Pattern formation and film rupture in a two-dimensional thermocapillary thin-film model of the B\'enard-Marangoni problem

TL;DR

This paper analyzes pattern formation and film rupture in a two-dimensional thermocapillary thin-film model derived from the Bénard-Marangoni problem, identifying bifurcation points and conditions leading to film rupture.

Contribution

It provides a detailed bifurcation analysis of stationary patterns in the model and establishes conditions under which film rupture occurs, supported by numerical experiments.

Findings

Flat surface destabilizes at critical Marangoni number M*

Square and hexagonal patterns bifurcate from the flat profile at specific M values

Solutions can exhibit film rupture with minimal height tending to zero

Abstract

We study two-dimensional, stationary square and hexagonal patterns in the thermocapillary deformational thin-film model for the fluid height $h$\begin{equation*} \partial_t h+\nabla\cdot\left(h^3\left(\nabla\Delta h-g\nabla h\right)+M\frac{h^2}{(1+h)^2}\nabla h\right)=0,\quad t>0,\quad x\in\mathbb{R}^2, \end{equation*} that can be formally derived from the B\'enard-Marangoni problem via a long-wave approximation. Using a linear stability analysis, we show that the flat surface profile corresponding to the pure conduction state destabilises at a critical Marangoni number $M^*$ via a conserved long-wave instability. For any fixed absolute wave number $k_0$, we find that square and hexagonal patterns bifurcate from the flat surface profile at $M=M^* + 4k_0^2$. Using analytic global bifurcation theory, we show that the local bifurcation curves can be extended to global curves of square and hexagonal patterns with constant absolute wave number and mass. We exclude that the global bifurcation curves are closed loops through a global bifurcation in cones argument, which also establishes nodal properties for the solutions. Furthermore, assuming that the Marangoni number is uniformly bounded on the bifurcation branch, we prove that solutions exhibit film rupture, that is, their minimal height tends to zero. This assumption is substantiated by numerical experiments.