- Email: [email protected]
Elucidating the relationship between the spreading coefficient, surface-mediated partial coalescence and the whipping time of artificial cream
Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
Elucidating the relationship between the spreading coefficient, surface-mediated partial coalescence and the whipping time of artificial cream Natalie E. Hotrum a,b,1 , Martien A. Cohen Stuart b , Ton van Vliet a , Serena F. Avino a , George A. van Aken a,c,∗ b
a Wageningen Centre for Food Sciences (WCFS), P.O. Box 557, 6700 AN Wageningen, The Netherlands Wageningen University, Laboratory of Physical Chemistry and Colloid Science, P.O. Box 8038, 6700 EK Wageningen, The Netherlands c NIZO Food Research, P.O. Box 20, 6710 BA Ede, The Netherlands
Received 10 November 2004; accepted 18 March 2005 Available online 10 May 2005
Abstract We studied the whipping of artificial creams composed of a blend of sunflower oil and hydrogenated palm fat stabilized by protein or a mixture or protein and low molecular weight (lmw) surfactant. It was found that an increased whipping speed, decreased protein concentration, and the addition of lmw surfactant leads to shorter whipping times. Further, shorter whipping times were observed for WPI-stabilized cream compared to cream stabilized by sodium caseinate. In all cases, the decrease in whipping time was due to a decrease in the length of the second stage of whipping, the stage characterized by the adhesion of fat droplets to the air bubble surface. The decrease in whipping time could be accounted for by considering the influence of the experimental variables on the fraction of bubble surface area at which fat droplet spreading is possible. The same changes in parameters that promote droplet spreading at the air/water interface cause a decrease in the whipping time of our model creams. Correlating the whipping time of cream with the spreading behavior of fat droplets at the air/water interface represents a new insight into the mechanisms involved in the whipping of cream. © 2005 Elsevier B.V. All rights reserved. Keywords: Whipped cream; Partial coalescence; Spreading coefficient; Low molecular weight surfactant; Dairy foam
1. Introduction With a current trend in food processing, being to provide the consumer with food products with reduced fat and lower levels of saturated fat, there is a demand for a deeper understanding of how fat gives structure and texture to foods. Whipped cream is an example of a food system for which the structural characteristics are dependent on the presence of crystalline fat. It is generally accepted that air bubbles in whipped cream are stabilized by a network of partially coalesced fat globules [1–8]. The transformation of cream ∗
Corresponding author. Tel.: +31 318 659 589; fax: +31 317 482 358. E-mail address: [email protected] (G.A. van Aken). 1 Present address: Numico Research, P.O. Box 7005, 6700 CA Wageningen, The Netherlands. 0927-7757/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfa.2005.03.004
into whipped cream by beating air into the system proceeds as follows. First, large air bubbles are introduced and broken up into smaller ones by the whipping action of the whisks. This initial foam is predominantly stabilized by protein [1,2]. This is followed by a gradual accumulation of fat droplets on the surface of the air bubbles. This process has been well characterized; a number of electron microscopy studies clearly show fat droplets adhered to the air bubble surface [2,5,7,9]. Finally, a partially coalesced network of fat droplets is built up, which acts to trap the air bubbles, retains the serum phase and gives stiffness to the whipped cream. A number of physical definitions for the endpoint of whipping can be found in the literature including, maximum overrun, index of globule clumping and maximum stiffness, to name a few. Regardless of the definition, it is generally accepted that further whipping action, after the endpoint of
72
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
whipping, leads to the formation of butter granules and foam collapse. In a study on the whipping of natural cream, van Aken [8] distinguishes the three stages in the whipping process described above based on the mechanical resistance required to maintain a constant whisk speed. Depending on the time to reach the endpoint of whipping, the three stages either appear as distinct phases in the mechanical resistance versus time plot, or may overlap to some extent. It was shown that at moderately low rates of the whisks, long whipping times were obtained even though the evolution of the overrun (introduction of air) was not significantly reduced. The long whipping times under these conditions were shown to be related to a long second stage of the whipping process, due to slow adherence of fat globules to the bubble surfaces [8]. While there is a general understanding of the physical changes that take place during the whipping of cream, the mechanisms driving the transformation of cream into whipped cream remain poorly understood. In this paper, we will focus on the interaction between fat droplets and air bubbles during the second stage of whipping. This is the stage during which fat droplets accumulate at the air bubble surface. During this stage, air–bubble interfaces are continuously expanded and compressed as bubbles form, break-up and coalesce; this dynamic nature of the air–bubble interface can lead to the necessary interfacial conditions for fat droplets to adhere to the air bubbles. Adsorbed fat droplets release some liquid fat, which spreads onto the bubble surface, leading to interfacial fat globule flocculation and aggregation [3,6], or surface-mediated partial coalescence between the neighboring adhered droplets. Thus, adhesion of fat droplets to the air bubbles is an important step in the build-up of the partially coalesced fat droplet network, which ultimately stabilizes the whipped cream. In addition to surface-mediated partial coalescence, shear-induced partial coalescence also takes place during the whipping of cream. Both partial coalescence mechanisms are represented schematically in Fig. 1. The latter partial coalescence mechanism is the result of sticking of emulsion droplets when they collide in the bulk as a result of convection [10,11]. Under conditions of laminar flow, shear-induced partial coalescence proceeds like orthokinetic flocculation [12]; the decrease in number of droplets, N, in time (-dN/dt) can be described by: −
dN 4 = ϕNΨα dt π
(1)
where ϕ is the volume fraction of the dispersed phase, Ψ is the velocity gradient (or shear rate), and α is the capture efficiency. Capture efficiency, α, represents the probability that two particles stick when they encounter each other, and is influenced by four parameters [12]: protruding crystals, contact angle, colloidal repulsion and permanence of junctions. Under turbulent conditions, such as can be expected during whipping, the rate of partial coalescence becomes far more complex than that predicted by (1); however, the
Fig. 1. Schematic depiction of two possible mechanisms for partial coalescence. In the surface-mediated mechanism (a), air bubbles play an important role. When Sdyn > 0 (2) fat droplets can attach to the protein-covered air bubble surface and interfacial flocculation takes place, partial coalescence (A) of the interfacial flocs ensues. If, at this point, the bubble bursts (C), a partially coalesced fat clump remains. If the air bubble remains stable (B), a fat clump from the bulk may partially coalesce with the adsorbed fat clump (D). The shear-induced mechanism for partial coalescence (b) is independent of the presence of air. Partial coalescence (E) occurs when two droplets collide and the fat crystals in the droplets pierce the film between them. Larger fat clumps can form (F) when small clumps collide and partially coalesce with other clumps or free droplets ((a) reproduced from [26]).
