J. of Supercritical Fluids 37 (2006) 142–150

Extraction of fat from pigskin with supercritical carbon dioxide

Eva Marıa Vaquero, Sagrario Beltran∗, Maria Teresa SanzArea de Ingenierıa Quımica, Facultad de Ciencias, Universidad de Burgos, Plaza Misael Banuelos s/n, 09001 Burgos, Spain

Received 21 July 2005; received in revised form 31 October 2005; accepted 3 November 2005

Abstracts

A supercritical fluid extraction process for reducing the fat content of pigskin is presented in this work. Carbon dioxide has been used assupercritical solvent for fat extraction. Efficiencies close to 100% were reached. The influence of process parameters – such as extraction temperature,pressure and time and solvent ratio – on the extraction yield was studied. The influence of other variables such as initial fat and water contentof the pigskin was also considered. The results showed that when using pigskin sheets, for non-interrupted extractions, there was an extractionyield limit, of around 60%, that could not be overcame even using an extraction pressure of 65 MPa. Some experiments that were interrupted forsampling of the pigskin showed that there was a positive influence of the intermediate depressurizations of the extractor on the extraction yieldand, in this case, values close to 100% could be reached.

creasing thes s, although,i

r coefficientt©

K

1

uhiMasbno

ftocF

ect-sedits-per-

tionscts. Itxic,sol-

kesoodm

tionas

prod-ture

sentxtrac-ess,

0d

The analysis of the results gave rise to the conclusion that the external mass transfer resistance could be easily eliminated by inolvent flow rate through the extractor. The internal mass transfer resulted to be the main resistance controlling the rate of the procesn the very first part of the extraction, it was the solubility that imposed the limit to the process rate.

A steady-state approximation of the extraction from solids was used to model the experimental data and to obtain the total mass transfehat would allow the scale up of the process.

2005 Elsevier B.V. All rights reserved.

eywords: Supercritical carbon dioxide; Extraction; Fat; Pigskin

. Introduction

Collagen is the most abundant protein in animals that makesp over 75% of the dry fat-free weight of their skin, and iseld to account for its tensile strength[1]. This fibrous protein

s widely used in the Food Industry for various applications.ost often, it is obtained from cattle skin. Nevertheless, it islways advisable to count on alternative sources to eliminatetrong dependences on one of them. Pig (Sus scrofa) skin coulde such an alternative; however, it has a high fat content thateeds to be reduced to meet the specifications required by mostf the applications in the Food Industry.

Although there are some methods to eliminate or reduce theat content of animal skin, we propose in this work a mild methodhat can achieve very high reductions, or even total elimination,f the fat contained in pigskin in order to achieve a protein con-entrate suitable to be used for most of the applications in theood Industry.

∗ Corresponding author. Tel.: +34 947 258810; fax: +34 947 258831.

This method uses supercritical carbon dioxide (SC-CO2) asa solvent that can selectively extract the fat without affing the protein. Carbon dioxide is the most commonly usupercritical fluid due to its mild critical properties – e.g.,critical temperature is 31◦C, which is low enough for processing collagen that may suffer damage under high tematures[2] – and because it is a gas under ambient condithat allows an easy separation of the processed produis also cheap and achievable with high purity; it is non-tonon-corrosive, non-flammable and considered a “greenvent”. Its main inconvenient is its lack of polarity that mait a poor solvent of polar solutes. Nevertheless, it is a gsolvent of fat [3], which is the solute whose removal fropigskin is to be studied in this work. Fat and oil extracwith SC-CO2 from a large variety of natural sources, suchdifferent types of seeds, spices, meat products, marineucts, skin, etc., has been widely reported in the litera[4–6].

The supercritical fluid extraction (SFE) processes preseveral advantages over traditional liquid solvent based etion methods including improved selectivity, expeditiousn

E-mail address: beltran@ubu.es (S. Beltran). automation and environmental safety. For these reasons, they

896-8446/$ – see front matter © 2005 Elsevier B.V. All rights reserved.oi:10.1016/j.supflu.2005.11.003

E.M. Vaquero et al. / J. of Supercritical Fluids 37 (2006) 142–150 143

are being extensively studied for separations involving naturalproducts.

