Porous Al2O3/Al metal ceramics prepared by the oxidation of aluminum powder under hydrothermal conditions followed by thermal dehydration: III. The reactivity of aluminum, the reaction mechanism of its oxidation with water vapor, and the microtexture of cermets

Article abstract of DOI:10.1023/A:1023373016664

The macrokinetics of aluminum oxidation under hydrothermal conditions at 150-250°C and water vapor pressures of ～0.5-4.5 MPa was studied. The apparent kinetic characteristics of the process were determined. The diffusion of water through the layer of a hydrated product was found to be a rate-limiting step. It was found that diffusion coefficients for various aluminum samples differed by almost two orders of magnitude; the main reaction steps were revealed. Changes in the texture of aluminum oxide in cermets in the course of the hydrothermal oxidation reaction were analyzed. This texture was found to depend on the ratio between the specific rate of the oxidation reaction and the rate of aging of the oxidation products under hydrothermal conditions.

Full text of DOI:10.1023/A:1023373016664

Kinetics and Catalysis, Vol. 44, No. 2, 2003, pp. 297–308. Translated from Kinetika i Kataliz, Vol. 44, No. 2, 2003, pp. 322–334.
Original Russian Text Copyright © 2003 by Tikhov, Potapova, Sadykov, Fenelonov, Tsybulya, Salanov, Ivanov, Kolomiichuk.
Porous Al2O3/Al Metal Ceramics Prepared by the Oxidation
of Aluminum Powder under Hydrothermal Conditions
Followed by Thermal Dehydration:
III. The Reactivity of Aluminum, the Reaction Mechanism
of Its Oxidation with Water Vapor, and the Microtexture of Cermets
S. F. Tikhov, Yu. V. Potapova, V. A. Sadykov, V. B. Fenelonov, S. V. Tsybulya,
A. N. Salanov, V. P. Ivanov, and V. N. Kolomiichuk
Boreskov Institute of Catalysis, Siberian Division, Russian Academy of Sciences, Novosibirsk, 630090 Russia
Received August 13, 2002
Abstract—The macrokinetics of aluminum oxidation under hydrothermal conditions at 150–250°C and water
vapor pressures of ~0.5–4.5 MPa was studied. The apparent kinetic characteristics of the process were deter-
mined. The diffusion of water through the layer of a hydrated product was found to be a rate-limiting step. It
was found that diffusion coefficients for various aluminum samples differed by almost two orders of magnitude;
the main reaction steps were revealed. Changes in the texture of aluminum oxide in cermets in the course of the
hydrothermal oxidation reaction were analyzed. This texture was found to depend on the ratio between the spe-
cific rate of the oxidation reaction and the rate of aging of the oxidation products under hydrothermal condi-
tions.
INTRODUCTION
tective film of aluminum oxide. At the point of a maxi-
mum, the specific rates of oxidation were comparable
for all of the tested types of aluminum [3, 5, 8]; this
should be observed when all the reactive islands merge
and the apparent rate of reaction is proportional to the
specific surface area of the metal. However, the specific
rates of reaction were different in different powders
under identical experimental conditions [6].
Previously [1, 2], analytic expressions and empirical
equations relating the pore volume and the specific sur-
face area to the composition of Al2O3/Al composites,
which depends on the conversion of aluminum at the
step of hydrothermal oxidation, were derived. There-
fore, an analysis of factors responsible for the conver-
sion of aluminum in this reaction (temperature, water
vapor pressure, and the specific reactivity of aluminum)
is of crucial importance in the development of
approaches to controlling the properties of cermets.
The conclusions about the nature of reaction prod-
ucts were even less consistent. Thus, according to data
[5], an X-ray amorphous hydroxide was formed at the
initial stage of the reaction; then, boehmite was formed,
and bayerite was predominant after attaining a maxi-
mum rate. Lur’e et al. [3] did not specially study the
nature of reaction products; however, they assumed a
layer-by-layer arrangement of reaction products as the
distance from the metal surface increased: oxide, boeh-
mite, and bayerite. According to Anan’in et al. [7], the
fraction of boehmite in the reaction products of highly
dispersed aluminum is high, whereas the fraction of
bayerite is low. No studies of the texture of oxides
formed after the thermal decomposition of hydroxides
were performed in the cited publications. Note that the
kinetics was studied at temperatures lower than 100°C
when the pressure of water vapor was approximately
constant. Yakerson et al. [9] studied the effects of addi-
tives on the nature of hydrothermal oxidation products
and the texture characteristics of the oxide obtained
The studies of aluminum oxidation with water have
a long history [3–8]. It was found that, at a temperature
lower than the boiling point of water, the reaction rate
passes through a maximum as aluminum undergoes
oxidation [3, 5, 8]. The value and position of a maxi-
mum rate depend on the temperature of experiments
and on the average particle size of aluminum powder.
Thus, in coarsely dispersed powders, the time and con-
version at which the rate reaches a maximum decreases
with temperature; ~3–5% conversion at 60–70°C is
reached in ~5–10 min [5, 7]. In submicron powders, a
broad maximum of the rate lies in the range from ~8 to
~20% conversion [8]. Similar activation energies of 74
[3] and 54–113 kJ/mol [7] were published. Zhilinskii et al.
[6] noted an increase in the activation energy from ~70
to ~108 kJ/mol with the dispersity of powders. The
majority of authors believe that the occurrence of a
maximum rate is due to the formation and growth of the after thermal decomposition. They found that pseudo-
reactive islands of a porous hydrated product in a pro- boehmite and bayerite resulted from the interaction
0023-1584/03/4402-0297$25.00 © 2003 MAIK “Nauka /Interperiodica”

