E616. The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells SHADIA J. IKHMAYIES

Transcription

1 E616 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells SHADIA J. IKHMAYIES ABSTRACT Solar cells based on thin films of polycrystalline materials are very promising in order to achieve better efficiency/cost ratios than the other counterparts. Among the thin film cells, CdTe based solar cell is the most promising candidate for photovoltaic energy conversion because of the high potentiality to realize low cost, high efficiency, reliable and stable solar cells. Thin film CdS/CdTe solar cells have an enormous potential as a commercially viable source of electricity in the near future. Due to its direct bandgap of 1.45 ev at room temperature, CdTe is able to absorb a significant fraction of the solar spectrum under AM1-AM1.5 conditions and so it is a promising candidate for terrestrial thin film photovoltaic manufacture. p-type CdTe is known to be used as an absorber in CdS/CdTe thin film solar cells, where n- type CdS works as the window layer. The absorber layer constitutes the core of the solar cell. So, optimizing the properties of this layer is a very important goal to get high performance solar cell. A lot of factors contribute in determining the properties of the CdTe layer in the CdS/CdTe thin film solar cell. The most important factors are the deposition technique, deposition temperature, sort of the substrate, impurities and dopants, CdCl 2 heat treatment, thickness and the back contact. In this chapter, a review of these factors and their influence on the properties of the CdTe layer and then on the performance of the CdS/CdTe thin film solar cell is performed. Key words: CdTe films, CdS/CdTe solar cell, CdCl 2 treatment, Dopants, Thickness, Back contact. Al Isra University, Faculty of Information Technology, Department of Basic Sciences- Physics, Amman (Jordan) Corresponding author : shadia_ikhmayies@yahoo.com

2 2 Energy Vol. 9: ============== 1. INTRODUCTION Polycrystalline thin film solar cells such as CdTe-based solar cells appear to be the most promising candidates for large scale application of photovoltaic energy conversion. The CdS/CdTe thin film solar cells can be prepared by simple deposition techniques and they have good physical and chemical stability (Antohe et al., 2012). The photoelectric conversion efficiency of small-area laboratory samples of CdS/CdTe solar cells reached 18.7% under AM1.5 solar radiation (Echendu et al., 2014; Paudel and Yan, 2013; Kosyachenko and Toyama, 2014). However, CdTe solar cells still need higher efficiency to lower the module cost and to increase the market share. There are several challenging points to improve the efficiency such as hightransparency window, low resistance contacts, low-defect density junction (Kim et al., 2010) and low interface and bulk recombination (Raadik et al., 2013). CdTe is an excellent absorber (Cousins, 2001) with large absorption coefficient (>10 4 cm -1 ) (Durose et al., 1999). It has a near-optimum direct bandgap of 1.45 ev at room temperature (Cousins, 2001; Gheluwe et al., 2005) which is suitable for efficient photo conversion. Therefore, an absorber layer of only a few micrometers thick is sufficient to absorb a maximum fraction of the AM1-AM1.5 solar spectrum (Potter et al., 2000a; Gheluwe et al., 2005). A 1ìm layer can absorb 99% of incident radiation in the visible range. It is, though, common to make the layers considerably thicker than this as homogeneity can be a problem for thinner layers (Cousins, 2001). CdTe can be grown to be n-type or p-type without extrinsic doping, simply by changing the composition (Echendu et al., 2014) and also it can be doped to be n- or p-type. Hence, it can form a monojunction (Durose et al., 1999). However, homojunction devices are impractical since most absorption of the solar spectrum occurs within 1-2 ìm of the CdTe surface. In addition, CdTe suffers from very high surface recombination velocities and this makes the surface recombination loss unacceptably high (Durose et al., 1999), which means that it is suitable only for use in a heterojunction solar cell. p-type CdTe commonly forms a p-n junction with the wide bandgap n- type compound semiconductor CdS, which forms the window layer in this heterojunction. CdS has a direct bandgap of 2.42 ev at room temperature and it has an acceptable lattice mismatch with CdTe (9.7%) (Antolin et al., 2013). In the superstrate configuration the CdS/CdTe thin film solar cell consists of a glass substrate, a transparent conducting oxide (TCO), the n- CdS window layer, the p-cdte absorber layer and the back contact. A schematic of the back contact/p-cdte/n-cds/tco/glass is shown in Fig. 1. The polycrystalline CdTe thin films can be deposited using a variety of different chemical and physical routes. The physical routes include thermal evaporation (Rimmaudo et al., 2013; Ikhmayies and Ahmad-Bitar, 2013; Gould and Bowler, 1988; Ikhmayies and Ahmad-Bitar, 2012), physical vapour

