The thermal oxidation rates of Al0.83In0.17N layers grown lattice-matched to GaN and the oxide's optical constants are studied. The ∼230 nm thick AlInN layers are placed into a horizontal furnace at elevated temperatures and exposed to either O2 (dry) or H2O vapor with an N2 carrier gas (wet). The samples are oxidized at different temperatures (830–870 °C) and at a constant time or at various times at a constant temperature of 830 °C. Spectroscopic ellipsometry is used to determine the oxide thicknesses, refractive index, and extinction coefficients. The oxidation rate for the wet conditions is faster than for the dry conditions, and both increase with temperature, as expected. However, the oxidation rate is also dynamic with time and can be fitted with the Deal–Grove model, suggesting reaction rate and diffusion-limited processes like other more mature semiconductors. Finally, the dry conditions produce oxides that expand more than the oxides produced under wet conditions. The ability to thermally oxidize Al0.83In0.17N layers lattice-matched to GaN is a promising process technique for producing new III-nitride-based electronic devices.

The ability to oxidize semiconductors plays a pivotal role in semiconductor devices by providing a means to fabricate and design new device architectures. The prime examples are the thermal oxidation of Si to form SiO21 or AlGaAs into Al2O3.2–5 These oxides are utilized in many ways, including as electrical isolating layers, waveguides, masking layers, and gate oxides. Currently, GaN-based power devices that can operate at high voltages, switching speeds, and temperatures are being developed.6–10 They are being produced by taking advantage of the availability of low defect density GaN substrates and the growth of thick and low-doped GaN drift layers to support high electric fields.

Another critical technology that would advance GaN-based power devices would be producing an electrically insulating oxide, analogous to SiO2 on Si. There have been some attempts at thermally oxidizing GaN to create insulating layers. The thermal oxidation of GaN forms Ga2O3 in both O2 and water vapors.11–13 Oxidation rates become appreciable at 800 °C and are reported as high as 1100 °C. However, this oxide has poor electrical properties such as a low bandgap, limiting its use as an insulator.14,15 Additionally, the thermal oxidation of AlGaN with low Al-content (<0.3) has been used to create thin, insulating oxide layers in high-electron-mobility transistors (HEMTs) to passivate and improve performance.16,17 In Ref. 17, the oxidation of Al0.27Ga0.73N at 900 °C in O2 results in an oxidation rate of 1.2 nm/h, an expansion of ∼1.12, and a diffusivity of 4.63 × 10−13 cm2/s. Comprehensive thermal oxidation rates of AlGaN are lacking.

AlxIn1−xN lattice-matched to c-plane GaN (x = 0.82) has become an area of interest for GaN-based power devices,18–21 and more recently, the thermal oxidation of these layers has been reported.22–25 Most researchers report on the oxidation of AlInN to form thin oxide layers in AlInN/GaN HEMTs (similar to AlGaN/GaN HEMTs) to suppress gate leakage. In Ref. 25, the Al0.82In0.18N films are subject to wet thermal oxidation (H2O in an N2 carrier gas) at high temperatures (900 °C) and are converted to a stable, robust oxide that is not limited to thin layers. The reported thermal oxide has a small amount of nitrogen remaining (∼1%), is insulating with resistivity >8 × 1014 Ωcm, can oxidize deeply (at least 250 nm), and has breakdown electric fields >2 MV/cm. Previous studies of thermal oxidation focused on thin oxide layer devices22–24 and did not determine comprehensive oxidation rates, which requires varying oxidation times and forming thicker oxide films.

The ability to oxidize a lattice-matched layer to GaN is beneficial for various devices beyond HEMTs. For example, the AlInN layer’s thickness on GaN is not confined by strain, which provides flexibility in oxide thicknesses. Additionally, the oxidation of an epitaxially grown AlInN layer on GaN avoids surface contamination that is a concern with deposited oxides but is useful for metal–oxide–semiconductor transistors. Thus, due to the limited available research literature regarding the thermal oxidization rate of Al0.82In0.18N, new studies are necessary for a more in-depth understanding into the subject so that the oxide can be used to create new GaN-based devices.

