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

Aug 29, 2026

Upload by Rizky

Melting temperature is essential for selecting and designing high-temperature materials, yet first-principles finite-temperature approaches demand extensive sampling and high CPU costs. SLUSCHI-UP provides a deployed web service that estimates melting temperatures using SLUSCHI small-cell solid–liquid coexistence simulations and statistical analysis of many short trajectories. It couples the SLUSCHI workflow to selectable pretrained universal machine-learning interatomic potentials with asynchronous GPU execution. Users submit structures via Materials Project IDs or POSCAR, run queued calculations without local software installation, and receive estimates with diagnostics. Validation on MeltBench indicates screening-level accuracy while preserving common uMLIP and finite-size limitations.

LanguageEnglish

CategoryTechnology

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Introduction

Motivation: Importance and difficulty of determining melting temperature

Computational strategies for estimating Tm

SLUSCHI background and small-cell coexistence approach

Move toward web-accessible, reproducible infrastructure with uMLIPs

SLUSCHI-UP: deployed web service workflow

arXiv :2606 .04973v1 [ cond-mat .mtrl-sci ] 3 Jun 2026 SLUSCHI-UP: A Web Infrastructure for SLUSCHI Melting-Temperature Calculations Using Universal Machine-Learning Interatomic Potentials Qi-Jun Hong 1, ∗ 1 Department of Materials Science and Engineering, Arizona State University, Tempe, Arizona 85287, USA (Dated: June 4, 2026) Melting temperature is a critical property for high-temperature materials design, but firstprinciples 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 meltingtemperature 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. I. INTRODUCTION Melting temperature, Tm , is one of the most important thermodynamic properties for materials selection, phase-diagram construction, and the design of hightemperature materials. It constrains the use of refractory ceramics, turbine and hypersonic materials, metallurgical systems, additive-manufacturing feedstocks, molten salts, geophysical minerals, and multicomponent alloys. Despite its importance, Tm remains difficult to determine in high-throughput settings. Experiments can be slow, hazardous, or unavailable for unstable and multicomponent compounds, while first-principles calculations are computationally demanding because they require extensive finite-temperature sampling. Conventional approaches such as thermodynamic integration and largecell solid–liquid coexistence simulations can consume on the order of 104–105 CPU-hours per material, making routine screening impractical. Several computational strategies have been developed to estimate melting temperatures. Free-energy methods compare the relative thermodynamic stability of solid and liquid phases, with representative approaches including particle-insertion chemical potentials, pairdistribution-function entropy, two-phase thermodynamics, thermodynamic integration, and generalized entropy ∗ [qhong@alumni.caltech.edu](qhong@alumni.caltech.edu) calculations [1–4] . Large-cell coexistence simulations directly evolve a solid–liquid interface in molecular dynamics and determine the melting temperature from interface motion or phase stability [5–8] . The small-size coexistence approach provides a statistical alternative: instead of stabilizing a large interface, it launches many small coexistence simulations and infers the melting point from the probability that an initially mixed solid–liquid configuration evolves

FAQs

What problem does SLUSCHI-UP address in melting-temperature calculations?

It reduces the high computational cost of first-principles melting calculations by using small-cell solid–liquid coexistence simulations combined with statistical analysis over many short molecular-dynamics trajectories.

How do users submit inputs to SLUSCHI-UP?

Users provide a crystal structure via a Materials Project identifier or pasted POSCAR content, choose a supported uMLIP backend, and submit a queued melting calculation through an email-verified web interface.

Which uMLIP backends does SLUSCHI-UP currently support, and how is accuracy evaluated?

Production supports mace-mpa-0-medium, Allegro-OAM-L, and DPA-3.2-5M-OMat24, with additional beta models available. Performance is assessed on MeltBench validation, reporting raw coexistence mean absolute errors in the screening range (about 178–327 K for production backends).

Aug 29, 2026

Upload by Rizky