Supercritical Fluid Crystallization of Adipic Acid Using Urea as Habit

Supercritical Fluid Crystallization of Adipic Acid Using Urea as Habit Modifier Giuseppe Caputo,*,† Renata Adami,†,‡ and Ernesto Reverchon† Dipartimento di Ingegneria Chimica e Alimentare, UniVersita` di Salerno, Salerno I-84084, Italy, and Department of Biochemistry & Food Chemistry, UniVersity of Turku, FIN-20014 Turku, Finland

CRYSTAL GROWTH & DESIGN 2008 VOL. 8, NO. 8 2707–2715

ReceiVed August 9, 2007; ReVised Manuscript ReceiVed May 13, 2008

ABSTRACT: The crystal morphology of adipic acid mediated by the action of urea as additive has been investigated using the supercritical antisolvent precipitation (SAS). The process was performed using acetone and carbon dioxide as solvent and antisolvent, respectively. The effect of urea concentration in the liquid solution and of precipitation pressure on the crystal habit was investigated; the products were characterized by scanning electron microscopy (SEM), X-ray powder diffraction (XRPD), differential scanning calorimetry (DSC), and infrared spectrometry (FT-IR). Urea was found to be very effective in modifying the crystal habit of adipic acid from needle-like shape with a length of several hundreds of micrometers to prismatic shape. Operating at 40 °C, 12 MPa, and a urea/adipic acid mass ratio of 12.5% prismatic crystals with a mean size of about 7.5 µm and a very narrow crystal size distribution were obtained. The habit modification was attributed to the preferential adsorption of urea to the fastest growing crystal face of the needle crystal, which inhibited the crystal growth along the corresponding direction. Introduction The manipulation of polymorphism, morphology, and size of crystals of micro- or nanometric dimensions is a relevant industrial challenge due to their strong influence on materials properties. Crystal morphology, or habit, can influence many downstream relevant properties, such as filterability, flowability, compactability, and dissolution time.1 The proper crystal habit and size are essential for manufacturing processes such as powder mixing and filling. For this reason, there is an increasing interest in the specialty chemicals industries to develop technologies and methods that can produce crystals at micro- and nanoscale levels with controlled properties: crystal size (CS), polymorphism, and habit. Crystallization is most commonly employed as the final step in the preparation of products with desirable crystal form and dimension. However, some products very often crystallize in acicular or needle shape, thus producing undesirable flow characteristics, low bulk density, caking, and difficulties in packaging and handling of the material.2,3 To avoid these problems, milling or crushing are often employed to eliminate the undesired shape of the particles and to reduce CS. However, milling may generate thermal degradation, contamination, differences among different product batches, and increase of the amorphous content that reduce the quality of the product. Alternative efficient processes that use supercritical fluids (SCFs) as processing medium for the production of micro- and nanosized particles have emerged.4–6 These processes try to take advantage of some specific properties of gases at supercritical conditions, such as enhanced solvent power, large diffusivities, and tunability of process conditions. SCFs are characterized by a continuously adjustable solvent power obtained by varying pressure and temperature. Moreover, their diffusivities can be about 2 orders of magnitude greater than those of liquids. As a result, SCF-based processes are characterized by fast mass transfer that allows operation at very high levels of supersaturation that cannot be obtained using conventional methods. Various SCF-based processes for microparticle generation have been developed to exploit these properties.4–6 Among them, † ‡

Universita` di Salerno. University of Turku.

semicontinous supercritical antisolvent (SAS) precipitation has emerged and been successfully applied to many organic and inorganic substances.5,7,8 The mean particle size that can be obtained by SAS ranges from about 0.05 µm to tens of micrometers and different morphologies can be produced depending on the processed material. It is a very rapid process compared with the traditional liquid-antisolvent crystallization, and for this reason, amorphous spherical particles are a common habit, and they are frequently the scope of the process.8 However, some substances can form crystals even at high pressure and large CO2 concentration with respect to the liquid solvent. In these cases, unfortunately, the size of the product is hard to control and large CS and wide distributions (CSDs) are usually obtained. A general processing strategy that can be used to modify CS and crystal habit is the use of habit modifiers to selectively change crystal growth kinetics during crystallization.2,3 This method is advantageous because crystal habit can be modified without onerous postprocessing operations. It is well-known that nucleation, growth, polymorphism, and habit of crystals can be significantly altered by the presence of low concentrations of additives purposely added to alter the crystallization process.2,3 Additives can reduce crystal growth rate and alter morphology by binding to crystal face and interfering with the propagation steps.9 In some cases, the additives substitute in the crystal lattice and disrupt the binding sequence.10,11 In either case, however, if the additive selectively interacts with a particular crystal face, it will slow only the growth rate of that face and alter the morphology of the crystal. The control of crystal morphology by the use of additives is a subject of great interest, which has been studied experimentally and through molecular modeling.11 Until now, the use of additives as habit modifiers during supercritical fluid crystallization has been studied in a very reduced number of works. In one of these works, the authors discussed the crystallization of chlorpropamide in the presence of urea as polymorph conversion additive using the rapid expansion of SC solution (RESS) process.12 Recently, Jarmer et al.13 proposed the use of poly(sebacic anhydride) as growth inhibitor of griseofulvin using a supercritical antisolvent process. They obtained a change of habit from acicular to bipyramidal due to a selective inhibition of crystal growth, but bimodal CSDs

