Influence of conduction band minimum difference between transparent conductive oxide and absorber on photovoltaic performance of thin-film solar cell

Conduction band offset (CBO) and valence band offset (VBO) of the buffer/absorber interface are well-known as important parameters to reduce the carrier recombination at the interface and achieve high efficiency solar cell.^1,2) Cu(In,Ga)Se₂ (CIGS) solar cell has been demonstrated high conversion efficiency of above 20% with a small area by manipulating the bandgap profile because its absorber bandgap is approximately 1.2 eV, and the CBO of CdS buffer/CIGS absorber is enough matched for high efficiency solar cell.^3–6) Currently, novel absorber materials of the thin-film solar cells such as Cu₂ZnSnS₄ (CZTS),⁷⁾ Cu₂O,⁸⁾ and SnS⁹⁾ have been intensively studied. Figure 1 depicts the band alignments between ZnO,^10–13) CdS,^10–13) Cu₂ZnSnS₄ (CZTS),¹⁰⁾ Cu₂ZnSnSe₄ (CZTSe),¹⁰⁾ CuInSe₂ (CISe),^10–13) CuGaSe₂ (CGSe),¹⁰⁾ Cu₂O,¹¹⁾ and SnS.¹²⁾ In the figure, the conduction band minimum (E_C) of ZnO is set to be 0.0 eV, and the numbers indicate the energy difference of the E_C of other materials from E_C of ZnO. ZnO as transparent conductive oxide layers (TCO) and CdS as buffer layers are commonly utilized in the thin-film solar cells with above absorbers. The absorbers possess different bandgap energies and electron affinities. Accordingly, it is greatly important to select appropriate materials as the TCO and buffer layers in solar cell structure, leading to the improved CBO of the buffer/absorber interface in order to reduce the carrier recombination, thereby increasing cell efficiency. In our previous study, the optimization of the CBO of buffer/absorber interface was quantitatively demonstrated.¹⁾ According to the result, high cell performance can be obtained when the E_C of the buffer layer is higher by 0.0–0.4 eV than that of absorber, whereas the cell performance sharply decreases when the E_C of buffer layer is below that of absorber. Namely, the cliff is formed at the buffer/absorber interface. We have recently reported the fabrication of CIGS solar cell on flexible stainless steel substrate with conversion efficiency of above 16% without anti-reflective layer by the variation of band gap energy of CIGS absorber to optimize the CBO of the buffer/absorber interface; however, the carrier recombination near the interface was still observed.^14,15)

Zoom In Zoom Out Reset image size

In this work, in order to further increase cell efficiency, it is considered that the E_C difference between TCO layer and the absorber, named ΔE_C-TA, should be scrutinized. Therefore, it is intriguing to investigate the influence of the ΔE_C-TA on photovoltaic performance as the useful knowledge for designing new material combination of buffer, absorber and TCO for high cell performance. The investigation was conducted using a device simulator, the Solar Cell Capacitance Simulator (SCAPS).¹⁶⁾ The optimized ΔE_C-TA for high photovoltaic performance was quantitatively revealed.

