betaFeedback

WorksPhysics and Astronomy

Altermagnetism and Weak Magnetism in the Insulating Distorted Perovskite Antiferromagnet NaOsO3_3

Abstract

The GdFeO3_3-type perovskite antiferromagnet NaOsO3_3, calculated here to be altermagnetic for all three typical collinear antiferromagnetic orders, was suggested early on to be a Slater-type insulator, due in large part to its continuous metal-insulator transition and its small energy gap. Below the Néel temperature, the gap opens along with ``weak magnetism'', accompanied by a sharp change in the magnetic susceptibility and resistivity. Without explicit correlation in the band structure calculation, and neglecting spin-orbit coupling (SOC), already a gap opens. Inclusion of a modest on-site Coulomb repulsion (U∼U\sim1 eV) is sufficient to eliminate a SOC-induced small band overlap, opening a gap similar to the experimentally observed gap of around 100 meV. Combined evidence supports the viewpoint that NaOsO3_3 lies in an unusual crossover region between Slater and Mott insulator. The unreported altermagnetism in NaOsO3_3 is demonstrated and its consequences are considered. The origin of the very weak magnetism has been investigated using a combination of ab initio calculations and symmetry analysis of the magnetic space group, confirming the origin lying in the Dzyaloshinskii-Moriya SOC buttressed by altermagnetic order. After determining the easy axis, our calculation leads to an Os spin canting angle of about 3∘^{\circ}, accounting for the observed weak magnetism and sharp change in the susceptibility. The altermagnetism spin-split bands (up to ∼\sim100 meV) are accompanied by a chiral-split magnon spectrum in both acoustic and optical modes in the THz range, and lead to significant anomalous Hall conductivity upon hole doping.

Introduction

Recently altermagnetism has become an active topic in condensed matter physics, since altermagnets (AMs) display unexpected properties, viz. spin-split bands in a zero net moment antiferromagnet (AFM), with or without the help of spin-orbit coupling (SOC). Other properties, observed in various topological systems as well as ferromagnet-like features with no stray field, show promising potential for spintronic applications [1, 2, 3, 4, 5]. In conventional AFMs, spin-opposite sublattices are mapped by translation tt or inversion PP symmetries, leading to Kramers degeneracy throughout the Brillouin zone (BZ). However, in AMs the AFM sublattice sites are connected by a combination of rotational CnjC_{nj} (nn-fold rotation about the j^\hat{j}-axis) and nonprimitive translation symmetries. As a result, the Kramers degeneracy is broken in some regions of BZ [6, 7, 8, 9, 10, 11, 12, 13], leading to spin-split bands in the selected regions in spite of vanishing net moment. After this phenomenon was theoretically reported in rutile AFM RuO2_2 by Ahn, Hariki, Lee, and Kuneš in 2019 [7], various unique features of this novel AFM have been systematically investigated by Šmejkal, Sinova, and Jungwirth in 2022 [1, 2], after which the phenomenon became known as altermagnetism.

Another particularity is that spin-splittings in the AM lead to chiral magnons that are anticipated to be a useful new property for applications to magnonics due to linear dispersion and the high energy scale of THz [14, 15]. Spontaneous crystal thermal Hall effects are proposed in both metallic or insulating AMs [16, 17], suggesting promising candidates for application to spin caloritronics. Additionally, the crystal chirality in a conducting AM induces anomalous Hall conductivity (AHC) [18, 19, 20, 21], a property that is more often associated with metallic ferromagnets.

In this paper we focus on the GdFeO3_3-type, quadrupled orthorhombic perovskites, NaOsO3_3 in particular. Perovskites typically exhibit three possible collinear AFM spin orders: G type (antialigned neighboring spins), C type (aligned along one direction and antialigned along a perpendicular direction), and A type (alternatively antialigned layers). In these orderings, the magnetic sublattices are connected by a combination of two-fold rotational C2jC_{2j} and non-primitive translation t12t_{\frac{1}{2}} symmetries characteristic of AMs [see Supplemental Material (SM) 22, for details]. The crystal structure and AFM alignment dictate that AFM GdFeO3_3-type perovskites are AMs in all of these collinear AFM orders [23]. In this paper we provide direct evidence that AFM NaOsO3_3 possesses spin-split bands and chiral magnons, the two most evident properties of AMs, and related properties are discussed.

Over last 15 years, GdFeO3_3-type NaOsO3_3 has been investigated on both experimental [24, 25, 26, 27, 28, 29, 30, 31] and theoretical sides [32, 33, 34, 35, 36, 37, 38, 39, 40], with an overview given in the next section. Perhaps the most unusual feature on NaOsO3_3 is that it was suggested [24] to be a promising candidate for a continuous Slater metal-insulator transition (MIT) [41] where a gap opens due to antiferromagnetic order with no relevant change of the crystal, either symmetry or internal structure. Previous theoretical work (see the next section) has shown that correlation corrections, though not large, are necessary to account for the insulating and AFM ground state that is observed. Correlations argue for a Mott transition involving correlation, Slater’s original picture [41] was based on a Hartree-Fock treatment with exchange being the underlying interaction. Since the appropriate picture of this transition is of primary interest in NaOsO3NaOsO_3, it is worthwhile to consider this rather high temperature MIT somewhat more broadly.

Some time ago Halperin and Rice provided a survey of such transitions from the point of view of a MIT with increasing temperature [42, 43], arising from a (most interestingly, indirect) bandgap closing, which seems to fit the current theoretical picture and much of the experimental data. At that time, their picture of a Mott transition was of a localized-to-itinerant electronic system, which would be first order (“no phase transition”). The Slater transition would be continuous and second order, the AFM order disappearing continuously to zero at the transition, unlike the discontinuous AFM order parameter at a Mott transition. Observed MITs have become much more complicated than these simple pictures as more interactions are included, viz. the temperature dependence of band energies due to electron-phonon coupling [44], and more complex compounds and active multiorbital electronic structures have produced ever richer behaviors [45].

However, the main emphasis of Halperin and Rice was other effects, of interactions and temperature. Assuming sharp band structures and a well defined gap Eg(T)E_g(T), they argued that (especially for indirect gaps) the electron-hole attraction would lead to an excitonic phase before the gap vanished. A great deal of research has been conducted on excitonic behavior in the meantime. There are other important effects they did not consider. The electron-phonon interaction, often strong in transition-metal oxides and not considered by them, leads to renormalization of the bands and gap from the band value,[44] [46] as well as providing lifetime effects thus broadening bands and somewhat smoothing the gap, and making the description of a sharp MIT (whether first or second order) a more involved task.

In this paper, after demonstrating band splitting and chiral magnons, we will focus on these questions of importance: (1) how does AFM figure into these phenomena or the properties of NaOsO3NaOsO_3? (2) what is the origin of the observed very weak magnetization? (3) how might one best consider the Slater versus Mott insulator question? The organization of the paper is as follows. Section II provides an overview of experimental data and theoretical modeling of NaOsO3NaOsO_3, with our Methods described in Sec. III. The main results from band modeling, on magnonic properties, the source of the weak magnetism, and anomalous Hall effect in the doped system are provided in several subsections of Sec. IV. The paper concludes with a Discussion and Summary in Sec. V.

Survey of data and previous theory

Shi et al. [24] discovered that NaOsO3NaOsO_3 becomes AFM at the relatively high Néel temperature of TN=410T_N = 410 K and correspondingly large Curie-Weiss temperature of ΘCW=−1950\Theta_{CW} = -1950 K. This was coincident with a continuous opening of a gap [24]. Neutron powder diffraction (NPD) measurements by Calder et al. indicated G-type AFM with the easy axis along the c^\hat{c}-direction [25], confirmed by resonant inelastic x-ray scattering (RIXS) [26] and several calculations based on density functional theory (DFT) [32, 33, 34, 35, 36]. Vecchio et al. used infrared and THz spectroscopy to monitor the energy gap [30], giving a value of the low temperature optical gap of ∼105\sim105 meV and a fully gapped Fermi surface . At the MIT the crystal symmetry and the volume remained unchanged [24, 25, 26, 27, 28, 29], supporting the suggestion of Shi et al. [24] that the AFM of NaOsO3NaOsO_3 is due to Slater transition.

Through RIXS, the spin gap was measured as 58–60 meV [26, 27, 28]. Measurements of non-resonant and resonant x-ray single-crystal diffraction at the Os L3L_3 edges showed the space-group forbidden (300) Bragg peak as well as the magnetic (330) peak emerged at TNT_N [29]. The forbidden peak was interpreted as the change in the electron density of Os, suggested to support the Slater mechanism [29]. High pressure experiments indicate that this small gap insulating state remains robust up to 35 GPa with the MIT critical temperature dropping to Tc≃300T_c \simeq300 K[31].

Vale et al. observed the magnetic excitation through RIXS that was weakly Landau-damped upon approaching the metallic region [27, 28]. This feature was well described by a weak AFM Fermi liquid picture, suggesting that NaOsO3NaOsO_3 is not well suited to the standard Slater picture [27, 28]. On the other hand, the Bragg reflections were well reproduced with uncorrelated abab initio calculations, suggesting that this aspect of NaOsO3NaOsO_3 can be accounted for within the Slater picture [29].

