Powder XRD Phase Identification

For the identification and characterization of materials using XRD data the most common method is based on pattern matching. For this method to work a large amount of reference data must be available to realize the pattern search and match algorithms. Obtained XRD powder patterns can have multiple compounds and phases present, all of which will generate peaks, and distinguishing the peaks manually is a laborious process. Computational resources and databases allow the reliable searching and matching of the varied phases and compounds patterns to the sample pattern. The formation of these patterns varies significantly enough between similarly structured materials making each unique.^[1][2] Typically pattern matching will require matching at least 3 peaks to a known pattern.^[2]

In principle, for a well-behaved sample the pattern search and match will be straightforward. Modern proprietary software has several automated functions for specific instrumentation and batch processing capabilities, i.e.: PANalyticals HighScore,^[3] which is one of the software packages used in Aalto University. In practice, supporting data is necessary for high confidence in the analysis. Such supporting data are the lattice parameters and elemental composition. These are, at a basic level, initial criteria to run the search functions and limit the search-space, unless 100s of hits are desired. The most common refinement method used for profile matching is the Rietveld method, it is effective in matching the found peaks, with varying widths and overlaps, to a modeled profile, a key advantage over the least-squares regression methods,^[4][5] this method is implemented in almost all diffraction pattern softwares.^[3][6][7] After refinement, peaks are discretely defined and fed to the search and match software with the supporting data, returning possible matches for phases.

Figure 1. Illustrative figure for the most simple phase identification procedure possible. Top: An unidentified XRD pattern, only raw data, obtained from a white powder. Bottom: Steps taken to analyze the pattern including baseline correction (dark green smooth line), peak search (blue contours and vertical lines) and phase match (green and yellow vertical lines). The substance identified were Ca(OH)₂ and CaO according to pattern matching. Figure: Patrik Eskelinen

Powder Pattern Databases

There are a range of databases available with crystallographic and diffraction data available. The majority of these databases are commercial or non-public, and such datasets have been kept since the early 20th century, i.e.: the Powder Diffraction File (PDF) since 1941.^[8] Table 1 below is compiled to provide an overview of the best-known databases for diffraction data, most of these are also included in International Union of Crystallography (IUCr) webpage,^[9] which is partially outdated. Observe that as of 2018 the ONLY well maintained open-access database available is COD, which can be implemented freely with both open-source and proprietary pattern matching software (i.e.: the free software FullProf,^[6] combined with QualX2,^[10] or Full Profile Search Match^[7]).

Abbreviations; ICDD – International Center for Diffraction Data, NIST – National Institute of Standards & Technology, ICDS – Inorganic Crystal Structure Database, COD – Crystallography Open Database, MPDS – Materials Phases Data System, JST – Japan Science Technology Corp, NIMS – National Institute for Materials Science , P2D2 – Predicted Powder Diffraction Database, CSD – Cambridge Structural Database, LBNL – Lawrence Berkeley National Laboratory

References