Water Interfaces

The adsorption of starch that had been hydrophobically modified with octenyl succinate anhydride (OSA) at the oil/water interface during emulsificatio...
8 downloads 0 Views 77KB Size
8770

Langmuir 2006, 22, 8770-8776

Adsorption of Hydrophobically Modified Starch at Oil/Water Interfaces during Emulsification Lars Nilsson* and Bjo¨rn Bergenståhl DiVision of Food Technology, Centre for Chemistry and Chemical Engineering, Lund UniVersity, Post Office Box 124, S-221 00 Lund, Sweden ReceiVed March 31, 2006. In Final Form: July 21, 2006 The adsorption of starch that had been hydrophobically modified with octenyl succinate anhydride (OSA) at the oil/water interface during emulsification was studied. The starch samples were of waxy barley origin and were varied in molar mass and degree of substitution (DS). The particle size of the emulsions was measured and the adsorbed amount of starch was determined through serum depletion. The results show that adsorption is governed by the relationship between interfacial area and OSA-starch concentration. The surface load of OSA-starch can in some cases become very high, reaching 16 mg/m2. The emulsification occurs under nonequilibrium and turbulent flow conditions. Under these conditions kinetic factors are likely to play an important role in the adsorption process. Turbulent flow favors transport to the interface of larger molecules over small ones, which could lead to higher surface loads by causing jamming at the interface. A model that treats the adsorption as a collision between particles in turbulent flow has been used, and it shows that the adsorption time of a polymer decreases with increasing polymer radius. It also shows that the time scale of adsorption is shorter than the time scales for configurational changes of macromolecules at interfaces and that emulsion droplet-droplet collisions are of similar time scales as adsorption, which gives further indications that kinetic factors are important during adsorption. The simulation results give a reasonable explanation to why large molecules such as OSA-starch can be efficient as emulsifiers.

1. Introduction The influence of macromolecules on the stability of colloidal systems has been extensively studied.1-4 The presence of macromolecules in colloidal dispersions may have both stabilizing and destabilizing effects. The stabilization may be due to the increased viscosity of the continuous phase caused by the presence of macromolecules (although that is rarely the case), but it can also be attributed to adsorption of macromolecules at the surface of the colloidal particles. Adsorbed macromolecules, which are soluble in the continuous phase, give rise to steric stabilization with mainly entropic contributions.5,6 On the other hand, destabilization may occur if the surface coverage of the adsorbing polymer is insufficient, which can allow the macromolecules to form bridges between particles. Nonadsorbing macromolecules may be depleted from the vicinity of the surface of the disperse phase and can give rise to depletion flocculation, which also gives a destabilizing effect on the system;7,8 hence it is clear that the adsorption behavior is crucial in understanding the role of a specific macromolecule in a dispersion. The most important measure of the adsorption is the adsorbed amount or surface load (Γ), which is commonly given as amount of macromolecule per surface area (milligrams per square meter) and can be presented as an adsorption isotherm where Γ is a function of the equilibrium bulk macromolecule concentration. * Corresponding author: e-mail [email protected]; tel +46 462229670; fax +46 462224622. (1) Freundlich, H. Kapillarchemie: Eine Darstellung der Chemie der Kolloide und Verwandter Gebiete; Akademische Verlagsgesellschaft: Leipzig, Germany, 1909. (2) Bergenståhl, B. In Gums and Stabilisers for the Food Industry; Phillips, G. O., Wedlock, D. J., Williams, P. A., Eds.; IRL Press Ltd.: Oxford, U.K., 1988; Vol. 4, pp 363-369. (3) Olsson, M.; Linse, P.; Piculell, L. Langmuir 2004, 20, 1611-1619. (4) Dickinson, E. Food Hydrocolloids 2003, 17, 25-39. (5) Napper, D. H. J. Colloid Interface Sci. 1977, 52, 390-407. (6) de Gennes, P. G. AdV. Colloid Interface Sci. 1987, 27, 189-209. (7) Feigin, R. I.; Napper, D. H. J. Colloid Interface Sci. 1980, 75, 525-541. (8) Klein, J.; Luckham, P. F. Nature 1984, 308, 836-837.