Several theoretical works based on the DFT have appeared, although nothing about the AM aspect. Several DFT+SOC+U calculations established SOC to be important, and proposing that an effective Coulomb repulsion Ueff=U−JHU_{eff} = U - J_H somewhat larger than 1 eV (mostly Ueff≈2U_{eff} \approx2 eV) seemed appropriate to account for the MIT [32, 33, 34, 37], implying that correlation, but only a modest amount, is necessary to account for the MIT. Here JHJ_H is the conventional Hund’s exchange integral. Zhou et al. pointed out the small change in the slope of the temperature dependence of the lattice parameter difference ∣c−a∣|c-a|, with onset below TNT_N (=TMIT= T_{MIT}) [37], although no change in volume versus temperature was observed. They suggested that this system is close to the Mott physics picture in spite of the very small gap. Alternatively, Kim et al. suggested a magnetically induced continuous change in the topology of the Fermi surface, called a spin-driven Lifshitz MIT, based on the coexistence of weak correlation and itinerant magnetism [38].

Measurements of isothermal magnetization showed a weak spontaneous magnetization of 0.005μB\mu_{\mathrm{B}}/Os at 5 K [24], while no evidence for spin canting was detected in the NPD Bragg peaks [25]. Although magnetic frustration is not an issue in perovskites, the unusually high Curie-Weiss temperature can be used to extract the empirical frustration factor, defined by Ramirez [47], of ΘCWTN≈4.8\frac{\Theta_{\mathrm{CW}}}{T_{\mathrm{N}}} \approx4.8. This value is comparable with that of the magnetically frustrated system YMnO3_3 [47], leaving this large ratio as an open question. Calculations based on DFT of non-collinear magnetic configurations indicated that G-type non-collinear ordering is energetically favored by several meV over collinear G-type ordering [34]. Currently the origin of the observed weak magnetism remains elusive.

Methods

We have used the experimentally observed lattice parameters of a=5.3842a = 5.3842, b=7.5804b = 7.5804, c=5.3282c = 5.3282 (in units of Å) in the GdFeO3_3-type 2×2×2\sqrt{2} \times2 \times\sqrt{2} perovskite structure (space group: No. 62 Pnma, with generators of inversion and two non-symmorphic operations) [24], as used in the previous paper by some of the current authors [33]. In this distorted perovskite structure, an O–Os–O axis rotates by 9° about the a^\hat{a} (c^\hat{c}) axis and tilts by 11° away from the b^\hat{b} axis. Although the distortions are moderate, they play a crucial role in the emergence of the AFM/MIT transition, as shown in previous calculations [33, 39, 40], and hence the low temperature AM state.

Our ab initio calculations, based on the generalized gradient approximation (GGA) of Perdew, Burke, and Ernzerhof [48], were carried out with the Vienna ab initio simulation package (VASP) [49, 50]. An energy cutoff of 600 eV was employed for the plane-wave basis set, and the BZ was sampled by a 16×12×1616 \times12 \times16 Monkhorst-Pack kk-point mesh [51]. SOC and correlation effects were treated within GGA+SOC, GGA+UU, and GGA+UU+SOC approaches, as implemented in VASP. Correlation effects in the half-filled t2gt_{2g} shell were included by the effective on-site Coulomb repulsion Ueff=U−JHU_{\mathrm{eff}} = U - J_{\mathrm{H}}, using the scheme proposed by Dudarev et al. [52] that depends only on UeffU_{\mathrm{eff}}. We varied UeffU_{\mathrm{eff}} in the 0.5–2 eV range to include correlation effects up to the value that is reasonable for Os ions in the small-gap semiconductor. As expected, increasing UeffU_{\mathrm{eff}} enlarges the energy gap and slightly varies the magnitude of the parameters studied here. However, the characteristic features induced by AM, e.g. chiral-split magnon and significant AHC around the Fermi energy EFE_{\mathrm{F}}, remain almost unchanged. In particular, compared to the existing experimental data, the choice of Ueff=0.5U_{\mathrm{eff}} = 0.5 eV in the GGA+UU+SOC calculations well reproduces the energy gap and the (maximum) values of

the magnon spectrum, as will be shown below. Thus, we have focused mainly on results for Ueff=0.5U_{\mathrm{eff}} = 0.5 eV.

Maximally localized Wannier functions [53] were constructed using the WANNIER90 program [54]. For the Wannierization, we used 56 orbitals including all of the Os 5d5d and O 2p2p orbitals in the quadruple perovskite cell, showing a good fit in the range of -9 eV to +7eV [22]. The symmetries of the Wannier Hamiltonian obtained above were analyzed, using the WANNsymm code [55]. From the symmetrized Hamiltonian, the magnetic exchange parameters and the Dzyaloshinskii-Moriya interaction (DMI) constants [56, 57] were calculated using the TB2J package [58] in which the local rigid spin rotation is treated as a perturbation within the single-particle Green’s function method [59].

Using the exchange constants, the spin dynamics was simulated by the UPPASD program [60]. For rigid band doping, the AHC can be calculated from ab initio calculations of the Berry curvature using the Wannier Hamiltonian, implemented in the WANNIERTOOLS code [61].

Results

Energetics and gap opening

To verify our results obtained from calculations based on the GGA, first we investigated energetics and compared with the experimental measurements and previous theoretical results. Consistent with the literature [25, 26, 32, 33, 34, 35], the AFM state has a much lower energy than the nonmagnetic and ferromagnetic states. At the GGA level, the G-AFM is energetically favored over the A-type by 36 meV/f.u. and the C-type by 55 meV/f.u., consistent with the previous calculations [33]. To find the easy axis, we carried out GGA+SOC calculations for three directions of the Néel vector N=m1−m2\mathbf{N} = \mathbf{m}_1 - \mathbf{m}_2 in terms of the two sublattice magnetic vectors [18], until SOC is added. Consistent with the experimental observations [25, 26], the N∥z^\mathbf{N} \parallel\hat{z} state has a little lower energy than the N∥x^\mathbf{N} \parallel\hat{x} and N∥y^\mathbf{N} \parallel\hat{y} states by 3.2 meV and 4.1 meV per f.u., respectively. Including the correlation effects, the energetics remains unchanged.

Figure 1 shows the band structures enlarged around EFE_{\mathrm{F}} for the GGA, GGA+SOC(N∥z^\mathbf{N} \parallel\hat{z}), and GGA+SOC(N∥z^\mathbf{N} \parallel\hat{z})+UU with Ueff=0.5U_{\mathrm{eff}} = 0.5 eV. In contrast to our previous results of the local spin density approximation (LSDA) [33], pure GGA opens a gap EgE_{\mathrm{g}} of 150 meV for the AFM state without a help of UU. It is rare for these two approaches, the GGA and the LSDA, to show opposite results regarding whether a system is metallic or insulating. The reason is unclear, but highlights yet another peculiarity of NaOsO3_3. Moreover, including SOC a small negative indirect gap around 50 meV appears between the R\mathit{R} and T\mathit{T} points in the middle panel of Figure 1, although the valence and conduction bands remain disjoint. After this, inclusion of a small on-site Coulomb repulsion is sufficient to re-open a gap. At $U_{\mathrm{eff}} = 0.5 eV, our calculated gaps are 29, 2, 70 (in units of meV) for N∥x^\mathbf{N}\parallel\hat{x}, y^\hat{y}, and z^\hat{z}, respectively. For the easy axis N∥z^\mathbf{N}\parallel\hat{z}, the energy gap is consistent with the experimental value of 60–100 meV. At Ueff=1(2)U_{\mathrm{eff}}=1(2) eV, the energy gap is about 0.2 (0.5) eV, at least twice larger than the experimental value. Thus, now we will focus on the results obtained from GGA+SOC(N∥z^\mathbf{N}\parallel\hat{z})+U at Ueff=0.5U_{\mathrm{eff}}=0.5 eV, unless mentioned otherwise. Considering a Hund’s integral JH≈0.5J_{\mathrm{H}}\approx0.5 eV typical for osm­ates [33], this UeffU_{\mathrm{eff}} corresponds to the Hubbard U≈1U\approx1 eV, around half of that indicated by previous DFT calculations [32, 33, 34, 37].

Image not extracted

Figure 1. Enlarged band structures of NaOsO3_3 with (top) GGA, (middle) GGA+SOC, and (bottom) GGA+SOC+U near EFE_{\mathrm{F}}. Here, Ueff=0.5U_{\mathrm{eff}}=0.5 eV and the Néel vector N∥z^\mathbf{N}\parallel\hat{z} were used. The spin characters are colored (red: spin-up, blue: spin-down), revealing the spin splittings along the R−ΓR-\Gamma line, indicating the altermagnetism character. Note that the splitting occurs only along the R−ΓR-\Gamma line (and inside the zone), but not along other symmetry directions. (For details, see § 3 B.) Band sticking is discussed in the text.

In the band structure plots enlarged in the −1-1 eV to 11 eV in Figure 1, there is band sticking along certain symmetry lines. Bands “stick together” at the top of the BZ, along the Z−U−R−T−ZZ-U-R-T-Z line, a common feature of non-symmorphic crystals [62]. However, bands also stick together along the X−S−YX-S-Y line in the basal plane, due to the second non-symmorphic group generator.

