Hot‐Pressing Transfer of Diffraction‐Grating Perovskite for Efficient Bifacial Perovskite Solar Cells

Compared with monofacial solar cells, the development of bifacial solar cells has garnered considerable attention for achieving higher power output by simultaneously harvesting direct and diffused light while incurring fewer additional manufacturing costs. Perovskite light absorbers, given their outstanding optoelectronic properties, present great opportunities for fabricating bifacial solar cells. However, one of the challenges in designing bifacial perovskite solar cells (PSCs) is that they suffer from optical and electrical losses due to insufficient light absorption in the perovskite layer. In this respect, in this work, a hot‐pressing transfer process is developed to fabricate a diffraction‐grating (DG) perovskite layer to form a bifacial PSC. The constructed DG structure on the perovskite layer improved its light‐harvesting efficiency, reduced its charge recombination, and enhanced its charge extraction properties. The bifacial DG PSC achieved power conversion efficiencies of 14.01% and 10.04% for the bottom and top illuminations, respectively, resulting in a bifaciality factor of 0.71. Therefore, the newly developed hot‐pressing transfer process for fabricating DG structures on perovskite layers is a promising technique for fabricating high‐efficiency bifacial PSCs with high bifaciality factors.


Introduction
Among various renewable energy forms, photovoltaic (PV) technologies have evolved as extant technologies because of their energy security, reliability, and ubiquitous nature.Over the past 7 years, the global installation of PV modules has skyrocketed to 760 GW, implying its key contribution to transiting global to decarbonized electricity generation. [1]With these advancements, the goal of achieving terawatt (TW)-scale PV modules will be accomplished in the upcoming years.

DOI: 10.1002/admi.202300808
For deploying PV modules at the TW scale, emerging PV technologies have attracted great attention as they have the potential to achieve high-power generation at low costs.One crucial factor to achieve sustainable PV modules is increasing the high-power output per unit area of solar panels at low manufacturing costs.Designing a bifacial solar cell is one of the easiest and most inexpensive strategies to achieve high-power output by simultaneously harvesting direct and diffused light. [2]Lately, perovskite light absorbers, given their long diffusion length, long carrier lifetime, and low surface recombination, have become a fascinating candidate in the development of bifacial PV modules. []Despite the tremendous advantages of perovskite solar cells (PSCs), a major challenge in achieving high efficiency in bifacial configuration is that they suffer from optical and electrical losses as only 70% of incident light is absorbed by perovskite light absorbers. [4]In the case of monofacial PSCs, the presence of a metal electrode increases the light absorption in the perovskite layer because the unabsorbed light can be reflected in the perovskite layer by the metal electrode.In contrast, in the case of bifacial configurations, a metal electrode should be replaced by a transparent electrode.Therefore, the unabsorbed light transmits through the transparent electrode rather than reflecting to the perovskite absorbing layer, resulting in incomplete light absorption in the perovskite layer.Thus, to address the above issues, the search for light-trapping strategies has received great attention, especially in the case of bifacial solar cells.Numerous studies have been conducted to construct microstructures and nanostructures in the perovskite layer to help ameliorate its optical properties. [5,6]Several methods, such as photolithography, [7] inkjet printing, [8] thermal evaporation, [9] and nanoimprint lithography (NIP), [10][11][12] have been adopted in the past few years to develop microstructures and nanostructures in the perovskite layer.However, each method has shortcomings that limit its real applications.The requisite developing and etching step in the photolithography process adversely affects the perovskite layer because of its solubility in various solvents. [13]Meanwhile, although inkjet printing is suitable for nondestructive and template-free patterning, it suffers from low pattern resolution and coffee ring effects during the filmforming process. [14]Further, because of the vapor pressure difference between organic and inorganic components involved in the formation of the perovskite layer, the exploration of the thermal evaporation method is limited. [15]On the contrary, although NIP is considered a conventional patterning method to pattern perovskite layers they cause contamination due to residual materials and damage from the fragility of conventional Si master. [16,17]n this work, we developed a hot-pressing transfer process that allows direct lamination of a diffraction -grating (DG) perovskite layer on the target surface.Unlike the NIP process, which relies on high-end and cost-intensive lithography methods to fabricate master molds and pattern the active layer, we employed a simple solution processing method to directly coat the active layer on a DG stamp and transfer it to the receiver substrate.Therefore, our newly developed process for fabricating a DG perovskite layer provided a facile and low-cost approach without complicated processes and facilities.Moreover, this hot-pressing transfer process is compatible with high-throughput printing techniques, such as roll-to-roll and offset printing.The DG perovskite layer increased the optical path length by diffracted light, thus enhancing the light absorption in the perovskite layer.Further, the DG perovskite layer increased the contact area between the perovskite and the charge extraction layer, thereby enhancing the charge extraction property.The power conversion efficiencies (PCEs) of the DG PSCs were 14.01% and 10.04% for the bottom and top illuminations, respectively, resulting in a high bifaciality factor of 0.71.Compared to pristine PSC, the PCEs of the DG bifacial PSCs herein increased by 16% and 61% for the bottom and top illuminations, respectively.Integrating the advantages of our newly developed hot-pressing transfer method in the fabrication of DG structures on the perovskite layer, this work opens up avenues for fabricating high-efficiency bifacial PSCs with high bifaciality factors.

