ARTICLE

https://doi.org/10.1038/s41467-021-26915-3 OPEN

Spin-neutral currents for spintronics

Ding-Fu Shao 1 ✉, Shu-Hui Zhang 2, Ming Li1, Chang-Beom Eom3 & Evgeny Y. Tsymbal 1✉

Electric currents carrying a net spin polarization are widely used in spintronics, whereas
globally spin-neutral currents are expected to play no role in spin-dependent phenomena.
Here we show that, in contrast to this common expectation, spin-independent conductance in
1234567890():,;

compensated antiferromagnets and normal metals can be efficiently exploited in spintronics,

provided their magnetic space group symmetry supports a non-spin-degenerate Fermi
surface. Due to their momentum-dependent spin polarization, such antiferromagnets can be
used as active elements in antiferromagnetic tunnel junctions (AFMTJs) and produce a giant
tunneling magnetoresistance (TMR) effect. Using RuO2 as a representative compensated
antiferromagnet exhibiting spin-independent conductance along the [001] direction but a
non-spin-degenerate Fermi surface, we design a RuO2/TiO2/RuO2 (001) AFMTJ, where a
globally spin-neutral charge current is controlled by the relative orientation of the Néel
vectors of the two RuO2 electrodes, resulting in the TMR effect as large as ~500%. These
results are expanded to normal metals which can be used as a counter electrode in AFMTJs
with a single antiferromagnetic layer or other elements in spintronic devices. Our work
uncovers an unexplored potential of the materials with no global spin polarization for utilizing
them in spintronics.

1 Department of Physics and Astronomy & Nebraska Center for Materials and Nanoscience, University of Nebraska, Lincoln, NE 68588-0299, USA. 2 College

of Mathematics and Physics, Beijing University of Chemical Technology, Beijing 100029, People’s Republic of China. 3 Department of Materials Science and
Engineering, University of Wisconsin-Madison, Madison, WI 3706, US. ✉email: [email protected]; [email protected]

NATURE COMMUNICATIONS | (2021)12:7061 | https://doi.org/10.1038/s41467-021-26915-3 | www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-26915-3

T
he field of spintronics utilizes the spin degree of freedom in coupling where the combined space inversion-time reversal
condensed matter for information processing and storage1. symmetry is broken, indicating that they can also be utilized in
Most spintronic applications rely on electric currents with spintronics despite globally spin-neutral currents.
sizable spin polarization for detection or manipulation of the
magnetic order parameter in spintronic devices. A typical and Results
widely used spintronic device is the magnetic tunnel junction Spin polarized conduction channels. To explore the possible use
(MTJ), where a longitudinal charge current spin polarized by one of spin-neutral currents in a spintronic device, we first consider
ferromagnetic metal quantum-mechanically tunnels into another ballistic conductance of a material under investigation. Since the
ferromagnetic metal through an insulating barrier layer2,3. Con- ballistic conductance is determined by the number of conduction
ductance of the MTJ is controlled by the relative magnetization channels, i.e. propagating Bloch states at the Fermi energy, it can
orientation of the two ferromagnetic electrodes, resulting in a provide an important characteristic of a spintronic device where
tunneling magnetoresistance (TMR) effect4. this material is used as a metal electrode32,33. In the absence of
Contrary to the spin-polarized currents, spin-neutral currents spin-orbit coupling, the ballistic conductance g per unit area
are usually considered impractical for spintronics due to being along the z direction can be obtained in terms of two spin
unable to directly interact with the magnetic order parameter. components as follows34.
This fact challenges spintronics based on non-ferromagnetic
g ¼ g " þ g # ¼ eh ∑ðN "k þ N #k Þ;
2
materials, such as compensated antiferromagnets, which normally kk
ð1Þ
do not support spin-polarized currents. Due to being robust
against magnetic perturbations, the absence of stray fields, and ! R ∂f
ultrafast spin dynamics, antiferromagnets are considered as out- N k ð k k Þ ¼ _2 ∑ jvσnz j ! dkz : ð2Þ
n ∂Eσn ð kÞ
standing candidates to replace the widely used ferromagnets in !
the next generation spintronics5–9. This promising route has been Here σ denotes the spin component ↑ or ↓, k is the wave
recently stimulated by the demonstrated control of the anti- !
vector in the three-dimensional Brillouin zone, N σk ð k k Þ is the
ferromagnetic Néel vector by spin-orbit torques10,11. However,
the absence of a net magnetization and hence spin-independent number of conduction channels (integer) at the transverse wave
!
conductance makes the electrical detection of the Néel vector vector k k ¼ ðkx ; ky Þ for spin σ, Eσn is energy for the n-th band,
using conventional methods, such as TMR measurements, !
∂Eσn ð k Þ
unfeasible5. So far, the electrical detection of the Néel vector has vσnz ¼ _∂k is the band velocity along the transport z direction,
z
been performed using anisotropic10,11 or spin-Hall12–15 magne- and f is the Fermi distribution function.
toresistance. Unfortunately, both methods suffer from relatively The net transport spin polarization is defined by
small signals easily influenced by perturbations16 and require "
g #
multiple in-plane terminals resulting in large device dimensions7. p¼g g
ð3Þ
Antiferromagnetic spin valves17–20 and antiferromagnetic
tunnel junctions (AFMTJs)21,22 have been theoretically proposed, and represents an important quantity useful in spintronics. For
!
promising, in some cases, sizable magnetoresistance effects. example, in a crude approximation of k k -independent transmis-
However, these magnetoresistance effects rely on perfect inter- sion between two ferromagnetic electrodes with spin polarizations
faces and switching the interfacial magnetic moment alignment p1 and p2 in an MTJ, the TMR effect is given by the well-known
between parallel and antiparallel. This mechanism is not robust 2p p
Julliere’s formula2 TMR ¼ 1p1 p2 . Clearly, a larger spin polariza-
against disorder and interface roughness inevitable in experi- 1 2

