The authors have investigated different methods for preparing the surfaces of freestanding, Ga-polar, hydride vapor-phase epitaxy grown GaN substrates to be used for homoepitaxial GaN growth by plasma-assisted molecular beam epitaxy (MBE). Cross-sectional transmission electron microscopy and secondary ion mass spectroscopy, respectively, were used to characterize the microstructure and to measure the concentrations of impurities unintentionally incorporated in the MBE-grown homoepitaxial GaN layers. Heating Ga-polar substrates to ∼1100 °C is as effective as a wet chemical clean for reducing impurity concentrations of oxygen, silicon, and carbon. The combination of an aggressive ex situ wet chemical clean with in situ Ga deposition and thermal desorption results in homoepitaxial GaN layer growth with very low residual impurity concentrations and without generating additional threading dislocations.

Gallium nitride-based devices are now commercially available, but many of these devices are grown heteroepitaxially on non-native substrates such as Si, SiC, and sapphire due to the limited availability and high cost of bulk single-crystal III-nitride substrates. The density of threading dislocations (TDs) in heteroepitaxial GaN is typically 108–1010 cm−2, so it is remarkable that commercial GaN-based devices like light-emitting diodes (LEDs) and high electron mobility transistors (HEMTs) function despite the high TD density. The generation of high densities of TDs in heteroepitaxial GaN is generally attributed to lattice mismatch and the difference in thermal expansion coefficients between the GaN layer and the underlying non-native substrate.1 In contrast, homoepitaxial GaN can be grown on low dislocation density, native GaN substrates without generating any additional threading dislocations.2,3

There have been recent positive developments in the production and availability of bulk nitride materials, and a variety of devices grown on freestanding GaN substrates have been demonstrated, such as LEDs,4,5 laser diodes,6–9 solar cells,10–12 vertical power p-n diodes,13 Schottky diodes,14 HEMTs,15–18 and resonant tunneling diodes (RTDs).19,20 However, the next generation of GaN-based devices, such as vertical power switches and RTDs require lower densities of extended and point defects. Bulk GaN crystals with very low TD densities can be grown by the high pressure21,22 and ammonothermal23,24 methods, yielding crystals with TD densities below 102 cm−2 for the former and 5 × 104 cm−2 for the latter. Hydride vapor-phase epitaxy (HVPE) has been used most frequently for producing freestanding GaN substrates due to its wider availability, faster growth rates, and larger and more uniform wafer size.25 While the TD density of HVPE-grown GaN substrates is typically 106–107 cm−2, several orders of magnitude greater than for high-pressure or ammonothermally grown GaN, homoepitaxial GaN growth on a properly prepared HVPE GaN substrate results in GaN layers with no new TDs generated at the regrowth interface (RI), yielding epitaxial layers with several orders of magnitude lower TD density than achievable by heteroepitaxy.2,3 Recipes for ex situ preparation of the HVPE GaN surface that result in high quality homoepitaxial growth have been published elsewhere.26,27 However, homoepitaxial growth on insufficiently cleaned or improperly prepared GaN substrates results in defective layers.27 It is therefore of great interest to identify effective and efficient methods of GaN substrate surface preparation which result in high quality homoepitaxial growth. In this work, we have investigated alternative, in situ methods and characterized their effect on the microstructure, impurity incorporation, and surface morphology of homoepitaxial GaN layers grown by rf-plasma assisted molecular beam epitaxy (rf-MBE) on freestanding Ga-polar HVPE-grown GaN substrates.

All samples were grown by rf-MBE in a Scienta-Omicron PRO-75 MBE deposition chamber with a base pressure of 2 × 10−11 Torr. The PRO-75 chamber is equipped with reflection high energy electron diffraction (RHEED); a Veeco Uni-Bulb rf-plasma source for supplying active nitrogen; dual-filament effusion cells for Ga, Al, and In; single-filament effusion cells for dopants (Si, Mg, and Be); and a six-pocket electron-beam evaporation source. However, during sample growth, only the nitrogen plasma source and one Ga effusion cell were used: all other sources were off (e-beam) or at standby conditions (effusion cells) to minimize unintentional contamination. The substrate temperature was monitored by a thermocouple mounted behind the wafer mount but not in direct contact with it. All temperatures stated here are thermocouple measurements.

