Supercritical Fluid Technology for Nanotechnology in Drug Delivery

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

In the past few decades, supercritical fluid technology has attracted the attention of both scientists and engineers (McHugh and Krukonis, 1994; Taylor, 1996; Brennecke, 1993; Hutchenson and Foster, 1995; Levelt Sengers, 1991; Kendall et al., 1999). Early studies on the application of supercritical fluid technology were primarily in extraction and chromatography. Extensive experimental and theoretical investigations have been aimed toward an understanding of the properties of supercritical fluid systems, particularly intermolecular interactions (solute–solvent, solvent–solvent, and solute–solute) in supercritical fluid solutions (Tucker, 1999; Jessop and Leitner, 1999; Sun, 2002). Much progress has also been made in the use of supercritical fluids and mixtures as reaction media for chemical synthesis and as alternative solvent systems for materials processing (Sun, 2002; Poliakoff et al., 1996; Kajimoto, 1999; Savage, 1999; Musie et al., 2001). Recently, several supercritical fluid processing techniques have found significant applications in the nanotechnology development for drug formulation and delivery, especially the production of nanosized drugs and pharmaceuticals. In fact, drug formulation and delivery-related applications have emerged as a new frontier in the development of supercritical fluid technology.

In this chapter, we provide background information on the supercritical fluid processing techniques relevant to drug formulation and delivery, highlight the recent advances and novel applications, and discuss the successful development of a new supercritical fluid rapid expansion technique for producing exclusively nanoscale drug particles.

Supercritical Fluid Processing Techniques

A supercritical fluid is defined as a solvent at temperature and pressure above the critical temperature and pressure, respectively, where the fluid remains a single phase. Among the most important properties of a supercritical fluid are the low and tunable densities, which can be easily varied from gas-like to liquid-like via a simple change in pressure at constant temperature or vice versa, and the unusual solvation effects at densities near the critical density. Generally, solute–solvent interactions in supercritical fluids are understood in terms of a three-density region solvation model (Sun and Bunker, 1995). In the low-density gas-like region, the solvation increases almost linearly with density at constant temperature. This behavior is probably dictated by short-range interactions in the inner solvation shell. Before the inner shell is saturated, the microscopic consequence of increasing density is the addition of solvent molecules to the solvation shell, which causes large incremental effects. In the near-critical density region, the solvation is nearly independent of changes in density. A supercritical fluid in the near-critical region may be considered as being macroscopically homogeneous but microscopically inhomogeneous (a mixture of solvent molecules and “free volumes”). Thus, changes in the bulk density primarily correspond to decreases in the free volumes, with little effect on the solute molecules. A further increase in the fluid density to reach the point where the free volumes are largely gone affects the microscopic solvation environment of the solute molecules in a way similar to that in a liquid solution (Sun and Bunker, 1995).

Commonly used supercritical solvents include CO₂, ethylene, ethane, fluoroform, and ammonia, although the flammability and toxicity of some of these may limit their uses for specific applications in pharmaceutical processing. Supercritical CO₂ is obviously a favored choice for its near-ambient critical temperature (∼31^°C) and relatively low critical pressure (73.8 bar), and for its nontoxic, nonflammable, abundant, and inexpensive characteristics. Since CO₂ is nonpolar, a polar modifier such as a cosolvent or a surfactant may be added to improve the solubility of some solute molecules (Sauceau et al., 2004; Ting et al., 1993).

Supercritical fluid technology has shown great promise in addressing many of the challenges facing the pharmaceutical industry in drug delivery systems, including particle generation and processing techniques; and issues such as controllable particle size and shape, clean, environmentally responsible, and scalable (Rogers et al., 2001; York, 1999; Kompella and Koushik, 2001; Subramaniam et al., 1997a; Jung and Perrut, 2001; Stanton et al., 2002; Fages et al., 2004; Young et al., 2000; Tan and Borsadia, 2001; Del Valle and Galan, 2005; York, 2004; Date and Patravale, 2004; Hu et al., 2004). Several supercritical fluid methods have been successfully developed, leading to the production of micron-sized particles of different shape, size, and morphology (McHugh and Krukonis, 1994; Taylor, 1996; Brennecke, 1993; Hutchenson and Foster, 1995; Levelt Sengers, 1991; Eckert et al., 1996; Johnston and Penninger, 1989; Squires and Paulaitis, 1987; Bright and McNally, 1992; Von Rohr and Treep, 1996). These particle design and formation processes offer many drug formulation options such as dry powders, nanoparticle suspensions, microspheres or microcapsules as drug carriers, and drug-impregnated excipients (Rogers et al., 2001; York, 1999; Kompella and Koushik, 2001; Subramaniam et al., 1997a; Jung and Perrut, 2001; Stanton et al., 2002; Fages et al., 2004; Young et al., 2000; Tan and Borsadia, 2001; Del Valle and Galan, 2005; York, 2004; Date and Patravale, 2004; Hu et al., 2004). Among widely investigated and most relevant techniques are SAS (supercritical anti-solvent), RESS (rapid expansion of supercritical solutions), and more recently the RESOLV (rapid expansion of a supercritical solution into a liquid solvent) (Sauceau et al., 2004; Ting et al., 1993; Rogers et al., 2001; York, 1999; Kompella and Koushik, 2001; Subramaniam et al., 1997a; Jung and Perrut, 2001; Stanton et al., 2002; Fages et al., 2004; Young et al., 2000).

Supercritical Anti-solvent (SAS) Process

SAS generally refers to the precipitation for particle formation in a compressed fluid at supercritical as well as subcritical conditions. The process is also called PCA (precipitation with compressed anti-solvent) or GAS (gas anti-solvent) in some literature. As with any precipitation process, the anti-solvent can be added to the solution (normal-addition precipitation) or the solution can be added to the anti-solvent (reverse-addition precipitation). The SAS method requires that the supercritical anti-solvent be miscible with the solution solvent and that the solute be insoluble in the supercritical anti-solvent. In the normal-addition SAS, a solute is dissolved in a liquid solvent, and then a supercritical anti-solvent is added to the solution in a partially filled closed container that is initially at ambient pressure. With the addition of the supercritical anti-solvent, both the volume of the solution/anti-solvent mixture and the pressure of the closed container increase. The decrease in solubility of the solute with increasing anti-solvent fraction in the mixture results in the precipitation of the solute. The precipitate is then washed with the anti-solvent to yield the desired particles. The size and size distribution of the particles depend on the selection of the solution/anti-solvent system, the solution concentration, the relative solution and anti-solvent quantities, the rate of the anti-solvent addition, and the degree of mixing (Reverchon, 1999). In the reverse-addition SAS, a liquid solution is sprayed through a nozzle into a supercritical anti-solvent. The rapid diffusion of the solvent from the solution droplets sprayed into the bulk supercritical fluid results in the precipitation of the solute. The precipitate is then washed with the anti-solvent and filtered to obtain the desired particles.

Supercritical anti-solvent methods have been used for preparing a variety of micron and submicron particles and fine powders from inorganics, polymers, pigments, proteins, pharmaceuticals, and even explosives (Reverchon, 1999; Debenedetti et al., 1993a; Reverchon and De Marco, 2004; Wang et al., 2005; Winters et al., 1996; Del Valle and Galan, 2005; Jovanovic et al., 2004; Chattopadhyay and Gupta, 2002). For example, fine particles of trypsin, lysozyme, and insulin proteins with diameters ranging from 1 to 5 µm were produced by a continuous flow supercritical anti-solvent process (Winters et al., 1996). In the preparation, a solution of the protein in DMSO was sprayed through a small orifice into concurrently flowing supercritical CO₂. The particle sizes could be varied via changing processing conditions; for example, larger particles were obtained by decreasing the pressure or increasing the temperature of the supercritical anti-solvent or by using a larger-diameter expansion nozzle. The biological activity of the micron-sized powders compared to the starting materials was hardly affected by the processing.