10.1021/cg700753u CCC: $40.75  2008 American Chemical Society Published on Web 07/19/2008

2708 Crystal Growth & Design, Vol. 8, No. 8, 2008

and griseofulvin crystals of several micrometers in mean size were obtained. In a recent work, Caputo et al.14 proposed the use of urea as habit modifier of sulfathiazole precipitated by SAS from an acetone solution. In the presence of urea, sulfathiazole habit changed from a plate shape to spherical semicrystalline particles with a very narrow particle size distribution and with a mean particles size ranging between 0.5 and 1.0 µm depending on the sulfathiazole concentration. Urea was also used by Caputo et al.15 to change the crystal morphology of chlorpropamide precipitated from ethylacetate. Therefore, the use of a crystallization modifier in the SAS process can be an intriguing strategy to manage properties of materials whose precipitation is difficult to control. This potentiality should be exploited in more detail, and new data are necessary. Adipic acid is a dicarboxylic acid used in plastic processing (production of nylon 6,6), as excipient in drugs (tablet lubricant, constituent of tablet coatings), and in food production (as flavoring, leavening, and pH control agent). Adipic acid imparts a slowly developing smooth, mild taste in supplementing foods with delicate flavors. It is practically nonhygroscopic, a property that has the advantage of prolonging the shelf life of powdered products (e.g., drinks, cake mixes, instant pudding) in which it is incorporated. In this work, adipic acid has been selected as model compound due to its capability to form hydrogen bonds with urea. Urea is an end product of human protein metabolism and possesses several features that make it an attractive additive: it is nontoxic, biodegradable, and odor-free and can be generally considered to be safe for the human body. Urea was selected as a crystal modifier for its molecular structure, selective interactions with adipic acid molecules, and solubility in acetone, which was used as solvent in this work. The scope of this work has been the SAS precipitation of adipic acid from acetone in the presence of urea used as an additive to control crystal habit and dimensions of crystals. The effects of the main process parameters on powder properties were investigated, and crystals were extensively analyzed using various techniques such as scanning electron microscopy (SEM), differential scanning calorimetry (DSC), infrared (FT-IR) spectroscopy, and X-ray powder diffraction (XRPD). Materials and Analytical Methods Adipic acid [(CH2)4(COOH)2] and urea [CO(NH2)2], purity 99.5%, were purchased from Sigma-Aldrich (Italy). Acetone (purity 99.8%) was purchased from Carlo Erba Reagenti (Italy). CO2 (purity 98%) was purchased from SON (Italy). All products were used as received. The solubility of adipic acid in acetone was measured and resulted at 20 °C in about 1.3 wt %. Untreated adipic acid was composed of boulder-like crystals with irregular shapes and a large CSD, ranging between 20 and 300 µm. Physical mixtures of adipic acid and urea were prepared by mixing in a mortar for 5 min the required amount of adipic acid and urea, namely, 6.25, 12.5, 25.0 and 27.5 wt % of urea. These mixtures were used to compare solid state properties with that of the precipitated samples. The following solid state analytical techniques were used. Scanning Electron Microscopy (SEM). A LEO 420 (USA) microscope was used for the analysis. Powders were dispersed on a carbon tab previously stuck to an aluminum stub. Samples were coated with gold-palladium (layer thickness 250 Å) using a sputter coater (Agar Scientific, UK). Several SEM images from different parts of the precipitation vessel were taken for each run to verify the powder uniformity. Particle size and particle size distribution (PS, PSD) were evaluated from SEM images using the Sigma Scan Pro 5 Software (Systat Software Inc., USA). Depending on the sample, about 200-400 crystals were considered in each size distribution calculation. Histograms representing the crystal size distribution were best fitted using Microcal Origin 7.0 Software (Microcal Software Inc., USA).

Caputo et al. X-ray Powder Diffraction (XRPD). A model D8 Discover, Bruker (USA), powder diffractometer with a Cu sealed tube source was used. To obtain the XRPD pattern of selected samples, powders were placed in the holder and flattened with a glass slide to ensure a good surface texture. The measuring conditions were Ni-filtered Cu KR radiation, λ ) 1.54 Å, 2θ angle ranging from 5° to 45° with a scan rate of 2 s/step and a step size of 0.05°. Differential Scanning Calorimetry (DSC). Calorimetric analysis has been performed using a DSC Mettler TC11 (USA). Temperature and enthalpy of fusion were calibrated with pure indium standard (melting point 156.6 °C). Powder samples (5 ( 0.5 mg) were accurately weighed, crimped in an aluminum pan, and heated from 25 to 170 °C at 10 °C/min under a nitrogen purge. Melting points were calculated using the software provided with the instrument. Fourier Transform Infrared (FT-IR) Spectroscopy. Spectra were obtained via M2000 FTIR (MIDAC Co) at a resolution of 0.5 cm-1. The scan wavenumber range was 4000-400 cm-1, and 16 scan signals were averaged to reduce the noise. The samples were ground and mixed thoroughly with potassium bromide as infrared transparent matrix. KBr discs were prepared by compressing the powders under force in a hydraulic press.

