Professional Documents
Culture Documents
Experimental Measurements of Small Triple Juction PV
Experimental Measurements of Small Triple Juction PV
1. INTRODUCTION
Multi-junction solar cells recently demonstrated conversion efficiencies above 40% at a
concentration (power density) of several hundred suns [1] (one sun ≡ 1 mW/mm2). High
concentration realized with inexpensive high-flux optics reduces the system share of costly
photovoltaic (PV) cells. Practical considerations then motivate cell miniaturization which can (1)
raise the concentration at which efficiency peaks [2-4], (2) facilitate passive heat rejection and (3)
permit all-glass compact miniaturized optics [5,6] for attaining high concentration at high collection
efficiency with completely passive heat rejection on cell areas of ~1 mm2.
2. EXPERIMENTAL DETAILS
We present measurements on commercial triple-junction GaInP2/GaAs/Ge solar cells7 (Fig 1),
generated with an ultrahigh flux real-sun fiber-optic mini-dish concentrator [8-11] (Fig 2), and
elucidate the principal cell parameters for appraising concentrator PV performance for broad ranges
of anticipated operating conditions.
The dual-axis tracking concentrator [8] (Fig 2a) includes transmissive (quartz-core) optical fibers
1.0 and 0.6 mm in diameter that comprise a localized irradiation probe (LIP) [10,11]. Delivered light
is confined to the circle delimited by the fiber tip (Figs 2b and 3a). This procedure can simulate the
intensity and type of non-uniform distributions produced by high-flux optics [3,5,6]. Uniform
irradiation of the cell's 1.0 mm2 active area within the busbars was produced with a square glass
kaleidoscope coupled between the fiber and the cell (Fig 2c). Solar irradiation on the cell Pin was
moderated with an iris (Fig 2a) and measured pyrometrically. The highest local concentration with
the 0.6 and 1.0 mm fibers was 3700 and 5100 suns, respectively. The light spectrum on the cell was
nearly invariant and close to the air mass 1.5 direct beam solar spectrum [8].
Current-voltage (I-V) curves were measured under: (a) uniform illumination of the full 1.0 mm2
active area within the busbars, (b) LIP with the 1.0 mm fiber, and (c) LIP with the 0.6 mm fiber.
Cells were mounted on a 1.6 mm thick gold-coated Kovar ceramic substrate which was thermally
bonded to a passive copper heat sink. The maximum temperature at the heat sink-substrate interface
was only 3 K above indoor ambient (actual cell junction temperature Tcell cannot be measured
directly and exceed that of the interface).
Tcell increases with Pin, Voc may decrease as Pin is raised. Because our passive heat sinks were
intended to simulate actual concentrator operation and therefore allowed junction temperatures that
are non-negligibly above ambient, we filtered the temperature effect by performing measurements in
a flash (μs) solar simulator where the cell was uniformly illuminated and actively cooled to within
±1 K of ambient (Fig 3a). With heating mitigated, the effect of Rs in lowering Voc below that of Eq
(3) becomes evident at Pin above ~260 mW, corresponding to a flux of ~190 suns with flash
measurements (where active cell area includes the 0.35 mm2 perimeter region not covered by the
busbar – see Fig 1), ~260 suns with the 1.0 mm kaleidoscope, and ~920 suns (localized) with the 0.6
mm fiber (Fig 3a). This signature of flux non-uniformity likely derives from the large busbar area
and the substantial parasitic dark current it contributes, which enhances the distributed character of
Rs losses [14]. The fact that plots of Voc against log(Pin) grow sub-linear at higher Pin (Fig 3a)
indicates that an improved front contact configuration could both enhance η and increase the
concentration at which η peaks, e.g., reconfigure the busbar toward diminishing dark current losses.
(E) The plot of η against log(Pin) (Fig 3b) (i) is linear at low flux, with a slope proportional to
Isc/Pin (here 0.138 A/W); (ii) decreases non-linearly at high Pin due to Rs dissipation; and (iii)
possesses a maximum that reflects this tradeoff. Under uniform illumination of the active 1.0 mm2
area within the busbars, η is maximum at ~1,000 suns (Fig. 3b) - considerably higher than the ~350
suns at which η of earlier 100 mm2 cells of the same nominal architecture peaked [12]: the
miniaturization of a solar cell raises the flux at which η is maximized.
