Theory of the Interfacial Dzyaloshinskii-Moriya Interaction in Rashba

Antiferromagnets

Alireza Qaiumzadeh,1 Ivan A. Ado,2 Rembert A. Duine,1, 3, 4 Mikhail Titov,2, 5 and Arne Brataas1
1
Center for Quantum Spintronics, Department of Physics,
Norwegian University of Science and Technology, NO-7491 Trondheim, Norway
2
Radboud University, Institute for Molecules and Materials, 6525 AJ Nijmegen, Netherlands
3
Institute for Theoretical Physics and Center for Extreme Matter and Emergent Phenomena,
Utrecht University, Princetonplein 5, 3584 CC Utrecht, Netherlands
4
Department of Applied Physics, Eindhoven University of Technology,
arXiv:1709.09205v2 [cond-mat.mes-hall] 10 May 2018

P.O. Box 513, 5600 MB Eindhoven, Netherlands

5
ITMO University, Saint Petersburg 197101, Russia
(Dated: May 14, 2018)
In antiferromagnetic (AFM) thin films, broken inversion symmetry or coupling to adjacent heavy
metals can induce Dzyaloshinskii-Moriya (DM) interactions. Knowledge of the DM parameters is
essential for understanding and designing exotic spin structures, such as hedgehog Skyrmions and
chiral Néel walls, which are attractive for use in novel information storage technologies. We introduce
a framework for computing the DM interaction in two-dimensional Rashba antiferromagnets. Unlike
in Rashba ferromagnets, the DM interaction is not suppressed even at low temperatures. The
material parameters control both the strength and sign of the interfacial DM interaction. Our
results suggest a route toward controlling the DM interaction in AFM materials by means of doping
and electric fields.

Relativistic spin-orbit coupling (SOC) is the founda- (a)

HM
tion of spin orbitronics, a rapidly developing branch of Top Layer
spintronics [1, 2]. Anisotropic magnetoresistance [3] and
the anomalous Hall effect [4] are established SOC trans- Bottom layer
port phenomena. More recent discoveries include the (b) (c)
spin Hall and inverse spin Hall effects [5], topological sur-
face states [2, 6, 7], spin-orbit torques [1, 8–11], and chi- ‫ܧ‬ఎ ‫ ܧ‬௦ఎ
ral domain walls and Skyrmions [12–18]. These phenom-
ena are essential to enable novel ultrafast, nonvolatile,
nanoscale spin-based storage and computation devices. ܽ݇ ܽ݇
The Dzyaloshinskii-Moriya (DM) interaction between
localized spins in a magnetic material is induced by SOC FIG. 1: (a) The Rashba-AFM model is a model for systems
[19, 20]. The DM interaction is important for stabilizing with inversion asymmetry: an AFM/HM bilayer (left) and an
AFM thin film asymmetrically embedded between two differ-
noncollinear magnetic structures in ferromagnets. The
ent insulator layers (right). (b) , (c) Electronic dispersion
original proposal by Dzyaloshinskii and Moriya, however, relations in the limit of strong exchange coupling (b) and in
concerned antiferromagnetic (AFM) systems with weak the limit of high SOC (c). The superscript η = ± refers to
magnetization, i.e., weak ferromagnets; this weak mag- the conduction (blue) and valence (red) bands, whereas the
netism can be explained in terms of antisymmetric ex- superscript s = ± specifies chiral bands. In these two extreme
change, which is also referred to as the DM interaction. limits, the band structure is almost isotropic for an AFM sys-
The energy associated with the DM interaction be- tem with in-plane anisotropy.
tween two neighboring spins in a lattice, S1 and S2 , can
be written in the form of the mixed product HDM =
−D12 · S1 × S2 , where D12 is called the DM vector. The In an ultrathin magnetic film with impurity- or interface-
DM interaction, therefore, favors a perpendicular orien- induced SOC, there is also an interfacial DM interaction
tation of neighboring spins. By contrast, the Heisenberg that corresponds to a DM vector pointing out along the
exchange interaction Hex = Jex S1 · S2 favors a collinear interface. This gives rise to so-called Néel-type textures
magnetic order, which can be either ferromagnetic (FM) [14, 21–26]. Both the sign and the amplitude of the DM
for Jex < 0 or AFM for Jex > 0. The competition be- vector are of critical importance for observing and engi-
tween the Heisenberg exchange interaction and the DM neering different chiral structures [27–40, 43–45].
interaction leads to the formation of exotic structures AFM materials have recently attracted considerable
such as chiral domain walls, helices, and Skyrmions [2]. attention as active spintronics elements [46, 47]. The
In the bulk of a noncentrosymmetric magnetic crys- absence of stray fields, the possibility of operating at ter-
tal, the DM vector points along one of the lattice vectors ahertz frequencies, and the existence of spin waves with
and gives rise to so-called Bloch-like structures [19–21]. opposite helicities make antiferromagnets promising can-
 2

