Photovoltaic Cells: Power at a Price

Sunlight in, electricity out. The quest to achieve this transformation efficiently has been driving scientists in the field of photovoltaics for decades.

From handheld calculators to Earth-orbiting satellites, power-producing solar cells have been used in a variety of applications for years. Nonetheless, the push is on to develop new materials, processes, and technologies that boost the performance and efficiency of solar cells and dramatically lower their costs.

Motivation to develop new power technologies comes from a variety of sources. Energy crises, such as last summer's power outages across the U.S. East Coast, "make people aware of just how much we depend on electricity, because when the power goes out, all hell breaks loose," says Michael Grätzel, a chemistry professor and director of the Institute of Physical Chemistry at the Swiss Federal Institute of Technology, Lausanne.

Grätzel has been studying solar power for two decades. But it's only just recently that the topic's popularity among the public has begun to surge, he notes. Power shortages, rising fuel prices, and concern about the environment and the extent of petroleum reserves are focusing attention on photovoltaics and other alternative energy resources.

Since the early 1950s, when researchers at Bell Laboratories made a silicon-based solar cell that generated enough electricity to power common electric appliances, technology enthusiasts have looked for applications. At first, the devices were used primarily by toy manufacturers to make solar-powered miniature ships, radios, and other playthings.

But in 1958, photovoltaic cells broke out of the novelty toy market when the U.S. Navy launched Vanguard I. The 3.5-lb Earth-orbiting satellite was the first to be powered by solar energy. For several years, Vanguard I's silicon-based solar cells reliably provided electricity to the satellite's radios, temperature sensor, and other equipment. With a successful track record in powering small satellites, photovoltaic devices soon were used in more demanding missions throughout the U.S. and Soviet space programs. Eventually, the technology came to be recognized as a standard method for generating electric power in space.

Other applications, a little more down-to-earth in nature, emerged in the 1970s. For example, solar cells began powering warning and safety lights on off-shore oil and gas rigs and in the rail industry. Lighthouses, buoys, and other navigational safety equipment also began drawing electricity from solar panels at that time. And in remote locations, far from conventional power stations and power lines, photovoltaic devices were used to provide electricity and to run telecommunications equipment.

NOWADAYS, photovoltaic systems are used in many consumer applications and can even be ordered through the Internet. From rooftop solar panels for homes and recreational vehicles to cellular-phone and laptop-computer rechargers, photovoltaic devices are available commercially. The price of such systems has fallen over the years, yet the technology remains expensive compared with other energy sources. In addition, for some types of photovoltaic cells, the efficiency, stability, durability, and other performance characteristics are much lower than theoretically predicted values.

To address these issues, researchers are examining a wide range of materials for use in photovoltaic devices. The list includes various crystalline forms of inorganic semiconductors and several types of electrolytes and organic polymers. Scientists and engineers also are focusing on novel solar cell designs and low-cost processing and manufacturing techniques.

At the heart of a solar cell is a light-sensitive, semiconducting material--most commonly silicon. Photons impinging on the semiconductor can excite electrons from the material's valence band to its conduction band. The process generates electron-hole pairs--meaning pairs of negative and positive charge carriers.

If the semiconductor has been doped with impurity atoms so that it contains positively charged (p-type) regions and negatively charged (n-type) regions, then electron-hole pairs created near the interface between the two types of regions (known as a p-n junction) will be affected by a potential difference across the interface. Electrons will migrate toward the positive side of the junction and holes toward the negative side, leading to a flow of electric current.

Solar cells made from wafers of crystalline silicon are common. Large, high-purity single crystals are used to make high-performance cells capable of converting some 25% of incident sunlight into electricity. Even higher efficiencies--roughly 30%--have been demonstrated. But these types of high-end cells, which have proven to be durable in space applications, are very costly because of demanding and energy-intensive crystal-growth and manufacturing processes.

Less expensive forms of silicon, including material with smaller crystal domain sizes and noncrystalline forms, have been studied for years and continue to be investigated. "But you pay a price for using low-cost materials in terms of impurities and defects, which reduce quality and performance," says Ajeet Rohatgi, director of the University Center of Excellence for Photovoltaics at Georgia Institute of Technology. Low-cost amorphous silicon, for example, has been used to make inexpensive solar cell watches and calculators. But such cells degrade with use; their already low efficiencies (around 7%) often fall to half of their initial value.

One area of focus for the Georgia Tech group is boosting performance in multicrystalline silicon cells. These types of devices use silicon having grain sizes of up to a few millimeters and tend to be roughly 10–12% efficient. In contrast, polycrystalline silicon cells, with micrometer-sized domains, exhibit lower efficiencies.

THE APPROACH taken at Georgia Tech is to develop custom-processing steps that are needed for solar cell manufacturing but also serve to enhance material quality. For example, the group uses a gettering procedure involving phosphorus that extracts impurities such as titanium, iron, chromium, and other transition-metal atoms away from the silicon bulk where they degrade cell performance. Somewhat similar to flypaper, the phosphorus layer attracts the impurities and immobilizes them.

