Rapid polymorphic screening using sessile microdroplets: competitive nucleation of mannitol polymorphs†

Ruel Cedeno, Romain Grossier, Nadine Candoni and Stéphane Veesler*
CNRS, Aix-Marseille Université, CINaM (Centre Interdisciplinaire de Nanosciences de Marseille), Campus de Luminy, Case 913, F-13288 Marseille Cedex 09, France. E-mail: stephane.veesler@cnrs.fr

First published on 22nd August 2024

---

---

Introduction

Polymorphism, the ability of a material to form multiple crystal structures, plays a crucial role in numerous fields, notably in the pharmaceutical industry in which the solid forms directly impact the bioavailability, processability, and overall quality of the final product.¹ For this reason, understanding the kinetics of competitive polymorphic nucleation is essential in controlling the polymorphic outcome.² Given that nucleation is inherently stochastic,³ experimental platforms allowing a large number of experiments are needed to access the probability distribution of polymorphic selectivity as a function of various experimental conditions.^4–6

In our previous work, we have developed a microfluidic platform that allows extraction of thermodynamic and kinetic parameters of nucleation in contracting sessile microdroplets.^3,7 In this work, we further extend its application in the study of polymorphism. Using D-mannitol in water as a model system, we demonstrate how our platform enables rapid screening of polymorphic outcomes. Moreover, it also allows statistical investigation of the influence of volume and supersaturation ratio S (i.e. c_sol/c_eq. where c_sol is the solution concentration and c_eq. is the saturation concentration of the β polymorph) on polymorphic selectivity. We then rationalize our observed polymorphic distributions using classical nucleation theory, revealing interesting insights into the interplay of thermodynamics and kinetics in the stochastic nucleation of polymorphs.

Materials and methods

D-Mannitol (≥98%), a crystalline excipient, was purchased from Sigma Aldrich and verified as the β form by Raman spectroscopy. The β form is the thermodynamically stable polymorph of mannitol, while α and δ forms are metastable (stability order: δ < α < β at room temperature).⁸ CCDC numbers are 224658, 224659 and 22460 for polymorphs α, β and δ respectively.

Microdroplet generation and nucleation time detection are based on a previously reported experimental setup and protocol in ref. 7, consisting in deliquescence/recrystallization cycling.

However, unlike NaCl that dissolves upon exposure to humidity (RH > 75%), D-mannitol does not absorb enough moisture to undergo deliquescence, preventing the use of the RH cycling technique. We thus modified the procedure by generating initially undersaturated arrays of sessile microdroplets on PMMA coated glass immersed in a thin layer of polydimethylsiloxane (PDMS) oil (10 cSt) under ambient conditions (1 atm, 25 °C). This is done at RH close to 100% to minimize possible evaporation during microdroplet generation. Once the desired number of microdroplets is generated, the RH is lowered to 10%, causing the droplets to evaporate by diffusion of water in oil and eventually nucleate. Using a tailor-made evaporation model⁹ (details in section S1 in the ESI†), the supersaturation ratio of the microdroplets can be obtained as a function of time. The crystallized droplets are analysed in situ at the end of the experiment using a Kaiser RXN1 Raman microscope system. Measurements are made at room temperature using a 785 nm laser.

Results and discussion

Polymorph screening and characterization

To demonstrate how our sessile microdroplet approach can accelerate polymorph screening, we used D-mannitol in water as a case study. The bottom-view image of the generated sessile microdroplets is shown in Fig. 1a (see the video in the ESI† for the whole process, droplet contraction until crystallization). Supersaturation at generation is S₀ = 0.25.

The resulting crystal from each microdroplet was then characterized using Raman spectroscopy. Three distinct spectra were obtained corresponding to the α, β, δ forms as shown in Fig. 1b. These spectra exhibit sufficiently resolved characteristic peaks for precise phase identification.

