Electrodeposition of CuInSe2 thin films from aqueous solution

Electrodeposition of CuInSe2 thin films from aqueous solution

Solar Energy Materials 18 (1989) 385-397 North-Holland, Amsterdam 385 ELECTRODEPOSITION OF CulnSe z THIN FILMS FROM AQUEOUS SOLUTION S.N. SAHU, R.D...

927KB Sizes 0 Downloads 12 Views

Solar Energy Materials 18 (1989) 385-397 North-Holland, Amsterdam

385

ELECTRODEPOSITION OF CulnSe z THIN FILMS FROM AQUEOUS SOLUTION S.N. SAHU, R.D.L. KRISTENSEN and D. HANEMAN Department of Condensed Matter Physics, University of New South Wales, P.O. Box 1, Kensington, Australia 2033 Received 10 February 1988; in final form 12 November 1988 CuInSe 2 thin films were cathodically electrodeposited on conducting substrates from aqueous solutions containing CuCI, InCl 3 and SeO2. Structural characterizations were carried out by microprobe, X-ray diffraction and electron microscopy studies. The presence of chalcopyrite phase CulnSe2 was confirmed from X-ray studies. Optical absorption studies indicated band gap values of about 1.1 eV. Electrical characterization was carried out by Hall effect and resistivity studies. The room temperature resistivity and mobility were found to be 2.15 × 10 -3 [~ cm and 8.1 cm2 V - ! s-1 respectively for p-type films. Diffusion of In into p-type films converted them to n-type. Photovoltaic and photoelectrochemical solar cells were fabricated with Mo/p-CulnSe2/ CdS/Au and Mo/n-CulnSe2/I i- I ~ / C configurations. The open circuit photovoltage and short circuit current densities were 188 mV and 0.056 mA cm -2 for photovoltaic cells and 172 mV and 2.7'5 mA cm -2 for photoelectrochemical cells under 100 mW cm -2 intensity of illumination, without optimisation.

1. Introduction In recent years, CuInSe2 has emerged as a promising semiconductor for photovoltaic applications because of its excellent photoactive properties such as high_ absorption coefficient and good durability. Energy conversion efficiencies exceeding 12~ have already been exhibited with photovoltaic (PV) cells [1,2] as well as single crystal photoe!ectrochemical (PEC) cells [3]. For a recent review see ref. [4]. However, for large area solar cell use, (polycrystanine) thin films are necessary to minimise wastage of material. The cost of thin film solar cells primarily depends on the preparatory methods and processing. Various methods have been used, e.g. vacuum evaporation [5-9], spray pyrolysis [10] and other techniques [4]. Although vacuum coevaporation has achieved success for obtaining high efficiency, scale-up is still a problem. Electrodeposition appears in principle to be economically attractive. It has already been attempted at room temperature from aqueous solution for binary [11-17] and ternary [18-22] semiconducting compounds and from non-aqueous baths [23]. Reasonable success has been attainted with binary compound deposition but ternary compound deposition still suffers from many problems associated with the control of solution composition, deposition potentials and pH. Earlier ternary semiconductor electrodeposition [18-22] involved plating of Cu and In either 0165-163,3/89/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

$.N. Sahu et al. / Electrodeposition of CulnSe z thin films

386

separately from two different baths or simultaneously from a single bath, then sulfurization/selenization (for CuInS2 or CuInSe2) in an atmosphere of H2S or H2Se with Ar or N2 as a medium. This process, though successful, requires special safety precautions for the highly poisonous gases. Success has been obtained in elecrodepositing Cu, In and Se simultaneously in a non-aqueous bath [23] but without stoichiometric composition. Recently [24] stoichiometric CuInSe2 films were obtained by pulse plating in aqueous solution but no solar cell performance was reported. Here we repo~'t successful electrodeposition of stoichiometric and reproducible CulnSeq films from an aqueous solution containing CaCI, InCl3 and SeO2, leading to photoelectrochernical and photovoltaic performance.