didates for use in the next generation of spin-based mem- (a) (b) *z
ory and processing devices [46, 47] with a nanoscale ele- q ,i Zn q ,i Zn
ment base. *b *a Xy
q ,i Zn
The coexistence of strong SOC and spin ordering
at the interfaces of AFM/heavy-metal (HM) bilayers q  k ,iZn  i Q l q ,i Zn *y
makes such heterostructures particularly promising for
low-dimensional spin orbitronics applications [48]. An
AFM thin film sandwiched between insulators is also a FIG. 2: (a) The Feynman diagram corresponding to Eq. (4)
common functional geometry for spin-orbitronics due to for a single-loop polarizability tensor. (b) The Feynman dia-
interfacial SOC. The interfacial SOC in such a system is gram corresponding to Eq. (5), describing the interfacial DM
interaction for a system with inversion symmetry breaking
effectively described by Rashba SOC [49]. For magnetic
in the z direction and an order parameter vector n in the x
films, the two-dimensional (2D) Rashba model captures direction.
the main physics and trends of SOC with a broken in-
version symmetry [49], such as fieldlike and dampinglike
spin-orbit torques [1, 9, 10], intrinsic spin Hall effects [5], D. The contribution of the itinerant electrons to the
intrinsic anomalous Hall effects [50], inverse Faraday ef- exchange stiffness A is extracted from the second-order
fects [51], and magnetic anisotropy [52]. expansion in spatial gradients.
In this Letter, we develop a framework for computing TheR action S defines the
∗
system partition function
the interfacial DM interaction in AFM layers with inver- Z = d[Φ∗ ]d[Φ]d[n]e−S[Φ ,Φ,n]/~ , where Φ is the Grass-
sion asymmetry. The typical system illustrated in Fig. 1 mannian coherent-state spinor and ~ is Planck’s constant.
is described on the basis of an effective 2D AFM-Rashba In the s-d approach to magnetic systems, the action is de-
Hamiltonian. We find that both the sign and the magni- composed into the sum S = SF + SB , where SF [Φ∗ , Φ, n]
tude of the DM vector depend on the ratio relating three is the fermionic action corresponding to the Hamiltonian
energy scales: the chemical potential, the s-d exchange of Eq. (1), which also includes the s-d coupling, and
interaction, and the Rashba SOC strength. In particular, SB [n] is the bosonic action describing the dynamics of
the strong dependence on the chemical potential suggests the localized spins (magnons) in the absence of itinerant
that the DM interaction can be tuned by modifying the electrons. In our model, it is the coupling between the
electron density by means of doping or voltage gating itinerant electrons and the local moments that determine
[54]. the DM interaction, which is also directly linked to the
A generic effective 2D Hamiltonian describing itinerant SOC of the itinerant electrons. We will not specify the
electrons in the AFM layer [see Fig. 1(a)] can be written bosonic part of the action SB [n] since it is irrelevant for
as the subsequent discussion.
The fermionic action reads
H = Hkin + Hsd + Hso , (1) Z Z ~β ZZ
where Hkin is the kinetic energy of the electrons, Hsd = SF = dτ dτ 0 drdr 0 Φ∗r,τ [−~G−1
r,τ ;r 0 ,τ 0 ]Φr 0 ,τ 0 , (2)
0
Jsd n · Γ describes an effective interaction with a strength
Jsd between the spins of the itinerant s electrons and the where β = 1/kB T is the inverse temperature and τ is
localized d electrons [53], and HSO describes the SOC. the imaginary time. The inverse Green’s function op-
The operator Γ is the direct product of the electron spin erator is ~ G−1 0 0
r,τ ;r 0 ,τ 0 = − (~∂τ + H) δ(r − r )δ(τ − τ )
operator and the sublattice position operator, which, in in terms of the Hamiltonian of Eq. (1). We com-
the case of an AFM system, accounts for the effects of pute the effective theory for the vector field n(r)
sublattice staggering. The unit vector n is the order by integrating out the fermionic degrees of freedom.
parameter, which can represent either the total magne- This standard procedure results in an additional, ef-
tization in a ferromagnet or the staggered magnetization fective contribution
R ~β R to the bosonic action
of the form
in an antiferromagnet. ∆SFeff [n] = 0 dτ dr −~Tr[ln(−G−1 )] . Below, we
We compute the electronic contributions to the DM analyze ∆SFeff [n] and its influence on the magnet in the
interaction parameter D and to the exchange stiffness A AFM-Rashba model.
in the following way. First, we evaluate how the itinerant In our system, the symmetry-breaking direction (z di-
electrons influence the magnetic subsystem by finding an rection) is perpendicular to the plane [31]. Without
effective action. We expand the effective action up to the loss of generality, we choose the x axis to be in
linear order with respect to the deviation of the spins the direction of the in-plane vector field n. A small
from their equilibrium direction. The corresponding sus- deviation of the unit vector from its equilibrium direc-
ceptibility tensor describes the influence of the electronic tion is, then, parameterized by
n = x̂ + δn, where
degrees of freedom on the localized magnetic moments. δn = −(δn2y + δn2z )/2, δny , δnz . The effective action
A linear expansion of the susceptibility tensor in spa- is obtained from a perturbation with respect to δn,
tial gradients of n defines the DM interaction strength [53] that holds irrespective of the value of the exchange
 3

