JOURNAL OF NON-CRYSTALLINE SOLIDS 5 (1971) 426--443 © North-Holland Publishing Co.

HEAT CAPACITIES AND THERMAL BEHAVIOR OF

ALKALI BORATE GLASSES

D. R. UHLMANN, A. G. KOLBECK and D. L. DE WITTE

Department of Metallurgy and Materials Science, Centerfor Materials Science andEngineering,
Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, U.S.A.

Received 17 July 1970

The heat capacities of selected glasses in the five alkali borate systems have been measured
over a range of high temperatures which includes the respective glass transition regions.
The heat capacities per gram atom at 350 °K show little variation with composition while
those at the low temperature ends of the glass transition, Tg , show some systematic
variations. When compared with the respective 3R values, the heat capacities at Tg- range
from about 0.75 for B2Ozto values in excess of unity for various alkali borate compositions.
The present data on heat capacity have been combined with previous data on the elastic
moduli, densities and thermal expansion coefficients to evaluate the Grfineisen constant,
7, for each composition for temperatures around ambient. For BeOz, 7 has a low value
of about 0.25. It decreases with additions of Li20 and increases with additions of the other
alkali oxides.

1. Introduction

The thermal properties o f glasses are o f interest both in their own right
and because o f their significance in understanding other properties. In recent
years, considerable concern has been directed to a particular thermal
property, the heat capacity, because o f its role in various discussions o f glass
transition p h e n o m e n a (refs. l, 2, e.g.) and its significance in interpreting
thermal expansion behavior on more physical grounds.
Despite the utility o f information on the heat capacities o f glasses, rela-
tively little data are available for inorganic oxide systems. A m o n g these,
B 2 0 3 has been the m o s t widely studied 8, 4, 9); and its behavior indicates a
n u m b e r o f noteworthy characteristics. First, there is a rapid change in the
heat capacity in the vicinity o f the glass transition, f r o m a b o u t 22 cal/mole °C
for the glass to a b o u t 31.5 cal/mole°C for the liquid. Secondly, the value o f
the heat capacity at the low-temperature end o f the glass transition is only
a b o u t 70% o f the D u l o n g and Petit 3R value, and rises slightly above the 3R
value in the liquid state.
In contrast to the latter behavior, the observations o f Haggerty et al. 5) on
two silicate glasses, which also showed significant changes in heat capacity

426
 HEAT CAPACITIES OF ALKALI BORATE GLASSES 427

in the glass transition region, indicated heat capacities at the low temperature
ends which were nearly the respective 3R values.
Turdakin and Tarasov 9) measured the heat capacities of B203 and three
sodium borate glasses at temperatures between about 60°K and 295°K.
Near the high temperature end of this range, the molar heat capacities of
the glasses were very nearly equal.
Rindone and his co-workers e6, 27) reported the heat content at 100°C in-
tervals of various alkali and alkaline-earth borate compounds and their
glasses. The alkali borate compositions investigated were: L i 2 0 . 2 B203,
Na20" 2 B2Oa, Na20" 4 B203, and K 2 0 . 4 B203. The principal attention of
these authors was directed to the energy relations among the materials,
rather than to the heat capacities which can be derived from the reported
data; and the large temperature interval between data points renders diffi-
cult a detailed discussion of the behavior around the glass transition region.
The low temperature heat capacities of crystalline and amorphous Na20"
4 B203 were reported by Westrum and Grimes28). The measurements do
not, however, extend to the glass transition region.
Recently, DeNeufville and Turnbull6) measured the heat capacities of a
series of GeO2-Ge glasses and found a similar close approximation to the
3R values for all the high-GeO2 glasses studied. In their study, the change in
heat capacity at the glass transition was found to increase with increasing
Ge concentration up to about 10 m o l e ~ Ge, and then remain more or less
constant with further Ge additions.
In the case of GeO2 and SiO2, there is little apparent change in the heat
capacity on going through the glass transition. This has been associated by
Angell 2) with departures from ideality of the glasses formed, and by DeNeuf-
ville and Turnbull 6) with the tetrahedral networks of these materials.
In the case of borate glasses, a number of anomalous characteristics have
been reported in property versus composition relations. The classic example
of this is the so-called boron oxide anomaly - a minimum in the thermal
expansion coefficient of sodium borate glasses first reported at about 16
mole~o Na20. It has recently 7) been noted, however, that no single composi-
tion can be uniquely identified with the minimum, and that the observed broad
minimum from about 10 to 30 m o l e ~ N a 2 0 may be associated with a com-
petition between two processes: the formation of BO4 tetrahedra, which
tend to decrease the expansion coefficient, and the introduction of modifying
cations, which tend to increase it. Beyond about 30 mole~o Na20, singly-
bonded oxygens are formed in significant numbers, and the expansion
coefficient increases markedly.
While this physical picture seems quite plausible, the thermal expansion of
borate glasses may alternatively be approached from a more physical basis.
428 D . R . UHLMANN, A . G . KOLBECK AND D. L. DE WITI'E