2. Experimental

2.1. Film preparation The electrodeposition technique involves a simple electrolysis cell consisting of a cathode, an anode, an electrolyte containing soluble species and an electrical circuit. The deposition of ions on the cathode depends upon the electrode potential, pH. concentration of individual ions, temperature and agitation of the electrolyte. The steps involved in the deposition process are (i) transport of ions to the electrode surface, (ii) desolvation, (iii) charge transfer, (iv) surface diffusion, (v) integration into the substrate lattice. The main difficulty associated with the deposition process is the individual deposition potential of the ions. If the individual potentials are quite different, then for s~raultaneous deposition, one must b~ng them to a common potential. A constant current or voltage should be applied for linear growth of the deposit. The individual deposition equations of Cu, In and Se for CuInSe2 deposition are given as: Cu(s) ~ Cu 2+ + 2e-ffi E ° + ( R T / 2 F ) ln(acn2÷/acu) = 0.10rsc E + 0.0295 !og(acu~./oeu),

(1)

In(s)*~In3+ + 3 e - - - E ° + (RT/3F)ln(aln3+/asn) --" - 0.58Vsc E + 0.0197 log(a,n3+/a,a), Se(s) + 3 8 2 0 ** a2SeO 3 + 4 8 + + 4e** HSeO~ + 4H + + =

(2)

(0 < p a < 5)

4e- + OH-

E ° + ( R T / 4 F ) ln(aaseo~/ase) + (3RT/4F) log CH+

= 0.50F~cF +0.0148 log(anseo:/ase ) - 0.0443pH.

(3)

Here the quantities £0 are the equilibrium electrode potentials expressed with reference to saturated caiomel electrode, a are activities of ions in the bulk solution

$.N. Sahu et al. / Electrodeposition of CulnSe 2 thin films

387

and in the deposit (the activity of an ion in the solid deposit is unity) and CH+ is the concentration of hydrogen ions. The quantities Cu(s), :~n(s) and Se(s) are the respective species in the solid deposit. From the above equations we note that the equilibrium electrode potentials for Cu, Se and In are far apart. For Cu and Se the potentials are positive whereas that of In is negative. It would be impossible to deposit the three elements simultaneously unless the individual potentials are brought closer to each other. Further, the potentials are also dependent on the pH of the electrolyte. The method used is to adjust the concentrations of individual ions and the pH of the electrolyte. As Se and Cu have a large positive potential, their activities should be very low whereas that of In should be large as it has a large negative potential. Although the potentials for the 3 elements are made coincident, the deposition rates nevertheless depend on the degree of the nobility of the ions. Thus Se, being most noble, should deposit first. Hence, we expect that the first few layers should be Se-rich. As Cu and Se concentrations are very low, during the deposition these species are likely to be depleted from the cathode-electrolyte interface and the deposition would be a diffusion limited process. Hence, stirring could improve the quality of the deposit. Based on the above arguments and experimentation we have used the electrolyte concentrations: CuCI = 0.37mM, I n C l 3 - 5.27mM and SeO2 = 0.9mM in distilled water. The chemicals used are of A-R grade. The electrolysis cell consists of molybdenum sheet as cathode, carbon as an anode and a useful electrical circuit as described in ref. [12]. The Mo cathode was first electropolished with 140:35 = H20:H204 by volume, washed with distilled water, ultrasonically cleaned with acetone and finally washed with distilled water jets and dried. The carbon anode whose area is larger than the cathode was, after each experiment, polished with fine emery paper, washed with distilled water, ultrasonically Cleaned with acetone and finally dried. Reproducible and stoichiometric CuInSe2 films were obtained at a constant current density of 6 mA cm -2 for 15 rain at a temperature of 23°C and at pH = 1.0. Annealing experiments were carried out both in vacuum and air at various temperatures. 2.2, Characterization

