电子工程师手册（英文版三）：CH060.PDF

Ramakumar, R., Barnett, A.M., Kazmerski, L.L., Benner, J.P., Coutts, T.J. “Power Systems and Generation” The Electrical Engineering Handbook Ed. Richard C. Dorf Boca Raton: CRC Press LLC, 2000 60 Power Systems and Generation 60.1 Distributed Power Generation Photovoltaics ? Wind-Electric Conversion ? Hydro ? Geothermal ? Tidal Energy ? Fuel Cells ? Solar-Thermal-Electric Conversion ? Biomass Energy ? Thermoelectrics ? Thermionics ? Integrated System Concepts ? System Impacts 60.2 Photovoltaic Solar Cells Solar Cell Operation and Characteristics ? Solar Cell Types and Their Optimization ? Crystalline Silicon ? III-V Semiconductors ? Thin-Film Solar Cells ? Dye-Sensitized Cells ? Module Technologies ? Photovoltaic Power Systems 60.3 Thermophotovoltaics Background ? Design Considerations of a TPV System ? Optical Control of Sub-bandgap Energies ? Development of PV Cells ? Status of System Development ? Systems and Applications 60.1 Distributed Power Generation Distributed generation (DG) refers to small (a few watts up to 1 MW) power plants at or near the loads, operating in a stand-alone mode or connected to a grid at the distribution or subtransmission level, and geographically scattered throughout the service area. Typically they harness unconventional energy resources such as insolation, wind, biomass, tides and waves, and geothermal. Small plants powered by site-specific conventional energy resources such as low-head and small hydro and natural gas are also included in this general group. Interest in DG has been growing steadily since the dramatic oil embargo of 1973. In addition to the obvious advantages realized by the development of renewable energy sources, DG is ideally suited to power small remote loads, located far from the grid. An entire family of small power sources has been developed and employed for space, underwater, and biomedical applications. Another niche for these systems is in energizing remote rural areas of developing countries. It is estimated that there are more than one million remote villages in the world with no grid connection and minimally sustained by locally available energy sources. Integrated renewable energy systems (IRES), a special subset of DG, are ideally suited for these situations. General Features DG will have one or more of the following features: ? Small size ? Intermittent input resource ? Stand-alone or interface at the distribution or subtransmission level ? Extremely site-specific inputs ? Located near the loads R. Ramakumar Oklahoma State University Allen M. Barnett AstroPower, Inc. Lawrence L. Kazmerski National Renewable Energy Laboratory John P. Benner National Renewable Energy Laboratory Timothy J. Coutts National Renewable Energy Laboratory ? 2000 by CRC Press LLC ? Remoteness from conventional grid supply ? Availability of energy storage and reconversion for later use Potential and Future Globally, the potential for DG is vast. Even extremely site-specific resources such as tides, geothermal, and small hydro are available in significant quantities. Assessments of the future for various DG technologies vary, depending on the enthusiasm of the estimator. However, in almost all cases, the limitations are economic rather than technical. Concerns over the unrestricted use of depletable energy resources and the ensuing environmental problems such as the greenhouse effect and global warming are providing the impetus necessary for the continued development of technologies for DG. Motivation Among the powerful motivations for the entry of DG are: ? Less capital investment and less capital at risk in the case of smaller installations ? Easier to site smaller plants under the ever-increasing restrictions ? Likely to result in improved reliability and availability ? Location near load centers decreases delivery costs and lowers transmission and distribution losses ? In terms of the cost of power delivered, DG is becoming competitive with large central-station plants, especially with the advent of open access and competition in the electric utility industry DG Technologies Many technologies have been proposed and employed for DG. Power ratings of DG systems vary from milliwatts to megawatts, depending on the application. A listing of the technologies is given below. ? Photovoltaics (PV) ? Wind-electric conversion systems ? Mini and micro hydro ? Geothermal plants ? Tidal and wave energy conversion ? Fuel cells ? Solar-thermal-electric conversion ? Biomass utilization ? Thermoelectrics ? Thermionics ? Small cogeneration plants powered by natural gas and supplying electrical and thermal energies The technology involved in the last item above is mature and very similar to that of conventional thermal power plants and therefore will not be considered in this section. Photovoltaics PV refers to the direct conversion of insolation (incident solar radiation) to electricity. A PV cell (also known as a solar cell) is simply a large-area semiconductor pn junction diode with the junction positioned very close to the top surface. Typically, a metallic grid structure on the top and a sheet structure in the bottom collect the minority carriers crossing the junction and serve as terminals. The minority carriers are generated by the incident photons with energies greater than or equal to the energy gap of the semiconductor material. Since the output of an individual cell is rather low (1 or 2 W at a fraction of a volt), several (30 to 60) cells are combined to form a module. Typical module ratings range from 40 to 50 W at 15 to 17 V. PV modules are progressively put together to form panels, arrays (strings or trackers), groups, segments (subfields), and ulti- mately a PV plant consisting of several segments. Plants rated at several MW have been built and operated successfully. ? 2000 by CRC Press LLC Advantages of PV include demonstrated low operation and maintenance costs, no moving parts, silent and simple operation, almost unlimited lifetime if properly cared for, no recurring fuel costs, modularity, and minimal environmental effects. The disadvantages are its cost, need for large collector areas due to the diluteness of insolation, and the diurnal and seasonal variability of the output. PV systems can be flat-plate or concentrating type. While flat-plate systems utilize the global (direct and diffuse) radiation, concentrator systems harness only the direct or beam radiation. As such, concentrating systems must track (one axis or two axis) the sun. Flat-plate systems may or may not be mounted on trackers. By 1990, efficiencies of flat-plate crystalline and thin-film cells had reached 23 and 15%, respectively. Efficiencies as high as 34% were recorded for concentrator cells. Single-crystal and amorphous PV module efficiencies of 12 and 5% were achieved by the early ’90s. For an average module efficiency of 10% and an insolation of 1 kW/m 2 on a clear afternoon, 10 m 2 of collector area is required for each kW of output. The output of a PV system is dc and inversion is required for supplying ac loads or for utility-interactive operation. While the required fuel input to a conventional power plant depends on its output, the input to a PV system is determined by external factors such as cloud cover, time of day, season of the year, geographic location, orientation, and geometry of the collector. Therefore, PV systems are operated, as far as possible, at or near their maximum outputs. Also, PV plants have inertialess generation and are subject to rapid changes in their outputs due to moving clouds. The current-voltage (IV) characteristic of an illuminated solar cell is shown in Figure 60.1. It is given as where I o and I s are the dark and source currents, respectively, k is the Boltzmann constant (1.38 × 10 –23 J/K), T is the temperature in K, and e is the electronic charge. Under ideal conditions (identical cells), for a PV module with a series-parallel arrangement of cells, the IV characteristic will be similar, except that the current scale should be multiplied by the number of parallel branches and the voltage scale by the number of cells in series in the module. The source current varies linearly with insolation. The dark current increases as the cell FIGURE 60.1 Typical current-voltage characteristic of an illuminated solar cell. III eV kT so =? ? ? ? ? ? ? ? ? ? ? ? ? ? exp 1 ? 2000 by CRC Press LLC operating temperature increases. Also, the larger the energy gap of the material, the smaller the dark current. The ratio of source current to dark current should be made as large as possible for improved operation. Single-crystal silicon is still the dominant technology for fabricating PV devices. Polycrystalline, semicrys- talline, and amorphous silicon technologies are developing rapidly to challenge this. Highly innovative tech- nologies such as spheral cells are being introduced to reduce costs. Concentrator systems typically employ gallium arsenide or multiple junction cells. Many other materials and thin-film technologies are under inves- tigation as potential candidates. PV applications range from milliwatts (consumer electronics) to megawatts (central station plants). They are suitable for portable, remote, stand-alone, and utility-interactive applications. PV systems should be con- sidered as energy sources and their design should maximize the conversion of insolation into useable electrical form. Power requirements of practical loads are met using an energy storage and reconversion system or utility interconnection. Concentrating systems have been designed and operated to provide both electrical and low- grade thermal outputs with combined peak utilization efficiencies approaching 60%. The vigorous growth of PV technology is manifested by a doubling of world PV module shipments in six years — from 42 MW in 1989 to 84 MW in 1995. Tens of thousands of small (<1 kW) systems are in operation around the world. Thousands of kilowatt-size systems (1 to 10s of kW) also have been installed and are in operation. Many intermediate-scale systems (10 to 100s of kW) and large-scale systems (1 MW or larger) are being installed by utility- and government-sponsored programs as proof-of-concept experiments and to glean valuable operational data. By 1988, nearly 11 MW of PV was interconnected to the utility system in the United States alone. Most were in the 1- to 5-kW range. The two major exceptions are the 1-MW Hesperia-Lugo project installed in 1982 and the 6.5-MW Carrisa Plains project installed in 1984, both in California. In Germany, a 340-kW system began operation in 1988 as part of a large program. Switzerland had a plan to install 1 MW of PV in 333 roof-mounted units of 3 kW each. By 1990, the installed capacity of PV in Italy exceeded 3 MW. Many nations have recognized the vast potential of PV and have established their own PV programs within the past decade. A view of the 300 kW flat-plate grid-connected PV system installed and operated by the city of Austin electric utility depart- ment in Austin, Texas is shown in Figure 60.2. From a capital cost of $7000/kW in 1988 with an associated levelized energy cost of 32￠/kWh, even with a business-as-usual scenario, a twofold reduction to $3500/kW by 2000 and an additional 3-to-1 reduction to $1175/kW by 2030 are being projected. The corresponding energy costs are 15 and 5￠/kWh, respectively. These FIGURE 60.2 A view of the city of Austin PV-300 flat-plate grid-connected photovoltaic system. (Courtesy of the city of Austin electric utility department.) ? 