| Infomation search |
 |
|
|
| Categories lists |
|
|
|
|
|
| A Brief History of Solar PV: the road from $200 a watt to $1.50 a watt |
| 2009-03-18 22:22:08
(has been browse 652 times)
|
| I’ve had a lot of people explain the history of solar power to me. You can read John Perlin’s book on photovolatics (in the reading list) or go looking around clean tech conferences for some older guys with receding hairlines and ponytails. Ask a few questions and you’ll get all sorts of numbers and anecdotes thrown at you.
That said, if you want a broad overview with some really important perspective, you should read the interview I conducted below with Ajeet Rohatgi, a photovoltaics pioneer, Georgia Tech professor, and founder of Suniva. He’s dedicated his life to building solar cells — even starting the DOE-sponsored Center for Excellence in Photovolatic Research and Education — and he’s fantastic at explaining the evolution of the industry.
Not everyone is going to agree with Rohatgi and the potential of thin-film and organic solar cells, but he does a great job of explaining why he’s put his money on silicon. It’s long interview, so I bolded some of the passages that I thought you’d be most interested in.
Alexis Madrigal: Can you trace the arc of photovoltaics for me over the last century, as you see it, focusing on the materials science evolution?
Ajeet Rohatgi: The photovoltaic effect was discovered 1839 by a 19 year old French scientist. He was playing with an electrochemical cell with 2 electrodes and saw that when the light shines, there was voltage. That’s the first time that anybody saw that you can generate voltage out of light.
People started understanding why this photoelectric effect was there… But from 1839 to roughly 1950, the efficiencies of these solar cell devices were still less than one percent. People were still using selenium at the time. The efficiencies were not going up.
For 100 years, they remained 1 percent or less until 1956 when Bell Labs produced the first silicon cell with efficiency of 6 percent. The New York Times had an article saying we have discovered man’s dream, directly harnessing the sun. The cost at that time though was very high, probably $200 a watt. But the efficiencies started to take off with the first Vanguard satellite. That had efficiencies of more than 10% and it kept on going.
For space, the cost was not an issue. Space technology helped the growth in PV. The main material was silicon. People were starting to see interesting effects on thin-film cadmium sulfide. They were trying to look at these thin films and they looked interesting, but the efficiency was a problem and sometimes the stability was a problem. Thin film technologies were coming around that time, but the performance was low.
Crystalline silicon started proving its robustness in space. Efficiencies were going up, but they were not being used for terrestrial applications, until the 1970s and the oil embargo and everybody woke up. We had a huge program in PV. PV funding rapidly went up and that’s when the PV terrestrial applications exploded. A lot of thin film started to come into play. Copper indium gallium selenide (CIGS), cadmium telluride. But their efficiencies remained on the low side and the stability remained on the low side.
Amorphous silicon… IF you’re looking for six percent like cells, they are flexible. You can deposit on flexible steel and plastics and they do have a niche play, the amorphous silicon, but they can’t do a whole lot of power generation because of the efficiency issues.
Cadmium telluride and CIGS were having trouble in the 70s and 80s.. CIGS made very good cells in the lab and all but when you tried to make high-performance manufacturable devices, you could only make 10 percent efficient cells.
Now with First Solar having great success CIGS, thin films have gotten more visibility. Then Applied Materials entered that business and they are having problems having efficiency. No one has come up with efficiencies over 10 percent. Looking at what is out there, I think thin films especially cadmium telluride, look like a contender long term.
If you go into the high-cost, high-performance technology, people using gallium arsenide and multijunction cells. 40 percent has been achieved.
With multijunctions, you capture the light in the right material, so you don’t lose the efficiency as much. If you have a one material system, take Silicon for example, you have a band gap in these semiconductor materials. 20 percent of the light just goes through the material. So if you put the solar spectrum in front of silicon, the energy will go clear through. Another 30 percent is lost because there are photons with an energy greater than the bandgap and that energy is converted to heat. So people said, “Why don’t I put different band gap material in series?”
Multijunctions can get you very high efficiencies. If you stack about 10 different materials, you can [theoretically] get about 65% efficiency. There are some practical problems. When you start stacking layers, there are interface defect problems because you are putting one material over the other. You run into the interface defects, which hurts the performance, when you start going more than three materials.
