KINETICS OF CLAY D E H Y D R A T I O N

By P. MURRAY* and J. WHITEr

* A t o m i c Energy Research E s t a b l i s h m e n t , Harwcll
t Refractories D e p a r t m e n t , University o f Sheffield
[Read 5th November, 1954]
ABSTRACT

Kinetic studies have shown that the isothermal dehydration of the clay
minerals proceeds according to a first ordcr law enabling velocity
constants to be evahmted for different temperatures. From these, the
Arrhenius parameters have been determined and on this basis, the clay
minerals may be classified into three main groups, viz., kaolinites,
secondary mica clays and montmorillonites. No significant difference is
shown by comparison of the Arrhenius plots for the kaolinites and
halloysites but the secondary mica clays decompose more rapidly than
the kaolinites at comparable temperatures. Evidence is provided to
illustrate that the constants obtained are associated with the basic process
involving interaction of hydroxyl groups in the clay mineral lattice,
Indirect evidence also suggests tnat the irreversibilityof the dehydration
is due to lattice collapse resulting in the formation of highly stable
configurations of Si4q, AP ,- and 0 2 - atoms. The first order nature
of the dehydration reaction is shown to be the underlying basis of the
type of curve obtained in differential thermal analysis. Two important conclusions are that:--(a) 'Ihe clay mineral is only approximately 70,~-o decomposed at the peak on the thermal analysis crave.
(b) Heating rate has a mar~ed effect on the peak temperature. Mathematical equations have been developed to obtain the effect of heating at
a constant rate on the progress of a reaction for which the isothermal
velocity constants are known. The effect of sample weight and dilution
factors are discussed, together with the kinetic problems associated with
mixtures of clay minerals.
EXPERIMENTAl.

The progress o f the d e h y d r a t i o n reaction was determined by repeated weighings at the reaction temperature, p r e l i m i n a r y heating o f
the clays having eliminated " i n t e r m e d i a t e " w a t e r w i t h o u t causing
loss o f fully c o m b i n e d water. "Ihe t e m p e r a t u r e s selected were
430~
for all clays o t h e r than bentonites and 530~ for the latter.
T h e m e t h o d s used in the t r e a t m e n t o f the experimental d a t a have been
described previously, ( M u r r a y and White, 1949). One p o i n t that
m a y be emphasised is that the end p o i n t loss in weight ( L ~ ) a t t a i n e d
at a given reaction t e m p e r a t u r e must be known accurately in deciding the o r d e r o f a reaction using the conventional m e t h o d o f
p l o t t i n g the log (fraction u n d e c o m p o s e d ) vs. time (t). Using the
value o f L ~ o b t a i n e d from the A t p l o t ; the n o r m a l plot o f the log
(fraction u n d e c o m p o s e d ) vs. time c o u l d be constructed to give a final
check on the value o f velocity constant K. This type o f plot showed
t h a t deviations from the first o r d e r law occurred at high degrees o f
d e c o m p o s i t i o n (approx. 90~
and there was a definite tendency for
some slowing up o f the reaction to t a k e place in this range. The
m o s t r e a s o n a b l e explanation for this deviation is that the lattice collapses at high d e c o m p o s i t i o n and prevents the last final traces o f water
255

256

e. MURRAY AND J. WHITE

being removed and hence the reaction appears to stop completely.

