- Email: [email protected]
Preliminary formulation development for aqueous surfactant-based soybean oil extraction
Industrial Crops and Products 62 (2014) 140–146
Contents lists available at ScienceDirect
Industrial Crops and Products journal homepage: www.elsevier.com/locate/indcrop
Preliminary formulation development for aqueous surfactant-based soybean oil extraction Linh D. Do, Travis L. Stevens, Tohren C.G. Kibbey ∗ , David A. Sabatini School of Civil Engineering and Environmental Science, University of Oklahoma, Norman, OK 73019, United States
a r t i c l e
i n f o
Article history: Received 13 June 2014 Received in revised form 1 August 2014 Accepted 14 August 2014 Available online 7 September 2014 Keywords: Oilseed extraction Soybean oil Aqueous extraction Surfactant extraction Interfacial tension
a b s t r a c t Soybean oils are increasingly being used for a range of non-food applications, including production of biofuels and oleochemicals. While most soybean oil is produced by hexane-based extraction methods, concern about environmental and health effects from hexane extraction has led to increased interest in development of aqueous extraction methods. Among aqueous methods, surfactant-based aqueous extraction of vegetable oils has shown particular promise as an alternative to hexane-based extraction methods. The objectives of this work were to explore the use of surfactant-based methods for the extraction of soybean oils, and to test whether the use of mixed anionic–cationic and anionic–cationic–nonionic surfactant mixtures could successfully be used to reduce the salinity requirements for surfactant-based extraction. All three formulations tested were capable of producing ultra-low (<0.01 mN/m) interfacial tensions with soybean oil. One of the formulations, a four-component (three surfactant, one hydrotrope) mixture, was able to reduce the salinity requirement from 5% down to 0.75%. A range of experiments was conducted to better understand the factors influencing extraction yield for surfactant-based extraction of soybean oil. Extraction experiments were conducted with a single extended surfactant system which has been used previously for extraction of other oilseeds. Extraction yields as high as 88.6% were observed for the conditions tested. Extraction yield was strongly dependent on salinity, and was found to increase with increasing shaker agitation rate, decreasing solid to liquid ratio, and decreasing particle size. © 2014 Elsevier B.V. All rights reserved.
1. Introduction While soybean oil continues to be predominantly used in edible products, its use in non-food applications has grown rapidly over the past two decades, and now comprises a substantial fraction of all use. For example, it has been reported that 20% of soybean oil consumption in the United States in 2011 was due to non-food applications, up from only 4% a decade earlier (Gunstone, 2013; SoyStats, 2014). Much of this increase can be attributed to rapidly growing production of biodiesel and other soy-based biofuels. Data from the National Biodiesel Board indicates that biodiesel production has increased 100-fold over the past decade (Biodiesel.org, 2014). Another significant non-food use of soybean oil is the production of oleochemicals and biosurfactants (e.g., Qingyi et al., 2011). Because solvent extraction is the primary method used for extraction of vegetable oils, the vegetable oil extraction industry is a major source of volatile organic compound emissions (Rosenthal
∗ Corresponding author. Tel.: +1 405 325 0580; fax: +1 405 325 4217. E-mail address: [email protected] (T.C.G. Kibbey). http://dx.doi.org/10.1016/j.indcrop.2014.08.026 0926-6690/© 2014 Elsevier B.V. All rights reserved.
et al., 1996). The annual hexane emissions from soybean oil extraction processes have been reported to be 210–430 million liters in the U.S. (Rosenthal et al., 1996). The use of hexane for vegetable oil extraction has led to both increased health concerns, and increased environmental regulations. Exposure to hexane has been shown to cause peripheral nerve damage, and hexane is also a potentially hazardous explosive material (Wan and Wakelyn, 1997). In 2001, the U.S. Environmental Protection Agency (EPA) established regulations on hexane emission (EPA, 2001). With rapidly increasing vegetable oil production, it has become more challenging for oilseed extraction plants to meet these regulations (EPA, 2006). Thus, there is a pressing need for development of environmentallyfriendly and sustainable oil extraction technologies. Aqueous-based oil seed extraction methods have advantages over hexane-based extraction, since aqueous extraction can be employed for either dry or wet oilseeds/plants without extensive drying of the starting material. A recently developed alternate approach has employed aqueous surfactant systems to enhance seed oil extraction (Do et al., 2009; Do and Sabatini, 2010; Phan et al., 2010a; Kadioglu et al., 2011). Surfactant-assisted aqueous extraction makes use of interfacial tension reduction between the surfactant solution and the extractable oil, mobilizing the oil
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
and allowing it to pass through the disrupted oilseed cell matrix (Campbell and Glatz, 2009; Do and Sabatini, 2010). New classes of surfactants, known as extended-surfactants, are able to produce ultralow interfacial tensions with a wide range of vegetable oils, making them attractive for developing a versatile system for bio-oil extraction that can be used across widely different oilseeds or plants (Salager et al., 2005; Do et al., 2009; Phan et al., 2010a; Witthayapanyanon et al., 2010). Although recent preliminary studies of surfactant-assisted aqueous extraction have shown promising oil extraction results (e.g., 85% extraction efficiency with corn germ oil and over 90% with peanut and canola oils (Naksuk et al., 2009; Do and Sabatini, 2010; Kadioglu et al., 2011)), the fundamental processes controlling the oil release mechanisms have yet to be verified and modeled. To date, the amount of added salt required to achieve good extraction efficiency in aqueous surfactant-based extractions has been relatively high (e.g., 5–10% by weight). High salinity increases potential for corrosion in production equipment, and may lead to more challenging wastewater treatment and disposal scenarios. The need to add significant quantities of salt also can potentially increase cost, all else being equal. Improved surfactant systems will be required to reduce the salinity level. Furthermore, work has not been reported to date examining the use of aqueous surfactant-based extraction methods for extraction of soybean oil. Soybeans have been reported to be among the most challenging oilseeds to extract, due to their high phospholipid content, which can cause stable emulsification of extracted oils, requiring additional processing steps (Owusu-Ansah, 1997; de Moura and Johnson, 2009). As such, the objectives of the present study were (1) to conduct preliminary tests of the feasibility of aqueous surfactant-based soybean oil extraction using surfactant formulations used in previous studies for extraction of other oilseeds; (2) to explore whether more complex surfactant formulations involving mixed anionic/cationic/nonionic surfactants can reduce salinity requirements for soybean oil extraction; (3) to conduct a preliminary assessment of contributions of system inputs to ultimate extraction efficiency as a starting point for future modeling efforts; and (4) to explore the potential impact of lecithin co-extraction on interfacial tension, with the aim of understanding its likely impact on surfactant solution reuse. The work described here involves a combination of interfacial tension measurements to determine the conditions necessary to create ultralow interfacial tensions with soybean oil (a requirement for high extraction efficiencies), and extraction studies under varying conditions. Parameters affecting extraction results are evaluated with preliminary Box–Behnken experiment design calculations. The significance of this work is that it takes a critical first step toward development of surfactant-based aqueous methods for soybean oil extraction, providing a proof of concept demonstration of the feasibility of surfactant-based extraction from soybeans, and offering initial insights that will guide future optimization work. The resulting methods should be well suited to meet the rapidly expanding demand for soy-based biofuels and other non-food soybean oil applications.
2. Experimental methods 2.1. Materials All formulations studied in this work were based on an extended propoxylated-ethoxylated anionic surfactant, C10 PO18 EO2 sulfate (22.55% active), which was provided by Huntsman Chemical Co. (The Woodlands, TX). The same surfactant has been used previously in studies of the extraction of corn (Kadioglu et al., 2011), peanut and canola oils (Do and Sabatini, 2010, 2011).
141
Two of the three surfactant formulations studied include the cationic surfactant Arquat HTL8-MS, which was provided by Akzo Nobel (Amsterdam). One surfactant formulation also includes the nonionic surfactant Plantacare 818 UP (50% active), which was provided by Cognis, and the cationic hydrotrope Berol 563 SA (85% active), which was provided by Akzo Nobel. Surfactants and their compositions are listed in Table 1. Anhydrous sodium chloride (>99.5% purity) used for salinity adjustment was purchased from Sigma–Aldrich (St. Louis, MO). Selected experiments explored the effect of addition of soy lecithin on interfacial tension. Soy lecithin (>99% purity) was purchased from Thermo Fisher Scientific (Waltham, MA). The soybean oil used for interfacial tension measurements was purchased from a local market in Norman, OK. All chemicals were used as received. Soybeans (9E482X) from Midland Genetics (Ottawa, KS) were used for all extraction experiments.
