Available online at www.sciencedirect.

com

Ceramics International 36 (2010) 2493–2498

www.elsevier.com/locate/ceramint

Short communication
Dehydration reactions and kinetic parameters of gibbsite
Boquan Zhu *, Binxiang Fang, Xiangcheng Li
The State Key Laboratory Breeding Base of Refractories and Ceramics, Wuhan University of Science and Technology,
Wuhan 430081, China
Received 18 November 2009; received in revised form 25 March 2010; accepted 18 June 2010
Available online 3 August 2010

Abstract
Dehydration reactions and the corresponding kinetics of gibbsite [g-Al(OH)3] were analyzed using non-isothermal thermoanalysis and the
Kissinger equation. It is concluded that the starting temperature of the dehydration reaction of g-Al(OH)3 rises with increasing heating rate of the
system. At a heating rate of 10 8C min1, g-Al(OH)3 has lost part of its crystalline water, and was completely transformed into boehmite (g-
AlOOH)) at about 317 8C. However, g-AlOOH did not lose the residual crystalline water entirely, and did not change into amorphous Al2O3 until
the system was above 700 8C. The kinetic energy needed to convert g-Al(OH)3 to g-AlOOH, and g-AlOOH to amorphous Al2O3, was calculated by
differential scanning calorimetry (DSC) with activation energies of 108.50 and 217.24 kJ mol1, pre-exponential factors of 2.93  109 and
8.30  1013, and reaction orders of 0.96 and 1.06, respectively. The kinetic parameters of dehydration reactions for g-Al(OH)3 obtained using the
derivative thermogravimetric method (DTG) are very similar to that of obtained by DSC.
# 2010 Published by Elsevier Ltd and Techna Group S.r.l.

Keywords: Thermal analysis; Activation energy; Alumina dehydration; Gibbsite dehydration; Kissinger equation

1. Introduction kinetic issues when dehydrating g-Al(OH)3 using X-ray

diffraction (XRD). Non-isothermal analysis, such as differential
Various forms of hydrated alumina include trihydrate and thermal analysis (DTA), differential scanning calorimetry
monohydrate variants. The trihydrate is divided into gibbsite [g- (DSC), and thermogravimetry (TG) are other important methods
Al(OH)3], bayerite (a-Al(OH)3), doyleite and nordstrandite; used to evaluate the phase transformation and the kinetics of
while the monohydrate form is divided into boehmite inorganic synthesization [13–15]. When compared with iso-
(g-AlOOH)), diaspore (a-AlOOH)) and psendoboehmite [1,2]. thermal methods, several advantages emerge in the use of non-
The trihydrate form of alumina, especially g-Al(OH)3, is not only isothermal techniques, such as simple operation, a broad
the main raw material source used to produce aluminum oxide by temperature testing range, and the applicability to reactions
the Bayer method in industry, but is also the hydrated product with a high rate of phase transformation with temperature [16].
from calcining r-alumina, an important binder used in refractory Perić et al. [17] investigated the dehydroxylation of g-Al(OH)3
castables [3–6]. The dehydration temperature, and the corre- into g-AlOOH by DSC analysis and found the activation energy
sponding kinetic parameters of g-Al(OH)3 and its crystalline ranged from 132.92 to 142.26 kJ mol1. However, the
phase transformation into alumina, have been the focus of endothermic peak temperature (257–277 8C) of transformation
research by the ceramics and refractories industries [7–10]. from g-Al(OH)3 into g-AlOOH in their research [17] is much
Thermal analysis and X-ray diffraction are usually used to lower than the results obtained at the same heating rate (300–
evaluate the phase transitions of alumina. For example, Yang and 320 8C) by other researchers [18,19].
Yan [11] reported the kinetic study of the phase transformation of In this study, the dehydration temperature of g-Al(OH)3 was
g-alumina to a-alumina using gravimetric analysis measure- determined, and its corresponding kinetic parameters were
ments at a constant temperature. Wang et al. [12] discussed the calculated by DSC analysis. Furthermore, derivative thermo-
gravimetric (DTG) analysis was adopted to verify the data
determined using DSC analysis. DTG values can be obtained as
* Corresponding author. Tel.: +86 27 68862566; fax: +86 27 68862606. the first derivative of mass loss to time when the corresponding
E-mail address: [email protected] (B. Zhu). TG data are measured [20,21].

