Operation of Solar Cells in aSpace EnvironmentSheila Baileya and Ryne RaffaellebaPhotovoltaic and Space Environments Branch, NASA Glenn Research Center, USAbNational Center for Photovoltaics, National Renewable Energy Laboratory (NREL), 1617 Cole Blvd.,Golden, CO 80401-3305
Contents
1. Introduction 8632. Space Missions and their Environments 865
2.1 The Air Mass Zero Spectrum 8652.2 The Trapped Radiation Environment 8672.3 Solar Flares 8682.4 The Neutral Environment 8692.5 The Paniculate Environment 8702.6 Thermal Environment 870
3. Space Solar Cells 8723.1 Radiation Damage in Space Solar Cells 873
4. Small Power Systems 8755. Large Power Systems 877References 879
1. INTRODUCTION
The beginning of the Space Age brought about a perfect application
for the silicon solar cell developed at Bell laboratory in 1953. Sputnik was
battery powered and remained active only a little over a week. The US
launched the first successful solar powered satellite, Vanguard 1, seen in
Figure 1, on March 17, 1958 [1]. The solar powered transmitter lasted six
years before it is believed that the transmitter circuitry failed.
Vanguard I had eight small panels with six p on n silicon solar cells, each
2 cm3 0.4 cm, connected in series. Each panel output was approximately
50 mW with a cell efficiency of B8%. This can be contrasted with the
International Space Station (ISS), see Figure 2, which has the largest photo-
voltaic power system ever present in space, with 262,400 n on p silicon
solar cells, each 8 cm3 8 cm, with an average efficiency of 14.2% on 8
Figure 1 3/12/1957, Senator Lyndon B. Johnson, chairman of the US SenatePreparedness subcommittee, holds the tiny 6.5-inch American test satellite.
Figure 2 The International Space Station (compare shuttle for size).
864 Sheila Bailey and Ryne Raffaelle
US solar arrays (each B34 m3 12 m) [2]. This can generate about
110 kW of average power, which after battery charging, life support, and
distribution, can supply 46 kW of continuous power for research experi-
ments on ISS.
2. SPACE MISSIONS AND THEIR ENVIRONMENTS
Space missions are defined by their trajectories. For Earth orbiting
missions these are roughly classified as low Earth orbit (LEO),
300�900 km, mid-Earth orbit (MEO), and geosynchronous (GEO),
35,780 km. The orbit’s size, defined by the semimajor axis; shape, defined
by the eccentricity; orientation, defined by the orbital plane in space (incli-
nation and right ascension of the ascending node); and the orbit within the
plane, defined by the argument of perigee, determine the space environ-
ment the spacecraft will encounter. NASA missions also involve interplan-
etary flight both toward and away from the Sun, planetary fly-bys, and
orbiting other planets, each with their own unique set of environments. In
addition both the moon and Mars may be sites for future human visits and
have their own individual conditions for surface power.
At the heart of our solar system is the Sun, which is both the source
of the solar irradiance which a solar cell converts to electricity and the
solar wind which is primarily a stream of protons and electrons moving
with a mean velocity by Earth of B400�500 km/s with a mean density
of approximately 5/cm3. In addition, the Sun is a dynamic body exhibit-
ing facula, plages, spicules, prominences, sunspots, and flares over time.
The only other source of radiation is galactic cosmic rays which emanate
from beyond our solar system. These consist of about 85% protons, about
14% alpha particles, and about 1% heavier nuclei [3]. The differential
energy spectra of the cosmic rays near the Earth tend to peak around
1 GeV/nucleon and the total flux of particles seen outside the magneto-
sphere at the distance of the Earth from the Sun (i.e., 1 AU) is approxi-
mately 4 per square centimeter per second.
