View Article Online / Journal Homepage / Table of Contents for this issue

PCCP

Dynamic Article Links

Cite this: Phys. Chem. Chem. Phys., 2012, 14, 47894795

PAPER

Published on 01 March 2012. Downloaded by Chungnam National University on 07/11/2014 08:14:00.

www.rsc.org/pccp

Design of energy band alignment at the Zn1xMgxO/Cu(In,Ga)Se2

interface for Cd-free Cu(In,Ga)Se2 solar cellsw
Chang-Soo Lee,a Liudmila Larina,a Young-Min Shin,a Essam A. Al-Ammarb and
Byung Tae Ahn*a
Received 6th February 2012, Accepted 7th February 2012
DOI: 10.1039/c2cp40355b
The electronic band structure at the Zn1xMgxO/Cu(In0.7Ga0.3)Se2 interface was investigated for its
potential application in Cd-free Cu(In,Ga)Se2 thin lm solar cells. Zn1xMgxO thin lms with various
Mg contents were grown by atomic layer deposition on Cu(In0.7Ga0.3)Se2 absorbers, which were
deposited by the co-evaporation of Cu, In, Ga, and Se elemental sources. The electron emissions from
the valence band and core levels were measured by a depth prole technique using X-ray and
ultraviolet photoelectron spectroscopy. The valence band maximum positions are around 3.17 eV for
both Zn0.9Mg0.1O and Zn0.8Mg0.2O lms, while the valence band maximum value for CIGS is 0.48 eV.
As a result, the valence band oset value between the bulk Zn1xMgxO (x = 0.1 and x = 0.2) region
and the bulk CIGS region was 2.69 eV. The valence band oset value at the Zn1xMgxO/CIGS
interface was found to be 2.55 eV after considering a small band bending in the interface region. The
bandgap energy of Zn1xMgxO lms increased from 3.25 to 3.76 eV as the Mg content increased from
0% to 25%. The combination of the valence band oset values and the bandgap energy of
Zn1xMgxO lms results in the at (0 eV) and cli (0.23 eV) conduction band alignments at the
Zn0.8Mg0.2O/Cu(In0.7Ga0.3)Se2 and Zn0.9Mg0.1O/Cu(In0.7Ga0.3)Se2 interfaces, respectively. The
experimental results suggest that the bandgap energy of Zn1xMgxO lms is the main factor that
determines the conduction band oset at the Zn1xMgxO/Cu(In0.7Ga0.3)Se2 interface. Based on these
results, we conclude that a Zn1xMgxO lm with a relatively high bandgap energy is necessary to create
a suitable conduction band oset at the Zn1xMgxO/CIGS interface to obtain a robust heterojunction.
Also, ALD Zn1xMgxO lms can be considered as a promising alternative buer material to replace
the toxic CdS for environmental safety.

Introduction
Cu(In,Ga)Se2 (CIGS) thin lm solar cells are expected to play a
leading role in the photovoltaic market due to their high
eciency of around 20%.1 The conventional structure of a CIGS
solar cell is ZnO/CdS/CIGS, where CdS deposited by chemical
bath deposition (CBD) has been regarded as the best buer
material for high eciency thus far. However, Cd is a toxic metal
element, and the CBD process produces much chemical waste.
Thus, for the ecologically sustainable development of the energy
system, a Cd-free material which can be deposited by a non-CBD
process should be developed for environmental safety.
Over the last decade, the Zn1xMgxO lm has attracted an
increasing interest as an alternative Cd-free buer material for
a

Department of Materials Science and Engineering, Korea Advanced

Institute of Science and Technology, 291 Daehak-ro, Yuseong-gu,
Daejeon, 305-701, Republic of Korea. E-mail: btahn@kaist.ac.kr
b
Department of Electrical Engineering, King Saud University, 11451,
Riyadh, 2454, Kingdom of Saudi Arabia
w Electronic supplementary information (ESI) available. See DOI:
10.1039/c2cp40355b

This journal is

the Owner Societies 2012

CIGS solar cells.25 The Zn1xMgxO lm has several advantages,

including its variable bandgap value. Indeed, the bandgap can be
varied in the range from 3.2 to 3.9 eV by controlling the chemical
composition of the ternary compound.69 The electronic band
structure at the Zn1xMgxO/CIGS interface can be optimized by
using the Zn1xMgxO buer with various bandgap energies to
enhance the eciency of CIGS solar cells.
To date, several methods have been developed for Zn1xMgxO
lm deposition, including RF magnetron sputtering,9 pulsed laser
deposition,6 ultrasonic spray pyrolysis,7 metal organic vapor phase
epitaxy,10,11 and atomic layer deposition (ALD).2 Among them,
the most common deposition method is RF magnetron sputtering.
However, highly energetic particles in plasma cause serious damage
to the surface of CIGS absorbers which aects the quality of the
Zn1xMgxO/CIGS interface, and in turn the device performance.
Recently, to avoid this problem, Zn1xMgxO lms have been
deposited by ALD. The associated processing parameters have
been well established.2 Because ALD is characterized by the
alternate exposure of chemical species with self-limiting surface
reactions, ALD has several advantages over other deposition
methods. These advantages include accurate composition control,
Phys. Chem. Chem. Phys., 2012, 14, 47894795

4789

Published on 01 March 2012. Downloaded by Chungnam National University on 07/11/2014 08:14:00.

