Langmuir 2007, 23, 5005-5013
5005
Morphological Changes in Adsorbed Protein Films at the Air-Water Interface Subjected to Large Area Variations, as Observed by Brewster Angle Microscopy Rong Xu, Eric Dickinson, and Brent S. Murray* Food Colloids Group, Procter Department of Food Science, UniVersity of Leeds, Leeds LS2 9JT, U.K. ReceiVed NoVember 9, 2006. In Final Form: February 7, 2007 Adsorbed films of proteins at the air-water interface have been imaged using Brewster angle microscopy (BAM). The proteins β-lactoglobulin (β-L) and ovalbumin (OA) were studied at a range of protein concentrations and surface ages at 25.0 °C and two pH values (7 and 5) in a Langmuir trough. The adsorbed films were periodically subjected to compression and expansion cycles such that the film area was typically varied between 125% and 50% of the original film area. With β-L on its own, no structural changes were observable at pH 7. When a low-area fraction (less than 0.01%) of 20 µm polystyrene latex particles was spread at the interface before adsorption of β-L, the particles became randomly distributed throughout the interface, but after protein adsorption and compression/expansion, the particles highlighted notable structural features not visible in their absence. Such features included the appearance of long (several hundred micrometers or more) folds and cracks in the films, generally oriented at right angles to the direction of compression, and also aggregates of protein and/or particles. Such structuring was more visible the longer the film was aged or at higher initial protein concentrations for shorter adsorption times. At pH 5, close to the isoelectric pH of β-L, such features were just noticeable in the absence of particles but were much more pronounced than at pH 7 in the presence of particles. Similar experiments with OA revealed even more pronounced structural features, both in the absence and presence of particles, particularly at pH 5 (close to the isoelectric pH of OA also), producing striking stripelike and meshlike domains. Changes in the dilatational elasticity of the films could be correlated with the variations in the structural integrity of the films as observed via BAM. The results indicate that interfacial area changes of this type, typical of those that occur in food colloid processing, will lead to highly inhomogeneous adsorbed protein layers, with implications for the stability of the corresponding foams and emulsions stabilized by such films. Overall, the experimental results are in broad agreement with the sorts of trends predicted by earlier computer simulations of protein films subjected to such compression and expansion.
Introduction Many food products contain emulsions and foams where the adsorbed layer that stabilizes the droplets and bubbles consists mainly of protein. The processing of such colloidal products may involve the application of high shear forces during stirring, pumping, and homogenization, as well as pressure and temperature cycling. These processes may result in considerable compression and expansion of the adsorbed protein film stabilizing the system. However, little is known about how such abrupt interfacial area changes may disrupt the structure of the adsorbed films, even though such disruption is likely to have important consequences for the stability of the systems. For example, recent work has indicated1,2 that common food proteins, even at quite high protein concentrations, have difficulty in preventing the coalescence of air bubbles when they are expanded at relatively modest rates and extents of expansion. Since the introduction about 15 years ago of Brewster angle microscopy3,4 (BAM) and more recently imaging ellipsometry,5 they have become established as fundamental methods of noninvasively studying the two-dimensional structure of lipid monolayers. More recently, BAM has also been used by various * To whom correspondence should be addressed. Tel: 44 (0)113 3432962. Fax: 44 (0)113 3432982. E-mail
[email protected]. (1) Murray, B. S.; Dickinson, E.; Lau, C. K.; Schmidt, E. Langmuir 2005, 21, 4622. (2) Murray, B. S.; Campbell, I.; Dickinson, E.; Maisonneuve, K.; Nelson, P. V.; So¨derberg, I. Langmuir 2002, 18, 5007. (3) He´non, S.; Meunier, J. ReV. Sci. Instrum. 1991, 62, 936. (4) Ho¨nig, D.; Mo¨bius, D. J. Phys. Chem. 1991, 95, 4590. (5) Harke, M.; Teppner, R.; Schulz, O.; Motschmann, H.; Orendi H. ReV. Sci. Instrum. 1997, 68, 3130.
workers for visualization of the morphology of adsorbed films of proteins + lipids at the air-water (A-W) interface.6-10 In general, protein films on their own, when adsorbed under quiescent conditions, do not exhibit the rich array of structures and phase characteristics of low-molecular-weight surfactants, which can pack together in domains of specific intermolecular spacing and orientation. When built up fairly slowly from low surface coverages to high coverages via adsorption or spreading, protein films as observed via BAM tend to form fairly nonstructured films. This is as might be reasonably expected of the thin, gel-like network of molecules of which they are thought to be composed. However, when such films are quickly compressed or expanded, features such as collapse, multilayer formation, fracture, and disintegration are possible, and this paper seeks to explore the occurrence of these possibilities in reality. Macroscopic fracture of protein films under such conditions has been observed recently by Jones and Middelburg.11 This study is part of a larger project investigating the change in stabilizing properties of adsorbed protein films when subjected to typical food processing conditions.1,2,12-15 In developing this experi(6) Murray, B. S. Colloids Surf., A 1997, 125, 73. (7) Murray, B. S.; Cattin, B.; Schu¨ler, E.; Sonmez, Z. O. Langmuir 2002, 18, 9476. (8) Rodrı´guez Patino, J. M.; Sa´nchez, C. C.; Rodrı´guez Nin˜o, M. R. Langmuir 1999, 15, 2484. (9) Mino¨nes Conde, J.; Rodrı´guez Patino, J. M.; Mino¨nes Trillo, J. Biomacromolecules 2005, 6, 3137. (10) Sa´nchez, C. C.; Rodrı´guez, Nin˜o, C. R.; Caro, A. L.; Rodrı´guez Patino, J. M. J. Food Eng. 2005, 67, 225. (11) Jones, D. B.; Middelberg, A. P. J. Chem. Eng. Sci. 2002, 57, 1711. (12) Murray, B. S.; Dickinson, E.; Du, Z.; Ettelaie, R.; Maisonneuve, K.; So¨derberg, I. In Food Colloids, Biopolymers and Materials; Dickinson, E., van Vliet, T., Eds.; Royal Society of Chemistry: Cambridge, 2003; p 165.
