Photovoltaic Behaviors Regulated by Band-Gap and Bipolar Electrical

Oct 17, 2016 - One investigated the regulation of band gap and bipolar electrical cycling on photovoltaic behaviors including photovoltaic spectral re...
0 downloads 4 Views 2MB Size
Article pubs.acs.org/JPCC

Photovoltaic Behaviors Regulated by Band-Gap and Bipolar Electrical Cycling in Holmium-Doped Bi5Ti3FeO15 Ferroelectric Films Yulong Bai, Jieyu Chen, Xin Wu, and Shifeng Zhao* School of Physical Science and Technology, & Inner Mongolia Key Lab of Nanoscience and Nanotechnology, Inner Mongolia University, Hohhot 010021, PR China ABSTRACT: The pure and Ho-doped Bi5Ti3FeO15 ferroelectric films with layered perovskite structure were prepared via a chemical solution deposition. One investigated the regulation of band gap and bipolar electrical cycling on photovoltaic behaviors including photovoltaic spectral response, open-circuit voltage, and short-circuit current for the films. It is shown that the photovoltaic response peak of Ho-doped Bi5Ti3FeO15 films shifts toward the visible region and the open-circuit voltage as well as shortcircuit current are improved compared with the pure Bi5Ti3FeO15 films, which is attributed to the narrowed band gap after doping with Ho. And the spectral response, open-circuit voltage as well as short-circuit current of Hodoped Bi5Ti3FeO15 films do not suffer from the fatigue like the pure Bi5Ti3FeO15 films with the action of bipolar electrical cycling. It just appears extended response peak without blue shift from the visible region. Likewise, Ho-doped Bi5Ti3FeO15 films still enjoy strong open-circuit voltage and short-circuit current even undergoing long period switching. Such antifatigue photovoltaic behaviors derive from the antifatigue ferroelectric, dielectric properties under illumination for Ho-doped Bi5Ti3FeO15 films. Thus, regulation of band gap and bipolar electrical cycling on the photovoltaic behaviors of ferroelectrics promotes their potential application in film solar cells.



INTRODUCTION Solar energy is regarded as one of the most reliable and abundant clean energy source. For a traditional semiconductor solar cell in a p−n junction, photons with energy higher than the band gap are absorbed to produce electron−hole pairs, which are separated by the internal field and collected at the electrodes. However, the p−n junction is not irreplaceable for the photovoltaic effect. A ferroelectric photovoltaic effect can exist in homogeneous noncentrosymmetric materials such as ferroelectrics under uniform illumination.1 Different from the photovoltaic effect observed in the p−n junction, the requirement is ferroelectric polarization rather than an asymmetric interface such as p−n junctions since the ferroelectric photovoltaic effect originates from the electron− hole separation at ferroelectric domain walls. Therefore, its open-circuit voltage is not limited by the band gap like p−n junction solar cells, but may be larger than the band gap of the ferroelectric materials, which is a bulk photovoltaic effect (BPVE).2,3 However, photovoltaic effects in ferroelectric materials are still limited by two factors. One is the low conversion efficiency in the visible region as only small amount of photoexcited carriers are collected at the electrode due to their large band gap corresponding to the ultraviolet region. The other is that the photovoltaic performance are depressed due to the breakdown of the critical periodic ferroelectric domain walls when the films are repetitively polarized to endure the ferroelectric fatigue accompanying with the reduced coercive field and remnant polarization.4 Therefore, further narrowing band gap and improving the ferroelectric fatigue © XXXX American Chemical Society

resistance are expected for the excellent ferroelectric photovoltaic effect. As a kind of environmental-friendly lead-free ferroelectric, Bi5Ti3FeO15 (BTFO) materials with a four-layered perovskite unit of (Bi3Ti3FeO13)2− sandwiched by two (Bi2O2)2+ layers along the c-axis5 are tested in the ferroelectric photovoltaic effect due to their excellent ferroelectric and leakage properties.6 Up to date, there have been few reports on the ferroelectric photovoltaic effects of BTFO films, in particular, photovoltaic behaviors regulated by band gap and bipolar electrical cycling. Therefore, this work aims to reveal the regulation of band gap and bipolar electrical cycling on the ferroelectric photovoltaic behaviors including photovoltaic spectral response, open-circuit voltage, and short-circuit current in Ho-doped BTFO films. On one hand, Ho-doping can destroy the strong inversion symmetry structure of BTFO films, which results in a narrower band gap corresponding to enhanced optical absorption in the range of the visible region. On the other hand, Ho3+ dopants are known as the grain growth inhibitor, suppressing oxygen vacancies, and then further improving the antifatigue properties.7 Thus, photovoltaic behaviors are regulated by band gap and bipolar electrical cycling. Received: August 5, 2016 Revised: October 4, 2016 Published: October 17, 2016 A

