Interface defects induced vertical magnetic anisotropy in Sr2FeMoO6 thin films

Applied Surface Science 422 (2017) 682–689

Contents lists available at ScienceDirect

Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc

Interface defects induced vertical magnetic anisotropy in Sr2 FeMoO6 thin films I. Angervo a,b,∗ , M. Saloaro a , H. Huhtinen a , P. Paturi a a b

Wihuri Physical Laboratory, Department of Physics and Astronomy, FI-20014 University of Turku, Finland University of Turku Graduate School (UTUGS), University of Turku, FI-20014 Turku, Finland

a r t i c l e

i n f o

Article history: Received 17 March 2017 Received in revised form 24 April 2017 Accepted 20 May 2017 Available online 31 May 2017 Keywords: SFMO thin films Magnetic force microscopy Magnetic domains Magnetic anisotropy

a b s t r a c t Pulsed laser deposition was used to fabricate high quality Sr2 FeMoO6 thin films on SrTiO3 and SrLaAlO4 single crystal substrates. The focus of our research has been on the magnetic properties of Sr2 FeMoO6 thin films and we have observed an outstanding shift from parallel in-plane magnetic anisotropy to perpendicular out-of-plane anisotropy, induced by the substrate. Our results also provide the first experimental evidence of magnetic stripe domain pattern in Sr2 FeMoO6 thin films. Compared with previous studies for magnetoresistive thin films, we propose a conspicuous mechanism behind the shift in magnetic anisotropy. Rather than being induced directly by the lattice mismatch, we find the possible contribution to perpendicular magnetic anisotropy from over-relaxation at the film substrate interface or from other crystalline defects such as low-angle grain boundaries. © 2017 Elsevier B.V. All rights reserved.

1. Introduction High Curie temperature (TC ), between 410–450 K, and 100% spin polarization along with magnetoresistive phenomenon have attracted a lot of interest to the double perovskite Sr2 FeMoO6 (SFMO) [1,2]. The magnetic attributes make SFMO an interesting material on its own, but also a valuable candidate as material for spintronics. Therefore, many studies have focused on the optimization of SFMO thin films, which are necessary for multilayer based applications. As a ferrimagnetic material, the magnetic domain structure is an important element for magnetic properties of Sr2 FeMoO6 thin films. Perhaps the most common embodiment of the effects of domains is the magnetic hysteresis, but domains are also the key component to understand any magnetic behaviour in ferromagnetic (ferrimagnetic) materials. Magnetic properties of SFMO, polycrystalline bulk and thin films, have been studied intensively, both theoretically and experimentally [3–11]. However, there appears to be very little research done to directly image magnetic domains in SFMO films. Only few studies have been carried out with magnetic force microscopy (MFM) [12,13]. Kalanda et al. have reported a correlation between single grains in SFMO thin films and domain structure as well as a correlation between clusters

∗ Corresponding author at: Wihuri Physical Laboratory, Department of Physics and Astronomy, FI-20014 University of Turku, Finland. E-mail address: ijange@utu.fi (I. Angervo). http://dx.doi.org/10.1016/j.apsusc.2017.05.171 0169-4332/© 2017 Elsevier B.V. All rights reserved.

in SFMO thin films and domain structure [12]. MFM has been used in various research projects to study magnetic domain structure in other magnetoresistive thin films, especially in La0.7 Sr0.3 MnO3 (LSMO) films, but also in other magnetoresistive materials [14–17]. In these materials, the common magnetic stripe domain pattern has been reported and the domain structure has been shown to be affected by the thin film thickness and choice of the substrate material [14,15,17]. Stripe like patterns indicate increased magnetic anisotropy [18]. Directly related to the magnetic domain structure, the magnetic anisotropy will also play a key role in magnetic phenomena in SFMO thin films. Magnetic anisotropy is usually divided into three separate categories: magnetocrystalline anisotropy, magnetoelastic anisotropy and magnetic dipolar anisotropy, all which contribute in the direction of the magnetic moment in thin films [19]. Magnetocrystalline anisotropy originates from the spin-orbit interaction and results with easy and hard magnetization axes relative to the lattice vectors. Magnetoelastic anisotropy refers to anisotropy where the direction of magnetization is altered by the structural stress in the crystalline lattice. Magnetic dipolar anisotropy, a.k.a. shape anisotropy, usually dominates the anisotropy in thin films and sets the magnetic moment to in-plane of the film surface. Magnetoelastic anisotropy, induced by the lattice mismatch between the substrate and film material, has been used to shift magnetic anisotropy from parallel to perpendicular of the film plane orientation [19–21]. Before the fabrication of SFMO spin valves the

I. Angervo et al. / Applied Surface Science 422 (2017) 682–689

phenomenon of magnetic anisotropy must be addressed, since anisotropy will have a role in multilayer spinvalves [22,23]. The substrate material and the film thickness have been shown to have a great importance on the structural and the magnetic properties of SFMO thin films [9,24–28] but, according to our knowledge no study has been conducted with MFM to study the effects of substrate material and the film thickness directly to magnetic domains. In this paper, magnetic anisotropy and domain structure of SFMO films on different substrates with different thicknesses has been studied. The results reveal a shift from parallel in-plane easy axis orientation to perpendicular in-plane orientation and also traces of stripe domain pattern. Additional magnetometric measurements provide more information about domain dynamics. The effects of pinning are also analyzed from the magnetic hysteresis measurements.

