🔬 materials science

Amplitude-conserving frequency demultiplexer based on a spin wave delay line

This paper presents a theoretical investigation into the spatio-temporal separation of spin-wave wavepackets based on their frequency, proposing a compact, amplitude-conserving frequency demultiplexer that utilizes a spin-wave delay line for analogue signal processing.

Original authors: Gaël Yann Thiancourt, Thibaut Devolder

Published 2026-11-03

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Gaël Yann Thiancourt, Thibaut Devolder

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the world of modern electronics, moving information from one place to another often feels like managing a crowded highway. When many different signals travel together, they must be sorted out so that each piece of data reaches its correct destination. This process, known as demultiplexing, is currently handled by complex electronic circuits that require significant power and space. However, a new line of research is looking to a different kind of wave for help: the spin wave. Unlike the electrical currents that flow through copper wires, spin waves are ripples of magnetization that travel through magnetic materials. Imagine the magnetic atoms in a solid material as a line of tiny compass needles; when one is nudged, it tips its neighbor, which tips the next, creating a ripple that moves through the material without the need for electrons to flow. These waves can carry information at microwave frequencies and, crucially, their speed depends on the frequency of the wave itself. This unique property suggests a way to sort signals not by complex electronics, but simply by letting them travel and naturally separate as they go.

Researchers at the University of Paris-Saclay have now provided a detailed theoretical blueprint for a device that uses this natural separation to sort signals. They focused on a specific setup: a narrow magnetic strip, just 60 nanometers thick and 5 micrometers wide, made from a cobalt-iron-boron alloy. This material was chosen because it remains stable even when temperatures change. The team simulated what happens when a burst of magnetic energy is sent into one end of this strip. They found that if the burst contains multiple frequencies, the different parts of the signal do not travel at the same speed. Instead, the higher-frequency components and the lower-frequency components drift apart as they move down the strip, much like runners with different paces spreading out over a long distance.

The study carefully mapped out how these wave packets behave as they travel. The researchers discovered that while the waves spread out and lose some strength over distance, the separation between different frequencies grows steadily. They established clear rules for how far the waves must travel and how long the initial signal burst should be to ensure the different frequencies can be picked up separately at the other end. For instance, they calculated that for a signal traveling 60 micrometers, a time gap of just a few nanoseconds is enough to distinguish between frequencies that are a few gigahertz apart. This separation happens passively, meaning the device does not need any external power or active control currents to perform the sorting; it relies entirely on the physics of the material and the geometry of the strip.

To prove this concept works in practice, the team simulated a complex signal containing three distinct frequencies: 7, 11, and 14 gigahertz. When this combined signal was sent through the magnetic strip, the different frequencies arrived at the receiving end at different times. The 7-gigahertz wave arrived first, followed by the 11-gigahertz wave, and finally the 14-gigahertz wave. By placing a "time gate" at the receiver to look at the signal only during specific short windows, the researchers showed that they could isolate each frequency perfectly. This demonstrated that the device could successfully untangle a mixed signal into its individual components.

However, there was a catch in the initial simulation. Because the waves travel at different speeds, they also suffer different amounts of energy loss. The waves that took longer to arrive had faded more than the faster ones, meaning the separated signals would have different strengths. To solve this, the researchers proposed a clever engineering tweak for the antennas that send and receive the waves. By designing the receiving antenna with a specific shape—a U-shaped structure made of two wires carrying opposite currents—they could tune how sensitive the antenna is to different frequencies. This design allowed them to boost the weaker, slower signals and balance the output so that all three separated frequencies arrived with equal strength.

The result is a theoretical design for a frequency demultiplexer that is both passive and compact, with a footprint of just 60 by 5 micrometers. While the work is currently a simulation, it offers a clear path forward for creating devices that can process multiple communication channels without the heavy power demands of current technology. The researchers note that building such a device would require solving the challenge of generating the necessary magnetic fields on a tiny chip, perhaps by using special layers of material to create the required magnetic environment. If these engineering hurdles can be overcome, this approach could lead to a new generation of information processors that sort and handle data using the natural behavior of magnetic waves.

Technical Summary: Amplitude-Conserving Frequency Demultiplexer Based on a Spin Wave Delay Line

Problem Statement
Communication systems require efficient separation of complex multi-tone signals for multichannel processing. While magnonics offers a promising platform for this by exploiting spin waves (SWs) with frequency-dependent group velocities, existing demultiplexing approaches have limitations. Previous implementations either required active current supplies or relied on complex geometries. Furthermore, the community has relatively sparse exploration of the dynamics of propagating SW wave packets, specifically regarding how excitation duration and dispersion relations govern their spatio-temporal evolution. The challenge lies in achieving space-time separation of frequency components without signal degradation or complex control mechanisms, though practical implementation requires addressing the generation of necessary bias fields.

Methodology
The authors present a theoretical investigation of SW wave packet generation, propagation, and collection in micrometer-wide magnetic conduits. The study utilizes a CoFeB alloy conduit (60 nm thick, 5 µm wide) with a transverse magnetic field applied in the width direction (Damon–Eshbach configuration).

The methodology combines:

Micromagnetic Simulations: Using the TetraX package to calculate exact dispersion relations and susceptibility maps.
Analytical Modeling: Deriving approximate expressions for dispersion relations (based on Kalinikos and Slavin) and wave packet evolution. This model accounts for group delay ( $\tau_g$ ), group-delay dispersion ( $\tau_D$ ), and excitation duration ( $\tau_0$ ).
Spatio-Temporal Analysis: Computing the transmission spectrum ( $\tilde{S}_{21}$ ) between antennas and analyzing the impulse response. The authors model the evolution of Gaussian wave packets, considering factors such as Gilbert damping-induced attenuation, dispersion-induced broadening, and decoherence.
Antenna Engineering: Investigating the use of specific antenna geometries (single-wire vs. $\sqcup$ -shaped) to compensate for amplitude variations caused by propagation losses.

Key Contributions

Clarification of Wave Packet Dynamics: The paper elucidates the roles of excitation duration and dispersion relations in the spatio-temporal evolution of SW wave packets. It establishes that wave packets carrying different frequencies separate linearly with distance ( $r$ ), while their temporal broadening scales only as $\sqrt{r}$ at large distances. This differential scaling is identified as the key mechanism for separation.
Spatio-Temporal Separation Criteria: The authors derive a specific criterion (Eq. 22) for the separability of wave packets, defining the minimum frequency spacing required to separate packets based on their arrival time differences and temporal widths.
Amplitude-Conserving Design: A significant contribution is the proposal of a method to conserve signal amplitudes during demultiplexing. By engineering the receiving antenna (using a $\sqcup$ -shaped filter) to have a frequency-dependent efficiency that inversely compensates for propagation losses, the system can recover the relative amplitudes of the separated signals.
Passive Device Concept: The work proposes a frequency demultiplexer based on a 60 $\times$ 5 $\mu m^2$ spin wave delay line that operates passively regarding signal processing (no active current supply for switching or modulation). However, the device requires a static transverse bias field (35 mT) to function, and the authors note that on-chip generation of this field remains a challenge for full integration.

Results

Dispersion and Velocity: The study confirms that the group velocity ( $v_g$ ) of the fundamental width mode ( $n=1$ ) varies significantly with frequency (e.g., peaking at ~30 km/s near 7 GHz for a 20 mT internal field), while the edge mode is unsuitable for long-distance transport due to low group velocity.
Wave Packet Evolution: Analytical models (Eq. 14) show excellent agreement with numerical integrations. The results demonstrate that for a 10 GHz input, wave packets arrive at times consistent with group delays (e.g., ~3 ns for 60 $\mu m$ ) and broaden due to dispersion.
Demultiplexing Demonstration: The authors simulate the separation of a complex signal containing three frequencies (7, 11, and 14 GHz). At a propagation distance of 60 $\mu m$ , the three components arrive at distinct time intervals (approx. 0–2.7 ns, 2.6–4.5 ns, and 4.5–6.5 ns), allowing for temporal gating.
Amplitude Recovery: While natural propagation leads to amplitude decay (with later arrivals being significantly weaker due to damping and broadening), the use of a tailored $\sqcup$ -shaped receiving antenna successfully compensates for these losses, resulting in a demultiplexed output where the relative amplitudes of the three frequency components are preserved.

Significance and Claims
The paper claims to provide a theoretical foundation for analogue frequency demultiplexing in space and time using spin waves. The significance lies in demonstrating that:

Frequency separation can be achieved in a single waveguide by exploiting the frequency dependence of group velocity, without active current supplies for the demultiplexing operation itself.
The separation of overlapping wave packets is governed by a balance between linear separation in time and sub-linear broadening.
It is possible to design a device that not only separates frequencies but also conserves the amplitude of the signal components, a critical feature for information processing.

The authors note that the operational bandwidth is limited to a few tens of GHz, determined by the ferromagnetic resonance (FMR) frequency, antenna geometry, and bias field. They explicitly acknowledge that while the concept is all-electrical in principle, the practical implementation of the required transverse bias field on-chip (without external magnets) remains a challenge for future material research, potentially solvable via exchange biasing. The work is presented as a step toward designing devices capable of filtering and processing multiple spin-wave channels in parallel.

More like this