DOI: 10.1021/acs.jpcc.6b07927 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C



EXPERIMENTAL PROCEDURE The pure and Ho-doped Bi5Ti3FeO15 films were prepared via chemical solution deposition. Bi(NO3)3·5H2O, Fe(NO3)3· 9H2O, Ho(NO3)3·9H2O, and butyl titanate [CH3(CH2)3O]4Ti were dissolved in glycol to form the precursor solutions. The molar ratio of Ho:Bi for BHTFO films was 0.25:4.25, which suggests that the Ho doping concentration is 5%. The above precursor solutions were deposited on Pt(111)/Ti/SiO2/Si wafers by spin coating. Finally, the films were annealed by the rapid thermal processor (RTP-500) at 700 °C under air for 10 min. These processes were repeated 18 times to form films with a thickness of 1 μm. The crystalline phases were characterized by X-ray diffraction (XRD, Panalytical Empyrean). And the ferroelectric properties were studied by a multiferroic tester system (Multi-Ferroic100 V, Radient Technology, USA). The optical reflectance spectrum was measured by an ultraviolet−visible (UV) spectrophotometer (Hitachi, U3900). In order to measure the electric properties, Au electron dots with diameter of 200 μm are sputtered on the surface of the films with a mask. The small electron dots can depress the influences of the leakage on the ferroelectric properties. And the orientation of measurement is out of the film plane. The experimental setup for photovoltaic behaviors is a self-designed system, as sketched in Figure 1. Photovoltaic response was measured using a phase-

locked amplifier (Stanford, SR830) and Xenon arc lamp (Osram,7ILX500) in conjunction with a grating monochromator operated in the wavelength range of 300−1100 nm. J−V curves (photovoltaic current density−voltage) were measured under the white-light illumination to obtain the open-circuit photovoltage VOC and short-circuit photocurrent density JSC using a source meter (Keithley 2400). Where Multi-Ferroic100 V was employed as a fatigue resource to offer a periodic switchable electrical field, one also can measure the ferroelectric properties. When measuring the properties after fatigue, one must first connect the films with the Multi-Ferroic100 V to offer periodic bipolar electrical cycling and then carry out the corresponding measurement. The effect of illumination on dielectric constant was measured under the same condition by a precision impedance analyzer (Agilent E4990A) with a wide range of frequencies from 100 Hz to 1 MHz. Above these measurements were carried out at room temperature.



RESULTS AND DISCUSSION Figure 2a shows the XRD patterns of the pure and Ho-doped BTFO films. It is shown that the films belong to the layered perovskite structure. This Aurivillius structure containing four perovskite layers is identified by indexing all the diffraction peaks on the basis of an orthorhombic cell (Joint Committee for Powder Diffraction Standard (JCPDS) no.38−1257). Neither Bi25FeO40 nor Ho and its oxides impurity phases are observed. The similar diffraction patterns between the pure and Ho-doped BTFO films reveal that the present dopants concentration does not reach solubility limit of the parent compounds. It also indicates that Ho3+ ions have entirely entered into BTFO lattice. However, there is a remarkable difference between the pure and Ho-doped BTFO films. Some diffraction peaks such as (006), (1113), (0012), (026), and (0018) appear for BHTFO films. The diffraction peaks (111), (119), and (200) shift toward high 2θ angles for BHTFO films, as shown in Figure 2b. All of these results suggest that the phase structure is distorted, and the interplanar distance decreases after doping with the Ho element, which is attributed to the fact that the smaller radius Ho3+ (0.901 Å) has successfully substituted out the bigger radius Bi3+ (1.03 Å). In order to confirm thus structural transformation, the Rietveld method by EXPGUI is carried out in an orthorhombic structure. The results of Rietveld refinements are shown in Figure 2c. The optimal goodness of fit (χ2) and R-factors (Rp, Rwp) are used as numerical criteria of the fitting quality. The

Figure 1. Schematic illustration of the test system for photovoltaic behaviors.