298
TIKHOV et al.
Table 1. Diffusion coefficients for aluminum metal of different origins in the reaction of hydrothermal oxidation at 150°C
and a vapor pressure of 0.5 MPa; characteristics of the parent aluminum samples
SAXS data for
Impurity concentrations**
parent samples*
D × 1013,
Sample
cm2/s
Na
K
Mg
Ca
Fe
d, Å I, arb. u.
V
S
V
S
V
S
V
S
V
S
PAVCh
PA-4
Foil
167
10
2
103
47
105
58
0.01
0.02
0.01
8.0
4.4
0.02
0.02
0.03
9.3
2.8
4.7
0.07
0.07
0.01
4.0
4.1
2.2
0.002
0.02
0.04
4.6
2.6
2.4
0.01
0.3
15.0
***
2.7
38
<10
13.1
0.2
* d is the average size of nonuniformity regions; I is the SAXS signal intensity.
** V is the bulk impurity concentration (according to chemical analysis data), wt %; S is the enrichment of a surface layer of ~20 Å with
impurities (according to SIMS data), arb. u.
*** The concentration is lower than the mass-spectrometric detection limit.
with water, and the ratio between them depended on tion and the nature of the reaction products, to compare
temperature and the amount of activator (In or Ga). In the specific reactivities of various types of aluminum
this case, the concentration of bayerite increased with metal, and to study the effect of the kinetics of hydro-
temperature and the amount of activators. However, thermal oxidation of aluminum on the microtexture of
the kinetics of hydrothermal oxidation was not studied the Al2O3/Al cermets obtained after calcination.
in [9].
In the only publication [7] where the oxidation pro-
cess was studied at 120–230°C, kinetic measurements
EXPERIMENTAL
Aluminum powders of PA-4 and PAVCh grades and
aluminum foil were used as starting reagents. The spe-
cific surface areas of the samples were 0.8, 1.2, and
0.0042 m2/g, respectively. Bulk impurity concentra-
tions were determined by atomic absorption spectro-
photometry (AAS-1) and atomic ionization spectrome-
try (Plasma Spectrovac). The chemical composition of
reagent surfaces was determined by secondary ion mass
spectrometry (SIMS) on an MS-7201 mass spectrome-
ter (the energy of an Ar+ beam was 4 keV at an etching rate
of ~6 Å/min and a current density of 10–20 µA/cm2; the
thickness of a test layer was ~20 Å). The powders were
rubbed into a holder of high-purity indium. Table 1
summarizes the concentrations of impurities in the
samples. The texture characteristics were determined
on an ASAP 2400 Micromeritics instrument using the
adsorption of nitrogen at 77 K.
The conditions of kinetic experiments were
described in detail elsewhere [10, 11]. We varied tem-
perature (150, 210, and 250°C), time (0.5–6.5 h), and
water vapor pressure (500–4500 kPa). The temperature
and water vapor pressure ranges were chosen because
mechanically strong porous cermets are formed under
these conditions [1, 7], and this was the main goal of
this work. The final conversions of aluminum varied
over the ranges ~0.1–4.0% for foil, ~7–25% for PA-4,
and ~23–52% for PAVCh. The measurement error in
were performed using a simplified procedure: the con-
version of aluminum after 2.5 h was determined at dif-
ferent temperatures. Two types of aluminum powders
of different dispersity were studied. It was found that
boehmite was the main product of hydrothermal oxida-
tion under these conditions; however, the Al(OH)3 tri-
hydrate appeared as the temperature was decreased.
The formation of product in the form of whiskers was
mentioned, which was presumably explained by the
occurrence of local centers of the preferred oxidation of
aluminum (like dislocations). Additional studies of the
resulting cermets after calcination in air revealed non-
monotonic changes in the specific surface area depend-
ing on the temperature of hydrothermal treatment.
However, the dependence of the rate of reaction on the
pressure of water vapor was not determined in [7], and
no rate equations, which make it possible to compare
quantitatively the specific reactivity of aluminum of
different genesis, were derived.
The previously obtained dependence of conversion
on temperature and water vapor pressure was described
by formal kinetics using the Roginskii–Schulz equation
for a compressible sphere, which assumes the absence
of detectable diffusion inhibition by a reaction product
[10]. However, based on a more detailed analysis of
kinetic curves, we subsequently concluded that the
derived relations are closer to diffusion kinetics.
The aim of this work was to describe the data in aluminum conversion was ±10% of the determined
terms of diffusion kinetics, to reveal a correlation value; it was mainly due to technical problems in the
between quantitative kinetic characteristics of the reac- determination of sample weights in a press mold before
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