Spin-splitting induced by altermagnetism

An early example of AM-induced band splitting in an AFM was demonstrated in RuO2_2 by Ahn et al. [7] In the NaOsO3_3 band structures displayed in Figure 1, this characteristic feature of AM appears, recall that conventional AFMs have degeneracy related by reversal of spins. Clearly visible are spin-splittings along the R−ΓR-\Gamma line, and at the top of the valence bands and the bottom of the conduction bands, within the range of around −1-1 eV to 11 eV. The characters of bands in this region are mostly the Os 5d5d orbitals, and band-resolved magnetization is evident. Since in other energy ranges the splittings are negligible, this implies that the magnitude of splitting significantly depends on strength of the exchange interaction in specific bands.

As far as we know, no established way exists for determining the magnitude of spin splitting, yet. We scanned the entire BZ to investigate the magnitude of splitting at a certain k⃗\vec{k}-point, since its magnitude varies across the BZ. In NaOsO3_3 which has the spin space group of P1ˉn1ˉm1ˉa∞m1P\bar{1}n\bar{1}m\bar{1}a\infty m1 [63, 64], the Kramers spin degeneracy survives on the kz(ky)=0k_z(k_y)=0 and πc\frac{\pi}{c} (πa\frac{\pi}{a}) planes due to the symmetry of {2∥m001∥t12012}\{2\parallel m_{001}\parallel t_{\frac{1}{2}0\frac{1}{2}}\} ({m⊥∥m010∥t0120}\{m_{\perp}\parallel m_{010}\parallel t_{0\frac{1}{2}0}\}), i.e., spin reversing operation and a mirror reflection about the zz (yy)-axis. Thus, the spin-splittings appear only inside the BZ and on the kx=0,πak_x=0,\frac{\pi}{a} planes [22], as sketched in the bottom panel of Figure 2.

Image not extracted

Figure 2. Top: maximum splitting on the kx−kyk_x-k_y planes, for kzk_z varying through the range of 0−πc0-\frac{\pi}{c}. The variation in splittings is approximately quartic, as indicated from a fit to the data and shown by the red dashed lines (see text for more discussion). Bottom: sketch of the region showing spin-splitting in the orthorhombic BZ. In the red-shaded or lined regions, the Kramers spin-degenerate bands survive.

Using bands at the GGA level, Figure 2 presents the maximum splitting for each kx−kyk_x-k_y plane, varying kzk_z between 00 and πc\frac{\pi}{c}. On most planes, the maximum splitting appears accidentally in the midway of the [01kz]−[00kz][01k_z]-[00k_z] line. (For details, see the SM [22].) At the GGA level the largest splitting is about 200 meV, occurring at the top of valence bands at the $k_z\approx 0.6\frac{\pi}{c}$. With GGA+SOC+$U$ these reduce to about a half of the GGA value. Compared with the splittings in prototype AMs RuO$_2$, MnTe and CrSb, that of GGA in NaOsO$_3$ is about one fifth of those, but is larger than or comparable to other candidate AMs [1, 2]`. The plotted maximum vanishes at the endpoints due to symmetry, and is approximately quadratic with some displacement of the apex from the center.

Weak magnetism

SOC induces spin canting, and hence weak magnetism (WM) in distorted AFMs via the Dzyaloshinskii-Moriya mechanism. Presence of this WM can be analyzed through symmetries of the magnetic space groups that depend on the direction of the Néel vector. Neglecting SOC, each magnetic space group has symmetries of

TABLE II. Each component (in units of μB\mu_B) of spin and orbital moment vectors for GGA+SOC(N∥z^\parallel\hat{z})+UU, obtained from the VASP. There are two-pairs of antialigned magnetic Os ions in the unit cell.

ionspin xxspin yyspin zzorbital xxorbital yyorbital zz
Os1−0.058-0.058−0.018-0.0181.1391.1390.0040.004−0.016-0.016−0.068-0.068
Os2−0.058-0.058−0.018-0.018−1.139-1.139−0.004-0.004−0.016-0.016−0.068-0.068
Os30.0580.058−0.018-0.0181.1391.139−0.004-0.004−0.016-0.016−0.068-0.068
Os40.0580.058−0.018-0.018−1.139-1.139−0.004-0.004−0.016-0.0160.0680.068
sum/(ion)00−0.018-0.0180000−0.016-0.01600

a combination of rotation C2jC_{2j}, translation t12t_{\frac{1}{2}}, inversion PP, and time reversal TT operations, as given in Table 1, obtained using the WANNsymm code [55]. In the each space group, the magnetic moment m\mathbf{m} or Hall vectors h\mathbf{h} exists when each vector is invariant for the all symmetry operations of the space group. These vectors are affected by the inversion PP (r⃗→−r⃗\vec{r}\to-\vec{r}) and the rotation C2jC_{2j} (r⃗⊥→−r⃗⊥\vec{r}_{\perp}\to-\vec{r}_{\perp}, components perpendicular to the j^\hat{j}-axis). On the other hand, the time reversal or translation operator does not change these vectors that are quantities obtained from an integration over the BZ. For a moment m\mathbf{m}, for example, operations of C2xmC_{2x}\mathbf{m} and C2xPmC_{2x}P\mathbf{m} produce (mx,−my,−mz)(m_x,-m_y,-m_z) and (−mx,my,mz)(-m_x,m_y,m_z), respectively. Since these should be equal to m\mathbf{m} to be invariant, my=mz=0m_y=m_z=0 and mx=0m_x=0. Thus non-zero m\mathbf{m} is not allowed in this case. Therefore, m\mathbf{m} is allowed only along the z^\hat{z} and y^\hat{y} for the cases of N∥y^\mathbf{N}\parallel\hat{y} and ∥z^\parallel\hat{z}, respectively, whereas m=0\mathbf{m}=0 for N∥x^\mathbf{N}\parallel\hat{x}.

Dir. N\mathbf{N}MSGsymmetries
x^\hat{x}Pn^ma.1P\hat{n}ma.1 (No. 62.441)C2zt12,C2zt12P;C2xt12,C2xt12P;C2yt12,C2yt12PC_{2z}t_{\frac{1}{2}}, C_{2z}t_{\frac{1}{2}}P; C_{2x}t_{\frac{1}{2}}, C_{2x}t_{\frac{1}{2}}P; C_{2y}t_{\frac{1}{2}}, C_{2y}t_{\frac{1}{2}}P
y^\hat{y}Pn^′m′aP\hat{n}'m'a (No. 62.446)C2zt12,C2zt12P;C2xt12T,C2xt12PT;C2yt12T,C2yt12PTC_{2z}t_{\frac{1}{2}}, C_{2z}t_{\frac{1}{2}}P; C_{2x}t_{\frac{1}{2}}T, C_{2x}t_{\frac{1}{2}}PT; C_{2y}t_{\frac{1}{2}}T, C_{2y}t_{\frac{1}{2}}PT
z^\hat{z}Pn^′ma′P\hat{n}'ma' (No. 62.448)C2zt12T,C2zt12PT;C2xt12T,C2xt12PT;C2yt12,C2yt12PC_{2z}t_{\frac{1}{2}}T, C_{2z}t_{\frac{1}{2}}PT; C_{2x}t_{\frac{1}{2}}T, C_{2x}t_{\frac{1}{2}}PT; C_{2y}t_{\frac{1}{2}}, C_{2y}t_{\frac{1}{2}}P

Table 1. Magnetic space groups (MSGs) and symmetries for each direction of the Néel vector N\mathbf{N} in NaOsO3_3. All directions have in common the identity II and inversion PP symmetries (not given here). No additional time reversal symmetry operation TT appears in N∥x^N\parallel\hat{x}, whereas this symmetry is involved for the other two directions.

To investigate this WM, GGA+SOC+UU calculations were carried out. Our results are given in tbl:2. The spin moments of the Os ions are 0.019z^ μB0.019\hat{z}\,\mu_B for N∥y^\mathbf{N}\parallel\hat{y} and −0.018y^ μB-0.018\hat{y}\,\mu_B for N∥z^\mathbf{N}\parallel\hat{z}, consistent with the symmetry analysis of the magnetic space group. Considering the total orbital moment of −0.016 μB-0.016\,\mu_B, the total moment is −0.034y^ μB-0.034\hat{y}\,\mu_B/f.u. for N∥z^\mathbf{N}\parallel\hat{z}. This magnitude is very similar to the observed value of 0.005 μB0.005\,\mu_B [24]. In fact, such a small value is somewhat sensitive to calculational methods (see below).

