Int. J. Rock Mech. Min. Sci. & Geomech. Abstr. Vol. 11, pp. 403--411. Pergamon Press 1974.

Printed in Great Britain

Thermal Expansion Behavior of

Igneous Rocks
DOROTHY RICHTER*
GENE SIMMONS*
New experimental data on the thermal expansion of several igneous rocks over
the range 25-550°C show that the observed thermal expansion is a fimction
o.f crack porosity, heating rate, and previous maxinmm temperature as well as
mineralooical composition and preferred crystal orientation. For heating rates
<2°C/rain and a maximum temperature <250°C; expansion curves are repro-
ducible, may show a small hysteresis but no permanent strain, and the coeffi-
cient of thermal expansion (~) is approximately that calculated fi'om the single
crystal values of the constituent minerals. For highly cracked rocks ~ is sionifi-
cantly less than the calculated value. For heating rates >2°C/min or
T > 350°C, new cracks and hence permanent strain are produced in the sample
and ~t is systematically decreased. Recyclin 9 to the same temperature produces
additional cracks until a steady state is reached after 2-5 cycles.

INTRODUCTION crease (by smaller increments) after successive heatings

until a permanent condition was reached after a fifth
The thermal expansion of igneous rocks appears not to cycle. Wheeler attributed the permanent length change
have been studied extensively despite the importance of to the mismatch of coefficients of thermal expansion of
the subject in such different disciplines as civil engineer- the component minerals of the rocks, which would
ing, architecture, and geophysics. The few existing expand and contract by different amounts causing minor
studies show conflicting and commonly divergent reorientations of the mineral grains.
results. Griffith [4] measured the thermal expansion of a suite
As part of a larger program of investigation of the of rocks over the range 0-260°C. His data were incom-
effect of microstructures on the physical properties of plete because he reported only heating curves and failed
terrestrial and lunar rocks, we have studied the thermal to indicate whether there were any permanent length
expansion of several igneous rocks at room pressure. In changes in the samples after heating. However, he did
this report we consider the effects of heating-cooling note that in several cases a change in the slope of the
rates, repeated cycling, microcracks, grain size and com- expansion curve could be attributed to a small crack in
position on the observed thermal expansion behavior of the specimen. He tentatively concluded, as did Sosman
igneous rocks over a temperature range of approxi- [5] in reporting the properties of quartz, that minute dis-
mately 25-550°C. Our new results provide a basis for continuities in the sample, such as cracks, cleavage
understanding the previous divergence of results and planes, and inclusions, can significantly affect thermal
also give more reliable estimates of the thermal expan- expansion curves.
sion coefficients of several rocks at these temperatures. The question of permanent changes of specimen
Noteworthy for their early contributions to the under- length caused by heating is central to obtaining valid
standing of the thermal expansion of igneous rocks are data on thermal expansion. Obviously, if simply heating
Bartlett [I], Adie [2], and Wheeler [3]. Wheeler used a sample produces irreversible effects, then the usual
meter-sized samples. His results for the thermal expan- meaning of thermal expansion is no longer adequate and
sion of Westerly granite and Sudbury diabase over the one must include information on the history of each
temperature range of 25-1000~C were typical of the early sample.
values and indeed gave correct orders Of magnitude, He In conflict with previous observations, Hockman and
observed two effects of heating rocks that have been con- Kessler's [6] detailed study of the thermal expansion of
firmed many times since (and we believe misinterpreted): 48 granites over the temperature range of - 2 0 - + 60°C
(1) that the samples did not return to their original showed no permanent length change of the samples after
dimensions after a first cycle to 1000'~C: and (2) that the heating. Most significantly for our interpretation of their
sample's length at room temperature continued to in- results, and in the light of our own data, the)' used a heat-
inq rate of 0.4°C/rain. They attributed slight irregulari-
*Department of Earth and Planetary Sciences, Massachusetts Insti- ties in the heating curves to moisture changes in the sam-
tute of Technology, Cambridge. Massachusetts 02139. U.S.A. ples during the test,
403
404 Dorothy Richter and Gene Simmons