mental conditions. Recent efforts have been aimed at exploring tion of the electrodes favors a larger TMR.
unconventional methods for the Néel vector detection based A large spin polarization p is generally expected for ferro-
on topological properties9,23–25 but require an experimental magnets where the finite net magnetization breaks time reversal
confirmation. symmetry T ^ . The latter flips the spin σ and changes sign of !
k k,
One promising direction is to create spin-polarized currents in ^ " ! # !
antiferromagnets. Recently, it has been predicted that certain types resulting in TN k ð k k Þ ¼ N k ðk k Þ. Compensated antiferromag-
of compensated antiferromagnets exhibit a momentum-dependent nets do not have net magnetization and hence (with some
spin splitting of the Fermi surface26–28, resulting in spin-polarized exceptions29–31) do not support the macroscopic spin-polarized
currents along certain crystallographic orientations29–31. These current. However, even though macroscopically the net transport
predictions indicate that these antiferromagnets can work as fer- spin polarization p is absent, microscopically the conductance
romagnets in spintronic devices, which broadens the range of could be spin polarized as reflected in the spin polarization of
!
materials useful for spintronics. conduction channels at k k :
Here, we embark on a different path and argue that globally ! N " N #
spin-independent conductance in compensated antiferromagnets pk ð k k Þ ¼ N k" þN k# : ð4Þ
k k
can be efficiently used in spintronics, provided their crystal
symmetry supports a non-spin-degenerate Fermi surface and thus If both electrodes in a two-terminal spintronic device are made
momentum-dependent spin polarization. While such a spin of materials with zero net spin polarization but have spin-
polarization is cancelled out in the net conductance due to being polarized conduction channels, this momentum-dependent spin
antisymmetric with respect to certain symmetry operations, its polarization will be reflected in the device conductance and can
presence in the momentum space can be functionalized if such an be functionalized through the antiferromagnetic Néel vector.
antiferromagnet is combined with another similar antiferro- Indeed, in the transport regime conserving spin (no spin-orbit
magnet in a spintronic device such as an AFMTJ. In this case, the !
coupling) and wave vector k k (no diffuse scattering), the device
resistance change of the AFMTJ occurs in response to the
conductance is largely affected by the spin matching of the
orientation of the antiferromagnetic Néel vector due to changing !
matching conditions between the spin-polarized conduction conduction channels k k of the electrodes. If their spin
channels in the two metal electrodes. These considerations can be polarization changes in response to the Néel vector rotation in
expanded to normal (nonmagnetic) metals with spin-orbit an antiferromagnetic electrode, this alters the net conductance of

2 NATURE COMMUNICATIONS | (2021)12:7061 | https://doi.org/10.1038/s41467-021-26915-3 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-26915-3 ARTICLE

Fig. 1 Spin-polarized conduction channels in different types of magnetic materials. a Schematics of the atomic structure (left) and spin-polarized Fermi
surface (right) for a ferromagnet. A spin-neutral current passing through the ferromagnet becomes spin polarized. b Schematics of the atomic structure
(left) and spin-degenerate Fermi surface (right) for a compensated antiferromagnet where the net magnetization is forbidden by ^ PT^ symmetry. A spin-
neutral current passing through this antiferromagnet remains spin neutral. c Schematics of the atomic structure (left) and non-spin-degenerate Fermi
^ and G
surface congruent for opposite spins (right) for a compensated antiferromagnet where the net magnetization is forbidden by glide symmetries G ^ .A
x z
spin-neutral current passing through this antiferromagnet remains globally spin neutral but has a momentum-dependent spin polarization. d Schematics of
an AFMTJ where two antiferromagnetic (AFM) layers are separated by a nonmagnetic (NM) barrier layer. The Néel vector (indicated by arrows) of the
bottom free layer can be switched resulting in the TMR effect.

