Papadomanolaki APL

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Molecular beam epitaxy of thick InGaN(0001) films: Effects of substrate

temperature on structural and electronic properties

E. Papadomanolaki,1,a) C. Bazioti,2 S. Kazazis,1 G. P. Dimitrakopulos,2 and E.


Iliopoulos,1,3

1
Physics Department, University of Crete, P.O. Box 2208, 7100, Heraklion, Greece
2
Department of Physics, Aristotle University of Thessaloniki, GR 541 26 Thessaloniki, Greece
3
Institute of Electronic Structure & Laser (IESL), Foundation for Research & Technology-Hellas (FORTH), P.O.

Box 1527, 71110 Heraklion, Greece

Indium gallium nitride films with compositions close to the middle of the miscibility gap and
thickness approximately up to 0.5 μm were epitaxially grown on GaN(0001) by plasma-
assisted molecular beam epitaxy at growth temperatures spanning the range of 400 to 590 °C.
Epilayers were characterized by x-ray diffraction, transmission electron microscopy and Hall
effect measurements. The effect of substrate temperature during growth, on the structural and
electronic properties of the films, was studied. Single phase films, with record high electron
mobilities were obtained at lower temperatures. Increased growth temperatures led to
epilayers with higher defect densities and phase separation. Strain relaxation through
sequestration layering and introduction of multiple basal stacking faults was observed at such
temperatures.

PACS: 61.05.Cp, 68.37.Lp, 73.61.Ey, 64.75.Qr, 81.05.Ea, 81.15.Hi, 88.40.Fh

a)
Electronic mail: elpapad@physics.uoc.gr
Ternary indium-gallium-nitride (InGaN) alloys possess a direct bandgap that can be tuned

across the largest part of the solar spectrum (0.7eV-3.4eV).1,2 Thus, they are very promising for the

development of high-efficiency solar cells.3,4 However, photovoltaic applications demand the

development of heteroepitaxial InGaN thin films with significant indium content (close to 0.5 InN

mole fraction), of adequate thickness (in the order of 500 nm) and good homogeneity. Composition

fluctuations are detrimental since they act as recombination centers for the photoexcited carriers.

These requirements are particular stringent to meet. The large mismatch (10.8%) between the binary

endpoint constituents, GaN and InN, restricts the critical thickness in the case of high indium content

alloy heteroepitaxy on GaN.5 Furthermore, the solid phase miscibility gap6 in this alloy system

results in phase separation7-9 and compositional inhomogeneities.10-12 Plasma-assisted molecular

beam epitaxy (MBE), due to its non-equilibrium growth mode, has the potential to deliver

homogeneous alloy films in the entire composition range,13,14 although most reported results15,16

concern epilayers of insufficient thickness for photovoltaic devices. Careful tuning of growth

conditions is important, since the MBE growth temperatures of GaN and InN are strongly different.17

In this work, a detailed study of the effect of substrate temperature on the structural and electronic

properties of thick (0001) InGaN epilayers, with sufficiently high InN mole fractions, is reported.

InGaN alloy films were epitaxially grown by radio-frequency plasma-assisted molecular

beam epitaxy (RF-MBE), on high resistivity Fe-doped GaN(0001)/c-Al2O3 commercial templates.

Active nitrogen species were supplied by an Oxford HD25 nitrogen RF plasma source,18 while

Knudsen cells were employed for the evaporation of group III atoms. Prior to InGaN film growth, a

GaN buffer layer of 100 nm thickness was deposited in all cases at standard RF-MBE conditions.19

Ex-situ high resolution x-ray diffraction (HR-XRD) measurements were performed using a Bede D1

triple-axis x-ray diffractometer. The composition of the films was determined by HR-XRD

measurements of the c- and α- lattice constants using (0002) and (10 1 5) Bragg reflections and

assuming that Vegard’s law and biaxial strain apply.20 Transmission electron microscopy (TEM)

observations of the heterostructures were performed in cross sectional geometry using a 200 kV

JEOL JEM 2011 microscope. Ultrathin TEM foil preparation was performed by the tripod polishing
technique in order to minimize the ion milling time and reduce beam-induced damage. Precise Ar+

ion thinning up to electron transparency was performed for a maximum duration of 10 min at 4.0-4.5

kV, using the GATAN PIPS. The electronic properties of the films were studied by resistivity and

Hall effect measurements at room temperature, using van der Pauw (vdP) method and a magnetic

field of 0.30 T.

