Modelling of evanescent field immunosensors
Several factors affect the viability of biosensor design. This thesis presents the development of a computer-based model that will enable the sources and effects of noise and variations in concentrations within an evanescent field immunosensor to be analysed. The model was developed as a series of modules, each representing one aspect of the sensor, which when linked provide a simulation of the whole sensor. A complete solution of the complex biochemical reactions involved in the immunoassay module was achieved using a Markov chain approach. More traditional methods of solving sets of equations, such as optimisation, genetic algorithms and simulated annealing, all failed to produce satisfactory results. Two alternative assays, a sandwich and a competitive assay, are presented. The light module details the modelling of the coupling into a planar monomode waveguide and calculation of fluorescence excited by the resulting evanescent field using standard electromagnetic formulae. However, both beam divergence and scattering from the immobilised antibody layer were incorporated into the model. Two alternative coupling techniques were modelled, prism coupling and coupling through a "resonant mirror" multilayer. The detection system modelled the amplification of the fluorescence by a photomultiplier tube. The resulting model represents the most rigorous modelling undertaken in this area and the potential applications and benefits of such a model were detailed. Analysis of noise within the sensor allowed the impact of variation in the physical parameters defining the sensor to be determined and compared. The model was used to compare different protocols and confirmed that the sandwich assay produced the more sensitive device. A study of the kinetic response of the assay determined that measurements could be performed at half the time taken to reach equilibrium without significant loss of sensitivity. An analysis of the effect of scattering at the waveguide surface showed this to be significant noise factor. An initial study of the impact of the humectant layer illustrated that this is an issue that merits further consideration.