Leland Teschler
Editor

Solar cells today can't generate electricity as cheaply as fossil fuel-fired generators -- but stay tuned. New developments in cell fabrication and manufacturing techniques may start to close the cost gap. Today, solar power accounts for less than 1% of worldwide electricity generation, according to the International Energy Agency. The IEA also says it costs at least 35 cents to produce a kilowatt hour of electricity from solar panels, compared with about three to five cents burning coal.

The typical process for fabricating solar cells looks a lot like that for making integrated circuits. Both solar cells and ICs start with silicon wafers sliced from silicon ingots. The surfaces undergo diffusion processes to create semiconductor junctions, then get conductors added. And the wafer processing in both cases has been a batch process.

Panels made this way account for about 90% of today's sales and are about 12 to 20% efficient at converting sunlight into electricity, depending on the weather. There are efforts underway to boost their efficiency and get their costs down. Typical techniques for improvement concentrate sunlight on the cells through use of mirrors or use nanoscale features that boost the active area of the cell real estate.

But some steps in the cell fabrication process are unavoidable. For example, the process of sawing an ingot of silicon into wafers accounts for about 20 cents of cost per watt or more in the finished panel. All in all, estimates are that silicon wafers account for up to half the cost of the finished silicon solar cell. And silicon solar cells used up 33% of the world's electronic grade silicon production in 2006.

This scenario is in the process of changing, however, as cell makers get serious about perfecting thin-film cell fabrication techniques. Industry analysts say more than 40 companies are now in the process of developing thin-film solar cells.

The idea behind thin-film is to fabricate cells that don't use as much silicon as those based on wafers.

An example comes from Evergreen Solar Inc., Marlboro, Mass., which is pursuing a technology called string-ribbon wafer cells. The technique draws two graphite strings through a silicon melt. Molten silicon is trapped between them via surface cohesion to form 3.2-in.-wide silicon ribbons.

Evergreen says it uses about 5 gm of polysilicon (that is, silicon in the form of multiple crystals) to produce 1 W of power, and figures 3 gm/W may be feasible in a few years. This compares to an average of about 9 gm/W for cells cut from silicon ingots in the usual way.

The advantage of the technique becomes evident by considering Evergreen spends about 50 cents/W on polysilicon (at current polysilicon prices) compared to about a dollar per watt for ordinary cells on wafers.

Another approach to eliminate use of silicon wafers employs semiconducting polymer material sprayed onto plastic substrates. Nanosys Inc. in Palo Alto, Calif., says it is working on a fabrication line for cells that use 200-nmthick films. Its first efforts load the polymers with CdSe and CdTe semiconductor nanocrystals (that is, small grains of crystalline material) to create P-N junctions that convert photons to electrical current. But other types are possibilities in the future.

ROLL-TO-ROLL GETS ROLLING
Though it has yet to produce any solar cells, Nanosys says its technology uses roll-to-roll techniques rather than more traditional IC manufacturing methods. Roll-to-roll production is often viewed as a promising way to manufacture cheap electronics of all kinds, not just solar cells. The basic idea is to use a flexible substrate that can be unrolled from a web, then selectively deposit layers of electronic material on it to create circuits. The usual deposition method envisioned for roll-to-roll electronics is analogous to ink-jet printing, with fine jets shooting material in patterns on the flexible substrate moving past. Once all the material has been deposited and cured, the web material is cut to shape for the final application.

A roll-to-roll process configured this way is attractive because its economies are closed to that of a newspaper press rather to the batch processes used to pattern ordinary IC wafers.

Silicon thin films, however, are mainly deposited by chemical vapor deposition (typically plasma enhanced (PE-CVD)) rather than by ink-jetting techniques. The resulting material can be amorphous silicon or different types of nanocrystalline silicon (that is, amorphous silicon containing small grains of silicon crystals). And though thin-film cells have looked attractive for some time, only recently have they gone into mass production. That's because the production processes have proven to be tricky, and doing so in a roll-to-roll scheme, even trickier.

One of the few firms so far able to devise a workable roll-toroll process for solar cells is the United Solar Ovonic subsidiary of Energy Conversion Devices Inc., Auburn Hills, Mich. It claims to be the largest manufacturer of triple-junction, amorphous silicon thin-film cells.

