Solar Energy Materials 16 (1987) 199-204 199
North-Holland, Amsterdam
P H O T O V O L T A I C CHARACTERISTICS OF T H I N F I L M S OF C u 2 S n S 3
Titilayo A. K U K U
Department of Electronic and Electrical Engineering, University of lfe, lle-lfe, Nigeria
Olaosebikan A. F A K O L U J O
Department of Electrical Engineering (Optics Section), Imperial College of Science and Technology,
London, SW7 2BT, UK
We report for the first time the optical absorption and photovoltaic characteristics of thin films
of Cu 2SNS3. Films obtained by the direct evaporation of the synthesized compound were observed
to be deficient in copper, while those grown in an atmosphere of copper vapour were observed to
be more stoichiometric. Both sets of films are characterised by indirect and direct absorption edge
properties. The indirect absorption edges are at 1.065 and 1.510 eV, while the direct band edges
are at 1.658 and 1.770 eV for the copper deficient and stoichiometric films respectively. The
stoichiometric film is characterised by p-type semiconducting properties while a Schottky type
solar cell developed from it is observed to have potential photovoltaic characteristics.
1. Introduction
C u 2 S n S 3 belongs to a group of compounds represented by the general formula
Al2B~vCVt(superscripts denote group in the periodic table and subscripts the num-
ber of atoms). These compounds are the ternary analogues of the widely used binary
AnB v~ semiconducting compounds [1]. The compound is characterised by a cubic
structure [1,2], with a lattice constant of a = 5.445 A. However, unlike the AHBvl
analogues, not much is known about the optical properties of thin films of this
compound. This paper reports for the first time the absorption properties of thin
films of the compound, and the preliminary results on the investigation of the
photovoltaic properties of the compound.
2. Experimental
Cu2SnS 3 powder samples were prepared by melting together in an initially
evacuated ampoule a stoichiometric mixture of 99.999% purity samples of Cu, Sn
and S. X-ray analysis with a Philips X-ray goniometer (Cu K~ radiation and nickel
filter) gave a lattice constant consistent with that reported in the literature [2].
Thin films were prepared by vacuum evaporation of the prepared compound in
Edwards high vacuum equipment capable of generating a vacuum of about 10 6
Tort. For absorption measurements, films were grown on clean soda glass or quartz
substrates maintained at temperatures between 423 and 523 K. X-ray analysis of the
200 TA. Kuku, O.A. Fakolujo/ Thinfilm Cu2SnS~
Table 1
Elemental composition of Cu2SnS3-1ike films prepared by two different methods
Element Type A (at.%) Type B (at.%) Expected (at.%)
S 40.911 26.569 28.130
Sn 48.489 34.767 34.707
Cu 10.600 38.664 37.163
films so prepared show that they are polycrystalline, and have essentially the same
structure as the base compound. However, an electron-probe microanalysis (JOEL
JSM-35 scanning electron microscope with microprobe attachment) showed that the
films are deficient in copper as detailed in table 1, and labelled as type A films. The
copper content of the films were therefore subsequently improved by using a double
evaporation technique, in which the Cu2SnS3 powder was evaporated in an atmo-
sphere of copper at a temperature of about 1373 K, at a vapour pressure of about
3 >( 10 -6 Torr [3]. Such films were found to be much more stoichiometric as also
shown in table 1 and labelled as type B films.
Room temperature transmission measurements were made in a Varian 2300
double beam spectrophotometer in the spectral range 350-800 nm. Film thickness
was determined with a Talystep instrument (Rank Taylor Hobson).
Schottky type diodes were fabricated by evaporation of an aluminium fihn,
whose thickness ranged between 0.6 and 1.0 ~m, on glass slides. This was followed
by the deposition of a C u 2 S n S 3 film of dimension 5 × 6 mm and 3-4 ~tm thick, on
substrates heated to 523 K. The top contact was made of vacuum evaporated
indium ( - 0 . 1 4 ~tm thick) which has a solar transmittance of about 70%. Ohmic
contact was made to this with very fine silver wire and silver dag. A tungsten
halogen lamp with 100 m W / c m 2 incident radiation was used for the photo 1--V
measurements with the devices held at about 300 K. The lamp was calibrated with a
solar meter (Dodge products, model 776).
3. Results and discussion
3.1. Absorption spectra
Fig. 1 shows the variation of absorption coefficient, a, with wavelength for the
copper deficient (type A) and stoichiometric (type B) Cu2SnS 3 films respectively.
The a values were obtained by the direct ratio of the transmitted radiation (T), by
two samples of thickness x = 0.16 ~m and 0.37 I~m for type A and x = 0.173 ~m
and 0.325 ~m for type B films, which were used in the equation:
T t / T 2 = e - ~ x , - x2).
Fig. 2 shows the square root of the absorption coefficient against photon energy
for the two sets of films. Plotted in this way, the curves display the well-known
TA. Kuku, O.A. Fakolujo / Thinfilm CueSnS~ 201
10
"3"
E 6
-f,
0
0 I I I I I I
450 500 550 600 650 700 750 800
Wavelength (nm,)
Fig. 1. Variation of absorption coefficient with wavelength for type A ( x ) and B (O) films.
300 xx X
20O
100
/, i II J i i i
.9 1.1 13 1,5 1.7 1.9 21 2.3
Photon e n e r g y (eV.)
Fig. 2. The dependence of the square root of the absorption coefficient on photon energy (eV) for type
A ( × ) and B (O) films.
