⇦Chinese Physics Letters, 2017, Vol. 34, No. 2, Article code 027701 Improved Polarization Retention of BiFeO$_{3}$ Thin Films Using GdScO$_{3}$ (110) Substrates * Shuai-Qi Xu(许帅骑), Yan Zhang(张岩), Hui-Zhen Guo(郭慧珍), Wen-Ping Geng(耿文平), Zi-Long Bai(白子龙), An-Quan Jiang(江安全)** Affiliations State Key Laboratory of ASIC and System, School of Microelectronics, Fudan University, Shanghai 200433 Received 28 October 2016 *Supported by the National Basic Research Program of China under Grant No 2014CB921004, and the National Natural Science Foundation of China under Grant No 61225020.
**Corresponding author. Email: aqjiang@fudan.edu.cn Citation Text: Xu S Q, Zhang Y, Guo H Z, Geng W P and Bai Z L et al 2017 Chin. Phys. Lett. 34 027701 Abstract Epitaxial ferroelectric thin films on single-crystal substrates generally show a preferred domain orientation in one direction over the other in demonstration of a poor polarization retention. This behavior will affect their application in nonvolatile ferroelectric random access memories where bipolar polarization states are used to store the logic 0 and 1 data. Here the retention characteristics of BiFeO$_{3}$ thin films with SrRuO$_{3}$ bottom electrodes on both GdScO$_{3}$ (110) and SrTiO$_{3}$ (100) substrates are studied and compared, and the results of piezoresponse force microscopy provide a long time retention property of the films on two substrates. It is found that bismuth ferrite thin films grown on GdScO$_{3}$ substrates show no preferred domain variants in comparison with the preferred downward polarization orientation toward bottom electrodes on SrTiO$_{3}$ substrates. The retention test from a positive-up domain to a negative-down domain using a signal generator and an oscilloscope coincidentally shows bistable polarization states on the GdScO$_{3}$ substrate over a measuring time of 500 s, unlike the preferred domain orientation on SrTiO$_{3}$, where more than 65% of upward domains disappear after 1 s. In addition, different sizes of domains have been written and read by using the scanning tip of piezoresponse force microscopy, where the polarization can stabilize over one month. This study paves one route to improve the polarization retention property through the optimization of the lattice-mismatched stresses between films and substrates. DOI:10.1088/0256-307X/34/2/027701 PACS:77.84.-s, 77.80.Dj, 77.55.-g © 2017 Chinese Physics Society Article Text With both large spontaneous ferroelectric polarization and antiferromagnetic ordering at room temperature, multiferroic BiFeO$_{3}$ (BFO) materials which have good ferroelectric fatigue endurance attract a great deal of attention,[1-3] especially for epitaxial thin films that usually show excellent electrical properties than for polycrystalline thin films. Moreover, integration of such thin films on silicon substrates is applicable to fabricate nanoscale devices, such as to the nonvolatile memories.[4-6] However, BFO thin films on many substrates show a poor retention property, because they always have a preferred polarization direction. That is, it shows a good retention on one polarization direction, while it worsens on the opposite direction.[7] The preferred polarization direction is thought to be caused by the lattice mismatch between BFO films and the substrates. Also, the frozen compensation charges may affect the film's short time retention,[4] which affects the reliability of data storage devices, and thus the work to improve the BFO retention is meaningful. To improve the BFO retention, the GdScO$_{3}$ (GSO) substrate in orthorhombic (110) orientation which is equivalent to the (001) orientation of a pseudocubic symmetry was chosen to reduce the lattice mismatch between the BFO film and the substrate. The lattice misfit between the substrate GSO and BFO is merely 0.2%, but 1.0 % with (001) SrRuO$_{3}$.[8] Therefore, the SrRuO$_{3}$ (SRO) bottom electrode must be thin enough to permit the epitaxial growth of BFO on it.[9] Both BFO and SRO films were grown by pulsed laser deposition (PLD) using a KrF excimer laser (Coherent Compex Pro 201 F, USA) with the wavelength of 248 nm and the repetition rate of 6 Hz. The 20-nm-thick SRO bottom electrode was grown at 620$^{\circ}\!$C under the oxygen pressure of 8 Pa. In comparison, the BFO was grown at 630$^{\circ}\!$C under the oxygen pressure of 10 Pa for 150 nm thickness. Lastly, a very thin gold electrode layer through a shallow mask was sputtered on the surface of BFO for the measurement of polarization. The crystal structure of the film was determined by x-ray diffraction (XRD, Bruker D8 Advance with Cu-K$\alpha$ radiation). The topography of the film was measured with an atomic force microscope (AFM, Bruker Icon) in the tapping mode. We also employed a transmission electron microscope (TEM, FEI TECNAI G2 F20 S-TWIN) to obtain a vivid microscopic structure of our film and to further confirm the quality of our film. With the short-pulse characterization technology that transfers domain switching currents into polarization-electric ($P$–$E$) hysteresis loops, the switched polarization proportion can be obtained by charge integration over time.[10] The switching currents can be measured by an in-series oscilloscope with voltage supplied by a pulse generator, where the film's leakage current can be subtracted for the leaky films. Finally, the piezoresponse force microscopy (PFM)[11] was used to write and image domains over different film areas, and domain retention was measured for more than one week.