In contrast to conventional screening methods, our approach offers several advantages: (1) the relatively high supersaturation level achieved in microdroplets facilitates the formation of metastable polymorphs; (2) the linear arrangement of the immobilized crystals facilitates systematic labeling of each sample with respect to its position in the 2D array, allowing us to generate a cartography of the polymorph as shown in Fig. 1c. (3) The statistical distribution of polymorphic selectivity can be analyzed as a function of supersaturation ratio S as shown in Fig. 1d. We use image acquisition to track the droplets as they evaporate, so we know when nucleation occurred and at what concentration. We use Raman spectroscopy to characterize the nucleated phase. (4) The open geometry microfluidics ensures the formation of crystal for each droplet, unlike closed microfluidics (chips or tubing) where droplets can remain metastable for several days to weeks. (5) The method is relatively rapid, as the entire experiment from droplet generation to characterization took less than 4 hours in this work.

Influence of droplet volume

In this communication, we focus on the influence of droplet volume on the cartography of polymorphism. The droplet volume is an important parameter in microfluidic crystallization as it has a direct impact on the surface area to volume ratio as well as the achievable supersaturation level (which influences the relative importance of homogeneous and heterogeneous nucleation mechanisms), and on the degree of thermodynamic confinement (i.e. reduction of effective supersaturation upon formation of pre-critical clusters in small volumes).¹⁰ Thus, it would be interesting to understand how the droplet volume impacts the polymorphic selectivity of D-mannitol.

The polymorphic distribution for various volumes at saturation (0.2, 0.3, 0.7, 1.5, and 5 nL) is shown in Fig. 2a. Notice that the physical mixtures (α + β, α + δ, β + δ) occupy a non-negligible fraction across different volumes. The formation of polymorphic mixtures can be interpreted in three different ways:

Case 1: independent nucleation: mixtures arise from the independent/concomitant nucleation event of each polymorph.

Case 2: Ostwald's rule of stages: the less stable polymorph nucleates first then gradually transforms to a more stable polymorph.

Case 3: surface nucleation: the more stable polymorph nucleates first, which then facilitates the nucleation of the less stable ones on its surface.

It should be noted that the method used to measure induction time makes it possible to observe simultaneously hundreds of droplets at the cost of a loss of resolution,¹¹ which makes it impossible to distinguish between the 3 cases described above.

The resulting distribution based on case 1, case 2, and case 3 representations is plotted in Fig. 2b, c, and d respectively, considering only the first polymorph that appeared according to case 2 or 3 respectively.

Among these three mixture interpretations, we believe case 3 is highly unlikely because as the stable form nucleates and grows, it quickly depletes the supersaturation level in the droplet which consequently dissolves the precursors/pre-critical clusters of the more soluble metastable forms. Therefore, in the following discussion, we will concentrate on cases 1 and 2, for which it is impossible to give a definitive answer.

Recall that the order of stability for D-mannitol polymorphs is δ < α < β. During droplet contraction, the experimental conditions inside each droplet move on the phase diagram from undersaturated to supersaturated. They first cross the solubility curve of form β, before crossing the solubility curve of form α and then δ. The same trend is expected for the metastable limit of the three forms. Therefore, we would expect that the stable β form will dominate at larger volumes. This is due to the lower surface area to volume ratio of larger droplets, which implies that their supersaturation ratio S increases more slowly (Fig. S2b in the ESI†), allowing more time in the stable β-form nucleation area. Meanwhile the least stable δ form will dominate at smaller volumes due to the higher surface area to volume ratio of smaller droplets, and so their supersaturation ratio S increases more rapidly (Fig. S2b in the ESI†). Hence, metastable products are favoured according to Ostwald's rule of stages, as previously observed by the Myerson group in a confined environment for sulfathiazole and glycine^12,13 and by Buanz et al. for crystallization in printed droplets of D-mannitol.¹⁴ As shown in Fig. 2b and c, the trend in % δ-form qualitatively agrees with our hypothesis, i.e., it tends to increase as the volume decreases (except at 0.2 nL). Surprisingly, the % β-form does not increase with volume, and the least stable δ-form dominates across all studied volumes. Interestingly, the medium stable α-polymorph follows the expected trend of the stable polymorph.

Predictions of classical nucleation theory

To gain insights into the observed dominance of the δ-form, we used the classical nucleation theory (CNT) to model the polymorphic selectivity as a function of supersaturation. Recall that CNT requires two parameters, the pre-exponential factor A (kinetic parameter) and the effective interfacial energy (between the crystal and solution) γ_eff (thermodynamic parameter). In the linearized form of the equation for heterogeneous nucleation, CNT can be written as:

	(1)

where J is the nucleation rate, ρ_s is the number density (formula units per m³) and k_bT is the thermal energy.

