PREVENTING CRYSTAL AGGLOMERATION OF PHARMACEUTICAL CRYSTALS

pharmaceutics Article

Preventing Crystal Agglomeration of Pharmaceutical Crystals Using Temperature Cycling and a Novel Membrane Crystallization Procedure for Seed Crystal Generation Elena Simone 1,2 1 2 3 4

*

ID

, Rahimah Othman 2,3 , Goran T. Vladisavljević 2

ID

and Zoltan K. Nagy 2,4, *

School of Food Science and Nutrition, University of Leeds, Leeds LS2 9JT, UK; [email protected] Department of Chemical Engineering, Loughborough University, Loughborough LE11 3TU, Leicestershire, UK; [email protected] (R.O.); [email protected] (G.T.V.) School of Bioprocess Engineering, Universiti Malaysia Perlis, Kompleks Pusat Pengajian Jejawi 3, Arau 02600, Perlis, Malaysia School of Chemical Engineering, Purdue University, West Lafayette, IN 47907-2100, USA Correspondence: [email protected]; Tel.: +1-765-494-0734

Received: 4 December 2017; Accepted: 18 January 2018; Published: 24 January 2018

Abstract: In this work, a novel membrane crystallization system was used to crystallize micro-sized seeds of piroxicam monohydrate by reverse antisolvent addition. Membrane crystallization seeds were compared with seeds produced by conventional antisolvent addition and polymorphic transformation of a fine powdered sample of piroxicam form I in water. The membrane crystallization process allowed for a consistent production of pure monohydrate crystals with narrow size distribution and without significant agglomeration. The seeds were grown in 350 g of 20:80 w/w acetone-water mixture. Different seeding loads were tested and temperature cycling was applied in order to avoid agglomeration of the growing crystals during the process. Focused beam reflectance measurement (FBRM); and particle vision and measurement (PVM) were used to monitor crystal growth; nucleation and agglomeration during the seeded experiments. Furthermore; Raman spectroscopy was used to monitor solute concentration and estimate the overall yield of the process. Membrane crystallization was proved to be the most convenient and consistent method to produce seeds of highly agglomerating compounds; which can be grown via cooling crystallization and temperature cycling. Keywords: agglomeration; membrane crystallization; temperature cycling; seeded crystallization

1. Introduction Crystal agglomeration is a common phenomenon for many chemical and pharmaceutical compounds. It is usually undesirable since the formation of agglomerates promotes the entrapment of mother liquid and, therefore, compromises the purity of the dried product. Furthermore, particle agglomeration can generate broad crystal size distributions (CSD) and induce the formation of a large amount of fines during storage and transport [1,2]. The effect of operating parameters on crystal agglomeration during crystallization processes has been studied via both experimental and modelling work [2,3]. The main variables affecting agglomeration were found to be solid content in suspension, particle size, stirring rate, and supersaturation at seeding or nucleation. Furthermore, both solvent and solute physical and chemical characteristics were found to strongly affect this phenomenon [1,4,5]. Recently, the effect of particle morphology on crystal agglomeration was studied using a novel image analysis routine [6]. An opportune choice of solvent, careful seeding and temperature cycling can largely decrease the Pharmaceutics 2018, 10, 17; doi:10.3390/pharmaceutics10010017

www.mdpi.com/journal/pharmaceutics

Pharmaceutics 2018, 10, 17

2 of 15

degree of agglomeration. In fact, repeated cycles of heating and cooling were found to promote de-agglomeration [7]. In some cases, the formation of regular agglomerates (mostly spherical) is desired, since they are easy to handle during downstream operations and can be more efficiently compacted [8]. Spherical agglomerates can be formed using a bridging liquid [9,10], a specific additive [11] or by cooling an oil-in-water emulsion where the crystallizing solution is localized within oil droplets [12]. Spherical agglomerates have recently been produced successfully in a continuous MSMPR (mixed suspension, mixed product removal) reactor [12] and in a microfluidic device [13]. Large agglomerates of poorly water soluble drugs can also be obtained in the form of crystanules, particles with mixed characteristics of both crystals and granules. These are produced by melting the insoluble drug and forming an emulsion consisted of drug droplets in water. The droplets are then solidified, by cooling, in the form of large agglomerates. The addition of a polymeric additive can improve the mechanical properties of these crystanules [14–16]. Membrane crystallization is a relatively new technique based on the use of a porous material as a semi-permeable barrier between two phases. The membrane can be used to create supersaturation by solvent evaporation, antisolvent or reactant addition, and mixing with a colder solvent [17,18]. The first membrane crystallization process dates back to 1917 when Kober used a polymeric membrane to evaporate water from aqueous solutions of ammonium sulfate and hydrochloric acid, and precipitated salts upon increasing the solution supersaturation [18]. Renewed interest in membrane crystallization aroused in the 80’s when microporous membranes for water treatment became popular [19]. Membrane distillation and reverse osmosis processes were proposed and applied [20,21]. More recently, membranes were used to crystallize proteins and macromolecules [22–24]. The presence of a membrane adds a supplementary resistance to mass transfer, but it also offers additional control over the nucleation kinetics (the nature of the membrane can significantly change the surface tension of the crystallized material) and polymorphic outcome of crystallization [25–27]. In this work, a flat isoporous nickel membrane installed in a stirred cell was used to produce micro-seeds of piroxicam monohydrate by reverse antisolvent addition. Piroxicam is a non-steroidal anti-inflammatory drug; it can crystallize in two different polymorphs or as monohydrate in the presence of water. This last form is characterized by high tendency of agglomeration. The membrane crystallization procedure used in this work allows producing non-agglomerated crystals of piroxicam monohydrate with narrow CSD, which can easily be filtered and dried [28,29], or used as seeds while still in slurry. The use of the membrane guarantees a higher polymorphic purity, lower tendency to agglomeration, and a narrower crystal size distribution compared to traditional batch crystallization techniques (cooling, antisolvent or reverse antisolvent addition) [28,29]. Seed crystals produced by membrane crystallization were compared to those prepared using traditional batch techniques. Furthermore, a specific cooling profile with temperature cycling was determined and applied in order to reduce agglomeration and promote crystal growth of the seed crystals. The experimental results show that a combination of specifically tailored membrane crystallization seeds and temperature cycling can efficiently reduce agglomeration and promote crystal growth of piroxicam monohydrate. 2. Materials and Methods Piroxicam (99% purity) was purchased from Hangzhou Hyper Chemicals Limited (Hangzhou, China). Acetone (99.98% purity) purchased from Fisher Scientific (Loughborough, UK) and de-ionized water (Millipore ultra-pure system) were used as solvent and antisolvent for the drug. A 400 mL jacketed glass vessel was used to carry out the 350 g cooling crystallization experiments. The vessel was equipped with an overhead polytetrafluoroethylene (PTFE) pitch blade stirrer (325 rpm was the stirring speed used in all experiments). A schematic of the rig used for the experiments is shown in Figure 1.

