Introduction

Tailoring multifunctional physical-chemical properties of oxide thin films represents a growing and challenging area of interest in materials science1. Search for multiferroic materials with co-existing room temperature ferromagnetism and ferroelectricity is another area of intense research activity today for developing devices that can be controlled by both magnetic and electric potentials2. Ability to stabilize oxides into a glassy state gives rise to many technologically relevant properties thus enabling a wide range of applications in electronics, ranging from gate dielectrics in CMOS devices to flash memories1,2,3,4,5. The glassy state differs from the crystalline ground state of a material in the way the atoms are arranged. Since many physical properties of solids, in particular magnetism, are closely related to the local atomic structure, it is of fundamental interest to unveil the underlying effects of amorphization on the magnetic structure of a material.

Magnetic oxides provide a rich variety of fascinating phenomena and among them the perovskites have recently attracted special attention due to the possible coupling between their electrical and magnetic properties leading to novel multifunctional materials6,7. Normally, in amorphous magnetic oxides, random distribution of magnetic ions affects the antiferromagnetic (AFM) superexchange interactions, which usually dominate in these oxides and causes frustration in the arrangement of the spins8,9,10,11. This induces a spin glass transition and a drastic suppression of the magnetic ordering temperature12,13. BiFeO3 is an example of such situation. In the crystalline state, BiFeO3 is antiferromagnetic, with a Neel temperature of 643 K14, whereas in the amorphous state, the material exhibits a transition from a paramagnetic to a spin glass state at 20 K13.

Here, we report a drastic deviation from the above pattern. We find robust soft ferromagnetism above room temperature in amorphized YCrO3-films obtained by Pulsed Laser deposition, while in contrast crystalline YCrO3 is known to be an antiferromagnetic insulator with Neel temperature, TN ~ 140 K. We have combined advanced spectroscopic techniques with first-principles calculations to reveal the underlying physics of this phenomenon. Such unusual transformation to ferromagnetism in amorphous oxides has been reported before15 in amorphous EuTiO3 with a TC = 5.5 K while its crystalline phase was antiferromagnetic with TN = 5.3 K.

The ability to tailor ferromagnetism above room temperature in oxide materials, induced by the disorder in the atomic lattice is a novel effect that deserves special attention and which might lead to many multifunctional technological applications.

Results

We have characterized the powder and the thin film by using X-ray diffraction (XRD), transmission electron microscopy (TEM) and a combined focused ion-beam (FIB)/scanning electron microscope (SEM). Figure 1(a) shows a typical film thickness determination from the focused ion-beam cross-section analysis, where the thickness of the film was found to be approximately 435 nm. In its powder and pellet forms, YCrO3 is found to be crystalline (Figure 1b) with orthorhombic symmetry belonging to space group Pnma in close agreement with studies by Ramesha et al.16. Unlike the powder, the pulsed laser deposited (PLD) thin film does not show any XRD intensity peaks (see Figure 1b) and is consistent with an amorphous phase. The TEM analyses and electron diffraction pattern obtained (inset in Figure 1a) also confirms the amorphous structure for the thin films. It should also be pointed out that we have tried to crystallize the films to see if their properties are reversible by growing thicker films, or annealing the sample at high temperatures, without any success. The thicker films remained amorphous and on annealing at higher temperatures the film started deteriorating in structure and also in its magnetic properties probably because of possible effects of phase separations, which is an undesired outcome. Furthermore, we could not detect any signature of impurities in the samples within the detection limits from Energy Dispersive Spectroscopy (EDS), indicating that the observed ferromagnetism is intrinsic.

Figure 1
figure 1

Structural analysis.

(a) Using Focused Ion-Beam assisted Scanning Electron Microscope; thickness of film was determined to be around 435.6 nm; (inset) TEM diffraction pattern of thin film shows a halo indicating complete amorphization. (b) XRD of YCrO3 powder and pellet (used as target) shows orthorhombic crystal structure while the experimental radial distribution function for the thin film indicates only an amorphous phase on the silicon substrate.

