Template for Electronic Submission to acs journals
Yüklə 71,67 Kb.
|
| Design and Evaluation of a Meso-scale Segmented Flow Reactor (KRAIC) Karen Robertson*, Pierre-Baptiste Flandrin, Anneke R. Klapwijk and Chick C. Wilson Department of Chemistry, University of Bath, Claverton Down, Bath, BA2 7AY
*Corresponding author: Department of Chemistry, University of Bath, Claverton Down, Bath, BA2 7AY k.robertson@bath.ac.uk
Design and Evaluation of a Meso-scale Segmented Flow Reactor (KRAIC) Karen Robertson*, Pierre-Baptiste Flandrin, Anneke R. Klapwijk and Chick C. Wilson Department of Chemistry, University of Bath, Claverton Down, Bath, BA2 7AY Corresponding author: k.robertson@bath.ac.uk
ABSTRACT: The design and development of a mesoscale flow reactor is presented, together with initial performance evaluation. The custom-designed KRAIC (Kinetically-Regulated Automated Input Crystallizer) uses liquid-segmented flow to achieve plug flow and mixing throughout the reactor length whilst helping prevent encrustation issues experienced by other precipitation reactors / crystallizers. Evaluation of the reactor for production of solid-state particle products has been carried out using cooling crystallization of fast growing succinic acid to explore the potential of the KRAIC design to mediate blocking caused during the precipitation process. The segmented flow environment is successful in preventing encrustation during an unseeded 2 h cooling crystallization experiment. Introduction Since the advent of flow crystallization in the 1980s there has been, until recently, little progress in translating this area into regular use in solid-state research or into production/manufacturing environments. This is due to the inherent problems associated with moving particles in a flowing environment; few pumps are compatible with even trace amounts of solids, any adhesion of particles to the reactor walls will ultimately lead to a loss of homogeneity of the product and failure of the reactor, on-line filtration methods are yet to be optimized to maintain the crystal size distribution (CSD),1 particle attributes and minimize effects of further solution washing over the precipitate for extended periods of time. In an ideal flow crystallizer, nucleation and growth occur at the same location and rate for a given set of parameters. Particles are well suspended with good mixing throughout the reactor whilst plug flow is achieved. Traditional flow crystallizers are either designed for large volumes (l/h) – such as continuous oscillatory baffled crystallizers (COBCs),2 Couette-Taylor crystallizers,3 mixed suspension mixed product removal (MSMPR) crystallizers4 – or for very small volumes (μl/h) in microreactors5 with a narrow internal diameter which makes them incompatible with any solids above the nanometre scale. There has been some significant progress recently in the development of meso-scale crystallizers which enable small scale production of μm-cm sized crystals without the need for kg of starting material. Meso-scale crystallizers are typically tubular employing either static mixers such as kenics-type mixers or segmented flow. Vacassy et al1 used liquid-segmented flow to generate a narrow particle size distribution (PSD) of CaCO3 (vaterite) through precipitation reaction. Jongen et al6 went on to use liquid-segmented flow to effectively generate narrow PSDs of metal oxalates. Issues with separating the immiscible carrier fluid in liquid-segmented flow has since restricted research in this area and focus has shifted to gas-segmented flow.7 Gas-segmented flow can however result in fouling issues which has led the authors to renew investigations into liquid-segmentation. Here we describe the design and evaluation of a new liquid-segmented meso-scale tubular reactor which has proven able to run for 5 h without fouling issues and with instant recovery of solids. The use of liquid-liquid segmented flow is employed for a multitude of purposes. By using a carrier fluid with favorable wetting properties the solution flow is physically removed from the walls of the reactor preventing any reaction with the walls of the reactor and consequent sedimentation or encrustation in precipitating systems. The non-slip boundary between the immiscible fluids generates bolus flow which provides mixing throughout the reactor length. The creation of discrete segments (slugs) prevents any back mixing in the system ensuring that all of the product has experienced the same conditions and reaction / crystallization time, imparting homogeneity. Separation of the immiscible carrier fluid prior to filtration has been achieved, overcoming the main challenge impeding further development of liquid-segmented crystallizers to date. Evaluation of the performance of the crystallizer with respect to its capabilities in crystallization of target molecular systems has been undertaken throughout its development using succinic acid (SA). SA is an ideal model system in this context due to its fast growing nature.8 This provides a challenging system which will quickly highlight any potential areas for blockages, paramount for the development of a reactor designed primarily for use in precipitation reactions / crystallizations. The development of the KRAIC platform is being carried