THERMAL EXPANSION AND ISOTHERMAL
COMPRESSIBILITY OF SOLID NITROGEN AND METHANE
DAVID CRAIG HEBERLEIN
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I would like to express my appreciation to the follow-
ing persons for their contributions towards the completion
S of this work:
Dr. E. D. Adams, the Chairman of my Supervisory
Committee, who assisted in the design and construction of
the apparatus and guided the course of this work;
Dr. T. A. Scott for his helpful suggestions in the
study of solid nitrogen;
Dr. J. S. Rosenshein for his helpful suggestions in
the study of solid methane;
Dr. J. W. Philp for his help in using the computer to
analyze the data taken in these experiments;
Doctors J. R. Gonano, P. N. Henriksen, M. F. Panczyck,
R. A. Scribner, and G. C. Straty, who all have rendered
valuable assistance to me; and,
Mr. B. McDowell, who provided valuable technical
assistance and produced the liquid helium necessary to
run the experiments.
I would like to thank my parents, Mr. and Mrs. F. A.
Heberlein, for their continued help and encouragement
during my undergraduate and graduate study. I also would
like to thank my aunt and uncle, Dr. and Mrs. R. L. Fairing,
for their help and encouragement during my graduate study.
I am grateful to my wife, Martha, for her understanding
and encouragement during a long, and,at times, frustrating,
TABLE OF CONTENTS
LIST OF FIGURES .
. . . . . iii
. . . . . vi
I. INTRODUCTION . . . . .
Nitrogen . . . . . .
Methane . . . . . .
II. THEORY . . . . . . .
Nitrogen . . . . . .
Methane . . . . . .
III. APPARATUS AND PROCEDURE . .
Cryostat . . . . . .
Gas Handling and Pressure System
Sample Chamber and Pressure Bomb
Temperature Measurements, Calibration
and Regulation . . . . . .
Procedure and Sample Measurements. .
IV. EXPERIMENTAL RESULTS AND DISCUSSION.
Nitrogen . . . . . . .
Methane . . . . . . .
REFERENCES. . . . . . . . .
BIOGRAPHICAL SKETCH . . . . . .
LIST OF FIGURES
1. Phase diagram of solid nitrogen. . . . 2
2. Phase diagram of solid methane . . . 6
3. Schematic drawing of apparatus . . .. 20
4. Pressure system. .. . . . . ... 25
5. Sample chamber and pressure bomb . . .. .31
6. Schematic drawing of three-terminal
resistance bridge. . . . . . .. 37
7. Specific heat versus temperature
for solid nitrogen . . . . . . 47
8. Relative length changes versus
temperature for solid nitrogen . . .. .48
9. Semilogarithmic plot of relative length
changes versus inverse temperature for
alpha nitrogen . . . . . . . 50
10. Linear expansion coefficient versus
temperature for solid nitrogen . . .. 52
11. Measured and calculated compressibilities
for solid nitrogen . . . . . ... 56
12. Gruneisen parameter versus temperature
for alpha nitrogen . . . . . .. .58
13. Specific heat versus temperature
for solid methane. . . . . ... 60
14. Relative length changes versus
temperature for solid methane. . . .. 61
15. Linear expansion coefficient versus
temperature for solid methane. . . . 63
16. Time versus temperature for a typical
warming curve of solid methane . . .. .65
17. Specific heat (C') versus temperature
for solid methane . . . . . . . 66
18. Specific heat versus temperature for
solid methane at low temperatures . . .. .68
19. Relative length changes versus
temperature for solid methane at
low temperatures. . . . . . . . 70
Solid nitrogen exists in three distinct phases. The
phase diagram as determined by Swenson1 is shown in Fig. 1.
The gamma phase found by Swenson in 1959 exists at low
temperatures and high pressures. Very little about the
gamma phase is known.
From its triple point to 35.6 K, nitrogen in its beta
phase consists of diatomic molecules arranged in a hexagonal-
close-packed lattice. The space group P63/mmc was determined
from x-ray studies by Boltz et al.2 in 1959. Molecular
rotation in the beta phase was confirmed from x-ray studies
by Jordan et al.3 and Streib et al. and from studies of the
infrared spectrum by Smith et al.5 The quenching of the
quadrupole coupling in beta nitrogen can only be explained
by either spherical rotation as found in liquid nitrogen or
by hindered rotation with the molecular axis inclined at an
angle of 54.7. Because the specific heat is found to have
a value 50% greater than that needed for free rotation and
the molecular volume is only 75% of that expected for freely
rotating molecules, the rotations are not free. Thus, there
is a hindered rotation in which the molecular axis makes an
angle of 54.70 with the crystal axis.
Phase diagram of solid nitrogen.
Pressure versus temperature.
10 20. 30 40 50 60
At 35.6 K, solid nitrogen at its vapor pressure under-
goes a first-order phase transition with a latent heat of
54.7 cal/gm-mole.67 The lower temperature, alpha phase
has a face-centered-cubic lattice in which the centers of
each molecule are displaced slightly from the lattice
points. X-ray measurements by Jordan et al.3 showed the
space group to be P213. Because alpha nitrogen has a rather
open structure, it is reasonable to assume that the nuclear
quadrupole coupling constant, e2qQ/h, differs very little
from its value in the free molecule. Using pure quadrupole.
resonance continuous wave techniques, Scott8 determined the
temperature dependence of e2qQ/h. The result of this study
and subsequent studies by DeReggi found the resonance
frequency to be extremely temperature dependent near the
alpha-beta phase transition.
Although most theories of solids are given for constant
volume processes, it is easier experimentally in most cases
to make measurements at constant pressure. In order to
distinguish the temperature dependence of a quantity such
as the quadrupole resonance frequency, which is measured at
constant pressure, from that caused by the expansion of the
lattice, the equation of state of the solid must be known.
To determine the equation of state, the temperature dependence
of the expansion coefficient and the isothermal compressi-
bility must be known.
Until recently, little experimental information other
than the specific heat was available on the thermal properties
of solid nitrogen. In 1966, Manzhelii, Tolkachev, and
Voitovichl0 (MTV) measured the expansion coefficient of
solid nitrogen from 22 to 44 K. In an attempt to extend
these measurements to 4.2K, the expansion coefficient was
measured from 4.2 to 38 K in this work. Recently, Bezuglyi,
Tarasenko, and Ivanov (BTI) determined the adiabatic
compressibility from 16 to 44 K from their measurements of
the velocity of sound in solid nitrogen. The isothermal
compressibility, KT, is simply related to the adiabatic
compressibility, KS, by
K = K + (2TvC1 (1)
T S P
where c is the expansion coefficient, T the temperature,
v the molar volume, and Cp the specific heat at constant
pressure. Thus, the isothermal compressibility of solid
nitrogen was measured in this work from 8 to 40 K as a
consistency check not only on the measurements of BTI,
but also as a check on the temperature dependence of the
The existence of a anomaly in the specific heat of
solid methane at 20.4 K was first observed in 1929 by
Clusius.6 Because the maximum in the specific heat appeared
to be finite, the transition was assumed to be of second-
order. X-ray diffraction studies by Schallamach12 showed
no change in the face-centered-cubic structure of solid
methane at 20.4 K, with the lattice constants differing
by approximately 1% on opposite sides of the transition.
This small change in the molecular lattice indicates that
the thermal anomaly involves a change in the orientational
ordering of the molecules of the lattice.
Pauling originally described the transition as
involving a sudden change from molecular oscillations in the
low temperature phase to free molecular rotation in the
higher temperature phase. This was consistent with the
approximate value of 3 cal/mole for the specific heat above
the transition. Proton spin resonance experiments by
Thomas et al. determined that this picture of the transi-
tion was not correct. The resonance line width observed by
Thomas et al. indicated no perceptible change at 20.4 K, and
only above 65 K did the line width and characteristic time
for spin-lattice relaxation both drop toward the values
observed in liquid methane. Studies of the Raman bands at
79 K by Crawfordl5 showed rotational wings of the same size
as those found in liquid methane. Thus, while the hindrance
to molecular rotation is small above 65 K, the phase change in
solid methane at 20.4 K takes place without the establishment
of free molecular rotation. Although the passage to free
rotation does occur, it occurs gradually at higher tempera-
tures, and without the appearance of a thermal anomaly.
The phase diagram of solid methane is shown in Fig. 2.
There are four solid phases in methane. The alpha phase has
been extensively studied and the alpha-beta and alpha-gamma
phase boundaries have been determined from the specific heat
measurements of methane under hydrostatic pressure by
0. 10 20. 30 40 50 60
Fig. 2. Phase diagram of solid methane.
Pressure versus temperature.