In another processing step, hydrogen is injected through a silicon nitride antireflection coating into the silicon bulk. Rohatgi explains that the procedure passivates (inactivates) crystal defects such as grain boundaries and lattice dislocations. Those kinds of defect sites are associated with so-called dangling silicon bonds (unsaturated valencies) that tend to promote electron-hole recombination. The recombination process quenches the photoexcitations and lowers the concentration of charge carriers.

"The key is to cut down the recombination of light-generated electron-hole pairs," Rohatgi asserts. "If you get rid of impurities and passivate defects, you can turn a bad material into a pretty good one." The idea is to do it without lengthening the processing time and introducing additional steps and costs, he emphasizes. That's why the Georgia Tech group couples the processing steps with a rapid thermal annealing treatment, which improves hydrogen passivation of defects. "It's a two-for-one deal," Rohatgi says. "We reduce the processing time and simultaneously get improved defect passivation."

The Georgia Tech center also focuses on computer modeling of photovoltaic devices. Researchers there use computational methods to optimize solar cell design parameters such as silicon resistivity and doping levels. And they also use computer methods to study the effect on cell performance of altering the thicknesses of various layers, including phosphorus diffusion layers and the blue silicon nitride antireflection coating commonly seen on solar cells.

Modeling efforts at Georgia Tech extend beyond single cells. Individual solar cells are connected--in series or in parallel, depending on voltage and current needs--to form modules, which in turn are wired together to form solar arrays. One of the world's largest rooftop photovoltaic array systems--one that's tied into a municipal power grid--was designed by Rohatgi and colleagues and is located on the university's campus. The roof of the aquatic center is covered with nearly three-quarters of an acre of solar panels. "It's a spectacular sight," Rohatgi comments. The system's performance is monitored continuously, and the data are compared with calculated values in an ongoing effort to improve photovoltaic modeling capabilities.

The 340-kW photovoltaic system sitting atop Georgia Tech's aquatic center has provided power to the building since the 1996 Olympic swimming competitions were held there. The giant array, which at the time of construction was the world's largest grid-connected rooftop system, consists of 2,856 1-meter 3 1-meter modules, each containing 72 multicrystalline silicon solar cells. According to Rohatgi, almost 30% of the power needed to run the swimming facility comes from the rooftop system. The eight-year-old cells are roughly 11% efficient. By comparison, Rohatgi, Vijay Yelundur, Brian Rounsaville, and their coworkers just recently reported laboratory-type, multicrystalline silicon cells that achieve on the order of 18% efficiency [Appl. Phys. Lett., 84, 145 (2004)].

REDUCING THE SIZE of semiconductor crystals to the nanometer scale can offer a number of advantages in photovoltaics compared with using larger crystal counterparts. For example, nanocrystals can be extremely sensitive to radiation and rather inexpensive to process. A major thrust in solar cell research nowadays, and the focus of Grätzel's research program in Lausanne, is developing efficient and long-lasting cells based on nanocrystalline oxide semiconductor films. These cells are sensitized to a broad spectrum of light by adsorbing a layer of dye molecules on the semiconductor surface.

Unlike conventional cells made from silicon, in dye-sensitized solar cells, light-absorption and charge-separation processes occur in separate molecular layers. Grätzel notes that separating the functions leads to simplified cell designs and reduces the need for ultra-high-purity semiconductors, which in turn lowers the cost of the cells. Another key difference is that in dye-sensitized cells, the semiconductor film is highly porous, and its pores are filled with an electrolyte or an organic hole conductor, leading to junctions with enormous contact area between interpenetrating materials.

As light shines on the thin-film cells, the sensitizer molecules, often members of the ruthenium-polypyridine family, undergo photoexcitations and inject electrons into the semiconducting oxide film--typically titanium dioxide (TiO₂, a negative-charge conductor). The dye molecules return to their original state via electron transfer from a redox material (a positive-charge conductor), which is then reduced at a counterelectrode.

Grätzel emphasizes that generating negative and positive charge carriers on opposite sides of the junction avoids the charge-carrier recombination problem encountered with silicon cells. He explains that recombination is impaired by the presence of the monolayer of dye molecules at the interface, which acts as an insulator. To date, the low-cost cells have proven to be greater than 10% efficient, making them comparable to some silicon solar cells.

But the story doesn't end there. It turns out that energy efficiency in dye-sensitized solar cells falls off significantly at temperatures above 80 °C. So Grätzel and coworkers designed a robust system consisting of an amphiphilic ruthenium dye and a polymer gel electrolyte that can take the heat. Last year, the group demonstrated that the new system maintains 94% of its initial efficiency value even after 1,000 hours at 80 °C [Nat. Mater., 2, 402 (2003)].

And just recently, the Lausanne team announced a way to avoid using volatile solvents to transport electrolytes (often the iodide/triiodide redox couple) through dye-sensitized solar cells. In place of a solvent-electrolyte system, which may leak or evaporate from large solar panels, Grätzel's research group made use of an imidazolium selenocyanate ionic liquid and measured conversion efficiencies on the order of 8% [J. Am. Chem. Soc., 126, 7164 (2004)].

