Bulk Altermagnetic Domains Overview

• Bulk altermagnetic domains are defined as distinct real-space regions in compensated magnets where the Néel vector selects symmetry-related states that control observable signals.
• Polarized neutron diffraction and magneto-optical imaging reveal that field-induced domain imbalances produce measurable nuclear–magnetic interference signals.
• Material studies indicate that probe geometry, strain, and local defects finely tune domain properties in systems like MnTe, α-Fe₂O₃, and CrSb.

Bulk altermagnetic domains are real-space regions of a compensated collinear magnet in which the altermagnetic order parameter, conventionally the Néel vector, selects one member of a set of symmetry-related states that remain distinct under time reversal. Their defining importance is not merely classificatory. In altermagnets, time-reversal-related antiphase domains can compensate spin-polarized transport, anomalous Hall-like signals, and magneto-optical responses when their populations are balanced, even though the underlying bulk crystal remains magnetically ordered. MnTe has become the benchmark system because bulk-sensitive polarized neutron diffraction, transmission XMCD spectromicroscopy, and magneto-optical Kerr imaging now converge on the existence of switchable bulk time-reversal-symmetry-breaking domains, while work on $\alpha$-Fe$_2$O$_3$, CrSb, and RuO$_2$ shows that the observability of such domains depends strongly on probe geometry, Néel-vector orientation, strain, and the stability of the magnetic ground state itself (Liu et al., 20 May 2026, Watanabe et al., 16 Apr 2026, Yamamoto et al., 25 Feb 2025, Yamamoto et al., 10 Mar 2026, Lee et al., 12 Feb 2026).

1. Symmetry definition and domain classes

In MnTe, the relevant order parameter is the Néel vector

$\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$

with sublattice sum

$\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$

Below $T_N$, MnTe orders in an $A$-type antiferromagnetic state with spins along the in-plane $\langle 1\bar{1}0\rangle$ directions. The bulk crystal then admits two distinct kinds of magnetic domains: three orientational domains related by the crystal $C_3$ symmetry, and, for each orientational state, a pair of antiphase domains related by time reversal $_2$0. The total domain manifold therefore contains six states. The antiphase pair carries opposite Néel vectors, denoted schematically as $_2$1 and $_2$2, and a balanced distribution of these time-reversal partners compensates altermagnetic signals at the macroscopic level (Liu et al., 20 May 2026).

The crucial distinction from conventional collinear antiferromagnets is that, in the altermagnetic case realized in MnTe, the two time-reversed states are not restored by translation or inversion. A balanced antiphase-domain population therefore yields zero net sample-averaged Néel-vector signal in probes sensitive to time-reversal breaking, whereas any imbalance produces a nonzero bulk response. This makes bulk altermagnetic domains the mechanism that decides whether altermagnetism is macroscopically visible, not merely a symmetry label for equivalent antiferromagnetic variants (Liu et al., 20 May 2026).

A broader terminological caution comes from thin-film Mn$_2$3Si$_2$4, where the authors distinguish “altermagnetic variants” from magnetic domains. There, three symmetry-related checkerboard realizations of the altermagnetic state are called variants, and each variant can still contain two magnetic domains of opposite Néel-vector direction. This distinction suggests that, in altermagnets generally, “variant” and “domain” need not coincide: a single crystallographic realization of the altermagnetic order can still support a time-reversed domain pair (Rial et al., 2024).

2. Reciprocal-space identification in bulk MnTe

The decisive bulk-domain probe in MnTe is polarized neutron diffraction in half-polarized mode. The experiment defines nuclear and magnetic structure factors $_2$5 and $_2$6, and the nuclear–magnetic interference vector

$_2$7

For MnTe’s collinear structure this reduces to

$_2$8

so $_2$9 is parallel to the Néel-vector direction. The measured interference component is therefore a direct manifestation of the net unit Néel vector in the sample: if antiphase partners are balanced, the interference vanishes; if one partner is favored, a nonzero interference signal appears. This is the central reciprocal-space criterion by which bulk antiphase-domain imbalance was established in MnTe (Liu et al., 20 May 2026).

