Journal of Food Engineering 95 (2009) 215226

Contents lists available at ScienceDirect

Journal of Food Engineering

journal homepage: www.elsevier.com/locate/jfoodeng

Review

Design and scale-up of pressurized uid extractors for food and bioproducts
C. Pronyk, G. Mazza *
Pacic Agri-Food Research Centre, Agriculture and Agri-Food Canada, 4200 Hwy 97, Summerland, BC, Canada V0H 1Z0

a r t i c l e

i n f o

a b s t r a c t
This article provides an in-depth review of the literature on the design, scale-up, and effects of scale on the extraction of food and bioproducts in pressurized uid extractors. The design of pressurized uid extraction systems such as supercritical CO2, pressurized solvent, and pressurized low polarity water (subcritical water) are similar. Knowledge of phase equilibria, mass transfer rate, and solubility data are important rst steps for the scale-up of extraction processes and equipment. The literature for the design, scale-up, and effects of scale on the extraction of bioproducts in pressurized uid extractors is examined with particular attention to the mass transfer principal and important parameters for extraction as they relate to the design and scale-up of xed bed pressurized uid extractors. Often when two scales of an extractor are examined, the scale-up has not been done uniformly, leaving the effects of the scale-up on extraction in doubt. There has been some success in design of a continuous pressurized uid extractor by utilizing a battery of vessels in series to operate on a quasi-continuous basis, and with the use of screw conveyors to produce a gas-tight plug of material, which allows the extraction to operate at the necessary elevated pressures and temperatures. Crown Copyright 2009 Published by Elsevier Ltd. All rights reserved.

Article history: Received 2 March 2009 Received in revised form 27 May 2009 Accepted 2 June 2009 Available online 6 June 2009 Keywords: Mass transfer Supercritical uids Subcritical water Pressurized solvent Fixed bed Extraction Leaching Counter-current Bioproducts

Contents 1. 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pressurized fluid extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Types of pressurized fluid extraction systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Basic design of pressurized fluid extractors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mass transfer and phase equilibria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Modeling of pressurized fluid extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. General model for the mass balance in a fixed bed extractor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. External mass transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Internal mass transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Effects of processing parameters on the extraction process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.1. Solubility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.2. Flow rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.3. Material properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.4. Length to diameter ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scale-up of pressurized fluid extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Continuous pressurized fluid extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Counter-current liquidliquid pressurized extraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Quasi-continuous solidliquid pressurized extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. Continuous counter-current solidliquid pressurized extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 216 216 216 218 218 218 219 219 219 220 220 220 220 221 222 222 223 224 225 225

3.

4. 5.

6.

* Corresponding author. Tel.: +1 250 494 6376; fax: +1 250 494 0755. E-mail address: Giuseppe.Mazza@agr.gc.ca (G. Mazza). 0260-8774/$ - see front matter Crown Copyright 2009 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jfoodeng.2009.06.002

216

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

Nomenclature ap Cf Cp Cs Cpo Cps Cso dE De DL dp specic surface area of a solid substrate particle (m1) concentration of solute in the solvent (kg/m3) concentration of dissolve solute in the solvent contained in the particle pores (kg/m3) concentration of the solute in the particle (kg/m3) initial concentration of dissolved solute within particle pores (kg/m3) concentration of solute on external surface of the particle (kg/m3) initial concentration of solute in the particle or adsorbed onto the particle surface (kg/m3) diameter of the extraction vessel (m) effective diffusion coefcient of the solid particle (m2/s) axial dispersion coefcient (m2/s) particle diameter (m) kf L r R t z u mass transfer coefcient in the uid phase (m/s) bed depth (m) radial position within the particle (m) particle radius (m) extraction time (s) axial position along the bed (m) interstitial velocity (m/s) porosity of the bed () porosity of the solid particle () residence time (s) U/e F/AE cross-sectional area of the extraction vessel (m2) supercial velocity (m/s) volumetric ow rate of the solvent (m3/s)

e ep s
u U AE U F

1. Introduction Given enough time and money almost any process or equipment can be made to work no matter how poor the design (Ritcey, 2004). The design of industrial-scale equipment is usually preceded by laboratory (bench) and pilot-scale systems. Sometimes the pilot-scale system is skipped, and work goes straight from the laboratory to industrial production. It is thought that time and money from the pilot-scale work can be applied to making the industrial plant work. This is most useful when there is sufcient data and the process is similar to one already in existence (Ritcey, 2004). However, with quality data and determination of scale-up factors, the design of pilot or industrial sized equipment is much more efcient. There is a large body of work in the literature dealing with the leaching (dissolution and removal of soluble components from a solid matrix)/extraction of bioproducts from solid organic materials with liquid solvents. Most of the work has been done in the laboratory with bench-scale equipment. There are general texts on the design of pressurized vessels or extraction systems (Bertucco and Vetter, 2001; King and Bott, 1993; Tzia and Liadakis, 2003), but there is little discussion on the aspects of scaling up a process or equipment. Such texts generally do not include any discussions on the performance of the equipment or on the appropriate scaling of the extraction process. In this paper the design, scale-up, and effects of scale on the extraction of bioproducts in pressurized uid extractors will be detailed by examining results available from the literature and patents. The mass transfer principles and important parameters for extraction will be discussed as they relate to the design and scale-up of xed bed pressurized uid extractors. 2. Pressurized uid extraction 2.1. Types of pressurized uid extraction systems Solidliquid extraction is a separation process involving the transfer of solutes from a solid matrix to a solvent. Solvents are chosen based on solubility characteristics of the desired solute. Ideally to achieve as pure a substance as possible, the solute should have high solubility in the solvent while other components in the solid matrix should not. Economics and safety are always a consideration and indeed, safer and less harmful solvents that are easy to remove, or recover, are gaining in popularity. There has been much interest in the eld of pressurized uids, as seen with the growing work with supercritical uid extraction (SFE) with CO2 (Daz-Reinoso et al., 2006; Herrero et al., 2006; Reverchon and De Marco,

2006; Shi et al., 2007) and the use of high pressure solvents including pressurized low polarity water (PLPW) (Kim and Mazza, 2009; Cacace and Mazza, 2007, 2006; Carabias-Martnez et al., 2005; Mazza and Cacace, 2005; Kaufmann and Christen, 2002; Smith, 2002). The versatility of pressurized solvents is excellent due to the physicochemical properties of the solvent, including density, diffusivity, viscosity, and dielectric constant, which can be controlled by varying the pressure and temperature of the extraction system. In this way the solvating power and selectivity of the solvent can be effectively controlled. Three methods of pressurized uid extraction have gained popularity in the research community as a means of extracting bioproducts. They are: supercritical uid extraction (SFE); pressurized solvent extraction (PSE, also known as pressurized liquid extraction (PLE), subcritical solvent extraction (SSE), or accelerated solvent extraction (ASE, Dionex trade name); and pressurized low polarity water (PLPW) extraction (also known as superheated water, subcritical water, pressurized hot water). All of these methods make use of vastly different pressurized solvents, each with a unique set of properties that inuence the extraction of the desired compounds. Supercritical uid extraction utilizes a solvent, usually CO2, although other substances such as water or ethylene can be used, near its thermodynamic critical point, where it possesses the density of a uid with the viscosity and diffusivity of a gas. By controlling the pressure of the supercritical uid, the solubility within the supercritical uid can be changed. This permits a high degree of selectivity, which allows for the fractionation of the extract by changing the pressure in the systems separators. Pressurized solvent extraction utilizes solvents such as hexane, methanol, or ethanol at temperatures above their boiling point, under high pressures, in order to increase the efciency of the extraction process with respect to extraction time, solvent consumption, and extraction yields. This is accomplished through improved solubility and mass transfer effects, in addition to a disruption of surface equilibria, such as reduced viscosity and increased solvent penetration into the sample matrix (Richter et al., 1996). PLPW extraction is a specic case of PSE, where water is utilized as the solvent. Water is a polar solvent, but if heated from 25 C to 200 C under pressure to maintain a liquid state, its dielectric constant decreases from 79 to 35, reaching values similar to solvents such as ethanol (24) or methanol (33) (Cacace and Mazza, 2006). 2.2. Basic design of pressurized uid extractors The solvents used for all three pressurized uid systems are vastly different, yet the basic design of the extraction equipment

