H. Nykänen, P. Mattila, S. Suihkonen, J. Riikonen, E. Quillet, E. Homeyer, J. Bellessa and M. Sopanen. Low energy electron beam
induced damage on InGaN/GaN quantum well structure. Journal of Applied Physics, 109, 083105, 2011. © 2011 American Institute of
Physics. Reprinted with permission from American Institute of Physics.
JOURNAL OF APPLIED PHYSICS 109, 083105 (2011)
Low energy electron beam induced damage on InGaN/GaN quantum
well structure
H. Nykänen,1,a) P. Mattila,1 S. Suihkonen,1 J. Riikonen,1 E. Quillet,1 E. Homeyer,2
J. Bellessa,2 and M. Sopanen1
1
Department of Micro- and Nanosciences, TKK–Aalto University School of Science and Technology,
P.O. Box 13500, FI-00076 AALTO, Finland
2
Laboratoire de Physique de la Matiére Condensée et Nanostructures, CNRS UMR 5586, University Lyon-1,
Villeurbanne 69622, France
(Received 3 January 2011; accepted 9 March 2011; published online 19 April 2011)
In this paper, low energy electron beam (5–20 keV, 0–500 lAs/cm2) induced damage on a GaN/
InGaN/GaN near-surface quantum well structure is studied. Exposure to low energy electron beam
is shown to significantly reduce the optical quality of the structure. It is also observed that reducing
the electron beam energy causes larger PL intensity reduction. This can be explained by
considering the beam penetration depth, which is shown to be smaller with lower e-beam energies.
The damage is believed to be attributed to enhanced dislocation mobility upon low energy electron
beam irradiation. However, further studies are needed to confirm the mechanism. These results
should be taken into consideration in low energy electron beam related sample characterization and
C 2011 American Institute of Physics. [doi:10.1063/1.3574655]
preparation. V
I. INTRODUCTION
Indium gallium nitride quantum wells (QWs) are currently widely utilized in optoelectronic devices. Due to their
high luminescence efficiency at near UV, blue, and green
wavelengths, InGaN QWs are used in light emitting diodes
and laser diodes.1,2 Further, the bandgap of InGaN can be
tuned from 0.7 to 3.4 eV by varying the indium content, thus
providing a full-solar-spectrum material system for multijunction solar cells.3 Combined with its remarkable resistance
against high-energy proton irradiation, InGaN material is also
very suitable for aerospace and nuclear energy applications.3
However, InGaN QWs are very sensitive to constant exposure to mega-electron-volt heavy ions as encountered in radiation environments and ion beam material analysis leading to
detrimental effects on the optical properties of InGaN QWbased devices.4 Studies on electron beam (e-beam) induced
damage in InGaN QWs have also been reported.5,6 They
relate to high e-beam energies ranging from 200 to 400 keV.
These studies indicate that long-time exposure to high energy
electrons can alter the local In concentration. However,
smaller energy electrons are utilized in various processes
such as electron beam lithography and scanning electron microscopy (SEM), which are both very important and widely
used methods in micro- and nanodevice fabrication and characterization. Low energy e-beam exposure has been demonstrated to reduce PL intensity in GaAs/AlGaAs (Ref. 7) and
AlN/AlGaN/GaN (Ref. 8) QW structures. There is still a very
limited amount of knowledge available on the effects of low
energy e-beam irradiation on InGaN QWs.
In this paper we present our studies on e-beam induced
damage on near-surface InGaN QWs. It is shown that the photoluminescence (PL) of InGaN QWs is significantly reduced
for samples exposed to an e-beam with low electron energies
a)
Electronic mail: henri.nykanen@tkk.fi.
0021-8979/2011/109(8)/083105/3/$30.00
(5–20 keV). Larger energy absorption in the surface region of
the sample is believed to be the main reason for the optical
degradation of the structure. To our knowledge, low energy ebeam damage to InGaN QWs has not been reported before.
