Incorporation of copper–indium back-end layers in the solution-based Cu(In, Ga)Se2 films: enhancement of photovoltaic performance of fabricated solar cells

The morphology and photovoltaic properties of the solution-based Cu(In, Ga)Se2 films are effectively improved via the incorporation of copper-indium back-end layers in the precursor films. The effects on the concentrations of bimetal-ions solutions to prepare copper-indium back-end layers are investigated in this study. The incorporation of copper-indium back-end layer in the precursor film enhances the internal diffusion between gallium-ions and indium-ions during selenization reaction. Hence, the porous structure in the back-contact region of prepared CIGS films becomes densified, and the bandgap distribution of films shows a gradient profile. The densified morphology and gradient bandgap reduce the carrier recombination and improve the carrier collection of solar cells. In contrast to the pristine precursor film, the precursor film with a copper-indium back-end layer increase the conversion efficiency of prepared solar cells from 8.34% to 11.13%. The enhancement of conversion efficiency is attributed to the improvement of short-circuit current density and fill factor from 25.70 mA cm−2 to 31.79 mA cm−2 and 57.65% to 65.70%, respectively. This study reveals that the photovoltaic properties of solution-based CIGS solar cells can be improved significantly via the incorporation of copper-indium back-end layers into the precursor films.


Introduction
Thin film solar cells are promising for solar power generation in recent years and Cu(In, Ga)Se 2 (CIGS) one of the potential candidates owing to direct bandgaps, high absorption coefficients, and superior cell performance. CIGS films with broad single-phase composition range and tunable bandgaps have attracted considerable attention as light absorbers for highly efficient photovoltaic devices [1][2][3]. Various vacuum processes are generally utilized to fabricate CIGS films, including co-evaporation processes and two-step fabrication processes that involve sputtering precursor and post-selenization reaction [4][5][6]. The efficiency of CIGS solar cells fabricated via co-evaporation method exceeds 20% [7]. Recently, CIGS-based solar cells prepared by the twostep process for the laboratory-scale exhibits high conversion efficiency of 22.9% [8]. However, vacuum processes for the preparation of CIGS films are marred by problems such as complicated manufacturing processes and low material utilization.
To address the problems of vacuum processes, researchers have developed nonvacuum processes for preparing CIGS films, including spray pyrolysis [9], electrodeposition [10,11], and solution coating [12,13] processes. These nonvacuum approaches are utilized to prepare the precursor film first and then hightemperature treatment in the selenium-containing atmosphere for the selenization reaction to fabricate CIGS films. The highest conversion efficiencies of CIGS solar cells based on a solution processes were 17.0%, the corresponding CIGS films were prepared by the low-temperature pulsed electron deposition method [10]. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Although the nonvacuum processes can be utilized to prepare the CIGS films, but the small particle sizes and high defect densities of solution-based CIGS films result in the poor photovoltaic performance of prepared solar cells. The photovoltaic performance of solar cells is influenced by the morphology and bandgap distribution of CIGS films [14,15]. The small sizes of CIGS particles induce the non-smooth surface and porous morphology of films. A rough surface of CIGS films leads to the inferior coverage of buffer layers to form additional shunt paths, results in the low open circuit potential and fill factor of solar cells [16,17]. The porous microstructure of CIGS films increases the possibility of carrier recombination during the carrier transfer, which deteriorated the photovoltaic conversion efficiency of solar cells [18]. On the other hand, the bandgap distribution of CIGS films influences cell performance. According to previous literature [19,20], gradual increasing the bandgap of CIGS films from surface to back-contact region enhances the collection of photo-generated current and increases the short current density of solar devices. The inverse bandgap distribution of CIGS surface reduces the carrier recombination in the space charge region, and improves the open circuit potential of solar cells [21].
In the present study, solution-based CIGS films with large grain size and bandgap gradient were successful synthesized by using precursor films that contained a copper-indium back-end layer. The copper-indium backend layer was prepared using a solution containing copper and indium ions. The effects of the metal ion concentration of the solution on the phase formation and morphological properties of the CIGS films were investigated. Furthermore, the photovoltaic characteristics of the obtained solar devices were studied in detail by using the prepared CIGS films.