Experimental Apparatus and Procedures A schematic representation of the SAS apparatus is reported in Figure 1. It consists of a modified HPLC pump (P1) equipped with a pulse dampener used to feed the liquid solution and a diaphragm high-pressure pump (P2) used to deliver carbon dioxide. A windowed vessel (P) of 0.375 dm3 with two quartz windows put along the longitudinal section was used as the precipitation chamber. It allowed visualization of the kind and number of fluid phases of the mixture formed during precipitation and following of the macroscopic evolution of the process from the liquid jet breakup to the deposition of precipitated particles. A video camera placed in front of one window allowed recording of the experiments. The acetone mixture with an adipic acid concentration of 1.26 wt % was sprayed in the precipitator through a thin wall stainless steel nozzle (60 µm in diameter, 800 µm in length) and SC CO2 was pumped through another inlet port located on the top of the chamber. CO2 was heated (H) to the process temperature before entering the precipitator. CO2 and acetone were demixed by a separator (S). A stainless steel sintered metal disk was put at the bottom of the chamber to collect the solid product, allowing the CO2-organic solvent solution to pass through. A typical SAS experiment started pumping pure CO2 into the precipitator. When antisolvent steady flow was established and the desired pressure was reached, pure liquid solvent was sent through the nozzle to the precipitation chamber at a flow rate of 1 mL/min in cocurrent mode with supercritical CO2, which flow rate was regulated to obtain a CO2 mole fraction of 0.99 inside the precipitator. The temperature of the precipitator was set at 40 °C in all the experiments. The pressure was set in the range 80-150 bar, as reported in Table 1. Then, the flow of pure acetone was stopped, and an acetone solution at 1 mL/min flow rate with adipic acid concentration of 1.26 wt % and urea concentration in the range 0-0.32 wt % was delivered. The experiment ended when the delivery of the liquid solution to the chamber was interrupted. However, supercritical CO2 continued to flow for an additional 30 min to wash out the residual content of acetone contained in the chamber. Precipitation experiments were repeated twice.

Results and Discussion Effect of Process Conditions on the Crystal Habit of Adipic Acid. Adipic acid normally crystallizes from aqueous solutions as flat, slightly elongated, hexagonal, monoclinic plates. According to Michaels and Colville,16 the predominant face of these plates is the (001) face (C), the elongated side faces are (010) faces (B), and the end faces are (110) faces (A) as shown in Figure 2. The linear, six-carbon dicarboxylic acid molecules are aligned end-to-end in a parallel array in the crystal, with their long axes parallel to the B face; therefore the C face is made up entirely of carboxyl groups, whereas the B

Supercritical Fluid Crystallization of Adipic Acid

Crystal Growth & Design, Vol. 8, No. 8, 2008 2709

Figure 1. Scheme of the apparatus used for SAS experiments. V1 ) solution vessel, V2 ) CO2 vessel, P1 ) pump for the solution, P2 ) pump for CO2, P ) precipitator, S ) separator, FM ) flow meter, C ) chiller, H ) heating bath. Table 1. Crystal Morphology and Size of Adipic Acid Particles Produced by SAS at Different Operating Conditionsa

a

expt

T, °C

P, bar

R7 M9 R6 M8 M4 M11 M7 M3 M10

40 40 40 40 40 40 40 40 40

80 120 150 120 120 120 120 80 90

Curea, wt %

0.08 0.16 0.24 0.32 0.16 0.16

urea/adipic acid wt %

morphology

crystal size, µm

6.25 12.5 18.5 25.0 12.5 12.5

needle-like needle-like needle-like needle-like tabular needle-like + tabular needle-like balloon-like tabular + balloon-like

225 250 220 1-20 7.5 (mode size) 10-100 >100 4.0 (mode size) >100

Cadipic acid, wt % ) 1.26 in acetone; CO2 molar fraction ) 0.99.

Figure 2. The morphology, faces, and orientation of the molecules with respect to the faces of adipic acid crystals according to ref 16. The linear six-carbon chains are perpendicular to the C face so that the face is made up entirely of carboxylic groups.

and A faces contains both carboxylic and hydrocarbon portions of the molecules. In this work, the SAS precipitation of pure adipic acid from 1.26 wt % acetone solutions was first performed at 40 °C and

80, 120, and 150 bar, as reported in Table 1 (experiments R7, M9, and R6). Figure 3a shows a representative SEM micrograph of adipic acid crystallized at 120 bar. As shown in the figure, at these conditions adipic acid crystallizes as long needles. The CSD is also reported in Figure 3b. The main axis of needles has been chosen as characteristic length, and about 200 crystals have been measured. The crystal population is unimodal; crystal sizes are between 100 and 750 µm with a mode of about 250 µm. The widths of the crystals were approximately an order of magnitude smaller than the lengths. Experiments at 80 and 150 bar lead to the same needle-like habit with a mode of 225 and 220, respectively. Operating SAS precipitation at 40 °C and pressure in the range 100-150 bar, nanoparticles formation has been frequently observed,18 but the formation of large crystals with a limited possibility of controlling their size has also been reported by several authors for some other compounds.18,19 A possible explanation of the formation of large crystals of adipic acid at these conditions of temperature and pressure can be found observing the precipitation process in a quartz windowed vessel. A frame of the video recorded during precipitation is reported in Figure 4. When the liquid solution was injected, a short liquid jet was formed at the nozzle exit; that is, the liquid practically


Caputo et al.

