Degumming of vegetable oil by membrane technology

Indian Journal of Chemical Technology Yol. 9, November 2002, pp. 529-534

Articles

Degumming of vegetable oil by membrane technology

N C Desai^a*, M H Mehta^b, A M Davee & J N Mehta^C

"University Department of Chemistry, Bhavnagar University, Bhavnagar 364 002, India bYice Chancellor, Gujarat Agriculture University, Sardar Krishinagar 385 506, India

cGSFC Science Foundation, Fertilizer Nagar 391 750, Vadodara, India Received 14 May 2001; revised received 7 May 2002; accepted 7 September 2002

Conventional method of degumming of crude vegetable oil, which involves the treatment of oil with steam followed by acid, is associated with many drawbacks, such as loss of oil, generation of wash waters, poor quality of degummed oil, etc. These drawbacks can be eliminated by the use of membrane technology being energy efficient, ecofriendly, and simple in operation. In order to explore the possibilities of utilizing membrane ba!ted separation process for the degumming of crude vegetable oil, this study has been conducted. The findings of the study indicate that membranes can remove phospholipids approximately 90% and above from castor oil, salicorniaseed oil and cottonseed oil. The permeability of membrane depends on the pressure applied. Ultrafiltration membranes exhibit higher permeability comparative to nanofiltration membrane. In addition to removal of phospholipids, membranes can simultaneously remove colour approximately 80% from the oil.

Crude vegetable oils contain phospholipids and other impurities such as free fatty acids (FFAs), colour, sterols, diglycerides, etc. in minor proportion.

Chemically phospholipids are glycerides esterified at I, 2 position by fatty acids and at 3 position by phosphoric acid residue. Phospholipids content of oils vary widely. Peanut oil contains -0.1 % phospholipids while soybean oil contains -3.0% phospholipids.

Phospholipids precipitate during storage of oil and impart unpleasant odour to the oil, which renders the oil unsafe for edible use. Therefore phospholipids and other impurities of oil such as FFAs, pigments, odorous compounds, etc are removed from the oil during refining to make oil safe for cooking and to increase its storabilityl. Conventionally phospholipids are removed from oils by hydration of oil (water degumming) followed by acid treatment of oil (acid degumming). This method of phospholipids removal is associated with some drawbacks, such as loss of oil, generation of acidic wash waters, etc²•3

. Moreover, phospholipids content of the oil cannot be brought down to the level required for refining the oil by physical method.

In the recent years, membrane technology has been widely accepted in various chemical process industries to supplement or replace conventional separation processes. The use of membrane

*For correspondence (E-mail: dnisheeth@indiatimes.com;

Fax: 0278-426706)

technology for refining crude vegetable oils has been documented recently4.8. Koseoglu4.⁶ reviewed applications of membrane technology in oil processing and also conducted a pilot plant study for degumming of crude vegetable oil by membrane technology. Bhowmick et al.⁷ has proposed degumming of rice bran oil using ceramic membrane.

Addition of surfactant in miscellas has also been proposed to enhance the separation of phospholipids by membrane process⁹. Kale et al.⁸ has done deacidification of vegetable oils by extraction of FFAs in solvent and then treating with membranes to separate FFAs from solvent. These studies indicate the potential of membrane technology for the processing of fatty oils because membrane based separation processes are simple in operation, ecofriendly and energy efficient. In the present communication membrane process is reported for the degumming of castor oil, cottonseed oil and salicomiaseed oil.

Experimental Procedure

Vegetable oils

Samples were collected from local suppliers which were filtered through microfilter (lOO)..lm) and were analyzed before use for phospholipids content in accordance with American Oil Chemists' Society (AOCS) method no. Ca-12-55⁹. Analysis of oils is given in Table 1.

Articles

So/vellts

Methanol and hexane solvents used were of laboratory grade and obtained from Mis S.D. Fine Chemicals Ltd, India.

Membranes

Two of the membranes i.e. Ultrafiltration membrane, UF, & UF" were obtained from CSMCRI, Bhavnagar. Other two membranes UF", &

Nanofiltration membrane-NF were obtained from Pennionics Ltd, Baroda. UF, & UF" membranes were prepared from polysulfone polymer (udel 3500) having molecular weight cutoff (MWCO) values 2,50.000 & 1,00,000 respectively, whereas UF", membrane was polyethersulfone having MWCO 10,000 and NF membrane was cellulose triacetate membrane having MWCO 1,000. MWCO value of the membrane has been reported as mentioned by suppliers. The membranes were washed with D M water before use to remove any stabilizer or preservative. The details of the membrane are shown in Table 2.

