1 Introduction
The first two decades of the new millennium have witnessed the surge of topological states of matter, whose electronic structures can be categorized by topological invariants. The prediction [
1-
6] and realization [
7-
11] of two-dimensional (2D) and three-dimensional (3D) topological insulators (TIs) have caused a paradigm shift to predict, understand and make use of quantum materials based on the topology of their band structures. Started with the 2D TI, quantum spin Hall (QSH) effect realized based on HgTe/CdTe quantum well [
6,
7] has revealed to the world the fundamental novelty and potential application of topological materials. QSH state is insulating in the bulk but has a pair of one-dimensional (1D) conducting edge states protected by time-reversal symmetry. Electrons in the 1D edge states move without elastic backscattering by nonmagnetic impurities, holding potential for dissipationless spintronics. Likely, a 3D strong TI is also insulating in the bulk and has 2D gapless topological surface states (TSSs). 3D TI was first realized based on Bi−Sb alloys [
8] and then on Bi
2Se
3 family [
5,
9-
11]. The robustness of topological protection to the TSSs from nonmagnetic perturbations has been experimentally demonstrated [
12-
14], pointing to feasible electronic and spintronic applications. Importantly, based on magnetically doped Bi
2Te
3 films, quantum anomalous Hall (QAH) effect was realized [
15,
16], a milestone towards low-power-consumption electronics without the need for applied magnetic field.
Besides insulators, quantum materials can also be metals and semimetals according to the detailed band structure around the Fermi level. After TIs, topological semimetals emerged as novel states of matter with degenerate band crossing close to which the band dispersion can be described by the massless 3D Weyl and Dirac equations [
17-
23].In a Dirac semimetal (DSM), the conduction and valence bands touch at discrete (Dirac) points with linear dispersion, forming bulk (3D) Dirac fermions. Given broken time-reversal or inversion symmetry, 3D Dirac fermion can be separated in the momentum space into two Weyl fermions (chiral massless fermions as a description of neutrinos with neglected mass in high-energy physics), resulting into topological Weyl semimetals (WSMs). Such nontrivial electronic features could bring into novel electrical and thermal transport behaviors such as anomalous Hall effect (AHE), anomalous Nernst effect, chiral anomaly signified by negative magnetoresistance and non-saturating magnetoresistance (see Refs. [
17,
24-
26] for comprehensive reviews). Such properties come from the enhanced Berry curvature hosted in the Dirac/Weyl type band structure, which exhibits extreme responses to external stimuli such as magnetic field, voltage or current bias, temperature gradient and optical excitation.
There have been hundreds of materials predicted as 3D strong TI [
27-
29] and dozens of them have been experimentally verified, usually through direct observation of their TSS Dirac cones by angle-resolved photoemission spectroscopy (ARPES) [
30-
32]. By comparison, magnetic TIs, especially intrinsic magnetic TIs, are limited in the material candidates. So far there is only (MnBi
2Te
4)·(Bi
2Te
3)
n (
) family which has been intensively studied as an intrinsic magnetic TI [
33-
39]. There are also many materials predicted and demonstrated as DSMs and WSMs, most of which are time-reversal invariant and only few materials have been studied as magnetic WSMs [
17,
24-
26]. Recently, there appeared several layered material families with hexagonal/Kagome lattices which host Dirac cones gapped by ferromagnetic (FM)/antiferromagnetic (AFM) order, such as Fe
3Sn
2 family [
40]. In the 2D limit, these systems with gapped Dirac cones can be viewed as Chern insulating phase with quantized anomalous Hall conductance [
41], given the Fermi level is positioned in the Dirac gap. In this sense, these materials share the same topological characters (the Chern number
) as intrinsic magnetic TIs. However, for 3D materials, the band structure is complicated by the coexisting trivial bands which locate at the same energy region as the Dirac gap, rendering such materials in metal phase with coexisting trivial and nontrivial conduction. Consequently, we feel it more appropriate to term such materials as magnetic topological metals. While quantized transport response from the edge conduction can be realized in intrinsic magnetic TIs by tuning the Fermi level in the gap of both bulk and surface bands, there is always transport contribution from the trivial bands in magnetic topological metals no matter where the Fermi level is. It is noted that there is no strict theoretical picture describing topological metals since the metallicity does not come from band topology but trivial bands. We choose this term only to emphasize its distinction from intrinsic magnetic TIs and topological SMs.
In this review, we focus on the recent progress in the exploration of these various kinds of intrinsic magnetic topological materials, categorized mainly into three groups: intrinsic magnetic (TIs), magnetic Weyl/Dirac semimetals and other magnetic topological metals. We will present representative materials for these novel topological states of matter, pay special attention to their characteristic band features such as the gap of topological surface state Dirac cone, gapped bulk Dirac cone, Weyl nodal point/line and Fermi arc, as well as the exotic transport responses resulting from such band features. There are also other intrinsic magnetic topological states of matter which have been proposed theoretically, yet lacking affirmative experimental evidence, such as topological Möbius insulators [
42-
44]. We briefly discuss the opportunities to explore new states of matter and novel physical properties based on intrinsic magnetic topological materials.
2 Intrinsic magnetic topological insulator
Intrinsic magnetic TIs provide an excellent platform for the study of exotic quantum states, such as QAH states, chiral Majorana fermions, and axion states [
33-
39], arising from the interplay between band topology and magnetism. Among them, QAH effect is of fundamental importance in the field of spintronics due to its non-dissipative properties in transport. One approach to realize it is to find a 2D TI that comprises long-range magnetic order. Introducing magnetism into the 2D TI can break the time-reversal symmetry, such that one direction of spin channels will be canceled. Although QAH effect has been proposed theoretically in the last century [
41], it is until 2013 when quantized edge resistance (
) was experimentally observed on Cr-doped (Bi, Sb)
2Te
3 thin films [
15]. The chemical doping results into inhomogeneity in the band structure (gap, carrier density) and consequently extremely low quantization temperature. Therefore, intrinsic magnetic states of matter with uniform long-range magnetic order are highly desired.
