Dynamic halide perovskite heterojunction generates direct current

Abstract

Here, we demonstrate a dynamic perovskite device capable of converting mechanical energy into direct current (DC) electrical energy, combining two concepts: carrier generation from the triboelectric effect and carrier separation through band energy level difference. By analyzing and comparing different perovskite (FAPbI₃, MAPbI₃, MAPbBr₃, PEA₂PbI₄, etc.) and charge transport layer (CTL) materials (spiro-MeOTAD, PTAA, TiO₂, SnO₂, etc.), the key rules for determining DC output and performances are identified: (1) a suitable band alignment (band position and bandgap) between perovskite and CTL can separate the carrier transfer; (2) a large difference in work function between two layers leads to high electrical potential difference; and (3) a high carrier concentration can enhance the DC power-generating performances. Furthermore, it is found that the light illumination acts as a stimulus to current output to a large extent, which is due to the coupling effect from triboelectric and photovoltaic effects. This study provides a set of key rules to explain the mechanism and to further improve the performance of the dynamic perovskite/CTL heterojunction.

---

Broader context

Harvesting environmental energy to generate electricity is a key scientific development of our time. Photovoltaic conversion and electromechanical transduction are two common energy-harvesting methods, which have drawn much attention. Organic–inorganic halide perovskite materials represent a recent key development in the photovoltaic field, which is attributed to their solution processability and excellent photovoltaic properties. Though static perovskite PN junctions have been widely studied and exploited in the photovoltaic field, the dynamic perovskite heterojunction has not been studied yet. More importantly, achieving the photovoltaic effect with mechanical energy harvesting in a single device is still challenging (i.e. alternative current (AC) for electromechanical transduction and direct current (DC) for photovoltaic conversion). Here, we demonstrate DC power from the dynamic perovskite/charge transport layer (CTL) heterojunctions via electrical carrier generation from triboelectrification between two layers, which is different from the electric energy generation by photon excitation in solar cells. More importantly, 1 sun illumination stimulates charge carrier concentration, which results in several hundred times higher current output than that in the dark due to the coupling triboelectric and photovoltaic effects. These findings open a new research area of triboelectric nanogenerator generating DC power.

Introduction

Harvesting energy from nature is a key scientific and technological concept.^1,2 Among different environmental energies, conversion of solar energy and mechanical energy to electric power is invaluable and thereby those technologies have drawn much attention.^3–5 Organic–inorganic halide perovskite materials have shown great potential in photovoltaic field since seminal works by Miyasaka et al. and Park et al.^6–8 In these devices, the perovskite layer acts as a light absorber and generates free carriers under light illumination, after which the free carriers are separated by the different charge transporting layers (CTL) due to the different bandgap alignments.^9–12

There also has been parallel progress in electromechanical properties of organic–inorganic halide perovskite (piezoelectric, triboelectric, and photoflexoelectric properties, etc.).^13–17 Using the triboelectric device as an example, contact electrification can provide the polarized charges on material surfaces and electrostatic induction can drive the charges to flow between two electrodes via changes in electric potential. Most of the triboelectric or piezoelectric devices can only produce alternative current (AC) outputs with the changed contact status. A rectifier is needed to convert AC to direct current (DC), which not only reduces the energy conversion efficiency but also increases the device size. It was also reported that DC output can be achieved by the dynamic Schottky and PN junctions (i.e. polymer/metal, inorganic/metal junctions) due to the directional carrier transfer between two materials.^18–25 Inspired by the electric generation based on triboelectric processes and the carrier directional transfer at static perovskite/CTL heterojunctions, we hypothesized that DC power generation could be achieved in response to applied mechanical energy through a dynamic perovskite/CTL heterojunction.

