Thermal oxidation of aluminum nitride powder

Article abstract of DOI:10.1111/j.1551-2916.2006.01065.x

The oxidation kinetics, morphology, and crystallinity of aluminum nitride (AIN) powder thermally oxidized in flowing oxygen were determined from 800° to 1150° C. At 800°C the oxidation became detectable with weight change. AIN powder was almost completely oxidized at 1050°C after only 0.5 h. Amorphous aluminum oxide formed at relatively low temperatures (800°-1000°C), with a linear oxidation rate governed by the oxygen-nitride interfacial reaction. Transmission electron microscopy displayed individual aluminum oxide grains which formed a discontinuous oxide layer at this temperature range. The aluminum oxide was crystalline at higher temperatures (>1000°C), as detected by X-ray diffraction, and the density of oxide grains increased with temperature.

Full text of DOI:10.1111/j.1551-2916.2006.01065.x

J. Am. Ceram. Soc., 89 [7] 2167–2171 (2006)
DOI: 10.1111/j.1551-2916.2006.01065.x
r 2006 The American Ceramic Society
ournal
J
Thermal Oxidation of Aluminum Nitride Powder
w
Zheng Gu and James H. Edgar
Department of Chemical Engineering, Kansas State University, Manhattan, Kansas 66506
Chongmin Wang
Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, Washington 99352
Dorothy W. Coffey
High Temperature Materials Laboratory, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831
The oxidation kinetics, morphology, and crystallinity of alumi-
num nitride (AlN) powder thermally oxidized in flowing oxygen
were determined from 8001 to 11501C. At 8001C the oxidation
became detectable with weight change. AlN powder was almost
completely oxidized at 10501C after only 0.5 h. Amorphous
aluminum oxide formed at relatively low temperatures
produced a device-grade insulator suitable as gates for field ef-
fect transistors. The oxidized AlN has low defect densities com-
parable to device-grade SiO , and a higher dielectric constant
2
1
3–15
than SiO2.
polycrystalline AlN substrate is improved after oxidation.
In addition, the adhesion of metals on sintered
11
There have been several previous studies on the oxidation of
AlN, in terms of oxidation kinetics, oxidation products, the on-
set oxidation temperature, and oxidation of AlN powders with
(
8001–10001C), with a linear oxidation rate governed by the
oxygen–nitride interfacial reaction. Transmission electron
microscopy displayed individual aluminum oxide grains which
formed a discontinuous oxide layer at this temperature range.
The aluminum oxide was crystalline at higher temperatures
1
6
different particle size. Brown and Norton observed a linear
weight increase with time for AlN powder (981% pure, particle
size o10 mm with a range of shapes) oxidized in air at 8501C up
to 4 h. Above 8501C, the weight increased parabolically with
time. The activation energy for the reaction-controlled process
was 271 kJ/mol, and that for the diffusion-controlled process
was 423 kJ/mol, which agreed well with the experimental value
for the diffusion of oxygen through the polycrystalline Al O ,
(
410001C), as detected by X-ray diffraction, and the density
of oxide grains increased with temperature.
I. Introduction
2
3
459 kJ/mol. X-ray diffraction (XRD), scanning electron micros-
copy (SEM), and infrared (IR) spectroscopy were used by Ra-
LUMINUM nitride (AlN) finds enormous applications in elec-
tronics and optoelectronic devices. Its favorable properties
include good thermal stability (melting point 427501C), high
1
A
17
manathan et al. to study the oxidation behavior of AlN
powder (median particle size 5 mm, specific surface area 1.2
ꢁ
1
thermal conductivity (3.2 W(cmꢀ K) ) and electrical resistivity
2
8
m /g) at temperatures from 9001 to 11501C up to 20 h. At all
(
44 ꢂ 10 O/cm), and low dielectric constant (9 at 1 MHz) and
ꢁ6 ꢁ1 2–6
temperatures, an initial linear oxidation rate was followed by a
parabolic process, with activation energies of 197 and 270 kJ/
mol, respectively. The activation energies were much lower than
coefficient of thermal expansion (4.03–6.09ꢂ 10 ꢀ K ).
