Search

Important changes in the ozone vertical profile over South Pole, Antarctica have occurred both during the recent period of measurements, 1986-1991, and since an earlier set of soundings was carried out from 1967-1971. From the onset of the 'ozone hole' over Antarctica in the early 1980s, there has been a tendency for years with lower spring ozone amounts to alternate with years with somewhat higher (although still depleted) ozone amounts. Beginning in 1989 there have been three consecutive years of strong depletion although the timing of the breakdown of the vortex has varied from year to year. Comparison of the vertical profiles between the two periods of study reveals the dramatic decreases in the ozone amounts in the stratosphere between 15-21 km during the spring. In addition, it appears that summer values are also now much lower in this altitude region.

Electrochemical concentration cell (ECC) ozone-sonde observations, made in recent years at ten stations whose locations range from the Arctic to Antarctica, have yielded a self-consistent ozone data base from which mean seasonal and annual latitudinal ozone vertical distributions to 35 km have been derived. Ozone measurement uncertainties are estimated, and results are presented in the Bass-Paur (1985) ozone absorption coefficient scale adopted for use with Dobson ozone spectrophotometers January 1, 1992. The data should be useful for comparison with model calculations of the global distribution of atmospheric ozone, for serving as apriori statistical information in deriving ozone vertical distributions from satellite and Umkehr observations, and for improving the satellite and Umkehr ozone inversion algorithms. Attention is drawn to similar results based on a less comprehensive data set published in Ozone in the Atmosphere, Proceedings of the 1988 Quadrennial Ozone Symposium where errors in data tabulations occurred for three of the stations due to inadvertent transposition of ozone partial pressure and air temperature values.

Ozone profiles obtained at Boulder, Colorado and Wallops Island, Virginia indicate that ozone was about 25% below normal during the winter and spring of 1992-93 in the 12-22 km region. This large ozone reduction in the lower stratosphere, though sometimes partially compensated by higher than normal ozone above 24 km, was responsible for the low total column ozone values observed across the United States during this period. Normal temperatures throughout the low ozone region suggest that transport-related effects are probably not the most important cause of the ozone deficits. This region of low ozone at Boulder corresponds closely with the location of the enhanced H2SO4/H2O aerosol from the Pinatubo eruption of 1991 as measured near Boulder and at Laramie, Wyoming. Trajectory analyses suggest that except at low altitudes in spring, air parcels on the days of the ozone measurements generally arrived at Boulder from higher latitude, although seldom higher than 60 deg N, and hence may have been subjected to heterogeneous chemical processing on the surface of Pinatubo aerosol droplets resulting in chlorine-catalyzed ozone destruction, a process which is believed to be more effective under the lower winter temperatures and sunlight levels of higher latitudes.

Atmospheric ozone vertical distributions, air temperatures, and wind speed and direction data are presented for 40 balloon electrochemical concentration cell ozone soundings made at Mirny, Antarctica, in 1988.

The progress and initial data on accelerated stress tests on small amorphous silicon test samples from several manufacturers are reported. The test samples include both single and tandem junction devices.

If thin film cells are to be considered a viable option for terrestrial power generation their reliability attributes will need to be explored and confidence in their stability obtained through accelerated testing. Development of a thin film accelerated test program will be more difficult than was the case for crystalline cells because of the monolithic construction nature of the cells. Specially constructed test samples will need to be fabricated, requiring committment to the concept of accelerated testing by the manufacturers. A new test schedule appropriate to thin film cells will need to be developed which will be different from that used in connection with crystalline cells. Preliminary work has been started to seek thin film schedule variations to two of the simplest tests: unbiased temperature and unbiased temperature humidity. Still to be examined are tests which involve the passage of current during temperature and/or humidity stress, either by biasing in the forward (or reverse) directions or by the application of light during stress. Investigation of these current (voltage) accelerated tests will involve development of methods of reliably contacting the thin conductive films during stress.

