Application of Solar Energy in Space

CHARLES E. BACKUS

MECHANICAL ENGINEERING FACULTY ARIZONA STATE UNIVERSITY ТЕМ PE, ARIZONA

13.1 INTRODUCTION

The “space age” was formally started with the launching of the Sputnik by the USSR in 1957. That satellite and many more in the early years of the space programs did not use solar energy for power, but it was not long before solar energy became the dominant source of power for space craft. On March 17, 1958 the US launched Vanguard I which contained a small array of available silicon solar cells to power its 5 mW backup transmitter. Two months later the Russians launched a satellite that was totally powered by solar cells. Since the fall of 1959, essentially all space crafts with a mission duration of more than a couple of weeks used solar cells as their prime power supply. This includes more than 1000 satellites with power outputs up to 20 kW. The demand for solar cells from 1959 through 1974 was completely dominated by the demand for space craft power supplies. In the US the annual demand for space rated solar cells has held rather consistently at about 50 kW/а. In 1975 the demand for terrestrial solar cells exceeded the space demand and the future will probably show that the space market is a specialty within the general solar cell field. This could possibly turn around if some of the proposed systems of using huge satellites with thousands of megawatts of capacity are used for terrestrial demands.

Several other energy conversion systems have been investigated for coupling with the solar source in space, but thus far have not been able to compete with the solar cell arrays. These conversion systems include thermo­dynamic vapor cycles, thermoelectrics, and thermionics.

When one considers the selection of an electric power supply for a space craft application, it is considerably different than terrestrial applications. Although one still uses the usual engineering criteria of the lowest cost system to meet the constraints, the unusual requirements for reliability and not interfering with the mission are severe for space craft. The best power system for a particular mission depends primarily on the power level re­quired and the length of the mission. For example, if the mission is less than one day with a modest power requirement of less than about 1 kW, then chemical storage batteries are probably the best. If the mission is between one day and one month in duration with power requirements of up to 100 kW, fuel cells would probably be the best choice. If the mission is more than a month, then the weight of the fuel for any system other than solar or nuclear

FIG. 13.1 The US Satellite Skylab. The solar cell arrays could provide more than 20 kW of power. Photograph courtesy of Jet Propulsion Laboratory, Pasadena, California.

makes it prohibitive. Radioactive-isotope thermoelectric generators (RTG) are used for some systems going away from the sun, such as a Jupiter flyby, or on missions that continuously require some power. Some of the solar cell powered satellites have a small RTG on board to supply power for the small amount of time the satellite may lose the sun, such as in an eclipse. For large power outputs, it would appear that a nuclear reactor would be more com­pact and cheaper than solar cell arrays. The anticipated crossover point in power size between a nuclear reactor and a solar array has been a function of time. In 1960, solar cell systems were only thought to be attractive up to power levels of 200 W. By 1963, this number had gradually been moved to about 500 W and indeed this power level of satellite was launched in August 1964 (Nimbus).

This was followed by the Orbiting Astronomical Observatory in 1966 with 1000 W of installed power, and eventually by Skylab (see Fig. 13.1) with more than 20 kW of power. Figure 13.2 shows one of the four panels that could supply a total of 100 kW to US satellites. Each of these quadrants is 30 ft wide and 100 ft long.

FIG. 13.2 One of four panels that will provide power to the US space station. Each of these panels is approximately 30 x 100 ft and will provide 25 kW. This array will “fold up” for launch and be deployed after the correct orbit is reached. Photograph courtesy of Lockheed Missiles and Space Co., Cupertino, California.

Solar cells used in space utilize the same principles of operation as described in Chapter 12 but are influenced by the following additional constraints:

Operation is with no maintenance.

Operation is in an intense radiation field of high energy particles.

Launch costs make light weight, high-efficiency cells attractive.

Lack of area and complexity of deployment makes high packing factors attractive.

The only cell cooling is by radiation.

These constraints are listed roughly in their order of priority and the first two have received almost all of the development effort. A high reliability requirement has imposed extremely tight specifications, inspections, and testings at each step of the processing. Each cell in an array must individually be tested in a solar simulator and display the identical voltage and current characteristics. A speck on the back of a cell that did not get covered with the evaporated contact or a chip off the corner of a cell, even if it has no effect on the electric output, would be grounds for rejection. A particularly trou­blesome area in reliability has been with the contact leads between cells. These interconnects must be reliably soldered on both the top and bottom contacts of the cells without shading a cell, increasing the space between the cells, or shorting out cells. They must withstand the vibrations of launch and the thermal cycles in space between no sun and full sun with no cooling other than radiation. They also must be transported and stored before launch in high moisture atmospheric conditions without corrosion or degradation in any way. Titanium-silver contacts, using flexible interconnects, have proven very reliable.