Figure 2. (a) XRD results for BTFO and BHTFO films. (b) Enlarged region ranging from 22° to 34°. (c) Final Rietveld refinement XRD for BTFO and BHTFO films. B

DOI: 10.1021/acs.jpcc.6b07927 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C results reveal that the phase structures for both films belong to mixture space groups including A21am and F2mm. However, the fractions of each space group are different for the pure and Ho-doped BTFO films. All of the unit cell parameters and reliability factors obtained from the Rietveld refinement analysis for both films are summarized in Table 1. It is shown that the

of leakage current toward the overall polarization, positive-up negative-down (PUND) measurements were carried out at a pulse width of 1 ms, the results of which are inserted in Figure 3, parts a and b, to confirm the ferroelectric properties. These results show that all the films present well-defined ferroelectric properties. However, the ferroelectric properties change obviously when the illumination and bipolar electrical cycling are applied. The influence of the illumination and bipolar electrical cycling on the ferroelectric properties for the pure and Ho-doped BTFO films is summarized in Table 2. For the pure BTFO films, the polarization values are depressed more than 60%. To be specific, the remenant polarization (Pr) decreases from 19.1 to 6.8 μC/cm2 and saturation polarization decreases from 48.2 to 18.0 μC/cm2 under dark field after undergoing 109 bipolar electrical cycling. The results suggest that the pure BTFO films appear obviously ferroelectric fatigue behaviors. Besides, after illumination, Pr and Ps values increase from 19.1 to 20.1 μC/cm2 and from 48.2 to 52.5 μC/cm2, corresponding to increasing rates 5.23% and 7.67%. This means that the polarization charges induced from illumination contributed to the enhanced ferroelectric properties. After bipolar electrical cycling, the illumination works more obviously than that without bipolar electrical cycling. Under illumination, Ps value increase from 18 to 21.7 μC/cm2, corresponding to increasing the rate 20.6%. Thus, results suggest that the depressed ferroelectric polarization from the ferroelectric fatigue is partially recovered, which is attributed to the fact that photons can activate the polarization charges locked in the dead layers derived from the bipolar electrical cycling. When the light is irradiated on the fatigued surface (dead layer) of ferroelectric materials, the optical energy is transported to the lattice due to the collision between photons and the surface lattice. Subsequently, two phenomena may occur: one is that the surface active energy is enhanced because part of the photon energy is absorbed by the lattice. Thus, the new domain nucleation probably is formatted with the favored orientation of polarization, which subsequently expands to recover the existed domains.8 Another is that the optical stress induced by the lattice misfit is formed at the interface between the electrode and the films. The optical strain is an important factor for activation of the dead layers since the strains can alter the equilibrium polarization state and enhance spontaneous polarization.9 At the same time the surface thermal energy

Table 1. Structural Parameters for the Pure and Ho-Doped BTFO Films Obtained by the Rietveld Refinements parameters

BTFO

BHTFO

2θ (scanning scope) scanning rate crystal system space group (A21am) fraction space group (F2mm) fraction a (Å) b (Å) c (Å) volume (Å3) Rp Rwp χ2

10−60° 0.02° orthorhombic 45.5% 54.5% 5.5027 5.4808 41.6469 1256.198 0.0580 0.0757 1.078

10−60° 0.02° orthorhombic 49.7% 50.3% 5.4755 5.4549 41.7523 1247.069 0.0610 0.0770 1.036

crystal structural composition, lattice parameters, and unit cell volume all change after doping with Ho. For BTFO films the fractions of A21am and F2mm are 45.5% and 54.5%, respectively. However, the concentration of the A21am group increases to 49.3%, and that of the F2mm group decreases to 50.7% for BHTFO films. The lattice parameters are a ∼ 5.5027 Å, b ∼ 5.4808 Å, and c ∼ 41.6469 Å, which are different from Ho-doped BTFO films with a ∼ 5.4755 Å, b ∼ 5.4549 Å, and c ∼ 41.7523 Å. Thus, the cell volume is also different for BTFO with ∼1256.198 Å3 and for BHTFO with ∼1247.069 Å3. Thus, decreased volume and changed phase composition are induced by the different ionic radius between dopants Ho3+ and matrix Bi3+. Generally, the distorted structure is accompanied by a change of the ferroelectric properties. Figure 3a shows P−E hysteresis of the pure BTFO films before and after bipole electrical cycling under dark and illuminated conditions. Figure 3b describes P−E curves for BHTFO films. The measured electrical field is in the range of −350 kV/cm to 350 kV/cm. In order to extract the real ferroelectric properties and properly represent the contribution