This paper investigates the oxidation rates for the thermal oxide of AlxIn1−xN lattice-matched to GaN (x = 0.83) and the resulting oxide's optical constants. Thermal oxides formed in O2 (dry) and by H2O vapor with an N2 carrier gas (wet) at elevated temperatures (830–870 °C) are studied. The oxidation rate increases with temperature, and the rate is faster under wet conditions than under dry conditions. Oxidation at different times and at a constant temperature of 830 °C shows an oxidation rate consisting of two regimes, similar to the oxidation of Si that can be fitted with the Deal–Grove model. The thickness of the grown oxide is greater than the consumed AlInN, and surprisingly, the dry conditions result in more significant expansion. Finally, the refractive index for the dry and wet environments is different.

The AlxIn1−xN samples are grown by metal-organic chemical vapor deposition (MOCVD) on template layers consisting of 4.5 μm, unintentionally doped GaN on sapphire substrates.25 The Al content is ∼0.83. The Al0.83In0.17N layers are ∼230 nm thick, as confirmed using spectroscopic ellipsometry. They are unintentionally doped and have an n-type conductivity and electron concentrations of ∼2 × 1017 cm−3. The growth conditions and materials characterization of these lattice-matched AlInN films have been reported previously.26–28 

The Al0.83In0.17N films are oxidized in the following way. The samples are cleaned with acetone and isopropyl alcohol, placed vertically in a quartz boat, and inserted into a horizontal quartz tube furnace. For dry oxidation, O2 flows directly into the furnace at ∼13 SCFH. For wet oxidation, N2 carrier gas flows ∼14 SCFH through an H2O bubbler set to 95 °C, which is connected to the furnace. The diameter of the quartz tube is 2.755-in. Two oxidation experiments are presented in this paper for both environments. In the first experiment, the samples are oxidized at various temperatures (830–870 °C) but for a constant time of 1 h for wet and 2 h for dry. In the second experiment, the samples are oxidized for various times but at a constant temperature of 830 °C. After the oxidation is complete, the oxide thickness is measured using spectroscopic ellipsometry, and the oxidation rate, oxide thickness, and refractive index are obtained. The oxide thickness was also verified using a profilometer to measure the step height after etching some of the dry oxide samples.

Spectroscopic ellipsometry is performed on a J.A. Woollam V-Vase ellipsometer for wavelengths ranging from 350 to 800 nm in increments of 1 nm and for incident angles of 60° and 70°. The thickness for different layers in each sample is determined along with the refractive index (n) and extinction coefficient (k) from the ellipsometry measurements using the ellipsometry software CompleteEase. The structure used to fit the ellipsometry data consisted of a sapphire substrate followed by a GaN layer, an AlInN layer, and after oxidation, an AlInO layer. Spectroscopic ellipsometry is performed twice to obtain accurate fits of the oxide layer. The first ellipsometry measurement is performed before the samples are oxidized to determine the starting thicknesses of GaN and AlInN. Additionally, the n and k for the GaN and Al0.82In0.18N layers are slightly tuned to match known n and k values and assure that the thicknesses are accurate.25,28

The AlInN layer is fit using the Tauc–Lorentz oscillator model, and the starting thickness of the samples is ∼230 nm. After oxidation, spectroscopic ellipsometry is performed a second time to determine the oxide thickness, n, and k for each sample. Again, the Tauc–Lorentz oscillator model is used to fit the oxide layer from 400 to 800 nm. The absorbing region of GaN near 367 nm is not included when fitting the oxide layer’s thickness, because it complicates the fit. One sample for each of the four different oxidation experiments is used to fit the Tauc–Lorentz oscillator parameters and establish a common n and k in addition to the thickness of the oxide. For the remaining samples, the only fitting variables are the thicknesses for the AlInN and oxide layers, while the Tauc–Lorentz model parameters related to n and k are held constant based on the first fitted sample. This method is used successfully for three of the four sample sets. For the wet oxides that varied over time, constant n and k did not provide accurate fits. For these samples, the n and k are varied, and further research is required to fully understand why the n and k change over time. Although the n and k varied for this sample set, the thicker oxides formed after 1 h, 1 h and 30 min, and 2 h showed similar n and k values. Figure 1 shows an example fit to the ellipsometry data using the Tauc–Lorentz model for a sample oxidized in dry conditions at 830 °C for 2 h, resulting in an oxide thickness of 55.24 nm. The plot shows the measured amplitude ratio, Psi, and phase difference, Delta, for two different angles 60° and 70°. The fit is represented by the dotted lines and confirms the accuracy of the fitting.