0 -- I -- u -- L- ---------
0 02.304 06
Rosenshein6 and from the p-v isotherms determined by
Stevensonl7 and Stewart.18 The delta phase was first
reported by Stevenson in 1957 and to date nothing more is
known about this phase. The beta-gamma phase boundary has
been observed to pressures as low as 200 atm. by Rosenshein,
or about 800 atm. lower than that observed by either
Stevenson or Stewart. Recent specific heat measurements
by Colwell, Gill, and Morrison19 (CGM) show a broad excess
heat capacity in solid methane at zero pressure centered
about 8 K. An extrapolation of the beta-gamma phase
boundary determined by Rosenshein to zero pressure gives
a transition temperature of 9.5 K. While the agreement
seems reasonable in view of the broadness of the lower
temperature transition, there exists no other experimental
evidence to extend the beta-gamma phase boundary to 8 K
at zero pressure.
With the exception of the early molar volume studies
of solid methane between 20 and 21 K reported by Heuse20
in 1936, virtually no information exists on the thermal
expansion of methane at or below 21 K. In this work, the
thermal expansion of methane was measured from 4.2 to
26 K not only to determine the lattice contribution to the
thermal expansion, but also to study the nature of the
thermal anomaly at 20.4 K and to determine if possible the
existence of a lower temperature phase at zero pressure.
In 1936 Pauling13 suggested that the transitions in
solid CH4, N2, 02, and CO2 were caused by a change from
oscillation to free rotation of the molecules. He approx-
imated the effect of the crystal field on the molecule
with the potential
V(9) = V (l cos29) (2)
in which 9 is the angle between a molecular axis and its
equilibrium position. This form is, however, a poor repre-
sentation of the dipole interaction which depends not only
on the relative orientation of the two dipoles, but also on
the direction of the vector separating their centers.
Furthermore, it has been shown by Kirkwood21 that a potential
of this form always leads to a prediction of a single
Kreiger and James22 showed that the interaction potential
V = Acos ij + Bcos29ij (3)
could lead to a single second-order transition, a single
first-order transition, two first-order transitions or a
second-order transition followed by a first-order transition
for increasing temperature. Both of these models are much
oversimplified in that they assume axial symmetry of the
molecules and they treat the interaction of the molecules
as dependent only on the orientations of the molecules in
space irrespective of the position of each molecule in
Kohin23 found the directional interaction potential for
solid nitrogen to be
V(9) = CIP2(cos9) (4)
where I is the average of P2(cosQi) for the neighbors of a
central molecule, 9 the angle that each molecule makes
between the molecular axis and the symmetry axis at its
lattice site, P2(cosO) = -(1/2) + (3/2)cos28, and C a func-
tion of the lattice constants and molecular parameters as
determined from nearest and next-nearest neighbor inter-
actions. Eq. (4) is a combination of the following terms:
a term due to the interaction of the quadrupole moments of
the molecules; a directional correction to the attractive
dispersion forces due to the anisotropic polarizabilities of
the molecules; and, a directional correction to the repulsive
If an assembly of classical rigid rotors with fixed
centers is taken as a model for a crystal of diatomic mole-
cules, the normalized rotational distribution function is
f(G,0) = exp(-V(9,0)/kT (5)
where f(89,)dw is the probability that the axis of the mole-
cule lies in the solid angle dw about (9,0). V(9,0) is the
orientational potential due to the crystal field. If one now
assumes that each molecule orients itself independently in
the average field of its neighbors and that the probability
distribution for molecular orientation is the same for each
lattice site, we can then require the rotational distribution
to be consistent with the crystal field to obtain a consistency
equation relating the temperature and the order parameter.
Since I has been defined as the assembly average of P2(cosS),
the classical consistency relation for I(T) becomes
I(T) = P2(9)exp(-CIP2(9)/kT)dw (6)
There are two solutions to Eq. (6). One solution corresponds
to rotational disorder, i.e. V(9,0) = 0 for all temperatures.
The other solution, corresponding to an ordered array, has an
extremum corresponding to a maximum in the temperature. The
calculated Tmax then corresponds to a theoretical upper bound
on the alpha-beta transition. Using a classical model similar
to this, Jansen and de Wette24 obtained a value for Tmax very
close to 35.6 K. Kohin pointed out, however, that when the
directional anisotropy of the intermolecular repulsive forces
are included as in Eq. (6), the calculated transition tempera-
ture becomes much higher.
By treating the diatomic nitrogen molecule as a rigid
rotor, Kohin solved the Schrtdinger equation to find the
eigenvalues for the directional potential given in Eq. (4).
By minimizing the directional energy of the crystal field,
the acceptable orientations of the molecules agree with
the observed structure for alpha nitrogen. Since the method
used by Kohin assumes that the molecules are stationary
and consequently near absolute zero, no attempt was made to
determine theoretically the structure for beta nitrogen.
Furthermore, it would be erroneous to attempt to extend this
theory to temperatures higher than 20 K without including
such effects as the lattice vibrations and direct correla-
tions between neighbors. Since there exists no microscopic
theory to explain the behavior of alpha nitrogen above 20 K,
we must turn to some phenomenological approaches based on
DeReggi, Canepa, and Scott found that the pure
quadrupole resonance frequency, Q', obeys an order parameter
relation of the form
= A[(Tc T)I1d, (7)
r d) = -l(T T) (8)
By plotting -Q (- (Q/T)-1 versus temperature, these authors
found that their experimental points gave a straight line
from approximately 20 K to the transition temperature.
Extrapolating this line to its intersection with the tempera-
ture axis gave a critical temperature Tc = 37.70 0.07 K.
From this behavior it appears that there exists a higher
order process in alpha nitrogen.- The process is interrupted
before completion by the first-order transition. A similar
behavior is found in solid hydrogen26 in which a change in
crystal structure interrupts the progress of a higher order
A phenomenological approach to explain the large rise
in heat capacity and the thermal expansion from 20 K to the
transition at 35.6 K has been proffered by MTV.10 A number
of solids behave in a similar manner as the melting point
is neared. The changes in the thermal properties other
than that which can be ascribed to the nominal expansion
of the lattice are explained by a rapid growth in the
number of vacancies. In analogy to the rapid growth in
the number of vacancies, MTV suggested that as the tempera-
ture rises there is a rapid growth in the number of
disoriented molecules in alpha nitrogen until the lattice
becomes unstable. At this temperature a phase transition
occurs, and a different crystal structure appears.
To explain the formation of these "orientational de-
fects", the free energy of the crystal would have to decrease.
In order for the free energy to decrease, the gain in
potential energy due to the new orientation about which the
molecule oscillates has to be exceeded by a gain in entropy
from the decrease in orientational order.
If it is assumed that the production of disoriented
molecules is accompanied by an increase in volume, then the
change in volume may be expressed as
Av = Kc (9)
where K is a constant of proportionality and c is the con-
centration of orientational defects. The concentration of
these orientational defects should obey an exponential law
of the form
c = Aexp(-U/RT) ,
where A is a degeneracy factor equal to the number of posi-
tions of disoriented molecules having the same energy, U
the activation energy to create the orientational defect, R
the natural gas constant, and T the temperature. Combining
Eq. (9) and Eq. (10), we find the relative change in length,
dl/1, to be given by
AL = Av = KAexp(-U/RT) (11)
1 3v 3v
or ln(Al/l) = ln(KA/3v) U/RT. (12)
By assuming that at some temperature, T1, the influence of
the orientational defects is negligibly small and that the
coefficient of thermal expansion connected with other
mechanisms remains well-behaved as the temperature is
raised above TI, then any major portion of the plot of
ln(l1/1) versus 1/T that gives a straight line can be said
to obey the predicted exponential behavior. From their
thermal expansion data, MTV found this behavior in the
interval 29.4-34.5 K using a T1 = 23 K. From the slope
of this line, MTV calculated an activation energy of
450 cal/mole. The difficulty with this approach is that
both the slope of ln(Al/l) versus 1/T and the interval
for which a straight line pertains depends on the choice of
T1. Ideally, Al would approach zero as T approaches 0 K,
but in reality the exponential behavior is only observed
for temperatures above 20 K. This approach is useful
because it is possible to determine in approximately which
temperature interval a mechanism with an exponential
temperature dependence begins to dominate the thermal
A similar approach by Bagatskii, Kucheryavy, Manzhelii,
and Popov27 (BKMP) to explain the excess heat capacity above
20 K gave an activation energy of 460 cal/mole. In view of
the approximations made by BKMP for the lattice contribution
to the specific heat and the inherent error in the deter-
mination of the activation energy by MTV, the agreement
appears fortuitous. The phenomenological approach of MTV
is of interest, however, because there appears to be
nothing in the potential found by Kohin to explain the
strong temperature dependence of the expansion coefficient,
the isothermal compressibility, the excess heat capacity,
or the pure quadrupole resonance frequency at temperatures
above 20 K.
