Effects of a semiconductor matrix on the band anticrossing in dilute group II-VI oxides

Abstract

The effect of a semiconductor matrix on the band anticrossing interaction is studied for four different dilute-oxide material systems: ZnSO, ZnSeO, ZnTeO, and ZnCdTeO. The choice of host material allows for independent control of the energy separation between the conduction band edge and the O energy level as well as the coupling parameter. The transition energies measured by photoreflectance and optical absorption are well explained by the band anticrossing model with the coupling parameter increasing from 1.35 eV for ZnSO to 2.8 eV for ZnTeO and showing approximately linear dependence on the electronegativity difference between O and the host anion.

Full text

This content has been downloaded from IOPscience. Please scroll down to see the full text.
Download details:
IP Address: 144.214.25.79
This content was downloaded on 16/10/2015 at 03:25
Please note that terms and conditions apply.
Effects of a semiconductor matrix on the band anticrossing in dilute group II-VI oxides
View the table of contents for this issue, or go to the journal homepage for more
2015 Semicond. Sci. Technol. 30 085018
(http://iopscience.iop.org/0268-1242/30/8/085018)
Home Search Collections Journals About Contact us My IOPscience

Effects of a semiconductor matrix on the
band anticrossing in dilute group II-VI oxides
MWełna
1,2
, R Kudrawiec
1
, Y Nabetani
3
, T Tanaka
4,5
, M Jaquez
2,6
,
O D Dubon
2,7
,KMYu
2,8
and W Walukiewicz
2
1
Department of Experimental Physics, Wroclaw University of Technology, Wybrzeze Wyspianskiego 27,
50-370 Wrocław, Poland
2
Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA
3
Department of Electrical Engineering, University of Yamanashi, Takeda 4-3-11, Kofu 400-8511, Japan
4
Department of Electricaland Electronic Engineering, Saga University, 1 Honjo, Saga 840-8502, Japan
5
PRESTO, Japan Science and Technology Agency (JST), Kawaguchi, Saitama 332-0012, Japan
6
Department of Mechanical Engineering, University of California, Berkeley, CA 94720, USA
7
Department of Materials Science and Engineering, University of California, Berkeley, CA 94720, USA
8
Department of Physics and Materials Science, City University of Hong Kong, Kowloon, Hong Kong
E-mail: monika.welna@pwr.edu.pl and robert.kudrawiec@pwr.edu.pl
Received 1 April 2015, revised 9 June 2015
Accepted for publication 22 June 2015
Published 28 July 2015
Abstract
The effect of a semiconductor matrix on the band anticrossing interaction is studied for four
different dilute-oxide material systems: ZnSO, ZnSeO, ZnTeO, and ZnCdTeO. The choice of
host material allows for independent control of the energy separation between the conduction
band edge and the O energy level as well as the coupling parameter. The transition energies
measured by photoreflectance and optical absorption are well explained by the band anticrossing
model with the coupling parameter increasing from 1.35 eV for ZnSO to 2.8 eV for ZnTeO and
showing approximately linear dependence on the electronegativity difference between O and the
host anion.
Keywords: II-VI semiconductors, band gap, highly mismatched alloy, intermediate band gap
(Some figures may appear in colour only in the online journal)
1. Introduction
Highly mismatched alloys (HMA) are a class of semi-
conductors that are formed by substituting constituent atoms
with isovalent atoms with distinctly different electronegativity
and/or ion size. Dilute-oxide II-O-VI HMAs are formed when
column VI atoms are partially replaced by oxygen. These
materials are a group II-VI equivalent of dilute group III-V
nitrides in which column V atoms are partially replaced with
nitrogen [1–8]. The electronic band structure of those II-O-VI
HMAs is described by the band anticrossing (BAC) model
that considers an interaction between localized states of O
atoms and extended states of the semiconductor matrix. The
band anticrossing interaction splits the conduction band into
two subbands E
−
and E
+
. The splitting energy and the dis-
persion relations for the subbands depend on the electro-
negativity and size mismatch between the host and O atoms
and the location of the O level relative to the conduction band
edge of the matrix.