202 ZA. Kuku, O.A. Fakolujo / Thin film Cu2SnS.+
100
BO
"7
E 60
M
/ /
b
i
40
/"
20
× •
,//
/ /
olo
0 , I ell, [l X l I I , l , I i
1.3 1.5 1.7 1+9 2.1 2.3 2.5
Photon energy(eV+)
Fig. 3. Variation of a 2 against photon energy for type A ( × ) and B (e) films.
linear dependence associated with absorption at an indirect gap, with corresponding
indirect energy gaps of 1.065 and 1.510 eV for type A and B films respectively.
However a plot of the square of the absorption coefficient against energy, shows
also a linear relationship especially in the high energy regime as shown in fig. 3. This
indicates that the high energy absorption in both sets of films may be characterised
by direct edge absorptions, with corresponding edges at 1.658 and 1.770 eV for type
A and B films respectively.
It is worth noting that both sets of films are characterised by high absorptivity
(o~ > 10 4) and also that both the indirect and direct absorption edges are well placed
in the visible range of the solar spectrum, which when compared with other
photovoltaic materials like Cu2S and CuInS 2 (having absorption edges in this
energy range [4]) suggests that Cu2SnS 3 may also be a suitable candidate for
photovoltaic cell development.
The values of the direct and indirect band gaps in type B films are about equal.
The well resolved absorption peaks at 2.138 and 2.214 eV are evident in the spectra
of type B film as observed in fig. 1. These peaks may be due to transitions to higher
energy states in the conduction band generated probably due to spin orbit splitting
of the electronic state of the component atoms. This will be further verified in future
studies.
3.2. Schottky barrier cells
For the diode configuration, only the behaviour of type B films with stoichiomet-
ric compositions has been investigated. Hot probe conductivity type measurements
77A. Kuku, O.A. Fakolujo / Thin film Cu2SnS 3 203
on such films show them to be p-type semiconducting, while a four point probe
resistivity measurement at room temperature gives a resistivity of 1.74 fa cm.
A typical current-voltage characteristic for the I n - C u 2 S n S 3 contact at T = 300
K is as shown in fig. 4. Two sections are easily distinguishable: (i) for V < 0.25 V
and (ii) for V > 0.25 V. These sections are characterised by different current-voltage
dependences [5]. However, both are characterised by an exponential relationship of
current density on voltage of the form [6]
J = J, [ e x p ( q V / n k T ) - 1],
where the components have their usual meanings.
The diode factor n was calculated to be 1.87 for region I and n = 3.4 for region
II. These values allow us to speculate (since only data at 300 K are available) that
regions I and II are governed by space charge region generation-recombination,
and thermionic emission mechanisms respectively [7].
10-3
10-A
10-5
0 0.1 012 O'.a 014 015 016 017 0'.8 019
Voltage (Volts)
Fig. 4. Dark I - V characteristics of the I n - C u 2SnS3 Schottky barrier cells.
204 T.A. Kuku, O.A. Fakolujo / Thinfilm Cu2SnS~
The saturation current, Js, was extrapolated to be 8.0 × 10 -6 A / c m 2. This was
employed to calculate the barrier height ~ from the relation [8]
J~=AR m*
mo
T2 exp(- k~ ),
where A is the rectifying area, R is the Richardson constant and m * ----0.1 m 0 [8]. q~
was obtained to be 0.73 eV at 300 K.
For typical rectifying areas of 0.22 cm 2, the cells are observed to be characterised
by an open circuit voltage of 0.17 V, a short circuit current of 0.3 mA and a fill
factor of 0.48, giving an efficiency of about 0.11%. This low efficiency is believed to
derive from the poor ohmic contact at the collector junction, and the low quality
barrier layer at the top contact. Though somewhat limited in its overall behaviour,
the CuzSnS 3 Schottky barrier cell has shown good stability. No degradation has
been observed for the devices tested, even after being heated to about 373 K under
room temperature environmental condition for periods of over 24 h. These are
encouraging observations especially if the cell output can be significantly increased.
4. Conclusions
Our investigation of the absorption properties of stoichiometric thin films of
Cu 2SnS 3 show that the films are characterised by both indirect and direct absorp-
tion edges well placed in the visible region of the spectrum at 1.510 and 1.770 eV
respectively. There is a shift in these edges to lower photon energies with copper
deficiency. Schottky type diodes fabricated from stoichiometric films show prom-
ising photovoltaic characteristics, with open circuit voltage of 0.17 V, a short circuit
current of 0.3 mA and a fill factor of 0.48, for a rectifying area of 0.22 cm 2, thus
giving an efficiency of 0.11%. We hope that improved efficiency may be obtainable
with improved deposition and contacting methods.
Acknowledgements
The authors wish to thank Prof. J.C. Anderson and Prof. Mino Green of Imperial
College, London, for financial support for the electron microprobe analysis. We are
grateful to Dr. G.A. Adegboyega for useful discussions.
References
[1] S.N. Aliev, Ya.B. Magomedov and N.V. Shchegol'kova, Sov. Phys. Semicond. 4 (1971) 1986~
[2] M. Khanafer, O. Gorochov and J. Rivet, Mater. Res. Bull. 9 (1974) 1543.
[3] R.C. Weast, CRC Handbook of Chemistry and Physics, 60th ed. (CRC Press Inc., Florida, 1979).
[4] L. Kazmerski, Solar Materials Science (Academic Press, New York, 1980) p. 508.
[5] C.R. Cromwell and S.M. Sze, Solid State Electron. 9 (1966) 1035.
[6] C.T. Sah, R.N. Noyce and W. Shockley, W. Proc. IRE 45 (1957) 1228.
[7] S. Ashok and K.P. Pande, Solar Cells 14 (1985) 14.
[8] A.A. Abdurakhimov, L.V. Kradinova, Z.A. Parimbekov and Yu.V. Rud, Sov. Phys. Semicond. 16
(1982) 156.