Pharmaceutics 2018, 10, 17 Pharmaceutics 2018, 10, x FOR PEER REVIEW

3 of 15 3 of 15

Figure Schematic of of the the rig Figure 1. 1. Schematic rig used used for for the the experiments. experiments.

temperature probe connected to a Huber Huber Ministat Ministat 125 thermoregulator thermoregulator (Huber A PT-100 temperature Kältemaschinenbau AG, Offenburg, Germany) was used to control the internal temperature of the Kältemaschinenbau Germany) vessel as well as the temperature in the jacket. Several Several process process analytical analytical technology technology (PAT) (PAT) tools were monitor the the experiments, experiments, including: (1) a Kaiser RXN1 Raman analyzer with immersion immersion used to monitor probe and 785 nm laser, laser, equipped equipped with iC Raman Raman 4.1 software, (2) a Mettler-Toledo particle particle vision and measurement (PVM) V819 probe with an on-line image acquisition software (version 8.3), and (3) a G400 Mettler-Toledo Mettler-Toledo Focused Focused Beam Beam Reflectance Reflectance Measurement (FBRM) probe equipped with iC FBRM version 4.3 software. Data was processed using using MatLab MatLab R2015, R2015, iC iC Raman Raman 4.1 4.1 and and Excel Excel 2013. 2013. Additionally,the the CryPRINS CryPRINS software software (Crystallization (Crystallization Process Process Informatics Informatics System, System, version version 2.0), which Additionally, allows real time temperature control and simultaneous monitoring of signals from different probes (FBRM, ATR-UV/Vis, thermocouple, conductivity probes and pH-meter), was used. 2.1. Seeded Batch Crystallization Experiments The choice of solvent for the the seeded seeded cooling cooling crystallization crystallization experiments was made based on available solubility data and past literature on piroxicam crystallization [30,31]. In order to nucleate and grow crystals of piroxicam monohydrate, a mixture of water and an organic solvent needs to be used. The addition addition of an organic solvent is necessary to avoid an excessive reduction of piroxicam solubility in the final solvent mixture, since this compound is almost insoluble in pure water. Acetone was selected as as the the organic organicsolvent solventfor forthis thiswork, work,and anda a20:80 20:80 w/w water-acetone mixture was used w/w water-acetone mixture was used as as solvent in the experiments presented In such solvent mixture, the piroxicam solubility solvent in the experiments presented herehere [30].[30]. In such solvent mixture, the piroxicam solubility still still increase considerably as temperature rises, which a necessarycondition conditiontotoincrease increase the the yield of increase considerably as temperature rises, which is is a necessary cooling crystallization processes. processes. ◦ C) Solutions of piroxicam solvent (saturation (saturation temperature of 40 °C) piroxicam with with concentration concentration of of 126 126 mg/g mg/g solvent were prepared C for 30 min in order to fully dissolve the solid piroxicam. After prepared and and heated heated up up to to 50 50 ◦°C that, solutions were cooled down to 37 ◦°C C and seed crystals were added. In order to reduce crystal agglomeration added toto thethe solution was setset to agglomeration immediately immediatelyafter afterseeding, seeding,the theamount amountofofseed seedcrystals crystals added solution was 2% of of thethe total mass ofof piroxicam dissolved. to 2% total mass piroxicam dissolved.An Anadditional additionalexperiment experimentusing using6% 6% seed seed crystals crystals was performed comparison, usingusing crystals made bymade membrane crystallization. Piroxicam monohydrate performed forfor comparison, crystals by membrane crystallization. Piroxicam was found to have slow of nucleation and of growth; for this reason a specific temperature monohydrate was very found to kinetics have very slow kinetics nucleation and growth; for this reason a profile designed for the seeded experiments to maximize the recovery of piroxicam specificwas temperature profile was designed for in theorder seeded experiments in orderyield to maximize the by allowing, at the same time,by growth and secondary nucleation. Afterand seeding, the solution was slowly recovery yield of piroxicam allowing, at the same time, growth secondary nucleation. After ◦ C/min) cooled to 10 ◦ Cwas (−0.1 and thentoleft 10 left h. During such period seeding,down the solution slowly cooled down 10 at °Cthat (−0.1temperature °C/min) andfor then at that temperature of solution was completely depleted bywas growth, secondary nucleation and, fortime 10 h.the During suchsupersaturation period of time the solution supersaturation completely depleted by growth, secondary nucleation and, partly, by agglomeration. After the isothermal period, in order to remove

the fine crystals generated by secondary nucleation and reduce agglomeration, temperature cycling was applied. Slurries underwent nine temperature cycles of 20 °C amplitude with heating/cooling rate of ±0.2 °C/min. The cycles’ amplitude and rates were chosen, based on preliminary experiments (see Supplementary Materials Figures S1 and S2), in order to accelerate the dissolution of fines and4 dePharmaceutics 2018, 10, 17 of 15 agglomeration [29]. In addition to the use of fixed amplitude cycles, direct nucleation control (DNC) was also tested. partly, by agglomeration. After theare isothermal in order to remove the fine crystals by The details of this control strategy describedperiod, elsewhere [7,32–35]. Experimental resultsgenerated for the two secondary nucleation and reduce agglomeration, temperature cycling was applied. DNC experiments performed are shown in the Supplementary Materials Figures S8–S11). Slurries underwent nine temperature cycles of 20 ◦ C amplitude with heating/cooling rate of ◦ C/min. 2.2. Seeds Preparation ±0.2 The cycles’ amplitude and rates were chosen, based on preliminary experiments (see Supplementary Materials Figures S1 and S2), in order to accelerate the dissolution of fines and Seed crystals were prepared using membrane crystallization, antisolvent crystallization, and de-agglomeration [29]. polymorphic transformation, as described in the following sections. Cooling crystallization with in In addition to the use of fixed amplitude cycles, direct nucleation control (DNC) was also tested. situ nucleation was also performed but seed crystals could not be produced in a reasonable batch The details of this control strategy are described elsewhere [7,32–35]. Experimental results for the two time because of the very slow kinetics of primary nucleation for piroxicam monohydrate. DNC experiments performed are shown in the Supplementary Materials Figures S8–S11). 2.2.1. Membrane Crystallization 2.2. Seeds Preparation Piroxicam microcrystals were prepared by reverse antisolvent addition using a flat, disc-shaped Seed crystals were prepared using membrane crystallization, antisolvent crystallization, and membrane installed in a stirred cell, as shown in Figure 2a. The cell and membranes were purchased polymorphic transformation, as described in the following sections. Cooling crystallization with in from Micropore Technologies Ltd. (Derby, UK). This apparatus is normally used for membrane situ nucleation was also performed but seed crystals could not be produced in a reasonable batch time emulsification, a gentle process for the formation of droplets and particles in the micron range [36– because of the very slow kinetics of primary nucleation for piroxicam monohydrate. 38]. The rotation speed of the stirrer was fixed to 1500 rpm, corresponding to a peak shear stress at 2.2.1. Membrane Crystallization the membrane surface of 17.5 Pa. A 24 V direct current (DC) motor (Instek model PR-3060, New Piroxicam microcrystals were prepared by reverse antisolvent addition using a flat, disc-shaped Taipei City, Taiwan) was used to drive the stirrer. The nickel membrane used for the experiments membrane installed in a stirred cell, as shown in Figure 2a. The cell and membranes were purchased had an operative area of 8.55 cm2 and an effective diameter of 3.3 cm. It presented ≈24,690 pores with from Micropore Technologies Ltd. (Derby, UK). This apparatus is normally used for membrane diameter of 40 μm, arranged hexagonally and spaced apart at a constant distance of 200 μm, as shown emulsification, a gentle process for the formation of droplets and particles in the micron range [36–38]. in the scanning electron microscope (SEM, Hitachi Ltd., Tokyo, Japan) image of Figure 2b.

(a)

(b)

Figure Figure2.2.(a) (a)Schematic Schematicofofthe therig rigused usedfor formembrane membranecrystallization; crystallization;(b) (b)Scanning Scanningelectron electronmicroscope microscope (SEM) image of the Ni membrane used for the experiments. (SEM) image of the Ni membrane used for the experiments.