Experimentally, the up/down incident-polarization intensity difference in half-polarized geometry isolates $_3$0 or $_3$1, whereas $_3$2 cannot be measured because neutrons do not probe the magnetic moment component parallel to $_3$3. Measurements were carried out in the $_3$4 scattering plane on $_3$5, $_3$6, $_3$7, and $_3$8, with $_3$9 and $_2$0 used as null checks. The interference channels are zero at $_2$1 and $_2$2 and finite at the mixed nuclear/magnetic reflections, with alternating sign matching the calculated structure factors. That pattern identifies genuine nuclear–magnetic interference rather than ordinary magnetic intensity (Liu et al., 20 May 2026).

The domain sensitivity appears most clearly in the comparison of zero-field cooling, $_2$3-axis field cooling, and oblique-field cooling. After zero-field cooling, $_2$4 and $_2$5 are zero within resolution, consistent with no net antiphase imbalance. After field cooling along $_2$6, $_2$7 becomes large and reverses sign when the cooling field is reversed, providing direct evidence that the switched object is the time-reversal-related antiphase state. After oblique-field cooling in the $_2$8-plane, $_2$9 becomes large while $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$0 is small, showing that field geometry reweights a different combination of orientational and antiphase populations. The temperature dependence of $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$1 at $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$2 under $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$3 mT field cooling terminates at $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$4, confirming the magnetic origin of the interference signal (Liu et al., 20 May 2026).

The same work also used the pure magnetic reflection $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$5 in full-polarized mode to show that the orientational-domain distribution is not perfectly balanced even before field selection. The observed negative $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$6 indicates a natural bulk orientational imbalance attributed likely to defects and local strain. Bulk MnTe is therefore not an ideal symmetric multidomain ensemble: orientational populations are already biased, while field cooling can additionally select time-reversal-related antiphase partners (Liu et al., 20 May 2026).

3. Real-space hierarchy from macroscopic to atomic scales

Real-space imaging in MnTe now spans a wide length-scale hierarchy. Scanning polar magneto-optical Kerr microscopy on bulk single crystals revealed large contiguous regions of positive and negative Kerr rotation, interpreted as two classes of time-reversal-symmetry-breaking domains. Some of these domains are macroscopic, approaching $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$7 mm in size. The contrast weakens continuously on warming and disappears above $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$8 K, while field cooling in approximately $\bm n=\bm S_{\mathrm A}-\bm S_{\mathrm B},$9 T along $\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$0 produces opposite majority-domain signs. Fitted line profiles give an observed wall width $\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$1, but the width is resolution-limited and therefore consistent with much narrower intrinsic walls. The same measurements also revealed bubble-like substructures on a scale of a few micrometers and partial reproducibility of wall positions after thermal cycling, indicating substantial pinning by local defects or strain (Watanabe et al., 16 Apr 2026).

Transmission XMCD spectromicroscopy provided an independent bulk-sensitive confirmation in a $\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$2 free-standing lamella extracted from a bulk single crystal. The measured XMCD spectrum across the Mn $\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$3 edges switches sign eight times across $\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$4 and twice across $\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$5, and the maximum dichroic amplitude is

$\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$6

in excellent agreement with the predicted $\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$7 expected if altermagnetic order extends through the lamella thickness. A surface-only ordered region of $\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$8 would instead yield less than about $\bm m_c=\bm S_{\mathrm A}+\bm S_{\mathrm B}.$9. The same images resolve micron-scale domains, with roughly $T_N$0 domains near the lamella center and about $T_N$1 domains near the edges, $T_N$2 Néel walls of width about $T_N$3, and winding textures consistent with $T_N$4 vortices or antivortices (Yamamoto et al., 25 Feb 2025).

At the atomic scale, MnTe is not a perfectly uniform ideal $T_N$5 $T_N$6-wave altermagnet. Atomic-resolution STEM shows ubiquitous inversion-symmetry-breaking Mn and Te displacements, and the Mn displacement vector map forms locally aligned but longer-scale segmented patterns explicitly described as domain-like structures. The dominant local structural motifs are associated with $T_N$7, $T_N$8, and lower-symmetry $T_N$9, while EMCD measurements on identified $A$0 and $A$1 motifs still show alternating local magnetic order on adjacent Mn layers. A plausible implication is that real bulk MnTe contains a multiscale texture in which macroscopic altermagnetic domains coexist with nanoscale structural variants that locally convert the ideal $A$2-wave state into $A$3-wave or mixed $A$4 spin-splitting regimes (Ren et al., 26 May 2026).

4. Coupling, switching, and partial domain selection

Switching of bulk altermagnetic domains in MnTe is enabled by a weak ferromagnetic moment coupled to the altermagnetic order. The spontaneous remanent moment observed after field cooling is approximately

$A$5

consistent with an earlier scale around $A$6Mn and corresponding to stray fields of only $A$7 mT. The proposed microscopic origin is a spin–orbit-coupling-enabled uncompensated interlayer Dzyaloshinskii–Moriya interaction that cants the spins and locks the sign of $A$8 to the sign of $A$9. A Zeeman coupling to this tiny $\langle 1\bar{1}0\rangle$0-axis weak ferromagnetic moment therefore lifts the antiphase-domain degeneracy during cooling and makes milli-Tesla-scale field selection possible (Liu et al., 20 May 2026).

The resulting state is not a perfect monodomain. The inferred net Néel-vector modulus reaches only about

$\langle 1\bar{1}0\rangle$1

after oblique field cooling at $\langle 1\bar{1}0\rangle$2 T and after field cooling at $\langle 1\bar{1}0\rangle$3 mT along $\langle 1\bar{1}0\rangle$4. The demonstrated switching is therefore best interpreted as partial domain selection or domain-population reweighting. Zero-field cooling gives essentially zero net antiphase imbalance in the interference channels, while field cooling reweights both orientational and antiphase populations. Oblique-field cooling favors a Néel vector approximately along $\langle 1\bar{1}0\rangle$5, perpendicular to the in-plane component of the applied field, consistent with a spin-flop-like orientational selection (Liu et al., 20 May 2026).

Real-space Kerr imaging shows that this controllability coexists with stability and pinning. Thermal cycling through $\langle 1\bar{1}0\rangle$6 changes the domain pattern substantially but not completely: some regions reverse sign, while some wall locations recur. Field cooling in $\langle 1\bar{1}0\rangle$7 T produces a map dominated by one Kerr sign and $\langle 1\bar{1}0\rangle$8 T the opposite sign, but an anomalous central region remains opposite to the trained majority state. Fine spatial structures in the Kerr magnitude are largely preserved when the sign is reversed, which the authors interpret as evidence that local sample properties set the amplitude landscape while the magnetic field primarily selects the sign of time-reversal breaking (Watanabe et al., 16 Apr 2026).

5. Material dependence and probe dependence beyond MnTe

The domain problem is strongly material and probe dependent. In $\langle 1\bar{1}0\rangle$9-Fe$C_3$0O$C_3$1, the decisive variable is the orientation of the Néel vector $C_3$2. Above the Morin transition, $C_3$3 lies in the basal plane and extended room-temperature domains show finite XMCD and XMLD. Below the Morin transition, $C_3$4, the same bulk domains become XMCD-dark in the chosen $C_3$5 geometry, but polarization-independent XAS still detects the reoriented bulk state. Domain walls of width $C_3$6 nm and meron textures with $C_3$7 nm cores locally rotate $C_3$8 into XMCD-allowed or XMCD-forbidden orientations, so the measurable altermagnetic response becomes a function of local spin texture rather than a simple binary marker of ordered versus disordered regions. In this system, the absence of XMCD from a bulk domain does not imply the absence of altermagnetism; it may instead reflect the wrong $C_3$9-orientation for that observable (Yamamoto et al., 10 Mar 2026).

