Corrosion of structural ceramics under sub-critical conditions in aqueous sodium chloride solution and in deionized water. Part II: Dissolution of Al 2O3-based ceramics

Article abstract of DOI:10.1111/j.1551-2916.2011.04513.x

Two types of alumina-based ceramics, pure solid-state sintered (SSS) alumina ceramics, and a liquid phase sintered (LPS) alumina were corrosion tested under both static and quasi-dynamic conditions in 0.5M NaCl solution and in deionized water reference medium at the temperature of 290°C. Static tests were also performed at 150° and 200°C in order to obtain the data for determination of kinetic parameters, and for calculation of the activation energies. The apparent activation energies of dissolution of the LPS alumina in deionized water and in 0.5M NaCl solution were identical and ranged around 49 kJ/mol. The SSS alumina ceramics corroded by grain-boundary attack and slow dissolution of alumina matrix grains, and the corrosion rates were negligible in both corrosion media. The LPS alumina, corroded by preferential attack and dissolution of calcium aluminosilicate grain-boundary glass. The alumina matrix remained relatively intact. The dissolution was markedly faster than in the SSS alumina. The rates of dissolution were found to be temperature dependent, but no influence of the corrosion medium was observed.

Full text of DOI:10.1111/j.1551-2916.2011.04513.x

J. Am. Ceram. Soc., 94 [9] 3044–3052 (2011)
DOI: 10.1111/j.1551-2916.2011.04513.x
r 2011 The American Ceramic Society
ournal
J
Corrosion of Structural Ceramics Under Sub-Critical Conditions in
Aqueous Sodium Chloride Solution and in Deionized Water. Part II:
Dissolution of Al O -Based Ceramics
2
3
w
Dagmar Galuskova
´
Vitrum Laugaricio – Joint Glass Center of the IIC SAS, TnU AD, FChFT STU and RONA j.s.c.,
TrenWı
´
n 911 50, Slovakia
ˇ
Miroslav Hnatko, Dugan Galusek,* and Pavol Sajgalı
´
k*
Institute of Inorganic Chemistry, Slovak Academy of Sciences, Bratislava 845 36, Slovakia
Two types of alumina-based ceramics, pure solid-state sintered
SSS) alumina ceramics, and a liquid phase sintered (LPS)
not been studied by any research group working in the area of
corrosion of structural ceramics.
(
alumina were corrosion tested under both static and quasi—
dynamic conditions in 0.5M NaCl solution and in deionized
water reference medium at the temperature of 2901C. Static tests
were also performed at 1501 and 2001C in order to obtain the
data for determination of kinetic parameters, and for calculation
of the activation energies. The apparent activation energies of
dissolution of the LPS alumina in deionized water and in 0.5M
NaCl solution were identical and ranged around 49 kJ/mol. The
SSS alumina ceramics corroded by grain-boundary attack and
slow dissolution of alumina matrix grains, and the corrosion
rates were negligible in both corrosion media. The LPS alumina,
corroded by preferential attack and dissolution of calcium
aluminosilicate grain-boundary glass. The alumina matrix
remained relatively intact. The dissolution was markedly faster
than in the SSS alumina. The rates of dissolution were found
to be temperature dependent, but no influence of the corrosion
medium was observed.
A related area, not covered in this study, is supercritical water
oxidation (SCWO): high-temperature and supercritical water
provide promising environment for various processes, e.g. chem-
ical reactions, salt separation processes, oxidation of organic
wastes, etc. Known for its extreme corrosiveness the SCWO
unite represent a significant challenge for material selection: no
material can withstand every conceivable acidic high-tempera-
ture solution. On the other hand, some materials possess a good
7
corrosion resistance in the absence of some species, thus giving
a chance for tailoring materials for specific needs.
Alumina is generally considered as an appropriate, and par-
ticularly from the economical point of view, suitable ceramics for
such oxidizing corrosion environments. To understand mecha-
nisms of degradation in more complex systems, model experi-
ments were carried out with simpler systems, where aggressive
acidic and caustic conditions were avoided at the very beginning
of the dissolution process. The corrosion experiments under static
and so called quasi-dynamic conditions were performed in deion-
ized water and in the 0.5M NaCl solution with neutral pH at
temperatures r2901C. For evaluation of the corrosion mecha-
nisms the emphasis was put on determination of the eluate chem-
istry combined with the chemical and phase analysis of formed
corrosion products. The effect of additives on corrosion resis-
tance of the liquid phase sintered (LPS) alumina was evaluated
and compared to high purity solid-state sintered (SSS) alumina.
I. Introduction
ORROSION of nonoxide ceramics in contact with aqueous
C
solutions under subcritical conditions was discussed in de-
tail in Part I of this paper. The chemical resistance of the ce-
ramics based on Si was found to be relatively low and not
3 4
N
1
sufficient for applied conditions and corrosion environment.
Generally, there exist significant differences in the mechanisms
of corrosion and the corrosion resistance of different groups of
ceramic materials (silicate, oxide, and nonoxide ceramic). The
chemical composition and microstructure are the key factors in
terms of their corrosion resistance.
II. Experimental Procedure
(1) Materials Characterization
Both tested aluminas were prepared from ultra fine (particle size
1
The papers dealing with corrosion resistance of alumina ce-
50 nm) and ultra pure (99.995 wt%) alumina powder Taimi-
2
,3
4,5
ramics report on the corrosion in aqueous acidic or caustic
solutions at high temperatures. The corrosive effect of the acids
on alumina ceramics decreases in the order H PO 4HCl4
SO , where the dominant corrosion mechanisms were iden-
tified as intergranular attack and dissolution of Al O . Whether
cron TM DAR (Taimei Ltd., Tokyo, Japan). The pure SSS al-
umina (material A) was prepared by sintering of slip cast green
bodies at 13501C for 1 h, achieving the relative density of 99.8%.
