Corrosion of structural ceramics under subcritical conditions in aqueous sodium chloride solution and in deionized water. Part I: Dissolution of Si 3N4-based ceramics

Article abstract of DOI:10.1111/j.1551-2916.2011.04553.x

The results presented in this paper compare the behavior of silicon nitride prepared with the addition of yttria as sintering additive, and of a sialon ceramic in contact with an aqueous solution of sodium chloride (NaCl) at TaC. Both studied ceramics dissolved by preferential attack of Si-N bonds in the matrix accompanied by the release of ammonia and formation of protective, mostly oxide layer of corrosion products at the surface. Increase of dissolution rate of sialon in aqueous NaCl solution at 290°C was observed in comparison to dissolution in deionized water, achieving the value 27 mmol·(m ²·h)^-1. Penetration of chloride anions through protective layer of corrosion products is proposed to contribute to destruction of the passivation layer and renewal of the exposure of vulnerable material surface to corrosion medium.

Full text of DOI:10.1111/j.1551-2916.2011.04553.x

J. Am. Ceram. Soc., 94 [9] 3035–3043 (2011)
DOI: 10.1111/j.1551-2916.2011.04553.x
r 2011 The American Ceramic Society
ournal
J
Corrosion of Structural Ceramics Under Subcritical Conditions in
Aqueous Sodium Chloride Solution and in Deionized Water. Part I:
Dissolution of Si N -Based Ceramics
3
4
w
Dagmar Galuskova
´
Vitrum Laugaricio – Joint Glass Center of the IIC SAS, TnU AD, FChFT STU, and RONA, j.s.c., 911 50 TrenWı
´
n, Slovakia
ˇ
Miroslav Hnatko, Dugan Galusek,* and Pavol Sajgalı
´
k*
Institute of Inorganic Chemistry, Slovak Academy of Sciences, 845 36 Bratislava, Slovakia
1
–4
The results presented in this paper compare the behavior of sil-
icon nitride prepared with the addition of yttria as sintering ad-
ditive, and of a sialon ceramic in contact with an aqueous
solution of sodium chloride (NaCl) at Tr2901C. Both studied
ceramics dissolved by preferential attack of Si–N bonds in the
matrix accompanied by the release of ammonia and formation
of protective, mostly oxide layer of corrosion products at the
surface. Increase of dissolution rate of sialon in aqueous NaCl
solution at 2901C was observed in comparison to dissolution in
deionized water, achieving the value 27 mmol (m h) . Pen-
etration of chloride anions through protective layer of corrosion
products is proposed to contribute to destruction of the passi-
vation layer and renewal of the exposure of vulnerable material
surface to corrosion medium.
and on the composition of the grain-boundary phase. Nickel
and Seipel studied the corrosion behavior of silicon nitride
5
3 4
(Si N ) ceramics in hot aqueous fluids, e.g. sulphuric acid. All
methods he applied for monitoring of corrosion indicate partial
or complete removal of the grain-boundary phase to be the main
corrosion mechanism. They conclude deliberately used additives
(e.g., sintering aids) or impurities makes advanced ceramics vul-
6
nerable to corrosion by hot aqueous fluids. Jacobson studied
corrosion of silicon (Si)-based ceramics in combustion environ-
ment. Five main types of corrosive degradation have been iden-
tified: passive oxidation, deposit-induced corrosion, active
oxidation, scale/substrate interactions, and scale volatility. Ad-
ditionally, nonoxide ceramics were discussed as promising ma-
terials for reactors for supercritical water oxidation units
2
ꢀ1
.
.
7
–11
(
SCWO reactors).
Surprisingly, under these conditions the highest corrosion
rates were not found at the highest temperatures, where the
density of corrosion media (water vapour) was low, but at sub-
critical temperatures o4701C with water of high density (above
I. Introduction
OME particularly advantageous, especially mechanic, proper-
ties predestine nonoxide ceramics for various applications,
3
12
S
200–300 kg/m ). At these temperatures, the concentrations of
1
ꢀ
such as reciprocating engine components, ball bearings, cutting
tools, nonautomotive wear components, and gas turbine en-
gines. However, high fracture strength, toughness, light weight,
or some other mechanical properties are sometimes insufficient,
if an excellent resistance to attack by the environment plays a
key role and may, in some cases, be the prime reason for the
selection of a particular material or a limiting factor of its use.
Practically all nonoxide ceramics are in nonequilibrium state,
thus reacting with environment in almost every application. In
spite of this, they still remain valuable and demanding materials
from the point of view of their corrosion resistance.
Chemistry and microstructure of a particular material as well
as the variability of actual chemical environment are crucial for
the evaluation of corrosion behaviour. Detailed understanding
of corrosion mechanisms and kinetics of the processes is inev-
itable for prevention of active degradation of material, which
results in decrease of its lifetime performance.
both H and OH are about three orders of magnitude above
their values in ambient water, so that water can be considered to
be both acidic and alkaline. It was reported that the solubility of
inorganic compounds under these conditions is high (subcritical
like): practically all nonoxide ceramics are converted into their
1
3
oxides or dissolve rapidly.
The results presented in this paper compare the behavior of
prepared with the addition of yttria as sintering additive,
3 4
Si N
and of a sialon ceramic in contact with an aqueous solution of
sodium chloride (NaCl) under conditions close to, or within the
1
4
subcritical region (1501–3001C). NaCl is a common by product
of the treatment of chlorinated wastes in SCWO process units
and is known as an aggressive corrosive agent of car parts due to
winter road maintenance, or in pumps for hot high-salinity so-
lutions used in units for desalination of seawater. The number of
prospective applications of water-based fluids is not final, and
has been appreciably growing, because supercritical water as the
1
5
Recent studies on the corrosion behavior of nonoxide ceramic
materials in acidic and basic aqueous solutions showed that
both the corrosion rate and the corrosion mechanism strongly
depend on the concentration and the temperature of the solution
reaction medium is an attractive solvent in view of the avail-
ability, ecological cleanliness, and practical safety.