O’Reilly et al [9] analyzed the effect of a range of
N-related defect levels associated with isolated N atoms, N–N
pairs, and larger clusters of N atoms introduced by replacing
As by N. It has been shown that for GaAsN alloys, the two-
level BAC model provides a good qualitative explanation;
however, for other alloys (GaPN, GaSbN), it is necessary to
include the details of the distribution of N-related defect
levels to obtain a quantitative understanding of the conduction
band structure in dilute nitride alloys. In another work, Mudd
et al [10] proposed a three-level BAC model, which takes into
account interaction between localized levels of isolated N
atoms and N–N pairs with the conduction band of the host
material and allows one to obtain a good quantitative expla-
nation for band gap dependence on N content in GaSbN
alloys.
Semiconductor Science and Technology
Semicond. Sci. Technol. 30 (2015) 085018 (6pp) doi:10.1088/0268-1242/30/8/085018
0268-1242/15/085018+06$33.00 © 2015 IOP Publishing Ltd Printed in the UK1

We believe that the schema of analysis in materials group
II-VI-O is similar to the one described previously for group
III-V-N, and application of the two-level BAC model pro-
vides excellent explanation of experimental results that will
be discussed in this paper. Although there are several reports
on band anticrossing in dilute oxides [11–14], there has been
no systematic study of the effects of the host compound on
the BAC interaction.
In this paper, we report a study of the effects of a
semiconductor matrix on the band anticrossing interaction in
different group II-VI dilute oxides. Four different material
systems have been investigated (ZnSO, ZnSeO, ZnTeO, and
ZnCdTeO) with different O compositions. Choice of these
host materials allows for separation of different contributions
to the strength of the band anticrossing interaction.
For HMA, the BAC Hamiltonian is given by
HEk C x
CxE
() (1)
BAC
MOM
OM O
⎛
⎝
⎜⎞
⎠
⎟
=
where E
M
(k) is the energy dispersion of conduction band in
the host material (e.g., ZnS, ZnTe, etc.) and E
O
is the energy
of the localized O states, xis the oxygen concentration and
C
OM
is a coupling constant that reflects the strength of
interaction between oxygen states and conduction band states
of the host material and is composition independent. The band
anticrossing interaction results in the formation of E
−
and E
+
subbands with dispersion relations given by
[
[]
Ek E E k
EEk xC
() 1
2()
() 4 . (2)
OM
OM
2
OM
2
=+
±− +
±
Equation (2) indicates that the subband separation
depends on the energy difference between the E
O
and the
conduction band edge E
M
(0) and on the coupling para-
meter C
OM
.
Figure 1shows the energy of the E
O
level relative to
E
M
(0) in the semiconductors used in our study [15]. The O
level is located at the same energy of 0.2 eV below the con-
duction band edge (CBE) in ZnTe and ZnS. Therefore, the
difference in the BAC interaction in ZnTeO and ZnSO can be
attributed solely to a difference in the coupling parameter for
those two alloys. On the other hand, the location of the O
level relative to the CBE can be varied with composition for
ZnCdTe alloys. Assuming that the coupling parameter
remains constant for small Cd content provides the ability to
study the effects of varying energy separation between E
M
(0)
and E
O
on the anticrossing interaction. We have also studied
ZnSeO alloys in which the O level is located above the CBE
of the ZnSe matrix and the electronegativity of Se has an
intermediate value between the electronegativities of S
and Te.
In [16], theoretical calculations of band alignment in II-
VI group semiconductors have been shown. The authors state
that the oxygen level is localized in various energy positions
depending on the host material. However, taking into account
a certain degree of uncertainty of theoretical predictions and
the fact that the obtained differences in the energy positions
are rather small, the assumption made in the BAC model, that
energy of O level in the absolute scale is constant for different
materials, is correct within this uncertainty.
2. Experimental details
ZnSO layers with different oxygen concentrations were
deposited on sapphire substrates in a dual gun sputtering
system at 240 °C. The background pressure was 5 mTorr and
pure argon atmosphere was maintained. The thickness and the
composition of the layers were measured by Rutherford
backscattering spectrometry (RBS). Good optical quality
ZnCdTeO and ZnTeO layers were deposited on ZnTe sub-
strates by molecular beam epitaxy (MBE). The ZnSeO layers
were grown on GaAs substrates with 100 nm thick ZnSe
buffer layers by MBE Further details regarding the film
deposition can be found in [13,17,18].