Results and Discussion
Figure 1 shows the schematics of the hot-pressing transfer process.In our work, we utilized a commercially available compact disc (CD) as an initial master mold with inherent grating structures to form a DG perovskite layer.Although the PDMS stamp, which regains its hydrophobicity regardless of surface modification, [18] is adopted in conventional transfer printing processes, we utilized a UV-curable Norland optical adhesive (NOA) resin as a DG stamp.In contrast to the NIL, we simply obtained a DG pattern on the perovskite layer by spin-coating the perovskite solution on the DG stamp.Then, the DG perovskite layer can be transferred onto the receiver substrate (i.e., the underlying perovskite layer) using the hot-pressing method.In this respect, the stamp should be hot-pressed at a temperature higher than the transition temperature (T r ) to selectively weaken the adhesion between the stamp and the perovskite layer to transfer the DG perovskite layer successfully onto the receiver substrate.In our approach, as we utilized a shape memory polyurethanebased stamp, hot-pressing at a temperature of 110 °C, which was higher than the T r of the stamp, underwent a phase transition to a rubbery state and reduced the conformal contact between the stamp and the perovskite layer, resulting in the transfer of the perovskite layer to the receiver substrate. [19,20]Finally, the bifacial PSCs were fabricated using a fluorine-doped tin oxide (FTO)/TiO 2 /MAPbI 3 /Spiro-OMeTAD/Transparent electrode structure.
We examined the surface morphology of the master mold and DG perovskite layer to confirm the successful replication of the DG structure on the perovskite layer using our transfer process.Figure 2 shows the top and cross-sectional scanning electron microscopy (SEM) images of the master mold and patterned perovskite layer.The master mold had a grating period of 1.5 μm and a 0.5 μm line width (Figure 2a,c).Meanwhile, as can be seen in Figure 2b,d, the DG perovskite layer obtained a grating period of ∼1.5 μm and ∼0.5 μm line width, consistent with those of the master mold.Atomic force microscopy (AFM) images of the master mold and perovskite layer are also shown in Figure 3a,b.The height of the master mold (≈0.16 μm) was replicated on the patterned perovskite layer.Both SEM and AFM results indicated the successful fabrication of a stacked DG perovskite layer via a hot-pressing transfer process.Furthermore, as can be seen in Figure S1 (Supporting Information), the presence of an iridescent reflection on the surface of the DG perovskite layer was due to the optical interference of light by the grating structure, whereas this effect was not visible in the pristine perovskite layer.
Two substrates were pressed together by applying heat and pressure to transfer the DG perovskite layer to the bottom perovskite layer.We investigated the possibility of changes in the crystal structure of the underlying perovskite layer due to the hot-pressing process.Figure 4 shows the X-ray diffraction (XRD) measurements for the DG and pristine perovskite films together with the spin-coated perovskite film that had not undergone the hot-pressing process.The peaks located at 14.08°, 20.02°, 28.4°, and 31.86°wereassigned to the (110), ( 112), (220), and (310) planes of a typical MAPbI 3 tetrahedral structure, respectively. [21]lthough the perovskite layers were subjected to pressure with elevated temperatures, no noticeable peak shifts were observed after hot-pressing.Moreover, a significant PbI 2 diffraction peak was not observed in our samples, attributed to the relatively short period of heating time during hot-pressing.This result implied that both the transfer printing and hot-pressing processes did not influence the crystalline structure and crystallinity of MAPbI 3 .The cross-sectional SEM images shown in Figure 5 represent the device structure of the PSC prepared in this study.During the hot-pressing process, the generation of locally concentrated stress and strain could local contact between the top and bottom perovskite layers, resulting in large voids at the interface. [22]However, as shown in the SEM images, the transferred perovskite layer exhibited no voids at the interface between the two perovskite layers.Moreover, compared with the single perovskite layer that was spin-coated on the same TiO 2 electron transport layer (ETL), the interface between the two perovskite layers could hardly be seen because the contact between them was tight.This result revealed explicitly that the DG perovskite layer was successfully transferred to the bottom perovskite layer using the hot-pressing process.