2.1 The Air Mass Zero SpectrumThe spectral illumination that is available in space is not filtered by our
atmosphere and thus is quite different from what is incident on Earth’s sur-
face (see Figure 3). Space solar cells are designed and tested under an Air
865Operation of Solar Cells in a Space Environment
Mass Zero (AMO) spectrum. This is in contrast to an Air Mass 1.5 as
reduced by 1.5 times the spectral absorbance of the Earth’s atmosphere,
which is the standard condition for testing terrestrial solar cells. Thus, cells
intended for use in space will be optimized for a somewhat different spec-
trum. The change in spectral distribution will typically result in a decrease
in overall cell efficiency, even though the intensity of light is somewhat
higher (i.e., 1367 W/m2 in space as compared to 1000 W/m2 on Earth). A
12% efficient silicon solar cell as measured under AM1.5 condition on Earth
would translate into an approximately 10% cell as measured under AM0.
As seen below the Sun resembles a blackbody with a surface tempera-
ture of 5800 K with a peak of spectrally emitted energy at 480 nm.
Approximately 77% of the emitted energy lies in the band from 300 to
1200 nm. The total energy received from the Sun per unit area perpen-
dicular to the Sun’s rays at the mean Earth�Sun distance (1 AU) is called
00.000
0.050
0.100
Spe
ctra
l irr
adia
nce
(mW
/cm
2·n
m)
0.150
0.200
0.250
1000 2000
Wavelength (nm)
3000 4000
AM1.5 (ASTM)
AMO (WMO)
Figure 3 The Air Mass Zero (AMO) spectrum (WMO) and the Air Mass 1.5 (ASTM)spectrum.
866 Sheila Bailey and Ryne Raffaelle
the solar constant. The current accepted value of the solar constant is
1367 W/m2. The solar intensity of course varies in time. However since
space solar cells are calibrated in near-space, the variation in the value of
the solar constant primarily affects the predicted solar cell operational
temperature in orbit.
2.2 The Trapped Radiation EnvironmentThe solar wind, solar flares and galactic cosmic rays all consist of charged
particles (electrons, protons, and ions). These interact with a planetary
magnetic field. Some planets have a very weak magnetic field or no mag-
netic field. Jupiter has a very large magnetic field. Jupiter, Saturn, and
Uranus have trapped radiation belts. The Earth’s magnetic field is 0.3 gauss
at the surface on the equator and does change over time even reversing
polarity every 10,000 years. The Earth’s magnetic poles do not coincide
with the poles determined by the axis of rotation, with approximately
an 11 degree difference. The total magnetic field of the magnetosphere is
determined by the internal magnetic field of the planet and the external
field generated by the solar wind. These interact with each other and pro-
vide the complex asymmetric pattern of the geomagnetic cavity. Charged
particles gyrate around and bounce along magnetic field lines, and are
reflected back and forth between pairs of conjugate mirror points in
opposite hemispheres. At the same time electrons drift eastward around
the Earth while protons and heavy ions drift westward. These regions of
trapped charged particles are called the Van Allen belts. An illustration of
the regions of trapped particles can be seen in Figure 4, where L is a
dimensionless ratio of the Earth’s radius equal to the radial distance
divided by the cos 2Λ, where Λ is the invariant latitude.
An example of the number of trapped particles as a function of energy
for both Earth and Jupiter can be seen in Figure 5 [4]. The Jupiter data
were provided by Pioneers 10 and 11 and Voyagers 1 and 2 during their
encounters with the planet.
The models that are used for the trapped electron and proton environ-
ment at Earth were developed by the US National Space Science Data
Center at NASA’s Goddard Space Flight Center from available radiation
measurements from space. The most recent models in use, AP8 for pro-
tons [5] and AE8 for electrons [6], permit long-term average predictions
of trapped particle fluxes encountered in any orbit and currently consti-
tute the best estimates for the trapped radiation belts, although they have
been noted to overestimate the radiation in certain low Earth orbits.
867Operation of Solar Cells in a Space Environment
2.3 Solar FlaresSolar activity, as measured by the number of sunspots, follows an 11-year
cycle between maxima. The cycle has an active 7-year period during
which solar flare events are probable and a quiescent 4-year period during
which solar flare events are rare. The last peak in activity occurred in
1
Inner zoneelectrons
Outer zoneelectrons
Trappedprotons
Solar flareprotons
Region
Syn
chro
no
us
3 5 7
L (Earth Radii)
9 11 13
Figure 4 Regions of trapped particles as a function of distance in Earth radii fromthe Earth’s centre.
