Contents lists available at ScienceDirect The Journal of Supercritical Fluids journal homepage: www.elsevier.com/locate/supflu Polymorphism in the co-crystallization of the anticonvulsant drug carbamazepine and saccharin using supercritical CO2 as an anti-solvent Isaac A. Cuadra, Albertina Cabañas, José A.R. Cheda, Concepción Pando⁎ Departamento de Química Física I, Universidad Complutense, E-28040 Madrid, Spain G R A P H I C A L A B S T R A C T A R T I C L E I N F O Keywords: Carbon dioxide Pharmaceutical co-crystals Supercritical anti-solvent Carbamazepine Saccharin Applications A B S T R A C T 1:1 Co-crystals of carbamazepine (CBZ) and saccharin (SAC) were obtained for the first time through the su- percritical anti-solvent (SAS) technique based on using supercritical CO2 as anti-solvent. The capability of SAS to produce the desired polymorphic form (two polymorphs are known) was assessed. Operational conditions in- vestigated were temperature (40.0 and 60.0 °C), pressure (10.0 and 15.0MPa), solvent choice and coformer concentration in the organic solution (CBZ: 30 and 15mg/mL; SAC: stoichiometric ratio). Co-crystals were characterized in terms of crystallinity and coformers interactions. No homocrystals were present. Using me- thanol, at 40.0 °C polymorph I was obtained with yields up to 65%; whilst at 60.0 °C a mixture of polymorphs was obtained. Mixtures of polymorphs were also obtained in the ethanol and dichloromethane experiments at the studied conditions while the dimethylsulfoxide experiments failed to produce any co-crystal polymorph. For comparison purposes, pure CBZ and SAC were also processed by SAS. 1. Introduction The production of co-crystals is reaching a major importance in the pharmaceutical industry not only to overcome some crucial drawbacks of new produced drugs such as poor solubility, inadequate dissolution profiles or short shelf-life, but also to improve the drug organoleptic and mechanical properties [1]. Moreover, multidrug co-crystallization is becoming an important field of research in the treatment of some complex disorders [2]. A broad and generally accepted definition of co- crystal would be “a stoichiometric multi-component system connected by non-covalent interactions where all the components present are solid under ambient conditions” [3]. A pharmaceutical co-crystal would therefore involve a bonding through supramolecular synthons of at least an active pharmaceutical ingredient (API) and another API (in https://doi.org/10.1016/j.supflu.2018.02.004 Received 18 December 2017; Received in revised form 2 February 2018; Accepted 2 February 2018 ⁎ Corresponding author at: Departamento de Química Física I, Facultad C. Químicas, Universidad Complutense, E-28040 Madrid, Spain. E-mail address: pando@quim.ucm.es (C. Pando). Abbreviations: API, active pharmaceutical ingredient; BCS, Biopharmaceutics Classification System; BPR, back pressure regulator; CBZ, carbamazepine; CSS, co-crystallization with supercritical solvent; DCM, dichloromethane; DSC, differential scanning calorimetry; DMSO, dimethylsulfoxide; EMA, European Medicines Agency; FDA, Food and Drug Administration; FTIR, Fourier transform infrared; GAS, gas anti-solvent; HPLC, high performance liquid chromatography; P, pressure; T, temperature; SAC, saccharin; SAS, supercritical anti-solvent; SEA, supercritical fluid enhanced atomization; SEM, scanning electron microscopy; PXRD, powder X-ray diffraction The Journal of Supercritical Fluids 136 (2018) 60–69 Available online 08 February 2018 0896-8446/ © 2018 Elsevier B.V. All rights reserved. T case of multidrug co-crystals), or a suitable coformer. A guidance and regulatory classification for pharmaceutical co-crystals has been re- cently given by the European Medicines Agency (EMA) [4] and the United States Food and Drug Administration (FDA) [5]. According to the recent rules, co-crystals are considered a drug polymorph rather than a new API and drug development and regulatory submissions are simplified [6]. In this study, we investigate the production of the 1:1 co-crystal of carbamazepine (CBZ) and saccharin (SAC) using the supercritical anti- solvent (SAS) technique; this co-crystal consists of CBZ and SAC mo- lecular units in the stoichiometric ratio 1:1 linked through the bonds shown in Fig. 1. CBZ (molar mass= 236.27 gmol−1) is a poorly water- soluble drug used primarily in the treatment of epilepsy and trigeminal neuralgia that belongs to BCS class II. CBZ has five known anhydrous polymorphs [7–19] (at room temperature the most stable polymorph is polymorph III, an enantiotropic pair of polymorph I), and several di- hydrate and solvate polymorphs. CBZ solubilities in water and simu- lated gastric fluid may be found in Refs. [20,21]. CBZ major problems are its slow rate of absorption when administered through oral route, and its tendency to adopt the dihydrate form, which reduces its solu- bility in water to nearly a half of that of its anhydrous form. Therefore, larger doses of the drug are required in order to be effective. Aiming to improve CBZ performance, the co-crystallization of CBZ with coformer SAC has been widely studied during the last decade. Saccharin (molar mass= 183.18 gmol−1) is widely used as coformer in the preparation of pharmaceutical salts (acting as a weak acid when combined with a sufficiently basic molecule), or co-crystals (remaining then a neutral molecule). SAC has one known polymorph [22] and good water solu- bility [21]. The first published structure of the CBZ-SAC co-crystal re- ported by Fleischman et al. [23] corresponds to the more stable poly- morph I [24] and was produced by slow evaporation from an equimolar solution of CBZ and SAC in ethanol. The structures of CBZ and SAC along with the co-crystal polymorphs are shown in Fig. 1. In polymorph I, the CBZ molecules are bond to each other through an amide–amide homosynthon and the saccharin molecules bond through a pyridine- carbonyl oxygen H-bond to a CBZ molecule and through a sulfonyl oxygen-amide H-bond to another CBZ molecule. The polymorph II of the co-crystal was discovered later by Porter et al. [25] using eva- poration crystallization from an ethanol solution with the aid of func- tionalized cross-linked polymers. In polymorph II, the amide–amide homosynthon of polymorph I established between the two CBZ mole- cules disappears and each molecule of CBZ bonds through a pyr- idine–carbonyl oxygen and an amide–carbonyl oxygen H-bond to one saccharin molecule; and through an amide–sulfonyl oxygen to another saccharin molecule (see Fig. 1). Huskić et al. [26] and Maeshwari et al. [27] reported the formation of other polymorphs, but no crystal- lographic information file of these polymorphs is available yet and most authors only take into account the co-crystal polymorphs I and II. A co-crystal can be prepared by several conventional methods with the aid of solvents (evaporative or cooling crystallization and reaction crystallization), directly from the solid state (mechanical grinding or melt crystallization) or using some more novel co-crystallization methods [6]. The use of conventional methods often presents scaling-up difficulties, leads to the presence of crystals of the individual compo- nents (homocrystals) in the final product and often involves post pur- ification steps to eliminate solvents [1,28–30]. To overcome these dif- ficulties several processes using supercritical CO2 have been developed for the preparation of pharmaceutical co-crystals [6,31]. These pro- cesses generally involve fewer steps, reduced amounts of organic sol- vents, and use CO2 that is considered a green solvent [32] because this fluid is innocuous, non-flammable, may be recycled and has readily accessible critical parameters (31 °C and 7.4MPa). Thus, the sustain- able process requirements of the pharmaceutical industry can be ful- filled using supercritical CO2 based processes. The CBZ-SAC co-crystal has already been produced using two supercritical techniques: super- critical fluid enhanced atomization (SEA) [33] and co-crystallization Fig. 1. Molecular structures of carbamazepine, saccharin, and the two polymorphs of the 1:1 carbamazepine-saccharin co-crystal showing the supramolecular synthons involved. I.A. Cuadra et al. The Journal of Supercritical Fluids 136 (2018) 60–69 61 with supercritical solvent (CSS) [34]. It is not clear though which polymorph was obtained using the SEA technique while the more stable polymorph I was produced using CSS. The SAS technique may be ap- plied to pharmaceuticals with relatively low solubility in supercritical CO2. Pharmaceuticals are dissolved in a polar organic solvent miscible with supercritical CO2. When the organic solution and the fluid are brought into contact, supersaturation takes place and precipitation starts. Solvent–free particles that exhibit narrow size distributions are obtained. In the case of pure drugs, properties such as particle size, morphology and polymorphism may be easily modified through the SAS technique by selecting different operational variables such as pressure, solution and CO2 flow rates and temperature [35–38]. At our laboratory, we have already assessed the use of the SAS technique to successfully coprecipitate the biocompatible polymer poly- vinylpyrrolidone and diflunisal [39], and to obtain a co-crystal of di- flunisal and nicotinamide [40]. Also, we have recently reviewed the preparation of pharmaceutical co-crystals using SAS and other super- critical techniques [31]. Each method was described and its application, advantages and disadvantages were discussed. Our aim in this paper is the preparation of the CBZ-SAC co-crystal using the SAS process and the investigation of the SAS co-crystal polymorphism. Due to the improved properties of the CBZ-SAC co-crystals, other authors have produced them using a wide range of conventional methodologies: grinding co-crystallization [27,41–43]; cooling crys- tallization [44,45]; membrane based anti-solvent crystallization [46,47]; reaction co-crystallization [48]; ultrasound-induced co-crys- tallization [49]; continuous hot melt extrusion co-crystallization [50] and anti-solvent co-crystallization [51,52]. We here would like to highlight the studies where polymorphism of the CBZ-SAC co-crystal was observed. Rager and Hilfiker [53] prepared this co-crystal using the reaction co-crystallization method. They were able to obtain pre- cipitates of pure CBZ-SAC co-crystal polymorph I when using pure ethylene glycol, dimethylsulfoxide (DMSO) or dimethylformamide and polymorph II when using pure acetone or dioxane. Wang et al. [54,55] used anti-solvent co-crystallization to prepare the CBZ-SAC co-crystal. As anti-solvent they used water, and as solvent they tried ethanol, acetone, methylacetate, ethylacetate and methanol; the co-crystal- lization was only successful using the latter. They found that kinetic parameters such as anti-solvent addition rate and agitation speed in- fluenced the polymorphism of the co-crystal obtaining a highly pure polymorph II of the co-crystal when addition rate and agitation speed were lower. Pagire et al. [21] developed a spherical crystallization process using a reverse anti-solvent method in which a solution pre- pared with a good solvent of the coformers (DMSO, ethanol or me- thanol) is added into an anti-solvent solution (water) with bridging li- quids (benzene, dichloromethane, DCM, or ethylacetate). They were able to produce a nearly pure polymorph II CBZ-SAC co-crystal phase and concluded that the formation of a specific polymorph depended on the supersaturation level achieved with respect to both the co-crystal and the reacting components. 2. Materials and methods 2.1. Materials The materials employed were CO2 (Air Liquide 99.98 mol% pure), carbamazepine (Sigma-Aldrich, meeting USP testing specifications), saccharin (Sigma-Aldrich, ≥99mol% pure), acetone (Sigma-Aldrich ≥99.8mol% pure), dichloromethane (Sigma-Aldrich ≥99.9 mol% pure), dimethylsulfoxide (Sigma-Aldrich ≥99.9 mol% pure), methanol (Fischer chemical ≥99.9mol% pure) and ethanol (PanReac 99.5 v/v% pure). 2.2. Supercritical fluid anti-solvent (SAS) precipitation and design of experiments Fig. 2 illustrates the SAS technique used to prepare the CBZ-SAC co- crystal, a complete description of the high-pressure equipment involved can be found elsewhere [40]. Compressed CO2 was cooled to −1 °C in a chiller and pumped into the system using a high-pressure pump (Thar- SCF high pressure Pump P-50) at a given rate. The CO2 passed through a heat exchanger, where temperature was raised to the desired ex- perimental value, before entering a 500mL precipitation chamber. A back pressure regulator (BPR, Thar-SCF ABPR-20) situated at the exit of the chamber kept the pressure constant throughout the experiment. When steady conditions were reached a 1:1 molar ratio solution of CBZ and SAC in one of the studied solvents (methanol, ethanol, DCM or DMSO) was injected using a high performance liquid chromatography (HPLC) pump (lab Alliance series III pump) into the precipitation chamber through a 100 μm