Surface electron accumulation in indium nitride layers grown by high pressure chemical vapor deposition

Article abstract of DOI:10.1016/j.susc.2007.07.018

Surface termination and electronic properties of InN layers grown by high pressure chemical vapor deposition have been studied by high resolution electron energy loss spectroscopy (HREELS). HREEL spectra from InN after atomic hydrogen cleaning show N-H termination with no indium overlayer or droplets and indicate that the layer is N-polar. Broad conduction band plasmon excitations are observed centered at 3400 cm^-1 in HREEL spectra with 7 eV incident electron energy which shift to 3100 cm^-1 when the incident electron energies are 25 eV or greater. The shift of the plasmon excitations to lower energy when electrons with larger penetration depths are used is due to a higher charge density on the surface compared with the bulk, that is, a surface electron accumulation. These results indicate that surface electron accumulation on InN does not require excess indium or In-In bonds.

Full text of DOI:10.1016/j.susc.2007.07.018

Surface Science 601 (2007) L120–L123
www.elsevier.com/locate/susc
Surface Science Letters
Surface electron accumulation in indium nitride layers grown
by high pressure chemical vapor deposition
*
R.P. Bhatta, B.D. Thoms , M. Alevli, N. Dietz
Department of Physics and Astronomy, Georgia State University, Atlanta, GA 30303, United States
Received 30 March 2007; accepted for publication 11 July 2007
Available online 21 July 2007
Abstract
Surface termination and electronic properties of InN layers grown by high pressure chemical vapor deposition have been studied by
high resolution electron energy loss spectroscopy (HREELS). HREEL spectra from InN after atomic hydrogen cleaning show N–H ter-
mination with no indium overlayer or droplets and indicate that the layer is N-polar. Broad conduction band plasmon excitations are
observed centered at 3400 cmÀ1 in HREEL spectra with 7 eV incident electron energy which shift to 3100 cmÀ1 when the incident elec-
tron energies are 25 eV or greater. The shift of the plasmon excitations to lower energy when electrons with larger penetration depths are
used is due to a higher charge density on the surface compared with the bulk, that is, a surface electron accumulation. These results indi-
cate that surface electron accumulation on InN does not require excess indium or In–In bonds.
ꢀ 2007 Elsevier B.V. All rights reserved.
Keywords: Electron energy loss spectroscopy; Vibrational spectroscopy; Indium nitride; Hydrogen; Surface structure; Surface electronic phenomena
The physical properties of InN have attracted consider-
able research interest because of its transport properties
such as high electron mobility and saturation velocity mak-
ing it a promising candidate for use in high-speed devices
[1,2]. In addition, InN–GaN–AlN alloys and heterostruc-
tures will enable unique optoelectronic devices, operating
from near infrared to ultraviolet wavelength regime [1,3].
However, the range of device applications strongly depends
on the material properties and quality and on the extent to
which indium-rich alloys and heterostructures can be
formed [3].
The growth of InN is challenging since InN and indium-
rich group III-nitride alloys exhibit a high equilibrium vapor
pressure of nitrogen during growth and a low decomposi-
tion temperature compared to GaN and AlN. These
challenges also lead to difficulties in growing structures con-
taining indium-rich III-nitrides embedded in layers of lower
indium content [4]. InN films have been grown by a number
of techniques including metal–organic chemical vapor depo-
sition (MOCVD) and molecular beam epitaxy (MBE) but
many of the fundamental electrical and optical properties
are still in debate including the correct band gap energy
[1,3]. High pressure chemical vapor deposition (HPCVD)
has been developed to address these challenges and has been
shown to be capable of growing high-quality InN [4–6].
Using HPCVD, stoichiometry controlled InN layers have
been grown at temperatures around 1100 K [6].
There have been several reports regarding surface elec-
tron accumulation on InN layers. A number of investigations
of carrier concentration as a function of film thickness have
measured charge accumulation layers with electron densities
over the range of 1.6–5.1 · 1013 cmÀ2 [7–9]. Lu et al. [7] also
reported ohmic contacts for a number of metals without
annealing and suggested that surface electron accumulation
due to pinning of the Fermi level above the conduction band
minimum was responsible. A series of papers by McConville
and coworkers [10–15] report surface electron accumula-
tion as measured by high resolution electron energy loss
spectroscopy (HREELS) and X-ray photoemission spectros-
copy on MBE-grown InN layers. Colakerol et al. [16] used
*
Corresponding author. Tel.: +1 404 4136045; fax: +1 404 4136025.
