A new stacked, continuous, scalable, high performance crystallizer

11

Introduction

 

Crystallization has been an important purification technique in the chemical and pharmaceutical industries for many years with over 90% of active pharmaceutical ingredients (APIs) being purified with one or more crystallization steps (1, 2). Crystallizer operation and control have continued to increase in importance over recent years due to continually increasing stringency of final product requirements for APIs by government agencies worldwide. The crystallization process can affect multiple crystal critical quality attributes (CQAs) including yield and purity (3-5), polymorph control (6, 7), and particle size distribution (PSD) (8) all of which can lead to variations in final product performance by affecting solubility, dissolution rate, and tablet hardness of final APIs (9). However, the increased levels of control required to continue to meet these standards cannot be realized without a fundamental understanding of key operating parameters of crystallization equipment including mixing uniformity and scale up (10).
As it is well known, batch crystallization is the oldest and most commonly used technique. However, adoption of continuous approaches are slowed by 1) existing know-how on batch processes, 2) preference towards relying on already expensed equipment instead of making new investments, 3) revalidation hurdles of converting existing API production processes to flow (2), and 4) lack of technological solutions that provide enough compelling arguments to encourage new technolgy adoption.
Despite the dominance of batch based processes, the USFDA has been showing increased support for continuous manufacturing equipment (11-13) as batch processes are known to have numerous shortcomings, such as issues with scale up, batch-to-batch inconsistencies, non-uniform mixing conditions, and high manufacturing and maintenance costs (2, 14-19). In comparison, continuous crystallization techniques – most commonly utilized through mixed solids mixed product removal (MSMPR) crystallizers – are known to be easier to scale up (20) and can operate at a higher level of supersaturation due to an increased uniformity of mixing (21).
Additionally, increasing the number of tanks in a cascade of MSMPRs is known to provide benefits to the CQAs due to the narrowing of the residence time distribution as the crystals fundamentally have varying probabilities of residence time in the vessel (22-25) as shown in Figure 1.

 

Despite this known benefit, most crystallization examples in the literature which utilize MSMPRs are operated with only 1-3 tanks (26-29) due to the increased complexity of operation associated with each additional tank added as the transfer lines between tanks are known to be prone to fouling and encrustation due to solid settling (3). Additionally, current MSMPRs have a typical volume as low as 100-150mL per tank. Thus a multistage process of just 3 stages requires ~0.5L of process solution to be filled up and an additional 1.5-3L to reach steady state. The need for waiting multiple residence times to reach steady state makes the requirements in terms of availability of process solution practically prohibitive in the early stage of process development thus creating an additional barrier to adoption.

 

KryZr, Zaiput’s Tower Crystallizer (TWC)

 

To be able to access the benefits of a narrow residence time distribution accompanied with utilizing an increased number of tanks in series, Zaiput has proposed a novel concept of vertically stacked tanks called the Tower Crystallizer (TWC) or KryZr. A number of features allow adequate control of slurry movement and control of the temperature profile across the device. The utilization of several tanks provides better plug flow characteristics that are expected to result in improved product quality.
The vertical stacking eliminates the need for external transfer lines and therefore removes one of the relevant sources of fouling faced by operation of cascaded MSMPRs. The demonstration of the new design concept has been recently provided with an 11-tank device. The operating principles behind KryZr can be seen in Figure 2.

 

