Finite-temperature micromagnetic model bridging atomic- and macro-scale magnetism

This paper presents and validates the Landau-Lifshitz-Bernoulli (LLBe) model, a multi-scale finite-temperature micromagnetic framework that seamlessly bridges atomic and macroscopic scales to accurately predict bulk magnetic properties from below to above the Curie temperature, as demonstrated by its application to heat-assisted magnetic recording.

The Gist

Imagine you are trying to predict how a crowd of people moves.

In the world of magnets, scientists have two main ways to look at this "crowd" (which is actually made of tiny atomic magnets):

1. The "Frozen Crowd" Model (Old Way): This model assumes the crowd is frozen in place. Everyone is holding hands tightly, and no one can let go or change their size. It works great when the room is cold, but if you turn up the heat, the model breaks because it doesn't know how to handle people letting go of each other or shrinking down.
2. The "Flexible Crowd" Model (New Way): This is the new model presented in the paper, called LLBe. It understands that when the room gets hot, the crowd changes. People might let go of hands, shrink in size, or grow back when it cools down.

Here is a simple breakdown of what the paper does and why it matters:

The Problem: The "Too Hot" Problem

Modern technology, from wind turbines to hard drives, relies on magnets. To make better devices, scientists use computer simulations.

• The Issue: Existing computer models are like a camera that only works in the dark. They are perfect for cold magnets (where everything is solid and stiff). But when things get hot—like in a hard drive that is being heated up to write data—these old models fail. They can't handle the temperature rising above a certain point (called the Curie temperature) where the magnetism starts to disappear and then reappear.
• The Gap: Scientists needed a way to connect the tiny, atomic world (where heat makes atoms wiggle) with the big, macro world (where we see the magnet as a whole object).

The Solution: The "LLBe" Model

The authors created a new mathematical recipe called the Landau-Lifshitz-Bernoulli (LLBe) model.

Think of the old models as a rigid robot that can only march forward. The new LLBe model is like a shape-shifting robot.

• It has a "Thermostat" for size: The most important part of this new model is that it allows the "size" of the magnetism to change. In the old models, the magnet's strength was locked at a fixed number. In the LLBe model, the magnet's strength can grow or shrink depending on the temperature and the magnetic field, just like a balloon expanding or deflating.
• It uses a "Memory" of the material: Instead of guessing how the magnet behaves when hot, the model takes real data (from experiments or atomic simulations) and uses it as a guide. It asks, "If the temperature is X and the field is Y, what should the magnet size be?" and then forces the simulation to match that reality.

How It Was Tested

The authors didn't just make up the math; they proved it works by playing "match the model":

1. The Cold Test: They simulated a cold, thin magnetic film. The new model gave the exact same results as the famous, trusted software used by experts today. This proved it works for normal, cold magnets.
2. The Hot Test: They simulated a block of Gadolinium (a magnetic metal) at temperatures where it is just about to lose its magnetism and just after it regains it. They compared their results to a different, established type of physics software used for hot magnets. The new model matched perfectly.

The Real-World Demo: "Heat-Assisted" Writing

To show off the model's power, they simulated Heat-Assisted Magnetic Recording (HAMR).

• The Scenario: Imagine trying to flip a switch on a very stubborn door. It's too hard to push. But if you heat the door hinge, it becomes soft and easy to push. This is how modern hard drives write data: they zap a tiny spot with a laser to heat it up, making it easy to flip the magnetic bit, then let it cool down to lock the data in place.
• The Result: The new model successfully simulated this process. It showed that at room temperature, the bit wouldn't flip. But when they "heated" the bit in the simulation to near its melting point, the bit flipped easily. This proves the model can handle the complex, multi-scale dance of heat and magnetism that happens in real hard drives.

The Bottom Line

This paper introduces a new tool that bridges the gap between the tiny atomic world and the big macro world. It is a single equation that works whether the magnet is freezing cold, boiling hot, or somewhere in between. It allows scientists to simulate how magnets behave in high-temperature situations (like in hard drives or new types of cooling materials) with much higher accuracy than before, without needing to switch between different, incompatible software programs.

Technical Summary

Technical Summary: Finite-Temperature Micromagnetic Model Bridging Atomic- and Macro-Scale Magnetism

Problem Statement
Modern magnetic technologies, ranging from permanent magnets in wind turbines to data storage in hard disk drives, are fundamentally limited by the physical properties of magnetic materials. Computational modeling is essential for accelerating material development; however, existing models face a fidelity gap across length and temperature scales.

• Conventional Micromagnetics: The standard Landau-Lifshitz (LL) equation assumes a continuous magnetization field with a fixed magnitude ($M_s$). This zero-temperature approximation fails at high temperatures, particularly near or above the Curie temperature ($T_c$), where the magnetization norm is not preserved.
• Existing Finite-Temperature Models: Approaches like the Landau-Lifshitz-Bloch (LLB) equation allow for magnetization length relaxation but introduce complexity. They require specific transverse and longitudinal damping coefficients and magnetic susceptibilities, and they struggle to incorporate predetermined temperature and field-dependent magnetization data ($M(H, T)$) directly, often relying instead on zero-field parameterizations.
• The Gap: There is a need for a unified model that can accurately describe magnetic behavior from the atomic scale to the macro scale, spanning temperatures from well below $T_c$ to well above it, without requiring complex, material-specific damping parameters for every regime.

Methodology
The authors propose a new multi-scale model, the Landau-Lifshitz-Bernoulli (LLBe) equation. This model integrates the phenomenological Landau-Lifshitz equation with a Bernoulli-type differential equation to handle the relaxation of the magnetization norm.

• The Governing Equation:
  The LLBe equation is defined as:
  $$\frac{1}{\gamma} \frac{\partial \vec{M}}{\partial t} = -\vec{M} \times \vec{H} - \alpha \vec{M} \times (\vec{M} \times \vec{H}) - \beta (|\vec{M}| - M(H, T)) \vec{M}$$
  Here, $\gamma$ is the gyromagnetic ratio, $\alpha$ is the damping constant, and $\beta$ is a coefficient governing the longitudinal diffusion.

  • The first two terms are identical to the standard LL equation, handling precession and transverse damping.
  • The third term is the novel addition. It drives the magnetization norm $|\vec{M}|$ toward an equilibrium value $M(H, T)$, which depends on the local effective field and temperature. This equilibrium data can be sourced from experimental measurements, mean-field theory (MFT), or atomic-scale models (e.g., Heisenberg or Ising).

• Effective Field Consistency:
  Unlike the LLB equation, the LLBe does not add new field contributions. It retains standard expressions for the effective field ($\vec{H}$), including applied, demagnetizing, exchange, and anisotropy fields, but replaces the fixed saturation magnetization $M_s$ with the local norm $|\vec{M}|$. The authors verify that the energy functionals associated with these fields remain consistent with the new field expressions using variational calculus.

• Numerical Implementation:
  The model is discretized using the Finite Difference Method for time derivatives and the Finite Element Method (FEM) for spatial discretization of the magnetic field and magnetization.

  • The demagnetizing field is solved via a scalar potential approach.
  • The exchange field is derived using the box method and variational calculus.
  • A linesearch algorithm is employed to solve the resulting linear system of equations at each time step, ensuring convergence to the equilibrium state.

Key Contributions and Results
The paper validates the LLBe model against established solvers and demonstrates its capability across different temperature regimes:

1. Low-Temperature Validation (Micromagnetics):
   The model was tested on a permalloy thin film (100 x 100 x 5 nm) at low temperatures. The LLBe simulation accurately reproduced the dynamic behavior of magnetization components ($x, y, z$) when compared to MUMAX3, a standard zero-temperature micromagnetic solver. This confirms that the LLBe naturally recovers conventional LL results when $T \ll T_c$.

2. High-Temperature Validation (Macro-Scale Magnetostatics):
   The model was applied to a Gadolinium (Gd) cube at 280 K and 300 K (spanning the $T_c$ of ~290 K) under an applied field. The results were compared against FEMCE, a solver for classic magnetostatics. The LLBe predicted the magnetization field with good quantitative agreement to FEMCE, demonstrating its ability to reproduce classic Maxwell magnetostatics for paramagnets at high temperatures.

3. Application: Heat-Assisted Magnetic Recording (HAMR):
   The authors simulated a HAMR scenario on a thin magnetic track (540 x 60 x 16 nm) with local heating.

   • Setup: A bit with upward magnetization was subjected to a 3 T opposing field.
   • Results: At room temperature (293 K) and intermediate temperatures (600 K), the bit did not flip. However, when the local temperature was raised to 700 K (10 K below $T_c$), the bit successfully flipped.
   • Significance: This demonstrates the model's ability to capture the reduction of coercivity due to heating, a critical mechanism in HAMR, using a single governing equation.

Significance and Claims
The paper claims that the LLBe model offers a "straight-forward" pathway for multi-scale modeling of temperature-dependent magnetic systems. Its primary significance lies in:

• Unified Framework: It uses a single equation to describe magnetic behavior from below to above the Curie temperature, bridging the gap between atomic-scale simulations and macro-scale magnetostatics.
• Direct Coupling: It allows for the direct integration of intrinsic material data ($M(H, T)$) from various sources (experimental, mean-field, or atomic models) without the need for complex parameterization of damping coefficients or susceptibilities required by the LLB model.
• Practical Utility: The model is presented as a suitable tool for simulating dynamic, high-temperature processes such as Heat-Assisted Magnetic Recording (HAMR) and for studying magnetocaloric materials, overcoming the limitations of traditional micromagnetics near $T_c$.

The authors conclude that this approach defines a new pathway for computational modeling of temperature-critical magnetic materials, offering a direct improvement over existing studies on microstructure in magnetocaloric systems.