**3.1 Solubility**

Over the last few decades, the solubilities of solids and liquids in supercritical fluids (SCF) have been measured extensively. For instance, in the facilities of University of Cádiz, the solubility of solid dyes like 1,4-dimethylaminoanthraquinone (Disperse Blue 14) in supercritical carbon dioxide has been determined in the pressure range of 100-350 bar and in the temperature range of 313-353 K and correlated with empirical and semi empirical equations based model and models based on thermodynamic aspects and the use of equations of state (Gordillo et al., 2003). The solubility of palmitic acid in supercritical carbon dioxide was determined experimentally in the pressure range, 100 to 350 bar, and the temperature range, 308 to 323 K. A cubic equation of state and an empirical equation were used to correlate the solubility of this fatty acid in supercritical carbon dioxide (Gordillo et al., 2004). Such information takes an important part of establishing the technical and economic feasibility of any supercritical fluid process. Most of the investigations on solubility have been concerned about binary systems consisting of a single solute in contact with a single SCF. The solubility of solutes in supercritical fluids is related to its physical and chemical properties such as polarity, molecular structure, and nature of the material particles, and it is also related to the operating conditions such as temperature, pressure, density of solvent and co-solvents, and solvent flow rate in the supercritical region. From the 90s to now, many

Particles Formation Using Supercritical Fluids 465

between measured and calculated solubility data varied (Gordillo et al., 2005b). Moreover, the results obtained in another work about the dye solubility correlation showed that the choice of group contribution method was more important than the choice of the equation of state used, Redlich-Kwong, Soave-Redlich-Kwong and Peng-Robinson (Gordillo et al.,

However in SAS process, according to usual practice in the ternary systems, it has been considered that the presence of non soluble drugs does not affect the solvent–CO2 equilibrium, therefore to represent the ternary equilibrium of drug–solvent–CO2, the solvent–CO2 pseudo-binary diagram is used. Volumetric expansion curves provide a mean to determine an allowed range of pressure for solubility measurements at a given temperature and for a given solvent. Thus it is possible to know prior to the analysis whether the operating conditions are above, near or below the mixture critical point (MCP). The mass transfer between CO2 and solution depends on the situation in this diagram. In Figure 2 it is shown a phase equilibrium diagram of the binary system NMP–CO2, at two temperatures, estimated using the equation of state of Peng–Robinson, for the system NMP-CO2. The data corresponding to this equilibrium diagram were obtained from the

On the other hand, the presence of the solute can induce changes in the phase diagrams of the binary solvent–SC-CO2 systems. These changes have been rarely measured and they are difficult to evaluate. However, when the solute has small interactions and a very low solubility in SC-CO2, its influence on the phase diagrams should be small (Kikic et al., 2006). In systems with strong interactions between CO2 and solute a drastic alteration of the phase

development of a computer program in Matllab 7.0 (Tenorio et al., 2008).

Fig. 2. P–x–y diagram of the CO2–NMP system (Tenorio et al., 2008)

improvement on environmental considerations, among others.

As it has been argued, the supercritical fluid technology has emerged as an important alternative to traditional processes of generation of micro and nanoparticles, offering opportunities and advantages such as higher product quality in terms of purity, more uniform dimensional characteristics, a variety of compounds to process and a substantial

2005c)

diagrams is possible.

**3.2 Precipitation** 

articles about solubility of drugs in supercritical fluids were published. At the University of Cádiz, the solubility of the antibiotic Penicillin G in supercritical carbon dioxide was measured at pressures from 100 to 350 bar and temperatures from 313.15 to 333.15 K using a dynamic flow apparatus. Moreover a new empiric equation was proposed to improve the correlation with experimental data relating neperian logarithm with pressure and temperature (Gordillo et al., 1999). The model has been applied on several systems and the obtained results allow affirm that the thermodynamic model applied to fluid–solid equilibrium calculations is useful to predict the behaviour of this system.

Kikic et al developed an estimation method based on the Peng–Robinson´s equation of state in order to calculate the solubility of drugs such as acetaminophen, acyclovir, atenolol, Carbamazepine, ibuprofen, naproxen, nimesulide, and sotalol hydrochloride in mixtures of CO2 and common organic solvents at a constant temperature but at variable pressure (Kikic et al., 2010). Wubbolts et al studied the systems p-acetamido phenol + ethanol + CO2 (Wubbolts et al., 2004). In this way, Muntó et al measured the solubility of the two nonsteroidal anti-inflammatory drugs ibuprofen and naproxen in CO2-expanded ethanol and CO2-expanded acetone. The obtained data reflected that naproxen solubility behavior was strongly dependent on the protic or aprotic nature of the organic solvent whereas for ibuprofen this solvent characteristic seemed to be less important (Munto et al., 2008). Tomasko et al. carried out a detailed review of solubilities of CO2 into polymers as well as of other thermodynamic and transport properties of CO2-polymer systems (Tomasko et al., 2003). Ugaonkar et al examined the rate of dissolution of carbamazepine, a hydrophobic drug for treating epilepsy, in supercritical CO2 and its partitioning into polyvinylpyrrolidone and concluded that partitioning occurs by surface adsorption and impregnation within the polymer matrix (Ugaonkar et al., 2011).

