🔬 mesoscale physics

Domain wall motion in ferromagnetic nanowires driven by a localized Gaussian thermal gradient

This study uses stochastic Landau-Lifshitz-Gilbert simulations to demonstrate that magnetic domain walls in ferromagnetic nanowires are primarily driven by magnonic spin-transfer torque rather than entropic torque when displaced from a localized Gaussian thermal gradient, revealing how laser parameters and material properties nonlinearly influence wall velocity to guide thermal control in spintronic devices.

Original authors: M. A. Jafar Pikul, M. A. S. Akanda, M. T. Islam

Published 2026-07-15

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: M. A. Jafar Pikul, M. A. S. Akanda, M. T. Islam

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

Imagine a tiny, super-thin magnetic wire, like a microscopic railroad track made of a special metal called Permalloy. On this track sits a "domain wall"—think of it as a tiny, invisible fence separating two neighborhoods of magnetic atoms that are pointing in opposite directions. In the world of future computer memory (called racetrack memory), we want to push this fence along the track to store and read data.

Usually, we push these fences with electricity, but that creates a lot of heat and wastes energy. So, scientists asked: What if we just used a laser to nudge the fence instead?

In this study, the researchers used a computer simulation to see what happens when they shine a very focused laser spot on the wire. But here's the twist: they didn't shine the laser on the fence. They shone it a little bit away from it, creating a hot spot that fades out like a Gaussian bell curve (a smooth hill of heat).

The Big Surprise: The Invisible Push
You might think that if the fence is sitting in a cool spot far from the laser, nothing would happen. After all, the temperature right where the fence is sitting is flat and steady, so there's no "slope" of heat to slide down.

But the simulation showed the fence started moving anyway! It zoomed toward the hot laser spot. Why?

The authors explain that the laser heats up the metal, creating tiny, invisible waves of heat called "magnons." Think of these magnons like a swarm of energetic bees buzzing away from the hot laser. Even though the fence is far away, these bees travel through the wire until they hit the fence. When they crash into it, they transfer a little kick of momentum, shoving the fence toward the heat. This is called "magnonic spin-transfer torque."

The paper explicitly rules out a different idea: the "entropic torque." This is a force that usually pushes things from cold to hot, but it only works if there is a steep temperature slope right where the object is. Since the fence was placed far away (specifically, 105 nm away, which is 7 times the width of the laser spot), the temperature slope at the fence's location was effectively zero. So, the "slope-pushing" force was negligible. The movement was almost entirely due to the "bee-kick" (magnonic torque).

How Fast Does It Go?
The researchers ran thousands of these simulations to see how different knobs changed the speed of the fence:

Hotter Laser = Faster Fence: When they cranked up the peak laser temperature from 200 K to 800 K (all safely below the metal's melting point of 850 K), the fence moved faster. In fact, the speed went up in a straight line with the temperature. At 300 K, the fence moved at about 2 meters per second; at 800 K, it hit nearly 8 meters per second.
Further Away = Slower Fence: If they moved the laser spot further away from the fence (increasing the distance $d$ ), the fence slowed down. This is because the "bee swarm" gets weaker as it travels through the wire; the metal's natural resistance (called damping) eats up the energy of the waves before they reach the fence.
The Goldilocks Damping: This is the most interesting part. The metal has a property called "Gilbert damping" (let's call it $\alpha$ $α$ ), which is like how sticky the wire is to the moving bees.
- If the wire is too smooth (low damping), not enough bees are created to push the fence.
- If the wire is too sticky (high damping), the bees get eaten up before they reach the fence.
- The simulation showed a sweet spot! The fence moved fastest when the damping was around 0.002 to 0.003. This "hump" in the speed graph is a clear signature that the bees (magnons) are doing the work, not some other force.

Two Separate Knobs
The researchers discovered something really cool about how we control this system. You might think that the "distance" between the laser and the fence is the only thing that matters. But they found that the width of the laser spot ( $\sigma_L$ ) and the distance ( $d$ ) act as two completely independent controls.

Imagine you have a flashlight.

If you make the beam wider (increase $\sigma_L$ ), you heat up a bigger area, creating more bees. The fence moves faster, even if you keep the distance the same.
If you move the flashlight further away (increase $d$ ), the bees have a longer journey and get weaker. The fence slows down.

The paper shows that you can't just say "move the laser 100 nm away" and expect a specific result. You have to know how wide the laser beam is, too. A wide beam 100 nm away acts differently than a narrow beam 100 nm away. The simulation results for different beam widths never collapsed into a single line, proving these are two separate levers to pull.

The Shape of the Fence
They also looked at how "stiff" the magnetic fence is (called uniaxial anisotropy, $K_u$ ). When they made the fence stiffer, it got physically narrower (from about 3 nm down to 2 nm), but it moved slower. However, the fence started spinning (precessing) faster. It's like a skater pulling in their arms: they spin faster but move across the ice differently.

What This Means
The authors suggest that for future devices like racetrack memory, we could use these findings to control data movement with heat instead of electricity. They found a "sweet spot" for operating: a laser temperature under 500 K, a distance between 50 and 150 nm, a laser width between 10 and 20 nm, and a damping value around 0.001 to 0.003.

In short, this paper simulates a scenario where a laser heats a spot on a wire, sending invisible heat-waves to push a magnetic fence. It proves that even without a temperature slope at the fence's location, the fence will move, and it shows us exactly how to tune the laser's heat, width, and distance to make that movement happen just right.

Technical Summary: Domain Wall Motion in Ferromagnetic Nanowires Driven by a Localized Gaussian Thermal Gradient

Problem Statement
Magnetic domain walls (DWs) in ferromagnetic nanowires are critical components for spintronic data storage and logic applications, such as racetrack memory. While DWs can be driven by magnetic fields, microwave excitation, or spin-polarized currents, these methods face limitations including the inability to move multiple walls simultaneously, frequency sensitivity, or high energy consumption and Joule heating. Spin caloritronics offers an alternative by utilizing temperature gradients to excite thermal magnons, which carry spin angular momentum and exert torque on DWs. However, the specific dynamics of DWs driven by a localized Gaussian thermal profile (simulating a focused laser spot) remain incompletely understood. Previous studies using broad laser profiles or uniform linear gradients have not fully resolved the independent roles of the laser spot width ( $\sigma_L$ ) and the distance between the laser and the DW ( $d$ ), particularly in regimes where the local temperature gradient at the DW is negligible.

Methodology
The authors investigate DW dynamics in a uniaxial Permalloy (Ni $_{80}$ Fe $_{20}$ ) nanowire ( $1024 \times 8 \times 8$ nm $^3$ ) containing a head-to-head DW. The system is subjected to a localized Gaussian temperature profile $T(x)$ generated by a laser spot, with a peak temperature $T_L$ and width $\sigma_L$ . The simulations are performed using the stochastic Landau-Lifshitz-Gilbert (sLLG) equation, implemented in MuMax3, at temperatures below the Curie point ( $T_c = 850$ K).

Key simulation parameters include:

Geometry: The DW is initially positioned at the wire center ( $X_{DW}$ ), while the laser is placed at a distance $d = N\sigma_L$ to the right.
Forces: The study focuses on the regime where the initial local temperature gradient at the DW is effectively zero, rendering the entropic torque (ET) negligible. Consequently, the motion is driven primarily by magnonic spin-transfer torque ( $\mu$ STT).
Variables: The authors systematically vary the peak laser temperature ( $T_L$ ), the separation parameter ( $N$ , varying $d$ ), the Gilbert damping constant ( $\alpha$ ), the uniaxial anisotropy ( $K_u$ ), and the laser width ( $\sigma_L$ ).
Analysis: Results are averaged over 15 independent simulations with different random seeds to account for thermal fluctuations.

Key Results

Temperature Dependence: The DW velocity ( $v$ ) and angular velocity ( $d\phi/dt$ ) increase linearly with the peak laser temperature ( $T_L$ ). This linear scaling supports a thermal mobility picture where velocity is proportional to the characteristic strength of the temperature profile ( $T_L/\sigma_L$ ).
Distance Dependence: As the distance $d$ between the laser and the DW increases (by increasing $N$ ), both translational and angular velocities decrease. This attenuation is attributed to the damping of the magnon current as it propagates from the hot region to the DW.
Damping Dependence: The velocity exhibits a non-monotonic dependence on the Gilbert damping constant ( $\alpha$ ). Velocity increases at low $\alpha$ due to enhanced thermal magnon excitation (amplitude scales as $\sqrt{\alpha}$ ), reaches a broad maximum around $\alpha \approx 0.002\text{--}0.003$ , and then decreases as higher damping shortens the magnon propagation length ( $\propto \alpha^{-1/2}$ ). This non-monotonic behavior serves as a distinct signature of the magnonic driving mechanism, distinguishing it from field-driven models.
Anisotropy Dependence: Increasing uniaxial anisotropy ( $K_u$ ) narrows the DW width (consistent with $\Delta \propto \sqrt{A_{ex}/K_u}$ ) and reduces translational velocity while increasing angular velocity, indicating different sensitivities of translation and precession to anisotropy.
Independence of Control Parameters: A critical finding is that the laser width ( $\sigma_L$ ) and the laser-to-DW distance ( $d$ ) act as independent control parameters. When plotting velocity against the physical distance $d = N\sigma_L$ , the curves for different $N$ values do not collapse onto a single trend. This demonstrates that $d$ alone cannot determine the dynamics; $\sigma_L$ independently controls the size of the heated region and the magnon generation rate, while $d$ controls the propagation attenuation.

Significance and Claims
The paper claims to clarify the mechanism of localized thermally driven DW motion, specifically in the regime where the entropic torque is negligible and magnonic spin-transfer torque dominates. By decoupling the effects of laser width and separation distance, the study provides a parameter map for thermally driven DW dynamics in strongly localized Gaussian regions. The authors suggest these findings may guide thermal control strategies for spintronic racetrack memory devices. They identify a useful operating range for such devices: peak temperatures below 500 K, laser-to-DW separations of 50–150 nm, laser widths of 10–20 nm, and damping constants around $\alpha \approx 0.001\text{--}0.003$ . The work emphasizes that in localized regimes, both the generation of magnons (controlled by $\sigma_L$ ) and their propagation (controlled by $d$ ) must be considered independently to optimize DW motion.

More like this