Miscellas Preparation

Oil miscellas were prepared by mIXIng oil in solvent under constant stirring in ratio of 25:75 (v/v).

Since castor oil is not miscible with hexane, it was dissolved in methanol in the same concentration.

Miscellas was vigorously stirred for 30 min and filtered through common filter paper to remove suspended particles. The temperature was maintained at 25°C to avoid evaporation of sol vent.

Permeation studies

Permeation experiments were carried out in UF test cell model No. 402 supplied by Amicon, USA. The

Table I- Analysis of seed oils

Seed Phosphol i pids Colour FFA

oils ppm index %

Castor 100 9.64 0.78

Cottonseed 440 377.80 3.30

Salicornia 250 11.l9 3.12

Indian J. Chem. Techno!.. November 2()02

test cell was the stirred UF cell, 7 cm diameterx 15 cm long, capable of withstanding pressure up to 75 psi. The cell capacity was 100 to 500 mL, depending upon the fittings used. A llatsheet membrane with a diameter of 4.3 cm (effective membrane area of 13.4 cm") was placed at the bottom of the vessel and was supported by a porous disc mounted on circular base plate. The plate in turn was attached to the test cell by coupling. The air cylinder was connected to the top

or

the test cell. A conduit was also provided in the bottom plate of the test cell to collect the permeate. Turbulence to minimize concentration polarization was created by magnetic stirrer bar turning just above the membrane.

Experiments were conducted with chargi ng of 200 mL crude oil miscellas into the ctll. The pressure.

temperature and rotation of the magnetic spi n bar were maintained at 50 psi, 25°C and 1,000 rpm respecti vely. The experiment was stopped when the permeate collection was approximately 100 mL.

Batch experiments with all the four membranes were conducted with these oil miscellas at the same operating conditions.

Analysis

Phospholipids content of feed and permeates oil were determined in accordance with AOCS method No.Ca-12-55^9. The surface topography of membrane samples before and after miscella processing was examined under Scanning Electron Microscopy (SEM, Leo-440i model, Leoelectrons Microscopy Ltd, UK). The samples were sputter coated with gold.

Results and Discussion

In a membrane based separation process, the performance of the membrane is explained in terms of permeability of membrane and rejection of solute by the membrane. Permeability or flux indicates the productivity of membrane. The permeability of membrane is reported in terms of volume of permeate passes through per area of membrane per time [either in gallons per square feet of membrane area per day (gfd) or liter per square meter of membrane area per hour (Lm-²h-')]. In the present study flux has been

Table 2-Specification of membranes used

Membrane Membrane Membrane type Source MWCO

No. identity

I UF1 Polysulfone, asymmetric CSMCRI 2,50,000

2 UF_II Polysulfone, asymmetric CSMCRI 1,00,000

3 UFIII Polyethersulfone, asymmetric Permionics 10,000 4 NF Cellulose triacelate, asymmetric Permionics 1,000

Desai ('101.: Degulllming of vegetahle oil by membrane technology

-+-UF1(MeOH) --I\\--. UF1(Hex.) ___ UF2(MeOH)

-.\" UF2(Hex.)

-e-UF3(MeOH) -UF3(Hex.)

--+-NF(MeOH) --NI'(Hex.)

~:1

,.t:;

S

...

...J 100

.s

] 75

~

50

I

Articles

2: L--,"'+!:::: .

o

Pr(>SSure in psi

Fig. I-Effect of transmembrane pressure 011 flux of pure solvents at 30 Illin

- + -UFl (Mt'OH-misc_) ISO

I

I -,.----UFI (HexrmiSc.)

--/lUFll (MeOH-misc_l 125 _ _ UFU (Hex.-misc.)

- - - UFill(Mt'OH-misc_) 100

..t:: p.

~UFill (Hex.-nusc.) E ^{,0 . /}

--... ^{/ '/ ell}

.,' NF (MeOH-misc.) ^...l 75

~

-+-NF(Hex.-nlisc.) .S )( ..,.-:r_ ,,'

ri: ::s SO ^/l"

~

... ^---

25

.. . .. -' ^_0"

~-.. -_.--_ ....

0

• •

0 25 50

Pressure in psi

Fig. 2-Effect of transmembrane pressure on flux of oil miscellas at 30 min

reported as Lm-²h-'. Rejection of solute indicates the efficiency of membrane process. It is reported as % rejection of the solute.

