Multiferroic tunnel junctions

Abstract

Magnetic tunnel junctions with ferroelectric barriers, often referred to as multiferroic tunnel junctions, have been proposed recently to display new functionalities and new device concepts. One of the notable predictions is that the combination of two charge polarizing states and the parallel and antiparallel magnetic states could make it a four resistance state device. We have recently studied the ferroelectric tunneling using a scanning probe technique and multiferroic tunnel junctions using ferromagnetic La_0.7Ca_0.3MnO₃ and La_0.7Sr_0.3MnO₃ as the electrodes and ferroelectric (Ba, Sr)TiO₃ as the barrier in trilayer planner junctions. We show that very thin (Ba, Sr)TiO₃ films can sustain ferroelectricity up till room temperature. The multiferroic tunnel junctions show four resistance states as predicted and can operate at room temperatures.

Graphical abstract

Keywords

multiferroic tunnel junction / ferroelectric film / tunneling magnetoresistance effect / tunneling electroresisitance effect / magnetoelectric coupling

1 Introduction

The demands for faster, smaller, and non-volatile electronics have pushed the limit of semiconductor devices to ever smaller dimensions. However, the challenge of power dissipation places a limit on the future device scaling. In addition, size effects may also post constraint on device miniaturization. These have led to an emergent need and development of new concepts for information processing and storage [1, 2]. For the next generation devices, multifunctional structures have been envisioned, which can perform new functionality, can be manipulated by several external independent controls, and can operate with low power [3, 4]. One of the approaches to multi-functional devices is to combine ferromagnetic (FM) based spintronics with ferroelectric (FE) based functionality [3, 4]. It can be used as memory cells with multiple resistance states or to combine logic and memory operations. It may also be able to achieve cross control of the spin and charge degrees of freedom. This offers significant opportunities for new device concepts.

Magnetic tunnel junctions (MTJs) is the most widely used spintronic device in computer read heads, sensors, and memories [5]. The parallel and antiparallel magnetization alignments of the two electrodes represent two logic states with different resistances (tunneling magnetoresistance, TMR). Ferroelectric tunneling refers to a tunnel junction using a ferroelectric as the barrier. The basic idea of an FTJ was raised by Esaki et al. [6] decades ago. However, it only aroused considerable research interest recently, largely because only in recent years experiments [7–9] have shown that ferroelectricity can persist down to the nanometer scale. This makes it possible to use ferroelectrics as tunneling barriers. The key property of an FTJ is the switching of junction resistance with the reversal of ferroelectric polarization, namely, tunneling electroresistance (TER) effect. As reviewed by Tsymbalet al. [10], the TER effect of an FTJ could at least be obtained through the following three effects: i) electrostatic effect which will change the depolarization potential through the asymmetrical screening of the two interfaces of the barrier [11]; ii) asymmetrical interface atomic arrangement that will influence the transmission probability by the ferroelectric ionic displacement at interfaces [12]; and iii) strain effect which will affect the barrier width due to the piezoelectricity effect [13]. Very recently large TER effect tested using a scanning probe technique has been reported on BaTiO₃ [14–16] and PbZr_xTi_1-xO₃ [17, 18] ferroelectric films.

Combining a magnetic tunnel junction with a ferroelectric barrier, i.e., a multiferroic tunnel junctions (MFTJs), will add new functionality beyond the single MTJs or FTJs. The junction resistance will be sensitive to not only the magnetization alignment of the electrodes, but also the polarization direction of the barrier. This makes MFTJ a quaternary-state device which can be switched by both electric and magnetic fields. Theoretical calculations on SrRuO₃/BaTiO₃/SrRuO₃ structure have predicted [19] about \sim 50% resistance differences between the two polarization states, making the MFTJ a practically useful device. Moreover, due to the mechanical coupling [20], chemical bonding [21] or spin dependent screening [22] at the ferromagnetic/ferroelectric interfaces, these artificial composites of ferroic materials may enable the magnetoelectric coupling effect (as reviewed by [3, 4, 23–26]), making it possible to control magnetization by electric field or control electric polarization by magnetic field.

