Elsevier

Thin Solid Films

Volumes 361–362, 21 February 2000, Pages 338-345
Thin Solid Films

Stability of Cu(In,Ga)Se2 solar cells: a thermodynamic approach

https://doi.org/10.1016/S0040-6090(99)00856-1Get rights and content

Abstract

Cu(In,Ga)Se2 (CIGS) based photovoltaic cells have demonstrated the highest solar energy conversion efficiencies ever for thin film devices. They also exhibit excellent stability in field tests and exceptional radiation hardness. The apparent paradox is that these results are obtained with a cell that contains a material that is chemically the most complex of the materials used in the various thin film solar cells. Moreover, the device itself contains many elements, compounds and interfaces, all potential focus for evolution or reaction. Because of their central importance, the basic scientific foundations for the remarkable lifetime and stability of those devices are discussed, especially but not exclusively from a chemical point of view. A first section is devoted to the assessment of the intrinsic thermodynamic stability of CIGS by a critical evaluation of available data. Its relationship with the formation energy of point defects is stressed. The chemical stability of the device interfaces are examined, including prospective buffer and window layers.

Introduction

Cu(In,Ga)Se2 (CIGS) based photovoltaic cells have demonstrated the highest solar energy conversion efficiencies ever for thin film devices, both for small area (18.8%) [1] and modules (12%) [2], [3]. They also exhibit excellent stability in field tests and exceptional radiation hardness.

The success of CIGS in solar cells is a challenge to common sense. Good ohmic contact on p-type semiconductors is notoriously difficult, but it did not appear to be so between Mo and CIGS. Polycrystalline semiconductors tend to have reduced electronic performances, as compared to single crystals, due to carrier trapping or recombination at grain boundaries. This does not seem to be a problem in CIGS where grain boundaries are easily passivated. Semiconductors are generally very sensitive to impurities, those being generally lifetime killers. CIGS seems to be relatively immune to most impurities. Ironically, the one found to diffuse from the glass substrate (Na), was also found to improve the quality of the CIGS films structurally and electronically. Moreover, the device itself contains many elements, compounds and interfaces, all potential focus for evolution or reaction. Last but not least, record efficiencies are obtained using a material that is chemically the most complex of the materials used in the various thin film solar cells (a/Si, c-Si, CdTe,....), a compound that has also metastable states and shows significant ionic conductivity. In the list above, one will recognise factors that plague or have plagued the development of other thin film solar cells.

Before discussing in details this apparent paradox, let us get an overview of the two questions addressed here: the intrinsic stability of CIGS and the global chemical stability of the device (for a detailed review, see [4], of which this work is a continuation, from where the elements of the overview were taken and where more references can be found).

As far as terrestrial application are concerned, Cu(In,Ga)Se2-based solar modules have proven their stability in long term outdoor tests as well as under accelerated lifetime test conditions [5], and actually it is not uncommon that cells and modules show some improvement during testing [6], implying that the device as prepared is not optimized. This further suggests a positive evolution of the interfaces with time. The sensitivity of the cell to humidity either via the ZnO window [7] or via the bare CIGS absorber material [8] is mainly a concern at the production level: properly encapsulated, the modules are stable for years in spite of the chemical complexity of a system incorporating about ten different elements in layers of few tens of nm.