II. EXPERIMENT
The QW structures used in this study were grown on cplane sapphire (Al2O3) substrates by metal organic vapor
phase epitaxy. Ammonia (NH3), trimethylindium, and trimethylgallium were used as the N, In, and Ga sources, respectively. First a 3-lm-thick undoped c-plane GaN buffer layer
was grown on sapphire with the well-known two-step growth
method.9 A 3-nm-thick InxGa1xN QW (emission wavelength of 470 nm) was grown on the buffer layer at 750 C
with a composition of x ¼ 0.12. The QW was covered by a
GaN capping layer (CL) having a thickness d of 10–40 nm.
The thickness of the InGaN QW and of the GaN CL were
measured by x-ray reflectivity (details elsewhere).10
Areas of 700 lm 700 lm of the sample surface were
exposed to an e-beam using a Zeiss Supra 40 SEM varying
the exposure dose from 0 to 500 lAs/cm2 and e-beam energy
from 5 to 20 keV. The lower limit of 5 keV results from the
SEM system being unable to measure the beam current accurately under 5 kV acceleration voltages. In this context, the
exposure means that a focused (2.1 nm cross section), dense
e-beam is constantly scanning across the target area in rapid
succession. Consequently, the total exposure dose can be
considered to be identical to the situation where a static
beam would be spread across the whole area. PL measurements were performed before and after e-beam exposures for
evaluating the damage caused to the samples. A monochromatic He–Cd laser emitting at 325 nm with power of approximately 80 W/cm2 was used as an excitation source.
Moreover, to exclude possible measurement errors from
laser power fluctuations, the measured PL intensities were
109, 083105-1
C 2011 American Institute of Physics
V
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
083105-2
Nykänen et al.
FIG. 1. Relative PL spectra of the sample area before and after 20 keV
e-beam exposure. The exposure dose was 204 lAs/cm2 and the GaN CL
thickness was 20 nm.
compared to an unexposed reference sample measured
simultaneously with the test areas.
III. RESULTS AND DISCUSSION
Figure 1 shows the PL spectra of the InGaN QW sample
test area before and after e-beam exposure with 20 kV acceleration voltage and with an exposure dose of 204 lAs/cm2.
The PL intensity is reduced by a factor of 4, whereas the
full-width at half-maximum and the peak position remain
virtually unchanged. The emission wavelength of QWs
grown on c-plane GaN is strongly dependent on the piezoelectric field caused by strain in the InGaN layer.11 The absence of an observable PL peak shift suggests that the
piezoelectric field is unaltered. Therefore, there is no evidence of strain relaxation due to the e-beam irradiation.
To better understand the phenomenon, further experiments were carried out. Figure 2 shows the PL intensity as a
function of the exposure dose (i.e., e-beam current exposure time exposure area) using a constant e-beam energy
of 20 keV. The increasing dose was realized in two ways,
J. Appl. Phys. 109, 083105 (2011)
FIG. 3. Relative PL intensity as a function of e-beam energy for three different GaN capping layer thicknesses d (10, 20, and 40 nm). Exposure dose
was kept constant (37 lAs/cm2).
namely increasing exposure time and increasing e-beam current, to study if their contributions were equivalent. As can
be expected, the PL intensity decreased exponentially with
increasing exposure dose. Similar behavior has been previously reported for GaAs/AlGaAs structures.7 For exposure
doses above 100 lAs/cm2, the remaining PL intensity is less
than 50% of the original PL, which attests for considerable
damage within the QW structure.
As electron penetration into a material is a function of
the e-beam kinetic energy, experiments on the behavior of
the PL intensity as a function of e-beam energy (5, 10, 15,
and 20 keV) were made. To gain more insight, the experiment was repeated on three different samples having different
CL thicknesses (10, 20, and 40 nm) as shown in Fig. 3. The
exposure dose was constant at 37 lAs/cm2. It can be seen
that the PL intensity decreases more with a thinner CL. Further, a peculiar behavior of the relative PL intensity is
observed as the curves have a distinctive minimum at 5 keV.