0272-8842/$36.00 # 2010 Published by Elsevier Ltd and Techna Group S.r.l.

doi:10.1016/j.ceramint.2010.07.007
2494 B. Zhu et al. / Ceramics International 36 (2010) 2493–2498

In order to simplify the description of the kinetic equations the commercial mathematical software, Origin 70 (Origin
in the paper, all the nomenclature used in them and the Lab, USA).
corresponding meanings are listed as follows:
3. Results and discussion
A – apparent pre-exponential factor
aA – conversion degree reaction extent of reactant A* 3.1. Dehydration reaction
b – heating rate
E – activation energy The TG and DSC curves of sample at the heating rate of
f ðaA Þ, f ðwÞ – functions of kinetic model 10 8C min1 are shown in Fig. 2. Two major endothermic
k – reaction rate constant peaks and one exothermic peak were observed on the DSC
n – reaction order curve between 100 and 800 8C. The first endothermic peak
R – universal gas constant (8.314 J mol1 K1) begins at 270 8C and ends at 317 8C, with a peak temperature
S – peak shape factor occurring at 300 8C. The second endothermic peak began at
T – thermodynamic temperature 475 8C and ended at 550 8C, with a minimum occurring at
Tp – peak thermodynamic temperature of DSC curve 520 8C. The exothermic peak was minor, with a maximum at
t – reaction time 725 8C. The TG curve corresponding to endothermic peaks
w – residual mass extent of system indicates a large mass loss through 317 8C, with other weight
wt , w0 – residual mass at t time and 0 time of system, changes being minor by comparison.
respectively The degree of g-Al(OH)3 dehydration with a heating rate of
10 8C min1 at different temperatures is given in Table 1. Up to
the end temperature of the first endothermic peak (317 8C), the
2. Experimental procedure cumulative mass loss of the sample was 23.87 wt.%, which
approached the theoretical mass loss from the conversion of
In order to avoid interference by impurities on the g-Al(OH)3 to g-AlOOH (23.08 wt.%). It is concluded that the
heating effect of g-Al(OH)3, analytically pure first endothermic peak (270–317 8C) is induced by the
{w[Al(OH)3] > 97.50%, Kermel, Tianjin, China} was uti- conversion of g-Al(OH)3 to g-AlOOH, and the second
lized in this study. The XRD pattern of this material (Fig. 1) endothermic peak (475–550 8C) is induced by the conversion
indicated a small amount of g-AlOOH exists in it (the from g-AlOOH to amorphous Al2O3, according to the
characteristic diffraction peaks of g-Al(OH)3 and g-AlOOH following equations respectively:
are 0.484505, 0.437383, 0.432421 nm and 0.610700,
g-AlðOHÞ3 ! g-AlOOH þ H2 Oð23:08 wt:%lossintheoryÞ
0.316050, 0.234460 nm, respectively). The sample was
ground into powder with a particle size less than (1)
0.045 mm and a constant mass of 10  0.2 mg used in
g-AlOOH ! 0:5Al2 O3 þ 0:5H2 O
every test. TG and DSC data were measured simultaneously
ð11:54 wt:% more loss in theory compared with g-AlðOHÞ3 Þ
using a differential scanning calorimeter (Model STA449/6/
G, Netzsch, Germany) with a flowing atmosphere of dry (2)
argon (50 cm3 min1) and a heating rate of 5, 10, 15 and It is also observed from Table 1 that the sample still exhibits
20 8C min1, respectively. All test data were processed using mass loss during the temperature range from 550–825 8C,

