[论文解读] From Photons to Electrons: Accelerated Materials Discovery via Random Libraries and Automated Scanning Transmission Electron Microscopy

Slautin, Boris, Barakati, Kamyar+11 more|arXiv (Cornell University)|Mar 21, 2026

Machine Learning in Materials Science被引用 0

一句话总结

该论文提出并展示了一种基于电子的、由 ML 支撑的方法，利用随机化学库和自动化 STEM 来加速高维材料发现，在有效覆盖方面比传统的光子方法更具优势。

ABSTRACT

The real-world implementation of materials prediction algorithms remains limited by persistent characterization bottlenecks in materials discovery, where photon-based probe techniques (e.g., XRD or Raman) impose long acquisition times and access latencies, restricting exploration to quasi-ternary composition spaces typically realized as compositional libraries. Here, we argue that a paradigm shift from photon- to electron-based characterization can realign materials characterization with modern high-throughput synthesis. We formulate cost functions and exploration strategies for STEM-based chemical and structural characterization and use Monte Carlo simulations to show that random chemical libraries, where compositionally distinct regions are co-located within a single specimen and interrogated in situ by electron spectroscopies, can sample high-dimensional composition and phase spaces with orders-of-magnitude greater effective coverage than conventional spread-library/X-ray approaches. We further demonstrate autonomous discovery on a laboratory STEM platform, where ML-based autotuning and scripted control enable iterative region selection and characterization without human intervention. Finally, we outline extensions to labeled or position-encoded libraries that preserve compositional and processing metadata, enabling joint exploration of composition and process spaces. Together, these results establish electron-based, ML-enabled STEM as a scalable pathway toward combinatorially rich materials discovery.

研究动机与目标

• Identify bottlenecks in current photon-based materials characterization that hinder high-throughput exploration.
• Propose a framework for electron-based (STEM) chemical and structural characterization using random libraries.
• Show that co-located compositionally distinct regions interrogated in situ yield higher effective coverage of composition/phase space.
• Demonstrate autonomous discovery on a laboratory STEM platform with ML-driven autotuning and scripted region selection.
• Outline extensions to labeled or position-encoded libraries to preserve metadata for joint composition and processing exploration.

提出的方法

• Formulate cost functions and exploration strategies for STEM-based chemical/structural characterization.
• Use Monte Carlo simulations to compare random libraries with conventional spread-library/X-ray approaches.
• Implement autonomous discovery on a lab STEM with ML-based autotuning for iterative region selection and characterization without human input.
• Demonstrate in situ interrogation via electron spectroscopies on a single specimen containing compositionally distinct regions.
• Discuss extensions to labeled or position-encoded libraries to preserve compositional and processing metadata.

实验结果

研究问题

• RQ1 Can random libraries interrogated by STEM achieve higher effective sampling of high-dimensional composition/phase spaces compared to traditional photon-based libraries?
• RQ2 How do ML autotuning and autonomous control influence the efficiency and throughput of STEM-based materials discovery?
• RQ3 What are viable strategies to preserve metadata for joint exploration of composition and processing spaces?
• RQ4 What is the potential impact of electron-based characterization on accelerating combinatorially rich materials discovery?

主要发现

• Monte Carlo simulations indicate random libraries can sample high-dimensional spaces with orders-of-magnitude greater effective coverage than conventional approaches.
• Autonomous, ML-enabled STEM workflows can perform iterative region selection and characterization without human intervention on a laboratory platform.
• Electron-based, in situ spectroscopy-enabled interrogation enables rapid exploration of compositionally distinct regions within a single specimen.
• The framework supports extensions to labeled or position-encoded libraries while preserving important metadata for joint composition and process exploration.