My solar journey and comparison of solar technologies
We all come to our particular world view by a series of decision points, some intentional, some not. My particular path to Amorphous Silicon (A-Si) was pretty intentional. When I got out of graduate school (I did my work in low temperature condensed matter physics) it was in the early '70's and I decided to go into alternative energy as a way of contributing to the welfare of the planet.
I started at a national lab that had a strong program in solar thermal energy, both for active collectors and later passive solar house design. While there I, along with others, pushed for a photovoltaic program at the lab. My particular interest was a-Si thin films. Others pushed for concentrators and ternary compound devices. However, upper lab management sincerely felt that nuclear power would produce all the electricity we would ever need. They were not unkind; they just had a different world view, based on the years of the Manhattan Project. They didn't see the need for a similar crash alternative for solar PV. This would have been around 1980.
Then the Reagan years hit, and solar budgets collapsed. I was fortunate to get a position at a university solar energy institute, which worked with the newly formed DOE to test PV modules and power conditioning systems (PCS's). It was my formal introduction to PV systems, a prelude to later PV manufacturing work. (When I say fortunate I mean it: recall the first Reagan budget cut the USDOE PV program from $150 million to $28 million. A lot of talented PV researchers were forever lost to the US effort).
One must be agile to stay in solar work. When there was a state budget crunch, the solar institute was contracted and I was again on the move. Fortunately, on my own time and money I had done some work on a-Si modeling and presented a paper at a PV conference. That is where I met the 3M solar group, including Frank Jeffrey. They made me an offer to work at 3M on their a-Si solar project. I had just accepted a US Fulbright in energy studies overseas, but 3M was gracious enough to extend my hiring for 7 months. I was finally in a-Si module production research! This was in the mid-80's.
When the 3M project started winding down, Frank and I went with it to Ames, Iowa. 3M had decided to form a joint effort with Iowa State University (ISU) and the state of Iowa, in order to keep a hand in the technology. Initially we were going to remain as 3M employees, but --- by now a familiar story-- this was not meant to be. Frank and I left 3M and formed Iowa Thin Film Technologies (ITFT), as a contract company to build the new production a-Si pilot line. Eventually the 3M-ISU institute withered -- each party thought they were going to be funded by the other -- and ITFT became heir to the technology as its actual practitioners. This was in the mid '90's. ITFT, now PowerFilm, Inc., remains friends with both 3M and ISU, and recognize especially the incubator that ISU provided for the early fledgling company.
Fast forward to 2011: PowerFilm, Inc. is now a publicly traded company and excelling in the production of flexible a-Si thin film solar modules. We have an excellent record for quality products and product support. In the solar field that is not easily come by--it is a field that in my long experience has been marked by more than its share of hustlers and charlatans—quick buck artists.
Well, that's how I came to a-Si thin film PV modules. I haven't regretted it. Now let's turn to the decision process why a-Si has a number of substantial advantages over other PV technologies.
First of all, PV solar is always going to be a mix of technologies. Different climates and applications need different solutions. Limited array area favors higher efficiency cells and solar tracking. Areas of higher direct solar flux favor concentrator arrays. For example, a utility installation in the US desert SW might favor a tracking, concentrator array. Their land area available might dictate this choice.
On the other hand, a house of modest roof area might favor high efficiency crystalline silicon modules, non-tracking and roof-mounted because of aesthetic considerations of the owner.
An architectural design involving flexible fabrics would favor a-Si flexible modules to integrate PV power. Similarly, a large roof area industrial or agricultural metal-roof building might opt for the lower cost per watt of a laminated thin-film PV roof coating. The Building Integrated PV area offers a lot of possibilities.
There are some general comments one can make about solar energy, however, that give a guideline for its development and application.
First, solar energy is a diffuse and intermittent resource. By diffuse I mean that maximum direct solar power is about a kilowatt per square meter (kW/m^2). To collect a lot of energy, a lot of surface area needs to be exposed to sunlight with as high an efficiency as possible. The efficiency is of course limited to 100%. The area is limited by the cost per unit area. So to get a lot of solar energy one has to cover a lot of area as cheaply as possible. Since the manufactured cost of a mass produced item is related roughly to its weight, one can in general say the thinner you can make something offers you the ability to get as much solar energy as cheaply as possible. (The PowerFilm (PF) a-Si thin-film technology is a roll-to-roll processed technology, with its roots in the 3M sand-paper making philosophy—“make it by the square mile, sell it by the square foot”!)