FIG. 1.

Ellipsometry data collected at two different angles (60° and 70°) for wavelengths ranging from 400 to 800 nm for the (a) amplitude ratio, Psi, and (b) phase difference, Delta. These data correspond to a sample that is oxidized in dry conditions at 830 °C for 2 h, resulting in an oxide thickness of 55.24 nm.

FIG. 1.

Ellipsometry data collected at two different angles (60° and 70°) for wavelengths ranging from 400 to 800 nm for the (a) amplitude ratio, Psi, and (b) phase difference, Delta. These data correspond to a sample that is oxidized in dry conditions at 830 °C for 2 h, resulting in an oxide thickness of 55.24 nm.

Close modal

Figure 2 shows a plot of oxidation rate vs temperature for temperatures ranging from 830 to 870 °C that are similar to that of previous studies.25 The samples that are used for the oxidation study over temperature have an indium content of 16.9%. The dry oxidation is carried out for 2 h at each temperature, and the wet oxidation for 1 h at each temperature. Figure 2(a) shows the trend for the dry oxidation, and Fig. 2(b) shows the trend for the wet oxidation. The limited thickness of the AlInN layer necessitates a shorter time for the wet oxidation because of the faster oxidation rate. The plot shows that, as the oxidation temperature is increased, the oxidation rate increases linearly. The oxidation rate for the dry oxidation ranges between 0.41 and 1.03 nm/min and from 0.77 to 1.56 nm/min for the wet oxidation. Except for the data point at 850 °C, both environments’ rate of oxidation follows a linear trend. It should be noted that the wet oxidation at 850 °C was repeated and produced similar values, and it is unclear why there is a dip in the oxidation rate. The oxidation rate is faster under wet conditions than under dry conditions. This contradicts the observation made in Ref. 25; however, the present study compares multiple sample sets, while in the previous study, there was an initial report with a limited number of samples.

FIG. 2.

The oxidation rate vs temperature of Al0.83In0.17N ranging from 830 to 870 °C. The rates are obtained in (a) dry O2 conditions after 2 h and in (b) wet H2O vapor in an N2 carrier gas after 1 h. The dotted lines are linear fits to the data.

FIG. 2.

The oxidation rate vs temperature of Al0.83In0.17N ranging from 830 to 870 °C. The rates are obtained in (a) dry O2 conditions after 2 h and in (b) wet H2O vapor in an N2 carrier gas after 1 h. The dotted lines are linear fits to the data.

Close modal

A second oxidation experiment is carried out by varying the oxidation times at a constant temperature to determine the oxidation rate’s dynamic nature. Figure 3 shows the oxide thickness vs time at 830 °C for both dry and wet oxidation conditions. The samples that are used for the oxidation study over time have an indium content of 17.15%. The dry oxidation is carried out for times ranging from 30 to 360 min. Due to the increased oxidation rate, the wet oxidation is performed at shorter times from 15 to 120 min. There is a faster oxidation rate at the beginning of the process and slower rates at longer times for both conditions. This oxidation rate behavior is similar to that of the thermal oxidation of Si and SiC and can be described by the oxidation kinetic behavior developed by Deal–Grove.29,30 The two rate regimes have different oxidation behaviors. The initial (short-time) oxidation rates are reaction rate limited, while the long-time oxidation rates are determined by the oxidant diffusion and concentration in the oxide.29 The appearance of the two regimes for AlInN suggests similar oxidation behavior.

FIG. 3.

Plot of the oxide thickness (nm) vs oxidation time (min) of Al0.83In0.17N at 830 °C. The samples are oxidized (a) for times ranging from 30 to 360 min in dry conditions and (b) for times ranging from 15 to 120 min in wet conditions. The data are fitted using the Deal–Grove model, which is represented by the dotted lines.

FIG. 3.

Plot of the oxide thickness (nm) vs oxidation time (min) of Al0.83In0.17N at 830 °C. The samples are oxidized (a) for times ranging from 30 to 360 min in dry conditions and (b) for times ranging from 15 to 120 min in wet conditions. The data are fitted using the Deal–Grove model, which is represented by the dotted lines.