As first discussed in the theoretical proposal of AFM TIs in 2010 [
52], both time-reversal symmetry
and fractional translation
are broken but the combination
is preserved in AFM TI, leading to a topologically nontrivial phase which shares with 3D strong TI similar topological
invariant and quantized magnetoelectric effect. The material realization of an intrinsic AFM TI was not initiated until 2017. “Magnetic extension” picture proposed that by inserting MnTe bilayer into the quintuple layer of Bi
2Te
3, the system tends to form septuple layers of MnBi
2Te
4, hosting a robust QAH state [
53,
54]. The material was first experimentally realized by molecular beam epitaxy (MBE) [
55]. Subsequent theoretical works revealed its colorful physics and properties [
56-
59]. Since the successful preparation of single crystal MnBi
2Te
4, the surge of intrinsic magnetic TIs based on MnBi
2Te
4 family started. Following the discovery of MnBi
2Te
4, a series of superlattices of this family were discovered, denoted as MnBi
2Te
4·(Bi
2Te
3)
n (
n = 1, 2, 3) [
60-
62]. In addition, we will briefly introduce other intrinsic magnetic TI candidates such as MnSb
2Te
4·(Bi
2Te
3)
n (
n = 1, 2) and EuSn
2As
2 families.
2.1 MnBi2Te4·(Bi2Te3)n
In 2013, Lee
et al. [
63] synthesized the polycrystalline powder of MnBi
2Te
4 by the flux-method. In 2017, from MBE growth of heterostructure composed of MnSe and Bi
2Se
3, it was found that the topological surface state of this structure is located on the surface of the whole system, rather than at the interface of the two materials like other topological heterostructures. It was realized that the layered structure of MnSe and Bi
2Se
3 is a new type of single crystal, MnBi
2Se
4. Such transformation is also applicable to MnBi
2Te
4[
53,
54,
64,
65], and it is MnBi
2Te
4 which is the focus of intrinsic magnetic TI study due to its desirable magnetic, electronic, and structural properties.
The structure of MnBi
2Te
4 was refined to be in the hexagonal space group
(No. 166) [
60]. Its minimum structural unit is composed of seven atomic layers with stacking order Te−Bi−Te−Mn−Te−Bi−Te, which is called a septuple-layer (SL) and the adjacent layers are bonded by van der Waals force, as shown in Fig.2(a). The unit cell of MnBi
2Te
4 is composed of three SLs stacked in the −A−B−C− fashion, and its lattice constant
is about
nm. Its Neel temperature
[
66], above which the AFM order is transformed into paramagnetic (PM) order [Fig.2(b)]. Neutron diffraction experiments point out that the ground state magnetic structure of MnBi
2Te
4 is the
A-type AFM phase [
66,
67]. The magnetic moment of each SL points out of plane, and the magnetic moments of adjacent layers are opposite. Of course, if Bi
2Te
3 quintuple-layers (QLs) is inserted between SLs, we can get MnBi
4Te
7, MnBi
6Te
10, and MnBi
8Te
13 superlattices [
60-
62,
68]. MnBi
4Te
7 can be regarded as a sandwich structure formed by inserting one QL into each SL. Similarly, MnBi
6Te
10 and MnBi
8Te
13 are formed by inserting two or three QLs in each SL respectively. Note that the space group of MnBi
2Te
4, MnBi
6Te
10, and MnBi
8Te
13 is
, but the space group of MnBi
4Te
7 is
. Since the distance between two SLs in MnBi
4Te
7 and MnBi
6Te
10 is larger than that in MnBi
2Te
4, their interlayer AFM coupling is weaker. The results of magnetic transport measurement show that the AFM-PM transition temperature of MnBi
4Te
7 is
[Fig.2(c)] and that of MnBi
6Te
10 is
[Fig.2(d)]. More interestingly, with further increasing SL spacing, the compound of MnBi
8Te
13 has become the first intrinsic FM TI with
[Fig.2(e)]. The lattice constants and magnetic transition temperatures of these different compounds are also summarized in Fig.2(f).
The band structure of MnBi
2Te
4, as the first intrinsic magnetic TI, has been intensively studied [
59,
69,
73-
76] and the TSS inside the bulk gap is the focus of attention. At the early stage, a sizable gap was found for the TSS Dirac cone with temperature-independent behavior [
59,
77,
78]. However, subsequent ARPES works with systematic photon-energy-dependent measurement and higher energy and momentum resolution have revealed the nearly gapless behavior of TSS [
69,
73-
76,
79-
83], showing sample and location dependence [Fig.3(a)]. Here we use the term “nearly gapless” to describe the experimental observation that the size of Dirac gap varying from being vanishing to dozens of millielectronvolts, being much smaller than expected by theoretical calculation [
56-
59]. Such behaviors suggest much reduced effective magnetic moments felt by the TSS, which may arise from surface magnetic reconstruction or TSS redistribution (extension to the bulk). Currently there are several proposed mechanisms which may lead to one of these two phenomena yet none of them has been experimentally validated. Please refer to our recent review for more details [
33].
Since the SLs and QLs in the heterostructure members of this family (MnBi
4Te
7, MnBi
6Te
10, MnBi
8Te
13) are combined by van der Waals forces, there are different terminations after cleaving the sample. As shown in Fig.3(b−d), the band structure on SL-termination is very similar to that of MnBi
2Te
4, and the band structure on QL- and double QL-terminations show hybridization features between the TSS and certain bulk bands [
70,
71,
79,
84-
92]. Again, no signature of sizable magnetic gap can be found for the TSS from all the different terminations of AFM members. The sizable magnetic gap of TSS was realized based on the SL-termination of FM MnBi
8Te
13, with the gap size decreasing monotonically with increasing temperature and closing right at the Curie temperature [
72].