View Article Online

precise thickness control, excellent lm conformality, and

uniformity over a large area. Furthermore, the ALD method
is more favorable for integration into in-line fabrication
processes compared to the wet CBD process.
Thus far, investigations of CIGS solar cells fabricated with a
Zn1xMgxO buer layer are rare. Recently, excellent results
were attained for CIGS solar cells utilizing a Zn1xMgxO buer
layer deposited by ALD.2,4,5 The eect of the chemical composition
of the Zn1xMgxO buer on the photovoltaic properties of CIGS
solar cells was also reported. The highest eciency of 16.2% was
demonstrated in a device fabricated with the Zn1xMgxO buer
layer with the bandgap energy of 3.6 eV.4 However, the conversion
eciency of CIGS solar cells using this buer material was not
as high as that of CIGS solar cells using the CdS buer layer
deposited by a wet chemical process.
One possible reason for the limitation of the cell eciency is
related to the unfavorable electronic band structure at the
Zn1xMgxO/CIGS interface. The energy barrier at the interface
strongly aects the recombination at the interface as well as the
charge transport across the junction. An appropriate conduction
band discontinuity at the Zn1xMgxO/CIGS interface is of
primary importance for the achievement of high eciency devices.
The performance of CIGS solar cells based on Zn1xMgxO
buers can be improved by adjusting the interface structure
parameters. For this purpose, the precise knowledge of the energy
band alignments at the Zn1xMgxO/Cu(In,Ga)Se2 interface
is required.
To the best of our knowledge, only a few studies have
attempted investigations of the electronic structure at the
Zn1xMgxO/Cu(In0.7Ga0.3)Se2 interface.3 X-Ray photoelectron
spectroscopy (XPS) emission measurements on a sputterdeposited Zn1xMgxO/CIGS interface (x = 0, 0.03, 0.06, and
0.17) were carried out. However, the experimental conditions,
calibration, and accuracy of the experimental measurements of
the valence band maximum (VBM) positions were not given in
those studies. The presented valence band spectra raise several
issues, such as the existence of signicant electron emissions
below the Fermi level and the location of the VBM of CIGS
absorber (it matched to the Fermi level). Also, the authors
did not provide any information about the composition of
the CIGS absorber. Therefore, the estimated values of the
conduction band oset (CBO) at the Zn1xMgxO/CIGS interfaces are questionable. Thus, the published experimental data
regarding the band alignment at the interface are controversial.
Therefore, it is necessary to clarify the energy band alignment at
the Zn1xMgxO/Cu(In,Ga)Se2 interface to enhance the conversion eciency in Cd-free thin lm CIGS solar cells.
In this study, we report on the investigation of band
discontinuities at the Zn1xMgxO/Cu(In0.7,Ga0.3)Se2 interfaces
using XPS and ultraviolet photoelectron spectroscopy (UPS)
depth prole analysis.

Experimental
Zn1xMgxO/CIGS heterointerfaces were fabricated by the ALD
deposition of Zn1xMgxO layers on CIGS absorbers at 120 1C
using a vertical hot-wall type ALD reactor. For the Zn1xMgxO
lm growth, diethylzinc [Zn(C2H5)2, DEZ] and bis-ethylcyclopentadienyl magnesium [Mg(C5H4C2H5)2, Mg(CpEt)2] were used
4790

Phys. Chem. Chem. Phys., 2012, 14, 47894795

as metal precursors while water vapor was used as a reactant.