2000 by CRC Press LLC estimates put the cost of energy from PV in par with the cost of energy from conventional plants in the early part of the twenty-first century. Wind-Electric Conversion Wind energy is intermittent, highly variable, and site-specific, exists in three dimensions, and is the least dependent upon latitude among all renewable resources. The power density (in W/unit area) in moving air (wind) is a cubic function of wind speed and therefore even small increases in average wind speeds can lead to significant increases in the capturable energy. Wind sites are typically classified as good, excellent, or outstanding, with associated mean wind speeds of 13, 16, and 19 mph, respectively. Aeroturbines employ lift and/or drag forces to convert wind energy to rotary mechanical energy, which is then converted to electrical energy by coupling a suitable generator. The power coefficient C p of an aeroturbine is the fraction of the incident power converted to mechanical shaft power, and it is a function of the tip speed- to-wind speed ratio λ as shown in Figure 60.3. For a given propeller configuration, at any given wind speed, there is an optimum tip speed that maximizes C p . Several types of aeroturbines are available. They can have horizontal or vertical axes, number of blades ranging from one to several, mounted upwind or downwind, and fixed- or variable-pitch blades with full blade control or tip control. Vertical-axis (Darrieus) turbines are not self-starting and require a starting mechanism. Today, horizontal-axis turbines with two or more blades are the most prevalent, and considerable work is underway to develop advanced versions of these. The electrical output P e of a wind-electric conversion system (WECS) is given as P e = η g η m AC p Kv 3 where η g and η m are the efficiencies of the electrical generator and mechanical interface, respectively, A is the swept area, K is a constant, and v is the wind speed incident on the aeroturbine. There are two basic options for wind-electric conversion. With varying wind speeds, the aeroturbine can be operated at a constant speed by blade-pitch control, and a conventional synchronous machine is then employed to generate constant-frequency ac. More commonly, an induction generator is used with or without an adjust- able var supply. In this case, the aeroturbine will operate at a nearly constant speed. Alternatively, the aeroturbine rotational speed can be allowed to vary with wind to maintain a constant and optimum tip speed ratio, and then a combination of special energy converters and power electronics is employed to obtain utility-grade ac. FIGURE 60.3 Typical aeroturbine characteristics. ? 2000 by CRC Press LLC The variable-speed option allows optimum efficiency operation of the turbine over a wide range of wind speeds, resulting in increased outputs with lower structural loads and stresses. All future utility-grade advanced turbines are expected to operate in the variable-speed mode and use power electronics to convert the variable-frequency output to constant frequency with minimal harmonic distortion. Large-scale harnessing of wind energy will require hundreds or even thousands of WECS arranged in a wind farm with spacings of about 2 to 3 diameters crosswind and about 10 diameters apart downwind. The power output of an individual WECS will fluctuate over a wide range, and its statistics strongly depend on the wind statistics. When many WECS are used in a wind farm, some smoothing of the total power output will result, depending on the statistical independence of the outputs of individual WECS. This is desirable, especially with high (>20%) penetration of WECS in the generation mix. While the output of WECS is not dispatchable, with large wind farms the possibility of assigning some capacity credit to the overall output significantly improves. Although wind-electric conversion has overall minimum environmental impacts, the large rotating structures involved do generate some noise and introduce visual aesthetics problems. By locating wind energy systems sufficiently far from centers of population, these effects can be minimized. The envisaged potential for bird kills turned out to be not a serious problem. Wind energy systems occupy only a very small fraction of the land. However, the area surrounding them can be used only for activities such as farming and livestock grazing. Thus, there is some negative impact on land use. Today, the cost of energy delivered by wind plants rivals those obtained from some nonrenewable sources. By 1990, wind became the most utilized and competitive option among all the solar energy technologies for the bulk power market at a cost of generation of about 8￠/kWh (or roughly 7￠/kWh in 1987 dollars). Ongoing research and development work in new design tools, advanced airfoils, site tailoring, operating strategies, array spacing, and improved reliability and manufacturability is expected to bring the cost of energy further down by a factor of 2 to 3. At around 1600 MW, nearly 90% of all the WECS installed in the world are in California. They are expected to generate nearly 3 billion kWh of electricity per year to the state’s utilities to which they are interconnected. Although their lack of control and the intermittent nature of wind-derived energy are not embraced enthusiastically by electric utilities, this gap is expected to be bridged very soon with appropriate computer controls and operating strategies. Wind energy is already an economical option for remote areas endowed with good wind regimes. The modularity of WECS, coupled with the associated environmental benefits, potential for providing jobs, and economic viability point to a major role for wind energy in the generation mix of the world in the decades to come. Hydro Hydropower is a mature but neglected and one of the most promising renewable energy technologies. In the context of DG, small (less than 15 MW), mini (less than 1 MW), and micro (less than 100 kW) hydroelectric plants are of interest. The source of hydropower is the hydrologic cycle driven by the energy from the sun. Most of the sites for DG hydro are either low-head (2 to 20 m) or medium-head (20 to 150 m). The global hydroelectric potential is vast. One estimate puts it at 31 GW for Indonesia alone! The installed capacity of small hydro in the People’s Republic of China was exceeding 7 GW by 1980. Both impulse and reaction turbines have been employed for small-scale hydro for DG. Several standardized units are available in the market. Most of the units are operated at constant speed with governor control and are coupled to synchronous machines to generate utility-grade ac. If the water source is highly variable, it may be necessary to employ variable-speed operation. If the speed variations are not large, induction generators can be used. Special variable-speed constant-frequency (VSCF) generation schemes may be needed if the range of speed variations is large (> ±10%). Permanent magnet generators provide another alternative, especially if the output is to be rectified and stored for later use in the case of very small units. Geothermal Geothermal plants exploit the heat stored in the form of hot water and steam in the earth’s crust at depths of 2000 to 8000 ft. By nature, these resources are extremely site-specific and slowly run down (depletable) over a period of years. For electric power generation, the resource should be at least around 250°C. Depending on the temperature and makeup, dry steam, flash steam, or binary technology can be employed. Of these, dry natural steam is the best since it eliminates the need for a boiler. ? 2000 by CRC Press LLC The three basic components of a geothermal plant are (1) a production well to bring the resource to the surface, (2) a turbine generator system for energy conversion, and (3) an injection well to recycle the spent geothermal fluids back into the reservoir. Worldwide deployment of geothermal plants reached 5000 MW by 1987 in 17 countries. Nearly one-half of this was in the United States. The Geysers plant north of San Francisco is the largest in the world with an installed capacity of 516 MW. In some developing countries, the Philippines for example, geothermal plants supply nearly 20% of their electrical needs. Tidal Energy The origin of tidal energy is the upward-acting gravitational force of the moon, which results in a cyclic variation in the potential energy of water at a point on the earth’s surface. These variations are amplified by topographical features such as the shape and size of estuaries. The ratio between maximum spring tide and minimum at neap can be as much as 3 to 1. In estuaries, the tidal range can be as large as 10 to 15 m. Power can be generated from a tidal estuary in two basic ways. A single basin can be used with a barrage at a strategic point along the estuary. By installing turbines at this point, electricity can be generated both when the tide is ebbing or flooding. In the two-basin scheme, generation can be time-shifted to coincide with hours of peak demand by using the basins alternately. As can be expected, tidal energy conversion is very site-specific. The largest tidal power plant is the single- basin scheme at La Rance in Brittany, France. It is rated at 240 MW and employs 24 vane-type horizontal turbines and alternator motors, each rated at 10 MVA. The plant has been in operation since 1966 with good technical and economic results. It has generated, on the average, around 500 GWh of net energy per year. The Severn estuary in the southwest of England and the Bay of Fundy in the border between the United States and Canada with the highest known tidal range of 17 m have been extensively studied for tidal power generation. There are several other possible sites around the world, but the massive capital costs required have delayed their exploitation. Fuel Cells A fuel cell is a simple static device that converts the chemical energy in a fuel directly, isothermally, and continuously into electrical energy. Fuel and oxidant (typically oxygen in air) are fed to the device in which an electrochemical reaction takes place that oxidizes the fuel, reduces the oxidant, and releases energy. The energy released is in both electrical and thermal forms. The electrical part provides the required output. Since a fuel cell completely bypasses the thermal-to-mechanical conversion involved in a conventional power plant and since its operation is isothermal, fuel cells are not Carnot-limited. Efficiencies in the range of 43 to 55% are forecasted for modular dispersed generators featuring fuel cells. The low (< 0.05 lb/MWh) airborne emissions of fuel cell plants make them prime candidates for siting in urban areas. The possibility of using fuel cells in combined heat and power (CHP) units provides the cleanest and most efficient energy system option utilizing valuable (or imported) natural gas resources. Hydrocarbon fuel (natural gas or LNG) or gasified coal is reformed first to produce hydrogen-rich (and sulphur-free) gas that enters the fuel cell stack where it is electrochemically “burned” to produce electrical and thermal outputs. The electrical output of a fuel cell is low-voltage high-current dc. By utilizing a properly organized stack of cells and an inverter, utility-grade ac output is obtained. Early MW-scale demonstration plants employed phosphoric acid fuel cells. Molten carbonate fuel cell systems have shown considerable promise in recent years with demonstrated efficiencies in the 50 to 55% range based on the higher heating value. Another competitor in the long range is the solid oxide fuel cell that can be intergrated with a coal gasifier and a steam bottoming cycle. Solar-Thermal-Electric Conversion The quality of thermal energy needed for DG employing solar-thermal-electric conversion necessitates concen- trated sunlight. Parabolic troughs, parabolic dishes, and central receivers are used to generate temperatures in the range of 400 to 500, 800 to 900, and >500°C, respectively. Technical feasibility of the central receiver system was demonstrated in the early ’80s by the 10-MWe Solar One system in Barstow, California. Over a six-year period, this system delivered 37 GWh of net energy to the ? 