Octupole-Driven Spin-Transfer Torque (OTT)
- Octupole-driven spin-transfer torque (OTT) is a current-induced torque mediated by non-dipolar octupolar order in antiferromagnets.
- OTT arises from sublattice-resolved transport in AFMTJs, where asymmetrical intra- and inter-sublattice tunneling yields a net staggered torque.
- Experimental PtMn₃-based AFMTJs demonstrate room-temperature, field-free switching with record-high TMR, highlighting OTT’s potential in scalable spintronics.
Searching arXiv for related work on octupole torque, AFMTJs, and altermagnets to ground the article with recent literature. arxiv_search(query="all-antiferromagnetic tunnel junction octupole PtMn3 tunneling magnetoresistance", max_results=10, sort_by="relevance") Octupole-driven spin-transfer torque (OTT) denotes a current-induced torque on antiferromagnetic order that is mediated by octupolar, rather than net-magnetization, structure in the electronic states. In the recent literature, the term is introduced for all-antiferromagnetic tunnel junctions (AFMTJs), where the reciprocal counterpart of tunneling magnetoresistance (TMR) survives despite zero net magnetization and a spin-neutral total current because transport is organized by sublattice-resolved spin currents and cluster magnetic octupole moments (Kang et al., 3 Sep 2025). Closely related work in d-wave altermagnets shows that injected magnetic octupoles can also generate torque through the magnetic octupole Hall effect (MOHE), extending established spin-Hall-torque phenomenology into a higher-multipole setting (Han et al., 2024). Taken together, these results place octupolar transport and octupolar torque among the central mechanisms now considered in antiferromagnetic spintronics.
1. Definition and conceptual scope
In conventional magnetic tunnel junctions with ferromagnetic electrodes, spin-transfer torque (STT) provides current-driven control of magnetization, while TMR provides electrical readout. For antiferromagnets, both effects were conventionally believed to vanish because of zero net magnetization. The AFMTJ work on PtMn||_3_3_3_3120^\circ(111)_3|$0, describe a spatially anisotropic distribution of spin density and encode the non-relativistic band spin-splitting unique to altermagnets. Landau theory identifies the relevant coupling as $|$1, where $|$2 is the Néel vector (Han et al., 2024).
These two settings should not be conflated. PtMn$|$3-based OTT concerns a chiral or non-collinear antiferromagnet in a tunnel-junction geometry, whereas the altermagnetic work concerns d-wave altermagnets driven by octupole Hall injection from an adjacent heavy metal. The common element is that the operative order parameter is octupolar rather than dipolar.
3. Microscopic origin of OTT in all-antiferromagnetic tunnel junctions
The microscopic mechanism proposed for OTT in AFMTJs is a sublattice-resolved tunneling asymmetry. When a vertical current passes through the junction, tunneling processes are separated into intra-sublattice and inter-sublattice channels. Intra-sublattice tunneling transfers electrons from sublattice $|$4 in the fixed layer into the same sublattice $|$5 in the free layer, whereas inter-sublattice tunneling connects different sublattices $|$6 (Kang et al., 3 Sep 2025).
First-principles DFT calculations indicate that the Bloch states at the Fermi level are anisotropic and sublattice-dependent, so electrons preferentially retain both their sublattice and spin character during tunneling. The torque efficiencies are then parameterized by $|$7 for intra-sublattice processes and $|$8 for inter-sublattice processes. If $|$9, the torques cancel by symmetry; a net torque appears only when $|$0.
The resulting torque is staggered across the three sublattices but is organized by the same $|$1 rotational symmetry as the underlying CMO. The consequence is collective switching of the cluster magnetic octupole moment without generating net magnetization. The paper therefore interprets OTT equivalently in two ways: as an imbalance between intra- and inter-sublattice spin currents across the AFMTJ, and as a torque on the non-zero net cluster octupole polarization of each PtMn$|$2 layer.
A recurrent misconception is that a spin-neutral total current excludes any reciprocal current-induced torque in an AFMTJ. The PtMn$|$3 result directly opposes that expectation: the total current can remain spin-neutral while the sublattice-resolved transport still produces a non-vanishing staggered torque because the relevant conservation and cancellation arguments operate at the sublattice level, not only at the net-current level.
4. Torque equations and dynamical description
For PtMn$|$4, the free energy in the monodomain limit for three sublattices is written as
$|$5
For the in-plane octupole modes, the effective free energy density is
$|$6
with
$|$7
Within this model, the net sublattice torque takes the form
$|$8
which has the same mathematical form as Slonczewski’s STT, but acts staggeredly on the three sublattices in accordance with the CMO. Micromagnetic simulation of three-sublattice coupled Landau-Lifshitz-Gilbert equations confirms deterministic octupole switching: $|$9 These relations formalize OTT as a current-driven dynamics of an octupolar order parameter rather than a conventional magnetization vector (Kang et al., 3 Sep 2025).