High purity Ar gas was used as a purging and carrier gas, and
its ow was controlled by a mass ow controller (100 sccm for
the precursors and reactant and 400 sccm for purging).
Because the vapor pressure levels of both DEZ and H2O are
high, the bubblers were maintained at 10 1C using cooling
jackets. However, the temperature of the Mg(CpEt)2 bubbler
was maintained at 60 1C to increase the vapor pressure. A
typical growth sequence for a ZnO monolayer is composed of
DEZ exposure of 0.1 s, Ar purging of 6 s, H2O exposure of
0.1 s, and Ar purging of 6 s. For a MgO monolayer, it is
composed of Mg(CpEt)2 exposure of 2 s, Ar purging of 6 s,
H2O exposure of 0.1 s, and Ar purging of 6 s. For Zn1xMgxO
lms, a cycle of the MgO ALD sequence was introduced once
every 4 and 9 cycles of the ZnO ALD sequence. The total
number of ALD cycles, which consisted of ZnO and MgO
ALD cycles, was xed around 300 for all of the samples, and
the thickness of the Zn1xMgxO lms was kept between 40 and
45 nm. Diethylzinc (99.9999%) and bis-ethylcyclopentadienyl
magnesium (99.9999%) were, respectively, purchased from
Techno-Semichem and DNF (Republic of Korea).
The CIGS absorbers were deposited by three stage co-evaporation of Cu, In, Ga, and Se elemental sources on Mo-coated soda
lime glass substrates. The composition was adjusted to
Cu(In0.7,Ga0.3)Se2. CIGS absorbers with this composition are
commonly used for the fabrication of high eciency solar cells.
The detailed procedures of the deposition process are available
in our article published earlier.12
After deposition, the morphology and thickness of the
CIGS and Zn1xMgxO lms were investigated by scanning
electron microscopy (SEM) using a FEI Nova230. To estimate
the bandgap energy of the Zn1xMgxO lms, the extinction
coecient (k) was measured by a spectroscopic ellipsometer
(Woollam M-2000D) in the wavelength range of 193 to 1000 nm.
The compositions of the CIGS absorbers were determined by
energy dispersive spectrometry. The Mg content in the Zn1xMgxO
lms was measured by an inductively coupled plasma-atomic
emission spectrometer (ICP-AES) using a Thermo ICAP-6500.
For the ICP-AES analysis, Zn1xMgxO lms were dissolved
into a mixture of 18% hydrochloric acid (20 mL) and 20%
nitric acid (10 mL). Then, the atomic emission peaks were
recorded. To measure the exact atomic concentrations of
Zn and Mg in the lms, ICP-MS reference solutions (1000 ppm
of each element in 3% nitric acid) were used for precise
quantications (SCP Science, Canada).
The XPS and UPS depth proles were taken using a Sigma
Probe, a Thermo Fisher VG Scientic spectrometer equipped
with monochromatic Al Ka X-ray source and a He discharge
lamp under a base pressure of about 1010 Torr. For the UPS
measurements, He I (21.2 eV) radiation was used. The samples
were held with Cu clips which were grounded to the sample
stage. In the XPS and UPS measurements, the binding energy
scales were calibrated using the binding energy positions of
the C 1s core level at 284.6 eV and the Au 4f7/2 core level at
83.98 eV, respectively. Sputter etching was carried out by the
EX05 argon ion gun. To prevent the samples from being
damaged and to provide gentle etching, the ion beam energy
and beam current were set at 1 kV and 1 mA, respectively.
To access the Zn1xMgxO/CIGS interface, a short etch period
This journal is

the Owner Societies 2012

View Article Online

Published on 01 March 2012. Downloaded by Chungnam National University on 07/11/2014 08:14:00.

of 50 s and 39 levels were set. The energy of the VBM positions

of the Zn1xMgxO and CIGS lms was determined at every etch
level from the emission spectrum through linear extrapolation of
the steep leading edge of the highest valence band peak to the
baseline. The core level binding energy positions were dened as
the center of the full width at half maximum (FWHM) of the
photoemission peak using the Avantage software package.

Results and discussions

1.

XPS, UPS depth prole analysis

The photoemission probes the surface region. The depth

prole is used to assess the electronic band structure of the
Zn1xMgxO/Cu(In0.7Ga0.3)Se2
interface.
Photoemission
spectra are acquired at dierent physical levels within the
Zn1xMgxO buer and CIGS absorber. To identify the interface and determine the valence band oset (VBO), the valence
band spectra and Zn 3d, Zn 2p, Mg 1s Ga 2p, Cu 2p, In 3d,
and In 4d electron emissions were monitored. The Zn 3d, Zn
2p, Mg 1s core levels identify the Zn1xMgxO buer layer,
while the Ga 2p, Cu 2p, In 3d, and In 4d core levels identify
the CIGS absorber. The spectral weight distribution of the
emission intensity in the XPS spectra changes with the etch
level and the Zn1xMgxO/CIGS interface can be identied by
the change of the electronic structure.
Fig. 1 shows the XPS depth proles of Zn0.9Mg0.1O/CIGS
and Zn0.8Mg0.2O/CIGS interfaces in the energy range of
026 eV. The data are presented as the photoemission intensity
versus the etch level. The binding energy scale is referred to the
Fermi level. As shown in Fig. 1, the XPS depth proles of the
Zn0.9Mg0.1O/CIGS and Zn0.8Mg0.2O/CIGS interfaces display
similar electronic band structures. The discussion provided below
relates to both the Zn0.9Mg0.1O/CIGS and Zn0.8Mg0.2O/CIGS
interfaces.
The XPS spectra which appear in the binding energy range
of 010 eV at each etch level are attributed to photoelectrons
from the valence bands. The electron emissions from the Zn
3d, Cu 3d, In 4d, and Ga 3d core levels were observed to be in
the range of 1022 eV. The peak centered at a binding energy
of 11 eV is ascribed to the Zn 3d subshell emission. The spin-orbital
splitting of Zn 3d3/23d5/2 is unresolved. A similar position for Zn
3d subshell emission was reported in earlier research.13 Note that
the Zn 3d core level emission originates from the Zn1xMgxO
buer layer and can be used for its identication. Given that the
photoionization cross sections are identical for the Cu 3d and Zn
3d core levels for Al Ka excitation and the binding energy dierence
in the Zn 3d and Cu 3d peak positions is too narrow, the Cu 3d
subshell emission is overlapped with the Zn 3d core level
emission.14 Indeed, Fig. 1 shows that the emission intensity
in the vicinity of the Zn 3d core level position, as expected, is
high at the beginning. It then drops at the interface and
remains stable as the etching continues.
For the spectra starting from etch level 15, the emission
peaks are located in the binding energy range of 18 to 20 eV. Given
that the Ga 3d core level has a photoionization cross-section that is
much smaller than those of the In 4d subshell for the energy
excitation of the Al Ka X-ray source, the emission centered at a
binding energy of 18.35 eV was ascribed to the In 4d subshell.14
This journal is