Jupiter: hard electron spectrumEarth: hard electron spectrum
0.11E+8
1E+7
1E+6
1E+5
1E+4
Num
ber
of p
artic
les
(nor
mal
ized
)
1E+3
1E+2
1E+1
1E+0
1 10
Energy (MeV)
100 1000
Electrons,Earth Protons,
Jupiter
Protons,Earth
Electrons,Jupiter
Figure 5 Normalized energy spectrum of trapped electron and proton radiationenvironment at Jupiter, compared to Earth.
868 Sheila Bailey and Ryne Raffaelle
2000. The most recent model, JPL 91, allows the spacecraft designer to
predict the proton integral fluence as a function of confidence level and
exposure time [7]. The exposure time must be correlated to the solar
maxima. The calculated integral fluence is at 1 AU from the Sun. For
missions at other radial distances from the Sun the fluence should be
modified by 1/R2 to produce the most probable estimate. These solar
flare proton events are associated with coronal mass ejections on the Sun.
They occur at highly localized places on the Sun and rotate with the Sun.
Because they follow the field lines of the interplanetary magnetic field,
only when the ejection occurs along the line from the Sun that intersects
the Earth will the protons propagate immediately to the Earth, arriving
approximately an hour after the flare. These protons are highly anisotropic
and therefore variations in proton flux can be as large as 100 from the
same flare at different points of the orbit. Protons can also reach Earth
when ejection occurs away from a field line but then the protons must
diffuse through the solar corona before they propagate to Earth. This
takes longer, up to 10 hours, and the flux tends to become isotropic. The
trapped proton population is also significantly effected by solar activity,
particularly in lower Earth orbits. The increased energy output of the Sun
expands the atmosphere and increases proton densities. There is a region
of trapped particles close to the surface of the Earth called the South
Atlantic Anomaly. For low altitude, low inclination orbits the South
Atlantic Anomaly may be the most significant source of radiation.
The electron environment is also influenced by the solar cycle.
Magnetic solar storms can raise the electron flux in the outer orbits by
an order of magnitude over a short time period. This short-term varia-
tion in electron flux is usually more significant for spacecraft charging
effects in GEO. AP8 and AE8 mentioned above have data for both solar
maximum and solar minimum. A more recent model based on the
CRRES spacecraft launched in July 1990 found differences between the
AP8 and AE8 predictions to be as high as a factor of 3 orders of magni-
tude particularly at low altitudes due to the highly variable proton belts
[8]. At high altitudes AE8 predictions are typically higher than CRRES
predictions.
2.4 The Neutral EnvironmentThe density, composition, pressure, and temperature change dramatically
as a function of altitude. Orbits lower than 200 km are generally not
stable due to atmospheric drag. At 300 km only proportionately 23% of
869Operation of Solar Cells in a Space Environment
the sea-level molecular nitrogen remains and only 10% of molecular oxy-
gen. 80% of the atmosphere at 300 km is highly reactive atomic oxygen.
Atomic oxygen erodes polymers and composites that might be used in
array substrates and also silver interconnects between solar cells [9]. The
atmospheric density is very small above 800 km.
2.5 The Particulate EnvironmentThe particulate environment is composed of both naturally occurring
meteoroids and man-made space debris. The particles of most concern to
space arrays are between 10�3 and 10�6 g, since those below 10�6 g do
not have sufficient energy to cause significant damage and those above
10�3 are less frequently encountered [10]. Meteoroids, whose origin is
either asteroids or comets, have an average velocity of 20 km/s and their
density varies with the Earth’s position around the Sun. Debris has of
course become more problematical as the number of launched spacecraft
and their relative time in orbit has increased. The relative damage of
orbital debris, except for very large objects, is less due to the reduced dif-
ference in orbital velocities for the debris that was created in that orbit.
The flux of meteoroids and orbital debris has been observed for a variety
of orbits [11]. Damage to the solar arrays of Mir and the Hubble Space
Telescope were noted in primarily the erosion and cracking of the cover
glass on the array and the erosion of the substrate rear surface thermal
control coating.
