- Email: [email protected]
Crystal formation of BaCl2 and NH4Cl using a supercritical fluid antisolvent
Journal of Supercritical Fluids 16 (2000) 235–246 www.elsevier.com/locate/supflu
Crystal formation of BaCl and NH Cl using a 2 4 supercritical fluid antisolvent Sang-Do Yeo *, Jae-Ho Choi, Tae-Jong Lee Department of Chemical Engineering, Kyungpook National University, Taegu 702-701, South Korea Received 7 June 1999; received in revised form 7 September 1999; accepted 8 September 1999
Abstract The supercritical antisolvent (SAS ) process has been used to produce crystals of barium chloride (BaCl ) and 2 ammonium chloride (NH Cl ) from solutions of dimethyl sulfoxide (DMSO). Crystallization was performed by 4 introducing carbon dioxide into the DMSO solutions at different injection rates. Variations in crystal properties such as particle size, crystal habit, and internal space lattice were observed in the SAS processed crystals. The average particle size of the crystals decreased with increasing carbon dioxide injection rate for both compounds. As the injection rate increased, the crystal habit of BaCl was modified from an equant to an acicular habit, and that of 2 NH Cl changed from an equant to a tabular habit. X-ray diffraction patterns showed the transition of the space 4 lattice of BaCl from the orthorhombic to hexagonal crystal system. The use of the SAS process for the separation 2 of BaCl and NH Cl mixtures in DMSO was also investigated. © 2000 Elsevier Science B.V. All rights reserved. 2 4 Keywords: BaCl ; Crystal formation; NH Cl; Supercritical fluid antisolvent 2 4
1. Introduction The supercritical fluid antisolvent (SAS ) technique has been used in a variety of particle formation processes [1,2]. The technology has been used primarily to produce ultrafine particles, which are difficult to handle by conventional size-reduction methods. The SAS process is based on the principle of solvent-induced phase separation in which the particle precipitation is caused by the addition of a gas or a supercritical fluid into a liquid solution. An important feature of the SAS process involves the complete removal of residual organic solvents and antisolvents from the solid material by reduc* Corresponding author. Tel.: +82-53-950-5618; fax: +82-53-950-6615. E-mail address: [email protected] (S.-D. Yeo)
ing the system pressure at the end of the process. Following the pioneering work of Gallagher et al. in 1989 [3], research has focused on producing a variety of pharmaceutical and polymeric compounds such as microparticulate protein powders [4,5], polymer particles and fibers [6–8], and the composite particles of polymeric pharmaceutical compounds [9,10]. In these studies, investigators have largely concentrated on producing particles of macromolecular compounds. The main objective of previous research was to resize and modify the morphology of the macromolecular compounds. Structural analysis of the particles produced has shown that it is difficult to form crystals of the larger molecules using the SAS process. During the SAS operation, the mixing process of the solvent and antisolvent generates high turbulence within the solution, which
0896-8446/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S0 8 9 6 -8 4 4 6 ( 9 9 ) 0 0 03 7 - 6
236
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
affects both the nucleation and growth of the resulting particles. These disturbed conditions, which exist during the particle formation process, interrupt the regular arrangement of molecules. Indeed, the crystallization of proteins and polymers requires tranquil conditions during the supersaturation and particle growth stages. Therefore, in normal SAS operation, the precipitated macromolecular compounds are more likely to exist in the amorphous state rather than the crystalline state unless the molecule itself has an exceptionally high level of crystalline structure, which might be induced by a molecular backbone of exceptional rigidity [6 ]. However, under carefully controlled experimental conditions, such as an extremely low contact speed between the solvent and antisolvent, the technique can provide a suitable environment for the nucleation and growth of crystallizable materials. In such conditions, the variation of the contact rate of solvent and antisolvent may affect both the crystal’s internal structure and its external characteristics, which would determine the physical behavior of the materials [11]. Previous research has shown that the slow injection of antisolvent into a solution produces somewhat different particles in terms of morphology and size compared with the continuous SAS operation, which involves extremely rapid contact between solvent and antisolvent[4,6 ]. In these systems, the particles are produced in either the amorphous or spherulitic state. These results encouraged us to investigate how the contact rate of antisolvent influences the system, particularly when the particles produced are crystalline in nature and they have well-defined crystal structures. In this study, therefore, we select inorganic crystalline compounds (barium chloride and ammonium chloride) in order to investigate the impact of SAS operating conditions on the modification of the internal crystalline structure (crystal space lattice) and their external charateristics (crystal habit and size).