2.2. Methods 2.2.1. Preparation of soy flour Soy flour was prepared by initial grinding in a nut grinder, followed by processing in a L’Equip (St. George, UT) Nutrimill Grain Mill. The resulting flour was sieved first through a 300 m sieve, followed by a 212 m sieve, yielding a flour with a particle size less than 212 m. This size is comparable to sizes used on the industrial scale, where soybeans are cracked and flaked to approx. 0.2 mm using rolling mills (Campbell and Glatz, 2009; Do and Sabatini, 2010).
2.2.2. Interfacial tension measurement Interfacial tension measurements were used to assess the abilities of surfactant formulations to produce the ultralow interfacial tensions necessary for effective extraction. Dynamic interfacial tension experiments were carried out using a spinning drop tensiometer (University of Texas, Model 500). Oil droplets 1–3 L in volume were injected into a 300 L capillary tube containing the aqueous surfactant solution. Interfacial tensions were recorded at 5 min intervals over the course of 30 min to ensure measurements reflected equilibrium values.
2.2.3. Aqueous oil extraction with surfactant solutions Soybean flour was added to surfactant solution in 50-mL centrifuge tubes. Samples were placed on a Cole-Parmer (Vernon Hills, IL) Ping-Pong horizontal shaker, model 51504-00. Except where noted, samples were equilibrated on the shaker for 30 min at rates ranging from 100 to 300 rpm. Slurry pH values were monitored and found to be between pH 6.5 and 7 for all experiments. Following equilibration, slurries were centrifuged at 4000 rpm for 30 min in a Thermo Fisher Scientific CL10 centrifuge. After centrifugation, the solid portion was retained and freeze-dried in a Labconco (Kansas City, MO) FreeZone 4.5 freeze dry system for subsequent oil residual analysis. All experiments were conducted in duplicate.
2.2.4. Soxhlet hexane extraction Hexane extraction of soy flour solids remaining after surfactantbased aqueous extraction (Section 2.2.3) was used to determine the amount of remaining oil, for calculation of aqueous extraction yield. Hexane extraction was also used to determine the initial oil content of the soy flour. Freeze-dried samples (Section 2.2.3) and prepared soy flours (Section 2.2.1) were extracted using hexane in a Soxhlet extraction apparatus following the AOCS Method 948.22 standard procedure (AOAC, 1995). After solvent extraction, excess hexane was evaporated from the extracted oil in a hot air oven at
142
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
Table 1 Surfactants studied. Type
Name
Composition
Extended anionic Cationic Nonionic Cationic hydrotrope
C10 PO18 EO2 sulfate Arquad HTL8-MS Plantacare 818 UP Berol 563 SA
Alkyl poly-propyleneoxide diethyleneoxide sodium sulfate Hydrogenated tallow (2-ethylhexyl) dimethylammonium methosulfate C8-16 fatty alcohol glucoside Quaternary cocoalkylamine ethoxylate methyl sulfate
75 ◦ C until no change in mass of the oil was observed. The extraction yield (efficiency) was calculated using Eq. (1): Y (%) = 100 ×
initial oil − remaining oil after extraction initial oil
(1) 3. Results and discussion
The total oil analysis gave an initial soybean oil content of 18.9% on a dry weight basis. The water content of soybean flours was determined to be 7.4%. Note that yield calculated with Eq. (1) does not distinguish between free and emulsified oil, but is simply a calculation of the efficiency of removal from oilseeds during extraction. Others have used the same definitions in previous published aqueous extraction studies (e.g., Rosenthal et al., 1998). While avoiding emulsification is desirable, methods of breaking emulsions exist, as do operational methods of reducing emulsification during extraction (e.g., Rosenthal et al., 1996). Furthermore, some non-food uses of soybean oil such as reverse microemulsion fuels may not require complete absence of emulsion in the starting product (e.g., Do et al., 2011; Kibbey et al., 2014). As such, Eq. (1) is a meaningful measure of the yield for this work. That having been said, it is important to note that none of the extractions conducted for this work exhibited more than a small fraction (∼5–10% or less) of emulsified oil immediately following extraction. That is, calculated yields from Eq. (1) were essentially equivalent to free oil yields for this work. 2.2.5. Assessing the impact of system conditions on extraction yield To explore the significant factors affecting the oil extraction efficiency, the Box–Behnken experimental design method with three factors and one center point was applied (Box and Behnken, 1960). The method provides a way of selecting sets of test conditions to conduct experiments evaluating the impact of conditions on an outcome (in this case, extraction efficiency). Once conditions have been selected, Eq. (2) is used to assess the significance of the individual conditions through nonlinear multivariable regression analysis: Y = b0 +
k i=1
bi xi +
k k
bij xi xj
2014). Additional experiments were conducted to evaluate rate of extraction, as well as the impact of particle size on extraction yield.
(2)
i=1 j=1
where Y is the extraction yield, bi and bij are coefficients, and x values are factors (i.e., the conditions being evaluated), coded so each x is equal to −1, 0 or 1, depending on the magnitude of the condition value. For this work, three factors were considered: solid to liquid ratio (SLR), salinity (S), and shaker agitation rate (A). Experimental combinations were tested for three values for each factor: SLR (0.1, 0.2, 0.3 g/mL), S (1, 5, 8%), and A (100, 200, 300 rpm). All experiments conducted to evaluate the impacts of system conditions were conducted using the baseline extended anionic surfactant formulation only (i.e., no added cationic surfactant or other additives). The three-factor Box–Behnken method with one center point used for this work corresponds to a total of 13 different combinations of experimental conditions tested, and a total of 10 coefficients fit: an intercept (b0 ), one each for SLR, S and A, one each for SLR2 , S2 , A2 , and then one each for the three cross-terms SLR × S, SLR × A and S × A. Coefficients and corresponding p values were determined using the MATLAB (Mathworks, Natick, MA) Statistical Toolbox. Note that similar methods have been used previously by others to study oil extraction processes (e.g., Campbell and Glatz, 2009; Rostami et al.,
3.1. Development of ultralow interfacial tension formulations using surfactant mixtures An important requirement for efficient extraction of oils from oilseeds is formulation of surfactant mixtures capable of achieving ultralow (i.e., <∼0.01 mN/m) interfacial tensions (IFTs) with the oil to be extracted. Previous work has found that vegetable oils often require very high salinities (i.e., high quantities of added salt) to achieve sufficiently low interfacial tensions with vegetable oils and anionic surfactants (Phan et al., 2010a,b). The reason for the high salinity requirement is the highly hydrophobic nature of the vegetable oils, and the relatively hydrophilic nature of the head groups of most anionic surfactants. Mixtures of anionic and cationic surfactants have shown great synergism in reducing the salinity level; however, their potential to precipitate or form liquid crystals has limited their use to date (Doan et al., 2003; Upadhyaya et al., 2006). One method of reducing the sensitivity to precipitation of anionic/cationic mixtures is to add nonionic surfactants to the mixture. Nonionic surfactants incorporated into mixed micelles fit in between the oppositely charged head groups of cationic and anionic surfactants, reducing opportunities for crystal formation and precipitation (Shiau et al., 1994). Table 2 shows the three formulations tested in this work. Formulation 1 corresponds to a single extended anionic surfactant used previously for extraction of corn, peanut and canola oils (Do and Sabatini, 2010, 2011; Kadioglu et al., 2011). Formulations 2 and 3 start with the same extended anionic surfactant and add a cationic surfactant (Formulation 2), or a cationic surfactant in addition to a nonionic surfactant, as well as a small quantity of a cationic hydrotrope (Formulation 3). Formulation 1 was selected based on the previous work, while Formulations 2 and 3 were selected on the basis of preliminary screening. Fig. 1 shows measured interfacial tensions for the three surfactant formulations in contact with soybean oil as a function of salinity (NaCl concentration) at 20 ◦ C. From Fig. 1, it can be seen that all three surfactant formulations are capable of producing ultralow interfacial values with soybean oil. With increasing NaCl concentrations, interfacial tension values pass through minima <0.01 mN/m for all three formulations, indicating that a traditional microemulsion phase I-III-II transition is occurring (Bourrel and Schechter, 1988). Optimal salinities (S*) producing the minimum interfacial tensions in Fig. 1 are listed in Table 2, along with the corresponding interfacial tensions. It can be seen that Formulation 1 requires the highest concentration of NaCl at 5%. Incorporating a small amount of cationic surfactant into the mixture (Formulation 2) – a mass concentration of 0.0075% cationic surfactant out of the 0.15% total surfactant – reduces the salinity from 5% to 2% NaCl, a 60% reduction in the required added salt. It is important to note, however, that formation of surfactant liquid crystal/mesophase was particularly prominent in Formulation 2 for salinities near and above the optimum value. Formulation 3 starts with the extended anionic surfactant and adds significant quantities of cationic surfactant
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
143
Table 2 Formulations studied, measured optimal salinity (S*), and corresponding interfacial tension (IFT*). Total surfactant concentration is equal to 0.15% in all systems. Formulation
1 2 3
Composition (weight fraction) C10 PO18 EO2 sulfate
HTL8-MS cationic
1.00 0.95 0.40
0.05 0.213
818 UP nonionic
0.327
and nonionic surfactant, as well as a small quantity of a cationic hydrotrope. From Fig. 1 and Table 2, it is apparent that ultralow interfacial tensions can be achieved with even lower salinities with Formulation 3 – a sodium chloride concentration of 0.75% – only 15% of the added salt needed for Formulation 1. Formulation 3 also produced the lowest interfacial tension of the three systems (Table 2). Furthermore, it should be noted that systems containing Formulation 3 exhibited almost no evidence of emulsification or surfactant mesophase formation. In contrast, small quantities of emulsion or mesophase were present for both Formulations 1 and 2. It is important to note, however, that Formulation 3 is a substantially different mixture than either Formulations 1 or 2, in that the extended anionic surfactant no longer comprises the majority of the surfactant mass. Methods for a priori design of complex formulations like Formulation 3 are greatly needed. 3.2. Assessment of factors impacting soybean oil extraction 3.2.1. Extraction time As described above, factors impacting soybean oil extraction were explored using the simplest surfactant system, Formulation 1 (Table 2). Fig. 2 shows the time required for extraction at a shaker agitation rate of 200 rpm. Note that extraction yield is observed to plateau near 30 min. For this reason, all subsequent experiments made use of a 30 min extraction time. Note that the time needed to achieve maximum extraction observed in Fig. 2 is comparable to times reported in previous studies of surfactant extraction for other oils (Naksuk et al., 2009; Do and Sabatini, 2010; Kadioglu et al., 2011), and shorter than times typically reported for aqueous enzymatic-assisted extraction (i.e., 1–5 h or longer) (Rosenthal et al., 1998; Campbell and Glatz, 2009).
