Presented at the 4th World Conference on Photovoltaic Energy Conversion, Waikoloa, Hawaii, May 7-12, 2006

ADVANCED III-V MULTIJUNCTION CELLS FOR SPACE

Richard R. King, Christopher M. Fetzer, Daniel C. Law, Kenneth M. Edmondson, Hojun Yoon, Geoffrey S. Kinsey, Dimitri D. Krut, James H. Ermer, Peter Hebert, B. Terence Cavicchi, and Nasser H. Karam

Spectrolab, Inc., 12500 Gladstone Ave., Sylmar, CA 91342 USA

ABSTRACT The weight and volume of a stowed solar array have III-V solar cells have become the dominant power always been paramount concerns for photovoltaic power generation technology in space, due to their unparalleled generation systems on satellites, due to the tremendous high efficiency, reliability in the space environment, and cost of lifting a payload into orbit. Now III-V multijunction ability to be integrated into very lightweight panels. As cell technology seems poised to combine the stellar remarkable as these attributes are, new types of space III- efficiencies of this type of cell with the high specific power V solar cells are continually reaching new heights in (W/kg) and very small stowage volume of flexible performance. Commercially-available multijunction solar photovoltaic blanket arrays. cells with 30% conversion efficiency under the AM0 space Reliability of space cells under the extreme conditions spectrum are just around the corner. Understanding of of launch and space operation is also critically important. radiation resistance and thermal cycling reliability has Radiation exposure, thermal cycling, vibration, atomic reached levels never before attained, and is resulting in oxygen, contamination from volatile materials, and new standards of reliability. A flurry of research activity electrostatic discharge are all part of operating in the has resulted in very-thin, flexible, and extremely space environment, and must therefore be part of the lightweight space solar cells and panels in several groups qualification process for space cells and panels. As a around the world, capable of being folded or rolled into a result of extensive environmental testing, space solar cells smaller stowage volume for launch than has been possible have reached new heights of predictable performance. to date. This approach combines the very high efficiency and reliability of III-V multijunction cells with the thin, High-Efficiency Multijunction Space Solar Cells flexible PV blanket functionality normally associated only with thin-film polycrystalline or amorphous PV technology. Figure 1 charts the AM0 efficiency distributions for the This paper discusses the latest developments in III-V last four generations of multijunction space cells produced space solar cell technology, and explores opportunities for at Spectrolab, Inc., the dual-junction (DJ), triple-junction still higher performance in the future. (TJ), improved triple-junction (ITJ), and ultra triple-junction (UTJ) solar cell products. Each successive generation of high-efficiency space solar cell has not only shifted upward INTRODUCTION in average efficiency, but has also had a striking reduction in width of the distribution, indicating better uniformity and The last decade has seen a remarkable explosion in reproducibility for each new type of these production space the technology of photovoltaic cells designed for use in cells. space [1-3]. During this relatively short time, multijunction 25% DJ Ultra Triple Junction III-V solar cells have become the standard for space TJ ITJ power generation, at long last fulfilling their theoretical UTJ 20% Improved Triple Junction promise to deliver higher efficiency than the best single- junction cells. Early concerns about controlling the composition and thickness of the multiple semiconductor 15% Triple Junction layers that make up a multijunction solar cell, and about Dual Junction the transparency and conductivity of tunnel junction layers, 10% have largely fallen by the wayside. Overcoming these

