- Email: [email protected]
Membrane-based simultaneous degumming and deacidification of vegetable oils
Innovative Food Science and Emerging Technologies 6 (2005) 203 – 212 www.elsevier.com/locate/ifset
Membrane-based simultaneous degumming and deacidification of vegetable oils Abdellatif Hafidia,*, Daniel Piochb, Hamid Ajanaa a Laboratoire Sciences des Aliments, Universite´ Cadi Ayyad, Faculte´ des Sciences Semlalia, BP 2390, Marrakech, Morocco Physico-Chemistry of Processes and Bioenergy Laboratory, Agrifood Systems Programme CIRAD-AMIS, TA 40/16, 34398 Montpellier Cedex 5, France
b
Received 20 January 2004; accepted 16 December 2004
Abstract An efficient membrane based process for simultaneous degumming and deacidification of vegetable oil was investigated. Appropriate crude oil conditioning allow the formation of submicronic aggregates, composed with soaps molecules resulting from the neutralisation of FFA and PL, which are retained when microfiltrating. Initial flux for the 0.8 Am membranes (~560 l/h m2) was about twice that of the 0.5 Am and about 10 times that of 0.2 Am membrane. The filtered oils showed good quality in the case of 0.2 and 0.5 Am membranes, but the use of 0.8 Am membranes has allowed some soaps to pass through. Two types of crude oils behaviour were noticed. Oppositely to some oils for which just simple neutralisation led to a satisfactory elimination of the phospholipids, others were very tough to refine. The operating pression seems not to affect the efficiency of the separation, whereas the stability of the vesicle-like aggregates is found to be greatly affected by the increase of the temperature above 25 8C. Beside the quasi-elimination of FFA, PL and water, minerals and pigments contents were also greatly lowered. When using an NaOH 20%, the lovibond yellow score lowered from around 28 to 10 in the case of sunflower oils and from ~34 to 6–20 in the case of soya and rapeseed oils. The monoglycerides were almost undetectable after membrane processing whatever the type of the conditioning used. After processing, the diglycerides contents, which ranges in the tested crude oils between 0.8% and 1.0%, showed almost no changes for two oils, whereas noticeable increases were obtained for the other oils. Total phytosterols contents were systematically reduced. The reductions vary from 3% to 44% upon the case. Neutralisation with an NaOH 20% lead to higher sterol losses in comparison to NaOH 40%. All the sterol components contents were found to be reduced in almost similar proportions. D 2005 Elsevier Ltd. All rights reserved. Keywords: Vegetable oils; Degumming; Deacidification; Microfiltration; Minor components; Quality Industrial relevance: Conventional oil recovery and purification processes are continuousely sought to be replaced by gentler processing conditions. Membrane based oil refining operating at low temperatures and without the generation of waste water offer an interesting and promising approach towards bgreenerQ technologies.
1. Introduction Excepting extra virgin olive oils and some speciality oils, vegetable oils must undergo refining operations to remove undesirable components granting them thus satisfactory purity and stability characteristics (acidity, colour, oxidative, and sensory). Classical edible oil refining processes include degumming, neutralisation, bleaching, and deodorisation. * Corresponding author. Tel.: +212 44 43 46 49; fax: +212 44 43 74 12. E-mail address: [email protected] (A. Hafidi). 1466-8564/$ - see front matter D 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ifset.2004.12.001
These series of operations aim mainly to remove phospholipids (PL), free fatty acids (FFA), pigments, hydroperoxides, and waxes. Such a processing can be operated according to two main schemes differing essentially in the manner the FFA are removed. In the so called chemical or alkali refining, they are turned to soaps by an alkali and, subsequently, removed from the neutral oil by centrifugations and washing steps, whereas in physical refining they are distilled during the deodorisation operation. The two processes are reputed to be very energy-consuming. In addition, all the refining operations subject oils to high temperatures; consequently, quantitative and qualitative
204
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
changes inevitably take place especially in components responsible of quality. High temperatures seem to be the most harmful parameter, which is responsible of hydrolytic, oxidative, and polymerisation alterations (Cert, Lanzon, Carelli, Albi, & Amelotti, 1994; Schulte, 1995). Structural changes are also related to the heating especially during bleaching (Schulte, 1995) and deodorisation (Cert et al., 1994). Devinant, Scamaroni, and Naudet (1980) and Jawad, Kocchar, and Hudson (1983) reported that, during physical refining, the stereoisomerisation and polymerisation depend on the time and the temperature. Some sterols can undergo a dehydration during bleaching (Schulte, 1995) and deodorisation (Cert et al., 1994) so h sitosterol can form the 3,5 stigmadiene. Total sterols are also reported to lower after neutralisation (Kochhar, 1983). Pasqualone and Catalano (2000) reported that neutralised oils showed losses up to 50% in total sterols. Catalano, De Leonardis, and Comes (1994) and Gomes (1992) have reported an increase of diglyceride amounts in refined olive oil. Most of these alterations are temperature-dependent. Improving the quality of the processed oils by reducing side effects and diminishing the energy consumption has been among the most active research areas throughout the past half-century. Due to its characteristics as non-polluting and energy saving technology, membrane processing has recently emerged as a promising alternative. Potential applications covering various aspects such as solvent removing (Bhanushali, Kloos, & Bhattacharyya, 2002; Koseoglu, Lawhon, & Lusas, 1990; White & Nitsch, 2000), edible oil degumming (Kim, Jong-Ho Kim, Lee, & Tak, 2002; Koseoglu, Rhee, & Lusas, 1990; Subramanian & Nakajima, 1997), deacidification and free fatty acids recovering (Kale, Katikaneni, & Cheryan, 1999; Raman, Cheryan, & Rajagopalan, 1996; Zwijnenberg, Krosse, Ebert, Peinemeann, & Cuperus, 1999), dewaxing (De, Das, Dutta, & Bhattacharyya, 1998), and triacylglycerols partitioning (Bornaz, Fanni, & Parmentier, 1995a, 1995b) were reported. In the very first works from the last century, Sen Gupta, a pioneer, demonstrated the efficiency of degumming the oil miscellas using ultrafiltration techniques (Gupta, 1977, 1986). Ajana, Pioch, and Graille (1993) demonstrated the possibility of a complete degumming and deacidification of crude vegetable oils subsequently to appropriate conditioning using membrane separation techniques. The neutralisation of the free fatty acids with appropriate soda concentration and under adequate conditions allow the formation of submicronic particles with onion-like structures as revealed by scanning electron microscopy (Largueze, Pioch, & Gulik-Kryzwicki, 2002). These structures are believed to be formed not only by fatty acid salts but also with phospholipids and water molecules. Due to their relatively important sizes, these aggregates can be easily removed by a microfiltration technique. Such structures are thought to be formed of hundreds of fatty acid salts and phospholipid bilayer stacks. Both oil conditioning and filtration can be operated at low temperature (20–25 8C).
These mild conditions which contrast with the conditions in industrial practices are expected to preserve the sensitive and bioactive components in edible oils. In the present paper, we report an assessment of the efficiency of such a processing by testing different crude oils and comparing their suitability for our membranebased purification process. The effects of some physicochemical parameters on the stability of the aggregates and therefore on the efficiency of the separation were studied. The impact of our simultaneous degumming and neutralisation process on some minor components of the oils was also investigated.
2. Materials and methods 2.1. Oil samples Soya, sunflower, and rapeseed crude oils were obtained from CEREOL, se`te, France. The crude oils were neutralised at room temperature (20–25 8C) (unless otherwise stated) by pouring slowly with a pipette an aqueous sodium hydroxide solution (20% or 40% w/v) under magnetic stirring (600 rpm) and then filtered as described in previous works (Hafidi, Pioch, & Ajana, 2004; Pioch, Largueze, Graille, Ajana, & Rouviere, 1998; Pioch et al., 1996). In the cases of acid conditioning, 0.1% or 0.3% (w/w) of the acid phosphoric 85% (Sigma) were added at room temperature (20–25 8C) to the crude oils under vigorous mixing during 20 min before neutralisation of the whole acidity with an appropriate NaOH solution. 2.2. Microfiltrations Dead end filtration experiments were performed with a stainless steel Gelman module (200 ml volume, 200 kPa pressure; Whatman cellulose filter (pore size: 2.5 Am, filtration area 16 cm2)). Crossflow filtration experiments were carried out at 25 8C with a laboratory stainless steel apparatus (1 kg oil samples, tubular alumina membrane 200 mm length, 40 cm2 area) under: continuous permeate recycling, 3.5 m/s as a tangential velocity and 200 kPa transmembrane pressure. Fluxes were estimated by direct measurements using an appropriate graduated cylinder. 2.3. Analytical methods In order to check the chemical composition of both starting and refined oils, French standard procedures (AFNOR, 1984) were applied: phosphorus NF T 60-227, soaps NF T 60-217, free fatty acids NF T 60-204, water NF T 60-367, and peroxide value NF T 60-220. Cations (Na, Ca, Mg, Fe, Cu) were analyzed after mineralisation following the ISO 6884: 1985 (ISO standards methods, 1985) standard method, comprising mineralisation and atomic emission spectrophotometry analysis (inductive
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
coupled plasma, ICP) with a Jobin Yvon JY spectrophotometer. The results are the average of two independent trials (both mineralisation and ICP analysis). The composition of sterols were determined according to the French standard method NF T 60-258 (AFNOR, 1984). After extraction and separation of the unsaponifiable fraction of the oil by thin layer chromatography, the sterols were silylated and analyzed using an SE-54 (Supelco) 30 m; 0.25 internal diameter capillary column under the following conditions: the oven initial temperature (180 8C) was kept for 8 min before starting a temperature program: 5 C/min until 260 8C which is maintained for 15 min; injector and detector temperatures were 280 and 290 8C, respectively. The a cholestane (sigma) was used as an internal standard. Partial glycerides were analyzed by gas chromatography according to the method described by Plank and Lorbeer (1992). A DB1 WTHN capillary column was used (7 m length, 0.32 mm internal diameter). The carrier gas (helium) flow rate was set at 2.9 ml/min; the temperature was held at 100 8C for 3 min before a rise of 10 8C/min until 340 8C, which is maintained for 5 min. Hexatriacontane was used as an internal standard. The silylation was performed with the bis(trimethylsilyl) trifluoroacetamide (BSTFA) and the trimethylchlosilane (TMCS). 2.4. Lovibond color A lovibond tintometer was used and the colour of oil was matched with a set of standard coloured numbered glasses, ranging in the scales from 0 to 70 red (R) and 0 to 70 yellow (Y). The results are expressed as R and Y values.