Although the lack of sizable magnetic gap of TSS obscures the realization of topological quantized transport at high temperature (say, at the level of AFM transition temperature), QAH effect has indeed been realized at low temperature [1.4 K, Fig.3(e)] based on 5 QLs films of MnBi
2Te
4, key evidence of a 2D Chern insulator [
93]. The characteristics of an axion insulator state were also observed at zero magnetic field based on 6 SLs [
45]. Under a perpendicular magnetic field (
), characteristics of high-Chern-number quantum Hall effect without Landau levels and contributed by dissipationless chiral edge states are observed, indicating a high Chern number Chern insulator with
(9, 10 SLs) [
94]. The
A-type AFM configuration exhibits layer Hall effect in which electrons from the top and bottom layers deflect in opposite directions due to the layer-locked Berry curvature, resulting in the characteristic of the axion insulator state (6 SLs) [
95,
96]. We envision that half quantized Hall transport at the level of
can be realized based on the SL-termination of FM MnBi
8Te
13 with sizable magnetic TSS gap [
72].
2.2 MnSb2Te4·(Sb2Te3)n
Since the successful synthesis of MnBi
2Te
4·(Bi
2Te
3)
n single crystals, elemental substitutions have been explored in order to manipulate its magnetic and electronic properties. It turns out the Bi site can be completely substituted by Sb atoms. The resulting MnSb
2Te
4·(Sb
2Te
3)
n family of materials are currently under intensive investigation. Theoretically, this family (
) is also predicted to host similar AFM ground state and AFM TI phase [
98,
99], yet there lacks consistency between/among experiments and calculations on the exact magnetic ground state and band topology of MnSb
2Te
4 [
100-
106]. Notably, ARPES results reveal significant hole doping for all the members studied so far, leaving the detailed TSS Dirac cone structure not straightforward to study [
105,
107].
The crystal structure of MnSb
4Te
7 adapts a space group of
. The Mn layer constitutes a long-range magnetic order with moments along the
c direction [Fig.4(a, b)] [
107] (
A-type AFM with
). ARPES measurement also reveals hole doping for the band structure with expected Dirac cone located at
above the Fermi level [Fig.4(c)]. Pressure experiments and DFT calculations have revealed multiple topological phases corresponding to various magnetic structures and the emergence of superconductivity [Fig.4(d)] [
98,
106-
110]. Similar hole doping and multiple magnetic topological phases have also been found in MnSb
6Te
10, an FM member of this family at its ground state [Fig.4(f, g)] [
111]. Considering the universal electron doping behavior in MnBi
2Te
4·(Bi
2Te
3)
n family, it is natural to expect carrier tunability and magnetic manipulation based onthe mutual substitution of Sb and Bi in Mn(Bi, Sb)
2Te
4·((Bi, Sb)
2Te
3)
n series. In fact, a tunable TSS Dirac gap varying from being gapless to larger than
has been reported in Sb doped MnBi
2Te
4, with its gap size proportional to the doping level [
112].
Except MnBi
2Te
4·(Bi
2Te
3)
n and MnSb
2Te
4·(Sb
2Te
3)
n families, it is noted that MnBi
2Se
4 in the
space group shares the same magnetic and topological properties of MnBi
2Te
4. This phase turns out to be unstable in the bulk crystal form. Recent efforts have succeeded in synthesizing ultrathin films of MnBi
2Se
4 using nonequilibrium MBE [
113]. Its magnetic structure, however, deviates from the expected
A-type AFMz structure and the response of TSS Dirac cone to the magnetic order remains to be investigated.
2.3 EuM2X2 (M = metal; and X = Group 14 or 15 element)
EuSn
2As
2 belongs to the group of compounds with formula AM
2X
2 (A = alkali, alkaline earth, or rare earth cation; M = metal; and X = Group 14 or 15 element). Here we focus on the A = Eu compounds with intrinsic AFM order. The M site can be occupied by various types of metals such as Mg, In and Sn. EuSn
2As
2, as an important member in intrinsic magnetic TI family, crystallizes in the hexagonal space group
. The Eu atoms are triangularly distributed and sandwiched by two honeycomb SnAs layers to form a layered structure [Fig.5(a)]. The magnetic moment provided by Eu atom forms an
A-type AFM configuration with
[
74,
114] [Fig.5(b)]. ARPES measurements have revealed a TSS Dirac cone locating ~ 0.4 eV above the Fermi level at the PM phase, suggesting a strong 3D TI phase [Fig.5(c)]. Yet no observable change of the TSS or carrier concentration can be found in the AFM state, indicating weak coupling between the Eu moments and low-energy bands [
74,
121]. Magnetic property and transport measurements report negative magnetoresistance and complicated magnetic transitions from an AFM state to a canted ferromagnetic state and then to a polarized FM state as the magnetic field increases [
121,
122]. Electrical resistance measurements under pressure reveal an insulator-metal-superconductor transition at low temperature around 5 and 15 GPa [Fig.5(d)]. A new
phase appears when the pressure is higher than 14 GPa. As the pressure continues to increase, the superconductivity persists up to 30.8 GPa with
maintaining a constant value ~ 4 K [
119]. It is also found that the pressure has an enhancement effect on the AFM transition temperature and negative magnetoresistance [
123].
For EuMg
2Bi
2, it crystallizes into the tetragonal CaAl
2Si
2 structure type with space group
(No. 164) [
117] [Fig.5(e)]. Magnetic property measurements revealed AFM transition temperature
with slight anisotropy and positive Curie-Weiss temperature indicating FM interaction between Eu atoms (
) [Fig.5(f)]. Like Mn−Bi−Te family, AFM configuration between FM layers of Eu is established. The difference is that the moments point out-of-plane in Mn−Bi−Te but in-plane for EuMg
2Bi
2. ARPES measurements and DFT calculations have revealed Dirac surface state features and nontrivial band topology [Fig.5(g, h)], suggesting EuMg
2Bi
2 as a magnetic topological insulator candidate [
115,
117].
EuIn
2As
2 crystallizes into the hexagonal space group
, containing layers of Eu
2+ cations separated by In
2As
22− layers along the crystallographic
-axis [
124] [Fig.5(i)]. Magnetic property and neutron diffraction measurements have determined a colinear AFM ground state with the moments lying in the
-plane [
118,
120,
124,
125] [Fig.5(j)]. Furthermore, a complicated broken helix order is reported by neutron diffraction, tripling the unit cell along
-axis. EuIn
2As
2 was predicted as a high-order topological axion insulator candidate [
120,
126]protected by the magnetic crystalline symmetry. Such a state has gapless TSS Dirac cone at the symmetry-protected termination and gapped ones at other surfaces [Fig.5(k−l)]. However, like other AM
2X
2 compounds, its hole-doping nature as observed by ARPES [
116,
118] and scanning tunneling microscope (STM) [
127] has prevented the detailed study on the TSS band structure, especially the gap behavior. Further chemical and band structure engineering are strongly called for to tune the chemical potential for access to the TSS Dirac point in this family.
