arXiv subjects

Qi-Jun Hong

Publications and source records attributed to Qi-Jun Hong.

17 recordsLinked to original sources

arXiv2026

SLUSCHI-UP: A Web Infrastructure for SLUSCHI Melting-Temperature Calculations Using Universal Machine-Learning Interatomic Potentials

Melting temperature is a critical property for high-temperature materials design, but first-principles melting calculations based on finite-temperature molecular dynamics can require substantial computational resources. The SLUSCHI method reduces this cost by using small-cell solid--liquid coexistence simulations and statistical analysis of many short molecular-dynamics trajectories. Here I present SLUSCHI-UP, a deployed web service for atomistic melting-temperature estimation that couples the SLUSCHI workflow to selectable pretrained universal machine-learning interatomic potentials (uMLIPs) and asynchronous GPU execution. Users submit a crystal structure through a Materials Project identifier or POSCAR input, select a uMLIP backend, and launch a queued melting calculation without local installation of simulation software. The current production interface supports mace-mpa-0-medium, Allegro-OAM-L, and DPA-3.2-5M-OMat24, while beta deployments expose additional models. On the compact MeltBench-10 validation set, the three production backends produce raw coexistence mean absolute errors in the range of approximately 178--327 K. Across the broader set of materials tested so far in MeltBench, the current deployed-job snapshot contains 119 raw uMLIP entries, and PBE-corrected Allegro-OAM-L predictions reach a mean absolute error of approximately 166 K. These values should be interpreted as screening-level infrastructure validation rather than a definitive ranking of uMLIPs. The results demonstrate that SLUSCHI-UP provides a practical, provenance-aware deployment layer between fast scalar melting-temperature predictors and much more expensive first-principles coexistence calculations, while retaining the usual limitations of uMLIP transferability, finite-size sampling, and high-temperature trajectory stability.

cond-mat.mtrl-sci↗

arXiv2026

On the Mathematics of Information-Thermodynamics

We present a validation of the asdf method, an information-theoretic framework for computing thermodynamic entropy from molecular configurations. The method reformulates entropy estimation as the Shannon entropy of a residual mapping distribution defined between two decorrelated microstates. We demonstrate analytically that for the closed-form Hamiltonians with known solutions, the classical ideal gas and the one-dimensional harmonic oscillator's entropy obtained from the compressibility of the residual mapping object reproduces the exact thermodynamic entropy. In each case, the conditional entropy of the residual mapping object with respect to an uncorrelated microstate is shown to coincide with the ensemble entropy derived from the canonical partition function. These results establish consistency between the asdf formalism and classical statistical mechanics for analytically solvable systems. We further discuss how the framework generalizes to interacting Hamiltonians. The analysis supports the interpretation of thermodynamic entropy as an information measure encoded geometrically in inter-microstate mappings and motivates application of the method to complex condensed phases.

cond-mat.stat-mech↗

arXiv2025

Thermodynamic Entropy as Information -- A compression-based demonstration of the Shannon-Boltzmann equivalence in condensed matter

We demonstrate that Shannon's information entropy and the thermodynamic entropy of Boltzmann and Gibbs are quantitatively equivalent for real condensed-matter systems. By interpreting atomic configurations as information sources, we compute entropy directly from the compressibility of molecular-dynamics trajectories, without physical partitioning or empirical modeling. A custom lossy-compression algorithm measures the minimum number of bits required to describe a microstate at finite precision, and this bit count maps exactly to thermodynamic entropy through the Shannon-Boltzmann relation. The method reproduces benchmark entropies for metals, semiconductors, oxides, and refractory ceramics in both solid and liquid phases, establishing information as the fundamental quantity underlying thermodynamic disorder. This equivalence unifies information theory and statistical mechanics, providing a general and computationally efficient framework for determining entropies and free energies directly from atomic data.

