A study of the impact of gate metals on the performance of AlGaN/AlN/GaN

heterostructure field-effect transistors

Jingtao Zhao, Zhaojun Lin, Quanyou Chen, Ming Yang, Peng Cui, Yuanjie Lv, and Zhihong Feng
Citation: Applied Physics Letters 107, 113502 (2015); doi: 10.1063/1.4931122
View online: http://dx.doi.org/10.1063/1.4931122
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/107/11?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Effects of rapid thermal annealing on the electrical properties of the AlGaN/AlN/GaN heterostructure field-effect
transistors with Ti/Al/Ni/Au gate electrodes
Appl. Phys. Lett. 105, 083501 (2014); 10.1063/1.4894093
Influence of the side-Ohmic contact processing on the polarization Coulomb field scattering in AlGaN/AlN/GaN
heterostructure field-effect transistors
Appl. Phys. Lett. 101, 113501 (2012); 10.1063/1.4752232
Polarization Coulomb field scattering in AlGaN/AlN/GaN heterostructure field-effect transistors
Appl. Phys. Lett. 98, 123512 (2011); 10.1063/1.3569138
Investigation of Ta Ti Al Ni Au ohmic contact to Al Ga N Ga N heterostructure field-effect transistora)
J. Vac. Sci. Technol. B 23, 322 (2005); 10.1116/1.1856479
Growth of high-performance GaN modulation-doped field-effect transistors by ammonia-molecular-beam epitaxy
J. Vac. Sci. Technol. A 18, 652 (2000); 10.1116/1.582255

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
197.207.21.231 On: Sat, 19 Sep 2015 10:11:53

APPLIED PHYSICS LETTERS 107, 113502 (2015)

A study of the impact of gate metals on the performance of AlGaN/AlN/GaN

heterostructure field-effect transistors
Jingtao Zhao,1 Zhaojun Lin,1,a) Quanyou Chen,1 Ming Yang,1 Peng Cui,1 Yuanjie Lv,2
and Zhihong Feng2
1

School of Physics, Shandong University, Jinan 250100, China

National Key Laboratory of Application Specific Integrated Circuit (ASIC), Hebei Semiconductor Research
Institute, Shijiazhuang 050051, China
2

(Received 4 May 2015; accepted 4 September 2015; published online 15 September 2015)
The fabrication processes of heterostructure field-effect transistors (HFETs) often introduce
stresses into the device structures. These stresses can significantly affect the electrical characteristics of the device. We investigated the additional strain induced by the gate metals (in this study,
Au, Cu, Fe, Al, and Ni) in AlGaN/AlN/GaN HFETs using the measured capacitance-voltage and
current-voltage characteristics. We found that the intensity of the additional strain of the AlGaN
barrier layer is greatly affected by the types of gate metals. For those metals that mainly undergo a
physical interaction with the AlGaN barrier layer, there is a negative correlation between the intensity of the additional strain and the Youngs modulus of the gate metal. Therefore, the gate metal
with a larger Youngs modulus will be more favorable for weakening polarization Coulomb field
scattering and increasing both the carrier mobility and the sheet carrier density in AlGaN/AlN/GaN
C 2015 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4931122]
HFETs. V

The etching or deposition of films during the fabrication

of heterostructure field-effect transistors (HFETs) often leads
to the introduction of stresses into the device structure. These
stresses can significantly affect the electrical characteristics
of the device.1 Stress has been intentionally designed into
many modern Si CMOS transistors, as process-induced
stresses can produce a major enhancement in mobility and
channel current.2,3
In GaN-based HFETs, the strain of the epilayers is crucial to the material characteristics and the performance of
the devices, by altering the sheet carrier density and the electron mobility of the two-dimensional electron gas (2DEG).4,5
There are many studies focusing on the strain caused by the
deposition of passivation layer,6,7 Ohmic contacts,810 and
electric field.11,12 However, the Schottky contact metals, as
the gate electrodes of the AlGaN/AlN/GaN HFETs, can also
directly determine the device performance and long term
reliability behavior.1316 Few literatures about the strain
induced by the gate metals, especially about the comparison
for different types of gate metals, have been reported.
In this work, the five different types of gate metals (Au,
Cu, Fe, Al, and Ni) were deposited on strained AlGaN/AlN/
GaN heterostructures. With the measured capacitancevoltage (C-V) curves, the current-voltage (I-V) characteristics, and the polarization effects of the AlGaN/AlN/GaN
HFETs, we have analyzed the influence of the gate metals on
the strain of the AlGaN barrier layer in AlGaN/AlN/GaN
HFETs.
The epilayers of AlGaN/AlN/GaN heterostructures
employed herein were grown by metal-organic chemical
vapor deposition (MOCVD) on a (0001) sapphire substrate.
The epilayers consist of, from bottom to top, 40 nm AlN
nucleation layer on the substrate, 2 lm undoped GaN, 0.5 nm
a)

