Paralinear oxidation of silicon nitride in a water-vapor/oxygen environment

Article abstract of DOI:10.1111/j.1151-2916.2003.tb03461.x

Three Si₃N₄ materials were exposed to dry oxygen flowing at 0.44 cm/s at temperatures between 1200° and 1400°C. Weight change was measured using a continuously recording microbalance. Parabolic kinetics were observed. When the same materials were exposed to a 50% H₂O-5% O₂ gas mixture flowing at 4.4 cm/s, all three types exhibited paralinear kinetics. The material was oxidized by water vapor to form solid SiO₂. The protective SiO₂ was in turn volatilized by water vapor to form primarily gaseous Si(OH)₄. Nonlinear least-squares analysis and a paralinear kinetic model were used to determine parabolic and linear rate constants from the kinetic data. Volatilization of the protective SiO₂ scale could result in accelerated consumption of Si₃N₄. Recession rates under conditions more representative of actual combustors were compared with the furnace data.

Full text of DOI:10.1111/j.1151-2916.2003.tb03461.x

High-Temperature Water Vapor Effects
J. Am. Ceram. Soc., 86 [8] 1256–61 (2003)
journal
Paralinear Oxidation of Silicon Nitride in a
Water-Vapor/Oxygen Environment
Dennis S. Fox,* Elizabeth J. Opila,*,† QuynhGiao N. Nguyen,*
Donald L. Humphrey,‡ and Susan M. Lewton§
John H. Glenn Research Center at Lewis Field, National Aeronautics and Space Administration,
Cleveland, Ohio 44135
species has been identified as Si(OH)4(g),19 which forms by the
following reaction:
Three Si3N4 materials were exposed to dry oxygen flowing at
0.44 cm/s at temperatures between 1200° and 1400°C. Weight
change was measured using a continuously recording mi-
crobalance. Parabolic kinetics were observed. When the same
materials were exposed to a 50% H2O–50% O2 gas mixture
flowing at 4.4 cm/s, all three types exhibited paralinear kinet-
ics. The material was oxidized by water vapor to form solid
SiO2. The protective SiO2 was in turn volatilized by water
vapor to form primarily gaseous Si(OH)4. Nonlinear least-
squares analysis and a paralinear kinetic model were used to
determine parabolic and linear rate constants from the kinetic
data. Volatilization of the protective SiO2 scale could result in
accelerated consumption of Si3N4. Recession rates under con-
ditions more representative of actual combustors were com-
pared with the furnace data.
SiO2͑s͒ ϩ 2H2O͑g͒ ϭ Si͑OH͒ ͑g͒
(2)
4
The simultaneously occurring oxidation and volatilization reac-
tions have been modeled using paralinear kinetics.16,20 The kinet-
ics can be described by
␣2kp Ϫ2kl⌬w1
2kl⌬w1
t ϭ
Ϫ ln 1 Ϫ
(3)
(4)
ͩ ͪ
ͫ
ͬ
2͑k2l ͒
␣kp
␣kp
⌬w2 ϭ Ϫ␤klt
2
For Si3N4, ␣ ϭ MWSiO /(MWO Ϫ MWN ) and ␤ ϭ MWSiN
/
2
2
2
1.3
3
MWSiO , where MW is the molecular weight, kp the oxidation
2
parabolic rate constant in units of specific weight squared versus
time (mg2/(cm4⅐h)), kl the linear rate constant (mg/(cm2⅐h)), ⌬w1
the weight gain from SiO2 growth, ⌬w2 the weight loss from SiO2
volatilization, and t the oxidation time.
I. Introduction
HE high strength, high temperature capability, and low density
The objective of this paper is to determine the oxidation
behavior of three Si3N4 materials in dry oxygen and in 50%
H2O–50% O2 from 1200° to 1400°C. Nonlinear least-squares
analysis and a paralinear kinetic model described in Ref. 16 are
used to determine kp and kl from the continuously measured weight
change data during the water-vapor exposures. Recession of the
substrate under the furnace conditions is estimated. Si3N4 reces-
sion rates from the furnace data are extrapolated to more realistic,
high-velocity conditions. These are compared with experimental
results from a high-pressure burner rig that operates at conditions
more typical of combustion applications.