An alternative point of view which gives rise to the same type of
deviation is that decomposition proceeds by a nucleation mechanism
and that impingement of the growing nuclei is responsible for the
falling off of the decomposition rate. It may be noted that a similar
effect has been found in the thermal decomposition of beryllium
hydroxide. (Murray and Thackray, 1954).
Having determined K (the velocity constant) values at different
temperatures, a plot oflogl0K vs. 1/T~
can be made for each clay
and the values of the Arrhenius parameters E, the activation energy,
and log10 PZ in the eqn.
E
log10 K=lOgl0 PZ
4.575T
can be determined. In the above equation, Z is the collision number
(i.e., number of collisions/sec, when there is only 1 mol. of
reactant/e.c.) and is usually taken as 1014. P is known as the steric
factor and is used to make allowance for any effect causing deviations
from ideal behaviour. The values of the Arrhenius parameters
obtained for the different types of clay are given in Table 1.
TABLE1--Arrhenius Constants for the Different Types of Clay Mineral
Clay
Supreme kaolin
K. Plastic kaolin
Newton Abbot
Raw Bottle clay
E. halloysite
D. D. halloysite
Kingb0nd
Kloidex
Wyoming bentonite
Pembina bentonite
Fulbond T

Type
Kaolinite
Kaolinite
Kaolinite
Fireclay
Halloysite
Halloysite
Secondary mica
Secondary mica
Na+ montmorillonite
Ca2+ montreorillonite
Ca2+ montmorillonite

E
(cals./mole of
water vapour)

lOglO PZ

44,830
36,690
40,660
38,990
36,740
44,160
40,260
34,800
57,640
36,600
41,200

9-07
6.96
7.98
7.56
6.90
8.82
8.17
6.74
10.43
5-82
6.92

It will be noted that there is a connection between the parameters

and as E increases so does logt0 PZ. The velocity constants are also
compared in Fig. 1 from which it will be seen that the results fall into
three main groups comprising kaolinites, secondary micas and
montmorillonite clays. A s regards variation within each group,
several features appear to be significant :--(a) The kinetic results do
not appear to show any significant difference between the kaolinites
and halloysites. (b) Secondary micas give much higher velocity
constants than the kaolinites and this would appear to be due to
differences in the log~0 PZ term rather than in the E values. (c) The
Ca 2+ montmorillonites have much lower E values than the Na+
montmorillonite clay examined, the E values being of the same order
as those of the kaolinite and secondary mica clays. This similarity

KINETICS OF CLAY DEHYDRATION

257

indicates that the combined water in the Ca z+ montmorillonites is

bound less strongly and hence on a comparison of the E values alone,
they should decompose in the same temperature range as the kaolinites and secondary micas. The fact that the Ca z+ clays do not
decompose until higher temperatures are used might indicate that the
Ca z+ ions exert an inhibiting influence on collisions between O H O

SUpJ~EIvlIE K4OLIN

X
+

RAW BOTTLE CLAY

/(mGSON~

XLOIDEX
~fOM I~'~ ~ENTONiT"~
pEblBINA B~NTOrOTE

F/
~"

"4

~d
o

.$

.&

,4.2

,,.

r
,.~o
~.]o
-r'x
FIG. 1. Comparison of the Arrhenius plots for kaolinites, secondary micas
and montmorillonites
~=MontmoriUonites;
~ =Secondary micas.
~-- ~ Kaolinites;
boo

groups, this effect being reflected in the log10 PZ term. This assumes
that the kinetic measurements are related to the actual reaction
between O H - groups in the clay mineral lattice.
In all the above experiments, the crucible was filled to a constant
level with the clay; the amount being regulated by tapping down to
this minimum volurne. Experiments with varying sample weight

258

P. MURRAY AND J. WHITE

showed that a decrease in weight of sample taken, by approximately

50%, results in an increase of approximately 50~o in K value and this
increase in K was reflected in a higher log10 PZ, the activation energy
being the same for the two cases.
MECIIANISM OF THE DEHYDRATION REACTION

