MEMBRANE TECHNOLOGY IN FOOD PROCESSING

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
501	1510	2020	50%
501	1510	2000	50%

Progress 10/01/01 to 09/30/06

Outputs
The COPE (corn oil and protein extraction) process developed in our laboratory is aimed at dry-grind ethanol plants as a method for producing at least three high-value coproducts (zein, corn oil and xanthophylls). This technology involves a combination of solvent extraction with ethanol and membrane technology. The basic process consists of (1) using in-house ethanol (e.g., produced in the dry-grind ethanol plant) to extract the zein and/or oil and/or xanthophylls in dry-milled corn, (2) membrane technology to separate the desired products, and (3) recycle of the ethanol (after further clean-up with a membrane if necessary) for further use in the process. The process was optimized so that traditional products are still marketable (e.g., DDGS residue and ethanol). The COPE-zein is comparable in amino acid analysis and SDS-PAGE profiles to commercial samples. Oil extraction was optimum with 95-100% ethanol at temperatures above 50C. A three-stage batch extraction or a single-stage column extraction could result in >93% yield of oil which is better than dry-determination milling or corn wet milling. An HPLC method for analysis of oil in ethanol miscella was developed. DDGS is not a suitable feedstock for the COPE process since oil quality was poor. On the other hand, corn-based COPE-oil has many unique health-promoting components in it due to the nature of the solvent used (ethanol instead of hexane) thus making it a more valuable product than conventional corn oil. The xanthophylls could be extracted with aqueous ethanol concentrations of 60% to 100%, although the optimum was 85% ethanol. Over 30 polymeric membranes and several ceramic membranes were evaluated, covering a wide range of chemistries and configurations. Conditioning the membrane was important for its long-term stability and performance. Many membranes that gave satisfactory flux and rejection in the first 1-7 days failed when tested over longer periods up to 10 weeks. After exhaustive testing with model and real zein, oil and xanthophylls extracts, we identified five polymeric membranes as being suitable for the COPE process. With co-funding from the Illinois Corn Marketing Board, Illinois Department of Commerce and Community Affairs and three private companies, two semi-works trials of the COPE process were conducted involving 5,000 lbs. and 11,000 lbs. of raw dry-ground corn. Additional trials on a bushel-scale were conducted over several months at a USDA Laboratory under a CRADA arrangement. These trials were used to obtain economic and scale-up data as well as to produce sufficient samples for evaluation. Our research in the COPE process has resulted in six graduate students theses (four MS and two Ph.D.), work by three post-doctoral associates, one full-time and three part-time technical assistants, three US patents and three patent applications, more than 25 peer-reviewed papers and abstracts, and a commercial licensing arrangement for the manufacture of zein and oil.

Impacts
Incorporation of the COPE process in a typical dry-grind plant producing 50 million gallons of ethanol per year will result in additional annual revenues of $40-60 million from zein, $15-20 million from oil and $20-40 million from xanthophylls. This will improve the profitability and economic viability of dry-grind ethanol plants and reduce the need for subsidies and tax waivers.

Publications

Cheryan, M. 2006. Method and system for extraction of zein from corn. U.S. Patent 7,045,607.
Cheryan, M. 2006. Corn oil and dextrose extraction apparatus and method. U.S. Patent Application No. 11/197,115. Pub. No. US2006/0029715 A1, Feb. 9, 2006.
Cheryan, M. 2006. Method and system for extraction of oil from corn. U.S. Patent Application No. 11/223,891. Pub. No. US2006/0063920 A1, March 23, 2006.
Cheryan, M. 2006. Method and system for corn fractionation. U.S. Patent Application No. 11/327,166. Pub. No. US2006/0173169 A1, Aug. 3, 2006.
Kwiatkowski, J.R. and Cheryan, M. 2005. Performance of nanofiltration membranes in ethanol. Sep. Sci. Technol. 40: 2651-2662.
Darnoko, D. and Cheryan, M. 2005. Carotenoids from red palm methyl esters by nanofiltration. JAOCS 83: 365-370.
Krishna Kumar, N.S., Cheryan, M. and Yea, M.K. 2004. Ultrafiltration of soy protein concen-trate: Performance and modelling of spiral wound and tubular polymeric modules. J. Membrane Sci. 244: 235-242.

Progress 01/01/05 to 12/31/05

Outputs
The recovery of additional co-products from the dry-grind ethanol process could impact the industry greatly, as most facilities today rely on subsidies and tax incentives. A process combining solvent extraction with membrane technology was developed to recover the oil. Several nanofiltration membranes were tested for flux and rejection of selected solutes in ethanol, as well as stability. The membranes were initially conditioned with pure solvent containing increasing concentrations of ethanol. Flux decreased with increase in ethanol concentration and increased at higher temperatures and pressures. The type of solute had an influence on membrane rejection profiles. The DK membrane from GE-Osmonics showed increasing rejection of polyethylene glycols (PEG) dissolved in ethanol from 29% at a molecular weight (MW) of 200 to 80% at MW 1,000. However, the MW of sugars and lipids had little or no effect on rejection with the DK membrane; their rejection averaged 87%. In contrast, the TFC-SR1 membrane from Koch showed higher rejections with higher MW compounds: lipid rejection increased from 19% to 71%, sugars from 35% to 85% and lipids from 77% to 89%. The TFC-SR2 membrane was much more open and showed lowest rejections of all these compounds. Flux generally showed opposite trends, with the DK showing the lowest flux and the SR2 the highest. Solutions of corn oil in ethanol and the corn extract yielded clear permeates when processed through these membranes. The retentate contained zein and other potentially high-value fractions soluble in ethanol, in addition to the corn oil.

Impacts
The data is sufficient for purposes of scaling up to pilot and semi-works scale. Apart from the additional revenue generated by corn oil (about $3 million for a 30 million gallon per year ethanol plant), other advantages of the COPE process are that low-energy and low-temperature separation and concentration methods are used, a substantial portion of the ethanol solvent is recycled and the corn oil contains high concentrations of nutraceuticals which are normally not extracted with hexane solvents.

Publications

Kwiatkowski, J.R. 2004. Membrane processing of ethanol-extracted corn oil. PhD thesis, University of Illinois, Urbana
Kwiatkowski, J.R. and Cheryan, M. 2005. Recovery of corn oil from ethanol extracts of ground corn using membrane technology. JAOCS. 82: 221-227.
Kwiatkowski, J.R. and Cheryan, M. 2005. Performance of nanofiltration membranes in ethanol. Sep. Sci. Technol. (In Press).
Cheryan, M. 2005. Membrane technology in the vegetable oil industry. Membrane Technology (International Newsletter, Elsevier). February: 5-7.

Progress 01/01/04 to 12/31/04

Outputs
A process has been developed for producing a higher-value coproduct (corn oil) in dry-grind ethanol plants using in-house materials only. The COPE (corn oil and protein extraction) process was optimized so that traditional products are still marketable (e.g., DDGS, CO2 and ethanol) while maximizing yield of oil. Factors that were tested for significance were temperature, concentration of ethanol in the extractant, time of extraction and solvent-to-solids ratio. Two potential feedstocks available in a dry-grind ethanol plant were studied: dry-ground whole corn and distillers dried grains with solubles (DDGS). With dry-ground whole corn, optimum conditions are a solvent-to-solids ratio of 4 ml/g corn, an ethanol concentration of 100%, 30 minutes of extraction time and a temperature of 50 degrees C. Under these conditions, a single batch extraction yields 3.2 g of oil per 100 g corn, which is an extraction efficiency of 81% of the oil in the corn. A 3-stage extraction, where the same corn is re-exposed to fresh ethanol in successive stages, results in a yield of 4.5 g oil/100 g corn, equivalent to an extraction efficiency of 95% and 2.5 lb. of oil per bushel of corn. This is better than current wet-milling technology that typically extracts 72% of the oil and dry de-germination methods that extract 65% of the oil. Our results show that pre-drying the corn is not necessary to extract the oil. Indeed, even 95% ethanol, which is cheaper to produce, was effective in extracting oil from whole corn. A single batch extraction with 95% ethanol, but with a higher solvent ratio of 8 ml/g corn, yielded 3.5 g oil/100 g corn (83% of the oil in the corn). As reported earlier, DDGS is not a good feedstock for ethanol extraction of the oil, since it contains too many undesirable by-products that are coextracted with the oil. The extract was dark brown in color and had off-odors. A method for the analysis of corn oil and free fatty acids in ethanol miscella was developed using gel permeation chromatography. Two Phenogel columns were connected in series to a refractive index detector and tetrahydrofuran was used as the mobile phase. Sample preparation involved only dilution with the mobile phase and filtration. Individual free fatty acids eluted as one peak. Reproducibility of the method was good with a relative standard deviation of less than 3.17%.