The membrane was placed at the bottom of the cell which was filled with 30 mL of deionized The rotation speed of the stirrer was fixed to 1500 rpm, corresponding to a peak shear stress water. A solution containing 15 g·L−1 of piroxicam in acetone was injected through the membrane at the membrane surface of 17.5 Pa. A 24 V direct current (DC) motor (Instek model PR-3060, New using a syringe pump (Harvard Apparatus model 11 Elite, Harvard Apparatus, Holliston, United Taipei City, Taiwan) was used to drive the stirrer. The nickel membrane used for the experiments had States). The feed flow rate2 through the membrane was 18 mL·min−1, corresponding to a an operative area of 8.55 cm and an effective diameter of 3.3 cm. It presented ≈24,690 pores with transmembrane flux of 4000 L m−2·h−1). The total volume of the feed solution injected was 6 mL and diameter of 40 µm, arranged hexagonally and spaced apart at a constant distance of 200 µm, as shown the volume mean diameter of microcrystals was between 25–35 μm, as measured by Malvern in the scanning electron microscope (SEM, Hitachi Ltd., Tokyo, Japan) image of Figure 2b. The membrane was placed at the bottom of the cell which was filled with 30 mL of deionized water. A solution containing 15 g·L−1 of piroxicam in acetone was injected through the membrane using a syringe pump (Harvard Apparatus model 11 Elite, Harvard Apparatus, Holliston, United States). The feed flow rate through the membrane was 18 mL·min−1 , corresponding to a transmembrane


5 of 15

flux of 4000 L m−2 ·h−1 ). The total volume of the feed solution injected was 6 mL and the volume mean diameter of microcrystals was between 25–35 µm, as measured by Malvern Mastersizer 2000 (with a Hydro 2000SM dispersant unit, Malvern Instruments Ltd., Malvern, UK) [29]. In this reverse antisolvent crystallization procedure, the membrane acts as a physical support for the generation and sustenance of a controlled supersaturation for the nucleation and growth of the crystals. The feed is injected into the aqueous phase through multiple equally spaced injection microjets where individual crystals can nucleate and grow before being transferred to the bulk as the result of the shear stress created by the stirrer. This procedure allows the production of crystals with very narrow CSD, as well as low agglomeration tendency. Crystals smaller than 25 µm can be obtained with the same apparatus [28], but the 25–35 µm size range was chosen to minimize particle agglomeration immediately after seeding. After each experiment, the membrane was sonicated in acetone for 30 min, washed with deionized water in an ultrasonic bath for 5 min and then stored in acetone. The slurry containing the monohydrate microcrystals was transferred from the cell to a beaker where crystals were allowed to settle at the bottom. Crystals were then collected with a pipette and transferred directly to the vessel for the seeded batch experiments. A small amount of sample was used to measure the CSD using the Malvern Mastersizer 2000 previously described. A saturated solution of piroxicam monohydrate in water at room temperature was used as dispersant. 2.2.2. Antisolvent Crystallization Seed crystals from antisolvent crystallization were produced by pumping water with a peristaltic pump (Masterflex L/S digital drive, Masterflex Technical Hoses Limited, Oldham, UK) at a rate of 2.5 mL/min into a solution of piroxicam in acetone with concentration of 35 mg/g of solvent. The solution was prepared by dissolving solid piroxicam at 40 ◦ C in the 400 mL jacketed vessel. The rate of antisolvent was chosen after a few preliminary experiments in order to obtain pure monohydrate piroxicam at the end of the batch. In fact, nucleation of form II of piroxicam was observed using both faster and slower cooling rates. During the experiments, FBRM and PVM were used to monitor nucleation, agglomeration and shape of the particles. The Malvern Mastersizer 2000 previously described was used to measure the CSD of the crystals obtained at the end of the experiments (after filtration and drying). 2.2.3. Polymorphic Transformation Piroxicam form I (fine powder as purchased) was suspended in water and left to transform to monohydrate. A total amount of 2.2 g of piroxicam was suspended in 500 g of water at 20 ◦ C. Raman spectroscopy (ThermoFisher Scientific, Waltham, MA, USA) was used to monitor, in situ, the polymorphic transformation of form I into piroxicam monohydrate. After transformation, crystals were filtered and dried, and their CSD was measured by the Malvern Mastersizer previously described. 2.2.4. Cooling Crystallization Solutions of piroxicam in a 20:80 w/w water-acetone mixture were prepared in the 400 mL vessel, as described in the Equipment section. Solutions containing 6.8, 7, 8.8, 11.4, and 14 mg of piroxicam per 1 g of solvent were heated up until the drug was completely dissolved and then cooled down to 5 ◦ C at a cooling rate of 0.5 ◦ C/min. Nucleation of monohydrate crystals was not detected either during the cooling profile or after leaving the solution at 5 ◦ C for few hours, indicating very slow primary nucleation kinetics. In conclusion, piroxicam monohydrate seeds could not be produced in a reasonable batch time by cooling crystallization with the equipment available, because of the extremely slow primary nucleation kinetics.


6 of 15

Pharmaceutics 2018, 10, x FOR PEER REVIEW

6 of 15

3. Results and and Discussion Discussion 3. Results 3.1. Seeds Characterization and Comparison 3.1. Seeds Characterization and Comparison Seeds obtained by membrane crystallization are shown in Figure 3. Crystals displayed the Seeds obtained by membrane crystallization are shown in Figure 3. Crystals displayed the Pharmaceutics 2018,color 10, x FOR PEER REVIEW monohydrate (polymorphic purity was checked with 6 of 15 characteristic piroxicam characteristic yellow yellow color of of piroxicam monohydrate (polymorphic purity was checked with Raman Raman microscopy and differential scanning calorimetry) and the shape of a prism with a rhombus base. microscopy and differential scanning calorimetry) and the shape of a prism with a rhombus base. 3. Results and Discussion

3.1. Seeds Characterization and Comparison Seeds obtained by membrane crystallization are shown in Figure 3. Crystals displayed the characteristic yellow color of piroxicam monohydrate (polymorphic purity was checked with Raman microscopy and differential scanning calorimetry) and the shape of a prism with a rhombus base.

Figure 3. Monohydrate crystals produced by membrane crystallization.

These seeds did not show significant agglomeration and they were characterized by a volume These seeds did notFigure show significantcrystals agglomeration and they crystallization. were characterized by a volume 3. Monohydrate produced by membrane weighted mean diameter measured using the Mastersizer in a range of 25–35 μm. Figure 4d shows weighted mean diameter measured using the Mastersizer in a range of 25–35 µm. Figure 4d shows the the seeds obtained by antisolvent crystallization; crystals were highly agglomerated and it was These seeds did notcrystallization; show significant crystals agglomeration theyagglomerated were characterized a volume seeds obtained by antisolvent were and highly andby it was difficult to difficult weighted to identify their shape. Ramanusing microscopy confirmed that piroxicam mean diameter measured the Mastersizer in a range of crystals 25–35 m.were Figurepure 4d shows identify their shape. Raman microscopy confirmed that crystals were pure μpiroxicam monohydrate monohydrate after filtration and drying. However,crystals Figure 4c highly showsagglomerated a PVM image the seeds obtained by antisolvent crystallization; were and ittaken was after after filtration and drying. However, Figure 4c shows a PVM image taken after nucleation where difficult to identify their shape. Raman microscopy confirmed that crystals were pure piroxicam nucleation where several monohydrate agglomerates were observed together with needle-shaped several monohydrate agglomeratesand were observed together with needle-shaped crystals of after piroxicam monohydrate drying. However, Figure 4c shows a PVM taken crystals of piroxicam after formfiltration II. Most likely, a mixture of monohydrate and formimage II was nucleated and form II. Most likely, a mixture of monohydrate and form II was nucleated and then converted nucleation where several monohydrate agglomerates were observed together with needle-shaped to the then converted to the stable monohydrate during the batch. stable monohydrate duringform the II. batch. crystals of piroxicam Most likely, a mixture of monohydrate and form II was nucleated and The sudden drop inthe total counts/measurement recorded the FBRM (Figure 4a,b) was probably converted stable monohydrate during the batch. by Thethen sudden drop to in total counts/measurement recorded by the FBRM (Figure 4a,b) was probably due to the change in crystal shape associated to the polymorphic transformation (from The sudden drop in total counts/measurement recorded by the FBRM (Figure 4a,b) wasneedle-shaped probably due to the change in crystal shape associated to the polymorphic transformation (from needle-shaped due to thepiroxicam change in crystal shape associated the polymorphic transformationagglomerates (from needle-shaped form II to cubic monohydrate). Thetoformation of monohydrate might also form II to cubic piroxicam monohydrate). The formation of monohydrate agglomerates might also form II to cubic piroxicam monohydrate). The formation of monohydrate agglomerates might also have contributed to the decrease in the total counts/measurement. have contributed to the to decrease in the total counts/measurement. have contributed the decrease in the total counts/measurement.

