Controversy over the surface-oxidized Cu goes back to 2016, where An et al. from Keio University published a paper that demonstrates significantly large spin-orbit torque as Cu/Py/substrate structure becomes “naturally” oxidizied from the Cu surface. This seems like a typical setup to characterize self-induced spin-orbit torque in Py assuming that Cu doesn’t induce any significant effect. But the result was surprising in that the surface-oxidized Cu seems to affect drastically on the spin-orbit torque. The measured torque efficiency is more or less comparable to heavy metals with large spin Hall effect such as Pt, W, Ta.

Since the beginning, I have thought it might be due to “orbital current”, which doesn’t necessarily require spin-orbit interaction (note that spin-orbit interaction in Cu is negligible compared to that of heavy elements). However, demonstrating this idea with an experiment wasn’t very clear a few years ago. Despite this, I published a paper on arXiv with Dr. J. Junyeon Kim from RIKEN-CEMS claiming that surface-oxidized Cu generates orbital current from the interface. In this work, we noticed that the torque efficiency depends critically on the interface quality, making ~ 100 times difference for different interfaces. Since the orbital current interacts directly with the lattice structure, such anomalous feature can be explained, at least qualitatively. Meanwhile, shortly after, Keio group also published a paper on arXiv from the observation that the sign of the spin-orbit torque changes for different ferromagnets, which is an expected behavior of “orbital torque”.

But the problem is that these experiments are still indirect evidences of the orbital current. I think this is a difficulty in studying transport in metallic systems. Too many things can happen. In this sense, cleaner systems like 2D materials might be better, where comparison of the theory and experiment is relatively straightforward due to simpler band structure described by a few orbital bases.

Recently, from the collaboration with Kläui-Lab in JGU Mainz, we publisehd a paper on PRL. We got an idea that if the orbital current is generated from the surface-oxidized Cu, then insertion of a Pt layer would efficiently convert the orbital current into spin current, which we can utilize for torque generation in a ferromagnet (we used an insulating magnet TmIG). The experiment clearly shows that a very large amount of spin current is generated after Pt-insertion, which goes beyond the amount that we expect from bare Pt. Also, by considering all possible mechanisms that might happen, we found that the orbital current mechanism is the only one that can consistently explain the experiment (see Supplementary Information).

I feel like the mysteries of surface-oxidized Cu are starting to be uncovered, finally after 4-5 years. I hope we find a more “direct” evidence of the orbital current in this system in the coming years.

Harnessing Orbital-to-Spin Conversion of Interfacial Orbital Currents for Efficient Spin-Orbit Torques
Shilei Ding, Andrew Ross, Dongwook Go, Lorenzo Baldrati, Zengyao Ren, Frank Freimuth, Sven Becker, Fabian Kammerbauer, Jinbo Yang, Gerhard Jakob, Yuriy Mokrousov, and Mathias Kläui
Phys. Rev. Lett. 125, 177201 (2020)