Farhad Moghadam loves to make semiconductor analogies when he talks about his corner of solar photovoltaics. The former Applied Materials and Intel exec is wearing a new hat these days as he marks his third month as president/CEO of Ascent Solar. The company just announced its latest conversion efficiency numbers for its copper-indium-gallium-(di)selenide thin-film PV cells (14.1% peak) and modules (11.7% peak, 10.5%-plus median) and will be ramping its roll-to-roll CIGS-on-polyimide process to volume scale in its Thornton, CO, factory early next year.
When I spoke to Moghadam on the phone last week, I joked about him being another “semiconductor refugee” in the solar biz. He laughed, and then remarked on the benefits of the “cross-pollination of semiconductor knowledge” with PV.
The realms of productivity, manufacturing efficiencies, yield
enhancement, and advanced process control are quite familiar to
Moghadam, and he relishes talking about some of the gory details of
process and manufacturing technologies that would have some solar chief
execs speed-dialing their CTO or VP of manufacturing.
Ascent’s
progress over the past year or so has been impressive, raising its
median “center-line” conversion efficiencies—for modules off the
company’s 1.5MW pilot production line--from around 8% last November to
current numbers of 10.5% or better. A goal of 12% for 2010 has been set.
Moghadam
told me that the Ascent team improved the efficiencies by dialing in
the process “all within the CIGS tools themselves,” without yet going
to a higher temperature deposition or other means. One optimization
example he shared had to do with reducing the widths of the screen
lines on the CIGS to 50 µm, thus decreasing the pitch and increasing
the photoactive absorption area of the material. Although he didn’t
mention it, the anecdote had a faint echo of the “shrinking linewidth”
mantra of the IC crowd.
Ascent continues to work with Ube
Industries, its Japanese polyimide material supplier, on a new
substrate material that will be able to withstand a deposition process
temperature of 550°C, a significant jump from the current 450ºC. This
higher-temperature substrate will provide another “knob to turn” (a
favorite Moghadam expression) down the line to push the company’s CIGS
efficiencies even higher.
The accelerated improvement in
efficiencies will boost the nameplate capacity of Ascent’s fab at no
additional capital and operational cost, according to Moghadam. “It’s a
tremendous cost-reduction knob.” (There’s that word again.) Since the
factory was designed for 10% efficiency, achieving 12% or in the future
15% adds more capacity without any additional outlay of cash.
Moghadam
likens it to an example from his semiconductor experience, albeit on a
significantly different scale. “This is becoming like the Samsung flash
memory fabs, which were designed for 100,000 wafer starts and by
tweaking the lithography and other things, they could get an extra
30,000 to 40,000 wafer starts with no additional costs.”
“You
have to identify where the bottleneck tools are in terms of
productivity,” he continued, “and where the yield factors are in terms
of efficiency. All of those go into the cost equation.”
Speaking
of yields, he said that Ascent’s are in the 85-87% range now, “and the
goal is to get to 95% in the next 12 months.” Although yields in the
mid- to high-90s remain a sine qua non in the semi world, he spoke
instead of a similarity between Ascent and the “packaging industry.”
“When
you get to the end of the line, you’ve made a lot of investment on that
polyimide,” he explained. “You can’t have any loss at the end with the
module. The yield has to be in the 95% or better range, or it doesn’t
make sense. ”
When asked whether Ascent has deployed in-line
process metrology tools on its line, “Absolutely!” was his enthusiastic
reply, as he referenced his Applied days when CpK metrics, statistical
process control, and real-time integrated metrology with
feed-forward/feedback capability became de rigeur in chipmaking fabs.
The TFPV company has come up with a monitoring and control system that
combines in-house tools with vendor solutions, along with some
university contributions.
“In the core evaporation system,
it’s very important to keep the pressure constant, for all the
elements, to get the same film across the 1000-foot-long roll,” he
said. “Not only that, but when you do the next roll, you should get the
same film across the width, across the length, and from roll to roll.”
As for “the film thickness uniformity, the
stoichiometry/composition—we have a way to measure the composition of
the film on all four elements (Cu, In, Ga, Se2) in real time inside the
chamber and then feed back,” he said, citing specialized software, SPC,
and data mining as some of the implements in Ascent’s APC toolkit.
“You can establish the norm and when things go out of control, the
equipment will adjust itself.”
Even though the replacement of
cadmium sulfide as the window or buffer layer in the film stack may not
be imminent and isn’t at the top of Moghadam’s priority list, Ascent
has ongoing R&D in that area, having won a DOE/NREL preincubator
award earlier this year to investigate new material solutions.
“We’ve
had some encouraging results but it’s not an easy plug and play,” he
told me, using another familiar expression from chipmakers’ jargon.
“When you do that [replace CdS with another film], you take a big step
backwards in efficiency because cad sulfide has been optimized for
years. You can’t pull that out and put in another material and expect
to get 10.5 to 11% efficiencies right off the bat.”
“We’re
working through the learning curve there, just like any materials
change; in the semiconductor industry (a-ha!), copper didn’t replace
aluminum [as the interconnect] right away, it took a number of years to
do that.” Noting Ascent’s “multipronged approach to replace cad
sulfide,” he points out that “the amount of cadmium we have is less
than 100 ppm in the full stack, well below the level specified for
industry toxicity levels.…I don’t see cadmium as a show-stopper.”
Ascent
is exploring the replacement of the CdS wet process with a dry
evaporation approach as well as possible changes in the material set to
compounds such as indium sulfide or zinc sulfide, but the fruits of the
NREL project won’t ripen for another two to three years. “If it happens
earlier, we’ll pull it in,” said Moghadam.
Looking ahead, March looms as a bellwether month (and 2010 a
critical year) for Ascent, based on other information that Moghadam
shared. As spring returns to Colorado, the first part of the 30MW line
should be ramped up and the webs rolling, module certification by TUV
should be complete, a new moisture/barrier encapsulant material (a
combo of internally developed secret sauce with off-the-shelf
components on top) should be ready with at least two suppliers
qualified, and an increasing stream of the company’s lightweight
flexible modules should be shipping to EIPV and BIPV developers and
direct customers.
Ascent’s new boss is having a ball in his
new job, saying he feels “reborn.” Moghadam welcomes the responsibility
and challenges that come with running the show, happy that “the buck
stops here.”
Ascent Solar is exhibiting at Solar Power International at Booth 1254
PHOTO OF FARHAD MOGHADAM COURTESY OF ASCENT SOLAR