the device. Thus, although the spin polarization of the charge vanishing net spin polarization p for the current along the z
! direction. On the other hand, due to the Néel vector pointing
current remains zero, the device conductance reflects the k k - !
dependent spin polarization of the conduction channels. along the z axis, the symmetry transformation G ^ z Eσn ð k k ; kz Þ ¼
Next, we identify magnetic space group symmetry require- !
! Eσn ð k k ; kz Þ conserves the spin σ ¼"; # of the conduction
ments for crystals to exhibit the k k -dependent spin polarization. !
modes at k k and there is no symmetry operation which would
Obviously, in ferromagnets, the conduction channels are spin
enforce pk ¼ 0. The presence of spin-polarized conduction
polarized due to the spin-dependent Fermi surface. Thus, passing
a spin-neutral current through a ferromagnetic material makes it channels in this type of antiferromagnets can be understood in
spin polarized (Fig. 1a). On the contrary, most compensated terms of two congruent (but not identical) up- and down-spin
antiferromagnets contain symmetries that not only prevent Fermi surfaces, which are transformed to each other by the
the net magnetization but also lead to a spin-degenerate symmetry transformation G ^ x (Fig. 1c). In this case, each
Fermi surface and thus spin-independent conduction channels. conduction channel is spin polarized (except high-symmetry
^T^ ! ^ x ), whereas the net conductance is spin
For example, if a compensated antiferromagnet exhibits P k k points invariant to G
symmetry, where P ^ and T^ are space inversion and time reversal neutral.
symmetries, respectively, pk ¼ 0 due to P ^ " ð!
^ TN # ! Due to the non-spin-degenerate Fermi surface and spin-
k k k Þ ¼ N k ð k k Þ.
polarized conduction channels, the globally spin-neutral conduc-
This property follows from the spin-degenerate Fermi surface due
tion of the compensated antiferromagnets can be exploited in
^ " ð!
^ TE #
!
to P n k Þ= En ð k Þ (Fig. 1b). The spin degeneracy also appears spintronic devices, such as AFMTJs. Fig 1d shows an AFMTJ
in compensated antiferromagnets with T ^^t symmetry (^t is half a which contains two identical antiferromagnetic electrodes sepa-
unit cell translation) in the absence of spin-orbit coupling. rated by a nonmagnetic insulating spacer. The antiferromagnets
The spin degeneracy is however broken in compensated are assumed to have spin-polarized conduction channels along the
antiferromagnets belonging to magnetic space groups with out-of-plane transport direction. The functionality of the AFMTJ
violated P^T^ and T^^t symmetries26. The vanishing net magnetiza- is controlled by the relative orientation of the Néel vector of the
tion in such antiferromagnets originates from the combination of two antiferromagnetic electrodes. In the parallel state, the spin-
some other magnetic space group symmetries of the crystal. For polarized conduction channels of the electrodes perfectly match,
example, Fig. 1c shows a collinear antiferromagnet with the Néel resulting in a low resistance state. In the antiparallel state, the spin
vector pointing along the z direction. The zero net magnetization polarized conduction channels are mismatched, resulting in a high
in this antiferromagnet is guaranteed by two glide symmetries G ^x resistance state.
^ ^ ^ ^
and Gz , where Gl ¼ fM l j^t g represents mirror symmetry M l with a
! Electronic structure of RuO2. To demonstrate this spintronic
mirror plane normal to vector l combined with translation ^t . functionality, we consider the recently discovered room-
The symmetry transformation G ^ x N " ðkx ; ky Þ ¼ N # ðkx ; ky Þ flips temperature antiferromagnetic metal RuO235 suitable for realiz-
k k
the spin and thus according to Eqs. (1) and (2) results in a ing the proposed AFMTJ. RuO2 exhibits interesting properties

NATURE COMMUNICATIONS | (2021)12:7061 | https://doi.org/10.1038/s41467-021-26915-3 | www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-26915-3

Fig. 2 Electronic properties of RuO2. a The atomic and magnetic structures of RuO2. b The calculated band structure of RuO2. c The number of ~ kk -resolved
conduction channels in the 2D Brillouin zone of RuO2 for spin up N"k (left) and spin down N#k (right). High-symmetry ~
kk -points
 are indicated. N"k and N#k can
be transformed to each other by the glide transformations G ^ or G^ . d Spin polarization of conduction channels p ~
x y k kk . Gray contrast indicates regions
" #
where Nk ¼ Nk ¼ 0 and thus pk undefined.

!
a P state number of conduction channels N σk ð k k Þ in the two-dimensional
AP state (2D) Brillouin zone of RuO2. As seen from Fig. 2c, the
Ru Ti
distributions of N "k and N #k in the ðkx ; ky Þ plane have congruent
O
shapes and are symmetric with respect to the G ^ y symmetry
^ x and G
b transformations, which enforce a zero global spin polarization in
2
the conductance along the z direction. On the contrary, as seen
!
E-EF (eV)

0 from Fig. 2d, the k k -dependent spin polarization pk remains

finite across a sizable portion of the 2D Brillouin zone. This
-2
demonstrates that RuO2 (001) exhibits a globally spin-neutral
conductance through spin-polarized conduction channels.
Fig. 3 Atomic and electronic structure of a RuO2/TiO2/RuO2 supercell.
a, b The atomic structure (a) and layer resolved density of states (DOS) (b)
of the RuO2/TiO2/RuO2 supercell. Each panel in b contains two atomic TMR in a RuO2/TiO2/RuO2 AFMTJ. Next, we design an AFMTJ
layers of MO2 (M= Ru, Ti) and has left and right subpanels corresponding using RuO2 (001) as electrodes and TiO2 (001) as an insulating
to up- and down-spin states, respectively. The arrows in (a) indicate the barrier layer. Due to both having a rutile structure and a similar
magnetic moments of Ru atoms. lattice constant, this AFMTJ is feasible in practice. Fig 3a shows
the atomic structure of the RuO2/TiO2/RuO2 (001) supercell,
which is used in our DFT and quantum transport calculations
such as spin splitting without spin-orbit coupling36, a crystal Hall and includes 8 TiO2 layers in the center and 10 RuO2 layers on
effect37, and a magnetic spin Hall effect31,38. RuO2 has a rutile each side. We find that a wide band gap of TiO2 is well main-
structure with an out-of-plane Néel vector (Fig. 2a) and magnetic tained in this heterostructure, and the Fermi energy EF is located
space group P42'/mnm', which contains glide G ^ x ¼ fM
^ x jð1 ; 1 ; 1Þg, deeply inside the band gap (Fig. 3b).
2 2 2
^ ^ ^
Gy ¼ fM y jð2 ; 2 ; 2Þg and mirror M z symmetries. In the absence of
1 1 1
The RuO2/TiO2/RuO2 (001) structure in Fig. 3a is then used as
spin-orbit coupling, the energy bands are spin degenerate at the the scattering region of the AFMTJ connected to two semi-
k-planes invariant to G ^ y , such as kx ¼ 0; π or ky ¼ 0; π.
^ x and G infinite RuO2 (001) electrodes for calculating transmission. The
2 2
transmission is obtained for parallel (Fig. 4a) and antiparallel
This is evident from our first-principles density functional theory
(Fig. 4b) alignments of the Néel vectors of the electrodes. For the
(DFT) calculations. As seen from Fig. 2b, the energy bands of ! !
RuO2 are spin degenerate along the Г–X, Г–Z, X–M, Z–R, and parallel-aligned AFMTJ, the k k -resolved transmission T σP ð k k Þ is
R–A directions lying in these glide-invariant planes. On the other shown in Fig. 4c for spin up (left panel) and spin down (right
hand, a large spin splitting appears along the directions away !
panel). The clearly seen spin asymmetry between T "P ð k k Þ and
from these planes, such as Г–M and Z–A. !
The spin-dependent band structure leads to the momentum- T #P ð k k Þ reflects the related asymmetry in the distribution of the
dependent spin polarization of the conduction channels along the spin-polarized conduction channels in RuO2 (Fig. 2c). The
z direction. We explicitly demonstrate this by calculating the suppressed transmission near the Brillouin zone corners is due to

