Demonstration of Solar Cell on a Graphite Sheet with Carbon Diffusion Barrier Evaluation

An amorphous Si (a-Si) solar cell with a back reflector composed of zinc oxide (ZnO) and silver (Ag) is potentially the most plausible and flexible solar cell if a graphite sheet is used as the substrate. Graphite supplies lightness, conductivity and flexibility to devices. When a graphite sheet is used as the substrate, carbon can diffuse into the Ag layer in the subsequent p-i-n process at 200–400 °C. To prevent this, we added an oxide layer as a carbon diffusion barrier between the carbon substrate and the back reflector. For the carbon diffusion barrier, silicon oxide (SiO2) or tin oxide (SnOx) was used. We evaluated the thermal stability of the back reflector of a carbon substrate using secondary-ion mass spectrometry (SIMS) to analyze the carbon diffusion barrier material. We confirmed the deposition characteristics, reflectance and prevention of carbon diffusion with and without the barrier. Finally, the structures were incorporated into the solar cell and their performances compared. The results showed that the back reflectors that were connected to a carbon diffusion barrier presented better performance, and the reflector with an SnOx layer presented the best performance.


Introduction
The interest in renewable energy sources is constantly increasing due to the concurrent increase in environmental issues [1]. Therefore, the research and market of solar cells, which use a representative clean energy, have continued to grow [2]. Devices are becoming smaller and lighter [3,4], and an emerging trend for devices is the adaptation for flexible appliances [5]. On this basis, we developed the idea of a solar cell integrated to a flexible device [6].
Materials that can potentially overcome the shortcomings of conventional solar cell substrates have been investigated. Silicon and glass, which are conventionally used as substrates for solar cells, are expensive and rigid. Moreover, the absence of electrical conductivity is a limitation to their application in solar cells [7]. In contrast, carbon materials are light, have high electrical conductivity, can be processed in various forms [8][9][10][11][12] and can also be flexible [13]; thus, a solar cell with a carbon substrate is lighter and more flexible [14]. The graphite sheet is one of the most popular commercial substrates among the flexible carbon materials [5,13]. In this work, an a-Si, which can easily be used with various substrates, was modeled as the final structure [15]. A back reflector which consisted of zinc oxide/silver (ZnO/Ag) was added to increase the light efficiency [16,17].
The high diffusivity of carbon at low temperatures (<800 • C) must be considered when it is used as a substrate [18,19]. There is a possibility that carbon could diffuse into the Ag layer, which is simultaneously used as a back reflector and back electrode, in the subsequent p-i-n formation process at 200-400 • C. If the carbon diffuses into the Ag layer, it may act as an impurity in the back electrode, thus degrading the current density (J c ) [20,21]. To prevent the diffusion of carbon, a diffusion barrier was added. For the barrier material, two types of oxide were used-silicon oxide (SiO 2 ), which is a typical dielectric oxide, and tin oxide (SnO x ), a transparent conductive oxide (TCO) widely used in solar cells [2,17]. In summary, a graphite sheet (flexible carbon substrate) was used to produce a flexible solar cell device, and the functionality of the carbon diffusion barrier at the back structure of an a-Si solar cell-which included a carbon substrate and a back reflector-was investigated.
To apply an inexpensive graphite sheet as a solar cell substrate, the interface with the back reflector and the Ag/graphite sheet were annealed at 200-400 • C, which accounted for the chemical stability. The effects of the SiO 2 and SnO x layers between Ag/graphite were compared in terms of deposition characteristics, reflectivity and the chemical stability of Ag. Finally, the electrical properties of the a-Si solar cells were measured.
The aim of this research was to take a first step toward demonstrating that solar cells can be applied to flexible devices using graphite sheets as substrates when the carbon diffusion barrier is added. By extending their range to flexible devices, new possibilities for solar cell applications become clear.