More recently, measurements on lunar rocks and on transformer (L VDT). The measuring system, designed to
terrestrial igneous r o c k s thought to be analogues of be mounted horizontally, was mounted vertically with
lunar rocks have increased our understanding of thermal the L V D T b e l o w the furnace. The leaf-spring pushrod
expansion of igneous rocks. holder was removed. These two modifications, described
Thirumalai and Demou [7-1 and Griffin and Demou in more detail by Baldridge et al. [10], greatly reduced
[8] have measured the thermal expansion of'lunar-ana- friction in the transducer assembly and permitted better
logue' materials. Thirumalai and Demou I-7] showed mounting of the sample. The transducer assembly was
that the thermal expansion of rocks in vacuum was the surrounded by a constant temperature water-jacketed
same as that at atmospheric pressure. They noted an im- shield to avoid errors from changes in ambient tempera-
mediately irrecoverable length change of 0.08 per cent in ture. A high-impedance a.c. amplifier was inserted
their sample of granodiorite after heating it to 350°C at between the displacement transdu.cer and the demodula-
5°C/min. Griffin and Demou [8] measured the thermal tor to improve the signal-to-noise ratio and the sensi-
expansion of several rocks from - 140 to 950°C. They tivity of the system. The signal from the transducer
showed that distinct changes in slope of the expansion exciter-demodulator was sent to a Hewlett-Packard
curve of a granodiorite corresponded to the ~-fl quartz digital data acquisition system which automatically
inversion point (573°C). took readings at 10 min intervals. The resolution of the
Baldridge and Simmons [9], Baldridge et al. 1-101, and system is 0.001r~C for temperature and 1 × 10-5 for
Todd et al. 1-11-1 have measured the thermal expansion strain.
of lunar rocks. Baldridge et al. [ 10] found that the mea- A silicon carbide tube furnace was suspended around
sured expansions of lunar samples are significantly lower the sample. The furnace control and sample tempera-
than the values calculated for the thermal expansion of tures were obtained from separate chromel-alumel ther-
an aggregate but that the observed expansion for Fairfax mocouples. The signal from the sample temperature
diabase was approximately equal to the calculated value. thermocouple, which was placed at the surface of the
They attributed the discrepancy for the lunar rocks to sample, was also connected to the data acquisition sys-
the large number of microfractures that exist in lunar tem.
rocks. Todd et al. 1-11] corroborate the results of Bal- The rate of temperature change, whether heating or
dridge et al. 1-10-1in finding that the thermal expansions cooling, is extremely important in measuring the ther-
of highly fractured igneous rocks from the Ries meteor- mal expansion. In our early work, samples were heated
ite impact crater are also much lower than the calculated at a rate of 5°C/min which we now believe is too high.
'intrinsic' values. Todd [12] used acoustic emission techniques to show
We now describe our measuring system and the exper- that a rate of 5°C/min, even at low temperature, could
imental techniques because both have significant effects introduce new cracks in rock. We have, therefore, used
on the results. a lower rate. The operational procedure used to obtain
the thermal expansions reported here, and the one that
w e recommend for use with rocks, combines a low rate
EXPERIMENTAL EQUIPMENT AND METHODS of change of temperature with long intervals of constant
We measured thermal expansion on a modified Brink- furnace temperature. In making a run, we change tem-
mann TD-IX differential dilatometer. Figure 1 is a sim- perature at a rate of 1-2°C/min for 30 min and then hold
plified block diagram of the apparatus. The measuring temperature constant for 60 rain. The period of constant
system of the dilatometer consisted of a fused silica tube, furnace temperature is sufficient for both length and
sample holder, and an inner fused silica rod which temperature of the sample to stabilize. Each data point
pushed an iron slug inside a linear variable differential is the average of the sample temperature and length dur-
ing the last 40 rain of the 60 min interval in which the
temperature remained constant.

:', H Sroo
IO

8
s,,,:, 7" -I I III %
x

~i~or ~1 I ~ ~l~ LV~T "i

~o 4

I
-I I 0 I00 200 300 400
T, °C
500 600 700 800 900

Fig. 1. Simplified block diagram of the experimental apparatus. Fig. 2. Thermal expansion of quartz parallel to the c-axis. Data from
LVDT stands for linear variable differential transformer. Kozu and Takane [13].
 Thermal Expansion Behavior of Igneous Rocks 405

m ~ t ~ r e &ills. The ends were ground flat and parallel

~" CN-Z
6 to 0.003 cm. The sides of the cylinders were not ground
18 but were used in the 'as-cored-condition'.
4 1.5
Modal compositions of our samples are given in Table
~ ,, 1. For petrographic descriptions of the rocks see [14-
¢ 18].
"r,
~ 2

RESULTS AND DISCUSSION

I I I Our experiments were designed to isolate the effects
0 I00 200 500 400 ~ 600
T, "C on thermal expansion behavior of such variables as
mineralogy, microcrack density, heating rate, and so on..
Fig. 3. T h e steady-state t h e r m a l e x p a n s i o n of C a p e N e d d i c k g a b b r o
( s a m p l e C N - | ) . C r o s s e s are h e a t i n g d a t a ; circles are c o o l i n g data. In this section, we report the data that allow us to estab-
N u m b e r s i n d i c a t e the n u m b e r of the t h e r m a l cycle. lish the relative importance of each variable.

.The dilatometer was calibrated with a cylinder of Effect of thermal gradient on the observed thermal expan-
monocrystalline quartz cut parallel to the c-axis, using sion
the data of Kozu and Takane [13] for the correct expan- The rate at which a sample is heated affects the appar-
sion of quartz between 25 ° and 450°C (shown in Fig. 2). ent thermal expansion. Because the sample is finite in
The data for each sample were then corrected for the size, a change of temperature at the surface is not instan-
slight expansion of the fused-silica measuring system of taneous throughout• the volume of the sample. There-
the dilatometer and adjusted to a reference temperature fore, a temporal gradient produces a spatial gradient in
of 25°C. Each sample data set was recalculated as rela- the sample that can produce cracks. We believe that the
tive expansion AL/Lo and a third degree polynomial was heating rates commonly used for measurements of ther-
fitted by least squares to the data set (Fig. 3). mal expansion of rocks are too large. For example, many
The linear thermal expansion of each rock core was workers have used 5°C/rain as the heating-cooling rate
measured along the long dimension of the sample only, in their thermal expansion measurements [7-9]. Others
and the volume expansion was taken to be three times have used even higher rates. Thirumalai and Demou
the linear expansion. Anisotropy of expansion was not [19] measured the thermal expansion of the Halecrest
investigated. (N.J.) and Charcoal granites over the range 20-400°C at
We estimate the precision of the instrument and tech- heating-cooling rates of 5, 20 and 50°C/min. Because the
nique from repeated measurements on samples with two higher heating rates produced large immediately
known thermal expansion (e.g. alumina, fused silica). irrecoverable microstructural damage, the coefficients of
Our measurements of temperature are precise to 10-3°C thermal expansion determined from the expansion-con-
and changes of sample length to 5 x 10- 5 in. Our deter- traction curves at the three rates are different.
minations of the coefficient of thermal expansion are re- Todd [12] demonstrated clearly that simply heating a
peatable to 2 per cent or less. granite at a rate of 5°C/min causes cracking. He heated
samples of Westerly granite at rates of 8, 5, and 2°C/min
Samples from room temperature to 300°C and studied the acous-
Sample preparation was kept as simple as possible in tic emissions. The 8°C/min rate produced the greatest
order to introduce the least possible amount of struc- number of emissions. The 5°C/min rate produced about
tural damage. Cylinders 1 cm in diameter and 4"5-5"0 cm half as many events as the 8°C/min rate. And, the rate
long were cut from large blocks with water-cooled dia- of 2°C/rain produced almost no acoustic events over the