In brief, a cascade of vertically arranged tanks—each containing an impeller on a common shaft—facilitate the transfer of the slurry between tanks. Perfluoropolymer disks serving as check valves are present between each tank to prevent back-mixing from occurring due to the down pumping action of the impellers. Most importantly, the first and last tanks include moveable surfaces allowing their volumes to be changed during operation. In the bottom tank, a piston is connected to a linear actuator. The controlled and fast inward displacement of the piston against a movable surface generates an upward flow with an adequate velocity to ensures the slurry can be moved from one tank to the next without the settlement of solids; the outward movement of the piston is precisely controlled to match the filling rate of the chamber, eliminating backflow of slurry between chambers. Similarly, in the top “ejection tank” the crystallized slurry is stored until the fast inward movement of the movable surface in this tank ejects it from the device. As there is the ability to add antisolvent to any tank, they all contain access ports able to be outfitted with a moveable surface to prevent the additional antisolvent addition from feeding forward throughout the cascade. As the use of process analytical technology (PAT) probes are known to be critical in order to obtain a final product with desirable properties (30), when not used to house a moveable surface these access ports can also be adapted to hold various PAT probes (i.e. Mettler Toledo, BlazeMetrics, etc.). Finally, a specific design of the internal features of the TWC allow the generation of a linear temperature gradient between tanks by setting the temperature of the first and last tank allowing the device to be capable of cooling and/or antisolvent crystallizations.
The design of the TWC was made with the thought of scale up mind as it is necessary to meet the same regulatory requirements of a final drug during all scales of development (10). As such, equipment which is unable to maintain crystal CQAs upon scale up is of little use to the pharmaceutical industry.
Since the interaction between mixing and crystallization can affect nucleation, growth, and slurry transfer (10), we first began our development effort by determining tank and impeller geometries suitable for obtaining a uniform distribution of particles within a given tank at 3 orders of magnitude: 8mL, 80mL, and 800mL tank volume. Individual tanks were machined in clear acrylic to allow the particles to be viewed while they were mixing. White polyethylene microspheres (WPMS, from Cospheric) with a density of 1.35 g/cm3 and diameter ranges of either 10-32um, 53-63um, 212-250um, 300-355um, or 425-500um were suspended in ethanol (Δρ = 0.56 g/cm3). Images were taken with a macroscopic lens (Canon EOS R100 equipped with an RF 100mm F2.8 Macro Lens) to allow us to visualize the particle distribution as seen in Figure 3.

 

Image analysis software was developed with a custom MATLAB script to determine skew of the vertical and horizontal particle distribution to determine when uniform mixing had been achieved.
It was determined that a dual impeller PBT3 was suitable for uniform solid suspension at all scales and the RPM requirements at varying scales followed the correlation proposed by Zwietering (31).

 

After suitable geometries had been determined, solid particle transfer was tested with a 3 stage MSMPR at the KryZr-10 scale. Particle transfer was compared to theory using the same particle size ranges of the single tank mixing studies. Particles were loaded into the first tank at slurries of up to 200mg solids per mL of tank volume and a similar image analysis software was used to track the number of particles in the second tank of the cascade and compare this concentration to theory. An example RTD curve can be found in Figure 4.

 

Finally, final liquid and solid RTDs were checked on the full 11 stage KryZr-10 system using a dye tracer injected into the first tank and a step change between crystallization yields for the antisolvent crystallization of glycine for the liquid and solid, respectively. Example results of these RTD experiments at a 30 minute residence time can be found in Figure 5.

 

As seen in the figure, both the liquid and solid exit the cascade in good agreement with theory.
Current work is being done in collaboration with an academic group (Myerson Lab at MIT) involving the comparison of antisolvent and/or cooling crystallization of various molecules to existing single stage MSMPR. Initial results highlight the fact that the increased number of tanks and the possibility to feed antisolvent at several locations combine to a provide narrower particle size distribution. A full report on this will be published in the near future. Meanwhile, we are furthering the design and testing of larger versions of the KryZr device for pilot and production applications.

 

Discussion

The TWC design proposed provides a number of advantages over existing approaches. Improved plug flow characteristics can be attained with a larger number of tanks in series (11 tanks). A very accurate control of supersaturation conditions can be achieved by taking advantage of the possibility to add antisolvent at several locations. This is an important feature for process conditions control; the smaller crystallization driving force at each stage favors growth as opposed to nucleation creating a basis for a narrower particle size distribution. The ability to control temperature gradients accurately provides an additional process control tool.
Finally, the scalability allows for simplification of the process development burden and cost as conditions optimized in at the laboratory scale are expected to be able to be reproduced at different scales without any substantial additional effort. The laboratory scale device with 11 stages has an internal volume of ~80mL thus drastically reducing the amount of process solution needed to develop a continuous process.

