NSPARENT OHMIC CONTACTS TO EPITAXIAL InP CELLS

S. J.Taylor, A. Leycuras and B. Beaumont Centre de Recherche sur I'Hetero-Epitaxie et ses Applications,

Centre National de la Recherche Scientifique, Sophia Antipolis, Rue Bernard Gregory, 06560 Valbonne, France

C. Coutal, J. C.Roustan and A.Azema Laboratoire de Physique de la Matiere Condensee

U.R.A. C.N.R.S. 190, Parc Valrose, 06108 Nice, France

ABSTRACT

In this paper we report the deposition of transparent ohmic contacts of indium tin oxide by laser ablation to homoepitaxial InP cells. This method provides technologically simple, low resistance contacts at low deposition temperatures and should not damage the junction of shallow emitter n+/p structures that are required due to the high surface recombination velocity of InP.

INTRODUCTION

Ill-V cells have achieved higher conversion efficiencies than any other materials [I] and will represent one of the most economical alternatives for use in terrestrial concentrator applications if manufacturing costs can be made competitive with other technologies. Of the available materials for Ill-V cells, inP has the advantage of a near optimal band gap for the solar spectrum and a high radiation resistance, which favours its use for space applications [2]. At present the front contact metallization of such cells necessitates the use of complicated and expensive photolithographic procedures as well as inevitably introducing a shadowing loss. The use of a transparent conducting oxide could potentially reduce the processing costs and eliminate this loss. However, transparent conducting oxides introduce other losses due to electrical resistance and optical absorption which have so far proven to be greater than the shadowing losses which are eliminated, so up to now no overall efficiency gain has been achieved through their use.

The most successful oxide to date for such applications has been indium tin oxide (ITO), which has seen widespread use in the production of thin film cells and has been used to form n/p cells from InP which have reached an efficiency of 16.1% (AMO) when used beneath an additional metal contact grid [3]. Although this Is an excellent result, bearing in mind the simplicity of the junction formation, it seems probable that the

sputtering technique used for the deposition of the IT0 introduces defects into the surface layer of the InP substrate, forming a buried homo-junction [4]. This may well be the cause of the lower values of V,, and consequently of conversion efficiency that have been obtained from sputtered lTO/lnP junctions in comparison with epitaxial InP cells [3][4]. We have investigated the application of IT0 contacts, deposited by pulsed laser ablation to homoepitaxial InP cells. We chose this deposition method because it has been shown to give low resistance contacts to Ill-V materials at low temperatures [5], indicating that contacts to shallow emitter n+/p InP homoepitaxial cells should be able to be formed without deteriorating the characteristics of the junction. There are several other methods potentially suitable for a production environment, notably reactive evaporation [6] and chemical vapour deposition [7]. It should also be noted that the n+/p configuration is at present the most efficient for an InP cell because of the relatively high surface recombination velocity of this material. However, the future application of a window layer such as Ino.flo.&s would favour a p/n structure [8], prohibiting the use of an IT0 contact since the latter is intrinsically n type. The development of p type transparent conductors would clearly be of equal interest.

This system nevertheless has the advantage that there is no potential problem of inter-diffusion on the group Ill sublattice and there is a considerable body of litterature already available. However, recent results '

indicate that the study of other oxides of the same family (such as Galno:, [9] and CdO [IO]) may enable the development of materials sufficiently transparent over a wider spectral range (which would be important for the application of such contacts to tandem systems such as InP/GalnAs) and equally applicable to other materials, in particular GaAs.

EXPERIMENTAL PROCEDURE

Shallow emitter n+/p InP cells with the structure shown in figure 1 were grown by atmospheric pressure

1914

CH3365-4/94/0000-1914 $4.00 0 1994 IEEE First WCPEC; Dec. 5-9, 1994; Hawaii

organometallic vapor phase epitaxy (OMVPE) using trimethyl-indium and phosphine (diluted 50% in hydrogen) as precursors, transported into the reactor by palladium diffused hydrogen. Growth was carried out at 625°C using a V/III ratio of about 50. Di-ethyl zinc and silane (100ppm in hydrogen) were used as p and n type dopants respectively.

InP emitter Si: 3 ~ 1 0 ~ ~ c m ~ 0.03pm

I InP base Zn: I ~ l O ~ ~ c m ~ 1.8pm I

Fig 1. Epitaxial structure of the InP cell.

The back contact to the p type InP substrate of Au/Zn (evaporated from the alloy of composition roughly Auo.sZn0.l by Joule heating) was annealed on a graphite element heated to 375°C for one minute before any further processing. This enables formation of a low resistance back contact while avoiding damage to the emitter (which would be caused by annealing at this temperature after deposition of the front contact). The appropriate alloy phase is distinguished by its rose colour. This procedure has been described previously by other authors [3][11][12].

