Topical Review

Approaches to tunnel magnetoresistance effect with

antiferromagnets
arXiv:2410.19513v1 [cond-mat.mes-hall] 25 Oct 2024

Katsuhiro Tanaka
Department of Physics, University of Tokyo, Hongo, Bunkyo-ku, Tokyo
113-0033, Japan

Takuya Nomoto
Department of Physics, Tokyo Metropolitan University, Hachioji, Tokyo
192-0397, Japan

Ryotaro Arita
Department of Physics, University of Tokyo, Hongo, Bunkyo-ku, Tokyo
113-0033, Japan
Center for Emergent Matter Science, RIKEN, Wako, Saitama 351-0198, Japan

28 October 2024

Abstract. The tunnel magnetoresistance (TMR) effect is one of the

representative phenomena in spintronics. Ferromagnets, which have a net spin
polarization, have been utilized for the TMR effect. Recently, by contrast, the
TMR effect with antiferromagnets, which do not possess a macroscopic spin
polarization, has been proposed, and also been observed in experiments. In this
topical review, we discuss recent developments in the TMR effect, particularly
focusing on the TMR effect with antiferromagnets. First, we review how the
TMR effect can occur in antiferromagnetic tunnel junctions. The Julliere model,
which has been conventionally utilized to grasp the TMR effect with ferromagnets,
breaks down for the antiferromagnetic TMR effect. Instead, we see that the
momentum dependent spin splitting explains the antiferromagnetic TMR effect.
After that, we revisit the TMR effect from viewpoint of the local density of states
(LDOS). We particularly focus on the LDOS inside the barrier, and show that the
product of the LDOS will qualitatively capture the TMR effect not only in the
ferromagnetic tunnel junctions but also in the ferrimagnetic and antiferromagnetic
tunnel junctions. This method is expected to work usefully for designing magnetic
tunnel junctions.
 2

1. Introduction antiferromagnetic TMR effect. In this review, we

will introduce the role of momentum-dependent spin
The tunnel magnetoresistance (TMR) effect [1] is one polarization of antiferromagnets for the TMR effect;
of the typical phenomena in the field of spintronics [2– we will show that the antiferromagnetic TMR effect
6]. The TMR effect is observed in multilayer systems, can be described by a finite spin splitting in the
magnetic tunnel junctions (MTJs), which have two momentum space which antiferromagnets breaking the
magnetic electrodes sandwiching an insulating thin time-reversal symmetry have.
film. By applying a voltage, the tunneling current The proposal of the TMR effect with antiferro-
can flow through the MTJ as a quantum mechanical magnets will increase the possibility of utilizing a wider
effect. Depending on the relative directions of the variety of magnetic materials and also insulating bar-
magnetic moments between two magnetic electrodes, riers compatible with those magnets. Given this situa-
the tunneling resistance can change. In addition to tion, it is highly desired to develop a method which can
the physics viewpoint, namely, the ballistic transport efficiently evaluate the TMR property of the antiferro-
properties of electrons in relation to the spin degrees of magnetic MTJs as well as the ferromagnetic MTJs for
freedom, the TMR effect has attracted attention from further exploration of the MTJs. In this topical review,
the application point of view. The MTJ is utilized we next focus on the local density of states (LDOS)
for the devices such as the magnetic random access in the MTJs. We will show that the product of the
memories (MRAM) or the magnetic head of the hard- LDOS of the barrier region can trace the TMR effect
disk drive. qualitatively, which can also be understood as a sim-
After the discovery of the TMR effect, the TMR ple extension of the Julliere model incorporating the
effect has been discussed using ferromagnets which has tunneling effect and the details of the materials.
a macroscopic spin polarization [1,7–25]. There, it has The remainder of this review is as follows;
been believed that the magnetic electrode with a large after briefly reviewing the ferromagnetic TMR effect
spin polarization on the Fermi level is important to first, we discuss the recent developments of the
generate a large TMR ratio. TMR effect with antiferromagnetic electrodes [35–
Recently, on the other hand, the TMR effect 54]. We see that the spin splitting in the
has taken renewed interest with antiferromagnets momentum space works for generating a finite TMR
which do not have a net spin polarization; recent effect in the antiferromagnetic electrodes, and refer
developments have proposed that a particular kind to the theoretical and experimental studies of the
of antiferromagnets can be utilized for the spintronic antiferromagnetic TMR effect. In the latter part,
phenomena, namely, they can be controlled by an we discuss a real-space approach to the TMR effect
external field and can be incorporated to the spintronic using the LDOS inside the insulating barrier based
devices even without net magnetization [26–34]. The on Ref. [45]. After confirming that the LDOS
TMR effect using antiferromagnets has been actually works as an indicator for the TMR effect with
proposed, and a finite TMR effect has been observed the lattice models, we present that the LDOS
in the all antiferromagnetic tunnel junctions based on will be useful for designing the MTJs using the
Mn3 Sn or Mn3 Pt [35–37]. ferromagnetic Fe/MgO/Fe and the antiferromagnetic
This topical review aims to understand such Ru1−x Crx O2 /TiO2 /Ru1−x Crx O2 tunnel junctions as
antiferromagnetic TMR effect, and explore a way concrete examples.
toward designing the MTJs. For the ferromagnetic
MTJs, the conventional Julliere model has still
2. Tunnel magnetoresistance effect with
somewhat survived, although a more microscopic
antiferromagnets: role of momentum
mechanism, namely, coherent tunneling, has been
dependent spin splitting
discussed for more closer understanding. However,
in the antiferromagnetic MTJs, the Julliere model 2.1. Tunnel magnetoresistance effect with
completely breaks down since antiferromagnets do not ferromagnetic electrodes
have a macroscopic spin polarization which needs to
generate the TMR effect in the model. Therefore, First, we review the conventional TMR effect with
we need another picture to simply describe the ferromagnets briefly. The simplest model to grasp
 3

(a) (b) Butler et al. gave a microscopic view [12] using the
Parallel Antiparallel bcc-Fe/MgO/Fe MTJ as a specific example. They
focused on the symmetries of the tunneling electrons.
The electronic states around the Fermi level of bcc-Fe
consist of the electrons with the ∆1 -symmetry, whose
E E E E
electron orbitals are rotationally symmetric around z-
axis. In addition, in the MgO barrier, this ∆1 state
has a lower decay rate than the states with other
EF EF EF EF symmetries like ∆2 states or ∆5 states. Therefore,
only the spin-polarized ∆1 state can tunnel through
the MgO barrier, which will lead to a large TMR ratio.
Utilizing this strategy, Yuasa et al. observed a large
Figure 1. Schematic view of tunnel magnetoresistance effect TMR ratio at room temperature in the Fe/MgO/Fe
with ferromagnets for (a) the parallel and (b) antiparallel epitaxial tunnel junction [15]; the epitaxial MTJ made
configurations from the Julliere model. Upper panels represent it possible for electrons to tunnel through the MTJ
the magnetic tunnel junctions. Red and blue rectangles show
keeping the orbital symmetry.
the magnetic electrodes, and gray rectangles show the insulating
barrier. Arrows show the directions of the magnetic moment in
each electrode. Lower panels represent the spin-resolved density 2.2. Tunnel magnetoresistance effect with
of states of the magnetic electrodes.
antiferromagnetic electrodes

the ferromagnetic TMR effect is the so-called Julliere (a) (b)

