Pressurized Organic Solvent Extraction with On-line Particle Formation by Supercritical Anti Solvent Processes

In this work, anovel on-line process for pressurised hot organic solvent extraction of antioxidants from plantsas well as precip itation of the extract with or without a carrier material in one step was developed. This process has been called OEPO,Organic solvent Ext raction and On-line particle formation. With this process, different products with a very low residual organic solvent concentration (< 50 ppm) can be obtained by the use of supercritical CO2 as anti solvent for solvent elimination.OEPO process consists of hyphenated Pressurized Liquid Extraction (PLE)-Supercritical Anti So lvent (SAS) precipitation, PLE-SAS co-precipitat ion and PLE-Supercritical Fluid Extraction of Emulsions (SFEE). OEPO process was successfully developed using Brazilian g inseng roots (Pfa ffiaglomerata)as a model case using ethyl acetate as extracting solvent. Results were compared, in terms of antioxidant activity o r morphology, with the ones obtained by each process separately.In addition, an optimizat ion study for antioxidants recovery was performed using ethyl acetate as extracting solvent during PLE process. Optimum PLE extracts were produced under moderate extraction temperature (373 K) and high static extraction t ime (15 min). Under this condition an ext raction yield o f 1% (dry basis, d.b.) and an antioxidant activity of 53% are obtained, which was approximately 14% higherthan that observed after PLE-SAS precipitation and after SAS precipitation performed in two steps (step one PLE ext raction; step two – SAS precipitation by the use of the ext ract solution produced by step one stored).Similar behavior (hyphenated process producing similar p roducts than the two step process done separately) was observed for PLE-SAS co-precip itation and PLE-SFEE indicating that the OEPO process developed in this work can be considered as a suitable and p romising process to obtain, in only one step, different products (precipitated extract, co-precipitated extract or encapsulated extract in suspension), direct ly from plant materials.


