All-solution-processed inorganic CsPbBr₃ solar cells and their bifacial-irradiation functions

Abstract

Solution-processed technologies for (semi)transparent top electrodes remain suboptimal, although see-through perovskite solar cells (s-PSCs) are required in realizing window-integrated photovoltaics. Herein, we choose an inorganic perovskite, CsPbBr₃, offering the best matching example with wavelength-selective transparency and weatherability, and present the simplest s-PSC excluding organic components, fluorine-doped tin oxide (FTO)/TiO₂/CsPbBr₃/single-walled carbon nanotube (SWNT). The semitransparent electrode is realized by solution-processed filter-transferred SWNT thin films with different transmittances of 60–80%T at 550 nm. The diode ideal factors range between 1 and 2, suggesting high heterojunction qualities as a single-diode model with hole-transfer-layer-free CsPbBr₃/SWNT. Under monofacial pseudo-sunlight irradiation through FTO, the increased electrical conductivities (densities) of SWNT improve power-conversion efficiencies/short-circuit currents, PCEs (FTO)/J_sc = 8.68/7.49 (60) > 8.18/7.33 (70) > 7.30%/6.91 mA cm⁻² (80%T). Through SWNT, the increased transparency improves PCEs (SWNT)/J_sc inversely as 4.21/3.79 (60) < 4.45/4.13 (70) < 4.56%/4.56 mA cm⁻² (80%T). Open-circuit voltages/fill factors are 1.48/0.79 through FTO and 1.33 V/0.84 through SWNT (60%T). A tradeoff exists between the conductivities and transparency to achieve high performance. Bifacial irradiation using light-emitting diodes shows close values of PCEs (bifacial) = 3.67 (60), 3.86 (70), and 3.71% (80%T) based on 32–35% of pseudo-sunlight power (100 mW cm⁻²), equivalent to the sums of the monofacial-irradiation PCEs (FTO) and PCEs (SWNT). Enhancement ratios of PCEs (bifacial) to PCEs (SWNT)/to PCEs (FTO) are 3.19/1.36 (60), 3.04/1.46 (70), and 2.81/1.53 (80%T). The bifacial function solves the monofacial tradeoff. The black color of SWNT is not a serious obstacle visually under exterior environments.

---

Introduction

For the long-term operation of organic–inorganic perovskite solar cells (PSCs), improving stability under realistic environments is a hot issue. Cs⁺ displaces the organic cations to form all-inorganic perovskite layers to address this issue.^1,2 CsPbI₃ with a small bandgap energy (E_g) of 1.7 eV, which has a similarity to CH₃NH₃PbI₃ of E_g = 1.5 eV in the iodide composition, struggles to maintain the thermodynamically stable perovskite phase because of the size mismatch between Cs⁺ and I⁻ in the tolerance factor.³ By increasing x of CsPbI_3−xBr_x from 0 to 3, the size mismatch is solved to stabilize all-inorganic perovskite structures, and simultaneously, E_g increases from 1.7 (729) to 2.3 eV (539 nm).^1,2 The increasing number of Br⁻ is a tradeoff; for instance, all-bromide CsPbBr₃ (x = 3) is long-life with the highest weatherability against moisture, heat, UV light, and/or oxygen, but the wide E_g of 2.3 eV cannot utilize long-wavelength visible lights of >539 nm and decrease the PCE up to 17% of the Shockley–Queisser limit.^1,2 Nevertheless, the wavelength-selective transparency is a strong merit when window-integrated see-through PSCs (s-PSCs) are constructed.⁴ Sublayers with a perovskite layer can be completed by bottom-up solution processes on a commercial indium-tin-oxide (ITO) or fluorine-doped-tin-oxide (FTO) substrate (Fig. 1a–c); however, solution-processed technologies for (semi)transparent top electrodes remain a challenge because the perovskite layers are solvent-sensitive to decompose by exposure to water and hydrophilic solvents.⁵

Fig. 1 (a–c) Schematic n-i-p PSCs with different electrodes using (a) metals, (b) carbon pastes, and (c) single-walled carbon nanotubes (SWNTs). Arrows show light pathways. (c) In this study, semitransparent SWNT thin films are employed to fabricate all-inorganic hole-transfer-layer (HTL)-free s-PSCs and to investigate bifacial-irradiation functions by systematically controlling transmittance values of the SWNT thin films to 60, 70, and 80%T at 550 nm. (d) Schematic procedures including a solvent-replacement process from water to toluene to transfer a SWNT thin film onto a solvent (water)-sensitive perovskite layer without decomposition.⁴¹

To date, the photovoltaic performance of PSCs has been significantly enhanced by comprehensive research using optically opaque electrodes (Fig. 1a and b). Such one-sided (monofacial) irradiation solar cells cannot fully utilize reflected irradiance (albedo) in realistic environments, and setting their panels requires optimal tilt angles for an ideal irradiance of AM 1.5G. Recently, the development of (semi)transparent electrodes for exploring bifacial-irradiation functions of PSCs has emerged as a new target of research, moving from basic research to future innovative applications.^4,6–10 Specifically, (i) (semi)transparent electrodes show potential for enhancing PCEs by bifacial irradiation (Fig. 1c), and (ii) see-through characteristics due to a high average visible-light transmittance are of particular interest in realizing window-integrated s-PSCs.⁴ The PCE enhancement effects (i) have been reported to only a limited extent by using narrow-E_g organic–inorganic PSCs but not s-PSCs. In 2021, Nasibulin et al. placed a dry-processed single-walled-carbon-nanotube (SWNT) thin film on a perovskite layer.⁶ In the hole-transfer-layer (HTL)-free PSC of ITO/SnO₂/phenyl-C₆₁-butyric acid/organic–inorganic perovskite/SWNT, the monofacial-irradiation PCEs through ITO and semitransparent SWNT were only reported to be 16.7 and 6.9%, respectively. In 2022, Grätzel et al. constructed FTO/SnO₂/organic–inorganic perovskite/spiro-OMeTAD by contacting as a top semitransparent electrode, carbon nanotubes on an FTO or ITO substrate, at a pressure of 4.7 kPa using two clips. Reflected irradiance was utilized as a bifacial function.¹¹ In 2023, Omer et al. fabricated FTO/TiO₂/organic–inorganic perovskite/spiro-OMeTAD using evaporated gold electrodes with thicknesses between 17.5 and 27.5 nm, and ITO was further deposited by magnetron sputtering to prepare a top semitransparent electrode of Au/ITO.¹² By bifacial irradiation using light-emitting diodes (LEDs) (40 mW cm⁻²) and pseudo sunlight (100 mW cm⁻²), the enhancement ratio was 1.14 as PCEs increased to 20.1% from 17.7% of the monofacial irradiation. In 2024, Zhang et al. excluded a bottom ITO/FTO substrate to form glass/SWNT/Cu:NiO_x/organic–inorganic perovskite/SnO₂/PCBM/SWNT with dry-processed semitransparent SWNT thin films in both electrodes.¹³ The bifacial-irradiation performance was investigated by using the different-transmittance SWNT between 75 and 95%T at 550 nm. The ratios (bifaciality factors) between monofacial-irradiation PCEs showed the highest value of 0.9827 through the back and front SWNT electrodes with the same transmittance at 85%T. Related to (ii), s-PSCs using wide-E_g perovskites such as CsPbBr₃,^8,14–17 FAPbBr₃ (FA = formamidinium ion),^18,19 and CsPbCl_2.5Br_0.5,²⁰ were reported; however, the bifacial-irradiation effect to enhance PCE (i) remains unexplored.

Mirror-like materials, used as opaque top electrodes, can reflect light back to perovskite layers, which is believed to nominally increase PCEs (Fig. 1a).²¹ Highly conductive low-cost silver and copper are promising electrodes but are corroded by halides migrated from perovskite layers through hole/electron-transfer layers (HTLs/ETLs).^22–25 Evaporated gold is a practical choice as a less-corrosive electrode; however, the use of gold raises serious concern for the commercialization of PSCs, because gold speculation has continuously increased its price,²⁵ and recovery technologies must be developed for disposal.^26,27 Conductive carbon can directly contact CsPbBr₃ layers to operate as a representative HTL-free PSC.^28–40 Most importantly, carbon electrodes prevent metal-like corrosion with halides and comprise elemental-risk-free ubiquitous materials. Among these, carbon pastes are cost-effective. The thick carbon-paste electrodes increase their electrical conductivity for obtaining high-performance PSCs; however, they result in opacity by absorbing all the visible light even if it passes through wavelength-selective-transparent perovskite layers (Fig. 1b).