3 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells 3 Fig. 1: The superstrate configuration used for CdTe/CdS heterojunction solar cells (Reprinted with permission from Durose et al., 1999; Copyright 1999, Elsevier) deposition (PVD) (Cousins, 2001; McCandless and Birkmire, 2000), hot-wall vacuum evaporation (HWVE) (Seto et al., 2001), close-space sublimation (CSS) (Williams et al., 2014; Yun and Cha, 2014; Paudel and Yan, 2013; Halliday et al., 1998a; Halliday et al., 1998b;; Abulfotuh et al., 1997; Kosyachenko and Toyama, 2014), molecular beam epitaxy (MBE) (Oehling et al., 1996), electrodeposition (Echendu et al., 2014; Khan and Zhang, 1995), pulsed laser deposition (PLD) (Pandey et al., 2005; Neretina et al., 2007) and r.f. sputtering (Cousins, 2001; Paudel and Yan, 2013; Islam et al., 2013; Dolog et al., 2004; Lee, 2011). The chemical routes include chemical vapour deposition (CVD) (Yi et al., 1988), chemical bath deposition (CBD) (Ubale et al., 2006; Sotelo-Lerma, 2001), metal-organic chemical vapour deposition (MOCVD) (Zoppi et al., 2006; Proskuryakov et al., 2009; Hartley et al., 2001) and spray pyrolysis (Boone et al., 1982; Khan and Zhang, 1995). In this chapter a review of the main factors that determine the properties of the CdTe absorber layer in CdS/CdTe thin film solar cells is performed. These factors include the sort of the substrate, deposition temperature, deposition technique, impurities and dopants, the CdCl 2 heat tratment, thickness and back contact. All of these must be optimized to get high performance solar cell. 2. MICROSTRUCTURE OF THE CDTE LAYER The microstructure of the CdTe absorber layer has a crucial role in determining the performance of the solar cell. Large grains are desirable for both solar cell performance and stability, since the grain boundaries interfere with current transport, act as recombination centres and are pathways for diffusion (Zoppi et al., 2006). The deposition technique, deposition temperature and substrate nature are all important factors that contribute in determining the microstructure of the CdTe layer. A discussion on the influence of these factors will be presented in the following paragraphs. There are several experimental results which show that films produced by different deposition techniques have different microstructure. In their as-deposited form thin films of CdTe always show a columnar grain structure with submicron grain size unless the films are deposited by high temperature

4 4 Energy Vol. 9: ============== processes (above 500ºC) such as CSS, screen printing or spray pyrolysis (Birkmire and Eser, 1997). Vapour deposited CdTe thin-films have columnar grain structure. Zoppi et al. (2006) found that in the MOCVD films grown at 350ºC, the grains are smaller than those in CSS CdTe films: for films of 1 ìm thickness the average grain size for CSS material is < 1.2 ìm, while that for MOCVD material is < 0.15 ìm. According to them, this large difference in grain size with the two methods may be due to the low temperature of MOCVD growth (350ºC compared to ~500ºC for CSS). Absorber layers laid down by PVD are usually ~5 ìm thick with a grain size which is small compared to that of CSS material. Electrodeposition method of CdTe film deposition, in common with PVD growth produces grains of size in general about 0.1 ìm or lower (Cousins, 2001) and sputtering produces CdTe films with typically small grains (Islam et al., 2013). In the case of CdTe films deposited by high temperature processes, the grain structure is still columnar in that the grain boundaries are normal to the substrate, but the grain sizes are much larger, on the order of film thicknesses 2-15 ìm, depending on the specific process (Birkmire and Eser, 1997). Not only the grain size depends on the deposition technique, but also the relation between grain size and thickness and/or the other deposition parameters. This relation was found to differ from one deposition technique to another. Cousins (2001) found that in the MOCVD films grown at 350ºC, the grains increase in size with thickness at half the rate that they do in CSS material. The texture of the CdTe films depends strongly on the type of processes used and in some cases completely random films can be obtained. As a general rule lower is the process temperature higher is the preferred (111) orientation (Birkmire and Eser, 1997). Luschitz et al. (2009) found that deposited CdTe films tend to grow in columnar layers with (111) orientation, when PVD or CSS is applied at low sample temperatures, where the (111) texture relates to the fast growth direction of zinc blend phase. Cousins (2001) found that electrodeposition method of CdTe film deposition, in common with PVD growth, usually shows very strong (111) orientation. In general, vapour deposited films show a somewhat pronounced (111) orientation (Birkmire and Eser, 1997). For any given growth method, the influence of the deposition temperature on the microstructure of the films is confirmed. The deposition temperature affects the overall grain size, morphology, texture and preferential orientation of the CdTe layer. Fig. 2, which was obtained by Luschitz et al. (2009), depicts the SEM images of CdTe films as formed at different substrate temperatures. As shown the grain size increases with temperature. This increase may be related to the increased atomic nucleation mobility during the deposition.

5 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells 5 Fig. 2: SEM pictures with a 25000X magnification of CdTe films as formed at different substrate temperatures. Left: top view, right: side view. Compact and rather defect free grains are observed at elevated temperatures (520ºC) but also at the transition range between the first and second growth regime (around 340ºC) (Reprinted with permission from Luschitz et al., 2009; Copyright 2009, Elsevier). At low deposition temperatures CdTe films tend to grow in columnar layers with (111) orientation, which is favourable for current transport through the film, since charge transport proceeds without crossing grain boundaries. If the deposition temperature is strongly increased, which will only be practicable for CSS deposition, the (111) texture is lost and large three-dimensional grains of nearly statistical orientation are formed (Luschitz et al., 2009). The ordering (111) preferred orientation in the asdeposited state of CdTe deposited by PVD decreases with increasing substrate

6 6 Energy Vol. 9: ============== temperature, possibly due to the higher surface mobility of arriving species (Cousins, 2001). The effect of the substrate nature on the microstructure of the CdTe layer can be observed through the change in phase, preferential orientation and grain size with changing the kind of substrate. Several authors studied the properties of CdTe films deposited on different substrates. The phase and grain size are different for different kinds of substrates as shown by Islam et al. (2013) and Seto et al. (2001). Islam et al. (2013) found that CdTe films grown on glass have been found as cubic phase, but CdTe films on FTO/CdS stacks were found to be in hexagonal phase. Seto et al. (2001) produced CdTe films on three different substrates: glass, SnO 2 /glass and CdS/SnO 2 /glass. Fig. 3 depicts the SEM images of these films. As the figure shows, the surface morphology of these three CdTe films is different. The grain size of the CdTe/glass is uniformly about 1ìm and that of the CdTe/ SnO 2 /glass film is in the range 2-6 ìm. On the other hand, the CdTe/CdS/ SnO 2 /glass film is composed of larger grains of size 3-15 ìm and exhibits a close-packed structure compared to the other CdTe films. So, as found by Cousins (2001), when CdTe is deposited onto CdS the relationship between Fig. 3: SEM micrographs of CdTe films deposited on: (a) d7059 glass, (b) SnO 2 -coated glass, (c) CdS/SnO 2 /glass (Reprinted with permission from Seto et al., 2001; Copyright 2001, Elsevier). the morphology and microstructure of the CdTe layer and the underlying CdS is not simple. The CdTe is nucleated on the CdS grains (i.e., grain boundaries propagate across the interface). As the grain structure is predominantly columnar, the CdS grain structure has a strong influence on the surface structure of the CdTe films (Cousins, 2001). In addition, the strain in the CdTe layer changes according to the sort of the substrate, where it depends on the thermal expansion coefficients of the materials and the structure of the substrate. Cousins (2001) relates the compressive strain in the as-deposited PVD films either to the difference in the lattice parameters of CdTe and CdS, or to the differences in thermal expansion coefficients between the two compounds, or to a combination of both.