AlN ceramic substrates are preferred over traditional alumina
substrates, because their thermal conductivity is nine times high-
16
7
2
those observed by Brown and Norton. One reason was pos-
sibly due to the different particle sizes and impurities of the
original AlN powders. A poorly crystalline spinel-type alumi-
num oxynitride formed at 9001C, which transformed into a-alu-
mina crystallites at higher temperatures. The oxide morphology
changed from a fine-grained and coherent structure at 9501C to
er. AlN’s high piezoelectric constants (e33 5 1.5 C/m ,
2
e
31 5 ꢁ0.60 C/m ), which are up to 10 times larger than con-
ventional III–V and II–VI compounds, are the highest among
the tetrahedrally bonded semiconductors. Therefore, the spon-
taneous and piezoelectric polarization of AlN greatly influences
1
the electrical and optical properties of devices.
1
7
a porous fissured structure at 11501C.
The study of AlN oxidation is of great interest both theoret-
ically and practically. Oxygen is a common impurity in AlN,
which has a detrimental effect on its properties: it reduces the
lattice parameters of AlN and markedly decreases its thermal
There are several different structures of aluminum oxide, and
rhombohedral a-Al O (corundum) is the thermodynamically
2
3
stable form for the terminal products in AlN oxidation. Other
possible forms of Al O include monoclinic y-Al O and te-
8
–10
2
3
2
3
conductivity.
ters the AlN lattice is an important first step for reducing its
Therefore, understanding how the oxygen en-
2 3
tragonal d-Al O , normally called transition aluminas. Duch-
1
8
1
1
esne et al. studied the sequence of transition alumina formed
during the oxidation of AlN powder in air over the temperature
range from 8501 to 11501C. Diffuse reflectance FTIR confirmed
that AlN powder was oxidized via the reaction sequence
effects. However, in some cases the oxide layer on AlN is ben-
eficial. For example, a significant improvement in water resist-
ance at ambient conditions was achieved for AlN powder after
heating it in air above 8001C, due to the inhibition of the reac-
tion between AlN and water at room temperature by the formed
1
2
Al O film. The thermal oxidation of AlN thin films on Si
2
4AlN þ 3O2 ! 2ðd þ yÞAl2O3 þ 2N2
3
!
2a ꢁ Al2O3 þ 2N2
N. Jacobson—contributing editor
Epitaxial aluminum oxide layer can be obtained by the ther-
1
9
mal oxidation of epitaxial AlN films. Geng and Norton dis-
played that the oxide layer grew by the Stranki–Krastonow
mechanism, that is, the initial formation of a uniform oxide
layer was followed by island formation. Epitaxially oriented is-
lands of triangular outline, with the long edge aligned in the
½2110ꢃ direction, appeared on the AlN surface after oxidation at
Manuscript No. 21188. Received November 26, 2005; approved February 6, 2006.
Research is partly sponsored by the Assistant Secretary for Energy Efficiency and
Renewable Energy, Office of Freedom CAR and Vehicle Technologies, as part of the
High Temperature Materials Laboratory User Program, Oak Ridge National Laboratory,
managed by UT-Battelle, LLC, for the US Department of Energy.
w
Author to whom correspondence should be addressed. e-mail: guzheng@ksu.edu
2
167

2
168
Journal of the American Ceramic Society—Gu et al.
Vol. 89, No. 7
8
001C. 20%–30% of the surface was covered with these islands,
wt%) and oxygen were the main impurities. Other impurities
included Ca, Si, Fe, and Ni. The oxygen content was independ-
ently measured by LECO analysis at 0.9 wt%.
2
with the island density B15/mm . Between the islands was a thin
uniform oriented oxide layer, as confirmed by the moire fringe
pattern observed in an electron microscope. Kang et al.
demonstrated the oxidation of AlN/sapphire (0001) into epit-
axial g-Al upon oxidation. A porous g-Al , in the form of
nanoscale crystalline domains, was initially nucleated on the
surface. As the oxidation temperature or time increased, the
crystalline domain size increased, accompanied by a decreasing
oxide porosity.