Percentage of Parts obstacles is now an accomplished fact, but similar 5% questions still arise as we look to the future of space photovoltaics. The historical perspective provided by the 0% development of today's triple-junction cells suggests that 19.0 20.0 21.0 22.0 23.0 24.0 25.0 26.0 27.0 28.0 29.0 we will find ways to tap the higher theoretical efficiency of new multijunction cell designs, and to achieve the level of Efficiency at load (%) control needed to grow even more complex cell Fig. 1. AM0 efficiency distributions of 4 generations of architectures, such as 5- and 6-junction solar cells. Spectrolab space solar cells. In Figure 2 the efficiency of Spectrolab space solar To achieve higher efficiencies, a more radical cells is plotted versus the year those cells were first flown departure from conventional 3-junction space cell on a spacecraft, stretching back to silicon cells in the late architecture is needed, one which has a higher theoretical 1960's. The trend of increasing beginning-of-life (BOL) efficiency ceiling. One such approach is the use of efficiency takes a sharp turn upward in the late 1990's, metamorphic, or lattice-mismatched materials, to tune the with the advent of multijunction cells, increasing by nearly bandgaps of the individual subcells of a multijunction cell one absolute percent in efficiency per year. This growth to the solar spectrum for maximum conversion efficiency rate is projected to continue with the new solar cells in [4-7]. Another approach is to divide the solar spectrum development today: the XTJ solar cell with 30% minimum more finely using a greater number of junctions, such as 5- average AM0 efficiency; and a new generation of cells and 6-junction cells [4,8]. Fig. 4 shows the partition of the with 4 to 6 junctions, the nJ cell with 33% efficiency. solar spectrum by the bandgaps in a AlGaInP/ GaInP/ AlGaInAs/ GaInAs/ GaInNAs/ Ge 6-junction cell. 35% Si Space PV Schematics of 5- and 6-junction solar cells, including a GaAs SJ nJ XTJ 4J to 6J ~1.1-eV GaInNAs subcell lattice-matched to Ge, are GaInP/GaAs/Ge MJ GaInP/GaAs/Ge 30% XTJ shown in Fig. 5. nJ ITJ UTJ GaInP/GaAs/Ge 2.0 1.8 1.6 1.41 eV 1.1 eV 0.67 eV GaInP/GaAs/Ge 100 100 25% DJ TJ GaInP/GaAs/Ge GaInP/GaAs/Ge 90 90 SJ

, 28ºC) 20% GaAs/Ge 2 SJ 80 80 GaAs/GaAs BSF - BSR Si 70 AM0 70 15% Textured Si Thin Si (64 um) 60 AM1.5G 60 m) µ

Large Area Si 2 (135.3 mW/cm AM1.5 Direct, low-AOD 10% (2 x 6 cm) 50 50 GaInP/GaAs/Ge MJ Generation Spectrolab Product Efficiency AM0 40 40 (mA/cm 5% GaAs SJ Generation Si Generation 30 30 0% 20 20 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Current Density per Unit Wavelength 10 10 Date of First Flight Fig. 2. Space solar cell efficiency versus the year each 0 0 new solar cell product was first launched into orbit. 350 550 750 950 1150 1350 1550 1750 1950 Wavelength (nm) The drive toward commercially-available space cells Fig. 4. Wavelength partition of the AM0 space, and with an average efficiency of 30% has resulted in terrestrial AM1.5 Global and AM1.5 Direct, low-AOD prototype quantities of XTJ cells with record AM0 spectra, by the subcell bandgaps of a 6-junction cell. efficiencies, giving confidence that the performance

targets of the XTJ cell can be met relatively soon. The contact AR AR distribution of 125 demonstration cells with average AM0 cap contact AR AR (Al)GaInP Cell 1 2.0 eV efficiency of 30.3%, and 31.0% maximum efficiency, is cap wide-Eg tunnel junction

shown in Fig. 3. The distribution of UTJ cells processed at (Al)GaInP Cell 1 2.0 eV GaInP Cell 2 (low Eg) wide-Eg tunnel junction 1.8 eV the same time is shown for comparison. These prototype wide-Eg tunnel junction AlGa(In)As Cell 2 AlGa(In)As Cell 3 XTJ cells still need to go through qualification, but their 1.7 eV 1.6 eV measured performance gives confidence that space solar wide-Eg tunnel junction wide-Eg tunnel junction Ga(In)As Cell 3 Ga(In)As Cell 4 cells with 30% minimum average efficiency will be 1.41 eV 1.41 eV commercially available soon. tunnel junction tunnel junction GaInNAs Cell 4 GaInNAs Cell 5 1.1 eV 1.1 eV 30 XTJ FS XTJ tunnel junction tunnel junction XTJ 2x2 Ga(In)As buffer Ga(In)As buffer UTJ Controls Avg. = 30.3% nucleation UTJ FS nucleation 25 Avg. = 28.4% Max = 31.0% UTJ 2x2 Ge Cell 5 Ge Cell 6 Max = 29.1% and substrate and substrate 20 0.67 eV 0.67 eV back contact back contact

