SEARCH

SEARCH BY CITATION

Keywords:

  • ferroelectric photovoltaics;
  • mutiferroics;
  • oxide photonics;
  • photodetectors

Leaky ferroelectric oxides can serve as optoelectronic circuit elements in which potential gradients due to the spontaneous polarization yield asymmetric and nonlinear photocarrier dynamics. Ferroelectric domains and domain walls should each influence photocarrier generation, separation, and transport differently, but the microscopic mechanisms are unknown. Here, we present scanning photocurrent images of epitaxial BiFeO3 thin films that reveal how the photoresponse depends on dynamic domain configurations. Locally, photocurrent direction is determined by local domain orientation, whereas the photocurrent magnitude is spectrally centered around charged domain walls associated with oxygen vacancy migration. Our observations demonstrate that photodetection can be electrically controlled by manipulating domains, suggesting non-volatile optoelectronic memory applications.

In non-centrosymmetric media, nonequilibrium photocarriers can be directed along polar directions. This yields a unidirectional photocurrent (Iph) parallel to the spontaneous polarization (PS), and thus high photovoltages (Vph)1–6 that are linearly proportional to channel length.4 Photovoltaic effects in ferroelectric oxides were previously investigated back in the 1960s,1, 2, 7 and a high Vph accompanied by nonlinear and unidirectional Iph have been recently observed in a small bandgap ferroelectric BiFeO3 (BFO).3, 4 This material is the only room–temperature multiferroic,8–10 and represents an interesting platform for investigating photon interactions with internal ferroic order.11–14 For example, bulk photo-diode effects3, 15 and photo-induced mechanical striction16 were recently observed in BFO monodomains with a switchable ferroelectric polarization. Meanwhile, in polydomain samples, it was also found that the sign and the magnitude of Iph and Vph are predominantly determined by domain walls (DWs), particularly when multiple DWs are regularly spaced.4 Polarization switching in rhombohedral BFO is closely associated with magnetoelectric coupling, and is manifested in ferroelastic switching associated with 71° or 109° DWs. The spatial extent of ferroic order induces changes in polarization, and must thus impose a local variation in the electrical potential and bandgap that govern the photoresponses. Nevertheless, the interplay between these internal states and the local photoresponse has not been fully addressed to date. Here we report a vectorial mapping of the local Iph, scanned over the static domain configuration of BFO thin films, manipulated by an applied E-field. By the spatially- and spectrally-resolved photoresponses of (001)-oriented BFO films, it was revealed that photodetection is determined by the local PS. Specifically, we captured direct images of the evolution of the local Iph on switching domains via ‘charged’ domain-wall motion,17–19 with the magnitude of Iph being enhanced across a reduced band gap.

The device structure of our BFO photodetector is shown in Figure1, where a pair of coplanar electrodes are patterned on (001) epitaxial BFO films that are 80 nm thick. The rectangular device channel which is 30 μm wide is defined via the 4 μm electrode gap. Note that we use the pseudocubic notation for rhombohedral BFO films. We performed scanning photocurrent microscopy (SPM)20, 21 by measuring short-circuit currents (IscIph at 0 V) while a 532 nm ( = 2.33 eV) laser beam of a given power is rastered over the area containing the BFO channel. As illustrated in Figure 1, the chopped laser beam is focused by a lens such that it illuminates the device which is placed on a piezo-scanner. A lock-in current measurement that is relayed with a 10 kHz chopper records Isc. Our diffraction-limited optics provide a laser spot of diameter ∼500 nm. This scanning optical excitation locally generates photocarriers via an interband transition at a given spot. They are subsequently transported toward the electrodes by drift-diffusion prior to recombination, and are finally collected as output signal Iph. Therefore, the spatially resolved Iph measurements, particularly in the zero external voltage, provide information about the local energy bandgap (Eg) and the inherent potential variation within the channel, which are in turn closely associated with the spatial distribution of PS along the current pathway.

thumbnail image

Figure 1. The schematic of the spatially-resolved photodetection of the BFO thin film. A diffraction-limited laser (∼500nm) is raster-scanned over the area containing device channel while the local Iph is measured through lock-in detection. The upper left inset shows an optical microscope image of the rectangular device channel of BFO photodetector with electrode-gap and width of 4 μm and the 30 μm, respectively.

