Polarization imprint effects on the photovoltaic effect in Pb(Zr,Ti)O
3thin films
ZhengweiTan,1,2JunjiangTian,1,3ZhenFan,1,2,a)ZengxingLu,4LuyongZhang,1 DongfengZheng,1YadongWang,1DeyangChen,1,2MinghuiQin,1,3MinZeng,1,3 XubingLu,1,3XingsenGao,1,3and Jun-MingLiu1,3,4
1Institute for Advanced Materials, South China Academy of Advanced Optoelectronics, South China Normal University, Guangzhou 510006, China
2Guangdong Provincial Key Laboratory of Optical Information Materials and Technology, South China Academy of Advanced Optoelectronics, South China Normal University, Guangzhou 510006, China
3Guangdong Provincial Key Laboratory of Quantum Engineering and Quantum Materials, South China Academy of Advanced Optoelectronics, South China Normal University, Guangzhou 510006, China
4Laboratory of Solid State Microstructures and Innovation Center of Advanced Microstructures, Nanjing University, Nanjing 210093, China
(Received 26 December 2017; accepted 1 April 2018; published online 12 April 2018)
The polarization imprint along with the photovoltaic (PV) effect has been studied in Pt/
Pb(Zr0.3Ti0.7)O3/SrRuO3ferroelectric capacitors. It is shown that the positive DC poling induces the imprint with a downward direction whereas the negative DC poling suppresses the imprint (i.e., rejuvenation). In the polarization up state, the imprinted capacitor exhibits degraded PV properties compared with the rejuvenated one. This may be because the imprint reduces the number of upward domains, thus lowering the driving force for the PV effect. In the polarization down state, however, the rejuvenated capacitor enters the imprinted state spontaneously. This rejuvenation-to-imprint transition can be further aggravated by applying positive voltages and ultraviolet illumination. It is proposed that the domain pinning/depinning, which are associated with the oxygen vacancies and trapped electrons modulated by polarization, voltage, and illumination, may be responsible for the polarization imprint and rejuvenation. Our study therefore sheds light on the correlation between the polarization imprint and the PV effect in the ferroelectrics and also provides some viable suggestions to address the imprint-induced degradation of PV performance.Published by AIP Publishing.
https://doi.org/10.1063/1.5020694
Recent years have witnessed a tremendous flurry of research interest in the ferroelectric photovoltaic (PV) effect that exhibits many fascinating features,1–4including above- bandgap photovoltage5and switchable photocurrent.6,7Thus, the ferroelectrics hold great promise for optoelectronic appli- cations, such as solar cells, photo-sensors, and electrically written/optically read memories. It is known that many fac- tors, such as polarization,6,8 domain structures,9 domain walls,10,11 and interface Schottky barriers,12,13 are playing roles in the ferroelectric PV effect. Among them, the polari- zation is of much concern because it can determine the sign and magnitude of photocurrent,8a characteristic that sepa- rates the ferroelectric PVs from the conventional PVs.
Ideally, the polarization can be reversibly switched between two energetically equivalent, opposite states. However, the degeneracy or bistability of the polarization states may be absent in realistic ferroelectric capacitors. How this issue influences the PV effect deserves much attention.
As the polarization loses its bistability, one polarization direction will be favored over the other. This phenomenon is known as the imprint. The imprint is generally caused by the internal bias fields (Eimp) with different origins, such as asymmetric ferroelectric/electrode interface barriers,14 non- uniformly distributed space charges,15and stress gradients.16 On the one hand, theEimptogether with the depolarization field (Edep; stemming from the unscreened polarization)
provides the driving force for the separation of photo- generated charge carriers.17,18 On the other hand, the PV effect often leads to a redistribution of space charges, because the charge migration, trapping, and re-emission can all occur during the PV process.19,20Consequently, theEimp
will be present (or changed), driving the polarization towards a preferential direction (i.e., the imprint). This in turn results in the variation of Edep and further modifies the PV effect.
As seen above, there is a subtle correlation between the imprint and the PV effect. Studying this correlation may thus provide insights into the interplay between the photocurrent, polarization, charged defects, and electric fields.