POROUS Al2O3/Al METAL CERAMICS
299
(a)
(b)
α
0.10
0.5 MPa
1.5 MPa
2.0 MPa
3.5 MPa
4.5 MPa
0.010
0.005
150°ë
150°ë
0.05
0
0.15
0
0.020
0.10
0.05
0.015
0.010
0.005
210°ë
210°ë
0
0.25
0
0.04
0.03
0.02
0.01
0
0.20
0.15
0.10
0.05
250°ë
250°ë
0
1
2
3
4
5
6
7
0
1
2
3
4
5
6
7
t, h
t – 0.5, h
Fig. 1. (a) Aluminum conversion (α) as a function of the time of hydrothermal oxidation at different temperatures and water vapor
pressures for the PA-4 powder. (b) The above function on the coordinates of a diffusion equation [13].
and after hydrothermal treatment and calcination. The the use of electron microscopy, we found that, even at
vapor pressure was kept constant during each particular the initial stages of reaction (t = 0.5–2.0 h), the surface
experiment. The procedures used for studying the sam- of aluminum was fully coated with a porous product
ples by scanning electron microscopy, X-ray diffraction layer ~1 µm in thickness. This product was signifi-
(XRD) analysis, small-angle X-ray scattering (SAXS), cantly different in morphology from the initial oxide
and adsorption techniques were described in detail pre- films of powders [12] and aluminum foil [13]. Thus, in
viously [2, 12]. The SAXS signal intensity was deter- our case, the region in which the reagent is completely
mined in an angle region of 7'–7° with reference to a coated with a product layer of aluminum hydroxide is
primary X-ray beam in air.
the most likely region of the occurrence of a topochem-
ical reaction of aluminum oxidation with water.
An insignificant change in the specific surface area
of the reagent in the tested region of aluminum conver-
sions is another circumstance that is important for the
subsequent kinetic analysis. This is approximately
obeyed for aluminum foil because of extremely low
aluminum conversions. This assumption was also sup-
ported by possible changes in the specific surface area
of aluminum powders, as estimated taking into account
their almost ideally spherical shape [12]. Indeed, the
reaction rate is proportional to the mass of the remain-
ing reagent, which depends on the conversion (α) and
mass (m0) of the starting reagent. Moreover, the mass of
the remaining reagent is proportional to its volume,
which depends on the particle radius (in the monodis-
perse approximation):
RESULTS AND DISCUSSION
1. General Analysis of Kinetic Curves
Figures 1a, 2a, and 2c demonstrate the time depen-
dence of the conversion at different pressures of water
vapor and temperatures for the PA-4 powder, the foil,
and the PAVCh powder, respectively. As compared to
previously published data [3], it can be seen that these
plots are not S-shaped curves, which are characteristic
of many topochemical reactions, with an induction
period at the initial stage of reaction. In the oxidation of
aluminum with water, this induction period is usually
related to the slow degradation of a protective oxide
film and the formation of a loose hydrated phase at the
initial stage of reaction [4]. Under the conditions of our
experiments, it is most likely that we skipped this stage,
as well as the stage at which the rate increased to its
w = km0(1 – α ) ρreag4/3πR3,
(1)
maximum in the first half hour of reaction, before the where ρreag is the true density of the reagent and R is the
analysis of the first experimental point. Moreover, with particle radius. On the other hand, to a first approxima-
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

300
TIKHOV et al.
α2–α20.5
0.020
α
0.007
0.006
0.005
0.004
0.003
0.002
0.001
(‡)
(b)
0.015
0.010
0.005
0
1 2 3 4 5 6 7
(c)
0
0.25
1
2
3
4
5 6 7
(d)
0.6
0.5
0.4
0.3
0.2
0.1
0
0.20
0.15
0.10
0.05
0
0
1
2
3
4
5
0
1
2
3
4
5
t, h
t – 0.5, h
Fig. 2. Aluminum conversion (α) as a function of the time of hydrothermal oxidation at 150°ë and a water vapor pressure of
0.5 MPa for (a) aluminum foil or (c) PAVCh; the above function on the coordinates of a diffusion equation [13] for (b) aluminum
foil or (d) PAVCh.
tion, the specific surface area of spherical particles is teristics similar to those of the PA-4 powder used in this
proportional to the squared particle radius:
study (by the extrapolation of data obtained below the
boiling point of water to T = 150°C using the equation
wsp = 1022 exp(–17500/RT) [3]). This comparison
showed that the extrapolated rate of reaction is higher
than the experimental value by two orders of magni-
tude. Usually, the majority of authors cited comparable
reaction rates (at least at temperatures lower than
100°C) for a maximum point when the reaction zone is
proportional to the specific surface area. The difference
between the reaction rate taken from [3], which was
extrapolated to high temperatures, and the experimental
value obtained in this work cannot be explained only by
the difference in the specific reactivities of different
aluminum powders or by the difference in water vapor
pressures. Most likely, this may be due to a change in
the reaction mechanism (the shape of a kinetic curve) at
high temperatures, as compared with temperatures below
100°C. Consequently, the equation derived by Lur’e et al.
[3] is inapplicable to our conditions, and the reaction
mechanism is significantly different. Note that at the initial
portion (t = 0–0.5 h) the reaction rate for the PA-4 powder
was much higher (~1 × 10–6 mol Al cm–2 g–1). Thus, the
experimental rate of aluminum oxidation at the initial
portion was much closer to the maximum point of the
rate of a topochemical reaction, which could be
assumed on the extrapolation of published data to be
150°C.
S
4πR2 (1 – α )2/3.
(2)
With a maximum change in α from 0 to 0.25 (see
Fig. 1a), the reagent surface area is changed by a factor
of 0.81, which lies within the limits of experimental
error in kinetic measurements. The surface roughness
of real particles usually results in a significant increase
in the magnitude of the specific surface area of the
reagent. However, it is most likely that the contribution
of roughness decreases the dependence of the specific
surface area on the conversion because a topochemical
reaction hardly results in smoothing the reagent sur-
face. Thus, to a first approximation, the specific surface
area of aluminum can be considered constant over the
entire ranges of tested aluminum conversions, temper-
atures, and water vapor pressures.
On this basis, we estimated the average apparent
specific rates of reaction in the region t = 0.5–6.5 h for
PA-4 with the use of the simplified expression
dα
dt SspMAl
1
–4
---------------------
wsp
=
× 10 mol Al cm–2 h–1,
(3)
where MAl is the molar weight of aluminum and Ssp is
its specific surface area.At 150°C, the specific rate of reac-
tion was found to equal ~0.04 × 10–6 mol Al cm–2 h–1. We
compared this value with the specific rate of reaction at
a maximum point (wsp ≈ 5 × 10–6 mol Al cm–2 h–1),
The results allowed us to conclude that under the
which was obtained for an ASD-4 powder with charac- conditions of our experiments the majority of experi-
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

