SHORT COMMUNICATION: ACCELERATED PUBLICATION
Research
199%-efficient ZnO/CdS/
CuInGaSe
2
Solar Cell with
812% Fill Factor
z
Ingrid Repins
1
*
,y
, Miguel A. Contreras
1
, Brian Egaas
1
, Clay DeHart
1
, John Scharf
1
, Craig L. Perkins
2
,
Bobby To
2
and Rommel Noufi
1
1
National Renewable Energy Lab, MS 3219, CO, USA
2
National Renewable Energy Lab, MS 3218, CO, USA
We report a new record total-area efficiency of 199% for CuInGaSe
2
-based thin-film solar cells. Improved
performance is due to higher fill factor. The device was made by three-stage co-evaporation with a modified
surface termination. Growth conditions, device analysis, and basic film characterization are presented.
Published in 2008 by John Wiley & Sons, Ltd.
key words: CIGS; thin film solar cells; record efficiency; fill factor; recombination; diode quality; saturation current;
surface
Received 20 November 2007; Revised 9 January 2008
INTRODUCTION
Record-efficiency devices are of interest for several
reasons. First, they provide a proof of concept for
developing products that require higher power per
area, lower cost per watt, or higher watts per kg.
Perhaps more importantly, understanding the sensi-
tivities and physical mechanisms that lead to improved
efficiency can help improve yield and efficiency for a
variety of deposition processes. This paper describes a
Cu(In,Ga)Se
2
(CIGS) solar cell with record 19 9%
total-area efficiency demonstrated at the National
Renewable Energy Laboratory (NREL).
FILM GROWTH
The device structure is as follows: soda-lime glass
(SLG) substrate, sputtered Mo back contact, three-
stage co-evaporated CIGS, chemical-bath-deposited
(CBD) CdS, sputtered resistive/conductive ZnO
bi-layer, e-beam-evaporated Ni/Al grids, MgF
2
anti-
reflective coating, and photolithographic device
isolation. These device layers have been described
in previous publications.
1–5
Figure 1 shows logged
CIGS deposition data from the 199% device, M2992.
The graph reflects slight optimizations to deposition
times and temperatures that have been made since
earlier publications. Deposition rates are shown in the
top portion of the graph. The thick lines represent
metal (Cu, In, Ga) rate setpoints, and the thin lines
show metals rate monitor data. Metals rates are read
from the left axis, and Se rate is read from the right
axis. The Se rate exhibits some unintentional
oscillations, due to mismatch between the control
parameters and the thermal mass of the newly filled
boat. Zero-rate background signals were removed
from the graph. The lower portion of Figure 1 shows
temperature data. The left axis of the graph presents
a full-scale view of the lamp and substrate tempera-
tures. The right axis shows an expanded view of the
substrate temperature, with the slight temperature dip
PROGRESS IN PHOTOVOLTAICS: RESEARCH AND APPLICATIONS
Prog. Photovolt: Res. Appl. 2008; 16:235–239
Published online 14 February 2008 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/pip.822
* Correspondence to: Ingrid Repins, National Renewable Energy
Lab, MS 3219, 1617 Cole Blvd. Golden, CO 80401-3393, USA.
y
E-mail: ingrid_repins@nrel.gov
z
This article is a U.S. Government work and is in the public domain
in the U.S.A.
Published in 2008 by John Wiley & Sons, Ltd.
illustrating the sample emissivity change during the
portion of the deposition when the sample is Cu rich.
6
Processing of the record device differed from that
previously described in three respects. Most notably,
the third stage was terminated without Ga.
7
Through
the majority of the deposition, Ga and In fluxes were
delivered in the typical Ga/(In þ Ga) 0 3 ratio.
However, during the last 1 0 s of the deposition, about
25 A
˚
of In were delivered in the absence of Ga. Second,
after the deposition was terminated, the sample was
subjected to a 25-min anneal in Se while the sample
temperature was maintained at 6008C. Third, a
2-min, 2008C air anneal was performed after the CdS
deposition. Similar anneals have yielded small
improvements in fill factor and voltage for devices
with efficiencies greater than 15%.
8
We believe that
these three empirical processing changes create a
near-surface region in the CIGS with reduced
recombination.
DEVICE CHARACTERIZATION
The current–voltage characteristics of the 199%-
efficient device were measured by the Device
Performance Group
9
at NREL under AM15 global
spectrum at 258C. Total-area device parameters are as
follows: open-circui t voltage (V
oc
) ¼ 0690 V, short-
circuit current density (J
sc
) ¼ 355 mA/cm
2
, fill factor
(FF) ¼ 812%, efficiency ¼ 199%, and device area ¼
0419 cm
2
. The uncertainty in the efficiency measure-
ment is estimated at 3% relative.
Figure 2 compare s the relative quantum efficiency
(QE) of the new record device with those of four earlier
record devices. The new device is within the envelope
of the older record devices in terms of CdS thickness
(deduced from absorption around 450 nm) and optical
bandgap (as evidenced by the long-wavelength
response cut-off). Normalizing the relative QE to the
measured short-circuit current implies that the
maximum value of the external QE is 965% at 824 nm.
