Nonlinear dynamics in magnonic Fabry-P' e rot resonators: Low-power neuron-like activation and transmission suppression

arXiv:2602.10650 · cond-mat.mtrl-sci, cond-mat.mes-hall · Submitted 2026-02-11 · Read on arXiv

Listen

Transcript

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

Kai: Today's paper: "Nonlinear dynamics in magnonic Fabry-P' e rot resonators".

Mira: The gist: The study reports that nonlinear dynamics in magnonic Fabry-Pérot resonators enable low-power neuron-like activation and frequency-selective nonlinear spin-wave absorption, highlighting their potential for neuromorphic magnonic computing architectures.

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

Paper summary: Kai: So we’re looking at this paper titled "Nonlinear dynamics in magnonic Fabry-Pérot resonators: Low-power neuron-like activation and transmission suppression". It’s about YIG films coupled to CoFeB nanostripes, and what they do when you push the excitation power.

Mira: Basically, the thesis is that these systems exhibit nonlinear behavior at low excitation powers, which leads to a systematic downshift of spin-wave transmission gaps as the power goes up. What matters is that this nonlinearity happens because of a strong spatial concentration of spin waves inside the resonator itself.

Lev: From an error correction standpoint, if you're trying to run this on real hardware, that low-power threshold for nonlinearity is actually pretty interesting because it suggests you don't need huge driving fields to get these kinds of effects going.

Kai: Right, so what it claims is that this power-dependent transmission gives off neuron-like activation and frequency-selective nonlinear spin-wave absorption. It suggests they’re not just linear filters anymore when you start pumping more energy into them.

Mira: And the implication there is that these resonators are being highlighted as compact low-power nonlinear elements for neuromorphic magnonic computing architectures. That means they could be building parts of a computer that mimics brain-like behavior using magnons.

Lev: If we're thinking about running this on actual quantum hardware, the fact that the nonlinearity is at low power suggests you might be able to keep the energy consumption down compared to other ways of trying to achieve nonlinear logic.

Kai: Exactly. The paper shows how increasing power causes these transmission gaps to move toward lower frequencies, with shifts up to fifty MHz between-fifteen dBm and five dBm. That’s a significant shift in frequency based on just the input power level.

Mira: And that shift is what enables the neuron-like activation behavior you mentioned, where increasing power enhances the transmitted spin-wave amplitude when it's already inside that linear gap range.

Lev: I wonder about the numbers, though, because this nonlinear enhancement relies on a specific resonance within the cavity. Does that mean if you just tune your excitation frequency slightly off that peak, the effect disappears?

Kai: The paper says this behavior is reduced by a strong spatial concentration of spin waves within the resonator. It’s not just about the input power; it’s about how tightly those spin waves are packed inside the structure.

Mira: And they quantify that concentration by saying the spin-wave amplitude is enhanced by about five times for the in-plane component and by more than a factor of eleven for the out-of-plane component compared to a bare YIG film.

Lev: That amplification factor sounds substantial; running error correction codes on signals that are amplified this much needs careful consideration regarding noise accumulation, right?

Paper summary: Kai: It does. But the paper points out that even low-amplitude, nominally linear in bare YIG spin waves produce a nonlinear response when their amplitude is resonantly enhanced inside the nanoscale resonator.

Mira: That resonant reduction of the nonlinear thresholds is what makes it promising for energy consumption compared to other neural network approaches. It suggests a way to get nonlinearity without needing massive power inputs.

Lev: So, if you were designing a system, would you look at this as a way to introduce selective suppression or enhancement based on the signal's amplitude before it goes further?

Kai: That’s exactly what they show. They have both response types: one where increasing power leads to enhancement in the linear gap, and another where higher excitation powers suppress the transmitted spin-wave signal just below the gap.

Mira: So we get this power-dependent transmission that acts like a threshold-dependent suppression or enhancement, which supports logical operations based on how much energy you put in.

Lev: It seems like a way to build magnonic logic where the decision—whether to amplify or suppress—is encoded in the amplitude of the spin wave itself, rather than just switching on or off a component.