Supercritical anti-solvent processing has also been used in the preparation of pharmaceutically important compounds, including salmetrol xinafoate (York and Hanna, 1996), sulfathiazole (Kitamura et al., 1997), methylprednisolone (Schmitt et al., 1995), and hydrocortisone acetate (Schmitt et al., 1995). In a modification to the typical supercritical anti-solvent process, step-wise addition of supercritical anti-solvent to ethanol solutions of sulfathiazole was used to control the nucleation and control processes (Kitamura et al., 1997). Sulfathiazole crystals with sizes ranging from tens to hundreds of microns, up to 2–6 mm, can be prepared by varying the timing and amount of supercritical anti-solvent added. The particle size was also found to be dependent on the operating pressure: smaller particles were formed due to the increase in the pressure of the initially added anti-solvent. The formation of smaller particles was attributed to faster nucleation during the process. In another modification to the SAS process, compressed gas was used as an anti-solvent to precipitate and prepare particles of several commercial pigments (Gao et al., 1998). Particles of ∼1 µm were obtained by using acetone as a solvent and compressed CO₂ as anti-solvent. Addition of anti-solvent as a compressed gas or a supercritical fluid did not significantly change the particle properties. However, a decrease in the pressure or an increase in the temperature of the anti-solvent (compressed gas) or an increase in the nozzle size resulted in the formation of larger particles. Antibiotic tetracycline was also successfully processed by SAS by using N-methyl-2-pyrrolidone (NMP) as the solvent. The mean particle size of precipitated particles was ∼150 nm (Reverchon and Della Porta, 1999).

Chattopadhyay et al. developed a batch supercritical anti-solvent micronization process enhanced by the addition of a vibrating surface in the precipitation vessel, named supercritical anti-solvent with enhanced mass transfer (SAS-EM) (Chattopadhyay and Gupta, 2001a,b). They produced griseofulvin (antifungal, antibiotic) particles as small as 130 nm and lysozyme (enzyme) particles of about 190 nm. Nanometric lysozyme particles with a minimum mean diameter of 180 nm were also produced by Muhrer et al. using the GAS process (Muhrer and Mazzotti, 2003). Snavely et al. produced insulin (antidiabetic) nanoparticles by SAS with the aid of an ultrasonic nozzle (Snavely et al., 2002).

Mechanistically, the particles are formed from the solution droplets, which collide into the fluid phase. Either the addition of anti-solvent into solution or vice versa, the knowledge on the phase diagram is crucial to the understanding of the mass transfer and nucleation phenomena in SAS. The precipitation of the solute occurs either by the dissolution of the anti-solvent into the solution, leading to swelling of solution droplets, or by evaporation of the solvent into anti-solvent, causing shrinkage of the droplets (Fages et al., 2004). Shekunov et al. studied ethanol–CO₂ SAS system using optical methods (image analysis, laser interferometry, and particle-image velocimetry) to determine if the anti-solvent swells or shrinks during precipitation. According to their results, nucleation occurred within the shrinking ethanol-rich droplets, and no swelling was observed under any operating conditions (Shekunov et al., 2000). Rantakylä et al. studied the influence of several parameters on the size of poly-lactic acid particles formed in SAS process (Rantakylä et al., 2002). The mean particle size was found to depend slightly on the temperature and pressure, but independent of nozzle exit velocity and nozzle diameter. The initial droplet sizes formed at the nozzle had no effect on the particle size (Rantakylä et al., 2002). Another study with the toluene–CO₂ system (without a third component to be crystallized) showed the dependence of droplet fate on the respective densities of the solvent and anti-solvent (Werling and Debenedetti, 2000). To optimize particle production from the droplets, it is generally advantageous to work in the single-phase zone at high pressures, where higher mass transfer and higher supersaturation ratios are achieved.

A major disadvantage of the SAS method is that particle formation is followed by a lengthy drying period, which often leads to particle agglomeration and aggregation (Dixon et al., 1993; Falk et al., 1997; Debenedetti, 1994; Vemavarapu et al., 2005). However, the problem may be minimized by intensive mixing of the solution with the supercritical anti-solvent, which leads to more efficient mass transfer and smaller droplet size. One way to achieve the intensive mixing in a slightly modified SAS is the use of an ultrasonic nozzle, a process in which the increased mass transfer rate due to sonic waves in an energizing gas stream leads to the formation of discrete submicron drug particles (Subramaniam et al., 1997a,b). With sound waves rather than inertial and frictional forces for droplet formation, large-diameter nozzles (instead of capillary or micro-orifice nozzles) could be used in the process for fine particle production (Subramaniam et al., 1997b). Another conceptually similar modification to the traditional SAS, the solution-enhanced dispersion by supercritical fluids (SEDS) technique (York, 1999; Hanna and York, 1998; Sloan et al., 1998), uses the supercritical fluid as both an anti-solvent and a “spray enhancer”, where a nozzle with two coaxial passages allows the simultaneous introduction of the drug solution and supercritical anti-solvent into the particle formation vessel with controlled temperature and pressure. This modified SAS process has been used to formulate a variety of drug particles, including nicotinic acid, paracetamol, salmeterol xinafoate, fluticasone propionate, and water-soluble proteins (York, 1999). The process has also allowed manipulation of particle size, shape, and morphology by changing the process working conditions. For example, polymorphs of salmeterol xinafoate and fluticasone propionate were obtained from the SEDS by processing in different regions of the supercritical phase with the same organic solvent (Hanna and York, 1998). The same process was also applied to prepare microfine particles of lysozyme from an aqueous solution of the protein (Sloan et al., 1998).

In addition to SEDS, other variations of SAS including precipitation with compressed anti-solvent (PCA) and aerosol supercritical extraction system (ASES) have been developed (Rogers et al., 2001; York, 1999; Kompella and Koushik, 2001; Subramaniam et al., 1997a; Jung and Perrut, 2001; Helfgen et al., 2000; Rogers et al., 2003). For example, Johnston and coworkers used the PCA method to prepare polymer particles and fibers (Dixon et al., 1993; Mawson et al., 1997a,b; Dixon et al., 1994; Yates et al., 1999). Microparticles and fibers of polystyrene were formed by spraying a toluene solution into liquid CO₂ (Yates et al., 1999). Increasing CO₂ density and decreasing temperature resulted in the formation of smaller particles ranging from 0.1 to 1 µm. An increase in the polymer concentration from 1 to 5 wt% resulted in formation of fibers with diameters of 20–60 µm under certain conditions.

The other variation of the SAS technique, ASES process, involves spraying fine solution droplets through an atomization nozzle into compressed CO₂. The dissolution of the supercritical fluid into the liquid droplets is accompanied by a large volume expansion and, consequently, the supersaturation and formation of small uniform particles (Dehghani and Foster, 2003; Tu et al., 2002). Foster et al. used the ASES technique to micronize and microencapsulate parahydroxybenzoic acid (p-HBA) and lysozyme with poly(L-lactic acid) (L-PLA) from various organic solutions (Tu et al., 2002). In these studies, the effects of various parameters, such as pressure, temperature, solution concentration, solvent system, and spraying velocity on the nature of the particles were determined. Effective size reduction of the particles was achieved at low-to-moderate temperatures in an essentially one-step process. In general, it was found that the high-molecular-weight compounds, L-PLA and lysozyme, precipitated as microspheres and nanospheres, whereas the lighter-weight compound, p-HBA, precipitated as crystalline particles resembling platelets (averaging 3 µm in length). The maximum encapsulation efficiencies obtained for p-HBA and lysozyme with L-PLA particles were 9.2 and 15.6%, respectively. The ASES technique has also been applied to the micronization of poorly water-soluble drugs to improve their dissolution rates, and for re-engineering of proteins, steroids, and antibiotics so that they can be administered via the respiratory tract or other routes (Reverchon, 1999; Foster et al., 2003; Reverchon, 2003; Reverchon and De Marco, 2006; Kikic and Sist, 2000). Proteins such as insulin, rhDNase, lysozyme, and albumin were precipitated by the ASES process from aqueous solutions with a uniform primary particle size of less than 500 nm (Bustami et al., 2000). The structure and biochemical integrity of the proteins could be retained, but these were dependent on the operating conditions and the modified anti-solvent used. There have also been many investigations on the feasibility of utilizing the ASES process for the processing of fragile and heat-labile molecules, such as proteins which are difficult to process by conventional techniques (Snavely et al., 2002; Foster et al., 2003; Bustami et al., 2003; Bustami et al., 2001).