The work on reliability testing of solar cells is discussed. Results are given on initial temperature and humidity tests of amorphous silicon devices. Calibration and measurement procedures for amorphous and crystalline cells are given. Temperature stress levels are diagrammed.

Attention is given to the construction, calibration, and performance of a repeatable measurement system for use in conjunction with the accelerated stress testing of a-Si:H cells. A filtered diode array is utilized to approximate the spectral response of any type of solar cell in discrete portions of the spectrum. It is noted that in order to achieve the necessary degree of overall repeatability, it is necessary to pay particular attention to methods of contacting and positioning the cells.

A technique for performing accelerated stress tests of large-area thin a-Si solar cells is presented. A computer-controlled short-interval test system employing low-cost ac-powered ELH illumination and a simulated a-Si reference cell (seven individually bandpass-filtered zero-biased crystalline PIN photodiodes) calibrated to the response of an a-Si control cell is described and illustrated with flow diagrams, drawings, and graphs. Preliminary results indicate that while most tests of a program developed for c-Si cells are applicable to a-Si cells, spurious degradation may appear in a-Si cells tested at temperatures above 130 C.

Research on the reliability of terrestrial solar cells was performed to identify failure/degradation modes affecting solar cells and to relate these to basic physical, chemical, and metallurgical phenomena. Particular concerns addressed were the reliability attributes of individual single crystalline, polycrystalline, and amorphous thin film silicon cells. Results of subjecting different types of crystalline cells to the Clemson accelerated test schedule are given. Preliminary step stress results on one type of thin film amorphous silicon (a:Si) cell indicated that extraneous degradation modes were introduced above 140 C. Also described is development of measurement procedures which are applicable to the reliability testing of a:Si solar cells as well as an approach to achieving the necessary repeatability of fabricating a simulated a:Si reference cell from crystalline silicon photodiodes.

Plans for the development of amorphous cell accelerated test measurement instrumentation are outlined. Diagrams for an 11-lamp ELH solar simulator and ac light source instrumentation are given. Examples of ac and dc analysis graphs are also provided.

A method is described of determining an equivalent circuit for solar cells which have degraded as a result of the formation of a rectifying Schottky barrier at the back contact. An excellent fit of experimental data has been achieved using SCEPTRE with an equivalent circuit derived from the shape of the measured current voltage characteristics. One key parameter of the Schottky barrier diode, the reverse saturation current, can be used to determine the barrier potential. The barrier potential increases as the cell is stressed with 0.5 volts being a typical experimentally determined value for a degraded cell.

Clemson University has been engaged for the past five years in a program to determine the reliability attributes of solar cells by means of accelerated test procedures. The cells are electrically measured and visually inspected and then subjected for a period of time to stress in excess of that normally encountered in use, and then they are reinspected. Changes are noted and the process repeated. This testing has thus far involved 23 different unencapsulated cell types from 12 different manufacturers, and 10 different encapsulated cell types from 9 different manufacturers. Reliability attributes of metallization systems can be classified as major or minor, depending on the severity of the effects observed. As a result of the accelerated testing conducted under the Clemson program, major effects have been observed related to contact resistance and to mechanical adherence and solderability. This paper does not attempt a generalized survey of accelerated test results, but rather concentrates on one particular attribute of metallization that has been observed to cause electrical degradation - increased contact resistance due to Schottky barrier formation. In this example basic semiconductor theory was able to provide an understanding of the electrical effects observed during accelerated stress testing.

Results of an ongoing research program into the reliability of terrestrial solar cells are presented. Laboratory accelerated testing procedures are used to identify failure/degradation modes which are then related to basic physical, chemical, and metallurgical phenomena. In the most recent tests, ten different types of production cells, both with and without encapsulation, from eight different manufacturers were subjected to a variety of accelerated tests. Results indicated the presence of a number of hitherto undetected failure mechanisms, including Schottky barrier formation at back contacts and loss of adhesion of grid metallization. The mechanism of Schottky barrier formation is explained by hydrogen, formed by the dissociation of water molecules at the contact surface, diffusing to the metal semiconductor interface. This same mechanism accounts for the surprising increase in sensitivity to accelerated stress conditions that was observed in some cells when encapsulated.