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

217

Fig. 1. Schematic diagram of a typical supercritical uid extractor with a single separator.

is quite similar. At their most basic all three systems consist of a solvent supply, a pump for transporting the solvent, a heater for heating the solvent, a pressure vessel where the extraction occurs, a means to control the pressure in the system, and a collection vessel for the extract (Figs. 1 and 2). Individual units may have different congurations, instrumentation, valves, by-passes, gas purge systems, and safety features not shown here. The supercritical uid extractor differs slightly from the other pressurized extraction systems. Supercritical CO2, which is a nonpolar solvent, is sometimes modied with polar solvents such as ethanol to reduce the polarity of the system and allow for the extraction of a wider range of bioproducts. The co-solvent enters the system after the main pump (Fig. 1) via a secondary pump (not shown). Other differences for supercritical uid extractors include a heat exchanger before the pump, which is necessary to cool the CO2 to maintain it in a liquid state so that it may be pumped, and there may be a secondary pressure regulator after the separator (collection vessel) to control the pressure and allow for fractionation of the extract (Fig. 1). Without this pressure regulator, the CO2 would go from a supercritical state to a gaseous one, and all of the extract would precipitate from the solvent into the separator. By maintaining a small pressure in the separator, some compounds may remain in the solvent. By adding additional separators, it is possible to fractionate the extract into several different compounds by altering the pressure within each separator (Daz-Reinoso et al., 2006; Esquvel et al., 1999; Wang et al., 2004).

Fig. 2. Schematic diagram of a typical pressurized solvent or PLPW extractor.

Because PLPW extraction is a special case of PLE, the design of both systems is similar (Fig. 2). The PLPW extraction system adds a condenser after the extraction vessel to cool the water down so that it does not ash to steam. The heat capacity of water is very large, so considerable effort is necessary to remove the excess energy. In comparison, most organic solvents have a much lower heat capacity (approximately 3545% lower). The lower heat capacity, coupled with the fact that most PLE systems are small laboratory-scale units using low volumes of solvent, means that the solvent will cool before exiting the extraction system provided that the length of tubing after the extractor is sufciently long. Richter et al. (1996) found that when using extraction cells between 3.5 and 32 mL in volume, the temperature of exiting solvent was less than 35 C, even if the extractor was at 100 C, so long as a length of tubing of approximately 30 cm was used after the extractor. If bigger scale PLE units, using larger volumes of high temperature solvent were constructed, the use of a condenser would become necessary. For many laboratory-scale PLE and PLPW extractors, the solvent heater before the extraction vessel is omitted, and heating occurs by placing the extraction vessel within an oven. To allow the solvent to come to the proper operating temperature, long lengths of tubing are coiled within the oven before the extraction vessel to increase residence time and surface area. The benet of this system is that the extraction vessel is heated at the same time without the necessity of supplemental heaters for the vessel itself. Due to the difculties of designing a suitable device for charging a solid material into a high pressure and potentially high temperature system, these extractors are usually constructed as a batch system, with extraction occurring in a vessel through a xed bed of material. Such equipment is most suitable for high value and low volume products such as pharmaceutical and nutraceutical products. As such, there has not been great acceptance of this technology within the food processing industry where low value, high volume processing is the norm. Design considerations similar for all three extractors with the solvent, operating temperature, and pressure ranges being the system variables. With the choice of a suitable pump and extraction capacity, only the dimensioning of the extraction vessel and optimizing of the processing parameters are required. Work in the laboratory with small-scale equipment is invaluable for determining the appropriate factors for scale-up, usually with the development of process models. Knowledge of phase equilibria, mass transfer rate, and solubility data are important for scale-up of the extraction process and equipment.

218

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

C U, Cf (L) z=L u, Cf (z + z) z + z z Particle void fraction (p)

Cs Cp Cf

Cp r Cs

u, Cf (z)

z=0 U, Cf = 0

Bed void fraction ()

Solid particle

Fig. 3. Diagram of a xed bed extractor showing a nite volume element, bed and particle characteristics, and the concentration proles of solute within the particles and surrounding solvent (adapted from del Valle and de la Fuente, 2006).

3. Mass transfer and phase equilibria 3.1. Modeling of pressurized uid extraction Modeling of the extraction process is a benecial step in the effective scale-up from laboratory to pilot or industrial-scale systems (Bulley et al., 1984; del Valle and de la Fuente, 2006). del Valle and de la Fuente (2006) present a review of several kinetic and equilibrium models of SFE of oilseeds. Some of the models in literature are empirical in nature and are simplistic, but they describe the extraction process very well. However, these models do not provide insight into the underlying mass transfer phenomena that occur during the extraction process, with no correspondence to the materials being extracted, and therefore are unsuitable for scaling up purposes (del Valle and de la Fuente, 2006; Reverchon and De Marco, 2006). Kinetic (Anekpankul et al., 2007; Ho et al., 2008; Kubtov et al., 2002) and diffusion models (Anekpankul et al., 2007; Ho et al., 2008) have been applied to PLPW extraction, but no modeling seems to have been done for PLE. Some of the most interesting mass transfer models are those based on mass balance equations for thin sections of a packed bed (del Valle and de la Fuente, 2006). Such models are able to utilize data from laboratory-scale units to determine the mass transfer within a larger unit. Bulley et al. (1984) present a simplied model of mass balance in a single element of a bed for the extraction of oil from aked canola seed. The use of the xed bed model for SFE for the extraction of oil-containing seeds by researchers has been reviewed by del Valle and de la Fuente (2006). The model is also applicable for extraction of other substances from organic matter in packed beds such as the PLPW extraction of mannitol from olive leaves (Ghoreishi et al., 2008). However, PLPW extraction and PLE are still in their infancy and larger units are still rare, so modeling from a system, or bed perspective is not common. 3.2. General model for the mass balance in a xed bed extractor The mass transfer within a xed bed, inside a cylindrical extractor, may be described by a set of partial differential equations (Akgerman and Madras, 1994). The extraction vessel is divided into

nite difference volume elements of height Dz (Fig. 3). The resulting equations may be solved using numerical techniques. The assumptions of the model are: (i) pressure losses and temperature gradients are negligible within the bed, i.e. system is isothermal and solvent density remains constant along the bed; (ii) the solute is subjected to axial dispersion in the solvent due to a concentration gradient along the bed; (iii) physical properties of the solvent remain constant; (iv) bed porosity (e) remains constant, hence interstitial velocity is also constant); (v) there are no concentration gradients over the cross-section of the extractor. The uid phase mass balance for a nite volume element in a xed bed is thus:

@Cf @2Cf @ C f kf ap C f C ps DL 2 u @t @z @z e

If bed porosity is e, then the volume of particles is 1 e. Assuming spherical particles, the surface area of the particles in a unit volume of bed (specic surface area) is dened as:

ap

31 e R

Combining Eqs. (1) and (2) gives:

@Cf @2Cf @Cf 3 1 e kf C f C ps DL 2 u @t @z @z R e

The associated boundary conditions are:

DL

@2Cf uC f @ z2

at z 0; for all t

3a

@Cf 0 at z L; for all t @z C f 0 at t 0; for all z

3b 3c

The axial dispersion coefcient is a measure of back-mixing in the bed during uid ow. Axial dispersion is often neglected (DL = 0) and represents no back-mixing within the extraction vessel, or the case of plug ow (del Valle and de la Fuente, 2006). It is desirable to avoid axial dispersion of the solute along the bed in order to maintain the driving forces for extraction at their maximum level

