This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Bottom-up design of spin-split and reshaped electronic band structures
in spin-orbit-coupling free antiferromagnets:
Procedure on the basis of augmented multipoles

Satoru Hayami1, Yuki Yanagi2, and Hiroaki Kusunose3 1Department of Applied Physics, The University of Tokyo, Bunkyo, Tokyo 113-8656, Japan
2Center for Computational Materials Science, Institute for Materials Research, Tohoku University, Sendai, Miyagi, 950-8577, Japan
3Department of Physics, Meiji University, Kawasaki 214-8571, Japan
Abstract

We propose an efficient microscopic design procedure of electronic band structures having intrinsic spin and momentum dependences in spin-orbit-coupling free antiferromagnets. Our bottom-up design approach to creating desired spin-split and reshaped electronic band structures could result in further findings of practical spin-orbit-coupling free materials exhibiting a giant spin-dependent and/or nonreciprocal transport, magneto-electric and elastic responses, and so on, as a consequence of such band structures. We establish a systematic guideline to construct symmetric/antisymmetric spin-split and antisymmetrically deformed spin-independent band structures in spin-orbit-coupling free systems by using two polar multipole degrees of freedom, i.e., electric and magnetic toroidal multipoles. The two polar multipoles constitute a complete set and describe arbitrary degrees of freedom in the hopping Hamiltonian, whose onsite and offsite degrees of freedom in a cluster are described as the so-called cluster and bond multipoles, respectively, and another degree of freedom connecting between clusters is expressed as momentum multipoles. By using these multipole descriptions, we elucidate simple microscopic conditions to realize intrinsic band deformations in magnetically ordered states: The symmetric spin splitting is realized in collinear magnets when cluster and bond multipoles contain the same symmetry of multipoles. The antisymmetric spin splitting occurs in noncollinear antiferromagnets when a bond-type magnetic toroidal multipole is present. Furthermore, the antisymmetric band deformation with spin degeneracy is realized in noncoplanar antiferromagnets. We exemplify three lattice systems formed by a triangle unit, triangular, kagome, and breathing kagome structures, in order to demonstrate the band deformations under the magnetic ordering. On the basis of the proposed procedure, we list up various candidate materials showing intrinsic band deformations in accordance with MAGNDATA, magnetic structures database.

preprint: APS/123-QED

I Introduction

The electronic band structures in solids play an important role in determining fundamental physical properties. In general, the electronic band dispersions εσ(𝒌)\varepsilon_{\sigma}(\bm{k}), which are characterized by the wave vector 𝒌\bm{k} and the spin σ\sigma, are classified according to the presence and absence of space-time inversion symmetry, where the spatial inversion operation, 𝒫\mathcal{P}, transforms εσ(𝒌)\varepsilon_{\sigma}(\bm{k}) as 𝒫εσ(𝒌)=εσ(𝒌)\mathcal{P}\varepsilon_{\sigma}(\bm{k})=\varepsilon_{\sigma}(-\bm{k}) and the time-reversal operation, 𝒯\mathcal{T}, transforms εσ(𝒌)\varepsilon_{\sigma}(\bm{k}) as 𝒯εσ(𝒌)=εσ(𝒌)\mathcal{T}\varepsilon_{\sigma}(\bm{k})=\varepsilon_{-\sigma}(-\bm{k}). In the presence of 𝒫\mathcal{P} and 𝒯\mathcal{T}, i.e., in the centrosymmetric paramagnetic state, the system has a twofold degeneracy, εσ(𝒌)=εσ(𝒌)=εσ(𝒌)\varepsilon_{\sigma}(\bm{k})=\varepsilon_{\sigma}(-\bm{k})=\varepsilon_{-\sigma}(\bm{k}) in the entire Brillouin zone. The spin-split band structure is realized once either 𝒫\mathcal{P} or 𝒯\mathcal{T} is broken: The breaking of 𝒯\mathcal{T} (𝒫\mathcal{P}) results in the (anti)symmetric spin splitting in momentum space, provided that the spin and momentum degrees of freedom are coupled with each other. This is refereed as the spin-momentum locking Rashba (1960); Sinova et al. (2004); Dresselhaus et al. (2008).

One of the microscopic key ingredients to connect the spin degree of freedom with kinetic motion of electrons is the spin-orbit coupling (SOC). For example, the relatively large SOC brings about the large antisymmetric spin splitting in the noncentrosymmetric materials, such as a polar semiconductor BiTeI Rashba (1960); Ishizaka et al. (2011); Bahramy et al. (2011) and monolayer transition-metal dichalcogenides, MX2MX_{2} (M=M= Mo, W and X=X= S, Se) Zhu et al. (2011); Wang et al. (2012); Ugeda et al. (2014); Kormányos et al. (2014). Although materials with the large SOC give rise to intriguing physical phenomena, such as the magnetoelectric effect Kimura et al. (2003); Fiebig (2005); Khomskii (2009); Furukawa et al. (2017); Saito et al. (2018), spin Hall effect Hirsch (1999); Sinova et al. (2004); Bernevig et al. (2006); Liu et al. (2012); Sinova et al. (2015), and nonreciprocal optics Kézsmárki et al. (2014); Toyoda et al. (2016); Morimoto and Nagaosa (2016); Ideue et al. (2017); Tokura and Nagaosa (2018); Aoki et al. (2019), it is usually nontrivial to control them microscopically, since the SOC is predominant in the complicated atomic orbitals and chemical composition. It prevents us engineering large spin splittings by tuning the built-in SOC of the materials constituted of moderately heavier elements.

On the contrary, recent studies indicate that even without relying on the SOC, a change of electronic state by magnetic orderings leads to a similar spin splitting depending on the crystal momentum Ahn et al. (2019); Naka et al. (2019); Hayami et al. (2019a, 2020a, 2020b); Yuan et al. (2020). It has been discussed that the symmetric spin splitting with respect to momentum is realized in a nonsymmorphic organic compound, κ\kappa-(BETD-TTF)2Cu[N(CN)2]Cl Naka et al. (2019); Hayami et al. (2020a) and a distorted tetragonal compound, RuO2 Berlijn et al. (2017); Ahn et al. (2019) with collinear-type antiferromagnetic (AFM) structures. The subsequent similar works also discuss the spin splitting based on the band calculation in one of the candidate materials, MnF2 Yuan et al. (2020). Moreover, antisymmetric spin splitting can be realized as well in a trigonal oxide Ba3MnNb2O9 with a noncollinear AFM structure Hayami et al. (2020b).

The spin splittings driven by the magnetic phase transition induce interesting physical phenomena through the anisotropic spin-dependent kinetic motions of electrons, for instance, it is proposed the spin current generation by an electric field Ahn et al. (2019); Naka et al. (2019); Hayami et al. (2019a) in collinear magnets and by a shear-type strain in noncollinear magnets Hayami et al. (2020b). It is also shown that nonreciprocal transport arising from the antisymmetrically reshaped band structure is expected in noncoplanar magnets Hayami et al. (2020b, c). Such a magnetic-order driven band deformation has an advantage in its flexible controllability, i.e., it can be accessible by external fields, pressure and temperature. Furthermore, due to their kinetic origin, large spin splittings can be expected for the materials even with the negligibly small SOC. This aspect is significant to extend the scope of materials and explore further efficient functional materials in the field of AFM spintronics Jungwirth et al. (2016); Baltz et al. (2018).

In the present study, we further develop the above scenario, and we provide a complete microscopic guideline to engineer spin- and momentum-dependent band structures in SOC free AFMs. Our guideline is essentially based on local symmetry, which is embodied by the concept of augmented multipoles, especially with the electric and magnetic toroidal multipoles Suzuki et al. (2018); Hayami and Kusunose (2018); Hayami et al. (2018); Watanabe and Yanase (2018); Kusunose et al. (2020). The analysis of couplings among these multipoles in the given Hamiltonian provides necessary ingredients for the band deformations, instead of performing band calculations. By introducing cluster-, bond-, and momentum-type electric and magnetic toroidal multipoles in a magnetic cluster, we can analyze which effective multipole coupling realizes the spin splitting and/or band deformation. Specifically, the symmetric spin splitting occurs in collinear AFMs when the Hamiltonian contains cluster and bond multipoles with the same symmetry. Similarly, the antisymmetric spin splitting is realized in noncollinear AFMs when a bond-type magnetic toroidal multipole is activated through the magnetic phase transition. Furthermore, the antisymmetric band deformation with keeping spin degeneracy is realized in noncoplanar AFMs. We exemplify three lattice systems consisting of a triangle unit, triangular, kagome, and breathing kagome structures, in order to demonstrate how the (spin-dependent) band deformations occur. Our analysis provides a simple prescription of bottom-up design for arbitrary electronic band structures from a microscopic viewpoint. This simple procedure promotes further findings of materials exhibiting a giant spin splitting and related physical responses in SOC free AFMs. As a fruitful outcome, we list up candidate materials showing intrinsic spin splitting and/or band deformations in accordance with MAGNDATA Gallego et al. (2016), magnetic structures database.

Refer to caption
Figure 1: Outline of the engineering procedure of the spin-split and reshaped band structures in terms of augmented multipoles.

II Outline

First, let us show the abstract procedure to engineer the spin-split and/or reshaped band structures by AFM. The overall guideline is summarized in Fig. 1, which consists of the following five parts:

  1. (i)

    Set an NN-site magnetic cluster to describe AFM structures, such as triangle, square, hexagon, and tetrahedron, in accordance with the crystallographic point groups.

  2. (ii)

    Perform irreducible decomposition of an arbitrary hermitian matrix in the cluster for onsite and nnth-neighbor bonds according to the point group. The independent N×NN\times N degrees of freedom are decoupled into the symmetry-adapted NN onsite degrees of freedom, and N(N1)N(N-1) off-diagonal ones in which half of them are for real part, and another half of them are imaginary part.

  3. (iii)

    Assign the augmented multipoles to each decomposed irreducible representation (irrep.), which gives intuitive view of microscopic degrees of freedom. In the decoupled spin and orbital basis, an introduction of two types of multipoles, electric and magnetic toroidal multipoles, is sufficient, which describe polar tensors with time-reversal even and odd, respectively. The onsite and real bond degrees of freedom are represented by the electric multipoles, whereas the imaginary bond degrees of freedom are represented by the magnetic toroidal multipoles. These multipoles are used to span the given Hamiltonian as the symmetry-adapted basis set.

  4. (iv)

    Decompose the hopping and mean-field Hamiltonians into a “scalar-product” form in terms of electric and magnetic toroidal multipoles. The hopping Hamiltonian is described by a linear combination of products between bond and momentum multipoles, while the mean-field Hamiltonian is described by a linear combination of products between cluster multipoles and Pauli matrices of spins.

  5. (v)

    Extract effective spin-multipole couplings by evaluating momentum-dependent spin moments. An effective coupling between cluster (molecular field) and bond multipoles induces momentum multipoles, which yields symmetric, antisymmetric spin splittings, and antisymmetric band deformations depending on the type of multipole couplings: The symmetric spin splitting is represented by momentum electric multipoles, and the antisymmetric spin splitting or spin-independent band deformation is represented by momentum magnetic toroidal multipoles.

Through the above procedure, the microscopic conditions (e.g., which part of hopping element is indispensable, or significant to obtain large splitting, deformation, etc.) for emergent symmetric and antisymmetric spin splittings and antisymmetric band deformations are systematically derived.

The rest of the paper is organized as follows. In Sec. III, we set a magnetic cluster and perform irreducible decomposition for onsite and bond degrees of freedom, which corresponds to the procedures (i) and (ii). The remaining procedures (iii) to (v) are explained in Secs. IV and V. In Sec. IV, we introduce the concept of three kinds of multipoles, cluster, bond, and momentum multipoles representing different electronic degrees of freedom. We describe a general condition of the spin-split band structure and asymmetric spin-degenerate band deformation in Sec. V. In Sec. VI, we show three examples by considering the periodic lattice systems comprised of the triangle unit. We discuss potential candidate materials to exhibit spin splittings and band deformations driven by the magnetic order and summarize the paper in Sec. VII. In two Appendices, we show the explicit expressions of the electric multipoles in Appendix A, and classification of multipoles under eleven Laue classes in Appendix B.

Throughout this paper, we focus on the limit of negligibly small SOC in order to extract intrinsic role of the multipole-spin couplings, and then we adopt the spin-orbital decoupled basis to express the electronic degrees of freedom.

III Magnetic cluster and irreducible decomposition of electronic degrees of freedom

Refer to caption
Figure 2: Schematic picture of a four-site square cluster. The sixteen independent matrix elements are decomposed into four onsite degrees of freedom hih_{i}, six plus six bond degrees of freedom tij=tij+itij′′t_{ij}=t_{ij}^{\prime}+it_{ij}^{\prime\prime} for the real and imaginary components. The corresponding matrix elements and irrep. under the point group D4hD_{4\rm h} are also shown.

Before introducing the multipole descriptions, we perform the irreducible decomposition of the internal electronic degrees of freedom by using the group theory. A magnetic cluster is introduced as a unit to represent the AFM structure. In other words, the magnetic cluster represents the minimal magnetic unit cell we focus on. In the following, we consider the single-orbital model and ignore the atomic orbital degree of freedom, although the extension to the multi-orbital system is straightforward. Then, a spinless basis wave function is represented by ϕ=(ϕ1,ϕ2,,ϕN)\phi=(\phi_{1},\phi_{2},\cdots,\phi_{N}) where ϕi\phi_{i} is the wave function at sublattice ii and NN is the number of sublattices. The Hamiltonian for one-body electronic state is represented by the N×NN\times N matrix except for the spin degree of freedom. As the Hamiltonian is hermitian matrix, its matrix elements are decomposed into the NN diagonal part, the N(N1)/2N(N-1)/2 real and imaginary off-diagonal parts. The NN diagonal elements correspond to the onsite degrees of freedom such as charge and spin densities, while the N(N1)N(N-1) off-diagonal elements correspond to the bond degrees of freedom representing off-site kinetic motion of electrons. For each part, the irreducible decomposition can be performed according to the point group symmetry. It is noted that the N(N1)N(N-1) bond degrees of freedom can be further decomposed into nnth-neighbor bond degree of freedom of the real and imaginary components.

Table 1: Irreducible decomposition of the Hamiltonian matrix for the representative clusters Hayami et al. (2019a). # represents the number of sublattice. The parentheses in the fifth and sixth columns represent the irrep. for each neighbor bond.
Cluster # PG Onsite Real bond Imaginary bond
Triangle 3 D3hD_{3\rm h} A1+E+A^{\prime+}_{1}\oplus E^{\prime+} A1+E+A^{\prime+}_{1}\oplus E^{\prime+} A2EA^{\prime-}_{2}\oplus E^{\prime-}
Rectangle 4 D2hD_{2\rm h} Ag+B1g+B2u+B3u+A^{+}_{g}\oplus B^{+}_{1g}\oplus B^{+}_{2u}\oplus B^{+}_{3u} (Ag+B2u+)(Ag+B3u+)(Ag+B1g+)(A^{+}_{g}\oplus B^{+}_{2u})\oplus(A^{+}_{g}\oplus B^{+}_{3u})\oplus(A^{+}_{g}\oplus B^{+}_{1g}) (B1gB3u)(B1gB2u)(B2uB3u)(B^{-}_{1g}\oplus B^{-}_{3u})\oplus(B^{-}_{1g}\oplus B^{-}_{2u})\oplus(B^{-}_{2u}\oplus B^{-}_{3u})
Square 4 D4hD_{4\rm h} A1g+B2g+Eu+A^{+}_{1g}\oplus B^{+}_{2g}\oplus E^{+}_{u} (A1g+B1g+Eu+)(A1g+B2g+)(A^{+}_{1g}\oplus B^{+}_{1g}\oplus E^{+}_{u})\oplus(A^{+}_{1g}\oplus B^{+}_{2g}) (A2gB2gEu)(Eu)(A^{-}_{2g}\oplus B^{-}_{2g}\oplus E^{-}_{u})\oplus(E^{-}_{u})
Hexagon 6 D6hD_{6\rm h} A1g+B1u+E1u+E2g+A^{+}_{1g}\oplus B^{+}_{1u}\oplus E^{+}_{1u}\oplus E^{+}_{2g} (A1g+B1u+E1u+E2g+)(A^{+}_{1g}\oplus B^{+}_{1u}\oplus E^{+}_{1u}\oplus E^{+}_{2g}) (A2gB2uE1uE2g)(A^{-}_{2g}\oplus B^{-}_{2u}\oplus E^{-}_{1u}\oplus E^{-}_{2g})
(A1g+B2u+E1u+E2g+)(A1g+E2g+)\oplus(A^{+}_{1g}\oplus B^{+}_{2u}\oplus E^{+}_{1u}\oplus E^{+}_{2g})\oplus(A^{+}_{1g}\oplus E^{+}_{2g}) (A2gB1uE1uE2g)(B1uE1u)\oplus(A^{-}_{2g}\oplus B^{-}_{1u}\oplus E^{-}_{1u}\oplus E^{-}_{2g})\oplus(B^{-}_{1u}\oplus E^{-}_{1u})
Tetrahedron 4 TdT_{\rm d} A1+T2+A^{+}_{1}\oplus T^{+}_{2} A1+E+T2+A^{+}_{1}\oplus E^{+}\oplus T^{+}_{2} T1T2T^{-}_{1}\oplus T^{-}_{2}
Octahedron 6 OhO_{\rm h} A1g+Eg+T1u+A^{+}_{1g}\oplus E^{+}_{g}\oplus T^{+}_{1u} (A1g+Eg+T1u+T2g+T2u+)(A^{+}_{1g}\oplus E^{+}_{g}\oplus T^{+}_{1u}\oplus T^{+}_{2g}\oplus T^{+}_{2u}) (A2gEgT1gT1uT2u)(T1u)(A^{-}_{2g}\oplus E^{-}_{g}\oplus T^{-}_{1g}\oplus T^{-}_{1u}\oplus T^{-}_{2u})\oplus(T^{-}_{1u})
(A1g+Eg+)\oplus(A^{+}_{1g}\oplus E^{+}_{g})
Cube 8 OhO_{\rm h} A1g+A2u+T1u+T2g+A^{+}_{1g}\oplus A^{+}_{2u}\oplus T^{+}_{1u}\oplus T^{+}_{2g} (A1g+Eg+T1u+T2g+T2u+)(A^{+}_{1g}\oplus E^{+}_{g}\oplus T^{+}_{1u}\oplus T^{+}_{2g}\oplus T^{+}_{2u}) (A2uEuT1gT1uT2g)(A^{-}_{2u}\oplus E^{-}_{u}\oplus T^{-}_{1g}\oplus T^{-}_{1u}\oplus T^{-}_{2g})
(A1g+A2u+Eg+Eu+T1u+T2g+)\oplus(A^{+}_{1g}\oplus A^{+}_{2u}\oplus E^{+}_{g}\oplus E^{+}_{u}\oplus T^{+}_{1u}\oplus T^{+}_{2g}) (T1gT1uT2gT2u)\oplus(T^{-}_{1g}\oplus T^{-}_{1u}\oplus T^{-}_{2g}\oplus T^{-}_{2u})
(A1g+T2g+)\oplus(A^{+}_{1g}\oplus T^{+}_{2g}) (A2uT1u)\oplus(A^{-}_{2u}\oplus T^{-}_{1u})

As an example, let us consider a square cluster consisting of four sublattice as shown in Fig. 2, which belongs to the point group D4hD_{4\rm h}. The Hamiltonian matrix in spinless space is generally represented by the 4×44\times 4 matrix as

\displaystyle\mathcal{H} =i,j=A,B,C,DciHijcj,\displaystyle=\sum_{i,j={\rm A,B,C,D}}c_{i}^{\dagger}H_{ij}c_{j}, (1)
H\displaystyle H =(hAtABtACtADtBAhBtBCtBDtCAtCBhCtCDtDAtDBtDChD),\displaystyle=\left(\begin{array}[]{cccc}h_{\rm A}&t_{\rm AB}&t_{\rm AC}&t_{\rm AD}\\ t_{\rm BA}&h_{\rm B}&t_{\rm BC}&t_{\rm BD}\\ t_{\rm CA}&t_{\rm CB}&h_{\rm C}&t_{\rm CD}\\ t_{\rm DA}&t_{\rm DB}&t_{\rm DC}&h_{\rm D}\\ \end{array}\right), (6)

where cic_{i}^{\dagger} (cic_{i}) is the creation (annihilation) operator at site ii. hih_{i} and tij=tjit_{ij}=t_{ji}^{*} (i,j=A,B,C,Di,j={\rm A,B,C,D}) are real and complex numbers corresponding to onsite and hopping terms, respectively. The Hamiltonian matrix HH in Eq. (6) is specified by giving sixteen independent model parameters consisting of four hih_{i} and twelve tijt_{ij}.

The matrix HH is decomposed into the onsite potential and the hopping parts. The sublattice-basis wave function {ϕA,ϕB,ϕC,ϕD}\left\{\phi_{\rm A},\phi_{\rm B},\phi_{\rm C},\phi_{\rm D}\right\} is decomposed into the irrep. of the point group D4hD_{\rm 4h} as A1gB2gEuA_{1g}\oplus B_{2g}\oplus E_{u}, each of which represents the molecular orbital belonging to its irrep. Then, the internal degrees of freedom is given by its direct product as (A1gB2gEu)(A1gB2gEu)=2A1g+B1g+B2g+2Eu+A2gB2g2Eu(A_{1g}\oplus B_{2g}\oplus E_{u})\otimes(A_{1g}\oplus B_{2g}\oplus E_{u})=2{A}_{1g}^{+}\oplus{B}_{1g}^{+}\oplus{B}_{2g}^{+}\oplus 2{E}_{u}^{+}\oplus{A}_{2g}^{-}\oplus{B}_{2g}^{-}\oplus 2{E}_{u}^{-}, where the subscript represents the spatial parity (even: gg, odd: uu) and the superscript represents the time-reversal parity (even: ++, odd: -). Among them, for the onsite part, four parameters (hA,hB,hC,hD)(h_{\rm A},h_{\rm B},h_{\rm C},h_{\rm D}) are decomposed into A1g+B2g+Eu+{A}_{1g}^{+}\oplus{B}_{2g}^{+}\oplus{E}_{u}^{+}. By applying the symmetry operation of D4hD_{\rm 4h} to the degrees of freedom in the magnetic cluster, one can find that the matrix elements in each irrep. are given by

A1g+\displaystyle{A}_{1g}^{+} :hA=hB=hC=hD,\displaystyle:h_{\rm A}=h_{\rm B}=h_{\rm C}=h_{\rm D},
B2g+\displaystyle{B}_{2g}^{+} :hA=hB=hC=hD,\displaystyle:h_{\rm A}=h_{\rm B}=-h_{\rm C}=-h_{\rm D},
Eu+\displaystyle{E}_{u}^{+} :hA=hB=hC=hD,\displaystyle:h_{\rm A}=-h_{\rm B}=-h_{\rm C}=h_{\rm D},
:hA=hB=hC=hD.\displaystyle:h_{\rm A}=-h_{\rm B}=h_{\rm C}=-h_{\rm D}. (7)

For the hopping part, the hopping parameters tijt_{ij} are divided into the real (tij=tji)(t_{ij}^{\prime}=t_{ji}^{\prime}) and imaginary (tij′′=tji′′)(t_{ij}^{\prime\prime}=-t_{ji}^{\prime\prime}) components. By performing the irreducible decomposition for each nnth-neighbor bond, the real part is decomposed into

A1g+\displaystyle{A}_{1g}^{+} :tAC=tAD=tBC=tBD,\displaystyle:t_{\rm AC}^{\prime}=t_{\rm AD}^{\prime}=t_{\rm BC}^{\prime}=t_{\rm BD}^{\prime},
B1g+\displaystyle{B}_{1g}^{+} :tAC=tAD=tBC=tBD,\displaystyle:-t_{\rm AC}^{\prime}=t_{\rm AD}^{\prime}=t_{\rm BC}^{\prime}=-t_{\rm BD}^{\prime},
Eu+\displaystyle{E}_{u}^{+} :tAD=tBC,tAC=tBD=0,\displaystyle:t_{\rm AD}^{\prime}=-t_{\rm BC}^{\prime},\ t_{\rm AC}^{\prime}=t_{\rm BD}^{\prime}=0,
:tAC=tBD,tAD=tBC=0,\displaystyle:t_{\rm AC}^{\prime}=-t_{\rm BD}^{\prime},\ t_{\rm AD}^{\prime}=t_{\rm BC}^{\prime}=0, (8)

for the first-neighbor bond, and

A1g+\displaystyle{A}_{1g}^{+} :tAB=tCD,\displaystyle:t_{\rm AB}^{\prime}=t_{\rm CD}^{\prime},
B1g+\displaystyle{B}_{1g}^{+} :tAB=tCD,\displaystyle:t_{\rm AB}^{\prime}=-t_{\rm CD}^{\prime}, (9)

for the second-neighbor bond.