model [1]. The Jullire model explains the TMR
effect using the macroscopic spin polarization of the ‘‘Parallel’’ ‘‘Antiparallel’’
magnetic electrodes on the Fermi level. Namely, in the
Julliere model, the tunneling conductance of the MTJ,
τDOS (E), is estimated as E E E E
X
τDOS (E) = DL,σ (E)DR,σ (E). (1)
σ
EF EF EF EF
Here, DL/R,σ (E) (σ =↑, ↓) is the bulk density of
states (DOS) with spin-σ of the left/right magnetic
electrodes. For simplicity, let us assume that the
left and right electrodes are the same material, and
D↑ (E)/(D↑ (E) + D↓ (E)) = x. Then, τDOS (E) for the Figure 2. Schematic picture of antiferromagnetic tunnel
parallel/antiparallel configuration, τDOS, P/AP (E), are magnetoresistance effect viewed from the Julliere model, which
gives the same tunneling conductance for the (a) “parallel”
written as and (b) “antiparallel” configurations. Upper panels show
τDOS, P (E) = (x2 + (1 − x)2 )(D↑ (E) + D↓ (E))2 , (2) the magnetic configurations of the magnetic tunnel junctions.
Lower panels represent the spin-resolved density of states of
τDOS, AP (E) = 2x(1 − x)(D↑ (E) + D↓ (E))2 , (3) antiferromagnetic electrodes.
2
and τDOS, P (E)−τDOS, AP (E) ∝ (2x−1) holds. When
we consider the ferromagnetic electrodes, we have the Next, we consider the case where the magnetic
spin polarization at the Fermi energy, namely, x ̸= 0.5, electrodes are antiferromagnetic materials. When we
and thus τDOS, P (E) > τDOS, AP (E) holds. naively see this in terms of the Julliere’s picture,
When the Fermi level is perfectly spin polarized, the total spin polarization, namely, the DOS of
namely, x = 0 or 1, τDOS, P (E)−τDOS, AP (E) takes the the majority and minority spins, are the same, as
largest value. Namely, the materials with half-metallic schematically shown in Fig. 2, and the TMR effect
electronic states will be useful for the realization of a should be zero. We can understand this situation also
large TMR ratio. From this viewpoint, half-Heusler by considering the x = 0.5 case in the discussion in
metals have been explored as a promising candidate Sec. 2.1; τDOS, P (E) − τDOS, AP (E) = 0 holds. Also,
for realizing the half-metallic TMR effect [55, 56]. we should be careful how we can define the parallel
In reality, however, one should consider the details and antiparallel configurations.
of the MTJ such as the modulation of the electronic Recently, however, it has been proposed that one
structure at the interface between the electrodes and can realize the TMR effect even with antiferromagnets,
the barrier, or the decaying properties inside the by examining the structures of the MTJs and the
insulators, to precisely understand the TMR effect, characteristics of antiferromagnets more closely. Here
while the above Julliere model ignores such details. we introduce the antiferromagnetic TMR effect for
 4

two cases; one is the TMR effect owing to the (PGP) (Fig. 3(b)). They calculated the transmission
interfacial structure of the MTJ, and the other exhibits for each configuration (Figs. 3(d) and 3(e)), and
a finite TMR effect originating from the magnetic and showed that the CuMnAs-based MTJ exhibits a finite
electronic properties of antiferromagnets breaking the TMR effect (Figs. 3(f) and 3(g)). Such type of the
time-reversal symmetry. MTJs with multiple sets of the interfacial structure
has also been discussed with an antiferromagnet
Mn2 Au [41, 48, 52] as well as the lattice model
(a) calculation [60].
It should be noted that one needs to precisely con-
trol the interface to realize the situation when we uti-
lize the antiferromagnets shown above. Otherwise, the
(b) TMR properties will be cancelled out, and the TMR ra-
tio will become almost zero in total. Such cancellation
has been found in a ferrimagnet/insulator/ferromagnet
tunnel junction [61]; the observed TMR ratio is small
(c)
owing to the multiple kinds of interfaces between the
ferrimagnet and the insulator, while a large TMR ra-
tio using the ferrimagnet has been predicted by first-
principles calculation [62].

(d)
(a) up down
(e) ky

(f) (g)

kx

(b) Parallel

Figure 3. First-principles calculation of the tunnel mag- ×

netoresistance effect in the CuMnAs/GaP/CuMnAs magnetic
tunnel junction (MTJ). (a), (b) Crystal structures of the
CuMnAs/GaP/CuMnAs MTJ. Termination of the GaP barrier
is (a) Ga-layer and (b) P-layer. (c) Schematics of the parallel and Antiparallel
antiparallel configurations of the CuMnAs/GaP/CuMnAs MTJ.
(d)–(g) Transmission properties in the CuMnAs/GaP/CuMnAs
MTJ. (d), (e) Energy dependence of the transmissions for the ×
MTJ with (d) P-termination and (e) Ga-termination. (f), (g)
Energy dependence of the TMR ratio calculated by (TP −
TAP )/(TP + TAP ) for the MTJ with (f) P-termination and (g)
Ga-termination, where TP and TAP are the transmissions for the Figure 4. Schematics of tunnel magnetoresistance effect
parallel and antiparallel configurations, respectively. Figures are with antiferromagnets breaking time-reversal symmetry. (a)
adopted from Ref. [39] (©American Physical Society (2017)). Schematic view of the spin splitting in the momentum space
at the Fermi level for antiferromagnets breaking time-reversal
symmetry. (b) Schematic view of the antiferromagnetic
2.2.1. Antiferromagnet with controlled interface
tunnel magnetoresistance effect for parallel and antiparallel
First, we introduce the antiferromagnetic TMR configurations.
effect in the MTJ whose interfacial structure is
well controlled. Stamenova et al. studied the 2.2.2. Antiferromagnet breaking time-reversal symme-
MTJ with a collinear antiferromagnet CuMnAs try Next we discuss the TMR effect with antiferro-
with tetragonal crystal structure [39], which is one magnets breaking the time-reversal symmetry [35–37,
of the promising candidate materials utilized for 42–44, 46–51, 53, 54]. Recently, it has been proposed
antiferromagnetic spintronics [57–59]. They considered that the antiferromagnets breaking the time-reversal
the CuMnAs/GaP/CuMnAs tunnel junction. They symmetry can show various phenomena typically ob-
examined multiple sets of configurations depending served in ferromagnets [63–69]. These antiferromag-
on the interfacial structures, namely, Ga-layers of the nets is free from the problem on the interface discussed
GaP-barrier face to the magnetic electrodes (GPG) in the previous section.
(Fig. 3(a)), or P-layers face to the magnetic electrodes We will explain the antiferromagnetic TMR effect
 5

(a) (b) (c)

(d)

(e) (f)

Figure 5. (a)–(d) First-principles calculation of tunnel magnetoresistance (TMR) effect in the RuO2 (001)/TiO2 (001)/RuO2
magnetic tunnel junction. Figures are adapted from Ref. [42] (©Shao et al. (2021), licensed under CC BY 4.0). (a), (b) Magnetic
states of the (a) parallel (P) and (b) antiparallel (AP) configurations and corresponding transmissions resolved by the momentum
and spins. (c) Energy dependence of the total transmissions of parallel and antiparallel configurations. (d) Energy dependence of the
TMR ratio. (e), (f) Results of first-principles calculation for a collinear antiferromagnet Mn5 Si3 . Figures are adapted from Ref. [43]
(©Šmejkal et al. (2022), licensed under CC BY 4.0). (e) Spin and sublattice resolved band structure. (f) Spin and sublattice
resolved density of states (left) and corresponding spin polarization and TMR effect.

using the momentum dependent spin splitting. Let realized in the antiferromagnetic MTJs.
us consider a collinear antiferromagnet breaking the This type of the antiferromagnetic TMR effect
time-reversal symmetry, which has been also called owing to the spin splitting in the k-space has been
altermagnet recently [68, 69]. The band structures proposed by Shao et al. [42]. They have presented that
and the shapes of the Fermi surfaces are different the MTJ with an altermagnetic candidate RuO2 [66,
between the majority and minority spin states; the 70–73] exhibits a finite TMR ratio by performing first-
typical spin-resolved Fermi surfaces projected onto principles calculation. RuO2 has two inequivalent Ru
a two-dimensional plane are schematically shown in sites in a unit cell, and due to the crystal structure,
Fig. 4(a). We note that these spin splittings give the its antiferromagnetic order breaks the time-reversal
same amount of the total spin polarization when it symmetry macroscopically. They have performed first-
is integrated over the whole Brillouin zone, and the principles calculation of the tunneling conductance for
magnetization is compensated in total. When the each of the parallel and antiparallel configurations in
MTJs using such antiferromagnets are constructed, the RuO2 /TiO2 /RuO2 MTJ (Figs. 5(a) and 5(b)).
both electrodes have the same momentum-dependent They have found the difference in the transmission
spin splitting for the parallel configuration. By between the parallel and antiparallel configurations
contrast, for the antiparallel configuration, two (Fig. 5(c)) and corresponding finite TMR ratio
electrodes have the opposite momentum-dependent (Fig. 5(d)) which will be comparable to the one
spin splitting each other (Fig. 4(b)). Therefore, observed in ferromagnetic MTJs. Šmejkal et al. have
the transmission for the parallel and antiparallel also focused on the altermagnet candidates such as
configurations have different momentum dependence, RuO2 or Mn5 Si3 [43]. Based on the momentum
which is still different even when integrated over the dependent spin splitting properties shown in Fig. 5(e),
Brillouin zone. Namely, a finite TMR effect can be they have shown that a finite TMR effect will be
 6