Structure and Materials
In this study, a graphite sheet (EYGS-121803, Panasonic, Osaka, Japan) was used as the carbon substrate. For the back reflector layer, ZnO and Ag were used at a ratio that showed effective light trapping [22][23][24][25]. For the carbon diffusion barrier, an oxide layer of SiO 2 or SnO x was used.
Before the thin film deposition, the substrate was cleaned to remove any contamination by organic matter using acetone, isopropyl alcohol (IPA) and deionized (DI) water. First, the substrate was immersed in acetone for 5 min, then immersed in IPA for 3 min. Finally, it was rinsed for more than 1 min in DI water. The remaining DI water was then removed using a nitrogen (N 2 ) blowing gun.
Each thin film was deposited using a magnetron sputter (KVS-T8860, Korea Vacuum limited, Daegu, Korea). The SiO 2 , SnO x and ZnO thin films were deposited using a radio frequency (RF) power magnetron sputter, a controlled argon (Ar) flow of 5 sccm and constant 100-W RF power. The process temperature was held at 25 • C. The Ag thin film was deposited by direct current (DC) magnetron sputter, with constant process pressure of 3 mTorr, Ar flow of 22 sccm and substrate temperature of 25 • C.
When the carbon diffusion barrier was assembled, 100 nm of SiO 2 or SnO x was first deposited. The thickness of the back reflector included 350 nm of Ag and 100 nm of ZnO. These values represented the highest reflection efficiency according to previous experiments.

Thermal Stability of the Back Reflector on Graphite Sheet
After the structure was assembled, annealing was performed to create an environment similar to the p-i-n process for carbon diffusion evaluation. A drying furnace (WINFUS) was used for 1 h at 200, 300 and 400 • C in an N 2 atmosphere.
Physical properties were observed using transmission electron microscopy (TEM, JEM 2100F, JEOL, Tokyo, Japan) and ultraviolet-visible spectroscopy (UV-vis, Cary5000, Agilent, Santa Clara, CA, USA). These analyses were performed with the samples before annealing. The annealed samples were used only for diffusion evaluation. Diffusion characteristics were determined using time-of-flight secondary-ion mass spectrometry (TOF-SIMS, TOF.SIMS-5, ION-TOF, Munster, Germany).

Solar Cell Fabrication
The electrical properties of the solar cell were measured by applying the graphite sheet as the substrate. The structures of the a-Si solar cells are shown in Figure 1. Aluminum (Al) grids and 130 nm indium tin oxide (ITO) constitute the top electrode, which is combined with a 400 nm p-i-n and a back structure. The back structure is the same as mentioned in Section 2.1, and all upper structures were deposited using plasma-enhanced chemical vapor deposition (PECVD). The upper structure was processed at 200−400 • C. First the n-layer and then the intrinsic layer were deposited at 300 mTorr in hydrogen (H 2 ) and silane (SiH 4 ) gas. In the p-layer, methane (CH 4 ) and diborane (B 2 H 6 ) gas were used in addition to H 2 and SiH 4 at 250 mTorr. Finally, the ITO layer was deposited in an Ar atmosphere at 300 mTorr. The solar cell performance was evaluated using the solar simulator under standard test conditions (STC = 1000 W/m 2 , 25 • C, AM1.5), which are the solar cell performance evaluation conditions specified by the International Electrotechnical Commission (IEC) [26,27]. The electrical properties of the solar cell were measured by applying the graphite sheet as the substrate. The structures of the a-Si solar cells are shown in Figure 1. Aluminum (Al) grids and 130 nm indium tin oxide (ITO) constitute the top electrode, which is combined with a 400 nm p-i-n and a back structure. The back structure is the same as mentioned