than.
‘42.. .t m“

I. .. 1;. i:
t h» .MEle
. .3: E i

. $3? ‘

1.1.0.!
(-0.):
iiiul! 0,1.

I). y .. I"!
3|
I

. ..
KN? .
.513.

5.... .2. .
.51.». 5“...)

‘

. arbbx:

; 1.: {3

3.53.51»
fl

.\.r)‘l-\5 ‘ .
i...1:.vl.c.x‘.£l. ; , .KKAI
Iii :0!» ;§.‘IA, 3| L. .$:\t\n.

)ill): .t‘h-S911 t ‘ 17.3!
. vaL I. ill

.. :Et.‘ .
4. 3.31.1? 1.1.55.1.)€\u!\~
.1. .1 . . .,V V ‘ 53.3.31: .. . I

‘ .. .Hu H;W§.Muw.vm

 

 

m lllllllllllllllHIHHHIHIllllHUI!“llllllllllllllll

3 3 1293 014203

This is to certify that the

dissertation entitled

Solute-Solvent Interactions in Nonaqueous
Solutions

presented by

Philip W. Schultz

has been accepted towards fulfillment
of the requirements for

Ph . D . d . Chemistry
egree m

 

 

 

 

Date 11/16/95

 

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

LIBRARY ,
Michigan State I
Unlverslty

 

 

 

PLACE N RETURN BOX Io romovo this checkout from your rocord.
TO AVOID FINES rotum on or baton dot. duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU In An mm Action/Equal Opportunity Imitation
mm:

SOLUTE-SOLVENT INTERACTIONS IN NONAQUEOUS SOLVENTS

By

Philip Wayne Schultz

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemistry

1995

ABSTRACT

SOLUTE-SOLVENT INTERACTIONS IN NONAQUEOUS SOLVEN TS

By

Philip Wayne Schultz

The interactions of the cyanate, thiocyanate and selenocyanide anions in the
nonaqueous solvents: methanol, formamide and N-methylformamide were investigated by
infrared and multinuclear N MR spectroscopies. In this series of protic solvents, the
cyanate anion is more highly solvated than the thiocyanate and selenocyanide anions. In
order to understand the complicated nature of the spectral envelopes for the stretching
modes, the nature of the experimental vibrational modes were studied by ab initio
calculations at the MP2 level, in response to perturbations of these anions by a variety of
ionic and hydrogen bonding interactions. These studies provided a firm basis for assigning
the perturbations, in frequency and absorptivity, by the formation of hydrogen bonds with
this set of anions. These assignments have been augmented with the synthesis and crystal
structure determination of model solvates. The three anions were studied also in
nitromethane, a relatively ‘inert’ solvent, so that the salt and protic solvent concentration
could be varied independently to help identify the component bands of the stretching modes
and to provide insight into the solvation phenomena. The concentration dependence of the
infrared vibrations and the NMR resonances yielded the stoichiometry of the anion-solvent
complex and the thermodynamics of the complex formation. In the amide solvents, the
cyanate anion is hydrogen bonded at the nitrogen with one and two solvent molecules.

These two interactions are also present in methanol solutions with an additional complex

where a methanol molecule is bound to each end of the anion. The thiocyanate and
selenocyanide anions are substantially less solvated than the cyanate anion with the
interactions occurring mainly at the nitrogen end. These assignments are consistent with
the prediction that the negative electrostatic potential is concentrated off the nitrogen atom.
These investigations suggest that the extent of self association of the protic solvent is an
important factor in the anion-solvent equilibria, resulting in the minimization of the entropy

loss upon hydrogen bond formation.

Acknowledgments

The author would like to express his sincere appreciation to Dr. Alexander I. Popov
and Dr. George E. Leroi for their direction and encouragement during the course of this
study. Additional thanks need to be expressed to Dr. James F. Harrison who provided the
author substantial assistance.

He also wishes to thank the research groups of Drs. Leroi and Harrison for their
encouragement and the many productive discussions. Great thanks also go to the NMR
support staff for their help with the NMR spectrometers and the extensive use of the
spectrometers. The author is deeply indebted to the computer facility for providing a
tremendous amount of computer resources for this work.

Dr. Karen C. Inman, the author’s wife, deserves great thanks for her
encouragement and her source of ideas to spur this work to completion and to give this
work a quality that would not have been possible without her. Her optimism was a
tremendous help.

Finally, the author wishes to acknowledge the Creator of this physical world who

has given us curious minds and provided us with a fascinating world to discover.

Table of Contents

List of Tables
List of Figures
List of Abbreviations
I. Introduction
A. Introduction
B. Solvent-Solvent Interactions
C. Ion-Solvent Interactions
D. Ion-Ion Interactions
E. Model Crystal Structures
F. Proposed Work

H. Experimental Methods
A. Materials
B. Spectroscopic Techniques
C. Theoretical Calculations
D. Data Treatment
111. Theoretical Calculations
A. Introduction
B. Results
OCN', SCN' and SeCN‘
Electrostatic effects on OCN‘, SCN' and SeCN'
Ionic and hydrogen bonding
Atomic charges in ionic and hydrogen bonded complexes
Molecular orbitals in ionic and hydrogen bonded complexes

Electrostatic potential in ionic and hydrogen bonded complexes

sowéww—

Vibrational Frequencies in ionic and hydrogen bonded complexes

viii
xiv

XX

14
31
36

37
38
39
45
46
48
49
55
55
58

70
74
87
92

I

8. Cation Solvation 98

C. Conclusions 98

IV. Solvation of OCN' 103

A. Introduction 104

B. Results 105

1. Infrared Measurements 105

2. Assignments 117

3. NMR Measurements 120

4. Thermodynamic Parameters 121

C. Conclusions 129

V. Solvation of SCN‘ and SeCN' 132

A. Introduction 133

B. Results 135

1. Infrared Measurements 135

2. Assignments 145

3. Thermodynamics 148

C. Conclusions 160

VI. Crystal Structures 162

A. Introduction 163

B. Results 164

1. Crystal Structures for NaNCS°(DMF)2 and NaNCS0(NMF)2 164

a. Amide Structure 172

b. Thiocyanate 172

c. Hydrogen Bonding 172

2. Summary of NaSCN°(DMF)2 and NaSCN-(NMF)2 173

3. Crystal structures for Na(l2C4)2NCS°MeOH and 173
Na(12C4)2NCS

a. Infrared Spectroscopy 183

b. Ab initio Calculations 183

4. Summary of Na(12C4)2NCS'MeOH and Na(12C4)2NCS 183

vi

VII. Conclusions and Future Work
A. Conclusions
B. Future Work

VIII. References

vii

185
186
187

189

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

List of Tables

AH°vap for selected solvents

Entropy of vaporization (AS° ) for selected solvents

vap

Physical properties of selected solvents at 25°C

Donor and acceptor numbers of selected solvents

Wavenumber and band width of the vCN mode of the thiocyanate

anion in various solvents

Assignments of the six component vCN bands in

tetrabutylammonium thiocyanate in methanol by Gill et al.
Assignment of the four component vCN bands of
tetrabutylammonium thiocyanate in methanol by Chabanel and

coworkers

Spectroscopic parameters for the vCN mode of the cyanate anion
alkali cyanates in DMSO

Thermodynamics of alkali cyanate ion pairs in DMSO at 298K

Table 10. Complexes of metal thiocyanates and their vibrational spectra

in various aprotic solvents

viii

10

13

14

21

22

24

Table 11.

Table 12.

Table 13.

Table 14.

Table 15.

Table 16.

Table 17.

Table 18.

Table 19.

Table 20.

MNDO calculations of the vCN stretch in LiNCS complexes

compared with the experimental frequency

Thermodynamic parameters for the ion pair, dimer and tetramer

formation of alkali thiocyanates in various solvents

Frequency (cm") of the vCN mode of LiSeCN in various

solvents
Parameters for NMR spectra

Frequencies of the stretching modes of OCN', SCN' and SeCN'

species in solution

Calculated equilibrium bond lengths, stretching frequencies and
atomic charges for OCN', SCN' and SeCN'

Potential energy distribution and force constants for the
stretching modes of OCN‘, SCN' and SeCN'

Effect of a positive charge on the bond lengths, force constants
and vibrational frequencies for OCN', SCN' and SeCN’,
calculated upon fixing a bare proton 3 A along the internuclear

axis for either end

Calculated bond lengths, stretching frequencies, integrated
intensities, force constants and atomic charges for OCN' in ionic

and hydrogen bonded cyanate complexes
Calculated bond lengths, stretching frequencies, integrated

intensities, force constants and atomic charges for SCN‘ in ionic

and hydrogen bonded cyanate complexes

ix

29

30

31

40

50

55

56

62

65

67

Table 21.

Table 22.

Table 23.

Table 24.

Table 25.

Table 26.

Table 27.

Table 28.

Table 29.

Calculated bond lengths, stretching frequencies, integrated
intensities, force constants and atomic charges for SeCN' in

ionic and hydrogen bonded cyanate complexes

Occupancies of 2px, 2py and 2pz atomic orbitals in alkali metal,
alkaline earth metal and hydrogen bonded cyanates

Calculated changes in bond lengths, force constants and
frequency shifts for selected ionic cyanates and selenocyanides

with respect to the free anions

Calculated bond lengths and vibrational frequencies at the HF/
d95v+** level for OCN’, LiNCO, and Li(HOH)3NCO

Infrared parameters for the vCN mode of ~O.1M NBu4OCN in

several solvents

Peak position, band width, and molar absorptivity parameters of
component bands of the VCN spectral envelope of ~0.1M OCN‘
in DMF, NM, NMF, FA and MeOH

Summary of peak positions and band widths of the curve-fitted
component bands of the vCN stretch of ~0.1M NBu4OCN/NM
solutions with NMF, FA or MeOH added

Assignments for the vCN and VCO & 50m modes for ~0.1M
OCN' in neat solutions with MeOH, FA or NMF and titration

studies with these solvents

14N and 170 NMR chemical shifts of ~0.1 NBu4OCN in DMF,
NM, NMF, FA and MCOH

68

73

96

99

106

106

116

119

121

Table 30.

Table 31.

Table 32.

Table 33.

Table 34.

Table 35.

Table 36.

Table 37.

Table 38.

Association constants, K1, K2, and KD for the formation of
OCN""HB, OCN""(HB)2 and BH'"BH, where HB can be NMF,
FA or MeOH determined by 14N and 170 NMR spectroscopy

Association constants, K1, K2, and KD for the formation of
OCN""HB, OCN""(HB)2 and BH"'BH, where HB can be NMF,
FA or MeOH determined by infrared spectroscopy

Thermodynamic parameters for the formation of 1:1 and 1:2

OCN""solvent complexes in nitromethane solutions

Summary of position and band widths for the VCN stretch of
~0.25M NBu4SCN in NM, DMF, NMF, FA and MeOH

Summary of the deconvolved parameters for the component
bands in ~0.25M NBu4SCN in NM, DMF, NMF, FA, and
MeOH

Summary of the peak position, band widths (cm“) and molar
absorptivities from the curve-fitted component bands of the vCN
stretch of 0.15M NBu4SCN with NMF, FA or MeOH added

Summary of the peak positions and band widths from the curve-
fitted component bands of the vCN stretch of 0.025M
NBu4SeCN with MeOH added

Association constants, K,+K2 and KD, for the formation of the
two 1:1 complexes and self association of the protic solvent,

determined by 1“N NMR measurements

Association constants (K,+K2 and KB) of SCN' with MeOH,
FA and NMF from the stretching modes of SCN'

xi

127

128

129

137

137

144

145

155

159

Table 39.

Table 40.

Table 41.

Table 42.

Table 43.

Table 44.

Table 45.

Table 46.

Table 47.

Individual association constants using the extinction coefficients

from ab initio calculations

Thermodynamic parameters for the association of SCN' with
MeOH, FA and NMF from the NMR measurements

Positional parameters (A) and equivalent isotropic thermal
parameters (A2) with estimated standard deviations in

parenthesis for bis(N-methylformamide)—sodium thiocyanate

Positional parameters and isotropic temperature with estimated
standard deviations in parenthesis factors for

bis(N,N-dimethylformamide)-sodium thiocyanate

Intermolecular distances (A) with estimated standard deviations

in parenthesis for bis(N-methylformamide)-sodium thiocyanate

Intermolecular distances with estimated standard deviations in
parenthesis for bis(N,N—dimethylformarnide)-

sodium thiocyanate

Atomic coordinates (104) and equivalent isotropic displacement
parameters (A2x103) with estimated standard deviations in
parenthesis for the Na(12C4)2NCS complex

Atomic coordinates (104) and equivalent isotropic displacement
parameters (A2x103) with estimated standard deviations in
parenthesis for the N a( 12C4)2NCS°MeOH complex

Selected bond lengths with estimated standard deviations in
parenthesis for the Na(12C4)2NCS complex

159

160

169

170

171

171

178

179

180

Table 48. Selected bond lengths with estimated standard deviations in 181
parenthesis for the N a( 12C4)2NCS°MeOH

Table 49. Ab initio calculations for axial and nonaxial hydrogen bonds 184
with SCN‘

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

List of Figures

. RDF for pure FA, NMF, and DMF

. The VCN stretch of NBu4SCN in methanol and the component

bands determined by Gill and coworkers for spectral envelopes

. Concentration dependence of the 39K chemical shift in various

solvents

. Concentration dependence of the 23N a chemical shift of

NaSCN, NaBr, NaBPh4, NaClO4 in DMF

. Experimental infrared band envelope for the VCN stretch of

NaSCN in THF with calculated component bands

. Pictorial view of the LiNCS ion pair, dimer, and tetramer

. Frequency (cm") of the vCN stretch in alkali and alkaline earth

thiocyanate ion pairs as a function of the polarizing power of

the metal cation

. Resonance structures for the thiocyanate anion

. VCN/P of various metal thiocyanates correlated to the polarization

of the metal cations

xiv

12

17

19

20

25

26

26

27

Figure 10.

Figure 11.

Figure 12.

Figure 13.

Figure 14.

Figure 15.

Figure 16.

Figure 17.

Figure 18.

Figure 19.

Figure 20.

Figure 21.

Figure 22.

Li(9-Crown-3)SCN

Symmetric (LiNCS)2 dimer (left) and asymmetric
[Li(TMPA)2NCS]2

Diagrammatic representation of the coordination in
(KNCS)2-DB24C8

Contour plots for the valence molecular orbitals of OCN‘
Contour plots for the valence molecular orbitals of SCN'
Contour plots for the valence molecular orbitals of SeCN‘
Atomic charges at O, C and N in MNCO species, where M can
be a hydrogen bond donor, an alkali metal cation, or an
alkaline earth metal cation, versus the Coulombic potential
provided by the cation or hydrogen bond donor; r is the
correlation coefficient with and without (in parenthesis) the
hydrogen bonded species

Contour plots for the valence molecular orbitals in LiNCO
Contour plots for the valence molecular orbitals in LiOCN
Contour plots for the valence molecular orbitals in BeNCO+
Contour plots for the valence molecular orbitals in Li2NCO+

Contour plots for the valence molecular orbitals in LizOCN“

Contour plots for the valence molecular orbitals in LiNCS

XV

32

34

35

59

60

61

72

75

76

77

78

79

80

Figure 23.

Figure 24.

Figure 25.

Figure 26.

Figure 27.

Figure 28.

Figure 29.

Figure 30.

Figure 31.

Figure 32.

Contour plots for the valence molecular orbitals in BeNCS+
Contour plots for the valence molecular orbitals in LizNCS+
Contour plots for the valence molecular orbitals in LiNCSe
Contour plots for the valence molecular orbitals in BeNCSe”
Contour plots for the valence molecular orbitals in LiZNCSe*
Contour plots for the 11: and 21: molecular orbitals in OCN'
Li”NCO‘, Be2*NCO' and Li+OCN',

Contour plots of a 1: molecular orbital in HF extending into the
ON bond of OCN', thereby increasing the electron density
surrounding the C-N bond in the OCN""HF complex

Electrostatic potential for OCN', SCN‘ and SeCN‘

Molecular electrostatic potential of OCN', OCN""HOH,
OCN""HF and Li‘NCO’

Ab initio frequencies of the vCN stretch in alkali cyanates versus
experimental frequencies of the vCN stretch of alkali cyanates in
DMSO

Figure 33. Ab initio frequencies of the VCN stretch in alkali thiocyanates

Figure 34.

versus experimental frequencies of the VCN stretch of alkali

thiocyanates in aprotic solvents

Plot of the FCN and FCO force constants versus their respective

bond lengths in ionic and hydrogen bonded OCN‘ complexes

xvi

81

82

83

84

85

86

88

9O

91

93

93

97

Figure 35.

Figure 36.

Figure 37.

Figure 38.

Figure 39.

Figure 40.

Figure 41.

Figure 42.

Figure 43.

Figure 44.

Figure 45.

Pictorial view of the Li(HOH)3NCO ion pair

vCN stretch of ~0.1M NBu4OCN in DMF, NM, NMF, FA, and
MeOH

vCN stretch of ~0.1M NBu4OCN in NM and the Fourier

deconvolution of the vCN stretch as a function of MeOH added

vCN stretch of ~0. 1M NBu4OCN in NM and the Fourier

deconvolution of the vCN stretch as a function of FA added

VCN stretch of ~0.1M NBu4OCN in NM and the Fourier
deconvolution of the vCN stretch as a function of NMF added

Absorptions for the component bands in the 2200-2100 cm'1
region for ~0.1M NBu4OCN/N M solutions as a function of
MeOH, FA or NMF concentration

\ICN & 250CN stretch for ~ 0.1M NBu4OCN in NM as a function
of MeOH added

VCN & 250CN stretch for ~ 0.1M NBu4OCN in NM as a function
of FA or N MF added

'4N and '70 NMR chemical shifts for 0.1M NBu4OCN as a
function of MeOH added and temperature in the case for the

l"N resonance

VCN for ~0.25M NBu4SCN in DMF, NM, NMF, FA and
MeOH

Fourier deconvolution of vCN for 0.25M NBu4SCN in MeOH

xvii

100

107

109

110

111

112

114

115

122

138

139

Figure 46. VCN stretch for ~0.25M NBu4SCN in DMF

Figure 47. VCN stretch for ~0. 1M NBu4SCN in NM with MeOH added

Figure 48. Fourier deconvolution of the vCN stretch for ~0.1M NBu4SCN
in NM as a function of MeOH added

Figure 49. Absorptivities for component bands of the vCN mode of ~0.1M
NBu4SCN in NM as a function of MeOH added

Figure 50. vCS stretch for ~0.1M NBu4SCN in NM as a function of
MeOH added

Figure 51. 1“N NMR chemical shift for ~0.1M NBu4SCN in NM as a

function of MeOH concentration and temperature

Figure 52. 1“N NMR line widths for ~0. 1M NBu4SCN in NM as a

function of MeOH concentration and temperature

Figure 53. Ortep structure for N aNCS°(DMF)2

Figure 54. Packing diagram for NaNCs-(DMF)2

Figure 55. Ortep plot for NaNCs-(NMF)2

Figure 56. Packing diagram for NaNCS°(NMF)2

Figure 57. Ortep plot for Na(12C4)2NCS

Figure 58. Packing diagram for Na(l2C4)2NCS

Figure 59. Ortep plot for Na(12C4)2NCS°MeOH

xviii

140

142

143

146

151

152

154

165

166

167

168

174

175

176

Figure 60. Packing diagram for Na(12C4)2NCS°MeOH 177

xix

1.

2.

Salts

Solvents

NBu4OCN
NBu4SCN
NBu4SeCN
NBu4NO3
NBu4Cl
NaSCN
KSeCN
KCl

BaO

P205

CaH2
NaClO4

NM
DMF
NMF
FA
MeOH

List of Abbreviations

Tetrabutylammonium cyanate
Tetrabutylammonium thiocyanate
Tetrabutylammonium selenocyanide
Tetrabutylammonium nitrate

T etrabutylammonium chloride
Sodium thiocyanate

Potassium selenocyanide
Potassium chloride

Barium oxide

Phosphorous pentoxide

Calcium hydride

Sodium perchlorate

Nitromethane
N,N-dimethylformamide
N-methylformamide
formamide

Methanol

XX

3.

Ligand

HMPA
H20
EtOH
PY
THF

AC
DMSO

12C4

Hexamethylphosphoric triamide
Water

Ethanol

Pyridine

Tetrahydrofuran

Acetonitrile

Acetone

Dimethylsulfoxide

1 2-Crown-4

xxi

Chapter I

Introduction

A. Introduction

Although electrolyte solutions play an important role in many biochemical and
industrial processes, our knowledge of the structure of such solutions, especially in the
case of nonaqueous and mixed solvents, is rather rudimentary.l Many types of interactions
may occur in an electrolyte solution. Their strength varies from weak van der Waals forces
to strong electrostatic attractions. This wide range of attractive forces and the plethora of
interaction sites result in a large number of solution equilibria that are difficult to sort out
experimentally. In an electrolyte solution, the types of interactions that occur can be broken
down into three broad categories: solvent-solvent, ion-solvent, and ion-ion. The relative
importance of the resulting equilibria depends upon the nature of the electrolyte and of the

solvent.

B. Solvent-Solvent Interactions

Interactions between neutral molecules have been of long standing interest. Van der
Waals invoked such interactions between gas molecules to explain the deviations from ideal
gas law behavior in 1873.2 The dispersion forces that produce the non-ideality are due to
transitory electric moments in molecules that arise from the motions of the electrons.3 In
nonpolar solvents, dispersion forces alone are responsible for the cohesion of the liquid
state. Larger deviations from the ideal gas law are seen in vapors of molecules with a
permanent dipole moment. In addition to the dispersion forces, polar molecules may
interact through dipole-dipole interactions and hydrogen bonding. The strength of cohesive
forces in the liquid state of polar molecules is reflected in the enthalpy of vaporization
(AH°V8P), as shown in Table 1. For molecules of similar molecular weights (MW), the
enthalpy of vaporization is dramatically larger for polar solvents than for nonpolar ones.

Moreover, the stronger attractive forces in polar solvents provide a local structure, or order,

Table l. AH°vap for selected solvents“

 

 

Molecule MW (amu) Aflow,p (kJ/mole)
CH4 16 8.2
C211,, 30 14.7
H20 1 8 40.6
CH3OH 32 35.3

 

 

that the nonpolar solvents do not have. The higher order of polar solvents is revealed by
the entropy of vaporization (AS°vap), as shown in Table 2. All nonpolar solvents have
approximately the same molar entropy of vaporization, known as Trouton's constant,
which illustrates that nonpolar solvents have a comparable amount of disorder.5 Since the
standard molar entropy of all vapors is roughly constant, this illustrates that nonpolar
solvents have a comparable amount of disorder in the liquid state. The larger entropy of
vaporization for polar solvents, such as water and methanol, reflects their greater

aggregation in the condensed phase.

Table 2. Entropy of vaporization (AS°,,,p) for selected solvents4

 

 

Solvent 115",“ UK" mole")
CCI4 86
C,H.,, 85
C6H6 87
H20 109
CquH 104

 

 

Hydrogen bonding between solvent molecules can result in a very strong
interaction. This phenomenon is one of the reasons that water has such unusual chemical

and physical properties.3 A hydrogen bond in a D-H---A system, where DH is the

hydrogen bond donor and A is the hydrogen bond acceptor, has a number of unique
characteristics. The H---A bond length is shorter than the sum of the van der Waals radii of
the two atoms and longer than the covalent bond length of HA. In the case of an ethanol-
pyridine hydrogen bond, the HA bond length is 1.80 A.3 This distance is dramatically less
than the van der Waals radius sum of 2.70 A. Moreover, the D-H---A bond angle is almost
linear, which may not be the most favorable dipole-dipole alignment.3 The strength of a
hydrogen bond, in the gas phase, can be deduced from the change in the stretching
frequency of the DH mode, since this bond weakens upon hydrogen bonding.6 In the gas
phase, the strength of hydrogen bonds lie between 10 and 35 kJ/mole.3

Table 3. Physical properties of selected solvents at 25°C7

 

 

Solvent MW (amu) bp ( 0 C) Dielectric Dipole Moment
Constant (D)
water 1 8 100 78 l .8
methanol 32 65 33 l .7
formamide 45 21 1 l 11 3.4
N-methyl- 59 l 80 l 82 3 .9
formamide
N ,N-dimethyl— 73 153 37 3.9
fomamide
nitromethane 61 101 37 3.6

 

 

Dispersion forces, dipole-dipole interactions, and hydrogen bonding help determine
a solvent’s unique set of physical properties. Examples of these physical properties for a
selected series of solvents are gathered in Table 3. The amide solvents qualitatively
show how intermolecular forces affect the physical properties of a solvent. Although

formamide (FA), N-methylformamide (NMF), and N,N-dimethylformamide (DMF) all

have large dipole moments, the strength of the cohesive forces is vastly different as seen
from the wide range of boiling points. The effect of hydrogen bonding can be clearly seen
by comparing N MF and DMF. These two solvents have the same dipole moment yet
greatly different dielectric constants. A comparison of nitromethane (NM) and NMF, two
solvents with similar molecular weights, shows the effect of hydrogen bonding on the
boiling point. The large difference in boiling points for N MF and FA suggests that the
hydrogen bonding is much stronger in FA than in NMF. An understanding of the structure
of these solvents would provide a basis for interpreting the differences in their physical
properties.

The structure of a number of solvents has been studied by X-ray diffraction,
neutron diffraction, and electron diffraction.8'l3 The role of hydrogen bonding, dipole-
dipole interactions, and dispersion forces in determining the three dimensional structure is
demonstrated by the amide solvents. A comparison of the radial distribution functions
(RDF) for FA, NMF, and DMF, as reproduced in Figure 1, shows the vastly different
structure of these solvents. The RDF shows both the local order, i.e. nearest neighbor
interactions, and longer range ordering. Beyond the immediate hydrogen bond interaction
at 290 pm (10"2 m), FA has longer range correlations (order) at 350-500 pm and at 700-
800 pm.9 Liquid FA consists of a three dimensional network of hydrogen bonded
molecules comprised of ring and chain structures. Liquid NMF also has an extensive
hydrogen bonding network, except that it is much more disordered at longer ranges. In
fact, N MP is essentially disordered beyond 500 pm.ll On the other hand, DMF, which has
no ability to form hydrogen bonds, displays no local order.14 The physical properties of
the amide solvents can be interpreted in light of their liquid structures. The dielectric
constant of NMF is much greater than that of FA (64%) yet the dipole moment in NMF is
only 15% greater than the dielectric constant of FA.7 Moreover, the dielectric constant of
NW is almost five times the dielectric constant of DMF, even though these solvents have

similar dipole moments. The wide range of dielectric constants in the amide solvents (37-

182) has been attributed to the differences in solution structure. FA and NMF have higher
dielectric constants than DMF, which reflects the greater amount of self association in the
hydrogen bonding amide solvents. The difference between FA and N MP has been
explained through the types of hydrogen bonding. Since FA has several ‘sites available for
hydrogen bonding, it can easily form cyclic dimers. The presence of cyclic dimers in FA
reduces the ability of this solvent to 'shield ionic charges despite the greater association seen
in neat FA versus neat NMF. These observations demonstrate that not only is an
understanding of the self association of a solvent helpful in understanding its physical
properties, but also the extent of the self association is important in determining the

solvent’s physical properties.

 

 

 

[Dill-11:10.1 / pm'I

 

 

 

 

L #1 l l 1 I 1
I] 200 400 500 800 IOCO
r / pm

Figure l. RDF for pure FA, NMF, and DMF”

 

C. Ion-Solvent Interactions
The strength of an ion-solvent interaction is determined by the properties of the ion

and the solvent. Ionic solvation is manifested primarily through ionic charge-solvent dipole
interactions. The donor—acceptor approach to molecular interactions has been helpful in
establishing an empirical scale that ranks the strength of ion-solvent interactions. In this
exposition, each solvent has a donor number and an acceptor number that are empirical
measures of the strength of a cation-solvent interaction and an anion-solvent interaction,
respectively.” The donicity of a solvent is defined as the negative molar enthalpy (in

kcal/mole) for the reaction of a donor (D) with SbC15 as a reference acceptor, both at 10‘3 M

in dichloroethane.‘5

D + SbC15 Z D-SbCls

The acceptor number (AN) is related to the NMR chemical shift of 31P in Et3P0 in the

solvent under study, where 8m is the 3'P chemical shift of EtjPO at infinite dilution in the
solvent relative to the 3'P chemical shift of Et3PO at infinite dilution in hexane and
55t3p0.SbC]5 is the chemical shift of the Et3POOSbC15 complex in 1,2 dichloroethane.l5
Therefore, the acceptor number is based on a scale from 0 to 100 where the acceptor

number of hexane is 0 and the EtaPO'SbCI5 complex is 100. The donor and acceptor

5cm, x 100

AN 5
5(Et3Po-8bC15)

 

numbers for many solvents have been determined. Generally, solvents that can hydrogen
bond have high acceptor numbers and solvents with amine and amide functions have large

donor numbers.”16

Table 4. Donor and acceptor numbers of selected solventsl6

 

 

Solvent Donor Number Acceptor Number
H20 l 8,30’1 55
MeOH I9 42
FA 24 40
NMF 24 32
DMF 27 16
NM 0 21

 

 

‘The donor number of 30 for water was determined spectroscopically”

Many techniques have been employed to investigate ion-solvent interactions. X-ray
diffraction has been used to study the hydration of many metal cations by determining the
metal-oxygen distance and the hydration number in aqueous solutions.‘4 Vibrational
spectroscopy has been employed to follow ion-solvent interactions by analysis of
perturbations in the normal vibrational modes of the ion or the solvent.18 The hydration of
the Al3+ ion has been studied by 27A1 and ‘70 NMR.19 Generally, ion-solvent interactions
observed by NMR spectroscopy are in the fast chemical exchange regime; however, the
solvent exchange for the Al3+(H20)n complex is unusually slow so that the hydrated
complex can be observed directly from the bulk solution.‘9

Popov and coworkers used 23Na NMR to determine the relative ion-solvent

complexation strengths of several nonaqueous solvents with Na+ (from NaClO4).20 The

strength of the solvation decreased in the order HMPA > H20 > DMSO > DMF ~ N MF ~
FA ~ MeOH > EtOH > PY >> THF > AC ~ AN >> NM.20 Except for pyridine, there is a
good correlation between the relative solvating ability of the solvents and their donor
numbers.20 The peculiarity with pyridine can be rationalized in terms of the 'hard-soft‘
interactions between the donors and acceptors. Pyridine is a softer base than the other
solvents in this study.”

Keil et a]. used proton NMR spectroscopy to determine that the hydration number
of Li+ is 4;22 this compares well with the solvation numbers of lithium in l-methyl-Z-
pyrrolidone (4.5),23 in acetone (4.3),24 and in DMF (3.8).25

Hertz and coworkers have obtained important structural information about ion-
solvent geometries from the study of relaxation times of magnetically active nuclei. The 7Li
and 'H relaxation times yielded a Li-O distance of 2.56 i 0.14 A for hydrated Li+ in
aqueous solutions.”5 NMR relaxation was also applied to determine the structure of the
fluoride ion hydrate and the solvation of I' by formamide.”28

Ion-solvent interactions can be observed directly by vibrational spectroscopy,
owing to the faster timescale of this technique.29 Cation hydration has been studied
through the metal—oxygen stretching mode {v(M-O)}. The frequency of the v(M-O)
vibration in metal hydrates correlates well with the metal-oxygen bond 1ength.3°'3' In
addition, the v(M-O) stretch has been used to determine the solvation number of Li+ in
various nonaqueous solvents?“32 Luck et al. found that alkali cations shift the v(OH)
stretch of H20 in nitromethane up to 60 cm‘l.33

The utility of infrared spectroscopy in the study of anion-solvent interactions is
exemplified by the thiocyanate anion. Since thiocyanate is a linear triatomic anion, it has
three fundamental vibrational modes: v3 {VCN}, vl {Va}, and the doubly degenerate v2
{850,}. It is important to point out at the onset that the labels VCN and VCS are purely

descriptive since the normal modes of these stretching vibrations are the result of

displacements in all the atoms in SCN‘.34 Perelygin et al. showed that the vibrational

10

modes of the thiocyanate anion are very sensitive to the nature of the solvent, especially
when the solvent can hydrogen bond with the anion.35 As shown in Table 5, hydrogen
bonding solvents, such as H20, D20, and MeOH, markedly affect the vCN stretch of the
thiocyanate anion. Protic solvents shift the vCN mode to higher frequencies and broaden the
spectral envelope. The shift and broadening of the vibrational modes of the SCN' anion
were also reported by Koda et al, who observed that the addition of water to a solution of
tetrabutylammonium thiocyanate (NBu4SCN) in DMF shifted both the vCN and vcS modes
to higher frequencies and broadened both spectral envelopes.36 Bacelon et al. assigned a
similar shift in tetrabutylammonium thiocyanate / phenol / CCl4 mixtures to a complex
where the phenol molecule is hydrogen bonded to the nitrogen end of the thiocyanate

anion.37'39

Table 5. Wavenumber and bandwidth of the VCN
mode of the thiocyanate anion in various solvents35

 

 

 

Solvent VCN (cm") AVCN (cm”, F WHH )
(CH3)2NCHO 2058 I 3
(C11,),so 2058 14
11,0 2076 36
13,0 2078 41
CH3OH 2068 5 l

 

Hochstrasser and coworkers studied the effect of hydrogen bonding solvents on the
vibrational relaxation of SCN‘ (as well as N3' and OCN‘) in D20 and MeOH, where they
found that the relaxation of the VCN mode was consistent with hydrogen bonding to the

nitrogen end of the thiocyanate anionm‘ Moreover, neutron diffraction studies of

ll

NaSCN in D20 and H20 by Kameda and coworkers show that the thiocyanate anion is
hydrogen bonded at the nitrogen end of the anion by 1.8 i- 0.2 water molecules with a
CN'"H20 angle of 120°.42

Gill et a1. inferred the solvation structure of the thiocyanate anion in a solution of
NBu4SCN in methanol from deconvolution of the infrared band envelope of the vCN
vibration.43 The spectral envelope was decomposed into six component bands as shown in
Figure 2. The analysis was based on the second and fourth derivatives of the vCN
absorption, which yielded the position and the number of component bands. These authors
assigned the component bands to hydrogen bonding by methanol to the nitrogen end and
the ON bond of the thiocyanate anion, based on infrared studies of the association of
phenol with various alkenes and alkynes,“4 where the hydrogen bond is pointed to the It
electrons of the unsaturated bond. In their assignment (Table 6), a hydrogen bond to the 7:
cloud of the CN bond lowers the frequency of the ch stretch, whereas a hydrogen bond to
the nitrogen end increases the frequency of the vCN stretch with respect to its value in the
free SCN’ anion. These novel assignments are a significant departure from results of
previous infrared studies of the thiocyanate anion in protic solvents and in ion pair
environments. Moreover, no crystal structures of any metal thiocyanates show such
bonding; rather, metal thiocyanate interactions are localized to either the nitrogen or the
sulfur end of the thiocyanate anion (vide infra).