Introduction
Nowadays, the demand for natural b ioactive co mpounds is increasing due to their use in the functional food industry. Natural co mponents fro m p lants are emp loyed, including different functional act ivit ies, for instance,antio xidantactiv i ty, antimicrobial activity, anti-cancer, o r neurodegenerative diseases prevention, among others [1]. Ginseng species is one of the most app reciated natural sources fo r th is kind of compounds.The most known Ginseng species in the world belongs t o t hePan ax genus, wh ich hav e been used for thousands years by folk medicine. Asian ginseng (Panax ginseng), A merican g inseng (Panax quinquefolius) roots arerenowned and widely used herbs in China, Un ited States, Canada, etc. [2].
Species of the genus Pfaffia (A maranthaceae) has been commercialized as substitutes for Panax (ginseng, Araliaceae). Due to the similar mo rphology of its roots to those of ginseng, they are popularly known as "Brazilian ginseng". Around 90 species of Pfaffia are known in Central and South America [3]. In Brazil, 27 species have been described, being Pfaffia glomerata the most important specie. Since besides similarity in appearance Brazilian ginseng roots (Pfaffia glomerata) extracts have also similar effects to ginseng, large amounts of this plant material are being exported for production of their extracts [4].
Different classical ext raction techniques have been applied to obtain antioxidant extracts from Pfa ffia glomerata roots [5][6][7]. Classical extraction methods are time-and solvent-consuming and may pro mote extract degradation during the extraction process. On the other hand, pressurized liquid ext raction (PLE) technique enables the rapid extraction (less than 30 min) of analytes in a closed and inert environment under high pressures (no higher than 20 MPa) and temperatures (298-473 K). A major advantage of PLE over conventional solvent extraction methods conducted at atmospheric pressure is that pressurized solvents remain in a liquid state well above their boiling points, allowing for high-temperature extraction. These conditions improve analyte solubility and the kinetics of desorption from matrices [8].
The lo w stability during ext raction, formu lat ion, purificat ion and storage of some class of bioactive co mpounds has influenced all these steps, which are being studied by different researchers interested in novel forms for processing these compounds with minimu m degradation [9].The most important bioactive principles of Ginseng are saponins. Gradual degradation was observed with further increase in temperature resulting in co mplete destruction of saponins at temperatures > 543 K. Hydrothermo lysis of triterpenoid and steroid saponins occurred upon heating in water at 473-513 K resulting in the production of aglycones, prosapogenins, and sugars [10,11]. Thus, the extract ing solvent elimination step also should be done quickly and using mild operation conditions of temperature. Evaporation step usually expose the extracts to a risk of degradation of b ioactive co mpounds catalysed by heat besides light and/or oxygen.
In the search for alternative solvent elimination processes that can keep the stability of the extracted co mpounds, we have focused our attention on the use of supercritical fluids. Supercrit ical CO 2 and organic solvents are miscible above a moderate p ressure and temperature, wh ile the co mpoundor class of compounds are not soluble in the mixture and it precipitates [12]. Denominated asSupercritical Anti So lvent (SAS) precipitation this method has been used extensively mainly to obtain small part icles with narrow part icle size distribution, which can be encapsulated, if required, by its co-precip itation together a carrier material (SA S co-precipit ation) or by the formu lation of an emulsion [Supercritical Flu id Extract ion of Emu lsions (SFEE)] [13,14].
The encapsulation of natural substances presents several advantages over the natural substance itself. First, they acquire controlled release behaviour and are able to maintain their stability for longer periods [15].
The development of hyphenated processes for combining bioactive compounds extraction with on-line pa rticle formation is rarely reported. Ibanez group in Spain, recently, have developed a hyphenated process to obtain dried powders of extracts from natural sources in one step. This process called Water Extract ion and Particle formation On-line (W EPO) similarly to ours also use PLE technique for bioactive compounds recovery, but employ water as extracting solvent. Since, in their case supercritical CO 2 are not suitable for solvent elimination due to the low solubility of water in CO 2 , this flu id is used as a dispersion mediu m and hot N 2 is used as drying agent [16]. Due to the similarit ies of WEPO process we have named our process asOrganic solvent Extraction and Part icle formation On-line (OEPO). Differently of WEPO process, OEPO process also permits the encapsulation of the extract immediately after their production.Indeed, OEPO p rocess consists of hyphenated PLE-SAS precipitation, PLE-SAS co-precipitation and PLE-SFEE.
In this study the OEPO device and procedures are described in detail and successfully developed using Brazilian ginseng roots (Pfaffia glomerata) as a model case.Results were compared,in terms of antio xidant activity or morphology, with the ones obtained by each process separately. In addition, an optimization study for antioxidants recovery was performed using ethyl acetate as extracting solvent during PLE process and with static extraction t ime (5-15 min) and extract ion temperature (353-413 K) as independent variables.The organic solvent ethyl acetate was chosen because it is a Generally Recognized as Safe (GRA S) solvent according to the US Food and Drug Admin istration (FDA) (to xicological class 3), it can be safely used in food applications [17,18].

Materials
Brazilian ginseng roots (Pfa ffiaglomerata) were cultivated in the experimental field of CPQBA (Camp inas, Brazil), where they were collected on March 25, 2004, being 3 years old. They were washed and dried in a forced air circulation dryer at 313 K for 5 days. The dried roots (8.9% mo isture) were then comminuted in a pulse mill (Marconi, model MA 340, Piracicaba, Brazil) fo r few seconds. Next, the particles of higher size were milled again, this time using a knife mill (Tecnal, model TE 631, Piracicaba, Brazil) for 2 s at 18,000rp m and finally, they were separated according to their size using sieves (Series Tyler, W.S. Ty ler, Wheeling, IL). The milled roots were stored in freezer (Metalfrio, model DA 420, São Paulo, Brazil) at 263 K. For the extraction assays, particles of 7.89 µm of d iameter, according to ASAE methodology [19], were used. The mo isture content of the dried roots was determined by the AOAC method (Method 4.1.03) [20].
Ethyl acetate (analytical grade) purchased from Merck (Darmstadt, Germany) was used as extracting solvent. Dich loro methane (analytical grade) purchased from Merck (Darmstadt, Germany) was used to prepare the Polyethylene glycol (PEG) solutions.
N-octenyl succin ic anhydride (OSA)-modified starch, kin dly provided by National Starch Food Innovation (Hamburg, Germany), was used as surfactant and carrier material.