E-mail address: bthoms@gsu.edu (B.D. Thoms).
0039-6028/$ - see front matter ꢀ 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2007.07.018

R.P. Bhatta et al. / Surface Science 601 (2007) L120–L123
L121
angle-resolved photoemission spectroscopy to observe quan-
tized energy levels in the surface electron accumulation layer
on InN films grown by radio frequency plasma-assisted
molecular beam epitaxy.
the normal to remove surface carbon contamination from
exposure to air during transport to the UHV chamber.
When Auger electron spectroscopy showed removal of
nearly all surface carbon, the sample was treated using
atomic hydrogen cleaning (AHC) [10]. Atomic hydrogen
exposures were performed by backfilling the vacuum cham-
ber with hydrogen to a pressure of 8.4 · 10À7 Torr in the
presence of a tungsten filament heated to 1850 K. The sam-
ple was positioned 20 mm from the filament for 20 min
(giving an exposure of 1000 L of H2) during which time
the sample temperature rose to about 350 K due to proxim-
ity to the heated filament. After this, the sample was heated
to 600 K while remaining in front of the tungsten filament
for an additional 20 min (an additional 1000 L of H2). Au-
ger electron spectroscopy (AES) and HREELS showed
that surface carbon and oxygen contaminants had been re-
moved. Low energy electron diffraction (LEED) yielded a
hexagonal 1 · 1 pattern over the entire surface of the film
using electron energies from 40 to 170 eV indicating good
surface order and c-axis orientation of the InN layer.
HREEL spectra from the AHC-treated InN layer were
acquired using a range of incident electron energies and
are shown in Fig. 1. In all the spectra, a strong loss feature
was observed at 560 cmÀ1 which is assigned to Fuchs–Klie-
wer surface phonon excitations in agreement with previous
Surface electron accumulation has been reported to be
due to a particularly low C-point conduction band mini-
mum in both InAs [17] and InN [14]. In the case of InAs,
the surface electron concentration is reported to depend
on surface reconstruction [17,18]. The carrier concentration
of an InN epifilm grown by MOCVD has been altered
by exposure to a microwave hydrogen plasma [19]. The
changes were suggested as due to the removal of surface
contaminants such as oxygen and by the migration of in-
dium clusters to the surface during etching. Indium clusters
have been reported to be present in the bulk of InN layers
[20,21] and both indium overlayers and droplets have been
reported on InN surfaces [22,23]. The influence of In over-
layers or droplets on electron accumulation at the InN sur-
face is not yet understood. Segev and Van de Walle [24]
have suggested that occupied surface states caused by In–
In bonds are the cause of electron accumulation on InN
surfaces.
In an earlier HREELS study of HPCVD-grown InN,
the authors reported surface N–H termination and con-
cluded that the film was N-polar [25]. The HREELS studies
reported here were performed on different HPCVD InN
layers grown under different process conditions, which also
exhibit N-polarity. We report here surface electron accu-
mulation on an InN layer as observed by HREELS on a
surface of known termination with no indium overlayers
or droplets.
The results presented here were acquired from an InN
layer grown at a temperature of 1120 K, a reactor pressure
of 15 bar, and an ammonia to trimethyl-indium precursor
ratio of 630 on a HPCVD-grown GaN buffer layer on a sap-
phire (0001) substrate. Details of the HPCVD reactor, the
growth configuration, as well as real-time optical character-
ization techniques employed have been published elsewhere
[4,5]. The spectra reported here were all acquired from the
same sample, however similar results were observed from
other InN films grown under similar conditions.
The ultrahigh vacuum (UHV) surface characterization
system [26] had a base pressure of 1.8 · 10À10 Torr. The
InN sample was rinsed with isopropyl alcohol and then
attached to a tantalum sample holder using tantalum clips
before introduction into the UHV chamber. Sample heating
was accomplished by electron bombardment of the back of
the tantalum sample holder. Sample temperature was mea-
sured with a chromel–alumel thermocouple attached to the
mount next to the sample. HREELS experiments were
performed in a specular scattering geometry with an inci-
dent and scattered angle of 60ꢁ from the normal and inci-
dent electron energy ranging from 7 eV to 35 eV. An
instrumental resolution of 60 cmÀ1 was obtained as deter-
mined from the elastic peak full width at half maximum.