Chabanel’s group has also studied the vcN stretch of NBu4SCN in MeOH. The vCN
mode was found to consist of only four component bands, and assignments of these
constituents (Table 7) were based on that group’s previous studies of alkali thiocyanate ion
pairs.45 The vCN mode for the free thiocyanate anion is assigned at 2058 cm".45 The lowest
frequency mode of the set is assigned to a structure similar to those seen in alkali
thiocyanate dimers, where two alkali cations bound to the nitrogen end of the thiocyanate

anion reduce the frequency of the vCN mode. A single hydrogen bond to the nitrogen or

12

sulfur end of the thiocyanate anion is assigned as raising the frequency of the vcN stretch,
which is analogous to the observations for M+NCS' and M+SCN’ ion pairs. Since both
CH3OHm'SCN and CH3OH---’SCN---HOCH3 should increase the frequency of the vCN
stretch, the authors were not able to conclusively assign the 2090 cm“ component.”
Previous studies of the ionic association of alkali thiocyanates in NMF revealed that
the vCN absorption of the thiocyanate anion is broad (~32 cm") and asymmetric.46
Moreover, the peak position and lineshape of the vCN stretch were found to be independent
of the cation used. Closer analysis of the spectral envelope suggested that the vCN stretch

was a convolution of three bands, at 2066, 2055, and 2045 cm", which are a result of

anion-solvent interactions.46

 

0.20
0.10 -
0.10 -
0.14
0.12
0.10
0.00 -
0.06 -
0.04
0.02 -

 

 

 

 

 

wavonumbor/cm ' ‘

Figure 2. The vCN stretch of NBu48CN in methanol and the
component bands determined by Gill and coworkers for the spectral envelopes“3

13

Table 6. Assignments of the six component vCN bands in
tetrabutylammonium thiocyanate in methanol by Gill et al.43

 

 

 

Assignment Peak position (cm")
noon,
' _ 2010
S- CE N
I
I
HOCH3
HOCH3
' _ ‘1 2034
3" CE N
S— CE N_ 2054
noon3
I _ 2068
S—CEN --H0CH3
i10CH3
uocn3
I _ 2076
s—CEN --H0CH3
S—CEN—--HOCH3 ano

 

 

14

Table 7. Assignment of the four component VCN bands
of tetrabutylammonium thiocyanate in
methanol by Chabanel and coworkers“5

 

 

 

‘ Assignment Peak position (cm°’)
,HOCH3
s— CE N" 2042
‘HOCH3
S—CE N‘ 2058
S-CE N'- - -HOCH3 2070
H3COH- - - “s—cs N
2090
or
H3COH- - - 's—CE N - - -H0CH3

 

D. Ion-Ion Interactions:

The vexing problem of electrolyte solutions is the determination of the degree of
ionic association and the factor(s) that affect this association. The two most important
solvent properties that have been shown to influence ionic association are the dielectric
constant and the solvating ability of the solvent.'6 The role of the dielectric constant is
obvious from Coulomb's law; solvents with a high dielectric constant are more effective at

screening opposing ions. A number of empirical scales have been proposed for a solvent’s

15

solvating ability, such as the donor number and the acceptor number.”47 These scales are
a measure of the strength of the ion-solvent interaction. Only solvents with both high
solvating ability and high dielectric constants will minimize ion pairing, whereas electrolyte
solutions in solvents with either low dielectric constants or low solvating ability will have
ionic association.

Ionic association in solvents of high dielectric constant (8 > 30) and donicity, has
been interpreted in terms of the Eigen mechanism”52 where A+ and B’ refer to the free

cation and anion, respectively, and S stands for the solvent molecule. A+SB' is
A+ + B' —‘-— A+SB' —"~— A+B'

a solvent-separated ion pair and A+B' is a contact ion pair. In solvents of a very low
dielectric constant (e < 20) and donicity, the ion pair can aggregate into dimers, (A*B')2,

tetrarners, (A*B')4, and higher order multimers, (A*B'),‘.53 In addition, an ion pair
A+B' —“— (A+B')2 —‘— (AIB'M _"~— (AIB'L.

can associate with either an anion or a cation to form a triple ion, which results in a charged

species as illustrated below.
A+ + A+B' "—‘~— AIB'A+

Ionic association was first studied by electrical conductance.54 Ostwald and
coworkers were able to distinguish free ions from ion pairs, since ion pairs do not conduct
current.l Electrical conductance has been recently applied to the study of ion pairing in
solutions of tetraalkylammonium salts and cryptate salts.”56 One drawback of this method

is that it cannot distinguish contact ion pairs from solvent separated ion pairs.

16

There are a variety of more modern methods to probe solutions. Of these,
multinuclear NMR and vibrational spectroscopy have had wide application to the study of
ionic association since these two techniques are sensitive to local electrostatic interactions
and are applicable to a wide variety of solution problems'w'60 One reason for this broad
application of NMR spectroscopy is that almost all elements have a magnetically active
isotope.59 Moreover, the resonance range of a given magnetically active nucleus is so large
that one rarely encounters overlapped signals from two different nuclei. Ionic association
greatly perturbs the electronic environment surrounding a nucleus, which can result in a
distinct difference between the chemical shift of the free ion and the ion pair. In addition to
this general advantage in specificity and the sensitivity of the chemical shift of a nucleus to
a local electric field, there are several special NMR methods that can be used to study
solution phenomena, such as nuclear relaxation and nuclear Overhauser techniques.57'58'60

The investigation of electrolyte solutions by NMR was pioneered in the mid- 19608
by Deverall and Richards in their study of the chemical shifts of alkali nuclei in a series of
salt solutions?“63 In all cases, the chemical shift depended on the concentration of the salt
and the counter ion used. These observations were extended by Bloor and Kidd, who
studied the concentration dependence of the 39K chemical shift of fourteen potassium salts
in aqueous solution.“'65 In all cases, the concentration dependence of the 39K chemical
shift, which results from cation anion interactions, was linear as shown in Figure 3.

The concentration dependence of the 39K chemical shift results from cation-anion
interactions. The linear dependence of the 39K chemical shift upon electrolyte concentration
is not indicative of contact ion pair formation, rather, it has been interpreted as collisional
ion pairs.57

In solvents of intermediate and low dielectric constants (8<30), contact ion pairs
exist.”70 The existence of contact ion pairs is observed in NMR spectroscopy by a

nonlinear dependence of the chemical shift on the solute concentration.” In the case of 1:1

17

 

0
v

”K chanted mm I”...

i

  

 

 

-l
o
. l A A A A A A

1 2 3 4 5 6 7
Potassium ion W’ mm 1120)

Figure 3. Concentration dependence of the ”K chemical shift in various solvents"3

electrOlytes undergoing exchange between free ions (M+ and X‘) and an ion pair (M+X‘),

the observed chemical shift can be expressed as a function of the mole fraction of the free
A + + B' = A+B'
ion (Xf) and the ion pair (Xip) and their respective chemical shifts, Sr and Sip.

Dabs = Xfo + Xipfiip = 5f + Xip(8ip - 5f) (1)

The ion pair formation constant is given by

18

. = (M+X') = C,-Cf
‘P (M*)(X‘> (3.21.

 

(2)

where Ct is the total electrolyte concentration, Cf is the concentration of the free ion, and
Y: is the mean ion activity coefficient of the salt. These two equations can be combined so

that 508. is a function of total electrolyte concentration, C,, with the parameters Kip, 719 5f,

—1 i (1 + 4K,I,y,2C,)“2
KipYiZCt

 

5bs =

O

(5.p - 5.) +5. (3)

and 61p. The ion pair formation constant can be determined by fitting equation 3 to the

observed chemical shift and the concentration of the electrolyte. The concentration
dependence of the 23Na chemical shift of NaBr, NaSCN, NaBPh4, and NaClO4 in
dimethylforrnarnide are illustrative of ion pairing.7| As shown in Figure 4, the chemical
shift of 23Na is nonlinear and of the form given by equation 3.

The relaxation time of a magnetically active nucleus can also be used to study ion
pairing?9 Hertz et al. showed that the ”F relaxation time is extremely sensitive to ion pair
formation. By measuring the concentration dependence of the relaxation time of the I9F
resonance in a solution of (CH3)4N*F in water or D20,”73 Hertz was able to determine
that the fluoride anion is ion paired in these solvents, even at low concentrations.73

Likewise, vibrational spectroscopy and more specifically infrared spectroscopy,
provides another perspective for studying ion-ion interactions. A significant advantage of
vibrational spectroscopy over magnetic resonance is the timescale of the measurement.
Since the lifetime of an ion-ion complex is on the order of 1098, ion pairs are not directly
observed by NMR spectroscopy; however IR spectroscopy with its faster timescale can

observe ion-ion aggregates directly29 via distinct vibrational bands corresponding to

19

different chemical environments.49 The strong electrostatic interactions within ion pairs that
affect the chemical shift of probe nuclei also perturb the vibrational modes of a molecular

ion involved in an ionic association.

5

oh

__A_. .

wk.
.\.\ .\o

\ .\-
4-1: \-

\-
\I SCN

 

 

ut-t
Nd!
U
3—1
u

10:!

Figure 4. Concentration dependence of the 23Na chemical shift
of NaSCN, NaBr, NaBPh4, NaClO4 in DMF7|

Although infrared spectroscopy can directly observe ions in multiple chemical
environments, the bands that result tend to be overlapped as shown in Figure 5.5' In order
to obtain quantitative information about each chemical species, the observed spectral
envelope must be deconvolved into its component bands. A common decomposition
procedure is to fit the experimental spectrum with synthetic Gaussian and/or Lorenztian
component bands.74 Petrucci et al. used Voight profiles (a Gaussian-Lorenztian product)

in order to fit the band envelope of the vcN stretch of N aSCN in THF. They were able to

20

identify three component bands that were assigned as a NaNCS ion pair (2057 cm"), a
dimer (2043 cm'l), and a triple ion (2074 cm'l), NaNCSNa".5|

 

 

 

 

 

Figure 5. Experimental infrared band envelope for the VCN stretch of NaSCN
in THF, with calculated component bands51

Chabanel and coworkers studied the ion pairing of alkali cyanates in
dimethylsulfoxide by following the vCN mode of the cyanate anion.” As shown in Table 8,
complexation of the cyanate anion with an alkali metal markedly affects the position and
intensity of the VCN band. The association was assigned as a M‘NCO’ interaction, with the
cation bound to the nitrogen end of the cyanate anion. This assignment is based on earlier
work with alkali thiocyanates.” The thermodynamics of these ion pairings were also
determined, and are summarized in Table 9. As is generally the case with ion pair

formations, the process is entropy driven.

21

Table 8. Spectroscopic parameters for the VCN mode
of the cyanate anion and alkali cyanates in DMSO75

 

 

 

 

Smies Position (cm”) Av,,L(FWHH, cm") Intensity (M" cm")
OCN' 2138 10.3 2150
LiNCO 2172 12.4 1920
NaNCO 2159 14.3 1650
KNCO 2148 13.1 1770
RbNCO 2 145 a a
CsNCO 2142 a a
adata not reported

The small enthalpy change accompanying the formation of an ion pair arises from
the Coulombic attraction between the cation and the anion, which produces a negative
enthalpy contribution, plus the enthalpy required to break the ion-molecule bonds involved
in ion solvation. In the alkali cyanate examples, the strong Coulombic attraction in the ion
pair is balanced by energy needed to disrupt the structure of the solvated ions. The positive
entropy contribution is the result of the loss of entropy due to the ion pair formation being
more than balanced by solvent molecules being liberated from the solvated ions. The
increased freedom of the solvent molecules is the predominant entropy contribution. The
study of the thermodynamics of ion pairing shows how important the solvent is in ion pair
formation.

Chabanel and coworkers have also studied the aggregation of LiNCO ion pairs to
form dimers and triple ions.7‘5'77 Addition of LiClO4 to a solution of LiNCO in DMSO
results in a new vibrational band at 2203 cm" and a reduction of the intensity of the ion pair

band at 2172 cm" in the vCN region. The 2203 cm‘1 band has been attributed to a triple ion

22

which may be LiNCOLi“, LizNCO“, or Li20CN+ since Chabanel and coworkers were not
able to conclusively identify this species.76 The same group has followed the infrared
spectrum of OCN' as function of LiNCO concentration in DMSO; new bands are observed
at 2189 cm", and 1330 cm’l at high LiNCO concentrations. Chabanel and coworkers have

assigned these bands to the lithium cyanate dimer, (LiNCO)2.77

Table 9. Thermodynamics of alkali cyanate ion pairs in DMSO at 298K75

 

 

Sfles K (M") AG" (kJmol’) AH" (kJmol’) AS" (JK"mol")
LiNCO 110:1:10 -ll.6i0.3 1.2:121 43:1:3
NaNCO 46 i 5 -9.5 :1: 0.5 3.2 i 1.5 43 i 5
KNCO 16 i 4 -6.9 :l: 0.8 0.5 :1: 1.5 25 :t 5

 

 

The sensitivity of the vCN stretching mode of the cyanate anion to its local
environment can be seen by comparing the solution studies of Chabanel and coworkers
with the gas phase studies of KNCO by Devere.78 The VCN stretch of KNCO in the vapor
phase consists of two bands at 2207 cm" and 2185 cm". In order to assign these bands,
Devere studied the vibrational spectrum of KNCO at several temperatures. Analysis of the
vapors by mass spectrometry shows that it is a combination of monomer and dimer
KNCO, and that the relative intensity of the (KNCO)2 correlates with the 2185 cm'1
infrared band. This is in contrast to the solution phase studies of the vCN stretch in lithium
cyanate, where the absorption due to dimers is lower in frequency than the ion pairs.
Moreover, the 59 cm‘1 difference between the VCN stretch of the KN CO ion pair in the gas
and solution phase reveals a substantial difference in polarization of the cyanate anion in

these media.

23

From these studies, it can be seen that the vCN stretch is quite diagnostic of the local
environment surrounding the cyanate anion. Ionic association affects the frequency, the
bandwidth, and the intensity the vCN mode. In solution, the frequency of the vCN vibration
increases in the order: OCN' < (MNCO)2 < MNCO < triple ion.

The ionic association of alkali thiocyanates has been studied in many nonaqueous
solvents.”82 Several types of ionic aggregates have been identified from their vibrational
spectra. These results, summarized in Table 10, show the assignments for ion pairs,
dimers, triple ions and tetramers. It is interesting to note that the vCN and vCS modes are not
shifted in the same direction upon complexation. In fact, this observation has been helpful
in identifying the structure in unknown metal thiocyanate complexes.83 Ion pairing in alkali
metal thiocyanates, which bond to the nitrogen end of the thiocyanate anion, increases the
frequency of both the vCN and vCS stretches. Ionic association to the sulfur end of the
thiocyanate anion increases the frequency of vCN mode but decreases the frequency of vCS
mode. Triple ions, where a metal cation is bound to each end of the thiocyanate anion,
have a VCN frequency above that of the corresponding ion pairs. Dimerization and
tetramerization of the ion pair reduces the frequency of the vCN mode below that in the free
anion . Pictorial views of the LiNCS ion pair, dimer, and tetramer are shown in Figure 6.

Through the careful study of many thiocyanate ion pairs, Chabanel et a1. has been
able to correlate the frequency of the vCS stretch with the polarizing power of the metal
cation. As shown in Figure 7, the polarizing power of the metal cation in a metal
thiocyanate ion pair is linearly related to the difference between the vCS stretch in the ion
pair and the vCS stretch in the free thiocyanate anion.85 In the extreme example of Al“, the
vCS stretch of the thiocyanate anion increases by more than 100 cm". Chabanel and
coworkers explain the changes in the vibrational frequencies of the thiocyanate ion in ion
pairs through the resonance structures that can be drawn for the thiocyanate anion. When a

positive ion interacts with the nitrogen end of the thiocyanate anion, the polarization in the

24

Table 10' complexes 0f metal thiocyanates and their vibrational
spectra in various aprotic sol vents79'82'84

 

 

Assignment vCN (cm") Vcs (cm'l)
SCN"I 2056 735
Ion Pairs
LiNCS‘I 2072 765
NaNCS‘ 2066 754
KNCS‘ 2056 742
RbNCSa 2056 742
CsNCSa ; 2056 742
AgSCNb I 2080 721
Triple Ions
Li’SCN'Li“ 2087 d
Na‘SCN'Na‘c 2081 “
Dimers
(LiNCS)2c 2034 788
(NaNCS)2c 2044 769
(KNCS);e 2042 758
Tetramer
(LiNCS); 1993 ‘1

 

 

“in DMF, bin DMSO, ° nitromethane, dobscured by solvent
and cin THF

25

”-0-. O 8
Ion Pair 0 C

O
0 Li

 

Dimer Tetramer

Figure 6. Pictoral view of the LiNCS ion pair, dimer, and tetramer85

anion shifts so that the negative charge on the nitrogen increases. As seen in the resonance
structures (Figure 8), the greater negative charge on the nitrogen of the thiocyanate anion is
accompanied by a reduction in the bond order of the CN bond and an increase in the bond
order of CS; the latter will give rise to an increase in the frequency of the of the vCS
stretch. Likewise, a positive electrostatic interaction to the sulfur end of the thiocyanate
anion will result in an increase in negative charge on the sulfur atom. This change in the
polarization in the anion will decrease the bond order of the CS bond and thereby lower the
frequency of the vCS stretch. This is observed for AgSCN ion pairs.80

The changes in the frequency of the VCN stretch that arise from ionic association
cannot be explained simply through the change in polarization of the anion since the
frequency generally increases whether the electrostatic interaction is on the nitrogen end or
the sulfur end of the thiocyanate anion. There have been several attempts to explain why

the VCN stretch does not reflect the resonance structures.”88 Recently, Chabanel and

 

 

 

A13”
850 — in”. ‘68“
OZnZ+
'e
3
8 800 ~ ca“
’ oMg2+
2+ 2*
Bo . CO
Li+
No+
750 - ”<2
.RD 1 4 l
P

Figure 7. Frequency (cm") of the vCS stretch in alkali
and alkaline earth thiocyanate ion pairs as a function of the
polarizing power of the metal cation85

 

Figure 8. Resonance structures for the thiocyanate anion

27

coworkers noted that there are two effects that can contribute to the shift of the vcN mode in
alkali and alkaline earth thiocyanate ion pairssz'gs First, the stabilization of the 0 molecular
orbitals, especially near the cation, in the M"NCS' ion pairs causes a strengthening in the
CN bond. Second, polarization from the sulfur to the nitrogen lowers the C-N bond order
as shown by the resonance structures. The authors derive this argument from their study of
a variety of metal thiocyanates, where they were able to correlate the shift in the vCN mode
to the polarizability of the metal ion (Figure 9) and the polarizing power of the metal cation,
using the following relation: AVCN = AP + BaP.82 Here, AVCN is the frequency shift of
the vCN mode from the value of the free SCN' in 98% THF and 2% DMF.82 A and B are

the fitted parameters with P and a the polarizability and polarizing power of the metal

 

.Tl

 

Figure 9. VCN/P of various metal thiocyanates
correlated to the polarization of the metal cations82

28

cation, respectively. The resulting vCN stretching frequency of a given metal thiocyanate
ion pair reflects a balance of the stabilization in the o orbitals and the reduction of the bond
order by the change in the polarization of the anion.”85 This balance is observed best in
the alkaline earth thiocyanates. The vcN stretch in the Be(NCS)+ and Ca(NCS)“ ion pairs
shift to higher frequency whereas the vCN stretch in a Ba(NCS)+ ion pair is at a lower
frequency than that of the free thiocyanate anion.82 The authors explain this difference in
the alkaline earth thiocyanates by proposing that the stabilization of the 0' orbitals falls off
faster with distance than the resonance effect, so that at long distances (or large cations) the
resonance effect predominates and lowers the frequency of the vCN stretch. The VCN stretch
of the dimers and tetramers are also lower than the free thiocyanate anion. In these
complexes, the coordination of the metal cations to the nitrogen end of the thiocyanate
anion is angular rather than linear, which results in an increase in the polarization of the
charge to the nitrogen end of the thiocyanate anion. The stabilization of the 0' orbitals will
be lessened since there is less overlap between the positive charge and the lone pair of the
nitrogen. The net result will be that the resonance effect will predominate and the
frequency of the VCN modes in these species will be lower than in the free ion.82

Chabanel and coworkers have modeled the ion pairing, dimerization, and
tetramerization in lithium thiocyanates through semi-empirical MN DO calculations.89
Although these results are not quantitatively reliable, the calculated vCN stretch reflects the
same trend seen in their previous experimental studies, where higher aggregation of LiN CS
reduces the frequency of the vCN stretch (T able 11).89 The energies of all the molecular
orbitals in the thiocyanate anion are stabilized by 5-6 eV in the ion pair and by an additional
0.5 eV in the dimer and tetramer, with respect to the thiocyanate anion. Moreover,
complexation of the thiocyanate anion by lithium increases the calculated bond length of
C-N from 1.177 A in the anion to 1.204 A in the dimer. The calculated atomic charge on

the sulfur atom is more positive in the order of tetramer > dimer > ion pair > free anion.

29

The atomic charge on the sulfur and the ON bond length reveal that a positive electrostatic
interaction on the nitrogen end of the thiocyanate anion results in change in the polarization
of the anion that can be explained by the resonance structures. Although the MNDO
calculations qualitatively reflect the electrostatic effects on the vCN stretch of the thiocyanate
anion, this level of computation poorly predicts the vCS mode in these complexes. Despite
the poor performance of the MN DO calculations on the vCS stretch, they provide support
for Chabanel's model of the electrostatic effects on the stretching modes of the thiocyanate

anion.

Table 11. MNDO calculations of the Von stretch in LiNCS complexes
compared with the experimental frequency85'89

 

 

Complex Calculated V0,, (cm") Experimental VCN (cm")
LiNCS 2313 2072

(LiNCS)2 2244 2034

(LiNCSM 2207 1993

 

 

The thermodynamics of the formation of several alkali thiocyanate ion pairs and
dimers were also determined; they are summarized in Table 12.70 Again, the ion pair
formation is entropically driven with a small positive enthalpy component. This large
entropy change results from the freeing up of solvent molecules that surround the unpaired
ions, thus providing greater overall disorder in the solution. The same is true for the dimer
and tetramer formation where the aggregation is also entropy driven. In fact, the larger
entropy term of the dimers and the tetramers is a result of greater release of solvent

molecules. The notable exception of the KNCS dimer is due to the poor solvation of the

30

K+ cation which is demonstrated by the small entropy increase as a result of the dimer
formation.

The ionic association of LiSeCN in several solvents has been studied by infrared
spectroscopy. Chabanel and coworkers used the vCN mode of SeCN‘ to identify the free
SeCN‘, the LiNCSe ion pair and the dimer, as shown in Table 13.”91 Ionic interactions
with SeCN' have a similar effect on the vCN mode as observed for the thiocyanate anion.
Ion pairing raises the frequency of the vCN stretch, whereas dimerization and tetramerization

lower the frequency of the VCN vibration.

Table 12. Thermodynamic parameters for the ion pair, dimer

and tetramer formation of alkali thiocyanates in various solvents70

 

 

 

 

Solvent Salt Kass (M' 1) AH" AS”
(It/mole") (JK’Imole'l)
Ion Pair
DMF LiNCS 2.4 3.5 34
Dimer
THF LiNCS 0.24 23 66
NaNCS 45 l 37
KNCS 60 -3 2
Tetramer
BuOEt LiNCS ‘ 29 l 30
‘data not reported

E. Model Crystal Structures
Several of the structures proposed for metal thiocyanate complexes in solution have

been confirmed through the use of single crystal X—ray crystallography. Clearly, the

 

31

crystal structure will not provide a perfect model of solution structure since it can emulate
only a limited number of the broad range of dynamic solute-solvent geometries present in a
solution. In spite of this drawback, a variety of solution interactions can be observed in
solids, such as ion-ion, ion-solvent, and solvent-solvent interactions. Since these types of
interactions in the crystal lattice can be easily varied by the appropriate synthesis, this
method can correlate a give crystal structure with complexation in solution by comparison

of the vibrational modes in each system.

Table 13. Frequency (cm") of the vCN mode of LiSeCN in various solvent9|

 

 

Solvent SeCN' LiNCSe (LiNCSe); (LiNCSe),
DMF’ 2066 2080 f '

MeCNb 2068 208 l f ‘
DEC‘ ‘ 2073 2040 ’
Bu20d ‘ 207 l 2029 '
TEA‘ ' f r 2000

 

 

‘dimethylformamide, bacetonitrile, cdiethylcarbonate, dbutylether
‘triethylamine, rnot observed or minor components

To understand the effect of ion pairing and higher order aggregation on the
thiocyanate anion in the solid state, the structure of a 'free' thiocyanate anion must be
obtained. Since the principle of electroneutrality must be maintained, a 'free' thiocyanate
anion can never be observed in the absence of a cation. As in the solution state, the
perturbation of the thiocyanate anion by a cation can be minimized through the use of bulky

cations. Crown ethers and cryptands have long been used to isolate cations from anions.92

32

LiSCN complexed by 4,7,13-tri0xa-l,10-diazabicyclo[8.5.5]icosane is just such a
system.93 According to the analysis of the crystal diffraction, the cryptand surrounds the
lithium cation in a three-dimensional structure so that the thiocyanate lies beyond the van
der Waals radius of the cation. The positive charge is well screened from the thiocyanate
anion, so this crystal provides a good model of the ‘free’ SCN‘ in the solid state. The CS
bond length is 1.72 i 0.03 A and the CN bond length is 0.97 i 0.02 A.93

The ion pair interaction and its effect on the geometry of the thiocyanate anion can
also be observed from crystal structures. The lithium thiocyanate ion pair is modeled by
the crown ether complex of LiSCN with 1,5,9-trioxacyclododecane (9-Crown-3) shown in
Figure 10.94 The lithium cation is coordinated to the nitrogen end of the thiocyanate anion.

The bond length of the CS bond is reduced to 1.635 :1: 0.006 A, the CN bond is extended

0.0

0

Figure 10. Li(9-Crown-3)SCN93

to 1.160 :1: 0.006 A, and the thiocyanate anion is almost linear at 178.3 i 05°. The lithium
cation is 1.958 10.009 A away from the nitrogen end of the thiocyanate anion and forms a
169.4 :1: 06° angle with the CN bond. The sulfur atom of the thiocyanate anion is not
within the van der Waal radius of any other atom. Bright and Trutes synthesized a similar

ion pair of RbNCS with 2,3,11,12-Dibenzo-1,4,7,10,13,16-hexaoxocyclo-octadeca-2,11-

33

diene (Dibenzo-l8-Crown—6).95 The ion pair structure is similar to the Li(9-Crown-3)SCN
structure.94 The rubidium cation is surrounded by the crown ether and the thiocyanate
anion is coordinated to the cation through the nitrogen end. In the rubidium complex, the
CS bond length is 1.58 i 0.03 and the CN bond length is 1.16 i 0.03; the thiocyanate
anion is significantly distorted from linearity, with an angle of 163 i 2°. The Rb-N bond
length is 2.94 d: 0.02 A and the Rb-N-C angle is 108.8 i- 0.02.. The sulfur atom of the
thiocyanate anion is 3.6-3.8 A from three of the carbon atoms belonging to one of the
benzene rings in the crown ether. The crystal structure of the UN CS complex supports
that proposed for the LiNCS ion pair in solutions, it is almost linear as would be expected,
and the perturbation of the anion's geometry follows the same trend as previous MNDO
calculations.89 On the other hand, the rubidium complex provides an example of the care
that must be exercised in drawing inferences to solutions from crystal structures, since
packing forces and other interactions, such as the benzene-thiocyanate anion interaction,
may occur in addition to the ion-ion interaction that one may wish to model.

A crystal structure having a distorted dimer-like structure was prepared by
Armstrong et al.96’97 The structure of this lithium thiocyanate tetramethylpropylene diarnine
(TMPA) is a distorted dimer that resembles two monomers in a loose association. This
structure differs from the symmetric dimer proposed by Chabanel85 where each thiocyanate
anion tilts from one cation towards another; both are shown below in Figure 11. The CN
bond lengths in the asymmetric dimer are 1.146 and 1.153 i0.002 A. The CS bond
lengths are 1.601 and 1.606 i 0.002 A. The shorter Li-N distances are 1.978 and 1.992 i
0.003 A with Li-N-C angles of 167.0 i 0.20 and 159.8 i 0.20, respectively. The longer
Li-N distances are 2.061 and 2.095 i 0.003 A with Li-N-C angles of 115.3 and 108.8 i
0.2°, respectively. In addition to solving the crystal structure for this complex, the authors
studied the vCN mode of the thiocyanate anion, which was measured at 2033 cm'1 in the
solid state. Moreover, infrared studies of dilute solutions of this complex in benzene,

where the dimer has dissociated into monomers, show that the vCN stretch is unaffected by

34

the monomer-dimer equilibrium and remains at 2033 cm". N o infrared bands in the 2060—
2070 cm'1 range, which would be indicative of LiNCS ion pairs, were observed in these
solutions. This careful approach of correlating the structure of solutions with the structure
of crystal complexes through the infrared vibrations shows that previously inferred solution

structures may be in error.

(I)

I
III
111
‘P
(I)
/

01

Li \
TMP

3’4,
>

Figure 11. Symmetric (LiNCS)2 dimer (left) and asymmetric [Li(TMPA)2NCS]2
dimer (right)”97

A more symmetric dimer structure was synthesized from KNCS and
6,8,9,10,12,13,20,21,23,24,26,27-dodecahydrodibenzo[b,n]-1,4,7,10,13,16,19,22-
octaoxacyclotetracosin (Dibenzo-24-crown-8).98 In this interesting complex shown
schematically in Figure 12, two potassium cations are coordinated by a single crown ether.
The CN bond length is longer than the LiNCS ion pair at 1.18 i 0.01 A and the CS bond
length is shorter than the LiNCS ion pair at 1.60 i 0.01 A. The two K-N-C bond angles
are 130.3 :1: 03° and 136.3 ‘1: 03°.

The triple ion [MNCSM]+, which has been identified in solution studies, is
supported by the KSCN acetonitrile oxide heptamer crystalline complex.99 In this complex
each [C is surrounded by a nitrile oxide ring. The thiocyanate anion forms a linear bridge

to two potassium cations through bonding at the nitrogen and sulfur ends

35

 

Figure 12. Diagrammatic representation of the coordination in (KNCS)2-DB24C899

Additional support is seen in bis(N-methylformamide) tetrakis(thiocyanato)
cobalt(II) mercury(lI),100 where the thiocyanate anion bridges the cobalt and mercury
cations. The cobalt cation interacts with the thiocyanate anion through the nitrogen end, the
mercury cation through the sulfur end. The CN bond length is 1.10 i 0.03 A and the CS
bond length is 1.67 :1: 0.02 A

The crystal structures of these thiocyanate complexes show that ionic association
perturbs the geometry of the thiocyanate anion in a way that is unique to each complex.

The perturbation of the thiocyanate anion by metal cations in solution has long been studied
by vibrational spectroscopy in order to identify possible structures of ionic association

complexes. The limited examples where the vibrational spectroscopy of known structures

36

with vibrational spectroscopy of unknown systems has shown that one can easily infer the
wrong structure from solution studies alone. More work needs to be done to correlate the
vibrational spectroscopy of crystal alkali thiocyanate complexes with solution values to .

determine if there is a strong relationship.

F. Proposed Work

The structure of anion-solvent complexes and the thermodynamics of anion
solvation are poorly understood, as exemplified by the investigations of the solvation of the
thiocyanate anion where there are some conflicting results. The research described in this
dissertation utilizes a variety of experimental and computational methods to help determine
the structure and thermodynamics of anion-solvent complexes. The thiocyanate anion,
along with the closely related OCN' and SeCN' anions, will be used to probe anion-solvent
interactions. Methanol and the amide solvents, NMF and FA, will be employed to allow a
wide variety of hydrogen bonding interactions. DMF is utilized as a reference.

The principal investigations of these interactions will be accomplished by a
combination of NMR and infrared spectroscopies. A qualitative picture of the number of
solvated species and the structures of the ion-solvates will be obtained by studying each
anion in the neat solutions of the hydrogen bonding solvents as well as in less interacting
solvents. Quantitative determination of the thermodynamics of the interactions will be made
with the help of an ‘inert’ solvent, nitromethane, where the anion and hydrogen bonding
solvent concentrations can be independently varied. Ab initio calculations of selected ion-
molecule species will be performed to help in the assignment of infrared bands and to
provide insight into the nature of the hydrogen bonding interactions. Finally, crystal
structures of appropriate solvate structures will be synthesized and their structures solved to

add additional support to proposed crystal structures and vibrational assignments.

Chapter 11

Experimental Methods

A. Materials

Methanol (absolute, EM Science, 500ml) was dehydrated by refluxing over CaH2
(5 g) for two hours. The dried solvent was distilled under a dry N 2 atmosphere onto 4A
molecular sieves (Linde). The fast 50 ml of the distillate was discarded and the next 300
ml was collected. N,N-dimethylformarnide (99%, EM Science), N-methylforrnamide
(99% Aldrich), and formamide (99% EM Science) were dried by adding 5 g of BaO
(Baker) to 500 ml of each solvent. Each mixture was allowed to sit for twelve hours, then
distilled over fresh BaO under reduced pressure onto 4A Molecular sieves. Nitromethane
(96%, Aldrich) was dried by the addition of 5g of CaH2 to 500 ml and allowed to sit for 24
hours. The dried solvent was distilled under reduced pressure onto fresh CaHz. All
solvents were stored in dark glass bottles in a dry box under a nitrogen atmosphere.

Tetrabutylammonium cyanate (>96%, Fluka) was dissolved into ethyl acetate, and
filtered to remove insoluble material. The solvent was removed and the purified cyanate
salt was dried under vacuum over P205 at 45°C for 24 hours. Tetrabutylammonium
thiocyanate (>99%, Fluka) was dried in a vacuum oven over P205 at 45°C for 24 hours.
Tetrabutylammonium selenocyanide was prepared from tetrabutylammonium bromide
(99% Aldrich) and potassium selenocyanide (98%, Aldrich). Potassium selenocyanaide
was dissolved in acetone and the solution was filtered to remove insoluble material. The
potassium selenocyanide was recovered from the acetone solution through precipitation by
ethyl ether. Tetrabutylammonium selenocyanide was prepared by mixing equirnolar
amounts of tetrabutylammonium chloride and potassium selenocyanide in acetone. The
solution was filtered to remove the KC] precipitate and the solvent was removed from the
filtrate to obtain NBu4SeCN. The crude tetrabutylammonium selenocyanide was
recrystalized in bis(2-ethoxy)ethane (Eastman). The purified tetrabutylammonium
selenocyanide was dried in a vacuum oven at 45°C for 24 hours. NaSCN (99.99%,

Baker) was dried at 120°C for 24 hours.

38

39

Solutions with known concentrations were prepared by weighing the solute on an
analytical balance into a 5 or 25 ml volumetric flask and diluting with the appropriate
solvent. All solution preparation was done in a dry box under a nitrogen atmosphere.
Ternary solutions of the tetrabutylammonium salt and hydrogen bonding solvent in
nitromethane were prepared from stock nitromethane solutions of the salt or the hydrogen
bonding solvent. Aliquots (l, 2, 3, and 4 ml) of the salt or hydrogen bonding solvent
solution were pipetted into 5ml volumetric flasks so that the concentration of the salt and of
the hydrogen bonding solvent could be varied independently.

Crystal solvates of NaSCN-NMF and N aSCN-DMF were prepared by slow
evaporation of saturated NaSCN (99.4% Baker) salt solutions in each solvent in a dry N 2
atmosphere. Na(12C4)ZSCN-MeOH was prepared by dissolving stoichiometric amounts
of N aSCN and 12-crown-4 (12C4, Aldrich, Inc.), in a 1:2 mole ratio, into a minimum of
dry MeOH. The solution was diluted to 15 ml with bis(2—ethoxy)ethane (Eastman). The
mixtures were allowed to slowly evaporate in a dry N2 atmosphere until the initial
formation of crystal solvates was observed. Once crystal solvates were visible, the
solutions were capped in order to slow the crystal growth. All crystals were harvested
from the mother liquor and placed in a sealed glass capillary. Likewise, Na(12C4)ZSCN
was prepared by dissolving stoichiometric amounts of NaSCN and 12-crown-4 (12C4,
Aldrich, Inc.), in a 1:2 mole ratio, into a minimum of dichloroethane (Baker) and diluted
with bis(2-ethoxy)ethane. The crystal structures were solved by Dr. Don Ward, the

Crystallographic Facility Manager.