Fortunately, methane with its high molecular symmetry
and relatively large intermolecular spacing is a favorable
case for studying orientational coupling. In methane the
ratio of the intramolecular distance to the molecular
separation is so small that one can expect the octopole-
octopole interaction between next neighbors to be dominant.
Keenan and James28 have shown for methane that the electro-
static interaction for non-overlapping molecules falls off
as rapidly as R-7 and by neglecting all interactions except
those between nearest neighbors in a crystal, they were able
to formulate a theory for solid methane in terms of a single
adjustable parameter, the effective octopole moment of the
In the model developed by Keenan and James, molecular
and lattice vibrations are neglected and the methane crystal
was treated as a face-centered-cubic array of spherical
rotors carrying a charge distribution with tetrahedral
symmetry. The statistical calculation is a classical ver-
sion of the self-consistent field approach in which the
conditions for self-consistency appear as a family of
integro-functional equations, one for each molecule in the
crystal. Neglect of quantum effects makes the results
applicable only to CD4. Three solutions of these equations
minimize the free energy in a particular temperature range.
At the lowest temperatures the stable phase has a tetragonal
symmetry with the molecules oscillating about equivalent
equilibrium orientations. As the temperature rises, the
crystal undergoes a first-order transition into a phase
with octahedral symmetry in which one molecule in four
rotates freely. As the temperature rises even higher the
crystal undergoes a second-order transformation to an
orientationally disordered phase. By assigning a value to
the molecular octopole moment to make the higher transition
temperature agree with the 27.4 K value observed in CD4, the
predicted lower transition temperature becomes 24.4 K as
opposed to the observed temperature of 22.2 K. The predictions
of this theory are in agreement with integrated heats of
transition, zero-point entropy and the optical properties
of all three phases.
An extension of this model to include solid CH4 has
recently been completed by Yamamoto and Kataoka.29 The
high and low temperature ordered phases are assumed to have
the same sub-lattice structure as that proposed by James
and Keenan, but differ in that all calculations were made
on the basis of quantum statistical mechanics in the sub-
space J = 4, where J is the rotational quantum number.
Since the various spin combinations of the hydrogens pro-
duce three separate spin species, A (meta), E (para), and
T (ortho), the nuclear spin species are treated separately.
Inasmuch as the high temperature equilibrium proportions
of A:E:T become frozen in at low temperatures, the results
for individual spin species can not be compared in any
other than a qualitative way with experimental measurements
on samples with "mixed" spin species. The most interesting
feature of this quantum mechanical treatment of solid
methane is the prediction that for a sample of spin species
A, two phase transitions, one at 20.7 K and the other at
16 K, are to be expected. This is to be compared with the
specific heat measurements on "mixed" crystals by CGM,
whose results showed transitions at 20.4 K and 8 K.
Thermodynamically the large specific heat anomaly in
methane has been described as a cooperative transition. To
describe what is meant by a cooperative transition, let us
assume for the moment that methane obeys a van der Waals
equation of state
(P + av2) (v b) = RT ,
in which the term av2 is a correction to the pressure to
allow for the attractions of the molecules, and b is a term
to allow for the fact that the molecules are of a finite
size. The term av-2 increases in importance as the volume
decreases, such that if the temperature is steadily
reduced it becomes thermodynamically advantageous for the
entire lattice to change abruptly in volume. The loss in
entropy from this change would be offset by the gain in
energy from the work done by the attractive forces. Thus
the presence of a term such as av-2 becomes increasingly
important with the progress of the change which it is
causing, and it is in this sense that the transition was
called cooperative by Fowler.30'31 Furthermore, the
presence of av-2 also points out that a transition of
this type can only be understood if some account of the
molecular interactions is included in the theory of
solids displaying this type of behavior.
The dominant interaction in solid methane has been shown
to be the octopole-octopole interaction which lead to a type
of orientational ordering of the methane molecules below
20.4 K. Before ordering can occur, the energy in excess of
that allowed for the new molecular orientation must first
be dissipated in the lattice. Consequently, there appears
a large change in the specific heat. The term cooperative
is applied to the specific heat anomaly in methane because
the transition is very broad.
III. APPARATUS AND PROCEDURE
In this section the design and construction of the
apparatus necessary to measure thermal expansion and
isothermal compressibility of heavy solidified gases will
be presented. A capacitance technique will be described
for measuring changes in the length of the sample corres-
ponding to changes in temperature or pressure.
The cryogenic considerations including appropriate
plumbing and electrical connections will be considered in
a section devoted to the cryostat. The attainment of
temperatures ranging from 4.2 to 90 K is discussed in
this section with particular emphasis on the control and
measurement of temperatures in the interval 52-77 K.
The sample gas handling and helium pressure systems
are discussed in the second section of this chapter.
Particular emphasis is placed on the purification of the
samples used and on the means of controlling the pressure
transmitted to the samples.
The third section of this chapter describes the sample
chamber and the pressure bomb. The design of the sample
chamber is dictated by the thermal properties of the samples
which were studied. The concomitant problems of the design
necessary for filling the sample chamber at temperatures
from 63 to 90 K and of making sample measurements from
4.2 to 40 K are discussed. The construction of a pressure
bomb which contained the fluid helium used to transmit
hydrostatic pressure to the sample is also presented in
A germanium resistance-thermometer was used to monitor
the temperature of the sample. The associated electronics
and plumbing used in measuring, calibrating, and regulating
temperatures in the interval 4.2-40 K are discussed in the
fourth section of this chapter.
The final section in this chapter describes the
experimental procedure used to form the sample and to
make measurements of the thermal properties of each sample.
The cryostat used to perform these experiments is
similar to that described by Walsh32 and is shown schemat-
ically in Fig. 3. The sample chamber and pressure bomb
were surrounded by two exchange chambers which in turn were
enclosed by two large stainless steel dewars. The outer
dewar was usually filled with liquid nitrogen. The required
sample temperatures determined the refrigerant to be used
in the inner dewar. Thus, liquid nitrogen was used as a
refrigerant to obtain temperatures from 52 to 90 K and
liquid helium was used as a refrigerant to attain temperatures
from 4.2 to 52 K. The refrigerant contained in the inner
dewar will be referred to hereafter as the main bath.
M C 0
* a) 04 >
H 1.4 14
*d Mu > >
4-4 U) M 0
0) 14 0)
vi H a 1a
) 0) 04
k C C
u) 4 C
e e r
C 3 z
H w 0) 0
E )d 0)
m O H
The outer exchange chamber consisted of a vacuum jacket
connected to the stem of the cryostat with a flange-type
seal using an indium gasket. The outer exchange chamber
was used to isolate the sample chamber from the main bath.
A cylindrical container was enclosed by the outer
vacuum jacket. This container had a volume of approximately
250 cm3 and could be filled with liquid from the main bath
by means of a modified "Hoke" valve. The liquid in this
container will be referred to hereafter as the inner bath.
A smaller vacuum jacket connected to the base of the
inner bath container with another flange-type seal formed
the inner exchange chamber. The inner chamber was used to
either isolate the sample from the inner bath, or, when
filled with exchange gas, to provide thermal contact
between the sample and the inner bath.
To form the solid samples,the entire sample system
was first cooled to 77 K by filling the main bath with
liquid nitrogen. The inner bath container was then filled
with liquid nitrogen and the temperature of the sample was
controlled by reducing the vapor pressure of the nitrogen.
Temperatures from 52 to 77 K could be maintained by pumping
the nitrogen with a model KC-46 Kinney pump. The sample
chamber and the inner bath had to be maintained at a
temperature below the triple point of the sample in order
to form a crystal. Therefore, to fill the sample chamber,
the sample within the filling capillary had to be heated
above its melting point. A 0.0225" O.D. stainless steel
capillary was run through a 0.25" O.D. stainless steel tube
which extended to the top of the cryostat. The filling
capillary was electrically grounded only at the top of the
sample chamber and was thermally isolated from the main
and inner baths, permitting the temperature of the entire
length of capillary to be raised by connecting a 6-volt
storage battery between the capillary and the cryostat
ground. The capillary to the pressure bomb was connected
in a similar manner such that, if it became blocked by
frozen impurities, moderate heating would establish
pressure transmission over the entire length of the
After forming the sample, the liquid nitrogen was
removed from the inner dewar. The sample was then cooled
to 4.2 K by transferring liquid helium into the inner
dewar. Two separate heating coils were wound on the
outside of the inner vacuum jacket. Using these heaters,
the entire inner chamber could be maintained at temperatures
much above that of the main bath. The larger heating
coil had a total resistance of 5000-ohms, and, when used
with a Heathkit Model PS-4 Regulated Power Supply, pro-
vided a course temperature control from 4.2 to 52 K. The
smaller coil had a total resistance of 1000-ohms and was
used as a fine control on the temperature.