7 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells 7 Figure 4 which is obtained by Seto et al. (2001) shows the X-ray diffractograms of the three aforementioned CdTe films deposited on the three different substrates. In the three cases, the (111) peak is the most intense peak, which according to them means that the crystals grow up preferentially along the <111> direction. The diffractograms of CdTe/glass and CdTe/SnO 2 /glass films are similar to each other, where the weak (220) and (311) peaks are observed with almost the same intensity. In CdTe/CdS/ SnO 2 /glass film, there are two additional peaks, which are the (331) and (422). Fig. 4: X-ray diffraction patterns of CdTe films deposited on three kinds of substrates: (a) d7059 glass, (b) SnO 2 -coated glass, (c) CdS/SnO 2 /glass (Reprinted with permission from Seto et al., 2001; Copyright 2001, Elsevier). 3. IMPURITIES, DEFECTS AND DOPANTS Concentrations and distribution of dopants and impurities in the CdTe absorber layer are correlated with solar cell efficiency. As these solar cell structures are minority carrier devices, any defect or impurity which controls the minority lifetime can have a profound effect on the cell efficiency (Halliday and Potter, 2000). The predominant intrinsic defects in CdTe are cadmium interstials (Cd i ) and cadmium vacancies (V Cd ). Energy levels associated with these defects are 0.02 ev below the conduction band and 0.15 ev above the valance band, respectively (Birkmire and Eser, 1997). These native defects are known to be electrically active and the double acceptor V Cd may well play an active part in doping. In addition, interstitials, particularly Cd i, are thought to be important in compensation of p-doping (Proskuryakov et al., 2009). The nominal stoichiometry of the CdTe affects the cell parameters; efficiency, V OC and J sc which depend on the Te/Cd precursor ratio. But the large difference in the vapour pressure of Cd and Te makes it difficult to control the stochiometry (Birkmire and Eser, 1997). An important parameter of the CdTe absorber layer is the concentration of uncompensated acceptors N a -N d which determines the width of the space

8 8 Energy Vol. 9: ============== charge region (SCR). It is known that the efficiency of the device can significantly decrease owing to recombination in the SCR (Kosyachenko and Toyama, 2014). CdTe can be doped extrinsically in both n- and p-type form. Indium in Cd site (In Cd ) forms a donor level at 0.60 ev below the conduction band, whereas Cu, Ag, Au in Cd site (Cu Cd, Ag Cd and Au Cd ) form acceptor levels 0.33 ev above the valance band. Dopants that occupy tellurium sites in the crystal lattice such as S, O and Cl will aid cadmium vacancy formation (Halliday et al., 1998b). The solid solution CdS x Te 1-x which is a result of the interdiffusion at the CdS/CdTe interface is also considered one of the important dopants in addition to its role in the redistribution of the impurities. In addition, when the back contact of the solar cell contains copper, copper becomes one of the impurities not only in the CdTe layer, but also in the transparent conducting oxide and the CdS window layer. Other dopants are used by different authors such as arsenic with chlorine (Proskuryakov et al., 2009). A brief idea about the most important impurities and dopants of the CdTe layer in the CdS/CdTe thin film solar cells will be given in the following paragraphs: 3.1. Chlor Cl-doping is typically necessary in order to attain sufficiently high p-type doping densities (Williams et al., 2014). Post-growth doping of the CdTe with Cl is a result of in-diffusion of chlorine, usually from a vapour or evaporated layer of CdCl 2. The substitutional centre Cl Te (group VII on a group VI site) is a donor, it is considered that chlorine acts to form the complex [Cl Te -V Cd ] which acts over all as a single acceptor on account of the cadmium vacancy being a double acceptor (Proskuryakov et al., 2009). A change in the concentration of cadmium vacancy-chlorine defect complexes may be related to the change in the photovoltaic efficiency (Halliday et al., 1998b). Carriers which are released from this deep centre are more likely to recombine before reaching the junction (Halliday et al., 1998b). Indeed, the presence of Cl is known to enhance S diffusion into CdTe (Williams et al., 2014). High concentration of Cl introduced in CdTe weakens the influence of background impurities and defects, thus making the space-charge density more uniform (Kosyachenko and Toyama, 2014) Oxygen (O 2 ) It has been shown that incorporating a small amount of oxygen in both CdTe and CdS layers is very beneficial for CdTe solar cell performance (Paudel and Yan, 2013). Oxygen is known to be important in controlling the junction position and its incorporation in CdS has a particular effect in encouraging the formation of shallow junctions in the CdTe. It may outdiffuse from the CdS into the CdTe and hence dope it (Proskuryakov et al.,