15 Fig. 5. Monolithic, series-interconnected, 5-junction and 6- junction cell structures. 10 Number of Cells

5 The measured quantum efficiencies of the individual subcells of a 6-junction cell are plotted in Fig. 6, 0 superimposed on the AM0 solar spectrum. The quantum 2 2 2 2 2 2 2 2 2 3 3 3 6 7 7 7 8 8 9 9 9 0 0 1 .6 .0 .4 .8 .2 .6 .0 .4 .8 .2 .6 .0 % % % % % % % % % % % % efficiency and AM0 spectrum are plotted with photon Efficiency at Max Power (AM0, 135.3 mW/cm2, 28ºC) energy on the x-axis, with the subcell response cutoff Fig. 3. AM0 efficiency distribution for 125 XTJ bare cells, showing the bandgap of each subcell clearly. Subcell 1 with areas of 4 cm2 and >26 cm2, tested using calibration (the top subcell) is current matched to the other subcells in standards traceable to balloon flight reference cells. part by making it very thin, since its bandgap difference 1.91-eV GaInP Cell 1 EQE 100 1.81-eV GaInP Cell 2 EQE 70 1.57-eV AlGaInAs Cell 3 EQE 1.39-eV GaInAs Cell 4 EQE subcell. However, even with this relatively low quantum 1.05-eV GaInNAs Cell 5 EQE 0.67-eV Ge Cell 6 EQE 6J, cumulative EQE 60 efficiency, the ~1.1-eV GaInNAs subcell is capable of 2 80 AM0 generating greater than 8 mA/cm , allowing it to be current 50 matched to the other subcells in the series-interconnected, eV))

2 . 6-junction cell stack. The GaInNAs subcell can be 60 40 influenced strongly by the thermal profile (annealing) that it sees after it is deposited, depending on its growth 30 40 conditions. The GaInNAs subcell in a 6-junction cell will

20 necessarily experience a substantial thermal treatment Energy (mA/(cm Quantum Efficiency (%) due to the growth of the upper 4 subcells on top. Fig. 8

20 Unit Photon Per Curr. Dens. 10 shows the quantum efficiency of AR-coated GaInNAs subcells after anneals with a high and low thermal budget. 0 0 100 0.5 1.0 1.5 2.0 2.5 3.0 3.5 GaInNAs subcell 5: Effect of upper subcell 90 Photon Energy (eV) growth for 5 & 6J cells Fig. 6. Measured quantum efficiency of the six individual 80 Low thermal budget 70 subcells that compose a 6-junction GaInP/ GaInP/ 2 Jsc = 8.5 mA/cm AlGaInAs/ GaInAs/ GaInNAs/ Ge solar cell. The sum of 60 V cy (AR-coated) (%) the quantum efficiencies is also plotted, along with the 50

AM0 current density per unit photon energy. 40 = 1.05 e g

E 30 from subcell 2 is not sufficient to allow subcell 2 to current 20 match the other subcells otherwise. The GaInP subcell 2 High thermal budget 10 2 can be seen to have significant response at high photon Jsc = 5.1 mA/cm Ext. Quantum Efficien Ext. Quantum energies (>2 eV) as a result. 0 The theoretical efficiency of 3-junction space solar 800 900 1000 1100 1200 1300 cells, as well as 4 to 6-junction cells, can be improved by Wavelength (nm) the use of a higher bandgap AlGaInP top subcell with a Fig. 8. Effect of post-growth annealing on measured disordered group-III sublattice, rather than GaInP [9-11]. quantum efficiency and integrated AM0 current density of Figure 7 plots the measured external quantum efficiency AR-coated GaInNAs solar cells. and photoluminescence data from recent investigations of AlGaInP subcells ranging up to 2.1 eV in bandgap. Fully integrated 6-junction solar cells with active GaInNAs subcells have been grown and tested [4,8]. Fig. SR - GaInP - 1.896 eV 80 9 plots measured light I-V curves of three 6-junction SR - AlGaInP - 1.945 eV AlGaInP/ GaInP/ AlGaInAs/ GaInAs/ GaInNAs/ Ge solar

R d-GaInP 70 1.95 eV AlGaInP cells with three different anneal conditions for the GaInNAs 2.0 eV AlGaInP subcell, and with measured Voc reaching over 5.3 volts. 60 2.1 eV AlGaInP2 6-junction AlGaInP/ GaInP/ AlGaInAs/ GaInAs/ GaInNAs/ Ge solar cells 8 50 7 AlGaInP 1