All samples were grown on Ga-polar (0001) surface of 10 × 10 mm freestanding GaN substrates grown by HVPE; the effect of various ex situ and in situ cleaning regimens on homoepitaxial GaN growth on the N-polar (000–1) surface has been reported previously.28–30 The density of TDs in the freestanding wafers was between 106 and 107 cm−2. The Ga-polar surfaces of the substrate wafers were finished with a chemical-mechanical polish, and the wafers were mounted in In-free wafer mounts which were custom-machined from a molybdenum plate. In addition to a 30-min degas of all substrate wafers at 600 °C under ultrahigh vacuum conditions in the analysis chamber, the wafers were also subjected to various ex situ and in situ surface preparation regimens as shown in Table I and described below. These preparation regimens included ex situ wet chemical treatment, in situ thermal cleaning, and in situ deposition and desorption (D&D) of metallic Ga based on a technique first described by Bermudez et al.31 The variations in pregrowth substrate temperature, number of Ga deposition and desorption cycles, and amount of Ga deposited per cycle are due to attempts to minimize the effect of substrate surface roughening, which was marked by a transition from streaky to spotty RHEED patterns. Such roughening was expected to lead to defective GaN layer growth, and attempts were made to avoid or mitigate roughening by using lower temperatures or fewer Ga deposition and desorption cycles. However, as described more fully below, the roughening induced was not so severe as to cause the generation of additional TDs in subsequently grown homoepitaxial layers.

Table I.

Pregrowth treatment of freestanding GaN substrates used in this investigation. Columns show sample ID; whether or not an ex situ wet chemical clean was performed; number of cycles of in situ metallic gallium D&D cycles performed; approximate number of monolayers of Ga deposited in each cycle; the maximum temperature of the substrate wafer prior to growth; and the character of the RHEED pattern immediately before the start of growth.

Sample ID Ex situ clean (Y/N) Ga D&D cycles Ga D&D (ML/cycle) Maximum TS (°C) Pregrowth RHEED
1100  Streaky 
45  1000  Spotty 
45  950  Spotty 
960  Spotty 
45  920  Spotty 
835  Streaky 
Sample ID Ex situ clean (Y/N) Ga D&D cycles Ga D&D (ML/cycle) Maximum TS (°C) Pregrowth RHEED
1100  Streaky 
45  1000  Spotty 
45  950  Spotty 
960  Spotty 
45  920  Spotty 
835  Streaky 

The substrate of sample A underwent no ex situ wet chemical clean: the as-received wafer was loaded directly into the MBE system. After degassing in the analysis chamber, it was transferred to the deposition chamber. Its temperature was then ramped first to 1100 °C, then to the growth temperature, 860 °C. The RHEED pattern from the substrate was streaky at all temperatures, indicating a smooth, epitaxial surface. Likewise, the RHEED pattern from the surface of the MBE-grown GaN layer was also streaky at all times during and after growth.

The substrate of sample B was also loaded directly into the MBE system without any ex situ wet chemical clean. Following 30-min degas in the analysis chamber and subsequent transfer to the deposition chamber, its temperature was ramped to 750 °C. The Ga shutter was then opened, and Ga metal was deposited on the substrate surface. The RHEED pattern dimmed noticeably due to the accumulation of metallic Ga on the surface. The Ga shutter was closed after 2 min, and the substrate temperature was ramped to 950 °C. At a substrate temperature of approximately 920 °C, the brightness of the RHEED pattern increased rapidly, indicating desorption of metallic Ga from the substrate surface. The substrate temperature was then reduced to 750 °C and one more cycle of Ga deposition and desorption was performed. After the metallic Ga was desorbed, the substrate temperature was increased to 1000 °C. The RHEED pattern, which had been streaky, changed to a spotty pattern, indicating a change from a smooth to a roughened surface. The substrate temperature was immediately reduced to 860 °C to prevent further roughening of the surface.