Individual 0.059" O.D. stainless steel capillaries
run through separate 0.25 O.D. stainless steel tubes
which extended from the outer vacuum jacket flange to the
top of the cryostat formed coaxial lines which were used
for capacitance leads. The leads to the heaters and
auxiliary thermometers were introduced through the pumping
tube into the outer exchange chamber. These leads were
made from #36 Advance wire and were thermally anchored to
the inner bath container. Vacuum-tight glass-to-metal
seals were used to bring electrical leads through the inner
bath container into the inner vacuum chamber. The leads
into the inner exchange chamber for the germanium
resistance-thermometer consisted of two separate 0.059" O.D.
stainless steel capillaries introduced through the pumping
tube. A 4" section of #36 Advance wire was attached to both
of these stainless steel leads before thermally anchoring
them to the sample chamber.
Gas Handling and Pressure System
Purification of gas samples was necessary before
introduction into the sample and pressure systems. Because
impurities such as oxygen and nitrogen in methane could
pass through a cold trap held at a temperature near the
triple point of methane, a special process was used to
purify the methane samples.
C.P. Grade methane, 99.0% pure, was transferred into
a one liter cylinder and then immersed in liquid nitrogen.
This quick freezing of the methane produced a shattered
solid with a large surface area. Since the major
contaminants had a vapor pressure much higher than that of
methane at 77 K, the residual vapors could be pumped away
until the pressure over the solid was that appropriate to
the pure substance. This process was repeated until the
newly formed solid had the correct vapor pressure.
C.P. Grade nitrogen, 99.7% pure, was purified by
passing the gas sample through a liquid nitrogen cold
trap before it entered the U-tube. Similarly, the helium
gas used to pressurize the bomb was passed through a
helium cold trap before reaching the U-tube.
The gas handling system, shown schematically in
Fig. 4, consists of a sample line and a pressure line,
both interconnected to a mercury U-tube pressure system.
The pressure system was of standard U-tube design and
has been adequately described by Straty.33
In operation the sample was admitted into the U-tube
and subsequently into the filling capillary. The pressure
of the gas in the U-tube was then raised, creating a
pressure difference between the sample cell and the U-tube.
This pressure head produced rapid condensation of the gas
into the sample chamber. After forming the sample, the
valve between the filling capillary and the U-tube was
closed and the gas remaining in the U-tube was pumped away.
In order to measure the isothermal compressibilities
of samples, helium gas from a helium cylinder with a high
pressure regulator was then introduced into the U-tube and
the pressure capillary. Since the volume of the U-tube was
0) 0 -1
S4 l 0U
M C.) a)
C 4 U)
SCfl -H (
1 01% B3
approximately twenty times the dead volume of the pressure
bomb, the height of the mercury on the gas side of the
U-tube was used to vary the pressure of the gas trapped
in the bomb. To prevent mercury from entering the
hydraulic lines, a high-level alarm was installed on the
oil side of the U-tube. Because the entire U-tube assembly
was electrically insulated from its supporting structure,
the mercury itself was used as a switching device to
actuate the alarm.
A Texas Instruments pressure gauge was connected
through appropriate valving into the gas side of the U-tube
and was calibrated against a dead weight gauge connected
into the oil side of the U-tube. Special care was taken
to insure that the mercury levels in the U-tube were
equal when calibrating the Texas Instruments pressure
gauge reading against the dead weight gauge pressure.
Sample Chamber and Pressure Bomb
Two different approaches were available for measuring
thermal expansion and compressibility in solid nitrogen and
methane. One approach, which has been used successfully in
studies of solid helium,34 was to measure P = P(T) for
fixed molar volumes. The second approach was to measure
V = V(T) for fixed pressures.
When solid is formed in a sample chamber, the solid
also plugs the filling tube, isolating the chamber from
external pressure sensing devices. As the solid is cooled
below its melting point, the internal pressure also de-
creases. Assuming a Debye solid, this change in pressure
can be expressed as
ap = SCv-ldT =9DV-lfCvd(T/GD) (14)
where d is the Gruneisen parameter, v the molar volume,
and 8D the Debye temperature. Between 90 K, the melting
point of methane, and 4.2 K, the change in the internal
pressure of methane would be approximately 103 atm. It
therefore seemed unreasonable to attempt to design a
chamber capable of withstanding 103 atm. at 90 K and
yet retaining sufficient sensitivity to measure changes
of less than 1 atm. at 4.2 K.
Fortunately, the heavier solidified gases have a
coefficient of thermal expansion on the order of 10-4K-1
over most of the temperature interval below their melting
points. It was more reasonable to design a chamber whose
volume would be determined by the solid sample which it
Since the crystalline fields in both solid methane
and nitrogen are isotropic,29 it was possible to consider
geometries for the sample chamber in which only length
changes would be observed. By choosing a chamber with
cylindrical geometry, one end of the chamber could be
fixed, and the active end would then be used as the moving
plate of a three-terminal capacitor. Relative changes in
length would be directly related to changes in capacitance
as monitored by a General Radio type 1620-A capacitance
measuring assembly. A brass bellows with a small Hooke's
law constant was chosen as the sample chamber, both for
its radial strength and for its ability to distend elas-
tically over 20% of its equilibrium length.
In making measurements of changes in length, it was
necessary to have the bellows distended from its equilibrium
position by the solid which it surrounded. Since the solid
contracted uniformly as it was cooled, any solid caught in
the convolutions of the bellows would contract at the
same rate as the total length of the sample.
The change in volume upon melting in both methane and
nitrogen is approximately 10% of the total liquid volume.
To maximize the capacitive sensitivity, the distance be-
tween the moving and the fixed plate, referred to as the
gap length (1g), must be minimal. To minimize the gap
length in spite of the large volume change upon melting,
two further considerations were incorporated in the design
of the sample chamber.
The easiest means of filling the bellows with solid
was to sublimate the gas coming through the filling capil-
lary directly into the sample chamber. For sublimation to
occur, a thermal gradient along the length of the sample
chamber had to be maintained. Since the filling capillary
would necessarily have to be at or above the melting
temperature of the sample gas, the heated capillary was
introduced through the less massive, copper plug at the
active end of the bellows. The more massive, supporting
structure for the fixed end of the bellows was thermally
anchored to the inner bath. The sample formed in this
manner was not homogeneous.
Once the bellows was filled with solid, it was
desirable to form as homogeneous a sample as possible. To
form a good crystal meant that the material trapped in
the bellows would have to be melted and frozen again
very slowly. To allow for this large excursion in length,
the active end of the bellows was allowed to push a spring
loaded, "fixed" capacitor plate away from the stops in
its supporting structure. As the liquid began to solidify,
the active end of the bellows returned slowly to a position
such that the fixed plate was again pushed against the
stops in its supporting structure, and the two plates
were no longer shorted to one another.
To apply hydrostatic pressure directly to the sample,
the area both inside and outside the bellows had to be
filled with fluid helium. To surround the sample with
fluid helium, a pressure capillary was introduced through
the copper support assembly into the sample area. To
contain the pressurizing fluid outside the bellows, the
entire sample chamber was enclosed inside a pressure
The sample chamber, surrounded by the pressure bomb,
is shown in Fig. 5. Copper was used extensively through-
out the sample chamber for two reasons. Copper has a high
thermal conductivity, insuring reasonable thermal
Sample chamber and pressure bomb
Fixed capacitor plate
Active capacitor plate
Bellows sample chamber
Large hollow copper cylinder
Detachable copper piece
Copper support assembly
Stainless steel thermal standoff
Thermometer mounting assembly
Copper cage assembly
Small hollow copper cylinder
A .. :* * * ... ,* ,* ** :
.. --'"-, t' ;."" """'" ::"'.':" a:"" '"""
s* q.* **.. *P. *: .*. .*s.. .* : *
*. -... ... ; -, ,j ,*,
M' II W [' ; -
G 1 __S :I -
equilibrium times. Also the thermal expansion of copper
is not only small below 77 K, but is also well known.
The sample chamber consisted of a brass bellows,
approximately 0.50" in length, with an I.D. of 0.350"
and an O.D. of 0.592". The active end of the bellows was
soldered to a copper plug. A cylindrical capacitor plate
with a diameter of 0.500" was fastened with epoxy to the
opposite end of the copper plug. The epoxy electrically
insulated the capacitor plate from ground.
The fixed end of the bellows was soldered into a
detachable copper piece. At the opposite end of this
piece a hub was tapped to receive a plug extending from
the copper supporting structure. The plug, when covered
with 0.001" Teflon tape, seated on a 450 surface in the
bottom of the hub, thus served as the pressure seal to
the chamber. To insure that the solid was exposed only to
an active portion of the bellows, a hollow copper cylinder,
closed only at its top, was mounted on top of the threaded
copper plug. The diameter of the cylinder was 0.350" to
insure a mechanically tight, but not gas tight, seal to
the inside of the bellows. The length of the cylinder was
0.200", implying that the closed end of the cylinder pro-
truded 0.072" into the active portion of the bellows.