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

219

(del Valle and de la Fuente, 2006). Axial dispersion may be avoided and hence neglected in the modeling by: maintaining a larger interstitial solvent velocity in the bed; by using sintered metal plates to cause P10 bar of pressure loss in the solvent at the inlet; to avoid gradients in solvent viscosity due to radial gradients in temperature by operating at near ambient temperatures and/or using smaller diameter extraction vessels; and by maintaining a small enough particle diameter to extractor diameter ratio (dp/dE < 0.1) to diminish the radial variations in bed porosity (Brunner, 1994). Fluid transport of the solute within or out of the particle is assumed to take place by diffusion through a network of pores (Fig. 3). The particle mass balance (of the uid within the pores) is thus:

been applied to the extraction of other thin particles like andrographolide, a diterpenoid lactone, from Andrographis paniculata leaves (Kumoro and Hasan, 2006). Kim and Mazza (2006) found that extraction of phenolic compounds from ax shives (expressed as a total phenolic compounds value) in PLPW was controlled by external mass transfer at low ow rates. Anekpankul et al. (2007) found that at low ow rates (between 1.6 and 2.4 mL/min) the extraction of the anthraquinone damnacanthal from roots of Morinda citrifolia was controlled by external mass transfer. At high ow rates (>2.4 mL/min), results suggested that internal mass transfer was the controlling factor. Some extraction processes will be controlled by a combination of factors depending on the extraction conditions. 3.4. Internal mass transfer The second stage occurs when the freely available solute present at the surface of the material is removed. The extraction is then controlled by the movement of the product within the material, or the internal mass transfer coefcient (De in Eq. (4)). The diffusion of the dissolved solute within the solid matrix is usually the rate limiting step in the extraction of most botanicals (Schwartzberg and Chao, 1982) and for most SFE (Kubtov et al., 2002). Machmudah et al. (2008) found that the extraction of oil from ground seeds (rosehip, loquat, and physic nuts) with SFE was largely controlled by the internal mass transfer of oil, especially at high pressures. Ho et al. (2008) found that internal diffusion was the controlling factor in the extraction of lignans from axseed meal with PLPW. Similar results were obtained for extraction of lignans from whole axseed (Cacace and Mazza, 2006). Kubtov et al. (2002) found that the extraction of essential oil from savory was controlled by internal mass transfer of the dissolved solute within the solid matrix during SFE, but was controlled by the external mass transfer of solute to solvent in PLPW extraction. This is partially explained by the fact that PLPW is more likely to extract organic material than SFE (Hawthorne et al., 2000). Kubtov et al. (2002) hypothesized that PLPW is more effective at altering sample matrices and displacing analytes from their original binding sites than supercritical CO2. The extraction process may be controlled by a combination of different processes and the exact mechanisms controlling the extraction may change depending on the extraction parameters (Anekpankul et al., 2007). 3.5. Effects of processing parameters on the extraction process Optimization of the extraction by altering the processing parameters is an important step in the scale-up procedure, often done with laboratory or pilot-scale units before nal scaling to production-scale systems. A great deal of work has been done on determining the operating parameters for SFE (Daz-Reinoso et al., 2006; Reverchon and De Marco, 2006), PLE (Kaufmann and Christen, 2002), and PLPW extraction (Cacace and Mazza, 2007; Kim and Mazza, 2009), for the optimization of the extraction of bioproducts from organic material. A balance must be achieved between maximum yield (ratio of amount extracted by total available), and maximum concentration. For high value products, a maximum yield may be desirable, but as value of the product decreases, the economics of the extraction would dictate an increase in the concentration of the extract. In many cases the extraction process in commercial operations is considered complete when 90% of the solute has been extracted (del Valle et al., 2005). Reduction of the extract volume by removing the solvent can be an expensive process, and the added cost is not compensated from the additional product recovered. The ideal operating conditions for an extraction system occur when the system is operating in

ep

  @ C p @ C s De ep @ @Cp r2 2 @t @t r @r @r

The associated boundary conditions are:

@Cp kf C f C s ep De @r

at r R; for all z and t

4a

@Cp @Cs 0; 0 at r 0; for all z and t @r @r C p C po ; C s C so at t 0; for all z and r

4b 4c

The concentrations, Cf, Cp, and Cs in Eqs. (3) and (4), may be utilized on a wt/wt basis rather than a wt/vol basis by incorporating the solvent and particle densities into Eqs. (3) and (4) (Bulley et al., 1984; Lee et al., 1986). Density of the solvent will vary with the processing parameters of temperature and pressure. Both are signicant for supercritical uids, however, for PLE and PLPW extraction, changes in density with pressure tend to be negligible due to the incompressible nature of liquids (no change in density with pressure). Experiments have shown that operating pressure is not a signicant extraction parameter for PLE (Choi et al., 2003; Kawamura et al., 1999; Ong and Len, 2003) and PLPW extraction (Gogus et al., 2005; Rovio et al., 1999; Soto Ayala and Luque de Castro, 2001) of bioproducts. Yet, there may be some benet of operating at higher pressures whereby the increased pressure aids in penetration of the solvent into the cells and pores of organic materials. This is weighed against the potential for collapsing of the bed, and a reduction in porosity leading to lower interstitial velocities. However, for PLPW extraction an increase in pressure from 10 to 600 MPa at 25 C would result in an increase in the dielectric constant from 79 to 93 (Haar et al., 1984). This would yield a slight increase in the polarity of the PLPW, which would negatively affect the extraction of non-polar compounds. In both systems there is a critical constraint where it is necessary to maintain the pressure at a sufciently high level in order to maintain the solvent in a liquid state, but there is little need to signicantly increase pressure above that level. 3.3. External mass transfer The extraction process takes place in two distinct stages governed by two different modes of mass transport. The rst stage occurs early in the process, when the oil or solute is freely present on, or diffused to the surface of the material, and extraction is controlled by the external mass transfer (coefcient kf in Eq. (3)). The product is freely available to be removed so long as the maximum solubility of the solute within the solvent has not been reached. Extraction of some substances and materials occur wholly within the rst stage of extraction. In this case the extraction would be controlled only by Eq. (3), and the transport within the particles (Eq. (4)), would be neglected. This is most often the case for SFE of oil bearing materials, when they are processed into thin akes (Bulley et al., 1984; Lee et al., 1986), although it has also

220

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

the rst stage of extraction, which is limited by the solubility of the solute dissolved in the solvent. 3.5.1. Solubility Solubility of a solute in a particular solvent is fundamental to the design of an extraction process. It controls the ideal operating conditions and sets the minimum amount of solvent necessary to complete an extraction (Shi and Zhou, 2007). When the solubility of the solute in the solvent has been reached, under suitably low ow rates, the extraction will be operating in the rst stage whereby the extraction is controlled by the external mass transfer. Solubility of a product can be determined from the slope of the linear (constant extraction rate, external mass transfer limiting stage) portion of the extraction curve (Bozan and Temelli, 2003; Gl and Temelli, 2005; zkal et al., 2005). Solubility is stndag dependant on solvent density. With supercritical uids this means that the solvating power of the uid may be altered with changes in the systems pressure (Fig. 4). Liquids are largely incompressible in the subcritical region, so pressure tends not to have a great inuence on solubility and extraction in PLE and PLPW systems (Smith, 2002). In this case, temperature has a more important effect on product solubility than pressure. As ow rate increases, there is the possibility in extraction systems that the free solute on the surface of the material will rapidly be removed and the mass transfer of the dissolved solute within the product will become the rate limiting step. The key is not to keep increasing the ow rate when there is no benet. Doing so will only decrease the concentration of the extract and use more solvent than is necessary to complete the extraction. 3.5.2. Flow rate Flow rate has a direct bearing on the mode of mass transfer for the system, and therefore the design of a xed bed extractor. When the ow rate is sufciently large there forms a concentration gradient between the solids surface and the solvent and the extraction falls into the second extraction stage whereby diffusion within the material becomes rate controlling. If the ow rate is lowered sufciently, the removal of solute from the materials surface decreases, and diffusion of the solute within the solid matrix is adequate to replace the solute removed from the material. The uid phase mass balance (Eq. (3)) contains a term which incorporates the ow rate of the extractor. As stated above, when an extraction is controlled only by external mass transfer, then the extraction is governed solely by Eq. (3), and the kinetics of the sys35 30 25 20 15 10 5 0