POROUS Al2O3/Al METAL CERAMICS
301
mental points occurred in a region after the maximum guished from an analysis of specific properties of the
point, when the reaction rate of aluminum oxidation test system [16]. Indeed, a subsequent detailed analysis
decreased with time.
of the kinetic curves revealed that in our case a change
in the derivative of conversion with respect to time was
much greater (Fig. 1a) than the possible change in the
specific surface area of the reagent.
The diffusion coefficients were estimated from the
values of k' based on the assumption that the specific
surface area of the reagent is approximately constant
using the equation proposed by Rozovskii [16]. More-
over, because this equation was derived on the approx-
imation of a one-dimensional problem, it was reason-
able to change to the bulk reagent (water vapor) con-
centrations. This allowed us to obtain the following
expression for the diffusion coefficient:
2. Diffusion Coefficients for Different
Aluminum Samples
It is well known that a decrease in the rate of a
topochemical reaction after reaching a maximum is
usually due to an increase in the thickness of a product
layer, which inhibits the reaction (the reaction rate is
proportional to the amount of a converted substance)
[16]. To estimate the coefficients of diffusion through a
product layer, we used the well-known parabolic equa-
tion [16, 17]
MAlOOHm20, Alk'
2C0ρAlOOHSAlM2Al
α2 – α20.5 = k'(t – 0.5),
(4)
M
AlOOHk'
------------------------------------------
D = ----------------------------------------- =
,
(5)
2Cv ρAlOOHS2spMA2 l
where α and α0.5 are the conversions of aluminum at
times t and 0.5 h, respectively. The modified parabolic
equation was used because at the initial portion (0–0.5 h)
the state of the catalyst surface (the mechanism of a het-
erogeneous reaction) is uncertain; this uncertainty
resulted in a significant change in the function α(t), as
compared with the subsequent points (Figs. 1, 2).
C0
------ ------------
, D is the diffusion coefficient
C0
where Cv =
=
SAl Sspm0
(cm2/s), ρAlOOH is the density of aluminum hydroxide
produced from aluminum, M is the molar weight of alu-
minum and the hydroxide, C0 is the reagent concentra-
tion at the outer surface of the product in a one-dimen-
sional approximation (mol/cm), SAl is the geometric
surface of the reagent, and Ssp is the specific surface
An analysis of kinetic curves linearized on the coor-
dinates of a parabolic equation (Figs. 1b, 2b, 2d) dem-
onstrated that the experimental points for the foil,
PAVCh, and, partially, PA-4 adequately lay along a
straight line. In the PA-4 powder, some of the points at
150 and 210°C significantly deviated from the linear-
ized curves. The standard deviation of k', which is
determined by the slopes of these kinetic curves
(Fig. 1b), may be up to 30% due to the deviation of the
specific surface area of the reagent from the initial
value. The effect of diffusion processes within a grain
of the porous cermet formed may be another factor that
can deviate linearized curves from the origin of the
coordinates and increase the scatter of experimental
points. Thus, the linearized curves for aluminum foil,
where this factor is insignificant, clearly passed through
the origin (Figs. 2a, 2b). The change of the porous
structure of the composite with time complicates the
quantitative description of the diffusion effect within a
grain [1]. However, these factors did not appear in the
PAVCh powder. It is likely that the transition to the dif-
fusion region was incomplete for PA-4. At the same
time, as a first approximation, it is believed that the
reagent diffusion through a hydrated product layer was
predominant in the reaction of aluminum with water
under our conditions.
area of the reagent (m2/g).
In Eq. (5), the density of aluminum hydroxide was
estimated from data on the density and composition of
AlOOH/Al cermets. At 150°C, it was 2.4, 2.7, and
2.4 g/cm3 for PA-4, PAVCh, and foil, respectively. For
PA-4, the density of aluminum hydroxide at 210 and
250°C was taken to be equal to 3.0 and 3.4 g/cm3,
respectively. These values are somewhat different from
the tabulated data in [18]. The reasons for this phenom-
enon were discussed previously [2]. The concentration
of water vapor was calculated using the ideal gas equa-
tion
m
P
H2O
----------
n
Cv = --- =
,
(6)
RT
V
where PH O is the pressure of water vapor in an auto-
2
clave at temperature T. The dependence of the constant
of a hyperbolic equation (k') on the pressure of water
vapor was preestimated (the apparent order of reaction
(m) (Fig. 3a) was 0.9). The sufficiently strong depen-
dence on the pressure of water vapor allowed us to con-
clude that, in accordance with [6], the diffusion of
water rather than aluminum cations is the rate-limiting
step of the process.
Note that these data were adequately linearized pre-
viously [10] on the coordinates of the Roginskii–
Schulz equation, which is based on the assumption that
the apparent rate of reaction is proportional to a change
Note that, according to the Kelvin equation, which
in the reagent surface area. In this case, a paradoxical determines the minimum size of pores in which water
situation occurred when the same experimental data are vapor can condense [14], the surface of cermets was
described by different rate equations within the accu- completely covered with water in the course of hydro-
racy of the experiment. The models should be distin- thermal oxidation at equilibrium pressures and all tem-
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

