AutoEMX combines SEM-EDS automation and machine learning for high-speed quantitative powder composition analysis

How to apply this

With automated SEM-EDS and unsupervised machine learning, one can try to rapidly estimate phase-specific compositions in multiphase powders.

1. Prepare powder samples representatively and set the automated SEM-EDS measurement conditions consistently.
2. Preprocess the measurement data to reduce signal distortion according to particle shape and position, and construct features to be used for each phase.
3. Use unsupervised machine learning to find clusters or latent structures in the data, separate phase candidates, and estimate each phase composition.
4. Compare against reference samples or independent analysis to check relative error, reproducibility, and sensitivity to changes in measurement conditions.

Data needed

Multiphase powder samples, automated SEM-EDS maps or spectral datasets, and, if possible, reference compositions or independent chemical analysis values are needed.

Checking the result

Check whether phase-specific compositions are similar across repeated measurements, compare results before and after correction, and confirm the relative error and error distribution against independent analysis.

What can go wrong

If particle shape or phase separation is not sufficiently reflected, it is easy to misinterpret averaged signals as phase-specific compositions.

이 가이드는 원문의 방법을 바탕으로 AI가 제안한 적용 방향입니다. 원문에 적힌 내용이 아니며 같은 결과가 재현된다는 보장도 아닙니다.