9 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells ). There is evidence for incorporation of oxygen into the lattice as a substitutional (O Te ) impurity with a greater concentration at the CdTe/CdS interface rather than the surface (Halliday et al., 1998b). It is also suggested that the incorporation of oxygen may enhance the p-type doping of CdTe and improve the cell efficiency (Paudel and Yan, 2013). The contribution of oxygen to doping occurs by acting as a sink for Cd atoms (Proskuryakov et al., 2009) and it forms a shallow acceptor (Cousins, 2001). In addition, Paudel and Yan (2013) found that the introduction of oxygen during CSS CdTe controls the grain size and leads to the growth of dense films. The influence of oxygen on the CdTe layer can be observed in Fig. 5, which was recorded by Hernández-Fenollosa et al. (2003) and it displays the front PL spectra (through the glass) of a CdS/CdTe solar cell treated at 390ºC during 20 min with 0.5 mbar HCl in a varying partial pressure of oxygen. According to these authors, band (a) in the figure is presumed to be due to Cl Te -V Cd. Because the intensity of this band always increases with the pressure of oxygen p[o 2 ], it is inferred that the presence of oxygen during processing increased the concentration of Cl Te and V Cd centers. This interpretation is consistent with the large enhancement of this band seen for p[o 2 ] = 20% compared to the oxygen-free case. Band (b) in Fig. 5 is presumed to be related to oxygen in CdTe and can be TeO 2 or V Te -related. This assignment is compatible with the observed increase of the intensity of band (b) with increasing p[o 2 ] (Hernández-Fenollosa et al., 2003). The broad band (c) ( ev) is due to Te complexes in CdS. The presence of oxygen increases the intensity of this band above that seen for oxygen-free processing. Perhaps oxygen enhances Te diffusion into CdS. Band (d) (red band) is due to cadmium vacancy-sulphur vacancy complex V Cd -V S (Hernández-Fenollosa et al., 2003). Fig. 5: Front surface PL as a function of the oxygen partial pressure used in processing. The spectra were taken from a normal part of the cell (Reprinted with permission from Hernández-Fenollosa et al., 2003; Copyright 2003, Elsevier).

10 10 Energy Vol. 9: ============== 3.3. The CdS x Te 1-x Solid Solution This alloy is a result of the interdiffusion of sulphur from the CdS into the CdTe layer at the interface and it has a lower bandgap than CdTe for low sulphur concentrations (Halliday and Potter, 2000). It is implicated in the doping of the CdTe (Proskuryakov et al., 2009) and it affects the impurity distribution (Halliday et al., 1998b) in this layer. Durose et al. (1999) noted that the degree of interdiffusion at the CdTe/CdS interface was maximized under conditions of excess Te. It is well known that the interdiffusion between CdS and CdTe at the interface has a significant role at grain boundaries and it leads to the formation of graded junctions instead of abrupt ones at the CdS/CdTe interface making the bandgap of the device to gradually vary from 2.42 ev to 1.48 ev (Echendu et al., 2014). So, the formation of this solid solution at the CdS/CdTe interface is crucial for the improvement of solar cell performance Copper (Cu) Copper is a well-known substitutional acceptor, Cu Cd (i.e., a group I on a group II site) (Proskuryakov et al., 2009). In most cases copper is used in making the back contact of the CdS/CdTe solar cell and copper is thermally diffused. According to Kosyachenko and Toyama (2014), copper can cause dramatic changes in the electrical properties of the CdTe film, from an insulating to a highly conductive state. But this depends on the conditions of thermal treatment Indium (In) To increase the conductivity of the CdS window layer, doping with indium was found to be effective (Ikhmayies and Ahmad-Bitar, 2010). Indium can diffuse to the CdTe absorber layer at the CdS/CdTe interface and go deeply in the material. In this case it replaces cadmium (In Cd ) and forms a donor level at 0.60 ev below the conduction band Gold (Au) When gold is used as a back contact material for the CdS/CdTe solar cell its diffusion inside the CdTe layer is expected. Gold acts as an acceptor, where it substitutes Cd (Au Cd ) and forms an acceptor level at 0.33 ev above the valence band Arsenic (As) with Chlorine (Cl) Proskuryakov et al. (2009) used co-doped CdTe:As,Cl in CdTe/CdS solar cell devices that have been grown by in-situ MOCVD method. They concluded

11 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells 11 that incorporation of As takes place at low concentrations by occupancy of sites which reduce native p-conductivity (e.g., by blocking V Cd ), while at higher concentrations, As is incorporated as an acceptor, probably substitutionally as As Te Sodium (Na) In the superstrate configuration, the presence of sodium in the CdTe layer is attributed mainly to the diffusion from the glass substrate if it is soda lime glass. Sodium is known as an acceptor impurity in CdTe, where it is expected to substitute Cd in the crystal lattice. Because of the enhanced compensation and segregation effects at the grain boundaries in polycrystalline thin films such as CdTe, doping becomes even more difficult. Difficulties encountered in the doping of CdTe do not affect its photovoltaic performance, but they do create problems in making low-resistance ohmic contact to the material. In fact, because of these characteristics of the doping mechanism, the grain boundaries in polycrystalline CdTe thin films can be made more p-type than the bulk (similar to the CuInSe 2 -based solar cells), reducing and even eliminating the minority carrier (i.e., electron) recombination at the grain boundaries (Birkmire and Eser, 1997). 4. CDCL 2 HEAT TREATMENT A vital step in the manufacture of CdTe/CdS cells is annealing with CdCl 2 (Halliday and Potter, 2000). That is, the majority of CdS/CdTe solar cells and all those with notable conversion efficiencies were subjected to a postgrowth treatment of the CdTe layer. In the as-grown state, the cells have a typical efficiency of 1-2% and it is known that a post-deposition annealing in the presence of CdCl 2 increases the efficiency by an order of magnitude (Hernández-Fenollosa et al., 2003; Durose et al., 1999). The increase in efficiency is due to the improvement of the microstructure of the CdTe layer and then in the open circuit voltage (V OC ), short circuit current (J sc ) (Paudel and Yan, 2013; Durose et al., 1999) and the fill factor (FF) (Paudel and Yan, 2013). The CdCl 2 heat treatment is a complex process which not only increases efficiency, but also affects the chemical and structural properties of the layers (Halliday and Potter, 2000). Variations of the process are many, but this so called activation or type conversion process usually involves annealing in the presence of CdCl 2 (Durose et al., 1999). The CdCl 2 can be deposited onto the CdTe at a post-deposition stage (usually by evaporation or dipping in a methanolic solution) and this step is usually followed by annealing in air or in vacuum (Hernández-Fenollosa et al., 2003; Durose et al., 1999) or in argon atmosphere. Annealing atmosphere, annealing temperature and annealing time all have effective influence on the solar cell performance.

