PDF Journal of Materials Chemistry C - NJU

(1)

Cite this:DOI: 10.1039/c9tc04384e

Enhanced photovoltaic eﬃciency and persisted photoresponse switchability in LaVO

₃

/

Pb(Zr

_0.2

Ti

_0.8

)O

₃

perovskite heterostructures†

Shengliang Cheng,âbZhen Fan, *âbLei Zhao,âbHaizhong Guo, ^c Dongfeng Zheng,âZoufei Chen,âMin Guo,âYue Jiang,âSujuan Wu, â

Zhang Zhang, âJinwei Gao, âXubing Lu, âGuofu Zhou,^bdXingsen Gao â and Jun-Ming Liu âe

For ferroelectric photovoltaics, it is challenging to enhance the power conversion eﬃciency (PCE) without sacrificing the photoresponse switchability. Here, we demonstrate that enhanced PCE and good photoresponse switchability can be simultaneously achieved in perovskite heterostructures comprising narrow-gap semiconductor LaVO3 (LVO) and ferroelectric Pb(Zr0.2Ti0.8)O3 (PZT). The LVO(24 nm)/

PZT(120 nm) based device exhibits aB5-fold enhancement in PCE compared with the PZT-only based device, which is attributed to the enhanced absorption from the LVO layer and the built-in field at the LVO/PZT interface facilitating the separation of photo-generated e–h pairs. In addition, the switched photovoltage of the LVO/PZT based device is above 1 V, which is as large as that of the PZT-only based device. This persisted photoresponse switchability is obtained because the polarization can be fully switched in the LVO/PZT based devices when the LVO thickness is less than 24 nm. Our finding therefore provides a promising route for the development of high-efficiency and highly switchable ferroelectric photovoltaic devices.

Introduction

The ferroelectric photovoltaic (FEPV) effect, which occurs in a crystal with inherent non-centrosymmetry, has attracted tre- mendous attention recently because of its many intriguing characteristics.^1,2 For example, the FEPV effect can yield above-bandgap photovoltages,^3,4 switchable photocurrents,⁵ and tensorial photovoltaic (PV) outputs.⁶Due to these unique features, the FEPV effect is promising for wide applications in optoelectronic memories,⁵ photo-actuators,⁷ photo-sensors⁸ and so forth. To date the FEPV effect has been investigated in a variety of ferroelectrics, such as BiFeO3(BFO),^9–11Pb(Zr,Ti)O3

(PZT),^12–14BaTiO315,16and LiNbO3.^3,17These ferroelectrics are all wide-gap materials (42.5 eV) with poor absorption of solar radiation. Therefore, the photocurrent and the power conversion efficiency (PCE) generated by the FEPV effect are typically very small, which is a serious issue faced by the FEPV effect. To improve the PCE, various approaches have been developed. One of the most studied approaches is narrowing the ferroelectric’s bandgap by cationic substitution or doping (i.e., bandgap engineering).^11,18,19 For example, by replacing half of the Fe³⁺

ions in BFO by Cr³⁺ ions, Nechacheet al.¹⁰obtained a double perovskite Bi2FeCrO6with a narrowed bandgap (1.4 eV), which exhibited an unprecedentedly high PCE of 8.1%. While narrowing the ferroelectric’s bandgap can effectively enhance the PCE, it also inevitably causes a deterioration in spontaneous polarization (Ps) and an increase in leakage current,¹⁹ which in turn degrades the photoresponse switchability (exceptions do exist: cobalt substitution in BiFeO₃could reduce the bandgap while enhancing the spontaneous polarization²⁰). It is therefore still challenging to achieve a FEPV effect with enhanced PCE while maintaining its good photoresponse switchability.

A solution to this challenge lies in integrating the ferroelectric with a narrow-gap semiconductor absorber to form a semiconductor/ferroelectric (SEM/FE) composite structure.^12,13 In the composite structure, the semiconductor is mainly responsible

aInstitute for Advanced Materials, South China Academy of Advanced Optoelectronics, South China Normal University, Guangzhou 510006, China.

E-mail: [email protected]

bGuangdong Provincial Key Laboratory of Optical Information Materials and Technology, South China Academy of Advanced Optoelectronics,

South China Normal University, Guangzhou 510006, China

cSchool of Physical Engineering, Zhengzhou University, Zhengzhou 450001, China

dNational Center for International Research on Green Optoelectronics, South China Normal University, Guangzhou 510006, China

eLaboratory of Solid State Microstructures and Innovation Center of Advanced Microstructures, Nanjing University, Nanjing 210093, China

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

c9tc04384e

Received 9th August 2019, Accepted 13th September 2019 DOI: 10.1039/c9tc04384e

rsc.li/materials-c

Journal of

Materials Chemistry C

PAPER

Published on 16 September 2019. Downloaded by NANJING UNIVERSITY on 9/25/2019 1:42:23 AM.

View Article Online

View Journal

(2)

for absorbing photons and generating electron–hole (e–h) pairs, while the ferroelectric provides an internal electric field arising fromP_sto separate the e–h pairs. Due to the synergy between the ferroelectric and semiconductor, the photocurrent and PCE can be enhanced. Simultaneously, good photoresponse switchability can in principle be maintained because thePsof the ferroelectric is not sacrificed. Therefore, the SEM/FE composite structure offers an opportunity to achieve both enhanced PCE and good photoresponse switchability for the FEPV effect.

