1. Introduction

Solar cells and modules based on polycrystalline Cu(In,Ga)Se₂ (CIGS) thin films are considered to be a leading candidate for low-cost power generation. The performance of high-efficiency CIGS solar cells is limited mainly by moderate open circuit voltage (V_oc). The V_oc can be increased by increasing band gap of the CIGS absorber, although this gives a corresponding photocurrent loss. Analysis and modeling of CIGS device has shown that the Shockley-Read-Hall recombination in space charge region of the absorber layer limits the open circuit voltage [1,2]. It has also been experimentally proved that an elevated defect density in the near surface region contributes to the recombination [3]. Numerical simulation conducted by Rau and Turcu has led to conclusion that the grading of the valence band close to the absorber surface provides a transport barrier for holes which reduces the density of holes at the interface and hence the interface recombination [4,5]. Therefore incorporating a surface layer of wide band gap, accommodated by a valence-band offset, is an effective way to increase V_oc without loss of photocurrent. Numerous researches on surface band gap engineering have been carried on. A grading structure of CIGS thin film has been achieved by varying Ga/In atomic ratio near the surface [6]. Grading the band gap of space charge region by surface sulfurization is more effective in comparison because more of the band gap increase is due to lowering of valence band in case of S incorporation. However, the sulfurization treatment using toxic H₂S gas [7,8] or difficultly controlled S vapor [9] is not preferred in industry application. Mokel et al. found an unusually large band gap at the free surfaces of Cu-poor grown CIS thin films [10]. Direct evidences of surface band gap widening in CIGS thin films have also been reported [11,12]. Those observations have led to the assumption that a Cu-depleted chalcopyrite phase segregation, the so called ordered vacancy compound (ODC), occurs at the surfaces of the films [13]. Those surfaces were considered type inversion, which leads to a shift of the regime p=n into the absorber and hence away from the defect rich CdS/CIGS interface resulting in a reduced recombination rate [4]. However, a high-quality ODC surface layer with conduction band alignment at CIGS/CdS interface [10] is difficult to be spontaneously obtained by using the regular three stage physical vapor deposition process especially in case of a relatively high process temperature or relatively low Se vapor pressure [14]. Commonly, a KCN solution is used to modify the surfaces of CIGS thin films to obtain Cu-depleted surface layers [15].

In the present study, we have proposed a simple method to fabricate the wide band gap surface layer by reconstructing the Cu-depleted absorber surface. A light controlling system is firstly used to monitor the surface layers of CIGS thin films. Effects of surface layer fabricated by our approach on properties of CIGS thin films and devices have been investigated in details.

2. Experiment

Polycrystalline CIGS thin films of 2μm in thickness were deposited onto Mo coated Na-free glass substrates using the three-stage co-evaporation process. In the first stage, the substrate temperature (T _sub) was kept at 350°C by reference thermocouples. During the second and the third stages, T _sub was normally ramped up to 520°C. Na was incorporated into the CIGS thin films by evaporating NaF, which has been demonstrated in our previous work [16]. After deposition of CIGS thin films, the substrate was cooled down and an In-Ga-Se (IGS) layer designed on the order of 10nm in thickness was then deposited onto the CIGS thin film at different temperatures. A light controlling system was utilized to precisely in situ monitor the thickness of IGS layer, which is of vital importance to our experiment. The monitoring method and results are demonstrated in details in the following. For simplicity, the IGS post deposition treatment will be referred as PDT. CIGS thin films with IGS layer of x nm post deposited at a certain T _sub, for instance 300°C, will be abbreviated in the form of IGS_300°C_x nm, and the sample without PDT is noted by IGS_0nm. The compositions of finished CIGS thin films were within the range of Cu/(In+Ga)=0.89±0.01 and Ga/(In+Ga)=0.39±0.01, which was determined by X-ray fluorescence (XRF) calibrated by an inductively coupled plasma spectroscopy. The cross sectional morphological properties were investigated by scanning electron microscopy (SEM) and the depth profiles were analyzed by auger electron spectroscopy (AES).

The CdS buffer layer was deposited onto the CIGS absorber by chemical bath deposition (CBD) technique using a CdI₂ (1.4×10⁻³M)-thiouea (0.14M)-ammonia (1M) aqueous solution. Approximately 20nm thick CdS layers were grown using a room temperature solution whose temperature rose to 80°C after one CBD run. Thick buffer layer were obtained by repeating this CBD process. CIGS solar cells were then completed by RF sputtering of i-ZnO/ZnO:Al front contacts, printing of Ag contact grids and mechanical scribing. There are 10×15 cells prepared on the substrate, and each one’s active area is 0.46 cm². No anti-reflection coating was applied. The current density–voltage (J-V) measurements were performed under the standard 1.5 AM spectrum with 100mW/cm² at 25°C. Average values of cells on the substrate were used for performance evaluation. Cells with very atypical characteristics were excluded from statistics.