the Owner Societies 2012

Fig. 1 The XPS depth proles of the Zn0.9Mg0.1O/CIGS (a) and

Zn0.8Mg0.2O/CIGS (b) interfaces in the energy interval of 026 eV.
The data are presented as the elemental intensity versus the etch level.
The binding energy scale is referred to the Fermi energy level.

The lower intensity emission at a binding energy of 19.55 eV

can be ascribed to the Ga 3d core level emission. The positions of
In 4d and Ga 3d core level emission are in agreement with the
data reported elsewhere.1416 The dierence in the binding energy
between Ga 3d and In 4d core level peaks was estimated to be
about 1.20 eV, which is consistent with the literature.1416 The In
4d and Ga 3d core level emissions originate from the CIGS lm
and identify the surface of the CIGS absorber.
The change in the emission intensities of the In 4d, Ga 3d,
and Zn 3d core levels with the etch levels reects the evolution
of the Zn1xMgxO/CIGS interface. The decrease in the intensity
of the Zn 3d subshell emission and the increase in the intensity of
the In 4d and Ga 3d core level emissions with increasing etch time
was observed at both the Zn0.9Mg0.1O/CIGS and Zn0.8Mg0.2O/
CIGS interfaces. As shown in Fig. 1, the interface region is located
within the range of levels from L15 to L21. The thickness of the
interface region is estimated to be about 15 nm from the etching
rate. Unfortunately, a direct determination of VBO at the
Zn1xMgxO/CIGS interface is not possible. We cannot nd the
valence band spectrum, for which the electron emissions from
both Zn1xMgxO and CIGS layers are resolved, as the valence
band emission of the Zn1xMgxO layer is masked by the valence
band emission of the CIGS absorber.
Phys. Chem. Chem. Phys., 2012, 14, 47894795

4791

Published on 01 March 2012. Downloaded by Chungnam National University on 07/11/2014 08:14:00.

View Article Online

1.1 Valence band discontinuity between Zn1xMgxO (x = 0.1

and 0.2) and CIGS. Fig. 2 shows the valence band spectra taken
at nine etch levels from L1 to L9 in the Zn1xMgxO (x = 0.1 and
x = 0.2) region and at nine etch levels from L31 to L39 in the
CIGS region. We can clearly distinguish two layers of materials,
the Zn1xMgxO buer and the CIGS absorber, due to the
change in the spectral weight distribution of the emission
intensity. It should be noted that the spectra are unchanged
for the etch levels from L1 to L9. We designated this region as
the bulk Zn1xMgxO region. The spectra taken at etch levels
from L31 to L39, which are indicative of the bulk CIGS
region, are indistinguishable as well. The VBO can be estimated
from the spectra shown in Fig. 2 in a crude approximation
without taking band bending into account. The VBM values of
3.18 eV and 3.17 eV were deduced for the Zn0.9Mg0.1O (Fig. 2a)
and Zn0.8Mg0.2O (Fig. 2b) buers from the set of spectra from
L1 to L9, while the VBM value of around 0.48 eV was estimated

Fig. 2 The representative valence band spectra of (a) the bulk

Zn0.9Mg0.1O region and CIGS, (b) the bulk Zn0.8Mg0.2O region and
CIGS are shown on an expanded scale. The spectra are aligned on the
binding energy scale to the Fermi level. Valance band spectra taken
from the buer layer (at etch levels from L1 to L9) and the CIGS
absorber (at etch levels from L31 to L39). A dierence in the energy of
the valence-band edges is visible for the Zn1xMgxO (x = 0.1 and
x = 0.2) and CIGS layers.

