Polarisation-dependent single-pulse ultrafast optical switching of an elementary ferromagnet

The study addresses how to achieve deterministic, single-pulse all-optical magnetization reversal in an elemental ferromagnet, specifically bulk fcc Ni. While ultrafast laser-driven demagnetization and all-optical switching have been demonstrated in ferrimagnets (e.g., GdFeCo) and ferromagnetic multilayers (e.g., Co/Pt), single-pulse switching in an elementary ferromagnet remained elusive. The purpose is to identify mechanisms and laser parameters that enable full magnetization reversal on sub-picosecond to picosecond timescales and to clarify the role of light-induced torques and polarization. Achieving such control would be important for low-energy, high-speed information technologies.

The work builds on extensive research since the discovery of ultrafast demagnetization in Ni, including applications to data storage and studies in rare-earth ferromagnets, ferrimagnets, and ferromagnetic thin films. Prior models include: (i) three-temperature models describing coupled electron, spin, and lattice subsystems; (ii) stochastic magnetization dynamics via LLG/LLB equations with laser heating; (iii) microscopic mechanisms such as the inverse Faraday effect (IFE) and broader light-induced magnetic torques; (iv) superdiffusive spin transport of hot electrons; and (v) spin-lattice angular momentum transfer mechanisms. Time-dependent DFT studies have provided insights into sub-100 fs demagnetization. Successful single-pulse switching has been achieved in GdFeCo via heating above compensation and differing sublattice relaxation times (weak helicity dependence), while Co/Pt multilayers require multiple pulses or dual-pulse schemes with helicity dependence. Realistic simulations to date largely addressed demagnetization, not single-pulse switching in elemental ferromagnets, motivating the present study.

The authors use a time-dependent multi-orbital tight-binding framework parameterized from DFT (implemented in the TITAN code) to propagate ground-state eigenvectors in real time by solving the time-dependent Schrödinger equation up to a few picoseconds. The Hamiltonian includes kinetic hopping (H_kin), electron-electron interactions via a Hubbard-like term (H_el-el), spin–orbit coupling (H_so), and the coupling to a time-dependent electric field represented through a vector potential A(t) (dipole approximation; spatial dependence neglected; quadratic and higher terms set to zero). The laser magnetic field is neglected. Pulses are either linearly or circularly polarized with central photon energy ħω = 1.55 eV (λ ≈ 800 nm), various widths (tens of fs up to 1 ps) and field intensities. For a right-handed circular pulse in the yz-plane: A(t) ∝ cos²(πt/τ)[sin(ωt) ŷ − cos(ωt) ẑ]; for linear polarization: A(t) ∝ E0 cos²(πt/τ) sin(ωt) û. The absorbed laser fluence is approximated by the change in system energy divided by the cross-section and stabilizes after the pulse. Calculations are performed for bulk fcc Ni (theoretical lattice constant 3.46 Å) with one atom per unit cell, a 22×22×22 k-point grid, and a Fermi–Dirac smearing temperature of 496 K. Time propagation uses an initial step Δt = 1 a.u., with adaptive control to keep relative and absolute wavefunction errors below 10⁻³. Convergence and stability checks were done by increasing k-point density, tightening tolerances, and small parameter perturbations. The ground-state spin moment is 0.51 μB and is initialized along [001] (z-axis). Magnetization dynamics (longitudinal and transverse components), energy absorption, and orbital-resolved contributions (s, p, d) are tracked in time.