2. Experimental Barium chloride (BaCl ) and ammonium chlo2 ride (NH Cl ) with a minimum purity of 99.0% 4
were purchased from Aldrich Chemical Co. Dimethyl sulfoxide (DMSO) was selected as a solvent for the chlorides because of its high affinity to both compounds and the antisolvent. Carbon dioxide was employed as an antisolvent for the DMSO solutions. All the chemicals were used without further purification. A schematic diagram of the experimental apparatus for the SAS process is shown in Fig. 1. The apparatus was made exclusively for batch-type operation, which has been described in the literature [4,6 ]. In this operation, the contact between the solvent and antisolvent occurs only by the injection of carbon dioxide into the liquid solution. The unit consists of three major parts: a crystallizing chamber (Jerguson Gauge, Model 19-T-40), a carbon dioxide supply system, and a depressurizing section. The crystallizing chamber is placed in an air chamber to maintain a constant temperature during crystal formation. Solutions of 3 and 5 wt% BaCl in DMSO and 10 wt% NH Cl in DMSO 2 4 were prepared. Normally, 10 cm3 of the DMSO solution were loaded into the crystallizing chamber, and its temperature was maintained at a constant value within ±0.2°C. Carbon dioxide was injected from the bottom of the chamber to achieve a pressure of 95.0 bar. The rate of carbon dioxide injection was controlled so that the pressure could increase at different rates in the crystallizing chamber. The rate of pressure increase ranged from 0.1 to 24.5 bar/min. The introduction of carbon dioxide and the pressure increase caused a concomitant expansion of the DMSO solution and the crystallization of dissolved compounds. The expansivity of DMSO when in contact with carbon dioxide can be found in the literature [4]. During the expansion, the solution became cloudy at a certain pressure, and it was recorded as a nucleation pressure. The nucleation pressure was measured as a function of temperature and the initial concentration of the DMSO solutions. The system was pressurized until the vapor (carbondioxide-rich) and liquid (DMSO-rich) phases merged to become a single phase, at which stage the crystallization procedure was assumed to complete. The precipitated crystals were collected on a filter located at the bottom of the crystallizing chamber. The carbon dioxide and solute-free DMSO mixture was then separated from the crys-
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
237
Fig. 1. Schematic diagram of the experimental apparatus for the supercritical antisolvent (SAS ) process.
tals by the use of a ventilation valve. During the ventilation procedure, the pressure inside the chamber was kept constant at c. 95 bar, by introducing pure carbon dioxide from the top of the chamber. This was done to prevent the redissolution of the precipitated crystals in DMSO below the nucleation pressure. Finally, the residual DMSO was removed by flowing carbon dioxide continuously through the crystallizing chamber for 2 h at 37°C and 95 bar. The morphology and composition of BaCl and 2 NH Cl crystals were examined using a scanning 4 electron microscope (SEM, Hitachi, Model S-4200) equipped with an energy dispersive X-ray
spectrometer ( EDS). The internal structure of the crystals was analyzed by using a X-ray diffractometer ( XRD, Philips, Model X’PERT ). The particle size of the crystals was estimated, based on the image size of randomly selected specimens on SEM photomicrographs.
3. Results and discussion 3.1. Nucleation pressure The nucleation pressures of the BaCl and 2 NH Cl solutions in DMSO are represented as a 4
238
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
at higher temperatures. These results can be explained by considering the variation of carbon dioxide levels in the DMSO-rich phase upon changing the pressure and temperature. During the injection of carbon dioxide into the crystallizing chamber, two distinct phases are created, i.e. a carbon-dioxide-rich phase and a DMSO-rich phase. The carbon dioxide concentration in the DMSO-rich phase will continuously increase as the pressure increases inside the chamber. At the nucleation pressure, the amount of the carbon dioxide in the DMSO-rich phase corresponds to the amount of antisolvent, which causes the supersaturation. This quantity of antisolvent needed to achieve supersaturation will decrease with increasing chloride concentration in the DMSO solution, and therefore, the nucleation pressure will decrease accordingly. The effect of temperature on the nucleation pressure follows the principle of vapor– liquid equilibrium. At an elevated temperature, a higher pressure is required to maintain a sufficient carbon dioxide concentration in the DMSO-rich phase to achieve supersaturation. This explains the increase of nucleation pressure required at a higher temperature.
Table 1 Variation of crystal habit and size of BaCl and NH Cl crystals 2 4 as a function of carbon dioxide injection rate and temperaturea Injection rate (bar/min) Fig. 2. Nucleation pressure of (a) BaCl and (b) NH Cl solu2 4 tions in DMSO as a function of chloride concentration at 25 and 35°C.
function of initial chloride concentration in Fig. 2. The solutions at concentration ranges of 1–5 wt% (BaCl ) and 5–10 wt% (NH Cl ) were investigated 2 4 at two temperatures, 25 and 35°C. The nucleation pressure decreased with increasing chloride concentration in DMSO and increased with temperature at constant concentration. For BaCl 2 solutions, the dependency of nucleation pressure on the initial chloride concentration was greater
BaCl 2 0.1 0.1 0.1 24.5 24.5 24.5 NH Cl 4 0.1 0.1 0.1 24.5 24.5 24.5
T (°C )
Crystal habit
Estimated crystal size (mm)
25 35 50 25 35 50
Equant Equant Equant Acicular Acicular Acicular
7.7 6.3 9.0
25 35 50 25 35 50
Equant Equant Equant Tabular Tabular Tabular
200 280 400 1.8 1.8 4.8
a The initial concentrations of BaCl and NH Cl in DMSO 2 4 were 5 and 10 wt%, respectively
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
3.2. Effect of operating conditions on crystal habit and size The crystallizations of BaCl and NH Cl were 2 4 performed by changing various operating conditions: injection rate of carbon dioxide into the crystallizing chamber, initial chloride concentration in DMSO, and temperature. The carbon dioxide injection rate is defined as the rate of pressure increase in the crystallizing chamber in bar/min. Two injection rates were used: slow injec-
239
tion (0.1±0.05 bar/min) and rapid injection (24.5±0.5 bar/min). Figs. 3 and 4 show selected SEM photomicrographs of BaCl and NH Cl crystals: unprocessed 2 4 particles (a) and SAS processed crystals (b–d). Observations indicated that the initial chloride concentration in DMSO appeared to have little effect on particle size. However, the carbon dioxide injection rate obviously controlled both the size of crystals and the crystal habit. The crystal habit and the estimated size of BaCl and NH Cl crystals 2 4
(a)
(b)
(c)
(d)
Fig. 3. SEM photomicrographs of BaCl crystals: (a) unprocessed particles; (b) processed with the slow injection operation of 2 0.1 bar/min pressure increase; (c) magnified version of a crystal obtained in (b); (d) processed with the rapid injection operation of 24.5 bar/min pressure increase. The initial concentration of DMSO and temperature were 5 wt% and 35°C, respectively.
