Gallium Arsenide
Solar Cells

Gallium arsenide (GaAs) is a compound semiconductor, a mixture of two elements, gallium (Ga) and arsenic (As). Gallium is a by-product of the smelting of other metals, notably aluminum and zinc, and is rare-rarer than gold, in fact. Arsenic is not rare but is poisonous. Gallium arsenide for solar cells has been developing synergistically with gallium arsenide for light-emitting diodes, lasers, and other optoelectronic devices.

Single-crystal silicon and single-crystal GaAs cells have been running nip and tuck in the race to the highest-efficiency PV device. In 1989 experimental silicon cells reached efficiencies of nearly 23% under unconcentrated sunlight; experimental GaAs cells reached efficiencies of almost 26% under unconcentrated light and 29% under concentrated light. Cells in commercial production can average as high as 20% efficient. It was this type of cell that powered the record shattering performance of the GM Sunraycer in the Solar Challenge car race across Australia in 1987. Their high efficiency and their resistance to radiation have also made GaAs cells favorites for powering satellites and other spacecraft.

Properties of Gallium Arsenide

Gallium arsenide is quite suitable for use in high-efficiency solar cells for several reasons:

• The GaAs band gap is 1.43 eV, nearly ideal for single-junction solar cells.
• Gallium arsenide has a high absorptivity and requires a cell only a few microns thick to absorb sunlight. (Crystalline silicon requires a cell 100 microns or more in thickness.)
• Unlike silicon cells, GaAs cells are relatively insensitive to heat. (Cell temperatures are often quite high, especially for concentrator applications.)
• Alloys made from gallium arsenide using aluminum, phosphorus, antimony, or indium have characteristics complementary to those of gallium arsenide, allowing great flexibility in high-efficiency cell design.
• Gallium arsenide is very resistant to radiation damage. This, along with its high efficiency, makes GaAs very desirable for outerspace applications.

In silicon cells, the p-n junction is formed by treating the top of a p-type silicon wafer with an n-type dopant. This process does not work with gallium arserude. Instead, all the active parts of a GaAs solar cell are made by growing sequences of thin, singlecrystal layers on a singlecrystal substrate. As each layer is grown, it is doped in different ways to form the p-n junction and to control other aspects of cell performance.

The GaAs layers are grown via one of two popular techniques: molecular beam epitaxy (MBE) or metal-organic chemical vapor deposition (MOCVD). In MBE, a heated substrate wafer is exposed to gas-phase atoms of gallium and arsenic that condense on the wafer on contact and grow the thin GaAs film. In MOCVD, a heated substrate is exposed to gas-phase organic molecules containing gallium and arsenic, which react under the high temperatures, freeing gallium and arsenic atoms to adhere to the substrate. In both of these techniques, single-crystal GaAs layers grow epitaxially-that is, the new atoms depositedon the substrate continue the same crystal lattice structure as the substrate, with few disturbances in the atomic ordering. This controlled growth results in a high degree of crystallinity and in high cell efficiency.

One of the greatest advantages of gallium arsenide and its alloys as PV cell materials is the wide range of design options possible. A cell with a GaAs base can have several layers of slightly different compositions that allow a cell designer to control the generation and collection of electrons and holes. (To accomplish the same thing, silicon cells have been limited to variations in the level of doping.) This degree of control allows cell designers to push efficiencies closer and closer to theoretical levels. For example, one of the most common GaAs cell structures uses a very thin window layer of aluminum gallium arsenide. The thin dimension of the layer allows charge carriers to be created close to the electric field at the junction.

Lowering the Cost of Gallium Arsenide Cells

The largest barrier to the success of GaAs cells for terrestrial use has been the purportedly high cost of a single-crystal GaAs substrate, estimated at as much as $10,000 per square meter. Although this sounds formidable, it is an oversimplification of the situation.

First, nobody makes square meters of singlecrystal gallium arsenide; the $10,000 cost is projected from that of much smaller substrates.

Second, GaAs cells are used primarily in concentrator systems. The typical concentrator cell is approximately 0.25 cm2 in area and can produce ample power under high concentrations. So, at $10,000 per square meter, the substrate for a typical GaAs cell would cost about 25Q. This cost is low enough to make GaAs cells competitive, assuming that module efficiencies can reach between 25% and 30%o and that the cost of the rest of the system can be reduced.

Researchers are exploring two approaches to lowering the cost of GaAs devices. The first approach is to fabricate GaAs cells on cheaper substrates like silicon or germanium, rather than on the more expensive GaAs substrates. The other is to grow GaAs cells on a removable GaAs substrate that can be reused to produce other cells.

Growing good-quality GaAs crystals for solar cells requires a substrate with a crystal structure matching that of gallium arsenide and with similar thermal properties, i.e., the amount of expansion and contraction that occurs with temperature. Naturally, gallium arsenide itself works best as a substrate. But silicon and germanium can also be used. These materials make cheaper and more durable substrates than gallium arsenide. But the slight mismatch in crystal structures between Si or Ge and GaAs causes imperfections in the growing GaAs crystal. One way around this problem is to grow a relatively thick buffer layer of gallium arsenide between the silicon and the active GaAs cell; the buffer layer then absorbs many of the imperfections in the crystal structure. Another way is to treat cells with hydrogen, as we treat semicrystalline silicon, which lessens the effects of imperfections.

The highest efficiencies of GaAs-on-Si cells obtained in 1989 exceeded 22%. Several laboratories are working to improve our understanding of the GaAs-on-Si material system, since GaAs-on-Si is becoming increasingly important not only for solar cells but also for highspeed microelectronic components-another example of the cross transfer of knowledge between photovoltaic and semiconductor research.

In another attempt to reduce costs, one U.S. company is growing thin films of single-crystal gallium arsenide onto thick, reusable substrates of the same material. These thin films can then be peeled off and incorporated in a PV device; and the substrate can then be used to grow several more thin films. Photovoltaic cells made from this process, which can deliver efficiencies of about 24%, were used to make an experimental flat-plate module at a record efficiency-greater than 20%.

©2002 Sandia National Laboratories|Site Contact|Privacy and Security|PV Disclaimer | Contact Us