LPS-sintered alumina was prepared by hot pressing at 14501C
and 20 MPa of the same powder containing 5 wt% of
3
4
H
2
4
6
2
3
alumina could be prone to corrosion also in neutral aqueous
environments remains questionable, and to our knowledge has
CaO ꢀ 5SiO
2
sintering additives (material AH), achieving the rel-
ative density 99.2%.
The specimens for corrosion tests were prepared by the same
procedure as described for nonoxide ceramics in the Part I of
N. Jacobson—contributing editor
1
this paper. Rectangular bars 3 mm ꢁ 4 mm ꢁ 50 mm were cut,
ground, and the tensile faces polished up to 6 mm. Before im-
mersion in corrosive medium the samples were ultrasonically
cleaned in acetone for 15 min, rinsed with distilled water, and
put into a cabinet dryer for 3 h at 1101C. The stainless-steel
PTFE-lined pressure reactors were used for the corrosion tests
both under static and quasi-dynamic conditions.
Manuscript No. 28728. Received October 6, 2010; approved February 17, 2011.
This work was financially supported by the APVV grant No. APVV 0171-06 and by the
grant of the Slovak Scientific Grant Agency VEGA under the contract number VEGA
2/0036/10.
*
Member, The American Ceramic Society.
w
Author to whom correspondence should be addressed. e-mail: galuskova@tnuni.sk
3
044

September 2011
2) Corrosion Experiments
Corrosion of Alumina-Based Ceramics
3045
(
investigated with the use of X-ray diffractometer (XRD, Bruker
D8 DISCOVER, Bruker AXS Inc., Madison, WI) specially
designed for measurement of thin layers, in the 2y interval
In order to study the mechanisms of dissolution of the oxide
ceramic materials, the conditions of the corrosion tests
identical to those carried out with nonoxide ceramics were
201–801 using CuKa radiation.
1
employed. The bars were placed in PTFE-lined pressure
corrosion reactors with the inside volume 26 cm filled with
3
ꢂ4
corrosion liquid, deionized water (conductivity 2575 10 S/m),
III. Results
or 0.5 mol/L NaCl solution and heated in laboratory drying
oven. Both static and quasi-dynamic tests were carried out
at the temperature of 2901C with maximum duration of the
test 480 h. Additional static tests at temperatures 1501 and
(
1) Microstructure
The microstructure of the high purity SSS alumina, the material
A consists mostly of equiaxed grains with diameter of about 0.5
mm (Fig. 1(a)) and residual porosity of only 0.2%. Pores are
smaller and less frequent comparing to the microstructure of the
material AH (Fig. 1(b)). The microstructure of the material AH
mostly consists of the matrix of fine equiaxed alumina
grains with diameter o1 mm and with residual porosity 0.8%.
Occasionally, a large platelet-shaped grains with the length of
several tens of micrometers were present, which document
that abnormal grain growth was active under the conditions
of hot pressing. This is believed to be the result of local melt for-
mation due to inhomogeneous distribution of dopants/impurities
in the material. Detailed microstructure characterization of
the material AH published elsewhere revealed the presence of
approximately 0.5 nm thick amorphous grain-boundary film
with Ca:Si atomic ratio ranging between 1:11 and 1:25. The
composition of glass in triple pockets fell roughly into the anort-
hite and mullite phase fields and contained between 0–17 mol %
2001C were applied in order to obtain the data for determina-
tion of kinetic parameters, and for calculation of the activation
energies. The ratio between the sample surface and the
volume of corrosive liquid (S/V) was held constant at
ꢂ1
0
.770.03 cm , i.e. two bars were placed in one reactor at the
same time with 20 mL of the liquid medium. The tests were
carried out in three reactors in parallel. Six specimens were
tested at each condition. After the test the samples were removed
from the reactors, rinsed with distilled water, dried, and
weighed. In parallel, the tests without the ceramic specimens
were carried out for given time interval, providing the eluate
labelled as a blank. The concentration of the particular element
detected in the blank was subtracted from the concentration de-
termined in the eluates taken after the corrosion tests with the
ceramic material.
8
9
2 2 3
of CaO, 62–77 mol % of SiO , and 17–38 mol % of Al O .
Crystalline anorthite was detected as the main secondary crys-
talline phase.
(
3) Quasi-Dynamic Tests
Quasi-dynamic test conditions were also applied for determina-
tion of the kinetic parameters in case of the material AH.
As a standard flow-through arrangement could not be used
due to high temperatures applied, an alternative arrange-
ment has been used, where every 22 h the reactors were cooled
down, the corrosive solution was removed and fresh solution
was added to the reactor. The primary aim was to avoid
formation of a passivation layer due to oversaturation of the
solution with respect to leached components, or at least to min-
imize precipitation reactions. The test was carried out at the
temperature of 2901C. The S/V ratio was identical to the
static tests.