The objective of this work was to identify the optimum pro-cess conditions to obtain a high yield for reducing the fat contentof pigskin by using SC-CO2 as a solvent. The pigskin was defat-ted as part of the treatment for its conditioning to be used asraw material to obtain a collagen paste widely used as sausagecasings[7]. The fat concentration of the non-treated pigskindepends on the zone of the animal body and the thickness of theskin. After stripping off mechanically most of the hypodermis,which is the skin layer with the largest fat content, an average fatcontent as large as 30% was obtained. This value should be low-ered down to at most 6%[7] for pigskin to be used as raw materialto obtain the collagen paste of interest. Some reduction could bedone by working at the appropriate pressure with the machinesused to strip off the skin from the carcass, by eliminating the fatwith some chemicals[7], etc. In this work, we propose to carryout fat reduction of pigskin by a mild and efficient method, asit is supercritical fluid extraction. Different experiments werecarried out to identify the resistances controlling the rate of theextraction process and to search for the process conditions thatovercame such resistances favoring the achievement of a highextraction yield.

2. Experimental

2

oodc s( epi-d arsc elyt rmis[ tains orkw wask n oft 35%p

thefi ately5 edi olledb na kind rmeb itionp nonm

2

malls withd ounn ples

was significant, thus, it was taken into account in data anal-ysis when several samples were withdrawn along an experi-ment. In the case of minced skin, 20 g were also randomlyselected for analysis. The skin samples were analyzed followingthe ISO-1442[10] ISO-1443[11] and ISO937:1978[12] stan-dards to determine their moisture, total fat and protein content,respectively. These analyses permitted the characterization ofthe pigskin used as raw material and to follow the course of theextraction.

2.3. Equipment and procedure

For some of the extraction experiments, the skin was partiallydried in a continuous circulating air dryer, SBANC, Edibon, S.A.The air was fed into the dryer unit at room temperature to avoidprotein damage in the pigskin.

The extraction experiments were carried out in a semi-pilotSFE-plant whose P&I diagram is presented inFig. 1 [13]. Theusual elements of an SFE-plant with solvent recycling wereinstalled, i.e.: pump, extractor, separator, heating and coolingsystems, and pressure dampers; rupture disks were installedfor safety and instruments for measurement and control of theprocess parameters. The size of the plant, with a 2-L extrac-tion vessel and a 1-L separator, allowed to study the validityof a SFE process for new applications and to obtain the pro-cess parameters for scale up. The maximum specifications oft ,F

erep xtrac-t eda tfl u-o thes c-i up to3

ed.A wascea ld.A actorw peri-m lysisa sur-i ess,i thep count,t onca

3

opti-m O

.1. Raw material and pretreatment

The raw material was pigskin provided by a Spanish fompany – Campofrıo Alimentacion S.A. – located in BurgoSpain). The pigskin consists of two major layers, theermis and the dermis. The dermis consists mainly of coollagen fibers. The hypodermis of the pig is a relativhick layer of adipose tissue located ventral to the de8]. This layer was mostly mechanically stripped off to obkin sheets about 5 mm thick. The pigskin used in this was mainly obtained from the back of the animal andept frozen in a vacuum bag until used. The compositiohis skin was around 20–35% fat, 40–50% water and 20–rotein.

The pigskin was cut into pieces before extraction. Forrst set of extraction experiments, small pieces of approximmm× 5 mm (that will be called “minced skin”) were prepar

n order to avoid the rate of the SFE process to be contry the mass transfer in the solid phase[9] and this way, obtaifirst evaluation of the possibilities of using SFE for pigs

efatting. Nevertheless, most of the experiments were perfoy using skin sheets, because some of the stages of the tradrocedures to obtain the collagen paste of interest requireinced skin[7].

.2. Sample analysis

The sampling method consisted in randomly cutting skin pieces from the skin sheets used as raw material, orrawn from the extractor, up to obtain 20 g that was the ameeded for analysis in triplicate. The weight of the sam

e

dal-

-t

he SFE-plant were:T = 200◦C, p = 65 MPa and solvent flow= 20 kg/h.In a SFE experiment, approximately 300 g of pigskin w

laced in the extractor that was later pressurized up to the eion pressure,p, with CO2. Then, the solvent was circulatt the desired extraction temperature,T, with a certain solvenow, F, and during a specific time,t. The solvent was continusly recycled to the extractor after removing the solute ineparator where the solvent power of CO2 was reduced by redu

ng pressure down to 6 MPa and increasing temperature23 K.

