Analysis of electronic structure and its effect on magnetic properties in (001) and (110) oriented La 0.7 Sr 0.3 MnO 3 thin films This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2013 J. Phys.: Condens. Matter 25 376003 (http://iopscience.iop.org/0953-8984/25/37/376003) Download details: IP Address: 130.232.105.177 The article was downloaded on 22/08/2013 at 11:34 Please note that terms and conditions apply. View the table of contents for this issue, or go to the journal homepage for more Home Search Collections Journals About Contact us My IOPscience
Transcript
Analysis of electronic structure and its effect on magnetic properties in (001) and (110)
oriented La0.7Sr0.3MnO3 thin films
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
IOP PUBLISHING JOURNAL OF PHYSICS: CONDENSED MATTER
J. Phys.: Condens. Matter 25 (2013) 376003 (9pp) doi:10.1088/0953-8984/25/37/376003
Analysis of electronic structure and itseffect on magnetic properties in (001) and(110) oriented La0.7Sr0.3MnO3 thin films
S Majumdar1,2, K Kooser3, T Elovaara1,4, H Huhtinen1, S Granroth3 andP Paturi1
1 Wihuri Physical Laboratory, Department of Physics and Astronomy, University of Turku, FI-20014Turku, Finland2 Nanomagnetism and Spintronics Group, Department of Applied Physics, Aalto University School ofScience, PO Box 15100, FI-00076 Aalto, Finland3 Laboratory of Materials Research, Department of Physics and Astronomy, University of Turku,FI-20014 Turku, Finland4 The National Graduate School in Material Physics (NGSMP), Helsinki, Finland
Received 5 July 2013, in final form 5 August 2013Published 21 August 2013Online at stacks.iop.org/JPhysCM/25/376003
AbstractEpitaxial thin films of half-metallic oxide La0.7Sr0.3MnO3 (LSMO) have been grown in twocrystalline orientations, one with the c-axis out-of-plane, the (001) orientation, and one withthe c-axis in-plane, the (110) orientation. For the (110) oriented growth, there is no polardiscontinuity at the substrate–film interface and hence no dead layer formation, whichimproves ferromagnetic interaction in the LSMO, making it highly attractive for spintronicapplications. In our experiments, with x-ray diffraction, x-ray photoelectron spectroscopy andmagnetic measurements, we have demonstrated that in the (110) oriented LSMO the lattice ismore relaxed, leading to less deformation of electronic density around the La atom or in theMnO6 octahedra. This improved crystal and electronic structure improves the ferromagneticproperties of the films, making the Curie temperature higher by almost 15 K, which is ofpotential interest for spintronics. However, substrate strain induced magnetic anisotropycauses domain formation with out-of-plane components in these films, which poses someconcern for practical spintronic devices.
(Some figures may appear in colour only in the online journal)
1. Introduction
Rare-earth manganese oxides with the general formulaR1−xAxMnO3 (where R is the trivalent rare-earth element andA is a divalent metal) are one of the most exciting classes ofcorrelated electron systems, where the coexistence of differentphases leads to exotic physical properties such as colossalmagnetoresistance [1]. Among the different manganites,La0.7Sr0.3MnO3 (LSMO) is one of the most studied materials,due to its 100% spin polarized conduction band making it apromising candidate for spintronic applications [2, 3]. LSMOis a half-metallic oxide where the spin up and down bands
are completely separated near the Fermi level in its lowtemperature ferromagnetic (FM) phase [4]. Although LSMOundergoes paramagnetic to ferromagnetic transition close to350 K, it already starts to lose its spin polarization, especiallyon the surface, above 200 K. Therefore at room temperatureonly a very small number of spin polarized carriers areexpected in the spintronic devices from LSMO electrodes,which poses a serious challenge for the functionality of oxidebased spintronic components for industrial applications [5].
Also, the tunnelling magnetoresistance (TMR) signalsfrom LSMO based magnetic tunnel junction devices areoften far too low compared to what is expected from
J. Phys.: Condens. Matter 25 (2013) 376003 S Majumdar et al
Julliere’s formula [6] from a fully spin polarized electrode.Therefore the question arises whether the electrodes arereally fully spin polarized. Some studies have suggested thatLSMO loses its half-metallicity due to interaction with theSrTiO3 (STO) substrates [7, 8]. Due to polar discontinuitytheory [9, 10], (001) oriented LSMO is deposited as alternatelayers of La0.67Sr0.33O(0.67)+ and MnO(0.67)−
2 , while STOlayers are composed of alternate SrO and TiO2 planes.The LSMO layers are charged, while the STO has nopolarization of the layers. At the STO–LSMO interface,therefore, a polar discontinuity occurs which can lead toan electronic re-distribution. This electronic re-distributionresults in a reduced or enhanced doping, depending onthe interfacial layer. This modified doping at the interfaceleads to either under- or overpopulation of eg states. Inboth cases, the double exchange (DE) mechanism, essentialfor metallic conduction and ferromagnetism, is hindered.Interface engineering between LSMO and STO has beensuggested as one of the solutions to this problem [11, 12].