(
2) Static Experiments
(A) Corrosion in Deionized Water: The results of weight
change determination obtained from experiments carried out in
deionized water are at all studied temperatures and for both
studied materials summarized in the Fig. 2. At 1501 and 2001C
the weight losses observed in material A were similar at both
(4) Methods of Analysis
The amounts of ions dissolved in the corrosion medium were
determined by inductively coupled plasma atomic emission spec-
troscopy (ICP-AES, Vista MPX, Varian Australia Pty Ltd.,
Mulgrave, Australia). The concentration of Al, Ca, and Si in the
aqueous sodium chloride eluates was determined by the internal
standardization technique, with beryllium as the internal stan-
dard. The amount of an element released into the solution
was expressed in terms of the value NL in [g/m ] according
2
i
to the Eqs. 1 and 2 for static and quasi-dynamic conditions,
respectively
1
ci
NLi ¼
(1)
(2)
S
wi
wi
V
ci
t
i
þ NL ti ꢂDt
NL ¼
S
V
where c is the concentration of an element (i) in solution in [mg/
i
L], S is the surface of the ceramic material in contact with cor-
rosive solution in [m ], V is the volume of the corrosive solution
is the mass fraction of the element (i) in the ceramic
2
3
in [m ], w
i
material, t is the time and Dt is the length of the sampling in-
terval during the quasi-dynamic test.
Corroded surfaces and cross sections of corroded specimens
were examined by scanning electron microscopy (SEM
EDX, Carl Zeiss SMT, GmbH, Jena, Germany, EVO 40 HV,
accelerating beam voltage 20 kV). Before the SEM analysis
the specimens were coated with gold. The phase composition
of corrosion products formed at the exposed surface was
Fig. 1. Microstructure of the material A (a) and of the material AH (b).

3
046
Journal of the American Ceramic Society—Galuskova´ et al.
Vol. 94, No. 9
Fig. 3. NL(Al) as a function of time for A (a) and AH (b) specimens
corroded in deionized water.
Fig. 2. Weight loss as a function of time for the material A (a) and the
material AH (b) corroded in deionized water.
temperatures (Fig. 2(a)) during the first 192 h. A slight weight
gain was determined at 2001C for longer corrosion times. How-
ever, at 2901C notable weight gain was observed after 192 h of
corrosion. The weight loss of corroded AH (LPS-alumina) spec-
imens was temperature dependent and linear in the time interval
studied. The maximum measured weight loss in AH was five
times higher than in the material A (Fig. 2(b)).
The analysis of deionized water after corrosion experiments
with the material A were focused on determination of aluminum
2 3
as the main building element of the Al O matrix. The concen-
trations were recalculated to the NL values (Fig. 3(a)) according
to Eq. (1) and compared with the NL(Al) values obtained for the
AH specimens (Fig. 3(b)). The NL(Al) values in the system A
were extremely low, and indicate low dissolution rate of alumina
matrix. At 1501C the NL(Al) values were close to the limit of
detection (0.095 mg/L) of the used analytical method for alu-
minum, especially for short dissolution times, and are therefore
not shown in the figure. The lowest but still detectable NL(Al)
were measured at 2901C, which corresponds to observed weight
gains shown in (Fig. 2(a)), and indicates precipitation of hy-
drated aluminas. The NL(Al) values in the system AH were
systematically by one order of magnitude higher than in the
system A. After initial enrichment with Al within the first 48 h of
the test at 2901C a steady state was achieved.
At lower temperatures the time dependence of NL(Al) goes
through a maximum between 48 and 96 h of the test, after which
the solution was depleted of aluminum. When evaluating the NL
values in AH, the presence of the intergranular calcium alum-
inosilicate glassy phase must be also considered. The exact
amount of aluminum incorporated into the glass containing
neighboring alumina grains. The calculated NL values are thus
normalized with respect to overall Al in the ceramic material,
including both the alumina in the matrix grains and the Al in
grain-boundary phase. Considering the extremely low dissolu-
tion rate of alumina matrix in the material A, the contribution
of aluminum from the matrix in the eluate is also likely to be
negligible in the material AH. The main source of Al is therefore
the aluminosilicate grain-boundary phase with low calcia-to-sil-
ica ratio and unknown aluminum content, and the triple pocket
glass of approximately anorthite composition. The NL(Al) val-
ues normalized with respect to the overall Al content in the
materials thus provides only strongly biased information on
dissolution rate of AH. The analysis of corrosion solutions was
therefore focused on determination of calcium and silicon
(Fig. 4). Nearly identical NL(Si) and NL(Ca) values were mea-
sured at 1501C (Fig. 4(a)), indicating congruent dissolution of
grain-boundary glass. Similar congruent and linear behavior
was observed also at 2001C in the initial 96 h of the test, after
which the solution became depleted of calcium. No calcium was
detected in the eluates from the corrosion experiments at 2901C
up to 288 h (Fig. 4(b)), despite the fact that the NL(Si) values
show approximately linear trend until steady state is achieved
after 192 h of the test.
Dissolved aluminum has a distinct tendency to form hydroxy
31
complexes. The activity of Al ions in water is strongly influ-
enced by pH. Below pH 5 4.5 the aluminum solubility increases,
but in the near-neutral pH range, the total concentration of dis-
31
10
solved Al becomes as low as 1 mmol/L. The pH values of the
water leaching medium after the corrosion experiments carried
out at 2901C was in the range from 6.1 to 7.1 (measured at
251C). Based on this information we can assume that under
these conditions all dissolved aluminum was present either in the
2
nominally CaO:SiO in molar ratio 1:5 is not known, and can-
not be determined due to overlapping of the electron beam with

September 2011
Corrosion of Alumina-Based Ceramics
3047
Fig. 5. The morphology of the specimen A corroded in deionized water
for 288 h at 2901C (a) The facetted grains represent precipitated corro-
sion products. Detailed scan of the circular area in the right hand side of
the micrograph (a) revealing partially dissolved a-alumina grains (b).