302
TIKHOV et al.
ln D, [cm2/s]
lnk', [h–1]
–20
–22
–24
–26
–28
–30
–32
–34
(a)
(b)
–5.2
–5.6
–6.0
–6.4
–6.8
–7.2
200°C
250°C
1.2
1.8
2.4
3.0
3.6
1.9 2.0 2.1 2.2 2.3 2.4
1/T × 103, K–1
lnPH O, [MPa]
2
Fig. 3. (a) Determination of the reaction order with respect to water for the oxidation of PA-4 from data shown in Fig. 1. (b) The
temperature dependence of diffusion coefficients on the Arrhenius coordinates for PA-4.
peratures. At lower vapor pressures, the radius of pores tically no peaks with (111) indices, as distinct from alu-
in which water condensation is possible decreased from minum powder samples. This difference was also
90 to 3 Å as the pressure was decreased from 1.5 to retained after partial oxidation of aluminum. This fact
0.5 MPa at 210°C or from 15 to ~1 Å as the pressure is indicative of a preferential orientation of crystallites
was decreased from 3.5 to 0.5 MPa at 250°C, as calcu- with respect to the surface of foil, probably due to the
lated using the Laplace–Kelvin equation. Thus, in the effect of rolling in the course of its manufacture. How-
majority of kinetic experiments under conditions of ever, this phenomenon cannot explain the considerable
hydrothermal oxidation, water can condense in only difference between the diffusion coefficients of the two
micropores, which usually occur in the interlayer space powdered samples.
of hydroxides, and the amount of water decreases with
We also determined bulk and surface impurity con-
temperature. However, as distinct from the reaction of
centrations in the test samples of aluminum. We found
iron oxidation with water [15], no considerable changes
that the bulk concentrations of alkaline-earth metal
were observed in the dependence of the conversion on
impurities in all of the samples were comparable (Table 1).
the time of hydrothermal oxidation under conditions of
The enrichment of an aluminum surface with impuri-
an equilibrium vapor pressure, nor in the dependence
ties, as compared with the bulk, is typical (Table 1). The
obtained at lower pressures (Fig. 1a). It is likely that,
surface concentrations of alkaline-earth metals in the
under conditions when the diffusion of water vapor is a
PAVCh and foil samples were several times higher than
rate-limiting step, the concentration of water at the sur-
that in PA-4. The foil and PA-4 differ from PAVCh in
face of a solid reagent (aluminum) is close to zero, and
that they have considerably higher bulk iron contents.
the apparent rate of reaction depends on both water
Generally, no correlation between the diffusion coeffi-
concentration at the outer surface of a hydroxo com-
cients and impurity concentrations was observed.
pound and the diffusion permeability of this compound.
The SAXS data, which were used for the analysis of
Table 1 indicates that the diffusion coefficients
the texture characteristics of oxide films that cover alu-
obtained at 150°C in different samples differ by almost
minum, allowed us to conclude that changes in the con-
two orders of magnitude. Because the reaction is lim-
centrations of regions with a nonuniform electron den-
ited by diffusion, the diffusion coefficient practically
sity and in diffusion coefficients exhibit similar patterns
characterizes the specific reactivity of aluminum. Note
(greater SAXS signal intensities correspond to greater
that considerable differences in the diffusion coeffi-
values of D) (Table 1). Indeed, previously [2], it was
cients of water at temperatures lower than 100°C were
found that SAXS data characterize both qualitatively
also observed previously in different aluminum pow-
(by the distribution of these regions) and quantitatively
ders. This was related to differences in the dispersity of
(by the intensity of signals) oxide rather than metal
powders and, as a consequence, in the degrees of crys-
components of cermets. The SAXS signal as an inte-
tallinity of protective oxide layers [6]. However, in our
grated characteristic depends on the sizes of both pri-
case, the average particle sizes of both of the powders
mary particles and oxide pores. Taking into account
were similar and equal to 20–25 µm, as evaluated using
the Coulter method [12].
that in this case SAXS characterizes oxide films, which
cover the surface (partially hydrated because of contact
Note that the X-ray diffraction patterns of aluminum with humid air) of starting samples, it is believed that
foil, which exhibits the lowest reactivity, contain prac- the diffusion coefficients essentially depend on the tex-
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

POROUS Al2O3/Al METAL CERAMICS
303
Boehmite
Bayerite
Aluminum
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
3
2
1
*
*
*
*
* *
*
*
*
0
20
40
60
80 2θ, deg
Fig. 4. X-ray diffraction patterns of aluminum powder samples after hydrothermal oxidation (HTO) and drying at 120°ë: (1)
PAVCh, T = 150°C, t = 0.5 h; (2) PAVCh, T = 150°C, t = 5.0 h; (3) PA-4, T = 250°C, t = 6.5 h.
HTO
HTO
HTO
HTO
HTO
HTO
ture characteristics of these films, primarily, on pore after 5 h; it transformed into boehmite. In the less reac-
size and volume (“free” area of the film). The relation
of the oxide film texture of the parent metal to the tex-
ture of the hydrated oxidation product can be explained
by a memory effect. In this case, the structure and tex-
ture characteristics of aluminum hydroxides depend on
the structural characteristics of particles formed at the
early stages of aluminum oxidation; these particles
were retained even after long-term aging. These ori-
ented growth processes, in which the particle structure
and texture of aging products essentially depend on the
structure and texture of parent reagent particles, are
well known [19]. It is likely that stronger correlations
can be obtained from a comparison of SAXS intensities
with the initial rates of oxidation or from a study of
hydrothermal oxidation products (aluminum hydrox-
ides), which was beyond the scope of this work. The
effect of product properties on kinetic characteristics was
also supported by the apparent activation energy, which
was formally found to be negative, about –16 kJ/mol,
which is close to zero (Fig. 3b). However, the negative
temperature dependence of diffusion coefficients can
also be explained by product consolidation and, as a
consequence, a decrease in the diffusion permeability
with temperature.
tive PA-4 powder, boehmite as the product of hydro-
thermal oxidation was observed at all the contact times
over the entire range of tested temperatures (Fig. 4).
Thus, under the conditions of our hydrothermal oxida-
tion experiments, boehmite, which is more stable, was
formed from bayerite. Such a reaction sequence of
hydroxides on aging under hydrothermal conditions
was also observed previously [19]. Under our condi-
tions, it is likely that the ratio between these two species
depends on the ration between the rates of bayerite for-
mation and rearrangement into boehmite. The higher
the specific reactivity of aluminum, the higher the
above ratio. An amorphous hydroxide is the most prob-
able primary porous product of oxidation. A halo at 20°
in the X-ray diffraction patterns (Fig. 4) supported the
occurrence of the above hydroxide in all of the samples
of oxidation products. In this case, it is most likely that
the formation of bayerite results from the aging of a
loose amorphous product; this behavior is typical of
hydroxide aging processes [19]. Note that the aging of
hydroxides should become slower with decreasing
pressure of water vapor. Indeed, a comparison of
Figs. 5a and 6a–6c clearly indicates that the size and the
degree of crystallinity of product particle agglomerates
decrease with decreasing pressure of water vapor in an
autoclave.
3. The Nature of Products and the Sequence
of Hydrothermal Oxidation Reaction Steps
Thus, the following main steps responsible for the
The X-ray diffraction data (Fig. 4) indicate that bay- chemistry of hydrothermal oxidation can be recog-
erite was formed at the initial stages of reaction in the
hydrothermal oxidation of the highly reactive PAVCh
powder. At the onset of reaction, bayerite disappeared
nized:
1) H2Og
H2Oads,lig (on AlOx(OH)y);
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

