betaFeedback

WorksPhysics and Astronomy

Interplay between breathing-mode distortions and magnetic order in rare-earth nickelates from abab initioinitio magnetic models

Abstract

We use density-functional theory calculations to explore the magnetic properties of perovskite rare-earth nickelates, R\mathcal{R}NiO3_3, by constructing microscopic magnetic models containing all relevant exchange interactions via Wannierization and Green's function techniques. These models elucidate the mechanism behind the formation of antiferromagnetic order with the experimentally observed propagation vector, and explain the reason previous DFT plus Hubbard UU calculations favored ferromagnetic order. We perform calculations of magnetic moments and exchange-coupling parameters for different amplitudes of the R1+R_1^+ breathing mode distortion, which results in expanded and compressed NiO6_6 octahedra. We find that the magnetic moment vanishes for the "short bond" nickels, i.e., the ones in the compressed octahedra. The inclusion of spin-orbit coupling demonstrates that the magnetic anisotropy is very small, while the magnetic moment of the short bond nickel atoms tend to zero even for the noncollinear case. Our results provide a clear picture of the trends of the magnetic order across the nickelate series and give insights into the coupling between magnetic order and structural distortions.

Introduction

Rare-earth (RE) nickelate perovskites, with the chemical formula RNiO3R\mathrm{NiO}_3, have received significant attention due to their rich phase diagram. As a function of the ionic size of RE element [1, 2, 3, 4], a combined structural and metal-insulator transition (MIT) occurs with temperature, as well as the formation of long-range magnetic order. Both can be tuned by external pressure[5], strain[6] and reduced dimensionality[7]. This tunability paves the way for engineering nanoelectronic and spintronic devices[9, 10] [8].

The structural transition, which occurs in all RE nickelates except LaNiO3_3, is characterized by a symmetry lowering from the orthorhombic PbnmPbnm to the monoclinic P21/nP2_1/n spacegroup upon cooling; it can be described mainly by a “breathing-mode” distortion, which creates expanded and compressed NiO6\mathrm{NiO}_6 octahedra arranged in a three dimensional checkerboard pattern. We refer to the two different Ni sites that result as residing in either the long bond (LB) or short bond (SB) octahedra[11] (Figure 1).

Image not extracted

Figure 1. Evolution of crystal structure of R\mathcal{R}NiO3_3 under the breathing mode distortion R1+\mathcal{R}_1^+, which leads to formation of short bond (SB) and long bond (LB) NiO6_6 octahedra. The rare-earth element R\mathcal{R} is denoted by yellow spheres. Crystal structures are visualized using the VESTA software[29].

This distortion strongly modifies the electronic configuration and corresponding magnetic moments of nickel atoms. One of the proposed descriptions of the resulting electronic structure[12] suggests that the initial electronic configuration of the Ni atoms, i.e., 3d73d^7 with spin S=12S = \frac{1}{2}, transforms to two charge-disproportionated Ni sublattices 3d63d^6 (S=0S = 0) and 3d83d^8 (S=1S = 1) under the distortion. However, this atomic picture does not consider the strong hybridization between Ni-3d3d and O-2p2p orbitals. Hence, later studies introduced a more complete description of the electronic configuration including oxygen (ligand) “holes,” denoted as L‾\underline{L}[13, 14], and describing the connected structural-MIT in nickelates as a site-selective Mott transition[13, 15, 16, 17, 18, 19]. The resulting SB and LB octahedra are described by d8L‾0d^8\underline{L}^0 and d8L‾2d^8\underline{L}^2 states with total spins tending

toward S=0S = 0 and S=1S = 1, respectively. Conventional density functional theory (DFT) based methods also support this interpretation, showing different magnetic moments of NiSB_{\mathrm{SB}} and NiLB_{\mathrm{LB}} atoms, while demonstrating nearly equal integrated charges around both Ni atoms inside the projector augmented wave (PAW) spheres[1, 15]. However, such an analysis will depend on the size of the augmentation region[20], and cannot fully capture the hybridization with the oxygens; a more complete picture may be obtained, e.g., via constructing Wannier functions for the (possibly hybridized) Ni 3d3d states[2].

At low temperatures, RE nickelates (again with the exception of LaNiO3_3) exhibit an antiferromagnetic (AFM) insulating ground state. The magnetic structure has a propagation vector of q=(14,14,14)\mathbf{q} = \left(\frac{1}{4},\frac{1}{4},\frac{1}{4}\right) (in pseudocubic notation), which was proven by neutron scattering for bulk systems[11, 21, 22, 23, 24] and resonant soft X-ray scattering (RIXS) for thin film heterostructures[25, 26]. The exact arrangement of magnetic moments in the systems is still under debate and strongly depends on the interpretation of experimental data[3, 18]. Particularly, neutron data suggests a collinear magnetic ordering, where the magnetic moment of SB nickel atoms vanish[18, 23, 24]. One candidate for the ground-state magnetic configuration is a T-type AFM (T-AFM) structure[1, 3], depicted in Figure 2(a). In this structure, the LB moments are arranged in an “up-up-down-down” (↑↑↓↓\uparrow\uparrow\downarrow\downarrow) pattern in all three Cartesian directions, and the SB moments tend to zero with increasing R1+R_1^+ amplitude.

Image not extracted

Figure 2. (a) T-type antiferromagnetic (AFM) and (b) non-collinear (NCL) magnetic configurations as a possible ground state magnetic orderings in RNiO3_3 with experimental propagation vector q=(14,14,14)\mathbf{q} = (\frac{1}{4}, \frac{1}{4}, \frac{1}{4}). Yellow and blue spheres correspond to NiSB_{\mathrm{SB}} and NiLB_{\mathrm{LB}} atoms respectively, which have smaller and larger magnetic moments corresponding to the length of the red arrows.

On the other hand, the corresponding RIXS spectra have been interpreted as a non-collinear (NCL) arrangement of spins with clearly non-zero SB magnetic moments[25, 26], also involving the ↑↑↓↓\uparrow\uparrow\downarrow\downarrow pattern in all three Cartesian directions [see Fig. 2(b)]. Here, all moments lie in the acac plane, with their direction reversed upon moving along [1,0,1][1, 0, 1]. This magnetic configuration can be considered as an orthogonal spin spiral. More importantly, the nearest neighbor SB and LB magnetic moments are perpendicular to each other, which can weaken interactions between them.