8% Al )

d-GaInP 2 40 1.944 eV peapeak,k (% ) 1.896 eV peak, 6262 meV meV fwhm fwhm 6 66 meV fwhm Anneal condition 1 30 1 AlGaInP16% 2 Al 5 2.0062.006 eV eV pea peak,k Anneal condition 2 66 66meV meV fwhm fwhm 20 4 (arb. units) 28% Al Anneal condition 3 AlGaInP 3 V = 5.334 V 2.112 eV peak, oc 10 2.112 eV peak 3 2

A No - Effiiciency Quantum External 7373 meV meV fwhm fwhm Jsc = 7.30 mA/cm (active area) Relative RTPL Intensity FF = 82.0% 0 0 2 prelim. active area eff. = 23.6%

(mA/cm Density Current 300 400 500 600 700 prelim. total area eff. = 20.9% 1 AM0, 0.1353 W/cm2, 28oC, 0.26 cm2 Wavelength (nm)

0 Fig. 7. External quantum efficiency (no AR coating) and 0123456 photoluminescence measurements of AlGaInP cells. Voltage (V) Fig. 9. Illuminated I-V curves of fully-active, monolithic 6- As a measure of solar cell quality, it is desirable to junction cells with Voc over 5.3 volts. achieve a small bandgap-voltage offset, (Eg/q) - Voc, so that Voc is as close to the bandgap as possible [4,8]. The Achieving the correct spectral balance for 6-junction cells offset between bandgap and open-circuit voltage was under the simulated solar spectrum is difficult, and is an measured to be a low 453 mV for AlGaInP cells with 1.94 active area of development for space solar simulators. eV bandgap, indicating long minority-carrier lifetime in The Voc measurements are quite accurate even if the cells these Al-containing subcells. are unbalanced in current, but the Jsc is strongly From the quantum efficiency measurements in Fig. 6, dependent on imbalance in the simulated spectrum. the GaInNAs subcell 5 is clearly the lowest performing Preliminary short-circuit current density is 7.3 mA/cm2 and active-area efficiency is 23.6%, for the heavily metallized grid pattern used on these early prototype 6-junction cells. GaInNAs subcells and fully-integrated 6-junction cells with an active ~1.1-eV GaInNAs subcell have now been grown on 100-mm-diameter Ge wafers at Spectrolab, with good uniformity in a production-scale MOVPE reactor.

Flexible, Lightweight Space Solar Cells and Panels

Recently, there has been renewed interest in fabrication of extremely thin III-V multijunction solar cells, for the large savings in weight that can be accomplished, and also for the potential for such thin III-V cells to be flexible [12-14]. They may be integrated into novel, flexible panel designs with not only a very low weight per unit area Fig. 11. Photograph of a flexible, 3-cell coupon composed due to the cells, but also very low weight for the rest of the of thin , full-size (>26 cm2) 3-junction cells. panel support structures. For instance, thin multijunction 0.5 cells may be laminated between flexible sheets with metal 0.45 foil circuitry, resulting in a solar photovoltaic blanket with 0.4 very high specific power (W/kg), and which may be rolled 0.35 Preliminary XT-10 LIV Test or folded into a small package with very high power per 2 2 3 0.3 Cell Area: 79.86 cm (3 x 26.6 cm ) unit stowage volume (W/m ). This concept is depicted in Voltage: 7.9 V 0.25 Fig. 10, together with a photograph of one of Spectrolab's Current: 0.46post bend A XT-10 LIV 0.2 thin, full-size, flexible 3-junction space cells. With this Current (A) Fill Factor: 0.83 flexible PV blanket configuration, the lightweight, flexible, 0.15 Overall Coupon Weight: 6.1 g small stowage volume features of CuInSe2 and amorphous 0.1 Max Power: 3.03 W Power Density: ~500 W/kg Si thin-film space solar panels can be achieved, but with 0.05 AM0 efficiencies approaching 30% due to the thin III-V 0 multijunction cells used. 012345678 Voltage (V) Flexible Fig. 12. Measured current-voltage characteristic of the photovoltaic (PV) blanket flexible 3-cell coupon with power densiy of ~500 W/kg.

so much that the only layers remaining are the III-V epitaxially grown layers [14]. The Ge or GaAs growth substrate is removed either by lapping or etching it off, or by an epitaxial liftoff process. An attractive part of this latter process is that the growth substrate can be reused, High-efficiency multijunction solar cells cutting 3-junction cell manufacturing costs markedly. A in flexible, polymer photograph of a ~10-µm-thick cell with only the III-V packaging epitaxial layers still present, after lapping and etching away Spectrolab's thin multijunction cell the Ge substrate, is shown in Fig. 13. With no Ge Fig. 10. Schematic of thin, flexible III-V multijunction solar substrate, these particular multijunction cells have no Ge cells, encapsulated in thin polymer layers to form a high subcell and are only 2-junction cells. specific power photovoltaic blanket, deployed from a rolled up configuration with very small stowage volume.