Impacts
The data obtained in this research is sufficient for purposes of scaling up to pilot and semi-works scale. Apart from the additional revenue generated by corn oil (about $3 million for a 30 million gallon per year ethanol plant), another advantage of the COPE process is that the product is corn oil and not germ, the extraction is accomplished with high yields using conventional oil extractors and there is no need to pre-wet or pre-dry the corn.

Publications

Kwiatkowski, J.R., Darnoko, D. and Cheryan, M. 2002. Analysis of corn oil in ethanol miscella by gel permeation chromatography. J. Liq. Chrom. and Rel. Technol. 25: 1379-1386.
Kwiatkowski, J.R. and Cheryan, M. 2002. Extraction of corn oil from ground corn using ethanol. JAOCS 79: 825-830.

Progress 01/01/03 to 12/31/03

Outputs
A process has been developed for producing higher-value products (corn oil and zein) in dry-grind ethanol plants using in-house materials only. The COPE (Corn Oil and Protein Extraction) process allows traditional products (e.g., DDGS, CO2 and ethanol) to be still marketed while also producing oil and zein. Conceptually, there are four stages in the COPE process: (1) Extraction of ground corn with ethanol under optimum conditions for oil or zein, (2) Separation of the ethanol extract from the residual meal (by filtration or centrifugation), (3) Desolventizing the meal and using it for conventional ethanol production, and (4) Concentration of the corn oil and/or zein, depending on process requirements, and recovering the ethanol for recycle using low-energy processes (e.g., membrane technology). Optimum conditions for zein extraction were 70% ethanol in water, extraction time of 30-40 minutes and temperature of 50C. The best yields (60% of the zein in corn) and highest zein content in the extracted solids (i.e., purity of the zein) were obtained at a solvent-to-solids ratio of 4-8 mL of 70% ethanol per gram corn. Zein concentration in the extract was higher at lower ratios. Multiple extraction of the same corn with fresh ethanol resulted in a zein yield of 85% after 4 extractions, while multiple extractions of fresh corn with the same ethanol resulted in high (15 g/L) zein concentration in the extract. Column extractions were best at 50C and 70% ethanol; a solvent ratio of 1 mL/g resulted in high zein concentrations in the extract (17 g/L) but yields were low (20%). Several commercial polymeric ultrafiltration membranes were screened for their performance with aqueous ethanol solutions. The method of conditioning the membranes had a major effect on solvent flux, membrane integrity and their pressure ratings. Gradual solvent exchange with successively higher concentrations increased in small doses appeared to work best with completely miscible solvents such as those studied here (ethanol-water mixtures). Rapid solvent exchange between water and high concentrations of alcohol disrupted the polymer matrix in many cases. Exposure to organic solvents significantly reduces the pressure rating of the membranes. Several membranes that provided acceptable rejection of ethanol-soluble proteins at low pressures (20 psi) lost its properties at higher pressures (60 psi) if conditioned incorrectly, and vice-versa. Zein purity in the ethanol extract is typically 40-60%. With membrane technology, a substantially purified zein containing 90% or more protein has been produced combining these technologies. A considerable amount of experimental work was required to determine the best pore size, the membrane and operating conditions. Additional work is needed to improve yield of zein and to get higher membrane flux.

Impacts
An efficient and potentially low-cost process for production of zein from dry-ground corn has been developed using membrane technology. For a 30 million gallon per year ethanol plant, the additional revenue generated by zein is about $11 million per year, an increase of about 25%. This should reduce the net cost of ethanol from corn.

Publications

Shukla, R., Cheryan, M. and Devor, R. 2000. Solvent extraction of zein from dry-milled corn. Cereal Chem. 77: 724-730.
Shukla, R. and Cheryan, M. 2001. Zein: The industrial protein from corn. Industrial Crops and Products. 13: 171-192.
Shukla, R. and Cheryan, M. 2002. Performance of ultrafiltration membranes in ethanol-water solutions: Effect of membrane conditioning. J. Membrane Sci. 198: 75-85.
Cheryan, M. 2002. Corn oil and protein extraction method. U.S. Patent 6,433,146.
Shukla, R. and Cheryan, M. 2003. Stability and performance of ultrafiltration membranes in aqueous ethanol. Sep. Sci. Technol. 38: 1533-1547.

Progress 01/01/02 to 12/31/02

Outputs
Two process improvements were evaluated in an operating corn wet mill for production of ethanol. The first was to study the use of membrane filtration for treating corn steep water. Flux increased with transmembrane pressure and became independent of pressure above 10 psi with flux being higher at higher cross-flow velocities. Average flux at 4X concentration factor over 24-hour operating cycles was 20 gallons per sq.ft. per day under optimum conditions. Capital cost of the membrane system is expected to be lower than current methods since it eliminates the heat pasteurizer and centrifuge, and reduces cooling requirements. Operating costs of the microfilter are about one-third of a heat pasteurization system and provides the opportunity to recover insoluble protein and starch for use in corn gluten feed. The second technology was a 7,000-liter continuous membrane recycle bioreactor (MRB). The MRB integrated ceramic microfiltration membranes in a semi-closed loop configuration with a stirred-tank fermentation vessel. Residence times of 7.5-10 hours with ethanol outputs of 10-11.5% (v/v) were obtained when the cell concentration was 60-100 g/L dry weight of yeast, equivalent to about 1-10 billion cells per mL. The performance of the membrane was dependent on the startup mode and pressure management techniques. A steady flux of 70 liters per square meter per hour (LMH) could be maintained for several days before cleaning was necessary. The capital cost of the MRB is $0.08-0.13 per gallon per year of ethanol capacity. Operating cost, including depreciation, energy, membrane replacement, maintenance, labor and cleaning, is $0.017-0.034 per gallon of ethanol.

Impacts
Microfiltered steep water could improve ethanol fermentation efficiency and reduce fouling of heat exchangers in the fermenters, beer still and steep evaporators. The benefits of the MRB include better productivity, a clear product stream containing no particulates or yeast cells which should improve subsequent stripping and distillation operations, and substantially reduced stillage handling.

Publications

Cheryan, M. 2000. Ultrafiltration. Encyclopedia of Separation Science. Edited by I.D. Wilson, E.R.Adlard, M.Cooke and C.F.Poole. Academic Press, NY. pp. 1797-1802.
Cheryan, M. 2000. Food technology. Membrane separations. Encyclopedia of Separation Science. Edited by I.D. Wilson, E.R.Adlard, M.Cooke and C.F.Poole. Academic Press, NY. pp. 2849-2855.
Escobar, J.M., Rane, K.D. and Cheryan, M. 2001. Ethanol production in a membrane bioreactor. Pilot-scale trials in a corn wet mill. Appl. Biochem. Biotechnol. 91-93: 283-296.
Rane, K.D. and Cheryan, M. 2001. Membrane filtration of corn steep water. Cereal Chem. 78: 400-404.
Bailly, M., Roux-De Balmann, H., Aimar, P., Lutin, F. and Cheryan, M. 2001. Production processes of fermented organic acids targeted around membrane operations: Design of the concentration step by conventional electrodialysis. J. Membrane Sci. 191: 129-142.

