1 ye. .... ‘..
‘-\, ' .I k» u
- A C. “1. A“. 7|-
~ ":4“ ?‘ ~.'-.I’\‘...
A RAMAN STUDY OF CRYSTALL-INE
ETHYLENES AND A DETERMINATION
OF THEIR STRUCTURE THROUGH
MODEL CALCULATIONS OF
THE LATTICE REGION
Thesis for the Degree Of Ph. 9;
MICHIGAN STATE UNIVERSITY
GLEN III R ELLIOTT
‘1972 e -
.........
I" LIBRARY
I Michigan State
‘ University
This is to certify that the
thesis entitled
A RAMAN STUDY OF CRYSTALLINE ETHYLENES AND A
DETERMINATION OF THEIR STRUCTURE THROUGH
MODEL CALCULATIONS OF THE LATTICE REGION
presented by
Glenn R. Elliott
has been accepted towards fulfillment
of the requirements for
Ph.D. degree in Chemistry
gfizm
Major professor
Date August 10, 1972
0-7639
alum:
‘ HUM: & SUNS'
I 800K BINDERY INC.
LIBRARY BINDERS
“I” main-ll
ABSTRACT
A RAMAN STUDY OF CRYSTALLINE ETHYLENES AND A
DETERMINATION OF THEIR STRUCTURE THROUGH
MODEL CALCULATIONS OF THE LATTICE REGION
BY
Glenn R. Elliott
In solid ethylene there are two possible orientations
of the two molecules per unit cell consistent with the
'crystal space symmetry. Extensive previous study of the
crystal has offered no conclusive evidence as to which
orientation is correct. Model calculations of the Raman
lattice spectra for the two orientations were undertaken
to evaluate which of the structures would more satisfactorily
reproduce the experimental observations. The investigation
gives strong evidence that the orientation with crystal
space symmetry P121/n1 (Czh)' the so-called "b—structure",
is correct.
The development of a general theory for model calcula-
tions of infrared and Raman lattice spectra has been com-
pleted and is discussed in some detail, and its applicabil-
ity to crystal structurecbtermination is indicated. The
model calculations were tested on the Raman lattice spectrum
of crystalline benzene, where the structure is well known.
The results not only indicate that calculations based on
the correct crystal structure can reproduce observed spectra,
Glenn Robert Elliott
but also that they can provide information useful for sym-
metry assignments where polarization measurements cannot
be obtained.
Several sets of carbon-carbon, carbon-hydrogen and
hydrogen-hydrogen interaction parameters were evaluated
during this investigation. It was determined that Williams'
parameter sets, most notably his most recent values, are
the best available for treating atom-atom nonébonded inter-
actions in solid ethylene.
Raman spectra of the intra- and intermolecular regions
of pure ethylene and ethylene-d4, and of the mixed crys-
tals ethylene (4%) in ethylene-d4 and ethylene-d4 (4%) in
ethylene, have been obtained. With the exception of the
fundamaflal region of pure ethylene, these Spectra have not
been reported previously. Assignments of all the factor
group components and of the matrix-isolated bands, in addi-
tion to several C-13 isotopic peaks, are given.
A RAMAN STUDY OF CRYSTALLINE ETHYLENES AND A
DETERMINATION OF THEIR STRUCTURE THROUGH
MODEL CALCULATIONS OF THE LATTICE REGION
BY
L
Glenn R. Elliott
A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Chemistry
1972
G--- 78"??? {a
JOHN STUART ORLEMANN
July 22, 1945 to August 19, 1969
"They wrote in the old days that it is sweet
and fitting to die for one's country. But in
modern.war there is nothing sweet nor fitting
in your dying."
Ernest Hemingway
Notes on the Next War
ii
.l
.l
ACKNOWLEDGMENTS
I must first acknowledge my parents whose example has
given me the dedication and perseverance that was necessary
to finally complete my formal education. Their constant
interest and encouragement is deeply appreciated.
I also want to thank Dr. George Leroi for the interest
and at times great patience with which he has viewed this
research. The personal interest which he has shown for
those in his research group has been important in creating
a relaxed and enjoyable atmosphere for doing research.
Gratefully acknowledged is the financial support ob-
tained from Michigan State University and the Office of Naval
Research. Deep appreciation also goes to the Department of
Chemistry for my appointment as an Assistant Instructor.
I want to thank Bonnie L. Marris for drawing the won-
derful portrait on the preceding page.
Finally, I wish to thank the countless people who have
made my stay here a very memorable one. Special mention
goes to my good friends Frank Chi, James Olson and John
Shock, and particularly to Mrs. Naomi Hack for numerous
diversions which have helped me maintain my sanity.
iv
CHAPTER
I.
II.
III.
IV.
TABLE OF CONTENTS
INTRODUCTION . . . . . . . . . . . . . . . .
TIEORY O O O O O O O O O O O O O O O O O O O
A.
B.
General . . . . . . . . . . . . . . . .
Intermolecular Lattice Vibrations . . .
Lattice Modes and Symmetry Coordinates
Crystal Dynamics . . . . . . . . . . .
Intensities of Lattice Vibrations . . .
Infrared Lattice Mode Intensities . .
Raman Lattice Mode Intensities . . .
EXPERIMENTAL . O O O O O O O O O O O O O 0 O
A.
B.
C.
Spectrometer . . . . . . . . . . . . . .
cryostat O O O O O O O O O O O O O O O 0
Sample Preparation . . . . . . . . . . .
BENZENE . C C O O O O O O O C O O O O O O O
A.
B.
C.
D.
Introduction . . . . . . . . . . . . . .
Crystal Structure and Symmetry Considera-
tions 0 O O O O O O O O O O O O O O O 0
Interaction Potential . . . . . . . . .
Results and Discussion . . . . . . . . .
ETHYIJENE . . O C O O I O O O O O O O O O O O
A.
B.
C.
D.
E.
Introduction . . . . . . . . . . . . . .
Crystal Structure and Symmetry Considera-
tions 0 O O O O C O O O O O O O O O O 0
Experimental Results . . . . . . . . . .
Theoretical O O O O O O O O O O O O O 0
Theoretical Results . . . . . . . . . .
REFERENCES 0 O O O O O O O O O O O O O O O O
V
Page
10
10
14
18
18
27
34
34
37
38
42
42
42
46
49
58
58
60
66
79
85
98
TABLE
II.
III.
IV.
V.
VI.
VII.
VIII.
IX.
XII.
XIII.
XIV.
LIST OF TABLES
Page
Lattice parameters and molecular orientations
for crystalline benzene . . . . . . . . . .
Correlation table for crystalline benzene .
Internal symmetry coordinates for benzene .
Atom-atom interaction potential parameters.
Calculated and experimental frequencies for
the librational lattice motions of benzene.
Comparison of model Raman lattice frequencies
and intensities with experimental values for
crystalline benzene . . . . . . . . . . . .
Lattice parameters and molecular orientations
for the two possible crystal structures of
SOlid ethYIene O O O O I O O O O O O O O 0
Correlation table for solid ethylene . . .
Internal symmetry coordinates for the libra-
tional lattice motions of ethylene . . . .
Observed
of solid
frequencies and relative intensities
and matrix isolated ethylene . . .
frequencies and relative intensities
and matrix isolated ethylene-d4 .
Observed
of solid
Observed Raman lattice frequencies and calcu-
lated frequency ratios for ethylene and
ethylene-d4 . . . . . . . . . . . . . . . .
Atom-atom interaction potentials used in
calculating the lattice dynamics of crystal-
line ethylene . . . . . . . . . . . . . . .
Minimum energy rotations and static lattice
energies for the two possible structures of
solid ethylene . . . . . . . . . . . . . .
vi
43
45
47
49
53
55
62
64
65
71
78
82
84
87
LIST OF TABLES (Continued).
TABLE Page
XV. Observed and calculated librational fre-
quencies for solid ethylene . . . . . . . . 89
XVI. Molecular polarizability tensor components
for ethYJ-ene O O C O O O O O O O O O O I O 90
XVII. Eigenvectors for Ag and B9 motions for both
possible ethylene structures along with their
corresponding observed frequencies . . . . 92
XVIII. Calculated and observed frequencies and rela-
tive intensities and calculated mean amplitudes
of libration for the two possible ethylene
crystal structures . . . . . . . . . . . . 94
XIX. Observed and calculated frequencies for
SOIid ethYIGDE-d4 . o I o o c o o 0 o o o o 97
XX. Calculated and observed frequencies and rela-
tive intensities and calculated mean amplitudes
of libration for the two possible ethylene-d4
crystal structures . . . . . . . . . . . . 97
vii
FIGURE
1a.
lb.
2.
LIST OF FIGURES
Page
Spray-on cell for Malaker refrigerator. . . 39
Condensation cell for Malaker refrigerator. 39
Raman spectrum of crystalline benzene in the
lattice region . . . . . . . . . . . . . . 51
Raman spectra of the intramolecular region of
solid ethylene . . . . . . . . . . . . . . 68
Raman spectra of the internal vibrations of
C2H4 (4%) iSOlated in C2134 o o o o o o o o 73
Raman spectra of the intramolecular region
Of SOlid ethy1ene-d4 o o o o o o o o o o o 75
Raman spectra of the intramolecular region
of matrix isolated ethylene-d4 (4%) in C2H4 77
Raman spectra of the lattice re ion of
ethylene and ethylene-d4, at 30 K. The peak
marked with an asterisk is a laser fluorescence
line . . . . . . . . . . . . . . . . . . . 81
viii
CHAPTER I
INTRODUCTION
In the past decade a great amount of work on the study
of crystals by optical spectroscopy has been carried out.
This work has been stimulated by the quantitative under-
standing of crystal motions brought about by the extension
of normal coordinate analysis from studies of free molecules
to its current deep involvement in predicting and describing
normal modes and frequencies of Optically active crystal
motions. Much of this work has been done on molecular
crystals, although experimental studies and calculations
have also been made on ionic, covalent and.metallic crystals.1
The primary interest in molecular crystals is based on the
simplicity of the form of the interaction. Molecular crys-
tals do not display the strong metallic or covalent bonds
or the long range electrostatic interactions of the other
crystal types, and in many cases dipole and quadrupole
interactions need not be considered. Molecular crystal
motions can be divided into two classes. Intramolecular or
internal motions can be described as the perturbation of
gas phase molecular motions by weak Van der Waals, quadru-
pole or dipole interactions between different molecules.
These same weak interactions restrict the free rotational
1
2
and translational motions of gaseous molecules, giving rise
to intermolecular or external motions. Initially molecule-
molecule interatomic interactions directly available from
second virial coefficients were used to form interatomic
potential functions.
Early papers have established a qualitative under-
standing of crystal motions by a group theoretical treat-
ment.2 For molecular crystals, group theory predicts the
appearance of lattice motions corresponding to rotations
and translations, and the splittings of fundamental motions
consistent with the number of molecules per primitive unit
cell and its symmetry termed factor group splitting. Also,
in cases where gas phase motions show degeneracy, group
theory predicts splittings due to site symmetry, site group
splitting. In the infrared, many studies of intramolecular
motions and less frequently of intermolecular motions have
been made. Intermolecular and intramolecular motions have
both been frequently studied with the Raman effect. Much
of this work has been reviewed by O. Schnepp.3 Where the
crystal structure has been known, the results have agreed
with group theoretical predictions. In cases where the
structure is not known, correlation of observed results with
group theoretical predictions can narrow down the choice of
possible crystal structures. The basic theory and the
infrared work in the area have been reviewed by Vedder and
Horniga and by Dows.a
With the development of improved instrumentation,
spectral features of crystals could be more thoroughly
3
characterized. The use of grating spectrometers and inter-
ferometers in the far infrared region have permitted studies
down to 20 cm-l. The use of laser light sources and iodine
filters in the Raman have permitted shifts as low as 10
cm- for single crystals and 3 cm.1 for gas phase spectra
to be observed. With the use of these more powerful light
sources and more sensitive detectors, the increased signal-
to-noise ratio permits resolution on the order of one half
cm"1 and observation of extremely weak vibrational motions.
Finally with the development of cryogenic instruments and
techniques, the cumbersome and restrictive handling of
cooled samples has been virtually eliminated. Thus, high
quality Spectra have been obtained in the intramolecular
and intermolecular regions for acetylene7 and anthracene
and naphthalene8 in the Raman and ethylene9 in the infrared.
Indeed, much more difficult systems have been studied Spec-
trosc0pically with good results. These include matrix
isolation studies of xenon dichloride10 in the Raman and
of free radicals such as OF,11 C1012 and monobromomethyl13
in the infrared, as well as fluorescence studies of iodinel.4
Our present ability to obtain high quality spectra permits
a full characterization in the case of crystal motions:
their frequencies, intensities, splittings and in many cases
their space group symmetry.