SAS and related supercritical fluid processing methods are advantageous over a liquid solution-based technique for the ability to prepare dry powders in a single step (Gallagher et al., 1992). In addition, several new techniques have evolved in which the primary use of supercritical fluid is to assist the nebulization of the solution while also acting as an anti-solvent to precipitate the solute. In these techniques that closely resemble classic micronization by spray drying, the supercritical fluid and solution are intimately mixed and sprayed in a drying atmosphere. For example, carbon dioxide-assisted nebulization with a bubble dryer (CAN-BD) (Sellers et al., 2001; Sievers et al., 2003) and supercritical fluid-assisted atomization (SAA) (Reverchon and Della Porta, 2003a,b) have been used to prepare drug particles and process water-soluble proteins and vaccines. The preparation of dry powders at low temperature is particularly important for pharmaceuticals, protein samples, and other materials that are thermally labile or shock sensitive. Because many compounds of interest have higher solubility in liquid solvents than in low-temperature supercritical fluids, the SAS method generally allows higher throughputs than the RESS method discussed below.

Rapid Expansion of Supercritical Solutions (RESS)

RESS process has been widely studied as an effective technique for particle formation (McHugh and Krukonis, 1994; Rogers et al., 2001; York, 1999; Kompella and Koushik, 2001; Subramaniam et al., 1997a; Jung and Perrut, 2001; Stanton et al., 2002; Fages et al., 2004; Krukonis, 1984; Lele and Shine, 1994; Matson et al., 1986a, 1987a,b; Teja and Eckert, 2000; Blasig et al., 2002; Yeo and Kiran, 2005; Mohamed et al., 1989b; Domingo et al., 1997; Petersen et al., 1986, 1987; Mawson et al., 1995; Weber and Thies, 2002; Ginosar et al., 2000; Rollins, 1999; Leuner and Dressman, 2000; Matson et al., 1986b, 1987c,d; Smith et al., 1986; Matson et al., 1989; Matson and Smith, 1989; Burukhin et al., 1998; Williams et al., 1998; Yeo et al., 1994; Lele and Shine, 1992; Mishima et al., 2000; Krober et al., 2000; Aniedobe and Thies, 1997). It differs from the SAS process in that in RESS the solute is dissolved in a supercritical fluid to form a solution, and then the solution is rapidly expanded through a small nozzle or orifice into a region of lower pressure or ambient air (Eckert et al., 1996; Teja and Eckert, 2000; Blasig et al., 2002; Yeo and Kiran, 2005; Mohamed et al., 1989b). The rapid reduction in pressure and thus density results in the precipitation of the solute. Experimentally, the supercritical solution can be generated either by heating and pressurizing a solution from room temperature or by continuously extracting the solute using an extraction column. The preparation of the solution at room temperature allows expansion to be performed at a known constant concentration, whereas the use of an extraction column is useful for solutes that are insoluble or sparingly soluble in the solvents. The temperature of the solvent can be the same as or different from the temperature at which the expansion is carried out. The extraction temperature and flow rate may be used to control the solute concentration. The expansion is driven by the drop in pressure, which can propagate at speeds up to the speed of sound in the expansion nozzle. Because solubilities in supercritical fluids can be up to 10⁶ times higher than those under ideal gas conditions, the rapid expansion from supercritical pressure to ambient pressure leads to extreme supersaturation. The rapid decrease in pressure results in homogeneous nucleation of the solute, leading to narrow size distributions in the processed materials. This technique can be used for the preparation of particles and films of inorganic, organic, and polymeric materials. Fibers of polymeric materials can also be prepared under the appropriate expansion conditions. Thus, variations in expansion conditions allow modifications in the product morphology (Rollins, 1999).

The first report on RESS processing was published by Krukonis in 1984 (Krukonis, 1984). The preparation of small particles or fibers of aluminium isopropoxide, dodecanolactam, polypropylene, β-estradiol, ferrocen, navy blue dye, and soybean lecithin using RESS was reported (Mohamed et al., 1989b). CO₂ was used as a supercritical solvent for all the samples except polypropylene, which was processed with propylene. Primary products for most of the samples were micron-size particles, with a few exceptions like dodecanolactam and polypropylene, for which needles ∼30 µm long and 1 µm diameter and fiber-like particles were obtained. Smith and coworkers carried out a series of RESS experiments aimed at evaluating the effects of processing conditions on product morphology from particles to films, size, and distribution (Matson et al., 1986b, 1987c,d; Smith et al., 1986).

Besides inorganic particles and films, several polymers have also been processed into particles and fibers via RESS. For example Matson, Smith, Peterson et al. used supercritical pentane for the RESS processing of polystyrene, polypropylene, poly(carbosilane), and cellulose acetate (Matson et al., 1987a; Petersen et al., 1986, 1987; Matson et al., 1986b, 1987d; Mishima et al., 2000). In these investigations, pre-expansion temperatures significantly higher or lower than the melting point of the polymer led to the formation of particles, while pre-expansion temperatures close to the polymer melting point resulted in fiber morphology. Lele and Shine reported preparation of fibers and particles of polycaprolactone, poly(methyl methacrylate), and a styrene/methyl methacrylate block copolymer using chlorodifluoromethane (Lele and Shine, 1992, 1994). The group found that the formation of particles was favored over fibers at a low polymer concentration, low pre-expansion temperature, high pre-expansion pressure, and small length/diameter ratio for the nozzle. The results were explained on the basis of the location of phase separation in the nozzle: particles if the precipitation of polymer-rich phase occurs late in the nozzle and fibers if the precipitation occurs upstream in the nozzle. Thies and coworkers (Aniedobe and Thies, 1997) also reported the impact of polymer concentration on product size and morphology for cellulose acetate in supercritical methanol and fluoroacrylate in supercritical CO₂. They found a transition from continuous fibers to particles as the polymer concentration decreased from more than 5 wt% to less than 1 wt% (Aniedobe and Thies, 1997).

In addition to production of inorganic and polymeric materials, RESS has been used for processing organic materials and pharmaceuticals. Naphthalene was initially used as a model compound to study the effects of RESS processing conditions on product size and morphology (Mohamed et al., 1989a; Liu and Nagahama, 1996; Tai and Cheng, 1997). CO₂ was used as a supercritical solvent, and the effects of pre- and post-expansion temperatures and pressures and naphthalene concentration were evaluated. The particle size increased from 2–35 µm to 4–51 µm with the increase in the pre-expansion temperature from 110 to 170^°C. This dependence was more pronounced at higher solute concentrations, and both particle size and size distribution decreased with the decrease in the post-expansion temperature. The effect of particle morphology as a function of particle composition was also studied for naphthalene and phenanthrene particles. These particles exhibited a systematic change from plate-like particles for pure naphthalene to needle-like particles for pure phenanthrene. Similar studies on the effects of RESS processing conditions on product morphology of salicylic acid were evaluated by Reverchon et al. (Reverchon et al., 1993). The primary product morphology of salicylic acid was needles with diameters in the 1–10 µm range and lengths of 5–170 µm. The needle diameter and length both increased with the increase in the pre-expansion temperature from 100 to 140^°C. A decrease in the post-expansion chamber temperature from 30 to –10^°C resulted in a transition from needle-like to particle-like products. Concentration effects were also seen in the case of salicylic acid: an increase in the concentration by a factor of 4 for a decrease in needle diameter from 1 to less than 1 µm and a corresponding decrease in needle length from 5–15 to 3–6 µm. The results were in good agreement with those in the formation of naphthalene particles.