(a)

(a)

(b)

Figure 4. Cont.

(b)

Pharmaceutics 2018, 2018, 10, 10, 17 x FOR PEER REVIEW Pharmaceutics

7 of 15


(c)

7 of 15

(d)

(c)

(d)

Figure 4. (a) Total counts and volume of water pumped in the vessel during antisolvent crystallization Figure 4. (a) Total counts and volume of water pumped in the the vessellength during antisolvent crystallization of(a) piroxicam (2.5 mL/min water flow); (b) Evolution distribution during the Figure 4. Total counts and volume of water pumped of in the chord vessel during antisolvent crystallization of piroxicam (2.5 mL/min water flow); (b) Evolution of theofchord length distribution during the experiment; (c) particle vision and measurement (PVM) image the crystals during nucleation and of piroxicam (2.5 mL/min water flow); (b) Evolution of the chord length distribution during the growth. crystals identified as(PVM) piroxicam formof II;the (d) crystals Crystals after filtration and experiment; (c)Needle-shape particle vision andwere measurement image during nucleation and experiment; (c) particle vision and measurement (PVM) image of the crystals during nucleation and Form II was not detected the end of the growth.drying. Needle-shape crystals wereatidentified as experiment. piroxicam form II; (d) Crystals after filtration and

growth. Needle-shape crystals were identified as piroxicam form II; (d) Crystals after filtration and drying. Form II was not detected at the end of the experiment. drying. Figure Form II5b wasshows not detected at the end ofcrystals the experiment. the monohydrate obtained at the end of the polymorphic transformation from form II to piroxicam monohydrate, in water. The transformation took almost

Figure 5b5bshows the monohydrate crystals obtained at the endatof theagglomerated. polymorphic two days and generated monohydrate crystals thatobtained were considerably Figure shows the fine monohydrate crystals end of thetransformation polymorphic from form II to piroxicam monohydrate, in water. The transformation took almost two days and transformation from form II to piroxicam monohydrate, in water. The transformation took almost generated finegenerated monohydrate that were considerably two days and fine crystals monohydrate crystals that wereagglomerated. considerably agglomerated.

(a)

(b)

Figure 5. (a) Raman signal during polymorphic transformation from form I piroxicam to monohydrate; (b) Microscopic image of the monohydrate crystals obtained at the end of the experiment.

(b) Figure 5a shows(a) the intensity of two different Raman peaks, after second derivative and smoothing calculation. The peak at 1405–1390 cm−1 was associated with solid piroxicam Figure 5. (a) Raman signal during polymorphic transformation from form I piroxicam to −1 was typicalfrom monohydrate, while theduring peak atpolymorphic 1550–1537 cm of form thetopolymorphic Figure 5. (a) Raman signal transformation formI.I During piroxicam monohydrate; monohydrate; (b) the Microscopic image of theI peak monohydrate crystals obtained at the end intensity of the form decreased, while the intensity for of thethe (b) transformation, Microscopic image of the monohydrate crystals obtained at the end of thepeak experiment. experiment. monohydrate increased. The polymorphic transition was also detectable qualitatively by a change of color of the slurry from white (form I) to bright yellow (monohydrate). The Malvern Mastersizer was Figure thethe intensity two different Raman peaks, after and smoothing used5a toshows determine the intensity crystalofsize of the Raman three types ofsecond seeds. Seeds produced by Figure 5a shows ofdistribution two different peaks, afterderivative second derivative and − 1 calculation. The peak at 1405–1390 cm was associated with solid piroxicam monohydrate, while the membrane crystallization were analyzed in form of slurry, while seeds from antisolvent addition and −1 smoothing calculation. The peak at 1405–1390 cm was associated with solid piroxicam −1 was typical polymorphiccm transformation were filtered and dried before being dispersed in a saturated the aqueous peak at 1550–1537 of form I. During the polymorphic transformation, intensity of −1 monohydrate, while the peak at 1550–1537 cm was typical of form I. During the polymorphic solution of piroxicam monohydrate. Figure 6 showsfor thethe CSDmonohydrate of the three types of seeds used for this the form I peak decreased, while the peak intensity increased. The polymorphic transformation, the intensity of the form I peak decreased, while the peak intensity for the work. Membrane crystallization generated narrower CSDs than antisolvent addition.

transition wasincreased. also detectable qualitatively by a change of color of the qualitatively slurry from white (form of I) monohydrate The polymorphic transition was also detectable by a change to bright yellow (monohydrate). The Malvern Mastersizer was used to determine the crystal size color of the slurry from white (form I) to bright yellow (monohydrate). The Malvern Mastersizer was distribution of the three of seeds. Seeds produced bythree membrane were analyzed by in used to determine the types crystal size distribution of the typescrystallization of seeds. Seeds produced form of slurry, while seeds from antisolvent addition and polymorphic transformation were filtered membrane crystallization were analyzed in form of slurry, while seeds from antisolvent addition and and dried before being dispersed in filtered a saturated solution of dispersed piroxicam in monohydrate. Figure 6 polymorphic transformation were and aqueous dried before being a saturated aqueous shows the of the monohydrate. three types of Figure seeds used for this work. crystallization generated solution ofCSD piroxicam 6 shows the CSD ofMembrane the three types of seeds used for this narrower CSDs than antisolvent addition. work. Membrane crystallization generated narrower CSDs than antisolvent addition.


8 of 15


8 of 15

Figure Figure6. 6. Crystal Crystal size size distribution distributionof ofthe thedifferent differentseeds seedsfrom fromMalvern MalvernMastersizer. Mastersizer.