As mentioned above, CsPbBr₃ is beneficial to realizing s-PSCs using the advantages of the highest wavelength-selective transparency, and simultaneously, its intrinsically high weatherability achieves long-term operation of s-PSCs by excluding organic HTLs and ETLs and by replacing corrosive metal electrodes with elemental-risk-free carbon. However, the bifacial-irradiation functions (i) of such all-inorganic s-PSCs (Fig. 1c) remain unexplored. It is expected that semitransparent SWNT thin films by solution processes will be a breakthrough technology to accelerate the comprehensive research of see-through photovoltaics because wide-field researchers can readily manipulate aqueous dispersion solutions of commercial SWNTs by using common and inexpensive laboratory tools. Therefore, in 2021, we reported a solution-processed press-free filter-transfer method of semitransparent SWNT thin films onto various substrates.⁴¹ A relationship between visible-light transmittance at 550 nm and sheet resistance values of the thin films was experimentally determined by controlling concentrations of SWNTs in their aqueous dispersion solutions.^41–43 We preliminarily revealed that a semitransparent SWNT thin film was transferable onto a solvent-sensitive CH₃NH₃PbI₃ layer by a careful treatment of solvents, i.e., replacement of water with hydrophobic solvents such as toluene and ethyl acetate (Fig. 1d).⁴¹ In 2022, the solution process was applied to prepare the semitransparent thin films stacked by silver nanowires for improving electrical conductivities.⁴⁴

In this study, we fabricate solution-processed all-inorganic s-PSCs. To construct the simplest HTL-free model of FTO/TiO₂/CsPbBr₃/SWNT (Fig. 1c), a semitransparent SWNT thin film is directly placed on a CsPbBr₃ layer, using our filter-transfer method (Fig. 1d),⁴¹ and the SWNT densities are adjusted to approximately 60, 70, and 80%T of visible-light transmittance at 550 nm. Diode ideal factors range between 1 and 2, indicating that the HTL-free PSC functions as a single-diode model to afford superior fill factor (FF) values of 0.77–0.84. By monofacial irradiation through FTO using pseudo sunlight (100 mW cm⁻²), PCEs are improved in the order of 8.68 (60) > 8.18 (70) > 7.30% (80%T) by increased electrical conductivities of SWNT. By monofacial irradiation through SWNT, the increased transmittance of SWNT improves PCEs in the reversed order of 4.21 (60) < 4.45 (70) < 4.56% (80%T). By bifacial irradiation via LED-light power between 32 and 35% of the pseudo sunlight, pseudo-sunlight-based PCEs are 3.67 (60), 3.86 (70), and 3.71% (80%T), practically equivalent to the sums of the monofacial ones through FTO and SWNT. It is revealed that a tradeoff between the transmittance and electrical conductivities of SWNT is essential in affecting the monofacial-irradiation performance, while bifacial irradiation can solve the tradeoff.

Results and discussion

Preparation of CsPbBr₃ layers

According to the two-step spin-coating method using two precursor solutions of PbBr₂ and CsBr,^33,45–48 a CsPbBr₃ layer was constructed on a mesoporous-TiO₂ layer. This was layered atop an FTO glass substrate covered by a compact-TiO₂ layer, denoted as FTO/c-TiO₂/mp-TiO₂. To suppress the generation of voids/pinholes in the CsPbBr₃ layer, the optimal concentrations of PbBr₂ must be determined. A base layer of PbBr₂ was formed on FTO/c-TiO₂/mp-TiO₂ by spin coating using three different concentrations of PbBr₂: 1.20, 1.30, and 1.40 mol L⁻¹, all dissolved in N,N-dimethylformamide solutions. A methanol solution of CsBr at a low concentration of 7.00 × 10⁻² mol L⁻¹ was spin-coated onto the previously prepared FTO/c-TiO₂/mp-TiO₂/PbBr₂ substrate. After heating at 250 °C, the phase change into a yellow perovskite of CsPbBr₃ was monitored by X-ray diffraction (XRD) patterns. In a typical example using 1.40 mol L⁻¹, an initial CsPb₂Br₅ phase appeared and was gradually changed into CsPbBr₃ by repeating the spin-coating and heating processes (Fig. S1a, ESI†). After nine repetitions, the strong XRD signal from CsPb₂Br₅ at 11.6° almost disappeared to leave single-component signals of CsPbBr₃. To obtain a similar XRD pattern (Fig. S1b and c, ESI†), the spin-coating numbers were reduced to 7 and 8 by decreasing the concentration to 1.20 and 1.30 mol L⁻¹, respectively. According to the previous reports, an excess number of spin-coating applications caused decreasing PCEs due to the accumulation of an impurity of CsBr-rich Cs₄PbBr₆,³³ while the presence of CsPb₂Br₅ at the interface of CsPbBr₃ grains improved PCEs as an effective electron transport layer.⁴⁹ Therefore, with the determined number of spin-coating repetitions, small-intensity XRD signals of CsPb₂Br₅ were permitted but the formation of the impurity phase of Cs₄PbBr₆ was inhibited (Fig. S1, ESI†). In top-view FE-SEM images (Fig. S2, ESI†), the base PbBr₂ layers showed many voids exposing TiO₂ particles. As a result, the formation of void-less PbBr₂ layers proved difficult within the practical concentration range between 1.20 and 1.40 mol L⁻¹ but the voids almost disappeared by the second spin-coating procedure using the methanol solution of CsBr, vide infra.

As mentioned above, the methanol solution of CsBr was repeatedly spin-coated on the precursor layer of PbBr₂ to almost change into perovskite CsPbBr₃ from CsPb₂Br₅. In top-view FE-SEM images at the same magnification (Fig. 2), voids of the precursor layer (Fig. S2, ESI†) were surprisingly diminished; however, some pinholes remained in the CsPbBr₃ layer prepared from the low-concentration PbBr₂ of 1.20 mol L⁻¹ (Fig. 2a) while they almost disappeared in the cases of 1.30 and 1.40 mol L⁻¹ (Fig. 2b and c). In magnified top-view FE-SEM images (Fig. 2, inset), large grains with dimensions up to 1 μm were observed. A significant difference was seen in a comparison of cross-sectional FE-SEM images of 1.20 and 1.40 mol L⁻¹ (Fig. 2d and e): the CsPbBr₃ layer was composed of accumulated grains on a mp-TiO₂ layer in the case of 1.20 mol L⁻¹. In contrast, the CsPbBr₃ layer was formed by monolayer-aligned large gains^33,45–48 in the case of 1.40 mol L⁻¹, and its thickness was ∼300 nm. The CsPbBr₃ grains showed similar average surface roughness measurements of 34.9 (1.20), 31.2 (1.30), and 32.7 nm (1.40 mol L⁻¹), based on atomic force microscope (AFM) images (Fig. S3, ESI†). A UV-Vis-near IR absorption spectrum of FTO/c-TiO₂/mp-TiO₂/CsPbBr₃ (1.40 mol L⁻¹) showed a sharp absorption at 505 nm, characteristic of a direct-transition semiconductor, and E_g was estimated as 2.36 eV from a Tauc plot (Fig. S4, ESI†).

Fig. 2 (a–c) Top-view FE-SEM images of CsPbBr₃ layers on mp-TiO₂/c-TiO₂/FTO using a PbBr₂ solution of (a) 1.20, (b) 1.30 or (c) 1.40 mol L⁻¹. Inset: magnified images. (d and e) Cross-sectional FE-SEM images in the cases of (d) 1.2 and (e) 1.40 mol L⁻¹.

Spontaneous transfer of a semitransparent SWNT thin film onto a CsPbBr₃ layer

In 2010, Kauppinen et al. developed a dry-processed filter-transfer method for applying semitransparent SWNT thin films to various substrates via pressurized contact,⁵⁰ a pioneering approach for transferring them onto solvent-sensitive organic–inorganic and inorganic perovskite layers.^6,13,51–55 Various solution-processed methods have been developed using aqueous dispersion solutions of commercial SWNTs.^42,56 It has been demonstrated that both dry- and solution-processed semitransparent SWNT thin films function as p-type semiconductors and high-conductive electrodes to fabricate p–n heterojunction silicon solar cells.^44,57–61 Nevertheless, conventional solution processes have suffered from poor solvent selectivity when applied to solvent-sensitive materials. To address this drawback, we successfully transferred a semitransparent SWNT thin film onto a CsPbBr₃ layer by the solvent-replacement process from water to toluene (Fig. 1d).^41,44,62 The filtrated thin film on toluene-wetted polytetrafluoroethylene (PTFE) membrane was placed on the CsPbBr₃ layer without any pressure with the assistance of the solvent-wettability and flexibility of the PTFE membrane. After the toluene evaporated, the thin film was spontaneously transferred onto the CsPbBr₃ layer by peeling off the PTFE membrane. To systematically investigate the photovoltaic performance of s-PSCs depending on the transparency of SWNT thin films, the SWNT densities were adjusted to exhibit three different transmittance values of approximately 60, 70, and 80%T at 550 nm when the thin films were transferred onto glass substrates to measure UV-Vis-near IR transmittance spectra (Fig. 3a). The thin films maintained high transmittance values in the visible and near-IR range at >500 nm while showing decreased transmittance in the UV region. We used commercial SWNTs with a nominal diameter of 1.5 nm; however, in top-view FE-SEM images, they formed bundles with an average width between 6 and 60 nm (Fig. S5, ESI†).⁴¹ The long-length bundles are effective in improving the electrical conductivities of the thin films. As transparency increased, average sheet-resistance values increased (Table 1); 47 < 65 < 115 Ω cm⁻² at transmittances of 62.3 > 70.7 > 79.3%T at 550 nm (Fig. 3a).^41,43 After the semitransparent thin films were transferred onto the CsPbBr₃ layers prepared from the different PbBr₂ concentrations of 1.20, 1.30, and 1.40 mol L⁻¹, the XRD patterns of CsPbBr₃ were maintained without significant change (Fig. S6, ESI†), and simultaneously, the specific yellow color of CsPbBr₃ was unchanged in an all-inorganic s-PSC of FTO/c-TiO₂/mp-TiO₂/CsPbBr₃/SWNT at 60%T (Fig. 3b). Consequently, the solution-processed filter-transfer method prevents the decomposition of solvent (water)-sensitive CsPbBr₃ during the s-PSC production.

