Magnets come in two kinds. Ferromagnets have all spins aligned in the same direction, producing a net magnetization that interacts with external fields. Antiferromagnets have spins aligned in alternating directions, producing zero net magnetization. These two categories have organized condensed matter physics for a century, and every magnetic material has been classified as one or the other (or a blend, like ferrimagnets).
Altermagnets are a third class, formally distinguished from both only in recent years. Like antiferromagnets, they have zero net magnetization — the spins cancel globally. But unlike conventional antiferromagnets, they split the electronic bands by spin in a momentum-dependent way. Electrons moving in one direction have one spin polarization; electrons moving in a perpendicular direction have the opposite polarization. The spin splitting follows the point-group symmetry of the crystal — d-wave, g-wave, or i-wave patterns in momentum space, analogous to the angular momentum labels of atomic orbitals.
This combination — no net magnetization, but momentum-dependent spin splitting — enables effects that neither ferromagnets nor antiferromagnets can produce alone. A ferromagnet splits spins uniformly in momentum space (all electrons shifted the same way), which produces spin-polarized currents but also stray magnetic fields and sensitivity to external fields. An antiferromagnet produces no spin splitting at all (the two sublattices compensate exactly in each momentum direction). An altermagnet splits spins along specific crystallographic directions without producing stray fields or net magnetization.
Sarkar, Sarkar, and Agarwal (arXiv 2602.23273, February 2026) show that this spin splitting enables a spin-splitter effect where impurity scattering produces transverse spin currents. An electric field drives longitudinal charge current; asymmetric scattering from impurities deflects opposite spins in opposite transverse directions. The result is a spin current perpendicular to the charge current, analogous to the spin Hall effect but with a critical difference: the extrinsic spin conductivity is time-reversal even.
Time-reversal even means the effect does not change sign when the magnetic configuration is reversed. In ferromagnets, spin Hall effects are time-reversal odd — they reverse with the magnetization. In altermagnets, the extrinsic spin-splitter current survives time reversal because the momentum-dependent spin splitting has the symmetry of a quadrupole, not a dipole. Reversing all spins maps the spin pattern onto itself (up to a rotation), preserving the scattering asymmetry that generates the current.
The practical implication is that altermagnetic spintronic devices could generate spin currents without the disadvantages of ferromagnets — no stray fields to disturb neighboring components, no sensitivity to demagnetization, no hysteresis. The applied material in this study is FeSb2, a d-wave altermagnet where the spin splitting follows a cos(2phi) pattern. The impurity scattering mechanism (skew scattering and side-jump) provides the dominant contribution, meaning the effect can be tuned by impurity engineering — choosing dopant type and concentration — without modifying the host crystal structure.
A century of binary classification — ferromagnet or antiferromagnet — missed a class of materials that was there all along. The third kind wasn't hidden by rarity or exotic conditions. It was hidden by a classification scheme that didn't have the right symmetry labels to distinguish it.