Experimental
Precursor films of CIGS absorbers were fabricated through a spin-coating process using metal ion-based solutions. Two different approaches, namely process A and process B, were adopted to prepare the precursor films of the CIGS absorbers. The structures of the precursor films prepared using these two processes are schematically illustrated in figure 1. In both processes, copper nitrate (Cu(NO 3 ) 2 ·3H 2 O), indium nitrate (In(NO 3 ) 3 ·3H 2 O), and gallium nitrate (Ga(NO 3 ) 3 ) were used as reactants. Process A was used to fabricate standard precursor films. In this process, the standard ion-based solution was prepared first. The molar ratio of copper ions (Cu 2+ ) to IIIA group elements (In 3+ and Ga 3+ ) and gallium ions (Ga 3+ ) to IIIA group elements in the standard ion-based solution were 0.85 and 0.3, respectively. The total concentration of metal ions in the standard ion-based solution was fixed at 0.60 M. Reactants in the calculated molar ratios were dissolved in ethanol solution. The prepared solution was spin-coated on Mo-coated soda-lime glass and dried at 200°C for 10 min. The aforementioned steps were repeated ten times to obtain precursor films of the desired thickness for preparing standard CIGS films.
Process B was used to prepare precursor films with a copper-indium back-end layer. For the preparation of the copper-indium back-end layer, an exclusive solution containing copper and indium ions was prepared in ethanol. The molar ratio of copper ions to indium ions was set as 1:1, and the concentration ratios of both metal ions were varied as 0.2, 0.4, and 0.6 M. The mixed solution was then spin-coated on a Mo-coated soda-lime glass substrate and dried to prepare the copper-indium back-end layer. Subsequently, the standard ion-based solution containing copper, indium, and gallium ions in the same molar ratios as those described in process A was spin-coated on the copper-indium back-end layer and dried at 200°C for 10 min. This spin-coating process was repeated nine times to prepare precursor films of the desired thickness. The as-prepared precursor films of CIGS absorbers obtained using processes A and B were subjected to reduction at 500°C for 60 min in a 5 vol% hydrogen atmosphere. The reduced films were then selenized with selenium vapor at 550°C for 40 min in a 5 vol% H 2 /95 vol% N 2 atmosphere for synthesizing CIGS films. Herein, the CIGS films prepared using process A are designated as sample A1, and those prepared through process B by using bimetal-ion solution concentration ratios of 0.2, 0.4, and 0.6 M are designated as samples B1, B2, and B3, respectively.
The crystalline phases of the prepared CIGS films were determined using an x-ray diffraction (XRD) system with Cu Kα radiation (λ=0.154 nm). Secondary ion mass spectrometry (SIMS, Cameca IMS-4f) was performed to determine the elemental depth profiles of the prepared CIGS films. The morphology of the prepared films was analyzed using a scanning electron microscopy (SEM) system (JEOL JSM-7600F) and an atomic force microscopy (AFM) system (Seiko E-sweep System). The photovoltaic performance of the precursor films prepared through processes A and B after selenization was analyzed by fabricating solar devices. The architecture of the solar cells comprised the following layers: SLG/Mo/prepared CIGS films/cadmium sulfide (CdS)/intrinsic zinc oxide (i-ZnO)/ tin-doped indium oxide (ITO)/Ni:Al. Chemical bath deposition was used to prepare the CdS buffer layers. The i-ZnO layers, ITO layers, and top electrodes were deposited using radio frequency magnetron sputtering. The external quantum efficiency of fabricated CIGS solar cells was analyzed for investigating the bandgap values and carrier collection of prepared films. The CIGS solar cells were characterized using current-voltage (I-V) measurements under an AM 1.5 spectrum with 1000 W m −2 irradiance at 25°C.

