CuInSe2 solar cell heterojunctions

CuInSe2 solar cell heterojunctions

Solar Cells, 28 (1990) 31 - 39 31 ELECTRICAL C H A R A C T E R I Z A T I O N OF ALL-SPUTTERED CdS/CuInSe2 S O L A R CELL H E T E R O J U N C T I O N...

390KB Sizes 4 Downloads 53 Views

Solar Cells, 28 (1990) 31 - 39



Departamento Electricidad y Electronica, Facultad de Ciencias Fisicas, Universidad Complutense, 28040 Madrid (Spain)

(Received August 8, 1988; accepted in revised form March 15, 1989)

Summary Current-voltage ( J - V ) and capacitance-frequency (C-F) measurements were taken for all-sputtered CdS/CuInSe2 solar cell heterojunctions, in the range 100 Hz - 10 MHz, to investigate the effect of interface states on the conduction mechanism. The samples analyzed had different compositions for the chalcopyrite CuInSe2 layer. All of the samples showed a hybrid tunneling/interface recombination mechanism with the major contribution coming from the interface recombination mechanism for devices having a stoichiometric or indium-rich CuInSe2 layer, and from the tunneling conduction mechanism for copper-rich devices. An estimation of the relative distribution of interface states has been performed. The values obtained for interface state density were as high as 6 X 1012 V -1 cm -2 for copper-rich heterojunctions, and were lower than 2 × 1011 V -1 cm -2 for devices having a stoichiometric or indium-rich CuInSe2 layer.

1. Introduction The CdS/CuInSe2 solar cell heterojunction has proven to be a very promising candidate for terrestrial photovoltaic applications. Efficiencies as high as 12% have been achieved using the three-sources evaporation technique [1], a growth procedure which is n o t easily scaled up to industrial production volumes. Sputtering, however, appears to be more suitable for use on an industrial scale, b u t the efficiencies reported have been no higher than 6% [2]. Although much effort has been devoted to the study of cell operation and to the modelling of evaporated devices [ 3 - 6 ] , to our knowledge there has been little w o r k on devices produced by sputtering [7]. An analysis of current transport in sputtered heterojunctions can be very useful in obtaining information a b o u t the loss mechanisms responsible for the lower efficiencies. These loss mechanisms are often related to the 0379-6787/90/$3.50

© Elsevier Sequoia/Printed in The Netherlands


presence of interface states which act as recombination centres, and in this c o n t e x t capacitance-frequency (C-F) measurements are useful in determining the relative distribution of those states [8]. In this paper, we present a study of the current-voltage ( J - V ) and capacitance-frequency (C-F) characteristics of all-sputtered CuInSe2/CdS solar cell heterojunctions.

2. Experimental details

2. 1. Materials and cell preparation Molybdenum back contact, p-CuInSe2 absorbing layer, and n-CdS window layer of heterojunction structures were grown in a sputtering system (GCA Vacuum Industries) described elsewhere [9]. Targets were as follows: (i) m o l y b d e n u m , 5 in diameter, 99.999% pure, supplied by CERAC; (ii) CdS, 5 in diameter, 99.999% pure, supplied by CERAC; (iii) CuInSe2, 3 in diameter, with 5 wt.% Se excess, 99.999% pure, supplied by CERAC. Heterojunctions were grown w i t h o u t breaking the vacuum according to the procedure described below. M o l y b d e n u m back contact 1 0 0 0 - 2000 A thick was sputtered on chemically cleaned alumina substrates (2.5 cm × 2.5 cm × 0.1 cm). A 2 3 pm thick CuInSe2 layer was then deposited in At/H2 atmosphere in order to control the stoichiometry of films (growth parameters: 10 mTorr argon pressure; 1.5% H2 partial pressure, 900 V target voltage, and 3 3 0 - 400 °C substrate temperature). The variation of substrate temperature caused changes in the physical properties of the CuInSe 2 layer, as described in a previous work [10]. Relevant properties at a few significative substrate temperatures are presented in Table 1. After substrate cooling, a CdS window layer 2 #m thick was grown at 14 mTorr argon pressure and 800 V target voltage. The resistivity of the CdS film ranged between 1 and 4 ~2 cm, and its transmittance was about 80% in the visible and N I R range [11].