This full characterization of the crystal motions now
offers a base to test the theoretical developments over
the past decade. These developments include the
4
calculations of the frequencies, intensities and Splittings
which are observed by spectrosc0pic techniques. Shiman-
ouchi, g£_3l,15 have applied the Wilson GF method16 to the
calculation of optically active crystal vibrations and
Walmsley and Pople17 have shown how to determine the inter-
molecular force field from molecule-molecule (center—center)
pair interactions. An extensive review of this field is
given by Venkataraman and Sahni.1 Recently a more sophis-
ticated calculation has been reported,18 the self-consis—
tent phonon method, which includes anharmonicity and ampli-
tude corrections allowing studies close to phase transitions.
Discussion of the sources of intensity of the lattice vibra—
tions is given by Schnepp19 for infrared absorption in—
tensities and by Cahill20 for Raman scattering intensities.
These are reviewed in detail by Richardson.21 Models used
to calculate exciton splittings are discussed by Taddei and
Giglio22 and by Blumenfeld, §£_§l,23
Most of the studies to date have been made on molecules
which possess a center of symmetry. These molecules offer
two major advantages. First, they usually crystallize in
simple structures which possess a center of symmetry and
have only two or four molecules per primitive unit cell.
This simplifies the calculation of the intermolecular force
field for external motions where the crystal structure is
known, and in cases where only the space group is known it
simplifies the calculation of the minimum energy orientation
through crystal symmetry. Second, in the majority of these
5
cases, the molecules occupy inversion sites within the
unit cells. Therefore, no mixing of librational and trans-
lational lattice motions can occur for zone center crystal
motions. Further, only the librational motions are Raman
active and only the translational motions are infrared
active. This permits the complete experimental and compu-
tational analysis of librational and translational lattice
modes to be carried out independently.
A great deal of progress has been made in the calcula-
tion of lattice modes. Early calculations on ethylene24
and on benzene and naphthalene25 indicated that the short
range hydrogen-hydrogen and hydrogen-carbon repulsion
forces were of primary importance in determining lattice
frequencies. Recent calculations on ethane, ethylene and
acetylene using short range repulsions and a diatomic ap-
proximation have shown reasonable results for translational
lattice motions.26 Calculations using diSpersion forces
and quadrupole-quadrupole interactions along with the short
range repulsive forces have been made on carbon dioxide,17
cyanogen27 and oxygen.28 For cyanogen and oxygen, a slight-
ly better fit was obtained by excluding the quadrupole—
quadrupole part from the interaction potential. This was
not possible for carbon dioxide since a center-center inter-
action was employed. Calculations by Shinoda and Enokido29
on carbon monoxide and by Goodings and Henkelman3° on nitro-
gen have used correction terms and have invoked anharmonicity
to explain the deviations from experimental data. Suzuki
6
and Schnepp31 have shown that anharmonicity can be very
important for molecules smaller than carbon dioxide. The
major conclusions drawn from these results are that the
dispersion part of the interaction potential is important.
the prOper choice of interaction potential parameters
greatly improves calculations, good fits can be obtained
without the inclusion of higher multipole moments, and
anharmonicity can cause large shifts from harmonic frequen-
cies for small molecules.
In light of these advances in the theory of crystal
lattice motions, we have undertaken a thorough investiga-
tion of the possibility of structure determination by com-
parison of model lattice spectra calculations and experi-
mental studies in the intermolecular lattice region. Our
investigation was made on Raman-active external motions
since the method of calculation of Raman scattering inten-
sities from molecules which possess an inversion center is
more solidly founded than is the calculation of their infra-
red absorption. The choice of crystals for this examina-
tion was based on their symmetry. Only crystals where the
molecules occupied sites of inversion symmetry were con-
sidered in order to insure that there would be no mixing of
translational and librational motions. Benzene was chosen
to test the theory for several reasons. First it has been
studied extensively and has been shown to obey group theor—
etical predictions. Second it is easy to handle. Finally
and most important, its crystal structure has been determined
7
with a high degree of accuracy at several temperatures with
both X-ray diffraction methods and neutron scattering
techniques. Benzene also meets the requirements of symmetry
by having an inversion center and by crystallizing in a
highly symmetric orthorhombic Space group with molecules on
sites of inversion symmetry.
The results of the calculations of the intermolecular
lattice spectra of benzene indicate that these calculations
can reproduce experimental lattice spectra of Raman active
librational motions within the accuracy of the lattice
parameters, orientations and potential functions. Having
determined the accuracy of the calculations, we feel more
confident in our crystal structure determination of ethylene.
Ethylene was a natural choice to test the theory since its
crystal structure is not known exactly. Packing considera-
tions and static lattice energy calculations indicated two
possible crystal structures for ethylene.24 These two
crystal structures did not differ in space group so group
theoretical predictions could not distinguish these possi-
bilities. The two possible crystal structures differed
only in the relative positions of the molecular planes of
the two molecules per primitive unit cell.
We have calculated the molecular orientations, static
lattice energies, librational frequencies and relative in-
tensities of the librational motions using several potential
functions obtained by different techniques. These calcu-
lations were made for each of the two possible crystal
8
structures of ethylene. The results of the calculations
indicate that the relative intensities offer the best test
of molecular structure, since they appear to be independent
of the choice of interatomic potential functions. The re-
sults of this study leave no doubt which of the two possible
crystal structures is correct. They also further support
the theory that crystal structures can be determined by
comparison of experimental data with model spectra calcula-
tions of the intermolecular lattice region.
CHAPTER II
THEORY
A. General
The vibrational spectra of the intermolecular region
of molecular crystals are a source of much information about
the forces acting between molecules and about the molecular
orientations within crystals. Because of their periodic
nature, crystals represent ideal systems with which to
study intermolecular interactions. This can be done by
comparing calculated properties of these intermolecular
interactions such as molecular orientations, elastic con-
stants, external lattice frequencies and fundamental split-
tings with experimental data obtained by X—ray and neutron
diffraction, compressibility measurements and infrared and
Raman spectra of the inter- and intramolecular regions.
The independent study of external lattice motions is justi-
fied on the basis of the normally large energy separation
between the intermolecular motions and intramolecular
motions. Energies of intermolecular motions rarely exceed
200 cm-1, while the lowest skeletal vibrations of simple
molecules are rarely less than 600 cm-1. Because of this,
coupling between motions of the same symmetry in these two
classes is negligible. Using the oriented gas approximation?2
9
10
in which solids are viewed as being composed of oriented
molecules whose physical properties remain unchanged from
the gas phase, independent of the perturbations of sur—
rounding molecules, it is not difficult to calculate lat-
tice mode intensities for the optically active librations
and translations. Relative lattice mode intensities are
highly orientation dependent. Therefore, calculated rela-
tive intensities and frequencies may permit assignment of
crystal structures in the absence of complete X-ray or
neutron diffraction data.
B. Intermolecular Lattice Vibrations
Lattice Modes and Symmetry Coordinates21
Intermolecular lattice vibrations can be classified as
two types. Translational lattice vibrations can be repre-
sented as translations of one sublattice with respect to
the others. This is best described as a shearing motion of
the sublattices against each other. The translational dis-
placement of the qth molecule in the pth cell is de—
noted by ra(p/q) where a represents the three directions
x,y,z. Rotational lattice vibrations can be represented
as the rotations of the molecules of one sublattice about
their centers of mass with reSpect to the rotations of the
molecules on the other sublattices. The librational dis-
h molecule in the pth cell is denoted
placement of the qt
by Aa(p/q) where a represents the three axes of rotation
x,y,z. The instantaneous position and orientation of a
11
molecule are given by
Ra(p/q) rgIP/qI + ra(p/q) and. (1)
O
Aa(p q) AaIp/q) + Nam/q) (2)
respectively, where the superscript 0 indicates the
equilibrium position or orientation. For systems with
linear molecules, the redundancy condition is given by
ZlAa(p/q)lz = z IA°
]2. (34)
where N is the number of unit cells in the path, Qn is
the normal coordinate of the transition from Y" to Y',
and g, is the instantaneous dipole moment Of the crystal.
The evaluation of the integrated absorption intensity for
the normal coordinate, Qn' requires the evaluation Of the
factor
. (35)
In the harmonic approximation, this factor vanishes except
for a change Of one in the quantum number of the normal
coordinate, Qn' and E.# O.
The determination of the integrated intensity requires
the determination Of the instantaneous dipole moment, E'
The instantaneous dipole moment can be expanded in terms of
the normal coordinates about their equilibrium positions to
give
I3 = i» Rim) + A} g (OLE/30pm; + (36)
20
neglecting higher order terms. The first sum represents
the contribution of the permanent moments over all the
molecules to the instantaneous crystal moment. This term
will not be considered here since it does not apply to the
systems to be discussed in this dissertation, crystals com-
prised Of molecules with a center Of symmetry. Substitution
Of equation (36) into equation (35) shows that the condi-
tion for infrared intensity of a given normal coordinate,
is
Qn'
=
, (37)
ul/BQ )0 75 0-
N n
is (a
The first factor on the left is non-zero for one quantum
transitions. In fact for translational lattice modes, it
equals (1.68 x 10‘15/MV")1/2, and for librational lattice
modes, it equals (16.8/IU")1/2 where M and I represent
the molecular mass and the moment of inertia for the normal
mode respectively.19 The second factor represents the sum
Of the derivatives of the contributions to the instantaneous
dipole moment Of the crystal by the molecules, 1, Qpi/Oon.
The instantaneous dipole moment Of molecule 1, Q}, will
in general consist Of three parts; the induced molecular
moment caused by the local electric field, the permanent
moment and the short range distortion moment.20 The dis-
tortion moment is not expected to be very large for small
amplitude (v = OI—> 1) vibrations at very low temperature.
21
It was found that in the pure rotational spectrum of nitro-
gen35 and hydrogen36 gases, that the distortion moment is
very small and decreases with temperature. Therefore, it
will not be considered here. The induced moment consists
Of dipole—induced moments and quadrupole-induced moments.
For molecules possessing permanent dipole moments, the
quadrupole-induced moments will generally be insignificant.
Permanent moment derivatives predominate over induced
moments for librational motions. However for translational
motions, they will not produce a change in the total instan-
taneous moment of the crystal since only spatial distribu-
tions are changed. For crystals composed of non-polar
molecules, the intensity must be wholly derived from in-
duced molecular moments. Therefore the remainder of this
section deals only with induced molecular moments.
The induced dipole moment is dependent upon the elec-
tric field at the perturbed site and the polarizability of
the molecule on that site by the relation
A = Q.§.- <33)
By substituting equation (38) into the second factor on the
right in equation (37), we obtain
- . . . i
(apf/aonIO = gj(a§}/aon)o + (agf/aQnIOED . (39)
If the molecule is located on a centrosymmetric site, the
i
second term vanishes because ED vanishes. Also, for trans-
lational motions, it vanishes because the spatial orienta-
tional remains unchanged and (Egg/Oon)o vanishes. For
22
librational motions, it must be considered. However, since
it is proportional to the anisotropy of the polarizability,
it may be small. The effect Of both of the terms in equa-
tion (39) will be discussed in terms of internal symmetry
coordinates next.
Since the normal coordinates can be related to in-
ternal symmetry coordinates by an orthogonality transform,
an, equation (39) becomes
(agf/aon)o = fi an(8H?/Bsm)o and (40)
i i
E (Og'/OQn)o = % an g (ag,/asm)o
= 2 L z [Oi(OEi/OS )0 (41)
mnmi’z'v m
+ (ng/Osm)§:] .
The sum over i is over all molecules, q, in all the unit
cells, p.
The first term in the brackets in equation (41) can be
expanded in terms of internal coordinates by
(aei/OsmIo g (BEi/BRa(p/q) I. (42‘
Z Z
q P
x (5Ra(p/q)/Bsm)o
NW2 5 a when 120: (agi/an QIp/qn. (43)
for translational motions, and similarly for librational
motions
23
(agi/asmIO = N'l/zg g Tgfiq g (651/8 Aa(p/q))o~ (44)
The field at molecule i is calculated to be the sum over
the contributions from the neighboring molecules, j; j
here represents all molecules (p/q),
if = .3; §i(j)IR(j).A(j)] . (45)
Equation (45) states that the electric field at site i is
due to the contributions from all the surrounding molecules,
j. Further it indicates that the contribution from each
molecule, j, is dependent upon the distance from site i
to molecule j, R(j), and the orientation of molecule j,
IA(j). Substituting this into equation (43), we Obtain
i _ -1/ . i
(as /asm)o - N 2g g ngm'aqu (p/q)/5Ra(p/q))o
+ Tgwi (as? (p/q)/5Ra(i) )o I. (46)
Because of the symmetric relationship between the internal
ccordinates Ra(p/q) and Ra(i) of the same internal sym-
metry coordinate, this sum vanishes unless Tm,aq and
Tm,ai differ in sign or magnitude.19 Substituting equation
(45) into equation (44), we note that in this case the
derivatives with respect to (Aa(p/q) will be nonvanishing
only if p/q denotes molecule j. Thus only one term of
the sum over j remains giving
24
i = -1/ u 1
(BE /Bsm)o N 2%. 531‘qu EXDIBE (p/qI/B AO(p/q))o. (47)
The second term in equation (41) will contribute when
librational motions are infrared active, i.e. when the sites
are non-centrosymmetric. Since it has only been evaluated
for linear molecules, we will discuss it only in that con-
text. The u v element of the polarizability tensor g}
for molecule 1 is
i _ . .
auv - ai_éuv + BaaVKIAu(1)1\V(1)I (48)
in which aav is the average polarizability, K is the
anisotropy of the polarizability and ‘Au(i) and [\v(i) are
the instantaneous direction cosines relating the molecular
axis of molecule 1 to the u and v crystal axes. The
. . i .
derivative Of the auv element Wlth reSpect to a symmetry
coordinate, Sm' becomes
' ’1 nu .
