Electrical magnon spectroscopy with foundry-fabricated nanoscale magnetic tunnel junctions

arXiv:2509.19483 · cond-mat.mes-hall · Submitted 2025-09-23 · Read on arXiv

Listen

Transcript

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

Kai: I'm Kai, and with me are Mira and Lev, guest researcher.

Mira: Today's paper: "Electrical magnon spectroscopy with foundry-fabricated nanoscale magnetic tunnel junctions".

Kai: Electrical detection of spin waves using nanoscale magnetic tunnel junctions demonstrates a novel method for reading out information encoded in spin waves using CMOS-compatible technology.

Mira: First, who's behind it and why it matters.

Paper summary: Kai: So, we're looking at Christopher Heins' paper, "Electrical magnon spectroscopy with foundry-fabricated nanoscale magnetic tunnel junctions." It seems like the big idea here is using CMOS technology to read out information encoded in spin waves. Mira, could you give us the quick rundown on what this whole thing is actually trying to achieve and why it matters?

Mira: Absolutely, Kai. The paper focuses on addressing the challenge of electrically detecting magnons, which are those collective excitations of magnetic moments in magnetically ordered materials. The core thesis is engineering a dynamic coupling between these spin waves and the magnetization state of a nanoscale magnetic tunnel junction or MTJ cell fabricated within a standard CMOS production line. It claims they can achieve high fidelity in transducing these spin-wave excitations into measurable electrical signals, even revealing spectral line widths down to a few hundred kHz.

Lev: From my angle as someone who thinks about running this on real hardware, the feasibility of integrating this with existing semiconductor processes is a huge hurdle for any quantum error correction researcher. If you can't reliably control and measure the magnonic states electrically, building robust systems becomes incredibly complex because you lose that clean experimental handle.

Kai: Right, so they're taking something theoretical—magnons—and trying to make them something tangible and measurable using existing chip fabrication techniques. What specifically makes this work with those nanoscale MTJs?

Mira: The paper details the system setup, showing a modification of an MRAM device fabricated on a three hundred-mm wafer at GlobalFoundries. They engineered the structure by modifying the MgO-free layer interface and increasing the free layer thickness compared to typical MRAM stacks, which changes how perpendicular magnetic anisotropy behaves so it prefers lying in the wafer plane. This customization turns those standard memory cells into nanoscale field sensors where the orientation of that free layer is determined by a local magnetic field.

Lev: If they're engineering that interface and thickness to modify the anisotropy, we need to be careful about how much control they really have over those parameters during manufacturing; any variation in fabrication could introduce noise that drowns out these subtle magnon signals.

Kai: That makes sense, so the physical structure is tailored specifically for this sensing job. How do they actually generate the magnons and then translate that motion into a measurable electrical change in that MTJ?

Mira: They model the magnonic system using a ferromagnetic Ni81Fe19 disk, which acts as a cavity for resonances characterized by radial and azimuthal mode numbers, denoted as (n, m). These magnons are excited by an-shaped microwave antenna surrounding the disk to create oscillatory magnetic fields. The crucial part is separating the output signal from the direct input; they circumvent this by exploiting nonlinear interactions within a vortex-state disk, specifically three-magnon splitting. This process allows them to isolate a pure magnon signal because it only occurs above a well-defined excitation amplitude of the initially excited mode.

Paper summary: Lev: Three-magnon splitting is certainly a sophisticated mechanism for isolating the desired signal from the excitation source; though, on real hardware, achieving that specific nonlinear interaction cleanly at the required amplitude without introducing other unwanted coupling channels would be extremely difficult to guarantee.

Kai: And then once they have that pure magnon signal, how does it actually get into those MTJs and create a readable electrical output? That’s where the transduction mechanism has to happen.

Mira: The detection relies on the fact that when magnons are excited, they generate oscillating dipolar fields both below and above the sample surface. If these dipolar magnetic fields influence the free layer magnetization of the MTJ cell, then transduction occurs. This oscillation results in a measurable change in resistance because precession generates parallel and antiparallel components of change in magnetization, denoted as plus or minus m z, with respect to the reference layer.