Somesignificant significantstatistics statisticscalculated calculatedwith withthe theMalvern Malvernsoftware softwareare areshown shownin inTable Table1.1.Membrane Membrane Some seedshave havethe thelower lowerspan span(calculated (calculatedasas[d(0.9) [d(0.9)−−d(0.1)]/d(0.5)] d(0.1)]/d(0.5)] and, and, therefore, therefore, related related to to the thewidth width seeds ofthe theCSD) CSD)and andthe thelower lowervolume volumeweighted weightedmean meandiameter. diameter.Furthermore, Furthermore, as as shown shown in inthe theoptical optical of micrographs membrane membrane seeds seeds are areless lessagglomerated agglomeratedcompared comparedto tothe thepolymorphic polymorphictransformation transformation micrographs andantisolvent antisolventones. ones. and Table MalvernMastersizer Mastersizerdata data seeds used, where d(0.1), and are d(0.9) the Table1.1. Malvern forfor thethe seeds used, where d(0.1), d(0.5),d(0.5), and d(0.9) theare particle particle diameters 10%, of the cumulative distribution, was calculated diameters at 10%, at 50% and50% 90%and of 90% the cumulative distribution, span wasspan calculated as [d(0.9)as− [d(0.9) − d(0.1)]/d(0.5), andand D[4,3] and D[3,2] are volume the volume surfaceweighted weighted mean mean diameters, d(0.1)]/d(0.5), and D[4,3] D[3,2] are the andand surface diameters, respectively. The table shows the average values of all the measurement replicates and corresponding respectively. The table shows the average values of all the measurement replicates and corresponding standard standarddeviations. deviations. Seeds Type d(0.1) (µm) d(0.5) (µm) d(0.9) (µm) Span(-) d(0.1) (µm) d(0.5) (µm) d(0.9) (µm) Span(-) Membrane 19.8 ± 1.23 32.5 ± 1.85 53.1 ± 3.62 1.03 ± 0.07 Membrane 19.8 ±28.8 1.23± 0.49 32.5 50.6 ± 1.85 ± 3.62 1.03 Transformation ± 0.83 53.187.5 ± 1.45 1.16±± 0.07 0.002 Transformation 0.49± 0.05 50.6 ± ± 1.45 ± ±0.002 Antisolvent 28.8 ±65.6 1210.83 ± 3.39 87.5 220.9 ± 12.85 1.16 1.28 0.07

Seeds Type

Antisolvent

65.6 ± 0.05

121 ± 3.39

220.9 ± 12.85

1.28 ± 0.07

D[4,3] (µm) 34.8 ± 4.25 34.8 4.25 55.0 ± ± 0.90 55.0 0.90 134 ±±5.15

D[3,2] (µm) D[3,2] (µm) 30.2 ± 1.73 30.2± ± 1.73 46.0 0.78 46.0± 2.67 ± 0.78 101

134 ± 5.15

101 ± 2.67

D[4,3] (µm)

3.2. Seeded Growth Experiments 3.2. Seeded Growth Experiments Similar seeded batch cooling crystallization experiments were carried out using the three types Similar seeded batch cooling crystallization were carried out using solid the three types of seed crystals prepared. Solutions saturated at experiments 40 °C were prepared by dissolving piroxicam ◦ of crystalssolvent prepared. at 40 C were prepared solid inseed the chosen at 50Solutions °C. Seedssaturated were added after cooling down toby37dissolving °C, and then thepiroxicam solutions in the further chosen solvent at 50 ◦to C.10 Seeds added after cooling 37 ◦ C,deplete and then the solutions were cooled down °C atwere a rate of −0.1 °C/min. In down order to fully supersaturation, were further cooled down 10 ◦ C atata 10 rate ofRaman −0.1 ◦ C/min. In order to used fully to deplete supersaturation, the temperature was keptto constant °C. spectroscopy was monitor the evolution ◦ the temperature was keptThe constant 10 C. Raman spectroscopy was used to monitor the evolution of of solute concentration. peak at intensity at 1443–1438 cm−1 (after calculation of second derivative −1 (afterof solute concentration. peakasintensity at 1443–1438 calculation of dissolved second derivative and and smoothing) wasThe used an indication of thecmamount piroxicam in solution. smoothing) wassolubility used as an indication of the amount of piroxicam dissolved solution.of Ansuch inferential An inferential curve was determined by measuring the values ofinintensity Raman solubility wassaturated determined by measuring the values of intensity ofatsuch Ramantemperatures. peak for several peak for curve several solutions of piroxicam monohydrate, different A saturated solutions of piroxicam different temperatures. was polynomial function was used monohydrate, to interpolateatthe data and is shown A in polynomial Figure 7b. function The solution used to interpolatewas theconsumed data and isby shown in Figure 7b. The solution supersaturation consumed by supersaturation growth, secondary nucleation and, partly, by was agglomeration in growth, secondary nucleation partly, by agglomeration in about 8–10 h, temperature indicating slow kinetics about 8–10 h, indicating slow and, kinetics of growth and nucleation. After that, cycling was of growthtoand nucleation. that, temperature cyclingfine was crystals. applied to agglomeration applied reduce crystalAfter agglomeration and dissolve A reduce total ofcrystal nine cycles of 20 °C and dissolvewas fineused. crystals. A total of nine cycles of 20 ◦ C amplitude was used. amplitude


9 of 15


9 of 15

(a)

(b)

(c) Figure 7.7. (a) (a) Total Total counts counts and and temperature temperature during during the the seeded seeded experiments experiments with with 2% 2% membrane membrane Figure crystals; (b) Intensity of the Raman peak corresponding to dissolved piroxicam (solute); (c) Trends for crystals; (b) Intensity of the Raman peak corresponding to dissolved piroxicam (solute); (c) Trends FBRM (Focused Beam Reflectance Measurement) counts and mean of the square weighted chord for FBRM (Focused Beam Reflectance Measurement) counts and mean of the square weighted chord lengthdistribution. distribution. length

The evolution of the total counts/measurement during the experiment is shown in Figure 7a The evolution of the total counts/measurement during the experiment is shown in Figure 7a while Figure 7c shows the most significant FBRM statistics. It can be noticed that after two cycles a while Figure 7c shows the most significant FBRM statistics. It can be noticed that after two cycles a stable oscillatory trend for the total counts/meas. and mean of the squared weighted chord length stable oscillatory trend for the total counts/meas. and mean of the squared weighted chord length distribution was reached. As shown in Figure 7b, the solute concentration during cycling remained distribution was reached. As shown in Figure 7b, the solute concentration during cycling remained close to equilibrium conditions, indicating a high yield at the end of the experiment. close to equilibrium conditions, indicating a high yield at the end of the experiment. The trends of the counts/meas. between 50–150 and 150–300 μm (corresponding to the size of The trends of the counts/meas. between 50–150 and 150–300 µm (corresponding to the size of monohydrate agglomerates) also reached a stable trend after two cycles. Microscopic images of the monohydrate agglomerates) also reached a stable trend after two cycles. Microscopic images of the crystals from filtered and dried samples (Supplementary Materials Figure S3) also showed a decrease crystals from filtered and dried samples (Supplementary Materials Figure S3) also showed a decrease in the number of agglomerates while cycling. in the number of agglomerates while cycling. Figure 8 shows the trends for the total counts/measurement for the other three seeded cooling Figure 8 shows the trends for the total counts/measurement for the other three seeded cooling and cycling experiments. In all cases, a longer time was needed to reach a stable oscillating trend in and cycling experiments. In all cases, a longer time was needed to reach a stable oscillating trend the total counts/measurement compared to the seeded experiments with 2% membrane seeds. In fact, in the total counts/measurement compared to the seeded experiments with 2% membrane seeds. nine cycles were not enough to reach a stable oscillating trend for the antisolvent addition and 6% In fact, nine cycles were not enough to reach a stable oscillating trend for the antisolvent addition and membrane seeds, as shown in Figure 8a,b. Figure 8c shows that in the case of the polymorphic transformation seeds, four temperature cycles were needed to reach a stable trend. The maximum amount of total counts reached after 10 h at 10 °C for this last type of seeds was around 2100 total


10 of 15

6% membrane seeds, as shown in Figure 8a,b. Figure 8c shows that in the case of the polymorphic transformation seeds, cycles were needed to reach a stable trend. The maximum Pharmaceutics 2018, 10, x FORfour PEERtemperature REVIEW 10 of 15 amount of total counts reached after 10 h at 10 ◦ C for this last type of seeds was around 2100 total counts/measurement,significantly significantlylower lowerthan thanthe themaximum maximumreached reachedfor for all all the the other other experiments experiments counts/measurement, (between 5000 and 7000 total counts/measurement). counts/measurement).