3. Results 3.1. Membrane separation efficiency Numerous crude oils were tested but only five samples, two sunflower oils (SFO1 and SFO2), two soya oils (SO1 and SO2), and a rapeseed oil (RO), which summarise the most featured behaviours, will be presented and discussed. The main aim of our membrane-based refining process is to produce in a single step a suitable oil for bleaching. Thus, FFA, phospholipids, soaps, and water must be removed almost completely. Only oils containing less than 10 ppm phosphorus, 30 ppm soaps, and less than 0.1% water are considered as appropriate for bleaching (Denise, 1983). Results depicted in Table 1 recapitulate the main characteristics of the crude oils and revealed acceptable quality of the processed oils. Crude oils FFA acid contents range between 0.5% and 1.5% with soya oils having the least content and rapeseed oil the highest. Phospholipids expressed in terms of phosphorus content vary from 43 to 140 ppm. These relatively low PL contents confirm that these crude vegetable oils are actually water degummed. The filtration of the conditioned crude oils removes almost all the PL,
205
soaps resulting from the neutralisation of FFA, and water. Excepting the case of the SO2, the filtration of the crude oils subsequently to a simple neutralisation (without a prior phosphoric acid pre-treatment) has led in almost cases to a quasi elimination of the initial acidity and also to a great reduction of the phospholipids content. The filtered oils are also almost free from soap molecules resulting from the alkali treatment and from the water added via the sodium hydroxide solutions. PL retentions vary from 74% to 98%; consequently, the phosphorus content did not exceed 10 ppm in most cases. Despite the very high soaps retention rates obtained for the simply neutralised and filtered oils, their quality can be considered satisfactory only when the soaps molecules were almost undetectable in the filtered oils. Simple neutralisations has led to an excellent oil quality (i.e. oils which can be directly bleached without any additional purification) only in the case of SFO2 whatever the NaOH concentration and in the cases of SO1 and RO when using an NaOH 20%. In fact, soaps retentions seem to be higher in the case of neutralisation with NaOH 20%. The use of 0.1% phosphoric acid pre-treatment prior to the neutralisation of the FFA resulted in an improvement of the retention of phospholipids and soaps especially in the case of SFO1 with both NaOH concentrations and only in the case of an NaOH 40% for the SO1 and RO. In these two latter cases, the soaps retention was still not complete and few amounts of soaps molecules were detected in the filtered oils (Table 1). The SO2 showed only very slight retention of phospholipids in the case of an NaOH 20%. The increase of the amount of the phosphoric acid pre-treatment to 0.3% allowed better retentions of phospholipids and soaps for all the oils. Even the SO2 showed an acceptable retention after a 0.3% H3PO4 pre-treatment when neutralised with an NaOH 20%. Beside the elimination of FFA, PL, and water, the processed oils showed lower amounts of pigments as revealed by the lovibond colour estimation. When using an NaOH 20%, the yellow score lowered from around 28 to 10 in the case of SFO1 and SFO2 and from ~34 to values ranging from 16 to 20 for the SO1 and RO. Slightly higher values were obtained when using an NaOH 40%. Similar decrease was observed in the red. The phosphoric acid pretreatment seems to allow slightly lower values. The SO2 is definitely an exception, just a very slight diminishing of the colour is observed. Minerals are known to be pro-oxidants and must preferably be eliminated from edible oils to guarantee good oxidative stability. As the membrane processing of the crude oils succeed in greatly lowering the iron, calcium, magnesium, and sodium cations especially when the acid phosphoric is used (Table 2). In spite of its straightforwardness, dead end filtration seems not to fit industrial applications because of very fast and severe fouling. In crossflow filtrations, the tangential flow sweeping continuously the membrane surface limits the deposition and the formation of a cake layer and therefore
206
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
Table 1 Physico-chemical characteristics of the crude and membrane processed oilsa [NaOH]
SFO1
Crude Processed
20%
40%
SFO2
Crude Processed
20%
40%
SO1
Crude Processed
20%
40%
SO2
Crude Processed
20%
40%
RO
Crude Processed
20%
40%
a
[H3PO4]
– 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3% 0% 0.1% 0.3% 0% 0.1% 0.3%
FFA (%)
Phosphorus (ppm)
Soaps (ppm)
Water content (%)
Colour Y
R
0.97 0.06 0.06 0.05 0.08 0.08 0.06 1.00 0.04 0.02 0.02 0.05 0.06 0.04 0.73 0.04 0.06 0.04 0.06 0.04 0.06 0.49 0.04 0.04 0.04 0.06 0.04 0.06 1.55 0.04 0.02 0.02 0.06 0.04 0.04
43 9.0 0.8 1.5 11.0 2.2 5.5 67 1.2 0.6 0.8 4.6 2.7 0.3 96 6.8 4.1 3.3 11.2 7.6 2.0 141 155 9.1 7.4 141 231 128 123 12.0 3.0 1.8 19 5.4 7.6
– 144 25 nd 98 nd nd – nd nd nd 24 nd nd – nd 38 nd 244 70 nd – 840 786 nd 740 1725 816 – 34 nd nd 416 230 nd
0.05 0.08 0.08 0.06 0.10 0.07 0.06 0.08 0.06 0.04 0.05 0.05 0.08 0.06 0.17 0.03 0.09 0.05 0.06 0.05 0.05 0.13 0.11 0.07 0.08 0.11 0.11 0.07 0.09 0.07 0.03 0.05 0.06 0.05 0.04
26 10 10 10 15 14 11 28 12 10 10 14 12 12 34 20 16 16 22 18 17 37 34 33 34 34 34 34 36 22 20 20 26 24 24
2.0 1.5 1.2 1.2 1.2 1.2 1.2 2.2 1.4 1.0 1.0 1.2 1.2 1.2 2.8 2.0 1.5 1.6 2.1 1.8 1.8 2.5 2.2 2.0 2.2 2.2 2.2 2.2 3.0 2.4 1.8 1.8 2.6 2.0 2.0
Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa pressure; 25 8C.
allows better fluxes. As shown in Fig. 1, three different pore sizes (0.8, 0.5, and 0.2 Am) were tested. The initial flux for the 0.8 Am membranes (~560 l/h m2) is about twice that of the 0.5Am and about 10 times that of 0.2 Am membrane. After 30 min of crossflow filtration, the fluxes are reduced to more than the half. After 2 h, a relative stabilisation of the fluxes is observed around 40, 30, and 15 l/h m2 for the 0.8, 0.5, and 0.2 Am membranes, respectively. Nevertheless, the quality of the filtered oils was not always acceptable. Relatively large amounts of soap molecules are found in the permeate when 0.8 Am pore size membrane is used (Table 3). This seems contradictory with the dead end filtration results where larger pore sizes (2.5 Am) allowed satisfactory retentions. On one hand, the cake formed in early instant of the dead end filtration which most likely play the role of the actual membrane and, on the other hand, the shear caused by pumping and circulation of the conditioned oil during crossflow filtration which probably led to mechanical break of the aggregates may explain these results.
3.2. Influence of the operating temperature To evaluate the effect of the operating temperature on the efficiency of our membrane separation process, crude oils were neutralised and filtered at temperatures raging between 25 and 80 8C. All the oils showed a similar behaviour: a decrease of retention of phospholipids and soaps was observed as the operating temperature was elevated. Results concerning the SFO2 are depicted in Table 4. As a result of the neutralisation, the FFA content is greatly reduced in the filtered oils in most cases and visibly has not been affected with temperature. However, at 80 8C, this reduction was less satisfactory; an undesirable hydrolysis of the triacylglycerols generating additional amounts of FFA may be the cause. At 25 8C, the phospholipids retention reached 84% to 98%, whereas it was only of 46% to 78% at 60 8C. Increasing amounts of soap molecules permeate across the membrane as the temperature was increased. Nevertheless, it seems that
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212 Table 2 Minerals contents in crude and membrane processed oilsa [NaOH] SFO1
Crude Processed
20%
40%
SO2
Crude Processed
20%
40%
RO
Crude Processed
20%
40%
207
Table 3 Quality characteristics of the crossflow filtered SFO1a
[H3PO4]
Fe (ppm)
Ca (ppm)
Mg (ppm)
Na (ppm)
– 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3%
4.6 0.3 0.1 0.1 0.1 0.1 0.1 7.0 0.7 0.2 0.1 1.3 0.3 0.3 5.8 1.2 0.6 0.8 2.9 0.4 0.6
76.0 1.5 0.7 1.2 11.5 1.3 1.8 77.0 47.0 15.0 11.0 70.0 66.5 0.9 53.0 23.0 7.0 3.7 36.0 6.8 7.7
39.0 0.4 0.7 0.4 4.0 2.5 3.3 39.0 25.0 6.9 3.2 35.0 12.6 0.1 58.0 27.5 10.0 6.9 13.7 5.0 3.6
15.0 6.5 4.2 2.8 5.7 6.0 2.1 4.3 37.0 5.3 2.6 61.5 137.0 3.8 21.0 12.0 7.2 4.8 17.0 15.3 3.5
a Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure; 25 8C.
phosphoric acid pre-treatments improve the retention of the aggregates. Indeed, a noticeable difference was observed and a relatively acceptable retention rates were obtained until 40 8C in these cases.
SFO 1 0.8 Am 0.5 Am 0.2 Am
FFA (%)
P (ppm)
Soaps (ppm)
Water (%)
0.97 0.06 0.04 0.04
43.3 10.6 2.6 0.5
– 1586 nd nd
0.046 0.128 0.042 0.034
a Crossflow microfiltration conditions: alumina membrane, continuous permeate recycling, 3.5 m/s tangential velocity, and 200 kPa transmembrane pressure.
structures. Excepting the case of the SO2 and as already shown (Table 1), the operating pressure (200 kPa) seems not to be destructive for the structures of the aggregates since good retentions were obtained. Thus, we have examined the influence of the transmembrane pressure on the efficiency of the retention to demonstrate its effects if any and try to explain the particular behaviour of SO2. Results in Table 5 show that the pressure in the range (50–200 kPa) seems not to have any noticeable effect on the retention of the aggregates. The permeation of soaps and phospholipids in the case of SO2 cannot be explained by a breakable behaviour of its aggregates. Nevertheless, a slight sensibility to the operating pressure, especially regarding the retention of phospholipids, was observed with the NaOH 20% simple neutralisations. In some cases, negative phosphorus retention values were obtained mainly when phosphoric acid pretreatment is used. This may be explained by a permeation of the phosphoric acid salts in the filtered oils, which thus contain bigger amounts of phosphorus than the crude ones.
3.3. Influence of the transmembrane pressure
Fluxes (l/h/m2)
As far as the aggregates formed in the crude oils subsequently to conditioning are assumed to be mainly formed by soaps, water, and phospholipids, they can be considered as a soft matter. Therefore, the driven pressure is expected to be a crucial parameter for the integrity of such
3.4. Effect of the conditioning on the partial glycerides contents The partial glycerides are natural components occurring in oils, they result from the triacylglycerols hydrolysis. Due to their surface activity, they can cause neural oil losses by
600
0.8 µm 0.5 µm 0.2 µm
500
400
300
200
100
0 0
50
100
150
200
250
300
Time (min) Fig. 1. Evolution of fluxes during microfiltration of the SFO1 conditioned with 0.3% phosphoric acid and neutralised with an NaOH 20%.