Figure 3. Ferroelectric polarization vs. electric field under different condition: no cycling, cycling and dark, illumination: (a) for BTFO films and insert curves corresponding to 2Pr PUND values vs electric field; (b) for BHTFO films and insert curves corresponding to 2Pr PUND values vs electric field. C

DOI: 10.1021/acs.jpcc.6b07927 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Table 2. Values of Ps, Pr, and 2Pr of Hysteresis, 2Pr of PUND under the Following Conditions: No Cycling and Cycling up to 109 and Dark and Illumination for BTFO Films and BHTFO Films BTFO film cycling ≥ 10

no cycling Ps Pr 2Pr hysteresis 2Pr PUND

BHTFO film 9

cycling ≥ 109

no cycling

dark

illumination

dark

illumination

dark

illumination

dark

illumination

48.2 19.1 37.3 35.2

52.5 20.1 48.4 45.2

18.0 6.8 13.7 11.4

21.07 9.7 16 13.1

52.1 27.6 53.8 50.9

60.8 38.6 59.8 55.9

34.1 17.8 37.7 30.3

45.1 23.8 45.1 43.9

Figure 4. Schematic illustration for the mechanism of the ferroelectric properties regulated by illumination and bipolar electrical cycling.

activates nucleation to switch ferroelectric domains.10 Thus, the polarization charges remove from the dead layers and the ferroelectric polarization is enhanced. Compared with the pure BTFO films, Ho-doped BTFO films show attractive antifatigue behaviors. Undergoing 109 bipolar electrical cycling, Pr and Ps just degrade to 17.8 and 34.1 μC/ cm2 compared with the original values of Pr ∼ 27.6 μC/cm2 and Ps ∼ 52.1 μC/cm2 under a dark field. The polarization values are depressed about 34%. Therefore, the antifatigue properties are improved for Ho-doped BTFO films. More interestingly, under illumination the polarization is enhanced more obviously. To be specific, the Pr value increases from 27.6 to 38.6 μC/cm2, with the increasing rate reaching 40%. In particular, electrical fatigue is recovered to a great extent under illumination. The Pr value increases from 17.8 to 23.8 μC/cm2, which is recovered to 86.2% of the original value of 27.6 μC/cm2 before bipolar electrical cycling. This is attributed to the effectively decreased band gap after doping with Ho, which results in the fact that more photons are absorbed by the lattice and surface of films. PUND results also show a similar tendency. It is shown that undergoing electrical cycling, the polarization of BTFO films is deteriorated with largely declining from the original values. Under illumination, the ferroelectric properties of BTFO films are slightly enhanced to some degree. In contrast, the switchable polarization of BHTFO films withstood same number switchable cycling is only slightly reduced. More interestingly, under illumination 2Pr values are enhanced obviously, which suggests that the illumination improves ferreoelectric properties from the fatigue behaviors. Thus, results further confirm the intrinsic ferroelectric properties and natural behaviors. Therefore, it is not difficult to draw a conclusion that Ho-doping and bipolar electrical cycling both influence the ferroelectric properties. And an interesting phenomenon is that under illumination both factors are working better. Thus, behaviors are attributed to the relationship between illumination and ferroelectric fatigue. In fact, the