T
of silicon-based ceramics, such as silicon nitride (Si3N4) and
silicon carbide (SiC), make them candidate materials for applica-
tion in the hostile environments of advanced aeropropulsion
engines. The dry oxidation behavior of Si3N4 has been studied in
great detail.1–6 If silicon-based ceramics are to be used in
combustion applications, their behavior under fuel-lean conditions
must be determined. Jacobson1 has calculated equilibrium prod-
ucts resulting from fuel-lean combustion of Jet A aviation fuel
(CH1.9185). The products include O2, H2O, CO2, and N2. Indepen-
dent of the fuel-to-air ratio, the amount of water vapor remains
almost constant at ϳ10%. In such an environment, the primary
oxidant is water vapor:7
II. Experimental Procedure
Si3N4͑s͒ ϩ 6H2O͑g͒ ϭ 3SiO2͑s͒ ϩ 2N2͑g͒ ϩ 6H2(g)
(1)
The materials used in this study are described in Table I. The
␣-Si3N4 is produced via chemical vapor deposition (Advanced
Ceramics Corp., Cleveland, OH). SN282 (Kyocera Industrial
Ceramics Corp., Vancouver, WA) is a high-strength ␤-Si3N4 with
an improved oxidation-resistant grain-boundary phase developed
for gas-turbine applications. The AS800 material (AlliedSignal
Ceramic Components, Torrance, CA) is an in situ-reinforced
␤-Si3N4 developed to enable a new generation of power systems.
A rare-earth-based sintering system is used in production of the
material, resulting in an apatite grain-boundary phase.
Others have recognized the need to study the oxidation of Si3N4
in water-vapor-containing environments.8–14 However, none have
measured kinetics continuously. In general, it has been found that,
above a few volume percent water vapor, the oxidation rate of
Si3N4 increases with water-vapor content. This also has been
shown for SiC.15 It has been demonstrated for SiC that, in addition
to the oxidation reaction, water vapor simultaneously reacts with
the growing SiO2 scale to form a volatile species.16–18 This
Specimens were cleaned using a detergent solution, distilled
water, acetone, and alcohol. Specimen weight was measured on an
analytical pan balance (Ϯ0.01 mg) before exposure and at the end
of the experiment. Oxidation experiments were conducted in a
vertical-tube furnace at three temperatures: 1200°, 1300°, and
1400°C (all Ϯ10°C). Fused-quartz furnace tubes were used
(Quartz Scientific, Fairport Harbor, OH). The oxidation kinetics
were followed using thermogravimetric analysis (TGA). Weight
change was continuously monitored using a recording microbal-
ance (Model C-1000, Cahn Instruments, Cerritos, CA). Specimens
were suspended on sapphire hangers inside a 2.5 cm ID vertical
D. Butt—contributing editor
Manuscript No. 187109. Received March 13, 2002; approved April 3, 2003.
*Member, American Ceramic Society.
†Cleveland State University Resident Research Associate at NASA GRC.
‡QSS Group, Inc., Brookpark, OH; Research Technician at NASA GRC.
§Work performed while a NASA/Ohio Aerospace Institute summer intern; cur-
rently at Department of Chemical Engineering, Purdue University, West Lafayette,
IN.
1256

August 2003
Manufacturer
Paralinear Oxidation of Silicon Nitride in a Water-Vapor/Oxygen Environment
1257
Table I. Silicon Nitride Materials
CVD Si3N4
SN282 Si3N4
AS800 Si3N4
Advanced Ceramics Corporation
Kyocera Industrial Ceramics
Corporation
AlliedSignal Ceramic
Components
Major additives/impurities†
Major
None
Lu
La, Y, Sr
Minor
ppm
C, O
W, C
C
H, Al, Fe
3.18
2.5 ϫ 0.7 ϫ 0.3
7.8
Yb, Ca, Mg, Y, Fe, Ba, Mn, Sr
Ba, Fe
Density (g/cm3)
TGA sample size (cm)
Surface area (cm2)
3.37
3.32
2.6 ϫ 1.3 ϫ 0.3
2.5 ϫ 1.3 ϫ 0.3
9.1
9.0
†Major is Ͼ1%, Minor is 0.1–1.0%.
furnace tube. Initial experiments were conducted with dry oxygen
flowing at 0.44 cm/s in a fused-quartz tube. Because of the high
sensitivity of the microbalance used to measure very small weight
changes, the flow rate was kept low to minimize noise in the
kinetic curve.
been reported. A National Institute of Standards and Technology
calibration standard was used to establish the appropriate correc-
tion factor for reporting thickness. It was determined that the
measured thickness was ϳ5.3% greater than the true thickness.
For the subsequent water-vapor–oxygen tests, an experimental
apparatus based on a design by Belton and Richardson21 was used.
A schematic is found in Ref. 15. The reaction gas was introduced
at the bottom of the fused-quartz furnace tube. Water was added to
the oxygen by first over-saturating the gas with deionized water. A
high flow rate of 4.4 cm/s was needed to enable saturation of the
gas stream. The gas then passed through a glass-bead-filled
saturator immersed in a water bath held at 81.7°C. The resulting
oxygen stream was saturated with 50% water vapor. A counterflow
of oxygen through the top of the TGA apparatus was used to keep
water from condensing on the microbalance and hang wire. At
least two specimens of each Si3N4 material were run at each of the
three test temperatures. Weight change versus time was continu-
ously measured during the 100 h. In this experiment, deposition of
volatile products on the hang down occurred.16 Control experi-
ments were conducted with the sapphire hanger alone, and a linear
weight gain was measured because of SiO2 volatilization from the
quartz tube and subsequent deposition on the hang down. This gain
was subtracted from the weight data acquired for each Si3N4
sample.