Full size image

Figure 2(a) shows the field dependence, at 300 K, of the magnetization for a crystalline sample. As can be seen, the linear field dependence of the magnetization indicates a simple paramagnetic behavior at room temperature. Figure 2(b) displays the field sensitive temperature dependence of the magnetization measured at fields as low as 50 and 100 Gauss in both the zero field cooled (ZFC) and field cooled (FC) states. The sample undergoes a magnetic transition below about 140 K. On cooling down from room temperature in an external field of 2.5 kOe through the transition temperature down to 50 K, a minor hysteretic loop was found which does not close (as shown in Figure 2(c)). Furthermore, we find that the loop exhibits a high coercive field value that is dependent on the strength of the external field at which the sample was cooled, rate of cooling and the lowest temperature reached. This phenomenon indicates that the crystal particles have strong anisotropy fields so that the maximum applied field was not enough to close the loop17 resulting in the observed minor loop characteristics. In fact, at low temperatures the crystalline powder exhibits features, which may be related to the magnetic canting in the antiferromagnetic state.

Figure 2
figure 2

Magnetization measurements.

(a) Paramagnetic behavior of YCrO3 powder at room temperature, (b) Temperature dependence of the magnetization for YCrO3 powder showing Zero Field Cooled (ZFC) and Field Cooled (FC) curves measured at low fields and the magnetic transition at 140°K, (c) the open minor hysteretic loop for the powder at 50 K upon field cooling from 300 K in an external field of 2.5kOe and (d) the hysteretic loop for the amorphous thin film (435.6 nm) at 300 K showing robust ferromagnetism with magnetic coercivity, HC = ~29 Oe (inset).

Full size image

For the thin film, amorphous phase, the scenario is different. The sample shows robust soft ferromagnetic behavior persisting to above 300 K with a finite low coercive field of about 29 Oe, as shown in Figure 2(d). In the crystalline form the super-exchange interaction between Cr atoms bridged by oxygen (Cr3+-O2−-Cr3+) explains the antiferromagnetic ordering of Cr-spin-moments. However, a microscopic picture of the intrinsic structure of the amorphous phase remains to be understood to reveal the underlying physics of the room temperature ferromagnetic ordering.

It is important to note that although we have presented the properties of a 435 nm film as a representative amorphized film, we see similar results for films in the thickness range up to around 500 nm. Thicker films in the micron regime develop a mixed composite structure with the onset of partial crystallinity in them and the magnetic behavior is no longer that of a soft ferromagnet.

We investigated the electronic structure of the amorphous YCrO3 by means of X-ray absorption spectroscopy (XAS) and X-ray emission spectroscopy (XES) on the beam line 8.0.1 at the Advanced Light Source at Berkeley Lab. The experimental setup is described in Ref. 18. Figure 3(a) shows the XAS spectrum of YCrO3 at the oxygen K-edge. It was found that the oxidation state of Cr atoms in the bulk (100–200 nm in depth) obtained from the fluorescence yield (TFY) detection is Cr3+ with a small contribution of Cr4+ as indicated by the small peak at 529.2 eV. The oxidation state of Cr on the surface at about 5–10 nm in depth that was resolved from total electron yield (TEY) detection is mainly Cr4+.

Figure 3
figure 3

Spectroscopy measurements.

(a) O K-edge XAS; (b) O K-edge XES and XAS; the black lines are intended for identify the band issues; and (c) Cr L-edge XAS.

Full size image

In Figure 3 (b), we plot the oxygen bulk XES measurement together with XAS results from the surface as well as bulk in a standard absorption-emission spectrum19,20. The data show the metallic character for the amorphous phase. However, we have indicated in the figure that the band-gap may not be closed in the surface region. A similar absorption-emission spectrum (not shown here) clearly indicates a band-gap in our crystalline sample. Thus, the amorphization is accompanied by an insulator-metal transition. However, we are not able to explain the metallicity simply by the stabilization of mixed Cr3+ and Cr4+ valence states because we observed a similar mixed valence also in the insulating crystalline phase. Furthermore, such conducting behavior could be rigorously proved by measuring the transport properties. However, the spectroscopy measurements that we present in this work is considered to be sufficient to unveil the electronic structure information needed to interpret the observed magnetic phenomenon.

Figure 3(c) shows the XAS spectra of YCrO3 at chromium L-edge. It further confirms that the oxidation state of Cr atoms in the bulk (100–200 nm in depth), also obtained from fluorescence yield (TFY) detection is Cr3+, while the oxidation state at surface (5–10 nm in depth) detected from total electron yield (TEY) detection is Cr4+.

Discussion

The results presented in Figures 2 and 3 indicate that the structural and electronic phase transitions of YCrO3 are accompanied by a magnetic phase transition. In the crystalline structure, we have a super-exchange interaction mediated by bridging oxygen atoms favoring anti-ferromagnetic ordering, which is consistent with its insulating character. In the amorphous phase, local spin moments on Cr atoms may interact via double exchange, which is the dominant exchange mechanism in a mixed valence state (between surface Cr4+ and bulk Cr3+) of metal ions. However, as we shall demonstrate later, this is only one of the contributions that may stabilize the ferromagnetic state near the surface region. In order to study further the nature of such magnetic and electronic transitions, we have also employed ab initio theory and the results will be presented in the following.