LSMO on (110) oriented STO substrates shows acompletely different interfacial arrangement. STO in this caseis a stack of SrTiO4+ and O4−
2 . LSMO (110) has layers ofLa0.67Sr0.33MnO4+ and O4−
2 , and due to this new constitution,there is no polar discontinuity and hence no magnetic deadlayer formation in this case. Some experiments show thatLSMO (110) oriented films indeed possess a higher magneticmoment and improved ferromagnetism compared to the (001)oriented films [13–15]. However, the electronic configurationsof the constituent atoms in these two cases have not yetbeen studied in detail. Hence, in the present paper, we havestudied the electronic configurations of the constituent La,Sr, Mn and O atoms of LSMO (001) and (110) orientedfilms using x-ray photoelectron spectroscopy (XPS) withtwo different incident photon energies. Two different photonenergies provide the possibility of studying the core–shellstructure of the constituent atoms at different probing depths.In addition, the magnetic properties of the same films weremeasured for the purpose of finding the correlation betweenelectronic arrangement and magnetic properties in these twocrystalline orientations of LSMO. Analysis of the results givesdeeper insight into the cause of the improved ferromagneticproperties of the LSMO (110) oriented films, and suggests thatstrain engineering between the LSMO and the substrate latticecan be very useful for tuning the electronic and magneticproperties in these half-metallic oxides.
2. Experimental details
The films were prepared by pulsed laser deposition (PLD)from a stoichiometric LSMO (x = 0.3) target which wassintered from micronsized powder prepared by the solid-statemethod, as described earlier [16, 17]. The LSMO filmswere grown on epitaxially polished (001) and (110) orientedSrTiO3 (STO) single crystal substrates without a buffer layerat a deposition temperature of 700 ◦C using an excimer XeCl308 nm laser with laser fluence of 2 J cm−2 and repetitionrate of 5 Hz. The oxygen pressure in the chamber was0.2 Torr and after deposition the atmospheric pressure of O2
was introduced, and the films were annealed for 10 min,followed by a cooling process down to room temperature ata rate of 25 ◦C min−1. To check the thickness of the films,we have performed x-ray reflection measurements and foundthat the same number of laser pulses produces different filmthicknesses for films grown in different orientations. Thecalculated thicknesses are 168 nm and 147 nm for (001) and(110) oriented films, respectively, indicating slightly slowergrowth of LSMO on the (110) oriented STO substrate.
The detailed structural characterization was made byx-ray diffraction (XRD) measurements using a PhilipsX’Pert Pro diffractometer with a Schulz texture goniometerusing Cu Kα radiation, an incident Ni-filter, 0.04 radSoller slit and a 0.18◦ thin film collimator. The x-rayphotoelectron spectroscopy (XPS) spectra were obtainedby synchrotron radiation excited high kinetic energyphotoelectron spectroscopy (HIKE) [18] carried out atbeamline KMC-1, BESSY II, Helmholtz-Zentrum Berlin. TheKMC-1 beamline is equipped with a Scienta R4000 highenergy electron analyser and double crystal monochromatorcovering photon energies from 2 to 10 keV [19]. The HIKEspectra were measured at grazing incidence geometry withnormal emission and the photon energies used were 2500and 7000 eV. All the experiments were performed at roomtemperature and no cleaving, sputtering or heating was carriedout prior to the measurements. The binding energy (BE) scalewas calibrated using the Fermi level with Au 4f (calibrationsample) or C 1s core levels.
Magnetic measurements were performed with a QuantumDesign magnetic property measurement system (MPMS) bymeasuring the temperature dependences of the zero-field-cooled (ZFC) and field-cooled (FC) magnetizations between5 and 350 K with different external magnetic fields of 20,50, 100, 150 and 250 mT. Virgin magnetization at 5 K as afunction of B and magnetic hysteresis curves were recordedin a field of ±500 mT at different temperatures of 5, 100, 200and 300 K.
3. Results and discussion
3.1. Structural analysis
To investigate the purity and growth orientations of theprepared thin films, XRD measurements were performedat room temperature. The crystal structure of our LSMOtarget is a rhombohedral perovskite with space group R3Cand hexagonal lattice parameters ah = 0.5473 nm and ch =
1.336 nm (pseudocubic ac = 0.3866 nm) [20, 17]. The latticeparameter of the cubic perovskite STO substrate is a =0.3905 nm, which is very close to the pseudocubic ac of theLSMO target, hence the LSMO film grows epitaxially on topof the STO substrate as presented in figure 1.