2. Experimental details Four SFMO thin films were deposited on SrTiO3 (STO) and SrLaAlO4 (SLAO) single crystal substrates, two films on each substrate with different thicknesses, with pulsed laser deposition ˚ for (PLD). The lattice parameters for STO are a = b = c = 3.901 Aand ˚ ˚ which produce the in-plane SLAO a = b = 3.756 Aand c = 12.636 A, lattice mismatch values with SFMO of −1.05% and −4.73%, respectively. The in-plane lattice mismatch between the substrate lattice parameter, asub , and the SFMO √ lattice parameter, aSFMO , is calculated using the formula  = ( 2asub − aSFMO )/aSFMO based on the values of SFMO lattice parameters published in [29]. The substrate is heated from room temperature to the deposition temperature at the rate of 25 ◦ C/min in 9 Pa Ar-atmosphere. Deposition temperature was 900 ◦ C, which has shown to result with smooth SFMO thin films [30,31]. We used XeCl excimer laser with laser wavelength of 308 nm and the pulse frequency adjusted to 5 Hz. Laser fluence was 1.41 J/cm2 . Film thickness was controlled by varying the pulse number. After the laser deposition, the annealing treatment was carried out by keeping the temperature constant for 10 min, before the temperature was decreased back to the room temperature with rate of 25 ◦ C/min. Thinner films were deposited with 500 and thicker films with 2000 PLD laser pulses. The thinner SFMO films fabricated on STO and SLAO with 500 pulses are named STO1 and SLAO-1 and the thicker films fabricated on STO and SLAO with 2000 pulses are named STO-2 and SLAO-2. SFMO target was prepared for pulsed laser deposition using solid state synthesis. Details for target preparation are reported elsewhere [31]. Surface structure and domain structure of the films were studied with an Innova atomic force microscope provided by Bruker. The atomic force microscopy (AFM) and the magnetic force microscopy (MFM) scans were performed at room temperature. The AFM imaging was done with 20 ␮m × 20 ␮m, 10 ␮m × 10 ␮m, 5 ␮m × 5 ␮m, 2 ␮m × 2 ␮m scan sizes from four different areas on the SFMO thin films using the tapping mode. The surface roughness was determined as the root mean square (RMS) roughness from 20 ␮m × 20 ␮m, 10 ␮m × 10 ␮m, 5 ␮m × 5 ␮m scans. The magnetic images were obtained from 20 ␮m × 20 ␮m, 10 ␮m × 10 ␮m and 5 ␮m × 5 ␮m scans with 25 nm, 50 nm, 100 nm and 200 nm liftheights. X-ray diffraction (XRD) measurements were done using a Philips X’Pert Pro MPD diffractometer with a Schulz goniometer. To check for the possible impurity phases and to determine the c-axis parameter,  − 2 measurements were done between 20◦ –114◦ . A detailed  − 2 scan was done for the (336) to determine the a-axis parameter and a texture  − scan for (204) peak at 2 = 57.106◦ to check the c-axis orientation. A detailed 2 −  scan was performed for the (204) peak and used for lattice defect analysis to obtain the scale of in-plane lattice distortion. The lattice parameter values are used

683

to calculate the strain in the films. Film thicknesses were determined with X-ray reflectivity measurements (XRR), conducted on STO-1 and SLAO-1 both fabricated with 500 pulses. The results indicate 30 nm thickness for the 500 pulse films on both STO and SLAO. In this thickness range, we assume linear growth rate, which means that 2000 pulses correspond to thickness of 120 nm, which is slightly less than in our earlier films grown with different growth parameters [10,28]. Besides MFM, we used a MPMS XL SQUID magnetometer provided by Quantum Design to study the magnetic properties. The zero field cooled (ZFC) and the field cooled (FC) magnetizations were measured as a function of temperature in 100 and 500 mT magnetic fields between 10–400 K temperature for all films. The Curie temperature was determined from the minimum of the first order derivative of 100 mT FC curve. Magnetic hysteresis loops were measured between ±1 T at 10 K. Saturation magnetization, Msat , and coercive field, Bc , were obtained from the hysteresis loops. The SQUID measurements were conducted twice, magnetic field aligned differently respect to the SFMO lattice vectors, for all the films with magnetic field parallel first to [110], the diagonal of a and b lattice parameters, and second to [001], c lattice parameter.

