Skip to main content

Full text of "Low Intensity Low Temperature (LILT) measurements and coefficients on new photovoltaic structures"

See other formats



David A. Scheiman and Phillip P. Jenkins , .• ,../<■ 

NYMAinc. <J^i mJL . 

Brook Park, Ohio, 44142 . ^ 6 U 

r ' 

David J. Brinker Joseph Appelbaum 

NASA LeRC Tel-Aviv University 

Cleveland, Ohio 44135 Tel-Aviv, Israel, 69978 


Past NASA missions to Mars, Jupiter and the outer planets were powered by radioisotope thermal generators 
(RTGs). Although these devices proved to be reliable, their high cost and highly toxic radioactive heat source has 
made them far less desirable for future planetary missions. This has resulted in a renewed search for alternate 
energy sources, some of them being photovoitaics (PV) and thermophotovoltaics (TPV). Both of these alternate 
energy sources convert light/thermal energy directly into electricity/ In order to create a viable PV data base for 
planetary mission planners and cell designers, we have compiled low intensity low temperature (LILT) I-V data on 
single junction and multi-junction high efficiency solar cells. The cells tested here represent the latest photovoltaic 
technology. Using this LILT data to calculate Short Circuit Current (\J t Open Circuit Voltage (VJ, and Fill Factor 
(FF) as a function of temperature and intensity, an accurate prediction of ceil performance under the AMO spectrum 
can be determined. When combined with Quantum efficiency at Low Temperature (QULT) data, one can further 
enhance the data by adding spectral variations to the measurements. This paper presents an overview of LILT 
measurements and is only intended to be used as a guideline for material selection and performance predictions. 
As single junction and multi-junction cell technologies emerge, new test data must be collected. Cell materials 
included are Si, GaAs/Ge, GalnP/GaAs/GaAs, InP, InGaAs/lnP, InP/lnGaAs/InP, and GalnP. Temperatures range 
down to as low as -180°C and intensities range from 1 sun down to .02 suns. The coefficients presented in this 
paper represent experimental results and are intended to provide the user with approximate numbers. 


With increasing concerns over the safety and cost of RTGs, alternate power sources are being sought. NASA's 
current stand on this issue is to avoid using nuclear power sources unless there is no feasible alternative. One 
such alternate source of power is photovoitaics, which are widely used today in both space and terrestrial power 
systems. Most solar cells are designed to operate at 1 sun intensity (AMO, 136.7 mW/cm 2 ) and moderate 
temperatures (20° to 80°C). As space exploratory missions extend beyond earth's orbit, temperature and intensity 
become a concern. Missions are being proposed for Mars, Jupiter, the outer planets, and beyond the solar system. 
At these distances, both intensity and array operating temperature drop, intensity changes inversely as the square 
of the distance. Temperature calculations fcre based on intensity and emissivity. The array temperature can be 
as low as -14G°C at 6 astronomical units (A.U.), i.e. Jupiter intensity is 5 mW/cm 2 and -130°C at 5.2 A.U. (1). A 
plot of Intensity vs distance is shown on the following page, this plot also includes relative array temperatures at 
various planetary distances. 

With early LILT measurements dating back 1 5-25 years, most of the available data is outdated. Solar cells have 
become more efficient and more reliable over a range of environmental conditions. Early LILT data was also 
performed using older techniques with limited temperature and intensity regulation, and less sensitive measuring 
equipment. Flight hardware costs continue to increase, which decreases their allowable design margins . Updating 
these measurements is crucial for the recent resurgence in PV for interplanetary missions. 

Most temperature effects on solar cell output are understood. As cell temperature drops open circuit voltage V M 
will increase linearly, and short circuit current l^ will decrease due to a shift in bandgap (the absorption coefficient 


also decreases with temperature). Fill Factor 
will tend to increase proportionally with voltage 
but there are many other mechanisms that 
contribute to its temperature dependence (2). 
The most important effect is that the dark 
current l decreases as temperature decreases. 
The temperature effects on voltage and current 
can be seen in the following equations (3,4): 

v - y kT 






V " J 




r ~i 


i(V)~ L-k 

e \lXTl _ ^ 




4 a ?' 

t e Tr 




Distance From tine Sun (A.U.) 

Figure 1 : Solar Intensity vs. Distance From the Sun 

where T is temperature, y is the ideality factor, 
typically between 1 and 2, k is Bottzman's 
constant, E^ is the bandgap, and q is the 
charge on an electron. As temperature 
decreases, the bandgap of the semiconductor 
material increases. This decreases the 
spectrum which can be absorbed and reduces 
the photocurrent. 