Two series of samples are studied in this work. Within each, all other growth parameters

(gallium, indium and active nitrogen arrival rates and deposition time) were kept constant, except for

the substrate temperature ( Tgr ) during InGaN layer deposition. In the first series (S1) the total arrival

rate of group-III atoms was above that of active nitrogen (N*) species; so growth conditions were

nominally slightly metal-rich. The corresponding arrival rates for Ga, In and N* were 115, 135 and

235 GaN-equivalent nm/h. The substrate temperature range investigated was from 400 °C to 590 °C.

In the second series (S2) the group-III atoms flux was kept the same, while the N* flux was increased

to 280 GaN nm/h. Therefore, in S2 growth took place under, nominally, slightly nitrogen rich

conditions. It has however to be emphasized, that the surface stoichiometry conditions, as well as the

actual growth rates, deviate from the nominal ones as substrate temperature increases, due to InGaN

decomposition and indium desorption phenomena. The kinetic mechanisms of high indium content

InGaN RF-MBE growth and the effect of substrate temperature on them will be reported in detail

elsewhere. All samples were grown with the same deposition time, corresponding to a nominal

thickness of 0.5 μm (at low substrate temperature).

In Fig. 1(a) representative examples of HR-XRD symmetric ω-2θ scans around the GaN

(0002) reflection are shown, for InGaN layers grown at different growth temperatures, belonging to

S1 series. As seen, the InGaN (0002) diffraction peaks, for samples grown at relative low

temperatures (below 500 °C) are sharp and narrow, indicative of homogeneous alloys. No phase

separation is detected by XRD, despite the fact that the compositions of those samples (InN mole

fractions ~ 0.5) are close to the middle of the miscibility gap. As the substrate temperature increases,

the peaks broaden and become more asymmetrical. Shoulder intensity gradually increases with

growth temperature. For yet higher substrate temperatures (Tgr > 550 °C) double peaks are present in
the XRD spectra, indicating the occurrence of phase separation phenomena. The situation is similar

for the case samples belonging to the S2 series, with representative examples shown in Fig.1(b). The

dependence of the full width at half maximum (FWHM) of the (0002) XRD rocking curves (RC) on

Tgr is depicted in Fig. 2. Best structural quality, as assessed by the RC FWHM, is achieved for

InGaN films grown below 500 °C. Above that temperature, the (0002) RCs broaden considerably.

For series S1 an apparent reduction is observed above 550 °C. However, this should not be taken as

an indication of structural improvement, since both the InN mole fraction, as well as the thickness of

these films, are reduced. The dependence of the InN mole fraction on Tgr is presented in the inset of

Fig. 2. Furthermore, films grown at a substrate temperature above 550 °C always demonstrated

double peaks in XRD.

The critical layer thickness (hcr) for epitaxy of InxGa1-xN layers on GaN, with adequate InN

mole fractions x, is very small. Theoretical estimates are ~ 1 nm, for x=0.4 in the case of critical

thickness for dislocation generation5 and ~ 2 nm, for x in the range of 0.4-0.6 for the case of critical

thickness for three-dimensional growth.21 Therefore plastic relaxation is expected in thick InGaN

epilayers. The relaxation in the samples was studied by reciprocal space maps (RSM) around the

