A Conversation with Applied Material’s Solar Head Charlie Gay
by Neal Dikeman
I had a chance to chat today with Dr. Charlie Gay, the President of Applied Materials’ (AMAT) solar division. You may recall, we broke the story in the blogosphere 5 years ago about Applied’s entry into solar, which was anchored with a highly touted and very aggressive strategy for turnkey large format amorphous silicon and tandem cell plants called SunFab.
Charlie reminded me that when they began 5 years ago, they did so along two major thrusts: The acquisition of Applied Films in June 2006 getting an inline coating system for deposition of silicon nitride passivation layers on crystalline and in parallel an internal project to adapt their large flat panel display manufacturing technology for photovoltaics.
They still like the large module format, for a simple reason, cost in the field for large scale solar farms is heavily about getting area costs down relative to power output. I was excited for another simple reason, when major capital equipment developers get involved, manufacturing maturity is not far behind, it forces everyone to rethink scale in different ways.
After a huge initial splash outselling everyone’s expectations in that SunFab concept, many industry analysts later kind of wrote them off as flash in the pan when they were reported having problems as implementations came in slower and smaller and harder than expected on their SunFab lines a couple of years ago, and a saw a major restructuring in 2009. But they’ve had success with that product anyways, EVERYONE saw a major restructuring in 2009, and more importantly the original vision of leading solar into mass manufacturing is still going strong, now across a range of products and technologies in thin film and crystalline manufacturing equipment. Let’s put it this way, in their annual report they call themselves the largest equipment manufacturer to the solar sector, they have $1.5 Billion in annual revenues in the Energy & Environmental division, which is heavily PV, and there are like 120 mentions of the word solar in their annual report, almost once per page.
So what I really wanted to talk to Charlie about was the future of PV manufacturing. He frames the future by drawing a mirrored parallel between photovoltaics and integrated circuit manufacturing, beyond just semiconductors:
- In IC, dozens to hundreds of device architectures exist, but basically one material, silicon.
- In PV, there is essentially one architecture: the diode, but dozens to hundreds of material choices.
But silicon has been the mainstay material of PV for a number of reasons. So we got into one of my favorite topics, the manufacturing improvement potential in crystalline silicon.
His version of Moore’s law for solar runs like this: the thickness of the solar cell decreases by half every 10 years. Today it’s 180 microns thick. The practical possibility exists to get down to about 40 microns, with some performance improvement by making it thinner, but we can’t go much below 40 without being too thin to absorb enough light. This fits with other conversations I’ve had suggesting that over the past couple of years most of the major crystalline solar manufacturers were working on paths to take an order of magnitude out of cell thickness.
If this comes to fruition, crystalline can literally wipe the floor with the existing thin film technologies. Basically think sub $1 per watt modules with the performance of high grade crystalline modules today. And as cost per watt equalizes, that higher efficiency starts to really tell, as since Balance of Systems costs have fallen at 10-12% per doubling of installed fleet, compared to module costs falling at 18-20%, in a world where BOS increasingly matters, the old saw about lower area cost per unit of power installed starts to actually bite for once. Think ultra thin high performance low cost large format x-Si modules with fancy anti reflective coatings and snazzy high grade modules with on module inverters or DC optimizers mounted on highly automated, low cost durable trackers. Think solar farms approaching effective relative capacity factors of 2.5-3 mm kW Hours per year per MW on 25 year systems at $2-3 per Watt installed. Possibly the only thing on the planet that could match shale gas.
In fact, the entire thesis of thin film as a business and venture capital prospect has been built on the premise that crystalline material costs were just too high to get to grid parity. I’ve got scads of early thin film business plans touting that. That thesis is under extreme pressure these days. I’d submit that if the industry 7 years ago had really understood how much improvement could be had, we’d have saved billions in potentially stranded thin film development.
Charlie says there are about a dozen different paths for enabling 40 micron cells. The most interesting approach to him is an epitaxial growth process on reusable silicon templates. A process which grows a thin layer of silicon on top of a reusable layer of silicon, using perhaps one mm thick silicon templates, etching the surface, and directly depositing silicon from trichlorosilane gas. The idea would be to rack templates into a module array, grow the cells in an oven to your 40 micron level, then glue the glass module to the back side, and then separate it off to form a “ready to go assemble” module. The challenge is basically oven and materials handling designs that get it cost efficient in high volume.