Ohgaki et al. also observed the formation of crystalline whiskers or needles when processing sigmasterol in supercritical CO₂ at pre-expansion pressures between 13 and 15 MPa, which changed to amorphous agglomerates at lower pre-expansion pressures (Ohgaki et al., 1990). The formation of agglomerates and needles was attributed to the subsequent growth of the primary particles (∼10 nm in diameter) on the substrate, and the difference in the growth mechanisms was thought to be due to differences in the surface properties of the crystals. Needles that collected on the substrate after 10 min exposure time using a pre-expansion pressure of 15 MPa had a diameter of 0.2 µm and lengths of 2–3 µm.

In another report, an antimicotic compound griseofulvin was processed using trifluoromethane as a supercritical solvent in RESS (Reverchon et al., 1995; Donsi and Reverchon, 1991). As in previous studies, the product morphology changed from needles to particles with increases in pre-expansion temperature. For example, lower pre-expansion temperature of 60^°C resulted in the formation of griseofulvin needles with 1.1 µm diameter and lengths of 13–36 µm; but at the higher pre-expansion temperature of 150^°C, only particles with a diameter of 1.1 µm were formed.

The effects of RESS processing conditions on the particle formation of steroid drugs progesterone and medroxyprogesterone acetate were also studied by using supercritical CO₂ (Alessi et al., 1996). For progesterone, the particle diameter decreased from 7.4 to 4.1 µm with the use of a smaller diameter nozzle of 30 µm instead of 100 µm. The increase in progesterone concentration from 4.55 × 10^–5 to 1.35 × 10^–4 mole fraction resulted in a decrease in particle size from 7.5 to 6.0 µm, and an increase in the post-expansion temperature from 40 to 60^°C resulted in a slight increase in particle size. The authors also found that a decrease in the post-expansion pressure from 50 to 1 bar resulted in a change of particle morphology from needle-like with a dendritic structure to more particle-like, and only a slight decrease in size, from 9.1 to 7.5 µm, with little effect on size distribution. Similarly, for n-octacosane processed using CO₂, an increase in pre-expansion temperature resulted in decreases in particle size and distribution (Griscik et al., 1995).

Foster et al. employed RESS processing for the micronization of ibuprofen, an anti-inflammatory drug (Charoenchaitrakool et al., 2000). They studied the effect of spraying distance, pre-expansion pressure, and nozzle length on particle size. The median particle size of ibuprofen precipitated by RESS was less than 2.5 µm. Although the particles obtained were aggregated, they were easily dispersed by ultrasonication in water. The pre-expansion pressure and nozzle length had no effect on particle size and morphology within the range of operating conditions studied. An increase in spraying distance resulted in a slight decrease in particle size and degree of aggregation. The degree of crystallinity of the processed ibuprofen was slightly decreased; as a result, the micronized product exhibited a higher disk intrinsic dissolution rate. The RESS process was also successful in processing ultrafine spherical particles (0.1–0.3 µm) of aspirin, as demonstrated by Huang et al. (Huang et al., 2005). Their results showed that extraction pressure and extraction temperature could significantly affect the morphology and size of the precipitated particles, whereas the nozzle diameter and pre-expansion temperature did not influence the RESS-produced particles.

One of the major limitations of the RESS process with supercritical CO₂ is the low solubility of polar drugs. To overcome this limitation, a modified process of rapid expansion of supercritical solution with solid cosolvent (RESS-SC) was proposed by Gupta (Thakur and Gupta, 2005, 2006a,b). The solid cosolvent (SC) enhances the solubility, provides a barrier for coagulation in the expansion chamber, and can be removed from the solute particles later by lyophilization (sublimation). For example, with the use of menthol as cosolvent, griseofulvin (an antifungal drug) solubility in supercritical CO₂ was increased by about 28-fold. Griseofulvin particles in the range of 50–250 nm were obtained by using RESS-SC, which is a 10-fold reduction in size compared with those from the conventional RESS processing (Thakur and Gupta, 2005). The same process was also applied to phenytoin drug using menthol solid cosolvent, resulting in a 400-fold increase in the phenytoin solubility (Thakur and Gupta, 2006b).

Similarly, Johnston et al. modified the RESS process by rapid expansion from supercritical to aqueous solution (RESAS), which produced nanosuspensions of water-insoluble drugs (Young et al., 2000; Hu et al., 2004; Chen et al., 2002). Phospholipids and other suitable surfactants were integrated into the process to increase the solute solubility and enhance the dispersion of the particles. The Johnston group reported the formation of cyclosporine nanoparticles by spraying the drug–CO₂ solution into a Tween-80 solution to prevent nanoparticle agglomeration. Cyclosporine particles formed by RESAS had a size of 500–700 nm and could be stabilized for drug concentrations as high as 6.2 and 37.5 mg/ml in 1 and 5 wt% Tween-80 solutions, respectively.

Besides CO₂, several other supercritical solvents, like ethane and ethylene, have been used in RESS processing. Supercritical ethane, ethylene, and ethylene/1.5% toluene cosolvent systems were used for the RESS processing of β-carotene particles with diameters of less than 1 µm (Chang and Randolph, 1989). β-Carotene was dissolved in supercritical ethylene and then expanded into air under pre-expansion conditions of 316 bar and 70^°C to produce particles with an average size of 1.0 µm and size distribution of 36%. Under similar pre-expansion conditions, when the expansion was expanded into a 10 wt% aqueous gelatin solution, small particles of 0.3 µm size and 34% size distribution were obtained.

Domingo and coworkers investigated the effect of nozzle characteristics on the particle properties of benzoic acid, salicylic acid, aspirin, and phenanthrene with supercritical CO₂ (Domingo et al., 1996, 1997). The authors used porous frits as alternatives to capillary nozzles and evaluated the particle morphologies as a function of processing conditions (Domingo et al., 1996, 1997). Although porous frit nozzles resulted in smaller particles with particle size of 0.1 µm, clogging of frits was one of the main problems encountered. The clogging was prominent in the case of aspirin, which could not be processed using frit nozzles. On the other hand, the use of capillary nozzles resulted in more needle-like particles with an average particle size of 2–8 µm.

RESS technique has also been applied to the production of multi-component particles for various applications, particularly for the controlled release of pharmaceuticals, medical devices, etc. (Reverchon and Adami, 2006). The method allows preparation of intimately mixed samples of components with very different properties. The most promising area of application for these composite materials is in the controlled release of drugs using polymer particles. For example, RESS processing of several polyhydroxy acids for controlled release was reported by Tom and Debenedetti (Tom and Debenedetti, 1991). Polymer microparticles of bioerodible polymers poly(L-lactic acid), poly(D,L-lactic acid), and poly(glycolic acid) with diameters ranging from 2 to 25 µm were successfully prepared via RESS processing with neat CO₂, CO₂ with 1 wt% acetone cosolvent, and chlorotrifluoromethane. Benedetti et al. also did a comparative study of RESS and SAS methods for the preparation of micron-size particles of the biocompatible polymer hyaluronic acid benzylic ester (Benedetti et al., 1997).

The ideal targeted morphology for controlled release applications is uniform spherical drug particles homogeneously distributed throughout the polymer particles. RESS process was applied for the preparation of uniformly distributed pyrene within the poly(L-lactic acid) microparticles (Tom et al., 1994; Knutson et al., 1996). Pyrene was chosen as a model organic solute because the pyrene distribution in poly(L-lactic acid) could be easily determined using fluorescence microscopy. Poly(D,L-lactic acid) was extracted using CO₂-CHClF₂ (60 wt% CO₂) at 200 bar and 55^°C in one column, and pyrene was extracted with neat CO₂ at 200 bar and 65^°C in the other column. The pyrene concentration was adjusted by diluting the extract to the desired concentration using CO₂. Both extraction columns were used in parallel and the two solutions were mixed and brought to the pre-expansion temperature and expanded through a 50 μm capillary nozzle. The uniform distribution of pyrene inside poly(D,L-lactic acid) microspheres was evident from the results.

Lovastatin, an anti-cholesterol drug, was successfully encapsulated in poly(D,L-lactic acid) microparticles by Debenedetti et al. (Debenedetti et al., 1993b). Two separate extraction columns for the solutes were used to prepare encapsulated particles. Simultaneous extraction of the polymer and drug solutes using supercritical CO₂ at 200 bar and 55^°C followed by expansion at 75–80^°C produced a range of polymer particles from spherical to oblong, containing lovastatin needles. A single extraction column packed with a mixture of both solutes was also used for the processing. However, the use of separate extraction columns for individual solutes was preferred because separate columns allow better control over the solute extraction parameters and final product properties.