strength Jsd . The effective action is conveniently estab- the electron contribution to the exchange stiffness:
−1
lished from the Dyson equation, G−1 = G0 − Σ, where
the unperturbed Green’s function refers to δn = 0 and ∂ 2 Πyy
k ∂ 2 Πzz
k
A=− =− . (6)
the self-energy ~Σ = Jsd δn · Γ δ(r − r 0 )δ(τ − τ 0 ) is intro- ∂kx2 k=0 ∂ky2 k=0
duced.
In our model, A describes the contribution to the AFM
Computing the self-energy up to the second order in
exchange interaction from a superexchange-type interac-
δn yields [57]
tion between the localized spins in the AFM layer via the
X itinerant spins.
∆S eff [n] = ~β δnak,νl Πab b
k,iνl δn−k,−νl , (3)
We should emphasize here that in this approach, we
k6=0,l
have ignored the spin fluctuations of the localized AFM
where the indices a, b = (y, z) denote the transverse vec- spins, which is a valid omission as long as the system
tor components with respect to the equilibrium x̂ direc- temperature is much less than the critical Néel tempera-
tion, νl = 2nπ/β denotes the bosonic Matsubara frequen- ture.
cies, and To model an AFM system with interfacial SOC, we use
the 2D AFM-Rashba Hamiltonian [10, 11] on a square
2 X
Jsd lattice:
Πab
k,iνl = Tr[Γa G0q,iωn Γb G0q+k,iωn +iνl ] (4)
2~β q,n H = γk τx σ0 + Jsd τz σ · n − αR τx (σ × k) · ẑ, (7)

is the dynamical susceptibility tensor, pictured schemat- where σ and τ are the vectors of Pauli matrices repre-
ically in Fig. 2a. Here, G0q,iωn = (i~ωn − H)−1 is the senting the spin and AFM sublattice degree of freedom,
equilibrium Green’s function, and the ωn = (2n + 1)π/β respectively; σ0 is the identity matrix; n is the staggered
are the fermionic Matsubara frequencies. We compute order parameter (the normalized Néel vector); and αR is
the sum over thePfermionic Matsubara frequencies by us- the strength of the Rashba SOC. The kinetic energy of
ing the identity n (i~ωn − E)−1 /β = f (E), where f (E) the itinerant electrons is γk = a2 t(k 2 − k02 ), where t is
is the Fermi distribution. the nearest-neighbor hopping energy and k0 = 2/a, with
By expanding the static limit, νl = 0, on the spin a being the lattice constant.
susceptibility to the second order in the wave vector k, The band structure of the AFM-Rashba Hamiltonian
we find both the electronic contribution to the symmet- of Eq. (7) is, in general, anisotropic. It is convenient to
ric Heisenberg exchange stiffness, which is determined parameterize the four spectral branches as follows:
by the symmetric terms in the diagonal elements of the q
susceptibility tensor, and the antisymmetric exchange Eks,η = η γk2 + Jsd2 + α2 k 2 + 2sα kξ ,
R R k (8)
interaction (DM interaction), which is determined by
the antisymmetric terms in the off-diagonal elements. where s, η = ±1 are the spin chirality and elec-
From the partition function of the canonical ensem- tron/hole band indices, respectively. We also introduce
eff
ble, Z = d[n]e−S [n]/~ = d[n]e−βF [n] , we obtain
R R
1/2
ξk = γk2 + Jsd 2
cos2 φ with the in-plane wave vec-
δn S eff /~ = βδn F [n], where the micromagnetic free en- tor k parameterized by the angle φ, such that k =
ergy, including the stiffness and the DM interaction,

