SPS Built of Lunar Material SRA Report for SSI

SOLAR POWER SATELLITE BUILT OF LUNAR MATERIALS Study conducted by: SPACE RESEARCH ASSOCIATES, INC. 22907 - NE 15th Place Redmond, Wa. 98053 FINAL REPORT Sept. 21, 1985 Study conducted for: SPACE STUDIES INSTITUTE 285 Rosedale Road Princeton, New Jersey 08540


FOREWORD This final report was prepared by Space Research Associates, Inc. (SRA) for the Space Studies Institute. It presents the results of work done between June 1984 and July 1985. Interim reports were presented to SSI in September 1984, January 1985, and June 1985. Significant contributions to this study were made by SRA personnel, five advisers, a consultant on lunar geology, and several volunteers from Seattle chapters of the L5 Society. Space Research Associates: Paul DuBose - study manager, GaAs concentrator Dani Eder - microwave systems Scott Finfrock - passive thermal control, silicon planar, GaAs cone. Hugh M. Kelso - study business manager, graphics, production Amjad Shariatmadar - Brayton & Rankine engines, structure Brian Tillotson - TPV, silicon planar, final report editor Steve Weiss - Stirling engine Consultant: Dr. Steve Gillett - lunar materials availability Advisory Board: Dr. Adam Bruckner - University of Washington Dr. John Freeman - Rice University Donald Hervey - Eagle Engineering Dr. Tom Mattick - University of Washington Gordon Woodcock - Boeing Aerospace Company L5 Volunteers: Carl Case - proposal preparation Art LaPella - proofreading, consistency Beth Means - proofreading SRA thanks the members of the advisory board for their time and their many helpful suggestions. The advisory board’s experience and judgement provided valuable guidance for the study team. A preliminary summary of the study was presented at the 1985 Princeton Conference on Space Manufacturing, which resulted in constructive input from many of the attendees. Dr. Gillett’s advice on the availability of relatively rare lunar materials was of great help. The work of the volunteers, especially Art LaPella, significantly improved the clarity and consistency of the reports. Thanks are also given to Joel Sercel of JPL for several helpful conversations and references.

1. EXECUTIVE SUMMARY High cost has been a major obstacle to development of a Solar Power Satellite (SPS) system. A major part of this cost is the expense of transporting material from Earth to geosynchronous orbit (GEO). The energy required to transport lunar material to GEO is less than 8% of that for Earth material. In addition, launch from the Moon should be more efficient than launch from Earth due to low lunar gravity and the effective lack of a lunar atmosphere. Thus, the cost of delivering materials from the Moon to GEO might be about one-fiftieth of the cost to deliver equivalent materials from Earth.(4) This suggests that the use of lunar materials could significantly reduce the cost of building an SPS system. General Dynamics Convair has studied the use of lunar resources for SPS construction.(1) The rules of that study allowed only minor changes to an earlier Boeing design(2) which had assumed that all materials came from Earth. Thus, the resulting design was not optimized for the use of lunar materials. Even so, it required only 10Z as much Earth mass as the Boeing design.(1) This document is the final report of Space Research Associates’ SPS design study emphasizing minimal use of Earth-supplied mass to achieve low cost. The ground rules of the study are listed in Table 1-1. The study objective was to provide a basis for comparison of alternative SPS concepts and subsystems in the context of lunar material utilization. TABLE 1-1 STUDY GROUND RULES o Any reasonably abundant chemical element present in lunar material could be mined, refined, and transported to the SPS construction site. All other materials must be transported from Earth. o The primary design requirement is minimal non-lunar mass. Low total mass is only a minor concern. o Scope of the study does not include design of transportation systems, materials processing, or assembly systems. o To enhance commercial feasibility, undemonstrated technology is to be avoided in the design wherever possible. o Only microwave transmission of power is to be considered. The peak intensity at the ground is not to exceed 300 W/m*2. Elements abundant on the Moon are aluminum, calcium, iron, magnesium, oxygen, silicon, and titanium. Less abundant (< 1Z) but potentially available are chromium, potassium, manganese, and sodium; though using them

does not seem unreasonable, there is assumed to be some technological or economic risk in planning to use these materials. A special case is nickel, which is a component of some micrometeorites. Though present in only tiny amounts in lunar soil, these micrometeorites are high quality steel and should be easy to separate magnetically from the soil.(7) Thus, using nickel steel is assumed to be a reasonably good risk. 1.1 SUMMARY OF RESULTS This section presents the broad, satellite-level results of the study, followed by technical results for each major satellite system. It is found that an SPS can be designed which uses less than one percent as much non-lunar material as the Earth baseline SPS, a factor of ten improvement over the results of the General Dynamics study. The total mass of the system is about eight percent greater than that of one made from Earth-derived materials. The best design uses silicon photovoltaic cells for power conversion. Its structure is primarily aluminum and uses aluminum oxide coatings for thermal control. Radiators are more than 99 percent composed of lunar material. A flywheel system is used for energy storage during eclipses. The design is suitable for largely automated construction. FIGURE 1.1-1, TOTAL MASS TO NON-LUNAR MASS COMPARISON

Figure 1.1-1 compares total mass and non-lunar mass of the Boeing SPS design, the General Dynamics SPS design, and the Space Research Associates SPS design. Figure 1.1-2 compares the estimated cost per SPS for each design, based on the assumption that non-lunar material is fifty times more costly than lunar material. The cost units are equivalent to thousands of metric tons of lunar material. Growth and contingency analyses were outside the scope of this study, so the mass estimate for the SRA design assumes the same growth factor, 26.72, as the Boeing design. FIGURE 1.1-2, ESTIMATED SPS COST COMPARISON An alternative to silicon cells is to use gallium arsenide cells with solar concentrators. This roughly triples total SPS mass, but non-lunar mass remains less than two percent of that for the Earth baseline. Four other pox^r conversion systems were investigated: thermophotovoltaic (TPV), Brayton, Rankine, and Stirling. These were found to use significantly more non-lunar material than either the silicon system or the gallium arsenide system. It appears that a lens made of aluminum mesh could economically increase the effective aperture of the SPS antenna. If this proves to be correct, the lower limit on SPS power could be as low as 200 MW per satellite. This is an area for further study.