equation remains valid in the sense that it gives an indication of the rate-determining system parameters. Shear-induced partial coalescence is independent of the presence of air bubbles in the system; it has been shown that cream can be churned into butter (a partial coalescence driven process) in the absence of air [6,13]. However, churning proceeds much faster in the presence of air [6], which may be explained by the additional effect of surface-mediated partial coalescence. Presumably, the presence of air bubbles increases the value of the capture efficiency, for example, by bringing the fat globules into contact by immobilizing them at the air bubble surface, or by improving the permanence of the junctions between partially coalesced fat globules due to oil spreading out of the adjacent droplets. The adhesion of fat droplets to the air/water interface is a critical precursor to surface-mediated mediated partial coalescence. The interaction of emulsion droplets with an air/water interface has been extensively studied by us [14–17]. This process involves the entering of droplets into the interface, followed by spreading of liquid oil over the interface. The spreading is self-enhancing, meaning the spreading of a first droplet promotes the entering and spreading of other droplets [14,15]. Whether or not oil spreading will occur is dependent on the balance of surface tensions (γ) at the three-phase boundary between the oil (O), water (W) and air (A) interfaces [14,18]. This balance of forces is described by the spreading coefficient, S: S = γAW − (γOW + γOA )
(2)
When S is positive, oil spreads. Triglyceride oils with medium to long chain fatty acids, such as soybean, sunflower, or liquid butter oil, have S > 0, and can, therefore, spread at the clean air/water interface [14,16,19]. Oil spreading is inhibited if γ AW becomes so low that the condition
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
γ AW > γ OW + γ OA is no longer satisfied, for example, due to the presence of an adsorbed protein layer [14]. This has important consequences for the adhesion of fat droplets to the air bubble surface during the whipping of cream, since air bubble surfaces are initially stabilized by adsorbed protein [1,2]. However, whipping is a dynamic process, and likely the surface tension of the air bubbles is also dynamic. For protein-stabilized air bubbles, Sdyn may become positive due to an increase in γ AW,dyn resulting from local surface expansion of the air bubbles during whipping. In this way, fat droplets can adhere to the dynamic bubble surface. In our work on the spreading behavior of proteinstabilized sunflower oil emulsion droplets at protein solution/ air interfaces [14,20], it was observed that oil droplet spreading occurred at protein concentrations and expansion rates that yielded steady-state surface pressures of less than or equal to 15 mN m−1 . It was, therefore, concluded that the air/water surface pressure, Π AW = 15 mN m−1 , represents a critical surface pressure, Π cr , below which spreading can occur for the sunflower oil/protein solution/air system. The relatively high value of Π cr indicates that the air/water interface does not need to be completely void of adsorbed protein for spreading to occur. Using known values for γ OW and γ OA one can show that Π cr is precisely the value satisfying the condition Sdyn = 0. Moreover, for protein-stabilized emul0 during spreading, sions, the value of Π cr implies γ OW = γOW 0 where γOW is the oil/water surface tension of the clean interface. This last point is not the case in the presence of the low molecular weight surfactants, Tween 20 and Span 80, which when present at sufficiently high concentrations, can cause an increase in Π cr of a few mN m−1 due to their ability to lower γ OW at the expanding oil/water interface during oil spreading, thus facilitating oil droplet spreading [17]. In this paper, we investigate the influence of whipping speed, protein type, protein concentration and the presence of low molecular weight (lmw) surfactants on the duration of the whipping process. The results are evaluated taking into account our previous work on oil droplet spreading at the air/water interface yielding a new model, which correlates the whipping time of cream with the spreading behavior of emulsion droplets at the air/water interface. The effect of slight clustering of natural milk fat globules, which is often reported to improve the whipping characteristics of natural cream, and probably originates from specific properties of the natural milk fat globule membrane, does not occur in our artificial cream and is, therefore, not included in this study.
2. Materials and methods 2.1. Materials Whey protein isolate (WPI) (BiPRO, lot no. JE 052-9420, Davisco Foods International, Le Sueur, MN 56058) and
73
sodium caseinate (sodium caseinate S, DMV International, Veghel, The Netherlands) were used in the preparation of protein solutions and emulsions. WPI and caseinate powders contained 95 and 86% protein, respectively (Biuret standard assay, in agreement with manufacturer’s specifications). Proteins were dissolved in 20 mM imidazole buffer (pH 6.7) containing 0.1 M NaCl by stirring for 1 h at room temperature. The chemicals used for the buffers were purchased from Merck (Darmstadt, Germany, analytical grade). For the study of the influence of lmw surfactant on the whipping properties of cream, either water-soluble Tween 20 (polyoxyethylene sorbitan monolaurate, Merck), or oil-soluble Span 80 (sorbitan monooleate, Fluka, Switzerland) was used. The fat phase of the emulsions consisted of a blend of 35 wt.% fully hydrogenated palm fat (Grinsted PS101, Danisco Cultor, Denmark) and 65% (w/w) sunflower oil (Reddy, Vandermoortele, Roosendaal, The Netherlands). 2.2. Emulsion preparation Emulsions were prepared containing 40 wt.% fat blend and 1 wt.% protein. The fat blend was warmed under gentle stirring at 65 ◦ C for 1 h to melt the fat. Prepared protein solutions were warmed to 65 ◦ C, prior to mixing with the fat blend. The mixture was then pre-homogenized at 65 ◦ C using an Ultraturrax (IKA, Staufen, Germany). Finally, the emulsions were prepared by homogenization at 2 MPa in a Rannie laboratory homogenizer, warmed at 65 ◦ C for 10 passes of the pump. After homogenization, emulsions were rapidly cooled in an ice bath to 7 ◦ C. Emulsions were stored overnight at 7 ◦ C before performing the experiments. The emulsions had a monomodal droplet size distribution with an average droplet diameter (d3,2 ) of ∼1 m, as determined by light scattering using a Coulter Laser LS230 (Coulter Electronics, Mijdrecht, The Netherlands). For emulsions with added Span 80, the required amount of surfactant was dissolved in the fat blend during warming. In the case of added Tween 20, an emulsion with a slightly higher fat and protein content was prepared. After homogenization and initial cooling, this emulsion was diluted with Tween 20, dissolved in buffer to give the desired concentration, followed by gentle mixing. 2.3. Whipping of emulsions Emulsions were whipped at 7 ◦ C in an automated whipping apparatus (Ledoux b.v., Dodewaard, The Netherlands). The same instrument was also used by van Aken [8] in a study on the aeration of natural whipped cream. A 250 mL emulsion sample was gently poured into a stainless steel or glass beaker (diameter, 9.6 cm). This was placed inside the thermostat jacket of the whipping apparatus, where the temperature was controlled within 0.5 ◦ C during whipping. The whipping action was provided by two whisks (radius, 20 mm; length, 113 mm; composed of six, 1 mm thick stainless steel wires) rotating in a planetary motion pattern. This motion of the whisks is designed to maximize the
74
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
Fig. 2. Power input transferred by the whisks vs. time during the whipping of WPI-stabilized model cream. Numbers near the curves denote the rotational speed of the whisks in rpm. Reproduced from [26].
incorporation of air and to maintain homogeneous whipping conditions over the entire volume. The electrical current required to maintain a constant rotational speed was monitored during the whipping process. This current is a measure of the energy dissipation in the system, and is the characteristic of its mechanical properties. As a result of whipping, the aerated emulsion eventually reaches a maximum stiffness, which is registered as a peak in the electrical current signal. We define this peak as the endpoint of whipping, and refer to the time to reach this peak as the whipping time. Unless otherwise stated, all experiments were performed using a rotational speed of 584 rpm. All experiments were performed in duplicate.
3. Results In Fig. 2, the power transferred by the whisks during whipping of emulsion, also referred to as (artificial) cream,
stabilized by 1% WPI is plotted as a function of time for various whipping speeds. The peak of maximum resistance defines the endpoint of whipping, and the time to reach this peak is the whipping time. The whipping times measured for our model creams were significantly longer and had lower values for the peak of maximum resistance than those reported for natural cream whipped using the same apparatus [8]. However, the shape of the curves in Fig. 2 and the microscopy images in Fig. 3 suggest that despite the longer whipping times, the whipping process for our artificial cream follows a similar mechanism to that of natural whipping cream. For example, for the cream whipped at a rotational speed of 584 rpm (Fig. 2), the three stages of whipping described by van Aken [8] can be clearly identified. In the first stage, large air bubbles are incorporated into the emulsion and broken up into smaller bubbles; this is characterized by a small increase in the power-input signal (Fig. 2). During this first stage, the foam is considered to be protein-stabilized [1,2]. A few emulsion droplets are observed to adhere to the air bubble surface, but the air bubbles are not yet fully coated (Fig. 3b). In the second stage of whipping, the coating of emulsion droplets at the air bubble surface gradually becomes denser (Fig. 3c) and there is a gradual slight increase in the power transferred by the whisks (Fig. 2), which indicates an increasing viscosity of the system. Finally, in the third, the size of the fat clumps increases a network of partially coalesced emulsion droplets is formed (Fig. 3d). This partially coalesced network gives stiffness to the cream, resulting in a sharp increase in the power input (Fig. 2). In Fig. 4, whipping time is re-plotted as a function of rotational speed for WPI-stabilized cream as well as for sodium caseinate-stabilized cream. For both systems, the whipping time begins to increase sharply for whipping speeds lower
Fig. 3. Light micrographs of the cream stabilized by 1% WPI whipped at a rotational speed of 584 rpm: (a) before whipping; (b) after 50 s; (c) after 290 s; and (d) after 473 s. Time to reach endpoint of whipping, 473 s; length of bar = 100 m.