realized when these antiferromagnets are used as the tion has been experimentally observed by Chen et
electrodes of the MTJs (Fig. 5(f)). al. [35]. Chou et al. have shown the TMR ef-
The TMR effect with the antiferromagnets break- fect in the CoFeB/MgO/Mn3 Sn MTJ at low tempera-
ing time-reversal symmetry has been discussed using ture [87]. Theoretical calculations have been performed
noncollinear antiferromagnets as well as the altermag- in RuO2 /TiO2 /CrO2 MTJ [50, 51], where they utilize
nets or the collinear antiferromagnets, and been ob- the half-metallicity of the ferromagnet CrO2 [88, 89].
served in experiments. One of such noncollinear anti-
ferromagnets is Mn3 Sn with a hexagonal crystal struc- 3. A real-space approach to antiferromagnetic
ture. The Mn atoms in Mn3 Sn form the kagome lat- tunnel magnetoresistance effect: local density
tice and the magnetic moments take the inverse tri- of states
angular structure which breaks the time-reversal sym-
metry [74–76]. This antiferromagnetic state can be In the previous section, we discussed the antiferromag-
characterized by the cluster magnetic octupole mo- netic TMR effect from the viewpoint of the spin split-
ment [77, 78] as described by the projected density of ting in the momentum space. In this section, based on
states onto the cluster magnetic octupole states shown Ref. [45], we discuss a real-space approach to capture
in Fig. 6(a). This magnetic structure has led to the ob- the TMR effect, particularly using the LDOS.
servation of various ferromagnetic-like transport phe- We show that the LDOS inside the barrier will
nomena [79–83]. Chen et al. found a finite TMR effect work as an indicator of the TMR effect. In the
in an all-antiferromagnetic tunnel junction based on original Julliere model, the product of the bulk
Mn3 Sn (Figs. 6(b) and 6(c)) [35]. They experimentally density of states is used to estimate the ferromagnetic
observed the difference in the tunnel resistance between TMR effect as discussed in Sec. 2.1. In a similar
when the cluster octupole moments of two magnetic manner, here we present that the product of the
layers align parallelly and antiparallelly (Fig. 6(d)), LDOS of the barrier region can trace the TMR
as well as first-principles calculation showing a finite effect qualitatively. In contrast to the Julliere model
TMR effect with Mn3 Sn. Theoretical approaches to which does not describe the antiferromagnetic TMR
the Mn3 Sn-based MTJ are also performed by Dong et effect, this method with the LDOS will work in the
al. [44], where they show the relative angle dependence ferrimagnetic and antiferromagnetic MTJs as well as
of the cluster magnetic octupole moments between two the ferromagnetic MTJs.
electrodes in addition to the parallel and antiparallel
alignments. 3.1. Local density of states
Another noncollinear antiferromagnet showing the
TMR effect is Mn3 Pt with a cubic crystal structure. Similarly to the Julliere model (Eq. (1)), we can
The Mn ions in the (111)-plane of the Mn3 Pt form consider the product of the LDOS inside the barrier
the kagome lattice and the antiferromagnetic struc- given as
ture breaks the time-reversal symmetry macroscopi- na X
X
cally [84–86]. Qin et al. observed a finite TMR ef- τLDOS = dL,i,σ dR,i,σ . (4)
fect in the Mn3 Pt/MgO/Mn3 Pt MTJ (Fig. 6(e) and i=1 σ

Fig. 6(f)) [36]. They utilize the exchange bias effect Here, L and R is the indices of the layers we focus
between Mn3 Pt and MnPt and present that the TMR on, and dL/R,i,σ is the LDOS at the i-th atom on
effect will reach even around 100% (Fig. 6(g)). A finite the L/R layer. The i-th atom in the L and R layers
TMR effect in the Mn3 Pt-based MTJ has been also share the same in-plane coordinates perpendicular to
observed by Shi et al. in the Mn3 Pt/Al2 O3 /Mn3 Pt the conducting path. The number of atoms in the layer
MTJ [37]. They have performed the switching of the is na .
magnetic moments of Mn3 Pt by applying the electric
current instead of the magnetic field (Fig. 6(h)). This 3.2. Lattice model calculation
result will lead to the switching of the antiferromag-
netic MTJs by the electric current as well as the mag- By a lattice model calculation, we show that τLDOS
netic field, which will be more suitable for further ap- qualitatively captures the transmission property. To
plication in antiferromagnetic spintronics devices. mimic the MTJ, we consider the two-dimensional
The momentum dependent spin splitting in square lattice consisting of three parts; the left
the time-reversal symmetry breaking antiferromag- and right electrodes semi-infinitely extend in the
net also expects us to realize the antiferromag- conducting direction, and the barrier region between
net/insulator/ferromagnet MTJ, in addition to the two electrodes has a length of L in the conducting
all antiferromagnetic tunnel junctions discussed above. direction. Each of the two electrodes and the barrier
The TMR effect in the Fe/MgO/Mn3 Sn tunnel junc- has a width of W in the direction perpendicular to
 7

(a) (b) (c) (d)

(e) (f) (g) (h)

Figure 6. (a)–(d) Tunnel magnetoresistance (TMR) effect in the Mn3 Sn/MgO/Mn3 Sn magnetic tunnel junction (MTJ). Figures
are adapted from Ref. [35] (©Chen et al. (2023), licensed under CC BY 4.0). (a) Projected density of states onto the two
states with opposite cluster magnetic octupole moments. (b) Schematics of the parallel and antiparallel configurations of the
Mn3 Sn/MgO/Mn3 Sn MTJ. (c) Transmission electron microscope (TEM) image of the Mn3 Sn/MgO/Mn3 Sn MTJ. Scale bar is
3 nm. (d) Magnetic field dependence of the tunneling resistance in the Mn3 Sn/MgO/Mn3 Sn MTJ. (e)–(g) TMR effect in the
Mn3 Pt/MgO/Mn3 Pt MTJ. Figures are adapted from Ref. [36] (©Qin et al., under exclusive licence to Springer Nature Limited
2023). (e) Schematics of the Mn3 Pt/MgO/Mn3 Pt MTJ. (f) TEM image of the Mn3 Pt/MgO/Mn3 Pt MTJ. Scale bar is 5 nm. (g)
Magnetic field dependence of the tunneling resistance in the Mn3 Pt/MgO/Mn3 Pt MTJ. (h) Changes of the resistance by applying
an electric current in the Mn3 Pt/Al2 O3 /Mn3 Pt MTJ. Figure is adopted from Ref. [37] (©Shi et al. (2024)).

the conducting path (Fig. 7(a)). We deal with the 3.3. Results
following Hamiltonian on this lattice given as;
X X †  3.3.1. Ferromagnetic electrode We first deal with the
H= εi ni − t ci,σ cj,σ + h.c. ferromagnetic tunnel junction to test the applicability
i ⟨i,j⟩ of the method with LDOS. The left electrode
X αβ † has si = (0, 0, 1) for all sites. In the right
−J (si · σ) ci,α ci,β . (5)
electrode, every site has si = (0, 0, 1) for the
i∈electrode,α,β
parallel configuration, and s = (0, 0, −1) for
Here, c†i,α /ci,α (α =↑, ↓) is the creation/annihilation the antiparallel configuration. The parallel and
operator of an electron with spin-α on site-i, ni = antiparallel configurations of the ferromagnetic MTJ
P † t x y z
α ci,α ci,α , and σ = (σ σ σ ) is the 2 × 2 Pauli
are schematically shown in Fig. 7(a);
matrices. The on-site energy is εi . The electron We present the TMR ratio, rt , on the plane of J
hopping between site-i and j is t; the nearest-neighbor and E in Fig. 7(b). Here the TMR ratio is given as
hopping is considered here, which is expressed by ⟨i, j⟩. TP − TAP
The localized spin moment on the i-th site in the rt = , (6)
TP + TAP
electrode is si , and J is the magnetic coupling constant.
where TP and TAP are the transmission for the parallel
We set εi = 10 for the barrier region and t = 1. We
and antiparallel alignments of the MTJ. When the
take the system size of the barrier region as L = 8 and
magnetic interaction J = 0, the whole system is
W = 160, and impose the open boundary condition on
nonmagnetic, and the TMR ratio becomes zero. By
the perpendicular direction to the conducting path.
introducing a finite magnetic interaction J, the parallel
We use the kwant package for the simula-
and antiparallel configurations are distinguished, and
tion, where the quantum ballistic transport is calcu-
the TMR ratio becomes finite.
lated [90].
In Fig. 7(c), we show the ratio, rc , defined by the
 8

