First Experimental Evidence: Heat Can Drive Electron Orbital Currents

• CuOₓ (naturally oxidized copper film): the orbital current source. Earlier studies showed it generates substantial orbital currents while producing negligible spin current — exactly the selectivity this experiment demanded.
• Pt (thin platinum film): the translator. Spin-orbit coupling in platinum converts orbital angular momentum into spin angular momentum.
• YIG (yttrium iron garnet): a magnetic insulator acting as a spin current gate, controlling the reflection and absorption of spin currents without adding unwanted charge transport.

The mechanism runs like this. A temperature gradient applied along the sample triggers the orbital Nernst effect in CuOₓ, accumulating orbital angular momentum near its surfaces. Pt converts that into spin. The spin current travels to the Pt/YIG interface, where its fate — reflected or absorbed — depends on the direction of YIG's magnetization. The reflected portion becomes an electrical signal through the inverse spin Hall effect in platinum. By measuring voltage differences between different magnetization orientations, and feeding the data into a macroscopic spin/orbital current diffusion model they developed, the researchers quantified the orbital Nernst contribution.

Three confounders, one clean separation

The signal is small and the noise sources are real. The team had to deal with three of them.

First, the charge Seebeck effect: the connecting wires, not just the sample, experience temperature differences and produce extra voltage. That background had to be identified and subtracted.

Second, an unintended out-of-plane temperature gradient. The heater geometry and the differing thermal conductivities of the films, substrate and surroundings inevitably push some heat perpendicular to the film plane. That contaminates the orbital Nernst signal with spin Seebeck voltages. The fix: the two responses have distinct symmetries under magnetic field reversal, so the researchers separated the symmetry-even in-plane response from the symmetry-odd out-of-plane one.

Third, platinum itself generates a transverse spin current under a temperature gradient — the conventional spin Nernst effect. Any CuOₓ contribution had to be distinguished from Pt's own.

The sign reversal that clinched it

Here is the evidence worth quoting. The team compared three structures: YIG/Pt, YIG/CuOₓ, and YIG/Pt/CuOₓ. The Pt-only sample showed a clear spin Nernst thermal signal. The CuOₓ-only sample showed almost nothing detectable — consistent with CuOₓ producing orbital, not spin, angular momentum that goes unread without a Pt converter.

Then the striking part. Combine Pt and CuOₓ, and the measured signal reversed sign. The estimated total spin current reaching the YIG interface more than doubled compared with the Pt-only sample. A sign reversal plus a two-fold enhancement is hard to explain without an additional heat-generated orbital current in CuOₓ, converted to spin by platinum.

It is transport evidence — indirect, model-dependent, but compelling within its framework.

What does this change for spintronics?

Not tomorrow's phone, and the authors say so themselves. The voltages measured here serve as detection signals, not usable electrical power. But devices generate heat whether we like it or not, and converting some of that waste warmth into angular momentum could give spintronic systems an extra operating mechanism — essentially scavenging a resource they already produce.

"The orbital Hall effect has been intensively studied in recent years, while much less is known about its thermal counterpart, the orbital Nernst effect," Dr. Qu said. "Our work provides experimental evidence that a temperature gradient can generate orbital angular momentum transport. This finding extends our understanding of orbital transport from electrical to thermal driving and may offer new possibilities for spintronic applications."

Note the hedging: "may offer." One study, small voltages, layered model. Still, a first experimental sighting of a predicted effect — with a four-times-larger response than platinum's — is how new research directions start. The paper is Wen-Yuan Li et al., Transport evidence for the orbital Nernst effect, Nature Communications (2026), DOI: 10.1038/s41467-026-77878-2.