N-doped homo-epitaxial GaN samples grown on freestanding GaN substrates have been investigated by micro-Raman spectroscopy. Quantitative analysis of the E2h and the A1(LO) modes’ behavior has been performed while intentionally increasing the carrier density using silicon doping. We noticed that as the carrier concentration increases up to 1.8 × 1018 cm−3, the E2h mode remains unchanged. On the other hand, when the doping gets higher, the A1(LO) position shifts to a higher frequency range, its width becomes larger, and its intensity drastically diminishes. This change in the A1(LO) behavior is due to its interaction and its coupling with the free negative charge carriers. Furthermore, we calibrated the A1(LO) frequency position shift as a function of the n-carrier concentration. We found out that for low n doping, the change in the A1(LO) position can be considered as a linear variation while in the overall doping range, a sigmoid growth trend with a Boltzmann fit can be tentatively applied to describe the A1(LO) position shift. This calibration curve can also be used to describe the coupling strength between the carriers and the A1(LO) phonon. Eventually, this study shows that micro-Raman spectroscopy is a powerful non-destructive tool to probe the doping concentration and the crystalline quality of GaN material with a microscopic spatial resolution.

Vertical-type GaN (gallium nitride) power devices are reported to show better electrical performance than their lateral counterparts because of fewer leakage paths, high current density, and better reliability.1–3 The growth of such devices requires good quality homo-epitaxially grown layers on native GaN substrates. To date, native GaN substrates have been manufactured with highly sophisticated technologies such as HVPE (Hydride Vapor Phase Epitaxy), the ammonothermal method, and low-pressure solution growth.4 In addition, homo-epitaxially fabricated vertical GaN devices have been known to sustain more than 1 kV5,6 of blocking voltage depending on the device drift layer thickness. Such an expected device performance is possible through an accurate control of the doping level and the crystalline quality of the active epilayer.7 Indeed, lower doping of the drift layer increases the breakdown voltage in the device, and homogeneous doping induces a uniform electric field distribution in the device. A physical characterization approach of the doping concentration and the structural defects is, therefore, necessary to understand the roots of the electrical performance limitations of vertical GaN devices.

Micro-Raman spectroscopy is a powerful optical non-destructive tool to study the crystalline quality of GaN material by probing different phonon vibrational modes. In backscattering geometry and for a hexagonal GaN structure, the E2h mode is analyzed to evaluate the biaxial stress in GaN8–10 because of its non-polar behavior while the A1(LO) mode is more sensitive to the doping level and allows us to determine the free negative charge carrier concentration due to its longitudinal and polar features.11–13 In the literature, many studies on GaN crystal quality using the Raman tool have been conducted mainly on hetero-epitaxial samples.14,15 Only a few of them have given detailed explanations of the physics underlying the effect of the charge carrier density on the homo-epitaxial GaN crystal quality when a doping concentration ranges between 1017 and 1019 cm−3.16,17 In this range, the A1(LO) behavior is reported to be drastically changing with the carrier density due to a strong coupling with the plasmons. Indeed, at high doping concentration, the A1(LO) mode becomes an LOPC mode (longitudinal optical phonon-plasmon coupled).18 However, to our knowledge, investigations in the low doping range of 1014–1017 cm−3 are still fully lacking in the literature. Vertical GaN SBDs (Schottky Barrier Diodes) have been reported to always suffer from high reverse leakage current and premature breakdown voltage. To solve that challenge, one solution is to lower the doping concentration of the drift layer below 1017 cm−3.19,20

Here, we used the micro-Raman tool to quantitatively investigate the quality of Si-doped homo-epitaxial GaN samples from an unintentionally doped state n < 1014 cm−3 to a highly doped one n > 1018 cm−3. By tracking the position and the FWHM (Full Width at Half Maximum) of the E2h and A1(LO) Raman modes as the carrier density increases, the quality of the homo-epitaxial samples was carefully probed. Furthermore, calibration of the A1(LO) position as a function of Si concentration was performed to quantify its sensitivity at low and high doping levels, and then we deduced the physical mechanism underlying the A1(LO) behavior vs the carrier density. Thus, this study shows that the micro-Raman tool is an efficient tool to examine the effect of doping concentration on the crystal quality in GaN material.

