Grid parity for solar, at last? July 21, 2011Posted by Maury Markowitz in solar.
Tags: grid parity, solar power
I watch PV developments as best I can, and I’m more than familiar with most of the leading-edge developments. That’s why this story was such a surprise to me. It appears that we might finally have a cell that will allow us to hit grid parity for real.
That cell comes from a small company in California, Alta Devices. They’ve been around for a while and their tech wasn’t a secret by any means. However, they just dropped a bomb – their latest samples are smashing efficiency records one after the other, and the cost of the system is among the lowest of the lows. It’s that one-two punch that makes this story so interesting. So, here we go…
The grid parity problem
The PV industry has long talked about “grid parity”, the point where the cost of power coming out of a panel is the same as what it would cost to buy it from the grid. The price of grid power varies widely, and in some cases solar is already cheaper; California and Hawaii come to mind. But in most locations, where power is dominated by cheap supplies like hydro or coal, solar simply can’t compete – yet. The industry’s been saying parity is five years off for much longer than five years, but that doesn’t stop them from talking about it so much it even has a name, Sack’s Law.
Power systems are measured in terms of “dollars per watt”, the amount of money you have to pay to buy a system capable of generating one watt of power under perfect conditions. Hydro and coal are about $2 a watt, nuclear is anything from $6 to $15 depending on who you ask, and large solar arrays are around $4. Of that $4, about 1/4 is labor and related costs, another chunk is engineering and permit related, and about $2 – 2.50 is the actual equipment in the field. Of that, the panels themselves are about 2/3rds.
So the easiest way to reach parity would be to lower the cost of the panels. And panels have come way down in price over the last ten years, but we’re still not at grid parity. Why? Because our attempts to lower the cost of the panels has driven up the relative cost of the rest of the system.
The cost of a solar panel is dominated by two inputs, the cost of the cells and the cost of the glass sheet on the front (although not all “panels” use them). Of these two, it’s the cells that really matter, and the price of the cell is dominated by the cost of the materials. Not surprisingly, there’s been a considerable amount of industry effort expended on using alternate materials that cost less.
The major effort here has been the “thin-film” cells. A conventional solar cell is roughly 200 to 500 µm thick – 1/4 to 1/2 of a millimeter, but for thin-film cells this might be as low as 1 µm. If nothing else changed, this would reduce the price of the cells by hundreds of times. Unfortunately conventional silicon doesn’t really work when it’s this thin, the light simply goes right through it (after all, silicon is also used to make windows).
Three thin-film candidates have been well explored over the last 25 years, amorphous silicon (a-Si), CdTe and CIGS. Of these, a-Si and CdTe have entered widespread production, while CIGS is just starting to become a player. The price reductions are dramatic; in the case of CdTe, First Solar is producing panels for well under $1 a watt, the first company to do so. Commercial quantities are available for not much more than $1 wholesale.
But there’s a downside… these panels all generate less electricity than conventional silicon cells. Which means the panels are larger. Which means you need to cover more land, which costs you more to buy. Which means you need more racking to put them up, and hire more people to install them. And when all is said and done, the price goes down maybe 1/3rd, not the 1/2 you might expect.
This is a big issue in some cases. If land isn’t cheap, or is limited in size, these panels can actually cost much more than conventional ones. Consider a residential installation where there’s 200 square feet to install on. With normal panels you’ll collect about 15% of the light on that space, enough room for maybe 2700 watts of panels. At $1.80 a watt, that’s about $5000 for panels. With thin-film panels we might get only 10% of the light, or 1800 watts, and at $1.25 that’s just under $2500.
A win, right? Wrong. That system might cost you $7 a watt installed once you pay for everything else as well. In that case we’ve saved only 1/5th of the system cost, but are generating 1/3rd less power. The price/performance ratio has gone down.
The other solution
Thin-films have dominated PV research, but they’re aren’t the only solution to the problem. Another solution is to use materials that are much more efficient, and reverse the balance-of-system problem noted above. The main line of attack in this effort has been GaAs.
GaAs has a number of advantages over silicon. Primary among them is that it has high “electron mobility”, which, as the name implies, means the electrons can move around the material much more easily than in silicon. This is why it’s used for telecommunications products (there’s a GaAs chip in my iPhone), because it can easily handle higher frequencies and power levels. There’s also the possibility of stacking several cells on top of each other, a trick called “tandem cells”, which dramatically improves efficiency. In addition, GaAs is not terribly sensitive to changes in temperature, unlike silicon, which sees dramatic decreases in performance with rising temperature. Ironically, this means that traditional solar panels are at their worst on bright sunny days.
On the downside, GaAs is extremely expensive, and extremely fragile. This makes the chips difficult to build, often requiring lots of manual steps, as well as expensive in any process anyone’s introduced so far.
For power generation use, you make tiny chips, about 1 cm across, reducing the material use and allowing you to “pot” them firmly to protect them. These chips are extremely efficient, around 40 to 50%, three times that of conventional panels. To get that efficiency, you use mirrors to shine more light on them, which they don’t mind because they can handle the higher temperatures. Then you hope that the reduction in “balance of system” costs, the big mirrors, is enough to make it worthwhile.
The fact that no commercial GaAs solar arrays are in production is an indication that they just haven’t managed to pull this off. So far the balance of system costs are actually higher than entire installations using conventional silicon.
So, you see where this is going…
Well obviously what we want is a system with the efficiency of the GaAs, the tiny material use of the thin-film approaches, and easy manufacturing. And that’s exactly what Alta announced. Using a new manufacturing system, they’re producing 1 µm thick GaAs cells that are peeled off a mould and remain flexible.
Better yet, using a new trick, they’ve managed by bump the efficiency even at this thin size, and now have the highest efficiency of any single-junction solar cell, at 28%. For comparison, conventional silicon is around 16 to 17%, a-Si and CdTe around 10 to 12%, and CIGS around 14%. So Alta’s solution is double the performance best thin-film systems currently available. Even the expensive sort of tandem GaAs chips are barely able to reach this performance level, under “one sun” the best available cells are currently around 30%, a number Alta claims they’ll be able to hit.
The pricing is currently confusing – they talk about the same price per watt as CdTe but generating three times the power, but it’s not clear what this would mean exactly. The most obvious reading is that these panels will cost about $1 a watt, and produce three times as much power per area.
So here we have a solution that can attack the price problem from both ends, it’s cheaper to produce and reduces installation costs as well.If production really does start this year, and they really do hit $1 a watt, this is a major advance in the solar story.
Update: I found this article, which talks about 60 cents/watt pricing. If that’s production costs, like CdTe that they use for comparison, then we should expect wholesale pricing around $1. Given balance of system at less than $1, system costs of $2 a watt or lower would be possible. This brings solar to parity with coal and hydro. For comparison, large PV systems are currently going in at around $4 a watt, so we’re talking about a twofold reduction, well down the Sack’s graph posted above.