ELSEVIER Solar Energy Materials and Solar Cells 43 (1996) 29-36

Solar Enerw Mabmals and Solar C.eUs

Investigation of the current break-down phenomena in solar cells

S.K. Sharma *, N. Srinivasamurthy, B.L. Agrawal Power Systems Group, ISRO Satellite Centre, Bangalore 560017, India

Abstract

Observed reverse current-voltage characteristics of the single crystal silicon and gallium arsenide solar cells have been analyzed. Physical mechanisms behind the junction break-down in silicon cells and current break-down in gallium arsenide cells have been identified. Preliminary estimates of the diffusion capacitance in GaAs cells have been presented.

Keywords: Junction break-down; Current break-down; Interband impact generation; Quantum mechanical tunneling; Diffusion capacitance

1. Introduction

Heteroface single crystal aluminium gallium arsenide/gallium arsenide (GaAs) solar cells offer potential advantages over single crystal silicon (Si) solar cells for space applications. Notable amongst them are: (i) higher conversion efficiency, (ii) higher radiation resistance to proton irradiation, and (iii) compatibility with high temperature applications.

Production level efficiency of the GaAs solar cells under 1 sun (air mass zero-135.3 m W / c m 2) illumination at 28°C, at beginning of life is 18 to 20%. The corresponding efficiency figures for the Si solar cells vary from 12 to 15% depending upon the cell fabrication technology, viz., back surface reflector (BSR), back surface field and reflector (BSFR), textured front surface and passivated emitter, etc.

The higher radiation resistance of GaAs solar cells to protons in space is a consequence of the lower proton damage coefficient of the GaAs material compared to Si. Typical 10 MeV proton to 1 MeV electron conversion factor for GaAs cell is only

* Corresponding author.

0927-0248/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved SSDI 0927-0248(95)00154-9

30 S.K. Sharma et al. / Solar Energy Materials and Solar Cells 43 (1996) 29-36

about 1000 as compared to about 3000 for Si. GaAs cells, therefore, are very attractive for orbits predominated by protons, where Si cells would suffer heavy degradation.

Compatibility of GaAs solar cells to high temperature applications arises due to the fact that the initial fill factor (FF) and open-circuit voltage (Voc) of GaAs cells are considerably higher. Typically, the FF of GaAs cells is 80 to 85% against 75 to 80% of Si cells, and Voc of GaAs cells is about 1 V against 550 to 600 mV for Si cells. Furthermore, the temperature coefficient of GaAs cell voltage is ~ - 1 . 9 mV/°C against ~ - 2.5 mV/°C for Si cells. These features of GaAs cells permit usage of this cell at higher temperatures compared to Si cells. In fact, GaAs solar cells are the primary candidate for concentrator solar arrays.

One of the main disadvantages of GaAs solar cells, other than high cost, has been the cell electrical power output degradation post-reverse bias stress test.

In this paper, we have attempted to explain the observed behaviour of GaAs solar cells under and post reverse bias test. Also included are the preliminary estimates of the cell diffusion capacitance (C D) which is an important parameter for the design of solar array string switching scheme. The observations on GaAs cells have been compared with the corresponding Si cell data.

2. GaAs versus Ge substrate

Early GaAs cells were produced with GaAs substrate (GaAs/GaAs cells) and have been used in some of the early spacecrafts, e.g., CS-3 & UOSAT-F, etc. [1,2]. High fragility and high cost of GaAs wafer have been the major bottlenecks in wide applications of GaAs solar cells for spacecraft power generation. Efforts to reduce the cell cost and enhance the cell/panel production yield led to deposition of GaAs active epitaxial layers on germanium (Ge) substrate (GaAs/Ge cells). Ge substrate offers the additional advantage of reduction in substrate thickness down to 90 microns limited only by the cell mechanical strength for assembly into solar panels. The interface between GaAs and Ge in GaAs/Ge cells is rendered electrically in-active. This has further relaxed electrical performance requirements of the Ge substrate, and thereby further reduction in GaAs/Ge cell cost. GaAs/Ge solar cells have been fully qualified for space applications, and millions of cells have already been used on various spacecrafts. Ge has more or less completely replaced the GaAs substrate for production of space quality GaAs solar cells.

3. Importance of solar cell reverse current-vol tage characteristics

Typical reverse current-voltage characteristics of GaAs and Si solar cells are shown in Fig. 1. Typical reverse break-down voltage (V R) for GaAs cells is about - 5 V as against - 3 0 V to - 8 0 V for Si cells. Further, the nature of break-down in GaAs solar cells is not similar to that of Si cells.