2.6 Thermal EnvironmentThe temperature of a solar cell in space is largely determined by the
intensity and duration of its illumination [12]. In the case of the US array
on the ISS, the operating temperature of the silicon solar cells is as high
as 55�C while under illumination, and as low as 280�C when in eclipse.
Similarly, as spacecraft venture further away from the Sun their average
temperatures will decrease. Likewise, as they move closer to the Sun their
average temperatures will rise. The average illuminated temperature at the
orbit of Jupiter is �125�C, whereas at the average orbit of Mercury the
temperature is 140�C.The orbital characteristics of a space mission are also a major source of
thermal variation for the associated photovoltaic arrays. The relative
amount of illumination versus eclipse time and the rate of change in the
temperature vary dramatically will the orbital path. The orbital path will
also affect the fraction of incident solar radiation returned from a planet
870 Sheila Bailey and Ryne Raffaelle
or albedo. The average albedo from the Earth is 0.34, but can range any-
where from 0.03 (over forests) to 0.8 (over clouds) [7].
The available power generated by a solar cell is directly related to its
operating temperature. An increase in temperature will result in a reduc-
tion in output power. Although there will be a slight increase in the
short-circuit current with increasing temperature, it will be overshadowed
by the decrease in the open-circuit voltage. A GaAs solar cell will experi-
ence a decrease of about 0.05 mW/cm2 per �C.The degradation of solar cell performance as a function of temperature
is expressed in terms of temperature coefficients. There are several different
temperature coefficients used to describe the thermal behavior of solar
cells. They are based in terms of the change in a characteristic cell measure-
ment parameter (i.e., Isc, Voc, Imp, Vmp, or η) as a function of the change in
temperature. The difference in the measured value at the desired tempera-
ture and a reference temperature is used to determine the coefficient. The
International Space Organization (ISO) standard reference measurement is
taken at 25�C. For most space solar cells, the change in output is fairly lin-
ear from 2100�C to 100�C.Temperature coefficients are often expressed as a normalized number.
For example, if the case of the efficiency temperature coefficient the nor-
malized value would be expressed as
β5I
ηdηdT
ð1Þ
or the fractional change in efficiency with temperature. Representative
temperature coefficients for the various types of cells used in space are
given in Table 1. The temperature coefficient is inversely related to band
gap and negative for the majority of space solar types.
Table 1 Measured temperature coefficients for various types of solar cells used inspace [26]Cell Type Temp (�C) η (28�C) 1/ηdη/dT (3 1023�C21)
Si 28�60 0.148 24.60
Ge 20�80 0.090 210.1
GaAs/Ge 20�120 0.174 21.60
2-j GaAs/Ge 35�100 0.194 22.85
InP 0�150 0.195 21.59
a-Si 0�40 0.066 21.11 (non-linear)
CuInSe2 240�80 0.087 26.52
871Operation of Solar Cells in a Space Environment
3. SPACE SOLAR CELLS
The first 30 years of space solar cell development focused on the of
silicon solar cells, although it was known even in the early days that better
materials existed [13]. The concept of a tandem cell was also proposed in
the early days to enhance the overall efficiency. An optimized three-cell
stack was soon to follow with a theoretical optimum efficiency of 37%
[14]. However, it was 40 years later before a multijunction solar cell flew
in space. Today silicon cells still fly in space but the cell of choice is a
multijunction solar cell. Table 2 shows some of the best efficiencies for
small area cells and the comparison to an AMO efficiency.