Specifically, with measurements of the densities, thermal expansion coeffi-

cients heat capacities and bulk moduli of the glasses, one can through the
GriJneisen relation relate these quanties and obtain more quantitative in-
formation about the expansion behavior. In the case of alkali borate glasses,
measurements have previously been carried out in this laboratory 7, s) of the
moduli, thermal expansion coefficients, and densities of glasses in all five
alkali borate systems.
In the present study, we shall report measurements of the heat capacities
as a function of temperature of several glasses in the N a 2 0 - B 2 0 3 system as
well as a number of glasses in other alkali borate systems. The heat capacity
and change in heat capacity at the glass transition will be discussed, as will
the thermal expansion behavior of the materials.

2. Experimental procedure
To facilitate comparison of various properties of the glasses, all but one
of the samples used in the present study were remelts of the samples used
previouslyT,S). As in the previous studies, all the compositions to be re-
ported are nominal compositions calculated from the batch; previous chem-
ical analyses of selected glasses have indicated true compositions within
_ 1 mole% alkali oxide of the nominal composition.
The remeltings were carried out in platinum crucibles at temperatures of
about 1050-1100°C for periods of 5 to 15 rain, after which the melts were
poured into heated iron molds. In the case of B203, fresh samples were
prepared by melting purified B203 obtained from the US Borax Company
at temperatures of 1300°C for 4 hr. In all cases, samples of 10-20 mg were
cut or broken from the bulk samples and used in the calorimetric investiga
tion.
The heat capacities were determined using a Perkin-Elmer DSC-1B
scanning calorimeter. This instrument is now so standard that its use does
not require detailed description here. In the present study, calibration runs
on standard samples were made before and after each sample run; and with
this the measured values of heat capacity should be accurate to about _+ 10%.
For each of the sodium borate glasses, several runs were carried out on a
given sample, and the reproducibility was found to be well within this
estimated accuracy.
In most runs, a heating rate of 10°K/min was used; in several cases, a
variety of heating rates were used to investigate the effect of such rates on
the glass transition. Most of the scans were made over the temperature range
between 350°K and 770°K. In several runs, however, a wider temperature
range (including lower temperatures) was used as well.
 H E A T C A P A C I T I E S OF A L K A L I B O R A T E G L A S S E S 429

3. Results

A typical form of the heat capacity versus temperature relation for alkali
borate glasses is shown in fig. 1. As shown there, at temperatures below the
glass transition region, the heat capacity increases slowly with increasing
temperature; and the heat capacity versus temperature relation exhibits
concave-downward curvature. For all the glasses studied, there was an
abrupt increase in the heat capacity on heating through the glass transition.
The results of the present study on the various glasses are summarized in
the figures to follow. Shown there are the heat capacities at 350°K (C35o)
and at the low temperature end of the glass transition (Crg_), as well as the
change in heat capacity on going through the glass transition (ACTg = Crg + --
-Cry_). These last quantities are defined in fig. 2; and the glass transi-
tion temperature (Tg) is taken as the temperature where the heat capacity
reaches ½( CTg_ + CTg +).

.600
C£(Tg+ ) = . 4 9 8 Tg ~ 72'0 eK
C p ( T 8 - ) =. 37~.5 Cp(350) = . 2 4 9
A
Cp=1
.25
500

I
.400
i oF
-J .300

- - ~ HE
AT
N
IG
.200
O-- S
EC
ONDH
EAT
N
IG

.100

I I I I I
300 400 500 600 700 800
TEMPERATURE (*K)

Fig. 1. Variation of heat capacity with temperature for 0.15 Na20-0.85 SiO2 glass.
Heating rate 10°K/rain. (O) first heating run; ((3) second heating run after
cooling in calorimeter.
430 D.R. UHLMANN~A.G. KOLBECK AND D. L. DE WITI'E

SCHEMATIC:

Cp VS TEMPERATURE

/
Cp(Tg+) - ---X-- - -

~p

Cp(Tg_)

TEMPERATURETgf
g" \ Tg,
Fig. 2. Heat capacity versus temperature (schematic), showing heat capacity at low-
t e m p e r a t u r e e n d o f the glass transition (CTg_), a n d heat capacity at
the h i g h - t e m p e r a t u r e e n d (CT,+).