is k(cos φ, sin φ, 0).
F [n] = d2 r A(∇n)2 − Dn · (ẑ × ∇) × n . By com-
R
Let us now analyze the expressions of Eqs. (5) and (6)
paring the microscopic free energy with the expression for the AFM-Rashba Hamiltonian of Eq. (7) in the limit
for the effective action, Eq. (3), we define the micromag- of weak spin-orbit interaction, {k0 αR , Jsd }  min{t, F },
netic parameters A and D, which characterize the free and at zero temperature. Using the relation Γa = τz σa ,
carrier contributions to the exchange stiffness and the we obtain the relation
DM interaction, respectively.  2
Upon expanding the off-diagonal elements of the tensor Ak0
D=− αR , (9)
Π to the first order in the wave vector k [see Fig. 2(b)], 4t
we obtain the relation
where the stiffness parameter is
∂Πyz ∂Πzy
D=i k =−i k
(
(5a) 2
tJsd 2 − 2F /16t2 , F < 4t,
∂ky k=0 ∂ky k=0 A= (10)
2
2πF 1,
2 X F > 4t,
J
=i sd Tr[Γy G0q,iωn Γz G0q,iωn vy G0q,iωn ], (5b)
2β q,n which is manifestly independent of the SOC strength in
the limit of weak spin-orbit interaction. Here, the Fermi
where vy = ~−1 ∂Hq /∂qy is the y component of the ve- energy F is measured with respect to the center of the
locity operator. From the second-order terms, we obtain energy gap [see Figs. 1(b) and (c)]. In this regime, the
 4

●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●
●
●
Hamiltonian of the Rashba SOC symmetry, the DM in-
10 ● teraction vanishes inside the gap but remains finite out-
■■
●
■■ ■
side the gap, even at zero temperature[59, 60]. Thus,
■■
■■■■
■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
0 ■
■■■■■■■
■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■●
●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●
■●
■■■■■■■■■■■■■■■■■■■■■■■■■■■■■● ■●
■● ■●
■● ■●
■●■ we conclude that the DM interaction exhibits a quali-
● ■ ■■■ ●●●●●●●●●●●●●●
D (pJ/m)

●●●●●●●●
■ ●●●●●
■■■
●●
●●●●●●● tatively different behavior in antiferromagnets compared
●●
●
●● with that in ferromagnets.
10 ●
●
●

● In Fig. 3, we illustrate the behavior of the DM interac-

●
tion on the basis of a numerical analysis of Eq. (5b) be-
● ● ● Jsd 1.0 eV, k0αR 0.1 eV
20 ● yond the weak SOC regime of Eqs. (9-11), where typical
●● ■ Jsd 0.1 eV, k0αR 1.0 eV
● material parameters are assumed: t = 3 eV, F = 2 eV,
0.0 0.5 1.0 1.5 2.0 2.5 3.0 Jsd = 1 eV, a = 0.4 nm and k0 αR = 0.1 eV [10]. The
ϵF (eV) results indicate a rather large interfacial DM interaction,
60 with D ' −1.6 pJ/m. Moreover, when the Fermi energy
●●●●●●●●●●●●●●●●●●●●●●●●●
●●●●●●
●
●●
●●●● ●●●●
●●●
●●
lies within the gap, i. e., |F | < Jsd , we find an order-
40 ●●
●●
● ●●●
●●●● of-magnitude enhancement of the DM interaction with
20
●
the opposite sign, D ' 12.7 pJ/m. Importantly, the esti-
mated strength of the DM interaction is at least an order
D (pJ/m)

●
■■■■■■■■■■■■■■■■■■■■
0 ■●
●■●
■●
■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
●●●●●
●●●●
■■■■■■
■ ■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
of magnitude larger than that for a Rashba-FM layer
●●●●
●●●●
●●●