Much of the company's manufacturing process is proprietary. But its overall approach to roll-toroll production is not. The company uses PE-CVD to lay down layers of the cell, as is normally the case with other thin-film techniques. The last step in the process sputters on indium tin oxide (ITO) to form a transparent conductive coating for attachment of electrical leads.

Deposition operations take place in a 300-ft-long series of interconnected chambers. There are 31 different vacuum deposition areas in the system that are interconnected to allow the thin stainless-steel webs to run through them continuously. The interconnection is such that all chambers can be held at the same vacuum. Six webs of thin stainless steel move through these chambers at the rate of about 0.025 miles/hour. One line thus configured produces about nine miles of solar-cell material every 62 hr.

There are six motors associated with each web, each controlled by a variable-speed drive. Four of the motors are 5-hp units, two for payout and two for takeup reels. Two more 2-hp motors handle an interleaf material that is inserted in web of finished solar cell material to protect the finished surface. Tension on the webs dynamically adjusts and a regeneration system minimizes the overall power requirements.

The most difficult part of the process, claim ECD engineers, is in getting uniform coatings deposited on the moving web. They also say overall limit to production capacity lies in how fast the system can make uniform depositions. Deposition rate depends on the speed of the web, and the deposition processes have been reoptimized as the web speeds have risen over time. Upon final qualification, the newest machine can make up to 30-MW-worth of solar material annually, they say.

Once the cell material is spooled up at the end of the line, it goes to finishing operations for cutting into finished lengths, grid wire installation, and lamination.

Want to buy a solar generator?
According to industry analysts at Solarbuzz.com, the latest average going rate for a 125-W solar module is $4.85/W-peak. (Watt-peak is the power output of a solar module when illuminated under standard conditions of 1,000 W/m2 intensity, 25°C ambient temperature, and a spectrum like that of sunlight that has passed through the atmosphere.) Cell makers are trying to drive module prices down to about $2/W-peak over the next decade. They say this is the level needed to compete in the grid-tied electricity market without any subsidies. The Solarbuzz survey also found the lowest thin-film module price was $3/W-peak for a 42-W module. They also caution that the price of the solar module typically accounts for between 50 and 60% of the overall cost of the installation.

Inside an amorphous silicon solar cell
Thin-film solar cells often use amorphous silicon as the basic photosensitive material, or amorphous silicon in combination with microcrystalline silicon (that is, amorphous silicon containing small grains of crystalline silicon). An example of such a cell is that from United Solar Ovonic LLC. It uses three layers of amorphous silicon created so that each has a different bandgap energy. The different bandgaps let each layer react to a different part of the sun's energy spectrum as a way to boost conversion efficiency. The cells absorb blue, green, and red light in layered thin films of amorphous silicon and germanium alloys containing hydrogen and fluorine.

The microcrystalline silicon has a high electron mobility and is relatively stable. The amorphous materials don't have problems with grain boundaries in the crystal lattice of crystalline silicon used in conventional solar cells. The imperfections can cause a lot of charge carriers to be absorbed in the silicon material rather than generate electrical current, and thus forces conventional cells to be relatively thick (50 to 100 microns) in order to generate a useful amount of electrical current. In contrast, the triple-layer amorphous cell used by United Solar is less than 1 micron thick.

Subhendu Guha, United Solar Ovonic president and COO, examines a sample of the company's thin-film flexible solar cell material that has emerged from football field-sized roll-to-roll processing machines.

A schematic of ECD's roll-to-roll process and a view of the real thing. The massive A-frame structures (below) hold diffusion chamber doors which roll back for service or process tweaking. Gases used in the CVD processes feed up through the floor to metering equipment visible beneath the diffusion chambers. The resulting three-junction solar cells are only about a micron thick.

Production personnel show a portion of the 14-in.-wide stainless-steel roll that will form the substrate for thin-film solar cells from United Solar Ovonic. Six rolls of material unwind under the control of variable-speed drives that are visible in the background. The web of finished material (right) is later tested and cut to shape for use on industrial buildings.