🔬 physics

Accurate prediction of the α→β phase transformation temperature in tin via full anharmonic treatment

This study demonstrates that employing a full anharmonic treatment via thermodynamic integration on an Atomic Cluster Expansion potential successfully predicts the α→β phase transformation temperature of tin (288 K) with high accuracy, whereas the quasi-harmonic approximation significantly overestimates it (377 K) due to its failure to capture substantial vibrational anharmonicity in the metallic β-phase.

Original authors: Petr Šesták, Matous Mrovec, Martin Friák

Published 2026-07-29

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Petr Šesták, Matous Mrovec, Martin Friák

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are holding a block of metal in your hand. You know that if you heat it up, it might melt, but what if it changes its very soul before it even gets hot enough to melt? This is the strange world of phase transitions, where materials switch from one crystal structure to another, like a chameleon changing colors but on an atomic scale. For decades, scientists have been trying to predict exactly when and why this happens, especially for a metal called tin. The challenge lies in the fact that atoms aren't just rigid balls bouncing around; they are more like a chaotic dance floor where everyone is jiggling, pushing, and pulling on each other in complex ways. To predict when the dance changes style, you have to account for every single wiggle and wobble. If you ignore the messy, chaotic parts of the dance (what scientists call "anharmonicity"), your predictions can be wildly off, sending you on a wild goose chase that is hundreds of degrees wrong. Getting this right matters because tin is everywhere, from soldering electronics to protecting ships from rust, and if it changes its mind unexpectedly, it can crumble into dust, ruining everything it holds together.

Now, let's dive into the story of how a team of researchers finally cracked the code for tin. Tin has two main personalities: a shiny, metallic "white tin" that likes to be warm, and a dull, semiconducting "gray tin" that prefers the cold. At a specific temperature of about 286 Kelvin (which is roughly 13.2°C or 56°F), these two personalities are supposed to swap places. For a long time, computer simulations trying to predict this swap were failing miserably. Some predicted it would happen way too hot, others too cold, and the gap between the computer's guess and reality was huge—sometimes off by nearly 100 degrees! The scientists suspected that the computers were using a simplified rulebook that treated the atoms like perfect springs that only stretch and compress in a neat, predictable way. This is called the "quasi-harmonic approximation." It's like assuming a trampoline only bounces straight up and down, ignoring the fact that if you jump hard enough, the fabric stretches sideways and twists.

In this new study, the researchers decided to throw out the simple rulebook and build a much more sophisticated one. They used a powerful machine-learning tool called an "Atomic Cluster Expansion" (ACE) potential. Think of this as teaching a super-smart AI to watch millions of atomic dances and learn the exact rules of how they interact, based on high-level physics calculations. Once the AI learned the rules, they ran two different simulations to see when the switch happens. The first simulation used the old, simple "perfect spring" rulebook. It predicted that the switch would happen at 377 K, which is way too hot. But the second simulation, which used the new AI model to account for the messy, real-world "anharmonic" jiggling and twisting of the atoms, told a different story. This time, the simulation predicted the switch would happen at 288 K. That is incredibly close to the real experimental value of 286 K!

The big discovery here is that the "messiness" of the atoms is the key. The researchers found that the white tin phase (the metallic one) is incredibly "anharmonic," meaning its atoms are wild dancers that don't just vibrate in place; they push and pull on their neighbors in complex ways that lower the energy required to stay in that phase. This extra stability, provided by the chaotic atomic motion, is what actually allows the white tin to survive at lower temperatures than the simple models predicted. In contrast, the gray tin phase is much more orderly and behaves almost like a perfect spring, so the messy effects don't matter as much for it. By ignoring this chaos, previous models were missing a huge piece of the puzzle.

The team didn't just guess this; they measured it through their simulations. They looked at how much heat the metal could hold and found that the white tin absorbs way more heat than the simple models said it should, a clear sign of this extra atomic chaos. They also watched the atoms move in real-time and saw that their forces and movements were completely different from the neat, predictable patterns of the simple models. While the researchers are very confident in these simulation results, they do note that the final answer depends on how accurate the underlying physics rules were to begin with. However, their work strongly suggests that if you want to predict when tin will change its mind, you absolutely cannot ignore the wild, anharmonic dance of its atoms. Without accounting for that chaos, you'll never get the temperature right.

Technical Summary: Accurate Prediction of the $\alpha \to \beta$ Phase Transformation Temperature in Tin via Full Anharmonic Treatment

Problem Statement
The accurate prediction of the $\alpha \to \beta$ (grey-to-white) phase transformation temperature in tin (Sn) remains a significant challenge for atomistic simulations. The experimental equilibrium transition occurs at 286 K (13.2 °C). However, existing theoretical approaches have historically failed to reproduce this value, often over- or underestimating the boundary by several hundred Kelvin. The difficulty arises from the extremely small static energy difference ( $\Delta E_{\alpha\beta}$ ) between the diamond cubic ( $\alpha$ ) and body-centered tetragonal ( $\beta$ ) phases, which is on the order of a few tens of meV/atom. Consequently, the predicted transition temperature is hypersensitive to the details of electronic structure calculations (e.g., exchange-correlation functionals) and the treatment of finite-temperature effects. Previous studies utilizing the Quasi-Harmonic Approximation (QHA) with standard Density Functional Theory (DFT) functionals (such as PBE) consistently overestimated the transition temperature (yielding values between 350–500 K), raising questions about whether this error stems from static energy inaccuracies or a fundamental breakdown of the harmonic approximation in the metallic $\beta$ phase.

