Quantitative dopant profiling of p - n junction in In Ga As ∕ Al Ga As light-emitting diode
using off-axis electron holography
Suk Chung, Shane R. Johnson, Ding Ding, Yong-Hang Zhang, David J. Smith, and Martha R. McCartney
Citation: Journal of Vacuum Science & Technology B 28, C1D11 (2010); doi: 10.1116/1.3244575
View online: http://dx.doi.org/10.1116/1.3244575
View Table of Contents: http://scitation.aip.org/content/avs/journal/jvstb/28/1?ver=pdfcov
Published by the AVS: Science & Technology of Materials, Interfaces, and Processing
Articles you may be interested in
Doping-dependent device functionality of InP/InAlGaAs long-wavelength light-emitting transistors
Appl. Phys. Lett. 99, 103502 (2011); 10.1063/1.3633345
InP/InAlGaAs light-emitting transistors and transistor lasers with a carbon-doped base layer
J. Appl. Phys. 109, 063106 (2011); 10.1063/1.3561368
Influence of residual oxygen impurity in quaternary InAlGaN multiple-quantum-well active layers on emission
efficiency of ultraviolet light-emitting diodes on GaN substrates
J. Appl. Phys. 99, 114509 (2006); 10.1063/1.2200749
Study of AlGaInP multiquantum-well/double heterostructure light-emitting diodes with In-added GaP window
layer regrown by antimony-based liquid phase epitaxy
J. Vac. Sci. Technol. A 22, 807 (2004); 10.1116/1.1705586
The influence of the p-n junction induced electric field on the optical properties of InGaN/GaN/AlGaN light
emitting diode
Appl. Phys. Lett. 74, 1376 (1999); 10.1063/1.123555

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.10 On: Fri, 06 Feb 2015 17:26:50

Quantitative dopant profiling of p-n junction in InGaAs/ AlGaAs
light-emitting diode using off-axis electron holography
Suk Chunga兲
School of Materials, Arizona State University, Tempe, Arizona 85287

Shane R. Johnson, Ding Ding, and Yong-Hang Zhang
Center for Nanophotonics, Arizona State University, Tempe, Arizona 85287 and Department of Electrical
Engineering, Arizona State University, Tempe, Arizona 85287

David J. Smith and Martha R. McCartneyb兲
Department of Physics, Arizona State University, Tempe, Arizona 85287

共Received 16 June 2009; accepted 14 September 2009; published 3 March 2010兲
The electrostatic potential profile across the p-n junction of an InGaAs light-emitting diode with
linearly graded AlGaAs triangular barriers has been measured using off-axis electron holography.
Simulations of the junction profile show small discrepancies with experimental measurements in the
region of the p-and n-doped AlGaAs barriers, which are located away from the InGaAs quantum
wells. Revised simulations reproduce the measurements reasonably when a carrier-trap density of
6 ⫻ 1016 cm−3 in the AlGaAs barriers is subtracted from the dopant concentrations. The presence of
oxygen impurities is considered as the most likely reason for the reduction in doping efficiency.
© 2010 American Vacuum Society. 关DOI: 10.1116/1.3244575兴

I. INTRODUCTION
AlGaAs/ GaAs heterostructures have received much attention in recent years due to their novel properties which
make them suitable for optoelectronic applications, including
laser diodes, light-emitting diodes 共LEDs兲, and high electron
mobility transistors.1,2 AlxGa1−xAs/ GaAs/ AlxGa1−xAs quantum wells 共QWs兲 with triangular AlGaAs barriers have been
extensively investigated, with the objective being to improve
the carrier transport properties through the use of the linearly
graded barriers.3,4 However, unintentional impurities located
in the AlGaAs barrier layers have been reported to degrade
the expected performance of devices based on
AlGaAs/ GaAs heterostructures.5,6 Studies to determine the
nature of these impurities have been conducted using a range
of techniques including deep-level transient spectroscopy,5,7
secondary-ion mass spectroscopy,8 and photoluminescence
measurements.5 According to these studies, it has been found
that the photoluminescence intensity is inversely proportional to the concentration of deep-level electron traps.5 The
threshold current of optical devices is also closely related to
the concentration of oxygen impurities.6
Dopant distribution and activation are critical to device
performance, since these parameters determine the electrostatic potential distributions within real devices. However, as
the dimensions of semiconductor devices decrease, direct experimental determination of the potential profile represents a
demanding challenge. Off-axis electron holography with the
electron microscope is an interferometric technique which
allows convenient access to both amplitude and phase information. Since the phase shift of the electron wavefunction
depends on the local electric field distributions within the
a兲