8. Spectroscopic Techniques
Infrared spectra were recorded on a N icolet 520P (N icolet, Inc.) FT IR
spectrometer. Three sets of windows were used to cover the 4800-400 cm‘l spectral

region: Can (4800-900 cm'l), Irtran-2 (4800—700 cm'l), and Polyethylene (650-400

40

cm‘l). The resolution in all cases was 1 cm‘l- The number of scans was varied from 200
to 10000 depending on the signal intensity of the band studied. Teflon spacers were used
to vary the pathlength of the cell, the pathlength value being determined by the fringe
method.'°' The cell was flushed with 1-2 ml of the new sample to be studied between each
sample, and the spectrometer was purged for five minutes prior to the collection of a new
spectrum

Multinuclear NMR spectra ('H, 13C, MN, '70, and 77Se) were obtained on VXR-
300 and a VXR-SOO FT-NMR spectrometers (Varian, Inc.). Each sample was placed in a
5 mm NMR tube with a coaxial insert containing an external reference and lock solvent.
The parameters used for each nucleus are summarized in the following table. In addition,
the 170 spectra were 64K zero filled in order to increase the resolution. The temperature
probe in the spectrometer was calibrated by the known temperature dependence of the

proton chemical shifts in methanol.

Table 14. Parameters for NMR spectra

 

 

Nucleus Acquisition Time Dely Reference
'H l s l s 1% TMS in CD3C1
13C l s 8 s 1M NBu4NO3 in (CD3)2CO
1“N 1 s 0 8 1M NBu4NO3 in (CD,)2CO
”O 0.025 s 0 8 D20
77Se 1 s 9 8 1M diphenylselenide in (CD,)2CO

 

A crystal of bis(N-methylformamide)-sodium thiocyanate, CSHmN3OZNaS, having
approximate dimensions 0.08 x 0.10 x 0.42 mm, was mounted on a Riguaku AFC6S

diffractometer with graphite monochromated MoKor radiation and a 2 kW sealed-tube

generator. Cell constants and an orientation matrix for data collection, obtained from a

41

least-squares refinement using the setting angles of 20 carefully centered reflections in the
range 15.10 < 20 <19.85° corresponds to a monoclinic cell with dimensions: a=7.597 (6)
A, b=6.454 (5) A, 0:20.255 (4) A, v=922.1 (9) A, 13:92.58 (3)°. For 2:4 and
F.W.=l99.20, the calculated density is 1.334 g/em3. Based on systematic absences of
hOl: 1:1th and 0k0: k¢2n and the successful solution and refinement of the structure, the
space group was determined to be P2,/c.

The data were collected at a temperature of -90 i 3°C using the 00-20 scan technique
to a maximum 20 value of 500°. Omega scans of several intense reflections, made prior to
data collection, had an average width at half-height of 032° with a take-off angle of 60°.
Scans of (0.89 + 0.30tan0)° were made at a speed of 8.0°/min (in omega). The weak
reflections (I <10.00(I)) were rescanned (maximum of 3 rescans) and the counts were
accumulated to assure good counting statistics. Stationary background counts were
recorded on each side of the reflection. The ratio of peak counting time to background
counting time was 2: 1. The diameter of the incident beam collimator was 0.5 mm and the
crystal to detector distance was 285.0 mm.

Of the 2077 reflections which were collected, 1926 were unique (Rim = 0.023).
The intensities of the three representative reflections which were measured after every 50
reflections declined by -0.60%. A linear correction factor was applied to the data to
account for this phenomenon. The linear absorption coefficient for MoKa is 3.2 cm". An
empirical absorption correction, based on azimuthal scans of several reflections, was
applied which resulted in transmission factors ranging from 0.95 to 1.00. The data were
corrected for Lorentz and polarization effects. A correction for secondary extinction was
applied (coefficient = 0.6501 1x 10°).

The structure was solved by direct methodsm'103 The non-hydrogen atoms were
refined anisotropically. The final cycle of full-matrix least-squares refinement was based

on 1170 observed reflections (I > 3.00 o(I)) and 162 variable parameters and converged

42

(largest parameter shift was 0.01times its esd) with unweighted and weighted agreement

factors of:

R = Ell Fol - chll / ZlFol = 0.034
Rw = [(2 w(l Fol - chl)2 / Z wFo2 )]”2 = 0.041

The standard deviation of an observation of unit weight was 1.49. The weighting
scheme was based on counting statistics and included a factor (p = 0.03) to downweight
the intense reflections. Plots of [(2 w(| Fol - chl)2 versus lFol, reflection order in data
collection, sin0/A, and various classes of indices showed no unusual trends. The
maximum and minimum peaks on the final difference Fourier map correspond to 0.21 and
0.26 e'lA3, respectively.

Neutral atom scattering factors were taken from Cromer and Waber.‘°" Anomalous
dispersion effects were included in Fcalcf"5 the values for Af’ and At” were those of
Cromer.10° All calculations were performed using the TEXSAN crystallographic software
package of Molecular Structure Corporation.107

A colorless plate crystal of bis(N,N-dimethylformamide)-sodium thiocyanate,
C7HMN302NaS, having approximate dimensions of 0.1 x 0.22 x 0.38 mm, was mounted
in a glass capillary. All measurements and data analysis was the same except for the
differences described below.

Cell constants and an orientation matrix for data collection, obtained from a least-
squares refinement using the setting angles of 22 carefully centered reflections in the range
13.27 < 20 < 22.63°, correspond to an orthorhombic cell with dimensions:
a=9.626:l:0.002 A, b=17.869i0.004 A, e=6.6701~0.007 A, v=1147:1 A3. For 2:4 and
F.W. = 227.26, the calculated density is 1.315 g/cm’. Based on the systematic absences

of Okl: k¢2n and h01: l¢2n packing considerations, a statistical analysis of intensity

43

distribution, and the successful solution and refinement of the structure, the space group
was determined to be Pbcm.

A total of 1211 reflections was collected. The intensities of the three representative
reflections which were measured after every 150 reflections declined by 1.50%. A linear
correction factor was applied to the data to account for the phenomena. The final cycle of
full-matrix least squares refinement was based on 665 observed reflections and 102
variable parameters and converged with R = 0.065 and Rw = 0.072.

A crystal of N a(12C4)2SCN, CI7H32NNaOSS, having approximate dimensions
0.17 x 0.35 x 0.40 mm, was mounted on a Riguaku AFC6S diffractometer with graphite
monochromated MOK01 radiation and a 2 kW sealed-tube generator. Cell constants and an
orientation matrix for data collection, obtained from a least-squares refinement using the
setting angles of 20 carefully centered reflections in the range 15.10 < 20 < 19.85°
corresponds to a monoclinic cell with dimensions: a =15.745 (3) A, b=l4.445 (3) A,
e=19.454 (3) A, v=4411.2 (14) A3, 01:90", [1:94.47 (2)°. For 2:8 and F.W.=433.49,
the calculated density is 1.305 g/cm3. The space group was determined to P21/n.

The data were collected at room temperature (~24°C) using the 03-20 scan technique
to a maximum of 500°. The weak reflections (I<10.00'(I)) were rescanned (maximum of
10 rescans) and the counts were accumulated to assure good counting statistics. Stationary
background counts were recorded on each side of the reflection. The ratio of peak counting
time to background counting was 2:1. The diameter of the incident beam collimator was
0.5mm and the crystal detector was 285.0 mm.

Of the 6011 reflections collected, 5765 were unique (Rint = 0.0756). The intensities
of the three representative reflections which were measured after every 50 reflections
declined by 50%. A sixth order correction factor was applied to the data to account for this
phenomena. The linear absorption coefficient for MoKor is 3.2 cm". An empirical

absorption coefficient, based on azimuthal scans of several reflections. The data were

corrected for Lorentz and polarization effects. A correction for secondary extinction was
applied (coefficient=0.6501 1x10'°).

The structure was solved by direct methods.1 '3" ‘4 The non-hydrogen atoms were
refined anisotropically. The final cycle of the full-matrix least squares refinement was
based on 5671 observed (I>2.000'(I)) reflections and 762 parameters and converged (the
largest parameter shift was 0.01 times its esd) with unweighted and weighted agreement

factors of:

R=211F.1-1F.11/21F.1=0.0881
Rw =10: w(| F01 — | F, |)2/Zw F02]"2 = 0.2020

A crystal of Na(12C4)ZSCN-HOMe, CI8H36NNaO9S, having approximate
dimensions 0.25 x 0.25 x 0.10 mm, was mounted on a Riguaku AFC6S diffractometer
with graphite monochromated MoKor radiation and a 2 kW sealed-tube generator. All
measurements and data analysis were the same except for the differences described below.
Cell constants and an orientation matrix for data collection, obtained from a least-squares
refinement using the setting angles of 20 carefully centered reflections in the range 15.10 <
20 < 19.85° corresponds to a monoclinic cell with dimensions: a =8.427 (3) A, b=20.695
(3) A, c=7.818 (2) A, V=1210.8 (6) A3, 01:90.01 (2)°, l3=117.37 (2)°, 7:89.99 (2). For
2:2 and F.W.=465.53, the calculated density is 1.277 g/cm3. The space group was
determined to P2,/m.

Of the 2395 reflections collected, 1176 were unique (Rim = 0.1064). The intensities
of the three representative reflections which were measured after every 50 reflections. A
linear correction factor was applied to the data to account for this phenomena.

The structure was solved by direct methodsm'103 The sodium, oxygen, and
carbon of the methanol were refined anisotropically. The C-O, C-C-O, C-O-C, C-H, H-H,

C-C-H and C-C-H were constrained to have the same bond lengths for each respective

45

group. The final cycle of the full-matrix least squares refinement was based on 1166
observed (I>2.000'(I)) reflections and 282 parameters and converged (the largest parameter

shift was 0.01 times its esd) with unweighted and weighted agreement factors of:

R=211F.1-IF.11/21F.1=0.0656
Rw = [(2 w(| F01 - 1 Fc 1)2/2w F02]"2 = 0.1724

The largest residuals in the electron density was +0296 and -0. 147 e.A'3.

C. Theoretical Calculations

Ab initio calculations were performed by Gaussian 92/DFT (Gaussian, Inc)'°° on
several Indigo workstations (Silicon Graphics, Inc.). The standard basis sets employed
were as follows: d95v+* for Li, C, N, O, and F 109; d95+* for S”; and the 6—31+G*”°
basis set for Na and Mg. The Watchers basis set with polarization functions was used for
K and Ca' ' "' '3 and a Huzinaga basis set for Se.114 The four term Huzinaga basis set with a
polarization function was employed for H.1 '5 Second order Moller-Plesset perturbation
theory (MP2) with only the valence orbitals active and standard convergence criteria were
used to optimize the geometry of the species under study. Vibrational frequencies and the
infrared intensities of the normal modes were computed analytically except in a few cases
where numerical computation was dictated by memory constraints. Atomic charges were
determined by the Natural Bond Orbital (NBO) analysis algorithm.l '3 The electrostatic
potentials for selected systems were generated by Gaussian 92 and stored in a cube file.
Visualization of the electrostatic potential and the molecular orbitals were accomplished by
Scianl ‘7 in order to create isosurfaces and contour plots. Isosurfaces of the molecular
orbitals in selected species were visualized by importing the Gaussian output file into

Scian. The potential energy distribution (P.E.D.) of the stretching modes was computed

46

for each normal coordinate, Q, by using the usual relation, equation 4.1 ‘2 The off diagonal
elements were ignored since the Fii terms are larger than the Fij terms (i¢j). The Lik matrix
roar-1,3,,
P'E'D'diag = ——u-T% (4)
Zr 111117:
elements were determined by converting the normal coordinates from the Gaussian output,

which are in cartesian coordinates, to internal coordinates.

D. Data Treatment

The number of vibrational components in a spectral envelope was determined by
Fourier self deconvolution, second derivative analysis, and curve fitting. Fourier self
deconvolution and second derivative analysis were performed via the PCIR program
(N icolet, Inc.). In the deconvolution procedure, the width of the resolved peak was
adjusted to 10-15 cm'l and the deconvolution factor was varied, 2-4. The deconvolution
procedure was applied to small spectral regions (100 cm'l) having low noise (<0.0005 A),
in order to minimize the noise in the deconvolved spectrum.

Spectral envelopes were fitted to Gaussian/Lorenztian bandshapes in order to
detennine the number of vibrational components and yield quantitative information about
each component band. Sigma Plot (v. 4.14, Jandel Scientific, Inc.) was used to curve fit
the infrared band envelopes. Spectral regions of interest, generally 100 cm", were
converted to ASCH files and imported as text files into Sigma Plot. The

Gaussian/Lorenztian bandshapes had the following form:

A = Zeal—1.10M + l,L(v)]
i (5)

where

47
I

L =
(V) 1+(v_vi72 (6)
W1

e-ln2(v-v,- )2/w,-2

 

and

GM = (7)
The calculated absorbance, A, is a sum of Gaussian, G(v), and Lorenztian, L(v) ,
lineshapes, where the contribution of each lineshape can be varied by the parameter lg. The
independent variable v is the wavenumber(cm'1) of the spectrum. The parameters V), w,
and ai are the peak position, width of the band, and intensity at the peak, respectively for a
component band. The parameters are adjusted through nonlinear regression in order to
minimize the difference between the calculated absorbance and the experimental
absorbance. In addition to the Gaussian/Lorenztian sums, a linear offset is added to the
regression model to compensate for subtraction errors between the sample and reference
spectra. The initial guess used for the fitted parameters is derived from the Fourier self
deconvolution and the second derivative analysis. These procedures yield the number of
vibrational components, their position, and relative intensities. The uncertainties in the
parameters and the residuals of multiple fits to a given experimental band were compared in

order to determine which fits were statistically significant.

Chapter III

Theoretical Calculations

A. Introduction

The OCN', SCN' and SeCN' anions have been employed extensively as probes in
the study of ionic associations and other interactions in solutions.75‘8"82'8°"19"20 The
stretching modes of these linear anions are very sensitive to the local electrostatic
environment.82 Ionic association markedly perturbs the electronic structure and bonding,
and the vibrational modes are shifted (in either direction) from the frequencies of the ‘free’
anion, where the values of the ‘free’ anion are those of tetrabutylammonium cyanate,
thiocyanate, or selenocyanide in a high dielectric aprotic solvent. Examples from the
experimental literature are collected in Table 15.75'77'8°82‘84"18" '9 The identification of a
number of solution species, such as ion pairs, dimers, tetramers, and triple ions has been
accomplished by analyzing the vibrational perturbations. These investigations have been
crucial in understanding the factors involved in ion pair, dimer and tetramer formation.”74

Ion pairing by alkali or alkaline earth metal cations with OCN', SCN‘ and SeCN'
anions generally raises the frequency of both stretching modes (VCN and vex) with respect to
the free anion.”82 The largest of these frequency shifts for alkali metal cyanate and
thiocyanate ion pairs are observed for the lithium complexes. Likewise in alkaline earth
metal thiocyanate ion pairs the greatest shift in the vibrational modes is found for the
beryllium ion pairs. Unusually, the VCN stretching vibration (v3) of the Ba2*NCS' ion pair is
red shifted from that of free SCN'.82

Dimerization of alkali metal thiocyanate and selenocyanide ion pairs, as well as
tetramerization of lithium thiocyanate, reduce the frequencies of the vCN stretches with
respect to those of the free anions, with the largest effect seen in lithium complexes?“ '8
The frequency of the ch (vl) stretching mode is higher than the ch stretch in the
corresponding ion painm‘"82 It should be noted that the labels vCN for v3 and ch for V]

have historically been employed as approximate descriptors of these stretching vibrations;

the normal modes involve motions of all three atoms.

49

50

Table 15. Frequencies of the stretchingsrgiodes of OCN‘, SCN' and SeCN‘ species in

 

 

 

solution 1,82,84,119.120
Species VCN (cm") ch" (cm")
OCN’b 2138 1283, 1196h
LIINCO'b 2172 1304, 1218h
NaiNCO'b 2159 1298, 1212h
ICNCO’b 2 148 1 290, 1204h
LIINCO'Li+b 2203 i
SCN" 2058 735
BCZINCSf 2 l 1 2 i
MgeNcsr 2083 789
CaZINCS'c 2063 780
BaZINCS’c 2047 774
LiJ'NCS‘ 2072 765
Na’NCS‘ 2065 754
KINCST 2058 742
Ag+SCN4i 2080 721
(LIINCS')2c 2034, 2051(R)j 788, 813(R)l
(N a”NCS')2° 2042, 2052(R)’ 769, 772(R)‘
(Li"NCS'),,c 1993 , 2022(R)l '
Li‘SCN'Li”r 2087 '
Na‘SCN'Na+r 2081 '
SeCN‘° 2066 54 l
LAINCSC'c 2080 '
(Li +NCSe')2‘ 2040 '

 

’X can be 0, S, or Se; °in DMSOm’ ; cin DMF”; °in dimethylthioforrnamideso"2°;
cin THF“; rin CH3N028‘; l‘in diethylcarbonate'”; hthe VCO is in Fermi resonance with the overtone
of the bending mode (280cm), resulting in two bands”; 'obscured by solvent; ’Raman bands denoted by (R)

51

Ion pairing interactions by a given cation produce substantially different
perturbations in the frequency of the two stretching modes in these anions. In Li+NCO’,
Li+NCS', and Li+NCSe' ion pairs, the frequencies of the vCN modes are higher than those in
the corresponding free anions by 34 cm", 14 cm’1 and 14 cm", respectively. The
frequency of the vCO stretch is increased by 21 cm’1 in LiNCO from OCN' and the
frequency of the vCS stretch is increased by 30 cm“l in Li“ NCS' from SCN'. Moreover, the
VCN stretch of Li*NCO‘Li" is blue shifted 65 cm’1 from that in OCN' whereas the vCN stretch
of Li”NCS'Li+ lies 29 cm’1 higher than that in SCN'. It appears from these experimental
studies that the VCN mode in OCN‘ is more sensitive to electrostatic interactions than the vCN

mode in either SCN' or SeCN’. Additionally, the vCS mode in SCN' is more sensitive to an

 

electrostatic interaction than is the vCO mode in OCN'.

The basis for these frequency shifts in alkali and alkaline earth metal cyanate,
thiocyanate and selenocyanide ion pairs and dimers are of interest since these characteristic
shifts have been used to identify numerous inorganic complexes of these anions. A
number of authors have discussed the basis for these frequency shifts.”75 318°”
N akamoto proposed that 0’ donation from the highest occupied 0' orbital in the SCN‘ anion
to the cation in an alkali thiocyanate ion pair raises the frequency of the vCN mode, since
electrons are removed or depleted from the 0' orbital, which is a weakly antibonding
orbital.88 Chabanel and coworkers suggested that the frequency shifts in ion pairs and
dimers could be rationalized through a valence bond approach.”82

MNDO calculations of lithium thiocyanate ion pairs by the Chabanel group support
their rationalization for the frequency shifts in alkali thiocyanates.89 These calculations
show that ion pairing to the nitrogen end of the thiocyanate anion by lithium cations shifts
the negative charge distribution of the anion to the nitrogen end. The molecular orbitals
belonging to the thiocyanate anion in a Li"NCS' ion pair are lowered in energy by 5-6 eV,
with the 40 orbital (valence orbital in SCN') showing the largest drop.” In dimers and

tetramers, the energy of the 40 orbitals in the SCN' anion is slightly more negative than in

52

the ion pair, with an even greater charge transfer from the sulfur to the nitrogen end of the
anion. Calculations for a Li”NCS' ion pair show that the ON bond is longer and that the
GS bond is shorter than those in SCN'. Dimerization and tetramerization of the Li*NCS'
ion pairs further shift the negative charge toward the nitrogen end of the anion, which
lengthens the ON bond and reduces the GS bond, with only a small concomitant
reduction in the energy for the molecular orbitals. The MNDO calculations of the
vibrational frequencies for lithium thiocyanates show a balance between charge transfer and
the stabilization of the molecular orbitals in ion pairs, dimers and tetramers and qualitatively
reflect the experimental trends for the vCN mode. The frequency of the VCN stretch is
predicted to decrease in the following order: LYNCS' > (Li"NCS')2 > (Li‘NCS'),. On the
other hand, the calculated frequencies for the vCS stretch in these lithium thiocyanate species
do not correlate well with experimental results. One peculiar result is that the empirical
model predicts that ion pairing should reduce the ON bond length, yet the calculations
show that the C-N bond in lithium thiocyanate ion pairs is lengthened. However, the
lengthening of the C-N bond in Li‘NCS’ ion pairs does not reduce the frequency of the vCN
stretch. Although the model proposed by Chabanel and coworkers to rationalize the
frequency shifts in alkali thiocyanates is based on extensive experimental and theoretical
studies, inconsistencies with the experimental observations remain.

The practice of employing molecular orbitals to explain bond lengths in diatomics
(ions and neutral molecules) has shown that the bond order (charge density) can be related

'2' While this approach is empirical in nature, it

to the bond length of a diatomic species.
provides an intuitive understanding of how the degree of bonding relates to the bond length
between two nuclei. The electrostatic theorem proposed (independently) by Feynman and
Hellmann states that the buildup of charge density between two nuclei produces attractive
forces between these two nuclei whereas electron density outside the internuclear axis

produces forces that pull the two nuclei apart.122 Berlin equated the attractive forces to

bonding regions and repulsive forces to antibonding regions.123 Bader has applied this

53

approach to explain bonding in both diatomic and polyatomic molecules and has shown a

'24 Howard

relationship between the electron density and bond lengths between two nuclei.
has correlated the C-N bond lengths in OCN' and SCN' anions with the ON bond
strengths of these anionsm He rationalized this correlation by showing that electron
density around the ON bond is greater in SCN’ than in OCN'.‘25

An obvious criticism of using ab initio or semi-empirical computations to model ion
pairing in solution systems is that these calculations lack the dielectric medium and
solvation effects that are present in solutions. The vibrational modes of K*NCO' and
K”NCS‘ ion pairs in the vapor and solution phases have been studied by Devore.12° He
assigned a band at 2206 cm‘1 in the infrared spectrum to the vCN stretching mode of
K”NCO' in the gas phase; this vibration is substantially blue shifted from its value of 2148
cm'1 in DMSO solutions and its value of 2124 cm" in the gas phase.”126 A band at 2185
cm‘1 that grows with higher K*NCO' densities is assigned to the infrared active vCN
stretching mode in (K’NCO’)2.78 Similarly, Devore assigned a band at 2070 cm‘1 to the vCN
stretch of the K"NCS' ion pair, and a band observed at 2055 cm" at higher K*NCS'
concentrations to the infrared active vCN mode of (K"NCS')2.78 The value of 2070 cm’1 for
the VCN stretch of KiNCS' in the gas phase is close to the 2058 cm’1 band attributed to the
KNCS ion pair in DMSO solutions.82 These experimental studies suggest that the OCN'
anion is much more sensitive to medium effects such as ion solvation and the dielectric
environment than the SCN‘ anion. It is also clear that great care must be used when
comparing theoretical predictions to experimental solution values.

The OCN' and SCN' anions have also been employed in investigations of anion
solvation, particularly in hydrogen bonding solvents, since the perturbation of the
vibrations upon ionic association are well known, although the basis for the frequency
shifts are not entirely understood.3°'°3"27"28 Several authors have assigned a blue shift in
the VCN mode of SCN’ in protic solvents to a hydrogen bond at the nitrogen end of the

thiocyanate anion.”43 In a recent study of the solvation of the thiocyanate anion in

54

methanol (MeOH), Gill and coworkers determined that the vCN absorption envelope of
SCN' consisted of six component bands.43 Based on earlier observations on acetylenic and
ethylenic stretching modes where hydrogen bonding to the it cloud of these bonds lowers
the frequencies of the carbon-carbon stretching modes“, they assigned red shifted bands
with respect to ‘free’ SCN' anion to complexes with MeOH hydrogen bonded to the it
cloud of the C-N bond. Blue shifted bands were assigned to a single MeOH molecule
hydrogen bonded to the nitrogen end and to one and two MeOH molecules hydrogen
bonded to the 11: cloud of the ON bond in conjunction with a hydrogen bond to the nitrogen
end of the thiocyanate anion.4°

Much vibrational spectroscopic work has been devoted to the study and
identification of ionic and hydrogen bonding complexes with OCN', SCN', and SeCN'
anions. Unfortunately, the number of theoretical calculations to support these vibrational
assignments and to understand the nature of the observed frequency shifts is very limited.
With the exception of the Chabanel group’s calculation of lithium thiocyanates”, described
above, theoretical studies of these anions have primarily focused on the structure and
energetics of ionic and hydrogen bonded complexes, without much discussion of how
these interactions perturb the anion.'29"°5 The purpose of our work is to explore the
correlation between structure and vibrational frequencies. Towards this goal we will
calculate the vibrational frequencies for alkali and alkaline earth metal XCN' complexes
(X=O, S, or Se) that have been identified in solutions, ascertain that the level of theory is
adequate to describe such electrostatic interactions, and apply the theory to hydrogen
bonded complexes in order to assist in the assignments of these species in experimental
solution studies. This investigation also provides insight into the basis for the frequency
shifts observed for ionic and hydrogen bonded complexes by predicting the structural
implications, changes in the force constants, shifts in the distribution of the negative
charge, and perturbations of the molecular orbitals accompanying ionic and hydrogen

bonding interactions.

55

B. Results
1. OCN, SCN', and SeCN'

Ab initio calculations of the OCN‘, SCN', and SeCN' anions provide a basis for
understanding how these anions interact with and are perturbed by a positive electrostatic
interaction. The equilibrium bond lengths, vibrational frequencies, and atomic charges for
these anions are summarized in Table 16. The C-N bond lengths decrease from OCN' to
SeCN', which reflects the greater triple bond character of SeCN‘. The structural parameters
for these anions are comparable to a number of previous theoretical

calculations.”203733.3032

Table 16. Calculated equilibrium bond lengths, stretching frequencies, and
atomic charges for OCN', SCN', and SeCN'

 

 

Anions d(CN) d(CX)‘ VCN cha Charge Charge Charge
A A cm" cm" @ X“ @ C @ N

OCN' 1.21 8 1.244 2155 1205 -0.887 0.796 -O.909

SCN' 1 .208 1.671 1979 745 -0.514 0. 179 -0.665

SeCN' 1.205 1.770 2003 617 -0.350 0.009 -0.659

 

 

awhere X can be 0, S, or Se

Interestingly, the frequency of the vCN stretching mode does not correlate with the
length of the ON bond in these anions. The reason for the deviation from Badger’s rule136
is that the normal modes for the vCN stretch in these anions are not localized to the C-N
bond, especially for OCN'. The vCN vibration in OCN' is similar to the asymmetric stretch

in N3' or CO2 in that the normal mode involves significant displacements of all atoms,

56

whereas the VCN stretches of SCN' and SeCN' are more localized to the C-N bond. The
dependence of the vCN and VCX modes of the OCN', SCN', and SeCN' anions on their
respective bonds can be characterized quantitatively by the RED, which relates the
contribution of each bond stretch to the potential energy of the normal mode.l '2 The
results, summarized in Table 17, show that the stretching modes of the anions involve both
the ON and C-X bonds. As the mass of the chalcogen increases, these modes more
closely resemble a pure C-N or C-X stretch. Yet, this analysis shows that even in the
extreme example of SeCN', the vCN and vcSe modes should not be thought of as a pure C-N

or C-Se vibration.

Table 17. Potential energy distribution and force constants for the stretching modes
of OCN', SCN', and SeCN'

 

 

Anions V”, v” Fa, (N/m) a“ (N/m)
OCN' 1350 1113
%CN 62 41
%CO 38 59
SCN' 1400 502
%CN 88 8
%CS 12 92
SeCN' 1432 471
%CN 90 7
%CSe 10 93

 

 

“where X can be 0, S, or Se

The predicted force constants for OCN', SCN', and SeCN' provide insight into the
nature of the ON and C-X bonds. The C-N force constant, FCN, varies in the following

order: OCN' < SCN' < SeCN’, showing that the SeCN' anion has the greatest triple

57

bond character of this series. The force constants for the C-X bond change in the
following order: OCN' > SCN' > SeCN'. These trends are consistent with the
resonance structures that can be drawn for these anions, such that greater triple bond
character in the ON bond comes at the expense of the C-X bond order.

The predicted frequencies of the VCN modes in OCN‘, SCN', and SeCN‘ correlate
well with observed values for these anions in aprotic solutions. As in the experimental
studies, the predicted vCN frequency decreases in the order OCN' > SeCN' > SCN’. The
calculated values for vCN, shown in Table 16, are within 3% of the experimental solution
frequencies (T able 15). The predicted frequency of the vCO mode in OCN' and the vCS
stretch in SCN' also compare well with experimental solution values for these anions. No
experimental comparisons can be made for the vcSe stretch due to the lack of experimental
results.

The atomic charges, as calculated by the NBO procedure and listed in Table 16,
show that the negative charge is delocalized in these anions. In each case, the nitrogen
atom has the largest concentration of negative charge. An indirect measure of the charge
density around the nitrogen can be obtained experimentally by 14N NMR spectroscopy.
The 1”N chemical shifts of OCN', SCN' and SeCN' are -288 ppm, —165 ppm, and -137
ppm, respectively.83 The large difference in the nitrogen atomic charge calculated between
OCN‘ and SCN’ (Table 16) is supported by the dramatic change in the 14N chemical shift
from OCN' to SCN'. The small predicted difference in the atomic charge on the nitrogen
atom between SCN' and SeCN' is also concordant with experiment. The differences in the
l“N chemical shifts of OCN', SCN', and SeCN' have been rationalized in terms of the
change in the it cloud surrounding the nitrogen nuclei in these molecules.83 It is important
to note that these calculated charges do not agree with the implications of the resonance
model (Figure 15). Although the force constants and the bond lengths for these anions
imply that the C-N bonds in the SCN' and SeCN' anions have a greater triple bond

58

character than the C-N bond in OCN', the calculated atomic charges on the chalcogens in
SCN' and SeCN’ are less than those for the nitrogen atoms.

The valence molecular orbitals for these anions, reproduced in Figures 13-15, show
striking differences in shape across the series. The electronic configurations of the valence
molecular orbitals are 102 20’2 30‘2 111:4 40‘2 21t4 for OCN', with the 11t° and 40’2 reversed
for SCN' and SeCN'. The 10' and 20 orbitals are primarily bonding in OCN', SCN', and
SeCN'. The 30' and 40 orbitals are primarily antibonding. There is a remarkable difference
between the 11: orbitals in OCN’ and in SCN' and SeCN'. The maximum of the 11: orbitals
shifts from the chalcogen end in OCN' toward the nitrogen end in SeCN' with an
intermediate structure in SCN'. The 2n M08 in OCN' are slightly bonding for the ON
bond and antibonding for the CO bond. On the other hand, the 211: M08 in SCN' and
SeCN' are slightly antibonding for the ON bond and slightly bonding for the CS and
C-Se bonds.

2. Electrostatic eflects on OCN', SCN' and SeCN

The two resonance structures that can be drawn for OCN', SCN‘ and SeCN',
diagrammed for SCN', suggest a number of implications that can be tested by ab initio
calculations. This model predicts that electrostatic interactions to the nitrogen end of these
anions should shift the negative charge density toward the nitrogen atom, with a
consequent reduction of the C-N bond order and reduction of the C-X bond length, where
X can be 0, S, or Se. The perturbations in the structure of these anions by electrostatic
interactions should be reflected in changes in the force constants and bond lengths. ‘
Likewise, an electrostatic interaction to the chalcogen end should produce an opposite shift
in the electron density, resulting in an increase in the strength of the ON bond and a
reduction in the strength of the C-X bond. The shift in the charge density of these anions

as the result of an electrostatic interaction can be followed by the NBC analysis, which

 

59

   

   

Figure 13. Contour plots for the valence molecular orbitals in OCN'
Green to Blue (-.1 to -1.7), Green to Red (.1 to 1.7); stepsize 0.1

   

   

11:my 27‘x.y

Figure 14. Contour plots of the valence molecular orbitals for SCN
Green to Blue (-0.1 to -1.7), Green to red (0.1 to 1.7); stepsize 0.1

 

16

 

111:,“y

61

Q)

20

Figure 15. Contour plots for the valence molecular orbitals in SeCN'

Green to Blue (-0.1 to -1.7), Green to Red (0.1 to 1.7); stepsize 0.1

@

 

62

calculates the atomic charges and determines the occupancy of each atomic orbital in the
complex.

The effect of a purely electrostatic interaction with these species can be calculated by
placing a bare proton, having no basis functions, at a fixed distance from the anion. The
validity of the valence bond model can be tested by determining the effect of such an
electrostatic interaction on the bond lengths. Here we limited the model to electrostatic
interactions along the molecular axis, at a fixed distance of 3 A from either end of the
anion, as shown in Table 18. An interaction to the nitrogen end of SCN' and SeCN'
slightly increases the ON bond lengths and decreases the GS and C-Se bond lengths. The
same type of interaction with OCN’ slightly reduces the ON bond length and decreases the

length of the CO bond. Electrostatic interactions to the chalcogen end of these anions all

Table 18. Effect of a positive charge on the bond lengths, force
constants and vibrational frequencies for OCN', SCN‘, and SeCN‘,
calculated upon fixing a bare proton 3 A along the internuclear axis

 

 

from either end.
Species d d a Vc~ vex“ FCN F ex“
XI 2 (cm") ( cm") (N/m) (N/m)
OCN‘ 1.218 1.244 2155 1205 1350 1113
OCN""H+ 1.217 1.222 2244 1263 1414 1260
NCO""H" 1.208 1.258 2185 l 185 1423 1061
SCN’ 1 .208 l .671 1979 745 1400 502
SCN""H+ 1.21 1 1.636 1997 803 1429 586
NCS""l-I+ 1.201 1 .668 2032 756 1451 525
SeCN' 1.205 1.770 2003 617 1432 471
SeCN""H+ 1 .208 l .740 1999 652 1438 525
NCSe"“H* 1.198 1.754 2064 652 1489 534

 

 

'where X can be 0, S, or Se

63

produce a reduction in the length of the ON bond. On the other hand, the C-0 bond length
in NCO""H+ increases whereas the C-S and C-Se bond lengths decrease in NCS""H+ and
NCSe’"'H+ complexes.

The results of these model calculations are in only modest accord with the
predictions of the resonance structures and show that these anions respond in a more
complicated manner to electrostatic interactions. The simple resonance model may have
been successful in explaining structural and vibrational perturbations from metal-chalcogen
interactions with SCN' and SeCN' because the metal cation does not approach along the
molecular axis of the anion. For example, the crystal structure of Ag*SCN' shows that the
AgSC angle is 104°.83 This nonlinear interaction lengthens the GS bond length and
slightly shortens the C-N bond length.

The perturbations to the stretching modes of OCN', SCN', and SeCN' by the same
electrostatic interaction reveal the complicated nature of these vibrations. As discussed
earlier, the vCN mode in these anions shifts by +89 cm", +18 cm", and -4 cm’1 for
electrostatic interactions to the nitrogen end of OCN‘, SCN', and SeCN', respectively. An
apparent dilemma is the predicted negative frequency shift of the vCN stretch for the
SeCN""H+ complex relative to that in SeCN' despite an increase in both FCN and FCSe force
constants. This negative frequency shift is the result of a change in the normal mode such
that the ‘reduced mass’ of the mode is lowered. On the basis of the potential energy
distributions, the greater sensitivity of the vCN mode in OCN' versus the SCN' and SeCN'
anions to an electrostatic interaction at the nitrogen end is a result of the greater dependence
of this mode on the CO bond. In the SCN' and SeCN' anions, the vCN mode is more
dependent on the ON bond, which is only slightly increased in length by these
interactions. The opposite trend is calculated for this mode when the interaction is at the
chalcogen end of these anions. Here, the vCN mode shifts by +30 cm", +53 cm", and +61
cm". These trends are similar to those observed experimentally for lithium ion pairs of

these anions. It is interesting to note that there are several examples where the frequency

for the Vcn mode is predicted to shift to higher energies despite an increase in the calculated
C-N bond lengths. The calculated ch frequencies more closely follow the calculated C-X
bond length changes. Again, the greater sensitivity of the vCS and vCSc modes with respect
to VCO to an electrostatic interaction at the nitrogen end is due to the larger dependence of

these modes on the C-S and C-Se bonds, respectively.

