Semicond. Sci. Technol. 6
(1991)
C76-C79. Printed in the UK
The orientation independence of the
CdTe-HgTe
valence band offset as
determined by x-ray
photoelectron
spectroscopy
eR
Becker, Y S Wut, A Waag, M M Kraus and G Landwehr
Physikalisches Insmut
der
Universitat Wurzburg, 0-8700 Wurzburg, Federal
Republic of Germany
Abstract.
We have investigated the valence band offset
(.1.Ev)
of the CdTe-HgTe
heterojunction for three orientations,
(100),
(110)
and (111)8, using
In
situ x-ray
photoelectron spectroscopy. The difference in energy between the
Cd
4d
and
Hg
5d
512
core levels,
.1.E
CL
' and consequently
.1.Ev
was found to
be
independent of
surface orientation and the surface structure
Immediately
prior
to growth
of
the
uppermost layer.
AEv
was found to be 0.37 ±
0.07
eV.
1.
Introduction
The valence band offset (!1Ev) of
an
abrupt
CdTe-HgTe
heterojunction
is
an
important
parameter for the fabrica-
tion of devices as
well
as having a large effect on the
band
structure
of
quantum
wells
and
superlattices (see, for
example, Meyer
et
al
[1
]). !1Ev for the
CdTe-
HgTe inter-
face
has recently been the subject
of
a certain
amount
of
controversy. The
common
anion
rule predicts a small
valence
band
offset which was initially corroborated by a
magneto-optical investigation by
Guldner
et
al
[2],
i.e.
their measurements were consistent with a valence
band
offset
of
40 meV, However, a much larger value
of
350 meV was reported later by two separate x-ray photo-
electron spectroscopy
(xps) investigations
[3,4].
The
valence
band
offset was also found to be independent of
growth sequence
and
the thickness
of
the uppermost
layer from 5 to
35
A [4]. In addition, XPS
and
ultraviolet
photoemission spectroscopy
(ups) measurements by
Sporken et
al
[5]
have shown that the valence
band
offset
is
temperature independent
and
therefore the discre-
pancy between these two values can
not
be explained as
the result
of
a temperature dependence. The ensuing
controversy, which has been extensively reviewed by
Meyer
et
al
[1], appears
to
be resolved in favour
of
the
larger value.
Self-consistent tight-binding
(SCTB) calculations by
Munoz
et
al
[6]
have predicted a large dependence on
orientation,
180
meV between the (100)
and
(HO) sur-
faces
(AEv = 0.46
and
0.28 eV, respectively).
In
contrast,
t Permanent address: Institute
of
Physics, Chinese Academy
of
Science, Beijing. People's Republic
of
China.
0268-1242191/12OC76
+04
$03.50 © 1991101' Publishing Lld
Van de Walle
et
al
[7],
using self-consistent local density
functional methods, predict no surface dependence,
i.e.
AEv = 0.27
and
0.28 eV, respectively.
In
fact a more
general study by the latter
authors
suggests that this
independence is a characteristic of a number of impor-
tant interfaces, e.g.
CdTe·-HgTe, AIAs-GaAs
and
Si-Ge.
Indeed this has been
shown
to be the case for the
GaAs-AIAs heterojunction [8].
More
recent calculations
by
Muiioz
et
al
[9]
resulted in a much smaller depen-
dence on orientation,
90 me V between the (100) and
(110) surfaces
(IlEv
= 0.46
and
0.37
eV,
respectively).
In
order to determine which
is
correct for the
CdTe-HgTe
heterojunction
we
have investigated the effect of surface
orientation as well as the effect of interface structure
on
the valence
band
offset.
2.
Experimental details
Epitaxial growth was carried
out
in a four-chamber
RIBER
2300, molecular beam epitaxial
(MBE)
system
which has been modified to permit the growth
of
Hg-
based materials.
The
vacuum in the growth chamber
is
better
than
6 x
lO-10
Torr
when no
Hg
has recently
been admitted. Three
MBE cells were employed, two of
which were commercial cells
and
which contained high-
purity CdTe
and
Te.
The
third cell, designed by us,
is
a
stainless steel cell for
Hg
which can be refilled without
breaking the vacuum.
The
flux
of
the latter cell
is
stable
to within
±
1.5
and
± 3 % over a period
of
2
and
30
h,
respectively. The growth
chamber
is
connected with the
xps chamber
(3
x
lO-lO
Torr)
with a transfer system
whose vacuum was better
than
1 x
lO-9
Torr.
The CdTe-HgTe heterojunctions were grown
on
(110) CdTe and
on
(l00)- and (11l)B-oriented CdTe
and
CdZnTe substrates which had been degreased, chemo-
mechanicaUy polished for several minutes, etched in a
weak solution of bromine
in
methanol and rinsed
in
methanol. Immediately prior to loading the substrates
into the
MBE
system, they were rinsed
in
de-ionized water,
briefly dipped in hydrochloric acid
and
then rinsed in de-
ionized water so as to remove all of the original oxide
and
carbon from the substrate surface.