Naproxen (6-methoxy-methyl-2-naphthaleneacetic acid), an anti-inflammatory drug, was used with poly(L-lactic acid) to prepare encapsulated microparticles (Kim et al., 1996). Simultaneous extraction of both naproxen and poly(L-lactic acid) solutes was accomplished by using a single extraction column with supercritical CO₂. The solution containing both the solutes was co-precipitated at pre-expansion conditions of 190 bar and 114^°C using a 50 µm diameter capillary nozzle. Naproxen microparticles ranging from size 2 to 5 µm were found to be embedded inside poly(L-lactic acid) spheres with 10–90 µm diameter.

The RESS techniques have been applied to processing of other pharmaceutical compounds, such as salicylic acid, aspirin, lecithin, nifedepin, lovastatin, paracetamol, ketoprofen, nicotinic acid, and drug–polymer systems (York, 1999; Kompella and Koushik, 2001; Subramaniam et al., 1997a; Jung and Perrut, 2001; Vemavarapu et al., 2005; Subramaniam et al., 1997b; Hanna and York, 1998; Sloan et al., 1998; Helfgen et al., 2000; Rogers et al., 2003). The morphology of the prepared materials was found to be dependent on the different processing conditions, which allows some control over the materials properties. Although particles of a few microns are generally obtained as primary products, smaller particles (100–300 nm) have been produced in RESS with the use of appropriate nozzles (Jung and Perrut, 2001). To better understand the relationships between processing conditions and materials properties, numerous efforts have been made to model and simulate the supercritical fluid processing technique. The results from these studies suggest that traditional RESS process generally produces micron-size particles with only few exceptions (Lele and Shine, 1992, 1994; Matson et al., 1987a; Petersen et al., 1986; Matson et al., 1986b, 1987c,d; Smith et al., 1986; Matson et al., 1989; Matson and Smith, 1989; Burukhin et al., 1998; Williams et al., 1998; Yeo et al., 1994). Mechanistically, several theoretical studies suggest that the traditional RESS should facilitate the formation of primarily nanoscale particles and that agglomerations during the rapid expansion process are responsible for the growth of the initially formed particles beyond the nanoscale and thus for the observation of micron-size particles as major products (Helfgen et al., 2000; Weber and Thies, 2002). Ginosar and coworkers have shown experimentally that both nanosize and micron-size particles are present in the expansion jet and that the particle sizes increase with the distance from the expansion nozzle, consistent with the theoretically predicted agglomeration mechanism (Ginosar et al., 2000). For example, they found that the nanosize particles increased in size from an average diameter of 10 to 22 nm as the distance from the expansion nozzle increased from 0.26 to 5 mm, while micron-size particles maintained a relatively constant average diameter of ∼ 0.7 μm (Ginosar et al., 2000). These theoretical and experimental results suggest that nanoscale particles could be obtained from the RESS process if the nanoparticles could be effectively “captured”.

RESOLV and Nanotechnology

Sun and coworkers made a procedure-wise simple modification, but mechanistically significant, to the classical RESS processing by rapidly expanding a supercritical solution into a liquid instead of air or gas phase, namely the rapid expansion of a supercritical solution into a liquid solvent (RESOLV), as illustrated in Scheme 1 (Sun and Rollins, 1998; Sun et al., 1999a,b). Alternatively in some cases, a reacting system has been used, in which one reactant is dissolved in the supercritical solution and one in the liquid receiving solution. However, these variations of RESOLV all involve expansion into a liquid to produce exclusively nanoscale particles, and the nanoparticles thus produced are generally of narrow size distributions (Pathak et al., 2004, 2005, 2006, 2007a,b; Sun et al., 2000, 2001, 2005; Meziani and Sun, 2002; Meziani and Sun, 2002, 2003; Sane and Thies, 2005; Meziani et al., 2003, 2005a,b, 2002, 2006).

Scheme 3.1

RESOLV for processing drug nanoparticles. (See Color Plate 5)

RESOLV for Nanoscale Semiconductor and Metal Particles

Semiconductor and metal nanoparticles were prepared by using RESOLV with a reacting system. While not necessarily applicable to the processing of drug nanoparticles, the results are highlighted to demonstrate the feasibility and potentials of the technique. A simple example is the preparation of cadmium sulfide (CdS) nanoparticles. In the experiment, a supercritical ammonia solution of Cd(NO₃)₂ was prepared for rapid expansion into a room-temperature ethanol solution of Na₂S. The reaction of Cd²⁺ ions contained in the sprayed supercritical ammonia solution with S²⁻ in the receiving liquid ethanol solution resulted in the formation of CdS nanoparticles (Sun and Rollins, 1998). The same method has been applied to the preparation of other semiconductor (PbS, Ag₂S, ZnS), metal (Cu, Ni, Co, Fe), and metal oxide (Fe₂O₃) nanoparticles (Sun et al., 1999a,b, 2000).

An important feature of the RESOLV method is that it requires no nanoscale templates (such as nanoscale cavities in micelles) for the formation of nanoparticles. The supercritical fluid rapid expansion itself provides the templating effect, creating nanoscale droplets of the expanding supercritical fluid solution in the receiving solution (Sun et al., 1999a,b, 2000). Thus, the method offers a “clean” way to produce nanoscale materials that are conjugated directly with biological species (Meziani et al., 2002, 2005b; Meziani and Sun, 2003). For example, it was demonstrated that the semiconductor and metal nanoparticles coated directly with natural protein species could be prepared via RESOLV process (Meziani et al., 2002, 2005b; Meziani and Sun, 2003). Shown in Figure 3.1 is a TEM image of the BSA-conjugated Ag₂S nanoparticles, corresponding to an average particle size of 6.3 nm and a size distribution standard deviation of 1.6 nm. The result from atomic force microscopy (AFM) analysis is also shown in Figure 3.1.

Figure 3.1

TEM (left) and AFM (right) images of BSA-conjugated Ag₂S nanoparticles prepared via RESOLV. (From Meziani and Sun, 2003.) (See Color Plate 6)

Supercritical CO₂ with its favorable properties is obviously the preferred clean solvent in RESOLV. However, the limitation of neat supercritical CO₂ is that it mostly dissolves only nonpolar solutes. Sun and coworkers have investigated the use of other CO₂-based supercritical solvent systems with RESOLV, including the stabilized CO₂ microemulsions (Sun et al., 2001; Meziani and Sun, 2002; Meziani et al., 2003, 2005a). For example, in an evaluation of the feasibility as supercritical solvent systems in the preparation of nanoparticles via RESOLV, the supercritical CO₂-based microemulsions stabilized by perfluoropolyether ammonium carboxylate (PFPE-NH₄) have been used to dissolve silver and copper salts in RESOLV for the preparation of pure nanoscale silver and copper nanoparticles (Sun et al., 2001; Meziani and Sun, 2002; Meziani et al., 2003, 2005a). These metal nanoparticles, as well as nanoscale metal sulfides (Ag₂S, CdS, and PbS), have been considered as model systems for the evaluation of the microemulsion-for-RESOLV approach.