1«1«1 Solar Power Conversion Systems Solar power conversion systems include subsystems which collect and intensify solar radiation and those which dissipate heat from the power conversion process. Since design of these subsystems is largely independent of the conversion systems they support, they are discussed separately after the power converters. Systems for converting sunlight into electricity may be categorized as those which require active cooling and those which do not. Two systems which use passive cooling were investigated: silicon planar and gallium arsenide concentrator. Four systems which require active cooling were investigated: thermophotovoltaic (TPV) conversion, Brayton cycle, Rankine cycle, and Stirling cycle. At least one SPS design was developed and evaluated using each system. As stated above, the passively cooled systems were found to require significantly less non-lunar material. Total specific mass and non- lunar specific mass for each power conversion system are compared in Figs. 1.1-3 and 1.1-4. These figures do not include masses of other SPS subsystems. FIGURE 1.1-3, TOTAL SPECIFIC MASS COMPARISON

FIGURE 1.1-4, NON-LUNAR SPECIFIC MASS COMPARISON Silicon Planar In the silicon planar configuration, non-concentrated sunlight falls onto a planar array of silicon cells. Cells are passively cooled by radiating from their own surfaces. They are partly protected from harmful radiation by thick silica glass windows and substrate. Cells are periodically annealed to repair damage from energetic particles. Previous silicon planar SPS designs have used plastic tape to connect solar cell panels in a flat array. The design developed in this study uses aluminum wire *o connect and support the panels, reducing non-lunar mass. The panels are angled to face the sun more squarely, which improves efficiency. The silicon planar design uses virtually no non-lunar materials. It is also the least massive conversion system overall, as shown in Figures 1.1-3 and 1.1-4. The mass shown may be optimistic by a factor of two if radiation damage cannot be repeatedly annealed out of silicon cells, but the silicon planar concept would still require the least non-lunar material. Gallium Arsenide Concentrator The gallium arsenide (GaAs) concentrator concept uses GaAs cells to produce electricity from concentrated sunlight. Neither gallium nor arsenic is abundant on the Moon, but GaAs cells can be used with highly concentrated sunlight. Thus, the mass of the cells - and hence the mass of non-lunar material - need be only a small fraction of the SPS mass. GaAs cells are more resistant to radiation damage than silicon, but still require periodic annealing or thick cover glass. The design used here has a Cassegrain optical system with a concentration ratio of 260. Al uni num primary and secondary mirrors focus sunlight onto a small GaAs cell which is fixed to the center of the primary mirror. Excess heat from the cell is radiated from the back surface of the primary, which constrains the design to use many small units. The cell operates at 200 C, producing 4.04 watts per unit. As shown in Figures 1.1-3 and 1.1-4, this design is second only to silicon planar in minimizing non-lunar materials. Further optimization of the design could probably reduce the GaAs system's total mass by a factor of at least two. Thermophotovoltaic (TPV) A TPV converter moves the spectral peak of concentrated sunlight from the visible region to the infrared, which is more efficiently converted to electricity by photovoltaic cells. The cells require active cooling. A parametric model of a TPV converter using silicon photovoltaic cells was developed. The result of optimizing this model for minimal non-lunar mass is shown in Figures 1.1-3 and 1.1-4. The high mass of the TPV system is due to active cooling of the silicon cells. The cooling system also accounts for more than 90% of the non-lunar mass. An advantage of TPV conversion over other photovoltaic systems is that the cells are enclosed within a substantial structure which protects them from harmful radiation, so no annealing is required. TPV technology is advancing rapidly, so it should be considered in future SPS studies. TPV would be especially appealing if a cooling system with very little non-lunar mass were developed. Brayton Cycle High temperatures in the Brayton design yield high efficiency and low total mass but make lunar material substitution difficult due to the need for advanced materials. The degree of lunar material substitution possible was conservatively pstimated for two different temperature cvcles. The hot cvcle has a turbine inlet temperature of 1617 K; the cold cycle, 1100 K. Both use helium as the working fluid. Estimates of total specific mass and non-lunar specific mass for the two temperature ranges are shown in Figures 1.1-3 and 1.1-4. The low temperature cycle is found to use less non-lunar material, but the increase in total mass is forty times larger than the decrease in non-lunar mass. Since neither this figure nor the fifty-to—one cost advantage of