The first reported MFTJ is another type of MFTJ employing a single phase multiferroic La_0.1Bi_0.9MnO₃ barrier in a Au/La_2/3Sr_1/3MnO₃ junction where the TMR is due to the spin filtering effect and the TER is related to the barrier potential change when polarization is reversed [27]. Recently, by using a Fe indentation technique, Garcia et al. have fabricated the nanoscale Fe (or Co)/BaTiO₃/La_0.7Sr_0.3MnO₃ ferroelectric/ferromagnetic/ferroelectric MFTJs and reported the control of carrier spin polarization by electrically reversing the ferroelectric polarization [28, 29]. Moreover, not only the magnitude, but also the signs of the TMR can be inverted by ferroelectric switching in Co/PbZr_0.2Ti_0.8O₃/La_0.7Sr_0.3MnO₃ MFTJs, showing a new method of spin control in spintronic devices [30]. The four resistance states have also been reported on all perovskite La_0.7Sr_0.3MnO₃/BiFeO₃/La_0.7Sr_0.3MnO₃ MFTJs where BiFeO₃ is an intrinsic multiferroic material [31]. However, all of these results showed small magnetoresistance and can work only at low temperatures.

To obtain both TMR and TER effects in a MFTJ, thin insulating barrier with persistent ferroelectricity and ferromagnetic electrodes with high spin polarization are critical. Here, we choose (Ba, Sr)TiO₃ as ferroelectric barrier which shows ferroelectricity down to one unit cell under proper strain condition [7] and half metallic La_0.7Ca_0.3MnO₃ (LCMO) and La_0.7Sr_0.3MnO₃ (LSMO) as ferromagnetic electrodes for the proper lattice match with the barrier [32]. Large TMR and TER effects have been observed in the LCMO-based MFTJs [33], and in LSMO-based MFTJs the four resistance states can be achieved even up to room temperature [34].

2 Experiment

Epitaxial (Ba, Sr)TiO₃/La_0.7Sr_0.3MnO₃ bilayers for piezoresponse force microscopy (PFM) measurements and La_0.7Ca_0.3MnO₃/(Ba, Sr)TiO₃/La_0.7Ca_0.3MnO₃ and La_0.7Sr_0.3MnO₃/(Ba, Sr)TiO₃/La_0.7Sr_0.3MnO₃ trilayers for MFTJ investigations were grown on SrTiO₃ (100) substrates by pulsed laser deposition. The heterostructures were deposited in situ at 750℃ in a 300 mTorr flowing oxygen. The laser energy density was about 1.5 J/cm². The epitaxial growth of the trilayers on SrTiO₃ is revealed by high-resolution transmission electron microscopy (TEM) and the TEM specimens were prepared using conventional mechanical polishing followed by Ar ion milling. Cross-sectional scanning TEM (STEM) was performed in a JEOL-2010F field-emission TEM/STEM operated at 200 keV. The ferroelectric property of the (Ba, Sr) TiO₃ ultrathin films were investigated by PFM [Asylum Research-MFP-3D atomic probe microscopy (AFM)].

The micron-scale MFTJ in the cross-strip geometry was fabricated by using photolithography and Ar ion milling in a process similar to what we have used before [35]. Sputtered MgO or SiO₂ was used as an insulating layer to allow making gold contact leads with the top electrode without shorting to the bottom electrode. The schematic drawing of the MFTJ structure is illustrated in Fig. 1(a). A four-probe method was used to do electrical transport measurement. The positive bias corresponds to the current flow from top to bottom electrode and the magnetic field was applied along the longitudinal direction of the bottom La_0.7Sr_0.3MnO₃ lead.