Methodology
To resolve these discrepancies, the authors constructed a high-fidelity Atomic Cluster Expansion (ACE) machine-learning interatomic potential trained on first-principles data computed using the PBE functional. This approach fixes the electronic reference at a single, well-defined level of theory, allowing for a rigorous comparison of different thermodynamic treatments on the same potential energy surface (PES).

The workflow involved three primary stages:

Potential Development: A training set comprising 4,427 configurations (including equilibrium, distorted, defective, and AIMD trajectories) was used to fit the ACE potential. The model was validated against DFT lattice parameters, elastic constants, and phonon dispersions, demonstrating high accuracy and dynamic stability for both phases.
Quasi-Harmonic Approximation (QHA): The free energies of $\alpha$ -Sn and $\beta$ -Sn were calculated using the QHA, which accounts for thermal expansion but neglects explicit phonon-phonon anharmonicity.
Full Thermodynamic Integration (TI): To capture explicit anharmonic effects, the authors employed thermodynamic integration using the Calphy software package. This method computes the free energy by integrating the energy difference between the ACE potential and a reference harmonic system over a coupling parameter.

By comparing the transition temperatures derived from QHA and TI on the identical ACE potential, the study isolates the contribution of explicit anharmonicity to the phase stability.

Key Results

Transition Temperature: The QHA approach predicted a transformation temperature of $T_{\alpha\beta}^{QHA} \approx 377$ K, overestimating the experimental value by nearly 90 K. In contrast, the full thermodynamic integration yielded $T_{\alpha\beta}^{TI} = 288$ K, which is within 2 K of the experimental 286 K.
Anharmonic Contribution: The shift from 377 K to 288 K directly quantifies the explicit anharmonic free energy ( $\Delta F_{anh}$ ). This contribution is substantial for the metallic $\beta$ -Sn phase (reaching approximately $-11$ meV/atom at 300 K) but negligible for the semiconducting $\alpha$ -Sn phase (approx. $-2$ meV/atom).
Physical Evidence of Anharmonicity: The study provides four independent lines of evidence confirming the strong anharmonicity of $\beta$ $β$ -Sn:
1. Heat Capacity: The calculated constant-pressure heat capacity ( $C_p$ ) of $\beta$ -Sn exceeds the classical Dulong-Petit limit ( $3R$ ) above 200 K, a feature only captured by the TI method, not QHA.
2. Vibrational Spectrum: Analysis of the velocity autocorrelation function (VACF) reveals a temperature-driven suppression of the 3–3.5 THz vibrational feature in $\beta$ -Sn and a shift of spectral weight to lower frequencies, phenomena absent in the harmonic QHA description.
3. Atomic Displacements: Real-space analysis shows anisotropic, non-Gaussian atomic displacement distributions in $\beta$ -Sn along specific crystallographic directions, whereas $\alpha$ -Sn remains harmonic.
4. Force Deviations: The normalized residual force amplitude ( $A_F$ ), representing the fraction of force not described by the harmonic approximation, reaches 0.58 for $\beta$ -Sn at 300 K compared to 0.24 for $\alpha$ -Sn.

Significance and Claims
The paper claims that the long-standing discrepancy in predicting the tin phase transition temperature is largely driven by the neglect of explicit lattice anharmonicity in the metallic $\beta$ -Sn phase. While previous studies attributed QHA overestimations primarily to errors in static 0 K energetics or exchange-correlation functionals, this work demonstrates that even with a standard PBE functional, capturing full vibrational anharmonicity is essential for achieving quantitative agreement with experiment.

The authors emphasize that while the absolute transition temperature prediction relies on the underlying 0 K energetics (which remain uncertain across different DFT functionals), the relative shift caused by anharmonicity is a robust physical effect. The study establishes that for soft, polymorphic materials with shallow potential energy landscapes, such as tin, full free-energy integration via machine-learning potentials is necessary to correctly predict phase boundaries. The results corroborate experimental observations of intrinsic anharmonicity in white tin, such as excess heat capacity and forbidden X-ray reflections, providing a microscopic theoretical explanation for these phenomena.

Technical Summary: Accurate Prediction of the α→β\alpha \to \betaα→β Phase Transformation Temperature in Tin via Full Anharmonic Treatment

More like this

Technical Summary: Accurate Prediction of the $\alpha \to \beta$ Phase Transformation Temperature in Tin via Full Anharmonic Treatment