12 12 Energy Vol. 9: ============== The presence of oxygen during the CdCl 2 heat treatment has beneficial effects on cell performance. It can give slightly higher efficiencies than from the simple chloride treatment (Hernández-Fenollosa et al., 2003). As mentioned in the previous section, oxygen has many benefits for the CdS/ CdTe solar cell. These are confirmed by Hernández-Fenollosa et al. (2003), where they say that oxygenation (i) enhances concentrations of Cl Te and V Cd in CdTe, (ii) enhances diffusion of Te into CdS with highly oxidizing conditions yielding Te-O complexes and (iii) can act to fill V S with O S. The influence of annealing temperature on the microstructure of the CdTe layer can be observed in Fig. 6 and Fig. 7 obtained by Lee (2011) for CdTe films with evaporated CdCl 2 layer. Fig. 6 shows the change in morphology and grain size as the annealing temperature increases and Fig. 7 shows the relations between the CdS/CdTe solar cell parameters: the open-circuit voltage (V oc ), short-circuit current density (J sc ) and fill factor (FF) with annealing temperature. According to Lee (2011), the characteristics of CdS/CdTe solar cells appear to be significantly influenced by the annealing temperature of CdTe films and the improvement in cell performance may be due to increasing grain size and improving crystallinity of CdTe films. However, the efficiency of cells annealed at temperatures higher than 425ºC decreased. Fig. 6: SEM micrographs of CdTe films annealed at different temperatures with CdCl 2 layer: (a) as-deposited, (b) T a : 375ºC, (c) T a : 400ºC, (d) T a : 500ºC (Reprinted with permission from Lee, 2011; Copyright 2011, Elsevier).

13 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells 13 Fig. 7: Influence of annealing temperature of CdTe films with evaporated CdCl 2 layer on the CdS/CdTe solar cell parameters (Reprinted with permission from Lee, 2011; Copyright 2011, Elsevier). The influence of annealing time was investigated by Halliday and Potter (2000), where they found that the efficiency of the CdS/CdTe solar cell peaks at short annealing times and drops off at long anneal times. In addition, Potter et al. (2000a) and Potter et al. (2000b) studied the influence of annealing time and found that a short (5-20 minute) anneal increases both V oc and J sc. The greater effect they got is the increase in J sc, possibly through a reduction in the resistance of the cell. Longer anneal times reduce the efficiency mainly by reducing V oc. According to Potter et al. (2000b), a more detailed study of the individual spectra for the samples with long anneal times reveals evidence of phase segregation (the formation of discrete regions of CdS within the CdTe) occurring after excess annealing. Also it is possible that the entire CdS layer has diffused into the CdTe after long anneal, which would produce homogeneous layer and could explain the reduced efficiency at long anneal times (Halliday and Potter, 2000). Numerous mechanisms have been proposed to explain the efficiency increase caused by the CdCl 2 anneal. These mechanisms include conductivity type conversion, diffusion enhancement, grain boundary passivation, the increase of CdTe crystallite grains (Zoppi et al., 2012) and recrystallization (Potter et al., 2000a), strain recovery, changing the texture and preferred orientation of the films, changing the mechanism of current transport,

14 14 Energy Vol. 9: ============== reduction of the recombination in the devices and the decrease of defect states at the interface CdS/CdTe (Zoppi et al., 2006). These effects may act singly or in combination and they will be briefly discussed in the following paragraphs Conductivity Type Conversion As mentioned before in this section, the as-deposited CdTe/CdS cells are inefficient in their action as solar cells and it is widely reported that the CdCl 2 heat treatment enhances performance by introducing p-type dopant centres into CdTe. This is generally a type conversion since this treatment affects the conversion of CdTe from n-type to p-type (Durose et al., 1999; Hartley et al., 2001). This is consistent with the tendency of CdTe to selfcompensate, making it difficult to dope p-type (Cousins, 2001). The possible explanations for this type of conversion are (i) the (V Cd + Cl Te ) + complex (Cousins, 2001), (ii) the shallow-acceptor-related centre, involving two chlorine atoms and a cadmium vacancy (V Cd -2Cl Te ) and (iii) a substitutional oxygen atom O Te. This type of conversion causes the CdTe material to become p-type and then results in the increase in the cell efficiency, electric field and V OC (Halliday et al., 1998a). The anneal following CdCl 2 deposition produces a redistribution of impurities, but air anneal does not significantly alter the impurity distribution (Halliday et al., 1998b). There is evidence to suggest the diffusion of both Cd and Cl species and also oxygen into the CdTe, all of which can have a profound effect on the defect chemistry of the system. Oxygen has been found to be a necessary adjunct to the process, which must take place in air (Hartley et al., 2001). In addition the CdCl 2 heat treatment alters the distribution of V Cd -Cl Te complexes which may allow increased recombination away from the interface (Halliday et al., 1998b) Interdiffusion Enhancement at the CdS/CdTe Interface A further effect of the CdCl 2 heat treatment is to enhance interdiffusion of the CdTe and CdS layers (Paudel and Yan, 2013; Durose et al., 1999), where CdS x Te 1-x forms on the CdTe side and CdTe y S 1-y forms on the CdS side. This intermixing at the CdS/CdTe interface is reported to have a number of effects: Interdiffusion reduces lattice mismatch at the CdS/CdTe junction and hence it reduces the interfacial recombination centres. Bowing of the bandgap with composition. Bandgap bowing with composition according to Pal et al. (1993) is represented by: E(x) = E 2 + (E 1 E 2 b)x + bx 2 (1)