To follow this in more detail, from Wannierization we obtained a symmetrized Wannier Hamiltonian. From this Hamiltonian, the spin moments of Os ions were calculated using the TB2J code [58]. For the easy axis N∥z^\mathbf{N}\parallel\hat{z}, the calculated local spin moments of the two Os ions are m1,2S=(±0.0521,−0.0182,∓0.9882)μB\mathbf{m}^{S}_{1,2}=(\pm0.0521,-0.0182,\mp0.9882)\mu_B, leading to mS=m1S+m2S=−0.018y^ μB\mathbf{m}^{S}=\mathbf{m}^{S}_{1}+\mathbf{m}^{S}_{2}=-0.018\hat{y}\,\mu_B per Os ion, consistent with the value obtained above. The local spin moment vectors m1,2S\mathbf{m}^{S}_{1,2} indicate a spin canting angle ϕ=cos⁡−1∣m1S∣∣mS∣≈3.2∘\phi=\cos^{-1}\frac{|\mathbf{m}^{S}_{1}|}{|\mathbf{m}^{S}|}\approx3.2^\circ from the c^\hat{c}-axis, not far from the rotation angle of the O–Os–O axis about the axis, similar to the value of 5.6∘5.6^\circ obtained from the non-collinear calculations [34]. In this system, the DMI is allowed due to broken inversion symmetry by the distortions of oxygen ions involved in the magnetic interaction. This is similar to the AM La2_2CuO4_4 [65], though in both cases the magnetic space groups possess inversion symmetry. Recently, Jo and colleagues suggested an intrinsic origin of WM in AM [66]. The anisotropic local symmetry of AM gives rise to anisotropic orbital moments, which in turn lead to an anisotropic gg-tensor. That is, the total moment ∥m∥∝Δg≈L∥−L⊥S0\lVert\mathbf{m}\rVert\propto\Delta g \approx\frac{L_{\parallel}-L_{\perp}}{S_{0}} for the two principle axes, where S0S_{0} is the magnitude of spin. Thus, WM can emerge in AM, even in the cases where the DMI is forbidden, as applied to the case of the AM NiF2_2 where the orbital moment is approximately ten times larger than the spin moment [66]. However, as shown in Table II, the total spin and orbital moments in NaOsO3_3 are nearly identical, in contrast to NiF2_2. In the OsO6_6 octahedron, there are three different Os-O bond lengths [33], but differences between them are only thousandths of Angströms, resulting in L∥≈L⊥L_{\parallel} \approx L_{\perp}. So, the contribution of the anisotropic gg-tensor in NaOsO3_3 is minimal, since NaOsO3_3 is close to the ideal cubic perovskite and insulating. As a result, the WM in NaOsO3_3 primarily arises from spin canting induced by the DMI, as will be shown below.

Before closing this subsection, we mention that atomic decomposition of the magnetic moments is approximate, with different methods giving somewhat different values and perhaps implications. As mentioned above, we have calculated the Os1 (say, with spin-up) moment by two methods (Na gives up its electron, and O ions lie between antiparallel Os moments and will have negligible induced moment). The moments on the other Os sites are equal in magnitude but with different directions, according to the space group. For Os1, VASP, integrating the spin density within a sphere of radius 1.413 Å, gives the moment (−0.058,−0.018,1.139)μB(-0.058, -0.018, 1.139)\mu_{\mathrm{B}}. From the Os-centered Wannier functions, which employ no sphere but include inter-atomic effects, the moment is (−0.052,−0.018,0.988)μB(-0.052, -0.018, 0.988)\mu_{\mathrm{B}}. The actual spin densities would differ by very little; the differences arise only due to the (subjective) assignment to atoms.

Magnonic properties

The magnetic exchange parameters JijJ_{ij} for the the classical spin Hamiltonian H=−∑i,jJijei⋅ej\mathcal{H}=-\sum_{i,j}J_{ij}\mathbf{e}_{i}\cdot\mathbf{e}_{j} are calculated as implemented in the TB2J code [58]. Here, ei=Si/∣Si∣\mathbf{e}_{i}=\mathbf{S}_{i}/\lvert\mathbf{S}_{i}\rvert is the normalized spin vector. For simplicity, JijJ_{ij} is denoted below by JjJ_{j} for a fixed iith Os ion.

Our calculated parameters with respect to the Os-Os distance are given in Figure 3. Due to the orthorhombic symmetry (12a≠12b≠12c)(\frac{1}{\sqrt{2}}a \ne\frac{1}{2}b \ne\frac{1}{\sqrt{2}}c), the two nearest neighbor (NN) interactions along the (101) and (010) directions differ only slightly, as shown in the top panel of Figure 3. These inter-sublattice interactions at dij≈3.8d_{ij} \approx3.8 Å lead to negative values of J1(≈J1′)=−4.6(−4.1)J_{1}(\approx J_{1}^{\prime})=-4.6(-4.1) meV, indicating that AFM order in NaOsO3_3 is induced by the NN exchange interactions. Around dij≈5.35d_{ij}\approx5.35 Å, four next nearest neighbor (NNN) interactions between

Magnetic exchange parameters and crystal structure of NaOsO3

Figure 3. Top: magnetic exchange parameters JijJ_{ij} versus the Os-Os distance dijd_{ij}, obtained for the GGA+SOC(N∥z^∥\lVert\hat{z}\rVert)+UU at Ueff=0.5U_{\mathrm{eff}}=0.5 eV. The Insets show (left) the two NN interactions separated by a tiny amount and (right) the four inequivalent NNN values. Note the difference in two (J2J_{2} and J2′J_{2}^{\prime}) of the NNN values (filled circles by the red color) at dij≈5.358d_{ij}\approx5.358 Å. Bottom: origin of the inequivalent J2J_{2} and J2′J_{2}^{\prime} depicted in the G-type AFM cell. In the cell, the large and small spheres indicate Os (red: spin-up, blue: spin-down) and O ions, respectively. The Na ions are not shown, for easier visualization.

intra-sublattices along the (100), (001), (111), and (1ˉ1ˉ1)(\bar{1}\bar{1}1) directions appear in the range of 0.5 meV – 0.9 meV, indicating spin-alignment NN coupling that augments the NN antiferromagnetic coupling. Higher order interactions are much smaller and affect only details of the magnon dispersion.

An unusual feature appears in two of the NNN interactions (denoted by J2J_{2} and J2′J_{2}^{\prime}) along the (111) and (1ˉ1ˉ1)(\bar{1}\bar{1}1) directions at dij=12a2+b2+c2≈5.358d_{ij}=\frac{1}{2}\sqrt{a^{2}+b^{2}+c^{2}}\approx5.358 Å, as shown in the top panel of Figure 3. Along the paths of the two intra-sublattice interactions J2J_{2} and J2′J_{2}^{\prime} the environments of oxygen ions are inequivalent due to the crystallographic anisotropy of the AM, leading to a difference in the J2J_{2} and J2′J_{2}^{\prime} values by about 0.2 meV. This is depicted in the bottom panel of Figure 3. This difference is small, about a fifth of that of RuO2_2 in magnitude [14], possibly because NaOsO3_3 is insulating. However, the difference is sufficient to lead to a unique feature of AMs arising in the magnon spectrum.

To calculate the magnon spectrum using the UPPASD code [60], the exchange integrals are considered up to the Os-Os distance dij=6d_{ij}=6 Å to include the J2J_2 and J2′J'_2 values. Figure 4 shows the magnon spectrum, calculated without including a magnetic anisotropy term that shifts the Γ\Gamma-point value to the experimental spin gap of ∼59\sim59 meV. There are four branches, two acoustic and two optic, since this unit cell contains two pairs of magnetic sublattices. Except for the spin gap, our calculated magnon curves are close to the measured values [26] that are depicted by symbols of the red squares in Figure 4. Note especially that our maximum energy is very near the measured maximum (at the XX point).

Magnon spectrum

Figure 4. Magnon spectrum for the GGA+SOC(N∥z^N\parallel\hat{z})+U at Ueff=0.5U_{\mathrm{eff}}=0.5 eV. Here, the exchange integrals JijJ_{ij} are included up to the Os-Os distance dij=6d_{ij}=6 Å. The AM character reflects in the bifurcation in the acoustic and optic modes along the R−ΓR-\Gamma line. The (red) squares denote the experimental values adapted from Ref. [26]. Inset: Blowup spectrum in the energy range above 130 meV along the R(111)−Γ−R′(1ˉ1ˉ1ˉ)R(111)-\Gamma-R'(\bar{1}\bar{1}\bar{1}) line where the bifurcations due to opposite chirality of the eigenvectors are clearly visible.

What is specific to NaOsO3_3 is that the difference in the J2J_2 and J2′J'_2 values is reflected on the bifurcation of the otherwise degenerate modes in both the acoustic and (more evident) optical branches along the R−ΓR-\Gamma line. As shown in the Inset of Figure 4 the spectrum along the R−Γ−R′R-\Gamma-R' line is split due to opposite chirality of the eigenvectors, due to a symmetry constraint similar to the one discussed in § 3 B. This has been anticipated as a promising feature for the application of magnonics [14]. Interestingly, the splittings are clearly distinguishable in both the acoustic and optical modes in the THz region, whereas altermagnetic RuO2_2 and MnTe have the splittings only in the acoustic mode [14, 15]. The largest splitting in NaOsO3_3 is about 5 meV in the optic mode around 150 meV (about 35 THz). Note that the bifurcation along the R−T−ZR-T-Z line is a trivial splitting between the optic modes.

