Electronic Properties of Fully Strained La1-x Sr x MnO3 Thin Films Grown by Molecular Beam Epitaxy (0.15 ≤ x ≤ 0.45).

The structural, electronic, and magnetic properties of Sr-hole-doped epitaxial La1-x Sr x MnO3 (0.15 ≤ x ≤ 0.45) thin films deposited using the molecular beam epitaxy technique on 4° vicinal STO (001) substrates are probed by the combination of X-ray diffraction and various synchrotron-based spectroscopy techniques. The structural characterizations evidence a significant shift in the LSMO (002) peak to the higher diffraction angles owing to the increase in Sr doping concentrations in thin films. The nature of the LSMO Mn mixed-valence state was estimated from X-ray photoemission spectroscopy together with the relative changes in the Mn L2,3 edges observed in X-ray absorption spectroscopy (XAS), both strongly affected by doping. CTM4XAS simulations at the XAS Mn L2,3 edges reveal the combination of epitaxial strain, and different MnO6 crystal field splitting give rise to a peak at ∼641 eV. The observed changes in the occupancy of the eg and the t2g orbitals as well as their binding energy positions toward the Fermi level with hole doping are discussed. The room-temperature magnetic properties were probed at the end by circular dichroism.

Introduction

Strongly correlated functional oxide perovskites have been known for their myriad properties such as two-dimensional electron gas, superconductivity, colossal magnetoresistance, half-metallicity, ferroelectricity, catalysis,[1−6] and so forth. Among the various known perovskites, La1–SrMnO3 (LSMO) with optimal doping has attracted the research community owing to room-temperature ferromagnetic metal (FM) and large spin-polarization.[5,7,8] This compound exhibits a rich electronic and magnetic phase diagram by tuning the hole-dopant Sr “x”.[9,10] For instance, the parent compounds LaMnO3 (x = 0) (Mn electronic configuration is 3d4: t2g3eg1) and SrMnO3 (x = 1) (3d3: t2g3eg0) show antiferromagnetic (AFM) (A-type and G-type, respectively) and insulating (I) properties.[11,12] By considering the charge neutrality, substituting La by divalent Sr in parent LaMnO3, the compound exhibits Mn mixed-valence states (Mn4+ and Mn3+: 3d3 and 3d4). The Mn 3d electronic band-filling percentage in the t2g (xy, yz, zx) and eg(x2 – y2, 3z2 – r2) orbitals differs with hole doping, influencing the macroscopic electronic and magnetic properties of the system. Furthermore, the structural distortion due to the Jahn–Teller (JT) effect adds an additional degree of freedom for electrons in selecting the preferential orbital occupancy.

For low doping (x < 0.17), an A-type AFM-I is observed.[12] In the doping regime 0.17 < x < 0.6, an FM behavior is favored, with the highest observable Curie temperature Tc ≈ 350 K for the optimal doping x ≈ 0.33.[13] Due to large Hund’s coupling and to the increase in the Mn–O–Mn bond angle, the probability of hopping electrons between Mn ions via O 2p orbitals enhances the double-exchange (DE) mechanism, giving rise to FM and metallic nature.[14−16] For doping levels x > 0.6, a charge-ordered AFM-I state is favored due to the suppression of DE. Distinctly from bulk and in addition to JT distortions, epitaxial strain imposed by the substrate in thin films adds additional constraints in selecting the orbital filling in the eg states, that is, tensile (compressive) strain that favors the in-plane (out-of-plane) x2 – y2 (3z2 – r2) orbitals.[17,18]

Although several studies have been performed on manganites, for instance, Hishida et al.[19] investigated the relationship between the valence-band X-ray photoelectron spectroscopy (XPS) spectral weights to the electrical conductivity of the entire Sr doping series of the polycrystalline La1–SrMnO3 samples, whereas Horiba et al.[20] studied the loss of spectral weight at the Fermi level and subsequent increase in the shoulder of the O 1s X-ray absorption spectroscopy (XAS) spectra and suggested pseudo-gap formation with an increase in Sr doping in PLD-grown La1–SrMnO3 films. However, discrepancies between the expected ferromagnetic and the actual paramagnetic insulating behaviors of LSMO[21] as a function of doping (e.g., in Horiba et al.,[20] only Sr ≈ 0.3 and 0.4 doping levels do show an FM behavior at room temperature) often question possible oxygen vacancies[22] as well as the La/Sr/Mn stoichiometry.[21,23−25] Hence, there is a need to investigate in-detail a systematic study of Sr-doped La1–SrMnO3 films by careful tuning of the La/Sr cation ratio using the powerful molecular beam epitaxy (MBE) technique and by fixing the thickness realizing how the physical properties in thin films are affected with respect to the epitaxial strain imposed by the underlying substrate.[26]