The variation of this glass transition temperature with composition in the

five alkali borate glass systems is shown in fig. 3. The values were obtained
with a uniform heating rate of 10°K/min. As seen there, the glass transition
temperature increases with increasing alkali concentration, and at least in
the Na20-B203 system reaches a maximum about 24-28 mole~ Na20.
These variations are similar in form to those reported previously 10-aa) for
the respective systems. For the Na20-B203 system, the present values are
higher by about 50 °K than the average of those reported previously; this is
undoubtedly associated with the relatively fast rates of temperature change,
and correspondingly short experimental times, in the present investigation.
As a measure of this effect, it was found that increasing the heating rate
from 5 to 40 °K/min resulted in an increase in the glass transition temperature
of about 40°K. For the Cs20-B203 and K 2 0 - B 2 0 3 systems, the present
data are higher by about 30 °K than those of a previous studylO), while for
the L i 2 0 - B 2 0 a system the present values are higher at 6 ~ Li20 but lower
at 16~o Li20 than those of the previous study1°). No previous data are avail-
able for glasses in the R b 2 0 - B 2 0 a system.
 HEAT CAPACITIES OF ALKALI BORATE GLASSES 431

[]

.700

o [] X •
0

uJ

I-
.600

,g

.500 I I I I
0 .10 20 .50 ,40

COMPOSITION--MOLE FRACTION ALKALI OXIDE

Fig. 3. Variation of glass transition temperature with composition in the alkali borate
systems. Heating rate 10°K/min. (O) Na~O-BzOa; (O) Cs20-B2Oa; (F1) Li20-BzOa;
(11) Rb20-BzOa; ( × ) K20-BeOs.

The change in heat capacity per gram, ACT~, on going through the glass
transition was found to vary with composition as shown in fig. 4a. As seen
there, ACT, for N a 2 0 - B 2 0 3 glasses seems not to vary appreciably with com-
position up to about 30~ Na20, with the exception of the value for the
0.15 Na20-0.85 B203 composition. The data for this composition have been
shown in fig. 1 above; and like those for all other NazO-B2Oa glasses, the
ACr~ value shown in fig. 4 represents the average value of two runs which
were in close agreement with each other. For Cs20-B203, Rb20-B203 and
K 2 0 - B 2 O 3 glasses, ACT, seems to decrease systematically with increasing
alkali content, while for Li20-B203 glasses, ACT, appears to decrease some-
what with small Li20 additions but increase strongly at higher Li20 con-
centrations.
When one considers the change in heat capacity per gram-atom, rather
than per gram the results shown in fig. 4b are obtained. As shown there, the
variation with composition and type of alkali is less pronounced than is the
ACT, behavior.
432 D. R. UHLMANN, A . G . KOLBECK AND D. L. DE WITTE

.400 !

.300
E~

e
I .200 --
[]
E}= x • x
tD
ta • x
,::3 .lO0
0 0 0

0 I I I I
0 .lO .20 ,30 .40
COMPOSITION -- MOLE FRACTION ALKALI OXIDE
(a)

20.0

o
0
i
0
~O
[]
-x
• i100 •
I0.0
o n 0 x
~E

I I I I
.I 0 .20 .30 ,40
COMPOSITION -- MOLE FRACTION ALKALI OXIDE

(h)
Fig. 4. (a) Variation with composition of the change in heat capacity per gram, ACT,,
on going through the glass transition. (b) Variation with composition of the change in
heat capacity per gram-atom, ACrgm, on going through the glass transition. (a) and (b):
( O ) NazO-BzOs; ((3) Cs20-B2Oa; (IS]) LiO-BzOa; (11) Rb~O-B2Oa; ( × ) K~O-BaOa.
 HEAT CAPACITIES OF ALKALI BORATE GLASSES 433

,400

.300 O
o
I Irn
X
X •

.200
• ×
{3
0 0

0
.I O0 --

0 I 1 I
0 .10 .20 .30 .40

COMPOSITION--MOLE FRACTION ALKALI OXIDE

(a)

25.0

20.0

o ~ ,,o
[]
[] • ~ X 0 O
o
I • •
t5.0 • X
o
4..
L3
o
S I 0.0 --

5.0 --

0 I I I I
0 • 10 .20 ,50 ,40

COMPOSITION--MOLE FRACTION ALKALI OXIDE

(b)
Fig. 5. (a) Variation with composition of the heat capacity per gram at 350°K. (b) Varia-
tion with composition of the heat capacity per gram-atom at 350°K. (a) and (b): (Q)
Na~O-B~Oa; ((3) Cs20-B~O3; (D) L&O-B~O3; ( I ) Rb~O-B~Oa; ( × ) K20-B~Oa.
434 D.R. UHLMANN, A.G. KOLBECK AND D.L.DE WITTE