(10 1 5) GaN reflection. The percentage of plastic relaxation R is defined as:

m s
R  100% (1)
l   s

where  s and  m the in-plane lattice constants of the substrate and the epilayer, and  l is the

corresponding value of a fully relaxed ternary compound of the same composition. In Fig. 3(a) the

values of R of all samples are presented as a function of growth temperature. As seen, in the cases of

low growth temperature and increased incident N* flux, a full relaxation is observed, while as Tgr is

increased, epilayers are only partially relaxed, despite the fact that their thickness is far greater than

the corresponding hcr. This is in agreement with previous observations in low In content epilayers.22

The significant difference in residual strain between the two cases should reflect differences in the

actual strain relaxation mechanisms taking place, dictated by the different growth kinetics. A very

low temperature leads to a three-dimensional (3D) growth mode, where island edges may play a
significant role. At intermediate temperatures, a two-dimensional (2D) mode is stabilized by the

presence of metal-adlayer terminated surfaces.23 Under such conditions, plastic relaxation can be

delayed due to the absence of suitable slip plane for gliding threading dislocations(TDs).24 A further

increase of Tgr resulted in even lower values of R, accompanied by a significant broadening of RSM

peaks, while this decrease was associated with a reduction of InN mole fraction and thickness of the

films.

To shed light on the different relaxation mechanisms, TEM observations were performed in

correlation to the HR-XRD results. Figure 4 illustrates weak beam dark field (WBDF) images

obtained near the [ 1120 ] zone axis from representative samples, given in order of decreasing

temperature. Images were recorded using the g0002 and g 1 1 00 reflections in order to observe TDs

with c- and a-type Burgers vector components respectively. Moreover, using g 1 1 00 , basal stacking

faults (BSFs) become slightly inclined and well visible.

The respective film thickness measurements showed an increase of thickness with decreasing

temperature as expected due to the enhanced indium desorption and InGaN decomposition at higher

Tgr. The nominal thickness was almost reached for the sample grown at Tgr = 475 °C, shown in Figs.

4(a) and 4(b), with a thickness of 440 nm and indium content x = 0.418. A general observation for all

samples is that the majority of TDs was of a-type, as attested by their extinction when using the

g0002 reflection, consistent with the gb invisibility criterion. Such TDs are associated to the strain

relaxation since they are connected to misfit dislocation segments lying on the basal plane. The film

grown at 475 °C exhibited a rather typical relaxation behavior, albeit with a very high a-type TD

density (in the high 1010 cm-2 range) that may be attributed to twist type mosaicity introduced at the

initial stages of growth.

At higher growth temperatures, the indium content as well as the epilayer thickness were

decreased, due to the increased indium desorption. Figures 4(c) and 4(d) illustrate an example of

such a sample grown at 565 °C with thickness 320 nm and indium content x = 0.267. It was found

that, with increasing growth temperature, a pronounced change of the relaxation mode gradually took

place compared to the previous sample. In particular, it was observed that the lower part of the
InGaN epilayer, with a thickness of ~30-40 nm, exhibited better structural quality with lower defect

content compared to the rest of the film. Such stratification is similar to what has been reported

previously by Liliental-Weber et al,25 for the case of InGaN films grown by metalorganic vapor

phase epitaxy (MOVPE). This sequestered InGaN (s-InGaN) layer is better visible with the g 1 1 00

reflection, as shown in Fig. 4(d). Another characteristic feature is the observation of multiple BSFs in

the main part of the film. Such BSFs could be associated to local variations of the indium

concentration. Upon further increase of the growth temperature to 580 oC, further reduction of the

InN mole fraction (to x = 0.182) and film thickness (to 250 nm) were accompanied by an increase of

the thickness of the s-InGaN layer, which was now more pronounced reaching up to 50 nm.