Similar behavior has been reported for GaAs/AlGaAs QW
structure.7
The energy dissipation as a function of depth (i.e.,
depth-dose function) can be approximated for low e-beam
energies by the Bethe–Bloch formula.12 The depth-dose
function for GaN is shown in Fig. 4. The curves are obtained
using cubic polynomial approximation for the energy loss
parameter,13 which is multiplied by the ratio of beam energy
(E) and Grün range (RG ):
dE fE
ða0 þ a1 y þ a2 y2 þ a3 y3 Þ;
¼
dx RG
FIG. 2. Relative PL intensity as a function of e-beam exposure dose. The
e-beam acceleration voltage (20 kV) and GaN CL thickness (20 nm) were
fixed.
(1)
where RG ¼ kE1:75 q1 (nm),13 k ¼ 45:7,14 q is the material
density (6.15 g/cm3 for GaN), and f ¼ 16:5 is the normalization coefficient. The coefficients a0 ¼ 0:62, a1 ¼ 6:10,
a2 ¼ 12:26, and a3 ¼ 5:65 were obtained from Ref. 13.
Variable y is the penetration depth x divided by the Grün
range.13
The minimum in the PL intensity at smaller kinetic
energy can be explained by the beam energy dissipation.
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
083105-3
Nykänen et al.
J. Appl. Phys. 109, 083105 (2011)
Consequently, thermal effects can be ruled out. All in all, the
damage mechanism in the kilo-electron-volt range e-beam
irradiation is believed to be different from the elastic scattering mechanism of the mega-electron-volt range irradiation.7
The study shows that (in the case of GaAs) annealing after 5–
25 keV irradiation does not recover the PL intensity, which it
does after mega-electron-volt range e-beam irradiation.
IV. CONCLUSIONS
FIG. 4. Relative energy dissipation calculated as a function of penetration
depth in GaN with three different e-beam energies (5, 10, and 15 keV).
With lower beam energies, a larger amount of energy is
absorbed closer to the surface. Thus, the PL intensity minimum should indeed be located at smaller kinetic energies for
near-surface QWs. Our calculations suggest that the agreement between the experimental data and Eq. (1) can be
slightly enhanced using smaller values of k, which decreases
the Grün range. Smaller values for k have been suggested
before.14 However, it should be noted that the formula for the
Grün range is only valid for e-beam energies between 5 and
25 keV.14 Therefore, this model cannot accurately predict the
location of the PL intensity minimum in Fig. 3, which could
be situated at less than 5 keV at least for the thinner CLs. To
gain more insight into the PL intensity reduction as a function of the CL thickness, further studies are needed.
Cathodoluminescence intensity reduction under 10 keV
e-beam irradiation has been observed in GaN layers.15 The
study suggests that the contributing mechanism is the introduction of nonradiative defects by dislocation diffusion. Further, similar intensity reduction has been observed in the
cathodoluminescence study of GaN/AlGaN QWs irradiated
with a 10 keV e-beam.8 The used e-beam current (0.1–2 nA)
and exposure time (30 min) correspond to our work. It was
assumed that e-beam exposure can cause vacancy diffusion
(through nonradiative recombination) forming dislocation
loops,16 which are known to degrade radiative recombination. Both of these studies give a plausible mechanism for
PL intensity reduction under the e-beam irradiation.
Hydrocarbon contamination in SEM has been
observed.17 It is possible that this contamination alters the
optical properties of the surface and causes the reduction of
the PL intensity. However, after annealing (at 600 C,
observed to desorb hydrocarbons),18 cleaning with aqua
regia (HNO3:HCl 1:3, efficiently removes metal contamination) and piranha (H2SO4:H2O2 3:1, efficiently removes organic contamination), the PL was not recovered. This
suggests that the possible surface contamination is not the
reason for the PL reduction.