4792

Phys. Chem. Chem. Phys., 2012, 14, 47894795

for the CIGS absorber from the set of spectra from L31 to
L39. Therefore, valence band discontinuities of 2.70 and
2.69 eV were deduced from the bulk Zn1xMgxO (x = 0.1
and x = 0.2) region and the bulk CIGS region, respectively,
when the band bending is ignored. It is found that the VBO
values for both samples are almost same within an experimental uncertainty of 0.1 eV (which is a commonly accepted
level for XPS and UPS). The results indicate that the change in
the chemical composition of Zn1xMgxO in the range of x
from 0.10 to 0.20 does not aect the electronic structure of the
valence band.
Note that the VBMs are determined with respect to EF.
Given that the composition of the CIGS absorber was adjusted
to Cu(In0.7,Ga0.3)Se2, which corresponds to a bulk bandgap
energy of 1.12 eV, the estimated VBM of 0.48 eV places the
EF in the CIGS at 0.08 eV below the middle of the bandgap of the
CIGS. Similar values of VBM in bulk CIGS with a conventional
composition were reported in other XPS (0.40 eV) and UPS
(0.47 eV) photoemission experiments.15,17
It can be seen from Fig. 2 that we did not nd any change in
the spectral weight distribution of the emission intensity in
both sets of spectra taken at nine etch levels from L1 to L9
(bulk Zn1xMgxO) and the set of spectra taken at nine etch
levels from L31 to L39 (bulk CIGS). Furthermore, the XPS
peak positions and FWHM values for Zn 2p, Cu2p, and In 3d
core levels from the bulk Zn1xMgxO region and bulk CIGS
region also were not changed with sputter etching as shown in
Fig S1 (ESIw). These results indicate that the valence band
structures are not aected by Ar+ ion sputter damage. In
Fig. 2, a VBO value of approximately 2.69 eV was determined
between the bulk Zn1xMgxO (x = 0.1 and x = 0.2) region
and the bulk CIGS region due to the nearly identical VBM
position of the Zn0.9Mg0.1O and Zn0.8Mg0.2O lms. These
results can be supported by a UPS analysis as well. Fig. 3
shows the UPS spectra taken from L6 to L10 for both
Zn0.9Mg0.1O and Zn0.8Mg0.2O lms, and the spectra from
the CIGS absorber. As shown in Fig. 3, the VBM position
of CIGS is 0.57 eV; moreover, the same VBM position of
3.17 eV was observed for both the Zn0.9Mg0.1O and Zn0.8Mg0.2O
lms. The resulting VBO value of 2.60 eV, which is comparable

Fig. 3 The UPS spectra taken from L6 to L10 for both the
Zn0.9Mg0.1O and Zn0.8Mg0.2O lms, and taken from L6 to L10 for
the CIGS absorber. Both Zn1xMgxO lms show nearly the same
VBM position of 3.17 eV.

This journal is

the Owner Societies 2012

View Article Online

Published on 01 March 2012. Downloaded by Chungnam National University on 07/11/2014 08:14:00.

to the values from the XPS analysis, was obtained. A small

dierence in the values determined from XPS and UPS VBO
measurements was also reported by Nakada et al. and Schmid
et al.15,16 Based on this result, we can deduce that the VBM of
Zn1xMgxO is nearly constant at 3.17 eV regardless of the
variation in the Mg content.
1.2 Band bending induced by the Zn1xMgxO (x = 0.1 and
0.2)/CIGS interface formation. To estimate the band bending
in the Zn1xMgxO buer and the CIGS absorber, we examined
the evolution of the VBMs and Zn 3d, Zn 2p, Mg 1s, In 3d, and
Cu 2p core level peak positions with etching time at the
Zn1xMgxO/CIGS interfaces. It is proven that cation core level
peak positions are primarily aected by band bending.18
Fig. 4 shows the evolution of the VBM position with the
Ar+ ion beam etching levels near and at the Zn0.8Mg0.2O/
CIGS interface. No shifts of the VBM position in the
Zn0.8Mg0.2O lm were observed. The Zn0.9Mg0.1O layer also
showed a similar trend near and at the Zn0.9Mg0.1O/CIGS
interface. Moreover, the positions of the cation core levels and
the FWHM values of the Zn 2p, Zn 3d, and Mg 1s subshell
emissions were also unchanged (data not shown). These
experimental results suggest that the band bending values in
the Zn0.8Mg0.2O and the Zn0.9Mg0.1O are too small to be
recorded within an acceptable level of experimental uncertainty.
It should be noted that the lack of change in the peak positions
and in their FWHM values indicates a stable chemical composition
of the Zn1xMgxO buer in the bulk and at the Zn1xMgxO/
CIGS interface. This nding is also conrmed by the atomic
percent XPS prole, which shows a uniform composition of the
Zn1xMgxO lm in bulk and monotonic parallel drops in the
values of the Zn, Mg, and O atomic percent across the interface.
These results are additional evidence that Ar+ ion etching did not
induce any change in the chemical composition or cause
sample damage under our experimental conditions.18
Unlike in the Zn1xMgxO lms, a small energy shift of the
VBM position was detected in the CIGS layer. The VBM
position moves to a smaller binding energy, from 0.62 to 0.48 eV,
upon an increase in the etch level at the interface region near
the CIGS. Note that the region of the etch levels from L15
to L16 is determined as the Zn0.8Mg0.2O/CIGS interface.

Fig. 4 Evolution of the VBM energy level with the Ar+ ion beam
etch level near and at the Zn0.8Mg0.2O/CIGS interface.