Similarly, the imaginary bond degree of freedom is decomposed into

A2g\displaystyle{A}_{2g}^{-} :tAC′′=tAD′′=tBC′′=tBD′′,\displaystyle:-t_{\rm AC}^{\prime\prime}=t_{\rm AD}^{\prime\prime}=t_{\rm BC}^{\prime\prime}=-t_{\rm BD}^{\prime\prime},
B2g\displaystyle{B}_{2g}^{-} :tAC′′=tAD′′=tBC′′=tBD′′,\displaystyle:t_{\rm AC}^{\prime\prime}=t_{\rm AD}^{\prime\prime}=t_{\rm BC}^{\prime\prime}=t_{\rm BD}^{\prime\prime},
Eu\displaystyle{E}_{u}^{-} :tAC′′=tBD′′,tAD′′=tBC′′=0,\displaystyle:-t_{\rm AC}^{\prime\prime}=t_{\rm BD}^{\prime\prime},\ t_{\rm AD}^{\prime\prime}=t_{\rm BC}^{\prime\prime}=0,
:tAD′′=tBC′′,tAC′′=tBD′′=0,\displaystyle:-t_{\rm AD}^{\prime\prime}=t_{\rm BC}^{\prime\prime},\ t_{\rm AC}^{\prime\prime}=t_{\rm BD}^{\prime\prime}=0, (10)

for the first-neighbor bond, and

Eu\displaystyle E_{u}^{-} :tAB′′=tCD′′,\displaystyle:-t_{\rm AB}^{\prime\prime}=t_{\rm CD}^{\prime\prime},
:tAB′′=tCD′′,\displaystyle:t_{\rm AB}^{\prime\prime}=t_{\rm CD}^{\prime\prime}, (11)

for the second-neighbor bond.

In general, the N×NN\times N matrix elements in NN sublattice cluster are also represented by the irreps. of the given point group. It is noted that such an irreducible decomposition is performed much more intuitively by using the multipole description, as will be shown in the next section. We summarize the irreducible decomposition for onsite and bond degrees of freedom in the representative clusters in Table 1 Hayami et al. (2019a).

IV Multipole description

In this section, we describe the concept of multipole. We introduce three kinds of multipole notations, which are necessary to describe the distinct electronic degrees of freedom in the tight-binding model. The cluster multipole is used to describe the onsite degree of freedom in Sec. IV.1, the bond multipole is for the bond degree of freedom in Sec. IV.2, and the momentum multipole is for a wave-vector-dependent form factor in periodic lattice systems in Sec. IV.3. Then, we show the correspondence between the irreps. explained in the previous section and these multipoles in Sec. IV.4.

IV.1 Cluster multipole

Refer to caption
Figure 3: (a) Collinear AFM order in a square cluster under the point group D4hD_{4{\rm h}}. The collinear order parameter (Qxy(c)σyQ^{\rm(c)}_{xy}\sigma_{y}) is decomposed into the electric-quadrupole-type alignment of point charges (Qxy(c)Q^{\rm(c)}_{xy}) and the spin along the yy direction (σy\sigma_{y}). (b) Coplanar magnetic order in a square cluster, which is regarded as a superposition of two collinear spin components with Qx(c)σxQ^{\rm(c)}_{x}\sigma_{x} and Qy(c)σyQ^{\rm(c)}_{y}\sigma_{y}.

The cluster multipole is defined to describe the onsite degree of freedom in the tight-binding Hamiltonian. As arbitrary onsite degrees of freedom are represented by a superposition of local potentials at each atomic site, all the anisotropic charge distributions in a cluster are systematically represented by using the spherical harmonics with the origin at the center of the cluster, which is related to the electric multipole degree of freedom Kusunose (2008); Kuramoto et al. (2009); Santini et al. (2009). Eventually, the anisotropic charge distributions on cluster sites are described as a cluster electric multipole Q~lm(c)\tilde{Q}^{(c)}_{lm}, which is given by

Q~lm(c)=i=1Nqi(lm)Olm(𝑹i),\displaystyle\tilde{Q}^{\rm(c)}_{lm}=\sum_{i=1}^{N}q^{(lm)}_{i}O_{lm}(\bm{R}_{i}), (12)

where Olm(𝑹i)=4π/(2l+1)RilYlm(𝑹^i)O_{lm}(\bm{R}_{i})=\sqrt{4\pi/(2l+1)}R_{i}^{l}Y^{*}_{lm}(\hat{\bm{R}}_{i}), 𝑹i=(Xi,Yi,Zi)\bm{R}_{i}=(X_{i},Y_{i},Z_{i}) is the position vector of iith cluster site, qi(lm)q^{(lm)}_{i} is the local electric charge of iith cluster site, and NN is the number of sites in a cluster. We omit electric charge unit e-e for notational simplicity. Ylm(𝑹^i)Y_{lm}(\hat{\bm{R}}_{i}) is the spherical harmonics as a function of angle 𝑹^i=𝑹i/|𝑹i|\hat{\bm{R}}_{i}=\bm{R}_{i}/|\bm{R}_{i}| with the azimuthal and magnetic quantum numbers, ll and mm (lml-l\leq m\leq l). In the following, we regard the symbol Olm(𝒓)O_{lm}(\bm{r}) as the harmonics of the point group such as cubic and hexagonal ones instead of the spherical harmonics, which are real functions given by linear combinations of OlmO_{lm} and OlmO_{l-m} as shown in Appendix A Kusunose (2008). Through this expression, we define qi(lm)q^{(lm)}_{i} for the specified electric multipole, and the corresponding matrix (operator) expression is given by Qlm(c)=iqi(lm)|ii|Q^{\rm(c)}_{lm}=\sum_{i}q^{(lm)}_{i}\ket{i}\bra{i} where |i\ket{i} is the atomic site basis.

Such a cluster multipole can also describe magnetic ordering patterns in a cluster Suzuki et al. (2017, 2019). In the spin-orbital decoupled basis, it is useful to express the magnetic structure as a linear combination of direct products of Qlm(c)Q^{\rm(c)}_{lm} and the Pauli matrices of spin 𝝈=(σx,σy,σz)\bm{\sigma}=(\sigma_{x},\sigma_{y},\sigma_{z}). Then, any types of magnetic orderings coupled with the corresponding molecular fields are expressed by a linear combination of Qlm(c)σμQ^{\rm(c)}_{lm}\sigma_{\mu}. Namely, the mean-field Hamiltonian of the AFM ordering is represented by

Hm=lmμ=x,y,zmlmμQlm(c)σμ,\displaystyle H_{m}=\sum_{lm}\sum_{\mu=x,y,z}m^{\mu}_{lm}Q^{\rm(c)}_{lm}\sigma_{\mu}, (13)

where the coefficient mlmμm^{\mu}_{lm} is a conjugate field of an order parameter in the AFM state. As Qlm(c)Q^{\rm(c)}_{lm} is time-reversal-even and σμ\sigma_{\mu} is time-reversal-odd, HmH_{m} is time-reversal-odd as it is a symmetry breaking term. From the multipole viewpoint, the ordering pattern is characterized by the type of the emergent multipole: The ferromagnetic structure corresponds to the isotropic electric monopole and the AFM structure corresponds to the anisotropic electric multipoles for l1l\geq 1.

In the case of the collinear AFM order, the mean-field Hamiltonian matrix is represented by the single component of σμ\sigma_{\mu} where μ\mu denotes the ordered moment direction, although it is taken to be arbitrary in the absence of the SOC. We show an example of the staggered AFM ordering in a square cluster under the point group D4hD_{\rm 4h} in Fig. 3(a). By decomposing this magnetic structure into the alignment of point charges and spin as shown in Fig. 3(a), and then, evaluating Qlm(c)Q^{\rm(c)}_{lm} via Eq. (12), one can find that the corresponding multipole is Qxy(c)Q^{\rm(c)}_{xy}.

Such a multipole description is also understood from a symmetry viewpoint. The mean-field matrix in a square cluster in Eq. (6) is represented by

Hm\displaystyle H_{m} =hσ(1000010000100001),\displaystyle=h\sigma\left(\begin{array}[]{cccc}1&0&0&0\\ 0&1&0&0\\ 0&0&-1&0\\ 0&0&0&-1\\ \end{array}\right), (18)

where σ=±1\sigma=\pm 1 for up and down spins. From Eq. (III), this matrix element except for σ\sigma belongs to the irrep. B2g+{B}_{2g}^{+} under the point group D4hD_{4{\rm h}}, which is the same irrep. of the Qxy(c)Q^{\rm(c)}_{xy}-type electric quadrupole (See also the correspondence between the irrep. under the point group and multipoles in Sec. IV.4Hayami et al. (2018). More intuitively, the real-space point charge alignment in a square cluster clearly indicates the presence of xyxy-type electric quadrupole; the positive charges are in the [110][110] direction, while the negative ones are in the [1¯10][\bar{1}10] direction.

In a similar manner, coplanar and noncoplanar magnetic structures are described by a linear combination of two and three components of σμ\sigma_{\mu}, respectively. Figure 3(b) shows an example of the coplanar spin structure in a square cluster where each spin points to the 110\langle 110\rangle radial direction. Also in this case, the plane including spins is taken to be arbitrary due to spin rotational symmetry. By using Eq. (12) for two spin components, one can find that the spin pattern in Fig. 3(b) is proportional to Qx(c)σx+Qy(c)σyQ^{\rm(c)}_{x}\sigma_{x}+Q^{\rm(c)}_{y}\sigma_{y}. The mean-field matrix is given in the form of

Hm\displaystyle H_{m} =h[σx(1000010000100001)+σy(1000010000100001)].\displaystyle=h\left[\sigma_{x}\left(\begin{array}[]{cccc}-1&0&0&0\\ 0&1&0&0\\ 0&0&1&0\\ 0&0&0&-1\\ \end{array}\right)+\sigma_{y}\left(\begin{array}[]{cccc}-1&0&0&0\\ 0&1&0&0\\ 0&0&-1&0\\ 0&0&0&1\\ \end{array}\right)\right]. (27)

It is apparent from Fig. 3(b) that Qx(c)Q_{x}^{\rm(c)} and Qy(c)Q_{y}^{\rm(c)} represent the xx and yy electric dipoles in the multipole language.

IV.2 Bond multipole

Refer to caption
Figure 4: The examples of bond multipoles in a square unit. The real and imaginary hoppings correspond to the presence of the electric monopole on the bond center and the magnetic toroidal dipole 𝒕(ij)\bm{t}_{(ij)} along the bond direction, respectively. See also Eqs. (33) and (34). From the left, the electric quadrupole Qv(b1)Q^{\rm(b1)}_{v}, magnetic toroidal dipole Tx(b1)T^{\rm(b1)}_{x}, electric dipole Qx(b1)Q^{\rm(b1)}_{x}, and magnetic toroidal hexadecapole T4zα(b1)T^{\alpha{\rm(b1)}}_{4z} are presented.

The bond multipole is introduced to describe the bond degree of freedom in the system, which corresponds to the off-diagonal hopping part in the tight-binding Hamiltonian Hayami et al. (2019b).

First, in order to get an intuitive insight into the relation between bond multipoles and hoppings, let us consider a two-site problem connected by the complex hopping t=t+it′′t=t^{\prime}+it^{\prime\prime}. The two sites are denoted as A and B, which are separated by the distance aa in the xx direction. By using the molecular-orbital basis {|ϕ1,|ϕ2}\{\ket{\phi_{1}},\ket{\phi_{2}}\} instead of the atomic site basis {|A,|B}\{\ket{{\rm A}},\ket{{\rm B}}\}, the real and imaginary hopping matrices are transformed as

Re[Ht]=(t00t),iIm[Ht]=(0it′′it′′0),\displaystyle{\rm Re}[H_{t}]=\left(\begin{array}[]{cc}t^{\prime}&0\\ 0&-t^{\prime}\\ \end{array}\right),\ \ i{\rm Im}[H_{t}]=\left(\begin{array}[]{cc}0&-it^{\prime\prime}\\ it^{\prime\prime}&0\\ \end{array}\right), (32)

where |ϕ1=(1/2)(|A+|B)\ket{\phi_{1}}=(1/\sqrt{2})(\ket{{\rm A}}+\ket{{\rm B}}) and |ϕ2=(1/2)(|A|B)\ket{\phi_{2}}=(1/\sqrt{2})(\ket{{\rm A}}-\ket{{\rm B}}). As the anisotropy of the molecular orbitals {|ϕ1,|ϕ2}\{\ket{\phi_{1}},\ket{\phi_{2}}\} is the same as the ss- and pxp_{x}-orbital wave functions, the ordinary atomic-scale multipole description in Ref. Hayami and Kusunose, 2018 can be applied. Then, by comparing the matrix elements in ss-pxp_{x} orbital basis, Re[Ht]{\rm Re}[H_{t}] corresponds to the electric monopole Q0Q_{0}, while Im[Ht]{\rm Im}[H_{t}] corresponds to the magnetic toroidal dipole TxT_{x} Hayami and Kusunose (2018). This result indicates that the real hopping is expressed as the electric monopole on the bond center, while the imaginary hopping is expressed as the magnetic toroidal dipole along the bond direction. This assignment of multipole moments on the bond center is reasonable from a symmetry viewpoint, since the real (imaginary) hopping is equivalent with the time-reversal-even scalar (time-reversal-odd polar vector), which corresponds to the electric monopole (magnetic toroidal dipole).

This result is generalized for arbitrary cluster systems. Any types of hoppings are represented by bond electric and magnetic toroidal multipoles, Q~lm(b)\tilde{Q}^{\rm(b)}_{lm} and T~lm(b)\tilde{T}^{\rm(b)}_{lm}, which are expressed as

Q~lm(b)\displaystyle\tilde{Q}^{\rm(b)}_{lm} =(ij)Nbondq(ij)(lm)Olm(𝑹(ij)),\displaystyle=\sum_{(ij)}^{N_{\rm bond}}q^{(lm)}_{(ij)}O_{lm}(\bm{R}_{(ij)}), (33)
T~lm(b)\displaystyle\tilde{T}^{\rm(b)}_{lm} =(ij)Nbond𝒕(ij)(lm)Olm(𝑹(ij)),\displaystyle=\sum_{(ij)}^{N_{\rm bond}}\bm{t}^{(lm)}_{(ij)}\cdot\bm{\nabla}O_{lm}(\bm{R}_{(ij)}), (34)

where NbondN_{\rm bond} is the number of bonds in a cluster and 𝑹(ij)\bm{R}_{(ij)} is the position vector at the ii-jj bond center. q(ij)(lm)q^{(lm)}_{(ij)} and 𝒕(ij)(lm)\bm{t}^{(lm)}_{(ij)} are the local electric charge and local magnetic toroidal dipole at 𝑹(ij)\bm{R}_{(ij)}, respectively, where 𝒕(ij)(lm)=itij′′(lm)𝒏ij\bm{t}^{(lm)}_{(ij)}=it^{\prime\prime(lm)}_{ij}\bm{n}_{ij} with 𝒏ij\bm{n}_{ij} being the unit vector connecting between ii and jj sites. The multipole assignment is independently performed per nnth-neighbor bond. The corresponding matrices (operators) of bond electric and magnetic toroidal multipoles are given by Qlm(b)=(ij)q(ij)(lm)|ji|Q^{\rm(b)}_{lm}=\sum_{(ij)}q^{(lm)}_{(ij)}\ket{j}\bra{i} and Tlm(b)=(ij)(𝒕(ij)(lm)𝒏ij)|ji|T^{\rm(b)}_{lm}=\sum_{(ij)}(\bm{t}^{(lm)}_{(ij)}\cdot\bm{n}_{ij})\ket{j}\bra{i}.

By using these matrices, we express any bond modulations in terms of bond multipoles. In particular, the bond magnetic toroidal multipoles, Tlm(b)T^{\rm(b)}_{lm}, represent the anisotropic current distribution including a loop-current distribution discussed in cuprates and iridates Shekhter_PhysRevB.80.214501; zhao2016evidence; murayama2020bond, as the imaginary hopping represents a local current along the bond.

Let us again consider an example in a square cluster under the point group D4hD_{4{\rm h}}, whose Hamiltonian is shown in Eq. (6). There are six real and imaginary bond degrees of freedom, which are assigned as six electric and magnetic toroidal multipoles, respectively. By using Eqs. (33) and (34), the first-neighbor four real (imaginary) bonds correspond to electric monopole Q0(b1)Q^{\rm(b1)}_{0}, electric quadrupole Qv(b1)Q^{\rm(b1)}_{v}, and electric dipoles (Qx(b1),Qy(b1))(Q^{\rm(b1)}_{x},Q^{\rm(b1)}_{y}) [magnetic toroidal hexadecapole T4xα(b1)T^{\alpha{\rm(b1)}}_{4x}, magnetic toroidal quadrupole Txy(b1)T^{\rm(b1)}_{xy}, and magnetic toroidal dipoles (Tx(b1),Ty(b1))(T^{\rm(b1)}_{x},T^{\rm(b1)}_{y})], while the second-neighbor two real (imaginary) bonds correspond to electric monopole Q0(b2)Q^{\rm(b2)}_{0} and electric quadrupole Qxy(b2)Q^{\rm(b2)}_{xy} [magnetic toroidal dipoles (Tx(b2),Ty(b2))(T^{\rm(b2)}_{x},T^{\rm(b2)}_{y})] where the integer (n=1,2)(n=1,2) in superscript represents the nnth-neighbor bond. The specific examples of Qv(b1)Q^{\rm(b1)}_{v}, Tx(b1)T^{\rm(b1)}_{x}, Qx(b1)Q^{\rm(b1)}_{x}, and T4zα(b1)T^{\alpha{\rm(b1)}}_{4z} are shown in Fig. 4. It is noted that the magnetic toroidal hexadecapole T4zα(b1)T_{4z}^{\rm\alpha(b1)} with the vortex-like alignment of 𝒕(ij)\bm{t}_{(ij)} in Fig. 4 is equivalent to the magnetic dipole along zz direction. Nevertheless, we use the higher-rank hexadecapole T4zα(b1)T_{4z}^{\rm\alpha(b1)} since we use the convention in this paper that all bond degrees of freedom are described by the electric or magnetic toroidal multipoles.

IV.3 Momentum multipole

Finally, we introduce momentum multipoles to represent the momentum dependence in crystals. In the 𝒌𝟎\bm{k}\to\bm{0} limit, the spherical harmonics as a function of 𝒌^=𝒌/|𝒌|\hat{\bm{k}}=\bm{k}/|\bm{k}|, Ylm(𝒌^)Y_{lm}(\hat{\bm{k}}), gives the anisotropic momentum distribution. As 𝒌\bm{k} is a polar vector with time-reversal odd, the even-(odd-)rank component in Ylm(𝒌^)Y_{lm}(\hat{\bm{k}}) corresponds to the electric (magnetic toroidal) multipoles, which are defined as

Qlm(m)(𝒌)\displaystyle Q^{\rm(m)}_{lm}(\bm{k}) =Olm(𝒌)forevenl,\displaystyle=O_{lm}(\bm{k})\ \ {\rm for\ even\ }l, (35)
Tlm(m)(𝒌)\displaystyle T^{\rm(m)}_{lm}(\bm{k}) =Olm(𝒌)foroddl.\displaystyle=O_{lm}(\bm{k})\ \ {\rm for\ odd\ }l. (36)

The explicit expressions are given by replacing 𝒓\bm{r} with 𝒌\bm{k} in Table 8 in Appendix A where the odd-rank multipoles in Table 8 should be replaced with Tlm(m)(𝒌)T_{lm}^{\rm(m)}(\bm{k}).

In general, the momentum dependence in crystals has periodicity and is represented by a superposition of trigonometric functions of 𝒌\bm{k}. In this case, the momentum form factor consists of the momentum multipoles up to the infinite rank belonging to the same irrep. For example, we consider the single-band tight-binding model on a simple square lattice under the point group D4hD_{4{\rm h}} with the lattice constant aa. The momentum form factor for the nearest-neighbor bond f(𝒌)f(\bm{k}) is given by a linear combination of momentum multipoles as

f(𝒌)\displaystyle f(\bm{k}) =cos(kxa)+cos(kya)\displaystyle=\cos(k_{x}a)+\cos(k_{y}a) (37)
=2a22(kx2+ky2)+a424(kx4+ky4)+\displaystyle=2-\frac{a^{2}}{2}(k_{x}^{2}+k_{y}^{2})+\frac{a^{4}}{24}(k_{x}^{4}+k_{y}^{4})+\cdots (38)
=c1Q0(m)(𝒌)+c2Qu(m)(𝒌)+c3Q4(m)(𝒌)+,\displaystyle=c_{1}Q^{(\rm m)}_{0}(\bm{k})+c_{2}Q^{(\rm m)}_{u}(\bm{k})+c_{3}Q^{(\rm m)}_{4}(\bm{k})+\cdots, (39)

where cic_{i} (i=1,2,i=1,2,\cdots) are the expansion coefficients. f(𝒌)f(\bm{k}) clearly consists of the multipoles belonging to the totally symmetric irrep. A1gA_{1g} under D4hD_{\rm 4h}. To specify the type of multipole, we use the lowest-rank multipole in the superscript of f(𝒌)f(\bm{k}) as the convention is often used in the field of superconductivity Nomoto et al. (2016); Sumita and Yanase (2020). In this case, f(𝒌)f(\bm{k}) is expressed as fQ0(𝒌)f^{Q_{0}}(\bm{k}).

Similarly, the form factors belonging to the other irreps. are also described by the different set of multipoles. In the case of D4hD_{\rm 4h}, when the sign of the hopping along the yy direction is opposite, the form factor is given by

f(𝒌)\displaystyle f(\bm{k}) =cos(kxa)cos(kya)\displaystyle=\cos(k_{x}a)-\cos(k_{y}a) (40)
=a22(kx2ky2)+a424(kx4ky4)+\displaystyle=-\frac{a^{2}}{2}(k_{x}^{2}-k_{y}^{2})+\frac{a^{4}}{24}\left(k_{x}^{4}-k_{y}^{4}\right)+\cdots (41)
=c1Qv(m)(𝒌)+c2Q4v(m)(𝒌)+,\displaystyle=c_{1}Q^{(\rm m)}_{v}(\bm{k})+c_{2}Q^{(\rm m)}_{4v}(\bm{k})+\cdots, (42)

which belongs to the irrep. B1gB_{1g}. In this situation, we denote f(𝒌)f(\bm{k}) as fQv(𝒌)f^{Q_{v}}(\bm{k}).

Moreover, when the imaginary hopping appears only in the xx direction under D4hD_{\rm 4h}, the form factor is given by

f(𝒌)\displaystyle f(\bm{k}) =sin(kxa)\displaystyle=\sin(k_{x}a) (43)
=akx+a33kx3+\displaystyle=ak_{x}+\frac{a^{3}}{3}k_{x}^{3}-+\cdots (44)
=c1Tx(m)(𝒌)+c2Txα(m)(𝒌)+,\displaystyle=c_{1}T^{(\rm m)}_{x}(\bm{k})+c_{2}T^{\alpha(\rm m)}_{x}(\bm{k})+\cdots, (45)

which belongs to the irrep. EuE_{u}. In this situation, we denote f(𝒌)f(\bm{k}) as fTx(𝒌)f^{T_{x}}(\bm{k}). Such form factors in Eqs. (40) and (43) can appear in the tight-binding Hamiltonian when the system has the sublattice degree of freedom, as the local site symmetry is lowered than the lattice symmetry.

IV.4 Irreducible representation of multipoles in crystal

In the crystal systems, a part of the rotational symmetry and/or inversion symmetry are lost due to the regular and discrete alignment of the ions. As a result, the multipole degrees of freedom belonging to the same irrep. are not distinguished from the symmetry viewpoint. In other words, the irrep. of the rotational group split into subgroups according to the point-group irrep. For example, some even-parity and odd-parity multipoles belong to the same irrep. in noncentrosymmetric crystals.

To avoid such confusion, we uniquely assign the multipoles of irrep. as in Sec. III by the following rules: Among the multipoles belonging to the same irrep., we adopt the lowest even-rank electric multipoles for time-reversal even quantities, whereas we adopt the lowest odd-rank magnetic toroidal multipoles for time-reversal odd quantities. In this convention, the momentum-type odd-rank electric multipoles and even-rank magnetic toroidal multipoles do not appear in the Hamiltonian, as will be clarified in Sec. V.

Following the above rules, the multipoles and the irrep. have a one-to-one correspondence. We summarize the corresponding multipole notations under 32 point groups in Appendix B, where we divide them into eleven Laue classes with the same number of the irreps., in Tables 9-19. The compatibility relation of multipole within the same Laue class is given in the same row in the table. On the other hand, for the compatibility relation of multipole between the different Laue classes, we use the group-subgroup compatibility relation and adopt the lower-rank multipole to assign the irrep. For example, by the relation between TdT_{\rm d} and TT, the electric monopole Q0Q_{0} belonging to A1A_{1} and the electric hexadecapole Q6tQ_{6t} belonging to A2A_{2} under the point group TdT_{\rm d} turn into the same irrep. AA under the point group TT. In this case, the multipole belonging to AA under TT is denoted as Q0Q_{0}.

