Simulating Phase Change: Boltzmann Law, Temperature, and the Liquid-Vapor Model

Understanding Phase Transitions

• 💡 This video explores phase transitions using a discretized liquid-vapor model simulation, where blue pixels represent molecules and white pixels are empty space.
• 🎯 Phase transitions are not chemical reactions but rather different ways molecules interact and arrange, like ice, water, and steam from H2O.
• 🔑 The simulation uses temperature (T) and chemical potential (C) as parameters, with chemical potential acting as a stand-in for pressure to control molecule count.

The Boltzmann Law and Free Energy

• 🧠 The Boltzmann law is introduced as a fundamental formula, stating that the probability of a microstate is proportional to the exponential of its negative energy divided by temperature.
• ⚡ Randomness is used as a proxy for uncertainty in systems with many particles, as predicting individual molecular behavior is computationally impossible.
• 📈 Phase transitions arise from the minimization of free energy (E - T*S), which represents a competition between minimizing energy and maximizing entropy, mediated by temperature.

Defining Temperature and Chemical Potential

• 🔬 Temperature is defined as the quantity that equalizes when two systems exchange energy, specifically as the inverse of the derivative of entropy with respect to energy (1/T = dS/dE).
• 🧪 The chemical potential is defined as minus T times the derivative of entropy with respect to the number of molecules (-T * dS/dN), equalizing when systems exchange molecules.
• ✅ The Boltzmann formula is derived by considering a small system in contact with a large heat bath, allowing for energy exchange at a fixed temperature.

Simulating Phase Change Dynamics

• 🛠️ The simulation samples from the Boltzmann distribution using Markov Chain Monte Carlo (MCMC) algorithms, specifically Kawasaki Dynamics (swapping molecules) or Glauber dynamics (adding/removing molecules).
• 🚀 Allowing the number of molecules to change simplifies the simulation, enabling real-time adjustments of temperature and chemical potential on a GPU.
• 📊 The model generates a phase diagram similar to H2O, featuring liquid, gas, and supercritical fluid phases, demonstrating how density changes with T and C.

Model Insights and Universality

• ✨ The model exhibits interesting behaviors like metastability, where a system can remain in a “wrong” macrostate without an external impulse, and droplet/bubble formation.
• 🧩 At the critical point, the simulation displays fractal-like structures and self-similarity, meaning it looks the same regardless of zoom level.
• 💡 This simplified model’s ability to recover complex real-world behavior illustrates the universality principle, suggesting that macroscopic behavior depends on only a few fundamental microscopic rules.
• 🔗 The liquid-vapor model is shown to be equivalent to the Ising model (magnetism) and related to the XY model (vortices), highlighting connections across statistical mechanics.