Twin Creeks March 17, 2012Posted by Maury Markowitz in solar.
Tags: solar power, thin-film
I’ve touched on the price inputs to PV before, and detailed some of the other pricing pressures in the article on Alta Devices. To date, most of the thin-film efforts haven’t panned out, or just aren’t ready yet. So when Twin Creeks Technologies came out of stealth mode this week, announcing their machinery is ready for installation right now, I was more than a bit surprised.
What was even more surprising was the product. They’re selling a system that you can bolt into existing solar cell production lines that transforms the output into a thin-film version. After the switchover, your line is producing 10 cells for every 1 you used to make, and all you add is a little time and electrical power. Better yet, the cells are partially flexible, so they can be embedded in flexible substrates.
This is BIG NEWS. I’ll try to do it some justice.
When the production methods for thin-film systems were first being explored in the 1980s, it looked like a win-win situation.
Conventional cells are made by melting down a pot of silicon and then drawing it out into a single large block. Once it cools off you saw it up into slices called wafers. Ideally you’d like to cut the cells as thin as possible, but you’re sawing them, so there’s only so thin you go before they break during the cut. And that’s way too thick for what you actually need. Worse, when you get down that small the saw itself is about the same width as the cell – which means that a full 1/2 of the silicon you spent all that money making ends up on the factory floor as sawdust. And that dust is mixed with silicon carbide, an abrasive used on the saws, so it’s expensive to recycle for new cells.
Instead of sawing wafers off of huge blocks of solid silicon, the thin-fim systems used a variety of molten-draw or even spray-on techniques to produce cells that were much thinner. There were two big advantages. One was that you could produce cells in a continuous fashion, which would, hopefully, lower production costs. Instead of waiting for the blocks of silicon to cool, you simply pulled a backing sheet, typically stainless steel, through the machine and rolled it up at the other end. The other advantage was that the cells used much less silicon, about 1/10th as much, and silicon is the major component of the price of a panel. And there’s no sawing, so there;s less, if any, waste.
Better yet, the cells were also expected to produce more power. Why? Imagine light hitting the front of a solar cell and knocking off an electron, the “photo excitation” event. The electron now has to travel to the front of the cell for collection on those little wires you can see. On the way it bumps into other atoms and any number of things that will cause it to lose energy. If the cell is 1/10th as thick, it will suffer 1/10th the losses, and come out the front with more energy. The prediction was that such cells would be around 25% efficient, in an era when conventional techniques were around 12% That’s impressive!
Sadly, that last prediction didn’t turn out. Far from it. The defects caused by the method of construction turned out to leave lots of “holes” left in the crystals, which sucked up electrons like a sponge. Efficiencies were terrible, 5 to 6% at best, and got worse over time as hydrogen got stuck in the holes. That’s improved to about 11% these days, but during that same time the traditional methods also improved performance, cells around 20% are widely available. Additionally, it turns out that light tends to go right through the cells, in spite of previous predictions to the contrary. There’s ways to combat this, but they tend to drive up costs.
That didn’t mean thin-film went away, instead it found a number of niche roles where its light weight was useful, or its flexibility. Uni-Solar, Oerlikon and other companies have made a go of it, with varying levels of success. But as costs have come down on conventional systems, and efficiencies improved, the thin-film companies simply haven’t been able to compete.
And that’s why the Twin Creeks solution is so different – it’s a thin-film system based on conventional cells and production lines. What they’ve invented, effectively, is a new type of saw.
And what a saw! Instead of the spinning metal thread or diamond saws in conventional systems, they have a particle accelerator that shoots hydrogen atoms into the cell. By controlling the energy of the particles, they can control exactly how deep into the cell they penetrate before they stop. So they create a layer of holes at a precise depth. When you heat the cell up, a layer of the cell pops off the top. Then you run it through again, pop off the next layer, and keep going until you’ve used up the cell.
Let’s put some numbers to this. Conventional cells back in the 1980s were about 350 micrometers (µm) thick, about five human hairs. Over time that’s been reduced down to as little as 180 µm, largely though the use of robots that could handle cells that were so fragile. Twin Creek’s system cuts off layers only 20 µm thick. So you get 9 or 10 cells per original cell.
But here’s the thing; the cell is otherwise unchanged. So if that cell was a high-end 20% efficiency model, well, you should get a high-end 20% efficiency cell that’s 1/10th as thick… avoiding all of the problems with the “conventional” thin-film techniques.
Now, sadly, it won’t cost 1/10th. The system adds more construction time and more equipment stations, and the particle accelerator it uses as a saw is both expensive and energy hungry. And there’s all the processing that happens after the sawing, which changes only slightly. You also have to apply all of the techniques needed to trap the light, like other thin-film approaches.
So they’re only 1/2 the price. Awww, darn!
But wait, there’s more. Once you get down below about 35 to 50 µm, silicon gets fairly flexible. That’s because the thicker cuts tend to have mechanical flaws that are subject to a mechanism known as “crack propagation”, but thinner cuts slice these right out. Conventional panel designs start with a very expensive sheet of glass that the cells are glued onto to make sure they never get bent. But at 20 µm you don’t care, you simply dip them in plastic to protect them from the elements. Bend away!
So not only are the cells cheaper, but the panel is cheaper too. And since the panel is flexible, you don’t need to mount it on big metal rails to make sure they don’t bend. And that, my friends, is one big part of the installation costs, mounting those rails.
They’re not the first to attempt this solution to cut cells. Similar systems have been used for a variety of tasks in the microprocessor industry for years, typically injecting atoms into the chips during construction. A number of university teams have used these devices on solar cells in the past, but they were way too slow, you’d need hundreds of them to make any sort of reasonable production line. That’s the problem Twin Creeks’ solved; their Hyperion 3 accelerator is “10 times more powerful” (100 mA at 1 MeV) than anything on the market today. Now you need 10 instead of 100, which is about the same number of conventional saws you might have.
Twin Creeks says cells from their line cost about 40 cents a Watt, half of what cells from existing conventional lines cost. In fact, conventional cells aren’t 80 cents any more, and are already approaching 45 cents. So unless they get to market really quickly, they might be trapped in the same market squeeze as other companies before them. But they do have another advantage, the flexibility. That might allow them to make cheaper panels, in spite of the cell price.
This is big news.
But they’re not alone
Almost all of the advantages of the Twin Creeks process are also shared by the Alta design. Alta’s system uses a coated plate that they grow the cell on instead of sawing, but like the Twin Creek concept, they pop off a layer around 20 µm thick and off you go.
Alta has a major advantage over Twin Creeks though, 28% efficiencies. If Twin Creeks panels work the same as conventional designs, you might expect a 19% efficient cell to give you a 15% efficient panel. In the case of Alta’s cells, you might get a 25% efficient panel. That’s a huge improvement, it means that you get almost 50% more power for every dollar you spend on installation.
But the question remains whether or not Alta can actually make their cells in a production setting. Twin Creeks implies they’re shipping their system now. Their technology is nothing new, it’s a development of existing production systems. And with the exception of that one box, the rest of the assembly line remains largely unchanged.
There’s also SilGen to consider, as they’re developing a system that appears to be very similar to Twin Creek’s and demonstrated cell thicknesses about the same range. Yet they don’t seem to have gone anywhere with it commercially. And 1366 has their molten-to-cell system on the market, although that’s much more radical.
The CEO states they expect a half dozen systems to be in production next year. I don’t see any reason to believe this isn’t true. Time will tell, as always, but needless to say I’ll be watching this story very carefully.