Emergence of ferromagnetism in antiferromagnetic TbMnO3 by epitaxial strain

by epitaxial strain X. Marti, V. Skumryev, C. Ferrater, M. V. García-Cuenca, M. Varela, F. Sánchez, and J. Fontcuberta Institut de Ciència de Materials de Barcelona (ICMAB-CSIC), Campus UAB, 08193 Bellaterra, Spain Departament de Física, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain and Institució Catalana de Recerca i Estudis Avançats (ICREA), 08010 Barcelona, Spain FAO and IN2UB, Universitat de Barcelona, Martí i Franquès 1, 08028 Barcelona, Spain

The coupling of ferroelectricity and ferromagnetism in a single material or heterostructure attracts much attention due to its promising applications in sensor and data storage technologies. The scarcity of materials which individually display this feature 1 triggered the development of varied strategies where, for instance, the coupling is mediated by elastic 2 or exchange-bias 3 effects. Particularly relevant is the discovery that ferroelectricity can be obtained subsequent to magnetic ordering. Antiferromagnetic orthorhombic rare earth manganites ͑RMnO 3 ͒ probably constitute the best known example of such materials. Either if the magnetic structure is collinear ͑E-type as in YMnO 3 ͑YMO͒ ͑Ref. 4͒ or ͑HoMnO 3 ͒ ͑Ref. 5͒ or noncollinear ͓spiral as in TbMnO 3 ͑TMO͔͒, 6 magnetic interactions give rise to atomic displacements destroying the center of symmetry and thus allowing an electrical polarization. 7 Due to this intimate link between magnetic structure and ferroelectricity, changes in the electrical polarization upon applied magnetic fields are to be expected. Indeed, it was the early experimental observation of such effects in bulk TMO ͑Ref. 8͒ that triggered intensive theoretical research and exploration of epitaxial thin films. However, investigation of magnetic properties on RMnO 3 films ͑Refs. 9-17͒ revealed the unexpected existence of a ferromagnetic response in otherwise bulk antiferromagnetic materials. Whereas in the case of E-type spin-ordered structures, its origin has been attributed to strain-induced deformation of the unit cell and subsequent unbalancing of magnetic interactions, [9][10][11][12]17 it also has been proposed domain walls boundaries to be relevant in the case of the spiral TbMnO 3 . 16 However, it turns out that the signatures of ferromagnetism reported 9-17 are in all cases intriguingly similar. Whether the origin of the ferromagnetism in both compounds is the same irrespective of their bulk magnetic structure is a critical open issue that we aim to disentangle in this paper.
We report here on the magnetic properties of epitaxial TMO thin films. Data shows that the ferromagnetic response can be selected at wish by tuning the unit cell volume ͑V C ͒ through the film thickness and deposition conditions. Of relevance here is that the obtained V C versus ferromagnetism correlation clearly resembles that reported for orthorhombic YMO thin films indicating that its origin does not depend on the bulk magnetic structure and it is strongly related to the unit cell distortion.
Thin films were deposited by pulsed laser deposition using a KrF excimer laser focused on a stoichiometric TbMnO 3 target with the fluence and the repetition rate set to 1.5 J / cm 2 and 5 Hz, respectively. Films were grown on ͑001͒-oriented SrTiO 3 ͑STO͒ substrates, heated at 785°C and placed at 5 cm from the target. A set of samples were grown at 0.1, 0.2, 0.3, and 0.4 mbar oxygen pressure; the laser pulses were adjusted to obtain films with thickness around 40 nm. This set of samples was completed with three thicker films grown at fixed pressure ͑0.3 mbar͒ and being 95, 137, and 140 nm thick. Crystal structure was investigated using four-circle x-ray diffractometer. Magnetic data were recorded using a superconducting quantum interference device with the magnetic field applied parallel to the sample surface. Figure 1͑a͒ shows the / 2 scan of a selected TMO sample ͑38 nm thick, 0.3 mbar͒. Only reflections corresponding to the substrate and to the TMO͑001͒ planes are observed indicating that the films are c-textured note the Pbnm settings is used. 18 -scans around the ͑111͒ reflections of the substrate and the film are shown in Fig. 1͑b͒. Four narrow substrate STO͑111͒ peaks, 90°in-plane spaced, are visible in Fig. 1͑b͒, together with four o-TMO͑111͒ reflections, located at 45°from each STO͑111͒ reflection; this indicates that the in-plane axes ͑a, b͒ of the film are 45°in-plane rotated respect to the ͓100͔ direction of the substrate. The broadening of the TMO͑111͒ peaks results from the existence of 90°i n-plane twinning, as we have exhaustively analyzed for o-YMO͑001͒/STO͑001͒ films. 19 Reciprocal space maps around the TMO͑208͒ and TMO͑028͒ reflections are shown in Figs. 1͑c͒ and 1͑d͒ Figure 1͑c͒ shows that the in-plane component of the ͑208͒ reflection is observed exactly at the bulk position thus implying that the a parameter is fully relaxed. In contrast, data in Fig. 1͑d͒ show that the ͑028͒ reflection is shifted toward larger K values indicating that b parameter is compressively strained. On the other hand, the c parameter is elongated as evidenced in both plots by the shorter L of the corresponding reflections. It turns out that, similarity to what was observed in o-YMO films, 11,18 epitaxial strain is anisotropic being zero, tensile, and compressive along the ͓100͔, ͓001͔, and ͓010͔ directions, respectively. These observations are common to all TMO samples produced. As a consequence of these distortions, the unit cell volume can be contracted up to 2%.