This journal is

the Owner Societies 2012

Fig. 5 Binding energy shift of the VBM and In 3d and Cu 2p core

level peak positions with the Ar+ ion beam etch level near and at the
Zn0.8Mg0.2O/CIGS interface.

As shown in Fig. 4, the VBO value of 2.55 eV can be deduced

as the dierence in the binding energy between the valence
band edges in the spectra taken at etch level 14 and etch level 17.
A similar valence band discontinuity of 2.56 eV was also
estimated at the Zn0.9Mg0.1O/CIGS interface from the depth
prole of the valence band. The transitivity rule for a band
discontinuity suggests a nearly at valence band alignment at the
Zn0.9Mg0.1O/Zn0.8Mg0.2O interface.
Fig. 5 shows the energy shifts of the In 3d, Cu 2p core level
emissions and the VBM position. The shifts of the In 3d core
level as well as the Cu 2p core level emission are parallel to that
of the VBM energy in the interface region near the CIGS.
Given that the values of the shifts are quite similar and
considering that the FWHM values of the In 3d and Cu 2p
core level emissions remained unchanged during the etching
process, we can conclude that the shifts in the interface region
near the CIGS were caused by band bending. Note that the
downward band bending in the CIGS at the Zn0.8Mg0.2O/
CIGS interface is consistent with typical band bending
for p-type semiconductors. A band bending value of around
0.14 eV was detected in the CIGS absorber and this value is in
agreement with those in the literature.17,19 The pinning position
of the Fermi energy level at the interface is derived as 0.62 eV
higher than the top of the valence band of CIGS at the
interface.
As shown in Fig. 2, the valence band oset at both the
Zn0.9Mg0.1O/CIGS and Zn0.8Mg0.2O/CIGS interfaces is about
2.69 eV when band bending is ignored. At this point, we can
correct this value of VBO using the energy shifts of the cation
emissions from the Cu 2p, In 3d, Zn 3d, Zn 2p, and Mg 1s
subshells at the Zn1xMgxO/CIGS interfaces. The VBO values
determined at the Zn0.8Mg0.2O/CIGS interface are listed in
Table 1. This table summarizes several combinations of core
level emissions from the CIGS absorber and the Zn0.8Mg0.2O
buer. The average VBO value of 2.60 eV was found at the
Zn0.8Mg0.2O/CIGS interface.
Therefore, the VBO value of 2.55 eV determined from the
depth prole in Fig. 4 is quite similar to the valence band
discontinuity estimated by the commonly accepted method,
which is based on the shifts in the cation core level emissions at
Phys. Chem. Chem. Phys., 2012, 14, 47894795

4793

View Article Online

Published on 01 March 2012. Downloaded by Chungnam National University on 07/11/2014 08:14:00.

Table 1 VBO values derived based on the shifts in the Cu 2p, In 3d,
Cu2p
Zn 3d, Zn 2p, and Mg 1s subshell emissions. EIn3d
VBM and EVBM are the
binding energies of the In 3d and Cu 2p core levels with respect to
Mg1s
Zn3d
the VBM energy of the CIGS absorber. EZn2p
VBM, EVBM and EVBM are the
binding energies of the Zn 2p, Mg 1s, and Zn 3d core levels with
respect to the VBM energy of the Zn0.8Mg0.2O lm. DECL denotes the
binding energy dierence between the core levels at the interface
CIGS
(L39, eV)
ECu2p
VBM

EIn3d
VBM

932.21

444.52

Zn0.8Mg0.2O
(L10, eV)

DECL
(L17, eV)

EZn2p
VBM
EMg1s
VBM
EZn3d
VBM
EZn2p
VBM
EMg1s
VBM
EZn3d
VBM

EZn2p
Cu2p
EMg1s
Cu2p
EZn3d
Cu2p
EZn2p
In3d
EMg1s
In3d
EZn3d
In3d

1018.84
1300.48
7.58
1018.84
1300.48
7.58

VBO/
eV
89.22
370.91
922.08
576.92
858.61
434.38

2.59
2.64
2.55
2.60
2.65
2.56

the interface (Table 1). Given that the two VBO values derived
from two dierent methods show a high degree of similarity,
we can conclude that a cli alignment of 2.55 eV is a good
approximation of the valence band discontinuity at the
Zn1xMgxO/CIGS interface.
2. Optical analysis of Zn1xMgxO (x = 0.1, 0.14, 0.17, 0.2,
and 0.25) lms
To estimate the optical bandgap energies of Zn1xMgxO lms,
the extinction coecients were obtained by spectroscopic
ellipsometry. The optical absorption coecients were derived
as functions of the incident photon energy from the extinction
coecients for 45 nm-thick Zn1xMgxO lms.
Given that the Zn1xMgxO compound is a direct semiconductor, the spectral response of the absorption coecient
with the photon energy near the absorption edge for direct
allowed transitions can be expressed as follows:
ahn

B
hn  Eg 1=2 ;
hn

where B is the absorption constant for a direct transition, h the

Plancks constant, n the frequency, and Eg the optical bandgap
energy.20 The plot of (ahn)2 versus the incident photon energy

Fig. 6 Analysis of the optical absorption edge of the Zn1xMgxO

lms in terms of the direct transition. The plots of (ahn)2 versus the
incident photon energy (hn) are depicted with solid lines.