Static perovskite/CTL heterojunctions have been widely studied and exploited in solar cells.^26,27 Here, we demonstrate for the first time that DC power can be generated at dynamic perovskite/CTL heterojunctions by electrical carrier generation from triboelectrification between two layers, which is different from the electric energy generation by photon excitation in solar cells. Key factors for determining the DC output and performances (i.e. voltage and current) are studied: (1) heterojunctions formed by materials with different band energy levels; (2) work function difference between the two sliding materials and (3) carrier concentrations and mobilities. In these dynamic perovskite/CTL devices, carriers in the heterojunction are separated during the movement, as confirmed by photoluminescence (PL) and time-resolved PL (TRPL) measurements. As a result, free electrons and holes generated by triboelectrification from the sliding motion can be directionally transferred by the heterojunction, forming DC power output. Furthermore, more electrons and holes are generated by photovoltaic effects under 1 sun illumination. Combining the triboelectric and photovoltaic effects, the current of a dynamic device under 1 sun illumination is about several hundred times higher than that in the dark. This new device can be used as a multi-energy (mechanical and solar energy) input system for self-powered electronic devices, such as portable electronics, wearable electronics, self-powered sensors, e-skin, and sustainable devices.

Results and discussion

Perovskite used here is FAPbI₃ (FA = formamidinium) with MACl (MA = methylammonium) as an additive to improve the quality and stability.²⁸ Structure and morphology of the perovskite film were characterized by X-ray diffraction (XRD) and scanning electron microscopy (SEM). As shown in Fig. S1a and b (ESI†), the perovskite film shows a pure phase with uniform grain size and good coverage. Moreover, the absorption and PL spectrum in Fig. S1c (ESI†) show that the perovskite has a bandgap of ∼1.5 eV. The dark current–voltage (IV) curve was measured by pressing the 2,2′,7,7′-tetrakis(N,N-di-p-methoxyphenyl-amine)9,9′-spirobifluorene (spiro-MeOTAD, for short spiro) on the surface of the perovskite film to ensure the electric contact. As shown in Fig. S2 (ESI†), the static perovskite/spiro heterojunction shows a nonlinear curve with a rectifying effect, which is attributed to the formation of heterojunction.²⁹ The performance of the dynamic perovskite/CTL junctions below are all measured in the dark if without further notification.

The electric power output of the dynamic perovskite/spiro heterojunction was studied based on the sliding contact mode.³⁰ As shown in Fig. 1a, the current and voltage are measured by pressing the spiro-coated electrode to make contact with the perovskite-coated electrode and sliding the spiro film on the perovskite surface. The IV curve of the device under the movement is shown in Fig. 1b. The fluctuation of the IV curve under movement compared with the static condition is indicative of the voltage and current output.²⁰ As shown in Fig. 1c and d, a DC voltage of ∼0.4 V and a current of ∼1.2 μA are achieved, which is consistent with the results from the IV curve. This DC power is assumed to be related to the band alignment between perovskite and spiro, which can directionally transfer carriers.

Fig. 1 Schematic device structure and output analysis. (a) Schematic illustration of the dynamic perovskite/hole transport layer (HTL) heterojunction device. (b) IV curves obtained from the dynamic sliding contact between the FAPbI₃ perovskite and the spiro layers, along with static (no sliding movement) condition, measured in the dark (c) voltage and (d) current output of the dynamic perovskite/spiro heterojunction device under continuous sliding movements. The contact area was ∼1 cm² and the applied force was ∼5 N.

To study the mechanism of carrier generation and transfer at the dynamic perovskite/spiro heterojunction, the Mott–Schottky, steady-state PL, and TRPL measurements were conducted on the perovskite/spiro heterojunction. As shown in Fig. 2a, the Mott–Schottky curve shows a built-in potential of 0.38 V for the static perovskite/spiro junction, which confirms the formation of a stable PN junction. To study the carrier dynamic of the perovskite/spiro junction, the PL and TRPL spectra of the device (see details in Experimental section) were measured under static and bending conditions. Fig. 2b shows that the PL intensity of the device reduces from ∼27 000 counts under static to ∼22 000 counts under bending, which is indicative of enhancement in carrier separation and transfer between perovskite and spiro.³⁷ TRPL data in Fig. 2c are fit with a bi-exponential decay equation, where relatively fast decay component (τ₁) is assigned to carrier transfer or separation on the surface and much slower component (τ₂) is due to radiative recombination of carriers in the bulk.^38–40 Under the bending condition, the slow decay component (τ₂) is slightly decreased from 87 ns (static) to 81 ns (bending), while τ₁ is significantly reduced from 27 ns (static) to 16 ns (bending), which supports that the friction induced by the bending can facilitate carrier separation and enhance carrier transferring at the perovskite/spiro interface.

