Liquid–Liquid Phase Transition in Supercooled Silicon

Vishwas V. Vasisht, Srikanth SastryAdvances 2013 · DOI · arXiv

Introduction

Silicon is the cornerstone of modern technology, yet its behavior in the liquid state reveals a complexity far beyond its well-known semiconducting properties. Like water, silicon is a "tetrahedral" liquid that exhibits thermodynamic anomalies, most notably a predicted liquid-liquid phase transition (LLPT) in the supercooled regime. This paper synthesizes decades of theoretical, computational, and rare experimental research to explore the "split personality" of silicon: its transition between a metallic high-density liquid (HDL) and a semiconducting low-density liquid (LDL).

The Challenge of the "No-Man's Land"

The primary obstacle in studying liquid silicon is its status as a "fast crystallizer." When liquid silicon is cooled below its melting point ($T_m = 1687 K$), it typically crystallizes almost instantly. This creates a "no-man's land" where the liquid state is highly metastable and difficult to probe experimentally. Consequently, much of our understanding is derived from Classical Molecular Dynamics (MD) simulations, specifically those utilizing the Stillinger-Weber (SW) potential. The SW potential is considered the gold standard in this field because it accounts for both two-body (distance) and three-body (angular) interactions, which are essential for capturing silicon's tetrahedral bonding tendencies.

Theoretical Framework and Methodology

To navigate the difficulties of sampling metastable states, researchers employ several advanced computational techniques:

  • Parallel Tempering Monte Carlo (PTMC): Used to equilibrate the system at low temperatures and high negative pressures without falling into local energy minima.
  • Restricted Ensemble Monte Carlo (REMC): Used to locate the "liquid spinodal" (the limit of liquid stability) by constraining density fluctuations to prevent cavitation.
  • Ab Initio MD (FPMD): Used to validate classical results and probe the electronic density of states (DOS) without empirical parameters.

The Two-Liquid Hypothesis and the Critical Point

The central finding of this review is the evidence for two distinct liquid phases in silicon:

  1. High-Density Liquid (HDL): A metallic melt characterized by a coordination number of approximately 5.1 to 6.
  2. Low-Density Liquid (LDL): A tetrahedral, semiconducting liquid with a coordination number near 4.

Simulations provide strong evidence for a liquid-liquid critical point (LLCP) located at negative pressure. For the Stillinger-Weber potential, this critical point is estimated at $T_c \approx 1120 \pm 12 K$ and $P_c \approx -0.60 \pm 0.15 GPa$. This suggests that the transition only becomes continuous when the liquid is under significant tension (stretched).

Dynamics and Electronic Structure

The transition from HDL to LDL is not merely structural; it is also dynamic and electronic. As the liquid enters the LDL phase, the self-diffusion coefficient $D$ drops by approximately two orders of magnitude, reflecting the increased structural order. This is often described as a "fragile-to-strong" transition, where the liquid's viscosity shifts from a rapid, non-Arrhenius change with temperature to a more stable Arrhenius behavior.

Electronically, the LDL phase shows a significant drop in the density of states at the Fermi level. Electronic states become localized, marking a metal-to-semiconductor transition within the liquid itself. This coupling of structural, dynamic, and electronic changes reinforces the validity of the two-liquid model.

Conclusion and Future Directions

While the Stillinger-Weber potential provides a robust qualitative match to ab initio simulations and limited experimental benchmarks, it still faces challenges in absolute density and diffusivity accuracy. The "holy grail" of this research remains experimental verification. Future studies using ultrafast techniques, such as femtosecond X-ray pulses, may finally allow scientists to "catch" the liquid-liquid transition in the brief window before crystallization occurs, confirming the existence of the hidden critical point in one of the world's most important materials.

Summary updated

Abstract

This chapter contains sections titled: Introduction Early Work on Metastable Silicon Scenarios for Liquids Displaying Thermodynamic Anomalies Recent Studies of Metastable Silicon Electronic Structure Critical Assessment of Classical Simulation Results Summary References
BibTeX
@inproceedings{vasisht2013liquidliquid,
  author = {Vishwas V. Vasisht and Srikanth Sastry},
  title = {{Liquid–Liquid Phase Transition in Supercooled Silicon}},
  booktitle = {Advances},
  year = {2013},
  doi = {10.1002/9781118540350.ch18},
}
Figure