There have also been systematic investigations on the experimental conditions and parameters required or suitable for the dissolution of different metal salts in water-in-CO₂ microemulsions (Meziani and Sun, 2002). In a RESOLV experiment for pure silver nanoparticles, as an example, AgNO₃ was dissolved in the water cores of the PFPE-NH₄-protected reverse micelles in supercritical CO₂. The microemulsion containing the aqueous AgNO₃ was rapidly expanded into an ambient solution with a reductant in the classical RESOLV arrangement for the nanoparticle formation. The results from this approach are more complicated than those from the use of neat supercritical solvent discussed in previous sections. The particle size distributions are generally broader with the rapid expansion of the microemulsion when the receiving liquid solution is close to neutral pH (Meziani et al., 2005a). However, in an optimization of the process, it was found that the use of basic liquid receiving solution resulted in two sets of silver particles, each of which with a reasonably narrow size distribution. Shown in Figure 3.2 are TEM images of the two sets of silver nanoparticles obtained from the basic receiving solution conditions (pH ∼ 11). The separation of the two sets of nanoparticles was accomplished via simple centrifuging. The set of a smaller average particle size represents the initially formed nanoparticles, and the other set of a larger average particle size probably represents those that grow from the initially formed nanoparticles. Such an assignment is supported by the fact that the average size of the set of smaller nanoparticles is dependent on the water-to-surfactant ratio (or the Wo value) of the pre-expansion microemulsion. A higher Wo value corresponds to a larger average size of the initially produced silver nanoparticles (Figure 3.3 ). This nice size correlation suggests that the reverse micelles in supercritical CO₂ before the rapid expansion still play a role in the post-expansion formation of nanoparticles, which may have significant implications to the effort on controlling particle sizes. Although the use of supercritical CO₂-based microemulsions in RESOLV complicates the overall technical platform, it does offer important flexibility in the nanoparticle production.

Figure 3.2

TEM images of the two sets of Ag nanoparticles. (From Meziani et al., 2005a.)

Figure 3.3

Variation of Ag nanoparticle sizes with the pre-expansion Wo values. (From Meziani et al., 2005a.)

RESOLV for Nanoscale Polymeric and Organic Particles

Polymeric nanoparticles (100 nm or less), especially those that are biocompatible and/or biodegradable, have attracted much attention for their various applications. In the field of drug delivery, as an example, these nanoparticles may be used to carry a wide range of drugs, proteins, vaccines, or other biological species and for the purpose of controlled release of drugs. Available methods for preparing polymeric nanoparticles include solvent-in-emulsion evaporation, phase separation, and spray drying. These methods are limited by issues such as excessive use of surfactant and solvent, unwanted impurities, and insufficient colloidal stability. Thus, it remains an ongoing challenge to find reliable and versatile techniques for clean and nontoxic production of well-dispersed organic and polymeric nanoparticles. RESOLV has been demonstrated as such a technique with the CO₂-soluble fluoropolymer poly(heptadecafluorodecyl acrylate) (PHDFDA) as model polymeric solute (Meziani et al., 2004; Sun et al., 2005). Experimentally, the fluoropolymer PHDFDA was dissolved in supercritical CO₂. The supercritical solution was rapidly expanded into ambient pure water to yield nanoscale PHDFDA particles. In order to prevent the particles from agglomeration, polymeric and surfactant stabilization agents were used in the aqueous suspension. For example, when an anionic surfactant sodium dodecyl sulfate (SDS) was presented, even at a relatively low concentration (20 mM), the aqueous-suspended PHDFDA nanoparticles could be protected to remain exclusively nanoscale, about 40 nm in average diameter according to scanning electron microscopy (SEM) analysis (Figure 3.4 ). Similarly, water-soluble polymers can be effective stabilization agents for the initially formed aqueous suspension of polymeric nanoparticles in RESOLV. In the presence of poly(vinyl alcohol) (PVA) in the aqueous receiving solution, as an example, the PHDFDA nanoparticles from RESOLV remained homogeneous without any significant precipitation (Sun et al., 2005). An SEM image of the PVA-protected PHDFDA nanoparticles is also shown in Figure 3.4.

Figure 3.4

SEM images of poly(HDFDA) nanoparticles from RESOLV with aqueous SDS solution (left) and aqueous PVA solution (right) at the receiving end of rapid expansion. (From Meziani et al., 2004.)

The same RESOLV processing was applied to produce nanoparticles from biodegradable and biocompatible polymers, including poly(L-lactic acid) (PLA) (Figure 3.5 a) and poly(methyl methacrylate) (PMMA) (Figure 3.5b), for stable aqueous suspensions of the nanosized particles (Meziani et al., 2006). For organic materials, the feasibility of RESOLV for the production of nanoscale particles has been demonstrated with 5,10,15,20-tetrakis(3,5-bis(trifluoromethyl) phenyl) porphyrin (TBTPP) as a model compound (Figure 3.5c) (Sane and Thies, 2005).

Figure 3.5

SEM images of the SDS-protected PLA nanoparticles (a), SDS-protected PMMA nanoparticles (b), and pluronic-protected TBTPP nanoparticles (c). (From Sane and Thies, 2005; Meziani et al., 2006.)

RESOLV for Nanosizing Drug Particles

For the formulation and delivery of drugs that are insoluble or poorly soluble in water, particle size reduction has emerged as a promising strategy. Particulate drug delivery systems are useful for administering drugs by various routes, including aqueous suspensions for oral, parenteral, or topical applications and dry powders for inhalation and other uses. The purpose of these delivery systems can be either immediate release or sustained release of the therapeutic agent conjugated with a polymer excipient. Since the particle size and morphology of a drug may directly affect its pharmacokinetics, it is important to tailor the drug particle size in a dosage to the route of administration. The RESOLV technique has demonstrated great potential in the production of exclusively nanoscale drug particles (Pathak et al., 2004, 2005, 2006, 2007a,b). Various anti-inflammatory (ibuprofen and naproxen), antifungal (amphotericin B), and anti-cancer (paclitaxel) drugs were selected for demonstration, for which CO₂ and CO₂–cosolvent systems were used. The experimental procedure of RESOLV for the production of drug nanoparticles was similar to that for polymeric nanoparticles. The drug particles produced in the RESOLV process are exclusively nanoscale (less than 100 nm), but without protection, these suspended nanoscale particles can form larger aggregates and precipitate. In the case of ibuprofen, naproxen, and paclitaxel nanoparticles, the water-soluble polymer poly(N-vinyl-2-pyrrolidone) (PVP) was found to be an effective stabilization agent. Figure 3.6 shows homogeneously distributed ibuprofen (40 nm in average diameter) and naproxen (64 nm in average size) nanoparticles obtained from RESOLV with aqueous PVP solution at the receiving end of the rapid expansion. The average particle size and size distribution were both found to be dependent on the PVP concentration in the aqueous receiving solution and PVP average molecular weight.

Figure 3.6

SEM images of the naproxen (left) and ibuprofen (right) nanoparticle samples obtained from RESOLV with the expansion into aqueous PVP solution. (From Pathak et al., 2004.)

Various other water-soluble polymers, surfactants, and food emulsifiers can also be used to protect the drug nanoparticles produced in RESOLV and stabilize the aqueous suspensions for formulation and other purposes (Pathak et al., 2006). The investigated stabilization agents, such as PVP, sodium dodecyl sulfate (SDS), poly(ethylene glycol) (PEG), and bovine serum albumin (BSA) protein are all biocompatible and/or biodegradable additives, with an accepted GRAS (generally referred as safe) status. In fact, these polymers have already found other biological and biomedical applications, such as drug-encapsulated polymeric particles and drug–polymer composites. The particle properties were found to be dependent on the type and properties of the stabilization agent used (Table 3.1 ). Figure 3.7 shows ibuprofen particles of various sizes obtained by using various stabilization agents. The difference in particle sizes might be due to the further growth of the initially formed drug particles, for which the less effective protection by the stabilizer might be responsible. Conjugation of drug nanoparticles with different stabilization agents (thus different surface properties) can have important implications in the delivery and targeting of these nanoparticles. For example, PEG-protected drug nanoparticles are often referred to as “stealth nanoparticles”, because they can prevent recognition and clearance from the blood stream to achieve longer circulation. Beyond the use of different stabilization agents, the properties of the drug nanoparticles could also be altered by other RESOLV experimental conditions (such as temperature, pressure, drug concentration, cosolvent amount). For example, while a higher pre-expansion pressure had only marginal effect, an increase in the pre-expansion temperature resulted in larger particle sizes (Pathak et al., 2004). Similarly, an increase in the dissolved drug concentration in supercritical CO₂ resulted in larger nanoparticles with different particle morphology (Pathak et al., 2006).