4 NATURE COMMUNICATIONS | (2021)12:7061 | https://doi.org/10.1038/s41467-021-26915-3 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-26915-3 ARTICLE

a P State b AP State
Electrode Barrier Electrode Electrode Barrier Electrode
Ti
Ru
O

c d
TP
↑
TP
↓
TAP
↑
TAP
↓

ky

kx
-4 -4
0 4×10 0 4×10
e f1500
-3 ●
10 ●
TMR=(TP-TAP)/TAP
Transmission

● 1000
● TMR (%)
●
10
-4 ◆◆◆●
◆● ● ● ●● ● 500 ● ● ● ● ● ●●● ● ●
◆ ◆◆ ● ●●● ● ● ● ●
◆ ● ●
●
10-5 ◆ ◆◆◆ ●●●●●
◆◆◆◆◆◆◆ 0

-0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4
E-EF (eV) E -EF (eV)
Fig. 4 Giant TMR in RuO2/TiO2/RuO2 AFMTJ. a, b The atomic and magnetic structures of RuO2/TiO2/RuO2 AFMTJ for parallel (P) (a) and antiparallel
(AP) (b) alignment of the Néel vectors. c, d The calculated ~
kk -resolved transmission in the 2D Brillouin zone for the AFMTJ in P (c) and AP (d) states.
e Total transmission as a function of energy for the AFMTJ in P (red dots) and AP (blue dots) states. f TMR as a function of energy.

! it also more robust against other types of disorder, as long as the

a larger decay rate of the evanescent states for k k away from the
crystallinity of the tunnel junction and the direct tunneling
zone center.
transport mechanism are maintained (see Section C of Supple-
For the antiparallel-aligned AFMTJ, the transmission
mental Material).
σ ! !
T AP ð k k Þ is blocked for the wave vectors k k with no conduction
!
channels in one of the spin states, i.e. N σk ð k k Þ ¼ 0 for σ ¼" or
Discussion
σ ¼#. These are the regions where pk ¼ ± 1 or N "k ¼ N #k ¼ 0 in The above properties are sustained in the presence of spin-orbit
Fig. 2c and d. As a result, only the states near the zone center coupling. This is due to the momentum-dependent spin polar-
where the spin channels are degenerate (enforced by G ^ y)
^ x and G ization in rutile antiferromagnets being inherited from the anti-
contribute to the transmission (Fig. 4d). This leads to the total ferromagnetic order rather than spin-orbit coupling27,31,37. In
transmission being much smaller for the antiparallel state (TAP) RuO2, the M ^ z symmetry transformation reverses the wave vector
than for the parallel state (TP) (Fig. 4e). At the Fermi energy EF , component kz , conserves the spin component σ z but flips σ x and
!
we find the TMR ratio ðT P  T AP Þ=T AP as large as ~500%. σ y . As a result, the conduction modes at k k are spin polarized
This value is comparable to the values obtained for the well- purely along the z axis and the x- and y-spin components vanish
known Fe/MgO/Fe (001) MTJs39,40 which are currently used in (see Supplementary Section D). Supplementary Fig. S5e shows the
magnetic random-access memories. The giant TMR appears not results of the DFT calculation for RuO2 in the presence of spin-
only for E ¼ EF but also for the energies around the Fermi level, orbit coupling. It is seen that the spin polarization of the most
with the smallest value of ~50% at E ¼ EF  0:25 eV (Fig. 4f). conduction channels is well preserved, indicating that the giant
This fact indicates that the large TMR will be sustained under an TMR is robust against spin-orbit coupling in the RuO2
applied bias voltage. based AFMTJ.
The predicted TMR is largely independent of the interface The spin-polarized conduction channels are not limited to
terminations and the relative alignment of the interface magnetic ^T
RuO2, but typical for a wide group of materials with violated P ^
moments, as follows from our explicit DFT calculations symmetry, including those with a noncollinear antiferromagnetic
(Supplementary Figs. S1 and S2). This distinguishes our results order29. For the antiferromagnets with magnetization being
from the previous findings21,22, where the interface termination compensated by combined mirror and/or rotation symmetries,
controls TMR and implies that the predicted TMR effect is likely the spin-polarized conduction channels can exist purely
less sensitive to the interface roughness than that in the previous due to the antiferromagnetic order, even in the absence of spin-
studies. The bulk origin of TMR in the proposed AFMTJs makes orbit coupling26–31. On the other hand, in compensated