The θ–2θ diffraction scans made on the LSMO films withdifferent growth orientations show only LSMO pseudocubicstructure peaks (00l) for the (001) sample and (hh0) forthe (110) sample, matching with corresponding substratepeaks and indicating complete epitaxial growth withoutgrowth in any other directions. Moreover, all the reflection
2
J. Phys.: Condens. Matter 25 (2013) 376003 S Majumdar et al
Figure 1. (a) X-ray pole figures of the {110}c reflection in the (001) oriented sample at 2θ = 32.60◦, and (b) {200}c and {024}h reflectionsin the (110) oriented sample at 2θ = 47.17◦. The faintly visible extra peaks on the sample (110) at ϕ = 0◦, 180◦ and ψ = ±45.2◦ arise fromthe substrate due to continuous radiation.
peaks in the (001) oriented sample are high intensitypeaks from the substrate with only poorly visible shouldersfrom LSMO, indicating almost complete overlap of theLSMO and substrate peaks. This means that the substrateinduced strain forces the pseudocubic LSMO lattice to matchwith the substrate lattice, producing a very small latticemismatch in the (001) oriented sample. For the (110) orientedsample, the LSMO peaks are slightly more visible under thesubstrate peaks, indicating more deviation in the pseudocubicLSMO lattice parameter compared to that of the substrate.The calculated pseudocubic lattice parameters ac for thedifferently oriented samples (001) and (110) are 0.3901 nmand 0.3863 nm, respectively. The (110) oriented sample latticeparameter is very close to the bulk value, which indicates amore relaxed lattice compared to the (001) oriented sample.
The x-ray pole figures (ϕ − ψ scans) were measured toobtain more information about the crystal orientations withinthe samples. The (001) oriented sample gives only a cubiclattice {110}c reflection pattern at 2θ = 32.60◦, whereas the(110) oriented sample shows a hexagonal {024}h reflectionpattern including a cubic {200}c reflection pattern at 2θ =47.17◦. The pole figures are shown in figure 1. The absenceof the hexagonal reflections in the (001) oriented samplegives even more indication of matching lattice parametersbetween film and substrate. In the (110) oriented sample,the appearance of the hexagonal {024}h reflections indicatesmore distortion in the LSMO pseudocubic lattice and higherlattice mismatch compared to the (001) oriented sample. Thelack of any extra peaks in the pole figures indicates thatsamples comprise only one crystalline orientation and arefully textured in the same direction as the substrate. Thestructural differences between the differently oriented samplesprobably arise from the substrate induced tensile strain, wherethe substrate strains the film’s lattice parameters to matchwith the substrate. For the (001) oriented sample the effect ofsubstrate induced tensile strain is stronger, because it affectsin the in-plane directions [100] and [010] and hence impactsdirectly on lattice parameters a and b as well as on the angle γbetween them. As for the (110) oriented sample, the in-planestrain affects the [001] and [110] directions, having less directeffect on the LSMO lattice parameters and angles. The less
effective substrate induced tensile strain in the (110) orientedsample allows the lattice to grow more relaxed compared tothe (001) oriented sample.
3.2. Electronic structure
In order to study the electronic structure of LSMO samplesin the two crystalline orientations, we have performed XPSmeasurements on them at room temperature. Here the resultsof the Hard x-ray photoelectron spectroscopy (HAXPES)measurements for two LSMO samples grown on (001) and(110) oriented STO single crystal substrates are presented.The photoelectron spectra of Mn 3p, Mn 3s, Mn 2p, O 1s,La 3d core levels and valence band, together with the surveyspectra, were recorded. Two different photon energies, 2500and 7000 eV, have been used to collect the photoelectronspectra. For these photon energies, the inelastic mean freepaths (IMFPs) λ for all measured photolines calculated withthe semiempirical model developed by Tanuma et al [21] arepresented in table 1.
First, we focus on the photoemission spectra of the Mn 3sstructure. The XPS spectra indicate core electron bindingenergy splitting for the Mn 3s levels [22]. The spectralsplitting of the 3s core level in transition metals and theircompounds is caused by the exchange coupling between the3s hole and the 3d electrons. According to the results of theMn 3s spectra of manganites La1−xSrxMnO3, the magnitudeof the Mn 3s splitting increases with decreasing Sr dopinglevel [23]. This implies that the value of the Mn 3s exchangesplitting may be used to determine the Mn valence in a morequantitative way. Similarly, in the case of the manganeseoxides (MnO, Mn2O3 and MnO2), the value of the exchangesplitting decreases from 6 to 4.7 eV as the Mn valenceincreases [24]. In figure 2, the normalized XPS spectra ofthe Mn 3s exchange splitting measured at photon energiesof 2500 and 7000 eV for LSMO (110) and (001) are shown.The maxima of exchange splitting for two different incidentenergies and different crystalline directions of LSMO givesimilar binding energies of 83.1 and 88.5 eV. Curve fittingof these spectra gives values of the Mn 3s exchange splittingequal to 5.3–5.4 eV. Also, the area ratios of the splitting
3
J. Phys.: Condens. Matter 25 (2013) 376003 S Majumdar et al
Table 1. Inelastic mean free paths (IMFPs) for measured photolines of LSMO samples with different binding energies (BEs).