Electronic mail: [email protected].

0003-6951/2015/107(11)/113502/5/$30.00

AlN interlayer, and around 19 nm undoped AlGaN barrier

layer with Al mole fraction of 24%. The Hall effect measurement on the heterostructure showed a 2DEG sheet density of
around 1.07  1013 cm2 with an electron mobility of
2040 cm2/V s at room temperature. Device process began
with device isolation utilizing Cl2/BCl3 inductively coupled
plasma (ICP) etching, followed by the source and drain
Ohmic contacts formation using a rapid thermal anneal
(RTA) of Ti/Al/Ni/Au metal stacks in an N2 ambient, then
deposited different types of gate metals by e-beam evaporation, which would be discussed in detail below. To eliminate
the interferences from other factors except gate metals, such
as Ohmic contact processing, we designed a relatively large
device structure. The source and drain regions were rectangular with 50 lm long and 100 lm wide, and the distance in
between is 100 lm. The gates are rectangular with 20 lm
long and 100 lm wide. One thing to point out about the device process is that the sample was not cut into pieces until
gate metal evaporation. To get clean surface, the samples
were treated by O2 plasma ashing for 2 min and diluted HCl
(HCl:H2O 1:1) for 1 min, followed by rinsing in deionized
(DI) water and blow drying in N2. To investigate the influence of the types of gate metals on strain in AlGaN/AlN/
GaN HFETs, the five types of metal stacks had been chosen
for the Schottky contact metal candidates, which are Au
(200 nm), Cu/Au (50 nm/150 nm), Fe/Au (50 nm/150 nm),
Al/Au (50 nm/150 nm), and Ni/Au (50 nm/50 nm, 50 nm/
150 nm, and 50 nm/300 nm), respectively. In this study,
150 nm Au was deposited on the gate metals for all the devices to eliminate the interferences from the additional strain
induced by the oxidation of the gate metals. During metal
evaporation, the temperature at the sample is kept at 80  C.
/s.
Each metal was deposited at a deposition rate of 23 A
4
The vacuum degree was about 2  10 Pa, and the evaporation distance was about 50 cm.

107, 113502-1

C 2015 AIP Publishing LLC

V

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
197.207.21.231 On: Sat, 19 Sep 2015 10:11:53

113502-2

Zhao et al.

To investigate the electrical characteristics, the C-V

measurements were carried out using an Agilent B1520A at
1 MHz, and the I-V measurements were carried out using an
Agilent B1500A semiconductor parameter analyzer at room
temperature. The transmission line method (TLM) was
adopted to investigate the Ohmic contact quality, and the
specific resistivity of Ohmic contacts was measured to be
1.3  105 X cm2.
Fig. 1(a) shows the C-V curves of the AlGaN/AlN/GaN
HFETs with the different gate metals. The 2DEG density can
be obtained from the C-V curves.17 Fig. 1(b) shows the
obtained 2DEG density under different gate biases by integrating the measured C-V curves. Table I summarizes the
parameters of the AlGaN/AlN/GaN HFETs with respect to
the different Schottky metals, in which VT is the threshold
voltage obtained by the C-V curve differentiation,17 n2D0
represents the 2DEG sheet carrier density under the Schottky
contact at zero gate bias, and ub is the evaluated Schottky
barrier height with the flat-band voltage from the forward
I-V characteristics.18 It can be readily observed that the barrier height for the different gate metals is different,19 and
there is a significant difference in the 2DEG density between
the devices with the different gate metals. The variation for
the 2DEG density is as large as 1.5  1012 cm2. Previous
studies have shown that the 2DEG in GaN-based heterostructures is induced by the spontaneous and piezoelectric polarization, and the interface charge density is sensitive to a large
number of different physical properties such as polarity, alloy

FIG. 1. The measured C-V curves (a) and the calculated 2DEG electron density n2D (b) under different gate biases for the AlGaN/AlN/GaN HFETs with
different types of the gate metals at room temperature.