There are also theoretical calculations which predict materials such as several Eu
5M2X6 (
M = metal,
X = pnictide) Zintl compounds [
128,
129], 2D EuCd
2Bi
2 [
130], and NiTl
2S
4 [
131] to be intrinsic magnetic TI candidates yet their growth, band structure, magnetic structure and band topology remain to be investigated.
3 Magnetic Weyl/Dirac semimetals
In a DSM, two doubly degenerate bands contact at discrete momentum points called Dirac points and disperse linearly along all directions around these points. The four-fold degenerate Dirac points need symmetries to ensure their existence, such as time-reversal symmetry
, inversion symmetry
, rotational symmetry and nonsymmorphic symmetry. In a DSM with
symmetry, when either
or
is broken, each doubly degenerate band is lifted, so that the Dirac cones can split into multiple Weyl cones, giving birth to WSMs. However, in 3D systems with AFM order that breaks both
and
but respect their combination
, four-fold degenerate Dirac points can still exist, resulting into AFM DSM [
132]. Such consideration has also been generalized to 2D systems [
133-
135].
In magnetic WSMs, spin-polarized conduction and valence bands touch at finite number of nodes, forming pairs of Weyl nodes. In each pair, the quasiparticles carry opposite chirality and can be viewed as the “source” (“+” chirality) and the “sink” (“−” chirality) of the Berry curvature. Odd pairs of Weyl nodes with opposite chirality can be expected in systems with
symmetry breaking, such as Co
3Sn
2S
2 [
136,
137] and Mn
3X (X = Sn, Ge) [
138-
140]; while for systems with time-reversal symmetry
, the total number of Weyl nodes pairs must be even. Noncentrosymmetric WSMs belong to this category, such as TaAs family [
141-
144]. If
and
symmetries are both preserved, Weyl nodes with opposite chirality can merge at the same momentum and form a four-fold Dirac point (assisted by additional crystal symmetry), such as Na
3Bi [
21,
145] and Cd
3As
2 [
23,
146,
147]. Due to non-zero Berry curvature, many novel physical properties such as giant AHE and giant anomalous Hall angle, chiral anomaly, anomalous Nernst effect will emerge in magnetic WSM, holding potential applications in spintronics field. In the early stage, several candidate materials were predicted, such as R
2Ir
2O
7 (R = Nd, Pr) [
20], HgCr
2Se
4 [
148]. Recent efforts have focused on Co
3Sn
2S
2 [
136,
137] and Mn
3X (X = Sn, Ge) [
138-
140] which clearly host the band structure and transport characters as expected by magnetic WSM. We will briefly introduce these magnetic materials.
3.1 FeSn
FeSn crystallizes in a hexagonal structure (
) with the Fe atoms forming a Kagome lattice [
150,
151,
153,
154]. Like Fe
3Sn
2, FeSn is formed by interlacing Fe
3Sn layer and Sn layer. The difference is that there is only one Kagome layer (Fe
3Sn layer) in a unit cell [Fig.6(a)]. It is closer to the 2D limit than Fe
3Sn
2. Below
[Fig.6(b)], the Fe spins form FM Kagome layers which are stacked antiferromagnetically along the
axis. The Dirac nodal line along the K−H line opens small energy gaps when SOC is considered, except at the H point where a gapless Dirac point (protected by
and
symmetry) still exist, rendering FeSn as an AFM DSM. Such gapless Dirac cones have been directly observed by ARPES [
150,
151] [Fig.6(c, d)]. Besides, the flat band because of the Kagome layer has also been observed directly by ARPES [Fig.6(e)]. Furthermore, in a planar tunneling spectroscopy measurement [
152], an anomalous enhancement in tunneling conductance within a finite energy range of FeSn has been observed in its Schottky heterointerface with Nb-doped SrTiO
3 [Fig.6(f)]. Such tunneling conductance peak is attributed to spin-polarized flat band localized at the FM Kagome layer at the Schottky interface.
3.2 Co3Sn2S2
Co
3Sn
2S
2 crystallizes in the
space group with a stacking order −Sn−S−Co
3Sn−S− from top to bottom. The central Co layer forms a 2D Kagome lattice with one Sn atom at the center of the hexagon, as shown in Fig.7(a). Co
3Sn
2S
2 is a ferromagnet with a curie temperature of 175 K and a magnetic moment of
. In magnetization measurement, the saturation field along
axis is low (0.05 T) but along in-plane is extremely high (> 9 T), confirming that the easy magnetic axis is
-axis [
48,
155]. Combining theory and experiments, Co
3Sn
2S
2 is an ideal FM WSM with three pairs of Weyl points whose energies are only ~ 60 meV above the Fermi level [
136,
137,
156-
161]. The Weyl nodes have been observed by ARPES after doping alkaline metal [Fig.7(c)]. Three Fermi arcs form a triangular-like loop around the K′ point near Fermi surface. Meanwhile, the electronic structure does not undergo obvious dispersion along the
direction, suggesting the nature of TSSs [Fig.7(b)]. Termination-dependent surface band structures of Co
3Sn
2S
2 were observed by using STM [
136]. Different surface potentials imposed by three different terminals will change the Fermi arc contour and Weyl node connectivity. On the Sn-termination, the Fermi arcs connect Weyl nodes within the same Brillouin zone, while on the Co-termination, the connectivity spans the two adjacent Brillouin zones. On S-termination, Fermi arcs overlap with the trivial surface-projected bulk bands. The topologically protected and unprotected electronic properties of WSMs Co
3Sn
2S
2 were verified.