Microcomputer controlled system gather current and voltage data. System consists of light source, microcomputer, programable dc power supply, analog/ digital interface, and data storage display equipment. Applies series of test loads to cell via programable dc power supply to obtain I/V characteristic curve and key cell-peformance parameter. Apparatus and programming technique are applicable to devices such as batteries and sensors.

The development of an accelerated test schedule for terrestrial solar cells is described. This schedule, based on anticipated failure modes deduced from a consideration of IC failure mechanisms, involves bias-temperature testing, humidity testing (including both 85-85 and pressure cooker stress), and thermal-cycle thermal-shock testing. Results are described for 12 different unencapsulated cell types. Both gradual electrical degradation and sudden catastrophic mechanical change were observed. These effects can be used to discriminate between cell types and technologies relative to their reliability attributes. Consideration is given to identifying laboratory failure modes which might lead to severe degradation in the field through second quadrant operation. Test results indicate that the ability of most cell types to withstand accelerated stress testing depends more on the manufacturer's design, processing, and worksmanship than on the particular metallization system. Preliminary tests comparing accelerated test results on encapsulated and unencapsulated cells are described.

The accelerated stress test results obtained on all terrestrial solar cells since the inception of the program are summarized. Tested cells were grouped according to the method used to form the conductive metallization layer: solder dipped, vacuum deposited, screen printed, and copper plated. Although metallization systems within each group were quite similar, they differed in numerous details according to the procedures employed by each manufacturer. Test results were summarized for all cells according to both electrical degradation and catastrophic mechanical changes. These results indicated a variability within each metallization category which was dependent on the manufacturer. Only one manufacturer was represented in the copper plated category and, although these showed no signs of detrimental copper diffusion during high temperature testing, their metallization was removed easily during high humidity pressure cooker testing. Preliminary testing of encapsulated cells showed no major differences between encapsulated and unencapsulated cells when subjected to accelerated testing.

The use of accelerated testing methods in a program to determine the reliability attributes of terrestrial silicon solar cells is discussed. Different failure modes are to be expected when cells with and without encapsulation are subjected to accelerated testing and separate test schedules for each are described. Unencapsulated test cells having slight variations in metallization are used to illustrate how accelerated testing can highlight different diffusion related failure mechanisms. The usefulness of accelerated testing when applied to encapsulated cells is illustrated by results showing that moisture related degradation may be many times worse with some forms of encapsulation than with no encapsulation at all.

An accelerated test schedule using a minimum amount of tests and a minimum number of cells has been developed on the basis of stress test results obtained from more than 1500 cells of seven different cell types. The proposed tests, which include bias-temperature, bias-temperature-humidity, power cycle, thermal cycle, and thermal shock tests, use as little as 10 and up to 25 cells, depending on the test type.

A drawback of conventional electrical evaluation procedures regarding solar cells is related to time consuming operations required in the determination of specific parameters, such as the maximum power point, from the plotted curves, and to the transfer of data to a large digital computer for analysis and manipulation. This is especially true when large numbers of cells must be measured. To overcome such drawbacks, a digital short interval tester was designed and constructed. The new tester provides rapid and accurate measurements at constant temperature. The utilization of a light shutter makes it possible to measure up to 200 data points along the I-V curve in less than a second, thus avoiding thermal effects. Because the system is digital, parameters such as the maximum power may be calculated directly from the data.

Major effort during this reporting period was devoted to two tasks: improvement of the electrical measurement instrumentation through the design and construction of a microcomputer controlled short interval tester, and better understanding of second quadrant behavior by developing a mathematical model relating cell temperature to electrical characteristics. In addition, some preliminary work is reported on an investigation into color changes observed after stressing.