Solar Energy Materials & Solar Cells 81 (2004) 249–259

Preparation and characterization of

Cu(In,Ga)(Se,S)

thin ﬁlms from electrodeposited

precursors for hydrogen production

Jennifer E. Leisch

a,b

, Raghu N. Bhattacharya

*, Glenn Teeter

John A. Turner

National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401-3393, USA

Department of Chemistry, Colorado School of Mines, 1500 Illinois St., Golden, CO 80401, USA

Received 28 March 2003; received in revised form 16 September 2003

Abstract

Semiconducting Cu(In,Ga)(Se,S)

thin ﬁlms were made from electrodeposited Cu(In,Ga)Se

precursors, followed by physical vapor deposition of In

, Ga, and Se. The bandgaps of these

materials were found to be between 1.6 and 2.0 eV, which spans the optimal bandgap
necessary for application for the top junction in photovoltaic multijunction devices and for
unassisted water photolysis. These ﬁlms were characterized by electron-probe microanalysis,
scanning Auger spectroscopy, X-ray diffraction, and photocurrent spectroscopy.
r

Keywords: Electrodiposition; CIGS; Hydrogen

1. Introduction

With a bandgap of around 1.05 eV, CuInSe

(CIS) ﬁlms have been studied

thoroughly for application to solar energy conversion devices. It has been shown that
incorporation of gallium into CIS to form Cu(In,Ga)Se

(CIGS) can increase the

bandgap up to 1.6 eV

[1]

. Materials with bandgaps in this range are of potential

interest to both solar cell researchers and those performing photoelectrochemical
water splitting. In the solar cell area, materials with this bandgap range can be used
in multijunction cells. For photoelectrochemical hydrogen production, effective

ARTICLE IN PRESS

*Corresponding author. Tel.: +1-303-384-6477; fax: +1-303-384-6432.

E-mail address:

raghu bhattacharya@nrel.gov (R.N. Bhattacharya).

0927-0248/$ - see front matter r 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.solmat.2003.11.006

voltages as low as 1.4 V

[2]

could be sufﬁcient to drive the water-splitting reaction.

This requires a semiconductor with a minimum bandgap of about 1.7 eV (assuming a
0.3-eV entropy loss for charge separation

[3]

). Additional criteria for photoelec-

trochemical devices include: the band edges of the semiconductor must overlap the
hydrogen and oxygen redox potentials, and the charge transfer across the
semiconductor/liquid interface must be fast enough to prevent band-edge migration

[4]

. The semiconductor surface must also be stable against corrosion both in the dark

and under illumination.

Thin-ﬁlm CIGS devices are well known for their high conversion efﬁciency even

when prepared as polycrystalline thin ﬁlms. Electrodeposition (ED) is a potentially
scalable technique for large-area thin-ﬁlm fabrication at reduced costs. In fact, CIGS
ﬁlms have been successfully prepared from electrodeposited precursors to yield solar
energy conversion devices with a 15.4% efﬁciency

[5]

. Sulfur incorporation has also

been used in CIGS materials to increase device performance and provide a graded
bandgap structure

[6]

. The purpose of this study is to incorporate sulfur throughout

the entire ﬁlm to obtain a higher-bandgap material suitable for water splitting. The
combination of lower system manufacturing cost (from ED) and higher efﬁciency
represents an important area of research for hydrogen production systems. In
addition, indium in the ﬁlm may provide some protection against corrosion. In
aqueous solution, indium can form a conducting oxide layer that stabilizes the
interface and may protect the underlying material

[7]

2. Experimental details

The ﬁlms in this research were prepared by ED of precursor ﬁlms, followed by

enrichment via physical vapor deposition. Details of the ED process have been
previously reported

[5]

. In short, the deposition was done in a three-electrode system,

using a Mo-coated glass working electrode. The deposition baths were made with
varying concentrations of CuCl

, GaCl

, InCl

, H

SeO

, and LiCl in a pH 2 buffer.

A constant applied potential was used to deposit the ﬁlms. These precursor ﬁlms
were then put into a vacuum chamber and processed at 600

C with an In

–Ga

mixture in a Se atmosphere for 1 h.

Electrodes were prepared from about 1-cm

sections of the ﬁnal materials. Front

contacts to the Mo were made using silver paint and copper wire. The edges were
then insulated with non-conducting epoxy, and the electrodes were baked in an oven
at 80

C for at least 1 h to fully cure the epoxy.