208
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
Table 4 Effect of the temperature on the efficiency of the separation (case of the SFO2)a [NaOH] [H3PO4] Temperature FFA Phosphorus Soaps Water (%, w/w) (8C) reduction retention retention content (%) (%) (%) (%) 20%
0
0.3
40%
0
0.3
25 40 60 80 25 40 60 80 25 40 60 80 25 40 60 80
96.0 94.0 95.0 92.0 96.0 94.0 90.0 84.0 93.0 94.0 80.0 88.0 96.0 96.0 91.0 69.0
98.1 82.9 55.1 49.0 92.9 87.9 78.6 22.7 83.7 74.2 46.2 3.0 95.6 98.2 68.1 6.7
100.0 99.9 98.5 99.1 100.0 100.0 99.9 98.9 99.9 99.8 99.5 98.9 100.0 99.9 99.8 99.0
0.03 0.07 0.05 0.02 0.02 0.03 0.07 0.08 0.04 0.06 0.04 0.03 0.04 0.04 0.05 0.06
a Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure.
emulsification during the refining operations. Excepting the SFO2, the monoglycerides were almost undetectable in all the tested crude oils and no increase was observed after processing whatever the type of the conditioning used (Table 6). Diglycerides content ranges in crude oils between 0.8% and 1.0% and seems to be somehow proportional to the FFA contents. After processing, the diglycerides contents showed almost no changes for two oils (SFO1 and SO1). Meanwhile, noticeable increases were obtained Table 5 Effect of the transmembrane pressure on the efficiency of the separation in the case of the SO2a [NaOH]
20%
[H3PO4] (%, w/w)
Pressure (kPa)
FFA reduction (%)
0
50 100 200 50 100 200 50 100 200 50 100 200 50 100 200 50 100 200
91.84 91.84 91.84 91.84 91.84 91.84 91.84 91.84 91.84 87.76 87.76 87.76 91.84 91.84 91.84 87.76 87.76 87.76
0.1
0.3
40%
0
0.1
0.3
P retention (%) 58.16 8.51 9.93 94.33 96.81 93.55 99.29 93.40 94.75 1.21 2.48 0.43 68.79 52.48 63.83 10.64 24.11 9.22
Soaps retention (%)
Water content (%)
98.80 98.97 98.60 98.96 98.84 98.69 100.0 99.95 100.0 99.07 98.75 98.77 98.80 98.23 97.13 98.75 98.93 98.64
0.078 0.161 0.112 0.078 0.088 0.074 0.063 0.056 0.082 0.101 0.116 0.110 0.078 0.161 0.112 0.068 0.088 0.067
a Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am; 25 8C.
for the other oils. These increases reached approximately 200% in the case of SO2 simply neutralised with an NaOH 20% but generally they vary from 30% to 60%. These variations seem not to be influenced neither by the NaOH nor the phosphoric acid concentrations. 3.5. Effect of the conditioning on the sterol composition Sterols are among the major components of the unsaponifiable fraction of the vegetable oils. Because of their beneficial effects on human health especially their action of blood cholesterol lowering, the phytosterols are desirable components in refined oils (Trautwein et al., 2003). Unfortunately, the processed oils (Tables 7 and 8) showed lower sterol contents comparatively to the crude ones. Total phytosterols reduction vary from 3% to 44% upon the case. Neutralisation with an NaOH 20% lead to higher sterol losses in comparison to NaOH 40%. Reductions of 17 to 23%, 25 to 44% and 22 to 35% were obtained when using an NaOH 20% for the SFO1, SO1, and RO, respectively, whereas they ranged from 3% to 18%, 21% to
Table 6 Effect of the processing on partial glycerides contentsa Samples SFO1
Crude Processed
[NaOH]
[H3PO4]
20%
0% 0.1% 0.3% 0% 0.1% 0.3%
40%
SFO2
SO1
Crude Processed
Crude Processed
20% 40%
0% 0% 0.1% 0.3%
20%
0% 0.1% 0.3% 0% 0.1% 0.3%
40%
SO2
Crude Processed
20% 40%
RO
Crude Processed
20%
40%
0% 0.1% 0% 0.1% 0.3% 0% 0.1% 0.3% 0% 0.1% 0.3%
Monoglycerides (%)
Diglycerides (%)
nd 0.01 0.01 0.01 0.01 0.01 nd 0.013 0.03 0.02 0.02 0.013 Tr. nd nd nd nd nd nd nd nd 0.02 0.012 nd nd nd 0.06 nd nd nd nd nd
0.84 0.75 0.84 0.71 0.96 0.74 0.44 1.05 1.67 1.57 1.62 1.29 0.78 0.76 0.72 0.66 0.82 0.70 0.71 0.67 1.96 0.97 1.37 1.17 1.00 1.01 1.40 1.42 1.50 1.40 1.25 1.10
a Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure; 25 8C.
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
209
Table 7 Effect of the processing on different sterol content (ppm) of the sunflower oil (SFO 1) and the soya oil (SO1)a SFO1 [NaOH]
SO1
Crude
Processed
–
20%
40%
Crude
Processed
–
20%
40%
[H3PO4]
–
0%
0.1%
0.3%
0%
0.1%
0.3%
–
0%
0.1%
0.3%
0%
0.1%
0.3%
Campesterol Stigmasterol D7 Campesterol Clerosterol h Sitosterol D5 Avenasterol D5,24 Stigmastanediol D7 Stigmastenol D7 Avenasterol Total
356 454 93 41 2641 39 37 524 100 4285
292 346 103 34 2072 25 nd 570 95 3537
281 323 105 33 1979 29 nd 561 107 3418
278 374 73 nd 2094 nd nd 375 nd 3299
400 459 111 39 2398 37 38 518 116 4117
292 360 90 27 2090 38 35 482 95 3511
324 398 108 34 2386 nd 36 576 76 3939
808 640 50 – 1859 50 – 43 – 3450
572 466 62 – 1410 47 – 30 – 2587
386 370 38 – 1076 28 – 36 – 1934
396 375 48 – 1106 26 – 23 – 1975
619 510 57 – 1463 47 – 32 – 2729
622 500 59 – 1367 44 – 33 – 2625
603 515 55 – 1358 44 – 20 – 2596
a
Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure; 25 8C.
25%, and 11% to 18% when using an NaOH 40%. In addition, these losses were more important in the cases of the phosphoric acid conditioning. As shown by the results depicted in Tables 7 and 8, all the sterol components contents were reduced in almost similar rates.
4. Discussion Demands for more environment-preserving technologies and more healthier foods are continuously growing especially in western countries. New attitudes emerge towards food processing and, consequently, more attention is paid to the influence of processing on the quality. The effects of classical refining operations on the quality and chemical composition of the oils were extensively studied. Vegetable oils are constituted predominantly of triglycerides; however, the presence of various classes of minor constituents in such oils is of a great importance. Identification and quantitation of such components, naturally present or generated by refining operations constitute an important mean either for characterisation as well for frauds detection purposes. Refining operations may eliminate, reduce, or alter the different components of the oils. PL must be removed from oils because of their emulsifying properties, which may complicate the subsequent operations, increase neutral oil losses, and may led to the formation of dark coloured
compounds during deodorisation (Denise, 1983). The FFA are also undesirable components, several works had outlined their prooxidant effects (Frega, Mozzon, & Lercker, 1999; Kiristakis & Tsipeli, 1992) contributing to the reduction of the oil shelf life. The alkali neutralisation lead to the transformation of the free fatty acids to soaps. These latter are known to be, like PL, very surface active components, which may self-assemble into wide variety of macromolecular aggregates. Our results showed quasi complete elimination of PL and FFA from the membrane-processed oils. As PL and soap molecules resulting from the alkali treatment are retained when microfiltrating, they must have somehow sizes beyond the membrane pore diameter (2.5 Am). In fact, both PL and soap molecules are known to be amphiphilic and consist of a polar headgroup and a nonpolar tails, which cause their preference for self assembly with the hydrophobic groups on one side and the hydrophilic groups on the other side. Surfactants are normally believed to exist as monomers up to a certain concentration above which they may self-assemble under specific conditions to form a wide range of organised macrostructures such as micelles, lamellar phases, liquid crystals, or crystals. Ajana et al. (1993) reported that conditioning the crude vegetable oils under specific condition lead to the formation of submicronic aggregates, which can be easily separated from the continuous glyceridic phase using a microfiltration technique. Using scanning electron microscopy, Largueze et al.
Table 8 Effect of the membrane processing on sterols content (ppm) of the rapeseed oila Crude
Processed
[NaOH]
–
20%
[H3PO4]
–
0%
0.1%
0.3%
0%
0.1%
0.3%
Brassicasterol Campesterol h Sitosterol Sitostanol Total
776 2578 3337 103 6795
715 1866 2628 76 5285
430 1715 2172 124 4441
555 2008 2535 89 5186
527 2116 2747 156 5547
639 2267 2922 66 5894
616 2343 2980 120 6059
a
40%
Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure; 25 8C.
210
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
(2002) showed that these submicronic particles are close packed multilamellar vesicles formed with hundreds of bilayer stacks. Analogous molecular organisations has recently drawn considerable attention and found to form when a lamellar phase is subjected to shear (Berghausen, Zipfel, Lindner, & Richtering, 1998; Diat, Roux, & Nallet, 1993; Richtering, 2001). Such structures are referred to as onion phases. At the best of our knowledge, it is the first time that such macromolecular structures are described in a nonpolar solvent and particularly in a glyceridic media. From our results, it is easy to distinguish between two groups of oils: in the first one, crude oils are easily freed from their PL and FFA contents by simple neutralisations; other oils are more difficult to refine and require additional phosphoric acid conditioning before neutralisation and microfiltration to get satisfactory quality, which may enable them to be bleached. This is assumed to be originated from higher levels of the so-called nonhydratable phospholipids. In fact, the elimination of such PL is known to be a very tough task. Factors like aging, incomplete ripeness, and enzymatic hydrolysis can lead to an enrichment of these nonhydratable PL in oils (List, Monts, & Lanser, 1992). These PL are often found to form complexes with divalent metallic cations. Phosphoric acid is often used to solubilise these metallic cations to enhance their hydratation. Segers (1979) had reported that an increase of the hydratable PL in the crude oils improved removal of the nonhydratable. Such cooperative elimination resulting from interactions between the two PL types was also found to improve the efficiency of a membrane degumming process (Subramanian et al., 1999). These authors outlined an ability of the hydratable PL to encapsulate the nonhydratable. From our results, we can notice that other components are also eliminated. Filtered oils water content was each time lower than that in originally crude ones. Not only the added water via the aqueous soda solutions is retained but also that in the crude oils is reduced. In fact, a quasicomplete retention of water was observed whatever the concentration of the NaOH solutions. Visibly, the formed onion-like aggregates can bsolubiliseQ relatively important amounts of water, which most likely is accommodated between the polar head groups of PL and FFA sodium salt molecules. We assume that this partitioning phenomenon between the glyceridic and the onion phases may concern, in various proportions, other oil components. Deme´ , Dubois, and Zemb (2002) reported that the dissolution of small molecules inside multilamellar vesicles can lead to their swelling. As revealed by the lovibond colour, pigments concentrations are also greatly lowered in almost all cases. Although, this method, which consists of visually matching the colour with red and yellow lovibond colour glasses, is somehow subjective, a relatively acceptable correlation can be made with the total pigments concentrations in the oils. In our case, the processing has led to a reduction of approximately 50% of the pigments. This may lead to substantial reduction of the amount of the bleaching earths
to be used to process such oils. The pigments reduction may be attributed to their molecular destruction with the alkali or/and their solubilisation into the onion phase. Other vegetable oil components are found to be partly eliminated with the aggregates. Monoglycerides are known to be surface active agents and are often used in industries as emulsifiers can be among the PL and soaps molecules forming the bilayers of the onion phases. Gaonkar and Borwankar (1991) reported a competitive adsorption behaviour at the vegetable oil–water interface between MG and lecithins and showed the possibility of forming mixed micelles. Oppositely to all the previously studied compounds, sterols must preferably be kept in the oils due to their numerous health beneficial effects especially on the coronary diseases (Moreau, Whitaker, & Hicks, 2002). Unfortunately, noticeable reductions were obtained. The interaction between sterols and phospholipids is a wellstudied phenomenon (McConnell & Radhakrishnan, 2003) and may explain the reduction of their rates in the filtered oils. Nevertheless, in the classical refining process such reduction is reported to concern mainly the neutralisation step (Kochhar, 1983; Pasqualone & Catalano, 2000) rather than the degumming step. Probably, the sterols in the glyceridic media do interact mainly with FFA salts rather than phospholipids. Verleyen et al. (2002) reported that only free sterols were eliminated with the soaps. The mechanisms of such partitioning between the bulk triglyceridic media and the onion phases has not been investigated. Thus, we believe that the elimination of a portion of the pigments, MG, sterols is facilitated either by their polar character but mainly by an enhanced solubilisation by the soap aggregates. In fact, none of these components is known to be eliminated to such extent by simple washing. The relatively low fluxes obtained when membrane processing vegetable oils is among the main hurdles impeding industrial applications. The tangential flow in the crossflow filtration permit the transport of the particles away from the membrane surface and thus limit the cake formation. This main advantage of the crossflow filtration over the dead end allows better fluxes. However, as far as the particles formed in the oils are considered as a soft matter, the shear stress imposed on aggregates because of their recirculation throughout the system may alter their morphology or even led to their breakage. Works on membrane bioreactors dealing with living cells has noticed breakage of microbial flocks due to pumping shear (Kim, Lee, & Chang, 2001). In our case, this may partially explain the need of smaller membrane pores sizes to get satisfactory retention in comparison to dead-end microfiltration. In the tested range (50–200 kPa), it seems that the transmembrane pression has not affected significantly the efficiency of the separation. However, the cake compaction, especially in the case of dead end filtration, may directly lead to flux reduction since the void fraction in the cake is diminished but may also explain the good
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
retention observed in these cases in spite of the relatively large pore sizes. Viscosity reduction can also improve the fluxes either with solubilising the oils in organic solvents (miscella phase) or with increased temperature. The miscella phase evidently cannot solve the problem since it rises safety concerns. In our case, the increase of temperature had negatively affected the effectiveness of the separation. The stability of the vesicle-like aggregates seems to be very temperature sensitive; increasing amounts of soaps and PL molecules permeate across the membrane when temperature was elevated above 25 8C. In fact, weak and short range forces are known to play the main role in self-assembled structures and also to be very temperaturesensitive. As the increase of temperature led to poor retention of soaps and phospholipids, we assume that this can be the result of increasing solubilisation of soaps and phospholipids as monomers or/and small clusters in the continuous triglyceridic phase and/or of a change of the structures of the aggregates. In fact, Buchanan, Egelhaaf, and Cates (2001) reported that at 30 8C onion phases can dmeltT into a sponge phase.