Here, various Sr-doped (0.15 ≤ x ≤ 0.45) epitaxial LSMO thin films of thickness 120 unit cells (uc’s) are deposited on 4° vicinal STO (001) substrates using the shutter MBE technique. The choice of vicinal substrates is due to induced uniaxial magnetic anisotropy with the easy axis parallel to the step edge direction being necessary for the design and development of anisotropic magnetoresistance sensors.[27,28] The structural, chemical, electronic, and magnetic properties are probed by means of X-ray diffraction (XRD), XPS, XAS, and X-ray magnetic circular dichroism (XMCD) spectroscopy techniques. XRD reveals the increase in tensile strain with an increase in doping “x”, whereas XPS and XAS confirm the mixed-valence state of Mn ions. Valence band (Vb) spectra show substantial changes in the eg, t2g orbital occupancy, and changes in the ferromagnetic order are observed by XMCD. Finally, the common attribution of the intense peak in XAS spectra at ∼641 eV is found to be due to the combination of epitaxial strain and different MnO6 crystal field splitting by CTM4XAS simulations instead of the formation of Mn2+ ions.

Experimental Details

Sr dopant concentration-dependent epitaxial LSMO (001) thin films of thickness 120 uc’s were coherently grown on to (001)-oriented 4°tw vicinal SrTiO3 (STO) substrates using the shutter layer-by-layer MBE technique.[29] The as-received substrates were sonicated in acetone and ethanol solutions and subsequently dried under a N2 atmosphere prior to be inserted into the MBE deposition system. The deposition of the films was made by maintaining the substrate temperature of about 720 °C and the ozone background pressure was kept at 2 × 10–6 mbar. After deposition, films were cooled down to room temperature under the same deposition pressure. By taking advantage of the in situ reflection high energy electron diffraction (RHEED) technique, thickness and Sr doping concentrations in the films were controlled. The growth rate of the films was about 1 uc per minute.

The structural characterization of all films was performed by monitoring both in situ RHEED oscillations and ex situ PANalytical X’Pert Pro XRD using monochromatic Cu Kα1 radiation (λ = 1.54056 Å) in the Bragg–Brentano configuration. Rocking curves of the films were measured using high-resolution XRD with incident optics composed of double Ge(440) and diffracted optics composed of a 3-axis monochromator. The chemical composition and electronic properties of the films were characterized using linearly polarized soft X-ray synchrotron radiation techniques such as XPS and XAS spectroscopies at the APE-HE beamline at Elettra synchrotron.[30] The XAS measurements at the Mn L2,3 edges were performed in the total electron yield (TEY) mode with a grazing angle of 45° with respect to the beam. The drain current of a reference mesh was measured simultaneously and independently from the sample for the alignment and normalization of the spectra. XMCD measurements were taken in circular polarization, by measuring the TEY signal at remanence after the application of an in-plane magnetic pulse of ±30 mT at each energy step. The magnitude of the applied magnetic field (±30 mT) is sufficient to saturate the sample, as it was observed from MOKE measurements (not presented here). The dichroic signal is calculated by considering the 75% degree of polarization and the 45° of angle incidence. Core-level and Vb XPS spectra were acquired using an Omicron E125 hemispherical analyzer with an incident photon excitation energy of 1000 and 470 eV, respectively.