Close modal

The plots for both dry and wet oxidation for the AlInN films can be fit using the Deal–Grove model. The thickness of the oxide layer, xo, is the solution to a differential rate equation and can be expressed as

x0=A2[1+tA24B]121,
(1)

where t is the time, B is the parabolic growth rate constant and is a function of oxidant diffusion rate and saturation, and A is the function of the oxidant diffusion and reaction rates.29 Equation (1) is used to calculate the oxide thickness over time in Figs. 3(a) and 3(b), where the A and B parameters fit the data. Fits for the wet oxidation give an A and B of 4.99 nm and 48.58 nm2/min, respectively, while for dry oxidation, the A and B are 138.08 nm and 115.99 nm2/min, respectively. The values of the parameters that contribute to A and B, such as reaction rates and oxidant diffusion, require further study. These data show that this oxidation kinetic behavior is similar to that in other semiconductors, even though this is a wide bandgap semiconductor that was previously assumed to be limited to thin oxide layers less than 10 nm.25 Past oxidation studies in thin AlInN layers22,23 concluded diffusion-limited oxidation rates at short times without performing a thickness vs time analysis.

Next, the diffusivities of the AlInN oxide that determine the oxidation rates are compared with those of silicon. The parameters that determine the oxidation of AlInN are not available in the literature, so values for Al2O3 (alumina) are used. For the O2 conditions at 830 °C, the parabolic growth rate constants, B, are ∼3 × 10−15 cm2/s for Si29 and 1.9 × 10−14 cm2/s found here for AlInN. B can be expressed as31 

B=2Dox(N/Nox),
(2)

where Dox is the diffusivity of the oxidant, N* is the equilibrium concentration of the oxidant in the oxide, and Nox is the concentration of oxidant molecules in the oxide layer. Nox is 2.25 × 1022 and 1.64 × 1022 cm−3 for SiO2 and Al2O3. The diffusivity data of O2 in Al2O3 are limited to high temperatures (>1500 K),32 so the diffusivity of O2 in alumina33 is extrapolated to 830 °C, which results in ∼4.9 × 10−17 cm2/s. For SiO2 under these conditions, Dox is ∼2.2 × 10−10 cm2/s. Using Eq. (2), N* are 3.2 × 1024 and 1.5 × 1017 cm−3 for AlInN oxide and SiO2, respectively. These values are orders of magnitude apart, which are not expected and are most likely from the choice of using the diffusivity of O2 in alumina, which is much lower than that of SiO2. To match the N* of SiO2, a diffusivity of ∼10−9 cm2/s is required, closer but still higher than the thermal oxide of Al0.27Ga0.73N.17 It is likely that factors such as the phase of the oxide, the inclusion of In, and grain boundaries help raise the diffusivity.34 Determining the diffusivities in these thermal oxides warrants future work.

The average thickness expansion for the samples oxidized at different temperatures in a dry environment for 2 h and a wet environment for 1 h is also determined. The oxide expansion is obtained by dividing the oxide thickness by the thickness of the consumed AlInN. This expansion is sometimes referred to as the Pilling–Bedworth ratio35 but will be referred to here as a thickness expansion ratio. Surprisingly, the average dry oxide expansion is far greater than the wet oxide expansion over the temperature range that is studied. The expansion of the dry oxide averages 2.08 nm/nm for the samples oxidized over temperature, while the expansion of the wet oxide averages 1.27 nm/nm. The different expansions for the two oxidation conditions indicate either different compositions, crystal structures, or voiding within the oxide layer. These hypotheses are further supported by the fact that the dry oxide can be wet etched, while the wet oxide cannot be.25 This indicates that there is a phase difference between the dry oxide and the wet oxide.