More striking is the exceptional tolerance of CIGS to defects of various origins. Primo, one should keep in mind that CIGS is a non-stoichiometric compound, with deviations from stoichiometry in the % range: PV-grade material is generally obtained with a Cu content between 22 and 24%. Hence the main kind of defects are of the intrinsic nature, largely above the free carrier concentration. The material is strongly self-compensated, but in a way that do not harm its electronic properties. This is quite surprising as generally, large deviations from stoichiometry are a problem for electronic applications as (e.g. in WOx, CuxSe, In2O3−x, ZnO1−x etc.) the valence electron unbalance is compensated by free carriers, so that relatively small deviations (in the 1018 cm−3 range) already result in electronic degeneracy. In chalcopyrites though, electron unbalance from the 4 e per atom rule which are larger by as much as three orders of magnitude, i.e. in the 1021 cm−3 range) result in non-degenerate semiconductors. This implies that the formation of the compensating defects is energetically favoured as compared to formation of electronic carriers [9]. When this is the case, the Fermi level position is self-stabilized between a lower and upper value, as has indeed been found in many compound semiconductors [10], [11]. Evaluation of the concentration of point defects in CIGS lead to the conclusion that many of the cationic defects have concentration above 1018 cm−3 [4]. At those concentrations, they most likely form complexes, a fact also supported by ab-initio calculations [12]. Parallel to that ability to accommodate intrinsic defects, and certainly not unrelated, the electronic properties of the material appear less sensitive to impurities than is usually found in other semiconductors [4], [13]. Secundo, we note the excellent radiation hardness of this type of solar cells, as compared to traditional space cells (Si, InP, InGaP) [14], [15]. Other studies concluded that a very efficient, low temperature, defect recombination mechanism is needed to explain the exceptional radiation hardness [16], [17]. Tertio, in spite of the high density of grain boundaries and other crystallographic defects, the transport properties of the free carrier did not seem to be affected, at least in the transversal direction. Accordingly, very high efficiencies could be obtained. An explanation for this feature was found involving the passivation of grain boundaries in air for the p-type material [18], [19].

Surely, all this fortunate tolerance of the compound must be grounded in some of its particular properties. The most prominent one, from a solid state chemistry point of view, is its non-stoichiometry. Though the actual homogeneity domain of Cu(In,Ga)Se2 is not well known (see the discussion below), it is much larger than in the other semiconductors that have been successfully used for solar energy conversion (e.g. CdTe, Cu2xS). The existence of an electronic material with excellent electronic properties, compatible with high efficiency solar energy conversion, is quite puzzling in this context as most of the defects or defect complexes present must be inactive with respect to carrier recombination. Thus, their corresponding energy levels must either be shallow or altogether outside the bandgap. Again, ab-initio calculations on some defect complexes have shown that trend: defect levels of the considered complexes are be shallower than that of isolated point defects [12].

The second most important characteristic of CIGS is its mixed conductivity. Migration of Cu is well documented in crystals and in thin films [20], [21], [22], [23] The room temperature Cu diffusion coefficient in CuInSe2 (CIS) is in the range of 10−13–10−10 cm2/s [24]. Regardless of the exact values, this implies that Cu can diffuse across the space charge region (SCR) of the CIGS absorber in a few minutes to several days at most [25]. Such a rapidly migrating species potentially could raise serious concerns as to the long-term stability of CIGS-based devices.

Overall, these facts outline the picture of a compound with a ‘soft lattice’ certainly related to the non-stoichiometry of the material and to its ionic conductivity. It is not surprising then that metastable electronic centers were found in this material. The most studied metastable level in CIGS is very different from those encountered for instance in a:Si on two aspects at least: they lead to an improvement of the device characteristics under operating conditions and they are reversible [26], [27], [28]. The ‘soft lattice’ picture is central to the model proposed to account for the resilience and self-stabilization of CIGS [4], [25], [28]. It is because the material is highly disordered, but in a way that the most intrinsic defects complexes are not lifetime killers, and because Cu is mobile, that further defects (impurities, or radiation induced) can be self-passivated. This model uses the similarity between the point defects in a solid and solutes in an electrolyte, an analogy developed earlier [9], in that picture, the buffering action of the defect pool is essentially similar to the role of a buffer solution in aqueous electrolytes, where complexes in large concentrations (e.g. acids with a concentration such that the pH is close to their pKA) are used to fix the ionic concentration, e.g. the pH. In other words, the defect complex pool acts as a reservoir, capable of receiving or delivering electronic and ionic charge carriers so that their net amount remains fixed. Such defect reactions therefore act as both electronic and chemical buffers, by controlling both the electron concentration and the Cu concentration, respectively. In this model, a central role is given to mobile Cu as a vector of the buffering equilibrium. The detailed exposition of that mechanism being beyond the scope of this paper, the readers are referred to the original work [4], [25].