A variety of cell types are listed in Table 2 because, while commercial
satellites use silicon or dual or triple junction GaInP/GaAs/Ge. there is a
marked interest in military applications of thin film cells. NASA also has
planned missions in which a large specific power (kW/kg) and lower cost
would be beneficial. The advantages of thin film solar cells are their large
specific power when deposited on a flexible, lightweight substrate with a
Table 2 Measured Global AM1.5 and measured or *estimated AMO efficiencies forsmall area cells
Cells
Efficiency(%)GlobalAM1.5
Efficiency(%)AM0
RadioAM0/AM1.5
Area(cm2) Manufacturer
c-Si 22.3 21.1 0.95 21.45 Sunpower [15]
Poly-Si 18.6 17.1* 0.92 1.0 Georgia Tech/
HEM[16]
c-Si film 16.6 14.8* 0.89 0.98 Astropower [17]
GaAs 25.1 22.1* 0.88 3.91 Kopin [17]
InP 21.9 19.3* 0.88 4.02 Spire [17]
GaInP(1.88ev) 14.7 13.5 0.92 1.0 ISE [18]
GaInP/GaAs/Ge 31.0 29.3 0.95 0.25 Spectrolab [18]
Cu(Ga,In)Se2 18.8 16.4* 0.87 1.04 NREL [15]
CdTe 16.4 14.7* 0.90 1.131 NREL [15]
a-Si/a-Si/a-SiGe 13.5 12.0 0.89 0.27 USSC [15]
Dye-sensitized 10.6 9.8* 0.92 0.25 EPFL [15]
*These are based on cells measured under standard conditions, courtesy or Keith Emery, NREL. Thecalculated efficiency uses the ASTM E490�2000 reference spectrum and assumes that the fill factordoes not change for the increased photocurrent. Quantum efficiencies corresponding to thetable entries were used in the calculations.
872 Sheila Bailey and Ryne Raffaelle
suitably lightweight support structure. Thin film solar cells are currently
lower in efficiency and require a larger area for the same power levels;
however, trade studies have identified several potential applications [19].
3.1 Radiation Damage in Space Solar CellsRadiation degradation in space is a complex issue. The degradation is
dependent on the type of particle, energy and fluence, shielding, and cell
design (layer thickness, number of junctions, etc.). In addition ground
based radiation measurements use monoenergetic, unidirectional beams of
particles (electrons or protons) and the simulation of the space solar envi-
ronment, especially for multijunction cells, is difficult. Historically the Jet
Propulsion Laboratory (JPL) has provided the format for determining
radiation damage in silicon and gallium arsenide space solar cells [20,21].
The elements needed to perform degradation calculations are degradation
data under normal 1 MeV electron irradiation, the effective relative dam-
age coefficients for omnidirectional space electrons and protons of various
energies with various cover-glass thickness, and the space radiation data
for the orbit of interest. As discussed in the section on the trapped radia-
tion environment, AP8 and AE8 are current NASA models for trapped
radiation. The models were based on observations from 43 satellites from
1958 to 1970 for AP8 and from 1958 to 1978 for AE8. They can return
an integral or differential omnidirectional flux for a set of energies. The
integral flux is the number of particles with energy greater than or equal
to the input energy. The models calculate a numerical derivative of the
integral flux to obtain the differential flux. For AP8 the energy range is
0.1 to 400 MeV with McIlwain L number ranging from 1.1 to 6.6. For
AE8 the energy range is 0.04 to 7 MeV with an L number from 1.1 to
11. The models calculate a numerical derivative of the integral flux to
obtain the differential flux. As mentioned in the section on solar flares the
models permit a choice of solar maximum or solar minimum. AP8 and
AE8 provide the largest coverage for Earth orbiting spacecraft and are
internationally available. Other models exist with narrower coverage: the
US Air Force model from the CRRES data, an ESA model based on the
SAMPEX spacecraft, and a Boeing model based on TIROS/NOAA
satellites.
In recent years the Naval Research Lab (NRL) has developed a model
of displacement damage dose based on the nonionizing energy loss
(NIEL) [22]. The NIEL gives the energy dependence of the relative
873Operation of Solar Cells in a Space Environment
damage coefficients, and because the NIEL is a calculated quantity, the
NRL method enables the analysis of a solar cell response to irradiation by
a spectrum of particle energies, as encountered in space, based on only
one or two ground measurements. The Solar Array Verification and
Analysis Tool (SAVANT) [23,24] computer program being developed at
NASA Glenn Research Center combines the NRL method with the
NASA space environment models to produce a user-friendly space solar
array analysis tool. Equator-S and COMETS satellite data have been ana-
lyzed using SAVANT. In the Equator-S mission, the model was successful
in predicting the onboard degradation of both GaAs/Ge and CuInSe2solar cells. This is the first time that the model has been applied to a thin
film technology. SAVANT and the onboard measurements agreed to
within a few percent over the entire mission. The COMETS mission
used GaAs/Ge solar cells as its main power source, and SAVANT accu-
rately modeled the power output of the arrays for the bulk of the mission
lifetime [25]. SAVANT is not currently available to the public.