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
75
Fig. 4. The influence of rotational speed on whipping time for creams stabilized by either WPI () or sodium caseinate (). Dotted lines serve to distinguish between data sets.
than 600 rpm. van Aken [8] attributes the strong dependence of whipping time on rotational speed to a variation in the rate of transformation of free fat globules into clumps driven by surface-mediated partial coalescence during the second stage of whipping. Fig. 4 also shows that WPI-stabilized creams give shorter whipping times than sodium caseinate-stabilized creams, a trend that was also observed by van Camp et al. [21] for butterfat emulsions stabilized by whey protein concentrate (WPC) compared to those stabilized by sodium caseinate. This observation suggests that the transition from free fat droplets to clumped fat proceeds faster for WPI-stabilized creams than for sodium caseinate-stabilized creams. Protein concentration had an effect on the whipping time of WPI-stabilized model cream (Fig. 5). There is a significant increase in the whipping time when the protein concentration is increased from 1 to 3 wt.%; however, the whipping time does not change much between the creams containing 0.6 and 1 wt.% WPI. These results are in agreement with van Camp et al. [21], who observed a decrease in whipping time with decreasing WPC concentration (from 3.1 to 0.42 wt.%), which below 1.6 wt.% appeared to level off to a plateau value. Further, these results are in qualitative agreement with Needs and Huitson [9], who observed a 50% decrease in whipping time for natural cream, from which the milk serum proteins had been removed, compared to unadulterated cream. We observed a slight decrease in the overrun of the
Fig. 5. The influence of protein concentration on whipping time () and overrun () of WPI-stabilized creams. Dotted lines serve to distinguish between the data sets. Whisk rotational speed: 584 rpm.
Fig. 6. Whipping time () and overrun () of WPI-stabilized creams with added (a) Tween 20 (in water phase) or (b) Span 80 (in oil phase). Dotted lines serve to distinguish between the data sets. The values measured in the absence of surfactant are shown on the primary y-axis before the break. Whisk rotational speed: 584 rpm.
emulsions with increasing protein concentration (Fig. 5). Needs and Huitson [9] reported no significant effect of protein concentration on the overrun, nor did van Camp et al. [21] in the case of WPC-stabilized model creams. This difference between our experimental results and the literature could be due to a difference in the whipping apparatus used. The influence of added lmw surfactant on the whipping time and overrun of the model creams was also investigated. In Fig. 6, the whipping times are shown for WPI-stabilized emulsions containing added Tween 20 (Fig. 6a) or Span 80 (Fig. 6b). In both the cases, we observed a decrease in whipping time with increasing surfactant concentration, the effect being stronger for Tween 20 than for Span 80, especially at lower concentrations. For the emulsions containing 2.7 and 5.5 mM Tween 20, the whipping time was decreased to such an extent that it was similar to the whipping time of natural cream whipped at the same rotational speed using the same apparatus as reported by van Aken [8]. A decrease in whipping time in the presence of added lmw surfactant has also been reported for recombined cream to which glycerol monostearate had been added [22]. Further, our results are in qualitative agreement with the observation reported in the literature, that destabilization of ice cream mix emulsions subjected to shear in the presence of air proceeds more rapidly in the presence of added emulsifier [23,24]. Over the range of concentrations used, the overrun was not influenced by the addition of Span 80 (Fig. 6b). This is also the case for Tween 20 (Fig. 6a), with the exception of the highest concentration (5.5 mM Tween 20), for which a somewhat lower overrun was found; this system also showed a different shape of the power input curve (see below). The power-input profiles for the creams with added surfactant (Fig. 7) were similar in shape to those for the WPIstabilized model cream in the absence of surfactants (Fig. 2). The main difference is that with the increasing surfactant
76
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
Fig. 7. Power input transferred by the whisks vs. time for the whipping of cream containing 1% WPI with added (a) Tween 20 or (b) Span 80. Numbers near the curves denote the lmw surfactant concentration in mM. Whisk rotational speed: 584 rpm ((a) reproduced from [26]).
concentration, a steady decrease in the length of the second stage of whipping is observed, which strongly suggests that surfactants play an important role in the adhesion and subsequent surface-mediated partial coalescence of emulsion droplets at the air/water interface. The sharpness of the peak in the third stage of whipping, the stage characterized by the build-up of a partially coalesced network, did not appear to differ between systems in the presence (Fig. 7) or absence (Fig. 2) of surfactant. The shape of power-input profile for the cream with the highest concentration of Tween 20 (5.5 mM, Fig. 7a) differed from the other samples. In view of the low overrun for this concentration, this may suggest that there is an optimum between accelerating the whipping process and forming a whipped cream with desirable structural characteristics.
4. Discussion We have demonstrated that whipping speed, protein type, protein concentration and the presence of lmw surfactant all influence the whipping time of our artificial creams (Figs. 4–6). The differences in whipping time were observed to result from a variation in the length of the second stage of whipping. This stage is characterized by the accumulation of fat globules at the air/water interface. Thus, our results strongly suggest that the whipping time is mainly determined by the rapidity and ease with which emulsion droplets can attach to and spread at the air/water interface. In the following discussion, we present a model in which the effect of the variables investigated on the whipping time should is explained in terms of the spreading behavior of emulsion droplets at the planar air/water interface. Brooker et al. [2] showed that the incorporation of air is not a sufficient stimulus per se to cause fat droplets to penetrate the air/water interface on a large scale. In our work on the spreading behavior of emulsion droplets at the planar air/water interface [14,15,20], we showed that an adsorbed
protein layer can effectively inhibit droplet entering and spreading under quiescent and dynamic conditions, provided it maintains a pressure exceeding a critical value, Π cr . During aeration, the air bubbles are initially protein-stabilized [1,2]; thus, it seems reasonable to assume that during whipping, fat droplet adsorption and spreading can only occur at the portion of the air bubble surface for which the air/water surface pressure is lower than Π cr . Moreover, adsorption and spreading of fat droplets at the air bubble surface is the precursor to surface-mediated droplet coalescence (Fig. 1). We, therefore, expect that manipulating the fraction of the total bubble surface area for which Sdyn > 0 will influence the ease and rapidity of fat droplet adhesion and spreading, ultimately governing the whipping time of the system. This leads to the following scenario. During the second stage of whipping, the air bubbles are in a dynamic state of formation, break-up and coalescence, during which their surfaces are continually expanded and compressed. As a result, at a given moment in time there will be a distribution of dynamic air/water surface pressures; ranging from low values for newly formed air bubble surfaces to high values for portions of the air bubble surface that are compressed. This distribution is represented schematically in Fig. 8a. The total area under the curve represents the sum of the surface area of the individual air bubbles, or the total air bubble surface available in the whipped cream. The Π cr required for the fat droplet spreading is indicated by a vertical dotted line. The area under the curve to the left of this line is shaded to indicate that for this fraction of the total air bubble surface, the air/water surface pressure will be low enough for adsorption and spreading of fat droplets to occur; the remainder of the bubble area is inactive. First, let us assume that the situation in Fig. 8a represents a reference case. If the surface pressure distribution is shifted to the left with respect to the reference case, as in Fig. 8b, we expect the fraction of the total bubble surface area, for which Sdyn > 0, to increase and the whipping time to decrease. In our experiments at the planar air/water interface [20], we observed that, for a protein solution, lower steady state surface pressures (i.e. a shift in the surface pressure distribution in Fig. 8a to the left) could be achieved by increasing the relative expansion rate, decreasing the protein concentration or using WPI instead of sodium caseinate. Similarly, for the creams investigated, increasing the rotational speed (Figs. 2 and 4), decreasing the protein concentration (Fig. 5) or using WPI instead of sodium caseinate (Fig. 4) resulted in shorter whipping times. Conversely, if the surface pressure distribution is shifted to the right with respect to the reference case, as in Fig. 8c, we expect the fraction of the total bubble surface area for which Sdyn > 0 to decrease and the whipping time to increase. This can be achieved by decreasing the relative expansion rate/rotational speed, increasing the protein concentration or using sodium caseinate instead of WPI. The experimental results presented in Figs. 4 and 5 confirm this, supporting the concept that the whipping time is related to the ease with
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
Fig. 8. Schematic representation of the distribution of surface pressure values present at the surface of air bubbles during whipping. Vertical line indicates the critical surface pressure for spreading. For a given system (a), the shaded area under the curve indicates the fraction of the air bubble surface area for which fat droplet spreading can occur. This area can be increased or decreased by shifting the surface pressure distribution to the left (b) or to the right (c), respectively. In addition, if the critical surface pressure is shifted to a higher value (d), the fraction of the bubble surface area for which Sdyn > 0 will increase. As explained in Section 4, increasing the proportion of the bubble surface area for which Sdyn > 0 leads to shorter whipping times, whereas decreasing the area leads to longer whipping times. Reproduced from [26].