Dzyaloshinskii-Moriya interaction

The two-ion Hamiltonian for the DMI term is given by

HDMI=Dij⋅[ei×ej],(1)\mathcal{H}_{DMI}=\mathbf{D}_{ij}\cdot[\mathbf{e}_i\times\mathbf{e}_j], \tag*{(1)}

where Dij=∥Dij∥\mathbf{D}_{ij}=\lVert\mathbf{D}_{ij}\rVert is the parameter representing strength of the DMI. (As above for JijJ_{ij}, DijD_{ij} will be denoted by DjD_j for a fixed ii-spin, as in JjJ_j.) We calculated DijD_{ij} from the total energy variation where the spin rotation is treated in linear response, implemented in the TB2J package [58]. In NaOsO3_3, our calculations show that the NN DijD_{ij} constants are dominant, with negligible longer range values. (For longer range values, see the SM [22].) The NN values are, for the two bent bonds from octahedron rotation and tilting, D1=(0.3753,−0.0722,0)\mathbf{D}_1=(0.3753,-0.0722,0) along the (101)(101) direction and D2=(0.1903,0,−0.2091)\mathbf{D}_2=(0.1903,0,-0.2091) along the (010)(010) direction, in units of meV. So, D1=0.38D_1=0.38 meV and D2=0.28D_2=0.28 meV, about 10% of the corresponding exchange parameters. Using a formula due to Jackeli and Khaliullin [67] for the canting angle ϕ=12tan⁡−1(DjJj)≈2.3∘\phi=\frac{1}{2}\tan^{-1}\left(\frac{D_j}{J_j}\right)\approx2.3^\circ, close to the value obtained from our calculated spin moment vectors in § 3 C.

Our calculated values of DijD_{ij} are smaller (by several tenths of meV) than those reported by Ntallis et al. [36]. The differences may be due to choices of computational methods. For the DFT part, Ntallis et al. applied a fully relativistic, full potential linear muffin-tin method implemented in the RSPT code [68]. Thereafter, different codes were used to obtain the various constants and the spin wave spectrum. Although differences among various approaches are common, there are two additional possible reasons for the relatively small discrepancy in this case.

Spin-orbit interactions were treated differently: at the first level, we used a SOC approach originally implemented in the linear augmented plane method [69], whereas Ntallis et al. adopted a fully relativistic scheme. While these two methods typically yield quite similar results for 5d elemental solids [69], subtle differences can still arise. Second, the DFT+U schemes employed in the two studies may have been different. Different choices of this functional can result in somewhat different results [70, 71], however the nearly (locally) cubic spin density of Os would show less difference than if the moment had low symmetry spin density. The larger values reported by Ntallis et al. led to an overestimation of the top magnon frequencies [36], which occur at the zone boundary. Our results show good agreement for the maximum of the spin wave spectrum, note specifically the XX and YY points. Whatever the source of the differences, it is satisfying that the ratios of D1J1\frac{D_1}{J_1} obtained both computational methods are similar.

Anomalous Hall effect

In an AM material, separation of majority and minority characters of bands in some regions of BZ can lead to the anomalous Hall effect [18], as observed in ferromagnetic metals. As discussed in § 3 C, by the symmetry of the magnetic space group the Hall vector h\mathbf{h} connecting the applied electric field to the Hall current density is allowed only along the z^\hat{z} and y^\hat{y} directions for the case of N∥y^\mathbf{N}\parallel\hat{y} and z^\hat{z}, respectively, whereas h\mathbf{h} is zero for N∥x^\mathbf{N}\parallel\hat{x}. Thus for our case of N∥z^\mathbf{N}\parallel\hat{z} nonzero anomalous Hall effect exists only for σzx\sigma_{zx}.

The AHC, calculated using GGA+SOC+UU for the easy axis N∥z^\mathbf{N}\parallel\hat{z}, is given in Figure 5 versus band filling in the range −1-1 eV to 1 eV. Several peaks with a large magnitude appear, with a maximum value of 450 S/cm at 0.5 eV above EFE_F, implying electron doping on the Na site could achieve this value. Below EFE_F, peaks with magnitude of 150–200 S/cm at −0.06-0.06 eV and −0.12-0.12 eV could be reached by doping of 0.11 holes and 0.29 holes per f.u., estimated by the rigid band approximation, possibly achieved by Na vacancies. These features remain nearly unchanged at Ueff=1U_{\mathrm{eff}}=1 eV, while the peaks below EFE_F are somewhat enhanced at Ueff=2U_{\mathrm{eff}}=2 eV, as given in SM [22]. This approximation would provide a reasonable estimate, since there is no Na character near EFE_F (see SM [22]). These values of AHC are much larger than those calculated for MnTe [20]. Note that the ab initio AHC values obtained from the Berry curvature provide a reasonable tendency, although such estimates can show quantitative disagreement with experimental values [19].

Image not extracted

Figure 5. Anomalous Hall conductivity σzx\sigma_{zx} for GGA+SOC(N∥z^\mathbf{N}\parallel\hat{z})+UU at Ueff=0.5U_{\mathrm{eff}}=0.5 eV. For the case of N∥z^\mathbf{N}\parallel\hat{z}, σxy=σyz=0\sigma_{xy}=\sigma_{yz}=0 by the magnetic space group symmetry (for details, see text). The (blue) dashed lines indicate doping levels of 0.29 holes, 0.11 holes, and 0.62 electrons (per f.u.), from left to right in order, estimated from the rigid band approximation. The zero value in the energy gap region shaded here indicates a topologically trivial insulating state.

Discussion and Summary

Our work supports that NaOsO3_3 is an example of an unusual small gap insulating altermagnet, whose G-type AFM order and spin dynamics are dominated by first and second neighbor exchange couplings. The quite weak net magnetization arises from Dzyaloshinskii-Moriya canting due to low symmetry bonding, aided by the altermagnetic ordering. To answer one question posed in the Introduction, our DFT work supports that NaOsO3_3 lies in a crossover regime between Slater and Mott insulator. As mentioned in the Introduction, this regime of a gap opening/closing MIT is more involved than the conventional Slater and Mott pictures, although we will continue to consider here this separation.

The distortion into the gadolinium orthoferrite structure already creates an unexpected separation of valence and conduction bands. The combination of SOC with a small Hubbard UU repulsion steadies the (T=0) gap, seemingly supporting the Mott picture. These features, relying on the small gap (or not), may instead reflect a small deficiency of the GGA functional. Studies of excitations or properties upon doping, which differ between Slater and Mott systems, could clarify this crossover regime.

Our values of the spin exchange parameters led to excellent agreement with the experimental magnon dispersion in the high energy region. The resulting spin-wave spectrum provides the novel example of a chiral magnon in the optic branch. With the anisotropy term obtained elsewhere, the magnon spectrum including the chiral anomaly is obtained quantitatively. Upon doping, NaOsO3_3 should show a substantial anomalous Hall effect.

Although emphasizing that our results give support to the picture that the magnetic transition in NaOsO3_3 is intermediate between Slater- and Mott-types (neither clearly applying) and accepting the usual characterization as a continuous volume transition (which favors the Slater picture), our work has not focused unduly on experimental details of the transition. However, in the data plotted [37] by Zhou et al. in their Fig. 2 (using the data adapted from Ref. [25]), a very small, but above stated experimental uncertainty, volume change occurs at the transition. Our estimate is ΔV/V≈2×10−4\Delta V/V\approx2\times10^{-4}. This discontinuity indicates an (extremely) weakly first order transition, which would likely entail a discontinuous opening of the gap (our calculations address only T=0 electronic and magnetic structure). This feature might support the characterization of Zhou et al. as a “muted Mott transition”. Such a small discontinuity, if true, may rival the smallest volume collapse transition ever reported, and might affect the analysis in terms of critical exponents and indeed in concept of such a delicate distinction between characters of magnetic ordering and gap opening.

A weakness of the above picture, in line with muted Mott transition picture, is that the seeming (very small) discontinuity in volume is not clear in the individual structural parameters, and indeed the Slater picture has been the conclusion of earlier works [24, 25, 29, 32, 39], in spite of the need to apply a modest onsite repulsion UU to obtain a gap. The relevance of our work to this question has not been great, except to point out so far neglected complexities where experimental data become even more relevant. Shi et al. have shown [24] that both the resistivi ity and magnetic susceptibility display an abrupt increase at TMITT_{\mathrm{MIT}}, not necessarily the signature of a continuous transition involving smooth onset of magnetic order and gap opening. Given the large size of the (low temperature) ordered Os moment ∼1μB\sim1\mu_{\mathrm{B}}, and the increasing susceptibility as TT is lowered in the conducting phase above TMITT_{\mathrm{MIT}} (linear behavior of 1/χ(T)1/\chi(T) indicative of spin fluctuations [24]), and the abruptly increasing susceptibility above the transition, the transition will have spin disorder-order character, an aspect not included in either the Slater or Mott pictures.

The measured resistivity also raises related questions. A continuous opening of the gap and of magnetic order could suggest less abrupt behavior, the gap opening decreasing conductivity sharply, while magnetic order reduces spin fluctuation scattering proportional to the order parameter while leaving phonon scattering unchanged. Here the altermagnetic character of NaOsO3_3 may become relevant, because both hole and electron bands bordering the gap are spin polarized and will enable spin scattering separate from that of the magnons. In addition, the electron band broadening due to phonons and spin fluctuations will be noticeable and have its own implications, especially for gap opening.

