Dehydration and dehydroxylation of alumina gels prepared from tri-sec-butoxyaluminium modified with short chain aliphatic acids

Article abstract of DOI:10.1016/S0040-6031(00)00572-4

The dehydration and dehydroxylation of alumina gels prepared from the hydrolysis of tri-sec-butoxyaluminium (ASB) at both 25 and 90°C, unmodified and modified with short chain aliphatic acids has been studied using a combination of differential thermal analysis (DTA) and thermogravimetry. The alumina phase in the gel was found to be dependent upon the temperature of hydrolysis and the modifying acid. The gels formed at 25°C from unmodified ASB contained alumina as gibbsite and at 90°C from the unmodified ASB a mixture of gibbsite and boehmite, whereas those gels formed from ASB at 25°C and modified with short chain aliphatic acids were amorphous. In DTA of the modified gels, the exotherms from the combustion of the aliphatic acid groups masked the endotherms of dehydration and dehydroxylation of the gels. Neither DTA nor thermogravimetric analysis (TGA) could determine the dehydroxylation temperature range. The application of differential thermogravimetric analysis (DTGA) enabled these temperatures to be obtained. DTGA of the gels formed from the acid modified ASB showed complex 'endograms'. Infrared spectroscopy was used to determine the molecular structures of the acetic acid modified gels. (C) 2000 Elsevier Science B.V.

Full text of DOI:10.1016/S0040-6031(00)00572-4

Thermochimica Acta 362 (2000) 37±48
Dehydration and dehydroxylation of alumina gels prepared from
tri-sec-butoxyaluminium modi®ed with short chain aliphatic acids
*
J.L. Szetu, R.L. Frost , J.T. Kloprogge, S.C. Russell, W. Martens
Centre for Instrumental and Developmental Chemistry, Queensland University of Technology, 2 George Street, Brisbane,
GPO Box 2434, Qld 4001, Australia
Received 20 December 1999; received in revised form 24 May 2000; accepted 28 May 2000
Abstract
The dehydration and dehydroxylation of alumina gels prepared from the hydrolysis of tri-sec-butoxyaluminium (ASB) at
both 25 and 908C, unmodi®ed and modi®ed with short chain aliphatic acids has been studied using a combination of
differential thermal analysis (DTA) and thermogravimetry. The alumina phase in the gel was found to be dependent upon the
temperature of hydrolysis and the modifying acid. The gels formed at 258C from unmodi®ed ASB contained alumina as
gibbsite and at 908C from the unmodi®ed ASB a mixture of gibbsite and boehmite, whereas those gels formed from ASB at
2
58C and modi®ed with short chain aliphatic acids were amorphous. In DTA of the modi®ed gels, the exotherms from the
combustion of the aliphatic acid groups masked the endotherms of dehydration and dehydroxylation of the gels. Neither DTA
nor thermogravimetric analysis (TGA) could determine the dehydroxylation temperature range. The application of differential
thermogravimetric analysis (DTGA) enabled these temperatures to be obtained. DTGA of the gels formed from the acid
modi®ed ASB showed complex `endograms'. Infrared spectroscopy was used to determine the molecular structures of the
acetic acid modi®ed gels. # 2000 Elsevier Science B.V. All rights reserved.
Keywords: Dehydration; Dehydroxylation; Differential thermal analysis; Differential thermogravimetric analysis; Alumina gel; Gibbsite;
Infrared emission spectroscopy; Thermogravimetric analysis; Tri-sec-butoxyaluminium
1
. Introduction
from the hydrolysates depends on the conditions of the
hydrolysis and the aging of the hydrolysates [4,5].
Aging simply means the changes in phase as either the
hydrolysate or the gel is left to stand in the presence of
water. Such aging can effect the rates of peptisation of
the alumina sols. Aging of the oxo-hydroxides as gels
causes structural changes in the sequence amorphous-
pseudoboehmite±bayerite±gibbsite [1,2].
Chemical modi®ers are used in controlling or
retarding the rate of hydrolysis and condensation so
that the desired material, a gel, colloidal solution or
precipitate is formed [6,7]. The methodology is essen-
tial in the process of obtaining a metal sol that is
suitable to be used in for example coating applica-
Alumina gels are readily prepared by the hydrolysis
of aluminium alkoxides [1,2]. Several phases of alu-
minium hydroxides exist according to the formula
Al(OH) ; gibbsite, nordstrandite and bayerite [3]. If
the formula is AlO(OH), then the phases are boehmite
or diaspore [3]. Pseudoboehmite is a hydrated form of
boehmite with slightly larger d-spacing. The chemical
and structural make-up of the oxo-hydroxides formed
3
*
Corresponding author. Tel.: ‡61-7-3864-2407;
fax: ‡61-7-3864-1804.
E-mail address: r.frost@qut.edu.au (R.L. Frost).
0040-6031/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 0 4 0 - 6 0 3 1 ( 0 0 ) 0 0 5 7 2 - 4