The DG perovskite layer offered advantages for the optoelectronic performance of PSCs by i) enhancing the light-harvesting efficiency by light diffraction of the grating structure and ii) improving the charge transport process by increasing the contact area between the perovskite and hole transport layer (HTL).First, we measured the adsorption and reflection spectra of the DG and pristine perovskite layers, as shown in Figure 6a,b, to investigate the effects of the DG structure on their optical properties.The absorbance of the DG perovskite film showed a similar trend to that of the pristine perovskite film.Because of the DG structure, the reflectance of the DG perovskite film was lower than that of the pristine perovskite film at the wavelength range of 300-700 nm.This indicated that light diffracted within the perovskite film could be trapped inside the absorber layer by total internal reflection, enhancing light harvesting.Figure 6c shows the light-harvesting efficiency (LHE) of the pristine and DG perovskite films.We calculated the LHE as follows: where R and A represent reflectance and absorbance, respectively. [23]Compared with that of the pristine perovskite film, the LHE of the DG perovskite film was improved because of the decrease in light loss via the DG structure.Therefore, the optical efficiency of the perovskite films could be enhanced by the DG effect of lower light reflection and higher light harvesting at the visible region.The amounts of light absorption were calculated using the 3D finite-difference time-domain (FDTD) simulations to further verify the optical diffraction effect of the grating structure inside bifacial PSCs with the pristine and DG perovskite layers.Figure 7 shows the spatial profile of the power absorbed per unit volume with respect to the xz-plane at a wavelength of 550 nm for bifacial devices with the pristine and DG perovskite layers.In the case of the bottom illumination, the incident light on the pristine and DG perovskite layers was mostly absorbed in the lower part of the perovskite layer near the mesoporous TiO 2 layer, as depicted in Figure 7a,b.Meanwhile, as shown in Figure 7c,d the incident light on the perovskite layers via top illumination was absorbed in the upper part of the perovskite layer near the Spiro-OMeTAD layer.Because of the lower light reflection and higher light harvesting via the DG structure, the amount of light absorption in the DG perovskite layer was enhanced compared with that in the pristine perovskite layer.This indicated that the grating structure could increase the light absorption and number of electron-hole pairs within the perovskite layer, thereby improving the J sc under both top and bottom illumination.
Second, to examine the effects of the DG structure on the charge transport process, we analyzed the steady-state photoluminescence (PL) spectra and time-resolved PL (TRPL) spectra of the pristine and DG perovskite layers.The peak at 770 nm corresponded to fluorescence emission by the MAPbI 3 layer.The PL intensity of the DG perovskite layer was lower than that of the pristine perovskite layer, as shown in Figure 8a.The strong quenching of the PL intensity of the DG perovskite layer indicated efficient charge extraction from perovskite to HTL and the suppression of charge recombination owing to the increased contact area between the DG perovskite and HTL. Figure 8b shows the TRPL decay profile of the pristine and DG perovskite layers.We fitted the TRPL data with multiexponential decay model to quantify the PL lifetime, and the values are presented in Table S1 (Supporting Information). 1 ,  2,  3 , and  4 corresponded to the fast, intermediate, and slow decay components, respectively.The decreased lifetime values of the DG perovskite layer indicated improved carrier extraction from to HTL.The average lifetime of the DG perovskite layer was much lower than that of the pristine perovskite layer, which suggested that photogenerated holes in the DG perovskite layer were more efficiently extracted and transported to HTL because of the presence of a grating structure that accelerates the charge separation. [24]In this respect, we used electrochemical impedance spectroscopy (EIS) to further elucidate the effect of the DG structure on the photogenerated charge carrier transport and recombination.Figure 8c shows the Nyquist plots measured in the dark under the bias voltage of 0.8 V.The inset shows the equivalent circuit model used for fitting the EIS data.The larger semicircle in the EIS data represented the charge recombination resistance (R rec ), [25] and a higher value of R rec corresponded to low carrier recombination.The higher R rec value of the PSC with the DG perovskite indicated that the device exhibited reduced charge recombination and enhanced charge separation.
Finally, for investigating the effects of the DG structure on the PCE of bifacial PSCs, PSCs were fabricated with the following device configuration: FTO/TiO 2 /MAPbI 3 /Spiro-OMeTAD/dielectric/metal/dielectric (DMD) transparent electrode.Figure 9a,b shows the J-V curves of the PSCs with the pristine and DG perovskite layers under the bottom (FTO side) and top (DMD side) illuminations, respectively.The PSC with the pristine perovskite exhibited PCE = 12.10%, J sc = 20.00mA cm −2 , V oc = 0.95 V, and FF = 0.62, whereas the PSC with DG perovskite showed PCE = 14.01%,J sc = 21.50 mA cm −2 , V oc = 0.99 V, and FF = 0.66 under bottom illumination.Under top illumination, the PSC with the DG perovskite exhibited a higher PCE of 10.04% (J sc = 14.50 mA cm −2 , V oc = 0.99 V, and FF = 0.70) compared with the PCE of 6.23% (J sc = 11.50 mA cm −2 , V oc = 0.90 V, and FF = 0.60) in the case of the pristine PSC.Regardless of the illumination direction, the PCE of the DG PSC was always higher than that of the pristine PSC.Compared with that of the pristine PSC, the PCE of the DG PSC increased by 16% (bottom illumination) and 61% (top illumination).These improvements were primarily attributed to the integration effect of the enhancement of the LHE and the charge transport process by the grating structure and showed good agreement with previous optical and electrical analyses.As revealed in the FDTD simulation results, the incident light via top illumination was mostly absorbed in the upper perovskite region near the HTL, whereas incident light via bottom illumination occurred in the lower perovskite region near the ETL.Therefore, the DG structure might have a significant effect under top illumination, and this might have been the reason for the substantial increase in the performance of the DG PSCs when illuminating from the top side.