III. Results
(1) Dry Oxygen
Each of the three Si3N4 materials exhibits parabolic behavior in
dry oxygen. The average parabolic rate constants in this environ-
ment are listed in Table II with the calculated activation energies
for the oxide formation. Parabolic rate constants for each sample as
a function of reciprocal temperature are shown in Fig. 1. The pure
CVD Si3N4 is the most oxidation-resistant material. The Lu-
containing SN282 Si3N4 is more oxidation resistant than the
La-containing AS800 Si3N4.
After TGA exposure, the surface oxide (Fig. 2) is characterized
using XRD. The oxide grown on CVD Si3N4 at 1200°C is
amorphous, with ␣-cristobalite observed after 1300° and 1400°C
exposures. After exposure of the Lu-containing SN282 Si3N4,
␣-cristobalite and Lu2Si2O7 are observed. Lu2SiO5 is also present
at T Ն 1300°C. The AS800 oxide is ␣-cristobalite, with Y2O3 also
present only at 1400°C. The cracks in the surface oxide are formed
from the ␤- to ␣-cristobalite transformation on cooling.
To compare TGA furnace recession rates with those measured
under conditions more representative of actual combustors, CVD
and AS800 Si3N4 were also exposed in the NASA GRC high-
pressure burner rig (HPBR) under fuel-lean conditions.17 Addi-
tional HPBR tests of SN282 Si3N4 have been conducted. The rig
itself is described in detail in Ref. 18. Weight change and sample
thickness measurements were made every 8–16 h. The linear rate
constants were obtained from the kinetic data, as were the
parabolic rate constants when available.
The surface oxide was characterized using X-ray diffractometry
(XRD) following TGA exposure of the as-received specimens.
Oxide morphologies were then studied using scanning electron
microscopy (SEM; Model JSM-840A, JEOL, Tokyo, Japan).
Polished cross sections of selected samples were prepared to
measure the SiO2 scale thickness. One of the ϳ2.5 cm ϫ 0.3 cm
edges was polished to a 15 ␮m finish for observation using SEM
and entailed oxide thickness measurements every 1 mm along the
25 mm edge. The average thickness for each selected specimen has
(2) 50% H2O–50% O2
In 50% H2O–50% O2 flowing at 4.4 cm/s, paralinear kinetics
are observed for all three Si3N4 materials (Fig. 3). Although CVD
Si3N4 initially gains weight in the water-vapor environment, a
change to a linear weight loss occurs after ϳ20 h. This is due to the
volatilization of the SiO2 scale that occurs concurrently with the
oxidation. The noise in the curves is due to the high sensitivity of
the microbalance scale required to measure the very small weight
changes (ϳ0.1 mg/cm2) and to the high 4.4 cm/s flow rate.
Once the paralinear nature of the reaction in 50% H2O–50% O2
is established, the parabolic oxidation rate constants (kp) and the
linear volatility rate constants (kl) can be determined. The fit of the
model to the data is shown as the solid line in Fig. 3. This is the
sum of the weight changes from the oxidation and volatilization
reactions, as determined using the nonlinear least-squares analysis
of the data. Table III summarizes the parabolic and linear rate
Table II. Average Parabolic Rate Constants (kp) of Si3N4 Determined from TGA Experiments in
Dry Oxygen
kp (mg2/(cm4⅐h))
CVD Si3N4
SN282 Si3N4
AS800 Si3N4
Temperature (°C)
1200
1300
1400
5.6 (Ϯ1.3) ϫ 10Ϫ6
1.6 (Ϯ0.6) ϫ 10Ϫ5
5.2 (Ϯ0.8) ϫ 10Ϫ5
1.8 (Ϯ0.9) ϫ 10Ϫ5
5.5 (Ϯ1.4) ϫ 10Ϫ5
1.2 (Ϯ0.4) ϫ 10Ϫ4
2.1 (Ϯ0.2) ϫ 10Ϫ4
3.4 (Ϯ0.3) ϫ 10Ϫ4
8.0 (Ϯ1.6) ϫ 10Ϫ4
Activation energy (kJ/mol)
227 Ϯ 44
204 Ϯ 79
135 Ϯ 31

1258
Journal of the American Ceramic Society—Fox et al.
Vol. 86, No. 8
Fig. 1. Parabolic rate constants (kp) for the three Si3N4 materials in dry
oxygen flowing at 0.44 cm/s
constants for the three Si3N4 materials from the paralinear data in
50% H2O–50% O2.
The volatility fluxes determined experimentally for the three
Si3N4 materials are plotted in an Arrhenius diagram in Fig. 4. The
activation energy (Ea) from a fit of all the data points (solid line)
is calculated to be 36 Ϯ 86 kJ/mol (95% confidence interval),
which indicates little temperature dependence. Representative
photographs after 1400°C exposures are shown in Fig. 5. Oxide
thickness measurements, for single specimens after exposure in
50% H2O–50% O2, are listed in Table IV. The oxides grown on
CVD Si3N4 at 1200° and 1300°C are too thin to measure using this
technique, as is that on SN282 Si3N4 at 1200°C.