Fig. 2 Mechanism of the dynamic perovskite/CTL heterojunction. (a) Mott–Schottky curve of the static perovskite/spiro junction. (b) PL and (c) TRPL of the perovskite/spiro heterojunction under static and bending conditions (solid lines represent the fit results based on bi-exponential decay equation). (d and e) Schematics and energy band diagram of the perovskite/spiro heterojunction before and after contact (static heterojunction). (f) Free charge generation by triboelectric effects and the directional charge transfer by mechanical energy (triboelectric potential).

Based on the above analysis, we interpret the DC power generation in dynamic perovskite/spiro heterojunctions through the following possible processes: (1) Fig. 2d shows the energy band levels of perovskite and spiro materials before the two layers come into contact. Perovskite is supposed to be n-type and spiro is a p-type material. (2) When the perovskite contacts with spiro at a static condition, the electrons in the n-type perovskite are attracted to the positive holes in the p-type spiro and they diffuse into the p-type materials. Similarly, the positive holes in the spiro are attracted and diffuse to the perovskite. As a result, an electrical potential is formed between two materials, as shown in Fig. 2e. (3) When the top material is sliding, free charge carriers are generated by triboelectric effects. Then, the electrons and holes can be directionally transferred to perovskite and spiro, respectively, forming a DC power (Fig. 2f).

To study the voltage and current performances, perovskites with different band levels and electronic properties (i.e. FAPbI₃, MAPbI₃, and MAPbBr₃) were applied in the dynamic heterojunction with spiro. Structure and thickness of FAPbI₃, MAPbI₃, and MAPbBr₃ were studied by XRD (Fig. S3, ESI†) and SEM (Fig. S4, ESI†), where the thickness of ca. 560 nm is observed for FAPbI₃ and MAPbI₃. As shown in Fig. 3a, the device based on FAPbI₃ shows the highest voltage of 0.4 V, while the MAPbI₃ and MAPbBr₃ show reduced voltages of 0.05, and ∼0 V, respectively. The current of the FAPbI₃ (∼1.2 μA, Fig. 1b) is also higher than the MAPbI₃ (∼0.6 μA) and the MAPbBr₃ (∼0.2 μA), as shown in Fig. S5 (ESI†). The band energy levels of the FAPbI₃, MAPbI₃, MAPbBr₃, and spiro are shown in Fig. 3b.³⁶ Interestingly, though the FAPbI₃ and MAPbI₃ have similar VB and CB levels, the voltage and current performances of the FAPbI₃ device are about 8 and 2 times higher than the device based on MAPbI₃, respectively. This result is different from the perovskite solar cells, in which the band energy levels play a key role in the carrier transfer between perovskite and CTL. In the dynamic device, the carrier generation and separation processes are dominated by the triboelectric effect at the interface. Thus, the band energy level is not the key point to control the voltage and current performances.

Fig. 3 Parameters affecting triboelectric voltage. (a) Voltage performance of FAPbI₃, MAPbI₃, and MAPbBr₃ with spiro. The contact area was ∼1 cm² and the applied force was ∼5 N. (b) Schematic illustration of band energy levels of FAPbI₃, MAPbI₃, MAPbBr₃, and spiro. (c–f) Topography images and (g) work functions of FAPbI₃, MAPbI₃, MAPbBr₃, and spiro. (h) Dependence of voltage output on applied force on the dynamic FAPbI₃/spiro heterojunction.