Most food dispersions are stabilized by macromolecular compounds. Often the macromolecular compound is a protein as proteins are surface-active and are abundant in most foods. The ability of adsorbed proteins to act as stabilizers of dispersions is 2-fold as they may provide both steric and electrostatic stabilization. Proteins are, however, less efficient stabilizers under conditions with low pH, high ionic strength, and high temperature. In most cases these conditions lower the solubility of the proteins, which in turn limits their ability to give rise to steric stabilization of the dispersion. The high ionic strength will also decrease the efficiency of electrostatic stabilization as the repulsion between dispersed particles is diminished. Thus other macromolecular compounds that retain their stabilizing properties under these conditions are of great interest, as the above conditions are common in many recombined food systems. Polysaccharides constitute the other large group of macromolecules that is abundant in nature and in foods. They are often high molar mass compounds and could thus provide steric stabilization in disperse systems. However, few native polysaccharides display sufficiently amphiphilic properties to form a coherent adsorbed layer. One of the most well-known exceptions to this is gum arabic, which is an extract from certain acacia trees and has been used as an emulsifier and emulsion stabilizer for many years.4,9 The amphiphilic character of gum arabic is mainly due to protein components that are closely associated with the polysaccharide structure of gum.4 Pectin is also considered to have a somewhat amphiphilic character, and partially demethoxylated pectin has been used to produce stable oil-in-water emulsions.10 The amphiphilic character of the demethoxylated pectin was, however, also proven to be due to the presence of a small proteinaceous fraction present in the pectin.10 (9) Islam, A. M.; Phillips, G. O.; Sljivo, A.; Snowden, M. J.; Williams, P. A. Food Hydrocolloids 1997, 11, 493-505. (10) Akhtar, M.; Dickinson, E.; Mazoyer, J.; Langendorff, V. Food Hydrocolloids 2002, 16, 249-256.

10.1021/la060870f CCC: $33.50 © 2006 American Chemical Society Published on Web 09/15/2006

Adsorption of Modified Starch during Emulsification

Chemically modified polysaccharides do, however, offer amphiphilic properties. Cellulose is the most abundant polysaccharide in nature and it is commonly modified to obtain amphiphilic properties through grafting of various groups onto the polysaccharide backbone. The grafted groups depend on the properties desired, such as water solubility and surface activity. Another abundant polysaccharide in nature is starch, which consists of two different polymers: amylose, which is a mainly linear polymer, and amylopectin, which is branched. The amylopectin structure consists of (1f4) glycosidically linked R-D-glucopyranose units and branches that are attached with (1f6) glycosidic links. The amylose structure consists of units of R-D-glucopyranose that are linked in the (1f4) configuration but also a few branches in the (1f6) position. Hydrophobically modified starch is an amphiphilic macromolecule and it offers properties that have many applications within the formation and stabilization of dispersed food systems such as emulsions. Due to its high molar mass and branched polymer structure, hydrophobically modified starch that is adsorbed at the interface will give rise to steric stabilization in emulsions and other dispersed systems. Starch that has been hydrophobically modified through esterification with dicarboxylic acids was first described in 1953 by Caldwell and Wurzburg11 and one of the common starches obtained in this manner is octenyl succinate anhydride(OSA-) starch. The substitution with OSA can occur at carbons 2, 3, and 6 in the glucose molecule and the typical degree of substitution (DS) for food applications is 0.01-0.03. It has been suggested that the substitution occurs preferentially in and around the amorphous branch points of the amylopectin.12 Although it has been around for some time, OSA-starch has not been extensively studied. Viswanathan has studied the effect of the degree of substitution on the emulsification activity and the enzymatic breakdown of OSA-starch and found that a higher DS does not necessarily give higher emulsification activity.13 Furthermore, the resistance to enzymatic degradation compared to native starch was proportional to the DS of the OSA-starch.14 Bao et al.15 have reported that the magnitude of the changes in physical properties such as pasting viscosity, gelatinization temperature, gel hardness and retrogradation of OSA-starch compared to native starch does not only depend on DS but also on the botanical origin of the starch. Ntawukulilyayo et al.16 studied the stabilizing effect of OSAstarch on pharmaceutical paracetamol suspensions and found that at a concentration of 0.1% (w/v) OSA-starch inhibited crystal growth over a one year period. Tesch et al.17 studied the stabilization of emulsions by OSA-starches and found that OSAstarch can be used to produce stable emulsions (30 vol % oil and a Sauter mean particle size of approximately 2 µm) at concentrations as low as 0.08 wt % (corresponds to 0.9 mg m-2). Hydrophobically modified polymers such as OSA-starch are often referred to as associative thickeners because they tend to form networks with other polymers in aqueous solution through hydrophobic interaction.18 This increases the viscosity of the system and can thus stabilize dispersed systems. However, the surface activity and adsorption behavior of OSA-starch at the (11) Caldwell, C. G.; Wurzburg, O. B. U.S. Patent 2-661-349, 1953. (12) Shogren, R. L.; Viswanathan, A.; Felker, F.; Gross, R. A. Starch/Sta¨rke 2000, 52, 196-204. (13) Viswanathan, A. J. EnViron. Polym. Degrad. 1999, 7, 191-196. (14) Viswanathan, A. J. EnViron. Polym. Degrad. 1999, 7, 185-190. (15) Bao, J.; Xing, J.; Phillips, D. L.; Corke, H. J. Agric. Food Chem. 2003, 51, 2283-2287. (16) Ntawukulilyayo, J. D.; De Smedt, S. C.; Demeester, J.; Remon, J. P. Int. J. Pharm. 1996, 128, 73-79. (17) Tesch, S.; Gerhards, C.; Schubert, H. J. Food Eng. 2002, 54, 167-174. (18) Ortega-Ojeda, F. E.; Larsson, H.; Eliasson, A.-C. Carbohydr. Polym. 2005, 59, 313-327.