Appl. Phys. Lett. 107, 113502 (2015)

TABLE I. Parameters of the AlGaN/AlN/GaN HFETs with the different
Schottky metals. UM is the work function of each metal, VT is the threshold
voltage, n2D0 is the 2DEG sheet carrier density under the Schottky contact
metals at zero gate bias, and ub is Schottky barrier height.
Metal
Al
Au
Cu
Ni
Fe

UM (eV)

VT (V)

n2D0 (1012 cm2)

ub (eV)

4.28
5.10
4.65
5.15
4.50

4.60
2.85
3.10
3.20
3.20

6.18
6.24
7.26
7.35
7.60

1.02
1.21
1.38
1.20
1.22

composition, barrier thickness, doping, and strain of the barrier layer.20,21 In this study, the mechanisms that affect the
2DEG densities should be only related to the Schottky contact processing, since the AlGaN/AlN/GaN heterostructure
material, device isolation, Ohmic contact processing, and the
device structure are exactly the same. As the surface of the
AlGaN barrier layer is significantly rough after being treated
by the O2 plasma ashing and diluted HCl, which allows
metal atoms to enter the AlGaN lattice structure, sometimes
even in the interstitial position,22 we believe that both chemical reaction and physical interaction probably exist at the
gate metal/AlGaN interface in or after the Schottky contact
processing.
For the devices with Al/Au gate metal, the previous
research shows that the Al/AlGaN interface undergoes a GaAl exchange chemical reaction driven by the large heat of
formation of AlN as compared to that of GaN at room temperature.23 With the Schottky contact metals removed, the
scanning electron microscope with energy dispersive spectrometer (SEM-EDS) measurements were performed under
the gate at room temperature. The experimental results are
consistent with the conclusion in Ref. 23. The detailed experimental results and discussion are given in supplementary
material.24
As shown in Fig. 1 and Table I, the 2DEG sheet carrier
density for Al/Au gate metals is lower than that of other
devices at zero gate bias. We deduce that the possible reason
for the lower 2DEG sheet carrier density is the chemical
reaction at the Al/AlGaN interface leads to a thinner AlGaN
barrier layer and affects the parameters of the surface donor
states of the AlGaN barrier layer, such as the surface electronic state density and energy level. While the chemical
reaction product as an intermediate at the Al/AlGaN interface increases the distance between the gate metal and the
2DEG channel, which causes weaker control ability of the
Schottky contact to the 2DEG and results in a high-dose negative shift of the threshold voltage. The above explanations
for the Al/Au Schottky gate device are tentative, and the further study is needed to understand the related phenomena.
For the other four devices, the threshold voltages are
around 3 V as listed in Table I. Studies have shown that
these metals would undergo less chemical reactions with the
AlGaN barrier layer compared to Al described above.2126
Therefore, the physical interactions should be the main cause
of the additional strain, whose intensity, as a result, would be
mainly determined by the physical properties of the gate
metals.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
197.207.21.231 On: Sat, 19 Sep 2015 10:11:53

113502-3

Zhao et al.

However, it should be noted that the direct measurement

of the strain in the AlGaN layer is almost impossible due to
its thickness (on the order of 20 nm).27 Previous studies have
proved that the polarization coulomb field (PCF) scattering
is one of the most important scattering mechanisms in GaNbased HFETs, and it has a very important influence on the
carrier mobility of the devices.17,2831 As the PCF scattering
is related to the strain variation of the AlGaN barrier layer,
its intensity can reflect the situation of the additional strain
of the barrier layer induced by the gate metals. The stronger
the intensity of the PCF scattering is, the larger the additional
strain is.
The I-V output characteristics of the AlGaN/AlN/GaN
HFETs for different types of the gate metals are shown in
Fig. 2, in which the values of the measured current with a
source-drain voltage of 0.1 V at different gate biases can be
obtained for calculating the carrier mobility. Using the same
method described in Ref. 28, we determined the electron mobility (l) of the 2DEG under different gate biases in the
AlGaN/AlN/GaN HFETs, as shown in Fig. 3. Previous studies
have demonstrated that the faster the electron mobility of the
2DEG increases with gate biases, the stronger the intensity of
the PCF scattering is.2831 Thus, we can see the intensity of
the PCF scattering (IP) varies greatly for the devices with different types of the gate metals, and IPAu > IPCu > IPNi > IPFe.
This means that the intensity of the additional strain to the