TABLE 1 Relevant CuInSe2 film properties and photovoltaic parameters of three representative p-CuInSe2 films and corresponding cells Cell

Ts a F i l m p r o p e r t i e s (°C) p


( ~ cm)

1747 (Type 1) 364 1752 (Type 2) 390 1741 (Type 3) 375

760 10 25

Chalcopyrite Chalcopyrite Chalcopyrite

aGrowth temperature of p-CuInSe2 films.

[Se] (%)

[Cu] (%)

[In] (%)

49.4 45.6 48.3

24.3 27.9 26.5

25.7 26.4 24.8

P h o t o v o l t a i c parameters

Voc (V)

Jsc (mA em -2 )


0.277 0.191 0.156

22 2.7 6.1

0.34 0.34 0.30

33 The sample was then removed from the sputtering chamber and placed in an evaporator allowing exposures to the atmosphere of no more than 1 rain. An aluminium grid 1000 - 2 0 0 0 A thick was then evaporated on the CdS layer. The aluminium grid was grown through a mechanical mask, so that top contact was n o t optimized. Short circuit current densities (J~c) can, therefore, only be considered as indicative. Heterojunctions were annealed in air at temperatures of 200 °C for periods ranging between 20 and 120 min in order to activate their photovoltaic response. 2. 2. Heterojunction characterization Heterojunction characterization consisted of J - V and C - F measurements. The d.c. characteristics were measured at variable temperature ( 2 0 0 - 350 K) in dark and at r o o m temperature under AM 1 illuminated conditions supplied b y a xenon lamp calibrated with a single crystalline silicon solar cell. The a.c. characterization was carried o u t b y measurement of the complex impedance, at variable temperature, in the frequency range 100 Hz - 10 MHz, using an HP 4 1 9 2 A impedance analyzer. Both techniques have been described elsewhere [12].

3. Results In studying the composition of the p-CuInSe2 layer, we analyzed three types of solar cell: T y p e 1 cells, based on stoichiometric or slightly indiumrich films; T y p e 2 cells, based on copper- and indium-rich and selenium-poor material; and Type 3 cells, produced with copper-rich and indium-poor CuInSe2 layer (see Table 1). Cells produced with stoichiometric or indiumrich material were electrically indistinguishable. Electrical characteristics of Type 2 and Type 3 heterojunctions were also indistinguishable. Hence, in the following we will only show electrical characteristics of T y p e 1 and Type 2 heterojunctions. Cells with excess selenium in the CuInSe2 layer always showed sphalerite structure [10] and will n o t be considered because of their vanishingly low photovoltaic response. 3.1. Current vs. voltage under illumination We present in Table 1 the relevant photovoltaic parameters of the three types o f cell. The cell parameters listed in this table correspond to three cells chosen from a set of 30 analyzed heterojunctions. Henceforth, we will focus the discussion on the analysis of measurements performed on the cells listed in Table 1. Type 2 and T y p e 3 cells showed very low J ~ values, as reported b y other authors [13]. The small Voc and F F values of the near stoichiometric based cells (Type 1) are probably related to the nonohmicity of the Mo/CuInSe2 back contact. This problem has been overcome b y other authors b y means of a two-layer CuInSe2 structure [14]. In this our first approach, we are chiefly interested in studying the transport mechanisms and interface properties o f all-sputtered cells rather than in achieving a high photo-








0.4 0,5 0.6 VOLTAGE (V)




Fig. I. J - V characteristics for several temperatures for a cell showing the influence of shunt and series resistance: 1, 254 K; 2, 280 K; 3, 298 K; 4, 313 K; 5 , 3 3 9 K; 6, 348 K; 7, 357 K.

voltaic response; of CuInSe2. The complication of interpretation of

the structure has therefore been grown using a single layer presence of two layers would have resulted in an additional the structure of the depletion region, thus making the J - V and C - F characteristics more difficult.