cond-mat.stat-mech↗

arXiv2025

Extending SLUSCHI for Automated Diffusion Calculations

We present an extension of the SLUSCHI package (Solid and Liquid in Ultra Small Coexistence with Hovering Interfaces) to enable automated diffusion calculations from first-principles molecular dynamics. While the original SLUSCHI workflow was designed for melting temperature estimation via solid-liquid coexistence, we adapt its input and output handling to isolate the volume search stage and generate one production trajectory suitable for diffusion analysis. Post-processing tools parse VASP outputs, compute mean-square displacements (MSD), and extract tracer diffusivities using the Einstein relation with robust error estimates through block averaging. Diagnostic plots, including MSD curves, running slopes, and velocity autocorrelations, are produced automatically to help identify diffusive regimes. The method has been validated through representative case studies: self- and inter-diffusion in Al-Cu liquid alloys, sublattice melting in Li_7La_3Zr_2O_12 and Er_2O_3, interstitial oxygen transport in bcc and fcc Fe, and oxygen diffusivity in Fe-O liquids with variable Si and Al contents. Viscosity and diffusivity are linked through the Stokes-Einstein relation, with composition dependence assessed via simple linear mixing. This capability broadens SLUSCHI from melting-point predictions to transport property evaluation, enabling high-throughput, fully first-principles datasets of diffusion coefficients and viscosities across metals and oxides.

cond-mat.mtrl-sci↗

arXiv2025

Structure and stability of 7:3 rare earth oxide-phosphates: a combined ab initio and experimental study

Rare earth oxide-phosphates (REOPs) form a largely unexplored family of refractory lanthanides and yttrium compounds with general formula RExOy(PO4)z. They are of interest for applications ranging from thermal barrier coatings to catalysts and magnetic materials. At least four REOPs phases were experimentally identified with RE/P ratios from 7:3 to 6:1, however the structures were solved only for 3:1 phases (RE3O3(PO4)). In this work we report the structure for the 7:3 phases (RE7O6(PO4)3) derived by ab initio analysis of models based on previously reported oxide-vanadate analogues. The most stable structures for all 7:3 REOPs were found to be isotypic, adopting monoclinic symmetry with space group P21/c. The structures were validated by comparison of their powder X-ray diffraction patterns to those of synthesized La, Pr, Nd, Sm, Eu, Gd and Tb 7:3 phases (Rietveld refinement for all except Tb). Ab initio analysis of thermodynamic stability showed that all 7:3 REOPs are unstable at 0 K toward decomposition to REPO4 and RE3PO7 or RE2O3. The entropy contribution stabilizes RE7O6(PO4)3 phases for light rare earth elements above 1000 K, however, starting with Dy, computationally predicted stabilization temperature is higher than estimated melting points of RE7O6(PO4)3, which is consistent with observed synthesis pattern.

cond-mat.mtrl-sci↗

arXiv2025

Achieving accurate entropy and melting point by ab initio molecular dynamics and zentropy theory: Application to fluoride and chloride molten salts

We have recently developed a breakthrough methodology for rapidly computing entropy in both solids and liquids by integrating a multiscale entropy approach (known as zentropy theory) with molecular dynamics (MD) simulations. This approach enables entropy estimation from a single MD trajectory by analyzing the probabilities of local structural configurations and atomic distributions, effectively addressing the long-standing challenge of capturing configurational entropy. Here, we demonstrate the power of this method by predicting entropies, enthalpies, and melting points for 25 binary and ternary chlorite- and fluoride-based molten salts using ab initio MD (AIMD) simulations. The strong agreement between our predictions and experimental data underscores the potential of this approach to transform computational thermodynamics, offering accurate, efficient, and direct predictions of thermodynamic properties across both solid and liquid phases.

cond-mat.mtrl-sci↗

arXiv2024

Computational investigation of formation enthalpies and phase stability for rare earth oxyphosphates

Rare earth phosphates have garnered significant interest due to their versatile properties, including high chemical stability, thermal resistance, luminescence, and the ability to adopt various crystalline structures. Density functional theory (DFT)-based ab initio methods have become essential tools for complementing experimental studies. In this paper, we performed DFT calculations on rare earth (RE; here considered as lanthanides + Y) oxyphosphates to examine their formation enthalpies and phase stability. The calculations were conducted using the GGA-PBE and r2SCAN exchange-correlation functionals. Our results indicate that both functionals predict similar phase stabilities for REPO4 and RE3PO7. However, the r2SCAN functional provides significantly more accurate formation enthalpies for the monazite and xenotime REPO4, aligning closely with experimental data. Furthermore, the inclusion of lattice vibrational entropy enhances the free energy predictions, leading to improved agreement with experimental observations on phase stability.

cond-mat.mtrl-sci↗

arXiv2024

Accelerating material melting temperature predictions by implementing machine learning potentials in the SLUSCHI package