T A B L E I. M O D A L COMPOSITIONS

Westerly Barre Cape Neddick Fairfax Frederick

granite granite gabbro diabase diabase

Quartz 29"3 30'7 -- ] --

K-feldspar 31"4 19'8 __ f 5'2
Plagioclase 31"3 36'5 56" I 49"0 44'0
Pyroxene -- -- 17"6 42'4 43"4
Amphibole -- -- 6-7 -- --
Muscovite 3"0 4"3 -- -- --
Biotite 3"8 7.3 7-0 1'0 t" 1
Olivine -- -- if8 -- 1"8
Opaques 0"8 0" 1 7"4 2"0 5'4
Apatite -- -- 3"2 -- --
Accessories 0"4 1"3 1"2 0"3 4"2

Total 99'0 100.0 100"0 99"9 99'9

N u m b e r of p o i n t s 1831 2092 1979 1602 1512

406 Dorothy Richter and Gene Simmons

TABLE2. COMPARISONOFMEASUREDANDCALCULATEDTHERMALEXPANSIONS
Measured value Calculated value

Sample ~ 200 I 3AL l :x~s 2o0= IZ~,K,V~t

= jTif XK,vl
(10~ .'C)- (10o oC)-1

•Uncracked' samples:
Cape Neddick gabbro (I) 25.5 19.5
Fairfax diabase
(5°C/min) 19-5 18.4
( 1°C/min) 17.8
Frederick diabase
(5°C/rain) 18.8 lcJ.0
Westerly granite 33"6 24-2
Barre granite 23"2 24-3
Ries granite 934 23.5 25"6
Highly cracked samples:
Ries granite 936 19-4 23-4

Lunar samples
62295 6'8 17.0
65015 10-9 18.4
68415 4-8 14.4

entire 300°C range. We conclude from Todd's data that expression

a thermal gradient of 2°C/min or less should be.used to
~,. = 3Z~d../(L.:AT) •
separate the dual effects of differing therma! expansions
of the component minerals and the heating rate on the where AL is the change in length from the initial length
thermal expansion of rocks. We have, therefore, used a Lo and ATis the change in temperature from the refer-
heating and cooling rate of l-2~C/min. ence temperature. In Table 2 are presented the mean
A comparison of the coefficients of thermal expansion coefficients of thermal expansion over the range 25-
of the Fairfax diabase determined at 5 and l°C/min 200°C of virgin samples of two granites, two diabases,
(listed in Table 2) shows a 10 per cent difference in values and one gabbro. By 'virgin" we mean that the samples
for the first cycle to 250°C. We attribute the difference have not been subjected to additional stresses after col-
in coefficients to the difference in heating-cooling rates. lection. The data in Table 2 were all taken at maximum
Precise measurements of thermal expansion must there- heating-cooling rates of l-2°C/min except the data for
fore be made at heating rates no greater than 2°C/min. the Frederick diabase and one sample of Fairfax dia-
base.
Thermal expansion of'viroin" rocks Also presented in Table 2 are the calculated values for
~,. based on Turner's equation for the expansion of an
If irreversible changes do not occur on heating or
aggregate [20].
cooling, then the thermal expansion of a polyphase
aggregate is a function of the thermal expansions and ~.,~xi K i V i
volumes of the component phases. The irreversible ~,. - ~ K i Vi
change that is most likely to occur while measuring the
thermal expansion of rocks is the creation of cracks where ~,. is the volume coefficient of the aggregate, ~; is
along grain boundaries. The stresses at grain boundaries the volume coefficient of the i'" phase, and Ki and V~are
that accompany change of temperature will produce the bulk modulus and volume fraction respectively of
cracks if the local stress exceeds the local strength. The the i'" phase. The assumptions inherent in Turner's
stress is caused by the mismatch of linear thermal expan-. equation are that (1) the distribution of phases is iso-
sion at grain boundaries which may be due to differing tropic, (2) the expansion of the phases is independent of
values of thermal expansion in the several phases (of a the size and shape of the grains, (31 no internal disrup-
.polyphase aggregate) or the anisotropy of thermal tions occur, and (4} the bulk moduli of the components
expansion in any aggregate. On the basis of our new data are approximately equal. Baldridge et al. [10] compared
described below, we define a "cracking temperature" (T~) values for ~ calculated with Turner's equation with the
as the minimum cycling temperature at which a sample values calculated from more complex formulae for the
first attains a permanent strain of 2 x 10 -4 (measured thermal expansion of an aggregate and found that they
at the reference temperaturel. It is perhaps a rough mea- agree within a few per cent.
sure of minimum grain boundary strength• It appears to In most cases, we found the intrinsic expansion to be
be a useful parameter for characterizing rocks. a good estimate of the measured thermal expansion of
The mean volume coefficient of thermal expansion (~,,) the four samples on the first cycle to 250°C, and
is derived from the relative expansion data by the observed little or no permanent length change at room
 Thermal Expansion Behavior of Igneous Rocks 407