NATURE COMMUNICATIONS | (2021)12:7061 | https://doi.org/10.1038/s41467-021-26915-3 | www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-26915-3

antiferromagnets with T ^^t symmetry, the spin-polarized con- quantum transport calculations, we have demonstrated such
duction channels appear due to the spin degeneracy lifted by functionality using a room-temperature antiferromagnetic metal
spin-orbit coupling. Such antiferromagnets can also be used in RuO2 as electrodes in a RuO2/TiO2/RuO2 AFMTJ and predicted a
AFMTJs if they have sizable spin-orbit splitting. giant TMR effect of ~500%. Our work uncovers an unexplored
These considerations can be expanded to normal metals with potential of the materials with no global spin polarization for
broken space inversion symmetry, where the band spin degen- utilizing them in spintronics. We hope therefore that our pre-
eracy is lifted by spin-orbit interaction. In these materials, the two dictions will stimulate experimental investigations of these
conduction channels with opposite spin polarizations are linked materials and the associated phenomena.
by the time reversal symmetry operation. For example, in topo- Note added: During the review of this manuscript, we became
logical metal TaN41,42, the conduction channels along the [001] aware of the relevant work by Šmejkal et al. posted recently48.
direction carry the spin polarization pointing along the same
[001] direction (Supplementary Fig. S6). This property is Methods
enforced by the M ^ z mirror symmetry. While the momentum- The atomic and electronic structures shown in Figs. 2a, b, 3, S5a, and S6a, b of the
dependent spin polarization in non-centrosymmetric normal systems are calculated using the projector augmented wave (PAW) method49
implemented in the VASP code50. A plane-wave cut-off energy of 500 eV and a
metals is fixed by their crystal symmetry and band structure, they !
16 × 16 × 16 k -point mesh in the irreducible Brillouin zone are used in the cal-
can be used in spintronics in conjunction with antiferromagnets. culations. The exchange and correlation effects are treated within the generalized
For example, the antiferromagnetic reference layer in the AFMTJ gradient approximation (GGA) developed by Perdew-Burke-Ernzerhof (PBE)51.
in Fig. 1d can be replaced by a normal metal layer. Alternatively, The GGA+U functional52,53 with Ueff = 2 eV on Ru 4d orbitals and Ueff = 5 eV on
one can create an antiferromagnet/normal metal interface. In Ti 3d orbitals is included in the calculations.
The transport properties shown in Fig. 4 and S1–S3 are calculated using the non-
such systems with a single antiferromagnetic layer, the spintronic equilibrium Green’s function formalism (DFT+NEGF approach)54,55, as imple-
functionality is controlled by the Néel vector orientation that mented the Atomistic Simulation Toolkit (ATK) distributed in the QuantumWise
regulates a matching of the conduction channels in the anti- package (Version 2015.1) (ATOMISTIX TOOLKIT version 2015.1 Synopsys
ferromagnetic and normal metal layers. QuantumWise (www.quantumwise.com). QuantumWise A/S is now part of
Synopsys, and from the upcoming version ATK will be part of the QuantumATK
Note that in junctions with a single antiferromagnetic layer, suite)56. The atomic structures are relaxed by VASP and the nonrelativistic Fritz-
reversal of the Néel vector is equivalent to the time-reversal Haber-Institute (FHI) pseudopotentials using a single-zeta-polarized basis. The
transformation which does not change the resistance. However, spin polarized GGA+U functional51,52 with Ueff = 2.3 eV on Ru 4d orbitals and
the resistance changes with rotation of the Néel vector, resulting Ueff = 5 eV on Ti 3d orbitals is included in the calculations. A cut-off energy of 75
!
in a tunnelling anisotropic magnetoresistance (TAMR) effect43. Ry and a 11×11×101 k -point mesh are used for the self-consistent calculations to
eliminate the mismatch of the Fermi level between the electrodes and the central
Another possibility is to utilize non-centrosymmetric insula- region. Unless mentioned in the text, the transmission is calculated using an
tors as a tunneling barrier layer in an AFMTJ. Due to the broken !