´
2
0,21
also
The AlN powder (original weight 1–2 g) was oxidized in an
open-ended horizontal tube furnace regularly used for oxidizing
silicon. The reactor tube was quartz, with a maximum operation
temperature of 12001C. The sample was contained in an open
ceramic boat. The oxidation was performed over the temperature
range from 8001 to 11501C up to 6 h under 1 SLPM of pure
flowing oxygen gas. The oxygen was not purified or dessicated
before using. All oxidations were performed at 1 atm. When the
furnace reached the desired temperature, the ceramic boat con-
taining the AlN powder was inserted into the hot spot slowly (the
process took about 5 min). After 0.5, 2, 4, and 6 h of oxidation,
the sample was removed from the furnace for weight measure-
ment. Therefore, the whole oxidation process was interrupted.
O
2 3
2 3
O
Using reflection high-energy electron diffraction (RHEED),
22
Abid et al. found that AlN powder was resistant to oxidation in
air up to 10001C, which was higher than that observed by other
researchers, such as the 8501C using weight change by Brown
and Norton, the 7001C confirmed by both weight change and
X-ray photoelectron spectroscopy results by Katnani and Papa-
thomas, and the 7501C via calculating the oxide thickness from
the aluminum Auger electron spectroscopy spectra by Wang
et al. For AlN powder heated at 10001C, an unidentified phase
was detected by RHEED, possibly an oxide or oxynitride. At
16
23
24
(2) Characterization
The weight change of the AlN powder after oxidation was
measured to an accuracy of 71 mg using an electrobalance.
The weight increase was normalized relative to the initial unox-
idized AlN weight to account for variation in the sample
amount. The oxidation kinetics were derived from the relative
weight increase with time. Powder XRD with CuKa radiation
was employed to detect the crystalline oxidation products. The
overall oxygen and nitrogen contents in the resultant oxidized
AlN powder were measured by LECO analysis, a solid-state IR
and thermal conductivity detector, with accuracy and precision
to low ppm levels. The morphology of the AlN powder before
and after oxidation was determined by a Hitachi S-3500 (Hitachi
High Technologies America, Inc., Pleasonton, CA, USA) SEM
and a JEOL 2010 (JEOL-USA, Inc., Peabody, MA, USA) high-
resolution TEM. Both techniques can provide information on
the morphology, with TEM giving a higher magnification. The
TEM was equipped with an Oxford ISIS energy-dispersive
X-ray microanalysis (EDS) system for content measurement.
Before TEM measurement, the AlN powder was dispersed in
acetone, and then ultrasonically agitated for uniform distri-
bution onto the TEM sample mounts. The AlN powder was
sputter-coated with a thin Pt–Au layer to eliminate the charging
problem in SEM to obtain a good image.
1
2 3
2001C, AlN was completely transformed to a-Al O .
In our previous study, oxidation of highly textured (0001)
25
polycrystalline AlN and low defect density single crystalline AlN
was studied between 9001 and 11501C. The N face consistently
oxidized faster than the Al face. Grain boundaries and defects
were clearly decorated by oxidation. In comparison, single crys-
talline AlN produced a more uniformly thick oxide layer, as
expected with a single crystal. A crystallographic relationship
2 3
was established between the Al O and single crystalline AlN by
transmission electron microscopy (TEM).
The goal of the present study was to examine the role of the
oxide morphology in the oxidation of AlN powder, and com-
pare it with the previous results observed in the thermal oxida-
tion of polycrystalline and single crystalline AlN substrates. A
key consideration was if a continuous or discontinuous oxide
layer formed, and how it affected the oxidation kinetics. The
oxidation kinetics were determined by weight change and com-
pared with that observed in other AlN powders. The properties
of the aluminum oxide formed, and the morphology before and
after oxidation were studied in detail for additional insights into
the oxidation process.
II. Experimental Procedure
III. Results
(1) Oxidation of AlN Powder
The AlN powder employed in this study had an average
agglomerated particle diameter of 1.8 mm with a narrow size
distribution curve, according to the vendor data. Carbon (0.06
SEM and TEM images for the original AlN powders are shown
in Fig. 1. AlN particles were nearly spherical, with a few facets.
The particle size (0.5–1 mm) was somewhat smaller than that
Fig. 1. (a) Scanning electron microscopy and (b) transmission electron microscopy images for as-received aluminum nitride powder before oxidation.

July 2006
Thermal Oxidation of Aluminum Nitride Powder
2169
reported by the vendor (1.8 mm), probably due to the sample
preparation procedures used in SEM and TEM.
