Continuous Production of Spherical Multicrystals ... - ACS Publications

May 1, 2017 - (8) Lu, J.; Litster, J. D.; Nagy, Z. K. Cryst. Growth Des. 2015, 15 (8),. 3645−3651. (9) Robertson, K.; Flandrin, P.-B.; Klapwijk, A. ...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/crystal

Continuous Production of Spherical Multicrystals by Extractive Crystallization in a Droplet Based Fluidic Device Martin Kalný, Anna Pittermannová, Aleš Zadražil, and František Štěpánek* Department of Chemical Engineering, University of Chemistry and Technology Prague, Technicka 5, 166 28 Prague 6, Czech Republic S Supporting Information *

ABSTRACT: A novel continuous fluidic method based on extractive crystallization for producing monodisperse spherical crystal aggregates with potential use in pharmaceutical engineering is presented. A model system of aqueous solution containing KCl as the precursor dispersed phase and 1-hexanol as the continuous extraction phase was investigated experimentally. Monodisperse droplets of the aqueous phase were generated by a fluidic T-junction in the stable dripping regime. Supersaturation was achieved by extraction of the aqueous phase into the organic phase. The effect of extraction rate on the size, morphology, and internal pore structure of the resulting spherical crystal aggregates was systematically investigated. It was shown that nearly monodisperse particles with a tunable size can be produced by adjusting the initial solute concentration. The dissolution rates of the resulting spherical aggregates were measured and compared with single cubic crystals of identical mass. It was shown that a four- to five-fold improvement in the dissolution rate constant can be obtained due to the internal porosity of the multicrystals, compared to single cubic crystals of identical size. antisolvent, or by a combination of these processes.15 Cooling crystallization in droplet-based fluidic systems has been implemented by on-chip Peltier elements16 or by means of a convective heat exchange.17 Antisolvent crystallization in a fluidic setup has also been implemented and shown to be useful, e.g., for the screening of polymorphs18 or for the crystallization of heat sensitive substances.19 However, due to the mutual miscibility of solvents typically used during antisolvent crystallization, the identity of the original droplets tends to be lost, and with it the benefits of spatial confinement. Evaporative crystallization in droplets requires an open configuration for the solvent vapors to escape, and such setup has been recently demonstrated by dropping aqueous solution into a hot oil bath.17 An alternative to evaporative crystallization without the drawbacks of generating large volumes of vapor at elevated temperatures is extractive crystallization.20,21 Unlike antisolvent crystallization, which relies on a pair of miscible solvents, extractive crystallization is based on liquid−liquid systems with limited miscibility, such that the solubility of solvent A (e.g., water) in solvent B (e.g., aliphatic alcohol) is significantly higher than vice versa.21,22 Supersaturation is created by extracting solvent A into solvent B, thereby increasing the concentration of the solute in droplets of solvent A. The

1. INTRODUCTION Batch crystallization in bulk typically yields particles with a relatively wide size distribution due to a distribution of nucleation and growth events in time and space. The polydispersity of the product is often further amplified by uncontrolled agglomeration or breakage of the crystals.1 By confining the crystallization process into a droplet, the crystal size can be a priori limited by the amount of dissolved material in each droplet, which makes it possible to achieve a narrower size distribution of the final crystals.2 Droplets can be formed in bulk by high intensity agitation of a two-phase liquid−liquid system.3 However, to achieve a narrower size distribution of the droplets, it is advantageous to use methods such as membrane emulsification,4 droplet formation in fluidic junctions (Tjunction, flow-focusing junction),5,6 or seeded crystallization.7 Spatial confinement of crystallization can be achieved not only by liquid−liquid systems, but also by air segmented fluid flow8,9 or by laser-induced crystallization.10 A special case where crystallization occurs within droplets is the so-called oiling out, where a single-phase liquid system phase-separates into a twophase one prior to crystallization.11−13 The spatial confinement of the crystallization process into a droplet should ideally allow direct control over the crystal size and morphology. In the ideal case, there would be a one-to-one correspondence between each droplet and the formed crystal.14 The driving force for crystal nucleation and growth is supersaturation of the solution, which is most commonly achieved by cooling, by solvent evaporation, by the addition of © 2017 American Chemical Society