The reverse current-voltage characteristics of the cell decide power dissipation by the cell during periods of shadow on solar array string. Power dissipation in the cell leads to

S.K. Sharma et al. / Solar Energy Materials and Solar Cells 43 (1996) 29-36 31

I i I I I I i i I I

- 2 8 - 2 6 -2 / . - 2 2 - 2 0 -18 -16 -1/, -12 -10 - 8 - 6

' i

- t . - 2

80

70

60

o x

/ . 0 - -

30 ~o

ZO

10

REVERSE VOLTAGE IV)

Fig. 1. Reverse current-voltage characteristics of silicon and gallium arsenide solar cells at room temperature.

rise in cell temperature, which, if excessive, may cause cell puncturing (shorting of p-n junction), hence loss of power, and also loss of insulation between solar cell and solar panel substrate. In case of Si cells, V R is large and the cell is forced to dissipate higher power leading to a high rise in cell temperature. The cell, however, need not necessarily be driven to break-down region as the reverse voltage across the cell is limited by the forward voltage (Vf) generated by the un-shadowed part of the solar array string. If Vf exceeds V R, the cell will operate in the break-down region, else not. Typical values of Vf for the operational spacecrafts vary from 28 V to 42 V, depending on the mission. The solar cell in the GaAs cell string under similar operational conditions, due to lower V R, shall be forced to operate in the break-down region. The power to be dissipated by the cell (for the same string current), however, is significantly lower than that for the Si cell and, hence, rise in GaAs cell temperature is considerably lower than that in Si cell. To this extent, lower V R of GaAs cells is an advantage for their use on solar arrays.

The electrical power output of GaAs cell may, however, degrade due to small current flow in the reverse biased mode [3], [4]. This will lead to subsequent reduction in solar array power output during non-shadow periods in orbit due to (i) degradation in cells exposed to reverse bias stress, and (ii) current and voltage mismatch between the cells exposed to reverse bias stress and those not, in the string. This is in contrast to the Si cell where no such observation has been reported over several years of their application for spacecraft power generation. Is a lower V R of GaAs solar cells responsible for their degradation under reverse bias stress? We shall examine this later in this paper. But before that let us briefly discuss the prevailing reverse bias test conditions for GaAs solar cells and observations thereof.

32 S.K. Sharma et al. / Solar Energy Materials and Solar Cells 43 (1996) 29-36

An in-line reverse bias screen test is conducted on 100% GaAs solar cells to filter out the weak cells. The cells are tested at current which is typically 35% of the cell short-circuit current (I~c) for about 5 s at room temperature, followed by the electrical performance measurement. The reverse bias screen test current of 0.35I~c is chosen to ensure that the cell operates in the break-down region. Cells degrading by more than 3% in electrical performance after this test are generally not considered for on board application. It is also reported that on an average only about 35% of the GaAs/GaAs cells that passed 0.351~c test go through the high reverse current test ( ~ 1.7Isc for 30 minutes at 90°C) and reverse bias cycling as against 100% for GaAs /Ge cells [3]. This has led to conclusion that the in-line reverse bias screen test such as at 0.35!~ c is adequate to stabilise the electrical performance of GaAs /Ge cells, and cell efficiency figures should be quoted only after this test. GaAs/GaAs solar cells, on the other hand, have to be reverse bias tested to the expected shadow conditions on the solar panel, followed by the electrical performance measurement to avoid uncertainty in the solar array performance in orbit. Alternative to this is the use of shunt diodes across group of series cells in the strings located in the shadow prone area on solar panel [3].

4. Analysis of the current break-down phenomena

To explain the observed reverse current-voltage characteristics of Si and GaAs solar cells shown in Fig. 1, let us recall the structure of GaAs and Si cells. Active region in Si cells extends typically upto 200 microns (base thickness) and base doping level is of the order of 1015 to 1016 a toms /cm 3. In the GaAs cell, the active region thickness is only about 5 microns and total cell thickness including substrate is 100 to 200 microns for GaAs /Ge cells and 250 to 300 microns for GaAs/GaAs cells. The doping level in substrate and active regions in GaAs cells are of the order of 1017 to 1018 a toms/cm 3.

The reverse voltage applied across Si cell increases the space charge layer width and the reverse saturation current due to thermally generated minority carries near and inside the space charge region, flows across the junction. The thermal generation rate is very small and so is the reverse saturation current. As the reverse voltage is increased, the minority carriers gain higher and higher kinetic energy and at a particular reverse voltage, the carder energy is sufficient to cause inter-band impact generation of electron-hole pairs [5]. The magnitude of the reverse voltage at the on-set of interband impact generation of electron-hole pairs is such that the kinetic energy of the thermally generated minority carriers is several times the value of band gap energy (1.2 eV for Si). This voltage is referred to as the reverse break-down voltage (VR). The electron-hole pairs generated through interband impact generation are also accelerated by the electric field and hence generate additional electron-hole pairs. The regenerative process so set in diverges and result into large current corresponding to p-n junction break-down. The junction break-down due to interband impact generation of electron-hole pairs at large V R is also referred to as "Avalanche" break-down. The steady increase in cell current with reverse voltage for Si cell prior to p-n junction break-down, shown in Fig. 1, is suggestive of the interband impact or avalanche break-down in Si cells.

