Quickstart Guide

This guide introduces how to use AutoEMX for typical workflows:

  • Identify how many phases you have in your sample, and measure their compositions via SEM-EDS

  • Fit and optionally quantify a single EDS spectrum to evaluate the model performance

  • Measure the particle size distribution in your sample via SEM

  • Quantify the extent of intermixing of precursors prior a solid-state reaction

Warning

If this is the first time AutoEMX is run on your microscope, note that there are a few steps required to set it up before AutoEMX can be properly run. Refer to Maintainer docs.

Warning

Ensure the EDS detector is periodically recalibrated for optimal EDS quantification. See EDS Detector Calibration.

Expected runtime on a normal desktop computer:

  • Quantification typically takes a 0.5-3 minutes per spectrum.

  • For multiple spectra, quantification is parallelized, so runtime does not

    scale linearly with the number of spectra. 100 spectra can be quantified in 10 minutes with 10 CPU units.

  • Composition analysis (clustering/statistical analysis) typically takes a few seconds.

Workflows

Each workflow can be run in two ways, which use the same engine and write the same sample folders:

  • GUI: the AutoEMX GUI, a point-and-click interface running locally in your browser. Create its double-click app once with python -m autoemx.gui --create-launcher (Tutorial).

  • Scripts: the scripts in autoemx/scripts/, which require a minimal set of user-defined parameters.

EDS compositional analysis for phase identification

With one click, AutoEMX handles the full workflow from EDS spectral acquisition and quantification, to rule-based filtering of the quantified compositions and unsupervised machine-learning analysis to identify the different phase compositions in your sample.

  • GUI: Acquisition, Quantification and Analysis tabs, with interactive 3D plots of the results (Tutorial).

  • Scripts: run_acquisition_quant_analysis.py (Tutorial).

See full workflow at:

A. Giunto et al., Accurate SEM-EDS Quantification, Automation, and Machine Learning Enable High-Throughput Compositional Characterization of Powders, Nature Communications 17, 9735 (2026). https://doi.org/10.1038/s41467-026-76633-x

Fit and quantify EDS spectra

Fit–and optionally quantify– EDS spectra acquired using AutoEMX or exported by commercial EDS software (.msa, .emsa, .msg spectra files), one at a time to evaluate the model performance, or in batch.

  • GUI: Single spectrum tab for one spectrum; Quantification tab for many spectra, including folders of exported spectra (Tutorial).

  • Scripts: fit_quant_single_autoemx_spectrum.py (Tutorial) or fit_quant_single_msa_spectrum.py (Tutorial) for one spectrum, which print the full process in the terminal, the employed fit parameters and their final values; quantify_external_spectra.py for exported spectra in batch (Tutorial).

Measure particle size distribution via SEM

Have AutoEMX collect multiple images, detect particles, and quantify their size distribution.

  • Scripts: collect_particle_statistics.py (Tutorial). Not available in the GUI yet.

Quantify the extent of intermixing in precursor powders

Use EDS to evaluate the extent of spatial intermixing of different precursor powders, known to affect the output of solid-state reactions. AutoEMX offers a method to quantify the intermixing, helping the rationalization of impurity formation in solid-state reactions. See for example Fig. 6 in:

Chem. Mater. 2025, 37, 6807−6822 (https://pubs.acs.org/doi/10.1021/acs.chemmater.5c01573)

  • GUI: as for the compositional analysis, with the precursor options (Tutorial, last section).

  • Scripts: Run_Acquisition_PrecursorMix.py (Tutorial).