The AlN powder was almost completely oxidized (495% con-
version) at 10501C after only 0.5 h, and the oxidation extent
remained almost constant at higher temperatures or longer time.
The majority of the AlN particles were covered with individ-
ual oxide grains after oxidation at 11501C for 6 h, as shown in
the TEM images in Fig. 3, and they tended to merge together.
EDS detected only Al and O on these features. In the SEM im-
ages in Fig. 3, some voids seemed to appear on the AlN particle
surface, which were probably caused by the out-diffusion of the
product gas through the oxide layer. After oxidation at 11001C
for 6 h, the AlN powder consisted of 0.26 wt% of N and 43.5
wt% of O according to the LECO analysis. If the remaining
Oxidation under the same condition was repeated several
times, and the results were highly reproducible. Assuming all the
nitrogen formed in the oxidation dissipates into the gas phase,
the complete oxidation of AlN should result in a 24.4% increase
in weight. The AlN powder oxidized at 8001C for 6 h had a
small weight change (0.5%), and no oxide-related XRD peaks
were detected. It either suggested little oxidation occurred, or a
thin aluminum oxynitride (oxygen might be substitutionally in-
corporated into AlN) or amorphous aluminum oxide layer
formed at this temperature, which were difficult for XRD to
identify. AlN powder was stable up to 8001C, in the same tem-
perature range as previous results measured by weight change,
2 3
56.24 wt% was Al, then in addition to Al O , other aluminum
oxide or aluminum oxynitride possibly formed (which was be-
yond the characterization capabilities in this paper), consistent
with the previous studies by Ramanathan, Duchesne, and
1
6
such as 8501C by Brown and Norton and 7001C by Katnani
2
3
17
18
and Papathomas. At 9001C, still no oxidation products were
observed by XRD. However, the relative weight change in-
creased by 1.6% after 6 h. Considering an AlN particle (density
2
2
Abid.
The mass increase was linear at 8001–10001C, as seen in
Fig. 4, with an activation energy of 157 kJ/mol. This value is
3
3
.255 g/cm ) with a 1.8 mm diameter (the average particle size for
1
lower than 271 kJ/mol calculated by Brown and 197 kJ/mol by
6
AlN powder), a 1.6% weight increase should result in a 41.3-
3
layer (density 3.987 g/cm ). It was thinner than
17
Ramanathan, which is most likely caused by the different par-
nm-thick Al O
2 3
the oxide layer we previously observed on a flat polycrystalline
AlN wafer under the same oxidation condition (136.4 and 176.2
nm on the Al-face and N-face, respectively).
ticle size and impurity levels of the original AlN powder, and the
different oxidant (oxygen in this study versus air in most other
research). Brown used a 981% pure AlN powder with the
2
5
16
New features, some of which were hexagonal shaped, were
observed on the AlN particles as shown in the TEM images in
Fig. 2, with the detection of mainly Al and O, and traces of N by
particle size o10 mm and a range of shapes. The AlN powder
1
that Ramanathan used was a pure powder of hexagonal phase
7
AlN with a median particle size of 5 mm and a specific surface
area of 1.2 m /g. AlN is susceptible to moisture. The AlN ox-
2
EDS. They were probably individual Al
O
2 3
grains formed on
the AlN particles, as the formation of a very thin uniform oxide
layer on spherical particles was less likely than the formation of
individual Al O crystallites. This was different from the results
idation rate is more than 10 times higher in wet air than in dry
air, possibly because the diffusion rate of water through alumi-
2
6
num oxide is much higher than that of oxygen. Untreated ox-
ygen was used in the experiment, which possibly contained a
high amount of humidity. The linear kinetics was consistent with
the observation of individual Al O grains on the AlN particles,
2
3
19
by Geng and Norton, who observed an initial formation of a
thin uniform oxide layer, and a subsequent nucleation and
growth of islands on AlN substrates. The main difference was
the shape of the AlN employed: ours were nearly spherical AlN
powders, which were more irregular than their AlN substrates.
Further characterization by the LECO analysis demonstrated a
bulk composition of 31.2 wt% N and 4.8 wt% O, which was
consistent with the measured 1.6% relative weight increase. As-
suming the remaining component to be Al, the oxidized powder
2
3
as the process was controlled by the surface reaction between
oxygen and exposed AlN.