Fig. 4. NL(Si) and NL(Ca) as a function of time for the specimen AH
corroded in deionized water at 1501, 2001C (a), and at 2901C (b).
48 h due to dissolution of grain-boundary glass and continue to
grow in the next 480 h. At the neutral pH aluminum is likely to
0
3
ꢂ 11,12
ꢂ
form of Al(OH) or as [Al(OH) ] ,
4
until the equilibrium was
be present in the soluble form of [Al(OH) ] as the consequence
4
3
1
ꢂ
reached. The initial precipitates were likely amorphous hydrox-
ides that could be either aged under hydrothermal conditions of
the test or formed during cooling of the autoclave, to yield crys-
talline corrosion products. The SEM analysis of the sample A
surface corroded in deionized water at 2901C for 288 h revealed
the presence of significant amount of corrosion products with
well-defined facets indicating their crystalline nature (Fig. 5(a)).
The XRD pattern of the precipitates removed from corroded
surfaces confirmed the presence of AlOOH phase in the form of
boehmite (Table I). A more detailed view of the area without
precipitates (Fig. 5(b)) revealed partially dissolved rounded
grains of a-Al O and the places where loosened corroded grains
2 3
detached from the alumina matrix as the result of the corrosion
process.
Dissolution of the material AH and the chemistry of the cor-
rosion solution in the initial stage of dissolution might be influ-
enced by the concentration of released Al ions. The results from
this study at 2901C show that NL(Si) increase rapidly in the first
of bonding the Al ions with OH in later stages of dissolution
when the steady state is achieved. The normalized amount of
aluminum released into the solution after 24 h was 36 mg/m
2
2
and after 480 h of dissolution 300 mg/m (Fig. 3(b)). The pres-
31
41
21
ence of Al and Si together with Ca ions in the solution
gives rise to a number of identifiable soluble/insoluble com-
pounds. Moreover, the solution is evidently depleted of calcium.
The depletion is observed after 96 h of the test at 2001C and
from the beginning of the dissolution process at 2901C. The
morphology of corroded surface of the specimen AH after only
8 h at 2901C (Fig. 6) revealed needle-shaped corrosion products,
which differ markedly from the angular crystals found on cor-
roded surface of the material A. The XRD of the corrosion
products removed from the corroded surface after 480 h
at 2901C confirmed the presence of crystalline phases of the
anorthite type (anorthite and dmisteinbergerite) with the molec-
ular formula CaO ꢀ Al
2
O
3
ꢀ 2SiO
2
(Table I), where Ca, Si
incorporated in the mineral structure originated from the
Table I. The Crystalline Phases Identified by the X-Ray Diffraction of the Corrosion Products Removed from the Corroded Surface
of the Ceramic Material after 480 h at 2901C
X-ray analysis of the corrosion products after corrosion of the material in
Specimens
Deionized water
0.5 mol/L NaCl
CP-A
Boehmite/AlOOH (83-2384)
Corundum/Al O (71-1123)
2
3
CP-AH
Anorthite/CaO–Al O ꢀ 2SiO (89-1459)
Analcime/Na(Si Al)O (74-2219)
2
2
3
2
6
Dmisteinbergerite/(CaAl Si )O (51-64)
2
2
8
Corundum/Al O (71-1123)
2
3
CP, corrosion products removed from the material.

3
048
Journal of the American Ceramic Society—Galuskova´ et al.
Vol. 94, No. 9
Fig. 6. SEM micrograph of the AH surface exposed to deionized water
for 8 h at 2901C.
grain-boundary glass together with Al of, very likely, the same
origin. The corundum phase identified in the corrosion product
originates from the uncorroded material removed together with
corrosion products during mechanical separation of the layer.
(B) Corrosion in an Aqueous Sodium Chloride Solution:
Unlike for dissolution in deionized water no weight-gain was
observed in the material A corroded in 0.5M NaCl solution at
2
901C (Fig. 7(a)). The time–weight loss dependences differed
only slightly, irrespective of the temperature. The time–weight
loss dependences of the material AH showed similar linear
trends at 1501 and 2001C. However, at 2901C the weight de-
creased significantly, similar to deionized water, and approxi-
mately linearly with time up to 192 h of the test (Fig. 7(b)). The
Fig. 8. NL(Al) as a function of time for A (a) and AH (b) specimens
corroded in 0.5M NaCl solution.
NL(Al) values in the aqueous sodium chloride solution for the
material A were very low up to 2001C, but at the temperature
2
(
901C significant increase of the NL(Al) with time was observed
Fig. 8(a)).
In the material AH the aluminum content increased rapidly,
and approximately linearly, in the initial period of the test, going
2
through a maximum of 230 mg/m at about 48 h, and followed
by possible formation of colloid Al(OH) resulting to depletion
3
2
of Al to 30 mg/m with extension of the test over 288 h
(
Fig. 8(b)). The initial aqueous aluminum concentration, to-
1
gether with the presence of Na ions in the NaCl solution are
likely to control the Si and Ca release, as well as precipitation of
Si and Ca-containing corrosion products in both aqueous solu-
tions. Calcium precipitated in deionized water eluates, while the
content of silicon increased linearly. In the NaCl solution Ca
remained in soluble form even during the corrosion at 2901C
(
Fig. 9), while solution became gradually depleted of silicon in
later stages of the material dissolution. In both corrosion solu-
tions the concentration of Si achieved the value 200750 mg/L at
2901C, which might indicate the equilibrium concentration of
silicon under the applied conditions.