4794

Phys. Chem. Chem. Phys., 2012, 14, 47894795

(hn) is tted using relation (1). The results of the tting are shown
in Fig. 6. The Eg value for the ZnO lm was determined to be
3.25 eV, which is in good agreement with the previously reported
value.21,22 The Eg value increased from 3.44 to 3.76 eV as the Mg
content increased from 10% to 25%. A widening of Eg with
increasing Mg content was also reported in earlier research.7,11
3. Estimation of the conduction band oset at Zn1xMgxO/
CIGS interfaces
The conduction band discontinuity at the Zn1xMgxO/CIGS
interface was calculated as follows: CBO = Eg (Zn1xMgxO) 
Eg (CIGS)  VBO. As mentioned in the experimental section, the
Ga/(Ga + In) atomic ratio in the CIGS absorbers was adjusted to
0.3. This composition results in a CIGS bandgap value of 1.12 eV.
The experimentally determined bandgap energies of Zn1xMgxO
lms from 3.44 to 3.76 eV were used for the CBO estimations.
The scheme of the energy band alignment at the
Zn0.8Mg0.2O/CIGS interfaces is presented in Fig. 7. A cli
CBO of 0.14 eV was deduced between Zn0.8Mg0.2O and
CIGS without considering the band bending in the interface
region. However, a at conduction band alignment was estimated at the Zn0.8Mg0.2O/CIGS interface due to the small
band bending which was induced by interface formation.
Zn1xMgxO (x = 0.1 and x = 0.2) lms (Fig. 2 and 3) showed
almost same VBM positions. The at VBO between ZnO and
Zn0.85Mg0.15O was estimated based on the transitivity rule by Rao
et al.23 Given these data, our experimental results, and the common
anion rule, we assumed that the Mg content does not aect the
valence band structure of the Zn1xMgxO lms. We calculated the
CBO values between the bulk Zn1xMgxO and CIGS region, and
conduction band discontinuities at the Zn1xMgxO/CIGS interfaces using determined bandgap energies of the Zn1xMgxO lms
(Fig. 6). Fig. 8 shows the calculated CBO values for the
Zn1xMgxO lms with various levels of Mg. The conduction band
alignment between the Zn1xMgxO and CIGS could be changed
from a cli to a spike by increasing the bandgap energy of the
Zn1xMgxO lm. It has been known that the CBO value should be
adjusted between 0 and 0.3 eV to ensure high eciency. Thus, the
Mg content in the Zn1xMgxO lms should be carefully
controlled between 20% and 30% to obtain a proper CBO value.

Fig. 7 Energy band diagram at the Zn0.8Mg0.2O/CIGS interface.

This journal is

the Owner Societies 2012

View Article Online

eciency. Our results suggest that ALD Zn1xMgxO lms

are promising candidates that can replace the toxic CdS buer
layer and can contribute to environmental safety.

Published on 01 March 2012. Downloaded by Chungnam National University on 07/11/2014 08:14:00.

Acknowledgements
This work was supported by the Engineering Research Center
Program (2011-0001000) and the Priority Research Center
Programs (2009-0094040 and 2010-0029714) through the National
Research Foundation of Korea (NRF) funded by the Korean
Ministry of Education, Science and Technology (MEST).

References
Fig. 8 The calculated CBO values using Zn1xMgxO lms with
various Mg content.

Conclusion
We investigated the energy band alignment at the Zn1xMgxO
(x = 0.1 and x = 0.2)/Cu(In0.7Ga0.3)Se2 interface using XPS
and UPS depth prole measurements. From the XPS measurement, the VBM positions of the bulk Zn1xMgxO (x = 0.1 and
0.2) region and the bulk CIGS region were determined at 3.18,
3.17, and 0.48 eV, respectively. Thus, the VBO value of approximately 2.69 eV was estimated between the bulk Zn1xMgxO region
and the bulk CIGS region without considering the band bending.
These results were also conrmed by UPS analysis, which indicated
the nearly same VBM position of 3.17 eV for both the Zn0.9Mg0.1O
and Zn0.8Mg0.2O lms. The same valence band discontinuity of
approximately 2.55 eV was directly determined at the Zn0.9Mg0.1O/
CIGS and Zn0.8Mg0.2O/CIGS interfaces after considering the
small band bending of 0.14 eV in the interface region. This
result suggests a nearly at valence band alignment at the
Zn0.9Mg0.1O/Zn0.8Mg0.2O interface according to the transitivity
rule for a band discontinuity.
Given that valence band discontinuity does not change with
the Mg content within an acceptable level of experimental
uncertainty, the dierence in the bandgap energy of the
Zn1xMgxO lms only contributes to the CBO at the
Zn1xMgxO/CIGS interfaces. Thus, the conduction band
alignment at the Zn1xMgxO/CIGS interfaces could be changed
from a cli to a spike according to the bandgap energy of the
Zn1xMgxO lm. In a CIGS solar cell with bandgap energy of its
buer layer of less than 3.67 eV, the cli CBO can lower the open
circuit voltage by enhancing the interface recombination. Therefore, to create a suitable conduction band alignment at the
Zn1xMgxO/CIGS interface, the Mg content in the Zn1xMgxO
lms should be carefully controlled.
This study is the rst to quantify energy band alignments at
ALD Zn1xMgxO (0.1 r x r 0.25)/CIGS interfaces. The
results can be used to adjust the electronic structure at the
Zn1xMgxO/CIGS interface so as to improve the device