Received: March 1, 2017 Revised: May 1, 2017 Published: May 1, 2017 3700

DOI: 10.1021/acs.cgd.7b00303 Cryst. Growth Des. 2017, 17, 3700−3706

Crystal Growth & Design

Article

Figure 1. Principle of extractive crystallization. The number of nucleation events in a single droplet depends on the initial solute concentration and the rate of extraction.

principle of this process, which shares some similarities with the recently reported extractive gelation,23 is shown schematically in Figure 1. For the first time, the present work implements extractive crystallization in a droplet-based continuous fluidic system. It makes use of the formation of monodisperse droplets in a Tjunction to precisely control the crystal size. Using the system KCl/water/1-hexanol as a model, we demonstrate that by systematically varying the initial solute concentration in the aqueous phase and the volumetric ratio between the aqueous and the organic phase, it is possible to control not only the size of the resulting crystals in a defined range but also their external morphology and internal structure. Most notably, we show that under a certain combination of parameters, it is possible to form spherical aggregates of cubic crystals with a percolating internal pore network, which possess superior dissolution properties over single cubic crystals of identical mass.

2. MATERIALS AND METHODS 2.1. Materials. 1-Hexanol, potassium chloride (p.a.), and poly(vinyl alcohol) were purchased from Sigma-Aldrich. Acetone p.a. was purchased from Penta, Czech Republic. Food-grade surfactant Abil EM 90 was obtained from Degussa. All chemicals were used as received. Deionized water (Aqual 25, 0.07 μS/cm) was used for all reactions and treatment processes. 2.2. Experimental Setup. The water-in-oil (w/o) emulsion was created using a T-junction (internal diameter 500 μm) made from PEEK (IDEX Health and Science, USA) connected to PTFE capillaries (Figure 2). The inner diameter of the capillary containing the emulsion was 800 μm, and the length was 100 cm (90 cm was thermostated). The flow rates of both phases were controlled by linear syringe pumps (neMESYS Low Pressure Syringe Module, Cetoni, Germany). The continuous phase was anhydrous 1-hexanol, which can absorb water up to 6.92% (w/w) at 20 °C, while the disperse phase was aqueous solution of KCl in which 1-hexanol is only sparingly soluble (0.70% w/w). The ability to absorb small amounts of water is common to all long-chain primary alcohols, although the water/alcohol equilibrium is different for each one.24 Therefore, the methodology presented in this work can be applied to other solvent systems than 1hexanol. The continuous phase contained 1.0% (w/w) of Abil EM 90, which is suitable for stabilizing w/o emulsions. The surfactant is used not only to stabilize the aqueous emulsion droplets but also helps to produce a more stable flow regime, therefore producing droplets with a narrower size distribution. The aqueous phase was prepared by diluting a stock solution of KCl by water (360.0 g/L) to the required concentration. The size of the droplets formed in the T-junction at

Figure 2. (a) Experimental setup used for the fabrication of w/o emulsion during extractive crystallization. (b) Droplets generated by the T-junction. The aqueous phase was dyed by methylene blue to enhance the contrast for illustration purposes.

different flow rates was calculated from the flow rate and the droplet formation time using the following equation:

dD =

⎛ 6Ft ⎞1/3 ⎜ ⎟ ⎝ π ⎠

(1)

where dD is the droplet diameter, F is the volumetric flow rate of the aqueous phase, and t is the period of droplet formation. The droplet formation period was calculated as an average of at least 10 droplets. 2.3. Parametric Analysis of Extractive Crystallization. The droplet size depends on the diameter of the T-junction (500 μm) and the flow rates of each phase. In order to control the final crystal size obtained after water extraction, three different concentrations of the feed KCl solution were used, namely, 90.00, 11.25, and 1.40 g/L. The volumetric flow rate of the continuous phase (1-hexanol) was fixed at 0.5 μL/s, and the flow rate of the disperse (aqueous) phase was varied from 0.005 to 0.012 μL/s, covering a range of w/o phase ratio Rv between 0.010 and 0.024. The Rv ratio is defined by the following equation: 3701