S* (%, w/v)
IFT* (mN/m)
5 2 0.75
0.0048 0.0104 0.0028
563 SA hydrotrope
0.06
were generally somewhat lower than the 90% extraction yields observed for some other oils under similar conditions with the same surfactant formulation (Do and Sabatini, 2010). Preliminary measurements found that the interfacial tension of aqueous surfactant following the extraction process (0.023 mN/m) was an order of magnitude higher than its initial value (0.0048 mN/m, Fig. 1). While this was still a low interfacial tension, the implication of the increase in interfacial tension is that extraction efficiency might be expected to decrease with subsequent uses, reducing the costeffectiveness of the system, potentially increasing materials and solution disposal costs. It was hypothesized that simultaneous extraction of soy lecithin, a natural surfactant present in soybeans, might be impacting the surfactant performance by changing the nature of the surfactant formulation during extraction. To test this hypotheses, interfacial tension was measured for a range of Formulation 1 systems containing 5% NaCl, with increasing lecithin fractions (Fig. 3). From Fig. 3, it is apparent that, while lecithin does have the potential to increase interfacial tensions at high weight fractions, the effect is negligible at low weight fractions. That is, even at 25% lecithin by weight, the interfacial tension of the surfactant solution in contact with soybean oil is essentially unchanged. This suggests that lecithin extraction is probably not the source of the higher interfacial tensions. One possible explanation for the higher interfacial tensions is that some of the most hydrophobic surfactant components in Formulation 1 may be adsorbing to the extracted soy flours. Note that nearly all commercial surfactants, particularly those containing PO or EO groups, contain a many individual surfactant components; loss of any of those components could potentially change performance. Future work is needed to test this hypothesis.
3.2.2. Effect of soy lecithin One observation from initial soybean extractions was that extraction yields, while greater than 80% under many conditions,
3.2.3. Box–Behnken analysis Table 3 shows the measured extraction yields for the thirteen experiments conducted to evaluate the effect of system conditions on extraction yield. Codes (x values for Eq. (2)) are given in
Fig. 1. Interfacial tensions of three surfactant formulations as a function of salinity (NaCl, %). Arrows indicate optimum salinity values. Surfactant formulations are given in Table 2.
Fig. 2. Oil extraction vs. time for Formulation 1, with SLR = 0.2, S = 5%, and A = 200 rpm.
144
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
Table 3 Experiments conducted to evaluate the impact of conditions on extraction yield. Codes (x values for Eq. (2)) are shown in parentheses. Note: −1, 0 and +1 correspond to low, medium and high values, respectively, for a given parameter. Factor (Code)a
Expt.
1 2 3 4 5 6 7 8 9 10 11 12 13 a
Extraction yield %
SLR, g/mL
S, %
A, rpm
0.2 (0) 0.1 (−1) 0.2 (0) 0.3 (+1) 0.2 (0) 0.1 (−1) 0.1 (−1) 0.3 (+1) 0.3 (+1) 0.2 (0) 0.1 (−1) 0.2 (0) 0.3 (+1)
1 (−1) 5 (0) 8 (+1) 5 (0) 5 (0) 1 (−1) 8 (+1) 8 (+1) 1 (−1) 1 (−1) 5 (0) 8 (+1) 5 (0)
100 (−1) 100 (−1) 100 (−1) 100 (−1) 200 (0) 200 (0) 200 (0) 200 (0) 200 (0) 300 (+1) 300 (+1) 300 (+1) 300 (+1)
37.1 80.3 57.1 67.4 81.5 43.2 52.2 50.0 31.9 45.6 88.6 67.1 72.9
SLR = solid to liquid ratio; S = salinity; A = shaker agitation rate.
Table 4 Coefficient (b) values for Eq. 2, and their statistical significance. Term (Eq. (2)) Intercept SLR S A SLR2 S2 A2 SLR × S SLR × A S×A
Coefficient, b 84.33 −5.50 9.70 5.50 −7.33 −31.50 −0.50 2.25 −1.25 2.00
p value <0.001 0.012 <0.001 0.007 0.017 <0.001 0.856 0.362 0.605 0.415
p < 0.05 X X X X X X
parentheses. Note that observed extraction yields in Table 3 vary considerably, ranging from a low of 31.9% to a high of 88.6%. Table 4 shows the results of the regression analysis of the data in Table 3 using Eq. (2). The table shows both the coefficient (b) values, and the significance (p value) of each coefficient. A lower p value corresponds to a greater impact on the fit. From the values in Table 4, it is apparent that the solid to liquid ratio (SLR), the salinity (S) and shaker agitation rate (A) all have low p values, and thus exhibit strong impacts on the fit, as do the quadratic terms SLR2 and S2 . Not surprisingly, salinity is the most significant factor influencing extraction yield (the p values for S and S2 are 0.0003 and 7 × 10−6 , respectively). This is because ionic surfactant performance varies significantly with salinity. None of the cross terms in Eq. (2) (e.g., S × A) have a significant impact on extraction yield. In
Fig. 3. Effect of soy lecithin on interfacial tension for Formulation 1, S = 5%.
addition, the A2 term does not exhibit a strong impact on extraction yield. Fig. 4 shows a comparison of the actual oil extraction yields (Table 3) and predicted oil extraction yields calculated from Eq. (2) and the coefficients in Table 4, illustrating the ability of Eq. (2) to capture the major factors impacting extraction in the system over a wide range of extraction efficiencies. Fig. 5 shows the response surface at a shaker agitation rate of 200 rpm. From Fig. 5, the strong quadratic nature of the dependence on salinity, S, is apparent. This is consistent with the low p value for the S2 term in Eq. (2) (Table 4). This result is also consistent with the fact that interfacial tension is expected to increase and thus surfactant performance is expected to decrease as salinity becomes significantly larger or smaller than optimum (e.g., Fig. 1). A weak quadratic dependence of solid to liquid ratio (SLR) is also observed in Fig. 5. Although the response surface in Fig. 5 suggests extraction yield may pass through a maximum with SLR, it is more likely that the quadratic term in the model simply reflects the fact that extraction efficiency is relatively stable at the two lower SLR values, only decreasing at the highest value. For example, additional measurements at A = 300 rpm, S = 5% found extraction efficiencies of 88.56, 87.63 and 72.92% for SLR = 0.1, 0.2 and 0.3, respectively. While these data suggest maximum extraction yield occurs at any
Fig. 4. Pure quadratic fit for the response surface model.
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
145
Fig. 5. SLR-S response surface at A = 200 rpm.