in Section 2.1, and all upper structures were deposited using plasma-enhanced chemical vapor deposition (PECVD). The upper structure was processed at 200−400 °C. First the n-layer and then the intrinsic layer were deposited at 300 mTorr in hydrogen (H2) and silane (SiH4) gas. In the p-layer, methane (CH4) and diborane (B2H6) gas were used in addition to H2 and SiH4 at 250 mTorr. Finally, the ITO layer was deposited in an Ar atmosphere at 300 mTorr. The solar cell performance was evaluated using the solar simulator under standard test conditions (STC = 1000 W/m 2 , 25 °C, AM1.5), which are the solar cell performance evaluation conditions specified by the International Electrotechnical Commission (IEC) [26,27].  Figure 2 shows the cross-sectional images of the multilayer samples. It can be confirmed that in the samples with a carbon diffusion barrier, the Ag thin film was more uniformly deposited. This phenomenon occurred because the carbon diffusion barrier reduced the influence of the graphite sheet roughness. The reflection efficiencies of the samples with and without the carbon diffusion barrier were compared. Figure 3 demonstrates the reflectance of the back reflectors. The reflectance was measured in the visible light region, which is the wavelength range of amorphous solar cells [28].  Figure 2 shows the cross-sectional images of the multilayer samples. It can be confirmed that in the samples with a carbon diffusion barrier, the Ag thin film was more uniformly deposited. This phenomenon occurred because the carbon diffusion barrier reduced the influence of the graphite sheet roughness. The electrical properties of the solar cell were measured by applying the graphite sheet as the substrate. The structures of the a-Si solar cells are shown in Figure 1. Aluminum (Al) grids and 130 nm indium tin oxide (ITO) constitute the top electrode, which is combined with a 400 nm p-i-n and a back structure. The back structure is the same as mentioned in Section 2.1, and all upper structures were deposited using plasma-enhanced chemical vapor deposition (PECVD). The upper structure was processed at 200−400 °C. First the n-layer and then the intrinsic layer were deposited at 300 mTorr in hydrogen (H2) and silane (SiH4) gas. In the p-layer, methane (CH4) and diborane (B2H6) gas were used in addition to H2 and SiH4 at 250 mTorr. Finally, the ITO layer was deposited in an Ar atmosphere at 300 mTorr. The solar cell performance was evaluated using the solar simulator under standard test conditions (STC = 1000 W/m 2 , 25 °C, AM1.5), which are the solar cell performance evaluation conditions specified by the International Electrotechnical Commission (IEC) [26,27]. The reflection efficiencies of the samples with and without the carbon diffusion barrier were compared. Figure 3 demonstrates the reflectance of the back reflectors. The reflectance was measured in the visible light region, which is the wavelength range of amorphous solar cells [28]. The reflection efficiencies of the samples with and without the carbon diffusion barrier were compared. Figure 3 demonstrates the reflectance of the back reflectors. The reflectance was measured in the visible light region, which is the wavelength range of amorphous solar cells [28].   The difference in the reflectance was not significant despite the carbon diffusion barrier. This result was attributed to the characteristics of the graphite sheet substrate, which was coated and presented a glossy surface. This feature appeared to attribute high reflectivity without a significant difference among the samples. Nevertheless, the reflectance was saturated at a higher value when the SnOx carbon diffusion barrier was used.