Fig. 1. XRD pattern of a sample high in g-Al(OH)3 used to study its decom-
position. Fig. 2. TG and DSC curves of a test sample heated at 10 8C min1.
 B. Zhu et al. / Ceramics International 36 (2010) 2493–2498 2495

Table 1
Quantitative relationship between temperature and dehydration of g-Al(OH)3 (heating rate of 10 8C min1).
Temperature intervals (8C)
<100 100–270 270–317 317–475 475–550 550–650 650–725 725–825
Mass loss (%) 0 5.92 17.95 3.57 2.69 1.38 1.64 0.22
Cumulative mass loss (%) 0 5.92 23.87 27.44 30.13 31.51 33.15 33.37

indicating that the complete dehydration of g-AlOOH does not the conversion from g-AlOOH to amorphous Al2O3. The
terminate at the temperature corresponding to its endothermic Kissinger method was used to calculate kinetic parameters of
peak, but is still occurring at higher temperatures. the two dehydration reactions [g-Al(OH)3 and g-AlOOH],
respectively.
3.2. Kinetic parameters of the dehydration reaction
3.2.1. Determination of kinetic parameters of the first stage
A general heterogeneous reaction is as follows: by DSC curves
Kissinger made some differential derivation based on
A ! BðsÞ þ CðgÞ (3)
Eqs. (4) and (6), deriving the following equation [26]:
The reaction rate equation is given by !  
b AR E 1
daA ln 2 ¼ ln  (7)
¼ k f ðaA Þ (4) Tp E R Tp
dt
or As for different heating rates (b), Tp is also different. A
dw linear plot of lnðb=T 2p Þ against 1/Tp can be obtained according
 ¼ k f ðwÞ (5) to Eq. (7). Furthermore, E and A can be calculated from the
dt
slope and intercept of the fitted line.
where f ðaA Þ ¼ ð1  aA Þn and f ðwÞ ¼ wn , and k is calculated Partial DSC curves of samples at heating rates of 5, 10, 15
according to the following equation: and 20 8C min1 showing where the maximum endothermic
  peak occurs are given in Fig. 3. All beginning, ending, and peak
E
k ¼ A exp  (6) temperatures of each endothermic curve shift to higher
RT
temperatures with an increase of heating rate; which means
The values of E, A and f ðaA Þ [or f ðwÞ] are the three kinetic that dehydration temperature of g-Al(OH)3 is not fixed, but
factors describing a reaction [22]. increases with increasing of heating rate. The fitted linear
Most of the methods that calculate kinetic parameters are equation between lnðb=T 2p Þ and 1000/Tp has a high coefficient
derived from Eqs. (4)–(6); among which, the Kissinger method of correlation (R2 = 0.996), as shown in Fig. 4 [Yand X stand for
is the one most frequently used in non-isothermal thermo- lnðb=T 2p Þ and 1000/Tp, respectively]; and results in a value of
analysis [23–25]. Dehydration of g-Al(OH)3 is divided into 108.50 kJ mol1 for E, and 2.93  109 for A, respectively. The
two stages: the conversion from g-Al(OH)3 to g-AlOOH, and activation energy of conversion from g-Al(OH)3 to g-AlOOH

Fig. 3. DSC curves of the same g-Al(OH)3 starting material at different heating Fig. 4. Plotting lnðb=T 2p Þ vs. 1000/Tp during the first dehydration stage
rates during the first dehydration stage. (between 200 and 400 8C).
2496 B. Zhu et al. / Ceramics International 36 (2010) 2493–2498

obtained in this study is lower compared with the result of Perić

et al. [17] (132.92–142.26 kJ mol1) mainly because of the
different decomposition temperature of g-Al(OH)3.