Reported experimental work employed static and quasi-
dynamic test procedures developed to evaluate the chemical
durability and to determine dissolution rates of tested ceramic
materials. Corrosion in deionized water as the reference corro-
sion medium was carried out simultaneously in order to reveal
potential influence of the ionic strength.
N. Jacobson—contributing editor
In works published previously by other authors the extent of
corrosion was monitored applying numerous techniques, relying
mostly on determination of weight loss or weight gain of cor-
roded specimen, followed by complementary measurement of
the corrosion layer thickness and composition with various
microscopy techniques. In this work the emphasis was put on
Manuscript No. 28727. Received October 6, 2010; approved March 9, 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 is gratefully acknowledged.
*
Member, The American Ceramic Society.
w
Author to whom correspondence should be addressed. e-mail: dagmar.galuskova@
tnuni.sk
3
035

3
036
Journal of the American Ceramic Society—Galuskova´ et al.
Vol. 94, No. 9
the analysis of specific elements released from the material into
the corrosion medium rather than on determination of the
weight change alone, with complementary detailed study of the
corrosion layer composition (phase and chemical). By combi-
nation of all mentioned techniques a set of data were obtained,
which facilitated determination of the kinetic parameters of cor-
rosion, together with broad analysis of corroded material and
formed secondary products.
quasi-dynamic test conditions were also applied. Due to high
temperatures used, typical flow-through experiment with fresh
corrosion medium continuously flowing around the tested ma-
terial could not be performed. Therefore 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 test was carried out in two reactors in
parallel at the temperature of 2901C, and with the same speci-
mens for the whole duration of test. The S/V ratio was identical
to the static tests. The content of released elements including the
II. Experimental Procedure
total concentration of ammonia (NH
and in molecular form) were determined in each eluate.
(A) NL value calculations: The amount of an element
released into solution was expressed in terms of the value NL in
(g/m ) according to the Eq. 1, which is often applied in studies of
3
)-bound N (both in ionized
(1) Materials
Two materials, Si N (S) and sialon (SA) were selected to study
3
i
4
the corrosion behavior of nonoxide, Si
compositions are given in Table I. The Si
3
N
4
-based materials. The
specimens were
i
2
3
N
4
1
6
prepared from the powder SN-E10 (UBE Industries Ltd.,
Tokyo, Japan) with Y O (4N label, Pacific Industrial Devel-
chemical durability of glasses
2
3
opment Corporation, Ann Arbor, MI) as the sintering aid. For
preparation of sialon with the nominal composition Si Al
the following commercial powders were used: a-Al O (Martox-
ci
NLi ¼
(1)
S
2
O N
4 4 4
wi
V
2
3
ide PS-6, Martinswerk GmbH, Bergheim, Germany); a-Si N /
3
4
i
where c is the concentration of an element (i) in solution in (mg/
LC12-S (H.C. Starck, Goslar, Germany); and AlN/grade C
(H.C. Starck). The powder batch was attrition milled in
isopropanol for 4 h. The powders were dried under infrared
lamp and sieved, respectively, through a 25 and 71 mm sieve to
3 4
prepare Si N and sialon powders with required granulometry,
blocks with dimensions of 50 mm ꢁ 50 mm ꢁ 5 mm prepared by
hot pressing at 17501C and pressure 30 MPa under nitrogen (N)
pressure of 0.1 MPa were cut and ground into rectangular bars
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
2
3
in (m ), and w is the mass fraction of the element (i) in the ce-
i
ramic material. This way the amount of the element released
into the solution is normalized with respect to its content in
corroded material. Identical NL values of all elements indicate
i
congruent dissolution of the material. The variation of the NL
i
values indicates either preferential dissolution and release of
some elements, or precipitation of corrosion products and de-
pletion of the solution of some elements.
Data acquired from the quasi-dynamic tests at the respective
time of corrosion t were recalculated according to the following
equation:
3
mm ꢁ 4 mm ꢁ 50 mm, their faces polished, and edges cham-
fered. The bars were then ultrasonically cleaned in acetone for
5 min, rinsed with distilled water, and dried for 3 h at 1101C.
1
(
2)Corrosion Experiments Under Static Conditions
ꢀ4
Deionized water (conductivity 2575 10 S/m) used as the ref-
erence medium was prepared in the reverse osmosis purification
system (CHEZAR, Bratislava, Slovak Republic). The 0.5 mol/L
NaCl solution was prepared by dissolving 29.9670.03 g of NaCl
ci
t
i
þ NL ti ꢀDt
NL ¼
(2)
S
wi
V
(
MERCK, Darmstadt, Germany) in 1000 mL of deionized wa-
ter. The bars were placed in teflon-lined pressure corrosion re-
3
actors with the inside volume 26 cm filled with corrosion liquid
The Eq. 2 includes cumulative effect of the amount of dis-
solved elements in the studied time interval Dt.
and heated in laboratory drying oven. Both static and quasi-dy-
namic tests were carried out at the temperature of 2901C with
maximum duration of the test 480 h. Additional static tests at
temperatures 1501 and 2001C were applied in order to obtain the
data for determination 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
(
4) Methods of Analysis
The contents of metallic elements (Si, Al, Y) released into the
solution were determined with the use of the Inductively Cou-
pled Plasma Atomic Emission Spectroscopy (ICP AES, Vista
MPX, Varian Australia Pty Ltd., Mulgrave, Australia) with ra-
dial viewing, equipped with V-groove nebuliser and Sturman–
Masters spray chamber. The presence of high concentrations of
ꢀ1
0
.770.03 cm . For determination of each experimental data
point two specimens were placed in one reactor with 20 mL of
the liquid medium. The tests were carried out in three reactors in
parallel, i.e. six specimens were tested under identical conditions.