240
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
(a)
(b)
(c)
(d)
Fig. 4. SEM photomicrographs of NH Cl crystals: (a) unprocessed particles; (b) processed with the slow injection operation of 4 0.1 bar/min pressure increase; (c) magnified version of a crystal obtained in (b); (d) processed with the rapid injection operation of 24.5 bar/min pressure increase. The initial concentration of DMSO and temperature was 10 wt% and 35°C, respectively.
produced using the two different injection rates are summarized in Table 1. In this table, the initial concentrations of BaCl and NH Cl in DMSO 2 4 were at 5 and 10 wt%, respectively. The change of carbon dioxide injection rate resulted in a modification of the crystal habit of BaCl (Fig. 3). A slow injection produced cubic2 shaped crystals termed an ‘equant habit’ while rapid injection generated flattened needle-like crystals termed an ‘acicular habit’ [12]. The variation in crystal habit results from the alteration of the relative growth rates of crystal faces, which are
governed by a number of factors. The degree of supersaturation of the mother liquor is known as one of the main factors that affects crystal habit. In general, it is known that as supersaturation is increased, the crystal form tends to change from granular to needle-like [13]. In the case of NH Cl, an equant habit was consistently observed 4 in the slow injection mode. In a rapid injection operation, the shapes of the crystals tended to flatten and this resulted in tabular habit accompanied by a tremendous reduction in crystal size. The size of the NH Cl crystals obtained from the 4
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
slow injection experiments was almost two orders of magnitude larger than crystals produced from the rapid injection operation. It was found that the crystallization temperature had practically no influence on the crystal habit or size within the temperature ranges investigated. 3.3. Effect of operating conditions on crystal structure The composition of the BaCl crystals was 2 examined using an energy-dispersive X-ray spectrometer. Fig. 5 shows the EDS spectra of unprocessed and processed BaCl crystals. The 2 stoichiometric analysis of BaCl crystals based on 2 the integration of these spectra is summarized in Table 2. The patterns of the two spectra are identical, which implies that the chemical composition of the BaCl crystals remains unchanged after the 2 SAS process. The internal structure of the crystals was analyzed using an X-ray diffractometer with CuK a radiation. Fig. 6 shows the X-ray diffraction patterns of unprocessed and processed BaCl crystals. 2 In these X-ray diffraction patterns, distinct differences are apparent, which are attributed to change in the arrangements of the molecules in the crystal lattices. The X-ray diffraction patterns were compared with the diffraction data from JCPDS– ICDD (Joint Committee of Powder Diffraction Standards–International Centre for Diffraction Data) cards. The result revealed that the internal structure of unprocessed BaCl crystal corresponds 2 to the orthorhombic space lattice. However, the processed BaCl crystal was found to possess a 2 hexagonal crystal structure. Fig. 3(c) also shows an agreement between the external shape of a single crystal and the systematic definition of the Table 2 Percentage of elements in a BaCl crystal based on the EDS 2 analysis Elements
Unprocessed (wt%)
Processed (wt%)
Ba Cl O S
75.25 22.30 1.80 0.52
69.02 25.29 4.86 0.83
241
hexagonal space lattice in the crystallographic classification (the hexagonal crystal system is defined as: the elementary lengths and axis angles of a single crystal unit are described as a=b≠c and a=b=90°, c=120° [14]). For reference purposes, according to the JCPDS data cards, BaCl may 2 exist in three different lattice spaces, which are orthorhombic, hexagonal and cubic. The variation of experimental conditions, such as temperature, initial DMSO concentration, and the carbon dioxide injection rate, did not have any recognizable effect on the X-ray diffraction patterns. Fig. 7 shows the X-ray diffraction patterns of unprocessed and processed NH Cl crystals. The 4 results indicated that two diffraction patterns were identical, and the internal structure of NH Cl 4 crystal was unchanged by SAS operation. According to the JCPDS diffraction data cards, the unprocessed and processed NH Cl crystals 4 both belong to cubic space lattice. Moreover, the carbon dioxide injection rate did not influence the diffraction pattern, and NH Cl crystallized in the 4 cubic space lattice form, regardless of the experimental conditions used. In fact, the cubic space lattice is the only possible crystal system for NH Cl as reported in the JCPDS data cards. 4 However, it was observed that the relative intensity of each peak in the diffraction patterns varied, depending upon the experimental conditions, including the injection rate. The results suggested that the SAS operation probably alters the orientation of the crystal faces. 3.4. Separation of BaCl and NH Cl mixtures in 2 4 DMSO The SAS process enables the separation of multicomponent mixtures if the nucleation of each component occurs at different pressures. As shown in Fig. 2, the nucleations of BaCl and NH Cl take 2 4 place at different pressure ranges depending upon the temperature and the concentration of each compound. The SAS experiments were conducted employing a DMSO solution containing both BaCl (5 wt%) and NH Cl (5 wt%) at 25°C. 2 4 According to Fig. 2, the nucleation pressures of BaCl and NH Cl at the concentration of 5 wt% 2 4 were 43.5 and 48.0 bar, respectively. During the
242
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
(a)
(b)
Fig. 5. EDS spectra of the (a) unprocessed and (b) processed BaCl crystals. 2
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
243
Fig. 6. X-ray diffraction patterns of the (a) unprocessed and (b) processed BaCl crystals. The diffraction patterns (a) and (b) 2 correspond to the orthorhombic and hexagonal crystal systems, respectively.