The corroded surfaces of the material AH were coated with a
compact crust of precipitates with scattered globular clusters
with diameters ranging from 10 to 200 mm (Fig. 10). The EDX
analysis confirmed the presence of Si, Al, and Na in the clusters.
The crust itself consisted of fine star-like dendritic crystallites
with the length of individual needles in the range from 1 to 2 mm
(
Fig. 10(b)). Detailed XRD analysis of the globular clusters re-
moved from the precipitated surface confirmed the presence of
analcime (Na(Si Al)O , PDF 74-2219) crystalline phase (Table
2
6
I), in accord with the measured concentrations of leached ele-
ments, which indicated precipitation of a silicate, calcium-free
phases from the NaCl solution after 196 h testing at 2901C.
Fig. 7. Weight loss as a function of time for the material A (a) and AH
(
b) corroded in 0.5M NaCl solution.

September 2011
Corrosion of Alumina-Based Ceramics
3049
Fig. 10. SEM micrograph of the globular cluster (a) and star-like
dendritic crystallites (b) on the surface of specimen AH corroded for
2
88 h in 0.5M NaCl at 2901C.
under static conditions, i.e. anorthite in deionized water and
analcime in aqueous sodium chloride solution. Interestingly, the
precipitates did not seem to cover the prismatic planes of abnor-
mally grown alumina grains, which then remain exposed to cor-
rosion solution (Figs. 12(a) and (b)). This effect might be related
to different quality of corroded surfaces: the abnormal grains
remain largely smooth and unaffected by the corrosion, prevent-
ing the formation of new phase nuclei. At the same time the
boundaries in fine-grained matrix are etched, creating uneven
surface facilitating nucleation and precipitation of new phases.
Apart from the precipitated layer of corrosion products the
EDX analysis revealed an approximately 400 mm thick surface
layer significantly depleted of calcium and silicon in specimens
corroded in both media. The layer could be clearly distinguished
also on SEM micrographs appearing brighter comparing to dark
uncorroded core, and according to EDX analysis the content of
both Ca and Si was close to detection limit of the method. The
layer is characteristic also by the presence of clearly distinguish-
able abnormal alumina grains, which are etched from the sur-
rounding matrix. The Ca and Si contents determined by EDX in
the uncorroded area beyond the depleted layer were, respectively,
0.46 and 1.03 at.%.
(D) Corrosion Rates and Activation Energies: As shown
by the results of the static tests, dissolution of the specimen A is
influenced by the temperature. Moreover, aluminum released
from the studied material has strong tendency to create complex
ions or to be incorporated into the corrosion products attached
to the specimen surfaces, especially those corroded at 2901C.
The rate constants of dissolution at 1501C for the material A
could not be determined as the concentrations of aluminum in
eluates were close to the limit of detection. Also determination
of the activation energies of dissolution of the material A from
only two allowable kinetic parameters would be rather ques-
tionable, and are therefore not considered.
Fig. 9. NL(Si) and NL(Ca) as a function of time for AH specimens
corroded in 0.5M NaCl at 1501, 2001C (a), and at 2901C (b).
(
C) Quasi-Dynamic Conditions: Both corrosion solu-
tions under static conditions at 2901C became enriched with
the metal ions transferred from the dissolved material until ther-
modynamically stable reaction products were formed to hinder
further dissolution process. In order to obtain data for deter-
mination of the initial dissolution rate of the material AH,
quasi-dynamic arrangement of the experiment was used. The
cumulative NL values of the leached aluminum, calcium, and
1
silicon calculated according to the Eq. (2) and their time de-
pendences in deionized water and in the 0.5 mol/L NaCl solu-
tion are, respectively, shown in Figs. 11(a) and (b). The
morphology of the corroded surfaces from both corrosion
solutions are shown in Fig. 12.
The total concentration of calcium in deionized water eluates
after subtraction of the blank experiment was close to zero. The
concentration of Al in the eluate after each 22 h period did not
exceed the average value of 3 mg/L in an early stage of the ex-
periment, then decreased to 1 mg/L. Such decrease is explained
either by the hindrance of the leaching process by the layer of
precipitates or by gradual exhaustion of the surface corrosion
layer with respect to the leached ions with extension of the test.
2
Nearly identical maximal cumulative NL(Al) value, 400 mg/m
was measured in both corrosion solutions. The total weight
losses measured after the tests in deionized water and in the
NaCl solution were respectively 0.002 (RSD 8%) and 0.004
2
g/cm (RSD 29%), although these values might be influenced by
the formation of new secondary phases attached to the surface
of specimens. The morphology of new reaction products dif-
fered, depending on the solution. Sporadically aggregated nee-
dle-shaped crystals were observed after the tests in deionized
water and the lamellas of various orientation covered the ceramic
surface in contact with the NaCl solution (Fig. 12). The phase
composition of the precipitates was identical to those formed
Dissolution processes of the material AH were evidently in-
fluenced by the temperature of the corrosion media. The rate

3
050
Journal of the American Ceramic Society—Galuskova´ et al.
Vol. 94, No. 9
Fig. 12. SEM micrograph of the specimen AH analyzed after corrosion
in deionized water (a) and in aqueous sodium chloride solution (b) under
quasi-dynamic conditions at 2901C. The exposed prismatic planes are
marked with arrows.
Fig. 11. Time dependences of NL(Al, Ca, Si) leached from the specimen
AH in deionized water (a) and in aqueous sodium chloride solution (b).