Normalized power as a function of altitude for solar cells in a 60�
orbit for 10 years with 75 μm cover glass can be seen in Figure 6. Note
that by using normalized power the higher radiation resistance of
CuInSe2 can be seen. Also, the larger particle flux can be noted for MEO
orbits.
00
5
10
15
20
Pow
er (
mW
/cm
2 )
25
30
35
5000 10,000 15,000 20,000
Altitude (km)
10 years, 60 degrees, 12 mil coverglass
25,000 30,000 35,000 40,000
Dual junction
GaAs
InPSilicon
CulnSe2
Figure 6 Normalised power versus altitude for a 60� orbit for 10 years with 75-μmcover glass. (Graph courtesy of Tom Morton, Ohio Aerospace Institute.)
874 Sheila Bailey and Ryne Raffaelle
4. SMALL POWER SYSTEMS
There has been considerable development over the last several years
in the development of small spacecraft. Many so-called microsats or satel-
lites whose total mass is less than 100 kg have been deployed. In fact,
satellites whose total mass is less than 10 kg (i.e., nanosats) and even satel-
lites weighing less that 1 kg (i.e., picosat) have already been tested in the
space environment (see Figure 7). Consequently, the demands on the
space power community to develop appropriately sized power systems for
these new classes of satellites have arisen.
It is true that a premium has always been placed on the efficiency of
space power systems and specifically the photovoltaic arrays. Specific
power (W/kg) or power per mass is one of the most important figures of
merit in judging a power system. The higher the specific power, the less
the spacecraft mass that has to be dedicated to the power system and the
more that can be used for the scientific mission. This is especially true in
the case of a small power system. There is an economy of scale savings
that can sometimes be recouped on a larger satellite. As power compo-
nents are reduced in size, so too is there capacity. In the case of a solar
cell this translates into their ability to gather light.
Figure 7 The SNAP-1 Surrey Nanosatellite Applications Platform was a 6-kg satellitewith imager and propulsion. (Picture courtesy of NASA and Surrey Satellite TechnologyLtd.)
875Operation of Solar Cells in a Space Environment
A number of approaches have been used to meet the demands of small
satellites. One such approach is the development of integrated power sup-
plies. These supplies combine both power generation and storage into
single devices. In fact, NASA, the Naval Research Labs, and others are
working to develop monolithically grown devices that combine mono-
lithically interconnect module (MIM) solar cells or micro-sized solar
arrays with lithium-ion energy storage. A first demonstration of this
Figure 8 Photograph of Starshine 3 satellite with a magnified region of the inte-grated power supply.
(2) (1959) 70�75.[2] L. Hague, J. Metcalf, G. Shannon, R. Hill, C. Lu, Performance of International
Space Station electric power system during station assembly, Proceedings of theThirty-first Intersociety Energy Conversion Engineering Conference, 1996, pp.154�159.
[3] E. Stassinopoulos, J. Raymond, The space radiation environment for electronics,Proc. IEEE 11 (1988) 1423�1442.
[4] S. Kayali, Space radiation effects on microelectronics, NASA Jet PropulsionLaboratories Course. See material on JPL web site at: ,http://nppp.jpl.nasa.gov/docs/Raders_Final2.pdf/..
[5] D.M. Sawyer, J.I. Vette, AP8 trapped proton environment for solar maximum andsolar minimum, Report NSSDA 76�06, National Space Science Data Center,Greenbelt, MD, 1976.
[6] J.I. Vette, AE8 trapped electron model, NSSDC/WDC-A-R&S 91�24, NationalSpace Science Data Center, Greenbelt, MD, 1991.