The case of Cd2_2Os2_2O7_7, another Os5+^{5+} osmate [72], presents an example of another unusual MIT, smooth in the resistivity and Hall response. The smoothness of the transition would not allow any discontinuous volume change. The underlying magnetic behavior in this pyrochlore structure osmate is quite different from that of G-type AFM NaOsO3_3. It seems likely that there is not yet any viable theory for this intermediate region between Slater and Mott ordering in a material specific manner.

Acknowledgments

We acknowledge useful communications with Young-Joon Song for comments on perovskite altermagnets, with Hyun-Woo Lee for discussion on anisotropy of orbital moments in altermagnets, with Changyoung Kim for spin splitting in the altermagnets, with Bongjae Kim for various DFT calculations of NaOsO3_3, and with M. D. Johannes on the character of the magnetic transition. H.S.C and K.W.L were supported by National Research Foundation of Korea (NRF) Grant (RS-2024-00392493), and M.C.J. was supported by NRF of Korea Grant No. NRF2019R1A2C1009588. K.H.A. was supported by Grant No. 25-17490S of the Czech Science Foundation and No. CZ.02.01.01/00/22 008/0004594 (TERAFIT) of the Czech Ministry of Education, Youth and Sports.

End of Paper

Supplemental Material

This Supplemental Material provides additional information to provide support to the text. The information is described in the figure captions.

Image not extracted

Figure S1. Three types of collinear antiferromagnetic order in the GdFeO3_3-type (gadolinium orthoferrite) perovskite structure. Each magnetic sublattice, marked by brown ellipses, is connected by a combination of two-fold rotational C2jC_{2j} and non-primitive translation t12t_{\frac{1}{2}} magnetic symmetries: [C2∥C2yt12(0 1 0)][C_2 \Vert C_{2y}t_{\frac{1}{2}}(0\,1\,0)], [C2∥C2zt12(1 0 1)][C_2 \Vert C_{2z}t_{\frac{1}{2}}(1\,0\,1)], and [C2∥C2xt12(1 1 0)][C_2 \Vert C_{2x}t_{\frac{1}{2}}(1\,1\,0)] for the G-, C-, and A-types, respectively. A fundamental result is that these perovskites are altermagnets in each of these orderings.

Image not extracted

Figure S2. Comparison of the Wannierized bands (brown symbols) with the ab initio bands, from the GGA+SOC(Nkzz^N_{k_z}\hat{z})+U at Ueff=0.5U_{\mathrm{eff}} = 0.5 eV, showing the very good representation over a 15 eV span. All Os-dd and O-pp orbitals were used in the Wannier basis set.

GGA band structure along the S–Γ–U–T–Γ line

Figure S3. GGA band Structure along the S−Γ−U−T−ΓS-\Gamma-U-T-\Gamma line showing spin-splitting inside the Brillouin zone. Here, the spin characters are colored by red (spin-up) and blue (spin-down).

Fine scale band structures along the lines surrounding the zone boundary

Figure S4. Fine scale band structures, within the GGA level, plotted along the line surrounding the zone-boundary over the range of −1-1 eV to 11 eV. These are on kx−kyk_x-k_y planes with values kz=14,12,k_z = \frac{1}{4}, \frac{1}{2}, and 34\frac{3}{4} (units of πc\frac{\pi}{c}), in frames (a), (b), (c) respectively. The red-solid (blue-dashed) lines indicate spin-up (down) bands. Spin-splittings occur along the lines parallel to the kx=0k_x=0 line on the each plane with a fixed kzk_z. The size of the splittings is clearly visible in the right hand panels of each frame.

Magnon spectrum along the R–Γ–X–S–Y–Γ–Z–U–R–T–Z path

Figure S5. Magnon spectrum from the GGA+SOC(N∥z^\mathrm{N}\parallel\hat{z})+U at Ueff=0.5U_{\mathit{eff}}=0.5 eV, including only the NN exchange integrals. This demonstrates that chiral splitting arises from the inequivalent NNN J2J_2 and J2′J'_2, which are not considered for this plot. Note that the trivial splitting between two optic branches along the R−T−ZR-T-Z line is about 0.3 meV, invisible here.

Figure S6. Each component of the Dzyaloshinskii-Moriya interaction(DMI) vectors Dij\mathbf{D}_{ij} (meV) versus the Os-Os distance dd (Å), indicating that the NN DMIs are dominant. These are obtained from the GGA+SOC(N∥z^N\parallel\hat{z})+U calculations at Ueff=0.5U_{\mathrm{eff}}=0.5 eV.

Figure S7. Total and atom-projected densities of states from the GGA+SOC(N∥z^N\parallel\hat{z})+U at Ueff=0.5U_{\mathrm{eff}}=0.5 eV, in units of states per eV per f.u. The vertical (blue) dashed lines denote the doping levels mentioned in the caption of Figure 5.

Anomalous Hall conductivity for two values of U

Figure S8. Anomalous Hall conductivity σzx\sigma_{zx} for GGA+SOC(N∥z^N\parallel\hat{z})+U at Ueff=1U_{\mathrm{eff}}=1 (left) and 2 (right) eV. In particular, the peaks located between –0.3 eV and EFE_{\mathrm{F}} are robust against variations of UU value studied here. The shaded area denotes the zero value in the energy gap region.

Magnetic exchange parameters versus Os–Os distance

Figure S9. Magnetic exchange parameters JijJ_{ij} versus the Os-Os distance dijd_{ij} for GGA+SOC(N∥z^N\parallel\hat{z})+UU at Ueff=1U_{\mathrm{eff}}=1 (left) and 2 (right) eV. The difference in the J2J_{2} and J2′J'_{2} values, located at the vertical dashed-line, is by about 0.12 (0.07) meV at Ueff=1U_{\mathrm{eff}}=1 (2) eV, invisible here. Note that the magnitude of the NN exchange parameters increases with increasing UeffU_{\mathrm{eff}} due to the enhancement of the Os local magnetic moments: 0.989, 1.130, and 1.338 (in units of μB\mu_{\mathrm{B}}) for Ueff=0.5U_{\mathrm{eff}}=0.5 eV, 1 eV, and 2 eV, respectively. This enhancement leads to an increase in the maximum energy of the magnon spectrum, as shown below.

Enlarged magnon spectrum around the optic modes

Figure S10. Enlarged magnon spectrum around the optic modes along the R(111)−Γ−R′(1ˉ1ˉ1ˉ)R(111)-\Gamma-R'(\bar{1}\bar{1}\bar{1}) line where the splittings due to opposite chirality of the eigenvectors are clearly visible, for GGA+SOC(N∥z^N\parallel\hat{z})+UU at Ueff=1U_{\mathrm{eff}}=1 (left) and 2 (right) eV. The largest splitting in the optic modes is about 2 (1) meV at Ueff=1U_{\mathrm{eff}}=1 (2) eV. However, the maximum energy for these values of UeffU_{\mathrm{eff}} is significantly larger than that of the experimental value given in the main text.

atomOs1Os2
orbitalxyxyyzyzz2z^2xzxzx2−y2x^2-y^2xyxyyzyzz2z^2xzxzx2−y2x^2-y^2
majority0.39350.17780.0357−0.04200.11180.60240.25300.0304−0.08820.0890
majority0.17780.75740.00270.0694−0.06720.25300.5485−0.04300.0195−0.0775
majority0.03570.00270.5329−0.2769−0.07680.0304−0.04300.8406−0.0957−0.0756
majority−0.04200.0694−0.27690.6200−0.02450.08820.0195−0.09570.4162−0.2034
majority0.1118−0.0672−0.0768−0.02450.81020.0890−0.0775−0.0756−0.20340.7063
minority0.29660.04360.0211−0.02360.05430.37290.07100.0155−0.02930.0128
minority0.04360.41410.01980.0108−0.01060.07100.3379−0.01750.0148−0.0528
minority0.02110.01980.3101−0.0707−0.02250.0155−0.01750.4315−0.0134−0.0119
minority0.02360.1080−0.07070.37370.0029−0.02930.0148−0.01340.3050−0.0638
minority0.0543−0.0106−0.02250.00290.41670.0128−0.0528−0.0119−0.06380.3639

Table S1. Occupation matrices for the Os 5d5d orbitals of NaOsO3_3 in GGA. These are represented in terms of the real harmonics, implemented in the all-electron full-potential FPLO program [K. Koepernik and H. Eschrig, Phys. Rev. B 59, 1743 (1999)]. The occupation matrices provide the trace of 3.114 (1.811) for the majority (minority) orbitals, resulting in about 4.915 electrons, significantly different from the 5d35d^3 configuration of Os5+^{5+}. This difference of real charge from formal charge is a well documented feature of transition metal oxides; for Os, the real charge is near 5 electrons for even higher formal valences (7+, 6+). The unexpected differences between the occupation matrix values and what would normally be expected of the majority t2g3t_{2g}^3 occupations are because the natural axes for the OsO6_6 octahedron are rotated away from the global coordinate system in which the occupation matrices are expressed.

