An Unintended Interface Made Mn3Sn Spin Signals Look More Than 100 Times Stronger
Researchers found that annealing Mn3Sn after adding platinum created an unintended ferromagnetic interface whose signal exceeded the antiferromagnet’s by more than 100 times. Changing the fabrication order exposed the material’s intrinsic response and measured spin-to-charge conversion at about one tenth of platinum’s efficiency.
Quick summary
- Heating a Mn3Sn and platinum stack created an unintended ferromagnetic compound at the interface.
- Its magnetic signal was more than 100 times stronger than the signal researchers wanted to measure.
- Annealing Mn3Sn before depositing platinum prevented the parasitic layer.
- The corrected samples converted spin current to voltage at roughly 10% of platinum's efficiency.
A fabrication step can sometimes become the experiment. Researchers at Martin Luther University Halle-Wittenberg and the Max Planck Institute for Microstructure Physics found that a standard heat treatment changed the interface between the chiral antiferromagnet Mn3Sn and platinum. The newly formed layer was ferromagnetic, and its response overwhelmed the signal attributed to Mn3Sn.
The result, reported in Advanced Materials and announced on 16 November 2026, is important because Mn3Sn is studied for ultrafast spintronics. Its manganese moments form a non-collinear, 120-degree arrangement. That chiral magnetic order can support spin-dependent transport while producing little external magnetic field, a useful combination for compact and interference-resistant devices.
The central finding is a correction to the measurement itself: the strongest signal did not belong to Mn3Sn, but to a compound unintentionally created while preparing the sample.
How the false signal appeared
The team deposited nanometre-thick Mn3Sn films and combined them with platinum or a nickel-iron alloy. Ultrashort laser pulses disturbed the magnetic order, and the researchers measured electrical voltages on picosecond timescales. In the platinum stack, however, the expected behavior did not appear cleanly.
Analysis showed that annealing the completed stack allowed atoms at the interface to react. The resulting ferromagnetic phase produced signals more than 100 times stronger than those of the antiferromagnet. This meant earlier-looking traces could be physically real while still describing the wrong material.
The researchers changed the sequence: they annealed and stabilized Mn3Sn first, then deposited platinum. That prevented the unwanted interlayer and exposed the intrinsic ultrafast response. In a separate nickel-iron and Mn3Sn structure, a spin current entering Mn3Sn was converted to an electrical voltage. The conversion efficiency was around 10% of the level typically obtained with platinum.
Why interfaces dominate spintronics
Spintronic devices move information using electron spin as well as electrical charge. Their active regions are often only a few atomic layers thick, so an interface is not a passive boundary. Atomic mixing, strain, oxidation or a new crystalline phase can define the measured response. A powerful signal is therefore not enough. Researchers must identify which layer generated it.
Reality check
This study does not demonstrate a finished memory chip or establish that Mn3Sn outperforms platinum. The corrected spin-to-charge conversion was about one tenth of platinum's typical efficiency. The value of the work is diagnostic: it identifies a major fabrication artifact and provides a process sequence that makes the intrinsic material response measurable.
The finding also warns against treating interface engineering as a final optimization step. In this experiment it determined whether the measurement described chiral antiferromagnetism or an accidental ferromagnet. Replication across different film thicknesses, substrates and deposition systems will be needed to show how general the corrected process is.
NewTqnia recently covered a rotating-field method that formed 1 to 5 nm crystals in soft magnetic alloys and an atomically thin photodiode used for 10 Gbit/s visible-light data. The new Mn3Sn result connects both themes: nanoscale structure and interfaces can determine whether an electronic material's apparent performance is genuine.
For future chiral spintronic components, the practical lesson is direct. Process order, thermal budget and interface chemistry must be measured alongside switching speed and voltage. Otherwise, a device may look exceptionally efficient because the intended material is not the one producing the signal.
Verified topics and entities
Sources and citations3 sources
External references used to support the reporting in this article.
Published by
NewTqnia Science Desk
An institutional editorial team within NewTqnia