Solar arrays in space are bombarded by radiation of all types and energies. Meteorite penetration has not proven to be a problem as was anticipated before the start of the space program. The charged particles in the radiation belts around the earth have provided the major degradation of the cells. Now that these belts have been mapped, the exposure to a satellite in a particular orbit can be fairly well estimated. These belts contain electrons with energies over 1 MeV and protons with energies over 30 MeV. These high energy particles impinging on a solar cell physically dislodge atoms from their crystalline sites and create large numbers of vacancies and interstitial atoms. The crystal imperfections thus created act as recombina­tion sites for the free electrons and holes lowering the overall efficiency of the cells. These types of defects could be fairly well annealed out, but this would be very difficult to do remotely in space. In order to minimize the effect of radiation, there has been rather extensive experimental investiga­tions using particle accelerators to bombard solar cells (Curtin and Statler, 1975). Figure 13.3 shows typical results from these types of experiments (Allison et al., 1975). The general conclusions shown in Table 13.1 from irradiation experiments have been substantiated by experience from flight data. These conclusions have resulted in most space cells being thin (~ 300 /xm), n on p silicon cells made from 10 П-cm base silicon. In addition, they usually have a transparent cover glass over the cells to transmit the photons but which absorbs many of the protons and electrons. Lithium-doped cells have proven to be advantageous for some applications. Lithium has a much higher diffusion rate than most doping materials and seems to anneal some of the radiation damage.

FIG. 13.3 Maximum power output of nonreflective (——- ), violet (—– ), and conven­ tional (—) cells as a function of 1-MeV electron irradiation fluence (Allison et al., 1975).

The processing of silicon solar cells for terrestrial applications was described in Chapter 12 and is similar to the steps for space cells. In the US essentially all of the single crystal silicon is grown with the Czochralski crucible process from purified silicon. Until the last few years, all cells manufactured in Western Europe used float-zone refined silicon. In this

TABLE 13.1

General Conclusions Concerning Radiation Damage to Solar Cel№

High resistivity cells (10 Q-cm) degrade less than lower resistivity cells (1 Q-cm). n on p base cells degrade less than ропи base cells.

Thin cells degrade less than thick cells.

Float-zone-refined silicon cells degrade less than crucible-grown silicon cells when ex­posed to protons and electrons.

Crucible-grown cells degrade less than float-zone silicon cells from photon exposure.

The radiation damage is manifested as a decreased infrared response of the cells.

The violet and CNR cells maintain their advantage over previous cells throughout irradiation. Lithium-doped (p on n) cells recover better from damage than conventional n on p cells at temperatures > 50°C.

a Curtin and Statler (1975).

method a R-F heater coil is moved slowly along a silicon bar maintaining a small molten zone of silicon within it. Silicon atoms preferentially solidify on the crystal behind the zone and the impurities build up in the molten zone as it moves. After several passes, the bar becomes a high-purity single crystal with all of the impurities concentrated at the ends which can be discarded. These single crystal rods are made to contain about 1015 atoms/cm3 of a desired p-type dopant such as boron. This doping density will provide a base resistivity of about 10 Q-cm.

Both crucible and float-zone silicon bars are frequently cut in their long direction in order to obtain a square cross-section bar with dimensions of about 4×4 cm. From this square bar, it is easy to cut individual cells as may be required for a particular satellite specification (e. g., 1 x 2, 2 x 2, or 2×4 cm). The rectangular cells are required to ensure a high packing factor for the cells on the array. The smaller cells are easier to attach to a curved surface. The surfaces of the rectangular wafers must be extremely smooth (lapped) to remove the saw marks and to ensure uniform thickness. The raw cells are subsequently cleaned and etched to remove any mechanical damage caused by previous processing steps. The blanks are then loaded into trays and put in a diffusion furnace to diffuse a л-type dopant, such as phosphorus, into the surfaces to provide the p-n junction. The junction depth is determined by the time and temperature in the furnace. The surface concentration of the impurity reaches about 1020 atoms/cm3, which is ap­proaching the silicon density, before removal from the furnace. Since only one side of the blank is desired to have a junction, it must again be etched on one side to remove the undesired junction.

The cells with junctions are mounted onto receptacles and covered with masks that have slits of the desirable front contact pattern. Figure 13.4

FIG. 13.4 Diffused silicon wafers are being loaded into fixtures for placing into the vacuum chambers for the evaporation of the metal contacts. Photograph courtesy of OCLI, Inc., City of Industry, California.

shows diffused wafers being loaded for application of front grids and back contacts. After being placed in a high vacuum chamber, titanium-silver contacts are evaporated through the front mask (on the junction side of the wafer) and evaporated completely over the back. The evaporation of the contacts does not make a good low resistivity bond to the silicon until it is heat treated (sintered) at about 600°C. The cells are then dipped into solder which will only adhere to the Ti-Ag deposition. The cells are again mounted onto fixtures and placed in a high vacuum for the evaporation of the anti­reflective coating.

Most space cells up through 1975 used SiO thin films as an antireflective coating. With the development of the violet cell that responds to wavelength down to 0.3 /im, the SiO was not transparent enough. Most cells using violet cell technology use Ta205 which is transparent to below 0.3 /от. Sometimes TiO* is used which has its absorption edge between SiO and Ta205. The need for the higher transmission in the 0.3 to 0.4 /ші range is seen in Fig. 13.5 that shows a comparison with the relative spectral responses of the con­ventional cell, the violet cell, and the CNR cell (Allison et al, 1975).

