Exchange biasing and electric polarization with Y Mn O 3

We report on the growth and functional characterization of epitaxial thin films of the multiferroic YMnO3. We show that using Pt as a seed layer on SrTiO3(111) substrates, epitaxial YMnO3 films (0001) textured are obtained. An atomic force microscope has been used to polarize electric domains revealing the ferroelectric nature of the film. When a Permalloy layer is grown on top of the YMnO3(0001) film, clear indications of exchange bias and enhanced coercivity are observed at low temperature. The observation of coexisting antiferromagnetism and electrical polarization suggests that the biferroic character of YMnO3 can be exploited in novel devices.


U. Lüders and J. F. Bobo
We report on the growth and functional characterizations of epitaxial thin films of the multiferroic YMnO 3 . We show that using Pt as a seed layer on SrTiO 3 ͑111͒ substrates, epitaxial YMnO 3 films ͑0001͒ textured are obtained. An atomic force microscope has been used to polarize electric domains revealing the ferroelectric nature of the film. When a Permalloy layer is grown on top of the Multiferroic materials, displaying simultaneously the occurrence of magnetic and ferroelectric orders, are nowadays of the highest interest because of possible applications where dual tuning of the magnetic and electric responses could be of relevance. 1,2 Whereas much effort is being focused to the scarce materials displaying ferromagnetic and ferroelectric responses, 3 the functional exploitation of the more abundant materials displaying antiferromagnetic ͑AF͒ and ferroelectric ͑FE͒ orders has received less attention. Hexagonal YMnO 3 ͑YMO͒ belongs to this class of materials. It displays a high temperature FE transition ͑at about 900°C͒ and a low temperature AF transition ͑T N Ϸ 70 K͒. Possible use of the FE properties of YMO in ferroelectric devices has been already highlighted 4 and YMO thin films have been subsequently grown with a variety of techniques. 4,5 Electrical polarization in YMO is along the hexagonal c axis 6 whereas the magnetic moments of Mn 3+ lie in the perpendicular plane, forming a triangular, geometrically frustrated, network of AF coupled spins. 7,8 Antiferromagnets are used in spintronics by exploiting the exchange bias ͑EB͒ with suitable ferromagnets. Therefore, YMO could be also used as pinning layer if significant EB could be stimulated across an interface with a ferromagnet. Indeed, it has been recently proposed that YMO displays EB when grown textured along the ͓112 គ 1͔ direction but almost no exchange bias was observed for ͓0001͔ films. 9 As the ͓0001͔ direction is the polarization direction it seemed that prospectives of YMO for multifunctional devices were rather limited. In this letter we will show that YMO films with a unique ͓0001͔-axis texture can be grown on ͑111͒oriented SrTiO 3 ͑STO͒ single crystalline substrates when buffered with a thin Pt layer and that these films display both ferroelectric character and a substantial EB. As FE and AF domains are coupled in YMO ͑Ref. 10͒ it turns out that YMO could be exploited for multifunctional operation.
Pt films were deposited on a STO͑111͒ substrate by dc sputtering. Previous to the deposition, substrates were annealed in situ for 5 min at 800°C and, afterwards, Pt ͑ϳ8 nm thick͒ was deposited at 5 ϫ 10 −3 mbar pure Ar and at substrate temperature of 500°C. The Pt buffered substrate was transferred to a pulsed laser deposition chamber and the YMO was subsequently grown, using a stoichiometric YMnO 3 target. Films were deposited at a substrate temperature of 800°C in 0.2 mbar of oxygen pressure. At the end of the growth the samples were cooled down and 1 atm of oxygen was introduced in the chamber at 500°C. A Permalloy ͑Ni 81 Fe 19 , Py͒ layer, 15 nm thick, was subsequently grown by dc sputtering at room temperature, under a magnetic field of 100 Oe, in order to build the Py/YMO/Pt//STO͑111͒ heterostructure.