Subsequently the front contact was processed either by traditional photolithographic techniques, resulting in a Au/Ge metal grid contact to the selectively etched GalnAs contact layer, or by deposition of an IT0 contact by laser ablation directly to the emitter of the InP cell after complete removal of the GalnAs contact layer by etching in 5H2S04:1H202:1H20. For the IT0 deposition, we employed an excimer laser operating at 193nm at a frequency of 5Hz with an energy of up to 350mJ per pulse directed at a powder target of 90°hln203/1 0°hSn02. The epitaxial structures were heated to 250°C under a pressure of 1 OmTorr of oxygen. This allowed deposition of a polycrystilline film of about 50nm of IT0 (in analogy to the recent results of the deposition of IT0 on InP by reactive evaporation obtained by Korobov et al. [SI). This thickness represents a compromise between the conflicting requirements of low optical absorption and low sheet resistance 131. Further details of the deposition process will be published later.

A ZnSIMgF2 anti reflection coating was applied to the traditionally processed cells but no further anti- reflection coating was applied to the IT0 contacts (IT0 has a refractive index of about 2 and hence serves as an anti-reflection coating, but this could be improved by application of a further MgF2 layer [3]).

A comparison of the photoluminescence intensity from separate samples of IT0 on InP (n type) and InP (n type) alone revealed no difference other than

that due to the difference in the refractive index (ie. no passivation of the InP surface but equally no degradation caused by the IT0 deposition).

RESULTS AND DISCUSSION

The spectral response of InP cells from the same growth run processed either with metal grid contact and anti-reflection coating or IT0 contact alone (hereafter cells a and b respectively) are illustrated in figures 2a) and 2b) respectively. Both show a high collection efficiency. The peaks in figure 2b) are interference resulting from the difference in the refractive indices of IT0 and InP.

1.0 1 7 1 , , 1 1 1 1 1 1 1 / / , # I , #

a) cell a with metal grid front contact

0.0 1.0 1.5 2.0 2.5 3.0

1.0 I I I I , I I I I , I I I , I I I , I

b) cell b with IT0 liont contact

0.0 1.0 1.5 2.0 2.5 3.0

enww ( e 9

Fig.2 Spectral response of the best InP cells measured: a) with metal grid front contact and ZnS/MgF2 anti-reflection coating or b) with IT0 front contact and no additional anti-reflection coating.

Figure 3a) and 3b) show the corresponding I-V characteristics of these cells (the former confirmed by measurements made at the Fraunhofer Institut, Freiburg) and their computer fits to the Shockley equation (using standard sym bois):

1915

The best cell with metallic front contact (cell a) had a conversion efficiency of 18.7Oh (AM1.5 at 25°C considering the active cell area only, ie. ignoring shadowing losses, which were around 20°h with the mask used). A reasonable comparison with the existing record for a similar structure of 21.9Oh [I41 can be made by taking 95Oh of the above the active area efficiency to allow for a reasonable shadowing loss of 5% under an optimised grid, giving an estimated value of 17.7Oh for the conversion efficiency for the total cell area.

3 - E -

-1

- - .

-2

- a) cell a with metal grid front contact - VOC = 831mV

Isc 1.04mA FF 820,

-

- active area = 0.0375cm2 q = 18.7% (AM 1.5,25"C)

-

-

' " " ' ' ' I

Fig. 3 Measured I-V characteristics of the best InP cells measured: a) with metal grid front contact and ZnS/MgF2 anti-reflection coating or b) with IT0 front contact and no additional anti-reflection coating. Dotted lines indicate computer fits using the Shockley equation. Note that these are not clearly visible because they are almost exactly superposed with the experimental results.

1 I I I I I I I I I

- b) cell b with IT0 front contact - VOC = 740mV

Isc = 1.28mA

FF=66% -totalarea = 0.0468" q = 13.77.h (AM 1.5,25'C)

z - E -

-1 - - - .

- -

-2. 2 I ' ' c a I '

The best cell processed with an IT0 contact and no additional anti-reflection coating (cell b) demonstrated a conversion efficiency of 13.8% for the total cell area. We estimate that a further percentage point could be added to this result by the addition of an additional MgF2 layer [3].

A detailed comparison of these two cells relies upon the assumption that their characteristics before processing were identical, which was not necessarily true in this case. An initial survey of the two series of processed cells (ie. with metal or IT0 contact) revealed a considerable variation in the characteristics obtained for each set. As a result, we are as yet unable to conclude as to whether the lower Voc of the cell in figure 3b) in comparison with that shown in figure 3a) results from damage inflicted during the deposition of the IT0 or simply a difference in the characteristics which was present before processing. This is an important point, because the motivation for using the laser ablation method is to avoid the damage which can be caused if IT0 is deposited by sputtering. However, an Auger profile confirms that the ITO/lnP interface of a processed cell, illustrated in figure 4, is extremely abrupt. Also, the photocurrent generated by cell b was indeed roughly 20% higher than that of cell a, in accordance with the gain in active surface area.