The calculations were carried out within the framework of density functional theory (DFT), as implemented in the VASP code (for more details see the methods section). We first investigated the electronic and magnetic properties of the ground-state crystalline structure of YCrO3, which has orthorhombic symmetry belonging to space group Pnma. The calculated lattice parameters (a = 5.6 Å, b = 7.49 Å and c = 5.2 Å) compare well with our experimental findings (a = 5.52 Å, b = 7.53 Å and c = 5.24 Å) and also other results in the literature21. The system with AFM ordering is found to have a finite band-gap and is more stable than the system with FM ordering by 136 meV/unit cell. The local moment on Cr is found to be 3μB, which is consistent with the high-spin solution from Hund's rule for a d3 configuration. In this system, the super-exchange interaction stabilizes the AFM state. This interaction is strongly dependent on the Cr-O-Cr bond angle, which is 146° for this orthorhombic cell. The calculated density of states for AFM ordering is shown in Figure 4(a).

Figure 4
figure 4

Theoretical results.

(a) Total DOS of crystalline phase, (b) Total DOS amorphous; (c) Cr-O-Cr bond angle distribution; and (d) total energy as a function of the number low spin states (local moment close to 1μB) on Cr atoms in the amorphous cell.

Full size image

To investigate the amorphous structure of YCrO3, we have employed ab inito molecular dynamics (MD) simulations on a 160-atoms supercell, which was constructed from the meta-stable cubic perovskite structure14. The system was heated at 6000 K until it melted and lost memory of the initial structure. This was achieved after 10 ps of a MD run consisting of 10000 1 fs ionic steps. Careful analysis of the evolution of the radial distribution function and mean square displacement, not shown here for brevity, was performed at this point to ensure a liquid structure. Thereafter, we cooled down the system from 6000 to 300 K in 3 ps by means of a 3000 more MD time steps. The last snapshot of this simulation was then structurally optimized at 0 K to fully quench into the amorphous state.

To rule out the possibility that the choice of temperature parameters in the MD melt-quench procedure could influence the amorphous structure, we also performed some quenches directly from starting configurations with randomly placed atoms in the supercell by means of the stochastic quench (SQ) technique22,23. The SQ and MD generated structures have equivalent radial and angle distribution functions as well as partial coordination numbers, which indicate consistency in our amorphous structures24.

The definition of the AFM state is ambiguous in the amorphous phase. The structural disorder will ensure an inescapable magnetic frustration that can result in a large set of almost degenerate magnetic configurations. Therefore a large number of different spin configurations with mixed parallel and anti-parallel alignments were calculated in the following protocol. First the system is fully optimized in the ferromagnetic state, with all spins polarized in the same direction. Then, we carried out a number of calculations with different initial magnetic arrangements, where no geometry optimization was performed. The local spins were allowed to relax but the total moment of the cell was kept fixed. In this way we could investigate the total energy change by varying only one degree of freedom and identify any tendency to prefer vanishing net moment in the cell. After that we successively allowed relaxation of the total spin and then the crystal geometry.

The outcome of this study was that the states with non-zero total magnetic moment in the supercell, that is, a ferromagnetic state, were systematically found to be the most energetically stable, indicating that the material should display partial FM ordering where some of the short-range interactions favor local AFM order. The average magnetic moment of Cr decreased to 2.5 μB in our most stable structures as compared to 3μB in the crystal. Bader charge analysis25,26,27 of the most stable structures showed that all Cr atoms are in the Cr3+ valence state, which is in good agreement with our XAS result for the bulk. The lowering of the average Cr moment was then found to be caused by a small set of magnetically frustrated Cr atoms in the low-spin state. The calculated FM order in absence of Cr4+ shows that the origin of the ferromagnetism in amorphous YCrO3 cannot only be the mixed valence states. In the following we discuss the electronic structure and the changes in the Cr-O-Cr bond angles to explore other possible mechanisms.