Results and Discussion

Structural Characterization

θ–2θ out-of-plane XRD was performed on the films with different Sr doping contents by aligning the substrate vicinal miscut angle with respect to the incident X-ray beam. The films have grown epitaxially on the STO substrate, showing the (00l) orientation (Figure a), and all the films show detailed finite-size structural effects, indicating a very high crystalline quality of the films. In addition, we have observed that the LSMO (002) film peak position (“*” in Figure a) shifts toward higher diffraction angles with an increase in Sr doping. The out-of-plane lattice parameter “c” calculated from the (002) LSMO peak is reduced from 0.39 to 0.382 nm with an increase in Sr doping from 0.15 to 0.45. Figure c shows the evolution of the c-axis lattice parameter of LSMO films with various Sr doping. By substituting Sr by La in the parent compound LaMnO3, the pseudocubic lattice parameter decreases from 0.3944 to 0.3805 nm for LaMnO3 to SrMnO3, respectively. As the films are grown on STO with the cubic lattice constant a = 0.3905 nm, the film experiences higher tensile strain and consequent reduction of the out-of-plane lattice constant “c” with an increase in Sr doping.[13,31] The estimated Poisson ratio for the films is ∼0.36 ± 0.02, which is a typical value for oxides and manganites.[32,33] The RHEED pattern shown in the inset of Figure a with well-defined sharp diffraction streaks corresponds to a long-range crystallinity order of the film.

Figure 1

(a) θ–2θ out-of-plane symmetrical XRD and the inset shows the RHEED pattern of the film, (b) ω-scan rocking curves measured along vicinal miscut (010) directions, (c) evolution of out-of-plane lattice parameter “c”, and (d) asymmetrical RSM of LSMO thin films for various Sr doping. The star in panel (c) indicates the bulk lattice parameter of LSMO with Sr doping concentration x = 0.3.

In addition, the quality of the films and the misalignment of the LSMO cell with the substrate were verified through rocking curve analysis. The omega scan rocking curves were obtained at the (002) peaks of LSMO by aligning the X-ray beam along the vicinal directions. Figure b depicts the rocking curves of all the various Sr-doped LSMO films measured along the miscut direction, showing a single sharp peak with full-width half maximum values in the range of 0.068 ± 0.02°. Such a value is similar to that of our STO substrates, indicating very high film crystallinity. Asymmetrical reciprocal space maps (RSMs) were obtained around the (013) Bragg reflections of both the LSMO film and STO substrate, as shown in Figure d. The scattering vectors Q of both film and substrate are vertically aligned for different Sr doping concentrations, confirming that the LSMO films are pseudomorphically grown on the STO substrate. Hence, the films are fully tensile-strained and the in-plane lattice constants “a” of the film match with those of the substrate.

X-ray Photoemission and Absorption Spectroscopy

To investigate the elemental composition and Mn oxidation states in LSMO films, XPS measurements are performed at the APE-HE beamline. Figure a shows the XPS survey scans of different Sr-doped LSMO thin films measured with linearly polarized photon excitation energy hν = 1000 eV. The XPS survey scan reveals all the expected elements with features of La 3d, Mn 2p, O 1s, Sr 3d, and La 4d edges. As the films are measured ex situ, except the C 1s peak, no other traces of impurities are observed on the film surface.

Figure 2

(a) XPS survey scan, (b) Mn 2p of La1–SrMnO3 thin films with different Sr doping concentrations x, (c) Mn 2p for 0.15 and 0.45 Sr with Mn3+ and Mn4+ peak-fitting components, and (d) Mn4+ content as a function of Sr doping. The measurements are performed with photon excitation energy hν = 1000 eV.

The Mn 2p core-level spectra of various Sr-doped LSMO films in the binding energy range of 670–635 eV are shown in Figure b. All the Mn 2p spectra show similar features with spin–orbit doublets Mn 2p3/2 and Mn 2p1/2 located around ∼641 and ∼653 eV, respectively, and the spin–orbit splitting of ∼12 eV. A small bump located at ∼662 eV is a charge transfer satellite peak of Mn 2p1/2.[34] Since the films were measured ex situ, the well-screened metallic peak at the lower binding energies is not visible, which otherwise is observed either on unexposed samples[35] or by hard X-ray photoemission.[19,36] In order to estimate the Mn valence state in the films, Mn 2p core-level spectra were fitted using mixed Gaussian-Lorentzian line shapes after subtracting the background intensity due to the presence of secondary electrons using the Shirley background. The comparison between the 0.15 and 0.45 Sr-doped LSMO films along with Mn3+ (orange shaded region) and Mn4+ (violet shaded region) fitting components is depicted in Figure c. By estimating the spectral weights of the Mn3+ and Mn4+ peaks, a systematic linear trend is found, that is, an increase in the percentage of Mn4+ content in correspondence to an increase in Sr-hole doping in the LSMO films, as shown in Figure d. Therefore, the MBE technique is advantageous in the fine tuning of doping concentrations, which is essential for the development of artificial multilayers and heterostructures.

