Solar Energy Materials 14 (1986) 51-59

North-Holland, Amsterdam

51

T H E P E R F O R M A N C E O F T H I N F I L M C u ~ S / C d 1_ y Z n y S SOLAR
CELLS
AND ITS D E P E N D E N C E ON T H E AMBIENT A T M O S P H E R E
DURING H E A T T R E A T M E N T
M. F A D L Y 1) and H. A F I F I 2)
I Helwan University, Faculty of Science, Helwan, Cairo, Egypt
2 National Research Centre, Physics Department, Solid State Physics Laboratory,
Dokki, Cairo, Egypt

Received 25 June 1985; in revised form 27 January 1986

Changes in the performance of Cu~S/Cd I _yZnyS solar cells have been studied after annealing
in air, nitrogen and hydrogen. The zinc content has two values: 5 and 10%. Nitrogen and
hydrogen treatments have been found to produce effects notably different to the effects of
annealing in air. The cell efficiency has been nearly doubled in the case of hydrogen treatment
compared to that of nitrogen. The results are consistent with previously reported measurements
which showed that the CuxS layer and the Cd l_yZnyS interface structure were unstable when
heat treated in air compared to the effect of annealing in either nitrogen or hydrogen.

I. Introduction

Much attention has been focused on thin film C u z S / C d S heterojunctions due to

their great promise as a low cost solar power converter. The conversion efficiency is
limited due to the mismatch of lattice constants and electron affinities of the two
components [1-2]. These mismatches can be reduced by the use of a Cd 1_yZnyS
film instead of the CdS layer [3]. The CuxS layer and its function are strongly
influenced by the past fabrication annealing, which is an important step in most
cell-manufacturing, procedures. The annealing conditions required to optimise the
cell characteristics have been the subject of some uncertainty.
In the present investigation we have used a chemical spray deposition process
[4-5] for producing a CdS layer with 5 and 10 mol% of ZnS. The CuxS layer has
been formed by the dip technique [3]. The fabricated cells have been annealed at
150C for different times in air, nitrogen and hydrogen. The cell performances have
been studied with respect to the different annealing parameters.
No attempts were made to increase the efficiency of our cells using transparent
grids, thin film of Cu on the CuxS surface or antireflection coatings. This should be
taken into consideration when comparing the parameters of our cells with those
reported in the literature.
0165-1633/86/$03.50 Elsevier Science Publishers B.V.
(North-Holland Physics Publishing Division)

52

M. Fadlv, H. Afifi / Thin film Cu ~S~ Cd I

v Zn.v S solar cells

2. Experimental details
Backwall cells (area: 1 cm 2) have been made using a bifilm structure. A
Cd 1 ,.ZnvS film (5-6 ~tm) is deposited on a layer of CdS doped with A1 (2-3 ~m)
that was previously deposited on an ITO (0.5 ~m) covered glass substrate as shown
in fig. 1. The starting solution for making C d t _ v Z n y S layer was a 0.1M solution of
Cd CI 2 5H 20, ZnCI 2 and ( N H :) 2CS. By varying the relative proportions of CdC1 z
5 H 2 0 and ZnC12 solutions and taking (NH2)2CS in a particular ratio one may get
starting solutions having different cadmium to zinc mole percentage ratios. The
films are prepared by spraying the mixture through a specially designed double
nozzle.sprayer. The spray rate is 8.5-9 cm3/min, the gas rate is 6.5 1/min and the
substrate temperature is 420C. The CuxS films have been formed by dipping the
alloy films in a heated solution ((99 + 1)C) of CuC1 with p H = 4 for 5 s. The films
are subsequently washed and rinsed by deionized water and dried in a stream of dry
nitrogen.
Heat treatment has been carried out for different times in a preheated furnace
maintained at (150 + 3)C. The employed atmosphere is pure and dry air, nitrogen
and hydrogen.
For electrical characterization, Hg on CuxS and silver paste on ITO were taken
as electrodes. The current-voltage ( J - V )
characteristics of the cells in dark and
light were plotted using an X - Y
recorder and a signal generator. The solar
simulator consists of a tungsten lamp at a filament temperature of 2800C with a
water filter. The equivalent radiation intensity of the simulator was 100 m W / c m 2 of
sunlight. The experimental work was done in Le Centre d'Etudes Electroniques des
Solides, Universit6 des Sciences et Techniques du Languedoc, Montpellier, France.

~
0 ~~
~

"

~Cd

Zn S
( 5-6 ~rn )
~ C d S(A[)
( 2_3,~m )
I T O (0.5/~m ~;
Pyre x (2/~m)

111{1
Ill u r N i n a t i o n
Fig. 1. Cross-sectionof CuxS/Cd 1_ ~,Zn~S solar cell.

