Combustion synthesis of Si3N4 by selective reaction of silicon with nitrogen in air

Article abstract of DOI:10.1111/j.1551-2916.2009.02929.x

Silicon nitride (Si₃N₄) was synthesized by a selective combustion reaction of silicon powder with nitrogen in air. The α/β-Si₃N₄ ratio of the interior product could be tailored by adjusting the Si₃N

Full text of DOI:10.1111/j.1551-2916.2009.02929.x

J. Am. Ceram. Soc., 92 [3] 636–640 (2009)
DOI: 10.1111/j.1551-2916.2009.02929.x
r 2009 The American Ceramic Society
ournal
J
Combustion Synthesis of Si3N4 by Selective Reaction of Silicon with
Nitrogen in Air
Jiang-Tao Li,z Lin Mei,w,z,y Yun Yang,z,y and Zhi-Ming Linz
zTechnical Institute of Physics and Chemistry, Chinese Academy of Sciences, Beijing 100190, China
yGraduated School of the Chinese Academy of Sciences, Beijing 100039, China
Silicon nitride (Si3N4) was synthesized by a selective combustion
reaction of silicon powder with nitrogen in air. The a/b-Si3N4
ratio of the interior product could be tailored by adjusting the
Si3N4-diluent content in the reactant mixtures. The synthetic b-
Si3N4 showed a well-crystallized rod-like morphology. Mechan-
ical activation greatly enhanced the reactivity of silicon powder,
and the slow oxidation of silicon at the sample surface promoted
the combustion reaction in air. The formation mechanism of
Si3N4 was analyzed based on a proposed N2/O2 diffusion kinetic
model, and the calculated result is in good agreement with the
experimental phenomenon.
formation mechanism of Si3N4 based on a proposed kinetic
model of N2/O2 diffusion.
II. Experimental Procedure
All reagents were obtained from the Chinese Chemical Market
and used as supplied: Si powder (purity 498.5 wt%, 0.4–0.7
wt% Fe, 45 mm, General Research Institute for Nonferrous
Metals, Beijing, China), a-Si3N4 (a-ratio 493.0 wt%, 1.5 wt%
oxygen, 5.0 mm, Shanghai Junyu Ceramic-Molded Product
Co. Ltd., Shanghai, China), and NH4Cl (purity 499.3 wt%,
Analytical Reagent, Beijing Chemical Corporation, Beijing,
China) as grinding aids. According to the mass ratio shown in
Table I, the reactant powders were dry blended and attrition
milled for different duration of time in a sealed agate jar, using
ZrO2 balls as the milling media, with a ball/charge weight ratio
of 10:1. The jar was made to vibrate at amplitude A 5 3–6 mm
and frequency f 5 24.3 Hz. Then the as-activated powders were
sieved through a 200-mesh screen and loaded loosely into a
horizontal quadrate graphite die of 220 mm length, 95 mm
width, and 70 mm height, which was placed in air. The sample
weight was fixed to 400.0 g and the initial porosity was estimated
to be 70%. The combustion reaction was initiated by igniting a
titanium powder compact placed on the top of the reactant. A
W-Re3/W-Re25 thermocouple was inserted directly into the
sample interior to record the temperature history of the com-
bustion reaction.
The phase assemblage of the as-synthesized product was
identified by X-ray diffraction (XRD; D/max-RB, Rigaku,
Japan) using CuKa radiation (l 5 1.54056 nm) with a step of
0.021. The phase content of a/b-Si3N4 was determined according
to Gazzara’s method.10 The microstructure of the product
was examined by scanning electron microscopy (SEM; JSM-
6460LV, JEOL, Japan). Fourier-transform infrared spectra
(FTIR; Excalibur 3100, Varian, Walnut Creek) were also used
to analyze the product, with a resolution of 0.20 cmꢀ1. Further
quantitative analysis was carried out by X-ray fluorescence spec-
troscopy (XRF; XRF-1800, Shimadzu, Kyoto, Japan) using
higher-order X-rays, where the percentage contents of Si, N,
O, Fe, and C were determined on BG–FP model.
