Room temperature ferromagnetism in epitaxial Cr₂O₃ thin films grown on r-sapphire

The sesqui oxide Cr₂O₃ is an antiferromagnetic (AFM) insulator below the Néel temperature 307 K. It crystallizes in the corundum structure with R$3¯$c space group and belongs to $3¯′m′$ magnetic point group with the easy axis of magnetization along the [0001] direction. Cr₂O₃ has been used in a broad range of applications, such as protective coatings on stainless steels,¹ as a catalyst² and as a solar thermal energy collector.³ Since Cr₂O₃ is a very insulating material, it can also be used as a tunnel barrier in magnetic tunnel junctions.^4,5 It was the first compound in which the linear magnetoelectric (ME) effect was theoretically predicted⁶ and experimentally observed.⁷ The ME effect refers to the appearance of electric polarization by the application of magnetic field and vice versa.⁸ Although ME effect is a well-studied phenomenon, recently there is a rejuvenation of interest in this area with the aim of use in technological applications like spintronics.^9,10 The ME effect of Cr₂O₃ is exploited to alter the magnetization of a ferromagnetic (FM) layer that is exchange biased with the adjacent AFM.¹¹ Recently, electrically switchable exchange bias has been realized at room temperature in Cr₂O₃ single crystal when in contact with a ferromagnetic film, utilizing an unconventional surface magnetism on the Cr₂O₃(0001) surface.^12,13

So far, the magnetoelectric property of Cr₂O₃ has been demonstrated only in single crystals.^12,14 Room temperature ferromagnetic (electric) hysteresis loops have been reported under external electric (magnetic) fields in Cr₂O₃ single crystals. Since the net magnetic moment induced by applied electric field in ME Cr₂O₃ is small,¹⁴ it is interesting and useful to see whether magnetic moment can be induced by other means, for example, by defects in the present case. Though many reports are available on the growth of Cr₂O₃ thin films, using different techniques,^15–17 in-depth study of their magnetic properties is lacking. For device applications, a good understanding of the structure property correlation in thin films is required. With this aim, we prepared epitaxial thin films of Cr₂O₃ on r-sapphire substrates and studied their magnetic properties using a Super-conducting Quantum Interference Device (SQUID) magnetometer. Here, we report on the epitaxial growth and magnetic properties of Cr₂O₃ thin films grown on single crystal r-sapphire (⁠$011¯2$⁠) substrate by pulsed laser deposition (PLD). The as grown films are well characterized using high-resolution (HR) X-ray diffraction (XRD) (2θ, Φ), TEM, and Raman spectroscopy. The films are found to be epitaxial with good crystallinity, as evidenced from the XRD and TEM data. The in-plane SQUID measurements performed (5–400 K) on the pristine samples revealed ferromagnetism with Curie temperature above 400 K. After oxygen annealing (OA), the FM is almost completely suppressed. The FM in the as deposited films is attributed to the strain present in the films as inferred from HR XRD and Raman measurements. The decrease in strain with oxygen annealing indicates that strain is due to point defects, such as oxygen vacancies. This study to understanding the origin of FM in Cr₂O₃ thin films is useful to better design the spintronic devices.

The Cr₂O₃ thin films were grown on r-plane sapphire (Al₂O₃) substrates by ablating high purity Cr target in oxygen ambience using PLD. The substrates were vapor cleaned in trichloro-ethylene at 250 °C and then ultrasonically cleaned by immersing in acetone and methanol each for 5 min. The PLD chamber utilized a KrF laser of 248 nm wavelength and 25 ns pulse duration. The laser energy density was kept at 3.2 J/cm² with a repetition rate of 5 Hz and the target to substrate distance was maintained at 4 cm during the deposition. The base pressure of the chamber before deposition was less than 1 × 10⁻⁶Torr. The thin film deposition was performed at a substrate temperature of 650 °C and 5 × 10⁻²Torr oxygen partial pressure. The cooling was carried under the same oxygen partial pressure at 10 °C/min to room temperature. The phase structure of the films was determined by X-ray diffraction θ-2θ scans using Rigaku X-ray diffractometer with Cu K_α (0.154 nm) radiation. The in-plane orientation of the films was determined by Φ scans using Panalytical X'Pert PRO MRD HR X-Ray Diffraction System. The JEOL 2010F TEM was used for imaging and determination of thicknesses of thin film layers. Selected area electron diffraction (SAED) was taken to further confirm the epitaxial nature and establish the epitaxial relationship between the film and the substrate. The in-plane magnetic properties were measured using a SQUID. Raman spectra were taken using a Horiba Labram HR800 system using a 532 nm wavelength laser.