2000 by CRC Press LLC Southern California Edison’s grid with an overall system efficiency in the range of 7 to 8%. With improvements in heliostat and receiver technologies, annual system efficiencies of 14 to 15% and generation cost of 8 to 12￠/kWh have been projected. Parabolic-dish electric-transport technology for DG was under active development at the Jet Propulsion Laboratory (JPL) in Pasadena, California, in the late ’70s and early ’80s. Prototype modules with Stirling engines reached a record 29% overall efficiency of conversion from insolation to electrical output. Earlier parabolic- dish designs collected and transported thermal energy to a central location for conversion to electricity. Advanced designs such as the one developed at JPL employed engine driven generators at the focal points of the dishes, and energy was collected and transported in electrical form. By far the largest installed capacity (nearly 400 MW) of solar-thermal-electric DG employs parabolic-trough collectors and oil to transport the thermal energy to a central location for conversion to electricity via a steam- Rankine cycle. With the addition of a natural gas burner for hybrid operation, this technology, developed by LUZ under the code name SEGS (solar electric generating system), accounts for more than 90% of the world’s solar electric capacity, all located in Daggett, Kramer Junction, and Harper Lake in California. Generation costs of around 8 to 9￠/kWh have been realized with SEGS. This technology uses natural gas to compensate for the temporal variations of insolation and firms up the power delivered by the system. This compensation may come during 7 to 11 P.M. in summer and during 8 A.M. to 5 P.M. in winter. SEGS will require about 5 acres/MW or can deliver 130 MW/mi 2 of land area. Biomass Energy Biological sources provide a wide array of materials that have been and continue to be used as energy sources. Wood, wood wastes, and residue from wood processing industries, sewage or municipal solid waste, cultivated herbaceous and other energy crops, waste from food processing industries, and animal wastes are lumped together by the term biomass. The most compelling argument for the use of biomass technologies is the inherent recycling of the carbon by photosynthesis. In addition to the obvious method of burning biomass, conversion to liquid and gaseous fuels is possible, thus expanding the application possibilities. In the context of electric power generation, the role of biomass is expected to be for repowering old units and for use in small (20 to 50 MW) new plants. Several new high-efficiency conversion technologies are either already available or under development for the utilization of biomass. The technologies and their overall conversion efficiencies are listed below. ? FBC (fluidized-bed combustor), 36–38% ? EPS (energy performance system) combustor, 34–36% ? BIG/STIG (biomass-integrated gasifier/steam-injected gas turbine), 38–47% Acid or enzymatic hydrolysis, gasification, and aqueous pyrolysis are some of the other technology options available for biomass utilization. Anaerobic digestion of animal wastes is being used extensively in developing countries to produce biogas, which is utilized directly as a fuel in burners and for lighting. An 80–20 mixture of biogas and diesel has been used effectively in biogas engines to generate electricity in small quantities. Biomass-fueled power plants are best suited in small (<100 MW) sizes for DG to serve base load and intermediate loads in the eastern United States and in many other parts of the world. This contribution is clean, renewable, and reduces CO 2 emissions. Since biomass fuels are sulphur-free, these plants can be used to offset CO 2 and SO 2 emissions from new fossil power plants. Ash from biomass plants can be recycled and used as fertilizer. A carefully planned and well-managed SRWC (short-rotation woody crop) plantation program with yields in the range of 6 to 12 dry tons/acre/year can be effectively used to mitigate greenhouse gases and contribute thousands of MW of DG to the U.S. grid by the turn of the century. Thermoelectrics Thermal energy can be directly converted to electrical energy by using the thermoelectric effects in materials. Semiconductors offer the best option as thermocouples since thermojunctions can be constructed using a p-type ? 2000 by CRC Press LLC and an n-type material to cumulate the effects around a thermoelectric circuit. Moreover, by using solid solutions of tellurides and selenides doped to result in a low density of charge carriers, relatively moderate thermal conductivities and reasonably good electrical conductivities can be achieved. In a thermoelectric generator, the Seebeck voltage generated under a temperature difference drives a dc current through the load circuit. Even though there is no mechanical conversion, the process is still Carnot- limited since it operates over a temperature difference. In practice, several couples are assembled in a series- parallel configuration to provide dc output power at the required voltage. Typical thermoelectric generators employ radioisotope or nuclear reactor or hydrocarbon burner as the heat source. They are custom-made for space missions as exemplified by the SNAP (systems for nuclear auxiliary power) series and the RTG (radioisotope thermoelectric generator) used by the Apollo astronauts. Maximum performance over a large temperature range is achieved by cascading stages. Each stage consists of thermocouples electrically in series and thermally in parallel. The stages themselves are thermally in series and electrically in parallel. Tellurides and selenides are used for power generation up to 600°C. Silicon germanium alloys turn out better performance above this up to 1000°C. With the materials available at present, conversion efficiencies in the 5 to 10% range can be expected. Whenever small amounts of silent reliable power is needed for long periods of time, thermoelectrics offer a viable option. Space, underwater, biomedical, and remote terrestrial power such as cathodic protection of pipelines fall into this category. Thermionics Direct conversion of thermal energy into electrical energy can be achieved by employing the Edison effect—the release of electrons from a hot body, also known as thermionic emission. The thermal input imparts sufficient energy (≥ work function) to a few electrons in the emitter (cathode), which helps them escape. If these electrons are collected using a collector (anode) and a closed path through a load is established for them to complete the circuit back to the cathode, then electrical output is obtained. Thermionic converters are heat engines with electrons as the working fluid and, as such, are subject to Carnot limitations. Converters filled with ionizable gases such as cesium vapor in the interelectrode space yield higher power densities due to space charge neutralization. Barrier index is a parameter that signifies the closeness to ideal performance with no space charge effects. As this index is reduced, more applications become feasible. A typical example of developments in thermionics is the TFE (thermionic fuel element) that integrates the converter and nuclear fuel for space nuclear power in the kW to MW level for very long (7 to 10 years) duration missions. Another niche is the thermionic cogeneration burner module, a high-temperature burner equipped with thermionic converters. Electrical outputs of 50 kW/MW of thermal output have been achieved. High (600 to 650°C) heat rejection temperatures of thermionic converters are ideally suited for producing flue gas in the 500 to 550°C range for industrial processes. A long-range goal is to use thermionic converters as toppers for conventional power plants. Such concepts are not economical at present. Integrated System Concepts DG technologies offer many possibilities for integrated operation. Integrated systems may be stand-alone with energy storage and reconversion or include grid connection. Also, both renewable and conventional systems can be integrated to achieve the required operational characteristics. Integrated renewable energy systems (IRES) that harness several manifestations of solar energy to supply a variety of energy and other needs have many advantages and applications worldwide. The complementary nature of some of the resources (insolation and wind, for example) over the annual cycle can be exploited by IRES to decrease the amount of energy storage necessary and lower the overall cost of energy. System Impacts Response of distribution systems to high penetrations of DG is not yet fully understood. Also, the nature of the response will depend on the DG technology involved. However, there are some general areas of potential impacts common to most of the technologies: (i) voltage flicker, imbalance, regulation, etc.; (ii) power quality; ? 2000 by CRC Press LLC (iii) real and reactive power flow modifications; (iv) islanding; (v) synchronization during system restoration; (vi) transients; (vii) protection issues; (viii) load following capability; and (ix) dynamic interation with the rest of the system. Since there are very few systems with high penetration of DG, studies based on detailed models should be undertaken to forecast potential problems and arrive at suitable solutions. Defining Terms Biomass: General term used for wood, wood wastes, sewage, cultivated herbaceous and other energy crops, and animal wastes. Distributed generation: Small power plants at or near loads and scattered throughout the service area. Fuel cell: Device that converts the chemical energy in a fuel directly and isothermally into electrical energy. Geothermal energy: Thermal energy in the form of hot water and steam in the earth’s crust. Hydropower: Conversion of potential energy of water into electricity using generators coupled to impulse or reaction water turbines. Insolation: Incident solar radiation. IRES: Acronym for integrated renewable energy system, a collection of devices that harness several manifes- tations of solar energy to supply a variety of energy and other needs. Photovoltaics: Conversion of insolation into dc electricity by means of solid state pn junction diodes. Solar-thermal-electric conversion: Collection of solar energy in thermal form using flat-plate or concentrat- ing collectors and its conversion to electrical form. Thermionics: Direct conversion of thermal energy into electrical energy by using the Edison effect (thermi- onic emission). Thermoelectrics: Direct conversion of thermal energy into electrical energy using the thermoelectric effects in materials, typically semiconductors. Tidal energy: The energy contained in the varying water level in oceans and estuaries, originated by lunar gravitational force. Wind-electric conversion: The generation of electrical energy using electromechanical energy converters driven by aeroturbines. Related Topic 22.1 Physical Properties References S.W. Angrist, Direct Energy Conversion, 4th ed., Boston, Mass.: Allyn and Bacon, 1982. R. C. Dorf, Energy, Resources, & Policy, Reading, Mass.: Addison-Wesley, 1978. J. J. Fritz, Small and Mini Hydropower Systems, New York: McGraw-Hill, 1984. J. F. Kreider and F. Kreith (eds.), Solar Energy Handbook, New York: McGraw-Hill, 1981. T. Moore, “On-site utility applications for photovoltaics,” EPRI J., p. 27, 1991. R. Ramakumar and J. E. Bigger, “Photovoltaic Systems,” Proceedings of the IEEE, vol. 81, no. 3, pp. 365–377, 1993. R. Ramakumar, “Renewable energy sources and developing countries,” IEEE Transactions on Power Apparatus and Systems, vol. PAS-102, no. 2, pp. 502–510, 1983. R. Ramakumar, “Wind-electric conversion utilizing field modulated generator systems,” Solar Energy, vol. 20, no. 1, pp. 109–117, 1978. R. Ramakumar, I. Abouzahr, and K. Ashenayi, “A knowledge-based approach to the design of integrated renewable energy systems,” IEEE Transactions on Energy Conversion, vol. EC-7, no. 4, pp. 648–659, 1992. R. Ramakumar, H. J. Allison, and W. L. Hughes, “Solar energy conversion and storage systems for the future,” IEEE Transactions on Power Apparatus and Systems, vol. PAS-94, no. 6, pp. 1926–1934, 1975. R. H. Taylor, Alternative Energy Sources for the Centralised Generation of Electricity, Bristol, U.K.: Hilger, 1983. The Potential of Renewable Energy, An interlaboratory white paper, prepared for the U.S. Department of Energy, Solar Energy Research Institute, Golden, Colo., 1990. ? 