Inside the previously described hollow cylinder was
inserted a similar, but smaller, hollow cylinder with a
closed top. This cylinder covered the open end of the
pressure capillary which extended through the copper
support system into the sample chamber. A pressed indium
O-ring seal was made between the base of this hollow
cylinder and the top of the threaded plug. This seal was
vacuum tight and prevented the sample gas from entering
the pressure line when the chamber was being filled. Be-
cause this seal was easily broken when the pressure in the
pressure capillary exceeded two atmospheres, it was possible
to pressurize the inside of the bellows after it had been
filled with the solid sample.
The copper supporting structure not only held the
detachable copper piece, but also served as a support for
a cylindrical copper "cage". The base of the supporting
structure was threaded to fit similarly tapped threads in
the base of the cage. The top of the cage contained the
spring loaded, fixed plate assembly. To give this plate
a fixed position, stops were machined a distance 0.317"
from the top of the cage. The top of the cage was threaded
to receive a 0.875" O.D. copper plate. Between this copper
plate and the fixed plate was inserted a small spring. The
spring was pliant enough so that a small pressure inside
the bellows could lift the fixed plate above the stops in
the cage, but was strong enough to insure that the plate
returned to the stops when it was no longer in contact with
The fixed plate consisted of a guard ring surrounded
by a Mylar-insulated capacitor plate. The capacitor plate
was made from a 0.486" O.D. copper piece with a 0.067" taper
per foot. This piece was wound with two layers of 0.001"
Mylar and then pressed into a similarly tapered hole in a
copper plate. The outside plate formed a guard ring for
the capacitor plate.
The entire sample system was mounted on top of a
steel pressure plug. The copper supporting structure was
silver soldered to the steel plug at the point where the
supporting structure passed through the steel plug. The
outside of the pressure plug was threaded to fit the female
threads of the pressure bomb. A 0.012" copper gasket was
used to seal the pressure plug to a knife-edge surface
machined into the bomb.
Steel was chosen as the material for the pressure
bomb for two reasons. Firstly, the strength of steel per-
mitted a minimal wall thickness of the pressure bomb.
Secondly, the thermal expansion of steel is very nearly
that for a ductile material such as copper which was to
be used as the sealing gasket. If the gasket contracted
more rapidly than the steel, then the bearing pressure on
the sealing gasket would be appreciably reduced at low
To permit easy access to the sample chamber, all
electrical and capillary connections into the bomb were
made through the pressure plug. The capacitor leads were
attached to glass-to-metal seals which were soldered to
the inside surface of the pressure plug.
As mentioned previously, the entire length of capil-
lary to the top of the sample chamber had to be heated in
order to fill the sample chamber. A stainless steel ther-
mal standoff extended 2.5" from the outside surface of the
pressure plug. The filling capillary was soldered to a
feed-through glass-to-metal seal at the end of the thermal
standoff. The capillary which was used to pressurize the
pressure bomb was silver soldered into the pressure plug.
A large copper cylinder was soldered to that part of
the sample support system which extended outside the
pressure plug. Holes were drilled into this cylinder to
form wells into which the thermometers were inserted.
Apiezon type N grease was forced around the thermometers
and into the wells to insure good thermal contact between
the cylinder and the thermometers. All electrical leads
were thermally anchored to this cylinder by soldering them
to glass-to-metal seals set in the large copper cylinder.
Capacitor leads were run on opposite sides of the sample
support system, and,wherever unshielded, were separated as
far as possible to reduce distributed capacitance. By
disconnecting the leads at the capacitor plates, the dis-
tributed line capacitance was measured to be 0.004 pf. The
distributed line capacitance affected the total capacitance
by less than 0.4%. As mentioned previously, the capacitance
leads were made from stainless steel and had a low temperature
coefficient such that the change in the distributed line
capacitance with temperature was essentially zero.
Temperature Measurements, Calibration, and Regulation
A "three-terminal" resistance bridge was used to measure
the resistance of the germanium resistance-thermometer. A
schematic drawing of the three-terminal resistance bridge
is shown in Fig. 6. The term three-terminal means that the
ground is important as well as the current leads to the
unknown and standard resistor. By placing the current leads
to the resistors across the ratio transformer and grounding
the center tap of the ratio transformer, ground acts as a
"guard point". At balance the detector signal is zero and
no current is drawn. Therefore, any resistive or capacitive
leakage to ground in the leads has no effect on the balance
of the bridge. At balance, the unknown resistance of the
germanium resistance-thermometer plus the lead resistance
is given by
R = S(l x) (15)
where R is the unknown resistance, S the resistance of the
standard resistor, and x the reading of the ratio trans-
former. The use of stainless steel thermometer leads in
the cryostat minimized both the temperature coefficient of
the lead resistance and the total line resistance, while
thermally isolating the sample system from room temperature.
The bridge was composed of a Model 120 PAR lockin
amplifier with a fixed oscillator frequency of 400 Hz, a
i----vw\_ *- 1
PAR CR4 differential preamplifier, a Gersch ST100-B iso-
lation transformer, a Cryocal S/N 87 germanium resistance-
thermometer, a wire-wound standard resistor (2000-ohms)
with a low temperature coefficient, and a General Radio
model 1493 ratio transformer.
The germanium resistance-thermometer was calibrated
against both the vapor pressure of hydrogen and against a
platinum resistance-thermometer. Between 12 and 20.4 K
temperatures were calibrated against the vapor pressure of
normal liquid hydrogen. All vapor pressure measurements
were taken within three to four hours after condensation
of the hydrogen gas into the vapor pressure bulb, and thus
represented values appropriate for normal liquid hydrogen.
The hydrogen was condensed into a copper bulb with a total
volume of 2.5 cm3. A strip of copper foil 0.010" thick,
0.50" wide and 10" long, was spiral wound to fit the inside
of the chamber and was silver soldered to the bottom surface
of the bulb. The large surface area provided good thermal
contact between the liquid and the chamber.
In the temperature interval 10-77 K, the germanium
resistance-thermometer was calibrated against a platinum
resistance-thermometer, No. 1634579. This platinum ther-
mometer had been calibrated in May, 1964 by the National
Bureau of Standards.
For the germanium resistance-thermometer, if the log R
is plotted against log T, the result is very nearly a
straight line. To obtain temperatures between 4.2 and 10 K
this straight line was used to find the temperature corres-
ponding to a measured value of the resistance. By using
the average value, -.0591, for the slope of this line in
the temperature interval 10-77 K, the temperature can be
T = ZAiR-0.591(i 1) (16)
where R is the resistance, Ai the constants to be deter-
mined, and n the number of terms in the expansion. Since
the resistance has already been expressed in terms of the
bridge reading, the temperature is then given by
n -0.591(i 1)
T = 2A[ [1 x 2000 (17)
A program developed by Philp35 permits the temperature to
be calculated by computer directly from the bridge reading.
This program gave a smooth temperature fit with a rms devia-
tion of 0.02 K when five terms were used in the expansion.
To regulate the temperature, the out-of-balance signal
from the lockin amplifier was used to heat the 1000-ohm coil
wound on the inner vacuum jacket against the heat leak to
the main bath. To maintain temperatures above 15 K, the
output of the lockin amplifier was supplemented by a con-
stant heating current supplied to the larger 5000-ohm coil
also wound on the outside of the inner vacuum jacket.
Procedure and Sample Measurements
The sample gas was condensed into the sample chamber
until the capacitance indicated that pressure was being
transmitted to the bellows. The liquid in the bellows
was then cooled into the solid. A thermal gradient was
established along the length of the chamber and more solid
was sublimated from the filling capillary into the sample
chamber. The heat needed to warm the capillary as well as
the warm fluid entering the bellows from the capillary
gradually raised the temperature of the entire sample
system. The heat supplied to the capillary would then
have to be reduced and the sample system allowed to cool
much below the melting temperature of the sample. This
process would be repeated until the capacitance and
temperature indicated that the bellows was distended from
its equilibrium position by the solid which it surrounded.
To insure that a good sample was formed, the sample
was warmed above its melting point and allowed to cool
slowly into the solid and then anneal for a period of from
eight to ten hours. With the sample thus formed, the
sample was cooled to 4.2 K over a two to three hour period.
To measure the expansion coefficient, heating and
cooling curves of the capacitance as a function of temperature
were taken. Near solid phase transitions, capacitance was
measured at three to four millidegree intervals. Away from
the solid phase transitions, capacitance was measured at
0.1 to 0.2 K intervals. Thermal equilibrium was indicated
when the capacitance no longer changed at a fixed temperature.
With the solid nitrogen samples, the temperature was main-
tained at 35.6 K for time periods ranging from one to three
hours to insure that the total length change associated with
the first-order alpha-beta transition was measured accurately.