The choice of RESS or SAS process depends on the active substance high or low solubility in the supercritical fluid. The very low solubility of solids in carbon dioxide makes the RESS process unattractive, since a very small amount of material is processed. A solvent mixture composed of carbon dioxide and a co- solvent (Bush et al., 2007; Hosseini et al., 2010) could be an alternative, since more material could be processed at high supersaturation rates in the RESS process. It is also possible to overcome the limitation of low solubility in CO2 by employing alternative organic supercritical solvents such as trifluoromethane or clorodifluoromethane.

However, the very low solubility of solids in carbon dioxide makes the SAS process very attractive because in this process the solute must not be soluble in this fluid. So, understanding the phase behavior of solvent-supercritical fluid system can therefore provide important information regarding the role of this supercritical fluid as a solvent or reaction medium in diverse applications.

The miscibility of a dense gas with a liquid solvent is a fundamental requirement of a lot of precipitation techniques which use a gaseous or supercritical antisolvent. Vapour–liquid equilibria and volumetric expansion data for the CO2-solvent binary system are a good starting point in order to design every supercritical process. While the vapour–liquid equilibrium data of solvents and CO2 are usually available, the solubility of solids in a mixture of a common solvent and CO2 are not. Gordillo et al developed and applied a thermodynamic model to several systems and the results obtained let affirm that the thermodynamic model applied to fluid–solid equilibrium calculations was useful to predict the behaviour of this system (Gordillo et al., 2005a). These authors proved that depending on the group contribution methods chosen to estimate the parameter critical the agreement

articles about solubility of drugs in supercritical fluids were published. At the University of Cádiz, the solubility of the antibiotic Penicillin G in supercritical carbon dioxide was measured at pressures from 100 to 350 bar and temperatures from 313.15 to 333.15 K using a dynamic flow apparatus. Moreover a new empiric equation was proposed to improve the correlation with experimental data relating neperian logarithm with pressure and temperature (Gordillo et al., 1999). The model has been applied on several systems and the obtained results allow affirm that the thermodynamic model applied to fluid–solid

Kikic et al developed an estimation method based on the Peng–Robinson´s equation of state in order to calculate the solubility of drugs such as acetaminophen, acyclovir, atenolol, Carbamazepine, ibuprofen, naproxen, nimesulide, and sotalol hydrochloride in mixtures of CO2 and common organic solvents at a constant temperature but at variable pressure (Kikic et al., 2010). Wubbolts et al studied the systems p-acetamido phenol + ethanol + CO2 (Wubbolts et al., 2004). In this way, Muntó et al measured the solubility of the two nonsteroidal anti-inflammatory drugs ibuprofen and naproxen in CO2-expanded ethanol and CO2-expanded acetone. The obtained data reflected that naproxen solubility behavior was strongly dependent on the protic or aprotic nature of the organic solvent whereas for ibuprofen this solvent characteristic seemed to be less important (Munto et al., 2008). Tomasko et al. carried out a detailed review of solubilities of CO2 into polymers as well as of other thermodynamic and transport properties of CO2-polymer systems (Tomasko et al., 2003). Ugaonkar et al examined the rate of dissolution of carbamazepine, a hydrophobic drug for treating epilepsy, in supercritical CO2 and its partitioning into polyvinylpyrrolidone and concluded that partitioning occurs by surface adsorption and

The choice of RESS or SAS process depends on the active substance high or low solubility in the supercritical fluid. The very low solubility of solids in carbon dioxide makes the RESS process unattractive, since a very small amount of material is processed. A solvent mixture composed of carbon dioxide and a co- solvent (Bush et al., 2007; Hosseini et al., 2010) could be an alternative, since more material could be processed at high supersaturation rates in the RESS process. It is also possible to overcome the limitation of low solubility in CO2 by employing alternative organic supercritical solvents such as trifluoromethane or

However, the very low solubility of solids in carbon dioxide makes the SAS process very attractive because in this process the solute must not be soluble in this fluid. So, understanding the phase behavior of solvent-supercritical fluid system can therefore provide important information regarding the role of this supercritical fluid as a solvent or