temperature (Figs. 3 and 5). The notable exception is the elastic waves. He also showed that the fracture strength
Westerly granite, for which the measured value is much depended significantly on residual stresses. Perhaps the
higher than the calculated value. Yet our measured value most striking demonstration of the effects of residual
compares well with the value measured by Wheeler [3] stress was the bending of a bar of lunar rock number
and independently, by Griffith [4]. The measurements ' 15555 as it was sawed from a larger piece (Patrick Butler,
were made at a heating rate of l-2°C/min; and the differ- personal communication). The bar was roughly 1 x 2 x
ence in values seems too great to be caused by the inac- 17 cm and the free end moved away from the larger
curacies inherent calculating the coefficient from Tur- piece by 5 cm as it was sawed. Moisture also has been
ner's equation. We do not yet understand the cause of shown to affect several physical properties. For example,
the discrepancy between the calculated and intrinsic Simmons and Nur [22] and Nur and Simmons [23]
values. showed the effects of moisture on the velocity of elastic
waves. The effects of moisture on the electrical proper-
The effects of repeated thermal cycling on the thermal ties of rocks are well known. And finally, Hockman and
expansion of igneous rocks Kessler [6] have examined directly the effect of moisture
Figure 3 shows the thermal expansion behavior of the on the thermal expansion of rocks. They found that irre-
Cape Neddick gabbro (Sample CN-I) for several tem- gularities in the heating curves of samples of granite over
perature cycles to successively higher temperatures. The the range - 20 to + 60°C would disappear if water were
heating rate for this set of data was 2°C/min or less. prevented from entering a specimen during the test.
Several features of the thermal expansion of this rock Whatever the cause of the first hysteresis loop, none
can be obtained directly from Fig. 3: (1) the coefficient of of the successive thermal cycles of the sample displayed
thermal expansion as a function of both temperature a similar feature. The heating curve of cycle 2 (to 300°C)
and the maximum temperature to which the sample has coincides with the cooling path of the first cycle. Similar
been exposed, (2) the temperature at which significant first cycle hysteresis has been observed for other samples
cracking begins to occur, (3) the permanent strain asso- of Cape Neddick gabbro, Fairfax diabase and Frederick
ciated with maximum temperatures of cycling, and (4) diabase; such behavior was not observed in either the
the influence of cracks on the thermal expansion of a Westerly or Barre granite.
rock. Cycle 2 (T,,.... = 300°C) of CN-I yielded a permanent
The heating of any virgin rock to a temperature of strain of 1"2 x 10-4 at room temperature. The length of
even 70°C produces a (very) few cracks. For example, di- the sample remained the same after another cycle to
rect measurement of the crack porosity produced by the 300°C (cycle 3), which indicates that the rock had
thermal cycling of Frederick diabase, at a rate of I'0°C/ reached a steady state of deformation after only one
min, to 70°C gives an increase of porosity of 7 x 10-6. cycle.
Furthermore, comparison of specially prepared The sample was then cycled to increasing values
petrographic sections of virgin samples of the Frederick of Tmax in increments of approx. 50°C and cycled
diabase with sections of the samples that had been four times between room temperature and each value of
heated to 70°C shows that a few characteristic cracks T,,,~,,. In Fig. 4 we summarize the permanent strain at
were produced by the heating. room temperature of the sample of Cape Neddick gab-
Whether such cracks affect the thermal expansion is bro as a function of the maximum temperature of each
unclear to us because the change is too small to be cycle. Note that no cracking occurred during an initial
detected experimentally with our dilatometer. cycle to 250°C. The first discernible cracking occurred
The thermal expansion data on the Cape Neddick during cycle 2 to 300~'C and produced a strain of 0.012
gabbro provide a good example of the progressive effect per cent. Cycling the sample to 350 and 380°C produced
on an igneous rock of several cycles to different tempera- steady increases of co that are approximately linear with
tures. We discuss the data on this rock in some detail as T,.,~,,. Note that the values of e0 for T z; 380'C are rela-
a typical example. tively small. During the cycles to 430°C the first drama-
On even the first cycle to 250°C, a very small hyster- tic increases in permanent strain occur. After four cycles
esis loop appears but the sample shows no permanent to 430~'C, the value ofe0 had tripled, increasing from 0"03
strain. We do not understand the cause of this hysteresis. to 0.09 per cent. After four cycles to 480°C. the eo in-
Todd's [12] results indicate that very few cracks should creased to 0-16 per cent and after three cycles to 530°C,
be produced in the rock when heated at such a low rate. a total of 18 thermal cycles, the sample of Cape Neddick
Furthermore, newly created cracks should cause a per- gabbro had increased its length at room temperature by
manent strain at ambient temperature. We suggest two 0"21 per cent. This total length change was confirmed by
possible explanations for this initial hysteresis: (I) r~si- measuring the sample length with a precision
dual stresses were partially relieved in the sample with- micrometer at the end of the experiment.
out creating new ones during the first cycle, and (2) the Also plotted in Fig. 4 are the values of e,, for another
loop is a function of moisture being driven from the rock sample of Cape Neddick gabbro (Sample CN-II) which
when heated. Both residual internal stresses and mois- was cycled once to 250°C and then five times to 550°C.
ture content affect the physical properties of rocks. For The expansion curves are shown in Fig. 5.'Approximate
example, Friedman [21] showed that residual stresses in steady state conditions were reached after the fifth cycle
the Barre granite affect the velocity and attenuation of to 550°C and the sample had a final length change of
408 Dorothy Richter and Gene Simmons
30 [
CN--IT