Photoline BE (eV) IMFP 2500 eV (A) IMFP 7000 eV (A)
Figure 2. Normalized PES spectra of the Mn 3s exchange splittingmeasured at photon energies of 2500 and 7000 eV, and for twoLSMO films on STO (110) and STO (001) substrates.
doublets have approximately equal value from 2.15 to 2.25.The changes of the degree of Mn–O covalency as well asthe formal valence of Mn ions in mixed valence manganites(La1−xSrxMnO3) should lead to different Mn 3s splittingdepending on the doping concentration. However, it has beenshown that Mn 3s splitting stays constant (around 5.3 eV)for the formal valence manganese ions in the range from+3.0 to +3.3 [25]. Hence, it may be concluded that themajority of Mn ions have the valence state+3 and the averageconcentration of Mn4+ ions in the different probing depthsof studied samples has a minor role. The presence of Mn2+
may be excluded due to the value of the exchange splitting,which is even lower than the corresponding value, 5.5 eV, forMn2O3, and because of the Sr2+ dopant, that should lead tothe increase of the Mn4+ content.
The XPS spectra of La 3d peaks show not only thespin–orbit splitting (≈17 eV), but both of the two main peaksare accompanied by high energy satellites with comparableintensity and a separation of 4.1–4.3 eV from the mainpeaks (shown in the structure of La 3d5/2 in figure 3(a)).A similar structure of La 3d photolines has been reportedalso in the case of LSMO single crystal measurements [26].The studies on lanthanum compounds have shown that thesatellites of the La 3d3/2,5/2 peaks arise from the electrontransfer process [27–29]. The transfer is induced by thesudden creation of an electron–hole pair in the inner core
level of La 3d, which tends to attract an outer electron froma ligand (in this case oxygen) to the empty La 4f energylevel. The final state of the main peak corresponds to theconfiguration 4f0
[La4+(3d95p6)∗] and the satellite peak tothe final state 4f1L1−
[La3+(3d95p64f1)∗+O1−(2p5)∗], whereL1− denotes a hole in the oxygen ligand, and the asteriskmarks the excited states. Comparison of the La 3d5/2 doubletstructure of the (110) and (001) oriented LSMO films showsclear differences. In the case of the (001) oriented LSMO,the satellite peak is remarkably enhanced. The estimation ofthe La 3d5/2,3/2 doublet peak areas for both samples showsa 4.5% higher peak area in the case of (001) oriented LSMOcompared to (110) oriented LSMO. In other words, the changeof the LSMO crystalline orientation from (110) to (001)causes the deformation of the electronic density between Laand surrounding O atoms and produces an enhanced electrontransfer from the O 2p orbital to the empty La 4f subshell toscreen the La 3d hole.
The photoelectron spectra of Mn 2p3/2 and Mn 3p areshown in figures 3(b) and (c), respectively. Manganese formsvery complex species with oxygen, in addition to simpleoxides of Mn2+, Mn3+, and Mn4+ [30]. Also the complexityof Mn 2p peaks is known to arise due to the very complicatedmultiplet splitting structure [31, 32]. The curve fitting of themeasured LSMO spectra based on the Mn2+, Mn3+ and Mn4+
peak parameters given in [32] showed good results composedof Mn3+ and Mn4+ multiplet peaks. The proportion of Mn2+
multiplet peaks was clearly vanishing. The curve fitting resultsare not presented here. In the case of (110) and (001) orientedLSMO films the Mn 2p and Mn 3p peaks have clearlydifferent widths. Both spectra of (001) oriented LSMO showthe broadening towards to the lower binding energy almostby 0.2 eV. This indicates a distinguishable re-distribution ofthe formal valency of Mn atoms. In accordance with theearlier studies [30, 32], the detected broadening towards tothe lower binding energy may be interpreted as an increaseof the Mn3+ concentration compared to the concentration ofthe Mn4+ ions. Earlier it was shown that a weak ligand field inmanganese compounds can give rise to a high spin state of Mn[33]. This means that due to the modified crystalline structure,there is an out-of-plane or in-plane change in the Mn–O bondlength of the (001) and (110) oriented LSMO films. IncreasedMn–O bond length in the LSMO (001) sample leads to aweaker ligand field and hence Mn3+ concentration in thissample is higher.