The miscibility of a dense gas with a liquid solvent is a fundamental requirement of a lot of precipitation techniques which use a gaseous or supercritical antisolvent. Vapour–liquid equilibria and volumetric expansion data for the CO2-solvent binary system are a good starting point in order to design every supercritical process. While the vapour–liquid equilibrium data of solvents and CO2 are usually available, the solubility of solids in a mixture of a common solvent and CO2 are not. Gordillo et al developed and applied a thermodynamic model to several systems and the results obtained let affirm that the thermodynamic model applied to fluid–solid equilibrium calculations was useful to predict the behaviour of this system (Gordillo et al., 2005a). These authors proved that depending on the group contribution methods chosen to estimate the parameter critical the agreement

equilibrium calculations is useful to predict the behaviour of this system.

impregnation within the polymer matrix (Ugaonkar et al., 2011).

clorodifluoromethane.

reaction medium in diverse applications.

between measured and calculated solubility data varied (Gordillo et al., 2005b). Moreover, the results obtained in another work about the dye solubility correlation showed that the choice of group contribution method was more important than the choice of the equation of state used, Redlich-Kwong, Soave-Redlich-Kwong and Peng-Robinson (Gordillo et al., 2005c)

However in SAS process, according to usual practice in the ternary systems, it has been considered that the presence of non soluble drugs does not affect the solvent–CO2 equilibrium, therefore to represent the ternary equilibrium of drug–solvent–CO2, the solvent–CO2 pseudo-binary diagram is used. Volumetric expansion curves provide a mean to determine an allowed range of pressure for solubility measurements at a given temperature and for a given solvent. Thus it is possible to know prior to the analysis whether the operating conditions are above, near or below the mixture critical point (MCP). The mass transfer between CO2 and solution depends on the situation in this diagram. In Figure 2 it is shown a phase equilibrium diagram of the binary system NMP–CO2, at two temperatures, estimated using the equation of state of Peng–Robinson, for the system NMP-CO2. The data corresponding to this equilibrium diagram were obtained from the development of a computer program in Matllab 7.0 (Tenorio et al., 2008).

On the other hand, the presence of the solute can induce changes in the phase diagrams of the binary solvent–SC-CO2 systems. These changes have been rarely measured and they are difficult to evaluate. However, when the solute has small interactions and a very low solubility in SC-CO2, its influence on the phase diagrams should be small (Kikic et al., 2006). In systems with strong interactions between CO2 and solute a drastic alteration of the phase diagrams is possible.

Fig. 2. P–x–y diagram of the CO2–NMP system (Tenorio et al., 2008)

#### **3.2 Precipitation**

As it has been argued, the supercritical fluid technology has emerged as an important alternative to traditional processes of generation of micro and nanoparticles, offering opportunities and advantages such as higher product quality in terms of purity, more uniform dimensional characteristics, a variety of compounds to process and a substantial improvement on environmental considerations, among others.

Particles Formation Using Supercritical Fluids 467

**Solubilization Cell**

**<sup>2</sup> Depressurization Cell**

**BPR**

In this way, Corazza et al. carried out an analysis of supersaturation of the system during expansion of the supercritical solution. For that the phase equilibrium problem was solved at system temperature and pressure for each specific position. Supersaturation values were very high in the free-jet expansion region, but depending on the preexpansion conditions, the supersaturation profile in the free jet region was quite different. These observations suggest that pre and post expansion conditions can have a remarkable effect on the characteristics of precipitated particles (Corazza et al., 2006). The results presented in this work indicate that the fluid residence time in the capillary region was very low, and thus the mechanism for microparticle formation could also be affected by mass transfer phenomena in addition to thermodynamic equilibrium. One more time, mass transfer must be studied

In SAS process, mass transfer occurs between a droplet of organic solvent and a compressed antisolvent. In miscible conditions, above mixture critical point, there is no obvious way to define the interface between the two fluids. Dukhin et al. has evidenced the transient existence of droplets at conditions slightly above the mixture critical point, due to the existence of a dynamic interfacial tension, so a description of mass transfer from a droplet

On the other hand, in the SAS process the solution is generally dilute and the equilibrium compositions of the binary and ternary mixtures are not significantly different. Accordingly to this, the solid present in the solution is not likely to affect the rates of mass transfer of CO2

Mass transfer depend on the densities differences between solvent and antisolvent, viscosity, diffusivity, droplet or particle diameter and solvent flow rate. Chong et al. developed a mathematical model form mass transfer between a droplet of organic solvent and a compressed antisolvent in complete miscibility in SAS process. Calculations using Peng-Robinson equation of state showed that droplets swell upon interdiffusion when the

**Solute dissolved in SC-CO**

from thermodynamic and hydrodynamic point of view.

even in miscible conditions seems reasonable (Dukhin et al., 2003).

and solvent to and from the droplet respectively.

**CO2**

Fig. 4. RESS process diagram

**4. SAS process 4.1 Mass transfer** 

Previously, it was discussed that the different particle formation processes using SCF are classified depending on how this SCF behaves, i.e., the supercritical CO2 can play the role as antisolvent (AntiSolvent Supercritical process, SAS) or solvent (RESS process).