This journal is

the Owner Societies 2012

1 I. Repins, M. A. Contreras, B. Egaas, C. DeHart, J. Scharf,

C. L. Perkins, B. To and R. Nou, Prog. Photovoltaics: Res. Appl.,
2008, 16, 235.
2 T. Torndahl, C. Platzer-Bjorkman, J. Kessler and M. Edo, Prog.
Photovoltaics: Res. Appl., 2007, 15, 225.
3 T. Minemoto, Y. Hashimoto, T. Satoh, T. Negami, H. Takakura
and Y. Hamakawa, J. Appl. Phys., 2001, 89, 8327.
4 T. Torndahl, E. Coronel, A. Hultqvist, C. Platzer-Bjorkman, K. Leifer
and M. Edo, Prog. Photovoltaics: Res. Appl., 2009, 17, 115.
5 C. Platzer-Bjorkman, T. Torndahl, A. Hultqvist, J. Kessler and
M. Edo, Thin Solid Films, 2007, 515, 6024.
6 T. Makino, Y. Segawa, M. Kawasaki, A. Ohtomo, R. Shiroki, K.
Tamura, T. Yasuda and H. Koinuma, Appl. Phys. Lett., 2001, 78, 1237.
7 X. Zhang, X. M. Li, T. L. Chen, J. M. Bian and C. Y. Zhang, Thin
Solid Films, 2005, 492, 248.
8 A. Ohtomo, M. Kawasaki, T. Koida, K. Masubuchi, H. Koinuma,
Y. Sakurai, Y. Yoshida, T. Yasuda and Y. Segawa, Appl. Phys.
Lett., 1998, 72, 2466.
9 T. Minemoto, T. Negami, S. Nishiwaki, H. Takakura and
Y. Hamakawa, Thin Solid Films, 2000, 372, 173.
10 T. Gruber, C. Kirchner, R. Kling, F. Reuss and A. Waag,
Appl. Phys. Lett., 2004, 84, 5359.
11 F. Y. Meng, Y. Chiba, A. Yamada and M. Konagai, Sol. Energy
Mater. Sol. Cells, 2007, 91, 1887.
12 K. H. Kim, K. H. Yoon, J. H. Yun and B. T. Ahn, Electrochem.
Solid-State Lett., 2006, 9, A382.
13 M. Bar, J. Reichardt, A. Grimm, I. Kotschau, I. Lauermann,
K. Rahne, S. Sokoll, M. C. Lux-Steiner, C. H. Fischer,
L. Weinhardt, E. Umbach, C. Heske, C. Jung, T. P. Niesen and
S. Visbeck, J. Appl. Phys., 2005, 98, 053702.
14 L. J. Brillson, Surface and Interface of Electronic Materials, Wiley-VCH
Verlag GmbH & Co. KGaA, Weinheim, 2010.
15 T. Nakada, M. Hongo and E. Hayashi, Thin Solid Films, 2003,
431432, 242.
16 D. Schmid, M. Ruckh, F. Grunwald and H. Schock, J. Appl. Phys.,
1993, 73, 2902.
17 L. Larina, D. Shin, J. H. Kim and B. T. Ahn, Energy Environ. Sci.,
2011, 4, 3487.
18 A. D. Katnani and G. Margaritondo, Phys. Rev. B: Condens.
Matter, 1983, 28, 1944.
19 M. Morkel, L. Weinhardt, B. Lohmuller, C. Heske, E. Umbach,
W. Riedl, S. Zweigart and F. Karg, Appl. Phys. Lett., 2001, 79, 4482.
20 J. I. Pankove, Optical Process inSemiconductors, Dover Publication,
Inc., New York, 1987.
21 U. Ozgur, Y. I. Alivov, C. Liu, A. Teke, M. A. Reshchikov,
S. Dogan, V. Avrutin, S. J. Cho and H. Morkoc, J. Appl. Phys.,
2005, 98, 1.
22 S. J. Pearton, D. P. Norton, K. Ip, Y. W. Heo and T. Steiner,
J. Vac. Sci. Technol., B, 2004, 22, 932.
23 G. V. Rao, F. Sauberlich and A. Klein, Appl. Phys. Lett., 2005,
87, 032101.

Phys. Chem. Chem. Phys., 2012, 14, 47894795

4795