These are the Ferraris of solar technology. They have more application in space where cost is not an issue and they are thinner cells. Their weight is less.
Then go on the other side, which is the organic solar cells. These are supposed to be very cheap and they are supposed to be very low cost. Right now they are 2-5 percent at best and they are very vulnerable to humidity and exposure to oxygen. They are exciting materials but they have a long way to go in terms of performance.
The way I define PV industry, you will have room for Ferraris, for gallium arsenide and multijunction. You’ll have Lexus and Bens, monocrystalline, ribbon crystalline. You’ll have Taurus and Camry, amorphous silicon and CIGS. Then you’ll have cars like Pintos and Yugos, which are organic. You can also have SUVs like concentrator cells.
Madrigal: Everyone says the key in solar PV is just getting the cost down. What technology do you think is most likely to get the cost to say, two dollars per watt?
Rohatgi: The idea is to reach grid parity. It’s tricky little term. How do you define grid parity? In the US grid parity is defined as 8 to 10 cents per kilowatt hour. But if you go to California, people pay 15 cents per kilowatt hour. In Connecticut, it is 20 cents per kilowatt hour. There are many areas where the cost of electricity is much higher. Grid parity could be different. In Japan, electricity costs 15-18 cents per kWh. In Italy it’s 24 cents per kWh.
Everybody is trying to get there. The way to get to grid parity is to have an installed system cost of about $3.50 a watt. Then, PV can generate electricity at 8 to 10 cents. There are two components: the cost of the panel, the cells, and the balance of systems, which is the permit, connecting to the grid, project management, etcetera. That number needs to come down dramatically. It’s as important as the efficiency. Right now, just the balance of system cost is about $3 a watt. With that, you can see that you can’t reach grid parity.
As you go from residential to commercial to utility, the balance of system cost goes down. The expectation is that balance of system cost will come down at some time to $1.50 to $2 a watt. That means your panels need to be at $1.50 a watt. That should be the selling price of the panels. If you put a 20-30% markup, your manufacturing costs should be about $1 to $1.25 a watt. Whichever technology can get to 1-$1.25 in watt manufacturing cost, That’s the technology that can be a winner for power applications.
If you now look at crystalline silicon today. Crystalline silicon prices have come dramatically. Now you can get modules which are maybe as low as $2.70-$3 a watt. As I pointed out, if you get to $1.50-$1.75, you’re pretty close to grid parity. Balance of systems cost will continue to come down.
First Solar is talking about manufacturing cost of $1.20 already, but when you have a lower efficiency, like 9% or 10%, your balance of system cost scales up with the area. The efficiency premium could be pretty high, 30-40-50 cents per watt when you are dealing with lower efficiency module.
Sometimes people get carried away with the manufacturing cost of the technology. You must also consider the efficiency. Cadmium telluride looks attractive. Silicon prices are coming down. I think the gallium arsenide cells are still very expensive and organics are not efficient or stable enough. This is the way I see at least near term. There will be a lot of focus on thin films and crystalline silicon.
Madrigal: You see a lot of players from the integrated circuit and semiconductor industry trying to move into photovoltaics. How much knowledge from that semiconductor world can really be applied to PV?
Rohatgi: This is one of the strengths of silicon. There is a lot of knowledge base from the integrated circuit industry. But the technology for making solar cells is very different. In the IC world, you make 1000 chips per wafer. In solar, 1 wafer is 1 cell. So the mindset has to be very different. In IC, the cost of processing is not an issue. You can make 1000 chips per wafer, so if it costs more to make the wafer, you can make more per chips per wafer. Here in PV you use belt printing. IC you use photolithography.
The technologies are different, but the knowledge base that has been developed for silicon are the same: how do you characterize silicon material? You can design and understand silicon devices much better than CIGS or amorphous silicon. This is what has come from the IC industry: the knowledge base, the materials database, the device modeling. It’s very difficult to model CIGS.
This is very interesting. This is good or bad for silicon. The good part is that we know silicon material very well. We know all the properties of silicon. But this sometimes ends up being a disadvantage for silicon because we know too much about it. We are not willing to give it the benefit of the doubt. If you look at where a lot of the investment is going, people will talk about 2% organic cells and say some day they will be 10-15%. Because people don’t know about those materials they are willing to give them the benefit of the doubt.
