Nanostructure engineering for ferroelectric photovoltaics

Abstract

Ferroelectric photovoltaics have attracted increasing attention since their discovery in the 1970s, due to their above-bandgap photovoltage and polarized-light-dependent photocurrent. However, their practical applications have been limited by their weak visible light absorption and low photoconductivity. Intrinsic modification of the material, such as bandgap tuning through chemical doping, has proven effective, but usually leads to the degradation of ferroelectricity. Recently, various nanostructures, such as multilayer heterojunctions, nanoparticles, vertically aligned nanocomposites and polar nanoregions, have been developed to enhance photovoltaic performance. These approaches enable the nanoassembly of materials in a lower-dimension manner to optimize the bulk photovoltaic effect whilst effectively preserving or even inducing ferroelectricity. This review highlights the fabrication processes of these emerging ferroelectric nanostructures and evaluates their photovoltaic performance.

---

Wenzhong Ji

Wenzhong Ji is a PhD candidate at the Research School of Chemistry (RSC), the Australian National University (ANU). He received his B.E. and M.E. degrees from Northwestern Polytechnical University in 2017 and 2020, respectively. His research focuses on the fabrication of ferroelectric nanostructures by the pulsed laser deposition technique and exploring their potential photoelectric applications.

Teng Lu

Teng Lu received his PhD degree from the ANU in 2018. He now works as an Australian Research Council (ARC) Discovery Early Career Researcher Award (DECRA) fellow at the RSC, ANU. His research focuses on the development of polar functional materials for energy conversion and storage, as well as the investigation of their structure–property relationships at different length/time scales and under various external stimuli, such as electric fields, pressure, temperature, and light.

Yun Liu

Distinguished Professor Yun Liu holds the prestigious Australian Research Council (ARC) Georgina Sweet Australian Laureate Fellowship and leads the Functional Materials Research Group at the RSC, ANU. Her research focuses on the understanding of the chemistry–structure–property relationships of solids at multiple length scales and their multi-functional coupling effects. Professor Liu pioneers the application of complex materials chemistry to design novel functional materials for use in the electronic technology, energy and environment. She has extensive expertise in ferroelectric, piezoelectric, pyroelectric, dielectric, magnetic, multiferroic, and spintronic materials, and their device applications.

---

1. Introduction

Since the discovery of the ferroelectric (FE) photovoltaic (PV) effect in the 1970s,^1,2 it has been intensively investigated both theoretically and experimentally.^3–5 Unlike traditional semiconductor PVs used for solar cells, which rely on p–n junctions for photo-excited charge carrier separation and transportation, the photovoltaic effect in FE materials is a bulk behavior – the internal electric field throughout the bulk region induced by the electric polarization provides a driving force. Thus, this effect is also called the bulk photovoltaic effect (BPE). One characteristic of the BPE is that the open-circuit photovoltage is far beyond the limitation determined by the material's bandgap (E_g);⁶ in some cases two to four orders of magnitude higher.⁷ These BPE materials therefore hold promising potential for third-generation solar cells⁸ exceeding Shockley–Queisser limitations.⁹ The BPE also presents polarized-light-dependent photocurrent arising from the coupling between the light polarization and materials’ spontaneous electrical polarization.

Coupling between reversible electric polarization and strain, magnetism and light in ferroelectricity would further push photoferroelectric applications beyond the scope of energy conversion.^10–12 Additionally, the intrinsic BPE enables simple device design. Therefore, FE materials have been used for applications in various photoelectronic devices¹³ such as ferroelectric photovoltaic synapses,^14,15 solid state memory devices,^16,17 and polarized-light-sensitive photodetectors,¹⁸ as well as for optically controlled electro-resistance and electrically controlled photovoltage in ferroelectric tunnel junctions.^19,20 The combination of BPE and pyroelectricity in FE materials has also been used in temperature/infrared sensors.^21,22

However, the main problems of FE materials are their large bandgaps and low electronic conductivity, which lead to limited utilization of visible light and relatively low photocurrent. Many strategies to narrow the bandgap have been proposed, such as ionic doping, but these influence the intrinsic ferroelectricity. In contrast, recently emerging nanostructure design strategies tend to be a more practical solution; that is, they are more controllable and can effectively maintain the ferroelectric features. This review aims to summarize the fabrication processes of these nanostructures and compare their bulk photovoltaic performance. Although many articles have reported the combination of ferroelectric materials with organic trihalide perovskites (OTPs) to achieve high power conversion efficiency, the photovoltaic performance mainly depends on the OTP.²³ Therefore, this review only focuses on the structure design of inorganic ferroelectric materials for enhancing or creating BPE.

In the second section of this review, the ballistic and shift current theories will be briefly introduced. In the third section, conventional methods for enhancing BPE through materials modification are summarized before discussing emerging nanocomposites formed by structures of various dimensionalities including multilayers, nanoparticles, and vertically aligned structures and their unique properties in sections 4.1, 4.2 and 4.3. In section 4.4, local polar structures are discussed separately, as these materials can be synthesized from a single raw material in one process, but form a polar nanoregions (PNRs). In the above ferroelectric structures, the materials are synthesized by chemical sol–gel, pulsed electron deposition (PED), pulsed laser deposition (PLD) or RF sputtering methods. In recent years, ferroelectricity in low-dimensional materials has generated much interest, and these materials can form more unique structures through exfoliation and transformation. Section 4.5 will briefly discuss the BPE in these materials. In the final two sections of the review. The photovoltaic performance of the different structures is outlined, and perspectives are provided for future research and technology development.

2. Bulk photovoltaic theory

The BPE has been observed in noncentrosymmetric crystals that belong to 20 point groups, including ferroelectrics and piezoelectrics. The origin of this effect in ferroelectric materials has been attributed to their bulk photovoltaic effect (BPE),^24,25 crystal structure,²⁶ domain walls,^6,27,28 Schottky barriers²⁹ and depolarization fields.^30,31

Two main theories have been proposed to explain the BPE in ferroelectrics: the ballistic and shift current theories.³² In the ballistic theory, photoexcited non-thermalized carriers have asymmetric momentum distribution in the conduction band, as shown in Fig. 1. In this theory, the carriers lose energy and descend to the bottom of the band over a free path l₀. In the shift current theory, the BPE is quantum-mechanical in nature. Taking into account the non-diagonal elements of the density matrix,³³ the BPE is not caused by the carrier movement in the band, but instead by virtual shift R in real space following the carrier band–band transition. These two theories are discussed in further detail below.

Fig. 1 Nonequilibrium carriers’ momentum distribution in (a) centrosymmetric crystals and (b) noncentrosymmetric crystals. Reprinted with permission,³⁴ Copyright 2014, American Physical Society.

2.1 Ballistic theory^25,34

When a homogeneous medium without an inversion symmetry center is under uniform illumination, a steady-state current j_pv will be generated in short-circuited electrodes; the current value depends on the intensity and polarization of the light. When the electrodes are disconnected, the current j_pv will generate an open-circuit voltage V_oc,where σ_d and σ_pv are the dark- and photoconductivity, respectively, and l is the distance between the two electrodes. Supposing that σ_pv ≫ σ_d, the photoinduced electric field is given by:

The σ_pv can also be expressed as

σ_pv = eI₀αφ(ħω)⁻¹(μτ)_pv (3)

where e is the elementary charge, I₀ is the incident light intensity, α is the absorption coefficient, φ is the quantum field, ħω is the incident photon energy, and μ and τ are the mobility and lifetime of the nonequilibrium carriers that are responsible for the photoconductivity.

The tensor properties of the linear BPE current are described by:

jⁱ_pv = αG_ijle_je_lI₀ (4)

where e_j and e_l are the components of the linear light polarization vector and G_ijl is the third-rank piezoelectric tensor. The corresponding scalar relations are:

j_pv = αGI₀ (5)

Based on eqns (3) and (5), the eqn (2) is modified to

For σ_pv ≫ σ_d, the photovoltaic field E_pv does not vary with I₀, but j_pv scales linearly with I₀. It is also the case that

j_pv = eI₀αφ(ħω)⁻¹ξl₀ (7)

where ξ is the photoexcitation asymmetry parameter and l₀ is the free path in the ballistic mechanism. Combine eqns (5) and (7), l₀ is expressed by

l₀ = Ge⁻¹ħω(φξ)⁻¹ (8)

Here, l₀ depends on the asymmetry parameter and the energy of the exciting light. The energy conversion efficiency η for the unit surface is determined by the ratio between the power dissipated in the load, Q_R = j_pv²R', where R' = σ_pv⁻¹l, and the light absorbed by the unit surface, Q_pv = αI₀l,

Considering eqns (2) and (9), the conversion efficiency is equal to

η = GE_pv (10)

For example, in a (001)-oriented BTO single crystal, G₃₁ = 3 × 10⁻⁹ cm V⁻¹, E_pv ≈ 100–200 V cm⁻¹ and η ≈ 10⁻⁶–10⁻⁷.

2.2 Shift current^35,36

In addition to the electron velocity distribution discussed above, there is another shift current that contributes to the BPE stemming from the carrier shift in real space.^32,35 This is described by:

j_pv = eI₀α(ħω)⁻¹R (11)

where R, the shift vector, represents the carrier displacement in real space per absorbed photon. The shift and ballistic current can coexist under uniform illumination, but the shift current has an atomic-scale relaxation time (≈ 0.1–1 fs), as described bywhere the first term is the linear bulk photovoltaic current from linearly polarized light, and the second term is the circular bulk photovoltaic current from circularly polarized light. i[ee*] determines the degree of circular polarization γ and is expressed as , where q is the photon wave vector. β^l_ijk is a third-rank tensor and β^c_ik is a second-rank tensor relevant to gyrotropic materials. Based on the two theoretical properties that the shift current does not have photo-Hall behavior and the circular current is purely ballistic, the shift current could be directly observed in the presence of a magnetic field.³⁶

In FE materials, the values of both l₀ and R are estimated to be on the order of 10–100 nm.^32,37 As l₀ represents the free path of the hot photocarriers in space, when the film thickness is less than this value, all photoexcited carriers can be efficiently collected by electrodes before they experience severe recombination. For example, when the thickness l of BTO is close to l₀ ≈ 90 nm, η could be significantly increased by five orders of magnitude.³⁴ In terms of shift current, compared to bulk materials, two-dimensional materials can reduce the exciton effect.³⁸ All the above demonstrate the importance and necessity of nanostructure design for achieving high PV performance in FE materials.

3 Intrinsic modification of bulk photovoltaic materials

The power conversion efficiency of ferroelectric thin films is far below the expected value compared with commercial solar cells. The low power conversion efficiency of FE materials (8.1% ³⁹) is attributed to their large bandgaps, which lead to poor light absorption of the solar spectrum and low electronic conductivity. The short-circuit current (I_sc) and open-circuit voltage (V_oc) are figures-of-merit to evaluate photovoltaic performance. Tremendous efforts have been made to improve these two parameters, particularly I_sc.

