Intuition

Push a spin-polarized current through a ferromagnetic film. Each electron carries an angular momentum in its spin. Inside the film the electron must align its spin to the local exchange field — and by conservation of angular momentum, what the electron loses, the magnetization gains. A current therefore exerts a torque on the magnetization, with no external magnetic field needed.

Predicted by Slonczewski (1996) and Berger (1996), demonstrated experimentally by Tsoi (1998) and Katine (2000), this spin-transfer torque (STT) has become the workhorse of modern spintronics. The same effect drives two seemingly opposite phenomena:

  • If the torque is dissipative-like, the current writes a new magnetization state. This is the writing mechanism of STT-MRAM.
  • If the torque is anti-dissipative, the current sustains a steady-state precession. This is the spin-torque nano-oscillator (STNO).

Formal definition

Add a Slonczewski term to the LLG equation for the free-layer reduced magnetization :

where is the polarization direction (the magnetization of the reference layer) and is the Slonczewski torque amplitude,

with the current density, the spin polarization, the free-layer thickness, and the electron charge. This is the LLGS equation.

The Slonczewski term has the same geometric form as Gilbert damping — both are double cross products — but with opposite signs depending on current direction. Hence STT acts like either extra damping or negative damping.

Key results

1. Current-driven switching

When STT acts as extra damping towards , the free layer relaxes to its anti-parallel or parallel state with very low energy cost. The Sun–Slonczewski critical current density to switch a uniaxial single-domain free layer is

Typical numbers: A/cm — enormous in absolute units, but easy to reach through a nanometer-thin pillar of nm diameter. This is the writing mechanism of STT-MRAM and the reason modern MRAM cells dispense with inductive write lines.

2. Current-driven oscillation (STNO)

When the current flows the other way, the Slonczewski term acts as negative damping: it amplifies precession instead of damping it. Steady-state is reached when negative STT damping exactly compensates Gilbert damping. The free layer then precesses indefinitely at a frequency set by and tunable by — typically GHz.

These spin-torque nano-oscillators are highly compact (few tens of nm), tunable, and CMOS-compatible — they are studied for microwave sources, RF mixing, and neuromorphic computing (a single STNO can emulate a neuron).

3. Macrospin picture and stability boundary

For a macrospin in the linearized regime, the LLGS equation reduces to an ordinary differential equation in the cone angle with an effective damping

  • → relaxation, switching path.
  • → growing precession, oscillator path.
  • → the critical current above.

This single sign change governs every STT device.

4. Spin-Hall variant

A separate but related mechanism — spin-orbit torque (SOT) — generates a spin current via the spin-Hall effect in a heavy metal (Pt, Ta, W) underneath the free layer. SOT decouples read and write paths, allowing better endurance and lower error rates, and is now displacing STT in the highest-performance MRAM cells.

Summary

A spin-polarized current carries angular momentum; flowing through a ferromagnet, it transfers that momentum to the local magnetization. Adding the Slonczewski term to LLG gives the LLGS equation of modern spintronics:

  • one current direction → writing, the basis of STT-MRAM;
  • the opposite direction → sustained precession, the basis of spin-torque nano-oscillators.

The critical current ( A/cm) is large in absolute units but easily reached in nanometer-thin pillars. STT, and its cousin SOT, are the physical mechanism behind almost every active spintronic device in production today.

Connections

References

  • J. C. Slonczewski, J. Magn. Magn. Mater. 159, L1 (1996).
  • L. Berger, Phys. Rev. B 54, 9353 (1996).
  • D. Ralph & M. Stiles, J. Magn. Magn. Mater. 320, 1190 (2008) — review.
  • A. D. Kent & D. C. Worledge, Nat. Nanotechnol. 10, 187 (2015).