(a)

(b)

(c) Figure 8. (a) (a) Total Totalcounts countsand and temperature profile seeded experiment 6% membrane Figure 8. temperature profile for for the the seeded experiment with with 6% membrane seeds; seeds; (b) 2% antisolvent addition seeds; (c) polymorphic transformation seeds. (b) 2% antisolvent addition seeds; (c) polymorphic transformation seeds.

In order to better compare the crystals obtained at the end of each cycling experiments, In order to better compare the crystals obtained at the end of each cycling experiments, microscopic microscopic images are shown in Figure 9. The narrowest CSDs were obtained with membrane seeds images are shown in Figure 9. The narrowest CSDs were obtained with membrane seeds and seeds and seeds from polymorphic transformation, as it can be noted in Figure 9a,b,d. The final degree of from polymorphic transformation, as it can be noted in Figure 9a,b,d. The final degree of agglomeration agglomeration at the end of each experiment was very similar, as shown in Figure 9. However, in at the end of each experiment was very similar, as shown in Figure 9. However, in Figure 9c few large Figure 9c few large agglomerates could still be identified in the crystals obtained using antisolvent agglomerates could still be identified in the crystals obtained using antisolvent seeds. Microscopic seeds. Microscopic images of crystals collected at different times during the seeded crystallization images of crystals collected at different times during the seeded crystallization experiments can be experiments can be found in the Supplementary Materials Figures S3–S7. found in the Supplementary Materials Figures S3–S7.


11 of 15


11 of 15

(a)

(b)

(c)

(d)

Figure 9. Microscopic images of crystals grown from: (a) membrane seeds (2% of the total mass of Figure 9. Microscopic images of crystals grown from: (a) membrane seeds (2% of the total mass piroxicam); (b) (b) membrane seeds (6%(6% of the totaltotal mass of piroxicam); (c) antisolvent seeds; (d) of piroxicam); membrane seeds of the mass of piroxicam); (c) antisolvent seeds; transformation seeds. (d) transformation seeds.

The CSDs and the main statistics for the crystals obtained at the end of the cycling experiments The CSDs and the main statistics for the crystals obtained at the end of the cycling experiments are shown in Figure 10 and Table 2. The highest increase in volume weighted mean diameter during are shown in Figure 10 and Table 2. The highest increase in volume weighted mean diameter during the cycling experiment was observed for the membrane seeds: +243% when 2% seeds were used, and the cycling experiment was observed for the membrane seeds: +243% when 2% seeds were used, and +205% for 6% seeds. +205% for 6% seeds. Table 2. Malvern Mastersizer data for the crystals at the end of the seeded experiments, where d(0.1), Table 2. Malvern Mastersizer data for the crystals at the end of the seeded experiments, where d(0.1), d(0.5), and d(0.9) are the particle diameters at 10%, 50% and 90% of the cumulative distribution, Span d(0.5), and d(0.9) are the particle diameters at 10%, 50% and 90% of the cumulative distribution, Span was calculated as [d(0.9) − d(0.1)]/d(0.5), and D[4,3] and D[3,2] are the volume and surface weighted was calculated as [d(0.9) − d(0.1)]/d(0.5), and D[4,3] and D[3,2] are the volume and surface weighted mean diameters, respectively. The table shows the average values of all the measurement replicates mean diameters, respectively. The table shows the average values of all the measurement replicates and corresponding standard deviations. and corresponding standard deviations. Sample Sample 2% Membrane Membrane 6% Membrane 2% Transformation Membrane 6% Transformation Antisolvent Antisolvent

d(0.1) (µm) d(0.1) 78.7(µm) ± 1.21 67.7 0.31 78.7 ± ±1.21 104 67.7 ± ±0.45 0.31 104 ±0.45 70.3 ± 1.07 70.3 ± 1.07

d(0.5) (µm) d(0.5) 115 ±(µm) 0.46 101±±0.46 0.70 115 148±±0.70 0.62 101 148 181±±0.62 1.59 181 ± 1.59

d(0.9) (µm) d(0.9) 166 ±(µm) 3.61 152±± 3.61 1.60 166 211±± 1.60 0.86 152 211 340±± 0.86 1.97 340 ± 1.97

Span(-) 0.76 ±Span(-) 0.04 (−25%) 0.83 ± ± 0.01 0.76 0.04 ((−19%) −25%) 0.72 0.83 ±±0.001 0.01 ((−38%) −19%) 0.72 (−38%) 1.49±± 0.001 0.01 (+16%) 1.49 ± 0.01 (+16%)

D[4,3] (µm) (µm) 119 D[4,3] ± 0.9 (+243%) 106 119±±0.74 0.9 (+205%) (+243%) 153 106 ±±0.64 0.74(+179%) (+205%) 153 0.64 (+46%) (+179%) 195 ± ± 1.38 195 ± 1.38 (+46%)

D[3,2] (µm) (µm) 110 ±D[3,2] 0.33 (+263%) 96.7 110±±0.63 0.33(+220%) (+263%) 142 0.60 96.7±± 0.63(+209%) (+220%) 142 ±±1.35 0.60(+36%) (+209%) 138 138 ± 1.35 (+36%)

A decrease in the span of the distribution (parameter related to its narrowness) was noticeable A decrease in the span the obtained distribution to its was noticeable for all seeds apart from theof ones by (parameter antisolvent related addition. In narrowness) particular, the polymorphic for all seeds apart from oneslowest obtained byof antisolvent addition. In particular, transformation seeds ledthe to the value the span and its highest decrease.the Thepolymorphic final crystal transformation seeds value of the span its highest decrease. size distributions for led eachtoofthe thelowest cycling experiments areand shown in Figure 10. The final crystal size distributions for each of the cycling experiments are shown in Figure 10.

Pharmaceutics 2018, 10, 17 Pharmaceutics 2018, 10, x FOR PEER REVIEW

12 of 15 12 of 15

Figure 10. 10. Crystal size distributions of the crystals obtained at the end of the seeded experiments. Figure Crystal Crystal size size distributions distributions (CSDs) (CSDs) measured measured with with the theMalvern MalvernMastersizer. Mastersizer.

