AVS 61st International Symposium & Exhibition
    Surface Science Tuesday Sessions
       Session SS-TuP

Paper SS-TuP13
Unveiling Hidden Information in Temperature-Programmed Desorption-Reaction Data: Identification of Desorbing Compounds by Their Desorption and Fragmentation Patterns

Tuesday, November 11, 2014, 6:30 pm, Room Hall D

Session: Surface Science Poster Session
Presenter: JuanCarlos Rodríguez-Reyes, Universidad de Ingeniería y Tecnología, UTEC, Peru
Authors: J.C.F. Rodríguez-Reyes, Universidad de Ingeniería y Tecnología, UTEC, Peru
J.-M. Lin, University of Delaware
J. Zhao, University of Delaware
A.V. Teplyakov, University of Delaware
Correspondent: Click to Email
Temperature-programmed desorption/reaction (TPD/R) experiments are very powerful for providing information in several fields related to surface science, including heterogeneous catalysis, thin film precursor analysis and thermal treatments. Data can be, however, very complex and difficult to extract, especially when competing reaction pathways yield products with similar spectrometric features. We present a mathematical method for analyzing TPD/R data, which is able to extract, for each desorbing compound, its desorption pattern and its fragmentation pattern. This methods is called multivariate curve resolution (MCR) and, briefly, requires the organization of data (e.g. "n" m/z traces followed over "t" temperature points) as a (n x t) matrix, which can be seen as the product of two matrices, (n x k) and (k x t), where k is calculated by considering variations and correlations between m/z traces. Interestingly, k turns out to be the number of components (compounds desorbing from a surface), and its value is limited not only by correlations between data points, but also by the fact that intensities in the two matrices cannot be negative. Therefore the two matrices correspond to the fragmentation pattern, (n x k) matrix, and to the desorption pattern, (k x t) matrix. Since it is a mathematical method, MCR is applicable even when no previous knowledge of the system under investigation is available. However, any available information can be used as constraints that guide the outcome, increasing the accuracy of the resolution. The usefulness of this method is demonstrated using datasets from a variety of surface reactions, including the reaction of ethyl halides with a Si(100) surface and the thermal decomposition of the TiN precursor tetrakisdimethylamido titanium, TDMAT.