We also calculated the bifaciality factor (), one of the key traits for evaluating bifacial solar cells, using the formula below: [3] top and  bottom were the PCE under the top and bottom illuminations, respectively.Typically,  ranges from 0 to 1 as the PCE of bottom-side illumination is generally higher than that of top-side illumination.The higher value of  for the DG PSC ( = 0.71) indicates that it had better albedo light-conversion capability than the pristine PSC ( = 0.51).For the realization of a bifacial PSC, a transparent top electrode is required to transmit light from both directions.However, the lack of efficient transparent electrodes limits the performance of bifacial PSCs.In our work, we used a DMD electrode as a transparent electrode to replace the commonly adapted indium-doped tin oxide electrode because they generally require a high kinetic energy deposition technique, which damages the underlying perovskite layer.The performance under top illumination yielded lower PCE than that of the bottom illumination, mainly due to the smaller J sc .
The smaller photocurrent when illuminating from the top side arose from the parasitic absorption of the colored Spiro-OMeTAD and the low transmittance of the DMD electrode, [26] as shown in Figure S3 (Supporting Information).Detailed optimization of the DMD electrode to balance the tradeoff between conductivity and transparency was beyond the scope of this study; therefore, the conventional DMD electrode with MoO 3 /Ag/MoO 3 configuration was adopted.This means that bifacial PSCs could be further optimized to improve the performance under top-side illumination, resulting in a higher bifaciality factor on par with those of commercialized bifacial Si solar cells.In future works, the PV performance of bifacial PSCs could be further improved by precisely optimizing the DMD electrode, among others.Bifacial PSCs can boost the power output because the bifacial PSCs harvest the reflected and diffused sunlight from the ground through the rear side.Generally, the reflected and dif-fused light is incident on the rear side of the bifacial PSCs at various angles, rather than just one angle.Therefore, we examine the change in J sc values according to the illumination at various angles.As shown in Figure 9c, the J sc of DG perovskite PSC exhibits little variation while the pristine PSC shows a linear decrease in J sc as the incident light angle increases.This result demonstrates that the DG PSCs can harvest the reflected and diffused light from the ground more efficiently compared to the pristine PSCs, thus resulting in high efficiency bifacial PSCs.Moreover, the additional power that can be gained by the bifacial PSCs depends on the rear albedo from the ground.We further evaluate the performance of bifacial PSCs under concurrent bifacial illumination with different albedo background.The photovoltaic characteristics of bifacial PSCs were measured under both standard illumination at FTO side and back reflection at DMD side.As shown in Figure 9d, after the introduction of albedo light, the PCEs of bifacial PSCs were increased by 22% and 45% with white paper and aluminum foil backgrounds, respectively.The increase in PCE was primarily attributed to the increase in J sc .Compared to the white paper, aluminum foil as a reflecting surface exhibits higher current density and efficiency due to its high albedo.Thus, the overall PCE of the bifacial PSC can be enhanced with a higher albedo background.Although relatively high albedo backgrounds were adopted to evaluate our bifacial DG PSCs, high albedo greater than 5 is feasible in real-world applications. [27]his implies that bifacial DG PSCs outperform monofacial PSCs under concurrent bifacial illumination.
In addition, monofacial PSCs comprising FTO/TiO 2 /MAPbI 3 /Spiro-OMeTAD/Ag electrodes were fabricated to elucidate the effectiveness of the DG structure.Figure S4 (Supporting Information) shows the J-V curves of the monofacial PSCs with pristine and DG perovskite layers and detailed PV characteristics are summarized in Table S2 (Supporting Information).As expected, the DG PSC yielded higher PCE than the pristine PSC, and the increases in J sc and FF were responsible for such relatively superior performance.As shown in the FDTD simulation results (Figure S2, Supporting Information), the increase in J sc was attributed to the constructive interference and the diffraction by the grating structure.
More importantly, the PCE of the bifacial DG PSC (14.01%) was higher than that of the monofacial pristine PSC (13.85%).Generally, bifacial PSCs show inferior PV performance compared with monofacial PSCs under bottom illumination because of the relatively lower electrical characteristics of the transparent electrode compared with the metal electrode. [28]However, the higher PCE of the bifacial DG PSC compared with that of the monofacial pristine PSC demonstrates that the DG structure is a promising technology for fabricating efficient bifacial PSCs.