lunar materials is very precise, neither temperature cycle has a clear cost advantage over the other. Rankine Cycle Two working fluids, potassium and steam, were evaluated for the Rankine engine. Each fluid was evaluated with two temperature ranges. The more promising specific mass estimates are shown in Figures 1.1-3 and 1.1-4. The high temperature steam cycle was found to require least non-lunar material among the Rankine designs, but the Rankine cycle requires much more non- lunar material than any other option. Stirling Cycle Performance of the Stirling engine was estimated using the goals of a current NASA demonstration project. As shown in Figures 1.1-3 and 1.1-4, its specific mass is rather high. However, the Stirling requires little non-lunar material because of its low operating temperatures. The state of Stirling engine technology is advancing rapidly, so it should not be ruled out for SPS applications. Concentrating Reflectors Standard technology for large reflectors in space is vapor deposited aluminum on facets of Kapton plastic. A similar design which uses vapor deposited aluminum on facets of aluminum foil was selected. This requires virtually no non-lunar material. Radiator Systems Three radiator concepts which might be used to dissipate heat from active cooling systems are the heat pipe radiator (HPR), the liquid droplet radiator (LDR), and the moving belt radiator (MBR), The heat pipe radiator was selected for all active cooling needs in this study because it represents least technological risk and requires little non-lunar material. The moving belt radiator may have even lower non-lunar material requirements, but was considered too risky to use in this study. Nonetheless, the MBR concept is promising and deserves further study. 1.1.2 Structural Material Aluminum was found to be the best structural material for this study because it is well understood, requires few non-lunar alloying agents, has good structural properties, and has relatively low mass per unit strength. Passive thermal control systems should suffice to keep eclipse-related thermal expansion within acceptable bounds.

1.1.3 Power Distribution The Power Distribution System (PDS) controls, conditions, and transmits power from the conversion system to the microwave transmission system. It also provides energy storage for eclipse periods and handles fault detection and isolation. Major PDS subsystems are discussed below. Main Buses Four options were considered for the main power buses: sheet aluminum conductors, refrigerated buses, superconducting buses, or microwave transmission. The sheet aluminum option requires essentially no non-lunar material, and resistance losses can be made arbitrarily small by increasing the width or thickness of the bus. It is also the simplest to manufacture. Thus, the sheet aluminum option was selected. DC-DC Converters Previous studies(l,2,5) have used an oscillator-transformer-rectifier system for DC to DC conversion at the microwave antenna. Transformers are about 98* efficient, making the overall conversion process about 96* efficient. Transformers require active cooling, so non-lunar material is needed for pumps and coolant. Modern power electronics provide DC-DC conversion at over 98* efficiency without using transformers.(6) This increased efficiency reduces non-lunar mass needed for cooling and reduces the mass of the power conversion system, so this conversion technology was selected. Energy Storage Stored energy is required during eclipses to maintain essential systems and to keep RF amplifier cathodes hot. Three energy storage technologies were considered: nickel-hydrogen, sodium chloride, and counter-rotating flywheels wrapped with glass fiber. The flywheel system requires much less non-lunar material per stored kilowatt-hour than the other systems. The energy density of a glass-wrapped flywheel was conservatively estimated from the best demonstrated performance of a steel flywheel. Electrical Slipring All major SPS studies to date have used a slipring-and-brush assembly as the electrical rotary joint. Typically these designs have used silver molybdenum sulfide brushes on coin silver to minimize abrasion, drag, and electrical losses. However, this calls for a large mass of non-lunar materials. The silver in the sliprings (10.6 metric tons for a 5 GW SPS) could cause supply problems on Earth. To decrease the mass of non-lunar material in the electrical slipring, a thin plating of coin silver on an aluminum slipring was selected. This design reduces the requirement for a 5 GW SPS from 11.8 metric tons of coin silver to 1.2 metric tons and gives equivalent electrical performance with less mass. It also reduces the total slipring mass.

1.1.4 Microwave Transmitter The microwave transmitter includes RF amplifiers, waveguides, power distribution systems, phase control systems, thermal control systems, and structure. It may also include microwave optics such as a lens or reflector. Major components of the microwave transmitter are described below. RF Amplifiers Two RF amplifiers, a klystron and a magnetron, were redesigned for minimal use of non-lunar material. The magnetron was found to contain more non- lunar material, but its higher efficiency reduces the size of the power conversion system. Thus the magnetron is preferred if the power conversion system contains much non-lunar material, and the klystron is preferred if the power conversion system contains little non-lunar material. Using this criterion, the klystron is appropriate for the silicon planar and gallium arsenide concentrator systems; the magnetron is preferred for all others. A third RF amplifier concept, the gyrocon, was dropped from consideration due to its low state of development. Waveguides The waveguide designed by General Dynamics(3), which has a conductive coating of aluminum in a foamed glass waveguide, was selected. It uses no non-lunar materials and remains within the close length tolerance over a wide temperature range. The alternative, an all-aluminum waveguide, could not maintain the required length in the severe thermal environment of the microwave transmitter. Microwave Lens A microwave lens design was developed so that SPS systems smaller than 2 GW could be economical. The lens is a Fresnel lens made of layers of aluminum wire mesh, and requires very little non-lunar material. Assuming that a reasonable design limit is that the lens constitutes no more than 50* of the SPS mass, then a power level as low as 200 MW can be achieved. Further study is required to verify the effectiveness of the lens design.

1.2 REFERENCES 1. Bock, E., et al.,Lunar Resources Utilization for Space Construction, Vols. 1-3, General Dynamics Convair, San Diego, Cal., NAS9-1556O, April 1979. 2. Woodcock, G. R., et al.,Solar Power Satellite - System Definition Study, Vols. 1-3, Boeing Aerospace Company, Seattle, Wash., NAS9-15196, Dec. 1977. 3. Bock, 1979, Vol. 3, p. A-29. 4. Woodcock, G. R., "Transportation Networks for Lunar Resources Utilization", Boeing Aerospace Company, Huntsville, Alabama, 1985. 5. Hanley, G. M., Satellite Power Systems Concept Definition Study, Vols. 1-7, Rockwell International, Downey, Cal., NAS8-32475, Sept. 1980. 6. Lauritzen, Peter 0., Professor of Electrical Engineering, University of Washington, Seattle, WA, personal communication. 7. Gillett, Steve, "Elements on the Moon", 1984.