Progress 01/01/01 to 12/31/01

Outputs
An HPLC method for simultaneous identification and quantitation of tocopherols and carotenoids in red palm oil was developed using a C-30 column and a photo diode detector. Methyl tertiary-butyl ether/methanol/water was the mobile phase. Gradient elution resulted in complete separation of five tocopherols such as alpha-, delta-, gamma-tocopherols and tocotrienols in palm oil in less than 10 minutes. Thirteen carotenoids were also detected and quantitated in red palm oil within 45 minutes. A method was also developed for simultaneous analysis of transesterification reaction products such as triglycerides, diglycerides, monoglycerides, methyl esters and glycerol using gel permeation chromatography. Sample preparation involved only dilution and neutralization. Reproducibility of the method was good with a relative standard deviation of 0.27% - 3.87%. Methyl esters (biodiesel) were produced by transesterification of palm oil with methanol in the presence of a catalyst (KOH). The rate of transesterification in a batch reactor increased with temperature up to 60 degrees C. The reactions appeared to be second order up to 30 minutes of reaction time. Reaction rate constants of the hydrolysis reactions were 0.018-0.191 (1/wt%.min). Activation energies were 6.4-14.7 kcal/mole and the optimum catalyst concentration was 1% KOH. With continuous transesterification systems, at 60 degrees C and a methanol to oil molar ratio of 6:1, the yield of methyl esters increased from 58.8% of theoretical yield at a residence time of 40 minutes to 97.3% at a residence time of 60 minutes. However, higher residence times decreased the production rate. During long term continuous operation, the continuous bioreactor displayed steady state conditions in terms of product profile and methyl ester concentration.

Impacts
The continuous transesterification process developed here has good potential in the manufacture of biodiesel. An additional coproduct, glycerol, is also produced which will further enhance the economic feasibility.

Publications

DARNOKO, D., CHERYAN, M., MOROS, E., JERREL, J. and PERKINS,E.G. 2000. Simultaneous analysis of palm carotenoids and tocopherols using C-30 column and photodiode array detector. J. Liquid Chromatog. 23: 1873-1885.
DARNOKO, D., CHERYAN, M. and PERKINS, E.G. 2000. Analysis of vegetable oil transesterification products by gel permeation chromatography. J. Liquid Chromatog. 23: 2327-2335.
DARNOKO, D. and CHERYAN, M. 2000. Kinetics of palm oil transesterification in a batch reactor. JAOCS. 75: 1263-1267.
DARNOKO, D. and CHERYAN, M. 2000. Continuous production of palm methyl esters. JAOCS. 75: 1269-1272.
CHERYAN, M. 1999. Membrane Processing. Wiley Encyclopedia of Food Science and Technology, 2nd edition. Edited by F.Francis. John Wiley, West Sussex, UK. pp. 1627-1628.

Progress 01/01/00 to 12/31/00

Outputs
Crude rice bran oil containing 16.5% free fatty acids (FFA) was deacidified by treating it with methanol. At the optimum ratio of 2:1 methanol:oil by volume, the concentration of FFA in the crude rice bran oil was reduced to 5.6%. A second extraction reduced FFA to 0.33% of the oil. The FFA in the methanol extract was recovered by nanofiltration using commercial membranes. The DS-5 membrane from Osmonics/Desal and the BW-30 membrane from Dow/FilmTec gave average FFA rejection of 93-96% and an average flux of 41 liters per square meter per hour (LMH) to concentrate the FFA from 4. 69% to 20%. The permeate could be nanofiltered again to recover more FFA. Preliminary design calculations indicate a 2-stage system can result in a final retentate stream with 20% FFA or more and a permeate stream with negligible FFA (0.20%) that can be recycled for FFA extraction. Net recovery of FFA would be 96%. The capital cost of the membrane plant would be $5.8 per ton oil processed/year. Processing cost is $2 per ton oil processed and $15 per ton FFA recovered. The feasibility of extraction of crude oil from distillers dried grains with solubles (DDGS) of corn was studied using anhydrous ethanol as a solvent. Total solids, fat, glycerol and protein extracted from DDGS increased with increase in ethanol-to-DDGS ratio at a temperature of 50 degrees C and an extraction time of 30 minutes. However, ethanol-to-DDGS ratios above 6 ml ethanol per g DDGS did not significantly increase the amount of oil extracted, but did increase the glycerol. At the optimum ethanol-to-DDGS ratio of 6, about 50% of the crude corn oil can be extracted from DDGS in a single extraction step.

Impacts
The membrane process has several advantages: it does not require an alkali for neutralization, no soapstock nor waste water is produced and effluent discharges are minimized. The FFA recovered is worth over $5.62 million (assuming a FFA value of $550/ton). Since the annual operating cost is only $111,900, the value gained in this membrane plant is $50 per dollar of operating cost.

Publications

V.Kale, Katikaneni, S.P.R. and Cheryan, M. 1999. Deacidifying rice bran oil by solvent extraction and membrane technology. JAOCS. 76: 723-727.
Chukwu, U. and Cheryan, M. 1999. Electrodialysis of acetate fermentation broths. Applied Biochem. Biotechnol. 77-79: 485-499.

Progress 01/01/99 to 12/31/99

Outputs
Clarification of starch hydrolysates is the largest application today of membrane technology in corn wet milling. The effect of operating parameters on fouling of a ceramic microfiltration membrane by corn starch hydrolysate of 95 dextrose equivalence was studied. Transmembrane pressures above 100 kPa had little or no effect on flux. Cross-flow velocity had a significant beneficial effect. The rate of flux decline was reduced significantly when the feed was adjusted from its natural pH of 4.2 to 10. However. this resulted in a dark brown clarified syrup (permeate). Scanning electron microscopy showed extensive fouling layers on the alumina surface with conventionally processed dextrose solutions and the least fouling layer with corn starch hydrolysate adjusted to pH 10. Maximum steady state flux for unconcentrated hydrolysate at its natural pH was 178 liters per square meter per hour obtained at low transmembrane pressures (103 kPa) and high cross-flow velocity (5 m/s). Adjustment of the pH to 10 can increase the flux by 40%.

Impacts
Membrane microfiltration results in a dextrose syrup that is superior in terms of turbidity, color and microbiological quality compared to conventional precoat filtration. It also eliminates filter aid and associated disposal costs, and reduces downstream processing cots. An investment of $500,000 in membranes can save $2-3 million per year for a 500 gallon per minute dextrose line.

Publications

SINGH, N. and CHERYAN, M. 1997. Fouling of a ceramic microfiltration membrane by corn starch hydrolysate. J. Membrane Science. 135:195-202.
RAMAN, L.P., CHERYAN, M., and RAJAGOPALAN, N. 1997. Consider nanofiltration for membrane separations. In Practical Engineering Perspectives: Distillation and Other Industrial Separations, Edited by G.F. Nalven. Amer. Inst. Chem.Engrs., New York. pp. 202-208.
KAILASAPATHY, K., SINGH, N., and CHERYAN, M. 1997. Carrageenan-thickened milk base by ultrafiltration. Milchwissenschaft. 52:491-495
SINGH, N. and CHERYAN, M. 1997. Membrane applications in corn wet milling. Cereal Foods World.42:520-525
SINGH, N. and CHERYAN, M. 1998. Membrane technology in corn refining and bio-products processing. Starch/Starke. 50:16-23.
SINGH, N. and CHERYAN, M. 1998. Dextrose filtration using microfiltration membranes In Advances in Filtration and Separation Technology, Volume 12. Edited by G.C. Chase, R. Herrera and K.L. Rubow, American Filtration & Separations Society, Northport, AL. pp. 273-280.