We now turn to the magnetic characterization of the films. Figure 2 shows the inverse susceptibility −1 ͑T͒ of the thickest TMO sample available ͑140 nm͒ measured after field cooled ͑FC͒ at two different magnetic fields ͑10 and 20 kOe͒ applied in the plane of the film. We focus our attention to the high temperature paramagnetic regime, T ӷ 40 K. We note that susceptibility values measured at different fields coincide thus indicating the absence of spurious ferromagnetic contributions. 20 The extrapolation of the linear regime of susceptibility curves to lower temperatures reveals that the intersection with the horizontal axis is around p = 0 in agreement with previous data on bulk TMO ͑Ref. 21͒ measured along the b-axis. In agreement with bulk measurements, the absence of a well-defined antiferromagnetic up-turn at −1 ͑T N ͒, visible in YMO films, 11 indicates that the paramagnetic susceptibility is dominated by Tb 3+ ions. 22 The slope of the inverse susceptibility versus temperature leads to a Curie constant C Ϸ 7.22 K which corresponds to and effective magnetic moment eff Ϸ 12͑2͒ B which is in agreement with that expected ͕ eff = ͱ͓ eff 2 ͑Tb͒ + eff 2 ͑Mn͔͒ =11 B ͖ for the contribution from Tb 3+ and Mn 3+ ions ͓ eff ͑Tb͒ = 9.72 B and eff ͑Mn͒ = 4.90 B ͔. Figures 3͑a͒-3͑c͒ show the zero-field cooled ͑ZFC͒ and the FC ͑ZFC-FC͒ magnetization curves of few representative samples having distinct V C values. A clear splitting of both curves is well visible revealing a net magnetization. The splitting emerges intriguingly close to the bulk antiferromagnetic ordering ͑T N ϳ 40 K͒ and at slightly increasing temperatures as the unit cell volume contracts. It suggests that the ferromagnetic ordering is closely related to the pristine antiferromagnetic ordering and, at the same time, it depends on the unit cell distortion. Indeed, data shows that the ZFC-FC splitting increases when shrinking V C . Identical be-  havior is found in all films as shown in Fig. 3͑d͒, where only FC curves are included for the sake of clarity. Note that data clearly resembles the behavior reported for YMO films 11 and Y 3+ is a nonmagnetic atom. Owing to this similarity and the fact that the net magnetic moment is observed well above the expected Tb 3+ ions ferromagnetic ordering, it turns out plausible that the ZFC-FC splitting is caused only by the Mn sublattice.
Following the criteria we already used for YMO films, 11 we plot in Fig. 4 the dependence of ͑T=25 K͒ on V C aiming to quantitatively evaluate how epitaxial strain is inducing the weak ferromagnetism. In contrast to our early work on YMO, 11 we used V c instead of the strain along ͓010͔ to correlate the magnetic response because the observed trend is identical but V c includes all structural distortions. We have subtracted the paramagnetic contribution of Tb 3+ ions to evaluate the contribution of the magnetically ordered Mn 3+ O 3 sublattice: this allows comparing the strain-induced magnetization to that reported for YMO films in Ref. 11. Data in Fig. 4 collects the main finding here, that is the ferromagnetic response of the TMO films is determined by the unit cell contraction and its behavior mimics that reported for YMO films. 11 It thus follows that the magnetic structure of the pristine bulk material, that is its spiral or collinear orders are not relevant to the magnetic response of strained YMO or TMO films. Indeed, the evident similarity between magnetic properties in strained YMO and TMO films suggest that the magnetic structure could be similar in both strained films thus vanishing the boundary between E-type ͑YMO͒ ͑Ref. 4͒ and spiral ͑TMO͒ ͑Ref. 6͒ magnetic orderings existing in bulk orthorhombic rare earth manganites. As suggested in Ref. 16, a possible scenario could be that epitaxial strain primarily suppresses the spiral order in TbMnO 3 in favor of the E-type as found in YMnO 3 . Subsequent strain-unbalanced magnetic interactions may lead to the observed weak ferromagnetic response as found in YMnO 3 . 11 In summary, we have shown that by gradually contracting the unit cell of epitaxial TMO͑001͒ thin films a ferromagnetic response gradually emerges. The present work shows that despite the fact that the bulk magnetic ground structure of TbMnO 3 is spiral, the changes in the magnetic structure upon epitaxial strain follow the same pattern as observed in the strained YMO films having a collinear ͑E-type͒ ground magnetic structure. Therefore, it follows that in TbMnO 3 can be made ͑weak͒ ferromagnetic under strain. Although it remains to be seen that the ferroelectric character is preserved, recent results 23 on the closely related YMnO 3 strongly suggest that strained TbMnO 3 films may become a new member of the reduced family of biferroic: ferromagnetic and ferroelectric, materials.