Electrical detection of spin-flip transition in metal/ heterostructure

arXiv:2605.00491 · cond-mat.str-el · Submitted 2026-05-01 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: We're looking at the paper: "Electrical detection of spin-flip transition in metal/ heterostructure".

Mira: Here's the gist of the paper. The gist The longitudinal magnetoresistance (MR) in thin metal films on an Ising-type antiferromagnetic insulator, Na5Co15.5Te6O36 (NCTO), shows steep changes with hysteresis at spin-flip transitions driven by magnetic fields along the easy axis of the NCTO crystal Magnetic Properties of NCTO The target material of this study is an antiferromagnetic insulator, Na5Co15.5Te6O36 (NCTO), which is regarded as a model material for an Ising antiferromagnet.

Kai: NCTO possesses a hexagonal crystal structure belonging to the space group P63/m.

Mira: The magnetic Co ions are divalent and occupy three distinct crystallographic sites: Co(1) (blue), Co(2) (green), and Co(3) (cyan).

Kai: The N´eel temperature (TN) of NCTO is approximately 50 K, and the magnetic moments tend to align along the c-axis owing to the Ising anisotropy.

Title and authors: Mira: Moving on to the improvements suggested by this study, they are pointing towards using AI systems to better predict the magnetic structure changes in materials like Na5Co15 point 5Te6O36 when subjected to external magnetic fields along its easy axis <ref:2605.00491#pg1>.

Lev: That sounds like a big leap for simulation work. How does an AI system actually predict these complex structural changes based on what the authors observed? What kind of modeling are they talking about?

Kai: The AI system would be modeling the observed "sudden change with clear hysteresis in the magnetic-field dependence of the magnetization" to make accurate predictions. That's a big deal because it moves us from just observing data to predicting behavior.

Mira: And another improvement is that this improved AI system can distinguish between different physical mechanisms causing magnetoresistance anomalies, specifically identifying that "the origin of the MR anomalies is not interface spin accumulation but is attributed to the magnetostriction of NCTO" by comparing results from Pt/NCTO, Pt/SiOx /NCTO, and Cu/NCTO heterostructures.

Lev: That's useful for filtering out noise in experimental data. If an AI can tell the difference between magnetostriction and spin accumulation based on the structure of the response across those three different interfaces, that’s a way to validate our physical models more rigorously.

Kai: Another improvement they suggest is that this system can predict the sign change in high-field magnetoresistance slopes between different metal interfaces based on the magnetic ordering of the substrate. For instance, they found that "the slope for Pt/NCTO is positive, while it is negative for Cu/NCTO at five and zero point zero one" in the high-field regime.

Mira: That prediction about the sign change based on substrate magnetic ordering is interesting because it links the macroscopic measurement directly back to the microscopic physics of how those two different metal interfaces interact with NCTO.

Lev: If an AI can predict those slopes, then we can run simulations much faster for real hardware designs. We need to know if the expected resistance change will be positive or negative before we even start fabricating things.

Kai: Finally, they suggest the AI system can model and quantify the contribution of weak anti-localization, or WAL, to low-temperature magnetoresistance by estimating "the phase coherence length of Pt is approximately fifty-five nm for Pt/NCTO and fifty nm for Pt/SiOx /NCTO" based on the observed MR behavior <ref:2605.00491#pg2>.

Mira: Quantifying the phase coherence length based on MR behavior is a way to get a concrete physical parameter that we can use in our theoretical models, rather than just relying on fitting data curves.

Lev: Knowing those coherence lengths helps us design better error correction codes because we know the limits of how well those quantum states are protected by spin-orbit interactions in the Pt layer.

Kai: So, these improvements focus on using AI to go beyond observation and actually predict the underlying physics—predicting structure changes, distinguishing mechanisms, predicting slope signs, and quantifying phase coherence length.