out in the context of the Centre for Continuous Manufacturing and Crystallisation (CMAC).9 This UK academic-industry consortium is targeting the underpinning research and development required to ease the translation of continuous crystallization into the industrial manufacturing of fine chemicals and pharmaceuticals. Reactor Design The kinetically regulated automated input crystallizer (KRAIC) is a modular flow crystallization and reaction chemistry apparatus with multiple functionality (Figure 1). In essence it is an open tubular reactor which operates on the principle of segmented flow to impart the desirable characteristics of flow crystallization / chemistry detailed above. Feed solutions and carrier fluid are pumped independently to a mixer piece whereupon the segmented flow is then propelled through three separate, independently temperature-controlled zones before emerging at the filtration site. Kinetic regulation is imparted by pump speed and reactor length. By varying the input speed of the solution(s) or carrier fluid flow the intensity of mixing and residence time can be optimized. Therefore solution-mediated kinetic changes such as polymorphic changes can be either maximized or eliminated as desired. The body of the reactor is constructed from one extruded length of fluorinated ethylene propylene (FEP) (15 m, 3.2 mm ID) through which the liquid-segmented flow proceeds. Gear pumps propel the feed solution(s) and carrier fluid through the tubing; each pump is independent enabling full control over mixing and segmenting conditions. A variety of mixing and segmenting configurations can be achieved by use of cross-pieces, combining either standard Y- and T-pieces or custom impinging jets but for this work a simple glass T-piece was used. This level of parameter control results in a highly flexible platform which can deliver a range of crystallization / reaction needs. Temperature control is achieved in the KRAIC through a combination of static and kinetic heat transfer media. The feed vessels (T1, Figure 1), mixer / segmentor (T4a) and tubing coils (T5-7) are heated using hotplate stirrers and either Drysyn®, water or graphite baths. The tubing between feed vessel and pump (T2), pump and mixer / segmentor (T3), mixer / segmentor and coil 1 (T4b) are heated / cooled 
Figure . Image of the KRAIC with labels highlighting each section, number labels refer to temperature zones listed in Table 1 
Figure . Schematic of KRAIC showing a single feed set-up using circulating baths via flexible tubing jackets designed in collaboration with Asynt (Figure 3).10 The tubing requiring temperature control is threaded through the tubing jacket, comprising two glass ends with o-ring fittings to secure tubing within, with an outer silicone tubing enabling flexibility in both position and length. A sealed bag of ice / water can be placed conveniently within coil 3 for forced precipitation.
Figure a. Flexible tubing jacket ends, b. in-situ with solution feed tubes inserted and circulator hose attached. In a single feed crystallization set-up the temperature zones (1-7, Table 1) are determined by the solubility profile of the product. Nucleation can be achieved through a decreasing temperature profile or by hot / cold cycling – in which the tube immediately prior to coil 1 is cooled below the metastable zone of the compound (i.e. below the solubility limit of the solution where precipitation is expected) resulting in fine precipitate, most, but not all, of which is re-dissolved in the higher temperature zone of coil 1. Once nucleation has been achieved the remainder of the crystallizer length is employed in crystal growth. Table . List of temperature zones within the KRAIC
An end-piece, placed on the outlet of the tubing, is designed both to facilitate the propulsion of the crystals into the filtration unit and to separate the carrier fluid from the solution (Figure 4). The widening of the outlet to the end-piece (the ‘mouth’) results in both a downward slope, which aids in maintaining the momentum of particles for transport into the filtration unit, and a sudden increase in volume. The slugs distort in response to this volume increase, causing turbulence within each slug thereby preventing sedimentation. Immediately prior to the opening in the end-piece, a series of small holes at the bottom of the tubing allow passage of the preferentially wetting carrier fluid but not the solution flow. This results in the separation of carrier fluid prior to filtration allowing instant recovery of carrier fluid. Fouling of the tubing upon loss of protecting carrier-fluid is mitigated by the introduction of a neutral solvent upstream from the carrier-fluid recovery system. By adding a miscible solvent which does not alter the solubility properties of the solution (a neutral solvent) the velocity of the solution is increased at the point of addition, reducing the likelihood of fouling by reducing contact time of the precipitate with the reactor walls. Additionally the reduced solid loading experienced at this point of the crystallizer further mitigates the possibility of fouling. The end-piece is designed such that it fits onto the end of the tubing of the crystallizer enabling facile exchange during experiments if fouling should occur. 