Dye-sensitized solar cells aren't just a laboratory curiosity. Greatcell Solar in Switzerland manufactures the solar cells in the form of architectural panels for building facades and roofs. And at Konarka Technologies, Lowell, Mass., researchers are developing low-cost manufacturing processes to produce the cells on flexible materials. Unlike rigid and fragile glass- and silicon-based cells, Konarka's products are fabricated on rolls of plastic sheeting and can be prepared in all sorts of sizes and shapes and applied to many kinds of surfaces.

Konarka was founded to commercialize a low-temperature procedure for sintering TiO₂ nanoparticles. The process, which serves to bridge the particles electrically so that nanoporous TiO₂ can function as an electron carrier, is a key step in fabricating inexpensive dye-sensitized solar cells. Company representatives note that the military is interested in the technology because it could enable the surfaces of tents, vehicles, and buildings to produce the power needed to recharge batteries.

With nanometer-scale properties at the center of so many research efforts nowadays, it's not surprising that a large variety of nanostructured materials are being scrutinized for photovoltaic applications. At the University of California, Berkeley, for example, chemistry professor A. Paul Alivisatos and coworkers paired semiconducting nanorods of cadmium selenide (CdSe) with the conjugated polymer poly(3-hexylthiophene) to form novel solar cells with energy conversion efficiencies of roughly 2%. The group demonstrated that device characteristics such as charge-carrier mobility and absorption spectrum could be tailored by controlling the length and radius of the nanorods via synthesis techniques.

Meanwhile, at Los Alamos National Laboratory, Victor I. Klimov and Richard D. Schaller found that by controlling the conditions under which very small (less than 10 nm) lead selenide (PbSe) nanocrystals are irradiated, they could increase the number of electron-hole pairs generated per incident photon, which in turn would raise solar cell current and efficiency [Phys. Rev. Lett., 92, 186601 (2004)].

Other types of semiconductor materials also are being examined for use in photovoltaic systems. At Lawrence Berkeley National Laboratory (LBNL), Kin Man Yu, Wladek Walukiewicz, and their coworkers have been exploring ways of increasing the efficiency of solar cells by designing materials that are sensitive to a broad spectrum of sunlight.

SOLAR RADIATION includes infrared, visible, and ultraviolet light, and corresponds to photons with roughly 1–4 eV of energy. The range of wavelengths absorbed by a semiconductor in a solar cell is determined by the material's band gap--the energy required to promote an electron from the valence band to the conduction band. Photons with energies below the band gap are not absorbed. If the photon energy is greater than the band gap, the photon is absorbed, but much of the energy is wasted as heat.

Yu explains that by stacking materials with dissimilar band gaps, for example, Si (1.1-eV band gap) and AlGaAs (1.7 eV), it should be possible to broaden the range of wavelengths absorbed by a semiconductor and boost the efficiency of a solar cell above 35%. As luck would have it, however, those compounds and other pairs of dissimilar materials often resist being stacked because of differences in their lattice parameters. But based on an unexpected result from their studies of extremely pure nitride materials, Yu and Walukiewicz found an excellent candidate for stacking.

"We were very surprised to find that InN has a very low band gap--around 0.7 eV--not 2 eV as was assumed for the past 30 years," Yu remarks. Indium nitride can be alloyed relatively easily with GaN (3.4-eV band gap), and alloys of various compositions of indium, gallium, and nitrogen can be stacked readily to form multilayer (multijunction) materials. "The implication is that by varying the proportions of indium and gallium, we can make a whole range of alloy compositions and cover the entire solar spectrum," Yu asserts. The Berkeley team and other research groups are working on it.

At LBNL, they're also working on semiconductors with an altogether different type of band structure: a single-layer multigap material. For this investigation, the team is examining semiconductors that are highly mismatched in terms of electronegativity. Yu explains that when a few percent of the arsenic atoms in a III–V semiconductor such as GaAs are replaced with nitrogen, the difference in electronegativity causes a significant decrease in the material's band gap. Based on computational and experimental investigations, the group attributes the electronic behavior to a band-splitting process that creates high-energy and low-energy band gaps. In some cases, the splitting process results in three bands, according to Yu.

"We found that the phenomenon is quite general and applies to other materials," Yu remarks. For example, band splitting occurs when oxygen is substituted for group VI atoms in II–VI semiconductors such as ZnTe. The same is true of oxygen-doped ZnMnTe.

The studies highlight distinct approaches to a common research goal: increasing the efficiency of photovoltaic cells by designing novel semiconductors. In the case of multijunction materials, distinct segments of the solar spectrum are absorbed using separate materials. In multigap materials, a single material absorbs multiple wavelengths of the sun's light.

From solar-powered calculators to solar system satellites, photovoltaic technology has proven itself to be reliable, durable, and in some cases fairly efficient--but at high cost. Some industry experts predict that over the next 25 years, the output from this alternative energy source will increase 1,000-fold. But those projections are predicated on cost reductions in solar cell materials and manufacturing methods and significant advances in performance relative to today's systems.