The optical properties of the studied materials were
determined using photoreflectance (PR) and optical absorp-
tion spectroscopies. PR is a very sensitive and nondestructive
modulation technique that has been widely and successfully
used to study optical transitions in semiconductors
[8,11,13,17,19,20]. In our study, we used a bright con-
figuration setup [21]. A 150 W halogen lamp was used as a
probe beam and the pump light sources were a series of
Figure 1. Position of the localized O level with respect to the
conduction and valence bands in group II-VI semiconductors. The
energy scale is relative to the vacuum level. The valence and
conduction bands’positions are taken after [15].
2
Semicond. Sci. Technol. 30 (2015) 085018 MWełna et al

different wavelength (442, 404, 325, or 266 nm) mechanically
chopped laser beams. PR signal was detected using a lock-in
system and a Si photodiode. Optical absorption measurements
were performed using a Perkin Elmer Lambda 950 UV/Vis/
NIR spectrophotometer.
The PR measurements were successfully carried out for
ZnSeO, ZnTeO, and ZnCdTeO alloys. In the case of the
ZnSO alloy films grown on sapphire substrates, the large
difference in the refractive index of the substrate and epilayer
resulted in strong Fabry–Perot (F-P) oscillations, making the
analysis of PR features unreliable [22–24]. Therefore, optical
absorption was used to determine the interband transitions in
this alloy. We note that since both PR and optical absorption
measure the energy of the optical transitions from the VBE to
E
−
afitting of equation (2) based calculations to the experi-
mental data requires that E
O
and E
M
(0) are taken relative to
the VBE of the matrix.
3. Results and discussion
Because of the small difference between S and O atoms,
ZnSO alloys can be synthesized in the entire composition
range [25]. Since we are only interested in the band antic-
rossing effects in dilute oxides, this study is limited to ZnSO
alloys with up to 8% oxygen content. The absorption edge
energy given by the energy difference between the E
−
sub-
band and the valence band edge was determined by fitting the
absorption curves using a procedure based on the BAC model
as described in [26]. The results are shown in figure 2(a). The
figure also includes the results of a previous study where the
absorption edge energy was determined using the standard
extrapolation procedure of the square of the absorption
coefficient (α) for parabolic bands [25]. All experimental data
are in good agreement with the BAC model (red dashed line)
with the BAC parameters as follows: E
O
= 3.5 eV relative to
the valence band edge of ZnS and C
OM
= 1.35 eV.
The systematic study of the ZnSeO films grown on GaAs
reveals two optical transitions from the valence band to the E
−
and E
+
subbands. The transition energies obtained from low
temperature PR measurements are in good agreement with
predictions found using the BAC model when small, although
significant effects of strain resulting from the lattice mismatch
between ZnSeO film and the GaAs substrate are included in
the calculations. Results of measurements and calculations
can be seen in figure 2(b). BAC parameters used in the cal-
culation are: E
O
= 2.96 eV relative to the valence band edge of
ZnSe and C
OM
= 1.5 eV. More details of the analysis of this
data can be found in [24].
ZnTeO alloys are very different than ZnSO despite
having the O level similarly located approximately 0.2 eV
below the CBE: there is a larger electronegativity and atom
size mismatch between O and Te than compared to the dif-
ference between O and S. ZnTeO layers with up to 1.6% O
content were synthesized. Because tellurium atoms are larger
than oxygen, good quality layers can be obtained for only a
few percent of oxygen (replacing Te). Figure 3(a) shows
energies of optical transitions obtained from analysis of PR
spectra (black diamonds) plotted together with results repor-
ted by Tanaka et al (open blue circles) [13] and BAC cal-
culations (red dashed line). Excellent agreement in the
calculations was obtained with the following BAC fitting
parameters: E
O
= 2.02 eV relative to the valence band edge of
ZnTe and C
OM
= 2.8 eV.
In order to study the effect of the O level location relative
to the CBE, we have studied BAC effects in ZnCdTe with up
to 11% Cd and up to 3.1% O. A significant advantage of
ZnCdTeO HMAs is that the alloys are lattice matched to
Figure 2. Energies of the optical transitions extracted from PR and
absorption spectra together with energies of transitions calculated
with the BAC model for different material systems (a) ZnSO and (b)
ZnSeO taking into account valence band splitting into heavy and
light holes (red and black dashed lines, respectively). Dashed grey
lines show energies of the localized states of oxygen. A red dot is the
band gap energy for ZnS.