Other LILT effects are not well known. Tandem cells in series must be current matched. As the band gap shifts 
with temperature, the current matching may be lost. As cells drop in temperature and intensity, these changes can 
be nonlinear. Cells may become shunted and/or carriers and dislocations may be "frozen out". Three common 
LILT phenomenon that lead to performance degradation include cell shunting, formation of a rear contact Schottky 
barrier, and the "broken knee" or llat spot" curve shape (5,6). 


The cells used for this experiment represent a broad range of new cell materials. Only one of the cells tested 
was obtained from a production run; all other cells were grown in research labs. These materials were grown on 
substrates which include Si, GaAs, Ge, and InP. The cells are: 

• GalnP/GaAs two-terminal monolithic tandem grown on GaAs. 

- GalnP cell on GaAs (inactive) 

- GaAs cell with a GalnP window layer. 

• InP/lnGaAs two-terminal monolithic tandem grown on InP. 

-InP cell 

- InGaAs cell with a InP window and grown latticed matched on InP 

• GaAs/Ge (passive Ge), GaAs grown on Ge. 

• Si 2 Q<cm with BSF. This a production cell. 


• 12 eV InGaAs (InP window, InP substrate) 

• GaSb (bottom cell of GaAs on GaSb tandem stack) 


The test consisted of measuring IV curves of solar cells at varying light intensities and temperature. The 



Quartz Window 



Test Plate 



Figure 2 LILT Test Setup 

temperatures ranged from 25°G to -185°C. The 
intensities ranged from 1 sun down to .03 suns, or 
equivalent distances of 1 to 6 au. I-V curves were run 
every 25°C at 2.8, 4.7, 11.5, 46, and 136.7 mW/crri 2 
intensities. The information included in this paper is 
only a summary of the data analysis. Figure 2 shows 
a diagram of the test setup. 

The tests were all conducted at NASA Lewis in the 
Solar Cell Evaluation Lab. A Spectrolab X-25 solar 
simulator was used to measure the cells. This 
simulator provides a close match to the AMO spectrum 
but ft is not exact. A monitor cell was placed outside 
the low temperature plate to correct for flicker in the arc 
lamp light source. All the cells were mounted to a test 
plate and placed in a closed environment with a quartz 
window and constant nitrogen purge. Temperature of 
the test plate was maintained by cooling with liquid 
nitrogen and heating with resistive heaters. Up to eight 
cells can be tested simultaneously with this setup. All 
of the cell measurements and temperatures are 
computer controlled. Cells were measured with 
standard 4-wire techniques and contacted using Kelvin 
probes; no epoxies or solders were used to contact the 

A single thermocouple embedded in the test plate is used for temperature control. Additionally, four witness cells 
of similar material and thickness as the test cells were mounted to the test plate and used as a temperature 
reference for the cells. A temperature measurement was made at the beginning and end of each IV curve so that 
accurate V^ vs T and I sc vs T correlations could be made. Typically, a temperature drift of less then 2° was 
observed during an IV curve. Each IV curve was performed from V^ to l sc . 

Light intensity was set up for 1 sun by adjusting the lamp intensity to match l 8C on a calibrated GaAs/Ge cell at 
the plane of the test cells. Intensity was decreased by using metal screens, which lower the amount of light on the 
cells without changing the spectrum. The cells were placed far enough behind the screens to avoid 'hot spots' on 
the individual cells. 


All the test data was used to calculate temperature coefficients for V^, l sc , and FF. The data analysis is 
presented by cell type. Any anomalies in the cells are shown in the plots of the data or mentioned in the text. All 
of the data are normalized to the value at 25°C so that they can be used independently of cell size. Temperature 
coefficients are presented in Tables I and II on the following pages. All of these cells were optimized for 1 sun 
or greater intensities. 


The GalnP/GaAs cell is a monolithic tandem cell consisting of series connected current matched cells. The cells 
are series connected using a tunnel junction. This cell had nearly linear temperature/intensity dependence to about 
-90°C t with peak efficiency at around -50°C. Below -90°C, the cell voltage flattened and then dropped to near room 
temperature values. A plot of this data at 1 sun is shown in Figure 3. This loss of output below -90°C can be 
attributed to the eventual current mismatch of the two cells, parasitic tosses in the iunmi junction, and additional 
voltage loss from changes in dark current. 

A GalnP cell and a GaAs cell with a GalnP window layer were measured separately. Data on these two 
individual cells show that the drop in current is due to limiting by the bottom cell. Both of these cells continue to 
operate well below -90°C and indicate that the probable loss in tandem performance could be in the tunnel junction. 