Electronic mail: suk.chung@asu.edu
Electronic mail: molly.mccartney@asu.edu

b兲

C1D11 J. Vac. Sci. Technol. B 28„1…, Jan/Feb 2010

sample, the electron holography technique can be used to
provide quantitative information about electrostatic potential
variations in materials, with nanometer-scale resolution.9
Off-axis electron holography has been widely used to measure potential profiles associated with dopant distributions in,
for example, Si p-n junctions,10,11 and Si transistors,12,13 including one device having a 30 nm gate length.14 The potential distributions present within compound semiconductor
devices are also of considerable interest, but only a very
limited number of experimental holography studies have so
far been reported.15,16 Moreover, very little attention has
been given to studying compound semiconductors which
have varying alloy compositions that cause variations in the
mean inner potential 共MIP兲 of the material, which in turn
complicate hologram interpretation.
In this article, off-axis electron holography has been used
to measure the electrostatic potential profile across an
In0.2Ga0.8As LED p-n junction which has linearly graded
AlxGa1−xAs barriers, where the concentrations of the Si and
Be dopants were also exponentially varied. A careful comparison between the experimental and simulated p-n junction
potential 共V p-n兲 distributions allowed the density of carrier
traps in the AlGaAs layers to be determined.
II. EXPERIMENTAL DETAILS
The samples were grown by molecular beam epitaxy
共MBE兲 in a VG V80H solid-source system, and the sample
cross section is schematically illustrated in Fig. 1. The epilayers were grown on a semi-insulating GaAs 共100兲 substrate
and consisted of 共I兲 1280-nm-thick Si-doped 共n-type兲 GaAs
contact layer, 共II兲 210-nm-thick Si-doped 共n-type兲
AlxGa1−xAs barrier layer where the Al mole fraction 共x兲 was
linearly graded from 0.05 up to 0.20 and back down to 0.05,
共III兲 170-nm-thick undoped active region containing three

1071-1023/2010/28„1…/C1D11/4/$30.00

©2010 American Vacuum Society C1D11

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.10 On: Fri, 06 Feb 2015 17:26:50

C1D12

Chung et al.: Quantitative dopant profiling of p-n junction in InGaAs/ AlGaAs light-emitting diode

C1D12

FIG. 1. Schematic of InGaAs LED sample structure showing Al and In
compositions and nominal Si and Be concentrations. Growth direction is
from left to right. Region I: Si-doped n-GaAs contact; region II: Si-doped
AlxGa1−xAs barrier; region III: undoped triple In0.2Ga0.8As/ GaAs QWs; region IV: Be-doped AlxGa1−xAs barrier; region V: Be-doped GaAs; region
VI: Be-doped p-GaAs contact. Only the part of region I is shown in
schematic.