3.2. Current vs. voltage in dark

J - V characteristics of all the cells analyzed showed a pronounced influence of series resistance (R,) and shunt resistance (R0h), as shown in Fig. 1. In order to determine the relevant junction parameters, the assumption has been made that junction behaviour can be modelled by the classical one-diode equivalent circuit, so that the following J - V relation can be written: Vexp -- JexpRs

+ Jo[ e x p ( C( Ve=p - Je~pRs)} -- 1] (1) Rsh where Je~p and Vexp are the values of current and voltage experimentally measured, J0, R~ and Rsh have their usual meanings, and C is a parameter that may or may n o t be temperature dependent and that will allow us to identify the conduction mechanism. All have been calculated from dark measurements following the m e t h o d described by Fuchs and Sigmund [15]. In our assumption, a single exponential is supposed to account for cell operation; this point was confirmed by experimental results for temperatures higher than 230 K. For lower temperatures, eqn. (1) did not fit the experimental data, probably because of these temperatures the Mo/CuInSe2 contact shows high rectifying behaviour due to a higher resistivity of the CuInSe2. The C factor showed a marked temperature dependence for Type 1 cells, but only a weak temperature dependence for Type 2 cells, as shown in Fig. 2. For Type 1 cells, as can be seen in Fig. 3, the J0 factor showed an exponential activation with temperature in the form Jexp =

J0 = Jo0 exp(--AE/KT)












T > 20 (..)




16 14

[ 3

I 4


-7 5


I 3

I 4 IO00/T


Fig. 2. C factor vs. 1000/T for Type 1 (o) and Type 2 (+) cells. Fig. 3. Jo vs. 1000/T for Type 1 cells. -3


5 o


,/ -6 250



Fig. 4. Jo vs. T for Type 2 cell. AE ranged between 0.44 and 0.46 eV for Type 1 cells. For Type 2 cells on the other hand, J0 followed the form, as in Fig. 4, Jo = Joo exp(T]To)


To values ranged between 14 and 24 K.

3.3. C a p a c i t a n c e - f r e q u e n c y m e a s u r e m e n t s C - F measurements at various d.c. forward bias voltages have been

performed in the t w o types of cell. For low values o f d.c. bias, there was an additional contribution to the capacitance, with respect to its value at zero, and reverse bias appeared in the low-frequency region, relaxing around 102 Hz. This behaviour occurred for all cells analyzed, and is illustrated in Fig. 5 for Type 1 cells. Measurements were performed at various tempera-

36 600




_0 O+



~-- 0


8O E o

z Q.



60 o

o 200

+ °o






I 2



I 4

++++-+c~-+-+.,~4.~ 0 5 6 7


Fig. 5. C a p a c i t a n c e (o) a n d losses (+) f o r w a r d bias.

v s.

f r e q u e n c y for t h e T y p e 1 cell at 0.2 V d.c.

tures to establish whether this additional contribution to the capacitance was due to deep levels or to interface states. Details a b o u t this point will be given below.

4. Discussion 4. I. Current-voltage characteristics None of the usual heterojunction conduction mechanisms [16] can account satisfactorily for the experimental data shown in Fig. 2. The temperature dependence of the C factor for the T y p e 1 cells is more pronounced than in a tunneling mechanism, in which C is constant and is weaker than in the interface recombination mechanism. This suggests a simultaneous contribution of the t w o mechanisms: carriers are thermally activated up to a certain height o f the barrier, at which the barrier is thin enough to be crossed by tunneling. This process is followed by carrier recombination through interface states in the usual manner. This thermal activation of the tunneling mechanism is also suggested b y the temperature dependence of J0, of the form Jo = J00 exp(--AE/KT), found for Type 1 cells, as shown in Fig. 3. Such a conduction mechanism (tunneling/interface recombination) has been reported by Miller and Olssen [6] as being responsible for the J - V characteristics of evaporated CdS/CuInSe2 Boeing cells; it has also been reported as being responsible for the J - V characteristics of all-sputtered CdS/Cu=S cells [12]. Following this model, an expression for the current can be given in the form: J = Joo e x p ( - - A E / K T ) { e x p ( C V ) -- 1}