(a) (a) (c) (d) 4 1 (e) (f) 6

(a) (d) 4 1 (e)
3
2
(a) (d) 4 3 4
2 1 (e)
(b)
20
-2 1 3 2

2 2
(b) 0-4 0 0 0
(c) 6 (g) 4 1 (h) (i)
(b) (d) Parallel 2
(b) 4 0 2 Antiparallel 1
2
-22 3

0 0 2
-2 1
0.3
1
-2
-4-2 0 0 1
-4
0
0 2
(c) 6 -6
-0.6 0 (g)-4
0.6 4-4 0 1 2
0 0
0 1 1 (h)20
0
0
0
density of states

(c) 4
6 (g) 4
Figure 7. (a), (b) Schematic figures of the magnetic tunnel junction with ferromagnetic electrodes for the (a) parallel and (b) 1 (h) 2
2
antiparallel configurations. (c) Tunnel magnetoresistance ratio, rt , defined by Eq. (6), with respect to J and E. (d) The ratio
2
4
defined by the product of the local density of states, rc (Eq. (7)). Figures are adopted from Ref. [45] (©American Physical Society
(2023)). 2
2
0
2 0 0.3
LDOS, as We show the TMR ratio rt (Eq. (6)) with respect 1
-2
τLDOS, P − τLDOS,0AP to J and E 0 for the configuration-1 and 2 in Figs. 8(e) 0.3
rc = , (7) -2
and 8(f), respectively. The ratio calculated by the
τLDOS, P + τLDOS, AP
-4 1
where τLDOS, P/AP is τ-2
LDOS, rc (Eq. (7)) for the configurtation-1 and 2
LDOS for the parallel/antiparallel 0
-2 shown in Figs. 8(g) and 8(h). By
are respectively 0
configuration. In this -6model calculations, we calculate
-4 comparing -4rt and rc for each of the two configurations,0 0
τLDOS using the L/2-th -0.6
(fourth) layer and0L/2 + 1-th 0.6confirm that
we 0 τLDOS can trace1 the qualitative 2TMR 0 0
(fifth) layer [91]. By comparing Fig. 7(b) and Fig. 7(c), 0
we find that rt and rc-6
density of states
are qualitatively similar to each
property also for the ferrimagnetic MTJs.
-4 that the interfacial spins roughly0 0
other. Namely, we can -0.6
estimate the TMR 0 property 0.6We remark 0 1 2 0
determine the TMR property. In the configuration-
using the LDOS inside the barrier. density of states 1, spins at the interface of the left and right
electrodes align antiparallelly (parallelly) in the
3.3.2. Ferrimagnetic electrode Next we present parallel (antiparallel) configuration (Figs. 8(a) and
the results of the simulations for the MTJs with 8(b)). This leads to an inverse TMR effect, namely,
ferrimagnetic electrodes. Unlike the ferromagnetic rt < 0, in some parameter regions of the configuration-
MTJs, we have multiple sets of the parallel and 1. On the other hand, in the configuration-2, interfacial
antiparallel configurations for the ferrimagnetic MTJs, spins of the two electrodes across the barrier region
which is owing to the sublattice degrees of freedom of take parallel (antiparallel) in the parallel (antiparallel)
ferrimagnets. Here we consider the ferrimagnet with directions (Figs. 8(c) and 8(d)), and thus rt > 0
two sublattices, A and B, with sA = (0, 0, 1) and sB = basically holds.
(0, 0, −0.5). In one configuration, we call configuration-
1, the sites in the different sublattices sandwich the 3.3.3. Antiferromagnetic electrode Finally, we discuss
barrier as shown in Figs. 8(a) and 8(b). In the other the TMR effect with antiferromagnetic electrodes.
configuration, we call configuration-2, the sites in the We deal with the antiferromagnetic electrodes by
same sublattices are facing to each other across the taking sA = (0, 0, 1) and sB = (0, 0, −1) in the
barrier region (see Figs. 8(c) and 8(d)). The parallel configuration-1 of the above ferrimagnetic electrode.
and antiparallel configurations of the ferrimagnetic We show the TMR ratio rt (Eq. (6)) and the ratio
MTJ are given by the relative directions of the spin defined by the LDOS rc (Eq. (7)) in Figs. 9(a)
moments in the same sublattice; when the spins in and 9(b), respectively. Also, in the case with the
the A-sublattice of the left and right electrodes align antiferromagnetic electrode, we confirm that the TMR
parallelly (antiparallelly), the MTJ takes the parallel ratio is qualitatively captured by the product of the
(antiparallel) configuration. LDOS inside the barrier.
 9
Configuration-1
Configuration-1
(a)
(a) Parallel
(a) Parallel

Configuration-1 Configuration-2 Configuration-1

(e) (f)
(a) 4 (b) 4 1 (c) 3
(b) Antiparallel
(b) Antiparallel
(b) 2 2
2
0 0 0
1
-2 -2

-4 -4 -1 0
Configuration-2
Configuration-2
(g) (h)
(c) (e) 4 (f) 4 1 (g)
(c) Parallel
(c) Parallel 0.08
2
2 2

0 0 0 0
1 0 2

-2 -2 Parallel
(d) Antiparallel Antiparallel
(d) Antiparallel
(d)
-4 -4 -1 0
0 1 2 0 1 2 0 1

Figure 8. Simulation results for the ferrimagnetic tunnel junctions with sA = (0, 0, 1) and sB = (0, 0, −0.5). (a), (b) Colormaps of
the tunnel magnetoresistance ratio, rt (Eq. (6)), as a function of J and E. (e), (f) Ratio defined by the product of the local density
of states (LDOS) inside the barrier, rc (Eq. (7)). Figures are adopted from Ref. [45] (©American Physical Society (2023)).

(a) (a) 4 1 (b) We should note that we have considered the case
where the time-reversal symmetry is preserved here;
8
2 configuration-1 and configuration-2 are equivalent with
each other. Hence the TMR effect will be cancelled
0 0
4
out if we consider the other configuration similarly to
-2
the ferrimagnetic MTJs. However, this approach itself
will be applicable to the TRS-broken antiferromagnets.
-4 -1 0 Namely, the configuration-1 and 2 are distinguished
when we use the antiferromagnets breaking the TRS.
(b) (c) 4 1 (d)
Parallel
In that case, a finite TMR will be realized.
Antiparallel
8
2
1
3.4. Related approaches to the tunnel
0 0
4
magnetoresistance effect with local density of states
0
-2 0 Here2 we have introduced a real-space LDOS approach
to the TMR effect. We have utilized the LDOS inside
-4 -1 0
0 1 2 0 1 the barrier
2 which has the information on the tunneling
decay and the details of the electrodes and the barrier.
We remark that another real-space approach using the
Figure 9. Simulation results for the antiferromagnetic tunnel LDOS was proposed for the TMR effect by Tsymbal et
junction with sA = (0, 0, 1) and sB = (0, 0, −1). (a) Colormap
of the tunnel magnetoresistance ratio rt as a function of J and al. [92, 93]; they evaluated the transmission with the
E. (b) Ratio defined by the product of the local density of states interfacial LDOS and the exponential decay factor.
(LDOS) inside the barrier, rc . Figures are adopted from Ref. [45] In addition, the momentum-resolved LDOS has been
(©American Physical Society (2023)). utilized to analyze the TMR effect [39, 94, 95]. Our
approach and these approaches introduced here will be
complementary to each other.
 10