According to first-principles calculation, the Weyl nodes in Co
3Sn
2S
2 locate close to the Fermi level and produce a giant anomalous Hall conductivity (AHC) (~1100 Ω
−1·cm
−1), which has been directly observed in transport measurement [Fig.7(d, e)] [
48,
161,
162]. Besides, giant anomalous Hall angle also emerges in this material. As shown in Fig.7(e), with increasing temperature, a maximum value of nearly 20% is reached around 120 K, which is at least one order of magnitude higher than that of conventional magnetic materials. Negative magnetoresistance is found in Co
3Sn
2S
2, as shown in Fig.7, when the magnetic field is applied in the in-plane direction, the longitudinal resistance is negative, and when the external magnetic field is applied in the out-of-plane direction, the longitudinal resistance changes from negative to positive, showing evidence of chiral anomaly [
48,
155,
161,
162]. In Co
3Sn
2S
2 thin film, a maximum Nernst thermopower of 3 µV·K
−1 is achieved [
50], demonstrating the possibility of application of hard magnetic topological semimetals for low-power thermoelectric devices.
3.3 Mn3X (X = Sn, Ge)
Mn
3X (X = Sn, Ge) has a hexagonal Ni
3Sn-type structure and crystalizes in the
space group. One unit cell consists of two sets of Mn layers stacked along
-axis and each Mn layer forms a breathing-type Kagome lattice with one Sn atom at the center of the hexagon, as shown in Fig.8(a). Mn
3Sn and Mn
3Ge are both chiral antiferromagnets which means Mn moments are forming a 120° ordering with a negative vector chirality [Fig.8(b)] [
164,
165]. The AFM transition temperature of Mn
3Sn and Mn
3Ge is
and
, respectively. Because the electronic structures of Mn
3Sn and Mn
3Ge are quite similar and the study of Mn
3Sn is more comprehensive, we will mainly focus on Mn
3Sn. Mn
3Sn possesses non-collinear AFM spin texture and strong SOC effect, which produce multiple pairs of Weyl points close to the Fermi level, according to first-principal calculation [
138,
166,
167] [Fig.8(c)]. However, ARPES spectra [
138] measured on Mn
3Sn lacks clear features of quasiparticle bands, likely due to strong correlation effect of Mn
electrons [Fig.8(d)].
Novel transport properties governed by the topological nature can serve as evidence for Weyl fermions. In Mn
3Sn, strongly anisotropic magnetoconductance was observed. The sign of magnetoconductance changed when rotated the direction of magnetic field from being parallel to perpendicular to the current direction, serving as strong evidence of chiral anomaly [
138,
164]. The large AHE is also a key characteristic of magnetic WSM. In the traditional sense, because the magnetic configuration of Mn
3Sn is AFM, there is no net magnetic moment in this material and the AHE will not emerge. But many reports revealed that Mn
3Sn exhibits a large AHE [
139,
140,
166,
168-
171]. Fig.8(e, f) show the temperature-dependence of zero-field Hall conductivity under different magnetic field and current directions [
140]. We can see that when the magnetic field and the current are applied along the (
) and (
) direction, the
will achieve a maximum value of nearly 130 Ω
−1·cm
−1 at
. In Mn
3Ge, by employing similar magnetic field and current direction, even higher AHC have been obtained [
139,
168]. The large AHE in Mn
3X is mainly caused by the non-zero Berry curvature produced by Weyl nodes [
164]. Besides chiral anomaly and large AHE, many other exotic physical properties such as large anomalous Nernst effect [
163,
164,
172-
174], planer Hall effect [
171,
175,
176], magnetic spin Hall effect and magnetic inverse spin Hall effect [
177] are also observed in Mn
3X. Furthermore, as shown in Fig.8(g), anomalous Nernst voltage
image mapped by scanning thermal gradient microscopy reveals the existence of magnetic domains. The orientation of these domains can be changed (written) by laser-induced local thermal gradient [
163], offering a chance to study spintronics phenomena in non-collinear antiferromagnets with spatial resolution.
3.4 Co2MnGa
A new family of magnetic WSM emerged among the magnetic Heusler alloys, i.e., the Heusler alloy WSMs [
181,
182]. It is an important family due to their rich transport properties and several superiorities. Firstly, the Curie temperatures of most Heusler compounds are above the room temperature [
183,
184]. Secondly, this kind of materials has a significant AHE and spin Hall effect arising from the large Berry curvature [
179,
181,
182,
185-
187]. Thirdly, Heusler compounds are usually soft magnetic materials, which means that their magnetization direction can be tuned by a weak magnetic field. These properties facilitate spin manipulation and applications in spintronics, as a result, these Heusler alloy WSMs have been widely studied.
As full Heusler compounds, Co-based Heusler materials have the formula of Co
2XZ (X = IVB or VB; Z = IVA or IIIA), here we focus on Co
2MnGa and Co
2MnAl. Co
2MnGa (Co
2MnAl) crystalizes in a face-centered cubic Bravais lattice (space group
, No. 225), as shown in Fig.9(a). The relevant symmetries are the three mirror planes and three
rotation axes. The Curie temperature of Co
2MnGa and Co
2MnAl are known to be ~700 K [
179] and 726 K [
183], respectively. Transport experiments showed that Co
2MnAl has a giant room-temperature AHE with the Hall angle (
) reaching a record value tan
= 0.21 at the room temperature among magnetic conductors [
178], as shown in Fig.9(b). This property results from the gapped nodal rings that generate large Berry curvature. Furthermore, for Co
2MnGa films, when the
EF is set in the magnetization-induced gap of the Weyl cones by the electronic doping, the highest anomalous Nernst thermopower of a record value 6.2 μV·K
−1 will be reached at room temperature [
188].
The Hopf link is originally a mathematical concept which consists of two rings on the two perpendicular planes, each passing through the center of each other, as shown in Fig.9(c). The symmetry of Co
2MnGa can protect this band crossing associated with the unusual linking-number (knot theory) invariant, giving rise to a variety of new types of topological semimetals [
179,
180,
185-
187,
189-
192]. Systematic ARPES investigation of the electronic structure of Co
2MnGa has been carried out and directly revealed three intertwined degeneracy loops in the material’s three-torus bulk Brillouin zone [Fig.9(d−f)]. In addition, the Seifert boundary states protected by the bulk-linked loops have been predicted and observed, while the links and knots in the electronic structure and the accompanied exotic behaviors remain unexplored.