The maximum change of material temperature in local
heating of GaN during low energy (5–15 keV) e-beam exposure is shown to be in the order of tens of degrees.19
To conclude, low energy (5–20 keV) e-beam induced
damage to InGaN QWs was studied. According to the
results, exposure to low energy e-beam severely degrades
the optical quality of InGaN QWs. Further, an interesting
discovery was that e-beam with lower kinetic energy can
cause more damage to the near-surface QW than a beam
with higher energy. Based on the depth-dose calculations,
this phenomenon can be explained by larger energy absorption into the surface region with smaller e-beam energies.
Because of the significance of the PL intensity reduction,
these results should be taken into consideration when fabricating or characterizing near-surface GaN/InGaN structures
using e-beam techniques.
ACKNOWLEDGMENTS
This work was supported by the Multidisciplinary Institute of Digitalization and Energy (MIDE). The authors
acknowledge the Academy of Finland (Project Nos. 127689,
138115, and 135157), for financial support.
S. Pearton, J. Zolper, R. Shul, and F. Ren, J. Appl. Phys. 86, 78 (1999).
S. Nakamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T. Matsushita, H. Kijoku, and Y. Sugimoto, Jpn. J. Appl. Phys. 35, L74 (1996).
3
J. Wu, W. Walukiewich, K. M. Yu, J. W. A. II, E. E. Haller, H. Lu, and
W. J. Schaff, J. Appl. Phys. 94, 6477 (2003).
4
M. Ali, O. Svensk, Z. Zhen, S. Suihkonen, P. Törmä, H. Lipsanen, M.
Sopanen, K. Hjort, and J. Jensen, Physica B 404, 4925 (2009).
5
T. Li, E. Hahn, D. Gerthsen, A. Rosenauer, A. Strittmatter, L. Reissmann,
and D. Bimberg, Appl. Phys. Lett. 86, 241911 (2005).
6
T. M. Smeeton, C. J. Humphreys, J. S. Barnard, and M. J. Kappers,
J. Mater. Sci. 41, 2729 (2006).
7
N. Tanaka, H. Kawanishi, and T. Ishikawa, Jpn. J. Appl. Phys. 32, 540
(1992).
8
B. Sieber, J. Phys.: Condens. Matter 18, 1033 (2006).
9
H. Amano, N. Sawaki, I. Akasaki, and Y. Toyoda, Appl. Phys. Lett. 48, 353
(1986).
10
S. Sintonen, S. Suihkonen, S. Svensk, P. Törmä, M. Ali, M. Sopanen, and
H. Lipsanen, Physica Status Solidi 7, 1790 (2010).
11
T. Takeuchi, S. Sota, M. Katsuragawa, M. Komori, H. Takeuchi,
H. Amano, and I. Akasaki, Jpn. J. Appl. Phys. 36, 382 (1997).
12
H. Bethe and J. Ashkin, Experimental Nuclear Physics, edited by E. Serge
(Wiley, New York, 1953).
13
T. Everhart and P. H. Hoff, J. Appl. Phys 42, 5837 (1971).
14
O. Kurniawan and V. K. S. Ong, Scanning 29, 280 (2007).
15
A. Amokrane, S. Dassonneville, B. Sieber, J.-L. Farvacque, B. Beaumont,
V. Bousquet, P. Gibart, J.-D. Ganiere, and K. Leifer, J. Phys.: Condens.
Matter 12, 10271 (2000).
16
J. M. Bonard, J.-D. Ganiere, L. Vanzetti, J. J. Paggel, L. Sorba, A. Franciosi, D. Herve, and E. Molva, J. Appl. Phys. 84, 1263 (1998).
17
M. T. Postek, Scanning 18, 269 (1996).
18
R. F. Egerton, P. Li, and M. Malac, Micron 35, 399 (2004).
19
L. Bakaleinikov, E. Galaktionov, V. Tret’yakov, and E. Tropp, Phys. Solid
State 43, 811 (2001).
1
2
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp