High-Volume Manufacturing of Flexible and Lightweight CIGS Solar Cells
J.S. Britt, S. Wiedeman, U. Schoop, D. Verebelyi
Global Solar Energy, Inc., Tucson, AZ 85747
ABSTRACT
Progress in scaling up CIGS thin film solar cell pro-
duction at a new Global Solar manufacturing line is re-
ported. A new plant with high throughput tools will in-
crease capacity from 4MWp/yr to 40MWp/yr in 2008. The
processes applied to make the product and the status of
the manufacturing processes during the start-up stage will
be reported.
Investigations of tool capability will be presented.
The tool cycle times and coating uniformity have been
investigated to ensure production readiness. Satisfactory
capability has been demonstrated for all coating proc-
esses and large area conversion efficiencies greater than
10% have been independently demonstrated through
substitution in the first-generation production line.
INTRODUCTION
In 1996 Global Solar began developing the technolo-
gies for roll-to-roll manufacture of products based on
CIGS deposited on a flexible substrate. The technologies
required include thin film coatings, roll-to-roll processing
equipment, thin film solar cells, cell electrical interconnec-
tion, packaging, and product reliability testing. In practice,
it is difficult to successfully address the required technolo-
gies separately due to the cross-dependencies and occa-
sionally conflicting goals (e.g. cost vs. performance).
Without an existing model to follow, much of the develop-
ment had to be conducted iteratively as advancements
within each technology were achieved [1,2].
Figure 1. Global Solar 10,220m
2
Tucson manufacturing
facility.
After successfully operating a 4.2MWp/yr CIGS pro-
duction line for several years, Global Solar initiated an
ambitious scale-up plan in 2006. Now in the final stages of
the first phase of the plan, 75MWp/yr of combined manu-
facturing capacity is being installed at new production
facilities in Tucson, AZ and Berlin, Germany. A 40MWp/yr
production line is being installed in a 10,220m
2
production
facility in Tucson, Arizona (Fig. 1). Equipment installation
is planned to be completed in this factory in August 2008.
Tool installation in the 35MWp/yr Berlin production line will
begin in May 2008 and should be completed in Q3 2009.
The new production tools represent an evolutionary
progression of the Global Solar CIGS technology. The
primary design goal for the new tools was decreased
manufacturing cost. Manufacturing cost will be reduced
by increased automation, higher material utilization, and
greater tool capacity. Greater tool capacity leads to re-
duced capital expenses applied to unit production and a
smaller factory floor area requirement. Although the Tuc-
son factory tool installation is not yet complete, a com-
plete line has been installed and process engineering for
each tool set has begun. Preliminary results of individual
process evaluations have been generated.
PRODUCTION PROCESSES
Global Solar employs a batch manufacturing process
based on webs 670m to 1000m in length and approxi-
mately 32cm in width. Substrate webs are stainless steel
approximately 25µm thick. Each web is processed inde-
pendently through the coating steps. Each web and man-
drel combination weighs between 40kg and 60kg. Be-
cause of the sizable weights, webs are loaded and
unloaded into coaters by crane and conveyed between
processes on carts.
The batch-style production facilitates independent
process optimization, and in particular allows each proc-
ess to be developed for optimal speed. Production bottle-
necks can be easily addressed by placing additional tools
where the bottlenecks arise. Batch production also miti-
gates the effect of unscheduled downtime if multiple tool
sets for each production step are available. Finally, batch
production permits off-line characterization between steps
for improved quality control.
Back Electrode
The back electrode structure is Chromium and Mo-
lybdenum. A thin Chromium coating is applied to enhance
adhesion of the Molybdenum coating to the stainless steel
web. Both metals are deposited by pulsed-DC sputtering.
Absorber
CIGS is deposited by multi-source co-evaporation of
the elements (Fig. 2). The effusion sources are loaded
with the least expensive forms of the metals (shot, wire,
978-1-4244-1641-7/08/$25.00 ©2008 IEEE
etc.). In practice, the effusion source control reactivity is
small due to the large source thermal masses of the
sources. The total deposition time for the CIGS coating
(thickness: 1.7µm) is 2.6 minutes.
Figure 2. CIGS deposition by co-evaporation
Buffer
The buffer layer is CdS and is deposited from solution
by a proprietary technique. The targeted CdS thickness is
approximately 80nm. The process effluent is treated by a
purification system to reduce metals and other contami-
nants below levels considered hazardous. Only solid
waste is generated.