3
8
J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
tions. Acids, chlorides, alcohols, bases, and chelating
agents are all substances that effectively produce a less
reactive precursor [8]. These ligands undergo either
one of the two reactions with the metal. Firstly,
nucleophilic substitution may occur where the chemi-
cal modi®er replaces an original ligand from the
metal, a reaction that usually occurs if the precursor
is a coordinately saturated metal [6,7],
chain aliphatic acids. The formation of such gels is
important in aluminium production where in the Bayer
process the effect of organics on alumina sols can be
important. In particular, the presence of organics
effects the solubility of the alumina in the caustic
solutions. One means of studying the effects of acids
on alumina sols is to synthesise the alumina sols in the
presence of aliphatic acids. This may be achieved in
two ways: (a) by peptising the unmodi®ed ASB with
the short chain acid or (b) by modifying the ASB with
the acid and then peptising with the same or another
acid. In order to understand the chemistry of these
sols, we have converted these sols to gels to determine
the structure, and in this paper, we report the thermal
analysis of alumina gels formed from the hydrolysis of
ASB modi®ed with short chain aliphatic acids.
XOH ‡ AlꢀOR† ! AlꢀOR†3� xꢀOX† ‡ xROH
3
x
The second reaction is nucleophilic addition, which
occurs only for unsaturated coordinated metal [6±10],
XOH ‡ AlꢀOR† ! AlꢀOR† ꢀXOH†
n� 3
3
3
The alumina sols and gels are prepared from the
hydrolysis of the modi®ed precursor with water. The
tri-sec-butoxyaluminium (ASB) reacts with the modi-
fying aliphatic acid according to the reaction:
2
. Experimental techniques
RCOOH‡AlꢀOR† ! AlꢀOOCR† ꢀOX† ‡ xROH
3
3� x
x
2
.1. Synthesis of alumina sols and gels
Sanchez et al. [11] indicated that the aluminium
sec-butoxide reaction with acetic acid is an exother-
mic reaction. A transparent alumina gel is obtained in
a faster time than in the similar reaction with a
titanium alkoxide. The acetic acid:alkoxide ratio of
Aluminium oxo-hydroxide hydrolysates were pre-
pared by the reaction of tri-sec-butoxyaluminium in
both cold (258C) and hot (908C) water. The water±
alkoxide ratio was kept at 100:1. The alkoxides were
introduced into the excess water with vigorous stir-
ring. Alkoxides were modi®ed with short chain ali-
phatic acids by the slow addition of the dry acid to the
alkoxide using an ice bath. The modi®ed alkoxide was
then hydrolysed in water to produce the hydrolysate.
The resulting precipitates were washed several times
with water to remove organic reaction products,
namely alcohol, then diluted in water to give slurries
with concentration of aluminium of ꢁ0.6 M. Hydro-
lysate slurries were agitated by the use of a magnetic
stirrer at room temperatures to prepare the sol. Nitric
acid dispersing agent was added in a ratio of 0.1 mol of
1:1 takes a few minutes to gel. This effect is due to the
increase in condensation rate. Comparing Ti(OR)4
(
Zˆ4, Nˆ6) to Al(OR) (Zˆ3, Nˆ8), the greater
4
difference in N and Z in¯uences the condensation
rate.
Several methods are available for the study of the
structure of alumina gels. One method is to use X-ray
diffraction. However, if the gels contain amorphous or
semi-amorphous phases where short range order pre-
vails, then XRD is not applicable. The molecular
spectroscopic techniques of infrared and Raman are
suitable for the study of alumina gels [12,13]. Solid
state NMR has also proved useful [12±16]. 27Al SS/
MAS NMR has proved useful in determining the
peptisation characteristics of alumina hydrolysates.
27Al NMR has proven useful for the determination
of the coordination state of the aluminium [12,13].
NMR is also useful in characterising amorphous
phases. The technique of thermal analysis has also
been used for a long time to study alumina phases
HNO per mole of Al. Samples of the sols produced
3
were collected from the peptisation vessel and allowed
to gel by solvent evaporation and by using a rotary
evaporator at 258C.
2.2. Thermal analysis
Differential thermal and thermogravimetric ana-
lyses of the crushed gels were obtained simulta-
neously using a Setaram DTA/TGA instrument,
[17,18]. In this work, we have prepared alumina gels
from tri-sec-butoxyaluminium modi®ed with short