(aafiv/asm), = 3N (2 aavIc[T;r"'uq/\3(i)+Tm'qu3(1)]o
(49)
The evaluation of the equations (46) and (47) re-
quires an expression for the contribution from molecule q
in unit cell p to the electric field at site i. In
molecules which do not possess an intrinsic dipole moment,
the major contribution to the infrared intensity for lat-
tice motions will be from quadrupole-induced moments. For
linear, non-polar molecules, the contribution to the elec-
tric field in the a crystallographic direction at site 1
due to molecule (p/q) is given by
25
i -.- 2 _ -
Ram/q} (39/2D‘H5dadz. Zdz. Aa da) (50)
where 9 is the quadrupole moment, D is the absolute
magnitude of the equilibrium position vector of molecule
d d are
11'?-
its direction cosines in the crystal-fixed coordinate sys-
(P/Q) relative to the molecule on site i, dx'
tem, d d ,, d its direction cosines in the coordinate
x" y 2"
system native to the molecule (p/q) and AX, Ay, A z are
as defined above. Thus
dz, = a31dX + aasz + a33dz = g [\ada (51)
where the elements of the matrix 2‘ are defined as follows
£‘E.
51
where {J is a vector in the molecule axis system and r,
is a vector in the crystal system. Substitution of equa-
tion (50) into equation (47) for example requires the rela-
tion
(5E;(P/q)/5I\B(P/q))o = (39/2D‘II10dad)dz.-2dz.5 -2d
- . A
E ob b o)Pq
(52)
where 6 represents a rotation axis parallel to one of the
x, y, 2 molecule fixed axes.
When a dipole moment exists, the contribution of the
dipole-induced moment and the intrinsic moments to the
infrared intensity must be considered. To consider the
effect of dipole-induced moments, it is only necessary to
introduce the relation for the contribution of the intrinsic
26
dipole moment on molecule (p/q) to the electric field at
site i. This is given by
i = 4 _
Ram/q) (M/D )( 3d,.da + Ag) (53)
where M is the magnitude of the dipole moment. If the
molecules possess an intrinsic dipole moment then libra-
tional motions will be infrared active since they vary the
orientation of the permanent molecular moments. The in-
stantaneous component Of the dipole moment on molecule i,
ul, along the a axis of the crystalebased system is
i - o . .
Ea - MlIAa (i) + Aa(l)], (54)
in which /\:(i) is the equilibrium direction cosine and
Aa(i) is the change in the direction cosine. Using equa-
tions (7), (9) and (40), we have
i _ '1/ .. i
(at wen). - N 2?. an g: g Tm,aq EON onto/mags)
Since g} depends only on A.a(p/q) where (p/q) refers to
molecule 1, equation (55) reduces to
i '1/ . .. i .
(5p. /aQn)o N 32 L iTm'ang /a Aa(1))° (56)
mnm
= 2 II
N gtznsz'aiL M. (57)
Evaluation Of the intrinsic moment contribution to infrared
intensity for librational motions is now possible by sub-
stituting equation (57) into equation (37).
27
Raman Lattice Mode Intensities21
The total intensity Of Raman scattering associated
with a transition from some state a to some state b is
given by the quantum mechanical expression
-1
Iab = NaC'(vi - ¥)4g|P|2(1 - exp(-hy{kT)) I (58)
where Na is the number of molecules in the initial state,
a, vi is the frequency Of the incident light, Vn is the
frequency Of the Raman shift, 9 is the degeneracy Of vi,
and C' is a constant.37 The induced transition moment,
|P|, for a given normal coordinate, Qn' is given by
IPI = IgIYbIon». (59)
The field on a particular molecule is due to the electric
field, g, of the incident radiation and the fields induced
by the surrounding molecules. The induced moment, 2, then
becomes
3 ' E gdlg ‘kgi Tik 2km ' (60)
where %d is the polarizability of the perturbed molecule,
2: is an identity matrix, Tik is an element of the reac-
tion field tensor, and %k is the polarizability Of the
molecules surrounding the perturbed site, i. The effec-
tive polarizability Of the crystal go can be expressed as
2&2'§%d [g'kii Tik %<]' (61)
28
In the infrared there were three factors which con-
tributed to the integrated intensity; the distortion moment,
the induced moment, and the permanent moment. As in the
infrared, the distortion moment is not expected to contrib-
ute appreciably to the Raman scattering. Also the perman-
ent moment has no significance in the Raman. Therefore,
the total scattering can be attributed to the induced
moment. First, we will examine the effect Of the induced
moment on the scattering from translational lattice vibra-
tions. This is expected to be small because translational
lattice vibrations do not change the orientation of the
molecules within the crystal. Thus, they are not expected
to change the polarizability Of the crystal, go’ to any
great degree. Next, we will examine the effect the induced
moment has on scattering from librational lattice vibrations.
Finally, the form Of the induced moment for polycrystalline
samples will be discussed.
Neglecting the distortion moment, we examine the Raman
scattering from translational lattice vibrations. For
translational motions, the orientation of the molecules in
the crystal and hence their polarizability remains unchanged.
Therefore, the induced moment for these translational mo-
tions is wholly dependent on how the reaction field tensor
varies with these motions. The reaction field tensor
varies with the normal coordinates, Qn = Qgcos(2wvnt),
according to
.. 0 o
Tik - Tik + (aTik/aon)° on + ... , (62)
29
where Tik is the instantaneous reaction field tensor and
Tik is the equilibrium value. The induced moment, g, of
the crystal then becomes
- o _ o .
5 ' 17‘ ERIE: kii Tikgac kii‘Mik/BQnMQn (213.1 Ex (”3)
Including the form Of the normal coordinate given above and
the time dependence Of the electric vector of the irradia-
tive light, §'= E?cos(2wvit), we obtain
g'- g gd[gfos(2wvit) -k§i Tikgk cos(2wvit)
_ o
kiiOTik/Oon)ocos(2wvnt)cos(2wvit)gk]E . (64)
The terms in the induced moment which give rise to Raman
scattering involve a change in frequency of (vi - vn) for
Stokes scattering and of (vi + vn) for anti-Stokes scatter-
ing. Since our interest is in the Stokes scattering, we
will examine only the (vi - vn) frequency condition. With
the relation
1
COS(2vvit)cos(2wvnt) I §{cos(27r(vi - vn)t)
(65)
+ cos(21T(vi + vn)t)]
and the condition that for scattering cos(27T(vi - vn)t) = 1,
we Obtain
= _ o o
13(Stokes) >13 93 z. (a'rik/aon)0 gkong /2. (66)
k#i
For centrosymmetric crystals, when (BTik/OQn)o is summed
30
over all the molecules in the infinite crystal, all Of its
terms vanish due to symmetry and the induced moment for
translational lattice vibrations is zero. Consequently,
Raman scattering for translational lattice vibrations
vanishes under the approximation Of infinite crystal and
no distortion moment. This is not true for non-centrosym-
metric crystals and therefore translational lattice motions
can be Raman active in this case.
In a similar fashion, the induced moment for librational
lattice vibrations can be derived. For librational motions,
the intermolecular spacing remains unchanged so the reac-
tion field tensor remains constant. However, the molecular
orientation changes with the normal coordinates according
to On - Qgcos(2vvnt). The molecular polarizabilities can
be written in terms Of the normal coordinates
= a9 (Bad/aonuog + , (67)
2‘4 m
With this relation and the time dependence of the normal
coordinate and incident radiation, the induced moment be-
comes
: 0 - 0 _ ‘ 0
g 2.30.“; Z.Tikg3< “z Tik(5%i/don)°on
l i¢k
X cos(2wvnt)]§P cos(2wvit)
O _ O
" Res/awn. ... <2Tvnt>12 Z Tikgk
k¢i
-k:iTik(Ogd/Oon)003 cos(2wvnt)]EP cos(2wvit). (68)
Since we are only interested in the terms which show
31
(vi - vn) frequency dependence, the induced moment reduces
to
: .. 0 O O
P i gi kii Tik(5gk/OQn)an §,/2
- o o o
+ >; (a%/aon)otg >3_ Tikgkmne /2 (69)
i k#i
O o o
= , - . 2. O
(gagging 1.321 legglmong/ (7 )
Substituting equation (61) into equation (70) we Obtain the
induced moment in terms of the crystal polarizability, gco
g,= (Egg/OQn)oQg§P/2 . (71)
This total induced moment for Raman scattering from libra-
tions can be expanded in terms of individual components of
the tensors and vectors
1/
= 0 2 2
2, Ion/2H2 [2(OapO/OQn)oEO] 1 , (72)
p o
where E0 is the component of the electric vector, E9, in
the 0 direction and apo is the pO component Of the
crystal polarizability tensor.
The square of the induced transition moment defined in
equation (59) is given by
1
(Plz = ||2- (73)
Since only 0% is dependent on Ya(Qn) and Yb(Qn),
32
equation (73) can be written as
1 2 2
2 = _. o
|p| 4 g [§(aapO/aon)oso] xl<‘1’a(Qn) IQnI‘l’b (on)>| .
(74)
In the harmonic oscillator approximation, the integral
(Ra(on)log|wb(on)> for librations with a = O and b = 1
1
has the value (16.8/Ivnc) /3, where I is the moment Of
inertia about the axis of rotation, vn is the frequency of
the normal coordinate on, and c is the speed of light in
centimeters per second. Substitution of equation (74) into
equation (58) gives
I U 2
Iab NaC (vi - vn)4g|l
x [2 [2(aapO/Bon)oE012](4I1 - eXPI-hvn/kTII)‘1 (75>
p O
. K'(vi - Vn)‘9[% [g (OapO/OQA)OEO]2]
x (vn[1 - exp(-hvn/kT)])-1 (76)
where K' = (NaC'c x 16.8/4) and 08 is the mass weighted
normal coordinate, Q; = 11/2Qn_
In our experimental system, we look at a random sample
Of crystallites. This necessitates an evaluation of the
last term in equation (76). By integrating over all pos-
sible angles of incidence of the exciting radiation, we
learn that the intensity can be evaluated with the assump-
tion that one-third Of the radiation is incident from each
of the crystal axes, X, Y, 2. Equation (76) is now evalu-
ated for the three cases, Ex = 1/3. Ey = 1/3. and E2 = 1/3
33
I I I _ 4 ' . I 0 2
Iab K (vi vn) g %[ %(OOPO/don)EO]
x (vn[1 - exp(-hvn/kT)]) 1 (77)
= (1/3)K'(v. - v n)4g[z 2 up 02]
p o
x (vn[1 - exp(-hvn/kT)])'1, (78)
where ago is the derivative Of the po polarizability
tensor component with respect to the mass weighted normal
coordinate, QB. In terms of the invariants
5' = (1/3)(a;{x + a§y + aéz) and (80)
(7')2 = I1/2)I(a;(x - ayy)2 + (ayy - HQZIZ
(81)
+(a'
zz - aéx)2 + 6[(d}'{y)2 + (a9z)2 + (a'zx)2]].
referred to respectively as the mean value and the
anisotropy Of the derivative tensor, equation (78) reduces
to
k(vi - v n)4 g 2
Iab a Vn [1 ~1exp(éhvn Zkt)] [4 5(5 ) + (7 )2 1 (82)
where K 3 (1/3)(3/2)K'. As we have shown, for centrosym-
metric cells only librations show Raman scattering in the
lattice region. Therefore, equation (82) is the only con-
dition for Raman scattering intensity for polycrystalline
samples in this region.