Doping-induced subtle electronic changes in LSMO films are directly reflected in the XPS Vb spectra analysis, as presented in Figure . Figure a shows the Vb spectra in the binding energy range of 0–10 eV, taken at a photon energy of 470 eV for various Sr-doped LSMO films. The spectral assignments from the Fermi level EF are designated to Mn 3d eg (0.5–1 eV), Mn 3d t2g (2–2.5 eV), and Mn 3d–O 2p hybridization (∼3.5 and ∼7.5 eV) and with an intense O 2p peak (∼5.6 eV), respectively.[20,37−40] Hong et al.[41] demonstrated that in perovskite systems (LaMnO3, LaCoO3, etc.), the O 2p nonbonding states can straddle between two transition metal and oxygen (TM-O) hybridization states and are associated with the antibonding and bonding states. Therefore, the highly intense peak at the BE of ∼5.6 eV is assigned to O 2p nonbonding states, whereas the strongly hybridized Mn 3d–O 2p peaks that are present at ∼3.5 eV belong to the antibonding state and denoted as Mn–O* and the weak shoulder at the BE ≈ 7.5 eV belongs to the bonding states of Mn–O.

Figure 3

(a) XPS valence-band spectra (10–0 eV binding energy range) taken at the photon energy hν of 470 eV, (b) binding energy peak positions of eg, t2g, and Mn–O* hybridization peaks, and (c) eg and t2g spectral weights for various Sr doping in films, respectively.

While O 2p nonbonding states and Mn–O* hybridization peaks are present almost at the same binding energies of ∼5.6 and ∼3.5 eV for all the Sr doping levels, interesting changes have been observed in the eg and t2g states. While the binding energy position of the eg state for 0.15 Sr doping is located at ∼1.1 eV, upon increasing the Sr doping concentration in the films, the eg state tends to move closer to the EF level, reaching ∼0.5 eV for 0.45 Sr. Similar to the eg state, the t2g state-binding energy position also moves toward EF, going from ∼2.45 to ∼1.9 eV for 0.15 to 0.45 Sr doping, as shown in Figure b. Subsequently, the spectral weights calculated under the eg (t2g) state reduces (enhances) with the increase in hole doping (Figure c), indicating the decrease (increase) in eg band filling.[42] This results in an increase in the electron hopping parameter, thus enhancing the conductivity and favoring the ferromagnetic order with hole doping, as it will be discussed in the XMCD section.

The changes in the electronic structure induced by different Sr doping contents in LSMO films were then probed by XAS at Mn L2,3 edges, as shown in Figure a. A reference signal acquired simultaneously with the spectra allowed us to correctly align the spectra in energy. As described in the introduction, the electronic configuration of the Mn 3d in a purely electron-doped (LMO) and hole-doped (SMO) system is t2g3eg1 (Mn3+) and t2g3eg0 (Mn4+), respectively, while the intermediate doping regime presents mixed Mn3+ and Mn4+ valence states. Therefore, the Mn XAS features will be sensitive to any changes in both doping concentrations. We therefore observed that by increasing the Sr doping level from 0.15 to 0.45, a shift of the main Mn L3 edge of 0.48 eV takes place toward the higher photon energy.

Figure 4

(a) Mn L2,3 edge XAS spectra of 120 uc LSMO films with various Sr doping concentrations and (b) XAS Mn L2,3 edges simulated for different valence states. Dashed vertical gray lines indicate the correspondence between experimental and simulated peaks for the different doping levels.