which fat droplets can adhere to and spread at the air/water interface. As an additional point, in an earlier paper [15] we observed that -lactoglobulin forms brittle adsorbed layers at the air/water interface, leading to entering and spreading of oil droplets over a much larger area in a single event than for -casein, which forms more fluid-like layers. This may serve as an additional explanation for the shorter whipping times observed for the WPI-stabilized compared to the sodium caseinate-stabilized creams (Fig. 2). The line of reasoning depicted by Fig. 8 can also be used to explain why the creams stabilized by either 0.6 or 1 wt.% WPI had similar whipping times, while the cream stabilized by 3 wt.% WPI had a significantly longer whipping time (Fig. 5). Spreading can occur as long as Sdyn > 0. If, for the 1 wt.% WPI cream, the majority of the surface pressure distribution is already to the left of the critical surface pressure, then shifting the distribution even further to the left by lowering the protein
77
content will no longer significantly influence the whipping time. However, increasing the protein concentration will shift the distribution to the right and, with that, the whipping time of the cream will increase. For the variables discussed above (rotational speed, protein type and protein concentration) the value of Π cr was constant. In previous work on spreading behavior of emulsion droplets at the planar air/water interface [17], we showed that surfactants facilitate droplet spreading by shifting Π cr to higher Π AW values. Correspondingly, in the whipping experiments (Fig. 7), we observed a significant decrease in the length of the second stage of whipping with increasing surfactant concentration. Thus, the decrease in whipping time in the presence of surfactants further strengthens our hypothesis that whipping time is dependent on the rapidity and ease with which emulsion droplets can adhere to and spread at the air/water interface. This is depicted schematically in Fig. 8d. It should be mentioned that the whipping time of the WPIstabilized cream is observed to begin to decrease (Fig. 6) at surfactant concentrations that are considerably lower than those required for an increase in the value of Π cr for -lactoglobulin stabilized emulsion reported in reference [17]. The reason for this is not completely clear. A possible explanation is that, in reference [17], the increase in the value for Π cr was measured as a function of the surfactant concentration in an emulsion that was added as a tiny droplet to a bulk solution of protein, which did not contain surfactant. Because the surfactant slowly diffuses out of the droplet, the actual surfactant concentration at the point of measuring Π cr will have been considerably lower than the initial concentration. Until now, researchers have tried to explain the influence of whipping speed, protein concentration, protein type and the presence of lmw surfactant on the whipping time of cream in terms of the static properties of the emulsion droplets and their adsorbed layers, such as adsorbed amount, relative strength of the adsorbed protein layer, size and orientation of fat crystals, or contact angle of the fat crystals at the oil/water interface. These properties of the emulsion droplets are of relevance to the probability of a fat crystal piercing the film between colliding droplets, which is an important parameter controlling the rate of shear-induced partial coalescence, and probably a significant parameter controlling the formation of a network of partially coalesced fat droplets during the third stage of the whipping process. However, we have shown that the whipping time is largely controlled by a process of surface-mediated partial coalescence. Therefore, an explanation for rate of partial coalescence based solely on the static properties of the emulsion droplets and oil/water interfaces fail when applied to emulsions sheared in the presence of air. The conceptual model presented in Fig. 8 demonstrates that the influence of rotational speed, protein concentration, protein type and the presence of lmw surfactants on the whipping time of cream (Figs. 4–6 and [8,9,21–23]) can be well accounted for in terms of the influence of these variables on the spreading behavior of emulsions. This model (Fig. 8), when combined with the existing knowledge of the factors promot-
78
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
ing shear-induced partial coalescence (1), can more fully account for the dependence of the whipping time of the creams studied in this work on the experimental system parameters (whisk rotational speed, protein concentration, protein type, lmw surfactant concentration). The industry has long been aware of the fact that in order to improve the whipping properties of recombined cream, lmw surfactants must be added. The mechanism of surface-mediated partial coalescence and how it can be controlled (Fig. 8) helps to explain why this is so. It would be interesting to determine if the model presented in Fig. 8 can be applied to explain why homogenized cream tends to have much longer whipping times than natural cream [5,7]. Conceivably, the abundant presence of small, surface active molecules in the milk fat globule membrane of fat globules in natural cream [6] could shift Π cr to a higher value compared to homogenized cream, where the droplets are predominantly protein-stabilized [25]. This is the subject of ongoing research. Acknowledgement Danisco Ingredients, Denmark is gratefully acknowledged for providing the fully hydrogenated palm fat. References [1] M. Anderson, B.E. Brooker, E.C. Needs, in: E. Dickinson (Ed.), Food Emulsions and Foams, Royal Society of Chemistry, London, 1987, p. 100. [2] B.E. Brooker, M. Anderson, A.T. Andrews, Food Microstruct. 5 (1986) 277. [3] D.F. Darling, J. Dairy Res. 49 (1982) 695.