5. Conclusion A promising membrane-based refining process of vegetable oils was tested. Crude oils were successfully degummed and deacidified in a single operation and can be directly bleached. Furthermore, no wastewaters were generated. In addition, the low operating temperature is expected to preserve the bioactive and temperature sensitive components and also to reduce the energy consumption. These advantages may directly contribute to costs reductions and environment preservation. A partitioning phenomenon of some vegetable oil minor components between the bulk triglyceridic media and the aggregates phases has been demonstrated. The comprehension of the mechanisms of such partitioning may help to reduce the loss of beneficial components.
Acknowledgments The authors wish to thank the IFS (International Foundation for Science; Grant No. E/1480) and the AUF (Agence Universitaire Francophone; LAF 313) for their financial support.
References AFNOR (1984). Recueil des normes franc¸aises: Corps gras, Graines ole´agineuses et produits de´rive´s (3rd ed.). Ajana, H., Pioch, D., & Graille, J. (1993). Raffinage d’huiles ve´ge´tales ou animales par filtration. French Patent Application, 93-02999. Berghausen, J., Zipfel, J., Lindner, P., & Richtering, W. (1998). Shearinduced orientations in a lyotropic defective lamellar phase. Europhysics Letters, 43, 683 – 689.
211
Bhanushali, D., Kloos, S., & Bhattacharyya, D. (2002). Solute transport in solvent-resistant nanofiltration membranes for non-aqueous systems: experimental results and the role of solute–solvent coupling. Journal of Membrane Science, 208, 343 – 359. Bornaz, S., Fanni, J., & Parmentier, M. (1995a). Filtration in hydrophobic media: 1. Evidence of molecular selection by cross-flow filtration of butter oil. Journal of the American Oil Chemists’ Society, 72, 1139 – 1142. Bornaz, S., Fanni, J., & Parmentier, M. (1995b). Filtration in hydrophobic media: 2. Triglyceride partition phenomenon as observed by tangential filtration of butter oil. Journal of the American Oil Chemists’ Society, 72, 1143 – 1148. Buchanan, M., Egelhaaf, S. U., & Cates, M. E. (2001). Swelling and dissolution of onion phases: The effect of temperature. Colloids and Surfaces. A, Physicochemical and Engineering Aspects, 183, 293 – 302. Catalano, M., De Leonardis, T., & Comes, S. (1994). Diacylglycerols in the evaluation of virgin olive oil quality. Grasas y Aceites, 45, 380 – 384. Cert, A., Lanzon, A., Carelli, A. A., Albi, T., & Amelotti, G. (1994). Formation of stigmasta-3,5-diene in vegetable oils. Food Chemistry, 49, 287 – 293. De, B. K., Das, R., Dutta, B. K., & Bhattacharyya, D. K. (1998). Membrane degumming and dewaxing of rice bran and its refining. Fett/Lipid, 100, 416 – 421. Deme´, B., Dubois, M., & Zemb, T. (2002). Swelling of a lecithin lamellar phase induced by small carbohydrate solutes. Biophysical Journal, 82, 215 – 225. Denise, J. (1983). Le raffinage des corps gras. (Westhoek edition). Dunkerque7 Editions des Beffrois. Devinant, G., Scamaroni, L., & Naudet, M. (1980). Isome´risation de l’acide linole´nique durant la de´sodorisation des huiles de colza et de soja. Revue Francaise Des Corps Gras, 27, 283 – 287. Diat, O., Roux, D., & Nallet, F. (1993). Effect of shear on lyotropic lamellar phase. Journal de Physique II, 3, 1427 – 1452. Frega, N., Mozzon, M., & Lercker, G. (1999). Effects of free fatty acids on oxidative stability of vegetable. Journal of the American Oil Chemists’ Society, 76, 325 – 329. Gaonkar, A. G., & Borwankar, R. P. (1991). Competitive adsorption of monoglycerides and lecithin at the vegetable oil–water interface. Colloids and Surfaces, 59, 331 – 343. Gomes, T. (1992). Oligopolymer, diglycerides and oxidized triglyceride contents as measures of olive oil quality. Journal of the American Oil Chemists’ Society, 69, 1219 – 1223. Gupta, A. K. S. (1977). Process for refining crude glyceride oils by membrane filtration. US Patent, 4,062,882. Gupta, A. K. S. (1986). Novel developments in refining of edible oils. Fette, Seifen, Anstrichmittel, 88, 79 – 86. Hafidi, A., Pioch, D., & Ajana, H. (2004). Soft purification of lampante olive oil by microfiltration—influence of oil conditioning on the permeate flux and cake properties. European Journal of Lipid Science and Technology, 106, 152 – 159. ISO standards methods (1985). Phosphorus and cations : ISO 6884. Jawad, I. M., Kochhar, S. P., & Hudson, B. J. F. (1983). Quality characteristics of physically refined soybean oil: Effects on component fatty acids and on triglycerides. Lebensmittel-Wissenschaft + Technologie, 17, 289 – 293. Kale, V., Katikaneni, S. P. R., & Cheryan, M. (1999). Deacidifying rice bran oil by solvent extraction and membrane technology. Journal of the American Oil Chemists’ Society, 76, 723 – 727. Kim, I., Kim, J., Lee, K., & Tak, T. (2002). Phospholipids separation (degumming) from crude vegetable oil by polyimide ultrafiltration membrane. Journal of Membrane Science, 205, 113 – 123. Kim, J., Lee, C., & Chang, I. (2001). Effect of pump shear on the performance of a crossflow membrane bioreactor. Water Research, 35, 2137 – 2144. Kiristakis, A., & Tsipeli, A. (1992). Relationship of the acidity of olive oil to its resistance to oxidation. Rivista Italiana Delle Sostanze Grasse, 69, 513 – 515.
212
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
Kochhar, S. P. (1983). Influence of processing on sterols of edible vegetable oils. Progress in Lipid Research, 22, 161 – 188. Koseoglu, S. S., Lawhon, J. T., & Lusas, E. W. (1990). Membrane processing of crude vegetable oils: pilot plant scale removal of solvent from oil miscellas. Journal of the American Oil Chemists’ Society, 67, 315 – 322. Koseoglu, S. S., Rhee, K. C., & Lusas, E. W. (1990). Membrane processing of crude vegetable oils: Laboratory scale membrane degumming, refining and bleaching. Proceedings of the World Conference on Edible Fats and Oils Processing: Basic Principles and Modern Practices (pp. 182 – 188). Champain7 American Oil Chemist’s Society. Largueze, C., Pioch, D., & Gulik-Kryzwicki, T. (2002). Structure of clusters of phospholipids and soaps during refining of vegetable oils. Meeting of the Euro Fed Lipids, Strasbourg 25–28 Nov. 2002. List, G. R., Monts, T. L., & Lanser, A. C. (1992). Factors promoting the formation of nonhydratable soybean phosphatides. Journal of the American Oil Chemists’ Society, 69, 443 – 467. McConnell, H. M., & Radhakrishnan, A. (2003). Condensed complexes of cholesterol and phospholipids. BBA-Biomembranes, 1610, 159 – 173. Moreau, R. A., Whitaker, B. D., & Hicks, K. B. (2002). Phytosterols, phytostanols, and their conjugates in foods: Structural diversity, quantitative analysis, and health-promoting uses. Progress in Lipid Research, 41, 457 – 500. Pasqualone, A., & Catalano, M. (2000). Free and total sterols in olive oils. Effects of neutralization. Grasas y Aceites, 51, 180 – 182. Plank, C., & Lorbeer, E. (1992). Quality control of vegetable oil methyl esters used as diesel fuel substitutes: Quantitative determination of mono-, di, and triglycerides by capillary GC. Journal of High Resolution Chromatography, 15, 609 – 612. Pioch, D., Largueze, C., Graille, J., Ajana, H., & Rouviere, J. (1998). Towards an efficient membrane based process for vegetable oils refining. Industrial Crops and Products, 7, 83 – 89. Pioch, D., Largueze, C., Graille, J., Hafidi, A., Ajana, H., & Graille, J. (1996). Membrane based vegetable oil refining process. Tangential
filtration at laboratory and pilot scales. World Conference and Exhibition on Oilseed and Edible Oils Processing, Istambul, Turkey, October 6–10th. Raman, L. P., Cheryan, M., & Rajagopalan, N. (1996). Deacidification of soybean oil by membrane technology. Journal of the American Oil Chemists’ Society, 73, 219 – 224. Richtering, W. (2001). Rheology and shear induced structures in surfactant solutions. Current Opinion in Colloid & Interface Science, 6, 446 – 450. Schulte, E. (1995). Determination of edible fat refining by HPLC of 3,5steradienes. Fat Science and Technology, 96, 124 – 128. Segers, J. C. (1979). Oil purification by adding hydratable phosphatides. U.S. Patent 4,162,260. Subramanian, R., & Nakajima, M. (1997). Membrane degumming of crude soybean and rapeseed oils. Journal of the American Oil Chemists’ Society, 74, 971 – 975. Subramanian, R., Nakajima, M., Yasui, A., Nabetani, H., Kimura, T., & Maekawa, T. (1999). Evaluation of surfactant aided degumming of vegetable oils by membrane technology. Journal of the American Oil Chemists’ Society, 76, 1247 – 1253. Trautwein, E. A., Duchateau, G., Lin, Y., Mel’nikov, S. M., Molhuizen, H. O. F., & Ntanios, F. Y. (2003). Proposed mechanisms of cholesterollowering action of plant sterols. European Journal of Lipid Science and Technology, 105, 171 – 185. Verleyen, T., Sosinska, U., Ioannidou, S., Verhe, R., Dewettinck, K., Huyghebaert, A., et al. (2002). Influence of the vegetable oil refining process on free and esterified sterols. Journal of the American Oil Chemists’ Society, 79, 947 – 953. White, L. S., & Nitsch, A. R. (2000). Solvent recovery from lube oil filtrates with a polyimide membrane. Journal of Membrane Science, 179, 227 – 267. Zwijnenberg, H. J., Krosse, A. M., Ebert, K., Peinemeann, K. -V., & Cuperus, F. P. (1999). Acetone-stable nanofiltration membranes in deacidifying vegetable oil. Journal of the American Oil Chemists’ Society, 76, 83 – 87.