The paper's summary: Kai: Wrapping up this paper "Electrical detection of spin-flip transition in metal/heterostructure," the main points are that high-field MR is dominated by weak anti-localization at low temperatures, while interface and surface roughness plays an important role in the negative MR in the intermediate temperature region.

Mira: They also emphasize that the electrical detection of spin-flip transition in metal/NCTO could contribute to the further development of antiferromagnetic spintronics because they successfully detected these magnetic properties electrically.

Lev: For us, this means we need to focus on integrating magnetostriction and interface roughness into our models when we are designing systems for real hardware. We can't ignore those effects just because the bulk material is an Ising antiferromagnet.

Kai: So, to summarize, the paper shows that the electrical detection of spin-flip transition in metal/NCTO has shown that weak anti-localization dominates at low temperatures and interface roughness is significant in the intermediate temperature region.

Mira: And they conclude that this research opens up avenues for using electrical detection of spin-flip transition in metal/NCTO to further develop antiferromagnetic spintronics. It’s a way to probe the underlying magnetic properties electrically without having to rely on other measurement techniques.

Lev: From an engineering standpoint, it means that if we want practical applications, we need reliable ways to model those strain effects and roughness contributions accurately within our system design pipeline.

Kai: So that's the whole picture of this paper "Electrical detection of spin-flip transition in metal/heterostructure." We have seen how it can be used to probe magnetic properties electrically.

Mira: And we've seen how it connects strain, interface roughness, and low-temperature effects all through the longitudinal magnetoresistance in this NCTO system.

Lev: It's a solid piece of work for experimental physics, but for us, the real value is in translating these complex findings into reliable models that can guide hardware development.

The paper's improvements: Kai: So we just finished talking about how this paper used electrical measurements to see magnetic transitions in Na5Co15 point 5Te6O36, and now we're looking at what they suggest for improving that work with AI.

Mira: Right, the improvements they propose are really about using AI to get better at predicting those magnetic structure changes when you apply a magnetic field along the easy axis of the material.

Lev: That sounds like an attempt to model that sudden change with clear hysteresis in the magnetization dependence on a real hardware level.

Kai: Exactly, and it moves us from just observing that jump to actually having an AI system predict what's going to happen before we even run the experiment. It's about modeling those specific magnetic structure changes accurately.

Mira: And another improvement they highlight is using that improved AI system to tell the difference between various physical mechanisms causing magnetoresistance anomalies.

Lev: They found this is important because the original paper attributed those anomalies to magnetostriction, which means changing the size of the material under a magnetic field, not just spin accumulation at an interface.

Kai: So, if the AI can compare results from Pt/NCTO, Pt/SiOx /NCTO, and Cu/NCTO heterostructures and correctly identify that magnetostriction is happening instead of interface spin accumulation, that's a big validation for our physical models.

Mira: Plus they suggest the AI system can predict the sign change in those high-field magnetoresistance slopes between different metal interfaces based on how the substrate is magnetically ordered.

Lev: That’s interesting because it means we can predict whether an interface will show a positive or negative MR slope just by knowing the magnetic state of the NCTO underneath. It’s about predicting outcomes before fabrication even starts.

Kai: Then there's quantifying the contribution of weak anti-localization, or WAL, to that low-temperature magnetoresistance using AI.

Mira: They estimate things like the phase coherence length for Pt as approximately fifty five nanometers for Pt/NCTO and fifty nanometers for Pt/SiOx /NCTO based on what they saw in the MR behavior.

Lev: Quantifying that phase coherence length is useful because we can actually use those numbers to design better error correction codes because we know the limits of how well those quantum states are protected by spin-orbit interactions in the Pt layer.

Kai: So it’s moving from just seeing a result to getting concrete physical parameters like those coherence lengths that you can plug into your simulation code.

Mira: And they also suggest using AI to predict the role of interface roughness in determining negative magnetoresistance in Cu/NCTO.

Lev: They link that negative MR directly to scattering caused by interface roughness, and they say this effect is significant in the intermediate temperature region. That’s a useful piece of information for us when we design interfaces for real systems.