6
of AlO octahedra (102 kJ/mol). Therefore, the mechanism of dis-
solution of silica-doped aluminas in HF is the same that of the
fused silica, while the mechanism of dissolution MgO codoped
materials with higher than equimolar Mg/Si ratio is similar to pure
sapphire. The values of activation energies obtained in this study
for the material AH with Ca/Si ratio 0.2 therefore confirm the re-
sults of chemical analysis of eluates, i.e. that the material corrodes
by preferential dissolution of the aluminosilicate grain-boundary
phase, or, more specifically by congruent dissolution of the
constants of dissolution at the two temperatures lower than
2
901C and also at 2901C were determined from the linear parts
2
of the experimental NL (mol/m ) versus time dependences ob-
tained from the analysis of the data acquired, respectively, from
static and quasi-dynamic tests. Additionally, calculations of the
rate constants were based on the element which was not incor-
porated to the secondary phases formed due to oversaturation
of the solution, i.e. Si and Ca in deionized water (Figs. 4(a) and
21
aluminosilicate network with simultaneous release of Ca ions.
11(a)) and NaCl solution (Figs. 9(a) and 11(b)), respectively.
Experimentally, the activation energies were obtained from the
slope of a plot lnk vs. 1/T using the Arrhenius equation
IV. Discussion
ln k ¼ ln A ꢂ Ea=ðRTÞ
(3)
It is generally accepted that the corrosion resistance of alumina
Even for high-
2
,3,13,14
ceramics is closely related to their purity.
purity alumina, attack can occur at grain boundaries where
small amounts of silica-based impurities segregate and are pref-
erentially dissolved by aqueous fluids. a-alumina crystals have
dense structure where aluminum-oxygen octahedra form an in-
terpenetrating lattice with strong interatomic bonds, which
makes the corrosion attack particularly difficult. Detailed in-
sight in SEM micrographs of the material A corroded at 2901C
(Fig. 5(b)), revealed that all Al O matrix grains at the surface
The values of rate constants and the activation energies for
both studied ceramic materials are summarized in Table II.
The rate constants of dissolution of the material A were al-
most three orders of magnitude lower than the rate constants of
the material AH. The apparent activation energy of dissolution
of the AH ceramic material in deionized water and in 0.5M
NaCl solution was 48.370.6 and 49.070.1 kJ/mol, respectively.
The value of the activation energies reported for dissolution of
alumina ceramics with purity 99.8% corroded by various acids
was determined to be 10.68 kJ/mol in HCl and 9.74 kJ/mol in
2
3
had round edges and in some places the grains have been de-
tached from the material. It cannot be excluded that some of
Al O grains were detached due to preferential decomposition of
2 3
6
3
H SO . Mikeska and Bennison determined the activation en-
2
small amounts of Si-impurities segregated at the grain bound-
aries. Dissolution of grains themselves could be assigned to pos-
sible metal-proton exchange reactions, the reversible exchange
4
2
ergies for alumina samples codoped with MgO and SiO dis-
solved in aqueous hydrofluoric acid in the range of 30–90 kJ/mol
depending on composition of the grain-boundary glassy phase.
They conclude that low activation energies, at the level of 30 kJ/
ꢂ
ꢂ
of OH for Al, yielding [Al(OH) ] . The increase of temperature
4
up to 2901C greatly influences speciation of Al in solution, re-
sulting in precipitation of new secondary phases. Alumina has
mol in silica-doped aluminas is close to that of dissolution of SiO
tetrahedra, while high activation energies characteristic for al-
4
5
been shown by Sato et al. to dissolve in aqueous solution at
uminas with Mg/Si ratio ꢃ 1 are similar to those of dissolution
1501–2001C containing NaOH by the following reactions:

September 2011
Corrosion of Alumina-Based Ceramics
3051
Table II. The Activation Energies of Dissolution of the Material AH, and Rate Constants Measured for Corrosion in Deionized
Water and in 0.5M NaCl Solution
Deionized water
0.5 mol/L NaCl
2
ꢂ1
2
ꢂ1
Specimen
k mmol(m ꢀ h)
E [kJ/mol]
k mmol(m ꢀ h)
E [kJ/mol]
Material A
2901C
2001C
1501C
0.002170.0001
0.00970.001
—
—w
0.01470.001
0.002670.0003
—
—w
Material AH
2
2
1
901C
001C
501C
5071
5.470.4
1.770.1
48.370.6
6772
9.773
2.170.3
49.070.1
wNot determined.
2
ꢂ
3ꢂ
4ꢂ
ꢂ
ꢂ
pH, Si(OH)
4
further dissociates into H SiO
2
4
/HSiO
4
/SiO
4
Al2O3 þ OH þ 2H2O ! AlðOHÞ þ AlOOH
(4)
(5)
4
1 18,19
ꢂ
and produces more H . In the presence of [Al(OH)
Ca , and Si(OH) various complexes such as Al(OH) H SiO
can be formed along with other corrosion products, e.g. anort-
hite (CaOꢀ Al ꢀ 2SiO ) confirmed by XRD at the surface of
AH corroded in deionized water. In the contrary, in aqueous
4
] ,
ꢂ
ꢂ
21
ꢂ
AlOOH þ OH þ H2O ! AlðOHÞ
4
4
3
3
4
with the second reaction being the faster of the two. According
to phase distribution diagram, the principal Al specie in the
presence of OH is the aluminate ion [Al(OH) ] . As our experi-
O
2 3
2
ꢂ
ꢂ 12
1
environment with Na ions, analcime (Na(Si Al)O ) was iden-
2 6
4
ments were carried out in the solutions with neutral pH,
amorphous Al(OH) precipitates can also form along with the
3
tified as the main reaction product (Table I), while calcium re-
mained in a soluble form. Calcium together with silicon were not
detected with EDX analysis in the corrosion affected, approxi-
mately 400 mm thick, layer. Similar dissolution rate constants
were acquired for deionized water and the NaCl solution.