The substrate of sample C was also loaded into the MBE system with no ex situ wet chemical clean. After degassing in the analysis chamber and transferring to the deposition chamber, its temperature was ramped to 750 °C, and it was then subjected to three cycles of Ga deposition at 750 °C and desorption at 950 °C. After the third such cycle, the temperature was reduced from 950 to 860 °C. Very soon after the substrate temperature reached 860 °C, the RHEED pattern, which had been streaky, became spotty.

The substrate of sample D underwent an aggressive, ex situ wet chemical clean prior to loading in the MBE system. This cleaning regimen has been described in detail elsewhere26,27 and consists of a solvent degrease and acid and base etches. We have found that GaN layers with no new TDs at the regrowth interface can be reliably grown on substrates cleaned in this manner. After loading into the deposition chamber, the substrate wafer was heated to 500 °C. The Ga shutter was opened for 5 s, and approximately 1.5–2 monolayers (ML) of metallic Ga was deposited on the substrate surface. The substrate temperature was then increased in order to desorb the Ga on the surface. The RHEED pattern changed from streaky to spotty at 960 °C, and the substrate temperature was immediately reduced to 860 °C.

The substrate for sample E also underwent an ex situ wet chemical clean identical to sample D. After degas in the analysis chamber and loading into the deposition chamber, it was subjected to two cycles of Ga deposition for 2 min at 750 °C and desorption at 920 °C, after which the substrate temperature was reduced to 860 °C. This procedure is similar to the Ga deposition and desorption treatment of samples B and C, but the lower desorption temperature was used in an attempt to avoid roughening of the substrate surface as evidenced by alteration of the RHEED pattern from streaky to spotty. While the streaky RHEED pattern persisted through the completion of the Ga deposition and desorption, a spotty pattern nevertheless emerged at 860 °C prior to initiation of growth.

The epitaxial GaN layers were grown identically for all five substrates. Growth was initiated by first exposing the substrate surface to the nitrogen plasma for 2 min, after which the Ga shutter was opened. In the four instances in which the RHEED pattern from the substrate surface changed from streaky to spotty (samples B–E), the pattern reverted to a streaky pattern within 60–90 s after the onset of GaN growth, and remained streaky for the remainder of growth. The GaN layers were grown without interruption and under constant Ga and active nitrogen fluences for 3 h at a constant substrate temperature of 860 °C. The Ga flux was approximately twice the active nitrogen flux, but no metallic Ga droplets were observed on the as-grown surface, indicating that the excess Ga had re-evaporated. The plasma conditions were maintained at 0.9 sccm and 300 W, resulting in a nitrogen-limited growth rate of approximately 180 nm/h. At the conclusion of the layer growth, all shutters were closed, the plasma was extinguished, and the substrate temperature was rapidly reduced to room temperature.

The samples were characterized by optical microscopy, atomic force microscopy (AFM), transmission electron microscopy (TEM), and secondary ion mass spectroscopy (SIMS).

A sixth sample, denoted X, was grown several weeks earlier under similar conditions on a substrate which underwent only the ex situ wet chemical clean and the 30-min degas in the analysis chamber. It was also characterized by SIMS and AFM, and the results are included here for comparison. We have used the pregrowth treatment of sample X to reliably and repeatably produce homoepitaxial GaN on freestanding Ga-polar substrates in which no new TDs are generated at the regrowth interface.2,3,27

Optical microscopy of the as-grown samples revealed smooth surfaces without evidence for Ga droplets. AFM was performed at three locations on each sample; the z-range and root-mean-square (rms) roughness from 20 × 20 μm regions at the center of each sample is shown in Table II, and three of the AFM images are shown in Fig. 1.

Table II.

Summary of SIMS and AFM analysis. Sample ID; integrated impurity sheet densities of oxygen, silicon, and carbon near the RI; z-range and rms roughness (Rq) from 20 × 20 μm fields of view near the sample center.