C=B +


(5) AkT This model reduces to a pure tunneling mechanism if f = 0 and to an interface recombination mechanism if B = 0. As stated above, Fig. 2 shows the linear behaviour found in plots of C vs. 1 0 0 0 / T for Type 1 and Type 2 cells, in good agreement with eqn. (5). In Table 2, fitting parameters B, f/A, AE and Joo are listed for the t w o analyzed cells and for a high-efficiency Boeing representative cell (Cell BAC788A) [6]. TABLE 2 Fitting parameters of the tunneling/interface recombination conduction mechanism for different solar cells (with To value for Type 2 cell)

Solar cell


f /A

(A cm -2 ) Type 1 Type 2 BAC-788A a

1.56 × 103 1.46 × 10 -I~ b 8.0 x 102

0.37 0.04 0.420



(V -1)


TO b (K)

7.8 14.6 12

0.45 -0.55

-22.4 --

aHigh efficiency (8.25%) Boeing cell (from ref. 6). b T 0 and J00 values for the pure tunneling conduction mechanism J = Jooexp(T/To)× exp(BV).

A few results should be pointed out: (i) Type 1 cellsshow relatively high f/A and A E values, close to those of the efficientBoeing cell; (ii) Type 2 cells(copper-richCuInSe2) show very low values of f/A, the C factor being almost temperature independent (Fig. 2). On the other hand, J0 have a temperature dependence in the form Jo = Joo exp(T/To), as in Fig. 4, characteristicof a pure tunneling conduction mechanism. The high f / A and AE values for Type 1 cells suggest a stronger contribution of the interface recombination mechanism. Type 2 cells, on the other hand, show a stronger contribution of the tunneling mechanism. C-F measurements can be helpful in finding the reasons for such differences in the cell conduction mechanism. 4.2. Capacitance-frequency measurements The low-frequency relaxation which appears under forward bias at 10 2 Hz (Fig. 5) does n o t show temperature dependence. This suggests that this capacitance contribution may be associated with interface states [8]. In the presence of interface states, deionization of the state at the Fermi level m a y occur via these interface states and probably through tunneling processes instead of the usual thermal activation. When a forward bias is applied to the heterojunction, the occupancy of the interface states will follow the variation of the bias voltage across the depletion region. The small alternating signal will then cause changes in the interface state density at the

38 Fermi level if it can respond to the frequency of the applied signal. The capacitance can therefore be expressed as the sum of the depletion capacitance Cd and the contribution due to interface states Cis, according to the following equation [ 17] : Cis C = C d "t"


1 + (COT)2

Taking the simplest approach, the interface state capacitance, at a frequency at which the levels can respond, may be related to interface state density at the Fermi level Nis(V) as follows [17] :

ci,(v) Nis(V)




Capacitance measurements at forward bias can then be used to obtain a relative distribution of interface states. In Fig. 6 the distribution is shown for Type 1 and Type 2 cells. The greater contribution of the tunneling mechanism in T y p e 2 cells seems to be related to the higher density of interface states probably associated with the copper excess at the interface [18]. All devices having a copper/indium composition ratio greater than 1, independently of whether they were rich or poor in indium, showed similar behaviour. In the presence of a large density of such gap states, a multistep tunneling through those states will result, energetically more favourable than the thermal activation over the barrier. This is consistent with the lower photovoltaic response observed in those cells. Interface states form recombination paths for the photogenerated carriers which reduce Jsc and hence Voc. 15 O ©


12 0



7 0

% z


O ~t





. I 0.1


I I 0.2 0.5 VOLTAGE ( V )

I 0.4

Fig. 6. Interface state distribution


vs. forward voltage for

Type 1 (*) and Type 2 (o) cells.