Bandgaps were determined using photocurrent spectroscopy, as previously

documented

[8]

. A 100-W tungsten bulb was used as the light source, and a

Princeton Technology 101 monochromator was used to step through wavelengths
from 500 to 1000 nm. Lamp spectra for normalization were obtained using a
thermopile with a low-noise ampliﬁer and a 495-nm ﬁlter, which eliminated second-
order interference in the data collection. The light was chopped at 41 Hz. A three-
electrode set-up with a Pt counter electrode and a saturated calomel electrode (SCE)
reference was used in an aqueous solution of 0.1 M Na

in pH 10 buffer.

ARTICLE IN PRESS

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

250

A Princeton 263A potentiostat and Stanford SR830 lock-in ampliﬁer interfaced with
a computer were used to register the photocurrent response as a function of
wavelength.

Precursor compositions were determined via inductively coupled plasma (ICP)

with a Varian-Liberty spectrometer. Electron-probe microanalysis (EPMA) was
done at both 10- and 20-keV accelerating voltages, using a JEOL WD/ED combined
microanalyzer. Auger electron spectroscopy (AES) survey scans and depth proﬁles
were done at 5.0 keV with a Physical Electronics PHI 670 Auger instrument. The
same instrument was used to obtain SEM images at a 20.0-keV accelerating voltage.
X-ray diffraction (XRD) spectra were obtained using a Scintag XGEN diffract-
ometer with a Cu Ka anode source.

3. Results and discussion

3.1. ICP and EPMA compositions

Four different precursors containing Cu, In, Ga, and Se were prepared in a

one-step ED. All had different compositions, with 0

oGa/(Ga+In)o1.

Table 1

shows the elemental concentrations in the precursor ﬁlm determined by ICP analysis.
All precursor ﬁlms were Cu-rich.

Table 2

shows the compositions of the ﬁnal ﬁlms after the enrichment step. EPMA

was used because of the lack of sulfur sensitivity in ICP analysis. EPMA analysis
showed the incorporation of sulfur into the ﬁlms and the enrichment of some of the
other elements during the PVD process. All ﬁlms were enriched in Ga and some S.
All ﬁlms became Cu-deﬁcient during subsequent processing.

3.2. Film morphology

Fig. 1

shows SEM images of the surfaces of two samples. These ﬁlms show similar

polycrystalline morphology. For example, Samples #2, #3, and #4 appear very
similar. Some ﬁlms show a darker background color in the ﬁlms, with lighter-colored
submicron-sized crystallites on the surface (

Fig. 1

). However, AES survey scans

showed that the composition of these two different areas is very similar. These
crystallites are most likely due to spitting during the PVD process. Sample #2

ARTICLE IN PRESS

Table 1
Precursor compositions of ED ﬁlms by ICP analysis

Sample

40.22

33.19

29.12

60.70

17.49

20.38

17.53

1.40

2.78

0.86

42.29

45.03

50.58

38.43

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

251

showed some bubbling in the ﬁlm, resulting in a few broken bubbles exposing the
molybdenum substrate through the ﬁlm.

SEM cross-section images were obtained, and that of Sample #2 is shown in

Fig. 2

. The thickness of these samples varied with processing. All initial, unprocessed

ARTICLE IN PRESS

Table 2
Atomic ratios obtained by EPMA (average of 10- and 20-kV spectra) and bandgaps measured by
photocurrent spectroscopy

Sample

21.32

22.22

8.01

5.30

2.25

2.72

1.41

0.14

29.51

23.43

37.75

42.37

36.92

39.72

39.09

37.26

10.00

11.90

13.74

14.93

1.61

1.76

1.93

2.0

Fig. 1. (a) SEM micrograph of Film 1, showing lighter-colored Ga-rich crystallites on the surface and
(b) SEM micrograph of Film 3.

Table 3
Atomic ratios in ﬁnal ﬁlms, normalized to copper

Sample

(In+Ga)

1.49

1.18

4.89

8.02

(Se+S)

2.20

2.32

6.59

9.84

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

252

ﬁlms were about 1 mm thick. After processing, Samples #1 and #2 became thicker
and gained material, whereas Samples #3 and #4 lost material.

The normalized compositions of the ﬁlms show that Cu(In,Ga)(Se,S)

is not the

only phase present in several of the ﬁlms. An excess of Ga and Se can be observed in
some samples, notably the two ﬁlms with the higher bandgaps—Samples #3 and #4
(

Table 3

AES depth proﬁles of Sample #1 in

Fig. 3

show two different types of Cu bonding

in the material. One form of copper is shown in the top part of the ﬁlm, whereas the

ARTICLE IN PRESS

Fig. 2. SEM cross-section of Sample #2, about 2.4 mm thick.

