Inferring Chemical Reaction Rates from a Sequence of Infrared Spectra
MetadataShow full item record
Many chemical compounds used by the energy and agricultural industries introduce large amounts of arsenic into the environment. As this poses serious health and environmental risks, designing safe and effective decontaminating agents remains an active research area. To do this, it is crucial to understand the chemical kinetics between arsenic and certain geochemicals at the molecular level; of particular interest are the reaction rate constants which describe the behaviour and properties of arsenic in relation to different chemicals. These rates can be inferred from a time series of individual concentration measures of all constituent chemicals in a mixture. However, current laboratory technology cannot produce such measures but instead produces time series of infrared spectra, from which individual chemical concentrations must be deconvoluted. Existing techniques to analyze such data focus on minimizing modeling assumptions and point estimation. In this thesis, we propose a fully specified parametric statistical model directly relating the rate constants to the spectra. This model drastically reduces the number of free parameters, offers statistically principled uncertainty estimates for parameters of interest and provides the added flexibility of incorporating important prior information, which current methodologies do not seem to account for. We further apply the model to experimental data in order to compare two plausible models of arsenic neutralization.
Cite this version of the work
Peter Starszyk (2016). Inferring Chemical Reaction Rates from a Sequence of Infrared Spectra. UWSpace. http://hdl.handle.net/10012/10207