3. Results 3.1. Lattice strain and crystalline defects The  − 2 scans between 20◦ − 114◦ are presented for all our SFMO thin films in Fig. 1(a). The results show clear (00l) peaks arising from SFMO thin films and the substrates. No impurity phase peaks are observed since the additional the small peaks observed around 44◦ and 52◦ arise from the sample holder. The pole figures of the texture analysis showed clear (204) and (132) peaks as is expected for fully textured c-axis oriented SFMO films [8,30]. The XRD results show films to be phase pure, highly textured and c-axis oriented. 2 −  scans for (204) are presented in Fig. 1(b) and (c) for STO2 and SLAO-2, respectively. 2 −  scans were also performed for thinner films but due to slight overlapping between the peaks of STO substrate and the SFMO film, accompanied by the smaller intensity from the thinner SFMO films, the analysis of the (204) peak in STO-1 becomes inconvenient. The broadening, in  angle, results from low angle grain boundaries in the in-plane orientation of the crystal lattice. The broadening was determined as −FWHM (full width at half maximum) by fitting a Gaussian distribution function with the section of 2 −  scan corresponding to the 2 value of the highest intensity of the SFMO (204) peak. The Gaussian fit resulted with  − FWHM = 0.692◦ and  − FWHM = 0.914◦ for STO-2 SLAO-2, respectively. SLAO-1 showed  − FWHM = 1.013◦ . These results are consistent with our previous work also showing larger  − FWHM values and therefore indicating larger spread of in plane crystalline orientations, i.e. more low angle grain boundaries, in the SFMO thin films fabricated on SLAO substrate [10]. Lattice parameters a and c were obtained by fitting a Gaussian distribution function to (336)- and (008)-peaks in the  − 2 scans, respectively. The results are presented in Fig. 1(d). The dotted lines ˚ cSFMO = 7.893 A˚ [29]. The represent the values of aSFMO = 5.575 Aand results show that thinner STO-1 film lattice parameter deviates from bulk values and film is strained due to compressive in-plane lattice mismatch, −1.05%, between the substrate and SFMO, showing strain value of −0.86%. The in-plane and out-of-plane strain, a and c , were calculated as l = (l − lSFMO )/lSFMO , where l is the SFMO lattice parameter, a or c, obtained from our XRD results and lSFMO represents the unstrained SFMO lattice parameter according to [29]. Once the thickness increases, the SFMO film on STO becomes more relaxed, with a value of −0.46%, and the lattice

684

I. Angervo et al. / Applied Surface Science 422 (2017) 682–689

Fig. 1. (a)  − 2-results obtained for our SFMO thin films show SFMO (00l) peaks, pointed with the Miller indices, along with the substrate peaks from STO and SLAO, marked with • symbol. The peaks from the sample holder are marked with  symbol. (b) and (c) Show the results of 2 −  scans for STO-2 and SLAO-2, respectively. (d) Presents the lattice parameters in SFMO films as a function of film thickness. Percentages are the values of in-plane strain, a and out-of-plane strain, c . Table 1 The consolidated results for structural measurements. SFMO film

 (%)

a (%)

c (%)

˚ a (A)

˚ c (A)

 − FWHM (◦ )

RMS (nm)

STO-1 STO-2 SLAO-1 SLAO-2

−1.05 −1.05 −4.73 −4.73

+0.90 +0.60 −0.13 −0.11

−0.86 −0.46 +0.05 +0.03

5.526 5.550 5.578 5.577

7.964 7.940 7.883 7.884

− 0.692 1.013 0.914

1.9 5.7 2.0 2.9

parameters in STO-2 are closer to the bulk values. On SLAO substrate, both SFMO films show lattice parameters values very close to the strain free values with only 0.05% and 0.03% in-plane strain, despite of the high in-plane lattice mismatch of −4.73% between the film and the substrate. This indicates some kind of interface relaxation in SLAO-1 and SLAO-2. It seems that the SFMO film becomes strain free immediately at the interface and grows with only minor strain. Our previous studies have shown the relaxation in the film substrate interface leading to the increase of in-plane lattice parameters despite of the compressive strain from the substrate, when the lattice mismatch is large enough [10]. In STO-1 and STO-2 strain remains, but STO-2 becomes more relaxed as the thickness increases. Similar results have been reported previously for SFMO thin films on different substrates [25,27,28]. Since XRD provides results from the film as an average obtained through entire depth of the film, it is likely that surface layers are already free of strain in thicker STO-2 film, because SFMO thin films, with thickness between 80 nm and 120 nm, are reported being fully relaxed [24,11]. The results for lattice parameters, lattice mismatches and strain values are also presented in Table 1. 3.2. Magnetic anisotropy The valuable magnetic properties were characterized with SQUID measurements and ferromagnetic transition above room temperature is confirmed. Fig. 2(a) and (b) present the ZFC/FC measurements results for STO-2 and SLAO-2, in 100 mT magnetic field, parallel to the SFMO [110] and [001] lattice vector, respectively. This means that the magnetic field is in plane of the film surface in Fig. 2(a) and out of plane in Fig. 2(b). The ZFC/FC measurements are normalized by dividing the magnetization values with magnetization at 10 K. The TC was obtained as the minimum of the first order derivative and TC values are listed in Table 2. The presented values are obtained from [110] field configuration, since no difference was observed in TC between these two configurations. Both films on SLAO have the TC value of 353 K. SFMO films on STO show an increase in TC from 350 K to 359 K with the increase of the film thickness. The TC of 350 K and above, determined from the minimum of the first order derivative, is quite high compared to some of our previous results, where we have observed the highest TC up to

