Highefficiency indium tin oxide/indium phosphide ...

Highefficiency indium tin oxide/indium phosphide solar cells X. Li, M. W. Wanlass, T. A. Gessert, K. A. Emery, and T. J. Coutts Citation: Appl. Phys. Lett. 54, 2674 (1989); doi: 10.1063/1.101363 View online: http://dx.doi.org/10.1063/1.101363 View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v54/i26 Published by the American Institute of Physics.

Additional information on Appl. Phys. Lett. Journal Homepage: http://apl.aip.org/ Journal Information: http://apl.aip.org/about/about_the_journal Top downloads: http://apl.aip.org/features/most_downloaded Information for Authors: http://apl.aip.org/authors

Downloaded 06 May 2013 to 192.174.37.50. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions

Highaefficiency indium Un oxJdelindium phosphide solar cens x. Li, M. W. Wanlass, T. A. Gessert, K. A. Emery, and T. J. Coutts Solar Energy Research In.l'titUle, 1617 Cole Bouleuard, Golden, Colorado 80401

(Received 3 January 1989; accepted for publication 1 May 1(89) Improvements in the performance of indium tin oxide/indium phosphide (ITO/lnP) solar cells have been achieved by using dc magnetron sputter deposited n-ITO onto an epitaxial p/p structure grown on good quality commercial p" hulk substrates. The composition of the sputtering gas has been investigated and the highest efficiency cells resulted when the surface of the epilayer was exposed to an Ar/R, plasma before depositing the bulk of the ITO in a more typical Ar/02 plasma. With He processing, record efficiencies of 18.9(10 global, 1000 W m J, 25°C (17.0% air mass zero) \vere achieved. Without III processing, the devices exhibited lower efficiencies and were unstabk. Type conversion of the IuP was shown to occur and was established as being associated with the ITO (possibly due to Sn donors) rather than sputter damage. These improvements in performance have resulted from the optimization of the doping, thickness, transport, and surface properties of the p-type base, as well as from better control over the ITO deposition procedure. i

High efficiencies I and excelleni radiation resistance 2 have recently been demonstrated for both InP shallow homojunction and indium tin oxide (ITO)/lnP cells.'.4 Although the former devices are well understood, there are still many uncertainties regarding the ITO/InP cells. Previously, ITO/InP cells have been fabricated simply by sputter depositing ITO onto bulk liquid-encapsulated Czochralski (LEe) substrates of the appropriate doping concentration. 5 The properties of the bulk substrates are inferior to those of epitaxial layers due to, for example, the effects of polishing and etching. Because oftheir thickness (about 400pm) they also contribute significantly to the series resistance of the device. The role of sputter damage is also an open question but has been discussed at length in the literature. 6 It is also possible that thermal damage to the surface of the substrate can occur during contact sintering 7 However, as will be shown, the damaging effects of thermal treatment arc evidently less than the benefits resulting from the elimination of adverse surface properties. Hence, by using good quality cpilayers, it has been possible to overcome these artifacts and produce devices with low series resistance and excellent conversion effkiency. Additionally, it has thus been possible to investigate, more directly, the elfects of any sputter damage (i.e., due to bombardment hy charged or neutral particles) and the influence ofthe sputtt.,·ing gas. It will he shown that sputter damage is insignific;;wt, ..mder the fabrication conditions used, although the composition of the sputtering gas has a very important effect. The devices are fabricated by growing a p i buffer layer on ap f -InP LEC substrate, using atmospheric pressure metalorganic vapor phasc epitaxy (APMOVPE). This is followed by the growth of the p-lnP base layer, which is 4 flm thick mid has an impurity concentration of (1-2) X ro'l em ". Full details of the substrates used and the epitaxial growth are given elsewhcfc. x Following growth of the base layer, a low resistance ohmic contact is formed Oil the back surface of the wafer. This is achieved using a thermally evaporated Au:Be (I wt. % Be) film of 150 nm in thickness which is sintercd in forming gas for 2 min at 390 0c. A high 2674