Our study is aimed at understanding the formation of these proposed complex magnetic orderings on a microscopic level in the RE nickelates. To this end we performed a systematic DFT plus Hubbard UU (DFT+UU) analysis of the magnetic properties of a series of RE nickelates: PrNiO3_3, SmNiO3_3 and LuNiO3_3. We will consider the collinear T-AFM and NCL order Figure 2(a) and (Figure 2b), as well as the ferromagnetic (FM) case, which was shown to have the lowest energy in previous DFT+UU studies[1, 2]. Spin models are constructed by obtaining Wannier functions for the hybridized Ni(3d3d)-O(2p2p) orbitals, and then using the local force theorem approach[27, 28] to extract exchange couplings. We analyze the evolution of the magnetic interactions under the R1+\mathcal{R}_1^+ structural breathing mode distortion. Our calculations reveal strong ferromagnetic nearest-neighbor and weaker antiferromagnetic next-nearest-neighbor couplings, which are induced due to magnetically active Ni dz2d_{z^2} and dx2−y2d_{x^2-y^2} states. Competition between these couplings plays the dominant role in determining the formation of magnetic order in R\mathcal{R}NiO3_3 systems and explains why previous theoretical calculations favored ferromagnetic ordering. The inclusion of spin-orbit coupling demonstrates that the magnetic anisotropy is very small, while the magnetic moments of SB nickel atoms tend to zero for finite R1+\mathcal{R}_1^+ amplitude even in the NCL case.

The remainder of this paper is organized as follows. Sec. II covers the methods and theories used in our calculations. In Sec. III, we present our main results for magnetic moments and interactions of selected rare-earth nickelate systems: PrNiO3_3, SmNiO3_3 and LuNiO3_3. We discuss the implications of our results on the proposed magnetic structures in Sec. IV. Finally, Sec. V concludes the paper.

II Theoretical framework

A Formalism

We construct the following spin model, which describes energetics of individual spins Si\mathbf{S}_i on nickel atoms interacting via isotropic exchange couplings JijJ_{ij}:

H^spin=∑i>jJijS^iS^j.(1)\hat{\mathcal{H}}^{\mathrm{spin}}=\sum_{i>j}J_{ij}\hat{\mathbf{S}}_i\hat{\mathbf{S}}_j. \tag*{(1)}

The exchange integrals can be calculated using the local force theorem approach[27, 28]

Jij=12πS2∑m,m′,n,n′×Im⁡∫−∞EFdε[Δimm′Gijm′n(ε)Δjnn′Gjin′m(ε)].(2)J_{ij}=\frac{1}{2\pi S^2}\sum_{m,m',n,n'}\times\operatorname{Im}\int_{-\infty}^{E_F}\mathrm{d}\varepsilon\left[\Delta_i^{mm'}G_{ij}^{m'n}(\varepsilon)\Delta_j^{nn'}G_{ji}^{n'm}(\varepsilon)\right]. \tag*{(2)}

where m,m′,n,n′m,m',n,n' are orbital indices running over Ni 3d3d (l=2l=2) states, EFE_F is the Fermi energy, S=12S=\frac{1}{2} is the spin quantum number[30], Δimm′=Hiimm′−Hiim′m′\Delta_i^{mm'}=H_{ii}^{mm'}-H_{ii}^{m'm'} is the on-site potential, and GG is the single-particle Green’s functions

Gijnn′(ε)=1Nk∑k,lciσnl(k)cjσn′l∗(k)ε−Eσl(k).(3)G_{ij}^{nn'}(\varepsilon)=\frac{1}{N_k}\sum_{\mathbf{k},l}\frac{c_{i\sigma}^{nl}(\mathbf{k})c_{j\sigma}^{n'l*}(\mathbf{k})}{\varepsilon-E_{\sigma}^{l}(\mathbf{k})}. \tag*{(3)}

Here NkN_k is the number of kk points in the first Brillouin zone (BZ), ciσnl∗(k)c_{i\sigma}^{nl*}(\mathbf{k}) [ciσnl(k)c_{i\sigma}^{nl}(\mathbf{k})] is the creation [annihilation] operator for the llth eigenstate in the basis of Wannier functions (see Sec. II B), with eigenvalue Eσl(k)E_{\sigma}^{l}(\mathbf{k}). The summation in Eq. (3) runs over all Ni(3d3d) and O(2p2p) orbitals; since nn and n′n' only consist of Ni(3d3d) orbitals, this corresponds to projecting the Wannier functions onto Ni(3d3d) states, thus capturing the overlap between Ni(3d3d) and O(2p2p).

The magnetic moment of a selected nickel site can also be evaluated as

mi=−1πIm⁡∫−∞EFdεTr⁡n[Giinn(ε)−Giin′n′(ε)].(4)m_i=-\frac{1}{\pi}\operatorname{Im}\int_{-\infty}^{E_F}\mathrm{d}\varepsilon\operatorname{Tr}_n\left[G_{ii}^{nn}(\varepsilon)-G_{ii}^{n'n'}(\varepsilon)\right]. \tag*{(4)}

Eq. (4) includes in the Ni moments the contribution from the overlap with oxygen. If we define a different set of Green’s functions by allowing nn and n′n' in Eq. (3) to run over all of the Wannier functions (not just nickel 3d3d), then Eq. (4) corresponds to the total magnetic moment per site, equivalent to integrating over the density of states near the Fermi level; by limiting the trace over just the Ni(3d) or just O(2p) Wannier functions, we can obtain separate moments on the nickels and oxygens, which may be compared to those integrated within the PAW spheres[20], as discussed in the next section.

Computational approach

We calculated the exchange couplings in Eq. (2) using DFT with the generalized-gradient approximation exchange-correlation functional of Perdew, Burke, and Ernzerhof[31] and the projector augmented wave (PAW) method[32] as implemented in the Vienna ab initio Simulation Package (VASP)[33, 34]. For the rare-earth atoms, we used PAW potentials corresponding to a 3+ valence state with ff-electrons frozen into the core; for Ni, the 3pp semi-core states were included as valence electrons. We set the energy cutoff of the plane-wave basis to 550 eV and the energy convergence criteria to 10−810^{-8} eV.

Correlation effects were taken into account on the static mean-field level using the DFT+UU method[35], where we considered an effective on-site Coulomb UU and Hund’s exchange JHJ_H interactions in the rotationally invariant form[36]. A value of JH=1J_H = 1 eV can be used for compounds with 3dd orbitals[37, 38], while the on-site Coulomb parameter depends on the specific system. Previous results showed that the best agreement with experiment can be obtained using a relatively small value of U=2U = 2 eV[1, 2, 39]. Therefore, we used U=2U = 2 eV and JH=1J_H = 1 eV for the majority of our DFT+UU calculations, and comment on how a larger UU (U=5U = 5 eV) may effect the results.