One strategy to enhance I_sc is reducing the materials’ bandgaps to match the solar spectrum. To achieve this, a large number of strategies have been proposed such as chemical doping,^40,41 ‘gap-state’ engineering⁴² and the use of ternary morphotropic phase boundary system⁴³ to introduce intermediate bands. In addition to the traditional perovskite ferroelectric materials LiNbO₃,¹ BaTiO₃ (BTO),³⁴ BiFeO₃ (BFO)^6,44,45 and Pb(Zr, Ti)O₃ (PZT),⁴⁶ many narrow bandgap ferroelectric materials such as [KNbO₃]_1−x[BaNi_1/2Nb_1/2O_3−δ]_x (KBNNO),⁴⁷ KBiFe₂O₅,⁴⁸ CaMnTi₂O₆,⁴⁹ BaFe₄O₇ ⁵⁰ and hexagonal manganite and ferrite^51–53 have also been explored for BPE in the visible light range. V_oc, on the other hand, can be improved by increasing the material thickness, but I_sc will deteriorate as a result due to the fact that the free path of the charge carrier contributing to PV is only around tens of nanometers (based on BPE theory, which is discussed above). One method employed to avoid this is taking advantage of the domain wall configuration.⁶ This is because, based on BPE theory, V_oc is inversely proportional to the intrinsic film conductivity (eqn (1)). Note that domain walls are more conductive than the bulk region, thus, when the measurement direction is perpendicular to the domain wall rather than in a parallel configuration, a higher V_oc can be obtained.²⁴

Several other methods have also been considered for boosting the BPE. In the FE film fabrication process, the introduction of electronic defects is inevitable. Agarwal⁵⁴ found that oxygen vacancies could be suppressed by isovalent and aliovalent co-doping on A- and B-sites to reduce the leakage current. Additionally, from a microscopic perspective, the interaction between polarization, lattice and orbital plays an important role in the ferroelectric photovoltaic process. Thus, manipulating the polar order has also been proven to be an effective strategy to enhance BPE, including rhombohedral-to-tetragonal structural transition,⁵⁵ FE-paraelectric (PE) phase transition⁵⁶ and polarization gradients in PbTiO₃ ⁵⁷ and KNbO₃.⁵⁸

In all these methods, the improvement in BPE comes from an intrinsic change within the materials. These methods usually require a precise synthesis process, often resulting in the material's physical and chemical properties changing in an unpredictable manner. In addition to the intrinsic properties, many external factors also influence the photovoltaic effect, including the electrode dielectric constant,⁵⁹ photoexcited electrons from low-work-function electrodes,⁶⁰ depletion layer thickness²⁹ and upconverter layer.⁶¹ For example, when the top electrode is nanoscale in size, photoexcited carriers are collected more efficiently, leading to an enhancement in external quantum efficiency by several orders of magnitude.^25,62 Recently, many nanostructures have been designed to combine different materials and create specific geometries for achieving a high BPE. In the following sections, we will present a variety of nanostructures that have been investigated by researchers. It is noteworthy that in addition to the depolarization field (E_dp), other coexisting built-in fields, originating from factors such as interfacial junctions, inhomogeneous chemical distribution or strain gradients, may also contribute to the overall photovoltaic effect.

4 Nanostructures

4.1 Multilayers (2–2 type)

Layer-by-layer deposition is the most straightforward method to achieve nano-structure design. In the context of the BPE, multilayer structures have been exploited to change the Schottky barrier at the FE film–metal interface to different junctions by inserting a semiconductor layer. In recent gradient-doped multilayers, additional internal electric fields induced by flexoelectric polarization or a p–n junction further enhance charge carrier separation. For high-quality epitaxial films, the superlattice allows researchers to manipulate the ordering state at an atomic level and tune the materials’ properties, such as their bandgap and dielectric constant, as desired.

4.1.1 Enhanced transport layers. In ceramics, the interfacial Schottky barrier has negligible impact on the overall photovoltaic output due to the large volume of the active layer, while in the vertical metal–FE thin film–metal geometry, the interfacial contact must be seriously considered and can even dominate the photovoltaic performance.^29,63,64 In metal/semiconductor systems, the Schottky barrier height can be easily estimated by the Fermi level of the metal and conductive band (valence band) of semiconductors, and its width is determined by the charge carrier concentration. When a wide-bandgap ferroelectric material is used to replace the semiconductor, both the barrier height and width are further affected by ferroelectric polarization. For example, the barrier height of PZT/SRO is estimated to be 1.2 eV, but the polarization contribution of PZT to the barrier height and width must be determined from further measurements.²⁹ The barrier widths of metal/ferroelectric films have been reported to be around several nanometers.^29,59,65 To precisely obtain these two parameters, the temperature-dependent I–V curve needs to be measured.⁶⁶ In detail, the thermal emission is expressed as follows:where A* is Richardson's constant, T is the temperature, q is the charge of an electron, k is Boltzmann's constant, Φ⁰_B is the potential barrier height at zero applied field, E_m is the electric field, and ε₀ and ε_op are the permittivity of free space and dynamic (high frequency) dielectric constant, respectively. This equation can be modified as follows:

When V = 0, Φ⁰_B is proportional to the slope of the ln(J/T²) − 1/T plot.^67,68 The depletion width can be calculated using the equation:

where ε_st is the static (low frequency) dielectric constant, is the built-in potential considering polarization charge contribution, and N_eff is the effective charge density. The Schottky barriers in two FE–metal interfaces with opposite directions impede the transport of photoinduced carriers towards the outside circuit. Band structure reconstruction by inserting a semiconductor layer has proven to be an effective way to mediate this process.

For example, Cao et al.⁶⁷ inserted a Cu₂O layer between PZT and Pt (top electrode). The Schottky barrier at the PZT/Pt interface was replaced by a n–n+ junction between PZT/Cu₂O, and on the Cu₂O/Pt side, the tunneling process dominates the conductance behavior. Due to the aligned depolarization and interfacial electric fields after the interface modification, the short-circuit current density increased from 0.04 to 4.8 mA cm⁻² and the power conversion efficiency (PCE) improved from 0.008% to 0.57%, as shown in Fig. 2a. Another metal oxide, ZnO, an n-type semiconductor with high electron mobility, has also been used for the electron transport layer in dye-sensitized solar cells.⁶⁹ In this work,⁷⁰ the ZnO layer was chosen to construct an n–n+ junction with BFO, as illustrated in Fig. 2b. The unidirectional driving force at the two interfaces and the additional photoexcited electrons from ZnO simultaneously enhance the photovoltaic performance. Although the authors claimed that the depolarization field had a minor impact on the photovoltaic process, a high energy conversion efficiency of 0.33% was obtained. Wurtzite ZnO also possesses spontaneous polarization induced by strain; this polarization originates from an asymmetric permanent static electric dipole and cannot be altered by an external electric field. When the ZnO layer was exploited to fabricate the PZT/ZnO heterostructure, a polarization-dependent interfacial coupling effect was observed at the p–n junction interface.⁶⁸ Not only was the PCE improved by 2 orders of magnitude, but it also exhibited a different V_oc and J_sc modulation behavior compared to pure PZT, as shown in Fig. 2c and d. Fig. 2e and f illustrate that both the depletion layer width and potential barrier height can be modulated by the polarization direction in PZT. In another study, the same behavior was observed in a Ag/ZnO/BFTO/NSTO structure.⁷¹ This continuous tunability of photovoltaic output is suitable for random-access memory applications.⁷² In addition to metal oxides, n-type reduced graphene oxide has also been used for constructing a p–n junction with p-type BFCO, which improved J_sc by 1000-fold without affecting the V_oc or the fill factor (FF).⁷³

Fig. 2 (a) Schematic charge transportation process with and without Cu₂O layer. E_Bi-bottom and E_Bi-top are Schottky-barrier-induced electric fields at the top and bottom FE–metal interfaces, respectively. E_DP is the depolarized field and E_PZT/Cu2O is the newly created field. Reprinted with permission,⁶⁷ Copyright 2012, American Chemical Society. (b) Energy band diagram of the Pt/BFO/ZnO/ITO device. Reprinted with permission,⁷⁰ Copyright 2014, AIP Publishing. (c) and (d) are the photovoltaic V_oc and J_sc hysteresis loops upon applying an external poling voltage sequence. (e) and (f) are schematic illustrations of the depletion layer thickness and the barrier height at the PZT/ZnO interface tuned by different polarization states in PZT. Reprinted with permission,⁶⁸ Copyright 2016, Springer Nature.

4.1.2 Gradient doped films. It has been found that the migration of oxygen vacancies affects the photovoltaic effect in BFO films.^74,75 Therefore, it is worth considering utilizing the distribution of oxygen vacancies to tune the BPE efficiency. Short-range ordered oxygen vacancies can induce incommensurate local structure (Fig. 3a) as reported by Schiemer et al. for Bi_1−xCa_xFeO_3−x/2.⁷⁶ To take advantage of , Zhang⁷⁷ investigated gradient-doped multilayer structures (Bi_1−xCa_xFeO_3−x/2). The amount of in each layer was controlled by the doping concentration, as shown in Fig. 3b. These two stacking geometries had opposite self-polarization direction and photocurrent (Fig. 3c). Due to the strong self-polarization and a gradient distribution superposition effect in BFCO-1, the open-circuit voltage and short-circuit current were 2 times and 32 times higher than those of pure BFO, respectively.

Fig. 3 (a) Electron diffraction pattern of Bi_0.8Ca_0.2FeO_2.9 and corresponding structural model. Reprinted with permission,⁷⁶ Copyright 2009, American Chemical Society. (b) Multilayer Bi_1−xCa_xFeO_3−5x (BCFO) structures with different oxygen vacancy gradient distributions. (c) Photovoltaic I–V curves for three devices. Reprinted with permission,⁷⁷ Copyright 2020, Elsevier. (d) Energy diagram of Au/multilayer structure (La, Co)-gradient-doped BiFeO₃/FTO with an oxygen-vacancy-induced field E_bi-gv. (e) Emerging flexoelectricity in the gradient-doped sample. Reprinted with permission,⁷⁸ Copyright 2021, American Chemical Society. (f) Illustration of photovoltaic process in the BNO-N₂/BNO-O₂ structure. (g) Kelvin probe force microscopy (KPFM) characterization of the BNO-N₂/BNO-O₂ interface. The top panel is the morphology, the middle panel is a cross-sectional surface potential image, and the bottom panel is the surface potential profile of the middle panel. Reprinted with permission,⁸² Copyright 2021, Royal Society of Chemistry.

The gradient distribution of in a (La, Co)-gradient-doped BiFeO₃ (BFLCO) multilayer was reported by Sun.⁷⁸ They first conducted XPS and Urbach energy characterization to demonstrate that the concentration increases with increasing the amount of the dopant (La). They then fabricated (Bi_1−xLa_xFe_0.975Co_0.025O₃, x = 0.05, 0.1 and 0.15) multilayer structures. The gradient distribution of established an extra built-in electric field, E_bigv, along with an interfacial field, E_bi-s, responsible for photogenerated carrier separation, as illustrated in Fig. 3d. In addition, another flexoelectric field, E_flexo, was created by the strain gradient, considering that the material unit cell volume was modulated by the doping level (Fig. 3e). E_flexo is described by the equation

where ε₀ is the permittivity of the free space, ε_r and a are the relative permittivity and lattice constant of the sample, e is the charge of an electron and ∂_u/∂_z is the strain gradient, which was estimated by the Williamson–Hall relation. E_bi-s, E_bigv and E_flexo work together to generate a photocurrent three times higher than a single-layer sample. In another work by the same authors,⁷⁹ a Bi_1−xSm_xFe_0.975Ni_0.025O₃ multilayer with similar mechanisms for enhanced BPE was also studied, confirming that this strategy is applicable for many material systems.

To achieve high BPE, removing centrosymmetry is essential.⁸⁰ It has been found that material symmetry can be broken by oxygen vacancies.⁸¹ Inspired by this idea, Mai⁸² successfully created a depolarization field in centrosymmetric BiNbO₄ deposited by a pulsed electron deposition (PED) technique after annealing in an oxygen-deficient atmosphere (BNO-N₂). Later, by combing BNO-N₂ with another layer annealed in oxygen (BNO-O₂), he constructed the BNO-O₂/BNO-N₂ homojunction illustrated in Fig. 3f. In this bilayer system, the depolarization field in BNO-N₂ and the p–n junction in the BNO-N₂/BNO-O₂ interface, which was proven by cross-sectional surface potential measurement as shown in Fig. 3g, synergically contributed to the PV performance. Under solar light illumination, the device generated a V_oc of 0.85 V and a J_sc of 200 μA cm⁻². Another centrosymmetric semiconductor, BiVO₄, which has a narrow bandgap of ∼2.6 eV and high organic pollution degradation efficiency under visible-light illumination, is widely used for photocatalysis.⁸³ During its fabrication, bismuth is prone to evaporate at high temperatures. Taking advantage of this property, Shi⁸⁴ introduced an internal electric field in multilayer BVO with a charge gradient distribution. The structure was fabricated through a multiple-step spin-coating and calcination method. The uneven volatilization of bismuth resulted in the Bi³⁺ content in the top layer being higher than that in the bottom layer. Consequently, the related gradient compressive stress led to local symmetry breaking and local polarization. Additionally, compared with classical ferroelectric materials, the sample exhibited extraordinary thermal stability.