This paper will first elaborate further on the stability of CIGS from the thermodynamic point of view of CIGS and its connection with the interactions between point defects. Lastly, the stability in connection to the other partner compounds in the device will also be discussed, completing the results already presented [4].

Section snippets

Intrinsic stability of CIGS

Looking back at what has been achieved with CIGS, it is surprising to notice how little relatively is actually known about the compound. For instance, the extension of the homogeneity region has not been determined precisely, nor are the variations of most of the compound's properties with composition accurately known.

Interfacial stability

We now turn to the discussion of the stability of the full device, i.e. of the stability of CIGS in conjunction with the other compounds that are present in the device. Values of the standard free energies of formation of the relevant materials were collected from various sources [41], [42], [43], [44], [45], [46], [47], [48] and are given in Table 1.

In this section, instead of making an hypothesis on the exact value of the stabilisation energy, we have considered reactions with the constituent

Conclusion

In short, what makes CIGS so special among semiconductors used for opto-electronic applications? The thermodynamics and the solid-state chemistry of the system provide some answers: the unusual stability of CIGS comes from its ability to accommodate defects chemically and electrically (radiation-induced, impurities, dislocations) rather than from the difficulty to actually form defects. In that sense, CIGS is elastic rather than plastic, flexible rather than hard. This endows a new type of

Acknowledgements

This work originates in the joint effort carried on by L. Kronik and D. Cahen (Weizmann Inst.), R. Herberholtz, U. Rau and H.W. Schock (IPE), and the author, where seminal idea where first formulated. It also owes much to stimulating discussions with J. Vedel (ENSCP). It was carried out with the support of the EC under Joule contract (#JOR3-CT97-0149). The author is grateful to S. Cassaignon and E. Clolus for the electrochemical measurements.

References (65)

  • M.L. Fearheiley

    Sol. Cells

    (1986)
  • B. Eisener et al.

    J. Cryst. Growth

    (1999)
  • V. Lyahovitskaya et al.

    J. Cryst. Growth

    (1999)
  • M.A. Contreras et al.

    Sol. Energy Mater Sol. Cells

    (1997)
  • D. Cahen et al.

    J. Phys. Chem. Solids

    (1991)
  • C. Chatillon

    J. Cryst. Growth

    (1993)
  • H. Rau et al.

    State Chem.

    (1970)
  • J.B. Mooney et al.

    Sol. Cells

    (1986)
  • H. Neumann

    Sol. Cells

    (1986)
  • P. Manca et al.

    Sol. Cells

    (1986)
  • M.A. Contreras et al.

    Prog. Photovoltaic Res. Appl.

    (1999)
  • K. Emery, K. Zweibel, Measurements done on 3651 cm2 Siemens Solar modules, private...
  • K. Zweibel, Thin Solid Films (1999) in...
  • J.F. Guillemoles, L. Kronik, D. Cahen, U. Rau, A. Jasenek, H.W. Schock, J. Phys. Chem. B, (1999) submitted for...
  • B. Dimmler et al.

    2nd World Conf. Photovoltaic Energy Conv., Vienna

    (1998)
  • H.S. Ullal et al.

    26th IEEE Photovoltaic Spec. Conf., New York

    (1997)
  • J. Wennerberg et al.

    2nd World Conf. Photovoltaic Energy Conv., Vienna

    (1998)
  • D. Hariskos et al.

    Inst. Phys. Conf. Ser. E

    (1998)
  • F.A. Kroger

    The Chemistry of Imperfect Crystals, 1st Edition

    (1964)
  • W. Walukiewicz

    Phys. Rev. B

    (1988)
  • S.B. Zhang et al.

    J. Appl. Phys.

    (1998)
  • S.B. Zhang et al.