The crystal structure was investigated by x-ray diffractometry ͑XRD͒ using Cu K␣ radiation. Pt film thickness was determined by modeling Laue fringes observed in XRD patterns, whereas profilometry was used in other cases. Magnetic and electric transport properties were measured by using a superconducting quantum interference device and a physical with a Nanoscope Dimension 3100 microscope ͑Digital In-struments͒ was used to measure the piezoelectric response. The bottom Pt electrode was connected to ground by using silver paste, while the voltages V dc and V ac , to polarize the sample and characterize the response, respectively, were applied to the upper surface of the YMO film through the microscope tip.
In Fig. 1͑a͒ we show a XRD /2 scan corresponding to a Py/ YMO͑90 nm͒ /Pt/ /STO͑111͒ sample. Diffraction peaks from the substrate and the Pt and YMO layers are found. Inspection of the Pt reflections indicates that the Pt film has a ͑111͒ texture and the observation of the associated Laue fringes reflects the flatness and crystal quality of the Pt films. The reflections ͑0001͒ of the hexagonal YMO phase are well visible. Interestingly, no traces of orthorhombic phase are found in spite of being the stable phase when YMO grows directly on STO͑111͒ substrates. 11 Out-of-plane lattice distances were evaluated from Bragg's peaks positions. The out-of-plane parameter of Pt is d ͑111͒ = 2.27 Å, almost coincident with that of bulk Pt whereas the out-of-plane d ͑0001͒ ϵ c parameter of YMO is 11.45 Å, somewhat expanded with respect to c bulk = 11.40 Å; the strain is ͑0001͒ = 0.44%, suggesting an in-plane compression. Indeed, the in-plane parameter of the YMO film, determined from XRD scans, matches that of Pt thus reflecting the epitaxial compressive stress on the YMO films. In Fig. 1͑b͒ we have collected the pole figures of the STO͑110͒, Pt͑111 គ ͒, and YMO͑112 គ 4͒ reflections measured along the pole ͓111͔ STO direction. Considering the poles of the Pt͑111 គ ͒ reflections, one notices the existence of two sets of peaks ͑labeled A and B͒ corresponding to two families of planes ͕111 គ ͖. It indicates that the Pt epitaxial film has grown in two domains ͑180°in-plane ro-tated͒; the epitaxial relationships with the substrate are ͓112 គ ͔Pt͑111͒ / ͓112 គ ͔STO͑111͒ and ͓112͔Pt͑111͒ / ͓112 គ ͔ STO͑111͒. The pole figure of the YMO͑112 គ 4͒ reflections shows six peaks as expected due to the P6 3 cm symmetry of YMO. Notice that this symmetry does not allow to discern the presence of one or two domains ͑180°in-plane rotated͒. However, since YMO grows on bidomained Pt layer, two domains also in the YMO layer are expected. The corresponding epitaxial relationships are ͓101 គ 0͔YMO/ ͓112 គ ͔Pt and ͓101 គ 0͔YMO/ ͓112͔Pt.
The magnetic hysteresis loops, measured at 100 and 2 K after a magnetic field ͑3 kOe͒ cooling procedure, are displayed in Fig. 2. The loop measured at 100 K is symmetric with coercivity ͑H C ͒ of only ϳ10 Oe, very similar to the loops measured at higher temperature. In contrast, the loop measured at 2 K is asymmetric, revealing that an exchange bias ͑H EB Ϸ 61 Oe͒ accompanied by a significant coercivity enhancement ͑H C Ϸ 98 Oe͒ develops at low temperatures. Figure 2 includes two magnetization loops measured consecutively ͑denoted as first and second͒. It can be appreciated that there is a significant reduction of H EB upon cycling the field. This training effect 12 is a characteristic of AF systems with low or multiaxial anisotropy, 13 which likely contributes to the fast decay of H EB upon heating. The temperature dependences of H EB and H C are shown in Fig. 2 ͑inset͒. We notice that H EB decays rapidly when increasing the temperature and vanishs at ϳ10 K.
