phys. stat. sol. (c) 3, No. 6, 1408 – 1411 (2006) / DOI 10.1002/pssc.

200565121

Effect of carrier gas on GaN epilayer characteristics

Y. S. Cho*, H. Hardtdegen, N. Kaluza, N. Thillosen, R. Steins, Z. Sofer, and H. Lüth
Institute of Thin Films and Interfaces (ISG-1), Center of Nanoelectronic Systems for Information
Technology, Research Center Jülich, 52425 Jülich, Germany

Received 25 July 2005, revised 5 January 2006, accepted 16 January 2006

Published online 10 May 2006

PACS 61.10.Nz, 68.55.Jk, 68.37.Ps, 78.55.Cr, 81.05.Ea, 81.15.Kk

Metalorganic vapor phase epitaxy (MOVPE) of GaN was performed using hydrogen (H2), nitrogen (N2)
and H2/N2mixtures thereof as the carrier gas in the high temperature buffer growth range. The effect of
carrier gas on the structural and morphological characteristics of the epilayers was systematically studied
using interference and atomic force microscopy (AFM), photoluminescence (PL) measurements at 2 K,
Raman spectroscopy and X-ray diffraction (XRD). The higher the N2 content in the carrier gas, the more
pinholes are observed, the lower compressive strain and the higher dislocation density in the layers. A car-
rier gas composition range was defined at which GaN layers with acceptable structural and morphological
characteristics are achieved.

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction
Despite their high dislocation density, GaN and related nitride compounds are extremely important mate-
rials for device applications in light emitting diodes (LEDs) [1], laser diodes (LDs) [2] and high power
and high frequency devices such as high electron mobility transistors (HEMTs) [3]. In order to grow
high quality GaN epilayers, a number of growth parameters have to be optimized and their influence on
growth need to be studied especially with respect to their influence on structural characteristics. Even
though the effect of most of the growth parameters on epilayer characteristics has been clarified the role
of the carrier gas is still discussed quite contradictorily. For the growth of GaN, most groups reported the
use of hydrogen (H2) as carrier gas. Amano et al. used nitrogen (N2) as the ambient gas [4]. Nakamura et
al. reported the use of a gas mixture of H2 and N2 as the sub-flow gas [5]. These GaN epilayers grown in
the presence of N2 show good uniformity and high quality. However, some other groups reported that
pure N2 is not only detrimental to the morphology and the structural quality of GaN but also to the elec-
trical properties [6]. Details of the underlying physical processes, however, are not yet understood.
In this paper, the influences of H2 and N2 as well as H2/N2 carrier gas mixtures are discussed on the
morphological and structural characteristics of GaN epilayers. The study is carried out to clarify why
metalorganic vapor phase epitaxy (MOVPE) of GaN in N2 atmosphere differs from that in H2.

2 Experiment
The growth of undoped GaN epilayers was performed on sapphire (0001) substrates by MOVPE in an
AIX 200/4 RF-S horizontal reactor. The samples were grown using the inverted inlet process [7] using
trimethylgallium (TMGa) and ammonia (NH3) as the Ga and N precursors, respectively. The growth
pressure was maintained at 200 mbar. For all samples a 25 nm thick low temperature GaN nucleation
layer (LT-GaN NL) was deposited at 550 °C in H2 environment. The LT-GaN NL was annealed at
*
Corresponding author: e-mail: [email protected], Phone: +49 2461 61 2991, Fax: +49 2461 61 2940