Fig. 3 (a) UV-vis-near IR transmittance spectra of different-density SWNT thin films transferred onto glass substrates. For example, SWNT thin films of 79.3, 70.7, and 62.3%T at 550 nm show averaged sheet resistance values of 115, 65, and 47 Ω cm⁻², respectively. (b) Photograph and schematics of an all-inorganic s-PSC of FTO/c-TiO₂/mp-TiO₂/CsPbBr₃/SWNT at 60%T. The visible-light transmittance values of SWNT thin films are systematically controlled at approximately 60, 70, and 80%T. The SWNT thin film can be cut with scissors to obtain suitable areas and shapes. To improve the electrical contact of SWNT with probes of a hand-made jig, vacuum-evaporated gold is used as an auxiliary electrode. Using the solar cell, the photovoltaic performance is evaluated via four different electrode areas of the SWNT thin film. (c and d) Photographs of the s-PSC at 80%T taken through environmental light in (c) clear and (d) cloudy weather. As shown in a red dotted line, a semicircle-shaped SWNT thin film is placed on a yellow CsPbBr₃ layer but its black color is not a serious obstacle visually. (a and b) Note that the photographs taken by reflecting light from a white substrate emphasize the black color.

Table 1 Photovoltaic parameter values of s-PSCs in monofacial irradiation through FTO and SWNT using pseudo sunlight^a

	J sc/mA cm^−2	V oc/V	FF	PCE (%)	Integrated Jsc/mA cm^−2	Hysteresis index (%)	R/Ω cm^−2	R s/Ω cm^2	Diode ideal factor
a Integrated Jsc values are estimated from EQE spectra (Fig. 4g and h). R is sheet-resistance values of SWNT thin films. Rs and diode ideal factors are based on Fig. S8 and S9, ESI, respectively.
Through FTO
60%T	7.49	1.48	0.79	8.68	6.85	7.72	47	6.2	1.9
70%T	7.33	1.40	0.79	8.18	6.80	2.19	65	7.0	1.6
80%T	6.91	1.34	0.79	7.30	6.57	4.28	115	18	1.5
Through SWNT
60%T	3.79	1.33	0.84	4.21	3.71	11.6
70%T	4.13	1.32	0.82	4.45	4.09	16.7
80%T	4.56	1.30	0.77	4.56	4.24	15.3

---

Monofacial-irradiation performance using pseudo sunlight through FTO and its dependence on PbBr₂ concentrations

By using s-PSCs at 60%T (Fig. 3b), the photovoltaic performance was investigated (Fig. S7, ESI†) under monofacial irradiation through FTO at an area of 4.90 × 10⁻² cm². To evaluate the inherent monofacial-irradiation performance, the reflecting light was prevented by a black substrate placed below the semitransparent SWNT electrode. The photovoltaic performance was improved by increasing the concentrations of PbBr₂ from 1.20 to 1.40 mol L⁻¹. In the J–V curves of 1.40 mol L⁻¹ (Fig. S7a, ESI†), the best PCE was 8.68% in the reverse scan, and other photovoltaic parameter values were J_sc = 7.49 mA cm⁻², V_oc = 1.48 V, and FF = 0.79. As the PCE decreased to 8.01% in the forward scan, the hysteresis index (HI) was 7.72%, calculated as (PCE_reverse – PCE_forward)/PCE_reverse × 100 (%). The PCE of 8.68% was listed as a high value compared to those of previously reported HTL-free opaque PSCs of FTO/TiO₂/CsPbBr₃/carbon paste (Table S1, ESI†).²⁸ In the J–V curves of 1.30 and 1.20 mol L⁻¹ (Fig. S7b and c,† ESI), the best PCEs were 7.41 and 7.36% in the reverse scans, respectively. The photovoltaic performances were inferior to the case of 1.40 mol L⁻¹; in a typical case of 1.20 mol L⁻¹, all the parameter values were decreased at J_sc = 7.08 mA cm⁻², V_oc = 1.38 V, and FF = 0.75 in the reverse scan, and the HI showed a low value of 2.99%. It is noted that insufficient contact is unavoidable at the interfaces between perovskite layers and carbon-paste electrodes composed of rigid graphite flakes with binders. To address this problem, in 2022, Zhang et al. improved the interfacial contact between a CsPbBr₃ layer and a carbon-paste electrode by adding a hot-pressing process.³⁰ The PCE was successfully improved from 7.40% to 8.34%. The PCE enhancement was mainly due to an increased J_sc from 6.80 to 7.18 mA cm⁻² and FF from 0.75 to 0.79 at almost the same V_oc of 1.45 V. The hysteresis was mitigated with the decreased HI from 14.9% to 6.8%. In the CsPbBr₃ layer at the optimal PbBr₂ concentration of 1.40 mol L⁻¹, the comparably high PCE (8.68%) and low HI value (7.72%) suggest that flexible SWNT bundles can sufficiently adhere to CsPbBr₃ large grains (Fig. 2e) with a surface roughness of >30 nm (Fig. S3, ESI†) even through press-free (spontaneous) transfer.

Monofacial-irradiation performance using pseudo sunlight through SWNT/FTO and its dependence on SWNT densities

At the optimal PbBr₂ concentration of 1.40 mol L⁻¹, the photovoltaic performance was investigated based on the transmittance values of 60, 70, and 80%T for SWNT (Fig. 4a–c). The photovoltaic parameter values are summarized in Table 1. When pseudo sunlight was irradiated through FTO, the best PCEs (FTO) decreased to 8.68 > 8.18 > 7.30% with an increase in transmittance values of 60 < 70 < 80%T. This change was associated with decreasing SWNT densities (Fig. 3a). In the most specific feature, as synchronized with the PCE (FTO) change, the systematic decrease in J_sc (FTO) and V_oc (FTO) was found in the order of J_sc (FTO) = 7.49 (60) > 7.33 (70) > 6.91 mA cm⁻² (80%T) and V_oc (FTO) = 1.48 (60) > 1.40 (70) > 1.34 V (80%T). The sheet resistance (R) of the SWNT thin films increasing in the order of 47 (60) < 65 (70) < 115 Ω cm⁻² (80%T) is mainly responsible for the decreasing trend of J_sc (FTO) and V_oc (FTO) because pseudo sunlight was irradiated through FTO with the same conditions. Indeed, affected by the sheet resistances, the series resistance (R_s) values were increased in the order of 6.2 (60) < 7.0 (70) < 18 Ω cm² (80%T) (Fig. S8, ESI†).

Fig. 4 (a–c) J–V curves (reverse scans) of s-PSCs at different transmittance values of (a) 60, (b) 70, and (c) 80%T under monofacial irradiation using pseudo sunlight (AM 1.5G, 100 mW cm⁻²) through FTO (red lines) and SWNT (black lines) at an area of 4.90 × 10⁻² cm². (d–f) Top-view SEM images of s-PSCs at different transmittance values of (d) 60, (e) 70, and (f) 80%T. (g–i) EQE spectra under monofacial irradiation through (g) FTO, (h) SWNT, and (i) an overlay of (g) and (h) at 60%T.

Despite the SWNT densities being changed between 60 and 80%T, FF (FTO) maintained the same high value of 0.79. Although the diode ideal factors have received less attention in recent PSC studies, they should be evaluated from J–V curves in the dark to understand why the s-PSCs can maintain such high FF via the different SWNT densities. As we reported in 2022,⁴⁴ a SWNT thin film (80%T) was directly placed onto an n-type silicon wafer to fabricate a p–n heterojunction solar cell by the same solution-processed filter-transfer method (Fig. 1d). The diode ideal factor (m) was estimated as 2.4 based on a J–V curve in the dark but largely deviated from m = 1 as a single-diode model. The bundle formation of SWNTs (Fig. S5, ESI†) weakened their inherent p-type semiconductor character and resulted in a low FF value of 44% by facilitated carrier recombination. When the SWNT thin film was doped using Nafion, the improved semiconductor character significantly decreased the m value to 1.3, and simultaneously, increased the FF value to 70%. Similarly, in PSCs, m values are a measure to judge the heterojunction qualities influencing the carrier recombination at the interfaces of ETL/perovskite and perovskite/HTL. In 2018, Liao et al. proposed a double-diode model to solve the dilemma of m ≫ 2 in many cases of PSCs.⁶³ In 2019, Ito et al. revealed that the m value drastically dropped from 5 to 1.4 by introducing a buffer layer of bathocuproine (BCP) between [6,6]-phenyl-C₆₁-butyric acid methyl ester (PCBM) and a silver electrode in a p–i–n PSC of FTO/NiO/CH₃NH₃PbI₃/PCBM/Ag.⁶⁴ By changing a Schottky-barrier diode of PCBM/Ag to an ohmic contact of PCBM/BCP/Ag, the m values ranged between 1 and 2, consistent with a single-diode model, and FF and J_sc values were improved to increase PCEs. The unnecessarily increased thickness of BCP resulted in decreased FF and J_sc because the increased R_s facilitated charge recombination at the interface of CH₃NH₃PbI₃/PCBM. Thus, when PSCs exhibit m ≫ 2, it is considered that Schottky-barrier diodes appear on both sides with the n- and p-type semiconductors as a double-diode model. In this study, the s-PSCs showed the typical J–V curves in the dark comprising recombination currents in the diode space-charge region, diffusion currents, and diode diffusion currents limited by series resistances.⁶³ Based on the linearities of the diffusion currents versus applied voltages (Fig. S9, ESI†), the estimated m values ranged between 1 and 2, showing an increasing trend as m = 1.9 (60) > 1.6 (70) > 1.5 (80%T) with densifying SWNTs (Table 1). Accordingly, the s-PSCs can be regarded as a single-diode model suggesting that the HTL-free SWNT electrode does not act as a Schottky-barrier diode. The large energy gap between CsPbBr₃ (−5.6 eV) and SWNT (−5.0 eV) for hole transfer is effective in suppressing charge recombination at the interface of CsPbBr₃/SWNT. However, this large gap also results in a decrease in V_oc. The p-type semiconductor character of SWNTs is weakened with the enhanced metal-like character by bundle formation. This is probably responsible for the increasing trend of m as a negative factor while increasing the SWNT densities endows a positive factor to decrease R_s. Thus, in the HTL-free SWNT electrodes, a tradeoff relationship between m and R_s values was found by changing SWNT densities. The high heterojunction qualities of mp-TiO₂/CsPbBr₃ and CsPbBr₃/SWNT showing m < 2 effectively suppressed charge recombination. As a result, the same high FF (FTO) values of 0.79 between 60 and 80%T would be obtained by the balanced tradeoff.