To measure the isothermal compressibility, the pressure
bomb was pressurized with fluid helium. As mentioned pre-
viously, the fluid helium surrounded the sample as well as
the sample chamber. The compressibility was measured by
maintaining a constant temperature and noting changes in
capacitance corresponding to changes in pressure. The
pressure was varied by changing the height of the mercury
in the U-tube. The pressure was read to 0.01 psi on a
Texas Instruments pressure gauge.
IV. EXPERIMENTAL RESULTS AND DISCUSSION
The raw data obtained from these experiments consist
of a series of values of capacitance corresponding to
various temperatures at zero pressure, and a series of
values of capacitance corresponding to various pressures
at a fixed temperature. To relate the changes in capaci-
tance to changes in length, we must first look at the
expression for a three-terminal capacitor using a guard
C 1f72 + ; Erw +l + Fw (18)
1g Ig + 0.22w [L 2rJJ
where C is the capacitance, E the permittivity, r the
radius of the plate with the guard ring, 1g the length
between the capacitor plates, and w the half width of the
distance between the inside radius of the guard ring and
the outside radius of the capacitor plate surrounded by
the guard ring. For w = 0.001", 1 = 0.020", and r = 0.486",
the second term in the above expression affected the total
capacitance by less than 1% and the ratio 1 dClby less
To calculate the expansion coefficient, we find that
a change in the gap length, dlg, corresponding to a tempera-
ture change, dT, is given by
dl = ITr2 dC] + 2Irr [dr] (19)
dT C2 dT C dTJ
Because the radial expansion of the copper capacitor plates
is small, rIdCj > 104 d, the second term on the right-hand
C dTJ r [daT
side of Eq. (19) shall be neglected. By noting that the
change in the gap length is equal and opposite to the change
in the sample length and that the coefficient of linear
expansion is given by 1 fdls, we find that
1 = 7-r2 dCl. (20)
A correction to this term to be considered is the change in
capacitance contributed by the expansion of the construction
material. Because the entire sample system with the ex-
ception of the bellows was made from copper, only the ex-
pansion of copper over a length equal to the plate spacing
plus the length of the sample had to be considered. A
calculation of this correction showed that it affected the
results by less than 0.5% and was well inside the scatter
of the data.
A third correction to be considered was a correction
for any tilt of one capacitor plate relative to the other
capacitor plate. A relation developed by Philp35 gives the
correction to the expansion coefficient as
C= C[ 1 r292l
sample measured [2--i J (21)
where r is the radius of the capacitor plate surrounded by
the guard ring, Q the angle of tilt, and 1 the gap length
at which the measurements.of capacitance were taken. A
measure of the effect of non-parallel plates was afforded
when the bellows, then filled with solid, became disengaged
from the spring loaded capacitor plate. As the sample
continued to cool, it was reasonable to assume that the
shape of the sample would remain intact. A typical value
of 48 pf was measured when the moving plate no longer
touched the ground of the guard ring. This value of
capacitance implied a plate spacing of less than 0.001"
and therefore a 9 = (0.001"/r). For a plate spacing of
0.020", this effect amounted to a correction of less than
To compute the expansion coefficient, the constant
1T-r211 = Co was calculated and inserted into a computer
program designed to determine the quantity C dC As
has been mentioned, the temperature had been fitted by
computer to a function of the bridge reading. Thus, with
the exception of computing Co, the computer calculated
the expansion coefficient as a function of temperature
from the raw data.
Because the fluid helium used to pressurize the
sample also filled the space between the capacitor plates,
the measured differences in capacitance caused by changes
in pressure had to be corrected for the accompanying
change in the dielectric constant of the fluid helium.
To calculate the isothermal compressibility, we find
that a change in the sample length, dls, corresponding to
a pressure change, dP, is given by
[dli = rgr2[dCl WTr2 d] (22)
dP T C2 LdPJ C TdpJ
Since (dC/dP)T is the quantity that is actually measured,
we must calculate (dc/dP)T from published data.
The dielectric constant, #(= E/eo, can be calculated
for a given pressure and temperature by using the Clausius-
0= 3M ( 1- 1) (23)
4T ( K + 2)
in which C( is the molar polarizability, M the molecular
weight, and S the density. This equation has been veri-
field from above 300 K (gas) down to 1.62 K (liquid He II).
Because the dielectric constant is equal to unity to within
1% over the pressure and temperature studied in this work,
this equation may be rewritten as
= [4i-a + 1 = 0.39 f+ 1 (24)
with f expressed in gm/cm3. A change in permittivity
corresponding to a change in pressure is then simply
[d = o =d 0.39Eo d[ ] (25)
LdPj LdPJ [dPJ
T dP T dPT
Because the relative change in volume is three times the
relative change in length, the expression for the iso-
thermal compressibility becomes
S= s = 3KErr2 31dCs 0.39 1dfg (26)
s dPJ T is CIdP C TdPJT
Because the density changed most rapidly with pressure at
low temperatures, and the compressibility itself was a slowly
varying function of temperature, the error in determining
what fraction of the total capacitive change was due to a
change in length of the sample was largest for the lowest
The Debye model for solids has been chosen for
convenience in comparing the results of the present deter-
mination of the expansion coefficient and the isothermal
compressibility with related quantities such as the specific
heat and the velocity of sound in solid nitrogen. For a
Debye solid, C the expansion coefficient, and KT, the
isothermal compressibility, are related to C the specific
heat at constant volume, by
Cv = 0v (27)
where T is the Gruneisen parameter and v is the molar
volume. The degree to which the ratio (v/K Cv varies from
a constant is a measure of the specific applicability of
this model to nitrogen and will be discussed later in terms
of the temperature dependence of the Gruneisen parameter.
The temperature dependence of the specific heat at
zero pressure for solid nitrogen is depicted in Fig. 7.
Comparison of the specific heat measurements given in
Fig. 7 to the temperature dependence of the relative changes
in length of nitrogen shown in Fig. 8 suggests that not
only is there an abrupt change in the lattice size corres-
ponding to a change in structure, but also that the lattice
must expand rapidly in the alpha phase near the transition
0 10 20. 30 40 50
nitrogen. Dashed curve represents behavior
0 10 20. 30 40 50
Fig. 7. Specific heat versus temperature for solid
nitrogen. Dashed curve represents behavior
expected from a Debye solid with 8D = 80K.
O -4 4
4J, 4 0)
-4 > -.4
(B H *
in order to accommodate the large amount of energy given to
the solid as it is warmed.
Recent thermal expansion measurements by MTV10 give
the relative change in volume at 35.6 K as 0.5%, which is
lower than the 0.8% value obtained earlier by Swenson.1
The magnitude of this change as determined by this work
gave a 1.4% change on warming and a 0.9% change on cooling.
Several efforts were made to reconcile the difference in
volume change when warming and cooling through the transition.
After cooling to initiate the transition, the transition
temperature was maintained for two to three hours. The
length of the sample, as detected by the capacitance bridge,
would change for the first hour, after which time no change
in length would occur as long as the transition temperature
was maintained. Cooling below the transition produced a
thermal contraction of the same magnitude as the thermal
expansion observed when warming through the same temperature
interval. Thus, there was no evidence of supercooling. An
analogous approach to determine if superheating had occurred
also produced a negative result. MTV also noticed the
presence of hysteresis in the vicinity of the phase transition.
The ratio l1/1 versus 1/T is shown in the semilogarithmic
plot given in Fig. 9. By choosing T1 as absolute zero, the
slope of ln(l/l) versus 1/T remains constant from 24 K to
the transition temperature. From the slope of this line
an activation energy of 300 cal/mole was calculated from
80 O- T1 = 0 K
S- T1 = 15 K
S- T = 23 K
1 2 3 4 5
Fig. 9. Semilogarithmic plot of relative
length changes versus inverse
temperature for alpha nitrogen.
Eq. (2) in Chapter II. By choosing T, = 15 K, the slope
remains constant from 20 K to the transition temperature
and gives an activation energy of 310 cal/mole. Finally,
by choosing the same T1 as MTV, T1 = 23 K, an activation
every of 370 cal/mole was found. This is to be compared
with the 450 cal/mole activation energy found by MTV and
the 460 cal/mole activation energy found by BKMP.
The temperature interval for which the slope remains
constant found in this work is much larger than that found
by MTV and compares favorably with the 24-35.6 K interval
in which the excess specific heat has been attributed to
orientational defects. This is also the same temperature
interval in which the pure quadrupole resonance frequency
has been found to obey the parameter relation given in
Eq. (8) in Chapter II. While the data do seem to obey
the postulated behavior for orientational defects, it is
apparent from the differences found in calculating activa-
tion energies corresponding to different choices of T1
that the experimental evidence of such behavior is far from
A point-by-point differentiation of length changes
corresponding to temperature changes gave the expansion
coefficient shown in Fig. 10. The scatter in the data
is a reflection of the limit of capacitive and temperature
sensitivity. As can be seen from the reproducibility of
the heating and cooling curves in both Fig. 8 and Fig. 10,
thermal equilibrium was established for all measurements.