A batch of eight homo-epitaxial GaN samples was grown by the MOCVD (Metal Organic Chemical Vapor Deposition) method21 on 50 mm diameter highly n-doped (around 2 × 1018 cm−3) freestanding GaN substrates provided by Saint Gobain Lumilog.22 GaN films were grown in a close-coupled showerhead reactor. Ammonia, trimethylgallium, and hydrogen carrier gas were used to grow the films at 1020°C at a growth rate of 2 µm h−1. Diluted silane was added to the vapor phase in order to dope the GaN films. Before the growth of the thick low-doped films, a thin highly doped GaN buffer layer of 0.1 µm thickness was grown on top of the substrate [see Fig. 1(a)]. The epilayer’s thickness was kept at 5 µm (except for sample Epi 04 that was 10 µm) with the Si carrier concentration ranging from a nonintentionally doped (nid) level <1014 cm−3 to a high-doped state of 1.8 × 1018 cm−3. As a stress-free sample reference, the non-intentionally doped sample was considered. Prior to the Raman measurements, the capacitance–voltage (C–V) method was used to determine the doping level (Nd–Na measurement). Micro-Raman spectroscopy measurements were carried out at room temperature. For that, a confocal micro-Raman spectrometer (Renishaw Invia model) was used in backscattering geometry, with a ×100 objective and a 2400 l/mm diffraction grating. All the samples were excited with a 532 nm laser source (maximum power of 10 mW to avoid local heating). The spectral and spatial resolutions were around 0.1 cm−1 and 1 µm, respectively. Prior to any measurements, calibration of the Raman shift was carried out with a silicon reference at 520.5 cm−1. During measurements, Raman mappings of 150 × 150 µm2 square dimension with a step size of 3 µm were made on different zones of each sample. An average of more than 2600 spectra extracted from each Raman map was obtained. The quantitative analysis of the E2h and the A1(LO) modes has been made by processing and fitting these spectra by a mixed Gaussian–Lorentzian function with the Wire 5 Renishaw software. In addition, we have drawn and visualized the E2h and the A1(LO) position maps. Finally, we then considered the mean values of the E2h and the A1(LO) position and the FWHM deduced from all the Raman maps in our analyses. Table I shows the summary of the values extracted from the data processing. The calculated errors are the systematic errors based on the mean values.

FIG. 1.

(a) Cross section image of the homo-epitaxial sample. (b) Optical image of the sample surface. (c) Raman mapping of the A1(LO) position of one zone of the sample.

FIG. 1.

(a) Cross section image of the homo-epitaxial sample. (b) Optical image of the sample surface. (c) Raman mapping of the A1(LO) position of one zone of the sample.

Close modal
TABLE I.

Summary of the E2h and the A1(LO) mean values derived from the Gauss–Lorentzian fitting.

SampleSi carrier concentration (cm−3)A1(LO) position (cm−1)A1(LO) width (cm−1)E2(h) position (cm-1)E2(h) width (cm−1)
Epi 01 ≤1014 733.3 ± 0.1 6.3 ± 0.1 567.1 ± 0.1 3.5 ± 0.1 
Epi 02 8 × 1015 733.4 ± 0.1 6.9 ± 0.1 567.1 ± 0.1 3.3 ± 0.1 
Epi 03 1.5 × 1016 733.6 ± 0.1 7 ± 0.1 567.1 ± 0.1 3.5 ± 0.1 
Epi 04 5 × 1016 734 ± 0.2 7.1 ± 0.1 567.1 ± 0.1 3.5 ± 0.1 
Epi 05 7 × 1016 734.2 ± 0.2 7.2 ± 0.1 567.1 ± 0.1 3.5 ± 0.1 
Epi 06 2.4 × 1017 737.3 ± 0.1 10.9 ± 0.3 567.1 ± 0.1 3.5 ± 0.1 
Epi 07 4.7 × 1017 743.6 ± 0.2 14.9 ± 0.7 567.1 ± 0.1 3.5 ± 0.1 
Epi 08 1.8 × 1018 764.2 ± 1.2 48 ± 1.8 567.1 ± 0.1 3.5 ± 0.1 
SampleSi carrier concentration (cm−3)A1(LO) position (cm−1)A1(LO) width (cm−1)E2(h) position (cm-1)E2(h) width (cm−1)
Epi 01 ≤1014 733.3 ± 0.1 6.3 ± 0.1 567.1 ± 0.1 3.5 ± 0.1 
Epi 02 8 × 1015 733.4 ± 0.1 6.9 ± 0.1 567.1 ± 0.1 3.3 ± 0.1 
Epi 03 1.5 × 1016 733.6 ± 0.1 7 ± 0.1 567.1 ± 0.1 3.5 ± 0.1 
Epi 04 5 × 1016 734 ± 0.2 7.1 ± 0.1 567.1 ± 0.1 3.5 ± 0.1 
Epi 05 7 × 1016 734.2 ± 0.2 7.2 ± 0.1 567.1 ± 0.1 3.5 ± 0.1 
Epi 06 2.4 × 1017 737.3 ± 0.1 10.9 ± 0.3 567.1 ± 0.1 3.5 ± 0.1 
Epi 07 4.7 × 1017 743.6 ± 0.2 14.9 ± 0.7 567.1 ± 0.1 3.5 ± 0.1 
Epi 08 1.8 × 1018 764.2 ± 1.2 48 ± 1.8 567.1 ± 0.1 3.5 ± 0.1 