Fresh specimens were used for each test. After the test the sam-
ples 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 concen-
tration determined in the eluates taken after the corrosion tests
with the ceramic material.
1
metallic ions in the corrosion solution, such as Na , might sig-
A simple internal standard-
1
7,18
nificantly disturb the analysis.
ization technique with beryllium as an internal standard was
therefore utilized for multielement analysis of the eluate solu-
tions containing 0.5 mol/L NaCl, to eliminate possible errors
19
introduced by the solution with high-salt content.
Si is known to react, under the conditions similar to that
of the corrosion test, with water, yielding silica, and NH
3 4
N
3
5
according to the following reaction :
Si3N4 þ 6H2O ) 3SiO2 þ 4NH3
(3)
(
3) Quasi-Dynamic Test
To avoid oversaturation of solution with regard to precipitation
of corrosion products, which could act as a passivation layer,
NH
3
dissolves in water until equilibrium between NH
3
gas
and ammonium ions is established. Not all of the dissolved NH
3
reacts with water to form ammonium ions. A substantial frac-
tion remains in the solution in molecular form. For quantifica-
tion of the total NH nitrogen (both in ionized and molecular
form) in eluates the spectrophotometric method with sodium
nitroprusside (UV-VIS-NIR spectrometer Cary 2300, Varian
Techtron Pty Ltd., Mulgrave, Australia) was employed accord-
ing to the standard STN ISO 7150-1.
Table I. Starting Compositions of Tested Materials
3
Composition of studied materials in wt%
Si N
3 4
Y O
2 3
a-Al O
2 3
Si N
3 4
AlN
Si N (S)
3
95
5
Sialon (SA)
47.9
32.9 19.2
4

September 2011
Corrosion of Silicon Nitride-Based Ceramics
3037
Corrosion surfaces and polished cross sections of corroded
specimens were examined by scanning electron microscopy
(
SEM/EDX, Carl Zeiss SMT GmBH, model EVO 40 HV,
Jena, Germany) at accelerating beam voltage 20 kV. Corrosion
products formed at exposed surfaces were investigated with the
use of X-ray diffractometer (XRD, Bruker AXS Inc., Madison,
WI) specially designed for measurement of thin layers, in the 2y
interval 201–801 using CuKa radiation.
III. Results
(1) Microstructure
Characteristic microstructure of specimen of the material S
consists predominantly of elongated homogeneously distributed
b-Si
boundary phases (lighter areas), as illustrated in Fig. 1(a). The
Si grain size ranges between 150 and 500 nm with randomly
3 4
N grains (dark gray areas) with low aspect ratio and grain-
3 4
N
scattered grains with diameter up to 2 mm. Some residual pores
in the range between 0.1 and 1.0 mm are present in the material.
The main phase is b-Si N , with small amount of a-Si N . Some
3
4
3
4
additional crystalline phases have been also detected, such as
YSiO N and Y Si
2
2
3 3 4
O N .
Microstructure of the sialon (SA) is relatively fine grained,
and characteristic for this type of nonoxide ceramics (Fig. 2(b)).
The grain size ranges usually from 1 to 1.5 mm, with small frac-
tion of randomly scattered grains with the size up to 10 mm. The
material is also characterized by the presence of residual poros-
ity up to 3%, and phase inhomogeneities comprising mainly
2 3
small fraction of unreacted a-Al O , slightly above the detection
limit of the XRD.
(2) Static Experiments
(
A) Corrosion in Deionized Water: The time depen-
dences of the weight change normalized with respect to the cor-
roded surface area for the tests carried out in deionized water at
1501, 2001, and 2901C are summarized in Figs. 2(a) and (b). At
Fig. 2. Weight loss measured for S and SA specimens in deionized wa-
ter at 1501 and 2001C (a) and at 2901C (b).
1501 and 2001C similar weight losses were measured for both
materials exposed to deionized water. Unlike for sialon speci-
mens, rapid loss of weight of Si N specimens was observed at
3
4
the temperature of 2901C up to 48 h of the test later followed by
a steady state until the end of the experiment. Irrespective of the
temperature, nearly no weight change was observed during the
dissolution of sialon ceramics (Figs. 2(a) and (b)). This might
indicate either the temperature independence of dissolution of
the sialon material, or that the state of saturation was achieved
soon after the exposure to deionized water.
However, various factors, especially inhomogeneous dissolu-
tion in some places (pitting corrosion) with simultaneous pre-
cipitation of reaction products and formation of passivation
layer at others, or random cracking and peeling off the passi-
vation layer might influence the weight loss-time dependences
significantly. We therefore do not consider the weight loss to be
a correct parameter for evaluation of corrosion rates and mech-
anisms, as will be discussed later.
Although the preliminary considerations based on weight loss
measurements could classify sialon as promising and chemically
resistant material, which could be applied under studied condi-
tions and in the studied temperature interval. The chemical
analysis of eluates did not support the expectations.
Reaction of Si
yielding NH and silicon dioxide as the reaction products. No
specific influence of the selected corrosion media on dissolution
of Si ceramics could be observed, irrespective of the temper-
3 4
N with water proceeds according to the Eq. 3
3
3 4
N
ature (Fig. 3). However, higher NL(N) values were detected for
dissolution of sialon in NaCl solution at 2901C, which was
nearly eight times higher than for dissolution in deionized water
(
Fig. 4) at the same temperature. Corrosion tests with sialon in
Fig. 1. Microstructure of the material S (a) and of the material SA (b).
0.5M NaCl were carried out only for temperature 2901C.

3
038
Journal of the American Ceramic Society—Galuskova´ et al.
Vol. 94, No. 9
Fig. 3. Time dependence of NL(N) values obtained for S specimens
dissolved under static conditions at all temperatures in both corrosion
media.
NH
901C was recalculated to the normalized amount of N and
compared with the amount of dissolved Si (Fig. 5(a)). Compa-
rable amounts of N and Si were leached from Si ceramics to
3
determined in eluates after corrosion of S specimens at
2
3 4
N
the solutions at all applied temperatures under static conditions.