340 K [10,11,28]. However, we have also reported higher TC values of 363 K in [31]. Saturation magnetization was studied with magnetic hysteresis measurements. The magnetic hysteresis loops obtained for STO-2 and SLAO-2 between ±0.5 T measured at 10 K, magnetic field parallel to the [110] and [001] lattice vector, are presented in Fig. 2(c) and (d), respectively. Diamagnetic background, arising from the substrate and the sample holder, has been subtracted from the results. The results for Msat are presented in Table 2. The presented values for SFMO films on STO are obtained from [110] configuration and for the films on SLAO values are obtained from [001] configuration. This is because the saturation of the magnetization in the films depends on the configuration. The saturation field is smaller for the films on STO in [110] configuration and for the films on SLAO in [001] configuration. The highest Msat value, 1.92 B /f.u., reported here is smaller but close to previously reported values around 2.0 B /f.u. [8,10,11,28,30]. The increase of TC and decrease of Msat are linked to crystalline defects, anti-site disorder (ASD) and oxygen vacancy concentration [3–7,11]. ASD refers to the misplacement of Fe and Mo ions, where Fe has switched position with Mo ion in the SFMO lattice. Higher Msat would suggest that the SFMO films on SLAO have smaller amount of ASD or oxygen vacancies [7,11]. The slightly higher TC and smaller Msat in STO-2 compared to other films could be related to increased oxygen vacancy concentration, since oxygen vacancies have been seen to increase TC and decrease Msat [7,11]. Structural characterization revealed higher strain in films on STO substrate, while both films on SLAO are almost entirely free of strain. According to previous theoretical analysis, the strain, induced by the lattice mismatch, is seen to be related to the ASD and oxygen vacancies stabilizing the films from formation of both defects [11]. Even though, ASD and oxygen vacancies are obviously present in our films, the strain values in our films are so small that the stabilizing effect of strain cannot alone explain the magnetic results. The effect is most likely overpowered by other mechanisms. Lattice mismatch induced defects, like low angle grain boundaries and dislocations, have stronger effect on the magnetic results based on our earlier work and explain our results here [10]. We have also observed a small ferromagnetic signal in STO-1 and STO-2 above TC . This suggests that some kind of ferromagnetic impurity trace with high magnetic susceptibility, below XRD detection limit of 1%

I. Angervo et al. / Applied Surface Science 422 (2017) 682–689

685

Fig. 2. ZFC/FC magnetization for STO-2 and SLAO-2 in 100mT field, magnetic field parallel (a) and perpendicular (b) to the film plane. Magnetic hysteresis between ±0.5 T for STO-2 and SLAO-2 measured at 10 K, magnetic field parallel (c) and perpendicular (d) to the film plane. Small peaks in ZFC curves at 50 K in (b) arise most likely from paramagnetic signal of oxygen.

Table 2 The consolidated results for magnetic measurements. SFMO film

TC (K)

Msat (B /f. u .)

Mirr (110) (B /f. u .)

Mirr (001) (B /f. u .)

Bc (110) (mT)

Bc (001) (mT)

STO-1 STO-2 SLAO-1 SLAO-2

350 359 353 353

1.14 1.43 1.37 1.92

0.219 0.208 0.184 0.192

0.011 0.055 0.318 0.526

32 39 23 20

40 15 69 62

of total sample volume, might be present in SFMO films on STO or it may arise from impurity in the measurement configuration. This signal is negligible compared to the signal from SFMO, but more profound in STO-1. Therefore, we address the possibility that the results for STO-1 might be affected by the small impurity signal due to the smaller film thickness. The ZFC/FC magnetization and the magnetization hysteresis have so far provided information about the TC and the Msat , which are very important properties for the quality of SFMO. The same measurements are now analyzed to provide further information about domain dynamics in the SFMO films. Comparison of the overall shape of the ZFC/FC and hysteresis curves performed with different magnetic field configuration reveals clear differences. When the field is in [110] configuration the ferro-paramagnetic transition appears sharper and magnetization is preserved better in SLAO-2 before the ferro-paramagnetic transition when temperature increases. However, in Fig. 2(b), with magnetic field parallel to [001] lattice vector, the similar tendencies can be obtained in STO-2 when compared to SLAO-2. The normalized ZFC/FC measurements suggest that magnetization irreversibility at low temperature has also increased dramatically in SLAO-2 when magnetic field configuration is changed from [110] to [001]. Magnetization irreversibility, Mirr = MFC (10K) − MZFC (10K), was obtained from ZFC/FC measurements and is presented in Table 2 for all the films and for both measurement configurations. Mirr (110) refers to the configuration with magnetic field parallel to the [110] lattice vector and Mirr (001) magnetic field parallel to the [001] lattice vector. Mirr values are above 0.2 B /f.u. for both films on STO in [110] configuration, but drop below 0.1 B /f.u., when the magnetic field is changed to the [001] configuration. The tendency is opposite