The SAS process (Figure 3) uses both the high power of supercritical fluids to dissolve the organic solvents and the low solubility of the compounds in supercritical fluids (Shekunov and York, 2000) to cause the precipitation of such compounds once they are dissolved in the organic phase. The dissolution of the supercritical fluid into the organic solvent goes along with a large volume expansion and, consequently, a reduction of the liquid density, and therefore, of its solvent power, causing a sharp rise in the supersaturation within the liquid mixture. Because of the high and uniform degree of supersaturation, small particles with a narrow particle size distribution are expected (Dukhin et al., 2005).

Fig. 3. SAS process diagram

In the RESS method, the sudden expansion of supercritical solution (solute dissolved in supercritical carbon dioxide) via nozzle and the rapid phase change at the exit of the nozzle cause a high super-saturation, thus causing very rapid nucleation of the substrate in the form of very small particles that are collected from the gas stream (Figure 4). Hence, the conditions inside the expansion chamber are one key factor to control particle size and the particles grow inside the expansion chamber to their final size. This result clarifies the influence of two important process parameters on particle size. Both, a shorter residence time and, hence, less time available for particle growth as well as a higher dilution of the particles in the expansion chamber result in smaller particles.

Harrison et al. performed RESS studies on benzoic acid, cholesterol and aspirin, in which the influence of several expansion parameters on the particle size were studied: the variation of the pre-expansion pressure and temperature, distance from the nozzle, and on the amount and type of co-solvent added. To characterize the supercritical CO2 expansion, a modelling to calculate pressure, temperature, density and velocity, along the nozzle was developed. The average particle diameter decreased with increasing pre-expansion pressure, and increased with increasing pre-expansion temperature. This is probably due to a lower mass flow rate, which is associated to a lower pre-expansion pressure or higher preexpansion temperature (Harrison et al., 2007).

Previously, it was discussed that the different particle formation processes using SCF are classified depending on how this SCF behaves, i.e., the supercritical CO2 can play the role as

The SAS process (Figure 3) uses both the high power of supercritical fluids to dissolve the organic solvents and the low solubility of the compounds in supercritical fluids (Shekunov and York, 2000) to cause the precipitation of such compounds once they are dissolved in the organic phase. The dissolution of the supercritical fluid into the organic solvent goes along with a large volume expansion and, consequently, a reduction of the liquid density, and therefore, of its solvent power, causing a sharp rise in the supersaturation within the liquid mixture. Because of the high and uniform degree of supersaturation, small particles with a

**Solvent**

In the RESS method, the sudden expansion of supercritical solution (solute dissolved in supercritical carbon dioxide) via nozzle and the rapid phase change at the exit of the nozzle cause a high super-saturation, thus causing very rapid nucleation of the substrate in the form of very small particles that are collected from the gas stream (Figure 4). Hence, the conditions inside the expansion chamber are one key factor to control particle size and the particles grow inside the expansion chamber to their final size. This result clarifies the influence of two important process parameters on particle size. Both, a shorter residence time and, hence, less time available for particle growth as well as a higher dilution of the

Harrison et al. performed RESS studies on benzoic acid, cholesterol and aspirin, in which the influence of several expansion parameters on the particle size were studied: the variation of the pre-expansion pressure and temperature, distance from the nozzle, and on the amount and type of co-solvent added. To characterize the supercritical CO2 expansion, a modelling to calculate pressure, temperature, density and velocity, along the nozzle was developed. The average particle diameter decreased with increasing pre-expansion pressure, and increased with increasing pre-expansion temperature. This is probably due to a lower mass flow rate, which is associated to a lower pre-expansion pressure or higher pre-

**Solute+Solvent**

**Separator**

**CO2**

antisolvent (AntiSolvent Supercritical process, SAS) or solvent (RESS process).

narrow particle size distribution are expected (Dukhin et al., 2005).

**Precipitator**

particles in the expansion chamber result in smaller particles.

expansion temperature (Harrison et al., 2007).

**CO2**

Fig. 3. SAS process diagram

Fig. 4. RESS process diagram

In this way, Corazza et al. carried out an analysis of supersaturation of the system during expansion of the supercritical solution. For that the phase equilibrium problem was solved at system temperature and pressure for each specific position. Supersaturation values were very high in the free-jet expansion region, but depending on the preexpansion conditions, the supersaturation profile in the free jet region was quite different. These observations suggest that pre and post expansion conditions can have a remarkable effect on the characteristics of precipitated particles (Corazza et al., 2006). The results presented in this work indicate that the fluid residence time in the capillary region was very low, and thus the mechanism for microparticle formation could also be affected by mass transfer phenomena in addition to thermodynamic equilibrium. One more time, mass transfer must be studied from thermodynamic and hydrodynamic point of view.