Preliminary coupons of flexible photovoltaic blankets have been built from thin, large-area (>26 cm2) 3-junction GaInP/GaInAs/Ge cells, and tested [13]. Figure 11 shows a photograph of a flexible 3-cell coupon conforming to the shape of a 190-mm-diameter metal cylinder, in much the same way as a large flexible panel might be rolled around the spindle of a stowage and deployment mechanism on a satellite. Light I-V measurements of this 3-cell coupon, plotted in Fig. 12, show a specific power of ~500 W/kg at the PV blanket (panel) level. The 3-junction cells shown in Figs. 10-11 are still thick enough to be free standing. Further gains in reduced weight and flexibility can be achieved by thinning the cell Fig. 13. Sixteen 1.5 cm x 1.5 cm ultra-thin solar cells on 50-µm-thick kapton sheet. Radiation and Environmental Reliability in the chart, voltage degrades more rapidly than current for this type of cell. The resulting relative power at the Multijunction III-V solar cells and panels have been maximum-power-point, NPmp, is 0.89 for a fluence of 5 x subjected to some of the most rigorous environmental 1014 cm-3, and 0.86 at 1 x 1015 cm-3. tests that any type of photovoltaic device has been asked The experience of the space solar cell industry with to endure. As a result of testing under punishing radiation, thermal cycling of III-V multijunction panels has led to new thermal, and mechanical conditions, our understanding of heights in durability for these panels. Figure 16 plots the degradation mechanisms is deeper than ever, and high- progression of the maximum number of thermal cycles efficiency space panels based on III-V multijunction cells passed by multijunction test coupons by year. are resulting in more stable and reliable operation than Temperature range and frequency of the cycles in this test previous cell and panel technologies. were adjusted to simulate LEO and GEO orbits. Under the Figure 14 plots the cumulative power generated on LEO simulation, the number of thermal cycles successfully orbit by ITJ solar cells with AM0 efficiency > 26.8%. The reached the 100,000 mark. ITJ panels on spacecraft have been operating nominally 120000 LEO thermal cycles 24000 over a nearly 4 year period. As more experience is gained GEO thermal cycles with ITJ cells and panels in space, confidence in the ITJ 100000 20000 cell continues to grow. LEO GEO 80000 16000

60000 12000

40000 8000

Cycles Survived -- -- Survived Cycles 20000 4000 -- Survived Cycles Maximum Number of Thermal Thermal of Number Maximum Thermal of Number Maximum

0 0

1996 1998 2000 2002 2004 2006

Year Fig. 16. Maximum number of thermal cycles passed by test coupons using III-V multijunction cells.

The ultimate goal of reliability testing is accurate prediction of the actual power produced on orbit. Over 3 Fig. 14. Cumulative power of ITJ (Improved Triple- years of data on the power delivered by solar panels Junction) solar cells on panel and in orbit. populated with Improved Triple-Junction (ITJ) cells from Spectrolab are now available [15], and are shown in Fig. Similarly, our understanding of the relationships 17. The telemetered power, i.e., the power output between multijunction device structure and radiation measured during panel operation in orbit, is highly stable, resistance has become stronger than ever, and end-of-life and is consistently slightly above the predicted power for (EOL) efficiency continues to climb along with beginning- the solar panels, by about 2%. This gives outstanding of-life (BOL) power. Radiation characteristics of the UTJ confidence for the actual performance of triple-junction (Ultra Triple-Junction) cell are shown in Fig. 15. As seen cells under the rigorous conditions of space. 6.0% 1.00 NIsc NImp NVoc 4.0% NVmp NPmp 0.95 2.0%