We
have found that, as
a result
of
this previous step. the newly formed oxide,
is
much more easily evaporated from the surface. This
is
accomplished by heating the substrates
at
temperatures
up to about
350
°C
while being monitored by reflection
high-energy electron diffraction (RHEED) as described
elsewhere [10]. Throughout this paper
we
consistently
use the convention
of
referring to the direction of the
incident electrons when referring
to
reconstruction in a
particular azimuth.
Approximately
0.1
and
2 1lm of CdTe were grown
on
CdTe and CdZnTe substrates, respectively. This growth
was initiated at
300
and 340 cC, respectively and contin-
ued while lowering the temperature to
230°C
where the
growth was completed. Then a thin layer, 6-40
A,
of
HgTe was grown
at
180°C.
For
the (100) and (1l1)B
orientations. this thin layer of HgTe was on a Te-
stabilized surface as
well
as a surface displaying attri-
butes
of
both Te and Cd stabilization. The former surface
structure was established by exposing the CdTe
film
to a
Te
flux
of 3 x
10
- 7
Torr
at
210
0C.
The latter surface
structure was established by evaporating Te from the
surface at about 340°C for several minutes while main-
taining a smooth surface as evidenced by the presence of
uniform streaks and the absence of spots in the RHEED
pattern. In the (100) case this mixture of Te and Cd
stabilization
of
this surface
is
characterized by half-order
reconstruction in the [011] and [010] azimuths. The
(1
t t)B surface
is
more complicated. Here the Te
(1
x
1)
stabilized surface undergoes a transition
to
(2j3
x
2j3)R30°
reconstruction upon evaporation
of
Te from
HgTe
CdTe
-----r,---
E~dTe
E v
_...;v
__
-+-
__
[CdTe
HgTe
y,
l1E
[~dTe
(
EH9
Te
_ EH9
Te
)
Hg5d
5/2
v
~
(
ECdTe_
ECdTe)
Cd4d
v
E~~;~5/2
LL
l1ECL
v
[CdTe
Cd4d
Figure 1. A schematic diagram of the relevant energy
levels
01
an abrupt CdTe-HgTe heterojunction.
Valence band offset of CdTe-HgTe
by
xps
the surface. The as-grown surface structure in the (110)
orientation could not be changed either with excess Te
flux
or
with
an
increase in temperature up to 340°C, and
therefore HgTe was grown only
on
the as-grown CdTe
surface.
The resulting heterojunctions were transferred via an
ultra-high-vacuum transfer system and investigated by
XPS
under nearly
in
situ conditions.
XPS
experiments were
preformed with a RIBER MAC 2 electron spectrometer
using
an
Mg
KO(
x-ray source (1253.6 e
V)
with
an
acceler-
ation voltage of
10
kV, a current
of
10
mA
and
without a
monochromator. The energy scans were repeated for at
least
12
h in order to achieve
an
acceptable signal-to-
noise ratio.
3. Results and
discussion
The valence band offset AEv is schematically shown in
figure 1 and
is
given by
AEv =
(E~:Xd512
-
E~gTe)
-
(E~~Id
-
E~dTe)
+ AE
CL
•
(1)
Therefore, in order
to
determine AE
v
,
we
have
to
mea-
sure these three binding energy differences for HgTe,
CdTe and the CdTe-HgTe heterojunction. Where
E~mSI2
and
E~~Id
are binding energies
of
the Hg 5d
sl2
and Cd 4d core levels in HgTe
and
CdTe, respectively.
E~gTc
and
E~dTe
are the energies of the valence band
maxima
in
HgTe
and
CdTe, respectively,
and
AEcL
is
the
binding energy difference between the Hg
5d5J2
and
Cd
4d core levels in the HgTe-CdTe heterojunction. This
procedure results in a value of 0.37 ± 0.07
eV
for AEv for
the (100) orientation. The large uncertainty
is
due pri-
marily to the difficulty in determining the position
of
the
valence band maximum (see figure
2).
As
can be seen by
comparing the XPS spectra for (100) and (110) CdTe in
figure
2,
the energy difference between the
Cd
4d core
level and valence band maximum in CdTe
is
independent
of these two surface orientations. The same
is
true for
80
60
:l
I'd
40
(Jl
I-
Z
:::J
0
20
u
0
12
9
(100l
edTe:
Full
I'n.
(t10)
CdTe:
Dotted
l,ne
6 3
BINDING
ENERGY
(eV)
o
Figure 2.
xps
spectra
of
(100) and (110) CdTe, showing the
Cd
4d
core
level peak and the CdTe valence band
maximum.
e77
C A Becker at
al
60
:J
'"
4Q
U1
r-
Z
::1
0
.......
r.\'J
U
0
B 6
ENERGY
(eV)
Figure
3.
An
XPS spectrum of a (110) CdTe-HgTe
heterojunctlon (full curve) and a least-squares fit of the XPS
spectra of a
(110)
CdTe and
(110)
HgTe film (dotted curve)
in the vicinity of the
Cd
4d
and Hg
5d
core
levels.
(111)8. This demonstrates
that
this energy difference is a
bulk property
and
is independent
of
orientation as is
usually assumed. Consequently all
orientation
or
inter-
face
effects
on
6.Ev
are
contained in
6.E
CL
which can be
determined with greater precision
than
the
position
of
the valence band.