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

phys. stat. sol. (c) 3, No. 6 (2006) 1409

970 °C for 10 minutes also in H2 carrier gas and ammonia. After the annealing process, growth was re-
sumed and the successive high temperature growth step was performed with different ratios of H2/N2
carrier gas mixtures - 0%, 50%, 70%, 80% and 100% - at 1100 °C for 60 minutes while the total flow
was kept constant at 9 slm. Since the carrier gas had a strong influence on the true surface temperature
[8] - the surface temperature drops successively by 50 °C from the pure H2 carrier gas to the pure N2
carrier gas sample -, heating control temperature was adjusted in order to compensate for this effect and
to achieve the same surface temperature 1100 °C for all five samples.
Growth was monitored by in-situ reflectometry combined with emissivity corrected pyrometry (EpiR-
DA (TT)). The surface morphology was observed ex-situ by Nomarski- on the micrometer scale and
atomic force microscopy (AFM) on the nanometer scale. The structural characteristics of the layers with
respect to strain were studied at first by PL at 2 K using a He-Cd laser with 325 nm wavelength and by
identifying the excitonic line position. The penetration depth of the laser is estimated to be approxi-
mately 100 nm. Since the substrates are transparent, the measurements were carried out from the back-
side of the wafer and from the front-side giving access to the strain state of the first 100 nm and the last
100 nm of GaN growth, respectively. The in-plane biaxial stress of the GaN epilayers was analyzed by
Raman spectroscopy at room temperature. At last the structural characteristics of the GaN epilayers were
evaluated by X-ray diffraction (XRD) using the symmetric (002) and (110) and asymmetric (102) reflec-
tion the full width at half maximum (FWHM) of which are sensitive to screw and mixed types of thread-
ing dislocations (TDs), to pure edge and mixed types of TDs, and to all types of TDs, respectively [9].

3 Results and discussion

3.488
He-Cd laser
tcoalescence = 65 sec.
EDBX peak position (eV)

Front-side PL T = 2K, λ = 325 nm

pure N2
RGaN/Rsapphire (a.u.)

tcoalescence = 320 sec. 3.484

H2/N2=20%/80%

tcoalescence = 550 sec. Back-side PL

3.480
H2/N2=30%/70%

tcoalescence = 720 sec.

H2/N2=50%/50% 3.476
less compressive strain
tcoalescence = 1810 sec.
pure H2
3.472
Fully relaxed GaN, EDBX = 3.466 ~ 3.473 eV
1500 3000 4500 6000 7500
0 20 40 60 80 100
time (sec.)
N2 content in carrier gas (%)

Fig. 1 The effect of N2 on the coalescence time and Fig. 2 The comparison of front- and back-side PL meas-
thickness. With increasing N2 content in the carrier gas, urement. At the interface between GaN epilayer and
the coalescence time (and therefore the “coalescence sapphire substrate, GaN has less compressive strain than
thickness”) decreases drastically. the surface of GaN epilayer.

Figure 1 shows the reflectance transients at 600 nm taken during growth for the sample series. Three
different growth regimes are observed. Before the first dotted line the identical growth steps for all sam-
ples are to be seen: the LT-GaN NL growth and the annealing phase. After the first dotted line, the coa-
lescence phase starts in which the nuclei grow together. Gradually oscillations form which reach a con-
stant intensity. The second dotted line denotes the end of the coalescence phase and the onset of com-
plete 2D growth. With increasing N2 content in the gas phase (from bottom to top) the coalescence time
and therefore the “coalescence thickness” decreases drastically from 1810 to 65 seconds. Also the
growth rate varies and therefore the layer thickness changes from 2.29 to 1.73 µm. Earlier reports point
out that the coalescence phase strongly influences the strain and the number of threading dislocations as
well as the point defects in the layer as well as the surface morphology [10]. Therefore the morphology
of the layers was investigated next.
Although somewhat contradictorily to observations made by other groups [6], the morphology on the
micrometer scale is smooth and specular. On the nanometer scale, however, the surface morphology is

www.pss-c.com © 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1410 Y. S. Cho et al.: Effect of carrier gas on GaN epilayer characteristics