To explore the factors that determine the voltage and current performances, piezoelectric force microscope, kelvin probe force microscope, Hall effect measurements were conducted on these samples. As shown in Fig. S6 (ESI†), although FAPbI₃ shows a larger piezoresponse value compared to MAPbI₃, and MAPbBr₃ in d₃₃ direction, it is a very small value compared to the reference (Al₂O₃) sample indicating that the DC power-generating performance is hard to be determined by the piezoelectric property of the perovskites. The topography images of FAPbI₃, MAPbI₃, MAPbBr₃, and spiro are shown in Fig. 3c–f, which shows uniform films with good coverage. As shown in Fig. 3g and Fig. S7 (ESI†), the work function, determined by surface potential, of FAPbI₃, MAPbI₃, MAPbBr₃, and spiro are 5.05, 5.46, 5.31, and 5.37 eV, respectively. It should be mentioned that the work function difference between FAPbI₃/spiro (0.32 eV) is larger than that of MAPbI₃/spiro (0.09 eV) and MAPbBr₃/spiro (0.06 eV), which proves a higher driving force for carrier transfer between FAPbI₃ and spiro than that between MAPbI₃ or MAPbBr₃ and spiro. Therefore, we attribute the large voltage performance in the FAPbI₃/spiro device to the large work function difference between FAPbI₃ and spiro, while the devices based on MAPbI₃ and MAPbBr₃ show significantly reduced voltages because of the small work function differences with spiro.³⁷ The electronic properties of these perovskites are studied by Hall effect and summarized in Table 1. The FAPbI₃, MAPbI₃, and MAPbBr₃ show similar Hall mobilities, while the carrier concentrations of FAPbI₃ and MAPbI₃ are 3 and 10 times higher than that of MAPbBr₃, respectively. Thus, the high carrier concentration can also contribute to the high voltage and current performances. It should be noted that the voltage and the current for MAPbI₃ are low although MAPbI₃ has a higher carrier concentration and mobility than FAPbI₃. We attribute this to the small work function difference between MAPbI₃ and spiro, which results in the low carrier separation at the interface.

Table 1 Carrier concentration and Hall mobility of FAPbI₃, MAPbI₃, MAPbBr₃, and PEA₂PbI₄

Materials	Carrier concentration (10^16 m^−3)	Hall mobility (10^−2 m^2 V^−1 s^−1)
FAPbI3	1.5	5.5
MAPbI3	4.8	6.4
MAPbBr3	0.44	6.2
PEA2PbI4	1.5	0.8

---

Spiro films with different areas were used to study the effect of contact area on the voltage and current performances. As shown in Fig. S8 (ESI†), with the contact area increasing from 0.3, 0.6, and to 1 cm², the current generated from the perovskite/spiro heterojunction increases from 0.6, 0.9 to 1.2 μA, while the voltage is hardly influenced. Furthermore, different applied forces were also applied to the dynamic perovskite/spiro heterojunction. As shown in Fig. 3h, in the applied force ranging from 10 to 40 N, the voltage increases with the applied force on the dynamic device, and the highest voltage of 1.3 V can be achieved under 40 N. The current also slightly increases with the applied force, as shown in Fig. S9 (ESI†).

To reveal the relationship for the DC power generation caused by the directional carrier transfer and band alignments, tunable heterojunction architectures between perovskite and electron transport layer (ETL) or HTL were studied. The detailed relationship between triboelectric voltage output and heterojunction architectures are shown in Fig. 4. As shown in Fig. 4a, the CB of PEA₂PbI₄ (PEA = phenylethylammonium) two-dimensional (2D) perovskite is lower than that of spiro, and its VB position is above that of spiro, which exhibits Type I alignment.^31–33 When the electric energy generated by the triboelectric effect, electrons, and holes on the contact surface will transfer to 2D perovskite in the same direction driven by the band energy difference. As a result, an AC power is observed in Fig. 4c and Fig. S10 (ESI†), which is due to the unseparated electron and hole transfer routes. For the FAPbI₃/spiro heterojunction (see Fig. 2f), the CB and VB levels of FAPbI₃ are all lower than those of spiro, which forms a Type II alignment.³⁴ Thus, the electrons will transfer to that of FAPbI₃, while the holes will spontaneously transfer to that of spiro. Unlike the Type I alignment, the Type II heterojunction shows separated electron and hole transfer routes.³⁵ Thus, a DC power is observed (see Fig. 1c). To further confirm the directional carrier transport by the band alignment, Type II alignment between TiO₂ ETL and FAPbI₃ was studied, as shown in Fig. 4b. In this TiO₂/FAPbI₃ junction, the electrons will transfer to TiO₂ and holes will transfer to FAPbI₃, which is opposite to that of the FAPbI₃/spiro heterojunction. As expected, a negative DC output is observed, as shown in Fig. 4d and Fig. S11 (ESI†). Based on these analysis of band alignments, we conclude that the DC power is due to the directional carrier transfer by the band alignment.