Sample ID [O]sh, RI (1013 cm−3) [Si]sh, RI (1013 cm−3) [C]sh, RI (1013 cm−3) z-range (nm) Rq (nm)
6.42  0.27  1.04  30.8  0.70 
2.15  0.09  0.10  31.4  0.74 
0.36  0.10  0.02  4.15  0.31 
0.80  0.24  0.03  13.7  0.81 
0.22  0.09  n/a  25.4  0.55 
6.91  0.23  0.46  —  — 
Sample ID [O]sh, RI (1013 cm−3) [Si]sh, RI (1013 cm−3) [C]sh, RI (1013 cm−3) z-range (nm) Rq (nm)
6.42  0.27  1.04  30.8  0.70 
2.15  0.09  0.10  31.4  0.74 
0.36  0.10  0.02  4.15  0.31 
0.80  0.24  0.03  13.7  0.81 
0.22  0.09  n/a  25.4  0.55 
6.91  0.23  0.46  —  — 
Fig. 1.

(Color online) AFM images from 20 × 20 μm fields of view near the center of: (a) sample A; (b) sample C; and (c) sample E. Z-ranges and rms roughnesses for all samples are given in Table II. All samples exhibit subnanometer rms roughness. Sample C has the smoothest surface morphology.

Fig. 1.

(Color online) AFM images from 20 × 20 μm fields of view near the center of: (a) sample A; (b) sample C; and (c) sample E. Z-ranges and rms roughnesses for all samples are given in Table II. All samples exhibit subnanometer rms roughness. Sample C has the smoothest surface morphology.

Close modal

The surface morphology and roughness values at the center of each wafer were representative of the other wafer locations. The z-range and rms roughness values were broadly consistent between the five samples, although sample C was markedly smoother compared to the other four samples. These results suggest that Ga deposition and desorption alone may be preferred for substrate cleaning if it is necessary to obtain the smoothest possible surfaces and interfaces, but more work is needed to confirm this possibility.

SIMS analysis of oxygen, silicon, and carbon impurity concentrations was performed on each of the five samples. Table II summarizes the sheet densities of each impurity at the regrowth interface of each sample, which was determined by integrating the impurity concentration with respect to depth near the interface. In every case, the O, Si, and C concentrations were at or near detection limits in the MBE-grown GaN layer. Significant concentrations of these impurities were only found near the regrowth interface between the MBE-grown epitaxial layer and the HVPE-grown GaN substrate. Figure 2 shows representative SIMS profiles from samples A and E. The impurity concentrations of sample A are very similar to those measured from homoepitaxial GaN layers on substrates prepared with the ex situ wet chemical clean, indicating that all surface preparations investigated here result in substrate surfaces at least as clean, or significantly cleaner, as that achieved by a wet chemical clean alone. In particular, the SIMS results indicate that heating an as-received substrate to 1100 °C is as effective at reducing impurity concentrations at the regrowth interface as an aggressive wet chemical clean, and that Ga deposition and desorption results in drastic reductions in impurity concentrations. The SIMS profiles suggest that prior to growth the distribution of impurities is restricted to the substrate surface; hence, heating the substrate surface under UHV conditions should result in desorption of impurities from the substrate surface and a reduction of the impurity concentrations at the regrowth interface. While three cycles of Ga deposition and desorption alone (sample C) resulted in very low sheet densities of O, Si, and C at the regrowth interface, the lowest impurity sheet densities appeared in sample E, which was subjected to a combination of wet chemical clean and Ga deposition and desorption. The carbon profile of sample E is particularly interesting, insofar as it completely lacks a peak at the regrowth interface, suggesting that the combination of the wet chemical clean and Ga deposition and desorption can completely remove adventitious carbon from the substrate's surface.

Fig. 2.

(Color online) Representative SIMS impurity profiles of oxygen, silicon, and carbon from samples A (a) and E (b). The impurity concentrations in the MBE-grown layers of all samples are at or near detection limits. Integrated sheet impurity densities of O, C, and Si near the regrowth interface are given in Table II. Note the absence of a C impurity peak at the regrowth interface in sample E.