Front Electrode
The front electrode is a transparent conducting oxide.
The oxide is deposited by pulsed-DC sputtering (Fig. 3).
The total TCO coating thickness is approximately 100nm.
Figure 3. Front electrode deposition by sputtering (a simi-
lar tool is applied for the back electrode)
Collection Grid and Slitting
The collection grid is formed by roll-to-roll screen-
printing of a silver ink (Fig. 4). The ink is thermally cured
in the same step, prior to re-wrapping the web. The nomi-
nal cell dimensions are 210mm x 100mm. Three cells are
printed across the width of the web. After printing, the
web is slit into three separate reels, so that each reel is
the width of a single solar cell, and contains between 3200
and 4800 solar cells.
Figure 4. Collection grid printing
Stringing
Single reels of printed cells are input to the stringer.
In the stringer, the cells are separated and then serially
attached to one another by bonding conductive ribbons
between the collection grid of one cell and the backside of
an adjacent cell. There are three ribbons per cell (Fig. 5).
Strings are comprised of up to 18 cells and have a maxi-
mum length of 2m and width 210mm. The strings are
electrically characterized and binned according to their
output characteristics. Finally, strings are packaged and
shipped to module manufacturers for integration into their
products. The majority of production in the new GSE fac-
tories will be CIGS strings. The strings are designed to
look like traditional Si solar cell strings to increase their
appeal to module manufacturers with products designed
for those strings.
Pmp (W): 39.5
Vmp (A): 7.3
Imp (A): 5.4
Voc (V): 10.3
Isc (A): 6.7
Figure 5. CIGS String and nominal string electrical char-
acteristics
PROCESS QUALIFICATION
The first tools for all process steps have been in-
stalled into the Tucson production line. For several proc-
ess steps, multiple tools have been installed. In the first
stage of process development, the capability of coating
uniformity has been demonstrated for each process. In
the following stage, efforts shift to process integration to
978-1-4244-1641-7/08/$25.00 ©2008 IEEE
optimize cell performance and meeting manufacturing
goals on the new production line.
Back Electrode
The thickness of the Mo back electrode should be op-
timized for maximum performance and minimal cost [1].
At Global Solar, the Mo coating thickness is characterized
by XRF (Fig. 6). Along the majority of the web length, the
Mo typical thickness uniformity is +/- 3% and the thinnest
coating occurs in the web center. Thinner Mo coating is
frequently observed in the first 100m of deposition. The
source of this effect is under investigation.
Figure 6. Contour plot of Mo thickness (a.u.) down and
across the web as measured by XRF (distance weighted
least squares fit)
Absorber
Coating uniformity of copper, gallium, and indium is
critical to achieving optimal CIGS string performance and
high yields. In practice, thermal evaporation cross-web
uniformity is more difficult to achieve than uniformity down
the web length. The cross-web CIGS uniformity is chiefly
determined by the design of the effusion sources, deposi-
tion zone geometry (location of sources and shielding),
and zone pressure. However, the new CIGS coaters and
effusion sources have been designed with increased de-
grees of freedom to permit better control of coating thick-
ness across the web than was allowed by the previous
generation of ClGS coaters.
Figure 7. Equivalent thicknesses of copper, indium, and
gallium in a CIGS film (across the web width) as meas-
ured by ex-situ XRF
The cross-web profiles of the elements are similar for
a typical CIGS film deposited in the new coaters (Fig. 7).
All elements are deposited from identical effusion sources
in nearly the same environment, so it not surprising that
the profiles are similar. When combined to make CIGS,
the thickness varies across the web width, but the com-
posite ratios Cu/(Ga+In) and Ga/(Ga+In) are relatively
uniform (Fig. 8).
Figure 8. Atomic ratios and thickness of a CIGS film
(across the web width) as measured by ex-situ XRF
The coating uniformity of Cu, Ga, and In along the
web length has been evaluated by XRF for CIGS-coated
webs up to 670m in length (Fig. 9). In this instance, the
web was sampled at identical cross-web locations down
the length of the web. The effusion sources contain rela-
tively large charges of the elements. The large thermal
masses provide stable evaporation rates within the re-
sponse time of the control loop, and uniformity along the
length of the web is generally excellent.
