Total Pattern Analysis Using PF-4 Relational Databases

John Faber

International Centre for Diffraction Data (ICDD),Newtown Square, PA 19073, USA

The fundamental underpinnings for qualitative and quantitative phase analysis have rested with the description of diffraction results in terms of a concise peak-list, i.e., d-spacing/peak intensity pairs. These are taken from the experimental data, usually after background subtraction and alpha-2 stripping. However, there are obvious short-comings in this approach. Information details concerning background variations, amorphous components in the scattering pattern and peak-width information have been purposefully removed from consideration (e.g. stress/strain). However, the effects of the instrumental resolution function and peak broadening persist.

In this paper, we will follow these historical developments and show qualitatively how Bragg peak resolution affects the quality of pattern matching using Smith-Snyder FOM criteria[1]:

,

where N is the number of observed Bragg reflections, is the average error, and is the number of possible reflections. Specifically, we will use calculated patterns obtained from the ICDD PDF-4 databases and show how symmetry and peak breadth affect the FOM. Residuals derived from unit cell least-squares analyses also correlate with these observations.

In contrast, we will examine methods for Total Pattern Analyses as implemented in DDView+, the front-end software in the PDF-4 relational databases [2-3]. We will explore how TPA helps us to recover search-indexing details that have been overlooked or lost using data reduction methods. Along the way, we will review several current developments in total pattern matching techniques.

[1] Smith, G. S. and Snyder, R. L., J. Appl. Cryst. 12, 60-65 (1979).

[2] J. Faber, C. A. Weth and J. Bridge, J. Powder Diffraction 19, 26-30 (2004).

[3] T. G. Fawcett, S. N. Kabbekodu, J. Faber, F. Needham and F. McClune, J. Powder Diffraction 19, 20-25 (2004).