Figure 2 shows the whole Raman spectra recorded at the center of each sample in backscattering geometry. Two main peaks can be observed: the E2h and the A1(LO) peaks. This confirms the hexagonal crystal structure of the films. The small peak at the shoulder of E2h can be ascribed to the E1 (TO) (558 cm−1) mode.23 The Raman spectra have been normalized with respect to the E2h peak intensity. As reported in the literature, the E2h mode can be used to probe the biaxial stress in GaN because of its non-polar behavior24 while the A1(LO) is more sensitive to the doping level and allows us to determine the free carrier concentration due to its longitudinal and polar features.8 

FIG. 2.

Raman spectra of the Si-doped epilayers.

FIG. 2.

Raman spectra of the Si-doped epilayers.

Close modal

The E2h mode behavior shows no change as the carrier density increases. From the quantitative point of view (Table I), the E2h position and width remain constant with average values of 567.1 and 3.5 cm−1, respectively, for all the studied samples. Indeed, as compared to Ref. 25, a Si-doping of up to 1.8 × 1018 cm−3 does not disturb the crystalline quality (namely, the stress and the arrangement of dislocations) of the GaN layers grown on the GaN substrate. We can therefore notice no influence of the Si impurities on the structural quality of the samples.

However, the A1(LO) mode was sensitive to the free carrier density. When zooming in on the A1(LO) mode features (Fig. 3), we clearly observe that as the Si concentration increases at moderate levels from 1014 to 7 × 1016 cm−3, the A1(LO) peak position shifts slightly by +0.9 cm−1 (deduced from Table I) as well as the width increases by a small amount of +0.9 cm−1 (deduced from Table I). This position shift and the width broadening, even though small, are only seen for the A1(LO) mode and not on the E2h mode. This reveals that these changes observed in the A1(LO) behavior are not due to the presence of dislocations or the residual stress effect. Thus, we conclude that they can be solely attributed to a variation in the doping level.

FIG. 3.

Zoomed-in view of the A1(LO) Raman peak as the carrier density increases.

FIG. 3.

Zoomed-in view of the A1(LO) Raman peak as the carrier density increases.

Close modal

On the other hand, at higher doping levels starting from 2.4 × 1017 up to 1.8 × 1018 cm−3; the shift of the A1(LO) peak position becomes important, and the width gets larger. Indeed, a gap of 30.9 and 41.7 cm−1 can be seen in the A1(LO) peak position and width respectively. Moreover, the peak intensity is decreasing when the doping increases. These changes in the A1(LO) behavior as a function of the n-carrier concentration are due to the coupling of the A1(LO) mode with the carrier oscillation, called plasmons, through their macroscopic electric field,11,19 changing the A1(LO) mode to the LOPC mode. Indeed, as the carrier density is lower, the coupling of their electric field with that of the A1(LO) mode is weak. This is the origin of the small variations in the A1(LO) position and its width for such a doping level. Meanwhile, when the carrier density increases, their electric field coupling with one of the A1(LO) modes becomes stronger, inducing a high frequency shift and broadening of the A1(LO) peak.