illumination can enhance the ferroelectric polarization for thus ferroelectric multilayered structures. Figure 4 schematically illustrates the mechanism of the ferroelectric properties regulated by illumination and bipolar electrical cycling. The light illumination incidents on the surface then generates electron−hole pairs at the interface, which can compensate the depolarizing field, further enhance the polarization. However, in individual ferroelectric films, electron−hole recombination limits the lifetime of the carriers, minimizing this enhancement. And photoexcited carriers are isolated on the nonferroelectric side of the interface at the ferroelectric films, minimizing recombination. This would result in physically separation, longlived screening charges, and potentially high Pr.11 After doping with Ho, induced structural distortion and strain release lead to relative small internal barrier, which is helpful to generate electron−hole pairs at the interface and further improve the polarization. After undergoing bipolar electrical cycling, charges injected from the electrode area agglutinate at the nucleation sites, which consequently results in the local phase decomposition. Thus, the “dead layer” is formed due to the existence of local phase decomposition. Since the bipolar electrical cycling induces the electron-injected effect, the “dead layer” is directly connected to the electrode. As reported, the “dead layer” mainly contributes to the ferroelectric fatigue behavior and degraded ferroelectric properties.12 Therefore, the ferroelectric fatigue behaviors occur after the sample undergoes bipolar electrical cycling. These dead layers depress the ferroelectric properties of the ferroelectric films and even make it become an inslulator at the ferroelectric−electrode interface.13 Thus, layers insulate the electrodes from the ferroelectric films and inhibit charge compensation, which generates a strong depolarization field in the ferroelectric films and then directly decreases the ferroelectric photovoltaic effect. Under illumination, photons can drive the mobility of carriers, thus decreasing the injected charge density near the electrodes. So the injected electrons D

DOI: 10.1021/acs.jpcc.6b07927 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 5. Relative dielectric constant vs frequency at the condition: no cycling, cycling and dark, illumination: (a) for BTFO films; (b) for BHTFO films.

Figure 6. (a) J−V curves in the dark field and under illumination at room temperature for BTFO and BHTFO films without electrical cycling and insert figure is enlarged center region. (b) J−V curves undergoing bipolar electrical cycling up to 109 and (inset) enlarged center region.

the measuring frequency of 0.4 MHz. After bipolar electrical cycling, the dielectric constant is found to decrease with the number of switching cycles accumulation, approximately declining 50%. Another point is that the dielectric constant under illumination also reduces slightly more than that without illumination. For BHTFO films, the ε value is shown in Figure 5b. After bipolar electrical cycling, the dielectric constant declines about 11%, which is less than that of the pure BTFO films. Under illumination, the dielectric constant also declines compared with dark conditions. Interestingly, under illumination, the value of dielectric constant is in variation, which is also corresponding to the results of ferroelectric properties. Thus, dielectric behaviors can be explained by space-chargelimited conduction (SCLC) model.14 It suggests that local phase decomposition causes dielectric declining in two ways. On one hand, the effective field applied in the films is significantly reduced after decomposition due to the low dielectric constant of the degraded layer. On the other hand, one can see that the most probable locations where phase decomposition occurs are the domain nucleation sites. The collapse of nuclei and the decrease in the number of the available nucleation sites during electrical cycling also depress the switching of the domains, which makes the induced carriers move more difficultly. Accordingly, the carriers are generated in ferroelectric layer by passing through internal barrier and

induced local phase decomposition and the collisions are depressed. Therefore, the near “dead layer” can be partially activated and the thickness of the “dead layer” declines. The activated “dead layer” can work as originally (without bipolar electrical cycling), which implies the recovery of the ferroelectric polarization to some extent. For Ho-doped BTFO films, the declined internal potential barrier is helpful for the moving of the electron−hole pairs, so local phase decomposition derived from the injected electrons is inhibited; subsequently, it forms more thin “dead layer” comparing with the pure BTFO films. Therefore, the antifatigue properties are improved after doping with Ho. Under illumination, induced electron−hole pairs at the interface more easily move through the thin “dead layer”. So for Ho-doped BTFO films after bipolar electrical cycling, the excited carriers compensate the polarization charges restrained in the “dead layer” under illumination. Thus, films exhibit ferroelectric fatigue recovery behavior. These results can be deduced by the difference of dielectric properties with Ho doping and bipolar electrical cycling under illumination. Relative dielectric constant dependence of BTFO films on frequency under dark field and illumination before and after bipolar electrical cycling is described in Figure 5a. It is shown that the relative dielectric constant εr of BTFO films gradually decreases with frequency increasing, then sharply decreases at E

DOI: 10.1021/acs.jpcc.6b07927 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 7. Before and after undergoing bipolar electrical cycling photoresponse with the wavelength starting at 300 to 1100 nm: (a) for BTFO films; (b) for BHTFO films.