Some recent studies already noted that the imprint with a staticEimpcould cause non-switchable PV behavior.18,21,22 However, much less attention has been paid to the induction and modulation of Eimp during the PV measurement and its consequent effects on the PV performance. In addition, to ensure the reliability of ferroelectric PV devices, it is indispensable to address the possible degradation of PV per- formance arising from the imprint. Therefore, a study con- cerning both the polarization imprint and the PV effect is demanded.
Herein, we have concurrently investigated the imprint and PV behaviors of the Pt/Pb(Zr0.3Ti0.7)O3(PZT)/SrRuO3 (SRO) ferroelectric capacitors. The PZT has been chosen for its robust ferroelectricity and strong ultraviolet (UV) photo- response.8It will be clearly shown that the PV effect is influ- enced by the imprint, while the imprint can be formed and modulated owing to the PV effect. The microscopic
a)Author to whom correspondence should be addressed: fanzhen@
m.scnu.edu.cn
0003-6951/2018/112(15)/152905/5/$30.00 112, 152905-1 Published by AIP Publishing.
respectively, while the oxygen pressure was fixed at 15 Pa. The circular Pt electrodes of200lm in diameter and10 nm in thickness wereex situdeposited on the PZT films by PLD at room temperature and high vacuum. The crystal structures were characterized by X-ray diffraction (XRD) and reciprocal space mapping (RSM) (PANalytical X’ Pert PRO). Atomic force microscopy (AFM) and piezoresponse force microscopy (PFM) were performed using a commercial scanning probe microscope (Cypher, Asylum Research). The polarization- voltage (P-V) hysteresis loops and the DC current-voltage (I-V) characteristics were measured with a Radiant ferroelectric workstation and a Keithley 6430 SourceMeter, respectively.
An UV light-emitting diode (LED) with the wavelength of 365 nm and the intensity of 52.3 mW/cm2was used as the light source. The light wavelength is approaching the bandgap of PZT (Fig. S1 in thesupplementary material). Additionally, the light intensity after transmission through the Pt electrode (trans- mittance:34%) is34.5 mW/cm2, which is able to induce noticeable PV responses in the PZT films.23
The XRD and RSM results (Fig. S2 in thesupplementary material) show that the PZT film exhibits a tetragonal phase with the lattice constants ofc¼4.13 A˚ and a¼3.99 A˚ . The AFM image (Fig. S3 in thesupplementary material) demon- strates that the PZT film has a relatively flat surface (rough- ness:420 pm). The PFM phase image shows that there is a clear phase contrast of 180 between the two regions poled with67 V [Fig.1(a)], indicating that the domains are switch- able. However, the PFM was not used to study the imprint effect because of the different electrostatic boundary condi- tions between the PFM tests and the macroscopicP-V mea- surements [Figs. S4 and S5 in thesupplementary material].
In the P-V measurements, the standard bipolar wave- forms were used [inset in Fig.1(b)]. The voltage was applied on the Pt top electrode, while the SRO bottom electrode was grounded. To obtain a saturatedP-V loop, a few trial mea- surements with the drive voltages increasing from 1 V to 4 V were performed. First, theP-Vloop of a typical Pt/PZT/SRO capacitor in the as-grown state was measured. Here, the
r
cive voltage (Vc) of2 V. One can further observe a gap in theP-Vloop in the as-grown state, corresponding to a differ- ence of10lC/cm2between the starting -Prand the ending -Pr. This may be a fingerprint of polarization imprint,14sug- gesting that some upward domains rotate to the downward direction during the delay period where the applied voltage is zero [see inset in Fig.1(b)]. Due to this downward imprint, the starting -Pris reduced, giving rise to the gap in theP-V loop. To confirm the existence of imprint and its downward direction, the positive-up negative-down method was used (see results in Fig. S6 in the supplementary material).
However, the conventional method of using the asymmetry of 6Vcto study the imprint is not adopted here, because the measured P-V loop would be shifted along the polarization axis by the software for centering, which may accompany with the shift along the voltage axis and thus deviate the apparent6Vcvalues from the actual coercive voltages.
