XRD Analyzer Documentation

Understanding the Analysis Pipeline

The XRD Analyzer platform implements a standard four-phase digital signal processing pipeline for powder X-ray diffraction data. Because all processing runs inside a WebAssembly virtual machine directly in your client's web browser, understanding how these subroutines execute helps you configure parameters for publication-quality results.

Phase 1: Pre-processing & Baseline Removal

Experimental XRD scans often contain backgrounds caused by thermal diffuse scattering, sample holder scatter, or fluorescence. Before fitting profiles, you must isolate the crystalline reflection signals. Our tool uses the SNIP algorithm from the pybaselines package to clip backgrounds iteratively. You can also strip $K\alpha_2$ radiation peaks using the Rachinger shift algorithm, preventing split peak profiles.

Phase 2: Peak Detection & Centroid Location

Once the background is subtracted, a peak-finding subroutine scans the spectrum. It uses the SciPy signal processing module to identify local maxima based on your custom height, prominence, and width parameters. The coordinates of these peaks are converted to crystallographic lattice d-spacings via Bragg's Law.

Phase 3: Mathematical Profile Fitting

Diffraction peaks are not simple delta functions; their shape represents the instrument geometry and crystallographic defects. Our system runs non-linear least-squares regressions to fit each detected peak with a Pseudo-Voigt model. The Pseudo-Voigt function is a linear combination of Gaussian and Lorentzian profiles, which allows it to model both instrument-induced shapes and grain-broadening effects.

Phase 4: Sizing & Crystallographic Phase Matching

The final phase calculates crystallite sizes using the Scherrer equation. If you upload crystal structure definitions in CIF format, our engine calculates theoretical Bragg reflections, overlays them on your plot, and computes Figure-of-Merit (FOM) coefficients to identify phase contributions.