4.1.3 Superlattices. The above-mentioned structures are generally synthesized by wet-chemical methods. The obtained films are usually polycrystalline and contain a number of defects. With the development of film deposition techniques, more delicate epitaxial multilayer structures can be fabricated using the pulsed laser deposition method (PLD).

Among the various FE materials, BFO has been intensively used for BPE due to its relatively narrow bandgap of ∼2.7 eV and high polarization of ∼90 μC cm⁻², but it can still only absorb 20% of solar irradiation. The double perovskite Bi₂FeCrO₆ was predicted to have a lower bandgap and maintain a large remanent polarization.⁸⁵ Although synthesizing ordered double-perovskite structures is difficult given the thermodynamic process, superlattice (SL) structuring opens a corridor to achieving an atomically well-ordered system. These materials show many exotic properties, such as superexchange interaction⁸⁶ and polar skyrmions.⁸⁷ Nechache et al.³⁹ found that ordered and disordered phases coexisted in PLD-synthesized Bi₂FeCrO₆ (BFCO) films. The ordered phase intensity ratio (R) and domain size (D) are controlled by the deposition temperature and growth time, respectively. These two parameters significantly affect the film bandgap, optical absorption behavior and FE polarization magnitude, as indicated in Fig. 4a–c. At the optimized temperature, by gradually changing the laser repetition during the deposition process, multilayer structures were constructed in which the absorption capability of each layer was superposed, and the bandgap could be reduced to the range of 1.43–2.51 eV. Finally, a recorded high PEC of 8.1% in ferroelectric material was achieved. Regarding this work, a first-principles calculation showed that the photovoltaic improvement is related to efficient electron–hole separation on the Fe and Cr sites,⁸⁸ and five other promising Bi-based double-perovskite oxides with similar mechanisms were predicted, including (Ti, V), (V, Cu), (Co, Cr), (Mn, Co) and (Fe, V). Based on density functional theory calculations, a narrow bandgap of 1.6 eV can also be obtained in the (BiCrO₃)m/(BiFeO₃)n superlattice,⁸⁹ which is correlated to the lattice strain, octahedral distortion and polarization discontinuity at the interface. As ferroelectric polarization is maintained in the system, this structure has potential for highly efficient BPE applications. In another system consisting of PbNiO₂ and PbTiO₃ nanolayers, theoretical calculations showed a much lower bandgap (1.5 eV) and 43-fold shift-current enhancement.⁹⁰

Fig. 4 (a–c) Tuneable absorption (a), bandgap (b) and ferroelectric polarization (c) in BFCO by ordering characteristics R and D. Reprinted with permission,³⁹ Copyright 2015, Springer Nature. (d) High-resolution cross-sectional STEM of the STO/BTO/CTO structure. (e) and (f) are the spectral photo-response and temperature-resolved relative permittivity measured at 100 kHz. Reprinted with permission,⁹¹ Copyright 2021, The American Association for the Advancement of Science.

Recently, a tricolor SL structure of (SrTiO₃)x/(BaTiO₃)y/(CaTiO₃)z, where x, y and z represent the number of unit cells, was experimentally investigated for BPE⁹¹ (Fig. 4d). It could generate a photocurrent three orders higher than a single BaTiO₃ layer with similar thickness. By studying the two-color SLs STO/BTO and BTO/CTO, the authors ruled out the influence of the interface and the clipping effect of the substrate. As shown in Fig. 4e and f, the tricolor structure demonstrated a lower bandgap and large dielectric permittivity (ε_r), which was further affected by the periodicity (x, y, z). In conjunction, the layered arrangement caused subtle displacement of the cations and therefore modified the electronic structure. It has been proven by others that high ε_r promotes exciton formation, which contributes to the photocurrent.

4.2 FE matrix-nanoparticle composites (0–3 type)

The photovoltaic effect consists of three key steps: photoexcited e–h pair generation, charge carrier separation and transportation. Given that the depolarization field in FE materials is responsible for charge separation and the electrodes for transportation, incorporating narrow bandgap semiconductors or taking advantage of surface plasmon resonance effect to capture more photons will compensate for the weak light absorption of FE materials. Yang et al.⁹² sputtered a metallic silver target to synthesize elliptical Ag₂O nanoparticles (Ag₂O-NPs) between two PZT layers, as shown schematically in Fig. 5a. The size was controlled by the sputtering time. The composite showed additional absorption from 350 to 800 nm, and more importantly, the electric polarization within PZT was retained. Although there was a p–n-junction-induced electric field E_bi at the left and right PZT-Ag₂O junctions, the e–h pairs generated from the Ag₂O NPs were mainly separated by the depolarization field in bulk PZT, as shown in Fig. 5b, and the photocurrent could be tuned via the magnitude of the external poling voltage. Finally, the authors observed an approximate 80-fold improvement of the photocurrent under illumination of a 150 W Xe bulb. The maximum light-conversion efficiency was 0.195% with J_sc = 0.47 mA cm⁻², V_oc = 0.69 V and FF = 0.6.

Fig. 5 (a) Schematic structure of PZT/Ag₂O-NPs/PZT and the photovoltaic process. (b) Energy band structure between Ag₂O-NPs and PZT with the polarization pointing left. E_bi is the electric field at the PZT-Ag₂O junction. Reprinted with permission,⁹² Copyright 2012, John Wiley and Sons. Energy band structure of CFO-PZT composite and the corresponding photoexcited carrier generation, separation and transportation process with (c) and without (d) an external magnetic field. Reprinted with permission,⁹⁴ Copyright 2016, AIP Publishing. (e) Atomic force microscopy (AFM) image of BWO nanonets grown on STO. (f) TEM image of the heterojunction between Bi₂Fe₄O₉ nanoparticles and BWO. Reprinted with permission,⁹⁸ Copyright 2024, AIP Publishing.

CoFe₂O₄ is a member of the magnetic ferrites with a bandgap of around 1.5 eV.⁹³ Pan et al.⁹⁴ used sol–gel and spin-coating to fabricate a 0–3 type CoFe₂O₄-Pb(Zr, Ti)O₃ (CFO-PZT) composite in which cubic CFO nanoparticles were uniformly distributed in the PZT matrix. Due to the optical bandgap of CFO being smaller than that of PZT, the CFO-PZT composite had a lower bandgap (∼3.48 eV) compared to that of the pure PZT film (∼3.6 eV). In this work, the interfacial barrier between the PZT and CFO impedes photo-induced barriers crossing the interfaces, but the photovoltaic effect can be manipulated by both electric and magnetic fields via a magnetostrictive effect in CFO and piezoelectric effect in PZT (Fig. 5c and d). This work demonstrates that multiferroic composite materials composed of independent ferroelectric and ferromagnetic materials possess strong coupling effects through interfacial stress transfer, and therefore have an impact on the PV effect. The ion implantation technique has been used to fabricate multiferroic materials by introducing transition-metal particles such as Fe⁹⁵ and Co⁹⁶ into BTO single crystals. Given the formation of narrow bandgap metal oxides,⁹⁷ their photovoltaic performance should be influenced, but this has not yet been studied.

When growing ferroelectric Bi₂WO₆ (BWO) on STO via PLD, Sun⁹⁸ found that BWO formed a cross-oriented nanonet topography due to the different interfacial free energies and anisotropy of the Aurivillius-type crystals. The morphology is shown in Fig. 5e. Then, mullite-type Bi₂Fe₄O₉ nanoparticles were deposited to create a p–n junction (Fig. 5f). The J_sc and V_oc in this structure were four and three orders of magnitude higher than those of pure BWO, respectively. The photocurrent and device resistance could both be modulated by the ferroelectric polarization. Compared with multilayer structures, this nanoparticle/nanonet system has a large interfacial junction area. p-type Sb₂Se₃ has also been chosen to form a p–n junction with BWO for photovoltaic applications.⁹⁹

In metal nanoparticles (NPs), the electric field of an incident light beam can polarize surface loosely bound electronic clouds, and this charge separation can be restored by the force from nuclei. Thus, a collective, in-phase conduction electron oscillation occurs, as shown in Fig. 6a. When the frequency of a photon matches the natural frequency of the electron oscillation, surface plasmon resonance (SPR) occurs. This changes the dielectric constant of the material and affects its optical properties.¹⁰⁰ This phenomenon has been found in noble metals, metal oxides (WO_3−x, MoO_3−x and NiO) and metal chalcogenides.¹⁰¹Fig. 6b displays the absorption bands, which are dependent on the material's intrinsic properties. Due to the high electron–hole pair generation rate, hot carrier transfer and reduced charge recombination, SPR fuels photo-related activities such as photocatalysis, photoelectrocatalysis and photodetection.^102–106

Fig. 6 (a) Schematic of surface plasmon resonance. Reprinted with permission,¹⁰⁰ Copyright 2016, RSC Publishing. (b) Absorption ranges of different metal nanoparticles. Reprinted with permission,¹⁰⁷ Copyright 2018, John Wiley and Sons.

To employ SPR, Maruyama et al.¹⁰⁸ fabricated BFO and Ag NP (10–20 nm) thin-film composites via a chemical coating process using colloidal silver and a BiFeO₃ metal–organic precursor solution (Fig. 7a). This composite could absorb light in the range of 400–600 nm, and although the absorption edge did not move after the addition of Ag NPs, the absorption capability in the wavelength range of 400–450 nm was strengthened due to the SPR of the Ag NPs. As a result, the photocurrent was enhanced approximately two-fold in the BFO/Au-NPs/BFO and four-fold in the BFO/Au-NPs/BFO/Au-NPs/BFO films relative to that of the pure BFO film, while the photovoltage was not affected by the incorporation of Ag NPs.

Fig. 7 (a) Schematic illustration of BFO/Au-NPs/BFO (BAB) and BFO/Au-NPs/BFO/Au-NPs/BFO (BABAB) structures grown on Pt/TiO_x/SiO₂/Si substrates. Reprinted with permission,¹⁰⁸ Copyright 2016, IOP Publishing on behalf of the Japan Society of Applied Physics. (b) UV-vis absorption of the BLFO and BLFO/AuNPs films. (c) Schematic scattering effect of metal nanoparticles caused light trapping. Reprinted with permission,¹⁰⁹ Copyright 2017, Royal Society of Chemistry. (d) Energy band diagram of BFO/ZnO and the photovoltaic mechanism under illumination. (e) Energy band diagram of BFO/Au-NPs/ZnO and emerging pyroelectric polarization charges. (f) UV–vis absorption spectra of BFO, BFO/ZnO and BFO/Au-NPs/ZnO. Reprinted with permission,¹¹² Copyright 2021, Elsevier. (g) HRTEM image of Au nanorods synthesized via a seed-mediated growth method. Reprinted with permission,⁷³ Copyright 2020, Elsevier. (h) Illustration of separation of photoexcited carriers in PZT/metal junctions. h_BB and e_BB represent UV-light-induced holes and electrons. h_sp and e_sp indicate visible-light-induced holes and electrons. Reprinted with permission,¹¹³ Copyright 2014, RSC Pub.

In a report by Li et al.,¹⁰⁹ the authors first proved that lanthanum doping could reduce the bandgap of BFO. Subsequently, the visible light absorption of BLFO (550–700 nm) was further enhanced after decorating the surface with Au nanoparticles, as shown in Fig. 7b. Au-PNs 70 nm in diameter were deposited by sputtering. The enhancement was attributed to the light scattering effect increasing the optical path length, as proposed in Fig. 7c, which occurred at large particle size (>40 nm),¹¹⁰ as well as the localized surface plasmon resonance (LSPR) intensifying the optical field. I_sc and V_oc could also be modulated by an external electric field, which provided evidence of BPE. In bulk form, the BTO-Au nanohybrid ceramic showed an additional absorption peak at 550 nm due to LSPR.¹¹¹ The photocurrent was seven times that of a pure BTO ceramic, yet the photovoltage had almost the same value (6 V), indicating negligible influence on the polarization.