Conclusion
In summary, we developed a facile technique to fabricate a DG structure on the perovskite layer using a simple spin-coating process.Then, this DG perovskite layer was transfer-printed to form bifacial PSCs using a hot-pressing process.The DG perovskite layer had advantages in improving light trapping and charge extraction properties and reducing the charge recombination to improve both J sc and FF of the device.When illuminated from the bottom electrode, the bifacial DG PSC showed a PCE of 14.01%.When illuminated from the top electrode, the same PSC achieved a PCE of 10.04%, with a bifaciality factor of 0.71.Both experimental and simulation results showed that the DG structure was required for fabricating efficient bifacial PSCs.The continued optimization of transparent electrodes and colorless HTLs can further enhance the PV performance of bifacial PSCs to a level similar to that of bifacial Si solar cells, with better light collection ability under top illumination.The simplicity of our newly developed process and the effectiveness of the DG structure in bifacial PSCs provide possibilities for the prospective future commercialization of PSCs.

Experimental Section
Master Mold Fabrication: Metal films over a CD were removed using 3 M tape and were then cleaned through ultrasonication in ethanol for 10 min.After drying on the hot plate at 80 °C for 10 min, polydimethylsiloxane (PDMS, Sylgard 184, Dow Corning) solution with a weight ratio of 10:1 (base:curing agent) was cast onto the CD, followed by degassing in a vacuum chamber for 1 h and curing at 80 °C for 2 h.The patterned PDMS film was peeled off from the CD, which served as a master mold.
Diffraction-Grating Stamp Fabrication: This process involves replicating the PDMS master mold to prepare a flexible thin DG stamp.An UVcurable NOA resin was dropped onto the PDMS master mold and covered with a polyethylene terephthalate (PET) substrate to replicate the pattern.An appropriate pressure was applied on the PET surface to remove the trapped air bubbles, followed by UV exposure for 5 min.Finally, a rigid thin NOA stamp covered with PET film was peeled off from the PDMS master mold, where the grating pattern from PDMS was replicated onto the NOA stamp.Then, a thin layer of chromium and silicon dioxide were deposited on the NOA stamp to modify the surface energy of the stamp using an E-beam evaporator at a deposition rate of 1.0 Å s −1 .
Monofacial and Bifacial Device Fabrication: FTO substrates were cleaned through ultrasonication in detergent, acetone, and ethanol for 10 min.ETL and the perovskite layer were subsequently deposited onto the FTO substrate according to the previous report. [24]These layers were herein referred to as the bottom half stack.The bottom and top half stacks were subjected to a hot-pressing process to fabricate the laminated perovskite layer, as described in Section 4.4.After the hot-pressing process, a HTL was deposited onto the perovskite layer with a solution containing (72.3 mg) of Spiro-OMeTAD, (28.8 μL) of 4-tert-butyl pyridine, and Hot-Pressing Transfer Process: A perovskite solution (1 M MAI:PbI 2 ) dissolved in N,N-dimethylformamide and dimethyl sulfoxide was spincoated directly onto the NOA stamp to compose a top half stack.The top MAPbI 3 layer was not fully annealed before hot-pressing, to decrease the local contact probability between the top and bottom MAPbI 3 layers, thereby hindering the formation of large void defects.Then, the fabricated top and bottom half stacks were laminated in a hot-pressing machine at a pressure of 0.2 MPa at 110 °C for 10 min.After the lamination process, the stamp was carefully peeled off from the bottom substrate and the perovskite layers were post-annealed at 100 °C for 10 min.
Characterization: The surface morphologies of the perovskite film were analyzed using SEM (SU8220, Hitachi) and AFM (NX20, Park System).X-ray diffractometers (Bruker AXS, D8-Discover) were acquired to determine the crystallographic changes in the perovskite layer.Steadystate PL spectra and TRPL were measured using MicroTime-200 (Picoquant).The current density-voltage characteristics of PSCs were measured under AM 1.5G illumination using Sol2A (Oriel).Absorbance and reflectance spectra were obtained using a UV spectrometer (Lambda 365).
Optical Analysis of Perovskite Devices: The optical characteristics within the fabricated PSCs were analyzed in a 3D FDTD simulation (Lumerical FDTD Solutions, ANSYS Inc., USA).The FDTD simulation was performed in the 3D region with the antisymmetric boundary conditions along the xaxis, the symmetric boundary conditions along the y-axis, and the perfectly matched layer boundary conditions along the z-axis to compute the spatial profile of the power absorbed per unit volume at specific wavelengths for PSCs.The maximum mesh step of the designed structures was set to 5 nm in both the x-and y-axis directions.An AM1.5 solar spectral source in the wavelength range of 300-800 nm was vertically incident on the designed structure as a plane wave of the top and bottom directions.The electric field and refractive index monitors were located in the designed structure to calculate the amount of light absorption for each layer of the PSCs.