The values of the heat capacity per gram at 350 °K and at the low-tem-
perature end of the glass transition vary with composition as shown in
figs. 5a and 6a. As seen there, the heat capacities at 350°K seem to increase
with increasing Li20 concentration, change relatively little (perhaps de-
crease slightly) with increasing N a 2 0 concentration, and decrease with in-
creasing K 2 0 , R b 2 0 and Cs20 concentrations. At Tg_, the heat capacities
in all systems seem to increase somewhat with small additions of alkali
oxide. With further additions, the heat capacities continue to increase for
Li20-B203 glasses, remain relatively unchanged for NaEO-B20 a glasses,
and decrease for K20-B203, Rb20-B203 glasses. The data for heat capac-
ities at the respective glass transition temperatures seem to show greater
variability about the mean trend curves than the data for the particular
temperature of 350 °K.
When the heat capacities per gram-atom at 350°K are considered (see
fig. 5b) little systematic variation with composition or with type of alkali is
observed. At the low temperature end of the glass transition, the heat capac-
ities per gram-atom seem to increase with small additions of all alkali oxides.
With further Li20 and Cs20 additions, the heat capacities at Tg_ continue to
rise, while with additional Na20 and K20 additions an apparent maximum
is observed.
.500 --
[]

.400 -- C3 x

X •
r"l • •
0 •
~c
* • •
I .3001 -- X

-I
.~ 0
u 0
L
o :" .200 -- 0

.I O 0 --

o I I I I
o .~o .2o .30 .40
COMPOSITION--MOLE FRACTION ALKALI OXIDE

Fig. (6a)
 HEAT CAPACITIES OF ALKALI BORATE GLASSES 435

300
0
X
[] 0

eo
i
[]

20.O
A
o

w
.J
£3

.<
o
I0.0

o I I I L
0 .10 .20 .30 .40

COMPOSITION--MOLE FRACTION ALKALI OXIDE

Fig. (0b)
Fig. 6. (a) Variation with composition of the heat capacity per gram at the low-tem-
perature end of the glass transition, CTg_. (b) Variation with composition of the heat
capacity per gram-atom at the low-temperature end of the glass transition. (0) Na20-B2Oa;
((3) Cs~O-B2Oa; (IS]) Li20-B~O3; ( I ) RbaO-BeOa; (×) K20-BeO3.

The heat capacities at the low temperature end of the glass transition may
be compared with the Dulong and Petit 3R values for the respective glasses.
The results for each composition studied in this investigation are shown in
fig. 7. As seen there, the ratio Cr,_/3R is about 0.7 for B203, and increases
to values of 0.8 to 0.9 with increasing N a 2 0 concentration. For compositions
in each of the other alkali borate systems, this ratio reaches values slightly in
excess of unity.
The fractional change in heat capacity on heating through the glass transi-
tion, ACr,/CT,_,varies between about 0.36 and 0.63 for the glasses studied.
It does not seem to show any systematic variation with composition (for a
given type of alkali) or with the alkali type.
436 D . R . UHLMANN~ A . G . KOLBECK AND D. L. DE WITTE

1.50 - -

B 0
I 1.00 -- x •
laJ
o~ q • •
[] • • x

3
i 0.50

I I I 1
0 JO .20 .30 .40

COMPOSITION--MOLE FRACTION ALKALI OXIDE

Fig. 7. Comparison of the heat capacity at the low temperature end of the glass transition
with the Dulong and Petit 3R limit. (Q) Na20-B2Oa; (©) Cs~O-B2Oa; (D) Li20-B203;
( I ) Rb~O-B203; ( × ) K~O-B203.

