Acid number, viscosity and end-point detection in a multiphase high temperature polymerisation process using an online miniaturised MEMS Fabry-Pérot interferometer.

Recent advances in the latest generation of MEMS (micro-electro-mechanical system) Fabry-Pérot interferometers (FPI) for near infrared (NIR) wavelengths has led to the development of ultra-fast and low cost NIR sensors with potential to be used by the process industry. One of these miniaturised sensors operating from 1350 to 1650 nm, was integrated into a software platform to monitor a multiphase solid-gas-liquid process, for the production of saturated polyester resins. Twelve batches were run in a 2 L reactor mimicking industrial conditions (24 h process, with temperatures ranging from 220 to 240 °C), using an immersion NIR transmission probe. Because of the multiphase nature of the reaction, strong interference produced by process disturbances such as temperature variations and the presence of solid particles and bubbles in the online spectra required robust pre-processing algorithms and a good long-term stability of the probe. These allowed partial least squares (PLS) regression models to be built for the key analytical parameters acid number and viscosity. In parallel, spectra were also used to build an end-point detection model based on principal component analysis (PCA) for multivariate statistical process control (MSPC). The novel MEMS-FPI sensor combined with robust chemometric analysis proved to be a suitable and affordable alternative for online process monitoring, contributing to sustainability in the process industry.

Process Monitoring of Moisture Content and Mass Transfer Rate in a Fluidised Bed with a Low Cost Inline MEMS NIR Sensor.

PURPOSE: The current trend for continuous drug product manufacturing requires new, affordable process analytical techniques (PAT) to ensure control of processing. This work evaluates whether property models based on spectral data from recent Fabry-Pérot Interferometer based NIR sensors can generate a high-resolution moisture signal suitable for process control. METHODS: Spectral data and offline moisture content were recorded for 14 fluid bed dryer batches of pharmaceutical granules. A PLS moisture model was constructed resulting in a high resolution moisture signal, used to demonstrate (i) endpoint determination and (ii) evaluation of mass transfer performance. RESULTS: The sensors appear robust with respect to vibration and ambient temperature changes, and the accuracy of water content predictions (±13 % ) is similar to those reported for high specification NIR sensors. Fusion of temperature and moisture content signal allowed monitoring of water transport rates in the fluidised bed and highlighted the importance water transport within the solid phase at low moisture levels. The NIR data was also successfully used with PCA-based MSPC models for endpoint detection. CONCLUSIONS: The spectral quality of the small form factor NIR sensor and its robustness is clearly sufficient for the construction and application of PLS models as well as PCA-based MSPC moisture models. The resulting high resolution moisture content signal was successfully used for endpoint detection and monitoring the mass transfer rate.

An ultra-compact particle size analyser using a CMOS image sensor and machine learning.

Light scattering is a fundamental property that can be exploited to create essential devices such as particle analysers. The most common particle size analyser relies on measuring the angle-dependent diffracted light from a sample illuminated by a laser beam. Compared to other non-light-based counterparts, such a laser diffraction scheme offers precision, but it does so at the expense of size, complexity and cost. In this paper, we introduce the concept of a new particle size analyser in a collimated beam configuration using a consumer electronic camera and machine learning. The key novelty is a small form factor angular spatial filter that allows for the collection of light scattered by the particles up to predefined discrete angles. The filter is combined with a light-emitting diode and a complementary metal-oxide-semiconductor image sensor array to acquire angularly resolved scattering images. From these images, a machine learning model predicts the volume median diameter of the particles. To validate the proposed device, glass beads with diameters ranging from 13 to 125 µm were measured in suspension at several concentrations. We were able to correct for multiple scattering effects and predict the particle size with mean absolute percentage errors of 5.09% and 2.5% for the cases without and with concentration as an input parameter, respectively. When only spherical particles were analysed, the former error was significantly reduced (0.72%). Given that it is compact (on the order of ten cm) and built with low-cost consumer electronics, the newly designed particle size analyser has significant potential for use outside a standard laboratory, for example, in online and in-line industrial process monitoring.

Data fusion strategies to combine sensor and multivariate model outputs for multivariate statistical process control.

Process analytical technologies (PAT) applied to process monitoring and control generally provide multiple outputs that can come from different sensors or from different model outputs generated from a single multivariate sensor. This paper provides a contribution to current data fusion strategies for the combination of sensor and/or model outputs in the development of multivariate statistical process control (MSPC) models. Data fusion is explored through three real process examples combining output from multivariate models coming from the same sensor uniquely (in the near-infrared (NIR)-based end point detection of a two-stage polyester production process) or the combination of these outputs with other process variable sensors (using NIR-based model outputs and temperature values in the end point detection of a fluidized bed drying process and in the on-line control of a distillation process). The three examples studied show clearly the flexibility in the choice of model outputs (e.g. key properties prediction by multivariate calibration, process profiles issued from a multivariate resolution method) and the benefit of using MSPC models based on fused information including model outputs towards those based on raw single sensor outputs for both process control and diagnostic and interpretation of abnormal process situations. The data fusion strategy proposed is of general applicability for any analytical or bioanalytical process that produces several sensor and/or model outputs. Graphical abstract.

Rapid multistep kinetic model generation from transient flow data.

Today, the generation of kinetic models is still seen as a resource intensive and specialised activity. We report an efficient method of generating reaction profiles from transient flows using a state-of-the-art continuous-flow platform. Experimental data for multistep aromatic nucleophilic substitution reactions are collected from an automated linear gradient flow ramp with online HPLC at the reactor outlet. Using this approach, we generated 16 profiles, at 3 different inlet concentrations and 4 temperatures, in less than 3 hours run time. The kinetic parameters, 4 rate constants and 4 activation energies were fitted with less than 4% uncertainty. We derived an expression for the error in the observed rate constants due to dispersion and showed that such error is 5% or lower. The large range of operational conditions prevented the need to isolate individual reaction steps. Our approach enables early identification of the sensitivity of product quality to parameter changes and early use of unit operation models to identify optimal process-equipment combinations in silico, greatly reducing scale up risks.