3. Ionic and hydrogen bonding

Ab initio calculations for cation or hydrogen bonded complexes with OCN’, SCN',
and SeCN' can be grouped into four categories: MNCX, MXCN, triple ions and dimers,
where M is an alkali or alkaline earth metal cation and X can be O, S, or Se. The
equilibrium bond lengths, vibrational frequencies, and atomic charges for OCN', SCN' and
SeCN‘ in ionic and hydrogen bonded complexes are summarized in Tables 19-21. Both
ionic and hydrogen bonding interactions dramatically perturb these anions. The largest
effects are predicted for the alkaline earth metal ion pairs, reflecting the large polarization in
the anions. On the other hand, the hydrogen bonded complexes are calculated to have
smaller changes in the bond lengths, vibrational frequencies, and atomic charges, owing to
the weaker electrostatic nature of the hydrogen bond. Interestingly, similar electrostatic
interactions with OCN‘, SCN' and SeCN‘ result in comparable structural changes.

The strong electrostatic interaction by alkali metal and alkaline earth metal cations in
MNCO, MNCS, and MNCSe ion pairs lead to predictions that can be rationalized by the
resonance model. The C-N bond lengths are increased and the C-X bond lengths are
decreased, with the notable exceptions of the K*NCO‘ and the K”NCS' ion pairs, where the
C-N bond lengths are slightly decreased. Comparison of the bond lengths in the
BehNCO', Be2*NCS', and Be21NCSe‘ ion pairs to their respective NCX’ anions shows
‘ that the ON bonds are lengthened by 0.022 A, 0.040 A, and 0.042A, respectively,
whereas the C-X bonds are shortened by 0.077 A, 0.125 A, and 0.124 A, respectively.
The same trend can be seen in the Li*NCO', Li”NCS', and Li“NCSe' ion pairs, where the

65

Hus—e we. GEE—58n— coaegmam. mqonoaam menace-=98. Waggon 583:3? 38o Sagas Ba ”2835 058mg 404
002. 5 mean 85 358mm: 338 8.258 83338

 

 

 

when-ma. A9 A8 <9. mm: <8 m8 kn: woo 9.9% gm figs

3 c» 2:; 535326. 2:; 5458? >3: >5: ©2 © 0. @Q
002- _.N_m Fun: Nam owq 50m um Sue :5 .908 chem 19me
Wmu+200- ~.NAO H-Haq whom 5: ~23 #5 55m S—O LUG— —.~©~ touc—
zm~+200. —.Nwo - — mN Nwoa Sou :uo um 7:00 awe L130 —. _ uN toduo
0m~+200. 5qu ~. Hem wwww Hqu 3.; Ho 73a Tam L .wmh foqw -9on
5+200. _ .NNO —.NO.~ Nwmm sow Sue m HANG 5.: .. H .NNN oboe touxm
23200. —.NNO —.N H m NNMO ~93 3A3 w 736 5 ~A .- — . fig 0b.: -94?“
“+200- 0&3 ~.N~o N~©u Sow ~Nqu 0 30m ~N83 Lwa ohm—O tau-mg
mum—.200. Hboo —.Nuw NNma mma 32 Na gum _ Eu tobmm Chou 10.x“;
=0:...200- _.N :4 ubwq NNOQ mi ~NAo wo T: — fl ~¢N 10.0% 9me to.ma~
3003-200- TN ~h _.Nwm NNoa muw ~93 mo («8 — Go tobqw O.m©~ rob-No
Pm+002- - ~ om fwow NNNm 0N0 sma Nam qu ~O~© tobac Chum L bub
2w+002. u .NON FNOG NNOQ ham 5cm Go Tgw Hows“ rename 9qu t ~ .omo
az002. ~.NO© _.Nua N:— m¢© 29* ~@_ :30 25¢ 10.x; 9.3— .995
20:...002. .979 _.~u~ 23 a; :ou «N 53 Ewe rpmma Ouwmm tag-mow
30031002. _.N T¢ ENMN N _ mm 2: _ ~©u um 5.3 23m to.mam O.mom tob— _
~1m+002-—L+ —. — cu _.Nww wwwu cuo waq .3 Gem ~Nwm raw-cow oqu tobau
m:..002-zm m. —.No~ —.NAu NNaN cow 5% mm 3N4 — 73 to.mmm chew toaqm
:0:...002-...:0= ~98 fuew NM _ .u w: SAN um Tgu fl 5m .90 ~ q 9me roach
A5+2000~ FNNQ _.Noo qum N3 u 53 E. Hume :3 t _ bag :0: toQ — N

wwoo 0 Gus o

 

66

 

Haw—a we. 823:3
5,.-
..200- rfiu :3 8.3 m3 5% 3 53 G _ u L .wwo 908 563
rm
1: .
..20O- Eu 5 rm 0 m 35 She Emu u :3 Eco - _ .8. 902 -943
mm-
:03.
..Z00- rm 3 rmuo mmom mum 53 we $8 52 .98.» came 5&3
mom.
54.-
..002- :3 _.wow N _ 3 GM Ben E :33 mg 6.30 9qu - _ .o-E

 

67

Haw—a N0. 0&8580 dozen—firm. £88306 35:398. Sam—£8 020358. 380 88:58 25 39am 033mg «2

m02- 5 mean 25 308mg @308 "380.258 85293

 

 

M‘Nfimfih mg “a <02 N02 <0M m0“ *0? k0“ Q50 05“ GS“
x» L 2:; 03532». as; 03532... >3: >33 ©2 © 0 © 0.
m02. _ .00m .03 030 uwu 03 N0 300 won -903 0. 0 ac -9u 3
wow-20m. _ .NAm _ .03 ~03 . mum :00 ~00 _ 00— 30 - 0 hmw 0.0g 98—
Zm~+20m. 0N3 ruoq N000 :3 02 00 $00 30 -_ bum 0.000 90.0
09220” 0.000 fume pong 5.5 000 0 _ 0.00 .30 - 0 . 0 0.0 0. 30 0.00m
E+Z0m. PM; 0.0; N03 :8 00$ 0 :5 000 L000 90?. 00%
2w+20w. 0.8 u 0.08 N000 03 wow 0 T: u 05 .9000 903 -903
520m. 0000 0.3m 500 m 0 0 000 u Ema mow -908 0.03 -9000
30-200. 0.~0~ .03 ~03 up: mom A 33 Km .930 9&3 -93»
EOE-20m- 0 .000 0000 00. 0 A00 0.: _ _ 300 000 -9000 0.03 -950
5.002. 0.» 0 0 0.000 00....— ~00 0.2 mm 3mm moo -903 0.0m.» -905.
30-002. rm?» 03m N000 ~00 «mm .0 :00 $0 -903 9 En -930
ICE-002. 0.000 #03 $3 nmo .r: E Ewa .0on -900» 9 50 -9u _ 0
500200.. :8 0.0: ~03 m _ q mum _ Gum E -980 0.0 0 0 -92 0
“2.00255“ 0. Gm #000 00% #00 000 um 33 3A -908 09% -930
EOE-0023.20: 0.000 roam ~03 0mm 30 0 75m H: -9000 930 -930
E...
inm-
Ecv .03 00mm 0000 03 mmm u 300 .50 - _. am 9.03 -903
m: .
..z0m-
3+. 0.3 _ .600 N00— m3 mg 0m 33 umm -9000 903 -9000
:01.
.28.
53 _.SN 3% E E a 3; E -3; 88 -930

:0:

 

68

0.9....” 0.. 9:858“. 00:05:03? £883.50 03:30.8. 580380 588.08. 38o 838:8 Ea 203.0 3808 000

0002. .: .200 BE 3.0302. 0308 8.380330 880.88

 

 

0.08.8 «.9 «.9 <9. mnz <8a m8. .82 .8. 0.0.9.0... QRSR QB»?
x. 0. as; .3330? as; .3530? 20: >3: ©.< ® 0. 000...
0002. . .000 . .000 0000 000 0 . 0 0. . .000 A0. 0.000 0.000 0.000
0034600. ..000 ..000 0000 .0.# 000 A00 .00A 000 .. . .000 0.000 0.000
2.08.2000. ..00. ..00. 0000 .000 000 00.0 .000 000 -..0.0 0.0.0 0..00
Om~+200o. ..00A . .000 .000 .000 000 00.0 .000 000 - . .000 0.000 0. . 0A
....+an0. ..0.. ..0 .0 0000 ..00 000 00k .t0 000 -.000 0. .00 0. . 00
20+2000. . .000 ..00. 00.0 0. 0 000 0.0 .000 00. 0.000 0. .00 0. . 00
.A+ZOmo. . .0 .0 ..000 0000 000 000 0. . .00. 000 0.000 0. .00 0.0.0
0..-2000- .. .00 . .000 0000 A00 000 .00. .0 .0 #00 0.000 0. .00 0.000
10......2000. ..000 ..00. 00.... 000 00. 0.0 .A00 “«00 0.000 0.000 0.00.
F3082. . .000 . .000 .000 . 00 000 #0. 0 .000 B«00 0.000 0.00 . 0.00 .
0..-.0002- ..00. . .000 0000 . 00 000 0.0 7900 B00 0.000 0.00 . 0.0.0
20......0002. ..000 . .000 0000 .00 0 . 0 0.. :00 h000 0.0.0 0.000 0.000
500.025.. .. . 00 . .000 0.00 000 000 00.0 .000 #00 0.0 .0 0.000 0.000
.uI...moOZ-.......u .. .0A ..000 0000 00.. 000 0.0 .050 A00 0.000 0. .00 0.000
ICE-0002-.....O: ..00. . .000 0000 000 0&0 P0 .000 R00 0.0.0 0.00. 0.000
5+
”.280.
E“... ..0A0 ..00. .000 0A0 0A0 00.0 .0.. 000 -...A0 0.000 0. .00
m: c
..ZOMo-
3.... . .000 . .000 00.0 000 00A 0.0 :00 000 0.00A 0. .00 0.000
:02.
..z8o-
30*... . .000 . .000 0000 000 000 0.0 it 00. 0.000 0.000 0.000

 

 

69

C-N bonds are lengthened by 0.002 A, 0.008 A and 0.006 A, with the ox bonds
decreased by 0.037 A, 0.057 A 0.051 A, respectively. The opposite trend in the bond
lengths is seen in Li*OCN' ion pairs. The C-N bond length is reduced by 0.020 A and the
C-0 bond length is lengthened by 0.025 A with respect to OCN'. Unfortunately, the same
comparisons cannot be made for the MSCN and MSeCN ion pairs, because the lowest
energy structures of these species are not linear, so that the simple resonance model would
not apply. In general, the calculated bond lengths for the linear MOCN, MNCO, MNCS,
and MNCSe ion pairs satisfy the predictions of the resonance model.

A given cation perturbs bond lengths of the anions in metal thiocyanate and
selenocyanide ion pairs more than those of the corresponding metal cyanates. This effect,
illustrated by the Be2+NCX’ and Li"NCX' ion pairs, can be rationalized in terms of the
simple valence bond model. Since the ON bonds in the SCN‘ and SeCN‘ anions have
greater triple bond character than that of the ON bond in OCN', the SCN' and SeCN'
anions are more easily polarized by an electrostatic interaction to the nitrogen end.

A single hydrogen bond to the nitrogen end of OCN', SCN' or SeCN', by HF
(hydrogen fluoride) or HOH and MeOH in the case of OCN’, shortens both the C-N and C-
X bond lengths in these anions. A hydrogen bond to the chalcogen end shortens the ON
bond and lengthens the C-X bond. The reduction of the C-N bond lengths in XCN""HD
complexes, where HD is a hydrogen bond donor, suggests that the hydrogen bonding
interaction cannot be interpreted as a simple electrostatic interaction as in ion pair
complexes.

The MZNCX* triple ions and (DH)2"‘NCX’ doubly hydrogen bonded complexes
lengthen the C-N bond and shorten the C-X bond. The (Li*)2NCO' triple ion is
qualitatively similar to the (Li“NCO‘)2 dimer in terms of the calculated bond lengths,
stretching frequencies and charge distribution, and thus serves as a good model species for
the dimer. Again, the response of these anions to two ionic or hydrogen bonding

interactions to the nitrogen end results in changes in the bond lengths that are qualitatively

 

70

predicted by the resonance model. The (Li*)2OCN‘ triple ion shows the opposite effect
from (Li”)2NCO', which also is consistent with the resonance model.

Both the ON and C-X bond lengths are calculated to shorten in the Li‘NCX'Li”
triple ions and DH"'NCX""HD hydrogen bonded species shorten. The largest changes are
seen in the Li+ complexes where the C-N bond is shortened by ~0.1 A and the C-X bond
by ~0.02 A.

4. Atomic charges in ionic and hydrogen bonded complexes
The shift in the distribution of the negative charge in OCN', SCN' and SeCN' in

response to an electrostatic interaction can be described by the atomic charges. The N BO
atomic charge analysis shows that the cyanate anion responds uniformly to the strength of
the Coulombic potential applied. The atomic charges on the nitrogen, carbon, and oxygen
atoms of OCN', when it forms a hydrogen bonded or ionic complex at the nitrogen end of
the anion, are plotted as a function of the Coulombic potential generated by the hydrogen
bond donor or the metal cation in Figure 16. Here, the Coulombic potential is defined as
the atomic charge of the metal cation or the hydrogen bond donor divided by the hydrogen-
nitrogen or cation-nitrogen distance. The linear correlations between the atomic charge on
the oxygen, the carbon or the nitrogen atoms and the Coulombic potential are 0.989,
0.976, and 0.960, respectively, reflecting the predominantly electrostatic nature of the
interaction. The largest change in the atomic charges is calculated for the nitrogen atom,
which draws electron density from both the oxygen and carbon atoms. The atomic charges
in OCN' also correlate well with the Coulombic potential when the ionic or hydrogen
bonding interaction is on the oxygen end. In this case the distribution of the negative
charge shifts towards the oxygen end. The change in the atomic charges in response to a
Coulombic potential is consistent with the resonance model for these interactions, where an
increase in the negative charge at the nitrogen atom in OCN' will come at the expense of the

negative charge on the oxygen atom. The atomic charges in MeOH, HOH and HF

71

complexes to the nitrogen end of the anions do not correlate as well with the Coulombic
potential generated by the proton of hydrogen bond. Indeed, the correlation between the
atomic charges and the Coulombic potential is improved when the hydrogen bonded
complexes are removed from the set. The difference is a reflection of the more complex
nature of the hydrogen bonding interaction. As discussed earlier, the ion pairing interaction
is primarily electrostatic in nature, with a small amount of a charge transfer involved for the
alkaline earth metal ion pairs. On the other hand, hydrogen bonding can result not only
from electrostatic and charge transfer interactions, but also may have some covalent
character. The sums of the atomic charges on the cyanate anion in the MeOH'"NCO',
HOH"'NCO' and FH'"NCO' species are 0952, -0.958 and-0.899, respectively, suggesting
the importance of charge transfer in these hydrogen bonded species. The same trends are
found for the atomic charges on the atoms in metal thiocyanates, metal selenocyanides,
hydrogen bonded thiocyanates, and hydrogen bonded selenocyanides, where the
interaction is to the nitrogen end of these anions. Correlations similar to those in Figure 16
are also observed for the ionic and hydrogen bonded SCN‘ and SeCN‘ complexes. For
these complexes the slopes of plots of the atomic charges on the nitrogen atoms versus the
Coulombic potential are larger than that for the OCN' system.

The calculated atomic charges in Li*NCX'Li+ triple ions and analogous hydrogen
bonded complexes, DH"'NCX""HD, show only small differences from their respective
anions (Tables 19-21). Electrostatic interactions to either end of the anion produce opposite
results, and the two effects approximately balance. The largest changes are predicted for
the lithium triple ions. The (Li+)2NCX' triple ions and analogous hydrogen bonded
species, (DH)2"'NCX' display effects similar to those calculated for ion pairs, where the
negative charge on the anion is substantially shifted to the nitrogen end. The opposite
effect is predicted for (Li+)2OCN' ion pairs, where the negative charge on the cyanate anion

is shifted to the oxygen end.

 

Alr-

72

  
 
  

  
    

 

 

1.500 ..
r = 0.989
(r= 0.987)
* BeNCO‘
1.000 - CaNCO’ MgNCO
HOH”‘NCO
OCN'
0.500 ..
2.“?
a 0.000 ..
3 l
9‘0 |
g r = 0.976
.2 (r = 0.995)
g ‘0.500 dr-
2 ' BeNCO‘
rNCO
OCN' HOHNCO KNCONaNOOL
‘: MeOH...NCO" . FH'NCO‘
'1 .000 db MCOHMNCO'. O muNCO-
OCN HOH‘"NCO‘
r = 0.960
. Charge @ N (r = 0.998:
-l.500 -
I Charge @ C BCNCO.
A Charge @ 0
-2.000 : : : :
0 5 10 ,5 20

Coulombic potential (J/C)

Figure 16. Atomic charges at O, C, and N in MNCO species, where M can be a
hydrogen bond donor, an alkali metal cation, or an alkaline earth metal cation, versus
the Coulombic potential provided by the cation or hydrogen bond donor; r is the
correlation coefficient with and without (in parenthesis) the hydrogen bonded species

 

73

The resonance model predicts that the polarization of the charge distribution of
OCN‘, SCN' and SeCN‘ is primarily in the 11: system. The calculations reveal that the ls
and 23 core orbitals are little affected by an electrostatic interaction, whether on the nitrogen
or chalcogen end. The pz orbital, which lies along the molecular axis, is slightly perturbed
by an ion pairing or a hydrogen bonding interaction. The greatest changes are seen in the
pK and py orbitals, as exemplified for the OCN’ species in Table 19. Ion pairing and
hydrogen bonding to the nitrogen end of these anions shift the pK and py electrons from the
chalcogen atoms to the nitrogen atom. The opposite trend is predicted when the ion pairing

or the hydrogen bonding interaction is at the oxygen end.

Table 22. Occupancies of 2px, 2p,, and 2pz atomic orbitals in alkali metal,
alkaline earth metal, and hydrogen bonded cyanates

 

 

Species Oxygen Carbon Nitrogen
2p... 2p. 2p. 2p” 2p. 2p, 2p... 2p. 2p,
OCN' 1.80, 1.80 1.53 0.79, 0.79 0.81 1.39, 1.39 1.46
Nitrogen species
Be2*NCO' 1.60, 1.60 1.55 0.64, 0.64 0.79 1.70, 1.70
Li‘NCO' 1.73, 1.73 1.54 0.72, 0.72 0.80 1.53, 1.53
FH'NCO' 1.78, 1.78 1.53 0.76, 0.76 0.80 1.44, 1.44 1.48
HOH"'NCO’ 1.77, 1.79 1.53 0.77, 0.77 0.80 1.42, 1.42 1.48
MeOI-PNCO' 1.77, 1.79 1.55 0.77, 0.77 0.80 1. 3, 1.42 1.49
Oxygen species
Li*OCN' 1.87, 1.87 1.58 0.83, 0.83 0.79 1.27, 1.27 1.46
FH"'OCN‘ 1.76, 1.83 1.58 0.81, 0.80 0.80 1.36, 1.35 1.43

HOH"‘OCN' 1.80, 1.80 1.52 0.80, 0.80 0.80 1.35, 1.37 1.46
MeOH"'OCN‘ 1.71, 1.81 1.63 0.81, 0.80 0.80 1.40, 1.37 1.43

 

 

 

74

5. Molecular orbitals in ionic and hydrogen bonded complexes
Evidence that the polarization of these anions in ion pairs is mainly through the 1:

system is provided by the visualization of the molecular orbitals, shown in Figures 17-21
for OCN', Li*NCO', Li*OCN', Be2"NCO' , LizNCO", and L120CN“. The 16 and 20'
valence molecular orbitals in OCN‘, SCN' and SeCN‘ in MNCX ion pairs are primarily
bonding in nature and are not substantially affected by an electrostatic interaction (Figures
22-27). The 36 and 4G valence orbitals, which are primarily antibonding, become slightly
bonding for both the C-0 and C-N bonds when perturbed by electrostatic interactions to the
nitrogen end. In Figure 28 which compares the 1: orbitals in OCN', LiNCO, LiOCN and

 

BeNCO‘, the extreme positions perpendicular to the intermolecular axis for the i0.26x10‘5
m3” and :1: 2.1x10'5 m‘3’2 contours of the 11: orbital and 21: orbital of OCN’ are referenced
by bars for each species shown. The Li+ cation primarily compacts the 11: orbital around
the C—0 bond in the Li"NCO‘ ion pair. The same electrostatic interaction pulls electron
density away from the ON bond in the weakly bonding 21: orbital. Electron density is
slightly pulled into the CO bond in the 21: orbital from the antibonding region. These
changes are more pronounced for Be2*NCO'. The 11: orbital for Be2+NCO' is substantially
compacted around the CO bond and is reflected by the substantial structural changes for
the anion in this ion pair. The shift of the 21: orbital to the Be2+ cation is more exaggerated
than for the Li” cation thereby removing additional electron density surrounding the C-N
bond and increasing the electron density around the C-0 bond over that of the Li+
interaction. In the Li‘OCN’ ion pair, the opposite effect is seen. The 11: and 21: orbitals are
shifted toward the oxygen end, which reduces the electron density surrounding the C-0
bond and increases the electron density around the ON bond. Similar shifts in the electron
density of the 11: and 21: molecular orbitals are also calculated for the MNCS and MNCSe
ion pairs. The maximum electron density of the 11: orbital shifts toward the nitrogen.
These interactions increase the bonding character of the GS and C-Se bonds by pulling

electron density from the chalcogens onto the OK bond. Likewise, the antibonding

75

   

3o 46

 

11:,"y 21:”

Figure 17. Contour plots for the valence molecular orbitals LiNCO
Green to Blue (-.1 to -1.6), Green to Red (.1 to 1.6); stepsize 0.1

76

   

lo 20

 

30 4o

 

11:”, 21:”

Figure 18. Contour plots for the valence molecular orbitals of LiOCN
Green to Blue (-.1 to -1.6), Green to Red (.1 to 1.6); stepsize 0.1

77

   

@D‘

   

11:,Ly 211:,“y

Figure 19. Contour plots for the valence molecular orbitals in BeNCO+
Green to Blue (-.1 to -1.6), Green to Red (.1 to 1.6); stepsize 0.1

78

   

21:x 21:},

ft»
' C0)

 

Figure 20. Contour plots for the valence molecular orbitals in LizNCO”
Green to Blue (-.1 to -1.7), Green to Red (.1 to 1.7); stepsize 0.1

79

26

 

36 4o

3)

 

 

21k zuy @

Figure 21. Contour plots for the valence molecular orbitals in LiZOCN“
Green to Blue (-.1 to -1.7), Green to Red (.1 to 1.7); Stepsize 0.1

80

   

26

”@j

   

11:x

Figure 22. Contour plots for the molecular orbitals for LiNCS
Green to Blue (-0.1 to -1.7), Green to Red (0.1 to 1.7); stepsize 0.1

81

@500

 

00 @=

   

Figure 23. Contour plots for the molecular orbitals BeNCS+
Green to Blue (-0.1 to -1.7), Green to Red (0.1 to 1.7); stepsize 0.1

82

 

 

83

   

16 26

 

36 46

  
   

   

@

-

@8

11:,“y 27R)!

25. Contour plots for the valence molecular orbitals for LiNCSe
Green to Blue (-0.1 to -1.7), Green to Red (0.1 to 1.7); stepsize 0.1

 

11:,“y

84

26

 

  
 

  

@.\\\ g

  

21:Ly

Figure 26. Contour plots for valence molecular orbitals for BeNCSe"
Green to Blue (-0.1 to -1.7), Green to Red (0.1 to 1.7); stepsize 0.1

85

 

 

9%)

@V

e

27%

Ky

27. Contour plots for valence molecular orbitals in

LizNCSe+, Green to Blue (-0.1 to -1.7), Green to Red (0.1 to 1.7)

stepsize 0.1

86

11: OCN' 21: OCN‘

 

2n OCN‘Li+

eeee

 

11: OCNBe2+ 11: ocmse2+

%@

 
   

11: Li+OCN' 21: 1.1+OCN‘

0

 

I e 28. Contour plots of the 11: and 21: Molecular orbitals in OCN‘,OCN‘Li*, OCN‘Bez”, and I.i+OCN‘.
(Green to purple: 0261th15 to -2.6x1015 m'm, green to red: 0.26x10‘5 to 2.6x10‘5 m‘m;
stepsize 0.26x10‘5 m'm)

87

character of the 21: orbital around the ON bond in these complexes is increased by pulling
electron density off the nitrogen atoms. Similar arguments can be made for the MzNCX+
triple ions. The non-axial interaction of the lithium cations has a dramatic effect on the 1:
orbitals in the plane of the complex, reducing the bonding character of these orbitals around
the C-N bond in these anions.

Hydrogen bonding interactions at the nitrogen end show similar trends in the atomic
charges and occupancies of the px, py and pZ atomic orbitals as the MN CX ion pairs, yet the
C—N bonds in these hydrogen bonded complexes are shorted relative to their free anions
rather than lengthened the ON bonds as in the ion pairs. The difference between hydrogen
bonding and ionic interactions is revealed by visualization of the molecular orbitals, shown
for the OCN""HF complex in Figure 29. In the hydrogen bonded anion complexes, the
molecular orbitals of the anion and the hydrogen bond donor are not separately localized, as
iS true for the ion pairing interactions. Rather, the molecular orbitals of both the anion and
the hydrogen bond donor are delocalized throughout the adduct. The dramatic reduction of
the C -N bond length in FH'"NCX' complexes results from the delocalization of the p" and
Py orbitals from HF into the C-N bond of the anions. The reduction of the ON bond
length in (Li*)2NCX' triple ions and doubly hydrogen bonded adducts, (DH)2"'NCX', is
3130 reflected by the molecular orbitals. The electrostatic interactions to the 1: cloud of the

C‘N bond pull electron density away from the anion thereby lengthening the bond.

6- Electrostatic potential in ionic and hydrogen bonded complexes
The electrostatic potential in OCN’, SCN' and SeCN' shows that the negative charge
is distributed throughout the anions. The cyanate anion has the most negative electrostatic
potel‘ltial, the selenocyanide anion the least, shown in Figure 30. This is consistent with
ion pairing studies of these anions, where OCN' forms stronger ion pair complexes than
SCN‘ and SeCN'.70 Ion pairing and hydrogen bonding dramatically alter the electrostatic

potential in the OCN‘, SCN' and SeCN' anions. The perturbation in the electrostatic

88

 

Figure 29. Contour plots of a 1: molecular orbital in HF extending into the C-N bond of OCN‘,
thel‘eby increasing the electron density surrounding the C—N bond in the OCN'MHF complex

(Green to Purple: -0.26x10‘5 to -1.6x1015 m'm, Green to Red: mama” to 1.6x1015 m‘m;
stepsize 0.2mm15 m'm)

89

potential by ionic and hydrogen bonding interactions to OCN' are shown in Figure 31. In
this series, the largest perturbation from the OCN' electrostatic potential is seen in the
lithium cyanate ion pair. The most negative electrostatic potential in Li"NCO' ion pairs is
on the oxygen; the same result holds for the Li+NCS' and the Li+NCSe‘ ion pairs. The
most negative electrostatic potential in OCN‘"HF is also on the oxygen, yet the nitrogen
end also has a substantial negative potential. The weakest hydrogen bonded complex,
OCN""HOH, retains significant negative electrostatic potential on the nitrogen end of the
anion. The electrostatic potential in SCN' and SeCN' anions respond to hydrogen bonding
interactions similarly.

The -1.36 J/C isosurface of the electrostatic potential in Li+NCO', shown at the
bottom in Figure 31, illustrates a peculiarity found in experimental studies of alkali
cyanates. No ion pair with the alkali metal bound to the oxygen end of the cyanate anion
has been identified experimentally. Although the calculated atomic charges suggest that ion
pairing to either the nitrogen or oxygen end of the cyanate anion are almost equally likely,
the electrostatic potential of OCN', shown at the top in Figure 30, predicts quite correctly
that the nitrogen end is preferred for electrostatic interactions. The calculated electrostatic
potential for Li”NCO' also demonstrates that the interaction with another Li+ cation will
occur at the oxygen end of the complex, thereby forming the Li*NCO'Li+ triple ion.
Chabanel and coworkers have observed this complex experimentally.75

The sensitivity of these anions to the strength of a hydrogen bonding interaction
provides important implications for solution complexes, especially for weak hydrogen
bonded complexes. The electrostatic potential correctly predicts that a 1:1 complex will be
formed at the nitrogen end by either hydrogen bonding or ionic interactions. More
importantly, these calculations provide predictions for the 1:2 complexes. Unlike the
situation for cations and strong hydrogen bond donors, for weak hydrogen bond donors a

second interaction at the nitrogen end will be more favorable than an electrostatic attraction

90

 

Figure 30. Electrostatic potential for OCN‘, SCN', SeCN’
Red to Green: 0125 to -0.275, stepsize 0.015

91

Isosurface .680 J/C

Isosurface -6.26 I/C
OCN‘-“HOH

Isosurface -5.85 J/C

Isosurface -1.36 J/C
Li*NCO‘

 

Figure 31. Molecular electrostatic potential of OCN‘, OCN""HOH, OCN'"'1-IF, Li*NCO‘

92

to the chalcogen end of the atom. Experimental investigations by Schultz et al. have
confirmed this prediction.137 Studies of the OCN' anion in methanol, formamide, and N -
methylformamide have shown that the methanol can hydrogen bonds to both ends of the
cyanate anion, but the amide solvents do not, even though the extent of solvation of the

cyanate anion is greater in formamide than in methanol.137

7. Vibrational Frequencies in ionic and hydrogen bonded complexes

Comparison of the predicted to the observed stretching frequencies in the alkali and
alkaline earth metal cyanates and thiocyanates, where possible, shows that there is good
agreement between the theoretical predictions and experimental results, as shown in
Figures 32 and 33. The calculations tend to overestimate the shifts in the vibrational modes
of the alkali cyanates by a factor of four. This difference has also been seen experimentally
by Devore, who found that the K*NCO' ion pair is very sensitive to the medium differences
between the gas and solution phases.78 Indeed, the predicted value for the frequency of the
Von stretch for K*NCO' is 2193 cm“, which is very close to the gas phase value of 2206
cm‘1 reported by Devore.78 The difference between predicted and experimentally observed
stretching frequencies for the thiocyanate ion pairs and dimers is much smaller than that
form OCN' systems, which is not surprising in knowing that the thiocyanate anion is less
sensitive to medium effects.

Due to computational costs, the (Li“NCS')2 and (Li”NCSe')2 complexes were not
studied. Rather, the (Li*)2NCS' and (Li*)2NCSe' triple ions were used to model the
frequency perturbation in the lithium thiocyanate and selenocyanide dimers. The calculated
frequencies of the stretching modes in (Li+)2NCS‘ and (Li+)2NCSe' qualitatively reproduce
the negative frequency shifts reported for the vCN mode of the dimers.“1 '9

Interactions with the nitrogen end of OCN‘, SCN', and SeCN' by an alkali metal,
alkali earth metal or hydrogen bond donor increase the frequency of both stretching modes.

The largest perturbations are predicted for the Be“ complexes, reflecting the large change

93

2350 l

2325 o LiNCOLi

 

2300 v:
o LiNCO

o NaNCO

2250 - /
2225

Ab initio v01 frequencies (cm")

OCN' r = 0.95
O

21ml a . e +
2125 2150 2175 2200 2225

Experimental frequencies for the v(CN) stretch (cm")

 

 

Figure 32. Ab initio frequencies of the VCN stretch in alkali cyanates versus
experimental frequencies of the Von stretch of alkali cyanates in DMSO75

2100

2075 l . LiNCSLi/

o BeNCS’
2050 -.

2025 - o CaNCS‘ /MgNCS

yLiNCS

2000 > e aNCS

1975 SCN’

 

o KNCS

Ab initio vm frequency (cm")

1950 ‘-

1925

o LizNCS’ r = 0.876
1900 i 1 ~ -- 1 1 4
2025 2050 2075 2100 2125

Experimental v(CN) frequency (cm")

 

 

Figure 33. Ab initio frequencies of the VCN stretch in alkali thiocyanates versus
experimental frequencies of the vCN stretch of alkali thiocyanates in aprotic

solvents (LizNCS* is compared with the experimental value for (LiNCS)2)"'82

94

in the bond lengths of the anions in Be2+NCX' ion pairs. Interestingly, perturbations by the
I—IF hydrogen bond donor produce frequency shifts as large as those calculated for the
alkali metal cation ion pairs.

The perturbation of OCN‘, SCN' and SeCN' frequencies caused by interactions of
alkali metal cations, alkaline earth metal cations and hydrogen bond donors at the chalcogen
end of the anion is more complex. For the OCN' anion, these interactions raise the
frequency of the vCN stretch, although the increase is not as large as those seen in analogous

complexes where the interaction is at the nitrogen end. Alkali metal cyanates have v00
stretching frequencies at higher energies with respect to OCN’. However, hydrogen
bonding at the oxygen of OCN’ red shifts the frequency of the VCO stretch. Ion pairing to
the sulfur atom in SCN' and to Se in SeCN‘ by Li+ shifts the vCN modes in these anions to
lower frequencies and the VCX modes to higher frequencies with respect to their free anions.
These results for the stretching modes in Li+SCN' and Li+SeCN' ion pairs are substantially
different from the experimental observation for the Ag‘SCN' ion pair.”120 The reason is
presumably the difference in the geometries. The LiSC and LiSeC angles in the Li+XCN’
ion pairs are acute (< 70°) and the anions are bent. Hydrogen bonding by HF, HOH or
MeOH with SCN‘ and SeCN’ shift the vCN stretch to higher frequencies and lower the
frequency of the vCX stretch with respect to the free anions.

Triple ions of the form Li”NCX'Li+ show the highest frequency stretching modes.
Indeed, the calculations predict that these complexes can be easily distinguished
spectroscopically from the free anion, ion pairs, dimers and tetramers, since they have the
highest frequency VCN stretch for ionic complexes of a given metal cation. The triple ions
of the form (Li+)2NCX', which we employ to approximate the dimer species, red shift the

. fre’quency of the VCN mode and blue shift the frequency of vCX with respect to Li*NCX' ion
Pairs - The red shifts in the vCN mode of (Li*)2NCS’ and (Li")2NCSe' are so large that the

VCN frequency in these anions is lower than that of their respective anions. The use of these

95

triple ions to approximate dimers was confirmed by the (Li*NCO')2 dimer, where the
calculated stretching modes are close to those of the (Li*)2NCO' triple ion.

The frequency shifts in the OCN', SCN', and SeCN' anions can be understood by
relating the frequencies to the force constants for the bonds. Selected examples of changes
in the bond lengths, force constants and vibrational frequencies are collected in Table 23,
where the complicated nature of the stretching modes for these anions is revealed. In the
OCN‘ complexes, there is little correlation between the frequency shifts of the vCN mode

and changes in the FCN force constants. For example, the FCN force constant in (Li*)2NCO'
is reduced by 93 N/m with respect to OCN', yet the frequency of the vCN mode is increased.
The greater dependence of the vCN mode on the ON bond stretching force constant for
selenocyanide complexes is illustrated by the Ca2*NCSe' and (Li*)2NCSe' species. In
Ca2+NCSe‘, the frequency of the VCN mode is slightly decreased with respect to SeCN’,
reflecting a balance of the reduction in the FCN and growth in the FCSe force constant. In
(Li+)2NCSe', the dramatic reduction in the FCN force constant is not similarly offset by the
che force constant, so that the predicted frequency of the VCN mode in this complex is
StIIbStantially lower than that in SeCN'. The vCO and vCSC modes in OCN' and SeCN'
c0111plexes exhibit similar complexities. For example, the reduction in the FCO force
constant for Li”OCN' is more than balanced by an increase in the FCN force constant,
reSulting in an increase in the frequency for the Vco stretch.

