1 Introduction

The fundamental physical properties of perovskite mixed valence manganites have attracted tremendous interest since the discovery of the colossal magnetoresistance (CMR) effect [1]. Significant progress has been achieved in understanding the novel properties of these compounds from the perspective of interactions involving charge, spin, orbital, and lattice due to considerable theoretical and experimental efforts conducted in recent years. The further developments of giant magnetoresistance (GMR) and tunneling magnetoresistance (TMR) have enriched spintronic science and technology. Thus, manganese perovskites are promising to meet the requirements of commercial applications in the field of spintronic devices such as the read/write heads of magnetic hard-disk drives and magnetic random access memories (MRAMs) [2].

Among the manganese perovskites, Sr doped LaMnO3 has been most widely investigated as a promising electrode material due to its good current collecting ability and excellent electrocatalytic activity for fuel [3–5], high temperature oxygen sensor [6], and highly active nanocatalyst [7]. Particularly, it is also an optimal material for spintronic devices due to almost 100 % spin-polarization [3, 8] and a Curie temperature above room temperature. Many emergent phenomena have not been found in the bulk state, such as abnormal magnetic anisotropy [9, 10] and domain configuration control [11], but inherent within La1-x Sr x MnO3 thin films due to the inevitable reconstruction of the charge, spin, orbital, and lattice states at the film-substrate interface [12, 13], which is thought to play an adverse role in device applications [14, 15]. Thus, to tailor the magnetic properties of epitaxial La1−x Sr x MnO3 thin films, we need to seek a deeper understanding of the origin of substrate-induced anisotropies, which are very sensitive to a number of parameters such as the film strain imposed by the underlying substrate [11, 16], substrate miscut angles [17, 18], and dead layer at the film-substrate interface [9, 19].

Generally, a biaxial anisotropy is observed in the ferromagnetic (FM) La0.7Sr0.3MnO3 (LSMO) thin films epitaxially grown on SrTiO3 (STO) (001) substrates due to the isotropic in-plane mismatch strain [20]. In the present study, we reported interest magnetic anisotropies in such LSMO thin films on STO (001). In particular, unexpected double-shifted magnetization curves were observed in the LSMO thin films on STO (001) and it originated from the competition between common biaxial and emergent uniaxial magnetic anisotropies as described by Stoner–Wohlfarth (S–W) model. The uniaxial anisotropy in the LSMO films could be caused by an intrinsic antiferromagnetic (AFM) dead layer at the LSMO/STO interface by means of the exchange coupling effect [21], which could be affected by the oxygen vacancy concentration in LSMO thin films. This novel feature offers a deeper understanding of the magnetic anisotropy observed in LSMO films, as well as a potential method of tuning that anisotropy.

2 Materials and methods

A series of 16 nm thick LSMO thin films were epitaxially grown on single-crystalline STO (001) substrates by the pulsed laser deposition (PLD, AdNaNo, China) method from a stoichiometric LSMO target, with a growth temperature of 800 °C, and an oxygen background pressure of 200 mTorr (1 Torr = 133.322 Pa). An excimer laser with a wavelength of 248 nm, repetition rate of 5 Hz, and an energy density of 2 J/cm2 was used. After film growth, the samples were cooled to room temperature at a rate of 20 °C/min in an oxygen environment equivalent to that employed during deposition. The magnetization reversal processes of the LSMO thin films were measured at room temperature by a typical transverse magneto-optic Kerr effect (MOKE, Durham NanoMOKE2, UK) configuration. The microstructure of LSMO film was carefully examined by the reciprocal space maps (RSM, high-resolution X-ray diffraction of PANalytical X’Pert MRD, the Netherland) and the high-resolution transmission electron microscopy (HRTEM, JEOL2011, Japan).