Fig. 4 Schematic illustration of different heterojunction architectures and voltage output. (a) Energy band schematic and (c) voltage performance of the dynamic 2D PEA₂PbI₄/spiro heterojunction. (b) Energy band schematic and (d) voltage performance of the dynamic FAPbI₃/TiO₂ heterojunction. (e) IV curve of dynamic perovskite/spiro heterojunction device under sliding and 1 sun illumination. Inset shows a schematic receiving mechanical and solar energies. (f) Energy band schematic showing the movement of charge carriers created by triboelectric and photovoltaic effects. The contact area was ∼1 cm² and the applied force was ∼5 N.

As the perovskite is an excellent photovoltaic material, the influence of the photovoltaic effect on the energy output from the triboelectric effect was studied. The IV curve for the static and dynamic device under 1 sun (100 mW cm⁻²) illumination is shown in Fig. 4e. Strikingly, the current of the dynamic device under light is ∼600 μA, which is higher than the sum of the triboelectric effect (∼1.2 μA in Fig. 1d) and photovoltaic effect (∼200 μA). The voltage of the dynamic device under light is ∼0.95 V, which is also higher than the decoupled triboelectric effect (∼0.4 V) and photovoltaic effect (∼0.65 V). We attribute the high current and voltage output to the coupling effect of triboelectric and photovoltaic effects as shown in Fig. 4f. The FAPbI₃ with a bandgap of ∼1.5 eV can generate more carriers under light illumination and the carriers can be efficiently separated at the dynamic perovskite/spiro, leading to enhanced current and voltage output. The influence of the thickness of the perovskite and the spiro films on the triboelectric–photovoltaic coupling effect is studied. As shown in Fig. S12 (ESI†), the perovskite film with a thickness of ∼600 nm shows the highest current and voltage. Lower performance for thinner perovskite films (∼70 nm and ∼250 nm) is attributed to the reduced light absorption, and for thicker perovskite film (∼900 nm), the reduced charge collection might result in low performances. A similar trend of thickness effect is observed for spiro films, where a thickness of 200 nm shows the highest performance.

The stability of the dynamic device is studied. The SEM image (Fig. S13, ESI†) of spiro film after 20 cycles under the applied force of 40 N shows small indentations compared to the pristine film, which no mechanical damage is observed for the perovskite film. No degradation in performance is observed for 22 cycles under 40 N as shown in Fig. S14 (ESI†). A high endurance is confirmed by continuous working for 4000 s (Fig. S15, ESI†), indicating the stable output of the dynamic device.

To explore the universal applications of this perovskite/CTL heterojunction, PTAA and SnO₂ are also applied, which shows voltages of 0.25 and 0.25 V, current of 0.2 and 0.3 μA, respectively, as shown in Fig. S16 and S17 (ESI†), confirming positive voltage output for PTAA and negative voltage output for SnO₂. The FAPbI₃/PEA₂PbI₄ dynamic device is also tested, which delivers DC outputs of 0.3 V and 0.25 μA as shown in Fig. S18 (ESI†). A FAPbI₃/spiro dynamic device with a SnO₂ ETL under the perovskite layer is measured (Fig. S19, ESI†), where the device shows a similar voltage (∼0.4 V) with the device without SnO₂.

Conclusion

This work introduces a new DC energy generation based on charge carrier generation by triboelectric effects and directional carrier transfer by dynamic perovskite/CTL heterojunction. A range of perovskite and CTL materials were applied to construct a set of rules for the design of this new device. Our findings suggest that: (1) the suitable alignment of band position and bandgap can help separate the carrier transfer; (2) at a given band alignment, the voltage output performance is highly dependent on work function difference between two materials and (3) a high carrier concentration can contribute to the enhanced DC power-generating performances. These findings provide a new prospect in the dynamic perovskite/CTL heterojunctions compared with the static perovskite/CTL junctions in solar cells.