[4] H.D. Goff, J. Dairy Sci. 80 (1997) 2620. [5] E. Graf, H.R. M¨uller, Milchwissenschaft 20 (1965) 302. [6] H. Mulder, P. Walstra, The Milk Fat Globule, Pudoc, Wageningen, 1974. [7] D.G. Schmidt, A.C.M. van Hooydonk, Scanning Electron Microsc. III (1980) 653. [8] G.A. van Aken, Colloids Surf. A 190 (2001) 333. [9] E.C. Needs, A. Huitson, Food Struct. 10 (1991) 353. [10] K. Boode, P. Walstra, Colloids Surf. A 81 (1993) 121. [11] M.A.J.S. van Boekel, P. Walstra, Colloids Surf. 3 (1981) 109. [12] P. Walstra, Physical Chemistry of Foods, Marcel Dekker, New York, 2003. [13] J.H. Labuschagne, Churning in the absence of air, Thesis, Wageningen Agricultural University, The Netherlands, 1963. [14] N.E. Hotrum, T. van Vliet, M.A. Cohen Stuart, G.A. van Aken, J. Colloid Interface Sci. 247 (2002) 125. [15] N.E. Hotrum, M.A. Cohen Stuart, T. van Vliet, G.A. van Aken, Langmuir 19 (2003) 10210. [16] N.E. Hotrum, M.A. Cohen Stuart, T. van Vliet, G.A. van Aken, Colloids Surf. A 240 (2004) 83. [17] N.E. Hotrum, Emulsion droplet spreading at air/water interfaces: mechanisms and relevance to the whipping of cream, Thesis, Wageningen University, The Netherlands, 2004. [18] W.D. Harkins, A. Feldman, J. Am. Chem. Soc. 44 (1922) 2665. [19] E.P. Schokker, M.A. Bos, A.J. Kuijpers, M.E. Wijnen, P. Walstra, Colloids Surf. B 26 (2003) 315. [20] N.E. Hotrum, M.A. Cohen Stuart, T. van Vliet, G.A. van Aken, in: E. Dickinson, T. van Vliet (Eds.), Food Colloids, Biopolymers and Materials, Royal Society of Chemistry, Cambridge, 2003, p. 192. [21] J. van Camp, S. van Calenberg, P. van Oostveldt, A. Huyghebaert, Milchwissenschaft 51 (1996) 310. [22] J.G. Zadow, F.G. Kieseker, Aust. J. Dairy Technol. 30 (1975) 114. [23] B.M.C. Pelan, K.M. Watts, I.J. Campbell, A. Lips, J. Dairy Sci. 80 (1997) 2631. [24] H.D. Goff, W.K. Jordan, J. Dairy Sci. 72 (1989) 18. [25] D.F. Darling, D.W. Butcher, J. Dairy Res. 45 (1978) 197. [26] N.E. Hotrum, M.A. Cohen Stuart, T. van Vliet, G.A. van Aken, in: E. Dickinson (Ed.), Food Colloids, Interactions, Microstructure and Processing, Royal Society of Chemistry, Cambridge, 2005, p. 317.
Elucidating the relationship between the spreading coefficient, surface-mediated partial coalescence and the whipping time of artificial cream Natalie E. Hotrum a,b,1 , Martien A. Cohen Stuart b , Ton van Vliet a , Serena F. Avino a , George A. van Aken a,c,∗ b
a Wageningen Centre for Food Sciences (WCFS), P.O. Box 557, 6700 AN Wageningen, The Netherlands Wageningen University, Laboratory of Physical Chemistry and Colloid Science, P.O. Box 8038, 6700 EK Wageningen, The Netherlands c NIZO Food Research, P.O. Box 20, 6710 BA Ede, The Netherlands
Received 10 November 2004; accepted 18 March 2005 Available online 10 May 2005
Abstract We studied the whipping of artificial creams composed of a blend of sunflower oil and hydrogenated palm fat stabilized by protein or a mixture or protein and low molecular weight (lmw) surfactant. It was found that an increased whipping speed, decreased protein concentration, and the addition of lmw surfactant leads to shorter whipping times. Further, shorter whipping times were observed for WPI-stabilized cream compared to cream stabilized by sodium caseinate. In all cases, the decrease in whipping time was due to a decrease in the length of the second stage of whipping, the stage characterized by the adhesion of fat droplets to the air bubble surface. The decrease in whipping time could be accounted for by considering the influence of the experimental variables on the fraction of bubble surface area at which fat droplet spreading is possible. The same changes in parameters that promote droplet spreading at the air/water interface cause a decrease in the whipping time of our model creams. Correlating the whipping time of cream with the spreading behavior of fat droplets at the air/water interface represents a new insight into the mechanisms involved in the whipping of cream. © 2005 Elsevier B.V. All rights reserved. Keywords: Whipped cream; Partial coalescence; Spreading coefficient; Low molecular weight surfactant; Dairy foam
1. Introduction With a current trend in food processing, being to provide the consumer with food products with reduced fat and lower levels of saturated fat, there is a demand for a deeper understanding of how fat gives structure and texture to foods. Whipped cream is an example of a food system for which the structural characteristics are dependent on the presence of crystalline fat. It is generally accepted that air bubbles in whipped cream are stabilized by a network of partially coalesced fat globules [1–8]. The transformation of cream ∗
Corresponding author. Tel.: +31 318 659 589; fax: +31 317 482 358. E-mail address: [email protected] (G.A. van Aken). 1 Present address: Numico Research, P.O. Box 7005, 6700 CA Wageningen, The Netherlands. 0927-7757/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfa.2005.03.004
into whipped cream by beating air into the system proceeds as follows. First, large air bubbles are introduced and broken up into smaller ones by the whipping action of the whisks. This initial foam is predominantly stabilized by protein [1,2]. This is followed by a gradual accumulation of fat droplets on the surface of the air bubbles. This process has been well characterized; a number of electron microscopy studies clearly show fat droplets adhered to the air bubble surface [2,5,7,9]. Finally, a partially coalesced network of fat droplets is built up, which acts to trap the air bubbles, retains the serum phase and gives stiffness to the whipped cream. A number of physical definitions for the endpoint of whipping can be found in the literature including, maximum overrun, index of globule clumping and maximum stiffness, to name a few. Regardless of the definition, it is generally accepted that further whipping action, after the endpoint of
72
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
whipping, leads to the formation of butter granules and foam collapse. In a study on the whipping of natural cream, van Aken [8] distinguishes the three stages in the whipping process described above based on the mechanical resistance required to maintain a constant whisk speed. Depending on the time to reach the endpoint of whipping, the three stages either appear as distinct phases in the mechanical resistance versus time plot, or may overlap to some extent. It was shown that at moderately low rates of the whisks, long whipping times were obtained even though the evolution of the overrun (introduction of air) was not significantly reduced. The long whipping times under these conditions were shown to be related to a long second stage of the whipping process, due to slow adherence of fat globules to the bubble surfaces [8]. While there is a general understanding of the physical changes that take place during the whipping of cream, the mechanisms driving the transformation of cream into whipped cream remain poorly understood. In this paper, we will focus on the interaction between fat droplets and air bubbles during the second stage of whipping. This is the stage during which fat droplets accumulate at the air bubble surface. During this stage, air–bubble interfaces are continuously expanded and compressed as bubbles form, break-up and coalesce; this dynamic nature of the air–bubble interface can lead to the necessary interfacial conditions for fat droplets to adhere to the air bubbles. Adsorbed fat droplets release some liquid fat, which spreads onto the bubble surface, leading to interfacial fat globule flocculation and aggregation [3,6], or surface-mediated partial coalescence between the neighboring adhered droplets. Thus, adhesion of fat droplets to the air bubbles is an important step in the build-up of the partially coalesced fat droplet network, which ultimately stabilizes the whipped cream. In addition to surface-mediated partial coalescence, shear-induced partial coalescence also takes place during the whipping of cream. Both partial coalescence mechanisms are represented schematically in Fig. 1. The latter partial coalescence mechanism is the result of sticking of emulsion droplets when they collide in the bulk as a result of convection [10,11]. Under conditions of laminar flow, shear-induced partial coalescence proceeds like orthokinetic flocculation [12]; the decrease in number of droplets, N, in time (-dN/dt) can be described by: −
dN 4 = ϕNΨα dt π
(1)
where ϕ is the volume fraction of the dispersed phase, Ψ is the velocity gradient (or shear rate), and α is the capture efficiency. Capture efficiency, α, represents the probability that two particles stick when they encounter each other, and is influenced by four parameters [12]: protruding crystals, contact angle, colloidal repulsion and permanence of junctions. Under turbulent conditions, such as can be expected during whipping, the rate of partial coalescence becomes far more complex than that predicted by (1); however, the
Fig. 1. Schematic depiction of two possible mechanisms for partial coalescence. In the surface-mediated mechanism (a), air bubbles play an important role. When Sdyn > 0 (2) fat droplets can attach to the protein-covered air bubble surface and interfacial flocculation takes place, partial coalescence (A) of the interfacial flocs ensues. If, at this point, the bubble bursts (C), a partially coalesced fat clump remains. If the air bubble remains stable (B), a fat clump from the bulk may partially coalesce with the adsorbed fat clump (D). The shear-induced mechanism for partial coalescence (b) is independent of the presence of air. Partial coalescence (E) occurs when two droplets collide and the fat crystals in the droplets pierce the film between them. Larger fat clumps can form (F) when small clumps collide and partially coalesce with other clumps or free droplets ((a) reproduced from [26]).