3.2.2. Determination of kinetic parameters of the first stage

by DTG curves
DSC curves indicate the energy changes of g-Al(OH)3
during dehydration, while TG curves reflect weight changes of
the process. Plotting the first derivative of weight loss versus
time, i.e. DTG curve, displays more sensitivity to weight
changes than temperature by giving greater definition to small
weight changes [20,27]. For that reason, DTG curves were used
to calculate kinetics parameters of dehydration of g-Al(OH)3 to
verify the accuracy of results derived from DSC curves.
Imitating the differential method of Kissinger, the following
equation based on Eqs. (5) and (6) was obtained: Fig. 6. Fitting lines of ln(  DTG) vs. 1000/T at different conversion degree
during the first dehydration stage.
 
dw E
ln  ¼ þ ln½A f ðaÞ (8)
dt RT rates, we can determine the corresponding DTG and T values at
certain conversion degrees, fit a line according to different
According to the definition, DTG ¼ dw=dt is the DTG value ln(  DTG) and 1/T values, then determine E from slope of the
at time t. Using a group of DTG curves at different heating fitted line.
The conversion of the first stage [g-Al(OH)3 to g-AlOOH]
occurs between 200 and 400 8C, so the degree of conversion
can be calculated presuming that dehydration starts at 200 8C
and ends at 400 8C. According to the temperature correspond-
ing to the given conversion degree (Fig. 5a), the time required to
reach the temperature can be determined, and the correspond-
ing DTG value plotted (Fig. 5b). Samples with different heating
rates result in different time and different DTG values at the
same conversion degree, then a line of ln(  DTG) against 1/T
can be fitted. In Fig. 6, 4 fitted lines of ln(  DTG) against
1000/T at conversion degree of 20, 40%, 60% and 80% are
represented respectively as follows:

a ¼ 20%; Y ¼ 26:43  13:92X ðR2 ¼ 0:9953Þ (9)

a ¼ 40%; Y ¼ 26:22  13:79X ðR2 ¼ 0:9952Þ (10)

a ¼ 60%; Y ¼ 25:90  13:68X ðR2 ¼ 0:9983Þ (11)

Fig. 5. Conversion degree (a) and DTG (b) curves of sample at different heating
rates during the first dehydration stage [g-Al(OH)3 to g-AlOOH]. Fig. 7. Diagrammatic sketch of definition for peak shape factor [22].
 B. Zhu et al. / Ceramics International 36 (2010) 2493–2498 2497