The surface of the InN sample was first cleaned by
bombardment of 1 keV Ar ions at an angle of 70ꢁ from
Fig. 1. HREELS from an atomic hydrogen cleaned InN sample. Spectra
were acquired in the specular direction with incident electron energies of
(a) 7 eV, (b) 15 eV, (c) 25 eV, and (d) 35 eV.

L122
R.P. Bhatta et al. / Surface Science 601 (2007) L120–L123
reports [10-15,25]. The HREEL spectrum taken using 7 eV
incident electron energy shows surface adsorbate loss fea-
tures at 870 and 3260 cmÀ1 due to the bending and stretch-
ing vibrations of a surface N–H species [25]. A small peak
observed near 1430 cmÀ1 is assigned to a combination loss
of the Fuchs–Kliewer phonon and the N–H bending vibra-
tion. With the incident electron energy increased to 15 eV,
the intensity of the N–H stretch decreased considerably
while the N–H bend was almost unchanged. The difference
in the dependence of the intensity of the two peaks on inci-
dent energy suggests that different scattering mechanisms
are contributing. According to Ibach and Mills [27], the di-
pole scattering cross-section decreases with the increasing
incident energy while the impact scattering cross-section
is less predictable and may even increase. Therefore, the
excitation of the N–H stretch is attributed primarily to di-
pole scattering while the N–H bend has a considerable con-
tribution due to impact scattering. As the incident energy is
increased further, all modes due to surface vibrations de-
crease due to reduction in surface sensitivity with an
increasing penetration depth of the electrons.
In general, the shift of the plasmon loss feature in
HREEL spectra with incident electron energy may be
attributed to three causes: dispersion, the kinematic factor,
and variation of the carrier concentration with depth. Fro-
itzheim et al. [33] have reported that if the parallel momen-
tum transfer vector is much smaller than the reciprocal
lattice vector then the wave vector dependence of dielectric
function e(x,q) can be ignored. Under the conditions used
in this study, the parallel momentum transfer vector is
small (60.04 AÀ1), so the role of dispersion in the shift of
˚
the plasmon peak position can be excluded.
According to Chen et al. [34], if the loss function con-
tains a broad frequency response due to large damping
constant, C, which is typical for low mobility p-type semi-
conductors, then the kinematic effects play a role in deter-
mining the peak position for incident energies less than
5 eV. Since the present work was performed on an n-type
semiconductor of higher carrier mobility (l ꢀ 210 cm2/
V s) using incident electron energies of 7–35 eV, the kine-
matic factor is not important in determining the energy
of the plasmon loss feature.
In–H stretching vibrations, typically found between
1650 and 1700 cmÀ1 [28–31], are not observed in these
spectra even with 7 eV incident electron energy. Since
hydrogen desorption may occur from indium sites at the
600 K temperature used for AHC, additional room temper-
ature hydrogen atom exposures were performed. HREEL
spectra from the room temperature H-dosed surface were
unchanged from those shown in Fig. 1 indicating that no
reactive indium exists on this surface. The presence of N–
H and lack of indium-related vibrations demonstrates N-
termination of the surface and implies N-polarity of the
film since N-termination would not be expected for In-po-
lar films [32]. Along with the observed 1 · 1 surface order,
this also shows that the surface is free of both indium over-
layers and droplets.
The observation of an N–H terminated surface free of
excess indium differs from the conclusion of several recent
reports. An investigation by Draxler et al. [23] using coax-
ial impact collision ion scattering spectroscopy (CAICISS)
determined that metallic In was present on N-polar InN
surfaces after cleaning by atomic hydrogen. Piper et al.
[22] also concluded that metallic In is sometimes present
on the InN surface after AHC based on X-ray photoemis-
sion spectroscopy. They assign a component of the In4d
peak to In–In bonds. However, the relative intensity of
the In–In feature relative to the In–N component and the
lack of surface sensitivity due to nearly 1500 eV photon
and electron energies suggest that this component is more
likely due to bulk indium clusters.