In essence, all you’d be doing is integrating a silicon ingot growth process directly into a module. Instead of growing ingots, cutting thick wafers, forming cells, then building modules from them, you grow cells racked into their own module personally instead of growing ingots first.
Hella cool. A process like that means using fairly manageable capital equipment and materials handling technology development in known device and module technologies we could literally rip the ever living guts out of crystalline manufacturing costs. And there are 11 more paths to play with???
The way he thinks about it, on a broader perspective more people are working in photovoltaic solar R&D today, by his estimate some 70,000 researchers and $3 billion per year, than in all of the prior PV history. And that means whereas perhaps five main innovations over 35 years drove almost all of crystalline PV manufacturing costs (screen printing, glass tedlar modules, adapting steel from tires for cutting wafers, silicon nitride processes, and fast metrology tools), in today’s world, Charlie thinks we see 5 equivalent innovations in PV manufacturing technology every 2 years.
So I asked him to comment on whether there were parallel cost-down opportunities for thin films or whether it is an also ran waiting to happen. He thinks there are. He mentioned organics. I pushed back hard, as organics have been written off by almost everyone for never seeing yield or performance, so where does he see the opportunity? He responded that he picked organics to keep me from narrowing the materials field prematurely to just A-Si, CdTe, CIGS, and GaAS. Silicon just like carbon can surprise us, e.g. bucky balls, carbon nanotubes, and just because early materials had stability and process issues, doesn’t mean we’ve exhausted the opportunities.
He says what he wants us to recall is that we are currently operating in PV manufacturing today with the materials that were on the radar in the energy crisis from 1974-1980. That is changing in the lab and universities these days. And given time the results will surprise us.
He draws a parallel
between photography and photovoltaics, both invented in 1839, both rely on sunlight acting on materials. In photography, people started off putting films on glass, then putting films on mylar, and running things continuously. Implying that in solar, we’re still on glass c. 1890.
He said to think about the original Ovonics/Unisolar vision in thinking about how you get to high speed continuous processing with thin film (think paper manufacturing, where done roll to roll it’s far more consistent than one-offs can be done). If that is still our ultimate thin film paradigm (got to love the chance to use the word “paradigm”), the stars are still in front of us with what thin film COULD do. And while roll to roll has had significant materials technology and process control challenges for the current class of materials, let’s go back to the mirror parallel to integrated circuits, in photovoltaics, one main device, scads of material options. Just a matter of R&D hours and time.
He markedly did NOT suppose that the current state of thin film devices could beat 40 micron crystalline silicon by themselves. It’s worth considering that we may look back and find that thin film, CdTe and First Solar (FSLR) were the stepping stones to 40 micron crystalline, not the other way around. Maybe my next question to Charlie is whether he and I should set up Neal and Charlie’s 40 Micron Solar Company of America yet.
Neal dikeman is a founding partner of Jane Capital Partners LLC, a cleantech merchant bank whose clients have included the technology arms of multinational energy companies. This article first appeared on CleanTechBlog and is reprinted with permission.
Thin film silicon would be great, though I doubt it will mop the floor with other thin film technology. The problem lies in its indirect band gap, which makes it a poor optical absorber (alpha ~ 15 cm-1). Other thin film technology focuses on materials with direct band gaps (for CIGS and CdTe alpha > 100,000 cm-1). This is a materials property and there is no way around it. So the solution is in the design. Efficient thin film silicon solar cells need complicated anti reflective layers on the front and back to keep the light trapped in long enough to get absorbed. The current laboratory record of thin film silicon solar cells is 16.7%, this is 45µm thick. You can see that here.
So currently thin film silicon solar cells don’t compete with CIGS. For them to be competitive, it’s going to take a very complicated design to trap the light more efficiently. It would be interesting to see what that does to the production cost.
“Think solar farms approaching effective relative capacity factors of 2.5-3 mm kW Hours per year per MW on 25 year systems at $2-3 per Watt installed. Possibly the only thing on the planet that could match shale gas.”
Certainly such costs do sound exciting. But gas fired plants can run all day. No storage cost or land cost built into discussion in article.So gas still seems to win handily…except in places where PV output match peak power use. Then such PV systems seem to beat gas peaker only power.
Thanks for the insights Jesse. It’s easy to lose sight of the underlying physics and engineering when talking about big trends, but they are a useful anchor to reality.