From the calculated electronic structure, maximally-localized Wannier functions[40] were generated using the wannier90 package[41]. We constructed Wannier functions for all occupied and unoccupied Ni 3dd and O 2pp states. This ensures that the obtained Wannier functions for both spin channels are well localized (Figure 3), and we can use them as an atomic-like basis for Eqs. (2) and (3). Note, that this choice is different to Ref. 2, were Wannier functions were only constructed for the occupied states. Since the ff electrons of the RE elements were frozen into core states, we have neglected their magnetic ordering which only occurs at very low temperatures[23, 24].

Image not extracted

Figure 3. Maximally localized Wannier functions of the long-bond (LB) and short-bond (SB) nickel atoms. (a) corresponds to dx2−y2d_{x^2-y^2} and (b) to dz2d_{z^2} orbitals for nearest neighbor LB – SB – LB nickel atoms, calculated for the T-AFM PrNiO3\mathrm{PrNiO_3} system at the experimental breathing mode amplitude R1+=0.045R_1^+ = 0.045 Å. Only spin majority orbitals at each nickel atoms are represented for clarity, spin moments are shown in (c).

To examine different magnetic orders we considered two types of cells: the primitive monoclinic cell containing 20 atoms, and a cell enlarged in the aa and cc directions containing 80 atoms (necessary for the AFM orderings); an 8×8×68\times8\times6 and 4×6×44\times6\times4 Monkhorst-Pack mesh for Brillouin-zone integration were used for the 20 and 80 atom cells, respectively. The number of kk-points within the first Brillouin zone for the evaluation of the Green’s function [Eq. (3)] and energy points for integration over a semicircular contour in the upper half of the complex energy plane [Eqs. (2) and (4)] were chosen to reach numerical convergence of exchange couplings within 0.1 meV.

For a systematic study of structural distortion, we used a symmetry-based mode decomposition analysis[42] where we varied the amplitude of specific modes expressed in terms of irreducible representations of the parent (cubic perovskite) Pm3ˉmPm\bar{3}m structure. The most significant modes that are present in the RE nickelates include octahedral rotation modes M3+M_3^+ (in-phase) and R4+R_4^+ (out-of-phase), and the breathing mode distortion R1+R_1^+[43]. The breathing mode, which is responsible for the formation of SB and LB octahedral (Figure 1), is the distortion most strongly coupled to the magnetic structure, so it will be the focus of this study. The main effect of the different RE atoms is to change the amplitude of the octahedral rotation modes. Specifically, smaller radii REs correspond to larger rotations. Thus the range in RE atoms in this study demonstrates the influence of octahedral rotations on the magnetic ordering.[44]

In our calculations, we start from the high-temperature orthorhombic PbnmPbnm structure, relaxed with non-magnetic DFT (with U=0U = 0)[1]; in this case, SB and LB octahedra have the same volume. We then gradually increase the R1+R_1^+ amplitude, calculating how the exchange couplings and magnetic moments are modified by the distortion. Thus we can study the evolution of magnetic characteristics with the transition from PbnmPbnm to the low-temperature insulating monoclinic phase P21/nP2_1/n.

Results

Magnetic moment analysis

We begin with the magnetic moments of PrNiO3\mathrm{PrNiO_3}, SmNiO3\mathrm{SmNiO_3} and LuNiO3\mathrm{LuNiO_3}, assuming FM or T-AFM ordering, using the on-site Green’s functions [Eq. (4)] in the Wannier basis. This procedure will correctly capture the hybridization effect between nickel and oxygen states and the corresponding distribution of magnetic moment density1.

Nickel magnetic moments versus R1+R_1^+ amplitude are shown in Figure 4(a)-(Figure 4c). Dashed vertical lines denote the experimental breathing-mode amplitudes for PrNiO3\mathrm{PrNiO_3} (0.0450.045 Å[45]) and LuNiO3\mathrm{LuNiO_3} (0.0750.075 Å[46]), and the theoretically predicted amplitude for SmNiO3\mathrm{SmNiO_3} (0.0600.060 Å[44]), where no experimental data on the low temperature phase is available. In the case of the high-temperature orthorhombic phase (R1+=0R_1^+ = 0 Å), in which all Ni octahedra have the same volume, FM order leads to magnetic moments of ∼0.9 μB\sim0.9\,\mu_B per nickel atom for all materials. Within the T-AFM order the PbnmPbnm symmetry is broken, causing a small variation of magnetic moments for the inequivalent Ni sites. The difference between the SB and LB moments strongly increases with larger R1+R_1^+ breathing mode amplitudes. In the FM case, the SB Ni magnetic moment smoothly decreases; for T-AFM, the moment quickly vanishes at a modest value of R1+R_1^+. The amplitude where the SB moment goes to zero depends on the rare-earth element, specifically it is larger for LuNiO3\mathrm{LuNiO_3} than PrNiO3\mathrm{PrNiO_3} and SmNiO3\mathrm{SmNiO_3}. In all cases, the R1+R_1^+ amplitudes at which the SB moment vanish are much smaller than the experimentally observed amplitudes for these compounds. Therefore, DFT+UU predicts mSB=0m_{\mathrm{SB}} = 0 for T-AFM order for all experimental structures. In contrast, mLBm_{\mathrm{LB}} increases with R1+R_1^+ and converges to ∼1.3 μB\sim1.3\,\mu_B for both FM and T-AFM order.

Image not extracted

Figure 4. Evolution of the SB and LB magnetic moments of (a) PrNiO3_3, (b) SmNiO3_3, and (c) LuNiO3_3 as a function of the R1+R_1^+ amplitude, calculated for different magnetic states using U=2U = 2 eV. Dashed vertical lines correspond to experimentally observed breathing mode amplitudes from Ref. 45 for PrNiO3_3 and Ref. 46 for LuNiO3_3, and the theoretically predicted amplitude for SmNiO3_3 from Ref. 44. The R1+=0R_1^+ = 0 structure corresponds to the high temperature orthorhombic PbnmPbnm structure. Additional dotted curves for PrNiO3_3 denoted by crosses represent results obtained with U=5U = 5 eV.

The Wannier-based analysis allows us to analyze the orbital contributions to total nickel magnetic moments, and we find that for both FM and T-AFM order, LB and SB moments originate almost entirely from Ni ege_g states. Since the corresponding Ni t2gt_{2g} states are fully occupied, they are not magnetically active.