Figure . Image of the end-piece of the KRAIC. Inset shows the holes for carrier fluid recovery and mouth The filtration unit comprises a carousel which holds a series of Buchner funnels and flasks, for smooth switching between sampling, which can then be connected to a diaphragm pump for on-line vacuum filtration. The residence volume of the KRAIC (excluding carrier fluid) at standard slugs sizes is 61 ml, i.e. the volume of solution within the KRAIC at any given time is 61 ml. This means that for an overall flow rate of 8.3 ml/min (as used for SA crystallization herein) 1 l of solution will enable a 2 h operation time. This is in stark contrast to traditional flow crystallizers such as COBCs which will typically use 1 l feed solution in 25 min. Experimental The crystallization of a single component molecular crystalline material in the KRAIC was investigated with succinic acid (SA). 130 g of SA were dissolved in 1 L of H2O at 40 °C. All tubing jackets and the mixer / segmentor bath were heated to 40 °C to prevent fouling prior to segmentation. Due to the efficacy of the tubing jackets the carrier fluid feed vessel did not require pre-heating. The three tubing coils were not actively heated thus a seamless cooling gradient was achieved throughout the crystallizer length, avoiding spikes of supersaturation. The ambient temperature was 18 °C and, at an overall flow rate of 8.3 ml/min, the temperature along the reactor length varied from 40 °C to 19.6 °C as shown in Figure 5. The calculated temperature gradient was corroborated by temperature probes at selected lengths, any discrepancy is due to unaccounted for insulation in coils 1 and 2. Prior to crystallization the tubing was primed with distilled water (40 °C feed) for 2 h. Figure . Calculated temperature gradient over reactor length in the KRAIC. Orange markers show experimentally observed temperatures.
In repeated experiments SA was observed to form crystals after 7 min in the top of the second coil (ca. 25 °C). Figure 6 illustrates how the temperature gradient correlates to the metastable zone (between cloud and clear points). At 25 °C the crystallization profile is significantly within the labile (precipitation) zone. Concentration points thereafter are estimated from final concentration analysis. The crystals could be seen to increase in size throughout the reactor length and mixing within the slugs was illustrated by tumbling of entrapped crystals. An average yield of 0.24 g (2 %) was collected for each residence time (RT); 56 ml of effluent in 9 min 15 sec. The nature of the solid form SA recovered was confirmed through PXRD and DSC (ESI) and crystal size distribution was evaluated visually by optical microscopy (Figure 7). Individual yields were measured every RT to ascertain the homogeneity of the crystallization rate within an experiment. The masses observed ranged from 0.16 – 0.32 g (ESI), this variation in obtained yield between residence times can be attributed to a gradually decreasing segment size during runs. The segments decreased in size over time resulting in an increased mixing efficiency which may have led to an increase in the number of nucleation events. Unexpectedly, PXRD analysis shows a mixture of both the α- and β-polymorphs of SA in the solid product; typically, only the β- form is expressed in solution-based crystallization. This is discussed further below. 