One-dimensional device simulation software SCAPS¹⁶⁾ ver. 3.2.1 developed by Gent University was used in this study. The structure of the thin-film solar cell in the simulation was TCO (0.3 µm)/buffer (0.05 µm)/absorber (2.0 µm), as shown in Fig. 2. The interface defects of ID_T-B and ID_B-A were inserted between the TCO and buffer layers and between the buffer and absorber layers, respectively. Here, the positive (negative) signs of energy difference denote that the E_Cs of TCO and buffer are higher (lower) than that of absorber. Table I summarizes input parameters for each layer in the simulation. The parameters of TCO, buffer, and absorber layers were based on ZnO, CdS, and CIGS, respectively. W is the thickness of the layer, ε_r is the relative permittivity, E_g is band gap energy, χ is the electron affinity, N_A and N_D denote acceptor and donor densities, N_t is the defect density, δ_e and δ_h are electron and hole capture cross sections, and E_d is the defect level from valence band maximum (E_V), respectively. The parameters were basically identical to values in simulation report in Ref. 17; however, small revisions were made to fit our case. The ΔE_C-TA value was varied from −0.8 to 0.8 eV by adjusting the E_C of TCO, namely varying the χ of TCO. The E_C difference between buffer layer and absorber, called ΔE_C-BA, was 0.1 eV, appropriate to obtain high cell efficiency.¹⁾ Recently, many research groups have proposed various materials used as the buffer layers of the thin-film solar cells such as CdS,¹⁸⁾ ZnO,¹⁹⁾ ZnS(O,OH),²⁰⁾ ZnS,²¹⁾ and InS,²²⁾ where they have different carrier densities. The carrier density of buffer layers, named N_D-B, in the simulation was therefore set to be 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³ to investigate the impact of N_D-B (i.e., different materials of buffer layer) on cell performance. Figure 3 displays the energy band diagrams of the solar cells with different N_D-B, in which the carrier concentration of the absorber is constant at 2.0 × 10¹⁶ cm⁻³. In the case of N_D-B of 1.0 × 10¹³ cm⁻³, the space-charge region (SCR) is formed in both near-surface area of the absorber and buffer layer, whereas in the case of N_D-B of 1.0 × 10¹⁸ cm⁻³, only near-surface region of the absorber is depleted. In addition, the interface defect was considered to be the important parameter; however, it was ambiguous. The ID_T-B and ID_B-A were therefore set to be 1.0 × 10¹³ cm⁻² in order to obtain conversion efficiency of 20% with ΔE_C-TA of 0.0 eV. Other input parameters not included in Table I were set to be as follows. Effective densities of conduction and valence bands were 2.2 × 10¹⁸ and 1.8 × 10¹⁹ cm⁻³, respectively. Mobility of electron and hole was 40 and 10 cm² V⁻¹ s⁻¹. Thermal velocity of electron and hole was 10⁷ cm/s. Defect type was neutral and distributed in Gaussian with characteristic energy of 0.1 eV. Pre-factor A_α was 10⁵ cm⁻¹ to obtain absorption coefficient (α) curve calculated by α = A_α(hν − E_g)^1/2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 4 illustrates cell performance parameters with different N_D-Bs of 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³ as a function of the ΔE_C-TA. It was revealed that the maximum efficiency in the case of N_D-B = 1.0 × 10¹³ cm⁻³ was almost the same as that in the case of N_D-B = 1.0 × 10¹⁸ cm⁻³. In both cases, conversion efficiency was drastically decreased, when the ΔE_C-TA was over 0.6 eV. This was because the spike was feasibly formed at the TCO/buffer interface, acting as a barrier impeding the collection of photo-generated carriers. As a result, the short-circuit current density (J_SC) was severely reduced. On the other hand, when ΔE_C-TA was lower than −0.2 and −0.4 eV for N_D-Bs of 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³, respectively, the efficiency was drastically decreased, primarily attributable to the decreases in fill factor (FF) and open-circuit voltage (V_OC). The J_SC was slightly changed since the recombination of photo-generated carriers hardly took place owing to the present of the electric field in the SCR. It was consequently suggested that the ΔE_C-TA values for high cell performance were optimized in the ranges from −0.2 to 0.6 eV and from −0.4 to 0.6 eV for N_D-Bs of 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³, respectively.

Zoom In Zoom Out Reset image size

Figure 5 depicts J–V curves of thin-film solar cells with different ΔE_C-TA values as well as the different N_D-Bs of (a) 1.0 × 10¹³ and (b) 1.0 × 10¹⁸ cm⁻³. In the case of N_D-B of 1.0 × 10¹³ cm⁻³, when ΔE_C-TA was less than −0.2 eV, the solar cells showed double diode characteristics, as seen in Fig. 5(a), thereby decreasing FF. Therefore, in spite of the decreases in the cell efficiency with the ΔE_C-TA in a range from −0.6 to −0.3 eV in Fig. 4, V_OC increased owing to the double diode characteristics, as depicted in Fig. 5(a). The excellent J–V curves which were realized with the ΔE_C-TA values in a range between −0.2 and 0.6 eV were almost overlapped, consistent with high efficiency in Fig. 4 in case of N_D-B of 1.0 × 10¹³ cm⁻³. On the other hand, in case of N_D-B of 1.0 × 10¹⁸ cm⁻³, when ΔE_C-TA was below −0.4 eV in Fig. 5(b), the J–V curves of solar cells exhibited double diode characteristics. The excellent J–V curves formed with the ΔE_C-TA values in a range between −0.4 and 0.6 eV were mostly overlapped, corresponding to high efficiency in Fig. 4 in case of N_D-B of 1.0 × 10¹⁸ cm⁻³. In comparison between N_D-B = 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³, the suitable ΔE_C-TA range for N_D-B of 1.0 × 10¹⁸ cm⁻³ was 0.2 eV wider than that for N_D-B of 1.0 × 10¹³ cm⁻³.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 6 shows the energy band diagrams of the solar cells under no bias voltage (black color lines) and threshold bias voltage (V_level) (red color lines) in the case of N_D-B = 1.0 × 10¹³ cm⁻³ and ΔE_C-TAs of (a) −0.2 and (b) −0.6 eV. The V_level was defined as the bias voltage to the level, where the conduction band (CB) peak energy of the spike at either the buffer/absorber interface or TCO/buffer interface was the same as the CB of the absorber bulk. In this case, the CB peak position of the spike at the buffer/absorber interface was the same as the CB of absorber bulk under the bias voltage of V_level, as seen in Fig. 6. The photo-generated carrier flow was consequently impeded by the spike if the bias voltage exceeded V_level. In the comparison between Figs. 6(a) and 6(b), the V_level for ΔE_C-TA = −0.6 eV was smaller than that for ΔE_C-TA = −0.2 eV, implying the easier recombination of photo-generated carriers, when ΔE_C-TA was decreased to the level lower than −0.2 eV. Furthermore, the solar cells showed double diode characteristic owing to the barrier between TCO/buffer in Fig. 5(a) with ΔE_C-TA of below −0.2 eV.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