Applied reverse voltage in the GaAs solar cell increases the space charge layer width

S.K. Sharma et al. / Solar Energy Materials and Solar Cells 43 (1996) 29-36

Table 1 Silicon and GaAs solar cell typical parameters at room temperature

33

Parameter Si cell GaAs cell

Input data Doping concentration (cm-3) - emitter ( ~ ) - base ( ~ ) Semiconductor permittivity (e~,F/cm) × 8.85418 × 10-14 Intrinsic carrier concentration (Ni, cm-3) VR (V) Electronic charge (q, C) Calculated parameters with V R = 0 Built-in potential (Vbi, V) 0.816 Space charge layer width (xpn, microns) 1.04 Maximum electric field (Ema X , V/cm) 1.57 × 104 Calculated parameters at break-down voltage, V R Space charge layer width ( Xpn, microns) 8.95 Maximum electric field (Ema x, V / c m ) 1.36× 105

1 X 1019 (Nd) 1X 1018 (Na) 1 × 1015 (Na) 5::< 1017 (Nd)

11,9 13.1 1.45× 101° 1.79× 10 6 -60 - 7 1.60218× 10-19

1.38

0.077

3.58 × 105

0.19

0.88 X 106

Where built-in potential (Vbi, V) = 0.0259 In (NaNd/Ni2), space charge layer width (xp,, cm) = [2e~(Vbi + VR)(Na+Nd) / qNaNd] I/2, maximum electric field strength (Ema~, V/cm)= (Vbi + VR)/Xpn. Using the corresponding values of xp. for Ema x calculation. Above equations taken from Refs. [5,7].

and the reverse saturation current, due to thermally generated minority carders, flows across the p-n junction, as in the case of Si cells. The magnitude of the reverse saturation current due to thermally generated minority carriers in GaAs cell is, however, smaller than in Si cell due to larger bandgap of the GaAs material.

Typical values of the space charge layer width (Xpn) and the maximum electric field strength (Ema x) calculated for Si and GaAs solar cells at room temperature are given in Table 1. Space charge layer width in GaAs cells, as shown in Table l, due to higher doping levels (1017 to 1018 a t o m s / c m 3) is an order of magnitude lower than that in Si cells (doping level of ~ 1015 atoms/cm3). The electric field strength ( V / c m ) in the space charge region of GaAs cell is, therefore, an order of magnitude higher than that in Si cell. Higher doping level in GaAs cells also results in lower V R in GaAs cells, compared to Si cells [5] Application of reverse voltage V R across the p-n junction increases the space charge layer width and enhances the electric field strength - - both being proportional to (Vbi + VR) 1/2, where Vbi is the built-in-potential[5], as shown in Table 1.

In contrast to Si cell, the space charge layer in GaAs cells is too thin to allow carders generate electron-hole pairs through interband impact while passing through the space charge region. The phenomenon that causes current break-down in GaAs cells is, therefore, different and can be attributed to the quantum mechanical tunnelling of

34 S.K. Sharma et al. / Solar Energy Materials and Solar Cells 43 (1996) 29-36

majority carriers [5]. The reasoning behind the quantum mechanical tunnelling in GaAs cells is as follows:

Due to a thin space charge layer, the majority carriers on either side of the p-n junction have high probability to tunnel through the space charge layer. The higher the doping concentration, the higher is the tunnelling probability of the majority carriers and, hence, the higher is the tunnel current. Unlike the interband impact generation process, tunnelling is, however, a non-regenerative process. Thus, the tunnel current rises rapidly as the reverse voltage is increased beyond a certain value causing Ema x to

exceed threshold value of ~ 7 × 10 5 V/cm, required for onset of quantum mechanical tunnelling in GaAs p-n junction [7]. The current at break-down will, however, not diverge. The magnitude of current due to tunnelling can, therefore, be large, but not infinite as in the case of interband impact generation in Si cells. The direction of current flow in GaAs cell at current break-down is opposite to that of the thermally generated minority carriers.

One may wonder whether tunnelling may take place in GaAs solar cells, even in the forward biased mode? It is indeed not true, as explained below.

The essential requirement for tunnelling to take place in p-n junction is that there must exist occupied states (initial states) on one side and empty (final states) on the other side of the p-n junction. Also, the initial and final states must have the same energy. Further, the tunnelling potential barrier height and tunnelling distance (width of the space charge layer) should be small [6,7]. In GaAs solar cells, as they are fabricated using highly doped but non-degenerate semiconductors, tunnelling conditions are easily satisfied in the reverse biased mode. In the forward biased mode, however, the empty states (final states) fall in the forbidden gap and hence all conditions for tunnelling are not satisfied. It is for this reason that in highly doped non-degenerate semiconductor p-n junctions, tunnelling contributes significantly to current flow in reverse biased mode but nothing in the forward biased mode.

