Content uploaded by Tom Tiedje
Author content
All content in this area was uploaded by Tom Tiedje on Feb 24, 2015
Content may be subject to copyright.
Effect of molecular beam epitaxy growth conditions on the Bi content
of GaAs1−xBix
X. Lu,1,a兲D. A. Beaton,1R. B. Lewis,1T. Tiedje,1,2 and M. B. Whitwick1
1Advanced Materials and Process Engineering Laboratory, Department of Physics and Astronomy,
University of British Columbia, Vancouver V6T 1Z4, Canada
2Department of Electrical and Computer Engineering, University of Columbia, Vancouver V6T 1Z4, Canada
共Received 7 April 2008; accepted 10 April 2008; published online 15 May 2008兲
We describe how the Bi content of GaAs1−xBixepilayers grown on GaAs can be controlled by the
growth conditions in molecular beam epitaxy. Nonstandard growth conditions are required because
of the strong tendency for Bi to surface segregate under usual growth conditions for GaAs. A
maximum Bi content of 10% is achieved at low substrate temperature and low arsenic pressure, as
inferred from x-ray diffraction measurements. A model for bismuth incorporation is proposed that
fits a large body of experimental data on Bi content for a wide range of growth conditions. Low
growth rates are found to facilitate the growth of bismide alloys with a low density of Bi droplets.
©2008 American Institute of Physics.关DOI: 10.1063/1.2918844兴
The bismide alloy, GaAs1−xBix, has a number of interest-
ing properties that make it potentially useful for devices. For
example, bismuth incorporation produces a much larger re-
duction in the band gap of GaAs than In or Sb alloying, for
the same increase in lattice constant.1Also, Bi is the heaviest
nonradioactive element, therefore, it has a large spin orbit
splitting, which is useful for spin-based semiconductor de-
vices. Applications of this new alloy have been held back
by practical difficulties in growing films with high Bi con-
tent, due to the strong tendency for Bi to surface segregate
and form droplets on the surface under conventional GaAs
growth conditions. The highest Bi content reported to date
in GaAs1−xBixis 8%.2,3In this paper, we explore the influ-
ence of the growth conditions on the Bi content in molecular
beam epitaxy 共MBE兲growth of GaAs1−xBix, with the goal
of identifying conditions for making films with large Bi
concentrations. In particular, we show how the Bi concen-
tration depends on the Bi and As fluxes and the substrate
temperature.
The GaAs1−xBixsamples were grown on undoped GaAs
共100兲substrates in a solid source MBE system, equipped
with effusion cells for Ga and Bi, and a two-zone valved
cracker source of As2. The substrate temperature is moni-
tored during growth by optical band gap thermometry.4The
Ga flux was measured with a retractable ion gauge, and cali-
brated from the film thickness measured after growth using
x-ray diffraction. The As flux was obtained from the flux
gauge reading and the calibrated Ga flux using literature val-
ues for the relative ionization efficiency of Ga and As.5The
Bi flux was calibrated by depositing a Bi film on an unheated
wafer and estimating the total amount of deposited Bi from
the volume of the Bi islands observed in atomic force micro-
scope 共AFM兲images. The Bi flux is believed to be a combi-
nation of monomers and dimers.6With in situ diffuse light
scattering, we were able to detect the formation of metallic
droplets on the surface during growth.
Earlier experiments showed that the Bi incorporation is
substitutional.7In this case, x-ray diffraction measurements
of the lattice constant can be used to determine the Bi con-
tent of the GaAs1−xBixepilayers. The Bi content and epilayer
thicknesses were determined by fitting high resolution 关004兴
-2
scans using RADS MERCURY software from Bede Scien-
tific. The surface morphology was measured by AFM.
Figure 1shows
-2
x-ray diffraction scans with Cu K
␣
radiation for three GaAs1−xBixepilayers with Bi content
x=1.4%, 5%, and 10%. The weak x-ray interference fringes
observed in the x-ray scans compared with the simulations
are believed to be due to non-uniformity in the Bi content in
the growth direction as reported earlier.7One source of non-
uniformity in composition in the growth direction is the de-
lay associated with the buildup of a Bi surface layer after the
Bi shutter is opened. Under steady state growth, the surface
is typically saturated with Bi. A typical AFM image of a
GaAs1−xBixepilayer is shown in Fig. 2. This sample is 30 nm
thick with a Bi concentration of 3.6%. The rms surface
roughness is 0.76 nm, with surface features elongated in the
a兲Electronic mail: xianfeng@physics.ubc.ca.
FIG. 1. High resolution x-ray 关004兴
-2
scans for GaAs1−xBixepilayers
with Bi content of 1.4%, 5%, and 10%. The corresponding sample thick-
nesses are 152, 56, and 30 nm, respectively. All samples show weak inter-
ference fringes.
APPLIED PHYSICS LETTERS 92, 192110 共2008兲
0003-6951/2008/92共19兲/192110/3/$23.00 © 2008 American Institute of Physics92, 192110-1
Downloaded 16 May 2008 to 142.103.140.77. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
关011
¯
兴direction. Although this sample had a mirrorlike
appearance to the eye, it also showed 0.3
m diameter Bi
droplets with the density of 1.5⫻107cm−2.