    Phys. Rev. B

    (1998)
  • B. Canava et al.
  • H.W. Schock et al.

    2nd World Conf. Photovoltaic Energy Conv., Vienna

    (1998)
  • T. Hisimatsu et al.

    2nd World Conf. Photovoltaic Energy Conv., Vienna

    (1998)
  • A. Polity et al.

    J. Appl. Phys.

    (1998)
  • M. Yakushev et al.

    Cryst. Res. Technol.

    (1996)
  • L. Kronik et al.

    2nd World Conf. Photovoltaic Energy Conv., Vienna

    (1998)
  • U. Rau et al.

    J. Appl. Phys.

    (1999)
  • K.D. Becker et al.

    Phys. Rev. B

    (1983)
  • K. Gartsman et al.

    J. Appl. Phys.

    (1997)
  • I. Lyubomirski et al.

    J. Appl. Phys.

    (1997)
  • Cited by (64)

    • The influence of relative humidity upon Cu(In,Ga)Se<inf>2</inf> thin-film surface chemistry: An X-ray photoelectron spectroscopy study

      2022, Applied Surface Science
      Citation Excerpt :

      Based on these two hypotheses, the following multistep mechanisms can be considered. By taking into account the phase diagram of CIGS [45], and starting from our CIGS layer, the withdrawal of In and Ga from the CIGS surface to form outermost indium and gallium oxides leads to an excess of Se in the compound and likely to the formation of Se0, as observed from 168 h to 864 h of ageing. Within the same time, a small amount of Cu(I) will react with the excess of Se to form binaries-based compounds of Cu(I) and Se (such as CuSe and Cu2Se) and/or Cu2O.

    • Thin film photovoltaic devices prepared with Cu<inf>3</inf>BiS<inf>3</inf> ternary compound

      2018, Materials Science in Semiconductor Processing
      Citation Excerpt :

      As observed on Fig. 2, when Cu was deposited on the material's surface, higher current density values were obtained compared with the values observed when Bi was deposited onto the surface. Although at this stage of this technology development, the possible reasons are unclear, we propose that this behavior can be attributed to the well-known fast Cu-out diffusion towards the surface, as has been reported for other Cu-based technologies [19]. Due to this fast Cu-out diffusion from the back towards the surface, the formation of small grains and voids at the rear region of the absorber is largely expected, as we will show later through cross-sectional SEM analysis of the corresponding layers.

    • Stability of Cu(In,Ga)Se<inf>2</inf> solar cells: A literature review

      2016, Solar Energy
      Citation Excerpt :

      Intrinsic instabilities of Cu(In,Ga)Se2, i.e. defect metastability, the mobility of copper as a constituent element and interface reactions between the layers of the final device, are potentially detrimental for the long-term stability of a CIGS solar cell. Guillemoles et al. (2000), Guillemoles (2000), however, argues that these fundamental instabilities are actually beneficial to the device: Possible interface reactions (back-contact/absorber, absorber/buffer, buffer/window) are thermodynamically or kinetically limited. In contrast to a-Si:H solar cells, defect metastabilities are self-annealed at room-temperature and are actually beneficial to the cell performance, since their respective energy states effectively increase the doping level in the CIGS absorber.

    • Photoluminescence study of the defect-induced recombination in Cu(In,Ga)Se<inf>2</inf> solar cell

      2013, Solar Energy
      Citation Excerpt :

      As a sort of important photovoltaic (PV) materials, CuInSe2 (CIS) or Cu(In,Ga)Se2 with the highest efficiencies reported to above 20% (EMPA, 2013) have been widely investigated for its thermal stability (Guillemoles, 2000), photo-induced deterioration (Haug et al., 2003; Bayhan and Bayhan, 2013), transport mechanisms under different conditions and the PV device functions (Bayhan and Bayhan, 2013), which are usually correlative to the improvement of phase structure within the bulk or interfaces of the multi-layers.

    View all citing articles on Scopus
    View full text