3.5 EuB6
The EuB
6 crystallizes in a similar body-centered-cubic-like crystal structure with space group
(No. 221) [Fig.10(a)]. EuB
6 is a soft FM semimetal which has a very small magnetic anisotropy energy so that the magnetization can be easily modulated by magnetic field [
193,
195-
197]. Electronic transport and magnetic susceptibility measurements showed that the system undergoes a paramagnetic to FM phase transition at about
and a new FM phase manifests below about
with moment oriented to the (111) direction [
198-
200]. The magnetotransport properties of EuB
6 have been widely studied around magnetic phase transition point, such as the metal-insulator transition, colossal magnetoresistance and quantum nematic phase [
201-
203].
It has been predicted that EuB
6 is a topological nodal-line semimetal when the magnetic moment is aligned along the (001) direction, and it turns out to be a WSM with three pairs of Weyl nodes when rotating the magnetic moment to (111) direction. Specifically, when the moment is in the (110) direction, a composite semimetal phase featuring the coexistence of a nodal line and Weyl points manifests [
195]. The electronic structures on the two different cleavage planes in EuB
6, i.e., the Eu- and B-terminated surfaces, have been investigated [
194,
204]. For the B-termination, in the FM state, obvious Zeeman splitting occurs for both the conduction and valence bands, which gives rise to the overlap of subbands and thus the band inversion at the time-reversal point X of the Brillouin zone [Fig.10(b)]. In this case, EuB
6 enters a topological semimetal state with an ideal electronic structure near
EF. The topological properties can be investigated by measuring the magnetotransport properties due to the correlation between the band structure and the local moments. Fig.10(c) shows the intrinsic anomalous Hall conductivity as a function of magnetization with different directions at 2 K [
193]. An intrinsic large anisotropic magnetoresistance of −18% at 0.2 T was observed and interpreted as the modification from the Berry curvature in a tilted Weyl cone [
205]. The theoretical prediction that a large-Chern-number QAH effect could be realized in its (111)-oriented quantum-well structure [
195] needs further investigations.
3.6 Fe3GeTe2
Fe
3GeTe
2 crystallizes in a hexagonal structure (
P6
3/
, No. 194) in which the layered Fe
3Ge substructure are sandwiched by two layers of Te atoms [Fig.11(a)]. Fe
3GeTe
2 is FM with Fe moments along the
axis and a Curie temperature of 204−230 K [Fig.11(b)] [
206,
211-
213]. ARPES measurements have revealed two pockets around
point and one at
point [Fig.11(c)]. Temperature-dependent ARPES spectra exhibits a massive spectral weight transfer in the FM state induced by exchange splitting [
207]. Orbital-driven nodal line along K−H protected by crystalline symmetry has been predicted [Fig.11(f)]. Introducing SOC will gap the nodal line and generate large Berry curvature [
210], an effective source of a large AHE in Fe
3GeTe
2. We note that Fe
3GeTe
2 is considered as a gapped nodal line semimetal with the Weyl point awaiting verification.
Fe
3GeTe
2 also contains very rich physical properties. Due to the gapped nodal line, negative magnetoresistance[
214,
215], anomalous Nernst effect [
216] and AHE were observed [
209,
210,
213]. Compared with other itinerant FM materials, Fe
3GeTe
2 has both large anomalous Hall factor and anomalous Hall angle [Fig.11(d, e)]. Due to the weak interlayer coupling, Fe
3GeTe
2 can be exfoliated into sheets with different number of layers. More importantly, its novel transport and magnetic properties show stability at room temperature and dependence on the number of layers, interlayer coupling and carrier density [
208,
209,
217-
223], holding potential in spintronics applications.
3.7 EuCd2As2
EuCd
2As
2 belongs to EuM
2X
2 (M = metal; and X = Group 14 or 15 element) family in which several members are studied as magnetic TI candidates (see Section 2).The exact band structure details and topological phase are sensitively related to the magnetic configuration. The crystal structure of EuCd
2As
2, with space group 164 (
), is shown in Fig.12. The Eu atoms form a simple hexagonal lattice at the 1a Wyckoff position. The As and Cd atoms at the 2b positions form the other four atomic layers with the sequence of −Cd−As−Eu−As−Cd− along the
c axis [
224,
229,
230]. Eu moments prefer an intralayer FM coupling and an interlayer AFM coupling along the
axis, i.e., an
A-type AFM, which doubles the unit cell along the
direction. Fig.12 and (c) show two such magnetic configurations by showing Eu atoms with magnetic moment directions along
(
A-type AFMc) and along
(
A-type AFMa).
A-type AFMc is proposed based on the anisotropic magnetic and transport properties [
224,
229].
A-type AFMa is proposed based on the resonant elastic X-ray scattering [
225,
231], first-principles calculations [
232] and magnetostriction measurements [
233]. Furthermore, neutron diffraction on isotopic
153Eu and
116Cd revealed a
FM order at zero field with the Eu moments pointing along the in-plane (210) direction with a
out-of-plane canting [magnetic space group
, Fig.12] [
226].
According to the first-principles calculation and symmetry analysis, various topological phases emerge based on different magnetic configurations in EuCd
2As
2. For
A-type AFMz, DSM phase exists with the gapless Dirac point protected by the
symmetry operation which is the product of inversion symmetry
, time reversal symmetry
and crystalline translation symmetry
L [
227,
234]. For
A-type AFMx, spin configuration breaks the
symmetry in the AFM state of EuCd
2As
2 and leads to an axion insulator with a hybridization gap of ~1 meV. Massless Dirac surface states appear on some surfaces protected by the mirror or
symmetries. For other surfaces without such symmetry, the surface states are gapped and the hinge states, associated with higher order TI states, emerge at the edges [
126,
235]. There are other calculations which predict EuCd
2As
2 as a WSM with a single pair of Weyl points very close to the Fermi level [
226,
230,
236]. Such Weyl phase can be generated in EuCd
2As
2 by applying a magnetic field
along the
axis [
236] or alloying with Ba at the Eu site to stabilize the FM configuration [
230]. In fact, the recently confirmed spin-canted structure as shown in Fig.12 naturally hosts such WSM phase [
226]. Spectroscopically, ARPES measurements have observed linear band crossings at the Fermi level and especially an “M”-shaped feature around
point [Fig.12], suggesting a nontrivial band inversion. Such features cannot distinguish between the semimetal and insulator phase as the gap is only ~1 meV, comparable to the thermal broadening effect at ~3 K. ARPES or scanning tunneling spectroscopy measurements at ultralow temperature are needed. Spin-resolved ARPES is also useful to examine the spin degeneracy of these linear bands and crossings.