Pressurized Li qui d Extraction (PLE)
The pressurized liqu id ext raction (PLE) system was designed and assembled at LASEFI/DEA/FEA (School of Food Engineering)/UNICAMP (University o f Camp inas). The solvent was pumped by a HPLC pu mp (Thermoseparati on Products, Model ConstaMetric 3200 P/F, Fremon i, USA) into the extraction cell, which was placed in an electrical heating jacket at a desired temperature, until the required pressure was obtained.
Dried and milled pieces of Brazilian ginseng roots (4.5 g with a moisture content of 8.9 %) were placed in a 6.57-cm 3 extraction cell (Thar Designs, Pittsburg, USA) containing a sintered metal filter at the bottom and upper parts. The cell containing the sample was heated, filled with extraction solvent (ethyl acetate) and then pressurized. The samp le was placed in the heating system for 6.5 min to ensure that the extract ion cell would be at the desired temperature during the filling and pressurization procedure. After pressurization, the sample with pressurized solvent was kept statically at the desired pressure for the desired time (static extraction time). The pressure of the extraction was set in all experiments to 12 MPa to simu late the conditions of the OEPO p rocess.Thereafter, the back pressure regulator (BPR) valve (Model #26-1761-24-161, Tesco, Elk River, USA) was carefully opened, keeping the pressure at an appropriate level for the desired flo w (1.0 cm 3 / min), to rinse the extraction cell with fresh extracting solvent for 20 min (dynamic ext raction time). After pressurized liquid extraction (PLE), the extracts were rap idly cooled to 268 K in ice water using glass flasks to prevent ext ract degradation. After extraction, depending on the aim different procedures were done. If the aim was to determine the extraction yield and the extract antioxidant activities, the solvent was evaporated using a rotary evaporator (Laborota, model 4001, Vertrieb, Germany) with vacuum control (Heidolph Instruments Gmbh, Vertrieb, Germany) and a thermostatic bath held at 313 K. On the other hand, if the aim was to use it in the particle formation processes, the 20 cm 3 of ethyl acetate extract solution produced in each experiment were directly stored. All ext racts (dried or not) were stored (263 K) in the dark prior to the next step (analysis or particle formation). Statistica' software (release 7, StatSoft, Tulsa, USA ) was used to calculate the effects of extraction temperature and static extract ion time on ext raction yield and antio xidant activity (Table 1). A ll extractions were performed in duplicate. Statistical analyses were performed using analyses of variance (ANOVA). The mean values were considered significantly different at p <0.05.The prediction of one set of optimal conditions for both response variables was done by using desirability function approach.