CHAPTER III
EXPERIMENTAL
A. Spectrometer38
The spectrometer used for these experiments consists
of a laser light source, a double monochromator, a photo-
multiplier, and the associated electronics and Optics. For
the laser light source, either a Spectra-Physics Model 140
Ar+ laser with a power output of approximately 1.5 watts
each in the 5145 R and the 4880 2 lines or a Coherent Radia-
tion Model 52BeO Ar+ laser with a power output of nearly
2.0 watts each in the 5145 R and the 4880 R lines was
employed. Both laser sources have excitation line widths
Observed to be less than 0.8 cm.1 at half height. The
laser lines drOp sharply to the base line and permit the
Observation of Raman scattering as close as 5 cm-1 to the
exciting line for gas samples. In our analysis, the most
highly scattering polycrystalline samples could be observed
to within 10 R Of the exciting wavelength, permitting studies
to within 40 cm‘1
of the exciting line for these cases.
When using the 4880 A laser line at high power, a fluor-
esence line at 4889 R appears with appreciable intensity.
This caused little difficulty in either the intermolecular
or intramolecular regions since all shifts were checked
34
35
with the 5145 2 laser line. No additional fluorescence lines
were observed in any of the regions of interest, so broad
band filters were not needed.
The monochromator used is a Spex Model 1400. This
double monochromator is, in effect, two Czerny-Turner 3/4-
meter grating Spectrometers mounted in tandem. This is
necessary to reduce stray light to a level acceptable for
investigations of the weak Raman effect. Two sets of
gratings were used. Initially, two Bausch and Lomb grat-
ings with 1200 grooves/mm., blazed at 5000 X in the first
order were used. A problem encountered with these gratings
was the appearance of grating ghosts near the exciting wave-
length, particularly when examining highly scattering poly-
crystalline samples. This problem was somewhat alleviated
by determining in advance the positions of the ghosts.
Since they are symmetric with respect to wavelength about
the exciting source, their positions and relative intensi-
ties could be characterized by placing a scatter plate or
ground glass joint in the sample area and focusing the
scattered Rayleigh light on the spectrometer entrance slit.
(Raman scattering, on the other hand, is symmetric in wave-
number about the exciting line.) The subsequent purchase of
a set of interferometrically ruled gratings from JobinéYvon
eliminated the prOblem of grating ghosts. These select
quality gratings were also ruled with 1200 grooves/mm. and
blazed at 5000 X in the first order. The spectrometer has
three slits which are 50 mm. high and can be bilaterally
36
adjusted from a few microns to 3 mm, and is equipped with a
speed control which permits the selection of several scan
speeds.
The spectrometer was calibrated using argon, neon,
and krypton gas discharge lamps. It was determined that a
linear relation between dial reading and actual wavelength
fit the calibration as well as any higher order polynomial
function. Thus a constant correction factor was used in
generating conversion tables from wavelength dial reading
of the monochromator to wavenumber shifts. However, be-
cause of the daily vagaries in the monochromator drive, it
is necessary to obtain the correction constant by deter-
mining the difference between the dial reading and the true
wavelength of the laser excitation line before each experi-
ment.
For the detector, either a selected ITT FW-130 Star-
Tracker end-on photomultiplier which has S-20 response or a
selected RCA C31034 with similar characteristics along with
slightly improved red response can be used. Both detectors
are cooled thermoehxxrically to -20°C in a cooler-photo-
multiplier housing assembly manufactured by Products for
Research. When cooled to -20°C a dark cound of 3-4 counts
per second can be achieved. A floating lens assembly pro—
vided by Spex allows precise alignment of the radiation from
the exit slit on the detector. With this assembly, good
signal-to-noise ratios can be obtained down to the 10-11
ampere output range on our Victoreen Model VTE-l micro-
microammeter.
37
B . Cryostat
A Malaker Corporation Cryomite Mark VII—C was used to
cool and control the temperature of all the solid samples.
This instrument works on the principle of adiabatic expan-
sion of helium gas, much like a standard refrigerator works
with freon. It is capable of controlling temperatures of
samples over a wide range from 250K up to room temperature.
The temperature control is stable to well within one de-
gree fluctuation. This control permits studies of spectral
features, such as frequencies, band widths and intensities,
as a function of temperature. It also has the advantage
that it is small and can be moved from sample preparation
areas to the sample region of our Spectrometer with ease.
Complete characterization of the Raman spectra of
solids requires polarized spectra of single crystal samples.
However difficulty in sample handling and preparation pre-
cludes the obtaining of single crystal samples in many
situations. When this is the case, as it is for ethylene,
it becomes necessary to study polycrystalline samples.
Since selection rules are based on the assumption of an
infinite crystal, crystallite size is important. If crystal-
lite size is too large, then complete random interaction of
polarized radiation with the sample,as assumed in the de-
velopment in the previous chapter, may not occur. 0n the
other hand, if the sample is highly powdered, then break-
down of selection rules, broadening of spectral lines, and
high Rayleigh scattering are possible. The temperature
38
control of the Cryomite helps circumvent these problems by
permitting the annealling of powdered samples and complete
shattering of samples obtained from the melt to give ac-
ceptable crystallite sizes. For these two types of sample
preparation, two cold head assemblies were designed for the
the Cryomite. The first is a Spray-on head where the gas
sample is allowed to diffuse into the cell, which is mounted
on the cold block of the cryostat, and quickly solidifies
on the copper face of the sample cell. This head is illus-
trated in Figure 1a. The sample formed by this means is
usually of a highly powdered form and needs to be annealled
before good spectra can be obtained. The second head as-
sembly, shown in Figure 1b, passes the gas through the cop-
per mounting block of the sample cell where it is liquified
before entering the cell. Once the sample cell is full,
the liquid is rapidly frozen to insure adequate shattering
to give a polycrystalline sample. Since, with this technique
the crystallites are relatively large with reSpect to
direct deposition, the sample suffers from none of the dis-
advantages of the spray-on technique. However, it may suf—
fer from insufficient shattering, giving rise to only par-
tially oriented crystallites. Of the techniques, the
second uses more sample requiring about 300 ml S.T-P. of
gas, nearly twice that required by the first.
C. Sample Preparation
All samples used were initially obtained at high purity.
Spectral Quality benzene (99+%) was obtained from Matheson,
39
Figure 1a. Spray-on cell for Malaker refrigerator.
a. Copper base which is attached to cold finger.
b. 1” Kovar seal
c. Spray-on nozzle.
d. 1/8" Kovar seal.
e. 1/16" thindwalled stainless steel tubing.
f. Silver soldered.
Figure 1b. Condensation cell for Malaker refrigerator.
a. Copper base which is attached to cold finger.
b. 1/2" Kovar seal.
c. Condensation chamber in center of cell.
d. 1/16" thindwalled stainless steel tubing.
e. Silver soldered.
40
Figure 1a.
it: e
Figure 1b.
41
Coleman and Bell and used without further purification.
The benzene was drawn into the second head assembly, Figure
1b, by vacuum distillation and then rapidly frozen and
taken to 30° K. The appearance of the sample and the
spectrum of the lattice region both indicated sufficient
shattering of the crystal to assure complete polycrystalline
behavior. Research Grade C2H4 (99.98%) was obtained from
The Matheson Company. The C2D4 (99+%) was obtained from
Merck, Sharp and Dohme of Canada. Both of these samples
were also used without further purification. Of the two
sampling techniques used for the ethylene study, the spray-
on technique was favored over the techniques of growing
samples from the melt for two reasons. The first reason
was the ease of handling of the sample by this method.
Second, resolution was very good with both techniques and
since mixed crystal work required the use of the gas deposi-
tion technique to minimize the possibility of cluster forma-
tion of the guest molecule, pure samples were also studied
this way for comparative purposes. In no case were impuri-
ties evident in the spectra.
CHAPTER IV
BENZENE
A. Introduction
Benzene lattice motions have been studied many times
both experimentally39"5 and theoretically.25r43'45 Its
crystal structure has been determined by X-ray diffraction
at 270°K by Cox47 and by neutron diffraction at 218°K and
138°K by Bacon, 32 31,43 It is because the structure and
dynamics of the crystal are so well defined that benzene
represents a good crystalline system with which to examine
the accuracy of the model calculations of torsional lattice
spectra. Here we will use only one of the most recent sets
of atom-atom interaction parameters determined by Williams.49
The evaluation of several sets of atom-atom interaction
parameters is left for the next chapter.
B. Crystal Structure and Symmetry Considerations
There is only one known phase of solid benzene. It
has orthorhombic symmetry and the crystal space group is
Pbca (D;g). The crystal lattice parameters and molecular
orientation for the two temperatures studied by Bacon, g£_al§3
are given in Table I. The listed direction cosines are
between the x, y, 2 molecule fixed coordinates (where z
42
43
Table I. Lattice parameters and molecular orientations
for crystalline benzene.
Molecular Orientation Lattice
2700K x y z Constants
a -0.2783 -0.6476 0.7094 7.460
b 0.9691 -0.1674 0.2238 9.666
c -0.0262 0.7434 0.6684 7.034
1380K x y z
a -0.3233 -0.6551 0.6830 7.39
b 0.9457 -0.1978 0.2580 9.42
c -0.0339 0.7293 0.6834 6.81
44
is chosen perpendicular to the plane of the molecule, x
lies along the carbon-hydrogen bond which is most closely
aligned with the b crystallographic axis, and y is
placed to complete an orthogonal right handed system) and
the a,b,c crystallographic axes.
In crystalline benzene there are four molecules per
unit cell, located on inversion centers at the corner and
face centers of the unit cell. These molecules are related
to each other by a series of screw axes parallel to the
crystal axes and glide planes perpendicular to these axes.
The direction cosines refer to the molecule located at the
origin, (0,0,0), and the set of all molecules equivalent to
it by pure lattice translation. This set of equivalent
molecules will be labeled set 1. The remaining molecules
at (%u %u 0), (%u 0, %0, and (0, %, $0 and the associated
sets of translationally equivalent molecules will be labeled
sets 2, 3, and 4 respectively. The molecules of set 1 are
related to those of sets 2, 3, and 4 by screw rotations
parallel to the a, b, and c crystallographic axes respec-
tively. Benzene has three rotational degrees of freedom.
Since there are four molecules per unit cell, there will be
twelve librational degrees of freedom. When these are
classified in terms of the factor group Dzh’ it is found
that there are three librational motions belonging to each
, and B3 .
of the irreducible representations Ag. B . B2 9
19 9
All twelve of these librational motions are permitted to
show'activity in the Raman according to selection rules,
45
Table II. Correlation table for crystalline benzene
Molecular Site Factor
pOint group D6h group Ci group D2h Activ1ty
Alg
2 2 2
(R2) A29 A9 x , y , z
B B xy
1 1
9 Ag 9
Bzg——' B29 xz
(RX.Ry) E1g Bag yz
E29
A1u
(T2) Azu Au inactive
Biu Biu X
Au
Bzu Bzu y
(TX.TY) Elu B3u z
E
311
46
since they transform according to the polarizability tensor
elements. The correlation table is shown in Table II along
with the transformation properties of the factor group and
the symmetry of the external motions of the D611 molecular
point group.
Symmetry coordinates were set up using linear combina-
tions of molecule based internal rotational coordinates.
These were right hand rotations about the molecule fixed
axes, x,y,z. These are denoted by xa(£/k) where the sub-
script a refers to the axis of rotation, x,y,z, and the
coordinate refers to the kth molecule in the 2th unit cell.
Since for the optically active lattice motions the mole-
cules in all the unit cells are in phase, the internal co-
ordinates are rewritten in the form
hams) = N’1/2 >3 )xa(£/k) (33)
where N is the number of unit cells in the crystal. Using
the screw rotations, linear combinations of these coordinates
may be constructed which have irreducible symmetry in the
factor group. The symmetry coordinates are listed in Table
III.
C. Interatomic Potential
In aromatic hydrocarbons there are three types of atom-
atom interactions; carbon-carbon, carbon-hydrogen, and
hydrogen-hydrogen. Each type displays an interaction po-
tential of the form
47
Table III. Internal symmetry coordinates for benzene.
Coordinate Symmetry
8.. = é-[xx<1> - xx(2>.- xx(3> + xx<4)1 Ag
51. = é- [5(1) + 5(2) - 5(3) - >3,(4)1
815 =§nz(1) - 5(2) + x203) - xz(4)1
Sm=%[kfl)+MW)-MQ)-MMH mg
Sn=%[§H)-§@)-§W)+5MH
$18 =%Hz(1) + xz(2) + )‘z(3) + xz(4)]
[xx(1) - x (2) + x (3) - xx(4)] 1329
[x (1) + A (2) + xy(3) + xy(4)]
m
N
o
II
hflH bflH~hflH
[gu>+xm>-xm)-xmn
Sm=%[MU)+&@)+MB)+&MN %9
s23 =%{>\1(1>- 5(2) + V3) - Vim
324 = é-[kz(1) + xz(2) - xz(3) - xz(4)]
48
)-D eXp(-C-Rij). (84)
where Rij is the interatomic distance and A, B, C, and
D are constants. If C = 0 and D = 12, then equation (84)
reduces to the Lennard-Jones [6-12] potential function.