Lev: So they are relying on the physical proximity and coupling between that external spin-wave field and the magnetic state of a nanoscale device; that coupling strength is going to be extremely sensitive to fabrication imperfections at the interface.

Kai: And they use what kind of equipment for reading this out? How do we actually see these tiny resistance changes?

Mira: They connect the MTJ cell to a spectrum analyzer via high-frequency ground-signal-ground probes. The electrical spectra resulting from three-magnon splitting show that magnons are only faintly visible compared to magneto-optic measurements, which the authors attribute to the narrow linewidth of the magnon signal measured with that analyzer. Specifically, they observe a spectral line width of "only one hundred eighty-eight kHz at a magnon frequency of three point five nine GHz."

Lev: An one hundred eighty-eight kHz linewidth is quite narrow for this kind of measurement; from an error correction standpoint, that narrowness suggests a potentially high cooperativity in the hybrid system, which is what they mentioned regarding strong coupling regimes.

Kai: That's what excites me about the potential for quantum applications here. So, looking at the overall picture of "Electrical magnon spectroscopy with foundry-fabricated nanoscale magnetic tunnel junctions," what do you think this work implies for future technology?

Mira: The proof-of-concept successfully connects state-of-the-art wafer production technology with lab spintronic ideas, establishing MRAM as a platform for hybrid spintronic and magnonic applications that are scalable and energy efficient within integrated circuits. Furthermore, the sub-MHz linewidth observed in electrical measurements suggests that magnons can be used as quantum transducers, which would significantly boost the cooperativity of hybrid magnon-qubit systems into those strong coupling regimes.

Paper summary: Lev: If they can demonstrate this level of electrical detection fidelity on a foundry line, it moves magnonics out of purely academic simulation and puts it squarely in the realm where hardware implementation becomes possible for certain types of information processing tasks.

Kai: It really does feel like they've taken a very abstract concept—spin waves—and built a concrete pathway to measure them using things that are already being mass-produced in semiconductor plants. That’s quite an achievement, Mira.

Mira: Precisely; the title itself, "Electrical magnon spectroscopy with foundry-fabricated nanoscale magnetic tunnel junctions," highlights that the innovation isn't just in the physics of magnons, but in successfully coupling that physics to a manufacturing reality using these specific MTJ structures.

Lev: For real hardware deployment, we need to consider how reliable those measurements are when scaling up beyond a single cell and how the noise floor from the CMOS environment will affect those extremely narrow lines they measured.

Kai: We'll have to see what the next steps are regarding scaling that detection capability up, because if this works reliably on a chip, we could start thinking about integrated spin-wave processors.

Mira: Yes, and we need to keep probing those underlying assumptions—especially how that three-magnon splitting works robustly across different materials used in the fabrication process.

Lev: I'd be interested to see if they can show stability over longer operational times, because for error correction, long coherence is everything.

Kai: Well, that brings us to the conclusion of this discussion on "Electrical magnon spectroscopy with foundry-fabricated nanoscale magnetic tunnel junctions." It’s a demonstration that hybrid spintronic systems are becoming a tangible reality within existing semiconductor infrastructure.

Mira: Indeed, the implication is that we are moving closer to realizing scalable, energy-efficient spin-wave processing and communication directly inside integrated circuits.

Lev: For researchers working on error correction, this suggests a viable physical mechanism for coupling quantum states via magnons that could eventually be implemented in silicon or CMOS platforms.

Kai: It really shows the convergence of different fields; we see the results of sophisticated physics being measured using technology that is already on the production line.

Mira: That's what makes this work compelling; it validates MRAM technology as a viable platform for these hybrid systems, opening pathways for scalable spin-wave processing within integrated circuits.

Lev: So, the path forward involves rigorous testing to see if those observed electrical signatures can be maintained and leveraged in larger, more complex devices.

Kai: That’s what we need to watch next; seeing how this works under more demanding conditions is key to moving past this initial proof-of-concept.