nozzle at a given flow rate. The precipita- tion chamber was equipped with a heating jacket to ensure that the experiment occurred at the desired temperature. After injection the fluid dissolves in the solution, the mixture becomes supersaturated and precipitation starts. Inside the chamber, a basket with a filter (frit, 2 μm) was placed in order to collect the microparticles obtained. The effluent exiting the BPR was directed to a cyclone separator where CO2 was separated from the waste. After spraying approximately 30mL of solution the HPLC pump was stopped and CO2 (three times the volume of the chamber at the same precipitation conditions) was allowed to flow through the chamber at 20 g/min to ensure that the product was dry and free of solvent. The characteristics of precipitates obtained using the SAS process may be modified by varying the different operational parameters: so- lution concentration, temperature, pressure, solution flow rate, super- critical CO2 flow rate, nozzle diameter, drying time and choice of sol- vent. In this research we will concentrate in the influence of temperature, pressure and concentration which have been shown to be important variables in previous studies [40], and the solvent election which has been shown to be a key factor in obtaining a specific poly- morph [56]. The chosen solvents were ethanol and methanol as polar protic solvents (dielectric constants of 24.55 and 33, respectively), and DCM and DMSO as polar aprotic solvents (dielectric constants of 9.2 and 46.7, respectively). CO2 flow rate was fixed to 20 g/min and so- lution flow rate to 1mL/min. As to the pressure and temperature con- ditions, operating in the supercritical region is usually convenient. However, there is no available phase equilibria data for the quaternary mixtures formed by CO2, the solvent, and the two coformers. Due to the low solubility of CBZ and SAC in supercritical CO2 [34,57,58] we decided to neglect their possible effect in the system phase diagram, and used the available data for the binary CO2+ solvent system to establish temperature and pressure conditions above the critical locus [59–61]. The effect of using a different solvent was assessed at 40.0 °C Fig. 2. Schematic representation of the supercritical fluid anti-solvent (SAS) technique used to obtain the 1:1 CBZ-SAC co-crystal. T, P, temperature and pressure measurement; BPR, back pressure regulator. I.A. Cuadra et al. The Journal of Supercritical Fluids 136 (2018) 60–69 62 and 10.0MPa, a condition slightly above the critical point of the mix- ture of CO2 with any of the studied solvents (see Table 1). As to the coformers concentration in the organic solution, solubility data of CBZ, SAC and the co-crystals polymorph I and II in the different solvents previously published [21] were taken into account. The concentration values shown in Table 1 were selected to be relatively close to the sa- turation values of the less soluble coformer in the solvent and always maintaining the 1:1 stoichiometric ratio. The effects of pressure, tem- perature and concentration were observed using methanol as solvent, reducing CBZ concentration to 15mg/mL, and increasing temperature up to 60.0 °C and pressure up to 15.0 MPa (see Table 2). 2.3. Solid characterization A JEOL-6335F JSM electron microscope working with an accel- erating voltage of 10 or 20 kV, was used to characterize the size and morphology of the microparticles obtained in the SAS experiments. Prior to analysis, samples were coated with gold in a Q150RS Rotary- Pumped Sputter Coater. Powder X-ray diffraction (PXRD), was used to study the crystalline form of carbamazepine and saccharin prior and after SAS treatment, to identify the co-crystal form obtained in the SAS experiments and to exclude the presence of homocrystals. PXRD patterns of the solids were obtained using a Philips X’pert, model MPD powder diffractometer with vertical goniometer θ-2θ. A Cu anode X-ray tube was powered at 40 kV and 40mA (Kα1 1.54056 Å). Scans were measured between 5° and 60° 2θ with a step size of 0.016° 2θ and a continuing time of 5 s per step. The detection limit is 2% mass. Differential scanning calorimetry (DSC) was used to obtain the melting point data of carbamazepine and saccharin prior and after SAS treatment, and the co-crystals obtained in SAS experiments. These data are used together with PXRD patterns to establish the polymorph or polymorphs present in the sample. Also, the broad peak associated to the dehydration of CBZ hydrate is used to detect the presence of this poorly soluble CBZ form. DSC thermograms were measured using a TA Instruments DSC model Q-20 connected to a refrigerating cooling system. Tightly sealed volatile aluminum pans containing the samples were heated at a 5 °C/min rate in dry nitrogen, flowing at 50.0 mL/min. A MT5 Mettler microbalance was used to weigh the samples, ranging between 3 and 10mg (with an error of ± 0.001mg). Temperature and enthalpy of the calorimeter were previously calibrated using standard samples of In (purity> 99.999%), Sn (> 99.9%) and benzoic acid (> 99.97%). As a complement of DSC, thermogravimetric analysis (TGA) was performed to study the stability of materials with temperature. A TA Instruments Q500 thermobalance was used. Samples were placed in Pt containers and heated at 5 °C/min under a 100mL/min N2 flow from room temperature to 400 °C. Fourier transform infrared (FTIR) spectroscopy was also used to study the intermolecular interactions between the coformers in the co- crystals. The functional groups in the co-crystal spectrum show differ- ences in the wavenumbers and intensity of the vibrational modes in comparison to those of the two coformers individual spectra. Furthermore, bands characteristic of the synthons established are ob- served. Carbamazepine, saccharin and co-crystal spectra were collected using the spectrum 100 Fourier-transform infrared spectrometer from Perkin Elmer with the universal attenuated total reflection (ATR) sampling accessory at a resolution of 4 cm−1. Samples were measured in the 4000–650 cm−1 range. 