adaptive k -point mesh. These parameters are confirmed to yield a good balance
space inversion symmetry and spin-orbit coupling, the evanescent between the computational time and accuracy.
states in these insulators are spin-polarized44,45. Therefore, the The tight-binding Hamiltonians of RuO2 and TaN are obtained using Wan-
nier90 code57 utilizing the maximally localized Wannier functions58. A 500 × 500 ×
Néel vector of the free antiferromagnetic layer can be used to !
500 k -point mesh and the adaptive smearing method59 are used to calculate the
control the matching between the propagating Bloch states in the !
k k -resolved ballistic conductance shown in Fig. 2c, d, S5e and S6c, d. The spin-
antiferromagnetic electrode and the evanescent gap states in the projected Fermi surfaces of RuO2 with spin-orbit coupling shown in Figs. S5b–d
barrier resulting in a TAMR effect. An additional useful func- are calculated using WannierBerri code60,61.
tionality of this kind of tunnel junctions may be provided by a Figures are plotted using VESTA62, FermiSurfer63, gnuplot64, and the SciDraw
scientific figure preparation system65.
switchable polarization of the non-centrosymmetric insulating
barrier layer if it is ferroelectric45.
The proposed use of spin-neutral currents in spintronics is Data availability
The data that support the findings of this study are available from the corresponding
feasible from the experimental perspective. For example, the author upon reasonable request.
proposed RuO2/TiO2/RuO2 (001) AFMTJ has all rutile structure
with a good match of the RuO2 and TiO2 lattice constants and
thus can be grown epitaxially preserving crystallinity of the Received: 29 March 2021; Accepted: 25 October 2021;
overall heterostructure. The Néel vector of the antiferromagnetic
free layer can be switched by a spin-orbit torque via the spin
current from an adjacent heavy metal layer generated by an in-
plane charge current8,46. With the in-plane writing path and
out-of-plane reading path, only two in-plane terminals and one References
out-of-plane terminals are required for such an AFMTJ, which is 1. Tsymbal, E. Y. & Žutić I. Spintronics Handbook: Spin Transport and
Magnetism, 2-nd edition. (CRC press, 2019).
desirable for nanoscale spintronic applications. In addition, the 2. Julliere, M. Tunneling between ferromagnetic films. Phys. Lett. 54A, 225–226
large magnitude of TMR indicates a possibility of a strong spin (1975).
transfer torque in the AFMTJs, which may be robust against 3. Moodera, J. S., Kinder, L. R., Wong, T. M. & Meservey, R. Large
disorder47 and may offer an alternative way to switch the Néel magnetoresistance at room temperature in ferromagnetic thin film tunnel
vector. junctions. Phys. Rev. Lett. 74, 3273–3276 (1995).
4. Tsymbal, E. Y., Mryasov, O. N. & LeClair, P. R. Spin-dependent tunneling in
In conclusion, we have proposed that globally spin-neutral magnetic tunnel junctions. J. Phys.: Condens. Matter 15, R109–R142 (2003).
currents flowing through oppositely spin-polarized conduction 5. Baltz, V. et al. Antiferromagnetic spintronics. Rev. Mod. Phys. 90, 015005
channels can be efficiently used in spintronics. Such currents exist (2018).
in compensated antiferromagnets and normal metals with the 6. Jungwirth, T., Marti, X., Wadley, P. & Wunderlich, J. Antiferromagnetic
magnetic space group symmetries which lift the spin-degeneracy spintronics. Nat. Nanotech. 11, 231–241 (2016).
7. Jungwirth, T. et al. The multiple directions of antiferromagnetic spintronics.
of the Fermi surface. In the heterostructures, such as anti- Nat. Phys. 14, 200–203 (2018).
ferromagnetic tunnel junctions or antiferromagnet/normal metal 8. Železný, J., Wadley, P., Olejník, K., Hoffmann, A. & Ohno, H. Spin transport
interfaces, these currents can be controlled by the Néel vector and spin torque in antiferromagnetic devices. Nat. Phys. 14, 220–228 (2018).
orientation providing a useful functionality for spintronics. Based 9. Šmejkal, L., Mokrousov, Y., Yan, B. & MacDonald, A. H. Topological
on first-principles density functional theory combined with antiferromagnetic spintronics. Nat. Phys. 14, 242–251 (2018).