For C ¥ 0 and D = 0, it becomes the Buckingham potential
function. From among the many sets of atom-atom interaction
potential parameters available in the literature, we have
chosen one of Williams most recent sets.49 Many other sets
of potential parameters will be tested in the next chapter.
The atom-atom interaction parameters used for the benzene
calculations are given in Table IV. Williams' parameters
were derived by fitting the A and B constants in equa-
tion (84) to permit agreement between observed and calculated
crystalline prOperties such as structure, elasticity con-
stants, compressibilities, and sublimation energies. The
magnitude of the exponents were taken from a calculation of
the interplanar spacing and compressability of graphite for
c--°C and a quantum mechanical calculation of the repulsion
of two hydrogen molecules for H---H. The C°"H eXponent
was taken as the mean of these two values. The potential
parameters have been modified and improved with the accumula-
tion of more data since they were first published. It is
because of this continued improvement and the amount of data
to which these parameters have been fit that this set of
potential constants was selected for the test of the model
calculations of Raman lattice spectra.
49
Table IV. Atom-atom interatomic potential parameters.
Interaction Parameters
type A B C D
H-H 1.74 1.704 3.74 0
C-H 9.68 7.703 3.67 0
C-C 53.90 74.620 3.60 0
A, units of 10-12 ergs 86; B, units of 10-10 ergs; C, units
of 3-1.
D. Results and Discussion
A polycrystalline benzene sample was prepared by
rapidly cooling the melt through the freezing point. The
crystal was examined using the 5145 R argon ion laser line
with a power output limited to approximately 100 milliwatts
to minimize localized heating effects. The sample was main-
tained at at approximately 400K. Spectrometer slits were
set at 15 microns, giving a resolution of approximately
0.25 cm-l. All reported frequencies are accurate to within
i1 cm_1. The spectra of the Raman lattice region of benzene
is shown in Figure 2.
Although twelve librational lattice modes are predicted
according to selection rules, only six are observed in the
present work. Ito and Shigeoka42 have measured the Raman
spectrum of benzene and fully deuterated benzene at various
temperatures down to that of liquid helium. They have
50
.&oov um coammu wofiuumH mnu Ca ocmucmn mcflaamummno mo Esuuommm cmEmm
.N wusmflm
51
"Is.
0
-§
‘}
~8
AllSNBlNI
Figure 2.
52
observed six bands at liquid helium temperature. (Dows
and coworkers have recently reported as many as eight,45
including two coincident bands.) In addition, Ito and
Shigeoka have extrapolated values for Raman librational fre-
quencies at 1380K. Since our calculated frequencies are
based on the 138°K lattice constants and molecular orienta-
tion, these extrapolated values permit a direct comparison
between experimental and calculated librational frequencies.
The calculated and experimental frequencies from this work
and the experimental 4°K and extrapolated 138°K frequencies
from Ito and Shigeoka are presented in Table V. The correla-
tion between the extrapolated and the calculated frequencies
in Table V is quite satisfactory
The calculation of librational intensities was under-
taken as described in Chapter II with one change. Since
we assumed oriented gas behavior for crystalline benzene,
the reaction field tensor was set equal to zero for the calcu-
lation. For the evaluation of the derivatives of the ele-
ments of the polarizability tensor with respect to the nor-
mal coordinates, eigenvectors obtained from the frequency
calculation were used. The values used for the molecular
polarizability tensor components of benzene were not im-
portant since only relative intensities are calculated, and
these are dependent only upon the relative differences of
the polarizability along axes perpendicular to the axis of
rotation. Thus one can choose arbitrary values for the di-
agonal elements of the molecular polarizability tensor,
53
Table V. Calculated and experimental frequencies for the
librational lattice motions of benzene.
Present Work Itoif’
Symmetry Calculated Experimental 4°K 138°K
(1380K) (400K)
Ag 95 98 (100) (90)
80 84 86 79
73 63 64 57
Blg 122 134 136 128
92 -- <86) <79)
77 -- 69 61
B29 107 -- (100) (90)
97 98 100 90
90 -- 86 79
B39 123 134 136 128
103 105 107 100
77 68 (86) (79)
. . . -1
FrequenCies are given in cm .
The experimental values in
brackets are suggested frequencies for transitions which are
forecast to be of low intensity in Frfihling's calculation,
ref. 41.
54
restricted by the symmetry requirement that the polarizabil-
ities along the two perpendicular axes in the molecular
plane are the same and differ from the polarizability along
the axis perpendicular to the molecular plane. The calcu-
lated frequencies and their respective relative intensities
are tabulated in Table VI along with their corresponding
experimental values and the irreducible symmetries accord-
ing to which they transform.
The calculated relative intensities show good agreement
with the experimental values within the restrictions of the
calculation. First, the calculation was restricted to
several temperatures where structural data are presently
available. The observed spectrum to which the calculated
spectrum was compared was obtained at approximately 400K.
No structural data is available for crystalline benzene at
this temperature. Small changes in the crystal structure
could change the form of the eigenvectors and hence the
intensities which correspond to them. Also, the inclusion
of the reaction field tensor could change the form of the
unit cell polarizability tensor. This would change the
polarizability tensor derivatives and thus change the in-
tensities corresponding to the various normal coordinates.
The reaction field tensor was not included because oriented
gas behavior was assumed. However Dunmur50 has shown that
there is significant interaction of molecules within con-
densed phases of several aromatic hydrocarbons, so as to
alter the values of their molecular polarizabilities by as
55
Table VI. Comparison of model Raman lattice frequencies
and intensities with experimental values for
crystalline benzene.
S mmetr Calculation (1380K) Experiment (400K)
y y Frequencies Intensities Frequencies Intensities
33 123 .296
g (.597) 134 .74
B 122 .301
19
329 107 .010
B 103 .049 105 .03
39
B2 97 .017
g (.189) 98 .33
A 95 .172
9
B1g 92 .001
B 90 .044
29
Ag 80 1.000 84 1.00
B1 77 .007
9 (.057) 68 .03
B 77 .050
39
Ag 73 .254 63 .36
O O -1 I O O O C
FrequenCies are in cm . All intenSities are given relative
to the most intense band. Values in brackets indicate
total calculated intensity which would fall within the
experimental band width.
56
much as ten percent of their gas phase values. However, no
data were given for benzene. Finally, in the crystal,
benzene is slightly deformed from its regular hexagonal
structure.47 Because of this, the low-lying E2g molec-
ular vibration can mix with the lattice motions and change
the form of eigenvectors. The degree of mixing induced
by this deformation is not known, and therefore cannot be
accounted for.
The use of calculated relative intensities along with
calculated frequencies allows the assignment of symmetries
to bands observed in polycrystalline samples where polari-
zation measurements are not possible. Our calculations
offer a basis from which we are able to make corrections in
the symmetry assignments of Ito and Shigeoka"2 and also in
the predicted band intensities calculated by Friihling.41
Our results indicate that all the Ag modes show appreciable
intensity and that none of the B39 modes are intensity
forbidden in disagreement with Frfihling's calculations.
Frfihling predicted the high frequency Ag mode and the low
frequency Bag mode to have low intensity. His predictions
were based on the assumption that these two bands, and two
others which we calculate to be extremely weak are rotations
purely about the axis perpendicular to the molecular plane.
Our analysis indicates that this is not true for any of the
librational motions. On the basis of Frfihling's calcula-
tions Ito and Shigeoka assigned B1 symmetry to the band
9
which they observed at 69 cm-l. Our calculations indicated
that this Big band is very weak and that the band
57
observed at 69 cm-1 by them and at 68 cm_1 by us has Bag
symmetry. They also indicate that the majority of the
intensity of the 98 cm"1
band arises from the Ag mode,
not the Egg as Ito and Shigeoka suggest.
In conclusion, model lattice calculations give a satis-
factory account of the librational lattice motion of crys-
talline benzene and also aid in assigning band symmetries.
The atom—atom interaction parameters chosen do not appear
to have caused any difficulties in the calculations. There
is some theoretical justification for subdividing the re-
pulsive interaction into a sum of atom-atom repulsions.51
However, the use of an atom-atom interaction to explain
dispersive forces is questionable, especially for highly
delocalized aromatic systems. Nevertheless, as a purely
empirical function the atom-atom interaction is very useful.
CHAPTER V
ETHYLENE
A. Introduction
Within the past decade much work has been done on
understanding interatomic potentials and their effect on
the dynamic motions within crystals. During these years
a large number of experimental and theoretical studies have
been carried out on lattice vibrations of molecular solids.
The initial thrust of much of this work was made with or-
ganic crystals such as ethylene,24o52 benzene,25o53 hexa-
methyl-tetramine,54 naphthalene,25:53v55 and antracene.54t55
In all these cases, except ethylene, the crystal structure
was known and the solids were easy to handle. More recently
interest has shifted to simpler molecular systems, most
generally solids of linear molecules, in an attempt to get
quantitative as well as qualitative agreement between ex-
perimental data and theoretical calculations.3°:31 Of the
organic systems, ethylene still stimulates a great deal of
interest because it is one of the simplest non-linear sys-
tems available.
However, ethylene also poses a perplexing problem to
lattice dynamicists. Lattice parameters and carbon positions
58
59
were determined in an X-ray investigation by Bunn,56 and
Halford and Brecher57 using polarized infrared light to
study a single crystal, determined solid ethylene to be
monoclinic, with Space group ng. Dows24 subsequently
pointed out that there are still two possible orientations
of the two molecules in the primitive unit cell consistent
with the C2; space group. In one case the two molecules
are related by a screw axis parallel to the a crystallo-
graphic axis, the "a-structure". In the other case they
are related by a screw axis parallel to the b crystallo-
graphic axis, the "b-structure". Much experimental and
theoretical work has been done to determine the correct
molecular orientation.22:23v24 These studies all indicate
that the "b-structure" more satisfactorily fits the data.
However, none offers a high degree of assurance that it is
the correct structure.
The present investigation utilizing Raman spectroscopy
was initiated in an effort to solve the problem of the
crystal structure, and to study the effect of the choice of
intermolecular force fields on lattice dynamic calculations.
Ethylene offers several advantages in regard to the second
part: (i) Atom-atom interatomic potentials have been ex-
tensively studied for carbon-hydrogen systems. For this
reason, there is a wide choice of potential parameters de-
termined by various methods to be examined.“3v53'63 (ii)
The lowest frequency intramolecular motion appears at 810
cm.1 and therefore coupling between internal and external
60
motions is expected to be negligible. (iii) The crystal
possesses an inversion center so that translational and
librational lattice motions can be treated separately.
B. Crystal Structure and Symmetry Considerations
As with benzene, there is only one known phase of
solid ethylene. This phase has been determined to be mono-
clinic with either P21/n11 (ng) or P121/n1 (C22) space
group. These two Space groups differ only in the direction
of the principal axis of symmetry. We shall refer to them
as the "a-structure" and the "b-structure" respectively, to
indicate the direction of the two-fold screw-axis (and the
normal to the glide plane) which is left in the resultant
crystal. Using X-ray powder diffraction, Bunn56 determined
the lattice parameters and the carbon positions as they ap-
pear projected on the ab-plane at 980K. Bunn made the as-
sumption of an orthorhombic structure with a Pnnm (Dzfiz)
space group, based on the lattice parameters and carbon
positions. Dows,24 using a purely repulsive potential,
later corrected the structure by determining energy minima
with respect to the hydrogen positions while maintaining
Bunn's carbon positions. He found that there were two mini-
mum energy orientations corresponding to rotations of the
two molecules per unit cell about their carbon—carbon axes,
first in the same direction and second in opposite direc-
tions. An examination of the earlier investigations re-
veals that all restricted the carbon-carbon bond to remain
61
in the ab—plane. We find that irrespective of potential
function, the true energy minimum consistent with monoclinic
symmetry is obtained only if the carbon-carbon bond is free
to rotate out of the ab-plane, within the restriction of
the symmetry elements of the Space group. Bunn did not ob-
serve this rotation because he was only looking at the pro-
jection on the ab-plane. The lattice parameters determined
at 980K by Bunn and the molecular orientations determined
by Bunn, by Dows (for the two monoclinic structures), and
by us (for the two monoclinic structures) are reported in
Table VII. The molecular orientations are defined in terms
of direction cosines as explained in Chapter IV. Here x
was chosen along the carbon-carbon bond, 2 was chosen per—
pendicular to the molecular plane, and y was chosen to
complete a right handed coordinate system.
In crystalline ethylene there are two molecules per
unit cell, located on inversion centers at the corners and
center of the unit cell. These molecules are related by
screw axes parallel to the a axis or the b axis and
glide planes perpendicular to these axes for the "a—struc-
ture" and the "b-structure" respectively. The direction
cosines refer to the molecule at the origin, (0,0,0), and
the set of molecules equivalent to it by pure lattice trans-
lations. The remaining molecules at (%, én %) and transla-
tionally equivalent to it form a second set of molecules.