3
Semicond. Sci. Technol. 30 (2015) 085018 MWełna et al

ZnTe substrates for the Cd/O atomic ratio of ∼3.5. Lattice
matching typically improves optical quality of alloys, elim-
inates strain, and enhances O incorporation. Energies of the
optical transitions between the VBE and E
−
for lattice mat-
ched ZnCdTeO layers measured by PR and optical absorption
are shown in figure 3(b).
The change in the energy gap between ZnTe and CdTe
originates mostly from the shift of the conduction band
(figure 1). Alloying of ZnTe with CdTe shifts the CBE by
about 8 meV downward per atomic percent of Cd. Therefore,
for our studied samples, the location of the O level relative to
CBE varies from 0.2 eV in ZnTe to 0.11 eV in
Zn
0.89
Cd
0.11
Te. Adopting this dependence of the location of
the O level on Cd composition, we use equation (2) to cal-
culate the energy difference between the E
−
subband and
VBE as a function of O content. Figure 3(b) shows that the
experimental results for the lattice matched layers can be well
explained by the BAC model with E
O
= 2.02 eV above the
VBE of ZnTe and a composition independent C
OM
of 2.8 eV.
As discussed previously, the shift of the conduction band with
Cd concentration has to be considered for the lattice-matched
ZnCdTe alloys [27]. The composition independent value of
C
OM
is understandable because in both ZnCdTeO and
ZnTeO, tellurium atoms are substituted by oxygen atoms and
the effect of the change of the lattice parameter is too small to
be detected. Note that the coupling parameter of C
OM
= 2.8 eV
used to explain our data is somewhat higher than the value
C
OM
= 2.2 eV used previously for CdTeO [28] and
ZnCdTeO [17].
The analysis of the optical data allowed us to determine
the coupling parameters for different group II-VI HMAs in
which O substitutes atoms with different electronegativities
and atom sizes. The coupling parameters obtained from fitting
of the experimental results along with values reported in the
literature are listed in table 1. There is a good agreement
between this work and previously determined values of dif-
ferent parameters [11,12,29]. The only exception is the C
OM
for ZnSeO, where our value of 1.5 eV is about 20% lower
than the previously determined values. Figure 4shows the
dependence of the C
OM
on the anion electronegativity [30].
There is a clearly observed increase of the coupling parameter
with an increase in electronegativity difference between O
and the host lattice anion. For a specific column in the peri-
odic table the electronegativity scales with the atom size [31],
and a similar dependence is shown between the value of the
coupling parameter and the atomic size of the host matrix
anions. Substitution of the host atom by an isovalent atom
with larger electronegativity leads to the formation of a
Figure 3. Comparison of experimental data obtained from PR and
absorption spectra with BAC calculations for (a) ZnTeO and (b)
ZnCdTeO. The dashed grey lines show energies of the localizedd
states of oxygen. A red dot is the band gap energy for ZnTe.
Figure 4. The coupling parameter of the band anticrossing in dilute
II-VI oxides plotted as a function of electronegativity difference. The
dashed line is a guide for the eyes.
4
Semicond. Sci. Technol. 30 (2015) 085018 MWełna et al

stronger local potential in the immediate vicinity of this atom.
The interaction between the localized oxygen level and con-
duction band of the host matrix can be seen as a stronger band
gap bowing for HMAs, which can be explained in terms of
the BAC model. The bigger the difference in electronegativity
between host and substitutional atoms, the stronger the
anticrossing interaction becomes. To successfully describe
such behavior of E
−
and E
+
bands in HMAs with a bigger
difference in electronegativity, a larger value for the coupling
constant is needed. One can attribute this increase with
stronger local potential of substitutional atom. It would be
interesting to find if a similar scaling is observed in less ionic
dilute nitride HMAs.
4. Conclusions
In summary, we have studied the band anticrossing effects in
four different HMA systems with different locations of the O
level relative the CBE of the host compound and different
electronegativity differences between O and the anions of the
host compound material. Optical transition energies are well
explained by the BAC interaction with the coupling parameter
scaling linearly with the electronegativity difference between
O and the host matrix anions.