However, the overall defect content was reduced due to the reduction of the indium content, as

shown in Figs. 4(e) and 4(f). The observed trend comprising an increase of the defect content from

475 oC to 565 oC, and then a reduction at 580 oC, is consistent with the (0002) RC FWHM

measurements given in Fig. 2. The appearance of shoulders (or splitting) of the HR-XRD ω-2θ scans

for high Tgr samples could be associated to the sequestration phenomenon,26 or possibly to indium

fluctuations at areas with high BSF density. Further studies are needed to clarify this issue.

The room temperature carrier Hall mobility and density, as a function of Tgr, are depicted in

Figs. 5(a) and 5(b) respectively. For low and intermediate growth temperatures, carrier mobilities of

around or above 100 cm2/Vs are measured for films with InN mole fractions in the range of 0.4 to

0.6. These values are much higher than up to now reported ones,27,28 for InGaN films of similar

composition and doping levels. A strong dependence of the electronic properties of the thick InGaN

epilayers on Tgr is observed. As seen, for both the S1 and S2 series, a similar strong monotonic

decrease of carrier mobility with Tgr is manifested. In the case of carrier concentration only a weak

dependence on Tgr is detected, with non-intentionally doping levels decreasing from ~2×1018 cm-3 to

the range of 5×1017- 1×1018 cm-3. Since there is also a dependence of the InN mole fraction on Tgr,

one would be tempted to attribute the functional dependence of mobility on Tgr to the difference of

the alloy binary endpoints carrier mobility. However, a detailed investigation excludes such

assignment. In Fig. 5(c), the experimental data, presented as a function of alloy composition, are
compared to the theoretical calculations29 of optical phonon, Coulomb and alloy scattering

mechanisms, for the same compositional range. As seen in Fig. 5(c), the major intrinsic mobility

limiting mechanism, for the InN mole fractions investigated, is alloy scattering. In the range of

compositions investigated, this mechanism has a rather weak dependence on In content, of different

functional character than the one observed in our data. It could only be partially responsible for a

small reduction of mobility in the cases of films grown in relatively low temperatures. The second in

importance intrinsic scattering mechanisms, namely Coulomb scattering, has an even weaker

dependence on composition and, given the decreasing carrier concentration on Tgr, should result in

the opposite effect. Therefore the reduction, of carrier mobility with increasing substrate temperature,

is correlated to the increasing density of structural defects and compositional inhomogeneities, in

epilayers grown at elevated temperatures. This is best described in Fig. 6, where the correlation of

measured mobility and carrier density values with the (0002) and (10 1 5) XRD RC FWHM, as well

as with epilayer growth temperature, is depicted.

In conclusion, the effect of substrate temperature during growth, by RF-MBE, on the

structural and electronic properties of thick InGaN epilayers, with high InN mole fraction, close to

the middle of the miscibility gap, was investigated. Single phase films, with record high electron

mobilities were obtained in lower temperatures, despite a high density of a-type TDs. Increased

growth temperatures led to sequestrated InGaN epilayers with inferior electrical properties and

compositional inhomogeneities. Despite the lowering of the indium content with increasing

temperature, the defect content was found to increase initially and to comprise multiple BSFs in

addition to the TDs. A growth temperature window of 450 to 500 °C is identified for the epitaxy of

high structural quality thick InxGa1-xN films with x~0.5.

Work co-financed by the European Union (European Social Fund – ESF) and Greek national funds

through the Operational Program "Education and Lifelong Learning" of the National Strategic

Reference Framework (NSRF) - Research Funding Program: THALES, project “NitPhoto”.