To attribute the valence contributions of each spectrum, simulations were performed using CTM4XAS software, as seen in Figure b.[43] First, we have simulated XAS spectra for different Mn valence states (Mn2+, Mn3+, and Mn4+). The ligand field multiplet theory, implemented in the CTM4XAS program, was adopted to simulate the theoretical Mn L2,3 edge spectra. The theoretical parameters are reported in Table S1. The hybridization of Mn with oxygen ligands was considered in the calculations by reducing the Slater integrals from their atomic values. The Mn3+ symmetry is considered to be D4h (due to a Jahn–Teller distortion) and the Mn2+ and Mn4+ to be Oh.[44] The charge transfer parameters for Mn2+, Mn3+, and Mn4+ were selected from previous work on MnO, LaMnO3, and MnO2, respectively.[45−49] The crystal field parameters Ds and Dt (tetragonal parameters) for the Mn3+, generated from the transition Oh to D4h distortion,[50,51] were selected with the opposite sign that indicates an axial compression (compression of the octahedral/tensile strain) in LSMO sample growth on STO substrates. The relative concentration of Mn3+ and Mn4+ was chosen from the previous Mn quantification performed on the Mn 2p XPS fit. From 0.25 to 0.35 Sr concentration, we observed the rise of a sharp intense peak at 641 eV (A in Figure a). We proposed two hypotheses about this observation: (i) Mn2+ formation and (ii) an increase in the JT distortion in Mn3+. The former is supported by the intense peak of the theoretical Mn2+ that is at around 641 eV, and moreover, there is a wide bibliography on the Mn2+ formation in an air-exposed LSMO system,[52] while the latter is supported by the previous XPS measurement in which the Mn2+ presence was not observed, in addition to the epitaxial strain in the film (Figure ). For this reason, we proposed another Mn3+*-simulated spectrum where the 10 Dq was increased to 2 eV: in this way, the final energy of d2 orbitals increased and it describes a stronger compression of MnO6 octahedra (Figure b). It is worth to note that the Mn3+* has an intense peak at 641 eV as the Mn2+. Hence, our simulations suggest that the epitaxial strain in the film leads to different crystal field splitting and would therefore attribute the peak present at ∼641 eV to an Mn3+ contribution instead of an Mn2+ one.

Finally, the energy splitting (Δ eV) between the L2 and L3 maximum peak intensity edges decreased from 10.8 eV for 0.15 Sr to 10.6 eV for 0.45 Sr doping. Such a similar energy difference trend was also observed previously in divalent Ca-doped manganite perovskite Pr1–CaMnO3 thin films and attributed the difference to the increase in the valence state of Mn.[53,54]

XMCD

In order to have a deeper understanding of the role of the Sr dopant on the changes in magnetic properties, XMCD measurements at the Mn L2,3-edge were performed at room temperature, that is, 300 K. Figure shows the XMCD spectra of the sample series with Sr doping from 0.15 to 0.45. The inset of Figure shows the maximum XMCD % that systematically increases with the increase in Sr content in films from 0.20 and reaches the maximum value of about 12–13% for 0.30 and then decreases. By considering the LSMO phase diagram,[55] the optimal doping regime (Sr ≈ 0.3) exhibits the highest ferromagnetic ordering, which is in good agreement with the XMCD results obtained here. The XMCD signal for the film grown with 0.15 and 0.45 Sr doping concentrations are almost flat, indicating a negligible FM order possibly due to the arising of the AFM one.

Figure 5

XMCD spectra at Mn L2,3 edges for the various doping levels of La1–SrMnO3 thin films measured at room temperature (300 K). The inset shows the XMCD % at the L3 edge as a function of Sr doping concentration.

Conclusions

We have successfully deposited epitaxial LSMO thin films by fine-tuning of Sr doping (0.15 ≤ x ≤ 0.45) in films using the MBE technique and extensively studied their structural, chemical, electronic, and magnetic properties using XRD and different synchrotron-based spectroscopy techniques. For all Sr concentrations, films were fully strained with the underlying STO substrate, thus showing a linear decrease in the out-of-plane lattice parameter “c” with an increase in Sr-hole doping content. Valence-band XPS shows that the spectral weight area of eg (t2g) decreases (increases) and the peak positions also move closer to EF as Sr-hole doping increases in films. The electronic states measured by XAS at Mn L2,3 edges also show significant shift toward higher photon energy with the increase in Sr doping levels. The differences in energy splitting between Mn L2,3 edges, that is, Δ eV from 10.8 to 10.6 eV correspond to the changes in the Mn valence state with Sr-hole doping. The additional shoulder peak in XAS present at ∼641 eV for the intermediate doping levels (0.25–0.35) was attributed to the epitaxial strain and crystal field splitting of the MnO6 octahedral network. This systematic study will serve as a template for the studies of various hole-doped manganites/perovskite systems.