o CN-I
20

)!I
o CN-II 2e_ a

c
I0--
9~
8~
7~
6--
~t" II/
o~ ii
2

5--
// 0 I00 200 300
T , "C
400 SO0 600
4~ / I
x I
Fig. 5. The thermal expansion of Cape Neddick $abbro (sample CN-
3~
w
o .~11 II). Crosses are heating data; circles are cooling data.
/I/
2--
.,~" iii cracked. In Table 3 we list the mean volume coefficient
(~5-2oo) measured after each cycle for the two samples
of Cape Neddick gabbro. The values were calculated
t.o
0"91
from the cooling curves. The values measured for the
0"8 first cycle, 25.5 x 10 -6 (°C) -1 for sample CN-I and
0"7
24.3 x 10 -6 (°C) -I for sample CN-II, differ by 5 per
O'E
cent, a value that exceeds our experimental error. Both
samples were heated at l-2°C/min; the difference in
0"~ I I I I I I
tOO 200 300 400 SO0 600 values is possibly due to a slight change in mineral com-
T, °C
position, or initial crack porosity, or both. The value for
Fig. 4. Permanent strain at room temperature (¢o) for Cape Neddick CN-I was slightly higher (but approximately constant)
gabbro (samples CN-I and CN-II) as a function of the maximum tem- during cycles 3-7 and then drops to 23-6 × 10 - 6 (°C)-l
perature (T,,~0 to which the sample was cycled. Circles are the data
for CN-I ; squares are the data for CN-II. upon cycling to T,,~ = 430°C. The coefficient decreased
again (to 21.6 x 10 -6) when the sample was cycled to
0.22 per cent. Comparison of the values of% for the two 480°C and dropped to 18"3 on the final cycles to 530°C,
samples of Cape Neddick gabbro shows that the degree a decrease of 33 per cent from the original ~,,.
of cracking depends on T,..... and the number of cycles to The ~5-2o0 of CN-II which was cycled once to 250°C
T,..... but is independent of the number of cycles to tower and then five times to 550°C, follows a different trend
ThL, • " but appears to be approaching a similar final value for
The curves in Figs. 3 and 5 show that the coefficient the coefficient of thermal expansion (Fig. 6). The coeffi-
of thermal expansion also change when a rock becomes cient after the first cycle to 550°C drops from 24.3 x

TABLE 3. E F F E C T S OF REPEATED THERMAL C Y C L I N G ON THE THERMAL EXPANSION OF C A P E N E D D I C K GABBRO

Sample: CN-I CN-II

T,,~ of T,,,,~ of
Cycle cycle ~5-2oo x 104 Cycle cycle ~5-2oo ~ 104
number (~CJ (°C- 1) number (°C) (°C- 1)

1 250 25.5 1 250 24"3

2 300 25.8 2 550 12"7
3 300 25.8 3 550 13"7
4 350 26.4 4 550 14"7
5 350 25.7 5 550 15'1
6 380 26.7 6 550 15"9
7 380 25.5 (Approximately 2 months passed)
8 430 23.6 B-I 550 18.5
9 430 22.8 B-2 550 16.8
I0 430 22.8 B-3 550 17.2
11 430 21.6 B-4 550 17.8
12 480 21-6 B-5 550 16.9
13 48O 18.3 B-6 550 17.6
14 480 18.5 B-7 550 17.6
15 480 18'5
16 530 17.1
17 53O 18.3
18 530 18.3
(Approximately 2 months passed}
B-I 530 22-0
B-2 53O 18.5
B-3 530 19.1
B-4 530 19.1
 Thermal Expansion Behavior of Igneous Rocks 409

10 - 6 (°C)-I to 12"7 x 10 - 6 (°C)-I. On successive !I ,

cycles, the coefficient gradually increases to 15.9 x 10 - 6 20 t3 WSI' / :
(°C)- ~, 35 per cent less than the original value. Our in- o CN-I
z~ Frfx
terpretation of these results is that during the first cycle o Frd
to 550°C, considerable cracking occurred but the com- / I

ponent crystals did not reach structural equilibrium on ,~-- /

cooling. The cracking process produces a small amount
of micro-debris along the crack so that adjacent grains
do not fit exactly together at room temperature, an ,, /
explanation suggested by Thirumalai and Demou I-7]
for their results that appear applicable to our results.
,' / "
During the later cycles to 550°C, the rock continued to x

crack, but by smaller amounts, and the individual crys-

tals apparently made slight adjustments in orientation.
We infer that the cracking during repeated cycling to the
same T,.... occurs at the high temperature end of the 0"8
0'7
cycle because on each cycle the increase in length at
550°C is larger than the increase at room temperature
0"E
0 - ~=
.¢Y
(see Fig. 5). 0"~ /

Figure 6 is a graph of :¢~s-2oo of the two samples of 0"]

/
Cape Neddick gabbro as a function of the number of
cycles to 540°C. The open symbols represent the coeffi-
cients obtained on the cycles discussed above. The
0-2
,~- ,1
closed symbols represent a new set of measurements
0.1 I I I I I
made at a later time. The second set of measurements o Ioo zoo 3oo 400 500 6oo
T, "C
was made by cycling the same samples to 540°C without
cycles to lower temperatures. The values for CN-I Fig. 7. P e r m a n e n t strain at r o o m temperature (%) as a function o f the
changed very little; the values for CN-II increased and m a x i m u m o f each thermal cycle for Westerly granite, Fairfax a n d Fre-
derick diabases, a n d C N - ] .
seemed to approach the values for CN-I.
The first cycle of both later sets of measurements pro-
duced anomalously high values for the coefficient of
thermal expansion. Successive cycles produced coeffi- which may have influenced the first cycle of the second
cients which return to the trend of the earlier measure- set of measurements is the adsorption of water into the
ments. Apparently a time dependent recovery occurred dry rock.
between the two sets of measurements. During that in- In Fig. 7 we compare the Eo (permanent length change
terval (approx. 2 months) the samples were exposed to at room temperature) observed during the thermal
room conditions. The sample length at the start of the cycling of the Westerly granite and the Fairfax and Fre-
second set of measurements was exactly the same as the derick diabases with CN-I. The quartz-rich Westerly
length at the end of the first set of measurements. Other cracks at lower temperatures than the other samples. We
workers who have observed a partial time dependent attribute this difference to the large (and anisotropic)
change in thermal expansion behavior [19] have attri- expansion of quartz relative to the other minerals in the
buted it to partial annealing of cracks. Another factor rock. Note that the quartz-free Frederick diabase was
cycled to 500°C before any permanent length change
occurred.
2'~
In Table 4 we list the coefficients of thermal expansion
for the Westerly and Barre granites, and the Frederick
CN-I • and Fairfax diabases. All of these rocks show a consis-
20
: .~..~-~- ; - - : - " tent behavior of a with increasing T,,~x.We interpret
•~ , , , O " ~ ~ ~_i1,~41,~ i F ~ , , l l , - - .11- these data to show clearly that the thermal expansion
coefficient of rocks decreases after thermal cycling to a
temperature greater than the cracking temperature.
A plot of steady-state 0q, (normalized with respect to
o 0c,,of the virgin sample) as a function of the permanent
strain (%) at room conditions is shown in Fig. 8 for
I J i I I I I I I I I I
I 2 3 4 .~ 6 7 8 9 I0 II 12 several rocks. A smooth curve has been drawn through
Number of cycles
the data such that those values for cycles which are not
Fig. 6. The coefficient of thermal expansion (%2~-200 x 106 °C- 1) of steady-state fall above the line. For instance, Westerly
the Cape Neddick gabbro as a function of the number of times it was
cycled to 530-550°C. Circles represent sample CN-I; squares represent
cycle 6 at T,..... = 400 did not quite reach steady-state.
sample CN-II. The closed symbols are data from a later (separate) set From this plot, we conclude that = is independent of the
of measurements made on the same cores. permanent strain for eo < 3 x 10- 4.
410 Dorothy Richter and Gene Simmons

T A B L E 4. E F F E C T S OF REPEATED THERMAL CYCLING ON THERMAl.

is 5-10 times greater than that of the latter. We found
EXPANSION
that T~ for the Cape Neddick gabbro is about 330c'C,
T,,,,of whereas T,. for the Frederick (average grain size <0.1
Cycle :t~5 2°° x 106 ram) diabase is around 560°C. These data are suggestive
Sample (cC) CC- ~)
that grain size is the major factor in the large differences
Fairfax 200 7"8 of T, for the Cape Neddick and Frederick specimens.
diabase 250 7'8
300 7"3 Thermal expansion behavior of hiqhly cracked rocks
350 7.8
350 7"8 Rocks which have been intensely cracked, such as by
400 8"3 impact processes, yield very low coefficients of thermal
400 18"7
450 19'7 expansion [10, I 1] compared to their calculated values.
450 18-7 In Table 2 we have listed the calculated and measured
%__.
coefficients of expansion reported previously [1 I], for
540 16.8
lunar samples and rocks from the Ries meteorite impact
Westerly 220 33'6 crater in Germany. Evidently, the pervasive cracks
granite 260 35.6 across the crystals, as well as along grain boundaries,
T,--.
300 33.2 allow the individual grains to expand into the pore space
310 32.9 without contributing to the overall expansion of the
330 32.2 aggregate. We cycled the shocked rocks over the range
400 33"0
450 32.4 25-350°C and observed no permanent strain (Eo) in any
500 28.1 of the samples at room temperature.
Frederick 270 18"8
diabase 340 18-3 CONCLUSIONS
400 18-8
450 19-7 New measurements of the thermal expansion of
510 19.3
L_, several igneous rocks as a function of heating rate, maxi-
550 18"5 mum temperature, and number of cycles, show the fol-
600 17'7 lowing:
600 14"7
(l) At low heating rates, atmospheric pressure, and
T_< 250°C, the thermal expansion of most rocks is
Effect of grain size approximated by calculations from single crystal values.
(2) Measurement of thermal expansion at rates of
The present set of data on the thermal expansion of
greater than about 2°C/rain may produce anomalously
igneous rocks is insufficient for a quantitative discussion
high values. Heating rates no greater than 2°C/min are
of the effect of grain size. However, our new data, as well
necessary for precise measurements of thermal expan-
as the results of ceramicists on several artificial aggre-
sion to eliminate cracking due to the stress produced by
gates [24, 25], suggests that such an effect does exist. The
thermal gradients in 6 mm diameter specimens.
Cape Neddick gabbro has a modal mineral composition
(3) The observed thermal expansion of a rock
somewhat similar to the Frederick diabase (Table l) and
depends inversely on the microcrack porosity. Once a
the calculated values of ~2.~-200 for the two rock types
rock has been heated to the point that different thermal
are nearly equal (Table 2). The grain size of the former
expansions of the component crystals have introduced
more than 3 × 10- 4 permanent strain, the coefficient of
cyc 7 Fr fx thermal expansion is reduced. Furthermore, shocked
cyc 3 CN - I
cyc 2 Wst
Cyc 5 C N - I
igneous rocks which initially contain high crack por-
'~ 1.0
cyc 7 C N - I
cyc 4 Wsf osities (such as lunar rocks) have lower values of ~t than