Two types of experimental procedures were followccording to the first experimental procedure, the extractionarried out continuously during an extraction time,t, in order tovaluate the influence of extraction pressure,p, temperature,T,nd time,t, and initial water content, on the extraction yieccording to the second experimental procedure, the extras depressurized several times during the extraction exent in order to withdraw a sample of pigskin, whose anallowed following the course of the extraction. The depres

zations of the extractor had a positive effect on the procncreasing the extraction yield; therefore, their influence onrocess has also been studied. To take this influence into ac

he total extraction time,t, was divided into several extractiycles, adding the cycle time,tc, and the number of cycles,Nc,s new process variables.

. Results and discussion

Three sets of experiments were carried out in order toize the pigskin defatting process by extraction with SC-C2.

144 E.M. Vaquero et al. / J. of Supercritical Fluids 37 (2006) 142–150

Fig. 1. Flowsheet of the SFE semi-pilot plant used in this work: high pressure pump (1); rupture disks (2 and 11); shutoff valves (3, 6, 9, 14, 18, 21 and 25); purgevalves (4, 15 and 20); thermostatic baths (5 and 13); discharge valves (7 and 17); extractor (volume = 2 L)(8); expansion valves (10); circulation pumps (12 and 23);separator (16); CO2 storage (19); check valves (22); cooling bath (24).

The first set of extraction experiments was carried out overminced skin and without depressurizing the extractor along theextraction process, that is, in one only extraction cycle. The sec-ond set of experiments was carried out also in one only extractioncycle, but the skin was not minced but cut into larger pieces.Finally, a third set of experiments was carried out in severalcycles and using also skin sheets.

In all cases, the process was supposed to occur through threeconsecutive steps: internal mass transport, solubility of the solutein the solvent and external mass transport. The experimenta-tion was designed to study the influence of the different processparameters and determine the resistance that controls the pro-cess.

3.1. One-cycle extractions over minced skin

The first set of extraction experiments was carried out overminced skin and in one only extraction cycle. The influenceof the extraction pressure,p, temperature,T, time, t, and CO2flow, F, on fat removal was studied. The results are presented inFigs. 2–4.

3.1.1. Influence of extraction pressure and temperatureFig. 2 shows the influence of pressure and temperature on

fat removal – expressed as percentage of the initial fat con-to ryingp hent ts ffi-c hem y

experience denaturational transitions under higher temperatures[14].

The large error bars in the experimental data are a conse-quence of the heterogeneity of pigskin regarding fat content.

3.1.2. Influence of extraction timeThe extraction experiments presented inFig. 2 do not show

very high yields because the extraction time was not largeenough (60 min). Nevertheless, if time increases, the extractionyield increases up to the desired values as can be observed inFig. 3. The gradient of the first part of the graph (≈10−3) is ofthe same order of magnitude as that obtained by King et al.[15]for lard extraction under similar conditions. Lard is the mostsimilar product to the fat isolated from pig; therefore, the equi-

F ntent,vs

ent – for an extraction time of 60 min and a SC-CO2 flowf 10 kg/h. Fat removal remains about constant when varessure at 20◦C, and increases slightly with pressure w

he extraction is carried out at 40◦C. Carbon dioxide is noupercritical at 20◦C, but liquid, that seems to be not as eient as SC-CO2 at T = 40◦C, when pressure increases. Taximum temperature essayed was 40◦C since collagen ma

ig. 2. Fat removal, expressed as percentage of initial pigskin fat cos. extraction pressure at different temperatures.F = 10 kg CO2/h; particleize = 5 mm× 5 mm; t = 60 min.

E.M. Vaquero et al. / J. of Supercritical Fluids 37 (2006) 142–150 145

Fig. 3. Fat removal vs. extraction time.T = 40◦C; p = 60 MPa;F = 10 kg CO2/h;particle size = 5 mm× 5 mm. The continuous line represents the extraction curveas predicted by the steady-state approximation of the extraction from solids(SSAES) described in Section3.4.

librium solubility could be the rate-limiting step of the processduring the first minutes, when the fat is very accessible to thesolvent.