Let us remark on the connection of the quantities introduced in the cluster and the lattice. Although we assign the multipoles to the electronic degrees of freedom by introducing the magnetic cluster, there is a situation where the lattice symmetry is higher than the cluster symmetry due to the additional operations combined with the translation. In such a situation, we replace the irreps. in a cluster with the corresponding ones in a lattice in accordance with the compatibility relation. Accordingly, the multipoles in a cluster are mapped onto those in a lattice. In Sec. VI, we exemplify this by considering the triangular and kagome lattices consisting of the triangle unit where the cluster and lattice symmetries are different with each other.

V Momentum-dependent spin splitting and band deformation

By using the multipole notations introduced in Sec. IV, we express the Hamiltonian in terms of multipoles in Sec. V.1. Then, we analyze systematically when and how the spin splitting and antisymmetric deformation in the band structure occur in Sec. V.2.

V.1 Hamiltonian

In the absence of the SOC, the single-orbital Hamiltonian consists of the hopping part without the spin dependence and the mean-field part gives rise to the symmetry breaking due to the magnetic ordering. The total tight-binding Hamiltonian is generally represented by

=𝒌σσγγc𝒌γσ[δσσ(HtQ+HtT)γγ+δγγHmσσ]c𝒌γσ,\displaystyle\mathcal{H}=\!\!\sum_{\bm{k}\sigma\sigma^{\prime}\gamma\gamma^{\prime}}c^{\dagger}_{\bm{k}\gamma\sigma}\left[\delta_{\sigma\sigma^{\prime}}(H_{t}^{Q}+H_{t}^{T})^{\gamma\gamma^{\prime}}+\delta_{\gamma\gamma^{\prime}}H_{m}^{\sigma\sigma^{\prime}}\right]c_{\bm{k}\gamma^{\prime}\sigma^{\prime}}, (46)

where c𝒌γσc^{\dagger}_{\bm{k}\gamma\sigma} (c𝒌γσc_{\bm{k}\gamma\sigma}) is the creation (annihilation) operator at wave vector 𝒌\bm{k} and sublattice γ\gamma. HtQH_{t}^{Q} and HtTH_{t}^{T} stand for the real and imaginary hopping matrices, respectively, which are represented by a linear combination of the product between bond and momentum multipoles as

HtQ\displaystyle H^{Q}_{t} =lmfQlm(𝒌)Qlm(b)(l:even),\displaystyle=\sum_{lm}f^{Q_{lm}}(\bm{k})Q^{\rm(b)}_{lm}\ \ \ (l:{\rm even}), (47)
HtT\displaystyle H^{T}_{t} =lmfTlm(𝒌)Tlm(b)(l:odd).\displaystyle=\sum_{lm}f^{T_{lm}}(\bm{k})T^{\rm(b)}_{lm}\ \ \ (l:{\rm odd}). (48)

In Eq. (47), the electric monopole contribution fQ0(𝒌)Q0(b)f^{Q_{0}}(\bm{k})Q^{\rm(b)}_{0} always appears in HtQH^{Q}_{t}, while the higher-rank contribution depends on the lattice symmetry. On the other hand, HtTH^{T}_{t} in Eq. (48) exists only in the absence of the local inversion symmetry. It is noted that the cross terms fTlm(𝒌)Qlm(b)f^{T_{lm}}(\bm{k})Q^{\rm(b)}_{lm} and fQlm(𝒌)Tlm(b)f^{Q_{lm}}(\bm{k})T^{\rm(b)}_{lm} do not appear in the Hamiltonian due to the time-reversal symmetry. The mean-field term HmH_{m} is represented by the cluster multipole as already introduced in Eq. (13).

V.2 Band deformation

Refer to caption
Figure 5: Schematic pictures of three types of band deformations; (a) the symmetric spin splitting, (b) the antisymmetric spin splitting, and (c) the antisymmetric band deformation with spin degeneracy. In (a) and (b), the red and blue curves represent the up-spin and down-spin polarized bands, respectively.

Let us first discuss the essential points for the spin-split band structures in AFMs without the SOC. We consider three types of band deformations, the symmetric spin splitting, the antisymmetric spin splitting, and the antisymmetric band deformation, which are categorized into the different symmetry classes. Note that the time-reversal symmetry is always broken, as we focus on magnetic orderings.

The first category is the symmetric spin splitting with respect to 𝒌\bm{k} when there is the spatial inversion symmetry in the system. In this situation, the spin-dependent band dispersion is described by the product of the even function of 𝒌\bm{k} and spin σ\sigma. This means that the symmetric spin splitting arises through the effective coupling between the momentum electric multipole Qlm(m)(𝒌)Q^{\rm(m)}_{lm}(\bm{k}) and σ\sigma. The lowest-order contribution is given by the rank-0 electric monopole, which merely corresponds to the momentum-independent Zeeman-like spin splitting in the band structure. In the following, we mainly focus on the higher-rank contribution for l2l\geq 2. The schematic example in the case Qxy(m)(𝒌)σkxkyσQ^{\rm(m)}_{xy}(\bm{k})\sigma\sim k_{x}k_{y}\sigma is shown in Fig. 5(a).

The second category is the antisymmetric spin splitting with respect to 𝒌\bm{k} in the absence of the spatial inversion symmetry and the product symmetry of time-reversal and spatial inversion operations in addition to the breaking of time-reversal symmetry. The functional form of the spin splittings is represented by the product of the odd function of 𝒌\bm{k} and spin σ\sigma. Thus, the antisymmetic spin splitting occurs when momentum magnetic toroidal multipole Tlm(m)(𝒌)T^{\rm(m)}_{lm}(\bm{k}) is coupled with σ\sigma. The schematic example in the case of Tx(m)(𝒌)σkxσT^{\rm(m)}_{x}(\bm{k})\sigma\sim k_{x}\sigma is shown in Fig. 5(b).

The third category is the antisymmetric band deformation with the spin degeneracy in the absence of spatial inversion symmetry, with preserving the product symmetry of time-reversal and spatial inversion operations. The band structure becomes asymmetric due to the contribution from the odd function of 𝒌\bm{k} in addition to the even function of 𝒌\bm{k}. In terms of the multipole description, the antisymmetric part of the band deformation is described by the emergence of the momentum magnetic toroidal multipole Tlm(m)(𝒌)T^{\rm(m)}_{lm}(\bm{k}) solely without spin dependence Dubovik and Tugushev (1990); Kopaev (2009); Yanase (2014); Hayami et al. (2014). The schematic example in the case of Tx(m)(𝒌)kxT^{\rm(m)}_{x}(\bm{k})\sim k_{x} is shown in Fig. 5(c).

To clarify a necessary condition of the microscopic model parameters for the band deformations in AFM orderings beyond symmetry argument, one can need to know when and how the momentum multipoles, Qlm(m)(𝒌)Q^{\rm(m)}_{lm}(\bm{k}) and Tlm(m)(𝒌)T^{\rm(m)}_{lm}(\bm{k}), become active and are coupled with spin σ\sigma. To examine such conditions, we introduce the following quantity at wave vector 𝒌\bm{k} in the magnetic unit cell,

Tr[eβ𝒌σμ]=s(β)ss!gsμ(𝒌),\displaystyle\mathrm{Tr}[e^{-\beta\mathcal{H}_{\bm{k}}}\sigma_{\mu}]=\sum_{s}\frac{(-\beta)^{s}}{s!}g_{s}^{\mu}(\bm{k}), (49)

where μ=0,x,y,z\mu=0,x,y,z, =𝒌𝒌\mathcal{H}=\sum_{\bm{k}}\mathcal{H}_{\bm{k}} and β\beta is the inverse temperature. By means of a sort of high-temperature expansion, the ssth order expansion coefficient of the μ\mu-component, gsμ(𝒌)g_{s}^{\mu}(\bm{k}), gives the corresponding effective multipole coupling as gsμ(𝒌)σμ/2g_{s}^{\mu}(\bm{k})\sigma_{\mu}/2. As the Hamiltonian in Eq. (46) consists of the cluster and bond multipoles in the matrix form, the ssth order expansion of eβ𝒌e^{-\beta\mathcal{H}_{\bm{k}}} can be described by the product of the ss-tuple of the matrices, Qlm(b)Q^{\rm(b)}_{lm}, Tlm(b)T^{\rm(b)}_{lm}, and Qlm(c)Q^{\rm(c)}_{lm}. It is noted that the 𝒌\bm{k} dependence arises from the momentum multipoles fQlm(𝒌)f^{Q_{lm}}(\bm{k}) and fTlm(𝒌)f^{T_{lm}}(\bm{k}), which are always coupled with the bond multipoles in the scalar form as Eqs. (47) and (48). This analysis can be applied to not only 𝑸=𝟎\bm{Q}=\bm{0} orderings but also finite commensurate 𝑸\bm{Q} orderings by choosing the appropriate minimal magnetic unit cell.

We present microscopic conditions for the band deformations from a multipole viewpoint in the cases of symmetric spin splitting, antisymmetric spin splitting, and antisymmetric band deformation with spin degeneracy in Secs. V.2.1, V.2.2, and V.2.3, respectively.

V.2.1 Symmetric spin splitting

The symmetric spin splitting, gsμ(𝒌)=gsμ(𝒌)g_{s}^{\mu}(\bm{k})=g_{s}^{\mu}(-\bm{k}) for μ=x,y,z\mu=x,y,z, occurs under the presence of the spatial inversion symmetry and the absence of the time-reversal symmetry. The conditions for the symmetric spin splitting are obtained by considering the space-time inversion properties (𝒫,𝒯\mathcal{P},\mathcal{T}); the product of ss-tuple of multipoles, which consists of the coupling between the bond and cluster multipoles, must have the same parities as those of the symmetric spin splitting, i.e., (𝒫,𝒯)=(+1,1)(\mathcal{P},\mathcal{T})=(+1,-1).

Since the bond multipoles consist of electric multipoles with (𝒫,𝒯)=(+1,+1)(\mathcal{P},\mathcal{T})=(+1,+1) and magnetic toroidal multipoles with (𝒫,𝒯)=(1,1)(\mathcal{P},\mathcal{T})=(-1,-1), while the cluster multipoles (electric multipoles) coupled with spin with (𝒫,𝒯)=(+1,1)(\mathcal{P},\mathcal{T})=(+1,-1), we obtain the following conditions for the product of ss-tuple of multipoles to realize the symmetric spin splitting:

  1. (i)

    Bond electric multipoles or even number of bond magnetic toroidal multipoles are involved.

  2. (ii)

    Odd number of cluster electric multipoles are involved.

  3. (iii)

    Trace of the sublattice degree of freedom (product of cluster multipoles) remains finite.

The conditions (i) and (ii) are required from the symmetry of the symmetric spin-split band dispersions, as mentioned above. The condition (ii) indicates that only the symmetric spin splitting occurs in collinear magnets. From the condition (iii), one can find that the symmetric spin splitting can occur when HtQH^{Q}_{t} and HmH_{m} contain the same symmetry of electric multipoles, while the term HtTH^{T}_{t} is not necessary. In other words, the momentum electric multipole can be coupled with σ\sigma through the higher-order coupling between the bond electric multipoles and cluster electric multipoles, which is necessary to yield the symmetric spin splitting.

Let us look at the example in an AFM with the collinear order parameter lmhlmzQlm(c)σz\sum_{lm}h_{lm}^{z}Q^{\rm(c)}_{lm}\sigma_{z}. The lowest-order contribution to Eq. (49) arises from the third order, which is proportional to

Tr[{Ht,{Ht,Hm}}σμ]mlmμTr[{Ht,{Ht,Qlm(c)}}],\displaystyle{\rm Tr}\left[\left\{H_{t},\left\{H_{t},H_{m}\right\}\right\}\sigma_{\mu}\right]\propto m_{lm}^{\mu}{\rm Tr}\left[\left\{H_{t},\left\{H_{t},Q_{lm}^{\rm(c)}\right\}\right\}\right], (50)

where {}\{\cdots\} is the anticommutator and Ht=HtQ+HtTH_{t}=H_{t}^{Q}+H_{t}^{T}. Here, the nonzero anticommutator between HtQH^{Q}_{t} (or HtTH^{T}_{t}) and HmH_{m} is essential to give rise to the spin-split band structure, since it gives a nontrivial coupling between the kinetic motions of electrons and the spin textures instead of the SOC. We can use the following relations among Qlm(b)Q^{\rm(b)}_{lm}, Tlm(b)T^{\rm(b)}_{lm}, and Qlm(c)Q^{\rm(c)}_{lm},

{Qlm(b),Ql′′m′′(c)}\displaystyle\left\{Q^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(c)}_{l^{\prime\prime}m^{\prime\prime}}\right\} =lmclmQlm(b),\displaystyle=\sum_{lm}c_{lm}Q^{\rm(b)}_{lm},
{Tlm(b),Ql′′m′′(c)}\displaystyle\left\{T^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(c)}_{l^{\prime\prime}m^{\prime\prime}}\right\} =lmclmTlm(b),\displaystyle=\sum_{lm}c_{lm}T^{\rm(b)}_{lm},
{Qlm(b),Ql′′m′′(b)}\displaystyle\left\{Q^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(b)}_{l^{\prime\prime}m^{\prime\prime}}\right\} =lmclmQlm(b)+lmclmQlm(c),\displaystyle=\sum_{lm}c_{lm}Q^{\rm(b)}_{lm}+\sum_{lm}c^{\prime}_{lm}Q^{\rm(c)}_{lm},
{Tlm(b),Tl′′m′′(b)}\displaystyle\left\{T^{\rm(b)}_{l^{\prime}m^{\prime}},T^{\rm(b)}_{l^{\prime\prime}m^{\prime\prime}}\right\} =lmclmQlm(b)+lmclmQlm(c),\displaystyle=\sum_{lm}c_{lm}Q^{\rm(b)}_{lm}+\sum_{lm}c^{\prime}_{lm}Q^{\rm(c)}_{lm},
{Qlm(b),Tl′′m′′(b)}\displaystyle\left\{Q^{\rm(b)}_{l^{\prime}m^{\prime}},T^{\rm(b)}_{l^{\prime\prime}m^{\prime\prime}}\right\} =lmclmTlm(b),\displaystyle=\sum_{lm}c_{lm}T^{\rm(b)}_{lm}, (51)

where clmc_{lm} and clmc^{\prime}_{lm} are expansion coefficients. These expressions are obtained from the comparison of the spatial and time-reversal parities of electric and magnetic toroidal multipoles in both sides. We omit the indices lml^{\prime}m^{\prime} and l′′m′′l^{\prime\prime}m^{\prime\prime} of clmc_{lm} and clmc^{\prime}_{lm} for notational simplicity. By using the first and third relations in Eq. (V.2.1), one can easily find that

gsμ(𝒌)mlmμfQ0(𝒌)fQlm(𝒌)\displaystyle g_{s}^{\mu}(\bm{k})\sim m_{lm}^{\mu}f^{Q_{0}}(\bm{k})f^{Q_{l^{\prime}m^{\prime}}}(\bm{k})
×Tr[{Q0(b),{Qlm(b),Qlm(c)}}]\displaystyle\quad\times{\rm Tr}\left[\left\{Q^{\rm(b)}_{0},\left\{Q^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(c)}_{lm}\right\}\right\}\right]

becomes nonzero only when HtH_{t} and HmH_{m} contain the same symmetry of multipole, i.e., l=ll^{\prime}=l, m=mm^{\prime}=m. Then, the functional form of the spin splitting is given by fQ0(𝒌)fQlm(𝒌)Qlm(m)(𝒌)f^{Q_{0}}(\bm{k})f^{Q_{lm}}(\bm{k})\sim Q^{\rm(m)}_{lm}(\bm{k}). In other words, the functional form of the spin splitting is characterized by the higher-rank momentum electric multipole Qlm(m)(𝒌)Q_{lm}^{\rm(m)}(\bm{k}). Note that the bond magnetic toroidal multipoles can also contribute to the spin splitting by the effective coupling as mlmμTr[{Tlm(b),{Tl′′m′′(b),Qlm(c)}}]m_{lm}^{\mu}{\rm Tr}\left[\left\{T^{\rm(b)}_{l^{\prime}m^{\prime}},\left\{T^{\rm(b)}_{l^{\prime\prime}m^{\prime\prime}},Q^{\rm(c)}_{lm}\right\}\right\}\right] when Tlm(b)Tl′′m′′(b)T^{\rm(b)}_{l^{\prime}m^{\prime}}T^{\rm(b)}_{l^{\prime\prime}m^{\prime\prime}} belongs to the same irrep. as Qlm(c)Q^{\rm(c)}_{lm}.

V.2.2 Antisymmetric spin splitting

In contrast to the symmetric spin splitting, the antisymmetric spin splitting, gsμ(𝒌)=gsμ(𝒌)g_{s}^{\mu}(\bm{k})=-g_{s}^{\mu}(-\bm{k}) for μ=x,y,z\mu=x,y,z, occurs only in noncollinear magnets. This is because in collinear magnets without the spin-orbit coupling the spin rotational operation [(𝒌,σ)(𝒌,σ)(\bm{k},\sigma)\to(\bm{k},-\sigma)] with combining the time-reversal operation [(𝒌,σ)(𝒌,σ)(\bm{k},\sigma)\to(-\bm{k},-\sigma)] ensures the spatial inversion symmetry [(𝒌,σ)(𝒌,σ)(\bm{k},\sigma)\to(-\bm{k},\sigma)Hayami et al. (2019a).

By similar argument as in the symmetric spin splitting, the conditions for the product of ss-tuple of multipoles are given as follows:

  1. (i)

    Odd number of bond magnetic toroidal multipoles are involved.

  2. (ii)

    At least, two spin components leading to noncollinear spin textures are involved.

  3. (iii)

    Trace of the sublattice degree of freedom (product of cluster multipoles) remains finite.

The conditions (i) and (ii) are required from the antisymmetric spin-split band dispersions under the breakings of spatial, time-reversal, and their product symmetries, which are obtained from the similar analysis in Sec. V.2.1. From the above conditions, the emergence of the antisymmetric spin splittings are due to the effective coupling between the bond magnetic toroidal multipoles and cluster electric multipoles.

Let us look at the example in a noncollinear AFM with the order parameter lm(hlmxQlm(c)σx+hlmyQlm(c)σy)\sum_{lm}(h_{lm}^{x}Q_{lm}^{\rm(c)}\sigma_{x}+h_{lm}^{y}Q_{lm}^{\rm(c)}\sigma_{y}). One of the contributions comes from the fifth order, which is proportional to

Tr[{Ht,{{Ht,Hm},{Ht,Hm}}}σz].\displaystyle{\rm Tr}[\left\{H_{t},\left\{\left\{H_{t},H_{m}\right\},\left\{H_{t},H_{m}\right\}\right\}\right\}\sigma_{z}]. (52)

In contrast to Eq. (V.2.1) where the spin is simply traced out leaving the anticommutator without spin dependence, the plural HmH_{m} terms depending on different component of spins are involved in this case. All the necessary anticommutator appearing in Eq. (49) are represented in the form,

{Xlmσμ,Ylmσν}\displaystyle\left\{X_{lm}\sigma_{\mu},Y_{l^{\prime}m^{\prime}}\sigma_{\nu}\right\} ={Xlm,Ylm}δμ,νσ0\displaystyle=\left\{X_{lm},Y_{l^{\prime}m^{\prime}}\right\}\delta_{\mu,\nu}\sigma_{0}
+i[Xlm,Ylm]κεμνκσκ,\displaystyle+i\left[X_{lm},Y_{l^{\prime}m^{\prime}}\right]\sum_{\kappa}\varepsilon_{\mu\nu\kappa}\sigma_{\kappa}, (53)

where [][\cdots] is the commutator, and i[Xlm,Ylm]i\left[X_{lm},Y_{l^{\prime}m^{\prime}}\right] is the hermite matrix. From the fact that the imaginary unit ii represents the time-reversal-odd scalar and the antisymmetric tensor εμνκ\varepsilon_{\mu\nu\kappa} changes the sign of the spatial parity, the commutation relation is given as follows:

i[Qlm(c),Ql′′m′′(c)]\displaystyle i\left[Q^{\rm(c)}_{l^{\prime}m^{\prime}},Q^{\rm(c)}_{l^{\prime\prime}m^{\prime\prime}}\right] =0,\displaystyle=0,
i[Qlm(b),Ql′′m′′(c)]\displaystyle i\left[Q^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(c)}_{l^{\prime\prime}m^{\prime\prime}}\right] =lmclmTlm(b),\displaystyle=\sum_{lm}c_{lm}T^{\rm(b)}_{lm},
i[Tlm(b),Ql′′m′′(c)]\displaystyle i\left[T^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(c)}_{l^{\prime\prime}m^{\prime\prime}}\right] =lmclmQlm(b),\displaystyle=\sum_{lm}c_{lm}Q^{\rm(b)}_{lm},
i[Qlm(b),Ql′′m′′(b)]\displaystyle i\left[Q^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(b)}_{l^{\prime\prime}m^{\prime\prime}}\right] =lmclmTlm(b),\displaystyle=\sum_{lm}c_{lm}T^{\rm(b)}_{lm},
i[Tlm(b),Tl′′m′′(b)]\displaystyle i\left[T^{\rm(b)}_{l^{\prime}m^{\prime}},T^{\rm(b)}_{l^{\prime\prime}m^{\prime\prime}}\right] =lmclmTlm(b),\displaystyle=\sum_{lm}c_{lm}T^{\rm(b)}_{lm},
i[Qlm(b),Tl′′m′′(b)]\displaystyle i\left[Q^{\rm(b)}_{l^{\prime}m^{\prime}},T^{\rm(b)}_{l^{\prime\prime}m^{\prime\prime}}\right] =lmclmQlm(b)+lmclmQlm(c),\displaystyle=\sum_{lm}c_{lm}Q^{\rm(b)}_{lm}+\sum_{lm}c^{\prime}_{lm}Q^{\rm(c)}_{lm}, (54)

where clmc_{lm} and clmc^{\prime}_{lm} are expansion coefficients where we again omit their indices, lml^{\prime}m^{\prime} and l′′m′′l^{\prime\prime}m^{\prime\prime}. By using the first and second relations in Eq. (V.2.1) and the sixth relation in Eq. (V.2.2), we find that

Tr[{Q0(b),{{Q0(b),Qlm(c)σx},{Tlm(b),Ql′′m′′(c)σy}}}σz]\displaystyle{\rm Tr}\left[\left\{Q^{\rm(b)}_{0},\left\{\left\{Q^{\rm(b)}_{0},Q^{\rm(c)}_{lm}\sigma_{x}\right\},\left\{T^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(c)}_{l^{\prime\prime}m^{\prime\prime}}\sigma_{y}\right\}\right\}\right\}\sigma_{z}\right]

becomes nonzero for the σz\sigma_{z} component perpendicular to the coplanar magnetic moments. The functional form of the spin splitting is then given by the active magnetic toroidal multipole fTlm(𝒌)(fQ0(𝒌))2Tlm(m)(𝒌)f^{T_{l^{\prime}m^{\prime}}}(\bm{k})(f^{Q_{0}}(\bm{k}))^{2}\sim T^{\rm(m)}_{l^{\prime}m^{\prime}}(\bm{k}). Note that the other multipole coupling can also contribute to the spin splitting, e.g.,

Tr[Tl′′′m′′′(b),{{Tl′′′′m′′′′(b),Qlm(c)σx},{Tlm(b),Ql′′m′′(c)σy}}σz]\displaystyle{\rm Tr}\left[T^{\rm(b)}_{l^{\prime\prime\prime}m^{\prime\prime\prime}},\left\{\left\{T^{\rm(b)}_{l^{\prime\prime\prime\prime}m^{\prime\prime\prime\prime}},Q^{\rm(c)}_{lm}\sigma_{x}\right\},\left\{T^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(c)}_{l^{\prime\prime}m^{\prime\prime}}\sigma_{y}\right\}\right\}\sigma_{z}\right]

as long as the quantity remains finite after tracing them out.

V.2.3 Antisymmetric band deformation with spin degeneracy

Finally, we discuss the antisymmetric band deformation with spin degeneracy, gs0(𝒌)=gs0(𝒌)g_{s}^{0}(\bm{k})=-g_{s}^{0}(-\bm{k}), which occurs in noncoplanar magnets without spatial inversion symmetry. The conditions for the effective multipole couplings are given as follows:

  1. (i)

    Odd number of bond magnetic toroidal multipoles are involved.

  2. (ii)

    Three spin components, which are necessary to represent noncoplanar spin structures, are involved.

  3. (iii)

    Trace of the sublattice and spin degrees of freedom remains finite.

The conditions (i) and (ii) are required to satisfy the symmetry for the antisymmetric band deformations. The condition (iii) indicates that the spin dependence is not important.