Band bending has been utilized for enhancing charge separation, as discussed in section 1.2. In a report by Zhang,¹¹² BFO/ZnO exhibits better photoelectric performance compared with BFO, due to the additional driving force E_bi as shown in Fig. 7d. After modifying the heterostructure with Au-NPs, light absorption was detected in a broad spectrum of 360–1000 nm (Fig. 7f). Interestingly, the BFO/Au-NPs/ZnO generated considerable pyroelectric current, which was negligible in the BFO/ZnO sample due to the screening effect (Fig. 7d and e). The authors also investigated the influence of the sputtering time on the photovoltaic effect due to the change of Au-NP grain size. In addition to nanoparticles, Au nanorods were synthesized using a seed-mediated growth method;⁷³ they were then dispersed in ethanol and deposited between BFCO and rCO. As a result, the photovoltage and photocurrent were increased by 12.5% and 66.2%, respectively, compared to the device without Au NRs.⁷³

Although artificial periodic structures or nanoparticle interfaces are generally required for SPR, Zheng et al.¹¹³ experimentally and theoretically proved that a rough surface can also introduce the same effect. A 300 nm PZT thin film was deposited on ITO via a sol–gel method, and although the bandgap of PZT is about 3.6 eV, meaning it can only absorb ultraviolet light (UV), the film could generate photocurrent under visible light irradiation after sputtering Ag as the top electrode. Through post-annealing to tune the Schottky barrier height at the PZT/Ag junction, they confirmed that the photocurrent originated from the photogenerated electron–hole pairs in the PZT surface due to enhanced near-fields, as illustrated in the bottom panel of Fig. 7h. This mechanism differs from the hot electrons generated in metal nanoparticles.¹¹²

4.3 Vertically aligned structures (1–3 type)

Nanowires are crystalline one-dimensional nanomaterials with high carrier mobility along the growth direction. For example, ferroelectric BiFeO₃ nanofibers demonstrated a photocurrent ∼10 times higher in comparison to the thin film.¹¹⁴ In addition, nanowires also have high flexibility, and BaTiO₃ nanowire (NW) arrays have demonstrated the harvesting of vibration energy.¹¹⁵ Thus, 1D structures serve as ideal configurations for realizing enhanced photovoltaic performance through polarization and strain coupling via the piezoelectric or flexoelectric effect. In the future, nanowires could be integrated into self-powered wearable devices.

Zhang et al.¹¹⁶ grew zinc oxide (ZnO) nanowires on a La-doped bismuth ferrite (BLFO) film. During fabrication, the BLFO was first spin-coated onto FTO, and ZnO seeds were then sputtered onto the film and grew into nanowires in a thermostat oven. Fig. 8a shows a cross-sectional SEM image of the BLFO/ZnO nanowire structure. A finite element analysis (FEA) in COMSOL showed that strain-induced positive piezo-charges accumulated on the BLFO/ZnO interface when the compressive strain was applied. This caused the depletion region to shift to the BLFO side and improved the e–h pair separation efficiency, as indicated in Fig. 8b (from stage 1 to stage 2). After poling the sample, the piezo-photoelectronic effect and the depolarization electric field synergistically enhanced the V_oc, J_sc and response speed (Fig. 8c), making it suitable for a future photodetector application.

Fig. 8 (a) Cross-sectional SEM image of BLFO/ZnO nanowires structure. (b) Energy band diagram of the BLFO/ZnO junction before (stage 1) and after (stage 2) applying compressive force on the top. (c) Photo-response of the device under different strain and polarization conditions. Reprinted with permission,¹¹⁶ Copyright 2020, American Chemical Society. (d) Schematic drawing of the fabrication process of aligned PZT nanowires/PDMS using an AAO template. (e) SEM image of hollow free-standing PZT nanowires after etching of AAO. Reprinted with permission,¹²⁰ Copyright 2022, American Chemical Society. (f) Illustration of preparation processes of three-dimensional ITO/BFO/WO₃. (g) Energy band structure of ITO/BFO/WO₃ and transportation process of carriers. Reprinted with permission,¹²² Copyright 2023, Elsevier.

The flexoelectric effect has been utilized for photovoltaic applications as discussed above.⁷⁸ The strain gradient induced flexoelectric polarization (P_flexo) is expressed by:

where ∂_eij/∂_xk is the strain gradient and μ_ijkl is the flexoelectric coefficient. The flexo-photovoltaic effect has also been demonstrated in the centrosymmetric materials such as SrTiO₃, TiO₂, Si, LaFeO₃ and BiVO₄.^117–119 Although FE materials have high flexoelectric coefficients, they are fragile and cannot tolerate large deformations. To address this issue, Zhang et al.¹²⁰ fabricated PZT hollow nanowires (NWs)/polydimethylsiloxane (PDMS) composites. After filling an anodic aluminum oxide (AAO) template with PZT, the AAO was then etched and replaced by flexible PDMS. For the comparison of photovoltaic performance, three different kinds of nanocomposites were investigated: PZT nanoparticles, randomly aligned PZT-NW and aligned PZT-NW embedded in a PDMS matrix. The flexophotovoltaic effect was demonstrated by upward or downward bending of the sample, which either enhanced or reduced the photocurrent. Due to the better charge transfer in the aligned PZT-NW/PDMS, it had the highest flexoelectric coefficient and best flexophotovoltaic effect. These results were consistent with the DFT calculations. The same strategy to take advantage of the flexoelectric effect has also been employed in a NaNbO₃ nanotube/epoxy composite.¹²¹

He et al.¹²² reported a TiO₂/BFO/WO₃ core–shell nanowire structure on the FTO substrate synthesized by hydrothermal and chemical liquid methods. The fabrication process is shown schematically in Fig. 8f. Considering the band structure of the three materials, the photoexcited electrons and holes in BFO were first separated by the internal electric field and then transported into the TiO₂ and WO₃ layers respectively, as shown in Fig. 8g. As previously mentioned, the non-thermalized carriers diffusion length is around tens of nanometres in ferroelectric materials. Therefore, in this device, the large surface area ensured sufficient light absorption, and the two transporting layers efficiently collected e–h pairs and reduced their recombination. Thus, for this specific structure, a high J_sc of 1.14 mA cm⁻², V_oc of 1.32 V, FF of 58% and PCE of 0.93% were obtained.

In contrast to this complex fabrication processes, self-assembled vertically aligned nanocomposites (SVANs) can be grown in a single process (Fig. 9a). This was first demonstrated for FE materials in 2004 in a perovskite BaTiO₃/spinel CoFe₂O₄ (CFO) system.¹²³ When one 0.65BaTiO₃–0.35CoFe₂O₄ ceramic target was ablated by a pulsed laser to grow a film on STO, in the resultant film, CoFe₂O₄ nanopillars were embedded in the BaTiO₃ matrix. A top-view TEM image is shown in Fig. 9b. This automatic phase separation was attributed to interfacial energy differences. In the following years, SVAN was successfully realized in many multiferroic systems including BaTiO₃–CoFe₂O₄,¹²⁴ BiFeO₃–CoFe₂O₄,¹²⁵ PbTiO₃–CoFe₂O₄ ¹²⁶ and BiFeO₃–NiFe₂O₄.¹²⁷ These composites decouple out-of-plane strain from in-plane strain and have large heterointerface areas. Thus, they are ideal platforms for investigating elastic interactions like magnetoelectric coupling.^128,129 SVANs provide new degrees of freedom in traditional metal oxide fabrication processes, as well as the integration of the functional properties of two different materials. However, their photovoltaic performance has been little researched. Given that the magnetophotovoltaic effect has been found in a CFO NPs/PZT composite,⁹⁴ the specific interface configuration of CFO-BTO SVAN should yield better performance.

Fig. 9 (a) Schematic drawing of a self-assembled vertically aligned nanocomposite on a substrate. (b) TEM top-view image of CFO-BTO SVAN; the dark regions are CFO pillars and the bright area is the BTO matrix. Reprinted with permission,¹²³ Copyright 2004, The American Association for the Advancement of Science. (c) Low-magnification annular-dark field STEM image of PTBNNO. The blue arrows indicate NiO columns. (d) High-resolution high-angle annular dark-field STEM image. Reprinted with permission,¹³⁰ Copyright 2020, American Chemical Society.

Another self-assembled VAN system was recently found in a (PbTiO₃)_x(BiNi_2/3Nb_1/3O₃)_1−x (PTBNNO) material¹³⁰ when it was deposited on a single-crystal STO substrate via PLD; two phases coexisted in the film: a nanocolumn NiO phase and a layered (Pb, Nb)-doped bismuth titanate Aurivillius phase. Compared with pure PTO, the layered structure in the SVAN (Fig. 9d) increased the resistance, which is favorable for a high open circuit voltage based on BPE theory.²⁴ In addition, the conductive NiO columns marked by blue arrows in Fig. 9c served as charge transport pathways during the photovoltaic process, which prevented severe degradation of the photocurrent. Consequently, the SVAN thin film exhibited improved photovoltaics compared to pure PbTiO₃ and Aurivillius phase films.

4.4 Polar nanoregions (PNRs)

To this point, this review has focused on the simple integration of ferroelectric materials with narrow bandgap semiconductors or noble metals to make full use of their different properties to obtain a high photovoltage and large photocurrent, but the interfaces between the two different materials must be carefully designed to ensure smooth electron transfer. Achieving both strong visible light absorption and the bulk photovoltaic effect within one material is highly desirable. However, narrow bandgap semiconductors usually rely on interfacial junctions for charge carrier separation, as built-in electric fields are inhibited by their high electron conductivity. Inspired by the concept of short-range ordered polar nano-regions (PNRs) in relaxor ferroelectrics, an electric field can be established in nanoscale clusters that arise from local inhomogeneities such as chemical or structural disorder.¹³¹ This section of the review explores methods used to introduce ferroelectric behavior into narrow bandgap semiconductors.

For the use of the PLD deposition method, many factors must be considered during the deposition of films, such as the laser fluence, pulse frequency, substrate temperature, oxygen pressure, distance between the target and substrate and the post-annealing process. Through delicately control of the deposition conditions, Chakrabartty et al.¹³² successfully deposited three kinds of films on STO substrates using the same Bi_1.4MnO₃ target. These were single-phase BiMnO₃, BiMn₂O₅, and mixed crystal phases (Fig. 10a). Compared with the single phases, the mixed phase consisting of ferroelectric BiMnO₃ and non-ferroelectric semiconductor BiMn₂O₅ had a stronger optical absorption and possessed a very high PCE of ∼4.2%. This high performance is attributed to abundant downward band-bending at grain boundaries (GBs). Through conductive AFM and KPFM, the authors confirmed that the photo-charge was mainly generated and transported in the GBs and was not related to topographic artifacts (Fig. 10b and c). The photocurrent can be further manipulated by using different numbers of bipolar voltage pulses, and considering the magnetic properties of BiMnO₃, this single-layer perovskite oxide film is expected to have multiferroic properties.

Fig. 10 (a) TEM image of the grain boundary between BMO₃ and BiMn₂O₅. Inset is an illustration of the device for photovoltaic measurement. (b) AFM morphology image. (c) Photocurrent mapping of (b). Reprinted with permission,¹³² Copyright 2018, Springer Nature. (d) HRTEM image of the mixed phase film with a green circle and red circles indicating Bi₄V₂O₁₁ and BiVO₄, respectively. At the right side are selected area electron diffraction (SAED) patterns of Bi₄V₂O₁₁ (top) and BiVO₄ (bottom). (e) Ferroelectric polarization versus electric field (P–E) hysteresis loop of films of different thicknesses. (f) I–V curve under AM 1.5G (100 mW cm⁻²) illumination. The top inset is the device configuration for characterizing PV performance. The middle inset is the reverse PV after applying opposite external electric field poling. Reprinted with permission,³¹ Copyright 2020, Royal Society of Chemistry. (g) Successive magnification of cross-sectional HRTEM image of ZCFLO. (h) I–V curves of ZCFLO films under AM 1.5G light (150 mW cm⁻²). Reprinted with permission,¹³⁶ Copyright 2023, John Wiley and Sons.