XRD of CVD Si3N4 at 1200°C shows the oxide to be amor-
phous. ␣-cristobalite is observed after 1300° and 1400°C expo-
sures. In comparison, Choi et al.,13 observed amorphous SiO2 after
steam exposure of CVD Si3N4 at 1000°–1300°C for 10 h. In the
present study, after exposure of the Lu-containing SN282 Si3N4,
␣-cristobalite and Lu2Si2O7 are observed using XRD. The AS800
oxides are identified as ␣-cristobalite, Y2O3, and La2Si2O7.
(3) High-Pressure Burner Rig
The only material to exhibit paralinear behavior in the HPBR
was AS800 Si3N4. Table V summarizes the parabolic and linear
rate constants for AS800 Si3N4 determined from the paralinear
data. The measured kp in the 0.6 atm PH O burner rig of ϳ5 ϫ
2
10Ϫ2 mg2/(cm4⅐h) is higher than the ϳ1 ϫ 10Ϫ3 mg2/(cm4⅐h)
measured in the 0.5 atm PH O TGA (1 atm ϭ 1 ϫ 105 Pa).
Impurities within the burner ri2g are thought to enhance oxidation.
The CVD Si3N4 and SN282 Si3N4 materials exhibit linear weight
loss. A weight-loss rate comparison under standard fuel-lean
conditions in the HPBR is shown in Fig. 6. As in the TGA
experiments, SN282 Si3N4 exhibits the lowest recession rates. This
is likely due to formation of Lu2Si2O7 in the surface oxide, which
results in a decrease of SiO2 volatility.
Fig. 2. Surface morphology of Si3N4 after 100 h at 1400°C in dry
oxygen: (A) CVD; (B) SN282; and (C) AS800.
IV. Discussion
(1) Parabolic Behavior in Dry Oxygen
The CVD Si3N4 is the most oxidation-resistant material (Fig. 1)
because of the formation of an oxynitride phase between the SiO2
and substrate.2–3 The mechanism is different in additive-
containing materials. In their study of the oxidation of hot-pressed
Si3N4 containing either MgO4 or Y2O3,5 Cubicciotti and Lau have
deduced that the rate-limiting step is the diffusion of additive
cations through the grain boundary to the surface oxide. MgSiO3
and Y2Si2O7 are observed in the cristobalite surface oxide.
Measured activation energies are 440 and 450 kJ/mol, respec-
tively. Clarke6 believes the growth of pure SiO2 creates a gradient
with the additive-containing silicate grain-boundary phase, driving
the cations to the surface oxide.
The SN282 Si3N4 oxidation mechanism is not clear. Lu2Si2O7
is present in the surface oxide of the SN282 Si3N4 material.
However, the 204 kJ/mol activation energy is lower than Cubic-
ciotti and Lau’s findings and higher than Deal and Grove’s 119
kJ/mol for single-crystal silicon. Others have studied the oxidation
of Si3N4 containing lanthanide rare-earth elements (which can also
include yttrium and scandium).22–24 There is no clear correlation
between the type of additive and oxidation resistance. Oxidation is

August 2003
Paralinear Oxidation of Silicon Nitride in a Water-Vapor/Oxygen Environment
1259
Fig. 4. Linear rate constants (kl) for all Si3N4 materials in 50% H2O–50%
O2 flowing at 4.4 cm/s ((‚) CVD, (ᮀ) SN282, and (E) AS800).
Fig. 3. Paralinear weight change kinetics for CVD, SN282, and AS800
Si3N4 at 1200°C in 50% H2O–50% O2 flowing at 4.4 cm/s. Solid line is the
result of the least-squares analysis of the data.
the oxide, resulting in a lower silicon activity. When the Table III
kl data are plotted on an Arrhenius diagram (Fig. 4), only slight
temperature dependence across the range of 1200°–1400°C is
observed for all three Si3N4 materials.
The measured kp values in 50% H2O–50% O2 in this study
(Table III) are 1–2 orders of magnitude higher than those measured
in dry oxygen (Table II), as is expected for SiO2 formers. Water
vapor enhances the oxidation rate in SiC26 and silicon.27 Very
small weight gains are measured early in the experiment and occur
at the ragged edge of the microbalance’s sensitivity. The kp is
determined from a curve fit to the data, leading to scatter in the
results and large standard deviations in the rate constants and
activation energies.
complicated in that it is a function of cation movement and the
nature of the grain-boundary phase. The La-containing AS800
Si3N4 is less oxidation resistant than SN282 Si3N4. The as-
received AS800 Si3N4 has various grain-boundary phases—apatite
and oxynitride. The oxide scale formed at 1200°C is solid and
uniform. At higher temperatures, the scale morphology is rough,
with glasslike nodules visible. The emergence of liquid phases
allows much more rapid oxygen ingress.