Considerable eﬀort has thus been devoted to investigating SEM/FE composite structures of various types, such as 0–3 type,^12,21 3–0 type,^22–241–3 type^25,26and 2–2 type (Fig. 1a–d).^13,27–29For the former three types, the semiconductors as either nanoparticles, matrices, or nanorods can lead to enhanced leakage currents. As a result, the actual switchable polarization in these composite structures will reduce even thoughP_sof the ferroelectric remains unchanged, which deteriorates the photoresponse switchability.

However, the 2–2 type is almost free of this leakage issue, making it a promising candidate for realizing both enhanced PCE and good photoresponse switchability.

So far a variety of 2–2-type SEM/FE composite structures have been investigated, including Cu2O/PZT,¹³ Si/PZT,²⁷ ZnO/

BFO,²⁸and BiVO4/BFO.²⁹Although the PCEs of these composite structures were enhanced to different degrees, most of them partially or even completely lost the photoresponse switchability, failing to meet the expectation for 2–2-type composite structures. Addressing this issue necessitates a comprehensive understanding of the effects of the semiconductor layer on both PV and polarization switching processes, which is however still lacking nowadays.

Herein, we design a 2–2-type perovskite heterostructure comprising a semiconducting LaVO₃ (LVO) upper layer on a ferroelectric Pb(Zr_0.2Ti_0.8)O₃ (PZT 20/80) lower layer (Fig. 1e), and conduct an in-depth study on the polarization switching and PV properties. PZT 20/80 is a prototype ferroelectric with large polarization (B80 mC cm ²),³⁰ while LVO is a Mott insulator with a suitable bandgap (1.1–1.8 eV)^31–33 for solar

absorption. Both PZT and LVO have perovskite structures with similar lattice constants (B4 Å),^30,31 enabling them to be conveniently integrated. The work functions of PZT and LVO are B4.6 eV^13,27andB4.0 eV,^31,33respectively, giving rise to a favorable energy band alignment when the LVO/PZT bilayer is sandwiched between Pt and SrRuO3 (SRO) electrodes.

Compared with the Pt/PZT/SRO device, the Pt/LVO/PZT/SRO device demonstrates aB5-fold increase in PCE without sacrificing the photoresponse switchability. The origins of the enhanced PCE and persisted photoresponse switchability are carefully analyzed in relation to the effects of the LVO layer on both the PV and polarization switching processes.

Results and discussion

PZT films grown on SRO-buﬀered STO substrates were first characterized. Fig. 2a shows the X-ray diﬀraction (XRD)y–2y pattern acquired from the PZT/SRO film [thickness (d_PZT) = 120 nm]. The distinct (00l) diffraction peaks suggest ac-oriented perovskite phase for PZT. To gain further information on the lattice parameters, reciprocal space mapping (RSM) was performed around the STO (103) reflection, as displayed in Fig. 2b.

The PZT (103) reflection corresponds to a single diffraction spot without splitting located atH= 0.996 andL= 2.788 (HandLare reciprocal space coordinates), indicating that the PZT film exhibits a tetragonal phase with lattice parameters: aPZT = 3.921 Å andcPZT= 4.202 Å. The AFM topography image of the PZT film shows a characteristic grid-like pattern ofa/c-domains (Fig. 2c), which is a typical feature of an epitaxial PZT film with a tetragonal phase.³⁴

To investigate the ferroelectricity of the PZT films, both piezoresponse force microscopy (PFM) and polarization–

voltage (P–V) hysteresis loop measurements were performed. As shown in Fig. 2d, a distinct phase contrast of 1801appears between the outer and inner boxes written using +9 V and 9 V, respectively, indicating that the ferroelectric domains can be switched.

Fig. 1 Schematics of SEM/FE composite structures of (a) 0–3, (b) 3–0, (c) 1–3, and (d) 2–2 types. (e) Schematic structure of the Pt/LVO/PZT/SRO device on the STO substrate.

(3)

Fig. 1d also illustrates that the as-grown region exhibits the same brown color as the 9 V-written region, suggesting that the as-grown region mainly contains the upwardc-domains. In addition, the square phase loop showing 1801 switching and the butterfly-like amplitude loop further confirm the domain switching in the PZT film (Fig. 2e). The macroscopicP–Vhysteresis loop measured on a Pt/PZT/SRO capacitor is displayed in Fig. 2f.

This P–V loop is near square (ESI,† Fig. S1), revealing a large remanent polarization ofB75mC cm ². All these results demonstrate the robust ferroelectricity of our PZT films.

Having obtained PZT films with good ferroelectricity, high- quality LVO semiconductor films were then fabricated for constructing the composite structures. Fig. 3a shows the XRD y–2ypattern of a LVO film [thickness (dLVO) = 60 nm] grown directly on the STO substrate. Only (00l) diffraction peaks from LVO and STO are observed, implying that the LVO film exhibits ac-oriented perovskite phase. The RSM around the STO (103) reflection (Fig. 3b) shows that the diffraction spots from LVO and STO have almost the same location on theH-axis, indicating that the LVO film is fully strained on the STO substrate.