- A single sub-picosecond laser pulse can reverse the magnetization of bulk Ni, accompanied by ultrafast demagnetization. Switching occurs only when the pulse is circularly polarized in a plane containing the initial magnetization direction; linear polarization or circular polarization in a plane perpendicular to the initial magnetization does not induce reversal. - Parameter space mapping reveals a switching region requiring a minimum pulse width of about 60 fs for fcc Ni and a finite absorbed fluence window that broadens with increasing pulse width (Fig. 1b). There is a lower intensity threshold; too high fluence can suppress stable switching despite stronger demagnetization. - Example dynamics: A 60 fs circularly polarized yz-plane pulse yields ≈30% demagnetization when absorbed fluence reaches ≈0.41 mJ cm⁻² at the end of the pulse; on longer timescales magnetization continues to decrease, reaching ≈25% of the ground-state moment by 5 ps. For 50 fs pulses, a threshold around E0 ≈ 6.5–6.8 E* (E* reference E = 9.7×10⁸ V m⁻¹) marks the onset of large-amplitude oscillations. For 100 fs pulses, magnetization reversal is observed with oscillation amplitudes and periods increasing with E0 up to ≈10–11 E*. - The mechanism is an ultrafast light-induced torque on the magnetization: during the pulse, an IFE-like effective field B_IFE points perpendicular to the polarization plane (x-direction for yz-plane polarization), generating a transverse torque that rotates M within the polarization plane. After the pulse, a second torque associated with a modified non-equilibrium magnetic anisotropy field B_MAE drives rotation out of plane (picosecond timescales). Reversing helicity flips the torque direction and rotation sense. If the polarization plane is perpendicular to the initial magnetization, the transverse torque cancels and no rotation occurs. - Orbital-resolved dynamics show rich intra-atomic behavior: sp and d moments exhibit a transient non-collinear state. Three regimes are identified for a 300 fs pulse: R1 (≤~60 fs) intra-orbital SOC-driven spin flips reduce magnetization; R2 (~60–165 fs) strong inter-orbital optical transitions lead to a transient ferromagnetic coupling between sp and d channels with substantial spin angular momentum transfer from d to sp, enabling z-component switching; R3 (~≥165 fs) partial recovery of d magnetization with near antiparallel alignment to sp and short-period oscillations that persist post-pulse.

The results directly address the challenge of achieving deterministic, single-pulse all-optical switching in an elemental ferromagnet by identifying the requisite polarization geometry and pulse parameter window. The discovery that only circular polarization in a plane containing the initial magnetization leads to switching demonstrates the pivotal role of light-induced torques, specifically an inverse-Faraday-like field during the pulse and a dynamically modified anisotropy afterwards. The mapping of pulse width and fluence clarifies practical constraints: a minimum duration (~60 fs) and a finite fluence window are necessary, countering the naive expectation that higher energy always enhances switching. The orbital-resolved analysis reveals that transient intra-atomic non-collinearity and inter-orbital spin and population transfers are central to reversal, offering a microscopic picture beyond effective thermodynamic models. These findings have relevance for ultrafast, energy-efficient spintronic device concepts where polarization control of single pulses can deterministically write bits in elemental ferromagnets.

This study demonstrates, via time-dependent tight-binding simulations parameterized from DFT, that single-pulse, helicity-dependent magnetization reversal is achievable in bulk Ni when using circularly polarized light in a plane containing the initial magnetization. A well-defined pulse-parameter window is identified, including a minimum pulse width (~60 fs) and an appropriate fluence range. The mechanism is attributed to ultrafast light-induced torques (IFE-like during the pulse and anisotropy-driven post-pulse), and the dynamics reveal transient intra-atomic non-collinearity with inter-orbital angular momentum transfer that mediates switching. These insights open avenues for optically controlled spintronic devices based on elemental ferromagnets. Future work could incorporate explicit dissipation channels (electron-phonon and lattice dynamics) to assess stabilization, explore different photon energies and materials, and perform experimental validation of the predicted polarization dependence and parameter windows.

- The simulations neglect explicit dissipation mechanisms (e.g., electron–phonon coupling, lattice heating/relaxation). The authors note that dissipation, active on picosecond timescales, could stabilize switched states. - Dipole approximation is used for light–matter coupling; spatial dependence and higher-order terms are ignored. The magnetic component of the laser field is neglected. - The effective non-equilibrium magnetic anisotropy field B_MAE is not explicitly defined or computed microscopically; its role is inferred from dynamics. - Results pertain to bulk fcc Ni at a single central photon energy (ħω = 1.55 eV) and specific computational settings; generalization to other materials and frequencies is not established within this work.