Magnetic transport experiments have provided more information on the interplay between magnetism and band topology in EuCd
2As
2. Negative magnetoresistance [Fig.12], as signature of chiral anomaly is observed along with AHE [Fig.12] [
225,
226,
228]. These transport results support as-grown EuCd
2As
2 in a semimetal phase, yet gate tunable transport is needed to verify the absence of gap close to the Fermi level. It was further reported that the Hall resistance shows a giant nonlinear behavior originating from a series of magnetic-field-induced Lifshitz transitions in the spin-dependent band structure [Fig.12] [
228]. Combined with band structure calculation, these results suggest that in EuCd
2As
2, electronic structure is extremely sensitive to the spin canting angle, with the magnetic field causing band crossing and band inversion and introducing a band gap when oriented along specific directions, offering an ideal platform for Berry curvature engineering.
4 Other magnetic topological metals
As introduced in the previous section, intrinsic magnetic TIs have nontrivial bulk band topology featured by a global bulk gap and TSS residing inside the bulk gap. Chemical potential can be tuned into the bulk gap to eliminate the transport contribution from the bulk bands, a key prerequisite to realize quantized Hall transport. There exist other magnetic systems which lack a global bulk gap in the whole momentum space but possess a locally nontrivial bulk gap and TSS inside. Such systems always exhibit metallic transport behavior contributed by trivial bulk bands. AHE is generally expected from the coexisting net magnetic moment and locally nontrivial topology. We term such materials as magnetic topological metals. It is noted that there is no strict theoretical scheme describing magnetic topological metal since the metallicity does not only come from band-topology-induced TSS but rather the trivial bulk bands. We choose this term only to emphasize its distinction from intrinsic magnetic TIs and topological SMs.
4.1 Fe3Sn2
Fe
3Sn
2 is a layered Kagome compound with a space group of
formed by interlacing two Fe
3Sn layers and one Sn layer. The Fe atoms in the Fe
3Sn layer form a Kagome structure, and the Sn atoms exhibit a honeycomb structure. The Sn atomic layer also exhibits a honeycomb distribution [Fig.13] [
239]. Fe
3Sn
2 is FM in the ground state with a Curie temperature of
~ 610 K [
237,
240-
242]. Due to the weak binding force between layers, Fe
3Sn
2 produces three different cleavage planes, Fe
3Sn-1-termination, Fe
3Sn-2-termination, and Sn-termination [
238,
243]. The experimentally observed band structures mainly come from Fe
3Sn-1-termination. The shape of the Fermi surface confirms the trigonal structure of Fe
3Sn
2. ARPES measurements have revealed two Dirac cone features at the corner of Brillouin zone, which are gapped by the SOC effect [Fig.13]. Such strong SOC also couples the magnetic and electronic structure of Kagome lattice, exhibiting a magnetization-driven giant nematic (two-fold-symmetric) energy shift [
244]. In the Kagome lattice, the destructive interference of the electron Bloch wave function can effectively localize the electrons to produce flat bands. Such flat bands are observed in Fe
3Sn
2, which are ~ 0.2 eV below the Fermi level [
243].
The coexistence of nontrivial band topology and FM order in Fe
3Sn
2 produces giant AHE [
40,
238,
245].The measured AHC is found to be temperature independent and persists above room temperature [Fig.13], which is suggestive of prominent Berry curvature from the time-reversal-symmetry-breaking electronic bands of the Kagome plane. Moreover, Fe
3Sn
2 shows complex magnetic bubbles and magnetic vortex structure like skyrmions[
246-
251]. These bubbles are 3D magnetic domains with complicated evolution of spin texture, which not only give rise to topological Hall transport response, but also show record-high temperature stability in magnetic racetrack memory devices [Fig.13].
4.2 RT6X6 (R = Rare earth metal; T = transition metal; X = Sn, Ge)
Layered Kagome compounds RT6X6 (R = rare earth metal, T = transition, alkali, alkaline earth metal, X=Sn or Ge) crystallize in the space group. As shown in Fig.14, T3X is the Kagome layer of T ions with one X atom at the center of the hexagon. In RX layer, the R atom lies at the center of the hexagons surrounded by the X atoms. X layer is a hexagonal layer only consisting of X atoms and separating each unit cell. In this system, the 4f electrons in the R element interact with the 3d electrons in the transition metal element T to generate a rich magnetic structure. Many novel physical properties are also found in this system, such as flat band, giant AHE and Nernst effects. Recent published articles focus mostly on RMn6Sn6 and RV6Sn6. Therefore, the following content will discuss these two systems.
Since Mn is a well-known magnetic metal, there are many magnetic configurations emerged due to the interaction between Mn 3d magnetic moment and
R 4f magnetic moments [Fig.14(b, e)] [
252,
255-
258]. When
R is a lanthanide element (
R = Gd−Tm, Lu), its magnetic configuration varies from FM to AFM. For
R = Gd to Ho, their magnetic configuration is ferrimagnetic, and for
R = Er, Tm and Lu, they possess AFM ground state. The direction of the magnetic moment of the
R element tends to be antiparallel to the magnetic moment of Mn, and the moment direction is variable for different
R elements. GdMn
6Sn
6 moment is arranged in-plane, and TbMn
6Sn
6 moment is arranged out-of-plane. DyMn
6Sn
6 and HoMn
6Sn
6 possess a conical magnetic structure. When
R is Er and Tm, the Mn and Er = Tm sublattices are independently ordered in an AFM manner because the strength of the magnetic coupling is weak. Since there is no 4f electrons in Lu and Y, they form in-plane FM and helical AFM along
c-axis. For
R = Gd to Ho, the Curie temperature of them is 435, 423, 393, and 376 K, respectively. For
R = Er to Lu and Y, the Neel temperature of them is 352, 347, 353, and 333 K, respectively. In general, the electronic structure is closely related to magnetic configuration, when magnetic configuration change, the electronic structure will also change. However, for the
RMn
6Sn
6 (
R = Gd−Tm, Lu, Y) system, even for the different
R, the band structure does not change significantly, indicating weak coupling between the low energy bands and magnetic moments.