SLR 0.2 or lower, fitting a quadratic equation to the data would put an erroneous maximum between SLR of 0.1 and 0.2. In general, extraction yield increases with decreasing SLR within the range studied here, consistent with improved accessibility of oilseed flour grains to extracting surfactant solution when SLR is lower. Fig. 6 shows the response surface at a fixed salinity of 5%. Note that, unlike salinity, the relationship between extraction yield and shaker agitation rate, A, is monotonic, with higher shaker agitation rates corresponding to greater extraction. This is also consistent with the high p value for the A2 term in Eq. (2) (Table 4), which indicates there is little curvature associated with the relationship between extraction yield and agitation. Note that Rosenthal et al. (1998) found that higher shaker agitation rates created more emulsification in non-surfactant-based aqueous extraction methods. It is probable that this would be a consideration with some surfactant-based methods as well at high enough shaker agitation rates, although appropriate surfactant choice could largely mitigate that concern. 3.2.4. Effect of particle size To evaluate the effect of grain size on extraction yield, soy flours with three different particle sizes were extracted using Formulation 1 at optimal salinity, and a shaker agitation rate of 200 rpm. The smallest size (<212 m) was the flour used for all extractions discussed previously. The intermediate size (212–300 m) corresponds to the flour which passed through the 300 m sieve, but was retained by the 212 m sieve. The largest size (>300 m) corresponds to flour processed by the grain mill, but which did not pass through the 300 m sieve. Fig. 7 shows the relationship between particle size and extraction yield. The results show greater extraction for smaller particles. This result is consistent
Fig. 7. Effect of particle size on extraction yield. Error bars correspond to measurement standard deviation from duplicate samples.
with the non-surfactant extraction results of Rosenthal et al. (1998), who found that extraction yield decreased with increasing particle radius, likely due to increased specific surface area with smaller particles and accessibility of disrupted cells on the particle surface. 4. Conclusions The results of this work show that soybean oil can be extracted using extended surfactant-based extraction processes; to the authors’ knowledge this work is the first published report of surfactant-based aqueous extraction of soybean oil. More complex surfactant mixtures formulated with extended anionic, cationic and nonionic surfactants can greatly reduce the salinity requirements for surfactant-based extraction of soybean oil. Extraction yields in this work were found to be as high as 88.6%, with higher yields corresponding to optimal salinity (S*), lower solid to liquid ratios (SLR), higher shaker agitation rates (A), and smaller particle sizes. Future work is needed to develop methods to aid in the design of extraction systems and complex surfactant formulations to further enhance oilseed extraction for soybeans and other materials, and also to better understand the factors that impact reuse of surfactant solutions for extraction, as a means of improving the costs at the industrial scale. The resulting methods are likely to be useful as demand for soybean-based biofuels continues to increase in coming years, particularly as pressures increase to reduce solvent usage in oilseed extractions. Acknowledgments This material is based upon work supported by the National Science Foundation under grant No. 1160053. Partial funding of this work was provided by the Sun Oil Company endowed chair (DAS) and by the industrial sponsors for the Institute for Applied Surfactant Research at the University of Oklahoma. References
Fig. 6. SLR-A response surface at S = 5%.
AOAC, 1995. Official method of analysis of the association of official analytical chemists. AOAC International, Arlington. Biodiesel.org, 2014. Biodiesel Production Statistics. National Biodiesel Board, http://www.biodiesel.org/production/production-statistics Bourrel, M., Schechter, R., 1988. Microemulsions and Related Systems: Formulation, Solvency, and Physical Properties. Marcel Dekker, Inc, New York. Box, G.E.P., Behnken, D.W., 1960. Some new three level designs for the study of quantitative variables. Technometrics 2 (4), 455–475. Campbell, K.A., Glatz, C.E., 2009. Mechanisms of aqueous extraction of soybean oil. J. Agric. Food Chem. 57 (22), 10904–10912.
146
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
de Moura, J.M.L.N., Johnson, L.A., 2009. Two-stage countercurrent enzyme-assisted aqueous extraction processing of oil and protein from soybeans. J. Am. Oil Chem. Soc. 86 (3), 283–289. Do, L.D., Sabatini, D.A., 2010. Aqueous extended-surfactant based method for vegetable oil extraction: proof of concept. J. Am. Oil Chem. Soc. 87 (10), 1211–1220. Do, L.D., Sabatini, D.A., 2011. Pilot scale study of vegetable oil extraction by surfactant-assisted aqueous extraction process. Sep. Sci. Technol. 46 (6), 978–985. Do, L.D., Withayyapayanon, A., Harwell, J.H., Sabatini, D.A., 2009. Environmentally friendly vegetable oil microemulsions using extended surfactants and linkers. J. Surfactants Detergents 12 (2), 91–99. Do, L.D., Singh, V., Chen, L.X., Kibbey, T.C.G., Gollahalli, S.R., Sabatini, D.A., 2011. Algae, canola, or palm oils-diesel microemulsion fuels: phase behaviors. Viscosity, and Combustion Properties. Int. J. Green Energy 8 (7), 748–767. Doan, T., Acosta, E., Scamehorn, J.F., Sabatini, D.A., 2003. Formulating middle-phase microemulsions using mixed anionic and cationic surfactant systems. J. Surfactants Detergents 6 (3), 215–224. EPA, 2001. 40 CFR part 63.GGGG. National emissions standards for hazardous air pollutants: Solvent extraction for vegetable oil production. EPA, 2006. $13.9 million clean air act settlement with multinational soybean and corn processor secures major pollution reductions in eight states. Clean Air Act: Agriculture-Related Enforcement Cases, News release: 10/26/2006. Gunstone, F.D., 2013. Oils and Fats in the Marketplace: Non-Food Uses, AOCS Lipid Library, http://lipidlibrary.aocs.org/market/nonfood.htm Kadioglu, S., Phan, T.T., Sabatini, D.A., 2011. Surfactant-based oil extraction of corn germ. J. Am. Oil Chem. Soc. 88 (6), 863–869. Kibbey, T.C.G., Chen, L.X., Do, L.D., Sabatini, D.A., 2014. Predicting the temperaturedependent viscosity of vegetable oil/diesel reverse microemulsion fuels. Fuel 116, 432–437. Naksuk, A., Sabatini, D.A., Tongcumpou, C., 2009. Microemulsion-based palm kernel oil extraction using mixed surfactant solutions. Ind. Crops Prod. 30 (2), 194–198. Owusu-Ansah, Y.J., 1997. Enzyme-assisted extractions. In: Wan, P.J., Wakelyn, P.J. (Eds.), Technology and Solvents for Extraction of Oilseeds and Nonpetroleum Oils. AOCS Press, Champaign, IL, pp. 323–332.
Phan, T.T., Harwell, J.H., Sabatini, D.A., 2010a. Effects of triglyceride molecular structure on optimum formulation of surfactant-oil-water systems. J. Surfactants Detergents 13 (2), 189–194. Phan, T.T., Witthayapanyanon, A., Harwell, J.H., Sabatini, D.A., 2010b. Microemulsion-based vegetable oil detergency using an extended surfactant. J. Surfactants Detergents 13 (3), 313–319. Qingyi, X., Mitsutoshi, N., Takeo, S., Zengshe, L., 2011. Soybean-based Surfactants and Their Applications, 1 online resource. INTECH Open Access Publisher. Rosenthal, A., Pyle, D.L., Niranjan, K., 1996. Aqueous and enzymatic processes for edible oil extraction. Enzyme Microb. Tech. 19 (6), 402–420. Rosenthal, A., Pyle, D.L., Niranjan, K., 1998. Simultaneous aqueous extraction of oil and protein from soybean: mechanisms for process design. Food Bioprod. Process. 76 (4), 224–230. Rostami, M., Farzaneh, V., Boujmehrani, A., Mohammadi, M., Bakhshabadi, H., 2014. Optimizing the extraction process of sesame seed’s oil using response surface method on the industrial scale. Ind. Crops Prod. 58 (0), 160–165. Salager, J.-L., Antón, R.E., Sabatini, D.A., Harwell, J.H., Acosta, E.J., Tolosa, L.I., 2005. Enhancing solubilization in microemulsions—state of the art and current trends. J. Surfactants Detergents 8 (1), 3–21. Shiau, B.J., Harwell, J.H., Scamehorn, J.F., 1994. Precipitation of mixtures of anionic and cationic surfactants. 3. Effect of added nonionic surfactant. J. Colloid Interf. Sci. 167 (2), 332–345. SoyStats, 2014. A Reference Guide to Important Soybean Facts and Figures. American Soybean Association, www.soystats.com Upadhyaya, A., Acosta, E.J., Scamehorn, J.F., Sabatini, D.A., 2006. Microemulsion phase behavior of anionic-cationic surfactant mixtures: effect of tail branching. J. Surfactants Detergents 9 (2), 169–179. Wan, P.J., Wakelyn, P.J., 1997. Technology and solvents for extracting oilseeds and nonpetroleum oils. AOCS press, 353 pp. Witthayapanyanon, A., Phan, T.T., Heitmann, T.C., Harwell, J.H., Sabatini, D.A., 2010. Interfacial properties of extended-surfactant-based microemulsions and related macroemulsions. J. Surfactants Detergents 13 (2), 127–134.