In our previous study on the oxidation of highly textured
2
5
AlN, we demonstrated that on the (0001) plane, the oxidation
rate was dependent on crystal polarity: the nitrogen polar plane
oxidized more rapidly than the aluminum polar plane. Different
orientations resulted in different oxidation rates, as evident from
the grain boundaries highlighted by different colors due to the
formation of oxide layers with different thickness. Furthermore,
the oxidation was faster near dislocation and other crystalline
defects. In comparison, in the case of nearly spherical AlN par-
ticles employed in the present study, a wide variety of crystal
planes are exposed to the oxidizing ambient, each of which may
2 3
was composed mainly of AlN with a small amount of Al O .
The weight change became significant when the oxidation
temperature was higher than 9001C. It reached 8.4% after ox-
2 3
idation at 10001C for 6 h, which equaled a 243.2-nm-thick Al O
layer, again assuming AlN particles with a 1.8 mm diameter. It
was comparable with that observed on polycrystalline AlN
2
5
(
268.6 and 308.9 nm on the Al face and N face, respectively).
2 3
Fig. 2. (a, b) Transmission electron microscopy images showing individual Al O grains (labeled by an arrow) formed on aluminum nitride powder
after oxidation at 9001C for 6 h.

2
170
Journal of the American Ceramic Society—Gu et al.
Vol. 89, No. 7
Fig. 3. (a, b) Scanning electron microscopy images showing voids (labeled by an arrow) and (c, d) transmission electron microscopy images showing
individual Al grains (labeled by an arrow) formed on aluminum nitride powder after oxidation at 11501C for 6 h.
2 3
O
oxidize at its own distinct rate. For example, the oxidation rate
of Si varies by a factor of 1.68 between the (100) and (111)
2
7
planes. From these observations and our experimental results,
we conclude that the oxidation of AlN powder does indeed pro-
ceed with the formation of a non-uniform oxide, as its growth is
mediated by a variety of exposed crystal planes. Nevertheless,
the linear rate expression can adequately model the oxidation
rate at 8001–10001C, as evident from Fig. 4, and the oxidation
reached almost completion rapidly at higher temperatures
1050 °C/6 h
(
10501C for 0.5 h).
No XRD peaks for aluminum oxide were detected until the
oxidation temperature reached 10501C, as shown in Fig. 5. The
1
0
8
6
4
2
0
1000 °C
1
000 °C/6 h
9
8
00 °C
00 °C
0
2
4
6
Oxidation Time (h)
20
30
40
50
60
70
80
90
Fig. 4. Relation between the percentage of relative weight increase and
oxidation time for aluminum nitride powder at temperatures from 8001
to 10001C.
Fig. 5. X-ray diffraction patterns for aluminum nitride powder
oxidized at 10001 and 10501C for 6 h.

July 2006
Thermal Oxidation of Aluminum Nitride Powder
2171
5
S. Strite and H. Morkoc, ‘‘Gallium Nitride, Aluminum Nitride and Indium
Nitride: A Review,’’ J. Vac. Sci. Technol. B, 10 [4] 1237–66 (1992).
only XRD peaks detected at 10001C can be attributed to AlN.
The distribution and intensities correspond to a completely ran-
dom AlN powder. The XRD patterns under 10501C indicate a
randomly oriented rhombohedral alumina layer, with all the
main peaks expected for this oxide clearly observed. In addition,
the intensities of all the peaks were consistent with those report-
ed in the JCPDS diffraction data, suggesting the formation of a
completely random aluminum oxide. Alumina peaks only ap-
peared at or above 10501C probably because, at lower temper-
atures, only amorphous alumina or intermediate species formed,
which were not sufficiently crystalline or too thin to be identified
by XRD. Thicker crystalline Al O formed at higher tempera-
6
S. Krukowski, M. Leszczynski, and M. Porowski, ‘‘Thermal Properties of the
Group III Nitrides,’’ EMIS Datarev. Series, 23, 21–8 (1999).
7
L. M. Sheppard, ‘‘Aluminum Nitride: A Versatile but Challenging Material,’’
Ceram. Bull., 69 [11] 1801–12 (1990).