[7] J. Feynman, G. Spitale, J. Wang, S. Gabriel, Interplanetary Proton Fluence Model:JPL 1991, J. Geophys. Res. 98(A8) (1993) 13281.
[8] H.B. Garrett, D. Hastings, The Space Radiation Models, Paper No. 94�0590, 32ndAIAA Aerospace Sciences Meeting, Reno, Nevada, 1994.
[9] R.C. Tennyson, Atomic oxygen and its effect on materials, In: The Behaviour ofSystems in the Space Environment, Kluwer Academic, 1993, pp. 233�257.
[10] Solar Array Design Handbook, vol. 1, JPL, 1976.[11] TRW Space & Technology Group, TRW Space Data, N. Barter (Ed.), Fourth ed.,
1996.[12] A.L. Fahrcnbruch, R.H. Bube, Fundamentals of Solar Cells, Academic Press,
Boston, 1983, Chapter 2.[13] J.J. Loferski, Theoretical considerations governing the choice of the optimum semi-
conductor for the photovoltaic solar energy conversion, J. Appl. Phys. 27 (1956)777.
[14] E.D. Jackson, Areas for improvement of the semiconductor solar energy converter.Trans. Conference On the Use of Solar Energy Tucson, Arizona, 1955, Vol. 5, p. 122.
[15] K. Bucher, S. Kunzelmann, The Fraunhofer ISE PV Charts: Assessment of PVDevice Performance, Report EUR 18656 EN, Joint Research Center, 1998, pp.2329�2333.
[16] M. Green, K. Emery, K. Bucher, D. King, S. Igari, Solar cell efficiencytables (version 11), Prog. Photovolt: Res. Appl. 6 (1998) 35�42.
[17] M. Green, K. Emery, D. King, S. Igari, W. Warta, Solar cell efficiency tables (version18), Prog. Photovolt: Res. Appl. 9 (2001) 87�293.
[18] R. King, M. Haddad, T. Isshiki, P. Colter, J. Ermer, H. Yoon, et al., MetamorphicGaInP/GaInAs/Ge solar cells, Proceedings of the Twenth-eighth IEEE PhotovoltaicSpecialist Conference, Anchorage, 2000, pp. 982�985.
[19] D. Murphy, M. Eskenazi, S. White, B. Spence, Thin-film and crystalline solar cellarray system performance comparisons, Proceedings of the Twenty-ninth IEEEPhotovoltaic Specialists Conference, 2000, Anchorage, pp. 782�787.
[20] H. Tada, J. Carter, B. Anspaugh, R. Downing, Solar Cell Radiation Handbook,third ed., JPL Publication, 1982, pp. 82�69.
[21] B. Anspaugh, GaAs Solar Cell Radiation Handbook, JPL Publication 96�9, 1996.[22] G.P. Summers, E.A. Burke, M.A. Xapsos, Displacement damage analogs to ionizing
[23] S. Bailey, K. Long, H. Curtis, B. Gardner, V. Davis, S.Messenger, et al., Proceedingsof the Second World Conference on Photovoltaic Solar Energy Conversion, Vienna,1998, pp. 3650�3653.
[24] T. Morton, R. Chock, K. Long, S. Bailey, S. Messenger, R. Walters, et al., TechicalDigest 11th International Photovoltaic Science and Engineering Conference,Hokkaido, 1999, pp. 815�816.
[25] S. Messenger, R. Walters, G. Summers, T. Morton, G. La Roche, C. Signorini,et al., A displacement damage dose analysis of the COMETS and Equator-S spacesolar cell flight experiments, Proceedings of the Sixteenth European PhotovoltaicSolar Energy Conference, Glasgow, 2000, pp. 974�977.
[26] P. Jenkins, T. Kerslake, D. Scheiman, D. Wilt, R. Button, T. Miller, et al., Firstresults from the starshine 3 power technology experiment, Proceedings of theTwenty-ninth IEEE Photovoltaic Specialists Conference, New Orleans, 2002, pp.788�791.