Results and discussion
3.1. Effects of copper-indium back-end layers on the phase formations and element distribution of prepared Cu(In, Ga)Se 2 films The precursor films prepared using processes A and B were selenized at 550°C for 40 min to synthesize CIGS films. Figure 2 illustrates the XRD and gracing incidence diffraction (GIXD) patterns of the CIGS films prepared under various conditions. All diffraction peaks of the prepared CIGS films adequately matched the JCPDS No. 35-1101 standard [22], as presented in figure 2(a). Well-crystallized GIGS films with a chalcopyrite structure were successfully synthesized by incorporating copper-indium back-end layers in the precursor films. Figure 2(b) displays the GIXD patterns of the CIGS absorbers prepared using precursor films with and without the copper-indium back-end layers. The GIXD patterns were compared with the JCPDS cards of CuInSe 2 (No. 89-5649) and CuGaSe 2 (No. 81-0903), as shown in figure 2(b). The diffraction peaks of sample A1 shifted to a high angle when the incident depth of x-ray was increased. Comparing the CuInSe 2 and CuGaSe 2 phases revealed a shift of peak positions, indicating an increase in gallium content from the surface toward the bottom region of the CIGS films. By contrast, the introduction of a copper-indium back-end layer broadened the diffraction peaks of the prepared CIGS films, as displayed in figure 2(b). The widening of the diffraction peaks might have been caused by the non-uniform distribution of indium and gallium elemental phases in the CIGS films. The aforementioned results indicate that the incorporation of a copper-indium back-end layer in the precursor films influenced the distribution of gallium and indium atoms in the prepared CIGS films.
To confirm the influence on the distribution of elements in the prepared CIGS films, SIMS analysis was conducted for all fabricated samples. Figure 3 presents the SIMS depth profiles of different elements in the prepared CIGS films. As indicated in figure 3(a), a high secondary ion intensity of copper atoms were observed in the CIGS absorbers prepared using precursor films with a copper-indium back-end layer. As the bimetal-ion solution concentrations were raised to prepare the copper-indium back-end layer, the secondary ion intensity of copper atoms was increased gradually. This phenomenon was caused by the high concentration of copper atoms present in the bimetal-ion solution. The distribution of selenium atoms is illustrated in figure 3(b); the depth profiles of selenium atoms in all samples were observed to be the same, regardless of whether a copper-indium back-end layer was present in the precursor films, which were the common distribution in the CIGS films [22][23][24]. As illustrated in figure 3(c), the distribution of gallium atoms exhibited a gradient profile from the backcontact region to the surface region in all samples. Fabrication of CIGS films using precursor films of copperindium back-end layer decreases the contents of gallium atoms in the surface region. Further increase of concentration of bimetal-ion solution to prepare copper-indium back-end layer decreases the contents of gallium atoms in surface region.
Notably, the distribution of indium atoms showed an inverse gradient profile compared with the depth profiles of gallium atoms, as shown in figure 3(d). When the concentration of the bimetal-ion solution was further increased to prepare copper-indium back-end layers, the gradient distribution of indium atoms was enhanced gradually. The gradient distribution of indium atoms can be attributed to the different reaction kinetics of binary selenides and the corresponding gradual migration of indium atoms to the surface region during the selenization process [25]. Meanwhile, the gallium atoms were gradually accumulated in the backcontact region [26,27]. Notably, when the bimetal-ion solution concentration was increased for preparing copper-indium back-end layers, the effects of interactive diffusion between indium and gallium atoms in the CIGS films were enhanced. The elemental distribution of indium and gallium atoms influences the bandgap (E g ) of CIGS films strongly. The E g value of the CIGS films was elevated by increasing the atomic ratios of gallium atoms to IIIA atoms (GGI value) [28,29]. When a copper-indium back-end layer incorporated into the precursor films, the gradient distribution of gallium atoms in the prepared CIGS films was enhanced; thus, the GGI ratios of the CIGS films were gradually increased from the surface to the back-contact region, and concurrently, the E g values of the prepared films exhibited a gradient distribution.

3.2.