Fig. 3. Depth proﬁle of Sample #1, showing two different types of Cu bonding in the material.

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

253

other form is fairly constant throughout the bulk. This may be due to the formation
of an amorphous layer on the surface of the ﬁlm. The Ga content decreases into the
ﬁlm as the In content increases, indicating that they are substituting for each other in
the lattice. This ﬁlm does not have a constant crystal composition; instead, the
CIGSS composition varies, with some sulfur incorporation causing additional
shifting in XRD peaks (

Fig. 4

Sample #2 (

Fig. 5

) shows a relatively even distribution of components throughout

the ﬁlm, with the exception of higher S content at the surface, with a complementary

ARTICLE IN PRESS

Fig. 4. XRD of Sample #1.

Fig. 5. Depth proﬁle of Sample #2, showing S substituting for Se in the crystal.

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

254

dip in the Se peak, which signiﬁes that S was substituted for Se in the crystal
structure. Sulfur remains at a small, but constant, amount throughout the bulk of
this ﬁlm, and according to EPMA analysis, the atomic ratios are very close to that of
Cu(In,Ga)(Se,S)

. The proﬁle also shows the presence of oxides of indium at the

surface of the ﬁlm.

The XRD spectrum for this ﬁlm in

Fig. 6

shows several CIGSS peaks, some with

slight shifts, indicating crystals with different sulfur incorporations. CIGSS peaks
are shifted to higher 2y values than CIGS due to a decrease in the lattice constant of
the material. This shift is most often examined in the CIGS XRD peak that occurs
around 27

[9]

. There appears to be a very small amount of GaSe present in the ﬁlm,

as well.

In Sample #3, AES depth proﬁling shows two distinct chemical states of Se and

Ga in the ﬁlm (

Fig. 7

). There is only a small amount of copper at the surface, an

uneven S distribution, and a small amount of In at the surface, grading to virtually
none in the rest of the ﬁlm. Oxides of indium are present at the surface of the ﬁlm.

XRD analysis given in

Fig. 8

shows the presence of Ga(Se,S) in the material. GaSe

formation has been shown to occur in Cu(In,Ga)Se

ﬁlms that are Cu-deﬁcient

[10]

It also shows peaks corresponding to CGSS. These peaks are broad, due perhaps to
differences in the different crystal compositions, with very small amounts of In
incorporated into the structure.

The AES depth proﬁle of Sample #4 (

Fig. 9

) indicates that In is barely present in

this ﬁlm. Ga(S,Se) appears in the XRD spectrum of the ﬁlm (

Fig. 10

), as does

CuGa(Se,S)

. Only one bandgap is evident in the photoresponse measurements of

2.0 eV, most likely arising from the CGSS.

ARTICLE IN PRESS

Fig. 6. XRD spectrum of Sample #2, showing small amounts of GaSe present in the ﬁlm.

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

255

3.3. Photoresponse measurements

Initial tests revealed the samples to be p-type semiconductor materials.

Photoresponse measurements showed a range of bandgap values varying with
composition, from 1.6 to 2.0 eV (

Fig. 11

). The sample fabricated from the precursor

ARTICLE IN PRESS

Fig. 8. XRD of Sample #3, showing two types of crystals present in the ﬁlm.

Fig. 7. AES depth proﬁle of Sample #3, showing only the Se and Ga proﬁles throughout the ﬁlm.

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

256

with no Ga showed the smallest bandgap values. The ﬁlm made from the precursor
with Ga/(Ga+In)=1 (no In) showed the highest bandgap. This result is expected,
because CIS traditionally has a bandgap of 1.05 eV and CuGaSe

has a bandgap of

1.7 eV

[11]

. As expected, the incorporation of sulfur into these ﬁlms increased the

bandgaps. The bandgap is seen to increase with atomic percent sulfur in the ﬁlm.

ARTICLE IN PRESS

Fig. 9. AES depth proﬁle of Sample #4, showing little or no In present throughout the ﬁlm.

Fig. 10. XRD of Sample #4, showing phases free of In in the ﬁlm, thus resulting in a higher bandgap.

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

257

However, the exact relationship is convoluted by the variation of other ﬁlm
components.