While the themembrane membraneseeds seedsand and the ones from polymorphic transformation generated narrow While the ones from polymorphic transformation generated narrow and and unimodal distributions, the antisolvent seeds led to a very broad CSD with a high amount of unimodal distributions, the antisolvent seeds led to a very broad CSD with a high amount of fines. fines. In conclusion, using membrane seeds and polymorphic transformation seeds allowed for In conclusion, using membrane seeds and polymorphic transformation seeds allowed for a narrowa narrow CSD with low tendency to agglomeration. The number of temperature cycles needed to reach CSD with low tendency to agglomeration. The number of temperature cycles needed to reach the the equilibrium size and to remove all agglomerates was around 2 for the membrane seeds (2% of the equilibrium size and to remove all agglomerates was around 2 for the membrane seeds (2% of the total total mass of piroxicam in solution at the moment of seeding), and 4 for the polymorphic mass of piroxicam in solution at the moment of seeding), and 4 for the polymorphic transformation transformation seeds. Therefore, shorter time wasmembrane needed when membrane seeds were used seeds. Therefore, a shorter batch atime wasbatch needed when seeds were used in the correct in the correct amount. in Furthermore, in order 2 g oftransformation polymorphic transformation seeds, a amount. Furthermore, order to produce 2 g to of produce polymorphic seeds, a fine powdered fine powdered sample of piroxicam form I needed to be suspended in water for over 40 hours, for sample of piroxicam form I needed to be suspended in water for over 40 hours, for complete conversion complete conversion to the form. On thethe other hand, producing the same amount of to the monohydrate form. Onmonohydrate the other hand, producing same amount of membrane seeds does not membrane seeds does not require any previous sample preparation and takes only a few seconds. require any previous sample preparation and takes only a few seconds. Additionally, membrane seeds Additionally, membrane seeds canmode be produced in a continuous with a similar apparatus [28] can be produced in a continuous with a similar apparatusmode [28] making this seed production making particularly this seed efficient production method particularly efficient for piroxicam and other highly method for piroxicam and other highly agglomerating compounds. agglomerating compounds. 4. Conclusions 4. Conclusions A novel procedure for seeds production was used in combination with temperature cycling to novel procedure seeds production used in(piroxicam combination with temperature cycling to growAcrystals of a highly for agglomeration prone was compound monohydrate). The membrane grow crystals of a highly prone compound (piroxicam monohydrate). membrane seeds were compared toagglomeration seeds obtained by conventional antisolvent addition andThe polymorphic seeds were compared to seeds obtained by conventional antisolvent addition and polymorphic transformation. Crystals of monohydrate could not be obtained by cooling crystallization with transformation. Crystals of monohydrate could not be obtained by cooling crystallization with in situ in situ primary nucleation. primary The nucleation. membrane seeds allowed a narrow crystal size distribution to be obtained at the end of the The membrane seeds agglomeration. allowed a narrow crystal sizeofdistribution to be obtained endcrystal of the batch without significant The quality the final crystals in termsatofthe final batch without significant agglomeration. The quality of the final crystals in terms of final crystal size size distribution was comparable to the one gained using seeds from a polymorphic transformation. distribution was comparable to be theproduced one gained polymorphic transformation. However, membrane seeds can in ausing fasterseeds and from more aefficient way compared to the However, membrane seeds can be produced in a faster and more efficient way compared to the polymorphic transformation ones. polymorphic transformation Furthermore, membraneones. seeds were used in combination with the direct nucleation control Furthermore, membrane seeds were used crystal in combination with the nucleation control strategy, which led to large crystals with a narrow size distribution anddirect no visible agglomeration. strategy, which led to large crystals with a narrow crystal size distribution and no In conclusion, this study shows that crystal agglomeration can be prevented efficiently byvisible using aagglomeration. specific seeds preparation and temperature cycling. Agglomeration is an undesired phenomenon conclusion, this study shows crystal can be prevented efficiently by using sinceIn it causes the broadening of the that crystal size agglomeration distribution and solvent incorporation. Membrane a specific seeds preparation and temperature cycling. Agglomeration is an undesired phenomenon crystallization is a fast and efficient way to obtain non-agglomerated crystals with narrow CSD, which since causes broadening of the crystaland sizedrying) distribution and solvent incorporation. Membrane can beitused as athe final product (after filtration or as seeds when larger crystals are required. crystallization is a fast and efficient way to obtain non-agglomerated crystals with narrow CSD, which can be used as a final product (after filtration and drying) or as seeds when larger crystals are


13 of 15

As shown by the experimental results, this method is particularly useful for compounds with a slow nucleation rate that cannot be generated at low supersaturation to avoid agglomeration. Supplementary Materials: The following are available online at www.mdpi.com/1999-4923/10/1/17/s1, Figure S1: Temperature profile and total counts evolution during a preliminary cycling experiments performed in order to establish the best cycles’ amplitude and rate, Figure S2: Microscopic images of crystals during the cycling experiment of Figure S1, Figure S3: Microscopic images of crystals during the cycling experiment 1 (2% membrane seeds), Figure S4: Microscopic images of crystals during the cycling experiment 2 (6% membrane seeds), Figure S5: Microscopic images of crystals during the cycling experiment 3 (antisolvent seeds), Figure S6: Microscopic images of crystals during the cycling experiment 4 (polymorphic transformation seeds), Figure S7: FBRM statistics during cycling experiments 1, 3 and 4, Figure S8: Temperature, total counts, Raman intensity and microscopic images of crystals for the DNC experiment 1, Figure S9: Temperature, total counts, Raman intensity and microscopic images of crystals for the DNC experiment 2, Figure S10: Microscopic images of crystals during the DNC experiment 1, Figure S11: Microscopic images of crystals during the DNC experiment 2, Table S1: Malvern Mastersizer data for the seeds and the crystals at the end of the DNC experiment 2. Acknowledgments: T. B. Hansen from Syddansk Universitet (Denmark) is acknowledged for the piroxicam solubility data and for useful discussion on the design of experiments. Financial support was provided by the European Research Council grant no. [280106-CrySys]. R. Othman gratefully acknowledges the financial support given for this work through the Ministry of Higher Education Malaysia. Author Contributions: Elena Simone designed and carried out the experiments, analyzed samples and data, and wrote the manuscript. Rahimah Othman prepared the membrane seeds used for the experiments and provided some of the Mastersizer measurements. Zoltan K. Nagy provided laboratory space, equipment and consumables. Both Zoltan K. Nagy and Goran T. Vladisavljević supervised the experimental work and data analysis and revised the manuscript before submission. Conflicts of Interest: The authors declare no conflicts of interest.

References 1. 2. 3. 4. 5. 6.

7. 8.

9.

10. 11.

Ålander, E. Agglomeration of paracetamol during crystallization in pure and mixed solvents. Ind. Eng. Chem. Res. 2004, 43, 629–637. [CrossRef] Lindenberg, C.; Scholl, J.; Vicum, L.; Mazzotti, M.; Brozio, J. L-Glutamic acid precipitation: Agglomeration effect. Cryst. Growth Des. 2008, 8, 224–237. [CrossRef] Yu, Z.Q.; Tan, R.B.H.; Chow, P.S. Effects of operating conditions on agglomeration and habit of paracetamol crystals in anti-solvent crystallization. J. Cryst. Growth 2005, 279, 477–488. [CrossRef] Brunsteiner, M.; Jones, A.G.; Pratola, F.; Price, S.L.; Simons, S.J.R. Toward a molecular understanding of crystal agglomeration. Cryst. Growth Des. 2005, 5, 3–16. [CrossRef] Ochsenbein, D.R.; Vetter, T.; Morari, M.; Mazzotti, M. Agglomeration of needle-like crystals in suspension. II. Modeling. Cryst. Growth Des. 2015, 15, 4296–4310. [CrossRef] Saleemi, A.N.; Steele, G.; Pedge, N.I.; Freeman, A.; Nagy, Z.K. Enhancing crystalline properties of a cardiovascular active pharmaceutical ingredient using a process analytical technology based crystallization feedback control strategy. Int. J. Pharm. 2012, 430, 56–64. [CrossRef] [PubMed] Chow, A.H.L.; Leung, M.W.M. A study of the mechanisms of wet spherical agglomeration of pharmaceutical powders. Drug Dev. Ind. Pharm. 1996, 22, 357–371. [CrossRef] Kawashima, Y.; Imai, M.; Takeuchi, H.; Yamamoto, H.; Kamiya, K.; Hino, T. Improved flowability and compactibility of spherically agglomerated crystals of ascorbic acid for direct tableting designed by spherical crystallization process. Powder Technol. 2003, 130, 283–289. [CrossRef] Thati, J. Particle Engineering by Spherical Crystallization: Mechanisms and Influence of Process Conditions. Ph.D. Thesis, KTH Royal Institute of Technology, Stockholm, Sweden, 2011. Available online: http://www. diva-portal.org/smash/record.jsf?pid=diva2:410938 (accessed on 1 July 2016). Yoshiura, H.; Nagano, H.; Hirasawa, I. New insights into additive structure effect on crystal agglomeration of L-valine. Chem. Eng. Technol. 2013, 36, 2023–2028. [CrossRef] Kawashima, Y.; Niwa, T.; Takeuchi, H.; Hino, T.; Itoh, Y.; Furuyama, S. Characterization of polymorphs of tranilast anhydrate and tranilast monohydrate when crystallized by two solvent change spherical crystallization techniques. J. Pharm. Sci. 1991, 80, 472–478. [CrossRef] [PubMed]


12.