I. Introduction
ILICON nitride (Si3N4) is a promising material for high-
S
temperature applications due to its high strength, high
toughness, excellent chemical stability, and wear resistance.1
For the synthesis of Si3N4 powder, direct nitridation, carbother-
mal reduction, vapor-phase reactions, and thermal decomposi-
tion are the most commonly used processes.2 Combustion
synthesis (CS) of Si3N4 powder is also an attractive technology
because of its greater energy efficiency, higher product purity,
and shorter production cycle.3 The theoretically calculated
nitrogen pressure required for carrying out the CS of Si3N4 is
over 104 MPa.4 However, the practical nitrogen pressure for CS
of Si3N4 has been decreased from 100 to 1.2 MPa due to the
continuous efforts of many researchers.5,6 It is demonstrated
that the gas–solid interfacial reaction between nitrogen and
silicon can be enhanced effectively by mechanical activation
(named hereafter as MA). These results indicate a potential of
realizing CS of Si3N4 under a much lower nitrogen pressure,
which is strongly desirable from the viewpoint of cost and safety.
Recently, the feasibility of synthesizing nitrogen ceramics by
combustion of several metallic powders in air has been reported.
For example, combustion of ultra-fine Al particle and (Al1C)
powders in air led to the formation of Al N instead of Al2O3.7,8
The TiN powder could also be synthesized by combustion of Ti
powder in air.9 These results suggest that single-phase nitrogen
ceramics could be synthesized through the selective nitriding
combustion of metallic powders with N2 in air, which provides
a possibility of synthesis and sintering of nitrogen ceramics in a
conventional kiln furnace without any atmosphere protection.
The present work attempts to synthesize Si3N4 powder by
combustion of nonmetallic powders containing Si/Si3N4/NH4Cl
in air, with the assistance of MA, and attempts to discuss the
III. Results
In order to identify the feasibility of the combustion reaction of
silicon powder in air, six experiments with different milling times
were performed, as shown in Table I. Only the reactants milled
for 16 h could be initiated to burn and self-propagated to all the
samples in the form of a combustion wave. After the combustion
reaction was over and the sample was cooled down, products
were collected from different locations, viz. the surface (H 50–7
mm) and the interior (H 5 7–35 mm), respectively, as shown in
Fig. 1. The whole product existed as a loose compact that could
be easily pulverized into powders by hand, while the samples
taken from the surface and interior showed different colors.
S. Danforth—contributing editor
Manuscript No. 25181. Received September 2, 2008; approved December 5, 2008.
This work was financially supported by National Natural Science Foundation of China
(Grant nos. 50502035 and 50772116).
wAuthor to whom correspondence should be addressed. e-mail: meilinipc@gmail.com
636

March 2009
Combustion Synthesis of Si3N4 in Air
637
Table I. Starting Compositions and Experimental Conditions
Starting composition
Milling time (h) and
Sample
Si:Si3N4:NH4Cl (wt%)
reaction characteristic
C1
C2
C3
60.5:25:14.5
80.5:5:14.5
35.5:50:14.5
0(N)w, 8(N), 12(N), 16(Y)
16(Y)
16(Y)
wThe abbreviation xx(Y/N) means that the combustion reaction could be re-
alized (Y) or not (N), with the reactants premilled for xx h.
Then the powder products were analyzed by XRD, and the
results are presented in Fig. 2. XRD patterns showed that the
product at the surface mainly consisted of a-Si3N4 and b-Si3N4,
except for a small amount of Si2N2O. However, the loose powder
taken from the interior was composed of single-phase Si3N4. This
is well consistent with the FTIR and XRF analysis results (see
Fig. 3). The absorption bands at 1039, 940, 915, 885, 576, and
440 cmꢀ1 were attributed to antisymmetric and symmetric
stretching vibrations of the Si–N bond, which further confirmed
that the powder collected from the interior was Si3N4, with only
4.72 wt% oxygen.11 FTIR analysis also demonstrated that a
mass of SiO2 existed in the surface product, corresponding to the
Fig. 2. X-ray diffraction patterns of the products from samples: (a) C1
surface, (b) C1 interior, (c) C2 interior, and (d) C3 interior. Indexed ac-
cording to JCPDS cards: 41-0360 (a-Si3N4), 33-1160 (b-Si3N4), and 84-
1813 (Si2N2O).
showed a coarse morphology. Because of the lower combustion
temperature, solid silicon powder may be converted into a-Si3N4
directly.
intensive absorption bands at 1094 and 494 cm .
ꢀ1 12 Meanwhile,
the oxygen content of the surface product increased considerably
to 26.8 wt%, although no diffraction peaks of SiO2 were found
in the XRD patterns. This suggests that SiO2 detected by FTIR
and XRF is amorphous, which may be caused by the rapid oxi-
dation-combustion of Si powder on the sample surface. There-
fore, Si3N4 was the dominant phase in the product block, and its
available scope is designated by a dashed circle shown in Fig. 1.