Conclusion: Kai: So, we've just finished looking at Christopher Heins' paper, "Electrical magnon spectroscopy with foundry-fabricated nanoscale magnetic tunnel junctions," which details how they built and measured spin waves using a standard semiconductor process. Mira, what are your thoughts on the title and the authors?

Mira: I think the title really zeroes in on the core innovation here, which is bridging that gap between fundamental magnonic physics and actual foundry fabrication techniques. The authors’ choice of terms highlights that they're not just doing a simulation; they're proving something can be done at scale with existing infrastructure.

Lev: From my side, I'm thinking about what this title means for hardware implementation because it suggests the physics isn't just theoretical anymore; it has a concrete fabrication pathway. It points toward creating real devices, not just models.

Kai: Exactly, Lev, and I think that’s where the excitement is—seeing those abstract spin-wave phenomena become something you can physically build and measure with standard tools. What about the authors themselves? What do they bring to this specific study?

Mira: The authors are combining expertise in condensed matter theory with hands-on experimental setup, which is crucial because it means they aren't just applying a known model; they're actively engineering the system to work. This dual perspective is what makes their methodology so compelling.

Lev: And from an error correction viewpoint, having authors who understand both the device physics and the underlying quantum mechanics of spin waves is vital for developing protocols that actually work on real hardware later on.

Kai: It sounds like this paper sets a new benchmark for experimental work where you can see direct results from a foundry line, which is pretty impressive. So, what are the big takeaways we should be hearing about regarding the world's impact?

Mira: The main implication is that MRAM technology isn't just for storage anymore; it’s being positioned as a platform for hybrid spintronic applications that can actually be scaled up and made energy-efficient within integrated circuits.

Lev: If they can maintain that electrical detection fidelity while scaling, it opens the door to building integrated spin-wave processors, which would be a major step toward realizing scalable quantum communication systems.

Kai: That really puts things into perspective; we're moving from small lab demonstrations to something that could genuinely be embedded in future chips. So, where do we go from here in this exciting area?

Mira: We need to keep focusing on those assumptions regarding how robust the three-magnon splitting interaction is across different fabrication materials, because that’s where the theoretical rigor needs to stay sharp.

Lev: I agree, and I'll be looking closely at their discussion on long-term stability and noise resilience, since real hardware needs coherence over time to do anything useful in quantum computing.

Helmholtz-Zentrum Dresden–Rossendorf Institute für Ionenstrahlphysik und Materialforschung, Technische Universität Dresden, Centre de Nanosciences et de Nanotechnologies, CNRS, GlobalFoundries Dresden Module One LLC & Co. KG

cond-mat.mes-hall

Submitted: 2025-09-23

Updated: 2026-10-01

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

Importance score: 79/100

The gist: Electrical detection of spin waves using nanoscale magnetic tunnel junctions demonstrates a novel method for reading out information encoded in spin waves using CMOS-compatible technology.

Key concepts

Magnons: Magnons are collective excitations of magnetic moments in a material, like spin waves. In this study, they were excited using a microwave antenna around a Ni81Fe19 disk. The key challenge was separating the magnon signal from the original microwave input.
Magnetic Tunnel Junction (MTJ): An MTJ is a nanoscale device consisting of two ferromagnetic layers separated by an insulating barrier, similar to MRAM. Here, it acts as a sensor where the orientation of one layer's magnetization is influenced by external magnetic fields generated by the magnons.
Three-Magnon Splitting: This is a nonlinear interaction between magnons within a vortex-state disk. It allows researchers to isolate the pure magnon signal from direct microwave input because this splitting only occurs above a specific excitation amplitude, providing a clean way to measure the spin wave's properties.