Appl. Phys. Lett. 54 (26), 26 June i 9a9

conductivity film of electroplated Au is then applied after etching thc Au:Be contact to remove any oxide. The ITO is deposited by de magnetron sputtering without further treatment of the epitaxial wafer. The main variable investigated was the composition of the sputtering gas and both argon/oxygen (0 cells) and argon/hydrogen (H cells) were used. Again, details are given in Ref. 8. Total ITO thickness is typically 55 urn and, for the H cells, only the initial 5 urn of ITO is deposited in the hydrogen-rich conclition, The remainder of the fUm is deposited under the usual oxygen-rich conditions to promote electrical conductivity and optical transmission. Following ITO deposition, the device substrates are processed into arrays of electrically isolated mesa solar cells with the cell areas and grid contacts defined using photolithography. The mesas have an area of 0.108 em" and the grid obscuration is ~ 5%. Grid contacts to the ITO are formed by electroplating approximately 3 pm of pure Au to the ITO after roughening the surface of the latter by etching for 5 s in it solution of 1 HF:l000 H 2 0 (by volume) to enhance contact adhesion. At this point, the performance of the cells is measured although the better cclls have an additional1ayer of MgFl applied to form a double-layer antireflection coating with the ITO. The MgF 2 is deposited using electron beam evaporation and its thickness is typically 76 nm. The performance characteristics of the highest efficiency ITO/hlP solar cdl made to datc are shown in Fig. 1. The three CUl-ves correspond to standard global, without and with a MgF2 coating, and air mass zero (AMO) reporting conditions.'l At 18.9% global and 17.0% AMO conversion efficiencies, this is the highest performance lTO/InP cell made, the highest performance for any cell incorporating an ITO layer and the highest by far for any cell having the junction fabricated using sputtering. In Fig. 2 the quantum efficiency characteristics of this cell are shown. The device exhibits an extremely high quantum efficiency at long wavelengths, indicating the excellent optoelectronic properties of the base, and is high even at the short-wavelength end

0003-6951/89/262674-03$01 .12 It is also known that hydrogen implantation in GaAt> passivates electrically active impurities and deep lying recombination centers. 12 This may be partially responsible for the improvement in performance observed in the present work. However, the major improvement has undoubtedly resulted from the use of the high quality pip"" epilayer structures, and the use of the much improved LEe substrates. The high efficiencies achieved for ITO/tnP cells have prompted much speculation that these devices are actually homojunctions with type conversion of the p-type base being caused by anyone of several possible mechanisms,6 The compensation of Zn acceptors by Sn donors has been suggested as a probable mechanism and several peripheral experiments were performed to test this hypothesis. Firstly, a semi-insulating F'e-doped InP substrate was subjected to the usual deposition of a 5 nrn hydrogen-rich interface layer of ITO. The initial mobility and the carrier concentration of the substrate were .~ 850 cm 2 V l S i and 109 cm 3,p type, respectively. After deposition of the ITO, the mobility and carrier concentration were 10 cm 2 V i S I and 5 X 10 18 em .\ n type, respectively (assuming an ITO thickness of 5 11m). The ITO interface layer was then removed by chemical etching and the total removal of the layer was confirmed ellipsometricallyo Hall measurements showed that the surLi et al.

2675

'.' '.' • '.' • '.' ............... '.7' ••


TABLE I. Summary of perfonnance data fodnP cells after removal of 5 nm onTO (and AI) thin films. Data illustrate that ITO is the principal cause of type conversion.

V)(' Cell type

Comments

(mV)

Hcells

50 nm ITO deposited in Ar/R2' then removed ill HF/H 2 O. A verage of 8 cells o cells 50 11m ITO deposited in ArlO" thell removed in HF/H 2O. Average of 6 cells Ar ollly 50 nm ITO deposited in Ar only (new target Uf;ed) , then removed ill HFIH 2 O. Average of 3 cells Al only 50 nm Al deposited in Ar only, then removed in Hl?/H 2 O. Average of 6 cells