DOI: 10.1021/acs.cgd.7b00303 Cryst. Growth Des. 2017, 17, 3700−3706

Crystal Growth & Design Rv =

Fd Fc

Article

(2)

where Fd is the volumetric flow rate of the dispersed phase and Fc is the volumetric flow rate of the continuous phase. The upper limit on the Rv ratio is given by the solubility of water in 1-hexanol (to guarantee sufficient capacity for a complete extraction of the aqueous phase), while the lower limit is restricted by the requirement on a reasonable production rate of the crystals (i.e., to avoid unnecessarily long gaps between individual droplets). Note that at the highest combined flow rate of both phases, which was 0.512 μL/s, the linear velocity in the outlet capillary was 1.02 mm/s, giving a residence time of approximately 980 s. This parameter can of course be simply regulated by choosing a capillary of different length. 2.4. Particle Characterization. The fabricated crystals were isolated by sedimentation, washed three times with acetone to remove any oil phase or surfactant residues, and then dried at 70 °C for 10 min in a vacuum oven. The size distribution and morphology of the produced particles was characterized by optical microscopy (Olympus BX41) and scanning electron microscopy (SEM JEOL JCM-5700). The circular equivalent particle diameter was calculated from the crystal area, evaluated by the image analysis software ImageJ. The internal pore structure of the particles was analyzed by X-ray microtomography (Skyscan micro-CT attachment for SEM). The dissolution rate of the obtained particles was measured by conductometry (Mettler Toledo SG23). Due to the small size and high solubility of the KCl crystals, dissolution in pure water was almost instant; hence, a mixture of water and acetone (1:4 v/v) was used as the dissolution medium to artificially slow down the process and discriminate between different particle morphologies, which was the main goal of the dissolution experiments. The dissolution rate constants of the obtained particles were measured and compared for a series of size fractions. For reference, the dissolution rates of three corresponding sieve fractions of commercially sourced KCl crystals (Sigma-Aldrich) were also included. These fractions with size ranges of 80−125, 180−200, and 200−300 μm were obtained by sieving (Retsch AS 200 basic sieve shaker) for 5 min at 100 Hz. The dissolution rate was measured at 20 °C in a stirred beaker (500 rpm) containing 100 mL of the water−acetone mixture and using approximately a 1 mg sample of the KCl particles (the inner diameter of the beaker was 57 mm; the impeller dimensions were 25 mm × 6 mm). The dissolution curves were normalized by the asymptotic conductivity at complete dissolution and regressed by a first-order dissolution kinetic model

y = 1 − exp(− kt )

Figure 3. Particle size distribution of produced KCl crystals at Rv = 0.024 (red, c = 1.40 g/L; green, c = 11.25 g/L; blue, c = 90.00 g/L). Each distribution was obtained by digital image analysis from at least 50 individual crystals.

Figure 5 compares the experimentally measured particle size with theoretical one obtained by considering the initial droplet diameter, the feed KCl concentration, and the single droplet mass balance: ⎛ ⎞1/3 c KCl ⎟⎟ d p = dD⎜⎜ ⎝ (1 − ε)ρKCl ⎠

(4)

where dD and dP is the droplet and the theoretical particle diameter, respectively, cKCl is the KCl concentration (w/v) in the feed, ρKCl is the true material density of KCl, and ε is the porosity of the particles. The theoretical curve shown in Figure 5 is based on the assumption of nonporous crystalline particles (ε = 0). From the plot, it can be seen that, with increasing KCl concentration, eq 4 systematically underestimates the actual particle size. As will be shown below, this is due to a porous internal structure of the produced KCl particles. The value of porosity calculated from eq 4 based on the experimental particle size is given by each data point in Figure 5. 3.2. Effect of Process Parameters on Particle Structure. The extractive crystallization process can yield distinct particle morphologies depending on the nucleation and growth events that take place within the droplet (Figure 1). The final morphologies can vary from single cubic crystals, shown in Figure 6a, to spherical porous multicrystal aggregates, such as the one shown in Figure 6b. An interesting feature of the spherical multicrystals is their percolating internal pore network, apparent in X-ray microCT images of the particles (a static X-ray micrograph is shown in Figure 7; a full 3D X-ray microCT scan can be found in Supporting Information 1). Such percolating pore network and the associated increased specific surface area could be beneficial for the incorporation of a second material into the particles or for dissolution rate enhancement. As illustrated in Figure 1, the final particle morphology depends on the number of nucleation events that occur in the droplet. Both nucleation and crystal growth rates depend on the supersaturation, but it is well-known that the power-law exponents of nucleation are generally higher than those for

(3)

where k is the dissolution rate constant, t is time, and y is normalized conductivity of the solution.