After the antireflective coating, the cells are essentially complete and must undergo a series of tests to meet specifications. They are checked for their physical dimensions to meet given tolerances. They are each electrically tested to ensure identical and specified J-V characteristics. There is an adhesive tape peel test performed. The peel test is to see if the antireflective coating or the contacts can be pulled off. If the cell meets all of the specifica­tions, then it is ready for assembly.

The cover glass is applied to the individual cells before they are soldered together. The added weight and shadowing effect these coverslips have on

WAVELENGTH Um) FIG. 13.5 Spectral response of nonreflective (——– ), violet (—— ), and conventional (—) cells (Allison et ai, 1975).

the cells is justified by the more stable output of the cells since they are shielded from most of the radiation. The coverslips are usually about the same thickness as the cells, i. e., 300 /лп, and are typically made of Corning 0211 microsheet or Corning fused silica. Sometimes the cover glasses have their own antireflecting coating on the outside surface and occasionally an optical filter on the underside to protect the coverslide adhesive from the solar UV irradiation. A clear adhesive must be between the coverslide and the cell to provide mechanical strength and optical coupling. Silicon adhe­sives are often used for bonding. Any degradation, such as discoloring, in either the coverslide or the adhesive due to the ultraviolet photons or the particle bombardment will, of course, result in a decrease of cell output. Figure 13.6 shows the adhesive being applied to some 2 x 2 cm silicon cells followed by the placement of the cover glass.

(a)

(b) FIG. 13.6 (a) A clear adhesive is applied to bond the coverslip onto the cell and provide optical coupling, (b) The coverslip is applied to provide radiation shielding for space cells. Photograph courtesy of TRW Systems, Inc., Redondo Beach, California.

The individual cells are usually soldered together into small panels which can be later soldered into larger arrays. The interconnects that go from the top contact on one cell to the bottom contact of an adjacent cell can be seen in Fig. 13.7. Also seen is the application of the adhesive that bonds the cells onto the substrate material.

FIG. 13.7 Adhesive is applied to bond back of the cells onto the substrate. The inter­connecting leads between the cells can be seen. Photograph courtesy of TRW Systems, Inc., Redondo Beach, California.

FIG. 13.8 The ESRO IV satellite using AEG-Telefunken cells that provide 60 W of power from the cells, which are mounted directly on the body of the satellite. Photograph courtesy of the European Space Research Organization, Noordwijk, The Netherlands.

The shapes of the final arrays vary a great deal depending on the power required for the mission, the physical constraints on the packing of the array for launch, and the deployment geometry in space so that it does not in­terfere with the mission function of the satellite. If the power requirement is not too high, the cells may be mounted directly onto the satellite’s surface (a “body” mount) as seen in Figs. 13.8 and 13.9. This type of mounting is not an efficient use of cells since only about one third of the cells will be facing the sun at any given time, but it does not require any tracking or locking onto the sun’s position. For higher power satellites one is forced to use extended surfaces such as “paddles” for the arrays as seen in Fig. 13.10. The French satellite shown in Fig. 13.11 uses a combination of body-mounted cells and paddles.

FIG. 13.9 A NASA satellite using body mounted cells from OCLI. Each of the flat section panels contains 770 cells and can produce 19.2 W. Each panel has two circuits of 7 parallel by 55 series connected cells. Photograph courtesy of TRW Systems, Inc., Redondo Beach, California.

FIG. 13.10 The US Mariner 64 satellite using paddle mounted cells to provide more area for the cells than is available on the body of the satellite. Photograph courtesy of Jet Propulsion Laboratory, Pasadena, California.

FIG. 13.11 The French satellite EOEL which uses a combination of body mounted and paddle mounted cells. Photograph courtesy of Centre National d’Etudes Spatiales, Paris, France.

FIG. 13.12 This 1500-W flexible array can be rolled up like a window shade for launch. The flexible substrate is Kapton. Photograph courtesy of Air Force Aero Propulsion Laboratory, Wright-Patterson AFB, Ohio.

FIG. 13.13 USAF satellite using the roll up flexible arrays. Photograph courtesy of Air Force Aero Propulsion Laboratory, Wright-Patterson AFB, Ohio.

For large power requirements, the storage and deployment in space of the large arrays present a sizeable problem. There are two techniques that are commonly used for these panels. One method is shown in Fig. 13.12 which uses a mechanism like a window shade. This 1500-W flexible array is rolled up into its stored position on an 8-in. diameter magnesium drum. The two 16 x 5.5 ft panels of the flight model (see Fig. 13.13) contain 34,500 solar cells bonded to a Kapton fiberglass laminated substrate material. The other technique for deployment of large panels is to use an accordion-type packing as shown in Fig. 13.14.

FIG. 13.14 The ESRO fold-up array in which the panels fold in an accordion fashion for storage during launch. Photograph courtesy of the European Space Research Organization, Noordwijk, The Netherlands.