3. Results and discussion 3.1. Influence of solvent The first experiments were conducted under the same conditions of pressure (10.0MPa) and temperature (40.0 °C) in order to assess the influence of the solvent. The other conditions used and a summary of the results obtained can be seen in Table 1. Each run was repeated three times to assess reproducibility. The yield shown in Tables 1 and 2 was calculated as an average of values obtained in the three experiments. Each value was estimated using a mass balance. The amounts of co- former injected in the precipitation chamber were calculated taking into account the concentration and flow of solution and the injection time. After precipitation the amount of powder in the precipitation chamber and the waste in the liquid-gas separator were weighed. The PXRD patterns of polymorphs I and II and those of precipitates obtained in runs 1–3 are shown in Fig. 3. The main diffraction peaks of both co-crystal polymorphs are highlighted, diffractions at 2θ=7.0, 14.1, 23.9 and 28.3° are characteristic of polymorph I, whilst the dif- fractions at 2θ=5, 11.5, 12.8, 15 and 25.7° correspond to the poly- morph II. In run 1 using methanol, only representative diffraction peaks of polymorph I are found. In runs 2 and 3, using ethanol and DCM, respectively, a mix of both polymorphs can be observed. Diffraction peaks associated to homocrystals did not appear in the PXRD patterns shown in Fig. 3 for runs 1–3. In comparison to the single crystals that may be obtained through conventional co-crystallization methods, the Table 1 Operational parameters and summary of the results for the 1:1 CBZ-SAC SAS co-crystallization using several solvents at 40.0 °C and 10.0MPa.a Run Solvent CBZ concentration (mg/mL) Yield (%) Polymorph Melting point (°C) Morphology 1 Methanol 30 65 I 174.7 Plate-like 2 Ethanol 15 55 I+ II 163.9; 172.1 Plate-like and needle-like 3 DCM 2.6 40 I+ II 172.1 Plate-like and needle-like 4 DMSO 65 – – – – a In all runs: solution flow rate= 1mL/min; CO2 flow rate: 20 g/min; nozzle diameter 100 μm; SAC concentration in equimolar ratio. Table 2 Operational parameters and summary of results for the 1:1 CBZ-SAC SAS co-crystallization using methanol.a Run T (°C) P (MPa) CO2 densityb (mg/mL) CBZ concentration (mg/mL) Yield (%) Polymorph Melting point (°C) Morphology 1 40.0 10.0 0.63 30 65 I 174.7 Plate-like 5 40.0 15.0 0.78 30 50 I 174.7 Plate-like 6 60.0 15.0 0.60 30 60 I+ II 164.9; 170.0 Needle-like 7 60.0 15.0 0.60 15 45 I+ II 166.0; 171.5 Needle-like a In all runs: Solution flow rate= 1mL/min; CO2 flow rate: 20 g/min; nozzle diameter 100 μm; SAC concentration in equimolar ratio. Each run was repeated three times to assess reproducibility. b CO2 density was taken from Ref. [67]. I.A. Cuadra et al. The Journal of Supercritical Fluids 136 (2018) 60–69 63 powder or microcrystalline samples resulting from SAS precipitation exhibit wider diffraction peaks and often show preferential orientation. This is particularly clear in run 2 using ethanol where the precipitate obtained shows less crystallinity. In run 4 where DMSO was the solvent, the experiment failed to produce any precipitate. A probable enhanced solubility of the co-crystal coformers in the supercritical CO2-DMSO system would explain the lack of precipitate in the precipitation chamber and the clogging formed in the BPR during the experiment. For these reasons, DMSO is not recommended as a solvent in the pre- cipitation of CBZ-SAC co-crystals using the SAS technique. Further characterization of the samples was carried out using dif- ferential scanning calorimetry and thermogravimetric analysis. DSC thermograms of commercial CBZ, SAC and the SAS precipitates are shown in Fig. 4 using the same scale for all thermograms. Melting transitions were measured reading the onset temperature of the peak. Porter et al. [25] carried out a thermogravimetric analysis of both polymorphs of CBZ-SAC co-crystals and found a single nonreversible transition for polymorph I with an onset temperature at 172 °C. For polymorph II a nonreversible transition with an onset at 168 °C was followed by a small non reversible transition with an onset at 172 °C. Threlfall [62] explained the different types of DSC traces which are likely to take place in a dimorphic system. The peak with an onset at 168 °C is characteristic of the less stable polymorph of the CBZ-SAC co- crystal (polymorph II). Precipitates obtained in runs 1–3 show melting temperatures below the melting points of untreated CBZ (189.9 °C) and SAC (227.8 °C). Also, the broad peak associated to the dehydration of CBZ hydrate does not appear in the thermograms [20]. The precipitate obtained using methanol (run 1) melted at 174.7 °C, a value similar to that reported for polymorph I [23,25]. In the SAS precipitate using ethanol (run 2) the transition appearing at 163.9 °C indicates the pre- sence of polymorph II. Contrary to the published behavior for pure polymorph II, in the ethanol precipitate the smaller transition is the first event suggesting the presence of a mixture of polymorphs I and II. As to the DCM precipitate (run 3), the PXRD pattern evidences the presence Fig. 3. PXRD patterns of left: CBZ-SAC co-crystals polymorphs I (bottom) and II (top) simulated from crystallographic information file and experimental patterns for untreated CBZ and SAC; right: experimental patterns from CBZ-SAC co-crystals obtained using methanol (bottom), ethanol (middle) and DCM (top). Fig. 4. DSC thermograms measured at a heating rate of 5 °C/min of untreated CBZ and SAC, SAS processed CBZ and the co-crystals obtained in runs 1, 2 and 3. I.A. Cuadra et al. The Journal of Supercritical Fluids 136 (2018) 60–69 64 of both polymorphs. However, the thermogram of run 3 only exhibits a melting transition at 172.1 °C. This could be due to a lower content of polymorph II in this sample. Overlays of DSC/TGA thermograms for runs 1–3 are included as Supplementary material. These thermograms evidence that mass loss in samples started with the melting transition. In the case of run 1 where the precipitate consisted only in polymorph I, an estimate for the heat of fusion can be obtained from peak integration. This value may be corrected taking into account the mass loss. These values are given in the Supplementary material. The morphology of both polymorphs has already been reported [25]. Polymorph I of the co-crystal presents a plate-like morphology whilst polymorph II exhibits a needle-like one. SEM images of the dif- ferent SAS precipitates are shown in Fig. 5. Here we can observe that the co-crystals obtained using methanol as a solvent (run 1) present mainly a plate-like morphology accompanied by some plate-like ag- glomerates. Crystals exhibit heterogeneous sizes with widths varying from 5 to 10 μm. Runs 2 and 3 present a more noticeable mixture of morphologies. In the precipitate obtained using ethanol as a solvent (run 2), a mixture of agglomerated plate-like and agglomerated needle- like morphologies can be observed. In