The average thickness expansion for the samples oxidized over different times at 830 °C for both dry and wet environments is also determined. In general, the oxide formed under dry conditions expands more than under wet conditions. For oxidation times greater than 30 min, the average expansion for the dry oxide was 2.01 nm/nm, while the average thickness expansion for the wet oxide was 1.05 nm/nm. For short oxidation times, the oxide expansion was more significant than for more extended periods for both conditions. This difference in expansion early in the oxidation process could be caused by several factors, such as dramatic changes in the crystal structure or possibly small errors in the thickness measurement due to the precision of the ellipsometer data, and is, therefore, not included in the average expansion. For thicker oxide layers, any errors have a smaller effect, and the ratio used to calculate the expansion is more accurate. The average expansion for the wet oxidation over temperature is different than the wet oxidation over time. One reason for this difference is that the indium composition in the starting AlInN samples when varying the temperature is at 16.9% and at 17.15% when varying the time. What is more likely though, is that the wet oxidation results in a different phase.

The expansion of the oxide layer from the initial film could pose a problem for devices. Any in-plane expansion and resulting strain are not large enough to cause delamination, even for the thickest layers investigated here. Non-delamination could be due to stain relief from dislocations or the expansion is mainly out of the plane (the c direction). Further structural analysis is required to determine the crystal structure of the oxide and of the oxide and the semiconductor interface. It is also interesting to note that this expansion is different from the oxidation of AlGaAs, which contracts.36 This is not unexpected, given the two semiconductors have different crystal structures and different oxidation conditions, especially a much lower typical oxidation temperature for AlGaAs, resulting in different oxide phases.

In addition to the rate of oxidation, the refractive index and extinction coefficient are obtained for wavelengths ranging from 400 to 800 nm for dry oxidation after 2 h at 830 °C and for wet oxidation after 2 h at 830 °C. Figure 4 shows a plot of the refractive index and extinction coefficients for the wet and dry oxides vs wavelength. The refractive index for dry and wet oxidation is distinct, with n at 500 nm being 1.548 for the dry oxide and 1.835 for the wet oxide. The difference in the refractive index is not unexpected, considering that the expansions for the two oxides are different. When fitting the dry oxidation thickness, the n and k remained constant for the different oxidation times. For the wet oxidation at different times, the n and k are varied to better fit the thicknesses. The n and k that are plotted for the wet oxide in Fig. 4(b) are for a thick oxide layer formed after 2 h of oxidation, since thicker wet formed oxide layers generally give more consistent n and k. The n at 500 nm is close to what is previously reported in Ref. 25 at ∼1.8. There is a slight increase in k at wavelengths less than 450 nm, which requires further study.

FIG. 4.

Refractive index (n) and extinction coefficient (k) vs wavelength for (a) dry oxidation at 830 °C after 2 h and (b) for wet oxidation at 830 °C for 2 h. The refractive index and extinction coefficients are determined from the spectroscopic ellipsometry. The extinction coefficient for the dry oxide is zero over these wavelengths.

FIG. 4.

Refractive index (n) and extinction coefficient (k) vs wavelength for (a) dry oxidation at 830 °C after 2 h and (b) for wet oxidation at 830 °C for 2 h. The refractive index and extinction coefficients are determined from the spectroscopic ellipsometry. The extinction coefficient for the dry oxide is zero over these wavelengths.

Close modal

In conclusion, Al0.83In0.17N oxidized in both dry and wet environments have different oxidation rates, and wet conditions are faster. These oxidation rates can be fitted with the Deal–Grove model and suggest that the oxidation process is similar to that in other oxidizable semiconductors. The expansion of the oxide is very different for the two conditions, suggesting either a different solid-phase structure, composition, or voiding. Determining this structure and its impact on insulating behavior will require future work so that the oxide can be intelligently integrated into semiconductor devices.

We would like to thank the technical staff (Anthony Jeffers and Grant Reed) at the Center for Photonics and Nanoelectronics for providing help with the oxidation furnace. We acknowledge funding from the U.S. National Science Foundation (Award Nos. 1408051, 1505122, 1708227, and 1935295) and the Daniel E. ’39 and Patricia M. Smith Endowed Chair Professorship Fund (N.T.).

The data that support the findings of this study are available within the article.