244
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
Fig. 7. X-ray diffraction patterns of the (a) unprocessed and (b) processed NH Cl crystals. Both the diffraction patterns (a) and (b) 4 correspond to the cubic crystal system.
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
245
Fig. 8. X-ray diffraction patterns of the (a) BaCl and (b) NH Cl crystals separated from the mixture of the two compounds in 2 4 DMSO solutions. The diffraction patterns (a) and (b) correspond to the patterns of pure BaCl and NH Cl, respectively. 2 4
246
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
pressure increase (up to 95.0 bar) caused by the carbon dioxide injection, two separate nucleation points were clearly observed at two different nucleation pressures: one at c. 43.5 bar and the other at c. 48.0 bar. In order to separate the two compounds, carbon dioxide was first injected up to 43.5 bar so that only BaCl precipitated from the 2 solution. After the NH Cl-containing solution was 4 removed from the crystallizing chamber, only BaCl crystals were collected. The removed DMSO 2 solution was then recharged to the crystallizing chamber, and the SAS experiment was repeated to produce NH Cl crystals. Fig. 8 shows the X-ray 4 diffraction patterns of BaCl and NH Cl crystals 2 4 obtained from the mixture of the two compounds in DMSO solution. A comparison of these diffraction patterns to those of Figs. 6 and 7 reveals that the BaCl and NH Cl were successfully separated 2 4 from the DMSO solution and that the crystal lattices of the separated BaCl and NH Cl were 2 4 hexagonal and cubic, respectively. In this research, it has been shown that the SAS process can be a useful technology to bring about a modification of the internal structure (space lattice) and external shape (crystal habit) of crystals, which will affect the physical behavior of the crystalline particles so obtained. The various crystal space lattices exhibited by a single compound may result in different physical properties, including the solubility and dissolution rates in solvents [11]. The crystal habit modification is of importance in industrial applications because it allows control of the type of crystals produced. Certain crystal habits are undesirable in commercial crystals for many reasons, such as poor appearance, inefficient flow characteristics or difficulty in handling. In order to obtain the desired crystal habit, habit modifiers are frequently employed in crystallizing processes, especially in those involving inorganic compounds[14]. In this point of view, there is a strong possibility that the SAS process can
provide an alternative method of controlling the internal and external structure of crystals and hence regulating the physical properties of crystalline particles.
Acknowledgements Analytical instruments (SEM, EDS and XRD) for this study were provided by the Korea Basic Science Institute, CKBSI.
References [1] E. Reverchon, J. Supercrit. Fluids 15 (1999) 1. [2] P.M. Gallagher, V.J. Krukonis, G.D. Botsaris, AIChE Symp. Series 284 (87) (1991) 96. [3] P.M. Gallagher, M.P. Coffey, V.J. Krukonis, N. Klasutis, New process to recrystallize compounds insoluble in supercritical fluids, in: K.P. Johnston, J.M.L. Penninger (Eds.), Supercritical Fluid Science and Technology, American Chemical Society, Washington, DC, 1989, p. 334. [4] S. Yeo, G. Lim, P.G. Debenedetti, H. Bernstein, Biotechnol. Bioeng. 41 (1993) 341. [5] M.A. Winters, B.L. Knutson, P.G. Debenedetti, H.G. Sparks, T.M. Przybycien, C.L. Stevenson, S.J. Prestrelski, J. Pharm. Sci. 85 (1996) 586. [6 ] S. Yeo, P.G. Debenedetti, M. Radosz, H.-W. Schmidt, Macromolecules 26 (1993) 6207. [7] S. Mawson, S. Kanakia, K.P. Johnston, J. Appl. Polym. Sci. 64 (1997) 2106. [8] L. Benedetti, A. Bertucco, P. Pallado, Biotechnol. Bioeng. 53 (1997) 232. [9] R.F. Falk, T.W. Randolph, Pharmaceut. Res. 15 (1998) 1233. [10] Y. Chou, D.L. Tomasko, GAS crystallization of polymerpharmaceutical composite particles, in: Proc. 4th Int. Symp. Supercritical Fluids Vol. A (1997) 55. [11] C.J. Price, Chem. Eng. Prog. 93 (1997) 35. [12] F.D. Bloss, Crystallography and Crystal Chemistry, Holt Rinehart & Winston, USA, 1971. [13] J.K. Haleblian, J. Pharm. Sci. 64 (1975) 1269. [14] J.W. Mullin, Crystallisation, Butterworth, London, 1972.