(2) Paralinear Behavior in 50% Water Vapor
The paralinear nature of the reaction kinetics of Si3N4 with 50%
H2O–50% O2 is obvious from the kinetic curves. Additional
evidence is provided by comparison of calculated weight change,
from SEM oxide measurements, with that actually measured. For
oxidation alone, the oxide thickness (x) and weight gain (in
mg/cm2) are related by the expression
The kl values are the more essential parameters derived from the
paralinear modeling. The recession calculations described below
are based on the SiO2 volatilization kinetics.
(3) Linear Weight-Loss Rates
The measured flux (Table III) can be compared with the
calculated flux.28 The appropriate relation for the flux of the
volatile species from a flat plate through a gaseous boundary layer
under laminar flow conditions is29
x ϭ 19.4⌬wt
(5)
when cristobalite is formed.25 Assuming no volatility and using the
measured oxide thickness, one calculates positive weight gains for
each of the materials (Table IV). However, there is a clear
discrepancy when the samples are weighed on the analytical pan
balance. Actual weight gains are much less than that calculated,
and, in two cases, overall weight losses are observed. The AS800
Si3N4 sample at 1400°C does not follow this trend because of
formation of a low-viscosity scale. It is therefore evident that,
although an oxide is continually being formed, volatilization of the
SiO2 by H2O is also occurring. The ultimate result of this
paralinear behavior is recession of the underlying Si3N4 substrate
and subsequent weight loss.
1/ 2
1/3
␳ЈvL
␩
D␳
J ϭ 0.664
(6)
ͩ ͪ ͩ ͪ
␩
␳ЈD
L
where J is the flux, ␳Ј the concentration of the major gas species,
v the linear gas velocity, L the sample length, ␩ the gas viscosity,
D the interdiffusion coefficient for the diffusing species in the
major gas species, and ␳ the concentration of the diffusing gas
species at the solid–gas interface. Gas concentration is calculated
using the ideal gas law. Hashimoto30 has measured thermody-
namic data for Si(OH)4. Krikorian31 and Allendorf et al.32 have
calculated data for the Si-O-H system. Using these data in a
free-energy minimization code (CHEMSAGE, GTT Technologies,
Sweden), one can calculate vapor pressures of Si-O-H species. For
The kl values measured for CVD Si3N4, AS800 Si3N4, SiO2,16
and CVD SiC16 under the same conditions are comparable. The kl
values for SN282 Si3N4 are slightly lower than those for the other
materials. This may be attributed to the formation of Lu2Si2O7 in
Table III. Average Parabolic (kp in mg/(cm2⅐h)) and Linear (kl in mg/(cm2⅐h)) Rate Constants of Si3N4 Determined from TGA
Experiments in 50% H2O–50% O2
CVD Si3N4
SN282 Si3N4
AS800 Si3N4
kp
kl
kp
kl
kp
kl
Temperature
(°C)
1200
8.8 (Ϯ1.0) ϫ 10Ϫ4 5.0 (Ϯ0.4) ϫ 10Ϫ3 9.3 (Ϯ6.7) ϫ 10Ϫ5 9.8 (Ϯ0.1) ϫ 10Ϫ4 1.7 (Ϯ0.1) ϫ 10Ϫ3 5.9 (Ϯ1.1) ϫ 10Ϫ3
3.5 (Ϯ2.4) ϫ 10Ϫ4 6.5 (Ϯ1.7) ϫ 10Ϫ3 4.2 (Ϯ2.5) ϫ 10Ϫ4 2.0 (Ϯ0.7) ϫ 10Ϫ3 8.6 (Ϯ1.8) ϫ 10Ϫ4 4.2 (Ϯ0.9) ϫ 10Ϫ3
1.0 (Ϯ0.1) ϫ 10Ϫ3 6.2 (Ϯ0.9) ϫ 10Ϫ3 4.5 (Ϯ2.2) ϫ 10Ϫ4 1.8 (Ϯ1.0) ϫ 10Ϫ3 5.2 (Ϯ0.9) ϫ 10Ϫ2 6.6 (Ϯ0.5) ϫ 10Ϫ3
1300
1400
Activation 20 Ϯ 211
energy
22 Ϯ 51
174 Ϯ 200
55 Ϯ 111
338 Ϯ 413
14 Ϯ 60
(kJ/mol)

1260
Journal of the American Ceramic Society—Fox et al.
Vol. 86, No. 8
temperature dependence of the measured flux is best described by
Si(OH)4 formation.