equation remains valid in the sense that it gives an indication of the rate-determining system parameters. Shear-induced partial coalescence is independent of the presence of air bubbles in the system; it has been shown that cream can be churned into butter (a partial coalescence driven process) in the absence of air [6,13]. However, churning proceeds much faster in the presence of air [6], which may be explained by the additional effect of surface-mediated partial coalescence. Presumably, the presence of air bubbles increases the value of the capture efficiency, for example, by bringing the fat globules into contact by immobilizing them at the air bubble surface, or by improving the permanence of the junctions between partially coalesced fat globules due to oil spreading out of the adjacent droplets. The adhesion of fat droplets to the air/water interface is a critical precursor to surface-mediated mediated partial coalescence. The interaction of emulsion droplets with an air/water interface has been extensively studied by us [14–17]. This process involves the entering of droplets into the interface, followed by spreading of liquid oil over the interface. The spreading is self-enhancing, meaning the spreading of a first droplet promotes the entering and spreading of other droplets [14,15]. Whether or not oil spreading will occur is dependent on the balance of surface tensions (γ) at the three-phase boundary between the oil (O), water (W) and air (A) interfaces [14,18]. This balance of forces is described by the spreading coefficient, S: S = γAW − (γOW + γOA )
(2)
When S is positive, oil spreads. Triglyceride oils with medium to long chain fatty acids, such as soybean, sunflower, or liquid butter oil, have S > 0, and can, therefore, spread at the clean air/water interface [14,16,19]. Oil spreading is inhibited if γ AW becomes so low that the condition
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
γ AW > γ OW + γ OA is no longer satisfied, for example, due to the presence of an adsorbed protein layer [14]. This has important consequences for the adhesion of fat droplets to the air bubble surface during the whipping of cream, since air bubble surfaces are initially stabilized by adsorbed protein [1,2]. However, whipping is a dynamic process, and likely the surface tension of the air bubbles is also dynamic. For protein-stabilized air bubbles, Sdyn may become positive due to an increase in γ AW,dyn resulting from local surface expansion of the air bubbles during whipping. In this way, fat droplets can adhere to the dynamic bubble surface. In our work on the spreading behavior of proteinstabilized sunflower oil emulsion droplets at protein solution/ air interfaces [14,20], it was observed that oil droplet spreading occurred at protein concentrations and expansion rates that yielded steady-state surface pressures of less than or equal to 15 mN m−1 . It was, therefore, concluded that the air/water surface pressure, Π AW = 15 mN m−1 , represents a critical surface pressure, Π cr , below which spreading can occur for the sunflower oil/protein solution/air system. The relatively high value of Π cr indicates that the air/water interface does not need to be completely void of adsorbed protein for spreading to occur. Using known values for γ OW and γ OA one can show that Π cr is precisely the value satisfying the condition Sdyn = 0. Moreover, for protein-stabilized emul0 during spreading, sions, the value of Π cr implies γ OW = γOW 0 where γOW is the oil/water surface tension of the clean interface. This last point is not the case in the presence of the low molecular weight surfactants, Tween 20 and Span 80, which when present at sufficiently high concentrations, can cause an increase in Π cr of a few mN m−1 due to their ability to lower γ OW at the expanding oil/water interface during oil spreading, thus facilitating oil droplet spreading [17]. In this paper, we investigate the influence of whipping speed, protein type, protein concentration and the presence of low molecular weight (lmw) surfactants on the duration of the whipping process. The results are evaluated taking into account our previous work on oil droplet spreading at the air/water interface yielding a new model, which correlates the whipping time of cream with the spreading behavior of emulsion droplets at the air/water interface. The effect of slight clustering of natural milk fat globules, which is often reported to improve the whipping characteristics of natural cream, and probably originates from specific properties of the natural milk fat globule membrane, does not occur in our artificial cream and is, therefore, not included in this study.
2. Materials and methods 2.1. Materials Whey protein isolate (WPI) (BiPRO, lot no. JE 052-9420, Davisco Foods International, Le Sueur, MN 56058) and
73
sodium caseinate (sodium caseinate S, DMV International, Veghel, The Netherlands) were used in the preparation of protein solutions and emulsions. WPI and caseinate powders contained 95 and 86% protein, respectively (Biuret standard assay, in agreement with manufacturer’s specifications). Proteins were dissolved in 20 mM imidazole buffer (pH 6.7) containing 0.1 M NaCl by stirring for 1 h at room temperature. The chemicals used for the buffers were purchased from Merck (Darmstadt, Germany, analytical grade). For the study of the influence of lmw surfactant on the whipping properties of cream, either water-soluble Tween 20 (polyoxyethylene sorbitan monolaurate, Merck), or oil-soluble Span 80 (sorbitan monooleate, Fluka, Switzerland) was used. The fat phase of the emulsions consisted of a blend of 35 wt.% fully hydrogenated palm fat (Grinsted PS101, Danisco Cultor, Denmark) and 65% (w/w) sunflower oil (Reddy, Vandermoortele, Roosendaal, The Netherlands). 2.2. Emulsion preparation Emulsions were prepared containing 40 wt.% fat blend and 1 wt.% protein. The fat blend was warmed under gentle stirring at 65 ◦ C for 1 h to melt the fat. Prepared protein solutions were warmed to 65 ◦ C, prior to mixing with the fat blend. The mixture was then pre-homogenized at 65 ◦ C using an Ultraturrax (IKA, Staufen, Germany). Finally, the emulsions were prepared by homogenization at 2 MPa in a Rannie laboratory homogenizer, warmed at 65 ◦ C for 10 passes of the pump. After homogenization, emulsions were rapidly cooled in an ice bath to 7 ◦ C. Emulsions were stored overnight at 7 ◦ C before performing the experiments. The emulsions had a monomodal droplet size distribution with an average droplet diameter (d3,2 ) of ∼1 m, as determined by light scattering using a Coulter Laser LS230 (Coulter Electronics, Mijdrecht, The Netherlands). For emulsions with added Span 80, the required amount of surfactant was dissolved in the fat blend during warming. In the case of added Tween 20, an emulsion with a slightly higher fat and protein content was prepared. After homogenization and initial cooling, this emulsion was diluted with Tween 20, dissolved in buffer to give the desired concentration, followed by gentle mixing. 2.3. Whipping of emulsions Emulsions were whipped at 7 ◦ C in an automated whipping apparatus (Ledoux b.v., Dodewaard, The Netherlands). The same instrument was also used by van Aken [8] in a study on the aeration of natural whipped cream. A 250 mL emulsion sample was gently poured into a stainless steel or glass beaker (diameter, 9.6 cm). This was placed inside the thermostat jacket of the whipping apparatus, where the temperature was controlled within 0.5 ◦ C during whipping. The whipping action was provided by two whisks (radius, 20 mm; length, 113 mm; composed of six, 1 mm thick stainless steel wires) rotating in a planetary motion pattern. This motion of the whisks is designed to maximize the
74
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
Fig. 2. Power input transferred by the whisks vs. time during the whipping of WPI-stabilized model cream. Numbers near the curves denote the rotational speed of the whisks in rpm. Reproduced from [26].
incorporation of air and to maintain homogeneous whipping conditions over the entire volume. The electrical current required to maintain a constant rotational speed was monitored during the whipping process. This current is a measure of the energy dissipation in the system, and is the characteristic of its mechanical properties. As a result of whipping, the aerated emulsion eventually reaches a maximum stiffness, which is registered as a peak in the electrical current signal. We define this peak as the endpoint of whipping, and refer to the time to reach this peak as the whipping time. Unless otherwise stated, all experiments were performed using a rotational speed of 584 rpm. All experiments were performed in duplicate.