Figure Image of SA crystals produced in the KRAIC Design Evolution – Optimizing Operation Evaluation of the reactor design, under continuous crystallization conditions, was carried out using a solution of SA thus ascertaining the capabilities of the KRAIC in a system in which the particles grow rapidly. The evolution of the operational design will be accounted with respect to findings in these SA crystallizations. Initial designs employed air-segmented flow to reduce complications of compatibility and separation of carrier fluid. Unfortunately, in an air-segmented system, the solution is the continuous phase and thus coats the walls of the tubing. This is the case no matter how high the contact angle is between the liquid and solid as is shown for an air-segmented aqueous solution in FEP (Figure 8). The combination of constant contact of the solution with the tubing walls and areas of reduced pressure, experienced adjacent to slugs of air, encourages precipitation on the walls of the tubing. This precipitation largely remains on the walls of the crystallizer causing multiple issues; disruption of segments, encouraging further nucleation and growth, influencing suspended particles (e.g. inducing polymorphism and a broadening crystal size distribution) and ultimately leads to blockage. Liquid-liquid segmentation was thus employed to combat this through choice of a carrier fluid (perfluoropolyether Galden SV110) which is highly inert with a very favorable contact angle with FEP making it the continuous phase. Thus the solution is physically removed from the crystallizer walls, preventing nucleation on the walls and inhibiting sedimentation (see Figure 8b).
Figure a. Schematic showing air-segmented flow, b. schematic showing liquid-segmented flow, c. air-segmented tubing coil showing disruption of segments due to precipitation. Even in the unaffected uppermost segments the angle of air-liquid interface shows wetting of the tubing by the aqueous solution, d. liquid segmented tubing coil showing aqueous slugs (with slurry), liquid-liquid interface angle reveals the transparent carrier fluid to be the continuous phase. The first design of the KRAIC lacked an end-piece and crystals were deposited directly from the end of the tubing. However it was found that crystals from each slug would adhere to the tubing walls at the exit of the tubing whilst the solution was deposited into the filtration unit. As more crystals were then propelled into the developing plug at the exit of the tubing a blockage would eventually form. The end-piece then went through many iterations as the initial design of the end-piece concerned only the split ‘mouth’ to induce turbulence. However the level of carrier fluid recovery was low and there were severe issues with filter cake formation. Coating of the particles with carrier fluid during filtration promoted adhesion of particles significantly reducing filter efficiency. This resulted in continued growth of the crystals in the filter funnel (at an exponential rate with respect to mass of solids in the filter funnel). A second design included the carrier-fluid recovery holes which efficiently separate the immiscible fluids through the disparity in wetting properties of FEP for these fluids. This not only enabled immediate high yielding recovery of the carrier fluid but also mitigated caking issues. Unfortunately this separation results in the removal of the physical barrier between the slurry and the crystallizer walls, enabling adhesion of the particles to the walls. In experiments with high concentration of succinic acid this resulted in a blockage in the end-piece within minutes. The third evolved design of the end-piece also includes an inlet for a neutral solvent (one which does not dissolve the precipitate or encourage further precipitation) immediately prior to the recovery holes. The increase in slurry velocity and reduction in solid loading resulting from this introduction significantly aids in mitigation of fouling of the crystallizer; in contrast to design two this configuration was successfully run for over an hour without encrustation issues. Three minutes after stopping the neutral solvent flow a blockage formed above the carrier fluid recovery holes, confirming the importance of this element. Prior to implementation of the novel tubing jacket designs, control over premature nucleation was attempted through the use of high feed vessel temperatures. Unfortunately the temperature drop experienced through unlagged tubing and exposed mixer / segmentor was underestimated and blockages occurred readily. This was mitigated by implementation