3 Results and discussion

To characterize the structural quality, X-ray diffraction θ–2\theta and \phi scans as well as high resolution cross sectional STEM have been conducted on selected samples. X-ray diffraction shows that all layers are grown epitaxially. Figure 1(b) is a high-resolution cross-sectional high-angle annular-dark-field (HAADF) STEM image of an as-grown La_0.7Ca_0.3MnO₃/Ba_0.5Sr_0.5TiO₃/ La_0.7Ca_0.3MnO₃ trilayer. Almost atomically smooth growth of the Ba_0.5Sr_0.5TiO₃ barrier was obtained even for our relatively thick La_0.7Ca_0.3MnO₃ bottom film (\sim30 nm), indicating high quality heterostructure growth.

To confirm the ferroelectricity of nanoscale (Ba, Sr)TiO₃ film, PFM and conducting atomic force microscopy (CAFM) have been performed at room temperature on several (Ba, Sr)TiO₃/La_0.7Sr_0.3MnO₃ samples. As illustrated in Fig. 2(a), these investigations were carried out based on a scanning probe microscopy (Asylum Research-MFP-3D AFM) using a conductive AFM tip in contact with the (Ba, Sr)TiO₃ surface. Figure 2(b) shows an out-of-plane PFM phase image at room temperature for a 3.5 nm Ba_0.95Sr_0.05TiO₃ thin film on top of a 30 nm La_0.7Sr_0.3MnO₃ bottom electrode. Here, a word “PSU” and the background were written by negative (-6.0 V) and positive (+6.0 V) voltages respectively. The observed “PSU” pattern indicates the occurrence of ferroelectric switching. Figure 1(b) shows a local voltage dependence of piezoresponse phase which suggests a \sim 180^∘ phase contrast and provides evidence for ferroelectricity of nanoscale (Ba, Sr)TiO₃ film. The current (I)–voltage (V) characteristic of this sample have also been tested by CAFM, as shown in Fig. 3. After polarize the ferroelectricity upward and downward, two obvious different I–V was observed. At -1 V, a large TER \sim30 000% was obtained.

In trilayer junction structures, due to the large junction area (∼10 μm×10 μm) in comparison with the scanning probe contact area (with a diameter of about several tens of nanometers), large TER has not been reported presumably due to the weak links and multidomain structures when the ferroelectric barrier is poled to reverse its directions using a large voltage. In addition, it was proposed that the large TER observed from the PFM measurement without the top electrode is due to the surface absorption of elements which acted as an additional dielectric layer on top of the ferroelectric layer [36]. By increasing the asymmetry of the junction, the dielectric layer will serve as a switch changing its barrier height dramatically when the electrostatic potential is changed by polarization reversal, leading to the giant TER values in PFM test.

Figure 4(a) shows the non-linear I–V characteristic of a La_0.7Sr_0.3MnO₃/Ba_0.95Sr_0.05TiO₃/La_0.7Sr_0.3MnO₃ MFTJ (∼10 μm×20 μm in area) at room temperature. The corresponding differential conductance G=\mathrm{d}I/\mathrm{d}V shows a parabolic dependence on bias voltage, as can be seen in the normalized conductance G(V)/G(0) shown in the inset of Fig. 4(a). As indicated by the solid line in the inset of Fig. 4(a), the good fitting of the conductance curve with Brinkman model [37] indicates that the electron tunneling dominates the transport process. Figure 4(b) shows potential diagrams drew based on the fitting parameters, the barrier thickness (d), average barrier height (\overline{\phi }) and the asymmetry in the barrier (\mathrm{\Delta }\phi ) for the curves with the barrier polarized downward and upward (with a poling voltage of V_{\mathrm{pulse}}\approx +1.4 V or -1.4 V). The differences between the barrier parameters after poling the ferroelectricity down or up suggest that the ferroelectricity can affect the electron transport properties of MFTJ, which will result in the TER effect. Similar to the theoretical prediction [19], the TER effect may be caused by the asymmetric Ba_0.95Sr_0.05TiO₃/La_0.7Sr_0.3MnO₃ interface which will induce asymmetric ferroelectric displacements in the barrier and lead to a change in complex band structure in Ba_0.95Sr_0.05TiO₃. Moreover, although \pm 1.4 V was used as poling voltages, we do not have direct evidence that the voltage is enough to completely poling the ferroelectric layer. Therefore, other switching mechanisms like the interfacial electrochemical modification by electric field may also take responsibility for the results [38–40]. More work is being carried on to understand the underlying mechanism.