Experimental section

Materials

Formamidinium iodide (FAI), methylammonium iodide (MAI), methylammonium bromide (MABr) were prepared by reacting 0.04 M formamidinium acetate (methylammonium ethanol solution for MAI and MABr) with 0.05 M HI (57 wt% in water, Sigma Aldrich) (0.05 M HBr for MABr) in ice bath for 2 h. Then the solution was evaporated at 70 °C for 1 h until solid powder was formed. The solid powder was washed with diethyl ether (99.0%, Samchun) and recrystallized using anhydrous ethanol for about 3 times. The resulting powder was dried under vacuum for 2 days. FAPbI₃ powders were synthesized by dissolving 1 M FAI and 1 M PbI₂ in γ-butyrolactone (99.5%, Samchun). Then the solution was heated at 130 °C for about 3 h until black crystals were formed. The FAPbI₃ crystals were washed by acetonitrile (DAEJUNG) and diethyl ether for about 2 times and then heated at 150 °C for about 30 min. Spiro-MeOTAD was purchased from Merck. Colloidal tin oxide solution (SnO₂, 15% in H₂O) was purchased from Alfa Aesar. Other chemicals were purchased from Sigma-Aldrich.

Device fabrication

FTO glass substrates were cleaned with acetone and ethanol under ultrasonication sequentially. The FAPbI₃ precursor solution was prepared by dissolving the pre-synthesized FAPbI₃ powder in a mixed solvent (DMF/NMP = 8 : 1), where 30 mol% MACl were included. The MAPbI₃ and MAPbBr₃ were prepared by dissolving PbI₂ with MAI and MABr in the same solvent (DMF/NMP = 8 : 1). The concentration of the perovskite solution was 1.6 M. The perovskite films were prepared by spin-coating 20 μL of the precursor solution on the FTO glasses at 4000 rpm for 20 s, where 1 mL of diethyl ether was dropped on the rotating substrate 10 s after spinning. Subsequently, the sample was annealed at 150 °C for 10 min. The spiro-MeOTAD layer was prepared by spin-coating of 20 μL of the stock solution comprising 60 mg of spiro-MeOTAD in 0.7 mL chlorobenzene (CB) including 25.5 μL of tBP (tert-butylpyridine) and 15.5 μL of the Li-TFSI solution (520 mg Li-TFSI in 1 mL acetonitrile (Sigma-Aldrich, 99.8%)) at 3000 rpm for 30 s. For a PTAA layer, PTAA was dissolved in CB with a concentration of 5 mg mL⁻¹, which was spin coated on the FTO substrate with a speed of 3000 rpm for 30 s. For the SnO₂ layer, a concentration of 4 mg mL⁻¹ was spin-coated at the speed of 4000 rpm for 20 s and then annealed at 180 °C for 30 min.

The samples for PL and TRPL were prepared on flexible PEN substrate. The perovskite film was prepared on the PEN/ITO substrate and spiro film was spin coated on PEN/ITO substrate as mentioned above.

Structure and surface characterization

X-Ray diffraction (XRD) results of the samples were obtained by a Rigaku SmartLab diffractometer, where Cu Kα radiation was used (λ = 1.5406 Å). The morphology of the samples was characterized by a scanning electron microscope (SEM) (JSM-7600F, JEOL). Work functions and topographies of the perovskites and spiro samples were obtained by using Kelvin probe force microscope (KPFM) (XE-100, Park Systems). Piezoresponse in d₃₃ direction of each perovskite sample was measured by piezoresponse force microscopy (PFM) (XE-100, Park Systems)

Optical characterization

Optical absorption spectra were measured by a UV-vis spectrometer (Lambda 45, PerkinElmer). Steady-state PL was measured using a fluorescence spectrometer (QuantaurusTau C11367-12, Hamamatsu). TRPL was performed on the perovskite films by a time-correlated single photon counting (TCSPC) system by FluoTim 200 (PicoQuant). The perovskite films were excited by 464 nm laser source and then the emitted PL from the samples was collected by a photon multiplier tube detector (PMA 182, PicoQuant-GmbH).