(4) Material Recession and Life Prediction
Many combustion applications require components to be used
for thousands of hours. If the component is fashioned from Si3N4
and is operating in a fuel-lean environment, some concerns need to
be addressed. The linear rate constant given in terms of weight loss
can be directly related to recession of the substrate. After an initial
period, the rate of SiO2 volatilization is equivalent to the rate at
which SiO2 is formed. Because at long times the substrate is
consumed to form the SiO2 at the same rate that the SiO2 is being
volatilized, substrate recession can be estimated using kl. A target
recession limit used by this laboratory for combustor liner appli-
cations is 2.5 ϫ 10Ϫ6 cm/h (10 mil/10 000 h (1 mil ϭ 10–3 in.)).
For CVD Si3N4 with a density of 3.2 g/cm3, this equates to a
maximum allowable kl of 8 ϫ 10Ϫ3 mg/(cm2⅐h). The values listed
in Table III are slightly lower than this allowable limit.
The TGA experiments provide an acceptable simulation of the
temperature and water-vapor partial pressures found in an engine
environment. However, engines operate at high total pressures.
The flow rates encountered in a turbine engine are also much
higher than the 4.4 cm/s used in the TGA experiments. At higher
flow rates and system pressures, flux estimates can be made using
a simplified form of Eq. (6). For this approximation, the diffusion
coefficient (D) is proportional to 1/Ptotal, and ␳Ј is proportional to
P
total. Assuming that Si(OH)4(g) is the only volatile species and is
formed via the reaction in Eq. (2), then ␳ is proportional to
PSi(OH)
:
4
v1/ 2P2H O
2
kl ϰ
(7)
1/ 2
P
total
For example, under the combustion conditions of 1200°C
(2200°F), gas velocity (v) of 21 m/s (63 ft/s), Ptotal of 600 kPa (6
atm), and PH O of 60 kPa (0.6 atm), the calculated kl is ϳ13 times
2
the furnace kl. Using AS800 Si3N4 as an example (3.32 g/cm3),
with its average kl of 5.9 ϫ 10Ϫ3 mg/(cm2⅐h) from the 1200°C
furnace data, the material’s recession rate would be 2.28 ϫ 10Ϫ5
cm/h (91 mil/10 000 h). This is an order of magnitude higher than
the acceptable limit. This increase is due to much higher combus-
tion gas velocity (v1/2 component in Eq. (7)).
The conditions described in the preceding paragraph are those
actually used for the high-pressure burner rig testing of all three
silicon nitrides under fuel-lean conditions. For an AS800 Si3N4
specimen exposed at 1232°C, the measured kl is 7.5 ϫ 10Ϫ2
mg/(cm2⅐h) (Table V). This correlates to a recession rate of 2.26 ϫ
10Ϫ5 cm/h (89 mil/10 000 h)—virtually the same rate predicted
above from the furnace results at 1200°C.
Si3N4 linear weight loss and surface recession result from SiO2
scale volatility as shown in the TGA experiments as well as in
more realistic burner rig tests. A considerable amount of substrate
recession can occur after only hundreds of hours of exposure in the
latter. Those proposing to use Si3N4 components in a fuel-lean
combustion application, such as a turbine engine, must therefore
be aware that SiO2 volatility can occur and result in loss of
component cross-sectional area.
Fig. 5. Surface morphology of Si3N4 after 100 h at 1400°C in 50%
H2O–50% O2: (A) CVD; (B) SN282; and (C) AS800.
V. Summary and Conclusions
these calculations, SiO2 (cristobalite) plus an initial gas com-
position of 50% H2O–50% O2 has been used. The calculated
vapor pressures of Si-O-H species are then used in the boundary
layer diffusion model, and Si3N4 mass loss rates are deter-
mined. The best agreement with the experimental data (Ea ϭ 36 Ϯ
86 kJ/mol) is for Si(OH)4 as the primary volatile species (Ea ϭ 69
kJ/mol). This calculated mass loss rate is shown in Fig. 4 as the dotted
line. SiO(OH)2 formation does not correlate with the experimental
data, because the calculated mass loss rates are ϳ10 times larger31 or
ϳ10 times smaller32 than those measured experimentally, and the
activation energy is high at 205 kJ/mol. Thus the magnitude and
Si3N4 exhibits parabolic kinetics when exposed to a dry-oxygen
environment flowing at 0.4 cm/s in a fused-quartz furnace tube at
1200°–1400°C. However, the material exhibits paralinear kinetics
when exposed in a 50 H2O–50% O2 gas mixture flowing at 4.4
cm/s. The material is oxidized by water vapor to form solid SiO2.
The protective SiO2, in turn, is volatilized by water vapor to form
primarily gaseous Si(OH)4. At long exposure times, the kinetics
can be approximated using the linear rate constants from the
volatilization reaction. As a result of this exposure, accelerated
consumption of Si3N4 occurs.