Figure 5 shows that, by reversing the barrier polarization repeatedly, the junction resistances measured at 10 μA bias between the two polarization states can be switched between each other directly. Here, the red and blue dots means that the switching was observed between parallel and antiparallel magnetic states [reached by R–H sweep (see Fig. 6)], respectively. Hence, after applying positive and negative voltages, two parallel and two antiparallel states i.e. a distinct four-state device was represented.

Figure 6(a) shows the magnetic field (H) dependence of the junction resistance (R) at a fixed voltage bias of 10 mV after aligning the ferroelectric polarization of Ba_0.95Sr_0.05TiO₃ barrier downward or upward. A sharp resistance change against the magnetic field is observed between two magnetic states (parallel and antiparallel), suggesting an almost entire flip of the magnetic domains. After reversing the polarization of barrier, the entire R–H curve shifts so that both parallel and antiparallel resistances switched to different values and four resistance states were observed. Figure 6(b) shows two resistance memory loops (measured at 10 mV) as a function of magnetic field for the two polarization states. A standard MTJ memory loop with two states (parallel and antiparallel) is obtained for both polarization states and it can be switched between two states by sweeping magnetic field. Similar to the R–H loops in Fig. 6(a), the resistance can be switched from one memory loop to aother by reversing the ferroelectricity of the barrier. It is worth mentioning that, from Fig. 6(b), the four states can be observed at zero magnetic and electric field, which means that the memory device is non-volatile.

Larger TER and TMR effect (up to \sim300%) can be obtained at lower temperatures in the trilayer junctions [33]. Figure 7 shows the R–V_p loops of a La_0.7Ca_0.3MnO₃/Ba_0.5Sr_0.5TiO₃/La_0.7Ca_0.3MnO₃ MFTJ (∼ 8 μm × 8 μm in area) in parallel magnetization state at 4.2 K. Here, each data point in the hysteresis loop is tested at a fixed voltage bias [-0.1 V for Fig. 7(a), -0.15 V for Fig. 7(b)] after pulsing a voltage. The TER values decreases from 45 % to 22 % with the increasing bias from -0.1 V to -0.15 V, and are all higher than the TER at room temperature shown in Figs. 5 and 6 (\sim2%) which are tested at even lower bias. The junction resistance changes when the polarization was switched by external voltage pulses and saturates afterward, indicating the coercive field for the polarization reversal should be reached. The switching voltage is slightly asymmetric, possibly due to the imprint of the ferroelectric layer.

4 Conclusions

In conclusion, we have fabricated all-oxide multiferroic tunnel junctions by pulsed laser deposition technique. Four distinct resistance states with large resistance differentials have been obtained and the device can operate at room temperatures. Using scanning probe technique, much larger TER effect can be obtained than that of the large size planner trilayer tunnel junctions. Further study on the origins of the differences should be conducted so that larger TER effect can also be obtained in trilayer junctions, which can be used in practical devices. These results pave the way for new multilevel and nonvolatile devices.