Contents lists available at ScienceDirect
Industrial Crops and Products journal homepage: www.elsevier.com/locate/indcrop
Preliminary formulation development for aqueous surfactant-based soybean oil extraction Linh D. Do, Travis L. Stevens, Tohren C.G. Kibbey ∗ , David A. Sabatini School of Civil Engineering and Environmental Science, University of Oklahoma, Norman, OK 73019, United States
a r t i c l e
i n f o
Article history: Received 13 June 2014 Received in revised form 1 August 2014 Accepted 14 August 2014 Available online 7 September 2014 Keywords: Oilseed extraction Soybean oil Aqueous extraction Surfactant extraction Interfacial tension
a b s t r a c t Soybean oils are increasingly being used for a range of non-food applications, including production of biofuels and oleochemicals. While most soybean oil is produced by hexane-based extraction methods, concern about environmental and health effects from hexane extraction has led to increased interest in development of aqueous extraction methods. Among aqueous methods, surfactant-based aqueous extraction of vegetable oils has shown particular promise as an alternative to hexane-based extraction methods. The objectives of this work were to explore the use of surfactant-based methods for the extraction of soybean oils, and to test whether the use of mixed anionic–cationic and anionic–cationic–nonionic surfactant mixtures could successfully be used to reduce the salinity requirements for surfactant-based extraction. All three formulations tested were capable of producing ultra-low (<0.01 mN/m) interfacial tensions with soybean oil. One of the formulations, a four-component (three surfactant, one hydrotrope) mixture, was able to reduce the salinity requirement from 5% down to 0.75%. A range of experiments was conducted to better understand the factors influencing extraction yield for surfactant-based extraction of soybean oil. Extraction experiments were conducted with a single extended surfactant system which has been used previously for extraction of other oilseeds. Extraction yields as high as 88.6% were observed for the conditions tested. Extraction yield was strongly dependent on salinity, and was found to increase with increasing shaker agitation rate, decreasing solid to liquid ratio, and decreasing particle size. © 2014 Elsevier B.V. All rights reserved.
1. Introduction While soybean oil continues to be predominantly used in edible products, its use in non-food applications has grown rapidly over the past two decades, and now comprises a substantial fraction of all use. For example, it has been reported that 20% of soybean oil consumption in the United States in 2011 was due to non-food applications, up from only 4% a decade earlier (Gunstone, 2013; SoyStats, 2014). Much of this increase can be attributed to rapidly growing production of biodiesel and other soy-based biofuels. Data from the National Biodiesel Board indicates that biodiesel production has increased 100-fold over the past decade (Biodiesel.org, 2014). Another significant non-food use of soybean oil is the production of oleochemicals and biosurfactants (e.g., Qingyi et al., 2011). Because solvent extraction is the primary method used for extraction of vegetable oils, the vegetable oil extraction industry is a major source of volatile organic compound emissions (Rosenthal
∗ Corresponding author. Tel.: +1 405 325 0580; fax: +1 405 325 4217. E-mail address: [email protected] (T.C.G. Kibbey). http://dx.doi.org/10.1016/j.indcrop.2014.08.026 0926-6690/© 2014 Elsevier B.V. All rights reserved.
et al., 1996). The annual hexane emissions from soybean oil extraction processes have been reported to be 210–430 million liters in the U.S. (Rosenthal et al., 1996). The use of hexane for vegetable oil extraction has led to both increased health concerns, and increased environmental regulations. Exposure to hexane has been shown to cause peripheral nerve damage, and hexane is also a potentially hazardous explosive material (Wan and Wakelyn, 1997). In 2001, the U.S. Environmental Protection Agency (EPA) established regulations on hexane emission (EPA, 2001). With rapidly increasing vegetable oil production, it has become more challenging for oilseed extraction plants to meet these regulations (EPA, 2006). Thus, there is a pressing need for development of environmentallyfriendly and sustainable oil extraction technologies. Aqueous-based oil seed extraction methods have advantages over hexane-based extraction, since aqueous extraction can be employed for either dry or wet oilseeds/plants without extensive drying of the starting material. A recently developed alternate approach has employed aqueous surfactant systems to enhance seed oil extraction (Do et al., 2009; Do and Sabatini, 2010; Phan et al., 2010a; Kadioglu et al., 2011). Surfactant-assisted aqueous extraction makes use of interfacial tension reduction between the surfactant solution and the extractable oil, mobilizing the oil
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
and allowing it to pass through the disrupted oilseed cell matrix (Campbell and Glatz, 2009; Do and Sabatini, 2010). New classes of surfactants, known as extended-surfactants, are able to produce ultralow interfacial tensions with a wide range of vegetable oils, making them attractive for developing a versatile system for bio-oil extraction that can be used across widely different oilseeds or plants (Salager et al., 2005; Do et al., 2009; Phan et al., 2010a; Witthayapanyanon et al., 2010). Although recent preliminary studies of surfactant-assisted aqueous extraction have shown promising oil extraction results (e.g., 85% extraction efficiency with corn germ oil and over 90% with peanut and canola oils (Naksuk et al., 2009; Do and Sabatini, 2010; Kadioglu et al., 2011)), the fundamental processes controlling the oil release mechanisms have yet to be verified and modeled. To date, the amount of added salt required to achieve good extraction efficiency in aqueous surfactant-based extractions has been relatively high (e.g., 5–10% by weight). High salinity increases potential for corrosion in production equipment, and may lead to more challenging wastewater treatment and disposal scenarios. The need to add significant quantities of salt also can potentially increase cost, all else being equal. Improved surfactant systems will be required to reduce the salinity level. Furthermore, work has not been reported to date examining the use of aqueous surfactant-based extraction methods for extraction of soybean oil. Soybeans have been reported to be among the most challenging oilseeds to extract, due to their high phospholipid content, which can cause stable emulsification of extracted oils, requiring additional processing steps (Owusu-Ansah, 1997; de Moura and Johnson, 2009). As such, the objectives of the present study were (1) to conduct preliminary tests of the feasibility of aqueous surfactant-based soybean oil extraction using surfactant formulations used in previous studies for extraction of other oilseeds; (2) to explore whether more complex surfactant formulations involving mixed anionic/cationic/nonionic surfactants can reduce salinity requirements for soybean oil extraction; (3) to conduct a preliminary assessment of contributions of system inputs to ultimate extraction efficiency as a starting point for future modeling efforts; and (4) to explore the potential impact of lecithin co-extraction on interfacial tension, with the aim of understanding its likely impact on surfactant solution reuse. The work described here involves a combination of interfacial tension measurements to determine the conditions necessary to create ultralow interfacial tensions with soybean oil (a requirement for high extraction efficiencies), and extraction studies under varying conditions. Parameters affecting extraction results are evaluated with preliminary Box–Behnken experiment design calculations. The significance of this work is that it takes a critical first step toward development of surfactant-based aqueous methods for soybean oil extraction, providing a proof of concept demonstration of the feasibility of surfactant-based extraction from soybeans, and offering initial insights that will guide future optimization work. The resulting methods should be well suited to meet the rapidly expanding demand for soy-based biofuels and other non-food soybean oil applications.
2. Experimental methods 2.1. Materials All formulations studied in this work were based on an extended propoxylated-ethoxylated anionic surfactant, C10 PO18 EO2 sulfate (22.55% active), which was provided by Huntsman Chemical Co. (The Woodlands, TX). The same surfactant has been used previously in studies of the extraction of corn (Kadioglu et al., 2011), peanut and canola oils (Do and Sabatini, 2010, 2011).
141
Two of the three surfactant formulations studied include the cationic surfactant Arquat HTL8-MS, which was provided by Akzo Nobel (Amsterdam). One surfactant formulation also includes the nonionic surfactant Plantacare 818 UP (50% active), which was provided by Cognis, and the cationic hydrotrope Berol 563 SA (85% active), which was provided by Akzo Nobel. Surfactants and their compositions are listed in Table 1. Anhydrous sodium chloride (>99.5% purity) used for salinity adjustment was purchased from Sigma–Aldrich (St. Louis, MO). Selected experiments explored the effect of addition of soy lecithin on interfacial tension. Soy lecithin (>99% purity) was purchased from Thermo Fisher Scientific (Waltham, MA). The soybean oil used for interfacial tension measurements was purchased from a local market in Norman, OK. All chemicals were used as received. Soybeans (9E482X) from Midland Genetics (Ottawa, KS) were used for all extraction experiments.
2.2. Methods 2.2.1. Preparation of soy flour Soy flour was prepared by initial grinding in a nut grinder, followed by processing in a L’Equip (St. George, UT) Nutrimill Grain Mill. The resulting flour was sieved first through a 300 m sieve, followed by a 212 m sieve, yielding a flour with a particle size less than 212 m. This size is comparable to sizes used on the industrial scale, where soybeans are cracked and flaked to approx. 0.2 mm using rolling mills (Campbell and Glatz, 2009; Do and Sabatini, 2010).