Thermal Stability of the Back Reflector
We evaluated the carbon diffusion between the back reflector and the graphite sheet as the annealing temperature varied from 200 to 400 °C, which is the temperature range for the subsequent p-i-n formation process. The SIMS analysis was performed to compare the diffusion of carbon according to different barrier structures and annealing temperatures.
The carbon peak was extracted, and the results are shown in Figure 4. In the absence of the carbon diffusion barrier (Figure 4a), carbon was detected in the Ag layer even before annealing-that is, as-deposited. The diffusion decreased significantly in the presence of the barrier. Moreover, no carbon peaks were detected in the Ag layer when the barriers were used.  The difference in the reflectance was not significant despite the carbon diffusion barrier. This result was attributed to the characteristics of the graphite sheet substrate, which was coated and presented a glossy surface. This feature appeared to attribute high reflectivity without a significant difference among the samples. Nevertheless, the reflectance was saturated at a higher value when the SnO x carbon diffusion barrier was used.

Thermal Stability of the Back Reflector
We evaluated the carbon diffusion between the back reflector and the graphite sheet as the annealing temperature varied from 200 to 400 • C, which is the temperature range for the subsequent p-i-n formation process. The SIMS analysis was performed to compare the diffusion of carbon according to different barrier structures and annealing temperatures.
The carbon peak was extracted, and the results are shown in Figure 4. In the absence of the carbon diffusion barrier (Figure 4a), carbon was detected in the Ag layer even before annealing-that is, as-deposited. The diffusion decreased significantly in the presence of the barrier. Moreover, no carbon peaks were detected in the Ag layer when the barriers were used.
After annealing, each layer was diffused, regardless of the presence of a carbon diffusion barrier. Moreover, as the annealing temperature increased, the diffusion of carbon became more noticeable. Figure 5 summarizes the SIMS depth profiles of the main components annealed at 400 • C compared to the values of the as-deposited samples. We focused on this case because it was the most critical temperature. We also compared the diffusion depth of carbon, which is shown in yellow on Figure 5.
Molecules 2019, 24, x; doi: FOR PEER REVIEW www.mdpi.com/journal/molecules p-i-n formation process. The SIMS analysis was performed to compare the diffusion of carbon according to different barrier structures and annealing temperatures. The carbon peak was extracted, and the results are shown in Figure 4. In the absence of the carbon diffusion barrier (Figure 4a), carbon was detected in the Ag layer even before annealing-that is, as-deposited. The diffusion decreased significantly in the presence of the barrier. Moreover, no carbon peaks were detected in the Ag layer when the barriers were used. After annealing, each layer was diffused, regardless of the presence of a carbon diffusion barrier. Moreover, as the annealing temperature increased, the diffusion of carbon became more noticeable. Figure 5 summarizes the SIMS depth profiles of the main components annealed at 400 °C compared to the values of the as-deposited samples. We focused on this case because it was the most critical temperature. We also compared the diffusion depth of carbon, which is shown in yellow on Figure 5. For the samples with a carbon diffusion barrier (Figure 5b,c), the carbon diffused approximately 25 nm into the SiO2 and SnOx layers. In the absence of the barrier (Figure 5a), carbon diffused approximately 113 nm into the Ag layer; therefore, carbon diffused better in the Ag layer than in the oxide layers. The results show that the barriers were effective in preventing carbon diffusion into the Ag layer.
When a carbon diffusion barrier is used, the structure becomes oxide/metal/oxide (OMO) [29,30]. This multilayer electrode structure is being actively researched for use in photoelectric devices [7]. Its advantages include low resistance and thermal and mechanical stability [30,31]. It also presents suitable characteristics for a flexible electrode [31]. The structure containing the carbon diffusion barrier is therefore beneficial not only to prevent carbon diffusion, but also to function as an effective electrode for solar cells.
The SIMS results show that the Ag peak saturates after the Ag layer (Figure 5a,c), but it is high only in the SiO2 thin film (Figure 5b). This result was attributed to the difference in density between SnOx and SiO2 thin films. The X-ray reflectivity (XRR) analysis showed that the densities of SiO2 and SnOx were 2.55 and 5.91 g/cm 3 , respectively. In addition, according to their molecular weights, the masses of SiO2 and SnOx were 9.98 × 10 −23 and 23.58 × 10 −23 g, respectively. Thus, they have a molecular density close to 2.55 × 10 −22 /cm 3 . Consequently, the diffusion of Ag is more likely to occur in the SiO2 layer due to its lower molecular mass. In addition, when the deposition cross-section was examined, the SnOx interface was cleaner than that of SiO2 and the influence of the interface was clear. For the samples with a carbon diffusion barrier (Figure 5b,c), the carbon diffused approximately 25 nm into the SiO 2 and SnO x layers. In the absence of the barrier (Figure 5a), carbon diffused approximately 113 nm into the Ag layer; therefore, carbon diffused better in the Ag layer than in the oxide layers. The results show that the barriers were effective in preventing carbon diffusion into the Ag layer.

Amorphous Silicon Solar Cell Demonstration
When a carbon diffusion barrier is used, the structure becomes oxide/metal/oxide (OMO) [29,30]. This multilayer electrode structure is being actively researched for use in photoelectric devices [7]. Its advantages include low resistance and thermal and mechanical stability [30,31]. It also presents suitable characteristics for a flexible electrode [31]. The structure containing the carbon diffusion barrier is therefore beneficial not only to prevent carbon diffusion, but also to function as an effective electrode for solar cells.
The SIMS results show that the Ag peak saturates after the Ag layer (Figure 5a,c), but it is high only in the SiO 2 thin film (Figure 5b). This result was attributed to the difference in density between SnO x and SiO 2 thin films. The X-ray reflectivity (XRR) analysis showed that the densities of SiO 2 and SnO x were 2.55 and 5.91 g/cm 3 , respectively. In addition, according to their molecular weights, the masses of SiO 2 and SnO x were 9.98 × 10 −23 and 23.58 × 10 −23 g, respectively. Thus, they have a molecular density close to 2.55 × 10 −22 /cm 3 . Consequently, the diffusion of Ag is more likely to occur in the SiO 2 layer due to its lower molecular mass. In addition, when the deposition cross-section was examined, the SnO x interface was cleaner than that of SiO 2 and the influence of the interface was clear.

