Application of combined multivariate techniques for the description of time‐resolved powder X‐ray diffraction data

• Published in: Journal of applied crystallography Vol. 50; no. 2; pp. 451 - 461
• Main Authors: Taris, Alessandra, Grosso, Massimiliano, Brundu, Mariarosa, Guida, Vincenzo, Viani, Alberto
• Format: Journal Article
• Language: English
• Published: 5 Abbey Square, Chester, Cheshire CH1 2HU, England International Union of Crystallography 01.04.2017; Blackwell Publishing Ltd
• Subjects: amorphous content; chemically bonded ceramics; Conversion; Crystal structure; Crystallography; Diffraction; Diffraction patterns; Evolution; Factor analysis; in situ X‐ray powder diffraction; Magnesium oxide; multivariate curve resolution; Potassium; Statistical methods; statistical total correlation spectroscopy; X rays
• ISSN: 1600-5767; 0021-8898; 1600-5767
• DOI: 10.1107/S1600576717001753

In this work, multivariate statistical techniques are employed to determine patterns and conversion curves from time‐resolved X‐ray powder diffraction data. For these purposes, time‐window statistical total correlation spectroscopy is introduced for the pattern matching of the crystalline phase and is shown to be effective even in the case of overlapping peaks. When combined with evolving factor analysis and multivariate curve resolution–alternating least squares, this technique allows a definite estimation of patterns and conversion curves. The procedure is applied to in situ synchrotron powder diffraction patterns to monitor the setting reaction of magnesium potassium phosphate ceramic (MKP) from magnesia (MgO) and potassium dihydrogen phosphate. It is shown that the phases involved in the reaction are clearly distinguished and their evolution is correctly described. The conversion curves estimated with the proposed procedure are compared with the ones determined with the peak integration method, leading to an excellent agreement (Pearson's correlation coefficient equal to 0.9995 and 0.9998 for MgO and MKP, respectively). The approach also allows for the detection and description of the evolution of amorphous phases that cannot be described through conventional analysis of powder diffraction data. A method combining different multivariate techniques is applied to the analysis of large datasets of time‐resolved X‐ray powder diffraction patterns. The method is based on time‐window statistical total correlation spectroscopy, which, together with evolving factor analysis and multivariate curve resolution, allows for the identification and the description of the time evolution of the crystalline as well as the amorphous fractions in the sample, in a semi‐automated fashion.