0D 30 6
-. 4 *p4 >
0 0U 0
0>O rJ C
CU 0 Ci
4*H C (
-4 oo -- 4 *
C U '
w 0 A
\ $0 U)41
-AC 40 0H
C'J CD 0 -T C 0
( N Oa l
(TSX- sCox I I o O H
^ 0 0 -)
;!r i- ^ Q<
~ I ac
The expansion coefficient is discontinuous at 35.6 K, as
is the specific heat, and the temperature dependence of
the expansion coefficient is similar to that of the
Also shown in Fig. 10 are the results of MTV. As
mentioned before, the measurements of MTV near the phase
transition exhibited hysteresis and no thermal expansion
data were reported from 34.5 to 36.5 K. Experimentally
the results quoted by MTV represent length changes
measured over 1 and 2 K intervals, except near the phase
transition, where measurements were taken over 0.25 K
intervals. Since no mention is made by MTV about their
limit of resolution in detecting changes in length, it is
fair to assume from the relatively small expansion
coefficient of alpha nitrogen that the large temperature
intervals were required to detect length changes.
Because length changes were measured over smaller
temperature intervals in this work, it is nct surprising to
find that this determination of the expansion coefficient
gave smaller values for I at lower temperatures and larger
values for o( at higher temperatures. What is surprising
is the degree to which the two results differ. The lowest
data point given by MTV at 21 K is 30% higher than that
measured in this work. The highest data point given by
MTV in the alpha phase of solid nitrogen at 34.5 K is 26%
lower than that found in this work. The differences in these
results become more acute when we consider the Gruneisen
parameter, but to compute the Gruneisen parameter we must
first determine the isothermal compressibility of solid
The isothermal compressibility, as measured in this
work, and the adiabatic compressibility as measured by
BTIll are shown as a function of temperature in Fig. 11.
The isothermal compressibility, KT, is related to the
adiabatic compressibility, KS, as given by Eq. (1) in
Chapter I. To compare the measured values of the adia-
batic compressibility with measured values of the
isothermal compressibility, the following three curves
are also plotted in Fig. 11. The first curve is the
adiabatic compressibility and was calculated from the
isothermal compressibility and thermal expansion measured
in this work. Agreement with the measured values of
BTI at temperatures near the transition is good, but the
agreement becomes worse as the temperature decreases. The
second curve is the isothermal compressibility as calculated
from the adiabatic compressibility measured by BTI and the
expansion coefficient measured by MTV. Agreement with the
isothermal compressibility measured in this work is good
at low temperatures, but becomes poorer as the alpha-beta
transition is approached. The third curve is the isothermal
compressibility as calculated from the measurements of the
expansion coefficient given in this work and from the
measurements of the adiabatic compressibility given by
BTI. Agreement with the measured isothermal compressibility
Fig. 11. Measured and calculated compressibilities
for solid nitrogen.
O Isothermal compressibility measured in
0 Adiabatic compressibility measured by
BTI, Ref. 11.
O Isothermal compressibility as calculated
from the adiabatic compressibility
measured by BTI and the expansion
coefficient measured in this work.
Z Adiabatic compressibility as calculated
from the isothermal compressibility and
expansion coefficient as measured in this
O Isothermal compressibility as calculated
from the adiabatic compressibility as
measured by BTI and the expansion coeffi-
cient as measured by MTV, Ref. 10.
( -UMp OT) A1InIgISSaHdwoo
is excellent from 28 K to the transition temperature, but
at temperatures less than 28 K the agreement becomes
poorer. In any event, the worst disagreement between the
measured and calculated compressibilities is less than
Having determined the isothermal and adiabatic
compressibility of solid nitrogen in its alpha phase, the
specific heat at constant volume, C can be simply related
to the specific heat at constant pressure, C as
Cv =[KS Cp (28)
Having found Cv, the Gruneisen parameter can now be
The temperature dependence of the Gruneisen parameter
was calculated from Eq. (27) using the isothermal compressi-
bility and the expansion coefficient measured in this work.
The temperature dependence of the Gruneisen parameter was
also calculated using the adiabatic compressibility measured
by BTI and the expansion coefficient measured by MTV.
These results are shown in Fig. 12. As can be seen in
Fig. 7, above 20 K the behavior of the specific heat for
alpha nitrogen is markedly different from that expected
for a Debye solid. It is not clear how the Gruneisen
parameter can remain constant over a temperature interval
where the specific heat differs appreciably from that of a
Debye solid. Yet, the Gruneisen parameter as calculated
from the measurements of BTI and MTV would seem to indicate
that alpha nitrogen is nearly a perfect Debye solid.
8 12. 16 20. 24. '28. 32. 36
Fig. 12. Gruneisen parameter versus temperature
for alpha nitrogen. 0's represent
values calculated from measurements of
MTV and BTI, Refs. 10 and 11.
The Gruneisen parameter as calculated solely from the
results found in this work is more easily understood.
Clearly, the Debye model predicts no first-order phase
change at any temperature below the Debye temperature, 9D.
The degree to which the Gruneisen parameter varies can be
considered a measure of how a particular solid deviates
from the behavior of an ideal Debye solid. As can be
seen from Fig. 12, the Gruneisen parameter has a strong
temperature dependence as the transition temperature is
From the nearly constant behavior of f below 18 K
found in this work, alpha nitrogen can be said to behave
like a Debye solid in the so-called "T3. region. There
is some question as to the absolute magnitude of I as
found in this work, because one would nominally expect
to have a value between 1 and 2. It is not clear, however,
what shape the expansion coefficient would have to assume
from 0 to 20 K to maintain a higher value for the Gruneisen
The specific heat of solid methane as measured by
Clusius6 is shown in Fig. 13. The solid undergoes a change
of phase at 20.4 K as is evidenced by the large I anomaly
in the specific heat.
Relative length changes in methane measured in this
work are shown in Fig. 14. The reproducibility of the
16 18 1020 22 24
Fig. 13. Specific heat versus temperature for solid
methane. O's represent data points given
by. Clusius, Ref. 6.
16 18 20 22 24.
Fig. 13. Specific heat versus temperature for solid
methane. 0's represent data points given
by Clusius, Ref. 6.
4) cQ) >
1) CU -4 '4-
V0 >U CC
N U) > 0
01 C 0
Q) 9) 4a
4 -4 )-4
4-> 4J 4J *4
Sr )l C)
r0 CU 4. C) ^
C C C
H r 1l i
warming and cooling curves establishes that all points
were taken in thermal equilibrium. Fast heating rates
displaced the warming curves to higher temperatures. Also
shown in Fig. 14 are the warming and cooling curves found
by Heuse.20 It is important to note the similarities in
the results of Heuse and the results presented in this work.
The transition is spread out below 20.4 K with the
consequence that neither result gives a maximum slope at
20.4 K. Furthermore, while nl/1 remains continuous, dl/dT
is discontinuous at 20.4 K.
The expansion coefficient of methane as measured in
this work is shown in Fig. 15. The circles and triangles
represent two samples of different length and formed from
different gas. The maximum in the expansion coefficient
is found to be at 19.8 K. It is also apparent that at
20.4 K the solid is entirely in the higher temperature
phase. Also shown in Fig. 15 are the thermal expansion
measurements of MTV. Unfortunately, MTV reported no
measurements below 22 K.
If for some reason the temperature calibration was
grossly in error, a determination of the specific heat would
show a maximum at a temperature different from that given
by the calibration. A crude determination of the specific
heat was effected by making simultaneous measurements of the
temperature and length changes at equal time intervals when
heating at a slow constant rate. The temperature dependence
of the expansion coefficient remained the same, but the
maximum slope in the time versus temperature plot, shown in
Linear expansion coefficient versus temperature
for-solid methane. O's and A's. represent values
for two different samples. O's represent data
points given by MTV, Ref. 10.
Fig. 16, occurred at 20.4 K. The specific heat at constant
pressure is proportional to dt/dT, which we shall call C'.
A plot of C' versus temperature is shown in Fig. 17. Since
points were taken at five minute intervals, the detailed
structure of the specific heat near 20.4 K could not be
determined. It is now obvious that the temperature cali-
bration was not in error, and that the temperatures at which
the specific heat and the expansion coefficient are maxima
are indeed different.
To explain the difference in the temperature at which
the specific heat is a maximum and the temperature at which
the expansion coefficient is a maximum, let us consider the
general thermodynamic relation
C = Cv + c2TvK 1 (29)
Eq. (29) is a different way of expressing Eq. (1) given
in Chapter I. If it is assumed that Cv is a smoothly-
varying function of temperature and that for small values of
the expansion coefficient Cv is approximately equal to C ,
then any "bump" in Cp must be attributed to the term (2TvKT.