Excluding both the effects of the kinematic factor and
dispersion, the shift of the plasmon peak position is attrib-
uted to a variation of carrier concentration with depth. As
McConville and coworkers [10–15] demonstrated by per-
forming semiclassical dielectric theory simulations, the po-
sition of the plasmon peak is related to the surface and
bulk plasma frequencies. The shift of the plasmon feature
to lower loss energy as the incident electron energy is in-
creased means that the plasma frequency is larger for the
surface than for the bulk and indicates an accumulation
of electrons on the surface. As seen in Fig. 1, the plasmon
peak position shifts most dramatically when the incident
electron energy is changed from 15 to 25 eV but there is al-
most no shift when incident energy changes from 7 to 15 eV
or 25 to 35 eV. The shift of the plasmon feature with inci-
dent energy is determined by the depth profile of the carrier
concentration along with the sampling depth of the elec-
trons. HREEL spectra reported by Mahboob et al. [14]
and Veal et al. [15] also show that most of the shift occurs
as the incident energy is changed from 15 to 30 eV and this
feature is reproduced in their simulations using a four or
five layer slab model with an electron concentration that
is large at the surface and decreasing deeper into the layer.
The HREEL spectra reported here are therefore consistent
with a surface electron accumulation layer.
In most of the reports of experimental observations of
surface electron accumulation on InN layers neither film
polarity nor surface termination is known. Piper et al.
[10,13] have reported that the MBE-grown samples on
which they observed surface charge accumulation were
75% In-polar and 25% N-polar as determined by CAICISS.
Although HREEL spectra were used in a number of cases
to study the surface electron accumulation, no surface
adsorbate vibrational loss peaks were observed in those
studies so surface termination could not be determined.
In the present work, HREEL spectra at low incident elec-
A broad loss feature, with a peak in the range of 3100–
3400 cmÀ1 for different incident electron energies, also ap-
pears in each spectrum. Following the assignment of
McConville and coworkers [10–15], we attribute these
broad features as due to conduction band plasmon excita-
tions. The plasmon peak energy decreases when the inci-
dent electron energy is increased.

R.P. Bhatta et al. / Surface Science 601 (2007) L120–L123
L123
[10] L.F.J. Piper, T.D. Veal, M. Walker, I. Mahboob, C.F. McConville,
H. Lu, W.J. Schaff, J. Vac. Sci. Technol. A 23 (2005) 617.
[11] I. Mahboob, T.D. Veal, L.F.J. Piper, C.F. McConville, H. Lu, W.J.
Schaff, J. Furthmu¨ller, F. Bechstedt, Phys. Rev. B 69 (2004) 201307,
R.
[12] T.D. Veal, I. Mahboob, L.F.J. Piper, C.F. McConville, H. Lu, W.J.
Schaff, J. Vac. Sci. Technol. B 22 (2004) 2175.
[13] L.F.J. Piper, T.D. Veal, I. Mahboob, C.F. McConville, H. Lu, W.J.
Schaff, Phys. Rev. B 70 (2004) 115333.
[14] I. Mahboob, T.D. Veal, C.F. McConville, H. Lu, W.J. Schaff, Phys.
Rev. Lett. 92 (2004) 036804.
[15] T.D. Veal, L.F.J. Piper, I. Mahboob, H. Lu, W.J. Schaff, C.F.
McConville, Phys. Stat. Sol. (C) 7 (2005) 2246.
[16] L. Colakerol, T.D. Veal, H.K. Jeong, L. Plucinski, A. DeMasi, T.
Learmonth, P.A. Glans, S. Wang, Y. Zhang, L.F.J. Piper, P.H.
Jefferson, A. Federov, T.C. Chen, T.D. Moustakas, C.F. McConville,
K.E. Smith, Phys. Rev. Lett. 97 (2006) 237601.
[17] M. Noguchi, H. Hirakawa, T. Ikoma, Phys. Rev. Lett. 66 (1991)
2243.
tron energy yield N–H surface vibrations, which indicate
both the surface termination and film polarity. The inten-
sity of the N–H vibrations, the lack of In–H vibrations
even after room temperature atomic hydrogen exposure,
and the presence of a 1 · 1 LEED pattern are evidence that
no indium was present on the surface. The electron accu-
mulation observed here is therefore not caused by indium
droplets, indium overlayers, or isolated In–In dimers on
the InN surface. This finding is consistent with the pinning
of the surface Fermi level in the conduction band but indi-
cates that In–In bonds are not the only source of surface
states in the conduction band.
In summary, InN layers grown by HPCVD and cleaned
by atomic hydrogen have been studied by HREELS. Spec-
tra taken with low incident electron energy demonstrated
that the surface is N–H terminated and free of indium over-
layers and droplets. Spectra acquired as a function of inci-
dent energy revealed the presence of a surface electron
accumulation. These results strongly suggest that the elec-
tron accumulation on InN surfaces is not due to excess
indium.
