Abstract
This paper provides a rigorous mathematical proof for a fundamental conjecture in crystal growth: that an initial state of growing crystals, moving with a normal velocity $\gamma(n)$, asymptotically approaches a Wulff shape. Utilizing the level set method and the theory of viscosity solutions to Hamilton-Jacobi equations, the authors demonstrate that this growth process minimizes surface energy relative to volume. The proof holds for any number of dimensions and accounts for complex topological changes, such as the merging of multiple crystals.
Introduction
The Wulff shape is a cornerstone of materials science, representing the unique geometry that minimizes the total surface energy of a crystal for a given volume. While it has long been observed that crystals grow into specific faceted shapes, a rigorous proof that arbitrary initial configurations converge to this optimal shape was historically elusive. Previous models were often restricted to two dimensions or required smooth initial conditions that could not account for crystals merging or filling internal voids.
Osher and Merriman (1997) addressed these gaps by modeling crystal growth as a dynamic evolution. They proved that if a crystal's growth velocity in a specific direction is proportional to its surface tension in that direction, the resulting interface will eventually become a perfectly scaled Wulff crystal.
Mathematical Framework
To handle the complexities of evolving interfaces, the authors employed two primary mathematical tools:
1. The Level Set Method
Traditional methods of tracking surfaces often struggle with "topological changes"—situations where two separate crystals merge into one or where a hole within a crystal closes. The level set method circumvents this by representing the crystal surface as the zero level set of a higher-dimensional function, $\phi(x, t)$. This allows the mathematical model to naturally handle changes in connectivity without requiring complex re-parameterization of the surface.
2. Viscosity Solutions to Hamilton-Jacobi Equations
Crystal growth frequently results in "kinks" or facets where the surface is not smooth, rendering standard calculus inapplicable. The authors modeled the growth using Hamilton-Jacobi equations. By applying the theory of viscosity solutions, they were able to identify unique, physically consistent solutions even in the presence of these singularities.
The Asymptotic Limit and Energy Minimization
The core of the paper is the proof that the Wulff shape acts as a "universal attractor."
Theorem 2: The Asymptotic Limit
The authors proved that any arbitrary initial shape can be "trapped" between two Wulff shapes—one contained within the initial crystal and one enclosing it. As time $t \to \infty$, the gap between these two scaling Wulff shapes vanishes. This forces the interior shape to converge to the Wulff geometry, regardless of how complex the initial state was.
Theorem 3.1: Dynamic Energy Flow
A significant discovery in this work is the link between the motion of the surface and a static optimization problem. The authors demonstrated that the quantity $E(\Gamma)$, defined as the surface energy divided by the enclosed volume (raised to the power of $(d-1)/d$ in $d$ dimensions), strictly decreases as the crystal grows. This characterizes the growth process as a form of gradient flow toward a minimum energy state.
Key Insights
- Self-Convexification: Even if the surface tension function $\gamma(n)$ is non-convex, the crystal evolution effectively "convexifies" itself. The resulting shape follows the convex envelope of the energy function, explaining why certain facets dominate the final geometry.
- Generalized Huygens' Principle: The authors established that the boundary of a growing crystal at any time $t$ can be determined by expanding the initial shape using the Wulff shape as the fundamental growth unit, analogous to wave propagation in optics.
- Extinction Estimates: The framework also allows for the calculation of "extinction time"—the moment a shrinking crystal (moving inwards) disappears based on its initial volume and the Wulff parameters.
Conclusion
Osher and Merriman's work provides a robust theoretical foundation for understanding crystalline morphology. By linking the level set method with energy minimization principles, they confirmed that the Wulff shape is the inevitable result of anisotropic growth. While the paper focuses on orientation-dependent velocity, it sets the stage for future research into curvature-driven motion and high-order numerical simulations of crystal formation.
Summary updated