According to the XRD and RSM results, the lattice parameters of LVO are determined as:a_LVO= 3.908 Å andc_LVO= 3.959 Å. This distorted orthorhombic structure found in the epitaxial LVO film grown on the STO substrate is consistent with those reported previously.^33,35 Additionally, Fig. 3c reveals a flat surface of the LVO film with a root-mean-square roughness (R_q) of onlyB80 pm.

Besides the good structural and morphological qualities, the LVO films also exhibit good optical and electrical properties.

Fig. 3d shows the absorption spectrum of the LVO film

illustrated by the Tauc plot of (ahn)²versus hn, whereais the absorption coeﬃcient andhnis the photon energy. The bandgap of LVO is extracted as B1.65 eV, consistent with the reported values.³¹This bandgap is much smaller than that of PZT (B3.6 eV; see our previous work in ref. 14). Through resistivity and Hall measurements, the LVO film is determined to be an n-type semiconductor with a carrier density ofB1.35 10¹⁹cm ³and a mobility ofB7.5 cm²V ¹s ¹. Fig. 3e compares the current–voltage (I–V) characteristics of a Au/LVO/Au planar metal/semiconductor/metal (MSM) structure in the dark and under illumination (AM1.5G, 100 mW cm ²). Upon illumination, the current is enhanced fromB2.410 ⁸ A toB9.5 10 ⁷A (@0.5 V), corresponding to a light-to-dark current ratio of B40. The photo-induced carrier density can be further estimated to be as large asB5.4 10²⁰ cm ³, agreeing well with the good photoresponsivity reported for high-quality LVO films.³²Therefore, the narrow bandgap and good photoresponsivity of our LVO films promise their suitability as the semiconductor layer for 2–2-type SEM/FE composite structures.

Then, SRO bottom electrodes, PZT layers (dPZT = 120 nm), LVO layers (dLVO= 12, 24, 33, and 42 nm), and Pt top electrodes were sequentially deposited on STO substrates, forming a series of Pt/LVO/PZT/SRO heterostructured devices. Note that the LVO layers are on top of the PZT layers because this configuration provides a favorable energy band alignment for the PV eﬀect (see detailed discussion later). Fig. 4a shows a typical AFM topography image of the LVO/PZT/SRO multilayered film. The film surface is relatively rough (Rq:B1.2 nm) with several small particles on it, probably due to the incoherent epitaxy of LVO on Fig. 2 (a) XRDy–2ypattern, (b) (103) RSM, and (c) AFM topography image of the 120 nm PZT film on the SRO-buffered STO substrate. In (c), the surface height values are relative values, and a negative value indicates a surface height which is below the average. (d) PFM phase image taken after electrically writing the outer (33mm²) and inner (1.51.5mm²) boxes using +9 V and 9 V, respectively. (e) Local PFM amplitude and phase hysteresis loops for the PZT film. (f) MacroscopicP–Vhysteresis loop of the Pt/PZT/SRO capacitor (frequency: 3.3 kHz).

Journal of Materials Chemistry C Paper

(4)

Fig. 4 (a) AFM topography image of the LVO/PZT/SRO multilayered film. (b) XRDy–2ypatterns of the LVO/PZT/SRO and LVO/PZT heterostructures on the STO substrate. (c) (103) RSM of the LVO/PZT heterostructure on the STO substrate. The red circle indicates the location of the LVO (103) peak. (d) Transmission spectra of the PZT/STO, LVO/STO, and LVO/PZT/STO samples, and insets showing the photographs of these samples. In (a–d),dPZT= 120 nm anddLVO= 33 nm.

Fig. 3 (a) XRDy–2ypattern, (b) (103) RSM, and (c) AFM topography image of the 60 nm LVO film on the STO substrate. (d) Tauc plot of (ahn)²versus hn for the LVO film, where the signals from the STO substrate are already extracted. (e)I–Vcharacteristics of the Au/LVO/Au planar MSM structure in the dark and under illumination (AM1.5G, 100 mW cm ²). The inset shows the linear plot of the darkI–Vcharacteristics.

(5)

PZT (see the ESI,†Fig. S2–S4, for evidence). To confirm that the desired phases were successfully formed, characterizations including XRD, RSM, and transmission electron microscopy (TEM) were conducted. For the complete LVO/PZT/SRO heterostructure, the (00l) peaks from PZT are clearly observed while those from LVO and SRO seem to overlap (see the blue line in Fig. 4b). To exclude the influence from SRO, the XRDy–2yscan and the (103) RSM were measured for a LVO/PZT heterostructure.

The (003) peak (red line in Fig. 4b) and (103) diffraction spot (Fig. 4c) from LVO are clearly identified, affirming the formation of the LVO phase. The high-resolution TEM cross-sectional view of the LVO/PZT/SRO heterostructure further reveals that the LVO film is successfully grown on the PZT film, forming a perovskite heterostructure (ESI,†Fig. S4).

The integrated LVO/PZT film is expected to produce enhanced optical absorption in the visible light spectrum compared with the PZT-only film, because the bandgap of LVO, which is in the visible light region, is narrower than that of PZT (1.65 eV versus3.6 eV). To verify it, the transmission spectra of the PZT/STO, LVO/STO, and LVO/PZT/STO samples were measured and compared. As shown in Fig. 4d, the transmittance decreases from PZT/STO to LVO/STO and to LVO/PZT/STO in the wavelength range of 380 to 900 nm, which is within the expectation that the LVO layer enhances the light

absorption for the LVO/PZT/STO heterostructure. Absorption edges below 400 nm are observed for all the samples, which may be mainly caused by the STO substrates since the bandgap of STO is 3.2 eV (B400 nm).