As illustrated in Figure 34, the predicted changes in the FCN and ch force constants
in the OCN', SCN', and SeCN' anion complexes can be related to the calculated ON and
C‘x bond lengths via Badger’s rule: F,j(rc - dij)3 = C; where F”. is the force constant, rc is
the equilibrium bond length, dij is an adjustable parameter and C is a constant.136 The FCN
force constants for the OCN', SCN', and SeCN' complexes have correlations of 1=0.809,
1:0-904, and r=0.963, respectively with the ON bond lengths in these species. The
relationships between the ch and the C-X bond lengths in these species are much better:

I:0-995, r=0.998, and r=0.999, respectively. The good empirical correlation between the

96

force constants and the bond lengths in these species provides an indirect connection
between the vibrational frequencies and structural parameters for these anions through the

force constants.

Table 23. Calculated changes in bond lengths, force constants and frequency shifts
for selected ionic cyanates and selenocyanides, with respect to the free anions

 

 

 

 

Species Ad(CN) Ad(CX)“ A Va: A vex“ AF (N AFCX"
A A (cm") (cm") (N/m) (N/m)
OCN'

Li‘NCO' +0.002 -0.037 +133 +145 +76 +258
Ca2*NCO' +0009 0051 +178 +17 1 +66 +385
OCN"“HF -0.009 -0.01 1 +101 +76 +125 +80

Li‘OCN' -0.020 +0.025 +73 +3 1 +147 —84
(Li*)2NCO' +0.025 -0.055 +1 19 +84 -93 +402

Li‘ NCO‘Li+ -0.021 -0.01 1 +176 +132 +242 +125
SeCN’

Li‘NCSe‘ +0.006 -0.051 +19 +250 +13 +1 1 l
Ca2*NCSe‘ +0.019 -0.074 -8 +217 -125 +184
SeCN""HF -0.008 —0.014 +65 +62 +82 +24
(Li*)2NCSe' +0.037 -0.079 -124 +125 -221 +206

Li‘NCSe'Li” -0.011 -0.017 +104 +146 +154 +23
“X can be 0 or Se

The hydrogen bonding and electrostatic interactions with the OCN', SCN‘ and
SeCN' anions perturb not only the vibrational frequencies of the stretching modes but also
the infrared molar absorptivites (see Tables 19—21). It is difficult to obtain a simple model
to easily explain these absorptivity changes since this property is related to the dipole
moment derivative. Nonetheless, these predictions for the absorptivities provide additional
clues for the identification of solution structures. The relative intensity predictions compare

well for previously identified solution species. For example, ionic interactions to the

97

 

 

1800 -

E
Z
V 1600 -
8

><
.2

Q:

E

O

U
'z’ _
U 1400
O
.5

:2

5

53

U)

S 1200 —
U

Q

8
c9.

8
u.
'2 1000 -
N

5
1.1..

800 1 1 1 1 1 1

 

1.150 1.175 1.200 1.225 1.250 1.275 1.300

Bondlengths for the C-N and C-0 bonds in OCN‘ complexes (ml 10"”)

Figure 34. Plot of the FCN and FCO force constants versus their respective bond lengths

in ionic and hydrogen bonded OCN' complexes

98

nitrogen end of either the cyanate or thiocyanate anions by alkali metal cations increase the
absorptivity of the VCN stretch, with the largest changes Li+ cations.

An important application of these calculations is the assignment of vibrational bands
where the identification is ambiguous. For example, the 2100-2300 cm'l infrared spectrum
of Li*OCN' in DMSO where LilClO‘,‘ has been added consists of three bands, at 2203 cm",
2189 cm'1 and 2172 cm".75’77 The highest frequency band has been assigned to the
Li“OCN‘Li+ triple ion and the lowest to the Li+NCO‘ ion pair.”77 The 2189 cm‘1 absorption
could be assigned as either Li“(OCN‘)2 or (Li+NCO')2.75'77 Of these two possibilities, the
(l.i"NCO')2 is more likely since the stretching modes of OCN' in this complex are shifted to

higher frequencies than LilOCN' ion pairs (Table 19).

8. Cation Salvation

As discussed earlier, the above calculations neglect the effects present in a dielectric
medium and the solvation of the cation and anion. Chabanel’s MNDO calculations show
that solvation of the lithium cation by water reduces the perturbation of the Von vibration in
lithium thiocyanate ion pairs, dimers and tetramers.89 Likewise, solvation of the cation in
the LiNCO ion pair by water, shown in Figure 35, does not shift the frequencies of the
stretching modes in the cyanate anion as great as that in the unsolvated ion pair (Table 24).
These ion pair models, which are calculated in a gas phase medium, exaggerate the .
perturbations of the OCN' and SCN' anions. Inclusion of cation solvation markedly

reduces the perturbation and should result in more quantitatively reliable results.

C. Conclusions

Ab intio calculations of ionic and hydrogen bonding interactions to XCN‘ anions at
the MP2 level using large basis sets such as the d95v+**, correctly predict a number of
trends seen experimentally in solution. The calculations successfully reproduce several

important experimental observations: (1) the vCN mode for the OCN‘ anion is more

99

Table 24. Calculated bondlengths and vibrational frequencies at the HF/d95v+** level
for OCN', LiNCO, and Li(HOH)3NCO

 

 

Species dc” dd ch Vm
OCN' 1.171 1.215 2380 1361
LiNCO 1.182 1.175 2420 1518
Li(HOH),NCO 1 . 178 l . 185 2412 1474

 

 

sensitive to ionic interactions than the VCN modes for SCN' and SeCN‘; (2) the vCS stretch in
SCN’ complexes is more perturbed by ionic interactions than is the vCO mode in OCN'
species; (3) the perturbation of the VCN mode of the OCN’ and SCN‘ anions in M+NCX‘
complexes increases as the charge density of the cation increases; (4) ionic interactions to
the nitrogen end of SCN' increase the absorptivity of the vCN mode. Generally, the
calculated frequency shifts for the stretching modes are much larger than those seen
experimentally. This discrepancy results from not incorporating the solvation of the ions
and the neglect of the dielectric medium surrounding the complex. The somewhat better
agreement of the frequency shifts for the VCN modes in SCN' and SeCN’ complexes,
compared to those of OCN', is due more to the nature of the RED. for this vibrational
mode than to the level of theory. (This phenomenon is supported by the small difference
between the experimental gas phase and solution phase values for the VCN stretch for
K"NCS' ion pairs.) Compared to the vCO mode of OCN', the more dramatic frequency
shifts calculated for the VCS and VCSC modes in SCN' and SeCN’ complexes are the result of
the greater dependence of these modes on the C-X bonds.

Despite the absence of a simple model that correlates the vibrational frequency shifts

in ionic and hydrogen bonded complexes, it is important to note that similar electrostatic

 

5»

Figure 35. Pictoral view of the Li(HOH)3NCO ion pair

lOl

interactions result in similar perturbations to the vibrational modes. For example, a single
ionic or hydrogen bond along the molecular axis to the nitrogen end of these anions will
increase the frequency of the vCN mode relative to the free anion. Two ionic or hydrogen
bonds to the nitrogen end of these anions will decrease the frequency of the VCN mode
below that of the single interaction. On the other hand, an interaction to each end of the
anion results in the greatest blue shift of the VCN mode among all the species studied. The
calculated frequency shifts strongly support previously assigned solution structures and
provide an important resource in assigning hydrogen bonded complexes with these anions.
The additional support from the predicted molar absorptivities provides added confidence in
the assignment of solution structures. The direction and magnitude of a frequency shift for
the stretching modes is dependent on the RED. for a given mode. The vCN mode for
OCN' is always blue shifted in ionic complexes since this mode is very dependent on the
C-0 bond. In the SCN' and SeCN‘ ions, negative frequency shifts are observed for dimer
and tetramer structures since the VCN mode in these anions is more dependent on the ON
bonds.

The calculations reveal the limitations of the valence bond model in understanding
the effect of electrostatic interactions on the OCN', SCN' and SeCN‘ anions. Although this
model successfully predicts the structural changes and shifts in the electron distribution for
ionic interactions in MNCX species, it is unable to explain many other interactions. For
example, the charge density in (Li+)2NCX' triple ions is not as greatly shifted to the
nitrogen end as in Be2"NCX' ion pairs, yet the ON bond in each complex is almost equally
weakened. For LilNCX'Li+ triple ions, the nitrogen atom has a slight increase in negative
charge compared to the free anion, yet the C-N bond is shortened, rather than slightly
lengthened. Finally, hydrogen bonding interactions to the nitrogen end of these anions
shorten the C-N bond rather than lengthening it, as ionic interactions do. A better

understanding of how these ionic and hydrogen bonding interactions affect the electronic

102

structure of the XCN' anions is obtained by following the perturbations to the molecular
orbitals, especially the 11: and 21: orbitals.

The shifts in the electron density of the 11: and 21: orbitals correlate well with the
structural changes in OCN', SCN', and SeCN' that result from electrostatic interactions.
The 11: and 21: orbitals are somewhat localized to the ON and C—X bonds in these anions.
For MNCO ion pairs, the 11: orbital is more bonding and the 21: orbital is less antibonding
around the C-0 bond relative to OCN'. The maximum electron density of the 21: orbital in
these ion pairs is pulled off the nitrogen atom, thereby reducing the bonding nature of this
orbital. These trends in electron density correlate well with the shortening of the C—0 and
lengthening of the ON bonds for these ion pairs with respect to OCN'. In Li*OCN’, the
electron density shifts towards the oxygen end, with the opposite effect on the C-0 and C-
N bond lengths relative to OCN'. Although all of the molecular orbitals are affected to
some degree by electrostatic interactions, the structural changes as a result of electrostatic
interactions are consistent with perturbations in the 11: and 21: valence molecular orbitals of
the anions in these complexes. Additional support for this model is found in the population
analysis. The occupancies of p, atomic orbitals are not substantially altered by electrostatic
interactions. Rather, the largest change in the atomic orbital occupancies are found in the px
and py orbitals which comprise the 11: and 21: molecular orbitals. This is consistent with
perturbations in the 11: and 21: valence orbitals of these anions having the greatest impact
on the structure changes in these anions

We have also applied the results of this theoretical investigation to our study of
solute-solvent interactions of OCN', SCN‘ and SeCN' anions with hydrogen bonding
solvents.'38"39 The unique frequency shifts and intensity changes predicted for the
stretching modes in these anions upon hydrogen bonding complexation have enabled us to

identify several specific hydrogen bonded complexes.

Chapter IV

Salvation of OCN'

A. Introduction

The cyanate anion has been used to probe ionic associations in alkali cyanate
solutions because its vibrational modes are accessible and sensitive to electrostatic
interactions.”77 The OCN‘ anion has three fundamental vibrations, all of which are
infrared active.88 The highest frequency mode, v3, which occurs at ~2140 cm", is
commonly labeled the vCN stretch.78'127 The lowest frequency mode, v2, which occurs at
~625 cm", is the doubly degenerate bending mode.”127 The vCO mode, V], which should
be observed at ~1250 cm", is in Fermi resonance with the first overtone of the bending
mode (250m)."‘88‘78'm The Fermi resonance couples and splits the VCO and 250CN
absorptions, resulting in two bands of similar intensity in DMSO solutions that are centered
at 1240 cm'l and displaced by 44 cm" on each side.77 Hereafter, absorption in this region
will be termed VCO & 2500,. It should be noted that the labels vCN and VC0 are convenient,
but somewhat misleading, since these vibrations are not localized at either the ON or CO
bonds.88"39 Rather, the vCN mode resembles the antisymmetric stretching vibration in CO2
or N 3188

Previous studies of alkali metal cyanates have shown that the stretching modes in
the cyanate anion are very sensitive to electrostatic interactions. For example, in lithium
cyanate ion pairs (LiNCO) in DMSO solutions the vCN mode increases by 34 cm‘l and the
vCO and 250CN absorption increases by 21 cm'1 from their respective absorptions in ‘free’
OCN'. Moreover, a number solution species (ion pairs, triple ions, and dimers) have been
identified by their characteristic vibrational frequencies. 77'88'78"27 Because of these unique
frequencies the factors involved in ionic association have been studied by vibrational
spectroscopy.

Therefore, the cyanate anion is a good probe of hydrogen bonding interactions
since the stretching modes in this anion are very sensitive to electrostatic interaction. A
further advantage of this anion is that each of the nuclei in the OCN' anion have
magnetically active isotopes that can be studied by multinuclear N MR spectroscopy. These

104

105

two complementary methods will be employed to obtain information about solute-solvent

interactions.

B. Results

1. Infrared Measurements

The VCN stretching vibration of the OCN' anion is sensitive to interactions in

solution. Hydrogen bonding broadens the band and shifts the peak frequency of the ch
band with respect to this mode in solutions of OCN’ in aprotic solvents, as shown in Figure
36 and summarized in Table 25. The bandshape of the spectral envelope is complex and
results from several overlapping component bands that can be resolved by techniques such
as Fourier self deconvolution. Quantitative information about the component bands was
obtained by fitting the observed spectral envelopes with Gaussian-Lorenztian bandshapes.
The resulting peak positions, bandwidths and absorbances are collected in Table 26. In the
aprotic solvents DMF and NM, the vCN mode consists of two component bands. The more
intense one is assigned as the ‘free’ vCN vibration, whereas the weaker component (4—5% of
the main band) is attributed to a hot band transition emanating from the first bending
vibrational excited state (v3 + v2 - v2), located below the fundamental.77 Dramatic changes
are observed in the 2200-2100 cm‘1 spectral envelopes for ~0.1M NBu4OCN in the
hydrogen bonding solvents: NMF, FA, and MeOH, with respect to the VCN mode in DMF
and NM. In NMF, the vCM spectral envelope is comprised of two bands, each about a
factor of two broader than the absorption in aprotic solvents. Only the higher frequency
component is observed in FA solutions. Interestingly, the vCN mode of OCN' in MeOH
consists of three constituent bands, two of which are similar to those seen in the amide
solvents, plus a new band at a higher frequency. Unfortunately, the vco & 250CN and 5001
vibrations are obscured by the hydrogen bonding solvents, so no information could be

obtained about these modes from the neat solution spectra.

Table 25. Infrared parameters for the vCN mode of ~0.1 M NBu4OCN in several solvents

106

 

 

VCN (cm") FWHH (cm‘l)
Solvent system $0.5 cm" 10.5 cm"
DMF 2136.5 8
NM 2142.0 9
NMF 2155.5 18
FA 2159.0 16
MeOH 2161.5 21

 

 

Table 26. Peak position, band width, and molar absorptivity parameters of component
bands of the vm spectral envelope of ~0.1M OCN' in DMF, NM, NMF, FA, and MeOH

 

 

Solvent system Peak Position vm (F WHH) Molar

( cm" ) (cm") absorptivity

(M"cm")

DMF 2136.5 10.1 7.9 :t0.1 21992125
2124.3 :1: 0.1 6.5 :1: 0.1 125 :1: 5

NM 2142.5i0.1 8.81201 211121:4

2130.4 :1: 0.2 8.1 i 0.2 92 :1: 4

NMF 2151.7i0.5 18.4:t0.1 1121 :75
2157.3 :1: 0.2 16.3 :1: 0.1 593 :1: 75

FA 2158.5i0.1 15.6iO.I 1652i4

MeOH 2151.8:t0.1 13.2:l:0.l 233i28
2160.6i0.2 12.1 21:07 418212189

2166.4 :1: 1.0 18.4 i 0.8 456 :1: 94

 

 

Molar absorbtivity (M"cm'l)

107

 

 

 

 

2500 —
2000 -1 FA NMF
1500 .. I
1000 _ MeOH
- \ DMF
500 _ \
O "" 1 r 1 l 1 l 1 l 1 l l l L

2190 2180 2170 2160 2150 2140 2130 2120 2110

Wavenumbers (cm")

Figure 36. VCN stretch of ~0.1 M NBu4OCN in DMF, NM, NMF, FA and MeOH

108

A clearer picture of the formation of anion-solvent adducts can be obtained by
adding either the protic solvent or the cyanate anion to a dilute solution of the other in a
weakly solvating (‘inert’) solvent, such as nitromethane (NM), has a low donor number.9
Infrared studies of the C-H stretching modes for NM show that this solvent only weakly
solvates anions.”0 Thus, the solvation of the cyanate anion by methanol can be followed
by measuring the 2200-2100 cm’1 spectral envelope of NBu4 OCN in NM as a function of
added MeOH. When no methanol is present, the vCN stretch of OCN' is a sharp band at
2142 cm“. As MeOH is added, the intensity of the 2142 cm'I band falls, with the spectral
envelope broadening and shifting to higher frequencies, as shown in Figure 37. Likewise,
similar curves are seen for the FA and NMF titration series, except that the extent of
complexation is lower (Figures 38 and 39). At intermediate concentrations of hydrogen
bonding solvent, an isobestic point is formed owing to the formation a 1:1 complex of the
OCN' anion with MeOH, FA or NMF. With higher concentrations of protic solvent, the
spectral envelope continues to broaden and shift to higher frequencies and the isobestic
point is lost. These results imply that the cyanate anion initially forms a 1:1 complex and
that an additional complex is formed at higher protic solvent concentrations. In order to
address the underlying composition of the spectral envelopes, the lineshapes were Fourier
deconvolved (Figure 40). The deconvolution procedure reveals that the absorbances of the
v3 fundamental and the (v3+v2-v2) mode of the OCN' anion dramatically decrease as the
concentration of MeOH increases. Concurrently, a new band at 2154 cm‘1 grows. Toward
the highest MeOH concentrations employed in this titration study (>2 M), the intensity of
the 2154 cm" band decreases and a new band at 2163 cm" grows. Similar results are seen
in the titration studies for NMF and FA. In these titrations, the free OCN’ absorption does
not decrease as fast with the amide solvents as with MeOH. The new bands that appear in
the amide titration series are not as blue shifted as those for the MeOH series

Quantitative information can be obtained by fitting the absorption envelopes to

Gaussian-Lorenztian sum components, which yields the absorbance, bandwidth, and peak

109

0.7 —

0.6 —

0.5 -~

0.4 A

0.3 —

Relative Absorbance

0.2 —

0.1 —

 

0.0 —

 

 

I I I I I I F T I I I 17 I I I I I I I I
2200 2190 2180 2170 2160 2150 2140 2130 2120 2110 2100
Wavenumbers (cm")

2.00 ——

1.75 — _,
2142 cm

/

1.50 —

2155 cm'1

    

1.25 —

 

 

Relative Absorbance
8
1

    

2163 cm"

   

 

 

 

l I I I I I I I I
2200 2190 2180 2170 2160 2150 2140 2130 2120 2110 2100

Wavenumbers (cm")

Figure 37. vCN stretch of 0.1M NBu4OCN in NM (top) and the Fourier

deconvolution of the vCN stretch (bottom) as a function of MeOH added

110

 

 

 

0.7 —
0.6 —1
0.5 —+
8 4
5
I—
o
(I) _
f2
0 0 3 —
.2
E —1
3; 0.2 -
0.1 —
0.0
I I I I I I I I I I I I l I I I I I I I

 

2200 2190 2180 2170 2160 2150 2140 2130 2120 2110 2100

Wavenumbers (cm“)

2143 cm‘1

    
  

21512 cm’1

0-75 ‘ 2157 cm"

Relative Absorbance

2131 cm'1

 
    

I I I I I I I I
2200 2190 2180 2170 2160 2150 2140 2130 2120 2110 2100

 

 

Wavenumbers (cm“)

Figure 38. VCN stretch of 0.1M NBu4OCN in NM(top) and the Fourier deconvolution of the

VCN stretch (bottom) as a function of formamide added

Relative Absorbance

Relative absorbance

0.6 —

0.5 -—

0.4 -

0.3 -—

0.2 —<

0.1—

111

 

 

0.0

 

 

l ' l ' l ' l ' l ' l ' I ' l ' I ' l

2200 2190 2180 2170 2160 2150 2140 2130 2120 2110 2100

 

 

Wavenumbers (cm")

/ 2142 cm'1

 

 

Tm

 

 

1‘1

2200 2190 2180 2170 2160 2150 2140 2130 2120 2110 2100

Wavenumbers (cm")

Figure 39. vcN stretch for 0.1M NBu4OCN in NM (top) and the Fourier

deconvolution of the VCN stretch (bottom) as a function of N-methylformamide added

112

 

 

 

 

 

 

 

 

 

0.6—
O
U
C 4
3 2154cm"
‘5 0.4—
(I)
.D
4: -
7; 2163cm"
a: 0.2—
00‘ 2143cm"
l'l'l'l‘l'l‘l‘l
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Concentration of MeOH (M)
0.6—
0
o .1
t:
:6 -1
'E 0.4+ 215lcm
8
.D a
<
7; 0,2- 2159cm"
a:
0.0— - 2142cm"
l‘l‘l'l'l'l'l'l
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Concentration of FA (M)
0.6—
0
E) 0.4—1
2
s - 2151cm'l
(I)
.0
<11 02—
B
0‘ - 2159cm"
0.0— O O 2142cm"
I‘TFTfil'l‘l'lTl

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

Concentration of NMF (M)

Figure 40. Absorptions for the component bands in the 2200-2100 cm’1 region for
~0.1M NBu4OCN/NM solutions as a function of MeOH (top). FA (middle)
or NMF (bottom)concentration.

113

position of each constituent band. The absorption curves in the methanol series can be well
simulated with four components: at 2130 cm" for the hot band, at 2143 cm’l for the free
anion, at 2154 cm‘1 for the first solvated complex, and at 2163 cm‘l for the second solvated
complex. The intensities of the fundamental constituents, plotted in Figure 41, are
qualitatively similar to those obtained by the Fourier self deconvolution of the observed
spectral envelopes. Similar titration experiments were performed by adding NMF and FA
to ~0.1M NBu4OCN in NM. Curve fitting and Fourier self deconvolution of the 2200-
2100 cm‘l spectral envelopes also show that the lineshape in each case is a convolution of
four component bands: the hot band (v3+v2-v2), the ‘free’ VCN band, a 1:1 hydrogen
bonded complex, and a 1:2 hydrogen bonded complex (from lowest to highest frequency).

The 1350-1180 cm’1 spectral region contains the vco & 2800., bands. In a 0.1 M
NBu4OCN solution in NM, these bands are centered at 1286 cm‘l and 1198 cm". Addition
of MeOH broadens the absorptions and shifts the maxima of the spectral envelopes to
higher frequencies. Moreover, new bands grow in with the increasing methanol
concentration, as illustrated in Figure 42. Fourier deconvolution is not necessary to
determine the positions of these new bands: at 1309 cm’1 and 1297 cm‘l for the higher
frequency component of the vcJO & 250CN absorption, and at 1219 cm" and 1209 cm’l for
the lower frequency component. The concentration dependence of the 1286 cm" and 1198
cm‘1 peaks is qualitatively similar to that of the 2142 cm’1 ‘free’ vcN stretch. Due to a strong
absorbance of the amide solvents, only the lower frequency component of the VCO & 280m
absorption is measurable. As in the studies of the vCN mode, the new bands that are a result
of amide-OCN' interactions are not as blue shifted as those found in the OCN""HOMe
complexes (Figure 41). In the FA titration, two new bands are observed at 1207 cm’1 and
1219 cm". Due to spectral interferences in the NMF titration, only a portion of the titration
could be collected which shows that a new band forms at 1207 cm‘1 as the NMF

concentration increases.

114

      

Absorbance (rel.)

 

 

0.08 -—
<'—‘ 1286 cm'1
0.06 ‘1 1297 cm'l a
‘—- 1198 cm"
0.04 —
1209 cm‘l
0.02 —
0.00
W 1 l l l I T j
1320 1300 1280 1260 1240 1220 1200 l 180

Wavenumbers (cm")

Figure 41. vCN and 250014 stretch for ~NBu4OCN in NM as a function of MeOH added

(a. 0 M, 0.04 M, C. 0.08 M, d. 0.12 M, and e. 0.16 M MeOH)

Rel. Absorbance

Rel. Absorbance

0.020

0.016 —

0.012 —«

0.008

 

0.000

115

  

1207 cm'1
1219 cm"

  

 

0.10
T

0.08 I

0.06 -

 

Wavenumbers (cm")

(a. 0 M, b. 0.04 M, c. 0.13 M, d. 0.17 M, 3.1 M FA)

   

1207 cm’l

  

 

0.04 —
0.02 -
0.00 —
l l I 7
1220 1210 1200 1190 1180

Wavenumbers (cm")

(a. 0 M, 0.04 M, 0.08 M. d. 0.51 M)

Figure 42. vCN and 2500: stretch for ~0.1 M NBu4OCN in NM as a function of

FA added (top) or NMF added (bottom)

116

The spectral envelopes of the vCO & 2500,, bands also can be modeled by Gaussian-

Lorenztian sums. The lower frequency band (1230—1180 cm") is more amenable to curve

fitting, since this spectral region has fewer interferences from the protic solvents. The

concentration dependence of the three bands in the 1230-1180 cm'1 region is similar to that

of the component bands in the VCN stretch, in that the absorbance of the band attributable to

‘free’ OCN’ plunges with increasing MeOH concentration, with the concomitant growth of

new bands. Qualitatively similar results are obtained in this spectral region from the NMF

and FA titrations. The results in both stretching spectral regions for the three protic

solvents are summarized in Table 27.

Table 27. Summary of peak positions and band widths of the curve-fitted component

bands of the vCN stretch of 0.1M NBu4OCN/NM solutions with NMF, FA, or

 

 

MeOH added.
VCN V10 86111 V60 & 250011“ Via Eco
Solvent Peak Position F WHH Peak Position F WHI-l
flstem ( cm") ( cm") (km/mo!) ( cm" ) (cm") (km/mag
NMF 2141.8 i 0.8 10.4 i 0.8 309 :1: 7 1198.2 i 0.5 11.2 i l 5.4 :1: 0.6
2150.8 :t 0.8 13.5 i 0.2 358 i 7 1207.2 i 0.2 5.0 i l 1.8 i 0.4
2159.2 :1: 0.2 13.9 i 0.8 320 i 20 b b h
FA 2142.0 i 0.5 10.6 i 1.2 322 :1: 7 1199.1 :1: 1.0 10.0 :1: 1 4.8 i 0.6
2150.6 :1: 0.9 12.9 :t 0.6 371 i 7 1207.1 1: 0.5 6.6 i l 2.4 i 0.6
2158.9 :1: 0.4 12.8 :t 1.0 330 :1: 20 1215.4 i 0.4 6.2 :t 1 0.8 i 0.2
MeOH 2142.8 :1: 0.7 10.5 i 1.6 311 i 4 1198.3 :1: 0.5 9.8 i 0.4 5.0 i 0.4
2154.0 :1: 0.9 12.3 i 1.2 399 i 4 1209.0 i 0.5 8.2 :t l 3.6 i 0.6
2163.3 :1: 0.4 14.0 i 0.3 450 :1: 20 1219.4 :1: 0.4 8.2 :1: 1 2.2 i 0.4

 

 

“only values for the lower frequency component are reported
bobscured by an NMF vibration

117

2. Assignments

The structure of the primary solvation sphere surrounding the cyanate anion in neat
solutions of NBu4OCN in MeOH, FA, and NMF can be understood by identifying the
component bands of the VCN stretch. The concentration behavior of the infrared
absorbances of the component bands in the titration studies of the cyanate anion with protic
solvents clearly shows that the OCN' anion is involved in two equilibria. Moreover, the
increase in the absorbance of the 2163 cm‘1 component in the MeOH titration comes at the
expense of the 2154 cm‘1 absorption (Fig. 41), suggesting that the 2154 cm’1 band can be
attributed to a 1:1 methanol:OCN' adduct and the 2163 cm‘1 band arises from a 2:1
MeOH:OCN' complex. Since the negative charge is distributed between the oxygen and
nitrogen ends of the anion, the methanol in these 1:1 complexes might be hydrogen bonded
to either the nitrogen or oxygen atom.

The OCN""HOMe and NCO""HOMe complexes can be easily distinguished from
one another by their infrared spectra. Ab initio calculations of these complexes show that
the frequency of the vCN mode will increase relative to free OCN' in either case. However,
these calculations predict that the frequency of the VCO stretch will increase in a
OCN""HOMe complex and decrease for a NCO"“HOMe adduct.I39 With respect to free
cyanate, it is also predicted that the absorptivity of the Vcn mode in OCN""HOMe will

139

increase and that of the vCO mode will decrease. The opposite trend is predicted for the
molar absorptivities of the stretching modes in a NCO""HOMe complex. From the
observed spectroscopic behavior, the 1:1 complex can be assigned with confidence as
OCN""HOMe. This assignment is supported by experimental studies of alkali cyanates,
where positive frequency shifts in the vCN and v00 & 250CN modes of OCN' with respect to
OCN‘ have been assigned to an alkali metal ion pairing to the nitrogen end of the cyanate
anion.”77 Likewise, the infrared spectra of OCN' in the NMF and FA titration series are

also consistent with the amide solvents hydrogen bonding to the nitrogen end of the cyanate

anion.

118

There are three possible ways that two MeOH molecules can interact with OCN’:
OCN""(HOMe)2, NCO“"(HOMe)2, and MeOH"'OCN""HOMe. Of these possibilities, the
NCO'"'(HOMe)2 complex, where two methanol molecules form a hydrogen bond to the
oxygen end, can be easily dismissed since no NCO""HOMe adducts were observed.
Moreover, the accompanying red shifts in the frequencies of the vCO & 280CN modes relative
to those in OCN' are not observed experimentally. Ab initio calculations for 1:2 complexes
of cyanate with water or hydrogen fluoride indicate that it is difficult to distinguish the three

139

structures based on the frequency shifts in the vCN mode alone. Again, the frequency of
the VCO mode and the molar absorptivities of the vCN and vCO modes provide the best way to
discriminate between the MeOH'"OCN"“HOMe and the OCN"“(HOMe)2 complexes. The
experimental frequency shift values and absorptivities in the MeOH, FA, and NMF titration
series suggest that both hydrogen bonding interactions are with the nitrogen end of the
cyanate anion in all three 2:1 solvates.

The vibrational assignments from the titration studies provide a firm foundation for
assigning the vibrational components of the vcN mode of 0.1M OCN' in neat solutions of
the anion in MeOH, FA, and NMF. They are collected in Table 28. The 2152 cm'l and
2157 cm’l components observed in neat N MF are similar to those seen in the titration
experiment. In neat FA solutions the 2151 cm‘1 band is absent; only the 2159 cm'1
component is observed. In neat MeOH solutions the component bands of the VCN mode are
shifted by ~2 cm‘1 from the values determined in the titration studies. The 2152 cm‘1
component can be attributed to the OCN""HOMe complex and the 2161 cm‘1 component to
the OCN""(HOMe)2. An additional component at 2166 cm’1 is observed in neat MeOH
solutions which can be assigned to the MeOH“'OCN""HOMe complex. The assignment is
based on ab initio calculations, which predict the frequency of the vCN mode will be higher
than those of the OCN""HOMe and the OCN""(HOMe)2 species.I39 Additional support is
provided by the experimental studies of alkali cyanates, where of the VCN frequency for the

LiOCNLi+ triple ion is the highest of all of lithium cyanate complexes studied to date.”77

119

Table 28. Assignments for the VCN and VCO& 25OCN modes for ~0.1M OCN' in neat
solutions with MeOH, FA, or NMF and in titration studies with these solvents

 

 

T01 vent system VCN component Vc~ component Vco & 260m component
bands in bands in titration bands in titration
Assignment neat solutions studies studies
NMF
,CH3
OCN'-"H" N 2152 cm'I 2151cm‘I 1207 cm‘l
‘CHo
.CHs
,N-CHo
,H
OCN' 1 2157 cm‘l 2159 cm'l ‘
N- CHO
CH3
FA
,H
OCN-...H-N , 2151cm" 1207 cm'1
‘CHo
V
’N-CHO
.° H
OCN 2159 cm" 2159 cm‘1 1215 cm'1
' H‘
1:1' CHO
H
MeOH
OCN'---H- Q 2152 cm'I 2154 cm'1 1209 cm'1
CH3
’O'CH3
,H
OCN- 2161 cm'1 2163 cm‘1 1219 cm‘I
H\
O ' CH3
Ha;
O‘H ”'0CN'WH'Q 2166 cm'1 b b
CH3

 

 

 

 

aobscured by solvent; bnot observed at the studied concentrations

120

3. NMR Measurements

Chemical shifts from the 14N and 170 NMR spectra of the cyanate anion in the five
solvents employed in this investigation are summarized in Table 29. The information
provided by NMR spectroscopy is less specific because the observed chemical shifts are a
population average over all species involved in the solution equilibria. Nonetheless, these
NMR results are consistent with the vibrational assignments. In NMF, the cyanate anion is
present in two hydrogen bonding environments. Relative to their values in aprotic
solvents, the decrease in the 14N chemical shift (more negative) and the increase in the 170
chemical shift (more positive) suggest that the electron density of the anion is shifted from
the oxygen end toward the nitrogen end. Ab initio calculations predict that a single or
double hydrogen bonding interaction to the nitrogen end will shift the distribution of the

'39 In FA, the further increase in the extent of the solvation

negative charge in this direction.
of the cyanate anion suggested by the infrared results is reflected by the increased positive
shift of the ”0 NMR chemical shift. On the other hand, the 14N chemical shift becomes
more positive with respect to the value in NMF solutions, which is not predicted by the
model ab initio calculations. This discrepancy shows that care must be employed in
correlating chemical shifts with atomic charges.

The l“N chemical shift of the OCN' anion has the most negative value in MeOH
solutions, where the '70 chemical shift has the smallest positive value among the protic
solvents. This pattern also fits well with the infrared results. The predicted atomic charges
on the anion in the MeOH"'OCN""HOMe species are similar to those of the free anion. The
increase in the negative charge on the oxygen end is reflected by a decrease in the ”0
chemical shift with respect to the values for the OCN' anion in the amide solvents. The
large negative |“N chemical shift of OCN' in MeOH solutions can also be explained through
the three hydrogen bonded species described earlier. The MeOH hydrogen bond is

stronger than those of NMF and FA, which results in a greater 1“N chemical shift in

121

OCN""HOMe and OCN""(HOMe)2 relative to the analogous NMF and FA species.
Although the l“N chemical shift in MeOH"'OCN°"'HOMe should be more positive than that
in OCN""HOMe and OCN""(HOMe)2, the observed chemical shift is dominated by
contributions from the OCN""HOMe and OCN""(HOMe)2 complexes.

Table 29. l“N and 170 NMR chemical shifts of ~0.1M NBu4OCN
in DMF, NM, NMF, FA, and MeOH

 

 

Solvent system "N chemical shift (10.02) ”0 chemical shift (10.1)
DMF -290.76 ppm 39.0 ppm
NM -293.28 ppm 38.4 ppm
NMF -301.25 ppm 42.9 ppm
FA -295.61 ppm 46.9 ppm
MeOH -310.65 ppm 41.1 ppm

 

 

4. Thermodynamic Parameters

The '4N and l7O chemical shifts of the cyanate anion were measured by NMR
spectroscopy for the MeOH, NMF, and FA titration series in NM in order to complement
the infrared studies. The temperature dependence of the l“N chemical shifts was also
determined. The concentration dependence of the l“N chemical shift for OCN’ upon
addition of MeOH, representative also of the NMF and FA systems, is shown in Figure
43. The concentration dependence displays a classic saturation curve for associating
complexes. The temperature dependence of the NBu4OCN/MeOH/NM dilution series

shows that the solvated complexes are more stable at lower temperatures.

 

l7O chemical shift (ppm)

14N chemical shift (ppm)

122

-302 — 1: -25.0°c
v -12.5°C

-300 _ v 00°C

-298 0 125°C
0 250°C

~296 —

-294 —

'292 l l l l 1 l l l

1
111
1.1

l

 

 

Concentration of MeOH (M)

42.5 —

42.0 _.