aluminate ions: these can be hydrothermally aged to yield crys-
talline products. Speaking in terms of Al speciation, the pH
value 6.970.3 determined in deionized water after the corrosion
tests falls into the neutral solution range. The dissolution equa-
tion (Eq. (6)) is then proposed to proceed as follows:
V. Conclusion
ꢂ
3þ
2
Al2O3 þ 6H2O ! 3AlðOHÞ þ Al
(6)
4
(
1) Static Conditions
31
Precipitation of Al ions in the form of boehmite (AlOOH),
as the new secondary phase covering the surface of the specimen
A corroded in deionized water, was detected by various analyt-
ical techniques. Summarizing the weight loss measurements and
the solution chemistry, the corrosion of the specimen A in both
corrosion solutions comprised the mentioned dissolution pro-
cess with partial precipitation of aluminum hydroxid-oxides.
However, the process was rather slow when comparing to dis-
solution of the material AH.
Dissolution in both corrosion media of the material A was
found to be an extremely slow, temperature-dependent process.
Partial dissolution of alumina matrix grains was documented by
rounding of the surface grain edges. Loosening and detachment
of some matrix grains explained by grain-boundary attack and
possible dissolution of small amounts of silicon-containing grain
boundary segregated impurities is considered as the principal
mechanism of material degradation. Hydroxyaluminates, iden-
tified by XRD as boehmite, precipitate to form the discontinu-
ous layer at the surface of the specimens corroded at 2901C in
deionized water.
The hypothesis that corrosion of the AH ceramics proceeds
via dissolution of the grain-boundary glassy phase is supported
by the eluate chemistry together with the analysis of corroded
surfaces and newly formed secondary phases. The mechanism of
dissolution reported for aluminosilicate oxides and glasses can
be therefore applied, as the presence of calcium aluminosilicate
Corrosion of the material AH is by three orders of magnitude
faster than that of the material A, and attacks preferentially the
calcium alumino-silicate grain-boundary glass. The initial dis-
2
ꢂ1
9
solution rates of 50 and 67 mmol(m ꢀ h) were determined at
the temperature 2901C in deionized water and in the aqueous
sodium chloride solution, respectively. The apparent activation
energies 48.370.6 kJ/mol (deionized water) and 49.070.1 kJ/
mol (NaCl solution) suggest metal-proton exchange reactions
glass phase was confirmed at the grain boundaries. Jantzen
1
5
et al. treats dissolution of such glasses through activated com-
plex as a rapid removal of univalent and divalent cations from
the near surface by ion exchange followed by the slower ex-
31
1
change of Al for 3H protons in solution. Divalent metal–
oxygen bonds break faster than trivalent metal–oxygen bonds,
4
and dissolution of SiO tetrahedra as the rate-controlling disso-
1
6
31
lution mechanism. In deionized water saturation of the corro-
sion solution was observed, accompanied by precipitation of
anorthite-type compounds, and by significant depletion of the
corrosion solution of Ca. In the aqueous sodium chloride solu-
tion analcime, Na(Si Al)O , was detected by XRD as the main
2 6
precipitated phase after 480 h of corrosion. Precipitation of
analcime was accompanied by depletion of the eluate of Si.
which break faster than Si–O bonds. The Al —proton ex-
change does not destroy the glass framework but it partially
liberates three adjoining silica tetrahedra to which it is bonded.
It is the detachment of the partially liberated silica that is the
rate-determining step, that is, partially detached silica dissolves
more readily than bonded or ‘‘attached’’ tetrahedral silica.
After Ca, Al, and Si are released from the grain-boundary
glass of the attacked ceramic material, various reactions can oc-
cur due to the complex solution chemistry. The hydrolysis of
Al(III) is characterized by a series of mononuclear as well as
(
2) Quasi-Dynamic Conditions
The quasi-dynamic conditions applied during the tests were not
able to prevent completely the attainment of the equilibrium
resulting in precipitation of secondary phases and partially cov-
ering the surfaces of corroded specimens. Needle-like precipi-
tates of anorthite assembled to spherical aggregates were found
at surfaces corroded in deionized water and the lamellas aggre-
gated to globular clusters of analcime were observed at the sur-
faces of the material AH exposed to the aqueous sodium
3
1
polynuclear species. It has been shown that [Al(OH
2
)
6
]
under-
21
goes successive deprotonation steps to yield [Al(OH)]
,
Al(OH) ] , and Al(OH)3(aq). In neutral solution and upon in-
1
17
[
creasing pH the amorphous Al(OH)
2
dissolves and the alum-
(aq)
3
ꢂ
17
inate ion, [Al(OH)
mostly Si(OH) . It is a weak acid, of which the first degree dis-
sociation produces H SiO and H at a pH o9, and at higher
4
]
is formed. Dissolved Si(IV) specie is
4
ꢂ
4
1
3

3
052
Journal of the American Ceramic Society—Galuskova´ et al.