These obtained results are consistent with the picture that would emerge from an analysis of the magnetization within the PAW spheres around Ni and O. The nickel moment from the Wannier analysis (and its dependence on R1+R_1^+) is in good agreement (∼0.01 μB\sim0.01\,\mu_B) with the moments in Ni PAW spheres. This indicates that the magnetism is localized around the atoms, instead of in the interstitial regions[47]. Similarly to the PAW estimation, the Wannier based oxygen magnetic moments for all magnetic configurations are very small (less than 0.05 μB0.05\,\mu_B). This suggests that, while there is significant Ni-O hybridization, the moment resides only at the Ni sites.

The magnitude of the magnetic moments depend on the choice of UU parameter of DFT+UU. Increasing UU leads to a stronger localization of the 3d3d electrons, and consequently gives comparably larger total magnetic moments of nickel atoms. The reason for this can be seen in the density of states, e.g., for PrNiO3\mathrm{PrNiO_3} (Figure 5), where a larger UU value leads to stronger splitting between spin channels of LB nickel ege_g states, enhancing their magnetic moment. To demonstrate this effect, we show magnetic moments calculated using U=5U = 5 eV for PrNiO3\mathrm{PrNiO_3} [see crosses in Figure 4(a)]. The resulting magnetic moments for PbnmPbnm are larger than those obtained with U=2U = 2 eV. For instance, the large R1+R_1^+ amplitude LB moments (for both magnetic orders) converge to ∼1.3 μB\sim1.3\,\mu_B for U=2U = 2 eV versus ≃1.5 μB\simeq1.5\,\mu_B for U=5U = 5 eV. Furthermore, mSBm_{\mathrm{SB}} for T-AFM order becomes zero at comparably larger value R1+≃0.05R_1^+ \simeq0.05 Å compared to ≤0.007\leq0.007 Å for U=2U = 2 eV, as was found in previous DFT+UU studies[1].

Image not extracted

Figure 5. Partial densities of states (DOS) for PrNiO3_3 with the experimental R1+=0.045R_1^+ = 0.045 Å, calculated with T-AFM order for (a) U=2U = 2 eV and (b) U=5U = 5 eV. Here, only t2gt_{2g} and ege_g states contributions of a single LB and SB nickel atom, as well as 2p2p states of the surrounding six oxygen atoms in NiO6_6 octahedra are shown. The different occupation of the ege_g spin channels of the LB nickel site carries the magnetic moment, while for SB both spin channels are equally occupied, leading to a zero moment.

B Exchange couplings

We now perform a quantitative analysis of the various magnetic interactions present in the RRNiO3_3 materials with respect to R1+R_1^+ amplitude. In Figure 6 we show a schematic of the exchange interactions included in our study. The LB and SB Ni sublattices are depicted as blue and yellow spheres, and the RE and oxygens are removed for clarity. Focusing on a LB site, there are two inequivalent nearest-neighbor interactions with SB sites in the abab plane (J1J_1). Also in the abab plane, we consider next-nearest-neighbor interactions with other LB sites J2J_2 and J3J_3, as well as next-next-nearest neighbor interactions J4J_4. In the cc direction, we have nearest neighbor interactions between LB and SB sites (J5J_5), next-nearest neighbor (J7J_7 and J6J_6) interactions.

Image not extracted

Figure 6. Magnetic model of RRNiO3_3 system with main exchange interactions plotted in (a) side and (b) top view. Blue atoms are long bond (LB) Ni, yellow are short bond (SB) (all other atoms removed). Arrows represent magnetic moments of nickel atoms in the T-AFM order. Corresponding interactions between SB nickels are also included, but not shown in the figure for clarity.

Ferromagnetic interactions J1J_1 and J5J_5

The calculated magnetic interactions via Eq. (2) depend on the reference magnetic configuration used in the underlying DFT calculation[48], since the electronic structure and the resulting single-particle Green’s functions will be different for FM and T-AFM order. Since the

T-AFM order leads to a suppression of the SB magnetic moment (Figure 4) the nearest-neighbor exchange interactions J1J_1 and J5J_5 are suppressed, as they couple two different types of nickel sites in the abab plane and along the cc direction (Figure 6). Therefore, for these interactions we focus on the FM state as the reference magnetic order. Resulting values as a function of R1+R_1^+ are given in Figure 7.

Image not extracted

Figure 7. Nearest neighbor (a) J1a/bJ_1^{a/b} and (b) J5J_5 exchange couplings (see Figure 6) between nickel atoms in RNiO3\mathcal{R}\mathrm{NiO}_3 as a function of breathing mode distortion R1+R_1^+ amplitude. The reference magnetic configuration is FM. The additional dotted curves for PrNiO3_3 denoted by crosses represent results obtained with U=5U = 5 eV.

In general, short range interactions originate from a combination of different mechanisms[37, 38]. Previous studies reported that the charge fluctuation between SB Ni and LB Ni sites with unequal spins will lead to ferromagnetic double exchange JDE<0J^{\mathrm{DE}} < 0 process, which competes with the conventional antiferromagnetic superexchange JSE>0J^{\mathrm{SE}} > 0[26]. The resulting coupling J=JSE+JDEJ = J^{\mathrm{SE}} + J^{\mathrm{DE}} can have either sign depending on which mechanism will prevail. We show in Figure 7 strong FM interaction (i.e., negative values) for both J1J_1 and J5J_5, implying ∣JSE∣<∣JDE∣\lvert J^{\mathrm{SE}}\rvert< \lvert J^{\mathrm{DE}}\rvert for both cases. Double exchange between the Ni atoms requires inequivalent charges on the sites[49, 50]. The strength of double exchange, and thus the FM couplings, would be expected to increasing with R1+R_1^+[26]. However, in our calculations the magnitude of J1J_1 and J5J_5 couplings gradually decrease with increasing R1+R_1^+ amplitudes. This suggests that other factors (e.g., charge transfer to oxygen states) are important in the resulting exchange interactions.

As a result of the presence of M3+M_3^+ and R4+R_4^+ rotation modes, the nearest neighbor exchange J1J_1 splits into two groups J1aJ_1^a and J1bJ_1^b for finite R1+R_1^+ (Figure 6). The value of this splitting depends on the rare-earth element R\mathcal{R} (c.f., solid and dashed curves in Figure 7). In particular, a stronger splitting J1a<J1bJ_1^a < J_1^b can be found for LuNiO3_3 compared to PrNiO3_3, where J1a≈J1bJ_1^a \approx J_1^b. Hence, this magnitude of the splitting is connected to the magnitude of the octahedral rotation modes M3+M_3^+ and R4+R_4^+. As mentioned in Sec. IIB, RE elements with smaller effective radii will cause stronger M3+M_3^+ and R4+R_4^+ rotations of NiO6_6 octahedra, modifying Ni–O–Ni angles, and resulting in a larger difference between J1aJ_1^a and J1bJ_1^b.

