(660d) Tackling Needle-like Particles in Industrial Crystallisation Via Solvent Screening, Polymorph Engineering, and Mechanical Impacts

Crystallisation is a vital unit operation in various industries such as food, fine chemicals, pharmaceuticals, agrochemicals, and animal health [1], [2], [3]. Product attributes of the crystallisation operation, including crystal shape and size distribution, have a significant impact on the performance of subsequent downstream units (e.g., drying, filtration). Particularly, needle-shaped crystal particles cause inefficiencies due to their flowability, compressibility, and clogging properties [4], [5]. Shape of crystals are influenced by solution properties, including the selected solvent, the supersaturation level, and impurities [1], [6]. Recent developments on computational crystal shape prediction tools, e.g., CrystalGrower [7], enabled convenient solvent screening to engineer the shape of crystalline products. For materials exhibiting polymorphism, tuning the solvent choice and supersaturation towards obtaining a certain product form with a desired shape is also an approach [8]. From a mechanical perspective, introduction of systematic breakage into crystallisation has shown to reduce the aspect ratio of needle-like crystals [9]. As a novel technique, spherical agglomeration has demonstrated to tackle the issues related with needle-like particles via merging fine needle particles into spherical agglomerates for improved processing [4]. This work aims to develop a toolbox to engineer needle-shaped crystals in industrial crystallisation processes, with using a confidential substance, referred to “Substance X” in this study, as an example case. Presented results include solvent screening, polymorph engineering, and shape engineering via integrated crystal breakage techniques.

Regarding the methodology, evaporative and cooling crystallisation techniques were employed. Solvent screening studies involved, by increasing relative polarity (RP), toluene, dichloromethane (DCM), acetone, acetonitrile, and ethanol. Evaporative crystallisation experiments involved 10 ml glass vials filled with saturated solutions of Substance X. The vials were enclosed with glass caps in slow evaporation experiments, whereas they were let natural evaporation in fast evaporation experiments. Crystal formation and growth processes were monitored with an automatised setup involving an RS Pro camera carried by a programmable robotic arm, controlled by an in-house developed Python script that allows periodic imaging of several crystallisation setups, concurrently (Figure 1a). Cooling crystallisation was facilitated by a recirculating water bath setup. The process was initiated by preparation of 10 ml saturated solutions, and followed by reducing their temperature to a level that corresponds to the desired supersaturation (Figure 1b). Mechanical breakage of isolated crystals was facilitated by a scalpel. Forms of Substance X were studied via X-ray powder diffraction (XRD). Product shapes and aspect ratios were analysed via stereomicroscopy and ImageJ software.

Regarding the outcomes of the solvent screening, two polymorphs of Substance X were confirmed via XRD. Form I is the thermodynamically stable form, and features needle-like particle shape. Form II is a metastable form, featuring plate-like particles. Form II was obtained with acetone and DCM in cooling crystallisation experiments, whereas the remaining solvents resulted in Form I. Regarding the appearance, Form II crystals characterised two-dimensional habits as plates in DCM, and elongated plates in acetone (Figure 2). The underlying factor for Form II production was deemed unrelated with the temperature and supersaturation level (S = c/c*). Accordingly, the same respective combination of 4°C and S = 1.6 yielded Form II in DCM but not in toluene and acetonitrile. Likewise, at 4°C and S = 1.3, Form II was obtained from acetone, while Form I was obtained from ethanol. Complementing the cooling crystallisation outcomes, only Form II crystals were obtained only in DCM in slow evaporative crystallisation experiments. In higher evaporation rates, all solvents including DCM resulted in Form I crystals, indicating the preference of Form II in lower supersaturations for DCM. Moreover, Form I and Form II were concomitantly obtained in toluene in evaporative crystallisation cases (Figure 3). Since cooling crystallisation studies indicated that the form preference is not due to the supersaturation level, it is currently hypothesized that specific solvent properties or solvent-solute interactions could be a major factor. Studies to explore the underlying factors for form preference are currently ongoing.

Based on the solvent screening outcomes, seeding was explored as an application for polymorph engineering. In fast evaporation experiments, Form II crystals produced in DCM were seeded to DCM and cross-seeded to acetonitrile. Single Form II crystals continued to grow in DCM, whereas they acted as a surface for heterogeneous nucleation of Form I in acetonitrile (Figure 4). This showcased seeding of Form II as an application for form engineering, whereas cross-seeding to a solvent with Form I tendency resulted in a preference of Form I nucleation over Form II growth in fast evaporation conditions. Cross-seeding to different solvents and cross-seeding with lower evaporation rates are currently studied, aiming to explore the underlying reasons for the observed behaviour, as well as to develop a structured workflow to utilise cross-seeding as an industrial shape engineering approach.