Compositional analysis and structural characterizations were carried out~by electron microprobe (EDAX) analysis (JEOL JXA-840) and a Philips X-ray powder diffractometer. The diffraction angles (20) for X-ray peak intensities were identified and the calculated d values (lattice spacings) were compared with standard d values. For microprobe as well as X-ray diffraction studies, the films deposited on Mo substrates were used directly. Optical absorption spectra were recorded with a Cary-17 spectrophotometer in order to estimate the band gaps of the CuInSe2 films. For this purpose the films were electro-deposited on transparent conducting glass. Thickness (1.2 /~m) was measured ~ t h a Sloan Dektak II profilometer. Electrical chalactefiz.ations were carried out by resistivity and Hall effect studies on films electrodeposited on transparent conducting glass, the resistance of which

388

S.N. Sahu et al. / Electrodeposition of CulnSe z thin films

was much greater than that of the film. Initially, the films were sub iected to hot-probe tests to determine whether they were n- or p-type. The as-deposited and annealed films (annealed in vacuum at 350°C for 2 h) were found to be p-type. Howe-;er, some of our interests in these films are for use in photoelectrochemical cells. Since good results for films [25] and polycrystalline s~rnples [26] have been obtained in iodine based electrolytes which require n-type specimens, some films were converted to n-type by In diffusion. For this purpose 0.2 /~m In was electroplated on p-CuInSe2 from an In2(SO4)3 bath and annealed in vacuum (10 -5 Torr) for 22 h at 350 ° C. Photoelectrochermcal cells with various configurations were fabricated. For photovoltaic heterojunctions, CdS ( ~ 1 /~m) and Au were deposited by thermal evapora~on. The polysulfide (S 2-, S~-) electrolytes for photoelectrochemical cells were prepared with 1M Na 2S, 1M S and 1M NaOH in distilled water. The counterelectrodes were graphite.

3. Results and discu~ion

As described earlier, the Se concentrations being low and the element more noble, it should deposit first, followed by Cu and then In. Visual observation showed that the first few layers of the deposit are brown. Then the colour changes to brown plus blue. Finally, the deposit becomes black and the individual colours are not distinguishable. This sequence is verified in fig. 1 which shows the voltammogram obtained during the deposition of C h l n ~ 2 from an aqueous chloride solution. A cathodic current began to flow at about +0.3 V and the peak at +0.15 V corresponds to the deposition of ~e (brown). The next peak ( - 0 . 2 V) corresponds to the deposition of Cu (blue) and the last to deposit was In (about -0.35 V). The current began to increase again at about - 0 . 6 V with the evolution of H2. It would not be surprising that initially Se and Cu may react to give Cu2_xSe in the deposit in the first few layers. Then with simultaneous deposition of Cu, In and Se, the formatio: of CuInSe2 is likely. It is expected that the as-deposited films might be spongy and non-stoichiometric. However, annealing should improve the quality and stoichiometry of the films. Electrodeposition was carried out for 15 min (1.2/zm thicl~ess) by varying the solution composition and electrolysis c-rrent density, keeping pH = 1.0 and the stirring rate ~ d temperature constant. At higher electrolysis current density (20-100 mA cm-2), vigorous H2 evolution was witnessed and no proper deposition could be obtained. In all the cases pinholes were seen in the deposit. Deposition at lower current density (0.5-1.0 mA cm -2) took a much longer time to obtain a reasonably thick film. Further, the colour of the deposits obtained at two extreme conditions were very different from the deposit obtained at 6 mA cm-2 which was black. The deposits at higher current density looked white and blue whereas at lower curren~ density they were brown. However, deposition of films at these extreme conditions was d~scarded as they were not stoichiometric. The bath composition was varied by changing the individual ion concentration while keeping fixed the current density (6.0 mA c m - 2 about - 0 . 4 VSCE), electrolysis temperature (23 ° C), stirring rate, and

$.N. Sahu et al. / Electrodeposition of CulnSe z thin films

389

I (rnAlcm2)

t

HSeO2* Cu +

t

[n3. H2

+0./+

+0.2

0 -0.2 V (SCE) {volts)