the one obtained using DCM (run 3) plate-like crystals seems to agglomerate in one direction. The pre- sence of plain plate-like and needle-like morphologies in runs 2 and 3 seems to confirm a mixture of both polymorphs of the CBZ-SAC co- crystal. Particles are much smaller in runs 2 and 3 using ethanol and DCM. Samples obtained were also analyzed using FTIR and results are presented in Fig. 6. As explained in detail in the Introduction, the bonds connecting the coformers to each other in the CBZ-SAC co-crystals change from polymorph I to polymorph II. As a consequence, the ab- sorption bands corresponding to the amide and carboxylic group pre- sent in the carbamazepine in the co-crystal polymorph I will be shifted in the co-crystal polymorph II. The stretching bands of the amide that take place at 3501 cm−1 (asymmetric) and 3434 cm−1 (symmetric) in polymorph I red shift to 3430 cm−1 and 3350 cm−1 in polymorph II. The vibration frequency at 1645 cm−1 in polymorph I, probably due to the asymmetric stretching of the amide carboxylic group, blue shifts to 1668 cm−1 in polymorph II [25]. In the FTIR spectra of the SAS pre- cipitates, we can see that the precipitate obtained using methanol shows the characteristic absorption bands of polymorph I (3500, 3434 and 1645 cm−1). Although the absorption bands obtained using ethanol (3424 cm−1 with shoulder at 3465, 3344 and 1657 cm−1) and DCM (3444, 3336 and 1657 cm−1), exhibit frequencies similar to those of polymorph II, PXRD patterns show the presence of both polymorphs in the precipitates. Only the runs where methanol was the solvent seemed to lead to a pure phase of co-crystal polymorph I. Polymorph I, and mixtures of polymorph I and the CBZ-dihydrate, were the only forms obtained using any of the studied solvents and a 1:1 molar ratio of CBZ and SAC in previous conventional anti-solvent experiments [21]. Polymorph I could also be expected because it has been previously reported that the generation of this polymorph by conventional anti-solvent technique is favoured by higher supersaturation conditions [55], conditions that are characteristic of the SAS technique. An advantage of not using water as an anti-solvent in the SAS process is that none of the precipitates con- tains the less soluble dihydrate form of CBZ. Experiments were per- formed using only supercritical CO2 as anti-solvent. At the studied conditions of temperature, CO2 molar fraction and pressure, ethanol, methanol, DCM and DMSO are fully miscible with CO2 forming one supercritical phase [59,61]. We can therefore expect that the different polymorphs obtained in runs 1–4 should also be a result from the so- lute-solvent interaction. The CBZ-SAC-solvent phase diagram is related to the different solubilities of the co-crystal coformers in the selected solvent. In the case of methanol, the solubilities of both coformers are relatively similar [21] and a congruently saturating system can be ex- pected. In the case of ethanol, DMSO and DCM, solubilities differ in one order of magnitude so a ternary phase diagram of an incongruently saturating system could be expected. In a previous investigation of the naproxen-nicotinamide co-crystal formation using the SAS technique [63] it was pointed out that incongruent saturating systems can lead to Fig. 5. SEM images of CBZ-SAC co-crystals obtained in runs 1, 2 and 3. I.A. Cuadra et al. The Journal of Supercritical Fluids 136 (2018) 60–69 65 the formation of homocrystals. In our case no homocrystals are formed but a mix of both co-crystal polymorphs is obtained as can be observed in the PXRD pattern shown in Fig. 3. 3.2. SAS precipitation of pure CBZ and SAC Fig. 7 shows SEM images of commercial and SAS processed co- formers at 40.0 °C and 10.0MPa using methanol as solvent (conditions of run 1). PXRD patterns of commercial CBZ and SAC are shown in Fig. 3. PXRD patterns of SAS processed CBZ and SAC were also obtained and are given as Supplementary material. DSC thermograms for com- mercial and SAS processed CBZ are shown in Fig. 4. Commercial CBZ consists in a mixture of polymorphs I and III as indicated by the two endothermic peaks at 150.2 and 189.9 °C and the comparison of its PXRD pattern to those of CBZ pure polymorphs. However, the ther- mogram of SAS processed CBZ exhibits one single endothermic peak at 188.7 °C and the SAS processed CBZ PXRD pattern is coincident with that of CBZ polymorph I. Therefore, commercial CBZ changed its polymorphism as a consequence of SAS treatment. CBZ morphology changed from agglomerate particles to needle-like structures. In pre- vious studies performed for pure CBZ using supercritical CO2 and the gas anti-solvent technique (GAS) [64], the starting polymorph III sample changed to mostly polymorph I regardless of the solvent used (acetone, ethylacetate and DCM). Several authors have pointed out the difficulty of producing distinct PXRD pure polymorphic forms of CBZ through SAS or GAS [64–66]. For instance, Padrela et al. [66] obtained a mixture of polymorphs II and III using GAS and starting with poly- morph III. On the other hand, saccharin kept its monoclinic form and its morphology changed from agglomerated particles to fragmented thin plates presenting heterogeneous size. 3.3. Influence of operational conditions In order to assess the influence of operational conditions further experiments with methanol as solvent were carried out varying tem- perature, pressure and coformer concentration. Table 2 shows the conditions studied and a summary of the results obtained. For com- parison purposes, results of run 1 and values for the CO2 density at the temperature and pressure conditions of the experiments are also in- cluded in this table. PXRD and FTIR results for runs 5–7 are shown in Fig. 8. Increasing the pressure in the precipitation chamber did not lead to a polymorphic Fig. 6. FTIR spectra of the CBZ-SAC co-crystals: run 1 (bottom), run 2 (middle) and run 3 (top). Fig. 7. SEM images of untreated CBZ and SAC, and SAS processed CBZ and SAC using conditions of run 1. I.A. Cuadra et al. The Journal of Supercritical Fluids 136 (2018) 60–69 66 change and the same reflections peaks and absorption bands of run 1 were found in the powder obtained in run 5. Increasing temperature though resulted in a mixture of polymorphs in the precipitate obtained in run 6. Reducing the concentration of the feeding solution at high temperature (run 7) produced a significant change in the PXRD pattern. Intensities of the characteristic reflections for the polymorph I de- creased and some