Fig. 1(a) shows the indexed θ-2θ X-ray diffraction pattern of a 230 nm thick Cr₂O₃ film grown on Al₂O₃(⁠$011¯2$⁠) substrate. The presence of only reflections corresponding to the (⁠$011¯2$⁠) plane of Cr₂O₃ and their higher orders indicate that the film was grown highly textured or epitaxial. It also eliminates the possibility of formation of other phases of chromium oxide within the detection limit of XRD measurements. The (⁠$011¯2$⁠) peak position of Cr₂O₃ film at 2θ = 24.352 (bulk value 24.494) indicates that the out-of-plane lattice constant is elongated compared to the bulk value with a tensile strain of 0.57% along that axis. The rocking curve (ω-scan) taken on the (⁠$011¯2$⁠) plane shown in Fig. 1(b) measures a full width at half maximum (FWHM) of 0.8°, indicating good crystallinity of the film. The XRD φ-scan measurements were performed to identify the epitaxial nature of the film and establish the epitaxial relationship between the film and substrate. Fig. 2 shows the Φ-scan excited on the (006) plane of the film and substrate, inclined at ψ = 57.61° from the (⁠$011¯2$⁠) plane. It should be noted that both Cr₂O₃ and Al₂O₃ crystallize in the corundum structure. The presence of a single peak at the same Φ-position as the substrate peak indicates that the r-plane of the film has grown epitaxially without any rotation in the basal plane. The epitaxial relations are determined as [$011¯2$] Cr₂O₃ ‖ [$011¯2$] Al₂O₃ (out-of-plane) and $[1¯1¯20]$ Cr₂O₃ ‖ $[1¯1¯20]$ Al₂O₃ (in-plane).

Fig. 3(a) shows the indexed SAED pattern acquired along the [$22¯01$] zone axis of the Cr₂O₃/Al₂O₃ heterostructure. The spot pattern confirms that the film was grown epitaxially. The SAED pattern also excludes the formation of any other oxide phases of chromium. The epitaxial relations obtained from the data are in agreement with those obtained from the XRD Φ-scan data. The inset of Fig. 3(a) shows the low magnification cross sectional TEM image of the heterostructure. It can be seen from the HRTEM image in the Fig. 3(b) that the interface between the film and the substrate is atomically clean, reaction free and so indicating that there is no secondary phase formation at the interface.

The magnetic field dependent magnetization (M vs H) loops measured in the −10 kOe to +10 kOe range on a 230 nm thick film are plotted in Fig. 4(a). To make sure that the magnetism is originating from the film instead of the substrate, we have performed measurements on the substrate and subtracted the diamagnetic contribution from the substrate. The M vs H loops and magnetization versus temperature (M vs T) data collected from the substrate are shown in Fig. S1.¹⁸ The film displayed typical ferromagnetic behavior in the measurement temperatures ranging from 5 K to 400 K, as seen in Fig. 4(a). Fig. 4(b) shows the enlarged view of the same in the −300 Oe to +300 Oe range. The hysteresis loops show finite coercivity that has decreased from 173 Oe at 5 K to 17 Oe at 400 K. The room temperature saturation magnetization (M_S) and coercivity are 0.12 emu/g and 25 Oe, respectively. The magnetic moment observed here is very small compared to that reported in Cu doped ZnO (Ref. 19) and Co doped CeO₂. (Ref. 20). The variation of M_S with the temperature is plotted in Fig. 4(c). The M_S is not a strong function of temperature in the measurement range of 5 K–400 K, which implies that the FM in this case could be induced by defects. In Cr₂O₃, the spin axis is along the c-axis with the Cr⁺³ spins aligned parallel and antiparallel to the c-axis. It is shown using first-principles calculations and magnetometry that the surface of c-plane of Cr₂O₃ has electrically switchable magnetization with orientation perpendicular to the plane. Based on the spin configuration, the r-plane should not exhibit any FM behavior. However, FM in otherwise AFM or non-magnetic materials has been well reported in the literature with reasons, including strain and defects.^21–23 

To investigate the origin of FM in our films, we have performed post OA on the sample at 650 °C for 1 h at 5 Torr partial pressure of O₂. Fig. 4(d) shows the effect of OA on the magnetic hysteresis loop measured at 300 K. It can be seen from the graph that the magnetic moment (0.01 emu/g) almost vanishes with OA. The M vs T data collected from the sample before and after OA can be seen in Fig. S2.¹⁸ These data suggest that oxygen vacancies are the reason for the observed FM in our pristine films and, filling the vacancies reduces the FM order. In order to determine whether the crystal structure has changed after the OA, we performed XRD on the sample and found that it remained the same. The plot in Fig. 5(a) shows the high-resolution 2θ-scan measured around the (012) peak of the pristine and OA sample. It can be clearly seen that the peak position shifted towards higher 2θ values (towards the bulk value of 24.494°) after OA. The pristine r-Cr₂O₃ film is found to have +0.57% strain in the out-of-plane direction, which is decreased to +0.31% after OA. The residual in-plane strain ε_xx (or) ε_yy calculated using the relation ε_zz = −2υ*ε_xx/(1 − υ) is about −0.85% for pristine and −0.46% for oxygen annealed sample, using the poison's ratio υ = 0.25.