15 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells 15 where E 1 and E 2 are the bandgaps of CdS and CdTe films, respectively, x is the compositional factor of the CdTe 1-x S x alloy and b is the bowing parameter. The value of b they obtained from experimental data is 1.7 ev, which they compared with that obtained theoretically (b = 2.0 ev). Thus, some compositions of CdS 1-x Te x have a lower bandgap than CdTe and will as such cause an increase in the optical absorption in the interface region (Proskuryakov et al., 2009). The relieve of the in-plane stresses, where these stresses are caused by lattice mismatch between CdS window layer and the CdTe absorber layer and relieving them improves V oc of the solar cell. CdCl 2 heat treatment prompts S diffusion into the CdTe layer, which reduces the strain at the junction. It may result in the entire consumption of the CdS window layer and hence degrading the performance of the solar cell. This area of research is still most active (Cousins, 2001). This interdiffusion is considered to be responsible for the displacement of the electrical p-n junction away from the metallurgical interface and into the CdTe (Durose et al., 1999). Interdiffusion at the CdTe/CdS interface is important because of its influence on the optical absorption in the structure due to bandgap changes. The effect of the interdiffusion of CdTe into CdS is to curtail the spectral response at its low end by reducing the window transmission. On the other hand incorporation of some CdS into the CdTe extends the spectral response to longer wavelengths (Durose et al., 1999). It is suspected that the interdiffusion may be responsible for important electrical changes undergone by the cell during CdCl 2 processing. For example, interdiffusion is associated with a decrease in the diode ideality factor, indicating a reduction in the interface state density (Durose et al., 1999) Grain Boundary Passivation Grain boundaries in CdTe thin films are generally considered to act as strong electrical recombination centres, due to the introduction of deep energy levels within the bandgap resulting from such defects. Grain boundaries may also act as barriers to current transport, or to cause significant leakage currents. It is therefore advantageous to device performance to minimize the impact of grain boundaries by maximizing the grain size in thin films (Lee, 2011). There are different explanations for the reasons of this grain boundary passivation after the CdCl 2 heat treatment: (i) the presence of an oxygenrelated centre such as O Te, which passivates the grain boundaries and helps

16 16 Energy Vol. 9: ============== in increasing the efficiency of CdS/CdTe solar cells (Halliday et al., 1998a); (ii) the CdCl 2 anneal causes sulphur diffusion into the CdTe, which passivates the CdTe grain boundaries (Potter et al., 2000a; Halliday and Potter, 2000); (iii) the CdCl 2 passivates the defect states at the grain boundaries near the interface (Potter et al., 2000a) and lowers the concentration of structural defects (Halliday et al., 1998b), so the number of defects at the interface of the CdS/CdTe decreases. Reduction of the density of defects is related to the enhancement of interdiffusion at the CdS/CdTe interface mentioned before in this section. Whilst excessive interdiffusion is detrimental to device performance (causing CdS consumption), a certain level of interdiffusion is thought to assist in the passivation of surface defects at the interface and therefore lead to reduced interface recombination; (iv) the upward band bending in the CdTe near to the grain boundaries after treatment leads to the repulsion of carriers, where this is proved by electron beam induced current (EBIC) measurements and impedance spectroscopy (Halliday and Potter, 2000); (v) the incorporation of impurities in grain boundaries (Halliday and Potter, 2000); (vi) the CdCl 2 heat treatment prompts Cl diffusion through grain boundaries and passivates them (Paudel and Yan, 2013). CdTe has a large surface recombination velocity at grain boundaries, free surfaces and interfaces (Halliday and Potter, 2000). Passivation of the grain boundaries by CdCl 2 heat treatment results in the passivation of nonradiative recombination centres. This can occur by different ways: (i) O Te sites which, as mentioned before, cause passivation of the grain boundaries thus reduce the non-radiative recombination of photogenerated carriers (Halliday et al., 1998b). Referring to Fig. 5 which was recorded by Hernández- Fenollosa et al. (2003) for the PL spectra of the front surface of CdS/CdTe, band (a) in Fig. 5 is presumed to be due to Cl Te -V Cd, where the presence of oxygen during processing increased the concentration of Cl Te and V Cd centres. The Cl Te -V Cd is expected to be a recombination site strongly coupled to the lattice. The coupling to the lattice is reduced after the CdCl 2 treatment possibly due to a reduction in the lattice distortion associated with this defect complex (Halliday et al., 1998a). (ii) As mentioned before, the CdCl 2 promotes S diffusion near the interface, causing grain boundary passivation, thus increasing the relative luminescence efficiency by reducing possible non radiative recombination centres (Potter et al., 2000b) Grain Growth and Recrystallization It is recognized that CdCl 2 heat treatment promotes grain growth and recrystallisation in the CdTe layer, but this depends on the growth temperature as well as the types of the substrates. Larger columnar grains grow in the plane of the film at the expense of smaller ones, thus increasing the average size. This process is driven by the minimization of grain boundary area and corresponding reduction in grain boundary energy (Cousins, 2001). A simple grain growth process creates a more uniform cell with fewer short