reflections of polymorph II increased. The PXRD pattern of run 7 shows nevertheless low crystallinity and conclusions about the polymorphs present and their ratio are difficult. Diffraction peaks associated to homocrystals did not appear in the PXRD patterns shown in Fig. 8 for runs 5–7. FTIR spectra for runs 6 and 7 show ab- sorption bands with frequencies similar to those of polymorph II. DSC curves for precipitates obtained in runs 1, 5, 6 and 7 are shown in Fig. 9. Precipitates obtained in runs 5–7 also show melting temperatures below the melting points of untreated CBZ and SAC and the broad peak associated to the dehydration of CBZ hydrate does not appear [20]. The almost identical curves for runs 1 and 5 indicate the presence of only polymorph I in these precipitates. However, DSC shows the presence of the transitions reported for polymorph II [25] when higher temperatures are used (runs 6 and 7). The fact that a pressure increase of 5MPa (and therefore a supercritical CO2 density increase from 0.63 to 0.78mg/mL) did not change the precipitate polymorphism seems to indicate that the anti-solvent density does not play a significant role in the polymorphic outcome. Nevertheless, a slight decrease in the precipitation yield was observed at the higher pressure. Increasing the temperature however, will influence the solu- bility of both coformers in the selected solvent, and will therefore affect the ternary phase diagram leading to different polymorphs and their mixtures, and could also lead to homocrystal formation. As may be seen in Fig. 9, the temperature increase led to a thermogram where the peaks characteristic of polymorph form II are present [25]. In the case of run 6, similar to run 2, the smaller transition is the first one at 164.9 °C. Reducing the concentration of both crystal coformers in the initial so- lution (run 7) resulted in an area increase of the first nonreversible peak (with onset at 166.0 °C) with respect to that of the second nonreversible peak (onset at 171.5 °C), indicating an increase in the polymorph II to polymorph I ratio. In summary, PXRD patterns and DSC thermograms confirm the presence of polymorph I for run 5 and a mixture of poly- morphs for runs 6 and 7. Overlays of DSC/TGA thermograms for runs 5–7 are included as Supplementary material. These thermograms evidence that mass loss in samples started with the melting transition. In the case of run 5 where the precipitate consisted only in polymorph I, the estimate for the heat of fusion obtained from peak integration and that resulting from taking into account the mass loss are also given in the Supplementary material. Fig. 10 shows SEM images of CBZ-SAC co-crystals obtained in runs 1, 5, 6 and 7. A change in pressure had an influence on the morphology of the SAS precipitate and a higher amount of agglomerate was ob- tained when pressure was increased in run 5 with respect to run 1. A temperature increase in runs 6 and 7 with respect to run 5 resulted in a Fig. 8. Left: PXRD patterns of run 5 (bottom), run 6 (middle) and run 7 (top); Right: FTIR spectra of run 5 (bottom), run 6 (middle) and run 7 (top). Fig. 9. DSC thermograms measured at a heating rate of 5 °C/min of the co-crystals ob- tained in runs 1, 5, 6 and 7. I.A. Cuadra et al. The Journal of Supercritical Fluids 136 (2018) 60–69 67 CO2 density reduction from 0.78 to 0.60mg/mL and a change of the morphology from plate-like to a needle-like agglomerate, thus con- firming the presence of polymorph II in the precipitate that had also been observed in the PXRD patterns, FTIR spectra and thermal analysis. Reduction of the organic solution concentration did not lead to a sig- nificant change in the morphology and SEM images for runs 6 and 7 are similar. Therefore, it may be concluded that changes in CO2 density in the range considered in this study influence the co-crystal morphology. However, these changes do not explain the polymorphic outcome. Neither the untreated or SAS processed crystals of CBZ or SAC were found in any of the co-crystal experiments. The SAS crystals and co- crystals images shown in Figs. 5, 7 and 10 demonstrate the reduction of size and the narrow size distribution obtained through this method. 4. Conclusions Starting from drug concentrations corresponding to the co-crystal stoichiometric composition, the production of 1:1 CBZ-SAC co-crystals using the SAS technique was achieved. SAS co-crystals exhibit the same crystal structure, morphology, FTIR spectra and melting point as those previously obtained by different methodologies. Within the detection limits, precipitates did not show the presence of homocrystals and were solvent free, thus, no further purification steps are required and co- crystals are readily produced in a one-step process with low environ- mental impact. The undesired dihydrate form of CBZ is avoided. Conditions to obtain polymorph I alone are clearly established and yields up to 65% are obtained. Unfortunately, precipitates of pure co- crystal polymorph II could not be obtained at the conditions and sol- vents used in this study. Variation of pressure and organic solution concentration in the system showed little influence in the polymorphic outcome of the precipitates while temperature and solvent selection were shown to be the key factors. These findings are in accordance with other SAS co-crystal investigations [63]. Changes in temperature or solvent will affect the ternary phase diagram and could lead to an in- congruent saturating system, thus producing different precipitate outcomes. The SAS process is therefore a suitable process for production of the pure CBZ-SAC co-crystal polymorph I as it operates at moderate tem- peratures using the green solvent supercritical CO2 avoiding therefore the degradation of the product and limiting the use of organic solvents. Acknowledgments We gratefully acknowledge the financial support of the Spanish Ministry of Economy and Competitiveness (MINECO), research project CTQ2013-41781-P. I.A.C. thanks MINECO for its support through a FPI grant. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.supflu.2018.02.004. References [1] C.C. Sun, Cocrystallization for successful drug delivery, Expert Opin. Drug Deliv. 10 (2013) 201–213. [2] R. Thipparaboina, D. Kumar, R.B. Chavan, N.R. Shastri, Multidrug co-crystals: to- wards the development of effective therapeutic hybrids, Drug Discov. Today 21 (2016) 481–490. [3] R. Thakuria, A. Delori, W. Jones, M.P. Lipert, L. Roy, N. Rodriguez-Hornedo, Pharmaceutical cocrystals and poorly soluble drugs, Int. J. Pharm. 