Table 2 4. Conclusions
Peak shape factor S and reaction order n of the first stage of g-Al(OH)3
dehydration.
The dehydration of g-Al(OH)3 is divided into two stages: (1)
Heating rate (8C min1) S n g-Al(OH)3 dehydrates partial crystalline water and transforms
5 0.460 0.85 into g-AlOOH, and (2) g-AlOOH dehydrates remaining water
10 0.552 0.94 and transforms into amorphous Al2O3 at higher temperatures.
15 0.627 1.00 Furthermore, the starting dehydration temperature of g-
20 0.688 1.04
Al(OH)3 shifts to higher temperatures as the heating rate of
the system increases.
Kinetic parameters, such as: activation energy, the pre-
a ¼ 80%; Y ¼ 24:60  13:17X ðR2 ¼ 0:9776Þ (12) exponential factor, and the reaction order of the two
dehydration reactions of g-Al(OH)3 were calculated using
where Y is the ln(  DTG) value, X is the 1000/T value, and a is the Kissinger equation, DSC and DTG, respectively. Values
the conversion degree. The slopes of the fitted lines at different produced by DSC and DTG were very similar.
conversion degree are similar to each other, and the activation
energy E of the first dehydration stage is calculated as Acknowledgement
113.40 kJ mol1 based on the average value of the 4 different
slopes. Financial support by National Natural Science Foundation
Obviously, the E value calculated from the DTG curves is of China (No. 50774057) is gratefully acknowledged.
very close to the one resulting from DSC curves
(108.50 kJ mol1), which means the two methods can be used References
to calculate the kinetics of g-Al(OH)3 dehydration.
[1] M.H. Zhang, Y.Q. Zhang, Research of crystalline phase transformation of
alumina by X-ray diffraction at high temperature, Pet. Process. Petrochem.
3.2.3. Reaction order of the first stage 10 (1979) 56–75.
In order to determine the reaction order, n, of the kinetic [2] H.S. Santos, T.W. Campos, P.S. Santos, P.K. Kiyohara, Thermal phase
sequences in gibbsite/kaolinite clay: electron microscopy studies, Ceram.
equation, Kissinger introduced the conception of peak shape Int. 31 (2005) 1077–1084.
factor, S, and defined it as the ratio of a/b shown in Fig. 7 [22]. [3] B.A. Scott, W.H. Horsman, Structural change during the production of
Details about the values a and b were defined as follows: corundum by calculation of gibbsite and their influence on fabrication,
For the triangle MNL shown in Fig. 7, the sides ML and NL Trans. Brit. Ceram. Soc. 69 (1970) 37–43.
are tangent lines to a DSC or DTA peak at the right and left [4] R. Salomão, V.C. Pandolfelli, The role of hydraulic binders on magnesia
containing refractory castables: Calcium aluminate cement and hydratable
inflection points (E and F), respectively. The bottom side MN is alumina, Ceram. Int. 35 (2009) 3117–3124.
a random straight line paralleling to the abscissa, which is [5] F.A. Cardoso, M.D.M. Innocentini, M.F.S. Miranda, F.A.O. Valenzuela,
divided into 2 line segments MH and NH by the altitude line V.C. Pandolfelli, Drying behavior of hydratable alumina-bonded refrac-
segment LH. The length of MH is a and the length NH is b. tory castables, J. Eur. Ceram. Soc. 24 (2004) 797–802.
According to the theoretical derivation of Kissinger, the [6] I.R. Oliveira, V.C. Pandolfelli, Castable matrix, additives and their role on
hydraulic binder hydration, Ceram. Int. 35 (2009) 1453–1460.
quantitative relationship between the reaction order (n) and the [7] I. Levin, L.A. Bendersky, D.G. Brandon, M. Rühle, Cubic to monoclinic
shape factor (S) is described as follows [22,28]: phase transformations in alumina, Acta Mater. 45 (1997) 3659–3669.
[8] Z.X. Gao, Z.Y. He, X.P. Zheng, Q.H. Fu, X.B. Shi, Phase transformation
n ¼ 1:26 S0:5 (13) and morphology of Bayer-gibbsite during heating, J. Chin. Ceram. Soc. 36
(2008) 117–123.
[9] H.S. Yen, J.L. Chang, P.C. Yu, Relationships between DTA and DIL
The peak shape factor, S, and the corresponding n, values of characteristics of nano-sized alumina powders during u- to a-phase
the first dehydration stage of g-Al(OH)3 at different heating rate transformation, J. Cryst. Growth 246 (2002) 90–98.
were calculated according to Fig. 7 and Eq. (13), respectively. [10] S.D. Vaidya, N.V. Thakkar, Study of phase transformations during hydra-
The calculated values for S and n are listed in Table 2. All the tion of rho alumina by combined loss on ignition and X-ray diffraction
technique, J. Phys. Chem. Solids 62 (2001) 977–986.
reaction order values of the samples at different heating rates [11] G.Q. Yang, D.S. Yan, The kinetics of phase transformation of alumina, J.
are close to 1 except at 5 8C min1. The average of the 4 Chin. Ceram. Soc. 5 (1966) 1–11.
reaction order values is 0.96. [12] H.P. Wang, B.G. Xu, P. Smith, M. Davies, L. Desilva, C. Vingate, Kinetic
modelling of gibbsite dehydration/amorphization in the temperature range
823–923 K, J. Phys. Chem. Solids 67 (2006) 2567–2582.
3.2.4. Determination of kinetic parameters of the second [13] J.P. Elder, Reconciliation of Arrhenius and iso-conversional analysis
stage by DSC and DTG curves kinetics parameters of non-isothermal data, Thermochim. Acta 272
The peak temperature, Tp, of the second endothermic peak (1996) 41–48.
of g-Al(OH)3 at heating rates of 5, 10, 15 and 20 8C min1, is [14] X.D. Li, B.Q. Zhu, Kinetics study of mullite formation from aluminum
505, 520, 531 and 535 8C, respectively. Using DSC and DTG sulfate and silica in molten sodium sulfate, J. Chin. Ceram. Soc. 36 (2008)
498–502.
data, kinetic parameters of the second dehydration stage of [15] N.S. Saxena, Phase transformation kinetics and related thermodynamic
g-Al(OH)3 were also determined using Eqs. (7), (8) and (13) and optical properties in chalcogenide glasses, J. Non-Cryst. Solids
as 217.24 kJ mol1 for E, 8.30  1013 for A, and 1.06 for n. 345–346 (2004) 161–168.
2498 B. Zhu et al. / Ceramics International 36 (2010) 2493–2498