Impedance measurements were performed on these materials to measure ﬂat-band

potentials in several pHs, as well as doping densities and resistances. For all samples
reported, the range of resistance in these materials is between 80 and 500 O. Doping
densities were calculated using the geometric area of the samples, assuming a
material dielectric of 10. These materials show a decrease in doping density with an
increase in AC frequency during Mott–Schottky analyses. At low AC frequency
(500 Hz) the average doping density (N

) was

B10

, while at higher frequency

(5000 Hz) the average N

value for this type of material was

B10

. At higher

frequency the N

value represents the bulk of the material, while the low frequency

measurements represent surface and near-surface regions. Our results indicate that
the synthesized material has more defects near the surface region, which could be due
to surface/solution interaction or the characteristic of this set of samples. The AES
depth proﬁles for this set of samples show a signiﬁcant variation of the ﬁlm
composition from the surface to the bulk region.

4. Conclusions

We prepared CIGSS ﬁlms from electrodeposited precursors. Precursor ﬁlms with

very little indium content resulted in CGSS after the PVD step. The prepared ﬁlms
consist of phases of CIGSS with varying composition, and some ﬁlms include other

ARTICLE IN PRESS

Fig. 11. Photoresponse measurements of ﬁlms with varying composition. Atomic percent sulfur in each
ﬁlm is shown.

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

258

secondary phases such as Ga(Se,S). All ﬁlms prepared in this manner show
polycrystalline morphology. Bandgaps up to 2.0 eV were obtained via this
preparation method, which is sufﬁcient for photoelectrochemical hydrogen produc-
tion. This research demonstrates the ability to synthesize materials with full range of
band gaps from a single material set.

At present, we are optimizing the processing conditions to obtain phase-pure

CIGSS ﬁlms. Future work will also include etching of the processed ﬁlms to improve
the surface characteristics.

Acknowledgements

This work was supported by the US Department of Energy Hydrogen, Fuel Cells,

and Infrastructure Technologies Program. Arturo Fernandezprovided XRD and
general help, and Bobby To provided EPMA analysis. J.L. would also like to thank
Miguel Contreras and Heli Wang for consultation regarding XRD analysis.

References

[1] M. Marudachalam, R.W. Birkmire, H. Hichiri, J.M. Schultz, A. Swartzlander, M.M. Al-Jassim,

Phases, morphology, and diffusion in CuIn

1 x

thin ﬁlms, J. Appl. Phys. 82 (2) (1997)

2896–2905.

[2] O. Khasalev, A. Bansal, J.A. Turner, High-efﬁciency integrated multijunction photovoltaic/

electrolysis systems for hydrogen production, Int. J. Hydrogen Energy 26 (2001) 127–132.

[3] J.R. Bolton, Solar photoproduction of hydrogen: a review, Sol. Energy 57 (1) (1996) 37–50.
[4] A. Bansal, J.A. Turner, Supression of band edge migration at the p-GaInP

O interface under

illumination via catalysis, J. Phys. Chem. B 104 (28) (2000) 6591–6598.

[5] R.N. Bhattacharya, J.F. Hiltner, W. Batchelor, M. Contreras, R.N. Nouﬁ, J.R. Sites, 15.4%

CuIn

1 x

-based photovoltaic cells from solution-based precursor ﬁlms, Thin Solid Films

361–362 (2000) 396–399.

[6] Y. Nagoya, K. Kushiya, M. Tachiyuki, O. Yamase, Role of incorporated sulfur into the surface of

Cu(InGa)Se

thin-ﬁlm absorber, Sol. Energy Mater. Sol. Cells 67 (2001) 247–253.

[7] C.A. Koval, R.L. Austermann, J.A. Turner, B.A. Parkinson, The effects of surface energetics and

surface oxide layers on the cyclic voltammetry of metallocenes at nonilluminated p-InP electrodes,
J. Electrochem. Soc. 132 (3) (1985) 613–623.

[8] J.D. Beach, In

1 x

N for photoelectrochemical water splitting, Ph.D. Thesis, Department of

Physics, Colorado School of Mines, 2001.

[9] D. Ohashi, T. Nakada, A. Kunioka, Improved CIGS thin-ﬁlm solar cells by surface sulfurization

using In

and sulfur vapor, Sol. Energy Mater. Sol. Cells 67 (2001) 261–265.

[10] M. Contreras, Internal Communication, 2002.
[11] S. Siebentritt, Wide gap chalcopyrites: materials properties and solar cells, Thin Solid Films 403–404

(2002) 1–8.

ARTICLE IN PRESS

J.E. Leisch et al. / Solar Energy Materials & Solar Cells 81 (2004) 249–259

259

Document Outline

Preparation and characterization of Cu(In,Ga)(Se,S)2 thin films from electrodeposited precursors for hydrogen production

M10

Document Outline