7-nm-thick In0.2Ga0.8As QWs embedded in GaAs, 共IV兲
210-nm-thick Be-doped 共p-type兲 AlxGa1−xAs barrier layer
where the Al mole fraction was linearly graded from 0.05 up
to 0.20 and back down to 0.05, 共V兲 40-nm-thick Be-doped
共p-type兲 GaAs layer, and 共VI兲 20-nm-thick Be-doped 共p+兲
GaAs contact layer. This electron holography study focuses
on the V p-n profile across regions I–V.
Samples suitable for observation by scanning transmission electron microscopy 共STEM兲 and off-axis electron holography were prepared using a Multiprep™ wedge-polishing
apparatus using a wedge angle of 2°. The holography
samples were mechanically polished without ion milling
down to thicknesses, which overlap the optimum range
needed for holography analysis.17,18 This thinning procedure
is critical for minimizing the possibility of an electrically
inactive layer that might otherwise result from extensive ion
milling.19,20 The optimum thickness range results from a
trade-off between the need for a strong phase signal against
the increasing background noise caused by increases in inelastic scattering with greater sample thickness.21 A thinner
sample area is necessary for STEM imaging and analysis.
Thus, Ar-ion milling was used for a few minutes at 3.5 keV,
with a current of 13 ␮A and a milling angle of 5°.
Off-axis electron holograms were recorded using a Philips
CM200 field-emission gun TEM operated at 200 keV, as reported elsewhere.22 In the Lorentz imaging mode, a biprism
voltage of ⬃110 V was used, giving rise to interference
fringes with a contrast of ⬃40% and a spacing of 3.8 nm. An
effective pixel size of 6 nm was obtained in the reconstructed phase image for a magnification of 20 k⫻ at the
charge coupled device camera, and lateral averaging of experimental profiles over ten adjacent pixels gave significant
improvement in the nominal sensitivity.9 During the holography observations, the samples were tilted by ⬃5° away
from the 关110兴 zone axis about the substrate normal in order
to minimize dynamical diffraction effects,23 and reference
holograms were used to correct for any nonlinearities in the

FIG. 2. 共a兲 Annular-dark-field STEM cross-sectional image of InGaAs LED
showing barriers and active layers. Note that the bands of darker contrast
correspond to the AlxGa1−xAs barrier layers in regions II and IV, and lines of
brighter contrast correspond to the three InGaAs QWs in region III. 共b兲
Corresponding composition profiles, from the line indicated in 共a兲, which
confirm Al and In concentrations across entire layer sequence.

imaging and recording system.24 Z-contrast STEM images
were recorded using a JEM2010F TEM equipped with an
annular-dark-field 共ADF兲 detector.
III. RESULTS AND DISCUSSION
A typical cross section of the InGaAs LED sample, recorded in the ADF STEM imaging mode, is shown in Fig.
2共a兲. In this Z-contrast imaging mode, the image intensity 共I兲
is proportional to the atomic number 共Z兲 of the specimen, as
given by the expression I ⬃ Z␣, where the constant ␣ has the
value of 1.7 in these experiments.25 Thus, the three InGaAs
QWs in the GaAs active layer 共region III兲 are clearly visible
as the lines of brighter contrast. Moreover, the bands of
darker contrast correspond to the graded AlGaAs barriers
共regions II and IV兲 which have Al mole fractions that are as
large as 0.2. Figure 2共b兲 shows the corresponding composition profiles for the line indicated in 共a兲, which are calculated
based on the measured Z-contrast intensities. Overall, the
profiles confirm the Al and In compositions across the entire
sequence of layers, which compare well with nominal values
when uncertainties in alloy compositions and experimental
measurements are considered. The top GaAs layer has darker
contrast than the GaAs buffer layer because of thickness curvature at the front edge of the sample due to ion milling. This
thickness variation was taken into account when calculating
elemental compositions.

J. Vac. Sci. Technol. B, Vol. 28, No. 1, Jan/Feb 2010

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.10 On: Fri, 06 Feb 2015 17:26:50

C1D13

Chung et al.: Quantitative dopant profiling of p-n junction in InGaAs/ AlGaAs light-emitting diode

FIG. 3. 共a兲 Reconstructed holographic phase image of InGaAs LED. The
black arrows indicate positions of three InGaAs QWs. 共b兲 Electrostatic potential profiles V, V p-n, and V0, from the line indicated in 共a兲, where V p-n was
extracted by subtracting V0 from V.