Cfeed - Cpermeate

% Rejection = xl 00 Cfeed

Membrane permeability

Permeability of different membranes for pure solvents and for oil miscellas was determined at 25 psi and 50 psi pressures. The results have been

presented in Figs 1 & 2. It is seen that flux of ultrafiltration (UF) membranes is higher than nanofiltration (NF) membrane at same pressure. It is due to the fact that UF membranes have larger pores than NF membrane. Larger pores permeate with higher rate; hence flux of UF membranes is higher.

Among UF membranes, membrane UF" with higher MWCO value (MWCO value for UF, is -2,50,000) has higher flux than the UF membrane UF_III, with low MWCO value (MWCO value for UF_III is -10,000).

Further, membrane flux also depends on the pressure applied. Therefore, flux values at 50 psi pressure are

Articles

-+-UFI gteOH) --UFI exane) --.\- URI WteOH) 250 -,:-. UFII exan1.

_w_UFIIl=H

---4--UFIIJ exane) 200

-

- I -NF fMeOH)

- N F Hexane) _..s::

-.

S

-..

--'---w. ..

^{--- - --. . • . ---.a}

;-:-.•...

---

..E

"0 ••••• _.J, •.•. -.. _-. " " - ' "

'.~'

~

50

'~: _!

+- - - - ! - - - ,

.• j-

0 ^I

-

30 60 90 120

Time in Min.

Fig. 3-Effect of time on the flux of pure solvents at 50 psi

- - UFl (Mi!OH.m&c.) _ _ UFI (He-x..-msc..) _ UFIl (MeOH-Ullsc.)

.-.:,' ... UFIl (Hec- misc.) ... )l .. UFIIl (MeOH-miSc.) - - UFIIl (Hex..-misc.) - t -NF (MeOH-trisc.)

- N F (Hec..mi8c.) .c:

...

S

.,J

.5

)( :r

~ 160

140

120 100

so

60

40 20

0

L_ ---- ^{--- -11- ----_. _ - ---}

... ,:' ... _., ...

30 60 90 120

Time in MiD..

Fig. 4-Effect of time on the flux of oil miscellas at 50 psi

higher than the flux values at pressure of 25 psi for all membranes. The effect of pressure on the flux of membranes is more pronounced in case of UF membranes. The flux of the membranes depends not only on the applied pressure, type of the membrane and its MWCO, but it also depends on the nature and solvent type, hence flux of methanol is higher than hexane.

The permeability of membranes was also determined at different times i.e. at 30, 60, 90 and

120 min to study the concentration polarization effect (Figs 3 & 4). It is seen from the results that the flux of membranes for pure solvents as well as for oil miscellas declines with time, however, the decrease in flux is marginal. Concentration polarization phenomenon is of reasonable importance in ultrafiltration and other membrane processes; because it adversely affects the flux of the membrane. During the present study no appreciable decrease in flux of the membrane has been observed up to 120 min. The

Desai et al.: Degumming of vegetable oil by membrane technology

Fig. 5-Electron micrographs showing surface topography of polysulfone membrane (a) before oil permeation (335x) (b) after oil permeation (376x) (c) sample (b) at higher magnification ( 1.05Kx).

membranes were studied by Scanning Electron Microscopy (SEM) before and after permeation experiment to observe solute deposition on the membrane, if any. Electron micrographs are given in

Articles

Table 3-Rejection patterns of phospholipids as a function of membrane type with various vegetable oils

Vegetable oil Phospholipids content. ppm Membrane Permeate Feed Rejection.

identity %

Castor UF_I 057 lOO 39.0

UF_II 061 100 41.0

UF_III 005 100 95.0

NF 006 lOO 94.0

Cottonseed UF_I 230 440 41.0

UF_II 250 440 48.0

UF_III 024 440 94.5

NF 044 440 90.0

Salicornia UF_I 100 250 40.0

UF_II 110 250 44.0

UF_III 012 250 95.0

NF 025 250 90.0

Fig. 5(a), (b) and (c). It is observed that topography of the membrane before and after permeation experiment is almost similar and there is no indication of solute deposition. These observations indicate that concentration polarization is a long-term phenomenon. It is necessary to conduct the study for a long time, giving sufficient time to build up solute concentration on the membrane side as to precipitate and deposit on the membrane surface.