With the enhanced light absorption, whether the PV properties can also be improved in the Pt/LVO/PZT/SRO devices is of interest.

Prior to measuring the PV properties,P–Vhysteresis loops were measured in order to obtain diﬀerent polarization states for the PV measurements. In the P–V loop measurements, triangular voltage pulses of Vmax (frequency: 3.3 kHz) were applied to the top electrodes. The Vmax values were determined from the maximum allowable voltages which were unable to cause sample breakdown. The voltage is defined to be positive when a positive bias is applied to the top electrode, while current flowing from top to bottom is defined to be positive. As presented in Fig. 5a–c, the switchable polarizations in the devices withdLVOof 0, 12, and 24 nm are all large despite the fact that the leakage current gradually increases withdLVO. However, the switchable polarization becomes much smaller and even vanishes as d_LVO increases to 33 nm and to 42 nm, and the leakage currents are large in these two devices, as deduced from the strongly distorted P–V loops (Fig. 5d and e). These different polarization switching behaviors are also supported by the PFM measurements (ESI,†Fig. S5).

Because of the different polarization switching behaviors and the

Fig. 5 (a–e)P–Vhysteresis loops, (f–j) photovoltaicI–Vcharacteristics, and (k–o) time-dependent photocurrents of (a, f and k) Pt/PZT/SRO, (b, g and l) Pt/LVO(12 nm)/PZT/SRO, (c, h and m) Pt/LVO(24 nm)/PZT/SRO, (d, i and n) Pt/LVO(33 nm)/PZT/SRO, and (e, j and o) Pt/LVO(42 nm)/PZT/SRO devices. In (f–o), the voltages for positive and negative poling are indicated in the brackets.

(6)

above-mentioned upward self-polarization (Fig. 2d), the polarization states of all the devices after applying Vmax pulses (i.e., negative poling) are always the polarization up (P_up) state, whereas those after applying +V_maxpulses (i.e., positive poling) depend on the magnitudes of the switchable polarizations and may not always correspond to the polarization down (P_down) state.

After negative (or positive) poling, theI–Vcharacteristics of the Pt/PZT/SRO and Pt/LVO/PZT/SRO (d_LVO = 12, 24, 33 and 42 nm) devices were measured in the dark and under illumination (AM1.5G, 100 mW cm ²). As mentioned above, the Pup

state is always established after negative poling for all the devices; therefore, by comparing the PV properties of different devices in the negatively-poled state, one can see how the PV properties are influenced by the LVO layers. As seen from Fig. 5f–j and 6a, the short-circuit current (ISC) in the negatively-poled state (i.e.,ISC_NP) firstly increases fromB4 to B8 nA asdLVOincreases from 0 to 24 nm, and then decreases to B7.6 andB7.4 nA asdLVOfurther increases to 33 and 42 nm, respectively. These photocurrents are quite stable and repeatable, as evidenced by the time-dependent photocurrent measurements (Fig. 5k–o). The photocurrent enhancement in the Pt/LVO(24 nm)/

PZT/SRO device as compared to the Pt/PZT/SRO device is con- firmed by the external quantum efficiency (EQE) measurement (ESI,†Fig. S6). In addition, the PCE in the negatively-poled state shows a similar dependence on dLVO to that of ISC_NP, as displayed in Fig. 6a. Notably, the device with dLVO = 24 nm exhibits the highest PCE ofB0.01%, which isB5 times as large as the PCE of the device without LVO.

Then, we focus on the eﬀect of the LVO layer on the photoresponse switchability of the Pt/LVO/PZT/SRO devices.

As seen from Fig. 5f, the Pt/PZT/SRO device exhibits a typical switchable FEPV eﬀect, and the PV behaviors in the negatively- poled and positively-poled states are relatively symmetric.

In the negatively-poled state,I_SC and the open-circuit voltage

(VOC) are 4 nA and 0.72 V, respectively. Upon positive poling, ISCandVOCare switched to 4.5 nA and 0.55 V, respectively.

The switchable PV behaviors are dramatically changed when the LVO layers are introduced. For the devices withd_LVO= 12 and 24 nm, bothI_SCandV_OCin the negatively-poled state can still reverse their signs upon positive poling (Fig. 5g and h). For the device withd_LVO= 33 nm, however, the signs ofI_SCandV_OC are not reversed while only their magnitudes change (Fig. 5i).

For the device withd_LVO= 42 nm, almost no diﬀerence in PV behavior is observed between the negatively- and positively- poled states (Fig. 5j). Clearly, the LVO layer has a great impact on the photoresponse switchability of the Pt/LVO/PZT/SRO device. To evaluate the photoresponse switchability quantita- tively, the switched photovoltage, i.e., |VOC_PP VOC_NP|, is employed. As shown in Fig. 6b, |VOC_PP VOC_NP| decreases very slightly fromB1.27 toB1.1 V asdLVOincreases from 0 to 24 nm, but it decreases apparently fromB1.1 V to almost zero as dLVO increases from 24 to 42 nm. Therefore, the photoresponse switchability largely persists when d_LVO is within 24 nm, but it degrades significantly asd_LVOfurther increases.