Langmuir, Vol. 22, No. 21, 2006 8771 Table 1. OSA-Starch Samples

sample

DS

B39-22 0.0224 B27-21 0.0213 B86-10 0.0104

homogen degree of initial molar molar massa rrmsa branching massa (106 g/mol) (106 g/mol) (nm) 0.0548 0.0502 0.0526

39 27 86

12 8.6 6.7

38 44 41

a From ref 21.The root-mean-square radii are for the homogenized samples.

interface is of crucial importance to the formation and stabilization of emulsions. In general, adsorption of amphiphilic polysaccharides has not been studied to the same extent as other biological macromolecules,that is, proteins. It is therefore of great interest to study the emulsification properties of OSA-starch and to investigate and describe the formation and properties of adsorbed OSAstarch layers. 2. Materials and Methods The three OSA-starch samples were provided by Lyckeby Sta¨rkelsen (Kristianstad, Sweden) and were of waxy barley origin containing 7% amylose and 93% amylopectin. The molar degree of branching and the molar degree of substitution (DS) were determined with 1H NMR. The samples for NMR were prepared by dispersing 100 mg of OSA-starch in 10 mL of D2O and placing it in a boiling water bath for 10 min. The samples were freeze-dried and approximately 80 mg of the freeze-dried sample was redissolved in 5 mL of DMSO-d6. The samples were placed in a boiling water bath for 10 min, after which the NMR measurements were performed. The NMR spectrometer (ARX 500, Bruker Fa¨llanden, Switzerland) operated at 500 MHz and the experimental method has been described in greater detail elsewhere.19,20 The molar mass distribution, average molar mass, and root-mean-square radius (rrms) of the OSA-starch samples were determined before and after homogenization by field flow fractionation (AsFlFFF-MALS-RI), which is described elsewhere.21 The properties of the various OSA-starch samples are given in Table 1 and the designation for the three samples is expressed as Bxx-yy, where xx is the initial molar mass (106 g/mol) and yy is the DS. OSA-starch solutions of 1% (w/v) were prepared by dispersing 1.0 g of OSA-starch in a phosphate buffer (10 mM, pH ) 6.0) containing 20 ppm NaN3 which was then diluted to 100 mL. The samples were then placed in a boiling water bath under stirring for 10 min, after which they were left overnight at room temperature. Emulsions with varying OSA-starch concentration were prepared with 5% (w/w) medium-chain triglyceride oil Miglyol 812 F (Sasol, Witten, Germany), in a buffer solution as above, by mixing for 3 min with a high-shear mixer, Ystral ×10/ 25 (Ystral, BallrechtenDottingen, Germany) followed by high-pressure homogenization in a lab-scale valve homogenizer at 15 MPa. The lab-scale valve homogenizer has been described elsewhere.22 All experiments were performed at room temperature. The Sauter mean droplet size (d32) in the emulsions was determined by light diffraction on a Coulter LS130 instrument(Beckman Coulter, High Wycombe, U.K.), and from this the emulsion surface area could be calculated. The adsorbed amount was determined through serum depletion and the emulsions were separated in two steps by mild centrifugation in order to avoid coalescence until a clear subnatant was obtained. The first separation step was carried out at 3400 G for 10 min and the second step at 7000 G for 15 min. The OSA-starch content was determined by enzymatic degradation according to the method of (19) Richardson, S.; Nilsson, G. S.; Bergquist, K.-E.; Gorton, L.; Minschnik, P. Carbohydr. Res. 2000, 328, 365-373. (20) Nilsson, G. S.; Bergquist, K.-E.; Nilsson, U.; Gorton, L. Starch/Sta¨rke 1996, 48, 352-357. (21) Nilsson, L.; Leeman, M.; Wahlund, K. G.; Bergenståhl, B. Biomacromolecules 2006 (in press). (22) Tornberg, E.; Lundh, G. J. Food Sci. 1978, 43, 1553-1558.

8772 Langmuir, Vol. 22, No. 21, 2006

Nilsson and Bergenståhl Table 2. Results from Adsorption Experiments with Different OSA-Starch Samplesa c0 d A app ceq Γ sample (mg/mL) (µm) (m2/mL) (mg/mL) (mg/m2) cadsorbed/c0 B39-22

B27-21

Figure 1. Emulsion surface area created by emulsification with different OSA-starches. The emulsions contained 5% (w/w) medium chain triglyceride (MCT) oil. Åman et al.23 The determination was carried out in the subnatant of the separated emulsion sample and in a reference sample containing the same initial OSA-starch concentration The reference sample was treated in the same way as the emulsion sample and thus went through the same steps of homogenization and centrifugation. As the method of Åman et al. was developed for nonmodified starches, the recovery was slightly lower for the OSA-starch, which was probably due to incomplete hydrolysis. The recovery varied between 85% and 89%. However, with the assumption that the recovery is equal in the emulsion sample and in the reference sample, this error can be minimized. 2.1. Calculation of the Surface Load of OSA-Starch. The adsorbed amount of OSA-starch is obtained from the difference between the amount in the reference sample containing no disperse phase and the amount in the subnatant after separation of the emulsion: cadsorbed ) creference - csub