-0.t,

"0.6

Fig. 1. Cyclic voltammogram for the deposition of CulnSe2 from aqueous solution (0.73mM CuCI, 5.27mM InCI 3, 0.9mM SeO2).

p H - 1.0. E D A X analysis was carried out for the deposits obtained at different solution composition and the results are shown in table 1. From this, it is clear that the as-deposited films do not show the correct stoichiometry. Annealing at a moderate temperature of 150 ° C decreases the Se content and increases the Cu and In content. Annealing at a higher temperature of 350 ° C for 2 h in vacuum (lo -5

Table 1 EDAX results for atomic composition of deposits from different solution compositions Film No. • 1

2 3 4 5

Solution composition (mM)

Atomic composition (~)

CuCI

InC! 3

SeO2

Cu

In

Se

0.73 0.73

5.27 5.27

0.9 0.9

21.9 22.7

22.1 22.7

56.0 54.0

0.73

5.27

0.9

25.2

23.2

51.5

0.73 0.73 0.98 0.73

5.27 5.50 6.78 6.03

1.2 1.0 0.9 1.2

22.5 21.4 24.5 11.2

14.1 21.4 44.2 67.4

63.4 57.1 31.3 21.3

Remarks As-deposited Annealed in air at 150 °C for 2 h Annealed in vacuum at 350°C for 2 h As-deposited As-deposited As-deposited As-deposited

390

$.N. Sahu et al. / Electrodeposition of CulnSe2 thin films

mfl

IH

gll~

~

~oI'I

tnllo

~

--,,4'11 ~ II

";dll m

01

=

_Nm

..~

~

01

01®

N

®

oE

~o

"-"

2O

b

C N

e¢q

¢q .....

rq

r~

Q

v s

~

O

Q

.~

O

O

¢.3

--

--IE ~ g ~"°

•%

£a L I

L L . . . . . . ,,. . . . . . . . . . . . .

I '.~;'i-i--I-i-:: s7 }is3

II

sl ii

~4

N e-

¢.3 O

ILX..._~_~,-..... ,~_ I i I Ii:1-'i--;|-1"--i|:

,

II

II

42 ii

II

3s

II

i~'~l

33ii 2e ,ii23 21ii,? ,6

20 Fig. 2. X-ray diffraction patterns for C u l n S e 2 filnm. (a) A n n e a l e d at 350 o C for two hours in v a c u u m ; (b) In-diffused at 350 o C for 22 hours in v a c u u m .

Torr), produces a near stoichiometric CuInSe2 film. It is possible that if free Se is deposited, then during annealing it reacts with Cu and In giving CuInSe2. For subsequent studies we have used the annealed (350 o C) films only. The as-deposited as well as the annealed films were always p-type due to the presence of excess Se or Cu. Conversion to n-type was achieved by IrL diffusion as described earlier. X-ray diffraction studies were carried out for the annealed and In-diffused films and the results are shown in fig. 2. The presence of tetragonal chalcopyrite phase CuInSe2 in (112) planes is clearly indicated at a diffraction an~e 20 = 26.7 ° for both annealed and In-diffused films. However, along with the chalcopyrite phase, sphalerite phase CuInSe2 is also present. In fact most of the other techniques of CuInSe2 thin film preparation do not show the presence of a single chalcopyrite phase. Besides the presence of CuInSe2, some extraneous compounds such as CuSe,

S.N. Sahu et al. / Electrodeposition of CulnSe 2 thin films

391

CuSe2 and In2Se3 are also indicated. After In diffusion, some extra peaks were identified which is expected because the unreacted compounds after self diffusion for a longer time may react with each other and give additional compounds such as CaSe 2. The as-deposited films do not show any X-ray peaks, but after annealing, the crystallize size h-~creases and X-ray peaks could be identified. The peaks for the In-diffused films are sharper than for the annealed films which can be attributed to grain growth due to the extra annealing. The presence of CuSe and CuSe2 in the electrodeposit is not surprising when one considers that the individual deposition potentials of Se and Cu are close to each other. Presumably, some Se and then Cu may deposit before much In deposits on the film. The mechanism of the compound formation can be given as CuCI =* Cu + + CI-, InCl3 =* In3 + + 3CI-, S e O 2 + H 2 0 =~

HSeO~ + O H - .