The measured photoelectron spectra of O 1s (figure 4)show that the broadening of the O 1s photolines at hν =7000 eV is caused by lower spectral resolution comparedto the measurements at 2500 eV. The spectra measured at
4
J. Phys.: Condens. Matter 25 (2013) 376003 S Majumdar et al
Figure 3. Photoelectron spectra of (a) La 3d5/2, (b) Mn 2p3/2 and(c) Mn 3p core levels measured with synchrotron radiation of2500 eV photon energy, and of two LSMO samples deposited on the(110) oriented and (001) oriented SrTiO3 substrates.
2500 eV have a clear high energy shoulder, that is missingin the spectra photoionized with energy of 7000 eV. Thishigh energy shoulder is most probably caused by the signalof the surface contaminant. The surface contamination maybe the result of physisorbed, chemisorbed, and structuralH2O together with hydroxide (OH−) and CO2 contributions.However, one can believe that the main profile of O 1s peaksdoes not contain a signal from the surface contamination.The change of the ionization energy from 2500 to 7000 eVincreases the bulk probing depth, but the main characteristic
Figure 4. Normalized photoelectron spectra of O 1s measured atphoton energies of 2500 and 7000 eV, and for two LSMO films onSTO (110) and STO (001) substrates. Also the subtraction result oftwo spectra at a photon energy of 2500 eV is given (dotted curve).
shape differences of O 1s spectra for LSMO samples on (110)and (001) oriented STO substrates remain the same. The maindifference for these two samples is the line width broadeningby 0.15 eV and enhanced asymmetry in the case of the LSMOfilm on (001) oriented STO substrate. The outcome of thesubtraction of the corresponding photolines shows a peak, themaximum of which is shifted by 0.6 eV from the centre of themain peak at a binding energy of 529.3 eV. It has been shownfor doped perovskite manganites [34–36], that the oxygen1s spectrum may contain three components arising from thethree different oxide layers in the crystal lattice, and in thecase of LSMO the binding energies of the three componentsfrom the higher binding energy side to the lower one wereassigned to Mn–O, La–O, and Sr–O, respectively. This meansthat due to the rearranged unit cell distortion of the LSMOfilm on (001) STO substrate, the Mn–O component of the O1s photoline is slightly shifted towards lower binding energyvalues compared to the (110) oriented LSMO.
In conclusion, two different crystalline orientations ofLSMO thin films cause different unit cell distortions of LSMOand modifications of the MnO6 octahedra, which affect thehopping of eg electrons along Mn3+–O2−–Mn4+ chains.Due to the out-of-plane or in-plane changes in the Mn–Obond length of the (001) and (110) oriented LSMO, thereexists an induced higher Mn3+ concentration in the (001)sample. The strain engineering to control the magnetic andmagnetotransport properties of LSMO film depending on itscrystalline properties provides an additional tuning parameterfor the functional characteristics.
3.3. Magnetic properties
To find the effect of a modified electronic structure onthe magnetic exchange interaction, we have measured themagnetic properties of these films under different magneticfields and a large temperature range. The temperaturedependences of magnetization (M(T)) curves measured indifferent external magnetic fields for the (001) and (110)oriented LSMO films are shown in figure 5. Because (110)surfaces show rectangular symmetry, the atomic sequences
5
J. Phys.: Condens. Matter 25 (2013) 376003 S Majumdar et al
Figure 5. Temperature dependence of the MZFC and MFC magnetizations measured in different external magnetic fields of 20 mT (a),50 mT (b), 100 mT (c), and 250 mT (d), for LSMO films in orientations (001) and (110). The insets show the first derivatives of the MFCcurves. The (110) oriented films are measured in two different in-plane directions [110] and [001].
Figure 6. Virgin curves measured up to 500 mT at 5 K (a), and the magnetic hysteresis loops (b)–(d) between ±200 mT fields attemperatures 5, 100, 200 and 300 K for (001) and (110) oriented LSMO films. The (110) oriented films are measured in two differentin-plane directions [110] and [001].
6
J. Phys.: Condens. Matter 25 (2013) 376003 S Majumdar et al
along two in-plane orthogonal directions [001] and [110]differ substantially, and therefore for LSMO (110) films boththese directions were measured. In LSMO (001) both in-planedirections, [010] and [100], display square symmetry andtherefore their magnetic properties are identical. Hence, it canbe concluded that in this sample anisotropy could only arisedue to magnetocrystalline or shape anisotropy.