B86-10

0.42 0.42 0.84 0.84 0.84 1.26 1.26 1.68 1.68 0.42 0.84 1.26 1.68 0.42 0.84 1.26 1.68

11 5.7 6.9 6.5 4.6 1.9 4.5 0.6 0.6 6.7 0.75 0.60 0.55 4.9 1.6 0.86 0.42

0.028 0.053 0.044 0.046 0.065 0.15 0.067 0.51 0.46 0.045 0.40 0.50 0.55 0.062 0.18 0.35 0.72

0.028 0.023 0.051 0.066 0.067 0.20 0.025 0.35 0.46 0.069 0.099 0.26 0.46 0.072 0.12 0.23 0.45

12.2 6.2 15.9 9.1 13.3 6.1 14.3 2.1 1.8 6.4 1.4 1.5 1.5 4.1 2.1 2.7 1.3

0.80 0.77 0.81 0.70 0.72 0.74 0.74 0.61 0.49 0.71 0.71 0.62 0.51 0.60 0.45 0.74 0.56

a c is the initial OSA-starch concentration, d is the Sauter mean 0 diameter of the emulsion droplets, A is the emulsion surface area, ceq is the apparent OSA-starch equilibrium concentration, Γ is the surface load of OSA-starch, and cadsorbed/c0 is the adsorption yield.

(1)

The surface load (Γ) is obtained by relating the adsorbed amount to the specific surface area of the emulsion: Γ)

cadsorbedd32 6φ

(2)

where d32 is the area-weighted droplet diameter and φ is the dispersed phase volume fraction.

3. Results The emulsions were produced with four levels of the three different starch samples. The particle size of the emulsions and the amount of adsorbed OSA-starch were determined. The emulsification performance is shown in Figure 1 as emulsion surface area created versus initial OSA-starch concentration (c0). The results show that it is possible to create an emulsion with the OSA-starch as the sole emulsifier. The Sauter mean diameter of the oil droplets was between 0.4 and 11 µm, which corresponds to an emulsion surface of 0.03-0.7 m2 mL-1. If we assume that the amount of available emulsifier limits the emulsification, the surface area should increase, that is, the droplet diameter should decrease, linearly with the initial emulsifier concentration. For samples B86-10 and B27-21 the amount of surface area created during emulsification increases strongly with increasing initial OSA-starch concentration, but for sample B3922 the surface area increase is less pronounced up to a concentration of about 1.25 mg mL-1. Replicates of sample B3922 also exhibits a rather large variation in the surface area created. (23) Åman, P.; Westerlund, E.; Theander, O. In Methods in carbohydrate chemistry; BeMiller, J. N., Manners, D. J., Sturgeon, R. J., Eds.; John Wiley & Sons Inc.: New York, 1994; Vol. 10, pp 111-115.

Figure 2. Adsorption isotherms for OSA-starches at the oil/water interface.

The results of the adsorption experiments are shown in Table 2. In all cases it appears that only a fraction of the respective OSA-starch sample adsorbs. The fraction adsorbed, the adsorption yield, varies between 45% and 81%.The adsorption yield is dropping with increasing OSA-starch concentration but seems to be equal between the OSA-starch samples and independent of differences in substitution. The adsorption isotherms for the OSA-starch samples are shown in Figure 2 as surface load (Γ) versus the apparent equilibrium concentration of OSA-starch in the bulk solution (apparent ceq). The surface load (Γ) ranges between 1 and 16 mg m-2 for the different samples. For samples B27-21 and B86-10 the surface load shows little concentration dependence and is approximately 1-3 mg m-2, which corresponds quite well to what is expected from a monolayer of an adsorbed macromolecule.24 However, at lower equilibrium concentrations these two samples show an increase in the surface load. The adsorption isotherm for sample B39-22 shows an even stronger dependence on the apparent equilibrium concentration than the two other samples and the surface load decreases strongly with increasing equilibrium concentration. At low apparent equilibrium concentrations the surface load can become very high (reaching (24) Fleer, G. J.; Cohen Stuart, M. A.; Scheutjens, J. H. M. H.; Cosgrove, T.; Vincent, B. Polymers at interfaces; Chapman & Hall: London, 1993.

