031
(1966).

 

113. M. J. D. Powell, The Computer Journal, 1, 155 (1964).

 

114. W. E. Wentworth. J. Chem. Educ., g2, 96, 162 (1965).
115. J. L. Dye, and V. A. Nicely, ibid.. 48, 443 (1971).
116. W. C. Hamilton, "Statistics in Physical Science;

Estimation, Hypothesis Testing and Least Squares,"
New York: Ronald Press Co., 1964.

117. E. Clementi. IBM J. Res. and Develop., 2, 2 (1965).