Vol. 94, No. 9
7
P. Kritzer, N. Boukis, and E. Dinjus, ‘‘Factors Controlling Corrosion in
Contribution to the Dissociation
and Stability Data Influencing Corrosion Processes,’’ J. Supercrit. Fluids, 15,
chloride solution. Irrespective to the corrosion solution used no
precipitation was observed at prismatic plates of abnormally
grown alumina grains, which remained exposed to corrosion
media. The surface leached layer with the thickness up to
High-Temperature Aqueous Solutions:
A
2
05–27 (1999).
I. J. Bae and S. Baik, ‘‘Abnormal Grain Growth in Alumina,’’ J. Am. Ceram.
8
Soc., 80, 1149–56 (1997).
ˇ
400 mm was nearly depleted of Ca and Si, as confirmed by the
9
P. SvanWarek, D. Galusek, F. Loughran, A. Brown, R. Brydson, A. Atkinson,
´
EDX analysing.
and F. Riley, ‘‘Microstructure-Stress Relationships in Liquid Phase Sintered
Alumina Modified by the Addition of 5 Weight % of Calcia–Silica Additives,’’
Acta Mater., 54, 4853–63 (2006).
Acknowledgments
10
C. A. J. Appelo and D. Postma, Geochemistry, Groundwater and Pollution, 2nd
edition, pp. 381–2. A.A. Balkema Publishers, Leiden, 2005.
S. Castet, J. L. Dandurand, J. Schott, and R. Gout, ‘‘Boehmite Solubility and
This publication was created in the frame of the project ‘‘Centre of excellence
for ceramics, glass, and silicate materials’’ ITMS code 262 201 20056, based on the
Operational Program Research and Development funded from the European
Fund of Regional Development. The authors also wish to express their thanks to
Dr. Robert Klement for helpful discussions.
11
Aqueous Aluminum Speciation in Hydrothermal Solutions (90–3501C): Experi-
mental Study and Modeling,’’ Geochim. Cosmochim. Acta, 57, 4869–84 (1993).
N. Moller, Ch. Christov, and J. Weare, ‘‘Thermodynamic Models of Alumi-
12
num Silicate Mineral Solubility for Application to Enhanced Geothermal Sys-
tems’’, Proceedings: Thirsty-First Workshop on Geothermal Reservoir Engineering,
SGP-TR-179, Stanford University, Stanford, CA, January 30–February 1, 2006.
References
13
K. Oda and T. Yoshio, ‘‘Hydrothermal Corrosion of Alumina Ceramics,’’
J. Am. Ceram. Soc., 80 [12] 3233–6 (1997).
1
ˇ
D. Galuskova, M. Hnatko, D. Galusek, and P. Sajgalı
tural Ceramics under Sub-critical Conditions in Aqueous Sodium Chloride Solu-
tion: Dissolution of Si -based Ceramics Part I,’’ J. Am. Ceram. Soc. (in press)
2011).
´
k, ‘‘Corrosion of Struc-
14
N. Boukis, N. Claussen, K. Ebert, R. Janssen, and M. Schacht, ‘‘Corrosion
Screening Tests of High-Performance Ceramics in Supercritical Water Containing
3 4
N
Oxygen and Hydrochloric Acid,’’ J. Eur. Ceram. Soc., 17, 71–6 (1996).
C. M. Jantzen, K. G. Brown, and J. B. Pickett, ‘‘Durable Glass for Thousands
(
15
2
M. Schacht, N. Boukis, and E. Dinjus, ‘‘Corrosion of Alumina Ceramics in
Acidic Aqueous Solutions at High Temperatures and Pressures,’’ J. Mater. Sci.,
of Years,’’ Inter. J. Appl. Glass Sci., 1 [1] 38–62 (2010).
J. Schott, O. S. Pokrovsky, and E. H. Oelkers, ‘‘The Link Between Mineral
16
3
5, 6251–8 (2000).
K. R. Mikeska and S. J. Bennison, ‘‘Corrosion of Alumina in Aqueous
3
Dissolution/Precipitation Kinetics and Solution Chemistry,’’ Rev. Mineral. Geo-
chem., 70, 207–58 (2009).
Hydrofluroric Acid,’’ J. Am. Ceram. Soc., 82 [12] 3561–6 (1999).
W. Genthe and H. Hausner, ‘‘Influence of Chemical Composition on the
17
4
L. Spadini, P. W. Schindler, and S. Sjo
¨
berg, ‘‘On the Stability of the
21
AlOSi(OH)
3
Complex in Aqueous Solution,’’ Aquat. Geochem., 11, 21–31 (2005).
R. K. Iler, ‘‘Effect of Adsorbed Alumina on Solubility of Amorphous Silica In
Corrosion of Alumina in Acids and Caustic Solutions,’’ J. Eur. Ceram. Soc., 9,
4
18
17–25 (1992).
T. Sato, S. Sato, and A. Okuwaki, ‘‘Corrosion Behavior of Alumina Ceramics
5
Water,’’ J. Colloid Inter. Sci., 43, 2399–408 (1973).
F. J. Hingston and M. Raupach, ‘‘Reaction Between Monosilicic Acid and
19
in Caustic Alkaline Solutions at High Temperatures,’’ J. Am. Ceram. Soc., 74 [12]
3
Aluminium Hydroxide. I. Kinetics of Adsorption of Silicic Acid by Aluminium
Hydroxide,’’ Aust. J. Soil Res., 5 [2] 295–309 (1967).
081–4 (1991).
6
&
L. Curkovic, M. F. JelaWa, and S. Kurajica, ‘‘Corrosion Behaviour of Alumina
2 4
Ceramics in Aqueous HCl and H SO Solutions,’’ Corr. Sci., 50, 872–8 (2008).