5. Degradation of GaAs cell under reverse bias stress

As mentioned earlier, the current at junction break-down in Si cells or current break-down in GaAs cells can be very large. This large current is localised at defect sites such as leakage sites where the electric field strength is high [5]. Degradation of GaAs cell electrical power output after reverse bias test occurs due to shorting of leakage paths in the p-n junction, particularly under front metallisation and edges, resulting in apparent decrease in cell shunt resistance. Leakage paths are introduced during front metallisation and cell cutting, etc. Once, the break-down region is reached, further increase in cell current leads to only increased dissipation in cell, without adding any additional short paths. To this extent, the cell electrical performance can be said to have stabilised after initial reverse bias stress test at 0.35Isc. Extent of the cell power output degradation after this test depends on the cell processing technology and tooling, etc. Introduction of a highly doped GaAs cap layer between front metallisation and A1GaAs window layer generally reduces leaky paths due to front metallisation.

The degradation in GaAs cell electrical performance post reverse bias stress test is,

S.K. Sharma et al. / Solar Energy Materials and Solar Cells 43 (1996) 29-36 35

therefore, a consequence of cell processing/fabrication constraints. Low reverse break- down voltage as such may not be responsible for the cell degradation, except that it leads to very thin space charge layer and thereby increases the probability of p-n junction shunting.

6. Cell diffusion capacitance

Solar cell diffusion capacitance (C D) is defined as the rate of change of excess minority carrier charge with cell forward voltage. Cell C D, depending on its magnitude, may pose constraints in the design of solar array string switching scheme. Detailed analysis of C D for Si solar cells has been carried out by Sharma et al. [8]. Although the analysis carried out by Sharma et al. [8] is for Si cells, the theory is equally applicable to GaAs solar cells as well. However, not much information is available with regard to the magnitude of C D in GaAs solar cells. Below, we give the preliminary estimates of C D for GaAs solar cells based on the theory of Sharma et al. [8].

As reported by Sharma et al. [8], C D is highest at Voc condition and is given by K t l s J T . Where K = 11604.85 K / V , t (s) is the excess minority carrier life time, I~ (A) is the cell short-circuit current, and T is the absolute temperature in °K. Typical values of t for GaAs material are of the order of nanoseconds (compared to microseconds for Si cells) and of Isc ~ 30 to 35 m A / c m 2 (as against 35 to 45 m A / c m 2 for Si cells). Simple calculations then show that typical magnitude of CD for GaAs cell shall be of the order of 10 n F / c m 2 as against 15 u F / c m : for Si-BSFR cell at room temperature. C D at any other voltage, as reported by Sharma et al. [8], shall be lower than the value obtained at Voc condition. Further, the qualitative variation of C o in GaAs cell with temperature and 1 MeV electron irradiation fluence shall be similar to Si cells as reported by Sharma et al. [8].

The above results suggest that C O should not pose any constraints on the design of the string switching scheme for solar arrays with GaAs cells.

7. Conclusions

Reverse current-voltage characteristics of Si and GaAs solar cells have been analyzed. The analysis suggests interband impact generation mechanism responsible for junction break-down in Si cells. Whereas in GaAs cells, it is the quantum mechanical tunnelling of majority carriers which leads to current break-down. The diffusion capacitance of GaAs cells is estimated to be three order of magnitude smaller than that of Si cells.

Acknowledgements

The authors are thankful to Deputy Director, ESA, ISAC and Director, ISAC, for granting permission to publish this work.

36 S.K. Sharma et a l . / Solar Energy Materials and Solar Cells 43 (1996) 29-36

References

[I] Mitsubishi GaAs Solar cells for space use, Marketing leaf-let from M/s . Mitsubishi Electric Corporation, Japan.

[2] EEV Ltd. GaAs Space Solar Power, Marketing Leaf-let from M / s . EEV Ltd., UK. [3] G.C. Datum and S.A. Billets, Proc. 22rid IEEE Photovoltaic Specialists Conf., 1991, pp. 1442-1428. [4] P.A. Illes et al., Proc. IEEE Phntovoltaic Specialist Conf., 1990, pp. 448-454. [5] C.T. Sah, Fundamental of Solid-State Electronics (World Scientific, NJ, 1993) pp. 408-451. [6] S. Wang, Solid-State Electronics (McGraw-Hill, NY, 1966) 372-379. [7] S.M. Sze, Semiconductor Devices Physics and Technology (Wiley, NY, 1985) pp. 225-229. [8] S.K. Sharma et al., Sol. Energy Mater. Sol. Cells 26 (1992) 169-179.