Separation of phospholipids from oil

The membranes employed in the study primarily exhibit distinctly different pore sizes basecl upon their MWCO values. Rejections of phospholipids from three vegetable oils (castor, cottonseed ancl salicornia oil) were obtained with respect to the membrane types and the results are given in Table 3. In general, the phospholipids rejection increases with the decreasing MWCO values of the membranes. It is noteworthy that when the pore-size is reduced by a factor of 0.4 from UF1 to UFII, the rejection is marginally increased by a factor of 1.05. However, for a reduction factor of 0.1 from UFII to UFIII. there is a sharp increase in the rejection by a factor of 2.32. These observations, as made in case of castor oil, are similarly applicable to the other two cases of cottonseed ancl salicornia oils.

The phospholipids rejection may also be influenced by the formation of reverse miscelles as reported by Koseoglu⁵, on account of their possessing both hydrophilic and hydrophobic ends. In a solvent like hexane, the phospholipid molecule tends to orient with hydrophobic ends facing outside. This results in likely enlargement of the molecules having relatively

Articles

~ UFJ-C.astor oil

. . ·9· . . UFI-Cs.oil

- - - 6 -UFJI.Castor oil - - - UFIl-<..s.oil - * -UFill-Castor oil --+--UFIII-Cs.oil --+._.-NF-Ca s tor oil

- - t . -NF-Cs.oil

100

1

'Cf< 90

...

i _.~ ^~l

60

~

"0 50~

t :l ⁱ

~

10

Indian J. Chem. Technol .. November 20n:?

•

o

25 50 75

Miscella recovery (%)

Fig. 6- Phospholipids rejection by membrane at different percent recovery of miscclla

larger size prior to separation. Thus, the ultimate results are intluenced by MWCO as well as the so- called reverse miscelle formed in the feed. UFI and UF" membranes have shown in relatively lower degree of separation indicating that most of reverse miscelles formed in the system were smaller than that of the pore size. The UF"I and NF membranes have shown >95% separation. though molecular size of phospholipids molecules is slightly larger than that of triglycerides. Phospholipids rejection by the membranes at different recovery levels of miscellas is shown in Fig. 6. It is evident from the results that rejection of phospholipids by membranes is marginally increasing at all stages of recovery. It indicates that concentration of phospholipids in the feed has practically a little effect on the performance of membranes within the range of experiments carried out.

Conclusion

An attempt has been made to separate phospholipids from vegetable oils by membrane technology. The process, commonly known as degumming, facilitates removal of the phospholipids to increase quality and stability of such oils. The membrane based separation is found to give advantages like simultaneous removal of phospholipids (-90%) and colour (-80%) at room temperature without adding additional reagents like acids. This was a laboratory scale study for the purpose of evaluating membrane feasibility for the conceptual separations. Pilot scale study will be

needed on the long-term stability of the permeated oils, membrane performance and process economics.

Acknowledgements

The authors thank GSFC Science Foundation, Vadodara for supporting the project. Mr. Kripal Singh for valuable suggestions. Institute of Plasma Research, Ganclhinagar, for SEM work and Central Salt & Marine Chemical Research Institute, Bhavnagar, for membrane related experimental work.

References

Applewhite T H. EIICI'l'Iopeili(l of' Cliellli('({1 Tecll/wlogl·. Vol 9. 3rd ed .. ediled by Othillcr Kirk ( JOl1l1 Wiley & Sons Inc.

New York). 19RO, 795.

2 Hodgson A S, Bailel' 's !lIiIlIs/rial Oilll/l{! Fa/ Pmdll(,/.\·. Vol 4, 5th ed, edited by Hui Y H. (John Wiley & Sons Inc. New York), 1996. 172.

3 Raman L P, Cheryan M & Rajagopalan N, .I Alii Oil CIiI'III Soc. 73 (1996) 219.

4 Koseoglu S S & Engelgall D E. .I Alii Oil Clielll SO('. 67 (1990) 239.

5 Koseoglu S S, Lawhon J T & Lucas E W . .I Alii Oil Che/ll Soc, 67 (1990) 315.

6 Koseoglll S S, Rhee K C & Lin L, J Melilbralle Sci. 134 (1997) 101.

7 Bhowmick D N & Subramaniyam . .I Oil Te('//lwl Ass(J('

!lIdia, 31 (1999) 193.

8 Kale V, Katikneni S P R & Cheryan M. J AIIl Oi! Chelll SOl'.

76 (1999) 723.

9 Subramanian R, Nakajima M, Yaslii A, Nabetani H. Kimura

T & Maekawa T, J Alii Oil Gelll Soc, 76 (1999) 1247.

10 Official methods and Recommended practices of tht:

American Oil Chemists' Society, 4th ed. (AOCS. Champaign, Illinois, USA) 1996.