Crystal formation of BaCl and NH Cl using a 2 4 supercritical fluid antisolvent Sang-Do Yeo *, Jae-Ho Choi, Tae-Jong Lee Department of Chemical Engineering, Kyungpook National University, Taegu 702-701, South Korea Received 7 June 1999; received in revised form 7 September 1999; accepted 8 September 1999
Abstract The supercritical antisolvent (SAS ) process has been used to produce crystals of barium chloride (BaCl ) and 2 ammonium chloride (NH Cl ) from solutions of dimethyl sulfoxide (DMSO). Crystallization was performed by 4 introducing carbon dioxide into the DMSO solutions at different injection rates. Variations in crystal properties such as particle size, crystal habit, and internal space lattice were observed in the SAS processed crystals. The average particle size of the crystals decreased with increasing carbon dioxide injection rate for both compounds. As the injection rate increased, the crystal habit of BaCl was modified from an equant to an acicular habit, and that of 2 NH Cl changed from an equant to a tabular habit. X-ray diffraction patterns showed the transition of the space 4 lattice of BaCl from the orthorhombic to hexagonal crystal system. The use of the SAS process for the separation 2 of BaCl and NH Cl mixtures in DMSO was also investigated. © 2000 Elsevier Science B.V. All rights reserved. 2 4 Keywords: BaCl ; Crystal formation; NH Cl; Supercritical fluid antisolvent 2 4
1. Introduction The supercritical fluid antisolvent (SAS ) technique has been used in a variety of particle formation processes [1,2]. The technology has been used primarily to produce ultrafine particles, which are difficult to handle by conventional size-reduction methods. The SAS process is based on the principle of solvent-induced phase separation in which the particle precipitation is caused by the addition of a gas or a supercritical fluid into a liquid solution. An important feature of the SAS process involves the complete removal of residual organic solvents and antisolvents from the solid material by reduc* Corresponding author. Tel.: +82-53-950-5618; fax: +82-53-950-6615. E-mail address: [email protected] (S.-D. Yeo)
ing the system pressure at the end of the process. Following the pioneering work of Gallagher et al. in 1989 [3], research has focused on producing a variety of pharmaceutical and polymeric compounds such as microparticulate protein powders [4,5], polymer particles and fibers [6–8], and the composite particles of polymeric pharmaceutical compounds [9,10]. In these studies, investigators have largely concentrated on producing particles of macromolecular compounds. The main objective of previous research was to resize and modify the morphology of the macromolecular compounds. Structural analysis of the particles produced has shown that it is difficult to form crystals of the larger molecules using the SAS process. During the SAS operation, the mixing process of the solvent and antisolvent generates high turbulence within the solution, which
0896-8446/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S0 8 9 6 -8 4 4 6 ( 9 9 ) 0 0 03 7 - 6
236
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
affects both the nucleation and growth of the resulting particles. These disturbed conditions, which exist during the particle formation process, interrupt the regular arrangement of molecules. Indeed, the crystallization of proteins and polymers requires tranquil conditions during the supersaturation and particle growth stages. Therefore, in normal SAS operation, the precipitated macromolecular compounds are more likely to exist in the amorphous state rather than the crystalline state unless the molecule itself has an exceptionally high level of crystalline structure, which might be induced by a molecular backbone of exceptional rigidity [6 ]. However, under carefully controlled experimental conditions, such as an extremely low contact speed between the solvent and antisolvent, the technique can provide a suitable environment for the nucleation and growth of crystallizable materials. In such conditions, the variation of the contact rate of solvent and antisolvent may affect both the crystal’s internal structure and its external characteristics, which would determine the physical behavior of the materials [11]. Previous research has shown that the slow injection of antisolvent into a solution produces somewhat different particles in terms of morphology and size compared with the continuous SAS operation, which involves extremely rapid contact between solvent and antisolvent[4,6 ]. In these systems, the particles are produced in either the amorphous or spherulitic state. These results encouraged us to investigate how the contact rate of antisolvent influences the system, particularly when the particles produced are crystalline in nature and they have well-defined crystal structures. In this study, therefore, we select inorganic crystalline compounds (barium chloride and ammonium chloride) in order to investigate the impact of SAS operating conditions on the modification of the internal crystalline structure (crystal space lattice) and their external charateristics (crystal habit and size).