The SLUSCHI (Solid and Liquid in Ultra Small Coexistence with Hovering Interfaces) automated package, with interface to the first-principles code VASP (Vienna Ab initio Simulation Package), was developed by us for efficiently determining the melting temperatures of various materials. However, performing many molecular dynamics simulations for small liquid-solid coexisting supercells to predict the melting temperature of a material is still computationally expensive, often requiring weeks and tens to hundreds of thousands of CPU hours to complete. In the present paper, we made an attempt to interface the SLUSCHI package with the highly efficient molecular dynamics LAMMPS code and demonstrated that it achieves a much faster melting temperature determination, outperforming the original VASP-based approach by at least one order of magnitude. In our melting temperature calculations, the LAMMPS simulations were performed based on the LASP (Large-scale Atomic Simulation) machine learning potentials which are pre-built using first-principles data. Besides the dramatic CPU time reduction for melting temperature predictions the calculated melting temperatures for various materials (simple and transition metals, alloys, oxides and carbide) are reasonably accurate. Analysis of the calculated results shows that 60% of the melting temperatures are within 200 K of experimental values with the RMSE value of 187 K which is slightly worse than the first-principles DFT RMSE value of 151 K. Therefore, interfacing SLUSCHI with LAMMPS molecular dynamics simulations makes it possible to quickly screen out the best candidates from numerous materials in a much more efficient way, and facilitate the rational design of materials within the framework of the materials genome paradigm.

cond-mat.mtrl-sci↗

arXiv2024

Materials Properties Prediction (MAPP): Empowering the prediction of material properties solely based on chemical formulas

Predicting material properties has always been a challenging task in materials science. With the emergence of machine learning methodologies, new avenues have opened up. In this study, we build upon our recently developed Graph Neural Network (GNN) approach to construct models that predict four distinct material properties. Our graph model represents materials as element graphs, with chemical formula serving as the only input. This approach ensures permutation invariance, offering a robust solution to prior limitations. By employing bootstrap methods to train on this individual GNN, we further enhance the reliability and accuracy of our predictions. With multi-task learning, we harness the power of extensive datasets to boost the performance of smaller ones. We introduce the inaugural version of the Materials Properties Prediction (MAPP) framework, empowering the prediction of material properties solely based on chemical formulas.

cond-mat.mtrl-sci↗

arXiv2024

A generalized approach for rapid entropy calculation of liquids and solids

We build a comprehensive methodology for the fast computation of entropy across both solid and liquid phases. The proposed method utilizes a single trajectory of molecular dynamics (MD) to facilitate the calculation of entropy, which is composed of three components. The electronic entropy is determined through the temporal average acquired from density functional theory (DFT) MD simulations. The vibrational entropy, typically the predominant contributor to the total entropy, even within the liquid state, is evaluated by computing the phonon density of states via the velocity auto-correlation function. The most arduous component to quantify, the configurational entropy, is assessed by probability analysis of the local structural arrangement and atomic distribution. We illustrate, through a variety of examples, that this method is both a versatile and valid technique for characterizing the thermodynamic states of both solids and liquids. Furthermore, this method is employed to expedite the calculation of melting temperatures, demonstrating its practical utility in computational thermodynamics.

physics.comp-ph↗

arXiv2023

Deep learning for CALPHAD modeling: Universal parameter learning solely based on chemical formula

Empowering the creation of thermodynamic and property databases, the CALPHAD (CALculation of PHAse Diagrams) methodology plays a vital role in enhancing materials and manufacturing process design. In this study, we propose a deep learning approach to train parameters in CALPHAD models solely based on chemical formula. We demonstrate its application through an example of calculating the mixing parameter of liquids. This work showcases the integration of CALPHAD and deep learning, highlighting its potential for achieving automated comprehensive CALPHAD modeling.

cond-mat.mtrl-sci↗

arXiv2022

Melting temperature prediction via first principles and deep learning

Melting is a high temperature process that requires extensive sampling of configuration space, thus making melting temperature prediction computationally very expensive and challenging. Over the past few years, I have built two methods to address this challenge, one via direct density functional theory (DFT) molecular dynamics (MD) simulations and the other via deep learning graph neural networks. The DFT approach is based on statistical analysis of small-size solid-liquid coexistence MD simulations. It eliminates the risk of metastable superheated solid in the fast-heating method, while also significantly reducing the computer cost relative to the traditional large-scale coexistence method. Being both accurate and efficient (at the speed of several days per material), it is considered as one of the best methods for direct DFT melting temperature calculation. The deep learning method is based on graph neural networks that effectively handles permutation invariance in chemical formula, which drastically improves efficiency and reduces cost. At the speed of milliseconds per material, the model is extremely fast, while being moderately accurate, especially within the composition space expanded by the dataset. I have implemented both methods into automated computer code packages, making them publicly available and free to download. The DFT and deep learning methods are highly complementary to each other, and hence they can be potentially well integrated into a framework for melting temperature prediction. I demonstrated examples of applying the methods to materials design and discovery of high-melting-point materials.