Furthermore, higher UU parameters will lead to a stronger localization of dd electrons, which influences not only the magnetic moments (Figure 4), but also the corresponding exchange interactions between them. Our calculations on PrNiO3_3 with U=5U = 5 eV show an enhancement of the nearest neighbor ferromagnetic J1J_1 and J5J_5 couplings (Figure 7). This is directly related to the fact that using a larger on-site Coulomb interaction parameter will decrease the AFM superexchange term, which is inversely proportional to UU. Double exchange is proportional to the transfer integral between Ni site, and does not depend directly on UU[50].

Antiferromagnetic interactions J4J_4 and J7J_7

Next-nearest-neighbor exchange interactions J4J_4 and J7J_7 connect the same type (SB or LB) of nickel atom (Figure 6). Similar to the nearest-neighbor J1J_1 interaction, the next nearest neighbor J4J_4 also splits into two groups J4aJ_4^a and J4bJ_4^b. The value of this splitting depends on the RE element via the octahedral rotation modes M3+M_3^+ and R4+R_4^+. These interactions are antiferromagnetic, i.e., positive (Figs. 8 and 9), and the dependence on the breathing mode distortion is different depending on the reference magnetic order. For the case of FM order, the next-nearest neighbor exchange interactions between LB nickel atoms gradually increase, while for the SB nickels they gradually decrease to zero; this is the same trend as the magnetic moments, see Figure 4. In contrast, for the T-AFM order, the SB magnetic moments are suppressed and thus the magnetic interactions between LB nickel atoms jump abruptly to a large value with R1+>0R_1^+>0. Further increasing R1+R_1^+ results in a small reduction in the magnitude of the exchange coupling. The behavior in Figure 8(a) hints at a strong coupling between R1+R_1^+ distortion and T-AFM order. Overall, exchange couplings between LB nickel sites at both reference magnetic configurations have the same order of magnitude. The values of J4J_4 and J7J_7 at R1+=0R_1^+=0 are about an order of magnitude smaller than J1/J5J_1/J_5, and thus the T-AFM magnetic configuration is fragile. Since the main mechanism of the next nearest interactions is superexchange, all these couplings weaken for a larger UU value, which is shown for J7J_7 of PrNiO3_3 in Figure 9.

Image not extracted

Figure 8. Next nearest neighbor J4a/bJ_4^{a/b} magnetic interactions between nickel atoms in RNiO3\mathcal{R}\mathrm{NiO}_3 as a function of breathing mode distortion R1+R_1^+ and reference magnetic order: (a) interactions between LB atoms in T-AFM order, while (b) and (c) correspond to interactions between LB and SB in FM order, respectively. Notation of magnetic couplings corresponds to Figure 6.

Image not extracted

Figure 9. Next nearest neighbor J7J_7 magnetic interactions between nickel atoms in RNiO3R\mathrm{NiO}_3 as a function of breathing mode distortion R1+R_1^+ and reference magnetic order: (a) interactions between LB atoms in T-AFM order, while (b) and (c) correspond to interactions between LB and SB in FM order, respectively. Additional dotted curve for PrNiO3_3 denoted by crosses represents results obtained with U=5U=5 eV. Notation of magnetic couplings corresponds to Fig. 6.

Summary and remaining exchange couplings

The values for the exchange interactions discussed above, calculated at the experimental R1+R_1^+ amplitudes[45, 46] (except for SmNiO3_3 where we use the theoretical value from Ref. 44), are given in Table I. Also included are the remaining couplings shown in Figure 6: J2J_2, J3J_3, and J6J_6; we

PrNiO3_3SmNiO3_3LuNiO3_3
R1+R_1^+0.045 Å[45]0.060 Å[44]0.075 Å[46]
J1a/J1bJ_{1a}/J_{1b}-103.9/-102.6-89.4/-81.4-83.6/-59.8
J5J_5-96.7-81.4-60.7
J4a/J4bJ_{4a}/J_{4b}38.7/37.533.0/28.329.8/17.9
J7J_741.326.97.7
J2J_23.57.510.0
J3J_39.911.616.0
J6a/J6bJ_{6a}/J_{6b}10.9/7.27.8/7.76.2/8.5

Table I. The main exchange couplings (in meV) for PrNiO3_3, SmNiO3_3 and LuNiO3_3, calculated (U=2U=2 eV) using experimental breathing mode amplitudes R1+R_1^+ for PrNiO3_3 and LuNiO3_3, and the theoretical one for SmNiO3_3. Notation of magnetic couplings corresponds to Figure 6.

see that these couplings are small compared to J1/J5J_1/J_5 and J4/J7J_4/J_7. This is because of the geometry of the active ege_g orbitals, which overlap with each other via oxygen atoms along the Cartesian axes (Figure 6). Our orbital resolved analysis of exchange interactions (Eq. (2)) demonstrates that the main contribution to in-plane J1J_1 and J4J_4 comes from dx2−y2d_{x^2-y^2} orbitals, while out-of-plane J5J_5 and J7J_7 are originated from dz2d_{z^2} states. Furthermore, it can be seen that the exchange couplings significantly increase going from Lu to Pr, e.g. when octahedral rotations are reduced.

C Anisotropy and non-collinear order

To shed light on the long-standing question of collinear versus noncollinear (NCL) order in rare-earth nickelates (Figure 2), we also performed an analysis of the magnetic anisotropy and NCL magnetic arrangements within DFT+U calculations taking spin-orbit coupling (SOC) into account. In the case of the proposed[25, 26] NCL magnetic order, nearest-neighbor magnetic moments are orthogonal to each other [Figure 2(b)]. Since mLB⋅mSB=0\mathbf{m}_{\mathrm{LB}}\cdot\mathbf{m}_{\mathrm{SB}}=0, the contributions of the strong isotropic ferromagnetic J1/J5J_1/J_5 couplings are eliminated, and magnetic frustration between FM nearest-neighbors and AFM next-nearest-neighbors couplings does not occur.

We perform calculations including SOC on the PbnmPbnm crystal structure of PrNiO3_3, SmNiO3_3, and LuNiO3_3 with breathing mode amplitudes given in Table I. We find that the SB moments for all three materials vanishes in these calculations, analogously to the collinear calculations in Figure 4. The LB moment is always nonzero and equals 1.19 μB\mu_{\mathrm{B}}, 1.24 μB\mu_{\mathrm{B}} and 1.28 μB\mu_{\mathrm{B}} for R=\mathcal{R}= Pr, Sm and Lu, respectively. Thus, the NCL configuration we find is equivalent to the collinear T-AFM order, with the LB magnetic moments free to align along the direction which minimizes the anisotropy energy in the systems; we can find this direction by calculating the anisotropic energy map[51].