The photovoltaic effect has also been created in BVO through either strain-induced polarization¹³³ or charge gradient distribution of volatile bismuth.⁸⁴ In other research, BVO was combined with ferroelectric Bi₄V₂O₁₁ to form a heterojunction and showed high-performance PV.¹³⁴ Mai et al.³¹ found that a local-chemistry-defect could introduce ferroelectric Bi₄V₂O₁₁ nanoparticles during the synthesis of non-ferroelectric BiVO₄ (BVO) film at a high temperature of 600 °C via PLD (Fig. 10d). However, the ferroelectricity was thickness-dependent. In 30 nm thick films, polar nanoparticles were randomly distributed in the matrix, and no overall polarization was detected, but when the thickness was comparable to the particle size of ∼10 nm, the composites showed typical ferroelectricity with remanent polarization 5 μC cm⁻², as shown in Fig. 10e. This might be due to the electric boundary condition change and the depolarization field stabilization of the PNRs. In terms of the photovoltaic effect, the narrow bandgap matrix BVO strongly absorbed visible light and generated e–h pairs, and these charge carriers were separated by PNRs. Therefore, the PCE was increased 1000 times compared with that of the BVO ceramic without PNRs. Additionally, the photocurrent was switchable by external electric field poling (Fig. 10f).

Based on the ballistic theory discussed in section 2.1, the photoexcited non-thermalized carriers in noncentrosymmetric crystals lose energy and descend to the conduction band minimum over the free path l₀. The value of l₀ is estimated to be on the order of 10–100 nm. This is the reason that when the thickness of thin films is close to l₀, the ferroelectric photovoltaic will be dramatically enhanced. In other words, high photocurrent in bulk photovoltaics can only be obtained in thin films with a top electrode–FE–bottom electrode configuration, but with a much lower open circuit voltage of ∼1 V. To overcome this limitation, Gao et al.^135,136 constructed small-size (5–20 nm) highly crystalline ferroelectric grains in an amorphous thick film, as shown in Fig. 10g. In their previous work,¹³⁵ Fe³⁺–Li⁺ doped ZnO showed an anomalous photovoltaic effect in which the ferroelectricity came from Fe³⁺–Li⁺ ionic pairs. Recently, a Cu dopant was added to control the degree of crystallinity and further reduce the bandgap.¹³⁶ The Zn_0.92−xCu_x(Fe_0.04Li_0.04)O (ZCFLO) was deposited on a Si substrate via a sol–gel method on top of an LaNiO₃ seed layer. After doping, ZCFLO had a much lower bandgap of 1.7 eV compared to 3.4 eV for pure ZnO. Under these circumstances, the amorphous area strongly absorbed light and generated e–h pairs, which were then separated by the ferroelectric nanocrystals. Fig. 10h depicts that the film (x = 0.05) after poling treatment exhibits a record-breaking PCE (14.4%) with an above-bandgap photovoltage (3.75 V), photocurrent (50.3 mA W⁻¹) and FF (83.4%) in an in-plane configuration with interdigital electrode spacing at 300 μm.

4.5 Low-dimensional materials

2D semiconductor materials with reduced dimensionality can be stacked together to form functional devices such as transistors¹³⁷ and photodetectors.¹³⁸ More recently, a bulk photovoltaic effect was demonstrated in the 2D materials CuInP₂S₆,¹³⁹ MoS₂ ¹⁴⁰ and α-In₂Se₃.³⁰ Apart from the intrinsic ferroelectricity driven by ionic displacement, such as that in the monolayer group-IV monochalcogenides MX (M = Ge, Sn; X = S, Se),¹⁴¹ SnTe,¹⁴² GeS nanosheets,¹⁴³ transition-metal dichalcogenide (MoTe₂ ¹⁴⁴), CuInP₂S₆ ^139,145 and quasi-2D Sn₂P₂S₆,¹⁴⁶ the charge redistribution at the interface also gives rise to a ferroelectricity-like behaviour in bilayer graphene/hexagonal boron nitride (h-BN) heterostructure,¹⁴⁷ twisted bilayer h-BN^148,149 and R-stacking MoS₂.¹⁴⁰ Given that 2D materials have small effective masses, high carrier mobilities and narrow bandgaps, they are expected to have high photovoltaic performance with both strong light absorption and low recombination rates.¹⁵⁰ Although at this stage, these materials cannot be scaled up for energy conversion, the various structure designs extend the realm of the BPE beyond traditional fabricated films and are suitable for minimalized photoelectronic applications.

Zhang et al.¹⁵¹ found that the polar crystal structure is the key factor in achieving high BPE, and therefore, in centrosymmetric bilayer tungsten disulfide (WS₂), there is no PV effect, while in WS₂, which is non-centrosymmetric but has a non-polar monolayer, the photocurrent is too low to be identified. WS₂ multi-wall nanotubes, in which mirror and rotational symmetries are broken, have a non-centrosymmetric and non-polar structure, as illustrated in Fig. 11a, and can even convert 60% of solar energy (photo energies above 1.45 eV based on the indirect bandgap) to electric power. As shown in Fig. 11b, the crossover of the power dependence photocurrent rules out a flexophotovoltaic effect or Schottky barrier contribution and is ascribed to the shift current origin. Later, Sun⁷² demonstrated the utilization of these multiwall WS₂ nanotubes in a photovoltaic random-access memory (PV-RAM) device, which was analogous to resistive random access memory (RRAM).¹⁵² As shown in Fig. 11c, in a single nanotube, the photovoltaic V_oc and J_sc were programmable by an external poling process. Based on this finding, the authors constructed an artificial vision system on randomly dispersed WS₂ nanotubes with multiple functions of sense, memory, computing, and power. Finally, they proposed a van der Waals sliding ferroelectric effect to explain the observed phenomenon, which meant that the electric polarization along the nanotube was created and modulated by the relative deformation of each monolayer, as shown in Fig. 11d.

Fig. 11 (a) Schematic crystal structure of bilayer and monolayer WS₂ and its multi-wall nanotube. The black dots in the bilayer mark the center of the inversion symmetry. The yellow planes and black line at the cross-section in the monolayer indicate the mirror planes and rotation axis. (b) Incident-light-power-dependence of the photocurrent. Reprinted with permission,¹⁵¹ Copyright 2019, Springer Nature. (c) Hysteresis behavior of photovoltaic J_sc and V_oc upon external voltage and current sweep. (d) In situ TEM image of layered structure in nanotubes before and after bias. Reprinted with permission,⁷² Copyright 2022, Springer Nature. (e) Crystal structure of monolayer WSe₂, BP and heterointerface of WSe₂/BP. The green lines and black dots represent the mirror plane and rotation axes, respectively. (f) Photon-energy-dependence of the photocurrent. Reprinted with permission,¹⁵³ Copyright 2021, The American Association for the Advancement of Science. (g) Optical microscope image (left) and scanning photocurrent microscopy image (right). Reprinted with permission,¹⁵⁵ Copyright 2019, Springer Nature. (h) Cross-sectional schematics of edge-embedded structures and their generation of opposite direction photocurrent. Reprinted with permission,¹⁵⁶ Copyright 2023, Springer Nature.

Akamatsu et al.¹⁵³ reduced the interfacial symmetry by stacking WS₂ and BP together. An electronic polarization was thus created parallel to the bilayer mirror plane, due to their different rotational symmetries as shown in Fig. 11e. As a result, a spontaneous photocurrent was observed along this polar direction, but was absent in the perpendicular direction. As shown in Fig. 11f, the photon energy dependence of the photocurrent indicates exciton resonance of WS₂ at 1.65 eV and 2.05 eV can enhance the photovoltaic response. This phenomenon and the polarized-light-dependent photocurrent are explained by band hybridization calculations and shift current theory. Recently, an ultra-fast BPE response was demonstrated in a MoS₂/BP bilayer taking advantage of the in-plane and out-of-plane dual-polarization.¹⁵⁴

In 2019, Wang et al. found a robust photocurrent at certain edges of Weyl semimetal (WTe₂) without symmetry constraints, as shown in Fig. 11g.¹⁵⁵ Later, Liang reported photocurrent in another interesting geometry called an edge-embedded structure.¹⁵⁶ The quasi-1D structure is marked by dashed lines in Fig. 11h. When the two layers were stacked together, the inversion symmetry at the bottom layer edge was broken by vdW coupling. Therefore, this structure showed a typical BPE photocurrent and it was polarized-light-dependent. Interestingly, the current had reversed polarity on opposite side edges. Through first-principles calculations, the authors found that the photo-carriers tended to accumulate at the bottom edge and then were separated by BPE. This specific structure for BPE was proved to be universal in many semiconductor homo- or heterostructures such as ReS₂/ReS₂, MoS₂/MoS₂ and WS₂/ReS₂.

5 Summary

In Table 1, we summarize the photovoltaic performances of all the previously mentioned materials, including their I_sc, V_oc and PCE values. In multilayers, band-structure engineering has been proven to be an effective strategy to align the internal electric field and achieve high photovoltaic performance. However, considering the large mismatch between the inserted metal oxide and ferroelectric perovskite layers, these devices are generally fabricated via sol–gel methods and contain large amounts of defects, which limit the energy conversion efficiency to below 1%. In gradient-doped films, charge gradient distribution can introduce an additional internal electric field for carrier separation. The pulsed laser deposition technique can produce high-quality epitaxial films. In particular, superlattice structure allows researchers to precisely control the ordering/disordering at the atomic level. By modifying the electronic structure, a narrow bandgap of ∼1.5 eV was experimentally achieved and theoretically predicted in many other ferroelectric thin films. For nanoparticle composites, integrating low-bandgap semiconductors into a ferroelectric matrix can slightly reduce the bandgap. In many reports, the localized surface plasmonic resonance effect of noble metal particles has been utilized for enhancing light absorption at specific visible-range wavelengths. In these 0–3 composites, the particle size has a great impact on photovoltaic performance.

Table 1 Summary of the photovoltaic performance of the above-mentioned ferroelectric devices

Structure	Fabrication method	Thickness or length (nm)	Remnant polarization Pr (μC cm^−2)	Light source		J sc (mA cm^−2)	V oc (V)	PCE (%)	Ref.
				Wavelength (nm)	Intensity (mW cm^−2)
a PCE is calculated based on the assumption that the fill factor is 25% for straight I–V curves.
Transport layers
ITO/PZT/Cu2O/Pt	Sol–gel	270	25	AM 1.5G	100	4.8	0.42	0.57	67
ITO/ZnO/BFO/Pt	Radiofrequency (RF) sputtering	150	30	435	22.3	0.34	0.2	0.33	70
Au/PZT/ZnO/ITO	Sol–gel	130	20	200–400	60	0.0424	0.681	0.012	68
Ag/ZnO/BFTO/NSTO	PLD	200	22	405	200	2.2	0.15	0.045	71
Gradient films
Ag/BCFO/FTO	Sol–gel	343	7	405	200	0.647	0.36	0.029^a	77
Au/BLCFO-g/FTO	Sol–gel	270	50	405	50	1.73	—	—	78
Au/BSFNO-g/FTO	Sol–gel	210	18	405	50	0.08	0.49	0.0194^a	79
ITO/BNO-N2/BNO-O2/LNO/Y2O3/Si	PED	300	10	AM 1.5G	100	0.2	0.85	0.0425^a	82
Au/BVO/Pt	Sol–gel	200	—	450		1.63	0.38	0.4	84
Superlattices
ITO/BFCO (2 Hz, 8 Hz, 14 Hz)/NSTO	PLD	285	35	AM 1.5G	100	20.6	0.84	8.1	39
ITO/SBC5555/NSTO	PLD	200	—	405	—	0.01103	0.058	—	91
FE matrix-nanoparticle composites
Pt/PZT/Ag-8s/Pt	Radiofrequency (RF) sputtering	400	20	AM 1.5G	100	0.47	0.69	0.195	92
Pt/CFO-PZT/Pt	Sol–gel	200	2.4	UV	60	0.00384	0.164	0.00026^a	94
Pt/BFO/BWO/NSTO	PLD	100	0.25	405	117	1.2	0.3	0.0769^a	98
Pt/Sb2Se3/BWO/NSTO	PLD + vapor transport deposition	140	0.0005	405	117	3.236	0.205	0.1417^a	99
SPR
Pt/BABAB/Pt	Sol–gel	500–550	95 (83 K)	450	1	0.001	0.31	0.01	108
ITO/AuNPs/BLFO/FTO	Sol–gel	70 + 476	17	AM 1.5G	100	0.02	0.4	0.002^a	109
Ag/ZnO/AuNPs/BFO/FTO	Sol–gel	80 + 50 + 340	—	405	60	0.1719	0.34	0.0244^a	112
				360	26	0.0908	0.66	0.0576^a
Ag/PZT/ITO	Sol–gel	300	21.15	AM 1.5G	100	1.32	0.82	0.42	113
Vertically aligned structures
ITO/ZnO nanowires/BLFO/FTO	Sol–gel	270 + 6000	18	405	100	0.419	0.411	0.043^a	116
Au/PZT NW/PDMS/ITO	Sol–gel	5 × 10^4		360	20	2.26 × 10^−7	2.65	7.49 × 10^−7^a	120
ITO/NN-NT/epoxy/Au	Hydrothermal	4 × 10^5–5 × 10^5	4.4	405	—	3.92 × 10^−5	0.22	—	121
Ag/WO3/BFO/TiO2/FTO	Hydrothermal + sol–gel	3000	—	AM 1.5G	100	0.1	1.4	0.069	122
ITO/65PTBNNO/NSTO	PLD	225	—	AM 1.5G	100	0.796	0.67	0.131	130
Polar nanoregions
ITO/Bi–Mn–O/NSTO	PLD	110 ± 10	1.85 (BMO)	AM 1.5G	100	7.03	1.48	4.2	132
ITO/BVO/LNO/YOT/Si	PLD	10	5	AM 1.5G	100	0.46	2.4	0.6	31
Au/Zn0.88(Fe0.06Li0.06)O/LNO/n^+-Si	Sol–gel	3 × 10^5 (in-plane)	1	400–830	150	5.115	5.06	10.3	135
Au/Zn0.87Cu0.05(Fe0.04Li0.04)O/LNO/n^+-Si	Sol–gel	3 × 10^5 (in-plane)	2	AM 1.5G	150	7.545	3.75	14.4	136
2D materials
Graphene/CIPS/graphene	Mechanical exfoliation (ME) and transfer	10	—	405	300	0.8	1.65	3.67	139
Au/graphene/α-In2Se3/graphene/Au	ME and transfer	18	—	520	1 × 10^5	0.032	0.32	0.00256	30
Au/WS2 nanotubes (device2)/Au	Chemical vapor transport and ME	5 × 10^3 (D: 90 nm)	—	632.8	1.39 × 10^7	8.56 × 10^4	0.3	0.0462^a	151