The NL values of N and Si determined in the corrosion medium
were identical, in other words the dissolution of Si
congruently indicating preferential degradation of Si
3
N
4
proceeds
4
N matrix
3
grains. Yttrium was not considered because concentrations in
both deionized water and NaCl solution are near the limits of
detection (0.01 mg/L) for applied analytical method. In the time
interval up to 48 h the time-concentration dependence of both
elements Si and N, is approximately linear, then the rate of
the dissolution reaction slows down as the equilibrium is
approached, and a layer of passivating precipitates is formed
at the surface. The XRD analysis confirmed the presence of
cristobalite (PDF 76–941) at corroded surfaces.
Fig. 5. Normalized amount of nitrogen and silicon as a function of
corrosion time for Si N (a) and sialon (b) specimens, in deionized water
at 2901C. The NL(Al) values (b) are included for the sialon.
3
4
NH was quantified also in the eluates from the corrosion of
3
sialon ceramics in deionized water (Fig. 5(b)). Already in the
first hours of dissolution significantly lower NL values of dis-
solved Si were measured in comparison to NL values of NH3
phase, cristobalite. The layer of precipitates was found to form
already after 48 h of the test, remaining compact during whole
time interval studied. The examination of the surface of sialon
ceramics corroded under the same conditions (Fig. 6(b)) re-
vealed the presence of unidentified phase, with crystal-like mor-
phology, similar to that observed by Nagae et al. However,
based on the XRD analysis these authors had obtained no
evidence of the compounds other than sialon at corroded sur-
faces. In our case the XRD analysis of corrosion products
removed from the specimen’s surface revealed the presence of
nitrogen and with very low NL of aluminum, ranging from only
0
2
.05 to 0.1 g/m . This indicates that the saturation with respect
to Si and Al was attained already in the early stage of the dis-
solution reaction as will be discussed in more detail below.
The SEM analysis of the specimen S corroded for 288 h at
20
2
901C (Fig. 6(a)) indicates the presence of newly formed SiO
2
2 3
a-aluminum oxide (Al O ) (PDF 89–7717) together with the
Si Al (PDF 48–1617) sialon: both phases were identical to
2
4 4 4
O N
that detected by XRD in the underlying uncorroded matrix.
High background at the XRD diffraction pattern implies the
presence of significant amount of an amorphous phase of un-
known composition (Fig. 6(c)), possibly aluminum hydroxides,
hydrated silica or amorphous aluminosilicates of variable com-
position. The EDX analysis of the amorphous layer confirmed
the presence of Si, Al, O, and N. However, due to small thick-
ness of the precipitated layer the analysis was likely to be
strongly influenced by the underlying substrate, and is therefore
not considered as a suitable indication of the true composition
of the precipitates. The presence of unidentified layer covering
completely the surface of sialon specimens explains the retarda-
tion of the rate of dissolution in an early stage of the corrosion
process in deionized water (Fig. 5(b)), as well as negligible
weight change as seen in Fig. 2(b).
(
B) Corrosion in an Aqueous NaCl Solution: In the first
9
6 h of dissolution in 0.5 mol/L NaCl solution at 2901C both
Fig. 4. Comparison of NL(N) values obtained under static conditions
for SA specimens dissolved in 0.5M NaCl at 2901C, and in deionized
water at 1501, 2001, and 2901C.
materials exhibited similar decrease of weight (Fig. 7). However,
unlike dissolution in deionized water the steady state was not

September 2011
Corrosion of Silicon Nitride-Based Ceramics
3039
Fig. 6. SEM image of the surface of specimen S (a) corroded in deionized water 288 h at 2901C and SA (b) after 24 h in deionized water. X-ray
diffraction pattern of corrosion products removed from surface of SA specimen, corroded 480 h in deionized water (c).
achieved. After 96 h further rapid change was observed on the
weight loss-time dependence of the material S, while in the case
of SA approximately linear trend was preserved in the whole
studied time interval. The normalized amounts of Si and N in
eluates from S specimens corroded in 0.5M NaCl solution do
not significantly differ from those measured in deionized water
eluates, indicating similar mechanism of congruent dissolution
of the Si N matrix in both media (Fig. 8(a)). The retardation of
the corrosion process in the latter hours of the test observed in
deionized water was not recorded in this case. The surface of
3 4
Si N ceramics was covered with a layer of corrosion products,
morphologically, and from the point of view of their phase
composition, identical to that found at the surface of S material
corroded in deionized water. The XRD of corrosion products
removed from the surface of the S specimen revealed the pres-
ence of crystalline cristobalite (PDF 76–941). The main differ-
ence observed when compared with corrosion in deionized water
was extensive cracking and peeling off of the layer of precipitates
(Fig. 8(b)). Delamination of the precipitated layer might prevent
or delay the attainment of the steady state and may be also re-
sponsible for the abrupt weight loss observed after 96 h.
The amount of N released into the corrosion media was
found to be one order of magnitude higher for SA than for the S
ceramic (Fig. 9(a)). In principle sialon is, similarly to S, also ex-
pected to dissolve congruently, but the release of aluminum into
3
4
1
the solution together with Si and Na ions is believed to shift the
equilibrium quickly toward the formation of corrosion products
(Fig. 9(b)) other than those found in deionized water. This
would explain lower concentrations of Si measured in NaCl so-
lution than in deionized water as well as extremely low NL(Al)
values determined in both corrosion media. Investigation of the
corrosion products removed from the surface of corroded sialon
ceramic by XRD (Fig. 10) revealed the presence of NaAl3
Si
corundum, PDF-71-1123) and boehmite (PDF-1-1283).
Cracks and places with peeled off layer, on the Fig. 9(b)
3
O
10(OH)
2
(paragonite, PDF-24-1047) together with a-Al
2 3
O
(
marked with arrows, are likely to be responsible for restarting
and accelerating of dissolution process in NaCl solution.