The effect of mechanical impact on aspect ratio was investigated via regrowth experiments involving isolated single Form I crystals of Substance X produced in acetonitrile (Figure 5). The results demonstrated that, after a breakage by a scalpel, the attrited sites on the crystal surface also acted as sites for secondary nucleation, resulting in a more branched growth for the crystal as a whole. Ultimately, the aspect ratio of crystals was reduced around 4 times compared to the starting crystal before the breakage (Figure 6). The results are in line with the literature, as reducing aspect ratios by branching were reported [10]. Conversely, the undamaged crystal (Figure 5) demonstrated a unidirectional regrowth without considerable secondary nucleation, increasing its aspect ratio from 62 to 143 throughout the regrowth. As a relevant industrial concept, the applicability of spherical agglomeration [4] to Form I of Substance X was tested via isolating and growing agglomerated chunks of needle-like crystals with multiple growth sites. The results demonstrated the applicability of spherical agglomeration to Substance X Form I as a scalable approach to reduce the aspect ratio of the product (Figure 7), which can be implemented to industrial processes via proper introduction of analogous breakage inducement apparatus.

This work demonstrates the applicability of solvent screening, polymorph engineering, and utilisation of mechanical impacts to tackle needle-shaped crystals in an industrial case. Current research directions include exploring the underlying factors for Form II production, and transferability of the approaches to certain impurities with similar effects, since adding impurities is an industrially more convenient switch to an existing process considering factors such as plant safety adjustments and material compatibility of the surrounding equipment. Holistically, the developed methods can guide new processes in their early-stage development, so that parameters yielding desired crystal product shapes with low economic and environmental costs can be selected, and the process can be feasibly designed to accommodate this selection.

References

[1] A. S. Myerson, D. Erdermir, and A. Y. Lee, Handbook of Industrial Crystallization, 3rd ed. Cambridge University Press, 2019.

[2] F. Zhang, B. Shan, Y. Wang, Z. Zhu, Z. Q. Yu, and C. Y. Ma, ‘Progress and Opportunities for Utilizing Seeding Techniques in Crystallization Processes’, Org Process Res Dev, vol. 25, no. 7, pp. 1496–1511, 2021, doi: 10.1021/acs.oprd.1c00103.

[3] K. A. Powell, D. M. Croker, C. D. Rielly, and Z. K. Nagy, ‘PAT-based design of agrochemical co-crystallization processes: A case-study for the selective crystallization of 1:1 and 3:2 co-crystals of p-toluenesulfonamide/triphenylphosphine oxide’, Chem Eng Sci, vol. 152, pp. 95–108, 2016, doi: 10.1016/j.ces.2016.06.005.

[4] C. W. Chen, H. L. Lee, K. L. Yeh, and T. Lee, ‘Effects of Scale-Up and Impeller Types on Spherical Agglomeration of Dimethyl Fumarate’, Ind Eng Chem Res, vol. 60, no. 30, pp. 11555–11567, 2021, doi: 10.1021/acs.iecr.1c02006.

[5] F. Civati, C. O’Malley, A. Erxleben, and P. McArdle, ‘Factors Controlling Persistent Needle Crystal Growth: The Importance of Dominant One-Dimensional Secondary Bonding, Stacked Structures, and van der Waals Contact’, Cryst Growth Des, vol. 21, no. 6, pp. 3449–3460, Jun. 2021, doi: 10.1021/acs.cgd.1c00217.

[6] M. Salvalaglio, T. Vetter, M. Mazzotti, and M. Parrinello, ‘Controlling and Predicting Crystal Shapes: The Case of Urea’, Angewandte Chemie, vol. 125, no. 50, pp. 13611–13614, Dec. 2013, doi: 10.1002/ange.201304562.

[7] P. R. Spackman, A. J. Walisinghe, M. W. Anderson, and J. D. Gale, ‘CrystalClear: an open, modular protocol for predicting molecular crystal growth from solution’, Chem Sci, vol. 14, pp. 7192–7207, 2023, doi: 10.1039/d2sc06761g.

[8] J. V. Parambil, S. K. Poornachary, R. B. H. Tan, and J. Y. Y. Heng, ‘Influence of solvent polarity and supersaturation on template-induced nucleation of carbamazepine crystal polymorphs’, J Cryst Growth, vol. 469, no. September 2016, pp. 84–90, 2017, doi: 10.1016/j.jcrysgro.2016.09.058.

[9] F. Salvatori and M. Mazzotti, ‘Experimental characterization and mathematical modeling of breakage of needle-like crystals in a continuous rotor-stator wet mill’, Cryst Growth Des, vol. 18, no. 10, pp. 5957–5972, 2018, doi: 10.1021/acs.cgd.8b00761.

[10] S. Vázquez-Martín, T. Kuhn, and S. Eliasson, ‘Shape dependence of falling snow crystals’ microphysical properties using an updated shape classification’, Applied Sciences (Switzerland), vol. 10, no. 3, 2020, doi: 10.3390/app10031163.