🔬 materials science

Classification and enumeration of solid-solid phase transition mechanisms

This paper introduces the "crystmatch" method and a formal tree-based classification system to rigorously and exhaustively enumerate all possible crystal-structure matches for solid-solid phase transitions, enabling the discovery of thousands of new deformation mechanisms within minutes.

Original authors: Fang-Cheng Wang, Qi-Jun Ye, Yu-Cheng Zhu, Xin-Zheng Li

Published 2026-11-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Fang-Cheng Wang, Qi-Jun Ye, Yu-Cheng Zhu, Xin-Zheng Li

Original paper licensed under CC BY 4.0 (http://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

Solids are not static monuments; they are dynamic arrangements of atoms that can rearrange themselves into entirely new structures. When a material transforms from one crystal form to another, such as when graphite turns into diamond under immense pressure, the atoms do not simply vanish and reappear. Instead, they shift, slide, and rotate to form a new pattern. This process is known as a solid-solid phase transition. For scientists, understanding exactly how these atoms move from their starting positions to their final destinations is crucial. This specific mapping of where every atom ends up is called a crystal-structure match. It acts as a blueprint for the transition, defining the path the material takes. While researchers have long known that these transitions follow the path of least resistance, or the lowest energy barrier, finding that specific path has been a formidable challenge. The problem is that there are countless ways atoms can rearrange themselves. In a chemical reaction, the number of ways atoms can connect is relatively small and predictable. In a solid, however, the sheer number of identical atoms and the infinite possibilities for how the crystal lattice can stretch or twist make the number of potential rearrangement paths astronomically large. For decades, scientists have struggled to navigate this vast landscape, often relying on guesswork or simulations that might miss the true, most efficient route.

A team of researchers at Peking University has now developed a rigorous method to map this entire landscape, ensuring that no possible rearrangement is overlooked. They created a new mathematical framework that treats the problem not as a chaotic jumble of possibilities, but as a structured tree. By organizing every conceivable way atoms can match up between two crystal structures, they proved that while the number of options is theoretically infinite, the number of physically meaningful ones is actually finite and manageable. Their approach, which they named "crystmatch," allows them to list every single valid rearrangement mechanism without needing to rely on human intuition or incomplete computer searches. This method is independent of how the crystal is measured or defined, meaning it provides a universal language for describing these transitions. The researchers demonstrated that their system can reproduce every known transition mechanism that scientists have discovered through experiments or previous simulations, and it does so in a matter of minutes on a standard computer. More importantly, the system reveals thousands of new, previously unknown candidates for how these transitions could occur, many of which involve significantly less energy or strain than previously thought possible.

The power of this new method lies in its ability to exhaustively search for the best path. In the past, scientists might use a technique called the nudged elastic band method to find the easiest route between two structures, but this technique requires the user to specify the starting and ending atom connections in advance. If the user guesses the wrong connection, the computer finds the easiest path for that specific guess, but it might miss the true, lowest-energy path hidden behind a different arrangement. The new method removes this guesswork. It systematically generates every possible connection, filters out the ones that would require impossible amounts of energy or cause atoms to travel too far, and presents a complete list of the remaining viable options. The researchers tested this on the transition between two common crystal structures, known as B1 and B2, which are found in materials like sodium chloride and cesium chloride. They found that their system could identify all the mechanisms that had been proposed over the last few decades, including those discovered through complex molecular dynamics simulations. In just over a minute of computer time, they generated nearly 800 unique structural matches, and in two hours, they expanded that list to nearly 10,000 non-repeating possibilities.

One of the most significant findings is that the method can predict which transition mechanism is most likely to occur based on experimental observations. For instance, in the B1-to-B2 transition, certain specific alignments between the crystal layers are frequently observed in real-world experiments. The researchers used their tool to calculate which of the thousands of possible rearrangements would produce these exact alignments. They discovered that only a handful of the generated mechanisms could match the experimental data without requiring any extra rotation of the crystal. Furthermore, by adjusting the parameters to account for different pressures, they showed how the preferred path changes. At low pressures, certain rearrangements are favored, but as pressure increases, the energy landscape shifts, making different mechanisms more efficient. This ability to see how the "lattice mismatch spectrum" changes with pressure helps explain why some transitions only happen under specific conditions, such as the high pressures needed to turn graphite into diamond. The researchers also applied this to the graphite-to-diamond transition, showing that the method can handle complex changes in density and strain that occur under extreme pressure.

The implications of this work extend beyond just understanding how crystals change shape. Because the method produces a complete database of all possible mechanisms, it provides a foundation for future research. Scientists can now use this list to calculate the energy barriers for these transitions with high precision, or to train artificial intelligence models to predict how new materials will behave. The researchers have made their tool, "crystmatch," freely available to the scientific community, allowing others to explore these transitions in their own materials. By turning an overwhelming problem of infinite possibilities into a finite, organized list, this work removes the guesswork from understanding how solids transform. It confirms that while the atomic world is complex, it is not chaotic; every transition has a set of defined paths, and now, for the first time, we have a way to see them all.

Technical Summary: Classification and Enumeration of Solid-Solid Phase Transition Mechanisms

Problem Statement
Solid-solid phase transitions (SSPTs) involve the rearrangement of atoms from one crystal structure to another, establishing an atom-to-atom correspondence known as a Crystal-Structure Match (CSM). While the CSM concept is central to understanding transition mechanisms, existing computational methods face significant limitations. Methods like the Nudged Elastic Band (NEB) can only calculate the Minimum Energy Path (MEP) for a user-specified CSM, potentially missing the true lowest-barrier path if the correct CSM is not intuitively guessed. Conversely, geometric optimization methods (e.g., minimizing strain or broken bonds) often yield suboptimal CSMs and fail to provide a comprehensive list of all possibilities. The core challenge is the sheer number of potential CSMs due to identical atoms and lattice deformation degrees of freedom, making exhaustive enumeration by human intuition or standard algorithms infeasible. Furthermore, existing descriptions of CSMs often rely on specific supercell choices, leading to redundancy and ambiguity.