According to Gazzara’s method, the products collected from
the interior of sample C1 contained 55.21 wt% a-Si3N4. For
tailoring the a/b-Si3N4 ratio of the product, different amounts of
Si3N4 diluent were introduced into samples C2 and C3, viz. 5
and 50 wt%, respectively. As shown in Figs. 2(c) and (d), the
products from sample C2 were b-Si3N4, while nearly single-
phase a-Si3N4 was obtained in sample C3, with only a trace of
b-Si3N4 and residual Si. This could be attributed to the decrease
of combustion temperature from 1983 to 1638 K caused by the
massive addition of Si3N4 diluent, therefore, the phase transfor-
mation from a-Si3N4 to b-Si3N4 was suppressed.
IV. Discussion
(1) Effects of MA on the Microstructure of Si Powder
To investigate the MA effect during milling, the reactant mix-
tures with the composition of C1 were mechanically milled for 0,
8, 12, and 16 h, respectively. XRD analysis of the milled reac-
tants was conducted and is shown in Fig. 5. It was found that no
new phases appeared in the reactant powders with an increase of
the milling time, while the integrated intensity of the diffraction
peaks of Si decreased significantly, which may have been caused
by the formation of amorphous and nanocrystalline Si particles.
This is consistent with the literature report.15 The degree of am-
orphization of silicon (R) could be calculated according to the
equation
R ¼ ð1 ꢀ I=I0Þ ꢁ 100%
(1)
Figures 4(a)–(d) show the SEM images of the combustion
products. Both samples C1 and C2 were composed of well-crys-
tallized rod-like grains, which are the typical morphology of
b-Si3N4. The amplified observation indicated that the Si3N4 rod-
like grains in sample C1 were B0.3 mm in diameter and B3 mm
in length, while those in sample C2 exhibited a larger diameter
(B1 mm) and a clearer hexagonal columnar morphology. The
combustion temperatures of the samples were considerably
higher than both the melting point of Si and the eutectic point
of the M–Si—O–-N system (M5 Fe, C). Therefore, an M–Si–
O–N liquid could be formed that is beneficial for the a to b
phase transformation and the b-Si3N4 grain growth by the VLS
(vapor–liquid–solid) mechanism.13,14 However, sample C3
where I is the intensity of the Si(111) plane measured from the
reactants milled for different times and I0 is the intensity mea-
sured from the nonmilled (i.e., 0 h) powder mixture. It was
found that the degree of amorphization of silicon increased con-
siderably with the milling time, as presented in Table II. When
the milling time reached 16 h, 50.63% silicon became amor-
phous. Two possible amorphization mechanisms, i.e., stress-in-
duced amorphization and crystallite refinement-induced
amorphization, have been proposed for the amorphization of
Si induced by ball milling.15 Especially for the introduction of
NH4Cl, the pulverizing efficiency was considerably improved in
surface
H
interior
25mm
Fig. 3. Fourier-transform infrared spectra of the products from sample
C1: (a) surface and (b) interior, where the inset shows the element con-
tents measured by X-ray fluorescence spectroscopy (XRF).
Fig. 1. Digital camera photograph of the transverse section of sample
C1; H represents the distance from the sample surface toward the interior.

638
Journal of the American Ceramic Society—Li et al.
Vol. 92, No. 3
Fig. 4. Scanning electron microscopic images of the interior products from different samples: (a) C1ꢁ 4.0 K, (b) C1 ꢁ 20 K, (c) C2 ꢁ 20 K, and (d)
C3 ꢁ 4.0 K.
the milling process.6 Therefore, the MA enhanced the reactivity
of Si powder reactants, which promoted the CS of Si3N4 in air.
Fig. 6. It is obvious that the burning of Si powder in air occurred
via a two-stage self-propagating regime. In the first stage, the
temperature increased following a low slope, reaching 1473 K at
2000 s, which can be attributed to the slow oxidation reaction of
Si powder.16 Then the second combustion took place; corre-
spondingly, the temperature increased quickly up to the maxi-
mum Tmax 51898 K. Therefore, it is concluded that the
oxidation of silicon powder at the sample surface induced the
subsequent nitridation of Si in the form of a combustion reaction.