10.1021/la063280q CCC: $37.00 © 2007 American Chemical Society Published on Web 03/27/2007
5006 Langmuir, Vol. 23, No. 9, 2007
mental study, we have also been motivated to obtain results for comparison with our earlier computer simulations16 of adsorbed films subjected to the same sorts of compression and expansion. In this previous work, Brownian dynamics simulations were performed of compressed and expanded protein layers, modeled as surface-active monodisperse particles possessing the capacity for forming interparticle bonds of varying strength at the interface. BAM has been used here to observe changes in the morphology of adsorbed films of β-lactoglobulin (β-L) and ovalbumin (OA) when subjected to compression and expansion cycles in a specially designed Langmuir trough, considering system variables such as film age, initial bulk protein concentration, and pH. As the major protein of whey, β-L is of widespread significance in many emulsified and foamed food products. Similarly OA, the major egg-white protein, has many important functions in food products, such as gelation and foaming, but it is of particular functional significance at the A-W interface due to its unique contribution to the whipping of egg white.17 Partly to combat the poor optical contrast obtained with BAM for proteins on their own, a low area fraction of microscopic solid particles was incorporated into the films in some experiments. This additional structural complexity also has practical significance because in many food products particles are known to form part of the adsorbed layer around emulsion droplets and bubbles. Such particles include, for example, droplets of another phase (e.g., lipid droplets), casein micelles, starch granules and crystals of fat, sugar, and other organic and inorganic particulate materials. Particles at the interface may help to stabilize or destabilize the system, mainly depending on their wetting characteristics, but their effects in extensively expanded or compressed protein films have not so far been systematically investigated. In particular, it is not known how the clustering of particles in a film might affect the structure and mechanical properties of the protein film itself and vice versa.15 A BAM experiment basically records the reflected intensity of p-polarized laser light at the Brewster angle due to the presence of a film at the interface.3,4 The incident beam is scanned over a certain area of the interface and the corresponding reflected intensities are captured on a CCD array to construct an image. The reflected intensity increases with increasing thickness and optical density of the film, but there is generally no simple relationship between these parameters and the gray levels registered in the digital image,20 so that it is usually not possible to attribute changes in reflected intensity to absolute changes in thickness or surface load. The resolution of BAM is typically around 1 µm, considerably less than fluorescence microscopy or scanning probe microscopy, but the technique has the great advantage of being completely noninvasive and not requiring labeling of the sample. In the following we describe the results of experiments when films of β-L and OA are allowed to adsorb at the A-W interface for different times, including the interfacial tension changes and the structuring that is observed via BAM, when the films are (13) So¨derberg, I.; Dickinson, E.; Murray, B. S. Colloids Surf., B. 2003, 30, 237. (14) Murray, B. S.; Dickinson, E.; Gransard, C.; So¨derberg, I. Food Hydrocolloids 2006, 20, 114. (15) Murray B. S.; Cox, A.; Dickinson, E.; Nelson, P. V.; Wang, Y. In Food Colloids: Self-Assembly and Material Science; Dickinson, E., Leser, M., Eds.; Royal Society of Chemistry: Cambridge, 2007, p 369. (16) Pugnaloni, L. A.; Ettelaie, R.; Dickinson, E. J. Colloid Interface Sci. 2005, 287, 401. (17) Mine, Y. Food Res. Int. 1996, 29, 155. (18) Garofalakis, G.; Murray, B. S. Food Sci. Technol. Today 1999, 13, 151. (19) Garofalakis, G.; Murray, B. S. Colloids Surf., B. 2001, 21, 3. (20) Garofalakis, G.; Murray, B. S. Langmuir 2002, 18, 4765.
Xu et al.
Figure 1. Schematic illustration of the Langmuir trough arrangement, as viewed from above. Key: (a) trough walls (made from anodized aluminum alloy); (b) location of Wilhelmy plate (when used) and also location and orientation of laser beam on the interface (when BAM used); (c) barrier made of single piece of PTFE (of side length 15 cm); (d) direction of stepper motor drive for compression of interface; (e) indication of the change in shape of barrier on compression.