tem prevails. If an extraction is limited by the external mass transfer from the solid to the solvent, then an increase in ow rate would result in an increase of the extraction rate of the solute. Conversely, if an extraction is limited by the internal mass transfer of the system, and the partitioning thermodynamics limit the extraction rate, then it would be expected that an increase in the ow rate would have little effect on the extraction rate of the solute (Eq. (4) does not contain a ow rate term). Ho et al. (2008) found that internal mass transfer was the controlling factor in the extraction of lignans from axseed meal with PLPW, as seen by the negligible effects of ow rate on the process. Shalmashi et al. (2008) found that increasing the ow rate during PLPW extraction increased the rate of caffeine removal from black tea leaves, which indicated that extraction was mostly controlled by the transfer from the solid to the solvent (rst stage). However, the most efcient ow rate was 2 g/min, which achieved the maximum yield of caffeine in the same amount of time, but with half the volume of solvent when compared to extraction with a ow rate of 4 g/ min. Knowledge of the mass transfer characteristics in the extraction vessel can aid in the optimization of the ow rate to minimize extraction time, while maximizing yield. 3.5.3. Material properties Physical properties of the material being extracted are important, as they play a role in the phase equilibria and mass transfer within the extraction vessel. Size reduction can improve extraction because of mass transfer in the form of diffusion, which is inversely proportional to the square of the characteristic dimension (Crank, 1975; Cussler, 1984; Osburn and Katz, 1944). In addition, with larger particles less cell walls will be undamaged, reducing solvent penetration and product extraction. Reducing the particle size will increase the surface area per unit mass between particle and solvent, but very small particles will produce smaller volume voids in the sample matrix, providing resistance to uid ow within the extractor. There are limits to suitable size reductions before porosity of the bed (e, Eq. (3)) decreases to such an extent that ow becomes difcult, and the possibility of channelling and nes in the extract decrease quality. If particle size is reduced sufciently, then internal diffusion ceases to be the controlling factor and external mass transfer from the particle to the solvent, becomes the controlling factor (Aguilera, 2003). 3.5.4. Length to diameter ratio Bed length to diameter ratio is an important consideration in the design of extraction vessels. Vessel cost is partially inuenced by volume, but mainly by the diameter (Laurent et al., 2001). In tall extractors back-mixing may occur, and large diameter vessels may result in heterogeneous extraction, resulting in radial effects within the extractor (Brunner, 1994; del Valle et al., 2004). For particle sizes between 0.4 and 0.8 mm the recommended length to diameter ratio is 6:1, while larger ratios of 9:1 have been used for the decaffeination of coffee (7 mm diameter particles) and smaller ratios of 3:1 for materials which may swell (Laurent et al., 2001). Eggers (1996) recommends a length to diameter ratio of 46 for the SFE of oilseeds with CO2. Many length to diameter ratios for experimental SFE units vary between 2 and 29 (del Valle et al., 2005; Meireles, 2003) and for PLPW extraction and PLE they generally fall in a similar range of 225 (Cacace and Mazza, 2007; Ho et al., 2008; Richter et al., 1996). But, with respect to cost, this value should vary between 5 and 7 (Meireles, 2003). With knowledge of the solvent to solid ratio and mass transfer rate, a system may be designed to produce a certain yield, or throughput per day. Mass transfer will dictate efcient ow rates for the extractor. Comparisons can be made between industrial units and experimental units if the solvent to solid ratio are the same, and scale-up is accomplished by increasing the vessel diameter and length to maintain

Solubility (mg/g)

40C 50C 60C

10

20

30

40

50

60

70

Pressure (MPa)
Fig. 4. Effect of pressure on the solubility of hazelnut oil in supercritical CO2 (adapted from zkal et al., 2005).

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

221

a constant ratio, and solvent is properly dispersed into the system (Meireles, 2003). However, this neglects to take into consideration the situations that may be encountered in a larger unit, such as, material agglomeration, increased pressure drop, uid channelling, and the increased difculty maintaining plug ow with increased vessel diameter.

4. Scale-up of pressurized uid extraction An effective model of the extraction system is crucial for determining the basic mass transfer data necessary for scale-up procedures. There have been many studies on the kinetics of extraction for high pressure extraction systems, but there has not been substantial analysis of the effect of process scale on the extraction rates (del Valle et al., 2004). When designing equipment certain inputs may be xed, such as temperature, pressure, ow rate, or pH, in order achieve a desired extraction efciency or rate. These are often determined in a laboratory using nothing more than glassware and chemicals, or other laboratory-scale equipment to produce product on the gram scale or smaller. The problem then is how to effectively scale-up to larger production runs, or produce equipment to do the same. Appropriate scale-up factors must be determined for the process being examined. It is not sufcient to scale everything up at the same rate, as the process is often constrained by certain key parameters, and all others do not contribute to the desired outcome. For instance, one of the key parameters in the operation of an extractor is the supercial velocity of the solvent in the extractor itself. This is the velocity across the cross-section of the extractor dened as the volumetric ow rate of the solvent divided by the cross-sectional area of the extractor. If the diameter is increased by a factor of ten, the volumetric ow rate would have to be increase by 100 to maintain the same supercial velocity (area is a squared term). At the same time, to maintain the same bulk density and porosity in the extractor, the mass of material would have to increase by 1000 (volume is a cubed term) if the length is also increased by a factor of 10 to maintain the same length to diameter ratio. This process is illustrated by Lagadec et al. (2000) who set out to scale-up a PLPW system to remove contaminates from soil. The capacity of a laboratory unit was scaled-up by a factor of 1000 to increase the amount of soil processed from 8 g to 8 kg. To maintain an equivalent bulk density, and hence bed porosity (e), the volume was increased by a factor of 1000. This was accomplished by scaling up the extraction vessel diameter and length by a factor of 10 (9.4 mm i.d. 100 mm long to 102 mm i.d. 1000 mm long). For removal of polycyclic hydrocarbons (PAHs) from contaminated soil the same solvent ow-to-solid ratio was used for the scale-up procedure. The ow was set at 0.5 mL/min and 0.5 L/min for the laboratory- and pilot-scale extractor respectively. At 275 C and the specied ow conditions, all PAHs were removed within 6070 min. However, when the ow rate was increased to 600 mL/min in the pilot-scale extractor, the removal of PAHs was completed in 3540 min, which indicates that the extraction was controlled by the external mass transfer of the solute to the solvent. When external mass transfer is the controlling factor, as it is in the rst stage of extraction, it is common practice during scale-up to keep the solvent to solid ratio constant (del Valle et al., 2005). If the maximum solubility of the solute in the solvent is maintained, then the extraction may be completed using the least amount of solvent possible. Whereas, when internal mass transfer is the controlling factor, as in the second stage, the practice is to keep the solvent ow-to-solid ratio constant in order to extend the contact time between the solvent and material (del Valle et al., 2005). Lagadec et al. (2000) kept the solvent ow-to-solid constant even though the extraction was controlled by the external

mass transfer from the solid to the solvent. In removal of soil contaminates only the efcacy of the extraction is important. As such, Lagadec et al. (2000) did not examine the extraction kinetics and no information on the extraction rates between the different scale extractors was given. Berna et al. (2000) modeled the extraction of essential oil from orange peels with supercritical CO2 using Lacks extended plug ow model as adapted by Sovov (1994). Experiments were conducted at two different scales (feedstock scaled-up by a factor of 20), while maintaining the same solvent ow-to-solid ratio and solvent-to-solid ratio. By studying the parameters from Sovovs (1994) model, Berna et al. (2000) were able to incorporate the effect of bed height into the model. Their results showed that there was similar behaviour in all experiments and that as long as there was homogeneous ow within the extractor, the height of the bed had little effect on the extraction at the same scale of operation. Berna et al. (2000) found that the addition of diatomaceous earth at the inlet of the extraction vessel promoted homogeneous ow, possibly by dispersing the solvent evenly across the diameter of the vessel and ensuring plug ow. In addition to issues of ow heterogeneity, there is a greater possibility for the agglomeration of particles with high oil contents at larger scales, which would increase the resistance to internal mass transfer of particles due to a larger effective diameter (Berna et al., 2000). Perrut et al. (1997) developed a model of supercritical CO2 extraction of sunower seed oil based on the results of a laboratory and pilot-scale system with an extraction vessel that was scaledup by a factor of 10 (from 0.15 103 to 1.5 103 m3). However, none of the other extractor dimensions were scaled proportionally, resulting in an altering of the length to diameter ratio from 4.4:1 in the laboratory-scale unit to 3.5:1 in the pilot-scale unit. Literature suggests that comparisons can be made between units if the scaleup is accomplished by increasing the vessel length and diameter to maintain a constant ratio (Meireles, 2003). There was no discussion by Perrut et al. (1997) about the effects of scale-up on the extraction kinetics. Simple scale-up by a set factor has been questioned by del Valle et al. (2004) who studied the effects of process scale on the oil extraction kinetics of supercritical CO2 extraction of rosehip seed. The rst series of experiments were carried out on a laboratoryscale system with a vessel volume of 5.0 105 m3 using 26 g of sample. The extraction conditions for the laboratory-scale system were 40 C and 30 MPa, or 50 C and 40 MPa at supercial CO2 velocities of 0.231.40 mm/s. Results were used to develop an one-dimensional, unsteady state model with axial dispersion of solute in the supercritical phase for a packed bed. It was a twostage model with extraction rate initially controlled by oil solubility and mass transfer from the material to the solvent, and by internal mass transfer within the particles during the latter stages of extraction. The model was used to simulate the extraction of a pilot-scale unit, which was scaled-up by a factor of 30 for the vessel volume, with recycling of the solvent. However, equivalent length to diameter ratio of the two vessels was not maintained as the inner diameter was scaled by a factor of 3 while the length was only scaled by a factor of 1.5. The extraction conditions for the pilotscale system were 40 C and 30 MPa, or 50 C and 40 MPa at a supercial CO2 velocity of 0.570.58 mm/s. Results showed that the extraction in the pilot-scale system was slower than those predicted from the model which utilized parameters from the kinetic data from the laboratory-scale unit. del Valle et al. (2004) attributed the discrepancy to ow heterogeneity in the extraction vessel, increased dispersion of solute between the extraction and separation vessels, entrainment of oil droplets in recycled gaseous stream, which would decrease the concentration gradient between the solute and solvent in the extraction vessel, or a combination of the three. Flow heterogeneity may occur due to radial variations of