Based on thed_LVO-dependent variation trend, the photoresponse switchability can be well correlated with the switchable polarization (Fig. 5a–e), as described as follows. Whend_LVOis within 24 nm, good photoresponse switchability is ascribed to the persistence of large switchable polarization. However, when dLVO is above 24 nm, the switchable polarization decreases significantly, thus deteriorating the photoresponse switchability.

To further demonstrate the correlation between the switchable photoresponse and polarization, bothVOCandISCas a function of pulse voltage were measured for the Pt/PZT/SRO and Pt/LVO/

PZT/SRO (dLVO= 24 nm) devices. As shown in Fig. 6c and d, these two devices exhibit similar hysteretic evolutions of VOCandISC, resembling their respectiveP–Vhysteresis loops. The switchable photoresponse is thus found to be controlled by the polarization,

Fig. 6 Summaries of (a)I_SCand PCE in theP_upstate and (b) |V_{OC_PP} V_{OC_NP}| of the Pt/LVO/PZT/SRO devices with diﬀerentd_LVO.V_OCandI_SCas a function of pulse voltage (V_p) for (a) Pt/PZT/SRO and (b) Pt/LVO(24 nm)/PZT/SRO devices.

(7)

which works not only in the Pt/PZT/SRO device but also in the Pt/LVO/PZT/SRO (dLVO= 24 nm) device.

The above results have illustrated that the integration of LVO into the PZT based FEPV devices can enhance the PCE and simultaneously retain the photoresponse switchability. To show the merits of our LVO/PZT based devices, we choose the one with the optimal d_LVO, i.e., 24 nm, and compare it with other 2–2-type SEM/FE composite structures in terms the overall PV performance. As summarized in Table 1, some composite structures, like Pt/Cu₂O/

PZT/ITO and Ag/a-Si/PZT/ITO, can exhibit high PCEs of above 0.5%, but their signs of V_OC and I_SC cannot be reversed. By contrast, although our LVO/PZT based device possesses only a moderate PCE ofB0.01%, it can exhibit fully reversibleVOCandISCand its switched photovoltage is the highest among all the composite structures.

Therefore, the LVO/PZT based devices can retain the promise of applications specifically for FEPV devices, such as electrically-written and optically-read non-destructive memories.⁵

The remaining task is to understand why the enhanced PCE and persisted photoresponse switchability can be simultaneously achieved in the Pt/LVO/PZT/SRO devices. As already discussed, the photoresponse switchability is determined by the switchable polarization. The polarization switching is likely to be influenced by the LVO layerviaa voltage partitioning eﬀect.⁴¹Specifically, when a voltage is applied to a Pt/LVO/PZT/SRO device, the voltage drop across the PZT layer is smaller than the applied voltage due to the voltage partitioned by the LVO layer. To fully switch the polarization in the PZT layer, the applied Vmax needs to be increased. The increment of Vmaxis typically proportional to dLVO whendLVO is small, which well explains the evolution of the P–V loops of the devices withdLVO= 0, 12, and 24 nm (Fig. 5a–c). However, whendLVO

is too large, the appliedVmaxcannot be further increased because it is limited by sample breakdown. The limitation of Vmax causes the polarization to be incompletely switched or even unswitched, agreeing with the polarization switching behaviors in the devices withdLVO= 33 and 42 nm (Fig. 5d and e). Therefore, there exists a maximum d_LVO(i.e., 24 nm) below which complete polarization switching can be achieved by increasing the appliedV_max. Because the switchable polarization determines the photoresponse switchability, good photoresponse switchability can thus be retained whend_LVOis within 24 nm.

We then proceed to analyze how the LVO layer leads to the PCE enhancement. Fig. 7a and b show the schematic energy band diagrams of the Pt/LVO/PZT/SRO heterostructure in the Pup and

Pdownstates, respectively. A Schottky barrier is formed at the PZT/SRO interface because SRO is a metal with a high work function of 5.2 eV⁴²while PZT 20/80 is an n-type semiconductor with a work function of 4.6 eV^13,27and an electron aﬃnity of 3.5 eV.⁴³This PZT/

SRO Schottky barrier was already demonstrated in our previous work.¹⁴ The Pt/LVO interface exhibits an Ohmic-like contact, as indicated by the linearI–Vcharacteristics observed in the Pt/LVO/

Pt device (inset in Fig. 3e). For the LVO/PZT interface, an n⁺–n junction may be formed because our LVO is heavily n-doped (see evidence from the resistivity and Hall measurements) with a work function of B4 eV,^31,33 apparently lower than that of PZT (B4.6 eV).^13,27 This work function difference is experimentally verified using scanning probe microscopy (SKPM), as shown in ESI,†

Fig. S7.

