NREL scientists Pauls Stradins, Brian Egaas, and David Young take a look inside their instrument, the Real-Time QE, which quickly measures how each solar cell responds to different wavelengths of light. Credit: Dennis Schroeder |
To
come up with a way to do something 1,000 times faster than it had been
done in the past, you have to count on some serendipity — not to mention
hard work, collaboration and good timing.
Such
was the case with three scientists from the U.S. Department of Energy’s
National Renewable Energy Laboratory, who somewhat accidentally
developed a way to assess the quality of solar cells at a speed that is
orders of magnitude faster than had been done before.
The
instrument, Real-time QE, licensed and embellished by Tau Science Corp.
as FlashQE, uses light-emitting diodes, high-speed electronics and
mathematical algorithms to measure the quantum efficiency of solar cells
up to 1,000 times faster than had been done before. The technology won a 2011 R&D 100 Award, as one of the year’s most significant innovations.
What
used to take 20 minutes — and therefore could be done only with random
samples of cells — now can be done in a second. That means every single
cell on a manufacturing line can be assessed and then sorted into bins
so the cells that respond best to, say, red or blue are kept together on
the same solar module. That way, a mismatched blue-response cell on a
module won’t put the brakes on all the work the red-response cells are
doing. And that means more efficient conversion of photons into
electricity at sunrise and sunset when the red wavelengths predominate.
Speed means putting every cell to the test
Quantum-efficiency
measurements indicate how well a solar cell converts the various
wavelengths of sunlight into electricity. More precisely, QE is the
ratio of the number of light-generated charge carriers collected by a
solar cell to the number of photons of a given energy that are shining
on the solar cell.
Today’s
solar cell manufacturing lines test each cell to determine useful cell
parameters such as how much current and voltage is generated. But those
tests give no information about how the cell responds to each color of
light in the solar spectrum.
Flash
QE’s ability to also test for each cell’s response to color allows
crucial extra information to be fed back into the production line. It
does it so fast, that cells of the same current and the same response to
particular colors can be sorted into particular bins. From these
sorted bins, spectrally matched modules can be made to optimize the
energy produced throughout a day.
Traditionally,
determining how a single cell responds to different wavelengths of
light has taken 20 minutes so only about one in 1,000 cells are plucked
from the manufacturing process for that extra test.
|
Flash QE, though, has the speed to supply that extra rich information for every cell.
It
likely will mean significant jumps in the efficiency values of future
solar modules and arrays that power the fast-growing solar industry as
well as much better manufacturing line diagnostics.
FlashQE
comes on the market at a time when solar manufacturers are working to
weed out any profit-robbing costs from their production lines, boost the
conversion efficiencies of solar cells, and move toward the U.S.
Department of Energy cost goals established within the “SunShot” initiative.
Insights, timing and serendipity
It started in some small labs in NREL’s Science and Technology Facility.
“I
almost forget what we were originally looking for,” principal
investigator David Young said, recalling the time seven years ago when
he was examining how different wavelengths of light penetrated to
different depths in a solar cell. “We just wanted to come up with a real
simple way of shining light of different colors.”
Enter Brian Egaas, who worked close by and was doing work on quantum efficiency.
“We
started looking at LEDs as the source of light, and I remember coming
into the lab one day and saying, ‘There are enough LEDs now that we can
probably get every color of the rainbow,'” Egaas said.
But
this work wasn’t an official project. So, they went to their group
leader, Rommel Noufi, who saw enough promise that he agreed to let them
have $1,000 to buy some LEDs.
|
Egaas
found a mom-and-pop shop in Vienna, Austria, that would supply them
with LEDs that spanned the solar spectrum — and let them buy just a
couple of each color, rather than the hundreds that are bundled together
from larger suppliers.
The
timing was fortuitous. LEDs spanning the solar cell spectrum wouldn’t
have been available a year or two earlier, and the computing power to
gather all the information needed in parallel wouldn’t have been
available much earlier than that either.
“This
invention came about at the time when it first could come about,” Young
said. “When enough LEDs were just coming onto the market, and when we
had enough high-speed computer capability to get all that data coming
out of the cells.”
“We mocked it up, and turned on one LED at a time, to make the measurements,” Egaas said.
“But
there was just too much noise in the quantum efficiency measurement,”
Young said. “Brian had the whole thing rigged up, and we tried to pick
up the speed of each individual measurement, but it was still taking 20
minutes or so to characterize each cell.”
Operate it like the human brain
Enter Pauls Stradins, who had a lab in the same corridor, and was keeping a casual eye on progress by Young and Egaas.
“Pauls
walks through our lab one day and says, ‘Do you realize you can run all
those lights at the same time at different frequencies?'” Young
recalled.
“When he said that, the light just kind of went on,” Young said. “We all realized, ‘Oh, yeah, that’s the way to do it.'”
“I’d been reading a book on how the brain works,” Stradins recalled.