collected at the top dot electron when the films are irradiated. Thus, the photoinduced carriers at the interface change the interfacial polarization and affect the dielectric properties.15 Since Ho-doping and bipolar electrical cycling can regulate the ferroelectric and dielectric properties, thus two factors can affect the photovoltaic effect driven by electrical polarization in ferroelectrics. This can be demonstrated by different photocurrent and photovoltaic response behaviors. Moreover, oxygen vacancies easily are assembled on the ferroelectric domains or grain boundaries, which can cause domain wall pinning and then hinder the switching of domains.16 Since the ferroelectric photovoltaic effect originates from the electron−hole separation at ferroelectric domain walls, the more easily the ferroelectric domains switch, the lower voltage is to observe ferroelectric photovoltaic effect. For the pure and Ho-doped BTFO films, the 2D nanostructure (Bi2O2)2+ layers can stabilize octahedron and product less defect (such as oxygen vacancies), which can depress the domain wall pinning.16 Therefore, the ferroelectric domain is easy to switch and the ferroelectric photovoltaic effect is obtained at a lower external voltage range of −10 to +10 V for the pure and Ho-doped BTFO films.17 Figure 6a shows the current density−voltage (J−V) characteristics of films before bipolar electrical cycling, which is indicative of the photovoltaic nature. The current−voltage characteristics were recorded under dark and illuminated conditions within a voltage range of −10 to +10 V. It is shown that BTFO films possess little dark current density in the measured range; to be specific, the values of short-circuit current (Jsc) and open-circuit voltage (Voc) for BTFO films are 4.1 × 10−4 mA/cm2 and −0.19 V at dark field. While under illumination, the improved photovoltaic effect is observed with the Jsc ∼ 5 × 10−4 mA/cm2 and Voc ∼ −0.58 V, respectively. For Ho-doped BTFO films, Jsc and Voc both have larger values than that of the pure BTFO films, which suggests a strong photovoltaic effect. The narrower band gap derived from Ho doping improves its light absorption. Therefore, when BHFTO films are illuminated, photoinduced carriers more easily are drifted away by the spontaneous polarization and produce photovoltage due to the relatively weak internal potential barrier. Thus, Jsc arises from the movement of carriers through BHTFO films. Whereas Voc originates from the difference in chemical potential between the different charge carriers accumulated on the electrode interfaces.18 So the general relation between Voc and Jsc can be described: Voc = nKT/q ×

ln(Jsc/j0), where n is the ideality factor, k is the Boltzman constant, T is the temperature, q is the electron charge, and j0 is the measured saturation current density. Therefore, the increased Jsc should result in an increase in Voc. In dark field, the abnormally somewhat high Voc and Jsc values in both the pure and Ho-doped BTFO films are intriguing, which greatly arouses our interest. The following concepts can help to understand the causes. On one hand, the dark current is attributed to the high ferroelectric polarization and leakage current density.19 The carriers under an electric field are induced to form current. Therefore, the dark current is understood as the Schottky-like barrier-to-Ohmic contacts resulting from the combination of carriers and polarization in unsymmetrical electrode.20 On the other hand, it exists low internal barrier potential and more carriers for Ho-doped BTFO films. When the same external field is applied, the more internal charges are separated and carried onto the electron. Thus, it is credible to observe the somewhat high dark current values of BTFO and BHTFO films. Figure 6b shows the J−V curves of the pure and Ho-doped BTFO films after bipolar electrical cycling. It shows a decrease of photovoltaic effect with the Voc and Jsc near to zero for the pure BTFO films. This suggests that after fatigue, there is a huge internal potential barrier, which inhabits the photongenerated carriers and enhances the thermal collide chances, so almost no electrons move toward the surface of ferroelectric materials. The observed short-current is just attributed to the fatigue current density j0 in the films due to the increase of leakage current. The increased fatigue current can be reflected by comparing the value of 2Pr hysteresis with that of 2Pr PUND. The obvious difference between thus two 2Pr suggests the larger leakage current. On the contrary, for Ho-doped BTFO films, in dark field after fatigue the Voc and Jsc values are −0.25 V and 6 × 10−3 mA/cm2, respectively. When under illumination Ho-doped BTFO films still possess strong photovoltaic effect, even undergoing long period switching. To confirm the presence of photogenerated carriers and exclude the leakage current, the PUND measurements are carried out. From the PUND results, the leakage current takes place a bit after fatigue, so the obtained short-current is dominated by the photocurrent. It is also evident from these results that effectively controlling the leakage currents in these films can lead to a higher value of Voc. Generally, it is proposed that the photovoltaic effect is induced by the depolarizing electric field, which separates the F

DOI: 10.1021/acs.jpcc.6b07927 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 8. (a) Reflectance spectrum for BTFO and BHTFO films from 250 to 800 nm. (b) (αhν)2 versus hv for plots used for the estimation of direct band gap for BFTO and BHTFO films.