Terminology

Analyzer — 12× in this paper · explained in 1 episode
Anisotropy — 6× in this paper · explained in 3 episodes
Fidelity — 6× in this paper · explained in 2 episodes
Core — 5× in this paper · explained in 2 episodes
Aperture — 2× in this paper · explained in 1 episode
Hybrid System — 2× in this paper · explained in 2 episodes
Alignment — 1× in this paper · explained in 1 episode
Benchmark — 1× in this paper · explained in 2 episodes
Benchmarking — 1× in this paper · explained in 3 episodes
Quantum Circuits — 1× in this paper · explained in 2 episodes
Verification — 1× in this paper · explained in 2 episodes

Summary

Electrical detection of spin waves using nanoscale magnetic tunnel junctions demonstrates a novel method for reading out information encoded in spin waves using CMOS-compatible technology. This work addresses the challenge of electrically detecting magnons, which are collective excitations of magnetic moments, by engineering a dynamic coupling between these spin waves and the magnetization state of a nanoscale magnetic tunnel junction (MTJ) cell fabricated within a state-of-the-art CMOS production line.

The gist

By engineering the dynamic coupling between spin waves and the magnetization state of the MTJ, this research demonstrates transduction of spin-wave excitations into measurable electrical signals with high fidelity, revealing spectral line widths down to a few hundreds of kHz.

System and Fabrication

The study utilizes a nanoscale magnetic tunnel junction (MTJ) cell, specifically a modification of a magnetoresistive random-access memory (MRAM) device, fabricated on a 300-mm wafer in the state-of-the-art CMOS production line at GlobalFoundries. The MTJ comprises two metallic ferromagnetic layers separated by a nonmagnetic tunnel barrier. This structure is engineered to serve as a sensitive magnetic field sensor where the orientation of the free magnetic layer of the MTJ is determined by a local magnetic field, while the second layer remains pinned (reference layer). The MRAM cells are customized to perform as nanoscale field sensors by engineering the MgO-free-layer interface and increasing the free layer thickness compared to standard MRAM film stacks, modifying perpendicular magnetic anisotropy so its magnetization prefers to lie within the wafer plane.

Magnon Excitation and Signal Separation

The magnonic system is modeled using a ferromagnetic Ni81Fe19 disk with 5 µm diameter and 50 nm thickness, which acts as a cavity for magnon resonances characterized by radial and azimuthal mode numbers (n, m). Magnons are excited by an omega-shaped microwave antenna fabricated around the disk to generate oscillatory magnetic fields. A crucial challenge addressed is the separation of output signals from direct input; this is circumvented by exploiting nonlinear interactions between magnons in the vortex-state disk, specifically three-magnon splitting. This process allows for separating a pure magnon signal from the direct microwave input because it is strongly nonlinear and only occurs above a well-defined excitation amplitude of the initially excited magnon mode.

Electrical Readout Mechanism

The detection mechanism relies on the fact that when magnons are excited, they create oscillating dipolar fields below and above the sample surface. If these dipolar magnetic fields influence the free layer magnetization of the MTJ cell, transduction is achieved. The magnetoresistance oscillates with the magnon frequency because precession generates parallel and antiparallel components of change in magnetization (±∆mz) with respect to the reference layer. This oscillation results in a measurable change in resistance, which can be detected using a spectrum analyzer connected to the MTJ via high-frequency ground-signal-ground (GSG) probes.

Spectral Characterization and Performance

Electrical characterization was performed by recording power spectra by contacting the MTJ cell to a spectrum analyzer. The electrical spectra resulting from three-magnon splitting show that magnons are only faintly visible in the electrical spectra compared to magneto-optic measurements, which is attributed to the narrow linewidth of the magnon signal measured with the spectrum analyzer. The resulting spectral line width observed in electrical measurements is only 188 kHz at a magnon frequency of 3.59 GHz. Furthermore, monitoring magnons over time reveals clear oscillations in the range of a few megahertz, and while individual split modes exhibit shifts, the sum of their frequencies remains constant as a function of time.

Conclusion and Outlook

The proof-of-concept successfully bridges state-of-the-art wafer-scale semiconductor production technology with lab-based spintronic ideas. The results establish MRAM technology as a viable platform for hybrid spintronic-magnonic applications, paving the way for scalable, energy-efficient spin-wave processing and communication within integrated circuits. The sub-MHz linewidth observed in electrical measurements boosts the cooperativity of hybrid magnon-qubit systems into the strong coupling regime, suggesting magnons can be used as quantum transducers.