Acknowledgments
This work was performed within the grant of the National
Science Centre ETIUDA no. 2013/08/T/ST3/00400 and
HARMONIA 2013/10/M/ST3/00638. The work performed at
LBNL was supported by the Director, Office of Science,
Office of Basic Energy Sciences, Materials Sciences and
Engineering Division, of the US Department of Energy under
Contract No. DE-AC02-05CH11231.
References
[1] Shan W 1999 Band Anticrossing in GaInNAs Alloys Phys.
Rev. Lett. 82 1221–4
[2] Wu J, Shan W and Walukiewicz W 2002 Band anticrossing in
highly mismatched III–V semiconductor alloys Semicond.
Sci. Technol. 17 860
[3] Buyanova I and Chen W 2004 Physics and Applications of
Dilute Nitrides (New York: CRC Press)
[4] Henini M 2005 Dilute Nitride Semiconductors (Amsterdam:
Elsevier) ch 10, pp 325–59
[5] Weyers M, Sato M and Ando H 1992 Red shift of
photoluminescence and absorption in dilute GaAsN alloy
layers Jpn. J. Appl. Phys. 31 L853
[6] Tisch U, Finkman E and Salzman J 2003 The anomalous
composition dependence of the bandgap of GaAsN Phys.
Status Solidi A195 528–31
[7] Kim T S, Cuong T V, Park C S, Park J Y, Lee J, Suh E-K and
Hong C-H 2003 Composition dependence of the band-gap
energy of GaAsN alloys J. Korean Phys. Soc. 43 273–6
[8] Kudrawiec R, Latkowska M, Misiewicz J, Zhuang Q,
Godenir A M R and Krier A 2011 Photoreflectance study of
N- and Sb-related modifications of the energy gap and spin–
orbit splitting in InNAsSb alloys Appl. Phys. Lett. 99
011904
[9] O’Reilly E P, Lindsay A, Klar P J, Polimeni A and Capizzi M
2009 Trends in the electronic structure of dilute nitride
alloys Semicond. Sci. Technol. 24 033001
[10] Mudd J J, Kybert N J, Linhart W M, Buckle L, Ashley T,
King P D C, Jones T S, Ashwin M J and Veal T D 2013
Optical absorption by dilute GaNSb alloys: influence of N
pair states Appl. Phys. Lett. 103 042110
[11] Shan W, Walukiewicz W, Iii J W A, Yu K M, Wu J,
Haller E E, Nabetani Y, Mukawa T, Ito Y and Matsumoto T
2003 Effect of oxygen on the electronic band structure in
ZnO
x
Se
1−x
alloys Appl. Phys. Lett. 83 299–301
[12] Chen W-Y, Lai C-W, Cheng C-C, Chen C-Y, Chyi J-I and
Hsu T-M 2012 Photomodulation reflectance study of
temperature dependence of the band gap of ZnSe
1-x
O
x
Phys.
Status Solidi C9187–9
[13] Tanaka T, Kusaba S, Mochinaga T, Saito K, Guo Q, Nishio M,
Yu K M and Walukiewicz W 2012 Molecular beam
epitaxial growth and optical properties of highly mismatched
ZnTe
1−x
O
x
alloys Appl. Phys. Lett. 100 011905
[14] Deng H-X, Li J, Li S-S, Peng H, Xia J-B, Wang L-W and
Wei S-H 2010 Band crossing in isovalent semiconductor
alloys with large size mismatch: First-principles calculations
of the electronic structure of Bi and N incorporated GaAs
Phys. Rev. B82 193204
[15] Van de Walle C G and Neugebauer J 2003 Universal alignment
of hydrogen levels in semiconductors, insulators and
solutions Nature 423 626–8
[16] Li J and Wei S-H 2006 Alignment of isovalent impurity levels:
oxygen impurity in II-VI semiconductors Phys. Rev. B73
041201
[17] Tanaka T, Nagao Y, Mochinaga T, Saito K, Guo Q, Nishio M,
Yu K M and Walukiewicz W 2013 Molecular beam
epitaxial growth of ZnCdTeO epilayers for intermediate
band solar cells J. Cryst. Growth 378 259–62
[18] Nabetani Y, Mukawa T, Okuno T, Ito Y, Kato T and
Matsumoto T 2003 Structure and optical properties of
Table 1. E
O
and C
OM
parameters obtained from fitting of experimental data using the BAC model for dilute II-VI oxides together with
previously reported data. The table also includes the electronegativity difference between O and the host lattice anions, as well as the size of
the atoms substituted by O atoms.