J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
39
operating with a heating rate of 1.08C per minute from
ambient temperatures to 6508C.
quality spectra come from very thin samples of the
alumina gel on the platinum plate.
The IES technique depends on the difference in
temperature between the sample and the detector for
its success. Thus, the greater the difference between
the sample temperature and the detector, the better is
the quality of the spectrum in terms of signal to noise.
Hence, spectra at the lower temperatures always con-
tain more noise. This means that the useful tempera-
ture range for the study of dehydroxylation is from
1008C and above. The dif®culty of lack of signal for
the low frequency region is not as pronounced. The
OH stretching region is noisier because it is in the
high-energy region of the spectrum. As the tempera-
ture goes up, more energy is available at higher wave
numbers (Planck's law), therefore at any temperature
(up to about 9008C), there is more energy at lower
wave number, which gives a higher signal/noise ratio.
It is for this reason that the signal gets better as the
temperature is increased. Emittance values vary from
0 to 1 with a scale equivalent to an absorption spec-
trum. The data were linearised with respect to con-
centration where required, by transforming to units of
� log10[1� emittance(n)]. Spectral manipulation such
as baseline adjustment, smoothing and normalisation
was performed using the Spectracalc software pack-
age (Galactic Industries, Salem, NH, USA).
2
.3. Infrared emission spectroscopy
FT-IR emission spectroscopy was carried out on a
Digilab FTS-60A spectrometer, which was modi®ed
by replacing the IR source with an emission cell. A
description of the cell and principles of the emission
experiment have been published elsewhere [19±21].
Approximately 0.2 mg of size fractioned alumina gel
(
6
<0.5 mm sample size) was spread as a thin layer on a
mm diameter platinum surface and held in an inert
atmosphere within a nitrogen purged cell with an
additional ¯ow of argon over the sample during heat-
ing. In the normal course of events, three sets of
spectra are obtained: ®rstly the black body radiation
over the temperature range selected at the various
temperatures, secondly the platinum plate radiation
is obtained at the same temperatures and thirdly the
spectra from the platinum plate covered with the
alumina gel sample. Normally only one set of black
body and platinum radiation spectra is required.
The emittance spectrum at a particular temperature
was calculated by subtraction of the single beam
spectrum of the platinum backplate from that of the
platinum‡sample, and the result ratioed to the single
beam spectrum of an approximate blackbody (gra-
phite). This spectral manipulation is carried out after
all the spectral data has been collected. The spectra of
a series of alumina gels may be obtained and in the
spectral manipulation, the same sets of black body and
platinum hot plate radiation are used for all the gel
spectra. The emission spectra were collected at inter-
vals of 258C over the range 100±6008C. The time
between scans (while the temperature was raised to the
next hold point) was approximately 100 s. Scans were
monitored until thermal equilibrium was obtained;
this usually required approximately 3±5 min. The
spectra were acquired by coaddition of 128 scans
for temperatures 100±3008C (approximate scanning
time 140 s) and 64 scans for temperatures 300±6008C
3. Results and discussion
3.1. Differential thermal analysis (DTA)
Fig. 1 shows the DTA of the alumina gels and the
results of the analyses are reported in Table 1. The
patterns are complex for all except the alumina gel
formed at room temperature from the unmodi®ed
ASB. When the ASB is modi®ed with short chain
aliphatic acids, the presence of the acetate together
with some residual alcohol generated during the
hydrolysis results in the observation of some large
exothermic peaks. The exotherms increase in size as
the length of the alkyl chain increases. These peaks
swamp the dehydroxylation of the alumina phases,
thus preventing any assessment of which alumina
phase is produced.
(
approximate scanning time 45 s), with a nominal
�
1
resolution of ꢁ4 cm . Good quality spectra can be
obtained provided the sample thickness is not too
large. If too large a sample is used then the spectra
become dif®cult to interpret because of the presence of
combination and overtone bands. In fact, the best
The alumina gel formed at 258C shows endotherms
at 85 and 2888C with a weak endotherm at 1688C. The