3. Results In Fig. 2, the power transferred by the whisks during whipping of emulsion, also referred to as (artificial) cream,
stabilized by 1% WPI is plotted as a function of time for various whipping speeds. The peak of maximum resistance defines the endpoint of whipping, and the time to reach this peak is the whipping time. The whipping times measured for our model creams were significantly longer and had lower values for the peak of maximum resistance than those reported for natural cream whipped using the same apparatus [8]. However, the shape of the curves in Fig. 2 and the microscopy images in Fig. 3 suggest that despite the longer whipping times, the whipping process for our artificial cream follows a similar mechanism to that of natural whipping cream. For example, for the cream whipped at a rotational speed of 584 rpm (Fig. 2), the three stages of whipping described by van Aken [8] can be clearly identified. In the first stage, large air bubbles are incorporated into the emulsion and broken up into smaller bubbles; this is characterized by a small increase in the power-input signal (Fig. 2). During this first stage, the foam is considered to be protein-stabilized [1,2]. A few emulsion droplets are observed to adhere to the air bubble surface, but the air bubbles are not yet fully coated (Fig. 3b). In the second stage of whipping, the coating of emulsion droplets at the air bubble surface gradually becomes denser (Fig. 3c) and there is a gradual slight increase in the power transferred by the whisks (Fig. 2), which indicates an increasing viscosity of the system. Finally, in the third, the size of the fat clumps increases a network of partially coalesced emulsion droplets is formed (Fig. 3d). This partially coalesced network gives stiffness to the cream, resulting in a sharp increase in the power input (Fig. 2). In Fig. 4, whipping time is re-plotted as a function of rotational speed for WPI-stabilized cream as well as for sodium caseinate-stabilized cream. For both systems, the whipping time begins to increase sharply for whipping speeds lower
Fig. 3. Light micrographs of the cream stabilized by 1% WPI whipped at a rotational speed of 584 rpm: (a) before whipping; (b) after 50 s; (c) after 290 s; and (d) after 473 s. Time to reach endpoint of whipping, 473 s; length of bar = 100 m.
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
75
Fig. 4. The influence of rotational speed on whipping time for creams stabilized by either WPI () or sodium caseinate (). Dotted lines serve to distinguish between data sets.
than 600 rpm. van Aken [8] attributes the strong dependence of whipping time on rotational speed to a variation in the rate of transformation of free fat globules into clumps driven by surface-mediated partial coalescence during the second stage of whipping. Fig. 4 also shows that WPI-stabilized creams give shorter whipping times than sodium caseinate-stabilized creams, a trend that was also observed by van Camp et al. [21] for butterfat emulsions stabilized by whey protein concentrate (WPC) compared to those stabilized by sodium caseinate. This observation suggests that the transition from free fat droplets to clumped fat proceeds faster for WPI-stabilized creams than for sodium caseinate-stabilized creams. Protein concentration had an effect on the whipping time of WPI-stabilized model cream (Fig. 5). There is a significant increase in the whipping time when the protein concentration is increased from 1 to 3 wt.%; however, the whipping time does not change much between the creams containing 0.6 and 1 wt.% WPI. These results are in agreement with van Camp et al. [21], who observed a decrease in whipping time with decreasing WPC concentration (from 3.1 to 0.42 wt.%), which below 1.6 wt.% appeared to level off to a plateau value. Further, these results are in qualitative agreement with Needs and Huitson [9], who observed a 50% decrease in whipping time for natural cream, from which the milk serum proteins had been removed, compared to unadulterated cream. We observed a slight decrease in the overrun of the
Fig. 5. The influence of protein concentration on whipping time () and overrun () of WPI-stabilized creams. Dotted lines serve to distinguish between the data sets. Whisk rotational speed: 584 rpm.
Fig. 6. Whipping time () and overrun () of WPI-stabilized creams with added (a) Tween 20 (in water phase) or (b) Span 80 (in oil phase). Dotted lines serve to distinguish between the data sets. The values measured in the absence of surfactant are shown on the primary y-axis before the break. Whisk rotational speed: 584 rpm.
emulsions with increasing protein concentration (Fig. 5). Needs and Huitson [9] reported no significant effect of protein concentration on the overrun, nor did van Camp et al. [21] in the case of WPC-stabilized model creams. This difference between our experimental results and the literature could be due to a difference in the whipping apparatus used. The influence of added lmw surfactant on the whipping time and overrun of the model creams was also investigated. In Fig. 6, the whipping times are shown for WPI-stabilized emulsions containing added Tween 20 (Fig. 6a) or Span 80 (Fig. 6b). In both the cases, we observed a decrease in whipping time with increasing surfactant concentration, the effect being stronger for Tween 20 than for Span 80, especially at lower concentrations. For the emulsions containing 2.7 and 5.5 mM Tween 20, the whipping time was decreased to such an extent that it was similar to the whipping time of natural cream whipped at the same rotational speed using the same apparatus as reported by van Aken [8]. A decrease in whipping time in the presence of added lmw surfactant has also been reported for recombined cream to which glycerol monostearate had been added [22]. Further, our results are in qualitative agreement with the observation reported in the literature, that destabilization of ice cream mix emulsions subjected to shear in the presence of air proceeds more rapidly in the presence of added emulsifier [23,24]. Over the range of concentrations used, the overrun was not influenced by the addition of Span 80 (Fig. 6b). This is also the case for Tween 20 (Fig. 6a), with the exception of the highest concentration (5.5 mM Tween 20), for which a somewhat lower overrun was found; this system also showed a different shape of the power input curve (see below). The power-input profiles for the creams with added surfactant (Fig. 7) were similar in shape to those for the WPIstabilized model cream in the absence of surfactants (Fig. 2). The main difference is that with the increasing surfactant
76
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
Fig. 7. Power input transferred by the whisks vs. time for the whipping of cream containing 1% WPI with added (a) Tween 20 or (b) Span 80. Numbers near the curves denote the lmw surfactant concentration in mM. Whisk rotational speed: 584 rpm ((a) reproduced from [26]).
concentration, a steady decrease in the length of the second stage of whipping is observed, which strongly suggests that surfactants play an important role in the adhesion and subsequent surface-mediated partial coalescence of emulsion droplets at the air/water interface. The sharpness of the peak in the third stage of whipping, the stage characterized by the build-up of a partially coalesced network, did not appear to differ between systems in the presence (Fig. 7) or absence (Fig. 2) of surfactant. The shape of power-input profile for the cream with the highest concentration of Tween 20 (5.5 mM, Fig. 7a) differed from the other samples. In view of the low overrun for this concentration, this may suggest that there is an optimum between accelerating the whipping process and forming a whipped cream with desirable structural characteristics.