Fig. 2.

(Color online) Representative SIMS impurity profiles of oxygen, silicon, and carbon from samples A (a) and E (b). The impurity concentrations in the MBE-grown layers of all samples are at or near detection limits. Integrated sheet impurity densities of O, C, and Si near the regrowth interface are given in Table II. Note the absence of a C impurity peak at the regrowth interface in sample E.

Close modal

All samples were characterized by cross-sectional TEM to determine the effects of the various cleaning regimens on the microstructure of the MBE-grown homoepitaxial GaN layers. Of particular concern was the effect of the Ga deposition and desorption, which in every instance caused the RHEED pattern to change from streaky to spotty, suggesting roughening of the substrate surface. This is consistent with an early report in which faceting of the (0001) GaN surface was observed when subjected to Ga deposition and desorption,31 but contrasts with the N-polar GaN surface, which appears much more robust against roughening from Ga deposition and desorption.28,29 Surprisingly, TEM analysis revealed that no new TDs were generated at the regrowth interface in any of the five samples (Fig. 3). Previous investigations have indicated that generation of additional TDs at the N-polar29 and Ga-polar27 homoepitaxial GaN regrowth interfaces can be avoided if the residual sheet densities of oxygen are less than ∼1.5 × 1014 and ∼5 × 1014 cm−2, respectively. The residual sheet densities of oxygen for all samples shown in Table II are clearly well below these thresholds.

Fig. 3.

Cross-sectional TEM images of the near-surface region (a) and the entire thickness of the MBE-grown homoepitaxial layers, the regrowth interface region, and the HVPE-grown GaN substrate (b) of sample A. Corresponding images from sample E are shown in (c) and (d). Note the relatively wavy surface of sample A, which is consistent with the AFM image of Fig. 1(a). No TDs are seen to have been generated at the regrowth interface of any of the samples, nor is the regrowth interface visible in any image from the five samples. The carbon and platinum layers deposited during preparation of the samples for TEM analysis are visible above the MBE GaN layers.

Fig. 3.

Cross-sectional TEM images of the near-surface region (a) and the entire thickness of the MBE-grown homoepitaxial layers, the regrowth interface region, and the HVPE-grown GaN substrate (b) of sample A. Corresponding images from sample E are shown in (c) and (d). Note the relatively wavy surface of sample A, which is consistent with the AFM image of Fig. 1(a). No TDs are seen to have been generated at the regrowth interface of any of the samples, nor is the regrowth interface visible in any image from the five samples. The carbon and platinum layers deposited during preparation of the samples for TEM analysis are visible above the MBE GaN layers.

Close modal

Moreover, the regrowth interface was not visible in the transmission electron micrographs of any of the samples investigated, and there was no evidence of any deleterious effects of the apparent roughening on the microstructure of the subsequently grown homoepitaxial GaN. It should be mentioned that care was taken in this present work to minimize the time that an apparently roughened surface was exposed to high temperature: once the RHEED pattern became spotty, the substrate temperature was reduced and GaN growth was initiated as expeditiously as possible. It is unknown at what point the substrate surface becomes irrecoverably damaged. However, it is clear from the results of the TEM analysis that none of the substrate surfaces in the present investigation were damaged beyond recovery by the pregrowth treatments employed.

Overall, these results indicate that homoepitaxial GaN layers with excellent crystal quality and low impurity incorporation can be grown on freestanding HVPE-grown Ga-polar GaN substrates by rf-MBE using any of several methods to prepare the substrate surface. Heating an as-received substrate wafer to 1100 °C results in a surface which is sufficiently clean to prevent TD generation at the regrowth interface, while Ga deposition and desorption is so effective at reducing O, Si, and C that it may be the preferred method of substrate preparation when homoepitaxial growth of very high purity GaN is desired. Even though the RHEED pattern indicated a roughening of the substrate surface by very gentle use of Ga deposition and desorption, a smooth surface and very high quality microstructure are recovered very quickly after the start of growth. Thus, the crystal grower has wide latitude to balance the requirements of a homoepitaxial GaN layer with the demands for simple and cost-effective methods for preparing the substrates.