4. Application of the approach with local the scattering approach [118]. The tunnel conductance
density of states to first-principles calculation: is calculated by the Landauer–Büttiker formula [119–
Fe/MgO/Fe tunnel junction 122], which relates the transmission to the conductance
as G = (e2 /h)T . The transmission is calculated as
In this section, we discuss an application of the method
1 X
discussed above to first-principles calculation. We have T = Tσ (k∥ ), (8)
Nk∥
discussed the role of the LDOS for evaluating the TMR k∥ ,σ
effect using the lattice model. However, to utilize where Tσ (k∥ ) is the partial transmission of the electron
this method for designing the MTJ, particularly the with the spin-σ and the in-plane momentum of k∥ . The
antiferromagnetic MTJs, we should demonstrate that in-plane k-mesh in the transmission calculation, Nk∥ ,
it works also for the materials. Here, we discuss the is 401 × 401.
Fe(001)/MgO(001)/Fe MTJ as an example, which is For the calculation of LDOS, we perform nscf
one of the representative MTJ to show a distinctive calculation of the scattering region after the scf
TMR effect [12–15, 96–108]. calculation. The k-point mesh of nscf calculation is
40 × 40 × 1.
4.1. System and Method
We use bcc-Fe as the magnetic electrode, whose lattice 4.2. Results and discussions
constant we use is 2.8665 Å. The insulating barrier The spin resolved LDOS of each layer of the scattering
consists of MgO, whose in-plane lattice constant is region with the parallel configuration is shown in
taken the same with that of bcc-Fe, and the c-axis Fig. 10(b). We confirm that the LDOS inside MgO is
lattice constant of MgO is set as 4.2113 Å. The small enough in the energy region we focus on, which
interface between Fe and MgO is given by the average shows that the transport discussed here is a tunneling
of the lattice constants of Fe and MgO. one.
We divide the whole MTJ into three regions, The total transmission of the parallel and
namely, the left and right leads and the scattering antiparallel configurations with respect to the chemical
region in between, and perform the electronic structure potential are shown in Fig. 11(a). In the whole energy
calculation for each of the three parts. The left region, we find that TP takes a larger value than TAP .
and right leads are Fe. The scattering region has The products of the LDOS inside the barrier for
seven monolayers (MLs) of MgO sandwiched by six the parallel and antiparallel configurations are shown
and five MLs of Fe as shown in Fig. 10(a). For the in Fig. 12. We take the product of the LDOS at the
calculation of the MTJ with antiparallel configuration, third and fifth layers in Fig. 12(b), and those at the
we deal with the doubled scattering region along the second and sixth layers in Fig. 12(c). For both cases,
conducting path for the smooth connection; we attach we can roughly capture the transmission properties
the supercell which consists of the scattering region shown in Fig. 11 in almost all energy ranges we focus
with the magnetic moments inverted to the original on here.
scattering region. The doubled scattering region is cut
in half, and return to the original scattering region
when calculating the transmission. 5. Application to antiferromagnetic tunnel
To perform the density functional theory calcu- junction: Cr-doped RuO2 electrode
lation [109, 110], we use the Quantum ESPRESSO
In addition to the ferromagnetic tunnel junction
(QE) package [111, 112]. We use the ultrasoft-type
discussed in Sec. 4, in this section, we show that the
pseudopotential obtained from pslibrary [113]. The
estimation of the TMR effect with the LDOS can work
exchange correlation is taken in by the Perdew–Burke–
also in the antiferromagnetic tunnel junctions. Based
Ernzerhof (PBE) generalized gradient approximation
on Ref. [123], here we discuss the antiferromagnetic
(GGA) [114]. The cutoff of the energy for the wave-
TMR effect using the rutile RuO2 electrode with Cr-
function is 100 Ry, and that for the charge-density is
doping. Recently an experiment have shown that
600 Ry. We perform self-consistent field (scf) calcula-
the Cr-doping into the altermagnetic candidate RuO2
tion with the k-point mesh of 20 × 20 × 20 for the lead
will be useful for controlling the magnetism in the
and 20 × 20 × 1 for the scattering region. The effect of
RuO2 systems [124]; the anomalous Hall effect in zero
the spin-orbit coupling (SOC) is neglected.
magnetic field has been observed in the Cr-doped
After the scf calculations of the leads and the
RuO2 , which should be contrasted to the observation
scattering region, we attach the three parts to
of the anomalous Hall component in RuO2 in a
construct the MTJ, and calculate the transmission. We
magnetic field [125]. We discuss the TMR effect in
use the pwcond routine contained in the QE package
the Ru1−x Crx O2 (001)/TiO2 (001)/Ru1−x Crx O2 MTJ.
to calculate the transmission [115–117], which takes
 11

(a) Fe MgO Fe

c
(b)

Figure 10. (a) Crystal structure of the scattering region for the Fe/MgO/Fe tunnel junction with seven monolayers of MgO barrier.
(b) Local density of states at each layer of the scattering region for the parallel configuration.

5.1. System and method

We use the Cr-doped RuO2 as the magnetic electrode
and nine MLs of the rutile-type TiO2 as the barrier.
The crystal structure of the center part of the MTJ
is shown in Fig. 13(a). We take a similar manner to
the case of the Fe/MgO/Fe MTJ discussed above using
the QE package [111, 112]. Namely, we separate the
MTJ into the left and right leads and the scattering
region, perform the scf calculation for each, and then
calculate the transmission connecting the three parts.
To calculate τLDOS (Eq. (4)), the LDOS at the fourth
Figure 11. Total transmissions for the parallel and antiparallel and sixth layers of TiO2 are used. We note that the
configurations of the Fe/MgO/Fe tunnel junction with respect parallel and antiparallel configurations of the MTJ are
to the energy. defined by the relative orientations of the magnetic
moments in the same sublattices between the left
(a) (b) and right magnetic electrodes as shown in Fig. 13(a),
similarly to the discussion in the model calculation of
the ferrimagnetic MTJ (Sec. 3). For the details of the
calculation, please see Ref. [123].

5.2. Results
We show the Cr concentration dependence of the
total transmission in the MTJ with parallel and
antiparallel configurations, TP and TAP , respectively,
in Fig. 13(b). We confirm that TP takes a larger
transmission than the antiparallel state in the region
Figure 12. Products of the local density of states for the parallel of 0.3 ≤ x ≤ 0.5, which gives a positive finite
and antiparallel configurations as a function of the chemical TMR ratio. The product of the LDOS with respect
potential E (measured from the Fermi energy EF ). (a) The case
with the third and fifth layers. (b) The case with the second and to the amount of Cr for the parallel/antiparallel
sixth layers. configuration, τLDOS, P/AP , is shown in Fig. 13(c).
We find that τLDOS, P takes a larger value than
τLDOS, AP in 0.3 ≤ x ≤ 0.5, which indicates that
the product of the LDOS inside the barrier region
 12

(a) 6. Summary and perspective

In summary, we have reviewed the tunnel magnetore-

sistance (TMR) effect with ferrimagnetic and antiferro-
magnetic electrodes as well as conventional ferromag-
netic electrodes. The antiferromagnets breaking the
time-reversal symmetry macroscopically can generate
a finite spin splitting in the momentum space with the
net spin polarization vanished, and the momentum de-
pendent spin splitting contributes to the realization of
a finite TMR effect. We have introduced examples
of the antiferromagnetic tunnel junctions showing the
TMR effect in experiments, in addition to the numeri-
cal studies.
In the latter part, we have introduced a method
to estimate the TMR effect using the local density
(b) (c) of states inside the barrier, and its demonstration
with the density functional theory calculation. This
approach may contribute to further exploration of the
magnetic tunnel junctions (MTJs), particularly the
antiferromagnetic tunnel junctions which cannot be
understood by the original Julliere model.
Recently, many antiferromagnets with broken
time-reversal symmetry have been found [126]. Those
antiferromagnets are expected to show ferromagnetic-
like behaviors including the TMR effect; we may
have a much room for developing antiferromagnetic
MTJs which is comparable or even superior to the
Figure 13. Tunnel magnetoresistance effect in the
Ru1−x Crx O2 (001)/TiO2 (001)/Ru1−x Crx O2 magnetic tunnel ferromagnetic MTJs. Furthermore, recent advances
junction (MTJ). (a) Crystal structure of the scattering region in antiferromagnetic spintronics have shown that the
used in the center of the MTJ, and the schematics of the parallel antiferromagnetic order can be manipulated by various
and antiparallel configurations of the MTJ. Arrows and arrows
described by broken lines are the magnetic moments of the two
external stimuli such as the electric current [127–135].
inequivalent Ru/Cr sites, Ru/Cr-A and Ru/Cr-B. (b), (c) Cr- By combining the antiferromagnetic TMR effect with
concentration dependence of (b) the total transmission in the these capability of manipulating the antiferromagnetic
MTJ and (c) the product of the local density of states calcu- order, we may be able to develop all antiferromagnetic
lated by Eq. (4), for the parallel and antiparallel configurations.
Figures are adopted from Ref. [123] (©American Physical Soci-
spintronic devices, which can possess the merit such as
ety (2024)). a high-speed operation or a smaller stray field.