2. Experimental Barium chloride (BaCl ) and ammonium chlo2 ride (NH Cl ) with a minimum purity of 99.0% 4
were purchased from Aldrich Chemical Co. Dimethyl sulfoxide (DMSO) was selected as a solvent for the chlorides because of its high affinity to both compounds and the antisolvent. Carbon dioxide was employed as an antisolvent for the DMSO solutions. All the chemicals were used without further purification. A schematic diagram of the experimental apparatus for the SAS process is shown in Fig. 1. The apparatus was made exclusively for batch-type operation, which has been described in the literature [4,6 ]. In this operation, the contact between the solvent and antisolvent occurs only by the injection of carbon dioxide into the liquid solution. The unit consists of three major parts: a crystallizing chamber (Jerguson Gauge, Model 19-T-40), a carbon dioxide supply system, and a depressurizing section. The crystallizing chamber is placed in an air chamber to maintain a constant temperature during crystal formation. Solutions of 3 and 5 wt% BaCl in DMSO and 10 wt% NH Cl in DMSO 2 4 were prepared. Normally, 10 cm3 of the DMSO solution were loaded into the crystallizing chamber, and its temperature was maintained at a constant value within ±0.2°C. Carbon dioxide was injected from the bottom of the chamber to achieve a pressure of 95.0 bar. The rate of carbon dioxide injection was controlled so that the pressure could increase at different rates in the crystallizing chamber. The rate of pressure increase ranged from 0.1 to 24.5 bar/min. The introduction of carbon dioxide and the pressure increase caused a concomitant expansion of the DMSO solution and the crystallization of dissolved compounds. The expansivity of DMSO when in contact with carbon dioxide can be found in the literature [4]. During the expansion, the solution became cloudy at a certain pressure, and it was recorded as a nucleation pressure. The nucleation pressure was measured as a function of temperature and the initial concentration of the DMSO solutions. The system was pressurized until the vapor (carbondioxide-rich) and liquid (DMSO-rich) phases merged to become a single phase, at which stage the crystallization procedure was assumed to complete. The precipitated crystals were collected on a filter located at the bottom of the crystallizing chamber. The carbon dioxide and solute-free DMSO mixture was then separated from the crys-
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
237
Fig. 1. Schematic diagram of the experimental apparatus for the supercritical antisolvent (SAS ) process.
tals by the use of a ventilation valve. During the ventilation procedure, the pressure inside the chamber was kept constant at c. 95 bar, by introducing pure carbon dioxide from the top of the chamber. This was done to prevent the redissolution of the precipitated crystals in DMSO below the nucleation pressure. Finally, the residual DMSO was removed by flowing carbon dioxide continuously through the crystallizing chamber for 2 h at 37°C and 95 bar. The morphology and composition of BaCl and 2 NH Cl crystals were examined using a scanning 4 electron microscope (SEM, Hitachi, Model S-4200) equipped with an energy dispersive X-ray
spectrometer ( EDS). The internal structure of the crystals was analyzed by using a X-ray diffractometer ( XRD, Philips, Model X’PERT ). The particle size of the crystals was estimated, based on the image size of randomly selected specimens on SEM photomicrographs.
3. Results and discussion 3.1. Nucleation pressure The nucleation pressures of the BaCl and 2 NH Cl solutions in DMSO are represented as a 4
238
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
at higher temperatures. These results can be explained by considering the variation of carbon dioxide levels in the DMSO-rich phase upon changing the pressure and temperature. During the injection of carbon dioxide into the crystallizing chamber, two distinct phases are created, i.e. a carbon-dioxide-rich phase and a DMSO-rich phase. The carbon dioxide concentration in the DMSO-rich phase will continuously increase as the pressure increases inside the chamber. At the nucleation pressure, the amount of the carbon dioxide in the DMSO-rich phase corresponds to the amount of antisolvent, which causes the supersaturation. This quantity of antisolvent needed to achieve supersaturation will decrease with increasing chloride concentration in the DMSO solution, and therefore, the nucleation pressure will decrease accordingly. The effect of temperature on the nucleation pressure follows the principle of vapor– liquid equilibrium. At an elevated temperature, a higher pressure is required to maintain a sufficient carbon dioxide concentration in the DMSO-rich phase to achieve supersaturation. This explains the increase of nucleation pressure required at a higher temperature.
Table 1 Variation of crystal habit and size of BaCl and NH Cl crystals 2 4 as a function of carbon dioxide injection rate and temperaturea Injection rate (bar/min) Fig. 2. Nucleation pressure of (a) BaCl and (b) NH Cl solu2 4 tions in DMSO as a function of chloride concentration at 25 and 35°C.
function of initial chloride concentration in Fig. 2. The solutions at concentration ranges of 1–5 wt% (BaCl ) and 5–10 wt% (NH Cl ) were investigated 2 4 at two temperatures, 25 and 35°C. The nucleation pressure decreased with increasing chloride concentration in DMSO and increased with temperature at constant concentration. For BaCl 2 solutions, the dependency of nucleation pressure on the initial chloride concentration was greater
BaCl 2 0.1 0.1 0.1 24.5 24.5 24.5 NH Cl 4 0.1 0.1 0.1 24.5 24.5 24.5
T (°C )
Crystal habit
Estimated crystal size (mm)
25 35 50 25 35 50
Equant Equant Equant Acicular Acicular Acicular
7.7 6.3 9.0
25 35 50 25 35 50
Equant Equant Equant Tabular Tabular Tabular
200 280 400 1.8 1.8 4.8
a The initial concentrations of BaCl and NH Cl in DMSO 2 4 were 5 and 10 wt%, respectively
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
3.2. Effect of operating conditions on crystal habit and size The crystallizations of BaCl and NH Cl were 2 4 performed by changing various operating conditions: injection rate of carbon dioxide into the crystallizing chamber, initial chloride concentration in DMSO, and temperature. The carbon dioxide injection rate is defined as the rate of pressure increase in the crystallizing chamber in bar/min. Two injection rates were used: slow injec-
239
tion (0.1±0.05 bar/min) and rapid injection (24.5±0.5 bar/min). Figs. 3 and 4 show selected SEM photomicrographs of BaCl and NH Cl crystals: unprocessed 2 4 particles (a) and SAS processed crystals (b–d). Observations indicated that the initial chloride concentration in DMSO appeared to have little effect on particle size. However, the carbon dioxide injection rate obviously controlled both the size of crystals and the crystal habit. The crystal habit and the estimated size of BaCl and NH Cl crystals 2 4
(a)
(b)
(c)
(d)
Fig. 3. SEM photomicrographs of BaCl crystals: (a) unprocessed particles; (b) processed with the slow injection operation of 2 0.1 bar/min pressure increase; (c) magnified version of a crystal obtained in (b); (d) processed with the rapid injection operation of 24.5 bar/min pressure increase. The initial concentration of DMSO and temperature were 5 wt% and 35°C, respectively.