Supercritical Anti Sol vent (SAS) Precipi tation Process
The experiments for ethyl acetate Pfaffiaglomerata roots extract precip itation were performed in a ho memade Supercrit ical Anti So lvent (SAS) equip ment designed and assembled at LASEFI/ DEA/FEA (School of Food Engineering)/ UNICAMP (University of Camp inas) emp loyi ng the ethyl acetate extract solution produced by PLE.
Liquid CO 2 , antisolvent, was fed fro m the cy linder through athermostatic bath (MA-184, Marconi, Piracicaba, Brazil) at 263 K to ensure the liquefaction of the gas and to preventcavitation, then it was pumped by an air-driven liquid pump(MaximatorGmbh, PP 111, Zorge, Germany) to the highpressurevessel (volu me of 500 cm 3 ; 6.8-cm internal diameter)via a no zzle.The nozzle consists of a 1/16" tube(inner diameter[i.d.]: 177.8millimeters) for the solution, placedinside a 1/8" tube for the CO 2 . Once the particle formationvessel reached steady state (temperature-313 K, pressure-10 MPa andCO 2 flo w rate-0.6 kg/h), the ethyl acetate extract solution was introduced into the vesselby a high-performance liquid chro matography (HPLC)pu mp (Model ConstaMetric 3200 P/F, ThermoseparationProducts, Fremont, USA) through the coaxial annular passageof the atomizer at a constant flow rate. The flow rate of the solution was set in all experiments to 1.0 cm 3 / min to simulate the conditions of the OEPO process. The vessel temperature (313 K) was maintained by a heating waterbath (MA 127BO, Marconi, Piracicaba, Brazil). CO 2 flo w rate (0.6 kg/h) was measuredusing a glass float rotameter (0.15-2.2 kg/h of CO 2 at0.1013 MPa/293 K; 16/286A/ 2, A BB, Warminster, PA)coupled to a flow totalizer (Model G0,6, LAO, Osasco,Brazil). When the desired amount (20 cm 3 ) of solution (ethyl acetate Pfaffiaglomerata roots extract) has been injected, which enabledthe collection of sufficient amount of p recip itated powderfor analysis, the HPLC pu mp was stopped and only pureCO 2 was fed. The flo w of CO 2 was maintained fo r 20 min forthe co mp lete removal of the solvent from the precipitator,wh ich was proven necessary by preliminary experiments.
Pfaffia glomerata roots extract precip itates were trapped by apaper filter fixed at the bottom of the precipitation vessel while the fluid mixtu re (CO 2 +ethyl acetate) flo wedto a second vessel (100-cm 3 glass flask). At the end, the precipitation vessel was slowly depressurizedto atmospheric pressure and particles were collectedand stored in the dark in a domestic freezer (263 K;Double Action, Metalfrio, São Supercritical Anti Solvent Processes Paulo, Brazil) until subsequentanalysis and characterizat ion. A heating system maintained at 353 K was used to heat the micro met ric valve toavoid the Joule-Thompson freezing effect that can lead toclogging of the throttling device during particle format ionprocedure.

Supercritical Anti Sol vent (SAS) Co-Preci pitati on Process
The experiments for ethyl acetate Pfaffiaglomerata roots extract co-precip itation with PEG were done in the SAS equipment previously described. The SAS co-precipitation procedure is very similar tothe SAS precip itation, differing that a carrier material was added to theethyl acetate extract solution produced by PLE. PEG, in this case, was the carrier material and dich loro methane was the solvent. Dich loro met hane was the selected solvent because it is a good solvent for PEG. Thus, in this case Pfaffia glomerata roots extract in PEG co-precip itates were trapped by a paper filter fixed at the bottom of the vessel while the fluid mixture (CO 2 +ethyl acetate+dichloromethane) exited the vessel.
The operating condition were the same that during SAS precipitation (10 MPa and 313 K, CO 2 flow rate of 0.6 kg/h, ethyl acetate solution flow rateof 1.0 cm 3 / min). The mass ratio between Pfaffiaglomerata roots extract and PEG investigated was 1:10

Supercritical Flui d Extracti on of Emulsions (SFEE) Process
Before init iating the SFEE process, an oil-in -water emu lsion must be prepared. In general, these emu lsions are prepared with the aid of surfactants.
Twentycubic centimeterof the ethyl acetate solution produced by PLE was dispersed into 80 cm 3 of an aqueous solution withOSA-modified starch surfactant (6 g/dm 3 ) by the aid of a h igh speed-stirring mixer(IKA® magic LA B ® , Staufen, Germany) with an engine power of 900 Watt processed during 4 min at 26.000 rp m. The mixer was cooled by ethylene glycol that circulates through a jacket, which allo ws to remove the heat generated by the equipment and to operate at temperatures lower than 298 K to avoid ethyl acetate evaporation.
The SFEE experiments were done also in the SAS equipment previously described. The SFFE procedure and operating conditions were the same that during SAS precipitation and SAS co-precip itation, only d iffering that instead of a solution,an emulsion with ethyl acetate extract solution produced by PLEwas used as the dispersed phase. Then, in this case the suspension containing Pfaffia glomerata roots extract encapsulated in OSA-starch micelles remained in the precipitation vessel while the fluid mixtu re (CO 2 +ethyl acetate) exited.
Afterwards, the suspensions obtained were further processed removing water to produce a dry powder. Th is was done by freeze-dry ing for 5 days at 60-100µHg and at -223K (Liobras, Liotop L101, São Carlos, Brazil).