3. RESULTS AND DISCUSSION 3.1. Effect of Process Parameters on Particle Size. The size distributions of the KCl particles fabricated by extractive crystallization are summarized in Figure 3. The equivalent particle diameter was evaluated by image analysis of at least 50 crystals for each set of experimental conditions. As can be seen from Figure 3, the size distribution of the particles is very narrow, which follows from the nearly monodisperse character of the droplets that are generally produced in a capillary Tjunction (Figure 2). Examples of crystals formed in the 500 μm T-junction for a fixed phase ratio Rv = 0.024 and varying initial KCl concentration are shown in Figure 4. For a fixed phase ratio Rv, which defines the droplet diameter (463 ± 9 μm in the case Rv = 0.024), the size of the crystals depends on the initial concentration of KCl in the feed and can therefore be finetuned. In the case of crystals shown in Figure 4, the mean particle diameters are 48, 115, and 234 μm for feed concentrations of KCl in the aqueous phase 1.40, 11.25, and 90.00 g/L, respectively. 3702

DOI: 10.1021/acs.cgd.7b00303 Cryst. Growth Des. 2017, 17, 3700−3706

Crystal Growth & Design

Article

Figure 4. Examples of crystals produced at constant Rv = 0.024 and varying initial KCl concentration: (a) c = 1.40 g/L; (b) c = 11.25 g/L; and (c) c = 90.00 g/L.

Figure 5. Comparison of the particle size of crystals produced at Rv = 0.024 with theoretical values calculated from the mass balance (eq 4) under the assumption of zero porosity.

Figure 7. X-ray micro-CT projection image of spherical aggregates. A full 3D micro-CT scan of the particle can be found in Supporting Information 1.

crystal growth.1 Therefore, nucleation dominates over crystal growth at high supersaturations, whereas crystal growth dominates over nucleation at lower supersaturations. The faster the water extraction from the droplet, the more likely multiple nucleation events occur. However, if the water extraction rate is slow, the first nucleus can grow into a single cubic crystal before other nucleation events may occur.

The extraction rate of water from the droplets depends on the droplet size (which defines the surface area available for mass transfer), the concentration of KCl, and the w/o phase ratio (Rv), which jointly determine the driving force for mass transfer. A systematic comparison of the effect of these parameters on the crystal morphology is shown in Table 1. In general, higher w/o phase ratio Rv favors the formation of single cubic crystals, whereas spherical aggregates tend to form

Figure 6. SEM micrographs showing examples of particle morphologies that can be obtained by extractive crystallization: (a) single cubic crystals and (b) spherical aggregates. 3703

DOI: 10.1021/acs.cgd.7b00303 Cryst. Growth Des. 2017, 17, 3700−3706

Crystal Growth & Design

Article

Table 1. Comparison of Particle Morphologies Obtained for Different Combinations of the Feed KCl Concentration and the w/o Phase Ratio Rv As Indicated

at lower Rv. For a given Rv, a higher initial solute concentration tends to favor the formation of particles consisting of a larger number of subcrystals. This can be explained by considering that for high initial KCl concentration, supersaturation is reached sooner (while the oil phase is not yet saturated by the aqueous phase, so the extraction rate is still high). Nucleation is therefore favored over crystal growth. Decreasing the feed concentration means that the droplets have to shrink more in order to become saturated. At the point when sufficient supersaturation is reached for nucleation to take place, the driving force for further extraction is already reduced due to the partial saturation of the continuous phase by water. 3.3. Effect of Particle Structure on Dissolution Behavior. For most industrially produced crystalline substances, the crystalline state is only an intermediate form, and their ultimate fate is to be dissolved at the point of use (e.g., in the gastro-intestinal tract for food additives and pharmaceuticals, in a chemical reactor for reagents, etc.). Hence, the ability to influence the dissolution behavior not only by the particle size but also by the particle structure is potentially interesting from the application point of view. The dissolution profiles of KCl crystals of identical mass but different structure (single cubic crystals vs porous spherical aggregates) are compared in Figure 8 (the crystallization conditions are specified in the figure caption). The dissolution rate constants evaluated from regression by eq 3 were found to be 0.0704 and 0.0262 s−1 for the spherical aggregates and the single cubic crystals,