[16] S. Kurajica, A. Bezjak, E. Tkalčec, Resolution of overlapping peaks and [22] R.Z. Hu, Q.Z. Shi, Thermal Analysis Kinetics, Science Press, Beijing,
the determination of kinetic parameters for the crystallization of multi- PRC, 2001, pp. 19–22, 111.
component system from DTA or DSC curves. I. Non-isothermal kinetics, [23] A.A. Duswalt, The practice of obtaining kinetic data by differential
Thermochim. Acta 288 (1996) 123–135. scanning calorimetry, Thermochim. Acta 8 (1974) 55–68.
[17] J. Perić, R. Krstulović, M. Vućak, Investigation of dehydroxylation of [24] S.M. Taylor, P.J. Fryer, A numerical study of the use of the Kissinger
gibbsite into boehmite by DSC analysis, J. Therm. Anal. 46 (1996) analysis of DSC thermograms to obtain reaction kinetic parameters,
1339–1347. Thermochim. Acta 209 (1992) 111–125.
[18] K.J.D. Mackenzie, J. Temuujin, J. Okada, Thermal decomposition of [25] I.W. Donald, Crystallization kinetics of a lithium zinc silicate glass studied
mechanically activated gibbsite, Thermochim. Acta 327 (1999) 103–108. by DTA and DSC, J. Non-Cryst. Solids 345–346 (2004) 120–126.
[19] J.M.R. Mercury, P. Pena, A.H. Aza, On the decomposition of synthetic [26] H.E. Kissinger, Variation of peak temperature with heating rate in
gibbsite studied by neutron thermodiffractometry, J. Am. Ceram. Soc. 89 differential thermal analysis, J. Res. Natl. Bur. Stand. 57 (1956) 217–221.
(2006) 3728–3733. [27] Y.F. Lee, D. Dollimore, The identification of the reaction mechanism in
[20] J. Yang, R. Miranda, C. Roy, Using the DTG curve fitting method to rising temperature kinetic studies based on the shape of the DTG curve,
determine the apparent kinetic parameters of thermal decomposition of Thermochim. Acta 323 (1998) 75–81.
polymers, Polym. Degrad. Stab. 73 (2001) 455–461. [28] N.S. Fatemi, R. Whitehead, D. Price, D. Dollimore, Determination of
[21] A.V. dos Santos, J.R. Matos, Dehydration studies of rare earth p-tolue- activation energy value from the maximum rate of reaction points
nesulfonate hydrates by TG/DTG and DSC, J. Alloys Compd. 344 obtained from non-isothermal experiments, Thermochim. Acta 78
(2002) 195–198. (1984) 437–440.