17 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells 17 circuits (Halliday et al., 1998a). Recrystallization is well known to accompany the CdCl 2 treatment. It was considered to be nucleated at the free surface with small new grains appearing at the grain interstices. Growth of these with random orientation was considered to affect the recrystallization itself (Durose et al., 1999). Figure 8 displays the AFM images obtained by Salavei et al. (2013) for two CdTe films of thickness 1.8 and 6 ìm deposited by thermal evaporation on CdS/ZnO/ITO stacks. The images of the thinner CdTe film show homogeneous morphology with similar but smaller grains; while the thicker one shows grains with sizes higher by order of magnitudes, but at the same time show a wide range of small and big grains. However, while the CdCl 2 heat treatment brought an overall increase in the grain size of up to one order of magnitude in the thin CdTe occurred. The enlargement it brought in the thick CdTe film is only from two to three times the initial size. Fig. 8: AFM pictures of 1.8 and 6 ìm as-deposited CdTe (top) and of 1.8 and 6 ìm treated CdTe (bottom) (Reprinted with permission from Salavei et al., 2013; Copyright 2013, Elsevier) Strain Recovery This strain is caused by the lattice mismatch between the CdTe and the underlying CdS film and also the difference in their thermal expansion coefficients. S diffusion into CdTe which is enhanced by CdCl 2 heat treatment will reduce strain at the junction (Halliday and Potter, 2000). Different authors (Zoppi et al., 2006; Islam et al., 2013) related the decrease in the material strain following this treatment with the reduction in the lattice parameter. Moutinho et al. (1998) measured the lattice parameter for CdCl 2 - annealed CdTe films under different annealing temperatures and found them larger than for a powder sample (a = Å), suggesting that the film is subjected to compressive stress in the plane parallel to the substrate surface. However, there was a clear reduction in a following a min

18 18 Energy Vol. 9: ============== heat treatment for all temperatures. When annealing was sustained beyond 20 min, the lattice parameter was nearly invariant for treatment temperatures below 500ºC. Treatment at this temperature induced greater variations in lattice parameter, with a decreasing further with longer annealing times. In addition, Islam et al. (2013) observed that for the planes corresponding to reflections in Fig. 9, microstrain values and dislocation densities are higher for untreated CdTe thin films. For the untreated samples deposited on glass A1 and B1 at substrate temperatures 250 and 300ºC, the microstrain values are å = and respectively, but for the CdCl 2 treated samples on glass at the same temperatures A2 and B2, the microstrain values are å = and respectively. For the samples deposited on FTO/CdS stacks at substrate temperatures 250 and 300ºC the microstrain values are å = and respectively before treatment and and respectively after CdCl 2 heat treatment Phase and Texture Figure 9, which was obtained by Islam et al. (2013), shows the XRD patterns of CdTe films deposited on glass and FTO/CdS stacks with and without CdCl 2 treatment. The films deposited on glass have been found as cubic phase but on FTO/CdS stacks as a hexagonal phase. In addition, CdCl 2 heat treatment causes rearrangement of the preferred orientation. This can only happen by (a) the formation of new grains which replace the old ones (recrystallization), or (b) the growth of some grains at the expense of others Fig. 9: XRD spectra for 1.0 ìm thick CdTe thin films; (a) deposited on top of glass substrate and (b) deposited on top of glass/fto/cds stacks at temperature 250ºC and 300ºC, respectively. A1, B1, C1 and D1 are untreated, but A2, B2, C2 and D2 are CdCl 2 treated (Reprinted with permission from Islam et al., 2013; Copyright 2013, Elsevier).

19 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells 19 (grain growth) (Zoppi et al., 2006). It is clear that the treated films possess better crystallinity of CdTe films deposited either on glass or FTO/CdS stacks. The diffraction peaks observed for all the films deposited on glass at around 2è = 23.76º, 38.16º and 46.49º are corresponding to (111), (220) and (311) planes of CdTe thin films, respectively, as confirmed by JCPDS X-ray powder file data ( ). On the other hand, the peaks observed for the CdTe films deposited on top of FTO/CdS stacks are around at 2è = 23.87º, 25.44º, 32.89º, 39.48º, 42.80º and 46.98º, corresponding to (002), (101), (102), (110), (103) and (201) planes as confirmed by JCPDS data ( ). By calculating the texture coefficient, Islam et al. (2013) found that the samples grown at 250ºC either on glass or FTO/CdS stacks are not affected significantly by CdCl 2 treatment. However, the films grown at 300ºC are largely affected. All the samples deposited on glass maintained a random orientation. However, the films deposited at 300ºC on FTO/CdS stacks are greatly affected by CdCl 2 treatment on the orientation of the films, for which the films have been found in the preferential (002) plane oriented. The highly preferential CdTe films are usually usable for high efficiency solar cell fabrication. All the other films are found in random orientation may be due to the thickness of the films, which are usually very thin (1 ìm) as it is reported that the film quality and crystallinity are increased with the increase of thickness (Islam et al., 2013) Changing Current Transport Mechanism This treatment is accompanied by a change in current transport mechanism from tunnelling/interface recombination to recombination in the depletion region (Durose et al., 1999). This suggests a decrease in the density of interfacial states. Beside these effects, CdCl 2 heat treatment decreases the inhomogeneity across the cell, lowers the series resistance (Durose et al., 1999), smoothes the CdTe surface, results in the formation of deep crevasses at the grain boundaries and in the case of CdS/CdTe thin film solar cells used in space applications, their stability against the ionizing radiations will be improved after CdCl 2 treatment (Antohe et al., 2012). All these effects would help to improve the minority carrier lifetime and consequently improve the performance of the solar cells (Paudel and Yan, 2013). 5. THICKNESS OF THE CDTE LAYER As mentioned in section 1, CdTe has a high absorption coefficient á >10 4 cm 1, which means that all the incident photons with energy greater than the bandgap will be absorbed within the first few microns of CdTe absorber layer. CdTe has a direct optical bandgap of 1.45 ev at room temperature which is very close to the optimum bandgap for solar cells. Hence, the