222

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

bed porosity and solvent viscosity, which could result from packing the bed with relatively large particles (dp/ue < 10, where dp is the particle diameter and ue is the internal diameter of the extraction vessel), or attempting heat transfer by conduction through the wall of the extraction vessel, respectively (Brunner, 1994). These would result in higher interstitial velocities near the wall, which would result in slower extraction in the centre of the extraction vessel (del Valle et al., 2004). This was manifested as a reduction in the period of external mass transfer in the pilot-scale system caused by earlier depletion of solute near the wall of the extraction vessel, causing an earlier drop in solute concentration in the exiting solvent. del Valle et al. (2004) concluded that radial variations in porosity could be more important with small diameter vessels on the laboratory-scale, whereas variations in viscosity could be more important when using large diameter vessels and high temperatures at the pilot-scale. Hence the caution in scaling up of extraction vessels by a constant factor because of changes in mass transfer phenomena related to extraction conditions that do not always depend on the mass transfer coefcient. Eggers and Sievers (1989) hypothesized that scale-up should be accomplished by altering the vessel height while maintaining the same supercial velocity in both extractors to ensure equivalent mass transfer properties. Results for the extraction of evening primrose seed oil (EPO) in a pilot-scale (200 L extractor volume) and production-scale unit (2000 L extractor volume) with supercritical CO2 differed. Extraction in the production-scale unit took longer but required a lower solvent to solid ration. Loading of EPO in the solvent increased from 0.63% in the pilot-scale, to 0.83% in the production-scale unit. The theoretical phase equilibrium for EPO in supercritical CO2 under the extraction conditions of the experiment was 0.9% (Eggers and Sievers, 1989). To account for this discrepancy, Eggers and Sievers (1989) proposed the use of higher velocities in larger scale extraction vessels compared to smaller scale vessels in order to keep the residence time (calculated as a function of bed depth and supercial velocity, s = L/U) equal in the vessels at the two scales. Ho et al. (2008) found that the yield of lignans from axseed meal increased with increasing bed depth, and was independent of the solvent to solid ratio. This was attributed to the residence time. It was suggested that longer residence times allowed sufcient time for the solvent to penetrate through the solid pores and allowed for increased yields. King et al. (1997) extracted EPO on a laboratory-scale unit (72 mL extraction vessel) and compared their results with those

of Eggers and Sievers (1989). The results from the laboratoryand production-scale units were in agreement. Only Eggers and Sievers (1989) results for the pilot-scale unit were lower. No explanation was given for these results. However, the bulk density was higher and the supercial velocity was lower in the laboratoryscale unit compared to the levels used by Eggers and Sievers (1989) for the pilot- and production-scale units. This would result in lower bed porosities in the laboratory-scale unit that would lead to a lower interstitial velocity that would be offset to a degree by the decrease in supercial velocity. It is hard to make concrete conclusions about the effect of scale when parameters that inuence the extraction are not held constant, or scaled in the appropriate fashion.

5. Continuous pressurized uid extraction Most applications of pressurized uid extraction in the literature operate as a batch process, whereby a material is extracted by immersion or percolation of solvent through a xed bed. This method of extraction is largely undesirable in industry due to necessary interruptions for loading and unloading of the extraction vessel, and the large amounts of solvent required to complete an extraction (Eggers and Jaeger, 2003). To be truly economical, and therefore desirable to industry, most processing must operate on a continual basis so that all equipment in a facility is being utilized to the fullest. There is great difculty to convert the technologies of pressurized solvent extraction to continuous operation. Design of a continuous system would be best done using information gained from a corresponding continuous laboratory-scale unit. However, it is more convenient to keep the product stationary in a xed bed. Design of a continuous extractor is still possible from information gained in experimentation with xed beds by studying the equilibrium distribution of oil between the material and solvent, and determination of the mass transfer rate of solute from the material to the solvent (Bulley et al., 1984). One area where continuous pressurized solvent extraction has achieved success is with counter-current liquidliquid extraction. 5.1. Counter-current liquidliquid pressurized extraction Counter-current contact is achieved when solvent rst comes into contact with the material that has been the most extracted, and exits the extractor in contact with the least extracted material. In this way the differences in concentration between the solute and solvent are greatest, providing the greatest driving force in the extraction process. Counter-current operation of an extractor is thus able to reduce the amount of solvent used, increases throughput, enables higher extract concentrations in the solvent, and lower residual concentrations in the material (Brunner, 2005). True counter-current contact is easily achieved when a liquidliquid extraction is conducted, or when a slurry of the material is created that may be pumped (Fig. 5). Systems may be constructed to include multiple separators, solvent regeneration, or reuxing of the extract (Singh and Rizvi, 1994; Brunner, 2005). Continuous counter-current pressurized solvent liquidliquid extractors have been studied by several researchers. Brunner (1998) reviewed and discussed the development process for supercritical countercurrent liquidliquid extractors. Singh and Rizvi (1994) have completed a design and economic analysis of a system for processing milk fat with supercritical carbon dioxide. Ibez et al. (2002) constructed a system for the concentration of sterols and tocopherols from olive oil using supercritical carbon dioxide. Ooi et al. (1996) used a continuous counter-current supercritical CO2 extractor to rene palm oil. The system was successful in removing the free fatty acids while retaining the carotenes in the oil which are

Fig. 5. Schematic diagram of a continuous counter-current system used for liquid liquid extraction with supercritical carbon dioxide without solvent regeneration and only one separator (adapted from Ooi et al., 1996).

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

223

Fig. 6. Schematic diagram of the setup and ow through a battery of extraction vessels. Each vessel can be bypassed for loading and unloading as indicated by the discharge bypass shown only for the middle extractor (adapted from Eggers and Jaeger, 2003).

