Blogger
Tom Cheyney
Tom Cheyney is former senior editor of PV-Tech / Photovoltaics International magazine. A veteran technology journalist / editor / blogger, he covered the semiconductor, microelectronics and solar sectors for many years - since fax machines were state of the art. His PV-Tech blog has become a must-read for industry insiders and observers. He was also chief editor of "The Rise of Thin-Film Solar Technology" book published in early 2010.
When I read the
brief report about Micrel's contract manufacturing of solar cells in
its San Jose semiconductor facility, I did a double take: "They're
running PV wafers and
semi wafers in the same fab?" I thought. "That
doesn't happen too often." As it turns out, Micrel may be the only
place where there's simultaneous volume processing of chips and
photovoltaic solar cells, according to company VP of worldwide
operations, Guy Gandenberger.
"I don't know of any other fab in the world that is integrating
solar cell manufacturing with IC manufacturing," he told me in a phone
interview earlier today. "Others have tried.... but no one's taken it
to a large-scale production level, except Micrel." Although Guy
wouldn't dilvulge the identity of the "fabless" solar customer, the
devices moving through his company's 150-mm line are die-level,
concentrator-type silicon cells made on
SEMI standard-grade wafers,
with what foundry manager Jim Crow (yes, that's really his name)
characterizes as "400 suns per die" conversion capability.
After Micrel "founders the wafers," Guy explains, the customer
tests the validity of the die, then creates a megamodule plate, where
they take the cells, put a (concentrator) lens on top of them, package
them with the die, and then build the plates. Each megamodule contains
several dozen plates, and a tower utilizes five megamodules. We "are
foundering the most efficient, most concentrated solar cell in the
industry for our customers...where a normal solar cell on a wafer might
get 10x efficiency, these are at least 100 times more efficient."
Micrel's not exactly a newbie in the solar industry, having worked
on an R&D level with a customer for at least 15 years, relates Guy.
But cell production has been progressively ramping for about a year and
a half, becoming "a steady fab filler on a day-to-day basis." About 20%
of the current 70% total capacity utilization has been offset by
solar-cell wafers, a number Guy would like to see in the 30-40% range
in 2008, possibly through the signing of additional foundry customers.
(Note that the company's own products account for more than two-thirds
of its capacity utilization at this point.)
Micrel's Gandenberger has reason to smile.
The optimization of the cell-manufacturing process has involved a
long learning curve, he says. "We have had to be extremely creative to
make the process work." About 95% of the equipment used for the
semiconductor products were suitable for solar, with the remaining
tools customized and adapted, but they "didn't have to buy any new
tools." Micrel's materials suppliers had to meet specific requirements
for the solar processes, and the equipment companies were consulted to
help make certain modifications to the toolsets.
The process is "very controlled, with the same type of quality
standards as the semiconductor industry," including stringent controls
of oxide integrity and particle cleanliness, Guy insists. "We hold the
production of solar cells at the same levels as chips, including die
and fab yields (as in better than 95%), wafer breakage, throughput, and
cycle times." Micrel does all of its own characterization and analysis
work, which helps ensure that the lifetime of the solar cells it fabs
"are far above any other cells on the market."
"It's not uncommon for a solar wafer to follow right after a
semiconductor wafer," notes Guy, sometimes in a merged lot. (That's
right, the solar and
semi wafers might be in the same boat!) The
various wafers can run alongside each other in an oxide furnace, with
no loss in efficiencies, he claims.
Although the solar-cell business has definitely helped Micrel's
margins, it's not the only nontraditional product running in the fab,
mentions Guy, with a little pride in his voice. Examples of other
components integrated (or at least integrating) into the process flows
include MEMS, surge protection devices, near-eye vision technologies,
and biometrics/biogenetics chips, with customers ranging from tiny
start-ups to more mature concerns. "We offer very unique processes,
that no one else will do" (and the higher margins that come with that
service), and have the capability to integrate a wide range of
technologies into the existing processes in the fab.
Guy and Micrel's position is do whatever it takes to "keep it at
the lowest cost structure possible," in order to compete successfully
at both the domestic and global levels: "I will offload costs onto
anything I can."
With the surge in new application-related foundry business, might
there be a need for, dare I say, added capacity in Micrel's future?
"That's something I pray for every day," quips Guy.