Figure 3共a兲 shows a reconstructed holographic phase image of the InGaAs LED. The sample thickness in the analyzed area was estimated using the ratio between the amplitude of the reference hologram and the reduced amplitude
caused by inelastic scattering.26 Using this relationship, the
sample thicknesses within the analyzed area were determined
to range from 100 to 200 nm. The values used for the inelastic mean free path in the calculations are the published values of 67 nm for GaAs and 77 nm for AlAs,18 a linear interpolation of these two values for the AlxGa1−xAs layers, and a
value of 62 nm for In0.2Ga0.8As which was determined by
fitting to the experimental measurements.
The electrostatic potential profile was then extracted from
the thickness and phase information using the expression:27

␾=

E0 + E
2␲e
⫻
Vt = CEVt = CE共V0 + V p-n兲t,
␭E
2E0 + E

共1兲

where e and ␭ are the charge and wavelength of the incident
electron, t is the sample thickness, E is the electron kinetic
energy, E0 is the electron rest energy, CE is an energy-related
constant 共0.007 28 rad V−1 nm−1 for 200 keV electrons兲, V is
the electrostatic potential, and V0 is the MIP. The electrostatic potential profile labeled V is shown in Fig. 3共b兲 for the
line profile indicated in Fig. 3共a兲. The measured phase shift
results not only from the dopant distribution across the p-n
junction but also from the variation in V0. Thus, its contribution to the electrostatic potential must be subtracted in order
to determine V p-n. The V0 profile in the AlGaAs layers was
calculated using the published MIP values for GaAs and

C1D13

FIG. 4. 共a兲 InGaAs LED V p-n profile comparisons between experiment and
simulations with and without carrier traps in AlGaAs barriers. The arrows
indicate regions of discrepancy between measurement and simulation without impurities. 共b兲 Concentration profiles of donors 共region II兲 and acceptors
共region IV兲 with and without carrier traps in AlGaAs barriers.

AlAs,18 and assuming that the MIP varied linearly with Al
mole fraction. Furthermore, the MIP difference of 0.2 V between the In0.2Ga0.8As and GaAs layers was utilized in the
calculation, which was determined by fitting to the measurement. Linearly graded MIP differences were used at the
InGaAs/ GaAs interfaces rather than abrupt values because
the phase images of the interfaces were not ideally sharp due
to insufficient sampling of holography measurements. The
result is also shown in Fig. 3共b兲, together with V p-n.
The V p-n profile was simulated using a one-dimensional
Poisson solver28 and with the published doping efficiency
values of 83% 共ASi兲 for Si-doped Al0.10Ga0.90As,29 and 75%
共ABe兲 for Be-doped Al0.65Ga0.35As.30 The experimental measurements provide the electrostatic potential profile within an
arbitrary voltage offset, which was chosen so that both the
measured and simulated V p-n profiles were zero in the GaAs
contact layer 共region I兲. By comparing the experimental and
simulated profiles in Fig. 4共a兲, it becomes apparent that the
experimental V p-n profile in the n- and p-doped AlGaAs barriers near the GaAs active layer shows some discrepancy
with the simulation, as indicated by the arrows.
The absolute accuracy of these measurements has been
improved by lateral averaging of the potential profiles to
approximately 0.03 V, which is close to the maximum discrepancy between the experimental and simulated profiles.

JVST B - Microelectronics and Nanometer Structures

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.10 On: Fri, 06 Feb 2015 17:26:50

C1D14

Chung et al.: Quantitative dopant profiling of p-n junction in InGaAs/ AlGaAs light-emitting diode

However, the shape and, most importantly, the slope of V p-n
near and throughout the active region of this p-n junction
structure 共which is depleted兲 is very sensitive to the doping
profile of the graded AlGaAs barriers on either side. Thus, it
is clear that the experimental holography results predict a
much reduced doping level in the graded AlGaAs barriers
near the barrier/active region interface than the nominal expected dopant levels that were based on prior calibrations of
Si- and Be-doped GaAs.
In order to improve the agreement with experiment, the
simulations were revised to include a reduction in doping
efficiency caused by carrier traps that could result from unintentional oxygen impurities in the AlGaAs layers. This effect can be described using the approximate expression:
n = A ⫻ Nnom − Ntrap ,