~
the same rock in an unshocked state.
(4) For T,.... ± 250'C and low heating rates some of
cyc II CN - I
our samples show a small hysteresis but no permanent
strain on the first and subsequent cycles. For these con-
cyc 15-I ditions, all cycles reproduce the first cycle.
(5) These results provide a framework with which to
$ 07

~
~ ,cycl6CN-!
cyc 6CN-O understand the previous discrepancies in values of ther-
F mal expansion reported by different investigators using
06
different experimental conditions and emphasize the
hypothesis that thermal expansion data of igneous rocks
0'5
2 must always be considered in light of the sample's pre-
E , X I0 4
vious history.
Fig. 81 Steady-statethermal expansion normalizedwith respect t9 the
value for the virgin sample as a function of co. Cyc = cycle: Frfx = Acknowledgements-ScottBaldridgehelped us in the beginning of this
Fairfax diabase: CN-I and 1I = Cape Neddick gahbro samples l and study. Frank Miller designed the additions to the electronic system
II : Wst = Westerlygranite. and maintained the equipment. David Riach machined the samples.
 Thermal Expansion Behavior of Igneous Rocks 411

Partial financial support was provided by the National Aeronautics 11, Todd T., Richter D., Simmons G. and Wang H. Unique characteri-
and Space Administration through contract NGR 22-009-540 and by zation of lunar rocks by physical properties. Proc. Fourth Lunar
the National Science Foundation contract GA 40756. Sci. Conf. 3, Pergamon Press, 2663 (19731
12. Todd T. Effect of cracks on elastic properties oflow porosity rocks.
Ph.D. thesis, Mass. Inst. of Tech., 320 (1973).
Received 8 April 1974. 13. Kozu S. andTakane K. Influence oftemperature on tbe axial ratio,
the interfacial angle, and the volume of quartz. Sci. Rep. Tohoku
Univ. 3, (3rd Ser.) 239 (1929).
REFERENCES 14. Faii'bairn H. W., Schlecht, W., Stevens R. E., Dennen W., Ahrens
L. and Chayes F. A cooperative investigation of precision and ac-
1. Bartlett W. H. C. Experiments on the expansion and contraction curacy in chemical, spectrochemical, and modal analysis of silicate
of building stone. Am. J. Sci. 22, 136 (1832). rocks. USGS Bull. 980, 71 (1951).
2. Adie A. J. On the expansion of different kinds of stone from an in- 15. Murthy V. R. Bed Rock Geology of the East Barre Area, Vermont.
crease of temperature with a description of the pyrometer used in Bull. 10 (1957).
making the experiments. Trans. Roy. Soc. Edinburtlh 13, 366 (1836). 16. Wandke A. A petrologic study of the Cape Neddick gabbro. Am.
3. Wheeler N. E. On the thermal expansion of rock at high tempera- J . Sci. 4, 295 (1922).
tures. Trans. Roy. Soc. Canada 3, 4 (1910). 17. Ross C. S., Foster M. D. and Myers A. T. Origin of dunites and
4. Griffith J. H. Thermal Expansion qf Typical American Rocks. Iowa olivine-rich inclusions in basaltic rocks. Am. Min. 39, 693 (1954).
Eng. Exp. Sta. Bull. 128 (1936). 18. Adams L. H. and Gibson R. E. The elastic properties of certain
5. Sosman R. B. The Properties of Silica. Amer. Chem. Soc. Mono- basic rocks and their constituent minerals. Geophys. 15, 713 (1929).
graph Series (1927). 19. Thirumalai K. and Demou S. G. Thermal expansion behavior of
6. Hockman A. and Kessler D. W. Thermal and moisture expansion intact and thermally fractured mine rocks. Preprint from Proc.
of some domestic granites. U.S. Nat. Bur. of Standards, RP 2087, 1973 Thermal Expansion Conf. (1973).
395 (1950). 20. Kingery W. D. Introduction to Ceramics. Wiley, New York (1960).
7. Thirumalai K. and Demou S. G. Effect of reduced pressure on 21. Friedman M. Residual elastic strain in rocks. Tectonophysics 15,
thermal expansion behavior of rocks and its significance to ther- 297 (1972).
mal fragmentation. J. Appl. Phys. 41, 5147 (1970). 22. Simmons G. and Nur A. Granites: Relation of properties in situ
8. Griffin R. E. and Demou S. G. Thermal expansion measurements to laboratory measurements. Science 162, 789 (1968).
of simulated lunar rocks. Proc. 1971 Thermal Expansion Conf. AlP 23. Nur A. and Simmons G. The origin of small cracks in igneous
Conf. Proc. No. 3 (1971). rocks. Int. J. Rock Mech. Min. Sci. 7, 307 0969).
9. Baldridge W. S. and Simmons G. Thermal expansion of lunar 24. Buessem W. R. and Lange F. F. Residual stresses in anisotropic
rocks. Proc. Second Lunar Sci. Conf. M.I.T. 3, 2317 (1971). ceramics. Inter. Ceram. 15, 229 (1966).
10. Baldridge W. S., Miller F., Wang H. and Simmons G. Thermal 25. Kuszyk J. A. and Bradt R. C. Influence of ~rain size on effects of
expansion of Apollo lunar samples and Fairfax diabase. Proc. thermal expansion anisotropy in Mg Ti20 ~. J. Am. Ceramic Soc.
Third Lunar Sci. Cm!f., M.I.T. 3, 2599 (1972). 56, 420 (1973).
 Geomechanics Abstracts