for the films on SLAO and Mirr values increase from below 0.2 B /f.u. values to above 0.3 B /f.u. values, when changing the magnetic field configuration from [110] to [001]. Small peaks in ZFC curves at 50 K arise most likely from of oxygen. The signal is detected only in [001] configuration, but this is only because sample chamber may capture a small amount of oxygen during the sample installation. The magnetization shows typical paramagnetic 1/T tendency at low temperature close to 10 K. However, oxygen has a complex susceptibility temperature dependence at low temperature. The peaks around 50 K in ZFC/FC measurements arise most likely from melting of solid oxygen accompanied by para-antiferromagnetic phase transition [32,33]. Oxygen has the melting point of 54.4 K [39]. The traditional paramagnetic 1/T tendency arises most likely from the sample holder or the substrate, since oxygen is paramagnetic only above para-antiferromagnetic transition temperature. Hysteresis loops show also clear differences both between the different films and the different magnetic field orientations. When comparing the films in Fig. 2(c), SLAO-2 loop is skewed and it shows saturation at higher magnetic field and clearly smaller coercivity compared to STO-2. These results are for configuration of magnetic field parallel to the lattice vector [110]. Once the magnetic field is changed into [001] orientation, as shown in Fig. 2(d), SLAO-2 shows higher coercivity, saturation at lower magnetic field and the loop appears less skewed when compared to STO-2 and also to SLAO2 in [110] field configuration. The changes in STO-2 are opposite. The comparison of films with different thicknesses shows that the changes in skewness between different magnetic field configurations in SLAO-1 are similar to SLAO-2, but in STO-1 there seemed to be no significant change in hysteresis loop skewness when field orientation is changed.

686

I. Angervo et al. / Applied Surface Science 422 (2017) 682–689

Fig. 3. The 5 ␮m × 5 ␮m AFM images obtained from SFMO thin films on different substrates(first row). The MFM images obtained at room temperature with 50 nm (second row) and 100 nm lift heights (third row).

The coercive field, Bc , obtained from hysteresis measurements is presented in Table 2 for all the films and for both measurement configurations. The results for different field configurations are indicated similarly as for Mirr by [110] and [001]. Both films on SLAO show an increase in the Bc from to 0.02 T to over 0.06 T, when the field configuration is changed from [110] to [001]. However, the STO-1 shows an increase in the Bc from 32 mT to 40 mT, while in STO-2 the Bc decreases from 39 mT to 15 mT. These and previous results will be discussed more thoroughly in a following discussion section.

3.3. Surface and magnetic domain structures Our results have now proved the high quality of our SFMO films. The high TC and ferrimagnetic signal well above room temperature allows us to detect long range magnetic forces with MFM and to analyze the domain structure. The smooth surface in thin films is a valuable result on its own but also crucial for MFM measurements. We begin by analyzing the surface structure results. The surface microstructure of SFMO thin films was studied with AFM. The subfigures in Fig. 3 labeled with AFM show the topography from SFMO films on STO and SLAO. The AFM results show smooth surface structure for all SFMO films, but on SLAO the thicker SFMO film is significantly smoother than on STO. The RMS roughness values for our films were 1.9 nm for STO-1, 2.0 nm for SLAO-1, 5.7 nm for STO-2 and 2.9 nm for SLAO-2, respectively. This clarifies that the films on SLAO are the smoothest, but all our SFMO films here have a relatively smooth surface structure. However, STO-2 has a rather high RMS value, which is roughly twice as high compared to the other three films. Similar smooth surface structures, with the smallest RMS roughnesses between 1 nm and 2 nm, have been previously reported for SFMO thin films [27,30,34]. Particle size on the film surface is also smaller in SFMO films on SLAO. Rectangular structures with similar orientation appear on the surface

of the thinner films on both substrates. The in-plane side lengths of the rectangular structures in SLAO-1 are approximately between 0.2 ␮m and 0.8 ␮m and between 0.3 ␮m and 0.6 ␮m in STO-1. The out-of-plane height is between 6 nm and 17 nm in SLAO-1 and between 5 nm 12 nm in STO-1. The similar rectangular structures have also been observed earlier in SFMO films [27,34–37] and they have been associated with island-like film growth [34,36]. It is possible that the film growth mechanism is slightly modified when the film thickness increases and therefore we do not see rectangular structures on STO-2 and SLAO-2. The MFM was obtained with same measurement configuration as the AFM and the results are presented in Fig. 3. In the presented MFM images, the lift height during the magnetic imaging was either 50 nm or 100 nm. With 25 nm lift height, MFM signal was greatly disturbed by the direct interaction with the surface in STO-2. This happened also with the other samples, but not as much as with STO2. This can be expected due to rougher surface microstructure with clearly higher color scale value of 40.9 nm in the AFM-image of STO2. When the lift height was increased to 200 nm, the magnetic signal was weak with relatively poor resolution. Hence, good quality MFM data was obtained with 50 nm and 100 nm lift heights, which is in good agreement with results observed in [12]. In thicker SFMO films, a clear MFM signal was detected with both lift heights. In these films, magnetic domains in the surface appear to include multiple grains. However, in SLAO-2, the magnetic structures appear to be smaller, meaning that the number of magnetic domains per unit area is larger. When the lift height is increased from 50 nm to 100 nm the resolution in the MFM images decreases and the details become less apparent, but the differences in MFM results seen with lower 50 nm lift height between STO-2 and SLAO2 are still visible. The thinner SFMO films also exhibit magnetic MFM signal. Domains appear to be more irregular and smaller in thinner films compared to thicker SFMO films. In STO-1, some of the smaller particles on surface give rise to signal seen as spots