E
O
(eV) C
OM
(eV) Electronegativity difference [30] Atomic radius difference (pm) [31]
This work Literature This work Literature
ZnTeO 2.02 1.98 [12] 2.80 3.20 [12] 1.45 75
ZnSeO 2.96 2.90 [11,29] 1.80 [11]
(at 80 K) (at 300 K) 1.50 1.95 [29] 1.18 55
ZnSO 3.50 —1.35 —1.02 40
5
Semicond. Sci. Technol. 30 (2015) 085018 MWełna et al

ZnSeO alloys with O composition up to 6.4% Mater. Sci.
Semicond. Process. 6343–6
[19] Motyka M, Kudrawiec R and Misiewicz J 2007 On the
deepness of contactless electroreflectance probing in
semiconductor structures Phys. Status Solidi A204 354–63
[20] Kudrawiec R, Luce A V, Gladysiewicz M, Ting M, Kuang Y J,
Tu C W, Dubon O D, Yu K M and Walukiewicz W 2014
Electronic band structure of GaN
x
P
y
As
1-x-y
highly
mismatched alloys: suitability for intermediate-band solar
cells Phys. Rev. Appl. 1034007
[21] Kudrawiec R and Misiewicz J 2009 Photoreflectance and
contactless electroreflectance measurements of
semiconductor structures by using bright and dark
configurations Rev. Sci. Instrum. 80 096103
[22] Kudrawiec R, Misiewicz J, Wachnicki Ł, Guziewicz E and
Godlewski M 2011 Contactless electroreflectance of ZnO
layers grown by atomic layer deposition at low temperature
Semicond. Sci. Technol. 26 075012
[23] Wełna M, Kudrawiec R, Kaminska A, Kozanecki A,
Laumer B, Eickhoff M and Misiewicz J 2013 Contactless
electroreflectance studies of free exciton binding energy in
Zn
1-x
Mg
x
O epilayers Appl. Phys. Lett. 103 251908
[24] Welna M, Kudrawiec R, Nabetani Y and Walukiewicz W 2014
Band anticrossing in ZnOSe highly mismatched alloy Appl.
Phys. Express 7071202
[25] Meyer B K, Polity A, Farangis B, He Y, Hasselkamp D,
Krämer T and Wang C 2004 Structural properties and
bandgap bowing of ZnO
1−x
S
x
thin films deposited by
reactive sputtering Appl. Phys. Lett. 85 4929–31
[26] Mayer M A, Sp eaks D T, Yu K M, Mao S S,
Haller E E and Walukiewicz W 2010 Band structure
engineering of ZnO
1−x
Se
x
alloys Appl. Phys. Lett. 97
022104
[27] Samanta B, Sharma S L and Chaudhuri A K 1995 Study of
the microstructure and optical properties of polycrystalline
Cd
1−x
Zn
x
Te thin films Vacuum 46 739–43
[28] Yu K M, Walukiewicz W, Shan W, Wu J, Beeman J W,
Scarpulla M A, Dubon O D and Becla P 2004 Synthesis and
optical properties of II-O-VI highly mismatched alloys
J. Appl. Phys. 95 6232–8
[29] Broesler R, Haller E E, Walukiewicz W, Muranaka T,
Matsumoto T and Nabetani Y 2009 Temperature
dependence of the band gap of ZnSe
1−x
O
x
Appl. Phys. Lett.
95 151907
[30] Allen L C 1989 Electronegativity is the average one-electron
energy of the valence-shell electrons in ground-state free
atoms J. Am. Chem. Soc. 111 9003–14
[31] Clementi E, Raimondi D L and Reinhardt W P 1967 Atomic
screening constants from SCF functions: II. Atoms with 37
to 86 electrons J. Chem. Phys. 47 1300–7
6
Semicond. Sci. Technol. 30 (2015) 085018 MWełna et al