(
3) Quasi-Dynamic Conditions
The formation of protective surface layer under static test con-
ditions at 2901C influences, and in some cases, effectively pre-
vents determination of kinetic parameters of the corrosion
process. In order to avoid saturation of the solution with re-
spect to the secondary phases formed at the surface of studied
materials in the course of corrosion quasi-dynamic arrangement
of the experiments was applied. The data were then used for
determination of the initial dissolution rates at 2901C. The
cumulative NL values of leached N and Si calculated according
Fig. 7. Weight loss measured for S and SA specimens in aqueous
sodium chloride solution at 2901C.

3
040
Journal of the American Ceramic Society—Galuskova´ et al.
Vol. 94, No. 9
Fig. 9. NL(Si) and NL(N) as a function of corrosion time for sialon
specimens in 0.5 mol/L NaCl solution at 2901C (a). SEM image of cor-
roded surface after 288 h at 290 1C (b). Cracks and the places with peeled
off layer of corrosion products are marked with arrows.
Fig. 8. Normalized amount of nitrogen and silicon as a function of
corrosion time for Si in 0.5 mol/L NaCl solution at all temperatures
a). SEM micrograph of corroded surface after 288 h at 290 1C (b).
3 4
N
(
Cracks and places with the peeled off-layer are marked with arrows.
to Eq. (2) and their time dependences in deionized water and in
were nearly three times higher comparing to NL(N) calculated
for dissolution in deionized water, and can be in part related to
increased ionic strength of the corrosion solution. In addition,
formation of precipitation products, AlOOH and paragonite
0
.5 mol/L NaCl solution are, respectively, shown in Figs. 11(a)
and (b).
Evaluation of experimental data obtained for the S ceramics
from quasi-dynamic tests at 2901C showed only negligible influ-
ence of corrosion media on the dissolution process. Speaking in
terms of the weight loss, the amount of S dissolved in deionized
3 3 2
NaAl Si O10(OH) different from those detected in deionized
water, deplete the solution of Al and Si promoting further dis-
solution of sialon (Fig. 11(b)). A layer of secondary crystalline
phases was found at the surface of SA specimen corroded in
both corrosion solutions even under quasi-dynamic conditions
at 2901C (Fig. 13).
2
water and in 0.5M NaCl solution was 0.005 and 0.006 g/cm ,
respectively, with the relative standard deviation (RSD) 7%.
The S material dissolves congruently with similar NL(N, Si)
values irrespective of the corrosion solution used. Analysis of
the surface of S specimen corroded under quasi-dynamic con-
ditions Figs. 12(a) and (b) revealed pitting corrosion. Higher
magnification images of attacked places, Fig. 12(b), revealed
needle-shaped imprints of dissolved Si N grains in residual
3
4
glassy phase.
The EDX analysis revealed significantly decreased concen-
tration of N in such places, which were at the same time mark-
edly enriched by Y and O. These observations together with the
chemical analysis of the eluate suggest that yttrium silicate
oxynitride grain-boundary glass phase is, at least under the
applied test conditions, resistant to attack by the used corro-
sion solutions.
Simultaneous congruent release of Si and N to deionized
water was observed in eluates from corrosion tests of the sialon
ceramics, while Al, detected in solution in significantly lower
amounts, reprecipitates in the form of hydroxid-oxides (such as
AlOOH) as confirmed by XRD where it was identified together
with a-Al O (corundum) phase, likely originating as a residuum
2
3
from the original, uncorroded material.
Fig. 10. X-ray diffraction pattern of the corrosion products removed
from the surface of SA specimen, corroded 480 h in 0.5M NaCl at
2901C.
1
There is, nevertheless, no doubt about the influence of Na
and Cl ions on dissolution of SA ceramics. The NL(N) values
ꢀ

September 2011
Corrosion of Silicon Nitride-Based Ceramics
3041
Fig. 12. SEM image of S specimen analyzed after corrosion in deion-
ized water (a) and in aqueous sodium chloride solution (b) under quasi-
dynamic conditions at 2901C. Detail of a pit: the Si N grains etched
3 4
away (b).
Fig. 11. Time dependences of NL(N) and NL(Si) leached from S and
SA specimen in deionized water (a) and in aqueous sodium chloride
solution (b).
3 4
glassy phase in the tested Si N ceramics. The corrosion rate
2
ꢀ1
21
1
4.271 mmol ꢂ (m ꢂ h) determined by Nickel et al. for Si N
3
4
powder corroded at 3001C show excellent agreement, within
error levels, with the findings obtained in this study, 15.070.2
(4) Corrosion Rate and Activation Energies
2
ꢀ1
.
mmol ꢂ (m ꢂ h)
As shown by the results of static tests the dissolution processes
of both studied ceramics are markedly influenced by the tem-
perature of corrosion media. The interpretation of corrosion
rates at temperatures above 1501C was influenced by saturation
of solution, which is the problem encountered also by other
Apparent activation energy of dissolution of sialon in deion-
ized water was 29.5710 kJ/mol and initial dissolution rate at
2
ꢀ1
2
901C was 10.6 mmolꢂ (m ꢂ h) which is comparable to disso-
lution rate determined for Si N ceramics. Nearly threefold in-
4
3
2
1
crease of dissolution rate of sialon in aqueous NaCl solution at
901C was observed in comparison to dissolution in deionized
authors. The rate constants of dissolution at the temperature
o2901C and at 2901C were determined from the linear parts of
2
2
ꢀ1
.