In the initial stages of the deposition Cu + + HSeO~ + 3H + + 5 e - = , CuSe + 2H20,

(3.1a)

Cu + + 2HSeO~ + 6H + + 9 e - = CuSe2 + 4 H 2 0 .

(3.1b)

or

At a later stage when In starts depositing, the possible reaction is: 2In3+ + 3HSeO~ + 9H + + 18e- ~* In2Se 3 + 6 H 2 0 ,

(3.2a)

In 3+ + HSeO~ + 3H + + 7 e - = , InSe + 2H20.

(3.2b)

or

When the three constituent elements start depositing simultaneously, the possible reaction is: Cu + 4- In3+ + 2HSeO~ + 6H + + 12e- =* CulnSe2 + 4H20.

(3.3)

There is a strong possibility of the presence of HSeO~ in an acidic aqueous electrolyte (pH < 5) containing SeO2 [12] in analogy with the HTeO~ species contained i~. an acidic TeO2 [15,27] solution. The HSeO~ species transfers its charge after adsorption [27]. As a rough verification of the proposed mechanism, i.e. eq. (3.3), a coulombic calculation was carded out taking a typical film deposited at the optimized conditions with an area of --- 1.38 cm2, deposited at 6.05 mA for 10 rain, i.e. 6.05 X 10 -3 A cm -2 × 1.375 c m 2 × 10 rain × 60 s rain -~ =4.99 C. The deposit was weighed as 1.2 nag ( + 10%). Assuming the deposit was entirely comprised of CuInSe2 (molecular weight 336.29 g mol-l), this amount corresponds to --3.6 × 10 -6 mol. Hence the deposition took place at a rate of 1.4 × 106 C tool -1 or -- 15 F (:t: 10%) for 1 mol. The proposed equation, (3.3), requires a 12 electron transfer for the cathodic deposition of CuInSe2; hence the two results are in

392

S. N. Sahu et ul. / Electrodeposition of CuInSez thin films

a fair agreement, taking into consideration the error in weighing and possil >le

hY&bogenevolution (which was visible) which would account for the difference in the1number of faradays.

Fig0 3. SEM mkrographs of CuIxrSe~ films: (a) as-deposited; (b) annealed; (c) In-diffused. The glob ules iI=(c) are balls of indium that have aggregated from the plated layer. _

393

$.N. Sahu et al. / Electrodeposition of CulnSez thin films

Fig. 3 (czntinued). Sce~ming electron microscopic (SEM) studies were carried out for the a s deposited, annealed and In-diffused films. Some micrographs are shown in figs. 3a-3c, They show a characteristic undulatory topography.

i

I=--"-'~I

[

I

I

I

I

I

I

I

I

I

I

'

,=

i

~

I

~

,~

i

, -i

,

i

0.6

0.5

AS

DE.POSITED

j,'

d UJ U

0.4

z

IZ

/ / { /

ANNEALED

0.3

ou l

m
0.1

O__L...~

1350

I

I

I. I 1:~50

i

_k i

I

i

1150

,

I= I

I

,

~

1050

WAVELENGTH

i

,

I

n

i

i

950 (r~m)

Fig. 4. Opticalabs~mt%n spectrafor the electrodeposiwxifilms.

i

I 850

n ~

S.N. Sahu et al. / Electrodeposition of CulnSez thin films

394

Annealed x ~

18 -

As deposited

16

14 12 d N

10

> ev

8

0 i .,~l 0.9 1.0

vl 1.1

I 1.2

I 1.3

I 1-4

I 1,5

hv(eV)

Fig. 5. (ah~,) 2 versus hJ, plot for CulnSe2 film where a is the absorption coefficient in arbitrary units and hi, is the photon energy.