2. SOLAR POWER CONVERSION SYSTEMS Solar power conversion systems include transducers which convert sunlight to electrical Energy, cooling systems to dissipate waste heat from the conversion process, and optical systems which filter or concentrate sunlight. Several power conversion concepts considered in this study require large concentrating reflectors and large active cooling systems. These subsystems are discussed in sections 2.7 and 2.8. Primary structure is discussed in chapter 3; it is not included in the mass estimates for any of the conversion systems. Most of the power conversion designs discussed in this chapter are sized to produce 9 GW of electricity at the SPS bus. This is assumed to correspond to 5 GW of usable electricity on the ground, i.e., overall electrical efficiency is assumed to be 55.62. This assumption is quite conservative, since earlier studies have achieved higher efficiency. Two categories of solar power converters were considered in this study: photovoltaic (PV) systems and heat engines. A third category, thermionic conversion, was not considered because of its low efficiency, high mass, and large non-lunar material requirements. Photovoltaic systems convert light energy directly to electrical energy. In all photovoltaic systems considered in this study, electricity is produced when photons of light form electron-hole pairs in a semiconductor material. The photovoltaic systems considered here are silicon planar, gallium arsenide concentrator, and thermophotovoltaic (TPV). Silicon planar was found to require least non-lunar material of all conversion systems considered, as shown in Table 2-1. TABLE 2-1 PERFORMANCE OF POWER CONVERSION SYSTEMS System Max. Temp. Z Eff. Total kg/kW Non-lunar kg/kW Si Planar N/A 14.34 2.22 0.0000001 GaAs Cone. N/A 9.5 6.93 0.0104 TPV N/A 20.5 13.0 0.08 Brayton 1617 X 43 3.01 0.43 Brayton 1100 X 35 10.77 0.234 Rankine, steam 1644 X 42 6.80 1.23 Rankine, potass. 1600 X 31 3.14 1.47 Stirling 720 X 25 34.0 0.66 Photovoltaic systems require less non-lunar material than heat engine systems and are simple to design and build. The semiconductor cells must be protected from radiation or periodically annealed to repair radiation damage- The study team assumed that repeated annealing is effective, but was unable to determine whether this is the case. This matter is discussed further in sections 2.1 and 2.2.

Heat engines convert thermal energy into mechanical energy which can be converted to electricity by a generator. Solar power systems based on heat engines use concentrated solar radiation to heat a working fluid. Some of the energy of the working fluid is extracted by a mechanical device such as a turbine and the remainder is rejected by a radiator. Use of these systems in space can offer good efficiencies but can also present problems with complex machinery, temperature limitations, and maintenance. Three types of heat engines were evaluated in this study: Brayton cycle, Rankine cycle, and Stirling cycle. A fourth type of heat engine, the magnetohydrodynamic (MHD) generator concept, appeared attractive due to mechanical simplicity and high theoretical efficiency. This concept was not thoroughly investigated, as it was felt that the high temperature required in the working fluid would make constructing such an engine from lunar materials impractical. A variation on the MHD concept, the liquid metal MHD generator, may prove suitable for construction from lunar materials after further development. The Brayton cycle was evaluated at two ranges of temperatures. The lower temperature range had lower non-lunar mass despite lower efficiency and higher total mass. The high temperature cycle required many non-lunar materials in hot and/or fast moving parts of the engine. Neither has a clear cost advantage, assuming a cost ratio of fifty to one for non-lunar vs. lunar materials. Many estimates of lunar material substitution for the Brayton system were based on intuitive assumptions, since detailed information on complex high-temperature machinery often could not be found. The Rankine cycle was studied using two different working fluids: potassium and water. The steam cycles gave better efficiencies but required more mass. The least non-lunar mass for the Rankine was achieved with the high temperature steam cycle. For both working fluids, the low temperature cycles required more non-lunar mass and more total mass. As for the Brayton system, many estimates of lunar material substitution in the Rankine were based on judgement rather than data. The Stirling engine has only recently been seriously considered for space power systems. It is potentially easier to construct from lunar materials than either Brayton or Rankine, but is more massive overall than either. Because relatively little has been done with the Stirling as a space power system, this study considers only one temperature cycle of the Stirling engine.