2
the experimental NL(N) (mol/m ) vs time dependences obtained
water, achieving the value 27.0 mmolꢂ (m ꢂ h)
from the analysis of the data acquired, respectively, from static
and quasi-dynamic tests. The activation energies were obtained
using the Arrhenius equation (Fig. 14)
ln k ¼ ln A ꢀ Ea=ðRTÞ
(4)
The values of rate constants and the activation energies are
summarized in Table II. The apparent activation energy of dis-
solution of Si N in deionized water and in 0.5M NaCl solution
3
4
is 73.678 and 69.3716 kJ/mol, respectively. Similar values of
2
2
a
E (73–110 kJ/mol) were estimated by Sato et al., and So-
1
0
3 4
miya, who concluded that dissolution of Si N ceramics under
hydrothermal conditions (below 3001C) is controlled by surface
chemical reactions. In contrary Yoshimura concluded that the
activation energies at the level 73–84 kJ/mol as determined in his
23
work correspond to diffusion of H O in amorphous silica.
2
Based on the literature data it is therefore not possible to make
unambiguous conclusions on the controlling mechanism. Very
low concentrations of yttrium determined in the corrosion
medium suggest a high chemical durability of yttrium oxy-nitride
Fig. 13. A layer of corrosion products observed by SEM on a SA spec-
imen corroded in NaCl solution. The numbered places were analyzed by
EDX.

3
042
Journal of the American Ceramic Society—Galuskova´ et al.
Vol. 94, No. 9
form of soluble silicic acid. Consequently, the equilibrium is
reached and solution becomes saturated with respect to possible
secondary phases, which precipitate at the surface of the mate-
rial and form a passivation layer.
The mechanism of dissolution of sialon in aqueous solutions
could not be established unambiguously, but based on the
results obtained in this work the following reaction (Eq. 6)
could possibly be proposed for the reaction process:
þ
Si2Al4O4N4 þ 2ðn þ 2ÞH2O þ 12H
!
(
6)
3þ
ꢀ
þ
4Al þ 4OH þ 2SiO2nH2O þ 4NH4
Ammonium detected in the eluates from sialon ceramics
could originate not only from cleavage of Si–N but also of
Al–N bonds. Exposure of AlN to water results in formation of
aluminum hydroxides and hydroxid-oxides, according to the
2
7,28
following reactions
:
ꢀ
AlN ꢀ þ2H2O ! AlOOHðamorphÞ þ NH3
(7)
(8)
3 4
Fig. 14. Arrhenius plot of ln k vs 1000/T for sialon and Si N .
AlOOHðamorphÞ þ H2O ! AlðOHÞ ðcrystalÞ
3
IV. Discussion
The Si–N–Si structure in Si -based compounds is rendered
Unlike for material S the presence of aluminum along with Si
dissolved in solution as silicic acid is likely to shift the equilib-
rium resulting in the formation of insoluble aluminosilicate or
hydroxyaluminosilicate phases at the very beginning of the
dissolution process. The formation of hydroxyaluminosilicates
requires the prior formation of Al(OH)3 solid. In aqueous
3 4
N
rigid by the necessity of N forming three rather than two bonds
unlike for flexible, adjustable Si–O–Si bridge bonds in silicon
2
4
dioxide. Schott et al. reviewed the mechanisms, which control
kinetics of mineral dissolution and precipitation in aqueous so-
lution. They concluded that water molecules coordinate to metal
centers and dissociative chemisorptions occurs with formation
environment formation of Al(OH) solids appears greatest at
3
2
9
pH above 6.0. The reaction between aluminum and silicic
acid in solution can be then defined according to the following
equation :
1
ꢀ
of positively (–OH ), negatively (–O ) charged and neutral
2
30
hydroxyl groups (–OH) depending on pH of the solution.
Thus in dissolution of the studied ceramic materials a series of
adsorption and hydrolysis reaction steps affecting the bonds of
Al þ SiðOHÞ , AlOSiðOHÞ3 þ Hþ
3
þ
2þ
(9)
4
SiN
4
tetrahedra might be involved. The bonding around the N
2
atoms is somewhat sp -like: that is, three hybrids of s, p , and p
The protective layer of precipitates slows down the process so
that further dissolution of S and SA ceramics is controlled either
by diffusion of NH or water molecules through precipitated
3
x
y
z
form three bonds to Si atoms, with the p orbitals (sticking out
25
of the plane) nonbonding. The free electron pair of N under
favorable condition of subcritical water is likely to react with H
1
layer or by solubility of the primary corrosion products.
In both studied ceramics corroded in NaCl solution the de-
struction of surface layer of corrosion products appeared to
contribute to more severe dissolution of the materials. Based on
similar observations of the influence of various, especially ha-
lide, anions, on the corrosion of stainless steels and nickel-based
ions along with chemisorptions of water molecules to Si atom
center.
3 4
Si N ceramics then in a contact with aqueous environment
2
dissolves as follows :
1
2,13
Si3N4 þ 3ðn þ 2ÞH2O þ 4Hþ ! 3SiO2 ꢂ nH2O þ 4NHþ (5)
alloys
one can speculate that the chloride anion, as an ag-
4
gressive agent in aqueous NaCl solution promotes destruction
of the protective oxide layer. Such local destruction of the layer
enables direct access of water, which then reacts with freshly
exposed surface of the material, thus enhancing severity of the
attack.
Based on the results shown above we propose that the Si–N
bonds in Si N and both the Si–N and Al–N bonds in sialon are
3 4
attacked preferentially under the conditions of the corrosion
test. Successive adsorption and hydrolysis reactions affecting the
The penetration of corrosion media through the precipitated
layer then results in its cracking and peeling off (the places
marked with arrows in the Figs. 8(b) and 9(b)). This would in
turn restart the corrosion process at freshly exposed surface.
Moreover, the sodium ions together with Al ions are believed to
shift the equilibrium towards precipitation of sodium alumino-
silicates, such as paragonite identified in precipitation layer.
Dissolution restarts due to depletion of solution from precipi-
tating elements. N is not incorporated into any secondary phase
detected by XRD, and remains in solution in the form of dis-
solved ammonium ion.