face layer had become strongly 11 type, and it was established that type conversion became stronger with increasing added hydrogen. Recent measurements of spreading resistance indicate that the thickness of the type converted region is -100 nm which indicates that the canier concentration in this layer is _10 17 cm ',depending on the precise ITO deposition conditions (Leo, the partial pressure of H2 and the deposition rate) This experiment was repeated using oxygen-rich, hydrogen-rich interface layers and without an interface layer (using only AI' as the sputtering gas)o In these experiments, bulk InP substrates were used with NrN[) of -3X lOt5 cm 3 In all three cases, the 5 nm layer of ITO was removed after deposition by etching in dilute HF/H 20 (1: 1000 by volume) and a Au grid was then electroplated onto the surfaceo Despite the absence of any ITO, in aU three cases a significant photo voltaic response was observed. The a cells and the H cells had efficiencies in the range 7-9% whereas the Ar cells only had an efficiency of about 6%. However, the ITO in the latter case was sputtered from a fresh target which had not been exposed previously to either oxygen or hydrogen. The quality of the ITO is therefore dubious. Hence, after deposition, the role of the ITO appears to be simply to act as a low-resistance transparent contact. These experiments confirm that the type conversion mechanism is due to the ITO although it is not yet clear whether because of compensation by Sn, or to plasma damage, or to some other mechanism. To investigate this, a film of Al was sputter deposited onto similar bulkp-InP substrates to a thickness of 5 nm, at the same rate as the ITO, using only argono After removal of the AI, considerable difficulty was experienced in plating the grid, as might be expected for a highly resistive surface layer. These devices gave a measurable photovoltaic response but with an efficiency of only 0.2% and a of only 293 m V. All the data from the above experiments are summarized in Table 1. The experiment was then repeated using a semi-insulating substrate and, after removal of the AI, the surface was not found to have become n typeo Hence, it seems probable that the cells treated with the ultrathin Al were simply Schottky barrier devices with a potential barrier formed between the Au grid and the p-InP, and with carrier 0

0

r:,c

2676

Appl. Phys. Lett., Vol. 54, No. 26, 26 June 1989

769 ±6 773 ±5 767 t 1 293 ± 14

J,,, (rnA cm-') 18,00

±1.1O 19.36 ±O.16 14.51 j 0.22 0.9 ±O.O9

Fill factor

Efficiency

(%)

(%)

56.6 ±O,g 55.3 ±I.l 52.3 0.1 52.S ± 2.8

7.85 ±0.60 8.28 ±0,27 5.83 ± 0.12 0.15 :1.. 0.05

collection only from the very ncar grid lines. Next, the experiment was repeated using Ar/02 and Ar/H2 as the sputtering gases. After removal of the Al the first of these substrates indicated a very weak increase in lz-type conductivity, whereas the second showed a significant increase in electron concentration, although not as strong as when ITO is used. Hence, it is concluded that the type conversion is mainly dependent on the ITO although it cannot be said with certainty that it is due to Sn compensationo Further experiments are presently under way using In 2 (}, to determine whether the Sn is essential. It has also been established that the sputtering gas influences the type conversion with hydrogen enhancing the effecto The latter observation, together with results on GaAs, suggests that hydrogen has a similar passivating role in InP as it does in other semiconductorso Further improvements in the performance of these cells may be possible by optimizing the growth of the H 2 -rich layer and by additional developments of the grids and AR coatingso This work was supported by the Department of Energy under contract No. DE-AC02-83CHlO093 and by the NASA Lewis Research Center under Interagency agreement Noo C-30005-K.

Ie. J. Keavney and M. B. Spitzer, AppL Phys. Lett. 52, 1439 (19R8). 'M. Yamaguchi, A. Yamamoto, Y. hoh, and C. Uemma. Proceedings 0/ the 2nd Infcmalional Photo1!o/taic Science and Engineering Conference, Beijing, PRe. (Adlle1d Advertising Co., Hong Kong, 1986), p. 573. 'M. W. Wanlass, To A. Gessert, K.A. Emery, and T. Jo Coutts, Proc. 20th IEEE Photovoltaic Specialists Conference, Las Vegas 1988. .,1. Weinberg, C. K. Swartz, R. E. Hart, and T. J. Coutts, Proc. 20th IEEE Photovoltaic Specialists Conference, Las Vegas 1988. 'T. J. Coutts and S. Naseem, App!. Phys. Lett. 46, 164 (1985). "1'. I. Coutts and M. Yamaguchi, "InP Based Solar Cells: A Critical Review of Their l'abrication, Performance and Operation," Current Topics in Photovo[taics, edited by T. J. Coutts and J. D. Meakin (Academic, London, 1958), VoL 3, p. 79. 7R, F. C. Farrow, J. Phys. D 8, L87 (1975). HM. w. Wanlass, T. A. Gessert, K. A. Emery, and To J. Coutts, NASA Conf. on Space Photovoltaic Research and Technology, NASA Lewis Research Center, Cleveland, OH 198R, p. 430 OK. A. Emery and C. R. Osterwald, Solar Cells 17,253 (1986). "'N. M. Johnson, Phys. Rev. B 31,5525 (1985). "G. Yaron, J. Levy, Y. Goldstein, and A. Many, J. App!. Phys. 59, 1232 ( 1986). lew. C. Duutremont-Smith, Mater. Res. Soc. Symp. Proc. 10,4 (1988).

Li et al.

2676