We show the example in a noncoplanar AFM with the order parameter lm(hlmxQlm(c)σx+hlmyQlm(c)σy+hlmzQlm(c)σz)\sum_{lm}(h_{lm}^{x}Q_{lm}^{\rm(c)}\sigma_{x}+h_{lm}^{y}Q_{lm}^{\rm(c)}\sigma_{y}+h_{lm}^{z}Q_{lm}^{\rm(c)}\sigma_{z}). One of the six-order contributions to the antisymmetric band deformation is given by

Tr[{{Ht,Hm},{{Ht,Hm},{Ht,Hm}}}].\displaystyle{\rm Tr}[\left\{\left\{H_{t},H_{m}\right\},\left\{\left\{H_{t},H_{m}\right\},\left\{H_{t},H_{m}\right\}\right\}\right\}]. (55)

By using Eq. (V.2.2), the contribution

Tr[{{Q0(b),Qlm(c)σz},{{Q0(b),Qlm(c)σx},{Tlm(b),Ql′′m′′(c)σy}}}]\displaystyle{\rm Tr}\left[\left\{\left\{Q^{\rm(b)}_{0},Q^{\rm(c)}_{lm}\sigma_{z}\right\},\left\{\left\{Q^{\rm(b)}_{0},Q^{\rm(c)}_{lm}\sigma_{x}\right\},\left\{T^{\rm(b)}_{l^{\prime}m^{\prime}},Q^{\rm(c)}_{l^{\prime\prime}m^{\prime\prime}}\sigma_{y}\right\}\right\}\right\}\right]

can remain finite. Although the noncoplanar magnets are rare as compared to the coplanar magnets, the antisymmetric band deformations can also be realized by applying the magnetic field to the coplanar AFMs without the spatial inversion symmetry in the out-of-plane-moment direction Hayami et al. (2020b).

We summarize the functional form of the band deformations and related magnetic textures in Table 2.

Table 2: Three types of band deformations and their functional forms: symmetric spin splitting (SS), antisymmetric spin splitting, and antisymmetric band deformation (BD). The necessary conditions of space-time parities of the system and magnetic textures are also shown.
Type form 𝒫\mathcal{P} 𝒫𝒯\mathcal{PT} magnetic textures
Symmetric SS fQlm(𝒌)σμf^{Q_{lm}}(\bm{k})\sigma_{\mu} \circ ×\times collinear
Antisymmetric SS fTlm(𝒌)σμf^{T_{lm}}(\bm{k})\sigma_{\mu} ×\times ×\times coplanar
Antisymmetric BD fTlm(𝒌)f^{T_{lm}}(\bm{k}) ×\times \circ noncoplanar

VI Application to triangular lattice systems

We apply the present scheme to specific lattice systems. We take three examples consisting of a triangle cluster: triangular, kagome, and breathing kagome structures. After introducing multipole degrees of freedom in the triangle cluster in Sec. VI.1, we show that spin splittings and band deformations are induced by the 120 AFM ordering on three specific lattices. We present the antisymmetric spin splitting on a triangular lattice in Sec. VI.2, the symmetric spin splitting on a kagome lattice in Sec. VI.3, and symmetric and antisymmetric spin splittings on a breathing kagome lattice in Sec. VI.4. We also show the effect of an external magnetic field on the noncollinear ordering on a breathing kagome lattice in Sec. VI.5.

VI.1 Triangle cluster

Refer to caption
Figure 6: Cluster and bond multipoles in a triangle cluster Hayami et al. (2020b). The correspondence between multipoles and matrix elements is shown. The red (blue) circles represent the positive (negative) onsite potential, and the red (blue) lines and orange arrows on each bond represent the positive (negative) real and imaginary hoppings, respectively. The gray lines represent no hoppings.

We consider a triangle cluster whose sublattice basis function consists of (|A,|B,|C)(\ket{{\rm A}},\ket{{\rm B}},\ket{{\rm C}}) as shown in Fig. 6. This cluster belongs to the point group D3hD_{3{\rm h}} and has nine multipole degrees of freedom. From the irreducible decomposition in Table 1 and corresponding multipole table in Table 15, three onsite degrees of freedom with A1+E+A^{\prime+}_{1}\oplus E^{\prime+} correspond to Q0(c)Q_{0}^{\rm(c)}, Qv(c)Q_{v}^{\rm(c)}, and Qxy(c)Q_{xy}^{\rm(c)}, three real bond degrees of freedom with A1+E+A^{\prime+}_{1}\oplus E^{\prime+} correspond to Q0(b)Q_{0}^{\rm(b)}, Qv(b)Q_{v}^{\rm(b)}, and Qxy(b)Q_{xy}^{\rm(b)}, and three imaginary bond degrees of freedom with A2EA^{\prime-}_{2}\oplus E^{\prime-} correspond to T3a(b)T_{3a}^{\rm(b)}, Tx(b)T_{x}^{\rm(b)}, and Ty(b)T_{y}^{\rm(b)}. It is noted that there are two settings in choosing the C2C^{\prime}_{2} rotational axis in D3hD_{3{\rm h}}. Here, we take the yy axis as the C2C^{\prime}_{2} rotational axis (See the column D3hD_{3{\rm h}} in Table 15).

The specific matrix elements for each multipole are shown in Fig. 6 com (a). In the following sections, we assume the noncollinear 120 AFM magnetic structure with the form of Qxy(c)σx+Qv(c)σyQ^{\rm(c)}_{xy}\sigma_{x}+Q^{\rm(c)}_{v}\sigma_{y} on the triangular, kagome, and breathing kagome structures. We implicitly assume that the spin rotational symmetry is spontaneously broken through the phase transition.

VI.2 Triangular

Refer to caption
Figure 7: Schematic pictures of the 120 AFM on a triangular lattice. The active multipoles are also shown.
Refer to caption
Figure 8: (Left panel) The band structure of the model on the triangular lattice at (a) m=0.2m=0.2 and (b) m=6m=6. The other model parameter is ta=1t_{a}=1. The dashed lines show the band dispersions and the color map shows the spin polarization of the zz component at each wave vector. (Right panel) The isoenergy surfaces at μ=2.5\mu=-2.5 and μ=6.5\mu=-6.5. The hexagon in the right panel represents the magnetic first Brillouin zone.

First, we consider the triangular lattice with the lattice constant aa, as shown in Fig. 7. It is noted that the symmetry of the triangular lattice D6hD_{6{\rm h}} is different from that of the triangle cluster D3hD_{3{\rm h}}, both of which belong to the same Laue class 6/mmm6/mmm, as shown in Table 15. In this case, from compatibility relation from D3hD_{\rm 3h} to D6hD_{\rm 6h}, the irrep. should be replaced as A1+A1g+A^{\prime+}_{1}\to A^{+}_{1g}, A1′′A1uA^{\prime\prime-}_{1}\to A_{1u}^{-}, and so on. Meanwhile, by looking the correspondence between the multipoles and the irrep. in the same row in Table 15, one can find that the same multipole notations, e.g., Q4aQ_{4a}, QzxQ_{zx}, are used for cluster and lattice systems.

The matrices of the hopping and mean-field Hamiltonians in the three-sublattice triangular system are given by

HtQ=fQ0(𝒌)Q0(b),\displaystyle H^{Q}_{t}=f^{Q_{0}}(\bm{k})Q^{\rm(b)}_{0}, (56)
HtT=fT3a(𝒌)T3a(b),\displaystyle H^{T}_{t}=f^{T_{3a}}(\bm{k})T^{\rm(b)}_{3a}, (57)
Hm=m(Qxy(c)σx+Qv(c)σy),\displaystyle H_{m}=-m(Q^{\rm(c)}_{xy}\sigma_{x}+Q^{\rm(c)}_{v}\sigma_{y}), (58)

where the form factors for the nearest-neighbor site are represented by

fQ0(𝒌)\displaystyle f^{Q_{0}}(\bm{k}) =6ta(coskxa+2cosk~xacosk~ya),\displaystyle=\sqrt{6}t_{a}(\cos k_{x}a+2\cos\tilde{k}_{x}a\cos\tilde{k}_{y}a),
fT3a(𝒌)\displaystyle f^{T_{3a}}(\bm{k}) =6ta(sinkxa2sink~xacosk~ya),\displaystyle=-\sqrt{6}t_{a}(\sin k_{x}a-2\sin\tilde{k}_{x}a\cos\tilde{k}_{y}a), (59)

with the hopping amplitude tat_{a}. Here and hereafter, we use the abbreviated notations k~x=kx/2\tilde{k}_{x}=k_{x}/2 and k~y=3ky/2\tilde{k}_{y}=\sqrt{3}k_{y}/2. We consider the first-neighbor hopping in Eq. (58), which is expressed by the electric monopole and magnetic toroidal octupole degrees of freedom. The presence of magnetic toroidal multipole, T3a(b)T^{\rm(b)}_{3a}, is attributed to the introduction of the sublattice degree of freedom by taking into account the magnetic unit cell, and it does not exist in the case of a single-site unit cell. As we will show below, T3a(b)T^{\rm(b)}_{3a} plays an important role for the emergent spin splitting as a result from the coupling with the noncollinear three-sublattice magnetic structures. The mean-field matrix HmH_{m} consists of two spin components to express the 120120^{\circ} noncollinear magnetic order with the amplitude mm.

It is noted that the active bond multipoles appearing in the hopping matrices, HtQH_{t}^{Q} and HtTH_{t}^{T} depend on the nature of hopping and the choice of the magnetic unit cell. For example, the further neighbor hoppings may bring about the other types of electric and magnetic toroidal multipoles, as shown in Table 1. Nevertheless, in the present triangular-lattice case, the further neighbor hoppings do not give rise to the other multipoles due to the lattice symmetry. Thus, the symmetric spin splitting does not appear even by taking account of further neighbor hoppings due to the lack of higher-rank electric multipoles in HtQH_{t}^{Q}. On the other hand, the antisymmetric spin splitting can occur according to the conditions given in Sec. V.2.2. The lowest-order contribution is given by

g5z(𝒌)=1323m2fT3a(𝒌)[(fT3a(𝒌))23(fQ0(𝒌))2].\displaystyle g^{z}_{5}(\bm{k})=-\frac{1}{3}\sqrt{\frac{2}{3}}m^{2}f^{T_{3a}}(\bm{k})\left[(f^{T_{3a}}(\bm{k}))^{2}-3(f^{Q_{0}}(\bm{k}))^{2}\right]. (60)

As fT3a(𝒌)kx(kx23ky2)f^{T_{3a}}(\bm{k})\propto k_{x}(k_{x}^{2}-3k_{y}^{2}) and fQ0(𝒌)1f^{Q_{0}}(\bm{k})\propto 1 in the 𝒌𝟎\bm{k}\to\bm{0} limit, the essential anisotropy is given by

g5z(𝒌)\displaystyle g^{z}_{5}(\bm{k}) 23m2(fQ0(𝒌))2fT3a(𝒌)\displaystyle\simeq\sqrt{\frac{2}{3}}m^{2}(f^{Q_{0}}(\bm{k}))^{2}f^{T_{3a}}(\bm{k})
=24m2ta3sink~x(cosk~ycosk~x)\displaystyle=24m^{2}t_{a}^{3}\sin\tilde{k}_{x}\left(\cos\tilde{k}_{y}-\cos\tilde{k}_{x}\right)
×(2cosk~xcosk~y+coskx)2\displaystyle\ \ \ \times\left(2\cos\tilde{k}_{x}\cos\tilde{k}_{y}+\cos k_{x}\right)^{2}
272m2ta3kx(kx23ky2)a3.\displaystyle\simeq\frac{27}{2}m^{2}t_{a}^{3}k_{x}\left(k_{x}^{2}-3k_{y}^{2}\right)a^{3}. (61)

In this way, the functional form of the antisymmetric spin splitting satisfying the magnetic space group symmetry is obtained from the effective multipole coupling. Moreover, one can obtain the model parameter dependence for the spin splitting. As is consistent with the discussion in Sec. V.2.2, the expressions in Eq. (VI.2) contain the product of the even number of order parameters as m2m^{2} and the bond magnetic toroidal multipole T3a(m)(𝒌)T_{3a}^{\rm(m)}(\bm{k}). The opposite spin alignment is realized by reversing the vector spin chirality; the sign of one spin components in HmH_{m} is reversed as Hm=m(Qxy(c)σxQv(c)σy)H_{m}=-m(Q^{\rm(c)}_{xy}\sigma_{x}-Q^{\rm(c)}_{v}\sigma_{y}). This is consistent with the analysis in Eq. (52), which results in opposite sign to Eq. (VI.2).

The effective multipole coupling leads to physical phenomena related with the inversion symmetry breaking Hayami et al. (2020b). For example, the active magnetic toroidal multipoles in the form of T3a(m)(𝒌)σzkx(kx23ky2)σzT^{(\rm m)}_{3a}(\bm{k})\sigma_{z}\sim k_{x}(k_{x}^{2}-3k_{y}^{2})\sigma_{z} in Eq. (VI.2), implies that a spontaneous threefold rotational nonreciprocity is induced by a magnetic field along the zz direction if one divides it as kx(kx23ky2)×σzk_{x}(k_{x}^{2}-3k_{y}^{2})\times\sigma_{z}.

The emergent antisymmetric spin splitting is confirmed by diagonalizing the Hamiltonian. We show the electronic band structure in Figs. 8(a) and 8(b). The result clearly shows the spin splitting along the M1M_{1}-Γ\Gamma-M2M_{2} line, while there is no spin splitting along the KK-Γ\Gamma-KK^{\prime} line irrespective of the value of mm, which is consistent with Eq. (VI.2).

VI.3 Kagome

Refer to caption
Figure 9: Schematic pictures of the 120 AFM on a kagome lattice. The active multipoles are also shown.
Refer to caption
Figure 10: (Left panel) The band structure of the model on the kagome lattice at ta=1t_{a}=1 and m=1m=1. The dashed lines show the band dispersions and the color map shows the spin polarization of the (a) xx and (b) yy components at each wave vector. (Right panel) The isoenergy surfaces at μ=1\mu=-1. The hexagon in the right panel represents the magnetic first Brillouin zone.

Next, we consider the 120 AFM on the kagome lattice with the lattice constant 2a2a, as shown in Fig. 9. The point group of the kagome structure is D6hD_{6{\rm h}}, which is the same as that of the triangular lattice. Owing to the different lattice geometry from the triangular lattice in the previous section, the different bond and momentum multipoles appear, as will be shown below.

The matrices of the hopping and mean-field Hamiltonians in the three-sublattice kagome system are given by

HtQ=fQ0(𝒌)Q0(b)+fQv(𝒌)Qv(b)+fQxy(𝒌)Qxy(b),\displaystyle H^{Q}_{t}=f^{Q_{0}}(\bm{k})Q^{\rm(b)}_{0}+f^{Q_{v}}(\bm{k})Q^{\rm(b)}_{v}+f^{Q_{xy}}(\bm{k})Q^{\rm(b)}_{xy}, (62)
HtT=0,\displaystyle H^{T}_{t}=0, (63)
Hm=m(Qxy(c)σx+Qv(c)σy),\displaystyle H_{m}=-m(Q^{\rm(c)}_{xy}\sigma_{x}+Q^{\rm(c)}_{v}\sigma_{y}), (64)

where the form factors for the nearest-neighbor site are represented by

fQ0(𝒌)\displaystyle f^{Q_{0}}(\bm{k}) =223ta(coskxa+2cosk~xacosk~ya),\displaystyle=2\sqrt{\frac{2}{3}}t_{a}(\cos k_{x}a+2\cos\tilde{k}_{x}a\cos\tilde{k}_{y}a),
fQv(𝒌)\displaystyle f^{Q_{v}}(\bm{k}) =43ta(cosk~xacosk~yacoskxa),\displaystyle=\frac{4}{\sqrt{3}}t_{a}(\cos\tilde{k}_{x}a\cos\tilde{k}_{y}a-\cos k_{x}a),
fQxy(𝒌)\displaystyle f^{Q_{xy}}(\bm{k}) =4tasink~xasink~ya,\displaystyle=4t_{a}\sin\tilde{k}_{x}a\sin\tilde{k}_{y}a, (65)

with the hopping amplitude tat_{a}. There are two differences from the case in the triangular lattice in Sec. VI.2: One is that the higher-rank electric multipoles are present, which indicates that the symmetric spin splitting can occur. The other is that there are no magnetic toroidal multipoles in the hopping matrix, since all the sublattice sites have the local inversion symmetry. Thus, any antisymmetric band deformations do not occur within the three-sublattice ordering in the kagome structure. We consider the 120AFM structure where the mean-field matrix HmH_{m} is the same as that in the case of triangular lattice.

The symmetric spin splitting due to the presence of Qv(b)Q^{\rm(b)}_{v} and Qxy(b)Q^{\rm(b)}_{xy} is given by

g3x(𝒌)\displaystyle g^{x}_{3}(\bm{k}) =m(2fQ0(𝒌)fQxy(𝒌)+2fQv(𝒌)fQxy(𝒌)),\displaystyle=m\left(2f^{Q_{0}}(\bm{k})f^{Q_{xy}}(\bm{k})+\sqrt{2}f^{Q_{v}}(\bm{k})f^{Q_{xy}}(\bm{k})\right), (66)
g3y(𝒌)\displaystyle g^{y}_{3}(\bm{k}) =m[2fQ0(𝒌)fQv(𝒌)\displaystyle=m\left[2f^{Q_{0}}(\bm{k})f^{Q_{v}}(\bm{k})\right.
12{(fQv(𝒌))2(fQxy(𝒌))2}].\displaystyle\ \ \ \left.-\frac{1}{\sqrt{2}}\left\{(f^{Q_{v}}(\bm{k}))^{2}-(f^{Q_{xy}}(\bm{k}))^{2}\right\}\right]. (67)

As fQ0(𝒌)1f^{Q_{0}}(\bm{k})\propto 1, fQv(𝒌)kx2ky2f^{Q_{v}}(\bm{k})\propto k_{x}^{2}-k_{y}^{2}, and fQxy(𝒌)kxkyf^{Q_{xy}}(\bm{k})\propto k_{x}k_{y} in the 𝒌𝟎\bm{k}\to\bm{0} limit, the essential anisotropy is given by

g3x(𝒌)\displaystyle g^{x}_{3}(\bm{k}) 2mfQ0(𝒌)fQxy(𝒌)122mta2kxkya2.\displaystyle\simeq 2mf^{Q_{0}}(\bm{k})f^{Q_{xy}}(\bm{k})\simeq 12\sqrt{2}mt_{a}^{2}k_{x}k_{y}a^{2}. (68)
g3y(𝒌)\displaystyle g^{y}_{3}(\bm{k}) 2mfQ0(𝒌)fQv(𝒌)62mta2(kx2ky2)a2.\displaystyle\simeq 2mf^{Q_{0}}(\bm{k})f^{Q_{v}}(\bm{k})\simeq 6\sqrt{2}mt_{a}^{2}\left(k_{x}^{2}-k_{y}^{2}\right)a^{2}. (69)

In contrast to the antisymmetric spin splitting in Eq. (VI.2), g3x(𝒌)g^{x}_{3}(\bm{k}) and g3y(𝒌)g^{y}_{3}(\bm{k}) are proportional to mm, which implies that the one spin component, i.e., the collinear spin structure, is sufficient to realize the symmetric spin splitting, as discussed in Sec. V.2.1. In fact, when we switch off one of the order parameters mQv(c)=0mQ^{\rm(c)}_{v}=0 or mQxy(c)=0mQ^{\rm(c)}_{xy}=0, g3x(𝒌)g^{x}_{3}(\bm{k}) or g3y(𝒌)g^{y}_{3}(\bm{k}) remains finite, i.e., the symmetric spin splittings for xx and yy spin components are independent with each other. Moreover, one can find the opposite direction of the AFM moment results in the opposite spin splittings.

The symmetric spin splitting as a result of the effective multipole coupling affects physical response tensors Hayami et al. (2018). For example, from the coupling between σx\sigma_{x} and Qxy(𝒌)kxkyQ_{xy}(\bm{k})\sim k_{x}k_{y} in Eq. (66), we can expect the magneto-elastic effect where a spontaneous xyxy-type shear stress is induced by a magnetic field along the xx direction or the spin-current generation where the spin current along the xx direction with the xx-spin component is generated by an electric field along the yy direction Naka et al. (2019); Hayami et al. (2019a).

The above analysis for the symmetric spin splitting is confirmed by calculating the electronic band structure. Figures 10(a) and (b) show the band structure at ta=1t_{a}=1 and m=1m=1 where the color map shows the spin polarization for the xx and yy components, respectively. The result clearly shows that the spin splittings are symmetric with respect to 𝒌\bm{k} and their functional forms are characterized by Qxy(m)(𝒌)σxQ^{\rm(m)}_{xy}(\bm{k})\sigma_{x} in Fig. 10(a) and Qv(m)(𝒌)σyQ^{\rm(m)}_{v}(\bm{k})\sigma_{y} in Fig. 10(b).

VI.4 Breathing kagome

Refer to caption
Figure 11: Schematic pictures of the 120 AFM on a breathing kagome lattice. The active multipoles are also shown.
Refer to caption
Figure 12: (Left panel) The band structure of the model on the breathing kagome lattice at ta=1t_{a}=1, tb=0.5t_{b}=0.5, and m=1m=1. The dashed lines show the band dispersions and the color map shows the spin polarization of the (a) xx, (b) yy, and zz components at each wave vector. (Right panel) The isoenergy surfaces at μ=1\mu=-1. The hexagon in the right panel represents the magnetic first Brillouin zone.

The last example is the 120 AFM on the breathing kagome lattice with the lattice constant a+ba+b, as shown in Fig. 11. The point group of the breathing kagome structure is D3hD_{3{\rm h}}, which is the same as that of the triangular cluster.

The matrices of the hopping and mean-field Hamiltonians in the three-sublattice breathing kagome system are given by

HtQ=fQ0(𝒌)Q0(b)+fQv(𝒌)Qv(b)+fQxy(𝒌)Qxy(b),\displaystyle H^{Q}_{t}=f^{Q_{0}}(\bm{k})Q_{0}^{\rm(b)}+f^{Q_{v}}(\bm{k})Q_{v}^{\rm(b)}+f^{Q_{xy}}(\bm{k})Q_{xy}^{\rm(b)}, (70)
HtT=fT3a(𝒌)T3a(b)+fTx(𝒌)Tx(b)+fTy(𝒌)Ty(b),\displaystyle H^{T}_{t}=f^{T_{3a}}(\bm{k})T_{3a}^{\rm(b)}+f^{T_{x}}(\bm{k})T_{x}^{\rm(b)}+f^{T_{y}}(\bm{k})T_{y}^{\rm(b)}, (71)
Hm=m(Qxy(c)σx+Qv(c)σy),\displaystyle H_{m}=-m(Q_{xy}^{\rm(c)}\sigma_{x}+Q_{v}^{\rm(c)}\sigma_{y}), (72)

where the form factors are represented by

fQ0(𝒌)\displaystyle f^{Q_{0}}(\bm{k}) =23ηtη(coskxη+2cosk~xηcosk~yη),\displaystyle=\sqrt{\frac{2}{3}}\sum_{\eta}t_{\eta}(\cos k_{x}\eta+2\cos\tilde{k}_{x}\eta\cos\tilde{k}_{y}\eta),
fQv(𝒌)\displaystyle f^{Q_{v}}(\bm{k}) =23ηtη(cosk~xηcosk~yηcoskxη),\displaystyle=\frac{2}{\sqrt{3}}\sum_{\eta}t_{\eta}(\cos\tilde{k}_{x}\eta\cos\tilde{k}_{y}\eta-\cos k_{x}\eta),
fQxy(𝒌)\displaystyle f^{Q_{xy}}(\bm{k}) =2ηtηsink~xηsink~yη,\displaystyle=2\sum_{\eta}t_{\eta}\sin\tilde{k}_{x}\eta\sin\tilde{k}_{y}\eta,
fT3a(𝒌)\displaystyle f^{T_{3a}}(\bm{k}) =23ηpη(sinkxη2sink~xηcosk~yη),\displaystyle=-\sqrt{\frac{2}{3}}\sum_{\eta}p_{\eta}(\sin k_{x}\eta-2\sin\tilde{k}_{x}\eta\cos\tilde{k}_{y}\eta),
fTx(𝒌)\displaystyle f^{T_{x}}(\bm{k}) =23ηpη(sinkxη+sink~xηcosk~yη),\displaystyle=\frac{2}{\sqrt{3}}\sum_{\eta}p_{\eta}(\sin k_{x}\eta+\sin\tilde{k}_{x}\eta\cos\tilde{k}_{y}\eta),
fTy(𝒌)\displaystyle f^{T_{y}}(\bm{k}) =2ηpηcosk~xηsink~yη,\displaystyle=2\sum_{\eta}p_{\eta}\cos\tilde{k}_{x}\eta\sin\tilde{k}_{y}\eta, (73)

for η=a,b\eta=a,b, pa=tap_{a}=t_{a}, and pb=tbp_{b}=-t_{b}. The hopping amplitudes are defined as tat_{a} within upward triangles and tbt_{b} within downward triangles. The mean-field matrix HmH_{m} is the same as that in Eqs. (58) and (64).