Figure 9. Equivalent thicknesses of copper, indium, and
gallium in a CIGS film (down the web length) as measured
by ex-situ XRF at the web center
Buffer
If the CdS coating thickness is less than optimal, Voc
and fill factor are reduced. If the coating thickness ex-
ceeds the optimal value, Jsc is reduced due to increased
absorption of light within the CdS coating. Global Solar
has developed a non-destructive optical technique for
qualification of the CdS coating thickness on production
webs. The characterization is performed after the CdS
has been applied on the CIGS coating, and before the
TCO coating has been applied. The CdS coating thick-
978-1-4244-1641-7/08/$25.00 ©2008 IEEE
ness is typically within specification in the utilized portion
of the web (web edges are not utilized) (Fig. 10).
0
3
0
6
0
9
0
1
2
0
1
5
0
1
8
0
2
1
0
2
4
0
2
7
0
3
0
0
25
125
225
3250.25
0.45
0.65
0.85
1.05
1.25
1.45
O
p
ti
c
a
l
m
e
tr
ic
(
a
.u
.)
Width (mm)
Length (m)
1.25-1.45
1.05-1.25
0.85-1.05
0.65-0.85
0.45-0.65
0.25-0.45
Spec. Limit
0.045 - 0.060
within Specification
Figure 10. Optical confirmation of CdS coating thickness
within specification down and across a web
Collection Grid
The collection grid has been designed to minimize re-
sistive losses, cell shading, and required Ag ink volume.
To first order, the design targets can be achieved through
proper screen design and tool setup, but other variables
such as ink pot life, screen wear, and environmental con-
ditions can lead the process out of control. Resistive
losses can be severe if the grid finger print geometry de-
viates substantially below the design goals for height and
width. On the other hand, excessive ink application adds
unnecessary product cost.
The ink print process has been characterized by opti-
cal profilometry. Cells were extracted at intervals from a
single printed reel 350m in length. The grid finger print
height and width were characterized at two locations on
each cell. The mean ink height and width down the web
was determined to be within acceptable limits (Fig. 11).
Figure 11. Collection grid finger average print characteris-
tics along the web length
Process Development
The first-generation production tools have been ap-
plied for process development utilizing a “hook and loop”
test methodology. In the initial “hook and loop” tests, indi-
vidual processes in the new tools have been evaluated
against first-generation tool processes. All thin film proc-
esses in the new tools have been demonstrated capable
of producing large-area solar cells (68cm
2
) with conver-
sion efficiency greater than 9% (Table 1). In the final
stage of the “hook and loop” tests, near completion now,
multiple new tool process steps are being evaluated on
single lots. The tests will conclude with the demonstration
of all new processes together in the final string form.
Table 1. Best cell electrical characteristics generated in
“hook and loop” tests
Process Voc (V) Isc (mA) FF (%) η (%)
Back electrode 0.570 2209 61.3 11.2
CIGS 0.504 2657 50.9 9.9
Buffer 0.592 2284 61.6 12.1
Front electrode 0.535 2379 55.4 10.3
SUMMARY
Global Solar is in the process of starting up new fac-
tories in Tucson, AZ and Berlin, Germany. In the first
phase, the annual capacity of the plants will be 40MWp/yr
and 35MWp/yr. A complete line of new tools has been
installed in Tucson and additional lines will be brought up
in Tucson and Berlin during the coming months. The new
tools are based on the prior generation of production
tools. The earlier tools have demonstrated average con-
version efficiency 10% over many production runs. The
improved control and capability of the new tools should
ultimately lead to even higher performance.
Tool capability has been assessed and found ade-
quate for process development to begin. “Hook and loop”
style tests have been conducted, and the results indicate
that conversion efficiencies >10% are achievable by all
coating processes in the new tools. Integrated process
development is now underway. String production will
commence after manufacturing robustness and repeat-
ability have been demonstrated.
ACKNOWLEDGEMENTS
We gratefully acknowledge support from NREL in this
effort under the Thin Film Partnership Subcontract ZXL-6-
44205-13 (TFPPP).
REFERENCES
[1] J.S. Britt, R. Huntington, J. VanAlsburg, S. Wiede-
man, M. E. Beck, “Cost Improvement for Flexible CIGS-
Based Product”, 4
th
World Conference on Photovoltaic
Energy Conversion (IEEE), 2006, pp. 388-391.
[2] J.S. Britt, E. Kanto, S. Lundberg, M. E. Beck, “CIGS
Device Stability on Flexible Substrates”, 4
th
World Confer-
ence on Photovoltaic Energy Conversion (IEEE), 2006,
pp. 352-355.
978-1-4244-1641-7/08/$25.00 ©2008 IEEE