The density of states of our lowest-energy amorphous structure is shown in Figure 4 (b). As can be seen, the system becomes metallic as no finite band gap is observed. To further understand the favoring of the FM state in the amorphous structure, we plot the Cr-O-Cr bond angle distributions of the amorphous and crystalline structures in Figure 4(c). We can see that the distribution of angles is very broad in the amorphous structure and that there are two main peaks. One peak is close to the Cr-O-Cr angle of the crystalline structure at 146°, meaning that this group of Cr pairs couple through the ideal AFM 180° superexchange mechanism. The other peak that is centered at 75° indicates that the interactions are closer to the ideal 90° FM superexchange over a large subsetset of Cr pairs. The metastable switching from AFM to FM as a function of Cr-O-Cr angle has been discussed by Ray et al.21 connecting with phonon modes in the simple cubic perovskite structure.

The FM superexchange interaction occurs also in the half-metal CrO2, with a Curie temperature of about 392 K28, in which the double exchange is also claimed to be the driving force for the ferromagnetic ordering. In fact, the distinction between these two exchange mechanisms is not well defined. The double exchange is a combination of Coulomb and kinetic-exchange and therefore the FM superexchange can be considered as a double exchange mechanism. Besides that, as discussed above, our spectroscopy measurements indicate that mixed valence states occur near the surface region where the double exchange interaction is certainly contributing to FM ordering. Therefore, we believe that both superexchange and double exchange are responsible for the observed FM state.

To investigate the changes in the local moments in the amorphous state, we have counted the number of low spin moments of the Cr atoms in the supercell for the fixed-spin calculation and correlated this information with the total energy of the systems. As a result we have the graphic presented in Figure 4(d). As can be seen, the more low-spin moments, the higher is the total energy of the supercell, showing that the high-spin solution is energetically favored in the amorphous phase.

In summary, we report on the discovery of room temperature ferromagnetism in amorphous YCrO3 thin films. Combining results from XAS, XES and first-principles calculations, we demonstrate that the amorphization affects the magnetic coupling between local spin-moments on Cr atoms. Bond angle disorder allows short range FM coupling, which percolates in the sample. In fact, our results indicate that the ferromagnetic ordering is a consequence of two exchange mechanisms, namely double exchange and FM superexchange. Furthermore, the high-spin solution for the local moments is energetically favored.

It should be emphasized again that disorder induced ferromagnetic ordering at room temperature leading to an electronic phase transition, is a counter-intuitive effect that might open up many possibilities for novel technological applications. Although we have used YCrO3 as a case study, we have similar results on amorphizing YbCrO3 and we believe our results are more general that would apply for a wide range of antiferromagnetic oxides that become metallic in the amorphous phase.

Methods

Computational details

Density functional theory (DFT) calculations were performed using the PAW (Projector Augmented Wave) method29 as implemented in the Vienna Ab initio Simulation Package (VASP)30. The PBE variant of the generalized gradient approximation (GGA) was used for the exchange-correlation functional31. The Kinetic energy cutoff of 600 eV and 5 × 5 × 5 k-points mesh generated by the Monkhorst-Pack method were found suitable to achieve the total energy convergence within the accuracy of 0.005 eV. In the PAW potentials, the electronic states 4s, 4p, 5s and 4d were treated as valence for Y, 3p, 3d and 4s states for Cr and 2s, 2p states for O. The convergence criterion for the electronic self-consistent cycle was fixed at 10−7 eV per cell and the forces on all ions were smaller than 10−5 eV/Å. Structural optimizations were performed by using a standard conjugate gradient method.

The synthesis methods

Since combustion reactions can yield large quantities of high quality powder, we used this method to produce enough powder for making a PLD target. Oxidants used in our work were Yttrium nitrate (Y(NO3)3·6H2O, 99.9%pure) and chromium nitrate (Cr(NO3)3·9H2O, 99.9% pure). Glycine (NH2CH2COOH) was used as fuel. These reactants were thoroughly mixed in the required molar ratios using a magnetic stirrer. Then the mixture was heated up till 100°C. On thermal dehydration a viscous liquid (gel) was formed. When this gel was further heated to ≈250°C, the viscous liquid swelled and auto-ignited, thus producing the powder which was calcinated at 600°C to obtain chemically pure and crystalline product. Due to a fuel-deficient ratio (1:0.5), the product was mainly YCrO4. To obtain YCrO3 the products were further heated at 800°C for 1 hour. The final powder was indexed on an.orthorhombic cell of YbCrO3 with lattice parameters a = 5.193 Å, B = 5.491 Å and c = 7.481 Å. Using spark plasma sintering technique this powder was used to cast a 1″ diameter pellet which was 99% dense. Using this pellet as the target, homogeneous dense PLD thin films were deposited at a constant energy laser pulse rate of 150 mJ/pulse.