Optical absorption spectra of the as-deposited and the annealed films were recorded at 300 K and are sho~n in fig. 4. Because the surface was not smooth it was difficult to identify interference peaks in the transmission spectra and hence carry out proper analysis for the absorption coefficient and refractive index. Hence the data was analysed using simple analysis without considering multiple reflections and the usual plot for direct band gap semiconductors is shown in fig. 5. This indicates a somewhat larger band gap than the usual 1.0 eV for thin films [28]. This may be due to the inadequate analysis or, if real, suggests the possibility of an increase in band gap due to a quantum size effect if the individual grain sizes are as small as about 5 nm. Resistivity studies were carried out for the annealed (p-type) and In-diffused (n-type) films in the liquid N 2 temperature range and are shown in figs. 6 and 7 respectively. It may be noted that the resistivity initially changes linearly with T-1 indicating the absence of ionised impurity scattering [29]. However, towards 77 K the resistivity remains covstant indicating freezing of carriers. The room temperature resistivity (1.11 × 10 -4 ~ cm) for n-type film~ s a s found to be lower than that (2.14 × 10 -3 0 cm) for p-type films. This could be attributed in part to an increase

395

$.N. Sahu et al. / Electrodeposition of CulnSe z thin films

6 c= p- 1.8

>

i

U~ 14,1 IZ:

1.'/ 1.6--

1.5 3.0

I

I

I

I

I

~

4.0

5.0

6,0

7,0

8.0

9.0

I_ 10.0

I_

I

I

11.0 12.0

13.0

10311• (K -1)

Fig. 6. Electrical resistivityversus inverse of temperature for p-type annealed CuInSe2 film.

in grain size after the long annealing treatment for In-diffused films. Hall measurements were carded out at room temperature, giving negative Hall voltages for n-type and positive for p-type films. The carrier concentration and mobilities have

1.20 1-10

,~ 1.00 E u C~

0.90 0.80 0.70

-

0.60

-

OC

0.50 I

3.0

4.0

,[

5.0

,

I

6.0

I

I

I

7.0 8.0 9.0 103/T (K "1 )

I

10.0

i

I

J

11.0 12.0 13.0

Fig. 7. Electricalresistivity for the In-diffuseACuInSe2 film.

S.N. Sahu et al. / Electrodeposition of CulnSez thinfilms

396

Table 2 Electrical properties ar,d photoactivity of annealed and In-diffused Film No.

1

Type of film

p

Carrier density (cm -3)

Mobility a) (cm2 V - 1 s - 1)

CulnSe

2

Photoactivity Dark

films

Light (100 mW cm -2)

Vo~

I~

Vo~

I~

(mV)

(mA ¢m -2)

(mV)

(mA cm- 2) 0.056 b) 2.75 c)

3.63 × 102o

8.08

2.9 b)

0.0 b)

188 b)

2.36 × 1021

23.76

9.0 c)

0.1 c)

172 c)

(annealed) 2

n (In diffused)

a) At 298 K. b) With p-CulnSe2/CdS junction, photovoltaic cell. c) With polysulfide electrolyte, PEC cell.

been calculated as/tp - 8.08 ~ 2 V-X s-X, p __ 3.63 x 102o cm -3, ~n " - 23.76 cm2 V -1 s -1 and n = 2.36 x 10 21 c m -3. Photovoltaic (PV) and photoelectre~hemical (PEC) solar cells were fabricated as described earlier and the open circuit photovoltage and short circuit current densities were measured under 100 mW c m - 2 intensity of illumination from a W halogen light source. The results are given in table 2 along with other electrical parameters. These values of PV performance are to be regarded as preliminary as the various processes of surface treatment have not been optimised. The best PEC performance is obtained using polyiodide electrolytes and stable performance has been reported [26] for thin CuInSe2 films made by coevaporation [9]. However, such films needed to be 4 pm thick as some 2 /tin thickness is chemically converted by the electrolyte. Since the electrodeposited films could not be made sufficiently thick they degraded quickly in iodide solution due to conversion of the entire film. Hence results are quoted only for polysulfide electrolytes. PEC performance is possible in iodide solution if the surface is protected by an appropriate layer and such work is in progress. At this stage the PEC activity can be regarded as a promising indicator of performance potential for electrodeposited films.