That the reaction appears as a first order one is due to the fact
that reactions of this kind in the solid state do not depend on the
collision of ions or ionic groups as a result of translation. The
ions will, in the main, remain vibrating about fixed positions and
reaction will occur when the energy of vibration reaches a certain
value relative to that possessed by a near neighbour at that particular
instant. In tbe kaolinite lattice, for example, the O H - ions occur in
tetrahedrally arranged groups and it is perhaps feasible to suggest
that decomposition occurs through interchange of hydrogen ions
between the molecules of these groups, the probability of the interchange occurring beiag constant at any temperature.
This would
lead naturally to a first order law. Whether or not propagation of
the reaction will occur round a reacted nucleus cannot be said. If it
does so, such propagation must be rapid relative to the rate of nucleation as only then would a first order law be obeyed.
The effect of vapour pressure is also a factor to be considered.
In experiments carried out at atmospheric pressure, it seems reasonable to assume that during the dehydration, the pressure of the water
vapour within the clay sample is approximately 1 atmosphere.
Workers who have studied decomposition of clays in vacuo, however,
have generally found that the dehydration temperature as given by
the "step" on the well known 3h dehydration temperature curves is
appreciably lower than at atmospheric pressure. It would appear
from this that at low pressures the dehydration reaction can proceed
with appreciable velocity at lower temperatures than are normally
observed. This would appear to indicate that the onset of the
reaction cannot be controlled entirely by kinetic factors as assumed in
the previous considerations of irreversibility, (Murray and White,
1949) and hence the question of the essential nature of the irreversibility is raised. It seems possible to account for the experimental
data in clay dehydration by postulating a reversible "first stage" of
the reaction, a simple loss of water occurring whenever the temperature or external conditions are such that the vapour pressure of the
clay exceeds that of its environment, lrreversibility could then be
accounted for by postulating that the first stage is immediately
followed by a second irreversible stage corresponding to collapse of
the lattice with the formation of new highly stable configurations of
the Si 4~ , A13+ and 0 2- ions. Some indirect evidence is provided by
the fact that isothermal experiments at temperatures less than those
used in previous kinetic work give K values which fall away steeply
from the straight lines of the Arrhenius plots as the temperatures of
the "steps" on the 3h dehydration temperature curves are approached. This behaviour is to be expected as a transition temper-

KINETICS OF CLAY DEHYDRATION

259

ature is approached from a higher temperature since the velocity of

the reaction must fall to zero at that temperature. Furthermore, this
fall will probably not be abrupt and discontinuous at the transition
temperature but will begin to manifest itself by a progressive drop
beginning at a ,.'lightly higher temperature. (See Murray and White,
1949, and Stone, 1952), the latter having shown that the peak on
d.t.a, curves can be displaced to higher temperatures by increasing the
external water vapour pressure.
CALCULATION OF DIFFERENTIAL CURVES FROM KINETIC DATA

To test the validity of the kinetic constants obtained, they have

been used to predict the endothermic peak temperatures which would
be obtained on a thermal analysis curve. This is possible since the
velocity constant K can be expressed as a continuous function of
temperature. Thus the variation of K with temperature for Supreme
kaolin is given by log10 K---9.07--9'T
w"'~" (T being the absolute
temperature). This relation holds in the experimental range studied
(i.e., over a range of 50~
Assuming that we can use this relation
to calculate velocity constants in the range 400~176
the progress of
the dehydration reaction under conditions of continuous temperature
rise can be deduced by numerical integration of successive one minute
interval periods, the temperature remaining constant until the end of
the period.
Expressing K in reciprocal seconds and taking At as one minute,
the equations reduce to
A L = = - 6 0 K . Loo
where L--:the fraction loss in weight during the interval
K - t h e velocity constant corresponding to the temperature T o
and Loo- the fraction of undecomposed material present at the
beginning of the interval.
A curve obtained in this way for Supreme kaolin at 10~ per
minute is very similar to plotting weight loss against temperature,
i.e., that obtained in thermal analysis. The percentage decomposition
becomes appreciable at 430~ reaches a maximum at 587~ and is
finally complete at 630~
The peak temperature obtained, viz.,
587~ is also of the right order, being slightly higher than those
reported in thermal analysis studies on kaolinite. The reason for the
maximum is also clearly seen. The percentage loss in weight per
minute at any temperature or in any interval is the product of two
factors:--(1) the appropriate K value and (2) the amount of undecomposed material remaining at that temperature. The K value
increases as the temperature increases and the amount of undecomposed material decreases. At the maximum, the product of these two
factors is a maximum and then starts to decrease since the amount
of undecomposed material is decreasing more rapidly than K is
increasing. The result is the asymmetric curve characteristic of d.t,a.
An important point revealed by this calculation also, is that the Clay

260

v.