The InP/lnGaAs cell is a monolithic tandem cell consisting of series-connected current matched cells. This ceil 
also had typical temperature/intensity dependence to about -90°C. This cell had a peak efficiency at near -90°C. 
Below -9.0°C, the cell voltage becomes nonlinear A plot of this data is shown in Figure 4. The voltage change 
does not coincide with the current drop. 

Plots of an InP cell and an InGaAs cell with an inP window layer measured separately show typical 
temperature/intensity dependence over the entire range of measurements. The voltage slope of both cells tends 
to lessen below -90°C. The current of the InGaAs cell changes very little with temperature. This is due to the shift 
at both ends of the spectrum. The InP window layer is shifting along with the band edge of the InGaAs cell which, 
when integrated over an AMO spectrum, shows little net change in current. This is clearly demonstrated in the 
QULT measurements (7). 

~~ 1.75 

O 1.50 

| 1.25 

cc 1.00 






* — "« 

+--+ FF 



' ' \/ 






Temperature (°C) 

Figure 3 GalnP/GaAs at 1 Sun 

Temperature (°C) 

Figure 4 InP/lnGaAs at 1 Sun 

Si Cells 

The Si cell is a 2fl«cm cell with a BSF. The 1 sun temperature data is shown in Figure 5. Below -100°C the 
voltage slope is much lower. This cell had typical temperature/intensity dependence over the entire range of 
measurements. Si efficiency increased by 70% from 25° down to -180°C, where it peaks. This cell tends to 
operate the best at low temperature due to its shift in bandgap. The bandgap shifts from 1 .21 eV up to 1 .45 eV, 
which is the optimum bandgap single-junction cells under AMO. 

InP Cell 

This InP cell had typical temperature/intensity dependence over the entire range of its measurements. The 
voltage slope did change at temperatures below -75°C, but the change was not as much as seen on the previous 
cells. The efficiency on this cell continued to rise over the entire temperature range, increasing by 30% from room 
temperature down to -180°C. 

InGaAs Cell 

The InGaAs cell is grown lattice matched^.72 eV) to InP with an InP window layer. The voltage also exhibits a 
prominent slope change below -100°C. The two InGaAs cells measured here had slightly different coefficients, 
which may be a function of their design {two different research labs). 

GaAs/Ge and GaAs Cell 
The GaAs/Ge cell was cut down from a large area cell and shows severe shunting at low intensities due to the 
cutting. Full area cells had no shunting problems. This cell also had a slope change in voltage below -75°C. The 
cell had a Sehottty barrier at temperatures below -125°C, seen as a bend in the IV curve near V^. 

Low Intensity measurements were conducted on all cells at every temperature recorded above. The behavior 
followed predicted performance within the ranges of the temperature coefficients presented above. 

of yandV^ 


The short circuit current varied linearly with intensity 
and the open circuit voltage varied with the linearly 
logarithm of 1 9C . The Fill Factor tended to follow V^. 
The GalnP/GaAs cell at room temperature and -90°C 
data follow typical temperature trends. The changes in 
voltage slope at lower temperatures reflect possible 
changes in dark current l as voltage is defined in 
equation 1. 


The basis for this paper is to attempt to create a data 
base for temperature coefficients for a wide variety of 
current cell structures. Use of these coefficients can be 
derived from the following equation: 

1 dPffi« _1 $m . 1 ** m i </FF 


dT + VL 

dT FF dT 


-50 -100 

Temperature (°C) 

-150 -200 

Figure 5 Si at 1 Sun 

From the above equation, which is based on the 

maximum power point, temperature correction can be 

applied directly. Simpler techniques apply correction 

to v oc. Iso and p max ( or FF ). then use curve fitting to generate the IV curve. This correction works well with normal 

IV curves, but does not accurately represent larger cells or arrays which contain steps or inconsistencies in the IV 

curve. The following two equations can be applied on a point by point basis to generate an approximate 

temperature corrected IV curve. 

r new tomp 

* v« 




V I 

moas meas 

1 dFF 
FF dT 



» L 


1 * . 

l ie dT 





The Fill Factor correction is applied to the voltage equation, but it could be used in the current equation if preferred. 
Second order equations can be substituted directly for the single coefficients. In all cases, voltage goes up and 
current goes down as temperature decreases. For use in arrays, series and parallel multipliers must also be used 
(series ceils add in voltage, parallel cells add in current). 