Moreover, as the A1(LO) mode evolves with the free carrier density, we calibrated its frequency position as a function of the carrier concentration. Figure 4 shows the evolution of the A1(LO) peak position as a function of the effective n-carrier density (in log scale). We found out that the overall trend of the curve follows a sigmoid growth trend, with a Boltzmann fitting as the best matching fit. In detail, we can see from the curve that at light doping levels of up to 7 × 1016 cm−3, the A1(LO) position does not shift significantly while at high doping levels of up to 1.8 × 1018 cm−3, its position shifts at a higher wavenumber. This means that at higher doping levels, as the carrier density is higher, the coupling with the A1(LO) mode is stronger, and therefore, the LOPC mode appears completely. In the transition region from 7 × 1016 to 2.4 × 1017 cm−3, the plot shows a concave curvature that may indicate the increase in the coupling between the carrier density and the A1(LO) mode. We suggest that this transition stands for the change in the A1(LO) mode to the LOPC mode.

FIG. 4.

Calibration of the A1(LO) position as function of the n-doping concentration in log scale.

FIG. 4.

Calibration of the A1(LO) position as function of the n-doping concentration in log scale.

Close modal

Most importantly, when zooming in on the light doping region (Fig. 5), we found out that a linear approximation could be considered as a fitting option, which we expressed in terms of linear equation as

ω1=1.4×1017n+ω0,
(1)

where n is the n-carrier concentration in cm−3 and ω1 [A1(LO) is the Raman shift in cm−1] and ω0 are offset values of 733.3 cm−1 deduced from the plot. The linear approximation in the light doping region may be due to the weak coupling strength between the A1(LO) electric field and the small carrier density in that region. This linear calibration for the light doping region is newly reported for thin homo-epitaxial GaN layers on freestanding substrates, and the considered doping level n < 1017 cm−3 falls within the useable doping range of GaN Schottky diodes.26 This light carrier doping is a crucial parameter to get a higher breakdown voltage of vertical GaN Schottky diodes. In the future, we intend to use this calibration to probe the doping level and the doping homogeneity of GaN epilayers for GaN on GaN vertical Schottky diodes by making 2D maps throughout their respective surface; because homogeneous high quality GaN drift layers (epilayers) are necessary for making high voltage vertical devices,27 this method would also be adapted to heteroepitaxial GaN samples.

FIG. 5.

Zoomed-in view of the light doping region n < 1017 cm−3 with a linear approximation fit.

FIG. 5.

Zoomed-in view of the light doping region n < 1017 cm−3 with a linear approximation fit.

Close modal

In this study, we show that Raman spectroscopy can be used to probe the structural and electronic properties of n-doped epilayers grown on freestanding GaN substrates by quantitative analysis of both the A1(LO) and E2h position and the FWHM as the carrier density increases from light doping to higher doping. The A1(LO) behavior changes drastically as a function of carrier concentration; especially at high doping, its position, intensity, and width shift significantly. Meanwhile, the E2h mode remains unchanged as the carrier concentration increases. Furthermore, calibration of the A1(LO) position as a function of carrier density was performed to monitor its sensitivity to the doping level. Most importantly, we showed that a linear approximation could be considered to express the A1(LO) peak position as a function of the n-doping concentration for n < 1017 cm−3. In the future, we will prospectively examine the homogeneity of the n doping concentration and the crystalline quality in the light doping region of vertical GaN Schottky diodes.

This work was supported by the French Technology Facility Network RENATECH, the French National Research Agency (ANR) through the project C-Pi-GaN (Grant No. ANR-18-CE05-0045) and the “Investissements d’Avenir” Program GaNeX (Grant No. ANR-11-LABX-0014), and the AuRA Region (Région Auvergne-Rhône Alpes) through the OptiGaN project.

The authors have no conflicts to disclose.