“The
brain has many similarities with a computer, but whereas a computer
does most things sequentially, the brain has a huge number of parallel
channels,” Stradins said. “When an image comes in, it doesn’t process it
‘one pixel, two pixels, three pixels,’ it processes it instantly — in
parallel.”
Applying
the brain’s parallel approach to the challenge ahead of them —
gathering quantum efficiency data from solar cells with a spectrum of
encoded LED light colors — proved to be the key.
“We
knew there were these mathematical things you can do to filter the
processes in real time,” Stradins said. “Because computers have so much
memory now, we could probably just download a whole chunk for one second
and get about a million points.”
By
arranging for each LED to blink at a different frequency, they could
determine how each solar cell generated current in response to certain
colors.
“We
arranged it so we could take our test cell and run it against a
pre-calibrated cell and learn the quantum efficiency of it,” Stradins
said.
“It was a true collaboration,” Egaas said. “There were pieces that everybody had that needed to come together.”
|
Over
the next few years, they brought in summer interns to work on a
prototype 10-LED device “held together by tape,” Young said.
Transferring the technology to private industry
Just
in time for a scientific conference, they got the first data that
proved that rich quantum efficiency information could be gathered almost
instantaneously from a solar cell. Young gave a talk on the instrument
at an IEEE Photovoltaic Specialists Conference in San Diego. He realized
many in the solar industry were intrigued by the promise of a fast
quantum-efficiency tool for analyzing solar cells in the lab and on the
manufacturing floor.
The
first commercial interest in the product came serendipitously. After
being alerted by a colleague that a start-up company was touring NREL
trolling for new ideas to market, Young had 10 minutes to write up some
notes, then “I gave four guys from Tau Science my off-the-cuff elevator
speech.”
“They just got it right like that,” Young said. “They knew the solar market would eat up a fast QE system.
“They licensed the product and now are selling it.”
Tau
Science made significant improvements to the instrument, patenting
their own ideas for LED optics and handling the vast amount of parallel
processed data needed for the technique.
“It’s
been a great collaboration,” Tau Science president Jamie Hudson, said,
adding that co-founder Greg Horner got to know NREL while he did some
post-doc work here.
“Quantum
efficiency is an extremely fundamental technique in solar cells, and
this is the first time it’s been able to be done at speeds to keep up
with the line,” Hudson said. “It tells you the spectral response of the
solar cell and also a lot of information about the front and back
surfaces. You’re able to look at every sample rather than just one out
of 1,000.”
Tau
Science’sfirst shipment of Flash QE was in early 2011 to Oregon State
University, which will use it in its pilot solar-cell production
facility.
Fast-blinking LEDs illuminate the cells in parallel
The
FlashQE system uses an electronically controlled full-spectrum light
source composed of an array of LEDs. Each LED emits a different
wavelength of light. The LEDs illuminate the cell simultaneously, rather
than the serial approach of a conventional system. The key to the
technology is that all the LEDs are flashed on and off at different
frequencies thereby encoding their particular response in the solar
cell. High-speed electronics and mathematics cleverly extract the
encoded information to reveal a full-spectrum quantum efficiency graph
of the cell. A wide variety of information is gathered in less than a
second — information about the ability of the front surface of the cell
to absorb high-frequency light, the quality of thin-film surface
coatings, the ability of the middle region of a cell to absorb a wide
range of wavelengths, how well the back surface absorbs lower-energy
light, the ability of the back surface to collect electrons.
For
multi-junction cells, Flash QE can detect how each of the layers
performs by using the light source itself to “electronically filter” the
light to only measure the response of the cell of interest.
Instant feedback is a competitive edge
Flash
QE is the quickest diagnostic tool for the quantum efficiency of solar
cells, yielding both a voltage current curve showing the amount of
power, and a spectral response gauge, diagnosing how the cells respond
to different wavelengths of light.
Manufacturers
can get a whole new insight into each of their cells, determining, for
example, why they’re not getting good responses from their reds.
Or
Flash QE can detect that the blue response is slowly getting worse and
worse — in real time, soon enough to alert workers that an adjustment
must be made to the line.
Flash
QE works for silicon cells, and also for multijunction cells that use
stacks of materials such as gallium and indium. “With Flash QE, you can
look at the individual responses of each of the layers,” Young said.
“It’s
fast enough to do spatial measurement mapping across the cell,” Egaas
said. “Is the response the same on the edges as it is in the middle? Is
there a cooling problem that makes the edge different? They can learn
that they have to cool it more slowly, change the process based on the
results.”
Like baking with constant vigilance
It’s
like baking bread, Stradins said. Automated bakeries can produce good
bread if the parameters are extremely tight, but if anything goes wrong,
a huge batch gets wasted.
The
family baker, able to take frequent peaks inside the oven, has better
quality control. That feedback, with bread or with solar cells, is a
powerful tool.
NREL’s
LED light source also is a stand-alone invention that could be licensed
by another company for probing things other than solar cells, ranging
from counterfeit bills to skin cancer.