Adsorption of Modified Starch during Emulsification

Langmuir, Vol. 22, No. 21, 2006 8773 Table 3. Surface Loads for Some Amphiphilic Polysaccharides Γ (mg m-2)

methoda

1.0 0.45-1.4,28 2.4,27 0.86-1.429 0.6-3.028 1.530 6.030 2.3-3.332 1-2.533 0.35-1.434 9.8-11.110 16b 6.4b 4.1b

S E, E, S E S S ST N R S S S S

polymer MC EHEC HM-EHEC HEC HM-HEC HM-DEX cholesterylpullulan HM-6-carboxypullulan demethoxylated pectin OSA-starch B39-22 OSA-starch B27-21 OSA-starch B86-10

Figure 3. Emulsion surface load (Γ) in emulsions with different levels of dynamic surface load (Γdyn); that is, amount of OSA-starch available per specific emulsion surface area.

levels of 16 mg m-2) and the variation between replicates is large. At these lower bulk concentrations is also where the emulsion surface area is smaller, that is, the droplet size is larger (Figure 1). The negative slope of the adsorption isotherms and large variation (Figure 2) suggests that the adsorption of the different OSA-starch samples, and in particular B39-22, is a nonequilibrium process. (Hence, we prefer the use of “apparent equilibrium concentration”.) A possibility is that the surface load may depend on polymer availability versus the specific emulsion surface area during emulsification rather than on equilibrium concentration. The dependence of the surface load on polymer availability versus the specific emulsion surface area has been suggested earlier by Walstra25 and we refer to this relation as the dynamic surface load. We are here assuming that the adsorption rate controls the adsorbed amount. If the polymers adsorb rapidly, there will not be sufficient time for rearrangement and a thick adsorbed layer is obtained. If on the other hand the polymers adsorb slowly, full rearrangement is possible and a thin layer is obtained. The surface load versus the dynamic surface load is shown in Figure 3. The results show that the surface load increases linearly with dynamic surface load for sample B39-22, while samples B27-21 and B8610 behave more ideally, that is, more surface area is created rather than high surface loads.

4. Discussion Native starch does not exhibit any amphiphilic properties. However, when low amounts of hydrophobic substituents, like OSA, are grafted onto the polymer backbone, amphiphilic characteristics are obtained that enable the OSA-starch to act as an emulsifier (Figure 1). Other modified polysaccharides also show amphiphilic properties, and there are several studies with different measuring techniques that have been applied on the adsorption behavior of these polymers on different surfaces. The polymers studied include methyl cellulose (MC),26 ethyl(hydroxyethyl) cellulose (EHEC),27-29 hydrophobically modified ethyl(hydroxyethyl) cellulose (HM-EHEC),28 hydroxyethyl cellulose (HEC),30 hy(25) Walstra, P. In Encyclopedia of emulsion technology Volume I: Basic Theory; Becher, P., Ed.; Marcel Dekker Inc.: New York, 1983; pp 57-127. (26) Saunders, F. L. J. Colloid Interface Sci. 1968, 28, 475-480. (27) Malmsten, M.; Lindman, B. Langmuir 1990, 6, 357-364. (28) Joabsson, F.; Thuresson, K.; Lindman, B. Langmuir 2001, 17, 14991505. (29) Kapsabelis, S.; Prestidge, C. A. J. Colloid Interface Sci. 2000, 228, 297305. (30) Tanaka, R.; Williams, P. A.; Meadows, J.; Phillips, G. O. Colloids Surf. 1992, 66, 63-72.

26

a Serum depletion (S), ellipsometry (E), neutron reflectivity (N), surface tension measurements (ST), and optical reflectometry (R). b Present study.

drophobically modified hydroxyethyl cellulose (HM-HEC),30 hydroxypropyl methylcellulose,31 hydrophobically modified dextran (HM-DEX),32 cholesterylpullulan,33 hydrophobically modified 6-carboxypullulan (HM-6-carboxypullulan),34 and demethoxylated pectin.10 The surface loads found and the method used are given for some of these systems in Table 3. The letters indicating the method are serum depletion (S), ellipsometry (E), surface tension measurements (ST), neutron reflectivity (N), and optical reflectometry (R). From Table 3 it is clear that most hydrophobically modified polysaccharides are able to adsorb giving a surface load around 1 mg m-2 while a few substances, demethoxylated pectin and HM-HEC, give significantly higher surface loads around 5-10 mg m-2.The results obtained for the surface loads of OSA-starches B27-21 and B86-10 are similar to the results in Table 3. OSAstarch B39-22, on the other hand, can give higher surface loads than what would typically be expected. Thus, it is clear that the adsorption behavior of OSA-starch may vary, depending on the properties of the sample, which makes them comparable either to surface-active polysaccharides giving rise to low adsorbed amounts, such as most modified celluloses, or to those giving rise to huge adsorbed amounts, such as demethoxylated pectin. From the adsorption isotherms obtained during the emulsification experiments (Figure 2), it is clear that the adsorption is not an equilibrium process. Adsorption of polymers to emulsion droplets during emulsification occurs under nonequilibrium conditions and thus kinetic factors are likely to play a considerable role in the adsorption. The results in Figure 3 suggest that the adsorption is governed by the amount of surface area created during emulsification and the amount of OSA-starch that is available for adsorption at this surface. When the surface that is created in the homogenizer is small in relation to the amount of polymer available, this could lead to high amounts of polymer being adsorbed as the surface available would be insufficient. The amount of surface area that is created does, on the other hand, to a great extent depend on the conditions in the homogenizer, which are not fully reproducible in a small-scale device, and the dynamic character of this process also agrees with the large variation between replicates or that the adsorption is not controlled by the equilibrium of an adsorbed state and a dissolved state. However, when the surface load is related to an area-normalized concentration (dynamic (31) Wollenweber, C.; Makievski, A. V.; Miller, R.; Daniels, R. Colloids Surf., A: Physicochem. Eng. Asp. 2000, 172, 91-101. (32) Rouzes, C.; Durand, A.; Leonard, M.; Dellacherie, E. J. Colloid Interface Sci. 2002, 253, 217-223. (33) Deme´, B.; Lee, L.-T. J. Phys. Chem. B 1997, 101, 8250-8258. (34) Paris, E.; Cohen Stuart, M. A. Macromolecules 1999, 32, 462-470.