When pseudo sunlight was irradiated through the semitransparent SWNT thin films, the inherent monofacial-irradiation performance was evaluated by similarly blocking the reflecting light from the rear FTO side (Fig. 4a–c and Table 1). Contrary to the irradiation through FTO, the best PCEs (SWNT) were improved in the reversed order of 4.21 (60) < 4.45 (70) < 4.56% (80%T) with decreases in the SWNT densities. The incident light was interrupted through the SWNT thin films by decreasing their transmittance values (increased densities) (Fig. 3a); thereby, J_sc (SWNT) values were reduced as 3.79 (60) < 4.13 (70) < 4.56 mA cm⁻² (80%T) with decreased numbers of photons passing through SWNT. V_oc (SWNT) values ranged between 1.30 and 1.33 V which were lower than V_oc (FTO) values while showing a similar decreasing trend to V_oc (FTO) with the SWNT densities. FF (SWNT) also shows a decreasing trend of 0.84 (60) > 0.82 (70) > 0.77 (80%T) with the SWNT densities. HI values were higher through SWNT than those through FTO. Thus, we have first revealed a systematic effect on the photovoltaic performance depending on the densities of SWNT thin films for changing their transmittance and sheet-resistance values. As discussed in the next subsection on bifacial-irradiation performance, considering that the incident light is almost absorbed in the perovskite layer near the interfaces of mp-TiO₂/CsPbBr₃ and CsPbBr₃/SWNT, the migration distances of charge-separated electrons to mp-TiO₂ and holes to SWNT are different between the monofacial irradiation through FTO and SWNT. However, at this stage, the negative mechanism due to the SWNT thin films remains unclear as to how V_oc (SWNT) drops and the HI values are elevated through SWNT. The statistical distributions of the photovoltaic parameter values via the monofacial irradiation through FTO and SWNT indicated that the J_sc and PCE values were synchronously changed with transmittance values of the SWNT thin films and were in a reverse relationship through FTO and SWNT (Fig. S10, ESI†). Densifying the SWNT thin films contributed to an increased trend of FF and V_oc. V_oc (FTO) was higher than V_oc (SWNT), while FF (FTO) was lower than FF (SWNT) at each transmittance value. As shown in the top-view SEM images of s-PSCs (Fig. 4d–f), the SWNT bundles directly adhered to the CsPbBr₃ layers, and their densities visually decreased with increased transmittance values. In the case of 60%T, the CsPbBr₃ layer seemed to be sufficiently covered by the bundles. Naked surfaces of the CsPbBr₃ grains were observed through the uncovered spaces and their areas were increased from 60 to 80%T. Accordingly, the contact area of the CsPbBr₃ layer with the bundles is an essential factor that influences the photovoltaic parameter values in the monofacial irradiation through not only SWNT but also FTO. As seen in the cross-sectional FE-SEM images (Fig. S11, ESI†), the SWNT bundles cannot penetrate through the narrow spaces between monolayer-aligned large gains.

Aiming for applications in tandem and window-integrated photovoltaics, s-PSCs of CsPbBr₃ have been reported limitedly.^14–17 For example, the top semitransparent electrodes were sputtered ITO,^14,15 a bilayer of ITO nanoparticles and Ag nanowires,¹⁶ and PEDOT:PSS;¹⁷ however, investigations focused solely on monofacial-irradiation performances through the bottom electrodes of FTO and ITO. Compared to the photovoltaic parameter values summarized in Table S2, ESI,† our solution-processed s-PSCs showed superior photovoltaic performance, although they were the simplest HTL-free structure excluding organic molecules.

External quantum efficiency (EQE) spectra were compared based on the transmittance values of SWNT (Fig. 4g–i). In the monofacial irradiation through FTO, EQEs (FTO) steeply increased at shorter wavelengths of ≤550 nm, derived from the absorption of photons having energies over the E_g of CsPbBr₃ (Fig. S4, ESI†). In a visible region, the EQEs (FTO) reached 90%, with maximum values at 92.2 (60), 91.5 (70), and 89.9% (80%T) at around 405 nm. In a UV range, they were abruptly decreased by absorption of the c-TiO₂/mp-TiO₂ layer with E_g = 3.2 eV (388 nm). In the visible region, the high EQEs (FTO) were consistent with the previously reported ones achieving high J_sc over 7 mA cm⁻².^{17,33,37–40} Based on the decreasing trend of J_sc (FTO) = 7.49 (60) > 7.33 (70) > 6.91 mA cm⁻² (80%T) from the J–V curves, the integrated J_sc(FTO) values from the EQE spectra showed a similar trend of 6.85 (60) > 6.80 (70) > 6.57 mA cm⁻² (80%T) (Table 1). The latter values are smaller than the former values by 0.64 (60), 0.53 (70), and 0.34 mA cm⁻² (80%T) with deviations of 9.3 (60), 5.3 (70), and 3.4% (80%T). When mp-TiO₂ layers were employed as a scaffold of CsPbBr₃, the inconsistency in J_sc was reported with a deviation of ∼30%.⁴⁸ Considering that the wavelength-scanning light for EQE spectra is weaker than the pseudo sunlight for J–V curves, the number of photons to reach the CsPbBr₃ would be influenced in the UV range by the absorber of mp-TiO₂. The UV-region EQE (FTO) significantly deteriorated in the case of the low-density SWNT thin film at 80%T (Fig. 4g). The difference in EQEs (FTO) in the UV region may be caused by a relationship between low numbers of photons passing through the c-TiO₂/mp-TiO₂ layer and the reduced hole-transfer ability (increased R_s) by decreasing the SWNT density (Fig. 4d–f). In contrast to EQEs (FTO), EQEs (SWNT) in the monofacial irradiation through SWNT were maintained without significant decay even in the UV region (Fig. 4h and i) because the incident light could directly reach the CsPbBr₃ layer without the UV-light absorber of c-TiO₂/mp-TiO₂. The EQEs (SWNT) ranged between 45 and 55% in the visible-light region between 400 and 550 nm, and exceeded 30% in the UV region. In not only the visible range but also the UV range, larger numbers of photons can pass through the decreased-density SWNT thin films. As a result, J_sc (SWNT) of the J–V curves increasing as 3.79 (60) < 4.13 (70) < 4.56 mA cm⁻² (80%T) exhibited near values to the integrated J_sc (SWNT) as 3.71 (60) < 4.09 (70) < 4.24 mA cm⁻² (80%T) (Table 1). It is revealed that the increased transmittance values are a critical factor in improving EQEs (SWNT). As a result, the positive EQE effect increased the monofacial-irradiation J_sc ratios in the order of J_sc (SWNT)/J_sc (FTO) = 0.506 (60) < 0.563 (70) < 0.660 (80%T).

Bifacial-irradiation performance using LED and pseudo sunlight and its dependence on SWNT densities

It has been regarded that the photo-reflection from metal-luster electrodes enhances PCEs when pseudo sunlight is irradiated through transparent FTO and ITO electrodes (Fig. 1a).²¹ We first investigated the positive effect due to similar reflection/scattering of pseudo sunlight into the CsPbBr₃ layer by metal-luster/white substrates placed below the rear semitransparent SWNT electrode. Contrary to our expectations, the PCE was never enhanced. To explore bifacial-irradiation functions, white light was irradiated using a commodity LED source through a 60%T-transmittance SWNT thin film. According to the previous reports,^11–13 to resemble reflecting light under realistic environments using the LED light, we controlled the light power between 30 and 40% of pseudo sunlight (100 mW cm⁻²). The distance between the LED source and the s-PSC at 60%T was adjusted by using the hand-made jig (Fig. 5a) to generate J_sc = 1.00 mA cm⁻² (Table 2), compared to the pseudo sunlight generating 3.79 mA cm⁻² (Table 1). Under the same condition, the light power ranged between 32 and 35% of the pseudo-sunlight power (100 mW cm⁻²) by calibrating using the Si photodiode. The monofacial-irradiation J-V curve profile was synchronized with an on/off mode of LED light through SWNT (Fig. 5b, black line). The pseudo sunlight and LED light were simultaneously irradiated through FTO (front) and SWNT (rear), respectively, where apertures for irradiation were in the same position. The bifacial-irradiation J–V curve profile (Fig. 5b, red line) was consistent with a superposition of the monofacial-irradiation ones through FTO (blue line) and SWNTs (black line). Interestingly, a profile nearly identical to that shown in Fig. 5b was obtained when bifacial irradiation was carried out from different aperture positions of the front and rear electrodes (Fig. S12, ESI†). This occurrence indicates that all of the photons from pseudo sunlight passing through FTO are captured by the CsPbBr₃ layer. However, there is potential to utilize further the LED light irradiated through the SWNT from the opposite side. This is why the PCE was never enhanced by the placement of white and metal-luster substrates in the first experiment. In 2023, Omer et al. simulated the energy flux in an organic–inorganic perovskite layer absorbing all visible light under bifacial irradiation. The energy flux was comparable to the sum of those under monofacial irradiation, with the highest concentration consistently found at the two interfacial regions. Nearly 80% of the incident light energy was absorbed within 100 nm of the perovskite-layer interfaces.¹² Omer's simulation supports our experimental results using the wavelength-selective-transparency all-inorganic CsPbBr₃ layer with a thickness of ∼300 nm (Fig. 2e and S11, ESI†).