3 Results

3.1 Magnetization measurement of LSMO thin films

To study the magnetic anisotropy of the films, angular-dependent in-plane hysteresis loops were measured as a function of the in-plane rotation angle α over the entire angular range. As the crystallographic directions STO [010] and STO [100] were identical, the angles α = 0° and 90° were defined corresponding to the STO [010] and STO [100] for simplification (the insets in Fig. 1). Generally, biaxial anisotropy is induced by the isotropic in-plane tensile strain imposed by the cubic STO (001) lattice and expected to be observed in LSMO thin films deposited on STO (001). Interestingly, our LSMO thin films presented an emergent uniaxial magnetic anisotropy, as shown in Fig. 1a. Three representative magnetization versus applied magnetic field (M–H) hysteresis loops acquired at α = 90°, 45°, and 0° are shown in Fig. 1b–d, respectively. Figure 1b clearly demonstrated that an unexpected double-shifted magnetization curve appeared along α = 90°, which was completely different from the hysteresis loops acquired along α = 45° and 0°.

Fig. 1
figure 1

(Color online) a Measured polar figures of the normalized remanent magnetization (M r /M s) for the LSMO thin films. b–d Kerr magnetization curves measured (dots) and corresponding curves simulated according to the S–W model (black lines) along α = 90° (b), 45° (c), and 0° (d). The arrows in the insets indicate the direction of the applied magnetic fields

Full size image

3.2 Micromagnetic modeling

Such double-shifted magnetization curves have already been studied by micromagnetic modeling, experimentally observed in some other magnetic bilayers, such as Co/NiMn, and explained from the perspective of phenomenological theory to be the result of competition between uniaxial magnetic anisotropy and biaxial magnetic anisotropy [22]. In order to understand the M–H hysteresis behavior observed in our LSMO thin films, the S–W model was used to simulate the M–H hysteresis loops of the films. By assuming a homogeneous magnetization in the ferromagnetic LSMO layer, the free-energy density f in the system with collinear uniaxial anisotropy (quadratic) and biaxial anisotropy (quartic) could be written as

$$ f = - M_{\text{s}} H\cos \left( {\theta - \alpha } \right) + K_{\text{u}} \sin^{2} \theta + K_{\text{b}} \sin^{2} \theta \cos^{2} \theta , $$
(1)

where M s is the saturation magnetization, which is 250 emu/cc (cc = cm3) at room temperature acquired from SQUID measurements (not shown here). K u and K b are the uniaxial anisotropy and biaxial anisotropy constants, obtained as 1.8 and 1.0 erg/cc, respectively, from a fitting of the analytical model to the experimental data in Fig. 1. θ and α are the angles of the magnetization vector and the applied magnetic field making with the axis of the uniaxial magnetic anisotropy, and α is the same as defined in our experiment. According to the S–W model, a double-shifted magnetization curve appears only when the values of K u and K b meet a rigorous requirement of 5K b > K u > K b > 0 [22]. As shown in Fig. 1, the shapes of the magnetization curves obtained from the S–W model fitted well with the experiments over the entire angular range, except that the coercive field strengths (H c) obtained were not always equal to those indicated by the experimental data. To validate this explanation further, the α-dependence of H c was investigated (Fig. 2). While over a range of α from 30° to 90°, the measured values of H c tended to agree with the results of the S–W model due to magnetic domain rotation, the large overestimation of H c in the range from 0° to 30° might be caused by anisotropy dispersion or domain wall motion [23], which were not taken into account in the simple S–W model.

Fig. 2
figure 2

(Color online) Angular dependence of H c for the LSMO thin films

Full size image

4 Discussion

To obtain a deeper understanding of the nature of this uniaxial magnetic anisotropy, we examined the structural information carefully. Figure 3a–c shows the reciprocal space maps around the (002), (013), and (103) STO reflections, respectively. The peaks corresponding to the films and to the substrate appeared for the same values of Q x, which illustrated that both films and substrate had the same in-plane parameter, and clearly revealed that the films were fully strained by the substrate. The films have been expected to be free from obvious structural defects, which was confirmed by HRTEM image given in Fig. 3d. The structural characterization revealed no observable difference in the strains along the [001] and [010] directions.