REFERENCES

1
J. Wu, W. Walukiewicz, K.M. Yu, J.W. Ager III, E.E. Haller, H. Lu, W.J. Schaff, Y. Saito, Y.
Nanishi, Appl. Phys. Lett. 80, 3967 (2002)
2
V.Y. Davydov, A.A. Klochikhin, R.P. Seisyan, V.V. Emtsev, S.V. Ivanov, F. Bechstedt, J.
Furthmüller, H. Harima, A.V. Mudryi, J. Aderhold, O. Semchinova, J. Graul, Phys. Stat. Solidi B
229, R1 (2002)
3
O. Jani, I. Ferguson, C. Honsberg, S. Kurtz, Appl. Phys. Lett. 91, 132117 (2007)
4
C.J. Neufeld, N.G Toledo, S.C.Cruz, M. Iza, S.P. DenBaars, U.K Mishra, Appl. Phys. Lett. 93,
143502 (2008)
5
D. Holec, P.M.F.J. Costa, M.J. Kappers, C.J. Humphreys, J. Cryst. Growth 303, 314 (2007)
6
I. Ho and G.B. Stringfellow, Appl. Phys. Lett. 69, 2701 (1996)
7
D. Doppalapudi, S.N. Basu, K.F. Ludwig, T.D. Moustakas, J. Appl. Phys. 84, 1389 (1998)
8
N.A. El-Masry, E.L. Piner, S.X. Liu, S.M. Bedair, Appl. Phys. Lett. 72, 40 (1998)
9
M. Rao, D. Kim, S. Mahajan, Appl. Phys. Lett. 85, 1961 (2004)
10
S. Chichibu, T. Sota, K. Wada, S. Nakamura, J. Vac. Sci. Tech. B 16, 2204 (1998)
11
S.M.S. Perreira, K.P. O’Donnell, E.J.D. Alves, Adv. Func. Mater. 17, 37 (2007)
12
J. Palisaitis, A. Lundskog, U. Forsberg, E. Janzén, J. Birch, L. Hultman, P.O.Å. Persson, J. Appl.
Phys. 115, 034302 (2014)
13
E. Iliopoulos, A. Georgakilas, E. Dimakis, A. Adikimenakis, K, Tsagaraki, M. Androulidaki, N.T.
Pelekanos, Phys. Stat. Solidi A 203, 102 (2006)
14
M. Kurouchi, T. Araki, H. Naoi, T. Yamaguchi, A. Suzuki, Y. Nanishi, Phys. Stat. Solidi B 241,
2843 (2004)
15
M. Moseley, B. Gunning, J. Greenlee, J. Lowder, G. Namkoong, W.A. Doolittle, J. Appl. Phys.
112, 014909 (2012)
16
Ž. Gačević, V.J. Gómez, N.García Lepetit, P.E.D. Soto Rodríguez, A. Bengoechea, S. Fernández-
Garrido, R. Nötzel, E. Calleja J. Cryst. Growth 364, 123 (2013)
17
E. Dimakis, E. Iliopoulos, K. Tsagaraki, A. Georgakilas, Appl. Phys. Lett. 86, 133104 (2005)
18
E. Iliopoulos, A. Adikimenakis, E. Dimakis, K. Tsagaraki, G. Konstantinidis, A. Georgakilas, J.
Cryst. Growth 278, 426 (2005)
19
E. Iliopoulos, M. Zervos, A. Adikimenakis, K. Tsagaraki, A. Georgakilas, Superlattices
Microstruc. 40, 313 (2006)
20
F.M. Morales, D. González, J.G. Lozano, R. García, S. Hauguth-Frank, V. Lebedev, V. Cimalla,
O. Ambacher, Acta Mater. 57, 5681 (2009)
21
W. Zhao, L. Wang, J. Wang, Z. Hao, Y. Luo, J. Cryst. Growth 327, 202 (2011)
22
A. G. Kontos, Y. S. Raptis, N. T. Pelekanos, A. Georgakilas, E. Bellet-Amalric, D. Jalabert, Phys.
Rev. B 72, 155336 (2005)
23
H. Chen, R. M. Feenstra, J. E. Northrup, T. Zywietz, J. Neugebauer, D. W. Greve, J. Vac. Sci.
Tech. B 18, 2284 (2000)
24
S. Srinivasan, L. Geng, R. Liu, F. A. Ponce, Y. Narukawa, S. Tanaka, Appl. Phys. Lett. 93, 5187
(2003)
25
Z. Liliental-Weber, K. M. Yu, M. Hawkridge, S. Bedair, A. E. Berman, A. Emara, J. Domagala, J.
Bak-Misiuk, Phys. Stat. Solidi C 6, S433 (2009)
26
Z. Liliental-Weber, K. M. Yu, M. Hawkridge, S. Bedair, A.E. Berman, A. Emara, D. R. Khanal, J.
Wu, J. Domagala, J. Bak-Misiuk, Phys. Stat. Solidi C 6, 2626 (2009)
27
B. N. Pantha, J. Li, J. Y. Lin, H. X. Jiang, Appl. Phys. Lett. 93, 182107 (2008)
28
M. Gunes, N. Balkan, D. Zanato, W.J. Schaff, Microelec. J. 40, 872 (2009)
29
L. Hsu, R.E. Jones, S.X. Li, K.M. Yu, W. Walkukiewicz, J. Appl. Phys. 102, 073705 (2007)
List of Figure Captions