Owing to the presence of bond electric and magnetic toroidal multipoles for l1l\geq 1, both the symmetric and antisymmetric spin splittings can occur. The lowest-order contribution to the symmetric spin splitting arises at s=3s=3 in Eq. (49) as

g3x(𝒌)\displaystyle g^{x}_{3}(\bm{k}) =m[2fQ0(𝒌)fQxy(𝒌)+2fQv(𝒌)fQxy(𝒌)\displaystyle=m\left[2f^{Q_{0}}(\bm{k})f^{Q_{xy}}(\bm{k})+\sqrt{2}f^{Q_{v}}(\bm{k})f^{Q_{xy}}(\bm{k})\right.
2fT3a(𝒌)fTy(𝒌)2fTx(𝒌)fTy(𝒌)],\displaystyle\ \ \ \left.-2f^{T_{3a}}(\bm{k})f^{T_{y}}(\bm{k})-\sqrt{2}f^{T_{x}}(\bm{k})f^{T_{y}}(\bm{k})\right], (74)
g3y(𝒌)\displaystyle g^{y}_{3}(\bm{k}) =m[2fQ0(𝒌)fQv(𝒌)12{(fQv(𝒌))2(fQxy(𝒌))2}\displaystyle=m\left[2f^{Q_{0}}(\bm{k})f^{Q_{v}}(\bm{k})-\frac{1}{\sqrt{2}}\left\{(f^{Q_{v}}(\bm{k}))^{2}-(f^{Q_{xy}}(\bm{k}))^{2}\right\}\right.
+2fT3a(𝒌)fTx(𝒌)12{(fTx(𝒌))2(fTy(𝒌))2}].\displaystyle\ \ \ \left.+2f^{T_{3a}}(\bm{k})f^{T_{x}}(\bm{k})-\frac{1}{\sqrt{2}}\left\{(f^{T_{x}}(\bm{k}))^{2}-(f^{T_{y}}(\bm{k}))^{2}\right\}\right]. (75)

It is easily confirmed that the effective multipole couplings with electric multipoles are the same as those in Eqs. (66) and (VI.3). There are additional effective multipole couplings with magnetic toroidal multipoles. When the limit of 𝒌𝟎\bm{k}\to\bm{0} is taken, the essential anisotropy is given by

g3x(𝒌)\displaystyle g^{x}_{3}(\bm{k}) m[2fQ0(𝒌)fQxy(𝒌)2fTx(𝒌)fTy(𝒌)]\displaystyle\simeq m\left[2f^{Q_{0}}(\bm{k})f^{Q_{xy}}(\bm{k})-\sqrt{2}f^{T_{x}}(\bm{k})f^{T_{y}}(\bm{k})\right]
62mtatb(a+b)2kxky,\displaystyle\simeq\frac{6}{\sqrt{2}}mt_{a}t_{b}(a+b)^{2}k_{x}k_{y}, (76)
g3y(𝒌)\displaystyle g^{y}_{3}(\bm{k}) m[2fQ0(𝒌)fQv(𝒌)\displaystyle\simeq m\left[2f^{Q_{0}}(\bm{k})f^{Q_{v}}(\bm{k})\right.
12{(fTx(𝒌))2(fTy(𝒌))2}],\displaystyle\ \ \ \left.-\frac{1}{\sqrt{2}}\left\{(f^{T_{x}}(\bm{k}))^{2}-(f^{T_{y}}(\bm{k}))^{2}\right\}\right],
32mtatb(a+b)2(kx2ky2).\displaystyle\simeq\frac{3}{\sqrt{2}}mt_{a}t_{b}(a+b)^{2}\left(k_{x}^{2}-k_{y}^{2}\right). (77)

Also in this case, the functional forms are similar to those in the case of the kagome lattice in Eqs. (68) and (69). In fact, g3x(𝒌)g^{x}_{3}(\bm{k}) and g3y(𝒌)g^{y}_{3}(\bm{k}) for the kagome and breathing kagome lattices are identical when we regard as a+b2aa+b\to 2a and tbtat_{b}\to t_{a}.

There are the contributions to the antisymmetric spin splitting in the zz-component due to the presence of bond magnetic toroidal multipoles. The lowest-order contribution is obtained at the fifth order in Eq. (49) as

g5z(𝒌)=m263(62fQ0(𝒌)2fT3a(𝒌)\displaystyle g_{5}^{z}(\bm{k})=\frac{m^{2}}{6\sqrt{3}}\left(6\sqrt{2}f^{Q_{0}}(\bm{k})^{2}f^{T_{3a}}(\bm{k})\right.
+62fQ0(𝒌)(fQxy(𝒌)fTy(𝒌)fQv(𝒌)fTx(𝒌))\displaystyle\ \ \ \left.+6\sqrt{2}f^{Q_{0}}(\bm{k})(f^{Q_{xy}}(\bm{k})f^{T_{y}}(\bm{k})-f^{Q_{v}}(\bm{k})f^{T_{x}}(\bm{k}))\right.
32[fQv(𝒌)2+fQxy(𝒌)2]fT3a(𝒌)22fT3a(𝒌)3\displaystyle\ \ \ \left.-3\sqrt{2}\left[f^{Q_{v}}(\bm{k})^{2}+f^{Q_{xy}}(\bm{k})^{2}\right]f^{T_{3a}}(\bm{k})-2\sqrt{2}f^{T_{3a}}(\bm{k})^{3}\right.
6[fQv(𝒌)2fQxy(𝒌)2]fTx(𝒌)\displaystyle\ \ \ \left.-6\left[f^{Q_{v}}(\bm{k})^{2}-f^{Q_{xy}}(\bm{k})^{2}\right]f^{T_{x}}(\bm{k})\right.
12fQv(𝒌)fQxy(𝒌)fTy(𝒌)\displaystyle\ \ \ \left.-12f^{Q_{v}}(\bm{k})f^{Q_{xy}}(\bm{k})f^{T_{y}}(\bm{k})\right.
+32fT3a(𝒌)[fTx(𝒌)2+fTy(𝒌)2]\displaystyle\ \ \ \left.+3\sqrt{2}f^{T_{3a}}(\bm{k})\left[f^{T_{x}}(\bm{k})^{2}+f^{T_{y}}(\bm{k})^{2}\right]\right.
+2fTx(𝒌)[fTx(𝒌)23fTy(𝒌)2]).\displaystyle\ \ \ \left.+2f^{T_{x}}(\bm{k})\left[f^{T_{x}}(\bm{k})^{2}-3f^{T_{y}}(\bm{k})^{2}\right]\right). (78)

All the terms contain the odd number of magnetic toroidal multipoles, as shown in Eq. (60). In the 𝒌𝟎\bm{k}\to\bm{0} limit, g5z(𝒌)g_{5}^{z}(\bm{k}) becomes

g5z(𝒌)23m2{fQ0(𝒌)[fQxy(𝒌)fTy(𝒌)fQv(𝒌)fTx(𝒌)]\displaystyle g_{5}^{z}(\bm{k})\simeq\sqrt{\frac{2}{3}}m^{2}\biggl{\{}f^{Q_{0}}(\bm{k})[f^{Q_{xy}}(\bm{k})f^{T_{y}}(\bm{k})-f^{Q_{v}}(\bm{k})f^{T_{x}}(\bm{k})] (79)
+fQ0(𝒌)2fT3a(𝒌)+132fTx(𝒌)[fTx(𝒌)23fTy(𝒌)2]}\displaystyle\quad+f^{Q_{0}}(\bm{k})^{2}f^{T_{3a}}(\bm{k})+\frac{1}{3\sqrt{2}}f^{T_{x}}(\bm{k})[f^{T_{x}}(\bm{k})^{2}-3f^{T_{y}}(\bm{k})^{2}]\biggr{\}}
=8m2tatb(tatb)sinkx(coskxcosky)\displaystyle=8m^{2}t_{a}t_{b}(t_{a}-t_{b})\sin k^{\prime}_{x}\left(\cos k^{\prime}_{x}-\cos k^{\prime}_{y}\right)
12(a+b)3m2tatb(tatb)kx(kx23ky2),\displaystyle\simeq-\frac{1}{2}(a+b)^{3}m^{2}t_{a}t_{b}(t_{a}-t_{b})k_{x}(k_{x}^{2}-3k_{y}^{2}), (80)

where kx=kx(a+b)/2k^{\prime}_{x}=k_{x}(a+b)/2 and ky=ky(a+b)3/2k^{\prime}_{y}=k_{y}(a+b)\sqrt{3}/2. From the expression in Eq. (79), the antisymmetric spin splitting occurs for ta0t_{a}\neq 0, tb0t_{b}\neq 0, and tatbt_{a}\neq t_{b}, i.e., the breathing structure is important.

The above analysis for the spin splittings is also confirmed by calculating explicitly the electronic band structure. Figures 12(a) and (b) show the band structure at ta=1t_{a}=1, tb=0.5t_{b}=0.5, and m=1m=1 where the color map shows the spin polarization for the xx and yy components, respectively. The result is similar to that in the kagome case in Fig. 10; the symmetric spin splittings are characterized by Qxy(m)(𝒌)σxQ^{\rm(m)}_{xy}(\bm{k})\sigma_{x} in Fig. 12(a) and Qv(m)(𝒌)σxQ^{\rm(m)}_{v}(\bm{k})\sigma_{x} in Fig. 12(b). In contrast to the result in the kagome system, the antisymmetric spin splitting occurs, as shown in Fig. 12(c), which is similar to that in the triangular case in Fig. 8.

VI.5 Effect of external magnetic field

We investigate the effect of an external magnetic field on the breathing kagome AFM. To this end, we add the Zeeman coupling term, 𝑯iσσciσ𝝈σσciσ-\bm{H}\cdot\sum_{i\sigma\sigma^{\prime}}c_{i\sigma}^{\dagger}\bm{\sigma}_{\sigma\sigma^{\prime}}c_{i\sigma^{\prime}}, to the Hamiltonian in Eq. (72), which is given in the multipole notation as

Hmag=Q0(c)𝑯𝝈.\displaystyle H_{\rm mag}=-Q_{0}^{\rm(c)}\bm{H}\cdot\bm{\sigma}. (81)

With this Zeeman term, the expansion procedure leads to the additional multipole couplings according to the symmetry reduction. There are mainly two types of additional couplings. One is the term proportional to the odd order of 𝑯\bm{H}, and the other is the term proportional to the even order of 𝑯\bm{H}.

For 𝑯[100]\bm{H}\parallel[100], the antisymmetric contributions proportional to HxH_{x} are given by

g0(𝒌)\displaystyle g^{0}(\bm{k}) mHxQxy(m)(𝒌)k2sin2ϕ,\displaystyle\sim mH_{x}Q^{\rm(m)}_{xy}(\bm{k})\sim k^{2}\sin 2\phi, (82)
gx(𝒌)\displaystyle g^{x}(\bm{k}) m2HxQv(m)(𝒌)k2cos2ϕ,\displaystyle\sim m^{2}H_{x}Q^{\rm(m)}_{v}(\bm{k})\sim k^{2}\cos 2\phi, (83)
gy(𝒌)\displaystyle g^{y}(\bm{k}) m2HxQxy(m)(𝒌)k2sin2ϕ,\displaystyle\sim m^{2}H_{x}Q^{\rm(m)}_{xy}(\bm{k})\sim k^{2}\sin 2\phi, (84)
gz(𝒌)\displaystyle g^{z}(\bm{k}) m3HxQxy(m)(𝒌)T3a(m)(𝒌)k5sinϕ,\displaystyle\sim m^{3}H_{x}Q^{\rm(m)}_{xy}(\bm{k})T_{3a}^{\rm(m)}(\bm{k})\sim k^{5}\sin\phi, (85)

where 𝒌=k(cosϕ,sinϕ)\bm{k}=k(\cos\phi,\sin\phi) and we omit the subscript of gsμ(𝒌)g^{\mu}_{s}(\bm{k}). Thus, one can expect the following additional effects under the [100] magnetic field: the xyxy-type shear stress in the form of Qxy(m)(𝒌)Q^{\rm(m)}_{xy}(\bm{k}), the symmetric spin splitting in forms of Qv(m)(𝒌)Q^{\rm(m)}_{v}(\bm{k}) and Qxy(m)(𝒌)Q^{\rm(m)}_{xy}(\bm{k}) for σx\sigma_{x} and σy\sigma_{y} spin components, respectively, and the antisymmetric spin splitting in the form of Qxy(m)(𝒌)T3a(m)(𝒌)Q^{\rm(m)}_{xy}(\bm{k})T^{\rm(m)}_{3a}(\bm{k}). Especially, the last additional antisymmetric spin splitting is related to the magnetoelectric effect, since Qxy(m)(𝒌)T3a(m)(𝒌)Q^{\rm(m)}_{xy}(\bm{k})T_{3a}^{\rm(m)}(\bm{k}) has the same symmetry as the electric dipole QxQ_{x} Hayami et al. (2018).

The symmetric contributions proportional to Hx2H_{x}^{2} are given by

g0(𝒌)\displaystyle g^{0}(\bm{k}) m2Hx2Qv(m)(𝒌)k2cos2ϕ,\displaystyle\sim m^{2}H^{2}_{x}Q^{\rm(m)}_{v}(\bm{k})\sim k^{2}\cos 2\phi, (86)
gx(𝒌)\displaystyle g^{x}(\bm{k}) mHx2Qxy(m)(𝒌)k2sin2ϕ,\displaystyle\sim mH^{2}_{x}Q^{\rm(m)}_{xy}(\bm{k})\sim k^{2}\sin 2\phi, (87)
gy(𝒌)\displaystyle g^{y}(\bm{k}) m3Hx2Q0(m)(𝒌)1,\displaystyle\sim m^{3}H^{2}_{x}Q^{\rm(m)}_{0}(\bm{k})\sim 1, (88)
gz(𝒌)\displaystyle g^{z}(\bm{k}) m4Hx2Qv(m)(𝒌)T3a(m)(𝒌)k5cosϕ.\displaystyle\sim m^{4}H^{2}_{x}Q^{\rm(m)}_{v}(\bm{k})T^{\rm(m)}_{3a}(\bm{k})\sim k^{5}\cos\phi. (89)

The obtained expressions indicate that magnetization in the yy component is spontaneously induced by applying the magnetic field even along the [100] direction. In this way, the multipole couplings under the magnetic field are obtained systematically. We summarize some of active multipoles under the magnetic field along various directions in Tables 3, 4, and 5.

Table 3: Some of active momentum multipoles when the magnetic field is applied in the xyxy plane, 𝑯=H(cosθ,sinθ,0)\bm{H}=H(\cos\theta,\sin\theta,0). The superscript (m)\rm(m) and (𝒌)(\bm{k}) in the multipoles Qlm(m)(𝒌)Q^{\rm(m)}_{lm}(\bm{k}) and Tlm(m)(𝒌)T^{\rm(m)}_{lm}(\bm{k}) are omitted for notational simplicity.
H(cosθ,sinθ,0)H(\cos\theta,\sin\theta,0) σ0\sigma_{0} σx\sigma_{x} σy\sigma_{y} σz\sigma_{z}
HcosθH\cos\theta mQxymQ_{xy} m2Qvm^{2}Q_{v} m2Qxym^{2}Q_{xy} m3QxyT3am^{3}Q_{xy}T_{3a}
HsinθH\sin\theta mQvmQ_{v} m2Qxym^{2}Q_{xy} m2Qvm^{2}Q_{v} m3QvT3am^{3}Q_{v}T_{3a}
H2cos2θH^{2}\cos 2\theta m2Qvm^{2}Q_{v} mQxymQ_{xy} m3,mQvm^{3},mQ_{v} m4QvT3am^{4}Q_{v}T_{3a}
H2sin2θH^{2}\sin 2\theta m2Qxym^{2}Q_{xy} m3,mQvm^{3},mQ_{v} mQxymQ_{xy} m4QxyT3am^{4}Q_{xy}T_{3a}
Table 4: Some of active momentum multipoles when the magnetic field is applied in the zxzx plane, 𝑯=H(sinθ,0,cosθ)\bm{H}=H(\sin\theta,0,\cos\theta). The superscript (m)\rm(m) and (𝒌)(\bm{k}) in the multipoles Qlm(m)(𝒌)Q^{\rm(m)}_{lm}(\bm{k}) and Tlm(m)(𝒌)T^{\rm(m)}_{lm}(\bm{k}) are omitted for notational simplicity.
H(sinθ,0,cosθ)H(\sin\theta,0,\cos\theta) σ0\sigma_{0} σx\sigma_{x} σy\sigma_{y} σz\sigma_{z}
HcosθH\cos\theta m2T3am^{2}T_{3a} m3QxyT3am^{3}Q_{xy}T_{3a} m3QvT3am^{3}Q_{v}T_{3a} m2H2Qvm^{2}H^{2}Q_{v}
HsinθH\sin\theta mQxymQ_{xy} m2Qvm^{2}Q_{v} m2Qxym^{2}Q_{xy} m3QxyT3am^{3}Q_{xy}T_{3a}
H2cos2θH^{2}\cos 2\theta m2Qvm^{2}Q_{v} mQxymQ_{xy} m3,m3Qvm^{3},m^{3}Q_{v} m2T3am^{2}T_{3a}
H2sin2θH^{2}\sin 2\theta m3QxyT3am^{3}Q_{xy}T_{3a} m2T3am^{2}T_{3a} m4QxyT3am^{4}Q_{xy}T_{3a} mQxymQ_{xy}
Table 5: Some of active momentum multipoles when the magnetic field is applied in the zyzy plane, 𝑯=H(0,sinθ,cosθ)\bm{H}=H(0,\sin\theta,\cos\theta). The superscript (m)\rm(m) and (𝒌)(\bm{k}) in the multipoles Qlm(m)(𝒌)Q^{\rm(m)}_{lm}(\bm{k}) and Tlm(m)(𝒌)T^{\rm(m)}_{lm}(\bm{k}) are omitted for notational simplicity.
H(0,sinθ,cosθ)H(0,\sin\theta,\cos\theta) σ0\sigma_{0} σx\sigma_{x} σy\sigma_{y} σz\sigma_{z}
HcosθH\cos\theta m2T3am^{2}T_{3a} m3QxyT3am^{3}Q_{xy}T_{3a} m3QvT3am^{3}Q_{v}T_{3a} m2H2Qvm^{2}H^{2}Q_{v}
HsinθH\sin\theta mQvmQ_{v} m2Qxym^{2}Q_{xy} m2Qvm^{2}Q_{v} m3QvT3am^{3}Q_{v}T_{3a}
H2cos2θH^{2}\cos 2\theta m2Qvm^{2}Q_{v} m3Qxym^{3}Q_{xy} m3,mQvm^{3},mQ_{v} m2T3am^{2}T_{3a}
H2sin2θH^{2}\sin 2\theta m3QvT3am^{3}Q_{v}T_{3a} m4QxyT3am^{4}Q_{xy}T_{3a} m2T3am^{2}T_{3a} mQvmQ_{v}

VII Discussion on Materials and Summary

The spin splittings and band deformations by the effective multipole-spin couplings irrespective of the SOC are ubiquitously found in various structures of magnetic materials with a variety of chemical compositions. The symmetric spin splitting in collinear AFM state has been studied for an organic κ\kappa-(BETD-TTF)2Cu[N(CN)2]Cl Naka et al. (2019); Hayami et al. (2019a, 2020a); Naka et al. (2020), transition metal oxide RuO2 Berlijn et al. (2017); Ahn et al. (2019), and transition metal fluoride MnF2 Yuan et al. (2020). Moreover, the antisymmetric spin splitting and band deformation have been studied for a transition metal oxide Ba3MnNb2O9 Lee et al. (2014); Hayami et al. (2020b). A similar analysis can be applied to the exchange Hamiltonian for insulating noncollinear magnets where an emergent Dzyaloshinskii-Moriya interaction without the SOC arises through the multipole couplings Cardias et al. (2020).

With the knowledge of the multipole couplings, we list up the candidate materials that could exhibit spin-split band structures at the onset of the AFM phase transition having ordering vector 𝑸=𝟎\bm{Q}=\bm{0} in Tables LABEL:tab:sym and LABEL:tab:asym, which are obtained in accordance with MAGNDATA, magnetic structure database Gallego et al. (2016). It is noted that the AFM materials with a finite ordering vector 𝑸\bm{Q} are also candidates, as exemplified for Ba3MnNb2O9 Hayami et al. (2020b). In the candidate materials, the symmetric spin splitting emerges in the magnetic-ordered-moment direction, while the antisymmetric spin splitting emerges in the direction perpendicular to the coplanar magnetic structure, when the present mechanism dominates over ordinary SOC one. Since the present mechanism of the spin splitting and band deformation does not rely on the presence of the SOC, we can explore a variety class of materials including simple compounds with lighter elements. The multipole-spin couplings can be flexibly tuned by temperature, pressure, and magnetic fields as discussed in this paper. The effect of the splitting and deformation is expected to be large as compared with those by the spin-orbit coupling origin since the magnitude of the effective coupling is characterized by the kinetic energy and the molecular field of AFM, which is the order of the Coulomb interaction. These advantages further promote the efficient engineering of spin-orbit-coupling free materials exhibiting a giant spin-dependent and/or nonreciprocal transport, magneto- electric and elastic responses.

To summarize, we have clarified the efficient bottom-up design procedure of electronic band structures in AFMs without the spin-orbit coupling. Our microscopic guideline to engineer the spin and momentum dependent band structures was established by introducing the concept of augmented multipoles consisting of the electric and magnetic toroidal multipoles. We showed that arbitrary Hamiltonians in the tight-binding model are decomposed into a scalar-product form in terms of electric and magnetic toroidal multipoles. The hopping Hamiltonian is expressed as a linear combination of products between bond and momentum multipoles, while the mean-field Hamiltonian is expressed as a linear combination of products between cluster multipoles and spins. By using such multipole degrees of freedom, we demonstrated that the spin-split and reshaped electronic band structures are caused by the effective multipole couplings. The symmetric spin splitting emerges in the presence of the coupling between cluster and bond multipoles with the same symmetry in collinear AFMs, whereas the antisymmetric one is induced by the coupling including a bond-type magnetic toroidal multipole in noncollinear AFMs. Furthermore, we found that the antisymmetric band deformation with spin degeneracy is realized in noncoplanar AFMs. We analyzed the band deformations under the AFM orderings by exemplifying three lattice systems consisting of a triangle cluster, triangular, kagome, and breathing kagome structures. Lastly, we listed candidate materials showing intrinsic band deformations driven by the magnetic orderings by referring MAGNDATA, magnetic structures database, which would be useful to unveil unexplored fascinating functional materials.