References

  1. [1]L. Šmejkal, J. Sinova, and T. Jungwirth, Emerging Research Landscape of Altermagnetism, Phys. Rev. X 12, 031042 (2022).
  2. [2]L. Šmejkal, J. Sinova, and T. Jungwirth, Beyond conventional ferromagnetism and antiferromagnetism: A phase with non-relativistic spin and crystal rotation symmetry, Phys. Rev. X 12, 040501 (2022).
  3. [3]For a review, see S. Shim, M. Mehraeen, J. Sklenar, S.-L. Zhang, A. Hoffmann, and N. Mason, Spin-Polarized Antiferromagnetic Metals, Annu. Rev. Condens. Matter Phys. 16, 103 (2024).
  4. [4]For a review, see C. Song, H. Bai, Z. Zhou, L. Han, H. Reichlova, J. Huguo Dil, J. Liu, X. Chen, and F. Pan, Altermagnets as a new class of functional materials, Nat. Rev. Mater. 10, 473 (2025).
  5. [5]H.-Y. Ma, M. M. Hu, N. Li, J. Liu, W. Yao, J.-F. Jia, and J. Liu, Multifunctional antiferromagnetic materials with giant piezomagnetism and noncollinear spin current, Nat. Commun. 12, 2846 (2021).
  6. [6]C. Wu, K. Sun, E. Fradkin, and S.-C. Zhang, Fermi liquid instabilities in the spin channel, Phys. Rev. B 75, 115103 (2007).
  7. [7]K.-H. Ahn, A. Hariki, K.-W. Lee, and J. Kuneš, Antiferromagnetism in RuO₂ as d-wave Pomeranchuk instability, Phys. Rev. B 99, 184432 (2019).
  8. [8]S. Hayami, Y. Yanagi, and H. Kusunose, Momentum-Dependent Spin Splitting by Collinear Antiferromagnetic Ordering, J. Phys. Soc. Jpn. 88, 123702 (2019).
  9. [9]L.-D. Yuan, Z. Wang, J.-W. Luo, E. I. Rashba, and A. Zunger, Giant momentum-dependent spin splitting in centrosymmetric low-Z antiferromagnets, Phys. Rev. B 102, 014422 (2020).
  10. [10]L. Šmejkal, R. González-Hernández, T. Jungwirth, and J. Sinova, Spin Wave Crystal time-reversal symmetry breaking and spontaneous Hall effect in collinear antiferromagnets, Sci. Adv. 6, eaaz8809 (2020).
  11. [11]I. I. Mazin, K. Koepernik, M. D. Johannes, and L. Šmejkal, Prediction of unconventional magnetism in doped FeSb₂, Proc. Natl. Acad. Sci. U.S.A. 118, e2108924118 (2021).
  12. [12]J. Krempaský, L. Šmejkal, S. W. D’Souza, M. Hajlaoui, G. Springholz, K. Uhlířová, F. Alarab, P. C. Constantinou, V. Strocov, D. Usanov, W. R. Pudelko, R. González-Hernández, P. Wadley, R. Sinova, D. Krieger, J. Minár, J. H. Dil, and T. Jungwirth, Altermagnetic lifting of Kramers spin degeneracy, Nature 626, 517 (2024).
  13. [13]S.-W. Cheong and F.-T. Huang, Altermagnetism with non-collinear spins, npj Quantum Mater. 9, 13 (2024).
  14. [14]L. Šmejkal, A. Marmodoro, K.-H. Ahn, R. González-Hernández, I. Turek, S. Mankovsky, H. Ebert, S. W. D’Souza, O. Šipr, J. Sinova, and T. Jungwirth, Chiral Magnons in Altermagnetic RuO₂, Phys. Rev. Lett. 131, 256703 (2023).
  15. [15]Z. Liu, M. Ozeki, S. Asai, S. Itoh, and T. Masuda, Chiral split magnon in Altermagnetic MnTe, Phys. Rev. Lett. 133, 156702 (2024).
  16. [16]X. Zhou, W. Feng, R.-W. Zhang, L. Šmejkal, J. Sinova, Y. Mokrousov, and Y. Yao, Crystal thermal transport in altermagnetic RuO₂, Phys. Rev. Lett. 132, 056701 (2024).
  17. [17]R. Hoyer, R. Jaeschke-Ubiergo, K.-H. Ahn, L. Šmejkal, and A. Mook, Spontaneous crystal thermal Hall effect in insulating altermagnets, Phys. Rev. B 111, L020412 (2025).
  18. [18]L. Šmejkal, A. H. MacDonald, J. Sinova, S. Nakatsuji, and T. Jungwirth, Anomalous Hall antiferromagnets, Nat. Rev. Matter 7, 482 (2022).
  19. [19]Z. Feng, X. Zhou, L. Šmejkal, L. Wu, Z. Zhu, H. Guo, R. González-Hernández, X. Wang, H. Yan, P. Qin, X. Zhang, H. Wu, H. Chen, Z. Meng, L. Liu, Z. Xia, J. Sinova, T. Jungwirth, and Z. Liu, An anomalous Hall effect in altermagnetic ruthenium dioxide, Nat. Electron. 5, 735 (2022).DOI
  20. [20]R.D. Gonzalez Betancourt, J. Zubáč, R. Gonzalez-Hernandez, K. Geishendorf, Z. Šobň, G. Springholz, K. Olejník, L. Šmejkal, J. Sinova, T. Jungwirth, S.T.B. Goennenwein, A. Thomas, H. Reichlová, J. Železný, and D. Kriegner, Spontaneous Anomalous Hall Effect Arising from an Unconventional Compensated Magnetic Phase in a Semiconductor, Phys. Rev. Lett. 130, 036702 (2023).DOI
  21. [21]H. Reichlova, R. L. Seeger, R. González-Hernández, I. Kounta, R. Schlitz, D. Kriegner, P. Ritzinger, M. Lammel, M. Leiviskä, A. B. Hellenes, K. Olejník, V. Petříček, P. Doležal, L. Horak, E. Schmoranzerova, A. Badura, S. Bertaina, A. Thomas, V. Baltz, L. Michez, J. Sinova, S. T. B. Goennenwein, T. Jungwirth, and L. Šmejkal, Observation of a spontaneous anomalous Hall response in the Mn5Si3 d-wave altermagnet candidate, Nat. Commun. 15, 4961 (2024).DOI
  22. [22]See Supplemental Material at [URL will be inserted by publisher] for figures of collinear antiferromagnetic orders, Wannier bands, fine band structures on kx − ky planes, magnon spectrum including only the NN exchange integrals, DMI vectors, and variations of certain results with varying Ueff.
  23. [23]For a review, see M. Nake, Y. Motome, and H. Seo, Altermagnetic perovskites, npj Spintronics 3, 1 (2025).
  24. [24]Y. G. Shi, Y. F. Guo, S. Yu, M. Arai, A. A. Belik, A. Sato, K. Yamaura, E. Takayama-Muromachi, H. F. Tian, H. X. Yang, J. Q. Li, T. Varga, J. F. Mitchell, and S. Okamoto, Continuous metal-insulator transition of the antiferromagnetic perovskite NaOsO3, Phys. Rev. B 80, 161104(R) (2009).DOI
  25. [25]S. Calder, V. O. Garlea, D. F. McMorrow, M. D. Lumsden, M. B. Stone, J. C. Lang, J.-W. Kim, J. A. Schlueter, Y. G. Shi, K. Yamaura, Y. S. Sun, Y. Tsujimoto, and A. D. Christianson, Magnetically Driven Metal-Insulator Transition in NaOsO3, Phys. Rev. Lett. 108, 257209 (2012).
  26. [26]S. Calder, J. G. Vale, N. Bogdanov, C. Donnerer, D. Pincini, M. Moretti Sala, X. Liu, M. H. Upton, D. Casa, Y. G. Shi, Y. Tsujimoto, K. Yamaura, J. P. Hill, J. van den Brink, D. F. McMorrow, and A. D. Christianson, Strongly gapped spin-wave excitation in the insulating phase of NaOsO3, Phys. Rev. B 95, 020413(R) (2017).DOI
  27. [27]J.G. Vale, S. Calder, C. Donnerer, D. Pincini, Y.G. Shi, Y. Tsujimoto, K. Yamaura, M. Moretti Sala, J. van den Brink, A.D. Christianson, and D.F. McMorrow, Evolution of the Magnetic Excitations in NaOsO3 through its Metal-Insulator Transition, Phys. Rev. Lett. 120, 227203 (2018).
  28. [28]J.G. Vale, S. Calder, C. Donnerer, D. Pincini, Y.G. Shi, Y. Tsujimoto, K. Yamaura, M. Moretti Sala, J. van den Brink, A.D. Christianson, and D.F. McMorrow, Crossover from itinerant to localized magnetic excitations through the metal-insulator transition in NaOsO3, Phys. Rev. B 97, 184429 (2018).DOI
  29. [29]N. Gurung, N. Leo, S. P. Collins, G. Nisbet, G. Smolentsев, M. García-Fernández, K. Yamaura, L. J. Heyderman, U. Staub, Y. Joly, D. D. Khayalvin, S. W. Loveysey, and V. Scagnoli, Direct observation of electron density reconstruction at the metal-insulator transition in NaOsO3, Phys. Rev. B 98, 115116 (2018).DOI
  30. [30]I. L. Vecchio, A. Perucci, P. Di Pietro, O. Limaj, U. Schade, Y. Sun, M. Arai, K. Yamaura, and S. Lupi, Infrared evidence of a Slater metal-insulator transition in NaOsO3, Sci. Rep. 3, 2990 (2013).DOI
  31. [31]R. Sereika, P. Liu, B. Kim, S. Kim, J. Zhang, B. Chen, K. Yamaura, C. Park, C. Franchini, Y. Ding, and H.-K. Mao, Aberrant electronic and structural alterations in pressure tuned perovskite NaOsO3, npj Quantum Mater. 5, 66 (2020).DOI