1.
C. J.
Frosch
and
L.
Derick
,
J. Electrochem. Soc.
104
(
9
),
547
(
1957
).
2.
J. M.
Dallesasse
,
N.
Holonyak
,
A. R.
Sugg
,
T. A.
Richard
, and
N.
El-Zein
,
Appl. Phys. Lett.
57
(
26
),
2844
2846
(
1990
).
3.
J. M.
Dallesasse
and
N.
Holonyak
,
Appl. Phys. Lett.
58
(
4
),
394
396
(
1991
).
4.
A. R.
Sugg
,
N.
Holonyak
,
J. E.
Baker
,
F. A.
Kish
, and
J. M.
Dallesasse
,
Appl. Phys. Lett.
58
(
11
),
1199
1201
(
1991
).
5.
F. A.
Kish
,
S. J.
Caracci
,
N.
Holonyak
,
J. M.
Dallesasse
,
A. R.
Sugg
,
R. M.
Fletcher
,
C. P.
Kuo
,
T. D.
Osentowski
, and
M. G.
Craford
,
Appl. Phys. Lett.
59
(
3
),
354
356
(
1991
).
6.
Y.
Saitoh
,
K.
Sumiyoshi
,
M.
Okada
,
T.
Horii
,
T.
Miyazaki
,
H.
Shiomi
,
M.
Ueno
,
K.
Katayama
,
M.
Kiyama
, and
T.
Nakamura
,
Appl. Phys. Express
3
(
8
),
081001
(
2010
).
7.
I. C.
Kizilyalli
,
A. P.
Edwards
,
H.
Nie
,
D.
Bour
,
T.
Prunty
, and
D.
Disney
,
IEEE Electron Device Lett.
35
(
2
),
247
249
(
2014
).
8.
T. J.
Anderson
,
B. N.
Feigelson
,
F. J.
Kub
,
M. J.
Tadjer
,
K. D.
Hobart
,
M. A.
Mastro
,
J. K.
Hite
, and
C. R.
Eddy
,
Electron. Lett.
50
(
3
),
197
198
(
2014
).
9.
H.
Ohta
,
N.
Kaneda
,
F.
Horikiri
,
Y.
Narita
,
T.
Yoshida
,
T.
Mishima
, and
T.
Nakamura
,
IEEE Electron Device Lett.
36
(
11
),
1180
1182
(
2015
).
10.
J. R.
Dickerson
,
A. A.
Allerman
,
B. N.
Bryant
,
A. J.
Fischer
,
M. P.
King
,
M. W.
Moseley
,
A. M.
Armstrong
,
R. J.
Kaplar
,
I. C.
Kizilyalli
,
O.
Aktas
, and
J. J.
Wierer
,
IEEE Trans. Electron Devices
63
(
1
),
419
425
(
2016
).
11.
S. D.
Wolter
,
B. P.
Luther
,
D. L.
Waltemyer
,
C.
Önneby
,
S. E.
Mohney
, and
R. J.
Molnar
,
Appl. Phys. Lett.
70
(
16
),
2156
2158
(
1997
).
12.
E. D.
Readinger
,
S. D.
Wolter
,
D. L.
Waltemyer
,
J. M.
Delucca
,
S. E.
Mohney
,
B. I.
Prenitzer
,
L. A.
Giannuzzi
, and
R. J.
Molnar
,
J. Electron. Mater.
28
(
3
),
257
260
(
1999
).
13.
P.
Chen
,
R.
Zhang
,
X. F.
Xu
,
Y. G.
Zhou
,
Z. Z.
Chen
,
S. Y.
Xie
,
W. P.
Li
, and
Y. D.
Zheng
,
Appl. Phys. A
71
(
2
),
191
194
(
2000
).
14.
H.
Kim
,
S.-J.
Park
, and
H.
Hwang
,
J. Vac. Sci. Technol. B
19
(
2
),
579
581
(
2001
).
15.
Y.
Nakano
and
T.
Jimbo
,
Appl. Phys. Lett.
82
(
2
),
218
220
(
2003
).
16.
H.
Liu
,
C.
Lee
,
W.
Hsu
,
L.
Tseng
,
B.
Chou
,
C.
Ho
, and
C.
Wu
,
IEEE Trans. Electron Devices
60
(
7
),
2231
2237
(
2013
).
17.
G.
Greco
,
P.
Fiorenza
,
F.
Giannazzo
,
A.
Alberti