共2兲

where n is the carrier concentration, A is the efficiency of
dopant activation, Nnom is the nominal doping impurity concentration 共Si or Be兲, and Ntrap is the carrier-trap density. The
electrostatic potentials for the experimental measurements
and simulations with and without carrier traps are compared
in Fig. 4共a兲. The potential simulation using Ntrap densities of
共6 ⫾ 1兲 ⫻ 1016 cm−3 agrees reasonably with experiment, especially in the p- and n-doped AlGaAs regions. The dopant
profiles with and without the presence of carrier traps are
compared in Fig. 4共b兲, showing that the effect of these traps
on doping efficiency becomes dominant as dopant concentrations decrease near the barrier/active region interface.
Finally, it can be asserted, although not proven here, that
these results are consistent with reports in the literature
where it has been found that unintentional oxygen impurities
in AlGaAs layers grown by MBE act as deep-level nonradiative centers,5,6 with typical concentrations ranging from 1
⫻ 1016 to 1 ⫻ 1018 cm−3 depending on growth conditions.6,8
It has also been reported that the typical carrier-trap densities
range from 1 ⫻ 1015 to 1 ⫻ 1017 cm−3 depending on Al mole
fraction and growth temperature.4,7
IV. CONCLUSIONS
Off-axis electron holography has been used to map the
electrostatic potential profile across the p-n junction of an
InGaAs quantum well LED with nanometer-scale resolution.
Comparisons of the measured and simulated p-n potential
profiles indicated carrier-trap concentrations of 6
⫻ 1016 cm−3 in the graded AlGaAs barriers of the LED away
from the InGaAs regions.
Overall, this study has confirmed the ability of electron
holography to provide useful quantitative information about
dopant distributions across graded heterojunctions. Moreover, electron holography measurements combined with
straightforward modeling should provide growers with a
powerful tool for monitoring the effect of deep-level impurity traps, such as oxygen, on the doping efficiency in complex heterostructures.