Effects of copper-indium back-end layers on the microstructure and morphology of prepared Cu(In, Ga)Se 2 films Figure 4 illustrates the top-view and cross-sectional micrographs of the CIGS films prepared using precursor films with and without copper-indium back-end layers. The CIGS film (sample A1) prepared using precursor films without any copper-indium back-end layer exhibited a flat and smooth surface, as displayed in figure 4(a). However, the cross-section image of the sample A1 showed small grain sizes and voids in the bottom region ( figure 4(b)). This non-dense microstructure in the bottom region has generally been observed in CIGS films prepared using nonvacuum processes [30,31]. The flat surface with few agglomerated particles was observed along with the reduction in voids from the bottom region of CIGS film prepared using a precursor film of a copper-indium back-end layer with 0.  increasing the concentration of bimetal-ion solution to 0.6 M, the particle sizes of agglomerated particles on CIGS surface were increased with the densification of cross-section microstructure, as shown in figures 4(g) and (h). The results reveal that the incorporation of copper-indium back-end layer in the precursor films effectively  improve the porous morphology in the bottom region of CIGS films. Meanwhile, the surface morphology of CIGS films was influenced by the bimetal-ion solution concentration.
To investigate the surface morphology of CIGS films, an AFM analysis was employed. Figure 5 depicts AFM micrographs of the CIGS films prepared using precursor films with and without copper-indium back-end layers. According to figure 5(a), Sample A1 exhibited a smooth surface with low root-mean-square (RMS) roughness. The CIGS film (sample B1) prepared using a precursor film with a copper-indium back-end layer, the RMS roughness value increased to 211 nm. When the bimetal-ion solution concentration was increased to 0.4 and 0.6 M for preparing copper-indium back-end layers, the RMS roughness of the resulting films increased to 223 and 252 nm, respectively. The surface roughness of the prepared CIGS films presented an upward trend with the bimetal-ion solution concentration, which is consistent with the rugged morphology observed in the previous SEM results (figure 4). Figure 6 presents a schematic of the effects of copper-indium back-end layers during post-selenization. According to the previous researches [32,33], the formation of various intermediates such as indium selenide and copper selenide as fluxing agents induced the internal diffusion of atoms to promote grain growth in the CIGS films during selenization reaction. The interaction phenomenon resulted in the diffusion of indium atoms toward the surface of CIGS films gradually, and gallium atoms accumulated in the bottom region of the CIGS films, as shown in figure 6(a). When the precursor film with a copper-indium backend layer was used, the internal atom diffusion was increased and promotion the grain growth of CIGS particles, resulting in the densification of prepared films, as displayed in figure 6(b). Figure 7 shows plots of the I-V curves of CIGS solar cells prepared from precursor films with and without a copper-indium back-end layer. Table 1 summarizes the photovoltaic parameters of the prepared CIGS solar devices. The conversion efficiency of solar devices based on the CIGS film (sample A1) was estimated to be 8.34%. The V oc , J sc , and FF values were 563 mV, 25.70 mA cm −2 , and 57.65%, respectively. When the CIGS film (sample B1) was prepared using precursor films with a copper-indium back-end layer, the conversion efficiency of the corresponding solar devices increased to 10.52%. The V oc , J sc , and FF values were 542 mV, 30.29 mA cm −2 , and 64.06%, respectively. Furthermore, when a copper-indium back-end layer was prepared using a 0.4 M bimetal-ion solution, the conversion efficiency of the CIGS cells increased to 11.13%. Further increasing the concentration of bimetal-ion solution for preparing a copper-indium back-end layer, the conversion efficiency of corresponding solar cells was decreased to 10.41%.    The relationship between the concentrations of the bimetal-ion solution used to prepare copper-indium back-end layers and the photovoltaic parameters of the CIGS solar cells is illustrated in figure 8. The V oc values of the prepared solar cells decreased gradually when the concentrations of the bimetal-ion solution used to prepare copper-indium back-end layers increased. The decrease in V oc values was caused by the rough surface morphology of the CIGS films, which was induced by the incorporation of the copper-indium back-end layer. The rough surface of the CIGS films was considered to result in a non-uniform coverage of the buffer layer, thereby forming shunt paths in the p-n junction region. Another reason for the decrease in V oc values was decreased gallium content on the surface of the prepared CIGS films. As the gallium content in the surface region decreased, the bandgap values of the CIGS films in the surface region were reduced and the V oc values were declined. The J sc and FF values of prepared solar cells increased significantly when CIGS films fabricated by the precursor film of copper-indium back-end layer. The improvement of J sc and FF values was attributed to the bandgap gradient and reduce amount of voids in the bottom region of prepared CIGS films which was observed in figures 4(b) and (f). According the previous research [34], the dense structure of the CIGS films restrained the carrier recombination, and the gradient distribution of bandgap enhanced the collection of photo-generated carriers. Hence, the J sc and FF values were elevated.