Progress 10/01/97 to 09/30/98

Outputs
A process was developed for partial de-solventizing and de-acidification of hexane-extracted vegetable oils using nanofiltration membranes. A model miscella containing 20% soybean oil with 2% free fatty acids was concentrated to 45% oil at an average flux of 9 liters per square meter per hour (LMH) at 2.76 MPa and 24C. The permeate from this stage was re-concentrated 9-fold at an average flux of 20 LMH. The net recovery of oil was more than 99%, with a simultaneous 40% reduction of FFA and over 50% reduction in the energy required for evaporation of hexane. De-acidification can also be done on the crude oil by first extracting free fatty acids (FFA) with methanol, and then separating FFA from methanol with nanofiltration membranes. The best membranes resulted in FFA rejection of >90% and flux of >25 LMH. A combination of high-rejection and low-rejection membranes could be used to result in a retentate stream of 35% FFA and a permeate stream with less than 0.04% FFA which can be recycled back to the extractor. No alkali is required and no soapstock is formed. All streams within the membrane process are recycled and little is discharged as effluent.

Impacts
(N/A)

Publications

Raman, L.P.; Cheryan, M. and Rajagopalan, N. 1996. Solvent recovery and partial deacidification of vegetable oils by membrane technology. Fette Wissenschaft Technologie. 98: 10-14
Raman, L.P.; Cheryan, M. and Rajagopalan, N. 1996. Deacidification of soybean oil by membrane technology. JAOCS. 73: 219-224.
Cheryan,M. 1996. Membranes in vegetable oil processing: still a slippery road. Membrane Technology (International Newsletter, Elsevier). 79 : 2.
Witjitra, K.; Shah, M.M. and Cheryan, M. 1996. Effect of nutrient sources on growth and acetate production by Clostridium thermoaceticum. Enzyme & Microbial Technol.19: 322-327.
Shah, M.M.; Akanbi, F. and Cheryan,M. 1996. Potassium acetate by fermentation with Clostridium thermoaceticum.Applied Biochem. Biotechnol. 63-65: 423-433
Cheryan, M.; Parekh, S.; Shah, M.M. and Witjitra, K. 1997. Production of acetate by Clostridium thermoaceticum. In Advances in Applied Microbiology, Volume 3. Edited by S.L. Neidleman and A.I. Laskin. Academic Press, NY. pp.1-33.

Progress 01/01/97 to 12/31/97

Outputs
Cross-flow microfiltration is an effective tool for the reduction of bacteria and fat in milk. There was no significant difference in concentration of casein, whey proteins, fat and lactose between the permeate and retentate in either the co-current permeate flow (CPF) mode or the conventional mode of microfiltration. The fat content of the microfiltered milk averaged 0.05%, and a bacterial reduction of 4-5 log cycles was obtained. With 1% fat milk, the average flux over a 10X level of concentration was 900 liters per square meter per hour (LMH) in the CPF mode, and 400 LMH in the conventional mode of microfiltration. Vinegar (9-12% w/v acetic acid) was concentrated by electrodialysis (ED) to produce triple-strength vinegar (30-33% acetate). At a current density of 174 Amps/sq.m, acetate flux was 520 g/sq.m./h (GMH) for 9.1% acetate and increased to 927 GMH at 31% acetate. Increasing feed concentration decreased water flux from 2.6 to 1.3 LMH and energy consumption from 1.2 to 0.7 kWh/kg acetate transported. Increasing current increased acetate flux, energy consumption and water flux. A multi-stage ED system could produce triple-strength vinegar at much lower costs per unit of single-strength vinegar.

Impacts
(N/A)

Publications

Pafylias,I., Cheryan,M., Mehaia M.A. and Saglam,N. 1996. Microfiltration of milk with ceramic membranes. Food Research International. 29: 141-146
Chukwu,U. and Cheryan,M. 1996. Concentration of vinegar by electrodialysis. J. Food Science 61: 1223-1226

Progress 10/01/95 to 09/30/96

Outputs
Several nanofiltration and reverse osmosis membranes were screened for separating acetic acid from model solutions. Flux increased with pressure and temperature and decreased with pH and concentration of acetate. Rejection increased with pH, probably depending on the degree of dissociation of the acetate. At higher pH, acetate rejection could be correlated with NaCl rejection. In a model system at pH 5.6 and 50 degrees C, NTR729 membrane from Nitto-Denko Company had an average acetate rejection of 40%, glucose rejection of 99% and flux of 60 liters per square meter per hour (LMH) at 250 psi. This information was used to design a process to separate and partially purify acetate from fermentation broths. Actual acetate rejection was 60%, glucose rejection was 99% and flux was 15 LMH at 200 psi, 30 degrees C, pH 5.6. The best downstream strategy is to clarify the fermentation broth by microfiltration (MF), recycle the cells for improving fermenter productivity, nanofilter the cell-free broth and then evaporate the permeate. The cost of ceramic MF and NF-purification is about $66/ton acetate, compared to the cost of potassium acetate ($950/ton) or calcium-magnesium acetate ($700/ton).

Impacts
(N/A)

Publications

RAMAN, L.P., CHERYAN,M. AND RAJAGOPALAN,N. Consider nanofiltration for membrane separations. Chemical Engineering Progress, 90 (3): 68-74 (1994).
HAN,I.S. AND CHERYAN,M. Nanofiltration of model acetate solutions. Journal of Membrane Science, 107 : 107 - 113 (1995).
HAN, I.S. AND CHERYAN,M. Downstream processing of acetate fermentation broths bynanofiltration. Applied Biochemistry and Biotechnology, 57/58: 19-27 (1996).

Progress 10/01/94 to 09/30/95

Outputs
The technical and economic feasibility of processing stillage from corn-based ethanol plants by membrane technology was investigated. If the goal was to recover 87% of the stillage as clarified permeate for recycle to upstream operations, the microfilter would operate at a concentration factor of 7.5X, with an average flux of 100 liters per square meter per hour. The membrane area needed is 1750 square meters for a 100 million gallon per year plant, costing about $3.76 million. The net operating cost would be 1.2 cents per gallon of ethanol. The by-products of ethanol fermentation were separated using electrodialysis (ED) and precipitation. Eds was used to separate the organic acids (lactate and succinate) from glycerol. Flux of lactate was 100-150 grams per square meter per hour per ampere current (GMHA), and of succinic acid was 30 GMHA. Lactic acid is precipitated and crystallized as the zinc salt. The zinc lactate is then acidified to produce the lactic acid. Lactate recoveries were typically 85% or greater. Glycerol also appeared in the concentrating stream of the ED unit due to diffusion, necessitating multiple stage operation to maximize yield of glycerol and purity of the organic acids. Glycerol transport through the ED membrane was a function of feed concentration.

Impacts
(N/A)

Publications

Progress 10/01/93 to 09/30/94

Outputs
Electrodialysis of model lactic acid solutions at various concentrations, pHs and modes of operation was evaluated in a bench-top system. The limiting current (determined with 0.6% NaCl) was 152-304 amps per square meter at Reynolds Numbers of 176 - 419. In the batch recycle mode, the mass of lactate transported increased linearly with time. The flux of lactate was 328 - 456 grams lactate per square meter per hour. Flux decreased with an increase in feed concentration and with an increase of pH. Water flux during ED was 2.8 - 3.8 liters per square meter per hour; it decreased with an increase in feed concentration and was not affected by pH. This water transport limits the maximum lactate concentration in the concentrating stream to 13.5% with feed concentrations of 4.5-9.5%. Coulomb efficiencies with NaCl and lactate solutions in batch, single-pass and feed-and-bleed modes were 0.55-0.8.