Figure 1(b) also shows that after the 3 V poling, the P-V loop becomes closed and is shifted along the positive voltage axis, suggesting that the3 V poling suppresses the downward imprint so that the stability of the Pup state is enhanced. Note that the upward imprint is not formed in the 3 V-poled state (see explanations in the supplementary material); therefore, this state is termed as the rejuvenated state. After the þ3 V poling, the P-V loop becomes open again and is shifted back. These results indicate that the posi- tive and negative poling can cause the polarization imprint and rejuvenation, respectively.
Having investigated the imprint, we proceed to study its effects on the PV properties. A Pt/PZT/SRO capacitor was first set in the imprinted state by theþ3 V poling [Fig.1(b)].
Then, the capacitor was applied with 4 V or þ4 V pulses (pulse width: 0.1 ms) to switch the majority polarization upward or downward. Next, theI-Vcurves under illumination were measured for the Pup andPdown states (voltage sweep rate: 0.02 V/s), respectively. After completing the above pro- cedures, the capacitor was set in the rejuvenated state by the 3 V poling [Fig.1(b)], and then, the same PV measurements were repeated.
Figure 2(a) displays the illuminated I-V curves in the Pup and Pdown states for the imprinted and rejuvenated capacitors. It is observed that theI-Vcurve in thePupstate of the imprinted capacitor is closer to the origin than that of the rejuvenated one, indicating the degradation of PV perfor- mance due to the imprint. More specifically, when the capacitor’s state is changed from rejuvenated to imprinted, the open-circuit voltage (VOC) decreases from 1.1 V to 0.94 V and the short-circuit current (ISC) decreases from 8 nA to6.2 nA. It is further found that theISCin thePup
state gradually decreases along with the imprint process [Fig. S7 in the supplementary material]. The reason for the observed imprint-induced PV performance degradation may
FIG. 1. (a) PFM phase images taken after poling the 55lm2region with 7 V and the middle 2.52.5lm2region withþ7 V. (b)P-Vloops of a typical Pt/PZT/SRO capacitor in the as-grown,3 V-poled andþ3 V-poled states (frequency: 10 kHz). The inset shows the schematic of standard bipo- lar waveforms.
be because in thePupstate the imprint causes some upward domains to rotate downward, reducing the driving force for the PV effect.
Another observation from Fig. 2(a) is that the illumi- natedI-Vcurves in thePdownstate of the imprinted and reju- venated capacitors almost overlap. This is understandable if the initial rejuvenated capacitor can easily transform into the imprinted one in this particularPdownstate (to be evidenced later). Additionally, the imprint direction is downward and thus negligible back-switching ofPdownwould occur.
We also recorded the P-V loops after each step of the PV measurements. In Fig. 2(b), one can observe that the closedP-Vloop of the rejuvenated capacitor remains closed after measuring the illuminated I-V curve in the Pup state, but it becomes open and is shifted leftward after the PV mea- surement in thePdownstate. This infers that the rejuvenated capacitor turns into an imprinted one after the PV measure- ment in thePdownstate.
Therefore, to address the imprint-induced PV perfor- mance degradation, some useful suggestions are given: (i) setting the ferroelectric capacitor in the rejuvenated state beforehand; and (ii) operating the PV effect in a particular polarization direction (e.g.,Pupin our case).
Then, we further identify the factors leading to the imprint during the PV measurements. Three factors, i.e., polarization state, UV illumination, and small applied vol- tages, were considered. A rejuvenated capacitor was set in thePuporPdownstate first, and then, three parallel treatments were performed: (i) leaving the capacitor in thePuporPdown
state, in the absence of illumination or applied voltages, for 5 min; (ii) illuminating the capacitor (without applying vol- tages) in thePuporPdownstate for 5 min; and (iii) sweeping small voltages repeatedly on the capacitor (without illumina- tion) for 5 min (0 to 1.1 V for the Pup state while 0 to þ0.7 V for the Pdown state; the maximum applied voltages were the same as those in the PV measurements). As shown in Fig. 3(a), when the rejuvenated capacitor is in the Pup
state, the P-V loop remains closed regardless of applying illumination or negative voltages. However, when the reju- venated capacitor is left in thePdownstate for 5 min, theP-V loop forms a gap and is shifted leftward [Fig. 3(b)]. Even larger gaps and shifts are observed after applying illumina- tion and positive voltages. Such transition from the rejuve- nated state to the imprinted one may be because some downward domains gradually become pinned in the Pdown
state, which is further aggravated by applying illumination or positive voltages.