For BHTFO films, the photovoltaic response peak is located at 500 nm, as shown in Figure 7b. It suggests that the response peak has a significant red-shift toward the visible region from 470 to 500 nm after doping with Ho. So the photovoltaic effect is improved due to the narrower band gap of Ho-doped BTFO films. Besides, the location of the photovoltaic response peak does not obviously shift but exhibits a broadened region after bipolar electrical cycling, which suggests that the modulation action of bipolar electrical cycling on the photovoltaic effect of BHTFO films is weaker than that of the pure BTFO films. That is to say, compared with BTFO films, the photovoltaic antifatigue properties of BHTFO films are improved, which is attributed the excellent ferroelectric fatigue resistance since thus BPVE originates from the spontaneous polarization of the ferroelectric films. The above results demonstrate that both bipolar electrical cycling and Ho doping regulate the ferroelectric photovoltaic response, which is attributed to the narrowed band gap after doping with Ho. That is to say, a certain photovoltaic response peak corresponds to a certain band gap for materials. The UV reflectance spectra of the as-prepared BTFO and BHTFO films are analyzed to assess the band gap, which is shown in Figure 8a. It is shown that the reflectance of Ho-doped BTFO films in visible light region (380−780 nm) is lower than that of BTFO films. It indicates that addition of rare earth Ho3+ ions can enhance the absorption of the visible and ultraviolet light in BHTFO films. Just as shown in Figure 8a, the reflectance edge of BHTFO films is mainly concentrated on the visible light region comparing with BTFO films. The band gap can be derived from the Tauc equation, (αhν)n = B (hν − Eg), where α, B, hν, and Eg denote the absorption coefficient, proportionality constant, photon energy, and band gap energy, respectively. n equals either 1/2 for an indirect transition, or 2 for a direct transition, which is 1/2 for BTFO and BHTFO films. Thus, Eg values of BTFO and BHTFO films are estimated from the tangent line in the plot of (ahv)2 versus hv as shown in Figure 8b. By extrapolation of the linear portion of the curve to intersect with the x-axis, the band gap of BFTO films may be estimated to be 3.01 eV and for BHTFO films to be 2.47 eV. The narrower band gap of BHTTO films is obtained, which is attributed to the rearrangement of molecular orbitals and induced distortion in the (Ti, Fe)O6 octahedron after doping with Ho3+. The strong symmetry structure is destroyed, which results in the band energy and angle change

photogenerated carriers.21 During electrical cycling, a new fatigue mechanism for ferroelectric photovoltaic effect is described, showing an increase of the internal potential energy and local phase decomposition derived from electrical cycling. The local phase decomposition weakens the screening effect and increases the electrode interfacial energy barriers, which results in the charge separation at the domain walls such as accumulating electrons (holes) on one side and depleting the other carriers. This local reduction of the recombination rate makes the walls act as current sources.22 As a result, carrier distributions do not change with applied voltage and the bands remain rigid. Therefore, it shows a linear J−V character for BTFO films after bipolar electrical cycling. So the pure BTFO films just present like a resistance device without any photoelectric response. While for BHTFO films, antifatigue effect of photovoltaic effect is enhanced. Undergoing the same number of cycles, for the noncentrosymmetric crystal BHTFO films, the transition probability that a electron jumps from the state with momentum of k to the state with momentum of k′ is different from the corresponding probability of the reverse process, which causes an asymmetric momentum distribution of the photogenerated carriers and a steady photocurrent.3 These also can be evidenced by the photovoltaic response spectrum. Figure 7 exhibits the photovoltaic response spectrum curves for BTFO and BHTFO films, including before and after bipolar electrical cycling, respectively. The halfway point of the full width at half-maximum (HFWHM) is used to express the best spectral response of the short-circuit current. Figure 7a describes the spectral response of BTFO films over a wavelength range of incident light from 300 to 1100 nm. It is shown that the response peak has a significant shift toward the ultraviolet region from 470 to 430 nm after bipolar electrical cycling. Thus, photovoltaic spectral response is attributed to the depression of the spontaneously polarization. Undergoing bipolar electrical cycling, the carriers need more energy to arrive the surface electrode due to the internal potential barrier induced by the dead layers and local phase decompose. So just the photons with high energy in the ultraviolet region can be absorbed to generate electron−hole pairs, which leads to the fact that photovoltaic spectral responses of the photocurrent of BTFO films have a noteworthy blue-shift behavior from the visible region. So the photovoltaic effect is depressed after bipolar electrical cycling. G