Membrane-based simultaneous degumming and deacidification of vegetable oils Abdellatif Hafidia,*, Daniel Piochb, Hamid Ajanaa a Laboratoire Sciences des Aliments, Universite´ Cadi Ayyad, Faculte´ des Sciences Semlalia, BP 2390, Marrakech, Morocco Physico-Chemistry of Processes and Bioenergy Laboratory, Agrifood Systems Programme CIRAD-AMIS, TA 40/16, 34398 Montpellier Cedex 5, France
b
Received 20 January 2004; accepted 16 December 2004
Abstract An efficient membrane based process for simultaneous degumming and deacidification of vegetable oil was investigated. Appropriate crude oil conditioning allow the formation of submicronic aggregates, composed with soaps molecules resulting from the neutralisation of FFA and PL, which are retained when microfiltrating. Initial flux for the 0.8 Am membranes (~560 l/h m2) was about twice that of the 0.5 Am and about 10 times that of 0.2 Am membrane. The filtered oils showed good quality in the case of 0.2 and 0.5 Am membranes, but the use of 0.8 Am membranes has allowed some soaps to pass through. Two types of crude oils behaviour were noticed. Oppositely to some oils for which just simple neutralisation led to a satisfactory elimination of the phospholipids, others were very tough to refine. The operating pression seems not to affect the efficiency of the separation, whereas the stability of the vesicle-like aggregates is found to be greatly affected by the increase of the temperature above 25 8C. Beside the quasi-elimination of FFA, PL and water, minerals and pigments contents were also greatly lowered. When using an NaOH 20%, the lovibond yellow score lowered from around 28 to 10 in the case of sunflower oils and from ~34 to 6–20 in the case of soya and rapeseed oils. The monoglycerides were almost undetectable after membrane processing whatever the type of the conditioning used. After processing, the diglycerides contents, which ranges in the tested crude oils between 0.8% and 1.0%, showed almost no changes for two oils, whereas noticeable increases were obtained for the other oils. Total phytosterols contents were systematically reduced. The reductions vary from 3% to 44% upon the case. Neutralisation with an NaOH 20% lead to higher sterol losses in comparison to NaOH 40%. All the sterol components contents were found to be reduced in almost similar proportions. D 2005 Elsevier Ltd. All rights reserved. Keywords: Vegetable oils; Degumming; Deacidification; Microfiltration; Minor components; Quality Industrial relevance: Conventional oil recovery and purification processes are continuousely sought to be replaced by gentler processing conditions. Membrane based oil refining operating at low temperatures and without the generation of waste water offer an interesting and promising approach towards bgreenerQ technologies.
1. Introduction Excepting extra virgin olive oils and some speciality oils, vegetable oils must undergo refining operations to remove undesirable components granting them thus satisfactory purity and stability characteristics (acidity, colour, oxidative, and sensory). Classical edible oil refining processes include degumming, neutralisation, bleaching, and deodorisation. * Corresponding author. Tel.: +212 44 43 46 49; fax: +212 44 43 74 12. E-mail address: [email protected] (A. Hafidi). 1466-8564/$ - see front matter D 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ifset.2004.12.001
These series of operations aim mainly to remove phospholipids (PL), free fatty acids (FFA), pigments, hydroperoxides, and waxes. Such a processing can be operated according to two main schemes differing essentially in the manner the FFA are removed. In the so called chemical or alkali refining, they are turned to soaps by an alkali and, subsequently, removed from the neutral oil by centrifugations and washing steps, whereas in physical refining they are distilled during the deodorisation operation. The two processes are reputed to be very energy-consuming. In addition, all the refining operations subject oils to high temperatures; consequently, quantitative and qualitative
204
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
changes inevitably take place especially in components responsible of quality. High temperatures seem to be the most harmful parameter, which is responsible of hydrolytic, oxidative, and polymerisation alterations (Cert, Lanzon, Carelli, Albi, & Amelotti, 1994; Schulte, 1995). Structural changes are also related to the heating especially during bleaching (Schulte, 1995) and deodorisation (Cert et al., 1994). Devinant, Scamaroni, and Naudet (1980) and Jawad, Kocchar, and Hudson (1983) reported that, during physical refining, the stereoisomerisation and polymerisation depend on the time and the temperature. Some sterols can undergo a dehydration during bleaching (Schulte, 1995) and deodorisation (Cert et al., 1994) so h sitosterol can form the 3,5 stigmadiene. Total sterols are also reported to lower after neutralisation (Kochhar, 1983). Pasqualone and Catalano (2000) reported that neutralised oils showed losses up to 50% in total sterols. Catalano, De Leonardis, and Comes (1994) and Gomes (1992) have reported an increase of diglyceride amounts in refined olive oil. Most of these alterations are temperature-dependent. Improving the quality of the processed oils by reducing side effects and diminishing the energy consumption has been among the most active research areas throughout the past half-century. Due to its characteristics as non-polluting and energy saving technology, membrane processing has recently emerged as a promising alternative. Potential applications covering various aspects such as solvent removing (Bhanushali, Kloos, & Bhattacharyya, 2002; Koseoglu, Lawhon, & Lusas, 1990; White & Nitsch, 2000), edible oil degumming (Kim, Jong-Ho Kim, Lee, & Tak, 2002; Koseoglu, Rhee, & Lusas, 1990; Subramanian & Nakajima, 1997), deacidification and free fatty acids recovering (Kale, Katikaneni, & Cheryan, 1999; Raman, Cheryan, & Rajagopalan, 1996; Zwijnenberg, Krosse, Ebert, Peinemeann, & Cuperus, 1999), dewaxing (De, Das, Dutta, & Bhattacharyya, 1998), and triacylglycerols partitioning (Bornaz, Fanni, & Parmentier, 1995a, 1995b) were reported. In the very first works from the last century, Sen Gupta, a pioneer, demonstrated the efficiency of degumming the oil miscellas using ultrafiltration techniques (Gupta, 1977, 1986). Ajana, Pioch, and Graille (1993) demonstrated the possibility of a complete degumming and deacidification of crude vegetable oils subsequently to appropriate conditioning using membrane separation techniques. The neutralisation of the free fatty acids with appropriate soda concentration and under adequate conditions allow the formation of submicronic particles with onion-like structures as revealed by scanning electron microscopy (Largueze, Pioch, & Gulik-Kryzwicki, 2002). These structures are believed to be formed not only by fatty acid salts but also with phospholipids and water molecules. Due to their relatively important sizes, these aggregates can be easily removed by a microfiltration technique. Such structures are thought to be formed of hundreds of fatty acid salts and phospholipid bilayer stacks. Both oil conditioning and filtration can be operated at low temperature (20–25 8C).
These mild conditions which contrast with the conditions in industrial practices are expected to preserve the sensitive and bioactive components in edible oils. In the present paper, we report an assessment of the efficiency of such a processing by testing different crude oils and comparing their suitability for our membranebased purification process. The effects of some physicochemical parameters on the stability of the aggregates and therefore on the efficiency of the separation were studied. The impact of our simultaneous degumming and neutralisation process on some minor components of the oils was also investigated.
2. Materials and methods 2.1. Oil samples Soya, sunflower, and rapeseed crude oils were obtained from CEREOL, se`te, France. The crude oils were neutralised at room temperature (20–25 8C) (unless otherwise stated) by pouring slowly with a pipette an aqueous sodium hydroxide solution (20% or 40% w/v) under magnetic stirring (600 rpm) and then filtered as described in previous works (Hafidi, Pioch, & Ajana, 2004; Pioch, Largueze, Graille, Ajana, & Rouviere, 1998; Pioch et al., 1996). In the cases of acid conditioning, 0.1% or 0.3% (w/w) of the acid phosphoric 85% (Sigma) were added at room temperature (20–25 8C) to the crude oils under vigorous mixing during 20 min before neutralisation of the whole acidity with an appropriate NaOH solution. 2.2. Microfiltrations Dead end filtration experiments were performed with a stainless steel Gelman module (200 ml volume, 200 kPa pressure; Whatman cellulose filter (pore size: 2.5 Am, filtration area 16 cm2)). Crossflow filtration experiments were carried out at 25 8C with a laboratory stainless steel apparatus (1 kg oil samples, tubular alumina membrane 200 mm length, 40 cm2 area) under: continuous permeate recycling, 3.5 m/s as a tangential velocity and 200 kPa transmembrane pressure. Fluxes were estimated by direct measurements using an appropriate graduated cylinder. 2.3. Analytical methods In order to check the chemical composition of both starting and refined oils, French standard procedures (AFNOR, 1984) were applied: phosphorus NF T 60-227, soaps NF T 60-217, free fatty acids NF T 60-204, water NF T 60-367, and peroxide value NF T 60-220. Cations (Na, Ca, Mg, Fe, Cu) were analyzed after mineralisation following the ISO 6884: 1985 (ISO standards methods, 1985) standard method, comprising mineralisation and atomic emission spectrophotometry analysis (inductive
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
coupled plasma, ICP) with a Jobin Yvon JY spectrophotometer. The results are the average of two independent trials (both mineralisation and ICP analysis). The composition of sterols were determined according to the French standard method NF T 60-258 (AFNOR, 1984). After extraction and separation of the unsaponifiable fraction of the oil by thin layer chromatography, the sterols were silylated and analyzed using an SE-54 (Supelco) 30 m; 0.25 internal diameter capillary column under the following conditions: the oven initial temperature (180 8C) was kept for 8 min before starting a temperature program: 5 C/min until 260 8C which is maintained for 15 min; injector and detector temperatures were 280 and 290 8C, respectively. The a cholestane (sigma) was used as an internal standard. Partial glycerides were analyzed by gas chromatography according to the method described by Plank and Lorbeer (1992). A DB1 WTHN capillary column was used (7 m length, 0.32 mm internal diameter). The carrier gas (helium) flow rate was set at 2.9 ml/min; the temperature was held at 100 8C for 3 min before a rise of 10 8C/min until 340 8C, which is maintained for 5 min. Hexatriacontane was used as an internal standard. The silylation was performed with the bis(trimethylsilyl) trifluoroacetamide (BSTFA) and the trimethylchlosilane (TMCS). 2.4. Lovibond color A lovibond tintometer was used and the colour of oil was matched with a set of standard coloured numbered glasses, ranging in the scales from 0 to 70 red (R) and 0 to 70 yellow (Y). The results are expressed as R and Y values.