cond-mat.mtrl-sci↗

arXiv2021

A melting temperature database and a neural network model for melting temperature prediction

I build a melting temperature database that contains approximately 10,000 materials. Based on the database, I build a machine learning model that predicts melting temperature in seconds. The model features graph neural network and residual neural network architecture. The root-mean-square errors of melting temperature are 90 and 160K for training and testing, respectively. The model is deployed online and is publicly available.

cond-mat.mtrl-sci↗

arXiv2020

Re-entrant melting of sodium, magnesium, and aluminum: General trend

Re-entrant melting (in which a substance's melting point starts to decrease beyond a certain pressure) is believed to be an unusual phenomenon. Among the elements, it has so far only been observed in a very limited number of species, e.g., the alkali metals. Our density functional theory calculations reveal that this behavior actually extends beyond alkali metals to include magnesium, which also undergoes re-entrant melting, though at the much higher pressure of ~300 GPa. We find that the origin of re-entrant melting is the faster softening of interatomic interactions in the liquid phase than in the solid, as pressure rises. We propose a simple approach to estimate pressure-volume relations and show that this characteristic softening pattern is widely observed in metallic elements. We verify this prediction in the case of aluminum by finding re-entrant melting at ~4000 GPa. These results suggest that re-entrant melting may be a more universal feature than previously thought.

cond-mat.mtrl-sci↗

arXiv2017

A Tetrahedron-tiling Method for Crystal Structure Prediction

Reliable and robust methods of predicting the crystal structure of a compound, based only on its chemical composition, is crucial to the study of materials and their applications. Despite considerable ongoing research efforts, crystal structure prediction remains a challenging problem that demands large computational resources. Here we propose an efficient approach for first-principles crystal structure prediction. The new method explores and finds crystal structures by tiling together elementary tetrahedra that are energetically favorable and geometrically matching each other. This approach has three distinguishing features: a favorable building unit, an efficient calculation of local energy, and a stochastic Monte Carlo simulation of crystal growth. By applying the method to the crystal structure prediction of various materials, we demonstrate its validity and potential as a promising alternative to current methods.

cond-mat.mtrl-sci↗

arXiv2017

Free energy calculation of mechanically unstable but dynamically stabilized bcc titanium

The phase diagram of numerous materials of technological importance features high-symmetry high-temperature phases that exhibit phonon instabilities. Leading examples include shape-memory alloys, as well as ferroelectric, refractory, and structural materials. The thermodynamics of these phases have proven challenging to handle by atomistic computational thermodynamic techniques, due to the occurrence of constant anharmonicity-driven hopping between local low-symmetry distortions, while maintaining a high-symmetry time-averaged structure. To compute the free energy in such phases, we propose to explore the system's potential-energy surface by discrete sampling of local minima by means of a lattice gas Monte Carlo approach and by continuous sampling by means of a lattice dynamics approach in the vicinity of each local minimum. Given the proximity of the local minima, it is necessary to carefully partition phase space by using a Voronoi tessellation to constrain the domain of integration of the partition function, in order to avoid double-counting artifacts and enable an accurate harmonic treatment near each local minima. We consider the bcc phase of titanium as a prototypical examples to illustrate our approach.

cond-mat.mtrl-sci↗

arXiv2017

An epicycle method for elasticity limit calculations

The task of finding the smallest energy needed to bring a solid to its onset of mechanical instability arises in many problems in materials science, from the determination of the elasticity limit to the consistent assignment of free energies to mechanically unstable phases. However, unless the space of possible deformations is low-dimensional and a priori known, this problem is numerically difficult, as it involves minimizing a function under a constraint on its Hessian, which is computionally prohibitive to obtain in low symmetry systems, especially if electronic structure calculations are used. We propose a method that is inspired by the well-known dimer method for saddle point searches but that adds the necessary ingredients to solve for the lowest onset of mechanical instability. The method consists of two nested optimization problems. The inner one involves a dimer-like construction to find the direction of smallest curvature as well as the gradient of this curvature function. The outer optimization then minimizes energy using the result of the inner optimization problem to constrain the search to the hypersurface enclosing all points of zero minimum curvature. Example applications to both model systems and electronic structure calculations are given.

cond-mat.stat-mech↗