We have investigated the effect of various ex situ and in situ methods to prepare the Ga-polar surface of freestanding, low dislocation density, HVPE-grown GaN substrates on the microstructure, impurity incorporation, and surface morphology of subsequent homoepitaxial GaN layers grown by rf-MBE. The surface preparation methods included: an aggressive ex situ wet chemical clean involving a solvent degrease and acid and base etches; in situ metallic gallium deposition and desorption; and in situ thermal cleaning at 1100 °C. All three methods resulted in homoepitaxial layers in which the regrowth interface was not visible by cross-sectional TEM and no new TDs were generated at that interface. Significantly lower impurity concentrations were achieved when Ga deposition and desorption was employed. The lowest impurity concentrations were found when both the ex situ wet chemical clean and in situ Ga deposition and desorption were used to prepare the substrate surface, but the smoothest surface morphology was obtained with Ga deposition and desorption alone.

The work at NRL was supported by the Office of Naval Research. The electron microscopy studies at ASU were supported under contract to Wyle Laboratories as part of Reliability Information Analysis Center Contract No. HC1047-05-D-4005 under the Air Force Research Laboratory Sensors Directorate Technical Task 261 (monitor: Chris Bozada). T.O.M. and D.J.S. acknowledge the use of the facilities at the John M. Cowley Center for High Resolution Electron Microscopy at Arizona State University.