Organic Sol vent Extraction and Particle Formati on On-line (OEPO) Process
The Organic solvent Ext raction and Part icle format ion On-line (OEPO) process combines the two different processes previously described: firstly a dynamic PLE process using organic solvents and secondly the elimination of the solvent by the precipitation of the extract, using supercritical CO 2 as a anti solvent. Thus, extraction and precipitation take place in the same system with a small time delay between these two processes. Figure 1 shows a scheme of the home-built equip ment designed to carry out the organic solvent extraction with particle format ion on-line (OEPO).
Fro m the ext raction cell, the ext ract solution is led to a Tmixer where it can be mixed with a solution containing a carrier material d issolved also in an organic solvent or with an aqueous solution of surfactant. Shortly afterwards the solution or emulsion is exited through the coaxial annular passage of the atomizer together with supercritical CO 2 into the precipitation vessel.All connections used for coupling the PLE system with the SAS equip ment were made using stainless steel tubes (1/16" and 1/ 8").
The extraction cell was filled with dried and milled pieces of Brazilian ginseng roots. The amount inserted of plant material was calculated in order to keep the same solvent volume to feed volu me rat io employed during the previous PLE experiments (20 cm 3 of solvent/4.5 g of roots). Theprocess started with a staticextraction period (selected after optimization) by filling the cellwith ethyl acetate at the desired temperature (selected after optimization)and pressure(12 M Pa), with valve V18 closed (see Figure 1).At the same t ime, CO 2 was pu mped through the system at thedesired temperature (313 K) and pressure (10 M Pa), with a constant flow rate(0.6 kg/h). The extract ion continued in a continuous flow mode (dynamic ext raction period) by opening valve V18 and setting the extracting solvent rate at the desired constant value. The ethyl acetate ext ract solution can meet first the solution containing PEG dissolved in dichloro methane or with an aqueous solution with OSA-mod ified starch surfactant (6 g/dm 3 ) pu mped with HPLC pu mp of the SAS equipment.The flo w rate of both solutionswas set in order to achieve a constant total flow rate of ethyl acetate extract solution plus the resulted solution or emulsion of 1.0 cm 3 / min. When the aimed process was PLE-SAS precipitation, the second HPLC pump was turned off, since in this process there is no need of any carrier or surfactant material addit ion.Afterwards theorganic solvent (ethyl acetate) or solvents (ethyl acetate+dichloromethane) fro m the solution or emulsion areexited through the vessel precipitating the product, which can be a: i) precipitated extract (product after PLE-SAS precipitation); ii) co-precipitated extract(product after PLE-SAS co-p recipitation) or iii)encapsulated extract in suspension (product after PLE-SFEE).
The suspensions obtained by PLE-SFEE were further freeze-dried to produce a dry powder as previously described.

Antio xidant Activity (AA)
The evaluation of antio xidant activ ity of the extracts was based on the coupled oxidation of β-carotene and linoleic acid. The technique developed by Marco [21] consisted of measuring the b leaching of β-carotene resulting fro m oxidation by the degradation products of linoleic acid. In short, the substrate of reaction was prepared using 10 mg of β-carotene (97%, Sig ma-Aldrich, St. Louis, USA), 10cm 3 of chloroform (99%, Ecibra, Santo Amaro, Brazil), 60mg of linoleic acid (99%, Sig ma-Aldrich, St. Louis, USA) and 200mg of Tween 40 (99%, Sig ma-Aldrich, St. Louis, USA). This solution was concentratedin rotary evaporator (Laborota, model 4001, Vertrieb, Germany), with vacuum control (Heidolph Instruments Gmbh, Vertrieb, Germany) and a thermostatic bath at 323 K, being then diluted in 50cm 3 ofdistilled water. The o xidation reaction was conducted using thefollowing procedure: to each 1 cm 3 of substrate, 2 cm 3 of distilled water and 0.05 cm 3 of ext ract diluted in ethanol (99.5%, Ecibra, Santo Amaro, Brazil) were added. The dilution used for AA determination was 0.02 g of extract/cm 3 of solvent. The mixture was placed in thermal bath (model TE 159, Tecnal, Piracicaba, Brazil Marconi, model MA159/300, Piracicaba, Brazil) at 313 K, and the product of reaction was monitored using a spectrophotomet er (Femto, model 800 XI, São Pau lo, Brazil Hitachi, model U-3010, To kyo, Japan) at 0, 1, 2 and 3 h of reaction, using absorbance at 470 n m. The antio xidant activity was determined in duplicate for each extract and calculated following the same calculation p rocedure done by Santos et al. [22]. Antio xidant activity for synthetic BHT (at the same concentration that the extract) was also determined for comparison.