The interfacial energy band alignments described above give rise to two built-in fields, namely,Ebi-L/PandEbi-P/S, at the LVO/

PZT and PZT/SRO interfaces, respectively.Ebi-L/PandEbi-P/Shave the same downward direction, and their magnitude can change due to the polarization modulation (Fig. 7a and b). Besides, the directions of Ebi-L/P and Ebi-P/S are the same as that of the depolarization field (E_dp) in theP_upstate, but they are opposite Table 1 Comprehensive PV performance of various SEM/FE composite structures. The illumination conditions are the same, i.e., AM1.5G with 100 mW cm ²

Device structure V_OC(V) I_SC(mA cm ²)

Can the signs of V_OC

andI_SCbe reversed? |V_{OC_PP} V_{OC_NP}| (V) Eﬃciency (%)

Pt/Cu₂O/PZT/ITO¹³ 0.42 4.8 No 0.1 0.57

Ag/a-Si/PZT/ITO²⁷ 1.28 2.56 No 0.15 1.25

Au/CH₃NH₃PbI₃/BFO/FTO³⁶ 1.62 1.74 No 0 1.48

graphene/BFO/Pt³⁷ 0.44 0.025 No 0 0.003

Au/BVO/BFO/FTO²⁹ 1 0.14 No 0 0.039

AZO/BFO/FTO³⁸ 0.63 0.13 No 0 0.02

ITO/BFO/NBT–BT/FTO³⁹ 0.71 0.045 No 0 0.009

Pt/BTO/BFO/FTO⁴⁰ 0.38 0.056 No 0 0.005

Pt/LVO/PZT/SRO (this work) 0.97 0.028 Yes 1.1 0.01

Fig. 7 Schematic energy band diagrams of the Pt/LVO/PZT/SRO heterostructure in the (a)P_upand (b)P_downstates.

(8)

to that ofEdpin thePdownstate. Therefore, only in thePupstate canEbi-L/P,Ebi-P/SandEdpcontribute constructively to the PV eﬀect, consistent with our observation that the PV outputs are larger in theP_upstate than in theP_downstate (Fig. 5f–j). Then, let us focus on theP_upstate. When the LVO layer is thin enough to allow it to be fully depleted, the photo-induced e–h pairs generated in the whole LVO layer can be effectively separated by E_bi-L/P. Under this circumstance, the increase ind_LVOenhances the PV performance. However, when the LVO layer becomes thicker than the depletion width, the e–h pairs outside the depletion region cannot be effectively separated. Additionally, the transport of electrons from the boundary of the depletion region to the Pt electrode relies only on diffusion, which may cause significant recombination.²⁸ Therefore, further increasing dLVO

results in the reduction of the PV performance. The above analyses can well explain the observed dependence of the PCE ondLVOin the Pup state for our Pt/LVO/PZT/SRO devices (Fig. 6a). It is also deduced that adLVOof 24 nm, where the highest PCE is achieved, may correspond to the depletion width in the LVO layer.

Although the present Pt/LVO/PZT/SRO devices exhibit both enhanced PCE and good photoresponse switchability, the achieved highest PCE (B0.01%) is still low. Nevertheless, this PCE level does not represent the upper limit of the PCE in 2–2-type SEM/FE composite structures. The PCE can be further improved by (i) using a semiconductor layer with stronger absorption and photoresponsivity, (ii) using a ferroelectric layer with a narrower bandgap if the photoresponse switchability can be compromised, and (iii) enhancing the quality of the SEM/FE interface to allow more fluent charge flow across the interface.

Conclusions

In summary, 2–2-type perovskite heterostructures based on LVO/PZT sandwiched between Pt and SRO electrodes were fabricated. The PZT films exhibit robust ferroelectricity with a large remanent polarization (B75mC cm ²), while the LVO films exhibit a narrow bandgap (B1.65 eV) and can produce abundant charge carriers upon illumination (B5.410²⁰cm ³@ AM1.5G, 100 mW cm ²). Then, the PV properties of the Pt/LVO/PZT/SRO devices with differentdLVOwere systematically investigated, and the device withdLVO= 24 nm exhibits the highest PCE (B0.01%) while maintaining good photoresponse switchability (|V_OC__PP V_OC__NP| = 1.1 V). The PCE enhancement is attributed to the enhanced absorption from the LVO layer and the built-in field at the LVO/PZT interface contributing to the separation of photo-induced e–h pairs. The persisted photoresponse switchability is due to complete polarization switching being available in the Pt/LVO/PZT/SRO devices whend_LVOis no larger than 24 nm.

Therefore, the integration of a narrow-gap semiconductor with a ferroelectric represents an effective approach to achieve simultaneously high PCE and good photoresponse switchability in FEPV devices.

Experimental

SRO, PZT, and LVO thin films were sequentially grown on STO (001) substrates using pulsed laser deposition (PLD) (KrF excimer laser,

l= 248 nm) with a laser fluence of 0.8–1.0 J cm ²at a repetition rate of 5 Hz. The SRO layers (B40 nm) were first deposited on the STO substrates at 6801C under an oxygen pressure of 18 Pa. Then, the PZT layers (B120 nm) were grown on top of the SRO layers at a lower temperature of 615 1C with unchanged oxygen pressure.

Subsequently, a series of LVO films with thicknesses ranging from 12 to 42 nm were grown on the PZT/SRO/STO heterostructures at 6151C under a vacuum (B10 ⁴Pa). After the film deposition, Pt electrodes (diameter:B200mm; thickness:B10 nm) wereex situ deposited on the films by PLD through a shadow mask at room temperature and under a vacuum (B10 ⁴Pa).