Fig. 5 (a) Photograph and schematics of a hand-made jig for bifacial irradiation using commercial LED light sources. The front-side irradiation is replaced with pseudo sunlight. (b) J–V curves of an s-PSC at 60%T under monofacial irradiation using an on/off mode of LED light through SWNT (black line) and using pseudo sunlight (AM 1.5G) through FTO (blue line). A bifacial-irradiation J–V curve (red line) using the on/off LED light through SWNT and the pseudo sunlight through FTO. The apertures for irradiation through the front (FTO) and rear (SWNT) electrodes are in the same position. (c–e) J–V curves of s-PSCs at different transmittance values of (c) 60, (d) 70, and (e) 80%T under monofacial irradiation through SWNT (red lines) and FTO (black lines) using LED light, and bifacial irradiation (blue lines) using LED light. The J–V curves are reverse scans. The irradiation area is 4.90 × 10⁻² cm². To control the LED-light power between 32 and 35% of the pseudo-sunlight power (100 mW cm⁻²), the distance between the LED source and the s-PSC is adjusted to generate J_sc (SWNT) = 1.00 mA cm⁻² in the case of 60%T.

Table 2 Photovoltaic parameter values of s-PSCs in bifacial and monofacial irradiation through FTO and SWNT using LED light

	J sc/mA cm^−2	V oc/V	FF	PCE (%)
a Monofacial irradiation. b To control the LED-light power between 32 and 35% of the pseudo-sunlight power (100 mW cm^−2), the distance between the LED source and the s-PSC is adjusted to generate Jsc (SWNT) = 1.00 mA cm^−2 in the case of 60%T.
Through FTO
60%T	2.53	1.37	0.78	2.70
70%T	2.44	1.32	0.82	2.65
80%T	2.34	1.33	0.77	2.42
Through SWNT
60%T	1.00^b	1.32	0.87	1.15
70%T	1.12	1.32	0.86	1.27
80%T	1.26	1.31	0.80	1.32
Bifacial irradiation
60%T	3.46	1.37	0.77	3.67
70%T	3.58	1.33	0.81	3.86
80%T	3.57	1.35	0.77	3.71

---

We focus on how a bifacial-irradiation effect is influenced by the different transmittance values of SWNT thin films. Based on the J–V curves through the LED light having lower power than the pseudo sunlight (Fig. 5c–e), the photovoltaic parameter values are summarized in Table 2. In the monofacial irradiation through SWNT, an increasing trend was seen in J_sc (SWNT) as 1.00 (60) < 1.12 (70) < 1.26 mA cm⁻² (80%T), while J_sc (FTO) through FTO showed a decreasing trend as 2.53 (60) > 2.44 (70) > 2.34 mA cm⁻² (80%T). These monofacial trends were essentially consistent with the cases using pseudo sunlight. Comparing the monofacial-irradiation J_sc (SWNT) and J_sc (FTO), bifacial-irradiation J_sc (bifacial) values were increased to 3.46 (60), 3.58 (70), and 3.57 mA cm⁻² (80%T), and were equivalent to the sums of J_sc(SWNT) + J_sc(FTO) within ±2% deviation: J_sc(SWNT) + J_sc(FTO) = 1.00 + 2.53 = 3.53 (60), 1.12 + 2.44 = 3.56 (70), and 1.26 + 2.34 = 3.60 mA cm⁻² (80%T). PCEs (bifacial) showed similar values of 3.67 (60), 3.86 (70), and 3.71% (80%T), and were close to the sums of the monofacial PCEs (SWNT) + PCEs (FTO) of 1.15 + 2.70 = 3.85 (60), 1.27 + 2.65 = 3.92 (70), and 1.32 + 2.42 = 3.74% (80%T). In the emission spectrum (Fig. S13, ESI†), the LED can compensate for the visible light of the pseudo sunlight but it excludes the UV light at <400 nm. Based on Tables 1 and 2, the monofacial J_sc (FTO) ratios using LED/pseudo sunlight showed close values of 0.338 (60), 0.333 (70), and 0.339 (80%T). Considering that the UV light included in pseudo sunlight is absorbed through the TiO₂ layers, the LED source emits visible light with 33–34% power compared to pseudo sunlight which can be estimated as the monofacial J_sc (FTO) ratios × 100 (%). The values are near calibrated ones of 32–35% by the Si photodiode.

Consequently, the specific photovoltaic feature of the see-through structure of FTO/c-TiO₂/mp-TiO₂/CsPbBr₃/SWNT is discovered; the increased transmittance of SWNT thin films can improve the monofacial-irradiation PCEs (SWNT) through SWNT but degrades PCEs (FTO) through FTO due to their increased sheet resistance and R_s. Considering that PCEs (bifacial) are practically equivalent to the sums of the monofacial PCEs (SWNT) and PCEs (FTO), bifacial irradiation can balance the merit and demerit in the monofacial-irradiation performance by solving a tradeoff between the transparency and electrical conductivities of the SWNT thin films. Most importantly, visible-light transparency of the s-PSCs can be improved from 60 to 80%T by reducing SWNT densities, and simultaneously, the PCEs (bifacial) can be maintained at almost the same values (Table 2). Therefore, we can positively apply the highest-transparency SWNT thin film at 80%T to construct various-E_g s-PSCs. The black color of the SWNT thin film is not a serious obstacle visually based on a photograph of the s-PSC at 80%T taken through the environmental light (Fig. 3c and d).^14–17 Note that the black color is visually emphasized by the photographs taken through reflecting light from a white substrate (Fig. 3a and b).

Future issues for practical applications

The solution-processed semitransparent SWNT thin films have pioneered a new status as a corrosion-free carbon electrode for all-inorganic s-PSCs: (a) enhancement ratios of the bifacial-irradiation PCEs to the monofacial-irradiation PCEs were 3.19/1.36 (60), 3.04/1.46 (70), and 2.81/1.53 (80%T) through SWNT/through FTO (Table 2); (b) UV light could be utilized through SWNT but mostly not through FTO (Fig. 4i). Advantage (a) may be beneficial for future applications for utilizing low-power reflected irradiance (albedo) under realistic environments. Nevertheless, we will continue to investigate the enhancement effect depending on layer thicknesses of perovskites with various E_g and irradiation light powers. Many reports have demonstrated that the long-term weatherability of CsPbBr₃ is significantly improved by covering it with hydrophobic thick-layer carbon-paste electrodes.^{33–35,37–40} However, for construction of the s-PSCs, the naked CsPbBr₃ surfaces are directly exposed to ambient air through the semitransparent SWNT thin films, as seen in the top-view FE-SEM images (Fig. 4d–f). In a preliminary experiment, the s-PSC at 60%T maintained the initial photovoltaic performance for at least two weeks under exposure to ambient air (Fig. S14, ESI†), but the performance dropped in a rainy season of harsh conditions with the relative humidity increasing to 90%. Regardless of the inherently superior weatherability, sealing using H₂O-barrier films is necessary for practical applications to prevent exposure of the CsPbBr₃ surfaces to high-humidity air. Moreover, for advantage (b), the sealing materials passing the UV light must be selected.

The degradation in PCEs by expanding the irradiation area is a common issue. By the expansion from 4.90 × 10⁻² to 1.56 × 10⁻¹ cm², the photovoltaic parameter values of the s-PSC at 60%T were decreased from J_sc = 7.49 mA cm⁻², V_oc = 1.48 V, FF = 0.79, and PCE = 8.68% to J_sc = 7.22 mA cm⁻², V_oc = 1.45 V, FF = 0.77, and PCE = 8.13% under pseudo-sunlight monofacial irradiation through FTO (Table S1, ESI†), as the J–V curve shown in Fig. S15, ESI.† The decreased J_sc value was responsible for the degraded PCE because the expanded area avoidably accumulated the electrical resistance of the SWNT thin film. In 2022, we reported that J_sc values were similarly decreased with increased irradiation areas from 3.1 × 10⁻² to 1.13 cm² in a p–n single heterojunction silicon solar cell using the same solution-processed SWNT thin film. However, the addition of stacked Ag nanowires as an auxiliary electrode on the thin film drastically suppressed the J_sc decreases.⁴⁴ As reported in 2021 by Zhang et al., a high-quality CsPbBr₃ layer prepared by a water-based spray-assisted growth method achieved a high monofacial-irradiation PCE of 8.21% in a large irradiation area of 1 cm² of FTO/TiO₂/CsPbBr₃/carbon paste (Table S1 and ref. 9, ESI†).³⁹

