A Reshaped Pulse Could Cut Magnetic Memory's Switching Energy Nearly 100-Fold

A conventional magnetic-field pulse needed 91.2 nanojoules to flip a bit in a simulated van der Waals magnet. A pulse reshaped by a 60-year-old branch of applied mathematics needed 0.94 nanojoules to do the same job. That gap, reported by physicists at the University of Edinburgh, is the difference a differently shaped field made in simulation, not a new material or a faster chip.

How Optimal Control Theory Reshapes a Magnetic-Field Pulse

Magnetic memory stores a bit by pointing a tiny magnetic region one way or the other. Switching that region from one stable orientation to its opposite is the basic act behind writing a "0" or a "1." For decades, engineers have leaned on current-driven methods, specifically spin-transfer torque and spin-orbit torque, sold under the names STT-MRAM and SOT-MRAM, because plain magnetic-field pulses were considered too energy-hungry and too hard to aim at a single nanoscale cell.

Researchers Mohammad H. Badarneh, PeiYu Cai, and Elton J. G. Santos revisited field-driven switching using optimal control theory, a mathematical framework for finding the most efficient route between a starting state and a target state. Instead of firing a fixed-shape pulse and letting the magnet fight its way to the new orientation, the team calculated how the field's strength and direction should change moment to moment so the magnetization drifts into place along its own natural rotation, minimizing the heat wasted along the way. According to the paper, published in Advanced Materials, the calculations also incorporated realistic experimental constraints, which the authors say makes the approach more relevant to devices that could eventually be built rather than an idealized textbook case.

From 91.2 to 0.94 Nanojoules: What the Simulation Found

The team tested the method computationally on three ultrathin van der Waals magnets, Fe₃GaTe₂, Fe₃GeTe₂, and CrSBr, materials built from weakly bonded atomic layers that researchers are already exploring for compact, fast electronics. In these simulations, the optimized pulses completed magnetization reversal in roughly one to ten picoseconds, using field amplitudes more than ten times weaker than standard switching pulses. Under the conditions studied, that translated to a drop from as much as 91.2 nanojoules for conventional field pulses to as little as 0.94 nanojoules for the optimal-control version, a reduction of roughly 97-fold on those two figures alone, per the University of Edinburgh's summary of the results.

By additionally tuning material-specific properties such as magnetic damping and anisotropy, the authors project that switching energy could eventually reach the femtojoule range, a thousand-fold smaller unit than a nanojoule, which would put the approach in competition with or ahead of established STT and SOT schemes. That projection is a further extrapolation beyond the 0.94-nanojoule result actually simulated, and the paper frames it as a design space to explore rather than a demonstrated number. The broader modeling also suggests the approach could bring magnetic memory closer to the Landauer limit, the thermodynamic floor on the energy required to process a single bit, though closer is not the same as reaching it.

Why This Is Still a Simulation, Not a Working Chip

Every figure in the study comes from computer modeling of three specific materials under idealized but constraint-aware conditions. The paper includes proposed device geometries and methods for delivering the shaped fields, offered as a starting point for other groups to test experimentally, but no physical device has switched using this method yet. The University of Edinburgh's own account of the study confirms the results are computer simulations intended to guide future experiments, not measurements from a built device. The authors also note the technique isn't limited to magnetic fields: the same optimal-control mathematics could in principle be applied to shape the electrical currents used in STT and SOT devices, or the ultrafast laser pulses being explored for optical switching, which would extend its relevance well beyond the three materials tested here. Whether any of those extensions hold up outside a simulation is the open question the next round of lab work will have to answer.