M. Fadly, H. Afifi / Thin film C u x S / C d I _ y Z n y S solar cells

53

3. Results and discussion

The effect of dipping time on cell performance has been studied [6]. The reported
results show that the 5 s dipping time gives the best cell performance. Therefore, the
dipping time used here is 5 s. The heat treatment studies [7] on C d S / C u 2 S solar
cells indicate that the junction formation is rapidly formed near a temperature of
180C while it is slowly formed at about 130C. Above 220C the junction will be
destroyed. Therefore, the C u x S / C d a _ y Z n y S solar cells were heat treated at 150C
which is between these two temperature extremes of 180C and 220C. The heat
treatment at 150C avoids the cell destruction and has a reasonable period for
treatment [8].
The effect of heat treatment at 150 in three different ambients (air, N 2 and H2)
on dark and light J - V characteristics of CuxS/Cd0.gZn0AS solar cell is demonstrated in fig. 2. It is clear that heat treatment is air gives inferior electrical
characteristics while heat treatment in N z and H 2 give better results. The inferior
characteristics may be due to a significant change in the stoichiometry of the CuxS
layer due to loss of copper. The Cu loss occurs primarily at the free surface of the

/ /"

//

.,/

I
/I

101o Zn

/'/

/I
/ / .I /

/"

/
/
./_,1"

//
/
i i1,1" i.I
I / ./ /
/

//"

///.

200
-

s
./7

/
i-"

/'

~,/-

I
400

V(mV)

//

/
/

==-T5

/
/
/
/

-15"
i

/
J
f

J ( m A/c m 2 )

air

3 0 min,

H2

2 0 rain.

- N2

20 min.

J
J

Fig. 2. Effect of heat treatment at 150C in air, N 2 and H2 on dark and light J - V characteristics of the
CuxS/Cdo.9Zno. 1 solar cell.

54

M. Fadly, H. Afifi / Thin film C u , S / C d j _ ,,Zn.~,S solar cells

/'

/ IIii

////
./ I/ /1"
//l;l/"

~ /o ZF)

,,
/' / P

I/

///
jo..

~
I

before
- -

heat treatmen'

H 2 20

rain.

N2 2 0

min.

j (mA/cm 2 )

Fig. 3. Dark and light J - V characteristics for the CuxS/Cd0.9Zn0.65S solar cell before and after
annealing at 150C in N 2 and H 2.

CuxS layer where a reaction with air makes it possible to form CdO, Cu 2804, CH 2
and Cu CO 3, or a combination of these, at the interface by diffusion. The literature
reported results and discussion under nearly the same conditions [7-9] and attributed these detrimental effects to the presence of oxygen.
The dark and light ( J - V ) characteristics for C u x S / C d l_yZn~,S ( y = 0.05 and
0.10) solar cells before and after annealing in N 2 and H 2 are given in figs. 3 and 4.
The observed effects induced by heat treatment are: (1) a large decrease in dark
forward-bias current, (2) crossing of the light and dark ( J - V ) curves, (3) a slight
increase of open-circuit voltage Vow, and (4) a decrease and increase in short-circuit
c u r r e n t ( J s c ) in case of N 2 and H 2 respectively.
The advantage of using a nitrogen or a hydrogen ambient during the annealing
stage is undoubtedly due to the avoidance of oxide formation with its consequent
lowering of the x value in the Cu~S layer. Heat treatment decreases the dark
forward-bias current by increasing the depletion layer width, hence the tunneling
forward current markedly decreases. The intersection of ( J - V ) curves in dark and
under illumination may be due to the fact that illumination makes the depletion
layer shrink and subsequently allows an increased tunneling current to flow. The
formation of a thin djurleite interlayer between CuxS and C d l _ y Z n y S may be

M. Fadly, H. Afifi / Thin film C u x S / C d I _ y Z n v S solar cells

//

/ /

,1//

i'
,I

/, , '

i i

100

200

.ff

-- 5

i f
/

/300
//

....................

/ /

//

///

//'

i /i
i Y

i'/
/I
/

/
/

' - , .....

~ ~ ~
--

i1/// i/ //
li ,i" /i'

10o/o Z n

55

_10~

-I-/

./~400

V (mY)

i'//
i ' i ' l

b e f o r e heat t r e a t m e n t
N 2 40

rain

-20
J (mA/cm 2 )

H 2 160 m i n .