Download figure to PowerPoint

Figures2a and Figures 2f show Isc-SPM images of the BFO channel after statically poling at 90 V for a few minutes in the [1–00] and [100] direction, respectively. The two-dimensional color contrast between a pair of electrodes signifies the spatial variation of the direction and magnitude of Isc. As expected, the major direction of Isc is immediately reversed in response to applied field E. This is also well presented in the Iph-V characteristics of the channel, as seen in Figure 2c and Figures 2h. We paid our particular attentions to the fact that the spatial distribution of the Isc amplitude within the channel is graded along the E-field direction in a unique manner, as also seen in the line profile across the channel in Figure 2b and Figures 2g. It is highest near the electrode that was negatively biased during the poling, but rapidly falls off at the contact, from both poling bias directions. In parallel, the corresponding in-plane PFM phase images, taken after the ±90 V poling, are shown in Figure 2d and Figures 2i, and they display the stripe domains mostly lying along the <010> direction. Out-of plane PFM measurements confirmed that the out-of plane components are pointing downward that is invariant upon the in-plane field application, i.e. from the surface to the substrate, as in Supplementary Information. Although the majority domains are switched in response to the applied E-field, we found the persistent presence of pinned domains near the (-) electrodes, regardless of the E-field directions, the poling conditions. These pinned Pnet are confronted with the switched Pnet, forming a saw-teeth patterned heads-on boundary between the two opposite Pnet's. Here, we emphasize that the relative location and spatial extent of such boundaries within the channel, identified in the PFM images, coincide with those of the higher Isc amplitude in the Isc-SPM images. Such areas, enclosed by the green dotted lines, are schematically expanded in Figure 2e and Figures 2j, and they typically are sub-divided into either 71° or 109° individual DWs22, 23 - hereafter we call this polar boundary, across which two opposite +Pnet/-Pnet vectors are facing by 180°, the global DW (GDW), as opposed to the local DW (LDW) at which two individual PS are interfacing. For example, the larger 180° GDWs is a bundles of numerous smaller 71° or 109° ferroelastic twinned LDWs.22 A closer inspection, based on the PFM images, classifies such GDWs into LDWs by the relative angles between the two interfacing PS vectors along with their statistical occurrences, as in Figure3a. It is particularly interesting to note the first case, where two adjacent electric dipoles are ‘head-to-head’ and ‘head-to-side’ across either LDWs. These head-to-head or head-to-side configurations are electrostatically unfavorable, resulting in a nominally ‘charged’ LDWs.17, 18, 19 The second and usual kinds of the neutral or uncharged LDWs are also observed. We find the areal fraction between the charged and uncharged LDWs is almost equal along the GDWs.

thumbnail image

Figure 2. Spatially-resolved, switchable photodetection. (a, f) The scanning photocurrent images of the BFO thin film under the short-circuit condition after the 90V electric poling in the [1-00] and [100] directions, respectively. (b, g) The Isc line profiles along the dotted lines in (a) and (f) (c, h) The corresponding Iph- V characteristics under the global illumination without using the lock in technique. (d, i) The in-plane PFM images under the same poling conditions as in (a) and (f). (e, j) Schematically expanded domain structures within the dotted lines of (d) and (i).

Download figure to PowerPoint

thumbnail image

Figure 3. Types of domain walls upon the domain switching and the spectrally resolved local photoresponses. (a) Classifications and occurrences of the observed domain walls according to the angle (71° and 109°) between the two polarization variants across the local domain wall and to the nominal charge neutrality. (b) The spatially-resolved, spectral photoresponses, taken from the different locations along the device channel. The pronounced enhancement in the Isc in the sub-bandgap is observed near the global domain wall and the (-) biased electrode, with the nearly absent feature near the (+) biased electrode.

Download figure to PowerPoint

In order to investigate the origin of the local Isc enhancement, we have taken the spectral response of the Isc while the supercontinuum laser was focused on different positions along the BFO channel, as shown in Figure 3b. They commonly show a strong upturn at 2.6–2.7 eV, which is consistent with optical absorption across the direct bandgap of BFO at 300 K.5, 24, 25 The spectral onset, however, varies at different locations, and at the GDWs it reads 2.0 eV with its shoulder peak at 2.3 eV. The similar, but less pronounced, shoulder peak is also seen near the (-) electrode, whereas it is nearly absent near the (+) biased electrode. This sub-bandgap feature in BFO, or generally in perovskite oxides, commonly resorts to the presence of oxygen vacancies (VO2+), and these VO2+ defect states are known to form energy levels below the conduction band edge of BFO by 0.5–0.7 eV, thus effectively lowering the energy bandgap.26, 27 Thus we discuss our principal observations within the framework of (1) the local bandgap reduction at the GDW that is spatially reconfigured upon the domain switching and (2) the associated spatial distributions of the mobile VO2+ charges, as follows in Figure4.

thumbnail image

Figure 4. Polarization vector configurable photodetection. (a, e, i) The in-plane piezoresponce force microscope images of BFO thin film after the +90V, –45V, –90V poling in the [1-00] direction, respectively, showing the GDW motion within the channels in a progressive manner. (b, f, j) The corresponding scanning short circuit photocurrent images under the same poling conditions as in (a, e, i). (c, g, k) The proposed band diagram in each poling condition. (d, h, l) The schematic domain configuration with oxygen vacancy distributions.