453 (2013) 101–125. [4] EMA, Reflection Paper on the Use of Cocrystals and Other Solid State Forms of Active Substances in Medicinal Products, (2015) http://www.ema.europa.eu/docs/ en_GB/document_library/Scientific_guideline/2015/07/WC500189927.pdf. [5] FDA, Guidance for Industry: Regulatory Classification of Pharmaceutical Co-crys- tals, (2011) https://www.fda.gov/downloads/Drugs/ GuidanceComplianceRegulatoryInformation/Guidance1054s/UCM281764.pdf. [6] D. Douroumis, S.A. Ross, A. Nokhodchi, Advanced methodologies for cocrystal synthesis, Adv. Drug Deliv. Rev. (2017), http://dx.doi.org/10.1016/j.addr.2017.07. 008. [7] J.P. Reboul, B. Cristau, J.C. Soyfer, J.P. Astier, 5H-5-dibenz(b, f)azepinecarbox- amide-5 (carbamazepine), Acta Crystallogr. B: Struct. Sci. 37 (1981) 1844–1848. [8] V.L. Himes, A.D. Mighell, W.H. Decamp, Structure of carbamazepine: 5H-dibenz (b,f)azepine-5-carboxamide, Acta Crystallogr. B: Struct. Sci. 37 (1981) 2242–2245. Fig. 10. SEM images of CBZ-SAC co-crystals obtained in runs 1, 5, 6 and 7. I.A. Cuadra et al. The Journal of Supercritical Fluids 136 (2018) 60–69 68 [9] J.N. Lisgarten, R.A. Palmer, J.W. Saldanha, Crystal and molecular structure of 5- carbamyl-5H-dibenzo(b,f)azepine, J. Crystallogr. Spectrosc. Res. 19 (1989) 641–649. [10] M.M.J. Lowes, M.R. Caira, A.P. Lotter, J.G. Vanderwatt, Physicochemical properties and X-ray structural studies of the trigonal polymorph of carbamazepine, J. Pharm. Sci. 76 (1987) 744–752. [11] M.D. Lang, J.W. Kampf, A.J. Matzger, Form IV of carbamazepine, J. Pharm. Sci. 91 (2002) 1186–1190. [12] A.L. Grzesiak, M.D. Lang, K. Kim, A.J. Matzger, Comparison of the four anhydrous polymorphs of carbamazepine and the crystal structure of form I, J. Pharm. Sci. 92 (2003) 2260–2271. [13] P. Fernandes, K. Shankland, A.J. Florence, N. Shankland, A. Johnston, Solving molecular crystal structures from X-ray powder diffraction data: the challenges posed by gamma-carbamazepine and chlorothiazide N,N,-dimethylformamide (1/2) solvate, J. Pharm. Sci. 96 (2007) 1192–1202. [14] K.S. Eccles, S.P. Stokes, C.A. Daly, N.M. Barry, S.P. McSweeney, D.J. O'Neill, D.M. Kelly, W.B. Jennings, O.M.N. Dhubhghaill, H.A. Moynihan, A.R. Maguire, S.E. Lawrence, Evaluation of the Bruker SMART X2S: crystallography for the non- specialist? J. Appl. Crystallogr. 44 (2011) 213–215. [15] J.B. Arlin, L.S. Price, S.L. Price, A.J. Florence, A strategy for producing predicted polymorphs: catemeric carbamazepine form V, Chem. Commun. 47 (2011) 7074–7076. [16] N. El Hassan, A. Ikni, J.M. Gillet, A. Spasojevic-de Bire, N.E. Ghermani, Electron properties of carbamazepine drug in form III, Cryst. Growth Des. 13 (2013) 2887–2896. [17] E.M. Horstman, S. Goyal, A. Pawate, G. Lee, G.G.Z. Zhang, Y.C. Gong, P.J.A. Kenis, Crystallization optimization of pharmaceutical solid forms with X-ray compatible microfluidic platforms, Cryst. Growth Des. 15 (2015) 1201–1209. [18] I. Sovago, M.J. Gutmann, H.M. Senn, L.H. Thomas, C.C. Wilson, L.J. Farrugia, Electron density, disorder and polymorphism: high-resolution diffraction studies of the highly polymorphic neuralgic drug carbamazepine, Acta Crystallogr. B: Struct. Sci. Cryst. Eng. Mater. 72 (2016) 39–50. [19] E. van Genderen, M.T.B. Clabbers, P.P. Das, A. Stewart, I. Nederlof, K.C. Barentsen, Q. Portillo, N.S. Pannu, S. Nicolopoulos, T. Gruene, J.P. Abrahams, Ab initio structure determination of nanocrystals of organic pharmaceutical compounds by electron diffraction at room temperature using a Timepix quantum area direct electron detector, Acta Crystallogr. A: Found. Adv. 72 (2016) 236–242. [20] Y. Kobayashi, S. Ito, S. Itai, K. Yamamoto, Physicochemical properties and bioa- vailability of carbamazepine polymorphs and dihydrate, Int. J. Pharm. 193 (2000) 137–146. [21] S.K. Pagire, S.A. Korde, B.R. Whiteside, J. Kendrick, A. Paradkar, Spherical crys- tallization of carbamazepine/saccharin co-crystals: selective agglomeration and purification through surface interactions, Cryst. Growth Des. 13 (2013) 4162–4167. [22] J.C.J. Bart, Crystal and molecular structure of saccharin (o-sulphobenzoicimide), J. Chem. Soc. B: Phys. Org. (1968) 376–382. [23] S.G. Fleischman, S.S. Kuduva, J.A. McMahon, B. Moulton, R.D.B. Walsh, N. Rodriguez-Hornedo, M.J. Zaworotko, Crystal engineering of the composition of pharmaceutical phases: multiple-component crystalline solids involving carbama- zepine, Cryst. Growth Des. 3 (2003) 909–919. [24] M.A. Oliveira, M.L. Peterson, R.J. Davey, Relative enthalpy of formation for co- crystals of small organic molecules, Cryst. Growth Des. 11 (2011) 449–457. [25] W.W. Porter III, S.C. Elie, A.J. Matzger, Polymorphism in carbamazepine cocrystals, Cryst. Growth Des. 8 (2008) 14–16. [26] I. Huskic, J.C. Christopherson, K. Uzarevic, T. Friscic, In situ monitoring of vapour- induced assembly of pharmaceutical cocrystals using a benchtop powder X-ray diffractometer, Chem. Commun. 52 (2016) 5120–5123. [27] C. Maheshwari, A. Jayasankar, N.A. Khan, G.E. Amidon, N. Rodriguez-Hornedo, Factors that influence the spontaneous formation of pharmaceutical cocrystals by simply mixing solid reactants, CrystEngComm 11 (2009) 493–500. [28] H.G. Brittain, Cocrystal systems of pharmaceutical interest: 2011, Cryst. Growth Des. 12 (2012) 5823–5832. [29] R.A. Chiarella, R.J. Davey, M.L. Peterson, Making co-crystals – the utility of ternary phase diagrams, Cryst. Growth Des. 7 (2007) 1223–1226. [30] J.H. ter Horst, M.A. Deij, P.W. Cains, Discovering new co-crystals, Cryst. Growth Des. 9 (2009) 1531–1537. [31] C. Pando, A. Cabañas, I.A. Cuadra, Preparation of pharmaceutical co-crystals through sustainable processes using supercritical carbon dioxide: a review, RSC Adv. 6 (2016) 71134–71150. [32] H. Machida, M. Takesue, R.L. Smith, Green chemical processes with supercritical fluids: properties, materials, separations and energy, J. Supercrit. Fluids 60 (2011) 2–15. [33] L. Padrela, M.A. Rodrigues, S.P. Velaga, A.C. Fernandes, H.A. Matos, E.G. de Azevedo, Screening for pharmaceutical cocrystals using the supercritical fluid en- hanced atomization process, J. Supercrit. Fluids 53 (2010) 156–164. [34] L. Padrela, M.A. Rodrigues, J. Tiago, S.P. Velaga, H.A. Matos, E.G. de Azevedo, Insight into the mechanisms of cocrystallization of pharmaceuticals in supercritical solvents, Cryst. Growth Des. 15 (2015) 3175–3181. [35] E. Reverchon, I. De Marco, Mechanisms controlling supercritical antisolvent pre- cipitate morphology, Chem. Eng. J. 169 (2011) 358–370. [36] G.W. Brun, A. Martin, E. Cassel, R.M.F. Vargas, M.J. Cocero, Crystallization