2.1 SILICON PLANAR 2.1.1 Introduction The silicon planar concept uses a large, modular array of silicon solar cells facing the sun. The sunlight is not concentrated nor is any attempt made to alter the spectrum. The cells are cooled by radiation from both surfaces. The cells are protected from harmful particle radiation by a thick window in front and a thick substrate behind. A modular box-frame structure maintains the shape of the solar array. Because the silicon planar concept does not require concentrated sunlight, it has an advantage over all other power conversion systems in that the SPS does not have to be precisely oriented to face the sun. Rather than being oriented perpendicular to the ecliptic plane, the SPS can be oriented perpendicular to the plane of its orbit. This results in reduced stiffness and control requirements, so the structure and attitude control systems are less massive than for other conversion systems. Several other studies have extensively reviewed the silicon planar option(l - 4). Most details presented here are based on the Boeing reference design(l,2). Lunar material substitutions for the cell covers, substrate, and cell interconnects are based on the studies by General Dynamics(3) and MIT(4,p3.5). Cell panels and the panel support system have been redesigned for better efficiency and reduced non-lunar mass. Silicon solar cells have been used for power production on satellites for many years. A 1984 Space Shuttle flight demonstrated a large solar array which is similar to the proposed SPS array. The vast amount of research that has been associated with silicon cells in space has resulted in a comparatively low technological risk for the silicon planar option. However, the useful lifetime of silicon cells in the GEO environment is questionable because repeated effective annealing silicon cells has not been demonstrated. High temperature silicon cells being developed now may solve this problem.(11) If repeated annealing cannot be effective, some redesign of the silicon SPS will be needed. Two possible changes are to use thicker cover glass and substrates, or to use a larger array area. Using a larger array is preferred, since doubling array size gives the same end of life (EOL) performance as increasing cover glass thickness by a factor of ten.(12) 2.1.2 Design Description The individual solar cells are 50 micron wafers of single crystal silicon fitted with aluminum contacts. These cells are connected into an 18 x 14 cell array which is placed between two plates of silicon dioxide glass to form a panel (Fig. 2.1-2). The glass serves to protect the cell from radiation damage as well as provide structural support. The 75 micron front plate is the optical cover and the 50 micron back plate is the cell substrate (Fig. 2.1-1). The optical cover is grooved to refract light around the grid fingers, increasing the effective efficiency by approximately 10X(2,p56).


In earlier studies, the panels were connected and supported by Kapton tape. Substitution of aluminum tape for Kapton was considered, but the tape adhesive still required a large non—lunar mass. It was decided to use a grid of aluminum wires to support the panels. Panels are mounted in front of the grid. Unlike earlier studies, this design puts no tensile stress on the panels. Cell substrate plates are 2 cm wider than cover plates, with holes drilled near the two sides. Support rods from the grid wires are connected to the panels at these holes. The panels are canted at 12 degrees to the SPS’s longitudinal axis (Fig. 2.1-4) to reduce solstice cosine losses. The cells never face more than 12 degrees away from the sun because the SPS is rotated 180 degrees at each equinox. At edges of the grid the support wires are connected to catenary cables which attach to the primary bay structure (Fig. FIGURE 2.1-3, PANEL ARRAY AND PRIMARY BAY STRUCTURE

FIGURE 2.1-4, PANEL DETAIL It is assumed that the silver grid fingers of the original design can be replaced with aluminum, based on Wang’s results(5,pp580-583). Wang found that silver contacts could be replaced by copper if a layer of aluminum were used to prevent copper from diffusing into the silicon. In this design, the aluminum layer is thicker so no copper is needed. The base efficiency for these cells is assumed to be 15.752 at AMO (Air Mass 0, refers to the solar spectrum unaltered by atmosphere, such as would be found at geo-synchronous orbit) and 25C, as in the Boeing study(2,p57). Allowing for increased efficiency due to the grooved cover plate and losses due to mismatch, radiation degradation, etc. (see the minimum efficiency is taken to be 14.342. 2.1.3 Design Summary Table 2.1-1 presents a design summary for an SPS which delivers 9 GW to the spacecraft bus. The design is based on standard panels from the Boeing reference design which have been modifieu to include various material substitutions. The panel design is detailed in the technical discussion that follows. Note that the mass summary includes the catenary tensioning system but that no attempt has been made here to estimate the mass of the supporting structure.

TABLE 2.1-1 MASS ANALYSIS OF 9 GW SILICON PLANAR SYSTEM 2.1.4 Silicon Planar Technical Discussion Cell Efficiency Modifiers The silicon cells used in this report are assumed to operate with the same base efficiency as those used in the Boeing study(2,p57). The loss factors are also assumed to be the same, except for the summer solstice cosine loss. The efficiency values and losses used are shown in Table 2.1-2. The figure for non-annealable radiation degradation is grossly uncertain. The value from the Boeing study was assumed for comparison purposes. TABLE 2.1-2 SILICON PLANAR EFFICIENCY MODIFIERS Silicon Solar Calls The Boeing reference design (1,2) assumed that 50 micron thick silicon solar cells with an efficiency of 15.75X would be available by 1985(2,p52). This has proved to be somewhat optimistic and such a cell has been projected for 1990 in this study. The state of the art in high efficiency silicon solar cells is changing so rapidly that any attempt to predict future achievements is risky at best. In any event it seems unlikely that standard silicon cell technology will produce efficiencies in excess of 20% for this application. This would imply that the reference cell used in this study is conservative by a factor of no more than 30% (this assumes, of course, that the projected values for the reference cell are in fact achieved). In the silicon planar option, the solar conversion system constitutes a majority of the overall system mass. Any significant improvement in the efficiency of silicon cells would, therefore, be of great value. The non-lunar mass would not be greatly affected, however, as the conversion system accounts for only a small percentage of the SPS total of non-lunar material. Of potential significance is the maturing of some alternate silicon cell approaches such as multi-gap and amorphous cells. Amorphous cells, to date, have achieved only relatively low efficiencies but are much less massive than standard cells(9,p67). Multi-gap cells offer higher efficiency but at the cost of higher mass(9,pp100-103). In either case it is the mass efficiency that will finally determine their relative merit. While neither of these two technologies are competitive now, it seems likely that one or both will become so in the future. A good example of current laboratory achievements in silicon cell technology appears in a paper by S. Matsuda et al (6). The National Space Development Agency of Japan and the Sharp Corporation collaborated to produce BSFR (Back Surface Field, Back Surface Reflector) silicon cells approximately 50 microns thick that operate at 13.23% efficiency (at AMO and 28 C). In addition these cells were exposed to a 1 Mev electron fluence of 1E+15 e/cm2 after which they were annealed at 60 C for 24 hours thereby restoring them to 82% of BOL (Beginning of Life power). Silicon Planar Dimensions and Masses for a 5 GW SPS This section presents the numerical details of the silicon planar design and the references from which they are taken. Solar Cells(4,p3.5) Material ■ Single Crystal Silicon Thickness ■ 50 microns Length » 7.7 cm Width » 6.4 cm Cover Plate(4,p5.73) Material » Fused Silica Thickness = 75 microns Length » 1.17 meters Width ■ 1.10 meters Substrate(4,p5.73) Material ■ Fused Silica Thickness * 50 microns Length - 1.17 meters Width « 1.12 meters Panel to Wire Connectors Material « Aluminum Radius ■ .28 mm Length » 1.3 meters per panel Support Grid Wires The radius of these wires was chosen to yield the same total mass as the aluminum tape considered earlier. This radius is quite conservative; each wire bears a tension of less than 6 newtons. Material ■ Aluminum Radius ■ .688 mm Length ■ 2.30 meters per panel (average) Cell Interconnects(4,p5.73-5.81) Material ■ Aluminum Mass Estimate » 1.47E-2 kg/panel