I. Angervo et al. / Applied Surface Science 422 (2017) 682–689

687

Fig. 4. (a) An example of Fourier transform profile obtained with ImageJ-program for STO-2 and SLAO-2. (b) 5 ␮m × 5 ␮m MFM image from SLAO-2 featuring traces of stripe domain pattern.

with high contrast. In SLAO-1, there were no similar type of small particles on the surface but some of the rectangular structures can be identified from the magnetic signal. Once the lift is increased, the MFM signal is still detected but only barely, been also clouded with random noise. Due to smaller thickness and therefore the smaller magnetic moment in the sample, the magnetic signal is almost but not totally lost in the thinner films with higher lift heights. It is crucial to acknowledge that MFM signal corresponds to not only the magnetic signal at the surface, but also to the signal below the surface layer. Therefore, the domain structure seen in MFM images is an average signal through the whole film and the differences in MFM images between films with different thicknesses might be exaggerated, because the details arising from smaller features in the domains in STO-2 and SLAO-2 become less apparent. To study MFM images from thicker SFMO films in more detail, we performed fast Fourier transforms to 5 ␮m × 5 ␮m and 10 ␮m × 10 ␮m sized MFM images with 50 nm and 100 nm lift heights using ImageJ program [38]. The idea behind this is to find an average size for magnetic structures seen in the MFM results. The Fourier transform produces a 2D function, where the amplitude corresponds to the amount of specific periodicity, or in this case size, present in the MFM data. The 360◦ integral radial profile was taken respect to the origin of the 2D function. The radial profiles of the transforms are used for size analysis. An example of radial profiles for 100 nm lift height and 10 ␮m × 10 ␮m size images is presented in Fig. 4(a) for STO-2 and SLAO-2. Fourier transform profile shows a maximum, which corresponds to approximately 2 ␮m and 1 ␮m diameter structures in the MFM images for SFMO film on STO and SLAO, respectively. This result comes directly from different sized structures seen in MFM images. This gives a value for comparison of magnetic structure and supports the conclusion of smaller domain size in SLAO-2. Fig. 4(b) shows another MFM image obtained from SLAO-2 with 100 nm lift height. The figure shows clear features of stripe domain patterns, which have been observed before in other magnetic thin films [14–17], but not in SFMO thin films. Stripe domains are lengthened structures of domains, which have similar orientation along the thin film. Similar stripe features are also visible in 50 nm lift height SLAO-2 image in Fig. 3, but they are more clouded, most likely due to magnetic distortion present in the film. The stripe domain structure in thin films is associated with perpendicular magnetization [15,18,39], which indicates increased magnetic anisotropy in SLAO-2. 4. Discussion Generally, higher Bc and higher Mirr , suggest stronger domain pinning, which can result from structural defects [40]. We have previously observed higher amount of structural defects in SFMO

films fabricated on SLAO [10]. These films showed skewness in the hysteresis loops and saturation in significantly higher fields and low remanence magnetization and it was pointed that skewness could be due to stronger domain pinning. Hysteresis loop here for SLAO-2 is more skewed when compared to the loops of STO-2 and SLAO-2 saturates in slightly higher magnetic field, when magnetic field is parallel to [110] lattice vector. However, once the magnetic field is changed parallel to [001] lattice vector, more skewness, along with smaller Bc and Mirr , is seen in STO-2 compared to SLAO-2. The magnetic hysteresis is a result of total pinning of magnetic domain wall motion. The domain wall motion consists of reversible and irreversible component, which play a role in any ferromagnetic system [40]. The observed Bc , Mirr and skewness can all be understood through domain pinning and changes in magnetic anisotropy. Skewness in the hysteresis loops accompanied by the small Bc suggest that measurements are done magnetic field parallel to the hard magnetization axis, while less skewed loops with higher Bc will result from measurements with magnetic field along the easy axis of magnetization [39]. This means that the magnetic easy axis is different between STO-2 and SLAO-2, as schematically shown in Fig. 5. This is supported by the MFM measurements, which revealed traces of stripe domains in SLAO-2 suggesting perpendicular magnetic anisotropy. No traces of stripe domains were observed in STO-2. Magnetization easy axis for STO-2 lies closer along the [110] and for SLAO-2 closer along the [001] lattice vector. The hysteresis results suggest similar magnetic anisotropy in both films on SLAO despite the thickness. The differences between hysteresis measurements in different magnetic field configurations in STO-1 were not as clear as in STO-2, which could possibly indicate a difference in magnetic anisotropy between STO-1 and STO-2. However, the small possible impurity signal may affect the magnetic measurements in STO-1, the arguments about the anisotropy in STO-1 should be approached with caution. Magnetic anisotropy has been addressed in previous studies in magnetoresistive thin films, for example in LSMO, but also in SFMO [20,21,41,42]. Studies report the effect of magnetocrystalline and strain induced magnetic anisotropy and argue that substrate induced lattice mismatch results with the differences in anisotropy [20,21]. The strain induced magnetic anisotropy, magnetoelastic anisotropy, is well known among magnetocrystalline and shape anisotropy, which all contribute to the direction of magnetization. The shape anisotropy sets the magnetization parallel to the thin film plane. The magnetoelastic effect has been used to turn anisotropy parallel to the normal of the thin film plane [21]. Our hysteresis results are quite similar to the results presented by Du et al. for SFMO [21]. The crucial difference is that the perpendicular magnetic anisotropy is observed in SLAO-1 and SLAO-2 which, according to our XRD results, are strain free or exhibit only minor strain, whereas earlier results report perpendicular magnetic