Conclusions

To construct s-PSCs and to explore their bifacial-irradiation performances, semitransparent SWNT thin films are transferred onto the wavelength-selective-transparent CsPbBr₃ layers. This transfer is achieved using a solution-processed press-free filter-transfer method, which includes the replacement of water with CsPbBr₃-compatible solvents. Thus, all-inorganic s-PSCs with the simplest HTL-free structure of FTO/TiO₂/CsPbBr₃/SWNT are successfully fabricated by all-solution processes. To elucidate the photovoltaic performance depending on the SWNT densities, its transmittance values at 550 nm are adjusted to approximately 60, 70, and 80%T. In monofacial irradiation through FTO using pseudo sunlight (100 mW cm⁻²), J_sc and PCE values are synchronously increased in the order of J_sc = 7.49 (60) > 7.33 (70) > 6.91 mA cm⁻² (80%T) and of PCE = 8.68 (60) > 8.18 (70) > 7.30% (80%T) with decreasing sheet resistance and R_s values of the SWNT thin films. In monofacial irradiation through SWNT, increased transparency of SWNT improves them in the reverse order of J_sc = 3.79 (60) < 4.13 (70) < 4.56 mA cm⁻² (80%T) and of PCE = 4.21% (60%T) < 4.45% (70%T) < 4.56% (80%T) by increased numbers of photons passing through SWNT. Diode ideal factors range between 1 and 2, suggesting that the s-PSCs are regarded as a single-diode model with high heterojunction qualities TiO₂/CsPbBr₃ and CsPbBr₃/SWNT to suppress charge recombination, thereby achieving superior FF values of 0.77–0.84. To resemble reflecting light under realistic environments, LED light power is controlled between 32 and 35% of pseudo sunlight (100 mW cm⁻²). Bifacial-irradiation functions are revealed by using the LED light, showing J_sc = 3.46 (60), 3.58 (70), and 3.57 mA cm⁻² (80%T) and PCEs = 3.67 (60), 3.86 (70), and 3.71% (80%T). These values are practically equivalent to the sums of the monofacial-irradiation ones through FTO and SWNT. Enhancement ratios of the bifacial-irradiation PCEs to the monofacial-irradiation PCEs were 3.19/1.36 (60), 3.04/1.46 (70), and 2.81/1.53 (80%T) through SWNT/through FTO. A tradeoff relationship between the transparency and electrical conductivity of SWNT is an essential factor in determining the monofacial-irradiation performance, while it is successfully solved by bifacial irradiation. As a result, the bifacial irradiation maintains a high PCE value even in the highest-transparency SWNT thin film at 80%T, and its black color will not be a serious obstacle visually. It is expected that low-power reflecting light under realistic environments will be utilizable by the bifacial function in the development of window-integrated s-PSCs using various-E_g perovskites.

Experimental procedures

Materials

For etching of FTO layers on a glass substrate with a thickness of 1.1 mm (AGC, 8.8 Ω cm⁻²), guaranteed-grade (GR) zinc powder, hydrochloric acid (35–37%), 2-propanol, and acetone were obtained from Kanto Chemical Co., Inc. To prepare c- and mp-TiO₂ layers, titanium diisopropoxide bis acetylacetonate (Sigma-Aldrich, 75 wt% in 2-propanol), ethanol (Kanto Chemical Co., Inc., GR), and TiO₂ paste (Mikuni-Color Ltd., SCI-80) were used. CsPbBr₃ layers were prepared using PbBr₂ (Tokyo Chemical Industry, >98.0%), CsBr (Kanto Chemical Co., Inc., ≥99.99%), N,N-dimethylformamide (FUJIFILM Wako Pure Chemical Corporation, super dehydrated, ≥99.5%), methanol (Kanto Chemical Co., Inc., GR) dehydrated by molecular sieves 3A (Kanto Chemical Co., Inc.). SWNTs (Meijo Nano Carbon, eDIPs EC-1.5) were dispersed in water using sodium dodecylbenzenesulfonate (Tokyo Chemical Industry, ≥95.0%). For solution-processed filter transfer of SWNT thin films, water is replaced with toluene (Kanto Chemical Co., Inc., GR) dehydrated by molecular sieves 4A.

Patterning of an FTO glass substrate

An FTO glass substrate was cleaned by ultrasonication in a detergent liquid (AS ONE, SDUN-N4) and washed with water, 2-propanol, and acetone. For patterning, the as-cleaned FTO layer was partially removed from a glass substrate by a reaction with zinc powder and hydrochloric acid (12%) through masking with vinyl chloride tape.

Preparation of FTO/c-TiO₂/mp-TiO₂/CsPbBr₃

A commercial 2-propanol solution (2.0 mL) of titanium diisopropoxide bis acetylacetonate (75 wt%) was diluted with ethanol (38 mL). To form a c-TiO₂ layer, the patterned FTO substrate was spray-coated at 450 °C using the precursor solution filtered through a polytetrafluoroethylene (PTFE) membrane filter (pore size: 0.22 μm) and further heated at 450 °C for 30 min. A commercial TiO₂ paste (0.60 g) was mixed with ethanol (3.4 g) and stirred for 2 h. The c-TiO₂ layer was treated with ozone for 30 min (Nippon Laser & Electronics Lab, UV-O₃ cleaner) and spin-coated with the precursor suspension (100 μL) through a PTFE membrane filter (pore size: 1.0 μm) at 3500 rpm (Kyowariken, K-35951). The spin-coated substrate was heated at 120 °C for 5 min on a hotplate. The substrate was further heated for 60 min after the temperature was increased to 550 °C for 30 min in an electric furnace (KDF S-70) to form a mp-TiO₂ layer. In a typical case, PbBr₂ (514 mg, 1.40 mmol) was dissolved in N,N-dimethylformamide (1.00 mL) by stirring at 90 °C and the solution (1.40 mol L⁻¹) was filtered through a PTFE-membrane filter (pore size: 0.22 μm). CsBr (149 mg, 7.00 × 10⁻¹ mmol) was dissolved in methanol (10.0 mL) by ultrasonication and the solution (7.00 × 10⁻² mol L⁻¹) was filtered through a PTFE membrane filter (pore size: 0.22 μm). In a typical case, the FTO/c-TiO₂/mp-TiO₂ substrate was treated with ozone for 30 min, spin-coated with PbBr₂ solution (50 μL) at 2000 rpm, and heated at 90 °C for 30 min on a hotplate to form a PbBr₂ layer. Note that a precipitate appeared in the high-concentration solution (1.40 mol L⁻¹) of PbBr₂ and was dissolved by heating before spin coating. The FTO/c-TiO₂/mp-TiO₂/PbBr₂ substrate was spin-coated with CsBr solution (90 μL) at 2000 rpm and heated at 250 °C for 5 min on a hotplate. The procedures were repeated nine times to almost complete the transformation from CsPb₂Br₅ to CsPbBr₃. The structural change was monitored by X-ray diffraction (XRD) measurements. The concentrations of PbBr₂ were changed between 1.20 and 1.40 mol L⁻¹ to compare morphologies of CsPbBr₃ layers and their photovoltaic performances. The preparation of FTO/c-TiO₂/mp-TiO₂/CsPbBr₃ was conducted under ambient air.

Press-free filter transfer (spontaneous transfer) of a semitransparent SWNT thin film onto FTO/c-TiO₂/mp-TiO₂/CsPbBr₃

SWNTs were stably dispersed in water using 0.50-wt% sodium dodecylbenzenesulfonate following the same procedures as in our previous report.⁴¹ A SWNT aqueous dispersion solution was prepared with a concentration of 2.5 × 10⁻³ wt% as a stock solution. The stock solution was diluted by the addition of water to adjust suitable SWNT densities of semitransparent thin films to transmittance values of 60, 70, and 80%T at 550 nm.⁴¹ As shown in Fig. 1d, the aqueous dispersion solution of SWNTs was filtered through an ethanol-filled hydrophobic PTFE membrane filter (Sumitomo Electric POREFLON HP-010-30, pore size: 100 nm). The SWNT thin film adhering to the PTFE membrane was washed to remove the surfactant of sodium dodecylbenzenesulfonate by passing water (5 mL) three times and ethanol (5 mL) three times under reduced pressure. The SWNT thin film could be cut with scissors to obtain suitable areas and shapes. The suitably shaped SWNT thin film, along with the PTFE membrane, was immersed in a dehydrated ethanol bath for over 10 min and further immersed in a dehydrated toluene bath for over 10 min. In this way, water could be replaced with toluene to prevent the decomposition of CsPbBr₃ layers in the following transfer processes: the toluene-wetted SWNT thin film on the PTFE membrane filter was placed on the CsPbBr₃ layer without any pressure and heated at 70 °C to evaporate toluene on a hotplate; by peeling the PTFE membrane filter, a semitransparent SWNT thin film was spontaneously transferred onto the CsPbBr₃ layer; by dropping a small portion of toluene on SWNTs and evaporation of toluene at 70 °C, their insufficient contact with the CsPbBr₃ surface was improved to prepare an HTL-free solar cell of FTO/c-TiO₂/mp-TiO₂/CsPbBr₃/SWNT. To improve the electrical contact of SWNT with probes of a hand-made jig (Fig. 5a), vacuum-evaporated gold was used as an auxiliary electrode (Fig. 3b).

Photovoltaic measurements

J–V curves were recorded on a source meter (Keithley 2400) using a solar simulator (Peccell PEC-L01) under irradiation of pseudo sunlight (AM 1.5G, 100 mW cm⁻²), where the irradiation intensity was calibrated by a Si photodiode (BS-520). External quantum efficiency (EQE) spectra were recorded on an action spectrometer (Peccell PEC-S20 with a Xe lamp (150 W)) by a Si photodiode reference (S1337–1010BQ). For the investigation of bifacial-irradiation performance, commercial LEDs emitting white lights (6 W) were used. The LED-light power was calibrated to range between 32% and 35% of the pseudo-sunlight power, using a Si photodiode (BS-520).

Characterization

A four-point probe technique (Kyowariken K-705RS) was used to measure the sheet resistances. The absorption/transmittance spectra of SWNT thin films and FTO/c-TiO₂/mp-TiO₂/CsPbBr₃ were recorded on UV-Vis-near-IR spectrophotometers (Shimadzu UV-2600 and 3600). The top views and cross-sectional images of the SWNT thin films, FTO/c-TiO₂/mp-TiO₂/CsPbBr₃, and FTO/c-TiO₂/mp-TiO₂/CsPbBr₃/SWNT were observed on a JEOL JSM-7600F field emission scanning electron microscope (FE-SEM). The surface morphologies of the CsPbBr₃ layers were investigated via atomic force microscope (AFM) images (Park Systems XE70). The structural change of the CsPbBr₃ layers was investigated via X-ray diffraction (XRD) patterns (Rigaku MiniFlex II, Cu Kα1 radiation).

Data availability

The data that support the findings of this study are available upon request from the corresponding authors.