304
TIKHOV et al.
1 µm
(‡)
1 µm
(‡)
1 µm
(b)
1 µm
(b)
1 µm
(c)
Fig. 5. Electron micrographs of Al O /Al composites pre-
2
3
pared from (a) PAVCh, (b) PA-4, and (c) aluminum foil by
hydrothermal oxidation at 250°C for 0.5 h followed by cal-
cination in air.
1 µm
(c)
2) H2Oads,lig (on AlOx(OH)y)
H2Oads
(on the surface of Al0);
Fig. 6. Electron micrographs of Al O /Al composites pre-
2
3
3) Al0 + H2Oads (on the surface of Al0)
pared from PA-4 by hydrothermal oxidation at 250°C and
the following vapor pressures, MPa: (a) 3.5 (t = 0.5 h),
HTO
(AlOx(OH)y)amorphous + ç2;
(b) 2.0 (t
= 3.5 h), and (c) 0.5 (t
= 6.5 h).
HTO
HTO
4) (AlOx(OH)y)amorphous
5) Al(OH)3 AlOOH + H2O.
Most likely, these steps are not elementary. The adsorp-
Al(OH)3;
compound of aluminum takes place at step (1). At step (2),
water diffusion becomes limiting after the formation of
tion of water (condensation) in the pores of a hydroxo a continuous product layer. Step (3) characterizes the
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

POROUS Al2O3/Al METAL CERAMICS
305
Ssp, m2/g Al2O3
oxidation of aluminum with water and is very fast.
Steps (4) and (5) characterize the evolution of reaction
products. So-called polynuclear hydroxo complexes
[19], whose structure can depend on external condi-
tions, impurity concentrations, etc., can be the loose
products that form amorphous aluminum hydroxide. A
more detailed description of the hydrothermal oxida-
tion reaction requires a much greater body of kinetic
information.
350
(a)
300
250
200
150
4. Effect of Product Aging under Hydrothermal
Conditions on the Microtexture of Aluminum Oxide
in Cermets after Calcination
Under conditions of hydrothermal oxidation, the
resulting hydroxide particles undergo aging simulta-
neously with the chemical processes described above.
This affects both the texture of these particles and the
final texture of aluminum oxide after calcination [2].
Indeed, the specific surface area of aluminum oxide in
cermets decreased (at the stage of hydrothermal oxida-
tion) from ~320 to ~200 m2/g for PAVCh, from ~250 to
~150 m2/g for PA-4, and from ~350 to ~70 m2/g for the
foil as the conversion increased (see Fig. 7 and Fig. 1 in
[2]). Anan’in et al. [7] observed a similar decrease in
the specific surface area of aluminum oxide in Al2O3/Al
cermets as the conversion of aluminum increased (upon
the corresponding quantitative treatment of experimen-
tal data). In this case, two tendencies can be recognized
in changes in the specific surface area of an oxide
phase:
0
5
10 15 20 25 30 35 40 45 50 55
(b)
400
300
200
100
150°ë
200°ë
0
0.5
1.0
1.5
20
25
α, %
(a) A decrease in the specific surface area with the
degree of reaction to an approximately constant level
(Fig. 7, Table 2).
(b) A decrease in the “constant level” of the specific
surface area with decreasing reactivity of aluminum
(Figs. 7 and 1 in [2]).
Fig. 7. Specific surface area (S ) of aluminum oxide in
sp
Al O /Al cermets formed after hydrothermal oxidation and
2
3
calcination as a function of aluminum conversion (α) in the
hydrothermal oxidation of (a) PAVCh and (b) foil samples.
Taking into account a correlation between the spe-
cific surface areas of hydroxides and oxides, we can
conclude that this phenomenon is analogous to the
aging of products deposited under hydrothermal condi-
tions, which was described in considerable detail for
the hydroxo compounds of aluminum [19, 20]. This
conclusion was also supported by the fact that boeh-
mite, which is most stable under these conditions, was
the final product of both hydrothermal oxidation at
150–250°C and aging at these temperatures [19, 20].
approached the H2 type, which is characteristic of PA-4
(Fig. 8), as the degree of reaction increased. Figure 9
shows the adsorption isotherms for cermets prepared
from PAVCh as comparative plots [21, 22]. A standard
N2 adsorption isotherm was used as a reference stan-
dard (the axis of abscissas) for constructing the com-
parative plots. This standard adsorption isotherm was
obtained by averaging over several nonporous samples
of different chemical nature [21]. This isotherm coin-
cided with the standard adsorption isotherm proposed
A correlation between the diffusion coefficients and by Gregg and Sing [14] in the region of relative pres-
the microtexture of aluminum oxide was found in a tex-
ture study based on a detailed analysis of nitrogen
adsorption–desorption isotherms at 77 K for cermets
prepared from PAVCh in comparison with PA-4 [2]. It
was impossible to study cermets prepared from the foil
because of the low oxide content.
Thus, we found that the shape of a hysteresis loop in
the PAVCh cermet formed at the early stages of hydro-
thermal oxidation was closer to the H3 type according
sures P/P0 < 0.7. The experimental adsorption deviated
from linearity in the region of minimum coverages
because of the effect of the nature of the surface, which
has no effect on specific surface areas measured from
the slopes of the linear plots [22]. Dotted lines in these
plots demonstrate two linear portions: first, in the
region of comparatively low P/P0 and, second, in the
region of high P/P0 for the desorption branch of the iso-
therm. The first linear portion in Figs. 9a and 9b is
to the IUPAC classification [14]. However, it extrapolated to the origin of the coordinates; this fact is
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