Figure 3shows the Bi content of a number of films made
under different growth conditions plotted as a function of the
Bi/As flux ratio. The symbols represent the experimental
data and the lines are fits to the model discussed below. The
solid circles and squares correspond to two groups of
samples grown at the same temperatures and Bi fluxes, but
with varying As fluxes. These results show that the Bi con-
centration increases as the As flux decreases. The two groups
of samples represented by the open symbols and solid tri-
angles are all grown at the same As fluxes, but at different
temperatures and Bi fluxes. The Bi content increases with
decreasing temperature 共open symbols兲, and increasing Bi
flux 共solid triangles兲. At large Bi fluxes, the concentration of
Bi in the film saturates and Bi droplets eventually form on
the surface during growth.
We have developed a model which provides a quantita-
tive description of how the Bi content depends on the growth
conditions. The model is intended to fit the data as a guide to
future experiments and is based on the following simplified
picture of the surface composition and structure during
growth. Due to the strong tendency for Bi to surface segre-
gate and the relatively low Bi vapor pressure at the growth
temperatures of interest, a metallic Bi surface adlayer accu-
mulates on the surface during growth. This is the primary
source of Bi atoms for incorporation into the GaAs during
film growth. The metallic Bi layer has a surface coverage
equal to
Bi and lies on top of the GaAs1−xBixfilm. Arsenic
has a much higher vapor pressure than Bi and does not ac-
cumulate on the surface except insofar as it bonds to Ga.
Since the As flux exceeds the Ga flux we expect that the film
will be primarily As terminated under the Bi adatom layer.
Three processes can be identified that affect the Bi in-
corporation. These processes are schematically illustrated in
the inset of Fig. 3. In the first process, a Ga atom may insert
between the As-terminated surface of the film and a metallic
Bi atom from the adatom layer, forming an As–Ga–Bi bond.
Bismuth incorporation into the film is associated with the
formation of Ga–Bi bonds. The rate of this process will be
proportional to
BiFGa共1−x兲, where FGa is the Ga flux. The
factor 共1−x兲means that we are excluding the second process
in which a Ga atom inserts between a Bi atom bonded to the
surface and a Bi atom in the surface adatom layer. A ratio-
nale for this assumption is that the large size of the Bi atoms
does not favor the formation of next-neighbor Bi bonds of
the form Bi–Ga–Bi. The third process involves the insertion
of an As atom into a Ga–Bi bond, thereby displacing a Bi
atom bonded to Ga, back into the surface wetting layer.
Since a Ga–Bi bond is broken in this process, we assume that
it is thermally activated with a rate proportional to
FAse−U1/kTx. Putting these processes together we obtain the
following rate equation:
dx
dt ⬀
BiFGa共1−x兲−aFAse−U1/kTx.共1兲
The constant ais a dimensionless fitting parameter that takes
into account the relative cross sections for the Ga and As
insertion reactions. In steady state, we have dx/dt= 0. In this
case the rate equation can be solved for x,
x=
BieU1/kTFGa
aFAs +
BieU1/kTFGa
.共2兲
The Bi surface coverage
Bi is needed before we can
evaluate Eq. 共2兲. The surface coverage has been measured on
GaAs as a function of temperature and flux and found to
obey a Langmuir isotherm with a surface binding energy of
U0=1.8⫾0.4 eV.8Although the earlier studies were carried
out at higher fluxes and higher temperatures 共450–600 °C兲
than the present experiments, we expect the surface coverage
FIG. 3. The dependence of Bi content on growth conditions plotted as a
function of the Bi/As flux ratio. The symbols represent the experimental
data and the lines are fits based on the proposed model. The solid circles
共with Bi flux of 0.05 nm−2 s−2兲and solid squares 共with Bi flux of
15 nm−2 s−1兲correspond to two groups of samples grown at the same tem-
peratures and Bi fluxes, but with varying As fluxes. The samples represented
by the open symbols 共with As flux of 2.2 nm−2 s−1 andBifluxof
0.16 nm−2 s−1兲and solid triangles 共with As flux of 2.2 nm−2 s−1兲are all
grown at the same As fluxes, but at different temperatures and Bi fluxes. The
two solid lines cut off when the As/Ga flux ratio reaches unity. Further
reduction in As flux will cause Ga droplets to form.
FIG. 2. 共Color online兲Typical AFM image of GaAs1−xBixepilayer with
Bi content of 3.6%. The sample is 30 nm thick. Image shows a
500⫻500 nm2scan. The rms roughness is 0.76 nm, with surface features
elongated in the 关011
¯
兴direction.
192110-2 Lu et al. Appl. Phys. Lett. 92, 192110 共2008兲
Downloaded 16 May 2008 to 142.103.140.77. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
to be described by a similar Langmuir model in our case. An
important difference is that, in the present case, the surface
of interest is on a growing film, which absorbs some of the
Bi from the surface layer as it grows. Therefore, it is reason-
able to replace the Bi flux 共or pressure兲in the expression for
the Langmuir isotherm with the net flux, in which the incor-
porated Bi has been subtracted as follows:
Bi =b共FBi −xFGa兲eU0/kT
1+b共FBi −xFGa兲eU0/kT .共3兲
When substituted into Eq. 共2兲, this equilibrium expression
for the surface coverage gives the correct low temperature
limit for the Bi concentration, namely x=FBi /FGa, when all
the incident Bi atoms incorporate into the film.