240
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
(a)
(b)
(c)
(d)
Fig. 4. SEM photomicrographs of NH Cl crystals: (a) unprocessed particles; (b) processed with the slow injection operation of 4 0.1 bar/min pressure increase; (c) magnified version of a crystal obtained in (b); (d) processed with the rapid injection operation of 24.5 bar/min pressure increase. The initial concentration of DMSO and temperature was 10 wt% and 35°C, respectively.
produced using the two different injection rates are summarized in Table 1. In this table, the initial concentrations of BaCl and NH Cl in DMSO 2 4 were at 5 and 10 wt%, respectively. The change of carbon dioxide injection rate resulted in a modification of the crystal habit of BaCl (Fig. 3). A slow injection produced cubic2 shaped crystals termed an ‘equant habit’ while rapid injection generated flattened needle-like crystals termed an ‘acicular habit’ [12]. The variation in crystal habit results from the alteration of the relative growth rates of crystal faces, which are
governed by a number of factors. The degree of supersaturation of the mother liquor is known as one of the main factors that affects crystal habit. In general, it is known that as supersaturation is increased, the crystal form tends to change from granular to needle-like [13]. In the case of NH Cl, an equant habit was consistently observed 4 in the slow injection mode. In a rapid injection operation, the shapes of the crystals tended to flatten and this resulted in tabular habit accompanied by a tremendous reduction in crystal size. The size of the NH Cl crystals obtained from the 4
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
slow injection experiments was almost two orders of magnitude larger than crystals produced from the rapid injection operation. It was found that the crystallization temperature had practically no influence on the crystal habit or size within the temperature ranges investigated. 3.3. Effect of operating conditions on crystal structure The composition of the BaCl crystals was 2 examined using an energy-dispersive X-ray spectrometer. Fig. 5 shows the EDS spectra of unprocessed and processed BaCl crystals. The 2 stoichiometric analysis of BaCl crystals based on 2 the integration of these spectra is summarized in Table 2. The patterns of the two spectra are identical, which implies that the chemical composition of the BaCl crystals remains unchanged after the 2 SAS process. The internal structure of the crystals was analyzed using an X-ray diffractometer with CuK a radiation. Fig. 6 shows the X-ray diffraction patterns of unprocessed and processed BaCl crystals. 2 In these X-ray diffraction patterns, distinct differences are apparent, which are attributed to change in the arrangements of the molecules in the crystal lattices. The X-ray diffraction patterns were compared with the diffraction data from JCPDS– ICDD (Joint Committee of Powder Diffraction Standards–International Centre for Diffraction Data) cards. The result revealed that the internal structure of unprocessed BaCl crystal corresponds 2 to the orthorhombic space lattice. However, the processed BaCl crystal was found to possess a 2 hexagonal crystal structure. Fig. 3(c) also shows an agreement between the external shape of a single crystal and the systematic definition of the Table 2 Percentage of elements in a BaCl crystal based on the EDS 2 analysis Elements
Unprocessed (wt%)
Processed (wt%)
Ba Cl O S
75.25 22.30 1.80 0.52
69.02 25.29 4.86 0.83
241
hexagonal space lattice in the crystallographic classification (the hexagonal crystal system is defined as: the elementary lengths and axis angles of a single crystal unit are described as a=b≠c and a=b=90°, c=120° [14]). For reference purposes, according to the JCPDS data cards, BaCl may 2 exist in three different lattice spaces, which are orthorhombic, hexagonal and cubic. The variation of experimental conditions, such as temperature, initial DMSO concentration, and the carbon dioxide injection rate, did not have any recognizable effect on the X-ray diffraction patterns. Fig. 7 shows the X-ray diffraction patterns of unprocessed and processed NH Cl crystals. The 4 results indicated that two diffraction patterns were identical, and the internal structure of NH Cl 4 crystal was unchanged by SAS operation. According to the JCPDS diffraction data cards, the unprocessed and processed NH Cl crystals 4 both belong to cubic space lattice. Moreover, the carbon dioxide injection rate did not influence the diffraction pattern, and NH Cl crystallized in the 4 cubic space lattice form, regardless of the experimental conditions used. In fact, the cubic space lattice is the only possible crystal system for NH Cl as reported in the JCPDS data cards. 4 However, it was observed that the relative intensity of each peak in the diffraction patterns varied, depending upon the experimental conditions, including the injection rate. The results suggested that the SAS operation probably alters the orientation of the crystal faces. 3.4. Separation of BaCl and NH Cl mixtures in 2 4 DMSO The SAS process enables the separation of multicomponent mixtures if the nucleation of each component occurs at different pressures. As shown in Fig. 2, the nucleations of BaCl and NH Cl take 2 4 place at different pressure ranges depending upon the temperature and the concentration of each compound. The SAS experiments were conducted employing a DMSO solution containing both BaCl (5 wt%) and NH Cl (5 wt%) at 25°C. 2 4 According to Fig. 2, the nucleation pressures of BaCl and NH Cl at the concentration of 5 wt% 2 4 were 43.5 and 48.0 bar, respectively. During the
242
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
(a)
(b)
Fig. 5. EDS spectra of the (a) unprocessed and (b) processed BaCl crystals. 2
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
243
Fig. 6. X-ray diffraction patterns of the (a) unprocessed and (b) processed BaCl crystals. The diffraction patterns (a) and (b) 2 correspond to the orthorhombic and hexagonal crystal systems, respectively.