41.5 —

41.0 e

40.5 -—

40.0 e

39.5 e

39.0 —

 

 

38-5 1 1 1 1 1 1 1 1
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

Concentration of MeOH (M)

Figure 43. l“N and 170 NMR chemical shifts for 0.1 M NBu4OCN as a function

of MeOH added and temperature in the case for the l“N resonance.

123

In order to extract thermodynamic parameters from the concentration dependence of
the l“N and '70 chemical shifts or the intensity of the component infrared bands of the vCN
or VCO & 500,, absorption, an appropriate model must be applied to the N MR and infrared
results. The concentration dependence of the intensities of the 2154 cm'l and 2163 cm'1
bands in the methanol dilution series implies that methanol initially forms a 1:1 complex
with the cyanate anion. At higher MeOH concentrations the 1:1 cyanate-methanol complex
associates with an additional methanol molecule, forming a 1:2 cyanate-methanol complex.
Thus the components of the VCN absorption provide an important clue to the equilibria
responsible for the concentration dependence of the N MR and infrared measurements.

In addition to multiple associations of the cyanate anion with the hydrogen bonding
solvents, both the self association of the hydrogen bonding solvent and ion pairing of the
NBu4OCN salt must be considered and possibly incorporated into the model. Previous
investigations of tetraalklyammonium salts in nitromethane have shown that ion pairing is
negligible at 0.1M.2| Therefore, the NMR and infrared investigations have been modeled
by three equilibria: (a) the self association of the hydrogen bonding solvent (limited to the
dimerization), (b) formation of a 1:1 cyanate-hydrogen bonding solvent adduct, and (c)
formation of a 1:2 cyanate-hydrogen bonding solvent complex, as illustrated in equations
(8)-(10). Although these equilibrium expressions are written with the hydrogen bonding
interaction at the nitrogen end of OCN', the determination of the equilibrium constants is
independent of the site of the hydrogen bonding interaction. At the low concentrations of
the cyanate anion employed in this work, the equilibrium constants can be written in terms
of the concentrations of the species in the equilibria; the activity coefficients can be
neglected since charge species are involved in both sides of the equilibria. Thus the

equilibria can be expressed by equations (1 l)-( 13).

124

 

 

K1
K2
OCN""HOMe + MeOH Z OCN""(HOMe); (9)
K1)
2MeOH ::_.> (MeOH)2 (10)
where
K1 = [OCN:"'HOMe] (11)
[OCN ][MCOH]
_ [OCN'"'(HOMC)2] (12)
2 [OCN""HOMe][MeOH]
K1) = M1 (13)
[MeOH]

Equation (14) gives the total concentration of OCN ’ in terms of the concentration of the free
OCN' anion, [OCN'], the concentration of the 1:1 complex, [OCN""HOMe], and the
concentration of the 1:2 complex, [OCN""(HOMe)2]. Likewise, the total concentration of
MeOH, CMcOH, can be written in terms of the concentrations of free MeOH, [MeOH], the

1:1 complex, the 1:2 complex, and the methanol dimer, [(MeOH),], equation (15).

COCN = [OCN'] + [OCN""HOMe] + [OCN""(HOMe)2] (l4)

CMeOH = [MeOH] + [OCN""HOMe] + 2[OCN""(HOMe)2] + 2[(MeOH)2] (15)

125

Substitution of equations (11)-(13) into equations ( 14) and ( 15) and solution of the
resulting equations simultaneously for [OCN'] yields a quartic expression in [MeOH],

equation (16).

(2K,K,KD)[Me0H]4 + (K,1(2 + ZKIKD)[MeOH]3 + (K,1<,(2COCN - CMCOH) + K, +
2KD)[MeOH]2 + (K,(cOCN - CMcOH) + l)[MeOH] + CMeOH = 0 (16)

The roots of equation (16) can be determined analytically, from which an expression for the
concentration of free methanol can be derived in terms of the association constants (K,, K2,
and KD), CMCOH, and COCN. An expression for [OCN'] can then be written in terms of the
association constants and the total MeOH and OCN' concentrations by substituting equation
(11), equation (12), and the solution for equation (16) into equation (14).

The observed chemical shift of either the l“N or the 17O nucleus of the cyanate
anion, 80b5, is a population average of those for the free anion, the 1:1 complex, and the 1:2
complex as shown in equation (17), where xfm is the mole fraction and 8f,” is the chemical
shift of the free anion, x1 is the mole fraction of the 1:1 complex and 8, is its chemical
shift, and x2 is the mole fraction of the 1:2 complex, which has the chemical shift 52. With

the appropriate substitutions, equation (17) can be rewritten (equation (18)) such that the
Bobs = Xfreesfree + XI51 + X292 (17)

60,, = K,[OCN‘][MeOH] (6, - 8m)/C0CN+
K,K2[OCN‘][MeOH]2 (8, — 51...) / COCN+ 5,,e (18)

observed chemical shift is a function of the association constants (K,, K2, and KD), the free
OCN’ concentration, the free MeOH concentration, and the limiting chemical shifts for

OCN', OCN""HOMe, and OCN""(HOMe)2. Following substitution of the expressions for

126

[OCN'] and [MeOH], the values for the equilibrium constants, including KD, can then be
calculated by fitting equation (18) to the chemical shifts observed at different methanol
concentrations. These values are summarized in Table 30.

The formation constants for the hydrogen bonded complexes can also be
determined from the infrared band absorptivities. For example, the concentration of the 1:1
OCN""HOMe complex is proportional to the absorbance of the 2154 cm’1 component
(A254) as shown in equation (19), where a is the molar absorptivity and l is the pathlength
of the infrared cell. Upon substitution for [OCN'] and [MeOH], the absorbance of the
2154 cm‘l component becomes a function of the molar absorptivity, pathlength, equilibrium
constants, and the total concentrations of OCN' and MeOH. The fitting of equation (19) to
the absorbance of the component band resulting from the 1:1 complex also provides the
formation constants of the solvated complexes and the self association of the hydrogen
bonding solvent (K,, K2, and KD). Likewise, the KI association constant can be
determined by relating the vCN and vCO & 280CN infrared bands that are attributable to the
‘free’ OCN' anion to the total methanol and cyanate concentrations, as shown in equation
(21). This neglects the self association of the hydrogen bonding solvent, since this
equilibrium is negligible when the protic solvent concentration is less than 0.5M, as shown
by the small equilibrium constant, KD, for this association. The results are listed in

Table 31.
A2,54 = 82,54 6 [OCN""HOMe] = Kl £1154 l [OCN']f[MeOH]f (19)
A2142 = 82.42 If [OCN']/(1 + K,[MeOH]) (20)
The good agreement between the association values obtained from the NMR and the

infrared measurements confirms that these spectroscopic measurements were modeled with

the correct equilibria. The equilibrium constants for the 1:2 complexes in the amide

127

Table 30. Association constants, K,,K2 and KO, for the formation of
OCN"“HB, OCN'"(HB)2, and BH'"BH where HB can be NMF, FA, or MeOH, determined by

l“N and 170 NMR spectroscopy

 

 

 

 

Solvent
Kl (M4) K2 (M'l) K0 (Md)
Temperature 1 i0. 2 ° C #
NMF
”N NMR
25.0° C 4.8 :1: 0.1 0.26 i 0.04 0.06 :1: 0.01
12.5° C 5.7 i 0.2 0.28 i 0.09 0.07 i 0.02
0.0° C 6.8 i 0.2 0.28 :t 0.02 0.07 i 0.02
-12.5° C 8.3 :t 0.2 0.31 :1: 0.12 0.09 i 0.02
-25.0° C 10.0 i 0.1 0.34 :t 0.06 0.09 i 0.02
170 NMR
25.0° C 2.7 i l 0.33 i 0.20 0.04 i 0.02
FA
l“N NMR
25.0° C 4.4 d: 0.1 ' 0.06 i 0.03
12.5° C 5.2 i 0.1 ‘ 0.07 i 0.03
0.0° C 5.9 :t 0.1 ' 0.11 i 0.03
-12.5° C 7.3 :1: 0.1 ' 0.10 i 0.03
-25.0° C 8.8 i 0.1 a 0.09 :1: 0.04
'70 NMR
25.0° C 3.2 i 1.0 0.53 i 0.20 0.06 i 0.03
MeOH
1“N NMR
25.0° C 8.8 d: 0.4 0.40 i 0.15 0.10 i 0.01
12.5° C 10.9 :1: 0.4 0.55 i 0.40 0.10 :t 0.01
0.0° C 13.5 i 1.0 0.71 :1: 0.17 0.16 i 0.02
-12.5°C 18.0: 1.0 0.98:0.17 0.15 $0.02
-25.0°C 21.41 1.1 l.24i0.13 0.17:t0.02
170 NMR
25.0° C 8.5 :t 3 0.47 :t 0.20 0.12 i 0.1

 

“not determinable

128

solvents should be considered approximate, since the measurements that are important in
their determination lie near the detection limit. It is interesting to note that the 1:1 OCN‘
"'HOMe adducts, which have the largest vCN frequency shifts relative to those observed for
the corresponding FA and N MF bands, have the highest formation constant for these
complexes. A similar result has been obtained for alkali cyanates and alkali thiocyanates,
where ion pairings that result in the largest frequency shifts of the Vcn mode, as in LiNCO

ion pairs, also have the greatest ion pair formation constants.7°'75'77'32'85

Am, = 82154€[OCN""HOMe] = K,e,,5,t [OCN'][MeOH] (21)

Table 31. Association constants, K1, K2, and KD, for the formation of OCN""HB and OCN"'(HB)2, and
HB'"HB, where l-IB can be NMF, FA, or MeOH, determined by infrared spectroscopy

 

 

Solvent
K1 (Md) K2 (MJ) K1) (M'l)
Temperature (24 :1: 1° C)

NMF

A“98 4.3 i 0.8

A2142 6.7 :1: 0.6

A2151 5.5 i 0.4 0.20 i: 0.05 0.10 i 0.01
FA

AW, 5.8 :1: 0.9

2.42 5.6 i- 0.7

A215, 5.0 i 0.4 0.46 i 0.07 0.13 i 0.03
MeOH

Am, 9.4 i 0.7

A2142 8.3 i 1.0

Am4 8.7 i 0.7 0.37 :1: 0.04 0.12 i: 0.02

 

 

129

From the temperature dependence of the association constants derived from the 1“N
NMR measurements, the enthalpy, AH°, and the entropy, AS°, for these complexation
equilibria were determined; they are summarized in Table 32. The negative enthalpy term is
a result of the ion-dipole attraction between the anion and the hydrogen bond donor. The
thermodynamic parameters for the complexation equilibrium in N MF, FA, and MeOH
show that the entropy change is an important consideration in the formation of the
complexes. There is little difference among the enthalpies and entropies of formation for
the 1:1 complexes of OCN' with NMF, FA, and MeOH, although the MeOH complex has
the highest AH,°. The dramatic reduction of K2 from K, appears to result primarily from

the decrease in the entropy of the system as demonstrated by the methanol complexes.

Table 32. Thermodynamic parameters for the formation of 1:1 and 1:2
OCN""solvent complexes in nitromethane solutions.

 

 

Solvent system AH,° AS,° AH2° ASZ°
(kJ/mol) (Jmol'l K") (kJ/mol) (.lmol"K°’)
NMF -9.0 i 0.3 -17 :1: 1 -4.7 i 5.2 -27.8 :1: 17
FA -8.5 :1: 0.3 -16 i 2 ‘ a
MeOH -12.1 i 0.7 -22i2 -12.1i0.6 47i2

 

 

'not determinable

C. Conclusions

The infrared and NMR spectroscopic studies of ternary nitromethane solutions
provide a consistent and coherent picture of the structure and thermodynamics of anion-
solvent complexes in the primary solvation sphere surrounding the cyanate anion by the

hydrogen bonding solvents MeOH, FA, and NMF. Our investigations reveal that

130

hydrogen bonding interactions in the amide solvents occur at the nitrogen end of the anion.
In moderately dilute, neat formamide solutions, each OCN' is hydrogen bonded to two FA
molecules, whereas in MeOH, presumably a stronger hydrogen bonding solvent, the
cyanate anion participates in both singly and doubly hydrogen bonded complexes. In neat
methanol solutions, two distinct 1:2 cyanatezmethanol adducts are observed.

An understanding of the difference between the hydrogen bonding interactions of
the amide and methanol solvents is provided by ab initio calculations of the molecular
electrostatic potential (MEP) of the cyanate anion in various electrostatic environments.’39
The MEP predicts the site of a potential electrostatic interaction.'“ For example, it predicts
that a single ionic or hydrogen bonding interaction will occur at the nitrogen end of OCN',
although population analysis suggests that the negative charge is almost equally distributed
between the oxygen and nitrogen ends. In FH“'NCO‘, which models a strong hydrogen
bond, the most negative MEP is found in a ring surrounding the oxygen end of the anion.
The weaker hydrogen bond in an OCN""HOH complex only slightly perturbs the MEP in
OCN' such that the minimum lies in an are that is slightly off the nitrogen end of the OCN'
anion. These calculations demonstrate that the MEP in OCN' is very sensitive to the
strength of an electrostatic interaction and that strength of the first hydrogen bonding
interaction will determine the site of the second hydrogen bond.139

From the association constant values, it is clear that the amide solvents form weaker
hydrogen bonds with OCN' than those by MeOH. Moreover, in the titration studies the
perturbations of the VCN and Vco & 250e,,l frequencies in the cyanate-methanol complexes
are larger that those observed for similar amide complexes, which also reflects the relatively
greater strength of the cyanate-methanol interaction. The weaker amide interactions with
the cyanate anion result in hydrogen bonding only to the nitrogen end, owing to the smaller
perturbation in the MEP for the OCN' anion.

The thermodynamic parameters obtained in the titration studies provide insight into

the factors that are involved in the solvation of anions. Interestingly, the enthalpy and

131

entropy of formation for the hydrogen bonded complexes have similar values, which is
probably a leveling effect due to the nitromethane solvent. The negative entropy values
show that this term is dominated by the loss of freedom of the cyanate anion and the protic
solvent upon formation of the complex, rather than by the release of nitromethane
molecules that weakly solvate the complexation partners. In the MeOH titration, the
entropy of formation for the second hydrogen bonded complex is more negative than that
for the first complex. This is not surprising since in the 1:1 complex the nitrogen end of
the anion will be less solvated by nitromethane than the free cyanate. Therefore, the
entropy of formation for the second complex, A82, will be even more dependent on the loss
of freedom of the complexation partners than on the nitromethane molecules liberated.

The extent of solvation of cyanate in neat solutions of MeOH, FA, and NMF can be
understood by considering the role of entropy in the formation of intimate ion-molecule
complexes. The overall entropy loss when these solvents form a hydrogen bond with the
cyanate anion will be smaller than for the association of two free particles, since these
solvents are highly self-associated. This can be clearly demonstrated by comparing the
degree of solvation of OCN' in NMF and FA solutions. Although the association constants
for cyanatezamide complexes are comparable in NM, in neat solutions the cyanate anion is
more solvated in FA than in NMF. Since FA is more self-associated than is NMF,” "'4 the
entropy loss upon cyanate solvation will not be as great in FA as it is in NMF.

Comparison of the solvation of OCN' in FA with that of MeOH also supports this
hypothesis. The perturbations in the VCN stretch show that the OCN""HOMe hydrogen
bond is much stronger than that with FA. Yet, the cyanate anion is more highly solvated in
neat FA solutions than in MeOH. This seemingly inconsistent can be rationalized by the
greater self association of FA relative to MeOH; the entropy loss upon hydrogen bonding

with OCN' will be smaller in FA solutions.

Chapter V

Salvation of SCN' and SeCN'

A. Introduction:

The thiocyanate and selenocyanide anions have been used to probe ionic association
in alkali and alkaline earth metal complexes in solution since the stretching modes for these
anions are easily observed and are sensitive to electrostatic interactions.50'70'74'8'5M4“44
Both the SCN' and SeCN‘ anions have three fundamental vibrations, all of which are
infrared active. The highest frequency mode (v3), which is commonly labeled the vCN
stretch, occurs at 2058 cm" for SCN' and 2066 cm’1 for SeCN' in DMSO.83'88'”2 The
lowest frequency mode (v,), which occurs at 465 cm’1 for SCN' and 420 cm" for SeCN', is
the doubly degenerate bending vibration, labeled 55m and 8%, respectively.”13 The
second stretching vibration (v,) occurs at 735 cm‘1 in SCN‘, commonly labeled VCS. The
comparable vCSc vibration for SeCN' is easily observed experimentally and is predicted to

'39 The justification for these labels was

have a very low dipole moment derivative.

demonstrated by ab initio calculations, from which, for example the potential energy

distribution for the vCN modes of these anions was found to be about 90% C-N stretch.139
The perturbation of these stretching modes by electrostatic interactions have been

”'70'74'3"9"'42"“ From these investigations, the structures

studied by an number of authors.
of numerous solution species have been identified by their unique vibrational fiequencies.
In general, alkali and alkaline earth metal cations interact with the nitrogen atom of SCN',
which raises the frequencies of both stretching modes with respect to the unperturbed
anions. Similar results have been reported for the limited studies of SeCN‘ systems.
Interactions to the sulfur end of SCN', as exemplified by the Ag”SCN' ion pair, lowers the
frequency of the VCN stretch and raises the frequency of the vCS mode relative to SCN'.l5 In
dimer systems, such as (LiNCS)2 and (LiNCSe)2, each anion is bridged by a two lithium
cations which reduces the frequency of the vCN vibration and increases the frequencies of
the vCS and VCSe modesm"

Likewise, hydrogen bonding interactions with these anions perturb the frequencies

of the stretching modes, from which inferences about structures of the complexes that are

133

134

formed can be drawn. Perelygin observed that hydrogen bonding solvents shift the peak
maximum and broaden the spectral envelope of the VCN stretch of SCN' compared to those
seen in aprotic solvents.3s Corset assigned a blue shift of the vCN mode when he added
phenol to a solution of NBu4SCN in CCl4 to the formation of hydrogen bond at the
nitrogen of SCN'.38 Gill and coworkers measured the infrared spectrum of NBu4SCN in
MeOH and deconvolved the Vet»: spectral envelope into six component bands.43 Another
study of SCN' in methanol by Bencheikh showed that only four bands are necessary to
simulate the VCN spectral envelope (2090, 2071, 2055, and 2037 cm").45 Recently,
Hochstrasser and coworkers have studied the vibrational relaxation of SCN' in D20 and
MeOH.“°’41 Their results are consistent with the solvent hydrogen bonding to the nitrogen
end of the thiocyanate anion. X-ray diffraction studies of NaSCN in methanol imply that
two methanol molecules are hydrogen bonded to the nitrogen of SCN‘, with a bond angle
of 120°.23

As the perturbations in the vibrational modes reveal, electrostatic interactions with
SCN' and SeCN' alter their electronic structure. All of the nuclei in the SCN' and SeCN'

9 and with the exception of 33S which has a large

anions have magnetically active isotopes,5
quadrupole moment, the shifts in electron density can be easily studied by multinuclear
NMR spectroscopy. l5N NMR spectroscopy has been employed to identify the isomers of
organo—thiocyanates and isothiocyanates. RNCS compounds are characterized by a l5NMR
shift of -275 ppm. whereas RSCN compounds have a l5N chemical shift of -100 ppm,
which are both large shifts from that of the free anion at -165 ppm.145 Ionic associations of
SCN' with Li” have been studied by l5N NMR spectroscopy, where interactions to either
the sulfur or nitrogen atoms were identified by the paramagnetic or diamagnetic shifts,
respectively.I46 Musikas et a1. studied the binding of SCN' to several lanthanide ions by
the perturbations in the chemical shift and relaxation time of the nitrogen nuclei.147

Similarly, the 77Sc chemical shifts for SeCN' are extremely sensitive ionic interactions.‘48

Ionic interactions to the nitrogen end increase the shielding from -273 ppm for SeCN' in

135

DMSO to -318 in a zinc complex with SeCN'. Interactions to the selenium end,
exemplified by a mercury complex, decrease the shielding to -191 ppm. In contrast to the
large shifts observed for the nitrogen and chalcogen nuclei, the 13C resonances in SCN‘ and
SeCN' are not very sensitive to bonding at either end of these anions.149

The quadrupolar relaxation of the 1“N resonance provides an additional dimension
of information about solution equilibria. Formation constants of several ionic complexes
have been determined from the change in the line widths of quadrupolar nuclei (Cl).'5‘“54
In the case of SCN' in H20, Au-Yeng found that the 14N line width is sensitive to the ion-
solvent interaction; the concentration dependence of the 1"N line width is consistent with the
electron distortion model, which relates the distortion in the paramagnetic shielding
contribution of the chemical shift to the relaxation rate of the resonance.34

In this paper, we address the structure of the primary solvation shell surrounding
the thiocyanate and selenocyanide anions in nonaqueous protic solvents. The protic
solvents used in this investigation are methanol (MeOH), formamide (FA), and
N-methylformamide (NMF); this series allows a variety of hydrogen bonding interactions
to be studied. NMR and infrared spectroscopies are employed to probe the interactions in
these solutions and provide data from which the thermodynamics of the equilibria which
exist in them can be calculated. Spectroscopic measurements are reported for the
thiocyanate and selenocyanide anions both in neat solutions of the appropriate salt and

protic solvents, and in an ‘inert’ solvent where the concentration of the hydrogen bonding

solvent and salt can be varied.

B. Results:
1. Infrared Measurements
The vCN stretching vibration of the SCN' anion is sensitive to electrostatic
interactions in solution. Hydrogen bonding broadens the absorption and shifts the peak

frequency of this mode compared to solutions of SCN' in aprotic solvents, as shown in

136

Figure 44 and summarized in Table 33. The bandshape of the spectral envelope, especially
for methanol solutions, is complex and results from several overlapping component bands.
Spectral enhancement techniques such Fourier self deconvolution and curve fitting can be
applied to these spectral envelopes to obtain the underlying band structure, shown in Figure
45. The peak positions and band widths are collected in Table 34. Closer inspection of the
VCN mode for SCN' shows that this mode is asymmetric (Figure 46.). Previous
investigations of SCN' have shown that this spectral envelope in aprotic solvents can be
simulated by a Voight lineshape or Gaussian-Lorenztian sums.“I44 A single Voight
lineshape does not simulate this bandshape well. A modest improvement is obtained when
a Gaussian-Lorenztian sum is applied, yet the large residuals suggest that this spectral
envelope is not correctly modeled. The gas phase infrared spectra of SCN' and OCN'
include a v3 + v2 - v2 transition (bending hot band) that is 6.5 cm’1 and 11.5 cm‘1 red shifted
from the fundamental transition.‘2°"55"5° In solution studies of OCN’, the v3 + v2 - v2 mode
is observed 10-12 cm'1 red shifted from the fundamental, with an absorptivity consistent
with that predicted by the Boltzman distribution for this mode.‘""“0 The asymmetry of the
VCN mode along with the gas phase studies of SCN' imply that the hot band transition (v3 +
v2 - v2) should be incorporated in the model for the VCN stretching mode. This model
(Figure 46) correctly simulates the vCN spectral envelope in DMF and NM solutions where
we have assumed that the absorptivity for the hot band transition is given by the Boltzman
distribution. The vCN mode of SCN‘ in the protic solvents is much broader than in the
aprotic solvents. In the amide solvents, NMF and FA, the spectral envelope was found to
consist of the three bands with and additional band at higher energy found in the methanol
solution, as shown in Table 34. The vCN stretch for 0.025 M SeCN' in aprotic solvents is
also asymmetric and is well modeled by a incorporating a bending hot band 6.5 cm‘l below
that of the fundamental.

The vCS stretching mode of SCN' is also perturbed by protic solvents when

compared to this vibration in the aprotic solvents: DMF and NM. As seen in earlier

Table 33. Summary of position and band widths for the VCN stretch of ~0.25M NBu,SCN

137

in NM, DMF, NMF, FA and MeOH

 

 

Solvent system Vm (cm”) Avm 1cm")
NM 2059.5 13.5
DMF 2056.5 12.5
NMF 2057.8 33.8
FA 2059.5 27.5
MeOH 2060.0 40.5

 

 

Table 34. Summary of the deconvolved parameters for the component bands in

~0.25M NBu4SCN in NM, DMF, NMF, FA, and MeOH

 

 

Solvent system VCN (cm") Avm 1cm" )
NM 2059.2 :1: 0.1 12.2 i 0.1
2052.0 i 0.1 12.4 :t 0.3

DMF 2056.7 i 0.1 11.8 i 0.1
2050.0 i 0.1 12.3 i 0.2

NMF 2048.0 i 0.5 25.4 :1: 0.2
2058.2 :1: 0.2 15.1 i 0.8

2067.5 i 0.5 20.4 :t 0.2

FA 2050.1 :t 0.8 23.0 :1: 1.2
2057.8 :1: 0.3 14.0 i 1.8

2065.3 i 1.0 20.8 :1: 1.0

M80" 2043.9 :1: 0.7 29.3 :1: 0.3
2057.5 i 0.1 20.2 i 2.0

2072.8 it 0.2 24.8 i 2.0

2090.9 i 0.6 15.8 :1: 1.0

 

 

Molar absorptivity (M"cm")

138

 

 

 

 

1000—
800“ DMF
NM
600—
NMF
1‘
400— FA ;l\
1
M H
200_ e0
0 ..r
l l 1 l I l l l
2150 2125 2100 2075 2050 2025 2000 1975 1950

Wavenumbers (cm")

Figure 44. VCN for~0.25M NBu4SCN in DMF, NM, NMF, FA, and MeOH

139

0.4 —
0.3 —
ES
V 0.2 —
8
t:
.8
‘3‘
.o
<
0.1 —
0.0 a
l I l I l l l l

 

 

2150 2125 2100 2075 2050 2025 2000 1975 1950

Wavenumbers (cm' I)

Figure 45. Fourier self deconvolution of vCN for ~0.25M NBu4SCN in MeOH

Molar absorptivity (M"cm")

140

 

 

 

 

 

1000 —
v, fundamental
800 —
600 —
— Exp. lineshape
400 ._ — — Symmetric lineshape
200 —
v_,+v,-v2
\ Residuals
0 _J . .-.. //V , \r "' a
I l l l l l l l

2 l 00 2090 2080 2070 2060 2050 2040 2030 2020

Wavenumbers (cm")

Figure 46. VCN stretch for ~0.25M NBu4SCN in DMF

141

studies, this mode is more sensitive to electrostatic interaction than the vCN vibration. It is
clear that new bands are formed at higher frequencies in both the NMF and MeOH
solutions, despite the low sensitivity for this mode.

A clearer picture of the formation of anion-solvent adducts can be obtained by
adding either the protic solvent or the chalcocyanate anion to a dilute solution of the other in
a weakly solvating (‘inert’) solvent, such as nitromethane (NM), for which the donor
number is low. Infrared studies of the OH stretching modes for NM show that this
solvent only weakly solvates anions.I40 The low donicity of this solvent also precludes it
from strongly interacting with electropositive species. Thus, the solvation of the
thiocyanate anion by methanol can be followed by measuring the 2100-2000 cm" spectral
enve10pe of NBu4SCN in NM as a function of added MeOH. When no methanol is
present, the vCN stretch of SCN' is a sharp absorption at 2059 cm". As MeOH is added,
the intensity of the 2059 cm‘1 peak falls, with concomitant broadening of the spectral
envelope but virtually no change in the peak frequency. There are two isobestic points,
observed above and below the peak maximum, which imply the formation of two
thiocyanate-methanol complexes (Figure 47). In order to address the underlying
composition of the spectral envelopes, the lineshapes were Fourier self deconvolved
(Figure 48). The deconvolution procedures reveals that the absorbance of the central vCN
component decreases as the concentration of MeOH increases. Concurrently, two new
bands grow, at 2070 cm‘1 and 2045 cm". Similar results are seen in the titration studies
with NMF and FA.

Quantitative information can be obtained by fitting the absorption envelopes to
Gaussian-Lorenztian sum components, which yields the absorbance, bandwidth, and peak
position of each constituent band. The absorption curves in the methanol series can be well
simulated with four components: at 2051 cm‘1 for the hot band, at 2059 cm" for the free
anion, at 2070 cm'1 for a methanol-thiocyanate complex, and 2047 cm'l for a second

methanol-thiocyanate complex. The absorptivities of the fundamental constituents, plotted

Absorbance (rel.)

142

1.4 —

 

 

 

' l l l T l ' l ' l ' l ' l ' l
2100 2090 2080 2070 2060 2050 2040 2030 2020

Wavenumbers (cm")

Figure 47. VCN stretch for 0.1M NBu4SCN in NM with MeOH added

143

Absorbance (rel.)

 

 

 

l l l 1 l l l l
2100 2090 2080 2070 2060 2050 2040 2030 2020

Wavenumbers (cm")

Figure 48. Fourier deconvolution of vCN stretch for ~0.1M NBu4SCN
in NM as a function of MeOH added

144

in Figure 49, are qualitatively similar to those obtained by the Fourier self deconvolution of
the experimental spectral envelopes. Similar titration experiments were performed by
adding N MF and FA to solutions of 0.15 M NBu4SCN in NM and are summarized in
Table 35. Curve fitting and Fourier self decovolution of the 2100-2020 cm’1 spectral
envelopes also show that the lineshape in each case is also a convolution of four bands.
The absorption envelope for SeCN' can also be well simulated by four bands: at 2162 cm"
for the hot band, at 2169 cm’1 for the free anion, at 2078 cm‘1 for the a methanol-
selenocyanide complex, and at 2061 cm‘1 for a second methanol-thiocyanate complex

(Table 36).

Table 35. Summary of the peak positions and bandwidths (cm"), and molar absorptivities
from the curve-fitted component bands of the vCN stretch of 0.15 M NBu4SCN
with NMF, FA, or MeOH added

 

 

Solvent system Vc~ V,,2 VCS VCS
Peak Position F WHH Peak Position F WHH
NMF 2050.0 1 1.0 17.4 1 0.8 739.5 1 0.6 11.4 :1: 0.8
2059. 2 1 0. 2 12.2 1 0.2 751.0 1 0.4 10.3 1 2.6
2068.8 1 1.4 13.2 1 2.2
FA 2050.711.0 15.81 2.8 7398104 12.8112
2059.2.101 11.8103 7514116 13.011.6
2067..2120 14.3 :1: 1.8
MeOH 2047.0 1 1.0 20.4 1 3.0 739.5 1 0.3 11.4 1 0.4
2059. 4 1 0.1 12.8 :1: 1.2 751.7 1 0.8 14.5 1 2.0
2070.5 1 0. 4 13.2 1 4.0

 

 

Hydrogen bonding interactions with these anions not only perturb the frequencies
of the stretching vibrations, but also absorptivity of these modes. The integrated
absorptivity for the vCN stretch increases by 30% in the methanol titration series. Likewise,

the integrated intensity for the VCN stretch in the NMF and FA titration series each increase

145

by 13%. Experimental studies have shown that ionic and hydrogen bonding interactions to
the nitrogen of SCN' increases the absorptivity the vCN stretch.4M They are supported by a
theoretical study by our group which predicts that electrostatic interactions to the nitrogen

of SCN' and SeCN' increase the molar absorptivity of the vCN mode in these anions.157

Table 36. Summary of the peak positions and bandwidths from the curve-fitted component
bands of the VCN stretch of 0.025 M NBu4SeCN with MeOH added.

 

 

Solvent system VCN v,,2
Peak Position F WHH
(cm" (cm")
MeOH 2060.7 1 1 6 7.93 1 1 80
2068.8 1 0 1 5.30 1 O 52
2077.5 1 0 4 7.69 1 O 84

 

 

The 770-720 cm’1 spectral region contains the vCS mode. In a 0.15 M NBu4SCN
solution in NM, this band is centered at 740 cm". Addition of MeOH broadens the
absorption and shifts the maximum of the envelope, with growth of a band at 752 cm", as
shown in Figure 49. This spectral envelope can be simulated adequately by two Gaussian-

Lorenztian sums.

2. Assignments
The structure of the primary solvation sphere surrounding the thiocyanate anion in
neat solutions of NBu4SCN in MeOH, FA and NMF can be understood by identifying the
component bands of the two stretching modes. The concentration dependence of the
infrared absorbances for the component bands in the titration studies (Figure 49) shows
clearly that the SCN' anion is involved in two equilibria that have 1:1 stoichiometry. The

most straight forward assignments are those for the unperturbed or ‘free’ SCN'; these

146

1.4 4

1.0 -

0.8 -

Absorbance (rel.)

2059 cm’1
0-4 ‘ 2070 cm"

2047 cm‘2
0.2 —

0.0 -—

 

 

1 1 l 1 l l l
0.0 0.5 1.0 1.5 2.0 2.5 3.0

Concentration of MeOI-I (M)

Figure 49. Absorptivities for component bands of the VCN mode

of ~0.1M NBu4SCN in NM as a function of MeOH added

147

bands occur at 2059 cm" for the ch mode and 740 cm‘1 for the vCS mode. Positive
frequency shifts for the VCN mode, upon complexation, can be due to interactions along the
molecular axis at the sulfur or nitrogen in SCN'.74 These two types of interactions can be
easily distinguished by looking at the perturbations of the vCS mode. An electrostatic
interaction to the nitrogen atom in SCN' will blue shift this mode whereas an electrostatic
interaction to the sulfur will result in a red shift of the VCS mode. Clearly, there is no band
lower in energy than that for the ‘free’ SCN', which implies that the 2070 cm’1 band must
be due to a SCN""HOMe complex. The red shifted bands in the vCN mode of SCN' have
previously been assigned to nonaxial interactions to the thiocyanate anion, where one or
two solvent molecules are associated with the complex. The concentration dependence of
the 2047 cm’1 band suggests that the stoichiometry of the complex is 1:1. Thus, the lower
frequency band is attributed to a nonaxial interaction to the nitrogen of the thiocyanate
anion. Theoretical calculations support these assignments. They predict that hydrogen
bonding to the nitrogen atom by hydrogen fluoride or water will increase the frequency of
the stretching modes when the interaction is along the molecular axis.I39 On the other hand,
a nonaxial hydrogen bonded complex will be characterized by a blue-shift for the VCN
stretch and a slight red-shift in the VCS stretch.158 Additional support for this unusual
hydrogen bonded complex is derived from the crystal structure of
Na+(12C4)2 SCN'-HOMe, where the C-N-O angle is 108°.47 The infrared spectrum of the
vCN mode in this complex is 10 cm‘1 lower than that of the VCN stretch in Na*( 12C4)ZSCN’.
Likewise, the same assignments can be applied to the component bands seen in the
selenocyanide titration. The low frequency band can be attributed to a non-axial interaction
to the nitrogen end of the anion and the high frequency component can be assigned to a
component that has an axial interaction to the nitrogen end.

The assignments for the titration experiments can be applied to the neat protic
solutions. In the amide solutions, the same number of vibrational components are found.

As in the titration experiments, a band above and below the component at 2058 cm", which

148

can be assigned to the ‘free’ SCN', are found. Again, this can be attributed to interaction
alone the molecular axis of the anion and a nonaxial interaction. It should be noted that our
experiments probably cannot differentiate between one and two nonlinear interactions since
the bands are so strongly overlapped. Similar structure is observed in methanol solutions
where the band at 2058 cm'I can be assigned to ‘free’ SCN', the band at 2044 cm‘1 to a
non-axial interaction, and 2073 cm". The additional band at 2091 cm‘l is probably due to

at MeOH"'SCN""HOMe complex since a red-shift in the VCS mode is not observed.

