Development of Combined Continuous Reaction to Crystallization Process Using PAT Tools

Introduction

Figure 1. Synthesis of (Z)-3-chloro-2-(phenylthio)-N-(p-tolyl)acrylamide

Traditionally the majority of active pharmaceutical compounds have been manufactured at large plant scales using batch methodologies. Continuous manufacturing is becoming an increasing attractive alternative due to apparent benefits such as smaller manufacturing footprints, increased process safety, lower manufacturing costs and increased supply-chain flexibility. In this work, a new approach to the design of continuous manufacturing platforms is being presented which is not restricted to using such conventional equipment but which is based around bespoke design and manufacture. As a model compound, 2-chloro-N-(4-methylphenyl)propanamide (S1), a precursor in the multi-step synthesis of (Z)-3-chloro-2-(phenylthio)-N-(p-tolyl)acrylamide (P1, Figure 1) is used. The aim of this work is to design a combined continuous reaction to crystallization process for S1 and overcome the existing challenges between the two process.

Solubility Studies for S1

Figure 2. Solubility data of S1 in toluene + hexane mixtures: *, w₂ = 1; +, w₂ = 0.8; ◊, w₂ = 0.6; □, w₂ = 0.4; ─, Phase Equilibria with NRTL equation calculated. w₂ is the mass fraction of toluene in binary toluene + hexane mixtures.

The solubility behaviour of S1 in ethyl acetate + hexane, toluene + hexane, acetone + hexane and butanone + hexane solutions were measured as the solubility characteristics have considerable influence upon the design and optimisation of any crystallisation process, affecting both the resulting crystal structure and crystal size distribution (CSD)¹. The solubility study is done using the polythermal method and conducted in EasyMax (Mettler Toledo) reactor with in situ Focused Beam Reflectance Measurement (FBRM G400, Mettler Toledo) to characterize the saturation temperature of S1. The experimental data are correlated with Apelbat, λh and the Phase Equilibria with NRTL (non-random two liquid) model equations using the function lsqcurvefit in Matlab’s optimization toolbox to enable both interpolation and extrapolation of the measured data, as shown in Figure 2.

Design of Continuous Reaction and Crystallization Process

Reaction: Using a modified procedure developed by Li et al², a batch process is employed where a-chloropropionyl chloride (12 mmol) was added dropwise to a suspension of p-toluidine (10 mmol) in toluene (20 mL) at 0^oC and the resulting solution was heated to reflux for 1 h with vigorous stirring.  After cooling, the solvent was removed under vacuum and the resulting off white solid was collected by filtration and washed thoroughly with cold cyclohexane. Spectral data of the obtained product matched that reported in the literature³. Therefore, it has been decided to use toluene as the reaction solvent as it can also be used as a crystallization solvent based on the solubility studies. However, this developed procedure produces a small amount of HCl as a by-product, which can be problematic for reaction product purity and its subsequent crystalisation steps. Therefore, it is decided to run the reaction in a novel 3D printed continuous biphasic reaction system in the presence of aqueous sodium hydroxide to neutralize HCl salt. The design of the biphasic reactor is based on the specific needs of the reaction which is capable of achieving extremely high dispersion rates between the two phases under the very low Reynolds numbers typically encountered in the pharmaceutical industry (Re < 5). Process steps are quantified in detail using methods such as computational fluid dynamics (CFD) leading to tailored equipment designs which can be manufactured with high precision using metallic 3D printing. The biphasic reactor is coupled with liquid-liquid separation such that both unit operations are designed using fundamental engineering principles combined with numerical analyses (based upon the requirements of the specific process), prior to being 3D printed and combined to create a bespoke, sealed, hard-piped manufacturing system.

Reaction to Crystallization: From the liquid-liquid separator, the organic phase is transferred in the feed/dissolution tank while aqueous phase goes to waste. In situ FTIR (React IR 15, Mettler Toledo) is used on the feed/dissolution tank to determine the concentration of S1 prior to crystallization and the crystallization kinetic data obtained (explained below) enables the adjustment of crystallization parameters for better control and set up of the process.

Crystallization: Using the solubility curve obtained from solubility studies for toluene, a continuously operated single stage mixed suspension, mixed product removal (MSMPR) crystallizer with intermittent withdrawal via dip pipe using an automated pressure supply^4-6 is set up for the cooling crystallization of S1. Acquisition of reliable kinetic correlations that describe how rates of nucleation and growth vary over the design space for the crystallization process is necessary for enabling robust prediction of crystallizer performance. Therefore, simple extraction of the growth and nucleation parameters from the single stage MSMPR operating at steady state is performed using the population balance model. Comprehensive experimental data for model fitting and parameter estimation is obtained by varying crystallizer operating conditions (temperature, supersaturation and suspension density) to induce changes in the rate-affecting variables for growth and nucleation. FBRM is used for real-time monitoring of size distribution and identification of steady state and hence enabling on-demand adjustment of operating conditions to transition multiple steady-states for rapid acquisition of kinetic data while PVM (ParticleView V819, Mettler Toledo) provides microscopy quality images in real time to monitor the crystal habit. Malvern Mastersizer is used as an offline analysis to determine the crystal size distribution (CSD) at steady-state while the concentration of mother liquor at steady-state is determined gravimetrically. By estimating the six kinetic parameters for crystal growth and nucleation, accurate prediction of CSD and concentration results at steady-state can be achieved.

References:

[1] Mullin, J. W. Crystallization; Butterworth-Heinemann, 1997

[2] Li, X.; Wang, D.; Wu, J.; Xu, W. Novel Approach to 3‐Methyl‐1H‐quinoxalin‐2‐ones. Synthetic communications 2005, 35 (19), 2553.

[3] Murphy, M.; Lynch, D.; Schaeffer, M.; Kissane, M.; Chopra, J.; O'Brien, E.; Ford, A.; Ferguson, G.; Maguire, A. R. Investigation of the synthetic and mechanistic aspects of the highly stereoselective transformation of [small alpha]-thioamides to [small alpha]-thio-[small beta]-chloroacrylamides. Organic & Biomolecular Chemistry 2007, 5 (8), 1228.

[4] Hou, G.; Power, G.; Barrett, M.; Glennon, B.; Morris, G.; Zhao, Y. Development and Characterization of a Single Stage Mixed-Suspension, Mixed-Product-Removal Crystallization Process with a Novel Transfer Unit. Crystal Growth & Design 2014, 14 (4), 1782.

[5] Power, G.; Hou, G.; Kamaraju, V. K.; Morris, G.; Zhao, Y.; Glennon, B. Design and optimization of a multistage continuous cooling mixed suspension, mixed product removal crystallizer. Chemical Engineering Science 2015, 133, 125.

[6] Morris, G.; Power, G.; Ferguson, S.; Barrett, M.; Hou, G.; Glennon, B. Estimation of Nucleation and Growth Kinetics of Benzoic Acid by Population Balance Modeling of a Continuous Cooling Mixed Suspension, Mixed Product Removal Crystallizer. Organic Process Research & Development 2015, 19 (12), 1891.