M icroscopy
Micrographs of the particles collected were taken by means of a scanning electron microscope (SEM ) (LEO 440i, Leica, Cambridge, USA) after coating with a thin gold film with the aid of a sputter coater (Polaron, SC 7620, Ring mer, U.K.).

Effects of PLE Process Variables on the Extraction Yiel d and Antioxi dant Acti vi ty
The effects of extract ion temperature and static ext raction time on the extract ion yield and on antio xidant activity were evaluated. The experimental values at various experimental conditions are presented in Table 2. In the variable ranges of 353-413 K and 5-15 min, the extraction yield variab le wassignificantly (95% confidence level, p < 0.05) affected by extraction temperature, static extract ion time and their interaction. On the other hand, only ext raction temperature was significant (95% confidence level, p < 0.05) with respect to the antioxidant activity.
The relationship of the extraction yield, ext raction temperature and static extraction time was linear. An increase in either of temperature and static time, wh ile the second variable remains constant, results in enhancement of the extract recovery. Moreover, the interaction between them had also a positive effect on the production of extract(Tab le 2). With regards of antio xidant activity, the increase of the extraction temperature beyond 373 K possibly might enhance the degradation of the bioactive compound extracted decreasing the antioxidant activity. An increase in extraction temperature is reported to improve the efficiency of extraction because of enhanced diffusion rate and solubility of analytes in solvents; neverth eless, high extraction temperatures may simultaneously increase the degradation rate of some ext racted compounds [8].
The use of high temperature pressurized solvents in saponin processing is still limited. Mazza group in Canada has been demonstrated that during PLE ext raction degradation of the some saponins from cow cockle seedsoccurs at 398 K [10,11]. Besides saponin degradation, the increase in the content of non-saponin compounds in the extracts with temperature can also helped with the decrease in the antio xidant activity.
Although the extraction yields beyond 373K had been larger, the antio xidant activities were smaller. This confirms that the ext racts obtained under higher temperatures contain other compounds classes that can be reducing the antio xidant activity.The prediction of one set of optimal conditions for both response variables was done by using desirability function approach (Figure 2). In particular, high desirability, within the experimental design values, has been achieved under moderate ext raction temperature (373 K) and high static ext raction t ime (15 min).Under this condition an ext raction y ield of 0.934 % (dry basis, d.b.) and an antio xidant activity of 52.96 % are obtained.
Co mparing with synthetic BHT (48.62 %) and with literature data (up to 25 %) emp loying supercritical CO 2 as extracting solvent [23], it is demonstrated the potentiality of our PLE ext ractsusing ethyl acetate as extract ing solvent.
The dynamic ext raction time of 20 min was decided after preliminary studies and based on literature data [8]. Obviously, the dynamic ext raction time play an important role on the ext raction yield, which will be evaluated in the future experiments.