Table 6: Symmetric spin-splitting materials listed in Ref. Gallego et al., 2016. SG, MSG, and MPG represent space group, magnetic space group, and magnetic point group, respectively. 𝒫\mathcal{P} stands for the presence (\circ) or absence (×\times) of the spatial inversion symmetry. # represents the serial number of space group. The symbol * shows that there are different magnetic patterns.
Crystal systems Materials # SG MSG MPG 𝒫\mathcal{P}
Monoclinic LiFeP2O7 44 P21P2_{1} P21P2_{1} 22 ×\times
*CaFe5O7 1111 P21/mP2_{1}/m P21/mP2_{1}/m 2/m2/m \circ
*CaFe5O7 1111 P21/mP2_{1}/m P21/mP2_{1}^{\prime}/m^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Nd2NaRuO6 1414 P21/nP2_{1}/n P21/cP2_{1}/c 2/m2/m \circ
LiFe(SO4)2 1414 P21/cP2_{1}/c P21/cP2_{1}/c 2/m2/m \circ
Li2Co(SO4)2 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Li2Mn(SO4)2 1414 P21/cP2_{1}/c P21/cP2_{1}/c 2/m2/m \circ
La2LiRuO6 1414 P21/nP2_{1}/n P21/cP2_{1}/c 2/m2/m \circ
Y2MnCoO6 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
FeCl5D2O(ND4)2 1414 P21/cP2_{1}/c P21P2_{1}^{\prime} 22^{\prime} ×\times
Ca2MnReO6 1414 P21/cP2_{1}/c P21/cP2_{1}/c 2/m2/m \circ
Sr2MnReO6 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Li3Fe2(PO4)3 1414 P21/nP2_{1}/n P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*Tb2MnNiO6 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*Tb2MnNiO6 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*Tb2MnNiO6 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*Tb2MnNiO6 1414 P21/cP2_{1}/c P21/cP2_{1}/c 2/m2/m \circ
Tl2NiMnO6 1414 P21/cP2_{1}/c P21/cP2_{1}/c 2/m2/m \circ
*Cu2(OD)3Cl 1414 P21/cP2_{1}/c P21/cP2_{1}/c 2/m2/m \circ
*Cu2(OD)3Cl 1414 P21/cP2_{1}/c P21/cP2_{1}/c 2/m2/m \circ
Sr2CoTeO6 1414 P21/nP2_{1}/n P21/cP2_{1}/c 2/m2/m \circ
Sr2Co0.9Mg0.1TeO6 1414 P21/nP2_{1}/n P21/cP2_{1}/c 2/m2/m \circ
Ho2CoMnO6 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*Tm2CoMnO6 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*Tm2CoMnO6 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Cu1.94Mn1.06BO5 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
KMnF4 1414 P21/aP2_{1}/a P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
RbMnF4 1414 P21/aP2_{1}/a P1¯P\bar{1} 1¯\bar{1} \circ
Li2FeP2O7 1414 P21/cP2_{1}/c P21/cP2_{1}/c 2/m2/m \circ
Co4(OH)2(C10H16O4)3 1414 P21/cP2_{1}/c P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Mn2ScSbO6 1414 P21/nP2_{1}/n P21/cP2_{1}/c 2/m2/m \circ
[CH3NH3] [Co(COOH)3] 1414 P21/nP2_{1}/n P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
La2CoIrO6 1414 P21/nP2_{1}/n P21/cP2_{1}/c 2/m2/m \circ
Fe3(PO4)2(OH)2 1414 P21/cP2_{1}/c P21/cP2_{1}/c 2/m2/m \circ
Cs2FeCl5{}_{5}\cdotD2O 1515 C2/cC2/c C2C2 22 ×\times
*BiCrO3 1515 C2/cC2/c P1¯P\bar{1} 1¯\bar{1} \circ
*BiCrO3 1515 C2/cC2/c C2/cC2/c 2/m2/m \circ
FeSO4F 1515 C2/cC2/c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Sr2CoOsO6 1515 B2/nB2/n C2/cC2/c 2/m2/m \circ
NaCrGe2O6 1515 C2/cC2/c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Na2BaFe(VO4)2 1515 C2/cC2/c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Orthorhombic SrMn(VO4)(OH) 1919 P212121P2_{1}2_{1}2_{1} P21P2_{1} 22 ×\times
BaCrF5 1919 P212121P2_{1}2_{1}2_{1} P212121P2_{1}^{\prime}2_{1}^{\prime}2_{1} 2222^{\prime}2^{\prime}2 ×\times
GaFeO3 3333 Pna21Pna2_{1} Pna21Pna^{\prime}2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
*Fe2O3 3333 Pna21Pna2_{1} Pna21Pna^{\prime}2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
*Fe2O3 3333 Pna21Pna2_{1} Pna21Pna^{\prime}2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
*[C(ND2)3]Cu(DCOO)3 3333 Pna21Pna2_{1} Pna21Pna2_{1} mm2mm2 ×\times
*[C(ND2)3]Cu(DCOO)3 3333 Pna21Pna2_{1} Pna21Pn^{\prime}a^{\prime}2_{1} mm2m^{\prime}m^{\prime}2 ×\times
Y2Cu2O5 3333 Pna21Pna2_{1} Pna21Pna2_{1} mm2mm2 ×\times
BaCuF4 3636 Cmc21Cmc2_{1} Cmc21Cm^{\prime}c^{\prime}2_{1} mm2m^{\prime}m^{\prime}2 ×\times
Ca3Mn2O7 3636 Cmc21Cmc2_{1} Cmc21Cm^{\prime}c2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
ErGe1.83 3636 Cmc21Cmc2_{1} Cmc21Cmc2_{1} mm2mm2 ×\times
Cu2V2O7 4343 Fdd2Fdd2 Fdd2Fd^{\prime}d^{\prime}2 mm2m^{\prime}m^{\prime}2 ×\times
BiFe0.5Sc0.5O3 4646 Ima2Ima2 Ima2Im^{\prime}a2^{\prime} mm2m^{\prime}m2^{\prime} ×\times
[C(ND2)3]Mn(DCOO)3 5252 PnnaPnna PnnaPn^{\prime}n^{\prime}a mmmm^{\prime}m^{\prime}m \circ
[C(ND2)3]Co(DCOO)3 5252 PnnaPnna PnnaPn^{\prime}na^{\prime} mmmm^{\prime}m^{\prime}m \circ
Fe1.5Mn1.5BO5 5555 PbamPbam PbamPbam mmmmmm \circ
Fe(N(CN2))2 5858 PnnmPnnm PnnmPnn^{\prime}m^{\prime} mmmm^{\prime}m^{\prime}m \circ
KCo4(PO4)3 5858 PnnmPnnm PnnmPnn^{\prime}m^{\prime} mmmm^{\prime}m^{\prime}m \circ
Mn(N(CN2))2 5858 PnnmPnnm PnnmPnn^{\prime}m^{\prime} mmmm^{\prime}m^{\prime}m \circ
*TmMn3O6 5959 PmmnPmmn PmmnPm^{\prime}m^{\prime}n mmmm^{\prime}m^{\prime}m \circ
*TmMn3O6 5959 PmmnPmmn PmmnPmm^{\prime}n^{\prime} mmmm^{\prime}m^{\prime}m \circ
*α\alpha-Mn2O3 6161 PbcaPbca PbcaPbca mmmmmm \circ
*α\alpha-Mn2O3 6161 PbcaPbca PbcaPbca mmmmmm \circ
Ca2RuO4 6161 PbcaPbca PbcaPbca mmmmmm \circ
CuFePO5 6262 PnmaPnma PnmaPnma mmmmmm \circ
NiFePO5 6262 PnmaPnma PnmaPnma mmmmmm \circ
CoFePO5 6262 PnmaPnma PnmaPnm^{\prime}a^{\prime} mmmm^{\prime}m^{\prime}m \circ
Fe2PO5 6262 PnmaPnma PnmaPnma mmmmmm \circ
CoSO4 6262 PnmaPnma PnmaPnma mmmmmm \circ
YCr0.5Mn0.5O3 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
*Mn2GeO4 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
*Mn2GeO4 6262 PnmaPnma PnmaPnma mmmmmm \circ
*Mn2GeO4 6262 PnmaPnma P21/cP2_{1}/c 2/m2/m \circ
NH4Fe2O6 6262 PnmaPnma PnmaPnma mmmmmm \circ
*NdMnO3 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
*NdMnO3 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
ErVO3 6262 PbnmPbnm P21/mP2_{1}^{\prime}/m^{\prime} 2/m2^{\prime}/m^{\prime} \circ
NiTe2O5 6262 PnmaPnma PnmaPnma mmmmmm \circ
(Tm0.7Mn0.3)MnO3 6262 PnmaPnma PnmaPnm^{\prime}a^{\prime} mmmm^{\prime}m^{\prime}m \circ
Cu4(OD)6FBr 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
Nd5Ge4 6262 PnmaPnma PnmaPnm^{\prime}a^{\prime} mmmm^{\prime}m^{\prime}m \circ
ErVO3 6262 PbnmPbnm P21/cP2_{1}/c 2/m2/m \circ
RbFe2F6 6262 PnmaPnma PnmaPnma mmmmmm \circ
Ca2PrCr2NbO9 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
Ca2PrCr2TaO9 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
DyVO3 6262 PbnmPbnm P21/mP2_{1}^{\prime}/m^{\prime} 2/m2^{\prime}/m^{\prime} \circ
NaOsO3 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
Ca2Fe0.875Cr0.125GaO5 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
La0.5Sr0.5FeO2.5F0.5 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
ScCrO3 6262 PnmaPnma PnmaPnma mmmmmm \circ
InCrO3 6262 PnmaPnma PnmaPnma mmmmmm \circ
TlCrO3 6262 PnmaPnma PnmaPnma mmmmmm \circ
*Co2SiO4 6262 PnmaPnma PnmaPnma mmmmmm \circ
*Co2SiO4 6262 PnmaPnma PnmaPnma mmmmmm \circ
Mn2SiO4 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
Fe2SiO4 6262 PnmaPnma PnmaPnma mmmmmm \circ
DyFeO3 6262 PnmaPnma Pna21Pn^{\prime}a^{\prime}2_{1} mm2m^{\prime}m^{\prime}2 ×\times
LaCrO3 6262 PnmaPnma PnmaPnma mmmmmm \circ
BiFe0.5Sc0.5O3 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
*NdFeO3 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
*NdFeO3 6262 PnmaPnma P21/cP2_{1}^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*TbFeO3 6262 PbnmPbnm PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
*TbFeO3 6262 PbnmPbnm PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
TbCrO3 6262 PbnmPbnm PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
TbPt0.8Cu0.2 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
NdNi0.6Cu0.4 6262 PnmaPnma PnmaPnm^{\prime}a^{\prime} mmmm^{\prime}m^{\prime}m \circ
[CH3NH3][Co(COOH)3] 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
LaMnO3 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
*NdMnO3 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
*NdMnO3 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
La0.75Bi0.25Fe0.5Cr0.5O3 6262 PnmaPnma PnmaPnma mmmmmm \circ
*Rb2Fe2O(AsO4)2 6262 PnmaPnma PnmaPnma mmmmmm \circ
*SmFeO3 6262 PbnmPbnm PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
*SmFeO3 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
*Rb2Fe2O(AsO4)2 6262 PnmaPnma PnmaPn^{\prime}ma^{\prime} mmmm^{\prime}m^{\prime}m \circ
Ca2Fe2O5 6262 PcmnPcmn PcmnPcm^{\prime}n^{\prime} mmmm^{\prime}m^{\prime}m \circ
TeNiO3 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
NdSi 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
PrSi 6262 PnmaPnma PnmaPnm^{\prime}a^{\prime} mmmm^{\prime}m^{\prime}m \circ
TmNi 6262 PnmaPnma PnmaPn^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
Y3Co3.25Al0.75 6363 CmcmCmcm CmcmCm^{\prime}cm^{\prime} mmmm^{\prime}m^{\prime}m \circ
CaIrO3 6363 CmcmCmcm CmcmCm^{\prime}cm^{\prime} mmmm^{\prime}m^{\prime}m \circ
LaCaFeO4 6464 CmceCmce CmcaCm^{\prime}c^{\prime}a mmmm^{\prime}m^{\prime}m \circ
Gd2CuO4 6464 AeamAeam CmcaCm^{\prime}ca^{\prime} mmmm^{\prime}m^{\prime}m \circ
Sr4Fe4O11 6565 CmmmCmmm CmmmCmm^{\prime}m^{\prime} mmmm^{\prime}m^{\prime}m \circ
YNi4Si 6565 CmmmCmmm CmmmCmm^{\prime}m^{\prime} mmmm^{\prime}m^{\prime}m \circ
*Y2SrCu0.6Co1.4O6.5 7272 IbamIbam IbamIb^{\prime}a^{\prime}m mmmm^{\prime}m^{\prime}m \circ
*Y2SrCu0.6Co1.4O6.5 7272 IbamIbam IbamIb^{\prime}a^{\prime}m mmmm^{\prime}m^{\prime}m \circ
*YBaMn2O5.5 7272 IcamIcam C2/mC2/m 2/m2/m \circ
*YBaMn2O5.5 7272 IcamIcam IbamIb^{\prime}a^{\prime}m mmmm^{\prime}m^{\prime}m \circ
Pr0.5Sr0.5CoO3 7474 ImmaImma ImmaIm^{\prime}m^{\prime}a mmmm^{\prime}m^{\prime}m \circ
Tetragonal MnPrMnSbO6 8686 P42/nP4_{2}/n P42/nP4_{2}/n 4/m4/m \circ
MnLaMnSbO6 8686 P42/nP4_{2}/n P2/cP2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
KyFe2-xSe2 8787 I4/mI4/m C2/mC2^{\prime}/m^{\prime} 2/m2^{\prime}/m^{\prime} \circ
TlFe1.6Se2 8787 I4/mI4/m I4/mI4/m 4/m4/m \circ
RbyFe2-xSe2 8787 I4/mI4/m C2/mC2^{\prime}/m^{\prime} 2/m2^{\prime}/m^{\prime} \circ
MnV2O4 8888 I41/aI4_{1}/a I41/aI4_{1}/a 4/m4/m \circ
SrMn2V2O8 110110 I41cdI4_{1}cd Iba2Ib^{\prime}a2^{\prime} mm2m^{\prime}m2^{\prime} ×\times
Ba2MnSi2O7 113113 P4¯21mP\bar{4}2_{1}m P4¯21mP\bar{4}2_{1}m 4¯2m\bar{4}2m ×\times
Ba2CoGe2O7 113113 P4¯21mP\bar{4}2_{1}m Cmm2Cm^{\prime}m2^{\prime} mm2m^{\prime}m2^{\prime} ×\times
Ca2CoSi2O7 113113 P4¯21mP\bar{4}2_{1}m P21212P2_{1}2_{1}^{\prime}2^{\prime} 2222^{\prime}2^{\prime}2 ×\times
CsCoF4 120120 I4¯c2I\bar{4}c2 I4¯I\bar{4}^{\prime} 4¯\bar{4}^{\prime} ×\times
CeMn2Ge4O12 125125 P4/nbmP4/nbm P4/nbmP4^{\prime}/nbm^{\prime} 4/mmm4^{\prime}/mm^{\prime}m \circ
CeMnCoGe4O12 125125 P4/nbmP4/nbm PbanPb^{\prime}an^{\prime} mmmm^{\prime}m^{\prime}m \circ
ZrCo2Ge4O12 125125 P4/nbmP4/nbm PbanPb^{\prime}an^{\prime} mmmm^{\prime}m^{\prime}m \circ
ZrMn2Ge4O12 125125 P4/nbmP4/nbm P4/nbmP4^{\prime}/nbm^{\prime} 4/mmm4^{\prime}/mm^{\prime}m \circ
CsMnF4 129129 P4/nmmP4/nmm PmmnPmm^{\prime}n^{\prime} mmmm^{\prime}m^{\prime}m \circ
MnF2 136136 P42/mnmP4_{2}/mnm P42/mnmP4_{2}^{\prime}/mnm^{\prime} 4/mmm4^{\prime}/mm^{\prime}m \circ
NiF2 136136 P42/mnmP4_{2}/mnm PnnmPnn^{\prime}m^{\prime} mmmm^{\prime}m^{\prime}m \circ
CoF2 136136 P42/mnmP4_{2}/mnm P42/mnmP4_{2}^{\prime}/mnm^{\prime} 4/mmm4^{\prime}/mm^{\prime}m \circ
Nd2NiO4.11 138138 P42/ncmP4_{2}/ncm P42/ncmP4_{2}/nc^{\prime}m^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
*Nd2NiO4 138138 P42/ncmP4_{2}/ncm P42/ncmP4_{2}/nc^{\prime}m^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
*La2NiO4 138138 P42/ncmP4_{2}/ncm PccnPc^{\prime}c^{\prime}n mmmm^{\prime}m^{\prime}m \circ
Sr2Mn2CuAs2O2 139139 I4/mmmI4/mmm I4/mmmI4/mm^{\prime}m^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
Mn2.85Ga1.15 139139 I4/mmmI4/mmm I4/mmmI4/mm^{\prime}m^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
EuCr2As2 139139 I4/mmmI4/mmm I4¯m2I\bar{4}m^{\prime}2^{\prime} 4¯2m\bar{4}2^{\prime}m^{\prime} ×\times
CaFe4Al8 139139 I4/mmmI4/mmm I4/mmmI4^{\prime}/mmm^{\prime} 4/mmm4^{\prime}/mm^{\prime}m \circ
Pr0.5Sr0.5CoO3 140140 I4/mcmI4/mcm FmmmFm^{\prime}m^{\prime}m mmmm^{\prime}m^{\prime}m \circ
NiCr2O4 141141 I41/amdI4_{1}/amd FdddFd^{\prime}d^{\prime}d mmmm^{\prime}m^{\prime}m \circ
Sr2Ir0.92Sn0.08O4 142142 I41/acdI4_{1}/acd IbcaIb^{\prime}c^{\prime}a mmmm^{\prime}m^{\prime}m \circ
Trigonal Mn2ScSbO6 146146 R3R3 P1P1 11 ×\times
Mn2FeMoO6 146146 R3R3 R3R3 33 ×\times
Mn2FeSbO6 148148 R3¯R\bar{3} P1¯P\bar{1} 1¯\bar{1} \circ
NiN2O6 148148 R3¯R\bar{3} R3¯R\bar{3} 3¯\bar{3} \circ
Li3Fe2(PO4)3 148148 R3¯R\bar{3} R3¯R\bar{3} 3¯\bar{3} \circ
Cr2S3 148148 R3¯R\bar{3} P1¯P\bar{1} 1¯\bar{1} \circ
NaMnFeF6 150150 P321P321 P321P32^{\prime}1 3232^{\prime} ×\times
GaFeO3 161161 R3cR3c CcCc^{\prime} mm^{\prime} ×\times
ScFeO3 161161 R3cR3c CcCc^{\prime} mm^{\prime} ×\times
MnTiO3 161161 R3cR3c CcCc^{\prime} mm^{\prime} ×\times
PbNiO3 161161 R3cR3c R3cR3c 3m3m ×\times
[NH2(CH3)2][FeCo(HCOO)6] 163163 P3¯c1P\bar{3}c1 C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
[NH2(CH3)2][FeMn(HCOO)6] 163163 P3¯c1P\bar{3}c1 C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Mn3Si2Te6 163163 P3¯1cP\bar{3}1c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Mn3Ti2Te6 163163 P3¯1cP\bar{3}1c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Na2BaCo(VO4)2 164164 P3¯m1P\bar{3}m1 P3¯m1P\bar{3}m^{\prime}1 3¯m\bar{3}m^{\prime} \circ
Nd3Sb3Mg2O14 166166 R3¯mR\bar{3}m R3¯mR\bar{3}m^{\prime} 3¯m\bar{3}m^{\prime} \circ
NiCO3 167167 R3¯cR\bar{3}c C2/cC2/c 2/m2/m \circ
CoF3 167167 R3¯cR\bar{3}c R3¯cR\bar{3}c 3¯m\bar{3}m \circ
FeF3 167167 R3¯cR\bar{3}c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
CoCO3 167167 R3¯cR\bar{3}c C2/cC2/c 2/m2/m \circ
Sr3LiRuO6 167167 R3¯cR\bar{3}c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
MnCO3 167167 R3¯cR\bar{3}c C2/cC2/c 2/m2/m \circ
FeCO3 167167 R3¯cR\bar{3}c R3¯cR\bar{3}c 3¯m\bar{3}m \circ
FeBO3 167167 R3¯cR\bar{3}c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Ca3Co2-xMnxO6 167167 R3¯cR\bar{3}c R3cR3c 3m3m ×\times
Ca3LiOsO6 167167 R3¯cR\bar{3}c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
[NH2(CH3)2]n[FeIIIFeII(HCOO)6]n 167167 R3¯cR\bar{3}c R3¯cR\bar{3}c^{\prime} 3¯m\bar{3}m^{\prime} \circ
Sr3NaRuO6 167167 R3¯cR\bar{3}c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Ca3LiRuO6 167167 R3¯cR\bar{3}c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*α\alpha-Fe2O3 167167 R3¯cR\bar{3}c C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*α\alpha-Fe2O3 167167 R3¯cR\bar{3}c P1¯P\bar{1} 1¯\bar{1} \circ
Hexagonal Cu4(OH)6FBr 176176 P63/mP63/m P21/mP2_{1}^{\prime}/m^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Fe2Mo3O8 186186 P63mcP6_{3}mc P63mcP6_{3}^{\prime}m^{\prime}c 6mm6^{\prime}mm^{\prime} ×\times
*Co2Mo3O8 186186 P63mcP6_{3}mc P63mcP6_{3}^{\prime}m^{\prime}c 6mm6^{\prime}mm^{\prime} ×\times
Mn2Mo3O8 186186 P63mcP6_{3}mc P63mcP6_{3}m^{\prime}c^{\prime} 6mm6m^{\prime}m^{\prime} ×\times
*Co2Mo3O8 186186 P63mcP6_{3}mc P63mcP6_{3}^{\prime}m^{\prime}c 6mm6^{\prime}mm^{\prime} ×\times
Mn5Ge3 193193 P63/mcmP6_{3}/mcm P63/mcmP6_{3}/mc^{\prime}m^{\prime} 6/mmm6/mm^{\prime}m^{\prime} \circ
*Mn3Sn 194194 P63/mmcP6_{3}/mmc CmcmCmc^{\prime}m^{\prime} mmmm^{\prime}m^{\prime}m \circ
*Mn3As 194194 P63/mmcP6_{3}/mmc CmcmCmc^{\prime}m^{\prime} mmmm^{\prime}m^{\prime}m \circ
*Mn3As 194194 P63/mmcP6_{3}/mmc CmcmCm^{\prime}cm^{\prime} mmmm^{\prime}m^{\prime}m \circ
*MnPtGa 194194 P63/mmcP6_{3}/mmc CmcmCm^{\prime}c^{\prime}m mmmm^{\prime}m^{\prime}m \circ
*MnPtGa 194194 P63/mmcP6_{3}/mmc CmcmCm^{\prime}c^{\prime}m mmmm^{\prime}m^{\prime}m \circ
*Mn3Sn 194194 P63/mmcP6_{3}/mmc CmcmCm^{\prime}cm^{\prime} mmmm^{\prime}m^{\prime}m \circ
*Mn3Ge 194194 P63/mmcP6_{3}/mmc CmcmCm^{\prime}cm^{\prime} mmmm^{\prime}m^{\prime}m \circ
*Mn3Ge 194194 P63/mmcP6_{3}/mmc C2/mC2^{\prime}/m^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Ba5Co5ClO13 194194 P63/mmcP6_{3}/mmc P63/mmcP6_{3}^{\prime}/m^{\prime}m^{\prime}c 6/mmm6^{\prime}/m^{\prime}mm^{\prime} \circ
*Pr3Ru4Al12 194194 P63/mmcP6_{3}/mmc CmcmCm^{\prime}c^{\prime}m mmmm^{\prime}m^{\prime}m \circ
*Pr3Ru4Al12 194194 P63/mmcP6_{3}/mmc C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
Nd3Ru4Al12 194194 P63/mmcP6_{3}/mmc CmcmCm^{\prime}c^{\prime}m mmmm^{\prime}m^{\prime}m \circ
Mn2.85Ga1.15 194194 P63/mmcP6_{3}/mmc P63/mmcP6_{3}^{\prime}/m^{\prime}m^{\prime}c 6/mmm6^{\prime}/m^{\prime}mm^{\prime} \circ
Cubic Cu2OSeO3 198198 P213P2_{1}3 R3R3 33 ×\times
Na3Co(CO3)2Cl 203203 Fd3¯Fd\bar{3} Fd3¯Fd\bar{3} m3¯m\bar{3} \circ
MnTe2 205205 Pa3¯Pa\bar{3} Pa3¯Pa\bar{3} m3¯m\bar{3} \circ
NiS2 205205 Pa3¯Pa\bar{3} Pa3¯Pa\bar{3} m3¯m\bar{3} \circ
Tb2C3 220220 I4¯3dI\bar{4}3d Fdd2Fd^{\prime}d2^{\prime} mm2m^{\prime}m2^{\prime} ×\times
Mn3Cu0.5Ge0.5N 221221 Pm3¯mPm\bar{3}m R3¯mR\bar{3}m 3¯m\bar{3}m \circ
*Mn3NiN 221221 Pm3¯mPm\bar{3}m R3¯R\bar{3} 3¯\bar{3} \circ
*Mn3NiN 221221 Pm3¯mPm\bar{3}m R3¯R\bar{3} 3¯\bar{3} \circ
Mn3Ir 221221 Pm3¯mPm\bar{3}m R3¯mR\bar{3}m^{\prime} 3¯m\bar{3}m^{\prime} \circ
Mn3Pt 221221 Pm3¯mPm\bar{3}m R3¯mR\bar{3}m^{\prime} 3¯m\bar{3}m^{\prime} \circ
Mn3GaN 221221 Pm3¯mPm\bar{3}m R3¯mR\bar{3}m 3¯m\bar{3}m \circ
Mn3ZnN 221221 Pm3¯mPm\bar{3}m R3¯mR\bar{3}m 3¯m\bar{3}m \circ
*Mn3AlN 221221 Pm3¯mPm\bar{3}m R3¯mR\bar{3}m^{\prime} 3¯m\bar{3}m^{\prime} \circ
*Mn3AlN 221221 Pm3¯mPm\bar{3}m CmmmCmm^{\prime}m^{\prime} mmmm^{\prime}m^{\prime}m \circ
Mn4N 221221 Pm3¯mPm\bar{3}m R3¯mR\bar{3}m^{\prime} 3¯m\bar{3}m^{\prime} \circ
Mn3(Co0.61Mn0.39)N 221221 Pm3¯mPm\bar{3}m R3¯R\bar{3} 3¯\bar{3} \circ
Ho2CrSbO7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/am^{\prime}d^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
Bi2RuMnO7 227227 Fd3¯mFd\bar{3}m FdddFd^{\prime}d^{\prime}d mmmm^{\prime}m^{\prime}m \circ
Gd2Sn2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}^{\prime}/amd^{\prime} 4/mmm4^{\prime}/mm^{\prime}m \circ
Tb2Ti2O7 227227 Fd3¯mFd\bar{3}m R3¯mR\bar{3}m^{\prime} 3¯m\bar{3}m^{\prime} \circ
Tb2Sn2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/am^{\prime}d^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
Nd2Hf2O7 227227 Fd3¯mFd\bar{3}m Fd3¯mFd\bar{3}m^{\prime} m3¯mm\bar{3}m^{\prime} \circ
Nd2Zr2O7 227227 Fd3¯mFd\bar{3}m Fd3¯mFd\bar{3}m^{\prime} m3¯mm\bar{3}m^{\prime} \circ
*Ho2Ru2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/am^{\prime}d^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
Er2Sn2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}^{\prime}/amd^{\prime} 4/mmm4^{\prime}/mm^{\prime}m \circ
Er2Pt2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}^{\prime}/amd^{\prime} 4/mmm4^{\prime}/mm^{\prime}m \circ
Er2Ti2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}^{\prime}/am^{\prime}d 4/mmm4^{\prime}/mm^{\prime}m \circ
Tm2Mn2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/am^{\prime}d^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
Er2Ru2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}^{\prime}/am^{\prime}d 4/mmm4^{\prime}/mm^{\prime}m \circ
Yb2Sn2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/am^{\prime}d^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
Yb2Ti2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/am^{\prime}d^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
*Ho2Ru2O7 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/am^{\prime}d^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
Cd2Os2O7 227227 Fd3¯mFd\bar{3}m Fd3¯mFd\bar{3}m^{\prime} m3¯mm\bar{3}m^{\prime} \circ
CdYb2S4 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/amd 4/mmm4/mmm \circ
CdYb2Se4 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/amd 4/mmm4/mmm \circ
Nd2Sn2O7 227227 Fd3¯mFd\bar{3}m Fd3¯mFd\bar{3}m^{\prime} m3¯mm\bar{3}m^{\prime} \circ
*Nd0.5Tb0.5Co2 227227 Fd3¯mFd\bar{3}m C2/mC2^{\prime}/m^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*Nd0.5Tb0.5Co2 227227 Fd3¯mFd\bar{3}m C2/mC2^{\prime}/m^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*NdCo2 227227 Fd3¯mFd\bar{3}m ImmaImm^{\prime}a^{\prime} mmmm^{\prime}m^{\prime}m \circ
*NdCo2 227227 Fd3¯mFd\bar{3}m C2/cC2^{\prime}/c^{\prime} 2/m2^{\prime}/m^{\prime} \circ
*NdCo2 227227 Fd3¯mFd\bar{3}m I41/amdI4_{1}/am^{\prime}d^{\prime} 4/mmm4/mm^{\prime}m^{\prime} \circ
TbCo2 227227 Fd3¯mFd\bar{3}m R3¯mR\bar{3}m^{\prime} 3¯m\bar{3}m^{\prime} \circ
Dy3Al5O12 230230 Ia3¯dIa\bar{3}d Ia3¯dIa\bar{3}d^{\prime} m3¯mm\bar{3}m^{\prime} \circ
Table 7: Antisymmetric spin-splitting materials listed in Ref. Gallego et al., 2016. The notations are the same as those in Table LABEL:tab:asym.
Crystal systems Materials # SG MSG MPG 𝒫\mathcal{P}
Monoclinic *Tb2MnNiO6 1414 P21/cP2_{1}/c P21P2_{1}^{\prime} 22^{\prime} ×\times
SrCo(VO4)(OH) 1919 P212121P2_{1}2_{1}2_{1} P212121P2_{1}2_{1}2_{1} 222222 ×\times
Orthorhombic Mn3B7O13I 2929 Pca21Pca2_{1} Pca21Pc^{\prime}a2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
Ni3B7O13Br 2929 Pca21Pca2_{1} Pca21Pc^{\prime}a2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
Ni3B7O13Cl 2929 Pca21Pca2_{1} Pca21Pc^{\prime}a2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
Co3B7O13Br 2929 Pca21Pca2_{1} Pca21Pc^{\prime}a2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
Tm2Cu2O5 3333 Pna21Pna2_{1} Pna21Pn^{\prime}a^{\prime}2_{1} mm2m^{\prime}m^{\prime}2 ×\times
CaBaCo4O7 3333 Pbn21Pbn2_{1} Pna21Pna^{\prime}2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
DyCrWO6 3333 Pna21Pna2_{1} P21P2_{1} 22 ×\times
Er2Cu2O5 3333 Pna21Pna2_{1} Pna21Pna2_{1} mm2mm2 ×\times
Tb3Ge5 4343 Fdd2Fdd2 Fdd2Fdd2 mm2mm2 ×\times
DyFeO3 6262 PnmaPnma P212121P2_{1}2_{1}2_{1} 222222 ×\times
TbFeO3 6262 PbnmPbnm P212121P2_{1}^{\prime}2_{1}^{\prime}2_{1} 2222^{\prime}2^{\prime}2 ×\times
*Cu3Mo2O9 6262 PnmaPnma P212121P2_{1}^{\prime}2_{1}^{\prime}2_{1} 2222^{\prime}2^{\prime}2 ×\times
*Cu3Mo2O9 6262 PnmaPnma Pmc21Pm^{\prime}c2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
FePO4 6262 PnmaPnma P212121P2_{1}2_{1}2_{1} 222222 ×\times
Fe3BO5 6262 PnmaPnma Pmc21Pm^{\prime}c2_{1}^{\prime} mm2m^{\prime}m2^{\prime} ×\times
Tetragonal U3Al2Si3 7979 I4I4 C2C2^{\prime} 22^{\prime} ×\times
Nd5Si4 9292 P41212P4_{1}2_{1}2 P41212P4_{1}2_{1}^{\prime}2^{\prime} 42242^{\prime}2^{\prime} ×\times
Ho2Ge2O7 9292 P41212P4_{1}2_{1}2 P41212P4_{1}2_{1}2 422422 ×\times
KMnFeF6 106106 P42bcP4_{2}bc Pba2Pb^{\prime}a2^{\prime} mm2m^{\prime}m2^{\prime} ×\times
FeSb2O4 135135 P42/mbcP4_{2}/mbc Pmc21Pmc2_{1} mm2mm2 ×\times
FePbBiO4 135135 P42/mbcP4_{2}/mbc Pmc21Pmc2_{1} mm2mm2 ×\times
Hexagonal Cu0.82Mn1.18As 174174 P6¯P\bar{6} P6¯P\bar{6}^{\prime} 6¯\bar{6}^{\prime} ×\times
*HoMnO3 185185 P63cmP6_{3}cm P63cmP6_{3}cm 6mm6mm ×\times
*HoMnO3 185185 P63cmP6_{3}cm P63cmP6_{3}cm 6mm6mm ×\times
*HoMnO3 185185 P63cmP6_{3}cm P63cmP6_{3}^{\prime}c^{\prime}m 6mm6^{\prime}mm^{\prime} ×\times
*HoMnO3 185185 P63cmP6_{3}cm P63cmP6_{3}^{\prime}cm^{\prime} 6mm6^{\prime}mm^{\prime} ×\times
*HoMnO3 185185 P63cmP6_{3}cm P63cmP6_{3}c^{\prime}m^{\prime} 6mm6m^{\prime}m^{\prime} ×\times
*YMnO3 185185 P63cmP6_{3}cm P63P6_{3}^{\prime} 66^{\prime} ×\times
*YMnO3 185185 P63cmP6_{3}cm P63cmP6_{3}cm 6mm6mm ×\times
*ScMnO3 185185 P63cmP6_{3}cm P63P6_{3} 66 ×\times
*ScMnO3 185185 P63cmP6_{3}cm P63cmP6_{3}c^{\prime}m^{\prime} 6mm6m^{\prime}m^{\prime} ×\times
LuFeO3 185185 P63cmP6_{3}cm P63cmP6_{3}c^{\prime}m^{\prime} 6mm6m^{\prime}m^{\prime} ×\times
YbMnO3 185185 P63cmP6_{3}cm P63cmP6_{3}^{\prime}c^{\prime}m 6mm6^{\prime}mm^{\prime} ×\times
Co6(OH)3(TeO3)4(OH)\sim0.90.9(H2O) 186186 P63mcP6_{3}mc P63mcP6_{3}^{\prime}mc^{\prime} 6mm6^{\prime}mm^{\prime} ×\times
Nd15Ge9C0.39 186186 P63mcP6_{3}mc P63mcP6_{3}m^{\prime}c^{\prime} 6mm6m^{\prime}m^{\prime} ×\times
TmAgGe 189189 P6¯2mP\bar{6}2m Amm2Am^{\prime}m^{\prime}2 mm2m^{\prime}m^{\prime}2 ×\times
Cubic U3P4 220220 I4¯3dI\bar{4}3d R3cR3c^{\prime} 3m3m^{\prime} ×\times
U3As4 220220 I4¯3dI\bar{4}3d R3cR3c^{\prime} 3m3m^{\prime} ×\times