Figure 3. SEM image (a) of adipic acid crystals obtained by SAS precipitation at 40 °C, 120 bar, and 1.26 wt % of adipic acid in acetone (expt M9) and the corresponding CSD (b).

occur, and crystal dimensions are mainly controlled by the residence time of the solute in the precipitator. It is worth noting that precipitation from expanded fluid phase is different from the one that produces nanoparticles from supercritical solutions, which is basically a gas-phase nucleation.17

Figure 4. Acetone/adipic acid solution injected at 40 °C and 120 bar in SC CO2. The liquid jet rapidly dissolves forming an expanded fluid phase in which adipic acid crystals are formed. Table 2. Crystal Size of Adipic Acid Particles Produced by SAS at Different Residence Timesa expt

T, °C

P, bar

residence time, min

length, µm

thickness, µm

R15 R13 R14

40 40 40

120 120 120

5 20 60

200 300 500

22.0 23.0 41.0

a

Cadipic

acid,

wt % ) 1.26 in acetone; CO2 molar fraction ) 0.99.

disappeared immediately below the nozzle tip. As shown in the figure, a clear solution filled completely the vessel, and no phase transitions were observed during the whole experiment. Only density fluctuation during liquid spraying was visible. After about five minutes from the start of the injection, the formation of large adipic acid crystals was observed mainly on the precipitator walls. A possible explanation of this fact is that, due to the partial solubility of adipic acid in the mixture SC CO2-acetone, adipic acid is partly solubilized in the fluid phase until supersaturation occurs from the expanded fluid phase (Figure 4). The partially solubility of adipic acid is confirmed by the observation that a quite large part of adipic acid is lost in the fluid phase. Indeed, the crystal yield ranges between 35% and 60% depending on the density of the fluid phase. The effect of residence time of adipic acid in the precipitator has been evaluated by performing experiments at residence times of 5, 20, and 60 min by changing the amount of the solution injected and taking constant all other parameters. As reported in Table 2, both crystal length and thickness increase to a large extent with residence time. Thus, liquid-like nucleation and growth within an expanded ternary solution acetone-CO2-adipic acid

Effect of the Presence of Urea. The needle-like habit of adipic acid is not unusual, and it was also observed by Colville et al. when cooling crystallization was carried out in the presence of a proper anionic surfactant.16 Surfactant can be physically adsorbed onto the A and B faces more strongly than onto the C face, causing retardation of the growth rates of the A and B faces. This behavior is related to the surfactant concentrations, and at higher concentration, extreme habit modifications were observed: the crystals becoming long thin rods or needles. Thus, the decrease in the growth rate of individual faces due to the effect of the additive is probably solely responsible for the habit modifications of crystals. On the other hand, using a cationic surfactant, chemiadsorbed most strongly onto the C face (which has the highest density of hydroxyl groups among the expressed faces of adipic acid), Colville et al. observed that the growth rate of the C face decreases to the greatest extent.16 Higher concentration of cationic surfactant also caused extreme habit modification, the crystals becoming very thin plates or flakes. Thus, a possible way to control adipic acid crystal size and to selectively modify crystal habit is to add a crystallization modifier, that is, an additive with the scope of growth inhibitor of the crystallization process. Therefore, in this work urea has been chosen because of its capability to form hydrogen bonds with carboxylic groups of adipic acid. Adipic acid was SASprecipitated from acetone at 40 °C and 120 bar and at different values of urea concentration, namely, 0.08, 0.16, 0.24, and 0.32 wt % (expt M8, M4, M11, and M7), corresponding to urea weight percentage ratio of 6.25, 12.5, 18.5, and 25.0 wt % with respect to adipic acid. Figure 5a shows the SEM micrograph of the crystals generated at the lowest urea concentration, that is, 0.08 wt %, where the crystal shape is acicular with sizes in the range 20-40 µm. The dimensions of crystals is about an order of magnitude smaller than that of crystals obtained at the same operating conditions without urea (Figure 3). When the urea concentration was increased to 0.16 wt %, prismatic crystals were obtained (Figure 5b) with a smaller crystal size. In Figure 6, the size distribution of these crystals is reported. The diagonal of prisms have been chosen as characteristic length and about 450 crystals have been measured. As shown in the figure, the crystal population is unimodal, and crystal sizes are between 2.5 and 30 µm with a mode of about 7.5 µm. The width and depth of prisms were of the same order of magnitude as the main



Figure 5. SEM images of adipic acid produced during SAS at 40 °C, 120 bar, in the presence of urea at (a) 0.08, (b) 0.16, (c) 0.24, and (d) 0.32 wt %.