The first six columns of Table I list the current
density–voltage (J–V) parameters for each device
graphed in Figure 2. Although V
oc
and J
sc
are similar to
the older devices, the new device demonstrates a clear
improvement in FF. In fact, the 812% FF exceeds
previous thin-film records for this parameter.
10
Diode
analysis of the J–V data
11
indicates that the improved
FF is due to decreased recombination, not series
resistance. The last three columns Table I compare
series resistance R, diode quality factor A, and diode
saturation current density J
0
for each of the devices of
Figure 2. A clear drop in diode quality factor and
saturation current has accompanied the recent effi-
ciency improvement. In Figure 3, the associated
increased slope of the log J–V plot is apparent, either
whether one views the raw data (Figure 3a), or data
Figure 1. Deposition data for CIGS film M2992.
Figure 2. Relative quantum efficiency of five record NREL
devices.
Published in 2008 by John Wiley & Sons, Ltd. Prog. Photovolt: Res. Appl. 2008; 16:235–239
DOI: 10.1002/pip
236 I. REPINS ET AL.
with the x-axis corrected for series resistance
(Figure 3b).
A decrease in recombination would normally be
expected to produce an increase in FF and V
oc
.
However, in the new device, the deposition was ended
with a small amount of In, without Ga. V
oc
is
determined by Ga content in the space-charge region
(SCR), including that at the surface of the CIGS;
12,13
therefore, we hypothesize that the method of reducing
recombination presented here is achieved at the price
of a slightly lower bandgap in a portion of the SCR, and
thus no V
oc
increase is achieved. Minimum bandgap,
which is the main determinant of J
sc
, occurs about
05 mm into the film and is unchanged by near-surface
variations.
FILM CHARACTERIZATION
Film composition and orientation are similar to those
of previous record devices.
1,2,14
As measured by
electron-probe microanalysis using 20-kV electrons,
the atomic Cu ratio, Cu/(In þ Ga), is 081. The atomic
Ga ratio, Ga/(In þ Ga), is 030. Auger electron
spectroscopy and sputter depth profiling indicate the
typical notch profile in the Ga ratio. X-ray diffraction
indicates a strong <220/204> orientation.
An examination of the effect of the modified surface
processing on film composition was performed via
Auger electron spectroscopy in conjunction with a
slow sputter depth profiling. Results are shown in
Figure 4. The circles in Figure 4 represent data from
the 199% device. The triangles show data from CIGS
made using an identical recipe but omitting the In-only
termination. Hollow symbols indicate Ga ratio and are
Figure 3. Log current plus photocurrent versus voltage,
without (a) or with (b) x axis correction for series resistance,
illustrating difference in slope between recent devices.
Table I. Comparison of parameters extracted from J–V analysis of recent record NREL devices
Device Area (cm
2
) Efficiency (%) V
oc
(mV) J
sc
(mA/cm
2
)FF(%)R (V-cm
2
) AJ
0
(mA/cm
2
)
C1068-2 0.450 18.8 678 35.2 78.7 0.41 1.30 5.3 10
8
S2051-A1 0.408 19.2 689 35.7 78.1 0.27 1.48 5.2 10
7
C1675-11 0.406 19.3 668 36.2 79.6 0.14 1.29 6.5 10
8
C1812-11 0.409 19.5 692 35.2 79.9 0.24 1.33 6.4 10
8
M2992-11 0.419 19.9 690 35.5 81.2 0.37 1.14 2.1 10
9
Figure 4. Atomic ratios near the CIGS surface, as measured
by Auger spectroscopy. Hollow symbols indicate Ga ratio
(left axis), and solid symbols indicate Cu ratio (right axis).
Circles represent data from the record device, and triangles
represent data from a film made with the identical recipe, but
omitting the In-only termination.
Published in 2008 by John Wiley & Sons, Ltd. Prog. Photovolt: Res. Appl. 2008; 16:235–239
DOI: 10.1002/pip
199% EFFICIENT Cu(In, Ga)Se
2
-BASED THIN-FILM SOLAR CELLS 237
read from the left axis. Filled symbols indicate Cu ratio
and are read from the right axis. The 199% CIGS
exhibits a slightly lower Ga ratio near the surface of the
film than does the comparison piece, suggesting that
some vestige of the In-only termination survives the
high-temperature processing. The Cu ratio for both
films shows reduced values near the surface, typical of
the defect chalcopyrites
15
found in high-efficiency
devices.
Figure 5 shows a scanning electron microscope
(SEM) cross sectio n and plan view of a portion of the
CIGS/Mo/SLG film that was not finished into devices.
As expected, the cross section (Figure 5a) shows large
grains extending from the back to the front of the film.
The CIGS is 22 mm thick via SEM cross-section or by
mechanical profilometer. This thickness is about
05 mm thinner than previous record devices. As
improved performance in the record device was due to
decreased recombination rather than decreased series
resistance, the thinner absorber layer is not likely an
essential characteristic of the improved device. An
atypical feature of the plan view (Figure 5b) is the
appearance of voids in the CIGS. Other work has
linked voids in CIGS with excess Se,
16
thus these voids
may be a product of the fluctuation of the Se rate to
high values (see Figure 1).