Fig. 7. Continuous counter-current supercritical CO2 extraction plant (adapted from Eggers and Sievers, 1989).

normally destroyed in traditional physical or chemical rening processes. Transfer of material into and out of counter-current liquidliquid extractors is not as difcult as for counter-current solidliquid extractors because the product can be pumped into the extractor in liquid form with the use of a piston or diaphragm pump. The biggest problem with continuous pressurized countercurrent solidliquid extraction systems is how to feed solid material from atmospheric pressure into a vessel at high pressure, then feed the material back to the atmosphere after extraction. There is a lack of reliable sluice systems for high pressure vessels for continuous charging and discharging of solids into the extraction vessel. To gain the benets of a continuous counter-current extraction without a suitable sluice system it is possible to conduct a quasicontinuous extraction system with multiple stages. 5.2. Quasi-continuous solidliquid pressurized extraction To create a quasi-continuous extraction system, several batch extractors are connected and operated in series to form a battery (Fig. 6). Fresh solvent is loaded into the vessel which contains

the most exhausted solid (Vessel 2). The solvent and extract from this vessel then passes through each vessel in succession, until it is discharged from the last one, which contains the most recently loaded material (Vessel 1). One vessel is bypassed to unload spent material and recharge with fresh material (Vessel 3). With this setup extraction occurs on a continuous basis, although loading and unloading is discontinuous to allow for depressurization/pressurization and opening of the pressure vessels. On the laboratory-scale the ASE extractor operates on a semi-continuous basis by setting up a series of vessels and batch extracting them one at a time (Richter et al., 1996). Most industrial-scale units are proprietary, and therefore detailed information of their design and operation are not available in the literature. However, several quasi-continuous pressurized extraction system designs have been proposed and discussed. Leyers et al. (1991) presented a detailed design and study of a system that used supercritical CO2 to decaffeinate coffee. The system consisted of a battery of four vessels operating in a continuous mode with three vessels operating at one time with one being loaded or unloaded with fresh coffee beans. The system was operated at 1435 MPa at temperatures of 70130 C with a

224

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

residence time of 612 h. Lack and Seidlitz (1993) proposed a similar design of a quasi-continuous system to decaffeinate coffee using a battery of four extraction vessels for while working for SchoellerBleckmann before moving on to form NATex. 5.3. Continuous counter-current solidliquid pressurized extraction The design of a continuous counter-current extraction system without a suitable sluice system is possible by using screw conveyors or extruders to pack the solid material into a gas-tight plug, preventing pressure losses along the moving bed (Eggers, 1988, 1994). Eggers et al. (1985) proposed a design for a continuous counter-current extraction system using supercritical CO2 for the extraction of oil bearing materials, which was later patented (United States Patent 4,675,133; Eggers and Schade, 1987). The system consists of a high-pressure extraction vessel placed between two screw presses (Fig. 7). The charging press serves to rst partially de-oil the material, then to build up the pressure to form a gastight plug. The plug of material that is formed within the press can reach temperatures of 6090 C due to the work of compression and wall friction (Eggers and Sievers, 1989). At the same time, the gas-tightness of the plug has been demonstrated for pressure differences up to 40 MPa for periods as long as 5 min, which is larger than the residence time within the screw press (Eggers and Sievers, 1989). The pressure in the extraction vessel inuences the pressure prole of the screw press. When the pressure in the extraction vessel increases, there is a corresponding pressure increase in the screw press which helps ensure the gas-tightness of the plug (Eggers, 1996). The discharging press acts in the same manner to form a gas-tight plug for the extraction vessel while allowing for the residue to be continuously removed. Extraction of the material occurs in the high-pressure vessel between the two screw presses. Supercritical CO2 enters the bottom of the extraction vessel and leaves at the top, moving opposite the ow of the material and thus extracting under counter-current conditions. Inside the extraction vessel there is a slowly revolving mechanical stirring mechanism, which aids in the transport of the material through the extraction vessel and prevents clogging. A pilot plant was produced that could process 225 kg/h of material (Eggers and Sievers, 1989; Eggers, 1996). Wingerson and Lehrburger (2004) designed and tested a continuous counter-current screw extractor to operate at elevated pres-

sures and temperatures for which a patent has been led (United States Patent Application# US 2006/0283995 A1; Wingerson (2006)). The goal was to design and validate a system to produce a puried cellulose product on a continuous basis by utilizing a process which minimizes the use of enzymes by fractionating biomass with aqueous media (water and alkali), at elevated temperatures up to 235 C. The extractor utilises a single threaded shaft that has a plurality of reaction zone segments along the length of the shaft that are separated from each other by dynamic plug segments. The shaft has a different thread pitch in the reaction zone segments than in the dynamic plug segments, which help to create the plugs. Wingerson and Lehrburger (2004) successfully achieved the formation of two plugs within the screw extractor using ground corn stover capable of retaining pressures of 1.4 MPa to over 6.9 MPa. They also achieved some success producing counter-current extraction and discharging the liquid and solid stream separately. However, they were as yet unable to operate on a completely continuous basis at the desired temperatures and pressures, or to produce a puried cellulose product. There has been some success in designing a sluice system for charging and discharging solids from a continuous counter-current pressurized uid extraction system. A group from Denmark has created a system that utilizes a so called particle pump, which is capable of transferring solid particles between two areas of different pressure (Christensen and Christensen, 2004; Thomsen et al., 2006, 2008). The particle pump, for which a patent application has been submitted (United States Patent Application# US 2004/0184900 A1; Christensen and Christensen, 2004), utilizes two sluice chambers, one of which is charged with a piston screw while the other is being discharged. Pressure tightness is achieved by the use of pistons or pressure locks in the pump. The system is capable of processing low density biomass such as straw on a continuous basis for the extraction of hemicellulose with PLPW for the purpose of ethanol production. The pilot plant system, for which a patent has been granted (WIPO Patent# WO/2007/009463; Christensen and Christensen, 2007), has a potential production capacity of 1000 kg/h although it has only been tested in the 50150 kg/h range (Thomsen et al., 2006, 2008). The system has been designed to operate as a three stage extraction/reaction process (Fig. 8). In the rst stage the biomass is pre-soaked in a 10 m long screw conveyer and soaking tank (1) at temperatures up to 100 C for residence times up to 30 min. Water for the pre-soaking is obtained

Fig. 8. Continuous counter-current PLPW extraction plant for the fractionation of biomass: (1) soaking tank; (2) belt conveyer; (3) particle pump; (4) screw conveyer extractor; (5) particle pump; (6) screw conveyer extractor/reactor; (7) exhaust particle pump (adapted from Thomsen et al., 2006).

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226

225

from the water outlet of the extraction system. During this stage all of the air is driven from the biomass as it is saturated with the process water, in addition there is some transfer of thermal energy which increases the temperature of the biomass. After the pre-soak the biomass is transferred to a hopper on a belt conveyer (2) where it is fed into the second stage by means of the particle pump (3) mentioned previously. The second stage occurs in the extractor (4), which is a screw conveyer operated at elevated pressures and temperature up to 200 C (Thomsen et al., 2006). The biomass moves upwards through the conveyer while the PLPW enters at the top of the conveyer and moves downward, counter-currently by gravity. The purpose of the second stage is the removal of most of the hemicellulose sugars (xylose and arabinose), which may be utilized for ethanol production by C5-sugar fermenting microorganisms (Thomsen et al., 2008). Biomass exits the second stage into the third stage through another particle pump (5). In the third stage, which is a screw conveyer that acts as an extractor/reactor (6), the remaining biomass is treated at 195 C at a residence time of 3 min with the addition of steam and water in the upper portion of the reactor, to improve the enzymatic convertibility of the remaining cellulose fraction for ethanol production. The remaining biomass exits the system through a nal particle pump. Water for the system enters at the end of the screw conveyer in the third stage and leaches through the system counter-currently to the ow of biomass. Hemicellulose recovery in the system ranged from 33% to 83% depending on the amount of water added to the extractors (Thomsen et al., 2008). Yields of cellulose and hemicellulose could not be maximized concurrently as highly digestible cellulose was only obtained under severe conditions, which resulted in larger amounts of hemicellulose degradation and formation of inhibitory fermentation compounds. However, the system has shown the potential under such conditions to still have an estimated ethanol production capability of 184192 kg ethanol/ton straw compared to the theoretical maximum if all cellulose and hemicellulose in the straw was hydrolysed to monomeric sugars of 318 kg ethanol/ton straw.

