What is the Appropriate Reference Spectrum for Characterizing Concentrator Cells?

Consensus standards for determining the efficiency of a concentrator cell or module have not been developed. NREL, Sandia National Laboratory, the Fraunhofer Institute for Solar Energy in Germany, and the Progress in Photovoltaics Efficiency Table authors have informally agreed upon concentrator-cell reference conditions. These conditions are 25∞C cell temperature, 1-sun = 1000 W/m total irradiance, and the ASTM E891-87 direct-normal reference spectrum. Deficiencies in the direct reference spectrum are discussed, and a more representative reference spectrum for evaluating concentrator cells is proposed. The spectrum was generated by the SMARTS model, and the atmospheric parameters are as close as possible to the existing direct spectrum, with the exception that the aerosol optical depth at 500 nm is reduced from 0.27 to 0.085. REVIEW OF EXISTING REFERENCE CONDITIONS Consensus standards were first proposed at the first and second terrestrial Photovoltaic Measurement Procedures Workshop in 1976 and documented in the Terrestrial PV Measurements Procedures [1-3]. The manual specified equipment and procedures for measurement of light I-V curves for one-sun and concentrator cells and modules [3]. For measurement of concentrator cells, the manual specified a direct-beam reference spectrum, a one-sun irradiance of 1000 W/m, a temperature of 28∞C, and an area defined as the area that is designed to be illuminated, which is normally the total area minus any peripheral bus bars or contacts [3]. For concentrator modules, the area was taken to be the cross-sectional area of the lens or mirror receiver. For the direct-beam reference spectrum, the atmospheric parameters were 2 cm precipitable water vapor, a turbidity of 0.12, an air mass (AM) of 1.5, and 0.34 cm of ozone [3]. The 1977 model was rather crude and was regenerated for global and direct conditions using a Monte Carlo computer model [4]. This new rigorous model was based on the United States Standard Atmosphere Mid-Latitude Summer profile, with 1.416 cm of precipitable water vapor and an ozone level of 0.344. An air mass of 1.5 at sea level with a sun–facing surface tilted at 37∞ and a wavelength-independent ground reflectivity of 0.2 were chosen [4]. The simplistic aerosol profile in the original reference spectrum [3] was difficult to reproduce and contained several errors in implementation. A rural aerosol profile was chosen in the rigorous model, and it corresponded to an aerosol optical depth at 500 nm of 0.27 or a visibility of 23 km. The choice of 0.27 for the turbidity was based on limited resource information, but was intended to be an average value in the continental United States and not an average value in locations where concentrators might be deployed. The results from the Monte Carlo model in Reference 4, along with the data from an undocumented simple model, were then incorporated into standards [5-7]. These standards have been in use by the photovoltaics community since about 1985. Reference conditions for rating concentrator cells and modules have been informally agreed upon by NREL, Sandia National Laboratory, the Fraunhofer Institute for Solar Energy in Germany, and the Progress in Photovoltaics Efficiency Table authors [8]. These conditions are 25∞C cell temperature, one-sun = 1000 W/m total irradiance, and the ASTM E891-87 direct-normal reference spectrum in Reference 7. Concentrator modules and systems have been rated at PVUSA with respect to their performance at a directnormal irradiance of 850 W/m, 1 m/s wind speed, and an air temperature of 20∞C [9]. Consensus standards for concentrator measurements are currently under development in the United States and Europe. MOTIVATION FOR CHANGING DIRECT REFERENCE SPECTRUM The U.S. Department of Energy (DOE) HighPerformance PV Project calls for a 33% concentrator module and a 40% concentrator cell to be developed [10]. The reference conditions to be used for measuring these efficiencies need to be clearly defined. Recent work has shown that the direct reference spectrum is not representative of sunny conditions in regions with a high annual direct-normal energy where concentrators might be deployed (the Sun Belt). In the past, this issue did not matter because at a given total irradiance and cell temperature under direct, global, or clear-sky natural sunlight, concentrator PV cells or modules produced the same shortcircuit current within ±2%. The reason stems from the fact that in the past, the only concentrator cells were singlejunction Si, GaAs, or independently measured multijunction cells that have a small spectra sensitivity. In contrast, the short-circuit current of the series-connected GaInP/GaAs/ Ge triple-junction cells are much more sensitive to spectral variations. The highest efficiency cell measured at NREL was 34.0±1.5% for solar fluxes between about 130 and 630 suns under the IEC global reference spectrum [5,6] and