In Figure 10(a), we show the interpolated anisotropic energy map for PrNiO3_3. To generate this plot, DFT+UU+SOC calculations were performed using magnetic moments constrained along various (uniformly distributed) directions, but keeping their mutual orientation consistent with T-AFM order. We find an easy axis along a direction close to the position of the apical oxygen atom in the NiO6_6 octahedra [green arrow in Figure 10(a)]. The anisotropy in the energy is very small (∼0.03\sim0.03 meV), which is expected for 3d3d systems with weak SOC[37, 38]. According to conventional theories of magnetocrystalline anisotropy (i.e., Bruno’s model[52, 53]), the orbital moment is maximized when the magnetization points along the easy axis. This is true for RNiO3_3 systems, showing the maximum value of orbital moment ML=0.13μBM_L = 0.13 \mu_B aligned with spin magnetic moment 1.19μB1.19 \mu_B along the direction of the easy axis.

Image not extracted

Figure 10. Anisotropic energy map for PrNiO3_3 system. The arrow indicates the energy minimum, which is close to the direction of apical oxygen position of MoO6_6 octahedra, depicted in (b) together with Wannier function of x2−y2x^2-y^2 symmetry, the valence electron in which state induces orbital magnetic moment.

The orbital moment is induced by ege_g electrons, which can be seen on the level of density matrix N^\hat{N}:

MLz=Tr⁡m(N^⋅L^z).M_L^z = \operatorname{Tr}_m(\hat{N} \cdot\hat{L}_z).

In particular, the valence electron occupying the dz2d_{z^2} will not produce any orbital moment (L^zdz2=0\hat{L}_z d_{z^2} = 0). On the other hand, due to a non-zero ⟨dxy∣L^z∣dx2−y2⟩=2i\langle d_{xy}|\hat{L}_z|d_{x^2-y^2}\rangle= 2i, and the fact that the calculated density matrix contains imaginary components between dxyd_{xy} and dx2−y2d_{x^2-y^2} states, an orbital moment is induced along the direction of the apical oxygen in the NiO6_6 octahedra (zz-axis) by the electron in the dx2−y2d_{x^2-y^2} state. This leads to the formation of the easy axis anisotropy found by in our DFT+UU+SOC calculations (Figure 10).

Discussion

The goal of the calculations in § 2 was to elucidate two key issues regarding the magnetic structure in RE nickelates: (a) why DFT+UU predicts a FM structure, while an AFM structure with propagation vector q=(14,14,14)\mathbf{q} = \left(\frac{1}{4}, \frac{1}{4}, \frac{1}{4}\right) is observed experimentally; and (b) what is the relation between the T-AFM magnetic structure determined from neutron diffraction data, and the NCL magnetic structure determined from RIXS.

We find that the key physical mechanism at play in these systems is the competition between strong nearest-neighbor ferromagnetic J1/J5J_1/J_5 and the much weaker next-nearest-neighbor antiferromagnetic J4/J7J_4/J_7. This competition causes magnetic frustration, and works to destabilize the long-range magnetic order. As discussed before, the R1+R_1^+ distortion will result in different magnetic moments on the LB and SB Ni atoms, with the SB magnetic moment suppressed (Figure 4). Therefore, the energy gain from forming ferromagnetic order will be directly proportional to the SB moment. Hence, the reason that DFT+UU favors FM order[1, 2] is the overestimation of short bond magnetic moments, leading to a larger contribution of nearest-neighbor FM couplings, which makes the FM solution energetically favorable. For a sufficiently small SB moment the FM J1J_1 and J5J_5 interactions are suppressed, and the next-nearest neighbor AFM couplings J4J_4 and J7J_7 will dominate; the result is the collinear T-AFM order proposed from neutron diffraction experiments[11, 21, 22, 23] (Figure 2). As we saw in Figure 4, the SB moments go to zero rapidly with increasing R1+R_1^+ amplitude in the T-AFM ordering to avoid the energy penalty associated with a finite FM J1/J5J_1/J_5 coupling.

The reason for the theory-experiment discrepancy, could be due to an overestimation of exchange splitting in DFT+UU, or problems in the refinement of experimental data. For a future analysis it would be of high interest to incorporate self-energy corrections to the Green’s function exchange coupling analysis from higher order theory, i.e. GW or dynamical mean-field theory, to rule out problems due to dynamic correlation effects[54].

Another approach to avoiding frustration is the proposed magnetic structure from RIXS experiments[25, 26]. The NCL ordering where the nearest-neighbor moments point in orthogonal directions, has no energy penalty for AFM versus FM ordering, and thus does not require the strong reduction of the SB magnetic moment needed for stabilization of collinear antiferromagnetic order. However, as discussed in § 2 C, we cannot stabilize this ordering in our calculations, as the SB moment vanishes at the experimental R1+R_1^+ amplitude, even if a SOC is included (thus a NCL order is allowed). The origin of the SB magnetic suppression can be related to the symmetry of the magnetic configurations. The SB magnetic moment is surrounded by four LB nickel atoms within the abab plane (Figure 2), and in the case of FM order, this will enhance local magnetic polarization on the SB nickel atom, increasing its magnetic moment. But since in T-AFM (and also the NCL order) half of the neighboring LB moments are aligned parallel, and the other half anti-parallel, the SB moment will be reduced. The valence electron in the magnetically active dx2−y2d_{x^2-y^2} states induces an orbital moment ML∼0.13 μBM_L \sim0.13\,\mu_B pointed at the apical oxygen position of the NiO6_6 octahedra. This leads to the formation of easy-axis anisotropy, giving preferable direction for LB magnetic moments alignment.

The analysis above is for the experimental R1+R_1^+ amplitudes (theoretical for SmNiO3_3); however, we see that these results are robust over a wide range of R1+R_1^+ amplitudes. Increasing the breathing distortion reduces FM nearest neighbor J1/J5J_1/J_5 interactions and magnetic moments of SB nickel atoms, thus it can strengthen long range T-AFM ordering in these systems. The same behavior was found in previous calculations2 by directly comparing relaxed R1+R_1^+ amplitudes in DFT+UU, which revealed R1+R_1^+ (FM) <R1+< R_1^+ (T-AFM).