4
0
J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
Fig. 1. DTA of alumina gels formed from tri-sec-butoxyaluminium: (a) unmodi®ed at 258C; (b) modi®ed with 1 mol of acetic acid and
hydrolysed at 908C; (c) modi®ed with 1 mol of acetic acid and hydrolysed at 258C; (d) modi®ed with 1 mol of propanoic acid and hydrolysed
at 258C; (e) modi®ed with 1 mol of butanoic acid and hydrolysed at 258C.
Table 1
Endotherms and exotherms of the synthesised alumina gels
Sample
Endotherm 1 T
8C), area (%)
Endotherm 2 T
(8C), area (%)
Endotherm 3 T
(8C), area (%)
Endotherm 4 T
(8C), area (%)
Endotherm 5 T
(8C), area (%)
(
Alumina gel (258C) unmodified
Alumina gel (908C) unmodified
85, 15.0
62, 14.7
168, 13.5
251, 23.0
244, 71.5
277, 43.4
404, 15.4
538, 3.0
Sample
Dehydration
Dehydroxylation
Compustion
Endotherm 1 T
Endotherm 2 T
(8C), area (%)
Endotherm 3 T
(8C), area (%)
Exotherm 1 T
(8C), area (%)
Exotherm 2 T
(8C), area (%)
(
8C), area (%)
Alumina gel (258C) modified
with 1 mol of acetic acid
78, 45
71, 31
81, 80
75, 35
160, 40
110, 43
178, 20
147, 38
224, 15
346, 11
394, 84
356, 25
370, 95
382, 89
424, 16
391, 75
467, 5
Alumina gel (908C) modified
with 1 mol of acetic acid
250/285/350, 38
Alumina gel (258C) modified
with 1 mol of propanoic acid
Alumina gel (258C) modified
with 1 mol of butanoic acid
238, 30

J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
41
areas of these endotherms are 15.0, 13.5 and 71.5%,
respectively. These values simply represent the
amount of thermal energy required to remove the
water in the gel. Most of the energy is used to remove
the water from the gel. The ®rst two endotherms are
attributed to water in the alumina gel and the third
to the dehydroxylation of the alumina phase.
Synthetic gibbsite shows endotherms at 261 and
2848C. Synthetic boehmite shows two dehydroxyla-
tion endotherms at 486 and 5008C [22,23]. Minor
endotherms are also observed at 223 and 4208C for
synthetic gibbsite [22]. It is concluded that the
endotherm at 2888C in the gel formed at 258C is
due to the dehydroxylation of gibbsite. This structure
was con®rmed by X-ray diffraction. The gel formed at
908C shows endotherms at 62, 251, 277, 404 and
Fig. 2. (a) Component peak analysis of the endotherms of the alumina gel formed from tri-sec-butoxyaluminium modi®ed with 1 mol of acetic
acid and hydrolysed at 908C, (b) Component peak analysis of the endotherms of the alumina gel formed from tri-sec-butoxyaluminium
modi®ed with 1 mol of acetic acid and hydrolysed at 258C, (c) Component peak analysis of the endotherms of the alumina gel formed from tri-
sec-butoxyaluminium modi®ed with 1 mol of propanoic acid and hydrolysed at 258C, (d) Component peak analysis of the endotherms of the
alumina gel formed from tri-sec-butoxyaluminium modi®ed with 1 mol of butanoic acid and hydrolysed at 258C.

4
2
J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
Fig. 2. (Continued ).
5
3
388C, which have areas of 14.7, 23.0, 43.4, 15.4 and
%. X-ray diffraction shows that this gel contains a
energy for the removal of the second water. A set of
three endotherms is observed at 250, 285 and 3508C.
The ®rst two endotherms are attributed to the dehy-
droxylation of gibbsite and the second to the dehy-
droxylation of boehmite. The boehmite is only a minor
component in this alumina gel. Normally the hydro-
lysis of ASB at 908C results in the formation of
boehmite. It is apparent that the presence of the acetate
suppresses the formation of boehmite. At 908C, the
sec-butanol formed upon hydrolysis is lost as it simply
boils off. Therefore, the exotherm at 3948C must be
due to the combustion of the acetate groups.
mixture of gibbsite and boehmite.
When the ASB is modi®ed with 1 mol of acetic acid
per mole of ASB, the DTA pattern becomes more
complex. Fig. 2a shows the component analyses of the
DTA pattern of an alumina gel formed from ASB
modi®ed with 1 mol of acetic acid per mole of ASB
and hydrolysed at 908C. Two endotherms are observed
at 78 and 1108C with relative areas of 15.5 and 21.5%.
The ®rst endotherm is associated with water in the gel
and the second which required almost 50% of the total