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

1.
L.
Sang
,
B.
Ren
,
M.
Sumiya
,
M.
Liao
,
Y.
Koide
,
A.
Tanaka
,
Y.
Cho
,
Y.
Harada
,
T.
Nabatame
,
T.
Sekiguchi
, and
S.
Usami
, “
Initial leakage current paths in the vertical-type GaN-on-GaN Schottky barrier diodes
,”
Appl. Phys. Lett.
111
(
12
),
122102
(
2017
).
2.
R.
Yeluri
,
J.
Lu
,
C. A.
Hurni
,
D. A.
Browne
,
S.
Chowdhury
,
S.
Keller
,
J. S.
Speck
, and
U. K.
Mishra
, “
Design, fabrication, and performance analysis of GaN vertical electron transistors with a buried p/n junction
,”
Appl. Phys. Lett.
106
(
18
),
183502
(
2015
).
3.
Y.
Zhang
,
M.
Sun
,
Z.
Liu
,
D.
Piedra
,
H.-S.
Lee
,
F.
Gao
,
T.
Fujishima
, and
T.
Palacios
, “
Electrothermal simulation and thermal performance study of GaN vertical and lateral power transistors
,”
IEEE Trans. Electron Devices
60
(
7
),
2224
2230
(
2013
).
4.
T.
Paskova
,
D. A.
Hanser
, and
K. R.
Evans
, “
GaN substrates for III-nitride devices
,”
Proc. IEEE
98
(
7
),
1324
1338
(
2009
).
5.
D.
Shibata
,
R.
Kajitani
,
M.
Ogawa
,
K.
Tanaka
,
S.
Tamura
,
T.
Hatsuda
,
M.
Ishida
, and
T.
Ueda
, “
1.7 kV/1.0 mΩ cm2 normally-off vertical GaN transistor on GaN substrate with regrown p-GaN/AlGaN/GaN semipolar gate structure
,” in
IEEE International Electron Devices Meeting (IEDM)
(
IEEE
,
2016
), pp.
248
251
.
6.
H.
Nie
,
Q.
Diduck
,
B.
Alvarez
,
A. P.
Edwards
,
B. M.
Kayes
,
M.
Zhang
,
G.
Ye
,
T.
Prunty
,
D.
Bour
, and
I. C.
Kizilyalli
, “
1.5-kV and 2.2 mΩ-cm2 vertical GaN transistors on bulk-GaN substrates
,”
IEEE Electron Device Lett.
35
(
9
),
939
941
(
2014
).
7.
H.
Fu
,
X.
Huang
,
H.
Chen
,
Z.
Lu
,
I.
Baranowski
, and
Y.
Zhao
, “
Ultra-low turn-on voltage and on-resistance vertical GaN-on-GaN Schottky power diodes with high mobility double drift layers
,”
Appl. Phys. Lett.
111
(
15
),
152102
(
2017
).
8.
F.
Demangeot
,
J.
Frandon
,
M. A.
Renucci
,
O.
Briot
,
B.
Gil
, and
R. L.
Aulombard
, “
Raman determination of phonon deformation potentials in α-GaN
,”
Solid State Commun.
100
(
4
),
207
210
(
1996
).
9.
J.-M.
Wagner
and
F.
Bechstedt
, “
Phonon deformation potentials of α-GaN and -AlN: An ab initio calculation
,”
Appl. Phys. Lett.
77
(
3
),
346
348
(
2000
).
10.
K. M.
Chen
,
Y. H.
Yeh
,
Y. H.
Wu
,
C. H.
Chiang
,
D. R.
Yang
,
Z. S.
Gao
,
C. L.
Chao
,
T. W.
Chi
,
Y. H.
Fang
,
J. D.
Tsay
, and
W. I.
Lee
, “
Stress and defect distribution of thick GaN film homoepitaxially regrown on free-standing GaN by hydride vapor phase epitaxy
,”
Jpn. J. Appl. Phys., Part 1
49
(
9
),
091001
(
2010
).
11.
T.
Kozawa
,
T.
Kachi
,
H.
Kano
,
Y.
Taga
,
M.
Hashimoto
,
N.
Koide
, and
K.
Manabe
, “
Raman scattering from LO phonon‐plasmon coupled modes in gallium nitride
,”
J. Appl. Phys.
75
(
2
),
1098
(
1994
).
12.
C.
Wetzel
,
W.
Walukiewicz
,
E. E.
Haller
,
J.
Ager
 III
,
I.
Grzegory
,
S.
Porowski
, and
T.
Suski
, “
Carrier localization of as-grown n-type gallium nitride under large hydrostatic pressure
,”
Phys. Rev. B
53
(
3
),
1322
(
1996
).
13.
D.
Olego
and
M.