subjected to compression/expansion cycles of magnitudes that are of significance to food colloid processing. Materials and Methods Materials. The β-L from bovine milk (∼90% PAGE, crystallized and lyophilized; lot no. 033K7003) and the OA from chicken egg white (>98% agarose gel electrophoresis; crystallized and lyophilized; lot no. 0146K7607) were from Sigma-Aldrich Chemical Co. (Poole, Dorset). Surfactant-free polystyrene latex particles, 20 µm diameter (lot No. 26902), were supplied by Duke Scientific Corporation (Palo Alto, CA) in the form of a suspension containing of 0.3% solid particles. A sample of this suspension was dried overnight and then redispersed in a mixture of chloroform and ethanol of 6:4 weight ratio immediately prior to use. All other reagents, including imidazole buffer, chloroform, ethanol, sodium chloride, hydrochloric acid, acetic acid, and acetone were AnalR grade from Sigma. Water from a Milli-Q system (Millipore Ltd, Watford, UK), free from surface-active impurities and with a conductivity of less than 10-7 S cm-1, was used throughout. Proteins were dissolved in pH 7.0 buffer solutions formed from 0.05 mol dm-3 imidazole + 0.05 mol dm-3 NaCl, adjusted to pH 7.0 by addition of HCl. Protein concentrations varying between 0.005 and 0.05 wt % were studied. A pH 5.0 buffer was also used in some experiments, formed from 0.01 mol dm-3 sodium acetate (+0.05 mol dm-3 NaCl) and adjusted to pH 5 by addition of acetic acid. Methods. Images of protein adsorbed films at the A-W interface were recorded using a BAM2plus Brewster angle microscope (NFT, Go¨ttingen, Germany), combined with a specialist Langmuir trough (see Figure 1) employing rhombic PTFE barriers for compression and expansion of the interface, as described previously.18-20 The design of this trough also allows easy compression and expansion of films at the oil-water interface. In the experiments reported here the trough containing the aqueous phase was made from anodized aluminum alloy with a hollowed-out base for circulation of water from a thermostatted bath. This arrangement gave excellent temperature control and reduced thermal convection at the A-W interface, which can disturb imaging of the interface. This feature of the trough was also designed to allow thermal processing of the adsorbed films. All experiments were carried at 25.0 ( 0.1 °C, unless stated otherwise. Images were collected from the center of the rhombic barrier, as indicated in Figure 1. The trough was first filled with buffer solution while the barriers were in such a position that the surface area enclosed by the barrier was 181 cm2. This is referred to throughout as the initial, starting area, A0. Where particles were employed, 0.0018 g of the dried latex particles were resuspended in 10 mL of the chloroform/ethanol mixture, and 80 µL of this suspension was spread via a microsyringe onto the A-W interface inside the barrier. The system was then left
Morphological Changes in Adsorbed Protein Films
Figure 2. Schematic diagram illustrating the compression and expansion cycle applied to each protein film. The film area (A), relative to the initial starting area (A0), i.e., on initial adsorption of protein, before any change in film area was applied, is plotted against time of compression or expansion. The time taken to acquire images (when the barrier was motionless) between each compression or expansion stage is not shown. Images were collected mainly at points (i) A/A0 ) 1, before any area change, (ii) A/A0 ) 0.48, (iii) A/A0 ) 1.24, and (iv) A/A0 ) 1, i.e., back to the original starting area. Occasionally images were collected at other stages as well. for 60 ( 10 min to allow the solvent to evaporate and the particles to disperse throughout the interface. After this time equal volumes ()25 mL) of protein solution (concentration ) 0.2 wt %) were injected into the four corners of the trough outside the barriers. Gentle stirring was applied by slowly sweeping a glass rod across the base of the trough for 2-3 min, sufficient to ensure a homogeneous protein concentration was obtained in the subphase. This point was taken as “zero” protein adsorption time. It should be mentioned that various other spreading solvents and spreading regimes were tested. For example, we attempted spreading latex particles at the interface after filling the trough with protein solution and rapidly expanding the interface, or aspirating the initially adsorbed film and then spreading the particles at the interface. With these alternative methods, however, extensive aggregation or precipitation of protein appeared to take place, presumably due to the solvent in the spreading solution. Therefore, the procedure was adopted of injecting the protein solution into the trough after a film of spread particles had been formed. It should also be mentioned that we tried a number of other types of particle, including silica particles of 0.02 and 0.08 µm diameter; also various polymer latices ranging in size from 0.15 and through to 2 and 10 µm diameter. With all these other particles, however, there were problems with either their extensive aggregation at the interface or their disappearance on spreading at the interface or on addition of the protein, so that they were not suitable for these experiments. After waiting for different adsorption times, the interface was compressed and expanded. The maximum area possible was 225 cm2, or A/A0 ) 1.24, as dictated by the dimensions of the barrier. The minimum area possible was 87.1 cm2, or A/A0 ) 0.48, as required to avoid the barrier touching the optics conveying the laser light to and from the interface. At different values of A/A0, the barrier motion was stopped and the interface was imaged via the BAM for 5 min. Figure 2 illustrates the typical compression/expansion cycles applied to the films. It was not possible to obtain images while the barrier was actually moving at the speeds employed because this caused movement of the interfacial features and consequent blurring of the images. Although in the apparatus used the films could be compressed or expanded at higher rates than those employed here,7 it was found that this made imaging difficult because of the persistence of reflected waves within the barrier for 10-20 s after the barrier motion had ceased. The laser output power was always kept at a given percentage (60%) of its maximum and the camera shutter speed fixed at 50 s-1. The resolution of the microscope employed was approximately 2 µm and the image dimensions 768 × 572 pixels. The analyzer and polarizer angles were set at zero. Although slightly improved resolution could sometimes be obtained by varying these settings, there was then the additional difficulty of reliably comparing systems under different settings. Minor changes in image brightness or contrast were made using the BAM2plus software. In experiments where the surface tension (γ) was monitored during compression/expansion cycles, a Wilhelmy plate of roughened mica
Langmuir, Vol. 23, No. 9, 2007 5007
Figure 3. Images of adsorbed films for 0.005 wt % β-L at pH 7 with polystyrene particles after compression to A/A0 ) 0.48 for (a) 4-h adsorption; (b) 15-h adsorption and after first compression cycle at 4 h. The large white arrows in (a) indicate the direction of compression in this example and all subsequent BAM images presented. The black arrow in (a) indicates the typical optical interference pattern that is present on all images (i.e., not film structure). was positioned in the interface at the center of the trough with its length parallel to the long dimension of the barrier, as indicated in Figure 1. The Wilhelmy plate was 3-4 cm wide and suspended from force transducer and digital recording device (Maywood Instruments, Basingstoke, UK) connected to a PC. Force data were recorded at a rate of 5 or 10 readings per second.