As can be seen from the graphs, the onset Curietemperature, TC,onset, is always above room temperature andcareful examination indicates that onset TC,onset is slightlyhigher (≈2–3 K) in (110) oriented than in (001) orientedLSMO film. The FC curve of both LSMO (110) directionsindicates also a slightly higher magnetic moment of ≈3 ×103 A m−1 at 300 K when compared to the LSMO (001).However, the TC determined from the minimum of the firstderivative, dM/dT , of the FC curve (insets of figure 5), is onaverage 12–15 K higher in (110) oriented film than that inthe (001) oriented film, depending on the magnetic field. Thisis in good agreement with the results reported earlier [14].Also the saturation magnetization in both in-plane directionsof (110) films is higher than in the (001) oriented LSMO filmwhen measured with small magnetic fields (approximately ofthe order of coercive field). However, with increasing appliedmagnetic field value the saturation moment value of LSMO(001) slightly surpasses the moment of the (110) oriented film.The higher Curie temperature and the enhanced ferromagneticproperties in LSMO (110) films can be explained by thepresence of a higher concentration of Mn4+ in them, asobserved from the XPS study. Higher Mn4+ concentrationfavours enhanced DE interaction between the Mn3+ andMn4+ ions leading to improved ferromagnetism. Also dueto less crystal deformation and more relaxed unit cells,deformation in MnO6 octahedra is lower in LSMO (110)films, which favours DE interaction. The higher magneticmoment at higher magnetic field in the LSMO (001) filmsmight be due to the alignment of the spins of the higher spinspecies Mn3+.
As can be seen from figure 5, the shape of the M(T)curve of LSMO (110) is much more gradual in the wholetemperature range 50–350 K without the clear saturationseen in LSMO (001) film at a low temperature range for allmagnetic field values. The magnetic transition in LSMO (001)film is clearly sharper and the saturation is more pronouncedin the low magnetic field range, however, the length of theM plateau shortens with increasing B, which at the sametime broadens the width of the magnetic transition also inLSMO (001) film. Therefore the shapes of the M(T) curves inLSMO (110) and LSMO (001) are almost similar in the B =250 mT field, but the magnetization is smaller below 280 Kin (110) oriented LSMO film. Also the ZFC and FC branchesof magnetization separate from each other at low magneticfields (20 mT) below 200 K for both LSMO (001) and (110)samples. This thermo-magnetic irreversibility is a signature offrustrated magnetic phase at low temperature. With increasingmagnetic field strength, the separation between the twobranches starts to reduce and the irreversibility temperaturedecreases. For the LSMO (001), the irreversibility vanisheseven faster than in the LSMO (110). The explanation of thisobservation will be discussed later.
To have a clear understanding of the difference inmagnetic behaviour in (110) oriented LSMO films in twoin-plane orthogonal directions [001] and [110], we havemeasured the M(H) virgin curve at 5 K and the M(H)hysteresis loops at 5, 100, 200 and 300 K (figure 6). Alsothe same measurements were performed for the (001) orientedLSMO films for comparison. LSMO (110) with a magneticfield applied in the [110] direction showed the sharpestincrease in M(H) curve and complete saturation of magneticmoment, around 200 mT field. Also the highest coercivityBc at 5 K (49.2 mT) and remanence (67% of Ms) wasobserved in this film. LSMO (110) with a magnetic fieldapplied in the [001] direction showed a gradual increase inmagnetization and a slightly suppressed saturation magneticmoment compared to the [110] direction. Also, the coercivity(38.2 mT at 5 K) and remanence (50% of Ms) were reducedin this case. For measurements up to higher magnetic fields,complete closure of the hysteresis loops can be obtained andthe coercive field values can be slightly different, however,the main observation still remains the same. The difference inthe magnetic properties between the two in-plane directionsclearly indicates that there is an in-plane magnetic anisotropyin the (110) oriented film. This points towards the fact thatthe (001) plane of the LSMO lattice is not perfectly alignedwith the (001) plane of the STO substrate and there is atilt of the LSMO (001) plane which results in a long bodydiagonal [111], forcing the easy axis out-of-plane. EarlierBoschker et al [37] experimentally verified this out-of-planeeasy axis formation through angle dependent magnetizationmeasurements. This out-of-plane easy axis results in magneticdomain formation. These magnetic domains, with eitheropposite in-plane magnetization direction or by a rotation ofthe magnetization out of the film plane, can cause reductionin the saturation and remanent magnetization as well as thecoercivity. Also these magnetic domains, aligned randomlyat high temperature and aligning themselves at differentdepinning temperatures, could result in more gradual M(T)curves for the LSMO (110) sample.