August 2003
Paralinear Oxidation of Silicon Nitride in a Water-Vapor/Oxygen Environment
1261
Table IV. Comparison of Calculated Weight Change from SEM Cross-Section Measurements
with Actual Measured Weight in 50% H2O–50% O2 Flowing at 4.4 cm/s
Calculated ⌬wt
(mg/cm2) from
oxide thickness
Measured ⌬wt
(mg/cm2) from
pan balance
Exposure
temperature (°C)
SEM oxide thickness
Si3N4 material
Time (h)
(␮m)
CVD
1400
1300
1400
1200
1300
1400
100
98
3.88 Ϯ 0.88
1.81 Ϯ 0.29
5.51 Ϯ 1.09
2.99 Ϯ 0.96
5.33 Ϯ 1.89
9.96 Ϯ 2.97
0.200
0.093
0.284
0.154
0.274
0.513
Ϫ0.274
ϩ0.029
ϩ0.009
Ϫ0.152
ϩ0.006
ϩ0.716
SN282
SN282
AS800
AS800
AS800
100
100
98
98
5D. Cubicciotti and K. H. Lau, “Kinetics of Oxidation of Yttria Hot-Pressed Silicon
Nitride,” J. Electrochem. Soc., 126 [10] 1723–28 (1979).
Table V. Parabolic and Linear Rate Constants
of AS800 Si3N4 Determined from HPBR
Experiments
6D. R. Clarke, “Thermodynamic Mechanism for Cation Diffusion through an
Intergranular Phase: Applications to Environmental Reactions with Nitrogen Ceram-
ics”; pp. 421–26 in Progress in Nitrogen Ceramics, NATO ASI Series, Series E:
Applied Sciences, No. 65. Edited by F. L. Riley. Martinus Nijhoff, The Hague, The
Netherlands, 1983.
kp
kl
Temperature (°C)
(mg2/(cm4⅐h))
(mg/(cm2⅐h))
7E. J. Opila and Q. N. Nguyen, “The Oxidation of CVD Silicon Carbide in Carbon
Dioxide,” J. Am. Ceram. Soc., 81 [7] 1949–52 (1998).
1163
1168
1232
1296
5.0 ϫ 10Ϫ2
7.0 ϫ 10Ϫ2
5.1 ϫ 10Ϫ2
4.8 ϫ 10Ϫ2
5.7 ϫ 10Ϫ2
5.7 ϫ 10Ϫ2
7.5 ϫ 10Ϫ2
1.0 ϫ 10Ϫ1
8S. C. Singhal, “Effect of Water Vapor on the Oxidation of Hot-Pressed Silicon
Nitride and Silicon Carbide,” J. Am. Ceram. Soc., 59 [1–2] 81–82 (1976).
9M. I. Mayer and F. L. Riley, “Sodium-Assisted Oxidation of Reaction-Bonded
Silicon Nitride,” J. Mater. Sci., 13, 1319–28 (1978).
10T. Sato, K. Haryu, T. Endo, and M. Shimada, “High-Temperature Oxidation of
Silicon Nitride-Based Ceramics by Water Vapour,” J. Mater. Sci., 22, 2635–40
(1987).
11M. Maeda, K. Nakamura, and T. Ohkubo, “Oxidation of Silicon Nitride in a Wet
Atmosphere,” J. Mater. Sci., 24, 2120–26 (1989).
12M. Maeda, K. Nakamura, and M. Yamada, “Oxidation Resistance of Silicon
Nitride Ceramics with Various Additives,” J. Mater. Sci., 25, 3790–94 (1990).
13D. J. Choi, D. B. Fishbach, and W. D. Scott, “Oxidation of Chemically-Vapor-
Deposited Silicon Nitride and Single-Crystal Silicon,” J. Am. Ceram. Soc., 72 [7]
1118–23 (1989).
14E. Proverbio, D. Rossi, and R. Cigna, “Influence of Water Vapour on High-
Temperature Oxidation of Al2O3–MgO-Doped Hot-Pressed Silicon Nitride,” J. Eur.
Ceram. Soc., 9, 453–58 (1992).
15E. J. Opila, “Oxidation Kinetics of Chemically Vapor-Deposited Silicon Carbide
in Wet Oxygen,” J. Am. Ceram. Soc., 77 [3] 730–36 (1994).
16E. J. Opila and R. E. Hann, “Paralinear Oxidation of CVD SiC in Water Vapor,”
J. Am. Ceram. Soc., 80 [1] 197–205 (1997).
17J. L. Smialek, R. C. Robinson, E. J. Opila, D. S. Fox, and N. S. Jacobson, “SiC
and Si3N4 Recession Due to SiO2 Scale Volatility Under Combustor Conditions,”
Adv. Compos. Mater., 8 [1] 33–45 (1999).
18R. C. Robinson and J. L. Smialek, “SiC Recession Caused by SiO2 Scale
Volatility under Combustion Conditions: I, Experimental Results and Empirical
Model,” J. Am. Ceram. Soc., 82 [7] 1817–25 (1999).
Fig. 6. Comparison of Si3N4 weight loss in the GRC high-pressure burner
rig under standard fuel-lean conditions (sample temperature of 1150°–
1330°C, gas velocity of 21 m/s, and total pressure of 600 kPa).