2.2.2. Interfacial tension measurement Interfacial tension measurements were used to assess the abilities of surfactant formulations to produce the ultralow interfacial tensions necessary for effective extraction. Dynamic interfacial tension experiments were carried out using a spinning drop tensiometer (University of Texas, Model 500). Oil droplets 1–3 L in volume were injected into a 300 L capillary tube containing the aqueous surfactant solution. Interfacial tensions were recorded at 5 min intervals over the course of 30 min to ensure measurements reflected equilibrium values.
2.2.3. Aqueous oil extraction with surfactant solutions Soybean flour was added to surfactant solution in 50-mL centrifuge tubes. Samples were placed on a Cole-Parmer (Vernon Hills, IL) Ping-Pong horizontal shaker, model 51504-00. Except where noted, samples were equilibrated on the shaker for 30 min at rates ranging from 100 to 300 rpm. Slurry pH values were monitored and found to be between pH 6.5 and 7 for all experiments. Following equilibration, slurries were centrifuged at 4000 rpm for 30 min in a Thermo Fisher Scientific CL10 centrifuge. After centrifugation, the solid portion was retained and freeze-dried in a Labconco (Kansas City, MO) FreeZone 4.5 freeze dry system for subsequent oil residual analysis. All experiments were conducted in duplicate.
2.2.4. Soxhlet hexane extraction Hexane extraction of soy flour solids remaining after surfactantbased aqueous extraction (Section 2.2.3) was used to determine the amount of remaining oil, for calculation of aqueous extraction yield. Hexane extraction was also used to determine the initial oil content of the soy flour. Freeze-dried samples (Section 2.2.3) and prepared soy flours (Section 2.2.1) were extracted using hexane in a Soxhlet extraction apparatus following the AOCS Method 948.22 standard procedure (AOAC, 1995). After solvent extraction, excess hexane was evaporated from the extracted oil in a hot air oven at
142
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
Table 1 Surfactants studied. Type
Name
Composition
Extended anionic Cationic Nonionic Cationic hydrotrope
C10 PO18 EO2 sulfate Arquad HTL8-MS Plantacare 818 UP Berol 563 SA
Alkyl poly-propyleneoxide diethyleneoxide sodium sulfate Hydrogenated tallow (2-ethylhexyl) dimethylammonium methosulfate C8-16 fatty alcohol glucoside Quaternary cocoalkylamine ethoxylate methyl sulfate
75 ◦ C until no change in mass of the oil was observed. The extraction yield (efficiency) was calculated using Eq. (1): Y (%) = 100 ×
initial oil − remaining oil after extraction initial oil
(1) 3. Results and discussion
The total oil analysis gave an initial soybean oil content of 18.9% on a dry weight basis. The water content of soybean flours was determined to be 7.4%. Note that yield calculated with Eq. (1) does not distinguish between free and emulsified oil, but is simply a calculation of the efficiency of removal from oilseeds during extraction. Others have used the same definitions in previous published aqueous extraction studies (e.g., Rosenthal et al., 1998). While avoiding emulsification is desirable, methods of breaking emulsions exist, as do operational methods of reducing emulsification during extraction (e.g., Rosenthal et al., 1996). Furthermore, some non-food uses of soybean oil such as reverse microemulsion fuels may not require complete absence of emulsion in the starting product (e.g., Do et al., 2011; Kibbey et al., 2014). As such, Eq. (1) is a meaningful measure of the yield for this work. That having been said, it is important to note that none of the extractions conducted for this work exhibited more than a small fraction (∼5–10% or less) of emulsified oil immediately following extraction. That is, calculated yields from Eq. (1) were essentially equivalent to free oil yields for this work. 2.2.5. Assessing the impact of system conditions on extraction yield To explore the significant factors affecting the oil extraction efficiency, the Box–Behnken experimental design method with three factors and one center point was applied (Box and Behnken, 1960). The method provides a way of selecting sets of test conditions to conduct experiments evaluating the impact of conditions on an outcome (in this case, extraction efficiency). Once conditions have been selected, Eq. (2) is used to assess the significance of the individual conditions through nonlinear multivariable regression analysis: Y = b0 +
k i=1
bi xi +
k k
bij xi xj
2014). Additional experiments were conducted to evaluate rate of extraction, as well as the impact of particle size on extraction yield.
(2)
i=1 j=1
where Y is the extraction yield, bi and bij are coefficients, and x values are factors (i.e., the conditions being evaluated), coded so each x is equal to −1, 0 or 1, depending on the magnitude of the condition value. For this work, three factors were considered: solid to liquid ratio (SLR), salinity (S), and shaker agitation rate (A). Experimental combinations were tested for three values for each factor: SLR (0.1, 0.2, 0.3 g/mL), S (1, 5, 8%), and A (100, 200, 300 rpm). All experiments conducted to evaluate the impacts of system conditions were conducted using the baseline extended anionic surfactant formulation only (i.e., no added cationic surfactant or other additives). The three-factor Box–Behnken method with one center point used for this work corresponds to a total of 13 different combinations of experimental conditions tested, and a total of 10 coefficients fit: an intercept (b0 ), one each for SLR, S and A, one each for SLR2 , S2 , A2 , and then one each for the three cross-terms SLR × S, SLR × A and S × A. Coefficients and corresponding p values were determined using the MATLAB (Mathworks, Natick, MA) Statistical Toolbox. Note that similar methods have been used previously by others to study oil extraction processes (e.g., Campbell and Glatz, 2009; Rostami et al.,
3.1. Development of ultralow interfacial tension formulations using surfactant mixtures An important requirement for efficient extraction of oils from oilseeds is formulation of surfactant mixtures capable of achieving ultralow (i.e., <∼0.01 mN/m) interfacial tensions (IFTs) with the oil to be extracted. Previous work has found that vegetable oils often require very high salinities (i.e., high quantities of added salt) to achieve sufficiently low interfacial tensions with vegetable oils and anionic surfactants (Phan et al., 2010a,b). The reason for the high salinity requirement is the highly hydrophobic nature of the vegetable oils, and the relatively hydrophilic nature of the head groups of most anionic surfactants. Mixtures of anionic and cationic surfactants have shown great synergism in reducing the salinity level; however, their potential to precipitate or form liquid crystals has limited their use to date (Doan et al., 2003; Upadhyaya et al., 2006). One method of reducing the sensitivity to precipitation of anionic/cationic mixtures is to add nonionic surfactants to the mixture. Nonionic surfactants incorporated into mixed micelles fit in between the oppositely charged head groups of cationic and anionic surfactants, reducing opportunities for crystal formation and precipitation (Shiau et al., 1994). Table 2 shows the three formulations tested in this work. Formulation 1 corresponds to a single extended anionic surfactant used previously for extraction of corn, peanut and canola oils (Do and Sabatini, 2010, 2011; Kadioglu et al., 2011). Formulations 2 and 3 start with the same extended anionic surfactant and add a cationic surfactant (Formulation 2), or a cationic surfactant in addition to a nonionic surfactant, as well as a small quantity of a cationic hydrotrope (Formulation 3). Formulation 1 was selected based on the previous work, while Formulations 2 and 3 were selected on the basis of preliminary screening. Fig. 1 shows measured interfacial tensions for the three surfactant formulations in contact with soybean oil as a function of salinity (NaCl concentration) at 20 ◦ C. From Fig. 1, it can be seen that all three surfactant formulations are capable of producing ultralow interfacial values with soybean oil. With increasing NaCl concentrations, interfacial tension values pass through minima <0.01 mN/m for all three formulations, indicating that a traditional microemulsion phase I-III-II transition is occurring (Bourrel and Schechter, 1988). Optimal salinities (S*) producing the minimum interfacial tensions in Fig. 1 are listed in Table 2, along with the corresponding interfacial tensions. It can be seen that Formulation 1 requires the highest concentration of NaCl at 5%. Incorporating a small amount of cationic surfactant into the mixture (Formulation 2) – a mass concentration of 0.0075% cationic surfactant out of the 0.15% total surfactant – reduces the salinity from 5% to 2% NaCl, a 60% reduction in the required added salt. It is important to note, however, that formation of surfactant liquid crystal/mesophase was particularly prominent in Formulation 2 for salinities near and above the optimum value. Formulation 3 starts with the extended anionic surfactant and adds significant quantities of cationic surfactant
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
143
Table 2 Formulations studied, measured optimal salinity (S*), and corresponding interfacial tension (IFT*). Total surfactant concentration is equal to 0.15% in all systems. Formulation
1 2 3
Composition (weight fraction) C10 PO18 EO2 sulfate
HTL8-MS cationic
1.00 0.95 0.40
0.05 0.213
818 UP nonionic
0.327
and nonionic surfactant, as well as a small quantity of a cationic hydrotrope. From Fig. 1 and Table 2, it is apparent that ultralow interfacial tensions can be achieved with even lower salinities with Formulation 3 – a sodium chloride concentration of 0.75% – only 15% of the added salt needed for Formulation 1. Formulation 3 also produced the lowest interfacial tension of the three systems (Table 2). Furthermore, it should be noted that systems containing Formulation 3 exhibited almost no evidence of emulsification or surfactant mesophase formation. In contrast, small quantities of emulsion or mesophase were present for both Formulations 1 and 2. It is important to note, however, that Formulation 3 is a substantially different mixture than either Formulations 1 or 2, in that the extended anionic surfactant no longer comprises the majority of the surfactant mass. Methods for a priori design of complex formulations like Formulation 3 are greatly needed. 3.2. Assessment of factors impacting soybean oil extraction 3.2.1. Extraction time As described above, factors impacting soybean oil extraction were explored using the simplest surfactant system, Formulation 1 (Table 2). Fig. 2 shows the time required for extraction at a shaker agitation rate of 200 rpm. Note that extraction yield is observed to plateau near 30 min. For this reason, all subsequent experiments made use of a 30 min extraction time. Note that the time needed to achieve maximum extraction observed in Fig. 2 is comparable to times reported in previous studies of surfactant extraction for other oils (Naksuk et al., 2009; Do and Sabatini, 2010; Kadioglu et al., 2011), and shorter than times typically reported for aqueous enzymatic-assisted extraction (i.e., 1–5 h or longer) (Rosenthal et al., 1998; Campbell and Glatz, 2009).