Kai: That’s what they are highlighting: that this structure could be used in neuromorphic computing architectures that utilize time-delay multiplexing, which is another way to handle computation.

Mira: And it connects back to their main point: magnonic Fabry-Pérot resonators can be used as compact low-power nonlinear elements for those kinds of devices. They are smaller and use less energy than what’s currently being explored for these types of functions.

Lev: The limitation they mention is that the simulations reproduce the experimental results with an exception of a minor frequency offset, which they attribute to small differences between their measured system and their model. So, you have to be careful about that modeling aspect when trying to replicate it exactly.

Kai: That’s a practical point. They also mention that the threshold power for nonlinear resonator behavior is reduced compared to using just a bare YIG film, which shows the advantage of having this structure.

Mira: It really boils down to showing how you can selectively introduce nonlinearity at specific nodes within a network of linear spin waves, rather than trying to make the whole system nonlinear at once.

Lev: That selective introduction is key for scaling up any hardware architecture, because you don't want every single component behaving nonlinearly all the time.

Kai: So this paper lays out a path toward magnonic neural networks where linear spin waves handle the all-to-all connectivity, and nonlinearity is introduced selectively at the resonator nodes.

Mira: And it suggests that this approach could substantially reduce energy consumption when compared to non-resonant neural networks, which makes it a compelling area for research in this field.

Conclusion: Kai: So we're wrapping up this look at these magnonic resonators that are showing nonlinearity when you pump more power into them, specifically focusing on how they act like tiny neurons and filters.

Mira: Yeah, the title itself is pretty direct; it's about seeing those nonlinear dynamics in these Fabry-Pérot structures and what that means for low-power operation.

Lev: From a practical standpoint, the authors are pointing to this as a way to do logic operations using amplitude encoding rather than just on or off signals.

Kai: It really boils down to showing how you can use these tiny cavities to selectively enhance or suppress spin waves based on their own power level, which is what we call neuron-like activation.

Mira: But the big idea here is that this happens at low excitation powers and due to a specific spatial concentration of the spin waves inside the resonator, which is what makes it different from just having a linear system.

Lev: That low threshold for nonlinearity is where I get curious; if we're building on this for actual error correction, how much noise are we talking about when you're running at these low power levels?

Kai: We’ll have to dig into those numbers next because that’s a big deal for making this viable in a real chip.

NanoSpin, Department of Applied Physics, Aalto University School of Science · University of Exeter

cond-mat.mtrl-sci, cond-mat.mes-hall

Submitted: 2026-02-11

Updated: 2026-02-11

Comments: 5 pages, 5 figures

Journal ref: Applied Physics Letters 129, 122403 (2026)

DOI: 10.1063/5.0327300

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 92/100

The gist: The gist: The study reports that nonlinear dynamics in magnonic Fabry-Pérot resonators enable low-power neuron-like activation and frequency-selective nonlinear spin-wave absorption, highlighting

Key concepts

Magnonics: Magnonics is a field focused on using magnetic materials like YIG to process information using spin waves instead of traditional electronics. It aims to create energy-efficient and scalable devices for computation and signal processing by exploiting the inherent magnetic properties of these systems.
Fabry-Pérot Resonator: This is a type of cavity structure formed by two mirrors, in this case, YIG films coupled with CoFeB nanostripes. It traps and modifies spin waves, creating specific resonant frequencies where spin waves can be strongly confined and interact nonlinearly.
Neuron-like Activation: This refers to a behavior where the transmission of a signal (spin wave) changes in a way that mimics how biological neurons fire. Specifically, increasing the input power causes an enhancement of transmission within certain frequency bands, similar to how neurons activate when stimulated above a threshold.
Spin-wave Transmission Gaps: These are specific frequency ranges where spin waves cannot propagate through the resonator in the linear regime. The paper observes that as excitation power increases, these gaps systematically shift toward lower frequencies due to nonlinear effects within the cavity.

Terminology

Activation Function — 3× in this paper · explained in 1 episode
Alignment — 2× in this paper
All-to-All Connectivity — 1× in this paper
Verification — 1× in this paper · explained in 2 episodes

Summary

The gist: The study reports that nonlinear dynamics in magnonic Fabry-Pérot resonators enable low-power neuron-like activation and frequency-selective nonlinear spin-wave absorption, highlighting their potential for neuromorphic magnonic computing architectures.