Results and Discussion The BAM results in a series of experiments under different conditions are presented separately for β-L and OA. Typically, results are compared for 0.005 wt % protein for short (ca. 4 h) and long (ca. 24 h) adsorption times, as well at two different pH values (pH 5 and 7) and with and without added latex particles. Morphological and Structural Characteristics of β-L Adsorbed Films. The first system to be studied was 0.005 wt % β-L at pH 7 without latex particles. After 4 and 21 h, the adsorbed film was compressed from A/A0 ) 1 to A/A0 ) 0.48, expanded to A/A0 ) 1.24, then expanded back to A/A0 ) 1, i.e., the original area, as indicated in the first part of the cycle in Figure 2. After each compression/expansion stage, BAM images were collected. However, as noted previously,7,18 no specific structures are observable under these sorts of conditions. At this bulk protein concentration, 4 h is long enough for an appreciable film of β-L to be adsorbed, close to the maximum surface packing,21 as evidenced by the increase and decrease in brightness on compression and expansion, respectively, indicating the presence of adsorbed material. Very similar results were obtained for the 21-h old film, the only difference being that the images were brighter, especially on compression to A/A0 ) 0.48, due to more extensive accumulation of protein at the interface for this longer adsorption time. Thus, for these solution conditions, no obvious surface structuring in the film was observable. Figure 3 shows examples of images obtained for the same solution conditions as above, 0.005 wt % β-L at pH 7, but now in the presence of added latex particles. The films were subjected to a similar compression/expansion cycle (A/A0 ) 1 to 0.76 to 0.48 to 1) after 4- and 15-h adsorption. Images are shown at the completion of the compression to A/A0 ) 0.48. At this compression stage, it was noticeable that the particles became (21) Murray, B. S. Prog. Colloid Polym. Sci. 1997, 103, 41. (22) Roth, S.; Murray, B. S.; Dickinson, E. J. Agric. Food Chem. 2000, 48, 1491. (23) Lipp, M. M.; Lee, K. Y. C.; Takamoto, D. Y.; Zasadzinski, J. A.; Waring, A. J. Phys. ReV. Letts. 1998, 81, 1650. (24) Murray, B. S.; Dickinson, E.; Du, Z.; Ettelaie, R.; Kostakis, T.; Vallet, J. In Food Colloids: Interactions, Microstructure and Processing; Dickinson, E., Ed.; Royal Society of Chemistry: Cambridge, 2005; p 259. (25) Ehsani, N.; Parkkinen, S.; Nystrio¨m, M. J. Membr. Sci. 1997, 123, 105. (26) Forciniti, D.; Hall, C. K. Biotechnol. Bioeng. 1991, 38, 986. (27) Ybert, C.; Lu, W. X.; Moller, G.; Knobler, C. M. J. Phys. Chem. B 2002, 106, 2004.
5008 Langmuir, Vol. 23, No. 9, 2007
Figure 4. Images of adsorbed films for 0.005 wt % β-L at pH 5 with polystyrene particles after 4-h adsorption at different film areas: (a) A/A0 ) 1, before any area change; (b) after compression to A/A0 ) 0.76; (c) after compression to A/A0 ) 0.48 and (d) after expansion back to A/A0 ) 1.
concentrated into bright bands. The long axes of the bands were generally orientated at right angles to the direction of compression on the first cycle (see Figure 3a). The bands persisted after expansion back to the original area and were still visible at the start of the second compression/expansion cycle after 15-h adsorption. They tended to be even more pronounced after compression of the aged film (Figure 3b). However, either as a result of the time gap between the first and second compression/ expansion cycles, or the additional compression and expansion, the orientation of the bands became more random, suggesting that rotation of these domains within the films was occurring. It is also worth mentioning that the close-spaced lines indicated in Figure 3a do not represent surface structure. They are present on all our BAM images and arise from an optical interference effect that is difficult to remove, by subtraction of background image for example. This is partly because the structures are dynamic to some extent, either changing their size and shape slightly and/or drifting in and out of view in the time it takes to acquire a single image. The effect can be reduced by appropriate processing of individual images but generally at the expense of reducing the clarity of any genuine surface structure effects. Therefore, in all of the following, we present the raw images, except for the occasional adjustment in brightness or contrast of the whole image. The susceptibility of such a protein film to the build up of corrugations or to their fracture on compression/expansion will obviously depend on the overall viscoelastic properties of the film. It is well known28 that protein film viscoelasticity generally increases as the pH is lowered toward to the isoelectric point (pI) of the protein (as long as the protein does not completely precipitate). For this reason, β-L was also studied at pH 5, close to its pI value of 5.6.22 Figures 4 and 5 show BAM images obtained for films adsorbed from 0.005 wt % β-L, as above, with latex particles, but now at pH 5 and after 4- and 20-h adsorption, respectively. After 4-h adsorption (Figure 4), the film was subjected to the same compression/expansion cycle as in Figure 3 (A/A0 ) 1 to 0.76 to 0.48 to 1). Much more extensive banding was observed at pH 5 (Figure 4) than at pH 7 (Figure 3), even for the relative short adsorption time. After the longer adsorption time (Figure 5), the banding was even more extensive. In the latter case, the film was (28) Murray, B. S. Curr. Opp. Colloid Interface Sci. 2002, 7, 426.
Xu et al.