A few words about the intermittent nature of solar energy, intermittency requires storage. However, because the energy use spectra of modern industrialized countries typically have their peak about 2pm, the solar inputs roughly coincide with the grid loads. In fact, it is generally conceded that until the solar electric inputs approach about 10% of the grid capacity, storage is not a worry. The grid can act as the storage for the PV inputs: the PV sends juice into the grid in daytime and the grid supplies at night. Hence, the first utility uses of PV are as peaking plants, replacing gas turbine fired units. With 10% of the installed capacity of the US at around 90 GW, we are still a ways from a storage problem. The whole question of energy storage is a lengthy one, with any number of possible solution scenarios: batteries; hydrogen electrolysis/fuel cells; pumped hydro; compressed air down mine shafts; etc. Again, it will be a mix--you have to use what is handy. Buckminster Fuller envisioned a planet with a completely intertwined solar grid, with the side of the planet in sunlight supplying not only itself but the side in darkness. With superconducting transmission lines this would be more than technically feasible. The big impediment would remain the ability of mankind to cooperate to that extent.
Crystalline silicon (X-Si) solar cells were first invented in the 1950's at Bell Labs. They rose steadily in efficiency from a couple of percent to commercially available terrestrial modules in the 14% efficient range. They work well and have high reliability. They are the dominant bulk of PV module sales, around 80%. Silicon is an abundant material. However, solar grade silicon is a portion of semiconductor silicon so there is a degree of supply/cost uncertainty due to other semiconductor competition. Also, X-Si must be sufficiently thick to capture enough light for PV conversion, typically not less than about 80 microns. X-Si wafers are wire-sawn from ingots or boules, so there is some wastage there, but the waste can be recycled. The wafers are discrete, and are assembled, now robotically except perhaps in China where labor is presently very cheap. Usually glass encapsulation on front is necessary, involving heavy and expensive packaging/shipping costs.
A-Si started out deposited on glass superstrates, and a number of a-Si modules are still made and shipped that way. A-Si shares the abundance of Si, although the A-Si precursor is actually a gas, silicon hydride, which is deposited in a glow discharge chamber. This SiH4 gas is also a semiconductor process gas. However, much less Si material is used in the thin film amorphous form of Si: typical A-Si thicknesses are around 0.5 micron—less than1/100 that of X-Si (the human hair is measured to be 40-120 micron)! The reason is due to the much higher light absorption of A-Si vs. X-Si. On the down side, A-Si is about half the X-Si efficiency, and the A-Si is subject to light degradation. On the other hand, the light degradation is reversible with module heating, and the spectral response of the A-Si favors blue light, making it perform better than X-Si on cloudy days (clouds transmit more blue light than red). In general, with all factors taken into consideration, A-Si requires about 50% more area than X-Si for the same delivered energy in kWh.
The best substrates for a-Si are flexible, due to the extreme thinness of the semiconductor material. This allows for roll to roll processing. The main substrate candidates have been stainless steel (SS) and polyimide (PI) plastic. The SS is typically around 6-8 mil thick, and the PI around 2 mil, although PF is shifting over gradually to 1 mil, with the attendant cost savings. The use of a continuous metal substrate, however, precludes the use of a printed monolithic interconnect like on PF's unique roll to roll electrically insulating PI substrates. Much like X-Si modules the SS substrate A-Si modules must be made from cut and re-wired pieces. One advantage of A-Si modules monolithically produced roll-to-roll like those of PF, is that finished modules, mounted to substrate of choice, does not have to be shipped. Light weight plastic encapsulated module material, shipped with a PSA adhesive backing, can be shipped anywhere in the world and substrate mounted locally.
The other competing thin film technologies are cadmium telluride (CdTe) and copper indium gallium diselenide (CIGS). CdTe's main supplier presently dominates the thin film module sales, about 14% of all module sales. CdTe is easy to apply and can be done in high volume. The hetero junction devices, CdS/CdTe, require thorough encapsulation. Their efficiency is about midway between that of X- and A-Si. CdTe is not done roll to roll, is fragile and is as costly to ship as X-Si. However, the main problem with CdTe material is the ultimate scarcity and cost of Te. Production cannot go much above a factor of two or so at present world CdTe production – based on current levels of Te availability/production.
CIGS can be deposited on web, either SS or PI although its deposition temperature of around 400C makes it a lot trickier on PI than A-Si at 260C. CIGS, like CdTe, has a higher rate of deposition than A-Si. Modules can have an efficiency perhaps a little better than CdTe. CIGS too is a hetero junction device with CdS, requiring excellent encapsulation due to the sensitivity of these materials to water vapor. The necessary flexible polymer encapsulant materials are more expensive than for PF flexible modules. The intrinsic voltage of the device material is about ¼ that of A-Si tandem junction material, requiring 4 times the number of cells and cell interconnects to achieve the same monolithic voltage. Finally, the supply of Indium is linked to that of silver, and the supply of Gallium is very limited.
So, in summary, relative to X-Si, CdTe and CIGS, A-Si lags a bit in efficiency and deposition throughput at present but does not share one or more of the following problems with the others: rigid substrates; non roll-to-roll fabrication; expensive shipping of finished products; discrete interconnects; expensive/heavy encapsulation; severe materials availability limitations; substantially more use of Si material; and metal substrates.
Derrick P. Grimmer © 2011