0.0%

0.90

-2.0% Spacecraft A Deviation Spacecraft B -4.0% 0.85 Spacecraft C

Spacecraft D -6.0% Fitted Normalized Degradation Normalized Fitted 0.80 -8.0%

0.0 1.0 2.0 3.0 4.0 0.75 Years in orbit 1.E+13 1.E+14 1.E+15 1.E+16 Fig. 17. Relative difference between the actual measured 1 MeV Electron Fluence [e/cm2] (telemetered) and predicted power of solar panels Fig. 15. Radiation degradation vs. 1 MeV electron fluence populated with ITJ cells, during normal operation in orbit for UTJ cells with >28.3% beginning-of-life efficiency. [15]. Deviation = (Ptelemetered / Ppredicted) - 1 . SUMMARY D. Krut, N. H. Karam, "High-Efficiency Space and Terrestrial Multijunction Solar Cells Through Bandgap Control in Cell III-V multijunction solar cells and panels for satellite Structures," Proc. 29th IEEE Photovoltaic Specialists Conf., New Orleans, Louisiana, May 20-24, 2002. power systems are entering an era of higher performance [3] P. R. Sharps, M. A. Stan, D. J. Aiken, F. D. Newman, J. S. never before seen with this cell technology. Quantitatively, Hills, and N. S. Fatemi, "High Efficiency Multi-Junction Solar Cells solar cell efficiencies of over 30% under the AM0 spectrum - Past, Present, and Future," Proc. 19th European Photovoltaic can be achieved reproducibly, and commercially-available Solar Energy Conf., Paris, France, 7-11 June 2004, p. 3569. 30% cells are just around the corner. Research is [4] R. R. King, D. C. Law, C. M. Fetzer, R. A. Sherif, K. M. underway on new multijunction cell structures such as 5- Edmondson, S. Kurtz, G. S. Kinsey, H. L. Cotal, D. D. Krut, J. H. and 6-junction cells, which can boost AM0 efficiency to the Ermer, and N. H. Karam, "Pathways to 40%-Efficient Concentrator 33-35% range. ~1.1-eV GaInNAs subcells and fully- Photovoltaics," 20th European Photovoltaic Solar Energy Conference and Exhibition, Barcelona, Spain, 6-10 June 2005. integrated 6-junction cells have now been grown [5] R. R. King, M. Haddad, T. Isshiki, P. Colter, J. Ermer, H. reproducibly and uniformly on 100-mm Ge substrates. Yoon, D. E. Joslin, and N. H. Karam, "Metamorphic Qualitatively, thin and flexible multijunction III-V cells with GaInP/GaInAs/Ge Solar Cells," Proc. 28th IEEE Photovoltaic radically different mechanical properties are being Specialists Conf., Anchorage, Alaska, Sep. 15-22, 2000. investigated in research organizations around the globe. [6] F. Dimroth, U. Schubert, and A. W. Bett, "25.5% Efficient These single-crystalline thin-film multijunction cells have Ga0.35In0.65P/Ga0.83In0.17As Tandem Solar Cells Grown on GaAs as high efficiency as their ~150-µm-thick counterparts, Substrates," IEEE Electron Device Lett., 21, p. 209 (2000). [7] M. W. Wanlass, S. P. Ahrenkiel, R. K. Ahrenkiel, D. S. approaching 30%, but their lightweight and flexibility allows Albin, J. J. Carapella, A. Duda, J. F. Geisz, Sarah Kurtz, T. them to be used in rollable or foldable arrays as is usually Moriarty, R. J. Wehrer, and B. Wernsman, "Lattice-Mismatched contemplated only for polycrystalline or amorphous thin- Approaches for High-Performance, III-V Photovoltaic Energy film solar cells with significantly lower efficiency. III-V Converters," Proc. 31st IEEE Photovoltaic Specialists Conf., Lake space cells are going through very thorough and rigorous Buena Vista, Florida, Jan. 3-7, 2005, p. 530. reliability testing, and are being proven out even under [8] R. R. King, C. M. Fetzer, K. M. Edmondson, D. C. Law, harsh conditions, such as 100,000 thermal cycles on a P. C. Colter, H. L. Cotal, R. A. Sherif, H. Yoon, T. Isshiki, D. D. panel coupon. The bottom line is the performance of a Krut, G. S. Kinsey, J. H. Ermer, Sarah Kurtz, T. Moriarty, J. Kiehl, K. Emery, W. K. Metzger, R. K. Ahrenkiel, and