8
J. H. Harris, R. A. Youngman, and R. G. Teller, ‘‘On the Nature of the Ox-
ygen-Related Defect in Aluminum Nitride,’’ J. Mater. Res., 5 [8] 1763–7 (1990).
9
G. A. Slack, L. J. Schowalter, D. Morelli, and J. A. Freitas Jr., ‘‘Some Effects
of Oxygen Impurities on AlN and GaN,’’ J. Cryst. Growth, 246 [3–4] 287–98
(
2002).
10
G. A. Slack and T. F. McNelly, ‘‘Growth of High-Purity AlN Crystals,’’
J. Crystal Growth, 34 [2] 263–79 (1976).
11
R. Yue, Y. Wang, Y. Wang, and C. Chen, ‘‘SIMS Study on the Initial Ox-
idation Process of AlN Ceramic Substrate in the Air,’’ Appl. Surf. Sci., 148 [1–2]
2
3
7
3–8 (1999).
12
tures. Thus, weight measurement is a more sensitive method
than XRD in detecting the onset of oxidation for the AlN pow-
der. A similar phenomenon was observed in the thermal oxida-
tion of single crystalline AlN. TEM results confirmed the
formation of amorphous oxide at 8001C and crystalline oxide
Y. Q. Li, T. Qiu, and J. Xu, ‘‘Effect of Thermal Oxidation Treatment in Air on
the Hydrolysis of AlN Powder,’’ Mater. Res. Bull., 32 [9] 1173–9 (1997).
J. Kolodzey, E. A. Chowdhury, G. Qui, J. Olowolafe, C. P. Swann, K. M.
13
Unruh, J. Suehle, R. G. Wilson, and J. M. Zavada, ‘‘The Effects of Oxidation
Temperature on the Capacitance–Voltage Characteristics of Oxidized AlN Films
on Si,’’ Appl. Phys. Lett., 71 [26] 3802–4 (1997).
2
8
at a higher temperature (10001C).
14
E. A. Chowdhury, M. Dashiell, G. Qiu, J. O. Olowolafe, R. Jonczyk, D.
Smith, A. Barnett, J. Kolodzey, K. M. Unruh, C. P. Swann, J. Suehle, and Y.
Chen, ‘‘Structural, Optical and Electronic Properties of Oxidized AlN Thin Films
at Different Temperatures,’’ J. Electron. Mater., 27 [7] 918–22 (1998).
IV. Conclusions
15
E. A. Chowdhury, J. Kolodzey, J. O. Olowolafe, G. Qiu, G. Katulka, D. Hits,
AlN powders were oxidized in flowing untreated oxygen over
the temperature range from 8001 to 11501C up to 6 h. Individual
M. Dashiell, D. van der Weide, C. P. Swann, and K. M. Unruh, ‘‘Thermally Ox-
idized AlN Thin Films for Device Insulators,’’ Appl. Phys. Lett., 70 [20] 2732–4
(1997).
amorphous Al
0001C, which changed to crystalline at higher temperatures.
The weight changed linearly with time between 8001 and
0001C, with an activation energy of 157 kJ/mol. At tempera-
2 3
O grains formed on AlN particles within 8001–
16
1
A. L. Brown and M. G. Norton, ‘‘Oxidation Kinetics of AlN Powder,’’
J. Mater. Sci. Lett., 17 [18] 1519–22 (1998).
17
S. Ramanathan, R. Bhat, D. D. Upadhyaya, and S. K. Roy, ‘‘Oxidation
Behavior of Aluminum Nitride Powder,’’ Br. Ceram. Trans., 94 [2] 74–8 (1995).
1
18
tures higher than 10001C (10501–11501C), 495% AlN was ox-
idized after only 0.5 h, and the conversion barely changed with
the further increase of temperature or time. The oxidation ki-
netics can be adequately explained by the morphology of the
formed aluminum oxide. Individual Al O grains left enough
D. J. Duchesne, K. W. Hipps, B. A. Grasher, and M. G. Norton, ‘‘The For-
mation of Transition Aluminas During Oxidation of AlN,’’ J. Mater. Sci. Lett., 18
11] 877–9 (1999).
[
19
Y. Geng and M. G. Norton, ‘‘Early Stages of Oxidation of Aluminum
Nitride,’’ J. Mater. Res., 14 [7] 2708–11 (1999).