Figure 8. Dissolution curves of single cubic crystals (Rv = 0.024; c = 90 g/L; dp = 254 μm) and spherical aggregates (Rv = 0.01; c = 90 g/L; dp = 316 μm) of identical mass. The curves show regression by eq 3.

respectively, meaning that the dissolution rate of the spherical aggregates is nearly 3× faster than that of single cubic crystals of comparable size in this particular case. The dissolution rate enhancement can be attributed to the higher surface area of the porous−spherical aggregates, which is decisive in the early stages of the dissolution process, and possibly also to eventual 3704

DOI: 10.1021/acs.cgd.7b00303 Cryst. Growth Des. 2017, 17, 3700−3706

Crystal Growth & Design

Article

possible to produce larger quantities of the particles without a loss of quality.

breakup of the aggregates, which may become significant in the later stages of the dissolution process as the necks connecting the individual subcrystals dissolve. A systematic view at the dissolution properties of the two types of crystal morphologies is provided in Figure 9 where the



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.cgd.7b00303. Three-dimensional visualization of the porous spherical aggregate depicted in Figure 7, obtained by X-ray microtomography (AVI)



AUTHOR INFORMATION

Corresponding Author

*Tel: +420 220 443 236. E-mail: [email protected]. ORCID

František Štěpánek: 0000-0001-9288-4568 Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS F.S. would like to acknowledge support by the Grant Agency of the Czech Republic (grant no. GACR 16-12291S).

Figure 9. Dependence of the dissolution rate constant on particle size for three different types of KCl particles: sieved KCl crystals sourced commercially, KCl crystals produced by extractive crystallization under conditions favoring spherical aggregates (Rv = 0.01) and single cubic crystals (Rv = 0.024).

REFERENCES

(1) Holaň, J.; Ridvan, L.; Billot, P.; Štěpánek, F. Chem. Eng. Sci. 2015, 128, 36−43. (2) Dombrowski, R. D.; Litster, J. D.; Wagner, N. J.; He, Y. AIChE J. 2010, 56 (1), 79−91. (3) Kovačík, P.; Kremlácǩ ová, Z.; Š těpánek, F. Microporous Mesoporous Mater. 2012, 159, 119−125. (4) Spyropoulos, F.; Lloyd, D. M.; Hancocks, R. D.; Pawlik, A. K. J. Sci. Food Agric. 2014, 94 (4), 613−627. (5) Leng, J.; Salmon, J. B. Lab Chip 2009, 9 (1), 24−34. (6) Bhamidi, V.; Lee, S. H.; He, G.; Chow, P. S.; Tan, R. B. H.; Zukoski, C. F.; Kenis, P. J. A. Cryst. Growth Des. 2015, 15 (7), 3299− 3306. (7) Briggs, N. E. B.; Schacht, U.; Raval, V.; McGlone, T.; Sefcik, J.; Florence, A. J. Org. Process Res. Dev. 2015, 19 (12), 1903−1911. (8) Lu, J.; Litster, J. D.; Nagy, Z. K. Cryst. Growth Des. 2015, 15 (8), 3645−3651. (9) Robertson, K.; Flandrin, P.-B.; Klapwijk, A. R.; Wilson, C. C. Cryst. Growth Des. 2016, 16 (8), 4759−4764. (10) Li, W.; Ikni, A.; Scouflaire, P.; Shi, X.; El Hassan, N.; Gémeiner, P.; Gillet, J.-M.; Spasojević-de Biré, A. Cryst. Growth Des. 2016, 16 (5), 2514−2526. (11) Codan, L.; Bäbler, M. U.; Mazzotti, M. Cryst. Growth Des. 2010, 10 (9), 4005−4013. (12) Lu, J.; Li, Y.-P.; Wang, J.; Li, Z.; Rohani, S.; Ching, C.-B. Org. Process Res. Dev. 2012, 16 (3), 442−446. (13) Takasuga, M.; Ooshima, H. Cryst. Growth Des. 2015, 15 (12), 5834−5838. (14) Yamaguchi, H.; Maeki, M.; Yamashita, K.; Nakamura, H.; Miyazaki, M.; Maeda, H. J. Biochem. 2013, 153 (4), 339−346. (15) Holaň, J.; Skořepová, E.; Heraud, L.; Baltes, D.; Rohlíček, J.; Dammer, O.; Ridvan, L.; Štěpánek, F. Org. Process Res. Dev. 2016, 20 (1), 33−43. (16) Teychené, S.; Biscans, B. Chem. Eng. Sci. 2012, 77, 242−248. (17) Quilaqueo, M.; Aguilera, J. M. Food Res. Int. 2016, 84, 143−149. (18) Kitamura, M.; Sugimoto, M. J. Cryst. Growth 2003, 257 (1−2), 177−184. (19) Park, M.-W.; Yeo, S.-D. Chem. Eng. Res. Des. 2012, 90 (12), 2202−2208. (20) Rajagopal, S.; Ng, K. M.; Douglas, J. M. AIChE J. 1991, 37 (3), 437−447.