Figure 3.7

SEM images of the ibuprofen nanoparticles obtained with SDS (a), PEG (b), and BSA (c) as stabilization agents in RESOLV. (From Pathak et al., 2006.)

Table 3.1 Parameters of ibuprofen nanoparticles with different stabilization agents.

Nanosizing Amphotericin B Particles

Amphotericin B (AmB), an amphipathic polyene macrolide, is practically insoluble in water but soluble in CO₂ with DMSO or methanol as a cosolvent (less than 4 vol%). These cosolvents make homogeneous mixtures with CO₂ at moderate temperature and pressure and are miscible with water, and they belong to class 3 (nontoxic) in the pharmaceutical guidelines. The solubility of AmB in the CO₂–cosolvent systems was evaluated spectroscopically under various temperature and pressure conditions by using a high-pressure optical cell with quartz windows (Meziani and Sun, 2002). In a typical RESOLV experiment, a solution of AmB in DMSO (3 mg in 0.57 mL) was added to the syringe pump, followed by filling the pump with liquid CO₂ to a pressure of 310 bar (total volume 50 mL, DMSO volume fraction 1%). The solution was pushed through the heating unit to reach the desired supercritical temperature of 40^°C before the expansion nozzle. The rapid expansion was carried out at pre-expansion pressure of 310 bar through a 50-µm orifice into ambient water. The as-produced aqueous suspension of AmB nanoparticles appeared homogeneous (Figure 3.8 ) and it remained stable for an extended period of time. The suspension was used to prepare specimen for SEM analyses. The SEM images suggest that the RESOLV process produced nanoscale AmB particles (Figure 3.9 ), with an average particle size of 38 nm diameter (particles treated as spheres) and a size distribution standard deviation of 7 nm.

 

Figure 3.8

AmB nanoparticles from RESOLV in a stable and optically transparent aqueous suspension. (From Pathak et al., 2007a.) (See Color Plate 7)

Figure 3.9

SEM images (different resolutions) of the AmB nanoparticles from RESOLV with DMSO as cosolvent into neat water. (From Pathak et al., 2007a.)

The use of methanol as a cosolvent with supercritical CO₂ in the RESOLV processing of AmB yielded similar nanoscale particles. The solution for rapid expansion contained 2 vol% methanol with an AmB concentration of 0.05 mg/mL, and the expansion was at 40^°C and 310 bar into ambient water. The as-produced aqueous suspension, appearing homogeneous initially, was used to prepare specimen for SEM analyses. The average particle size of 39 nm and the size distribution standard deviation of 8.5 nm from the SEM images (Figure 3.10) are comparable with those of the AmB nanoparticles obtained with DMSO as cosolvent. However, a significant difference is that the aqueous suspension from methanol as cosolvent was not stable, exhibiting precipitation soon after its formation. The precipitate was aggregates of AmB nanoparticles, as suggested by the SEM results. The different stabilities in the aqueous suspensions of the nanoparticles associated with the use of different cosolvents is an interesting topic for further investigations.

Figure 3.10

SEM images of the AmB nanoparticles from RESOLV with methanol as cosolvent: aqueous suspension right after the expansion (left), and with PVA as a stabilizer (right). (From Pathak et al., 2007a.).

The initially formed drug nanoparticles in RESOLV could be protected from agglomeration by using a stabilization agent in the aqueous suspension. For example, when an aqueous solution of PVA polymer (50 mg/mL) instead of neat water was used at the receiving end of the rapid expansion, the aqueous suspension of AmB nanoparticles from the same RESOLV process remained stable without precipitation. As shown in Figure 3.10, the AmB nanoparticles from the PVA-stabilized suspension are larger (average size of 64 nm diameter and size distribution standard deviation of 12 nm), but distributed homogeneously. PVA is an effective stabilizer commonly used in drug formulation. Various other polymeric and oligomeric stabilizers may also be used with RESOLV to protect the drug nanoparticles from agglomeration (Pathak et al., 2004, 2005, 2006, 2007a,b).

The amount of cosolvent in the supercritical solution for rapid expansion was found to have significant effects on the properties of the produced AmB nanoparticles. For example, an increase of DMSO from 1 to 2 vol% led to increases in the average particle size and crystallinity. As shown in Figure 3.11 , the AmB particles thus produced were faceted and square in shape, indicating a higher crystallinity. This was confirmed by the melting results from differential scanning calorimetry (DSC). The nanoparticles obtained with less cosolvent showed a 7.2^°C decrease in melting point from that of the bulk AmB, but those obtained with more cosolvent had only a 5.2^°C decrease (consistent with larger particles and higher crystallinity). The particle sizes as measured diagonally are on average 225 nm with a size distribution standard deviation of 49 nm. Similar changes in AmB particle sizes and crystallinity were observed with an increase in the amount of cosolvent methanol from 2 to 4 vol% in the supercritical solution for rapid expansion. The particles were also faceted (Figure 3.11) and larger (on average 143 nm diagonally with a distribution standard deviation of 40 nm). The higher crystallinity was also confirmed by DSC analysis. Mechanistically, the formation of more crystalline structures could simply be due to the significant increase in particle sizes, namely that only very small AmB particles could be made into largely amorphous under the RESOLV processing conditions. However, probably the higher cosolvent content also alters the rate of crystallization, thus producing more faceted particles (York, 1999; Jaarmo et al., 1997). The effect of solvent or solvent mixture on the formation of stable and unstable polymorphs is well recognized in the literature (Shekunov and York, 2000).

Figure 3.11

SEM images of the AmB nanoparticles from RESOLV with higher cosolvent concentrations, DMSO (left) and methanol (right). (From Pathak et al., 2007a.)

The variation in the morphology or crystallinity of the different AmB nanoparticles is an interesting feature in the nanosizing of drug particles via RESOLV. AmB is known to exhibit nonlinear concentration dependence in its pharmacodynamic characteristics, and thus higher doses are typically not recommended for toxicity considerations (Klepser et al., 1997; Bekersky et al., 2002; Andes, 2003). Because of the unusual biopharmaceutical property of AmB, its protein binding in plasma increases with increasing drug concentration and continuous infusion or less frequent administration of higher doses is generally preferred (Imhof et al., 2003; Lewis and Wiederhold, 2003).

Nanosizing Paclitaxel Particles

Paclitaxel, as one of the best antineoplastic drugs, is found to be effective or highly effective against a wide spectrum of cancers including ovarian cancer, breast cancer, lung cancer, colon cancer, head and neck cancer, etc. (Rowinsky et al., 1990; Lopes et al., 1993). Like many other anti-cancer drugs, it has difficulties in clinical administration due to its poor solubility in water and most pharmaceutical reagents. In its current clinical application, an adjuvant called Cremophor EL has to be employed, which has been found to be responsible for many serious side effects. Nanoparticle encapsulation of the drug has been considered as an alternative to realize a controlled and targeted delivery of the drug with better efficacy and less side effects.

 

The RESOLV processing technique was applied to produce paclitaxel nanoparticles suspended in aqueous buffer (Pathak et al., 2007b). The drug has some solubility in supercritical CO₂, with experimentally estimated solubility of 1.1×10^–6 to 6.3×10^–6 mole fraction at pressure and temperature ranges of 140–340 bar and 311–330 K, respectively. In a typical experiment, a solution of paclitaxel (2 mg) was added to the syringe pump, followed by CO₂ filling to a pressure of 310 bar. The solution was pushed through the heating unit to reach the desired supercritical temperature of 40^°C before the expansion nozzle. The rapid expansion was carried out at a pre-expansion pressure of 310 bar through a 50-µm orifice into ambient water solution. In the case of neat water at the receiving end of the rapid expansion, the nanoscale drug particles agglomerated over time to form larger aggregates and precipitated from the suspension. In the presence of a stabilization agent in the aqueous suspension, however, the initially formed paclitaxel nanoparticles in RESOLV could be protected from the agglomeration. For example, when an aqueous solution of PVP polymer (average molecular weight ∼ 40,000, 1 mg/mL) was used at the receiving end of the rapid expansion, no agglomeration and precipitation were observed following the RESOLV process. The suspended nanoparticles were used to prepare specimens for SEM characterization, which shows well-dispersed nanoparticles (Figure 3.12 a). According to a statistical analysis of the SEM images, the paclitaxel nanoparticles are of an average size of ∼ 40 nm in diameter.