---

Vertically aligned structures are suitable for fabricating flexible electronic devices by combining FE materials with polymers. In these structures, the piezo- or flexoelectric effect coupled with the photovoltaic makes the photocurrent tunable through strain and electric fields. In the past, complex nanostructures have usually required delicate control of the fabrication process. However, emerging self-assembled vertically aligned nanocomposites and polar nanoregions can be synthesized in a single step through either a physical vapor or a wet chemical method, making them facile and practical for high-performance FE-PV applications. Self-assembled vertically aligned nanocomposites such as those consisting of ferroelectric and magnetoelectric phases have attracted more attention for their ferric properties. In terms of PV applications, the conductive columns can significantly reduce resistance and improve photovoltaic performance. Compared with the above structures, the introduction of local polar nanoregions into narrow bandgap materials is the most promising method to simultaneously achieve high open-circuit voltage and large short-circuit current with a PEC exceeding 1%. More recently, ferroelectricity in low-dimensional materials, particularly two-dimensional van der Waals materials, has been intensively studied due to their various stacking geometries and interfacial configurations. Given that BPE is typically associated with the crystal structure, in low-dimensional materials, the ferroelectricity originates from the charge distribution. This has led to the discovery of many intriguing phenomena and further advancement the development of BPE theory.

6. Perspectives for future research

Improving the collection efficiency of hot photocarriers and enhancing light absorption remain critical challenges in ferroelectric photovoltaics. Addressing these issues requires reducing the scale of ferroelectric materials, providing appropriate charge transport channels and exploring narrow bandgap materials. In narrow bandgap semiconductors, ferroelectric-like behavior has been successfully demonstrated through the formation of polar nanoregions (PNRs). In these systems, light is absorbed by the matrix, and the photo-excited electron–hole pairs are separated by PNRs and then transported through conducive regions. Therefore, in the future, two research directions are expected to receive more attention in efforts to pursue ferroelectric photovoltaics with high power conversion efficiency. The first is low-dimensionalization and integration of ferroelectric materials to maximize the bulk photovoltaic effect. The second involves introducing ferroelectricity into narrow bandgap semiconductors through local chemical deviation, defects or strain engineering.

Data availability

All data used in this review were obtained from publicly available literature, as cited in the references. No new data were generated or analysed.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors thank the Australian Research Council (ARC) for funding support through the discovery program (DP200100159, DP230100462, DP250101852, DE240100032, FL210100017), and Linkage program (IH24010009).