J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
43
When the ASB is modi®ed with 1 mol of acetic acid
per mole of ASB and hydrolysed at room temperature,
the DTA pattern shown in Fig. 1c results. The curve
signi®cantly larger. The thermal energy required to
remove the water of hydration for the unmodi®ed gels
is ꢁ15% of the total thermal energy. The values for the
acid modi®ed gels vary from 31 to 80%. The second
dehydration endotherm is attributed to water hydrogen
bonded to the acid groups adsorbed on the surface of
the alumina gels.
®
tting is shown in Fig. 2b. Three endotherms are
observed at 78, 160 and 2248C with relative areas
of 22.5, 20.0 and 7.5%. The ®rst two endotherms are
attributed to the loss of water in the gel and the third to
the dehydroxylation of gibbsite, which is partially
masked by the large exotherm at 3948C. This large
exotherm is attributed to the combustion of not only
the acetate groups but also some sec-butanol which
may be retained in the gel. The DTA pattern for this
gel (Fig. 2b) is very different from the pattern for the
gel shown in Fig. 1b. It is proposed that the gel formed
contains aluminium oxyacetate and not gibbsite. X-
ray diffraction shows that the gel contains an amor-
phous phase.
3.2. Thermogravimetric analysis (TGA)
The thermogravimetric patterns of the alumina gels
are shown in Fig. 3. Theoretically, gibbsite and boeh-
mite should have weight losses of 34.6 and 15%. The
experimental weight losses are often less than these
values, and for synthesised gibbsite and boehmite,
weight losses of 26.4 and 12% were obtained. The
weight losses and the temperature range are reported
Fig. 1d and e shows the DTA patterns from the ASB
modi®ed with propanoic and butanoic acids and
clearly no dehydroxylation endotherms are observed.
Fig. 2c and d illustrates the component analyses of the
DTA patterns. Clearly the exotherms from the com-
bustion of retained alcohol and the aliphatic acid
groups dominate these ®ts and in the butanoic case
prevents the observation of the expected endotherms.
DTA patterns showing the dehydroxylation of the
alumina gels in the 250±3508C temperature range
would be expected. As for the gel formed from the
acetic acid modi®ed ASB, the gels are amorphous.
The dehydration of the alumina gels formed with
and without modi®cation with short chain aliphatic
acids is different. The alumina gels formed from the
hydrolysis and subsequent peptisation with nitric acid
show a dehydration temperature of 858C, whereas the
alumina gel hydrolysed at 908C dehydrates at 628C.
This difference in temperature relates to the bonding
mechanism of the water in the gel. It is proposed that
the water in the unmodi®ed 258C gel is hydrogen
bonded to the alumina particles, whereas in the unmo-
di®ed 908C gel, the water is simply surrounding/
hydrating the alumina particles. For the 258C acid
modi®ed gels, two dehydration endotherms are
observed; ®rstly in the 70±808C temperature range
and secondly in the 150±1708C temperature range. For
the 908C acetic acid modi®ed alumina gel, the second
dehydration temperature is 1108C. The ®rst dehydra-
tion temperature range is similar to that for the unmo-
di®ed gels. However, the area of the endotherm is
Fig. 3. TGA of alumina gels formed from tri-sec-butoxyalumi-
nium: (a) unmodi®ed at 258C; (b) unmodi®ed at 908C; (c) modi®ed
with 1 mol of acetic acid and hydrolysed at 908C; (d) modi®ed with
1
mol of acetic acid and hydrolysed at 258C; (e) modi®ed
with 1 mol of propanoic acid and hydrolysed at 258C; (f) modi®ed
with 1 mol of butanoic acid and hydrolysed at 258C.

