Evolution of speckle during spinodal decomposition

TL;DR

This paper investigates the evolution of speckle patterns during spinodal decomposition using numerical simulations of the Cahn-Hilliard-Cook equation, revealing scaling behaviors and a violation of a known autocorrelation bound.

Contribution

It provides a detailed analysis of speckle intensity covariance and correlation functions during spinodal decomposition, including new scaling functions and the violation of Fisher-Huse bound.

Findings

Speckle intensity covariance scales with elta t/ar{t} and elta t/ar{t}^{1-n}.

The speckle covariance equals the square of the two-time structure factor.

Autocorrelation exponent pprox 4.47 exceeds the Fisher-Huse bound.

Abstract

Time-dependent properties of the speckled intensity patterns created by scattering coherent radiation from materials undergoing spinodal decomposition are investigated by numerical integration of the Cahn-Hilliard-Cook equation. For binary systems which obey a local conservation law, the characteristic domain size is known to grow in time $\tau$ as $R = [B \tau]^n$ with n=1/3, where B is a constant. The intensities of individual speckles are found to be nonstationary, persistent time series. The two-time intensity covariance at wave vector ${\bf k}$ can be collapsed onto a scaling function $Cov(\delta t,\bar{t})$, where $\delta t = k^{1/n} B |\tau_2-\tau_1|$ and $\bar{t} = k^{1/n} B (\tau_1+\tau_2)/2$. Both analytically and numerically, the covariance is found to depend on $\delta t$ only through $\delta t/\bar{t}$ in the small-$\bar{t}$ limit and $\delta t/\bar{t} ^{1-n}$ in the large-$\bar{t}$ limit, consistent with a simple theory of moving interfaces that applies to any universality class described by a scalar order parameter. The speckle-intensity covariance is numerically demonstrated to be equal to the square of the two-time structure factor of the scattering material, for which an analytic scaling function is obtained for large $\bar{t}.$ In addition, the two-time, two-point order-parameter correlation function is found to scale as $C(r/(B^n\sqrt{\tau_1^{2n}+\tau_2^{2n}}),\tau_1/\tau_2)$, even for quite large distances $r$. The asymptotic power-law exponent for the autocorrelation function is found to be $\lambda \approx 4.47$, violating an upper bound conjectured by Fisher and Huse.