Atomic Diffusion and Band Lineups at InGaAsonInP Heterointerfaces

1
Atomic Diffusion and Band Lineups at
InGaAs-on-InP Heterointerfaces
P.E. Smith, S.H. Goss, M. Gao, and L.J.
Brillson The Ohio State University, Columbus, OH
M.K. Hudait, Y. Lin, and S.A. Ringel The Ohio
State University, Columbus, OH
IV. Analysis
Abstract We used secondary ion mass spectrometry
(SIMS), cathodoluminescence spectroscopy (CLS),
and an analysis of the secondary electron
thresholds (SETs) to determine the effects of
growth conditions on interdiffusion and band
lineups at InGaAs-on-InP heterointerfaces.
III. Results

• Ohio-State-grown, lattice-matched, InGaAs-InP
  double heterostructures have InP-on-InGaAs
  interface with 20 sec. P soak. InGaAs-on-InP
  interface has As soak ranging from 20-150 sec.
  during MBE growth at 4850C.

The trend of a decrease in DEvac at the
InGaAs-on-InP interface is consistent with a
decrease in DEc.
Heterostructure has broadened interface with
increased As soak time.

• I. Motivation
• The InP-InGaAs system is an important material
  for optoelectronics and electrical devices.
• Diffusion and reactions during source switching
  sequences can introduce defects and/or
  interlayers that affect local bonding and
  fundamental heterojunction properties.
• Despite an abundance of research and the
  commercial importance of InP based devices there
  remains debate on the physical and electronic
  microstructure of the InGaAs-InP interface.

InP
InGaAs
InP

• SIMS depth profiles show a broadened (diffused)
  As distribution at the InGaAs-on-InP interface
  with increased As soak time.
• InGaAs-on-InP interface broadening ranges from 0
  to 8 nm for soak times of 20 to 150 sec.
  respectively based on instrumental broadening of
  20 nm measured for the most abrupt interfaces.
• InP-on-InGaAs, control interface shows no
  broadening and is abrupt to within 15 nm
  instrumental broadening.
• Soak times of 60, 90, and 120 sec. show
  monotonically increasing broadening.

Evac
4.38 eV
4.5 eV
Abrupt structure has a theoretical DEc of 0.25 eV
and a DEvac of 0.13 eV based on c 4.50 V for
InGaAs and 4.38 V for InP.
EC
DEc 0.25 eV

• II. Experimental Techniques
• SIMS analysis performed on a PHI TRIFT III
  time-of-flight mass spectrometer.
• JEOL UHV scanning electron microscope (SEM)
  provides CLS analysis and determines SETs of
  in-situ cleaved, selectively diffused, InGaAs-InP
  double heterostructures.
• Complementary techniques including XRD and
  photoconductive decay (PCD) measurements give
  insight into interface structure.

InGaAs
Ev
InP
EC

• XRD shows broadening and shifting of InGaAs peak
  with respect to InP increases in heterostructures
  with long As soak times.
• Rocking curves are consistent with SIMS result
  showing a broadened, more disordered, interface
  with increased As exposure.

DEc 0
Diffused interface shows a lower DEvac consistent
with a decreased DEc.
InGaAs
Ev
InP
-

PHI TRIFT III TOF SIMS

• CLS results are consistent with this model
• In abrupt heterostructures with short soak times
  CLS measured 10 keV InGaAs (absolute) and 25 keV
  InP/InGaAs (relative) intensities are higher
  since a barrier to carrier diffusion exists DEc
  from InGaAs to InP and n-type band-bending from
  InP to InGaAs.
• Subtracting c from the SET data yields a
  lowered DEc. The barriers to carrier diffusion
  are removed and both 10 keV induced diffusion
  from InGaAs into InP and 25 keV induced diffusion
  from InP into InGaAs are increased. This reduces
  the measured InGaAs (10 keV) and InP/InGaAs (25
  keV) intensities.

• SET measurements show the change in vacuum level
  energy (Evac) across the InGaAs-on-InP interface.
• Evac changes positively in the abrupt, short
  soak time, heterostructures, and negatively in
  the more diffuse, longer soak time, structures.
• The interface region is defined as the region
  from which 90 of backscattered electrons come
  from as calculated via Monte Carlo simulations.

• Primary ion beam sputters and increases charged
  ion yield while interlaced analysis beam ejects
  ions.
• Ions are collected and separated according to
  mass, resulting in high-resolution elemental
  depth profiles.

InP
InGaAs
JEOL JAMP-7800F SEM

• UHV SEM measures beam-induced, depth-dependent,
  optical (0.7-7.0 eV) photon transitions (CLS) and
  secondary electrons.
• Sample may be cleaved in-situ yielding a clean
  (110) surface verified by Auger Electron
  Spectroscopy (AES).

• The change in SET across the interface region
  (from InGaAs to InP) plotted versus As soak time
  shows a decrease for more diffused structures.
• Several measurements performed on each
  heterostructure give consistent results.

• An effective change in DEc can be the result of
  internal strain, interfacial defects, or a
  chemical interlayer i.e. InAsP.
• PCD measured t increases due to As-defect
  related persistent photoconductivity with a
  systematic dependence on transition properties.
• XRD measurement consistent with the formation of
  a compressively strained InAsP layer at the
  InGaAs-on-InP interface.

• Low temperature (T 10K) CL peaks can be
  integrated to provide a measure of the total
  radiative recombination in a semiconductor layer.
• 25 keV beam (top) preferentially (75) excites
  the InP buffer layer and substrate. Less
  radiative recombination occurs in the InP
  relative to the InGaAs region in samples with
  longer As soaks.
• 10 keV beam (bottom) preferentially (90)
  excites the the 500 nm InGaAs region. Less
  radiative recombination occurs in the InGaAs for
  long As soaks.

V. Conclusions

• Cross-sectional SET measurements have been used
  to measure DEc.
• An increasing anion (As) soak time decreases the
  effective InGaAs-on-InP ?EC systematically.
• SIMS depth profiles reveal interfacial
  broadening with increasing soak time consistent
  with a chemical transition layer, e.g., InAsP or
  InGaAsP.

• SET determined from difference of the
  semiconductor and analyzer work function added to
  any applied bias1.
• SET values minus bulk electron affinities (c)
  yields conduction band edge EC values and hence
  DEc band offset changes across the interface.

Acknowledgments
This work is supported by the Office of Naval
Research, the National Aeronautics and Space
Administration, the Department of Energy, and the
National Science Foundation.
Left SEM image showing SC layer structure and
spot mode analysis. Right SET and
extrapolation determining DE.

• Room temperature PCD measured lifetimes (t) are
  well in excess of the 7 ms theoretical value due
  to an As-defect related persistent
  photoconductivity2.
• Minority carrier t is extended in structures
  with a diffused InGaAs-on-InP interface due to
  long As soaks.
• 77K PCD t increase to a convergent 170 ms due to
  reduced deep level emission rates.

Assume linear onset, analyzer error reflected in
2s confidence bands
1. Y. Sakai, M. Kudo, and C. Nielsen, J. Vac.
Sci, Technol. A 19, 1139- 1142 (2001). 2. M. K.
Hudait, Y. Lin, S. H. Goss, P. Smith, S. Bradley,
L. J. Brillson, S. W. Johnston, R. K.
Ahrenkiel, and S. A. Ringel. Submitted to Appl.
Phys. Lett. 2005.