It can be observed inFig. 3that the error bars decrease alongextraction time, what indicates that the fat content becomeshomogeneously low as fat extraction progresses.

3.1.3. Influence of solvent ratioFig. 4 presents the influence of the SC-CO2 flow on fat

removal for a given amount of skin. It may be observed thatfat removal is similar no matter the solvent flow used – thus, theamount of solvent used in 60 min – for a given mass of pigskinand under fixed operation conditions (p = 490 bar;T = 40◦C).That is, the amount of fat removed does not depend on theamount of CO2 used for extraction indicating that the equilib-rium solubility is not the extraction rate-limiting step duringthe main time of the experiments, although, as concluded fromFig. 3, it could be controlling during the first minutes.

Fig. 4 also shows that the extraction yield is about constantwhen changing the solvent rate – then, changing the Reynoldsnumber – through the particle bed, indicating that the external

Fp

Fig. 5. Fat removal vs. extraction pressure at different temperatures.F = 10 kg CO2/h; particle size = 20 cm× 2 cm;t = 60 min.

mass transport in the supercritical phase is neither the controllingstep.

3.2. One-cycle extractions over skin sheets

As indicated in the experimental section, cutting the skin insmall pieces, as in the set of experiments reported in Section3.1, was not recommended if the traditional procedure to obtainthe collagen paste of interest is to be followed[7]. Therefore,the extraction experiments reported in this section were carriedout over larger skin pieces; specifically, pieces of around 20 cmlong× 2 cm wide were chosen since they were easily placedin the extractor to allow a well distributed solvent flow. Thisset of experiments was designed to study the influence of theextraction pressure, temperature and time on the extraction yieldby performing the extraction in one only cycle.

3.2.1. Influence of extraction pressure and temperatureThe influence of temperature and pressure on fat removal is

presented inFig. 5. Fat removal was about constant around 35%when extractions were carried out at 20◦C and increased withpressure at 40◦C; however, a yield lower than 50% was obtainedin all cases.

3.2.2. Influence of extraction timethe

i ed inF tiont hisl ovese mis,t longti g thet akest f thep

rriedo mod-u gh

ig. 4. Fat removal vs. SC-CO2 flow (F). p = 490 bar;T = 40◦C; t = 60 min;article size = 5 mm× 5 mm.

Some additional experiments were carried out to studynfluence of the extraction time. The results are presentig. 6where it can be observed that even fairly large extrac

imes did not allow a limit of 60% yield to be overcame. Timit is supposed to be reached because the solvent only remasily the fat of the cells mechanically broken in the hypoder

hat is easily accessible to the solvent, but fat is located all ahe pigskin, in the epidermis, dermis and hypodermis[1]. Thats, the fat has to exit the cells and be transported all alonhickness of the skin before entering the fluid phase, what mhe internal mass transport the step that limits the rate orocess.

This assumption was confirmed by some experiments caut by using the pigskin as a membrane in a flat membranelus, where SC-CO2 was fed to evaluate its permeation throu

146 E.M. Vaquero et al. / J. of Supercritical Fluids 37 (2006) 142–150

Fig. 6. Fat removal vs. extraction time.T = 40◦C; p = 65 MPa;F = 10 kg CO2/h;particle size = 20 cm× 2 cm.

such a membrane. The skin was placed in the modulus with theepidermis or the hypodermis facing the SC-CO2 flow. Theseexperiments showed that the stratum corneum, that is the mostexternal layer of the epidermis, is a barrier for the SC-CO2, asit is for substances that could enter the animal in vivo from theexterior, since it is the first protection of the animal from theexternal agents. Therefore, the extraction of the fat contained inpigskin sheets only takes place through one of the faces of theskin, the hypodermis face. That is, the fat of the hypodermis canbe removed easier by the solvent than the fat from the dermis othe epidermis.

3.2.3. Influence of pigskin initial water contentThe influence of the skin water content on the fat extraction

yield is presented inFig. 7. The water content of the pigskinwas reduced about 50% of its initial content in a drier unit atroom temperature to avoid protein damage. The fat extractionyield obtained when using dried skin did not show a noticeableincrease compared to non-dried skin, and it was also limited to aapproximate value of 50% (seeFig. 7). SFE over dried productsusually results in larger efficiencies than SFE over products thacontain water. However, when dealing with natural products,where the solutes often have to cross cell membranes to reac

F skinT

the fluid phase, if there is not enough water in the system, thelipid pillars of the membranes may close their pores makingthem impermeable. If enough water is available, the membranesbecome permeable[9].