Kagome lattice usually hosts three typical band features: flat band over the whole Brillouin zone, Dirac cones located at the Brillouin zone corners, and the saddle points located at the Brillouin zone boundary. Such features have indeed been observed in YMn
6Sn
6 and others by ARPES [
259,
260]. The strong correlation between magnetism and Kagome lattice can produce many novel physical properties. In TbMn
6Sn
6, its Kagome lattice features an out-of-plane magnetic ground state, so it is predicted to support the intrinsic Chern topological phase. In STM measurement, the Dirac cone with a Chern gap [Fig.14] and topological edge state are detected, implying its non-trivial topological nature [
253].
The coexistence of nontrivial band topology and variation of magnetic structure results in novel transport behavior. In YMn
6Sn
6, a large room temperature anomalous transverse thermoelectric effect of ≈ 2 µV·K
−1 is realized, larger than all canted AFM material studied to date at the room temperature [
261]. In addition, topological Hall effect is observed in the transverse conical spiral phase of YMn
6Sn
6 and ErMn
6Sn
6 with similar magnetic configuration [
262-
264]
. Large anomalous Hall conductivity is also observed in many
RMn
6Sn
6 compounds such as LiMn
6Sn
6, TbMn
6Sn
6, DyMn
6Sn
6, and HoMn
6Sn
6, as shown in Fig.14(d, e) [
252,
254,
258,
262,
264].
In isostructural
RV
6Sn
6 compounds, V atoms have no magnetic moments, so that
RV
6Sn
6 magnetic configuration is different from
RMn
6Sn
6. The magnetic configuration is determined to be out-of-plane AFM for
R = Tb−Ho and in-plane AFM for
R = Er and Tm. because Lu and Y also possess no magnetic moment, so the compounds for
R = Lu and Y are PM metals [
265]. Typical band features such as Dirac cone, saddle point, and flat bands are also observed in this family [
266]. Furthermore, TSS Dirac cones emerge from the nontrivial bulk band topology and can be tuned in binding energy via potassium deposition [
267].
4.3 EuAs3
EuAs
3 crystallizes in a monoclinic structure (space group
, No. 12). As shown in Fig.15, the moments of Eu are oriented along with
axis [
268]. The specific heat, electrical conductivity, susceptibility measurements [
269], neutron diffraction [
270], X-ray scattering technique [
271,
272] and μSR [
273] studies showed that EuAs
3 orders in an incommensurate AFM state at
, and goes through an incommensurate-commensurate lock-in phase transition at
, reaching a collinear AFM ground state. Electrical transport studies showed an extremely anisotropic magnetoresistance related to the magnetic configuration [
274].
Recently, the magnetism-induced topology of EuAs
3 has been demonstrated and the origin of extremely anisotropic magnetoresistance has been discussed [
268]. An unsaturated extremely anisotropic magnetoresistance of 2 × 10
5% at
and
has been observed, as shown in Fig.15. Meanwhile, through the DFT calculations and transport measurements, it is demonstrated that EuAs
3 is a magnetic topological massive Dirac metal at AFM ground state. ARPES results probed by different photon energies verify that EuAs
3 is a topological nodal semimetal in PM state [Fig.15(d, e)], this is related to the extremely anisotropic magnetoresistance. For 3 K ≤
T ≤ 30 K, the concentration of hole carriers is larger than that of electron carriers. Upon decreasing the temperature
the concentration of electron carriers is suddenly enhanced, accompanied by a sharp increase in the mobility of hole carriers, indicating a possible Lifshitz transition [Fig.15].
5 Perspective
In this review, we have gone through several intrinsic magnetic topological states of matter by introducing their representing materials. The interaction between magnetic order and band topology in these materials brings forth characteristic band features such as Dirac gap, Weyl point, Fermi arc, hinge/corner state and so on, produces large Berry curvature and enables novel topological transport responses including quantum anomalous Hall effect, intrinsic anomalous Hall effect, anomalous Nernst effect, negative magnetoresistance as the signature of chiral anomaly and so on. Intrinsic magnetic topological insulators are of fundamental and practical importance because of the potential for the development of dissipationless spintronics, information storage and quantum computation. However, so far only Mn(Bi,Sb)
2Te
4·((Bi,Sb)
2Te
3)
n family is firmly verified as intrinsic magnetic topological insulator. For this family of materials, the lack of sizable magnetic gap hinders the realization of quantum anomalous Hall effect at the expected temperature. It is thus highly desired to search for new material systems hosting such topological state. Instead of incorporating magnetism into established topological systems like the way how Mn(Bi,Sb)
2Te
4·((Bi,Sb)
2Te
3)
n and magnetically doped Bi
2(Se,Te)
3 families are realized, we envision that looking for band topology based on known ferromagnets or antiferromagnets will be more efficient to realize intrinsic magnetic topological insulator. Recent high-throughput calculations and magnetic space group analyses [
275-
280] have predicted a large number of new magnetic topological materials which provide guidance for experiment.
While we are concentrating on the interplay between magnetism and band topology in these quantum states of matter, it is well known that magnets host many ordered phases such as spin density wave, charge density wave, superconductivity, nematicity and so on. The interplay between band topology and these orders could generate exotic states such as axionic charge-density wave [
281], chiral Majorana fermions [
282] and the unknown which deserved future theoretical and experimental investigation. Furthermore, besides ferromagnetism and antiferromagnetism, recently a third basic magnetic phase dubbed altermagnetism [
282-
284] has been developed to describe some supposed antiferromagnets with mysterious behaviors such as anomalous Hall effect [
285-
287], spin polarized bands [
288] and spin splitting torque [
289-
292]. Novel topological states of matter based on altermagnets remains to be explored.