An alternative non‐vacuum and low cost ESAVD method for the deposition of Cu(In,Ga)Se2 absorber layers

In this article, an environmentally friendly and non‐vacuum electrostatic spray assisted vapor deposition (ESAVD) process has been developed as an alternative and low cost method to deposit CIGS absorber layers. ESAVD is a non‐vacuum chemical vapor deposition based process whereby a mixture of chemical precursors is atomized to form aerosol. The aerosol is charged and directed towards a heated substrate where it would undergo decomposition and chemical reaction to deposit a stable solid film onto the substrate. A sol containing copper, indium, and gallium salts, as well as thiourea was formulated into a homogeneous chemical precursor mixture for the deposition of CIGS films. After selenization, both XRD and Raman results show the presence of the characteristic peaks of CIGSSe in the fabricated thin films. From SEM images and XRF results, it can be seen that the deposited absorbers are promising for good performance solar cells. The fabricated solar cell with a typical structure of glass/Mo/CIGSSe/CdS/i‐ZnO/ITO shows efficiency of 2.82% under 100 mW cm−2 AM1.5 illumination.

1 Introduction Chalcopyrite CuInGaSe 2 (CIGS) thin film solar cells processed by co-evaporation method have reached above 20% power conversion efficiency [1][2][3][4], and it is one of the most promising commercial solar modules. In order to make the CIGS solar industry competitive and sustainable in the long-term, CIGS based photovoltaic technology using low cost and non-vacuum processes such as electrodeposition [5,6], hydrazine [7,8], quantum dots [9,10], and other wet chemical precursor methods involving direct liquid coating of metal salt [11], metal sulfide [12,13], and metal oxide [14] have been developed. Among these methods, hydrazine is a relatively very toxic and dangerous solvent, which limits the mass production of this method. Chalcogenide nanocrystal method normally would involve complex chemical synthesis and purification process of nanocrystals, it is also rather challenging to be scaled up. Electrostatic spray assisted vapor deposition (ESAVD) has demonstrated to be suitable for the low cost and non-vacuum deposition of metal oxide [15,16], various sulphide [17], and chalcogenide [18]. Figure 1 shows the setup of ESAVD deposition. ESAVD is a non-vacuum chemical vapor deposition based process in which a mixture of chemical precursors is atomized to form aerosol. The aerosol is charged and directed towards a heated substrate where it would undergo decomposition and chemical reaction to deposit a stable solid film onto the substrate. ESAVD is a scalable process and it is a promising way to deposit large area oxide or sulphide thin films due to its characteristics such as low cost, simplicity, versatility, and environmentally friendly [19]. It can be operated in open atmosphere and can easily be adapted to large area deposition using multiple spray atomizers. ESAVD process tends to use more environmental friendly precursors. Most of the precursors would have been reacted and converted to coatings, 2 Experimental In this article, an environmental friendly and sustainable ESAVD process has been exploited and developed for the deposition of CIGS absorber layers. Metal salts of copper, indium, and gallium, as well as thiourea-based chemical were formulated into a chemical precursor mixture. For depositing CIGS films, a mixture of solution containing CuCl 2 , InCl 3 , Ga(NO 3 ) 3 , and thiourea was used and the deposition was performed at a temperature range of 250-450 8C. The as-deposited CIGS layers were selenized under vacuum at 550 8C for 30 min inside a dedicated graphite box in order to form CIGSSe layer. After selenization, 50 nm thick of CdS layer was deposited on top of CIGSSe absorber using chemical bath deposition (CBD) method. CdS deposition was performed using a solution containing 0.0015 M CdSO 4 , 1.5 M NH 4 OH, and 0.075 M thiourea. The samples were immersed in a well-sealed glass bottle in water bath at 60 8C, CdS thin films in the thickness range of 40-50 nm were deposited. On top of CdS, a bilayer composed of i-ZnO(50 nm)/ITO(250 nm) was sputtered as a window layer. Intrinsic ZnO was RF-sputtered at room temperature, with RF power density, O 2 /Ar 2 flow rates and sputtering pressure at 1.1 W cm À2 , 3/11 sccm, and 0.266 Pa, respectively. A sintered ceramic target with In 2 O 3 :SnO 2 ¼ 90:10 wt% was used to sputter ITO thin films. The sputter deposition was performed under the conditions of temperature, RF power density, O 2 /Ar 2 flow rates, and sputtering pressure of 140 8C, 1.3 W cm À2 , 1/8 sccm, and 0.67 Pa, respectively. Finally, on top of ZnO/ITO window layer, a patterned Al layer is thermally evaporated on top of ITO layer as a front electrode. The dimension of individual solar cells is 4 mm Â 4 mm.
The structural characterization of CIGS/CIGSSe layers was carried out using a Siemens D500 X-ray diffraction (XRD) system with a copper source (wavelength l ¼ 1.54 Å'). Raman measurement was carried out using Horiba-Jobin-Yvon LabRam spectrometer with HeNe (632.8 nm) laser excitation. The composition was analyzed using Fischer XAN250 X-ray fluorescence spectrometer. The surface morphology of the films was characterized by FEI XL-30 scanning electron microscopy (SEM). Solar cells were measured under AM1.5 simulated solar light with intensity of 100 mW cm À2 .