References

1. A. M. Glass, D. von der Linde and T. J. Negran, Appl. Phys. Lett., 1974, 25, 233–235.
2. W. T. H. Koch, R. Munser, W. Ruppel and P. Wurfel, Solid State Commun., 1975, 17, 847–850.
3. Y. Yuan, Z. Xiao, B. Yang and J. Huang, J. Mater. Chem. A, 2014, 2, 6027–6041.
4. P. Lopez-Varo, L. Bertoluzzi, J. Bisquert, M. Alexe, M. Coll, J. Huang, J. A. Jimenez-Tejada, T. Kirchartz, R. Nechache, F. Rosei and Y. Yuan, Phys. Rep., 2016, 653, 1–40.
5. Z. Dai and A. M. Rappe, Chem. Phys. Rev., 2023, 4, 011303.
6. S. Y. Yang, J. Seidel, S. J. Byrnes, P. Shafer, C.-H. Yang, M. D. Rossell, P. Yu, Y.-H. Chu, J. F. Scott, J. W. Ager, L. W. Martin and R. Ramesh, Nat. Nanotechnol., 2010, 5, 143–147.
7. V. M. Fridkin, Photoferroelectrics, Springer, Berlin, 1979.
8. G. Conibeer, Mater. Today, 2007, 10, 42–50.
9. W. Shockley and H. J. Queisser, J. Appl. Phys., 1961, 32, 510–519.
10. C. Paillard, X. Bai, I. C. Infante, M. Guennou, G. Geneste, M. Alexe, J. Kreisel and B. Dkhil, Adv. Mater., 2016, 28, 5153–5168.
11. S. Pal, N. V. Sarath, K. S. Priya and P. Murugavel, J. Phys. D: Appl. Phys., 2022, 55, 283001.
12. H. Huang and J. F. Scott, Ferroelectric Materials for Energy Applications, John Wiley & Sons, Weinheim, 2018.
13. J. Kreisel, M. Alexe and P. A. Thomas, Nat. Mater., 2012, 11, 260–260.
14. S. Cheng, Z. Fan, J. Rao, L. Hong, Q. Huang, R. Tao, Z. Hou, M. Qin, M. Zeng, X. Lu, G. Zhou, G. Yuan, X. Gao and J. M. Liu, iScience, 2020, 23, 101874.
15. Y. Wang, F. Wang, Z. Wang, J. Wang, J. Yang, Y. Yao, N. Li, M. G. Sendeku, X. Zhan, C. Shan and J. He, Nano Res., 2021, 14, 4328–4335.
16. R. Guo, L. You, Y. Zhou, Z. S. Lim, X. Zou, L. Chen, R. Ramesh and J. Wang, Nat. Commun., 2013, 4, 1990.
17. D. Li, D. Zheng, C. Jin, W. Zheng and H. Bai, ACS Appl. Mater. Interfaces, 2018, 10, 19836–19843.
18. Y. Peng, X. Liu, Z. Sun, C. Ji, L. Li, Z. Wu, S. Wang, Y. Yao, M. Hong and J. Luo, Angew. Chem., Int. Ed., 2020, 59, 3933–3937.
19. W. J. Hu, Z. Wang, W. Yu and T. Wu, Nat. Commun., 2016, 7, 10808.
20. X. Long, H. Tan, F. Sánchez, I. Fina and J. Fontcuberta, Nat. Commun., 2021, 12, 382.
21. K. Zhao, B. Ouyang and Y. Yang, iScience, 2018, 3, 208–216.
22. K. Zhao, B. Ouyang, C. R. Bowen and Y. Yang, Nano Energy, 2020, 77, 105152.
23. Y. Liu, S. Wang, Z. Chen and L. Xiao, Sci. China Mater., 2016, 59, 851–866.
24. A. Bhatnagar, A. R. Chaudhuri, Y. H. Kim, D. Hesse and M. Alexe, Nat. Commun., 2013, 4, 2835.
25. J. E. Spanier, V. M. Fridkin, A. M. Rappe, A. R. Akbashev, A. Polemi, Y. Qi, Z. Gu, S. M. Young, C. J. Hawley, D. Imbrenda, G. Xiao, A. L. Bennett-Jackson and C. L. Johnson, Nat. Photonics, 2016, 10, 611–616.
26. L. Kola, D. Murali, S. Pal, B. R. K. Nanda and P. Murugavel, Appl. Phys. Lett., 2019, 114, 183901.
27. D. S. Knoche, M. Steimecke, Y. Yun, L. Mühlenbein and A. Bhatnagar, Nat. Commun., 2021, 12, 282.
28. H. Matsuo, Y. Kitanaka, R. Inoue, Y. Noguchi, M. Miyayama, T. Kiguchi and T. J. Konno, Phys. Rev. B, 2016, 94, 214111.
29. Z. Tan, L. Hong, Z. Fan, J. Tian, L. Zhang, Y. Jiang, Z. Hou, D. Chen, M. Qin, M. Zeng, J. Gao, X. Lu, G. Zhou, X. Gao and J.-M. Liu, NPG Asia Mater., 2019, 11, 20.
30. S. M. Nahid, S. Nam and A. M. van der Zande, ACS Nano, 2024, 18, 14198–14206.
31. H. Mai, T. Lu, Q. Sun, R. G. Elliman, F. Kremer, T. Duong, K. Catchpole, Q. Li, Z. Yi, T. J. Frankcombe and Y. Liu, Mater. Horiz., 2020, 7, 898–904.
32. B. S. I. Sturman and V. M. Fridkin, Photovoltaic and Photorefractive Effects in Noncentrosymmetric Materials, Gordon and Breach, Philadelphia, 1992.
33. S. M. Young and A. M. Rappe, Phys. Rev. Lett., 2012, 109, 116601.
34. A. Zenkevich, Y. Matveyev, K. Maksimova, R. Gaynutdinov, A. Tolstikhina and V. Fridkin, Phys. Rev. B:Condens. Matter Mater. Phys., 2014, 90, 161409.
35. V. I. Belinicher, E. L. Ivchenko and B. I. Sturman, Sov. Phys. – JET, 1982, 56, 359–366.
36. A. M. Burger, R. Agarwal, A. Aprelev, E. Schruba, A. Gutierrez-Perez, V. M. Fridkin and J. E. Spanier, Sci. Adv., 2019, 5, eaau5588.
37. Z. Gu, D. Imbrenda, A. L. Bennett-Jackson, M. Falmbigl, A. Podpirka, T. C. Parker, D. Shreiber, M. P. Ivill, V. M. Fridkin and J. E. Spanier, Phys. Rev. Lett., 2017, 118, 096601.
38. R. Fei, L. Z. Tan and A. M. Rappe, Phys. Rev. B, 2020, 101, 045104.
39. R. Nechache, C. Harnagea, S. Li, L. Cardenas, W. Huang, J. Chakrabartty and F. Rosei, Nat. Photonics, 2015, 9, 61–67.
40. X. Liu, J. Tu, Y.-W. Fang, G. Xi, H. Li, R. Wu, X. Liu, D. Lu, J. He, J. Zhang, J. Tian and L. Zhang, J. Am. Chem. Soc., 2024, 146, 13934–13948.
41. S. Das, S. Ghara, P. Mahadevan, A. Sundaresan, J. Gopalakrishnan and D. D. Sarma, ACS Energy Lett., 2018, 3, 1176–1182.
42. H. Matsuo, Y. Noguchi and M. Miyayama, Nat. Commun., 2017, 8, 207.
43. L. Wu, A. Podpirka, J. E. Spanier and P. K. Davies, Chem. Mater., 2019, 31, 4184–4194.
44. S. Y. Yang, L. W. Martin, S. J. Byrnes, T. E. Conry, S. R. Basu, D. Paran, L. Reichertz, J. Ihlefeld, C. Adamo, A. Melville, Y.-H. Chu, C.-H. Yang, J. L. Musfeldt, D. G. Schlom, J. W. Ager and R. Ramesh, Appl. Phys. Lett., 2009, 95, 062909.
45. T. Choi, S. Lee, Y. J. Choi, V. Kiryukhin and S.-W. Cheong, Science, 2009, 324, 63–66.
46. L. Pintilie, I. Vrejoiu, G. Le Rhun and M. Alexe, J. Appl. Phys., 2007, 101, 064109.
47. I. Grinberg, D. V. West, M. Torres, G. Gou, D. M. Stein, L. Wu, G. Chen, E. M. Gallo, A. R. Akbashev, P. K. Davies, J. E. Spanier and A. M. Rappe, Nature, 2013, 503, 509–512.
48. G. Zhang, H. Wu, G. Li, Q. Huang, C. Yang, F. Huang, F. Liao and J. Lin, Sci. Rep., 2013, 3, 1265.
49. G. Gou, N. Charles, J. Shi and J. M. Rondinelli, Inorg. Chem., 2017, 56, 11854–11861.
50. G. Zhang, J. Hou, M. Zhu, G. Huang, D. Li, Y. Fang and T. Zeng, J. Mater. Chem. C, 2020, 8, 16234–16240.
51. H. Han, S. Song, J. H. Lee, K. J. Kim, G.-W. Kim, T. Park and H. M. Jang, Chem. Mater., 2015, 27, 7425–7432.
52. H. Han, D. Kim, K. Chu, J. Park, S. Y. Nam, S. Heo, C. H. Yang and H. M. Jang, ACS Appl. Mater. Interfaces, 2018, 10, 1846–1853.
53. H. Han, D. Kim, S. Chae, J. Park, S. Y. Nam, M. Choi, K. Yong, H. J. Kim, J. Son and H. M. Jang, Nanoscale, 2018, 10, 13261–13269.
54. R. Agarwal, Y. Sharma, S. Hong and R. S. Katiyar, J. Phys. D: Appl. Phys., 2018, 51, 275303.
55. F. Wang, I. Grinberg and A. M. Rappe, Appl. Phys. Lett., 2014, 104, 152903.
56. L. You, F. Zheng, L. Fang, Y. Zhou, L. Z. Tan, Z. Zhang, G. Ma, D. Schmidt, A. Rusydi, L. Wang, L. Chang, A. M. Rappe and J. Wang, Sci. Adv., 2018, eaat3438.
57. C. Lin, Z. Zhang, Z. Dai, M. Wu, S. Liu, J. Chen, C. Hua, Y. Lu, F. Zhang, H. Lou, H. Dong, Q. Zeng, J. Ma, X. Pi, D. Zhou, Y. Wu, H. Tian, A. M. Rappe, Z. Ren and G. Han, Nat. Commun., 2023, 14, 2341.
58. Y. Sun, Z. Zhang, X. Fang, Z. Huang, D. Zhou, H. Tian, G. Han and Z. Ren, APL Mater., 2023, 11, 091236.
59. M. Qin, K. Yao and Y. C. Liang, Appl. Phys. Lett., 2009, 95, 022912.
60. J. Zhang, X. Su, M. Shen, Z. Dai, L. Zhang, X. He, W. Cheng, M. Cao and G. Zou, Sci. Rep., 2013, 3, 2109.
61. H. Mai, T. Lu, Q. Li, Q. Sun, K. Vu, H. Chen, G. Wang, M. G. Humphrey, F. Kremer, L. Li, R. L. Withers and Y. Liu, ACS Appl. Mater. Interfaces, 2018, 10, 29786–29794.
62. M. Alexe and D. Hesse, Nat. Commun., 2011, 2, 256.
63. L. Pintilie, C. Dragoi and I. Pintilie, J. Appl. Phys., 2011, 110, 044105.
64. A. Abdelsamie, L. You, L. Wang, S. Li, M. Gu and J. Wang, Phys. Rev. Appl., 2022, 17, 024047.
65. L. Pintilie and M. Alexe, J. Appl. Phys., 2005, 98, 124103.
66. L. Pintilie, I. Vrejoiu, D. Hesse, G. Lerhun and M. Alexe, Phys. Rev. B:Condens. Matter Mater. Phys., 2007, 75, 104103.
67. D. Cao, C. Wang, F. Zheng, W. Dong, L. Fang and M. Shen, Nano Lett., 2012, 12, 2803–2809.
68. D. Pan, G. Bi, G. Chen, H. Zhang, J. Liu, G. Wang and J. Wan, Sci. Rep., 2016, 6, 22948.
69. P. Tiwana, P. Docampo, M. B. Johnston, H. J. Snaith and L. M. Herz, ACS Nano, 2011, 5, 5158–5166.
70. Z. Fan, K. Yao and J. Wang, Appl. Phys. Lett., 2014, 105, 162903.
71. J.-M. Yan, K. Wang, Z.-X. Xu, J.-S. Ying, T.-W. Chen, G.-L. Yuan, T. Zhang, H.-W. Zheng, Y. Chai and R.-K. Zheng, J. Mater. Chem. C, 2021, 9, 3287–3294.
72. Y. Sun, S. Xu, Z. Xu, J. Tian, M. Bai, Z. Qi, Y. Niu, H. H. Aung, X. Xiong, J. Han, C. Lu, J. Yin, S. Wang, Q. Chen, R. Tenne, A. Zak and Y. Guo, Nat. Commun., 2022, 13, 5391.
73. H. Renuka, P. Joshna, W. A. Wani, B. H. Venkataraman, K. Ramaswamy and S. Kundu, Mater. Sci. Semicond. Process., 2020, 109, 104937.
74. H. T. Yi, T. Choi, S. G. Choi, Y. S. Oh and S. W. Cheong, Adv. Mater., 2011, 23, 3403–3407.
75. T. Yang, J. Wei, Y. Guo, Z. Lv, Z. Xu and Z. Cheng, ACS Appl. Mater. Interfaces, 2019, 11, 23372–23381.
76. J. Schiemer, R. Withers, L. Norén, Y. Liu, L. Bourgeois and G. Stewart, Chem. Mater., 2009, 21, 4223–4232.
77. Y. Zhang, H. Zheng, X. Wang, H. Li, Y. Wu, Y. Zhang, H. Su and G. Yuan, Ceram. Int., 2020, 46, 10083–10088.
78. Z. Sun, J. Wei, T. Yang, Y. Li, Z. Liu, G. Chen, T. Wang, H. Sun and Z. Cheng, ACS Appl. Mater. Interfaces, 2021, 13, 49850–49859.
79. Z. Sun, J. Wei, Y. Li, Z. Liu, M. Xiahou, G. Chen, L. Zhao and Z. Cheng, Inorg. Chem. Front., 2023, 10, 1315–1327.
80. M. Yang and M. Alexe, Nature, 2019, 570, 310–311.
81. W. Dong, D. Cortie, T. Lu, Q. Sun, N. Narayanan, W. Hu, L. Jacob, Q. Li, D. Yu, H. Chen, A. Chen, X. Wei, G. Wang, M. G. Humphrey, T. J. Frankcombe and Y. Liu, Mater. Horiz., 2019, 6, 1717–1725.
82. H. Mai, T. Lu, Q. Sun, J. Langley, N. Cox, F. Kremer, T. Duong, K. Catchpole, H. Chen, Z. Yi, T. J. Frankcombe and Y. Liu, J. Mater. Chem. A, 2021, 9, 13182–13191.
83. A. Malathi, J. Madhavan, M. Ashokkumar and P. Arunachalam, Appl. Catal., A, 2018, 555, 47–74.
84. L. Shi, Z. Li, W. Zhao, Z. Wang, Y. Peng, C. Gao, W. Hua, J. Wang, X. Yang, W. Fei and Y. Zhao, Nano Energy, 2023, 114, 108594.
85. D. Meng, Y. Xiao, H. He, Y. Liao, H. Zhang, J. Zhai, Z. Chen, L. W. Martin and F. Bai, J. Am. Ceram. Soc., 2019, 102, 5234–5242.