The study so far, proved that SC-CO2 removed the fat frompigskin, but either particle reduction, or any other procedureto improve the internal mass transport, is necessary in order toobtain the desired extraction yields. An increase of the extrac-tion temperature was not advisable and a reduction of the watercontent or an increase of the extraction pressure or extractiontime did not produce the required efficiency for large particles;therefore, if particle reduction wanted to be avoided, a differentextraction procedure needed to be assayed.

Some experiments carried out with skin sheets of differentsizes, showed that this parameter did not have an influence onthe extraction yield. Only if the skin was minced, the extractionyield was affected. Therefore, the next set of experiments wasperformed by placing one only piece of skin in the extractorabout 20 cm long× 20 cm wide.

3.3. Extractions performed in several cycles over skinsheets

Some experiments were designed to follow the course of theextraction by determining the remaining amount of fat in thepigskin at different extraction times. In these experiments, thee alledc inei

sultso t thee henw ande s canb att -COd laterdf rgere non-i

s seto s,N tionp n fatr

3ime

( rentc ilarf 00%w cana actiona in pre-v ithi rger

ig. 7. Fat removal vs. extraction pressure for partially dried and non-dried= 40◦C; F = 10 kg CO2/h; particle size = 20 cm× 2 cm;t = 60 min.

r

n

t

h

.

xtractor was depressurized after a fixed extraction time (cycle time,tc) in order to withdraw a pigskin sample to determts fat content.

The first observation made when analyzing the rebtained with this new experimental procedure was thaxtraction yield of 60%, that was the maximum obtained working without depressurizing the extractor, was overcameven total elimination of fat was achieved in some cases ae observed inFigs. 8–12. A plausible explanation could be th

he membranes of the fat cells were swollen when the SC2iffused into the cell during an extraction cycle, and wereestroyed or somehow modified during depressurization[16],

avoring the fat–solvent contact and, thus, leading to a lafficiency on the extraction process than in the case of

nterrupted extractions.The influence of two new variables was evaluated in thi

f experiments, i.e.: the cycle time,tc, and the number of cyclec. In addition to these variables, the influence of extracressure, solvent flow, and initial fat and water content, oemoval was evaluated.

.3.1. Influence of cycle time and number of cyclesFig. 8shows fat removal as a function of total extraction t

a) and as a function of the number of cycles (b) for diffeycle times.Fig. 8a shows that the tendencies were very simor all the experiments of this set. Yields between 90 and 1ere obtained in all cases after about 4 h of extraction. Itlso be observed that the error bars decrease as fat extrdvances, as was observed in the experiments reportedious sections.Fig. 8b shows that fat removal increases wncreasing cycle time up to a cycle time of 60 min but la

E.M. Vaquero et al. / J. of Supercritical Fluids 37 (2006) 142–150 147

Fig. 8. Fat removal against total extraction time (a) and as a function of the number of cycles (b) for different cycle time.T = 40◦C; p = 35 MPa;F = 10 kg CO2/h;skin sheet size = 20 cm× 20 cm.

cycle times gave rise to similar results. That is, the number ofintermediate depressurizations seems not to be very importantas long as there is at least one – for instance, after the extractionof the superficial fat, after about 60 min – that helps to modifythe membranes of the fat cells to improve the contact of internalfat and solvent. An extraction experiment made with one onlydepressurization confirmed that assumption.

A cycle time of 60 min was considered as the most conve-nient to follow the course of the extraction in the followingexperiments. Shorter cycle times required a larger number ofdepressurizations leading to a substantial loss of on-stream time.

Sampling in the experiments presented inFig. 8, was madeto follow the course of the extraction along the extractor axis;however, we did not found a clear axial concentration profile fora cycle time of 20 min or longer. The length of the error bars inthose experiments is, thus, not a consequence of the existence ofsuch a concentration profile, but is due to the heterogeneity ofthe skin. (ObserveFig. 12where the fat content of the skin hasbeen plotted versus the amount of CO2 used for extraction: theerror bars of non-extracted skin are similar or even larger thanthe error bars of data obtained after extraction.)