References

[1] R. Ramesh and N. A. Spaldin, Nat. Mater., 2007, 6(1): 21

[2] J. A. Hutchby, R. Cavin, V. Zhirnov, J. E. Brewer, and G. Bourianoff, Computer, 2008, 41(5): 28

[3] J. Ma, J. M. Hu, Z. Li, and C. W. Nan, Adv. Mater., 2011, 23(9): 1062

[4] J. P. Velev, S. S. Jaswal, and E. Y. Tsymbal, Philos. Trans. R. Soc. A, 2011, 369: 3069

[5] J. S. Moodera, L. R. Kinder, T. M. Wong, and R. Meservey, Phys. Rev. Lett., 1995, 74(16): 3273

[6] L. Esaki, R. B. Laibowitz, and P. J. Stiles, IBM Technical Disclosure Bulletin, 1971, 13: 2161

[7] D. A. Tenne, A. Bruchhausen, N. D. Lanzillotti-Kimura, A. Fainstein, R. S. Katiyar, A. Cantarero, A. Soukiassian, V. Vaithyanathan, J. H. Haeni, W. Tian, D. G. Schlom, K. J. Choi, D. M. Kim, C. B. Eom, H. P. Sun, X. Q. Pan, Y. L. Li, L. Q. Chen, Q. X. Jia, S. M. Nakhmanson, K. M. Rabe, and X. X. Xi, Science, 2006, 313(5793): 1614

[8] D. D. Fong, G. B. Stephenson, S. K. Streiffer, J. A. Eastman, O. Auciello, P. H. Fuoss, and C. Thompson, Science, 2004, 304(5677): 1650

[9] C. Lichtensteiger, J.M. Triscone, J. Junquera, and P. Ghosez, Phys. Rev. Lett., 2005, 94(4): 047603

[10] E. Y. Tsymbal and H. Kohlstedt, Science, 2006, 313(5784): 181

[11] M. Y. Zhuravlev, R. F. Sabirianov, S. S. Jaswal, and E. Y. Tsymbal, Phys. Rev. Lett., 2005, 94(24): 246802

[12] J. P. Velev, C. G. Duan, K. D. Belashchenko, S. S. Jaswal, and E. Y. Tsymbal, Phys. Rev. Lett., 2007, 98(13): 137201

[13] H. Kohlstedt, N. A. Pertsev, J. R. Contreras, and R. Waser, Phys. Rev. B, 2005, 72(12): 125341

[14] V. Garcia, S. Fusil, K. Bouzehouane, S. Enouz-Vedrenne, N. D. Mathur, A. Barthélémy, and M. Bibes, Nature, 2009, 460(7251): 81

[15] A. Gruverman, D. Wu, H. Lu, Y. Wang, H. W. Jang, C. M. Folkman, M. Y. Zhuravlev, D. Felker, M. Rzchowski, C. B. Eom, and E. Y. Tsymbal, Nano Lett., 2009, 9(10): 3539

[16] A. Chanthbouala, A. Crassous, V. Garcia, K. Bouzehouane, S. Fusil, X. Moya, J. Allibe, B. Dlubak, J. Grollier, S. Xavier, C. Deranlot, A. Moshar, R. Proksch, N. D. Mathur, M. Bibes, and A. Barthelemy, Nat. Nanotechnol., 2012, 7(2): 101

[17] P. Maksymovych, S. Jesse, P. Yu, R. Ramesh, A. P. Baddorf, and S. V. Kalinin, Science, 2009, 324(5933): 1421

[18] A. Crassous, V. Garcia, K. Bouzehouane, S. Fusil, A. H. G. Vlooswijk, G. Rispens, B. Noheda, M. Bibes, and A. Barthelemy, Appl. Phys. Lett., 2010, 96(4): 042901

[19] J. P. Velev, C. G. Duan, J. D. Burton, A. Smogunov, M. K. Niranjan, E. Tosatti, S. S. Jaswal, and E. Y. Tsymbal, Nano Lett., 2009, 9(1): 427

[20] H. Zheng, J. Wang, S. E. Lofland, Z. Ma, L. Mohaddes-Ardabili, T. Zhao, L. Salamanca-Riba, S. R. Shinde, S. B. Ogale, F. Bai, D. Viehland, Y. Jia, D. G. Schlom, M. Wuttig, A. Roytburd, and R. Ramesh, Science, 2004, 303(5658): 661