Figure 5. Images of adsorbed films for 0.005 wt % β-L at pH 5 with polystyrene particles after 20-h adsorption at different film areas: (a) A/A0 ) 1, before any area change; (b) after compression to A/A0 ) 0.48; (c) after expansion to A/A0 ) 1 and (d) after expansion to A/A0 ) 1.24. The arrows in (b) highlight examples of bands that appear to consist of intermittent chains of bright particles filling darker ‘channels’.
also subjected to a final expansion to A/A0 ) 1.24 and occasionally very thick bright bands were observed, such as that shown in Figure 5d. It was possible that the more narrow bands had merged to form these thicker bands. Thus, in the presence of latex particles, structuring within the films, which was not visible in the absence of the particles, became observable at both low and high adsorption times. One immediate question that arises is whether the structure is induced by the particles, or it is simply that the particles highlight structures that are formed already, without them. This question will be addressed later, but it should be noted that the particles are initially present at a low area fraction overall, with, on average, rather large distances between them. This average spacing does not change significantly as the protein is added and subsequently adsorbs at the interface after the particles have been spread. Furthermore, it is noticeable that, as protein adsorption proceeds, the latex particles become increasingly less mobile in the interface, so that certainly after 4-h adsorption from 0.005 wt % β-L there is very little relative movement of neighboring particles in the absence of film compression/expansion. There is some overall slow drift of sections of the interface, which agrees with the observed rotation of the banded domains, indicating that there is some capacity for more rearrangement of sections of the interface. Thus, initially at least, it seems reasonable to assume that the latex particles become embedded in the protein film as it builds up around them and that they remain essentially trapped in their relative local positions by the highly viscoelastic, gellike film that β-L forms on adsorption under such conditions.24 A greater degree of structuring was observable at pH 5 than at pH 7, and this is consistent with the presence of a film of higher viscoelasticity. However, the nature of the bright bands at this lower pH deserves further comment. It was noticeable that some of the bands appeared to consist of intermittent chains of bright particles filling darker ‘channels’sFigure 5b is an example of an image that shows this particularly well. Darker regions in the BAM images indicate less thick or less optically dense interfacial material. Thus, it is tempting to ascribe these feature to cracks or fissure in the protein film into which the latex particles concentrate. Piling up of protein to form thicker layers could also form bands of higher intensity, but in this case one might expect them to be of uniform intensity along their length. Thus, these images may provide further evidence to support the idea
Morphological Changes in Adsorbed Protein Films
Figure 6. Images of adsorbed films for 0.005 wt % β-L at pH 5 without polystyrene after compression to A/A0 ) 0.48 for (a) 4-h adsorption; (b) 15-h adsorption and after first compression cycle at 4 h.
that the particles simply get caught in features generated in the protein film, rather than the particles being the origin of those structural features. Experiments were also conducted at 10 times higher β-L concentration at pH 7, i.e., 0.05 wt %, without particles, to check if more rapid adsorption to higher protein surface loads would result in the sorts of features observed on compression/expansion of the 0.005 wt % β-L systems with particles. However, no structural features were observed, as found at 0.005 wt % without particles. Thus, it seems clear that limited protein adsorption at 0.005 wt % was not the reason that the band structuring was not observable at pH 7 this lower concentration. The similarity between the above types of structures observed via BAM and the behavior predicted by the earlier simulations16 of particle monolayers is worth noting. In most of the simulations, the films were subjected to similar area changes: uniaxial compression to A/A0 ) 0.60 or expansion to A/A0 ) 1.40. An increasing tendency for wrinkling of the films occurred when permanent bonds of increasing stiffness were introduced between the particles. The wrinkles formed at right angles to the direction of compression (as in the BAM images) and increased in width and thickness as material was forced to protrude into the aqueous phase. One effect of this was sometimes to decrease the net coverage of the interface on compression, leaving bare regions of interface. This supports the idea that in the BAM imaging the particles could highlight the same sort of wrinkling by falling into the adjacent gaps formed beside the wrinkles. The particles are much more visible than the wrinkles because the diameter of the particles is much greater than the initial thickness or width of the protein wrinkles. On uniaxial expansion, the simulations also predicted the formation of wide fissures in the films, when there was relatively strong bonding between the particles. Simulations were also performed for homogeneous (purely dilatational) compression of the films. Here the same sorts of structures were predicted, but they had no preferred orientation. The diamond shape of the barrier used in our Langmuir trough means that the compression is not purely dilatational. At low compression ratios it starts off as largely uniaxial, but on increasing compression ratios increasing shear is introduced into the deformation. Therefore, this may also be part of the explanation why the specific orientation of the bands perpendicular to the direction of compression is lost on increasing compression and subsequent expansion. It should also be noted that such wrinkling has been observed experimentally via fluorescence microscopy, for close packed monolayers of lung surfactants,23 for example. To investigate further the possible effect of the particles on the film structure, the preceding experiments at pH 5 were repeated without latex particles. Panels a and b of Figure 6 show examples of BAM images obtained after 4- and 20-h adsorption, respectively, for the same compression/expansion cycles as in Figure 3 after compression to A/A0 ) 0.48 without particles. Some similar banding structure was observed, but again, as at
Langmuir, Vol. 23, No. 9, 2007 5009