Appl. Phys. Lett. 107, 113502 (2015)

FIG. 3. The relationship between the electron mobility of the 2DEG and the
applied gate bias for the AlGaN/AlN/GaN HFETs with different types of the
gate metals at room temperature.

AlGaN barrier layer varies greatly due to different types of

the gate metals.
To describe the variation of the additional strain quantitatively, with the measured electrical characteristics and the
polarization effects of the HFETs, we estimated the average
in-plane strain (e (a  a0)/a0) of the AlGaN barrier layer
under the gate for the AlGaN/AlN/GaN HFETs (the calculating method is described in detail in Ref. 32), as listed in
Table II. Compared with the value of 1.07  1013 cm2,
which is the 2DEG sheet density of the Hall measurement
for the heterojunction material, the 2DEG sheet density
under the Schottky contact for the different gate metals is
much lower. So, the additional strain of the AlGaN barrier
layer induced by the gate metals is compressive strain, which
makes the polarization and the in-plane strain of the barrier
layer weaker. The calculated result (eAu < eCu < eNi < eFe) is
consistent with the electron mobility results reflected by the
PCF scattering.
Since the mass of the Schottky contact metals with the
same thickness is different, to determine the relationship
between the additional strain of the AlGaN barrier layer with
the mass of the gate metals, we prepared three Ni/Au
Schottky contact devices with same structure but different
thicknesses of Au layers. The detailed experimental results
are given in supplementary material,24 from which it can be
seen that the 2DEG sheet carrier density and the intensity of
the PCF scattering are little affected by the mass of the gate

TABLE II. Calculated parameters and Youngs modulus of the AlGaN/AlN/

GaN HFETs with the different Schottky metals. rtotal is the total polarization
sheet charge density, e: e (a  a0)/a0 is the in-plane strain, a is the strained
a-axis lattice constant of the AlGaN barrier layer, and a0 is the equilibrium
value of the lattice parameter. CTE: Coefficient of thermal expansion of
each metal and E: Youngs modulus of each metal.
Metal

FIG. 2. The output I-V characteristics of the AlGaN/AlN/GaN HFETs with

different types of the gate metals at room temperature.

rtotal (1012 cm2)

e (%)
CTE (106 K1)
E (GPa)
CTE/E (106 K1 GPa1)

Au

Cu

Ni

Fe

7.25
0.184
14.2
78.5
0.181

7.72
0.234
16.5
129.8
0.127

8.05
0.270
13.4
199.5
0.067

8.38
0.305
11.8
211.4
0.056

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
197.207.21.231 On: Sat, 19 Sep 2015 10:11:53

113502-4

Zhao et al.

FIG. 4. Schematic drawing of the polarization charge distribution and the

stress induced by the gate metals.