T-

i 10nm , _

In (MNN)

0 (KLL) -

Fig. 4

Determination of the possible advantage to be gained from the use of an IT0 contact requires an accurate evaluation of the resistive losses and the possible drop in Voc mentioned above.

Hall measurements of IT0 layers deposited on glass substrates under similar conditions revealed a resistivity of 4x1 04Q2cm, comparable to the best results obtained elsewhere [5]. Evaluation of the contact resistance of IT0 layers deposited on InP has been made difficult by the fragility of the resin used to pattern appropriate contacts when using an HF selective etch. However the curve fit to figure 3b) using the Shockley equation reveals an estimated overall resistance of 4.34C2cm2 (of which we calculate that l.lC2cmZ can be attributed to the sheet resistance assuming uniform current generation over the cell surface) in comparison

Auger profile of the ITO/lnP interface.

1916

with 0.39 C2cm2 in the case of cell a. This implies a contact resistance at least two orders of magnitude greater than the values usually achievable [Ill. Three point measurements were used to check that this resistance was not simply a residual resistance due to the conductive resin used to make a contact to the IT0 for measurements. So far the process has not been optimized to minimize the contact resistance and it is possible that a substantial reduction in this value could be achieved. If indeed the overall series resistance can be reduced to about 1 ncm2 and the high Voc of epitaxial cells can be maintained, these cells may achieve high enough efficiencies to be interesting for one sun space applications (though concentrator applications can tolerate correspondingly smaller series resistances as the concentration ratio increases). Of course, the series resistance could be further decreased by using an additional metal grid on the IT0 but this would remove the benefit of the simplification of the process.

Certainly, the possibilities of transparent ohmic contacts in general have not been fully explored and there are obvious advantages to be gained from their use if their electrical and optical properties can be further improved.

CONCLUSION

Homoepitaxial n+/p InP cells with transparent indium tin oxide front contacts deposited by laser ablation or with traditionally processed metal grid front contacts have yielded conversion efficiencies of 13.8% (without additional anti-reflection coating) and approximately 17.7% (with a two layer anti reflection coating) respectively. Further investigation of the effect of the IT0 deposition process on the Voc of the cells and minimisation of the contact resistance at the ITO/lnP interface is required to assess the maximum yield available from such a cell. However, good quality IT0 material has been produced and the initial photovoltaic results are encouraging. The laser ablation process for the deposition of transparent conducting oxides is a good alternative to sputtering.

ACKNOWLEDGEMENTS

The authors would like to thank M.Leroux for photoluminescence measurements, J.Derrien and co- workers (CNRS Luminy) for the Auger profile and S.Kunzelmann (Fraunhofer Institute, Freiburg) for cell calibration. S.T. gratefully acknowledges the financial support of the JOULE programme Commission of the European Communities.

REFERENCES

[I] K.A.Bertness, S.R.Kurtz, D.J.Friedman, A.E.Kibbler, C.Kramer et J.M.Olson, Appl, Phys. Lett. 65, 989 (1994)

[2] M.Yamaguchi, K.Ando, A.Yamamoto and C.Uemura, J. Appl. Phys. 58, 568 (1985)

[3] T.A.Gessert, X.Li, T.J.Coutts, M.W.Wanlass et A.B.Franz a lnP and Related Materials for Advanced Electronic and Optical Devices N, 474 (SPIE Vol. 1144, 1989)

[4] J.K.Luo et H.Thomas, J. Electron. Mater. 22, 1311 (1 993)

[5] J.P.Zheng et H.S.Kwok, Appl. Phys. Lett. 63, 1 (1 993)

[6] V.Korobov, Y.Shapira, B.Ber, K.Faleev and D.Zushinskiy, J. Appl. Phys. 75, 2264 (1994)

[7] T.Maruyama and K.Fukui, J. Appl. Phys. 70, 3848 (1991)

[8] R.K.Jain, LWeinberg et D.J.Flood, Appl. Phys. Lett. 74,2948 (1 993)

[9] R.J.Cava, J.M.Phillips, J.Kwo, G.A.Thomas, R.B.van Dover, S.A.Carter, J.J.Krajewski, W.F.Peck, Jr., J.H.Marshall et D.H.Rapkine, Appl. Phys. Left. 64, 2071 (1 994)

[ IO] W.Gao, A.Sameen-Khan, P.R.Berger, R.G.Hunsperger, G.Zydzik, H.M.O’Bryan, D.Sivco et A.Y.Cho, Appl. Phys. Lett. 65, 1930 (1994)

[I I] T.A.Gessert et T.A.Coutts, ct Optimization and Fabrication of Metal Contacts for Photovoltaic Solar Cells>>, Chapter 13 of a lnP and Related Materials: Processing, Technology and Devices >> (A. Katz ed., Artech House, 1992)

[I21 V.G.Weizer et N.S.Fatemi, Proc. 23rd /€€E PVSC, 779 (1 993)

[I31 C.J.Keavney, V.E.Haven et S.M.Vernon, Proc. 21st I€€€ PVSC, 141 (1990)

1917