20
●●●
●●●
●●
●●
● with the same parameters [58].
●●
●●
●
For small Fermi energies, the DM interaction is pos-
●
●
40 ●
●
●●
● sd J 1.0 eV itive and almost independent of the Fermi energy (see
●●
●●
●●
●●●
●●●
●
■ J sd 0.1 eV the top panel in Fig. 3), as might be expected from
60 ●●●● ●●
●●●●
Eq. (11). In the metallic regime (i. e., for Fermi ener-
0.0 0.5 1.0 1.5 2.0 2.5 3.0 gies well above the gap), the DM interaction is nega-
k0 αR (eV) tive, and its strength decreases in proportion to −2 F , in
good agreement with Eqs. (9) and (10). The sign inver-
FIG. 3: Numerical results for the DM interaction as a function
sion of the
p DM interaction is rather sharp and occurs
of the Fermi energy (top) and of the SOC strength (bottom) 2 + α2 k 2 . The bottom panel of Fig. 3 also
for different values of the s-d exchange interaction. We set at F = Jsd R 0
F = 2 eV and t = 3 eV in all cases. confirms that the strength of the DM interaction is lin-
early proportional to αR in the weak SOC regime, in
agreement with Eq. (9). Thus, we conclude that the DM
energy gaps at k = 0 and k = ±k0 are equal to 8t and interaction in an AFM material may vary by orders of
2Jsd , respectively. magnitude depending on the material parameters, as il-
For the case in which the Fermi energy lies within the lustrated in Fig. 3.
band gap, i. e., for |F | < Jsd , and in the limit of k0 αR  Our calculations also show that, unlike in FM-Rashba
Jsd  t, we obtain systems [58], the temperature dependence of the DM in-
 2
k0 teraction in the AFM-Rashba model of Eq. (7) is weak
D= αR , (11) due to a large contribution to the DM interaction from
2π
the valence bands (not shown). As we have already dis-
which is of the opposite sign compared with the metallic cussed, this is correct if the system temperature is much
regime of Eqs. (9) and (10). less than the Néel temperature, i. e., in a regime in which
It is instructive to compare the results of Eqs. (9)- the spin fluctuations of the AFM layer are suppressed.
(11) with those for a Rashba-FM system [58]. If both Controlling the DM interaction is essential for engi-
FM chiral bands are occupied (F > Jsd ), the interfa- neering chiral magnetic structures. If the DM interaction
cial DM interaction in the Rashba-FM model vanishes at parameter exceeds a certain critical value, which is de-
zero temperature due to the exact cancellation between termined by the Heisenberg exchange interaction and the
the Fermi surface and Fermi sea contributions [58]. How- uniaxial anisotropy, then the ground state changes from
ever, such a cancellation is absent in the AFM-Rashba a collinear configuration to either a helimagnetic state or
model, giving rise to a finite result for Eq. (9), by virtue a Skyrmion lattice. A weaker DM interaction enables the
of an additional contribution from the valence bands. For stabilization of isolated Skyrmions in a metastable state
F < Jsd , i. e., when only the lowest FM chiral band is [16, 18]. In chiral magnets, the sign of the DM inter-
occupied, the DM interaction in the weak SOC limit and action determines the direction, or handedness, of spin
at zero temperature is finite and linearly proportional to rotation.
A [26, 55, 56, 58]. The asymmetry of the spin-wave dispersion in spin-
On the other hand, in the Dirac model of an polarized electron energy-loss spectroscopy and Brillouin
FM/topological-insulator bilayer characterized by the light scattering is a measure of the DM interaction in FM
 5