been used to achieve true continuous ow. Most scale-up has occurred with the relatively older pressurized uid extraction technology of SFE. The younger technologies of PLPW extraction and PLE are just now gaining popularity in the research community, and modeling and scale-up is sure to be forthcoming. References
Aguilera, J.M., 2003. Solidliquid extraction. In: Tzia, C., Liadakis, G. (Eds.), Extraction Optimization in Food Engineering. Marcel Dekker, Inc., New York, NY, pp. 3556. Akgerman, A., Madras, G., 1994. Fundamentals of solid extraction by supercritical uids. In: Kiran, E., Sengers, J.M.H.L. (Eds.), Supercritical Fluids: Fundamentals for Application. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp. 669695. Anekpankul, T., Goto, M., Sasaki, M., Pavasant, P., Shotipruk, A., 2007. Extraction of anti-cancer damnacanthal from roots of Morinda citrifolia by subcritical water. Separation and Purication Technology 55, 343349. Berna, A., Trrega, A., Blasco, M., Subirats, S., 2000. Supercritical CO2 extraction of essential oil from orange peel; effect of the height of the bed. The Journal of Supercritical Fluids 18, 227237. Bertucco, A., Vetter, G., 2001. High Pressure Process Technology: Fundamentals and Applications. Elsevier Science BV, Amsterdam, The Netherlands. Bozan, B., Temelli, F., 2003. Extraction of poppy seed oil using supercritical CO2. Journal of Food Science 68, 422426. Brunner, G., 1994. Gas Extraction: An Introduction to Fundamentals of Supercritical Fluids and Application to Separation Processes. Springer, New York, NY. Brunner, G., 1998. Industrial process development countercurrent multistage gas extraction (SFE) process. The Journal of Supercritical Fluids 13, 283301. Brunner, G., 2005. Supercritical uids: technology and application to food processing. Journal of Food Engineering 67, 2133. Bulley, N.R., Fattori, M., Meisen, A., Moyls, L., 1984. Supercritical uid extraction of vegetable oil seeds. Journal of the American Oil Chemists Society 61, 1362 1365. Cacace, J.E., Mazza, G., 2006. Pressurized low polarity water extraction of lignans from whole axseed. Journal of Food Engineering 77, 10871095. Cacace, J.E., Mazza, G., 2007. Pressurized low polarity water extraction of biologically active compounds from plant products. In: Shi, J. (Ed.), Functional Food Ingredients and Nutraceuticals: Processing Technologies. Taylor & Francis Group, Boca Raton, FL, pp. 135155. Carabias-Martnez, R., Rodrguez-Gonzalo, E., Revilla-Ruiz, P., Hernndez-Mndez, J., 2005. Pressurized liquid extraction in the analysis of food and biological samples. Journal of Chromatography A 1089, 117. Choi, M.P.K., Chan, K.K.C., Leung, H.W., Huie, C.W., 2003. Pressurized liquid extraction of active ingredients (ginsenosides) from medicinal plants using non-ionic surfactant solutions. Journal of Chromatography A 983, 153162. Crank, J., 1975. The Mathematics of Diffusion. Clarendon Press, Oxford, England. Christensen, L.H., Christensen, B.H., 2004. Method for transfer of particulate solid products between zones of different pressure. US Patent Application# US 2004/ 0184900 A1. Christensen, B.H., Christensen, L.H., 2007. Method and apparatus for conversion of cellulosic material to ethanol. WIPO Patent# WO/2007/009463. Cussler, E.L., 1984. Diffusion: Mass Transfer in Fluid Systems. Cambridge University Press, Cambridge, UK. del Valle, J.M., Rivera, O., Mattea, M., Ruetsch, L., Daghero, J., Flores, A., 2004. Supercritical CO2 processing of pretreated rosehip seeds: effect of process scale on oil extraction kinetics. The Journal of Supercritical Fluids 31, 159174. del Valle, J.M., de la Fuente, J.C., Cardarelli, D.A., 2005. Contributions to supercritical extraction of vegetable substrates in Latin America. Journal of Food Engineering 67, 3557. del Valle, J.M., de la Fuente, J.C., 2006. Supercritical CO2 extraction of oilseeds: review of kinetic and equilibrium models. Critical Reviews in Food Science and Nutrition 46, 131160. Daz-Reinoso, B., Moure, A., Domnguez, H., Paraj, J.C., 2006. Supercritical CO2 extraction and purication of compounds with antioxidant activity. Journal of Agricultural and Food Chemistry 54, 24412469. Eggers, R., 1988. Gasdichtigkeit von rapssaat unter mechanischer druckbeaufschlagung. Fat Science Technology 90, 184188. Eggers, R., 1994. Extraktion von fettrohstoffen mit berkritischem CO2. Fat Science Technology 96, 513518. Eggers, R., 1996. Supercritical uid extraction of oilseeds/lipids in natural products. In: King, J.W., List, G.R. (Eds.), Supercritical Fluid Technology in Oil and Lipid Chemistry. AOCS Press, Champaign, IL, pp. 3564. Eggers, R., Jaeger, P.T., 2003. Extraction systems. In: Tzia, C., Liadakis, G. (Eds.), Extraction Optimization in Food Engineering. Marcel Dekker, Inc., New York, NY, pp. 95136. Eggers, R., Schade, E.G., 1987. Process for apparatus for the recovery of fats and oils. US Patent# 4, 675, 133. Eggers, R., Sievers, U., 1989. Current state of extraction of natural materials with supercritical uids and developmental trends. In: Johnson, K.P., Penninger, J.M.L. (Eds.), Supercritical Fluid Science and Technology. American Chemical Society, Washington, DC, pp. 478498. Eggers, R., Sievers, U., Stein, W., 1985. High pressure extraction of oil seed. Journal of the American Oil Chemists Society 62, 12221230.

6. Summary There is little information in the literature about the effect of scale on the extraction process in pressurized uid extractors. Often when two scales of an extractor are examined there is no explicit discussion of the effects or the scale-up has not been done uniformly, leaving the effects of the scale-up on extraction in doubt. In general laboratory-scale extractions may be utilized to determine the mass transfer and phase equilibria data for a particular extraction vessel and material. With appropriate models, this data may be used to examine the effects of scale. Caution should however be taken as the effects of scale-up of extractors may not be accounted for in most models. Homogeneous ow is a crucial assumption in most models of extraction but in larger extraction vessels radial variations in bed porosity and viscosity could produce ow heterogeneity. This would result in higher interstitial velocities near the vessel wall, resulting in slower extraction in the centre of the extraction vessel. The goal in the design of pressurized uid extractors is not just to scale the process up, but to produce a system able to operate on a continual basis. There has been some success in this area by utilizing a battery of vessels in series to operate on a quasi-continuous basis, and with the use of screw conveyors to produce a gas-tight plug of material, which allows the extraction to operate at the necessary elevated pressures. Others have utilized the material itself to create a plug that can hold the pressure in the system to operate on a continual basis. Design of a suitable sluice system for charging solid materials between zones of different pressure is another method that has