4. D i s c u s s i o n

Besides B203, previous heat capacity data are available for a number of
alkali borate glasses: five sodium borates g, 26,28); a single potassium bo-
rate26); and a single lithium borate26). In all cases where detailed com-
parisons are possible, the results of the present investigation are in good
agreement with those reported previously.
Comparing figs. 5a and 5b, it is seen that much of the variation at 350 °K
of the heat capacity per unit mass with composition and type of alkali seems
to be associated with the variation of the molecular weight with composition.
A similar effect of molecular weight is seen in the data for Cr,_ (figs. 6a and
6b), but in this case some systematic variations are noted as well. These are
not readily interpretable in any direct manner at the present time. Some part
of the differences between figs. 5 and 6 may be associated with the variation
of the glass transition temperature with composition and type of alkali (fig. 3).
This is not, however, sufficient by itself to account for all the observed
differences; and consideration must also be given to factors such as the varia-
tion of the heat capacity with temperature at these relatively elevated tem-
peratures, its variation with composition and type of alkali, and the like.
The values of the heat capacity at the low temperature end of the glass
transition range from 0.73 to 1.11 of the Dulong and Petit 3R limit. B20 3
was at the low end of this range, while various alkali borate glasses were at
the high end. In the case of other inorganic oxide glasses studied to date 5, 8),
 HEAT CAPACITIES OF ALKALI BORATE GLASSES 437

the heat capacities of the glasses at Tg have been reported as about 0.9-1.0 of
the 3R limit. For most of the compositions studied in the present investiga-
tion, the glass transition temperature was sufficiently close to the upper
temperature limit of the calorimeter that little can be said with confidence
about the temperature dependence of liquid heat capacity. Consequently,
little can be said about the previous suggestion s) that when vibrational
modes are fully excited at Tg (when Crg_ is near the 3R limit), the heat
capacity of the liquid seems independent of temperature. Previous data are
generally also not of a form to permit firm conclusions to be drawn at this
point.
The heat capacity versus temperature relations for the glasses were de-
scribable over the range of measurement as straight lines or curves with
concave-downward curvature. Even for the glasses whose heat capacities
at Tg_ exceeded the 3R limit, the heat capacity was observed to increase
significantly with increasing temperature as the glass transition is approached.
Similar behavior in form was observed in the previous study of oxide glasses
by Haggerty et al.5), and by Rindone and his co-workers 28, 27).
The present measurements of the compositional dependence of the heat
capacity can be combined with previous measurements on the same glasses
of the thermal expansion coefficient, density, and bulk moduli to obtain
information about the vibrational characteristics of the alkali borate glasses.
In doing this, we shall make use of the Grtineisen relation 14):

c~ = T K T C , / V . (1)

Here e is the volume thermal expansion coefficient; K r is the isothermal

compressibility; Cv is the specific heat at constant volume; V is the specific
volume; and Y is Griineisen's constant. For a quasi-harmonic solid, a
Grtineisen constant may be defined for each mode and may be related to the
mode frequencies, as:
d In vj (2)
7j - d In V'

where vj is the frequency of t h e j t h mode. In Griineisen's original treatment,

yj was taken as the same for all modes:

y = yj, for a l l j . (2a)

Consistent with this assumption, y was treated as independent of temperature.

For the more general case, where the individual mode 7i's are not taken
as equal, the overall Grtineisen constant - appropriate for eq. (1) - may
438 D.R. UHLMANN, A.G.KOLBECK AND D.L. DE WITTE

be written:
7jc (vj)
J - 7 (T). (2b)
7 - 2 Co(vii
J
Each individual 7j is thus weighted by the heat capacity of the mode, thereby
reflecting its degree of excitation.
In using eq. (1) to evaluate 7, it should first be noted that for materials like
the glasses studied here, the specific heat at constant volume may well be
taken as the specific heat at constant pressure, Cp, measured in experiments
such as those reported here. This may be seen from the relation:
Cp - C v = TVo~2/KT . (3)

From this ( C p - C v ) is typically of magnitude l0 -4 cal g - l ° K - ~ for the

glasses of this study, which is negligible in comparison with the measured
values of Cp.
In many determinations of elastic moduli of materials, including those to
be cited in the present paper, high frequency techniques are employed. From
such measurements, values of the adiabatic compressibility, K s , are obtained.
This quantity is related to the isothermal compressibility, as:

Ks = KTCv/C p . (3a)

Since Cv~ Cp for these materials, values of K s may well be substituted for
K T in evaluating the Grtineisen relation. In the present case, the combined
use of Cp and K s data gives an exact value for 7 - as may be seen by com-
bining eqs. (1) and (3a).
The Griineisen constants for the various glasses are calculated from eq. (1)
by combining the measured heat capacities of the present study with the
values of specific volumes, thermal expansion coefficients and bulk moduli
determined in previous studies 7, s) on the same glasses. The temperatures at
which the various properties were measured covered some range of tem-
perature: heat capacities (350°K); specific volumes and moduli (300°K);
and thermal expansion coefficients (77-300 °K). Over the temperature range
of the property measurements, the only variation which could be significant
in the present application is that of the thermal expansion coefficient. The
data used here, which cover a somewhat lower temperature range than the
other properties, may be lower by some 10 or 15~ than the values appropriate
for 300-350°K. The likely variation of the properties are then not expected
to affect the estimation of 7 to any large extent; and in any case, this should
not effect significantly the compositional dependence of 7, since the tem-
peratures of the property measurements were the same for all compositions.
 HEAT CAPACITIES OF ALKALI BORATE GLASSES 439