Compared to the two in-plane directions of (110) orientedLSMO, (001) oriented LSMO shows a much more gradualM(H) curve and visibly smaller remanence (27% of Ms)and coercivity (18.1 mT at 5 K), although the magneticmoment value is higher in this sample. This clearly suggeststhat the ferromagnetic interaction in LSMO (001) is rathersoft in nature, and the higher magnetic moment in thissample is the result of spin alignment of higher magneticmoment spin species Mn3+ ions rather than improved DEinteraction, which is also supported from our XPS data.Now, the magnetization in a (001) sample increases as afunction of magnetic field due to both domain wall motionand domain rotation. The higher magnetic moment in LSMO(110) at small applied magnetic fields suggests that domainwall motion in these films is fairly easy. However, the highercoercivity and remanence in the (110) sample then suggeststhat domain rotation is restricted compared to the LSMO(001) films due to some kind of domain pinning mechanism.Domain wall motion is fairly easy in this sample and thereforethe substantially higher pinning energy barrier in LSMO (110)
7
J. Phys.: Condens. Matter 25 (2013) 376003 S Majumdar et al
indicates random orientations of the magnetic domains withalternate out-of-plane and in-plane components. This can leadto nucleation of magnetic domains resulting in slower changein magnetization with magnetic field reversal and a hardferromagnetic property. Previously, Uspenskaya et al [38]have shown that magnetization reversal occurs via both thenucleation and motion of domain walls and inhomogeneousrotation of magnetization from multiple nucleation centres.Hence our result suggests that in LSMO (110) films,both paramagnetic–ferromagnetic phase transformation andmagnetization reversal is more gradual, due to its domainnucleation, and a substantially higher depinning field isrequired for rotating the domains. This depinning energyis even higher when a magnetic field is applied along the[110] axis, as the remanence and coercivity are substantiallyhigher in this case. Previously it was shown that LSMOhas an easy axis aligned with the maximum tensile straindirection [39]. However, in the present study as the tilt ofthe (110) vector causes a partially out-of-plane easy axis,we can postulate that the projection of the easy axis inthe in-plane direction maximizes the magnetization for amagnetic field applied along the [110] direction. Also, in arecent article [40], the authors suggested that the rotationof oxygen octahedra in different lattice directions can leadto magnetocrystalline anisotropy behaviour of LSMO thinfilms. In the present study, from the XPS results, we haveshown that due to modified Mn–O distance, MnO6 octahedraare distorted differently in the two crystal orientations ofLSMO which can lead to magnetocrystalline anisotropy.Hence the observed thermo-magnetic irreversibility, higherremanence and coercive field in LSMO (110) are directconsequences of substrate induced strain anisotropy, while inLSMO (001), magnetocrystalline anisotropy induced by themodified MnO6 octahedra plays a major role in decreasingremanent magnetization and coercivity. Therefore, we canconclude that the increased ferromagnetic property in LSMO(110) is an added advantage for spintronic application, whilemagnetic domain formation and the high depinning field couldpose challenges in practical devices.
4. Conclusions
In conclusion, the structural, electronic and magneticproperties of LSMO films in two different crystallineorientations, (001) and (110), are studied. Our results showthat the (001) oriented LSMO shows stronger substrateinduced tensile strain, due to effects in both in-planedirections, [100] and [010], which affect directly the latticeparameters a and b as well as the angle γ betweenthem. However, for the LSMO (110), the in-plane strainaffects the [001] and [110] directions, affecting the LSMOlattice parameters and angles less directly. The less effectivesubstrate induced tensile strain in the (110) oriented sampleallows the lattice to grow more relaxed compared to the(001) oriented sample. This results in a higher populationof higher spin species Mn3+ in the (001) oriented LSMO,and changing the electronic density around the La ion andincreasing the Mn–O bond length, modifying the MnO6
octahedra in them. Comparatively, LSMO (110) shows lessdistorted MnO6 octahedra and a higher amount of Mn4+ ions,leading to improved ferromagnetism. However, strain inducedanisotropy causes magnetic domain formation in these filmswith an out-of-plane component which results in a decreasedsaturation moment and gradual magnetization curves, whichshould be carefully taken into account for practical spintroniccomponents.
Acknowledgments
The Wihuri Foundation, the Turku Collegium for Scienceand Medicine (TCSM) and the Academy of Finland areacknowledged for financial support. We also acknowledge theHelmholtz-Zentrum Berlin—Electron storage ring BESSY IIfor provision of synchrotron radiation at beamline KMC-1.
References
[1] Dagotto A, Hotta T and Moreo A 2001 Phys. Rep. 344 1[2] Bibes M and Barthelemy A 2007 IEEE Trans. Electron
Devices 54 1003[3] Majumdar S and Majumdar H S 2012 Org. Electron. 13 2653[4] Park J H, Vescovo E, Kim H J, Kwon C, Ramesh R and
Venkatesan T 1998 Nature 392 794[5] Garcia V, Bibes M, Barthelemy A, Bowen M, Jacquet E,
Contour J P and Fert A 2004 Phys. Rev. B 69 052403[6] Julliere M 1975 Phys. Lett. A 54 225[7] Kavich J J, Warusawithana M P, Freeland J W, Ryan P,
Zhai X, Kodama R H and Eckstein J N 2007 Phys. Rev. B76 014410
[8] Freeland J W et al 2007 J. Phys.: Condens. Matter 19 315210[9] Harrison W A, Kraut E A, Waldrop J R and Grant R W 1978
Phys. Rev. B 18 4402[10] Baraff G A, Appelbaum J A and Hamann D R 1977 Phys. Rev.