Fig. 3
figure 3

(Color online) a–c Reciprocal space maps (RSM) measured around the (002), (013), and (103) STO reflections, respectively. d A typical and amplified HRTEM images for the La0.7Sr0.3MnO3/SrTiO3 (001) (LSMO/STO) system. The dashed line in (d) indicates the sharp LSMO/STO interface

Full size image

Based on the literature [24], the presence of uniaxial anisotropy in the LSMO thin films could be due to several possible origins, such as the strain-induced uniaxial anisotropy in LSMO/NdGaO3 (NGO) owing to the non-equivalent strain imposed by the substrate, step-induced uniaxial anisotropy in LSMO/STO [17] owing to the formation of elongated structures along the step-edge direction, and the orthorhombic crystal structure induced in LSMO/(LaAlO3)0.3(SrAl0.5Ta0.5O3)0.7 (LSAT) [10]. However, based on the RSM and HRTEM results, the fully strained LSMO thin film has a tetragonal phase, and the strain is equal in the two in-plane directions. Besides, the easy and hard axes are always aligned with the STO [100] and STO [010] directions and have no relation to the step direction [25]. All these results indicated that none of the above-reported explanations were applicable to our case. Thus, the most reasonable explanation was an exchange coupling-induced uniaxial anisotropy acting between the FM LSMO and an AFM LSMO dead layer at the LSMO/STO interface. Interfacial dead layers of LSMO thin films have been studied with regard to a number of different substrates [26]. And the results displayed very different effects due to the different magnetic states of the dead layer, which could be tuned by a number of factors such as strain and oxygen deficiency [19].

An AFM dead layer with relatively large AFM anisotropy has been demonstrated for the LSMO/STO system [19], which induced an intrinsic exchange bias effect. More recently, even a spin glass dead layer has been detected in the LSMO/LaSrAlO4 (LSAO) system [9]. An AFM phase at the LSMO/STO interface has also been predicted from first principles [27, 28] and detected in other study [29]. These studies indicated that the dead layer at the interface could be an A-type AFM with sufficiently low AFM anisotropy under a particular tensile strain and oxygen deficiency. When the spins in the FM layer rotated, the spins in the AFM layer could be dragged and rotated at the same time, which induced the uniaxial anisotropy by quenching the FM along the AFM easy axis [21]. Other FM/AFM bilayer systems also reported this effect [22]. It was not easy to exactly demarcate the AFM dead layer in our LSMO thin films by TEM due to the almost undifferentiated lattice structure between the interface dead layer and the above FM layer. However, it could verify our explanation indirectly that the AFM easy axis always had a certain relationship with the crystallographic direction [25].

If the uniaxial anisotropy was induced by the AFM dead layer, the magnetic states of LSMO could be strongly tuned according to the degree of oxygen deficiency, and the hysteresis behavior should then be tunable according to the degree of oxygen deficiency. To further verify our hypothesis, the sample was placed back in the PLD chamber and annealed at 800 °C in an oxygen background pressure of 200 Torr for 10 min and cooled to room temperature at a rate of 20 °C/min in the same oxygen environment to partially eliminate the oxygen deficiency. The film was then measured by MOKE under the same conditions as employed previously, and the typical results are shown in Fig. 4. As expected, the double-shifted magnetization curves disappeared along the STO [100] crystallographic direction, and the LSMO thin film became nearly isotropic, accompanied by some decrease in H c along the STO [010] crystallographic direction, which was a significant feature exhibited in exchange coupling systems [21]. Therefore, our samples demonstrated that the magnetic anisotropy could be remarkably tuned according to the degree of oxygen deficiency and our hypothesis would be supported.

Fig. 4
figure 4

(Color online) Kerr magnetization curves of the annealed LSMO thin films measured along the SrTiO3 (STO) [100] (dots) and [010] directions (squares)

Full size image

5 Conclusion

In summary, we demonstrated that the unexpected double-shifted magnetization curves exhibited in single-layer FM LSMO thin films epitaxially grown on STO substrates were the result of the presence of a novel uniaxial anisotropy. The uniaxial anisotropy was induced by exchange coupling between the FM LSMO and the AFM LSMO dead layer at the LSMO/STO interface, and it was evidenced by the observation that the magnetic anisotropy of LSMO could be manipulated by the degree of oxygen deficiency.