systems [27, 30, 42]. Although the interfacial DM interac- [7] T. Zhang, P. Cheng, X. Chen, J-F. Jia, X. Ma, K. He,
tions in a few FM/HM bilayers have been experimentally L. Wang, H. Zhang, X. Dai, Z. Fang, X. Xie, and Q-K.
studied in recent years [29–42], we are not aware of sim- Xue, Phys. Rev. B 79, 094422 (2009).
ilar measurements in AFM heterostructures. However, [8] A. Manchon and S. Zhang, Phys. Rev. B 79, 094422
(2009).
the first observation of a bulk DM interaction in an AFM [9] A. Qaiumzadeh, R. A. Duine, and M. Titov, Phys. Rev.
system, namely, the noncentrosymmetric α − Cu2 V2 O7 , B 92, 014402 (2015).
has recently been reported in Ref. [43] based on inelastic [10] J. Železný, H. Gao, A. Manchon, F. Freimuth, Y.
neutron scattering. Very recently, a large DM interac- Mokrousov, J. Zemen, J. Mašek, J. Sinova, and T. Jung-
tion in Fe/Ir bilayers on Rh(001) has been predicted on wirth Phys. Rev. B 95, 014403 (2017); J. Železný, H.
the basis of ab initio calculations [61]. We hope that our Gao, K. Výborný, J. Zemen, J. Mašek, A. Manchon,
work will stimulate new experiments and ab initio works J. Wunderlich, J. Sinova, and T. Jungwirth, Phys. Rev.
Lett. 113, 157201 (2014).
on such AFM heterostructures. [11] H. B. M. Saidaoui and A. Manchon, arXiv:1606.04261.
In summary, we have computed the DM interaction, [12] S. Mühlbauer, B. Binz, F. Jonietz, C. Pfleiderer, A.
both analytically and numerically, using an effective Rosch, A. Neubauer, R. Georgii, and P. Böni, Science
model for AFM layers with interfacial Rashba SOC. In 323, 915 (2009).
the AFM-Rashba model, the induced interfacial DM in- [13] W. Jiang, G. Chen, K. Liu, J. Zang, S. G. E. te Velthuis,
teraction appears as a Lifshitz-type invariant term. Our and A. Hoffmann, Phys. Rep. 704, 1 (2017).
[14] A. Thiaville, S. Rohart, É. Jué, V. Cros, and A. Fert,
results show that both the sign and the strength of the
Europhys. Lett. 100, 57002 (2012).
DM interaction may be tuned by modifying the electron [15] S. Emori, U. Bauer, S-M. Ahn, E. Martinez, and G. S.
density, e. g., by applying a gate voltage or through dop- D. Beach, Nat. Mater. 12, 611 (2013).
ing. This tunability implies that a rich variety of chi- [16] V. Flovik, A. Qaiumzadeh, A. K. Nandy, C. Heo, and T.
ral magnetic structures can emerge in layered AFM/HM Rasing, Phys. Rev. B, 96, 140411(R) (2017).
systems with different interfacial charge densities, s-d ex- [17] A. Qaiumzadeh, L. A. Kristiansen, and A. Brataas, Phys.
change interactions, and SOC interactions. Rev. B 97, 020402(R) (2018).
[18] D. Yudin, D. R. Gulevich, and M. Titov, Phys. Rev. Lett.
119, 147202 (2017).
[19] I. Dzyaloshinsky, J. Phys. Chem. Solids 4, 241 (1958); J.
Acknowledgments Exp. Theor. Phys. 32, 1547 (1957).