Kai: So, these improvements take what was found experimentally—the MR jumps and the slope differences—and use AI to predict the physics underneath them, distinguishing between strain effects and surface roughness effects.

Mira: It's about building a predictive tool that connects those microscopic magnetic changes in NCTO directly to the macroscopic electrical response we measure in these metal bilayers.

Lev: If we can get that level of prediction, it drastically cuts down on the experimental time needed to figure out what's actually going on at a material interface.

Kai: So, what this means for us is that understanding the magnetic properties of materials like NCTO isn't just about measuring them; it’s about building predictive models using AI that can tell us exactly how those magnetic properties influence the electrical signals we get from our quantum devices.

Conclusion: Kai: So to wrap up, this paper on "Electrical detection of spin-flip transition in metal/heterostructure" shows that we can actually probe magnetic behavior electrically in these thin films using things like longitudinal magnetoresistance.

Mira: Exactly, and it’s interesting because it connects the magnetic structure changes happening inside the Na5Co15 point 5Te6O36 to measurable electrical signals, which is a pretty neat way to look at condensed matter physics.

Lev: It definitely gives us some concrete parameters like the phase coherence lengths for Pt films, which is useful for figuring out what those quantum states are actually doing in the hardware we build.

Kai: Right, and they've also highlighted that interface roughness plays a role in negative magnetoresistance at intermediate temperatures, which is something we need to keep in mind when designing those metal interfaces.

Mira: So what this means for us is that we can use electrical detection of spin-flip transition in metal/heterostructure to get real physical insights into how strain and disorder affect magnetic ordering.

Lev: It gives us a way to validate our simulation assumptions about strain effects by seeing if the measured MR jumps match the predicted magnetostriction coefficients.

Kai: And it confirms that even in complex antiferromagnetic insulators like NCTO, there are clear, measurable magnetic transitions driven by external fields along the easy axis.

Mira: It opens up a path for using electrical detection of spin-flip transition in metal/heterostructure to further develop antiferromagnetic spintronics because we can actually see the magnetism electrically.

Lev: For real hardware, it means we need to build better models that incorporate both magnetostriction and scattering from roughness when designing those metal layers.

Kai: We’re ready for the next paper on arXiv, which is about detecting channel-substitution attacks on quantum machine learning pipelines in untrusted clouds.

Mira: That sounds like a completely different field, moving into security and quantum computing applications.

Department of Basic Science, The University of Tokyo

cond-mat.str-el

Submitted: 2026-05-01

Updated: 2026-08-18

Comments: 8 pages, 4 figures

DOI: 10.1103/b1d1-ls28

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 74/100

The gist: The gist The longitudinal magnetoresistance (MR) in thin metal films on an Ising-type antiferromagnetic insulator, Na5Co15.5Te6O36 (NCTO), shows steep changes with hysteresis at spin-flip transitions

Key concepts

Ising Antiferromagnet: Na5Co15.5Te6O36 (NCTO) is an antiferromagnetic insulator that behaves like an Ising model. Its magnetic moments prefer to align along the c-axis due to strong magnetic anisotropy, making it a key material for studying spin dynamics under external fields.
Magnetostriction: This effect describes how the physical dimensions of the metal film change when the underlying NCTO crystal structure undergoes a magnetic transition. This structural change in NCTO alters the resistance of the adjacent metal layers, leading to measurable changes in magnetoresistance.
Weak Anti-Localization (WAL): WAL is a quantum interference effect seen in Pt films at low temperatures. It arises from strong spin-orbit interaction, causing electrons to scatter back and forth coherently, which leads to a positive contribution to the magnetoresistance observed below 50 K.
Interface Roughness Scattering: Roughness on the surface of the NCTO crystals creates scattering centers at the metal interfaces. This scattering is identified as a primary cause for negative magnetoresistance in Cu/NCTO, especially in intermediate temperature regions where electron-phonon scattering is suppressed.