F ervalp cyclet

Fig. 10. Fat removal vs. extraction time for different extraction pressure. Extrac-tion parameters:T = 40◦C; F = 10 kg CO2/h; t = 180 min; cycle time = 90 min;particle size = 20 cm× 20 cm. The lines represent the extraction curves as pre-dicted by the model described in Section3.4.

The extraction rate curves of some of the experiments pre-sented inFig. 8have been plotted inFig. 9 in order to evaluatethe different transport resistances. A constant extraction rate canbe observed during the very first part of the extraction for the

Fig. 11. Fat removal as a function of extraction time for different solvent flow.T = 40◦C;p = 35 MPa; skin sheet size = 20 cm× 20 cm; cycle time = 60 min. Thelines represent the extraction curves as predicted by the model described inSection3.4.

ig. 9. Extraction rate curve. Amount of fat extracted in a certain time intlotted against time of extraction for experiments carried out at different

ime.T = 40◦C; p = 35 MPa;F = 10 kg CO2/h; skin sheet size = 20 cm× 20 cm.

148 E.M. Vaquero et al. / J. of Supercritical Fluids 37 (2006) 142–150

Fig. 12. Fat content – expressed as fat/protein weight ratio – as a function of thetotal extraction time for different initial skin fat content.T = 40◦C; p = 35 MPa;F = 10 kg CO2/h; particle size = 20 cm× 20 cm;tc = 60 min. The lines representthe extraction curves as predicted by the model described in Section3.4.

experiment carried out attc = 10 min, indicating that mass trans-fer is constant and determined by the transition at the interfacebetween the solid and the fluid. In the second part of the extrac-tion, the internal mass transport becomes the main resistance.For the rest of the experiments, carried out at largertc, there areno data at an extraction time low enough to observe the part ofthe graph where the extraction rate is constant, although it isassumed to be constant during the first minutes.

3.3.2. Influence of extraction pressureFig. 10presents fat removal as a function of pressure. It may

be observed that the extraction curves obtained at 35 and 55 MPare not very different. Fat solubility in SC-CO2 increases withpressure at constant temperature, but, in general, the fat solbility isotherm at 40◦C does not present a large influence ofpressure in the range of pressures studied here[17] and, in anycase, the solubility is only limiting the rate of the process duringthe first extraction minutes, that is when this effect could havebeen observed if data had been taken at low extraction timeDuring the main time of extraction, the internal mass trans-port has been shown to be the resistance controlling the ratof the process, and to improve such internal mass transpora pressure decrease would be more convenient since it woucause a viscosity decrease and a diffusivity increase. Taking intaccount these considerations, an extraction pressure of 35 MPw

3t of

s rrieo cyt ue tt tiont than1 byt ntfl eo that

the external diffusion of fat in the fluid phase can be eliminatedas a significant resistance to the global mass transfer just byincreasing the solvent flow above 10 kg CO2/h.

3.3.4. Influence of initial fat content of pigskinFig. 12 presents the amount of fat in the skin, in a protein

bases (protein is the only non-extracted component in pigskin),as a function of the amount of solvent used for extraction offat from skin sheets with different initial fat content. It may beobserved that the amount of fat in the skin, after flowing around100 g CO2/g protein through it, is still related to the initial fatcontent indicating that the equilibrium solubility may be limitingthe rate of the extraction in the first part of the process. After500 g CO2/g protein have passed through the extractor, all thesamples have a similar fat content no matter what the initialcontent was.

The evolution observed inFig. 12is probably a consequenceof the skin structure. When the initial fat content of the pigskinwas high, it was because the hypodermis layer was thick. Thefat content of the skin increases from the epidermis to the hypo-dermis while the cell compactness decreases[18]; therefore, thefat of the hypodermis should be easier to extract than the fatof the epidermis because the internal mass transport should befaster through the non-compacted cells. Once the hypodermisfat has been eliminated, the fat of the epidermis has a long patht fluidp ble tot er ctionr g thefi

3ied

s sam-p ionc nt ofw ence

F iffer-es

as selected for the rest of the experiments of this set.