The data presented in this paper presents a brief overview of the temperature and intensity characteristics of new 
cell technologies. The temperature coefficients will help create a database for mission planners. This work is a 
continuation of the QULT and LILT measurements published previously (7,8). A comparison of the results of this 
paper with those obtained by QULT show^that l 8C obtained with temperature-dependent spectral response is in 
good agreement with 1^ dependence measured with an AMO simulator. It should be noted that temperature 
coefficients tend to vary among similar cells, and the spectrum of the X-25 simulator does not exactly match the 
AMO spectrum (it contains more infrared and less ultraviolet). 

The coefficients are indicated for the typical characteristics of cells showing common trends. These common 
trends are; higher bandgap cells have lower coefficients; voltage increases and current decreases with lowering 
temperature; V^ is proportional to the log of intensity, current is directly proportional to intensity, and fill factor tends 
to drift up to a peak and drop down. 

Although multi-junction cells offer higher efficiency than single cells, they do present problems if used over a wide 
range of temperatures. Monolithic tandem cells must be designed to match current over a wide range of 


temperatures, where changes in temperature cause a 
shift in bandgap. In both tandem cells presented here, 
the bottom cell current remained relatively flat, this is 
due to the bandgap shift of both cells, the spectral 
window to the bottom cell remained constant. Tandem 
cells measured here worked well together to -90°C and 
then started to drift nonlineaiiy. 

Most of the cells measured exhibited two slope 
curves for V^ vs temperature. This characteristic is £ 

O -.005 


as well as Jm. Different 

5 -.010 


. ©■- 



a q InGaAs Bottom 

■■- --■ GalnP/GaAs 
*~* InP 
o- — © $1 

indicative of a change in the i 
recombination mechanisms affect different voltage 
ranges and temperatures, i.e., Hall Schottky Read, 
tunneling recombination, junction recombination, and 
surface recombination. The voltage slope at lower 
temperatures tended to be less then near room 
temperature. Within the range of temperatures 
measured for most cells, a peak in fill factor peak could 
be observed; this required a second order equation for 
curve fitting. 

The plots shown in Figure 6 indicate that the voltage 
coefficients tend to increase linearly as a function of the 

log of intensity and that their slope also increase with decreasing bandgap. This trend can be mathematically 
demonstrated. It can be used to extrapolate temperature coefficients for a wide range of intensities. 

The authors would like to graciously thank National Renewable Energy Labs, Applied Solar Energy Corporation, 
Spire Corporation, Boeing Corporation, and JX Crystals for providing cells which were used for these 
measurements. The authors intend to continue to add to this data as new requirements and cells become 

Relative Intensity (1*AM0) 

Figure 6 V^ Coefficient vs Log(Intenstty) on Cells 


1) Ho, J.C., Barteis, F. T. C M and Kirkpatrick, A. R., 1970, "Solar Cell Low Temperature, Low Solar Intensity 
Operation". Conference record of the 8th IEEE Photovoltaic Specialists Conference, pp. 150-154. 

2) Hovel, Harold ^Semiconductors and Semimetals, Volume 1 1 , Solar Cells", Academic Press, 1975, pp 166-180 

3) Hu, Chenming, and White, Richard M., 1983, "Solar Cells, From Basic to Advanced Systems", McGraw-Hill , pp 

4) Yang, Edward S M "Fundamentals of Semiconductor Devices", McGraw Hill Book Company, 1978, pp 103-108. 

5) Stella, Paul M., Pool, Frederick S., Nicolet, Marc A., and lies, Peter A., 1994, "PV Technology for Low Intensity, 
Low Temperature, (LILT) Applications", First World Conference on Photovoltaic Energy Conversion, pp. 2082- 

6) Weizer, V. G., and Broder, J. D., 1982, "On the Cause of the Flat-Spot Phenomenon Observed in Silicon Solar 
Cells at Low Temperature and Low Intensities", J, Appl. Phys., Vol. 53, No. 8, August 1982, pp. 5926-5930. 

7) Jenkins, Phillip, Scheiman, David, and Brinker, David, 1994, "Prediction Short Circuit Current of Single Junction 
and Multi-Junction Solar Cells at low temperature for Planetary Missions", First World Conference on Photovoltaic 
Energy Conversion, pp. 2012-2015. 