The experimental estimates, used to fit magnon branches in RIXS experiment for thin film NdNiO3_3 samples, revealed that the in-abab-plane next-nearest-neighbor interaction J4J_4 is stronger that the nearest neighbor J2J_2 coupling J4≃2J2J_4 \simeq2J_23. In our calculations we also observe J4≫J2J_4 \gg J_2 (Table 1). Furthermore, we found strong spacial anisotropy depending on RE element, which splits interactions in the abab plane into two groups: nearest-neighbor J1a/J1bJ_{1a}/J_{1b}, next-nearest-neighbor J2/J3J_2/J_3 and next-next-nearest neighbor J4a/J4bJ_{4a}/J_{4b} (Figure 6). Moreover, we see a strong difference between abab plane (J1/J4J_1/J_4) and out-of-plane cc-direction (J5/J7J_5/J_7) exchange couplings depending on RE element.

Finally, the results provide a clear picture of the trends of the magnetic order across the nickelate series. While the larger octahedral distortions (R1+R_1^+, M3+M_3^+) present for smaller RE ions make the system more susceptible to the insulating breathing-mode-distorted state4, we find that they also reduce the exchange couplings, making the formation of the magnetically ordered state less favorable. This is in agreement with the experimentally-observed reduction in stability of magnetic order in the low temperature insulating phase across the series towards LuNiO3_35.

Conclusions

Based on DFT+UU calculations, Wannier orbitals and Green’s function techniques, we constructed microscopic magnetic models for selected rare-earth nickelate systems (PrNiO3_3, SmNiO3_3, and LuNiO3_3) and studied the evolution of their parameters under the R1+R_1^+ breathing mode distortion and reference magnetic configuration. Our theoretical results help to clarify several issues related to these systems. The Wannier function based analysis demonstrates that the magnetic moments reside on nickel atoms, and they are induced by active Ni ege_g states. The main exchange interactions are strong ferromagnetic interactions arising from double exchange between the nearest neighbor nickels, and somewhat weaker antiferromagnetic interactions between the next-nearest neighbor nickels induced by a kinetic superexchange mechanism. The competition between these, together with the R1+R_1^+ distortion, determines the magnetic order.

The stronger nearest-neighbor interactions are the reason that DFT+UU calculations find the ferromagnetic state to be lowest in energy. The stabilization of collinear T-AFM order, which is compatible with the experimental symmetry, is possible due to suppression of short bond magnetic moments with the onset of the breathing mode distortion, eliminating the contribution of the ferromagnetic interactions. We find that even if spin-orbit coupling is included and a non-collinear ordering is allowed, the SB Ni moments are still suppressed, and T-AFM order is restored with an easy axis magnetization pointed towards the apical oxygen. Thus our work determines the reason for the general disagreement between the DFT-UU predicted magnetic order (FM) and experiment (AFM), as well as provides a clear explanation how the q=(14,14,14)\mathbf{q} = (\frac{1}{4},\frac{1}{4},\frac{1}{4}) magnetic order emerges in rare-earth nickelates, and how the trends evolve across the series.

Acknowledgments

D.I.B. is grateful to S. A. Nikolaev (Tokyo Institute of Technology) for helpful discussions. The calculations have been performed using the facilities of the Flatiron Institute. The Flatiron Institute is a division of the Simons Foundation. C.E.D. acknowledge support from the National Science Foundation under Grant No. DMR-1918455.