Si–N bonds result in production of NH
3
, which dissolves in
solution changing pH in our case from 6.270.3 to 9.670.1
while Si is gradually released and in solution present in the
2
6
Table II. The Activation Energy of Dissolution of S and SA
Ceramics and Rate Constants Measured for Corrosion in
Deionized Water and 0.5M NaCl
Deionized water
0.5 mol/L NaCl
2
ꢀ1
2
ꢀ1
)
Samples
k (mmol ꢂ (m ꢂ h)
)
E (kJ/mol) k (mmol ꢂ (m ꢂ h)
E (kJ/mol)
Si N /S/
3
4
2
2
1
901C 15.070.2
001C 0.6970.07
501C 0.1170.01
73.678
20.170.7
0.6370.03
0.1670.01
69.3716
V. Conclusions
Two kinds of tests, both under static and quasi-dynamic condi-
tions, were applied in order to understand dissolution mecha-
nism of Si N and sialon ceramics under subcritical conditions
in neutral aqueous solutions. The corrosion rates for both ce-
ramics are increasing with temperature. The tests carried out
under static conditions enabled identification of precipitated
corrosion products after saturation was achieved under the
Sialon/SA/
3
4
2
2
1
901C 10.670.3
29.5710 27.072
—
—
—
001C
501C
1.870.3
1.470.2
—
, not determined.

September 2011
Corrosion of Silicon Nitride-Based Ceramics
3043
Concentration, Temperature and Composition,’’ J. Eur. Ceram. Soc., 27, 3573–88
2007).
J. Schilm, M. Herrmann, and G. Michael, ‘‘Leaching Behavior of Silicon Nit-
conditions of most severe attack at 2901C. The corrosion tests
performed under quasi-dynamic conditions facilitated determi-
nation of initial dissolution rates at 2901C and calculations of
activation energies. Both studied ceramics dissolved by prefer-
ential attack of Si–N bonds in the matrix accompanied by the
(
3
ride Materials in Sulphuric Acid Containing KF,’’ J. Eur. Ceram. Soc., 24, 2319–
27 (2004).
4
B. Seipel and K. G. Nickel, ‘‘Corrosion of Silicon Nitride in Aqueous Acidic
Solutions: Penetration Monitoring,’’ J. Eur. Ceram. Soc., 23, 595–602 (2003).
release of NH and formation of protective, mostly oxide layer
3
of corrosion products at the surface.
5
K. G. Nickel and B. Seipel, ‘‘Corrosion Penetration Monitoring of Advanced
Ceramics in Hot Aqueous Fluids,’’ Mater. Res., 7, 125–33 (2004).
6
In Si N
3 4
ceramics, the severity of corrosion attack increased
N. S. Jacobson, ‘‘Corrosion of Silicon-Based Ceramics in Combustion Envi-
ronments,’’ J. Am. Ceram. Soc., 76 [1] 3–28 (1993).
with time and temperature of corrosion. Corrosion pits were
formed at the surface in the initial stage of the process. Yttrium
silicate oxynitride grain-boundary phase was found to be highly
7
P. Kritzer, ‘‘Corrosion of High-Temperature and Supercritical Water and
Aqueous Solutions: A Review,’’ J. Supercrit. Fluids, 29, 1–29 (2004).
8
M. Nagae, T. Yoshio, and K. Oda, ‘‘Corrosion Behavior of Structural Ceram-
ics in Supercritical Water,’’ Adv. Sci. Technol., 45, 173–7 (2006).
resistant in both tested media. The Si
preferentially by congruent dissolution of Si
irrespective of the used corrosion solution, with initial dissolu-
N
3 4
ceramics thus corroded
9
K. Oda, T. Yoshio, Y. Miyamoto, and M. Koizumi, ‘‘Hydrothermal Corrosion
of Pure, Hot Isostatically Pressed Silicon Nitride,’’ J. Am. Ceram. Soc., 76, 1365–8
3 4
N
matrix grains
(
1993).
2
ꢀ1
tion rates of 15.0 and 20.1 mmol ꢂ (m ꢂ h) in deionized water
and NaCl solution, respectively, as determined at 2901C. The
apparent activation energies in deionized water and in the NaCl
solution were 73.678 and 69.3716 kJ/mol, respectively, indi-
cating the corrosion was controlled either by surface chemical
10
S. Somiya, ‘‘Hydrothermal Corrosion of Nitride and Carbide of Silicon,’’
Mater. Chem. Phys., 67, 157–64 (2001).
1
1
M. Yoshimura, J. Kase, and S. Somiya, ‘‘Hydrothermal Oxidation of Si
Powder,’’ J. Mater. Res., 1, 100–3 (1986).
3 4
N
12
P. Kritzer, N. Boukis, and E. Dinjus, ‘‘Factors Controlling Corrosion in High-
Temperature Aqueous Solutions: A Contribution to the Dissociation and Solu-
bility Data Influencing Corrosion Processes,’’ J. Supercrit. Fluids, 15, 205–27
2
reactions or by diffusion of H O in protective silica layer. Me-
chanical destruction of protective SiO layer in NaCl solution is
2
considered as the main reason for renewal of dissolution of
(
1999).
13
¨
H. Weingartner and E. U. Franck, ‘‘Supercritical Water as a Solvent,’’ Angew.
Chem. Int. Ed., 44, 2672–92 (2005).
Si N
3 4
at 2901C under static conditions.
The original assumption that the corrosion resistance of
Si N -based materials can be enhanced by dissolution of al-
14
M. V. Fedotova, ‘‘Structural Features of Concentrated Aqueous NaCl Solu-
tion in the Sub- and Supercritical State at Different Densities,’’ J. Mol. Liq., 143,
3
5–41 (2008).
R. W. Shaw, T. B. Brill, A. A. Clifford, C. A. Eckert, and E. U. Franck,
3
4
15
umina in the crystal structure, creating sialon in the process, was
not confirmed. The likely mechanism of corrosion of sialon ce-
ramics was preferential disruption of Si–N and Al–N bonds in
‘
‘Supercritical water a medium for chemistry,’’ Chem. Eng. News, 69, 26–39 (1991).