S* (%, w/v)
IFT* (mN/m)
5 2 0.75
0.0048 0.0104 0.0028
563 SA hydrotrope
0.06
were generally somewhat lower than the 90% extraction yields observed for some other oils under similar conditions with the same surfactant formulation (Do and Sabatini, 2010). Preliminary measurements found that the interfacial tension of aqueous surfactant following the extraction process (0.023 mN/m) was an order of magnitude higher than its initial value (0.0048 mN/m, Fig. 1). While this was still a low interfacial tension, the implication of the increase in interfacial tension is that extraction efficiency might be expected to decrease with subsequent uses, reducing the costeffectiveness of the system, potentially increasing materials and solution disposal costs. It was hypothesized that simultaneous extraction of soy lecithin, a natural surfactant present in soybeans, might be impacting the surfactant performance by changing the nature of the surfactant formulation during extraction. To test this hypotheses, interfacial tension was measured for a range of Formulation 1 systems containing 5% NaCl, with increasing lecithin fractions (Fig. 3). From Fig. 3, it is apparent that, while lecithin does have the potential to increase interfacial tensions at high weight fractions, the effect is negligible at low weight fractions. That is, even at 25% lecithin by weight, the interfacial tension of the surfactant solution in contact with soybean oil is essentially unchanged. This suggests that lecithin extraction is probably not the source of the higher interfacial tensions. One possible explanation for the higher interfacial tensions is that some of the most hydrophobic surfactant components in Formulation 1 may be adsorbing to the extracted soy flours. Note that nearly all commercial surfactants, particularly those containing PO or EO groups, contain a many individual surfactant components; loss of any of those components could potentially change performance. Future work is needed to test this hypothesis.
3.2.2. Effect of soy lecithin One observation from initial soybean extractions was that extraction yields, while greater than 80% under many conditions,
3.2.3. Box–Behnken analysis Table 3 shows the measured extraction yields for the thirteen experiments conducted to evaluate the effect of system conditions on extraction yield. Codes (x values for Eq. (2)) are given in
Fig. 1. Interfacial tensions of three surfactant formulations as a function of salinity (NaCl, %). Arrows indicate optimum salinity values. Surfactant formulations are given in Table 2.
Fig. 2. Oil extraction vs. time for Formulation 1, with SLR = 0.2, S = 5%, and A = 200 rpm.
144
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
Table 3 Experiments conducted to evaluate the impact of conditions on extraction yield. Codes (x values for Eq. (2)) are shown in parentheses. Note: −1, 0 and +1 correspond to low, medium and high values, respectively, for a given parameter. Factor (Code)a
Expt.
1 2 3 4 5 6 7 8 9 10 11 12 13 a
Extraction yield %
SLR, g/mL
S, %
A, rpm
0.2 (0) 0.1 (−1) 0.2 (0) 0.3 (+1) 0.2 (0) 0.1 (−1) 0.1 (−1) 0.3 (+1) 0.3 (+1) 0.2 (0) 0.1 (−1) 0.2 (0) 0.3 (+1)
1 (−1) 5 (0) 8 (+1) 5 (0) 5 (0) 1 (−1) 8 (+1) 8 (+1) 1 (−1) 1 (−1) 5 (0) 8 (+1) 5 (0)
100 (−1) 100 (−1) 100 (−1) 100 (−1) 200 (0) 200 (0) 200 (0) 200 (0) 200 (0) 300 (+1) 300 (+1) 300 (+1) 300 (+1)
37.1 80.3 57.1 67.4 81.5 43.2 52.2 50.0 31.9 45.6 88.6 67.1 72.9
SLR = solid to liquid ratio; S = salinity; A = shaker agitation rate.
Table 4 Coefficient (b) values for Eq. 2, and their statistical significance. Term (Eq. (2)) Intercept SLR S A SLR2 S2 A2 SLR × S SLR × A S×A
Coefficient, b 84.33 −5.50 9.70 5.50 −7.33 −31.50 −0.50 2.25 −1.25 2.00
p value <0.001 0.012 <0.001 0.007 0.017 <0.001 0.856 0.362 0.605 0.415
p < 0.05 X X X X X X
parentheses. Note that observed extraction yields in Table 3 vary considerably, ranging from a low of 31.9% to a high of 88.6%. Table 4 shows the results of the regression analysis of the data in Table 3 using Eq. (2). The table shows both the coefficient (b) values, and the significance (p value) of each coefficient. A lower p value corresponds to a greater impact on the fit. From the values in Table 4, it is apparent that the solid to liquid ratio (SLR), the salinity (S) and shaker agitation rate (A) all have low p values, and thus exhibit strong impacts on the fit, as do the quadratic terms SLR2 and S2 . Not surprisingly, salinity is the most significant factor influencing extraction yield (the p values for S and S2 are 0.0003 and 7 × 10−6 , respectively). This is because ionic surfactant performance varies significantly with salinity. None of the cross terms in Eq. (2) (e.g., S × A) have a significant impact on extraction yield. In
Fig. 3. Effect of soy lecithin on interfacial tension for Formulation 1, S = 5%.
addition, the A2 term does not exhibit a strong impact on extraction yield. Fig. 4 shows a comparison of the actual oil extraction yields (Table 3) and predicted oil extraction yields calculated from Eq. (2) and the coefficients in Table 4, illustrating the ability of Eq. (2) to capture the major factors impacting extraction in the system over a wide range of extraction efficiencies. Fig. 5 shows the response surface at a shaker agitation rate of 200 rpm. From Fig. 5, the strong quadratic nature of the dependence on salinity, S, is apparent. This is consistent with the low p value for the S2 term in Eq. (2) (Table 4). This result is also consistent with the fact that interfacial tension is expected to increase and thus surfactant performance is expected to decrease as salinity becomes significantly larger or smaller than optimum (e.g., Fig. 1). A weak quadratic dependence of solid to liquid ratio (SLR) is also observed in Fig. 5. Although the response surface in Fig. 5 suggests extraction yield may pass through a maximum with SLR, it is more likely that the quadratic term in the model simply reflects the fact that extraction efficiency is relatively stable at the two lower SLR values, only decreasing at the highest value. For example, additional measurements at A = 300 rpm, S = 5% found extraction efficiencies of 88.56, 87.63 and 72.92% for SLR = 0.1, 0.2 and 0.3, respectively. While these data suggest maximum extraction yield occurs at any
Fig. 4. Pure quadratic fit for the response surface model.
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
145
Fig. 5. SLR-S response surface at A = 200 rpm.