These two sets of molecules will be referred to as sets 1
and 2, reSpectively.
62
Table VII. Lattice parameters and molecular orientations
for the two possible crystal structures of
solid ethylene.
Bunn56 Molecular Orientations Lattice
980K x y z Constantsa
a .809 -0588 O 4.87
b .588 .809 0 6.46
C 0 0 1.0 4.14
Molecular Orientations
"a-structure" "b-structure“
Dows24 x y z x y z
a .809 -.525 .265 .809 -.515 .283
b .588 .722 -.365 .588 .709 -.389
c 0 .451 .892 0 .482 .876
1128:“ x y z x y z
a .782 -.592 .194 .825 -.532 .190
b .614 .679 -.403 .542 .652 -.530
c .106 .434 .894 .158 .541 .826
aAll unit cell angles were observed to be 90°.
63
Ethylene has three rotational degrees of freedom,
three translational degrees of freedom, and twelve internal
degrees of freedom. Since there are two molecules per unit
cell on Sites of inversion symmetry, only the intramolecular
modes which are Raman active in the gas phase will be Raman
active in the solid. Also, each of these Raman active in-
ternal modes may appear as doublets consisting of an A9
and a B9 symmetry component. Among the external modes
only the rotational degrees of freedom will be active in
the Raman spectrum of the crystal. Since there are two
molecules per unit cell we expect six Raman active libra-
tional motions, three each of Ag and B9 symmetry. All
the remaining intra- and intermolecular motions in the
solid are infrared active. The correlation table for ethyl-
ene is given in Table VIII.
Symmetry coordinates were set up for the librational
motions using linear combinations of the moleculeébased
internal rotational coordinates. These were right handed
rotations about the molecule-fixed axes, x,y,z. These are
denoted by xa(£/k) where a = x,y,z and the coordinate
refers to the kth molecule of the 3th unit cell. The zero
wave vector coordinates are of the form
x (k) = N’l/Z z A (z/k), (85)
a 2 a
where N is the number of unit cells in the crystal. The
symmetry coordinates are listed in Table IX. Symmetry co-
ordinates were not set up for the internal motions since
Table VIII.
64
Correlation table for solid ethylene.
Molecular Point Site Factor
h Group Ci Group Czh
Group D2
V1,V2,V3
stV6
V8
V4
v9,v10
ViiaViz
(internal modes)
(external mode)
1B
39
1Blu
31.1
19 3A9
18 3A <
29 g 39
9
3A
113 3A < u
211 u
1B 3Bu
’20
w
IR(1-acous.)
IR(Z-acous.)
R and IR denote Raman and infrared activity respectively.
65
Table IX. Internal symmetry coordinates for the librational
lattice motions of ethylene.
Coordinate Symmetry
"a-structure"
s7 (2)-1/2[>\X(1) + xx(2)] Ag
s. <2>"/2ny<1> - 51(2)]
s. (2>"/2[xz<1> - xz<2>1
310 (2)-1/2HX(1) - xx(2)1 139
Sn (2)“/2ny(1) + 5(2)]
s12 (2)-”nigh + xz<2n
"b-structure"
s7 <2>"/mx<1) - xx<2n Ag
S. (2)-1/2My(1) + 3(2)]
s. - (2)"/2[xz<1> - 42(2)]
s1. <2)’1/2[x2<1) + xx(2)1 39
sh <2>“/2[xy<1) - xy<2>1
S12 (2)-1/2[7\z(1) + 02(2)]
66
no normal coordinate calculations were made on them.
C. Experimental Results
In all cases polycrystalline samples of either pure or
mixed crystals were prepared by annealling the solid films
formed upon gas deposition. The crystals, maintained at
approximately 300K, were examined with the 4880 R and the
5145 R argon ion laser lines using a power output of approx-
imately 1 watt. Spectrometer slit settings of from 15
microns (0.25 cm_1) to 150 microns (2.5 cm-l) were used.
The reported frequencies are accurate to i 1 cm-l.
The Raman spectra of polycrystalline ethylene and
ethylene—d4 in both the intra- and intermolecular regions,
and the mixed crystal spectra of 4% C2D4 in C2H4 and of 4%
C2H4 in C2D4 at approximately 30°K were obtained. The
spectral observations are entirely consistent with the C2;
space group with two molecules per primitive unit cell
located on sites of Ci symmetry. The Six Raman active
modes in the gas phase, 3Ag, 2Blg, and 1B2g' give rise to
six Raman active doublets, an A9 and a B9 component for
each gas phase mode. Also there are six Raman active lat-
tice modes, 3A9 and 3B9: librations corresponding to
rotations in the gas phase.
The factor group components of all six Raman active
modes observed in the intramolecular region for C2H4 in the
present study are depicted in Figure 3. The relative in-
tensities are correct as shown with the exception of the v6
67
.ocwamcum oflaow mo cofimou HmHsomHOEmHucH map No muuowmm cmfimm
.m oudmflm
68
.m muomflm
onom 00.0..” L. (9.0m comm ..r 00.0—
new.
JEN..:(/\|JE
09.: man. 0mg 59
q q u q
n is...
N\
nmo mac
5?
69
mode at 1222 cm-1, which should be reduced by a factor of
twenty. The v6 mode was not observed by Blumenfeld23 in
the solid, probably because of its low intensity. It does
agree with the weak absorption in the infrared assigned to
it by Jacox.65 It was possible for Jacox to observe this
absorption because of the highly powdered nature of her
sample and the selection rule breakdown which accompanies
it. Several of Blumenfeld's choices for factor group com-
ponents have been corrected on the basis of our observed
values. Also C-13 isotopic shifts were Observed for v2,
v5, and v8. For v3, C-13 probably lies under the 1328 cm-1
component. The band at 1660 cm.1 is probably due to 2v10,
This assignment is supported by the assignment by Halford
and Brecher of the 828 cm-1 infrared absorption of crystal—
line ethylene as the V10 ethylene motion.57
Five of the C2H4 fundamentals were observed in the
C2D4 matrix. The V, band was not seen because of its low
intensity. With the exception of the C-H stretching
motions, all the factor group split components of C2H4 are
very nearly split symmetrically about the isolated peaks.
These modes also show very little in the way of site and
factor group shifts from the gas phase values. The C-H
stretching modes on the other hand show large site shifts.
The matrix isolated spectra of CzH4 are shown in Figure 4.
The additional peak at 3003.9 cm.1 has been assigned as an
overtone of the C=C stretching motion of the C2D4 host
lattice at 1508.8 cm-l, in Fermi resonance with the
70
isolated C2H4 v1 stretching mode. Such an assignment is
unprecedented. However, the use of cubic terms in the
anharmonicity calculation indicates that the overtone should
be observed at 3002.: 2 cm-l. Further, such an overtone
has the correct symmetry for interaction. Finally, the
peak is not present in either of the pure crystal studies.
The observed frequencies and relative intensities of the
guest C2H4 in C2D4 and of the pure C2H4 solid state bands
are reported in Table X.
In C2D4, as for C2H4, all the factor group components
of the Raman active modes have been observed in the intra-
molecular region. See Figure 5. C-13 isotopic Shifts were
observed for v1, v2, v5, and v8. Again the C-13 band for
v3 is probably hidden under the 972 cm“1 component. The
relative peak intensities and splittings are Similar to
those observed in C2H4, although in all cases the split-
tings of the factor group components in C2D4 are less than
those in C2H4. The relative intensities are correct as
shown in Figure 5 with the exception that the v2 peak has
been reduced by one half.
All the C2D4 peaks were observed when perdeuteroethyl-
ene was isolated in C2H4 matrix at a mole ratio of 1:25.
Within experimental error, all the factor group components
are Split symmetrically about the isolated peaks shown in
Figure 6. The site Shifts are small in all cases. C-13
isotopic shifts are seen for v1, v2, and v3. The experi—
mental results for the C2D4 Spectra are listed in Table XI.
Table X.
Observed frequencies and relative intensities of
solid and matrix isolated ethylene.
Crystal
Matrix Isolated
I Imax cm-1 I Imax
2995.5
1596.7d .011
v2 (A ) c-C 1614.6 .386 1618.0 .199
9 1621.7 .003
. 1328.5 .578 1330.5 .022
”3 (Ag) CH2 S°l° 1348.0 .234 1337.5 .719
a 3059.4d 009
v5 (B ) asym. C-H 3103 3065.9 '399 3068.0 .217
19 3068.3 '
1222.4 .003
v6 (B19) CH2 rock. 1222 1226.7 .004 —-— --
936.5d .008
941.6 .137 944.9 .154
V8 (Bzg) CH2 wag.
951.5 .020
aObserved frequencies taken from W. L. Smith and I. M. Mills,
J. Chem. Phys. 49, 2095 (1964).
bObserved frequency taken from B. L. Crawford, Jr., E. Lan-
caster,
and R. G. Inskeep, J. Chem. Phys.
21, 678 (1953).
CObserved frequency taken from T. Feldman, J. Romanko, and
H. L. Welsh, Can. J. Phys. 34, 737 (1956).
dThese frequencies are attributed to C-13 isotopic shifts.
72
QVN
GOSH
owumaomfl Amvv vmno mo mcoflumunfl> accumucfl may mo manommm cmEmm
.v musmfim
73
.v wusmflm
nkom 000m. . no on meow nmo— o5—
9;
‘r~
- Ns
JE J
m) H). N\.
03: man—
nno nmo
t m>
74
.volmcwfihzuo Uflaom mo commou HMHSUOHOEMHUGA 0:» mo muuommm cmfimm
.m musmflm
75
.m musmflm
o—MN mmNN : OWNN 8mm . F 99 mm!
a q J~ . 7 JM 4 4
, s
m). H) N). ‘
.
_
oak CNN
76
.vmuu Ca Amvv vvlwcmamsuw
woumHomfl xfluuma mo coflmmu HmHsomHoamuucH mnu mo muuommm cmemm
.o onamwm
.m ousmflm
77
o—mm comm nnmu mmmm .p 9n— nav—
up
.1) . . as a
E 4
Table XI.
78
solid and matrix isolated ethylene—d4.
Observed frequencies and relative intensities of
Gas Crystal Matrix Isolated
cm-1 cm.1 I Imax cm-1 I/Imax
2236.1b .016 2237.2b .017
v1(A ) sym. C-H 2260 2244.5
9 2246.0 .687 2245.7 .723
1487.4b .022 1487 9 019
V2(Ag) 9‘9 1518 ig2§33 .857 1510.1 1.000
. 972.7 .352
v3(Ag) CH2 SCi 985 987.2 .197 980.9 .525
2290.6b .023
v5(B ) asym. C-H 2310 2302.0 2303.3 .455
1g 2305.5 1.000
996.9 .015
v6(319) CH2 rock. 1011 1002.1 .033 999.9 .040
b
773.7 .009 b
v8(B2 ) CH2 wag. 785 778.7 .290 781.3 °gg§
9 784.1 .030 ° '
aObserved frequencies taken from W. J. Lehman, J. Mol. Spectry.
1, 1 (1961).
b
These frequencies are attributed to C-13 isotopic shifts.
79
All the lattice modes have been observed for both
ethylene and ethylene-d4. The relative intensities of the
lattice modes are similar in the two cases. The ratio of
the frequencies between corresponding bands in the two iso-
topes indicates that all bands are due to librations, as
was eXpected. Because of the small moment of inertia about
the C-C axis, one would expect librations about this
axis to have the largest frequency and to Show large Shifts
on deuteration. They should also be weak because of the
small difference in polarizability components perpendicular
to the C-C axis. This is seen to be the case for the
weak, high frequency pair of lattice bands although the fre-
quency ratio is not 1.41 as would be expected for librations
purely about the C-C axis. Figure 7 contains the spectra
of the lattice region. The observed lattice frequencies for
ethylene and ethylene-d4 and the calculated ratios of fre—
quencies upon isotopic substitution are listed in Table XII.
D. Theoretical
For the calculations of the crystal lattice energy,
lattice constants, and force constants, zeroth, first, and
second derivatives of the potential with respect to the
proper parameters were required. For these calculations,
ten different sets of semi-empirical atom-atom potential
functions of the type
U.. = —A(R.
13 lj )-D eXP(-C-Ri.) (85)
J
80
Figure 7. Raman spectra of the lattice region of ethylene
and ethylene—d4, at 300K. The peak marked with
an asterisk is a laser fluorescence line.
81
1
I
‘75
l
l
160
VXu
I
I
>=meb2~
1
ISO
CM"
1
l
100
Figure 7.
82
Table XII. Observed Raman lattice frequencies and calcu-
lated frequency ratios for ethylene and
ethylene-d4.