An
XPS spectrum for a (110)
CdTe-HgTe
heterojunc-
tion is shown in figure
3
and
the corresponding spectra
for CdTe and HgTe epitaxial films in the region
ofthe
Cd
4d and
Hg
5d core levels can be
~een
in figure
4.
6.E
CL
can-
not be determined directly for the heterojunction due to
the overlapping
of
the
Cd
4d,
Hg
5d
3
/
2
and
Hg
5d core
levels. First the HgTe
and
CdTe
spectra were
co~bined
and fitted to the heterojunction spectrum by means of a
least-square procedure. This least-square fit as well as the
heterojunction spectrum are plotted in figure
3.
In
order
to more accurately determine the position of the peak
60
::I
III
40
lrl
I-
Z
::J
0
20
U
0
B 6
ENERGY
(eV)
Figure 4.
xps
spectra of (110) CdTe and (110) HgTe films
in
the vicinity
of
the
Cd
4d
and
Hg
5d
core levels.
C78
Table 1. The energy difference
between the
Cd
4d
and Hg
5d
s/2
core levels
for
the (100),
(110)
and
(111)
B orientations.
Orientation
(100)
(110)
(111
)8
2.96 ± 0.03
3.01
± 0.03
2.98 ±
0.03
due
to
the
Hg
5d
~/2
core level it was resolved from the
HgTe spectrum as depicted by the dotted curve in figure
4.
This
was
accomplished by assuming
that
this peak
is
symmetricaL i.e. that the high-energy flank is a mirror
image
of
the low-energy flank. This should
be
a good
assumption because only minimal inelastic scattering.
is
expected under the low-energy flank.
6.E
CL
was then
determined
by
measuring the energy difference between
the peak centres
at
half maximum
of
the Cd
4d
and
Hg
5d
s
/
2
core levels.
The
results for the (100), (110)
and
(llJ)B
orienta-
tions are
tabulated
in table
1.
Two different interface
structures were investigated for the
(100)
and
(lll)B
orientations.
in
the first case HgTe was grown
on
(100)-(2 x 1)
and
(111)B-(1 x 1) Te-stabilized
CdTe
sur-
faces respectively, whereas
in
the second case HgTe was
grown on a mixed surface with half-order reconstruction
in
both
the
[OH]
and
[010] azimuths for the
(lOO)
orientation
and
(2v13 x
2j3)R30°
reconstruction for
(111)B. In each case the two corresponding values are the
same within the experimental uncertainities given in
table
1.
As
can
be seen in table
1,
6.E
CL
and
therefore tlEv
for the CdTe-HgTe heterojunction
is,
within experimen-
tal uncertainty
(±0.03
eV), independent
of
orientation
and
interface structure.
4.
Conclusions
In
conclusion, by means
of
in
situ
XPS
experiments we
have shown
that
6.E
CL
(and therefore
that
tlEv) for the
CdTe-
HgTe heterojunction
is
independent
of
the surface
orientation
and
the surface reconstruction immediately
prior to the
growth
of
HgTe, whether Te-stabilized
or
a
mixture of
Cd
and
Te
stabilization. These results agree
with the self-consistent local density predictions
of
Van
de Wane
et
at [7]
but
not
with the SCTB calculations
of
Muiioz
et
01
[6, 9]. Furthermore, 6.Ev has been deter-
mined
to
be 0.37 ± 0.07
eV,
in good agreement with the
literature
[3-5].
Acknowledgments
This project was supported by the Bundesministerium
fUr
Forschung und Technlogie and the Deutsche
F orsch ungsgemeinschaft.
References
[1] Meyer J
R,
Hoffman C A and Bartoli F J 1990
Semieond.
Sei.
Technol. 5 S90
[2] Guldner
Y,
Bastard G, Vieren J
P,
Voos
M,
Faurie J P
and Million A 1983 Phys. Rev. Left.
51
907
[3]
Kowalczyk S P, Cheung ] T, Kraut E A and Grant R
W 1986
Phys. Rev. Left. 56
1605
[4] Faurie J P, Hsu C and Due T M
1987
J.
Vae.
Sci.
Technol.
A 5 3074
[5J Sporken
R,
Sivananthan
S,
Faurie J
P,
Ehlers D H.
Fraxedas J,
Ley
L.
Pireaux J J and Caudano R
1989
Valence band offset of CdTe-HgTe by XPS
J.
Vac. Sci. Technol. A 7 427
[6J Muiioz
A,
Sanchez-Dehesa J
and
Flores F 1987 Phys.
Rev. B 35 6468
[7]
Van de Walle C G and Martin R M
1988
Phys Rev. B
374801
[8]
Hirakawa K, Hashimoto Y and Ikoma T 1990 Appl.
Phys. Lett.
57
2555
(9]
Munoz
A, Sanchez-Dehesa J and Flores F 1988
Phys.
Rev. B 37 4803
[10]
Wu Y
S,
Becker C
R.
Waag
A,
Bieknell-Tassius R N
and Landwehr G
1991
J.
Appl. Phys. 69268
e79