MURRAY AND J. WHITE

is not completely decomposed at the peak. The rate of return of

the curve in thermal analysis to the base line is thus not wholly governed by factors such as thermal diffusivity as has been previously
assumed.
The curve also shows that two characteristics may be used for
estimation of the amount of clay mineral present, viz., the peak
height which is the maximum percentage of clay decomposing per
minute, or the area under the curve which represents the total percentage decomposition. In this instance, the peak height is 13.3%
per minute and the area under the curve represents 100% decomposition. If we had started'with 50% of clay, the curve obtained
would be half the size so that both peak height and the area under
the curve are proportional to the amount of mineral present. Both
these quantities are, of course, used in the estimation of amounts of
minerals from thermal analysis curves.
Curves for different heating rates may also be obtained by the same
procedure as described above. In the case of 5~
the 10~
interval is kept the same so that in this case At--2 mins. and the
equation becomes A L = - - 120 K L~.
For a heating rate of 20~
keeping the same 10~ interval,
At= 89 rain., and the appropriate equation is A L = - - 3 0 K Loc. As
the heating rate is increased, the peak temperature is increased and
the peak height is lowered. The differences between the peak
temperatures for various heating rates are large and of the order of
20~ for rates of 10~ and 20~
Furthermore the lower the
heating rate, the sharper and more pronounced is the ,peak. Slower
heating rates therefore increase the sensitivity of differential methods
though this effect cannot be perhaps fully utilised in thermal methods
due to the necessity of maintaining a temperature differential between
specimen and blank. Since variations in heating rate also have an
effect on the height of the peak, the results also show that the area
under the curve is a more reliable index of the amount of mineral
present.
The above method of numerical integration although essentially
correct in principle was thought to be to some extent approximate,
and accordingly, a more rigorous mathematical treatment has been
worked out (Murray and White, 1955). Thus it was found that the
peak temperature on the differential curve for a heating rate of
10~
is related to the isothermal kinetic constants by the
equation
-E
- E

RTm 2

6 PZ

e RT---~

where Tm is the peak temperature in ~

and E and PZ are the
Arrhenius parameters.
For other heating rates, e.g., 5~ and
20~
the equation is similar, the only difference being the
numerical factor in front of the right hand term which is 12 and 3
respectively. Peak temperatures calculated from these equations
were in good agreement with those obtained using the previous

KINETICS OF CLAY DEHYDRATION

261

numerical integration method, the latter being lower by approximately 5~

The peak temperature for all the clays investigated as
determined from the above equations showed that (1) the peak
temperatures for the kaolinites, the fireclay and the halloysites
investigated were approximately the same, (2) the peak temperatures
for the secondary micas were 25~176 lower than those for the
T[IdPs

.!

55"o

9r

c,~

756

7,

FIG. 2. Comparativedifferentialweight loss curvesfor the montmorillonites

and kaolin. A--Wyoming bentonite, B--Pembina bentonite, C--Supreme
kaolin.
kaolinites and (3) the peak temperatures for the Ca 2+ montmorillonites were lower than that of Wyoming Bentonite by approximately
20o-25oc.
So far, in the treatment of the differential curves, only the effect of
the Arrhenius parameters has been considered, the differences in peak
heights for the various clays being due to differences in E and log10
PZ. To obtain a complete comparison of the curves of different
clays however, the relative amounts of combined water, i.e., the
different L ~ values must be taken into account. The relative
heights of the different peaks may be obtained by multiplying the
D