dissolution rate constants are compared for a series of size fractions of particles obtained by extractive crystallization and corresponding sieve fractions of commercially sourced KCl crystals. This makes it possible to assess the relative contribution of particle size and particle microstructure on the dissolution behavior. From the scaling of the dissolution rate with particle size, it is obvious that the dissolution rate enhancement achieved by forming porous spherical aggregates would be difficult to match by the classical bulk crystals even if very small sieve fractions were used. Interestingly, also the single cubic crystals produced by extractive crystallization dissolve faster than sieved crystals sourced commercially, probably owing to their internal porosity (see Figure 4 and Table 1) and the correspondingly higher surface area.

4. CONCLUSION A new method for the continuous production of crystalline particles with a narrow size distribution by extractive crystallization using a fluidic T-junction to generate monodisperse water-in-oil emulsion droplets has been developed. This method makes it possible to fabricate crystals of precisely defined size by tuning the initial droplet size and solute concentration. Depending on the process conditions, particles of diverse morphology can be prepared. These range from single cubic crystals, through cuboid or pyramidal crystals composed of a few subunits, to highly porous spherical aggregates with a percolating pore network. These crystals possess not only a narrow size distribution but also unique properties in terms of dissolution rate enhancement due to increased surface area and internal porosity. These properties could provide product benefits such as increased bioavailability of poorly soluble pharmaceutical compounds, improved redispersion properties of “instant” products, or reduced salt or sugar loads in food products without the loss of taste onset. The scalability of this method by parallelization25 makes is 3705

DOI: 10.1021/acs.cgd.7b00303 Cryst. Growth Des. 2017, 17, 3700−3706

Crystal Growth & Design

Article

(21) Weingaertner, D. A.; Lynn, S.; Hanson, D. N. Ind. Eng. Chem. Res. 1991, 30 (3), 490−501. (22) Gomis, V.; Ruiz, F.; Boluda, N.; Saquete, M. D. J. Chem. Eng. Data 1999, 44 (5), 918−920. (23) Pittermannova, A.; Ruberova, Z.; Zadrazil, A.; Bremond, N.; Bibette, J.; Stepanek, F. RSC Adv. 2016, 6 (105), 103532−103540. (24) Góral, M.; Wiśniewska-Gocłowska, B.; Mączyński, A. J. Phys. Chem. Ref. Data 2006, 35 (3), 1391−1414. (25) Riche, C. T.; Roberts, E. J.; Gupta, M.; Brutchey, R. L.; Malmstadt, N. Nat. Commun. 2016, 7, 10780.

3706

DOI: 10.1021/acs.cgd.7b00303 Cryst. Growth Des. 2017, 17, 3700−3706