[20] T. Moriya, Phys. Rev. 120, 91 (1960); Phys. Rev. Lett.
4, 228 (1960).
The research leading to these results was supported [21] M. Heide, G. Bihlmayer, and S. Blügel, Phys. Rev. B 78,
by the European Research Council via Advanced Grant 140403(R) (2008).
No. 669442, “Insulatronics,” and by the Research Coun- [22] A. Fert, Mater. Sci. Forum 59-60, 439 (1990); A. Fert
cil of Norway through its Centres of Excellence fund- and P. M. Levy, Phys. Rev. Lett. 44, 1538 (1980).
ing scheme, Project No. 262633, “QuSpin.” We also [23] A. N. Bogdanov and D. A. Yablonskii, Sov. Phys. JETP
68, 101 (1989); 69, 142 (1989).
acknowledge the support received from the Dutch Sci-
[24] A. Crépieux and C. Lacroix, J. Magn. Magn. Mater. 182,
ence Foundation, NWO/FOM 13PR3118; the European 341 (1998).
Commission; and the Russian Science Foundation under [25] A. N. Bogdanov and U. K. Rößler, Phys. Rev. Lett. 87,
Project No. 17-12-01359. R. A. D. is part of the D- 037203 (2001).
ITP consortium, a program of the Netherlands Organi- [26] J-H. Moon, S-M. Seo, K-J. Lee, K-W. Kim, J. Ryu, H-
sation for Scientific Research (NWO) that is funded by W. Lee, R. D. McMichael, and M. D. Stiles, Phys. Rev.
the Dutch Ministry of Education, Culture and Science B 88, 184404 (2013); K-W. Kim, H-W. Lee, K-J. Lee,
and M. D. Stiles, Phys. Rev. Lett. 111, 216601 (2013).
(OCW). M.T. acknowledges support from an ITMO vis-
[27] Kh. Zakeri, Y. Zhang, J. Prokop, T.-H. Chuang, N. Sakr,
iting professor fellowship. W. X. Tang, and J. Kirschner, Phys. Rev. Lett. 104,
137203 (2010); Kh. Zakeri, Y. Zhang, T.-H. Chuang, and
J. Kirschner, ibid 108, 197205 (2012).
[28] V. E. Dmitrienko, E. N. Ovchinnikova, S. P. Collins, G.
Nisbet, G. Beutier, Y. O. Kvashnin, V. V. Mazurenko,
[1] A. Manchon, H. C. Koo, J. Nitta, S. M. Frolov, and R. A. I. Lichtenstein, and M. I. Katsnelson, Nat. Phys. 10,
A. Duine, Nat. Mater. 14, 871 (2015). 202 (2014).
[2] A. Soumyanarayanan, N. Reyren, A. Fert, and C. [29] A. K. Chaurasiya, C. Banerjee, S. Pan, S. Sahoo, S.
Panagopoulos, Nature (London) 539, 509 (2015). Choudhury, J. Sinha, and A. Barman, Sci. Rep. 6, 32592
[3] W. Thomson, Proc. R. Soc. Lond. 8, 546 (1857). (2016).
[4] E. Hall, Philos. Mag. 12, 157 (1881). [30] J. Cho, N-H. Kim, S. Lee, J-S. Kim, R. Lavrijsen, A.
[5] J. Sinova, S. O. Valenzuela, J. Wunderlich, C.H. Back, Solignac, Y. Yin, D-S. Han, N. J. J. van Hoof, H. J. M.
and T. Jungwirth, Rev. Mod. Phys. 87, 1213 (2015). Swagten, B. Koopmans, and C-Y. You, Nat. Commun.
[6] P. Roushan, J. Seo, C. V. Parker, Y. S. Hor, D. Hsieh, 6, 7635 (2015).
D. Qian, A. Richardella, M. Z. Hasan, R. J. Cava, and [31] M. Belmeguenai, J-P. Adam, Y. Roussigné, S. Eimer, T.
A. Yazdani, Nature (London) 460, 1106 (2009).
 6