People sometimes think, “Oh, silicon is not as exciting,” but this has a much better chance than taking an organic. It’s kind of funny how it works. People get more excited because the new materials have the overtones of some very exciting materials or discoveries.
There’s nothing wrong about different materials. I see how difficult it is to increase the efficiency of a well-established material like crystalline silicon. That’s what tells me that if you go into work with materials like cadmium telluride, it is even harder.
Madrigal: What about the talk about concentrated photovoltaics. You hear people talk about how solar cells have two different purposes — capturing light and converting light — so if you could capture light with cheap materials and then convert light with just a little bit of the best possible light converting material, you’d have an efficient system that’s also cheap. It all seems so logical
Rohatgi: It is a good contender, but here are the problems and why the concentrator market is 1 percent of the PV market.
All the logic is good. The advantage of concentrators is that you can displace the cost of material with the cost of optics. You need 500x less photovolatic material, say. So, if you make an expensive cell or material, that’s the way to go. The optics is much cheaper than the cell by itself. It’s very logical that’s the way to go.
But here is what happens, when you go that route and you put 500 suns, the cell gets very hot, so you have to add some heat dissipation mechanism and that starts to add cost. If the cell gets hot, efficiency decreases. When you pump in 500 suns on a cell, you need to dissipate the heat. That means you’re adding extra cost to the cell.
That’s part of the problem. The second part is the way you concentrate the light is you have to track the light. The higher the concentration, the more precise your tracking needs to be. Now you need a tracker to move with the sun to get high concentration. That adds cost and has reliability issues because you need all thse moving parts for 30 years.
The third big problem for concentrators is that they cannot be used very efficiently in places that are cloudy or dusty. Concentrators can only capture 2-4 degree angle. They are good for locations like Phoenix desert, where there is a lot of direct sunshine. They are not good for places like Seattle. They are very location specific. You don’t want to put them in Japan.
So you’ve got three problems. You’ve got heat dissipation, tracking, and the geographical location. In the end, it all comes down to dollars per watt. You can get more wattage, but when you start adding things, it gets more expensive.
Madrigal: One thing I find interesting about Suniva and you is that you’ve been working with all kinds of photovoltaic materials at Georgia Tech for years and when you go to start a company, you pick monocrystalline silicon, the most well-known and kind of boring-seeming technology.
Rohatgi: Here’s the reason I chose monocrystalline. I work with all the different materials. I work with ribbon silicon and cast multicrystalline. For decades, I have been working with all three materials together and a lot of my research has been to make my lower quality materials go to higher efficiency.
I said I want to differentiate my company. How do I bring that? Our knowledge is in the area of making high efficiency. This is what I have been doing all my life, trying to make record efficiency cells. This is what I want to do: make the highest efficiency cells with the lowest cost possible.
You’re lucky if you can get to 17, 18 percent with cast. I can have a better competitive edge in monocrystalline. I believe efficiency will play a very big role. If I went with cast or ribbon, my performance will be limited by the material, not by my knowledge.
In my lab, I have made manufacturable 18% type cells. I thought I can start with an advantage over everyone else with the manufacturable cell. But I have so much knowledge and technology to go to 20 percent and continue on that path. The company started in 2007 and we have made a 20 percent cell in our R&D which has been certified by the National Renewable Energy Laboratory and this cell has been made by low cost screen printing technology. This is very good, actually. If we can transfer this cell to manufacturing line with the low cost technology — I have done the cost analysis — we will get very close to grid parity. This cell can also be thinner. It has an advantage that even if you thin it down, it won’t lose it’s performance.
Already solar modules have come down to $2.70. Grid parity, you need to get down to $1-1.50. I raise the efficiency from 17 to 20 percent and thin the wafer down 20 to30 percent. When I improve efficiency, my balance of system comes down. Also, by thinning the wafer, the wafer is almost 70% of the cost, if I can thin it down by 30%, I can drop the cost. So, really, you have three places where you make a huge dent in the cost. Because of efficiency, your dollar per watt is low.
The key here is to get there with low cost technology. I can make in my lab 23 or 24 percent cell if you want, if you allow me to use high cost technology. I’m trying to get there without adding to many processing steps.
Our plan is that the line we will have next year in 2010, it may not go straight from 17 to 20. We are hoping we will go past 18 by the end of this year and closer to 20 percent by 2010. |
|
|