References

  1. [1]A. Hampel and C. Ederer, Phys. Rev. B 96, 165130 (2017).
  2. [2]J. Varignon, M. N. Grisolia, J. Íñiguez, A. Barthélémy, and M. Bibes, npj Quantum Materials 2, 21 (2017).
  3. [3]S. Catalano, M. Gibert, J. Fowlie, J. Íñiguez, J.-M. Triscone, and J. Kreisel, Rep. Prog. Phys. 81, 046501 (2018).
  4. [4]Y. Lu, Z. Zhong, M. W. Haverkort, and P. Hansmann, Phys. Rev. B 95, 195117 (2017).
  5. [5]X. Obradors, L. M. Paulius, M. B. Maple, J. B. Torrance, A. I. Nazzal, J. Fontcuberta, and X. Granados, Phys. Rev.
  6. [6]S. Catalano, M. Gibert, V. Bisogni, O. E. Peil, F. He, R. Sutarto, M. Viret, P. Zubko, R. Scherwitzl, A. Georges, G. A. Sawatzky, T. Schmitt, and J.-M. Triscone, APL Materials 2, 116110 (2014).
  7. [7]A. V. Boris, Y. Matiks, E. Benckiser, A. Frano, P. Popovich, V. Hinkov, P. Wochner, M. Castro-Colin, E. Detemple, V. K. Malik, C. Bernhard, T. Prokscha, A. Suter, Z. Salman, E. Morenzoni, G. Cristiani, H.-U. Habermeier, and B. Keimer, Science 332, 937 (2011). C. Y. B. L. Y. Sun, H. Yang, and X. Zhang, Electrochimica Acta 174, 41 (2015).
  8. [9]J. Shi, Y. Zhou, and S. Ramanathan, Nature Communications 5, 4860 (2014).
  9. [10]S.-W. Cheong and M. Mostovoy, Nature Materials 6, 13 (2007).
  10. [11]J. A. Alonso, J. L. García-Muñoz, M. T. Fernández-Díaz, M. A. G. Aranda, M. J. Martínez-Lope, and M. T. Casais, Phys. Rev. Lett. 82, 3871 (1999).
  11. [12]I. I. Mazin, D. I. Khomskii, R. Lengsdorf, J. A. Alonso, W. G. Marshall, R. M. Ibberson, A. Podlesnyak, M. J. Martínez-Lope, and M. M. Abd-Elmeguid, Phys. Rev. Lett. 98, 176406 (2007).
  12. [13]S. Johnston, A. Mukherjee, I. Elfimov, M. Berciu, and G. A. Sawatzky, Phys. Rev. Lett. 112, 106404 (2014).
  13. [14]R. J. Green, M. W. Haverkort, and G. A. Sawatzky, Phys. Rev. B 94, 195127 (2016).
  14. [15]H. Park, A. J. Millis, and C. A. Marianetti, Phys. Rev. Lett. 109, 156402 (2012).
  15. [16]B. Lau and A. J. Millis, Phys. Rev. Lett. 110, 126404 (2013).
  16. [17]A. Subedi, O. E. Peil, and A. Georges, Phys. Rev. B 91, 075128 (2015).
  17. [18]G. Haule and L. Pascut, G, Scientific Reports 7, 10375 (2017).
  18. [19]O. E. Peil, A. Hampel, C. Ederer, and A. Georges, Phys. Rev. B 99, 245127 (2019).
  19. [20]G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999).
  20. [21]J. L. García-Muñoz, J. Rodríguez-Carvajal, and P. Lacorre, Phys. Rev. B 50, 978 (1994).
  21. [22]D. J. Gawryluk, Y. M. Klein, T. Shang, D. Sheptyakov, L. Keller, N. Casati, P. Lacorre, M. T. Fernández-Díaz, J. Rodríguez-Carvajal, and M. Medarde, Phys. Rev. B 100, 205137 (2019).
  22. [23]A. Muñoz, J. A. Alonso, M. J. Martínez-Lope, and M. T. Fernández-Díaz, J. Solid State Chem. 182, 1982 (2009).
  23. [24]M. T. Fernández-Díaz, J. A. Alonso, M. J. Martínez-Lope, M. T. Casais, and J. L. García-Muñoz, Phys. Rev. B 64, 144417 (2001).
  24. [25]V. Scagnoli, U. Staub, A. M. Mulders, M. Janousch, G. I. Meijer, G. Hammerl, J. M. Tonnerre, and N. Stojic, Phys. Rev. B 73, 100409 (2006).
  25. [26]Y. Lu, D. Betto, K. Fürsich, H. Suzuki, H.-H. Kim, G. Cristiani, G. Logvenov, N. B. Brookes, E. Benckiser, M. W. Haverkort, G. Khaliullin, M. Le Tacon, M. Minola, and B. Keimer, Phys. Rev. X 8, 031014 (2018).
  26. [27]A. Liechtenstein, M. Katsnelson, V. Antropov, and V. Gubanov, Journal of Magnetism and Magnetic Materials 67, 65 (1987).
  27. [28]V. V. Mazurenko and V. I. Anisimov, Phys. Rev. B 71, 184434 (2005).
  28. [29]K. Momma and F. Izumi, J. Appl. Crystallogr. 44, 1272 (2011).
  29. [30]For simplicity, we use spin quantum number S = 1/2 in Eq. (2) during the whole range of breathing distortions. It was revealed that very large breathing distortion amplitudes are needed to achieve spins 0 and 1 for SB and LB magnetic moments13,14, which is much higher than our considered cases.
  30. [31]J. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 (1996).
  31. [32]P. E. Blöchl, Phys. Rev. B 50, 17953 (1994).
  32. [33]G. Kresse and J. Furthmüller, Computa. Mater. Sci. 6, 15 (1996).
  33. [34]G. Kresse and J. Furthmüller, Phys. Rev. B 54, 11169 (1996).
  34. [35]V. I. Anisimov, J. Zaanen, and O. K. Andersen, Phys. Rev. B 44, 943 (1991).
  35. [36]A. I. Liechtenstein, V. I. Anisimov, and J. Zaanen, Phys. Rev. B 52, R5467 (1995).
  36. [37]D. I. Badrtdinov, O. S. Volkova, A. A. Tsirlin, I. V. Solovyev, A. N. Vasiliev, and V. V. Mazurenko, Phys. Rev. B 94, 054435 (2016).
  37. [38]D. I. Badrtdinov, V. V. Mazurenko, and A. A. Tsirlin, Phys. Rev. B 100, 214401 (2019).
  38. [39]A. Mercy, J. Bieder, J. Íñiguez, and P. Ghosez, Nat. Commun. 8, 1677 (2017).
  39. [40]N. Marzari and D. Vanderbilt, Phys. Rev. B 56, 12847 (1997).
  40. [41]G. Pizzi, V. Vitale, R. Arita, S. Blügel, F. Freimuth, G. Géranton, M. Gibertini, D. Gresch, C. Johnson, T. Koretsune, J. Ibañez-Azpiroz, H. Lee, J.-M. Lim, D. Marchand, A. Marrazzo, Y. Mokrousov, J. I. Mustafa, Y. Nohara, Y. Nomura, L. Paulatto, S. Poncé, T. Ponweiser, J. Qiao, F. Thöle, S. S. Tsirkin, M. Wierzbowska, N. Marzari, D. Vanderbilt, I. Souza, A. A. Mostofi, and J. R. Yates, Journal of Physics: Condensed Matter 32, 165902 (2020).
  41. [42]J. M. Perez-Mato, D. Orobengoa, and M. I. Aroyo, Acta Crystallographica Section A 66, 558 (2010).
  42. [43]P. V. Balachandran and J. M. Rondinelli, Physical Review B 88, 054101 (2013).
  43. [44]A. Hampel, P. Liu, C. Franchini, and C. Ederer, npj Quantum Materials 4, 5 (2019).
  44. [45]M. Medarde, M. T. Fernández-Díaz, and P. Lacorre, Phys. Rev. B 78, 212101 (2008).
  45. [46]J. A. Alonso, M. J. Martínez-Lope, M. T. Casais, J. L. García-Muñoz, M. T. Fernández-Díaz, and M. A. G. Aranda, Phys. Rev. B 64, 094102 (2001).
  46. [47]D. I. Badrtdinov, S. A. Nikolaev, A. N. Rudenko, M. I. Katsnelson, and V. V. Mazurenko, Phys. Rev. B 98, 184425 (2018).
  47. [48]X. Zhu, A. Edström, and C. Ederer, Phys. Rev. B 101, 064401 (2020).
  48. [49]C. Zener, Phys. Rev. 82, 403 (1951).
  49. [50]S. Keshavarz, Y. O. Kvashnin, D. C. M. Rodrigues, M. Pereiro, I. Di Marco, C. Autieri, L. Nordström, I. V. Solovyev, B. Sanyal, and O. Eriksson, Phys. Rev. B 95, 115120 (2017).
  50. [51]A. H. Abdeldaim, D. I. Badrtdinov, A. S. Gibbs, P. Manuel, H. C. Walker, M. D. Le, C. H. Wu, D. W. Wardecski, S.-G. Eriksson, Y. Kvashnin, A. A. Tsirlin, and G. J. Nilsen, Phys. Rev. B 100, 214427 (2019).
  51. [52]P. Bruno, Phys. Rev. B 39, 865 (1989).
  52. [53]Magnetism. From fundamentals to nanoscale dynamics (Springer, Berlin Heidelberg, 2006).
  53. [54]Y. O. Kvashnin, O. Grånäs, I. Di Marco, M. I. Katsnelson, A. I. Liechtenstein, and O. Eriksson, Phys. Rev. B 91, 125133 (2015).

Paper details

vlnc:W_wzqywdjk
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents