← all systems · see it in The System
Antiferromagnet
A magnetically ordered material in which neighbouring magnetic moments settle in opposition, on two or more sublattices whose moments cancel, so the bulk has a vanishing total magnetization; the order exists below the Neel temperature and gives way to a paramagnet above it. Its order is read through the staggered (Neel) vector L = S1 - S2 rather than a net moment; its kinds run from collinear two-sublattice magnets (NiO, MnO, chromium, hematite with canting) through zigzag, helical, non-collinear and non-coplanar orders to frustrated magnets on triangle-based lattices, layered van der Waals antiferromagnets down to a few atomic layers, and synthetic antiferromagnets built from coupled ferromagnetic films.
Properties published scienceour proposal
Sourced fields, and our own readings.
| boundary | Inside: the magnetic ions of the crystal, their moments, the exchange couplings between neighbours and the ordered sublattice pattern with its spin waves and domains. Outside: the lattice's elastic and vibrational life (which the order couples to), the heat bath, applied fields, light, currents, neighbouring ferromagnetic layers, and the probes (neutrons, light) that read it. The edge is the Neel temperature in time and the sample's surfaces and interfaces in space; the paramagnet above TN is the same material without the system. |
| nested level | Built from magnetic ions (Ni2+, Fe, Mn, Cr, Co...) in a crystal, their unpaired electron spins and the exchange between them; it is one of the ordered states of magnetism (with ferromagnet and ferrimagnet), held in a crystal, and in devices it is a layer inside spin valves and memories. |
| scale | molecular |
Measures (45) published science
| Neel temperature, NiPS3 (bulk) | ~155 K |
| Rayleigh-wave phase velocity, PM to AFM, NiPS3 220 nm film | ~1920 to ~1815 m/s |
| Rayleigh-wave phase velocity drop across TN, NiPS3 220 nm film | ~6 % |
| Rayleigh-wave phase velocity, PM to AFM, NiPS3 260 nm film | 1610 to 1500 m/s |
| Rayleigh-wave phase velocity drop across TN, NiPS3 260 nm film | ~7.3 % |
| effective modulus reduction implied by the velocity drop | ~12 % |
| Rayleigh-wave hardening above TN on cooling | ~0.13 (220 nm); ~0.10 (260 nm) GHz |
| Rayleigh-wave magnetoelastic softening across TN | ~0.23 (220 nm); ~0.30 (260 nm) GHz |
| Rayleigh frequency variation deep in the AFM phase | ~0.05-0.07 GHz |
| Rayleigh-wave frequency, NiPS3 220 nm film | ~4.2 (measured); ~4.6 (calculated) GHz |
| calculated Rayleigh frequency drop (DFT plus finite elements) | 4.603 to 4.578 (0.54%) GHz |
| c66 (DFT), NM to AFM | 47.4 to 55.4 GPa |
| c35 (DFT), NM to AFM | -0.5 to +0.4 GPa |
| independent elastic constants of monoclinic NiPS3 | 13 count |
| Rayleigh velocity sensitivity to a 20% rise in c44 | +4.85 (c33: +1.94) % |
| in-plane anisotropy of the softening (random crystal edges) | ~1 % |
| linewidth deviation of the 255 cm-1 Raman mode at TN | ~0.3 cm-1 |
| NiPS3 film thicknesses studied by BMS | ~220 and ~260 nm |
| lowest temperature at which monolayer NiPS3 order was looked for | 25 K |
| two-magnon Raman peak, bulk NiPS3 (AFM phase) | ~550 cm-1 |
| antiferromagnetic resonance (characteristic clock) | THz scale (ferromagnetic resonance: GHz) frequency |
| zone-edge spin-wave frequency, KNiF3 | 20 THz |
| magnetic oscillation, YMnO3 (three-sublattice order) | 1.3 THz |
| heating of NiO by a THz pump pulse | <10 mK |
| field needed to reorient spins | tens to hundreds T |
| domain-wall velocity limit | ~100 (AFM); ~1 (SAF); ~0.1 (bare FM) nm/ps |
| pulse lengths recorded by a CuMnAs bit cell | ~10 ms to ~100 ps (eight orders of magnitude) |
| field up to which CuMnAs memory was insensitive | 12 T |
| SAF interlayer exchange against crystal inter-sublattice exchange | two orders of magnitude smaller ratio |
| frustration ratio f = thetaW/Tc | ~1 (non-frustrated 3D); much greater than 1 (strongly frustrated; review focus f >~ 10) ratio |
| temperature where short-range order sets in, frustrated magnets | ~|thetaW|/10 K |
| Weiss constant of spin ice Dy2Ti2O7 | 0.5 K |
| susceptibility at the Neel temperature | maximum (staggered susceptibility diverges) shape |
| blocking temperature against Neel temperature | usually lower comparison |
| ground states of an Ising antiferromagnetic triangle | 6 of 8 states |
| Neel temperature, NiPS3 (resonant X-ray diffraction) | 155.5 K |
| critical exponent beta, NiPS3 (bulk) | 0.367 +/- 0.013 dimensionless |
| exchange constants, NiPS3 (DFT, four-state mapping) | J1 = 3.81, J2 = 0.15, J3 = -14.6 (negative = antiferromagnetic) meV |
| interplane exchange, NiPS3 (inelastic neutrons, Scheie et al., as reported) | -0.38 (ferromagnetic) meV |
| Neel temperature, Mn3Sn | ~430 K |
| anomalous Hall conductivity, Mn3Sn | ~20 (room temperature); >100 (low temperature) 1/(ohm cm) |
| net moment of Mn3Sn | ~0.002 Bohr magnetons per Mn |
| critical current density for Mn3Sn switching | 2e11 (Pt), 5e10 (W) A/m^2 |
| upper bound on RuO2 moment (muSR) | 1.14e-4 (bulk, Kessler); 7.5e-4 (11 nm films, Kessler); 4.8e-4 (bulk, Hiraishi) Bohr magnetons per Ru |
| time to destroy chromium's spin density wave after photoexcitation | sub-100 fs |
Open contentions (6) open contention
Where sources or readings disagree. Both stay on the map.
- C-3689 When an electric current 'switches' an antiferromagnet, is it really turning the magnetic order, or is some of the signal just heat?
booked: The map keeps the competing readings; none is ruled.
the other reading: electrical signals are torque switching of the Neel vector; signals are (in part) thermal artefacts - C-3695 Is an altermagnet a kind of antiferromagnet or a third basic kind of magnet?
booked: The map keeps the competing readings; none is ruled.
the other reading: altermagnets are a kind of antiferromagnet; altermagnetism is a third basic phase beside ferro- and antiferromagnetism - C-3774 In the layered magnet NiPS3, do neighbouring layers point the same way or opposite ways?
booked: The map keeps the competing readings; none is ruled.
the other reading: layers coupled antiferromagnetically; layers coupled ferromagnetically - C-3775 Does magnetic order split one of NiPS3's vibrations in two?
booked: The map keeps the competing readings; none is ruled.
the other reading: the splitting is intrinsic to the order; it is not resolved, or depends on the sample - C-3776 Does the magnetism of NiPS3 behave as two-dimensional or three-dimensional?
booked: The map keeps the competing readings; none is ruled.
the other reading: 2D XY / XXZ; 3D Heisenberg (bulk) - C-3791 Is ruthenium dioxide magnetic?
booked: The map keeps the competing readings; none is ruled.
the other reading: RuO2 is an antiferromagnet (altermagnet); RuO2 is nonmagnetic
Cross-scale patterns seen here (18) our proposal
Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Antiferromagnet. Each links to every other system that shows it.
- same role, different builds (structural)
The job: an opposed, compensated pair of moments that cancels its own stray field (used in reference layers). Built two ways: inter-atomic exchange inside a crystal antiferromagnet, and carrier-mediated coupling across a non-magnetic spacer in a synthetic antiferromagnet, 'of a physically distinct o - same shape, opposite sign (structural)
One mechanism, two signs: the nearest-neighbour exchange on a bipartite lattice gives ferromagnetic order with one sign and antiferromagnetic order with the other ('Depending on the sign of that interaction'). At the next scale down, in NiPS3 'different bonds contribute with opposite signs' of dJ/du - symmetry breaking (name the cut) (structural)
Before: the paramagnet above TN, no preferred direction. The cut: the Neel temperature. After: a Neel vector chosen (NiPS3: ferromagnetic zigzag chains along a, opposed along b, layers opposed along c). The chosen direction sets what follows: in exchange bias the ferromagnet's field breaks the symme - the whole sets the local reference (structural)
Which way is 'up' on a site is set by the whole: a moment's direction is defined by its sublattice in the staggered pattern (L = S1 - S2), not locally. The NiPS3 paper reads the two scales apart: optical phonons register bond-scale exchange and short-range correlations that persist above TN, while a - trade one property for another (structural)
No net moment buys robustness and speed and is paid for in handles: 'the absence of net magnetization and stray fields eliminates crosstalk' and makes manipulation faster, but control 'requires exceedingly high magnetic fields' and the order cannot be seen by magnetometry. Kinds settle along the dia - Umwelt: each system carves its own slice of one shared medium (structural)
Each probe carves its own slice of one order. Magnetometry sees nothing ('cannot be detected with conventional magnetometry'); neutrons see the pattern; optical phonons see short-range correlations; acoustic surface waves see long-range order; Raman two-magnon scattering sees order in two layers but - who opens the door (structural)
Who opens the door for the Neel spin-orbit torque: the receiver's own physics. The current only grips the order where the crystal's inversion-partner sites coincide with the opposite spin sublattices (CuMnAs); the door is the crystal's symmetry. For exchange bias the door is opened at an earlier tim - borrowed-then-held (partial)
Exchange bias: while cooling through TN the reference direction is supplied from outside (the applied field and the ordered ferromagnet's effective field); after ordering the antiferromagnet holds that direction itself and supplies it to the ferromagnet. Both phases are shown in the source; whether - conservation law plus closure (partial)
The elastic wave equation (-rho omega^2 u = div sigma) predicts the Rayleigh dispersion only once a stiffness tensor closes it; the magnetic state supplies the closure (c_ij computed separately in the AFM and NM states, c_eff = c0 - lambda^2 chi_m). - holding / entrainment (partial)
The two sublattices do not precess at their own anisotropy pace (GHz, as a ferromagnet would) but at the THz resonance set by the exchange between them: the coupled whole sets the tick of each part. - namespace completeness (the map predicts its gaps) (partial)
The spin-symmetry map of 2D magnets (Ising, XY, Heisenberg) predicted which layered antiferromagnets keep order in one layer: Ising FePS3 keeps it to the monolayer; the XY class 'cannot have a long-range order at finite temperatures', and monolayer NiPS3 was then measured to lose it. A slot predicte - smooth to sharp in finite time (partial)
Approaching TN the staggered susceptibility diverges while the ordinary susceptibility only peaks; in NiPS3 the Rayleigh frequency rises smoothly on cooling, then drops at ~150 K and saturates; interlayer shear shows critical behaviour near TN. - The carry (partial)
The wait face: a domain wall in an antiferromagnet cannot outrun the magnon velocity ('the limit is set by the magnon velocity'), so a 100 nm bit switches on a picosecond scale; in a ferromagnet the Walker breakdown is the slower limit (~0.1 nm/ps). - The Payment to Entropy (partial)
Form (thermal, small): writing the order with the magnetic part of a THz pulse pays almost no toll: the NiO volume warms by less than 10 mK, while optical excitation through spin-orbit coupling deposits heat in other degrees of freedom. The toll depends on which door the energy enters by. No end-to- - causal ceiling (surface)
With no net moment, any probe that couples only to a net magnetization reads nothing: a limit set by the order's own structure, not by instruments; but other channels (neutrons, SHG, sound) reach past it, so it is a channel limit, not a limit on what can be known. - coordination-before-mapping (surface)
Order across a whole crystal with nothing but neighbour-to-neighbour exchange: each moment opposes its neighbours and a lattice-wide staggered pattern results, with no stand-in for the whole anywhere in it. - partner-becomes-part (surface)
In a synthetic antiferromagnet two ferromagnetic films, each a magnet of its own, couple antiparallel through a spacer and act as one artificial antiferromagnet; no inheritance is involved. - templating (surface)
The same antiferromagnetic pinning layer (IrMn, PtMn) is swapped across GMR and TMR stacks and keeps pinning; the Ising-on-a-bipartite-lattice model is the template the field reuses for every simple antiferromagnet.
Valence: what it trades, and with whom (11) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Phonon via sys:acoustic-wave · Antiferromagnet gain
- Electron via sys:electricity · Antiferromagnet gain
- Photon via sys:electromagnetic-field · Antiferromagnet mixed
- Electromagnetic field via sys:electromagnetic-field · Antiferromagnet even
- Neutron via real:neutron · Antiferromagnet even
- Thermal energy (heat, temperature and heat flow) via sys:thermal-energy · Antiferromagnet loss
- Human (Homo sapiens) via sys:electromagnetic-field · Antiferromagnet even
- Magnon via sys:exchange-interaction · Antiferromagnet mixed
- Ferromagnet via sys:exchange-interaction · Antiferromagnet even
- Crystal via sys:matter · Antiferromagnet gain
- Exciton via sys:matter · Antiferromagnet even
Partners and parts published science
Built from / builds (8)
- Electron
- Nickel atom
- Magnon
- Exchange interaction
- Crystal
- Magnetic domain
- Competition: superconductivity vs antiferromagnetic order (cuprates)
- Ferromagnetic J1 vs antiferromagnetic J2 in a frustrated chain
Sits inside / holds (3)
expressed by (4)
floor filled by (15)
- Electron
- Photon
- Nickel atom
- Thermal energy (heat, temperature and heat flow)
- Acoustic wave (sound: pressure waves in matter)
- Phonon
- Magnon
- Ferromagnet
- Mineral
- Exchange interaction
- Spin (intrinsic angular momentum)
- Domain wall (the bubble wall)
- Crystal
- Magnetic domain
- Spin valve
represents (1)
Kinds (13)
- Magnet
- State of matter
- Helical antiferromagnet (helimagnet)
- Collinear antiferromagnet
- Zigzag antiferromagnet
- Non-collinear and non-coplanar antiferromagnet
- Geometrically frustrated magnet
- Layered (van der Waals) antiferromagnet
- Synthetic antiferromagnet (SAF)
- Speromagnet
- Altermagnet
- Itinerant (spin-density-wave) antiferromagnet
- Magnetoelectric antiferromagnet
wired (6)
Notes published science
- An antiferromagnet's magnetism cancels itself out, so a magnet cannot feel it and a magnetometer cannot see it; in 2026 a team heard it instead: a sound wave running along the surface of a NiPS3 flake slows by about 6 to 7% the moment the spins lock into their zigzag pattern at about 155 K (src:ebrahim-nataj-2026).
- One layer too thin: NiPS3 keeps its long-range antiferromagnetic order down to two atomic layers but loses it in a single layer, even at 25 K, as the maths of two-dimensional XY magnets had predicted; its Ising cousin FePS3 keeps order in one layer (src:kim-2019-nips3).
- Because the two opposite sublattices drive each other, an antiferromagnet rings at terahertz frequencies, a thousand times faster than a ferromagnet's gigahertz, and its domain walls can move about 100 nanometres per picosecond; a micron-size memory cell counted pulses from 10 milliseconds down to 100 picoseconds (src:gomonay-2017-concepts).
- A leading candidate for a whole new class of magnet turned out not to be magnetic at all: muon studies put RuO2's moment below about one ten-thousandth of a Bohr magneton per atom, and a re-measurement traced the earlier neutron signal to neutrons bouncing twice inside the crystal (src:kessler-2024-ruo2).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
antiferromagnet (reader's view: the host crystal, its moments, the exchange between them, the order, its waves, what it does to its surroundings)
- L1: The host crystal and its magnetic ions: transition-metal compounds, especially oxides (NiO, MnO, hematite), metals (Cr), alloys (FeMn), layered NiPS3 with a Ni honeycomb
- L2: Local moments: unpaired electron spins on each ion (Ni2+ in NiPS3)
- L3: Exchange between neighbours: its sign sets antiparallel order; its anisotropy (Ising, XY, Heisenberg); frustration on triangles
- L4: The order: opposed sublattices, Neel vector L, below TN; zero net magnetization
- L5: Dynamics: magnons, THz resonance, inertial motion of L, fast domain walls
- L6: Response on the surroundings: lattice stiffness, light, a pinned ferromagnet, resistance, no stray field
the boardroom (shared floors)
- L4: The order: two (or more) sublattices with opposed moments, read through L = S1 - S2 below the Neel temperature (NiPS3: zigzag, ~155 K); every influence ends as a change in its size, direction or domains
- L5: Dynamics of the order: spin waves (magnons; two gaps in NiPS3), THz antiferromagnetic resonance raised by exchange, inertia-like motion of L, domain walls up to the magnon velocity
- L6: What the order does to its surroundings: no stray field; pins a neighbour ferromagnet; retunes the lattice's stiffness; marks light and neutrons; changes resistance. The top floor: nothing in an antiferromagnet reads, regulates or means
way in: A neighbouring ferromagnet (the interface)
- L1: A ferromagnetic film grown on the antiferromagnet
- L2: Interfacial exchange between the two layers' boundary spins
- L3: On cooling through TN in a field, the ordered ferromagnet's effective field breaks the symmetry and sets the antiferromagnet's domains
way in: Applied magnetic field
- L1: A magnetic field reaching the crystal (lab magnets, a neighbour ferromagnet's field, a THz pulse's magnetic component)
- L2: Zeeman coupling acts on each sublattice; with no net moment the order is only weakly moved
- L3: Sublattices cant; a ferrimagnet-like net moment appears; reorienting the spins needs fields that scale with exchange (tens to hundreds of tesla); with the field along the easy axis the sublattices first flop (the spin-flop field, exchange-enhanced compared with a ferromagnet's coercive field)
way in: Electric current (spin-orbit torques)
- L1: A current pulse through the antiferromagnet (CuMnAs) or a heavy-metal neighbour
- L2: Staggered non-equilibrium spin polarization on inversion-partner sites (inverse spin galvanic effect)
- L3: A Neel spin-orbit torque, commensurate with the order, that rotates the Neel vector
way in: Electric field (the magnetoelectric effect)
- L1: An electric field (a bias voltage across top and bottom electrodes) in a magnetoelectric antiferromagnet such as Cr2O3
- L2: The field displaces the magnetic ions relative to their ligands
- L3: The two sublattices become inequivalent: a net ferrimagnetic moment is induced; with exchange bias or the anomalous Hall effect as read-out, sublattices in thin Cr2O3 films have been switched electrically
way in: Heat (thermal fluctuations)
- L1: The heat bath of the crystal at temperature T
- L2: Thermal agitation of each moment against the exchange that orders it
- L3: Above the Neel temperature thermal energy destroys the order (paramagnet); below, the order grows; in a 2D XY layer fluctuations forbid long-range order at any finite temperature
way in: Light and THz radiation