Device characterizations

The voltage and current analysis of dynamic devices were measured using a Keithley 2400 source meter and Keithley 6514 electrometer. The CTL films were pressed and sliding on the fixed perovskite film to make the electric contact (see Video S1 under room light, ESI†). All the films were prepared on conductive substrates as mentioned in the device fabrication part and the applied force was generated using a setup (Fig. S20, ESI†).

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Research Foundation of Korea (NRF) grants funded by the Ministry of Science and ICT (MSIT) of Korea under contracts NRF-2016M3D1A1027663 and NRF-2016M3D1A1027664 (Future Materials Discovery Program) and NRF-2018R1A2A1A19021947 (the Basic Science Research Program). This research was in part supported by Energy Technology Program of the Korea Institute of Energy Technology Evaluation and Planning (KETEP), funded by the Ministry of Trade, Industry & Energy (No. 20193091010310).

References

1. R. Hinchet, H.-J. Yoon, H. Ryu, M.-K. Kim, E.-K. Choi, D.-S. Kim and S.-W. Kim, Science, 2019, 365, 491–494.
2. F.-R. Fan, Z.-Q. Tian and Z. L. Wang, Nano Energy, 2012, 1, 328–334.
3. M. Seol, S. Kim, Y. Cho, K. E. Byun, H. Kim, J. Kim, S. K. Kim, S. W. Kim, H. J. Shin and S. Park, Adv. Mater., 2018, 30, 1801210.
4. M.-M. Yang, D. J. Kim and M. Alexe, Science, 2018, 360, 904–907.
5. B. O'regan and M. Grätzel, Nature, 1991, 353, 737–740.
6. A. Kojima, K. Teshima, Y. Shirai and T. Miyasaka, J. Am. Chem. Soc., 2009, 131, 6050–6051.
7. H.-S. Kim, C.-R. Lee, J.-H. Im, K.-B. Lee, T. Moehl, A. Marchioro, S.-J. Moon, R. Humphry-Baker, J.-H. Yum and J. E. Moser, Sci. Rep., 2012, 2, 591.
8. Q. Dong, Y. Fang, Y. Shao, P. Mulligan, J. Qiu, L. Cao and J. Huang, Science, 2015, 347, 967–970.
9. J. Chen and N. G. Park, Adv. Mater., 2019, 31, 1803019.
10. S. I. Seok, M. Grätzel and N. G. Park, Small, 2018, 14, 1704177.
11. N.-G. Park, ACS Energy Lett., 2019, 4, 2983–2985.
12. D. Luo, R. Su, W. Zhang, Q. Gong and R. Zhu, Nat. Rev. Mater., 2019, 5, 1–17.
13. R. Ding, X. Zhang and X. W. Sun, Adv. Funct. Mater., 2017, 27, 1702207.
14. D. B. Kim, K. H. Park and Y. S. Cho, Energy Environ. Sci., 2020, 13, 2077–2086.
15. L. Shu, S. Ke, L. Fei, W. Huang, Z. Wang, J. Gong, X. Jiang, L. Wang, F. Li, S. Lei, Z. Rao, Y. Zhou, R. K. Zheng, X. Yao, Y. Wang, M. Stengel and G. Catalan, Nat. Mater., 2020, 19, 605–609.
16. G. Hu, W. Guo, R. Yu, X. Yang, R. Zhou, C. Pan and Z. L. Wang, Nano Energy, 2016, 23, 27–33.
17. H. Askari, A. Khajepour, M. B. Khamesee, Z. Saadatnia and Z. L. Wang, Nano Today, 2018, 22, 10–13.