6 NATURE COMMUNICATIONS | (2021)12:7061 | https://doi.org/10.1038/s41467-021-26915-3 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-26915-3 ARTICLE

10. Wadley, P. et al. Electrical switching of an antiferromagnet. Science 351, 42. Weng, H., Fang, C., Fang, Z. & Dai, X. Topological semimetals with triply
587–590 (2016). degenerate nodal points in θ-phase tantalum nitride. Phys. Rev. B 93, 241202
11. Bodnar, S. Y. et al. Writing and reading antiferromagnetic Mn2Au by Néel (2016).
spin-orbit torques and large anisotropic magnetoresistance. Nat. Commun. 9, 43. Park, B. G. et al. & Jungwirth, T. A spin-valve-like magnetoresistance of an
348 (2018). antiferromagnet-based tunnel junction. Nat. Mater. 10, 347–351 (2011).
12. Hoogeboom, G. R., Aqeel, A., Kuschel, T., Palstra, T. T. M. & van Wees, B. J. 44. Mishra, S., Thulasi, S. & Satpathy, S. Spin polarization via electron tunneling
Negative spin Hall magnetoresistance of Pt on the bulk easy-plane through an indirect-gap semiconductor barrier. Phys. Rev. B 72, 195347
antiferromagnet NiO. Appl. Phys. Lett. 111, 052409 (2017). (2005).
13. Chen, X. Z. et al. Antidamping-torque-induced switching in biaxial 45. Tao, L. L. & Tsymbal, E. Y. Perspectives of spin-textured ferroelectrics. J. Phys.
antiferromagnetic insulators. Phys. Rev. Lett. 120, 207204 (2018). D: Appl. Phys. 54, 113001 (2020).
14. Fischer, J. et al. Spin Hall magnetoresistance in antiferromagnet/ heavy-metal 46. Manchon, A. et al. Current-induced spin-orbit torques in ferromagnetic and
heterostructures. Phys. Rev. B 97, 014417 (2018). antiferromagnetic systems. Rev. Mod. Phys. 91, 035004 (2019).
15. Baldrati, L. et al. Full angular dependence of the spin Hall and ordinary 47. Saidaoui, H., Manchon, A. & Waintal, X. Robust spin transfer torque in
magnetoresistance in epitaxial antiferromagnetic NiO(001)/Pt thin films. Phys. antiferromagnetic tunnel junctions. Phys. Rev. B 95, 134424 (2017).
Rev. B 98, 024422 (2018). 48. Šmejkal, L., Hellenes, A. B., González-Hernández, R., Sinova, J. & Jungwirth,
16. Chiang, C. C., Huang, S. Y., Qu, D., Wu, P. H. & Chien, C. L. Absence of T. Giant and tunneling magnetoresistance effects from anisotropic and valley-
evidence of electrical switching of the antiferromagnetic Néel vector. Phys. dependent spin-momentum interactions in antiferromagnets. arXiv 2103,
Rev. Lett. 123, 227203 (2019). 12664 (2021).
17. Núñez, A. S., Duine, R. A., Haney, P. & MacDonald, A. H. Theory of spin 49. Kresse, G. & Joubert, D. From ultrasoft pseudopotentials to the projector
torques and giant magnetoresistance in antiferromagnetic metals. Phys. Rev. B augmented-wave method. Phys. Rev. B 59, 1758–1775 (1999).
73, 214426 (2006). 50. Kresse, G. & Furthmuller, J. Efficient iterative schemes for ab initio total-energy
18. Saidaoui, H. B. M., Manchon, A. & Waintal, X. Spin transfer torque in calculations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996).
antiferromagnetic spin valves: From clean to disordered regimes. Phys. Rev. B 51. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation
89, 174430 (2014). made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).
19. Xu, Y., Wang, S. & Xia, K. Spin-transfer torques in antiferromagnetic metals 52. Dudarev, S. L., Botton, G. A., Savrasov, S. Y., Humphreys, C. J. & Sutton, A. P.
from first principles. Phys. Rev. Lett. 100, 226602 (2008). Electron-energy-loss spectra and the structural stability of nickel oxide: An
20. Manchon, A. Spin diffusion and torques in disordered antiferromagnets. J. LSDA+U study. Phys. Rev. B 57, 1505–1509 (1998).
Phys. Condens. Matter 29, 104002 (2017). 53. Anisimov, V. I., Zaanen, J. & Andersen, O. K. Band theory and Mott
21. Merodio, P., Kalitsov, A., Ba, H., Baltz, V. & Chshiev, M. Spin-dependent transport insulators: Hubbard U instead of Stoner. I. Phys. Rev. B 44, 943–954 (1991).
in antiferromagnetic tunnel junctions. Appl. Phys. Lett. 105, 122403 (2014). 54. Taylor, J., Guo, H. & Wang, J. Ab initio modeling of quantum transport
22. Stamenova, M., Mohebbi, R., Seyed-Yazdi, J., Rungger, I. & Sanvito, S. First- properties of molecular electronic devices. Phys. Rev. B 63, 245407 (2001).
principles spin-transfer torque in CuMnA|GaP|CuMnAs junctions. Phys. Rev. 55. Brandbyge, M., Mozos, J. L., Ordejón, P., Taylor, J. & Stokbro, K. Density-
B 95, 060403 (2017). functional method for nonequilibrium electron transport. Phys. Rev. B 65,
23. Šmejkal, L., Železný, J., Sinova, J. & Jungwirth, T. Electric control of Dirac 165401 (2002).
quasiparticles by spin-orbit torque in an antiferromagnet. Phys. Rev. Lett. 118, 56. Smidstrup, S. et al. QuantumATK: An integrated platform of electronic and
106402 (2017). atomic-scale modelling tools. J. Phys. Condens. Matter 32, 015901 (2019).
24. Shao, D.-F., Gurung, G., Zhang, S.-H. & Tsymbal, E. Y. Dirac nodal line metal 57. Pizzi, G. et al. Wannier90 as a community code: new features and applications.
for topological antiferromagnetic spintronics. Phys. Rev. Lett. 122, 077203 J. Phys. Cond. Mat. 32, 165902 (2020).
(2019). 58. Marzari, N., Mostofi, A. A., Yates, J. R., Souza, I. & Vanderbilt, D. Maximally
25. Shao, D.-F., Zhang, S.-H., Gurung, G., Yang, W. & Tsymbal, E. Y. Nonlinear localized Wannier functions: Theory and applications. Rev. Mod. Phys. 84,
anomalous Hall effect for Néel vector detection. Phys. Rev. Lett. 124, 067203 1419 (2012).
(2020). 59. Yates, J. R., Wang, X., Vanderbilt, D. & Souza, I. Spectral and Fermi surface
26. Hayami, S., Yanagi, Y. & Kusunose, H. Momentum-dependent spin splitting properties from Wannier interpolation. Phys. Rev. B 75, 195121 (2007).
