Abstract
Crystallisation by precipitation is a very common technique in industry, however the crystals produced are often of poor quality, characterised by broad particle size distribution and morphological inhomogeneities. These problems can be overcome by the use of ultrasound. Sonocrystallisation, or ultrasound-assisted crystallisation, has already demonstrated its general benefits in terms of reduction in crystallisation time, smaller and more homogeneous crystals and repeatability. However, there are many issues still unaddressed, such as poor knowledge of the mechanism of sonocrystallisation and of the governing parameters of the whole system. This is largely attributed to lack of systematic studies, varying frequency and power while using the same equipment, as well as difficulties in comparing results from different systems. Furthermore, ultrasound is still considered uneconomical in industrial scale, due to its high energy demand. Therefore, there is the necessity of optimising the use of ultrasound for guaranteeing a high-quality product with the lowest energy consumption. This thesis therefore focused on studying the effects of sonocrystallisation for two different crystallisation systems by precipitation: the antisolvent crystallisation of sodium chloride and the reactive crystallisation of ZIF-8, a type of Metal-Organic Frameworks (MOFs). In both systems, frequency, power and sonication time was systematically varied, and the effects on the product obtained investigated. The effect of changing supersaturation was also studied. The comparison between a simpler and a more complex system could help on obtaining general conclusions useful for clarifying the mechanism of sonocrystallisation. The antisolvent sonocrystallisation of sodium chloride revealed that sonication caused a reduction of the crystal size up to 10 times with respect to unsonicated conditions. This was achieved within a few seconds of sonication (5 s for high supersaturation and 15 s for low supersaturation), indicating that the effects of ultrasound were on the nucleation, causing the rapid formation of small and regular crystals. Phenomena of sono-fracture was excluded because the crystals formed were morphologically regular, very different from fragments. Adding a subsequent step in silent conditions or using intermittent ultrasound bursts coupled with silent conditions showed a limited crystal growth. Hence, ultrasound uses the majority of the solute forming new small crystals and leaving in the solution only a small amount of sodium chloride. The effects of different frequencies and powers were more evident at lower supersaturations, with lower frequency being more effective at reducing crystal size. However, under these conditions, although a small average crystal size was obtained, the particle size distributions were bimodal at low power. At high supersaturation, this bimodality disappeared even at low powers. For the reactive crystallisation of ZIF-8 at high excess of ligand, it was confirmed that the action of ultrasound caused a reduction of the crystal size up to nanometre. Furthermore, it was demonstrated again that the action of sonication was on the early stages of the crystallisation. However, the effects of ultrasound resembled a lot the results obtained when only mixing was applied, suggesting that sonication was not directly influencing the chemistry of the reaction. In addition, the BET surface area appeared to be related to the power used, reaching a maximum and decreasing beyond a certain power. This suggests that a competition between micromixing, causing an increase of the BET surface area, and shockwaves, responsible of breaking the framework during its formation. On the other hand, when a low excess of ligand was used, it was revealed that the formation of by-products was accelerated when mixing or sonication were applied. However, if shorter reaction time was applied, it was possible to obtain a reasonable quality of ZIF-8 at low excess of ligand. This new insight explains why low excess of ligand was considered unsuitable for ZIF-8 production, as all reported studies in the literature have applied mixing and used long reaction times. Overall, the results from this thesis have demonstrated commonalities between antisolvent and reactive crystallisation, that is reducing the sonication time to a brief interval at the beginning of the crystallisation can achieve the same size reduction as sonicating for the entire process. This is a key result in view of optimising sonocrystallisation processes. In addition, lower supersaturations or excess ligands are more sensitive to the crystallisation condition. The effect of frequency and power is more significant under antisolvent crystallisation, whereas under reactive crystallisation mixing is the dominant effect and therefore no frequency or power effect is observed. These differences may explain why there are inconsistencies in the reported frequency and power effects in the literature.