2000 by CRC Press LLC ? 2000 by CRC Press LLC NIAGARA FALLS ELECTRICAL TECHNOLOGY SHOWPLACE t the close of the 19th century, Niagara Falls, New York, represented a showplace for displaying how far the electrical engineering profession had come in one short decade. Here, electrical engineers were confronted with one of the great technical challenges of the age — how to harness the enormous power latent in Niagara’s thundering waters and make it available for useful work. Years of study and heated debate preceded the start-up of the first Niagara Falls Power Station in the summer of 1895, as engineers and financiers argued about whether electricity could be relied on to transmit large amounts of power the 20 miles to Buffalo and, if so, whether it should be direct or alternating current. The success of the giant polyphase alternating current generators made clear the directions that electric power technology would take in the new century, and the attraction of novel industries that consumed great amounts of electricity, such as aluminum and other electrochemical manufacturers, showed the vast potential for growth and change that electricity held for the future. (Courtesy of IEEE Center for the History of Electrical Engineering.) The discovery of how to use electricity to make aluminum in 1886 gave Niagara Falls its first major consumer of power — the Pittsburgh Reduction Company, known today as the Aluminum Company of America (ALCOA). (Photo courtesy of IEEE Center for the History of Electrical Engineering.) A 60.2 Photovoltaic Solar Cells Allen M. Barnett and Lawrence L. Kazmerski Photovoltaic solar cells are semiconductor diodes that are designed to abso rb sunlight and convert it into electricity. The absorption of sunlight creates free minority carriers, which determine the solar cell current. These carriers are collected and separated by the junction of the diode, which determines the voltage. Photo- voltaic solar cells have been the power supply of choice for satellites since 1958; 350 kilowatts of solar cells were sold for space applications in 1998. The widespread use of photovoltaic solar cells for terrestrial applications began during the oil crisis of 1973. The market for these solar cells has grown from 240 kilowatts in 1976 to 160 megawatts in 1998. Space solar cells cost approximately 100 times as much as terrestrial solar cells, so the revenue difference between the two markets is not as great as the power generation difference. Solar Cell Operation and Characteristics The basic operation of a solar cell is shown in Fig. 60.4. Photons of light are absorbed by the semiconductor material and each photon that is absorbed generates an electron-hole pair. The generated minority carriers diffuse to the junction where they are collected. The number of collected carriers determines the current. The voltage is determined by the junction characteristics. The equivalent circuit is shown in Fig. 60.5. The charac- teristic curve of the photovoltaic solar cell can be determined by first calculating the collected minority carriers and then separately calculating the current–voltage characteristic of the diode. Superposition can be used to combine them. The maximum current for any solar cell is dependent on the bandgap of the absorbing semiconductor and the solar spectrum. Any photon with an energy greater than the bandgap can be expected to generate one electron-hole pair, which will lead to one collectable minority carrier. The absorption coefficient of the semi- conductor material determines the thickness required to absorb the sunlight with energies greater than the bandgap. As examples, a silicon thickness of 0.5 mm will absorb 93% of the sunlight with an energy above its FIGURE 60.4 Operation of a solar cell. FIGURE 60.5 Equivalent circuit of a solar cell. ? 2000 by CRC Press LLC bandgap, while a gallium arsenide thickness of 5 μm will absorb 96% of the sunlight with an energy above it bandgap. The photogenerated minority carriers must diffuse to the junction to be collected. These carriers will recombine as a function of the distance that they travel in accordance with the following formula n = n 0 e x/L , where x is the distance traveled and L is the diffusion length for the minority carrier. (60.1) The other determinant of current is the spectrum of the light. The two most commonly used spectra in modeling are AM1.5 for terrestrial applications and AM0 for space applications. AM1.5 is the spectrum of the sunlight after it has passed through an equivalent thickness of 1.5 atmospheres. The integrated power of this spectrum is 970 W m –2 . AM0 is the spectrum of the sunlight before it passes through any of the atmosphere (0 atmospheres) and has an integrated power of 1353 W m –2 . AM1.5D is used to describe direct sunlight for applications that use optics to focus the sunlight onto the solar cell, these solar cells are called concentrators and the available power to the optical surface is 752 W m –2 . The solar cell voltage is determined by the diode junction. The most common and most efficient junction is the pn junction. The diode characteristic is determined by J = J 0 (e qV/kT – 1). J 0 is the reverse saturation current of the solar cell and is most commonly described as (60.2) This is the case where surface recombination can be ignored and the device thickness is much greater than the minority carrier diffusion length. Unfortunately, surface recombination is an important factor in a number of semiconductor materials, including silicon. Also, the minority carrier diffusion length can be greater than the solar cell thickness. The equation for J 0 then becomes the more formidable (see [Hovel, 1976] or [Fahrenbruch and Bube, 1983]). Superposition of the theoretical maximum generated current as a function of energy gap and the diode characteristic leads to the solar cell curve shown in Fig. 60.6. This is the diode curve offset by the light-generated current (I L ). The maximum current is called the short circuit current, I sc , while the maximum voltage is called the open circuit voltage, V oc . The maximum power point is shown in Fig. 60.6 as the product of V mp and I mp as the IV curve is traced by varying the load. The maximum power is often described by V oc × I sc × FF, where FF is called the fill factor and is defined as V mp × I mp /V oc × I sc . FIGURE 60.6 Current voltage characteristic of solar cell showing superposition. LkTqD=μ=ττ Jq D L n N q D L n N p pD n nA 0 1 2 1 2 =+ ? 2000 by CRC Press LLC Efficiency is defined as the maximum power divided by the power in the solar spectrum (for the solar cell area). The theoretical maximum efficiency as a function of photon energy and solar spectrum can be calculated and is shown in Fig. 60.7. The effect of energy gap is embedded in term n i in the equation for J 0 . Solar Cell Types and Their Optimization When one reviews the defining equation for an ideal solar cell, (60.3) it becomes apparent that there are two types of solar cell design improvements: 1. Those that improve the light-generated current, I L 2. Those that improve the open-circuit voltage by decreasing the diode saturation current, I 0 The full impact of new solar cell designs is often hard to separate from the three parameters used to characterize the performance of any solar cell: short-circuit current, open-circuit voltage, fill factor and their product, P max . However, these parameters merely summarize, and often obscure, the potential performance of any particular design. Light-generated current, together with all the various recombination processes, are the fundamental determinants of solar cell performance. A high efficiency solar cell design must maximize light- generated current and minimize losses in the bulk base and emitter, within the collecting junction, and at the surfaces. Because of these requirements, a particular device design for any material is typically optimized to obtain, first, light-generated current, then open-circuit voltage, and finally fill factor. Many examples of this approach can be found for a variety of solar cell materials, including silicon, GaAs, amorphous silicon, CuInSe 2 , CuInGaSe 2 , CdTe, and InP. More recently, however, an approach that emphasizes open-circuit voltage, rather than short-circuit current, has been followed. Notable examples are present high-efficiency crystalline silicon and GaAs solar cells. This FIGURE 60.7 Single-junction solar cell efficiencies for the standard spectra. II e I qV kT L =? ? ? ? ? ? ? ? () 0 1 ? 2000 by CRC Press LLC philosophy is due, in part, to the fact that for any photovoltaic material, optimizing the light-generated current is, in general, an easier task to accomplish than optimizing the voltage-sensitive parameters. Therefore, a better measure of the potential performance of a material or design is the actual achieved open-circuit voltage, relative to the theoretical open-circuit voltage, rather than short-circuit current, which often has a value approaching the theoretically expected current. Although the open-circuit voltage of a pn junction solar cell is influenced by the operating temperature, the light-generated current and the minority carrier losses, the minority carrier loss term visually dominates since it can change by orders of magnitude for different solar cell designs, materials, and fabrication processes. Minority carriers can recombine in each of the various regions of the solar cell — in the base, within the junction depletion region, in the emitter, and at either the front or the rear surface. For completeness, it should be mentioned that recombination at each surface consists of two parts: recombination at the ohmic contact and recombination at the free surface (if the metallization does not cover the whole surface, such as the front of the solar cell). In the limiting thick-base case, recombination at the surfaces and within the junction and emitter of the solar cell will be minimized in comparison to the base recombination term, and minority carrier losses will be controlled solely by the minority carrier lifetime in the base “bulk” region. For the thin-base device, however, base recombination will be reduced with volume, and minority carrier losses will be increas- ingly influenced by the surfaces and contacts. Solar cell design innovations that significantly impact high-efficiency performance include: 1. Increased photon absorption with surface texturing and antireflection coatings to reduce top surface reflection 2. Surface passivation and low-recombination emitters that decrease both bulk and surface recombination near the top surface 3. High-low junctions and heterojunctions for reducing surface recombination, particularly at the back surface 4. Reduced-area and heterojunction contacts to reduce contact recombination losses 5. High-injection conditions that can lead to reduced J 0 6. And for polycrystalline materials, grain-boundary passivation to reduce recombination at the defects at the grain boundaries The first design innovation contributes to improved performance primarily by increasing the light-generated current. The second innovation leads to increases in light-generated current for a direct bandgap material and to increased voltage in all materials. The next four lead to decreases in minority carrier recombination, and increased open-circuit voltage. In addition to the pn junction, there are a number of other ways to make semiconductor diode junctions. These approaches, in addition to the pn or homojunction, are heterojunction, metal/semiconductor junction, metal/insulator/semiconductor (MIS), semiconductor/insulator/semiconductor (SIS), and electrolyte/semicon- ductor. Band diagrams for these six structures are shown on Fig. 60.8. For a detailed discussion of the physics of these junction types, see [Fonash, 1981; Green, 1982; and Hovel, 1976]. The junction is designed to maximize the voltage at any current; accordingly the objective is to minimize J 0 . The pn junction leads to the minimum values for I 0 , in part due to the lack of interface states and the ability to tune the doping. Other junction types can approximate, but usually fall short of the values achieved by the pn junction. The status of the technologies — including an examination of the problems, opportunities, and issues with the various technical approaches — is summarized in this chapter. Bulk crystalline (Si, GaAs) and thin-film Cu-ternaries and multinaries, CdTe, hydrogenated amorphous silicon and dye-sensitized, CuInSe 2 , CdTe tech- nologies are reviewed. Applications for flat-plate and concentrator modes of operation are considered. This chapter can only present a brief overview, and the reader is directed to several other sources for more detailed evaluations and technical discussions [Kazmerski, 1997]. It can also only provide a “snapshot” of technologies that are changing with advancing R&D and manufacturing. Following is a discussion of specific solar cells. ? 