Author contributions

M. K. and M. I. designed all of the experiments, analyzed the data, and wrote the manuscript. H. D. prepared the HTL-free CsPbBr₃ solar cells and investigated their monofacial and bifacial performances. H. T., I. W., and R. A. assisted in the experiments.

Conflicts of interest

The authors declare no competing financial interests.

Acknowledgements

The authors thank Dr Masahide Kawaraya, the invited senior researcher at the National Institute of Advanced Industrial Science and Technology (AIST), for his advice on the preparation and evaluation of perovskite solar cells. This work was partially supported by Takahashi Industrial and Economic Research Foundation.

References

1. S. Yang, Y. Duan, Z. Liu and S. Frank) Liu, Recent advances in CsPbX₃ perovskite solar cells: Focus on crystallization characteristics and controlling strategies, Adv. Energy Mater., 2023, 13, 2201733.
2. S. Ullah, J. Wang, P. Yang, L. Liu, S.-E. Yang, T. Xia, H. Guo and Y. Chen, All-inorganic CsPbBr₃ perovskite: a promising choice for photovoltaics, Mater. Adv., 2021, 2, 646–683.
3. S. Lim, S. Han, D. Kim, J. Min, J. Choi and T. Park, Key factors affecting the stability of CsPbI₃ perovskite quantum dot solar cells: A comprehensive review, Adv. Mater., 2023, 35, 2203430.
4. D. B. Ritzer, B. A. Nejand, M. A. R. Preciado, S. Gharibzadeh, H. Hu, A. Diercks, T. Feeney, B. S. Richards, T. Abzieher and U. W. Paetzold, Translucent perovskite photovoltaics for building Integration, Energy Environ. Sci., 2023, 16, 2212–2225.
5. J. Zhuang, J. Wang and F. Yan, Review on chemical stability of lead halide perovskite solar cells, Nano-Micro Lett., 2023, 15, 84.
6. A. Elakshar, S. Tsarev, P. M. Rajanna, M. Tepliakova, L. Frolova, Y. G. Gladush, S. M. Aldoshin, P. A. Troshin and A. G. Nasibulin, Surface passivation for efficient bifacial HTL-free perovskite solar cells with SWCNT top electrodes, ACS Appl. Energy Mater., 2021, 4, 13395–13400.
7. A. S. R. Bati, Y. L. Zhong, P. L. Burn, M. K. Nazeeruddin, P. E. Shaw and M. Batmunkh, Next-generation applications for integrated perovskite solar cells, Commun. Mater., 2023, 4, 00325.
8. P. Patil, S. S. Sangale, S.-N. Kwon and S.-I. Na, Innovative approaches to semi-transparent perovskite solar cells, Nanomaterials, 2023, 13, 1084.
9. Z. Song, C. Li, L. Chen and Y. Yan, Perovskite solar cells go bifacial—mutual benefits for efficiency and durability, Adv. Mater., 2022, 34, 2106805.
10. S. Rahmany and L. Etgar, Semitransparent perovskite solar cells, ACS Energy Lett., 2020, 5, 1519–1531.
11. C. Zhang, M. Chen, F. Fu, H. Zhu, T. Feurer, W. Tian, C. Zhu, K. Zhou, S. Jin, S. M. Zakeeruddin, A. N. Tiwari, N. P. Padture, M. Grätzel and Y. Shi, CNT-based bifacial perovskite solar cells toward highly efficient 4-terminal tandem photovoltaics, Energy Environ. Sci., 2022, 15, 1536–1544.
12. M. I. Omer, T. Ye, X. Li, S. Ma, D. Wu, L. Wei, X. Tang, S. Ramakrishna, Q. Zhu, S. Xiong, J. Xu, C. Vijila and X. Wang, Two quasi-interfacial p–n junctions observed by a dual-irradiation system in perovskite solar cells, npj Flexible Electron., 2023, 7, 00256.
13. J. Zhang, X.-G. Hu, K. Ji, S. Zhao, D. Liu, B. Li, P.-X. Hou, C. Liu, L. Liu, S. D. Stranks, H.-M. Cheng, S. R. P. Silva and W. Zhang, High-performance bifacial perovskite solar cells enabled by single-walled carbon nanotubes, Nat. Commun., 2024, 15, 2245.
14. X. Jiang, C. Geng, X. Yu, J. Pan, H. Zheng, C. Liang, B. Li, F. Long, L. Han, Y.-B. Cheng and Y. Peng, Doping with KBr to achieve high-performance CsPbBr₃ semitransparent perovskite solar cells, ACS Appl. Mater. Interfaces, 2024, 16, 19039–19047.
15. T. Das, R. Nag, N. K. Rana, M. Nayak, R. Paramanik, A. Bera, S. K. Saha and A. Guchhait, Effect of transition metal doping in the ZnO nanorod on the efficiency of the electron transport layer in semitransparent CsPbBr₃ perovskite solar cells, Energy Fuels, 2023, 37, 10642–10651.
16. B. Parida, S. Yoon and D.-W. Kang, Room-temperature solution-processed 0D/1D bilayer electrodes for translucent CsPbBr₃ perovskite photovoltaics, Nanomaterials, 2021, 11, 1489.
17. W. Chen, J. Zhang, G. Xu, R. Xue, Y. Li, Y. Zhou, J. Hou and Y. Li, A semitransparent inorganic perovskite film for overcoming ultraviolet light instability of organic solar cells and achieving 14.03% efficiency, Adv. Mater., 2018, 30, 1800855.
18. J. Barichello, D. D. Girolamo, E. Nonni, B. Paci, A. Generosi, M. Kim, A. Levtchenko, S. Cacovich, A. D. Carlo and F. Matteocci, Semi-transparent blade-coated FAPbBr₃ perovskite solar cells: A scalable low-temperature manufacturing process under ambient condition, Sol. RRL, 2023, 7, 2200739.
19. W. Yue, H. Yang, H. Cai, Y. Xiong, T. Zhou, Y. Liu, J. Zhao, F. Huang, Y.-B. Cheng and J. Zhong, Printable high-efficiency and stable FAPbBr₃ perovskite solar cells for multifunctional building-integrated photovoltaics, Adv. Mater., 2023, 35, 2301548.
20. T. Liu, X. Zhao, P. Wang, Q. C. Burlingame, J. Hu, K. Roh, Z. Xu, B. P. Rand, M. Chen and Y.-L. Loo, Highly transparent, scalable, and stable perovskite solar cells with minimal aesthetic compromise, Adv. Energy Mater., 2023, 13, 2200402.
21. G. Perrakis, A. C. Tasolamprou, G. Kakavelakis, K. Petridis, M. Graetzel, G. Kenanakis, S. Tzortzakis and M. Kafesaki, Infrared-reflective ultrathin-metal-film-based transparent electrode with ultralow optical loss for high efficiency in solar cells, Sci. Rep., 2024, 14, 548.
22. J. Li, Q. Dong, N. Li and L. Wang, Direct evidence of ion diffusion for the silver-electrode-induced thermal degradation of inverted perovskite solar cells, Adv. Energy Mater., 2017, 7, 1602922.
23. X. Li, S. Fu, W. Zhang, S. Ke, W. Song and J. Fang, Chemical anti-corrosion strategy for stable inverted perovskite solar cells, Sci. Adv., 2020, 6, eabd1580.
24. N. N. Udalova, E. M. Nemygina, E. A. Zharenova, A. S. Tutantsev, A. A. Sudakov, A. Y. Grishko, N. A. Belich, E. A. Goodilin and A. B. Tarasov, New aspects of copper electrode metamorphosis in perovskite solar cells, J. Phys. Chem. C, 2020, 124, 24601–24607.
25. R. Chen, W. Zhang, X. Guan, H. Raza, S. Zhang, Y. Zhang, P. A. Troshin, S. A. Kuklin, Z. Liu and W. Chen, Rear electrode materials for perovskite solar cells, Adv. Funct. Mater., 2022, 32, 2200651.
26. E. S. Akulenko, M. Hadadian, A. S. -Aarnio and K. Miettunen, Eco-design for perovskite solar cells to address future waste challenges and recover valuable materials, Heliyon, 2023, 9, e13584.
27. R. G. Charles, A. Doolin, R. G. -Rodríguez, K. V. Villalobos and M. L. Davies, Circular economy for perovskite solar cells—drivers, progress and challenges, Energy Environ. Sci., 2023, 16, 3711–3733.
28. X. Zhang, Z. Yu, D. Zhang, Q. Tai and X.-Z. Zhao, Recent progress of carbon-based inorganic perovskite solar cells: From efficiency to stability, Adv. Energy Mater., 2023, 13, 2201320.
29. C. Dong, B. Xu, D. Liu, E. G. Moloney, F. Tan, G. Yue, R. Liu, D. Zhang, W. Zhang and M. I. Saidaminov, Carbon-based all-inorganic perovskite solar cells: Progress, challenges and strategies toward 20% efficiency, Mater. Today, 2021, 50, 239–258.
30. T. Zhang, C. Liu, Z. Li, B. Zhao, Y. Bai, X. Li, W. Liu, Y. Chen, Z. Liu and X. Li, Improved perovskite/carbon Interface through hot-pressing: A case study for CsPbBr₃-based perovskite solar cells, ACS Omega, 2022, 7, 16877–16883.
31. Y. Yuan, G. Yan, R. Hong, Z. Liang and T. Kirchartz, Quantifying Efficiency Limitations in All-Inorganic Halide Perovskite Solar Cells, Adv. Mater., 2022, 34, 2108132.