262

P. MURRAY AND J. WHITE

ordinates (calculated in the integration methods as % weight loss/

rain.) by factors proportional to the L ~ values of the clays concerned.
For this purpose, a factor for supreme kaolin with L ~ - - 1 1 . 9 was
taken as unity and the factors for the other clays were calculated on
this basis. Typical curves obtained in this way for supreme kaolin
are shown in Fig. 2 the main point being that on this basis the peak
height for supreme kaolin is approximately 3.5 times that for Wyoming Bentonite. Calculations have been made using the kinetic data
for both supreme kaolin and Wyoming bentonite in varying weights
and as expected, the peak temperature was lowered with decrease in
weight and furthermore the results showed that the effect became
more marked the greater the difference in sample weight.
TABLE 2--Comparison of calculated and experimental differential Peak
Temperatures

D.t.a.
Peak Temp.*

Clay
Supreme kaolin
K. Plastic kaolin
Newton Abbot
D. D. halloysite
E. halloysite
Kloidex
Pembina bentonite
Wyoming bentonite
Fulbond T

554
554
539
539
531
526
663
690
663

Sample weight
1 grn. in all cases.
Heating rate of
inert sample
10~
*All figures given are clay

D.t.a.
Peak Temp.*
(B.C.R.A.)

Diff.
Peak Temp.*
(calc.)

577
656

589
584
590
595
593
561
691
717
691

554
556
700
S~mple weight
0.7 - 0.8 gms.
Heating
rate
12~

Samples packed
to constant dens i t y. Heating
rate tO~

temperatures in ~

COMPARISON WITtI RESULTS OBTAINED BY THE D.T.A. METHOD

Two considerations must be borne in mind when comparing the

above differential curves with thermal analysis results. (1) The
comparison assumes that the heat value for the reaction for a given
quantfty of water evolved is constant over the entire reaction range
and does not require or imply any knowledge of the absolute'value of
the heat of reaction. (2) The differential weight loss curves are for
assumed heating rates of 5~ 10~ and 20~
in the clay sample.
In thermal analysis the heating rate is generally controlled with
reference to the inert material so that the heating rate in the clay
sample is variable and falls almost to zero as the decomposition
proceeds. This has an effect on the peak temperature and also under
these conditions, the variation of peak temperature with heating rate
(of the inert material) should be smaller. Nevertheless the kinetic
interpretation suggests that the heating rate should be controlled
within narrow limits.

KINETICS

OF

CLAY

DEHYDRATION

263

Generally the peak temperatures of the differential weight loss

curves are higher than the figures obtained by other workers using the
d.t.a, method. More exact data on these specific bond clays have
been obtained; check results were also carried out at B.C.R.A. and
a comparison is given in Table 2.
The following specific features may be noted. Comparing the
two sets of thermal analysis results, it will be noted that the B.C.R.A.
results are approximately 20 ~ higher but are consistent in that they
indicate a lower peak temperature for the halloysites than for
the kaolinites. This is m agreement with the results of
Grimshaw, Heaton and Roberts (1945) who found a difference
of 34 ~ in the peak temperatures. The predicted peak temperatures
are up to 50 ~ higher and in addition no difference is predicted between the kaolinites and the halloysites. As regards
the general difference between the calculated curves and the
thermal analysis results, one point which may be important is
that in the latter case the heating rate in the clay specimen itself
falls off considerably as the reaction proceeds; this would be
expected to result in some lowering of the peak temperature
although it is unlikely that this would be of the order given by
the results in Table 1.
In the thermal analysis work, the sample weight was the same
(0.8 g) for the kaolinites and the halloysites, whereas in the kinetic
work which was carried out on a constant volume basis, the sample
weights were 2.25 g for Supreme kaolin, 8.15 g for E halloysite and
6.5-7-5 g for D.D. halloysite. For a given clay the weight of sample
taken (keeping the packing density constant but reducing the volume)
is related to the peak temperature of the differential weight loss curve,
the lower the sample weight, the lower the peak temperature. It
seems reasonable, therefore, to account for the discrepancies of peak
temperatures on this basis that the effect of sample weight is responsible and that curves derived from kinetic data on the same
weight of sample would show a difference in peak temperatures
between kaolinite and halloysite of the order found in thermal
analysis, and in fact using a constant packing density basis no
difference showed between D.D. halloysite and K Plastic kaolin.
(See also Grimshaw, Heaton and Roberts, 1945). The effect of
packing density is also important, thus this was kept constant in the
above work and the volume reduced for the lower sample weights. It
is possible, however, that in thermal analysis the packing density
variation consequent on keeping the weight constant will give larger
effects than these found here and may have some bearing on the
general trend for the kinetic results to be higher than in practice.
Whether or not this effect results in any modification of the conclusions reached concerning the peak temperature of the other clays
has also been examined using the results for Supreme kaolin, and the
peak temperatures were converted to equal weights of 2.25g; the
results obtained showed that all the "kaolinite" type clays gave peak
temperatures on this basis lower than that of Supreme kaolin, and