244
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
Fig. 7. X-ray diffraction patterns of the (a) unprocessed and (b) processed NH Cl crystals. Both the diffraction patterns (a) and (b) 4 correspond to the cubic crystal system.
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
245
Fig. 8. X-ray diffraction patterns of the (a) BaCl and (b) NH Cl crystals separated from the mixture of the two compounds in 2 4 DMSO solutions. The diffraction patterns (a) and (b) correspond to the patterns of pure BaCl and NH Cl, respectively. 2 4
246
S.-D. Yeo et al. / Journal of Supercritical Fluids 16 (2000) 235–246
pressure increase (up to 95.0 bar) caused by the carbon dioxide injection, two separate nucleation points were clearly observed at two different nucleation pressures: one at c. 43.5 bar and the other at c. 48.0 bar. In order to separate the two compounds, carbon dioxide was first injected up to 43.5 bar so that only BaCl precipitated from the 2 solution. After the NH Cl-containing solution was 4 removed from the crystallizing chamber, only BaCl crystals were collected. The removed DMSO 2 solution was then recharged to the crystallizing chamber, and the SAS experiment was repeated to produce NH Cl crystals. Fig. 8 shows the X-ray 4 diffraction patterns of BaCl and NH Cl crystals 2 4 obtained from the mixture of the two compounds in DMSO solution. A comparison of these diffraction patterns to those of Figs. 6 and 7 reveals that the BaCl and NH Cl were successfully separated 2 4 from the DMSO solution and that the crystal lattices of the separated BaCl and NH Cl were 2 4 hexagonal and cubic, respectively. In this research, it has been shown that the SAS process can be a useful technology to bring about a modification of the internal structure (space lattice) and external shape (crystal habit) of crystals, which will affect the physical behavior of the crystalline particles so obtained. The various crystal space lattices exhibited by a single compound may result in different physical properties, including the solubility and dissolution rates in solvents [11]. The crystal habit modification is of importance in industrial applications because it allows control of the type of crystals produced. Certain crystal habits are undesirable in commercial crystals for many reasons, such as poor appearance, inefficient flow characteristics or difficulty in handling. In order to obtain the desired crystal habit, habit modifiers are frequently employed in crystallizing processes, especially in those involving inorganic compounds[14]. In this point of view, there is a strong possibility that the SAS process can
provide an alternative method of controlling the internal and external structure of crystals and hence regulating the physical properties of crystalline particles.
Acknowledgements Analytical instruments (SEM, EDS and XRD) for this study were provided by the Korea Basic Science Institute, CKBSI.
References [1] E. Reverchon, J. Supercrit. Fluids 15 (1999) 1. [2] P.M. Gallagher, V.J. Krukonis, G.D. Botsaris, AIChE Symp. Series 284 (87) (1991) 96. [3] P.M. Gallagher, M.P. Coffey, V.J. Krukonis, N. Klasutis, New process to recrystallize compounds insoluble in supercritical fluids, in: K.P. Johnston, J.M.L. Penninger (Eds.), Supercritical Fluid Science and Technology, American Chemical Society, Washington, DC, 1989, p. 334. [4] S. Yeo, G. Lim, P.G. Debenedetti, H. Bernstein, Biotechnol. Bioeng. 41 (1993) 341. [5] M.A. Winters, B.L. Knutson, P.G. Debenedetti, H.G. Sparks, T.M. Przybycien, C.L. Stevenson, S.J. Prestrelski, J. Pharm. Sci. 85 (1996) 586. [6 ] S. Yeo, P.G. Debenedetti, M. Radosz, H.-W. Schmidt, Macromolecules 26 (1993) 6207. [7] S. Mawson, S. Kanakia, K.P. Johnston, J. Appl. Polym. Sci. 64 (1997) 2106. [8] L. Benedetti, A. Bertucco, P. Pallado, Biotechnol. Bioeng. 53 (1997) 232. [9] R.F. Falk, T.W. Randolph, Pharmaceut. Res. 15 (1998) 1233. [10] Y. Chou, D.L. Tomasko, GAS crystallization of polymerpharmaceutical composite particles, in: Proc. 4th Int. Symp. Supercritical Fluids Vol. A (1997) 55. [11] C.J. Price, Chem. Eng. Prog. 93 (1997) 35. [12] F.D. Bloss, Crystallography and Crystal Chemistry, Holt Rinehart & Winston, USA, 1971. [13] J.K. Haleblian, J. Pharm. Sci. 64 (1975) 1269. [14] J.W. Mullin, Crystallisation, Butterworth, London, 1972.