2000 by CRC Press LLC Crystalline Silicon Silicon continues to be the foundation of the PV industry. The material is abundant, and Si solar cells have demonstrated reliability in both space and terrestrial environments [Green, 1996]. Although the properties of this semiconductor might make it non-ideal from a theoretical view, Si solar cells have reached among the highest efficiencies among the PV material options. The evolution of the high-efficiency Si solar cell is illustrated in the device cross sections shown in Fig. 60.9. The relatively simple, conventional p-n junction has given way to more complicated designs and structures — all aimed at capturing every incident photon, maximizing electron-hole generation, and generating maximum currents. This has called for improving antireflection approaches on the front surfaces, providing for multiple passes for the light, incorporating back surface electric field to reflect noncollected carriers, and minimizing obscuration of the incident light on the front surfaces. The evolution of these designs has included metal/insulator/n-type/p-type (MINP), passivated-emitter solar cells (PESC), single-sided and doubled-sided buried contact (SSBC and DSBC), point contact, and bifacial cells (see [Green, 1996]). Efficiencies up to 24% have been verified with monocrystalline Si cells. It should be noted that although performance enhancement is attained through these more complex designs, there is also a potentially significant increase in manufacturing cost. A major cost factor for the Si solar cell is that associated with a high-perfection wafer. This has directed attention toward less energy-intensive process, which sacrifice the crystalline order and higher device efficiency for the benefits of lower energy production and perhaps the utilization of lower purity feedstock Si. Casting (and some sister technologies) has become conventional in the current Si manufacturing industry. Bulk and ribbon approaches for sheet-Si material have undergone extensive investigations and development over the past 20 years. Among the more developed technologies is the Edge Film-Fed Growth (EFG) process, which involves the shaping of Si through a special die and forming a connected octagon of flat sheets. The cells are cut from the connected structure by lasers, and large area solar cells (100 cm 2 ) with efficiencies exceeding 14% have been produced. Treatments of the multi- and polycrystalline Si with hydrogen, lithium, aluminum, arsenic, and phosphorus have been used with varying degrees of success to minimize the effects of active defects and boundary surfaces. There is considerable effort on identifying treatment processes that benefit commercially produced polycrys- talline Si products and to implement them into manufacturing. The best bulk multicrystalline Si cells have reported 18.6% efficiency. FIGURE 60.8 Band diagrams for various solar cell junctions: (a) homojunction; (b) heterojunction; (c) metal/semicon- ductor; (d) metal/insulator/semiconductor (MIS); (e) semiconductor/insulator/semiconductor (SIS); and, (f) electro- lyte/semiconductor. ? 2000 by CRC Press LLC For better materials utilization, thin-film Si has always been of keen interest to the photovoltaics community. Early work in this area was limited to cells having efficiencies in the 5% regime, much below expectations. With the evolution of better cell engineering and understanding of Si deposition, higher efficiencies have been realized. Among these are a 20.6% (CVD of Si on Si) 47 μm thick active layer, a 47 μm thinned Si cell (CVD of Si on Si) with efficiency of 21.5%, and a PERL structure. However, the economy of these cells is not expected to meet large-scale production requirements in their current technologies. (Figure 60.10 presents some structures of advanced thin-film Si designs.) They serve as indicators that high efficiencies can be reached with thin-Si layers. The economy of films of Si requires high-performance and low-cost materials, process, and production methods. AstroPower Silicon-Film? solar cells have reached 16.6%. This thin-Si solar cell represents a first-phase com- mercial design. Other more advanced designs are also illustrated in Fig. 60.10. Si-based solar cells continue to lead the industry in performance, reliability, and availability. Major issues and concerns for Si include: ? Silicon feedstock (materials availability, solar-grade Si, competitiveness with electronic technologies) ? Manufacturing costs (yields, complexity) ? Manufacturing capacity (current output, demands, plans for increased capacity) ? Research (materials production, processing, solar cell design, thin films) FIGURE 60.9 Evolution of silicon solar cell designs. ? 2000 by CRC Press LLC III-V Semiconductors Semiconductors such as GaAs, GaAlAs, GaInAsP, InSb, and InP have been receiving attention for photovoltaics because they have exceptional characteristics that offer energy conversions exceeding one third of the sun’s energy. Although cost is the overriding consideration for terrestrial applications, some compelling arguments can be made for their use in concentrators. Space markets have expanded for these materials, since solar cell cost has been less a factor compared to performance. Because of the ability to adjust the electro-optical properties (e.g., bandgap), these materials lend themselves extremely well to multiple bandgap cell designs. Fig. 60.11 presents structures for two-, three-, and four-terminal approaches. Solar cells in the 30 to 34% efficiency range have been realized for these structures, and research continues in order to bring about better performances with lower complexities [Friedman, 1998]. There has been some recent attention directed toward the two- terminal, two-junction tandem — with the best efficiency of 30.3% under non-concentrator conditions. While these robust performing solar cells boast the highest efficiencies, the cost demands for terrestrial markets are still impeding their acceptance in competition with other approaches. Dominant issues include: ? Cost (materials, manufacturing, and processing) ? Industry (primarily directed toward space applications currently) ? Research (materials, processing, cell engineering) Emerging from the development of lower-bandgap III-V cells is the resurgent thermophotovoltaic (TPV) technology [Coutts and Ward, 1998]. These are photovoltaic devices designed to work with infrared or thermal sources rather than sunlight. Early considerations of this technology used semiconductors with bandgaps in the range of 0.9 to 1.1 eV (primarily Ge and Si), that are matched to black-body temperatures above 2000K, which is difficult to engineer real systems. The development of devices in the 0.5 eV range correspond to temperatures of approximately 1500K, more suitable for system realization. The complete TPV system includes: (1) a fuel and a burner that is non-direct solar, (2) a radiator using either selective or broadband emitters, (3) a long-wave photon recovery mechanism, (4) a PV cell or converter, and (5) a waste-heat recuperation system. TPV represents a growing and intensive research area for photovoltaics. FIGURE 60.10 Silicon-Film? products: (a) current design; (b) thinner solar cell design; and (c) thin, interconnected solar cell design. ? 2000 by CRC Press LLC System efficiencies have promise of better than 40%. Certainly, an attractive system quality is that it offers “24-hour power” potential, directly from a fueled burner. Selective emitters, primarily rare-earth oxides (e.g., Yb 2 O 3 for Si, Ho 2 O 3 for III-V alloys) and broadband emitters, black and gray-body types (e.g., SiC-based for 1300 to 1700K), are under consideration. The TPV cells are central to the system, and include InGaAs, GaSb, and InGaAsSb. The GaSb cells have been introduced into commercial products. Thin-film semiconductor cells based on Ge and chalcopyrites are research projects. The radiation sources for terrestrial applications remain the most intriguing. These vary from biomass related (e.g., burning of wood-powder) to industrial waste heat (e.g., large furnaces used in float-glass production). In any case, TPV is in the area of next-generation technol- ogies that could make a very large contribution to energy generation. Thin-Film Solar Cells The arguments for thin-film solar cells for terrestrial PV applications are primarily based on materials utiliza- tion, large-scale manufacturing potential, and better energy economy for production. The focus of this section is on the major approaches based on copper indium selenide, cadmium telluride, hydrogenated amorphous silicon (α Si:H), and dye-sensitized cells (see [Kazmerski, 1997]). Solar cells made from these materials generally have lower voltages than predicted by the energy gap because the diode technology is not an ideal pn junction. Additionally, there may be grain boundary losses due to additional minority carrier recombination, parallel diodes, and parasitic resistance, as shown in Fig. 60.12. FIGURE 60.11 Multijunction structures for high-efficiency cells: (a) two terminal; (b) three terminal; and, (c) four terminal. ? 2000 by CRC Press LLC Cu-Ternaries and Multinaries Interest in Cu-ternary semiconductors began in the early 1970s for solar cells. CuInSe 2 and CuInS 2 (and their various alloys) have dominated R&D. Advantageous properties include: (1) suitable bandgaps for homojunc- tions or heterojunctions; (2) direct bandgap transitions minimizing requirements for absorber-layer thickness; (3) n- and p-type semiconductor types achievable; (4) lattice and electron affinity matches with window layer partners; (5) high optical absorption coefficients; and (6) stable electro-optical properties. Most of the emphasis has been on modifications of CdS/Cu(In,Ga)Se 2 (or CIGS) heterojunction devices. The cross-sectional representations in Fig. 60.13 indicate the relative complexity in structure. Each of the layers, thicknesses, interfaces, and compositions are ascribed to the engineering of the cell for optimal performance and reliability. The best research cells have been demonstrated as high as a remarkable 18.8% for these true, polycrystalline thin films. Certainly, the positive and perhaps unique factors that favor this thin-film technology are stability and large-area production potential — with performance characteristics similar to those for laboratory devices. (Commercial modules with better than 12% efficiency and 4 ft 2 areas are available, using Cu(In,Ga)(S,Se) 2 active layers.) The variety of techniques (vacuum and non-vacuum deposition) used to make the cells speaks to the potential of low-cost manufacturing. Recently, a 15% “Cd-free” ZnO/CIGS research device has been reported. This and other Cu ternaries are undergoing research: CuGaSe 2 and CuInS 2 are prime contenders. The issues and concerns with CuInSe 2 and alloys include: ? Research (chemical paths to materials realization, window heteropartner, process development, minority carrier properties, contacts, role of sodium, alloy compositions) ? Complexity (manufacturing costs, control) ? Stability (increase in efficiency with light exposure) ? Device issues (low open-circuit voltage, high short-circuit current) ? Scale-up ? Manufacturing base (enhancing the current embryonic commercial base and products) FIGURE 60.12 Equivalent circuit of a solar cell with grain boundaries. FIGURE 60.13 Device cross sections for Cu(In,Ga)Se 2 solar cell designs. ? 