Uncertainty-Aware Machine Learning for Small-Angle X‑ray Scattering Analysis in Autonomous Experimentation

Area of Science:

• Materials Science
• Nanotechnology
• Data Science

Background:

• Small-angle X-ray scattering (SAXS) offers real-time nanoscale structural insights during synthesis.
• Automated SAXS data analysis is crucial for autonomous materials discovery but remains challenging.
• Current methods lack the speed and reliability needed for closed-loop experimentation.

Purpose of the Study:

• To develop a machine-learning approach for automated SAXS data analysis.
• To enable fast, reliable, and uncertainty-aware parameter extraction for closed-loop nanoparticle synthesis.
• To improve the efficiency of autonomous nanomaterials discovery workflows.

Main Methods:

• A Random Forest (RF) regression model was trained on synthetic SAXS data.
• The model predicts nanoparticle radius, polydispersity, and background parameters directly from intensity profiles.
• Uncertainty quantification (UQ) was integrated using ensemble standard deviation.

Main Results:

• The RF model accurately predicts structural parameters on synthetic data.
• Combining fit-quality metrics with UQ thresholds reliably identifies accurate estimates.
• Experimental SAXS data from gold nanoparticle synthesis was classified using UQ.
• RF-based analysis in a closed-loop campaign showed faster convergence than conventional fitting.

Conclusions:

• Machine-learning-driven SAXS analysis significantly enhances autonomous nanomaterials synthesis.
• Uncertainty-aware analysis improves the efficiency and robustness of closed-loop workflows.
• This approach accelerates the discovery and optimization of nanomaterials.