cpl-34-2-027701-fig1.png
Fig. 1. (a) X-ray diffraction patterns of the GSO substrate, the SRO/GSO film and the BFO/SRO/GSO film, (b) AFM surface morphology of the 20 nm SRO/GSO film (roughness 0.182 nm), and (c) AFM surface morphology of the 350 nm BFO/SRO/GSO film (roughness 0.365 nm).
cpl-34-2-027701-fig2.png
Fig. 2. (a) Low-magnification TEM image of the BFO/SRO/GSO film with the protection Cr layer. (b) HRTEM image of the BFO/STO interface from the region indicated by the yellow dotted rectangle shown in (a). (c, d) Diffraction irradiated by two electron beams along the direction vertical to the sample's grown direction.
Figure 1(a) gives the 2$\theta$-scan XRD patterns of the GSO substrate, SRO/GSO and BFO/SRO/GSO films. The BFO (001) diffraction peak overlaps the GSO (110) reflection. From these patterns, we find that the BFO thin film exhibits a pure perovskite phase without any impurity. Figures 1(b) and 1(c) provide the surface topographies of the SRO and BFO/SRO on GSO substrates. The roughness of either SRO or BFO is within a few nanometers. These verified the perfect epitaxial lattice matching between BFO and GSO and the high crystallinity of the films. Figure 2(a) shows the cross-sectional TEM image of the BFO/SRO/GSO film. The enlarged image of the interface of BFO/SRO in Fig. 2(b) shows the good epitaxial nature of BFO on the SRO/GSO substrate. From the zone axis of each SEAD pattern in Figs. 2(c) and 2(d), we can make sure that the orientation of the BFO grown on SRO/GSO substrates is (001) and we can also confirm the result of XRD patterns.
cpl-34-2-027701-fig3.png
Fig. 3. (a) Initial domain switching current with and without the leakage current in it. Inset: the voltage dependence of the film's leakage current. (b) Pure $P$–$V$ hysteresis loop of the BFO thin film.
Initial domain switching currents with and without the leakage current are shown in Fig. 3(a). The inset shows the film's leakage current dependence of applied voltage. Each leakage current was measured after the presetting pulse of 8 V with the duration time of 2 μs. The solid line gives the polynomial algorithm simulation of the asymmetric nonlinear current dependence, which will be used for the leakage current subtraction from the total domain switching current for the derivation of a $P_{\rm f}$–$V_{\rm f}$ hysteresis loop. The leakage current grows quickly when the voltage increases, as shown in the current-voltage dependence figure, and the total domain switching current is the mixture of the film leaky current and the ionic displacement current, as shown in Fig. 3(a). The solid line shows the total positive/negative domain switching currents mixed with leakage currents. After deduction of the leakage currents using the above polynomial algorithm formulization, the pure ionic displacement current transient with time was obtained, as shown by the dotted line. After this leaky current subtraction, the trail of negative current shown by the dotted line approaches zero unlike the solid line. Finally, the domain switching current without leakage current was integrated over time from which we derived the accurate $P_{\rm f}$–$V_{\rm f}$ hysteresis loop, as shown in Fig. 3(b). In addition to the leakage current subtraction discussed above, we also debated the voltage drop across the contact resistance between the film and electrode, thus the loop faithfully shows the inner domain switching characteristics of the BFO layer only. With this technique, we measured the retention of the BFO films on both STO and GSO substrates over a time of 500 s, as shown in Fig. 4. The downward domain of the BFO film on STO substrates is unchanged after writing. However, upward domains lost nearly 60% polarization after 0.1 s. In contrast, both domains on GSO substrates kept unchanged over our testing time of 500 s.[12] From this study, it is obtained that BFO films grown on STO substrates in large lattice mismatch suffer from a serious retention problem for the upward polarization degradation.[13] However, the BFO films grown on GSO (110) substrates with SRO bottom electrode can overcome this problem, where the lattice mismatch between BFO and GSO is largely reduced. This improvement is helpful for the improvement of BFO retention for the memory application.