2000 by CRC Press LLC Cadmium Telluride Since the 1960s, CdTe has been a candidate photovoltaic material. Evaporation, spraying, screen printing/sin- tering, and electrodeposition have been used to produce efficient solar cells. Inherent to most solar cell processing is a chemical treatment in CdCl 2 : methanol solutions at high temperature (~400°C). The beneficial effects of this process have been attributed to enhanced grain size, identification, evolution of a p-i-n or heterojunction, surface alteration/passivation, alteration of shallow and/or deep electronic levels, improvement in morphology, and the formation of an interfacial CdSTe layer. The best solar cell efficiency has been measured at 15.8%. There is some concern that this record efficiency has not been exceeded since its report more than 6 years ago. There is a small manufacturing base. Modules in excess of 9% have been confirmed. The devices are relatively stable, although there are concerns with humidity. Concerns for the environmental effects of Cd surround this (and other technologies). A great deal of validation (fire testing, leaching) testing has been undertaken by the industry and research laboratories. The reader is directed to the literature about this sensitive area for Cd-based cells. Issues and concern for CdTe include: ? Research (process development, modeling, interface optimization, contacts, chemical treatments, role of oxygen) ? Substrates (cost of borosilicate glass and use of sodium glasses) ? Cadmium (environmental concerns and availability) ? Stability (Cu diffusion, contact oxidation, contact degradation, humidity) ? Scale-up (cell vs. module performance levels) ? Manufacturing base Hydrogenated-Amorphous Silicon In contrast to more perfect crystalline materials, amorphous semiconductors have neither short- nor long- range structural order. At its introduction, a-Si:H seemed to be the ideal photovoltaic candidate. Its bandgap can be varied over tenths of eVs by changing the hydrogen content. Because its physics are considerably different from its single-crystal relative, its absorption characteristics make it about 100 times more effective in absorbing the sun’s irradiance. It also has benefited technologically because there are other electronic technologies (tran- sistors, flat-panel displays) that have enhanced knowledge and understanding of its properties. The evolution of the a-Si:H cell is illustrated in Fig. 60.14. The development of various device structures has been an integral part of improving solar cell performance. The inherent light instabilities [Staebler and Wronski, 1977] have been minimized by engineering of the layer thick- nesses, and by the use of multiple or tandem structures. The origins and cure for the light instabilities have not been completely identified, but “stabilized” cells and modules having less than 10% change in output characteristics have been produced. Many solar cells and modules with efficien- cies exceeding 10% have been reported and confirmed. Amorphous silicon-based solar cells and modules have serious problems with several stability mechanisms. The stabilized efficiencies of research solar cells are about 33% less than other thin-film options (i.e., 12% vs. 18%). Although solar cell and module engineering have mini- mized the Staebler-Wronski effect, the stability issue remains a major research, manufacturing, and consumer acceptance issue. Module design has requirements beyond those for crystalline Si and the ingress of any environmental entity has major influence because of the large surface-to- volume ratios involved. Stability and reliability of a device is the major issue relating to amorphous Si technology. FIGURE 60.14 Amorphous Si:H module designs, showing interconnections. ? 2000 by CRC Press LLC These issues may require the development of new methods of producing the films, new film structures (e.g., nanocrystalline Si or multilayers), or a new a-Si technology. Issues and concerns with the amorphous silicon technology primarily relate to the overriding stability issues. However, a variety of other concerns do accompany this technology: ? Stability and reliability ? Research (modeling, characterization, analysis, deposition research) ? Production costs (capital costs) ? Manufacturing capacity (vacuum based costs, yields, production volume) Dye-Sensitized Cells Among the thin-film technologies on the near horizon is the dye-sensitized solar cell. The device, illustrated in Fig. 60.15, utilizes an oxide semiconductor (TiO 2 ) having a bandgap in the 3-eV range that is insensitive to the solar spectrum. The extension of the photoresponse across the visible portion of the spectrum is achieved by separation of the two steps of the photovoltaic process. The oxide semiconductor in the electrochemical system is sensitized by a monolayer of an electroactive dye having an optical absorption band extending across the width of the visible spectrum. Charge separation occurs by electron loss from the photoexcited dye to the semiconductor substrate. Following absorption of a photon, the excited state of the dye is such that relaxation FIGURE 60.15 Device cross section for dye-sensitized solar cell. ? 2000 by CRC Press LLC by electron loss to the semiconductor substrate is possible, leaving the dye molecule as an oxidized species. The original ground state of the dye is restored by charge transfer reactions with the redox electrolyte. The solar cell circuit is completed by a metallized counterelectrode at which a reduction reaction with a redox system takes place. Research cells with better than 10% efficiency have been confirmed. Such cells have substantially better conversion efficiencies at lower light intensity levels. The technology is promising because it involves a low-cost processing of large areas. The solar cell continues to show growth in performance and in stability, a parameter that has been of concern to research and manufacturing groups. This technology is still in its infancy, and represents one area of thin films that is an alternative to conventional junction solar cell approaches. Module Technologies Solar cells are typically electrically connected into series and par- allel strings to deliver a desired voltage and current and encapsu- lated into a supporting structure for environmental protection and strength. The module is a composite structure involving mechanical, optical, and electrical optimization that has required the collaborating and overlapping knowledge of physics, chemis- try, materials science, and engineering to ensure viability [Lasnier and Ang, 1990]. The materials used for support and encapsulation depend on both the solar cell type and the application/installation. The module construction determines not only its cost, but also its lifetime. Module design has occupied a significant portion of the development efforts in photovoltaics, and the complexities and details for all photovoltaic materials are beyond the scope of this review. However, module issues are almost as important as solar cell issues because they directly impact the performance, lifetime, and cost of the photovoltaic technology [Wohlgemuth, 1995]. The module is central to meeting not only the efficiency goals (e.g., 15 to 25% for modules), but also the system cost (e.g., $1.00 to $1.50/W) and system lifetime (e.g., >30 years) in the 2010 to 2030 timeframe. (See Fig. 60.16.) Even the most developed and commercialized of the photovol- taics approaches, crystalline Si, has required many redirections of its module construction to meet either operational or cost limi- tations over the past 25 years. Two examples highlight this ability and necessity for change. First, the module has been redesigned to eliminate framing to decrease materials cost, and improve loading (weight) and aero- dynamics when configured into the array. In this same area, some modules, traditionally only the DC delivery system, have integrated the inverter into its construction to meet AC energy requirements. The second example has to do with the encapsulation of the Si cells in a polymer — commonly ethylene vinyl acetate (EVA). This encapsulant was identified through several development programs as a cost-effective and environment-resistant material for PV modules in the late 1970s. In the mid-1980s, the EVA in some modules was observed to turn yellow and brown. Subsequently, new and improved polymer encapsulants have also been developed to replace the original EVA formulations and UV-absortion glass has been implemented. The evolving thin-film and concentrator technologies have additional complexities and, likely, as yet unknown problems. These advanced technologies are much different from their flat-plate silicon cousins. Consider the thin-film solar cells. Because the surface-to-volume ratios in these solar cell types are extremely high compared with bulk counterparts, materials and environmental interactions are not only enhanced, but affect relatively larger portions of the structures. These structures are also more complex, beyond the numerous interfaces that are inherent to the device itself. The cross sections for a-Si:H integrated module designs, shown in Fig. 60.14, illustrate areas of concern for shunting (bridges), contact openings, electromigration, interdiffu- sion, delamination, and microdefects that affect macroscale electrical behavior. Moreover, concentrators present module designs and complexities that have little relationship to their “one-sun” relatives. FIGURE 60.16 75 W, 36 solar cell module. ? 2000 by CRC Press LLC Photovoltaic Power Systems The photovoltaic module is the power generating component of a power system. Systems can be generically divided into stand-alone and grid-connected. A stand-alone system provides power directly to a load and usually includes storage. Stand-alone systems can vary greatly in complexity. A relatively simple system is a lighting system that includes a battery, battery charge regulator, the photovoltaic module(s), and the lights. More complex stand-alone systems include a range of battery charging systems, provide power for a whole house, provide power to a telecommunications repeater, or provide power to a satellite. All of these systems have battery and the battery charge control electronics in common. The simplest stand-alone system is a water pump that can be directly connected to the photovoltaic module. In this case, no electricity storage is required because the pumped water can be stored if it is not immediately used. Electricity grid-connected systems generally exist in two basic forms. Systems that immediately convert the generated electricity from direct current (dc) to alternating current (ac) and synchronize this power with the electricity grid are the most fundamental for safety reasons. These systems are designed to not supply electricity to the grid when there is a grid power outage. The most complex photovoltaic systems are grid-connected systems with storage — sometimes called uninterruptible power supplies. These systems disconnect from the utility grid when there is a power outage but continue to provide electricity to the load. The load can often be a home, commercial building, or a sensitive load, such as a computer. The systems include the ability to synchronize with the grid, the ability to provide electricity to the grid when the amount generated is greater than the load, battery charging, and the conversion of dc to ac. A schematic of one of these systems is shown in Fig. 60.17. All other systems can be derived from this complex system by removing components. Defining Terms Photovoltaic effect: Conversion of photons to electricity. Solar cell: Diode that converts sunlight to electricity using photovoltaic effect. Fill factor: A measure of the relative squareness of the solar cell diode curve. Electricity grid: Transmission and distribution system for centrally generated electricity. References T.J. Coutts and S. Ward, Thermophotovoltaic Solar Cells, Scientific American, 1998. A.L. Fahrenbruch and R.H. Bube, Fundamentals of Solar Cells, Academic Press, New York, 1983. S.J. Fonash, Solar Cell Device Physics, New York, Academic Press, 1981. D. Friedman, Proc. 2nd World Conference on Photovoltaic Solar Energy Conversion, Vienna, Austria, IEEE Press, 1998. M.A. Green, Silicon Solar Cells, University of New South Wales Press, Sydney, Australia, 1996. M.A. Green, Solar Cells: Operating Principles, Technology and System Applications, Prentice-Hall, Englewood Cliffs, NJ, 1982. H.J. Hovel, Semiconductors and Semimetals, Vol. 2, Academic Press, 1976. L.L. Kazmerski, Photovoltaics: A Review of Cell and Module Technologies, Renewable and Sustainable Energy Reviews, chap. 1, pp. 71, 1997. FIGURE 60.17 Grid-connected photovoltaic system. ? 