(2) Analysis of the Combustion Temperature Profile
Two contrast experiments were performed to deduce the reac-
tion mechanism, in which the sample C1 mechanically activated
for 16 h was ignited in air and 0.1 MPa nitrogen atmosphere,
respectively. Interestingly, the reactants placed in air underwent
a combustion reaction propagating to the whole sample,
whereas the combustion reaction could not be sustained in a
pure nitrogen atmosphere. It is believed that the O2 in air plays a
critical role in the full combustion of silicon powder. This is why
large amounts of Si2N2O and SiO2 were detected by XRD and
FTIR on the surface area of the product. It can be concluded
that the highly activated Si powder at the sample surface first
undergoes an oxidation reaction, accompanied by a mass of heat
release. Once the temperature of the local parts reached the ig-
nition point of the Si–N2 reaction, the fully CS was initiated and
propagated through the whole sample. This is in good agreement
with the recorded combustion temperature profile shown in
(3) Kinetic Analysis of the Si–N2-Selective Reaction
It is generally regarded that the final products after combustion
of Si in air should mainly be oxides instead of nitrides, because O2
is a much stronger oxidizer than N2 in a redox reaction. Mean-
while, the decomposition temperature of NH4Cl is as low as 618
K, and so it is reasonable to neglect the contribution of NH4Cl to
the nitridation of silicon powder due to its decomposition during
the slow oxidation process (see Fig. 6). However, the present re-
sults indicated that the dominating phase of the inner products
was Si3N4, while oxides such as Si2N2O and SiO2 were detected
only in the surface products. To explain this phenomenon, a N2/
O2 diffusion kinetic model is proposed as follows:
Typically, a cylindrical reaction zone with d 5 30 mm is cho-
sen for this study. It is assumed that N2/O2 should infiltrate
continuously from the sample surface to the combustion wave
front after the gas–solid combustion reaction is initiated, as
shown in Fig. 7. As the pore diameter of the green compacts is
much larger than the mean free path of N2/O2 molecules, the
infiltration of N2/O2 in the porous sample can be considered as
normal diffusion.17 Assuming that the infiltration paths are
Table II. Degree of Amorphization of Silicon in the Reactants
Milled for Different Duration of Time
Milling time (h)
0
8
12
16
Integrated diffraction 42285(I0) 20813 15306 12 121
intensity (I)
Degree of
amorphism (R) (%)
0
15.78
37.65
50.63
Fig. 5. X-ray diffraction patterns of the reactants milled for different
times.

March 2009
Combustion Synthesis of Si3N4 in Air
639
Fig. 6. Temperature profile for the combustion synthesis of Si3N4 from
sample C1.
Fig. 8. Infiltration flux Ji of N2/O2 at different distances H.
composed of a large amount of cylindrical pores with diameter
a, the flux of N2/O2 at the infiltration distance H can be
expressed as
the interior of the sample, which is far higher than that at the
surface (i.e., 31.88/8.58 5 3.72). Therefore, the combustion re-
action between Si and N2 will be the predominant reaction in the
interior of the sample. If the final products consist of only Si3N4
and SiO2, the molar ratio of Si3N4/SiO2 should be (6.99/2):(1/
1) 5 3.49 and, correspondingly, the weight content of SiO2
should be 10.92 wt%. Actually, O2 is overwhelmingly consumed
at the sample surface because of the lower infiltration velocity;
thus, the oxygen content of the final products detected by XRF
analysis was well below 10.92 wt%. This result demonstrates
that the kinetic discrepancy of N2/O2 diffusion enhanced the
selective nitridation reaction of Si in the interior of the sample.
Furthermore, in order to confirm the influence of N2/O2
diffusion on the phase stability of the products, two equilibrium
reactions in the Si–N–O system in the measured temperature
range are considered
ꢀ
ꢁ
CSDi
H
C ꢀ CiðHÞ
C ꢀ Cið0Þ
Ji ¼
ln
(2)
where C is the concentration of ambient air, Ci(0) is the
concentration of N2 and O2 at the surface, and Ci(H) is the
concentration of N2 and O2 at distance H. In the present case,
C 5 40.88 mol/m3, CN(0) 5 31.88 mol/m3, and CO(0) 5 8.58
mol/m3. In particular, both CN(H) and CO(H) equal to 0, be-
cause the infiltration velocity of N2/O2 is much lower than the
reaction velocity. S is the overall areas of diffusion pores. Di is
the diffusion coefficient as given by18
rffiffiffiffiffiffiffiffiffiffi
2
3
8RT VP
Di ¼
a
(3)
4Si3N4ðsÞ þ 3O2ðgÞ ¼ 6Si2N2OðsÞ þ 2N2ðgÞ
2Si2N2OðsÞ þ 3O2ðgÞ ¼ 4SiO2ðsÞ þ 2N2ðgÞ
(5)
(6)
pMi
t
where R is the gas constant, T is the temperature of the porous
sample, Mi is the molecular weight of gas, VP is the porosity of
sample, and t is the tortuosity factor. As for cylindrical diffusion
pores, t 5 1.