quite different. For pure H2 carrier gas, the surface exhibits the well known straight atomic steps. How-
ever, the higher the N2 content the more the steps become round and pits are observed. This may be due
to a different surface stoichiometry in H2 and N2 carrier gas as has been observed surface photoabsorp-
tion (SPA) spectra [11]. However still studies need to be carried out which relate the spectra to surface
reconstructions.
Since the morphology development in the coalescence phase is closely related to the compressive
strain in the layer [12], the strain was investigated next. The GaN band gap increases in case of biaxial
in-plane compression compared to fully relaxed GaN. The donor bound exciton energy (EDBX) therefore
increases with increasing compressive strain in the layers. In Fig. 2 the EDBX is plotted as a function of
nitrogen content for PL performed on the back-side and on the front-side of the sample. The value for
fully relaxed GaN according to reference values [13, 14] is also included. The results of back-side PL
reveal that near the interface between GaN epilayer and sapphire substrate there is less compressive
strain. There are more point defects near the interface due to LT-GaN nucleation layer, therefore it has
less compressive strain than at the surface. For both front- and backside measurements the position of the
EDBX gradually moves towards lower energies with increasing nitrogen content as the layers show de-
creasing compressive strain. However above 70% N2 content the observed strain relaxation becomes
bigger. The decrease in coalescence thickness with increasing nitrogen content leads to early strain re-
laxation. The coalescence time correlates with the size of the nuclei before neighboring grains coalesce:
the smaller the thickness the smaller the grain size and the more dislocations can be found. Also it be-
comes apparent that the strain is smaller close to the substrate compared to near the surface. It means that
thermally induced stress for non-coalesced films should be very small [12].
Raman measurements are well suited for the non-destructive characterization of the overall stress in
the materials. The intense E2 phonon mode of GaN was chosen as a reference peak for the strain deter-
mination. Figure 3 shows the Raman shift of the E2 phonon mode and biaxial stress in the layers as a
function of nitrogen content in the carrier gas. It was reported by Koh et al. [15] that in-plane biaxial
stress (σ) is given by
σ [GPa ] = -0.22 ¥ (ω - ω 0 ) , (1)
where ω is the measured E2 phonon mode for a sample and ω0 is that for the reference sample, ω0(E2) =
566.34 cm–1. Our reference sample was a 300-µm-thick free-standing GaN substrate prepared by hydride
vapor phase epitaxy (HVPE). The commercial reference sample exhibited a TD density less then 1×108
cm–2 and resistivity less then 30 MΩ⋅cm. At first the Raman shift and biaxial stress gradually decrease
with increasing nitrogen content. Above 70% nitrogen content the decrease becomes more sudden. It
should be noted in Fig. 3 that the Raman shift and the biaxial stress are minimal for the GaN epilayer
grown in pure N2 carrier gas. These Raman results agree well with PL results. Also these Raman and PL
results are well matched with previous reports [16]. It means that we can influence the stress using car-
rier gas. The structural quality was investigated with XRD at last.
The FWHM of XRD ω-scans for the (002), (110) and (102) reflections are presented in Fig. 4. The
FWHM of all three reflections increase with increasing N2 content in carrier gas indicating an increase of
TDs. Above 70% N2 content in carrier gas, the increase becomes stronger for the (110) and (102) reflec-
tions compared to the (002) reflection. Obviously, mainly the edge dislocation density increases with in-
creasing N2 content. The investigations show that a minimal amount of H2 is necessary in the reactor in
order to obtain layers of sufficient structural and morphological quality. Apparently the higher edge disloca-
tion density correlates with strain relaxation due to the coalescence of small grains in the early stage of high
temperature growth [10]. By intentionally using carrier gas mixtures the strain in the layers can be tailored.
In future transmission electron microscopy (TEM) must still be carried out to confirm the results observed.

4 Conclusions
MOVPE of GaN was performed using H2, N2 and H2/N2 mixtures thereof as the carrier gas in the high
temperature buffer growth range. Based on the overall results, an increase of N2 content in the carrier gas

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-c.com

phys. stat. sol. (c) 3, No. 6 (2006) 1411

leads to a change in the surface morphology of the GaN epilayers on the nanometer scale. The shortened
coalescence time affects the TD density in a detrimental way to the layers and reduces the compressive
strain. A sufficient structural and morphological quality of the layers can nevertheless be obtained as
long as the H2 content in the gas phase is at least 30% in the coalescence phase of growth. The influence
of carrier gas on strain can be employed to tailor the strain in the layers which is an important issue in
fabrication of optical and electrical devices. Further TEM studies are necessary to study the evolution of
TDs and point defects during growth.