Impacts
(N/A)

Publications

Progress 10/01/92 to 10/30/93

Outputs
Models have been derived to predict the minimum process time where purification of a macromolecule is done by ultrafiltration (UF) in the concentration mode followed by continuous diafiltration (CD). It requires only knowledge of the parameters of the film theory (gel-polarization) model for UF flux. When both UF and CD times are included in the optimization, the optimum value of C(subscript B)(superscript *) (the protein concentration to begin diafiltration) is much lower than when CD alone is considered. For example, with soybean water extracts, C(subscript B)(superscript *) for the combined process is 13% rather than the 18.6% obtained for CD only. Similarly, with cottage cheese whey, C(subscript B)(superscript *) for CD alone is 15.4% versus 10.2% for the combined process. The model could also be used to minimize the volume of permeate generated (e.g., in situations where the diafiltration solvent is limited in availability and/or where waste disposal must be kept to a minimum). For soybean extracts being purified from an initial 61.9% protein to 95% protein, the combined process is marginally better in terms of lowest permeate volume and lowest process time compared to DD. With cottage cheese whey, which required a much greater amount of purification to increase protein content from 6.6% to 95.5%, the combined process was clearly superior in minimizing both permeate volume and process time.

Impacts
(N/A)

Publications

Progress 10/01/91 to 09/30/92

Outputs
A bench-top electrodialysis (ED) unit with 20 cell pairs was used to separate lactic acid produced from fermentation of whey permeate by Lactobacillus bulgaricus. In the batch recycle mode, mass of lactate transported increased linearly with time, with an average flux of 550 grams lactate per square meter per hour. A small transport of sugars occurred (17-27 gm(superscript -2)h(superscript -1)). Lactose concentration increased in the concentrating stream, while galactose concentration decreased. Water flux during ED was about 4 Lm(superscript -2)h(superscript -1). Simulated electrodialysis of batch and continuous fermentations indicated good potential of this technique in recovering and concentrating lactic acid from a fermentation broth. The maximum concentration of lactate in the concentrating stream is limited by the current density and voltage, and by the simultaneous water transport for a given feed concentration. In our short-term experiments, the presence of microbial cells appeared to affect the transport properties, causing a decrease in lactate and water flux with increase in cell concentration. Higher lactate concentrations are possible in the concentrating stream if higher feed concentrations are used, although the coulomb efficiency decreases at higher feed concentration.

Impacts
(N/A)

Publications

Progress 10/01/90 to 09/30/91

Outputs
An empirical model was developed to describe the flux of a complex protein suspension under ultrafiltration and diafiltration conditions. Flux decreased during ultrafiltration, but increased during diafiltration as the permeable solids concentration decreased. The "gel" model based on film theory was modified to describe the flux in terms of both retained and permeable solutes. The least-time processing strategy for a given end product concentration was also modelled and identified for this protein system. In general, ultrafiltration followed by diafiltration is best if protein purification is the goal. The above model was tested with skimmilk, which was ultrafiltered by a combination of concentration and diafiltration to produce purified total protein isolates with the desired protein-lactose ratio. The rejection of protein and lactose followed expected trends. However, rejection of ash increased to 100% towards the end of the process, due to insolubility of salts or their binding to proteins. At the natural pH of skimmilk, the minimum ash content is 7-8% of the retentate solids, resulting in a limiting protein content of 90%. Thus this model and design concepts can be used to optimize a purification process to minimize the installed membrane area and/or process time.

Impacts
(N/A)

Publications

Progress 10/01/89 to 09/30/90

Outputs
The feasibility of using reverse osmosis to concentrate fermentation broths was studied in tubular and spiral-wound modules with thin-film composite and cellulose acetate membranes. Flux increased linearly with transmembrane pressure with both lactic acid model solutions and lactic acid fermentation broths, and was relatively unaffected by flow rate. Osmotic pressures of 1% lactate solutions were 280-560 kPa, depending on the pH or degree of dissociation. Rejections increased with applied pressure. With composite membranes, rejections of lactate and residual sugars were greater than 97% above pH 5.6. In contrast, with cellulose acetate membranes, flux was generally lower and lactic acid rejection was proportional to the degree of dissociation at lower pressures. This phenomenon has interesting possibilities for designing continuous bioreactors. A low-pressure cellulose acetate membrane scan be attached to the fermentor to remove lactic acid as it is formed, while recycling the sugars and other fermentation broth components. The dilute acid can then be concentrated using composite membranes operated in a cascade-recycle mode. A substantial amount of energy can be saved using reverse osmosis. Concentrating lactate from 10 g/1 to 120 g/1 by reverse osmosis would require about 13 kcal/1 of feed solution. In contrast, a state-of-the-art seven-effect evaporator would require about 140 kcal/1 feed solution, with about the same level of initial investment for both.

Impacts
(N/A)

Publications

Progress 10/01/88 to 09/30/89

Outputs
The effect of operating parameters (transmembrane pressure, flow rate, temperature and feed concentration) on the performance of spiral-wound and a tubular thin-film composite reverse osmosis (RO) membrane during the concentration of milk was studied. No permeation was observed until the applied transmembrane pressure exceeded the osmotic pressure of the feed (0.70 MPa). With all modules, flux increased linearly with applied transmembrane pressure up to 2.1-2.8 MPa. Flux then became asymptotic and decreased at much higher pressures. The pressure at which maximum flux was obtained was higher with higher flow rate. In comparison to cellulose acetate, the composite membrane gave higher flux and better rejection of salts and sugars. The difference in permeate quality was greater at higher concentration factors. Flux declined with feed concentration, as expected for an osmotic-pressure limited system. Under otherwise equivalent operating conditions, higher flux could be obtained with the tubular unit, but at the expense of higher energy consumption. Substantial energy savings are possible by incorporating RO in certain milk evaporation and dehydration plants, depending on the design of the systems. Even if only 25% of the condensed, evaporated or dried milk products were processed with RO, it would represent an annual savings of at least 300 billion BTU. The development of sanitary composite membrane modules should enhance this potential even further.

Impacts
(N/A)

Publications

Progress 10/01/87 to 09/30/88

Outputs
A process has been developed for the manufacture of khoa using reverse osmosis (RO) to preconcentrate the milk. Flux in spiral-wound cellulose acetate membranes was pressuredependent up to 27 kg/cm(superscript 2) and then became independent of pressure. No permeation was observed until a pressure of 6 kg/cm(superscript 2) was applied due to the osmotic pressure of milk. Flow rate affected flux only in the pressure-independent region. The average flux when concentrating cow's milk from 12.5% total solids to 31% total solids was 8-10 liters per square meter per hour at 30(degrees)C. Khoa manufactured from pre-concentrated (31% total solids) whole milk was typical in flavor and texture compared to khoa made in the traditional open-pan boiler. The mineral and ash contents of RO-khoa were slightly lower due to permeation of these compounds through the RO membrane. The overall economics and feasibility of a continuous khoa manufacturing operation based on reverse osmosis is attractive primarily because of the large savings in energy compared to traditional open-pan boilers. To pre-concentrate milk from 15% to 31% total solids, a mechanically-scraped open-pan kettle requires 455 kcal/kg of milk, while a double-effect vacuum evaporator consumed less than half that energy and a mechanical vapor recompression/single-effect evaporator requires less than one-third the energy. Reverse osmosis requires less than 10 kcal/kg milk, plus about 15 kcal/kg milk for HTST pasteurization.