Figure 3.12

(a) Paclitaxel nanoparticles (average size ∼ 40 nm) from RESOLV with expansion into a somewhat higher concentration PVP solution. (b) Paclitaxel nanoparticles (average size ∼ 500 nm) from RESOLV with expansion into a somewhat lower concentration PVP solution. (From Pathak et al., 2007b.)

The particle sizes and size distribution were found to be dependent on the concentration of the stabilization agent PVP, which allowed the variation/manipulation of these particle parameters. For example, the use of PVP at a somewhat lower concentration of 0.33 mg/mL under otherwise similar conditions resulted in larger paclitaxel nanoparticles (Figure 3.12b), with an average particle size of about 500 nm. In both cases the RESOLV processing yielded stable aqueous suspensions (solution-like) with homogeneously dispersed paclitaxel nanoparticles. There have been reports (Vandana and Teja, 1997; Nalesnik et al., 1998; Suleiman et al., 2005) that the use of a cosolvent such as ethanol or ethyl acetate with supercritical CO₂ increases the solubility of paclitaxel under comparable pressure and temperature conditions. These reported findings were confirmed in our experiments, though their application to the RESOLV processing of paclitaxel nanoparticles requires detailed investigation.

The nanosized drug particles from the RESOLV processing are in aqueous suspensions, readily compatible with established in vitro assays and in vivo tests (Pathak et al., 2007b; Bharadwaj and Prasad, 2002). For the in vitro cytotoxicity assay, actively growing MDA-MB-231 cells (10,000–20,000) were plated in 12-well cluster dishes in regular growth media and allowed to adhere overnight. For paclitaxel nanoparticles of 38 or 530 nm in average size, the particle suspension was sonicated in a water-bath, diluted directly into cell growth medium, and added to the cells at a desired paclitaxel concentration. The same procedures were used with a paclitaxel solution in ethanol at equimolar concentration. In assays for all three samples, cells were harvested at various time points from 4 h of plating to 72 h in culture. Cells treated without any additions were used as controls. At the time of harvest, the growth medium was replaced and rinsed with PBS. Cells were fixed and stained in 0.5% crystal violet in 50% ethanol at room temperature for 15 min, washed to remove excess dye, and allowed to dry. The dye was extracted with 50% ethanol for 15 min, and the absorbance was recorded at 540 nm.

The evaluation was based on the anti-proliferation effects on the human malignant breast cell line MDA-MB-231 (Pathak et al., 2007b; Mahadev et al., 2002). Treatment of cells with small (38 nm) and larger (530 nm) paclitaxel nanoparticles resulted in marked inhibition of cell proliferation. The estimated IC₅₀ values (based on the results after 48 h exposure) are 100 nM (paclitaxel equivalent concentration in the nanoparticle samples), comparable with what is reported in the literature for paclitaxel dissolved in DMSO against the same cancer cell line. The use of paclitaxel ethanol solution under the same experimental conditions yielded similar results (Figure 3.13 ). However, the control experiments on PVP only (0.33 and 1.0 mg/mL) showed no toxicity to the cells, consistent with the known conclusion that the polymer is generally nontoxic.

Figure 3.13

Cytotoxicity assays of the paclitaxel nanoparticles (smaller 38 nm and larger 530 nm average sizes) and the paclitaxel ethanol solution at equimolar concentration (0.5 µM) against MDA-MB-231 cells (normalized to the cell growth of the control after 72 h culture), with significant inhibition in comparison with the controls (p values <0.05). Error bars indicate standard deviation, and the data points are offset slightly on purpose for easier viewing. (From Pathak et al., 2007b.)

A known function of paclitaxel is to stabilize microtubules, thus interfering with the formation of mitotic spindle and inhibiting cell division and inducing apoptosis (Pathak et al., 2007b; Bharadwaj et al., 2005) as observed in the cytotoxicity assay above. For MDA-MB-231 cells treated with the paclitaxel nanoparticles of small or larger sizes (10 µM drug concentration) for 24 h, there was marked disruption of microtubule architecture and nuclear morphology according to confocal microscopy images (Figure 3.14 ) (Pathak et al., 2007b; Bharadwaj et al., 2005; Hartsel and Bolard, 1996). The disruption of microtubule architecture and the accumulation of alpha-tubulin around the nucleus were evident in drug-treated cells. Furthermore, nuclear examination revealed the presence of fragmented, diffusely stained, and multinucleated nuclei, which are consistent with apoptosis (Figure 3.14). Colocalization of microtubules with fragmented DNA was also detected.

Figure 3.14

The cellular organization of microtubule network in MDA-MB-231 cells with and without paclitaxel treatment (10 µm for all scale bars). Experimentally, the cells were plated in Nunc chamber slides and allowed to attach overnight. After the treatment with the drug (10 µM concentration for all) in the different forms for 6–24 h, or no treatment as control, the cells were fixed with paraformaldehyde and stained for microfilaments with anti-actin antibody followed by rhodamine-conjugated antibody, and then reacted with DAPI to stain nuclei (DNA). Samples were mounted using Prolong Antifade kit (Molecular Probes) for imaging on a Zeiss LSM 510 confocal microscope. (From Pathak et al., 2007b.) (See Color Plate 8)

The interference of paclitaxel nanoparticles with the cell cycle progression was probed in terms of flow cytometry. For MDA-MB-231 cells treated 24 h with the nanoparticles of a small (38 nm) or larger (530 nm) average size, there was a marked increase (about sixfold) in G₂-M phase and a concomitant decrease in G₀–G₁ populations (Figure 3.15 ). This is consistent with the blockade of mitosis. In a comparison with paclitaxel dissolved in ethanol, the nanoparticles appeared to be more effective, with a higher population of cells accumulated in the G₂-M phase (about 70 vs about 50% for the paclitaxel ethanol solution).

Figure 3.15

Flow cytometry analysis of MDA-MB-231 cells after 3 and 24 h incubation with different paclitaxel formulations (nanoparticles of sizes averaging 38 and 530 nm, and the paclitaxel ethanol solution) at equimolar concentrations and the control (without paclitaxel). A graphical interpretation of the percentage of cells in each cell cycle phases G1, S, and G2-M is shown for different paclitaxel formulations. (From Pathak et al., 2007b.)

Summary and Perspective

There has obviously been exciting progress over the past few years in the use of supercritical fluid processing techniques as alternative or unique solutions to many problems in drug formulation, especially with respect to the production of micron-sized and nanoscale drug particles and related systems. These techniques outperform conventional ones in many areas, including moderate operation conditions, low levels of residual solvents, products with targeted properties, etc. As demonstrated primarily at the laboratory scale, the SAS method is often preferred in the processing of drugs and polymers with limited solubilities in supercritical CO₂, while the RESS method is more advantageous as a single-step process and with much reduced organic solvent usage. The potential of the RESS method could be enhanced by a better drug solubility in supercritical CO₂ through the use of cosolvent. The RESOLV process has demonstrated great potential in producing exclusively nanoscale drug particles, especially those drugs with little or no solubility in water. The produced drug particles are not only protected from agglomeration but also offer the much needed flexibility in their further processing such as coupling with biological species and for specific formulation and delivery requirements.

Despite the progress, there are still enormous challenges in the supercritical fluid processing of drug particles and related materials and systems. For example, the manipulation or even control of particle morphology, size, dispersion, crystallinity, and the drug distribution within polymeric particles is still in need of major improvement and optimization. The modeling and mechanistic understanding of the particle formation in the supercritical fluid processes, and also the influence of processing parameters and conditions on the product properties are generally considered as being only at the early stage. Equally important is the issue on the use of the technology at the industrial scale, for which a likely solution is through collaborations between industry and academia. In any case, the future is bright for the application of supercritical fluid technology in drug formulation and for enhanced drug delivery and other related applications.