8) Scheiman, David A., Jenkins, Phillip P. f Brinker, David J., 1995, "Low Intensity Low Temperature (LILT) 
Measurements on New Photovoltaic Structures", 30th IECEC, 95-353, Vol. I, pp.327-332 



TABLE 1 : Temperature Coefficients of Multi-BandaaD Cells and SubCells 


1 Sun 

.33 Suns 

.1 Suns 

.04 Suns 

.02 Suns 





-6.6E-6T - 8.29E-4 
alnP window) 


7.02E-5T + .00296 


-7.0E-6T - 8.94E-4 

-0.00257 ,1 > 
-0.00137 (5 » 



-1.1584E-5T-. 00265 

-3.70E-6T - 


-7.16E-6T - 

-0.00256 (^, 




-9.02E-6T - 

-1.72E-5T - 

f 9.61E-4 


.001 1 m 




-0.00325 (1 » 

-0.00263 <2, 


4.78E-6T + 

2.66E-6T + 
-5.86E-6T - 

-2.16E-5T - 











-ooDies* 61 



1.182E-5T ■ 

-4.42E-6T - 




f 5.87E-4 

dVoo/dT 25°,-100° 
dlsc/dT 25°,-100° 
dFF/dT 25°,-100° 

dVoo/dT 25°,-180° 
dVa/dT -100°,-180« 
dlsc/dT 25°,-180° 
dFF/dT 25°,-180° 
GaAs Bottom Cell (G. 

dVoc/dT 25°,-125° 
dVoc/dT -75V180 
dlsc/dT 25°,-180° 
dFF/dT 25°,-180° 


-0.00136 (6) 




-0.00246 (7) 


-5.92E-6T - 8.54E-4 
InP window) 

dVoc/dT 25°,-100° 
dlso/dT 25V100 
dFF/dT 25°,-100* 

dVoo/dT 25°,-75° 
dVoc/dT -75°,-180° 
dlsc/dT 25°,-180° 
dFF/dT 25°,-180° 

InGaAs Bottom Cell I 

dVa/dT 25°,-100° 
dVoc/dT -100V180* 
dlsc/dT 25°,-180° 
dFF/dT 25°,-180° 


-1.114E-5T-. 00226 

Notes: (1) 25°,-100°C; (2) 25°,-75°C; (3) 25»,-125°C; (4) -75°,-180°C; (5) -100°,-180°C; (6) -125V180°C; (7) 25V180°C 
dVoc/dT = (V/V)/°C 
dlsc/dT « (A/A)/°C 
dFF/dT = (%/%)/°C 

Table II: Temperature Coefficients of Single Junction Cells 



1 Sun 

.33 Suns 

.1 Suns 

.04 Suns 

.02 Suns 

GaAs Concentrator Cell 


-7.2E-6T + 1.5E-5 

dVo/dT 25°,-125° 
dVo/dT to -180° 
dlgc/dT 25°,-180° 
dFF/dT 25°,-180° 
GaAs/Ge (cut from 6 


-1.242E-5T + 1.53E-4 
-7.78E-6T - .001 

dVoo/dT 25°,-75° 
dVo/dT -75°,-180° 
dlsc/dT 25V180 
dFF/dT 25°,-180° 

dVoc/dT 25°,-75° 
dVo/dT -75V180 
dlsc/dT 25°,-180° 
dFF/dT 25°,-75° 
Si (2Q-cm.) 


-4.24E-6T + 2.61E-4 
-1.492E-5T + 3.13E-4 


-.00140 w 

2.98E-6T + 6.75E-4 
-8.90E-6T-. 00152 
laver . .72 eV) 


9.96E-6T + .00144 
-1.176E-5T-. 00130 

8.66E-6T - 9.0E-4 

-0.00589' 1 ' 
-0.0044 <S| 

3.38E-6T + S.73E-4 
9.64E-6T + 9.16E-4 


1.28E-5T + .00108 

dVoo/dT 25°,-75° 
dVoc/dT -75V180 
dlsc/dT 25V180 
dFF/dT 25°,-180° 
InQaAs (InP window 1 

dVo/dT 25°,-100° 
dVoo/dT -100°,-180 < 
dlsc/dT 25°.-180° 
dFF/dT 25°,-180° 

dVoo/dT 25°,-75° 
dVoc/dT -75°,-180° 
dlsc/dT 25V180 
dFF/dT 25°,-180° 

' -0.00348 
3.38E-06T + 2.12E-4 
-1.55E-5T-. 00138 


-3.34E-5T + 2.41E-4 
-8.20E-6T - .00225 

Notes: (1) 25°,-100°C; (2) 25° 
dVoc/dT » (V/V)/°C 
dlgc/dT = (A/A)/°C 
dFF/dT «(yj%)/°C 

-75°C; (3) 25°,-125°C; (4) -75°,-180°C; (5) -100V180°C; (6) -125V180°C; (7) 25 o ,-180 o C