Author Contributions

H.S., K.S., and C.H. conceived the experiments; K.S., Z.X., R.N., A.T., and J.M. fabricated the samples; C.H carried out the experiments; T.D assisted with the experimental setup; J.-V.-K and A.K performed micromagnetic simulations; R-H and A-W prepared and recorded SEM and TEM images; C.H., J-V.-K, K.S., and H.S visualised the results; all authors analyzed the data and discussed the results. H.S., K.S., and J-V.-K wrote the original draft of the paper; all authors reviewed and edited the paper.

Improvements for AI systems

As a fastidious researcher, I have analyzed this paper, Electrical detection of magnons with nanoscale magnetic tunnel junctions, and identified several high-leverage areas where its core findings—specifically, the electrical transduction of spin waves into measurable signals via CMOS-integrated MRAM sensors—can be directly applied to improve AI systems.

Here are the specific improvements and capabilities for an advanced AI system:

) Improved AI System Capabilities

The primary improvement lies in developing a new class of neuromorphic hardware and sensing architectures that exploit magnonic dynamics for ultra-low power, high-speed computation and sensory feedback. Specifically, the improved system can perform:

[] Use the electrical detection mechanism to create a novel, highly sensitive sensor layer integrated directly onto or adjacent to conventional CMOS chips.
[] Implement a low-power Magnon Sensor Module capable of detecting subtle changes in magnetic state dynamics (spin waves) within nanoscale memory arrays (MRAM).
[] Utilize the non-linear three-magnon splitting phenomenon as a mechanism for signal purification, allowing the system to extract pure magnon information from noisy microwave excitation signals.
[] Develop Magnon-Enhanced Analog Circuits where the free layer magnetization of an MTJ acts as a tunable element whose resistance state is modulated by magnon frequency/amplitude, enabling ultra-fast analog tuning of digital logic gates.

) Specific Enhancements to AI System Architecture

The paper suggests several concrete architectural shifts:

[] Create a Magnonic Readout Layer for in-memory computing: Integrate the MRAM cell structure (CoFeB/MgO) not just as storage, but as a dynamic sensor layer. This layer can be used to perform real-time, high-resolution monitoring of the local magnetic environment (e.g., detecting localized spin wave patterns generated during complex neural network inference).
[] Design a Frequency-Selective Filter for signal processing: Leverage the nonlinear three-magnon splitting to design computational blocks that inherently separate desired signal frequencies (the split modes) from broadband noise and direct excitation signals, leading to higher fidelity in complex spintronic computations.
[] Implement Energy-Efficient Spin Transducers: Replace power-hungry charge-based readout circuits with the electrical readout of the MTJ resistance change induced by magnon precession. This allows AI accelerators to operate at significantly lower energy costs, crucial for edge computing and mobile AI devices, as suggested by the paper's focus on overcoming limitations of charge-based electronics.
[] Develop Magnetically Tunable Logic Gates: Use the free layer's susceptibility to out-of-plane fields (as demonstrated in Fig. 1g) to create logic elements whose switching behavior is modulated by the presence or frequency of a local magnon field, effectively creating a hybrid spintronic/electronic logic gate.

) Specific Applications in AI Systems

The improved system can be applied to:

[] Edge AI Accelerators: Deploy the low-power, high-fidelity sensors on edge devices to monitor environmental magnetic fields or biological spin dynamics (e.g., in magnetoencephalography or neuromorphic sensing).
[] Neuromorphic Computing Primitives: Utilize the nonlinear magnon dynamics (three-magnon splitting) to implement biologically inspired computational tasks like pattern recognition and associative memory, where the signal processing inherently mirrors biological systems.
[] Quantum Transducer for Qubit Control: Given the observation that a 200 kHz linewidth boosts cooperativity, this system can be used as a highly sensitive quantum transducer to couple magnonic states directly to spin qubits (as hinted in the Outlook section), facilitating scalable quantum computing architectures.
[] High-Resolution Magnetic Imaging Sensors: Use the ability to resolve spatial features (down to hundreds of nanometers) through the detection of dipolar fields generated by magnon modes, leading to nanoscale magnetic field mapping capabilities beyond current limitations.