226

C. Pronyk, G. Mazza / Journal of Food Engineering 95 (2009) 215226 Machmudah, S., Kondo, M., Sasaki, M., Goto, M., Munemasa, J., Yamagata, M., 2008. Pressure effect in supercritical CO2 extraction of plant seeds. The Journal of Supercritical Fluids 44, 301307. Mazza, G., Cacace, J.E., 2005. Extraction of phytochemicals. US Patent Application. Pub. No. US 2007/0014912 A1; Pub. Date, January 18, 2007 (Filing Date: May 13). Meireles, M.A.A., 2003. Supercritical extraction from solid: process design data (20012003). Current Opinion in Solid State and Materials Science 7, 321330. Ong, E.S., Len, S.M., 2003. Pressurized hot water extraction of berberine, baicalein and glycyrrhizin in medicinal plants. Analytica Chimica Acta 482, 8189. Ooi, C.K., Bhaskar, A., Yener, M.S., Tuan, D.Q., Hsu, J., Rizvi, S.S.H., 1996. Continuous supercritical carbon dioxide processing of palm oil. Journal of the American Oil Chemists Society 73, 233237. Osburn, J.O., Katz, D.L., 1944. Structure as a variable in the application of diffusion theory to extract. Transactions of the American Institute of Chemical Engineers 40, 533556. zkal, S.G., Salgn, U., Yener, M.E., 2005. Supercritical carbon dioxide extraction of hazelnut oil. Journal of Food Engineering 69, 217223. Perrut, M., Clavier, J.Y., Poletto, M., Reverchon, E., 1997. Mathematical modeling of sunower seed extraction by supercritical CO2. Industrial & Engineering Chemistry Research 36, 430435. Reverchon, E., De Marco, I., 2006. Supercritical uid extraction and fractionation of natural matter. The Journal of Supercritical Fluids 38, 146166. Richter, B.E., Jones, B.A., Ezzell, J.L., Porter, N.L., 1996. Accelerated solvent extraction: a technique for sample preparation. Analytical Chemistry 68, 10331039. Ritcey, G.M., 2004. Development of industrial solvent extraction processes. In: Rydberg, J., Cox, M., Musikas, C., Choppin, G.R. (Eds.), Solvent Extraction Principles and Practice. Marcel Dekker, Inc., New York, NY, pp. 277337. Rovio, S., Hartonen, K., Holm, Y., Hiltunen, R., Riekkola, M.L., 1999. Extraction of clove using pressurized hot water. Flavour and Fragrance Journal 14, 399404. Schwartzberg, H.G., Chao, R.Y., 1982. Solute diffusivities in leaching processes. Food Technology 36, 7376. Shalmashi, A., Golmohammad, F., Eikani, M.H., 2008. Subcritical water extraction of caffeine from black tea leaf of Iran. Journal of Food Process Engineering 31, 330 338. Shi, J., Zhou, X., 2007. Solubility of food components and product recovery in the supercritical uid separation process. In: Shi, J. (Ed.), Functional Food Ingredients and Nutraceuticals: Processing Technologies. Taylor & Francis Group, Boca Raton, FL, pp. 4573. Shi, J., Kassama, L.S., Kakuda, Y., 2007. Supercritical uid technology for extraction of bioactive components. In: Shi, J. (Ed.), Functional Food Ingredients and Nutraceuticals: Processing Technologies. Taylor & Francis Group Boca Raton, FL, pp. 343. Singh, B., Rizvi, S.S.H., 1994. Design and economic analysis for continuous countercurrent processing of milk fat with supercritical carbon dioxide. Journal of Dairy Science 77, 17311745. Smith, R.M., 2002. Extractions with superheated water. Journal of Chromatography A 975, 3146. Soto Ayala, R., Luque de Castro, M.D., 2001. Continuous subcritical water extraction as a useful tool for isolation of edible essential oils. Food Chemistry 75, 109 113. Sovov, H., 1994. Rate of the vegetable oil extraction with supercritical CO2: I. Modelling of extraction curves. Chemical Engineering Science 49, 409414. Thomsen, M.H., Thygesen, A., Jrgensen, H., Larsen, J., Christensen, B.H., Thomsen, A.B., 2006. Preliminary results on optimization of pilot scale pretreatment of wheat straw used in coproduction of bioethanol and electricity. Applied Biochemistry and Biotechnology 129132, 448460. Thomsen, M.H., Thygesen, A., Thomsen, A.B., 2008. Hydrothermal treatment of wheat straw at pilot plant scale using a three-step reactor system aiming at high hemicellulose recovery, high cellulose digestibility and low lignin hydrolysis. Bioresource Technology 99, 42214228. Tzia, C., Liadakis, G., 2003. Extraction Optimization in Food Engineering. Marcel Dekker, Inc., New York, NY. Wang, B.J., Lien, Y.H., Yu, Z.R., 2004. Supercritical uid extractive fractionation study of the antioxidant activities of propolis. Food Chemistry 86, 237243. Wingerson, R.C., 2006. Apparatus for the separation and treatment of solid biomass. US Patent Application# US 2006/0283995 A1. Wingerson, R.C., Lehrburger, E., 2004. Completing pre-pilot tasks to scale up biomass fractionation pre-treatment apparatus from batch to continuous. Inventions and Innovations Final Report, US Department of Energy, Washington, DC.

Esquvel, M.M., Ribeiro, M.A., Bernardo-Gil, M.G., 1999. Supercritical extraction of savory oil: study of antioxidant activity and extract characterization. The Journal of Supercritical Fluids 14, 129138. Ghoreishi, S.M., Shahrestani, R.G., Ghaziaskar, S.H., 2008. Subcritical water extraction of mannitol from olive leaves. Proceedings of World Academy of Science, Engineering and Technology 33, 114124. Gogus, F., Ozel, M.Z., Lewis, A.C., 2005. Superheated water extraction of essential oils of Origanum micranthum. Journal of Chromatographic Science 43, 8791. , ., Temelli, F., 2005. Solubility behavior of ternary systems of Gl-stndag lipids, cosolvents and supercritical carbon dioxide and processing aspects. The Journal of Supercritical Fluids 36, 115. Haar, L., Gallagher, J.S., Kell, G.S., 1984. NBC/NRC Steam Tables: Thermodynamic and Transport Properties and Computer Programs for Vapor and Liquid States of Water in SI Units. Hemisphere Publishing Corporation, Washington, DC. Hawthorne, S.B., Grabanski, C.B., Martin, E., Miller, D.J., 2000. Comparisons of soxhlet extraction, pressurized liquid extraction, supercritical uid extraction and subcritical water extraction for environmental solids: recovery, selectivity and effects on sample matrix. Journal of Chromatography A 892, 421433. Herrero, M., Cifuentes, A., Ibaez, E., 2006. Sub- and supercritical uid extraction of functional ingredients from different natural sources: plants, food-by-products, algae and microalgae. A review. Food Chemistry 98, 136148. Ho, C.H.L., Cacace, J.E., Mazza, G., 2008. Mass transfer during pressurized low polarity water extraction of lignans from axseed meal. Journal of Food Engineering 89, 6471. Ibez, E., Benavides, A.M.H., Seorns, F.J., Reglero, G., 2002. Concentration of sterols and tocopherols from olive oil with supercritical carbon dioxide. Journal of the American Oil Chemists Society 79, 12551260. Kawamura, F., Kikuchi, Y., Ohira, T., Yatagai, M., 1999. Accelerated solvent extraction of paclitaxel and related compound from the bark of Taxus cuspidate. Journal of Natural Products 62, 244247. Kaufmann, B., Christen, P., 2002. Recent extraction techniques for natural products: microwave-assisted extraction and pressurised solvent extraction. Phytochemical Analysis 13, 105113. Kim, J., Mazza, G., 2006. Optimization of extraction of phenolic compounds from ax shives by pressurized low-polarity water. Journal of Agricultural and Food Chemistry 54, 75757584. Kim, J.-K., Mazza, G., 2009. Extraction and separation of carbohydrates and phenolic compounds in ax shives with pH-controlled pressurized low polarity water. Journal of Agricultural and Food Chemistry http://pubs.acs.org/doi/abs/ 10.1021/jf803467y. King, M.B., Bott, T.R., 1993. Extraction of Natural Products Using Near-Critical Solvents. Blakie Academic & Professional, Bishopbriggs, Glasgow. King, J.W., Cygnarowicz-Provost, M., Favati, F., 1997. Supercritical uid extraction of evening primrose oil kinetic and mass transfer effects. Italian Journal of Food Science 9, 193204. Kubtov, A., Jansen, B., Vaudoisot, J.F., Hawthorne, S.B., 2002. Thermodynamic and kinetic models for the extraction of essential oil from savory and polycyclic aromatic hydrocarbons from soil with hot (subcritical) water and supercritical CO2. Journal of Chromatography A 975, 175188. Kumoro, A.C., Hasan, M., 2006. Modelling of supercritical carbon dioxide extraction of andrographolide from Andrographis paniculata leaves by employing integral desorption concept. International Journal of Engineering and Technology 3, 13 20. Lack, E., Seidlitz, H., 1993. Commercial scale decaffeination of coffee and tea using supercritical CO2. In: King, M.B., Bott, T.R. (Eds.), Extraction of Natural Products Using Near-Critical Solvents. Blakie Academic & Professional, Bishopbriggs, Glasgow, pp. 101139. Lagadec, A.J.M., Miller, D.J., Lilke, A.L., Hawthorne, S.B., 2000. Pilot-scale subcritical water remediation of polycyclic aromatic hydrocarbon- and pesticidecontaminated soil. Environmental Science & Technology 34, 15421548. Laurent, A., Lack, E., Gamse, T., Marr, R., 2001. Separation operations and equipment. In: Bertucco, A., Vetter, G. (Eds.), High Pressure Process Technology: Fundamentals and Applications. Elsevier Science BV, Amsterdam, The Netherlands, pp. 351403. Lee, A.K.K., Bulley, N.R., Fattori, M., Meisen, A., 1986. Modelling of supercritical carbon dioxide extraction of canola. Journal of the American Oil Chemists Society 63, 921925. Leyers, W.E., Novak, R.A., Linnig, D.A., 1991. The economics of supercritical coffee decaffeination. In: McHugh, M.A. (Ed.), Proceedings of the Second International Symposium on Supercritical Fluids. John Hopkins University, Baltimore, MD, p. 261.