of tubing jackets and a mixer / segmentor bath, ensuring that the solute remains in solution until segmentation has been achieved. Discussion As nucleation is stochastic in nature12 it is of particular interest that, whilst nucleation has not been intentionally induced, the point of observed nuclei formation was consistent in repeated runs. This may be due to the sterile environment created by segmented flow. Both the carrier and solution feeds were filtered before immersion into the pump to ensure dust / nuclei did not enter the KRAIC from the feed vessels. The diaphragm pump used for vacuum filtration of effluent was placed on the bench and the housing of the KRAIC sits on rubber feet, minimizing vibrations. The temperature of any given point of the KRAIC remains constant throughout each experiment. The consistency of nucleation point achieved in the KRAIC is a significant benefit in controlled continuous crystallization studies. The major source of uncontrolled nucleation in any flow reactor is interaction of the solution with imperfections in the walls of the reactor. In segmented flow the carrier fluid coats the walls of the reactor and so any imperfections in the walls are not experienced by the solution flow. It is therefore presumed that nucleation occurs only once the system has left the metastable zone and the supersaturation ratio necessitates precipitation. This suggests that segmented flow can be used as a tool to prevent any unwanted nucleation in a steady state system. The expression of α-SA has only been reported once previously through solution-based crystallization methods.13 Standard methods of crystallizing α-SA include grinding14 and melt15 crystallization. In air-segmented experiments α-SA was not observed whereas in liquid-segmented experiments where either dichloromethane or PFPE was used carrier fluid α-SA can be observed in the PXRD patterns, occurring alongside the β-SA solid form. The only other reported example of α-SA crystallization through solution-based methods is by spray-drying techniques. In both liquid-segmented flow crystallization and spray-drying there is no interaction with a solid interface. This may imply that the interaction of the growth solution with a solid interface is detrimental to the formation/growth of α-SA. Further studies into this phenomenon are planned. The KRAIC can be used for process intensification of chemical reactions in a similar way to microfluidics. Due to the larger bore of tubing employed in the KRAIC the mixing efficiency will be reduced with respect to microfluidic systems thus having a negative effect on its process intensification potential. However the larger bore of tubing makes it ideal for precipitation reactions, currently inaccessible in microfluidics; it also lends obvious benefits in scale of production. Liquid-segmented flow has been used extensively in microfluidics but its application for the production of solids has thus far been restricted to nanoparticles with few exceptions.1, 16 Flow crystallization has been used for many years but to date the use of these platforms in a truly continuous manner is very rare and flow crystallizers are typically designed to produce kg/h product which is incompatible with most lab-scale applications. The use of any flow reactor for any significant time period (hours or days) necessitates that no adhesion of the particles onto the walls of the reactor takes place. As soon as any particle adheres to / grows on the walls of the reactor this not only affects the subsequent material passing through the reactor having a deleterious effect on the homogeneity of the product, in time it will almost certainly lead to blockage of the reactor. As the KRAIC uses liquid-segmented flow to prevent the precipitate from coming into contact with the walls of the reactor it is possible that this can be used as a truly continuous reactor without the need for periodic cleaning. This is only possible due to the design of an end-piece which separates the carrier-fluid prior to filtration and uses the momentum of a tertiary miscible flow to prevent sedimentation or encrustation of the particles. Conclusions A flow reactor capable of processing precipitation reactions / crystallizations has been developed, and its initial performance and design evaluation shows that this device has great promise in the area of continuous crystallization applications. The modular nature of the KRAIC, range of configurations possible and independent control over feed solutions results in a highly versatile platform. A wide range of crystallizations / reactions can be performed using this apparatus