Appendix A Expressions of electric multipoles

In this appendix, we show the multipole expressions by using the cubic and hexagonal harmonics up to rank 44 in Table 8.

Table 8: The correspondence between electric multipoles and cubic and hexagonal harmonics up to rank 44. The expressions for rank-0-2 harmonics are common. (lm)(lm) and (lm)(lm)^{\prime} stand for (1)l(Olm+Olm)/2(-1)^{l}(O_{lm}+O^{*}_{lm})/\sqrt{2} and (1)l(OlmOlm)/2i(-1)^{l}(O_{lm}-O^{*}_{lm})/\sqrt{2}i, respectively.
Cubic harmonics
rank symbol Definition correspondence
0 Q0Q_{0} 11 (00)(00)
11 QxQ_{x}, QyQ_{y}, QzQ_{z} xx, yy, zz (11)(11), (11)(11)^{\prime}, (10)(10)
22 QuQ_{u}, QvQ_{v} 12(3z2r2)\frac{1}{2}(3z^{2}-r^{2}), 32(x2y2)\frac{\sqrt{3}}{2}(x^{2}-y^{2}) (20)(20), (22)(22)
QyzQ_{yz}, QzxQ_{zx}, QxyQ_{xy} 3yz\sqrt{3}yz, 3zx\sqrt{3}zx, 3xy\sqrt{3}xy (21)(21)^{\prime}, (21)(21), (22)(22)^{\prime}
3 QxyzQ_{xyz} 15xyz\sqrt{15}xyz (32)(32)^{\prime}
QxαQ_{x}^{\alpha} 12x(5x23r2)\frac{1}{2}x(5x^{2}-3r^{2}) 122[5(33)3(31)]\frac{1}{2\sqrt{2}}[\sqrt{5}(33)-\sqrt{3}(31)]
QyαQ_{y}^{\alpha} 12y(5y23r2)\frac{1}{2}y(5y^{2}-3r^{2}) 122[5(33)+3(31)]-\frac{1}{2\sqrt{2}}[\sqrt{5}(33)^{\prime}+\sqrt{3}(31)^{\prime}]
QzαQ_{z}^{\alpha} 12z(5z23r2)\frac{1}{2}z(5z^{2}-3r^{2}) (30)(30)
QxβQ_{x}^{\beta} 152x(y2z2)\frac{\sqrt{15}}{2}x(y^{2}-z^{2}) 122[3(33)+5(31)]-\frac{1}{2\sqrt{2}}[\sqrt{3}(33)+\sqrt{5}(31)]
QyβQ_{y}^{\beta} 152y(z2x2)\frac{\sqrt{15}}{2}y(z^{2}-x^{2}) 122[3(33)+5(31)]\frac{1}{2\sqrt{2}}[-\sqrt{3}(33)^{\prime}+\sqrt{5}(31)^{\prime}]
QzβQ_{z}^{\beta} 152z(x2y2)\frac{\sqrt{15}}{2}z(x^{2}-y^{2}) (32)(32)
44 Q4Q_{4} 52112(x4+y4+z435r4)\frac{5\sqrt{21}}{12}\left(x^{4}+y^{4}+z^{4}-\frac{3}{5}r^{4}\right) (4)123[5(44)+7(40)](4)\equiv\frac{1}{2\sqrt{3}}[\sqrt{5}(44)+\sqrt{7}(40)]
Q4uQ_{4u} 7156[z4x4+y4237r2(3z2r2)]\frac{7\sqrt{15}}{6}\left[z^{4}-\frac{x^{4}+y^{4}}{2}-\frac{3}{7}r^{2}(3z^{2}-r^{2})\right] 123[7(44)5(40)]-\frac{1}{2\sqrt{3}}[\sqrt{7}(44)-\sqrt{5}(40)]
Q4vQ_{4v} 754[x4y467r2(x2y2)]\frac{7\sqrt{5}}{4}\left[x^{4}-y^{4}-\frac{6}{7}r^{2}(x^{2}-y^{2})\right] (42)-(42)
Q4xαQ^{\alpha}_{4x} 352yz(y2z2)\frac{\sqrt{35}}{2}yz(y^{2}-z^{2}) 122[(43)+7(41)]-\frac{1}{2\sqrt{2}}[(43)^{\prime}+\sqrt{7}(41)^{\prime}]
Q4yαQ^{\alpha}_{4y} 352zx(z2x2)\frac{\sqrt{35}}{2}zx(z^{2}-x^{2}) 122[(43)7(41)]-\frac{1}{2\sqrt{2}}[(43)-\sqrt{7}(41)]
Q4zαQ^{\alpha}_{4z} 352xy(x2y2)\frac{\sqrt{35}}{2}xy(x^{2}-y^{2}) (44)(44)^{\prime}
Q4xβQ^{\beta}_{4x} 52yz(7x2r2)\frac{\sqrt{5}}{2}yz(7x^{2}-r^{2}) 122[7(43)(41)]\frac{1}{2\sqrt{2}}[\sqrt{7}(43)^{\prime}-(41)^{\prime}]
Q4yβQ^{\beta}_{4y} 52zx(7y2r2)\frac{\sqrt{5}}{2}zx(7y^{2}-r^{2}) 122[7(43)+(41)]-\frac{1}{2\sqrt{2}}[\sqrt{7}(43)+(41)]
Q4zβQ^{\beta}_{4z} 52xy(7z2r2)\frac{\sqrt{5}}{2}xy(7z^{2}-r^{2}) (42)(42)^{\prime}
Hexagonal harmonics
rank symbol Definition correspondence
3 QzαQ_{z}^{\alpha} 12z(5z23r2)\frac{1}{2}z(5z^{2}-3r^{2}) (30)(30)
Q3aQ_{3a} 104x(x23y2)\frac{\sqrt{10}}{4}x(x^{2}-3y^{2}) (33)(33)
Q3bQ_{3b} 104y(3x2y2)\frac{\sqrt{10}}{4}y(3x^{2}-y^{2}) (33)(33)^{\prime}
Q3uQ_{3u}, Q3vQ_{3v} 64x(5z2r2)\frac{\sqrt{6}}{4}x(5z^{2}-r^{2}), 64y(5z2r2)\frac{\sqrt{6}}{4}y(5z^{2}-r^{2}) (31)(31), (31)(31)^{\prime}
QzβQ_{z}^{\beta}, QxyzQ_{xyz} 152z(x2y2)\frac{\sqrt{15}}{2}z(x^{2}-y^{2}), 15xyz\sqrt{15}xyz (32)(32), (32)(32)^{\prime}
44 Q40Q_{40} 18(35z430z2r2+3r4)\frac{1}{8}(35z^{4}-30z^{2}r^{2}+3r^{4}) (40)(40)
Q4aQ_{4a} 704yz(3x2y2)\frac{\sqrt{70}}{4}yz(3x^{2}-y^{2}) (43)(43)^{\prime}
Q4bQ_{4b} 704zx(x23y2)\frac{\sqrt{70}}{4}zx(x^{2}-3y^{2}) (43)(43)
Q4uα,Q4vαQ_{4u}^{\alpha},Q_{4v}^{\alpha} 104zx(7z23r2)\frac{\sqrt{10}}{4}zx(7z^{2}-3r^{2}), 104yz(7z23r2)\frac{\sqrt{10}}{4}yz(7z^{2}-3r^{2}) (41)(41), (41)(41)^{\prime}
Q4uβ1,Q4vβ1Q_{4u}^{\beta 1},Q_{4v}^{\beta 1} 358(x46x2y2+y4)\frac{\sqrt{35}}{8}(x^{4}-6x^{2}y^{2}+y^{4}), 352xy(x2y2)\frac{\sqrt{35}}{2}xy(x^{2}-y^{2}) (44)(44), (44)(44)^{\prime}
Q4uβ2,Q4vβ2Q_{4u}^{\beta 2},Q_{4v}^{\beta 2} 54(x2y2)(7z2r2)\frac{\sqrt{5}}{4}(x^{2}-y^{2})(7z^{2}-r^{2}), 52xy(7z2r2)\frac{\sqrt{5}}{2}xy(7z^{2}-r^{2}) (42)(42), (42)(42)^{\prime}

Appendix B Multipole notations under 11 Laue classes

We show the multipole notations per each Laue class in Tables 919.