Terminology

Hysteresis — 1× in this paper · explained in 3 episodes

Summary

The gist The longitudinal magnetoresistance (MR) in thin metal films on an Ising-type antiferromagnetic insulator, Na5Co15.5Te6O36 (NCTO), shows steep changes with hysteresis at spin-flip transitions driven by magnetic fields along the easy axis of the NCTO crystal

Magnetic Properties of NCTO

The target material of this study is an antiferromagnetic insulator, Na5Co15.5Te6O36 (NCTO), which is regarded as a model material for an Ising antiferromagnet. NCTO possesses a hexagonal crystal structure belonging to the space group P63/m. The magnetic Co ions are divalent and occupy three distinct crystallographic sites: Co(1) (blue), Co(2) (green), and Co(3) (cyan). The N´eel temperature (TN) of NCTO is approximately 50 K, and the magnetic moments tend to align along the c-axis owing to the Ising anisotropy. At TN, the Co(1) sublattice becomes a partially disordered antiferromagnetic (PDA) state, where one of the Co(1) spins in each Co(1) triangle remains disordered. As temperature further decreases, the disordered moments become frozen, and a frozen PDA (FPDA) state emerges below approximately 20 K. When strong magnetic fields are applied from zero field along the c-axis, the magnetic moments of Co(1) flip in the magnetic-field direction, and the induced ferromagnetic states are stabilized. The spin-flip transition is accompanied by a sudden change with clear hysteresis in the magnetic-field dependence of the magnetization. The hysteresis during the spin-flip transition is notably large at very low temperatures, which is a unique magnetic property of this material associated with the spin-frozen state.

Magnetoresistance Mechanisms and Observations

The study investigated longitudinal MR, measured under a magnetic field applied parallel to the electric current, for metal/NCTO bilayers. Steep changes in the MR spectra with hysteresis were observed during the spin-flip transition in both Pt/NCTO and Cu/NCTO. The similar MR jumps observed for Pt/NCTO, Pt/SiOx /NCTO, and Cu/NCTO indicate that the origin of the MR anomalies is attributed to the magnetostriction of NCTO rather than interface spin accumulation. A notable difference in the MR spectra between Pt and Cu was observed in the high-field regime, where magnetic moments were fully aligned along the magnetic-field direction. The opposite signs of the slopes of the MR for Cu/NCTO and Pt/NCTO are prominent below ∼ 50 K, suggesting a weak antilocalization effect due to the strong SOI of the Pt films.

Role of Strain and Interface Effects

The MR anomalies at the spin-flip transition are attributed to strain effects via magnetostriction upon the magnetic-structure change of the Co nets in NCTO, because similar MR jumps are observed in Pt/NCTO, Pt/SiOx /NCTO, and Cu/NCTO heterostructures. The dimensions of the metal/NCTO heterostructure can slightly change at the spin-flip transition via magnetostriction, which affects the resistance of the metal films. The magnetostriction coefficient λ was estimated to be 1.4 × 10−4 from the 10 K data. This value is more than an order of magnitude larger than the reported values at the spin-flop transition for NiO [23] and Cr2O3 [12], but is an order of magnitude smaller than that reported for MnTe [24].

Distinguishing Contributions in High Fields

In the high-field regime, the different MR spectra between Pt/NCTO and Cu/NCTO are associated not with the magnetic structure change of NCTO, but with the material of the metal layers. The slope for Cu/NCTO is positive at 150 K but gradually decreases and becomes negative below about 60 K. The difference in the MR slope between Pt/NCTO and Cu/NCTO becomes prominent below TN(=48.6 K), suggesting that the magnetic order of NCTO is related to the different MR slopes between Pt/NCTO and Cu/NCTO. The positive MR contribution in Pt/NCTO below 50 K is attributed to weak anti-localization (WAL).