 

Conclusion

 

A novel tower crystallizer design has been presented. Data collected on an 11 stage, vertically stacked MSMPR demonstrates excellent mixing, ability to effectively move slurries, and accurate temperature control. Importantly, the device has been developed with the necessity of scale up in mind; key hurdles involving solid mixing and slurry transfer have already been dealt with to ensure reproducibility of crystal CQAs upon scale.
The ability to obtain uniform solid mixing has been demonstrated on tank volumes at 3 different orders of magnitude and the lab scale device has shown to be able of isokinetic withdrawal at slurry concentrations of up to 200mg of solids per mL. Current work is ongoing regarding the comparison of the TWC to existing MSMPR equipment and the slurry transfer is being tested at larger scales.

 

References and notes

 

1. Alvarez AJ, Myerson AS. Continuous Plug Flow Crystallization of Pharmaceutical Compounds. Crystal Growth & Design. 2010 May 5;10(5):2219–28.
2. Chen J, Sarma B, Evans JMB, Myerson AS. Pharmaceutical Crystallization. Crystal Growth & Design. 2011 Apr 6;11(4):887–95.
3. Myerson AS, Deniz Erdemir, Lee AY. Handbook of industrial crystallization. Cambridge, United Kingdom ; New York, Ny Cambridge University Press; 2019.
4. Tung HH, Paul EL, Midler M, McCauley JA. Crystallization of Organic Compounds. John Wiley & Sons; 2009.
5. Yemane Andemichael, Chen J, Clawson JS, Dai W, Diederich A, Downing SV, et al. Process Development for A Novel Pleuromutilin-Derived Antibiotic. Organic Process Research & Development. 2009 Jul 2;13(4):729–38.
6. Chemburkar SR, Bauer J, Deming K, Spiwek H, Patel K, Morris J, et al. Dealing with the Impact of Ritonavir Polymorphs on the Late Stages of Bulk Drug Process Development. Organic Process Research & Development. 2000 Sep;4(5):413–7.
7. Llinàs A, Goodman JM. Polymorph control: past, present and future. Drug Discovery Today. 2008 Mar;13(5-6):198–210.
8. Mullin JW. Crystallization. Oxford: Butterworth-Heinemann; 2001.
9. Gao Z, Rohani S, Gong J, Wang J. Recent Developments in the Crystallization Process: Toward the Pharmaceutical Industry. Engineering. 2017 Jun;3(3):343–53.
10. Paul EL, Tung HH, Midler M. Organic crystallization processes. Powder Technology. 2005 Feb; 150(2):133–43.
11. Yang X, Acevedo D, Mohammad A, Pavurala N, Wu H, Brayton AL, et al. Risk Considerations on Developing a Continuous Crystallization System for Carbamazepine. Organic Process Research & Development. 2017 Jul 8;21(7):1021–33.
12. Lee SL, O’Connor TF, Yang X, Cruz CN, Chatterjee S, Madurawe RD, et al. Modernizing Pharmaceutical Manufacturing: from Batch to Continuous Production. Journal of Pharmaceutical Innovation. 2015 Mar 19;10(3):191–9.
13. Chatterjee S. FDA Perspective on Continuous Manufacturing IFPAC Annual Meeting Baltimore. 2012.
14. Ganko K, Berliner MD, Rhyu J, Wu L, Braatz RD, Leyffer S. Population Balance Model-based Dynamic Multiobjective Optimization of Yeast Cell Manufacturing. IFAC-PapersOnLine. 2024;58(14):860–7.
15. Ranodolph A, Larson MA. Theory of Particulate Processes. Academic Press; 1988.
16. Lawton S, Steele G, Shering P, Zhao L, Laird I, Ni XW. Continuous Crystallization of Pharmaceuticals Using a Continuous Oscillatory Baffled Crystallizer. Organic Process Research & Development. 2009 Nov 20;13(6):1357–63.
17. Kessel M. The problems with today’s pharmaceutical business—an outsider’s view. Nature Biotechnology. 2011 Jan;29(1):27–33.
18. Schaber SD, Gerogiorgis DI, Ramachandran R, Evans JMB, Barton PI, Trout BL. Economic Analysis of Integrated Continuous and Batch Pharmaceutical Manufacturing: A Case Study. Industrial & Engineering Chemistry Research. 2011 Sep 7;50(17):10083–92.