  32. [32]Y. Du, X. Wan, L. Sheng, J. Dong, and S. Y. Savrasov, Electronic structure and magnetic properties of NaOsO3, Phys. Rev. B 85, 174424 (2012).
  33. [33]M.-C. Jung, Y.-J. Song, K.-W. Lee, and W. E. Pickett, Structural and correlation effects in the itinerant insulating antiferromagnetic perovskite NaOsO3, Phys. Rev. B 87, 115119 (2013).
  34. [34]V. N. Antonov, D. A. Kukusta, and L. V. Bekenov, Electronic structure and resonant inelastic x-ray scattering in osmates. I. Perovskite NaOsO3, Phys. Rev. B 105, 155144 (2022).
  35. [35]S. Middey, S. Debnath, P. Mahadevan, and D. D. Sarma, NaOsO3: A high Neel temperature 5d oxide, Phys. Rev. B 89, 134416 (2014).
  36. [36]N. Ntallis, V. Borisov, Y. O. Kvashnin, D. Thonig, E. Sjöqvist, A. Bergman, A. Delin, O. Eriksson, and M. Pereiro, Connection between magnetic interactions and the spin-wave gap of the insulating phase of NaOsO3, Phys. Rev. B 104, 134433 (2021).
  37. [37]J.-S. Zhou, X.-Y. Li, and M. D. Johannes, Evidence of electronic structural change at the metal-insulator transition in the perovskite NaOsO3, Phys. Rev. B 108, 195115 (2023).
  38. [38]B. Kim, P. Liu, Z. Ergönenc, A. Toschi, S. Khmelevskyi, and C. Franchini, Lifshitz transition driven by spin fluctuations and spin-orbit renormalization in NaOsO3, Phys. Rev. B 94, 241113(R) (2016).
  39. [39]S. Mohapatra, C. Bhandari, S. Satpathy, and A. Singh, Effect of structural distortion on the electronic band structure of NaOsO3 studied within density functional theory and a three-orbital model, Phys. Rev. B 97, 155154 (2018).
  40. [40]C. Bhandari and S. Satpathy, Effect of epitaxial strain on the optical properties of NaOsO3, J. Phys. Chem. Solids 128, 265 (2019).
  41. [41]J. C. Slater, Magnetic Effects and the Hartree-Fock Equation, Phys. Rev. 82, 538 (1951).
  42. [42]B. I. Halperin and T. M. Rice, The Excitonic State at the Semiconductor-Semimetal Transition, Solid State Phys. 21, 115 (1968).
  43. [43]B. I. Halperin and T. M. Rice, Possible Anomalies at a Semimetal-Semiconductor Transition, Rev. Mod. Phys. 40, 755 (1968).
  44. [44]P. B. Allen and M. Cardona, Temperature Dependence of the Direct Gap of Si and Ge, Phys. Rev. B 27, 4760 (1983).
  45. [45]M. Imada, A. Fujimori, and Y. Tokura, Metal insulator transitions, Rev. Mod. Phys. 70, 1039 (1998). P. B. Allen and V. Heine, Theory of Temperature Dependence of Electronic Band Structures, J. Phys. C 9, 2305 (1976).
  46. [47]A. P. Ramirez, Geometrical frustration, Handbook of Magnetic Materials 13, 423 (2001).
  47. [48]J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized gradient approximation made simple, Phys. Rev. Lett. 77, 3865 (1996).
  48. [49]G. Kresse and J. Hafner, Ab initio molecular dynamics for liquid metals, Phys. Rev. B 47, 558 (1993).
  49. [50]G. Kresse, J. Furthmüller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54, 11169 (1996).
  50. [51]H. J. Monkhorst and J. D. Pack, Special points for Brillouin-zone integrations, Phys. Rev. B 13, 5188 (1976).
  51. [52]S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys, and A. P. Sutton, Electron-energy-loss spectra and the structural stability of nickel oxide: An LSDA+U study, Phys. Rev. B 57, 1505 (1998). This correlation functional uses only the difference U−J in terms of the spherically averaged values of the interaction matrices Uα,β and Jα,β, α, β are d orbital indices. Since the Os moments, hence spin densities, have nearly (locally) cubic symmetry, this functional should be applicable.
  52. [53]N. Marzari and D. Vanderbilt, Maximally localized generalized Wannier functions for composite energy bands, Phys. Rev. B 56, 12847 (1997).
  53. [54]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. Lihm, D. Marchand, A. Marrazzo, Y. Mokrousov, J. I. Mustafa, Y. Nohara, Y. Nomura, I. Paulatto, S. Poncé, T. Ponweiser, J. Qiao, F. Thölе, S. S. Tirkkin, M. Wierzbowska, N. Marzari, D. Vanderbilt, I. Souza, A. A. Mostofi, and J. R. Yates, Wannier90 as a community code: new features and applications, J. Phys.: Condens. Matter 32, 165902 (2020).
  54. [55]G.-X. Zhi, C.-C. Xu, S.-Q. Wu, F.-L. Ning and C. Cao, WannSym: A symmetry analysis code for Wannier orbitals, Comput. Phys. Commun. 271, 108196 (2022).
  55. [56]I. A. Dzyaloshinskii, A Thermodynamic Theory of Weak Ferromagnetism, J. Phys. Chem. Solids 4, 241 (1958).
  56. [57]T. Moriya, Anisotropic Superexchange Interaction and Weak Ferromagnetism, Phys. Rev. 120, 91 (1960).
  57. [58]X. He, N. Helbig, M. J. Verstraete, and E. Bousquet, TB2J: a python package for computing magnetic interaction parameters, Comput. Phys. Commun. 264, 107938 (2021).
  58. [59]A. I. Liechtenstein, M. Katsnelson, V. Antropov, and V. Gubanov, Local spin density functional approach to the theory of exchange interactions in ferromagnetic metals and alloys, J. Magn. Magn. Mater. 67, 65 (1987).
  59. [60]B. Skubic, J. Hellsvik, L. Nordström, and O. Eriksson, A method for atomistic spin dynamics simulations: Implementation and examples, J. Phys.: Condens. Matter 20, 315203 (2008).
  60. [61]Q. Wu, S. Zhang, H.-F. Song, M. Troyer, and A. A. Soluyanov, WannierTools: An open-source software package for novel topological materials, Comput. Phys. Commun. 224 405 (2018).
  61. [62]MnSi, Phys. Rev. B 70, 075114 (2004).
  62. [63]Z. Xiao, J. Zhao, Y. Li, R. Shindou, and Z.-D. Song, Spin Space Groups: Full Classification and Applications, Phys. Rev. X 14, 031037 (2024).
  63. [64]X. Chen, J. Ren, Y. Zhu, Y. Yu, A. Zhang, P. Liu, J. Li, Y. Liu, C. Li, and Q. Liu, Enumeration and Representation Theory of Spin Space Groups, Phys. Rev. X 14, 031038 (2024).
  64. [65]S-W. Cheong, J. D. Thompson, and Z. Fisk, Metamagnetism in La2CuO4, Phys. Rev. B 39, 4395 (1989).
  65. [66]D. Jo, D. Go, Y. Mokrousov, P. M. Oppeneer, S.-W. Cheong, and H.-W. Lee, Weak Ferromagnetism in Al-ternagnets from Alternating g-Tensor Anisotropy, Phys. Rev. Lett. 134, 196703 (2025).
  66. [67]G. Jackeli and G. Khaliullin, Mott Insulators in the Strong Spin-Orbit Coupling Limit: From Heisenberg to a Quantum Compass and Kitaev Models, Phys. Rev. Lett. 102, 017205 (2009).
  67. [68]J. Wills, M. Alouani, P. Andersson, A. Delin, O. Eriksson, and O. Grechnyev, Full-Potential Electronic Structure Method, Springer Series in Solid-State Sciences (Springer-Verlag, Berlin, 2010), Vol. 167.
  68. [69]A. H. MacDonald, W. E. Pickett, and D. D. Koelling, A linearized relativistic augmented-plane-wave method utilising approximate pure spin basis functions, J. Phys. C: Solid State Phys. 13, 2675 (1980).
  69. [70]K.-W. Lee and W. E. Pickett, Correlation Effects in the High Formal Oxidation-state Compound Sr2CoO4, Phys. Rev. B 73, 174428 (2006).
  70. [71]E. R. Ylvisaker, K. Koepernik, and W. E. Pickett, Anisotropy and magnetism in the LSDA+U method, Phys. Rev. B 79, 035103 (2009).
  71. [72]Y. Feng, Y. Wang, D. M. Silevitch, S. E. Cooper, D. Mandrus, P. A. Lee, and T. F. Rosenbaum, A continuous metal-insulator transition driven by spin correlations, Nat. Commun. 12, 2779 (2021).

Paper details

vlnc:W_mshuc28t
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