Particle Size Distributions from Electron Microscopy Images: Avoiding Pitfalls

Area of Science:

• Materials Science
• Catalysis
• Data Analysis

Background:

• Accurate particle size distribution is vital for heterogeneous catalysts.
• Manual compilation from electron micrographs and histogram representation are common.
• Transparent selection criteria for histogram bin width (w) are often lacking.

Purpose of the Study:

• To demonstrate data-driven estimators for selecting histogram bin width (w).
• To avoid operator bias in particle size distribution analysis.
• To reveal hidden features in particle size data.

Main Methods:

• Analysis of synthetic data to validate bin width selection.
• Survey of published particle size distribution data.
• Application of statistically founded bin width estimators.

Main Results:

• Operator's bias can be avoided using raw data-based estimators for bin width (w).
• Many published studies used excessively large bin widths.
• Statistically founded methods revealed features missed in original analyses, including bimodal distributions.

Conclusions:

• Data-driven bin width selection is essential for accurate particle size distribution.
• Implementing statistically founded estimators enhances objectivity and reveals subtle data characteristics.
• A suggested workflow ensures unbiased generation of particle size distributions from experimental data.