Now, we propose a microscopic mechanism to explain all the above observations. It is reasonably thought that the oxygen vacancies (OVs) are formed during the growth of PZT films in the oxygen-deficient atmosphere, and the OVs are mainly distributed in the near-surface region owing to the higher rate of oxygen loss in this region.24,25 The posi- tively charged OVs can create an electric field (i.e., Eimp) pointing downward, pinning the domains located beneath the OVs to the downward direction [States (i) and (ii) in Fig.4].
To verify the role of OVs, an imprinted capacitor was annealed in the vacuum at 300C for 1 h and then its P-V loop was measured. The X-ray photoelectron spectroscopy (XPS) results show that the vacuum-annealed PZT’s surface has a higher concentration of OVs [Fig. S8(a) and (b) in the supplementary material]. Moreover, the gap of the P-Vloop becomes larger in the vacuum-annealed capacitor [Fig. S8(c) in the supplementary material], providing evidence for the correlation between the OVs and the imprint. Additionally, the imprint almost disappears in the SRO/PZT/SRO capaci- tors, where the OVs near the top interface may be largely reduced during the growth of SRO top electrodes (Fig. S9 in the supplementary material). This further confirms that the imprint is caused by the OVs near the top interface.
After the negative DC poling, the electron injection and trapping may occur in the near-surface region.26,27Thus, the positive charges of OVs may be compensated by the elec- trons, vanishing the Eimp. In this rejuvenated state [States (iii) and (iv) in Fig.4], all the domains can be switched and therefore the closedP-Vloops are observed [Fig.1(b)]. After the positive DC poling, however, all the domains are switched downward and the electron detrapping occurs. As a result, the downward Eimpowing to the positively charged OVs is formed again, turning the capacitor back into the imprinted state [State (ii) in Fig. 4; see P-V loops in Fig.
1(b)]. Note that the migration of OVs under applied voltages may be neglected; otherwise, the negative (positive) poling would accumulate (deplete) OVs near the Pt/PZT interface, enhancing (suppressing) the imprint, which contradicts our observation.
In thePdownstate of the rejuvenated capacitor [State (iv) in Fig.4], the negative polarization charges repel the trapped electrons due to the Coulombic interactions.19,20 This may cause the electron detrapping, and consequently, the uncom- pensated OVs induce the downward Eimp. Therefore, the
FIG. 2. (a) IlluminatedI-Vcurves of theþ3 V-poled (imprinted; “I”) and 3 V-poled (rejuvenated; “R”) capacitors in thePupandPdownstates. (b) P-V loops of the initial rejuvenated capacitor (“R, initial”) and those obtained after the PV measurements in thePupandPdownstates (“Pup, light”
and “Pdown, light,” respectively).
FIG. 3.P-Vloops measured after different treatments for the rejuvenated capacitor whose initial state is (a)Pupand (b)Pdown. Here, “W/O,” “light,”
and “small6V” represent without applying voltages or illumination, apply- ing illumination, and applying small voltages, respectively (see the main text for details).
rejuvenated capacitor left in the Pdown state will gradually transform to the imprinted one [State (v) in Fig.4; seeP-V loops in Fig.3(b)]. Applying positive voltages and illumina- tion facilitates the electron detrapping, further aggravating the imprint [States (vi) and (vii) in Fig.4; seeP-Vloops in Fig.3(b)]. Particularly, the UV illumination induces the PV process, during which the electrons are detrapped and move towards the bottom electrode under theEdep. In thePupstate [State (iii) in Fig. 4], however, the positive polarization charges tend to attract the electrons, favoring the rejuvenated state.19,20Also due to such interactions, the trapped electrons are largely stable against the small negative voltages and illumination [seeP-Vloops in Fig.3(a)].