cpl-34-2-027701-fig4.png
Fig. 4. (a) The relaxation-time dependences of downward polarization for BFO thin films after a presenting downward voltage pulse. Inset: the relaxation-time dependences of positive switchable polarization for the bismuth ferrite thin film after different cycles of negative and positive square voltage. (b) The relaxation-time dependences of upward polarization after a presenting upward voltage pulse.
In Fig. 4(b), the upward polarization of the BFO/GSO film though has a large retention improvement drops slightly at a characteristic time of 10$^{-3}$ s. After that, the polarization is stable with time. In comparison, the retention time for a BFO/STO thin film is shorter than 1 s. In addition to the retention improvement through the adjustment of the lattice mismatch between the film and substrates, the upward retention property of the BFO/GSO film can be further increased by the bipolar cycles of the sample under positive and negative square-pulse voltages. The effect of the cycling number on BFO/GSO retention is presented in the inset of Fig. 4(a). The polarization degradation with retention time in the inset of Fig. 4(a) is thought to be caused by the frozen compensation charges in BFO/GSO thin films. All these three fitting lines show the logarithmic time dependence. According to Ref. [14], the changes of polarization over time are thought to obey the equation $$\begin{align} P(t)=P(0)-m\ln t+\Delta P ,~~ \tag {1} \end{align} $$ where $P(0)$ is the polarization at $t=0$, $t$ is the time, $m$ is the decay rate, and $\Delta P$ is the polarization depending on each test's condition. The values of each $m$ with various loops have been calculated and given in the inset of Fig. 4(a). With the increase of the pre-cycling numbers, the value of $m$ decreases until leveling off at a stable value, which can be explained by the frozen-compensation-charge theory. The lattice mismatch between BFO and GSO is so small that the frozen charge is much more movable than those within the BFO/SRO/STO thin films. The short-time retention property is improved when the voltage cycling number increases. Alternatively, PFM was carried out to perform a longtime domain retention observation. Its tip can apply a voltage on the local film area, thus we can write a domain pattern through tip scanning over the chosen area. After writing a square domain on one specific area, we can scan this area again after a retention time to image the polarization variants of the written domains through a piezoelectric effect under a readout voltage much smaller than the write voltage. Figures 5(a1)–5(a4) and 5(b1)–5(b4) give the out-of-plane and in-plane PFM images of the written domains over one 5$\times$5 μm$^{2}$ area on a BFO/SRO/GSO thin film, respectively. As shown in Figs. 5(a1) and 5(b1), periodical stripe domain patterns are obtained. When $\pm$20 V voltages were applied on the film surface, most of the domains tend to align along the direction of the electric field, as displayed in Figs. 5(a2) and 5(b2). Figures 5(a3), 5(a4) and 5(b3), 5(b4) show the changing domain patterns after 1 h and 11 h, respectively. The strip domain patterns remain clear, especially from out-of-plane images in Figs. 5(a3) and 5(a4). The slight change shown by green arrows may be caused by the longtime thermal noise to float the PFM tip during the measurement. In Figs. 5(b2)–5(b4), the in-plane domain reverses immediately under +20 V but seems to be unchanged under $-$20 V, which is different from the out-of-plane domain reversing and thus we use the out-of-plane figure only to express domain changing later.
cpl-34-2-027701-fig5.png
Fig. 5. The first 8 patterns ((a1)–(b4)) are in and out-of-plane PFM image of the same $5\times5$ μm$^{2}$ area before writing five strip patterns and after 10 min, 1 h and 11 h by applying alternating 20 V and $-$20 V voltage (the dark strip correspond to downward polarization). The following 9 patterns ((c1)–(e3)) are $1\times1$ and $2\times2$, $5\times5$, $15\times15$ μm$^{2}$ square domain areas measured immediately after applying $-$20 V writing voltage, after 7 days and 50 days. Both PFM images are measured under the voltage of 6 V and 75 kHz.
cpl-34-2-027701-fig6.png
Fig. 6. The relaxing time dependences of 4 different domain squares' areas in Fig. 5.