3. Thermodynamics

In addition to multiple associations of the thiocyanate and selenocyanide anions
with the hydrogen bonding solvents, both the self association of the hydrogen bonding
solvent and ion pairing of the NBu4SCN and NBu4SeCN salts in nitromethane must be
considered and possibly incorporated into the model for the solution equilibria. Previous
investigations of the tetraalklyammonium salts in nitromethane have shown that ion pairing
is negligible at 0.1M.48 Therefore, the NMR and infrared investigations have been
modeled by three equilibria: (a) the self association of the hydrogen bonding solvent
(limited to the dimerization) and (b) formation of two vibrationally distinct 1:1 thiocyanate-
hydrogen bonding solvent adducts, as illustrated for MeOH in equations (22)-(24).
Although these equilibrium expressions are written with the hydrogen bonding interaction
at the nitrogen end of SCN', the determination of the equilibrium constants is independent
of the site of the hydrogen bond. At the low concentrations of SCN' and SeCN' employed
in this work, the equilibrium constants can be written in terms of the concentrations of the
species in the equilibria; the activity coefficients can be neglected since a negatively charged
species is on each side of the equilibrium. Thus the equilibria can be expressed by

equations (25)-(27).

SCN + MeOH :22 SCN"’HOMe

K2
SCN' + MeOH = scrxt
.HOMe

KD
2MeOH == (MeOH)2

where

[SCN""HOMe]1
[SCN‘MMeon]f

 

[SCN'°“HOMe]2
[SCN']f[MeOH]f

 

2 =

[(MeOH)2]
[MeOH]f2

(22)

(23)

(24)

(25)

(26)

(27)

Equation (28) gives the total concentration of SCN‘ in terms of the concentration of the free

SCN' anion, [SCN'],, the concentration of the first 1:1 complex, [SCN""HOMe],, and the

concentration of the second 1:1 complex [SCN""HOMe]2. Likewise, the total concentration

of MeOH, [MeOH]T, can be written in terms of the concentrations of free MeOH,

[MeOH],, the two 1:1 complexes, and the methanol dimer, [(MeOH),], equation (29).

[SCN']T = [SCN'], + [SCN""HOMe], + [SCN""HOMe]2

[MeOH]T = [MeOH]f + [SCN""HOMe]l + [SCN"”HOMe]2 + 2[(MeOH)2]

(28)

(29)

150

Substitution of equations (25)-(27) into equations (28) and (29) and solution of the
resulting equations simultaneously for [SCN‘]f yields a cubic expression in [MeOH],,
equation (30). The roots of equation (24) can be determined analytically, from which an
expression for the concentration of free methanol can be derived in terms of the association
constants, KI+K2, and KD. It should be noted that due that KI cannot be separated from
K2, only the sum can be determined. An expression for [SCN‘]f can be written in terms of
the association constants and the total MeOH and SCN' concentrations by substituting

equation (25), equation (26), and the solution for equation (30) into equation (28)

{2KD(K. + K2)}[MeOH],3 + (KI + K, + l)[MeOH]f2 +
{(K,+K2)([SCN']T-[MeOH]T)+1}[MeOH]f - [MeOH]T = 0 (30)

The observed chemical shift of either the 14N or the 77Se (for SeCN') resonance in
these anions, 60b5, is a population average of those for the free anion and both 1:1
complexes as shown in equation (31), where the xfm is mole fraction and of,“ is the
chemical shift of the free anion, x, is the mole fraction of the first 1:1 complex and 5, is its
chemical shift, and x2 is the mole fraction of the second 1:1 complex, which has the
chemical shift 82. With the appropriate substitutions, equation (31) can be rewritten
(equation(32)) such that the observed chemical shift is a function of the association
constants (K1, K2, and K0), the free SCN' concentration, the free MeOH concentration,
and the limiting chemical shifts for the free SCN' and the 1:1 complexes. Following
substitution of the expressions for [SCN‘]f and [MeOH],, the values for the equilibrium
constants, including KD, can then be calculated by fitting equation (32) to the chemical
shifts observed at different methanol concentrations. These values are summarized in Table

37 and illustrated by the MeOH titration in Figure 51.

151

0.30 -—

0.25 A

0.20

0.15

Re]. Absorbance

0.10

0.05

 

 

 

Wavenumbers (cm")

Figure 50. vCS stretch for ~0. 1M NBu4SCN in NM as a function of MeOH added

152

 

 

 

-25.0“C
-176 — D-12.5°C
00°C
-174- .12.5°C
25.0°C
-172—
g 470—
a.
5'
c
".5 468*—
5”
§
g -m
0
«Z
-164-
-162-
‘lwl'l'l‘l'l'l‘ll'l
00 05 1.0 15 20 25 3.0 35 4.0

Concentration of MeOH (M)

Figure 51. 14N NMR chemical shift for ~0.1M NBu4SCN in NM as a function of

MeOH concentration and temperature

' 153

5 = Xfreesfree + X151 + X252 (31)

obs

(K16,+K282)[SCN']f[MeOH],/[SCN‘]T + 51... (32)

obs

The line width of the 14N resonances can also be used to obtain association
constants for these adducts. Under the condition of fast exchange between the free (1) and
the bound (b) sites, and in the absence of exchange broadening, the spin lattice relaxation
time (lff 101,3), equation (33) is the population average of the relaxation times for the free
(l/T”) and the bound species (l/T 1b)' For quadrupolar nuclei, such as 1“N, in asymmetric
environments the longitudinal relaxation rate is very fast and is the dominant relaxation
pathway.159 In such systems, the approximation l/Tlobs = 1/1‘ 20115 can be made, thereby
relating the line width of the MN resonance to the association constants for the complex
formation since the line width of an NMR resonance (Av) is proportional to the reciprocal
of the transverse relaxation rate, equation (34). Again, the relationships for [SCN']f and
[MeOH]f discussed previously can be substituted for the concentrations of free SCN' and

MeOH. Figure 52 shows the 14N line width dependence on temperature and MeOH for

 

 

 

SCN'.
1 = .711 + x_b z 1 (33)
Tlobs Tl f T1 b T20bs
A _ 1 _ xf xb _ [SCN']f[MeOH]f[ 1 1 ] 1
V1/2 - — - — + —— — — - — + —
“T2 “T1 f KT] b “[MeOHlT T] b Tl f KT] f

(34)

”N linewidth (Hz) for SCN‘

Figure 52.

 

154

 

 

500 a
-25.o°c
400 ~
300 7 425°C
‘ o.o°c
200 — , . 125°C
v e 0
, 0 ° - o 25.0°c
‘ . O
100 — H
l l l ‘ l ' l ' l l l ' 1
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Concentration of MeOH (M)

l4N line width of NBu4SCN in NM as a function of MeOH and temperature

Table 37. Association constants, Kl + K2 and KD for the formation of the two

155

1 :1 complexes and self association of the protic solvent, determined by l“N NMR measurements

 

 

 

Solvent system "N chemical shift ”N chemical shift "N line width l“N line width
"(111(2) K1) (K1119) K1)
MeOH
250°C 0.83 1 0.01 0.10 1 0.01 0.84 1 0.03 0.10 1 0.01
12.5°C 1.05 1 0.01 0.11 1 0.01 0.95 1 0.05 0.10 1 0.01
0.0°C 1.37 1 0.01 0.13 1 0.01 1.34 1 0.05 0.14 1 0.01
—l2.5°C 1.70 1 0.02 0.15 1 0.01 1.64 1 0.09 0.15 1 0.01
-250°C 2.07 1 0.02 0.17 1 0.01 2.12 1 0.07 0.14 1 0.01
FA
250°C 0.76 1 0.01 0.07 1 0.01 0.74 1 0.02 0.07 1 0.01
125°C 0.90 1 0.01 0.07 1 0.01 0.92 1 0.05 0.08 1 0.01
0.0°C 1.05 1 0.01 0.09 1 0.01 1.01 1 0.04 0.08 1 0.01
—12.5°C 1.27 1 0.01 0.10 1 0.01 1.25 1 0.05 0.09 1 0.01
-250°C 1.46 1 0.02 0.11 1 0.01 1.43 1 0.05 0.10 1 0.01
NMF
250°C 0.71 1 0.01 0.05 1 0.01 0.70 1 0.04 0.04 1 0.01
125°C 0.85 1 0.01 0.06 1 0.01 0.85 1 0.05 0.05 1 0.01
0.0°C 0.94 1 0.01 0.07 1 0.01 0.94 1 0.06 0.06 1 0.01
-12.5°C 1.00 1 0.01 0.07 1 0.01 0.99 1 0.03 0.06 1 0.01
-250°C 1.18 1 0.03 0.09 1 0.01 1.22 1 0.09 0.07 1 0.01

 

Similarly, the 77Sc and 14N resonances in NBu4SeCN are sensititive to electrostatic

interactions and the concentration dependences can be fit to the models described earlier.

Applying these association models, the K1+K2 parameter from the 77Se chemical shift is

0.68 1 0.06 M'l with a value of 0.69 1 0.01 M‘1 from the 1“N resonance. As seen in the

previous l“N experiments for SCN’, the line width of the nitrogen resonance for SeCN'

increases as a function of MeOH added, yet the line width for the 77Sc resonance of SeCN'

remains constant throughout the titration series. This observation precludes the line

broadening in the 1“N resonances SeCN' from being a result of exchange broadening.

The formation constants for the hydrogen bonded complexes can also be

determined in a number of ways from concentration dependence of the infrared band

156

absorptivities. The absorptivity of the 2059 cm‘1 component can be related to [SCN']f and
[MeOH]f as shown in equation (23), where 82059, is the molar absorptivity and I? is the
pathlength of the infrared cell. Upon substitution for [MeOH],, the absorbance of the 2059
cm" component becomes a function of the molar absorptivity, pathlength, equilibrium
constants, and the total concentrations of SCN‘ and MeOH. The fitting of equation (35) to
the absorbance of the v2059 band provides the formation constants of the solvated
complexes and the self association constants for the solvated complexes. Likewise, the
absorbances for the blue and red shifted component band in the methanol titration can also
be related to the total SCN' and MeOH concentrations, shown in equation (36). Nonlinear
regression of the band absorptivities to equation (36) yields the sum of the association
constants (K,+K2), the self association constant for MeOH (KD), and the product of the
molar absorptivity for that band and association constant for the complex (3207019)-
Likewise, a similar relationship can be derived for the red shifted component, described in

equation (37). The integrated intensity

82059€[SCN']T

(35)
1+ (K, + K2)[MeOH]f

 

A2059 = 320593150811 =

82070€K1[SCN-]T[MCOH]f
1 + (K, + K2)[MeOH]f

 

A2070 3' €2070€[SCN-WHOMC]I = €2070£K1[SCN-]f[MCOH]f =

(36)

A _ €2047€K2[SCN-]T[MCOH]f
20‘” 1 + (K, +K2)[MeOH]f

 

(37)

157

of the vCN mode provides fourth method of obtaining the association constants, since the
molar absorptivity of the complexes are larger than that for the unperturbed anion. Here,
the integrated absorptivity (fl) is the sum of all the component band absorbances, equation
(38). Equation (38) can be rewritten so that the integrated absorbance is a function of the
total concentrations for SCN' and MeOH by substituting the relationship for [MeOH],. It
should be noted that the inclusion of the second solvate cannot be distinguished from only
one 1:1 equilibria. The parameter obtained by fitting this model to the experimental data
provides additional confidence that the vCN spectral envelope was correctly simulated since
the model for the integrated absorptivity makes few assumptions about the equilibria
involved in the vcN spectral envelope. If the vCN mode was simulated with only two bands
Gaussian/Lorenztian bands, the association constant for this complex is substantially

smaller than the results shown here. These results are collected in Table 38.

fl = A... + 21, + A =efm€[SCN'], + e,€[SCN'"'HOMe], + e,€[SCN""HOMe]2
(38)

_ ascmT
A _ 1 + (K1 +K2)[MeOH]f(£°°° + (K,e, +K282)[MeOH],) (39)

 

The vCS mode, in the titration series, was found to be a composite of two bands.
The concentration dependence of these component bands can fit to several models. The
model that best fits the experimental data and is consistent with previous results assumes
that the 740 cm'1 is a composite band with contributions from the ‘free’ SCN' and one of
the solvated complexes and the 751 cm" is a sum of a NBu,+ absorption and the other

solvate complex. The absorbance for the 740 cm’l can be related to the total concentrations

158

of SCN' and MeOH, equation (40). The associations constants from this mode are in
excellent agreement with those obtained for the VCN mode, Table 39.

The infrared and N MR experiments only yield a sum of association constants
(K,+K,) since molar absorptivities and chemical shifts for the specific complexes are
convolved with the association constants in the equilibrium model. Theoretical calculations
of a number of SCN'/HOH and HF adduct have been computed by our group, which
provide predictions for the changes in absorptivity upon hydrogen bond formation.'39 It
turns out, quite fortuitously, that the actual values for the molar absorptivities do not need
to be known, as shown in equation (41). The ratio of the two adduct bands, A2070 and
A204,, for the methanol series, can be related to the association constants for these
complexes multiplied by their respective molar absorptivities. Likewise, a similar
relationship can be derived for the vCS stretch. The individual association constants, using

this assumption are collected in Table 40.

31308-11

A740 = EfrcchCN-]f + EZZISCN-WHOMC]2 = I + (K +K )[MCOH]
l 2 T

 

(Efrec + 82K2UVICOI-Hf )
(40)

A2070 = 82070[SCI‘I_"'I'I()IVIC]1 = 82070K1
A2047 82047[SCN-WHOMC]2 82047K2

 

(41)

Even though K, and K2 cannot be separated in the NMR experiment, it was found
that the temperature dependence of these association constants fit the usual Gibb’s free
energy relationship. The computed enthalpy and entropy values from these association
constants are a convoluted from the two equilibria that they represent. Despite the difficulty
in interpreting the meaning and significance of these parameters, they are fairly similar to

results that we observed for complexes of the cyanate anion with these protic solvents.I37

Table 38. Association constants (K, + K2, and KB) of SCN’ with MeOH, FA, and NMF

from the stretching modes of SCN'

 

 

 

 

 

 

Solvent system A2053 Ablue shift Aredshift Integrated A740
absorptivity
MeOH
K,~1-K2 0.86 1 0.02 0.86 1 0.01 0.86 1 0.01 0.83 1 0.01 0.82 1 0.04
KD 0.09 1 0.01 0.10 1 0.01 0.10 1 0.01 0.09 1 0.01 0.10 1 0.01
FA
K,+K2 0.71 1 0.01 0.70 1 0.01 0.72 1 0.01 0.72 1 0.02 0.68 1 0.09
KB 0.06 1 0.01 0.06 1 0.01 0.06 1 0.01 0.07 1 0.01 0.06 1 0.01
N MF
K,+K2 0.67 1 0.01 0.62 1 0.01 0.71 1 0.01 0.71 1 0.02 0.67 1 0.06
KD 0.06 1 0.01 0.05 1 0.01 0.05 1 0.01 0.05 1 0.01 0.05 1 0.01
Table 39. Individual association constants using the extinction coefficients from
ab initio calculations.
VCN Vcs Vcs
Solvent system K, (M") K2 (M") K2 (M”) K2 (M")
MeOH 0.46 1 0.04 0.48 1 0.08 0.40 1 0.04 0.38 1 0.07
0.31 1 0.02 0.30 1 0.02 0.40 1 0.02 0.41 1 0.02
NMF 0.27 1 0.02 0.16 1 0.10 0.40 1 0.02 0.51 1 0.30

 

 

160

Table 40. Thermodynamic parameters for the association of SCN' with MeOH, FA and
NMF from the NMR measurements

 

 

"N chemical shift "N line width
Solvent system AH“; (kJ/mol) ASH," (Jmol"K") AHM" (kJ/mol) AHM" (Jmol"K")
MeOH-..114107 -4013-..118108 -4113
FA “82103 -3011 “80106 -2912
NMF -5.8107 -2212 -6.2108 2413

 

 

Here, the AS,+2 values are larger than those for the cyanate association and the enthalpy

values for the thiocyanate complexes are smaller than those for the cyanate complexes.

C. Conclusions

The interactions of the thiocyanate and selenocyanide anions are weak. In fact, the
strength of these interaction are an order of magnitude weaker than our previous
investigation of the solvation of the cyanate anion by a number of protic solvents.137
Needless to say, the study of such weak complex is difficult. Here, we have studied
anion-solvent complexes by a number of infrared and NMR techniques in order to be
confident that the correct structure of the primary solvation sphere has been obtain. The
concentration dependence of the integrated intensity for the VCN mode, the I4N chemical
shift and the 14N line width all fit to a 1:1 equilibrium model. It should be noted that these
models do not discriminate between two 1:1 equilibria and a single 1:1 equilibrium. The
models that we applied to the infrared spectral envelopes are not without ambiguity. For
example, the VCN mode in the titration experiments could have been modeled by only three
bands (the hot band, the ‘free’ SCN‘ and a single SCN‘/MeOH complex) instead of four
bands, but the fitted parameters from this procedure are not self consistent. The peak
position and band widths of the component bands are not constant. Additionally, this

spectral envelope may be fit with more than four components as seen by Gill and co-

workers.43 Although four bands are more than adequate to simulate this spectral envelope,

161

due to the small residuals, a large number of bands could be convolved in this lineshape.
The vCS stretch, which is very sensitive to electrostatic interaction, is very diagnostic for the
number of vibrational components. This mode precludes a large number of vibrational
components. Without applying a two 1:1 equilibrium model to deconvolve the stretching
modes, the association constants would not be in agreement with those obtained from the
integrated absorbance, the l“N chemical shift and the l“N line widths. Therefore, we
believe that we have correctly determined the solution equilibria for our titration
experiments by the harmony found between the various models.

The SCN' and SeCN' anions are much less solvated than OCN‘. Whereas the OCN'
anion has 1.5 to 2 solvent molecules associated with it in its ion pair, a significant
proportion of the SCN' and SeCN' anions are not hydrogen bonded in neat solution of
MeOH, FA and NMF. The titration studies show that the enthalpy is smaller and the
entropy is larger for SCN' and SeCN' as compared to OCN'. Gas phase ab initio studies of
these anions show that the cyanate anion has a more negative electrostatic potential off the
nitrogen atom than SCN' and SeCN‘.4° The smaller electrostatic potential for SCN' and
SeCN' results in a smaller hydrogen bond enthalpy formation. Even though the charge of
the OCN', SCN' and SeCN' anion is -1, the distribution of this charge is also important in
the how these anions order or structure the solvent molecules surrounding them. The
smaller electrostatic potentials for SCN' and SeCN' do not order the nitromethane molecules
surrounding them as well as OCN'. Thus upon formation of a hydrogen with SCN' or
SeCN', fewer solvent molecules will be released thereby resulting in a more negative

entropy than for OCN'.

Chapter VI

Crystal Structures

A. Introduction

The environment of a cation or anion in solution depends on many factors. A
variety of experimental techniques have been used to determine or infer the structures
surrounding anions and cations in electrolyte solutions. Of these, infrared spectroscopy
has been widely applied to the study of solution equilibria, and has been used to identify
such species such as ion pairs, dimers and tetramers.85 Ionic and hydrogen bonding
interactions with the thiocyanate anion perturb its vibrational modes. The perturbation of
the stretching modes, vCN and vcs, are unique for a given structure. For example, ionic
interactions to the nitrogen end of the thiocyanate anion increase the frequencies of both the
stretching modes, whereas an interaction to the sulfur end of the anion raises the frequency
of the vCN mode and lowers the frequency of the vCS mode, with respect to the unperturbed
anion.”144 Negative frequency shifts for the vCN vibration, relative to ‘free’ SCN‘, are the
result of two electrostatic interactions to the It cloud of the C-N bond, as illustrated by
dimers and tetramers.85

Positive vCN frequency shifts for SCN""HOMe complexes have been attributed to a
hydrogen bond along the molecular axis of the anion and negative frequency shifts have
been assigned to nonaxial interactions at the C-N bond of the anion.43 Our group has also
attributed a red-shifted vCN vibration, as compared that of the unperturbed anion, to a single
nonaxial interaction at the nitrogen of SCN'. Such an unusual interaction was first
identified by Gill and coworkers in neat methanol solutions of tetrabutylammonium
thiocyanate, where they attributed red-shifted bands to one and two hydrogen bonding
interactions to the C-N bond.43

Infrared studies of solutions of alkali metal thiocyanates in NMF have shown that
the cation and anion are not ion paired.160 Moreover, deconvolution of the vcN stretch for
the thiocyanate anion shows that it is in three vibrationally distinct environments. Solid
solvates of NaSCN with DMF and NMF were prepared in order to provide a pictorial

understanding of potential solution structures and to determine what effect ionic and

163

 

164

hydrogen bonding interactions have on the vibrational modes of SCN'. Crown ethers bind
to alkali metal cations. Solution studies have shown that crown ethers reduce ion pairing in
electrolyte solutions by isolating the cation from the anion.3°'55'5°"°' Likewise, these

ligands can be employed to minimize electrostatic interactions in crystal structures.

B. Results
1. Crystal structures for NaNCSXDMF), and NaNCS-(NMF),

The crystal structures for both bis(N,N-dimethylformamide)-sodium thiocyanate
and bis(N-methylformamide)-sodium thiocyanate reveal linear chains of sodium cations
bridged by the amide solvents and thiocyanate anions, as shown in Figures 53-56. The
positional parameters for each structure are collected in Tables 41-42. Each sodium cation
is coordinated by four amide molecules and two thiocyanate anions in a distorted octahedral

'62 The amide solvents are coordinated

orientation, similar to that seen for N aSCN-2H20.
to the sodium cations through the carbonyl oxygen with Na*'"O distances of 2.39-2.43 A
and 2.46-2.48 A for the NMF and DMF solvates, respectively. The intermolecular bond
lengths are collected in Tables 43 and 44. The thiocyanate anions are coordinated through
the nitrogen end of the anion with Na*"'N distances of 2.50-2.53 A and 2525-2526 A for
the NMF and DMF solvates, respectively. In the N MF solvate the thiocyanate anions are
in a trans orientation whereas the thiocyanate anions are in a cis configuration around the
sodium cation in the DMF solvate. The bond angles for bridging ligands (Na“'"L"'Na*) are
80—90°. The bond angles between across a sodium cation for trans ligands (L,"'Na*"'L2) are
about 160° for those in the DMF solvate and 170° for the NMF solvate. The greater
distortion from octahedral geometry in the DMF solvate is reflected in the longer

internuclear distance (Na“'"Na") of 3.335 1 0.005 A versus 3.235 1 0.003 A for the NMF

solvate.

165

J

A

K 0
$05 N13 (2,4? / l 0.21 C22 0
Cfi>~w $81" 0? 04,9 M

C240
C12 K
Na]

P394): (2 {Cb—(B

.\

xii

25$

I.

O D.

Figure 53. Ortep structure for NaNCS'(DMF)2

166

 

 

 

I 1 I 1
_’ ‘ ' O y.
B o . B o
o . o
o . e
0‘. o \. o
.\,. e .‘ o .’ e .
. ‘7 .. .‘o ../,l -
O ‘ . . .ql. 0
Q . o. .
e . . e .1

W
.I' 0
O
o“ <
. C
O
T “‘0’
6
0
0 o
O

 

 

. ‘0 O.
. O ‘. 9 i .0
0 ’ 0A 0 ‘

 

 

 

 

 

 

Figure 54. Packing diagram for NaNCS-(DMF)2

167

 

1-.
\V)
V
1
.\
0 v1 7 =
&’ \
Hi2
0’9 3 03013/
H13 N1 14b \_ H23
1
e— 0 . 021
H12 m 0“ c1 \‘l N23
H14c N,
\ ){1
C24
‘. Na]
H24d 112511 H246
I -
\U 1" \:)

Figure 55. Ortep plot for NaNCS-(NMF)2

168

 

Figure 56. Packing diagram for NaNCS'(NMF)2

169

Table 41. Positional parameters (A) and equivalent isotropic thermal parameters (A2),
with estimated standard deviations in parentheses, for bis(N-methylformamide)—sodium thiocyanate

 

 

atom x y z 3,,I

S1 0.3166(1) 0.0840(1) 0.55631(3) 395(4)
Nal 0.0024(1) -0. 1370(1) 0.75548(5) 225(4)
01 1 -0.2287(2) 0.1067(3) 0.731 1(1) 300(9)
021 0.1020(3) 0.1267(3) 0.83077(8) 308(9)
N1 0.1589(3) 0.1030(4) 0.6784(1) 2.9( 1)
N13 -O.4938(3) 0.1019(4) 0.7774(1) 3. 1( 1)
N23 0.1678(3) 0.0798(5) 0.9398(1) 3.5(1)
C1 0.2209(3) 0.0963(4) 0.6270(1) 2.4(1)
C12 -0.3910(4) 0.1052(4) 0.7272(2) 2.9(1)
C 14 -0.4270(5) 0. 1005(6) 0.8456(2) 4.2(2)
C22 0.1273(4) 0.1929(5) 0.8874(1) 4.2(2)
C24 0. 1904(6) -0. 1427(6) 0.9382(2) 4.2(2)
H12 -0.455(4) 0.108(4) 0.684(1) 3.8(7)
H 13 -0.603(4) 0.104(5) 0.771(2) 4.7(8)
H 14a -0.479(5) 0.218(7) 0.870(2) 7(1)
H14b -O.307(5) 0.107(5) 0.850(2) 5.7(9)
H 14c -O.462(5) -0.019(7) 0.868(2) 7(1)
H22 0.121(4) 0.342(5) 0.897(1) 3.5(7)
H23 0.188(4) 0.140(5) 0.976(2) 4.0(7)
H24a 0.l7(1) -0.19(l) 0.892(3) 3(1)
H24b 0.12(1) -0.22( 1) 0.971(4) 6( 1)
H24c 0.31(1) -0.17(1) 0.949(4) 5(1)
H24d 0.08( 1) ~0.20( 1) 0.927(5) 5(1)
H24c 0.23(1) -0.18( 1) 0.983(4) 3(1)
H24f 029(1) -0. 18(1) 0.905(5) 6(1)

 

 

170

Table 42. Positional parameters (A) and equivalent isotropic temperature factors (A2), with estimated
standard deviations in parentheses, for bis(N,N-dimethylfonnamide)-sodium thiocyanate

 

 

 

atom x y z B(eq)
S 1 0.8444(2) 0.0823(1) 1/4 3.9( 1)
Na] 0.5073(3) 1/4 1/2 3.2(1)
.01 1 0.3224(5) 0.2761(3) 3/4 3.9(3)
O21 0.5478(5) 0.151 1(3) 3/4 3.2(2)
N 1 0.6536(7) 0.1789(4) 1/4 4.5(4)
N13 0.1279(5) 0.2049(3) 3/4 2.2(1)
N23 0.6559(6) 0.0407(3) 0.7200 2.4(1)
C 1 3/4 0.1387(4) 1/4 3.4(4)
C12 0.1947(7) 0.2690(3) 3/4 2.4( 1)
C 14 0.198(1) 0.1327(5) 3/4 4.0(2)
C15 -0.0229(9) 0.2021(5) 3/4 3.6(2)
C22 0.6429(8) 0. 1059(4) 0.807(2) 2.6(2)
C24 0.772(1) -0.0093(5) 3/4 3.9(2)
C25 0.558(1) 0.0139(7) 0.568(2) 3.5(2)
H12 0.135(5) 0.313(3) 3/4 1(1)
H14d 0.274(8) 0. 136(4) 3/4 3(2)
H 14c 0.170(8) 0.1 1 1(4) 0.60(2) 14(2)
H15d -0.048(8) 0.256(5) 3/4 6(2)
H15e -0.048(5) 0.172(3) 0.639(9) 5(1)
H22 0.720(8) 0.122(4) 089(1) 2(2)
H24a 0.729(9) -0.062(5) 3/4 5(2)
H24b 0.835(6) 0.005(3) 0.66( 1) 7(2)
H25d 0.62(l) 0.000(5) 0.46(2) 4(2)
H25e 0.55( 1) -0.038(7) 0.61 (2) 6(2)
H25f 0.50(1) 0.041(5) 053(2) 2(2)

 

171

Table 43. Intermolecular distances (A), with estimated standard deviations in parentheses,
for bis(N-methylformamide)—sodium thiocyanate

 

 

atom atom distance atom atom distance
8 1 C 1 1.637(3) N23 C22 1.312(4)
Nal Nal 3.235(3) N23 C24 1.447(5)
Nal 01 1 2.392(2) N23 H23 084(3)
Na] 01 1 2.392(2) C 12 H12 097(3)
Nal 021 2.386(2) C14 H14a 1.00(4)
Nal 021 2.425(2) C 14 H 14b 092(4)
Nal N1 2.534(3) C 14 H14c 094(3)
Nal N 1 2.502(3) C22 H22 098(3)
01 1 C12 1.232(3) C24 H24a 099(6)
021 C22 1.232(3) C24 H24b 1.00(7)
N1 C 1 1.162(3) C24 H24c 094(7)
N13 C 12 1.311(4) C24 H24d 095(8)
N13 C14 1.448(5) C24 H24c 097(7)
N 13 H13 083(3) C24 H24f l .05 (8)

 

Table 44. Intermolecular distances (A), with estimated standard deviations in parentheses,
for bis(N,N—dimethylformamide)-sodium thiocyanate

 

 

atom atom distance atom atom distance
S1 C1 1.640(5) C 12 H12 098(5)
Nal Nal 3.335(3) C14 H14d 074(8)
Na] 0]] 2.483(4) C14 H14c 1.1(1)
Nal O21 2.462(4) C15 H15d 099(9)
Nal N 1 2.526(5) C 15 H 15e 095(5)
01 1 C12 1.236(8) C22 H22 095(8)
021 C22 1.279(9) C24 H24a 1.02(8)
N1 C 1 1.174(8) C24 H24b 088(6)
N13 C12 1.314(7) C24 H25d 1.0(1)
N13 C14 l.45(1) C24 H25e 1.0(1)
N23 C22 1.308(9) C24 H25f 0.8( 1)
N23 C24 1.45(1)
N 23 C25 1.46(1)

 

a. Amide Structure

172

In both solvates, one amide molecule is perpendicular to the N a“"'N a” axis. In the
case of the DMF solvate, this DMF molecule is fixed perpendicular to the Na"'“Na+ axis in a
mirror plane. The second DMF molecule is distorted out of the mirror plane. No such
conditions are exists in the NMF solvate, yet one N MF molecule is roughly perpendicular
and the other roughly parallel to the Na"'"Na+ axis. The carbonyl bond lengths, 1.232 (3) A
in the NMF solvate and 1.238 (8) A and 1.279 (9) A for the DMF solvate, is slightly
longer in the DMF solvate. The bond distances between the carbonyl carbon and the
nitrogen are in much closer agreement of about 1.31 A. The alkyl carbon-nitrogen bond
lengths are ~1.45 A, and the oxygen-carbon-nitrogen bond angles are ~125°. Each of the

DMF and NMF molecules, in these solvates, are nearly planar.

b. Thiocyanate

The thiocyanate anion displays small differences between these two solvates. The
anion is linear in the DMF solvate (ZSCN' 179.8 1 06°) and slightly distorted in the NMF
solvate (ASCN‘ 177.5 1 02°). The C-N bond is slightly longer in the DMF solvate
( 1.174 1 0.008 A) than in the NMF solvate (1.162 1 0.003 A). There is no substantial
difference between the C-S bond lengths for the NMF solvate (1.637 1 0.003 A) and the
DMF solvate (1.640 1 0.005 A).

c. Hydrogen bonding
Owing to the acidic proton in NMF, a hydrogen bonding interaction with this
solvate is possible. A search of intermolecular distances out to 3.6 A reveals a potential
hydrogen bond to the sulfur end of SCN' from NMF. The hydrogen bond, S l"'H23, is
2.58 1 0.03 A lOng with a heavy atom bond (Sl'"N23) distance of 3.364 1 0.003 A. This
interaction is distorted from linearity and has a bond angle of 157.5°, shown in Figure 54.
Interestingly, this interaction links the sodium chains together. Such an interaction is not

observed in the DMF solvate, nor would it be expected due to the lack of an acidic

173

hydrogen. This type of non-axial interaction to the sulfur end is observed for hydrates of

various alkali metal thiocyanates.

2. Summary of NaSCN 0( DMF )2 and NaSCNO(NMF)2 crystal structures
As with other solvates, the neutral solvent molecules form part of the coordination
sphere surrounding the cation. Not surprisingly, the thiocyanate anion coordinates to the
sodium cation through the nitrogen atom. The nonaxial cation interactions, in these
solvates, lengthen the C-N bond and shorten the GS compared to SCN' in crystals having
no electrostatic perturbation. These structures may represent the structure of solvated
N aN CS dimers in NMF and DMF at extremely high electrolyte concentrations, where

NaNCS will dimerize.

3. Crystal structures for Na(12C4)2NCSOH0Me and Na(12C4),NCS

Previous investigations of 12—crown-4 have shown that the cavity is too small for
Na” cations, yet solution and crystal structure studies have shown that this ligand can easily
bind to Na“ with 1:1 and 1:2 (cationzligand) stiochiometry.'°3"°5 For the Na(l2C4),SCN
and the Na( 12C4)4SCN0HOMe crystals grown in this work, each the sodium cation is
surrounded by two 12-crown-4 ligands (Figures 57-60). The positional parameters for
each structure are collected in Tables 45 and 46. In each case, the thiocyanate anion is not
within van der Waal contact of the cation. The Na-O distances between the ligands and the
Na+ cation are comparable, with an average value of 2.50 (1) A in the solvated complex and
2.50 (3) A in the Na( 12C4),SCN complex (Tables 47 and 48). The ligands surrounding
the Na+ cation in the solvate complex are disordered shown in Figure 59. There are two
conformations for the ligands, that are at 50% occupancy which are ~45° out of phase from

each other. The l2-crown-4 ligands in the Na(12C4),,SCN complex are not disordered,

174

 

Figure S7. Ortep plot for Na(12C4)2NCS

175

 

 

 

 

 

Figure 58. Packing diagram for Na(12C)4NCS

176

 

Figure 59. Ortep plot for Na(l2C4)2NCS-HOMe

177

 

 

 

 

 

 

 

Figure 60. Packing diagram for Na(12C4)2NCS°HOMe

Table 45. Atomic coordinates (10‘) and equivalent isotropic displacement parameters (A2 x 10“)

178

for the Na(12C4)2SCN complex.