Acknowledgments
τLDOS can capture the qualitative TMR property in
the Ru1−x Crx O2 (001)/TiO2 (001)/Ru1−x Crx O2 MTJ. We thank Xianzhe Chen, Tomoya Higo, Shinji Miwa,
We have shown that the TMR property is roughly and Satoru Nakatsuji for fruitful discussions and col-
captured by the product of the LDOS inside the barrier laborative work. KT is grateful to Takashi Koret-
by first-principles calculation of the Fe/MgO/Fe sune and Yuta Toga for discussions. This work was
ferromagnetic MTJ (Sec. 4) and the antiferromagnetic supported by JST-Mirai Program (Grant No. JP-
Ru1−x Crx O2 (001)/TiO2 (001)/Ru1−x Crx O2 MTJ, in MJMI20A1), JST-CREST (Grant No. JPMJCR23O4),
addition to the lattice model calculation discussed in JST-ASPIRE (Grant No. JPMJAP2317), JSPS-
Sec. 3. These results expect us to apply this method KAKENHI (Grant No. 21H04437, No. JP21H04990,
using the LDOS to computational search for the MTJs. No. JP22H00290, No. JP24K00581), and the RIKEN
We may be able to avoid first-principles calculations of TRIP initiative (RIKEN Quantum, AGIS, Many-body
the transmission itself which often need huge numerical Electron Systems). We use the VESTA [136] soft-
costs, and instead we can grasp the TMR property only ware to visualize the crystal structure, whose input
by calculating the LDOS of the MTJs. file is generated with the help of the XCrySDen soft-
ware [137].
 13

References Yan H, Qin P and Liu Z 2024 Adv. Mater. 36 2310379

[35] Chen X, Higo T, Tanaka K, Nomoto T, Tsai H, Idzuchi
[1] Julliere M 1975 Phys. Lett. A 54 225–226 ISSN 0375-9601 H, Shiga M, Sakamoto S, Ando R, Kosaki H, Matsuo
[2] Žutić I, Fabian J and Das Sarma S 2004 Rev. Mod. Phys. T, Nishio-Hamane D, Arita R, Miwa S and Nakatsuji S
76(2) 323–410 2023 Nature 613 490–495
[3] Felser C, Fecher G and Balke B 2007 Angew. Chem. Int. [36] Qin P, Yan H, Wang X, Chen H, Meng Z, Dong J, Zhu M,
Ed. 46 668–699 Cai J, Feng Z, Zhou X, Liu L, Zhang T, Zeng Z, Zhang
[4] Chappert C, Fert A and van Dau F N 2007 Nat. Mater. 6 J, Jiang C and Liu Z 2023 Nature 613 485–489
813–823 [37] Shi J, Arpaci S, Lopez-Dominguez V, Sangwan V K,
[5] Bader S and Parkin S 2010 Annu. Rev. Condens. Matter Mahfouzi F, Kim J, Athas J G, Hamdi M, Aygen C,
Phys. 1 71–88 Arava H, Phatak C, Carpentieri M, Jiang J S, Grayson
[6] Hirohata A, Yamada K, Nakatani Y, Prejbeanu I L, Diény M A, Kioussis N, Finocchio G, Hersam M C and
B, Pirro P and Hillebrands B 2020 J. Magn. Magn. Khalili Amiri P 2024 Adv. Mater. 36 2312008
Mater. 509 166711 ISSN 0304-8853 [38] Merodio P, Kalitsov A, Béa H, Baltz V and Chshiev M
[7] Maekawa S and Gafvert U 1982 IEEE Tran. Magn. 18 2014 Appl. Phys. Lett. 105 122403
707–708 [39] Stamenova M, Mohebbi R, Seyed-Yazdi J, Rungger I and
[8] Slonczewski J C 1989 Phys. Rev. B 39(10) 6995–7002 Sanvito S 2017 Phys. Rev. B 95(6) 060403
[9] Miyazaki T and Tezuka N 1995 J. Magn. Magn. Mater. [40] Železný J, Zhang Y, Felser C and Yan B 2017 Phys. Rev.
139 L231–L234 ISSN 0304-8853 Lett. 119(18) 187204
[10] Moodera J S, Kinder L R, Wong T M and Meservey R [41] Jia X T, Cai X L and Jia Y 2020 Sci. China Phys., Mech.
1995 Phys. Rev. Lett. 74(16) 3273–3276 Astron. 63 1–7
[11] Mathon J 1997 Phys. Rev. B 56(18) 11810–11819 [42] Shao D F, Zhang S H, Li M, Eom C B and Tsymbal E Y
[12] Butler W H, Zhang X G, Schulthess T C and MacLaren 2021 Nat. Commun. 12 1–8
J M 2001 Phys. Rev. B 63(5) 054416 [43] Šmejkal L, Hellenes A B, González-Hernández R, Sinova J
[13] Mathon J and Umerski A 2001 Phys. Rev. B 63(22) 220403 and Jungwirth T 2022 Phys. Rev. X 12(1) 011028
[14] Parkin S S, Kaiser C, Panchula A, Rice P M, Hughes B, [44] Dong J, Li X, Gurung G, Zhu M, Zhang P, Zheng F,
Samant M and Yang S H 2004 Nat. Mater. 3 862–867 Tsymbal E Y and Zhang J 2022 Phys. Rev. Lett.
[15] Yuasa S, Nagahama T, Fukushima A, Suzuki Y and Ando 128(19) 197201
K 2004 Nat. Mater. 3 868–871 [45] Tanaka K, Nomoto T and Arita R 2023 Phys. Rev. B
[16] Djayaprawira D D, Tsunekawa K, Nagai M, Maehara H, 107(21) 214442
Yamagata S, Watanabe N, Yuasa S, Suzuki Y and Ando [46] Cui Q, Zhu Y, Yao X, Cui P and Yang H 2023 Phys. Rev.
K 2005 Appl. Phys. Lett. 86 092502 B 108(2) 024410
[17] Ikeda S, Hayakawa J, Ashizawa Y, Lee Y M, Miura K, [47] Gurung G, Shao D F and Tsymbal E Y 2023 Extraor-
Hasegawa H, Tsunoda M, Matsukura F and Ohno H dinary Tunneling Magnetoresistance in Antiferromag-
2008 Appl. Phys. Lett. 93 082508 netic Tunnel Junctions with Antiperovskite Electrodes
[18] Ikeda S, Miura K, Yamamoto H, Mizunuma K, Gan H, (Preprint arXiv:2306.03026)
Endo M, Kanai S, Hayakawa J, Matsukura F and Ohno [48] Jia X, Tang H M and Wang S Z 2023 Phys. Rev. B 108(10)
H 2010 Nat. Mater. 9 721–724 104406
[19] Worledge D C, Hu G, Abraham D W, Sun J Z, Trouilloud [49] Jiang Y Y, Wang Z A, Samanta K, Zhang S H, Xiao R C,
P L, Nowak J, Brown S, Gaidis M C, O’Sullivan E J Lu W J, Sun Y P, Tsymbal E Y and Shao D F 2023
and Robertazzi R P 2011 Appl. Phys. Lett. 98 022501 Phys. Rev. B 108(17) 174439
ISSN 0003-6951 [50] Chi B, Jiang L, Zhu Y, Yu G, Wan C, Zhang J and Han
[20] Scheike T, Wen Z, Sukegawa H and Mitani S 2023 Appl. X 2024 Phys. Rev. Appl. 21(3) 034038
Phys. Lett. 122 112404 ISSN 0003-6951 [51] Samanta K, Jiang Y Y, Paudel T R, Shao D F and
[21] Tsymbal E Y, Mryasov O N and LeClair P R 2003 J. Phys.: Tsymbal E Y 2024 Phys. Rev. B 109(17) 174407
Condens. Matter 15 R109–R142 [52] Zhu C, Jia X and Tang H M 2024 J. Magn. Magn. Mater.
[22] Zhang X G and Butler W H 2003 J. Phys.: Condens. 597 172036 ISSN 0304-8853
Matter 15 R1603–R1639 [53] Zhu M, Dong J, Li X, Zheng F, Zhou Y, Wu K and Zhang
[23] Itoh H and Inoue J 2006 J. Magn. Soc. Jpn. 30 1–37 J 2024 Chin. Phys. Lett. 41 047502
[24] Yuasa S and Djayaprawira D D 2007 J. Phys. D: Appl. [54] Shao D F and Tsymbal E Y 2024 npj Spintronics 2 13
Phys. 40 R337–R354 [55] Sakuraba Y, Nakata J, Oogane M, Kubota H, Ando Y,
[25] Butler W H 2008 Sci. Technol. Adv. Mater. 9 014106 Sakuma A and Miyazaki T 2005 Jpn. J. Appl. Phys. 44
[26] MacDonald A H and Tsoi M 2011 Phil. Trans. R. Soc. A L1100
369 3098–3114 [56] Sakuraba Y, Hattori M, Oogane M, Ando Y, Kato H,
[27] Gomonay E V and Loktev V M 2014 Low Temp. Phys. 40 Sakuma A, Miyazaki T and Kubota H 2006 Appl. Phys.
17–35 Lett. 88 192508
[28] Jungwirth T, Marti X, Wadley P and Wunderlich J 2016 [57] Wadley P, Howells B, Železný J, Andrews C, Hills V,
Nat. Nanotechnol. 11 231–241 Campion R P, Novák V, Olejnı́k K, Maccherozzi F,
[29] Baltz V, Manchon A, Tsoi M, Moriyama T, Ono T and Dhesi S S, Martin S Y, Wagner T, Wunderlich J,
Tserkovnyak Y 2018 Rev. Mod. Phys. 90(1) 015005 Freimuth F, Mokrousov Y, Kuneš J, Chauhan J S,
[30] Železnỳ J, Wadley P, Olejnı́k K, Hoffmann A and Ohno H Grzybowski M J, Rushforth A W, Edmonds K W,
2018 Nat. Phys. 14 220–228 Gallagher B L and Jungwirth T 2016 Science 351 587–
[31] Siddiqui S A, Sklenar J, Kang K, Gilbert M J, Schleife 590
A, Mason N and Hoffmann A 2020 J. Appl. Phys. 128 [58] Olejnı́k K, Schuler V, Martı́ X, Novák V, Kašpar Z, Wadley
040904 P, Campion R P, Edmonds K W, Gallagher B L, Garcés
[32] Amin O J, Edmonds K W and Wadley P 2020 Appl. Phys. J, Baumgartner M, Gambardella P and Jungwirth T
Lett. 117 010501 ISSN 0003-6951 2017 Nat. Commun. 8 15434
[33] Fukami S, Lorenz V O and Gomonay O 2020 J. Appl. [59] Godinho J, Reichlová H, Kriegner D, Novák V, Olejnı́k
Phys. 128 070401 K, Kašpar Z, Šobáň Z, Wadley P, Campion R, Otxoa
[34] Chen H, Liu L, Zhou X, Meng Z, Wang X, Duan Z, Zhao G, R, Roy P E, Železný J, Jungwirth T and Wunderlich J
 14