pH 7, much less structuring was observed in the absence of particles compared to in their presence. However, overall it is clear that there is more structuring at pH 5 than at pH 7, with or without particles, which is consistent an adsorbed film possessing a higher elastic modulus at pH 5 than at pH 7. Morphological and Structural Characteristics of OA Adsorbed Films. OA was adsorbed from 0.005 wt % solution and studied at pH 7 and 5 as for the β-L. Films were subjected to the same compression/expansion cycles as in Figures 3-6 (A/A0 ) 1 to 0.76 to 0.48 to 1), after 4-h adsorption and 20-21-h adsorption. In the absence of added latex particles at pH 7, no structuring was observed, only changes in intensity reflecting changes in the overall amount of protein adsorbed, as for β-L. Experiments were also conducted at pH 7 at 10 times higher OA concentration, i.e., 0.05 wt %, without particles, to check if lack of protein adsorption at the lower concentration was the reason for a lack of observed structuring. But again, as with as with β-L, this did not seem to be the case. At pH 5, after 4- and 21-h adsorption, the images obtained were essentially similar as those obtained at pH 7, although the occasional narrow bright band appeared, similar to those observed with β-L at this lower pH (see Figure 6). Also noticeable was the presence of a number of small, rounded brighter objects randomly situated throughout the film. Experience suggested that these were not dust particles, which are generally much larger and nonuniform. We postulate that these were protein aggregates. Such features have been observed previously with commercial whey protein,7 and OA is especially known to readily coagulate at interfaces.24 The isoelectric pH of OA is 4.825 and hydrophobic interactions are more predominant at a pH value closer to the isoelectric point (pI) of the protein.26 Consequently, the formation of stronger films is more likely at pH 5 than pH 7, as with β-L, but with OA this can lead additionally to surface coagulation and loss of surface activity. A few experiments were carried out at 0.05 wt % OA, and the main feature observed, particularly on compression, was a higher density of these small bright objects, merely reflecting the higher protein load and therefore the higher density of OA aggregates. When particles were included in the OA films at pH 7 the increase in the film structuring observed was remarkable. Figures 7 and 8 show representative images for 4- and 17-h old films, respectively, when subjected to similar compression/expansion cycles. Even the younger (4 h) films showed considerable structure, though it was clearly more pronounced for the older (17 h) films. The top left corner of Figure 7a illustrates the type of small bright objects observable before any compression, also in the absence of latex particles, as discussed above, that are characteristic of the OA films and that are thought to be aggregates of OA. The latex particles themselves tend to appear somewhat smaller, as can be seen in Figures 5 and 6, for example, for β-L. Figure 7d, for instance, seems to have two distinct populations of smaller and larger objects, though it must be admitted that it is not possible to unequivocally distinguish aggregates of protein from aggregates of latex particles in these images. However, the striking feature of the OA + particle systems is the appearance also of highly oriented, close-packed fiberlike structures, along with the small, bright, particulate features. In Figures 7 and 8 a few more images at different stages of compression/expansion have been included to capture all the different types of feature typically observed. For example, Figure 8b, after expansion to A/A0 ) 1.24, appears to show the closely packed fibers seen on compression to A/A0 ) 1 (Figure 8f) in a more expanded, less orientated form. The density of structures formed makes it difficult to tell how much of the observed brightness is due to particles
5010 Langmuir, Vol. 23, No. 9, 2007
Figure 7. Images of adsorbed films for 0.005 wt % OA at pH 7 with polystyrene particles after 4-h adsorption at different film areas: (a) A/A0 ) 1, before any area change; (b) after compression to A/A0 ) 0.48; (c) after expansion to A/A0 ) 0.76; (d) after expansion to A/A0 ) 1, (e) after expansion to A/A0 ) 1.24, (f) after compression to A/A0 ) 1.
Figure 8. Images of adsorbed films for 0.005 wt % OA at pH 7 with polystyrene particles after 17-h adsorption at different film areas: (a) A/A0 ) 1, before any area change; (b) after compression to A/A0 ) 0.76; (c) after compression to A/A0 ) 0.48; (d) after expansion to A/A0 ) 1, (e) after expansion to A/A0 ) 1.24, (f) after compression to A/A0 ) 1.
or coagulated protein. It is difficult to imagine how the initial low density of latex particles in the interface could alone generate all the structures observed or could induce them. Indeed, on closer examination, it does appear that some of the fiberlike stands are fissures with included particles, rather like the features observed with β-L at pH 5 (e.g., see Figure 5), though such fissures were much less prevalent with β-L. These sorts of fine,
Xu et al.
Figure 9. Images of adsorbed films for 0.005 wt % OA at pH 5 with polystyrene particles after 4-h adsorption at different film areas: (a) A/A0 ) 1, before any area change; (b) after compression to A/A0 ) 0.48; (c) after expansion to A/A0 ) 0.76; (d) after expansion to A/A0 ) 1; (e) after expansion to A/A0 ) 1.24, (f) after compression to A/A0 ) 1.
multiple-banded structures have been observed27 via light scattering microscopy for insoluble monolayers of 2-hydroxytetracosanoic, compressed at the A-W interface beyond the collapse surface pressure. The authors attributed them to bilayers suspended beneath the film, dangling into the aqueous phase. Thus, again it appears that structuring that is not visible without particles becomes much more visible with particles. With OA the structuring appears much more pronounced than with β-L, as might be expected for a protein that is more susceptible to forming aggregated films with stronger associative interactions between the adsorbed molecules. On this basis, even more pronounced structuring might be expected at pH 5, where electrostatic repulsion between the protein molecules is reduced. Indeed this is the case, as shown in Figures 9 and 10, for OA films at pH 5 after 4 and 22 h adsorption, respectively. Here large aggregates and fiberlike structures are observable even before any compression is applied (see Figures 9a and 10a). (In Figures 9 and 10 slightly different compression/expansion cycles were employed to try and capture the full range of structuring possible for the maximum area variation accessible with our Langmuir trough arrangement). Longer, thicker, and more coherent fiberlike structures are observed. Interfacial Dilatational Elasticity of β-L and OA Films on Compression/Expansion. In order to gain insight into the mechanical state of the films during the compression/expansion cycles, the dilatational elasticity () of the films was measured by monitoring the surface tension (γ) during such cycles. These measurements were conducted using the same apparatus under identical conditions as in the collection of the BAM images, but in separate experiments. This was necessary because it was not possible to measure γ and operate the BAM simultaneously due to the Wilhelmy plate interfering with the optical path of the laser. Also, if the laser was incident on the interface close to the Wilhelmy plate, very slight movements of the plate introduced
Morphological Changes in Adsorbed Protein Films
Langmuir, Vol. 23, No. 9, 2007 5011
Figure 10. Images of adsorbed films for 0.005 wt % OA at pH 5 with polystyrene particles after 22-h adsorption at different film areas: (a) A/A0 ) 1, before any area change; (b) after compression to A/A0 ) 0.48; (c) after expansion to A/A0 ) 1.24, (d) after compression to A/A0 ) 1.