metals. According to the schematic diagram of the polarization charge distribution and the stress induced by the gate
metals in Fig. 4, compressive strain of the AlGaN barrier
layer can be caused by uniaxial and biaxial stress induced by
the gate metals. As the mass of the gate metals has little
effect on the device performance, biaxial stress would be the
main factor of the additional strain of the AlGaN barrier
layer, and it is closely related to the types of gate metals.
Based on the analysis above, the intensity of additional
strain is mainly related to the physical properties of the gate
metals. The compressive stress of the AlGaN barrier layer
induced by the gate metals is probably formed as follows:
the gate metal is deposited at a high temperature, and as it
cools, it will undergo residual tensile stress while the AlGaN
layer beneath would be under compression. The additional
strain caused by the gate metal is a combination of both the
coefficient of thermal expansion (CTE) mismatch and the
Youngs modulus mismatch. Youngs modulus predicts how
much a material sample extends under tension or shortens
under compression,33 and it is a physical quantity associated
with the strain of the multilayer materials according to
earlier studies.3436 During the gate metal (Fe, Cu, Ni, and
Au) deposition, there are the physical interactions between
the gate metal and the AlGaN barrier layer. The gate metal
with a larger Youngs modulus is corresponding to a smaller
deformation, and exerts a small force to the AlGaN barrier
layer, which results in the additional strain of the AlGaN barrier layer being small too. Table II shows that the maximum
discrepancy among the CTE of the different Schottky metals
is 39.8%, which is much less than the maximum discrepancy
of 169.3% for the Youngs modulus of those Schottky metals, and the experimental results show a negative correlation
between the additional strain of the AlGaN barrier layer and
the Youngs modulus of the gate metal. We deduce that
Youngs modulus is the dominant factor that determines the
additional strain. The larger the Youngs modulus of the gate
metal is, the weaker the additional strain to the AlGaN barrier layer is. As a result, the gate metal with a larger Youngs
modulus will induce higher both the 2DEG electron density
and the electron mobility, which results in the larger saturated output channel current. It is confirmed from the measured output I-V characteristics of Fig. 2 that the carrier
mobility curves of Fig. 3 are also consistent with the additional strain induced by the gate metals.
In summary, we have investigated the additional strain
induced by the gate metals in the AlGaN/AlN/GaN HFETs

Appl. Phys. Lett. 107, 113502 (2015)

by the measured electrical characteristics. It was found that

biaxial stress associated with the types of gate metals is the
main source of the additional strain. For those metals that
mainly undergo a physical interaction with the AlGaN barrier layer, there is a negative correlation between the intensity of the additional strain and the Youngs modulus of the
gate metal. This finding is useful for AlGaN/AlN/GaN
HFETs to weaken PCF scattering and increase both the carrier mobility and the 2DEG electron density by optimizing
the gate metal.
This work was supported by the National Natural
Science Foundation of China (Grant Nos. 11174182 and
61306113) and the Specialized Research Fund for the
Doctoral Program of Higher Education (Grant No.
20110131110005).
1