Figure 6. CSD of adipic acid precipitated by SAS process at 40 °C, 120 bar, in the presence of urea at 0.16 wt % (expt M4).

diagonal. Thus, an effective change of crystal habit and size was obtained. When the urea concentration was increased to 0.24 wt %, two different crystal populations were obtained, one with a prismatic shape and dimensions of about 20 µm and one with an acicular shape and dimensions of several micrometers (Figure 5c). Thus, in this case, urea is only partially able to block the crystal growth. Finally, at the highest level of urea concentration (0.32 wt %), the crystal population is composed only by needles several micrometers long up to 1 mm (Figure 5d). Therefore, it is evident that the concentration of urea in the adipic acid solution plays a definite role in the precipitation mechanism. A concentration of 0.16 wt % seems an optimum value below and above which the modification effect is largely reduced. This behavior is not new and unexpected, since some studies report growth enhancement occurring at low impurity levels followed by a reversal at higher levels when the interactions between additives molecules become dominant with respect to the interaction between additive and solute molecules.2 Similar results were also observed in a previous14 work concerning the SAS precipitation of sulfathiazole in the presence of urea. The effect of urea is likely due to the formation of hydrogen bonds between urea and adipic acid. The mechanism responsible for the habit modification can be a selective interaction of urea on C faces of adipic acid crystals promoted by hydrogen bonds, which reduces the growth of these faces with respect to A and

Figure 7. X-ray powder diffraction (XRPD) spectra of adipic acid precipitated by SAS at 40 °C and 120 bar: (a) no urea; (b) 0.08 wt % urea; (c) 0.16 wt % urea; (d) 0.24 wt % urea; (e) 0.32 wt % urea; (f) pure urea. A1, A2, and A3 are characteristics peaks of adipic acid, P is a characteristic peak of adipic acid precipitated in the presence of urea, and U is a characteristic peak of pure urea.

B faces, producing a prismatic habit with comparable dimensions of all edges. SAS processed powders were first characterized by XRPD to determine whether there were any significant changes of crystal structure due to specific interactions between adipic acid and urea within the crystals. Figure 7 shows the changes in XRPD patterns of adipic acid after SAS crystallization in the presence of urea at various concentrations. For comparison purposes, the patterns of pure urea and pure adipic acid were also reported. The XRPD pattern of pure adipic acid is essentially the same as that obtained by SAS without urea (data not shown). Adipic acid exhibits three main characteristic peaks at 21.4°, 25.8°, and 31.1° 2θ, indicated in Figure 7 as A1, A2, and A3, respectively. Pure urea (pattern f) exhibits one characteristic peak at U ) 22.3° 2θ. XRPD patterns of adipic acid processed with urea at various concentrations are indicated with letters b, c, d, and e. A1, A2, and A3 characteristic peaks for adipic acid were observed in all spectra, although the peak intensity is reduced with respect to those in the pattern a when the urea concentration increases. The characteristic peak of urea (U) was not observed in the diffraction patterns of adipic acid except in samples obtained with the higher concentration of urea (0.32 wt %), where a very low intensity peak is present (pattern e). Thus, when adipic acid is processed in the presence of urea


Caputo et al.

Figure 8. X-ray powder diffraction (XRPD) spectra of adipic acid/ urea physical mixtures at (a) 25 and (b) 12.5 wt % urea.

between 0.08 and 0.24 wt %, no urea residue can be detected by XRPD. Also, no shifts in the adipic acid peak positions (A1, A2, and A3) were observed, which demonstrated that urea was not incorporated into the adipic acid crystal structure.13,21 Indeed, a significant and systematic shift in the diffraction peak is expected when an additive is incorporated into the parent crystal structure.2,13 Furthermore, this spectra provides support for the crystal growth inhibition mechanism played by urea, because growth-inhibiting additives that operate by a selective adsorption mechanism typically do not alter the crystal structure. However, a peak at 27.6° 2θ (P) was observed in the diffraction patterns c, d, and e of adipic acid precipitated with urea that is not present in the patterns a and b. The intensity of peak P changes linearly with the initial concentration of urea, suggesting that urea might modify the crystal structure of adipic acid when precipitated by SAS in the presence of urea. To gain insight into the effect of urea, we have prepared and analyzed by XRPD various physical mixtures obtained by mechanically stirring adipic acid and urea in the desired concentrations corresponding to those used in the precipitation process, namely, 6.25, 12.5, and 25 wt % urea with respect to adipic acid. Figure 8 shows the XRPD patterns of adipic acid/ urea mixtures at 25 and 12.5 wt %. Both patterns present the main characteristic peaks of adipic acid (A1, A2, and A3) and the characteristic peak of urea (U), whereas, the peak P at 27.6° 2θ observed in adipic acid samples obtained by SAS precipitation (patterns c-e) was not present. This indicates that peak P is characteristic of adipic acid powders obtained by SAS precipitation in the presence of urea in the given range of concentration. Differential Scanning Calorimetry. Typical thermograms for pure adipic acid, pure urea, SAS-processed adipic acid in the presence of urea, and adipic acid/urea physical mixtures are shown in Figure 9. The thermograms of pure products, as expected, showed a single endothermic peak corresponding to the melting point of adipic acid at 151 °C (Figure 9a) and of urea at 134 °C (Figure 8b). SAS-processed adipic acid showed the same melting temperature of the pure product. Also, sample obtained with urea at 6.25 wt % had only one peak at 151 °C, as the pure product. When adipic acid samples processed in the presence of urea at 12.5 and 25 wt % were analyzed, two thermal events were observed, as shown in the DSC curves reported in Figure 9c, an endothermic peak at low temperature (110 °C) and a second endothermic curve, which is not isothermal and peaks at higher temperature. This behavior is typical of eutectic solid mixtures, where the first peak corresponds to the eutectic melting and the second peak corresponds to solid-liquid transition in which temperature is a function of the composition. When the concentration of urea was increased from 12.5 to 25 wt %, the eutectic peak area increased and the solid-liquid transition peak area decreased: the sample composition approaches the eutectic composition at which only the first peak should be present. This behavior indicates that the SAS-precipitated samples contain a residual content of urea that

Figure 9. DSC thermograms of various samples: (a) pure adipic acid; (b) pure urea; (c) SAS-processed adipic acid in the presence of urea at 6.25, 12.5, and 25 wt %; (d) physical mixture of urea/adipic acid at 6.25, 12.5, 25.0, and 27.5 wt %.

is proportional to the initial concentration of urea in the solution. These findings were confirmed by a comparison with the thermal behavior of urea/adipic acid physical mixtures reported in Figure 9d. Physical mixtures prepared with a urea/adipic acid ratio from 6.25 to 25 wt % have eutectic peaks the area of which is proportional to the initial urea concentration. Samples prepared with urea at 27.5 wt % showed only a peak at 110 °C, attesting the close proximity to the eutectic concentration of the mixture of adipic acid/urea. SAS precipitation carried out at low urea concentration (i.e., 6.25 wt %) resulted in the formation of pure adipic acid, probably because urea is completely extracted by CO2 during the washing step. Indeed, urea is slightly soluble in SC CO2 with a molar solubility of 1 × 10-6 at 40 °C and 80 bar.20 Precipitation carried out at higher urea concentration (i.e., 12.5 and 25 wt %) leads to mixtures with a composition minor to the eutectic. A comparison between the DSC curves of the SAS-precipitated sample at 12.5 wt % and the physical mixture at the same concentration shows that the eutectic peak area of the physical mixture is slightly higher than that of the precipitated sample. This fact indicates that during precipitation, part of the urea is extracted by CO2. The formation of a solid mixture characterized by eutectic mixture upon SAS precipitation supports the hypothesis that urea acts by a selective adsorption mechanism and is not incorporated in the crystal structure. However, DSC analysis is not able to confirm the formation of a new polymorphic form of adipic acid when precipitated in the presence of urea because the formation of the eutectic peak hides the possible polymorphic transition. FT-IR Spectroscopy Analysis. To further study the possibility of an interaction of urea with adipic acid in the solid state, information was gathered using FT-IR spectroscopy. From the structure of adipic acid and urea, it can be assumed that the possible interaction would occur between the amide hydrogen atom and the carboxylic oxygen of adipic acid. Therefore, in this case, any sign of the interaction would be reflected by shifts in N-H and CdO vibrations depending on the extent of interactions. The FT-IR spectra of SAS-processed adipic acid, SAS-processed adipic acid in the presence of urea and urea/ adipic acid physical mixture are shown in Figure 10. SAS-



Figure 11. DSC thermograms of SAS-processed adipic acid in the presence of urea at (a) 12.5 and (b) 25 wt % before and after washing with water. Figure 10. FT-IR adsorption spectra of (a) pure SAS-processed adipic acid, (b) SAS-processed adipic acid in the presence of urea at 25 wt %, and (c) adipic acid/urea physical mixture at 25 wt %.