A closely related formalism appears in d-wave altermagnets, where injected nonequilibrium magnetic octupole density $_3$0 produces a field-like torque
$_3$1
The full magnetic-octupole torque is written as
$_3$2
and, for arbitrary Néel-vector angle $_3$3, the total torkance is
$_3$4
An important consequence is that the octupolar torque can be nonzero in symmetry configurations where spin Hall spin-orbit torque vanishes (Han et al., 2024).
5. Experimental realization in PtMn$_3$5$_3$6MgO$_3$7PtMn$_3$8 junctions
The reported experimental platform is the full stack Pt(5 nm)/PtMn$_3$9(7 nm)/MgO(2 nm)/PtMn$_3$0(10 nm)/Pt(5 nm)/Ru(10 nm) on oxidized Si, patterned into nanoscale pillars with diameters of 50–200 nm. The experiments are performed at room temperature and without external magnetic field (Kang et al., 3 Sep 2025).
Bidirectional, repeatable, field-free current-pulse switching is observed at a current density $_3$1 MA/cm$_3$2. The resistance changes sharply during switching, and the TMR reaches a record-high value of $_3$3 at room temperature. The switching remains robust up to $_3$4 K, corresponding to the magnetic phase transition of PtMn$_3$5. Pulse-width-dependent measurements yield an energy barrier of $_3$6 at room temperature. NV magnetometry finds domain sizes of 100–170 nm, suggesting abrupt switching in devices smaller than or comparable to the domain size.
| Quantity | Ferromagnetic MTJ | All-antiferromagnetic MTJ (PtMn$_3$7-based) |
|---|---|---|
| Reading | TMR | TMR via octupole polarization |
| Writing | STT | OTT via cluster octupole moment |
| TMR at room temperature | up to $_3$8 (best) | $_3$9 |
| Switching current | Few MA/cm$_3$0 | $_3$1 MA/cm$_3$2 |
| Magnetic order | Net $_3$3 | Net $_3$4; order in CMO |
The experimental significance is twofold. First, the same AFMTJ architecture supports both electrical readout and electrical writing entirely within the antiferromagnetic regime. Second, the data show that the absence of net magnetization does not preclude practical tunnel-junction functionality when the relevant order parameter is octupolar.
6. Relation to magnetic octupole Hall torques in d-wave altermagnets
The altermagnetic theory provides a parallel route to octupolar torque that does not rely on tunneling through an AFMTJ. In that setting, the nonequilibrium octupole density is injected from an adjacent heavy metal by the magnetic octupole Hall effect. The MO current density is written as
$_3$5
with MO current operator $_3$6, and the magnetic octupole Hall conductivity is evaluated using a Kubo formula (Han et al., 2024).
For fcc Pt, the calculated values at the Fermi level are
$_3$7
The same work states that the intrinsic spin Hall conductivity of Pt is about
$_3$8
so the MO Hall conductivity is comparable in magnitude to the spin Hall conductivity.
This comparison is important for the interpretation of OTT and related octupolar torques. It indicates that higher-order multipole transport is not necessarily a weak correction to dipolar spin transport. In the altermagnetic geometry, the heavy metal acts as an MO current source and the d-wave altermagnet as the torque detector, directly generalizing the conventional spin Hall phenomenology of heavy-metal spin generation and ferromagnetic detection. The work therefore frames electrical control of altermagnetic configurations as a realistic consequence of octupole transport, not merely a symmetry-allowed abstraction.
7. Significance, misconceptions, and emerging directions
The immediate device implication of OTT is the prospect of all-antiferromagnetic memory elements that combine TMR-based readout with current-induced writing. The PtMn$_3$9 study identifies insensitivity to external fields, scalability to nanoscale dimensions, elimination of dipolar crosstalk between bits, potential operation in the THz regime due to antiferromagnetic dynamics, and room-temperature functionality as advantages of AFMTJs using OTT. It further presents the system as a pathway towards deeply scaled magnetic memory and room-temperature terahertz technologies (Kang et al., 3 Sep 2025).
A second implication emerges from the altermagnetic MOHE analysis. Because octupolar torque can remain finite in symmetry configurations where spin Hall torque vanishes, octupolar control enlarges the set of electrically addressable antiferromagnetic states. That work also suggests that altermagnets may form a basis for “multipoletronics,” namely devices using higher-order multipole transport rather than only spin currents, and identifies a broader multipoletronics/orbitronics platform (Han et al., 2024).
Two misconceptions are now difficult to sustain. The first is that tunnel-junction spin-transfer phenomena necessarily require net magnetization. The second is that higher-order multipoles are only descriptive order parameters with little transport relevance. The available results instead show that octupole order can organize both transport coefficients and current-induced torques at technologically relevant magnitudes. What remains open is the comparative mapping between tunnel-mediated OTT in AFMTJs and Hall-mediated octupole torque in altermagnets: the current literature establishes both mechanisms, but their unification is, at present, best regarded as a plausible implication rather than a completed theory.