2018 Nat. Commun. 9 4686 [88] Schwarz K 1986 J. Phys. F: Met. Phys. 16 L211
[60] Saidaoui H B M, Waintal X and Manchon A 2017 Phys. [89] Korotin M A, Anisimov V I, Khomskii D I and Sawatzky
Rev. B 95(13) 134424 G A 1998 Phys. Rev. Lett. 80(19) 4305–4308
[61] Jeong J, Ferrante Y, Faleev S V, Samant M G, Felser C [90] Groth C W, Wimmer M, Akhmerov A R and Waintal X
and Parkin S S 2016 Nat. Commun. 7 10276 2014 New J. Phys. 16 063065
[62] Miura Y and Shirai M 2014 IEEE Trans. Magn. 50 1–4 [91] In the lattice model calculation [45], the open boundary
[63] Noda Y, Ohno K and Nakamura S 2016 Phys. Chem. condition is used in the direction perpendicular to
Chem. Phys. 18(19) 13294–13303 the conducting direction, y-direction. To exclude the
[64] Okugawa T, Ohno K, Noda Y and Nakamura S 2018 J. boundary effect, the product of the LDOS is taken only
Phys.: Condens. Matter 30 075502 around the center in the y direction, y = W/2, instead
[65] Naka M, Hayami S, Kusunose H, Yanagi Y, Motome Y of summing up the product over all y coordinates.
and Seo H 2019 Nat. Commun. 10 4305 [92] Tsymbal E Y and Belashchenko K D 2005 J. Appl. Phys.
[66] Ahn K H, Hariki A, Lee K W and Kuneš J 2019 Phys. 97 10C910
Rev. B 99(18) 184432 [93] Tsymbal E Y, Belashchenko K D, Velev J P, Jaswal S S,
[67] Šmejkal L, González-Hernández R, Jungwirth T and van Schilfgaarde M, Oleynik I I and Stewart D A 2007
Sinova J 2020 Sci. Adv. 6 eaaz8809 Prog. Mater. Sci. 52 401–420 ISSN 0079-6425
[68] Šmejkal L, Sinova J and Jungwirth T 2022 Phys. Rev. X [94] Belashchenko K D, Tsymbal E Y, van Schilfgaarde M,
12(3) 031042 Stewart D A, Oleynik I I and Jaswal S S 2004 Phys.
[69] Šmejkal L, Sinova J and Jungwirth T 2022 Phys. Rev. X Rev. B 69(17) 174408
12(4) 040501 [95] Flores G G B, van Schilfgaarde M and Belashchenko
[70] Berlijn T, Snijders P C, Delaire O, Zhou H D, Maier T A, K D 2024 Tunneling magnetoresistance in MgO tunnel
Cao H B, Chi S X, Matsuda M, Wang Y, Koehler M R, junctions with Fe-based leads in empirically corrected
Kent P R C and Weitering H H 2017 Phys. Rev. Lett. density functional theory (Preprint arXiv:2404.08103)
118(7) 077201 [96] Yuasa S, Fukushima A, Nagahama T, Ando K and Suzuki
[71] Zhu Z H, Strempfer J, Rao R R, Occhialini C A, Pelliciari Y 2004 Jpn. J. Appl. Phys. 43 L588
J, Choi Y, Kawaguchi T, You H, Mitchell J F, Shao- [97] Tusche C, Meyerheim H L, Jedrecy N, Renaud G, Ernst
Horn Y and Comin R 2019 Phys. Rev. Lett. 122(1) A, Henk J, Bruno P and Kirschner J 2005 Phys. Rev.
017202 Lett. 95(17) 176101
[72] Fedchenko O, Minár J, Akashdeep A, D’Souza S W, [98] Waldron D, Timoshevskii V, Hu Y, Xia K and Guo H 2006
Vasilyev D, Tkach O, Odenbreit L, Nguyen Q, Phys. Rev. Lett. 97(22) 226802
Kutnyakhov D, Wind N, Wenthaus L, Scholz M, [99] Matsumoto R, Fukushima A, Nagahama T, Suzuki Y,
Rossnagel K, Hoesch M, Aeschlimann M, Stadtmüller Ando K and Yuasa S 2007 Appl. Phys. Lett. 90 252506
B, Kläui M, Schönhense G, Jungwirth T, Hellenes A B, ISSN 0003-6951
Jakob G, Šmejkal L, Sinova J and Elmers H J 2024 Sci. [100] Heiliger C, Zahn P, Yavorsky B Y and Mertig I 2008 Phys.
Adv. 10 eadj4883 Rev. B 77(22) 224407
[73] We note that the magnetism of RuO2 is controversial; [101] Heiliger C, Zahn P and Mertig I 2007 J. Magn. Magn.
some recent studies have argued that RuO2 is unlikely Mater. 316 478–480 ISSN 0304-8853 proceedings of the
to have antiferromagnetic order stably [138–140]. Joint European Magnetic Symposia
[74] Tomiyoshi S 1982 J. Phys. Soc. Jpn. 51 803–810 [102] Rungger I, Reily Rocha A, Mryasov O, Heinonen O and
[75] Nagamiya T, Tomiyoshi S and Yamaguchi Y 1982 Solid Sanvito S 2007 J. Magn. Magn. Mater. 316 481–483
State Commun. 42 385–388 ISSN 0038-1098 ISSN 0304-8853
[76] Brown P J, Nunez V, Tasset F, Forsyth J B and [103] Wang S G, Ward R C C, Du G X, Han X F, Wang C and
Radhakrishna P 1990 J. Phys.: Condens. Matter 2 9409 Kohn A 2008 Phys. Rev. B 78(18) 180411
[77] Suzuki M T, Koretsune T, Ochi M and Arita R 2017 Phys. [104] Bose P, Zahn P, Henk J and Mertig I 2010 Phys. Rev. B
Rev. B 95(9) 094406 82(1) 014412
[78] Suzuki M T, Nomoto T, Arita R, Yanagi Y, Hayami S and [105] Raza T Z, Cerdá J I and Raza H 2011 J. Appl. Phys. 109
Kusunose H 2019 Phys. Rev. B 99(17) 174407 023705 ISSN 0021-8979
[79] Nakatsuji S, Kiyohara N and Higo T 2015 Nature 527 [106] Autès G, Mathon J and Umerski A 2011 Phys. Rev. B
212–215 84(13) 134404
[80] Ikhlas M, Tomita T, Koretsune T, Suzuki M T, Nishio- [107] Masuda K, Tadano T and Miura Y 2021 Phys. Rev. B
Hamane D, Arita R, Otani Y and Nakatsuji S 2017 Nat. 104(18) L180403
Phys. 13 1085–1090 [108] Scheike T, Xiang Q, Wen Z, Sukegawa H, Ohkubo T, Hono
[81] Higo T, Man H, Gopman D B, Wu L, Koretsune T, van’t K and Mitani S 2021 Appl. Phys. Lett. 118 042411 ISSN
Erve O M, Kabanov Y P, Rees D, Li Y, Suzuki M T, 0003-6951
Patankar S, Ikhlas M, Chien C L, Arita R, Shull R D, [109] Hohenberg P and Kohn W 1964 Phys. Rev. 136(3B)
Orenstein J and Nakatsuji S 2018 Nat. Photon. 12 73– B864–B871
78 [110] Kohn W and Sham L J 1965 Phys. Rev. 140(4A) A1133–
[82] Higo T and Nakatsuji S 2022 J. Magn. Magn. Mater. 564 A1138
170176 ISSN 0304-8853 [111] Giannozzi P, Baroni S, Bonini N, Calandra M, Car R,
[83] Nakatsuji S and Arita R 2022 Annu. Rev. Condens. Matter Cavazzoni C, Ceresoli D, Chiarotti G L, Cococcioni M,
Phys. 13 119–142 ISSN 1947-5462 Dabo I, Corso A D, de Gironcoli S, Fabris S, Fratesi
[84] Krén E, Kádár G, Pál L, Sólyom J and Szabó P 1966 Phys. G, Gebauer R, Gerstmann U, Gougoussis C, Kokalj
Lett. 20 331–332 ISSN 0031-9163 A, Lazzeri M, Martin-Samos L, Marzari N, Mauri F,
[85] Krén E, Kádár G, Pál L and Szabó P 1967 J. Appl. Phys. Mazzarello R, Paolini S, Pasquarello A, Paulatto L,
38 1265–1266 ISSN 0021-8979 Sbraccia C, Scandolo S, Sclauzero G, Seitsonen A P,
[86] Krén E, Kádár G, Pál L, Sólyom J, Szabó P and Tarnóczi Smogunov A, Umari P and Wentzcovitch R M 2009 J.
T 1968 Phys. Rev. 171(2) 574–585 Phys.: Condens. Matter 21 395502
[87] Chou C T, Ghosh S, McGoldrick B C, Nguyen T, Gurung [112] Giannozzi P, Andreussi O, Brumme T, Bunau O, Nardelli
G, Tsymbal E Y, Li M, Mkhoyan K A and Liu L 2024 M B, Calandra M, Car R, Cavazzoni C, Ceresoli D,
Nat.Commun. 15 7840 Cococcioni M, Colonna N, Carnimeo I, Corso A D,
 15