- L1: Photons from optical to THz: femtosecond laser pulses, intense THz pulses (above 1 MV/cm, 0.33 T), probe beams
- L2: Coupling: the THz magnetic field's Zeeman torque on the spins, or optical excitation acting through spin-orbit coupling and electron transitions
- L3: An impulse on the Neel vector: precession at THz frequencies, or inertial switching between metastable states by a pulse shorter than the barrier time
way in: Neighbouring moments (exchange through the ligands)
- L1: The neighbouring magnetic ions and the ligand orbitals between them (Ni2+ on a honeycomb bridged by P2S6 units; transition-metal oxides)
- L2: Exchange between neighbours: its sign decides antiparallel (antiferro) or parallel (ferro) alignment; in NiPS3 intralayer exchange dominates
- L3: The coupling's strength and anisotropy: XY-type (easy-plane) in NiPS3, Ising in FePS3, Heisenberg in MnPS3; on triangles the bonds cannot all be satisfied (frustration)
way in: Neutrons (the probe)
- L1: A beam of neutrons, each carrying a magnetic moment
- L2: Magnetic scattering by the ordered moments of transition-metal oxides (nickel, iron, manganese oxides)
- L3: The staggered pattern read as diffraction: neutron diffraction is the main detector of the order in bulk crystals; neutrons also read NiPS3's magnon gaps
way in: The lattice: strain and sound
- L1: The crystal's strain and its vibrations: optical phonons (out-of-phase motion in the cell) and long-wavelength acoustic and Rayleigh surface waves
- L2: Exchange depends on bond length and angle: optical modes modulate single bonds (dJ/du), uniform strain modulates all bonds at once (dJ/de)
- L3: Magnetoelastic energy: strain shifts the balance of the order; interlayer shear couples directly to the order parameter near TN; in the strict 2D limit small distortions sway stability
way out: Anomalous Hall voltage (non-collinear antiferromagnets)
- L1: A Hall voltage in a circuit: the readout used for electrical switching of Mn3Sn at room temperature
- L3: A transverse Hall voltage as large as in ferromagnetic metals, switchable by a few hundred oersted or, in bilayers, by current
- L6: A non-collinear (chiral) order such as Mn3Sn's three-sublattice 120-degree pattern, with almost no net moment
way out: Marks on light (magneto-optics, SHG, Raman, THz emission)
- L1: Polarization, frequency-doubled and inelastically scattered light; THz radiation
- L3: Effects quadratic in L (magnetic linear dichroism, the Voigt effect), second-harmonic generation that tells magnetic space groups, two-magnon Raman scattering, THz emission from precession
- L6: The order and its precession
way out: Order into the crystal: the lattice softens
- L1: The crystal's sound: a Rayleigh surface wave in NiPS3 travels ~6-7.3% slower below TN; optical phonons barely move
- L3: The elastic tensor is renormalized (c_eff = c0 - lambda^2 chi_m): in DFT c66 rises 47.4 to 55.4 GPa, c12 and c23 fall, c35 changes sign; the Rayleigh velocity follows c33 and c44
- L6: Long-range order sets in; uniform strain modulates all exchange couplings at once (exchange striction)
way out: Pinning a neighbour ferromagnet (exchange bias)
- L1: The ferromagnet's magnetization pinned: the reference layer of spin valves and read heads
- L3: Reversing the ferromagnet would cost a Neel domain wall in the antiferromagnet, so its loop shifts
- L6: The antiferromagnet's domain state, fixed below TN (and below the blocking temperature)
way out: Resistance (electrical readout)
- L1: A resistance read by a circuit: multi-level, neuron-like memory cells
- L3: Anisotropic magnetoresistance, even in the moment and so allowed in antiferromagnets
- L6: The Neel-vector direction after writing
way out: Scattered neutrons
- L1: Neutrons leaving with changed direction and energy
- L3: Magnetic diffraction peaks and inelastic magnon scattering
- L6: The ordered pattern and its spin waves