.3.3. Influence of solvent ratioFig. 11 shows fat removal as a function of the amoun

olvent used. The extraction curve for the experiments caut at a flow of 5 kg CO2/h lead to a lower extraction efficien

han the rest of the experiments. This difference could be dhe saturation of the solvent when the flow is low – saturahat would not take place when the solvent flow is larger0 kg CO2/h – or to a limitation of the rate of the process

he external mass transport.Fig. 11shows that when the solveow is equal or superior to 10 kg CO2/h, there is no influencf the solvent rate on the extraction yield. This indicates

a

u-

.

et,ldoa

d

o

hrough the dermis and hypodermis to reach the bulk of thehase, since the stratum corneum resulted to be impermea

he SC-CO2 as mentioned in Section3.2.2. This is probably theason why the internal mass transport controls the extraate at the end of the process but it is not so important durinrst extraction minutes.

.3.5. Influence of initial water content of pigskinIn Section3.2.3, this influence was evaluated by using dr

kin. In this case, we have used some natural pigskinles with different initial water content. The fat extracturves were similar in all cases indicating that the amouater contained in natural skin has not a noticeable influ

ig. 13. Water removal as a function of the amount of solvent for dnt initial skin water content.T = 40◦C; p = 35 MPa;F = 10 kg CO2/h; particleize = 20 cm× 20 cm;tc = 60 min.

E.M. Vaquero et al. / J. of Supercritical Fluids 37 (2006) 142–150 149

Table 1Mass transfer coefficients obtained from experimental data modeling

Extraction parameters Results of the model

Pressure (bar) Temperature (◦C) Solvent flow (kg CO2/h) Cycle time (min) K (kg m/min) S.D.a

350 40 10 10 1.3× 10−5 5× 10−2

350 40 10 30 2.1× 10−5 9× 10−2

350 40 10 40 9.9× 10−6 3× 10−2

350 40 10 60 1.2× 10−6 15× 10−2

350 40 5 60 6.8× 10−6 3× 10−2

350 40 15 60 1.1× 10−5 4× 10−2

350 40 20 60 1.8× 10−5 2× 10−2

220 40 10 90 7.2× 10−6 5× 10−2

350 40 10 90 1.0× 10−5 7× 10−2

550 40 10 90 9.2× 10−6 7× 10−2

a Standard deviation between the experimental and calculated values of the mean concentration of fat in pigskin at a timet (cm) are given.

on fat extraction. The water extraction curves are presentedin Fig. 13 in this case, since a co-extraction of water tookplace in all the experiments reported in this work. It can beobserved that the water extraction curves tend to reach a limit,and were similar in all cases. The skin recovered the initialaspect and properties, regarding the elaboration of the colla-gen paste of interest, when soaked in water after fat- and water-extraction.

3.4. Data analysis and modeling

The experimental data previously reported have been mod-eled to obtain the total mass transfer coefficients according tothe steady-state approximation of the extraction from solids pro-posed by Bruner[9]. The process here studied admits most ofthe simplifications of this model. The integration of the differen-tial equation resulting from the mass balances, the equilibriumrelations, and the extraction kinetics, assuming the total masstransfer coefficientK to be constant, provides the equation towhich experimental data were fitted:

cm − cf

c0 − cf≈ exp

(−KA

mst

)(1)

wherems is mass of pig skin in the extractor,c0 the initial meanconcentration of fat in pigskin,cm the mean concentration of fati oft erc

r thei eret nr of the te ot

n inT t andt lateva

s oft cyc

time or solvent flow used, what could be expected since these twoparameters did not have a very strong influence on the extractionyield.

Some of the extraction curves as predicted by the model havebeen plotted with lines in some of the graphs along the text.It can be observed that in spite of being a very simple model,it can represent the extraction curves within the experimentalerror.

Acknowledgements

The authors gratefully acknowledge the Ministry of Scienceand Technology for financial support through grant PPQ2001-1024 and Campofrıo Alimentacion S.A. (Spain) for supplyingthe pigskin required for this work.

References

[1] E.G. Cleary, Skin, in: W.D. Comper (Ed.), Extracellular Matrix, vol. 1,Harwood Academic Publishers GmBH, Amsterdam, 1996, pp. 77–109.