4
4
J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
Table 2
Percentage weight losses during dehydration and dehydroxylation of alumina gels
Alumina gel
Dehydration
(wt.% loss)
Dehydration
2 (wt.% loss)
Dehydration
3 (wt.% loss)
Dehydroxylation
1 (wt.% loss)
Dehydroxylation
2 (wt.% loss)
1
Temperature range
25±908C
90±1108C
110±2008C
200±3008C
26.4 (34.6)
300±>4008C
Gibbsite (syn)
Boehmite (syn)
12 (15)
6.7
4.3
Alumina gel (258C) unmodified
Alumina gel (908C) unmodified
Alumina gel (258C) modified with
10.3
5.7
23.0
21.3
5.0
23.0
14.5
14.5
14.5
1
mol of acetic acid
Alumina gel (908C) modified with
mol of acetic acid
Alumina gel (258C) modified with
mol of propanoic acid
Alumina gel (258C) modified with
mol of butanoic acid
6.0
11.5
10.8
ꢁ11
27.8
22.3
ꢁ12
18
1
19.0
ꢁ9
18.7
ꢁ28
1
ꢁ5
1
in Table 2. The thermal analysis patterns for the
alumina gels prepared from ASB and not modi®ed
show the simplest weight loss with three steps. The
unmodi®ed gel formed at 908C contains signi®cantly
less adsorbed water. The TGA curves for the alumina
gels formed from ASB modi®ed with short chain
aliphatic acids show almost a continuous weight loss
and the in¯ections in the TGA curves are dif®cult to
de®ne. The weight loss for both alumina gels formed
from the 908C hydrolysate and without and with 1 mol
of acetic acid is ꢁ6 wt.%. The 258C acetic acid gels
show a weight loss of ꢁ23%. The 258C propanoic acid
gel shows a weight loss of 19.0% at 908C and a further
weight loss of 10.8% over the 110±1908C temperature
range. The weight losses for the alumina gel formed
from ASB modi®ed with butanoic acid are dif®cult to
determine with precision.
turn reacts with the acetic acid to produce the ester and
liberate water. This reaction is highly exothermic. This
series of reactions gives rise to the formation of
Al(OH)(CH COO) . The TGA curves shows dehy-
3
2
droxylation at a higher temperature than the other gels
in the range 250±2808C. It is proposed that this
dehydroxylation is that of the hydroxyls from the
Al(OH)(CH COO) . A fourth step is observed for this
3
2
gel in the 325±4008C range and this is attributed to the
loss of the acetates. XRD shows that the alumina in
this gel is amorphous.
The weight loss for the alumina gel formed from the
propanoic acid appears in three steps over the 25±
1108C, 110±3008C and 350±4008C range. These steps
cannot be observed in the DTA patterns, because
of the exothermic peaks. It is likely that the
Al(OH)(CH CH COO) is formed in a similar manner
3
2
2
The TGA curves show a weight loss on steps even
though the steps are dif®cult to determine. The second
weight loss is in the 90±1108C range and the average
weight loss is 12.5%. For the alumina gel formed with
the modi®cation of ASB with propanoic acid, it is
dif®cult to determine the breaks in the TGA curve.
Consequently, the weight loss of 29.8% is for the ®rst
two dehydration steps. For the alumina gel formed
from ASB modi®ed with 1 mol of acetic acid and
hydrolysed at 258C, a third step is observed with a
weight loss of 14.5%. Thus, it would appear that the
dehydroxylation of this gel is different. Acetic acid
reacts with the ASB to produce sec-butanol, which in
to the Al(OH)(CH COO) . The weight loss in the
3 2
dehydration steps is 29.8% and the weight loss in
the two dehydroxylation steps is 22.3 and 18.7 wt.%.
XRD shows the alumina gel to be composed of
amorphous alumina. For the alumina gels prepared
from the modi®cation of ASB with butanoic acid, the
TGA curve shows almost a continuous weight loss
over the entire range from 25 to 4258C. Table 2 shows
estimates of the weight loss over the appropriate
temperature ranges.
One method of studying the dehydroxylation steps
in gels, which show large exotherms in the DTA
patterns, is to study the differential thermogravimetric

J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
45
spond to the DTA endotherms in Fig. 1a. For the
alumina gel formed from the modi®cation of ASB
with 1 mol of acetic acid and hydrolysed at 908C, two
endograms are observed at 90 and 2258C with a very
broad continuum over the 300±4508C temperature
range. The endotherms in the DTA patterns were
observed at 110 and 2908C. For the alumina gels
formed from the modi®cation of ASB with 1 mol of
acetic acid and hydrolysed at 258C, four weight
changes are observed at 90, 175, 280 and 3908C.
The ®rst weight loss corresponds well with the
endotherm in Fig. 1c. However, no other endotherms
could be observed in the DTA pattern. This example
clearly demonstrates the usefulness of the technique
of differential thermogravimetry. Neither DTA nor
TGA could identify the dehydroxylation tempera-
tures. For this particular gel, four weight losses were
observed.
The DTGA pattern for the alumina gel formed from
the modi®cation of ASB with 1 mol of propanoic acid
shows an intense endogram at 958C with two small
peaks at 165 and 2058C and a broad endogram at
3
908C. The pattern resembles that of the acetic acid
produced gel. X-ray diffraction shows that this gel
contains amorphous alumina that on aging transforms
to pseudoboehmite. The DTGA pattern for the alu-
mina gel formed from the modi®cation of the ASB
with butanoic acid is complex with endotherms
observed at 90, 170, 220, 275 and 3908C. The ®rst
endogram corresponds to the loss of water from the
gel. This gel must contain several chemical species
that dehydroxylate or debutanate at these endograms.
Nevertheless the DTGA pattern shows weight losses
occurring over speci®c temperature ranges which
could not be observed with either DTA or TGA.
Fig. 4. DTGA of alumina gels formed from tri-sec-butoxyalumi-
nium: (a) unmodi®ed at 258C; (b) unmodi®ed at 908C; (c) modi®ed
with 1 mol of acetic acid and hydrolysed at 908C; (d) modi®ed with
1
1
1
mol of acetic acid and hydrolysed at 258C; (e) modi®ed with
mol of propanoic acid and hydrolysed at 258C; (f) modi®ed with
mol of butanoic acid and hydrolysed at 258C.
patterns. Fig. 4 shows the differential thermogravi-
metric patterns of the alumina gels. It should be
remembered that the differential of the TGA curves
determines the points of in¯ections in these curves.
Thus, the peaks in the differential thermogravimetric
analysis (DTGA) patterns represent the points of
maximum weight loss and the temperature range over
which this weight loss is occurring. These negative
peaks do not necessarily correspond with the
endotherms in the DTA curves. However, the DTGA
curves eliminate the problem of the large exotherms in
the DTA patterns masking the perhaps smaller
endotherms.
3.3. Infrared spectroscopy
Acetic acid in butyl ether shows two peaks at 1720
�
1
and 1761 cm which are attributed to the symmetric
and antisymmetric C=O stretching of the undisso-
ciated acetic acid (Fig. 5) [24]. Upon reacting the
dry acetic acid with ASB in the ratio of 1:1 in butyl
ether at a concentration of 0.06 M, these bands shift to
�
1
1603 and 1645 cm . These bands are attributed to the
monodentate bonding of the acetate group to the
aluminium in ASB [24]. In the alumina gels formed
from the modi®cation of ASB with 1 mol of acetic
The TGA for the alumina gel formed without
modi®cation of the ASB and hydrolysed at 908C
are observed at 60 and 2758C. These values corre-