688

I. Angervo et al. / Applied Surface Science 422 (2017) 682–689

Fig. 5. The schematic image of the main subjects that produce the in-plane and out-of-plane magnetic anisotropy. Perpendicular magnetic anisotropy and stripe domain pattern were observed for SLAO-2 while in-plane magnetic anisotropy and no stripe domain were observed for STO-2. When compared to STO-2, SFMO films on SLAO also showed more in-plane lattice distortion, as can be seen from the irregular lattice structure on the cleaved terrace.

anisotropy being induced by tensile strain [21]. The small elongation of the in-plane lattice vectors, seen in Fig. 1(d), suggests, that some level of strain could be present in SLAO-1 and SLAO-2 and this might have some contribution to the perpendicular magnetic anisotropy in our samples. The immediate over-relaxation at the film substrate interface could also produce a sufficient tensile strain to turn magnetization easy axis from in-plane to out-of-plane respect to the film surface. Since XRD provides results, which correspond to average signal of the sample, the strained interface layer must be negligible in size, compared to the total film volume. This is because both SLAO-1 and SLAO-2, with significant difference in film thickness, appear to be strain free. This could mean that overrelaxation accompanied by tensile strain serve as trigger for the perpendicular magnetic anisotropy. In order to reach a fully noncontroversial discussion with our results, we also wish to address the possibility that both SFMO films, SLAO-1 and SLAO-2, are free of strain. As has been indicated by our XRD results. Our XRD results have not only shown that SFMO films, with significant range in thickness, on SLAO are essentially free of strain, but also exhibit larger amount of in-plane lattice distortion, such as low-angle grain boundaries, when compared with film on STO. This is represented in Fig. 5 with the more irregular grid in the SFMO film on SLAO substrate. Since shape anisotropy usually dominates the magnetic anisotropy in thin films and sets the magnetic moment parallel to the film plane, this effect must be overcome. The magnetocrystalline anisotropy is linked to spin-orbit interaction and gives rise to easy and hard magnetization axes relative to the lattice vectors. If a magnetic domain, containing structural inplane lattice distortion, is formed, magnetic moment is forced to align slightly out from the easy axis. This will increase the amount of anisotropy energy within the domain. Therefore, a more favorable direction for magnetization can be found in the normal of the plane surface. We believe that larger amount of lattice mismatch induced defects in the in-plane lattice orientation in SLAO-2 and SLAO-1 may give rise to perpendicular magnetic anisotropy in our films. It is also possible that the crystalline defects disturb the long range shape anisotropy in SFMO films on SLAO and thus ease the formation of perpendicular magnetic anisotropy. The hysteresis and the ZFC/FC measurements showed larger Bc and Mirr in SFMO thin films on SLAO compared to the films fabricated on STO. Larger Bc and Mirr are both characteristics of stronger domain pinning resulting from frictional forces in the sample, preventing the change in magnetization. The structural defects, impurities, grain boundaries, can act as pinning centers for magnetic domains. We found in the AFM/MFM results that SLAO-2 had showed smaller magnetic domain structures and particles on the surface were smaller. The XRD results also showed larger amount of lattice distortion in SLAO-2. This indicates larger amount of pinning centers for SLAO-2 and larger amount of active magnetic domain walls acting with the pinning centers. Due to weak magnetic signal in MFM, it is difficult to draw conclusions about the domain size in thinner films. However, based on the magnetometric measurements the domain pinning appears respectively similar to the films