3. Results and discusions

3.1 Monitoring details of PDT

Mono wavelength of 1100nm was used to monitor the thickness of IGS layer due to strong absorption of IGS material at visible spectrum. The monitoring glass was kept at 300°C to drive off Se. Reflective light monitoring mode was adopted in order to avoid contaminating photoelectric sensor by Se. Additionally, since refractive index of IGS thin film strongly depends on the Ga/(In+Ga) atomic ratio, deposition rates of In and Ga were regulated by crystal oscillator before PDT respectively based on compositional measurement by XRF and finally the Ga/(In+Ga) of IGS thin film was controlled around 0.4. Firstly, a thick IGS layer was deposited onto the monitoring glass. As shown in Fig. 1(a) , the light monitoring curve contains 18 peaks, and the peak width extends along with increase in deposition time. It is mainly because the anti-evaporation of Se on the chamber wall decreases the deposition rates, which may affect the Ga/(In+Ga) simultaneously as well. Figure 1(b) shows the optical constants of as deposited IGS layer which were determined by spectrometric measurements. The thickness of IGS layer calculated by software is 1437nm based on dielectric functions obtained from the transmission and reflection spectra. The thickness of IGS thin film measured by SEM is 1455nm as shown in Fig. 1(c), which basically agrees with the calculated result, thus the compositional ununiformity in depth could be negligible. After calibration by SEM result, the thickness of IGS layer corresponding to one peak in the light monitoring curve is around 80nm with errors within 1nm, suggesting a precise controlling for thickness of IGS layer. It is convenient to control the thickness of IGS layer for PDT at half peak positions and peak positions.

 

Fig. 1 (a) Monitoring curve of IGS layer. (b) The refractive index of IGS material with Ga/(In+Ga)~0.4. (c) cross-sectional SEM image of IGS layer and the thickness measured by SEM is 1.455μm.

Download Full Size | PDF

3.2 SEM results

Figure 2 shows the cross sectional and surface SEM image of CIGS thin films with PDT at different temperatures. It is observed that there is no clear difference on structural properties between IGS_400°C_80nm, IGS_500°C_80nm and IGS_0nm as shown in Figs. 2(a), 2(b) and 2(c). The post deposited IGS thin films are probably transformed into chalcopyrite phase due to Cu diffusion at relatively high T _sub above 400°C, thus a clear surface layer could not be observed. While the IGS_300°C_80nm shows fine grains on the surface and the IGS_300°C_320nm clearly exhibits a double-layer structure consisting of large (bottom) and small (upper) crystal grains, which implies poor lattice interdiffusion at T _sub of 300°C. According to the Cu-In-Se phase diagram [17], those small grains formed at 300°C should be mostly in In₂Se₃ or Ga₂Se₃ phase. Obviously, the small grains would bring more surface defect states and influence the formation of p-n junction during chemical bath deposition (CBD) [18]. Figure 2(f) shows the surface SEM image of IGS_500°C_320nm. The triangular-shaped grains suggest the (112) plane of a chalcopyrite structure, indicating the surface material of IGS_500°C_320nm is chalcopyrite phase.

 

Fig. 2 Cross-sectional SEM images of (a) IGS_0nm, (b) IGS_500°C_80nm, (c) IGS_400°C_80nm, (d) IGS_300°C_80nm, (e) IGS_300°C_320nm and (f) surface SEM image of IGS_500°C_320nm.

Download Full Size | PDF

3.3 Depth profiles analysis

Figure 3 shows the SIMS depth profile of IGS_0nm. It is observed that the upper region (close to front surface) of thin film presents higher Cu intensity, which is probably resulted by excess Cu_{2- x}Se formed on the surface during deposition stage 2. The Cu concentration distribution is not beneficial to the spontaneous formation of the Cu-poor surface layer during stage 3 since the In and Ga has higher activity in case of high Cu concentration.

 

Fig. 3 SIMS depth profile of IGS_0nm.