Impacts
(N/A)

Publications

Progress 10/01/86 to 09/30/87

Outputs
The performance of an industrial hollow-fibre module processing skimmilk was studied. Flux displayed typical asymptotic behavior with increasing transmembrane pressure, indicative of concentration polarization. In the polarized pressure-independent region, flux was significantly affected by velocity and feed concentration; temperature had a small positive effect. The Leveque and Grober laminar flow models underpredicted the experimental flux. The model was modified to include a concentration-dependent diffusivity term and to account for non-Newtonian rheological behavior at high concentrations. The correlation that best fitted the data was Sh = 0.087 (RE)(Sc) where Sh is the Sherwood Number, Re is the Reynolds number and Sc is the Schmidt number. Milk was concentrated in a spiral-wound reverse osmosis system using cellulose acetate membranes. Flux was pressure-dependent up to 27 atm and then became independent of pressure. No permeation was observed until a pressure of 6 atm was applied due to the osmotic pressure of milk. Flow rate affected flux only in the pressure-independent region. The average flux was 8-10 liters/square meter/hour when concentrating whole milk up to 31% total solids. Energy for the reverse osmosis process was only 8-16 kcal/kg milk, while evaporation using MVR systems used 135 kcal/kg milk and a double effect evaporator used 209 cal/kg milk.

Impacts
(N/A)

Publications

Progress 01/01/86 to 09/30/86

Outputs
Flux in hollow fiber ultrafiltration of skimmilk was modelled over the entire pressureflux region using the resistance approach. The Hagen-Poiseuille model was adequate to describe water flux data: effect of temperature could be largely accounted for by viscosity of permeate. To describe skimmilk data, the model was modified to include a term due to resistance of the polarized layer (Rp) and Rf , a resistance due to fouling by specific membrane-solute interactions. Rf was relatively independent of operating parameters, but Rp was a function of velocity, temperature and feed concentration. The membrane and fouling layer appeared to contribute little resistance to flux compared with the polarizedlayer. Different industrial concentraion processes for fruit juices were reviewed. Compared to evaporation, freeze concentration is potentially a superior method. Reverse osmosis is probably the most energy eficient, although many flavor compounds appear to pass through most commercial membranes. Ultrafiltration has excellent potential for clarification and "cold pasteurization" of fruit juices. The feasibility of harvesting yeast cells using cross-flow membrane filtration was studied with three modules. Flux declined in an exponential manner with time. Lower fouling rates were observed at lower concentrations, lower transmembrane pressure and higher velovity. Media components played a relatively greater role in the fouling phenomenon at low cell concentrations.

Impacts
(N/A)

Publications

Progress 10/01/84 to 09/30/85

Outputs
The performance and energy consumption of pilot-scale spiral-wound and hollow fiber ultrafiltration modules processing skim milk and acid whey were evaluated. With cheese whey preacidified to pH 3 and prefiltered, a typical exponential flux decay was observed with both units. Flux was affected by pressure in the first 60 minutes of operation; for the spiral unit, flow rate had a beneficial effect only at pressures above 135 kPa. After 3-5 hours of operation, flux became independent of pressure for both modules; flow rate affected the flux of hollow fibers but not of spiral membranes. In contrast to cheese whey, both modules showed flow fouling tendencies with skim milk, despite the higher concentration of solids. Flux became independent of pressure at 60-150 kPa with hollow fibers and at 20-250 kPa with spiral-wound, reaching maximum flux of 40 and 23 L/sq.m/h respectively. The flux-Reynolds Number relationship indicated that while both units appear to operate in laminar flow, the spiral-wound benefited significantly from the additional turbulence induced by the spacers in the feed channel. In general, regardless of whether cheese whey or skim milk was processed, the spiral wound module required twice the membrane area and twice the pressure as the hollow fibers for a certain concentration duty, but needed less than half the recirculation energy and required less expenditure for membrane replacement.

Impacts
(N/A)

Publications

Progress 01/01/84 to 12/30/84

Outputs
Bovine casein micelles isolated from raw uncooled skim milk by ultracentrifugation at 50,000 x g for 60 min were treated with chymosin and pepsin in both soluble and immobilized forms. The release of nonprotein nitrogen and sialic acid soluble in 2% and 12% trichloro-acetic acid was studied. Soluble chymosin released a maximum amount of 7.56 mg nonprotein nitrogen per gram casein soluble in 2% trichloroacetic acid and 4.67 mg nonprotein nitrogen per gram casein soluble in 12% trichloroacetic acid. The corresponding figures for immobilized chymosin were 6.46 and 4.20. Pepsin generally released higher quantities of nonprotein nitrogen from the micelle. Both glyco-kappa-casein and carbohydrate-poor kappa-casein exist on the micelle surface with at least 90% of the micelle's glyco-kappa-casein being on its surface. Sialic acid analysis of trichloroacetic acid filtrates corroborated the nonprotein nitrogen data. Skimmilk was concentrated by ultrafiltration (UF) to volume concentration ratios (VCR) of 1.0, 1.3, 2.0, 3.0 and 4.0, corresponding to protein contents of 3.04-11.30 (% w/w, N x 6.38) and calcium concentration of 28-99.5 mM in the retentate.

Impacts
(N/A)

Publications

Progress 01/01/83 to 12/30/83

Outputs
The fouling of ultrafiltration membranes was studied. Acidifying pastuerized cottage cottage cheese whey to pH2 or 3 followed by conventional centrifugal clarification (e.g. 5000g for 20-30 sec) significantly improved flux over that of unadjusted whey. Neurtalizing whey to pH 7 was also just as effective provided resulting insoluble particles could be removed efficiently. Operating at 50C was slightly more beneficial than 40 or 30C. High flow rates were beneficial only if the transmembrane pressure was below some critical value (about 50 psig/335 kP in our case). At pressures higher than this, high flow rates significantly increased fouling rates. After prolonged operating time, neither flow rate nor pressure appeared to have a significant effect, due perhaps to a change in the mode of fouling from surface deposition to pore adsorption. Studies were conducted on the surface structure of casein micelles. Bovine casein micelles were isolated from raw uncooled skimmilk and racted with soluble and immobilized neuraminidase. Almost all of the sialic acid was released with soluble neuraminidase whereas only 88% was released with immobilized neuraminidase. It appears that most of the glyco-kappa-casein fraction of milk is on the micelle surface. The secondary phase of milk coagulation was studied using theimmobilized enzyme technique.

Impacts
(N/A)

Publications

Progress 01/01/82 to 12/30/82

Outputs
A process was developed for the manufacture of blends of dairy and vegetable proteins by co-extraction and co-ultrafiltration. Defatted soy flour proteins were extracted using either sweet or acid whey as the solvent. The whey/soy slurry was separated in a two-stage extraction process using a centrifugal desludger. The proportion of whey protein in the blend was determined by the number of extractions and the total amount of whey used. The combined extracts were processed by ultrafiltration to increase the protein content of the blends. Solubility characteristics of the blends indicated potentially good functional properties. The effect of four operating variables (enzyme concentration, substrate concentration, flow rate, reaction volume) on the performance of a continuous stirred tank reactor(CSTR)-hollow fiber membrane reactor was studied for the continuous hydrolysis of proteins (Promine-D) using the enzyme Pronase. Based on a residence time distribution study, the reactor system was modelled as an ideal CSTR in combination with Michaelis-Menten equation of enzyme kinetics. This kinetic model correlated conversion with a space time parameter modified to include all four independent variables. An empirical model based on curvilinear regression analysis was also developed. Both models predicted conversion fairly well, although the kinetic model slightly underpredicts at high conversion.