.80

.SO

>.,
z
w
w 0 •
.40
X
0
P0 o oan
( Q
.20 El

0 [ I I 1
0 .tO .20 .50 .40

COMPOSITION -- MOLE FRACTION ALKALI OXIDE

Fig. 8. Compositional dependence of the Grfineisen constant for alkali borate glasses at
temperatures near ambient. (Q) NazO-B203; ((3) Cs20-B203; ([]) Li20-B~O3; ( . )
Rb20-B2Oa; ( × ) K20-B2Oa.

The variation of 7 with composition for the various alkali borate glass
systems is shown in fig. 8. As shown there, the Griineisen constant of B203
is about 0.25, and increases with increasing additions of NazO, K 2 0 , R b 2 0
and Cs20, but decreases to somewhat lower values with additions of Li20.
For comparison with the magnitudes of 7 shown in the figure, it might be
noted that the Gruneisen constants of most materials (admittedly crystalline)
lie between 1 and 314).
Taken at face value, the present results would indicate that the vibrational
frequencies in these glasses are significantly less sensitive to volume than are
the frequencies in most other materials [recall eq. (2) above]. To the extent
to which one can represent the cohesion in these glasses in terms of a power
law potential:
(a ( r ) = - A r - " + B r - " , (4)

and where the Grfineisen equation of state 15) provides a reasonable descrip-
tion (as it does for many materials), then ~ can be related to the exponent of
the repulsive term, as 15):
= ~(n + 2). (5)

While this has usefully been employed in a sizable number of cases, in the
present instance, it presents difficulties for many of the compositions.
44O D. R. UHLMANN, A . G . KOLBECK AND D. L. DE WITTE

Specifically, for ?<0.33, the values of n predicted by eq. (5) would be neg-
ative; and even for ? < 0 . 5 (corresponding to n < l ) , the calculated com-
pressibility would be negative.
Similarly small room-temperature values of ? had been noted by Griin-
eisen 15) for a number of crystalline materials - e.g., arsenic, zircon, calcite and
beryl. This was associated with group formation in the materials, and a
breakdown in the assumption that the ?j's are the same for all modes (that
all modes have the same volume dependence of frequency). Among glasses,
the most widely studied in this regard has been fused silica (refs. 16, 17, e.g.),
which exhibits anomalous low temperature heat capacity behavior, elastic
moduli which increase with increasing temperature and decrease with in-
creasing pressure, a thermal expansion coefficient which is negative over a
range of low temperatures, and a Grtineisen constant which is negative over
a range of temperature and reaches large negative values at low temperature.
The physical bases of much of this behavior has been associated 18) with the
transverse motions of oxygen atoms; and the low temperature negative ex-
pansion in compounds having the zincblende structure (also tetrahedral
coordination) has been similarly explained on the basis of more elegant
calculations 19, 20). It remains to be demonstrated, however, whether these
weak transverse modes are sufficient of themselves to explain the thermal
expansion behavior of fused silica, or indeed whether descriptions in terms
of rather localized modes are appropriate for representing the low-frequency
behavior.
Some of the present results can be discussed in similar terms. The low
values of ? for B203 and many of the binary borate glasses reflect the pre-
sence of vibrational modes with small or negative ?j (recall that most
vibrational modes in typical materials are characterized by ?j between 1 and
3). In the case of B203, the structure of the glass is composed of planar BO 3
triangles apparently joined in boroxyl-type arrangements 21). The small value
of ? observed here at temperatures near ambient may well reflect modes,
characterized by weak restoring forces, associated with the transverse mo-
tions of the oxygen atoms (normal to the planes formed by the B-O-B
linkages). These modes need not, of course be localized to a single B - O - B
group of atoms.
In the light of the fact that the low value of ? was observed in a relatively
high temperature range, it seems clear that there must be an appreciable
number of vibrational modes having small (or possibly even negative) ?j.
From the fact that the thermal expansion of B203 remains positive at low
temperatures (4-16°K), in contrast to the behavior of tetrahedral glass
formers22), one can conclude that 7 also remains positive.
The magnitude of the Grtineisen parameter at 15°K can be estimated
 HEAT CAPACITIES OF ALKALI BORATE GLASSES 441