8774 Langmuir, Vol. 22, No. 21, 2006

surface load), this large variability is eliminated. The specific emulsion surface area was not constant in the experiments and the emulsification performance (Figure 1) showed large variations in the amount of surface area ccreated by B39-22. The general concentration dependence of amphiphilic macromolecule adsorption is usually quite weak. The concentration dependence of protein adsorption is well established35-38 and higher protein bulk concentrations tends to give higher surface loads. In the present study, it seems that when the amount of OSA-starch in relation to surface area is high, it can in some cases generate quite high surface loads. The generation of high surface loads can be due to several reasons and below we have listed some possible explanations to this behavior. 4.1. Polydispersity of the Polymer. Many polymer samples are to some degree polydisperse. The effect of polymer polydispersity on adsorption is that at equilibrium a selectivity in polymer adsorption can occur at the interface,24 which means that as the surface approaches saturation high molar mass polymers adsorb preferentially over low molar mass ones.39,40 The reason for this is believed to be the entropy gain that is achieved by having the low molar mass polymers in the solution rather than at the interface. The driving force for the selectivity also depends on the solution concentration and becomes larger in more dilute solutions. This selectivity effect could cause the surface load to become higher as the surface load usually increases with the molar mass of the polymer.24 However, in our case it is a less likely explanation as the theory applies under equilibrium conditions and fully reversible adsorption. Equilibrium conditions are not met during the adsorption in this study, which is mentioned above and shown by the adsorption isotherms in Figure 2. 4.2. Formation of Multilayers at the Interface. When the surface is saturated and completely covered by a monolayer of adsorbed macromolecules, it is possible that additional macromolecules could interact with the monolayer and additional layers could be formed on top of the monolayer. The formation of multilayers is known for proteins and has been reported by several authors.36,41-43 It typically occurs when the solubility of the protein is very low and subsequently the equilibrium concentration in the bulk solution becomes very low. For hydrophobically modified polymers the formation of a multilayer could be due to hydrophobic interaction between nonadsorbed substituents on the polymer backbone leading to some form of aggregation at the surface. This phenomena has been reported for hydrophobically modified pullulans by Duval-Terrie´ et al.44 and has also been suggested for cholesteryl pullulans by Deme´ and Lee.33 In the case of OSA-starch, hydrophobic interaction could also give rise to the formation of inclusion complexes between adsorbed and nonadsorbed OSA-starch. The inclusion complex occurs when a hydrocarbon chain enters the hydrophobic cavity of amylose45 and to some extent the branches of amylopectin.46 (35) Andrade, J. D. In Surface and Interfacial Aspects of Biomedical Polymers; Andrade, J. D., Ed.; Plenum Press: New York, 1985; Vol. 2, Protein adsorption. (36) Young, B. R.; Pitt, W. G.; Cooper, S. L. J. Colloid Interface Sci. 1988, 125, 246-260. (37) van Eijk, M. C. P.; Cohen Stuart, M. A. Langmuir 1997, 13, 5447-5450. (38) McClellan, S. J.; Franses, E. I. Colloids Surf., B: Biointerfaces 2003, 28, 63-75. (39) Kolthoff, I. M.; Gutmacher, R. G. J. Phys. Chem. 1952, 56, 740-745. (40) Cohen Stuart, M. A.; Scheutjens, J. H. M. H.; Fleer, G. J. J. Polym. Sci.: Polym. Phys. Ed. 1980, 18, 559-573. (41) Malmsten, M. J. Colloid Interface Sci. 1998, 207, 186-199. (42) Graham, D. E.; Phillips, M. C. J. Colloid Interface Sci. 1979, 70, 415426. (43) Lassen, B.; Malmsten, M. J. Colloid Interface Sci. 1996, 180, 339-349. (44) Duval-Terrie´, C.; Huguet, J.; Muller, G. Colloids Surf., A: Physicochem. Eng. Asp. 2003, 220, 105-115. (45) Thompson, J. C.; Hamori, E. J. Phys. Chem. 1971, 75, 5. (46) Lundqvist, H.; Eliasson, A.-C.; Olofsson, G. Carbohydr. Polym. 2001, 49, 43-55.

