2026 MRS Spring Meeting にて発表

2026年4月26日〜5月1日にアメリカ・ハワイ州ホノルルで開催される「2026 MRS Spring Meeting & Exhibit」にて、Matlantis Inc. のメンバーが2件の口頭発表を行います。
MRS Spring Meetingは、Materials Research Society（MRS）が主催する材料科学分野における世界最大級の国際会議です。本大会は4月26日から5月1日までの6日間、ホノルル市内4カ所の公式会場にまたがって開催され、学術界・産業界・政府機関・国立研究所から材料研究者が集結し、分野横断的な議論と研究交流を行います。今回の会議では、持続可能な製造、先端計測、エネルギー材料などの分野で最先端の成果が発表される予定です。
当社からは、汎用機械学習ポテンシャル（uMLIP）「PFP」を活用した触媒・電池材料開発の最新研究成果について、以下の2件の発表を行います。

発表1

Accelerating Discovery of High Entropy Alloy Catalysts for Nitrate Reduction to Ammonia with a Universal Machine Learning Interatomic Potential

Ammonia is critical as a fertilizer feedstock and an emerging hydrogen carrier, but the energy-intensive, carbon-heavy Haber-Bosch process has driven interest in alternatives such as electrocatalytic nitrate reduction (NO₃RR), which also helps mitigate nitrate pollution in water. High-entropy alloys (HEAs) are attractive catalyst candidates due to their diverse active-site environments, yet their vast compositional space is intractable for conventional DFT-based screening. In this work, we use a universal machine learning interatomic potential to rapidly map Gibbs free energies along the NO₃RR pathway across thousands of HEA surfaces (e.g., FeCoNiCuZn), and then leverage graph-neural-network latent embeddings as descriptors to train ML models that predict nitrate and ammonia binding energies and identify compositions with optimal reactant–product binding balance—an approach readily extensible to other challenging electrocatalytic reactions such as CO₂ reduction.

Date & Time (local): Sunday, April 26, 2026, 8:30 AM
Session: EN02.01 — ML- and AI-Enabled Design of Sustainable Materials
Symposium: EN02 — Emergent Materials Paradigms—Catalytic, Photonic and Energy Conversion Technologies for a Sustainable Future
Venue: Sheraton Waikiki, Lanai Room, 2nd Floor
Presenter: Joshua Young (Matlantis, Inc.)

発表2

Accelerating Discovery of High Entropy Alloy Catalysts for Nitrate Reduction to Ammonia with a Universal Machine Learning Interatomic Potential

Discovering solid electrolytes for all-solid-state batteries is bottlenecked by the fact that high ionic conductivity arises from an intricate coupling of structural and vibrational properties, which single-modality machine learning models struggle to capture. To address this, we introduce a multimodal deep learning framework that integrates several physically motivated representations—long-range crystal structure, local atomic environments, and lattice dynamics—within a single architecture, learning cross-modal correlations to build a richer, more physically grounded material description. The approach outperforms single-modality baselines while offering interpretable insights into the factors governing ionic mobility, providing a practical route to accelerate solid electrolyte screening and uncover the structure–dynamics–property relationships that guide rational design of next-generation energy storage materials.

Date & Time (local): Tuesday, April 28, 2026, 8:45 AM
Session: MT01.03 — Physics-Informed, Expert Guided AI-Driven Experiments
Symposium: MT01 — Interfacing AI with Prior Knowledge and Human Expertise for Data-Efficient Autonomous Materials Research
Venue: Hilton Hawaiian Village, Tapa Conference Center, Iolani Suite VI, 2nd Floor
Presenter: Qingjie Li (Matlantis, Inc.)

イベント詳細・お申込み
イベントの詳細およびお申込みは、下記リンクよりご確認ください。
https://www.mrs.org/meetings-events/annual-meetings/2026-mrs-spring-meeting-exhibit