Lett. 38 237[11] Yamada H, Ogawa Y, Ishii Y, Sato H, Kawasaki M, Akoh H
and Tokura Y 2004 Science 305 646[12] Ishii Y, Yamada H, Sato H, Akoh H, Ogawa Y, Kawasaki M
and Tokura Y 2006 Appl. Phys. Lett. 89 042509[13] Infante I C, Sanchez F, Fontcuberta J, Wojcik M, Jedryka E,
Estrade S, Peiro F, Arbiol J, Laukhin V and Espinos J P2007 Phys. Rev. B 76 224415
[14] Minohara M, Furukawa Y, Yasuhara R, Kumigashira H andOshima M 2009 Appl. Phys. Lett. 94 242106
[15] Boschker H, Kautz J, Houwman E P, Siemons W,Blank D H A, Huijben M, Koster G, Vailionis A andRijnders G 2012 Phys. Rev. Lett. 109 157207
[16] Huhtinen H, Raittila J, Paturi P, Salminen J andZakhvalinskii V S 2002 J. Phys.: Condens. Matter 14 7165
[17] Majumdar S, Huhtinen H, Majumdar H S, Laiho R andOsterbacka R 2008 J. Appl. Phys. 104 033910
[18] Kover L 2010 J. Electron Spectrosc. Relat. Phenom. 178 241[19] Schaefers F, Mertin M and Gorgoi M 2007 Rev. Sci. Instrum.
78 123102[20] Majumdar S, Huhtinen H, Majumdar H and Paturi P 2012
J. Alloys Compounds 512 332[21] Tanuma S, Powell C J and Penn D R 1997 Surf. Interface
Anal. 48 25[22] Fadley C S, Shirley D A, Freeman A J, Bagus P S and
Mallow J V 1969 Phys. Rev. Lett. 23 1397[23] Saitoh T, Bocquet A E, Mizokawa T, Namatame H,
Fujimori A, Abbate M, Takeda Y and Takano M 1995 Phys.Rev. B 51 13942
[24] Kowalik M, Zalecki R and Kolodziejczyk A 2010 Acta Phys.Pol. A 117 277
J. Phys.: Condens. Matter 25 (2013) 376003 S Majumdar et al
[25] Galakhov V R, Demeter M, Bartkowski S, Neumann M,Ovechkina N A, Kurmaev E Z, Lobachevskaya N I,Mukovskii Y M, Mitchell J and Ederer D L 2002 Phys. Rev.B 65 113102
[26] Mannella N et al 2006 J. Electron Spectrosc. Relat. Phenom.153 37
[27] Jørgensen C K and Berthou H 1972 Chem. Phys. Lett. 13 186[28] Howng W Y and Thorn R J 1978 Chem. Phys. Lett. 56 463[29] Park K H and Oh S J 1993 Phys. Rev. B 48 14833[30] Oku M and Hirokawa K 1975 J. Electron Spectrosc. Relat.
Phenom. 7 465[31] Gupta R P and Sen S K 1974 Phys. Rev. B 10 71[32] Nesbitt H and Banerjee D 1998 Am. Mineral. 83 305[33] Blomberg M R A and Siegbahn P E M 1997 Theor. Chem.
Acc. 97 72
[34] Liang J J and Weng H S 1993 Indust. Eng. Chem. Res. 32 263[35] Choi J, Zhang J, Liou S H, Dowben P A and Plummer E W
1999 Phys. Rev. B 59 13453[36] Ibrahim K, Liu F, Qian H, Guo L, Xian D, Xiong G and Wu S
2000 Mater. Sci. Eng. B 76 26[37] Boschker H, Kautz J, Houwman E P, Koster G, Blank D H A
and Rijnders G 2010 J. Appl. Phys. 108 103906[38] Uspenskaya L, Nurgaliev T and Miteva S 2010 Acta Phys. Pol.
A 117 207[39] Boschker H, Mathews M, Houwman E P, Nishikawa H,
Vailionis A, Koster G, Rijnders G and Blank D H A 2009Phys. Rev. B 79 214425
[40] Boschker H, Mathews M, Brinks P, Houwman E, Vailionis A,Koster G, Blank D H A and Rijnders G 2011 J. Magn.Magn. Mater. 323 2632