32. J. Duan, T. Hu, Y. Zhao, B. He and Q. Tang, Carbon-electrode-tailored all-Inorganic perovskite solar cells to harvest solar and water-vapor energy, Angew. Chem., Int. Ed., 2018, 57, 5746–5749.
33. J. Duan, Y. Zhao, B. He and Q. Tang, High-purity inorganic perovskite films for solar cells with 9.72% efficiency, Angew. Chem., Int. Ed., 2018, 57, 3787–3791.
34. R. Guo, Y. Zhao, Y. Zhang, Q. Deng, Y. Shen, W. Zhang and G. Shao, Significant performance enhancement of all-inorganic CsPbBr₃ perovskite solar cells enabled by Nb-doped SnO₂ as effective electron transport layer, Energy Environ. Mater., 2021, 4, 671–680.
35. W. Zhu, X. Yao, S. Shi, Z. Zhang, Z. Song, P. Gao, T. Wang, Y. Ba and P. Dong, Mixed-bromide halide exchange toward efficient and stable carbon-electrode CsPbBr₃ solar cells, ACS Appl. Energy Mater., 2023, 6, 9798–9804.
36. Y. Zhang, F. Sheng, H. Zhang, X. Ding, L. Zhi, Y. Li, X. Cao, X. Cui and J. Wei, Preparation of CsPbBr₃ films with tens of micrometer-scale grains and preferential orientation for perovskite solar cells, ACS Appl. Energy Mater., 2023, 6, 12190–12197.
37. J. Duan, Y. Wang, X. Yang and Q. Tang, Alkyl-chain-regulated charge transfer in fluorescent inorganic CsPbBr₃ perovskite solar cells, Angew. Chem., Int. Ed., 2020, 59, 4391–4395.
38. Q. Zhou, J. Duan, J. Du, Q. Guo, Q. Zhang, X. Yang, Y. Duan and Q. Tang, Tailored lattice “tape” to confine tensile interface for 11.08%-efficiency all-inorganic CsPbBr₃ perovskite solar cell with an ultrahigh voltage of 1.702 V, Adv. Sci., 2021, 8, 2101418.
39. Z. Z. Y. Ba, D. Chen, J. Ma, W. Zhu, H. Xi, D. Chen, J. Zhang, C. Zhang and Y. Hao, Generic water-based spray-assisted growth for scalable high-efficiency carbon-electrode all-inorganic perovskite solar cells, iScience, 2021, 24, 103365.
40. Y. Liu, T. Xiang, B. Zhang, J. Wang, X. Yu, Y. Xiao, J. Xiao, Z. Ku and Y. Peng, Defect passivation and fermi level modification for >10% evaporated all-Inorganic CsPbBr₃ perovskite solar cells, ACS Appl. Energy Mater., 2022, 5, 8049–8056.
41. M. Ishizaki, D. Satoh, R. Ando, M. Funabe, J. Matsui and M. Kurihara, Solution-processed chemically non-destructive filter transfer of carbon-nanotube thin films onto arbitrary materials, Adv. Mater. Interfaces, 2021, 8, 2100953.
42. E. Muramoto, Y. Yamasaki, F. Wang, K. Hasegawa, K. Matsuda and S. Noda, Carbon nanotube–silicon heterojunction solar cells with surface-textured Si and solution-processed carbon nanotube films, RSC Adv., 2016, 6, 93575–93581.
43. H. Shirae, D. Y. Kim, K. Hasegawa, T. Takenobu, Y. Ohno and S. Noda, Overcoming the quality–quantity tradeoff in dispersion and printing of carbon nanotubes by a repetitive dispersion–extraction process, Carbon, 2015, 91, 20–29.
44. M. Funabe, D. Satoh, R. Ando, H. Daiguji, J. Matsui, M. Ishizaki and M. Kurihara, A solvent-compatible filter-transfer method of semi-transparent carbon-nanotube electrodes stacked with silver nanowires, Sci. Technol. Adv. Mater., 2022, 23, 783–795.
45. C. Liu, T. Zhang, Z. Li, B. Zhao, X. Ma, Y. Chen, Z. Liu, H. Chen and X. Li, Crystallization kinetics engineering toward high-performance and stable CsPbBr₃-based perovskite solar cells, ACS Appl. Energy Mater., 2021, 4, 10610–10617.
46. J. Feng, X. Han, H. Huang, Q. Meng, Z. Zhu, T. Yu, Z. Li and Z. Zou, Curing the fundamental issue of impurity phases in two-step solution-processed CsPbBr₃ perovskite films, Sci. Bull., 2020, 65, 726–737.
47. X. Wan, Z. Yu, W. Tian, F. Huang, S. Jin, X. Yang, Y.-B. Cheng, A. Hagfeldt and L. Sun, Efficient and stable planar all-inorganic perovskite solar cells based on high-quality CsPbBr₃ films with controllable morphology, J. Energy Chem., 2020, 46, 8–15.
48. X. Cao, G. Zhang, L. Jiang, Y. Cai, Y. Gao, W. Yang, X. He, Q. Zeng, G. Xing, Y. Jia and J. Wei, Water, a green solvent for fabrication of high-quality CsPbBr₃ films for efficient solar cells, ACS Appl. Mater. Interfaces, 2020, 12, 5925–5931.
49. H. Li, G. Tong, T. Chen, H. Zhu, G. Li, Y. Chang, L. Wang and Y. Jiang, Interface engineering using a perovskite derivative phase for efficient and stable CsPbBr₃ solar cells, J. Mater. Chem. A, 2018, 6, 14255–14261.
50. A. Kaskela, A. G. Nasibulin, M. Y. Timmermans, B. Aitchison, A. Papadimitratos, Y. Tian, Z. Zhu, H. Jiang, D. P. Brown, A. Zakhidov and E. I. Kauppinen, Aerosol-synthesized SWCNT networks with tunable conductivity and transparency by a dry transfer technique, Nano Lett., 2010, 10, 4349–4355.
51. S. Seo, K. Akino, J.-S. Nam, A. Shawky, H.-S. Lin, H. Nagaya, E. I. Kauppinen, R. Xiang, Y. Matsuo, I. Jeon and S. Maruyama, Multi-functional MoO₃ doping of carbon-nanotube top electrodes for highly transparent and efficient semi-transparent perovskite solar cells, Adv. Mater.Interfaces, 2022, 9, 2101595.
52. I. Jeon, A. Shawky, S. Seo, Y. Qian, A. Anisimov, E. I. Kauppinen, Y. Matsuo and S. Maruyama, Carbon nanotubes to outperform metal electrodes in perovskite solar cells via dopant engineering and hole-selectivity enhancement, J. Mater. Chem. A, 2020, 8, 11141–11147.
53. I. Jeon, S. Seo, Y. Sato, C. Delacou, A. Anisimov, K. Suenaga, E. I. Kauppinen, S. Maruyama and Y. Matsuo, Perovskite solar cells using carbon nanotubes both as cathode and as anode, J. Phys. Chem. C, 2017, 121, 25743–25749.
54. Z. Li, S. A. Kulkarni, P. P. Boix, E. Shi, A. Cao, K. Fu, S. K. Batabyal, J. Zhang, Q. Xiong, L. H. Wong, N. Mathews and S. G. Mhaisalkar, Laminated carbon nanotube networks for metal electrode-free efficient perovskite solar cells, ACS Nano, 2014, 8, 6797–6804.
55. S. Yoon, I. H. Lee, J. Han, J. Bahadur, S. Lee, S. Lee, D. S. Kim, B. Mikladal, E. I. Kauppinen, D.-W. Kang and I. Jeon, Semi-transparent metal electrode-free all-inorganic perovskite solar cells using floating-catalyst-synthesized carbon nanotubes, EcoMat, 2024, 6, e12440.
56. Z. Wu, Z. Chen, X. Du, J. M. Logan, J. Sippel, M. Nikolou, K. Kamaras, J. R. Reynolds, D. B. Tanner, A. F. Hebard and A. G. Rinzler, Transparent, conductive carbon nanotube films, Science, 2004, 305, 1273–1276.
57. X. Hu, P. Hou, C. Liu and H. Cheng, Carbon nanotube/silicon heterojunctions for photovoltaic applications, Nano Mater. Sci., 2019, 1, 156–172.
58. D. D. Tune and B. S. Flavel, Advances in carbon nanotube–silicon heterojunction solar cells, Adv. Energy Mater., 2018, 8, 1703241.
59. D. D. Tune, N. Mallik, H. Fornasier and B. S. Flavel, Breakthrough carbon nanotube–silicon heterojunction solar cells, Adv. Energy Mater., 2020, 10, 1903261.
60. R. Xie, N. Ishijima, H. Sugime and S. Noda, Enhancing the photovoltaic performance of hybrid heterojunction solar cells by passivation of silicon surface via a simple 1-min annealing process, Sci. Rep., 2019, 9, 12051.
61. P. Wadhwa, B. Liu, M. A. McCarthy, Z. Wu and A. G. Rinzler, Electronic junction control in a nanotube-semiconductor Schottky junction solar cell, Nano Lett., 2010, 10, 5001–5005.
62. T. Furusawa, R. Ando, H. Daiguji, J. Jang, M. Funabe, J. Matsui, M. Ishizaki and M. Kurihara, Low-temperature edge-fusing phenomenon of silver microplates and solution-processed low-resistivity top-contact electrodes, ACS Appl. Electron. Mater., 2022, 4, 5538–5549.
63. P. Liao, X. Zhao, G. Li, Y. Shen and M. Wang, A new method for fitting current–voltage curves of planar heterojunction perovskite solar cells, Nano-Micro Lett., 2018, 10, 5.
64. N. Shibayama, H. Kanda, T. W. Kim, H. Segawa and S. Ito, Design of BCP buffer layer for inverted perovskite solar cells using ideal factor, APL Mater., 2019, 7, 031117.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4se00845f

---