The crystalline phases and lattice parameters of the films were measured by X-ray diﬀraction (XRD) and reciprocal space mapping (RSM) (X’Pert PRO, PANalytical). The film morphology, domain switching, and surface potential were characterized by atomic force microscopy (AFM), piezoresponse force microscopy (PFM), and scanning Kelvin probe microscopy (SKPM), respectively. The AFM, PFM, and SKPM measurements were carried out on a commercial atomic force microscope (Cypher, Asylum Research) with Pt-coated silicon tips (EFM Arrow, Nanoworld).

The microstructures of the films were studied using transmission electron microscopy (TEM) (Tecnai G2-F20). The resistivity and Hall eﬀect were measured by the Van der Pauw method with a physical property measurement system (PPMS) (Quantum Design). The P–Vhysteresis loops were recorded with a ferroelectric workstation (Precision Multiferroic, Radiant). No extra pre-poling was needed for measuring the P–V loops. The I–V characteristics were measured using a source meter (6430, Keithley). The optical transmission properties were characterized with a UV/vis spectrophotometer (UV1800, Shanghai Jinghua). For the photovoltaic measurements, a xenon lamp (CEL-HXUV300, Beijing Zhongjiaojinyuan) with an AM1.5G filter was used as the light source. The light intensity was calibrated to be 100 mW cm ² using an optical power meter (CEL-NP2000, Beijing Zhongjiao- jinyuan). For the EQE measurement, a series of optical filters with different selection wavelengths were used to obtain mono- chromatic light. All the electrical measurements were performed at room temperature.

Author contributions

Z. F. designed and supervised this project. S. C., L. Z., Z. C., Y. J., and S. W. prepared the samples. S. C., H. G., D. Z., and X. G.

conducted the XRD, AFM, and TEM measurements. S. C., M. G., Z. Z., J. G., X. L., G. Z., and J. L. conducted the electrical measurements. S. C. and Z. F. wrote the manuscript. All authors contributed to the scientific discussion and edited the manuscript.

Conflicts of interest

There are no conflicts of interest to declare.

Acknowledgements

The authors would like to thank National Key Research Program of China (No. 2016YFA0201002), State Key Program for Basic

(9)

Researches of China (No. 2015CB921202), National Natural Science Foundation of China (No. 51602110, 11674108, 51431006, 11574365, and 51561135014), Guangdong Innovative Research Team Program (No. 2013C102), Science and Technology Project of Guangdong Province (No. 2016B090918083 and 2017B030301007), Natural Science Foundation of Guangdong Province (No. 2016A030308019), and Science and Technology Project of Shenzhen Municipal Science and Technology Innovation Committee (GQYCZZ20150721150406).

X. G., X. L., and Z. F. acknowledge the Project for Guangdong Province Universities and Colleges Pearl River Scholar Funded Scheme 2014, 2016, and 2018, respectively.

Notes and references

1 Y. Yuan, Z. Xiao, B. Yang and J. Huang,J. Mater. Chem. A, 2014,2, 6027–6041.

2 Z. Fan, K. Sun and J. Wang, J. Mater. Chem. A, 2015, 3, 18809–18828.

3 A. M. Glass, D. von der Linde and T. J. Negran,Appl. Phys.

Lett., 1974,25, 233–235.

4 A. M. Glass, D. von der Linde, D. H. Auston and T. J. Negran, J. Electron. Mater., 1975,4, 915–943.

5 R. Guo, L. You, Y. Zhou, Z. S. Lim, X. Zou, L. Chen, R. Ramesh and J. Wang,Nat. Commun., 2013,4, 1990.

6 G. Dalba, Y. Soldo, F. Rocca, V. M. Fridikin and P. Sainctavit, Phys. Rev. Lett., 1995,74, 988.

7 P. Poosanaasa, K. Tonookab and K. Uchino, Mechatronics, 2000,6, 467–487.

8 S. C. Lai, K. Yao and Y. F. Chen, Appl. Phys. Lett., 2013, 103, 092903.

9 T. Choi, S. Lee, Y. J. Choi, V. Kiryukhin and S. W. Cheong, Science, 2009,324, 63–66.

10 S. Y. Yang, J. Seidel, S. J. Byrnes, P. Shafer, C. H. Yang, M. D. Rossell, P. Yu, Y. H. Chu, J. F. Scott and J. W. Ager, Nat. Nanotechnol., 2010,5, 143–147.

11 R. Nechache, C. Harnagea, S. Li, L. Cardenas, W. Huang, J. Chakrabartty and F. Rosei,Nat. Photonics, 2015,9, 61–67.

12 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.

13 D. Cao, C. Wang, F. Zheng, W. Dong, L. Fang and M. Shen, Nano Lett., 2012,12, 2803–2809.

14 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. Liu,NPG Asia Mater., 2019,4(11), 20.

15 A. Zenkevich, Y. Matveyev, K. Maksimova, R. Gaynutdinov, A. Tolstikhina and V. Fridkin,Phys. Rev. B: Condens. Matter Mater. Phys., 2014,90, 161409.

16 X. Hu, X. Yan, L. Gong, F. Wang, Y. Xu, L. Feng, D. Zhang and Y. Jiang,ACS Appl. Mater. Interfaces, 2019,11, 7379–7386.