3. Results 3.1. Membrane separation efficiency Numerous crude oils were tested but only five samples, two sunflower oils (SFO1 and SFO2), two soya oils (SO1 and SO2), and a rapeseed oil (RO), which summarise the most featured behaviours, will be presented and discussed. The main aim of our membrane-based refining process is to produce in a single step a suitable oil for bleaching. Thus, FFA, phospholipids, soaps, and water must be removed almost completely. Only oils containing less than 10 ppm phosphorus, 30 ppm soaps, and less than 0.1% water are considered as appropriate for bleaching (Denise, 1983). Results depicted in Table 1 recapitulate the main characteristics of the crude oils and revealed acceptable quality of the processed oils. Crude oils FFA acid contents range between 0.5% and 1.5% with soya oils having the least content and rapeseed oil the highest. Phospholipids expressed in terms of phosphorus content vary from 43 to 140 ppm. These relatively low PL contents confirm that these crude vegetable oils are actually water degummed. The filtration of the conditioned crude oils removes almost all the PL,
205
soaps resulting from the neutralisation of FFA, and water. Excepting the case of the SO2, the filtration of the crude oils subsequently to a simple neutralisation (without a prior phosphoric acid pre-treatment) has led in almost cases to a quasi elimination of the initial acidity and also to a great reduction of the phospholipids content. The filtered oils are also almost free from soap molecules resulting from the alkali treatment and from the water added via the sodium hydroxide solutions. PL retentions vary from 74% to 98%; consequently, the phosphorus content did not exceed 10 ppm in most cases. Despite the very high soaps retention rates obtained for the simply neutralised and filtered oils, their quality can be considered satisfactory only when the soaps molecules were almost undetectable in the filtered oils. Simple neutralisations has led to an excellent oil quality (i.e. oils which can be directly bleached without any additional purification) only in the case of SFO2 whatever the NaOH concentration and in the cases of SO1 and RO when using an NaOH 20%. In fact, soaps retentions seem to be higher in the case of neutralisation with NaOH 20%. The use of 0.1% phosphoric acid pre-treatment prior to the neutralisation of the FFA resulted in an improvement of the retention of phospholipids and soaps especially in the case of SFO1 with both NaOH concentrations and only in the case of an NaOH 40% for the SO1 and RO. In these two latter cases, the soaps retention was still not complete and few amounts of soaps molecules were detected in the filtered oils (Table 1). The SO2 showed only very slight retention of phospholipids in the case of an NaOH 20%. The increase of the amount of the phosphoric acid pre-treatment to 0.3% allowed better retentions of phospholipids and soaps for all the oils. Even the SO2 showed an acceptable retention after a 0.3% H3PO4 pre-treatment when neutralised with an NaOH 20%. Beside the elimination of FFA, PL, and water, the processed oils showed lower amounts of pigments as revealed by the lovibond colour estimation. When using an NaOH 20%, the yellow score lowered from around 28 to 10 in the case of SFO1 and SFO2 and from ~34 to values ranging from 16 to 20 for the SO1 and RO. Slightly higher values were obtained when using an NaOH 40%. Similar decrease was observed in the red. The phosphoric acid pretreatment seems to allow slightly lower values. The SO2 is definitely an exception, just a very slight diminishing of the colour is observed. Minerals are known to be pro-oxidants and must preferably be eliminated from edible oils to guarantee good oxidative stability. As the membrane processing of the crude oils succeed in greatly lowering the iron, calcium, magnesium, and sodium cations especially when the acid phosphoric is used (Table 2). In spite of its straightforwardness, dead end filtration seems not to fit industrial applications because of very fast and severe fouling. In crossflow filtrations, the tangential flow sweeping continuously the membrane surface limits the deposition and the formation of a cake layer and therefore
206
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
Table 1 Physico-chemical characteristics of the crude and membrane processed oilsa [NaOH]
SFO1
Crude Processed
20%
40%
SFO2
Crude Processed
20%
40%
SO1
Crude Processed
20%
40%
SO2
Crude Processed
20%
40%
RO
Crude Processed
20%
40%
a
[H3PO4]
– 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3% 0% 0.1% 0.3% 0% 0.1% 0.3%
FFA (%)
Phosphorus (ppm)
Soaps (ppm)
Water content (%)
Colour Y
R
0.97 0.06 0.06 0.05 0.08 0.08 0.06 1.00 0.04 0.02 0.02 0.05 0.06 0.04 0.73 0.04 0.06 0.04 0.06 0.04 0.06 0.49 0.04 0.04 0.04 0.06 0.04 0.06 1.55 0.04 0.02 0.02 0.06 0.04 0.04
43 9.0 0.8 1.5 11.0 2.2 5.5 67 1.2 0.6 0.8 4.6 2.7 0.3 96 6.8 4.1 3.3 11.2 7.6 2.0 141 155 9.1 7.4 141 231 128 123 12.0 3.0 1.8 19 5.4 7.6
– 144 25 nd 98 nd nd – nd nd nd 24 nd nd – nd 38 nd 244 70 nd – 840 786 nd 740 1725 816 – 34 nd nd 416 230 nd
0.05 0.08 0.08 0.06 0.10 0.07 0.06 0.08 0.06 0.04 0.05 0.05 0.08 0.06 0.17 0.03 0.09 0.05 0.06 0.05 0.05 0.13 0.11 0.07 0.08 0.11 0.11 0.07 0.09 0.07 0.03 0.05 0.06 0.05 0.04
26 10 10 10 15 14 11 28 12 10 10 14 12 12 34 20 16 16 22 18 17 37 34 33 34 34 34 34 36 22 20 20 26 24 24
2.0 1.5 1.2 1.2 1.2 1.2 1.2 2.2 1.4 1.0 1.0 1.2 1.2 1.2 2.8 2.0 1.5 1.6 2.1 1.8 1.8 2.5 2.2 2.0 2.2 2.2 2.2 2.2 3.0 2.4 1.8 1.8 2.6 2.0 2.0
Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa pressure; 25 8C.
allows better fluxes. As shown in Fig. 1, three different pore sizes (0.8, 0.5, and 0.2 Am) were tested. The initial flux for the 0.8 Am membranes (~560 l/h m2) is about twice that of the 0.5Am and about 10 times that of 0.2 Am membrane. After 30 min of crossflow filtration, the fluxes are reduced to more than the half. After 2 h, a relative stabilisation of the fluxes is observed around 40, 30, and 15 l/h m2 for the 0.8, 0.5, and 0.2 Am membranes, respectively. Nevertheless, the quality of the filtered oils was not always acceptable. Relatively large amounts of soap molecules are found in the permeate when 0.8 Am pore size membrane is used (Table 3). This seems contradictory with the dead end filtration results where larger pore sizes (2.5 Am) allowed satisfactory retentions. On one hand, the cake formed in early instant of the dead end filtration which most likely play the role of the actual membrane and, on the other hand, the shear caused by pumping and circulation of the conditioned oil during crossflow filtration which probably led to mechanical break of the aggregates may explain these results.
3.2. Influence of the operating temperature To evaluate the effect of the operating temperature on the efficiency of our membrane separation process, crude oils were neutralised and filtered at temperatures raging between 25 and 80 8C. All the oils showed a similar behaviour: a decrease of retention of phospholipids and soaps was observed as the operating temperature was elevated. Results concerning the SFO2 are depicted in Table 4. As a result of the neutralisation, the FFA content is greatly reduced in the filtered oils in most cases and visibly has not been affected with temperature. However, at 80 8C, this reduction was less satisfactory; an undesirable hydrolysis of the triacylglycerols generating additional amounts of FFA may be the cause. At 25 8C, the phospholipids retention reached 84% to 98%, whereas it was only of 46% to 78% at 60 8C. Increasing amounts of soap molecules permeate across the membrane as the temperature was increased. Nevertheless, it seems that
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212 Table 2 Minerals contents in crude and membrane processed oilsa [NaOH] SFO1
Crude Processed
20%
40%
SO2
Crude Processed
20%
40%
RO
Crude Processed
20%
40%
207
Table 3 Quality characteristics of the crossflow filtered SFO1a
[H3PO4]
Fe (ppm)
Ca (ppm)
Mg (ppm)
Na (ppm)
– 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3% – 0% 0.1% 0.3% 0% 0.1% 0.3%
4.6 0.3 0.1 0.1 0.1 0.1 0.1 7.0 0.7 0.2 0.1 1.3 0.3 0.3 5.8 1.2 0.6 0.8 2.9 0.4 0.6
76.0 1.5 0.7 1.2 11.5 1.3 1.8 77.0 47.0 15.0 11.0 70.0 66.5 0.9 53.0 23.0 7.0 3.7 36.0 6.8 7.7
39.0 0.4 0.7 0.4 4.0 2.5 3.3 39.0 25.0 6.9 3.2 35.0 12.6 0.1 58.0 27.5 10.0 6.9 13.7 5.0 3.6
15.0 6.5 4.2 2.8 5.7 6.0 2.1 4.3 37.0 5.3 2.6 61.5 137.0 3.8 21.0 12.0 7.2 4.8 17.0 15.3 3.5
a Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure; 25 8C.
phosphoric acid pre-treatments improve the retention of the aggregates. Indeed, a noticeable difference was observed and a relatively acceptable retention rates were obtained until 40 8C in these cases.
SFO 1 0.8 Am 0.5 Am 0.2 Am
FFA (%)
P (ppm)
Soaps (ppm)
Water (%)
0.97 0.06 0.04 0.04
43.3 10.6 2.6 0.5
– 1586 nd nd
0.046 0.128 0.042 0.034
a Crossflow microfiltration conditions: alumina membrane, continuous permeate recycling, 3.5 m/s tangential velocity, and 200 kPa transmembrane pressure.
structures. Excepting the case of the SO2 and as already shown (Table 1), the operating pressure (200 kPa) seems not to be destructive for the structures of the aggregates since good retentions were obtained. Thus, we have examined the influence of the transmembrane pressure on the efficiency of the retention to demonstrate its effects if any and try to explain the particular behaviour of SO2. Results in Table 5 show that the pressure in the range (50–200 kPa) seems not to have any noticeable effect on the retention of the aggregates. The permeation of soaps and phospholipids in the case of SO2 cannot be explained by a breakable behaviour of its aggregates. Nevertheless, a slight sensibility to the operating pressure, especially regarding the retention of phospholipids, was observed with the NaOH 20% simple neutralisations. In some cases, negative phosphorus retention values were obtained mainly when phosphoric acid pretreatment is used. This may be explained by a permeation of the phosphoric acid salts in the filtered oils, which thus contain bigger amounts of phosphorus than the crude ones.
3.3. Influence of the transmembrane pressure
Fluxes (l/h/m2)
As far as the aggregates formed in the crude oils subsequently to conditioning are assumed to be mainly formed by soaps, water, and phospholipids, they can be considered as a soft matter. Therefore, the driven pressure is expected to be a crucial parameter for the integrity of such
3.4. Effect of the conditioning on the partial glycerides contents The partial glycerides are natural components occurring in oils, they result from the triacylglycerols hydrolysis. Due to their surface activity, they can cause neural oil losses by
600
0.8 µm 0.5 µm 0.2 µm
500
400
300
200
100
0 0
50
100
150
200
250
300
Time (min) Fig. 1. Evolution of fluxes during microfiltration of the SFO1 conditioned with 0.3% phosphoric acid and neutralised with an NaOH 20%.
208
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
Table 4 Effect of the temperature on the efficiency of the separation (case of the SFO2)a [NaOH] [H3PO4] Temperature FFA Phosphorus Soaps Water (%, w/w) (8C) reduction retention retention content (%) (%) (%) (%) 20%
0
0.3
40%
0
0.3
25 40 60 80 25 40 60 80 25 40 60 80 25 40 60 80
96.0 94.0 95.0 92.0 96.0 94.0 90.0 84.0 93.0 94.0 80.0 88.0 96.0 96.0 91.0 69.0
98.1 82.9 55.1 49.0 92.9 87.9 78.6 22.7 83.7 74.2 46.2 3.0 95.6 98.2 68.1 6.7
100.0 99.9 98.5 99.1 100.0 100.0 99.9 98.9 99.9 99.8 99.5 98.9 100.0 99.9 99.8 99.0
0.03 0.07 0.05 0.02 0.02 0.03 0.07 0.08 0.04 0.06 0.04 0.03 0.04 0.04 0.05 0.06
a Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure.
emulsification during the refining operations. Excepting the SFO2, the monoglycerides were almost undetectable in all the tested crude oils and no increase was observed after processing whatever the type of the conditioning used (Table 6). Diglycerides content ranges in crude oils between 0.8% and 1.0% and seems to be somehow proportional to the FFA contents. After processing, the diglycerides contents showed almost no changes for two oils (SFO1 and SO1). Meanwhile, noticeable increases were obtained Table 5 Effect of the transmembrane pressure on the efficiency of the separation in the case of the SO2a [NaOH]
20%
[H3PO4] (%, w/w)
Pressure (kPa)
FFA reduction (%)
0
50 100 200 50 100 200 50 100 200 50 100 200 50 100 200 50 100 200
91.84 91.84 91.84 91.84 91.84 91.84 91.84 91.84 91.84 87.76 87.76 87.76 91.84 91.84 91.84 87.76 87.76 87.76
0.1
0.3
40%
0
0.1
0.3
P retention (%) 58.16 8.51 9.93 94.33 96.81 93.55 99.29 93.40 94.75 1.21 2.48 0.43 68.79 52.48 63.83 10.64 24.11 9.22
Soaps retention (%)
Water content (%)
98.80 98.97 98.60 98.96 98.84 98.69 100.0 99.95 100.0 99.07 98.75 98.77 98.80 98.23 97.13 98.75 98.93 98.64
0.078 0.161 0.112 0.078 0.088 0.074 0.063 0.056 0.082 0.101 0.116 0.110 0.078 0.161 0.112 0.068 0.088 0.067
a Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am; 25 8C.