, and
F.
Roccaforte
,
Nanotechnology
25
,
025201
(
2013
).
18.
M. R.
Peart
,
N.
Tansu
, and
J. J.
Wierer
,
IEEE Trans. Electron Devices
65
(
10
),
4276
4281
(
2018
).
19.
M.
Kuramoto
,
S.
Kobayashi
,
T.
Akagi
,
K.
Tazawa
,
K.
Tanaka
,
T.
Saito
, and
T.
Takeuchi
,
Appl. Sci.
9
,
416
(
2019
).
20.
M. R.
Peart
,
D.
Borovac
,
W.
Sun
,
R.
Song
,
N.
Tansu
, and
J. J.
Wierer
,
Appl. Phys. Express
13
(
9
),
091006
(
2020
).
21.
M.
Miyoshi
,
M.
Yamanaka
,
T.
Egawa
, and
T.
Takeuchi
,
Jpn. J. Appl. Phys.
58
(
SC
),
SC1006
(
2019
).
22.
M.
Alomari
,
F.
Medjdoub
,
J.
Carlin
,
E.
Feltin
,
N.
Grandjean
,
A.
Chuvilin
,
U.
Kaiser
,
C.
Gaquiere
, and
E.
Kohn
,
IEEE Electron Device Lett.
30
(
11
),
1131
1133
(
2009
).
23.
P.
Kordoš
,
M.
Mikulics
,
R.
Stoklas
,
K.
Cico
,
A.
Dadgar
,
D.
Grutzmacher
, and
A.
Krost
,
J. Electron. Mater.
41
,
3013
(
2012
).
24.
M.
Eickelkamp
,
M.
Weingarten
,
L.
Rahimzadeh Khoshroo
,
N.
Ketteniss
,
H.
Behmenburg
,
M.
Heuken
, and
H.
Kalisch
,
R. H.
Jansen
and
A.
Vescan
,
Phys. Status Solidi C
8
(
7–8
),
2213
2215
(
2011
).
25.
M. R.
Peart
,
X.
Wei
,
D.
Borovac
,
W.
Sun
,
N.
Tansu
, and
J. J.
Wierer
,
ACS Appl. Electron. Mater.
1
(
8
),
1367
1371
(
2019
).
26.
G.
Liu
,
J.
Zhang
,
X.-H.
Li
,
G. S.
Huang
,
T.
Paskova
,
K. R.
Evans
,
H.
Zhao
, and
N.
Tansu
,
J. Cryst. Growth
340
(
1
),
66
73
(
2012
).
27.
D.
Borovac
,
W.
Sun
,
R.
Song
,
J. J.
Wierer
, and
N.
Tansu
,
J. Cryst. Growth
533
,
125469
(
2020
).
28.
D.
Borovac
,
W.
Sun
,
M. R.
Peart
,
R.
Song
,
J. J.
Wierer
, and
N.
Tansu
,
J. Cryst. Growth
548
,
125847
(
2020
).
29.
B. E.
Deal
and
A. S.
Grove
,
J. Appl. Phys.
36
(
12
),
3770
3778
(
1965
).
30.
Y.
Song
,
S.
Dhar
,
L. C.
Feldman
,
G.
Chung
, and
J. R.
Williams
,
J. Appl. Phys.
95
(
9
),
4953
4957
(
2004
).
31.
G. E.
Anner
, Planar Processing Primer, 1st ed. (
Springer
,
New York
,
1990
).
32.
R. H.
Doremus
,
J. Appl. Phys.
100
(
10
),
101301
(
2006
).
33.
Y.
Oishi
and
W. D.
Kingery
,
J. Chem. Phys.
33
(
2
),
480
486
(
1960
).
34.
A. H.
Heuer
,
J. Eur. Ceram. Soc.
28
(
7
),
1495
1507
(
2008
).
35.
J. S.
Dunn
,
Proc. Royal Soc. London
111
(
757
),
203
209
(
1926
).
36.
K. D.
Choquette
,
K. M.
Geib
,
C. I. H.
Ashby
,
R. D.
Twesten
,
O.
Blum
,
H. Q.
Hou
,
D. M.
Follstaedt
,
B. E.
Hammons
,
D.
Mathes
, and
R.
Hull
,
IEEE J. Sel. Top. Quantum Electron.
3
(
3
),
916
926
(
1997
).