DOI: 10.1021/acs.jpcc.6b07927 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 9. Schematic illustration of energy band and internal depolarization electric fields distribution in BTFO and BHTFO films: (a) the revised band structure induced by remnant piezopotential when the electrodes are metal; (b) the original situation; (c) the regulation of illumination of xenon lamp; (d) the regulation of bipolar electrical cycling; (e) the superimposed effect of illumination and bipolar cycling.

depolarization field at the same external electric field. The inserted impurity level lowers the band gap. Under illumination, a wider energy range region of photons are absorbed and the excited electrons (holes) are separated to drift toward either formed side. That is to say, the stronger depolarization field accelerates the movement of the carriers. Finally, stabler and larger ferroelectric photocurrents are observed for BHTFO films. Since the ferroelectric photovoltaic effect originates from the electron−hole separation at ferroelectric domain walls, this process is very efficient with internal quantum efficiency.13 Therefore, the depolarization field plays a key role in the electron−hole separation under the ferroelectric photovoltaic effect. After undergoing electrical cycling, the depolarization field is degraded for BTFO and BHTFO films. Therefore, a big potential barrier is formed at the interface between the dead layer and original layer due to the switching of the domain walls. The high potential barrier hinders the move of holes (electrons), which leads to a decreased photovoltaic effect for the pure BTFO films. The antifatigue properties are improved after doping with Ho. It forms a relative lower potential barrier than the pure BTFO films at the interface of dead layer, just as showed in Figure 9d. The completely separated electrons (holes) with kinetic energy can overcome the potential barrier, finally forming a stable open voltaic and short current. When given xenon lamp radiation, the ferroelectric properties of the fatigued films are rebuilt by transforming the dead layers to the active layers; thus, the ferroelectric photovoltaic effect is recovered from the fatigue. Figure 9e detailedly describes the recovery process of Ho-doped BTFO films, including the potential barrier decreased at the dead layer interface. At the mean time, the increased depolarization field enhances the separation of the photo induced electrons (holes). In a word, it is difficult to observe the separation and mobility of photoinduced carriers due to the decreased Edp field and reconstructed interfacial potential barrier caused by the electric fatigue behavior regulated by the bipolar electrical cycling, which is directly confirmed by the decreased values of JSC and VOC. However, these situations can be improved after doping with Ho because the thin dead layers are activated by absorbing the major part of visible light. This is attributed to the fact that the imported impurity level changes the band structure. Besides, the enhanced antifatigue properties for BHTFO films can induce a low potential barrier undergoing

after doping with Ho. Just these changes induce Ti 3d orbital to be intensely hybridized with the O 2p orbital below the Fermi level.23 Therefore, narrower band gap is obtained after doping with Ho, which is consistent with the photovoltaic response spectrum and improved photovoltaic output. In order to understand the photovoltaic behaviors regulated by Ho doping and bipolar electrical cycling, the schematic energy band of a typical unit is drawn as well as the corresponding illustration of the photoinduced carriers’ emission process under xenon lamp irradiation, as shown in Figure 9. The band structure diagram based on the different work function is able to consistently explain various ferroelectric photovoltaic behaviors. When it builds the band energy structure, the remnant piezopotential is a steady-state effect (as long as the strain is held) since BTFO films are typical piezoelectric materials. Thus, it will make a constant influence on the band structure.24 For the case of a metal−piezoelectric material (Pt-BTFO) interface, a Schottky contact is presented. The remnant piezopotential caused by screening charges in the metal side only exists in a very narrow region (