306
TIKHOV et al.
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
2
1
Adsorption
Desorption
20
10
–10
–20
1.0
0
0.2
0.4
0.6
0.8
P/P0
Fig. 8. Adsorption–desorption isotherms of nitrogen at 77 K for Al O /Al composites prepared from PAVCh; t = (1) 0.5 or
HTO
2
3
(2) 5.0 h. The gas volume was converted to normal temperature and pressure (NTP).
indicative of the absence of micropores. (Strictly speak- sponds to the remainder surface after the filling of
ing, the extrapolated comparative plot in Fig. 9 inter- mesopores in aggregates (Souter). The mesopore volume
sects the axis of abscissas at the point of ~0.0005 due to (Vs), which was determined by the extrapolation of the
the contribution of ultramicropores, whose volume is second linear portion to the axis of ordinates, also
extremely small.) Note that, previously, the linearization
of this portion for cermets prepared from PA-4 was
somewhat different; however, the contribution of
micropores to the total pore volume was insignificant [2].
decreased. This fact suggests that aging processes
occur simultaneously with the chemical process of alu-
minum oxidation and affect the texture properties of the
product: an increase in the size of primary particles, a
change in the character of their packing, and an
increase in the size of aggregates formed by these par-
ticles.
Table 2 summarizes the results of an analysis of
adsorption isotherms for cermets prepared from PAVCh
(Fig. 8). It can be seen in Table 2 that the outer surface
of aggregates in the cermets prepared from the PAVCh
powder decreased with the time of hydrothermal oxida-
tion. This outer surface was found from the slope of the
The specific pore volumes (Va) and the specific sur-
face areas of aluminum oxides in cermets (SAl O ) are
2
3
second linear portion of the isotherm, which corre- also given in Table 2. These data were obtained from
Table 2. Microtexture characteristics of the Al2O3/Al composites prepared from the PAVCh powder according to nitrogen
adsorption–desorption isotherms
Pore volume of the Specific surface area
Texture characteristics of Al2O3
composite cm3/g of the composite, m2/g
Conditions
Al2O3 con-
Sample of hydrothermal
tent, wt %
Va, Al O , SΣ, Al O , SBET, Al O
,
oxidation
2
3
2
3
2
3
Vs
Va
SBET
SΣ
Souter
cm3/g
m2/g
m2/g
PAVCh
(0.5 h)
150°C, 0.5 h,
38.7
58.4
0.094
0.090
0.094
0.090
139
121
129
135
14.8
11.6
0.243
0.154
333
231
359
205
0.5 MPa
PAVCh
(5.0 h)
150°C, 5.0 h,
0.5 MPa
Note: The pore volume was estimated using the balance equation V = V (1 – y) + V y, where y is the oxide fraction in the cermet. The
s
Al
a
specific surface area was estimated using the balance equation S = S (1 – y) + S
y. V is the specific volume of micropores and
a
Σ
Al
Al2O3
mesopores within aggregates. V is the mesopore volume. S is the total specific surface area of composites found by a comparative
s
Σ
analysis of adsorption isotherms. S
is the outer surface area of aggregates.
outer
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

POROUS Al2O3/Al METAL CERAMICS
307
Adsorption of N2 (liquid), cm3/g
the composition of samples on the assumption that both
the pore volume and the specific surface area of the
oxide component are much greater than those of pow-
dered aluminum, as was found previously [2]. A com-
parison between the texture properties of aluminum
oxides obtained from PAVCh (Table 2) and PA-4 [2]
suggests that, given the same synthesis time, both the
total specific surface area and the pore volume refer-
enced to the weight of the oxide were greater in Al2O3
from PAVCh. This can easily be explained by the much
higher rate of oxidation of the PAVCh aluminum.
Under comparable conditions of aging and oxidation of
PAVCh under equal conditions, the fraction of smaller
product particles is much higher; this results in the
observed texture differences. Another difference
between hydrothermal oxidation products from PAVCh
and PA-4 is as follows: the former exhibited a much
greater (by a factor of 4 to 5) outer surface (and hence
a smaller size) of aggregates of primary particles. Con-
sequently, under comparable external conditions of
hydrothermal oxidation, higher specific reactivity
resulted in the formation of smaller particles, which
formed smaller aggregates. This can clearly be seen by
comparing the scanning electron micrographs of
PAVCh and PA-4 (Fig. 5). It is likely that aging pro-
cesses rather than nonuniformities in the growth of
product particles at individual local points, as assumed
by Anan’in et al. [7], are primarily responsible for the
formation of elongated well-crystallized aggregates.
The behavior of particles formed in the hydrothermal
oxidation of the foil (Fig. 5c) is inconsistent with the
above pattern. This can be explained by the consider-
able change in the conditions of water vapor condensa-
tion on the surface of a plate as compared with pores
between the particles of powdered aluminum; this
change can affect the conditions of hydroxide aging.
0.14
(‡)
0.12
0.10
0.08
0.06
0.04
0.02
0
0.12
0.0004
0.0008
(b)
0.0012
0.10
0.08
0.06
0.04
0.02
0
0.0004
0.0008
0.0012
Standard adsorption of N2 (liquid), cm3/m2
Fig. 9. Comparative plots of the adsorption–desorption iso-
therms of nitrogen at 77 K for composites prepared from
Turning back to the reasons for differences in diffu-
sion coefficients, we believe that the texture character-
istics of the aluminum oxide formed (and hence the
hydroxide) can significantly affect the diffusion of
water through the product layer. As noted above, a
denser product more strongly inhibits the oxidation
reaction. Indeed, the average specific volume of
micropores and mesopores in aluminum oxide from
PAVCh (Table 2) was greater by 20–30% than that for
PA-4 (Table 2 in [2]).
PAVCh (Fig. 8); t
= (a) 0.5 or (b) 5.0 h. Open and closed
HTO
circles correspond to adsorption and desorption, respec-
tively. The gas volume was converted to the volume of liq-
uid condensed at 77 K. Data for the standard N adsorption
2
isotherm (see the text) are plotted on the abscissa.
ACKNOWLEDGMENTS
This work was supported by the Russian Foundation
for Basic Research (project no. 99-03-32853). We are
grateful to A.Ya. Rozovskii for helpful discussions of
the results.Yu.V. Potapova acknowledges the support of
INTAS (grant no. YFS-00-199).
Thus, we studied the macrokinetics of the hydro-
thermal oxidation of aluminum at elevated tempera-
tures, found the apparent kinetic parameters of the pro-
cess, and compared the diffusion coefficients of various
aluminum metal samples. We made assumptions about
the factors responsible for differences in the diffusion
coefficients and about the nature of the main reaction
steps. We analyzed the relationship between the micro-
texture and the specific reactivity of the resulting cer-
mets.
REFERENCES
1. Tikhov, S.F., Fenelonov, V.B., Sadykov, V.A.,
Potapova, Yu.V., and Salanov, A.N., Kinet. Katal., 2000,
vol. 41, no. 6, p. 907.
2. Tikhov, S.F., Zaikovskii, V.I., Fenelonov, V.B.,
Potapova, Yu.V., Kolomiichuk, V.N., and Sadykov, V.A.,
Kinet. Katal., 2000, vol. 41, no. 6, p. 916.
KINETICS AND CATALYSIS Vol. 44 No. 2 2003

