Search for content and authors
 

A robust methodology for high throughput identification of mixture of crystallographic phases from powder diffraction data

Laurent A. Baumes 1Manuel Moliner 1Remi Gaudin 2Nicolas Nicoloyannis 2Avelino Corma 1

1. Instituto de tecnologia quimica CSIC-UPV (ITQ), av de los naranjos, Valencia 46022, Spain
2. ERIC University of Lyon II Lumiere, 5 Av Pierre Mendes France, Lyon 69676, France

Abstract

The knowledge of the structure at atomistic/molecular level of a material is required for any advanced understanding of its performance. Fortunately, characterization techniques are developing rapidly, with a major effort for parallelization and high-throughput (HT). On the other hand, the corresponding high volume of generated data requires an automatic data analysis system not to slow down the whole process. Most characterization responses take the form of “series” where the measurements follow an ordered variable; eg temperature (TGA, TPD), time (GC), wave length (photoluminescence, IR, UV), voltage (Voltammetry), angles (XRD)… Computing techniques for the treatment of such naturally ordered data, represents an actual challenge in data mining research.


Here we focus on zeolite since they are versatile materials for an increasing number of applications. The automatic determination of crystalline phases through X-ray diffraction data from powder represents a study of great interest for both chemistry and computer science. The algorithms initially created for multi-dimensional data are not effective when series are considered. We propose an approach called Adaptable Time Warping. ATW is a learning process which uses a genetic algorithm. According to a given classification problem, ATW approach adapts itself for better capturing data specificity. ATW effectiveness is demonstrated through numerous tests: 20 benchmarks and 3 real datasets of zeolite investigations [1, 2, 3]. The originality and power of the technique are clearly demonstrated when the complete phase diagram corresponding to the research space of the zeolite ITQ33 is automatically revealed (see Table1). ATW appears as innovative and a very promising methodology considering the numerous potential applications in chemistry.

Baumes_Table.gif

1. M. Moliner et al., Microp. Mesop. Mater. 2005, 78, 73-81.

2. A. Corma et al., Nature. 2002, 418, 514-517

3. A. Corma et al., Nature. 2006, 443, 842-845

 

Legal notice
  • Legal notice:
 

Related papers

Presentation: Invited at E-MRS Fall Meeting 2007, Symposium G, by Laurent A. Baumes
See On-line Journal of E-MRS Fall Meeting 2007

Submitted: 2007-06-18 12:44
Revised:   2009-06-07 00:44