Final Activity Report Summary - INTROSPECT (INTegrated approaches to chemical pROcess and product quality monitoring using SPECTroscopy)
The research carried out through this Early stage training (EST) programme focussed on overcoming the limitations encountered in the extraction of information from spectroscopic measurements for monitoring particulate, e.g. powders, suspensions etc, processes. The aim was to develop novel information extraction methodologies through a multi-disciplinary engineering science approach that integrated measurement strategies, multivariate statistical modelling techniques, i.e. chemometrics, and fundamental models of light propagation in particulate media for maximal information extraction to facilitate effective process monitoring and control.
The research projects were chosen so that the early stage researchers trained through this programme would acquire unique multi-disciplinary skills. The research training was provided through the following four research projects:
1. estimation of chemical information in latex suspensions using radiative transfer theory to remove multiple scattering effects. Accurate estimation of concentrations of chemical components in turbid samples using spectroscopic techniques is an open-end problem that poses considerable challenges to applied scientists. This problem is of tremendous practical importance and is encountered in diverse areas, such as in monitoring polymerisation and fermentation processes and in medical diagnostics. Through this project, which was carried out in conjunction with projects two and three, a measurement-based methodology was developed whereby multiple measurements, in conjunction with the radiative transfer equation, were used to separate the absorption and scattering effects. The first step in this approach was to invert the measurements using the radiative transfer equation to obtain the bulk absorption and scattering properties. The second step was to extract the individual species' information from the bulk absorption coefficient using partial least squares regression models. The effectiveness of this approach was theoretically tested using simulations, as well as through experimental datasets of two-component and four-component model latex suspensions. The studies showed that models built using this methodology led to significant improvements in performance compared to traditional chemometrics' methods which used a single measurement combined with empirical pre-processing methods.
2. optical properties of growing biological suspensions. In conjunction with the first project, the measurement-based method for separating absorption and scattering effects using a single integrating sphere setup was used to monitor the optical property changes during a bacterial growth cycle for a simple bacillus subtilis system. It was shown that the scattering and absorption properties could be consistently extracted from such measurements and that these measurements were robust to variations in sample thicknesses and cell-holder refractive index which could be explicitly taken into account through light propagation theories.
3. a spatially resolved near-infrared measurement system for non-invasive optical characterisation of particulate systems. An alternative measurement configuration for the abovementioned methodology that was more suitable for online monitoring applications was the Spatially resolved measurement (SRM) system. A SRM system for the wavelength range of 1 000 to 1 700 nm was developed, along with associated methodologies for extracting bulk optical properties. This system was tested using epoxy-based and gelatine-based tissue phantoms.
4. inversion of spectroscopic measurements to estimate particle size distributions. This research focussed on developing regularisation techniques to extract particle size distribution from spectroscopic measurements of dense particulate systems. The methodology was based on using the bulk scattering coefficient obtained through inverting the radiative transfer equation and solving the resulting Fredholm integral equation via regularisation techniques.
The research projects were chosen so that the early stage researchers trained through this programme would acquire unique multi-disciplinary skills. The research training was provided through the following four research projects:
1. estimation of chemical information in latex suspensions using radiative transfer theory to remove multiple scattering effects. Accurate estimation of concentrations of chemical components in turbid samples using spectroscopic techniques is an open-end problem that poses considerable challenges to applied scientists. This problem is of tremendous practical importance and is encountered in diverse areas, such as in monitoring polymerisation and fermentation processes and in medical diagnostics. Through this project, which was carried out in conjunction with projects two and three, a measurement-based methodology was developed whereby multiple measurements, in conjunction with the radiative transfer equation, were used to separate the absorption and scattering effects. The first step in this approach was to invert the measurements using the radiative transfer equation to obtain the bulk absorption and scattering properties. The second step was to extract the individual species' information from the bulk absorption coefficient using partial least squares regression models. The effectiveness of this approach was theoretically tested using simulations, as well as through experimental datasets of two-component and four-component model latex suspensions. The studies showed that models built using this methodology led to significant improvements in performance compared to traditional chemometrics' methods which used a single measurement combined with empirical pre-processing methods.
2. optical properties of growing biological suspensions. In conjunction with the first project, the measurement-based method for separating absorption and scattering effects using a single integrating sphere setup was used to monitor the optical property changes during a bacterial growth cycle for a simple bacillus subtilis system. It was shown that the scattering and absorption properties could be consistently extracted from such measurements and that these measurements were robust to variations in sample thicknesses and cell-holder refractive index which could be explicitly taken into account through light propagation theories.
3. a spatially resolved near-infrared measurement system for non-invasive optical characterisation of particulate systems. An alternative measurement configuration for the abovementioned methodology that was more suitable for online monitoring applications was the Spatially resolved measurement (SRM) system. A SRM system for the wavelength range of 1 000 to 1 700 nm was developed, along with associated methodologies for extracting bulk optical properties. This system was tested using epoxy-based and gelatine-based tissue phantoms.
4. inversion of spectroscopic measurements to estimate particle size distributions. This research focussed on developing regularisation techniques to extract particle size distribution from spectroscopic measurements of dense particulate systems. The methodology was based on using the bulk scattering coefficient obtained through inverting the radiative transfer equation and solving the resulting Fredholm integral equation via regularisation techniques.