Lattice Vibrations
Ethylene Ethylene-d4 v v
Vobs(cm-1) vobs(cm-1) C2H4/ C2D4
Raman 73 60 1.22
90 75 1.20
97 78 1.24
114 95 1.20
167 123 1.36
177 135 1.31
Infrareda - — _
73 69.5 1.05
110 104 1.06
aInfrared frequencies obtained from M. Brith and A. Ron,
J. Chem. Phys. 59, 3053 (1969).
83
were used. A, B, C, and D are selected constants and Rij
is the distance between the i and j nonbonded atoms.
As mentioned previously, for C = 0 and D = 12 the po-
tential reduces to the Lennard-Jones [6—12] potential. For
C # 0 and D = 0 it becomes the Buckingham potential.
The values of A, B, C, and D chosen for calculations in
this work for the hydrogen-hydrogen, carbon-hydrogen, and
carbon-carbon nonbonded interactions are listed in Table
XIII.
The constants A, B, C, and D of the sets of potential
functions I, II, III, and IV were obtained from hetero-
geneous data on gas and solid phases of various molecules.
See the references in Table XIII for more details. The
constants of sets V through X were obtained by Williams
by using a least-squares method applied to the results of a
systematic investigation of several series of crystalline
hydrocarbons. The least-squares method has been described
in Chapter IV. Parameter sets V and VIII are derived
from a least-squares fit applied to data on aromatic and
non-aromatic hydrocarbons. Sets VI and IX are derived
from the same combined data for the aromatic crystalline
hydrocarbons, and sets VII and X are derived from data
on non-aromatic crystalline hydrocarbons only.
The calculations of the librational lattice frequencies,
eigenvectors, and lattice parameters were made using the
Wilson GF method16 as discussed in Chapter II. Summations
were carried to six angstroms, giving approximately 80% of
84
00 0 00.0 0.00005 0.000 0 50.0 0.0000H 0.00H 0 V5.0 0.005H 5.00 X
0v 0 00.0 0.000V5 0.000 0 50.0 0.0055 0.00 0 v5.0 0.v05u v.5H XH
00 0 00.0 0.005H5 0.0H0 0 50.0 0.0000 0.HHH 0 v5.0 0.H5H0 v.00 HHH>
00 0 00.0 0.000H0 0.000 0 50.0 0.000ufi 0.00H 0 V5.0 0.0000 0.00 HH>
00 0 00.0 0.00vv5 0.000 0 50.0 0.0va 0.00H 0 V5.0 0.000v 0.00 H>
00 0 00.0 0.00000 0.000 0 50.0 0.0550 0.00H 0 v5.0 0.0000 0.50 >
fi0 0 00.0 0.00000 0.000 0 0H.v 0.0000v 0.¢0H 0 00.v 0.0000v 0.50 >H
00 0H 0 0.000H00 0.500 0 v0.0 0.0H5vv 0.¢0fi 0 00.0 0.0000 00.0w HHH
00 0 00.0 0.000500 0.500 0 00.v 0.000H0 H.HOH 0 00.v 0.0000 00.00 HH
00 0a 0 0.000000 0.050 0H 0 0.00000 0.00H 0H 0 0.0000n 0.H0 H
D U m < D U m d D U m 0
.000 010 01m Elm umm
.mcwamnum mafiaamummuo
0o moflemcmv mofluuma 0:0 mcflumasoamo CH wows mHmHucmuom cofluomumucfi EouMIEou<
.HHHN wHQmB
85
the static lattice energy.62 Minimum energy orientations
and force constants were calculated for each of the two pos-
sible structures with each of the ten sets of potential
functions. For each of these cases a set of frequencies,
eigenvectors, and mean amplitudes of libration was calcu-
lated. Further, these eigenvectors were used to calculate
the relative intensities of these lattice modes according
to the method described in Chapter II. The oriented gas
model assumption was made. Therefore the reaction field
tensor was not included in the calculations of the polariz—
ability element derivatives.
E. Theoretical Results
The energy was minimized with respect to the molecular
orientation by allowing the two molecules per unit cell to
rotate within the restrictions imposed by the symmetry of
the space group. The final orientation of the molecules is
explained in terms of rotations of the molecules from a
position with the molecular plane in the ac-plane and the
carbon-carbon bond parallel to the a crystallographic axis.
The three angles of rotation, ¢ , B , and 9, taken in that
order, represent right-handed rotations about axes parallel
to the a, b, and c crystallographic axes. Space group
symmetry requires that both the B and 6 rotations for
the two molecules in the unit cell be equal in magnitude
but opposite in Sign and the 0 rotations be equal in sign
and magnitude for the "a-structure". For the "b—Structure"
86
the 0 and 9 rotations are equal in Sign but opposite in
magnitude, while the B rotations are equal in Sign and
magnitude. The list of rotations corresponding to energy
minima for the ten sets of potential functions, along with
the calculated static lattice energies, are given in Table
XIV. Only the rotation angles for set 1 molecules are given.
These rotations can be correlated with the molecular orienta-
tions given in Table VII by substituting the angles of
rotation into the orientation matrix
cos Bcose cos esin Bsin 0 sin Bcos ¢cos 9
-cos ¢sin 9 +sin ¢sin 9
cos Bsine Sin ¢sin Bsin 9 cos ¢sin Bsin 6 (86)
+cos ¢cos 6 -sin ¢sin 6
-sin B sin ¢COS B cos 0cos B .
The variations of the ¢, B, and 9 rotations with the
choice of potential set are greater for the "a-structure"
than for the "b-structure", (in degrees, 2.6 gs. 1.0, 18.6
gs, 1.2, and 2.2 Kg, 1.8 respectively). This indicates a
greater anisotropy of force on the "a—structure" unit cell
than on the "b-structure" unit cell. More important is the
fact that regardless of the choice of potential function,
the calculated energy for the "b-Structure" is less than
that of the "a-structure". However this only offers a
qualitative indication that the "b-structure" is better.
Knowledge of the frequency behavior of the zone center lat-
tice librations and their relative intensities is needed to
87
Table XIV. Minimum energy rotations and static lattice
energies for the two possible structures of
solid ethylene.
Minimum Energy
Potential Struc- Orientations Energya
Set ture m B 9 (Kcal/Mole)
I a 26.1 12.5 40.1 -4.441
II a 24.9 12.1 39.7 -4.021
III a 25.9 10.3 38.1 -3.833
IV a 25.1 12.1 38.5 -4.816
V a 26.1 -3.9 40.3 -3.398
VI a 24.3 10.5 38.3 -3.535
VII a 26.9 -3.9 40.1 -3 .290
VIII a 26.5 -3.9 40.3 -3.070'
IX a 26.7 -3 .9 40.1 -2.823
X a 25.9 -6.1 38.1 -2.963
I b 33.6 -9.1 32.5 -4.491
II b 33.0 -8.7 31.9 -4.124
III b 33.2 -9.1 32.1 -3.950
IV b 34.0 -9.7 33.3 -4.885
V b 33.4 -8.7 32.3 -3.664
VI b 32 .2 -8.5 31.5 -3.813
VII b 33.8 -9.1 21.7 -3.546
VIII b 33.4 -8.9 32.5 -3.284
IX b 33.6 -8.9 32.9 -2.998
X b 33.2 -9.1 33.3 -3.149
aStatic lattice energies are determined by summing atom-atom
interactions out to six angstroms. Williams has shown that
this gives a value of approximately 80% of the true value.62
For comparison, 80% of the sum of the latent heats of fusion
and vaporization (-4.037) is -3.230 Kcal/mole.
88
provide a more sensitive test, since higher derivatives are
utilized.
The static crystal energy was also minimized with re-
spect to the crystal unit cell dimensions. The small
variations of the unit cell dimensions for the "b-structure"
case indicate that the lattice parameters Obtained at 980K
should be substantially independent of temperature. Further,
this indicates that the lattice frequencies should show
little variation with temperature. Thus a comparison of
frequencies observed at 30°K and frequencies calculated
using the 980K unit cell parameters should be a reliable
method of determining which of the potential set is best.
The frequencies calculated using each of the sets of
interatomic potentials are listed in Table XV, along with
the Observed values. The use of interatomic parameter sets
I, II, III, and IV to calculate the force constants for
librational motions lead to imaginary frequencies in sever-
al cases. Their failure in calculating librational fre-
quencies is due to their "softness“. They overemphasize
the role of the attractive terms at small intermolecular
distances. For this reason, they will not be considered
further. The calculations using the remaining sets of
interatomic parameters give frequencies which vary consider-
ably in their ability to fit the observed values. The "b-
structure" frequencies Show a Slight improvement over the
"a-structure" frequencies in reproducing the observed fre-
quencies. However, the improvement is not good enough to
89
Table XV. Observed and calculated librational frequencies
for solid ethylene.
Struc-
- -1
Set ture Calculated FrequenCies (cm )
a 88 94 99 111 239 245
I
b imag 21 52 68 116 137
a 97 102 105 117 236 247
II
b (several imaginary frequencies)
a 96 101 111 133 247 263
III
b 37 68 96 99 164 173
a 105 110 125 135 259 269
IV
b imag 61 96 97 130 155
a 75 97 101 158 252 260
V
b 81 98 120 128 200 202
a 142 144 148 174 316 337
VI
b 97 116 144 150 249 252
a 80 103 106 162 247 257
VII
b 88 101 126 134 200 200
a 72 91 94 145 226 234
VIII
b 77 91 111 120 181 182
a 70 87 89 136 206 214
IX
b 75 87 105 114 167 169
a 79 94 99 146 200 224
X
b 82 90 110 124 173 173
Obs. 73 90 97 114 167 177
90
offer any assurance that the "b-structure" is correct.
The relative intensities of Raman scattering from the
librational motions were calculated using the eigenvectors
obtained from the frequency calculation and the diagonal
components of the molecular polarizability tensor calculated
by J. F. Harrison66 using uncoupled Hartree-Fock perturba-
tion theory. A description of the quantum mechanical
calculation has been published.67 The diagonal components
of the tensor are given in Table XVI.
Table XVI. Molecular polarizability tensor components for
ethylene.
Tensor a
Component Value
QXX 9.387
4.894
0£in
322 4.842
aAll val es were calculated by Harrison66 and have the
units ( )3.
For the discussion of the results of the intensity
calculation, we will refer only to values calculated using
the eigenvectors generating from set X potential parameters.
This can be done without loss of generality because the in-
tensity calculations are more or less independent of the
choice of potential function. The calculation of relative
librational intensities is dependent on three factors: (1)
91
the molecular polarizability tensor components, (2) the
molecular orientations, and (3) the eigenvectors. Since
the molecular polarizability tensor will remain unchanged,
it will not effect a difference in the relative intensities
calculated with different potential functions. The varia-
tion of the molecular orientations with the choice of
potential function is quite small and for the "b-structure"
is effectively negligible as can be seen in Table XIV.
Thus it will have very little if any effect on the calcu-
lated relative intensities Obtained with different potential
functions. Finally, the eigenvectors show a high degree of
reproducibility, irrespective of the choice of potential
function. With the exception of the first four sets of
interaction parameters, which do not consistently produce
real frequencies, the eigenvectors for the normal modes ob-
tained using the various potential functions are Shown in
Table XVII. Variations are apparent. However, large devia-
tions do not exist. It appears that for a self-consistent
set of interatomic potential parameters, eigenvectors, like
structure, show a high degree of stability.
The calculated frequencies, relative intensities, and
mean amplitudes of libration are given in Table XVIII.
Goodings30 has shown that a librational amplitude of 26 de-
grees in solid nitrogen gives an anharmonicity of approxi-
mately seventy percent. However, we feel that librations
with amplitudes of less than ten degrees should Show an
acceptable low anharmonicity for our calculations. The
Table XVII.
Eigenvectors for A
92
and B
9
motions for
both possible ethylene structures, along with
their corresponding observed frequencies.
AgMotions Frequency Bg Motions
87 S10 S12 58 S9 S11
"a—structure"
.535 -0024 -0012 .532 0027 0024
.538 .007 -.014 177 .533 .033 -.008
.535 .024 ‘0015 .532 .029 .024
.535 -.024 -.014 .532 .029 .025
.535 .024 -.015 167 .532 .029 .024
.534 .027 -.018 .535 .022 .016
0063 .198 .105 -0083 .124 .199
-.030 .197 -.114 97 -.019 .099 .218
.068 .199 .104 -.084 .129 .194
.066 .196 0110 -0085 .126 .197
.067 .193 .116 114 -.084 .125 .198
.076 .204 .090 -.060 .124 .201
.001 .097 .219 .024 -.182 .139
.020 .102 .216 73 .080 -.196 .108
.006 -.096 .219 .026 -.178 .145
.001 -.101 .217 .026 -.181 .141
.002 -.107 .214 90 .027 -.181 .140
.013 -.084 .226 .024 -.183 .138
Cont.