with the same substrate. The easy axis measurements show higher Bc values of 62 mT and above for SFMO films on SLAO compared to the highest easy axis value of 40 mT for SFMO film fabricated on STO. We also measured higher Mirr easy axis values for films on SLAO compared to the easy axis values for SFMO films on STO. However, Mirr values are also affected by the smaller Msat in STO-1 and STO-2. 5. Conclusions We have fabricated two sets of SFMO thin films on STO and SLAO single crystal substrates. The films exhibit a clear ferrimagnetic signal with high TC , well above room temperature, and good crystalline quality. We have focused our research on magnetic properties and observed a shift from parallel in-plane magnetic anisotropy to perpendicular anisotropy along with magnetic stripe domains. The perpendicular magnetic anisotropy is observed in SFMO film fabricated on SLAO with high compressive in-plane lattice mismatch between the film and the substrate. Our results indicate that the shift in anisotropy is induced by tensile over-relaxation at the film substrate interface or by lattice mismatch induced structural defects which were confirmed with XRD measurements. Also, magnetometric measurements, conducted in the easy axis configuration, indicate stronger domain pinning in the films fabricated on SLAO substrate. Acknowledgments The Jenny and Antti Wihuri foundation and University of Turku Graduate School are acknowledged for financial support. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]

K.-I. Kobayashi, et al., Nature 395 (1998) 677. D.D. Sarma, et al., Phys. Rev. Lett. 85 (2000) 2549. J. Navarro, et al., Phys. Rev. B 67 (2003) 174416. B.J. Park, et al., J. Magn. Magn. Mater. 272–276 (2004) 1851. A.S. Ogale, S.B. Ogale, R. Ramesh, T. Venkatesan, Appl. Phys. Lett. 75 (1999) 537. J.L. Alonso, et al., Phys. Rev. B 67 (2003) 214423. M. Hoffmann et al., arXiv:1504.02629 (2015). M. Saloaro, S. Majumdar, H. Huhtinen, P. Paturi, J. Phys. Condens. Matter 24 (2012) 366003. M. Saloaro, S. Majumdar, H. Huhtinen, P. Paturi, EPJ Web of Conferences vol. 40 (2013) 15012. M. Saloaro, et al., J. Phys. Condens. Matter 27 (2015) 386001. M. Saloaro, et al., ACS Appl. Mater. Interfaces 8 (2016) 20440. N.A. Kalanda, et al., J. Electron. Mater. 45 (2016) 3466. N. Kumar, et al., Mater. Lett. 118 (2014) 200. Y. Wu, et al., Appl. Phys. Lett. 75 (1999) 2295. J. Dho, N.H. Hur, J. Magn. Magn. Mater. 318 (2007) 23. Q. Lu, C. Chen, A. Lozanne, Science 276 (1997) 2006. J. Dho, et al., Appl. Phys. Lett. 82 (2003) 1434. C. Kittel, Phys. Rev. 70 (1946) 965. M.T. Johnson, et al., Rep. Prog. Phys. 59 (1996) 1409. C. Kwon, et al., J. Magn. Magn. Mater. 172 (1997) 229. C. Du, et al., Phys. Rev. Lett. 110 (2013) 147204. K. Lee, et al., J. Appl. Phys. 107 (2010) 073905. S. Dijken, et al., Appl. Phys. Lett. 87 (2005) 022504.

I. Angervo et al. / Applied Surface Science 422 (2017) 682–689 [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34]

T. Fix, et al., J. Appl. Phys. 98 (2005) 023712. R. Boucher, J. Phys. Chem. Solids 66 (2005) 1020. R. Boucher, J. Magn. Magn. Mater. 301 (2006) 251. H. Jalili, N.F. Heinig, K.T. Leung, J. Chem. Phys. 132 (2010) 204701. I. Angervo, et al., 20th International Conference on Magnetism vol. 75 (2015) 1011. S. Nakamura, K. Oikawa, J. Phys. Soc. Jpn. 72 (2003) 3123. P. Paturi, M. Metsänoja, H. Huhtinen, Thin Solid Films 519 (2011) 8047. I. Angervo, et al., Appl. Surf. Sci. 396 (2017) 754. Y.A. Freiman, et al., Phys. Rep. 401 (2004) 1. S. Gregory, Phys. Rev. Lett. 40 (1978) 723. R.P. Borges, et al., Thin Solid Films 429 (2003) 5.

689

[35] H. Deniz, et al., J. Mater. Sci. 50 (2015) 3131. [36] A. Venimadhav, M.E. Vickers, M.G. Blamire, Solid State Commun. 130 (2004) 631. [37] J.H. Song, J.-H. Park, Y.H. Jeong, J. Appl. Phys. 97 (2005) 46105. [38] C.A. Schneider, W.S. Rasband, K.W. Eliceiri, Nat. Methods 9 (2012) 671. [39] J.M.D. Coey, Magnetism and Magnetic Materials, Cambridge University Press, 2010. [40] D. Jiles, D. Atherton, J. Magn. Magn. Mater. 61 (1986) 48. [41] Y. Suzuki, H.Y. Hwang, S.-W. Cheong, R.B. van Dover, Appl. Phys. Lett. 71 (1997) 140. [42] N. Kumar, et al., Integr. Ferroelectr. 157 (2014) 89.