de Gironcoli S, Delugas P, DiStasio R A, Ferretti A, R and Moser S 2024 Absence of magnetic order in
Floris A, Fratesi G, Fugallo G, Gebauer R, Gerstmann RuO2 : insights from µSR spectroscopy and neutron
U, Giustino F, Gorni T, Jia J, Kawamura M, Ko diffraction (Preprint arXiv:2405.10820)
H Y, Kokalj A, Kü”çükbenli E, Lazzeri M, Marsili M,
Marzari N, Mauri F, Nguyen N L, Nguyen H V, de-
la Roza A O, Paulatto L, Poncé S, Rocca D, Sabatini
R, Santra B, Schlipf M, Seitsonen A P, Smogunov A,
Timrov I, Thonhauser T, Umari P, Vast N, Wu X and
Baroni S 2017 J. Phys.: Condens. Matter 29 465901
[113] Dal Corso A 2014 Comput. Mater. Sci. 95 337–350 ISSN
0927-0256 https://dalcorso.github.io/pslibrary/
[114] Perdew J P, Burke K and Ernzerhof M 1996 Phys. Rev.
Lett. 77(18) 3865–3868
[115] Smogunov A, Dal Corso A and Tosatti E 2004 Phys. Rev.
B 70(4) 045417
[116] Corso A D and Conte A M 2005 Phys. Rev. B 71(11)
115106
[117] Dal Corso A, Smogunov A and Tosatti E 2006 Phys. Rev.
B 74(4) 045429
[118] Joon Choi H and Ihm J 1999 Phys. Rev. B 59(3) 2267–
2275
[119] Landauer R 1957 IBM J. Res. Dev. 1 223–231
[120] Landauer R 1970 Phil Mag. 21 863–867
[121] Büttiker M 1986 Phys. Rev. Lett. 57(14) 1761–1764
[122] Büttiker M 1988 IBM J. Res. Develop. 32 317–334
[123] Tanaka K, Nomoto T and Arita R 2024 Phys. Rev. B
110(6) 064433
[124] Wang M, Tanaka K, Sakai S, Wang Z, Deng K, Lyu Y, Li
C, Tian D, Shen S, Ogawa N, Kanazawa N, Yu P, Arita
R and Kagawa F 2023 Nat. Commun. 14 8240
[125] Feng Z, Zhou X, Šmejkal L, Wu L, Zhu Z, Guo H,
González-Hernández R, Wang X, Yan H, Qin P, Zhang
X, Wu H, Chen H, Meng Z, Liu L, Xia Z, Sinova J,
Jungwirth T and Liu Z 2022 Nat. Electro. 5 735–743
[126] Nomoto T, Minami S, Yanagi Y, Suzuki M T, Koretsune
T and Arita R 2024 Phys. Rev. B 109(9) 094435
[127] Železný J, Gao H, Výborný K, Zemen J, Mašek J,
Manchon A, Wunderlich J, Sinova J and Jungwirth T
2014 Phys. Rev. Lett. 113(15) 157201
[128] Liu Z Q, Chen H, Wang J M, Liu J H, Wang K, Feng
Z X, Yan H, Wang X R, Jiang C B, Coey J M D and
MacDonald A H 2018 Nat. Electro. 1 172–177
[129] Chen X Z, Zarzuela R, Zhang J, Song C, Zhou X F,
Shi G Y, Li F, Zhou H A, Jiang W J, Pan F and
Tserkovnyak Y 2018 Phys. Rev. Lett. 120(20) 207204
[130] Meinert M, Graulich D and Matalla-Wagner T 2018 Phys.
Rev. Appl. 9(6) 064040
[131] Hajiri T, Ishino S, Matsuura K and Asano H 2019 Appl.
Phys. Lett. 115 052403 ISSN 0003-6951
[132] Tsai H, Higo T, Kondou K, Nomoto T, Sakai A, Kobayashi
A, Nakano T, Yakushiji K, Arita R, Miwa S, Otani Y
and Nakatsuji S 2020 Nature 580 608–613
[133] Tsai H, Higo T, Kondou K, Sakamoto S, Kobayashi A,
Matsuo T, Miwa S, Otani Y and Nakatsuji S 2021 Small
Sci. 1 2000025
[134] Higo T, Kondou K, Nomoto T, Shiga M, Sakamoto S, Chen
X, Nishio-Hamane D, Arita R, Otani Y, Miwa S and
Nakatsuji S 2022 Nature 607 474–479
[135] Deng Y, Liu X, Chen Y, Du Z, Jiang N, Shen C, Zhang
E, Zheng H, Lu H Z and Wang K 2022 Natl. Sci. Rev.
10 nwac154 ISSN 2095-5138
[136] Momma K and Izumi F 2011 J. Appl. Cryst. 44 1272–1276
[137] Kokalj A 1999 J. Mol. Graphics Modelling 17 176–179
ISSN 1093-3263
[138] Smolyanyuk A, Mazin I I, Garcia-Gassull L and Valentı́ R
2024 Phys. Rev. B 109(13) 134424
[139] Hiraishi M, Okabe H, Koda A, Kadono R, Muroi T, Hirai
D and Hiroi Z 2024 Phys. Rev. Lett. 132(16) 166702
[140] Keßler P, Garcia-Gassull L, Suter A, Prokscha T, Salman
Z, Khalyavin D, Manuel P, Orlandi F, Mazin I I, Valentı́