Acknowledgements This work was supported by the Australian Research Grants Scheme. S.N.S. acknowledges support of a Welch Foundation Scholarship.

References [1] W.E. Devaney, R.A. Mickelsen and W.S. Chen, Proc. 18th IEEE Photovoltaic Specialists Conf. (1985) p. 1733.

S.N. Sahu et al. / Electrodeposition of CulnSe 2 thin films

397

[2] R.A. Mickelsen, W.S. Chen, Y.R. Hsiao and V.E. Lowe, IEEE Trans. Electron De~ices 31 (1984) 542.

[31 S. Menezes, Appl. Phys. Letters 45 (1984) 148. [4] D. Haneman, CRC Crit. Rev. Solid State Mater. Sci. 15 (1988), in press. [51 R.W. Birkmire, L.C. DiNetta, P.G. Lasswell, J.D. Meakin and J.E. Phillips, Solar Cells 16 (1986) 419.

[61 N.G. Dhere, M. Lourenco, R. Dhere and L.L. Kazmerski, Solar Cells 16 (1986) 369. [7] J. Piekaszewski, J.J. Loferski, R. Beaulieu, J. Beall, B. Roessler and J. Schewchun, Solar Energy Mater. 2 (~q80) 363.

[81 N. Romeo, V. Canevari, G. Sberveglieri, A. Bosio and L. Zanotti, Solar Cells 16 (1986) 155. [91 J. Szot and D. Haneman, Solar Energy Mater. 11 (1984) 289. [10] C.R. Abemathy, C.W. Bates, Jr., A.A. Anani, B. Baba and G. Smestad, Appl. Phys. Letters 45 (1984) 890. K.T.L. de Silva, D.J. Miller and D. Haneman, Solar Energy Mater. 4 (1981) 233. S. Ch~,ndra and S.N. Sahu, J. Phys. D. (Appl. Phys.) 17 (1984) 2115. S. Chandra, O.N. Srivastava and S.N. Sahu, Phys. Status Solidi (a) 88 (1985) 497. R.A. l~udrean and R.D. Rauh, Solar Energy Mater. 7 (1982) 385. [15] M. Pamcker, M. Knster and F.A. Kroger, J. Electrochem. SOc. 125 (1967) 566. [161 L.E. Lyons, O. Morris, D.H. Horton and J.G. Keyes, J. Electroanal. Chem. 168 (1984) 101. [17] R.N. Bhattacharya, K. Rajeshwar and R. Noufi, J. Electrochem. SOc. 131 (1984) 939. [18] V.K. Kapur, B.M. Basol and E.S. Tseng, Solar Cells 16 (1986) 65. [19] G. Hodes, T. Engelhard and D. Cahen, Thin Solid Films 128 (1985) 93. [20] T.L. Chu, S.S. Chu, S.C. Lin and J. Yue, J. Electrochem. SOc. 131 (1984) 2182. [211 C.D. Lokhande and G. Hodes, Solar Cells 21 (1987) 215. [22] G. Hodes and D. Cahen, Solar Cells 16 (1986) 245. [23] R.N. Bhattacharya and K. Rajeshwar, Solar Calls 16 (1986) 237. [24] C.D. Lokhande, J. Electrochem. SOc. 134 (1987) 1727. [25] J. Szot and D. Haneman, J. Appl. Phys. 59 (1986) 2249. [26] D. Haneman and J. Szot, Appl. Phys. Letters 46 (1985) 778. [27] C. Sdla, P. Boncorps and J. Vedel, J. Electrochem. SOc. 133 (1986) 2043. [28] J. Szot, G.J. Storr and D. Haneman, J. Appl. Phys. 60 (1986) 4032. [291 S.M. Wasim, Solar Cells 16 (1986) 289.

[11] [12] [13] [14]