Negative Magnetoresistance Analysis

The negative MR observed in Cu/NCTO below TN is likely ascribed to scattering due to interface roughness. Atomic force microscopy observations revealed that the surface root-mean-square roughness of the as-grown NCTO crystals was at most 4 nm. Roughness-induced scattering at the interfaces and the film surfaces could contribute to the negative MR, especially at low temperatures where electron–phonon scattering is suppressed and the mean free path becomes long. The resistivity of the Cu film is significantly higher than the bulk value (∼ 1 µomega cm [35]) which may be consistent with strong roughness scattering as the origin of the negative MR in Cu/NCTO.

Conclusion

The successful electrical detection of the magnetic properties of an underlying Ising antiferromagnet could contribute to the further development of antiferromagnetic spintronics. The study concludes that the high-field MR in Pt/NCTO is dominated by weak anti-localization at low temperatures, whereas interface/surface roughness plays an important role in the negative MR in the intermediate temperature region. The electrical detection of the magnetic properties of an underlying Ising antiferromagnet could contribute to the further development of antiferromagnetic spintronics.

--- Page 1 ---

Electrical detection of spin-flip transition in metal/Na5Co15.5Te6O36 heterostructure Hirotsugu Tagami,∗ Takuya Kawada, and Yuki Shiomi Department of Basic Science, The University of Tokyo, Tokyo 153-8902, Japan (Dated: August 19, 2026 We report on the longitudinal magnetoresistance (MR) in thin metal films on an Ising-type antiferromagnetic insulator, Na5Co15.5Te6O36 (NCTO) Steep changes in the MR spectra with hysteresis were observed at spin-flip transitions driven by magnetic fields applied along the easy axis of the NCTO crystal The MR jumps almost follow step-like changes in magnetization at the spin-flip transition At very low temperatures where Co moments are partially frozen, the MR anomalies exhibit a tunnel-magnetoresistance-like shape The observed MR anomalies at the spin-flip transition are attributed to strain effects via magnetostriction upon the magnetic-structure change of the Co nets in NCTO because similar MR jumps are observed in Pt/NCTO, Pt/SiOx /NCTO, and Cu/NCTO heterostructures Interestingly, we found that the high-field slopes of the MR spectra show opposite signs between Pt/NCTO and Cu/NCTO at low temperatures Because similar MR spectra are observed for Pt/SiOx /NCTO in which the interface magnetic interaction is negligible, the weak antilocalization due to the strong spin-orbit interaction of the Pt films is likely to contribute to the low-temperature MR

--- Page 2 ---

2 In this study, we have investigated longitudinal MR, that is MR measured under a magnetic field applied parallel to the electric current, for metal/NCTO bilayers.

Improvements for AI systems

Improved AI systems can accurately predict magnetic structure changes in materials like Na5Co15.5Te6O36 when subjected to external magnetic fields along its easy axis by modeling the observed sudden change with clear hysteresis in the magnetic-field dependence of the magnetization.
The improved AI system can distinguish between different physical mechanisms causing magnetoresistance anomalies, specifically identifying that the origin of the MR anomalies is not interface spin accumulation but is attributed to the magnetostriction of NCTO by comparing results from Pt/NCTO, Pt/SiOx /NCTO, and Cu/NCTO heterostructures.
The system can predict the sign change in high-field magnetoresistance slopes between different metal interfaces based on the magnetic ordering of the substrate, noting that the slope for Pt/NCTO is positive, while it is negative for Cu/NCTO at 5 and 0.01 in the high-field regime.
The AI can model and quantify the contribution of weak anti-localization (WAL) to low-temperature magnetoresistance by estimating the phase coherence length of Pt is approximately 55 nm for Pt/NCTO and 50 nm for Pt/SiOx /NCTO based on the observed MR behavior.
The system can predict the role of interface roughness in determining negative magnetoresistance in Cu/NCTO by linking it to scattering due to interface roughness and noting that this effect is significant in the intermediate temperature region.