Devolder, J-V. Kim, S. M. Cherif, A. Stashkevich, and E. Fullerton, C. Leighton, A. MacDonald, D. Ralph, D.
A. Thiaville, Phys. Rev. B 91, 180405(R) (2015). Arena, H. Durr, P. Fischer, J. Grollier, J. Heremans,
[32] H. S. Körner, J. Stigloher, H. G. Bauer, H. Hata, T. T. Jungwirth, A. Kimmel, B. Koopmans, I. Krivorotov,
Taniguchi, T. Moriyama, T. Ono, and C. H. Back, Phys. S. May, A. Petford-Long, J. Rondinelli, N. Samarth, I.
Rev. B 92, 220413(R) (2015). Schuller, A. Slavin, M. Stiles, O. Tchernyshyov, A. Thi-
[33] K. Di, V. L. Zhang, H. S. Lim, S. C. Ng, M. H. Kuok, J. aville, and B. Zink, Rev. Mod. Phys. 89, 25006 (2017).
Yu, J. Yoon, X. Qiu, and H. Yang, Phys. Rev. Lett. 114, [49] S. Grytsyuk, A. Belabbes, P. M. Haney, H-W. Lee, K-J.
047201 (2015). Lee, M. D. Stiles, U. Schwingenschlögl, and A. Manchon,
[34] H. Yang, A. Thiaville, S. Rohart, A. Fert, and M. Phys. Rev. B 93, 174421 (2016).
Chshiev, Phys. Rev. Lett. 115, 267210 (2015). [50] N. Nagaosa, J. Sinova, S. Onoda, and A. H. MacDonald,
[35] H. T. Nembach, J. M. Shaw, M. Weiler, E. Jué, and T. N. P. Ong, Rev. Mod. Phys. 82, 1539 (2010).
J. Silva, Nat. Phys. 11, 825 (2015). [51] A. Qaiumzadeh and M. Titov, Phys. Rev. B 94, 014425
[36] A. Hrabec, M. Belmeguenai, A. Stashkevich, S. M. (2016).
Chérif, S. Rohart, Y. Roussigné, and A. Thiaville, Appl. [52] K-W. Kim, K-J. Lee, H-W. Lee, and M. D. Stiles, Phys.
Phys. Lett. 110, 242402 (2017). Rev. B 94, 184402 (2016).
[37] R. M. Rowan-Robinson, A. A. Stashkevich, Y. Roussigne, [53] J. Schliemann, Phys. Rev. B 67, 045202 (2003); A. Auer-
M. Belmeguenai, S-M. Cherif, A. Thiaville, T. P. A. Hase, bach, Interacting Electrons and Quantum Magnetism
A. T. Hindmarch, and D. Atkinson, Sci. Rep. 7, 16835 (Springer, New York, 1998).
(2017). [54] A. D. Caviglia, M. Gabay, S. Gariglio, N. Reyren, C.
[38] X. Ma, G. Yu, X. Li, T. Wang, D. Wu, K. S. Olsson, Z. Cancellieri, and J.-M. Triscone, Phys. Rev. Lett. 104,
Chu, K. An, J. Q. Xiao, K. L. Wang, and X. Li, Phys. 126803 (2010).
Rev. B 94, 180408(R) (2016). [55] F. Freimuth, S. Blügel, and Y. Mokrousov, Phys. Rev. B
[39] H. Yang, G. Chen, A. A. C. Cotta, A. T. N’Diaye, S. A. 96, 054403 (2017); J. Phys. Condens. Matter. 26, 104202
Nikolaev, E. A. Soares, W. A. A. Macedo, A. K. Schmid, (2014).
A. Fert, and M. Chshiev, arXiv:1704.09023. [56] T. Kikuchi, T. Koretsune, R. Arita, and G. Tatara, Phys.
[40] M. Baćani, M. A. Marioni, J. Schwenk, and H. J. Hug, Rev. Lett. 116, 247201 (2016).
arXiv:1609.01615. [57] In general, the effective action has two extra terms
[41] X. Ma, G. Yu, S. A. Razavi, S. S. Sasaki, X. Li, K. Hao, related to the spin susceptibilities with a long wave-
S. H. Tolbert, K. L. Wang, and X. Li, Phys. Rev. Lett. length nature, i. e., k = 0, which have no contri-
119, 027202 (2017). butions in the exchange P stiffness and DM
interaction:
[42] M. Belmeguenai, H. Bouloussa, Y. Roussign, M. S. Ga- ∆S0eff [n] = ~β δna χa + l δnaνl Πab b
iνl δn−νl . The static
bor, T. Petrisor, jr., C. Tiusan, H. Yang, A. Stashkevich, related to the first-order term in δna
spin susceptibility P
and S. M. Chérif, Phys. Rev. B 96, 144402 (2017). is χa = (Jsd /~β) q,n Tr[Γa G0q,iωn ], and the dynamic
[43] G. Gitgeatpong, Y. Zhao, P. Piyawongwatthana, Y. Qiu, spin susceptibilityPrelated
 to the second-order term in
L. W. Harriger, N. P. Butch, T. J. Sato, and K. Matan, δna is Πab iνl = q,n − (Jsd /2~β)Tr[Γx G0q,iωn ]δab +
Phys. Rev. Lett. 119, 047201 (2017). 2
(Jsd /2~β)Tr[Γa G0q,iωn Γb G0q,iωn +iνl ] , where δab is the
[44] L. Udvardi and L. Szunyogh, Phys. Rev. Lett. 102, Kronecker delta function.
207204 (2009). [58] I. A. Ado, A. Qaiumzadeh, R. A. Duine, A. Brataas, and
[45] J. Gayles, F. Freimuth, T. Schena, G. Lani, P. Mavropou- M. Titov, arXiv:1804.03739.
los, R.A. Duine, S. Blügel, J. Sinova, and Y. Mokrousov, [59] R. Wakatsuki, M. Ezawa, and N. Nagaosa, Sci. Rep. 5,
Phys. Rev. Lett. 115, 036602 (2015). 13638 (2015); T. Koretsune, N. Nagaosa, and R. Arita,
[46] T. Jungwirth, J. Sinova, A. Manchon, X. Marti, J. Wun- ibid 5, 13302 (2015).
derlich, and C. Felser, arXiv:1705.10489; O. Gomonay, [60] Y. Tserkovnyak, D. A. Pesin, and D. Loss, Phys. Rev. B
T. Jungwirth, and J. Sinova, Phys. Status Solidi RRL 91, 041121(R) (2015).
11, 1700022 (2017); T. Jungwirth, X. Marti, P. Wadley, [61] S. Meyer, B. Dupé, P. Ferriani, and S. Heinze, Phys.
and J. Wunderlich, Nat. Nanotechnol. 11, 231 (2016). Rev. B 96, 094408 (2017); A. Belabbes, G. Bihlmayer,
[47] V. Baltz, A. Manchon, M. Tsoi, T. Moriyama, T. Ono, F. Bechstedt, S. Blügel, and A. Manchon, Phys. Rev.
and Y. Tserkovnyak, Rev. Mod. Phys. 90, 015005 (2018). Lett. 117, 247202 (2016).
[48] F. Hellman, A. Hoffmann, Y. Tserkovnyak, G. Beach,