20 20 Energy Vol. 9: ============== thickness required for an absorption layer of CdTe cells makes the cost of material relatively low. Clearly one of the main goals of today s solar cell research is using less semiconductor material by making the cells thinner. Thinning will not only save material, but will also reduce the recombination loss as well as lower production time and the energy needed to produce them. All of these factors will decrease the production cost (Sharafat Hossain et al., 2011). But there are many factors that determine the lower and upper limits of the thickness of the CdTe absorber layer. Very thin CdTe layer suffers from lower crystallization, different possible effects on material diffusion within the interfaces (Salavei et al., 2013), extension of the depletion region to the back contact and also may suffer from micro shorts due to the presence of pinholes, which results in low values of V OC and FF. On the other hand, very thick CdTe layer may suffer from large series resistance R s. Mykytyuk et al. (2012) found that when the CdTe absorber layer is very thin, it is impossible to avoid a noticeable decrease of the short circuit current density J sc as compared with a typical thickness of the absorber layer. The loss in J sc is 19-20% when the thickness is 0.5 ìm compared to 5% for a thickness of 2-3 ìm (Mykytyuk et al., 2012). They also showed that the absorptivity of solar radiation power increases with increasing the thickness of CdTe. The absorptivity in CdTe with thickness of 1 ìm is about 97% and H % at thickness of 0.5 ìm. On the other hand Sharafat Hossain et al. (2011) and Islam et al. (2013) found that when the CdTe absorber layer was decreased to the extreme limit of 1 ìm the proposed cell has shown acceptable range of efficiency at this thickness. Maintaining high efficiency in cells with CdTe thickness <1 ìm will require overcoming losses in junction quality and understanding the influence of the back contact. The restrictions on the thickness of CdTe in CdS/CdTe heterojunction must be ascertained taking into account all types of losses. For the substrate CdTe/CdS solar cells, Williams et al. (2014) found that the CdCl 2 treated devices with the thickest CdTe (10 ìm) suffered from lower FF (30-35%) for all annealing temperatures. Indeed, for the series of devices annealed at 580ºC, the series resistance R S increased with CdTe thickness. They attributed this result to the higher probability that there exist grain boundaries perpendicular to current flow in thicker CdTe films, or simply due to the intrinsic resistance of the CdTe itself. Also, when they used a lower annealing temperature of 540ºC, the R S for thinner CdTe films (3-6 ìm) was equivalent to that for the 10 ìm CdTe films that had been annealed at 580ºC. According to them (Williams et al., 2014), this implies that high R S is attributed to CdTe films being undertreated. Salavei et al. (2013) recorded that for solar cells with 1 and 1.8 ìm CdTe layer thickness, lower efficiencies are connected to the low current density due to the extension of the depletion region to the back contact.

21 The CdTe Absorber Layer In CdS/CdTe Thin Film Solar Cells 21 From the review of the literature, it is found that CdTe solar cells are typically produced with thicknesses between 4 and 6 ìm. This assures good coverage avoiding pin-holes and delivering high open circuit voltage devices. On the other hand, because of the high absorption coefficient, 2 ìm is enough for absorbing all the light and converting sunlight into electricity, but in this case device performance is lower and the structural and electrical properties of thin absorbers are different (Salavei et al., 2013). This is confirmed by Durose et al. (1999) who recorded that the CdTe need only be 1-2 ìm thick, but may be thicker to ensure homogeneity. Williams et al. (2014) restricted the best thickness of this layer in the range 3-6 ìm. The thickness of the CdTe layer in the CdS/CdTe thin film solar cells has considerable influence on the efficiency through affecting texture, grain size and energy losses. A clear correlation between cell efficiency and the texture of the CdTe film is observed by Luschitz et al. (2009) and for the asgrown samples there is a relationship between preferred orientation and the layer thickness (Zoppi et al., 2006). Thin layers are [111] oriented as shown by different authors such as Zoppi et al. (2006) and Salavei et al. (2013). As the layers thicken, the texture becomes more randomized. In polycrystalline films, the increase in film thickness results in the increase of grain size. Or in other words, thicker films have larger grains than thinner ones. Zoppi et al. (2006) found that for the as-grown CdTe films, the increase of crystallite size with the layer thickness can be described by a simple power law of the form: r(x) = kx y + c (2) where y H 0.5 for CdTe films prepared by CSS and about 1.5 for MOCVD samples and it is assumed that the line passes close to the origin. Hence a superlinear relation is obtained for the MOCVD samples and a sublinear (square root behaviour) relation for the CSS material. It might be inferred that grain size development during MOCVD growth is very different from that during CSS. These differences might tentatively be ascribed to the growth temperature which is 350ºC for MOCVD samples and 500ºC for the CSS samples. 6. THE BACK CONTACT The back contact is an important issue that strongly affects both efficiency and stability of the CdS/CdTe solar cell. The application of a back contact to the CdTe absorber layer can influence the junction region even if the absorber is 10 ìm thick. All contacts to p-cdte have a barrier which has the opposite sense to that of the p-n junction. This is because CdTe has a high electron affinity ( = 4.5 ev) (Jarkov et al., 2013; Durose et al., 1999) and no metal exists with a high enough work function (W m ) to form an ohmic contact