for the other oils. These increases reached approximately 200% in the case of SO2 simply neutralised with an NaOH 20% but generally they vary from 30% to 60%. These variations seem not to be influenced neither by the NaOH nor the phosphoric acid concentrations. 3.5. Effect of the conditioning on the sterol composition Sterols are among the major components of the unsaponifiable fraction of the vegetable oils. Because of their beneficial effects on human health especially their action of blood cholesterol lowering, the phytosterols are desirable components in refined oils (Trautwein et al., 2003). Unfortunately, the processed oils (Tables 7 and 8) showed lower sterol contents comparatively to the crude ones. Total phytosterols reduction vary from 3% to 44% upon the case. Neutralisation with an NaOH 20% lead to higher sterol losses in comparison to NaOH 40%. Reductions of 17 to 23%, 25 to 44% and 22 to 35% were obtained when using an NaOH 20% for the SFO1, SO1, and RO, respectively, whereas they ranged from 3% to 18%, 21% to
Table 6 Effect of the processing on partial glycerides contentsa Samples SFO1
Crude Processed
[NaOH]
[H3PO4]
20%
0% 0.1% 0.3% 0% 0.1% 0.3%
40%
SFO2
SO1
Crude Processed
Crude Processed
20% 40%
0% 0% 0.1% 0.3%
20%
0% 0.1% 0.3% 0% 0.1% 0.3%
40%
SO2
Crude Processed
20% 40%
RO
Crude Processed
20%
40%
0% 0.1% 0% 0.1% 0.3% 0% 0.1% 0.3% 0% 0.1% 0.3%
Monoglycerides (%)
Diglycerides (%)
nd 0.01 0.01 0.01 0.01 0.01 nd 0.013 0.03 0.02 0.02 0.013 Tr. nd nd nd nd nd nd nd nd 0.02 0.012 nd nd nd 0.06 nd nd nd nd nd
0.84 0.75 0.84 0.71 0.96 0.74 0.44 1.05 1.67 1.57 1.62 1.29 0.78 0.76 0.72 0.66 0.82 0.70 0.71 0.67 1.96 0.97 1.37 1.17 1.00 1.01 1.40 1.42 1.50 1.40 1.25 1.10
a Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure; 25 8C.
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
209
Table 7 Effect of the processing on different sterol content (ppm) of the sunflower oil (SFO 1) and the soya oil (SO1)a SFO1 [NaOH]
SO1
Crude
Processed
–
20%
40%
Crude
Processed
–
20%
40%
[H3PO4]
–
0%
0.1%
0.3%
0%
0.1%
0.3%
–
0%
0.1%
0.3%
0%
0.1%
0.3%
Campesterol Stigmasterol D7 Campesterol Clerosterol h Sitosterol D5 Avenasterol D5,24 Stigmastanediol D7 Stigmastenol D7 Avenasterol Total
356 454 93 41 2641 39 37 524 100 4285
292 346 103 34 2072 25 nd 570 95 3537
281 323 105 33 1979 29 nd 561 107 3418
278 374 73 nd 2094 nd nd 375 nd 3299
400 459 111 39 2398 37 38 518 116 4117
292 360 90 27 2090 38 35 482 95 3511
324 398 108 34 2386 nd 36 576 76 3939
808 640 50 – 1859 50 – 43 – 3450
572 466 62 – 1410 47 – 30 – 2587
386 370 38 – 1076 28 – 36 – 1934
396 375 48 – 1106 26 – 23 – 1975
619 510 57 – 1463 47 – 32 – 2729
622 500 59 – 1367 44 – 33 – 2625
603 515 55 – 1358 44 – 20 – 2596
a
Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure; 25 8C.
25%, and 11% to 18% when using an NaOH 40%. In addition, these losses were more important in the cases of the phosphoric acid conditioning. As shown by the results depicted in Tables 7 and 8, all the sterol components contents were reduced in almost similar rates.
4. Discussion Demands for more environment-preserving technologies and more healthier foods are continuously growing especially in western countries. New attitudes emerge towards food processing and, consequently, more attention is paid to the influence of processing on the quality. The effects of classical refining operations on the quality and chemical composition of the oils were extensively studied. Vegetable oils are constituted predominantly of triglycerides; however, the presence of various classes of minor constituents in such oils is of a great importance. Identification and quantitation of such components, naturally present or generated by refining operations constitute an important mean either for characterisation as well for frauds detection purposes. Refining operations may eliminate, reduce, or alter the different components of the oils. PL must be removed from oils because of their emulsifying properties, which may complicate the subsequent operations, increase neutral oil losses, and may led to the formation of dark coloured
compounds during deodorisation (Denise, 1983). The FFA are also undesirable components, several works had outlined their prooxidant effects (Frega, Mozzon, & Lercker, 1999; Kiristakis & Tsipeli, 1992) contributing to the reduction of the oil shelf life. The alkali neutralisation lead to the transformation of the free fatty acids to soaps. These latter are known to be, like PL, very surface active components, which may self-assemble into wide variety of macromolecular aggregates. Our results showed quasi complete elimination of PL and FFA from the membrane-processed oils. As PL and soap molecules resulting from the alkali treatment are retained when microfiltrating, they must have somehow sizes beyond the membrane pore diameter (2.5 Am). In fact, both PL and soap molecules are known to be amphiphilic and consist of a polar headgroup and a nonpolar tails, which cause their preference for self assembly with the hydrophobic groups on one side and the hydrophilic groups on the other side. Surfactants are normally believed to exist as monomers up to a certain concentration above which they may self-assemble under specific conditions to form a wide range of organised macrostructures such as micelles, lamellar phases, liquid crystals, or crystals. Ajana et al. (1993) reported that conditioning the crude vegetable oils under specific condition lead to the formation of submicronic aggregates, which can be easily separated from the continuous glyceridic phase using a microfiltration technique. Using scanning electron microscopy, Largueze et al.
Table 8 Effect of the membrane processing on sterols content (ppm) of the rapeseed oila Crude
Processed
[NaOH]
–
20%
[H3PO4]
–
0%
0.1%
0.3%
0%
0.1%
0.3%
Brassicasterol Campesterol h Sitosterol Sitostanol Total
776 2578 3337 103 6795
715 1866 2628 76 5285
430 1715 2172 124 4441
555 2008 2535 89 5186
527 2116 2747 156 5547
639 2267 2922 66 5894
616 2343 2980 120 6059
a
40%
Dead end microfiltration conditions: cellulose membrane pore size: 2.5 Am, 200 kPa transmembrane pressure; 25 8C.
210
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
(2002) showed that these submicronic particles are close packed multilamellar vesicles formed with hundreds of bilayer stacks. Analogous molecular organisations has recently drawn considerable attention and found to form when a lamellar phase is subjected to shear (Berghausen, Zipfel, Lindner, & Richtering, 1998; Diat, Roux, & Nallet, 1993; Richtering, 2001). Such structures are referred to as onion phases. At the best of our knowledge, it is the first time that such macromolecular structures are described in a nonpolar solvent and particularly in a glyceridic media. From our results, it is easy to distinguish between two groups of oils: in the first one, crude oils are easily freed from their PL and FFA contents by simple neutralisations; other oils are more difficult to refine and require additional phosphoric acid conditioning before neutralisation and microfiltration to get satisfactory quality, which may enable them to be bleached. This is assumed to be originated from higher levels of the so-called nonhydratable phospholipids. In fact, the elimination of such PL is known to be a very tough task. Factors like aging, incomplete ripeness, and enzymatic hydrolysis can lead to an enrichment of these nonhydratable PL in oils (List, Monts, & Lanser, 1992). These PL are often found to form complexes with divalent metallic cations. Phosphoric acid is often used to solubilise these metallic cations to enhance their hydratation. Segers (1979) had reported that an increase of the hydratable PL in the crude oils improved removal of the nonhydratable. Such cooperative elimination resulting from interactions between the two PL types was also found to improve the efficiency of a membrane degumming process (Subramanian et al., 1999). These authors outlined an ability of the hydratable PL to encapsulate the nonhydratable. From our results, we can notice that other components are also eliminated. Filtered oils water content was each time lower than that in originally crude ones. Not only the added water via the aqueous soda solutions is retained but also that in the crude oils is reduced. In fact, a quasicomplete retention of water was observed whatever the concentration of the NaOH solutions. Visibly, the formed onion-like aggregates can bsolubiliseQ relatively important amounts of water, which most likely is accommodated between the polar head groups of PL and FFA sodium salt molecules. We assume that this partitioning phenomenon between the glyceridic and the onion phases may concern, in various proportions, other oil components. Deme´ , Dubois, and Zemb (2002) reported that the dissolution of small molecules inside multilamellar vesicles can lead to their swelling. As revealed by the lovibond colour, pigments concentrations are also greatly lowered in almost all cases. Although, this method, which consists of visually matching the colour with red and yellow lovibond colour glasses, is somehow subjective, a relatively acceptable correlation can be made with the total pigments concentrations in the oils. In our case, the processing has led to a reduction of approximately 50% of the pigments. This may lead to substantial reduction of the amount of the bleaching earths
to be used to process such oils. The pigments reduction may be attributed to their molecular destruction with the alkali or/and their solubilisation into the onion phase. Other vegetable oil components are found to be partly eliminated with the aggregates. Monoglycerides are known to be surface active agents and are often used in industries as emulsifiers can be among the PL and soaps molecules forming the bilayers of the onion phases. Gaonkar and Borwankar (1991) reported a competitive adsorption behaviour at the vegetable oil–water interface between MG and lecithins and showed the possibility of forming mixed micelles. Oppositely to all the previously studied compounds, sterols must preferably be kept in the oils due to their numerous health beneficial effects especially on the coronary diseases (Moreau, Whitaker, & Hicks, 2002). Unfortunately, noticeable reductions were obtained. The interaction between sterols and phospholipids is a wellstudied phenomenon (McConnell & Radhakrishnan, 2003) and may explain the reduction of their rates in the filtered oils. Nevertheless, in the classical refining process such reduction is reported to concern mainly the neutralisation step (Kochhar, 1983; Pasqualone & Catalano, 2000) rather than the degumming step. Probably, the sterols in the glyceridic media do interact mainly with FFA salts rather than phospholipids. Verleyen et al. (2002) reported that only free sterols were eliminated with the soaps. The mechanisms of such partitioning between the bulk triglyceridic media and the onion phases has not been investigated. Thus, we believe that the elimination of a portion of the pigments, MG, sterols is facilitated either by their polar character but mainly by an enhanced solubilisation by the soap aggregates. In fact, none of these components is known to be eliminated to such extent by simple washing. The relatively low fluxes obtained when membrane processing vegetable oils is among the main hurdles impeding industrial applications. The tangential flow in the crossflow filtration permit the transport of the particles away from the membrane surface and thus limit the cake formation. This main advantage of the crossflow filtration over the dead end allows better fluxes. However, as far as the particles formed in the oils are considered as a soft matter, the shear stress imposed on aggregates because of their recirculation throughout the system may alter their morphology or even led to their breakage. Works on membrane bioreactors dealing with living cells has noticed breakage of microbial flocks due to pumping shear (Kim, Lee, & Chang, 2001). In our case, this may partially explain the need of smaller membrane pores sizes to get satisfactory retention in comparison to dead-end microfiltration. In the tested range (50–200 kPa), it seems that the transmembrane pression has not affected significantly the efficiency of the separation. However, the cake compaction, especially in the case of dead end filtration, may directly lead to flux reduction since the void fraction in the cake is diminished but may also explain the good
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
retention observed in these cases in spite of the relatively large pore sizes. Viscosity reduction can also improve the fluxes either with solubilising the oils in organic solvents (miscella phase) or with increased temperature. The miscella phase evidently cannot solve the problem since it rises safety concerns. In our case, the increase of temperature had negatively affected the effectiveness of the separation. The stability of the vesicle-like aggregates seems to be very temperature sensitive; increasing amounts of soaps and PL molecules permeate across the membrane when temperature was elevated above 25 8C. In fact, weak and short range forces are known to play the main role in self-assembled structures and also to be very temperaturesensitive. As the increase of temperature led to poor retention of soaps and phospholipids, we assume that this can be the result of increasing solubilisation of soaps and phospholipids as monomers or/and small clusters in the continuous triglyceridic phase and/or of a change of the structures of the aggregates. In fact, Buchanan, Egelhaaf, and Cates (2001) reported that at 30 8C onion phases can dmeltT into a sponge phase.