4
6
J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
Fig. 5. Infrared spectra of: (a) butyl ether; (b) 0.6 M acetic acid in butyl ether; (c) ASB/acetic acid 1:1 as 0.6 M in butyl ether; (d) acetic acid
modi®ed alumina gel at 2508C; (e) 3008C; (f) 3508C (M represents the monodentate C=O antisymmetric stretch, A the free acetate, S the
monodentate CˆO symmetric stretch, AS the antisymmetric stretch, BE is butyl ether and CH the CH stretching vibrations of acetic acid).
Fig. 6. Infrared emission spectra of alumina gel formed from the 258C hydrolysis of ASB with acetic acid over the temperature range 200±
7
508C.

J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
47
acid the antisymmetric stretching band is observed at
�
4. Conclusion
1
587 cm . The symmetric stretching mode is
1
observed at 1464 cm . However, this is the overlap
�
1
The alumina gel formed from the nitric acid pepti-
sation of the hydrolysate formed from the reaction of
tri-sec-butoxyaluminium with water is shown to be
gibbsite. Peptisation at 908C resulted additionally in
the formation of some pseudo-boehmite as determined
by X-ray diffraction and infrared spectroscopy.
When the alumina gels are formed from the modi®ca-
tion of the ASB with acetic acid and peptised at 258C,
complex gels involving Al(OH)(CH COO) were pro-
duced. The calcination of this gel to 6008C produced
amorphous alumina. When the dehydroxylation
endotherms are masked by the exotherms due to the
combustion of the aliphatic acid and the alcohol in the
DTA patterns, the differential thermogravimetric
curves were used to determine these phase changes.
Neither DTA nor TGA could be used to determine the
complex patterns.
�
1
of this band with the 1430 cm HCH bending mode.
What the spectra clearly show is that the alumina gels
contain acetate groups bonded to the aluminium in a
monodentate coordination.
Spectra d to f also show the loss of hydroxyls from
the gel. In the 2508C spectrum two bands are observed
�
1
at 3642 and 3475 cm using curve ®tting techniques.
The higher frequency mode is attributed to the OH
stretching vibration of aluminium hydroxyls. The
lower frequency vibration may be due to the hydroxyl
groups in the gel, which are hydrogen bonded to
3
2
�
1
acetate groups. Alternatively, the 3475 cm
band
may be due to hydroxyls attributed to water in the
gel. The hydrogen bond causes the frequency to be
shifted to lower values. These bands are observed at
�
1
3
707 and 3142 cm
in the 3008C spectrum. As
thermal degradation of the gel is occurring, these
hydrogen bonds are being removed and the frequency
Acknowledgements
�
1
of the hydroxyl stretch is then observed at 3700 cm
It is proposed that a phase change is occurring between
50 and 3008C, which involves the loss of acetate
.
The Centre for Instrumental and Developmental
Chemistry of the Queensland University of Techno-
logy is gratefully acknowledged for ®nancial support
for this project.
2
groups. The difference between the two spectra at 250
and 3008C in the hydroxyl-stretching region is
observed in the relative intensities of the two bands
�
1
at 3642 and 3475 cm . The intensities are in the ratio
of 1.7±1.5 in the 2508C spectrum and are 1.7±0.7 in
the 3008C spectrum. The intensity of the band at
References
[
[
[
1] B.B. Yoldas, J. Appl. Chem. Biotechnol. 23 (1973) 803.
2] G.C. Bye, J.G. Robinson, Kolloid Z. 198 (1964) 53.
3] W.A. Deer, R.A. Howie, J. Zussman, An Introduction to the
Rock Forming Minerals, Addison-Wesley, Essex, UK, Long-
man, Harlow, London, 1996, pp. 571±572.
�
1
ꢁ
3475 cm has decreased by more than 50% in this
temperature range. The loss of acetate groups may be
observed by the decrease in intensity of the CH
stretching vibrations at ꢁ2927 and 2861 cm� . The
intensity of these bands is approaching zero as the
dehydroxylation occurs. In the 3508C spectrum, two
1
[
[
4] E. Calvet, P. Boivinet, M. Noel, H. Thibon, A. Maillard, R.
Tertain, Bull. Soc. Chim. Fr. (1953) 99.
5] D. Papee, R. Tertain, R. Biais, Bull. Soc. Chim. Fr. (1958)
�
1
weak bands are observed at 3397 and 3632 cm . The
infrared emission spectra show that some acetate
groups are retained up to ꢁ4508C. Such observations
1301.
[6] M. Anast, A. Wong, J.M. Bell, B. Ben-Nissan, Thin-®lm coa-
tings via the sol±gel process, Proceedings of the Australian
Ceramic Society, Austceram Proceedings 1990, Key Engi-
neering Materials, 1990, pp. 427±432.
®
t well with the DTGA pattern where an endogram is
observed in the range 325±3758C (Fig. 4c). Also a
weak band is observed at 1335 in the 2508C spectrum
which shifts to 1320 in the 3508C spectrum and is lost
by 4008C (Fig. 6). This band is attributed to the Al±O±
C bond. Importantly, XRD showed the gel to be
amorphous, the infrared emission spectra shows some
rearrangement of the molecular structures upon ther-
mal degradation.
[
[
7] E.M. Rabinovich, J. Mater. Sci. 20 (1985) 4259.
8] J. Livage, M. Henry, C. Sanchez, Sol±Gel Chemistry of the
Transitional Metal Oxides, Vol. 18, Pergamon Press, Oxford,
1988, p. 259.
[
9] C. Brinker, G. Scherer, Sol±Gel Science: The Physics and
Chemistry of Sol±Gel Processing, Academic Press, San
Diego, CA, 1990.
[
10] B.E. Yoldas, Modi®cation of polymer±gel structures, J. Non-
Cryst. Solids 63 (1984) 145.