For simplicity, given that the mass transfer properties (i.e. diffusivity, viscosity, etc.) in the liquid phase are identical regardless of the solid form, we can suppose that parameter A (related to the mass transfer rate towards the nucleus) is similar for all polymorphs. This assumption is similar to that of Sato¹⁵ and Deij et al.¹⁶ Consequently, the relative nucleation rate ln(J/A) is only a function of γ_eff. In principle, γ_eff can be measured from the probability distributions of nucleation time for each polymorph. Unfortunately, given the dominance of the δ-form, the number of data points for other polymorphs is not sufficient to extract reliable statistical distribution. For this reason, we decided to apply the empirical correlation of Mersmann¹⁷ which correlates the interfacial energy (between crystal and solution) γ_SL with the solubility, written as

	(2)

with K as an empirical parameter, c_sol and c_eq. as the equilibrium solid-state and liquid-phase concentrations (in mol m⁻³), respectively, and N_A as Avogadro's number. γ_SL is used for homogeneous nucleation in CNT, eqn (1), for more details see ref. 18. Using the solubility data of Su et al.¹⁹ (values in Table S1 in the ESI†) together with eqn (2), we obtained γ_SL as 9.6, 8.6 and 8.1 mJ m⁻² for β, α and δ, respectively, in agreement with the rule: the smaller the interfacial energy, the larger the solubility. The resulting relative nucleation rate ln(J/A) and the relative polymorphic selectivity (i.e. the probability of forming the polymorph, eqn (5) of ref. 20) are plotted in Fig. 3a and b respectively.

To interpret the results in Fig. 3, remember that nucleation time is inversely proportional to the droplet volume (t_n = 1/JV). Under continuous evaporation, smaller droplets can therefore achieve higher supersaturation within the nucleation zone. Thus, Fig. 3b suggests that larger droplets would favor the α-form while the smaller ones would favor the δ-form. Reassuringly, this behavior is coherent with what we observe in Fig. 2b and c, showing the predominance of polymorphs α and δ, except that of the 0.2 nL dataset. While one could speculate several possible explanations, such a change in trend at very low volumes could be likely due to confinement effects¹⁰ which lower the effective supersaturation and consequently promote the α-form. Moreover, diminishing droplet volume provides a higher surface to volume ratio, potentially affecting nucleation mechanisms and rates.

Indeed, our microfluidic platform and modeling approach reveal interesting insights into the impact of volume on the stochastic nucleation of mannitol polymorphs. Moreover, our approach can be extended to study other polymorphic materials of interest.

Conclusions

In this work, we developed a rapid polymorphic screening approach based on contracting sessile microdroplets, which offers several advantages: (1) achieves very high supersaturation to facilitate formation of metastable forms; (2) allows systematic labeling of samples; (3) gives access to the statistical distribution of polymorphic selectivity as a function of experimental conditions; (4) ensures the formation of crystal for each droplet, addressing the problem of uncrystallized droplets in traditional microfluidics.

Thanks to this platform, we studied the competitive nucleation of D-mannitol polymorphs and investigated the effect of droplet volume on polymorphic selectivity. We showed that our observed polymorph distributions at different volumes are qualitatively consistent with the predictions of the classical nucleation theory except for very small volumes where thermodynamic confinement or surface effects could play a substantial role.

Overall, our microfluidic approach can be a promising tool not only for routine screening of pharmaceutical polymorphs in the industrial context but also in the fundamental understanding of the mechanisms underlying the competitive nucleation of polymorphs.

Data availability

Data for this article are available at zenodo at https://doi.org/10.5281/zenodo.12665873.

Author contributions

Ruel Cedeno: formal analysis (equal); methodology (equal); writing – review & editing (equal). Romain Grossier: supervision (equal); formal analysis (equal); conceptualization (equal); writing – review & editing (equal). Nadine Candoni: supervision (equal); conceptualization (equal); writing – original draft (equal). Stéphane Veesler: supervision (equal); conceptualization (equal); writing – original draft (equal).