At 19.8 K, where the expansion coefficient is a maximum, the
magnitude of this term is 1.37 cal/mole-K. The specific
heat, as measured by Clusius6 and given in Fig. 13, exhibits
a small bump at 19.8 K with a height of approximately 1.6
cal/mole-K above that assumed to be contributed by Cv. These
two quantities are in reasonable agreement. Rosenshein16
reported double peaks in the specific heat in all the upper
transitions which he measured. The smaller peak usually
0 0 0 0
O (O O M
(u\O ) 3
Fig. 17. Specific heat (C') versus temperature for
occurred 0.25 K below the main peak in the specific heat.
While the specific heat measurements reported in this work
were much too crude to resolve and confirm this structure
in the specific heat at 19.8 K, it does appear that the
expansion coefficient measured in this work is not at
variance with existing measurements of the specific heat.
It is also helpful to look at the general properties
of the transition. The transition is of the coQperative
type, in which the modes of motion of each molecule are
directly affected by those of its neighbors. There is
no noticeable change in the crystal structure at 20.4 K,
but as the transition temperature is approached, each
molecule is reoriented in the lattice. In order for this
reorientation to occur, the lattice must expand. From the
results of this work, it appears that the greater part of
the lattice expansion occurs before 20.4 K and that most
of the energy assumed by the lattice goes into the new
rotational modes allowed for the new orientation of each
Recent specific heat measurements by CGM,19 which are
shown in Fig. 18, exhibit a broad excess heat capacity
centered about 8 K. Because equilibriation times become
longer at lower temperatures, varying from several hours
at 5 K to approximately 20 minutes at 11 K, only a few
experimental points were reported by CGM. Specific
heat measurements by Rosenshein16 have shown the presence
of a second, lower temperature transition to pressures
\ \ <"
(-om/ ) IVHH DI
0 0) )
(x o o IS OI3D
\ n+J 3
\ i 4-1cT^l
^ ^ s
xo a) .
V u a -i
Y -^ e v
()I-3-oni/-[Bo) ivatH 3I3I39aS < 0
as low as 200 atm. An extrapolation of the lower temperature
phase boundary determined by Rosenshein to zero pressure
gives a transition temperature of 9.5 K. Because the transi-
tion becomes broader at low temperatures, the agreement is
reasonable. It must be pointed out, however, that there
is no other experimental evidence to verify that the high
pressure phase observed by Rosenshein is the same as the
excess heat capacity found by CGM.19
Relative changes in length as a function of temperature
as determined in this work are shown in Fig. 19. It is
readily apparent that negative expansion occurs below 8.75 K.
Calculation of the expansion coefficient does not clarify
the issue. In fact, the expansion coefficient as calculated
from Fig. 19 has a negative minimum at 7.5 K. It is
difficult to compare this behavior to existing specific
heat measurements in any other than a qualitative manner.
Attempts to explain the long equilibriation times and
the peculiar thermal behavior below 12 K in terms of hydrogen
spin conversion are inconclusive. The sample apparently
does reach thermal equilibrium, as was evidenced in this
work. The length of the sample, as measured by the capaci-
tance, would change rapidly when warmed or cooled. The
length would continue to change slowly when the temperature
was held at the new value. Thermal equilibrium was estab-
lished in a time related to the temperature. As was
noticed by CGM, the lower the temperature, the longer the
equilibriation time. Attempts to ascribe a rate constant
to the equilibriation time similar to that found for ortho-
para conversion in liquid and solid hydrogen were unsuccess-
ful. If conversion were taking place, it would have been
an irreversible process, as in liquid and solid hydrogen
where the para hydrogen becomes frozen in the lattice. If
the negative expansion observed below 8.75 K were caused
by spin species conversion, then by warming the sample to
a higher temperature, the conversion would continue to
occur and the sample would continue to expand. The results
of Fig. 19 do not bear this out. It would appear that the
best approach to determine the nature of the low temperature
properties of solid methane would be to make more sensitive
specific heat or thermal expansion measurements at pressures
up to the lower limit of the phase boundary determined by
1. C. A. Swenson, J. Chem Phys. 23, 1963 (1955).
2. L. H. Boltz M. E. Boyd, F. A. Mauer, and H. S. Reiser,
Acta Cryst. 12, 247 (1959).
3. T. H. Jordan, H. W. Smith, W. E. Streib, and W. N.
Lipscomb, J. Chem Phys. 41, 756 (1964).
4. W. E. Streib, T. H. Jordan, and W. N. Lipscomb, J. Chem.
Phys. 37, 2962 (1962).
5. A. L. Smith, W. E. Keller, and H. L. Johnston, Phys.
Rev. 79, 728 (1950),
6. K. Clusius, A. physik.Chem. B3, 41 (1929).
7. W. F. Giaque and J. O. Clayton, J. Am. Chem. Soc. 55,
8. T. A. Scott, J. Chem. Phys. 36, 1459 (1963).
9. A. S. DeReggi, Ph.D. dissertation, University of Florida
10. V. G. Manzhelii, A. M. Tolkachev, and E. I. Voitovich,
Phys. Stat. Sol. St. 13, 351 (1966).
11. P. A. Bezuglyi, L. M. Tarasenko, and Yu. S. Ivanov,
Sov. Phys. Sol. St. 10, 1660 (1969).
12. A. Schallamach, Proc. Roy. Soc. (London) A171, 569 (1939).
13. L. Pauling, Phys. Rev. 36, 430 (1930).
14. J. Thomas, N. Alpert, and H. Torrey, J. Chem. Phys. 18,
15. A. Crawford, Can. J. Phys. 30, 81 (1952).
16. J. S. Rosenshein, Ph.D. dissertation, Massachusetts
Institute of Technology (1963).
17. R. Stevenson, J. Chem. Phys. 27, 656 (1957).
18. J. W. Stewart, J. Phys. Chem. Solids 12, 122 (1959).
19. J. H. Colwell, E. K. Gill, and J. A. Morrison, J. Chem.
Phys. 36, 2223 (1962).
20. W. Heuse, Z. physik. Chem. A147, 266 (1936).
21. J. G. Kirkwood, J. Chem. Phys. 8, 205 (1940).
22. T. J. Kreiger and H. M. James, J. Chem. Phys. 22, 796
23. B. C. Kohin, J. Chem. Phys. 33, 882 (1960).
24. L. Jansen and F. W. de Wette, Physica 21, 83 (1955).
25. A. S. DeReggi, P. C. Canepa, and T. A. Scott, J. Mag.
Resonance 1, 144 (1969).
26. J. C. Raich and R. D. Etters, J. Phys. Chem. Solids 29,
27. M. E. Bagatskii, V. A. Kucheryavy, V. G. Manzhelii, and
V. A. Popov, Phys. Stat. Sol. 26, 453 (1968).
28. T. A. Keenan and H. M. James, J. Chem. Phys. 31, 12
29. T. Yamamoto and Y. Kataoka, J. Chem. Phys. 48, 3199
30. R. H. Fowler, Proc. Roy. Soc. (London) A149, 1 (1935).
31. R. H. Fowler, Proc. Roy. Soc. (London) A163, 450 (1937).
32. P. J. Walsh, M. S. Thesis, University of Florida (1963).
33. G. C. Straty, Ph.D. dissertation, University of Florida
34. M. F. Panczyck, R. A. Scribner, G. C. Straty, and E. D.
Adams, Phys. Rev. Letters 19, 1102 (1967).
35. J. W. Philp, Ph.D. dissertation, University of Florida
36. J. C. Maxwell, A Treatise on Electricity and Magnetism
(Dover Publications, Inc., New York) 1, 308 (1954).
37. A. A. Maryott and F. Buckley, Natl. Bur. Standards Cir.
537, 29 (1953).
David Craig Heberlein was born in San Antonio, Texas,
on January 8, 1942. He attended high school in Arlington,
Virginia, graduating from Washington-Lee High School in
1959. After four years at the University of Virginia,
he received a B. S. in Physics degree. Since entering
the University of Florida in 1963, he has worked as a
graduate teaching assistant and as a graduate research
Mr. Heberlein is married to the former Martha Lois
Walkden of Lake Worth, Florida. He is a member of
Sigma Pi Sigma.
This dissertation was prepared under the direction of
the chairman of the candidate's supervisory committee and
has been approved by all members of that committee. It was
submitted to the Dean of the College of Arts and Sciences
and to the Graduate Council, and was approved as partial
fulfillment of the requirements for the degree of Doctor
Dean, Col /e/ of/Arts and Sciences
Dean, Graduate School
/(" l74'-'\ /f^