[18] L.O. Olsson, C.B.M. Andersson, M.C. Hakansson, J. Kanski, L.
Ilver, U.O. Karlsson, Phys. Rev. Lett. 76 (1996) 3626.
[19] S.P. Fu, T.J. Lin, W.S. Su, C.Y. Shieh, Y.F. Chen, C.A. Chang, N.C.
Chen, P.H. Chang, J. Appl. Phys. 99 (2006) 126102.
[20] J.C. Ho, P. Specht, Q. Yang, X. Xu, D. Hao, E.R. Weber, J. Appl.
Phys. 98 (2005) 093712.
[21] M. Drago, P. Vogt, W. Richter, Phys. Stat. Sol. (a) 203 (2006) 116.
[22] L.F.J. Piper, T.D. Veal, P.H. Jefferson, C.F. McConville, F. Fuchs, J.
Furthmuller, F. Bechstedt, H. Lu, W.J. Schaff, Phys. Rev. B 72 (2005)
245319.
Acknowledgement
The authors would like to acknowledge support of this
work by GSU-RPE.
[23] M. Draxler, M. Walker, C.F. McConville, Nucl. Instr. Meth. B 249
(2006) 886.
[24] D. Segev, C.G. Van de Walle, Europhys. Lett. 76 (2006) 305.
[25] R.P. Bhatta, B.D. Thoms, M. Alevli, V. Woods, N. Dietz, Appl.
Phys. Lett. 88 (2006) 122112.
References
[26] V.J. Bellito, B.D. Thoms, D.D. Koleske, A.E. Wickenden, R.L.
Henry, Surf. Sci. 430 (1999) 80.
[27] H. Ibach, D.L. Mills, Electron Energy Loss Spectroscopy and Surface
Vibrations, Academic, New York, 1982.
[28] N. Nienhaus, S.P. Grabowski, W. Monch, Surf. Sci. 368 (1996) 196.
[29] U.D. Pennino, C. Mariani, A. Ammoddeo, F. Proix, C. Sebenne, J.
Elect. Spectr. Rel. Phen. 64 (1993) 491.
[30] X. Hou, S. Yang, G. Dong, X. Ding, X. Wang, Phys. Rev. B 35
(1987) 8015.
[31] J. Fritsch, A. Eckert, P. Pavone, U. Schroder, J. Phys. Condens. Mat.
7 (1995) 7717.
[32] C.K. Gan, D.J. Srolovitz, Phys. Rev. B 74 (2006) 115319.
[33] H. Froitzheim, H. Ibach, D.L. Mills, Phys. Rev. B 11 (1975) 4980.
[34] P.J. Chen, J.E. Rowe, J.T. Yates Jr., Phys. Rev. B 50 (1994) 18134.
[1] A.G. Bhuiyan, A. Hashimoto, A. Yamamoto, J. Appl. Phys. 94
(2003) 2779.
[2] S.K. O’Leary, B.E. Foutz, M.S. Shur, L.F. Eastman, Appl. Phys.
Lett. 88 (2006) 152113.
[3] K.S.A. Butcher, T.L. Tansley, Superlattice. Microstruct. 38 (2005) 1.
[4] N. Dietz, M. Alevli, V. Woods, M. Strassburg, H. Kang, I.T.
Ferguson, Phys. Stat. Sol. (b) 242 (2005) 2985.
[5] V. Woods, N. Dietz, Mat. Sci. Eng. B 127 (2006) 239.
[6] M. Alevli, G. Durkaya, A. Weerasekara, A.G.U. Perera, N. Dietz, W.
Fenwick, V. Woods, I. Ferguson, Appl. Phys. Lett. 89 (2006) 112119.
[7] H. Lu, W.J.Schaff.F. Eastman, C.E. Stutz, Appl. Phys. Lett. 82 (2003)
1736.
[8] L.F.J. Piper, T.D. Veal, C.F. McConville, H. Lu, W.J. Schaff, Appl.
Phys. Lett. 88 (2006) 252109.
[9] C.S. Gallinat, G. Koblmuller, J.S. Brown, S. Bernardis, J.S. Speck,
G.D. Chern, E.D. Readinger, H. Shen, W. Wraback, Appl. Phys.
Lett. 89 (2006) 032109.