19E. J. Opila, D. S. Fox, and N. S. Jacobson, “Mass Spectrometric Identification of
Si(OH)4 from the Reaction of Silica with Water Vapor,” J. Am. Ceram. Soc., 80 [4]
1009–12 (1997).
20C. S. Tedmon Jr., “The Effect of Oxide Volatilization on the Oxidation Kinetics
of Cr and Fe–Cr Alloys,” J. Electrochem. Soc., 113 [8] 766–68 (1967).
21G. R. Belton and F. D. Richardson, “A Volatile Iron Hydroxide,” Trans. Faraday
Soc., 50, 1562–72 (1962).
Recession rates determined from the furnace experiments can be
used to estimate substrate recession under more realistic combustion
conditions. Component recession, predicted from TGA results and
observed under fuel-lean combustion conditions (T ϭ 1200°C, P ϭ 6
atm, vgas ϭ 20 m/s), is on the order of 2.5 ϫ 10Ϫ5 cm/h (1 mil/100
h). Engine designers must be aware of the possible rapid recession of
Si3N4 under turbine engine combustion conditions. Performance of
turbine vanes or blades, with their thin trailing edges, are especially at
risk from this mechanism of degradation.
22D. M. Mieskowski and W. A. Sanders, “Oxidation of Silicon Nitride Sintered
with Rare-Earth Oxide Additions,” J. Am. Ceram. Soc., 68 [7] C-160–C-163 (1985).
23M. K. Cinibulk and G. Thomas, “Oxidation Behavior of Rare-Earth Disilicate–
Silicon Nitride Ceramics,” J. Am. Ceram. Soc., 75 [8] 2044–49 (1992).
24H.-J. Choi, J.-G. Lee, and Y.-W. Kim, “Oxidation Behavior of Hot-Pressed Si3N4
with Re2O3 (Re ϭ Y, Yb, Er, La),” J. Eur. Ceram. Soc., 19, 2757–62 (1999).
25D. S. Fox, “Oxidation Behavior of Chemically-Vapor-Deposited Silicon Carbide
and Silicon Nitride from 1200° to 1600°C,” J. Am. Ceram. Soc., 81 [4] 945–50
(1998).
26E. J. Opila, “Variation of the Oxidation Rate of Silicon Carbide with Water-
Vapor Pressure,” J. Am. Ceram. Soc., 82 [3] 625–36 (1999).
27B. E. Deal and A. S. Grove, “General Relationship for Thermal Oxidation of
Silicon,” J. Appl. Phys., 36 [12] 3770–78 (1965).
Acknowledgment
28E. J. Opila, J. L. Smialek, R. C. Robinson, D. S. Fox, and N. S. Jacobson, “SiC
Recession Caused by SiO2 Scale Volatility under Combustion Conditions: II,
Thermodynamics and Gaseous-Diffusion Model,” J. Am. Ceram. Soc., 82 [7]
1826–34 (1999).
The authors would like to offer their sincere thanks to Ralph G. Garlick and
Dereck F. Johnson of the NASA Glenn Research Center at Lewis Field for the X-ray
diffraction results and chemical analysis of the starting materials, respectively.
29W. M. Kays and M. E. Crawford, Convective Heat and Mass Transfer; p. 139.
McGraw-Hill, New York, 1980.
References
30A. Hashimoto, “The Effect of H2O Gas on Volatilities of Planet-Forming Major
Elements: I. Experimental Determination of Thermodynamic Properties of Ca-, Al-,
and Si-Hydroxide Gas Molecules and Its Application to the Solar Nebula,” Geochim.
Cosmochim. Acta, 56, 511–32 (1992).
1N. S. Jacobson, “Corrosion of Silicon-Based Ceramics in Combustion Environ-
ments,” J. Am. Ceram. Soc., 76 [1] 3–28 (1993).
2H. Du, R. E. Tressler, K. E. Spear, and C. G. Pantano, “Oxidation Studies of
Crystalline CVD Silicon Nitride,” J. Electrochem. Soc., 136 [5] 1527–36 (1989).
3L. U. J. T. Ogbuji and E. J. Opila, “A Comparison of the Oxidation Kinetics of SiC
and Si3N4,” J. Electrochem. Soc., 142 [3] 925–30 (1995).
31O. H. Krikorian, “Thermodynamics of the Silica–Steam System”; p. 481 in
Symposium on Engineering with Nuclear Explosives, Vol. 1 (Las Vegas, NV, Jan.
14–16, 1970), unpublished.
32M. D. Allendorf, C. F. Melius, P. Ho, and M. R. Zachariah, “Theoretical Study
of the Thermochemistry of Molecules in the Si-O-H System,” J. Phys. Chem., 99 [41]
4D. Cubicciotti and K. H. Lau, “Kinetics of Oxidation of Hot-Pressed Silicon
Nitride Containing Magnesia,” J. Am. Ceram. Soc., 61 [11–12] 512–17 (1978).
15285–93 (1995).
Ⅺ