Download Full Size | PDF

In order to investigate the effects of PDT at different temperatures on Cu concentration in surface layer, AES depth measurements were performed on IGS_400°C_80nm, IGS_500°C_80nm and IGS_0nm. As can be seen in Fig. 4 , the Cu/(In+Ga) in the near surface region of IGS_0nm is around 0.7, whereas the Cu/(In+Ga) in surface layer of IGS_400°C_80nm decreases to around 0.4 which is close to the CIGS β phase (Cu(In,Ga)₃Se₅) in stoichiometry [13], suggesting a desirable Cu depleted surface layer. It is noted that Cu concentration near the surface of IGS_500°C_80nm remains high, revealing that the T _sub is a key factor for formation of Cu depleted surface layer. Although the Cu concentration in surface layer could be decreased by post-depositing a thicker IGS layer, the triangle-hole structure of Cu-poor chalcopyrite phase formed on the surface would lead to more surface defects as shown in Fig. 2(f). Therefore, the Cu concentration in the surface layer could be optimized by post depositing a IGS thin layer with proper thickness at different temperatures to obtain surface layer with large band gap, conduction band alignment at CIGS/CdS interface and fewer surface defect structures.

 

Fig. 4 The AES compositional depth profiles of IGS_0nm, IGS500°C_80nm and IGS400°C_80nm.

Download Full Size | PDF

3.4 Device performance

The photovoltaic performance parameters of 3 sets of CIGS devices with PDT at 300°C, 400°C and 500°C are summarized in Fig. 5 . As the thickness of IGS increases systematically, the IGS_400°C set cells are greatly improved in performance and reach the best performance at 80nm. In particular, the fill factor (FF) and V_oc of IGS_400°C_80nm are enhanced remarkably, confirming that the performance was improved by surface band gap widening [4]. As the thickness of IGS layer continues to increase, performance of IGS_400°C cells become deteriorated. It is mainly because of too low Cu concentration and increased defect states in surface layer due to excess IGS post deposited. However, the performance of IGS_400°C cells shows a benign tolerance to excess thickness of IGS layer. It is noted that the performance of IGS_400°C_160nm is still comparable with that of IGS_0nm. For IGS_500°C cells, the improvement of performance due to PDT is not clear. The AES result has shown that the wide band gap surface layer is not successfully fabricated by PDT at 500°C due to strong lattice diffusion [19], which agrees well with the device performance results. It is observed that the device parameters decrease after PDT at 300°C, which is probably due to the excess defect states brought by small grains on the surface as shown in Fig. 2(d). Additionally, CIGS thin film was surface sulfurized by annealing in S vapor at 520°C. Performance of CIGS solar cells with sulfurization treatment is shown for comparison with that of IGS_400°C_80nm in Fig. 5. The comparable efficiencies confirm that the PDT could be a successful alternative to surface sulfurization.

 

Fig. 5 Device parameters of CIGS solar cells with PDT at 300°C, 400°C, 500°C and solar cells with surface sulfurization treatment (maked with “★”).

Download Full Size | PDF

We have also applied PDT on a CIGS thin film with Cu/(In+Ga)=0.99. Generally, a high Cu/(In+Ga) composition ratio would lead to segregation of Cu_{2- x}Se phase on the surface, which results in low interface barriers and even shunts the solar cells. Dipping in the KCN solution is commonly known as an effective method to remove Cu from the surface of CIGS thin film [15]. However, it is not recommended due to high toxicity of KCN solution. In our experiment, the efficiency of high Cu/(In+Ga) ratio device is remarkably improved after PDT at 400°C as shown in Fig. 6 . AES measurements performed on the surface of the thin film show that the Cu/(In+Ga) of the surface decrease from 1.1 before PDT to 0.92 after PDT, which unveils that the exceeding Cu_{2- x}Se phase on the surface has been transformed into CIGS. Thus another effective method of eliminating excess Cu is suggested. It is of practical significance to deposit CIGS thin films with large area in industry, since the local Cu-rich might occur on the surface due to spatial ununiformmity of evaporation source or low Se pressure etc.

 

Fig. 6 J-V curves of CIGS device with PDT and without PDT. The Cu/(In+Ga) atomic ratio for the device is 0.99.

Download Full Size | PDF

4. Conclusion

The surface band gap of CIGS thin film has been optimized by varying Cu concentration in surface layer using PDT. Precise monitoring of thickness of IGS layer by light controlling method plays important roles in investigating the effects of PDT on CIGS thin films and devices quantitatively. The SEM results and depth profiles analysis suggest that both substrate temperature and thickness of IGS layer are important to fabricate high-quality Cu-depleted surface layer. The device parameters are greatly improved by optimizing the Cu concentration in surface layer, further more the Cu_2-xSe phase on the surface could be eliminated by PDT in case of high Cu/(In+Ga), which could not be done by other methods for increasing surface band gap such as Ga grading or surface sulfurization. Additionally, the PDT is easy to be implemented in CIGS deposition process. Thus a simple and effective method for grading surface band gap of CIGS thin films is suggested.