3 Results and discussion XRD is used to obtain information on crystalline quality and phase purity of ESAVD deposited CIGS absorber. The XRD of the asdeposited layer is shown in Fig. 2. The peaks at 2u of 28.38 and 46.98 belongs to (112) and (204)/(220) orientations of polycrystalline chalcopyrite CIGS structure, which indicates clearly the formation of CIGS in the as-deposited films. Figure 3 shows the XRD of the films after selenization. Overall, the characteristic peaks of CIGS phases become sharper and narrower which indicated the formation of films with better crystallinity. Since Se has larger atomic diameter as compared with S, the two main peaks at 2u of 28.38 and 46.98 have been shifted to lower 2u of 26.98 and 44.68, respectively, indicating the incorporation of Se in the CIGS crystal structure. The new peak at 53.08 also matches the crystal structure of CIGSe. The results indicate that CIGSSe structure has been formed after selenization. XRD results of CIGS films also show the dominant grow in 112 direction after selenization. Grain boundaries of CIGSSe layer may act as recombination centers for photogenerated charge carriers resulting degradation of device photovoltaic performance. It is desirable to have grain sizes about the order of the film thickness to minimize such recombination effects. The grain size of the CIGSSe film for a high efficiency solar cell is usually larger than 1 mm. Figure 4 shows the cross-section SEM image of ESAVD deposited CIGSSe layer after selenization. SEM image shows large grains with size around hundreds of nm have been obtained in the top layer of CIGSSe absorber after selenization and very small grains exist at the CIGS/Mo interfaces. The possible reason for these smaller grains might be attributed to the limited diffusion of Se through the dense CIGS films.
In order to determine the uniformity of the absorber layer, the thickness and composition were measured on five different points including four corners and one center point on a sample with size of 2 cm Â 2 cm by XRF. Table 1 shows the XRF measurement revealing that the thickness of CIGSSe layer is circa. 1.33 mm before selenization and 1.12 mm after selenization. The element atomic ratios of Cu, In, Ga, and S before selenization are 23.31%, 30.23%, 9.58%, and 36.67%, respectively. After selenization, element atomic ratios of Cu, In, Ga, S, and Se are 25.30%, 26.46%, 10.38%, 27.35%, and 10.51%, respectively. Relative standard deviation (RSD) of thickness and atomic ratios of Cu, In, Ga, S, and Se are all below 10% indicating good uniformity of the thickness and composition in ESAVD deposited CIGSSe film.
Raman analysis is able to detect and distinguish between phases, which may not be distinguishable by diffraction techniques through their characteristic scattering peaks. In order to clearly identify the secondary phases (if any) and to verify the purity of chalcopyrite CIGS phase, Raman was also utilized to characterize CIGSSe thin films. Figure 5 shows the as deposited CIGS thin film exhibiting two characteristic peaks of CIGS, the peak at 298 cm À1 is associated with the "A1" mode of lattice vibration for the   chalcopyrite structures, and another peak at 348 cm À1 is associated with the "E" mode of lattice vibration for the chalcopyrite structures. The absence of other Raman peaks indicates the purity of the CIGS phase. After selenization, due to the incorporation of Se element in CIGS thin film, there are four characteristic peaks appeared on Raman spectra, two are related to CIGS phase and another two due to CIGSe. The Raman peaks at 173 cm À1 is associated with the "A1" mode of lattice vibration for the CIGSe structures, and another peak at 222 cm À1 is associated with the "B 2 /E" mode of lattice vibration for the CIGSe structures.
After finishing the Al deposition and mechanical scribing, a standard glass/Mo/CIGSSe/CdS/ZnO/ITO solar cell was yielded. Two probes with magnetic micropositioners were used to measure the device efficiency under light illumination. Thus far, the fabricated device shows short circuit current (I sc ) of 15.1 mA cm À2 , open circuit voltage (V oc ) of 0.41 V, fill factor (FF) of 0.46 and conversion efficiency 2.82% (as seen in Fig. 6).
As compared to the high efficiency (20%) co-evaporated CIGS solar cells [3], low FF, V oc , and especially I sc limit the efficiency of ESAVD deposited solar cells. From the SEM result and XRF data, it can be seen that the grain size of our absorber is smaller than the desirable value and the metal/ S þ Se ratio of 1.64 is a little higher than the optimum value range of CIGS absorber for high efficiency solar cells. In our future work, we will optimize our selenization setup and selenization conditions to increase the grain size and decrease the metal/(S þ Se) ratio in the absorber. The low V oc of CIGS solar cell is possibly due to shunt current resulted by the presence of pin-holes, agglomerates or impurities in and near the junction. The V oc can be improved by adjusting the precursor formulation and deposition conditions. Further optimizing the contact between top electrode and window layer will decrease the series resistance of the device and lead to a higher current and device efficiency.

Conclusions
In conclusion, CIGS thin films have been deposited by non-vacuum ESAVD method. After selenization, the deposited thin films show the presence of the characteristic peaks of CIGSSe from XRD and Raman. From SEM images and XRF results, it can be seen the deposited absorber is promising for good performance solar cells. The fabricated solar cell with a typical structure of glass/Mo/CIGSSe/CdS/i-ZnO/ITO shows efficiency of 2.82% under 100 mW cm À2 AM1.5 illumination.