Then, we illustrate how the imprint influences the PV effect. Prior to it, by analyzing the conduction mechanisms, we have confirmed that the PV effect is a bulk effect rather than the Schottky barrier effect or the domain wall effect (Fig. S10 in thesupplementary material). Thus, we mainly consider the fields existing in the film bulk as the driving force for the PV effect. In thePupstate, all the domains are oriented upward in the rejuvenated capacitor [State (iii) in Fig. 4], giving rise to a large Edep to drive the PV effect, whereas in the imprinted capacitor [State (i) in Fig.4], some domains are pinned downward due to the Eimp. Therefore, the overall field driving the PV effect, whose direction is downward, is decreased in the imprinted capacitor [Fig.
2(a)]. In thePdownstate, the rejuvenated capacitor [State (iv) in Fig.4] may enter the imprinted state upon illumination [State (vii) in Fig.4]. This leads to almost the same PV prop- erties in thePdownstate [Fig.2(a)].
In summary, both the ferroelectric and PV properties of the Pt/PZT/SRO capacitors were studied. It was observed that the capacitor could become imprinted (rejuvenated) after the DC poling with þ3 V (3 V). In the Pup state, the imprinted capacitor exhibited smaller photovoltage and pho- tocurrent than the rejuvenated one. This may be because the imprint reduces the number of upward domains, thus lower- ing the driving force for the PV effect. In the Pdown state, however, there was a spontaneous transition from the rejuve- nated state to the imprinted one, which could be further aggravated by applying positive voltages and UV illumina- tion. It was proposed that the oxygen vacancies, together
with the trapped electrons modulated by the polarization, voltage, and illumination, may play significant roles in the polarization imprint and rejuvenation. Our study therefore contributes to understanding the correlation between the polarization imprint and the PV effect in the ferroelectrics and also provides some useful suggestions to address the imprint-induced degradation of PV performance.
Seesupplementary material for the absorption spectra, XRD and RSM results, AFM image, PFM amplitude and phase loops, PFM phase images, PUND results, time- dependent photocurrents, XPS spectra and P-V loops of the un-annealed and annealed PZT films, P-V loops of SRO/
PZT/SRO capacitors, and darkI-Vcharacteristics.
The authors thank the National Key Research Program of China (Nos. 2016YFA0201002 and 2016YFA0300101), the State Key Program for Basic Researches of China (No.
2015CB921202), the National Natural Science Foundation of China (Nos. 51602110, 11674108, 51272078, and 51332006), the Science and Technology Planning Project of Guangdong Province (No. 2015B090927006), and the Natural Science Foundation of Guangdong Province (No.
2016A030308019). X. Gao and X. Lu acknowledge the Project for Guangdong Province Universities and Colleges Pearl River Scholar Funded Scheme 2014 and 2016, respectively.
1P. Lopez-Varo, L. Bertoluzzi, J. Bisquert, M. Alexe, M. Coll, J. Huang, J.
A. Jimenez-Tejada, T. Kirchartz, R. Nechachei, F. Rosei, and Y. Yuan, Phys. Rep.653, 1 (2016).
2Z. Fan, K. Sun, and J. Wang,J. Mater. Chem. A3, 18809 (2015).
3J. Li, B. Huang, E. N. Esfahani, L. Wei, J. Yao, J. Zhao, and W. Chen,npj Quantum Mater.2, 56 (2017).
4F.-T. Huang, B. Gao, J.-W. Kim, X. Luo, Y. Wang, M.-W. Chu, C.-K.
Chang, H.-S. Sheu, and S.-W. Cheong, npj Quantum Mater. 1, 16017 (2016).
5K. Yao, B. K. Gan, M. Chen, and S. Shannigrahi,Appl. Phys. lett. 87, 212906 (2005).
6W. Ji, K. Yao, and Y. C. Liang,Adv. Mater.22, 1763 (2010).
7H. T. Yi, T. Choi, S. G. Choi, Y. S. Oh, and S.-W. Cheong,Adv. Mater.
