Nonlinear mode interactions under parametric excitation in a YIG microdisk

This paper experimentally and theoretically demonstrates how nonlinear interactions between quantized spin-wave modes in a YIG microdisk, driven by two-tone parallel pumping, exhibit controllable non-commutative behaviors dependent on mode selection and detuning, offering a scalable platform for neuromorphic and unconventional computing.

The Gist

Imagine a tiny, circular drum made of a special magnetic material (called YIG), so small it's invisible to the naked eye. This drum doesn't just vibrate with sound; it vibrates with magnetic waves (called spin waves).

In a big, flat sheet of this material, these waves are chaotic and messy, like a crowded dance floor where everyone is moving at once. But because our drum is so tiny and confined, the waves can only dance in specific, organized patterns. Think of these patterns like the specific notes a guitar string can play.

This paper is about what happens when we try to play two different notes on this magnetic drum at the same time, but with a twist: we don't just hit the drum; we "tickle" it with a magnetic field to make the waves grow.

Here is the story of their discovery, broken down into simple concepts:

1. The Setup: The Magnetic Drum

The researchers built a microscopic disk and placed a tiny antenna on top. They sent radio signals (microwaves) into it.

• The Trick: They used a technique called "parallel pumping." Imagine trying to push a child on a swing. If you push at the exact right moment in their swing cycle, they go higher and higher. That's what they did with the magnetic waves. They tuned their radio signals to exactly twice the frequency of the wave they wanted to create, causing the wave to grow exponentially until it hit a limit.

2. The Single Note (One Tone)

First, they played just one "note" (one frequency).

• The Result: The wave grew until it hit a "ceiling." Why? Because as the wave got stronger, its own pitch changed slightly (a phenomenon called a frequency shift). It shifted just enough that the radio signal was no longer pushing it at the perfect moment, so it stopped growing.
• The Discovery: Some waves shifted their pitch up when they got loud, while others shifted down. It depended on the shape of the wave inside the tiny disk.

3. The Two-Note Experiment (The Real Magic)

Then, they tried playing two notes at once. They sent two different radio frequencies to excite two different magnetic waves simultaneously.

This is where things got weird and fascinating. They discovered that the order in which they turned the notes on mattered immensely.

The "Non-Commutative" Surprise

In math, $2 + 3$ is the same as $3 + 2$. This is called "commutative."
In this magnetic drum, order matters.

• Scenario A: Turn on Note 1, let it settle, then turn on Note 2.
• Scenario B: Turn on Note 2, let it settle, then turn on Note 1.

The Result: The final state of the drum was completely different in Scenario A compared to Scenario B.

• Sometimes, the second note would crush the first one, silencing it.
• Other times, the second note would help the first one grow even louder.
• It was as if the drum had a memory. The history of how you got to the current state changed the current state itself.

4. The "Social" Analogy: The Dance Floor

To understand why this happens, imagine two dancers on a small stage (the two magnetic waves).

• Self-Shift (s-NFS): When a dancer spins fast, they get dizzy and change their rhythm slightly.
• Mutual Shift (m-NFS): When two dancers are on stage together, they affect each other.
  • If they are "friendly" (positive interaction), one dancer's spinning helps the other spin faster.
  • If they are "rivals" (negative interaction), one dancer's spinning pushes the other off balance.

The researchers found that depending on whether the dancers were "friendly" or "rivals" (which depends on the specific shape of the wave), the outcome changed.

• If Dancer A starts first, they get comfortable. When Dancer B enters, they might push Dancer A off the stage.
• If Dancer B starts first, they get comfortable. When Dancer A enters, they might actually help Dancer B spin faster.

The final "dance" depends entirely on who started the music.

5. Why Does This Matter? (The Big Picture)

You might ask, "Who cares about magnetic drums?"
The authors suggest this is a blueprint for unconventional computing.

• Memory and Logic: Because the system remembers the order of inputs (A then B vs. B then A), it acts like a simple brain. It can process information not just by what you tell it, but when you tell it.
• Pattern Recognition: They showed that by mixing different waves, the system could naturally sort or classify inputs. This is similar to how a neural network (AI) learns, but done with physical waves instead of software code.
• Scalability: If you can do this with two waves, you can do it with ten, or a hundred. The number of possible "states" grows exponentially, creating a massive playground for new types of computers that are faster and more energy-efficient than today's silicon chips.

Summary

The paper demonstrates that in a tiny magnetic disk, waves don't just add up; they interact in complex, history-dependent ways. By carefully controlling the timing and frequency of radio signals, scientists can create a system that "remembers" the sequence of events. This turns a simple magnetic disk into a powerful, programmable machine for the next generation of computing.

Technical Summary

1. Problem Statement

Spin-waves (SWs) are promising candidates for unconventional computing due to their high nonlinearity. However, in extended magnetic films, the magnon spectrum is continuous and degenerate, leading to complex, intractable phenomena (e.g., Bose-Einstein condensation, chaos) involving an unmanageable number of modes.

• The Challenge: While confined nanostructures quantize the SW spectrum, making the number of modes trackable, the specific mechanisms governing nonlinear interactions between multiple parametrically excited modes remain poorly understood.
• The Gap: Existing models often rely on macrospin approximations or single-mode theories. There is a lack of a comprehensive framework that explains how pairs of modes interact, how their steady states depend on excitation sequences, and how to quantitatively map these interactions for potential computing applications.

2. Methodology

The authors employed a combination of advanced experimental spectroscopy and a rigorous theoretical framework based on the Normal Modes Model (NMM).

Experimental Setup:

• Sample: A 1 µm diameter, 52 nm thick Yttrium Iron Garnet (YIG) disk magnetized in-plane.
• Excitation: Parallel pumping using a gold antenna generating a radio-frequency (rf) field parallel to the static magnetic field ($H_0 = 27.1$ mT).
• Technique:
  • Single-tone spectroscopy: Exciting one SW mode at a time to map instability regions (Arnold tongues).
  • Two-tone spectroscopy: Simultaneously exciting two distinct modes ($\omega_A$ and $\omega_B$) using two rf tones.
  • Pulse Modulation: The rf signals were pulse-width modulated (PWM) with a 3 µs delay between tones to control the temporal sequence (e.g., applying $\omega_A$ before $\omega_B$ vs. vice versa).
• Detection: A Magnetic Resonance Force Microscope (MRFM) was used to detect the longitudinal magnetization component, providing a signal proportional to the SW intensity in the steady state.

Theoretical Framework:

• The study utilizes a general theory of parametric excitation in confined structures (based on Ref. [35]).
• The nonlinear Landau-Lifshitz-Gilbert (LLG) equation is expanded on the eigenbasis of the disk's SW modes.
• The model focuses on 4-magnon scattering terms, specifically:
  • Self-Nonlinear Frequency Shift (s-NFS, $N_{hh}$): The frequency shift of a mode due to its own amplitude.
  • Mutual Nonlinear Frequency Shift (m-NFS, $N_{hn}$): The frequency shift of one mode caused by the amplitude of a second mode.

3. Key Contributions

1. Experimental Demonstration of Non-Commutativity: The paper experimentally proves that the final steady state of a two-mode system depends on the temporal sequence of excitation (i.e., the system is non-commutative: State($A \to B$) $\neq$ State($B \to A$)).
2. Identification of Dominant Mechanisms: The authors identify that the interplay between s-NFS and m-NFS is the primary driver of these complex dynamics. The sign of these coefficients (positive or negative) dictates whether modes suppress or enhance each other.
3. Quantitative Spectroscopic Mapping: They developed a method to extract nonlinear coefficients (NFS ratios) directly from experimental data by analyzing the shape and overlap of instability regions in the two-tone spectroscopy maps.
4. Generalized Theory Application: They successfully applied a reduced theoretical framework (projecting dynamics onto the eigenmode basis) to quantitatively reproduce complex experimental phase diagrams, moving beyond simple macrospin models.

4. Key Results

A. Single-Tone Behavior (Single Mode):

• The system exhibits typical Arnold tongue instability regions.
• The steady-state intensity is determined by the s-NFS. As the mode amplitude grows, its frequency shifts. The steady state is reached when this shifted frequency is critically detuned from the pump frequency ($2\omega_{mode}$).
• Modes exhibit different behaviors based on the sign of their s-NFS:
  • Positive s-NFS: The tongue shifts one way; intensity grows until the shift matches the detuning.
  • Negative s-NFS: The tongue shifts the opposite way.
• This behavior is attributed to the competition between static and dynamic field contributions in the confined geometry.

B. Two-Tone Interactions (Mode Pairs):

• Non-Commutative Behavior: When exciting pairs like T3 and T2, the final state depends entirely on which mode is excited first.
  • Example: If T2 (negative s-NFS) is excited first, it remains stable. If T2 is excited second (after T3), it is strongly suppressed in specific frequency regions.
  • Mechanism: The first mode shifts the resonance frequency of the second mode via m-NFS. If the signs of the shifts are opposing, the second mode may be pushed out of the instability region, suppressing the first mode.
• Commutative Behavior: Pairs like T3 and T1 (both positive s-NFS) show commutative behavior where the final state is independent of the sequence.
• Phase Diagrams: The authors constructed phase diagrams showing regions of:
  • Zero excitation ($z$): No mode is active.
  • Uncoupled modes ($u_h, u_n$): Only one mode is active.
  • Coupled modes ($c$): Both modes are active.
  • The boundaries of these regions are defined by the NFS coefficients and pump detuning.

C. Quantitative Agreement:

• By fitting the ratios of NFS coefficients ($N_{hh}/N_{hn}$) to the experimental data, the theoretical phase diagrams matched the experimental MRFM intensity maps with remarkable accuracy.
• The study revealed that NFS ratios vary significantly depending on the specific pair of modes chosen, highlighting the importance of spatial mode profiles.

5. Significance and Implications

• Model Platform for Nonlinear Dynamics: This system provides a highly controllable, scalable platform for studying nonlinear phenomena. Unlike chaotic systems in extended films, the number of modes here is finite and trackable.
• Unconventional Computing: The ability to map temporal rf inputs to specific programmable steady states (based on excitation sequence and frequency) establishes a route toward reservoir computing and neuromorphic architectures. The system can perform classification tasks by leveraging the nonlinear mixing of modes.
• Scalability: The authors suggest that extending this to larger sets of modes (via frequency multiplexing) could lead to chaotic dynamics with a number of stable solutions scaling as $2^p$ (where $p$ is the number of modes), offering immense computational capacity.
• Broad Applicability: The theoretical approach (normal mode expansion) is not limited to magnonics but can be applied to other dynamical systems with nonlinear mode interactions, such as hydrodynamics and opto-mechanics.

In summary, the paper bridges the gap between complex nonlinear magnetism and practical computing applications by demonstrating that the nonlinear interaction between quantized spin-wave modes is deterministic, controllable, and governed by measurable frequency shift parameters.