On the other hand, Fig. 7 shows the energy band diagrams of the solar cells under no bias voltage (black color lines) and bias voltage of V_level (red color lines) in the case of N_D-B = 1.0 × 10¹⁸ cm⁻³ and ΔE_C-TAs of (a) −0.4 and (b) −0.6 eV. In this case, the CB peak position of the spike at the TCO/buffer interface was the same as the CB of absorber bulk under the bias voltage of V_level, as depicted in Fig. 7. It was obviously revealed that the V_level for N_D-B = 1.0 × 10¹⁸ cm⁻³ was higher than that for N_D-B = 1.0 × 10¹³ cm⁻³, meaning that the carrier was easier to recombine in the case of N_D-B = 1.0 × 10¹³ cm⁻³. In addition, Fig. 8 displays the V_level with different N_D-Bs of 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³ as a function of ΔE_C-TA. When the ΔE_C-TA was in a range of 0.0–0.2 eV for N_D-B = 1.0 × 10¹⁸ cm⁻³, the V_level could not be calculated because of the limit of the simulator; however, its value was certainly over 0.9 V. According to Fig. 4, the efficiency of the thin-film solar cells was drastically decreased at the ΔE_C-TAs below −0.2 and −0.4 eV for N_D-Bs of 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³, respectively, which was correlated to the V_level of 0.6 V in both cases, as demonstrated in Fig. 8. It was therefore suggested that the V_level should be 0.6 V or higher for high cell performance. It is also noted that this V_level of 0.6 V for high cell performance is very well consistent with the voltage at maximum power point (V_max) of the J–V curve at the optimized range of ΔE_C-TA in Fig. 5. In addition, we calculated in the case of ΔE_C-BA = 0.0 eV. It was revealed that the results were very similar tendency to above simulation results.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Finally, it was disclosed that the appropriate ΔE_C-TA values for high cell performance were in the ranges from −0.2 to 0.6 eV and from −0.4 to 0.6 eV for N_D-Bs of 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³, respectively, at which the V_level was as high as V_max or higher than it, thus leading to high cell performance. The results clearly explained the effect of ΔE_C-TA on the performance of thin-film solar cell. From the viewpoint of this theoretical analysis, it can contribute to the improvement of the photovoltaic performance by adopting these results in the design of new material combination of buffer, absorber and TCO for high cell performance in the future study.

The device simulation of solar cells was performed using one-dimensional device simulator SCAPS, and the impact of the ΔE_C-TA in thin-film solar cell on cell performance was scrutinized. The results were quantitatively revealed that with ΔE_C-TA above 0.6 eV for both N_D-Bs of 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³, the spike was formed at the TCO/buffer interface, thus acting as a barrier which impedes the collection of photo-generated carriers, severely decreasing cell efficiency. In contrast, with ΔE_C-TA below −0.2 and −0.4 eV for N_D-Bs of 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³, the J–V curves of the solar cells demonstrated double diode characteristics, thereby decreasing cell efficiency. Eventually, the appropriate ΔE_C-TA values for high cell performance were quantitatively suggested to be in the ranges from −0.2 to 0.6 eV and from −0.4 to 0.6 eV for N_D-Bs of 1.0 × 10¹³ and 1.0 × 10¹⁸ cm⁻³, respectively. The results are the useful knowledge for designing and applying new material combination of buffer, absorber and TCO for high cell performance in the future study.

We would like to thank Professor Burgelman and the Department of Electronics and Information Systems at the University of Gent, Belgium, for the development of the SCAPS software package and allowing its use.