Fig. 4. Dark and light J - V characteristics for the Cu:~S/Cdo.gZno.lS solar cell before and after
annealing at 150C in N 2 and H 2.

responsible for the increase of Voc. This could be explained by the change of the
indirect band gap of chalcocite to the direct band gap of djurleite. Nakama [10]
reported that at 300 K the band gap energy for Cu~S changes from 2.3 to 1 eV when
x varies from 1.8 to 2. The change in the band gap energy of CuxS from 1.16 to 1.8
eV due to a change in x was reported by other investigators [11,12]. Rothwarf et al.
[13] used the following equation to calculate the Vo~ for a CdS/Cu~S solar cell:

qVo = Eg - Ax + k T In Js~ - k T ln( qN~2S,) - k r ln( A J A z ) ,

where q is the electronic charge, Aj is the junction area, A . is the normal area, N~2
is the effective density of states at the band edge of CdS, S l is the interface
recombination velocity which depends on the lattice mismatch between CuxS and
CdS, Eg is the band gap of CuxS, and Ax is the difference in electron affinity
between CuxS and CdS = 0.2 eV. The Vo~ shows an increase in its value when the
high value of the direct band gap for djurleite replaces the low value of the indirect
band gap for chalcocite in the above equation. The presence of the djurleite
interface layer and the change in the band gap energy of Cu~S were reported by
Oktik et al. [14] in their study of Cd t _ y z n y s / c u x s solar cells. The decrease in J~c
in the case of N 2 annealing could be attributed to a decreased electric field in the

56

M. Fadly, H. Afifi / Thin film Cu~S/Cd l ~,ZnvS solar cells

E
ij
E
L,'
0

2O

121

500

400

~3oo
D

200

5/o

Zn

10/o

100
~,

100

200
Annealing

Zn

3 0 0 t (rnin.
trme

Fig. 5. Variation of Jsc and V~: with annealing time at 150C in N 2 for C u ~ S / C d I vZnvS solar cells
(y = 0.05 and 0.10).

CdS at the interface, which results in an increased recombination through interface

states [15].
The variation of J.~c and Vo~ with annealing time at 150C in N 2 and H 2 for
C u ~ S / C d 1 yZnyS ( y = 0.05 and 0.10) cells are displayed in figs. 5 and 6. In the
case of cells with y = 0.1 and a N 2 ambient, Jsc and Voc show a high decrease and
increase in the first 40 min of the annealing time respectively and then they saturate.
For cells with y = 0.05, the Vo~ decreases from 240 to 160 mV in the first 40 min of
annealing after which it increases until it reaches its value before annealing ( - 240
mV) after 160 min of annealing. Also, the Js~ shows a marked decrease with
annealing time as shown in fig. 5. The variation in Voc and Js~ for cells with y = 0.1
causes the efficiency to go through a maximum after 40 rain of annealing, and
decreases on prolonged heat treatment as shown in fig. 7. The efficiency for cells
with y = 0.05 shows a drastic decrease with annealing time. Therefore the optimal
annealing period in N 2 ambient for cells with y = 0.05 is less than that for the cells
with y = 0.1. The heat treatment in H 2 gives the best electrical characteristics for
the two cell types. The cell performances J~c, Vo~ and ~/ show an increase with
annealing time as shown in figs. 6 and 8.
The strong decrease in J~c and the slight increase in Voc with increasing N 2
annealing time, may be due to the formation of a djurleite interlayer and the change
in stoichiometry of the copper sulphide layer. This is in good agreement with
N a k a y a m a [16] who postulated that heat treatment after barrier layer formation
produces the djurleite interlayer. Also, longer exposure to heat treatment produces

M. Fadly, H. Afifi / Thin film C u x S / C d I _ ~.ZnrS solar cells

57

E
E

z~

20

H2~_

>o
5O0

16

z.O0

12

3O0

20O

4-

I
100

I
200

5/o Zn

100
I 10/ Zn
I
300

Annealing

t(min.
time

Fig. 6. Variation of J~: and Vo with annealing time at 150C in H 2 for C u x S / C d l _ v z n y S solar cells
( y = 0.05 and 0.10).

(O/o)

%)

N2
10/o Zn

o 5/o Z n

1.5

).5

100

200

t (rain.)

anneolir~g

time

Fig. 7. Variation of cell efficiency ~/ with annealing time in H 2 for C u s x S / C d l _ y Z n y S

( y = 0.05 and 0.10).

solar cells

58

M. Fadly, H. A/ifi / Thin film C u ~ S / C d I _ y Z n y S solar cells

3.5
H2
o

2.5

1,5
n

0.5

100

5 %

Z n

200
annealing

t (rni n.)
time

Fig. 8. Variation of cell efficiency ~ with annealing time in H 2 for C u x S / C d l _ y Z n y S

(y = 0.05 and 0.10).