by combining the measured 22) linear thermal expansion coefficient,

3 x 10-6°K -1, with an extrapolation of the heat capacity data 9) obtained
at temperatures of 60 °K and higher.
F r o m the latter data, the heat capacity versus temperature relation for
B203 is seen to exhibit concave-downward curvature at the lowest tempera-
tures studied (60-100°K). Even a linear extrapolation of the data to lower
temperatures could, then, represent an overestimate of the heat capacity
- and taking the heat capacity at 15 °K to be the same at 60 °K would clearly
represent an overestimate of Co. With the former assumption, a heat capacity
at 15 °K of about 1.2 cal m o l e - 1 ° K - ' is estimated, while with the latter, the
estimated value is about 3.7 cal m o l e - 1 oK - 1. With these respective value for
ct and Cv, and taking the compressibility as independent of temperature*,
is estimated to lie between 0.25 and 0.75, or perhaps even higher. It is
suggested, therefore, that in contrast with fused silica whose 7 decreases to
progressively more negative values with falling temperature, the GriJneisen
constant for B203 seems to increase somewhat with falling temperature
below ambient.
The observed variations of ~ with temperature for a number of crystals
and glasses have been summarized by Leadbetter 17), and indicate a variety
of behavior. This behavior, like that of the borate glasses studied in this
investigation, is not amenable to simple explanation. Certain plausible sug-
gestions can be made, but even the general form of the data remains unex-
plained. The results on B203 indicate directly, however, that the high fre-
quency modes are less sensitive to volume changes than the lower frequency
modes.
In considering the ~ versus composition relations, it should first be recalled
that the addition of alkali oxide to B203 results in a change of boron coor-
dination from 3-fold to 4-fold, with the fraction of bron atoms in tetrahedral
coordination increasing continuously with increasing alkali oxide content up
to about 30-35 mole~o alkali oxide 23, 24). This change in boron coordination
was expected by Warren 25) to result in a decrease in the thermal expansion
coefficient. The smaller expansion was qualitatively associated with a change
in the local bonding character from 2-dimensional to 3-dimensional, with an
attendant increase in the tightness of the structure. For additions of alkali
oxide larger than about 30-35 m o l e ~ , singly-bonded oxygens are produced
in sizable concentrations. These have been regarded as weak links in the
B - O network, whose presence would result in a higher expansion coefficient.
The importance of other factors than the boron coordination is emphasized

* The compressibility is expected to decrease somewhat with falling temperature; but this
variation should be small in comparison with the change in ct and Cv.
442 D . R . UHLMANN, A. G. KOLBECK AND D. L. DE WITTE

by the following observation 7): while the thermal expansion does decrease
on adding alkali oxides to B203, a decrease of only about 15% is found in
the Cs20-B203 system, while a decrease of more than a factor of 3 is found
in the LizO-B203 system. The direct effect of the modifying cations is ex-
pected to be an increase in the expansion coefficient by the introduction of a
restoring force for motions of oxygen atoms in the oxygen-alkali directions.
Considering the effect of alkali additions on the Griineisen constant, it
should again be noted that a satisfactory description must await further
developments. Again, however, it seems clear that at least two effects must
be considered: the change in character of B-O coordination polyhedra from
triangles to tetrahedra, and the introduction of modifying cations into the
structure. In terms of a highly simplified view, the cations should result in a
decrease in the density of low frequency modes characterized by relatively
large ?'s, and hence a decrease in the overall ?. Such an effect should be
largest in the case of Li ions, which have the most pronounced effect on the
modulus and thermal expansion, and smallest for Rb and Cs. To the extent
to which this simple picture provides an appropriate description of the effect
of cations, the measured ? versus composition relations shown in fig. 8
suggest that the change in boron coordination has the effect of increasing ?.
This suggestion must, however, be regarded as a highly tentative one.