Download figure to PowerPoint

First, we assume that the persistent shoulder peaks at 2.3 eV near the GDWs and the (-) electrode of Figure 3b suggest that, during the +90V poling sequence, these positive VO2+ charges are activated to migrate toward the (-) biased electrode, and subsequently to accumulate near the GDWs, as illustrated in Figure 4a–d. Drift of these VO2+ charges in perovskite oxide ferroelectrics under the static E-field has been directly observed by electrocoloration.28, 29 This positive charge accumulation can locally screen the applied E-field near the (-) biased electrodes during the domain switching, and provide an electrostatic driving force to impede coherent domain switching, resulting in the local domain pinning. This local domain pinning gives rise to the local potential minimum at the GDW, as in Figure 4c. Extended crystal defects, such as twin DWs, grain boundaries and dislocations, are known to serve as the local occupancy sites for the mobile VO2+, as directly observed from various perovskite oxide ferroelectrics.30–38 For example, the VO2+ charges in BaTiO3 are preferentially incorporated in twin walls by modifying the local cation valences and displacements.31 To date, the charged DW in BFO has not been directly observed, and they have only been theoretically predicted to exhibit metallic electricity.17 It was predicted that these charged DWs can be stabilized by the cation valency exchanges that are provided by the local incorporation of foreign elements or self-doping by oxygen vacancies,17,18 allowing the local charge compensation at the charged DWs. In our observations, we suggest that the locally enhanced Isc at the GDW indicates that these VO2+ are accumulated at the GDW, effectively lowering the local energy bandgap, to yield the significantly higher photocarrier generation rate. These non-equilibrium photocarriers are transported to the electrode along the potential slopes, set by the Pnet vector, by drift-diffusion, as in Figure 4c. Here the local potential variation is inherently configured by vectorially manipulating the Pnet vectors with the applied E-field: the Isc direction is determined by the relative strength between the two opposite Pnet vectors within the channel, and the Isc magnitude is modulated by the photocarrier generation rate across the local bandgap. Upon the -90 V poling, the essentially same features with the +90 V poling case, but in a reversibly hysteretic manner, are observed, as evident in Figure 4i-l.

In summary, we demonstrate photodetection characteristics in leaky ferroelectric BFO thin films, where the spatial configuration of the internal order is vectorially coupled with light to produce spatially resolved Iph and Vph. The spontaneous polarization in ferroelectric materials, can serve as a unique optoelectronic memory element, since it is inherently imprinted and reconfigurable with the applied E-field. In our study, we show such photodetections on the spatially configurable Pnet vectors using a focused light source of a few hundred nanometers in size. We foresee that the use of near-field optics can bring such probing length scales down to an individual polarization or domain on the nanometer scale.39, 40, 41

Experimental Section

  1. Top of page
  2. Experimental Section
  3. Supporting Information
  4. Acknowledgements
  5. Supporting Information

Scanning Photocurrent Measurements: Spatially-resolved scanning Iph mapping of BFO (001) thin film was performed using a focused 532 nm laser in diameter of 500 nm. While continuously varying the in-plane position of the BFO channel using nm scale-controllable piezo scanner (PI instruments), the Isc is lock-in measured as a function of the laser illumination position in the excitation frequency 10 kHz, relayed by the optical chopper placed between the laser source and the BFO channel.

Piezoresponse Force Microscopy: Nanoscale pieozoelectric domain measurements were performed using a Digital Instruments Nanoscope-V Multimode scanning probe microscopy. Commercially available Ti/Pt coated Si tips (MikroMasch) were used with a typical scan rate of ∼ 10μm sec−1. The tip was oriented along the <110> direction to discriminate sensitively the four possible in-plane polarization. All data were acquired under ambient conditions at room temperature. In addition, we took PFM images rotating azimuthally by 90° in order to confirm the in-plane polarization.

Supporting Information

  1. Top of page
  2. Experimental Section
  3. Supporting Information
  4. Acknowledgements
  5. Supporting Information

Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements

  1. Top of page
  2. Experimental Section
  3. Supporting Information
  4. Acknowledgements
  5. Supporting Information

We thank J. F. Scott at the University of Cambridge for general comments on the manuscript. M.H.J. acknowledges the support from the Basic Science Research Program through the NRF (2010-0017853), the Nano Original & Fundamental Technology R&D Program (2010-0019195), the Priority Research Centers Program through the NRF (2010-0029711), the MEST-AFOSR NBIT (No. K20716000006-07A0400-00610), and the WCU program (R31-2008-000-10059-0). C.H.Y. acknowledge the support by the NRF (2010-0013528).

Supporting Information

  1. Top of page
  2. Experimental Section
  3. Supporting Information
  4. Acknowledgements
  5. Supporting Information

Detailed facts of importance to specialist readers are published as ”Supporting Information”. Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors.

FilenameFormatSizeDescription
adma_201102816_sm_suppl.pdf569Ksuppl

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.