Nilsson and Bergenståhl

This could give rise to multilayer adsorption through interaction between substituents of adsorbed OSA-starch and OSA-starch in solution. This is, however, not a likely explanation because if multilayer adsorption was to explain the high surface loads observed, we would expect the surface load to increase with an increasing apparent equilibrium concentration. However, the results in Figure 2 show the opposite relationship. 4.3. Configuration of Macromolecules at the Interface. Proteins have been reported to adsorb in different orientations depending on adsorption conditions and the three-dimensional structure of the protein. Bovine serum albumin (BSA) has been reported to adsorb in different orientations, that is, side-on or end-on.38,42 The orientation of adsorbed BSA is believed to be due to the three different domains within the protein and the varying charge density in those domains. Thus this adsorption behavior is believed to be dependent on the pH and concentration.38 Fibrinogen has been reported to change its orientation depending on the surface load.43 At low surface loads fibrinogen adsorbed in a rather flat orientation, while at higher surface loads it reoriented to a more upright position that accommodated more protein in the adsorbed layer. IgG has also been reported to be able to adsorb in an end-on orientation.43 It is known that proteins are able to unfold and spread at the interface after adsorption if adequate interfacial area is available and a necessary spreading time is allowed.37,47 If the spreading rate of a macromolecule is not high enough, then fast filling of the interface could prevent it and the macromolecules would have to pack closer together at the interface. This could for instance cause the macromolecules to adsorb in an “upright” configuration. The reconfiguration of adsorbed macromolecules can be divided into two regimes with different time scales. The first regime can be described as the reorientation of the entire macromolecule at the interface, that is, it rotates or “rolls over”. If the macromolecule is asymmetric this could cause it to occupy a different amount of interfacial area per molecule. Time scales for this configurational change have been reported for proteins and are in the range of 10-7-10-4 s.48 Monte Carlo simulations of polymer adsorption have shown times for this initial regime of slightly less than 10-7 s for a flexible 160-mer and 10-8 s for a flexible 40-mer.49 Thus, the time scale is likely to be even higher for larger macromolecules. The second regime can be described as major structural changes of the macromolecule such as unfolding and spreading at the interface, which cause an increase in the interfacial area occupied per molecule. The time scale for this regime range from 10 s for flexible proteins such as β-casein up to 103 s for globular proteins,50 but times as short as 10-2 s have been reported for bovine R-lactalbumin.47 For polysaccharides Deme´ and Lee33 saw a change in the configuration of cholesteryl pullulan after adsorption where the surface load was constant but the adsorbed layer decreased in thickness over time and became more condensed. The surface tension decreased during this time and the phenomenon was attributed to the diffusion of the hydrophobic cholesteryl groups through the adsorbed polymer matrix toward the surface. In the present study we believe that the configurational changes or reorientation of the polymers at the interface is the likely explanation to why surface loads can become very large and we will elaborate on this further. (47) Engel, M. F. M.; van Mierlo, C. P. M.; Visser, A. J. W. G. J. Biol. Chem. 2002, 277, 10922-10930. (48) Adamczyk, Z. J. Colloid Interface Sci. 2000, 229, 477-489. (49) Ka¨llrot, N.; Linse, P. Manuscript in preparation. (50) Walstra, P. Physical Chemistry of Foods; Marcel Dekker Inc: New York, 2003.

Adsorption of Modified Starch during Emulsification

Langmuir, Vol. 22, No. 21, 2006 8775

4.4. Mass Transport in Turbulent Flow. The main emulsification in our case takes place in the valve homogenizer, which operates under turbulent flow conditions. The transport of molecules or polymers to an interface is, under quiescent conditions, diffusion-controlled, which favors the adsorption of small polymers as the diffusion coefficient is inversely proportional to polymer radius. During high-pressure homogenization conditions, diffusional transport of polymers to the interface becomes less important and instead turbulencecontrolled convective transport will dominate.25,51 If the turbulence is assumed to be isotropic, the theoretical description of the flow becomes scalable and has been described by Kolmogorov,52 Levich,51 and others. The adsorption process could be viewed as the collision of two particles, polymer and emulsion droplet, caused by the turbulent flow. The collision frequency between these particles can be described in different ways depending on the length scale relationship between particles and the turbulent eddies. The Kolmogorov theory provides scaling laws for the turbulence, and what is usually referred to as the Kolmogorov scale describes the magnitude of the turbulence microscale and is given by

λ)

() ν3 

1/4

(3)

where λ is the length scale of the smallest eddies, υ is the kinematic viscosity (meters2 per second), and  is the energy dissipation rate per unit mass (watts per kilogram). If the radii of the particles are small in comparison to the Kolmogorov scale, the particle collision frequency has been described by Saffman and Turner.53 For the collision between polymers and droplets, their equation could be expressed as

(ν)

N ) 1.3(rp + rd)3

1/2

npnd

(4)

where N is the collision frequency, rp and rd are the radii of the polymer and the emulsion droplets, respectively (meters), and np and nd are the number concentrations of polymer and emulsion droplets, respectively (per cubic meter). If the radii rp and rd on the other hand are equal to or larger than λ, then the collision frequency can be described54 as

N ) 1.4π(rp + rd)7/31/3npnd

(5)

By assuming that the collision efficiency equals the free available surface, we can from eq 4 or 5 express the surface load as a function of time as

