Algorithms for error-compensated kinetic determinations without prior knowledge of reaction order or rate constant

Jan Anders Larsson, Purdue University

Abstract

This thesis describes development and evaluation of two new algorithms for error-compensating predictive kinetic determinations. With the new algorithms it is possible to calculate reaction order, rate constant, and initial and final values of detector signal from several signal vs. time data points from a single kinetic run. Curve-fitting methods are used to obtain values of these parameters that give the best fit of the model to the data. The first procedure presented uses nonlinear regression of an equation expressing signal as a function of time. The second algorithm utilizes a linearized version of the rate equation and is intended primarily to provide initial estimates of these kinetic parameters for the first curve-fitting method. However, under some circumstances, the linearized method can provide sufficiently reliable results that subsequent processing by other methods is not needed. Although intended primarily to be used to compute signal changes between t = 0 and $\infty$, the methods can also be used to determine reaction orders and rate constants. Although the first algorithm fails for reaction orders of zero and unity, it works well for orders between these values and orders greater than unity. The second algorithm provides less reliable results than the nonlinear curve-fitting method for some situations (eg. reaction orders greater than about two, low data densities) but has the advantages that it is faster and is applicable to reaction orders near unity. Simulated data with different levels of superimposed noise, reaction orders, rate constants, signal change and data densities are used to evaluate the algorithms. Results from quantitation of acetoacetate in urine are also presented.

Degree

Ph.D.

Advisors

Pardue, Purdue University.

Subject Area

Analytical chemistry

COinS