2000 by CRC Press LLC F. Lasnier and T.G. Ang, Photovoltaic Engineering Handbook, Adam Hilger Publishing, Bristol, UK, 1990. D.L. Staebler and C.R. Wronski, Applied Physics Letter, Vol. 31, 292–294, 1977. J. Wohlgemuth, Proc. 24th IEEE Photovoltaic Specialists Conference, New York, IEEE Press, pp. 901–904, 1995. Further Information Chopra, K.L. and S.R. Das, Thin Film Solar Cells, Plenum Press, New York, 1983. R. Hill, Applications of Photovoltaics, Adam Hilger Publishing, Bristol, UK, 1989. A. Luque and G.L. Araujo (Eds.), Physical Limitations to Photovoltaic Energy Conversion, Adam Hilger Publishing, Bristol, UK, 1990. J.A. Mazer, Solar Cells — An Introduction to Crystalline Photovoltaic Technology, Kluwer Academic Publishers, 1997. A. Ricaud, Photopiles Solaires, PPUR, EPFL, Lausanne, Switzerland, 1997. R.J. Van Overstraeten and R.P. Mertens, Physics, Technology and Use of Photovoltaics, Adam Hilger, Bristol, UK, 1986. S.R. Wenham, M.A. Green, and M.E. Watt, Applied Photovoltaics, Centre for Photovoltaics and Systems, Aus- tralia, 1994. 60.3 Thermophotovoltaics John P. Benner and Timothy J. Coutts Thermophotovoltaic (TPV) is the title given to a process of generating electric power by photovoltaic conversion of the energy radiated from a thermal source. The source of energy can be a hydrocarbon or hydrogen burner, concentrated solar energy, radioisotope or nuclear reactor heating the emitter to an operating temperature in the range of 1000°C to 1800°C. Two significant attributes of TPV systems, relative to photovoltaic solar energy conversion, are that the source is in close proximity to the cell and thus delivers a high power density and that the emission spectrum lies in the infrared. TPV cells share many design features with concentrator solar cells, but are generally made from semiconductors with smaller bandgaps in order to achieve peak response better matched to the spectrum of the emitter. Silicon devices, however, are still viable for systems with emitter temperatures in the upper end of the above range. The final component essential for high conversion efficiency is some mechanism for controlling the flux of sub-bandgap photons such that their energy is not lost. This can be achieved with either selective emission or optical systems to return unused energy to the emitter. Theoret- ically, TPV systems could achieve conversion efficiencies approaching 30%, far higher than any other candidates for energy conversion in this temperature range. In the near term, a system efficiency of 10% is a practical goal. System design must balance the conversion efficiency and power density from the PV converter. This important trade-off will be described later. TPV systems are now commercially available operating at power levels in the range of several tens of watts to about 500 W. Market analyses indicate that TPV may be competitive for a variety of distributed generation applications of up to about 100 kW. In addition, many industrial processes operate in a range of temperatures suitable for TPV conversion. As the technology matures, these may provide the opportunity for co-generation of heat, as well as electricity. In this type of application, TPV units of several megawatt electric generating capacity would be reached to correspond with the scale of the waste heat from associated manufacturing processes. Background The first reference to the concept of TPV appeared in 1961 and attributes the original idea to Professor Pierre Agrain [1]. In a series of lectures at MIT in the early 1960s, Agrain assumed that radiation from the emitter that is not useful for conversion in the photocell could be returned to the emitter. Photocells were available in several materials systems, but only silicon had reached conversion efficiencies of a level for consideration in power conversion. Silicon’s limited utilization of the emission spectrum from early systems was addressed in ? 2000 by CRC Press LLC 1964 with proposed use of selective emitters, development of rare-earth oxide systems for selective emission, and with use of germanium photocells [2–4]. By the mid-1970s, system designs of up to 1 kW with projected efficiencies of 6 to 7% were brought to the stage of working prototypes [5]. Interest waned as the benefits did not appear to outweigh the challenges in thermal management and structural stability through thermal cycling to the required emitter temperatures. More than a decade later, the advances in high-efficiency solar photovoltaics and innovations in selective emitter development prompted renewed interest in TPV systems. Solar-to-electric energy conversion efficiencies greater than 30% were achieved in multiple-junction structures. These devices exploit the range of properties that can be obtained from III–V semiconductors. The ability to tune the photocell to a desired response characteristic provided the degree of freedom necessary to address the remaining system design issues effectively. Design Considerations of a TPV System Figure 60.18 is a cross-sectional drawing of a proto- type TPV unit under development by McDermott Technologies for use as a portable power source and battery charger [6]. One of the major advantages of TPV is the wide variety of fuels that can be selected for powering the system. This system will use diesel fuel, greatly simplifying logistics for the planned use by the Army. The combination of burner, radiator, gas flow channels, and emitter is designed to optimize the temperature uniformity over the emitter surface and also to isolate the PV cells from any of the hot com- bustion products. At a flame temperature of 1700K, the TPV system can be designed for very low NO x emission. The steady burn contributes to quiet and reliable operation. Burner efficiency is improved by thermal recuperation. This consists of a heat exchanger that recirculates a high percentage of the heat of the combustion products to raise the temper- ature of the incoming combustion air and fuel. Note that the exhaust is mixed with the air flowing over the cooling fins for the PV cells, dropping the final outlet temperature to only 30 to 50°C over ambient. The cylindrical geometry and flow from the burner, up through the radiator, then down through the channel between the radiator and the emitter, are key features in the thermal design. The energy emitted depends on emitter temperature in proportion to T 4 . System performance will clearly require a high degree of uniformity of emitter surface temperature. In this particular system, the emitter produces a black-body spectrum. The front surface of the cell is coated with a selective filter that reflects long-wavelength light back to the emitter. This not only minimizes the cooling load on the PV array, but also returns useful energy to the emitter, maintaining more efficient energy conversion. The solar cells are made of GaSb. These usefully absorb energies to a maximum wavelength of about 1.8 μ. Figure 60.19 shows the emission spectrum for several temperatures, overlaid by the cut-off energies for silicon and GaSb. For an emitter temperature of 1700K, the silicon device is able to use only about 5% of the available flux, while the device with a bandgap at about 0.7 eV can use 30%. When corrected for obscuration of the front contact grid, this portion of the available spectrum that the cell can use is called the spectral utilization factor. A silicon photovoltaic cell can achieve efficiencies of about 25% under illumination by the solar spectrum. The major losses are in relaxation of carriers generated by high-energy photons back to the bandgap and nonabsorbing sub-bandgap energies. The photovoltaic cell in a TPV system can be quite efficient — 40 to 50% — in converting the usable energy into electricity, since most of this energy arrives at close to the bandgap energy for a selective radiator. For this reason, the silicon device will be more efficient than the 0.7 eV device. FIGURE 60.18 Schematic of the McDermott Technologies portable TPV generator. (Used with permission of the Amer- ican Institute of Physics.) ? 2000 by CRC Press LLC And, as mentioned, some proportion of the sub-bandgap photons are not lost, but recirculated. However, the power density produced by the silicon-based system will be quite small. The overall system efficiency is more critically dependent on the efficiency of the spectral control components — the efficiency of recirculating the unused energy. This challenge is further compounded by the broad bandwidth needed for spectral control. Figure 60.20 plots the trade-off between conversion efficiency and power density for PV cells of various cut- off wavelengths and a black-body emitter at 1700K. The calculations assume that 100% of the sub-bandgap energy is usefully returned to the emitter. If this figure is reduced to 50%, the theoretical silicon-based system efficiency will drop to under 10%, while the 0.7 eV system efficiency will fall to 25%. The power density is, however, unaffected. The practical challenges in efficiently recirculating the sub-bandgap energies highlight the importance of the development of low-bandgap TPV cells. For high-efficiency III-V-based PV cells, such as the GaSb device, economic considerations demand that TPV systems operate a PV cell at power densities of about 1 W cm –2 . The geometry of the TPV system also has a major impact on power density for the converter. For example, it would be technically desirable to evacuate the region between the emitter and cell to reduce convection losses or to place a filter in this region, but economic considerations may render these improvements impracticable. In the cylindrical geometry, if the FIGURE 60.19 Black-body emission spectrum and spectral utilization for GaSb and Si photovoltaic cells. FIGURE 60.20 Calculated power density and converter efficiency for TPV operating at three different emitter temperatures. ? 