Cardona
, “
Raman scattering by coupled LO-phonon plasmon modes and forbidden TO-phonon Raman scattering in heavily doped p-type GaAs
,”
Phys. Rev. B
24
(
12
),
7217
(
1981
).
14.
M.
Park
,
J. J.
Cuomo
,
B. J.
Rodriguez
,
W.-C.
Yang
,
R. J.
Nemanich
, and
O.
Ambacher
, “
Micro-Raman study of electronic properties of inversion domains in GaN-based lateral polarity heterostructures
,”
J. Appl. Phys.
93
(
12
),
9542
9547
(
2003
).
15.
A.
Hushur
,
M. H.
Manghnani
, and
J.
Narayan
, “
Raman studies of GaN/sapphire thin film heterostructures
,”
J. Appl. Phys.
106
(
5
),
054317
(
2009
).
16.
H.
Harima
,
H.
Sakashita
, and
S.
Nakashima
, “
Raman microprobe measurement of under-damped LO-phonon-plasmon coupled mode in n-type GaN
,”
Mater. Sci. Forum
264-268
,
1363
1366
(
1998
).
17.
E.
Richter
,
T.
Stoica
,
U.
Zeimer
,
C.
Netzel
,
M.
Weyers
, and
G.
Tränkle
, “
Si doping of GaN in hydride vapor-phase epitaxy
,”
J. Electron. Mater.
42
(
5
),
820
825
(
2013
).
18.
H.
Yamamoto
,
K.
Agui
,
Y.
Uchida
,
S.
Mochizuki
,
T.
Uruma
,
N.
Satoh
, and
T.
Hashizume
, “
Evaluation of carrier concentration reduction in GaN-on-GaN wafers by Raman spectroscopy and Kelvin force microscopy
,”
Jpn. J. Appl. Phys., Part 2
56
(
8S1
),
08LB07
(
2017
).
19.
H.
Harima
,
H.
Sakashita
,
T.
Inoue
, and
S.-i.
Nakashima
, “
Electronic properties in doped GaN studied by Raman scattering
,”
J. Cryst. Growth
189-190
,
672
676
(
1998
).
20.
B. J.
Baliga
, “
Gallium nitride devices for power electronic applications
,”
Semicond. Sci. Technol.
28
(
7
),
074011
(
2013
).
21.
G.
Gomila
and
O. M.
Bulashenko
, “
Effects of the epitaxial layer thickness on the noise properties of Schottky barrier diodes
,”
J. Appl. Phys.
86
(
2
),
1004
1012
(
1999
).
22.
K.
Fujito
,
S.
Kubo
,
H.
Nagaoka
,
T.
Mochizuki
,
H.
Namita
, and
S.
Nagao
, “
Bulk GaN crystals grown by HVPE
,”
J. Cryst. Growth
311
(
10
),
3011
3014
(
2009
).
23.
V. S.
Ban
, “
Mass spectrometric and thermodynamics studies of the CVD of some III–V compounds
,”
J. Cryst. Growth
17
,
19
30
(
1972
).
24.
M.
Amilusik
,
D.
Wlodarczyk
,
A.
Suchocki
, and
M.
Bockowski
, “
Micro-Raman studies of strain in bulk GaN crystals grown by hydride vapor phase epitaxy on ammonothermal GaN seeds
,”
Jpn. J. Appl. Phys., Part 2
58
(
SC
),
SCCB32
(
2019
).
25.
X.
Zhihao
,
Z.
Jincheng
,
D.
Huantao
,
Z.
Zhongfen
,
Z.
Qingwei
,
X.
Hao
, and
H.
Yue
, “
Stress, structural and electrical properties of Si-doped GaN film grown by MOCVD
,”
J. Semicond.
30
(
12
),
123003
(
2009
).
26.
T.
Munir
,
A. A.
Aziz
, and
M. J.
Abdullah
, “
Epilayer thickness and doping density variation effects on current-voltage (IV) characteristics of n-GaN Schottky diode
,” in
IEEE International Conference on Semiconductor Electronics
(
IEEE
,
2006
), pp.
892
895
.
27.
K. P.
Schoen
,
J. M.
Woodall
,
J. A.
Cooper
, and
M. R.
Melloch
, “
Design considerations and experimental analysis of high-voltage SiC Schottky barrier rectifiers
,”
IEEE Trans. Electron Devices
45
(
7
),
1595
1604
(
1998
).