Online measurement and population balance modelling of stirred nano-milling

Productions of fine particles in a wet mill is important to many industries including the pharmaceutical, chemical and synthetic industry. The particle size distribution of fine particles is one of the most important properties that determine the usefulness and the quality of the final product. It is therefore important that the final suspension from the wet milling process has a narrow size distribution to meet the product specification. This requires a good real-time particle size distribution measurement system and a good process model of the manufacturing process. Particle size measurement system that can cope with all particle sizes and concentration is not yet available and population balance modelling is not well developed for the wet milling process. These issues are addressed in this thesis. Ultrasonic particle sizing techniques have the potential for real time monitoring of the particle size distribution evolution in a stirred media mill. Hardware and software for ultrasonic measurement of the real time particle size distribution were developed during this project. The particle size distribution is determined by comparing the measured attenuation spectra with the prediction from an ultrasonic wave model and the size distribution adaptively fitted by minimizing the difference between the fitted and predicted spectra. Because no single model is valid for all particle size distributions and volume concentrations, the Hybrid© model is introduced in this work to automatically pick the best model depending on the suspension’s properties. Using the Hybrid© model, the instrument was validated for several suspensions and it gave excellent particle size distributions agreement with the reference particle size distributions. All instrument functions are controlled through a graphical user interface (GUI) software developed during this project. This will make it easy to transfer the findings of this project to other workers to use in their research. The ultrasonic instrument developed was used to monitor the particle size distributions in a stirred media mills using different milling conditions. Samples were taken for offline analysis using the Malvern Mastersizer and Zetasizer instruments to validate the measured size distributions. The mill was operated in circuit mode with the suspensions continuously circulated between a 2200 ml mixing tank and the 550 ml milling chamber of the mill. The experimental results showed that generally the particle sizes decreased with time during milling. The particle sizes in the milling tank are smaller compared to the particle sizes in the mixing tank. This observation agrees with the fact that particle grinding is only done in the milling chamber; therefore it can be concluded that particle breakage took place only in the milling chamber while mixing and aggregation were the main mechanisms responsible for particle size change in the mixing tank. The results show that the smaller the grinding media size higher the particle breakage rate due to increased shear generated by smaller grinding media as well as an increase in the grinding media number density as the media size reduces for the same grinding loading. The results also show that the particle breakage rate increases with increasing mill speed but final particle size distribution is independent of the mill speed. This shows that the breakage behaviour of the particles is not a function of the mill speed. The grinding media loading however have minimal effect on both the breakage rate and the final particle size distribution. One of the main factors limiting the application of population balance modelling to wet milling is the lack of a phenomenological breakage model for wet milling. In this work, a phenomenological breakage model linking the breakage rate to the process parameters is derived. A population balance model for the circuit mode was developed and validated in this work. The parameters of the population balance model were determined by adaptively minimizing the difference between the experimental size distribution and the size distribution predicted by the population balance model. The fitted parameters show that particle breakage rate increases with increasing mill speed. The number of daughter particles produced on breakage of a single particle decreases with increasing mill speed at low mill speed before increasing with increasing mill speed at higher mill speed. The degree of attrition of particles reduces with increasing the mill speed. The particle breakage rate decreases with increasing grinding media size. However, the breakage kernel power law exponent is higher for the higher grinding media size. The higher the grinding media size, the higher the number of particles produced per breakage event and the breakage behaviour changes from attrition to something more complex with the production of high proportion of small particles and little proportion of larger particles. The breakage rate increases with the increase in the grinding media loading but the daughter distribution function is independent of the grinding media loading.