CrSb provides a contrasting case in which the Néel vector is along $_2$00. Three-dimensional ARPES established a bulk $_2$01-wave altermagnetic splitting up to $_2$02 eV near the Fermi level, with symmetry-enforced horizontal nodal planes at $_2$03 and $_2$04 and three vertical nodal planes containing $_2$05. The paper explicitly notes that the probed area likely contains two domains with opposite spin splittings, which would reduce the measured spin polarization, yet such opposite $_2$06-axis domains do not affect spin-integrated ARPES spectra. Here the momentum-space geometry of the bulk splitting is essentially domain-invariant, while the spin sign is domain-dependent. A plausible implication is that CrSb is unusually clean for separating bulk altermagnetic symmetry from domain-specific spin labeling, even though direct real-space domain imaging was not provided (Yang et al., 2024).

These cases together show that “bulk altermagnetic domain” is not a probe-independent object. In MnTe, time-reversal-related antiphase domains are directly distinguishable in reciprocal space by nuclear–magnetic interference and in real space by Kerr or XMCD contrast. In hematite, bulk domains can be XMCD-bright or XMCD-dark depending on $_2$07. In CrSb, opposite domains mainly reverse the spin sign while leaving spin-integrated spectra nearly unchanged. This suggests that bulk-domain observability is set jointly by symmetry, domain population, and the tensor character of the probe.

6. Limitations, disputed cases, and emerging directions

Not every candidate altermagnet supports robust intrinsic bulk domains under realistic conditions. For RuO$_2$08, first-principles work argues that unstrained bulk RuO$_2$09 is most likely nonmagnetic at realistic small or zero $_2$10, with a magnetic phase transition appearing only near $_2$11 eV without SOC and $_2$12 eV with SOC. Epitaxial strain can stabilize magnetic order in specific film orientations, especially (100) and to some extent (110), but the same studies conclude that intrinsic bulk altermagnetic domains are not expected to be robust or generic in unstrained bulk RuO$_2$13. Complementary spin-torque and FORC measurements on RuO$_2$14(100) further show that the $_2$15-odd altermagnetic-like response is strongest in strained films and approaches zero as the lattice relaxes toward the bulk limit, while the strain-stabilized magnetic contribution disappears between $_2$16 and $_2$17 nm thickness. In this material class, bulk-domain language is therefore at best conditional on strain or other extrinsic stabilization (Lee et al., 12 Feb 2026, Jia et al., 24 Jun 2026).

Even in the benchmark material MnTe, limitations remain. Polarized neutron diffraction extracts only the net vector $_2$18, not the full set of six domain populations; the field-selected state remains only partially polarized; and the continued increase of the weak ferromagnetic moment above the apparent neutron saturation near $_2$19 mT is left open. Magneto-optical imaging resolves only the two time-reversal-related Kerr classes, not all three rotational variants within each class, while transmission XMCD remains a thickness-integrated projection. Atomic-resolution work further shows that local inversion-breaking variants are pervasive, so the bulk altermagnetic state sampled by macroscopic transport or optics is likely an average over a structurally inhomogeneous symmetry landscape rather than a single ideal $_2$20 phase (Liu et al., 20 May 2026, Watanabe et al., 16 Apr 2026, Yamamoto et al., 25 Feb 2025, Ren et al., 26 May 2026).

Theoretical work on altermagnetic domain walls extends the bulk-domain problem into dynamics. In a two-dimensional easy-axis altermagnet with bulk domains $_2$21, magnons bound to the domain wall inherit the bulk chiral splitting and acquire an orientation dependence proportional to $_2$22, where $_2$23 is the wall angle relative to the crystal axes. Because these bound modes are gapless Goldstone excitations and can move the observable scale from THz bulk magnons to microwave-accessible wall spectra, they provide a plausible route to spectroscopic readout of bulk-domain symmetry through domain walls rather than through the domains alone (Zeng et al., 4 Jan 2026).

Taken together, current work supports a layered picture of bulk altermagnetic domains. At the coarsest scale they are time-reversal-related or orientationally related regions whose population controls whether macroscopic time-reversal-odd observables survive. At intermediate scales they are shaped by weak ferromagnetic couplings, strain, defects, and surface or interface selection. At the finest scales, at least in MnTe, they coexist with local structural variants that change the symmetry class of the altermagnetic spin splitting itself. The resulting bulk state is therefore best understood as a multiscale domain landscape rather than a uniformly ordered compensated magnet.