N. H. Karam, solar panel on orbit. Telemetry of the power delivered to "Metamorphic III-V Materials, Sublattice Disorder and the spacecraft by solar arrays on orbit has shown that Multijunction Solar Cell Approaches with Over 37% Efficiency," actual power delivered over a period > 3 years is not only Proc. 19th European Photovoltaic Solar Energy Conf., Paris, stable, but exceeds the predicted power by ~2%. As a France, 7-11 June 2004, p. 3587. result of all that the space PV community has learned over [9] R. R. King, N. H. Karam, J. H. Ermer, M. Haddad, P. the past decade, III-V multijunction space solar cells are Colter, T. Isshiki, H. Yoon, H. L. Cotal, D. E. Joslin, D. D. Krut, R. reaching new heights in efficiency, reliability, and Sudharsanan, K. Edmondson, B. T. Cavicchi, and D. R. Lillington, "Next-Generation, High-Efficiency III-V Multijunction Solar Cells," versatility. Proc. 28th IEEE Photovoltaic Specialists Conf., Anchorage, Alaska, Sep. 15-22, 2000, p. 998. ACKNOWLEDGMENTS [10] Takaaki Agui, Tatsuya Takamoto, and Minoru Kaneiwa, "Investigation on AlInGaP Solar Cells for Current Matched The authors would like to thank Henry Yoo, John Multijunction Cells," Proc. 3rd World Conf. on Photovoltaic Energy Merrill, and Donna Senft of the Air Force Research Conversion, Osaka, Japan, May 11-18, 2003, p. 670. Laboratory (AFRL/VS), Jennifer Granata, Robert Francis, [11] Tatsuya Takamoto, Takaaki Agui, Kunio Kamimura, and Dean Marvin of the Aerospace Corporation, Randall Minoru Kaneiwa, Mitsuru Imaizumi, Sumio Matsuda, and Masafumi Yamaguchi, "Multijunction Solar Cell Technologies - Speth of the National Aeronautics and Space High Efficiency, Radiation Resistance, and Concentrator Adminstration (NASA), and Robert Cravens, Rina Applications," Proc. 3rd World Conf. on Photovoltaic Energy Bardfield, Dave Peterson, Jerry Kukulka, Mark Takahashi, Conversion, Osaka, Japan, May 11-18, 2003, p. 581. Jason Yen, Takahiro Isshiki, Moran Haddad, and the [12] T. Takamoto et al., "Flexible Thin-Film III-V Multijunction entire multijunction solar cell team at Spectrolab. This Solar Cells," Proc. 19th European Photovoltaic Solar Energy work was supported in part by AFRL/VS under DUS&T Conf., Paris, France, 7-11 June 2004, p. 3689. contract # F29601-98-2-0207, by NASA under the [13] K. M. Edmondson, D. C. Law, G. Glenn, A. Paredes, R. Exploration System Research and Technology (ESR&T) R. King, and N. H. Karam, "Flexible III-V Multijunction Solar Blanket," 4th World Conf. on Photovoltaic Energy Conversion, program contract # NNC05CB06C, and by Spectrolab. Waikoloa, Hawaii, May 7-12, 2006. This work is dedicated in memory of David E. Joslin. [14] D. C. Law, K. M. Edmondson, N. A. Siddiqi, A. Paredes, G. S. Glenn, E. L. Labios, R. R. King, M. H. Haddad, T. D. Isshiki, REFERENCES and N. H. Karam, "Lightweight, Flexible, High-Efficiency III-V Multijunction Cells," 4th World Conf. on Photovoltaic Energy [1] M. Meusel, C. Baur, W. Guter, M. Hermle, F. Dimroth, A. Conversion, Waikoloa, Hawaii, May 7-12, 2006. W. Bett, T. Bergunde, R. Dietrich, R. Kern, W. Köstler, M. Nell, W. [15] J.S. Fodor, S.W. Gelb, Z. Maassarani, J.S. Powe, J.A. Zimmermann, G. LaRoche, G. Strobl, S. Taylor, C. Signorini, G. Schwartz, "Analysis of Triple Junction Solar Arrays After Three Hey, "Development Status of European Multi-Junction Space Years in Orbit," Proc. of 4th World Conf. on Photovoltaic Energy Solar Cells with High Radiation Hardness," Proc. 20th European Conversion, Waikoloa, Hawaii, May 7-12, 2006. Photovoltaic Solar Energy Conf., Barcelona, Spain, 6-10 June 2005. [2] R. R. King, C. M. Fetzer, P. C. Colter, K. M. Edmondson, J. H. Ermer, H. L. Cotal, H. Yoon, A. P. Stavrides, G. Kinsey, D.