The existence of EB also provokes changes of the anisotropic magnetoresistance ͑AMR͒ of the upper-lying Py layer. 14,15 AMR was measured by rotating the magnetic field in the film plane while recording the film resistance R as a function of the angle between the measuring current and the applied magnetic field ͑H a ͒. In Fig. 3͑a͒ we include the R͑͒ cycles recorded at several temperatures ͑100, 50, and 2 K͒, above and below T N ͑ϳ70 K͒. Prior to any measurement, the sample was field cooled ͑3 kOe͒ from room temperature to the measuring temperature and then the field was reduced to the selected H a value. Data included in Fig. 3 were obtained when the cooling field was fixed at 45°from the current direction and H a = 40 Oe. We note that this value is smaller that the largest H C value determined from the magnetization loops ͑Fig. 2͒.
At T = 100 K, R͑͒ displays the common cos 2 dependence expected for a system where the magnetization follows the magnetic field ͓Fig. 3͑a͔͒. However, at 50 K, R͑͒ displays a clear departure of the cos 2 dependence due to the  2. ͑Color online͒ Magnetization loops vs applied magnetic field, measured at 100 K ͑open symbols͒ and 2 K ͑close symbols͒, after cooling the sample from 150 K in a field of 3 kOe. Two consecutive loops recorded at 2 K are displayed. Inset: temperature dependence of the exchange bias ͑closed squares͒ and the coercivity ͑open circles͒ fields.
presence of the H EB , which becomes more prominent when cooling down ͓Fig. 3͑b͔͒. In fact, at 2 K the presence of an H EB has dramatically modified the magnetization response when H a is rotated ͓Fig. 3͑c͔͒. The strong departure of the R͑͒ curve from the cos 2 dependence indicates that, at 2 K, the H EB should be much larger than the measuring field ͑40 Oe͒, and then larger than the values extracted from the magnetization loops. Observation of larger H EB values in transport measurements results from the different ways in which the presence of an H EB is revealed. 14 Notice that in transport measurements exchange bias is monitored via its impact on the magnetization when rotating H a but there is no need to switch the magnetization as in the common magnetic measurements and thus training effects are less important.
We turn now to the ferroelectric response. First, a background was set over a 2 ϫ 2 m 2 square by scanning it with V dc = + 6 V in contact mode using a tip velocity of 0.8 m/s. Inside, a smaller square ͑1 ϫ 1 m 2 ͒ was polarized with V DC = −6 V using the same tip velocity. Finally, the PFM response is recorded by using an ac signal ͑V ac = 3.5 V͒, and scanning over the outer square with a tip velocity of 0.4 m/s.
In Fig. 4 we show the PFM phase image of a YMO͑150 nm͒ / Pt/ / STO sample. As expected, it displays contrast over the polarized square ͑bright area of 1 ϫ 1 m 2 at the center͒, due to the different phases of the PFM response for the up and down domains; although the observed phase contrast is rather small ͑ϳ3°͒, the obvious change of contrast confirms the piezoresponse of the film. Traces of the substrate-induced steps are well visible in the PFM image ͓as well as in the topographic image ͑not show͔͒. We stress that contrast in the PFM image remains roughly constant for several hours after poling, thus suggesting a low leakage.
In summary, we have shown that YMO films can be used to induce exchange bias on suitable ferromagnetic layers while preserving the electric polarization. As magnetic and ferroelectric domains in YMO are coupled 10 our results strongly suggest that multiferroic materials with antiferromagnetic and ferroelectric orders can be successfully used to modify the exchange bias, and subsequently the transport and magnetic properties of a ferromagnetic layer, by electric fields.
Financial support by the MEC of the Spanish Government ͑Project Nos. NAN2004-9094 and MAT2005-5656͒, and FEDER and the STREP project ͑Grant No. 033221͒ of the EU are acknowledged. One of the authors ͑A.R.͒ acknowledges Generalitat de Catalunya for a grant. The authors are thankful to J. M. Pérez for technical support.
FIG. 3. ͑Color online͒ Angular dependence of the magnetoresistance ͑AMR͒ of the Py film at ͑a͒ 100 K, ͑b͒ 50 K, and ͑c͒ 2 K. Measurements were done using a magnetic field of 40 Oe, and after field cooling ͑3 kOe͒ applied at 45°with respect to the measuring current.