Table 9: Multipoles under Laue class m3¯mm\bar{3}m. The upper and lower columns represent even-parity electric and odd-parity magnetic toroidal multipoles, respectively. We omit the numerical coefficients of the basis functions.
OhO_{\rm h} OO TdT_{\rm d} MP basis functions
A1g+A^{+}_{1g} A1+A^{+}_{1} A1+A^{+}_{1} Q0Q_{0} 11
A2g+A^{+}_{2g} A2+A^{+}_{2} A2+A^{+}_{2} Q6tQ_{6t} (ky2kz2)(kz2kx2)(kx2ky2)(k_{y}^{2}-k_{z}^{2})(k_{z}^{2}-k_{x}^{2})(k_{x}^{2}-k_{y}^{2})
Eg+E^{+}_{g} E+E^{+} E+E^{+} QuQ_{u} 13(3kz2k2)\frac{1}{\sqrt{3}}(3k_{z}^{2}-k^{2})
QvQ_{v} kx2ky2k^{2}_{x}-k^{2}_{y}
T1g+T^{+}_{1g} T1+T^{+}_{1} T1+T^{+}_{1} Q4xαQ_{4x}^{\alpha} kykz(ky2kz2)k_{y}k_{z}(k_{y}^{2}-k_{z}^{2})
Q4yαQ_{4y}^{\alpha} kzkx(kz2kx2)k_{z}k_{x}(k_{z}^{2}-k_{x}^{2})
Q4zαQ_{4z}^{\alpha} kxky(kx2ky2)k_{x}k_{y}(k_{x}^{2}-k_{y}^{2})
T2g+T^{+}_{2g} T2+T^{+}_{2} T2+T^{+}_{2} QyzQ_{yz} kykzk_{y}k_{z}
QzxQ_{zx} kzkxk_{z}k_{x}
QxyQ_{xy} kxkyk_{x}k_{y}
A1uA^{-}_{1u} A1A^{-}_{1} A2A^{-}_{2} T9uT_{9u} kxkykz(kx2ky2)(ky2kz2)(kz2kx2)k_{x}k_{y}k_{z}(k_{x}^{2}-k_{y}^{2})(k_{y}^{2}-k_{z}^{2})(k_{z}^{2}-k_{x}^{2})
A2uA^{-}_{2u} A2A^{-}_{2} A1A^{-}_{1} TxyzT_{xyz} kxkykzk_{x}k_{y}k_{z}
EuE^{-}_{u} EE^{-} EE^{-} T5uT_{5u} 3kxkykz(kx2ky2)\sqrt{3}k_{x}k_{y}k_{z}(k_{x}^{2}-k_{y}^{2})
T5vT_{5v} kxkykz(3kz2k2)-k_{x}k_{y}k_{z}(3k_{z}^{2}-k^{2})
T1uT^{-}_{1u} T1T^{-}_{1} T2T^{-}_{2} TxT_{x} kxk_{x}
TyT_{y} kyk_{y}
TzT_{z} kzk_{z}
T2uT^{-}_{2u} T2T^{-}_{2} T1T^{-}_{1} TxβT_{x}^{\beta} kx(ky2kz2)k_{x}(k_{y}^{2}-k_{z}^{2})
TyβT_{y}^{\beta} ky(kz2kx2)k_{y}(k_{z}^{2}-k_{x}^{2})
TzβT_{z}^{\beta} kz(kx2ky2)k_{z}(k_{x}^{2}-k_{y}^{2})
Table 10: Multipoles under Laue class m3¯m\bar{3}.
ThT_{\rm h} TT MP basis functions
Ag+A^{+}_{g} A+A^{+} Q0Q_{0} 11
Eg+E^{+}_{g} E+E^{+} QuiQvQ_{u}-iQ_{v} 13(3kz2k2)i(kx2ky2)\frac{1}{\sqrt{3}}(3k_{z}^{2}-k^{2})-i(k^{2}_{x}-k^{2}_{y})
Qu+iQvQ_{u}+iQ_{v} 13(3kz2k2)+i(kx2ky2)\frac{1}{\sqrt{3}}(3k_{z}^{2}-k^{2})+i(k^{2}_{x}-k^{2}_{y})
Tg+T^{+}_{g} T+T^{+} QyzQ_{yz} kykzk_{y}k_{z}
QzxQ_{zx} kzkxk_{z}k_{x}
QxyQ_{xy} kxkyk_{x}k_{y}
AuA^{-}_{u} AA^{-} TxyzT_{xyz} kxkykzk_{x}k_{y}k_{z}
EuE^{-}_{u} EE^{-} T5uiT5vT_{5u}-iT_{5v} 3kxkykz(kx2ky2)+ikxkykz(3kz2k2)\sqrt{3}k_{x}k_{y}k_{z}(k_{x}^{2}-k_{y}^{2})+ik_{x}k_{y}k_{z}(3k_{z}^{2}-k^{2})
T5u+iT5vT_{5u}+iT_{5v} 3kxkykz(kx2ky2)ikxkykz(3kz2k2)\sqrt{3}k_{x}k_{y}k_{z}(k_{x}^{2}-k_{y}^{2})-ik_{x}k_{y}k_{z}(3k_{z}^{2}-k^{2})
TuT^{-}_{u} TT^{-} TxT_{x} kxk_{x}
TyT_{y} kyk_{y}
TzT_{z} kzk_{z}
Table 11: Multipoles under Laue class 4/mmm4/mmm. We take the xx ([110][110]) axis as the C2C_{2}^{\prime} rotation axis for D2dD_{\rm 2d} (D2dD_{\rm 2d}^{\prime}).
D4hD_{\rm 4h} D4D_{4} D2dD_{\rm 2d} D2dD^{\prime}_{\rm 2d} C4vC_{\rm 4v} MP basis functions
A1g+A^{+}_{1g} A1+A^{+}_{1} A1+A^{+}_{1} A1+A^{+}_{1} A1+A^{+}_{1} Q0Q_{0} 11
A2g+A^{+}_{2g} A2+A^{+}_{2} A2+A^{+}_{2} A2+A^{+}_{2} A2+A^{+}_{2} Q4zαQ_{4z}^{\alpha} kxky(kx2ky2)k_{x}k_{y}(k_{x}^{2}-k_{y}^{2})
B1g+B^{+}_{1g} B1+B^{+}_{1} B1+B^{+}_{1} B2+B^{+}_{2} B1+B^{+}_{1} QvQ_{v} kx2ky2k^{2}_{x}-k^{2}_{y}
B2g+B^{+}_{2g} B2+B^{+}_{2} B2+B^{+}_{2} B1+B^{+}_{1} B2+B^{+}_{2} QxyQ_{xy} kxkyk_{x}k_{y}
Eg+E^{+}_{g} E+E^{+} E+E^{+} E+E^{+} E+E^{+} QyzQ_{yz} kykzk_{y}k_{z}
QzxQ_{zx} kzkxk_{z}k_{x}
A1uA^{-}_{1u} A1A^{-}_{1} B1B^{-}_{1} B1B^{-}_{1} A2A^{-}_{2} T5uT_{5u} kxkykz(kx2ky2)k_{x}k_{y}k_{z}(k_{x}^{2}-k_{y}^{2})
A2uA^{-}_{2u} A2A^{-}_{2} B2B^{-}_{2} B2B^{-}_{2} A1A^{-}_{1} TzT_{z} kzk_{z}
B1uB^{-}_{1u} B1B^{-}_{1} A1A^{-}_{1} A2A^{-}_{2} B2B^{-}_{2} TxyzT_{xyz} kxkykzk_{x}k_{y}k_{z}
B2uB^{-}_{2u} B2B^{-}_{2} A2A^{-}_{2} A1A^{-}_{1} B1B^{-}_{1} TzβT_{z}^{\beta} kz(kx2ky2)k_{z}(k_{x}^{2}-k_{y}^{2})
EuE^{-}_{u} EE^{-} EE^{-} EE^{-} EE^{-} TxT_{x} kxk_{x}
TyT_{y} kyk_{y}
Table 12: Multipoles under Laue class 4/m4/m.
C4hC_{\rm 4h} C4C_{4} S4S_{4} MP basis functions
Ag+A^{+}_{g} A+A^{+} A+A^{+} Q0Q_{0} 11
Bg+B^{+}_{g} B+B^{+} B+B^{+} QvQ_{v} kx2ky2k^{2}_{x}-k^{2}_{y}
QxyQ_{xy} kxkyk_{x}k_{y}
Eg+E^{+}_{g} E+E^{+} E+E^{+} QyziQzxQ_{yz}-iQ_{zx} kykzikzkxk_{y}k_{z}-ik_{z}k_{x}
Qyz+iQzxQ_{yz}+iQ_{zx} kykz+ikzkxk_{y}k_{z}+ik_{z}k_{x}
AuA^{-}_{u} AA^{-} BB^{-} TzT_{z} kzk_{z}
BuB^{-}_{u} BB^{-} AA^{-} TxyzT_{xyz} kxkykzk_{x}k_{y}k_{z}
TzβT_{z}^{\beta} kz(kx2ky2)k_{z}(k_{x}^{2}-k_{y}^{2})
EuE^{-}_{u} EE^{-} EE^{-} Tx+iTyT_{x}+iT_{y} kx+ikyk_{x}+ik_{y}
TxiTyT_{x}-iT_{y} kxikyk_{x}-ik_{y}
Table 13: Multipoles under Laue class mmmmmm.
D2hD_{\rm 2h} D2D_{2} C2vC_{\rm 2v} MP basis functions
Ag+A^{+}_{g} A+A^{+} A1+A^{+}_{1} Q0Q_{0} 11
B1g+B^{+}_{1g} B1+B^{+}_{1} A2+A^{+}_{2} QxyQ_{xy} kxkyk_{x}k_{y}
B2g+B^{+}_{2g} B2+B^{+}_{2} B1+B^{+}_{1} QzxQ_{zx} kzkxk_{z}k_{x}
B3g+B^{+}_{3g} B3+B^{+}_{3} B2+B^{+}_{2} QyzQ_{yz} kykzk_{y}k_{z}
AuA^{-}_{u} AA^{-} A2A^{-}_{2} TxyzT_{xyz} kxkykzk_{x}k_{y}k_{z}
B1uB^{-}_{1u} B1B^{-}_{1} A1A^{-}_{1} TzT_{z} kzk_{z}
B2uB^{-}_{2u} B2B^{-}_{2} B2B^{-}_{2} TyT_{y} kyk_{y}
B3uB^{-}_{3u} B3B^{-}_{3} B1B^{-}_{1} TxT_{x} kxk_{x}
Table 14: Multipoles under Laue class 2/m2/m.
C2hC_{\rm 2h} C2C_{\rm 2} CsC_{\rm s} MP basis functions
Ag+A^{+}_{g} A+A^{+} A+A^{\prime+} Q0Q_{0} 11
Bg+B^{+}_{g} B+B^{+} A′′+A^{\prime\prime+} QzxQ_{zx} kzkxk_{z}k_{x}
Bg+B^{+}_{g} B+B^{+} A′′+A^{\prime\prime+} QyzQ_{yz} kykzk_{y}k_{z}
AuA^{-}_{u} AA^{-} A′′A^{\prime\prime-} TzT_{z} kzk_{z}
BuB^{-}_{u} BB^{-} AA^{\prime-} TyT_{y} kyk_{y}
BuB^{-}_{u} BB^{-} AA^{\prime-} TxT_{x} kxk_{x}
Table 15: Multipoles under Laue class 6/mmm6/mmm. For D6hD_{\rm 6h}, we take the yy and xx axes as the C2C_{2}^{\prime} and C2′′C_{2}^{\prime\prime} rotation axes, respectively com (b); Aroyo et al. (2006a, b). We take the xx (yy) axis as the C2C_{2}^{\prime} rotation axis for D3hD_{\rm 3h}^{\prime} (D3hD_{\rm 3h}). The sign and coefficient in two dimensional irrep. are chosen so as to satisfy the mutual relationship between two components.
D6hD_{\rm 6h} D6D_{6} C6vC_{\rm 6v} D3hD_{\rm 3h} D3hD_{\rm 3h}^{\prime} MP basis functions
A1g+A^{+}_{1g} A1+A^{+}_{1} A1+A^{+}_{1} A1+A^{\prime+}_{1} A1+A^{\prime+}_{1} Q0Q_{0} 11
A2g+A^{+}_{2g} A2+A^{+}_{2} A2+A^{+}_{2} A2+A^{\prime+}_{2} A2+A^{\prime+}_{2} Q6sQ_{6s} kxky(3kx2ky2)(kx23ky2)k_{x}k_{y}(3k_{x}^{2}-k_{y}^{2})(k_{x}^{2}-3k_{y}^{2})
B1g+B^{+}_{1g} B1+B^{+}_{1} B2+B^{+}_{2} A1′′+A^{\prime\prime+}_{1} A2′′+A^{\prime\prime+}_{2} Q4bQ_{4b} kzkx(kx23ky2)k_{z}k_{x}(k_{x}^{2}-3k_{y}^{2})
B2g+B^{+}_{2g} B2+B^{+}_{2} B1+B^{+}_{1} A2′′+A^{\prime\prime+}_{2} A1′′+A^{\prime\prime+}_{1} Q4aQ_{4a} kykz(3kx2ky2)k_{y}k_{z}(3k_{x}^{2}-k_{y}^{2})
E1g+E^{+}_{1g} E1+E^{+}_{1} E1+E^{+}_{1} E′′+E^{\prime\prime+} E′′+E^{\prime\prime+} QzxQ_{zx} kzkxk_{z}k_{x}
QyzQ_{yz} kykzk_{y}k_{z}
E2g+E^{+}_{2g} E2+E^{+}_{2} E2+E^{+}_{2} E+E^{\prime+} E+E^{\prime+} QvQ_{v} 12(kx2ky2)\frac{1}{2}(k_{x}^{2}-k_{y}^{2})
QxyQ_{xy} kxky-k_{x}k_{y}
A1uA^{-}_{1u} A1A^{-}_{1} A2A^{-}_{2} A1′′A^{\prime\prime-}_{1} A1′′A^{\prime\prime-}_{1} T7uT_{7u} kxkykz(3kx2ky2)(kx23ky2)k_{x}k_{y}k_{z}(3k_{x}^{2}-k_{y}^{2})(k_{x}^{2}-3k_{y}^{2})
A2uA^{-}_{2u} A2A^{-}_{2} A1A^{-}_{1} A2′′A^{\prime\prime-}_{2} A2′′A^{\prime\prime-}_{2} TzT_{z} kzk_{z}
B1uB^{-}_{1u} B1B^{-}_{1} B1B^{-}_{1} A1A^{\prime-}_{1} A2A^{\prime-}_{2} T3bT_{3b} ky(3kx2ky2)k_{y}(3k_{x}^{2}-k_{y}^{2})
B2uB^{-}_{2u} B2B^{-}_{2} B2B^{-}_{2} A2A^{\prime-}_{2} A1A^{\prime-}_{1} T3aT_{3a} kx(kx23ky2)k_{x}(k_{x}^{2}-3k_{y}^{2})
E1uE^{-}_{1u} E1E^{-}_{1} E1E^{-}_{1} EE^{\prime-} EE^{\prime-} TxT_{x} kxk_{x}
TyT_{y} kyk_{y}
E2uE^{-}_{2u} E2E^{-}_{2} E2E^{-}_{2} E′′E^{\prime\prime-} E′′E^{\prime\prime-} TzβT_{z}^{\beta} 12kz(kx2ky2)\frac{1}{2}k_{z}(k_{x}^{2}-k_{y}^{2})
TxyzT_{xyz} kxkykz-k_{x}k_{y}k_{z}
Table 16: Multipoles under Laue class 6/m6/m.
C6hC_{\rm 6h} C6C_{6} C3hC_{\rm 3h} MP basis functions
Ag+A^{+}_{g} A+A^{+} A+A^{\prime+} Q0Q_{0} 11
Bg+B^{+}_{g} B+B^{+} A′′+A^{\prime\prime+} Q4aQ_{4a} kykz(3kx2ky2)k_{y}k_{z}(3k_{x}^{2}-k_{y}^{2})
Q4bQ_{4b} kzkx(kx23ky2)k_{z}k_{x}(k_{x}^{2}-3k_{y}^{2})
E1g+E^{+}_{1g} E1+E^{+}_{1} E′′+E^{\prime\prime+} Qzx+iQyzQ_{zx}+iQ_{yz} kzkx+ikykzk_{z}k_{x}+ik_{y}k_{z}
QzxiQyzQ_{zx}-iQ_{yz} kzkxikykzk_{z}k_{x}-ik_{y}k_{z}
E2g+E^{+}_{2g} E2+E^{+}_{2} E+E^{\prime+} Qv+iQxyQ_{v}+iQ_{xy} 12(kx2ky2)+ikxky\frac{1}{2}(k_{x}^{2}-k_{y}^{2})+ik_{x}k_{y}
QviQxyQ_{v}-iQ_{xy} 12(kx2ky2)ikxky\frac{1}{2}(k_{x}^{2}-k_{y}^{2})-ik_{x}k_{y}
AuA^{-}_{u} AA^{-} A′′A^{\prime\prime-} TzT_{z} kzk_{z}
BuB^{-}_{u} BB^{-} AA^{\prime-} T3aT_{3a} kx(kx23ky2)k_{x}(k_{x}^{2}-3k_{y}^{2})
T3bT_{3b} ky(3kx2ky2)k_{y}(3k_{x}^{2}-k_{y}^{2})
E1uE^{-}_{1u} E1E^{-}_{1} EE^{\prime-} Tx+iTyT_{x}+iT_{y} kx+ikyk_{x}+ik_{y}
TxiTyT_{x}-iT_{y} kxikyk_{x}-ik_{y}
E2uE^{-}_{2u} E2E^{-}_{2} E′′E^{\prime\prime-} Tzβ+iTxyzT_{z}^{\beta}+iT_{xyz} 12kz(kx2ky2)+ikxkykz\frac{1}{2}k_{z}(k_{x}^{2}-k_{y}^{2})+ik_{x}k_{y}k_{z}
TzβiTxyzT_{z}^{\beta}-iT_{xyz} 12kz(kx2ky2)ikxkykz\frac{1}{2}k_{z}(k_{x}^{2}-k_{y}^{2})-ik_{x}k_{y}k_{z}
Table 17: Multipoles under Laue class 3¯m\bar{3}m. We take the xx (yy) axis as the C2C_{2}^{\prime} rotation axis for D3dD_{\rm 3d}^{\prime} and D3D_{\rm 3}^{\prime} (D3dD_{\rm 3d} and D3D_{\rm 3}). For D3dD^{\prime}_{3{\rm d}} and C3vC_{3{\rm v}} (D3dD_{3{\rm d}} and C3vC^{\prime}_{3{\rm v}}), we take the yzyz (xzxz) plane as the σv\sigma_{v} or σd\sigma_{d} mirror plane. The sign and coefficient in two dimensional irrep. are chosen so as to satisfy the mutual relationship between two components.
D3dD_{\rm 3d} D3dD_{\rm 3d}^{\prime} D3D_{3} D3D_{3}^{\prime} C3vC_{\rm 3v} C3vC_{\rm 3v}^{\prime} MP basis functions
A1g+A^{+}_{1g} A1g+A^{+}_{1g} A1+A^{+}_{1} A1+A^{+}_{1} A1+A^{+}_{1} A1+A^{+}_{1} Q0Q_{0} 11
A1g+A^{+}_{1g} A2g+A^{+}_{2g} A1+A^{+}_{1} A2+A^{+}_{2} A2+A^{+}_{2} A1+A^{+}_{1} Q4bQ_{4b} kzkx(kx23ky2)k_{z}k_{x}(k_{x}^{2}-3k_{y}^{2})
A2g+A_{2g}^{+} A1g+A_{1g}^{+} A2+A_{2}^{+} A1+A_{1}^{+} A1+A_{1}^{+} A2+A_{2}^{+} Q4aQ_{4a} kykz(3kx2ky2)k_{y}k_{z}(3k_{x}^{2}-k_{y}^{2})
Eg+E^{+}_{g} Eg+E^{+}_{g} E+E^{+} E+E^{+} E+E^{+} E+E^{+} QzxQ_{zx} kzkxk_{z}k_{x}
QyzQ_{yz} kykzk_{y}k_{z}
Eg+E^{+}_{g} Eg+E^{+}_{g} E+E^{+} E+E^{+} E+E^{+} E+E^{+} QvQ_{v} 12(kx2ky2)\frac{1}{2}(k_{x}^{2}-k_{y}^{2})
QxyQ_{xy} kxky-k_{x}k_{y}
A1uA^{-}_{1u} A2uA^{-}_{2u} A1A^{-}_{1} A2A^{-}_{2} A1A^{-}_{1} A2A^{-}_{2} T3bT_{3b} ky(3kx2ky2)k_{y}(3k_{x}^{2}-k_{y}^{2})
A2uA^{-}_{2u} A1uA^{-}_{1u} A2A^{-}_{2} A1A^{-}_{1} A2A^{-}_{2} A1A^{-}_{1} T3aT_{3a} kx(kx23ky2)k_{x}(k_{x}^{2}-3k_{y}^{2})
A2uA^{-}_{2u} A2uA^{-}_{2u} A2A^{-}_{2} A2A^{-}_{2} A1A^{-}_{1} A1A^{-}_{1} TzT_{z} kzk_{z}
EuE^{-}_{u} EuE^{-}_{u} EE^{-} EE^{-} EE^{-} EE^{-} TxT_{x} kxk_{x}
TyT_{y} kyk_{y}
Table 18: Multipoles under Laue class 33. C3i=S6C_{\rm 3i}=S_{6}.
C3iC_{\rm 3i} C3C_{3} MP basis functions
Ag+A^{+}_{g} A+A^{+} Q0Q_{0} 11
Eg+E^{+}_{g} E+E^{+} Qzx+iQyzQ_{zx}+iQ_{yz} kzkx+ikykzk_{z}k_{x}+ik_{y}k_{z}
QzxiQyzQ_{zx}-iQ_{yz} kzkxikykzk_{z}k_{x}-ik_{y}k_{z}
Eg+E^{+}_{g} E+E^{+} QviQxyQ_{v}-iQ_{xy} 12(kx2ky2)ikxky\frac{1}{2}(k_{x}^{2}-k_{y}^{2})-ik_{x}k_{y}
Qv+iQxyQ_{v}+iQ_{xy} 12(kx2ky2)+ikxky\frac{1}{2}(k_{x}^{2}-k_{y}^{2})+ik_{x}k_{y}
AuA^{-}_{u} AA^{-} TzT_{z} kzk_{z}
EuE^{-}_{u} EE^{-} Tx+iTyT_{x}+iT_{y} kx+ikyk_{x}+ik_{y}
TxiTyT_{x}-iT_{y} kxikyk_{x}-ik_{y}
Table 19: Multipoles under Laue class 1¯\bar{1}.
CiC_{\rm i} CC MP basis functions
Ag+A^{+}_{g} A+A^{+} Q0Q_{0} 11
AuA^{-}_{u} AA^{-} TxT_{x} kxk_{x}
AuA^{-}_{u} AA^{-} TyT_{y} kyk_{y}
AuA^{-}_{u} AA^{-} TzT_{z} kzk_{z}
Acknowledgements.
This research was supported by JSPS KAKENHI Grants Numbers JP15H05885, JP18H04296 (J-Physics), JP18K13488, JP19K03752, JP19H01834, and JP20K05299.

References

  • Rashba (1960) E. I. Rashba, Sov. Phys. Solid State 2, 1109 (1960).
  • Sinova et al. (2004) J. Sinova, D. Culcer, Q. Niu, N. A. Sinitsyn, T. Jungwirth, and A. H. MacDonald, Phys. Rev. Lett. 92, 126603 (2004).
  • Dresselhaus et al. (2008) M. S. Dresselhaus, G. Dresselhaus, and A. Jorio, Group Theory: Application to the Physics of Condensed Matter (Springer-Verlag, Berlin Heidelberg, 2008).
  • Ishizaka et al. (2011) K. Ishizaka, M. Bahramy, H. Murakawa, M. Sakano, T. Shimojima, T. Sonobe, K. Koizumi, S. Shin, H. Miyahara, A. Kimura, et al., Nat. mater. 10, 521 (2011).
  • Bahramy et al. (2011) M. S. Bahramy, R. Arita, and N. Nagaosa, Phys. Rev. B 84, 041202(R) (2011).
  • Zhu et al. (2011) Z. Y. Zhu, Y. C. Cheng, and U. Schwingenschlögl, Phys. Rev. B 84, 153402 (2011).
  • Wang et al. (2012) Q. H. Wang, K. Kalantar-Zadeh, A. Kis, J. N. Coleman, and M. S. Strano, Nat. Nanotech. 7, 699 (2012).
  • Ugeda et al. (2014) M. M. Ugeda, A. J. Bradley, S.-F. Shi, H. Felipe, Y. Zhang, D. Y. Qiu, W. Ruan, S.-K. Mo, Z. Hussain, Z.-X. Shen, et al., Nat. Mater. 13, 1091 (2014).
  • Kormányos et al. (2014) A. Kormányos, V. Zólyomi, N. D. Drummond, and G. Burkard, Phys. Rev. X 4, 011034 (2014).
  • Kimura et al. (2003) T. Kimura, T. Goto, H. Shintani, K. Ishizaka, T. Arima, and Y. Tokura, Nature 426, 55 (2003).
  • Fiebig (2005) M. Fiebig, J. Phys. D: Appl. Phys. 38, R123 (2005).
  • Khomskii (2009) D. Khomskii, Physics 2, 20 (2009).
  • Furukawa et al. (2017) T. Furukawa, Y. Shimokawa, K. Kobayashi, and T. Itou, Nat. Commun. 8, 954 (2017).
  • Saito et al. (2018) H. Saito, K. Uenishi, N. Miura, C. Tabata, H. Hidaka, T. Yanagisawa, and H. Amitsuka, J. Phys. Soc. Jpn. 87, 033702 (2018).
  • Hirsch (1999) J. E. Hirsch, Phys. Rev. Lett. 83, 1834 (1999).
  • Bernevig et al. (2006) B. A. Bernevig, T. L. Hughes, and S.-C. Zhang, Science 314, 1757 (2006).
  • Liu et al. (2012) L. Liu, C.-F. Pai, Y. Li, H. Tseng, D. Ralph, and R. Buhrman, Science 336, 555 (2012).
  • Sinova et al. (2015) J. Sinova, S. O. Valenzuela, J. Wunderlich, C. H. Back, and T. Jungwirth, Rev. Mod. Phys. 87, 1213 (2015).
  • Kézsmárki et al. (2014) I. Kézsmárki, D. Szaller, S. Bordács, V. Kocsis, Y. Tokunaga, Y. Taguchi, H. Murakawa, Y. Tokura, H. Engelkamp, T. Rõõm, et al., Nat. Commun. 5, 1 (2014).
  • Toyoda et al. (2016) S. Toyoda, N. Abe, and T. Arima, Phys. Rev. B 93, 201109(R) (2016).
  • Morimoto and Nagaosa (2016) T. Morimoto and N. Nagaosa, Phys. Rev. Lett. 117, 146603 (2016).
  • Ideue et al. (2017) T. Ideue, K. Hamamoto, S. Koshikawa, M. Ezawa, S. Shimizu, Y. Kaneko, Y. Tokura, N. Nagaosa, and Y. Iwasa, Nat. Phys. 13, 578 (2017).
  • Tokura and Nagaosa (2018) Y. Tokura and N. Nagaosa, Nat. Commun. 9, 1 (2018).
  • Aoki et al. (2019) R. Aoki, Y. Kousaka, and Y. Togawa, Phys. Rev. Lett. 122, 057206 (2019).
  • Ahn et al. (2019) K.-H. Ahn, A. Hariki, K.-W. Lee, and J. Kuneš, Phys. Rev. B 99, 184432 (2019).
  • Naka et al. (2019) M. Naka, S. Hayami, H. Kusunose, Y. Yanagi, Y. Motome, and H. Seo, Nat. Commun. 10, 4305 (2019).
  • Hayami et al. (2019a) S. Hayami, Y. Yanagi, and H. Kusunose, J. Phys. Soc. Jpn. 88, 123702 (2019a).
  • Hayami et al. (2020a) S. Hayami, Y. Yanagi, M. Naka, H. Seo, Y. Motome, and H. Kusunose, JPS Conf. Proc. 30, 011149 (2020a).
  • Hayami et al. (2020b) S. Hayami, Y. Yanagi, and H. Kusunose, Phys. Rev. B 101, 220403(R) (2020b).
  • Yuan et al. (2020) L.-D. Yuan, Z. Wang, J.-W. Luo, E. I. Rashba, and A. Zunger, Phys. Rev. B 102, 014422 (2020).
  • Berlijn et al. (2017) T. Berlijn, P. C. Snijders, O. Delaire, H.-D. Zhou, T. A. Maier, H.-B. Cao, S.-X. Chi, M. Matsuda, Y. Wang, M. R. Koehler, et al., Phys. Rev. Lett. 118, 077201 (2017).
  • Hayami et al. (2020c) S. Hayami, T. Okubo, and Y. Motome, arXiv:2005.03168 (2020c).
  • Jungwirth et al. (2016) T. Jungwirth, X. Marti, P. Wadley, and J. Wunderlich, Nat. Nanotech. 11, 231 (2016).
  • Baltz et al. (2018) V. Baltz, A. Manchon, M. Tsoi, T. Moriyama, T. Ono, and Y. Tserkovnyak, Rev. Mod. Phys. 90, 015005 (2018).
  • Suzuki et al. (2018) M.-T. Suzuki, H. Ikeda, and P. M. Oppeneer, J. Phys. Soc. Jpn. 87, 041008 (2018).
  • Hayami and Kusunose (2018) S. Hayami and H. Kusunose, J. Phys. Soc. Jpn. 87, 033709 (2018).
  • Hayami et al. (2018) S. Hayami, M. Yatsushiro, Y. Yanagi, and H. Kusunose, Phys. Rev. B 98, 165110 (2018).
  • Watanabe and Yanase (2018) H. Watanabe and Y. Yanase, Phys. Rev. B 98, 245129 (2018).
  • Kusunose et al. (2020) H. Kusunose, R. Oiwa, and S. Hayami, arXiv:2007.08039 (2020).
  • Gallego et al. (2016) S. V. Gallego, J. M. Perez-Mato, L. Elcoro, E. S. Tasci, R. M. Hanson, K. Momma, M. I. Aroyo, and G. Madariaga, J. Appl. Crystallogr. 49, 1750 (2016).
  • Kusunose (2008) H. Kusunose, J. Phys. Soc. Jpn. 77, 064710 (2008).
  • Kuramoto et al. (2009) Y. Kuramoto, H. Kusunose, and A. Kiss, J. Phys. Soc. Jpn. 78, 072001 (2009).
  • Santini et al. (2009) P. Santini, S. Carretta, G. Amoretti, R. Caciuffo, N. Magnani, and G. H. Lander, Rev. Mod. Phys. 81, 807 (2009).
  • Suzuki et al. (2017) M.-T. Suzuki, T. Koretsune, M. Ochi, and R. Arita, Phys. Rev. B 95, 094406 (2017).
  • Suzuki et al. (2019) M.-T. Suzuki, T. Nomoto, R. Arita, Y. Yanagi, S. Hayami, and H. Kusunose, Phys. Rev. B 99, 174407 (2019).
  • Hayami et al. (2019b) S. Hayami, Y. Yanagi, H. Kusunose, and Y. Motome, Phys. Rev. Lett. 122, 147602 (2019b).
  • Nomoto et al. (2016) T. Nomoto, K. Hattori, and H. Ikeda, Phys. Rev. B 94, 174513 (2016).
  • Sumita and Yanase (2020) S. Sumita and Y. Yanase, Phys. Rev. Research 2, 033225 (2020).
  • Dubovik and Tugushev (1990) V. Dubovik and V. Tugushev, Phys. Rep. 187, 145 (1990).
  • Kopaev (2009) Y. V. Kopaev, Physics-Uspekhi 52, 1111 (2009).
  • Yanase (2014) Y. Yanase, J. Phys. Soc. Jpn. 83, 014703 (2014).
  • Hayami et al. (2014) S. Hayami, H. Kusunose, and Y. Motome, Phys. Rev. B 90, 024432 (2014).
  • com (a) It is noted that the basis function belonging to the E+E^{\prime+} representation is expressed as (Qv,Qxy)(Q_{v},-Q_{xy}) so as to satisfy the mutual relationship between two components, although we adopt (Qv,Qxy)(Q_{v},Q_{xy}) following in Ref. Hayami et al., 2020b.
  • Naka et al. (2020) M. Naka, S. Hayami, H. Kusunose, Y. Yanagi, Y. Motome, and H. Seo, Phys. Rev. B 102, 075112 (2020).
  • Lee et al. (2014) M. Lee, E. S. Choi, X. Huang, J. Ma, C. R. Dela Cruz, M. Matsuda, W. Tian, Z. L. Dun, S. Dong, and H. D. Zhou, Phys. Rev. B 90, 224402 (2014).
  • Cardias et al. (2020) R. Cardias, A. Bergman, A. Szilva, Y. O. Kvashnin, J. Fransson, A. B. Klautau, O. Eriksson, and L. Nordström, arXiv:2003.04680 (2020).
  • com (b) When we take the xx and yy axes as the C2C_{2}^{\prime} and C2′′C_{2}^{\prime\prime} rotation axes, respectively, the irreps. (B1g,B2g,B1u,B2u)(B_{1g},B_{2g},B_{1u},B_{2u}) are replaced with (B2g,B1g,B2u,B1u)(B_{2g},B_{1g},B_{2u},B_{1u}).
  • Aroyo et al. (2006a) M. I. Aroyo, J. M. Perez-Mato, C. Capillas, E. Kroumova, S. Ivantchev, G. Madariaga, A. Kirov, and H. Wondratschek, Z. Kristallogr. 221, 15 (2006a).
  • Aroyo et al. (2006b) M. I. Aroyo, A. Kirov, C. Capillas, J. Perez-Mato, and H. Wondratschek, Acta Crystallogr., Sect A 62, 115 (2006b).