Tensioning System(4,p3.4) Material 3 Aluminum Mass Estimate = 3.21E-3 kg/panel System Performance Cells per Panel « 252 Cell Area per Panel 3 1.24 square meters Efficiency « 14.34Z Solar Constant * 1353 W/m‘2 Panel Output * 240.6 W Power Required to Bus » 9.0 GW Panels per SPS 3 37.4 million Panel Area = 49.23 square kilometers Mass Analysis Densities (kg/m3) Aluminum: 2.7E+3 Fused Silica: 2.2E+3 Silicon: 2.36E+3 Panel System Item Mass(kg) Mass(kg) Photovoltaic Cells 1.47E-1 5.50E+6 Cover Plate 2.12E-1 7.93E+6 Substrate 1.45E-1 5.42E+6 Interconnects 1.47E-2 5.50E+5 Support Wires 9.23E-3 3.45E+5 Panel to Wire Connectors 8.65E-4 3.24E+4 Tensioning System 3.21E-3 1.20E+5 TOTAL 5.32E-1 1.99E+7

2.1.5 Silicon Planar References 1. Solar Power Satellite: System Definition Study, Vol. 3, Phase I, Final Report, DI80—25037—3, Boeing Aerospace Company, Seattle, 1979. 2. Solar Power Satellite: System Definition Study, Parts 1&2, Vol. II, Technical Summary, D180—22876—2, Boeing Aerospace Company, Seattle, 3. Lunar Resources Utilization for Space Construction, Final Report, GDC- ASP79-OO1, General Dynamics Convair Division, 1979. 4. Extraterrestrial Processing and Manufacturing of Large Space Systems, Vol. 1, NASA-CR-161293, Space Systems Laboratory, MIT, 9/79. 5. Wang, Jing-xiao, "An Investigation of Copper Contact for Use in Silicon Solar Cells", 17th IEEE Photovoltaics Specialists Conference, 1984, pp. 580-583. 6. Matsuda, S., et al, "Development of Ultrathin Si Solar Cells", 17th IEEE Photovoltaics Specialists Conference, 1984, pp 123-127. 7. Rusch, D., "New Flexible Substrates with Anti-Charging Layers for Advanced Light-Weight Solar Array", Photovoltaic Generators in Space, European Space Agency, ESA SP-140, 1978, pp 41-48. 8. Kutateladze and Borishanskii, A Concise Encyclopedia of Heat Transfer, Pergamon Press, USA, 1966. 9. Zweibel, K., Basic Photovoltaic Principles and Methods, Solar Energy Research Institute, Von Nostrand Reinhold Company, USA, 1984. 10. Advanced Propulsion System Concepts for Orbital Transfer Study, Final Report, Vol. 2, Study Technical Results, Boeing Aerospace Company, Seattle, Wash., D180-26680-2, 1981. 11. Horne, W. E., Boeing Aerospace Company, personal communication, June 20, 1985. 12. Systems Definition: Space Based Power Conversion Systems, Final Report, Boeing Aerospace Company, Seattle, NAS8-31628, November, 1976.