20
2
3
H. C. Kang, S. H. Seo, and D. Y. Noh, ‘‘Synchrotron X-Ray Scattering Study
on Oxidation of AlN/Sapphire,’’ Mater. Res. Soc. Symp. Proc., 590, 195–9 (2000).
surfaces for further oxidation, leading to an interface reaction-
controlled process. Owing to the small size of AlN particles, it
quickly reached complete oxidation at high temperatures
410001C). The wide variety of crystal planes present on AlN
powders resulted in the initial formation of non-uniform oxide.
21
H. C. Kang, S. H. Seo, H. W. Jang, D. H. Kim, J. W. Kim, and D. Y. Noh,
‘Synthesis of Epitaxial g-Al Thin Films by Thermal Oxidation of AlN/Sap-
phire(0001) Thin Films,’’ Appl. Phys. A, 77 [5] 627–32 (2003).
‘
2 3
O
22
(
A. Abid, R. Bensalem, and B. J. Sealy, ‘‘The Thermal Stability of AlN,’’
J. Mater. Sci., 21 [4] 1301–4 (1986).
A. D. Katnani and K. I. Papathomas, ‘‘Kinetics and Initial Stages of Oxida-
23
tion of Aluminum Nitride: Thermogravimetric Analysis and X-Ray Photoelectron
Spectroscopy Study,’’ J. Vac. Sci. Technol. A., 5 [4] 1335–40 (1987).
P. S. Wang, S. G. Malghan, S. M. Hsu, and T. N. Wittberg, ‘‘The Oxidation of
Acknowledgments
24
The TEM work was performed at the EMSL, a national scientific user facility
sponsored by DOE’s Office of Biological and Environmental Research and located
at Pacific Northwest National Laboratory, operated for DOE by Battelle. Support
from the National Science Foundation via award DMR-0408874 is greatly
appreciated.
An Aluminum Nitride Powder Studied by Bremsstrahlung-excited Auger Electron
Spectroscopy and X-Ray Photoelectron Spectroscopy,’’ J. Mater. Res., 10 [2]
302–5 (1995).
Z. Gu, J. H. Edgar, S. A. Speakman, D. Blom, J. Perrin, and J. Chaudhuri,
‘‘Thermal Oxidation of Polycrystalline and Single Crystalline Aluminum Nitride
25
Wafers,’’ J. Electron. Mater., 34 [10] 1271–9 (2005).
H.-E. Kim and A. J. Moorhead, ‘‘Oxidation Behavior and Flexural Strength
26
of Aluminum Nitride Exposed to Air at Elevated Temperatures,’’ J. Am. Ceram.
Soc., 77 [4] 1037–41 (1994).
J. D. Plummer, M. D. Deal, and P. B. Griffin, ‘‘Thermal Oxidation and the Si/
References
1
27
O. Ambacher, ‘‘Growth and Applications of Group III-Nitrides,’’ J. Phys. D:
Appl. Phys., 31 [20] 2653–710 (1998).
B. Luo, J. W. Johnson, O. Kryliouk, F. Ren, S. J. Pearton, S. N. G. Chu, A. E.
Nikolaev, Y. V. Melnik, V. A. Dmitriev, and T. J. Anderson, ‘‘High Breakdown
M–I–M Structures on Bulk AlN,’’ Solid State Electron., 46 [4] 573–6 (2002).
G. A. Slack, ‘‘Nonmetallic Crystals with High Thermal Conductivity,’’ J. Phys.
Chem. Solids, 34 [2] 321–35 (1973).
D. K. Gaskill, L. B. Rowland, and K. Doverspike, ‘‘Electrical Transport Prop-
erties of AlN, GaN and AlGaN,’’ EMIS Datarev. Series, 11, 101–16 (1994).
SiO2 Interface’’; pp. 287–364 in Silicon VLSI Technology Fundamentals, Practice
and Modeling, Edited by J. D. Plummer, M. D. Deal, and P. B. Griffin. Prentice
Hall, New Jersey, 2000.
2
28
J. Chaudhuri, N. Nyakiti, R. G. Lee, Z. Gu, J. H. Edgar, and J. G. Wen,
‘‘Thermal Oxidation of Single Crystalline Aluminum Nitride,’’ Mater. Character-
ization, in press.
3
&
4