by collinear antiferromagnetic ordering. J. Phys. Soc. Jpn. 88, 123702 (2019). 60. Tsirkin, S. S. High performance Wannier interpolation of Berry curvature
27. Yuan, L.-D., Wang, Z., Luo, J.-W., Rashba, E. I. & Zunger, A. Giant and related quantities with WannierBerri code. npj Comput. Mater. 7, 33
momentum-dependent spin splitting in centrosymmetric low-Z (2021).
antiferromagnets. Phys. Rev. B 102, 014422 (2020). 61. Destraz, D. et al. Magnetism and anomalous transport in the Weyl
28. Yuan, L.-D., Wang, Z., Luo, J.-W. & Zunger, A. Prediction of low-Z collinear semimetal PrAlGe: Possible route to axial gauge fields. npj Quantum Mater. 5,
and noncollinear antiferromagnetic compounds having momentum- 5 (2020).
dependent spin splitting even without spin-orbit coupling. Phys. Rev. Mater. 5, 62. Momma, K. & Izumi, F. VESTA 3 for three-dimensional visualization of crystal,
014409 (2021). volumetric and morphology data. J. Appl. Crystallogr. 44, 1272–1276 (2011).
29. Železný, J., Zhang, Y., Felser, C. & Yan, B. Spin-polarized current in 63. Kawamura, M. FermiSurfer: Fermi-surface viewer providing multiple
noncollinear antiferromagnets. Phys. Rev. Lett. 119, 187204 (2017). representation schemes. Comp. Phys. Commun. 239, 197–203 (2019).
30. Naka, M. et al. Spin current generation in organic antiferromagnets. Nat. 64. Williams T. & Kelley, C. Gnuplot 5.2: An interactive plotting program,
Commun. 10, 4305 (2019). http://www.gnuplot.Info (2017).
31. González-Hernández, R. et al. Efficient electrical spin-splitter based on non- 65. Caprio, M. A. LevelScheme: a level scheme drawing and scientific figure
relativistic collinear antiferromagnetism. Phys. Rev. Lett. 126, 127701 (2021). preparation system for Mathematica. Comp. Phys. Commun. 171, 107–118 (2005).
32. Landauer, R. Spatial variation of currents and fields due to localized scatterers
in metallic conduction. IBM J. Res. Dev. 32, 306–316 (1988).
33. Datta, S. Electronic Transport in Mesoscopic Systems (Cambridge University Acknowledgements
Press, 1995). The authors thank Bo Li for helpful discussions. This work was supported by the Vannevar
34. Schep, K. M., Kelly, P. J. & Bauer, G. E. W. Ballistic transport and electronic Bush Faculty Fellowship (ONR grant N00014-20-1-2844) (C.B.E) and by the National
structure. Phys. Rev. B 57, 8907–8926 (1998). Science Foundation (NSF) through the MRSEC (NSF Award DMR-1420645) and EPSCoR
35. Berlijn, T. et al. Itinerant antiferromagnetism in RuO2. Phys. Rev. Lett. 118, RII Track-1 (NSF Award OIA-2044049) programs (E.Y.T.). S.-H.Z. thanks the support of
077201 (2017). National Science Foundation of China (NSFC Grant No. 12174019). Computations were
36. Ahn, K.-H., Hariki, A., Lee, K.-W. & Kuneš, J. RuO2 Antiferromagnetism in as performed at the University of Nebraska Holland Computing Center.
d-wave Pomeranchuk instability. Phys. Rev. B 99, 184432 (2019).
37. Šmejkal, L., González-Hernández, R., Jungwirth, T. & Sinova, J. Crystal Hall
Author contributions
effect in collinear antiferromagnets. Sci. Adv. 6, eaaz8809 (2020).
D.-F.S. and E.Y.T. conceived this project. D.-F.S. performed symmetry analysis and
38. Bose, A. et al. Tilted spin current generated by the collinear antiferromagnet
calculated electronic properties for RuO2 and TaN. S.H.Z calculated the ballistic con-
RuO2. arXiv 2108, 09150 (2021).
ductance of RuO2 and TaN. D.-F.S. and M.L. calculated the transport properties of RuO2/
39. Yuasa, S., Nagahama, T., Fukushima, A., Suzuki, Y. & Ando, K. Giant room-
TiO2/RuO2 AFMTJ. D.F.S., S.H.Z., M.L. and E.Y.T. analyzed the results and discussed
temperature magnetoresistance in single-crystal Fe/MgO/Fe magnetic tunnel
with C.B.E. D.-F.S. and E.Y.T. wrote the manuscript. All authors contributed to the final
junctions. Nat. Mater. 3, 868–871 (2004).
version of the manuscript.
40. Parkin, S. S. P. et al. Giant tunnelling magnetoresistance at room temperature
with MgO (100) tunnel barriers. Nat. Mater. 3, 862–867 (2004).
41. Zhu, Z., Winkler, G. W., Wu, Q., Li, J. & Soluyanov, A. A. Triple point Competing interests
topological metals. Phys. Rev. X 6, 031003 (2016). The authors declare no competing interests.

NATURE COMMUNICATIONS | (2021)12:7061 | https://doi.org/10.1038/s41467-021-26915-3 | www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-26915-3

Additional information Open Access This article is licensed under a Creative Commons
Supplementary information The online version contains supplementary material Attribution 4.0 International License, which permits use, sharing,
available at https://doi.org/10.1038/s41467-021-26915-3. adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Correspondence and requests for materials should be addressed to Ding-Fu Shao or Commons license, and indicate if changes were made. The images or other third party
Evgeny Y. Tsymbal. material in this article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
Peer review information Nature Communications thanks the anonymous reviewer(s) for
article’s Creative Commons license and your intended use is not permitted by statutory
their contribution to the peer review of this work.
regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
Reprints and permission information is available at http://www.nature.com/reprints
licenses/by/4.0/.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations. © The Author(s) 2021

8 NATURE COMMUNICATIONS | (2021)12:7061 | https://doi.org/10.1038/s41467-021-26915-3 | www.nature.com/naturecommunications