(F) Reducing cell size can raise η at high flux primarily due to the decrease in both absolute
photo-current and front grid resistance (given that the metal finger height and depth, as well as the
grid density, are roughly independent of cell size). Other contributions to Rs (e.g., contact, emitter
and tunnel diode resistances) exhibit a far weaker (even negligible) dependence on cell size [2]. As a
consequence, there is a clear trend but no simple scaling relation between effective Rs and cell area.
Our experimental results regarding higher concentration values at which η peaks in smaller cells are
consistent with calculations of optimized cell areas that account for the tradeoff between resistive
and perimeter recombination losses in single and triple-junction cells [2,4].
(G) The cell parameter most sensitive to Rs, and hence to irradiation distribution, FF, remains
roughly independent of both Pin and flux distribution up to Pin ~1.0 W (Figs 3c). FF and η worsen
with more strongly localized Pin at higher concentration as Rs effects are expressed more prominently
[11]. Under highly localized irradiation, η (with the 0.6 mm fiber) peaks at lower Pin (~800 mW) but
at higher localized flux (~2830 suns) – consistent with the above-noted benefit of cell
miniaturization. Nonetheless, a prediction of the concentration at which η would peak for a cell of
diameter 0.6 mm is not straightforward, mainly due to possibly larger relative losses from perimeter
recombination.
4. CONCLUSIONS
This report constitutes characterization studies of ultra-small ultra-efficient multi-junction
concentrator solar cells, with particular attention to sensitivity to concentration and flux distribution.
The adequacy of the metallization and busbar design had remained to be established. The irradiation
protocols and flux maps used for cell interrogation here can correspond to situations encountered in
concentrator photovoltaics. As such, they shed light on anticipated performance penalties (or the lack
thereof), as well as how cell architecture could be judiciously reconfigured.
REFERENCES
[1] R.R. King, D.C. Law, K.M. Edmonson, C.M. Fetzer, G.S. Kinsey, H. Yoon, R.A. Sherif and N.H. Karam, Appl.
Phys. Lett. 90 (2007) 183516.
[2] C. Algora and V. Diaz, Prog. Photovoltaics 8 (2000) 211.
[3] A. Bett and H. Lerchemueller, in Concentrator Photovoltaics, eds. A. Luque and V.M. Andreev, p. 301-319
(Springer, Heidelberg, 2007).
[4] K. Nishioka, T. Takamoto, T. Agui, et. al.., Sol. Energy Mat. Sol. Cells 90 (2006) 1308.
[5] R. Winston and J.M. Gordon, Opt. Lett. 30 (2005) 2617.
[6] S. Horne, G. Conley, J. Gordon, D. Fork, P. Meada, E. Schrader and T. Zimmerman, IEEE 4th World Conf. on
Photovoltaic Energy Conv. Proc. 1, 694-697 (IEEE, Piscataway, NJ, 2006).
[7] Spectrolab Inc., 12500 Gladstone Ave., Sylmar, CA, G. Glenn, Technical prospectus and private communications.
[8] J.M. Gordon, E.A. Katz, D. Feuermann and M. Huleihil, Appl. Phys. Lett. 84 (2004) 3642.
[9] J.M. Gordon, E.A. Katz, W. Tassew and D. Feuermann, Appl. Phys. Lett. 86 (2005) 073508.
[10] E.A. Katz, J.M. Gordon and D. Feuermann, Prog. Photovoltaics 14 (2006) 297.
[11] E. A. Katz, J. M. Gordon, W. Tassew and D. Feuermann, J. Appl. Phys. 100 (2006) 044514.
[12] R.R. King, R.A. Sherif, G.S. Kinsey, S. Kurtz, C.M. Fetzer, K.M. Edmonds, D.C. Law, H.L. Cotal, D.D. Krut, J.M.
Ermer and N.H. Karam, 3rd Int. Solar Conc. Conf. for the Generation of Electricity or Hydrogen, Proc. NREL/CD-520-
38172 (2005).
[13] M. S. Carpenter, M. R. Melloch, M. S. Lundstrom, and S. P. Tobin, Appl. Phys. Lett. 52 (1988) 2157.
[15] V.M. Andreev, V.A. Grilikhes and V.D. Rumyantev, Photovoltaic Conversion of Concentrated Sunlight (Wiley,
Chichester, 1997).