Amorphous Silicon Solar Cell Demonstration
The solar cell was assembled so the feasibility of the carbon substrate and the barrier layer effect on the device characteristics could be examined. Due to the graphite sheet used as the substrate, the solar cells also showed flexibility by being able to bend. The performance of the solar cells is shown in Figure 6 and Table 2.
Molecules 2019, 24, x FOR PEER REVIEW 6 of 8 The solar cell was assembled so the feasibility of the carbon substrate and the barrier layer effect on the device characteristics could be examined. Due to the graphite sheet used as the substrate, the solar cells also showed flexibility by being able to bend. The performance of the solar cells is shown in Figure 6 and Table 2.  The performance of solar cells can be compared using conversion efficiency (Eff), which can be expressed by Equation (1). This equation shows that the ratio of incident light energy to electrical energy can be converted into electrical energy. Here, Pinput is the solar energy power incident on the solar cell, Poutput is the electrical energy output, fill factor (FF) is the charging factor, Imax is the maximum current, Vmax is the maximum voltage, Isc is the short-circuit current and Voc is the open voltage when the resistance is flowing indefinitely.
The overall efficiency is not significantly lower than the efficiency of the a-Si solar cells described in previous papers, and all samples showed similar values [32]. Moreover, because the open voltage values were the same, it was concluded that the short-circuit current density (Jsc) and FF affected the energy conversion efficiency. The reason for this result was attributed to the reduction of the influence of the lower part of the structure while the influence of the upper structure was applied.

Conclusions
In this study, graphite sheet was used as a substrate in solar cells for light and flexible devices. The effects of a carbon diffusion barrier on the backside of a solar cell where the back reflector was bonded to a carbon substrate were investigated. The thermal stability of the back reflectors on the graphite sheet was examined using SIMS with varying barrier layers and annealing temperatures up to 400 °C. We observed that the carbon diffused into Ag as the processing temperature increased; therefore, SiO2 or SnOx were used as carbon diffusion barriers. Moreover, 350 nm Ag and 100 nm ZnO were combined in the back reflector layer to increase the efficiency on the back of the cell.  The performance of solar cells can be compared using conversion efficiency (E ff ), which can be expressed by Equation (1). This equation shows that the ratio of incident light energy to electrical energy can be converted into electrical energy. Here, P input is the solar energy power incident on the solar cell, P output is the electrical energy output, fill factor (FF) is the charging factor, I max is the maximum current, V max is the maximum voltage, I sc is the short-circuit current and V oc is the open voltage when the resistance is flowing indefinitely.
The overall efficiency is not significantly lower than the efficiency of the a-Si solar cells described in previous papers, and all samples showed similar values [32]. Moreover, because the open voltage values were the same, it was concluded that the short-circuit current density (J sc ) and FF affected the energy conversion efficiency. The reason for this result was attributed to the reduction of the influence of the lower part of the structure while the influence of the upper structure was applied.

Conclusions
In this study, graphite sheet was used as a substrate in solar cells for light and flexible devices. The effects of a carbon diffusion barrier on the backside of a solar cell where the back reflector was bonded to a carbon substrate were investigated. The thermal stability of the back reflectors on the graphite sheet was examined using SIMS with varying barrier layers and annealing temperatures up to 400 • C. We observed that the carbon diffused into Ag as the processing temperature increased; therefore, SiO 2 or SnO x were used as carbon diffusion barriers. Moreover, 350 nm Ag and 100 nm ZnO were combined in the back reflector layer to increase the efficiency on the back of the cell.
Three samples were prepared to compare the characteristics of the backs of the solar cells. The first one had only an as-deposited back reflector layer, while the second and third had an SiO 2 or SnO x carbon diffusion barrier.
When the carbon diffusion barriers were used, the film was deposited more evenly because the influence of the substrate was reduced. The reflection efficiency of the back reflector was maintained. It was possible to prevent carbon diffusion even when the structure of the back of the cell was only several nanometers thin. This experiment is a basic study that demonstrates the applicability of a graphite sheet to a solar cell. The practical use of the device would be possible by conducting further studies about the aging of the cell, such as temperature and light stability of the internal and external environments.