13.

14.

15.

16.

17. 18. 19. 20. 21. 22. 23. 24.

25. 26.

27. 28.

29.

30. 31. 32.

33.

14 of 15

Peña, R.; Nagy, Z.K. Process intensification through continuous spherical crystallization using a two-stage mixed suspension mixed product removal (MSMPR) system. Cryst. Growth Des. 2015, 15, 4225–4236. [CrossRef] Toldy, A.I.; Badruddoza, A.Z.M.; Zheng, L.; Hatton, T.A.; Gunawan, R.; Rajagopalan, R. Spherical crystallization of glycine from monodisperse microfluidic emulsions. Cryst. Growth Des. 2012, 12, 3977–3982. [CrossRef] Abioye, A.O.; Kola-Mustapha, A.; Ruparelia, K. Impact of in situ granulation and temperature quenching on crystal habit and micromeritic properties of ibuprofen-cationic dextran conjugate crystanules. Int. J. Pharm. 2014, 462, 83–102. [CrossRef] [PubMed] Abioye, A.O.; Kola-Mustapha, A.; Chi, G.T.; Ilya, S. Quantification of in situ granulation-induced changes in pre-compression, solubility, dose distribution and intrinsic in vitro release characteristics of ibuprofen-cationic dextran conjugate crystanules. Int. J. Pharm. 2014, 471, 453–477. [CrossRef] [PubMed] Abioye, A.O.; Chi, G.T.; Simone, E.; Nagy, Z.K. Real-time monitoring of the mechanism of ibuprofen-cationic dextran crystanule formation using crystallization process informatics system (CryPRINS). Int. J. Pharm. 2016, 509, 264–278. [CrossRef] [PubMed] Chabanon, E.; Mangin, D.; Charcosset, C. Membranes and crystallization processes: State of the art and prospects. J. Member. Sci. 2016, 509, 57–67. [CrossRef] Kober, P.A. Pervaporation, perstillation and percrystallization. J. Member. Sci. 1995, 100, 61–64. [CrossRef] Wu, Y.; Kong, Y.; Liu, J.; Zhang, J.; Xu, J. An experimental study on membrane distillation-crystallization for treating waste water in taurine production. Desalination 1991, 80, 235–242. [CrossRef] Azoury, R.; Garside, J.; Robertson, W.G. Crystallization processes using reverse osmosis. J. Cryst. Growth 1986, 79, 654–657. [CrossRef] Zhang, X.; Zhang, P.; Wei, K.; Wang, Y.; Ma, R. The study of continuous membrane crystallization on lysozyme. Desalination 2008, 219, 101–117. [CrossRef] Curcio, E.; Di Profio, G.; Drioli, E. Membrane crystallization of macromolecular solutions. Desalination 2002, 145, 173–177. [CrossRef] Di Profio, G.; Di Curcio, E.; Cassetta, A.; Lamba, D.; Drioli, E. Membrane crystallization of lysozyme: Kinetic aspects. J. Cryst. Growth 2003, 257, 359–369. [CrossRef] Di Profio, G.; Fontanova, E.; Curcio, E.; Drioli, E. From tailored supports to controlled nucleation: Exploring material chemistry, surface nanostructure, and wetting regime effects in heterogeneous nucleation of organic molecules. Cryst. Growth Des. 2012, 12, 3749–3757. [CrossRef] Caridi, A.; Di Profio, G.; Caliandro, R.; Guagliardi, A.; Curcio, E.; Drioli, E. Selecting the desired solid form by membrane crystallizers: Crystals or cocrystals. Cryst. Growth Des. 2012, 12, 4349–4356. [CrossRef] Di Profio, G.; Curcio, E.; Ferraro, S.; Stabile, C.; Drioli, E. Effect of supersaturation control and heterogeneous nucleation on porous membrane surfaces in the crystallization of L-glutamic acid polymorphs. Cryst. Growth Des. 2009, 9, 2179–2186. [CrossRef] Hansen, T.B.; Qu, H. Formation of Piroxicam polymorphism in solution crystallization: Effect and interplay of operation parameters. Cryst. Growth Des. 2015, 15, 4694–4700. [CrossRef] Othman, R.; Vladisavljević, G.T.; Simone, E.; Holdich, R.G. Preparation of microcrystals of piroxicam monohydrate by antisolvent precipitation via microfabricated metallic membranes with ordered pore arrays. Cryst. Growth Des. 2017, 17, 6692–6702. [CrossRef] [PubMed] Othman, R. Production of Functional Pharmaceutical Nano/Micro-Particles by Solvent Displacement Method Using Advanced Micro-Engineered Dispersion Devices. Ph.D. Thesis, Loughborough University, Loughborough, UK, 2016. Liu, G.; Hansen, T.B.; Qu, H.; Yang, M.; Pajander, J.P.; Rantanen, J.; Christensen, L.P. Crystallization of piroxicam solid forms and the effects of additives. Chem. Eng. Technol. 2014, 37, 1297–1304. [CrossRef] Abu Bakar, M.R.; Nagy, Z.K.; Saleemi, A.N.; Rielly, C.D. The impact of direct nucleation control on crystal size distribution in pharmaceutical crystallization processes. Cryst. Growth Des. 2009, 9, 1378–1384. [CrossRef] Saleemi, A.N.; Rielly, C.D.; Nagy, Z.K. Comparative investigation of supersaturation and automated direct nucleation control of crystal size distribution using ATR-Uv/Vis spectroscopy and FBRM. Cryst. Growth Des. 2012, 12, 1792–1807. [CrossRef] Yang, Y.; Song, L.; Nagy, Z.K. Automated direct nucleation control in continuous mixed suspension mixed product removal cooling crystallization. Cryst. Growth Des. 2015, 15, 5839–5848. [CrossRef]


34.

35.

36.

37.

38.

15 of 15

Kacker, R.; Salvador, P.M.; Sturm, G.S.J.; Stefanidis, G.D.; Lakerveld, R.; Nagy, Z.K. Microwave Assisted Direct Nucleation Control for Batch Crystallization: Crystal Size Control with Reduced Batch Time. Cryst. Growth Des. 2016, 16, 440–446. [CrossRef] Simone, E.; Zhang, W.; Nagy, Z.K. Application of process analytical technology-based feedback control strategies to improve purity and size distribution in biopharmaceutical crystallization. Cryst. Growth Des. 2015, 15, 2908–2919. [CrossRef] Morelli, S.; Holdich, R.G.; Dragosavac, M.M. Chitosan and Poly (Vinyl Alcohol) microparticles produced by membrane emulsification for encapsulation and pH controlled release. Chem. Eng. J. 2016, 288, 451–460. [CrossRef] Imbrogno, A.; Dragosavac, M.M.; Piacentini, E.; Vladisavljević, G.T.; Holdich, R.G.; Giorno, L. Polycaprolactone multicore-matrix particle for the simultaneous encapsulation of hydrophilic and hydrophobic compounds produced by membrane emulsification and solvent diffusion processes. Colloids Surf. B Biointerfaces 2015, 135, 116–125. [CrossRef] [PubMed] Vladisavljević, G.T.; Wang, B.; Dragosavac, M.M.; Holdich, R.G. Production of food-grade multiple emulsions with high encapsulation yield using oscillating membrane emulsification. Colloids Surf. A Physicochem. Eng. Asp. 2014, 458, 78–84. [CrossRef] © 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

PREVENTING CRYSTAL AGGLOMERATION OF PHARMACEUTICAL CRYSTALS

Recommend Documents