A. Helebrant, ‘‘Kinetics of Corrosion of Silicate Glasses in Aqueous Solu-
16
tions,’’ Ceram.-Silik., 41 [4] 147–51 (1997).
M. Iglesias, T. Vaculovic, J. Studynkova, E. Poussel, and J. M. Mermet, ‘‘In-
17
(
Si, Al)ON tetrahedra, with aluminum released into solution
2
fluence of the Operating Conditions and of the Optical Transition on Non-Spectral
Matrix Effects in Inductively Plasma-Atomic Emission Spectrometry,’’ Spectroc-
him. Acta B, 59, 1841–50 (2004).
precipitating in the form of insoluble hydroxyaluminosilicates
once the saturation concentration was attained. Apparent acti-
vation energy of dissolution of sialon in deionized water was
18
A. Montaser and D. W. Golightly, Inductively Coupled Plasmas in Analytical
Atomic Spectrometry, 3rd edition, VCH Publishers Inc, New York, 1992.
2
9.5710 kJ/mol and initial dissolution rate at 2901C was 10.6
1
9
2
ꢀ1
D. Galuskova, ‘‘Koro
roztoku chloridu sodneho,’’ Ph.D. Thesis, Slovak Technical University, Bratislava,
010.
zia kongtruken y´ ch keramicky´ ch materialov vo vodnom
´ ` ´
mmol ꢂ (m ꢂ h) which is comparable to dissolution rate deter-
´
N
3 4
2 3
mined for Si ceramics. Protective Al O layer was formed in
2
20
early stage of interactions of material with deionized water. The
presence of aluminum in aqueous NaCl solution together with
Si and Na ions resulted in early achievement of saturation with
M. Nagae, Y. Koyama, S. Yasutake, and T. Yoshio, ‘‘Corrosion Behaviour of
Sialon Ceramics in Supercritical Water,’’ J. Am. Ceram. Soc., 89 [11] 3550–3
(2006).
1
21
K. G. Nickel, U. Da
3 4
Si N ,’’ Key Eng. Mater., 89–91, 295–300 (1994).
22
¨
umling, and K. Weisskopf, ‘‘Hydrothermal Reactions of
3 3 2
respect to formation of corrosion products, NaAl Si O10(OH)
and AlOOH. In addition, nearly threefold increase of dissolu-
tion rate of sialon in aqueous NaCl solution at 2901C was
observed in comparison to dissolution in deionized water,
T. Sato, Y. Tokunaga, T. Endo, and M. Shimada, ‘‘Corrosion of Silicon Nit-
ride Ceramics in Aqueous Chloride Solutions,’’ J. Am. Ceram. Soc., 71 [12] 1074–9
1988).
(
23
M. Yoshimura, J. I. Kase, M. Hayakawa, and S. Somiya, ‘‘Oxidation Mech-
anism of Nitride and Carbide Powders by High-Temperature, High-Pressure
2
ꢀ1
achieving the value 27.0 mmol ꢂ (m ꢂ h) . Penetration of ag-
gressive chloride anions through protective layer of corrosion
products is proposed to contribute to destruction of the layer
and renewal of the exposure of vulnerable material surface to
corrosion medium.
Water,’’ Ceram. Trans., 10, 337–54 (1990).
J. Schott, O. S. Pokrovsky, and E. H. Oelkers, ‘‘The Link between Mineral
24
Dissolution/Precipitation Kinetics and Solution Chemistry,’’ Rev. Mineral. Geo-
chem., 70, 207–58 (2009).
S. Wild, P. Grieveson, and K. Jack, ‘‘The Crystal Structures of Alpha and Beta
25
Silicon and Germanium Nitrides’’; pp. 385–93 in Special Ceramics, 5, Edited by P.
Popper. Academic Press for the British Ceramic Research Association in London,
New York, 1972.
´
S. Gin, Ch Jegou, P. Frugier, and Y. Minet, ‘‘Theoretical Consideration on the
Application of the Aagaard-Helgeson Rate Law to the Dissolution of Silicate
Acknowledgment
26
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.
Minerals and Glasses,’’ Chem. Geol., 255, 14–24 (2008).
P. Bowen, J. G. Highfield, A. Mocellin, and T. A. Ring, ‘‘Degradation of
27
Aluminum Nitride Powder in an Aqueous Environment,’’ J. Am. Ceram. Soc., 73
[3] 724–8 (1990).
28
L. M. Svedberg, K. C. Arndt, and M. J. Cima, ‘‘Corrosion of Aluminum
Nitride (AlN) in Aqueous Cleaning Solutions,’’ J. Am. Ceram. Soc., 83 [1] 41–6
(
2000).
References
29
31
D. Dobrzynski, ‘‘Chemistry of Neutral and Alkaline Waters with Low Al
Solubility. The Evidence from
1
J. Schilm, W. Gruner, M. Herrmann, and G. Michael, ‘‘Corrosion of Si
3
N
4
-
Activity Against Hydroxyaluminosilicate HAS
Ground and Surface Waters of the Sudetes Mts. (SW Poland),’’ Aquat. Geochem.,
13, 197–210 (2007).
B
Ceramics in Aqueous Solutions. Part I: Characterisation of Starting Materials and
Corrosion in 1 N H SO ,’’ J. Eur. Ceram. Soc., 26, 3909–17 (2006).
J. Schilm, M. Herrmann, and G. Michael, ‘‘Corrosion of Si
Aqueous Solutions. Part II: Corrosion Mechanisms in Acids as a Function of
2
4
2
30
L. Spadini, P. W. Schindler, and S. Sjo
Complex in Aqueous Solution,’’ Aquat. Geochem., 11, 21–31 (2005).
21
berg, ‘‘On the stability of the AlOSi(OH)
3
N
3 4
-Ceramics in
¨
&