2000 by CRC Press LLC converter array is separated far from the emitter, then the view factor is reduced. The view factor is the ratio of the photon flux per unit area impinging on the converter to that emanating from the emitter. The TPV system has several potential advantages, including: ? Versatile fuel usage ? High efficiency ? High power density ? Quiet ? Clean, low NO x emission ? No moving parts Optical Control of Sub-bandgap Energies Advances in selective emitters provided a major influence in rekindling interest in TPV development. Several investigators have shown the utility of rare-earth oxides such at holmia, ytterbia, erbia in modifying the emission spectrum from a broad band to one that selectively emits a substantial portion of the energy in a narrow band around a characteristic resonant frequency. Ytterbia emits with good selectivity at a peak wavelength of 980 nm — close to the bandgap of silicon (1070 nm). Given the relative maturity, efficiency, and low cost of silicon solar cells, development of an effective ytterbia emitter and surrounding system for use of silicon converters presents a pathway with potential for good performance and quick development for TPV. Unfortunately, the bulk properties of the substrate for these materials also contribute to the total radiation. Even a very low value of out-of-band emissivity integrated over a broad spectrum presents unacceptable losses. Mantles, similar to those used in camping lanterns, alleviate this loss by eliminating most of the bulk of the support structure. Nelson was the first to observe that fibrous emitters for TPV systems yield the same improvement [7]. However, it is difficult to scale the mantle structure to larger sizes and maintain acceptable mechanical performance. Several solutions for design of mechanically robust emitters are in development. One example uses selectively emitting fibers embedded in a ported ceramic block that allows the radiating fibers to work within the flame while the ceramic substrate remains relatively cool. The alternative to selective emission is to reflect sub-bandgap energy back to the emitter. Several types of filters show potential for this task, including dielectric layers, plasma filters, metallic reflectors on the back surface of the cell, or some combination of these. An ideal filter will have 100% transmission up to the band edge of the semiconductor and 100% reflection for lower energies. Even very small absorption losses in the filter system can produce unacceptable cooling loads at typical TPV system power densities. A new class of systems for optical control is under development using geometrical feature sizes on the order of the wavelength of the light. Patterning a metal film to produce a high-density array of antenna elements can achieve an inductive resonance that produces a bandpass filter. In a somewhat similar way, producing a fine periodic surface structure on the emitter material can produce wavelength-selective behavior. Development of PV Cells High-performance silicon solar cells are more widely available and lower in cost than other types of photovoltaic devices. For this reason, they remain in consideration for a number of TPV prototype systems. As discussed, choosing silicon for the PV converter places stringent requirements on other subsystems. TPV system design can take one of two paths to improve on Si-based converters; namely, either raise the emitter temperature (which will worsen thermal management problems) or seek PV devices with smaller bandgaps and commen- surately longer wavelength response. During the early phase of TPV research, only Ge cells offered a longer wavelength response. However, the intrinsic carrier concentration of germanium is too high for this device technology ever to reach a high efficiency [8]. Development of very high efficiency tandem PV cells for use with concentrated solar power provided the other major stimulus for renewed interest in TPV. Two designs were particularly important in that they took the approach of developing low bandgap booster cells of GaSb or Ga x In 1–x As for use under existing high- performance devices [9–11]. Both GaSb and Ga 0.47 In 0.53 As have bandgaps of about 0.7 eV, corresponding to a ? 2000 by CRC Press LLC cut-off wavelength near 1.8 μm. This response enabled new designs and applications for TPV using lower emitter temperatures and less-demanding photon recuperation. Ongoing research in TPV converters is advancing on two fronts. First, materials are in development with even lower bandgaps in the Ga x In 1–x As y Sb 1–y system and in Ga x In 1–x As. As shown in Figure 60.20, maximum power densities should be achieved for devices at about 0.5 eV or less. The availability of devices over a narrow range of bandgaps opens the possibility of further system efficiency gains through use of cascade multiple- junction converters. A two-cell stack with bandgaps of 0.7 eV and 0.53 eV should be current-matched for a black-body emitter operating at about 1500°C. The use of such tandem cells will likely place an even greater premium on achieving exceedingly uniform temperature profile over the emitter area. The other avenue of converter development is in the creation of monolithic interconnected modules. These devices are fabricated in semiconductor layers grown on semi-insulating substrates (or an isolation layer on a conducting substrate), using processing to isolate mesas into individual cells and then to connect the cells in series to produce a higher-voltage, lower-current circuit. This greatly simplifies the balance-of-system design and assembly because the converter size is limited only by dimensions of available substrates rather than current handling capability for the single high-flux device. The use of a semi-insulating substrate in the design may also provide the simplest method for optical and thermal control by incorporating a back surface reflector. This option is essentially precluded in other converter designs because of excessive free-carrier absorption losses in the relatively thick substrate. All of the TPV converters described above present some challenge in reaching a cost target of about $1 per watt. In the case of silicon, the power densities are quite small. For the GaSb or GaInAs devices, the cost of the substrate (GaSb or InP) and device processing leaves little margin. Thus, as in solar photovoltaics, there is growing research interest in thin-film technologies and in low-cost substrate alternatives. Status of System Development Despite the elegance of the physics of the emitters, optical control, and PV converters, the engineering of components and system designs to minimize optical and thermal losses is probably the major near-term barrier. Monte Carlo analyses of photon paths have provided insights into the mechanisms of photon loss for charac- teristic Lambertian emission profiles. These analyses yield some surprising results. Off-normal angle of inci- dence and internal reflection can multiply the effective cross section of absorbing or reflecting materials. As mentioned, the emitter operates at relatively high temperatures and must maintain a highly uniform temper- ature profile. Considerations of view factor and power density require close spacing between hot and cool components. Thus, some designs include evacuation of the region between emitter and converter. Small oversights in estimating conductive losses have also broadened the gap between predicted and achieved performance. Systems and Applications One class of systems, used for explanation here, is intended for military use. The Army consumes large quantities of batteries for communications and a range of other applications. In these kinds of applications, systems of less than 5% conversion efficiency using diesel fuel compare favorably to batteries. Remote communication systems also benefit from new approaches for power delivery. For example, many are currently powered by hybrid systems that may contain combinations of small diesel engines, photovoltaics, thermoelectrics, and batteries. In many remote applications, risk of environmental damage from spilled liquid fuels limits choices for power generation. TPV systems offer efficient conversion of energy from gaseous sources. This offers greatly reduced maintenance, reduced delivery cost of maintaining the fuel supply, and improved environmental acceptance. In the commercial sector, one of the early applications might well be for recreational vehicles and boats. Quiet operation carries premium value. One of the next early opportunities may well be for self-powered appliances. A furnace can be designed to extract the electricity needed to power the blower motor by TPV conversion of radiant emission from the burner. More than a million households in the U.S. experience power outages annually, leaving their homes without electric power to drive circulation circuits on heating units. The ? 2000 by CRC Press LLC Gas Research Institute estimates that a unit costing about $500 added to furnaces could capture much of this market to make furnaces self-powered units. The power demands for these will be in the range of 200 to 500 W. This path leads to larger energy units that could provide electricity, heat, and hot water for a home. The PV cell array for TPV systems are likely to represent about half of the system cost for units using crystalline silicon or III–V devices. How might the applications for TPV systems change if thin-film photovol- taics or other advances significantly drop projected system costs? One possibility may be in co-generation in industrial processes. For example, float-glass manufacturing processes 600 tons of material each day for a typical line. These factories melt feedstock in crucibles 100 ft long by 30 ft wide operating at 1500°C. Projected technology advances both in improving energy efficiency of glass manufacturing as well as in TPV create the potential for covering the top of this melt unit with an umbrella of TPV converters. A large part of the electric demand for the factory might be met by the co-generation unit. The engineering problems may be challenging, however, even if the benefits are potentially great. Defining Terms Recuperator: A heat exchanger that extracts energy from the combustion products to heat the incoming fuel and air. In TPV systems, optical control is sometimes called photon recuperation. Optical control: Technology such as a selective emitter or filter used to minimize loss of unusable sub-bandgap photons by the photovoltaic converter. View factor: Ratio of the photon flux per unit area impinging on the converter to that emanating from the emitter. Spectral utilization factor: The fraction of the incident energy that the photovoltaic converter can use to generate electricity. References 1. D. C. White, B. D. Wedlock, and J. Blair, Recent advances in thermal energy conversion, 15th Annual Proceedings, Power Sources Conference, May 1961, pp. 125–132. 2. D. C. White and R. J. Schwartz, P-I-N structures for controlled spectrum photovoltaic converters, Pro- ceedings NATO AGARD Conference, Cannes, France, March 1964. 3. E. Kittl, Thermophotovoltaic energy conversion, Proceedings 20th Annual Power Sources Conference, May 1966, pp. 178–182. 4. R. W. Beck and E. N. Sayers, Study of Germanium Devices for Use in a Thermophotovoltaic Converter, Progress Report No. 2 (Final Report) DA28-043-AMC-02543(E), General Motors Corporation, 1967. 5. E. Kittl and G. Guazzoni, Design analysis of TPV-generator system, Proceedings 25th Power Sources Symp., May 1972, pp. 106–109. 6. C. L. DeBellis, M. V. Scotto, L. Fraas, J. Samaras, R. C. Watson, and S. W. Scoles, Component development for 500 watt diesel fueled portable thermophotovoltaic (TPV) power supply, Thermophotovoltaic Gener- ation of Electricity: Fourth NREL Conference, AIP Confrence Proceedings 460, Woodbury NY. 7. R. E. Nelson, U.S. Patent No. 4,584,426, filed July 1984, issued April 1986. 8. J. L. Gray and A. El-Husseini, A simple parametric study of TPV system efficiency and output power density including a comparison of several TPV materials, Thermophotovoltaic Generation of Electricity: Second NREL Conference, AIP Confrence Proceedings 358, Woodbury NY, pp. 3–15. 9. L.M. Fraas, J. E. Avery, P. E. Gruenbaum, and V. S. Sundarum, Fundamental characterization studies of GaSb solar cells, Conference Record of the 22 nd IEEE Photovoltaic Specialists Conference, IEEE, New York, 1991, pp. 80–89. 10. M. W. Wanlass, T. A. Gessert, G. S. Horner, K. A. Emery, and T. J. Coutts, InP/Ga 0.47 As 0.53 monolithic, two-junction, three-terminal tandem solar cells, Proc. 10 th Space Photovoltaic Research and Technology Conference, NASA, 1989, pp. 102–116. 11. M. W. Wanlass, J. S. Ward, K. A. Emery, T. A. Gessert, C. R. Osterwald, and T. J. Coutts, High-performance concentrator tandem cells based on IR-sensitive bottom cells, Solar Cells, 30, 363, 1991. ? 2000 by CRC Press LLC Further Information Proceedings from the NREL Conferences 1, 2, 3, and 4 on Thermophotovoltaic Generation of Electricity; AIP Conference Proceedings Volumes 321, 358, 401, and 460, American Institute of Physics, Woodbury, NY; T. J. Coutts and M. C. Fitzgerald, Thermophotovoltaics, Scientific American, September, 1998, pp. 90–95. ? 2000 by CRC Press LLC