[21] C. G. Duan, S. S. Jaswal, and E. Y. Tsymbal, Phys. Rev. Lett., 2006, 97(4): 047201

[22] J. M. Rondinelli, M. Stengel, and N. A. Spaldin, Nat. Nanotechnol., 2008, 3(1): 46

[23] C. A. F. Vaz, J. Hoffman, C. H. Anh, and R. Ramesh, Adv. Mater., 2010, 22(26–27): 2900

[24] C. W. Nan, M. I. Bichurin, S. X. Dong, D. Viehland, and G. Srinivasan, J. Appl. Phys., 2008, 103(3): 031101

[25] K. F. Wang, J. M. Liu, and Z. F. Ren, Adv. Phys., 2009, 58(4): 321

[26] M. Bibes, J. E. Villegas, and A. Barthelemy, Adv. Phys., 2011, 60(1): 5

[27] M. Gajek, M. Bibes, S. Fusil, K. Bouzehouane, J. Fontcuberta, A. E. Barthélémy, and A. Fert, Nat. Mater., 2007, 6(4): 296

[28] V. Garcia, M. Bibes, L. Bocher, S. Valencia, F. Kronast, A. Crassous, X. Moya, S. Enouz-Vedrenne, A. Gloter, D. Imhoff, C. Deranlot, N. D. Mathur, S. Fusil, K. Bouzehouane, and A. Barthélémy, Science, 2010, 327(5969): 1106

[29] S. Valencia, A. Crassous, L. Bocher, V. Garcia, X. Moya, R. O. Cherifi, C. Deranlot, K. Bouzehouane, S. Fusil, A. Zobelli, A. Gloter, N. D. Mathur, A. Gaupp, R. Abrudan, F. Radu, A. Barthélémy, and M. Bibes, Nat. Mater., 2011, 10(10): 753

[30] D. Pantel, S. Goetze, D. Hesse, and M. Alexe, Nat. Mater., 2012, 11(4): 289

[31] M. Hambe, A. Petraru, N. A. Pertsev, P. Munroe, V. Nagarajan, and H. Kohlstedt, Adv. Funct. Mater., 2010, 20(15): 2436

[32] J. H. Park, E. Vescovo, H. J. Kim, C. Kwon, R. Ramesh, and T. Venkatesan, Nature, 1998, 392(6678): 794

[33] W. J. Hu, K. Chen, Z. D. Zhang, X. X. Xi, and Q. Li, unpublished

[34] Y. W. Yin, M. Raju, W. J. Hu, X. J. Weng, X. G. Li, and Q. Li, J. Appl. Phys., 2011, 109: 07D915

[35] E. T. Wertz, and Q. Li, Appl. Phys. Lett., 2007, 90(14): 142506

[36] M. Y. Zhuravlev, Y. Wang, S. Maekawa, and E. Y. Tsymbal, Appl. Phys. Lett., 2009, 95(5): 052902

[37] W. F. Brinkman, R. C. Dynes, and J. M. Rowell, J. Appl. Phys., 1970, 41(5): 1915

[38] J. Z. Sun, K. P. Roche, and S. S. P. Parkin, Phys. Rev. B, 2000, 61(17): 11244

[39] H. Kohlstedt, A. Petraru, K. Szot, A. Rudiger, P. Meuffels, H. Haselier, R. Waser, and V. Nagarajan, Appl. Phys. Lett., 2008, 92(6): 062907

[40] J. Choi, J. S. Kim, I. Hwang, S. Hong, I. S. Byun, S. W. Lee, S. O. Kang, and B. H. Park, Appl. Phys. Lett., 2010, 96(26): 262113

Rights & permissions

© Higher Education Press and Springer-Verlag Berlin Heidelberg 2012