CONTENTS Geological factors of importance in

underground excavations 205A
General 197A Construction methods 205A
Economics 197A Groundwater problems 207A
Contracts and specifications 197A Experimental and numerical techniques 207A
Books 197A
Bibliographies 197A Surface Stroctores 207A
Conferences 197A Embankments and embankment dams 207A
Dams other than embankment dams 208A
Properties of Rocks and Soils 198A Foundations 208A
Texture, structure, composition and density 198A Slopes 209A
Fracture processes in rocks 199A Earth retaining structures 210A
Strength characteristics 199A Base courses and pavements of roads,
Shear deformation characteristics 199A railways and airfields 210A
Friction in rocks 200A Groundwater problems 210A
Physico-chemical properties 200A Influence of dynamic loads due to explosions
Permeability and capillarity 200A or earthquakes 210A
Compressibility, swelling and consolidation 200A Experimental and numerical techniques 210A

Geology 201A Comminution of Rocks 211A

Tectonic processes 201A Rock fracture under dynamic stresses 211A
Tectonic stress and strain 201A Drilling 211A
Earthquake mechanisms and effects 201A Blasting 212A
Frost action, permafrost and frozen ground 202A Cutting 212A

Rock and Soil Iml~rovementTechniques 213A

Hydrogeology 202A
Bolts and anchors 213A
Groundwater 203A
Grouting and freezing 213A
Chemical and physical changes due to
Groundwater control 214A
water 203A
Soil stabilization 214A

Underground Excavations 203A Site Investigation and Field Observation 214A

Mines 203A Geophysical techniques 215A
Tunnels 204A Presentation and interpretation of data 215A
In-situ stresses in ground and stress
around underground openings 204A Subjects Peripheral to Rock Mechanics 215A
Temporary and permanent supports 204A Snow and ice mechanics 215A
Explanation of Abstract Format
The information contained in the abstract entries themselves is described in the following example

Abstract number ,422

Author , H U D S O N , JA T R A N S P . R O A D RES. LAB. C R O W T H O R N E , BERKS, G.B

C R O U C H , SL D E P T . CIV. M I N E R . E N G . U N I V . M I N N E S O T A , U.S.A
FAIRHURST, C D E P T . CIV. M I N E R . E N G . U N I V . M I N N E S O T A , U.S.A

l Affiliation
Title , Soft, stiff s e r v o - c o n t r o l l e d testing m a c h i n e . A r e v i e w with reference to
r o c k failure. 23F, I T , 54R.
I Number of references

Namberoftables

Number of figures
Source , EN( ;NG. GEOLOGY, V6, N3, 1972, P 1 5 5 - 1 8 9 .

1 Page

Number

Volume

Title of journal
Abstract , T h i s r e v i e w c o n t a i n s a brief h i s t o r y of testing m a c h i n e s a n d a
detailed discussion of the principles i n v o l v e d in r o c k f a i l u r e . . .

F o r this a n d future issues, t h e r e a r e t h r e e f o r m s of p r e s e n t a t i o n for the references. T h e y a r e as follows: (1) Reference c o n t a i n i n g a u t h o r ,

affiliation w h e n k n o w n , title, a n d s o u r c e of p u b l i c a t i o n ; (2) As for (1), b u t with a n e x t e n d e d title; (3) As for (1), b u t with a n a b s t r a c t .

Subject and Author Index

C u m u l a t i v e yearly subject a n d a u t h o r indexes a r e b e i n g issued at the e n d o f e a c h year.

Codes for Coax'ties

A Austria E Spain,Balearic Islands, IS Iceland RCH Chile
ADN Aden Canary Islands, RH Haiti
AL Albania Spauish. Guinea. J Japan RI Indone~a
AND Andorra SPanish Sahara JA Jamaica RL Lebanon
AUS Australia. Norfolk Islands EAK Kenya JOR Jordan RNR North Rhodesia
EAT Tanganyika RNY Nyautland
EAU Uganda K Cambodia RSM San Matino
B Belgium EAZ Zanzibar, Peraba KWT Kuwait RSR Southern Rhodesia
BDS Barbados EIR Republic of Ireland RU RuandaUrundi
BG Bulgaria EQ Ecuador L Luxembourg
BH British Honduras ET Egypt LAO Lags
BL Basutoland S Sweden
SD Switzerland
BP Bechuanaland F France MA Morocco
SF Finland
BR Brazil FL Liechtenstein MC Monaco SCP Singapore
BRG British Guiana MEX Mexico
SK Sarawak
BRN Bahrain GB Great Britain and MS Mauritius
BRU Brunei Northern Ireland SME Surinam
SP BritishSomaliland
BS Bahamas GBA Alderney N Norway SU Unionof Soviet
BUR Burma GBG Guernsey NA Netherlands Antilles
Socialist Republics
GBJ Jersey NGN Netherlands New Guinea SWA South West Africa
GBM Isleof Man NIC Nicaragua SY Seychenes
(" Cuba GBY Malta, Gozo NL NetherlandsIHollandl SYR Syria
CB BelgianCongo GBZ Gibraltar NZ New Zealand
CDN £anada GCA Guatemala
CH Switzerland GH Ghana P Portugal T Thailand
CL Ceylon GR Greece.Crete. PA Panama TD Trinidad and Tobago
CNB British North Borneo. Dodecanese Islands PAK Pakistan TN Tunisia
Labuan PE Peru TR Turkey
CO Colombia Pl Philippine Islands
CR CostaRica H Hungary PL Poland USA UnitedStates of America
CS Czechoslovakia HK Hong Kong PTM Malaya
CY Cyprus PY Paraguay
I Italy, Sardinia Sicily VN Viet-Nam
IL Israel
D German.~ IND India R Rumania YU Yugoslavia
DK Denmark IR lran IPersial RA Argentina
DOM Dominican Republic IRQ lraq RC Formosa ZA Unionof South Africa

ii