18. K. H. Kim, B. Kumar, K. Y. Lee, H. K. Park, J. H. Lee, H. H. Lee, H. Jun, D. Lee and S. W. Kim, Sci. Rep., 2013, 3, 2017.
19. S. Lin, Y. Lu, S. Feng, Z. Hao and Y. Yan, Adv. Mater., 2019, 31, e1804398.
20. G. C. Yoon, K.-S. Shin, M. K. Gupta, K. Y. Lee, J.-H. Lee, Z. L. Wang and S.-W. Kim, Nano Energy, 2015, 12, 547–555.
21. H.-K. Park, K. Y. Lee, J.-S. Seo, J.-A. Jeong, H.-K. Kim, D. Choi and S.-W. Kim, Adv. Funct. Mater., 2011, 21, 1187–1193.
22. J. Liu, A. Goswami, K. Jiang, F. Khan, S. Kim, R. McGee, Z. Li, Z. Hu, J. Lee and T. Thundat, Nat. Nanotechnol., 2018, 13, 112–116.
23. H. Shao, J. Fang, H. Wang, L. Dai and T. Lin, Adv. Mater., 2016, 28, 1461–1466.
24. Z. Hao, T. Jiang, Y. Lu, S. Feng, R. Shen, T. Yao, Y. Yan, Y. Yang, Y. Lu and S. Lin, Matter, 2019, 1, 639–649.
25. P. K. Nayak, S. Mahesh, H. J. Snaith and D. Cahen, Nat. Rev. Mater., 2019, 4, 269–285.
26. C. Ma and N.-G. Park, Chem, 2020, 6, 1254–1264.
27. M. Kim, G.-H. Kim, T. K. Lee, I. W. Choi, H. W. Choi, Y. Jo, Y. J. Yoon, J. W. Kim, J. Lee and D. Huh, Joule, 2019, 3, 2179–2192.
28. H. Shao, J. Fang, H. Wang, H. Zhou and T. Lin, J. Mater. Chem. A, 2017, 5, 8267–8273.
29. X. Li, T. H. Lau, D. Guan and Y. Zi, J. Mater. Chem. A, 2019, 7, 19485–19494.
30. M. E. Kamminga, H.-H. Fang, M. R. Filip, F. Giustino, J. Baas, G. R. Blake, M. A. Loi and T. T. M. Palstra, Chem. Mater., 2016, 28, 4554–4562.
31. R. K. Misra, B.-E. Cohen, L. Iagher and L. Etgar, ChemSusChem, 2017, 10, 3712–3721.
32. J. Qing, X.-K. Liu, M. Li, F. Liu, Z. Yuan, E. Tiukalova, Z. Yan, M. Duchamp, S. Chen, Y. Wang, S. Bai, J.-M. Liu, H. J. Snaith, C.-S. Lee, T. C. Sum and F. Gao, Adv. Energy Mater., 2018, 8, 1800185.
33. A. N. Cho and N. G. Park, ChemSusChem, 2017, 10, 3687–3704.
34. Y. Zhu, T. Wan, X. Wen, D. Chu and Y. Jiang, Appl. Catal., B, 2019, 244, 814–822.
35. T. Zhang, J. Wu, P. Zhang, W. Ahmad, Y. Wang, M. Alqahtani, H. Chen, C. Gao, Z. D. Chen and Z. Wang, Adv. Opt. Mater., 2018, 6, 1701341.
36. Y. Lu, Z. Hao, S. Feng, R. Shen, Y. Yan and S. Lin, Science, 2019, 22, 58–69.
37. D.-Y. Son, S.-G. Kim, J.-Y. Seo, S.-H. Lee, H. Shin, D. Lee and N.-G. Park, J. Am. Chem. Soc., 2018, 140, 1358–1364.
38. J.-W. Lee, S.-H. Bae, Y.-T. Hsieh, N. De Marco, M. Wang, P. Sun and Y. Yang, Chem, 2017, 3, 290–302.
39. C. Ma, M.-F. Lo and C.-S. Lee, J. Mater. Chem. A, 2018, 6, 18871–18876.
40. R. Xu, Q. Zhang, J. Y. Wang, D. Liu, J. Wang and Z. L. Wang, Nano Energy, 2019, 66, 104185.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0ee03487h

‡ These authors contributed equally to the work.

---