C. Wood and D. Jena, Polarization Effects in Semiconductors: From Ab

Initio Theory to Device Applications (Springer, New York, 2007).
2
K. Rim, J. L. Hoyt, and J. F. Gibbons, IEEE Trans. Electron Devices 47,
1406 (2000).
3
S. Shan, J. S. Yuan, and Z. J. Shen, IEEE Electron Device Lett. 58, 1517
(2011).
4
Y. F. Zhang, I. P. Smorchkova, C. R. Elsass, S. Keller, J. P. Ibbetson, S.
Denbaars, U. K. Mishra, and J. Singh, J. Appl. Phys. 87, 7981 (2000).
5
N. Goyal and T. A. Fjeldly, J. Appl. Phys. 113, 014505 (2013).
6
J. Bernat, P. Javorka, M. Marso, and P. Kordos, Appl. Phys. Lett. 83, 5455
(2003).
7
C. M. Jeon and J. L. Lee, Appl. Phys. Lett. 86, 172101 (2005).
8
H. Tokuda, T. Kojima, and M. Kuzuhara, Appl. Phys. Lett. 101, 082111
(2012).
9
H. Tokuda, T. Kojima, and M. Kuzuhara, Appl. Phys. Lett. 101, 262104
(2012).
10
F. Gonzalez-Posada Flores, C. Rivera, and E. Mu~
noz, Appl. Phys. Lett. 95,
203504 (2009).
11
C. Rivera and E. Mu~
noz, Appl. Phys. Lett. 94, 053501 (2009).
12
A. F. M. Anwar, R. T. Webster, and K. V. Smith, Appl. Phys. Lett. 88,
203510 (2006).
13
J. D. Guo, M. S. Feng, R. J. Guo, F. M. Pan, and C. Y. Chang, Appl. Phys.
Lett. 67, 2657 (1995).
14
Q. Z. Liu, L. S. Yu, S. S. Lau, J. M. Redwing, N. R. Perkins, and T. F.
Kuech, Appl. Phys. Lett. 70, 1275 (1997).
15
H. S. Venugopalan and S. E. Mohney, Appl. Phys. Lett. 73, 1242 (1998).
16
L. S. Yu, D. J. Qiao, Q. J. Xing, S. S. Lau, K. S. Boutros, and J. M.
Redwing, Appl. Phys. Lett. 73, 238 (1998).
17
J. Z. Zhao, Z. J. Lin, T. D. Corrigan, Z. Wang, Z. D. You, and Z. G.
Wang, Appl. Phys. Lett. 91, 173507 (2007).
18
Y. J. Lv, Z. J. Lin, L. G. Meng, Y. X. Yu, C. B. Luan, Z. F. Cao, H. Chen,
B. Q. Sun, and Z. G. Wang, Appl. Phys. Lett. 99, 123504 (2011).
19
J. H. Shin, J. H. Park, S. Y. Jang, T. Jang, and K. S. Kim, Appl. Phys. Lett.
102, 243505 (2013).
20
O. Ambacher, J. Smart, J. R. Shealy, N. G. Weimann, K. Chu, M. Murphy,
W. J. Schaff, L. F. Eastman, R. Dimitrov, L. Wittmer, M. Stutzmann, W.
Rieger, and J. Hilsenbeck, J. Appl. Phys. 85, 3222 (1999).
21
O. Ambacher, B. Foutz, J. Smart, J. R. Shealy, N. G. Weimann, K. Chu,
M. Murphy, A. J. Sierakowski, W. J. Schaff, L. F. Eastman, R. Dimitrov,
A. Mitchell, and M. Stutzmann, J. Appl. Phys. 87, 334 (2000).
22
S. N. Mohammad, J. Appl. Phys. 97, 063703 (2005).
23
C. I. Wu and A. Kahn, J. Vac. Sci. Technol., B. 16, 2218 (1998).
24
See supplementary material at http://dx.doi.org/10.1063/1.4931122 for
complete experimental results and discussion.
25
J. P. Ao, D. Kikuta, N. Kubota, Y. Naoi, and Y. Ohno, IEEE Electron
Device Lett. 24, 500 (2003).
26
S. Arulkumaran, T. Egawa, H. Ishikawa, M. Umeno, and T. Jimbo, IEEE
Trans. Electron Devices 48, 573 (2001).
27
S. Choi, E. Heller, D. Dorsey, R. Vetury, and S. Graham, J. Appl. Phys.
113, 093510 (2013).
28
Y. J. Lv, Z. J. Lin, Y. Zhang, L. G. Meng, C. B. Luan, Z. F. Cao, H. Chen,
and Z. G. Wang, Appl. Phys. Lett. 98, 123512 (2011).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
197.207.21.231 On: Sat, 19 Sep 2015 10:11:53

113502-5
29

Zhao et al.

C. B. Luan, Z. J. Lin, Y. J. Lv, L. G. Meng, Y. X. Yu, Z. F. Cao, H. Chen,

and Z. G. Wang, Appl. Phys. Lett. 101, 113501 (2012).
30
Y. J. Lv, Z. H. Feng, T. T. Han, S. B. Dun, G. D. Gu, J. Y. Yin, B. C.
Sheng, B. Liu, Y. L. Fang, S. J. Cai, Z. J. Lin, C. B. Luan, and Q. H. Yang,
Appl. Phys. Lett. 103, 113502 (2013).
31
J. T. Zhao, Z. J. Lin, C. B. Luan, Y. Zhou, M. Yang, Y. J. Lv, and Z. H.
Feng, Appl. Phys. Lett. 105, 083501 (2014).

Appl. Phys. Lett. 107, 113502 (2015)

32

J. T. Zhao, Z. J. Lin, C. B. Luan, Q. Y. Chen, M. Yang, Y. Zhou, Y. J. Lv,

and Z. H. Feng, Superlattices Microstruct. 79, 21 (2015).
A. D. McNaught and A. Wilkinson, Compendium of Chemical
Terminology, 2nd ed. (Blackwell Science, Oxford, 1997), p. 12.
34
C. H. Hsueh, S. Lee, and S. Lee, J. Appl. Mech. 70, 151 (2003).
35
S. M. Hu, J. Appl. Phys. 70, R53 (1991).
36
H. C. Liu and S. P. Murarka, J. Appl. Phys. 72, 3458 (1992).
33

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
197.207.21.231 On: Sat, 19 Sep 2015 10:11:53