Figure 4. Numerical solutions for eqs 6 (rp, rd < λ) and 7 (rp, rd g λ), showing the estimated adsorption time as a function of polymer radius in turbulent flow.

per square meter per second), cp is the initial polymer concentration in the solution (milligrams per cubic meter), and 1 - θ describes the saturation of the interface where θ ) (Γ/Γmax). Similar approaches as above have been described by Wågberg and Ha¨gglund55 and by Walstra.25 The energy dissipation rate is in our case largely dependent on the homogenization pressure and the volume in which the energy is dissipated in the homogenizer. A typical value of  in a high-pressure homogenizer, according to Walstra,25 is 109 W kg-1 and a calculation of λ, assuming  ) 109 W kg-1 and υ ) 10-6 m2 s-1, gives a value of 180 nm. This suggests that eq 7 is the more accurate description as the OSA-starch is of similar size as λ and the emulsion droplets are larger than λ. To determine the influence of polymer size on the adsorption kinetics, eqs 6 and 7 can be solved numerically for different polymer radii according to

∆Γ ) q(rp,rd)cp(t)[1 - θ(t)]∆t

(8)

with the following boundary conditions:

cp(t) ) cp(0)

(AΓ λ.

the droplet half-life t(1/2), that is, the time it takes for all droplets to collide once, can be derived from eq 5 by rearranging it to

dn ) - Knrd-2/3 dt

(12)

for rd g λ, where K ) 4.2φ(2)1/3 and φ is the volume fraction of the dispersed phase. From eq 12 the droplet half-life can then be expressed as

t(1/2) )

ln 2 2/3 r K d

(13)

The results of these calculations are shown in Figure 5 as droplet half-life as a function of droplet diameter. The most noticeable thing with Figure 5 is that the time scale of droplet half-life is comparable to the time scale for polymer adsorption shown in Figure 4. This suggests that kinetic factors probably are of great importance during the emulsification as the time scales of adsorption and droplet-droplet collision are quite similar. The kinetics of macromolecular adsorption in emulsions has to our knowledge not been widely studied. Proteins have been somewhat studied, but very few of these studies deal with the relation between protein concentration and surface area available for adsorption. Tornberg et al.56,57 have studied the surface load of different proteins on emulsion droplets as a function of different homogenization conditions. The authors found that for emulsions with a small surface area (large particles) the surface load of soy protein, whey protein concentrate (WPC), and blood plasma could become substantially higher than in emulsions with a large surface area (small particles), and as the surface area increased, the surface load decreased. Tcholakova et al.58 have studied the adsorption of WPC at emulsion droplets as a function of protein concentration. In this study the results suggest that the surface load is determined by the droplet size at low protein concentrations. (56) Tornberg, E. J. Sci. Food Agric. 1978, 29, 867-879. (57) Tornberg, E.; Olsson, A.; Persson, K. In Food Emulsions, 3rd ed.; Friberg, S. E., Larsson, K., Eds.; Marcel Dekker Inc.: New York, 1997; pp 279-359. (58) Tcholakova, S.; Denkov, N. D.; Sidzhakova, D.; Ivanov, I. B.; Campbell, B. Langmuir 2003, 19, 5640-5649.

5. Conclusion The OSA-starch samples used in this study produce reasonably stable emulsions within the given concentration range. At higher apparent equilibrium concentrations adsorption behavior followed the Langmuir isotherm, with surface loads of 1-3 mg m-2 for two samples. For one sample in particular the adsorption behavior was shown to be more complex. This sample could give very high surface loads reaching approximately 16 mg m-2. This adsorption behavior could be due to different mechanisms such as polymer polydispersity, formation of multilayers at the interface, or the polymer configuration at the interface. For reasons presented above we believe that the high surface loads are mainly due to jamming at the interface caused by the polymer orientation. The interfacial jamming could be caused by the relationship between surface area created during homogenization and amount of polymer available for adsorption, that is, when the emulsion surface area is small and the polymer concentration is high. The emulsification occurs under nonequilibrium and turbulent flow conditions and these factors most likely have a profound effect on the adsorption behavior. Under these conditions kinetic factors are likely to play a large role in the adsorption process. Turbulent flow favors transport to the interface of larger molecules over small ones, which could lead to higher surface loads and limit changes in configuration and unfolding at the interface after adsorption. A model that treats the adsorption as a collision between particles in turbulent flow was used and it shows that the adsorption time of a polymer decreases with increasing polymer radius. The time scale of adsorption is lower than the time scales for configurational changes of macromolecules at interfaces. Thus it is likely that the macromolecules are unable to optimize their configuration before jamming occurs at the interface. The modeling results also give a reasonable explanation to why large molecules such as OSA-starch can be efficient in the formation of emulsions. The model also shows that the adsorption of macromolecules within the given size range and emulsion droplet-droplet collision are of similar time scales, which gives a further indication that kinetic factors are important during adsorption. Acknowledgment. The Center for Amphiphilic Polymers, Lund, Sweden, is acknowledged for financial support. LA060870F