The Gibbs-free energy changes of reactions (5) and (6) under
the partial pressure of N2/O2, i.e., P (N2) and P (O2), can be
expressed as
In the present case, the measured values are as follows:
Tmax 5 1898 K, VP 5 60%, a 5 30.0 mm, thus, T 5 Tmax
/
2 5 949 K, S 5 0.25pd2, and VP 5 423.9 mm2. According to
Eqs. (2) and (3), Ji can be given as
DG1ðP; TÞ ¼ ꢀ249 000 ꢀ ½1014 ꢀ 16:629 ln PðN2Þ
2:72 ꢁ 10ꢀ4
þ 24:942 ln PðO2ÞꢂT
(7)
(8)
JN
¼
¼
(4--1)
(4--2)
2
2
H
3:87 ꢁ 10ꢀ5
DG2ðP; TÞ ¼ ꢀ2284 936 þ ½422 þ 16:629 ln PðN2Þ
ꢀ 24:942 ln PðO2ÞꢂT
JO
H
According to Eq. (4), the dependence of Ji on the infiltration
distance H was calculated and is shown in Fig. 8. Provided that
no reactions occur between N2/O2 and Si during the infiltration
process, the calculated infiltration flux ratio of N2/O2 is 6.99 at
Fig. 7. Ilustration of the N2/O2 diffusion kinetic mode: (A) longitudinal
section, (B) cross section.
Fig. 9. Phase stability diagram as a function of the partial pressure of
nitrogen and oxygen at T 5 1898 K.

640
In an equilibrium state,
DGðP; TÞ ¼ 0
Journal of the American Ceramic Society—Li et al.
Vol. 92, No. 3
References
1F. L. Riley, ‘‘Silicon Nitride and Related Materials,’’ J. Am. Ceram. Soc., 83 [2]
245–65 (2000).
(9)
2H. Lange, G. Wotting, and G. Winter, ‘‘Silicon Nitride from Powder Synthesis
to Ceramic Material,’’ Ang. Chem. Int. Ed. Engl., 30, 1579–97 (1991).
3K. Hirao, Y. Miyamoto, and M. Koizumi, ‘‘Synthesis of Silicon Nitride by a
Combustion Reaction Under High Nitrogen Pressure,’’ J. Am. Ceram. Soc., 64 [7]
C-60–1 (1986).
According to Eqs. (7)–(9), the phase stability diagram is con-
structed at T 5 1898 K, which is the maximum measured value
during the combustion reaction in air. Thus, the phase stability
diagrams are constructed as a function of nitrogen and oxygen
partial pressures, as shown in Fig. 9. The value of the lnP(O2) is
ꢀ1.56 in air (marked as area A), which leads to the formation of
Si2O2N and SiO2 in the surface products. As shown in Fig. 8, the
infiltration flux of N2/O2 decreased remarkably with an increase
in the diffusion distance. Because of the consumption at the
sample surface, the partial pressure of O2 decreased quickly to a
much lower value than that of N2 (marked as area B). Therefore,
the synthesis of single-phase Si3N4 in the sample interior is quite
favorable.
4Z. A. Munir and J. B. Holt, ‘‘The Combustion Synthesis of Refractory Nitrides
– Part I: Theoretical Analysis,’’ J. Mater. Sci., 22, 710–4 (1987).
5I. G. Cano and M. A. Rodriguez, ‘‘Synthesis of b-Si3N4 by SHS: Fiber
Growth,’’ Scr. Mater., 50, 383–6 (2004).
6H. B. Jin, Y. Yang, Y. X. Chen, Z. M. Lin, and J. T. Li, ‘‘Mechanochemical–
Activation–Assisted Combustion Synthesis of a-Si3N4,’’ J. Am. Ceram. Soc., 89 [3]
1099–102 (2006).