17 M. Carrascosa, A. Garcı´a-Caban˜es, M. Jubera, J. B. Ramiro and F. Agullo´-Lo´pez,Appl. Phys. Rev., 2015,2, 040605.

18 H. Ye, W. Liao, C. Hu, Y. Zhang, Y. You, J. Mao, P. Li and R. Xiong,Adv. Mater., 2016,28, 2579–2586.

19 H. Matsuo, Y. Noguchi and M. Miyayama, Nat. Commun., 2017,8, 207.

20 P. Machado, M. Scigaj, J. Gazquez, E. Rueda, A. Sa´nchez-Dı´az, I. Fina, M. Gibert-Roca, T. Puig, X. Obradors, M. Campoy- Quiles and M. Coll,Chem. Mater., 2019,31, 947–954.

21 Q. Liu, Y. Zhou, L. You, J. Wang, M. Shen and L. Fang,Appl.

Phys. Lett., 2016,108, 022902.

22 Z. Xiao, Q. Dong, P. Sharma, Y. Yuan, B. Mao, W. Tian, A. Gruverman and J. Huang,Adv. Energy Mater., 2013, 3, 1581–1588.

23 F. Huang and X. Liu,Appl. Phys. Lett., 2013,102, 103501.

24 L. T. Quynh, C. N. Van, Y. Bitla, J. Chen, T. H. Do, W. Tzeng, S. Liao, K. Tsai, Y. Chen, C. Wu, C. Lai, C. Luo, Y. Hsu and Y. Chu,Adv. Energy Mater., 2016,5, 1600686.

25 H. Zhou, X. Chen, G. Wu, F. Gao, N. Qin and D. Bao,J. Am.

Chem. Soc., 2010,132(6), 1790–1791.

26 Y. Xu, B. Feng, M. Xue, Z. Li, Q. Xiong, J. Zhang, J. Duan, X. Wang and H. Wang,Int. J. Energy Res., 2017,41, 1847–1854.

27 F. Zheng, Y. Xin, W. Huang, J. Zhang, X. Wang, M. Shen, W. Dong, L. Fang, Y. Bai, X. Shen and J. Hao,J. Mater. Chem.

A, 2014,2, 1363–1368.

28 Z. Fan, K. Yao and J. Wang,Appl. Phys. Lett., 2014,105, 162903.

29 X. Li, S. Zhao, Q. Lu and Z. Wang,Int. J. Hydrogen Energy, 2017,42, 18246–18251.

30 H. N. Lee, S. M. Nakhmanson, M. F. Chisholm, H. M. Christen, K. M. Rabe and D. Vanderbilt,Phys. Rev.

Lett., 2007,98, 217602.

31 L. Wang, Y. Li, A. Bera, C. Ma, F. Jin, K. Yuan, W. Yin, A. David, W. Chen, W. Wu, W. Prellier, S. Wei and T. Wu, Phys. Rev. Appl., 2015,3, 064015.

32 H. Zhang, M. J. Brahlek, X. Ji, S. Lei, J. Lapano, J. W. Freeland, V. Gopalan and R. E. Herbert, ACS Appl. Mater. Interfaces, 2017,9(14), 12556–12562.

33 M. Jellite, J. Rehspringer, M. Fazio, D. Muller, G. Schmerber, G. Ferblantier, S. Colis, A. Dinia, M. Sugiyama, A. Slaoui, D. Cavalcoli and T. Fix,Sol. Energy, 2018,162, 1–7.

34 I. Vrejoiu, G. L. Rhun, N. Zakharov, D. Hesse, L. Pintilie and M. Alexe,Philos. Mag., 2006,86(28), 4477–4486.

35 H. Rotella, U. Lu¨ders, P.-E. Janolin, V. H. Dao, D. Chateigner, R. Feyerherm, E. Dudzik and W. Prellier, Phys. Rev. B:

Condens. Matter Mater. Phys., 2012,85, 184101.

36 P. Zhao, L. Bian, L. Wang, J. Xu and A. Chang,Appl. Phys.

Lett., 2014,105, 013901.

37 Y. Zang, D. Xie, X. Wu, Y. Chen, Y. Lin, M. Li, H. Tian, X. Li, Z. Li, H. Zhu, T. Ren and D. Plant,Appl. Phys. Lett., 2011,99, 132904.

38 W. Dong, Y. Guo, B. Guo, H. Liu, H. Li and H. Liu,Mater.

Lett., 2013,91, 359–361.

39 F. Wu, Y. Guo, B. Guo, Y. Zhang, H. Li and H. Liu,J. Phys. D:

Appl. Phys., 2013,46, 365304.

40 J. Zhang, H. Zheng, Y. Zhang, G. Yuan, W. Gao, X. Liu, G. Yin, Y. Gu and W. Zhang,Mater. Lett., 2015,156, 98–100.

41 A. K. Tagantsev, M. Landivar, E. Colla and N. Setter,J. Appl.

Phys., 1995,78, 2623.

42 X. Fang and T. Kobayashi,Appl. Phys. A: Mater. Sci. Process., 1999,69, S587–S590.

43 J. F. Scott, K. Watanabe, A. J. Hartmann and R. N. Lamb, Ferroelectrics, 1999,225(1), 83–90.