processed adipic acid and pure adipic acid (data not shown) have the same spectra. Their characteristic peaks at 1709 and 1230 cm-1 are assigned due to stretching of CdO and C-O groups. The broad band from 3500 to 2500 cm-1 is due to the stretching of O-H bonds. Pure urea is characterized by two characteristics peaks at 3445 and 3340 cm-1 assigned due to stretching of the two NH2 groups, at 1680 cm-1 due to stretching of CdO, and two peaks at 1637 and 1469 cm-1 due to NH2 scissors and C-N stretching, respectively. SAS-processed adipic acid in the presence of urea and urea/adipic acid mixtures showed characteristics peaks of adipic acid, confirming the stability of the product when treated with SC CO2. SASprocessed adipic acid in the presence of urea and physical mixtures also showed characteristics peaks of urea at 3445 and 3440 cm-1, whereas the three peaks at lower wave numbers are superimposed to that of adipic acid. FT-IR analysis showed no defined interactions between adipic acid and urea since no new peaks or shifts of peaks were observed. These results further indicate the absence of well-defined interaction between adipic acid and urea as already indicated by XRPD and DSC. Moreover, physical mixtures and precipitated samples with urea show practically the same FT-IR spectra. Thus, after precipitation, urea remains entrapped in the adipic acid sample as an impurity, which can be detected as low intensity peaks at 3445 and 3440 cm-1. Comparing spectra b and c of Figure 10, which refer to precipitated sample and physical mixture at the same initial concentration of urea (i.e., 25 wt %), one can observe that the intensity of peaks at 3445 and 3440 cm-1 are lower for the precipitated sample, confirming that after precipitation, urea is partially extracted by CO2. Washing of Crystals. Upon crystallization, urea is partially extracted by CO2 in the final washing step of the process, and the remaining part is mixed with adipic acid forming a solid mixture. After precipitation, adipic acid crystals obtained at 12.5 and 25 wt % urea were washed with water (in which urea is largely soluble and adipic acid practically insoluble), filtered, dried, and analyzed by DSC and XRPD. In Figure 11, DSC thermographs of washed samples are reported. Urea is no longer present after washing, and the melting peak of adipic acid overlaps that of pure adipic acid shown in Figure 9c. XRPD spectra reported in Figure 12 show that after washing the position of the characteristic peaks of pure adipic acid is unchanged and the large peak P present in the sample at 25 wt % urea (Figure 7) is no longer present. Thus, it can be concluded that urea is not incorporated into the

Figure 12. X-ray powder diffraction (XRPD) spectra of SAS-processed adipic acid in the presence of urea before and after washing with water: (a) 12.5 wt % urea after washing; (b) 12.5 wt % urea before washing; (c) 25 wt % urea after washing, (d) 25 wt % urea before washing. A1, A2, and A3 are characteristics peaks of adipic acid; P is a characteristic peak of adipic acid precipitated in the presence of urea.

parent crystal structure and crystal structure is not changed. The modification induced by urea on the crystal habit and size of adipic acid can be attributed to a preferential adsorption mechanism of urea to the fastest growing crystal face of needle crystals, which in the case of adipic acid is the face in which the higher concentration of carboxylic groups is present. Peak P in the adipic acid might be related to the presence of urea that is modified by the SAS process and behaves as an impurity for the adipic acid in the powder. Effect of Pressure in the Presence of Urea. The effect of pressure on the precipitation of adipic acid in the presence of urea was also investigated. Experiments were carried out at 40 °C and 80, 90, and 120 bar at a fixed concentration of urea of 0.16 wt % (experiments M3, M10, and M4 of Table 1). Precipitation carried out at 120 bar results in the formation of prismatic crystals of 7.5 µm mean size, as previously described (Figures 5b and 6). As pressure is isothermally reduced to 80 bar, the resulting crystal habit changes significantly. This can be readily seen by comparing SEM images reported in Figure 5b to that reported in Figure 13a. Particle morphology is completely different in shape and size from that obtained at 120 bar. Spherical particles with a uniform morphology and with a mode of about 4 µm were obtained. In Figure 14, the particle size distribution is also reported. Spherical microparticles are a morphology that has been frequently observed during SAS experiments with various materials.14,22 It is due to the jet breakup of the liquid solution and the subsequent drying of the droplets formed upon atomization. This mechanism develops when temperature and pressure conditions lead to the formation of a vapor phase that is below the mixture critical point. Liquid droplets can form at the nozzle tip, and sustained by surface tension, their lifetime is long enough to allow the precipitation of the solute within each droplet.23,24 In the case of adipic acid,


Caputo et al.

Figure 13. SEM images of adipic acid produced by SAS at 40 °C, 80 bar, in the presence of urea at 0.16 wt %: (a) 3000×; (b) 5000×; (c) 10000× enlargements.

Figure 14. Particle size distribution of adipic acid precipitated by SAS process at 40 °C, 80 bar, in the presence of urea at 0.16 wt % (expt M3).

the surface of microparticles is composed of very small crystals less than 1 µm in length (Figure 13c). This fact can be explained considering that two mechanism in series are responsible of the formation of the particles: the first leads to the formation of spherical particles, and the second leads to the formation of crystals on the particle surface. The effect of urea superimposes to this complex mechanism. It is likely that urea acts as growth inhibitor of adipic acid during the “droplet confined” crystallization, producing small prismatic crystals that are visible on the surface of the expanded particles in Figure 13b. When the precipitation is operated at the same conditions without urea, long needles are formed because of the lack of the growth inhibition effect provided by urea. Precipitation carried out at 90 bar resulted in the formation of both crystals with near prismatic shape and microparticles, probably because these operating conditions represent a transition between the two observed mechanisms. The solid state properties of samples obtained at 80 and 90 bar were also analyzed by DSC and XRPD. Results are very similar to those obtained at 120 bar. Particularly, DSC thermograms also in this case revealed the presence of a eutectic peak. Conclusions The supercritical antisolvent crystallization of adipic acid from acetone using CO2 as antisolvent and urea as habit modifier has been studied. The use of urea allows a close control of crystal size and habit of adipic acid. Indeed, adipic acid crystal habit was modified from needle-like to prismatic with the addition of urea at 12.5 wt %. The size of the crystals was reduced from several tens of micrometers to 7.5 µm. However, the effect of urea largely depends on its concentration with respect to adipic acid. Urea at 12.5 wt % seems an optimum value below and above which the modification effect is largely reduced. After precipitation, when crystals were washed with water in which urea is largely soluble, it was observed that urea is