[2] C.L. Jenkins, R.T. Raines, Insights on the conformational stability ofcollagen, Nat. Prod. Rep. 19 (2002) 49–59.

[3] T. Bamberger, J.C. Erickson, C.L. Cooney, S.K. Kumar, Measurementand model prediction of solubilities of pure fatty acids, pure triglyc-erides, and mixtures of triglycerides in supercritical carbon dioxide, J.Chem. Eng. Data (30) (1988) 327–333.

romPro-, pp.

reat-ing a

Fluids

e

im.

per-inger,

[ t and

n the pigskin at a timet, cf the concentration of fat in the bulkhe fluid,A the mass transfer area andK is the total mass transfoefficient.

Eq. (1) is able to represent the extraction process aftenitial extraction period. The first part of the extraction, whhe equilibrium solubility is likely to be limiting the extractioate, is not expected to be represented as well as the restxtraction where the internal mass transport controls the rahe process.

The results of the experimental data modeling are showable 1where the values of the total mass transfer coefficienhe standard deviation between the experimental and calcualues of the mean concentration of fat in pigskin at a timet (cm)re given.

Table 1shows that the total mass transfer coefficient ihe same order of magnitude in all cases, regardless the

ef

d

le

[4] E. Lack, B. Simandi, Supercritical fluid extraction and fractionation fsolid materials, in: A. Bertuco, G. Vetter (Eds.), High Pressurecess Technology: Fundamentals and Applications, Elsevier, 2001537–575.

[5] G. Gavend, B. Vulliermet, C. Perre, M. Carles, Process for the tment of skins, leathers or collagen-containing sheet materials uspressurized dense fluid, World Patent WO9408054 (1994).

[6] A. Marsal, P.J. Celma, J. Cot, M. Cequier, Supercritical CO2 extractionas a clean degreasing process in the leather industry, J. Supercrit.16 (2000) 217–223.

[7] M.A. Checa-Andres, Procedimiento de obtencion de una pasta dcolageno, Spanish Patent ES2017564 (1991).

[8] M. Mowafy, R.G. Cassens, Microscopic structure of pig skin, J. AnSci. 41 (1975) 1281–1290.

[9] G. Bruner, Gas Extraction: An Introduction to Fundamentals of Sucritical Fluids and the Application to Separation Processes, SprNY, USA, 1994.

10] International Organization for Standardization: ISO 1442:1997, MeaMeat Products—Determination of Moisture Content.

150 E.M. Vaquero et al. / J. of Supercritical Fluids 37 (2006) 142–150

[11] International Organization for Standardization: ISO 1443:1973, Meat andMeat Products—Determination of Total Fat Content.

[12] International Organization for Standardization: ISO 937:1978, Meat andMeat Products—Determination of Nitrogen Content.

[13] O. Ganado, E.M. Vaquero, R. Murga, S. Beltran, J.L. Cabezas, Designof a SCFE pilot plant to be operated up to 70 MPa and 475 K, and someapplications, in: A. Bertuco (Ed.), Proceedings of the 4th InternationalSymposium on High Pressure Technology and Chemical Engineering,vol. I, Associazione Italiana Di Ingegneria Chimica, Chem. Eng. Trans.2 (2002) 205–210.

[14] P.L. Privalov, E.I. Tiktopulo, V.M. Tischenko, Stability and mobility ofthe collagen structure, J. Mol. Boil. 127 (1979) 203–216.

[15] J.W. King, J.H. Johnson, J.P. Friedrich, Extraction of fat tissues formmeat products with supercritical carbon dioxide, J. Agric. Food Chem.37 (1989) 951–954.

[16] C. Perre, A.L. Hans, M. Dedieu, O. Gand, L. Saldinari, L. Dutel,From skin to leather in dense pressurized CO2, in: G. Brunner, I.Kikic, M. Perrut (Eds.), Proceedings of the 6th International Sympo-sium on Supercritical Fluids, Tome 1, ISASF, Versailles, 2003, pp. 419–424.

[17] J.P. Friedrich, Supercritical CO2 extraction of lipids from lipid-containing materials, US Patent 4,466,923 (1984).

[18] A.W. Ham, Histology, sixth ed., J.B. Lippincott Company, Philadelphia,1969.