-1
1100
GaN (110)
Ref. : ω(E2) = 566.34 cm
568.8 1000
GaN (102)
GaN (002)

FWHM (arcsec.)
900
Raman Shift (cm )

568.5
-1

800
568.2 0.55
700
Biaxial Stress (Gpa)

0.50

0.45
567.9 0.40
600
0.35
500
567.6 0.30

0.25
400
0.20
567.3 0 20 40 60 80 100 300
N 2 content in carrier gas (%)
567.0 0 20 40 60 80 100
0 20 40 60 80 100 N2 content in carrier gas (%)
N2 content in carrier gas (%)

Fig. 3 Raman shift of the E2 phonon mode and biaxial Fig. 4 FWHM of X-ray rocking curves of the GaN
stress in the GaN epilayer. epilayers.

Acknowledgements The authors would like to thank Dr. M. von der Ahe and Dr. E. Hollmann for fruitful discus-
sion about XRD measurement, Dr. G. Crecelius for assistance with Raman and K. Wirtz for the technical support as
well as German Academic Exchange Service (DAAD) for the research grant for Ph. D. degree to Y. S. Cho.

References
[1] T. Mukai, H. Narimatsu, and S. Nakamura, Jpn. J. Appl. Phys. 37, L479 (1998).
[2] S. Nakamura, M. Senoh, S. Nagahama, T. Matsushita, H. Kiyoku, Y. Sugimoto, T. Kozaki, H. Umemoto,
M. Sano, and T. Mukai, Jpn. J. Appl. Phys. 38, L226 (1999).
[3] P. Javorka, A. Alam, M. Wolter, A. Fox, M. Marso, M. Heuken, H. Lüth, and P. Kordoš, IEEE Electron Device
Lett. 23, 287 (2002).
[4] H. Amano, N. Sawaki, and I. Akasaki, J. Cryst. Growth 68, 163 (1984).
[5] S. Nakamura, Y. Harada, and M. Senoh, Appl. Phys. Lett. 58, 2021 (1991).
[6] H. X. Wang, Y. Amijima, Y. Ishihama, and S. Sakai, J. Cryst. Growth 233, 681 (2001).
[7] H. Hardtdegen, N. Kaluza, R. Steins, R. Schmidt, K. Wirtz, E. V. Yakovlev, R. A. Talalaev, and Yu. N.
Makarov, J. Cryst. Growth 272, 407 (2004).
[8] H. Hardtdegen, N. Kaluza, R. Steins, Y. S. Cho, Z. Sofer, M. Zorn, K. Haberland, and J.-T. Zettler, phys. stat.
sol. (b) 242, 2581 (2005).
[9] B. Heying, X. H. Wu, S. Keller, Y. Li, D. Kapolnek, B. P. Keller, S. P. DenBaars, and J. S. Speck, Appl. Phys.
Lett. 68, 643 (1996).
[10] T. Böttcher, E. Einfeldt, S. Figge, R. Chierchia, H. Heinke, D. Hommel, and J. S. Speck, Appl. Phys. Lett. 78,
1976 (2001).
[11] Y. Kobayashi and N. Kobayashi, J. Cryst. Growth 189/190, 301 (1998).
[12] S. Einfeldt, T. Böttcher, S. Figge, and D. Hommel, J. Cryst. Growth 230, 357 (2001).
[13] B. Monemar, Phys. Rev. B 10, 676 (1974).
[14] K. Kornitzer, T. Ebner, K. Thonke, R. Sauer, C. Kirchner, M. Kamp, V. Schwegler, M. Leszczynski, I. Grze-
gory, and S. Porowski, Phys. Rev. B 60, 1471 (1999).
[15] E. K. Koh, Y. J. Park, E. K. Kim, C. S. Park, S. H. Lee, J. H. Lee, and S. H. Choh, J. Cryst. Growth 218, 214
(2000).
[16] D. Li, H. Chen, H. B. Yu, Y.J. Han, X. H. Zheng, Q. Huang, and J. M. Zhou, J. Cryst. Growth 263, 76 (2004).

www.pss-c.com © 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim