by Katherine Derbyshire, Contributing Technical Editor, Solid State Technology
Cost per watt (CPW) is probably the most important metric in the solar industry. It defines an installation’s cost and the point at which the site breaks even. It provides a basis for comparison among photovoltaic technologies and between installations, as well as between solar energy and other sources of electricity. Though other factors contribute to sales, the system with the lowest cost per watt will receive careful consideration for any project.
To improve CPW, manufacturers can either reduce cost — by improving yield, increasing throughput, and pulling all the other levers available to a manufacturing process — or they can increase the wattage available from a given panel area. In turn, increasing output power requires improved conversion efficiency: the panel must capture more incident photons, convert more of them to free carriers, and deliver more of those carriers to the panel’s terminals. Each of these steps — capture, conversion, and transport — shaves points from the ultimate efficiency and offers opportunities for performance improvements.
The search for improved efficiency begins at the front, or sunward, surface of the cell, where the average incident solar radiation is about 1000 W/m2. Though this number is a convenient rule of thumb, it is only an average. At high altitudes or in space, less atmosphere blocks the sun’s light. In winter and in the morning and evening, the sun is lower in the sky and less energy reaches the panel. Differences in atmospheric absorption, whether due to location or to weather and pollution, change both the intensity and the spectrum of incident light hitting the panel. These differences affect the economic model for solar energy in fairly intuitive ways: Arizona and New Mexico have access to a much larger solar resource than Washington and Oregon. More subtly, however, atmospheric effects change the performance of particular cell designs.
Concentrators, multijunction cells capture more light
For example, concentrator photovoltaic (CPV) systems boost efficiency and reduce cost by using lenses to focus light onto the cells. [Not to be confused with thermal concentrating solar power (CSP) systems, which focus the sun's light to heat a reservoir of molten salt or another fluid, then use the heat to drive steam turbines.] CPV systems aim to reduce overall cost by reducing the amount of semiconductor needed. A panel incorporating 10× magnifying lenses can capture the same amount of light as a standard panel, but with one-tenth of the semiconductor area. Lenses are relatively inexpensive, while semiconductor material is the most expensive component of a panel. CPV systems can reduce cost, even if the size of the solar panel as a whole stays the same. Depending on the design, CPV lenses can range from 5× concentration to 500× or more (see Fig. 1).
Figure 1: Concentrating photovoltaic panel. (Courtesy of SolFocus)
Because of their reliance on lenses, however, CPV systems work best in direct sunlight. Power output drops sharply on overcast days, or even if a cloud passes over the sun. If the angle between the incident light and the lens changes, the lens will focus less light onto the center of the cell. Systems with concentrations of 10× or more use motors to track the sun across the sky, maintaining the correct relationship between the sun, the lens, and the semiconductor. Both the cost of these moving parts and the electricity needed to drive them increase the overall CPW of CPV systems. Moving parts add a potential reliability concern as well; conventional solar panels have none. High concentration systems may also need active cooling and high capacity heat sinks, which add further cost and complexity.
To make the most effective use of high photon fluxes supplied by the lens array, high magnification concentrator systems use high-efficiency multijunction solar cells based on GaAs and related III-V semiconductors. A single junction cell is most efficient when photon energy is near the cell’s bandgap. Photons with lower energy fail to excite free carriers and are lost. Photons with more energy than the bandgap dissipate the excess as heat. In multijunction cells, each junction is optimized for a different slice of the solar spectrum. Research-oriented III-V multijunction cells hold the record for solar conversion efficiency, at 42%. Commercial cells used in CPV receivers can achieve conversion efficiencies as high as 32%; the thermodynamic limit for a single-junction cell is about 31%.
III-V based multijunction cells are either deposited by MOCVD or MBE, or are assembled by mechanically stacking several monolithic cells. First used in space applications, they still occupy the high end of the market, where high performance commands a price premium. Thin-film silicon cells, in contrast, occupy the low end. They are generally deposited by hot-wire CVD, a low-cost, high-deposition-rate technology suited to the large glass panels used for these cells. Yet they too can benefit from multijunction technology. The best thin-film silicon cells incorporate amorphous silicon, microcrystalline silicon, and even SiGe, achieving efficiencies as high as 15%; single-junction amorphous silicon cells are less than 10% efficient.
Both concentrating lenses and multijunction cells, used together or separately, seek to increase output power by capturing a larger fraction of the incident photons. A third approach, Sanyo’s heterojunction with intrinsic thin layer (HIT) design broadens the very definition of available light. HIT cells, unique among cell designs, capture light from the back surface as well. These cells are made by depositing amorphous silicon on both sides of a crystalline silicon wafer. Where most cells have an opaque metallic contact layer, HIT cells have additional active area. Unlike other designs, they can exploit light reflected from the ground or roof under the panels. Commercial cells have demonstrated 20% efficiency, while research-class cells have reached 22%.