308
TIKHOV et al.
3. Lur’e, B.A., Chernyshev, A.N., Perova, N.N., and Svet- 13. Tikhov, S.F., Chernykh, G.V., Sadykov, V.A., Sala-
lov, B.S., Kinet. Katal., 1976, vol. 27, no. 6, p. 1453.
4. Lur’e, B.A., Chernyshev, A.N., and Svetlov, B.S., Kinet.
Katal., 1977, vol. 28, no. 4, p. 842.
5. Zhilinskii, V.V., Lokenbakh, A.K., and Lepin’, L.K., Izv.
Akad. Nauk Latv. SSR, 1986, no. 2, p. 151.
6. Zhilinskii, V.V., Lokenbakh, A.K., and Lepin’, L.K., Izv.
Akad. Nauk Latv. SSR, 1988, no. 5, p. 622.
nov, A.N., Alikina, G.M., Tsybulya, S.V., and
Lysov, V.F., Catal. Today, 1999, vol. 53, p. 639.
14. Gregg, S. and Sing, K., Adsorption, Surface Area, and
Porosity, New York: Academic, 1982.
15. Rozovskii, A.Ya., Bashkirov, A.N., Kagan, Yu.B., and
Pokrovskaya, E.G., Kinet. Katal., 1961, vol. 11, no. 6,
p. 830.
7. Anan’in, V.N., Belyaev, V.V., Romanenkov, V.E., and
Trokhimets,A.I., Vestsi Akad. Nauk BSSR (Khim.), 1989,
no. 5, p. 17.
16. Rozovskii, A.Ya., Kinetika topokhimicheskikh reaktsii
(The Kinetics of Topochemical Reactions), Moscow:
Nauka, 1980.
8. Lyashko, A.P., Medvinskii, A.A., Savel’ev, G.G.,
Il’in, A.P., and Yavorovskii, N.A., Kinet. Katal., 1990,
vol. 31, no. 4, p. 967.
9. Yakerson, V.I., Dykh, Zh.L., Subbotin, A.N., Gud-
kov, B.S., Chertkova, S.V., Radin, A.N., Boevskaya, E.A.,
Tertichnik, E.A., Golosman, E.Z., and Sarmurzina, R.G.,
Kinet. Katal., 1995, vol. 36, no. 6. p. 918.
10. Tikhov, S.F., Potapova, Yu.V., Sadykov, V.A., and
Fenelonov, V.B., React. Kinet. Catal. Lett., 2001, vol. 73,
no. 1, p. 55.
11. Tikhov, S.F., Fenelonov, V.B., Zaikovskii, V.I.,
Potapova, Yu.V., and Sadykov, V.A., Micropor. Mesopor.
Mater., 1999, vol. 33, p. 137.
17. Barre, P., Cinetique Heterogene, Paris: Gauthier-Villar,
1976.
18. Fenelonov, V.B., Kinet. Katal., 1994, vol. 35, no. 5,
p. 795.
19. Krivoruchko, O.P., Zolotovskii, B.P., Plyasova, L.M.,
Buyanov, R.A., and Zaikovskii, V.I., React. Kinet. Catal.
Lett., 1982, vol. 21, p. 103.
20. Buman, R.K., Mironovich, A.A., Chertov, V.M., Tsy-
rina, V.V., and Tsielen, U.A., Izv. Akad. Nauk Latv. SSR,
1988, no. 6, p. 691.
21. Zagrafskaya, R.V., Karnaukhov, A.P., and Fenelo-
nov, V.B., Kinet. Katal., 1976, vol. 17, no. 3, p. 130;
1979, vol. 20, no. 2, p. 465.
12. Tikhov, S.F., Salanov, A.N., Palesskaya, Yu.V.,
Sadykov, V.A., Kustova, G.N., Litvak, G.S., Rudina, N.A., 22. Fenelonov, V.B., Romannikov, V.N., and Derevyan-
Zaikovskii, V.A., and Tsybulya, S.V., React. Kinet.
Catal. Lett., 1998, vol. 64, no. 2, p. 301.
kin, A.Yu., Micropor. Mesopor. Mater., 1999, vol. 28,
p. 57.
KINETICS AND CATALYSIS Vol. 44 No. 2 2003