To characterize the domain change, four square domain areas with different areas (1$\times$ μm$^{2}$, $2\times2$ μm$^{2}$, $5\times5$ μm$^{2}$ and $15\times15$ μm$^{2}$) were written on a BFO/SRO/GSO thin film and tested after 7 days and 50 days as shown in Figs. 5(c1)–5(e3). Clearly, each domain maintained its shape and had a sharp boundary. Thus a merely very small change can be found precisely (signed in the picture by the yellow arrow). To accurately characterize the variation of each domain, we figure out the area of each squared domain by setting a fixed threshold. Figure 6 shows the areas of the squared domains varying with relaxation times. We can find that each squared domain tends to be stable after 7 days. The $1\times1$ μm$^{2}$ squared domain degrades slightly faster than other domain areas, which implies that small domains are harder to maintain than large domains as their polarization may be easily affected by the opposite polarization around them.[15] In summary, the retention properties of the ferroelectric BFO/SRO/GSO films have been studied. The BFO/STO/GSO thin film grown exhibits excellent retention performance in polarization test, while the positive polarization of BFO/SRO/STO thin films degenerates abruptly during the first second time, which is attributed to the preferred polarization direction caused by the larger lattice mismatch. Moreover, with increasing the alternate voltage cycles, the short-time retention property of BFO/SRO/GSO films can be improved, which can be explained by the frozen-compensation-charge theory. Moreover, the retention test of PFM reveals that the written domains can maintain their shapes for 50 days or even half a year. Our work paves one route to improve the polarization retention of the BFO thin film for application in ferroelectric random access memory. References Nanoscale Domain Control in Multiferroic BiFeO3 Thin FilmsSynthesis and ferroelectric properties of epitaxial BiFeO3 thin films grown by sputteringThe Nature of Polarization Fatigue in BiFeO3The improved polarization retention through high-field charge injection in highly strained BiFeO 3 thin films with preferred domain orientationsElectrical reliability and leakage mechanisms in highly resistive multiferroic La0.1Bi0.9FeO3 ceramicsDomain switching kinetics in ferroelectric-resistive BiFeO 3 thin film memoriesThickness-dependent retention behaviors and ferroelectric properties of BiFeO3 thin films on BaPbO3 electrodesEpitaxial strain and electric boundary condition effects on the structural and ferroelectric properties of BiFeO 3 filmsReduced Coercive Field in BiFeO3 Thin Films Through Domain EngineeringNanosecond-range imprint and retention characterized from polarization-voltage hysteresis loops in insulating or leaky ferroelectric thin filmsReverse-poling effects on charge retention in Pb(Zr,Ti)O3(001)/LaNiO3(001) heterostructures180° Ferroelectric Stripe Nanodomains in BiFeO 3 Thin FilmsRetention Characteristics of Bi 3.25 La 0.75 Ti 3 O 12 Thin FilmsFerroelectric domain inversion and its stability in lithium niobate thin film on insulator with different thicknesses
[1] Chu Y H et al 2006 Adv. Mater. 18 2307
[2] Das R R et al 2006 Appl. Phys. Lett. 88 242904
[3] Baek S H, Folkman C, Park M et al 2011 Adv. Mater. 23 1621
[4] Liu X B, Ding N F, Jiang A Q and Yang P X 2012 Appl. Phys. Lett. 100 132901
[5] Wang S Y, Qiu X, Gao J et al 2011 Appl. Phys. Lett. 98 152902
[6] Meng J W, Jiang J, Geng W P et al 2015 Jpn. J. Appl. Phys. 54 024102
[7] Lee C C and Wu J M 2007 Appl. Phys. Lett. 91 102906
[8] Johann F, Morelli A, Biggemann D et al 2011 Phys. Rev. B 84 094105
[9] Shelke V, Mazumdar D, Srinivasan G et al 2011 Adv. Mater. 23 669
[10] Jiang A Q, Liu X B and Zhang Q 2011 Appl. Phys. Lett. 99 142905
[11]Zavaliche F, Yang S Y, Zhao T, Chu Y H et al 2007 Phase Transit. 79 991
[12] Jo W, Kim D C and Hong J W 2000 Appl. Phys. Lett. 76 390
[13] Chen Z, Liu J, Qi Y J, Chen D Y, Hsu S L et al 2015 Nano Lett. 15 6506
[14] Kang B S et al 2002 Jpn. J. Appl. Phys. 41 5281
[15] Shao G, Bai Y, Cui G, Li C et al 2016 AIP Adv. 6 075011