completely extracted and crystals were unchanged. Thus it can be concluded that urea is not incorporated into the parent crystal structure. The modification induced by urea on the crystal habit and size of adipic acid can be attributed to a preferential adsorption mechanism of urea to the fastest growing crystal face of needle crystals. The obtained experimental results motivate further work and provide directions for future investigations about the potential of the use of habit modifiers in the supercritical antisolvent precipitation. The authors believe that coupling the well-known tunability of SCF-based processes with the use of “tailor-made” additives can greatly enhance the control of crystal habit and polymorphism of a large variety of products. Acknowledgment. The authors acknowledge Dr. Domenico Ercolino for the help in performing the micronization tests and part of the analyses and Dr. Andrea Sorrentino for performing FT-IR analyses. Renata Adami acknowledges Academy of Finland (Finland) for the financial support.

References (1) Weissbuch, I.; Leiserowitz, L.; Lahav, M. In Crystallization Technology Handbook, 2nd ed.; Mersman, A., Ed.; Marcel Dekker: New York, 2001; p 315. (2) Mullin, J. W. Crystallization, 3rd ed.; Butterworth Heinemann: London, 1993; pp 254-256 (3) Meenan, P. A. In Handbook of Industrial Crystallization, 2nd ed; Myerson, A. S. Ed.; Butterworth Heinemann: Boston, MA, 2002; p 67. (4) Jung, J.; Perrut, M. J. Supercrit. Fluids 2001, 20, 1. (5) Reverchon, E.; Caputo, G. Trends Chem. Eng. 2003, 8, 183. (6) Shariati, A.; Peters, C. J. Curr. Opin. Solid State Mater. Sci. 2003, 7 (4-5), 371–383. (7) Thiering, R.; Dehghani, F.; Foster, N. R. J. Supercrit. Fluids 2001, 21 (2), 159–177. (8) Reverchon, E. J. Supercrit. Fluids 1999, 15, 1. (9) Cabrera, N.; Vermilyea, D. In Growth and Perfection of Crystals; Doremus, R. H. Roberts, B.W.; Turnbull, D., Eds.; Wiley: New York, 1958; p 393. (10) Rosenberg, F. J. Cryst. Growth 1986, 76, 618. (11) Myerson, A. S.; Jang, S. M. J. Cryst. Growth 1995, 156, 459. (12) Vemavarapu, C.; Mollan, M. J.; Needham, T. E. AAPS Pharm. Sci. Technol. 2002, 3 (4), 1. (13) Jarmer, D. J.; Lengsfeld, C. S.; Anseth, K. S.; Randolph, T. W. J. Pharm. Sci. 2005, 94 (12), 2688. (14) Caputo, G.; Reverchon, E. Ind. Eng. Chem. Res. 2007, 46 (12), 4265– 4272. (15) Caputo, G.; Adami, R.; Ercolino, D.; Sesti Osse´o, L.; Reverchon, E. Proc. V Int. Symp. High Pressure Processes Technol. Chem. Eng., SegoVia (Spain) 2007, 24–27. (16) Michaels, A. S.; Colville, A. R. J. J. Phys. Chem. 1960, 64, 13. (17) Reverchon, E.; De Marco, I.; Torino, E. J. Supercrit. Fluids 2007, 43, 126–138. (18) Yeo, S.; Kim, M.; Lee, J. J. Supercrit. Fluids 2003, 25, 143–154. (19) Muhrer, G.; Meier, U.; Fusaro, F.; Albano, S.; Mazzotti, M. Int. J. Pharm. 2006, 308, 69–83. (20) Catchpole, O. J.; Tallon, S. J.; Dyer, P. J.; Lan, J. S.; Jensen, B.;

Supercritical Fluid Crystallization of Adipic Acid Rasnussen, O. K.; Grey, J. B. Fluid Phase Equilib. 2005, 237, 212– 272. (21) Shekunov, B. Y.; Bristow, S.; Chow, A. H. L.; Cranswick, L.; Grant, D. J. W.; York, P. Cryst. Growth Des. 2003, 3, 603. (22) Reverchon, E.; Caputo, G.; De Marco, I. Ind. Eng. Chem. Res. 2003, 42 (25), 6406–6414.

Crystal Growth & Design, Vol. 8, No. 8, 2008 2715 (23) Lengsfeld, C. S.; Delplanque, J. P.; Barocas, V. H.; Randolph, T. W. J. Phys. Chem. B 2000, 104, 2725. (24) Chavez, F.; Debenedetti, P. G.; Luo, J. J.; Dave, R. N.; Pfeffer, R. Ind. Eng. Chem. Res. 2003, 42, 3156–3162.

CG700753U

Supercritical Fluid Crystallization of Adipic Acid Using Urea as Habit

Recommend Documents