Figure 11. (a) Change in surface tension (∆γ) for 0.005 wt % β-L at pH 5, after 20-2-h adsorption, when compressed from A/A0 ) 1 to 0.76. The smooth curve is the fit to the data. (b) The corresponding dilatational elasticity () versus change in area strain ∆ ln A, calculated from the fitted curve in (a). Table 1. Surface Tension Values (γ) Measured for the Principal Systems Studieda γ/(0.5 mN m-1 pH 5
pH 7
protein
4h
20 ( 2 h
4h
20 ( 2 h
β-L OA
54.6 59.2
58.1 61.9
55.6 65.6
54.5 62.2
a 0.005 wt % β-L and OA at 25 °C, at pH 5 and 7, after film ages of 4 and 20 ( 2 h, before any compression/expansion cycle was applied.
interfacial movements that blurred the images. All these measurements of γ were conducted in the absence of latex particles. Table 1 gives the absolute values of the surface tension measured for the main systems studied before any compression/ expansion cycle was applied. Figures 11 and 12 each give just one example of the typical changes in surface tension ∆γ for a compression and expansion stage, respectively. In all cases ∆γ was more noisy on compression than expansion, as can be seen by comparing Figure 11a with Figure 12a. This probably reflects structural fluctuations during collapse of the films, as they are
Figure 12. (a) Change in surface tension (∆γ) for 0.005 wt % β-L at pH 5, after 20-22-h adsorption, when expanded from A/A0 ) 0.48 to 0.76. The smooth curve is the fit to the data. (b) The corresponding dilatational elasticity () versus change in area strain ∆ ln A, calculated from the fitted curve in (a).
largely already in a condensed state, at or close to the equilibrium surface tension, in the compression part of the cycle. Nevertheless, due to the large number of data points collected, it was still possible to fit the noisy compression ∆γ versus time data to curves that described the overall trends in the data, which for compression always involved a further decrease in γ (increase in surface pressure). The smoother expansion data were easier to fit, where there was always an increase in γ (decrease in surface pressure) during the expansion. Usually, a simple double exponential decay (or rise) in ∆γ was sufficient to fit the data. However, the exact values of the fitting parameters used for each stage of the compression or expansion are not important. We do not consider it useful here to try and interpret them in any meaningful way, for example, to ascribe the quantities to particular relaxation mechanisms in the films, which are very complex.21 Rather the equation fitted to the ∆γ versus time data was simply differentiated to obtain the corresponding dilatational elasticity data, ) dγ/(d ln A). However, again it is interesting to compare the general behavior of the experimental results with those of the simulations.16 In the simulations, the stress in the plane of the interface can be calculated and correlated with observed structural rearrangements in the films, though the stress is obtained in reduced units, and so cannot be quantitatively compared with measured ∆γ in any meaningful way. Nevertheless, it is interesting that the simulations also show much larger fluctuations in the interfacial stress on compression than on expansion. The individual fluctuations corresponded to the desorption of individual particles (or small clusters) which release the stress by creating space in the interface. The fluctuations were stronger for stronger adsorption energies of the particles to the interface. Figures 11b and 12b give examples of the experimental values of calculated as described above, as obtained from the fits to the ∆γ versus time data in Figures 11a and 12a. As such, the calculated dilatational elasticities should strictly be referred to as complex elasticities because there will be viscous and elastic contributions to the magnitude of the stress (γ) change on deformation (A). Moduli calculated in this way are frequently used in interpreting surface pressure-area curves of adsorbed or spread molecules, where the gradient of the curve is called the ‘equilibrium’ film modulus, at some arbitrary slow rate of compression/expansion. Comparison of Figure 11a with Figure 12a shows that there was greater noise in the data on compression
5012 Langmuir, Vol. 23, No. 9, 2007
Xu et al.
Figure 13. Plots of the average dilatational elasticity (av) for 0.005 wt % β-L at the beginning of the various parts of the compression/ expansion cycles, versus A/A0 for (a) 4-h at pH 7; (b) 20-22-h at pH 7; (c) 4-h at pH 5 and (d) 20-22-h at pH 5. The arrows indicate the sequence of the area changes.
Figure 14. Plots of the average dilatational elasticity (av) for 0.005 wt % ΟΑ at the beginning of the various parts of the compression/ expansion cycles, versus A/A0 for (a) 4-h at pH 7; (b) 20-22-h at pH 7; (c) 4-h at pH 5, and (d) 20-22-h at pH 5. The arrows indicate the sequence of the area changes.