2.2 GALLIUM ARSENIDE CONCENTRATOR SYSTEM 2.2.1 Introduction The gallium arsenide (GaAs) concentrator option is based on replacing the bulk of the solar cells used in a planar configuration with an aluminum concentrator. Such a system would typically have a concentration ratio in the 100 - 1000 range and is therefore suitable to a strategy of minimizing the use of non-lunar materials. Although GaAs is not available on the Moon, its ability to operate efficiently at high concentration ratios and high temperatures allows a system to be designed which employs minimal amounts of non-lunar mass (less than one percent). Like silicon, GaAs cells must be protected from radiation or must be periodically annealed. GaAs is more likely to be annealable than silicon, but questions remain as to whether the process is workable.(18) An advantage of the GaAs concentrator concept is that its mass is not greatly affected by whether GaAs cells can be repeatedly annealed. Two factors account for this: the high radiation resistance of GaAs (roughly three times that of silicon) and the relatively small size of the cells. Thus, very thick cover glass could be added without significantly affecting the mass (total or non-lunar) of the SPS. It was assumed here that GaAs could be annealed. Three major areas were considered in this analysis: the concentrator, the cooling system and the optimization of concentrator dimensions. These areas are addressed in the technical discussion (section 2.2.4), along with supporting comments on the cell itself, the structure and manufacturing. 2.2.2 Design Description The concentrator system used in this study consists of a parabolic dish for the primary reflector and a hyperbolic mirror for the secondary reflector. This makes up a typical Cassegrainian concentrator. In addition a small Compound Parabolic Concentrator (CPC) has been added around the photovoltaic cell as a tertiary reflector. This, along with an oversized secondary reflector, is designed to compensate for imperfections in the mirrors which cause dispersion in the reflected light. (See Fig. 2.2-1) The secondary reflector is located at a point where the radius of the image striking it is the same as that of the photovoltaic cell. This point is very near the focal point of the primary reflector. The two reflectors are connected by four aluminum legs. The Ga’s photovoltaic cell is attached directly to the primary reflector which serves as a radiator. The Design Summary allows for a thin layer of possibly non-lunar material between the cell and the primary reflector to serve as an insulator and/or adhesive. All three reflectors are made of aluminum and have a coating of vapor deposited aluminum to enhance reflectivity. In addition the back surface of the primary reflector has been anodized to increase the emissivity. This

layer of aluminum oxide greatly increases the capacity of the radiator and is a significant fraction of the concentrator mass (see The concentrators are supported by a wire grid with the same mass per unit area as that of the silicon planar configuration. FIGURE 2.2-1, GALLIUM ATJENIDE CONCENTRATOR SYSTEM 2.2.3 Design Summary Table 2.2-1 summarizes the results from the technical discussion that follows. The overall efficiency is 9.5Z. Some other key values are 0.84 reflectivity for the mirrors, 0.8 emissivity for the radiator and 15 percent efficiency of GaAs cells operating at 200 C and with a concentration ratio

of 260. The mass calculations are based on 2.228 billion cells each supplying 4.04 watts to the power distribution system for a total of 9.00 gigawatts. TABLE 2.2-1 MASS ANALYSIS OF 9 GW GALLIUM ARSENIDE CONCENTRATOR SYSTEM 2.2.4 Gallium Arsenide Technical Discussion System Performance Solar Constant 1.353 kW/m2 Collector Area 3.142E-2 m2 Collector Efficiency 9.5Z Cell Output 4.04 W Power Required to Bus 9.00 GW Collectors per SPS 2.228 billion Total Collector Area 70.0 square kilometers Concentrator Design Many types of concentrators were considered, however, the Cassegrainian appears to be far superior in light of the design limits. Lenses were ruled out because they tend to be massive and may require large amounts of non- lunar material. A simple parabolic dish reflector offers some advantage over the Cassegrainian, however, it would add considerable difficulties to the design of the cooling system because the radiator would be facing the sun and would be separate from the primary reflector. The Compound Parabolic Concentrator (CPC) also has advantages over the Cassegrainian, in particular its tolerance to pointing error and its one piece construction.

These benefits, however, are far outweighed by its disadvantages. The CPC, particularly at high concentration ratios, is a long and therefore massive device. In addition its shape does not lend itself to effective radiation of heat and a secondary radiator would be needed. As a result, unless pointing accuracy or precision construction prove to be insurmountable problems for the Cassegrainian, the CPC is disqualified on grounds of weight. Radiator Design Approximately 80 percent of the light absorbed by the solar cell must be disposed of as waste heat. Two methods of achieving this have been considered in detail. The first, a disk radiator, has the advantage of all aluminum construction and simplicity of design but limits the system to small cell sizes and/or relatively low concentration ratios. The second option, heat pipes, allow the use of larger cells and higher concentration ratios but they require significant amounts of non-lunar material and have a considerable technological risk associated with them. Both radiator systems are compatible with a simple annealing scheme. A sheet of reflective material is positioned behind cells to be annealed. This reflects heat back to the radiator, raising the temperature of the cell. This annealing system has not been designed in detail, but should require virtually no non-lunar material. Disk Radiator The major advantage of the disk radiator is that it is already present in the form of the primary reflector. By making the primary reflector perform double duty as both reflector and radiator it is possible to cool the solar cell without any additional structures. As aluminum alone is not an efficient radiator (15,pp2-7), it is necessary to coat the back surface of the primary reflector with a high emissivity material. Aluminum oxide was chosen for this purpose because it has a reasonably high emissivity, consists totally of lunar materials and is easily applied by anodization. The thermal and mechanical properties of A1203 are, however, considerably different from those of aluminum. It was necessary, therefore, to require that the radiator be thick with respect to the A1203 coating. It is estimated that 25 microns of A1203 will be required to give the desired emissivity of .8 (14,ppll98-1201), however this emissivity value is heavily dependent on the method of anodization. A precise knowledge of this value would be essential for a more accurate determination of the mass efficiency of the concentrator. Heat Pipes Two different configurations were considered for the application of heat pipes. The first involved placing several heat pipes as ribs in the primary reflector. This had all the advantages of using the primary reflector as the radiator, most notably little or no reradiation between unit cells.

