Recursion formulas for nonlinear density fluctuations near the glass transition

TL;DR

This paper derives and analyzes nonlinear recursion formulas for density fluctuations near the glass transition, improving upon mode-coupling theory by providing more accurate critical points and temperature predictions.

Contribution

The paper introduces a simplified nonlinear recursion formula for density fluctuations, offering improved predictions of critical points and temperatures over traditional mode-coupling theory.

Findings

Recovers simulation results well in the β-relaxation stage

Predicts a lower critical temperature than MCT, addressing the high T_c problem

Provides detailed analysis of the Debye-Waller factor and memory function evolution

Abstract

The time-convolutionless mode-coupling (TMCT) equation for the intermediate scattering function $f_{\alpha}(q,t)$ derived recently by the present author is transformed into a simple nonlinear recursion formula for a generating function $\Omega_{\alpha}(\bm{q},t)(=-\ln[f_{\alpha}(q,t)]/q^2)$, where $\alpha=c$ stands for a collective case and $\alpha=s$ for a self case. By employing the same simplification on the nonlinear memory function as that proposed by the mode-coupling theory (MCT), the simplified asymptotic recursion formula is then derived and is numerically analyzed for different temperatures under the initial conditions obtained from the simulation. In a liquid state the numerical results are shown to recover the simulation results well. Although they can describe the simulation results well in the $\beta$-relaxation stage even for lower temperatures, they do not agree with those in the so-called $\alpha$-relaxation stage because of the simplified model. The coupling parameter $\lambda^{(\alpha)}$ dependence of the Debye-Waller factor $f_{\alpha}$ is also discussed. The critical point is found as $\lambda_c^{(c)}=2e(\simeq 5.43656)$ and $f_c=e^{-1/2}(\simeq 0.60653)$, while MCT gives $\lambda_c^{(c)}=4.0$ and $f_c=1/2$. Then, the critical temperature $T_c$ is shown to be definitely lower than that predicted by MCT. Thus, it is emphasized that the present theory can improve the high $T_c$ problem appeared in MCT. The time evolution of the memory function and that of the diffusion coefficient are also investigated within asymptotic formulas.