Introduction to Magnonics and Nonlinearity

Magnonics is emerging as a promising platform for designing energy-efficient and scalable devices for information processing Many existing magnonic devices and proof-of-concept demonstrations operate in the linear regime, where spin waves propagate and interfere without mutual interaction While this regime is well understood and technologically valuable, the most versatile and powerful magnonic functionalities arise from nonlinear dynamics Accessing the nonlinear regime enables a wide range of magnonic phenomena, including amplitude-dependent frequency shifts, multimagnon scattering, parametric amplification, formation of solitons and other self-localized modes, and generation of coherent spin-wave frequency combs These effects form the basis for various applications such as reconfigurable on-chip signal processing, microwave generation and control, neuromorphic computing, and quantum transduction The inherent nonlinearity of magnetization dynamics manifests itself in a strong magnon-magnon interaction

Fabrication and Experimental Setup

The system under investigation consists of a low-damping yttrium iron garnet (YIG) film coupled to a ferromagnetic metal nanostripe Dynamic dipolar coupling between the two materials forms a magnonic cavity with a modified spin-wave dispersion The resonator consists of an 85-nm-thick YIG film with a series of Co40Fe40B20 (CoFeB) nanostripes patterned on top Electronbeam lithography followed by magnetron sputtering and liftoff was used to fabricate 30 nm thick and 15 µm long CoFeB nanostripes with a width range from 350 to 1000 nm A 1 nm Ta/3 nm TaOx spacer between the CoFeB and YIG films blocked exchange while mediating dynamic dipolar coupling between their magnetizations Spin waves in the YIG film were launched via a 1.5-µm-wide Au microwave antenna at excitation powers between −15 and +5 dBm Spin-wave transmission through the resonator was measured using SNS-MOKE microscopy A 3 mT magnetic field was applied parallel to the antenna to establish the Damon-Eshbach geometry Micromagnetic simulations were performed using MuMax336

Observation of Nonlinear Behavior

Using super-Nyquist sampling magneto-optical Kerr effect (SNS-MOKE) microscopy together with micromagnetic simulations, the researchers observed a systematic downshift of the spin-wave transmission gaps as the excitation power increases This nonlinear behavior occurs at low power levels, reduced by a strong spatial concentration of spin waves within the resonator The resulting power-dependent transmission enables neuron-like activation behavior and frequency-selective nonlinear spin-wave absorption Increasing the excitation power causes the transmission gaps to move toward lower frequencies, with shifts of up to 50 MHz between −15 dBm and 5 dBm

Neuron-like Activation and Suppression

The nonlinear shift gives rise to qualitatively different transmission behaviors depending on the excitation frequency When the excitation frequency lies within the transmission gap in the linear regime, increasing the power leads to a nonlinear enhancement of the transmitted spin-wave amplitude This response resembles neuron-like activation functions commonly used in artificial neural networks In contrast, for frequencies just below the transmission gap, higher excitation powers suppress the transmitted spin-wave signal This nonlinear suppression can serve as a magnonic limiter, protecting downstream components from high-power microwave inputs Both response types can enable nonlinear magnonic logic, where threshold-dependent suppression or enhancement of spin-wave transmission supports logical operations based on amplitude encoding

Mechanism of Nonlinear Enhancement

The pronounced nonlinearity of the magnonic Fabry-Pérot resonators at relatively low powers arises from the resonant concentration of the spin-wave energy within the cavity The spin-wave amplitude is enhanced by about five times for the in-plane component and by more than a factor of eleven for the out-of-plane component compared to the bare YIG film Consequently, even low-amplitude, nominally linear in bare YIG spin waves produce a nonlinear response when their amplitude is resonantly enhanced inside the nanoscale resonator The resonant reduction of the nonlinear thresholds suggests that the energy consumption in such devices will be substantially reduced in comparison to non-resonant neural networks

Conclusion and Future Directions

The study concludes that both experiments and simulations reveal a systematic frequency downshift of the spin-wave transmission gaps with increasing excitation power This enables the realization of neuron-like activation functions and frequency-selective nonlinear spin-wave absorption Compared to a bare YIG film, the threshold power for nonlinear resonator behavior is reduced, highlighting the potential of these nanoscale resonators as low-power nonlinear elements for neuromorphic magnonic computing networks The results suggest a route toward magnonic neural networks where all-to-all connectivity is mediated by interference of linear spin waves, while nonlinearity is selectively introduced at resonator nodes Alternatively, nonlinear magnonic Fabry-Pérot resonators could be used within neuromorphic computing architectures that utilize time-delay multiplexing The resonant reduction of the nonlinear thresholds suggests that the energy consumption in such devices will be substantially reduced in comparison to non-resonant neural networks

Author Contributions

A.L. performed the experiments A.L. and K.G.F. conducted the micromagnetic simulations L.F. assisted in the SNS-MOKE measurements A.V.S., V.V., and S supervised the work All authors discussed the results and contributed to writing of the manuscript

Data Availability

The data that support the findings of this study are available from the corresponding authors upon reasonable request The project also received funding from the European Union’s Horizon Europe research and innovation program under Grant Agreement No. 101070347-MANNGA The project also received funding from the Research Council of Finland (Grant No. 357211) and the UK Research and Innovation (UKRI) under the UK government’s Horizon Europe funding guarantee (Grant No. 10039217) as part of the Horizon Europe (HORIZON-CL4-2021-DIGITAL-EMERGING-01) under Grant Agreement No. 101070347 Conflict of Interest The authors have no conflicts to disclose Author Contributions A.L. performed the experiments, A.L. and K.G.F conducted the micromagnetic simulations, L.F assisted in the SNS-MOKE measurements, and A.V., V.

Improvements for AI systems

textbf Neuron-like Activation Functions in Magnonic Circuits :: This system can emulate neuron-like activation behavior by exploiting the power-dependent frequency shift of the transmission gaps. The system can exhibit thresholdactivation behavior analogous to biological and artificial neurons, enabling logical operations based on amplitude encoding through nonlinear spin-wave suppression or enhancement.
textbf Frequency-Selective Nonlinear Spin-Wave Absorption :: The system can perform frequency-selective nonlinear spin-wave absorption by leveraging the nonlinear dynamics within the resonator. This allows for nonlinear magnonic logic, where threshold-dependent suppression or enhancement of spin-wave transmission supports logical operations.
textbf Low-Power Neuromorphic Magnonic Computing :: The system can realize compact low-power nonlinear elements for neuromorphic magnonic computing architectures. By using the resonant concentration of energy within the cavity, the threshold power for nonlinear resonator behavior is reduced, leading to a substantial reduction in energy consumption compared to non-resonant networks.
textbf Magnonic Limiter Function :: The system can function as a magnonic limiter, protecting downstream components from high-power microwave inputs when excitation powers are increased just below the transmission gap. This suppression mechanism allows for amplitude-dependent control of transmitted spin waves based on power level.
textbf Resonator Node Nonlinearity :: The system supports a architecture where all-to-all connectivity is mediated by interference of linear spin waves, while nonlinearity is selectively introduced at resonator nodes. This selective introduction of nonlinearity enables efficient and potentially lower-energy neuromorphic networks compared to non-resonant alternatives.

Abstract

We report on nonlinear spin-wave dynamics in magnonic Fabry-Pérot resonators composed of yttrium iron garnet (YIG) films coupled to CoFeB nanostripes. Using super-Nyquist sampling magneto-optical Kerr effect microscopy and micromagnetic simulations, we observe a systematic downshift of the spin-wave transmission gaps as the excitation power increases. This nonlinear behavior occurs at low power levels, reduced by a strong spatial concentration of spin waves within the resonator. The resulting power-dependent transmission enables neuron-like activation behavior and frequency-selective nonlinear spin-wave absorption. Our results highlight magnonic Fabry-Pérot resonators as compact low-power nonlinear elements for neuromorphic magnonic computing architectures.