Effects of concentration of the bimetal-ion solution used to prepare copper-indium back-end layers on photovoltaic performance of prepared Cu(In, Ga) Se 2 solar cells
For investigating the diode parameters of prepared CIGS cells, the current density (J) is expressed using the diode equation [35]: where J 0 and q denote the saturated current density and electron charge respectively, A denotes the diode factor which identified the mechanism of carriers recombination in prepared solar cells, k and T denote the Boltzmann constant and temperature (K) respectively, R s denotes the series resistance, J sc denotes the short-circuit current density, and G denotes the shunt conductance. The G values were extracted from the slope in the reverse bias region of the J-V curves and were used to calculate J ′ =J-GV. While R s G = 1, equation (1) can be simplified as below: Differentiating equation (2) with respect to J′ yields the below equation The value of series resistance (R s ) can be calculated from the intercept of the plots dV/dJ′ versus (J′+J sc ) −1 and shown in figure 9(a). Furthermore, equation (2) can be rewritten as equation (4): The value of diode factor (A) and saturated current density (J 0 ) can be respectively calculated from the slope and intercept of the equation (4) (as shown in figure 9(b)). Table 2 summarizes the obtained diode parameters of prepared CIGS solar cells. According to the results in table 2, the diode parameters of fabricated CIGS solar cells were decreased when CIGS films prepared from precursor films with copper-indium back-end layer. As the concentration of bimetal-ion solution was increased  Further increasing the concentration of bimetal-ions solution to 0.6 M, all diode parameters of prepared solar cells were slightly increased. The increase in diode parameters was attributed to the decrease of bandgap in CIGS surface and the addition of surface roughness. The low bandgap value in the p-n junction region enhanced the possibility of carrier recombination, and the rough surface of CIGS films induced the formation of additional shunt paths. Therefore, the diode parameters of CIGS solar cells were deterioration. According to the aforementioned results, the preparation of CIGS films by using precursor films with appropriate concentrations of a copper-indium back-end layer can effectively enhance the photovoltaic characteristics of CIGS solar cells through a reduced leakage current and suppressed carrier recombination.

Conclusions
The incorporation of copper-indium back-end layers in the precursor films was an effective approach to improve the morphology and photovoltaic characteristics of solution-based CIGS films. The particle sizes and bandgap gradient of prepared CIGS films were increased gradually when the concentration of bimetal-ion solution was raised. The increase in the CIGS particle sizes reduced the formation of grain boundaries and suppressed the carrier recombination. The bandgap gradient of CIGS films enhanced the carrier collection; thereby, the photo-generated current of solar cells was increased. As the copper-indium back-end layer was prepared from the 0.4 M of bimetal-ion solution, the shunt conductivity and saturated current density of prepared solar cells effectively reduced to 1.86 mS cm −2 and 7.68×10 −4 mA cm −2 respectively. Hence, the conversion efficiency of CIGS solar cells was significantly improved to 11.13%. Further increasing the concentration of bimetal-ion solution to 0.6 M, the rough surface with low bandgap of prepared CIGS films resulted in the decrease of open circuit potential and deterioration of the photovoltaic conversion efficiency of solar cells. The present study revealed that the incorporation of a copper-indium back-end layer into the precursor film for preparing CIGS films is an effective method for improving the photovoltaic performance of CIGS solar cells.