To further confirm that OA is decreasing the amount of strain present in the films, we have measured Raman spectra at room temperature on the pristine and OA samples. The Raman vibrations are very sensitive to the amount of stress present in the films thus giving the Raman shift, which is a direct evidence for the presence of strain in the films. For Cr₂O₃, the vibrations with symmetry A1_g and E_g are Raman active. Out of the seven active modes (2Al_g+5E_g), one A1_g and four E_g are usually observed. The most intense peak corresponds to Al_g mode (out-of-plane vibration) and can be used for strain estimation.^24,25 It is evident from the graphs in Fig. 5(b) that with the OA, the Al_g mode for the pristine sample at 561.6 cm⁻¹ is shifted to 558.6 cm⁻¹ approaching bulk value of 552 cm⁻¹. The change in the peak position of Al_g can be written as Δω = ω_stressed − ω_bulk. The strain is tensile when Δω is negative and compressive when Δω is positive. We found a Δω of −9.6 cm⁻¹ for the pristine and −6.6 cm⁻¹ for the OA sample, indicating that the strain is tensile and also decreased with the OA. Thus the results from Raman measurements are in good qualitative agreement with those obtained from XRD. From these results we can conclude that FM is induced by the strain due to defects, such as oxygen vacancies.

Defects play a critical role in controlling the physical properties of the material. For instance, defect induced FM in otherwise nonmagnetic and un-doped oxides like TiO₂, ZnO, and SnO₂, is widely reported in the literature often attributed to point defects, such as oxygen vacancies, cation vacancies, and cation interstitials.^26–29 In ZnO, it is shown that room temperature ferromagnetism (RTFM) can be created by a controlled introduction of Zn vacancies, where magnetic coupling is provided by carriers from oxygen vacancy and Zn interstitial clusters.³⁰ In PLD grown HfO₂ films, highly anisotropic FM was induced by defects.²³ A review of FM in various oxide systems is made recently.³¹ Since defect generation is controlled by strain, strain engineering can be used to tune magnetic properties in thin films. Strain induced FM and Néel temperature change in AFMs have been demonstrated theoretically and experimentally in previous studies.^32–34 

Strain induced FM in many systems in the form of thin films and nanoparticles is well reported in the literature. A tensile uniaxial strain of the order of 2%, in AFM thin films of LaMnO₃, is found to change the orbital state and magnetic ground state significantly.³² The epitaxial growth of Cr₂O₃ on sapphire occurs via lattice matching epitaxy with a misfit strain of −4%. In thin films, there are three different sources of strain: lattice, thermal mismatch, and defect induced. In our case, when the thickness of the film is about 230 nm, most of the lattice misfit strain relaxes through the generation of dislocations and the contribution of thermal mismatch is very small. The residual strain in our pristine sample, which is experimentally observed to be +0.57%, could be mostly due to defects, such as oxygen vacancies and related point defects. Under near equilibrium growth conditions, Cr₂O₃ has no specific type of point defects, whereas oxygen partial pressure determines the type of defects. When Cr₂O₃ is grown under a partial pressure of 5 × 10⁻²Torr O₂, we expect some oxygen deficiency in the films and thus defect induced strain. When the sample is annealed in oxygen ambient, most of the oxygen vacancies are diffused out and thus reducing the amount of strain in the film. It is well known that antiferromagnets are piezo magnetics and strain induces magnetic moment. Recently, it is found from theoretical studies that the two unpaired electrons due to oxygen vacancy are localized on the surrounding Cr atoms and hence enhancing the magnetic moment of the Cr atoms surrounding the oxygen vacancy.³⁵ We speculate that in addition to the increased magnetic moment of the Cr atoms surrounding the oxygen vacancy, the local compressive strain is inducing a canting of the magnetic moments away from the c-axis towards the r-plane and hence resulting in RTFM in the r-plane of Cr₂O₃. It should be noted that the angle between the r-axis and the c-axis in bulk Cr₂O₃ crystal is 57.61° and r-axis is perpendicular to r-plane. To get a clear understanding of the origin of RTFM in r-plane Cr₂O₃, more advanced experimental and theoretical studies are required.

We have investigated RTFM in epitaxially grown Cr₂O₃ thin films on r-sapphire by using PLD. The XRD (2θ and Φ) and TEM data confirm that the films are grown epitaxially with the underlying sapphire substrate. Further TEM study indicated that there is no secondary phase formation in the film and also the interface between the substrate and film is clean and sharp. The epitaxial relations between the substrate and film were established as [$011¯2$] Cr₂O₃ ‖ [$011¯2$] Al₂O₃ (out-of-plane) and $[1¯1¯20]$ Cr₂O₃ ‖ $[1¯1¯20]$ Al₂O₃ (in-plane). The as deposited films exhibited ferromagnetic nature with Curie temperature above 400 K. The room temperature coercivity and saturation magnetization are 25 Oe and 0.12 emu/g, respectively. The ferromagnetic nature almost vanished after oxygen annealing. The high resolution XRD and Raman measurements indicate that the pristine films are strained, and the strain was reduced by oxygen annealing. The FM nature in the pristine films is attributed to strain induced defects, such as oxygen vacancies.

The authors acknowledge the use of the Analytical Instrumentation Facility (AIF) at North Carolina State University, which is supported by the State of North Carolina. The corresponding author would like to thank Dr. S.S. Rao and Dr. S. Nori for suggestions and critical manuscript reading and Yi-Fang Lee for help in Raman measurements. Part of this research is supported by the National Science Foundation (DMR-1304607) and Army Research Office.