C1D14

ACKNOWLEDGMENTS
This work was partially supported by the Department of
Energy Grant No. DE-FG02-04ER46168. We acknowledge
use of facilities within the John M. Cowley Center for High
Resolution Electron Microscopy at Arizona State University.
H. Kroemer, Proc. IRE 45, 1535 共1957兲.
Z. I. Alferov and R. F. Kazarinov, Patent Application No. 950840 30
March, 1963.
3
C. L. Allyn, A. C. Gossard, and W. Wiegmann, Appl. Phys. Lett. 36, 373
共1980兲.
4
M. E. Daniels, P. J. Bishop, K. D. Jensen, B. K. Ridley, D. A. Ritchie, M.
Grimshaw, E. H. Linfield, G. A. C. Jones, and G. W. Smith, J. Appl. Phys.
74, 5606 共1993兲.
5
K. Akimoto, M. Kamada, K. Taira, M. Arai, and N. Watanabe, J. Appl.
Phys. 59, 2833 共1986兲.
6
Y. Mihashi, M. Miyashita, N. Kaneno, M. Tsugami, N. Fujii, S. Takamiya, and S. Mitsui, J. Cryst. Growth 141, 22 共1994兲.
7
K. Yamanaka, S. Naritsuka, K. Kanamoto, M. Mihara, and M. Ishii, J.
Appl. Phys. 61, 5062 共1987兲.
8
S. Naritsuka, O. Kobayashi, K. Mitsuda, and T. Nishinaga, J. Cryst.
Growth 254, 310 共2003兲.
9
M. R. McCartney and D. J. Smith, Annu. Rev. Mater. Sci. 37, 729
共2007兲.
10
M. R. McCartney, D. J. Smith, R. Hull, J. C. Bean, E. Voelkl, and B.
Frost, Appl. Phys. Lett. 65, 2603 共1994兲.
11
H. Lichte, P. Formanek, A. Lenk, M. Linck, C. Matzeck, M. Lehmann,
and P. Simon, Annu. Rev. Mater. Sci. 37, 539 共2007兲.
12
M. A. Gribelyuk, M. R. McCartney, J. Li, C. S. Murthy, P. Ronsheim, B.
Doris, J. S. McMurray, S. Hegde, and D. J. Smith, Phys. Rev. Lett. 89,
025502 共2002兲.
13
M. G. Han, P. Fejes, Q. Xie, S. Bagchi, B. Taylor, J. Conner, and M. R.
McCartney, IEEE Trans. Electron Devices 54, 3336 共2007兲.
14
N. Ikarashi, T. Ikezawa, K. Uejima, T. Fukai, M. Miyamura, A. Toda, and
M. Hane, J. Appl. Phys. 103, 114514 共2008兲.
15
A. T. Tilke, A. Lenk, U. Muhle, C. Wagner, C. Dahl, and H. Lichte, IEEE
Trans. Electron Devices 52, 1067 共2005兲.
16
Z. H. Wu, A. M. Fischer, F. A. Ponce, W. Lee, J. H. Ryou, J. Limb, D.
Yoo, and R. D. Dupuis, Appl. Phys. Lett. 91, 041915 共2007兲.
17
M. G. Han, J. Li, Q. Xie, P. Fejes, J. Conner, B. Taylor, and M. R.
McCartney, Microsc. Microanal. 12, 295 共2006兲.
18
S. Chung, D. J. Smith, and M. R. McCartney, Microsc. Microanal. 13,
329 共2007兲.
19
W. D. Rau, P. Schwander, F. H. Baumann, W. Höppner, and A. Ourmazd,
Phys. Rev. Lett. 82, 2614 共1999兲.
20
D. Cooper, A. C. Twitchett, P. K. Somodi, P. A. Midgley, R. E. DuninBorkowski, I. Farrer, and D. A. Ritchie, Appl. Phys. Lett. 88, 063510
共2006兲.
21
A. Lenk, H. Lichte, and U. Muehle, J. Electron Microsc. 54, 351 共2005兲.
22
M. R. McCartney, D. J. Smith, R. F. C. Farrow, and R. F. Marks, J. Appl.
Phys. 82, 2461 共1997兲.
23
M. Gajdardziska-Josifovska, M. R. McCartney, W. J. de Ruijter, D. J.
Smith, J. K. Weiss, and J. M. Zuo, Ultramicroscopy 50, 285 共1993兲.
24
W. J. de Ruijter and J. K. Weiss, Ultramicroscopy 50, 269 共1993兲.
25
S. Hillyard and J. Silcox, Ultramicroscopy 58, 6 共1995兲.
26
M. R. McCartney and M. Gajdardziska-Josifovska, Ultramicroscopy 53,
283 共1994兲.
27
J. Li, M. R. McCartney, and D. J. Smith, Ultramicroscopy 94, 149
共2003兲.
28
G. Snider, 1D Poisson program, available at http://www.nd.edu/~gsnider/.
29
S. Chung, S. R. Johnson, Y.-H. Zhang, D. J. Smith, and M. R. McCartney,
IEEE Trans. Electron Devices 56, 1919 共2009兲.
30
S. R. Johnson, Y. G. Sadofyev, D. Ding, Y. Cao, S. A. Chaparro, K.
Franzreb, and Y.-H. Zhang, J. Vac. Sci. Technol. B 22, 1436 共2004兲 共Be
doping efficiencies for lower Al content AlGaAs do not appear to be
available in the literature兲.
1
2

J. Vac. Sci. Technol. B, Vol. 28, No. 1, Jan/Feb 2010

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.10 On: Fri, 06 Feb 2015 17:26:50