solar cells

cell degradation due to phase transformation. The changes in stoichiometry can

profoundly alter the carrier mobility and the recombination kinetics as well. The
maximum efficiency in the C u x S / C d S system occurs when x is very close to 2.00.
Deviations from this value may occur during heat treatment. These deviations take
place by copper ion diffusion and oxidation. Palz et al. [17] noted a substantial loss
in Jsc when thin film C u x S / C d S cells are exposed to oxidation. The oxidation of the
CuxS brings about a transformation from chalcocite to djurleite. The copper ion
diffusion increases the width of the depletion layer on the CdS side. This gives rise
to a marked decrease in Jsc and increase in the series resistance of the cell. Shiozawa
et al. [18] showed that copper diffusion is appreciable at 150C. Martinuzzi et al.
[19] saw evidence of diffusion of Cu atoms into CdS and of Cd diffusion into the
CuxS layer by chemical analysis.
The cell performances Jsc, Vo~ and ~/are improved by heat treatment in hydrogen
as shown in figs. 6 and 8. This improvement may be due to the amelioration of the
CuxS layer stoichiometry, due to the possibility of the reaction between hydrogen
and sulfur, and the reduction of oxidation conditions. The reduction reaction could
be increased with annealing time which gives rise to an increased x value. The
primary effect of varying x is to change the properties of the Cu~S phases present
in the layer. Palz et al. [17] observed that Jsc for chalcocite is about a factor of 10
higher than that for djurleite, with an almost linear variation with x in the range
1.95 ~< x < 2.00. Thus raising x would decrease carrier density and raise the effective
absorption coefficient, tending to increase J~.

M. Fadly, H. Afifi / Thin film C u x S / C d I _ y Z n y S solar cells

59

4. Conclusion
Heat t r e a t m e n t of C u x S / C d t _ y Z n v S solar cells at 150C in N 2 causes a decrease
in cell p e r f o r m a n c e due to changes in the CuxS layer stoichiometry in the direction
of lower x value. The d e g r a d a t i o n of the stoichiometry takes place b y the diffusion
of copper ions u n d e r a c o n c e n t r a t i o n gradient a n d t e m p e r a t u r e effects. This will
reduce the CuxS layer p h o t o n a b s o r p t i o n coefficient, a n d c o n s e q u e n t l y the light
generated c u r r e n t in the cell.
The cell p e r f o r m a n c e s Jsc, Voc a n d ~ are i m p r o v e d b y a n n e a l i n g in h y d r o g e n
a m b i e n t which is n o t the case b y a n n e a l i n g in a n N 2 atmosphere. This is because of
the a m e l i o r a t i o n of the CuxS layer stoichiometry a n d the complete e l i m i n a t i o n of
oxidation conditions.

Acknowledgement
T h e authors would like to t h a n k Dr. J. Bougnot a n d Dr. S. D u c h e m i n for their
technical assistance.

References
[1] L.C. Burton, Appl. Phys. Len. 35 (1979) 780.
[2] F. Zapien Nataren, B. Bouchikhi, S. Chandrosekhar, J.P. Crest and S. Martinuzzi, J. Cryst. Growth
59 (1982) 409.
[3] A. Banerjee, S.R. Das, A.P. Thakoor, H.S. Randhawa and K.L. Choppa, Solid State Electron. 22
(1979) 495.
[4] R.R. Camberlin and H.S. Skarman, J. Electrochem. Soc. 113 (1966) 86.
[5] C.W. Wu and R.H. Bube, J. Appl. Phys. 45 (1974) 648.
[6] M. Fadly and H. Afifi, to be published.
[7] B.G. Caswell and J. Woods, Phys. Status Solidi (a) 44 (1977) K 47.
[8] Th6se present6e h l'Universit6 Des Sciences et Techniques du Languedoc, Etudes des traitements
termiques des Cellules Solaires Cu2S/CdS par Marian Minoun (1980).
[9] B. Baron, A.W. Catalano and E.A. Fagen, Proc. 13th IEEE Photovoltaic Specialists Conf. (1978) p.
406.
[10] N. Nakama, J. Phys. Soc. Japan 25 (1968) 290.
[11] L.D. Partain, P.S. McLeod, J.A. Duisman, T.M. Paterson, D.E. Sawyer and C.S. Dean, J. Appl.
Phys. 54 (1983) 6708.
[12] A. Rothwarf and H. Windawi, IEEE Trans. Electron. Dev. ED-28 (1981) 64.
[13] A. Rothwarf and A. Barnett, IEEE Trans. Electron. Dev. ED-24 (1977) 381.
[14] S. Oktik, G.J. Russel and J. Woods, J. Cryst. Growth 59 (1982) 414.
[15] A.L. Fahrenbruch and R.H. Bube, Fundamentals of Solar Cells (Academic Press, New York, 1983)
p. 455.
[16] N. Nakayama, Jap. J. Appl. Phys. 8 (1969) 450.
[17] W. Palz, J. Besson, T.N. Duy and J. Vedel, 10th Photovoltaic Specialists Conf. (1973) p. 69.
[18] L.R. Shoozawa, G.A. Sullivan and F. Augustine, 7th Photovoltaic Specialists Conf. (1986) p. 39.
[19] S. Martinuzzi, F. Cabane-Brouty and J.F. Bretzner, 9th Photovoltaic Specialists Conf. (1972) p. 111.