Comparison betweenthe PLE Extracts, SAS Precipitated Extract and the Preci pitated Extract Obtained from the OEPO Process
The extract obtained under optimu m PLE conditions was used as a control to compare antio xidant activity with the SAS precipitate ext ract and the precipitated extract obtained fro m the OEPO process (PLE-SAS precip itation) The pressure during the PLE p rocesswere set in all experiments to 12 MPa, given that, this condition permits the coupling of SAS precipitation (and other anti solvent processes) after PLE.
A slightly difference (14.24% lo wer) was observed between the PLE and OEPO samples, meaning thatthe OEPO process effects the antioxidantactivity of the extracts. SAS precip itation and OEPO process presented no significant difference (< 5%) between the samples, as expected. This difference between the samples can be attributed to the solubilization of some co mpounds in supercritical CO 2 during anti solvent process [24]. Further improvements in our equipment in order to determine this loss will be done.
A reference point for the anti solvent processes is the mixtu re critical point for the binary system CO 2 -organic solvent. In addition to the complete miscibility of the selected organic solvent in the supercritical anti solvent, a partial solubilization of some class of compounds can be aimed for fractionation purposes [25]. Recently, good results were obtained to fract ionate the antio xidants fro m methanol ic extract solution obtained from grape wastes without degradation and with the complete elimination of the solvent residues [26]. In our case, the fractionation phenomenon was undesired, but tuning the supercritical CO 2 density, we canprobably reduce the observed difference or even improve the antioxidant activity. Figure 3 shows the pictures of SAS co-precipitated ext ract and the co-precipitated ext ract obtained fro m OEPO p rocess (PLE-SAS co-p recip itation) obtained by scanning electron microscope (SEM).

Comparison between the SAS Co-Preci pitated Extract and the Co-Preci pitated Extract Obtained from the OEPO Process
No significant differences were observed between the samples, meaning that the OEPO process does not have positive nor adverse effect on the morphology of the co-precipitated extracts.
Recently, our research group evaluated the influence of several process variables during SAS co-precipitation of bixin-rich extract also in PEG [27]. Taking into account this study, we selected operating conditions that effectively encapsulate the extract minimizing extract and carrier material losses with supercritical CO 2 flo w. Obviously, as well as occurred during SAS precipitation the extract loss may have reduced the antioxidant activity. Once again, this antioxidant activity change can be avoided with the optimization of the anti solvent process.