Figure 1 .
Figure 1.Schematic illustration of the hot-pressing transfer process to make a diffraction-grating structure on the perovskite layer.

Figure 2 .
Figure 2. Top and cross-sectional SEM images of a,c) the master stamp and b,d) the diffraction-grating perovskite layer.

Figure 3 .
Figure 3. AFM images of a) the master stamp and b) the diffraction-grating perovskite layer.

Figure 5 .
Figure 5. Cross-sectional SEM images of PSCs with a) the pristine perovskite layer and b) the diffraction-grating perovskite layer.

Figure 6 .
Figure 6.a) Absorbance spectra, b) reflectance spectra, and c) LHE of the patterned and pristine perovskite layers.

Figure 7 .
Figure 7. Spatial profiles of absorbed per unit volume with respect to the xy-plane: a,c) Case of the pristine perovskite layer for the bottom and top illuminations, respectively.b,d) Case of the diffraction-grating perovskite layer for the bottom and top illuminations, respectively.

Figure
Figure a) PL, b) TRPL, and c) EIS measurements of the pristine and diffraction-grating perovskite layers.

Figure 9 .
Figure 9. a,b) J-V of the PSCs with the pristine and diffraction-grating perovskite layers under the bottom and top illuminations, respectively, c) Incident angle versus normalized J sc , and d) Bifacial performance of PSCs with various backgrounds.