Unfortunately there were also several problems with this design. The very complex temperature gradient in the radiator was felt to present severe problems to the design of the primary reflector. The resulting complex thermal expansion pattern and fatigue resulting from thermal cycling seem likely to degrade the optics. In addition, as the radiator design (radial heat pipes on a one sided disk) is inefficient, it was necessary to make the dish relatively thick (and therefore massive) to minimize the temperature drop. Finally, it appeared that mass efficiency increased significantly as the number of heat pipes increased and as concentrator size decreased. Optimizing in this way alleviated the first two problems to some degree but also decreased the minimum heat pipe radius. For a unit cell radius of 10 - 100 cm the heat pipe dimensions became so small as to be impractical. Larger heat pipes could be used but that would incur a severe mass penalty. For these reasons focus was placed on the second option, a single finned heat pipe projecting axially back from the primary reflector. The major benefits of this approach are simplicity and economy of scale (using a larger heat pipe). At the same time, however, analysis of the reradiation became difficult. Reradiation between fins (taken here to be coaxial disks) on a single pipe was easily evaluated (6,p635) but for radiation between concentrators no simple approximation was found. As a standard of comparison it was estimated that 25 percent of the heat emitted from a heat pipe/fin assembly would be reabsorbed by other units. Using this assumption and foregoing the requirement that the fins be thick relative to any high emissivity coating gave mass efficiency results that were approximately equivalent to that of the disk radiator. In addition to the problems mentioned above there were some difficulties encountered that were common to both applications of heat pipes. The first was transferral of heat from the comparatively large solar cell to the much smaller radius heat pipe. The most promising method seemed to be wrapping the evaporator portion of the pipe into a spiral disk which would cover the back of the solar cell. There is not much information, however, on complex geometries for heat pipes and this application remains to be proven acceptable. A second problem lies in the choice of a proper structural material I working fluid combination for this temperature (assumed to be approximately 470 K). Water seems to be the most likely candidate for a working fluid, particularly when striving for a low non-lunar mass. Unfortunately, few lunar materials are compatible with water at temperatures above 100 C (1,ppl03-106) because generation of even small amounts of oxides or non-condensable gas can seriously impair the performance of a heat pipe(l,pl03). There is evidence to suggest that some steels are workable but this is not yet sufficiently documented(1,ppll0-lll). This leads to a final problem: the lifetime of the heat pipe. Currently most space applications require lifetimes of 5 to 10 years and even that has proved to be a real challenge for heat pipe design(1,pp!08-110). The 30 year lifetime required by the SPS will place a much greater strain on the heat pipes as well as on the verification process. Construction For analysis, the primary reflector is approximated as a circular dish. In practice, however, it is more likely to be constructed as a square or hexagonal section of a panel to facilitate assembly. As these panels will

serve as radiators they will have a relatively high operating temperature and will therefore be subject to a fair amount of thermal expansion. In addition, due to the occasional eclipsing of the SPS, there will be thermal cycling. This combination will result in stress on the secondary reflector as well as the SPS structure. This is expected to be one of the more severe problems to be dealt with in the mechanical design of the system. The design used here assumes that the primary reflector has constant thickness for ease of manufacturing. A tapered disk would be more efficient as a radiator, reducing total mass by over half. Members of the Advisory Committee believe that an effective means of manufacturing tapered primary reflectors could be found. Thus, the total mass estimated here is probably conservative by a factor of two. Optimization The thickness of the parabolic dish was the limiting parameter for optimization of the concentrator size. System mass appeared to increase almost linearly with thickness, suggesting that thickness should be as small as practical. It was also necessary, however, to make the thickness large compared to that of the high emissivity coating. As a result the thickness was set at 0.25 mm (10 times that of the coating). With thickness fixed, the maximum concentration ratio (CP) at which the cell can be effectively cooled is a function of disk radius. The maximum CP was found to increase as disk radius decreased. High concentration ratio corresponds to low non-lunar mass, so a small disk size is preferred. Scanty data made it infeasible to optimize the solar cell efficiency with respect to concentration ratio and operating temperature. It was assumed that reasonable progress in technology could produce a GaAs cell 50 microns thick that would have an operating efficiency of 15 percent after all losses (see Table 2.1-2) at 470 K and with a concentration ratio in the 100 - 500 range. Hot spots within the solar cell were not considered in detail so an intermediate value was thought to be best for the concentration ratio. A CR of 260 was chosen as it was in the middle of the range and proved to be convenient for calculations. If the desired efficiency can not be achieved with this value a sub-optimum concentration ratio could be used. The mass of the solar cell is a very small fraction of the concentrator mass and if the concentration ratio was decreased by a factor of two or three the non- lunar mass would still be very good. Furthermore it would probably still be within the error bars of the values shown. Gallium Arsenide Concentrator Parameters Primary Reflector/Radiator Design « Parabolic Dish Material ■ Aluminum Radius » 10 cm Thickness « 0.25 mm Focal Length * 40 cm Secondary Reflector Design * Hyperbolic Dish Material * Aluminum Radius * 3.15 cm Thickness * 0.10 mm Distance from Primary » 38.2 cm Supports for Secondary Number = 4 Length = 38.8 cm Radius » .28 mm Tertiary Reflector Design - CPC Material » Aluminum Maximum Radius » 3.15 cm Minimum Radius » 0.62 cm Height - 3.0 cm Thickness » 0.10 mm High Reflectivity Coating Location =« All Mirror Surfaces Material * Vapor Deposited Aluminum Thickness * 1 micrometer Reflectivity » 0.84 High Emissivity Coating Location « Radiator Face Material * Aluminum Oxide Thickness « 25 micrometers Emissivity • 0.8 Glass Cover Plate Material » Fused Silica Radius - 6.2 mm Thickness » 0.2 mm