Forms I-III and dihydrate carbamazepine (CBZ) were prepared and confirmed by X-ray powder diffraction (XRPD) and differential scanning calorimetry (DSC). Influences of supersaturation (σ), stirring, anti-solvent (H2O), and polymer type on the resultant polymorph are discussed. For a CBZ ethanol solution at 5 ºC, more than 10 h was required to form crystals when σ was 0.5, while less than 2s was required when σ was increased to 9.0. Very fine needle-shaped Form II crystals were obtained when σ ≥ 7.5. Higher stirring rates facilitated the formation of Form III CBZ. Continuous heterogeneous crystallization of Form III on Polyvinyl alcohol (PVA, MW 89,000-98,000) was achieved in a one-stage mixed suspension mixed product removal (MSMPR) crystallizer at 15 ºC and 300 rpm. At 5 ºC and 40 rpm, only Form II crystals were obtained. However, Form II CBZ gradually transformed to Form III within 2 residence times, and the transition process was irreversible.
An automated multi-stage continuous reaction system with in-line PATs for a high viscosity reactive crystallization process was developed in the present study. Data acquisition hardware and Labview software were used as the local control system. A “forward-backward” burst pumping strategy was developed to smoothly transfer the highly viscous hot slurry from one vessel to the next. In addition, a comparative analysis between a plug flow reactor (PFR) and continuous stirred-tank reactors (CSTRs) in series revealed that to achieve the same conversion, the latter would require more volume than the former, but less than a single CSTR. For a second-order reaction, the value of the Damköhler number necessary to achieve conversion of 90.0% in a single CSTR is 90. It is reasonable to approximate a PFR using CSTRs in series to obtain a high yield with a smaller reaction volume (compared with a single CSTR). As the probes could not be positioned in the hot slurry due to fouling issues, in-line focused beam reflectance measurement and React IR were used to monitor the crystal size and reactant concentration in the vessel containing the cold slurry. E-factors of batch and continuous processes were also compared and the continuous reaction could obtain a lower E-factor because less waste was generated.
An automated continuous clarification bypass system was developed to remove the suspended particulate matter (SPM) in the prereaction material. Compared to commercially available duty/standby filters, the proposed clarification bypass system is able to self-clean and does not require detachment of the filter and manual cleaning. In a stainless steel (SS) filter, the effects of flow direction, ultrasonication, and viscosity were investigated. The data showed that the filtration performance could not meet the requirement of high clarification efficiency because of the high switch frequency. The deposition of crystals on the SS filter medium, and not the SPM, was the primary cause of the pressure build-up. Experiments with PTFE filter elements with comparable pore size and surface area to the stainless-steel filter were performed, and improved filtration performance was observed. At the beginning of the SPM filtration process with the PTFE filter elements, three filtration mechanisms occurred. As the filtration cake formed on the filter element surface, straining gradually dominated the filtration process, while the effects of impingement and entanglement became negligible.