Impacts
(N/A)

Publications

Progress 01/01/81 to 12/30/81

Outputs
Water extracts of defatted soy flour was prepared by multiple extraction and centrifugation. Ultrafiltration of extracts in hollow-fiber units showed significant deviation from ideal behavior for removal of nonprotein components, probably due to solute-solute-and solute-membrane interactions. These interactions affected the retentate composition as well as the yield of protein. Membrane adsorption losses increased with feed concentration and were higher for the XM than for the PM series of membranes. A retentate washing step was necessary to produce a true protein isolate. Overall yield of protein was 86% of that in the defatted soy flour. Flux was significantly affected by pressure up to the limits of the unit. Higher temperatures also increased flux. However, flow rate had practically no effect and no hysteresis effects were observed upon lowering the pressure from the higherst to lowest values. The dat indicated that concentration polarization or mass transfer was not rate-controlling and flux behavior could be modelled by momentum transfer considerations. A modified power law version of the Poiseuille equation J = A (DeltaP(T) n, best fit the data, when J is the flux, DeltaP(T) is the transmembrane pressure and A and n are constant characteristic of a particular membrane-feed combination. A and n decreased with increasing solids in the feed, and A increased while n decreased with increasing temperature. Beta-D-Galactosidase was immobilized in a hollow fiber ultrafiltration module.

Impacts
(N/A)

Publications

Progress 01/01/80 to 12/30/80

Outputs
Functional properties of a full-fat soy protein produced by ultrafiltration (FU)of water extracts of whole soybeans was studied. Protein Dispensibility Index (PDI) was higher than that of the raw material and a commercial soy isolate in the acidic and neutral pH region. A significant difference in salting out at pH 6.7 was observed depending on the order of mixing of ingredients; protein dispersed after NaCl was dispersed showed much larger salting out effects than vice versa. At 0.01-0.15 moles calcium per liter, addition of calcium chloride resulted in PDI of 6-20%, while tricalcium phosphate addition resulted in PDI of 81-89%. Phytic acid had a significant effect on protein solubility in the acidic pH region and it may mask the true effects of low levels of Ca 1 + protein solubility in the acidic pH region and it may mask the true effects of low levels of Ca 2 + on solubility characteristics of soy products. Excellent emulsification activity and stability in the acidic and isoelectric pH regions was observed for the UF product when evaluated at 1.9% protein concentration and 50% oil in the emulsion. Viscosity of emulsions were significantly affected by pH, shear rate and cold storage. Brookfield viscosity at 6 RPM and 25 degrees C was 108 cp at pH 6.7 and 348 cp at pH 3.0, and much higher at pH 4.2 (26,975 cp) and pH 5.5 (45,650 cp). In contrast, Promine-D emulsions were more viscous at pH 6.7 (2156 cp) and pH 3.0 (2556 cp) but far less viscous at pH 4.2 (4555 cp) and pH 5.5 (6070 cp).

Impacts
(N/A)

Publications

Progress 01/01/79 to 12/30/79

Outputs
The mechanism of fouling of ultrafiltration membranes during processing of cottage cheese whey was studied. A mathematical model of the flux decline was developed and used to determine the contribution of individual proteins and the aqueous environment to fouling. In general, proteins dissolved in a salt-free environment resulted in a higher flux and slower rate of flux decline than those dissolved in whey dialysate systems, implicating salts as a major contributor to the fouling phenomenon. X-ray analysis of the membrane surface indicated adsorption or binding of salts by the membrane; this could result in a salt-bridge and enhance adsorption of proteins by the membrane, resulting in faster fouling. Alpha-lactoglobulin appeared to be the cause of initial fouling and the major contributor to the concentration polarization layer, while Beta-lactoglobulin seemed to be the major contributor to long-term fouling. The scanning and transmission electron microscopes revealed an assymetric ultra-structure on the membrane surface, with a membrane skin of 0.1 Mum and a support or spongy layer of 100-200 Mum. The ultrastructure of the protein fouling deposit on the membrane further supported the hypothesis of the effects of salts on the fouling phenomenon. Functional properties of a full-fat soy protein product produced by ultrafiltration were studied. Protein solubility was greater than 90% at acidic and neutral pH. Phytic acid, sodium chloride and calcium chloride greatly affected solubility characteristics.

Impacts
(N/A)

Publications

Progress 01/01/78 to 12/30/78

Outputs
A purified and functional protein-fat product was produced by ultrafiltration (UF) of water extracts of soybeans. Elimination of phytic acid did not follow expected behavior for a nonrejected solute that is freely permeable through the membrane, but was governed by the environment and state of binding to rejected species. UF to a volume concentration ratio (VCR) of 5 resulted in a 25% reduction of phytic acid at pH 2, and a 65%, 43% and 27% reduction at pH 6.7, 8 and 10 respectively, despite a removal of 80% of the solvent. Dilution at the same pH and reultrafiltration eliminated about 92% of the phytic acid at pH 6.7 and more than 80% at pH 8 and 10. Reultrafiltration had little effect on the phytate-protein complex at pH 2. During UF, protein, fat, ash and total solids concentration increased linearly with VCR, but some loss in yields was evident probably due to membrane adsorption. Binding of minerals to proteins prevented removal of more than 50-60% ash. Trypsin inhibitor activity of final product was reduced by 79%, mostly due to heat treatments during extraction. Protein Dispersibility Index of greater than 94% at pH 7 has been observed for the freeze-dried product, which typically assays, on a dry basis, 59% protein, 34% fat, 2.8% ash, and 0.064% phytic acid and 0.6% oligosaccharides. The electron microscope is being used to elucidate the mechanism of fouling of UF membranes. Preliminary findings are protein and salts are major causative factors of flux decline.

Impacts
(N/A)

Publications

Progress 01/01/77 to 12/30/77

Outputs
Performance characteristics of the soybean water extract-ultrafiltration (UF) system were optimized. Initially, flux increased linearly with transmembrane pressure and in this region, activation energy for both flux and viscosity were similar (3400 cal/mole), confirming the validity of the pore flow model. At pressures higher than 0.8-1.0 kg/cmy, flux approached an asymptotic value. The onset of pressure-independence occurred at lower pressures with lower flow rates (or lower Reynold's Numbers, Re) and lower temperatures (or higher Schmidt Number, Sc). Other data indicated concentration polarization and formation of a hydrodynamic gel on the membrane surface, which limited flux. Theoretical laminar flow mass transfer models were found to predict Sherwood Numbers (Sh) and flux lower than experimental values. The correlation that fit the data best was Sh=0.18 (Re)y47 (Sc)y3DT. A purified, full-fat, high-protein concentrate and flour was produced by UF of soybean water extracts. 83% of the protein, fat and ash was recovered during the extraction. A 5-fold volume reduction to a total solids of 12-14% could be achieved, with a 96% reduction in undesirable oligosaccharide concentration using a two-stage UF proccess. Membrane rejection of protein and fat is close to 100%. Trypsin inhibitor was reduced by 60% by a combination of blanching and UF. Excellent functional properties are envisaged since protein dispersibility index at pH 6.7 was 87%.

Impacts
(N/A)

Publications

Progress 01/01/76 to 12/30/76

Outputs
A new soybean product was prepared by ultrafiltration of aqueous soybean extracts, using methods reported previously. Initial permeate flux of 38-43 liters/my/hr (LMH) at alkaline pH could be increased to 60 LMH by operating at low pH. Flux dropped in a log-linear manner with volume concentration ratio (VCR), to 12-15 LMH at 14% solids. High viscosity of the concentrate prevented concentration to more than 17% solids. Spray drying of the concentrate resultedin a powder containing 57% protein, 32% fat. Carbohydrate analysis indicates that under optimum conditions, 70% of sucrose and stachyose and 78% of raffinosecan be removed by ultrafiltration to VCR 5. Diluting the concentrate and reultrafiltration to VCR 5 effected further purification. About 76% of the trypsin inhibitor was eliminated during processing and the protein dispersibility index was 87%. Organoleptic properties of the reconstituted powder are acceptable. Studies of phytic acid removal and functional propertiesof the product are in progress. Ultrafiltration of dairy systems was also studied. Skimmilk concentrated in a pilot scale ultrafiltration unit was used to make cottage cheese. A VCR of 3 was found optimum in terms of flux rates and curd handling characteristics. A firmer curd and increased production per cheese vat was obtained. The amount of whey from individual vats was reduced thereby reducing potential pollution problems.