Comparison between the Dried S FEE Enc apsulate d Extract and the Dried Encapsul ated Extract Obtained from the OEPO Process
OEPO p rocess was also effective for the production of encapsulated extract. In Figure 4, SEM images of freeze-dried encapsulated ext racts suspensions in water obtained by SFEE p rocess and PLE-SFEE process, respectively. It is observed that both produced dried particles were spherical. Otherwise, the particles produced by OEPO process presented higher degree of porosity in their surface and higher particle d iameter than SFEE part icles.
Briefly, in the SFEE process the emulsion and the supercritical CO 2 are injected into the precipitation vessel continuously, and CO 2 diffuses through the aqueous phase to the drop, extracting organic solvent out of the drop and precipitating the solute dissolved into the organic phase due to the anti solvent effect of the carbon dio xide [28].
In SFEE process, like in the previously reported anti solvent processes, operating conditions, in general, are selected in order to facilitate maximu m extraction of the organic phase with min imu m extract and carrier material losses due to dissolution in CO 2 . In contrast, process variables like pressure and temperature are more closely related to the capacity to eliminate the remain ing organic solvent from the products than to the final particle size [29].
The preparation of the emulsion has demonstrated to be the crucial point for the production of fine particles. The similarity between particle sizes of the in itial emu lsion and the final suspension suggests that the final part icle size is dependent of the original droplet size of the emulsion [28][29][30]. Thus, the production of larger particles by OEPO process can be associated to the quality of the emulsion prepared using the T-mixer co mpared with that emulsion prepared using high speed-stirring mixer.  The fabrication of water-in-o il emulsions is a process with widespread applications in formu lation engineering. The idea to create the emulsion in line and operate the process in a continuous mode is not new, existing several commercial in-line dispersing technologiesalready available [31]. Better results could be obtained with theoptimization of this point during OEPO process. Further experiments will be done in this direction. In addition, the use of ultrasound irradiation in this stage and also in the PLE processin order to try to enhance the extract's antio xidant activity will be studied in the future for the imp rovement of OEPO p rocess.
Ev idently, the OEPO process can be used for other plant materials. The use of other o rganic solvents as extracting solvents though is limited since they have to be miscible in the supercritical fluid. Another limitation for the choice of the organic solvent is the international regulations regarding the safety of consumers. Others class 3 solvents (like ethyl acetate) that meet both requirements are: ethanol, acetone, Dimethyl sulfo xide, among others. With regard to the choice of the carrier material, their bioco mpatibility and lack of toxicity are of course important considerations as well their low solubility in the supercritical flu id. Polyethylene glycol (PEG),used in this study, has been extensively used for precipitation and co-precipitation studies with supercritical flu ids due to achieve both needs [13].A disadvantage of the use of PEG, poly (3-hydroxybutyrate-co-3-hydro xyvalerate) (PHBV), among others, is that they are only d issolved in dichloro methane (class 2 solvent -which means that it can be used with products for hu man consumption, although subject to a legal concentration limit in the final product). Thus, biodegradable polymers that can be dissolved in lower to xic solvents such as poly-lactic acid (PLA) (dissolves in ethyl acetate) should be preferred for PLE-SAS co-precipitation [32]. For SFEE during the hyphenated PLE-SFEE process, some additionalbenefits can be obtained if the surfactant material can alsobe used as carrier material. N-octenylsucci nic anhydride (OSA)-modified starch surfactant, employed in this study, isbeen expansively used because they are suitable for food and nutraceutical applicationsand they are also capable of providing a double functionalityas a surfactant for the emulsion stabilizat ion and as a carrier materialin the final product [14]. Another versatility of our process is related to the possibility of change the pressure during the precip itation stage, wh ich can as prev iously reported, promote a fract ionation/purification of the extract produced. For this, we have to set the pressure of the previous stage (PLE ext raction) higher or at least equal to the next stage (Anti solvent process). In this study, we have set the PLE p ressure at 12 MPa (120 bar) due to the range usually used for the anti solvent processes studied be in the range of 8-12 M Pa [13][14][15]. Obviously, together with a fractionation/purificat ion the organic solvent elimination rate will also be changed producing products with different residual organic solvent concentrations. This analysis of the residual ethyl acetate concentration was not done in this work, but at the operating conditions employed during the anti solvent processes it is expected a concentration lower than 50 ppm, wh ile the conventional solvent evaporation results in a residual content of around 500 pp m. [33,34].

Conclusions
Organic solvent Ext raction and Particle formation On-line (OEPO) process was described in detail and successfully developed using Brazilian g inseng roots (Pfa ffia glomerata) as a model case. This novel process consists of an on-line process for pressurized hot organic solvent extraction of plant materials and precipitation of the ext ract with or without a carrier material by organic solvent elimination in one step, based on the use of supercritical CO 2 as anti solvent.
The use of PLE conditions employing ethyl acetate as extracting solvent for obtaining antioxidants from Pfaffia glomerata roots set at 12 MPa and 373 K under a static extraction time o f 15 min was selected for further coupling with SAS precip itation, SAS co-precip itation with PEG and SFEE using OSA-modified starch as surfactant/carrier material. Indeed, OEPO process consists of hyphenated PLE-SAS precipitation, PLE-SAS co-precipitation and PLE-SFEE. Under this PLE condition an ext raction yield of 0.934% (dry basis, d.b.) and an antio xidant activity of 52.96%were obtained, which was slightly higher (14.24%) that was observed after PLE-SAS p recipitation and after SAS precipitat ion performed in two steps (step one -PLE extraction; step two -SAS precipitation by the use of the extract solution produced by step one stored). Similar behavior (hyphenated process producing similar products than the two step process done separately) was observed for PLE-SAS co-precip itation and PLE-SFEE indicating that the OEPO process developed in this work can be considered as a suitable and promising process to obtain, in only one step, different products (precipitated extract, co-precip itated extract or encapsulated extract in suspension) with desired antioxidant activity and particle size, directly fro m plant materials.