Methodology
The authors propose a rigorous, cell-independent formalism and a computational framework named crystmatch to classify and exhaustively enumerate all noncongruent CSMs.

Cell-Independent Formalism:
- The authors define CSMs using set theory and commutative diagrams to establish intrinsic properties independent of unit cell choices.
- A CSM is decomposed into a Deformation Gradient ( $S$ ) and a Shuffle (an atom-to-atom correspondence with no deformation).
- The concept of Sublattice Match (SLM) is introduced, defined by a triplet $(\tilde{L}_A, \tilde{L}_B, S)$ , where $\tilde{L}_A$ and $\tilde{L}_B$ are sublattices of the initial and final structures, respectively. This organizes all CSMs into a tree structure based on multiplicity ( $\mu$ ) and SLM.
- Congruence relations are defined to group CSMs that differ only by rigid transformations (rotations/translations) or symmetry operations, ensuring only unique mechanisms are enumerated.
Matrix Representation:
- SLMs are represented by Integer Matrix Triplets (IMTs) $(H_A, H_B, Q)$ , where $H_A$ and $H_B$ are Hermite Normal Forms (HNF) of integer matrices describing the sublattices, and $Q$ is a unimodular matrix. This representation eliminates redundancy.
- Shuffles are represented by Permutations with Class-wise Translations (PCT), consisting of a permutation $p$ and a set of translation vectors $t_i$ within the shuffle lattice.
Pruning Criteria:
To ensure the enumeration space is finite, the authors apply three physical constraints:
- Multiplicity ( $\mu$ ): An upper bound $\mu_{max}$ is set based on empirical observations that realistic mechanisms rarely require supercells larger than $\sim 10^1$ atoms.
- Strain Energy ( $w$ ): A threshold $w_{max}$ filters out deformation gradients with excessive elastic energy.
- Shuffle Distance ( $d$ ): A threshold $d_{max}$ limits the displacement of atoms, defined via a weighted root-mean-square distance of the standard shuffle.
Algorithms:
- Algorithm 1: Enumerates all noncongruent IMTs (SLMs) by sampling the deformation gradient space, rounding to the nearest integer matrices, and verifying constraints.
- Algorithm 2 & 3: Enumerate all noncongruent PCTs (shuffles) for a given SLM. Algorithm 3 optimizes the shuffle distance to find "bottom PCTs" (minimal $d$ ), while Algorithm 2 uses a branch-and-bound strategy (via Split and Fill functions) to exhaustively list all PCTs satisfying the distance constraint.

Key Contributions

Formalism: A mathematically rigorous, cell-independent definition of CSMs, SLMs, and their congruence relations, proving that only a finite number of noncongruent CSMs exist under practical physical constraints.
crystmatch Method: A computational tool that exhaustively enumerates all possible CSMs for a given SSPT, solving the CSM optimization problem under any geometric criterion without prior bias.
Database Generation: The ability to generate a complete database of CSMs, including thousands of new candidates, which can serve as inputs for high-throughput energy barrier calculations or machine learning.

Results
The method was applied to two case studies: the B1–B2 transition (NaCl to CsCl) and the graphite-to-diamond transition.

B1–B2 Transition:
- The enumeration reproduced all known mechanisms (Buerger, WTM, TKED, TGS, etc.) within 10 CPU minutes.
- It identified that the TGS mechanism has the lowest strain energy among enumerated candidates.
- The method systematically inferred deformation gradients consistent with experimental Orientation Relationships (ORs), revealing new deformation gradients ( $S_1, S_2$ ) that satisfy the Okai-Fujiwara-Onodera (OFO) OR without rotation.
- By increasing the multiplicity limit ( $\mu_{max}=12$ ), the authors found new candidates with lower strain energy than TGS and new OR-consistent gradients, demonstrating the method's ability to go beyond known mechanisms.
- Exhaustive enumeration of shuffles (using weighted atomic masses) revealed hundreds of thousands of CSMs, including the TKED mechanism, which was previously difficult to identify via pure geometric optimization.
Graphite-to-Diamond Transition:
- The study analyzed the effect of pressure (0–50 GPa) on lattice mismatch.
- Results showed that as pressure increases, the strain energy of deformation gradients with small shuffle distances decreases significantly, while others exhibit non-monotonic behavior.
- This suggests that the dominant SSPT mechanism may shift depending on pressure, a conclusion supported by the "lattice mismatch spectrum" generated by the method.

Significance and Claims
The paper claims that crystmatch provides a systematic, unbiased solution to the long-standing problem of identifying the correct CSM for SSPTs. By rigorously proving the finiteness of the search space and providing algorithms to traverse it, the method:

Reproduces all known deformation mechanisms and CSMs.
Reveals thousands of new candidate mechanisms that were previously inaccessible to intuition or partial optimization.
Serves as a foundation for comparing experimental phenomena, performing high-throughput energy barrier calculations (e.g., via NEB), and training machine learning models for transition prediction.
Integrates seamlessly with existing computational workflows with negligible computational cost compared to electronic structure calculations.

The authors emphasize that the method does not replace energy calculations but rather provides the necessary geometric inputs (the CSMs) to ensure that energy barrier calculations are performed on the correct, and potentially lowest-barrier, pathways. The tool is open-sourced and available as a command-line utility.

Technical Summary: Classification and Enumeration of Solid-Solid Phase Transition Mechanisms

More like this