86. K. Ueda, H. Tabata and T. Kawai, Science, 1998, 280, 1064–1066.
87. S. Das, Y. L. Tang, Z. Hong, M. A. P. Gonçalves, M. R. Mccarter, C. Klewe, K. X. Nguyen, F. Gómez-Ortiz, P. Shafer, E. Arenholz, V. A. Stoica, S.-L. Hsu, B. Wang, C. Ophus, J. F. Liu, C. T. Nelson, S. Saremi, B. Prasad, A. B. Mei, D. G. Schlom, J. Íñiguez, P. García-Fernández, D. A. Muller, L. Q. Chen, J. Junquera, L. W. Martin and R. Ramesh, Nature, 2019, 568, 368–372.
88. D. Kim, H. Han, J. H. Lee, J. W. Choi, J. C. Grossman, H. M. Jang and D. Kim, Proc. Natl. Acad. Sci. U. S. A., 2018, 115, 6566–6571.
89. S. Zhang, H. Y. Xiao, S. M. Peng, G. X. Yang, Z. J. Liu, X. T. Zu, S. Li, D. J. Singh, L. W. Martin and L. Qiao, Phys. Rev. Appl., 2018, 10, 044004.
90. F. Wang, S. M. Young, F. Zheng, I. Grinberg and A. M. Rappe, Nat. Commun., 2016, 7, 10419.
91. Y. Yun, L. Mühlenbein, D. S. Knoche, A. Lotnyk and A. Bhatnagar, Sci. Adv., 2021, 7, eabe4206.
92. X. Yang, X. Su, M. Shen, F. Zheng, Y. Xin, L. Zhang, M. Hua, Y. Chen and V. G. Harris, Adv. Mater., 2012, 24, 1202–1208.
93. D. Sharma and N. Khare, Appl. Phys. Lett., 2014, 105, 032404.
94. D.-F. Pan, G.-Y. Chen, G.-F. Bi, H. Zhang, J.-M. Liu, G.-H. Wang and J.-G. Wan, Appl. Phys. Lett., 2016, 108, 222902.
95. M. Maksutoğlu, S. Kazan, N. I. Khalitov, V. I. Nuzhdin, R. I. Khaibullin, L. R. Tagirov, V. V. Roddatis, K. E. Prikhodko and F. A. Mikailzade, J. Magn. Magn. Mater., 2015, 373, 103–107.
96. M. Algueró, J. Ricote, M. Torres, H. Amorín, A. Alberca, O. Iglesias-Freire, N. Nemes, S. Holgado, M. Cervera, J. Piqueras, A. Asenjo and M. García-Hernández, ACS Appl. Mater. Interfaces, 2014, 6, 1909–1915.
97. B. Leedahl, D. A. Zatsepin, D. W. Boukhvalov, R. J. Green, J. A. McLeod, S. S. Kim, E. Z. Kurmaev, I. S. Zhidkov, N. V. Gavrilov, S. O. Cholakh and A. Moewes, J. Appl. Phys., 2014, 115, 053711.
98. S. Sun, D. Yang, J. Shen, R. Chen, H. Huang, R. Ti and X. Yin, Appl. Phys. Lett., 2024, 124, 202902.
99. S. Sun, D. Yang, W. Han, J. Shen, J. Zhang, H. Huang, G. Wang, W. Zi, X. Li, H. Qi, X. Yin and Y. Lu, Chem. Eng. J., 2024, 486, 150235.
100. J. Jana, M. Ganguly and T. Pal, RSC Adv., 2016, 6, 86174–86211.
101. J. Li, Z. Lou and B. Li, Chin. Chem. Lett., 2022, 33, 1154–1168.
102. W. Hou and S. B. Cronin, Adv. Funct. Mater., 2013, 23, 1612–1619.
103. Y.-L. Huang, W. S. Chang, C. N. Van, H.-J. Liu, K.-A. Tsai, J.-W. Chen, H.-H. Kuo, W.-Y. Tzeng, Y.-C. Chen, C.-L. Wu, C.-W. Luo, Y.-J. Hsu and Y.-H. Chu, Nanoscale, 2016, 8, 15795–15801.
104. J. Zhao, S. Xue, R. Ji, B. Li and J. Li, Chem. Soc. Rev., 2021, 50, 12070–12097.
105. Z. Wang, D. Cao, L. Wen, R. Xu, M. Obergfell, Y. Mi, Z. Zhan, N. Nasori, J. Demsar and Y. Lei, Nat. Commun., 2016, 7, 10348.
106. P. Joshna, S. R. Gollu, P. M. P. Raj, B. P. Rao, P. Sahatiya and S. Kundu, Nanotechnology, 2019, 30, 365201.
107. S. Kim, J. M. Kim, J. E. Park and J. M. Nam, Adv. Mater., 2018, 30, 1704528.
108. R. Maruyama, W. Sakamoto, I. Yuitoo, T. Takeuchi, K. Hayashi and T. Yogo, Jpn. J. Appl. Phys., 2016, 55, 10TA14.
109. F. Z. Li, H. W. Zheng, M. S. Zhu, X. A. Zhang, G. L. Yuan, Z. S. Xie, X. H. Li, G. T. Yue and W. F. Zhang, J. Mater. Chem. C, 2017, 5, 10615–10623.
110. C. Noguez, J. Phys. Chem. C, 2007, 111, 3806–3819.
111. R. Su, D. Zhang, M. Wu, F.-t. Li, Y. Liu, Z. Wang, C. Xu, X. Lou, Q. Yu and Y. Yang, J. Alloys Compd., 2019, 785, 584–589.
112. Y. Zhang, H. Su, H. Li, Z. Xie, Y. Zhang, Y. Zhou, L. Yang, H. Lu, G. Yuan and H. Zheng, Nano Energy, 2021, 85, 105968.
113. F. Zheng, P. Zhang, X. Wang, W. Huang, J. Zhang, M. Shen, W. Dong, L. Fang, Y. Bai, X. Shen, H. Sun and J. Hao, Nanoscale, 2014, 6, 2915–2921.
114. L. Fei, Y. Hu, X. Li, R. Song, L. Sun, H. Huang, H. Gu, H. L. Chan and Y. Wang, ACS Appl. Mater. Interfaces, 2015, 7, 3665–3670.
115. A. Koka, Z. Zhou and H. A. Sodano, Energy Environ. Sci., 2014, 7, 288–296.
116. Y. Zhang, L. Yang, Y. Zhang, Z. Ding, M. Wu, Y. Zhou, C. Diao, H. Zheng, X. Wang and Z. L. Wang, ACS Nano, 2020, 14, 10723–10732.
117. M.-M. Yang, D. J. Kim and M. Alexe, Science, 2018, 360, 904–907.
118. M. Wu, Z. Jiang, X. Lou, F. Zhang, D. Song, S. Ning, M. Guo, S. J. Pennycook, J. Y. Dai and Z. Wen, Nano Lett., 2021, 21, 2946–2952.
119. P.-W. Shao, M.-C. Lin, Q. Zhuang, J. Huang, S. Liu, H.-W. Chen, H.-L. Liu, Y.-J. Lu, Y.-J. Hsu, J.-M. Wu, Y.-C. Chen and Y.-H. Chu, Appl. Catal., B, 2022, 312, 121367.
120. B. Zhang, D. Tan, X. Cao, J. Tian, Y. Wang, J. Zhang, Z. Wang and K. Ren, ACS Nano, 2022, 16, 7834–7847.
121. F. Yu, J. Tian, F. Jiang, Y. Liu, C. Li, C. Wang, Z. L. Wang and K. Ren, Nano Res., 2023, 16, 11914–11924.
122. J. He, J. Li, Y. He, Y. Ren, S. Li, S. Xing, R. Gao, W. Cai and C. Fu, J. Alloys Compd., 2023, 960, 170692.
123. H. Zheng, J. Wang, S. E. Lofland, Z. Ma, L. Mohaddes-Ardabili, T. Zhao, L. Salamanca-Riba, S. R. Shinde, S. B. Ogale, F. Bai, D. Viehland, Y. Jia, D. G. Schlom, M. Wuttig, A. Roytburd and R. Ramesh, Science, 2004, 303, 661–663.
124. R. V. Chopdekar and Y. Suzuki, Appl. Phys. Lett., 2006, 89, 182506.
125. H. Zheng, F. Straub, Q. Zhan, P.-L. Yang, W.-K. Hsieh, F. Zavaliche, Y.-H. Chu, U. Dahmen and R. Ramesh, Adv. Mater., 2006, 18, 2747–2752.
126. I. Levin, J. Li, J. Slutsker and A. L. Roytburd, Adv. Mater., 2006, 18, 2044–2047.
127. Q. Zhan, R. Yu, S. P. Crane, H. Zheng, C. Kisielowski and R. Ramesh, Appl. Phys. Lett., 2006, 89, 172902.
128. A. Chen, Z. Bi, Q. Jia, J. L. MacManus-Driscoll and H. Wang, Acta Mater., 2013, 61, 2783–2792.
129. W. Zhang, A. Chen, Z. Bi, Q. Jia, J. L. MacManus-Driscoll and H. Wang, Curr. Opin. Solid State Mater. Sci., 2014, 18, 6–18.
130. L. Mühlenbein, C. B. Singh, A. Lotnyk, C. Himcinschi, Y. Yun, N. Ramakrishnegowda, D. S. Knoche, X. Li and A. Bhatnagar, Nano Lett., 2020, 20, 8789–8795.
131. D. Lee, H. Lu, Y. Gu, S.-Y. Choi, S.-D. Li, S. Ryu, T. R. Paudel, K. Song, E. Mikheev, S. Lee, S. Stemmer, D. A. Tenne, S. H. Oh, E. Y. Tsymbal, X. Wu, L.-Q. Chen, A. Gruverman and C. B. Eom, Science, 2015, 349, 1314–1317.
132. J. Chakrabartty, C. Harnagea, M. Celikin, F. Rosei and R. Nechache, Nat. Photonics, 2018, 12, 271–276.
133. X. Liu, F. Zhang, P. Long, T. Lu, H. Zeng, Y. Liu, R. L. Withers, Y. Li and Z. Yi, Adv. Mater., 2018, 30, e1801619.
134. W. S. Dos Santos, M. Rodriguez, A. S. Afonso, J. P. Mesquita, L. L. Nascimento, A. O. T. Patrocínio, A. C. Silva, L. C. A. Oliveira, J. D. Fabris and M. C. Pereira, Sci. Rep., 2016, 6, 31406.
135. C. Gao, Y. Zhao, W. Li, L. Jing, Y. Qiao, Z. Wang, J. Sheng, L. Wang and W. Fei, Nano Energy, 2022, 102, 107685.
136. C. Gao, W. Li, L. Jing, Z. Wang, L. Shi, J. Sheng, L. Wang, Y. Zhao and W. Fei, Adv. Funct. Mater., 2023, 33, 2213178.
137. S. Das, A. Sebastian, E. Pop, C. J. Mcclellan, A. D. Franklin, T. Grasser, T. Knobloch, Y. Illarionov, A. V. Penumatcha, J. Appenzeller, Z. Chen, W. Zhu, I. Asselberghs, L.-J. Li, U. E. Avci, N. Bhat, T. D. Anthopoulos and R. Singh, Nat. Electron., 2021, 4, 786–799.
138. Q. Qiu and Z. Huang, Adv. Mater., 2021, 33, 2008126.
139. Y. Li, J. Fu, X. Mao, C. Chen, H. Liu, M. Gong and H. Zeng, Nat. Commun., 2021, 12, 5896.
140. D. Yang, J. Wu, B. T. Zhou, J. Liang, T. Ideue, T. Siu, K. M. Awan, K. Watanabe, T. Taniguchi, Y. Iwasa, M. Franz and Z. Ye, Nat. Photonics, 2022, 16, 469–474.
141. R. Fei, W. Kang and L. Yang, Phys. Rev. Lett., 2016, 117, 097601.
142. K. Chang, J. Liu, H. Lin, N. Wang, K. Zhao, A. Zhang, F. Jin, Y. Zhong, X. Hu, W. Duan, Q. Zhang, L. Fu, Q.-K. Xue, X. Chen and S.-H. Ji, Science, 2016, 353, 274–278.
143. K. Kushnir, M. Wang, P. D. Fitzgerald, K. J. Koski and L. V. Titova, ACS Energy Lett., 2017, 2, 1429–1434.
144. S. Yuan, X. Luo, H. L. Chan, C. Xiao, Y. Dai, M. Xie and J. Hao, Nat. Commun., 2019, 10, 1775.
145. F. Liu, L. You, K. L. Seyler, X. Li, P. Yu, J. Lin, X. Wang, J. Zhou, H. Wang, H. He, S. T. Pantelides, W. Zhou, P. Sharma, X. Xu, P. M. Ajayan, J. Wang and Z. Liu, Nat. Commun., 2016, 7, 12357.
146. Z. Liu, H. Wang, M. Li, L. Tao, T. R. Paudel, H. Yu, Y. Wang, S. Hong, M. Zhang, Z. Ren, Y. Xie, E. Y. Tsymbal, J. Chen, Z. Zhang and H. Tian, Nature, 2023, 613, 656–661.
147. Z. Zheng, Q. Ma, Z. Bi, S. De La Barrera, M.-H. Liu, N. Mao, Y. Zhang, N. Kiper, K. Watanabe, T. Taniguchi, J. Kong, W. A. Tisdale, R. Ashoori, N. Gedik, L. Fu, S.-Y. Xu and P. Jarillo-Herrero, Nature, 2020, 588, 71–76.
148. K. Yasuda, X. Wang, K. Watanabe, T. Taniguchi and P. Jarillo-Herrero, Science, 2021, 372, 1458–1462.
149. M. V. Stern, Y. Waschitz, W. Cao, I. Nevo, K. Watanabe, T. Taniguchi, E. Sela, M. Urbakh, O. Hod and M. B. Shalom, Science, 2021, 372, 1462–1466.
150. C. Wang, L. You, D. Cobden and J. Wang, Nat. Mater., 2023, 22, 542–552.
151. Y. J. Zhang, T. Ideue, M. Onga, F. Qin, R. Suzuki, A. Zak, R. Tenne, J. H. Smet and Y. Iwasa, Nature, 2019, 570, 349–353.
152. A. Chanthbouala, V. Garcia, R. O. Cherifi, K. Bouzehouane, S. Fusil, X. Moya, S. Xavier, H. Yamada, C. Deranlot, N. D. Mathur, M. Bibes, A. Barthélémy and J. Grollier, Nat. Mater., 2012, 11, 860–864.
153. T. Akamatsu, T. Ideue, L. Zhou, Y. Dong, S. Kitamura, M. Yoshii, D. Yang, M. Onga, Y. Nakagawa, K. Watanabe, T. Taniguchi, J. Laurienzo, J. Huang, Z. Ye, T. Morimoto, H. Yuan and Y. Iwasa, Science, 2021, 372, 68–72.
154. Z. Zeng, Z. Tian, Y. Wang, C. Ge, F. Strauß, K. Braun, P. Michel, L. Huang, G. Liu, D. Li, M. Scheele, M. Chen, A. Pan and X. Wang, Nat. Commun., 2024, 15, 5355.
155. Q. Wang, J. Zheng, Y. He, J. Cao, X. Liu, M. Wang, J. Ma, J. Lai, H. Lu, S. Jia, D. Yan, Y. Shi, J. Duan, J. Han, W. Xiao, J.-H. Chen, K. Sun, Y. Yao and D. Sun, Nat. Commun., 2019, 10, 5736.
156. Z. Liang, X. Zhou, L. Zhang, X.-L. Yu, Y. Lv, X. Song, Y. Zhou, H. Wang, S. Wang, T. Wang, P. P. Shum, Q. He, Y. Liu, C. Zhu, L. Wang and X. Chen, Nat. Commun., 2023, 14, 4230.

---