4
8
J.L. Szetu et al. / Thermochimica Acta 362 (2000) 37±48
[
11] C. Sanchez, J. Livage, M. Henry, J. Babonneau, Chemical
modi®cation of chemical precursors, J. Non-Cryst. Solids 100
Parkyns (Eds.), Programming of Vacuum Microbalance
Technology, Plenum Press, New York, 1973, 35 pp.
[19] A.M. Vassallo, P.A. Cole-Clarke, L.S.K. Pang, A.J. Palmisa-
no, Appl. Spectrosc. 46 (1992) 73.
(
1988) 65.
[
[
12] T.J. Sato, Appl. Chem. 11 (1961) 207.
13] T. Assih, A. Ayral, M. Abenoza, J. Phalippou, J. Mater. Sci.
[20] R.L. Frost, K. Finnie, B. Collins, A.M. Vassallo, Infrared
emission spectroscopy of clay minerals and their thermal
transformations, in: R.W. Fitzpatrick, G.J. Churchman, T.
Eggleton (Eds.), The Proceedings of the Tenth International
Clay Conference, Adelaide, Australia, CSIRO, Melbourne,
Australia, 1995, pp. 219±224.
2
3 (1988) 3326.
14] E. Morgado, Y.L. Lam, L.F. Nazar, J. Colloid Interf. Sci. 188
1997) 257.
[
[
[
(
15] J.T. Kloprogge, P.J. Dirken, J.B.H. Jansen, J.W. Geus, J. Non-
Cryst. Solids 160 (1993) 144.
16] R.J.M.J. Vogels, J.T. Kloprogge, P.A. Buining, D. Seykens,
J.B.H. Jansen, J.W. Geus, J. Non-Cryst. Solids 191 (1995)
[21] R.L. Frost, A.M. Vassallo, Clay and Clay Minerals 44 (1996)
635.
3
8.
[22] R.L. Frost, J.T. Kloprogge, S.C. Russell, J. Szetu, Appl.
Spectrosc. 53 (1999) 423.
[
[
17] E. Lodding, The gibbsite dehydroxylation fork, in: R.F.
Schwenker, P.D. Gran (Eds.), Thermal Analysis, Vol. 2:
Inorganic Materials and Physical Chemistry, Academic Press,
New York, 1969, pp. 1239±1250.
[23] R.L. Frost, J.T. Kloprogge, S.C. Russell, J. Szetu, Appl.
Spectrosc. 53 (1999) 572.
[24] K. Nakamoto, in: Infrared and Raman Spectra of Inorganic
and Coordination Compounds, Wiley, New York, 1986,
p. 233, 237.
18] J. Rouquerol, F. Rouquerol, Thermal decomposition of
aluminium hydroxides, in: S.C. Beva, S.J. Gregg, N.D.