5. Conclusions

The heat capacities of selected glasses in the five alkali borate systems have
been measured over a range of temperature including the respective glass
transition regions. The heat capacities per gram at a given temperature
(350 °K) were found to vary rather systematically with composition, reflecting
in large measure the variation of molecular weight with composition. The
heat capacities per gram-atom were found correspondingly to display rela-
tively little variation with composition. The calorimetric glass transition
temperatures were observed to increase with increasing alkali oxide con-
centrations in all systems, with the Tg's being highest for the N a 2 0 - B 2 0 3
and Li20-B203 glasses and lowest for the Cs20-B203 and Rb20-B203
glasses. The heat capacities per gram-atom at the low temperature end of the
glass transition when compared with the respective 3R values ranged from
about 0.75 for B203 to ratios in excess of unity for a number of borate com-
positions.
Combining the present values of heat capacity with data obtained pre-
viously on the elastic moduli, densities and thermal expansion coefficients
of the same glasses, the Griineisen constant has been evaluated for each
composition for temperatures around ambient. The Grtineisen constant, y, for
 HEAT CAPACITIES OF ALKALI BORATE GLASSES 443

B203 was f o u n d to be low ( a b o u t 0.25). W i t h ad d i t i o n s o f L i 2 0 , 7 was

observed to decrease s o m e w h a t , while for additions o f all o t h er alkali oxides,
~, was f o u n d to increase. N e i t h e r the m a g n i t u d e n o r the v ar i at i o n o f the
GriJneisen c o n s t a n t can simply be i n t e r p r e t e d in satisfactory detail at the
present time.

Acknowledgements

T h e au t h o rs are h a p p y to a c k n o w l e d g e stimulating discussion with

J. T. F o u r n i e r , R. R. Shaw and R. J. L a n d r y o f the A m e r i c a n Optical C o m -
pany. F i n a n c i a l s u p p o r t for the w o r k was p r o v i d e d by the A d v a n c e d Re-
search Projects A g e n c y u n d e r C o n t r a c t A R P A SD-90 and by the U n d e r -
g r a d u a t e Research O p p o r t u n i t i e s P r o g r a m at M . L T .

References
1) G. Adam and J. H. Gibbs, J. Chem. Phys. 43 (1965) 139.
2) C. A. Angell, J. Am. Ceram. Soc. 51 (1968) 117.
3) S. B. Thomas and G. S. Parks, J. Phys. Chem. 35 (1931) 2091.
4) N. E. Schmidt, Russ. J. Inorg. Chem. (Eng. Transl.) 11 (1966) 241.
5) J. S. Haggerty, A. R. Cooper and J. H. Heasley, Phys. Chem. Glasses 9 (1968) 47.
6) J. DeNeufville and D. Turnbull, to be published;
see J. DeNeufville, PhD. Thesis, Harvard University, Cambridge, Mass., 1969.
7) D. R. Uhlmann and R. R. Shaw, J. Non-Crystalline Solids 1 (1969) 347.
8) R. R. Shaw and D. R. Uhlmann, to be published;
see R. R. Shaw, ScD. Thesis, Massachusetts Institute of Technology, Cambridge,
Mass., 1967.
9) V. A. Turdakin and V. V. Tarasov, Russ. J. Inorg. Chem. (Eng. transl.) 11 (1966) 501.
10) K. H. Karsch, Glastechn. Ber. 35 (1962) 234.
11) E. Jenckel, Z. Elektrochem. 41 (1935) 211.
12) E. J. Gooding and W. E. S. Turner, J. Soc. Glass Technol. 18 (1934) 32.
13) M. D. Sampsoen, Ann. Phys. (Paris) 9 (1928) 35.
14) J. C. Slater, Introduction to ChemicalPhysics (McGraw-Hill, New York, 1959).
15) E. Gr~ineisen, in: Handbuch der Physik, Vol. 10 (1926) p. 1.
16) O. L. Anderson and G. J. Dienes, in: Non-Crystalline Solids, Ed. V. D. Frechette
(Wiley, New York, 1960).
17) A. J. Leadbetter, Phys. Chem. Glasses 9 (1968) 1.
18) H. T. Smyth, J. Am. Ceram. Soc. 38 (1955) 140.
19) M. Blackman, Phil. Mag. 3 (1958) 831.
20) A. Bienenstock and G. Burley, J. Phys. Chem. Solids 24 (1963) 1271.
21) R. L. Mozzi and B. E. Warren, to be published.
22) J. T. Krause and C. R. Kurkjian, J. Am. Ceram. Soc. 51 (1968) 226.
23) P. J. Bray, in: Interaction of Radiation with Solids, Ed. A. Bishay (Plenum Press,
New York, 1967).
24) P. J. Bray and J. G. O'Keefe, Phys. Chem. Glasses 4 (1963) 37.
25) B. E. Warren, J. Appl. Phys. 13 (1942) 602.
26) G. S. Smith and G. E. Ridone, J. Am. Ceram. Soc. 44 (1961) 72.
27) D. R. Stewart and G. E. Rindone, J. Am. Ceram. Soc. 46 (1963) 593.
28) E. F. Westrum and D. M. Grimes, J. Am. Chem. Soc. 79 (1957) 1799.