7A. Gromov and V. Vereshchagin, ‘‘Study of Aluminum Nitride Formation by
Superfine Aluminum Powder Combustion in Air,’’ J. Eur. Ceram. Soc., 24, 2879–
84 (2004).
8T. Tsuchida and T. Hasegawa, ‘‘TG–DTA–MS Study of Self-Ignition in Self-
Propagating High-Temperature Synthesis of Mechanically Activated Al–C Pow-
der Mixture,’’ Thermochim. Acta, 276, 123–9 (1996).
9G. H. Liu, K. X. Chen, H. P. Zhou, K. G. Ren, and H. B. Jin, ‘‘Dynamically
Controlled Formation of TiN by Combustion of Ti in Air,’’ J. Am. Ceram. Soc.,
90 [9] 2918–25 (2007).
10P. C. Gazzara and D. R. Messier, ‘‘Determination of Phase Content of Si3N4
by X-Ray Diffraction Analysis,’’ Am. Ceram. Bull., 56, 777–80 (1977).
11T. Jie, M. Ye, Y. C. Wu, and L. D. Zhang, ‘‘Fourier Transform Infrared
Spectroscopy and Raman Spectrum Analyses of Monocrystalline a-Si3N4 Nano-
wires,’’ J. Chin. Ceram. Soc., 36 [1] 44–8 (2008).
V. Conclusions
Nitriding CS of Si3N4 in air was realized through the selective
reaction of Si with N2. The a/b-Si3N4 ratio of the interior prod-
ucts could be easily tailored by adjusting the Si3N4-diluent con-
tent in the reactant mixtures, and nearly single-phase a-Si3N4
and b-Si3N4 powders were synthesized at the Si3N4-diluent con-
tents of 50 and 5 wt%, respectively. The as-synthesized b-Si3N4
showed a well-crystallized rod-like morphology.
12J. M. Qian, J. P. Wang, G. J. Qiao, and Z. H. Jin, ‘‘Preparation of Porous SiC
Ceramic with a Woodlike Microstructure by Sol–Gel and Carbothermal Reduc-
tion Processing,’’ J. Eur. Ceram. Soc., 24, 3251–9 (2004).
13W. K. Li, D. Y. Chen, B. L. Zhang, H. R. Zhuang, and W. L. Li, ‘‘Effect of
Rare-Earth Oxide Additives on the Morphology of Combustion Synthesized Rod-
Like b-Si3N4 Crystals,’’ Mater. Lett., 58, 2322–5 (2004).
The reactivity of Si powder was enhanced remarkably by MA
treatment. The surface Si powder exposed to air was prone to
oxidation with a mass of heat release, which mainly contributed
to the subsequent nitridation of Si powder in air. The discrep-
ancy in the penetrability of N2/O2 in the porous media is
responsible for the different phase assemblages formed at differ-
ent regions along the gas infiltration direction. Based on the
proposed N2/O2 diffusion kinetic model, the infiltration flux
ratio of N2/O2 to the sample interior is 6.99, which is responsible
for the selective reaction of Si powder with nitrogen in the
interior of the samples.
14D. Y. Chen, B. L. Zhang, H. R. Zhuang, and W. L. Li, ‘‘Combustion Syn-
thesis of Network Silicon Nitride Porous Ceramics,’’ Ceram. Int., 29, 363–4 (2003).
15T. D. Shen, C. C. Koch, T. L. Mclormick, R. L. Nemanich, J. Y. Huang, and
J. G. Huang, ‘‘The Structure and Property Characteristics of Amorphous/Nano-
crystalline Silicon Produced by Ball Milling,’’ J. Mater. Res., 10 [1] 139–48 (1995).
16T. Tsuchida, T. Hasegawa, T. Kitagawa, and M. Inagaki, ‘‘Aluminum Nitride
Synthesis in Air from Aluminum and Graphite Mixtures Mechanically Activated,’’
J. Eur. Ceram. Soc., 17, 1793–5 (2004).
17F. Xia, X. L. Qian, X. Yang, G. K. Liu, and X. Q. Sun, ‘‘Model on Gas Diffu-
sion in Porous Electrode in ZrO2 Oxygen Sensor,’’ J. Sens. Technol., 2, 123–8 (2001).
18F. Zhao, T. J. Armstrong, and A. V. Virkar, ‘‘Measurement of O2–N2 Effec-
tive Diffusivity in Porous Media at High Temperatures Using an Electrochemical
Cell,’’ J. Electrochem. Soc., 150 [3] A 249–56 (2003).
&