Even ordinary single surface, single junction cells seek to increase light capture. While the wafers used for ICs are polished to a mirror shine, the wafers used for solar cells are etched to texture the surface and increase light absorption. The most efficient crystalline silicon cells, from SunPower, move all conducting lines to the back side, where they do not shadow the active surface.
Companies with conventional front contact designs would like to reduce shadowing as well. To this end, they are examining ink-jet printing and other alternatives to screen-printed silver paste contacts. These approaches typically use non-contact printing to form the initial contact between metal and silicon, followed by a plating step to achieve the desired line thickness. Compared to screen printing, these methods offer narrower lines with higher aspect ratios, reducing both shadowing and line resistance. Narrow contact lines are especially important for advanced cell designs. If contact fingers are spaced more closely, and the junction itself is closer to the surface, then carriers have less distance to travel through silicon before reaching low-resistance metal. Resistive losses go down. Unless narrower lines are used, however, the efficiency gains from improved carrier transport are offset by the increase in shaded area.
Optimized cells make the most of their photons
Ample opportunities for efficiency improvements remain even after photons reach the cell’s active surface. To understand these, it’s helpful to break efficiency down into several independent components. Overall conversion efficiency (h) is defined as the ratio of the power striking the cells to the power delivered to the load. [The efficiency of an installation also depends on losses between the panel and the ultimate customer. Though these "balance of system" losses can be substantial, this article only considers the power measured at the terminals of the cell.] Power, in turn, is the product of voltage and current, V * I. In analyzing cell performance, engineers consider the open circuit voltage (Voc), the short circuit current (Isc) or current density (Jsc), and the ratio of the maximum power actually delivered to the theoretical maximum power (Voc * Isc). This last ratio is known as the fill factor (FF).
When photons with energy in excess of the bandgap strike a conventional inorganic solar cell, they excite free carriers in the form of electron-hole pairs. When these carriers encounter the p-n junction at the heart of the cell, the junction potential sweeps electrons to one side and holes to the other, from which point they diffuse to the nearest metallic contacts. Because of the photovoltaic effect, excess holes accumulate in the p-doped portion of the cell, and excess electrons in the n-doped portion. These excess carriers create a potential field, offsetting the built-in potential of the junction. When no load is attached — the open circuit condition — the potential across the cell (Voc) will be the difference between the light-generated bias and the built-in junction bias. [In so-called third-generation cells, such as dye-sensitized cells, the carrier generation and transport mechanisms are different and the chemical potential plays an important role. Interested readers should see reference .]
Not all carriers excited in the bulk actually reach the junction. Many of them recombine along the way and are lost. To reach the junction, the carrier’s recombination lifetime must exceed the necessary transport time. While solar cell performance, like IC performance, depends on silicon quality, the important quality metrics are slightly different. Point defects and impurities are less critical for solar cells than for ICs. Solar cells have large features, so a single defect is unlikely to cripple a cell as it might a transistor or an IC. Still, carrier mobility and lifetime are critical performance parameters.
Silicon quality has become especially important in recent years, as the solar industry’s rapid growth has strained polysilicon supplies. Manufacturers would like to use so-called upgraded metallurgical grade (UMG) silicon to reduce their cost. As the name implies, the quality of UMG silicon lies somewhere between electronics-grade and metallurgical-grade material. It might be melted in a crucible with electronics-grade silicon, producing intermediate-grade wafers, or might serve as a substrate for deposition of a higher quality epitaxial layer. So far, however, no consensus specification for this mid-grade silicon has emerged.
Because grain boundaries act as recombination centers, crystal quality affects efficiency as well. Cz-grown, single-crystal silicon wafers make the most efficient silicon solar cells, amorphous silicon cells on glass are the least efficient. In between, manufacturers try to balance crystal quality against manufacturing cost. Thin-film silicon cells — both amorphous and polycrystalline — incorporate substantial amounts of dissolved hydrogen. Hydrogen helps to passivate grain boundaries and other defect sites, improving Voc. In amorphous silicon, it appears to reduce porosity, reducing surface area and therefore surface recombination.
Recombination losses also depend on the spectral response of the cell. Lower energy photons are absorbed near the front surface, while those with more energy travel further. Since recombination rates change over the thickness of the cell — due to surface defects, for example — the amount of energy collected depends on the placement of the junction relative to the front surface.
Figure 2: a-Si/μ-SI tandem thin-film solar cell. (Courtesy of ULVAC Inc.)
Voc has only a slight dependence on light intensity. The open circuit potential is an equilibrium value whenever light is shining on the cell. Light intensity dependence is reflected in the other important cell parameter, the short circuit current (Isc). As the name implies, this is the current that flows when the cell is short circuited, meaning that the voltage across it is zero. The more photons strike the cell, the more current it produces. To allow comparisons between cells with different areas, short circuit current density (Jsc) is generally used instead of Isc. However, the overall conversion efficiency is calculated based on Isc.
Optical losses are the primary reason for reduced Jsc. One of the most important of these is absorption in the glass layer protecting the cell from the elements. Low-iron glass transmits about 15% more sunlight than standard sheet glass. However, manufacturing low-iron glass requires special oven cleaning and a higher melting temperature. Until recently, solar industry demand was too small to support dedicated low-iron glass factories. The material was a specialty product carrying comparatively high prices. An NREL study estimated that capacity in excess of 2GW/year would justify construction of a dedicated glass plant, cutting glass costs from more than $23.62/m2 to under $5.00/m2. The largest solar cell companies expect to reach 1GW or more of capacity in the few years. According to SolarBuzz, a market analysis firm, world solar cell production reached 3.4GW in 2007.
Further optical losses occur when light passes through the cell without being absorbed. Transmission losses were not a significant concern until recently: the silicon wafer could be as thick as needed to maximize light absorption. When the current solar boom brought a serious shortage of polysilicon, manufacturers were forced to reduce wafer thickness to keep costs under control. According to Erik Saver of REC, speaking at this spring’s IEEE Photovoltaic Specialists’ Conference (San Diego), manufacturers are extracting 50% more wafers per ingot than they were in 2005. Thinner wafers, in turn, have brought a new appreciation of the role of the back surface in both light capture and recombination. Traditionally, the back contact has been a screen-printed coating of aluminum or silver paste, heat-treated to ensure contact with the underlying silicon. More advanced designs replace this with a dielectric passivation coating, reducing back-surface recombination. Though a coat of conductive paste still follows, laser-firing is used to make contact with the silicon in only a few clearly defined locations. Over the rest of the cell, the dielectric/metal bilayer serves as an excellent optical mirror, reflecting unabsorbed light back into the cell.
Thin film cell designs must pay particular attention to light trapping and transmission losses. In multi-junction silicon structures, the interfaces between layers can act as both reflectors and recombination centers. Substantial research effort is focused on optimizing these interfaces. In CdTe (cadmium telluride) and CIGS (copper-indium-gallium-selenide) cells, a distinct absorber layer contributes to light capture. Most thin-film cells are deposited on glass or plastic, using a conductive oxide layer as a back contact. The optical properties of this layer define the cell’s light capture.
The theoretical maximum power of a cell is the product of Voc and Isc. However, by definition, the maximum current and voltage cannot occur simultaneously. Voc is the voltage when current is zero, and Isc is the current when voltage is zero. The third performance parameter, fill factor, is the ratio of the maximum power actually obtained to the theoretical maximum. It is a measure of the ideality of the cell, or the “squareness” of the I-V characteristic. As such, fill factor is degraded by resistive losses within the cell. Leakage across the junction (or junctions), resistive losses at the front and back contacts, and resistive losses at series connections to other cells all reduce output power and fill factor.
One of the challenges cell designers face is the need to improve voltage, current, and fill factor simultaneously. Surface texturing, for example, can introduce surface recombination sites at the same time that it improves light capture. Replacing a backside conductive film with a passivating dielectric can reduce recombination, but increase contact resistance.
Finally, it’s important to remember that efficiency is less important than CPW. In the IC market, more capable chips can earn a substantial premium, justifying the expensive leading-edge processes used to manufacture them. All solar cells, in contrast, produce the same end product. Electricity from a III-V multijunction concentrator photovoltaic system is identical to electricity from a low-end amorphous silicon panel. To succeed, a design must be not only efficient, but cost effective. At the moment, the most commercially successful designs seem to be those that combine moderate efficiency with low or moderate cost. – K.D.
The basic physics of solar cells and solar cell efficiency are discussed at length in most introductory texts. A good Web-based resource is http://pvcdrom.pveducation.org/.
 Brian A. Gregg and Mark C. Hanna, “Comparing organic to inorganic photovoltaic cells: Theory, experiment, and simulation,” J. Applied Physics, vol. 93, pp 3605-3614 (2003).
 B. Yan, et. al., “Correlation of Material Properties and Open-Circuit Voltage of Amorphous Silicon Based Solar Cells,” MRS Symp. Proc., vol. 762, paper A7.4 (2003).
 M. S. Keshner and R. Arya, “Study of Potential Cost Reductions Resulting from Super-Large-Scale Manufacturing of PV Modules,” NREL Subcontract Report SR-520-36846 (2004). http://www.nrel.gov/docs/fy05osti/36846.pdf
In 2001, Katherine Derbyshire founded Thin Film Manufacturing, a consulting firm helping the industry manage the interaction between business forces and technology advances. Previously, she was Managing Editor of Semiconductor Online from 1998 to 2001, and Chief Technical Editor of Solid State Technology from 1994 to 1998. She has engineering degrees from the Massachusetts Institute of Technology and the University of California, Santa Barbara. Thin Film Manufacturing, PO Box 82441, Kenmore, WA 98028, Email: firstname.lastname@example.org, http://www.thinfilmmfg.com.