SLR 0.2 or lower, fitting a quadratic equation to the data would put an erroneous maximum between SLR of 0.1 and 0.2. In general, extraction yield increases with decreasing SLR within the range studied here, consistent with improved accessibility of oilseed flour grains to extracting surfactant solution when SLR is lower. Fig. 6 shows the response surface at a fixed salinity of 5%. Note that, unlike salinity, the relationship between extraction yield and shaker agitation rate, A, is monotonic, with higher shaker agitation rates corresponding to greater extraction. This is also consistent with the high p value for the A2 term in Eq. (2) (Table 4), which indicates there is little curvature associated with the relationship between extraction yield and agitation. Note that Rosenthal et al. (1998) found that higher shaker agitation rates created more emulsification in non-surfactant-based aqueous extraction methods. It is probable that this would be a consideration with some surfactant-based methods as well at high enough shaker agitation rates, although appropriate surfactant choice could largely mitigate that concern. 3.2.4. Effect of particle size To evaluate the effect of grain size on extraction yield, soy flours with three different particle sizes were extracted using Formulation 1 at optimal salinity, and a shaker agitation rate of 200 rpm. The smallest size (<212 m) was the flour used for all extractions discussed previously. The intermediate size (212–300 m) corresponds to the flour which passed through the 300 m sieve, but was retained by the 212 m sieve. The largest size (>300 m) corresponds to flour processed by the grain mill, but which did not pass through the 300 m sieve. Fig. 7 shows the relationship between particle size and extraction yield. The results show greater extraction for smaller particles. This result is consistent
Fig. 7. Effect of particle size on extraction yield. Error bars correspond to measurement standard deviation from duplicate samples.
with the non-surfactant extraction results of Rosenthal et al. (1998), who found that extraction yield decreased with increasing particle radius, likely due to increased specific surface area with smaller particles and accessibility of disrupted cells on the particle surface. 4. Conclusions The results of this work show that soybean oil can be extracted using extended surfactant-based extraction processes; to the authors’ knowledge this work is the first published report of surfactant-based aqueous extraction of soybean oil. More complex surfactant mixtures formulated with extended anionic, cationic and nonionic surfactants can greatly reduce the salinity requirements for surfactant-based extraction of soybean oil. Extraction yields in this work were found to be as high as 88.6%, with higher yields corresponding to optimal salinity (S*), lower solid to liquid ratios (SLR), higher shaker agitation rates (A), and smaller particle sizes. Future work is needed to develop methods to aid in the design of extraction systems and complex surfactant formulations to further enhance oilseed extraction for soybeans and other materials, and also to better understand the factors that impact reuse of surfactant solutions for extraction, as a means of improving the costs at the industrial scale. The resulting methods are likely to be useful as demand for soybean-based biofuels continues to increase in coming years, particularly as pressures increase to reduce solvent usage in oilseed extractions. Acknowledgments This material is based upon work supported by the National Science Foundation under grant No. 1160053. Partial funding of this work was provided by the Sun Oil Company endowed chair (DAS) and by the industrial sponsors for the Institute for Applied Surfactant Research at the University of Oklahoma. References
Fig. 6. SLR-A response surface at S = 5%.
AOAC, 1995. Official method of analysis of the association of official analytical chemists. AOAC International, Arlington. Biodiesel.org, 2014. Biodiesel Production Statistics. National Biodiesel Board, http://www.biodiesel.org/production/production-statistics Bourrel, M., Schechter, R., 1988. Microemulsions and Related Systems: Formulation, Solvency, and Physical Properties. Marcel Dekker, Inc, New York. Box, G.E.P., Behnken, D.W., 1960. Some new three level designs for the study of quantitative variables. Technometrics 2 (4), 455–475. Campbell, K.A., Glatz, C.E., 2009. Mechanisms of aqueous extraction of soybean oil. J. Agric. Food Chem. 57 (22), 10904–10912.
146
L.D. Do et al. / Industrial Crops and Products 62 (2014) 140–146
de Moura, J.M.L.N., Johnson, L.A., 2009. Two-stage countercurrent enzyme-assisted aqueous extraction processing of oil and protein from soybeans. J. Am. Oil Chem. Soc. 86 (3), 283–289. Do, L.D., Sabatini, D.A., 2010. Aqueous extended-surfactant based method for vegetable oil extraction: proof of concept. J. Am. Oil Chem. Soc. 87 (10), 1211–1220. Do, L.D., Sabatini, D.A., 2011. Pilot scale study of vegetable oil extraction by surfactant-assisted aqueous extraction process. Sep. Sci. Technol. 46 (6), 978–985. Do, L.D., Withayyapayanon, A., Harwell, J.H., Sabatini, D.A., 2009. Environmentally friendly vegetable oil microemulsions using extended surfactants and linkers. J. Surfactants Detergents 12 (2), 91–99. Do, L.D., Singh, V., Chen, L.X., Kibbey, T.C.G., Gollahalli, S.R., Sabatini, D.A., 2011. Algae, canola, or palm oils-diesel microemulsion fuels: phase behaviors. Viscosity, and Combustion Properties. Int. J. Green Energy 8 (7), 748–767. Doan, T., Acosta, E., Scamehorn, J.F., Sabatini, D.A., 2003. Formulating middle-phase microemulsions using mixed anionic and cationic surfactant systems. J. Surfactants Detergents 6 (3), 215–224. EPA, 2001. 40 CFR part 63.GGGG. National emissions standards for hazardous air pollutants: Solvent extraction for vegetable oil production. EPA, 2006. $13.9 million clean air act settlement with multinational soybean and corn processor secures major pollution reductions in eight states. Clean Air Act: Agriculture-Related Enforcement Cases, News release: 10/26/2006. Gunstone, F.D., 2013. Oils and Fats in the Marketplace: Non-Food Uses, AOCS Lipid Library, http://lipidlibrary.aocs.org/market/nonfood.htm Kadioglu, S., Phan, T.T., Sabatini, D.A., 2011. Surfactant-based oil extraction of corn germ. J. Am. Oil Chem. Soc. 88 (6), 863–869. Kibbey, T.C.G., Chen, L.X., Do, L.D., Sabatini, D.A., 2014. Predicting the temperaturedependent viscosity of vegetable oil/diesel reverse microemulsion fuels. Fuel 116, 432–437. Naksuk, A., Sabatini, D.A., Tongcumpou, C., 2009. Microemulsion-based palm kernel oil extraction using mixed surfactant solutions. Ind. Crops Prod. 30 (2), 194–198. Owusu-Ansah, Y.J., 1997. Enzyme-assisted extractions. In: Wan, P.J., Wakelyn, P.J. (Eds.), Technology and Solvents for Extraction of Oilseeds and Nonpetroleum Oils. AOCS Press, Champaign, IL, pp. 323–332.
Phan, T.T., Harwell, J.H., Sabatini, D.A., 2010a. Effects of triglyceride molecular structure on optimum formulation of surfactant-oil-water systems. J. Surfactants Detergents 13 (2), 189–194. Phan, T.T., Witthayapanyanon, A., Harwell, J.H., Sabatini, D.A., 2010b. Microemulsion-based vegetable oil detergency using an extended surfactant. J. Surfactants Detergents 13 (3), 313–319. Qingyi, X., Mitsutoshi, N., Takeo, S., Zengshe, L., 2011. Soybean-based Surfactants and Their Applications, 1 online resource. INTECH Open Access Publisher. Rosenthal, A., Pyle, D.L., Niranjan, K., 1996. Aqueous and enzymatic processes for edible oil extraction. Enzyme Microb. Tech. 19 (6), 402–420. Rosenthal, A., Pyle, D.L., Niranjan, K., 1998. Simultaneous aqueous extraction of oil and protein from soybean: mechanisms for process design. Food Bioprod. Process. 76 (4), 224–230. Rostami, M., Farzaneh, V., Boujmehrani, A., Mohammadi, M., Bakhshabadi, H., 2014. Optimizing the extraction process of sesame seed’s oil using response surface method on the industrial scale. Ind. Crops Prod. 58 (0), 160–165. Salager, J.-L., Antón, R.E., Sabatini, D.A., Harwell, J.H., Acosta, E.J., Tolosa, L.I., 2005. Enhancing solubilization in microemulsions—state of the art and current trends. J. Surfactants Detergents 8 (1), 3–21. Shiau, B.J., Harwell, J.H., Scamehorn, J.F., 1994. Precipitation of mixtures of anionic and cationic surfactants. 3. Effect of added nonionic surfactant. J. Colloid Interf. Sci. 167 (2), 332–345. SoyStats, 2014. A Reference Guide to Important Soybean Facts and Figures. American Soybean Association, www.soystats.com Upadhyaya, A., Acosta, E.J., Scamehorn, J.F., Sabatini, D.A., 2006. Microemulsion phase behavior of anionic-cationic surfactant mixtures: effect of tail branching. J. Surfactants Detergents 9 (2), 169–179. Wan, P.J., Wakelyn, P.J., 1997. Technology and solvents for extracting oilseeds and nonpetroleum oils. AOCS press, 353 pp. Witthayapanyanon, A., Phan, T.T., Heitmann, T.C., Harwell, J.H., Sabatini, D.A., 2010. Interfacial properties of extended-surfactant-based microemulsions and related macroemulsions. J. Surfactants Detergents 13 (2), 127–134.