19. Simon LL, Pataki H, Marosi G, Meemken F, Hungerbühler K, Baiker A, et al. Assessment of Recent Process Analytical Technology (PAT) Trends: A Multiauthor Review. Organic Process Research & Development. 2015 Jan 8;19(1):3–62.
20. GUILLEMETFRITSCH S. High-quality nickel manganese oxalate powders synthesized in a new segmented flow tubular reactor. Solid State Ionics. 2004 Jun;171(1-2):135–40.
21. Ferguson SH, Morris GA, Hao H, Barrett M, Glennon B. Characterization of the anti-solvent batch, plug flow and MSMPR crystallization of benzoic acid. 2013 Dec 18;104:44–54.
22. Park K, Kim DY, Yang DR. Operating Strategy for Continuous Multistage Mixed Suspension and Mixed Product Removal (MSMPR) Crystallization Processes Depending on Crystallization Kinetic Parameters. Industrial & Engineering Chemistry Research. 2016 Jun 22;55(26):7142–53.
23. Jones AG. Crystallization Process Systems. Elsevier; 2002.
24. Randolph AD, Chand Deepak, Iskander M. On the narrowing of particle‐size distributions in staged vessels with classified product removal. AIChE Journal. 1968 Sep 1;14(5):827–30.
25. Zhang H, Quon J, Alvarez AJ, Evans J, Myerson AS, Trout B. Development of Continuous Anti-Solvent/Cooling Crystallization Process using Cascaded Mixed Suspension, Mixed Product Removal Crystallizers. Organic Process Research & Development. 2012 Mar 29;16(5):915–24.
26. Li J, Trout BL, Myerson AS. Multistage Continuous Mixed-Suspension, Mixed-Product Removal (MSMPR) Crystallization with Solids Recycle. Organic Process Research & Development. 2015 Dec 29;20(2):510–6.
27. Ferguson S, Ortner F, Quon J, Peeva L, Livingston A, Trout BL, et al. Use of Continuous MSMPR Crystallization with Integrated Nanofiltration Membrane Recycle for Enhanced Yield and Purity in API Crystallization. Crystal Growth & Design. 2013 Dec 19;14(2):617–27.
28. Liu F, Li H, Li Y. Breakage-Facilitated Mixed-Suspension-Mixed-Product-Removal (MSMPR) Crystallization of Pharmaceutical Compounds. Crystal Growth & Design. 2024 Jan 31;24(4):1591–602.
29. Alvarez AJ, Singh A, Myerson AS. Crystallization of Cyclosporine in a Multistage Continuous MSMPR Crystallizer. Crystal Growth & Design. 2011 Oct 5;11(10):4392–400.
30. Yu ZQ, Chew JW, Chow PS, Tan RBH. Recent Advances in Crystallization control. Chemical Engineering Research and Design. 2007 Jan;85(7):893–905.
31. Zwietering ThN. Suspending of solid particles in liquid by agitators. Chemical Engineering Science. 1958 Jun 1;8(3):244–53.

 

 

 

Figure 1. Demonstration of the narrowing of the residence time distribution curve by increasing the number of tanks in series.

 

Figure 2. Schematic of main features of the TWC (left), picture of the KryZr-10 (right).

 

Figure 3. High speed images of the particle distribution in single acrylic tanks using 425-500um WPMS with a density of 1.35g/cm3 suspended in ethanol at the KryZr-10 (a), KryZR-100 (b), and KryZr-1000 (c) scales.

 

Figure 4. Experimentally obtained RTD for the second tank of an MSMPR cascade using 212-250um WPMS with a density of 1.35g/cm3 suspended in ethanol with an initial slurry concentration of 200mg solids per mL in the first tank. This high speed photography data confirms the ability to adequately move slurry within the device.

 

Figure 5. Normalized RTD curves using a liquid dye tracer (left) and a solid step change between crystallization yields for the antisolvent crystallization of glycine (right) for the full 11 stage KryZr-10 system with a 30 minute residence time demonstrating the device follows theory.