1.
P.
Gibart
,
Rep. Prog. Phys.
67
,
667
(
2004
).
2.
D. F.
Storm
 et al,
J. Cryst. Growth
305
,
340
(
2007
).
3.
D. F.
Storm
 et al,
J. Vac. Sci. Technol. B
23
,
1190
(
2005
).
4.
A.
Chakraborty
,
B. A.
Haskell
,
S.
Keller
,
J. S.
Speck
,
S. P.
DenBaars
,
S.
Nakamura
, and
U. K.
Mishra
,
Jpn. J. Appl. Phys.
44
,
L173
(
2005
).
5.
K.
Okamoto
,
H.
Ohta
,
D.
Nakagawa
,
M.
Sonobe
,
J.
Ichihara
, and
H.
Takasu
,
Jpn. J. Appl. Phys.
45
,
L1197
(
2006
).
6.
C.
Skierbiszewski
,
H.
Turski
,
G.
Muziol
,
M.
Siekacz
,
M.
Sawicka
,
G.
Cywinski
,
Z. R.
Wasilewski
, and
S.
Porowski
,
J. Phys. D: Appl. Phys.
47
,
073001
(
2014
).
7.
Y.
Aolki
,
M.
Kuwabara
,
Y.
Yamashita
,
Y.
Takagi
,
A.
Sugiyama
, and
H.
Yoshida
,
Appl. Phys. Lett.
107
,
151103
(
2015
).
8.
K.
Okamoto
,
J.
Kashiwagi
,
T.
Tanaka
, and
M.
Kubota
,
Appl. Phys. Lett.
94
,
071105
(
2009
).
9.
T.
Miyoshi
,
S.
Masui
,
T.
Okada
,
T.
Yanamoto
,
T.
Kozaki
,
S.-I.
Nagahama
, and
T.
Mukai
,
Appl. Phys. Express
2
,
062201
(
2009
).
10.
Y.
Kuwahara
,
T.
Fujii
,
Y.
Fujiyama
,
T.
Sugiyama
,
M.
Iwaya
,
T.
Takeuchi
,
S.
Kamiyama
,
I.
Akasaki
, and
H.
Amano
,
Appl. Phys. Express
3
,
111001
(
2010
).
11.
Y.
Kuwahara
 et al,
Appl. Phys. Express
4
,
021001
(
2011
).
12.
M.
Mori
 et al,
Proc. SPIE
8262
,
82620Z
(
2012
).
13.
I. C.
Kizilyalli
,
A. P.
Edwards
,
O.
Aktas
,
T.
Prunty
, and
D.
Bour
,
IEEE Trans. Electron Devices
62
,
414
(
2015
).
14.
Y.
Wang
 et al,
Semicond. Sci. Technol.
26
,
022002
(
2011
).
15.
E.
Frayssinet
 et al,
Appl. Phys. Lett.
77
,
2551
(
2000
).
16.
D. F.
Storm
,
J. A.
Roussos
,
D. S.
Katzer
,
J. A.
Mittereder
,
R.
Bass
,
S. C.
Binari
,
D.
Hanser
,
E. A.
Preble
, and
K.
Evans
,
Electron. Lett.
42
,
663
(
2006
).
17.
D. J.
Meyer
 et al,
IEEE Electron Device Lett.
34
,
199
(
2013
).
18.
P.
Kruszewski
 et al,
Semicond. Sci. Technol.
29
,
075004
(
2014
).
19.
T. A.
Growden
,
D. F.
Storm
,
W.
Zhang
,
E. R.
Brown
,
D. J.
Meyer
,
P.
Fakhimi
, and
P. R.
Berger
,
Appl. Phys. Lett.
109
,
083504
(
2016
).
20.
D.
Li
,
L.
Tang
,
C.
Edmunds
,
J.
Shao
,
G.
Gardner
,
M. J.
Manfra
, and
O.
Malis
,
Appl. Phys. Lett.
100
,
252105
(
2012
).
21.
I.
Grzegory
,
J. Phys. Condens. Matter
13
,
6875
(
2001
).
22.
M.
Bockowski
 et al,
J. Cryst. Growth
281
,
11
(
2005
).
23.
R.
Dwiliński
,
R.
Doradziński
,
J.
Garczyński
,
L. P.
Sierzputowski
,
A.
Puchalski
,
Y.
Kanbara
,
K.
Yagi
,
H.
Minakuchi
, and
H.
Hayashi
,
J. Cryst. Growth
310
,
3911
(
2008
).
24.
R.
Dwiliński
,
R.
Doradziński
,
J.
Garczyński
,
L. P.
Sierzputowski
,
A.
Puchalski
,
Y.
Kanbara
,
K.
Yagi
,
H.
Minakuchi
, and
H.
Hayashi
,
J. Cryst. Growth
311
,
3015
(
2009
).
25.
T.
Paskova
and
K. R.
Evans
,
IEEE J. Sel. Top. Quantum Electron.
15
,
1041
(
2009
).
26.
D. F.
Storm
,
D. A.
Deen
,
D. S.
Katzer
,
D. J.
Meyer
,
S. C.
Binari
,
T.
Gougousi
,
T.
Paskova
,
E. A.
Preble
, and
K. R.
Evans
,
J. Cryst. Growth
380
,
14
(
2013
).
27.
D. F.
Storm
,
M. T.
Hardy
,
D. S.
Katzer
,
N.
Nepal
,
B. P.
Downey
,
D. J.
Meyer
,
T. O.
McConkie
,
L.
Zhou
, and
D. J.
Smith
,
J. Cryst. Growth
456
,
121
(
2016
).
28.
D. F.
Storm
,
D. S.
Katzer
,
D. J.
Meyer
, and
S. C.
Binari
,
J. Appl. Phys.
112
,
013507
(
2012
).
29.
D. F.
Storm
,
T. O.
McConkie
,
D. S.
Katzer
,
B. P.
Downey
,
M. T.
Hardy
,
D. J.
Meyer
, and
D. J.
Smith
,
J. Cryst. Growth
409
,
14
(
2015
).
30.
D. F.
Storm
,
D. J.
Meyer
,
D. S.
Katzer
,
S. C.
Binari
,
T.
Paskova
,
E. A.
Preble
,
K. R.
Evans
,
L.
Zhou
, and
D. J.
Smith
,
J. Vac. Sci. Technol. B
30
,
02B113
(
2012
).
31.
V. M.
Bermudez
,
R.
Kaplan
,
M. A.
Khan
, and
J. N.
Kuznia
,
Phys. Rev. B
48
,
2436
(
1993
).