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

RC acknowledges the financial support of ANR-FACET, ANR-19-CE08-0014-02.

References

1. D. Mangin, F. Puel and S. Veesler, Org. Process Res. Dev., 2009, 13, 1241–1253 CrossRef CAS .
2. R. Cedeno, A. Cruz-Cabeza, R. Drummond-Brydson, M. K. Dudek, K. Edkins, K. Fichthorn, A. R. Finney, I. Ford, J. M. Galloway, R. Grossier, J. Kim, C. Kuttner, L. Maini, F. Meldrum, M. Miller, P. Morris, S. O. Nilsson Lill, B. Pokroy, S. Price, I. B. Rietveld, J. Rimer, K. Roberts, J. Rogal, M. Salvalaglio, J. Sefcik, W. Sun, S. Veesler, P. Vekilov, H. Wheatcroft, M. Whittaker and R. Zhao, Faraday Discuss., 2022, 235, 508–535 RSC .
3. R. Cedeno, R. Grossier, N. Candoni, N. Levernier, A. E. Flood and S. Veesler, J. Chem. Phys., 2023, 158, 194705 CrossRef CAS PubMed .
4. M. Lambert, R. Grossier, M. Lagaize, T. Bactivelane, V. Heresanu, B. Robert, N. Candoni and S. Veesler, J. Cryst. Growth, 2023, 616, 127252 CrossRef CAS .
5. S. Zhang, C. J. J. Gerard, A. Ikni, G. Ferry, L. M. Vuillard, J. A. Boutin, N. Ferte, R. Grossier, N. Candoni and S. Veesler, J. Cryst. Growth, 2017, 472, 18–28 CrossRef CAS .
6. G. Peybernès, R. Grossier, F. Villard, P. Letellier, N. Candoni and S. Veesler, Cryst. Growth Des., 2020, 20, 3882–3887 CrossRef .
7. R. Cedeno, R. Grossier, M. Lagaize, D. Nerini, N. Candoni, A. E. Flood and S. Veesler, Faraday Discuss., 2022, 235, 183–197 RSC .
8. F. M. Penha, A. Gopalan, J. C. Meijlink, F. Ibis and H. B. Eral, Cryst. Growth Des., 2021, 21, 3928–3935 CrossRef CAS PubMed .
9. R. Cedeno, R. Grossier, V. Tishkova, N. Candoni, A. E. Flood and S. Veesler, Langmuir, 2022, 38, 9686–9696 CrossRef CAS PubMed .
10. R. Grossier and S. Veesler, Cryst. Growth Des., 2009, 9, 1917–1922 CrossRef CAS .
11. R. Grossier, V. Tishkova, R. Morin and S. Veesler, AIP Adv., 2018, 8, 075324 CrossRef .
12. A. Y. Lee, I. S. Lee and A. S. Myerson, Chem. Eng. Technol., 2006, 29, 281–285 CrossRef CAS .
13. I. S. Lee, A. Y. Lee and A. S. Myerson, Pharm. Res., 2008, 25, 960–968 CrossRef CAS PubMed .
14. A. Buanz, M. Gurung and S. Gaisford, CrystEngComm, 2019, 21, 2212–2219 RSC .
15. K. Sato, J. Phys. D: Appl. Phys., 1993, 26, B77–B84 CrossRef CAS .
16. M. A. Deij, J. H. ter Horst, H. Meekes, P. Jansens and E. Vlieg, J. Phys. Chem. B, 2007, 111, 1523–1530 CrossRef CAS PubMed .
17. A. Mersmann, J. Cryst. Growth, 1990, 102, 841–847 CrossRef CAS .
18. M. Ildefonso, N. Candoni and S. Veesler, Cryst. Growth Des., 2013, 13, 2107–2110 CrossRef CAS .
19. W. Su, H. Hao, B. Glennon and M. Barrett, Cryst. Growth Des., 2013, 13, 5179–5187 CrossRef CAS .
20. W. Sun and G. Ceder, CrystEngComm, 2017, 19, 4576–4585 RSC .

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ce00648h

This journal is © The Royal Society of Chemistry 2024