of the films, as compared to expansion. This was the case for all films studied. It is sufficient for our purposes here to take the initial value of measured at the start of each compression or expansion stage to indicate the mechanical strength of the film in order to compare the different systems. Because there is some noise in the data, the first five values of , i.e., over the first 0.5-1 s of the compression/expansion were averaged. The standard deviation (σ) about the mean value, av, was taken as indicative of the typical noise in the data (see Supporting Information). Figures 13 and 14 are plots of av versus the corresponding values of A/A0 at the start of the compression or expansion, for β-L and OA, respectively. Data are shown for films of age 4 and 21 h at pH 5 and 7. The arrows on the lines connecting the points indicate the sequence of changes in A/A0. Figures 13 and 14 reveal a number of general trends. For both proteins, peaks in occur at the start of the expansion after compression to the minimum area applied (A/A0 ) 0.48). Minima in generally occur between these two peaks, where the film is most expanded. This is understandable, since after compression to the minimum area the surface concentration and packing of adsorbed material will be greatest, but lowest when the film is most expanded. In all cases, values of for OA tend to be lower than those for β-L films under the same conditions, particularly for the case of the aged films. This includes the first measurement
of , just at the start of the first compression. Thus, the OA films appear to offer less resistance to deformation and this correlates with their greater apparent tendency to form fractures, etc., as evidenced by the BAM images. On the other hand, for the 20-h films was always much higher than for the 4-h films, particularly for β-L, despite the fact that the aged films generally exhibited more structuring and fracturing. Aging is expected to lead to stronger films due to further unfolding of the proteins with time and the increased development of intermolecular cross-linking. This was observed by us earlier7 in similar expansion experiments with β-L films, and the effect has been illustrated many times before,28 where both the shear and dilatational interfacial viscoelasticities continue to increase long after the surface protein concentration has essentially stabilized. β-L in particular seems capable of developing stronger cross-linking with time, perhaps even via disulfide bond interchange.29 Similarly, was found always to be much larger at pH 5 than at pH 7. As has already been mentioned, this can be explained by the reduction in the net protein charge at pH 5, leading to more close-packed films with increased capacity for intermolecular cross-linking.28 Overall then, the measured dilatational elasticities agree with the trends observed in the BAM images. However, despite the apparent structuring and fracturing apparently occurring through(29) Dickinson, E.; Matsumura, Y. Int. J. Biol. Macromol. 1991, 13, 26.
Morphological Changes in Adsorbed Protein Films
out the compression/expansion cycles at pH 5, for example, and the large changes in the values of , particularly for the 20-h films, it is interesting that the values after one complete cycle are close the values before the cycle, at corresponding values of A/A0. Thus, the changes in due to the large area changes appear to be reversible. This behavior has also been noted in the shear rheology of adsorbed protein films; although the films appear to have yield stresses30 or to be quite shear thinning,28 there is little hysteresis on remeasurement of the apparent shear moduli. This suggests a ‘healing’ mechanism operates in these condensed films, even though the films are broken into fragments. Under the conditions studied here there is always a sufficiently high protein concentration in the bulk to allow further adsorption to quickly fill any gaps in the film, and therefore perhaps to link together the fragments of pre-existing films before the film rheology is remeasured.
Conclusions BAM of adsorbed films of β-L and OA at the A-W interface has shown that extensive structuring occurs on compression and expansion of the films within the range of 48-124% of the initial film area. The structural features become much more apparent to the observer when a low area fraction of latex particles is initially present in the film before compression/expansion. These features include striated structures at which the particles appear to accumulate. The observed structures may be fissures in the adsorbed protein film or corrugations, i.e., thicker regions of protein. With both these proteins the features appear to form much more easily at pH 5 than pH 7, presumably due to pH 5 being closer to the isoelectric pH and therefore associated with the formation of initially stronger films. More interfacial structuring appears to occur with OA than with β-L, including the formation of a greater proportion of more spherical protein aggregates, consistent with the capacity for OA to undergo surface coagulation more readily. The dilatational extent of the strains of compression/expansion applied in this study were somewhat arbitrary. While the area changes were partly dictated by the geometry of the apparatus, they are, however, certainly representative of the extents of compression/expansion rates relevant to food foam processing, where flow deformations and bulk pressure changes commonly occur. Consequently, these observations can explain the high (30) Martin, A.; Bos, M.; Cohen-Stuart, M. C.; van Vliet, T. Langmuir 2002, 18, 1238. (31) Hotrum, N. E.; Cohen-Stuart, M. A. C.; van Vliet, T.; van Aken, G. A. Langmuir 2003, 19, 10210.
Langmuir, Vol. 23, No. 9, 2007 5013
susceptibility of protein-stabilized bubbles to coalescence under such conditions, as described elsewhere.1 Obviously, however, different specific morphological changes might be observed if substantially different rates and extents of interfacial area were to be applied, and hence this should be the subject of further study. For example, Hotrum et al.31 have clearly illustrated the gross fracture of protein films when they are continuously expanded with a close-packed layer of emulsion droplets injected beneath the film. Overall, computer simulations16 of adsorbed protein films, modeled as adsorbing particles possessing interparticle bonds of varying strengths, show remarkably similar behavior to the film structuring observed here experimentally. The formation of compressed bands of adsorbed material, adjacent to depleted regions of interface, and also cracks and fissures on expansion, appear to be typical of adsorbed films where strong attractive interactions between the film components can occur. The simulations also indicated that variations in the structuring occur depending upon the speed of compression or expansion. It is not simple to translate the simulation speeds (or times) into real values, but these variations indicate that the compression/ expansion speed is an important experimental variable that also should be tested further. Finally, one might ask how the typical thermal processing applied to foods might affect the susceptibility of such films toward aggregation, fracture, etc., and also does similar behavior occur at the oil-water interface? Such aspects are the subject of continuing study in our group. Acknowledgment. The authors would like to thank BBSRC (BBS/B/0501X) for funding this research and Mr. Phillip V. Nelson for designing and constructing the Langmuir trough apparatus. Supporting Information Available: Values of the av values plotted in Figures 13 and 14 and the associated noise (σ); the complete set of images for the compression/expansion cycle for 0.005 wt % β-L at 25 °C and pH 7 without polystyrene particles after 4-h adsorption, with polystyrene particles after 4-h adsorption, and with polystyrene particles after 15-h adsorption; the complete set of images for the compression/expansion cycle for 0.005 wt % β-L at pH 5 without polystyrene particles after 4-h adsorption and after 20-h adsorption; the complete set of images for the compression/expansion cycle for 0.005 wt % OA at pH 5 without polystyrene particles after 4-h adsorption and 21-h adsorption. This material is available free of charge via the Internet at http://pubs.acs.org. LA063280Q