Impacts
(N/A)

Publications

Progress 01/01/75 to 12/30/75

Outputs
Aqueous extracts of soybeans were prepared using optimum procedures for soluble protein recovery, as determined and reported previously. Preliminary laboratory-scale ultrafiltration (UF) trials were conducted, using a laboratory-scale Amicon apparatus, to investigate feasibility of producing stable soy emulsions reduced in concentrations of nutritionally undesirable compounds. Soy protein concentrates (SPC) containing 60% protein (dry basis) were prepared. Phytic acid concentration in this SPC was less than 30% of that present in the original soy extract. A Romicon pilot-scale hollow fiber UF unit(15 square feet of membrane) was obtained and three batches of purified soy SPC produced. Using extracts containing 3.4% TS, initial flux rates were typically 38-40 liters meter(-2) hour(-1). After removal of 80% of the volume as a protein-free permeate (0.7-0.9%TS), flux was typically reduced to 50% of original. Proximate analysis of this retentate indicated 13.5-14% TS, containing approximately 50-55% protein (DB). Diafiltration of retenate effected further purification. Analyses conducted to date on freeze dried permeate and retentate solids indicate that phytic acid can be reduced (dry basis) to approximately one-third initial value by removing 80% of soy extract volume as permeate and then washing retentate with a volume of water equaling volume of permeate removed.

Impacts
(N/A)

Publications

Progress 01/01/74 to 12/30/74

Outputs
Addition of anti-gelling agents (sodium sulfite, sodium hexametaphosphate) to extraction water was shown to have no significant effect on yield of crude protein in water extracts of soaked and unsoaked beans (Amsoy 71). Gelation of soy protein during filtration was not found to interfere with extractability of soy solids. The procedure giving greatest yields of solids and protein was to extract beans using a 1:10 bean:water (b:w) ratio and to reextract the resultingfilter cake so that final b:w was 1:20. Retention of extractable solids by the filter cake was due to absorption and adsorption, with the former predominant. Adsorption of extractable solids by the filter cake did not conform with the linear form of the BET equation. Spray dried soymilk was prepared from a 1:20 b:w extract. Processing included hydration, grinding, filtration, heating(310F for 28 sec), vacuum concentration to 16% solids (TS), and drying (pressure nozzle, inlet temperature 310F). Solubility of this powder met standards for finest grade non-fat dry milk. A flavor panel found soymilk flavor significantly improved by vacuum treatment of thermally processed extract. Homogenization of pre-dried fluid gave a preparation more susceptible to oxidation in storage at 110F, as evaluated by expert panelists.

Impacts
(N/A)

Publications

Progress 01/01/73 to 12/30/73

Outputs
The effect of soybean to water ratio on the recovery of total solids and proteinin water extracts of soybeans was studied. Extracts in each case were prepared by adding beans to 90 C H(2)O with proper ratio, holding for 3 minutes, grindingthe mixtures for 5 minutes in a Waring Blender and subsequently filtering the resultant slurry. Varying the bean to water ratio from 1:6 to 1:10 and to 1:20 resulted in recoveries of total solids in the extracts (filtrates) of 53.3%, 65.2% and 71.3% respectively; protein recoveries in the extracts were 59.1%, 70.5% and 77.1% respectively. Rinsing, washing and reextracting the filter cakes from 1:10 bean - water ratio preparations using an amount of water in eachcase necessary to obtain a final bean - H(2)O ratio of 1:20 resulted in protein recoveries of 81:5, 81:7 and 84.7% respectively. No significant differences were found in the compositions of the moisture free extracts. Addition of 0.01,0.1 and 0.2 g sodium sulfite to 350 ml of 90C extraction water before blending resulted in modest, but significant increases in the amount of total solids and protein recovered in the extracts. Sulfite could be detected organoleptically (decreasing the palatability) when recovery of proteins was the greatest.

Impacts
(N/A)

Publications

Progress 01/01/72 to 12/30/72

Outputs
Thermal inactivation of soybean trypsin inhibitor (STI) in water extracts of Amsoy soybeans was studied at 200F and under high temperature short time heat treatment (HTST) conditions in the temperature range of 250F-310F at pH 2.0, pH 6.5 and pH 7.5. With HTST, holding time of 0, 14, 28, 42, 56 sec were employed with 6 sec come up and 6 sec cooling time. When water extracts of soybeans were heated at pH 6.5, the temperature required to inactivate 90% STI appeared to be a log function of the holding time, with 60 min required at 200F, 56 sec at 290F and 23 sec at 310F. However, this relationship was not found to exist at other levels of inactivation nor when heating was done at pH 2or pH 7.5. The overall kinetics of thermal inactivation of STI is complicated by the heterogenous nature of STI, with different STI fractions having differentheat labilities depending also on pH of heating. However, 90% destruction of STI was accomplished with 21-24 sec hold at 310F whether heating was done at pH2, 6.5 or 7.5. From this work it would appear that at temperatures above 250F,more exposure time would be required to inactivate STI than would be required for adequate sterilization of the soy extracts. Urease activity in soy extractswas more easily destroyed by heat than the STI.

Impacts
(N/A)

Publications

Progress 01/01/71 to 12/30/71

Outputs
Soymilk free from the lipoxygenase induced painty, oxidized flavor was obtained 1) by extraction of the soybeans with water at 95C, 2) extraction with 25C water at pH 2, 3) and extraction with 25C water at pH 9.5. The efficiency of extraction was poorer with the 95C extraction than with the other two extraction methods. Using soaked dehulled soybeans and 95C extraction, approximately 66% of the total solids, 75% of the proteins and 77% of the lipidswere recovered in the soymilk. Soymilks prepared by extraction at 95C rated highest in flavor preference by an untrained panel over those prepared by the other extraction techniques. Flavoring materials, such as vanilla and chocolate, added to soymilk improved their acceptance ratings, but even with flavored milks, preference was given to soymilks prepared by the 95C extraction. A good correlation between the flavor of soymilk and the number of volatile components found in the soymilk was evident from CLC patterns of the CS(2) soluble volatiles. During the 5 min extraction period at 95C, approx 62%of the soybeans Trypsin Inhibitor (STI) was inactivated. Subsequent heating of the soymilk at pH 6.5 to 200F for 50 min or 250F for 3 min was sufficient to inactivate STI. STI inactivation by heat was found to be pH dependent.

Impacts
(N/A)

Publications

Progress 01/01/70 to 12/30/70

Outputs
Methods for assaying lipoxidase activity and trypsin inhibitor activity of soybeans and soymilk were developed. A study of changes in composition during maturation of 4 vegetable varieties of soybeans indicated that trypsin inhibitoractivity was present from the onset of bean development, but the lipoxidase activity in the bean was absent until thy bean contained approximately 150 mg. solids. With mature beans, optimum extractions of solids, lipid and protein were obtained by water extraction of ground beans at 65C - 75C. The % recovery in the extracts were approximately 80% of the solids, 93% of the protein and 83% of the lipid. Above 75C, the extraction efficiency was reduced, whether dry whole beans, soaked beans, or ground beans were wet milled by a standard procedure. An extraction temperature of 75C was not sufficient to prevent lipoxidase activity, so extracts possessed the typical painty flavor.Cold extraction at pH 2, effectively inhibited the development of the painty flavor and extraction yields at this pH were similar to pH 6.5.

Impacts
(N/A)

Publications