93
Table XVII. (Continued)
Ag Motions Frequency B Motions
58 S9 S12 S7 810 S11
"b-structure"
.537 -0019 “”0005 .539 -0004 -0004
.537 -.019 -.003 167 .539 -.004 -.002
.536 -0021 -0010 .539 -.004 -0009
.537 -0020 -0008 .539 -0004 -0007
.536 -.018 -.018 .538 -.005 -.018
0046 0221 -0012 0011 0053 .236
.046 .221 -.015 97 .007 .054 .236
.049 .220 -.021 .022 .055 .236
0047 0221 “.017 .018 .053 .236
.044 .221 -.013 114 .020 .054 .236
.039 .220 ‘0024 .041 .053 .236
.014 .010 .243 -.007 -.215 .058
.083 .014 .243 9° -.008 -.215 .059
.027 .019 .242 -.005 -.215 .060
0022 .015 0243 -0007 -0216 .059
.026 .011 .243 73 -.005 -.215 .060
.045 .020 .242 -.003 -.216 .059
These eigenvectors were generated from the last six sets of
interaction parameters listed in Table XIII.
The frequencies
indicate the observed frequency to which each eigenvector
belongs.
94
Table XVIII. Calculated and Observed frequencies and rela-
tive intensities and calculated mean amplitudes
of libration for the two possible ethylene
crystal structures.
Frequency (Amplitude) Relative Intensity
Obs. Calc. "a" Calc. "b" Obs. Calc. Calc.
Ila N "b N
177 224(10.9) 173(12.4) .002 .006 .001
167 200(11.5) 173(12.4) .010 .001 .006
114 146(5.6) 124(6.1) .197 .057 .177
97 99(7.6) 110(6.5) .418 .535 .628
90 94(7.2) 90(8.1) 1.000 .619 1.000
73 79(8.8) 82(8.6) .190 1.000 .314
I
111'“,
I'll-I'll:
95
comparison of the observed values with the calculated fre-
quencies and their relative intensities for both structures
gives strong support to the "b-Structure". The inability of
the "a-structure" calculations to predict the correct
ordering of the intensities, along with the prediction that
one of the strong low frequency librations Should be in-
tensity forbidden, indicate that the "a-structure" cannot
be the correct structure. In addition the ordering and
magnitudes of the calculated "b-Structure" intensites are no
more than fifteen percent in error. This is completely
within the limits of accuracy expected. The limitations on
the accuracy of the relative intensities include the accur-
acy with which the molecular polarizability tensor is known,
as well as the several approximations used in classical
lattice dynamical calculations. Presently there are no ex-
perimental values for the molecular polarizability compo-
nents with which to compare the theoretical values used.
Also, Dunmur5° has shown that perturbations do indeed exist
in solids and can alter gas phase molecular polarizabili-
ties by as much as ten percent. These perturbations are
neglected in our oriented gas approach. Finally, calcula-
tions of eigenvectors may be as much as five percent in
error due to the form of the interaction potential function.
Similar results have been obtained for ethylene-d4.
In this case the electronic structure of the molecule and
of the crystal will not be changed from the ethylene case
results. Therefore the molecular orientations and static
96
lattice energies will be the same as those reported in Table
XIV. Only potential sets VIII, IX, and K were used to cal-
culate the ethylene—d4 frequencies, which are compared to
the observed values in Table XIX. As was noted for ethylene,
the "b-Structure" frequencies are in slightly better agree—
ment with the observed values than are the "a—structure" fre—
quencies. However, the improvement is not Significant. Rel-
ative intensities were caluclated as for ethylene. The cal-
culations were based on the eigenvectors generated with set
IX interaction parameters, and the results compared to ex-
periment in Table XX. Again the calculations for the "b-
structure" provide strong evidence that this molecular orien-
tation represents the correct crystal structure. The "a-
structure" calculations cannot predict the proper ordering
or relative intensities of the observed spectra. The con-
clusions drawn from the ethylene-d4 calculations agree with
those drawn from the ethylene calculations, that the "b-
structure" must be the correct one.
In conclusion, the calculations at all levels indicate
the "b-structure" to be correct. Within the limitations of
the model, the intensity calculations indicate with a high
degree of assurance that the "a—structure" cannot satisfac-
torily predict the Observed spectra. Further we have indi-
cated that intensity calculations offer a powerful tool for
the determination not only of symmetry assignments, but in-
deed of crystal structure where an ambiguity remains from
diffraction experiments. This is particularly significant
since, unlike energy and frequency calculations, the rela-
tive intensity predictions are effectively independent of
the choice of intermolecular potential.
97
Table XIX. Observed and calculated frequencies for solid
ethylene-d4.
Struc- . -1
Set ture Calculated FrequenCies (cm )
a 61 75 78 122 161 166
VIII
b 63 77 91 102 128 129
a 59 72 74 114 147 152
IX
b 62 74 86 98 118 119 5
i
a 67 77 82 122 144 159 i
X _
b 70 80 92 108 124 124 9
!.
Obs. 60 75 78 95 123 135
Table XX. Calculated and Observed frequencies and relative
intensities and calculated mean amplitudes of
libration for the two possible ethylene-d4 crys-
tal structures.
Frequency (Amplitude) Relative Intensity
Obs. Calc. "a" Calc. "b" Obs. Calc. Calc.
"a" "b.
135 152(9.4) 119(10.7) .013 .011 .001
123 147(9.6) 118(10.8) .082 .003 .010
95 114(5.3) 98(5.8) .198 .027 ‘.195
78 74(7.9) 86(6.2) .504 .621 .609
75 72(7.0) 74(7.9) 1.000 .951 1.000
60 59(9.2) 62(9.0) .170 1.000 .316
REFERENCES
11.
12.
13.
14.
REFERENCES
G. Venkataraman and V. C. Sahni, Rev. Mod. Phys. 42,
409 (1970).
D. F. Hornig, J. Chem. Phys. 16,1063 (1948), H.
Winston and R. S. Halford, J. Chem. Phys. 17, 607
(1949).
O. SchnePP. Advances in Atomic and Molecular Physics,
Vol. 5, Ed. D. R. Bates and I. Estermann, (Academic
Press Inc., New York, 1969).
S. Bhagavantum and T. Venkatarayudu, Proc. Indian
Acad. Sci. 9A, 224 (1939).
W. Vedder and D. F. Hornig, Advances in Spectroscopy,
Vol. .)II, Ed. H. W. Thompson,‘(Interscience, New York,
1959
D. A. Dows, Physics and Chemistry of the Organic Solid
State, Vols. I and II, Ed. D. Fox, M. M. Labes and A.
Weissberger, (Interscience, New'York, 1963 and 1965).
D. A. Hatzenbuhler, Ph.D. thesis, Michigan State Uni-
versity, 1970.
M. Suzuki, T. Yokoyama, and M. Ito, Spectrochim. Acta
24A, 1091 (1968).
M. Brith and A. Ron, J. Chem. Phys. 59, 3053 (1969).
D. Boal and G. A. Ozin, Spectroscopy Letters 4(3&4),
43 (1971).
L. Andrews and J. I. Raymond, J. Chem. Phys. 55, 3078
(1971).
L. Andrews and J. I. Raymond, J. Chem. Phys. 55,3087
(1971).
D. W. Smith and L. Andrews, J. Chem. Phys. 55,5295
(1971).
R. B. Kurzel, J. I. Steinfeld, D. A. Hatzenbuhler, and
G. E. Leroi, J. Chem. Phys. 55, 4822 (1971).
98
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
99
T. Shimanouchi, M. Tsuboi, and T. Miyazama, J. Chem.
Phys. éé, 1597 (1961).
E. B. Wilson, J. C. Decius, and P. C. Cross, Molecular
Vibrations, (McGraw-Hill Book Company, Inc., New York,
1955).
S. H. Walmsley and J. A. Pople, Mol. Phys. 8, 345
(1964).
J. C. Raich and R. D. Etters, J. Chem. Phys. 55, 3901
(1971). ,
O. Schnepp, J. Chem. Phys. 46, 3983 (1967).
J. E. Cahill,ONR Contract N00014-68-A-0109-0003, NR 014-
203, Technical Report No. 7(32), (1968).
P. M. Richardson, Ph.D. thesis, University of Pennsyl-
vania, 1968.
G. Taddei and E. Giglio, J. Chem. Phys. 53, 2768 (1970).
S. M. Blumenfeld, S. Paddi Reddy, and H. L. Welsh,
Can. J. Phys. gg, 513 (1970).
D. A. Dows, J. Chem. Phys. 36, 2836 (1962).
I. Harada and T. Shimanouchi, J. Chem. Phys. 44, 2016
(1966).
Y. A. Schwartz, A. Ron, and S. Kimmel, J. Chem. Phys.
22. 99 (1971).
P. M. Richardson and E. R. Nixon, J. Chem. Phys. 49,
4276 (1968).
J. C. Laufer and G. E. Leroi, J. Chem. Phys. 55, 993
(1971).
T. Shinoda and H. Enokido, J. Phys. Soc. Japan 26, 1353
1969 .
D. A. Goodings and M. Henkelman, Can. J. Phys. 23,
2898 (1971).
M. Suzuki and O. SchnePP. J. Chem. Phys. 55, 5349
(1971).
G. C. Pimentel, J. Chem. Phys. 12, 1536 (1951).
s. H. Walmsley, J. Chem. Phys. 48, 1438 (1968).
R. S. Mulliken, J. Chem. Phys. 2, 14 (1939).
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
100
J. P. Col a and J. A. A. Ketelaar, Mol. Phys. 1, 14,
343 (1959 ; J. A. A. Ketelaar and R. P. H.
Rettschnick, ibid. Z, 191 (1964).
z. J. Kiss, H. P. Gush, and H. L. Welsh, Can. J. Phys.
62, 362 (1959).
R. H. Hester, Raman Spectroscopy, Vol. I, Ed. H. A.
Szymanski, (Plenum Press, New York, 1965).
R. E. Miller, K. L. Trueil, R. R. Getty, and G. E.
Leroi, ONR Contract N00014-68-A-0109-0003, NR 014-203,
Technical Report No. 4(29), (1968).
S. C. Sirkar, Indian J. Phys. 66, 189 (1936).
I. Ichishima, J. Chem. Soc. Japan 16, 391 (1949).
A. Frfihling, J. Chem. Phys. 36, 1119 (1950); Ann.
Phys.-6, 401 (1951).
M. Ito and T. Shigeoka, Spectrochim. Acta 66, 1029
(1966).
N. Rich and D. A. Dows,.Mol. Phys. 6, 111 (1968).
I. Harada and T. Shimanouchi, J. Chem. Phys. 26,
2708 (1967).
W. D. Ellenson, M. Nicol, L. Hsu, and D. A. Dows,
Twenty-seventh Symposium on Molecular Structure and
Spectroscopy. Columbus, Ohio, 1972, Paper BB-2.
D. A. Oliver and S. H. Walmsley, Mol. Phys. 11, 617
(1967).
E. G. Cox, Rev. Mod. Phys. 66, 159 (1958).
C. E. Bacon, N. A. Curry, and S. A. Wilson, Proc. Roy.
Soc. (London) A279I 98 (1964).
D. E. Williams, Transactions of the American Crystal-
lographic Association 6, 21 (1970).
D. A. Dunmur, M01. Phys. 66, 109 (1972).
D. P. Craig, R. Mason, P. Pauling, and D. P. Santry,
Proc. Roy. Soc. (London) A286, 98 (1965).
D. A. Dows, J. Chem. Phys. 66, 2833 ( 962).
T. Shimanouchi and I. Harada, J. Chem. Phys. 31, 2651
(1964).
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
101
W. Cochran and G. S. Pawley, Proc. Roy. Soc. (London)
A280, 1 (1964). .
G. S. Pawley, Phys. Stat. Sol. 66, 347 (1967).
C. W. Bunn, Trans. Faraday Soc. 36, 23 (1944).
C. Brecher and R. S. Halford, J. Chem. Phys. 66, 1109
(1961).
N. G. Parsonage and R. C. Pemberton, Trans. Faraday
Soc. 66, 311 (1961).
A. M. Liquori, E. Giglio, and L. Mazzarella, Nuovo
Cimento B55, 476 (1968).
L. Bartell, J. Chem. Phys. 66, 827 (1960).
A. I. Kitaigorodskii, J. Chem. Phys. 66, 6 (1966).
D. E. Williams, J. Chem. Phys. 36, 3770 (1966).
D. E. Williams, J. Chem. Phys. 47, 4680 (1967).
B. L. Crawford, J. E. Lancaster, and R. G. Inskeep,
J. Chem. Phys. 61, 678 (1953).
M. E. Jacox, J. Chem. Phys. 66, 140 (1962).
J. F. Harrison, private communication.
J. F. Harrison, J. Chem. Phys. 32, 3321 (1968).
LIB
((nitfififlraflfiflnfljng(wiim(s
"'TITJ'ITMEHHM