Equation 共2兲provides a good description for the con-
centration of incorporated Bi as a function of the growth
conditions, as shown in Fig. 3. In this figure, the various
solid and broken lines are computed from the model in
Eq. 共2兲, with b=8.5⫻10−11 nm s, a=2.5⫻108,U0=1.3 eV,
and U1=0.8 eV. The surface binding energy 共1.3 eV兲is
similar to the value reported earlier 共1.8⫾0.4 eV兲if the
experimental uncertainties are taken into account. The solid
lines in Fig. 3show that the Bi content increases with de-
creasing As flux, when the Bi flux and the substrate tempera-
ture are held constant, in agreement with the model. The
broken lines show the dependence on Bi pressure and
substrate temperature, with the As flux held constant. In
particular, the broken line at 300 ° C shows that the Bi
content saturates at high Bi flux in the model, when other
growth parameters are kept fixed, in agreement with the
measurements.
In Eqs. 共2兲and 共3兲, the Ga flux, which is proportional to
the growth rate, enters only as a ratio with the Bi and As
fluxes. This means that if all three fluxes are scaled by the
same parameter, the Bi content will remain the same. One
could conclude that the growth rate is not a critical factor in
Bi incorporation. However, it turns out that the growth rate is
important in controlling the formation of Bi droplets. Bi at-
oms incident on the sample surface can do one of three
things: incorporate into the film, evaporate back into the va-
por or attach to a Bi droplet on the surface. In order to
prevent formation of Bi droplets, the flux of deposited Bi
atoms must be less than the sum of the rates of evaporation
and incorporation or
0⬍FBi −xFGa ⬍EBi.共4兲
The evaporation rate
Bi is controlled by the substrate tem-
perature and the Bi surface coverage
Bi which is typically
close to one, in the growth of high Bi concentration films. If
the inequality in Eq. 共4兲is violated, that is if the Bi flux
exceeds the incorporation rate by more than the rate of Bi
evaporation, then droplets will form. As the data in Fig. 3
show, in order to achieve a high Bi content, a low growth
temperature is required. At low temperatures, the Bi evapo-
ration rate will also be low. This means that, for low tem-
perature growths, the Bi flux must be rather precisely con-
trolled to match the incorporation rate, as the excess Bi has a
low evaporation rate. The ratio of the Bi evaporation rate to
the Bi incorporation rate can be viewed as a measure of the
process latitude. The process latitude can be maximized by
slowly growing, so that any excess Bi has a chance to evapo-
rate. In this case, the Bi flux does not have to be as accu-
rately matched to the incorporation rate. By reducing the
growth rate to 0.07
m/h, we have been able to grow mir-
rorlike films with Bi concentrations as high as 10%, and a
reduced density of Bi droplets 共1.7⫻106cm−2兲.
In conclusion, we have explored the effect of the growth
conditions on the Bi concentration in GaAs1−xBixgrown by
MBE. A model has been developed that describes the mea-
sured Bi content as a function of the Bi and As fluxes and the
substrate temperature. We find that low growth rates facili-
tate the growth of films with high Bi concentrations and low
Bi droplet density.
This work was supported by the Natural Sciences and
Engineering Research Council of Canada and Zecotek
Photonics.
1S. Tixier, M. Adamcyk, T. Tiedje, S. Francoeur, A. Mascarenhas, P. Wei,
and F. Schiettekatte, Appl. Phys. Lett. 82, 2245 共2003兲.
2K. Alberi, O. D. Dubon, W. Walukiewicz, K. M. Yu, K. Bertulis, and A.
Krotkus, Appl. Phys. Lett. 91, 051909 共2007兲.
3K. Bertulis, A. Krotkus, G. Aleksejenko, V. Pačebutas, R. Adomavičius,
G. Molis, and S. Marcinkevičius, Appl. Phys. Lett. 88, 201112 共2006兲.
4M. K. Weilmeier, K. M. Colbow, T. Tiedje, T. Van Buuren, and L. Xu,
Can. J. Phys. 69, 422 共1991兲.
5V. V. Preobrazhenskii, M. A. Putyato, O. P. Pchelyakov, and B. R.
Semyagin, J. Cryst. Growth 201, 170 共1999兲.
6R. F. C. Farrow, J. Vac. Sci. Technol. B 1, 222 共1983兲.
7P. Wei, S. Tixier, M. Chicoine, S. Francoeur, A. Mascarenhas, T. Tiedje,
and F. Schiettekatte, Nucl. Instrum. Methods Phys. Res. B 219,671
共2004兲.
8E. C. Young, S. Tixier, and T. Tiedje, J. Cryst. Growth 279,316共2005兲.
192110-3 Lu et al. Appl. Phys. Lett. 92, 192110 共2008兲
Downloaded 16 May 2008 to 142.103.140.77. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp