Hugo Bocquet,Peter M. Derlet
Abstract:Magnetic systems can exhibit thermally activated transitions whose timescales are often described by an Arrhenius law. However, robust predictions of such timescales are only available for certain cases. Inspired by the harmonic theory of Langer, we derive a general activation rate for multi-dimensional spin systems. Assuming local thermal equilibrium in the initial minimum and deriving an expression for the flow of probability density along the real unstable dynamical mode at the saddle point, we obtain the expression for the activation rate that is a function of the Gilbert damping parameter, $\alpha$. We find that this expression remains valid for the physically relevant regime of $\alpha\ll1$. When the activation is characterized by a coherent reorientation of all spins, we gain insight into the prefactor of the Arrhenius law by writing it in terms of spin wave frequencies and, for the case of a finite antiferromagnetic spin chain, obtain an expression that depends exponentially on the square of the system size indicating a break-down of the Meyer-Neldel rule.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the prediction of thermal activation processes in magnetic systems, especially how to accurately calculate the activation rate of multi - dimensional spin systems under the condition of low damping (Gilbert damping parameter \( \alpha \ll 1 \)). Specifically, the paper aims to extend the application range of Langer theory, make it applicable to the physically relevant low - damping regime, and provide a general expression for the activation rate.
### Main problems
1. **Prediction of thermal activation time scale**:
- Magnetic systems can remain in a dynamically stagnant state for a long time at low temperatures, but occasionally thermal - activation transitions occur. The time scale of these transitions can usually be described by the Arrhenius law, that is, the activation rate \( \Gamma \propto e^{-\Delta E / k_B T} \), where \( \Delta E \) is the energy barrier.
- However, for systems with multiple degrees of freedom, this prediction is only valid in certain specific cases. Especially when multiple dimensions and complex energy landscapes are involved, the prediction becomes difficult.
2. **Applicability under low - damping conditions**:
- Under low - damping conditions (\( \alpha \ll 1 \)), the assumptions of existing theories such as Langer theory may no longer be applicable, because these theories are usually based on precessional dynamics close to determinism.
- The goal of the paper is to verify and extend Langer theory so that it can handle thermal activation processes under low - damping conditions.
3. **Accurate calculation of activation rate**:
- In addition to the influence of the energy barrier, the activation rate also depends on the pre - factor \( \nu \), which requires considering the dynamic characteristics of the system, especially the influence of the Gilbert damping parameter \( \alpha \).
- By introducing the concept of local thermal equilibrium and analyzing the probability density flow at the saddle point, the paper derives an activation rate expression related to \( \alpha \).
### Solutions
- **Generalizing Langer theory**: By evaluating the time for the system to reach thermal equilibrium near the local minimum and comparing it with the activation time, the validity of Langer theory under low - damping conditions is verified.
- **Analysis of probability density flow**: Assume that there is local thermal equilibrium at the initial minimum, and derive an expression for the probability density flowing along the only truly unstable mode at the saddle point.
- **Explicit expression of activation rate**: Finally, an activation rate expression containing the Gilbert damping parameter \( \alpha \) is obtained, and it is proved that this expression is still valid in the physically relevant range (\( \alpha \ll 1 \)).
### Formula summary
- Activation rate formula:
\[
\Gamma = \nu e^{-\Delta E / k_B T}
\]
where the pre - factor \( \nu \) is:
\[
\nu = \frac{|\kappa|}{2\pi} \sqrt{\frac{\prod_{n = 1}^{2N} \varepsilon_n^{(m)}}{\prod_{n = 1}^{2N} |\varepsilon_n^{(s)}|}}
\]
Here, \( \varepsilon_n^{(m)} \) and \( \varepsilon_n^{(s)} \) are the eigenvalues of the Hessian matrix at the local minimum and saddle point respectively, and \( \kappa \) is the growth rate of the unstable mode at the saddle point.
- For coherent transitions, the pre - factor \( \nu \) can be further simplified to:
\[
\nu_{\text{coh}} = \frac{|i\omega_0^{(s),1}|}{2\pi} \sqrt{\frac{\prod_q |w_q^{(m),1} w_q^{(m),2}|}{\prod_q |w_q^{(s),1} w_q^{(s),2}|}}
\]
Through these methods, the paper successfully solves the problem of predicting the thermal activation rate of magnetic systems under low - damping conditions.