with a high degree of kinetic control. The use of liquid-segmented flow in the KRAIC enables plug flow to be generated and mitigates encrustation whilst providing a sterile environment in which primary nucleation can occur free of external stimulus; nucleation is thus induced in a consistent and reproducible manner, a critical advantage on delivering selectivity and consistency in the solid form particles produced. Design of a system for separation of the carrier-fluid prior to filtration, without impact on the precipitate, enables continuous use of this crystallizer. The small scale of the KRAIC makes it ideal for research labs and the production of high value and rare chemicals. As most of the components of the KRAIC are common lab consumables it is also an economical alternative to standard flow equipment available from commercial suppliers. Acknowledgements The authors would like to thank the EPSRC Centre for Innovative Manufacturing in Continuous Manufacturing and Crystallisation (grant No.: EP/I033459/1), Metastable Materials programme grant (No.: EP/K004956/1) and the University of Bath (P. B. F.) for funding. Specific acknowledgement for aid in optimization of the design go to Dr Chris Price and Prof Jan Sefcik at the University of Strathclyde, Prof Zoltan Nagy at the University of Purdue and Martyn Fordham and Arran Solomonsz of Asynt. References 1. Vacassy, R.; Lemaître, J.; Hofmann, H.; Gerlings, J. H., AIChE Journal 2000, 46 (6), 1241-1252. 2. (a) Brown, C. J.; Adelakun, J. A.; Ni, X.-w., Chemical Engineering and Processing 2015, 97, 180-186; (b) Lawton, S.; Steele, G.; Shering, P.; Zhao, L.; Laird, I.; Ni, X.-W., Org Process Res Dev 2009, 13 (6), 1357-1363; (c) McGlone, T.; Briggs, N. E. B.; Clark, C. A.; Brown, C. J.; Sefcik, J.; Florence, A. J., Org Process Res Dev 2015, 19 (9), 1186-1202; (d) Siddique, H.; Brown, C. J.; Houson, I.; Florence, A. J., Org Process Res Dev 2015, 19 (12), 1871-1881. 3. (a) Jung, W. M.; Kang, S. H.; Kim, K. S.; Kim, W. S.; Choi, C. K., J Cryst Growth 2010, 312 (22), 3331-3339; (b) Jung, W. M.; Kang, S. H.; Kim, W. S.; Choi, C. K., Chem Eng Sci 2000, 55 (4), 733-747. 4. (a) Morris, G.; Power, G.; Ferguson, S.; Barrett, M.; Hou, G.; Glennon, B., Org Process Res Dev 2015, 19 (12), 1891-1902; (b) Powell, K. A.; Saleemi, A. N.; Rielly, C. D.; Nagy, Z. K., Chemical Engineering and Processing 2015, 97, 195-212; (c) Yang, Y.; Nagy, Z. K., Chem Eng Sci 2015, 127, 362-373. 5. (a) Köhler, J. M. K., Andrea, Micro-Segmented Flow: Applications in Chemistry and Biology. Springer: 2014; pp 149-200; (b) Shestopalov, I.; Tice, J. D.; Ismagilov, R. F., Lab on a Chip 2004, 4 (4), 316-321. 6. Guillemet-Fritsch, S.; Aoun-Habbache, M.; Sarrias, J.; Rousset, A.; Jongen, N.; Donnet, M.; Bowen, P.; Lemaı̂tre, J., Solid State Ionics 2004, 171 (1–2), 135-140. 7. (a) Jiang, M.; Zhu, Z. L.; Jimenez, E.; Papageorgiou, C. D.; Waetzig, J.; Hardy, A.; Langston, M.; Braatz, R. D., Cryst Growth Des 2014, 14 (2), 851-860; (b) Lu, J.; Litster, J. D.; Nagy, Z. K., Cryst Growth Des 2015, 15 (8), 3645-3651. 8. Mullin, J. W.; Whiting, M. J. L., Ind Eng Chem Fund 1980, 19 (1), 117-121. 9. Crystallisation, E. C. f. I. M. i. C. M. a. 10. Asynt http://www.asynt.com/. 11. Klapwijk, A. W., C, Crystal Growth and Design Manuscript ID: cg-2016-00465q 2016. 12. Brandel, C.; ter Horst, J. H., Faraday Discussions 2015, 179, 199-214. 13. Carver, K. M.; Snyder, R. C., Industrial & Engineering Chemistry Research 2012, 51 (48), 15720-15728. 14. (a) Chikhalia, V.; Forbes, R. T.; Storey, R. A.; Ticehurst, M., European Journal of Pharmaceutical Sciences 2006, 27 (1), 19-26; (b) Trask, A. V.; Shan, N.; Motherwell, W. D. S.; Jones, W.; Feng, S. H.; Tan, R. B. H.; Carpenter, K. J., Chemical Communications 2005, (7), 880-882. 15. Rieck, G. D., Recueil des Travaux Chimiques des Pays-Bas 1944, 63 (9), 170-180. 16. Aimable, A.; Jongen, N.; Testino, A.; Donnet, M.; Lemaître, J.; Hofmann, H.; Bowen, P., Chemical Engineering & Technology 2011, 34 (3), 344-352.
FOR TABLE OF CONTENTS USE ONLY Design and Evaluation of a Meso-scale Segmented Flow Reactor (KRAIC) Karen Robertson*, Pierre-Baptiste Flandrin, Anneke R. Klapwijk and Chick C. Wilson
The kinetically regulated automated input crystalliser (KRAIC) is a small scale flow crystallization/reaction platform. Using segmented flow a homogenous crystallization environment is created away from any solid-liquid interface resulting in more repeatable nucleation events and preventing fouling. Evaluation and optimization of the apparatus has been carried out using fast growing succinic acid as a model system. Yüklə 71,67 Kb. Dostları ilə paylaş:
|