 

A tom x y z U(eq)‘ Occ.
8(1) 3486(3) 1228(3) 4913(2) 88(1) 1
8(2) 6448(3) 5939(3) 136(2) 97(2) 1
Na( 1) 8768(3) 6620(3) 2825(2) 49(1) 1
Na(2) 1274(3) 1620(3) 2186(2) 43(1) 1
0(101) 8139(5) 7140(6) 3889(4) 62(2) 1
0(104) 8639(5) 5327(5) 3708(4) 59(2) 1
0(107) 10178(5) 5968(5) 3256(4) 52(2) 1
0(1 10) 9665(5) 7815(5) 3439(4) 52(2) 1
0(201) 8778(5) 5373(5) 1927(4) 49(2) 1
0(204) 7303(4) 6157(5) 2425(3) 48(2) 1
0(207) 7943(5) 7970(5) 2252(3) 52(2) 1
0(210) 9410(5) 7164(5) 1771(4) 52(2) 1
0(301) 1842(5) 1 1 15(6) 1058(4) 56(2) 1
0(304) 1355(5) 2940(5) 1374(3) 51(2) 1
(X307) ~16 1 (5) 2283(5) 1873(4) 53(2) 1
(X310) 330(5) 461(5) 1566(3) 50(2) 1
0(401) 1262(5) 2788(5) 3157(4) 51(2) 1
0(404) 2732(4) 2083(5) 2596(3) 46(2) 1
(X407) 21 18(5) 272(5) 2669(3) 48(2) 1
0(410) 640(4) 980(5) 3212(3) 48(2) 1
N( 1) 4040(8) 2986(9) 4618(6) 88(4) 1
N(2) 6195(9) 78 1 8( 10) 378(6) 91(4) 1
C( 1) 3791(8) 2257(9) 4759(6) 54(3) 1
C(2) 6300(8) 7046(12) 269(6) 67(4) 1
C(102) 8145(12) 6521(12) 4454(8) 73(5) 1
C(103) 7997( 10) 5607(12) 4165(8) 72(4) 1
C(105) 9419( 10) 4995(12) 4021(8) 70(4) 1
C(106) 1009100) 5061(11) 3513(9) 67(4) 1
C(108) 10678(9) 6570(9) 3727(7) 57(4) 1
C(109) 10550(8) 7529( 10) 3465(8) 59(4) 1
C(1 1 1) 9404(9) 81 1 1(1 1) 4090(7) 67(4) 1
C(1 12) 8455(9) 8077(10) 4044(6) 66(4) 1
C(202) 7967(8) 5136(1 1) 1622(8) 57(4) 1
C(203) 7329( 10) 5248( 10) 2147(8) 61(4) 1
C(205) 6841(9) 6805(8) 1998(7) 48(3) 1
C(206) 7043(8) 7757(9) 2245(6) 50(3) 1
C(208) 8208(9) 8222(9) 1593(6) 57(4) 1
C(209) 9174( 10) 8107(10) 1643(8) 60(4) 1
C(2“) 9319(12) 6562(10) 1196(8) 63(4) 1
C(212) 9391(9) 5588(10) 1435(8) 57(4) 1
C(302) 1778(10) 1809(10) 541(7) 63(4) 1
C(303) 1936( 10) 2732(1 1) 864(8) 62(4) 1
C(305) 559(9) 3273(1 1) 1097(8) 63(4) 1
C(306) -69(9) 3205(9) 1647(7) 58(4) 1
C(308) -695(9) 1714(10) 1415(7) 54(4) 1
C(309) ~527(8) 749(10) 1621(7) 57(4) 1
C(31 1 ) 525(9) 245(9) 874(6) 51(3) 1
C(312) 1476( 10) 254( 10) 854(7) 63(4) 1
C(402) 2090(8) 2968(12) 3498(7) 62(4) 1
C(403) 2720( 10) 293 1 (10) 2959(7) 63(4) 1
C(405) 3240( 10) 1 385( 10) 2964(8) 62(4) 1
C(406) 2991(9) 464(10) 2663(7) 60(4) 1
C(408) 1861(9) -42(1 1) 3325(7) 58(4) 1
C(409) 923( 10) 56( 10) 3326(8) 57(4) 1
C(4“) 691(10) 1541(10) 3820(7) 60(4) 1
C(412) 2537(9) 3608(7) 642(9) 54(3) 1

 

'U(eq) is defined as one third of the trace of the orthogonalized U,,- tensor

179

Table 46. Atomic coordinates (10‘) and equivalent isotropic displacement
parameters (A2 x 103), with estimated standard deviations in parenthesis,
for the Na(12C4)2SCN0MeOH complex.

 

 

x y z U(eq) Occ.
Na( 1) 0 5000 5000 57(1) 1
8(2) 1283(3) 2500 3003(4) 120(1) 1
C(3) 3077(12) 2500 2863(12) 92(2) 1
N(4) 4409(12) 2500 2750(14) 140(3) 1
GOOD -7624(7) 5405(2) 299(6) 68(2) 0.50
C(102) 721 1 (10) 5428(5) -2292(10) 84(4) 0.50
C(103) 5459(11) 5741(5) -3487(14) 78(6) 0.50
O(104) 4070(6) 5395(2) 3324(7) -64(2) 0.50
C(105) 2400(8) 5723(4) 4004(9) -76(3) 0.50
C(106) 2418(12) 621 1(3) -2596(1 1) 75(4) 0.50
O(107) 2802(6) 5906(2) -823(6) 65(2) 0.50
C(108) 3349(9) 6336(4) 790(1 1) 86(3) 0.50
C(109) 5274(1 1) 6500(3) 1633(14) 76(4) 0.50
O( 1 10) 6324(6) 5926(2) 2168(7) 66(2) 0.50
C(1 l l) 8145(8) 5987(4) 2539(9) 65(3) 0.50
C( 1 12) 8305(13) 5993(4) 728(13) 86(5) 0.50
O(201) 6269(6) 5325(2) -2246(6) 57(1) 0.50
C(202) 5159( 10) 5783(4) -3661(12) 56(4) 0.50
C(203) 3254(9) 5631(4) -4295(8) 67(3) 0.50
O(204) 2821(7 5659(2) -2725(7) 63(2) 0.50
C(205) 2566( 1 l) 6297(3) -221 2(10) 64(4) 0.50
C(206) 2692(8) 6286(4) -258(10) 76(3) 0.50
O(207) 4403(6) 6060(2) 1 168(6) 68(2) 0.50
C(208) 581 1(10) 6526(3) 1821 ( 12) 68(3) 0.50
C(209) 7545(9) 6194(4) 2855(8) 66(2) 0.50
O(210) 7797(7) 5720(2) 1652(7) 63(2) 0.50
C(21 l) 8378(1 1) 5994(3) 364(1 1) 60(4) 0.50
C(212) 8102(8) 5523(4) -1 179(1 1) 70(3) 0.50
O(301) 7563(9) 2867(5) 6386(9) 185(5) 0.50
C(302) 8853(14) 2500 6936(14) 136(3) 1

 

 

U(eq) is defined as one third of the trace of the orthogonalized Uij tensor

180

Figure 47. Selected bond lengths, with estimated standard deviations in parenthesis, for the

 

 

Na(12C4)2SCN complex
Atoms Bond length

N(1)-C(l) 1.163(14)
N(2)-C(2) 1 . 15(2)
S(l)—C(l) 1.598(14)
S(2)-C(2) 1 .64(2)
Na( 1 )-O(204) 2.469(8)
Na(1)-O(1 10) 2.476(8)
Na(1)-O(101) 2.479(9)
Na(1)-O(210) 2.484(8)
Na(1)-O(107) 2.496(8)
Na(l )-O(201) 2.51 1(8)
Na(1)-O(207) 2.551(8)
Na(1)-O(104) 2.556(8)
Na(2)-O(404) 2.463(8)
Na(2)-O(410) 2.480(8)
Na(2)-O( 307) 2.486(8)
Na(2)—O(304) 2.486(8)
N a(2)-O(3 10) 2.487(8)
Na(2)-O(407) 2.499(8)
Na(2)-O(401) 2.533(8)
Na(2)-O(301) 2.541(8)
O(101)-C(102) l.42(2)
O(101)-C(112) 1.466(14)
O(104)-C(105) 1.41(2)
O(104)-C(103) l.45(2)
O(107)-C(106) 1.413(14)
O(107)—C(108) 1.451(13)
O(110)—C(111) 1.428( 13)
O(110)-C(109) 1.451(13)
O(201)-C(202) 1.408(13)
O(201)-C(212) 1.445(14)
O(204)-C(205) 1.414(12)
O(204)-C(203) 1.421(14)
O(207)-C(208) 1.428(13)
O(207)-C(206) 1 .449(13)
O(210)-C(21 1) l.42(2)
O(210)-C(209) 1.429(14)
O(301)-C(312) 1.414(14)
O(301)-C(302) 1 .419(14)
O(304)-C(305) 1.410(13)
O(304)-C(303) l.43(2)
O(307)-C(306) 1.414(14)
O(307)-C(308) 1.435(13)
O(310)-C(309) 1.424(13)
O(310)-C(3l 1) 1.439(13)
O(401)-C(4l2) 1.409(13)
O(401 )-C(402) 1 .440(14)
O(404)-C(403) 1 .414(14)
O(404)-C(405) 1 .443( 14)
O(407)-C(406) 1 .404( 13)
O(407)-C(408) 1 .442( 13)
O(410)-C(409) 1 .420( 14)
O(410)-C(41 l) 1.430(14)

 

181

Figure 48. Selected bond lengths, with estimated standard deviations in parenthesis,
for the Na( 12C4)2SCN complex

 

 

Bond Length (4)
Na(1)-O(1 10)#1 2.460(5)
Na(1)—O(1 10) 2.460(5)
Na(1)-O(101) 2.475(5)
Na(l)—O(101)#1 2.475(5)
Na(1)-O(104)#l 2.483(5)
Na(1)-O(104) 2.483(5)
Na(l)-O(204)#l 2.485(5)
Na(1)-O(204) 2.485(5)
Na(1)-O(107) 2.503(5)
Na(1)-O(107)#l 2.503(5)
Na(1)-O(207)#1 2.513(5)
Na(1)-O(207) 2.513(5)
S(2)—C(3) 1.565(9)
C(3)—N(4) 1.166(9)
O(101)-C(1 12) 1.426(6)
O(101)-C(102) 1.432(6)
C(102)-C(103) 1.484(8)
C(103)-O(104) 1.427(6)
O(104)-C005) 1.427(6)
C(105)-C(106) 1.489(8)
C(106)-O(107) 1.419(6)
O(107)-C(108) 1.436(6)
C(108)-C(109) 1.483(8)
C(109)-O(l 10) 1.424(6)
O(110)-C(1 1 1) 1.430(6)
C(1 1 1)-C(1 12) 1.483(8)
O(201)-C(202) 1.430(6)
O(201)—C(212) 1.437(6)
C(202)-C(203) 1.480(8)
C(203)-O(204) 1.433(6)
O(204)-C(205) 1.424(6)
C(205)-C(206) 1.482(8)
C(206)-O(207) 1.436(6)
O(207)-C(208) 1.429(6)
C(208)-C(209) 1.474(7)
C(209)-O(210) 1.441(6)
O(210)-C(21 1) 1.424(6)
C(2] 1)-C(212) 1.484(8)
O(301)—C(302) 1.230(10)
O(301)<H(301) 0.950(2)
C(302)-O(301)#2 1.230(10)

 

182

but they have two distinct configurations in this complex. The coordination sphere
surrounding the Na+ cation is a distorted octahedral arrangement with O-Na*-O angles of
69.5 (5)°, 84.8(5)°, 109.7 (1.5)°, and 144.6 (l.4)° that correspond to adjacent oxygens
belonging to the same ligand, cis oxygens across two rings, opposite oxygen atoms on the
same ring, and trans oxygens across two rings, respectively. The C-O-Na+ angles fall into
two categories for Na(12C4)2SCN, with angles of 109.1 (11)° and 115.2 (14)°. The
analogous angles in the methanol solvate are closer at 109.9 (12)° and 113.3 (15)°.

In the ligands of these complexes, the mean aliphatic C-C bond lengths are
comparable with a distance of 1.488 (5) A for the solvate complex and 1.492 (20) A for the
unsolvated complex. Likewise, the CO bond lengths are similar with a distance of
1.448 (5) A in the solvated complex and 1.428 (20) A in the other. The C-C-O bond
angles also fall into two categories of 108.0 (12)° and 112.8 (9)° for the Na(12C4)2SCN.
These bond lengths are in good agreement with those seen in other crown complexes.“ '67

The thiocyanate anion is in two different environments in the Na(12C4)ZSCN
complex. The CS and C-N bondlengths are slightly different (1.598 (14) & 1.64 (2) and
1.163 ( l4) & 1.15 (2), respectively), although the large uncertainties for the C-S bond
make the comparison difficult. Each of the anions is almost linear, with bond angles of
176.2 (13) and 178.4 (14). Neither thiocyanate anion points toward the Na” cation,
indicating that the cation is screened from the anion; the closet approach between the cation
and anion is approximately 6 A. In the methanol solvate, the thiocyanate anion is also well
isolated from the sodium cation. Hydrogen bonding by methanol slightly lengthens the
C-N bond to 1.166 (9) A with respect to the unperturbed anion in Na(12C4)2SCN, with a
slight reduction for the GS bond Hydrogen bonding by methanol forms a AONC angle of
113°. The anion in the solvate complex is also nearly linear at 179.0 (3)°. In each
structure, the thiocyanate anion is greater than 5.7 A from any sodium cation and the anion

is not orientated toward any cation through either the nitrogen or sulfur end.

183

a. Infrared
The infrared spectra of these crystals were obtained. The nonaxial hydrogen

bonding interaction in the methanol solvate reduces the frequency of the Vcn vibration from
2045 cm‘1 to 2032 cm“. This reduction in frequency is consistent with the lengthening of
the C-N bond observed in the crystal structures. In the analysis of solutions, such negative
frequency shifts found in alkali metal thiocyanate dimers have been attributed to the
weakening of the ON bond, resulting from each thiocyanate anion being bridged by two
lithium cations.7°'77 A crystal structure of a lithium thiocyanate that has a VCN vibration <

2040 cm‘l also contains nonaxial interactions with the thiocyanate anion.97

b. Ab initio calculations
Previous theoretical investigations by our group have shown that the potential

energy distribution for the vCN mode is approximately 90% C-N stretch. Ab initio
calculations that involve ionic or hydrogen bonding interactions along the molecular axis
predict an increase in the frequency of the vCN mode. When a hydrogen bond interaction by
hydrogen fluoride or water is fixed at 113°, the ON bond is lengthened and the GS bond
is shortened as compared to the free anion or a comparable hydrogen bonded complex. In
such nonaxial complexes, the frequency of the vcN mode is reduced. Therefore, it is likely
that the lengthening of the C-N bond and the red shift of the vCN vibration are the result of
the hydrogen bond interaction, and are not due to a difference in the packing forces

between these two crystal structures.

4. Summary of Na(12C4)2NCS'Me0H and Na(12C4)2NCS
The crystal structures for Na(12C4)2SCN and Na(l2C4),SCN0HOMe clearly
show the effect of hydrogen bonding upon the SCN' anion. A nonaxial hydrogen bond to
the nitrogen of SCN’ lengthens the C-N bond. This structural change is manifest by a

reduction in the frequency of the Vcn stretching vibration. These crystal structures provide

184

an important connection between our theoretical calculations and our solution studies.

Theoretical calculations of model SCN""HOMe complexes predict similar trends; see Table

49. N onaxial interactions to the SCN' anion lengthen the C-N bond and shorten the CS

bond, with a concomitant decrease in the frequency of the vcN vibration and an increase in

the frequency of the vCS vibration.

Table 49. Ab initio calculations for axial and nonaxial hydrogen bonds with SCN'

 

 

System dCN (A) dcs (A) VCN (cm") Vcs (cm-l)
SCN' 1.208 1.671 1979 745
SCN""HF 1.202 1.653 2047 805
SCN'~
' HF 1.213 1.651 1966 776
SCN""HOH l .206 1.660 2010 777
SCN'
1.212 1.651 1970 778

HOH

 

 

Chapter VII

Conclusions and Future Work

 

A. Conclusions

The solvation of the OCN', SCN’ and SeCN' anions by MeOH, FA and NMF has
been studied primarily by NMR and infrared spectroscopies. From these experimental
measurements, we have been able to identify the structure of the primary solvation sphere
surrounding these linear triatomic anions in these protic solvents. Each of the anions has
the same charge, yet there is a great difference in the extent of solvation and the structure of
the solvation complexes. The OCN' anion is the most solvated followed by SCN‘ and then
SeCN'. The extent of solvation correlates with the electrostatic potential calculated for these
anions. Our thermodynamic studies imply that the entropy is a significant component of
the solvation free energy in the primary solvation sphere. We have assumed that the
measured entropy of formation for the 1:1 complexes is a sum of entropy loss
accompanying the complex formation and entropy gain for NM molecules liberated from
the solvation sphere surrounding these anions, in NM solutions. Thus, a small reduction
in the electrostatic potential of these anions may result in a large change in the strength of
the anion-solvent complex, since both the enthalpy of formation will be reduced and the
entropy of formation will become more negative.

The combination of the infrared and NMR techniques not only resulted in good
agreement between the methods, but was indispensable in resolving the band structure for
the stretching modes of the SCN' and SeCN’ anions. The vibrational band envelopes for
these two chalcocyanates in protic solvents is extremely convoluted and difficult to sort.
Indeed, we were able to fit these spectral envelopes with more than one model, but only
one of these models gave calculated association constants in good agreement with both
stretching modes, the l°N chemical shifts, and MN line widths. Ab initio calculations of
anion-solvent and ion pair complexes not only provided a firm foundation for the
vibrational assignments, but yielded insight into the nature of solvation via the electrostatic

potential. In the case of the cyanate anion, the strength of the hydrogen bond donor

186

187

dictates whether a second hydrogen bond will occur on the nitrogen end or the oxygen end.
Finally, the crystal structures for Na(12C4)2SCN and Na(12C4)2SCN-MeOH provided us

with additional confidence for the nonaxial hydrogen bonding interaction with SCN'.

B. Future Work
As with most research projects, we have raised a more questions than we have

answered. Therefore, there are a number projects that can be easily identified.

1. We have shown that the distribution of the negative charge is important in
determining the strength of the complex. It would be interesting to extend these
studies to CN‘ and N31

2. This investigation has focused on hydrogen bonding interactions, since this
intermolecular force is sufficiently strong to perturb vibrational modes and
chemical resonances. Although ion-dipole interactions are substantially weaker,
it might be possible to observe specific anion solvent interactions with a very
polar solvent such as ethylene or propylene carbonate. The latter solvent has a
very high acceptor number.

3. The use of cryptands might provide an additional synthetic route to prepare
isolated anions for solvate synthesis.

4. Our ab initio predicted showed that the perturbation by metal cations should lead
to larger changes in the vibrational frequencies than are observed
experimentally. This discrepancy could result from the lack of incorporating
solvation or the absence of dielectric medium background. As computing
power increases, these possibilities can be examined.

5. The dCN bond lengths in OCN', SCN' and SeCN' did not correlate as well with
the FCN (force constants) as did dCX with ch. A study of the potential energy

curves for the ON and C-X bonds as a function of the electrostatic perturbation

 

188

would should show whether this is a result of a change in the curvature of the
well or that there is poor correlation between bond length and bond strength.

. We have shown how the 11: and 211: orbitals are important for the electrostatic
perturbation caused by ionic and hydrogen bonding interactions. Another
perspective would be to calculate the density difference between the complexes
and the respective atoms to see how the bond length changes correlate with

changes in electron density.

 

List of References

References

' Barthel, J., Gores, J., Schmeer, G., Wachter, R., Topic in Curr. Chem, 1983, III, 33.

2 van der Waals, J.D., W, University of Lieden, 1873.

3 Huyskens. P0 Luck. W.. ZcegerS-Huyskens 1.me
WM, Springer-Verlag, Berlin, 1991.

4 Lide, D., Handlxxzk Qf Chemistry and Physics, 21" Egition, CRC Press, Boca Raton, 1991.
5 Popovych, D., Nonaguegus Solution Chemism, Wiley, New York, 1981.
6 Novak, A., Structure and Bonding, 1974, 18, 177.

7 Riddick. J., Wu Wiley. New York.
1986.

8 Tanaka, Y., Ohtomo, N., Arakawa, K., Bull. Chem. Soc. Jpn., 1985, 58, 270.

°Ohtaki, H., Itoh, S., Rode, B., Bull. Chem. Soc. Jpn., 1983, 56, 2117.

”’ Ohtaki, H., and Itoh, S., Z. Naturfrosch., 1985, 40A, 1351.

” Ohtaki, H., Itoh, S., Rode, B., Bull. Chem. Soc. Jpn., 1986, 59, 271

'2 Radnai, T., Itoh, 8., Yamaguchi, T., lshiguro, S., Rode, B., Bull. Chem. Soc. Jpn, 1983, 56, 3406.
'3 Radnai, T., Itoh, S., Ohtaki, H., Bull. Chem. Soc. Jpn., 1988, 61, 3845.

”Ohtaki, H., Yamatera, H., W, 1992, Elsevier, Amsterdam.

‘5 Gutmann, V., W; Plenum, New York, 1980.
'° Tanaka, N., Ohtaki, H., Tamamushi, R., WW5, Elsevier, 1983, 203.

'7 Justice, J.C., J. Chem. Phys., 1968, 65, 353.

'8 Hohn, B., Olander, J., Day, M., J. Phys. Chem, 1969, 73, 3880.

‘9 Greenberg, M., Bodner, R., Popov, A., J. Phys. Chem. 1973, 77, 2449.

2° Chuang, H., Soong, L., Leroi, G., Popov, A., J. Soln. Chem, 1989, I8, 759.

2' Popov, A., Pure & Appl. Chem, 1979, 51, 101.

22 Keil, R., Johnson, D., Fryling, M., O’Brien, J., Inorganic. Chem, 1989, I8, 759.

190

191

2" Woepper, J., Popov, A., J. Am. Chem. Soc., 1969, 91, 4352.

2“ Wong, M., McKinney, J., Popov, A., J. Phys. Chem, 1971, 75, 56.

25 Lassigne, C., Baine, P., J. Phys. Chem, 1971, 75, 3188.

2° Mazitov, R., Muller, K., Hertz, H., Z. Phys. Chem. N. F., 1984, 64.

27 Ahlrich, R., Bohm, H., Hertz, H., Muller, K., Z. Phys. Chem N. F., 1984, 73.
28 Finter, C., Hertz, H., Z Phys. Chem. N.F., 1986, 73.

2° Laidler, K., Pure & Appl. Chem, 1990, 62, 2221

3° Kanno, H., J. Raman Spec, 1987, 18, 301.

3‘ Kanno, H., J. Phys. Chem., 1988, 92, 4232.

’2 Baum, R., Popov, A., J. Soln. Chem, 1975, 4, 441.

33 Kleeberg, H., Heinje, G., Luck, W., J. Phys. Chem, 1986, 90, 4427.

3“ Paoli, D., Lucon, M., Chabanel, M., Spectrochimica Acta, 1977, 34A, 1087.

’5 Perelygin, 1., Mikhailov, G., Russ. J. Phys. Chem, 1990, 64, 1060.

3“ Koda, S., Goto, J., Chikusa, T., Nomura, H., J. Soln. Chem, 1989, 93, 4959.
37Bacelon, P., Corset, J., de Loze, C., J. Soln. Chem, 1983, I, 1.

33 Bacelon, P., Corset, J., de Loze, C., J. Soln. Chem, 1983, I, 13.

3° Bacelon, P., Corset, J., de Loze, C., J. Soln. Chem, 1983, I , 23.

4° Owrutsky, J., Raftery, D., Hochstrasser, R., Annu. Rev. Chem, 1994, 45, 519
4‘ Owrutsky, M., Sarisky, M., Culver, J., Yodh, A., Hochstrasser, R., J. Chem. Phys., 1993, 98, 5499.
‘2 Kameda, Y., Takahashi, R., Usuki, T., Uemura, 0., Bull. Chem Soc. Jpn., 1994, 67, 956.
‘3 Bachelin, M., Gans, P., Gill, J., J. Chem Soc. Faraday Trans, 1992, 88, 3327.
4‘ Kuhn, L., Bowman, R., Spectra. Acta, 1967, 23A, 189.

‘5 Bencheikh, A., We, Nantes, 1987.

4° Soong, L., Ph.D. Thesis, MSU, 1988.

‘7 Griffiths, T., Pugh, D., Coord. Chem. Rev., 1979, 29, 129.

‘8 Dogonadze, R., Kalman, B., Kornyshev, A., Ulstrup, J., W, Elsevier,
Amsterdam, 1986, Part B.

°°Laidler, K.J., Pure & Appl. Chem, 1990, 43, 275.

5° Irish, D., Tang, S., Talts, H., Petrucci, S., J. Chem. Phys., 1979, 83, 3268.

 

192

5' Saar, D., Petrucci, S., J. Phys. Chem, 1986, 90, 3326.

52 Xu, M., Eyring, E., Petrucci, S., J. Phys. Chem, 1986, 90, 6125.

53 Chabanel, M., Lucon, and Paoli, D., J. Phys. Chem, 1981, 85, 1058.
5‘ Ostwald, W., Z. Phys. Chem, 1888, 2, 270.

5‘ Tawarah, K., Mizyed, S., J. Soln. Chem, 1989, I8, 387.

5° Tawarah, K., Mizyed, S., J. Include. Phen., 1988, 6, 583.

i .n , Marcel Dekker,

 

’7 Popov, A., Hallenga, K., 11 m - It. 51011
New York, 1991.

58 Delpuech, J., Analyst, 1992, 117, 267.

59 Mason, J., W, Plenum, New York, 1987.

6" Martin, M., Martin, G., Delpeuch, J., W, Heyden and Sons, London, 1980.
6' Deverall, C., Richards, R., Mol. Phys.,l966, 10, 551.

62 Deverall, C., Richards, R.E., Mal. Phys., 1969, I6, 421.

63 Deverall, C., Prog. Nucl. Magn. Reson. Spectrosc., 1969, 4, 235.

6‘ Bloor, E., Kidd, R., Can. J. Chem, 1968, 46, 3425.

65 Bloor, E.G., Kidd, R.G., Can. J. Chem, 1972, 50, 3962

6" Mamantov. G.. Popov, A.. We VCH. New York.
1994.

‘57 Gutmann, V., Wychera, B., Inorg. Nuclear Chem. Lett., 1966, 2, 257.
“8 Mayer, U., Gutmann, Gerger, W., Monatsh. Chem, 1975, 106, 1235.
6° Chabanel, M., Menard, C., Guihenauf, G., Comptes Rendus, 1971, 272C, 253

7°Gardiner,D.J., i ain _ c' o ' _o ‘ o -.01 1J9 1.x.__'_-- '1 01 1‘: En,
Academic Press: New York, 1977, chapter 9.

7‘ Coetzee, J.F., and Ritchie, C.D., W115, Marcel Dekker, Inc.: NY,1975, Vol 11,
p.271

72 Hertz, H., Muller, K., Chemica Scripta, 1989, 29, 277.

73 Hertz, H., Muller, K., Chemica Scripta, 1989, 29, 291.

7‘ Chabanel, M, Z., J. Phys. Chem, 1984, 88, 1441.

7’ Chabanel, M., Rannou, J., J. Chimie Phys., 1980, 77, 201.
7" Rannou, J., Ph.D. Dissertation, L’universite de Nantes, 1979.

77 Rannou, J., Masse, G., Chabannel, M., C. R. Acad Sc. Paris, 1978, 93.

193

78 Devore, T., J. Mol. Struct., 1987, I62, 273.

7° Perelygin, I., Osipov, I., Gryaznov, 8., Russ. J. Phys. Chem, 1985, 59, 1462.

8" Paoli, D., Lucon, M., Chabanel, M., Spect. Acta., 1977, 34A, 1087.

8' Paoli, D., and Chabanel, M., J. Chem Res., 1991, 360

“2 Chabanel, M., Bencheikh, A., Puchalska, D., J. Chem. Soc. Dalton Trans, 1989, 2193.

83 Golub, A., Kohler, H., Skopenko, V., W, Elsevier, Amsterdam, 1986.
8‘ Bacelon, P., Corset, J ., de Loze, C., J. Soln. Chem, 1980, 9, 129.

8’ Chabanel, M., Pure & Appl. Chem, 1990, 62, 35.

8" Kharitonov, Y., Tsintsadze, G., Porai-Koshits, M., Russ. J. Inorg. Chem, 1965, 10, 427.

”7 Karitonov, Y., Tsintsadze, G., Porai-Koshits, M., Dok. Akad. N. SSSR, 1964, 160, 1351

”8 Nakamoto, K., . , Wiley, New York, 1978.

 

“9 Squalli, O., Martins Costa, M., Cartier, A., Chabanel, M., J. Mol. Struct., 1994, 303, 11.
9” Paoli, D., Chabanel, M., J. Chem Research, 1991, 360.

9' Goralski, P., Legoff, D., Chabanel, M., J. Organomet. Chem, 1993, 456, l.
92 Melson.G..C_m131matron£hemisnxoLMaem§x§liLCQmmnd§ Plenum. New York. 1979.

9" Abou-Hamdan, A., Hounslow, A., Lincoln, S., Hambley, T., J. Chem. Soc. Dalton Trans, 1987, 489.
9‘ Groth, P., Acta Chem. Scand, 1987, 41A, 355
95 Bright, D., Truter, M., J. Chem Soc., 1970, 1544.

96 Barr, D., Doyle, M.J., Drake, S.R., Raithby, P.R., Snaith, R., Wright, D.S., J. Chem. Soc. Chem.
Commun., 1988, 1415.

97 Armstrong, D., Kandelwal, A., Raithby, P., Snaith, R., Stalke, D., Wright, D., Inorg. Chem, 1993,
32, 2132.

98 Mercer, M., Truter, M., J. Chem. Soc. Dalton, 1973, 2469.
99 Casini, G., de Sarlo, F., Donati, D., Guama, A., Orioli, P., J. Chem Research, 1984, 2.

“’° Kinoshita, H., Ouchi, A., Bull. Chem. Soc. Jpn., 1986. 59. 3495-

“" Robinson, J.W. NEW, CRC Press: Boca Raton, 1991.
Boston, 500.

"’2 Gilmore, C.J., J. App. Cryst., 1984. 17. 42-

“” Beurskens, P.T., Technical Report 1984/1 Crystallography Laboratory, Toemooiveld, 6525 Ed
Nijmegen, Netherlands.

"’4 Cromer, D.T. and Waber, J.T., WWW, Vol IV, The Kynoch
Press: Birmingham, 1974, Table 2.2A

194

”’6 Cromer, D.T., WWW, Vol. IV, The Kynoch Press:
Birmingham, 1974, Table 2.3.1

“’7 TEXSAN - TEXRAY Structure Analysis Package, Molecular Structure Corporation, 1985.

'08 Gaussian 92/DFI', Revision F.3, Frisch, M., Trucks, G., Schlegel, H., Gill, P., Johnson, B., Wong,
M., Foresman, J., Robb, M., Head-Gordon, M., Replogle, E., Gomperts, R., Andres, J., Raghavachari,
K., Binkley, J., Gonzalez, C., Martin, R., Fox, D., Defrees, D., Baker, J., Stewart, J., and Pople, J.,
Gaussian, Inc., Pittsburgh, PA, 1993.

“’9 Dunning, T. H., and Hay, P. J., Modern Theoret. Chem, Plenum: NY, 1976.

”0 Hehre, W. J ., Ditchfield, R., and Pople, J. A., J. Chem. Phys., 1972, 56, 2257.

"' Poirer, R., Kari. R.. and Csizmadia. I. G., W Elsevier: Amsterdam.
1985.

"2 Wachters, A., J. Chem. Phys., 1970, 52, 1033.

113Gianolo, L., Pavani, R., and Clementi, E., Chim. Ital., 1978, 108, 181.

”‘ Sakai, Y., Tatewaki, H., and Huzinaga, S., J. Comput. Chem. 1982, 2, 278.
”5 Huzinaga, S., J. Chem. Phys., 1965, 42, 1293.

”6 Glendening, B., Reed, A., Carpenter, J., and Weinhold, F., NBO Version 3.1.

”7 Pepke, B., Murray, J., Lyons, J., and ku, T-Y., Supercomputer Computation Research Institute at
Florida State University.

”8 Colthup, N. B., Daly, L. H., and Wiberley, S. E, ”WWW
W , Academic Press, Inc.:

“9 Paoli, D. and Chabanel, M., J. Chem Research, 1991, 360.

'2" Gans, P., Gill, J.B., and Fearnley, D.P., J. Chem. Soc. Dalton Trans., 1981, 1708.

'2' Lin, B., and McLean,J., J. Chem. Phys., 1980, 72, 3418.

'22 Feynmann, R., Phys. Rev., 1939, 56, 340.

'23 Berlin, T., J. Chem. Phys., 1951, I9, 208.

124 Bader, R.F'w” WW, Clarendon Press: Oxford, 1994.
'25 Howard, s.'r., Mol. Phys., 1987,85, 395.

‘26 Gruebele, M., Polak, M., Saykally, R.J., J. Chem. Phys., 1987, 86, 6631.

'27 Brooker, M.H., and Wen, N., Can. J. Chem, 1993,71, I764.

'28 Kameda, Y., Takahashi, R., and Usuki, U., Bull. Chem. Soc. Jpn., 1994, 67, 956.

'29 Probst, M., W Wiley: NY. 19% 553.
'3" Parrini, F., and Morales, R., J. Mol. Struct. (Theochem), 1993, 282, 59.

‘3' Cai, Z-L., Chem. Phys., 1983, 170. 33-

 

195

"‘2 Baranovskii, V.I., Sizova, O.V., Invanova, N.V., and Ermoleava, T.T., Z Struk., 1986, 27, 9'.
m Perelygin, I.S., and Shatokhin, S.A., Z. Struk. Khim., 1990, 31, 886.

'3‘ Pyykko, P., and Zhao, Y., J. Phys. Chem, 1990, 94, 7753.

”5 Yu., D., Rauk, A., and Armstrong, D., J. Phys. Chem, 1992, 96, 6031.

”6 Badger, R., J. Chem. Phys., 1934, 2, 128.

”7 Schultz, P.W., Leroi, GE, and Popov, A.I., J. Am. Chem. Soc., in press, 1995.

”8 Schultz, P.W., Leroi, GE, and Popov, A.I., J. Am Chem. Soc., submitted.

'39 Schultz, P.W., Leroi, Harrison, J.F., Mol. Phys., submitted, 1995.

‘4" Ramana, K., and Surjit, S., J. Mol. Struct., 1989, 194, 73.

'4' Lin, K-J, Cheng, M-C., and Wang, Y., J. Phys. Chem, 1994,98, 11685.

”2 Perelygin, IS, and Klimchuk, M.A., Z Struck. Khimi, 1983, 24, 67.

”3 Firman, P., Xu, M., Eyring, B., Petrucci, S., J. Phys. Chem, 1992, 96, 8631.

”4 Touaj, K., Chabanel, M., Trans. Faraday Soc., in press, 1995.

"5 Giffard, M, Cousseau, J., and Martin, G.J., Chem. Soc. Perkin Trans. 11, 1985, 157.

"6 Vaes, J., Chabanel, M., and Martin, M.L., J. Phys. Chem, 1978, 82, 2420.

”7 Musikas, C., Cullerdier, C., and Chachaty, C., Inorg. Chem, 1978, I7, 3610.

"8 Pan, W-H., Fackler, J.P., Kargol, J .A., and Burmeister, J.L., Inorg. Chim Acta, 1980, 44, L95.
”9 Kargol, J.A., Crecely, R.W., Burmeister, J.L., Inorgan. Chem, 1979, 18, 2532.

'5" Nicholas, A.M., and Wasylishen, RE, Can. J. Chem, 1987, 65, 951.

'5' Owens, G., Guarilloff, P., Steel, B.J., and Kurucsev, T., Aust. J. Chem, 1995, 48, 207.
‘52 Berman, H.A., and Stengle, T.R., J. Phys. Chem, 1975, 79, 1001.

'53 Miller, A.G., Franz, J.A., and Macklin, J.W., J. Phys.Chem., 1985, 89, 1190.

'5‘ Au—Yeng, S.C., J. Mag. Res., 1991, 92, 10.

'55 Polak, M., Gruebele, M., Saykally, R.J., J. Chem Phys., 1987, 87, 3352.

'56 Sherman, W.F., Wilkinson, G.R.. WW. ed. Hallam. HE.
Wiley: New York, 1973.

‘57 Paoli, D., Maryvonne, L., and Chabanel, M., Spectrochim. Acata, 1978, 34A, 1087.

'58 Schultz, P.W., Leroi, G.E., Ward, D., manscript in preparation. 1995-
'” Abragam, A., W, Oxford University Press: New York, 1961.

196

'6" Soong, L-L., W, Michigan State Univeristy, 1988.

'6' Lin, J.D., Popov, A.I., J. Am. Chem. Soc., 1981, 103, 3773.

”’2 Mereiter, K., Preisinger, A., Z. Kristall., 1984, 169, 95.

"‘3 Izatt, R.M., Pawlak, K., Bradshaw, J.S., Chem. Rev., 1991, 91, 1721.
”‘4 Christensen, J.J., Eatough, D.J., Izatt, J., Chem. Rev., 1974, 74, 351.

"‘5 Izatt, R.M., Bradshaw, J.S., Nielson, S.A., Lamb, J.D., Christensen, J.J., Chem. Rev. . 1985, 85.
271.

"‘6 Melson. G.. W. Plenum Pressz New York. 1979-
”” Eileen, M., Eick, H.A., Acata Cryst. Sec. B., 1982, B38. 1821.