5. Conclusion A promising membrane-based refining process of vegetable oils was tested. Crude oils were successfully degummed and deacidified in a single operation and can be directly bleached. Furthermore, no wastewaters were generated. In addition, the low operating temperature is expected to preserve the bioactive and temperature sensitive components and also to reduce the energy consumption. These advantages may directly contribute to costs reductions and environment preservation. A partitioning phenomenon of some vegetable oil minor components between the bulk triglyceridic media and the aggregates phases has been demonstrated. The comprehension of the mechanisms of such partitioning may help to reduce the loss of beneficial components.
Acknowledgments The authors wish to thank the IFS (International Foundation for Science; Grant No. E/1480) and the AUF (Agence Universitaire Francophone; LAF 313) for their financial support.
References AFNOR (1984). Recueil des normes franc¸aises: Corps gras, Graines ole´agineuses et produits de´rive´s (3rd ed.). Ajana, H., Pioch, D., & Graille, J. (1993). Raffinage d’huiles ve´ge´tales ou animales par filtration. French Patent Application, 93-02999. Berghausen, J., Zipfel, J., Lindner, P., & Richtering, W. (1998). Shearinduced orientations in a lyotropic defective lamellar phase. Europhysics Letters, 43, 683 – 689.
211
Bhanushali, D., Kloos, S., & Bhattacharyya, D. (2002). Solute transport in solvent-resistant nanofiltration membranes for non-aqueous systems: experimental results and the role of solute–solvent coupling. Journal of Membrane Science, 208, 343 – 359. Bornaz, S., Fanni, J., & Parmentier, M. (1995a). Filtration in hydrophobic media: 1. Evidence of molecular selection by cross-flow filtration of butter oil. Journal of the American Oil Chemists’ Society, 72, 1139 – 1142. Bornaz, S., Fanni, J., & Parmentier, M. (1995b). Filtration in hydrophobic media: 2. Triglyceride partition phenomenon as observed by tangential filtration of butter oil. Journal of the American Oil Chemists’ Society, 72, 1143 – 1148. Buchanan, M., Egelhaaf, S. U., & Cates, M. E. (2001). Swelling and dissolution of onion phases: The effect of temperature. Colloids and Surfaces. A, Physicochemical and Engineering Aspects, 183, 293 – 302. Catalano, M., De Leonardis, T., & Comes, S. (1994). Diacylglycerols in the evaluation of virgin olive oil quality. Grasas y Aceites, 45, 380 – 384. Cert, A., Lanzon, A., Carelli, A. A., Albi, T., & Amelotti, G. (1994). Formation of stigmasta-3,5-diene in vegetable oils. Food Chemistry, 49, 287 – 293. De, B. K., Das, R., Dutta, B. K., & Bhattacharyya, D. K. (1998). Membrane degumming and dewaxing of rice bran and its refining. Fett/Lipid, 100, 416 – 421. Deme´, B., Dubois, M., & Zemb, T. (2002). Swelling of a lecithin lamellar phase induced by small carbohydrate solutes. Biophysical Journal, 82, 215 – 225. Denise, J. (1983). Le raffinage des corps gras. (Westhoek edition). Dunkerque7 Editions des Beffrois. Devinant, G., Scamaroni, L., & Naudet, M. (1980). Isome´risation de l’acide linole´nique durant la de´sodorisation des huiles de colza et de soja. Revue Francaise Des Corps Gras, 27, 283 – 287. Diat, O., Roux, D., & Nallet, F. (1993). Effect of shear on lyotropic lamellar phase. Journal de Physique II, 3, 1427 – 1452. Frega, N., Mozzon, M., & Lercker, G. (1999). Effects of free fatty acids on oxidative stability of vegetable. Journal of the American Oil Chemists’ Society, 76, 325 – 329. Gaonkar, A. G., & Borwankar, R. P. (1991). Competitive adsorption of monoglycerides and lecithin at the vegetable oil–water interface. Colloids and Surfaces, 59, 331 – 343. Gomes, T. (1992). Oligopolymer, diglycerides and oxidized triglyceride contents as measures of olive oil quality. Journal of the American Oil Chemists’ Society, 69, 1219 – 1223. Gupta, A. K. S. (1977). Process for refining crude glyceride oils by membrane filtration. US Patent, 4,062,882. Gupta, A. K. S. (1986). Novel developments in refining of edible oils. Fette, Seifen, Anstrichmittel, 88, 79 – 86. Hafidi, A., Pioch, D., & Ajana, H. (2004). Soft purification of lampante olive oil by microfiltration—influence of oil conditioning on the permeate flux and cake properties. European Journal of Lipid Science and Technology, 106, 152 – 159. ISO standards methods (1985). Phosphorus and cations : ISO 6884. Jawad, I. M., Kochhar, S. P., & Hudson, B. J. F. (1983). Quality characteristics of physically refined soybean oil: Effects on component fatty acids and on triglycerides. Lebensmittel-Wissenschaft + Technologie, 17, 289 – 293. Kale, V., Katikaneni, S. P. R., & Cheryan, M. (1999). Deacidifying rice bran oil by solvent extraction and membrane technology. Journal of the American Oil Chemists’ Society, 76, 723 – 727. Kim, I., Kim, J., Lee, K., & Tak, T. (2002). Phospholipids separation (degumming) from crude vegetable oil by polyimide ultrafiltration membrane. Journal of Membrane Science, 205, 113 – 123. Kim, J., Lee, C., & Chang, I. (2001). Effect of pump shear on the performance of a crossflow membrane bioreactor. Water Research, 35, 2137 – 2144. Kiristakis, A., & Tsipeli, A. (1992). Relationship of the acidity of olive oil to its resistance to oxidation. Rivista Italiana Delle Sostanze Grasse, 69, 513 – 515.
212
A. Hafidi et al. / Innovative Food Science and Emerging Technologies 6 (2005) 203–212
Kochhar, S. P. (1983). Influence of processing on sterols of edible vegetable oils. Progress in Lipid Research, 22, 161 – 188. Koseoglu, S. S., Lawhon, J. T., & Lusas, E. W. (1990). Membrane processing of crude vegetable oils: pilot plant scale removal of solvent from oil miscellas. Journal of the American Oil Chemists’ Society, 67, 315 – 322. Koseoglu, S. S., Rhee, K. C., & Lusas, E. W. (1990). Membrane processing of crude vegetable oils: Laboratory scale membrane degumming, refining and bleaching. Proceedings of the World Conference on Edible Fats and Oils Processing: Basic Principles and Modern Practices (pp. 182 – 188). Champain7 American Oil Chemist’s Society. Largueze, C., Pioch, D., & Gulik-Kryzwicki, T. (2002). Structure of clusters of phospholipids and soaps during refining of vegetable oils. Meeting of the Euro Fed Lipids, Strasbourg 25–28 Nov. 2002. List, G. R., Monts, T. L., & Lanser, A. C. (1992). Factors promoting the formation of nonhydratable soybean phosphatides. Journal of the American Oil Chemists’ Society, 69, 443 – 467. McConnell, H. M., & Radhakrishnan, A. (2003). Condensed complexes of cholesterol and phospholipids. BBA-Biomembranes, 1610, 159 – 173. Moreau, R. A., Whitaker, B. D., & Hicks, K. B. (2002). Phytosterols, phytostanols, and their conjugates in foods: Structural diversity, quantitative analysis, and health-promoting uses. Progress in Lipid Research, 41, 457 – 500. Pasqualone, A., & Catalano, M. (2000). Free and total sterols in olive oils. Effects of neutralization. Grasas y Aceites, 51, 180 – 182. Plank, C., & Lorbeer, E. (1992). Quality control of vegetable oil methyl esters used as diesel fuel substitutes: Quantitative determination of mono-, di, and triglycerides by capillary GC. Journal of High Resolution Chromatography, 15, 609 – 612. Pioch, D., Largueze, C., Graille, J., Ajana, H., & Rouviere, J. (1998). Towards an efficient membrane based process for vegetable oils refining. Industrial Crops and Products, 7, 83 – 89. Pioch, D., Largueze, C., Graille, J., Hafidi, A., Ajana, H., & Graille, J. (1996). Membrane based vegetable oil refining process. Tangential
filtration at laboratory and pilot scales. World Conference and Exhibition on Oilseed and Edible Oils Processing, Istambul, Turkey, October 6–10th. Raman, L. P., Cheryan, M., & Rajagopalan, N. (1996). Deacidification of soybean oil by membrane technology. Journal of the American Oil Chemists’ Society, 73, 219 – 224. Richtering, W. (2001). Rheology and shear induced structures in surfactant solutions. Current Opinion in Colloid & Interface Science, 6, 446 – 450. Schulte, E. (1995). Determination of edible fat refining by HPLC of 3,5steradienes. Fat Science and Technology, 96, 124 – 128. Segers, J. C. (1979). Oil purification by adding hydratable phosphatides. U.S. Patent 4,162,260. Subramanian, R., & Nakajima, M. (1997). Membrane degumming of crude soybean and rapeseed oils. Journal of the American Oil Chemists’ Society, 74, 971 – 975. Subramanian, R., Nakajima, M., Yasui, A., Nabetani, H., Kimura, T., & Maekawa, T. (1999). Evaluation of surfactant aided degumming of vegetable oils by membrane technology. Journal of the American Oil Chemists’ Society, 76, 1247 – 1253. Trautwein, E. A., Duchateau, G., Lin, Y., Mel’nikov, S. M., Molhuizen, H. O. F., & Ntanios, F. Y. (2003). Proposed mechanisms of cholesterollowering action of plant sterols. European Journal of Lipid Science and Technology, 105, 171 – 185. Verleyen, T., Sosinska, U., Ioannidou, S., Verhe, R., Dewettinck, K., Huyghebaert, A., et al. (2002). Influence of the vegetable oil refining process on free and esterified sterols. Journal of the American Oil Chemists’ Society, 79, 947 – 953. White, L. S., & Nitsch, A. R. (2000). Solvent recovery from lube oil filtrates with a polyimide membrane. Journal of Membrane Science, 179, 227 – 267. Zwijnenberg, H. J., Krosse, A. M., Ebert, K., Peinemeann, K. -V., & Cuperus, F. P. (1999). Acetone-stable nanofiltration membranes in deacidifying vegetable oil. Journal of the American Oil Chemists’ Society, 76, 83 – 87.