23, 3403 (2011).
8L. Pintilie, I. Vrejoiu, G. L. Rhun, and M. Alexe, J. Appl. Phys.101, 064109 (2007).
states [States (i), (ii), (v), (vi), and (vii)] and the rejuvenated states [States (iii) and (iv)].
9Y. Zhou, L. Fang, L. You, P. Ren, L. Wang, and J. Wang,Appl. Phys.
Lett.105, 252903 (2014).
10S. Y. Yang, J. Seidel, S. J. Byrnes, P. Shafer, C. H. Yang, M. D. Rossell, P. Yu, Y. H. Chu, J. F. Scott, J. W. Ager III, L. W. Martin, and R.
Ramesh,Nat. Nanotechnol.5, 143 (2010).
11A. Bhatnagar, A. R. Chaudhuri, Y. H. Kim, D. Hesse, and M. Alexe,Nat.
Commun.4, 2835 (2013).
12L. Fang, L. You, Y. Zhou, P. Ren, Z. S. Lim, and J. Wang,Appl. Phys.
Lett.104, 142903 (2014).
13S. Y. Yang, L. W. Martin, S. J. Byrnes, T. E. Conry, S. R. Basu, D. Paran, L. Reichertz, J. Ihlefeld, C. Adamo, A. Melville, Y. H. Chu, C. H. Yang, J.
L. Musfeldt, D. G. Schlom, J. W. Ager, and R. Ramesh,Appl. Phys. Lett.
95, 062909 (2009).
14J. Lee, C. H. Choi, B. H. Park, T. W. Noh, and J. K. Lee,Appl. Phys. Lett.
72, 3380 (1998).
15A. K. Tagantsev, I. Stolichnov, N. Setter, and J. S. Cross,J. Appl. Phys.
96, 6616 (2004).
16Y. Zhou, H. K. Chan, C. H. Lam, and F. G. Shin,J. Appl. Phys. 98, 024111 (2005).
17L. Pintilie, V. Stancu, E. Vasile, and I. Pintilie,J. Appl. Phys.107, 114111 (2010).
18F. Liu, I. Fina, D. Gutierrez, G. Radaelli, R. Bertacco, and J. Fontcuberta, Adv. Electron. Mater.1, 1500171 (2015).
19D. Dimos, W. L. Warren, M. B. Sinclair, B. A. Tuttle, and R. W.
Schwartz,J. Appl. Phys.76, 4305 (1994).
20W. L. Warren, D. Dimos, G. E. Pike, B. A. Tuttle, M. V. Raymond, R.
Ramesh, and J. T. Evans,Appl. Phys. Lett.67, 866 (1995).
21H. Fan, C. Chen, Z. Fan, L. Zhang, Z. Tan, P. Li, Z. Huang, J. Yao, G.
Tian, Q. Luo, Z. Li, X. Song, D. Chen, M. Zeng, J. Gao, X. Lu, Y. Zhao, X. Gao, and J.-M. Liu,Appl. Phys. Lett.111, 252901 (2017).
22P. P. Biswas, C. Thirmal, S. Pal, and P. Murugavel,J. Appl. Phys.123, 024101 (2018).
23M. Qin, K. Yao, Y. C. Liang, and S. Shannigrahi,J. Appl. Phys. 101, 014104 (2007).
24G. L. Yuan, L. W. Martin, R. Ramesh, and A. Uedono,Appl. Phys. Lett.
95, 012904 (2009).
25K. Szot and W. Speier,Phys. Rev. B60, 5909 (1999).
26Z. Fan, H. Fan, Z. Lu, P. Li, Z. Huang, G. Tian, L. Yang, J. Yao, C. Chen, D.
Chen, Z. Yan, X. Lu, X. Gao, and J.-M. Liu,Phys. Rev. Appl.7, 014020 (2017).
27P. Li, Z. Huang, Z. Fan, H. Fan, Q. Luo, C. Chen, D. Chen, M. Zeng, M.
Qin, Z. Zhang, X. Lu, X. Gao, and J.-M. Liu,ACS Appl. Mater. Interfaces 9, 27120 (2017).