of caffeine by supercritical antisolvent (SAS) process: analysis of process parameters and control of polymorphism, Cryst. Growth Des. 12 (2012) 1943–1951. [37] N. Esfandiari, Production of micro and nano particles of pharmaceutical by su- percritical carbon dioxide, J. Supercrit. Fluids 100 (2015) 129–141. [38] R. Campardelli, E. Reverchon, I. De Marco, Dependence of SAS particle morphol- ogies on the ternary phase equilibria, J. Supercrit. Fluids 130 (2017) 273–281. [39] F. Zahran, A. Cabañas, J.A.R. Cheda, J.A.R. Renuncio, C. Pando, Dissolution rate enhancement of the anti-inflammatory drug diflunisal by coprecipitation with a biocompatible polymer using carbon dioxide as a supercritical fluid antisolvent, J. Supercrit. Fluids 88 (2014) 56–65. [40] I.A. Cuadra, A. Cabañas, J.A.R. Cheda, F.J. Martinez-Casado, C. Pando, Pharmaceutical co-crystals of the anti-inflammatory drug diflunisal and nicotina- mide obtained using supercritical CO2 as an antisolvent, J. CO2 Util. 13 (2016) 29–37. [41] A. Jayasankar, A. Somwangthanaroj, Z.J. Shao, N. Rodriguez-Hornedo, Cocrystal formation during cogrinding and storage is mediated by amorphous phase, Pharm. Res. 23 (2006) 2381–2392. [42] A. Jayasankar, D.J. Good, N. Rodriguez-Hornedo, Mechanisms by which moisture generates cocrystals, Mol. Pharm. 4 (2007) 360–372. [43] S.R. Bysouth, J.A. Bis, D. Igo, Cocrystallization via planetary milling: enhancing throughput of solid-state screening methods, Int. J. Pharm. 411 (2011) 169–171. [44] M.B. Hickey, M.L. Peterson, L.A. Scoppettuolo, S.L. Morrisette, A. Vetter, H. Guzman, J.F. Remenar, Z. Zhang, M.D. Tawa, S. Haley, M.J. Zaworotko, O. Almarsson, Performance comparison of a co-crystal of carbamazepine with marketed product, Eur. J. Pharm. Biopharm. 67 (2007) 112–119. [45] Z. Rahman, R. Samy, V.A. Sayeed, M.A. Khan, Physicochemical and mechanical properties of carbamazepine cocrystals with saccharin, Pharm. Dev. Technol. 17 (2012) 457–465. [46] G. Di Profio, V. Grosso, A. Caridi, R. Caliandro, A. Guagliardi, G. Chita, E. Curcio, E. Drioli, Direct production of carbamazepine-saccharin cocrystals from water/ ethanol solvent mixtures by membrane-based crystallization technology, CrystEngComm 13 (2011) 5670–5673. [47] A. Caridi, G. Di Profio, R. Caliandro, A. Guagliardi, E. Curcio, E. Drioli, Selecting the desired solid form by membrane crystallizers: crystals or cocrystals, Cryst. Growth Des. 12 (2012) 4349–4356. [48] A. Alhalaweh, L. Roy, N. Rodriguez-Hornedo, S.P. Velaga, pH-dependent solubility of indomethacin-saccharin and carbamazepine-saccharin cocrystals in aqueous media, Mol. Pharm. 9 (2012) 2605–2612. [49] I. Tomaszewska, S. Karki, J. Shur, R. Price, N. Fotaki, Pharmaceutical character- isation and evaluation of cocrystals: importance of in vitro dissolution conditions and type of coformer, Int. J. Pharm. 453 (2013) 380–388. [50] H. Moradiya, M.T. Islam, G.R. Woollam, I.J. Slipper, S. Halsey, M.J. Snowden, D. Douroumis, Continuous cocrystallization for dissolution rate optimization of a poorly water-soluble drug, Cryst. Growth Des. 14 (2014) 189–198. [51] S. Kudo, H. Takiyama, Production method of carbamazepine/saccharin cocrystal particles by using two solution mixing based on the ternary phase diagram, J. Cryst. Growth 392 (2014) 87–91. [52] M. Nishimaru, S. Kudo, H. Takiyama, Cocrystal production method reducing de- position risk of undesired single component crystals in anti-solvent cocrystalliza- tion, J. Ind. Eng. Chem. 36 (2016) 40–43. [53] T. Rager, R. Hilfiker, Cocrystal formation from solvent mixtures, Cryst. Growth Des. 10 (2010) 3237–3241. [54] I.-C. Wang, M.-J. Lee, S.-J. Sim, W.-S. Kim, N.-H. Chun, G.J. Choi, Anti-solvent co- crystallization of carbamazepine and saccharin, Int. J. Pharm. 450 (2013) 311–322. [55] M.J. Lee, I.C. Wang, M.J. Kim, P. Kim, K.H. Song, N.H. Chun, H.G. Park, G.J. Choi, Controlling the polymorphism of carbamazepine-saccharin cocrystals formed during antisolvent cocrystallization using kinetic parameters, Korean J. Chem. Eng. 32 (2015) 1910–1917. [56] P. Subra-Paternault, C. Roy, D. Vrel, A. Vega-Gonzalez, C. Domingo, Solvent effect on tolbutamide crystallization induced by compressed CO2 as antisolvent, J. Cryst. Growth 309 (2007) 76–85. [57] R. Bettini, L. Bonassi, V. Castoro, A. Rossi, L. Zema, A. Gazzaniga, F. Giordano, Solubility and conversion of carbamazepine polymorphs in supercritical carbon dioxide, Eur. J. Pharm. Sci. 13 (2001) 281–286. [58] D. Bolten, M. Türk, Micronisation of carbamazepine through rapid expansion of supercritical solution (RESS), J. Supercrit. Fluids 66 (2012) 389–397. [59] S.N. Joung, C.W. Yoo, H.Y. Shin, S.Y. Kim, K.P. Yoo, C.S. Lee, W.S. Huh, Measurements and correlation of high-pressure VLE of binary CO2-alcohol systems (methanol ethanol, 2-methoxyethanol and 2-ethoxyethanol), Fluid Phase Equilib. 185 (2001) 219–230. [60] C.J. Chang, C.Y. Day, C.M. Ko, K.L. Chiu, Densities and P-x-y diagrams for carbon dioxide dissolution in methanol, ethanol, and acetone mixtures, Fluid Phase Equilib. 131 (1997) 243–258. [61] A. Vega-Gonzalez, R. Tufeu, P. Subra, High-pressure vapor-liquid equilibrium for the binary systems carbon dioxide plus dimethyl sulfoxide and carbon dioxide plus dichloromethane, J. Chem. Eng. Data 47 (2002) 492–495. [62] T.L. Threlfall, Turning DSC charts of polymorphs into phase diagrams: a tutorial paper, Org. Process Res. Dev. 13 (2009) 1224–1230. [63] C. Neurohr, A. Erriguible, S. Laugier, P. Subra-Paternault, Challenge of the super- critical antisolvent technique SAS to prepare cocrystal-pure powders of naproxen- nicotinamide, Chem. Eng. J. 303 (2016) 238–251. [64] M. Moneghini, I. Kikic, D. Voinovich, B. Perissutti, P. Alessi, A. Cortesi, F. Princivalle, D. Solinas, Study of the solid state of carbamazepine after processing with gas anti-solvent technique, Eur. J. Pharm. Biopharm. 56 (2003) 281–289. [65] D. Meng, J. Falconer, K. Krauel-Goellner, J. Chen, M. Farid, R.G. Alany, Self-built supercritical CO2 anti-solvent unit design, construction and operation using car- bamazepine, AAPS PharmSciTech 9 (2008) 944–952. [66] L. Padrela, J. Zeglinski, K.M. Ryan, Insight into the role of additives in controlling polymorphic outcome: a CO2-antisolvent crystallization process of carbamazepine, Cryst. Growth Des. 17 (2017) 4544–4553. [67] NIST Standard Reference Database Number 69, which can be accessed electro- nically through the NIST Chemistry Web Book (http://webbook.nist.gov/ chemistry/). I.A. Cuadra et al. The Journal of Supercritical Fluids 136 (2018) 60–69 69