FIG. 1 (color online): Representative examples of HR-XRD ω-2θ scans around the (0002) Bragg
point for InGaN films grown at different substrate temperatures. (a) and (b) corresponds to series S1
and S2 cases.

FIG. 2 (color online): InGaN films’ (0002) rocking curves FWHM versus Tgr. Red squares and blue
dots correspond to series S1 and S2 respectively. In the inset the dependence of the InN mole
fraction on substrate temperature is shown.

FIG. 3 (color online): (a) Percent of plastic relaxation R of InGaN films as a function of substrate
temperature (red squares and blue dots correspond to S1 and S2 series’ samples). Representative
(10 1 5) RSMs of a fully relaxed epilayer [case A] and of an epilayer with R=60% [case B] are shown
in (b) and (c) respectively. In each RSM the solid lines labelled R=0 and R=1 depict the locus of
expected peak positions of samples fully strained to the GaN substrate and of fully relaxed ones,
correspondingly.

FIG. 4: Cross sectional WBDF TEM images near the [ 1120 ] zone axis of InGaN thin films using
g0002 [(a), (c), (e)] and g 1 1 00 [(b), (d), (f)]. (a) and (b) illustrate a sample grown at Tgr = 475 °C. a-
type TDs are observed to emanate from the InGaN/GaN interface. (c) and (d) show a sample grown
at Tgr = 565 °C. In (d), the sequestered InGaN region (s-InGaN) is indicated, as well as multiple
BSFs at the main film area. Finally, (e) and (f) show a sample grown at Tgr = 580 °C. An increase of
the s-InGaN thickness is observed [note that the whole film thickness is not visible in (e) and (f) due
to the ion thinning of the TEM foil].

FIG. 5 (color online): The carrier Hall mobility and density as a function of growth temperature is
shown in (a) and (b) correspondingly. Red squares and blue dots correspond to series S1 and S2
respectively. In (c) the Hall mobility of all samples is presented as a function of InN mole fraction
and compared to the theoretical prediction (Ref.30) of major scattering mechanisms: alloy, optical
phonon, and Coulomb scattering (calculated for n=1×1018 cm-3)

FIG. 6 (color online): Correlation of RT carrier mobility and density with the (0002) and (10 1 5)
XRD rocking curves’ FWHM, as well as, with epilayers’ growth temperature. The symbol colours
represent the Tgr, with deep blue and deep red corresponding to the lowest and to the highest Tgr,
respectively. The symbol sizes in (a) and (b) are proportional to the RC FWHM of (0002) and
(10 1 5) reflections, respectively.
Figure 1 of 6, E. Papadomanolaki et al.
Figure 2 of 6, E. Papadomanolaki et al.
Figure 3 of 6, E. Papadomanolaki et al.
Figure 4 of 6, E. Papadomanolaki et al.
Figure 5 of 6, E. Papadomanolaki et al.
Figure 6 of 6, E. Papadomanolaki et al.

You might also like