CONCLUSIONS
A new record efficiency of 199% was demonstrated
for a CIGS solar cell. The device exhibits significantly
lower recombination and higher fill factor than earlier
devices. Slight modifications to the CIGS surface are
believed to be responsible for the improved perform-
ance.
Acknowledgements
This work was performed for the US Department of
Energy Photovoltaics program under contract
DE-AC36-99GO10337 to NREL. The authors would
like to thank F. Hasoon at NREL for discussion of film
growth, and T. Moriarty and K. Emery at NREL for
cell characterization.
REFERENCES
1. Ramanathan K, Contreras MA, Perkins CL, Asher S,
Hasoon FS, Keane J, Young D, Romero M, Metzger W,
Noufi R, Ward JS, Duda A. Properties of 192% effi-
ciency ZnO/CdS/CuInGaSe
2
, thin-film solar cells. Pro-
gress in Photovoltaics Research and Applications 2003;
11: 225–230.
2. Contreras MA, Egaas B, Ramanathan K, Hiltner J,
Swartzlander A, Hasoon F, Noufi R. Progress toward
20% efficiency in Cu(In,Ga)Se
2
polycrystalline thin-film
solar cells. Progress in Photovoltaics Research and
Applications 1999; 7: 311–316.
3. US Patent No. 5,441,897 (15 August 1995) and US
Patent No. 5,436,204 (25 July 1995).
4. Contreras MA, Tuttle JR, Gabor A, Tennant A, Rama-
nathan K, Asher S, Franz A, Keane J, Wang L, Scofield J,
Noufi R. High efficiency Cu(In,Ga)Se
2
-based solar cells:
processing of novel absorber structures. Conference
Figure 5. SEM cross section (a) and plan view (b) of M2992.
Published in 2008 by John Wiley & Sons, Ltd. Prog. Photovolt: Res. Appl. 2008; 16:235–239
DOI: 10.1002/pip
238 I. REPINS ET AL.
Record of the 24th IEEE Photovoltaics Specialists Con-
ference 1994; 68–75.
5. Contreras MA, Romero MJ, To B, Hasoon F, Noufi R,
Ward S, Ramanathan K. Optimization of CBD CdS
process in high-efficiency Cu(In,Ga)Se2-based solar
cells. Thin Solid Films 2002; 403–404: 204–211.
6. Kohara N, Negami T, Nishitani M, Wada T. Preparation
of device-quality Cu(In,Ga)Se
2
thin films deposited by
coevaporation with composition monitor. Japanese
Journal of Applied Physics 34: L1141–L1144.
7. Abushama J, Noufi R, Johnston S, Ward JS, Wu X.
Improved performance in CuInSe
2
and surface-modified
CuGaSe
2
solar cells. Conference Record of the 31st
IEEE Photovoltaics Specialists Conference 2005;
299–302.
8. Wang X, Li S, Kim WK, Yoon S, Craciun V, Howard JM,
Easwaran S, Manasreh O, Crisalle OD, Anderson TJ.
Investigation of rapid thermal annealing on Cu(In,-
Ga)Se
2
films and solar cells. Solar Energy Materials
and Solar Cells 2006; 90: 2855–2866.
9. See, for example http://www.nrel.gov/pv/measurements/
device_performance.html
10. Kessler J, Bodegard M, Hedstrom J, Stolt L. Baseline
Cu(In,Ga)Se
2
device production: control and statistical
significance. Solar Energy Materials and Solar Cells
2001; 67: 67–76.
11. Sites JR, Mauk PH. Diode quality factor determination
for thin-film solar cells. Solar Cells 1989; 27: 411–417.
12. Dullweber T, Hanna G, Rau U, Schock H. A new
approach to high-efficiency solar cells by band gap
grading in Cu(In,Ga)Se
2
chalcopyrite semiconductors.
Solar Energy Materials and Solar Cells 2001; 67:
145–150.
13. Dullweber T, Hanna G, Shams-Kolahi W, Schwartzlan-
der A, Contreras MA, Noufi R, Schock HW. Study of the
effect of gallium grading in Cu(In,Ga)Se
2
. Thin Solid
Films 2000; 361–362: 478–481.
14. Contreras MA, Romero MJ, Noufi R. Characterization of
Cu(In,Ga)Se
2
materials used in record performance solar
cells. Thin Solid Films 2006; 511–512: 51–54.
15. Schmid D, Ruckh M, Grunwald F, Schock HW. Chal-
copyrite/defect chalcopyrite heterojunctions on the basis
of CuInSe
2
. Journal of Applied Physics 1993; 73(6):
2902–2909.
16. Beck ME, Swartzlander-Guest A, Matson R, Kean J,
Noufi R. CuIn(Ga)Se
2
-based devices via a novel for-
mation process. Solar Energy Materials and Solar Cells
2000; 64: 135–165.
Published in 2008 by John Wiley & Sons, Ltd. Prog. Photovolt: Res. Appl. 2008; 16:235–239
DOI: 10.1002/pip
199% EFFICIENT Cu(In, Ga)Se
2
-BASED THIN-FILM SOLAR CELLS 239