264

p. MURRAY AND J. WHITE

no distinction could be drawn between K Plastic kaolin, Newton

Abbot ball clay, Raw Bottle clay and the halloysites.
It is also of interest to compare the ratio of the peak heights obtained for the clays by the differential weight loss method with the
ratio as found by the d.t.a, method. The height of the Supreme
kaolin peak is 3.5 times that of the Wyoming bentonite peak. This
is approximately the same ratio as is indicated by a comparison of the
peak heights of d.t.a, curves determined on equal weights of the two
clays and would appear to indicate that the heat value of the reaction
in terms of moles of water formed is of the same order. This is
confirmed by the values obtained by Allison (1954) (600-900 for kaolinites and 700-1100 cals./gm, of water for montmorillonites). The p e a k
heights for the halloysites (particularly E. halloysite) are lower than
that for Supreme kaolin, while the thermal analysis data on equal
weights of the clays indicates that the peak heights of the halloygites
are either of the same order or higher. This was thought to indicate
a higher heat value for the decomposition of halloysite than kaolinite
but Allison (lot. cir.) found little difference. It should be pointed
out, in addition, that the total water contents assumed are based on
values derived above 430~
In the case of halloysities, the water
lost between 110~ and 430~ is abnormally high and, as appreciable
time is required to eliminate this water at 430~ some of this might
contribute to the peak height observed on a heating curve. The peak
height of a differential weight loss curve is expressed as a rate but the
effect of the 110~ to 430 ~ water may be reflected in a higher L ~
factor.
CONCLUSIONS

Summarising this work on the kinetics of clay dehydration and its

application to thermal analysis, it may be stated that the fact that a
first order kinetic law is obeyed isothermally is the underlying basis
of the type of curve obtained in thermal analysis. The most important conclusions are (a) the clay is only approximately 70% decomposed at the thermal analysis peak, (b) heating rate has a marked
influence on the peak temperature and (c) the effects of sample weight
and packing density are important and would appear to be at least
partially responsible for the thermal analysis differences between
kaolinites and halloysites.
Acknowledgment. The authors are grateful to the British Iron and Steel
Research Association for the award of a research bursary to one of them (P.M.)
to enable this work to be carried out.
REFERENCES
Allison, E. B. 1954 A.E.R.E. Report, M / R 1408.
Barrett, L. R. 1946 Trans. Brit. Ceram. Soc., 45, 340.
Grimshaw, R. W., Heaton, E. and Roberts, A. L. 1945 Trans. Brit. Ceram. Soc.,
44, 76.
Guggenheim, E. 1926 Phil. Mag., 7, 538.
Murray, P. and White, J. 1949 Trans. Brit. Ceram. Soc., 48.
Murray, P. and White, J. 1955 Trans. Brit. Ceram. Soc., In press.
Murray, P. and Thackray, R. W. 1954 A.E.R.E. Report, M/R 1410.
Stone, R. L. 1952 J. Amer. Cer. Soc., 35, 90.