5. Conclusions C-F


and J - V characteristics of all-sputtered CdS/CuInSe2 heterohave been presented. Two kinds of cells, based on CuInSe2

39 but with different compositions, have been analyzed. A hybrid model (tunneling/interface recombination) accounts for the current transport m e c h a n i s m o f t h e t w o t y p e s o f cells. I n cells b a s e d o n s t o i c h i o m e t r i c or i n d i u m - r i c h CuInSe2 ( T y p e 1) a significant c o n t r i b u t i o n was m a d e b y t h e i n t e r f a c e s t a t e r e c o m b i n a t i o n m e c h a n i s m a n d t h e r e was b e t t e r p h o t o v o l t a i c r e s p o n s e , while in c o p p e r - r i c h C u I n S e : cells ( T y p e 2) a g r e a t e r c o n t r i b u t i o n was m a d e b y t h e t u n n e l i n g m e c h a n i s m a n d t h e r e was v e r y p o o r p h o t o v o l t a i c r e s p o n s e . A n e x p l a n a t i o n t o this b e h a v i o u r has b e e n given in t e r m s o f a h i g h e r d e n s i t y o f i n t e r f a c e s t a t e s in T y p e 2 cells, d e d u c e d f r o m C - F m e a s u r e m e n t s .

Acknowledgment This w o r k w a s p a r t i a l l y f i n a n c e d b y t h e S p a i n - U . S . A . J o i n t C o m m i t t e e under Grant CCA-8411/046.

References 1 W. E. Devaney, R. A. Mickelsen and W. S. Chen, Proc. 18th IEEE Photovoltaic Specialists' Conf., Las Vegas, NV, October 21 - 25, 1985, IEEE, New York, 1986, p. 1733. 2 T. C. Lommasson, H. Talieh, J. D. Meakin and J. A. Thornton, Proc. 19th IEEE Photovoltaic Specialists" Conf., New Orleans, LA, May 4 - 8, 1987, IEEE, New York, 1987, p. 1285. 3 A. Rotwarf, IEEE Trans. Electron Devices, ED-29 (1982) 1513. 4 M. Eron and A. Rotwarf, J. Appl. Phys., 57 (1985) 2275. 5 C. Goradia and M. Ghalla-Goradia, Sol. Cells, 16 (1986) 611. 6 W. A. Miller and L. C. Olsen, IEEE Trans. Electron Devices, ED-31 (1984) 654. 7 N. Romeo, A. Bosio, V. Canevari and L. Zanotti, Proc. 7th Commission o f the European Communities Conf. on Photovoltaic Solar Energy, Sevilla, October 27 - 31, 1986, Reidel, Dordrecht, 1987, p. 656. 8 J. Santamaria, E. Iborra, I. Martil, G. Gonzalez Diaz and F. Sanchez Quesada, Semicond. Sci. Technol., 3 (1988) 781. 9 I. Martil, G. Gonzalez Diaz and F. Sanchez Quesada, Thin Solid Films, 114 (1984)

327. 10 I. Martil, J. Santamaria, E. Iborra, G. Gonzalez Diaz and F. Sanchez Quesada, J. Appl. Phys., 62 (1987) 4163. 11 I. Martil, G. Gonzalez Diaz and F. Sanchez Quesada, Sol. Energy Mater., 12 (1985) 345. 12 E. Iborra, J. Santamaria, I. Martil, G. Gonzalez Diaz and F. Sanchez Quesada, Sol. Energy Mater., 17 (1988) 279. 13 J. A. Thornton and T. C. Lommasson, Sol. Cells, 16 (1986) 165. 14 R. Noufi and J. Dick, J. Appl. Phys., 58 (1985) 3884. 15 D. Fuchs and H. Sigmund, Solid-State Electron., 22 (1986) 791. 16 M. Arienzo and J. J. Loferski, J. Appl. Phys., 51 (1980) 3393. 17 Tavakolian and R. E. Hollingsworth, J. Vac. Sci. Technol. A, 4 (1986) 488. 18 T. J. Coutts and J. D. Meakin, Current Topics in Photovoltaics, Academic Press, London, 1986, Chapter 2.