4. Discussion We have demonstrated that whipping speed, protein type, protein concentration and the presence of lmw surfactant all influence the whipping time of our artificial creams (Figs. 4–6). The differences in whipping time were observed to result from a variation in the length of the second stage of whipping. This stage is characterized by the accumulation of fat globules at the air/water interface. Thus, our results strongly suggest that the whipping time is mainly determined by the rapidity and ease with which emulsion droplets can attach to and spread at the air/water interface. In the following discussion, we present a model in which the effect of the variables investigated on the whipping time should is explained in terms of the spreading behavior of emulsion droplets at the planar air/water interface. Brooker et al. [2] showed that the incorporation of air is not a sufficient stimulus per se to cause fat droplets to penetrate the air/water interface on a large scale. In our work on the spreading behavior of emulsion droplets at the planar air/water interface [14,15,20], we showed that an adsorbed
protein layer can effectively inhibit droplet entering and spreading under quiescent and dynamic conditions, provided it maintains a pressure exceeding a critical value, Π cr . During aeration, the air bubbles are initially protein-stabilized [1,2]; thus, it seems reasonable to assume that during whipping, fat droplet adsorption and spreading can only occur at the portion of the air bubble surface for which the air/water surface pressure is lower than Π cr . Moreover, adsorption and spreading of fat droplets at the air bubble surface is the precursor to surface-mediated droplet coalescence (Fig. 1). We, therefore, expect that manipulating the fraction of the total bubble surface area for which Sdyn > 0 will influence the ease and rapidity of fat droplet adhesion and spreading, ultimately governing the whipping time of the system. This leads to the following scenario. During the second stage of whipping, the air bubbles are in a dynamic state of formation, break-up and coalescence, during which their surfaces are continually expanded and compressed. As a result, at a given moment in time there will be a distribution of dynamic air/water surface pressures; ranging from low values for newly formed air bubble surfaces to high values for portions of the air bubble surface that are compressed. This distribution is represented schematically in Fig. 8a. The total area under the curve represents the sum of the surface area of the individual air bubbles, or the total air bubble surface available in the whipped cream. The Π cr required for the fat droplet spreading is indicated by a vertical dotted line. The area under the curve to the left of this line is shaded to indicate that for this fraction of the total air bubble surface, the air/water surface pressure will be low enough for adsorption and spreading of fat droplets to occur; the remainder of the bubble area is inactive. First, let us assume that the situation in Fig. 8a represents a reference case. If the surface pressure distribution is shifted to the left with respect to the reference case, as in Fig. 8b, we expect the fraction of the total bubble surface area, for which Sdyn > 0, to increase and the whipping time to decrease. In our experiments at the planar air/water interface [20], we observed that, for a protein solution, lower steady state surface pressures (i.e. a shift in the surface pressure distribution in Fig. 8a to the left) could be achieved by increasing the relative expansion rate, decreasing the protein concentration or using WPI instead of sodium caseinate. Similarly, for the creams investigated, increasing the rotational speed (Figs. 2 and 4), decreasing the protein concentration (Fig. 5) or using WPI instead of sodium caseinate (Fig. 4) resulted in shorter whipping times. Conversely, if the surface pressure distribution is shifted to the right with respect to the reference case, as in Fig. 8c, we expect the fraction of the total bubble surface area for which Sdyn > 0 to decrease and the whipping time to increase. This can be achieved by decreasing the relative expansion rate/rotational speed, increasing the protein concentration or using sodium caseinate instead of WPI. The experimental results presented in Figs. 4 and 5 confirm this, supporting the concept that the whipping time is related to the ease with
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
Fig. 8. Schematic representation of the distribution of surface pressure values present at the surface of air bubbles during whipping. Vertical line indicates the critical surface pressure for spreading. For a given system (a), the shaded area under the curve indicates the fraction of the air bubble surface area for which fat droplet spreading can occur. This area can be increased or decreased by shifting the surface pressure distribution to the left (b) or to the right (c), respectively. In addition, if the critical surface pressure is shifted to a higher value (d), the fraction of the bubble surface area for which Sdyn > 0 will increase. As explained in Section 4, increasing the proportion of the bubble surface area for which Sdyn > 0 leads to shorter whipping times, whereas decreasing the area leads to longer whipping times. Reproduced from [26].
which fat droplets can adhere to and spread at the air/water interface. As an additional point, in an earlier paper [15] we observed that -lactoglobulin forms brittle adsorbed layers at the air/water interface, leading to entering and spreading of oil droplets over a much larger area in a single event than for -casein, which forms more fluid-like layers. This may serve as an additional explanation for the shorter whipping times observed for the WPI-stabilized compared to the sodium caseinate-stabilized creams (Fig. 2). The line of reasoning depicted by Fig. 8 can also be used to explain why the creams stabilized by either 0.6 or 1 wt.% WPI had similar whipping times, while the cream stabilized by 3 wt.% WPI had a significantly longer whipping time (Fig. 5). Spreading can occur as long as Sdyn > 0. If, for the 1 wt.% WPI cream, the majority of the surface pressure distribution is already to the left of the critical surface pressure, then shifting the distribution even further to the left by lowering the protein
77
content will no longer significantly influence the whipping time. However, increasing the protein concentration will shift the distribution to the right and, with that, the whipping time of the cream will increase. For the variables discussed above (rotational speed, protein type and protein concentration) the value of Π cr was constant. In previous work on spreading behavior of emulsion droplets at the planar air/water interface [17], we showed that surfactants facilitate droplet spreading by shifting Π cr to higher Π AW values. Correspondingly, in the whipping experiments (Fig. 7), we observed a significant decrease in the length of the second stage of whipping with increasing surfactant concentration. Thus, the decrease in whipping time in the presence of surfactants further strengthens our hypothesis that whipping time is dependent on the rapidity and ease with which emulsion droplets can adhere to and spread at the air/water interface. This is depicted schematically in Fig. 8d. It should be mentioned that the whipping time of the WPIstabilized cream is observed to begin to decrease (Fig. 6) at surfactant concentrations that are considerably lower than those required for an increase in the value of Π cr for -lactoglobulin stabilized emulsion reported in reference [17]. The reason for this is not completely clear. A possible explanation is that, in reference [17], the increase in the value for Π cr was measured as a function of the surfactant concentration in an emulsion that was added as a tiny droplet to a bulk solution of protein, which did not contain surfactant. Because the surfactant slowly diffuses out of the droplet, the actual surfactant concentration at the point of measuring Π cr will have been considerably lower than the initial concentration. Until now, researchers have tried to explain the influence of whipping speed, protein concentration, protein type and the presence of lmw surfactant on the whipping time of cream in terms of the static properties of the emulsion droplets and their adsorbed layers, such as adsorbed amount, relative strength of the adsorbed protein layer, size and orientation of fat crystals, or contact angle of the fat crystals at the oil/water interface. These properties of the emulsion droplets are of relevance to the probability of a fat crystal piercing the film between colliding droplets, which is an important parameter controlling the rate of shear-induced partial coalescence, and probably a significant parameter controlling the formation of a network of partially coalesced fat droplets during the third stage of the whipping process. However, we have shown that the whipping time is largely controlled by a process of surface-mediated partial coalescence. Therefore, an explanation for rate of partial coalescence based solely on the static properties of the emulsion droplets and oil/water interfaces fail when applied to emulsions sheared in the presence of air. The conceptual model presented in Fig. 8 demonstrates that the influence of rotational speed, protein concentration, protein type and the presence of lmw surfactants on the whipping time of cream (Figs. 4–6 and [8,9,21–23]) can be well accounted for in terms of the influence of these variables on the spreading behavior of emulsions. This model (Fig. 8), when combined with the existing knowledge of the factors promot-
78
N.E. Hotrum et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 260 (2005) 71–78
ing shear-induced partial coalescence (1), can more fully account for the dependence of the whipping time of the creams studied in this work on the experimental system parameters (whisk rotational speed, protein concentration, protein type, lmw surfactant concentration). The industry has long been aware of the fact that in order to improve the whipping properties of recombined cream, lmw surfactants must be added. The mechanism of surface-mediated partial coalescence and how it can be controlled (Fig. 8) helps to explain why this is so. It would be interesting to determine if the model presented in Fig. 8 can be applied to explain why homogenized cream tends to have much longer whipping times than natural cream [5,7]. Conceivably, the abundant presence of small, surface active molecules in the milk fat globule membrane of fat globules in natural cream [6] could shift Π cr to a higher value compared to homogenized cream, where the droplets are predominantly protein-stabilized [25]. This is the subject of ongoing research. Acknowledgement Danisco Ingredients, Denmark is gratefully acknowledged for providing the fully hydrogenated palm fat. References [1] M. Anderson, B.E. Brooker, E.C. Needs, in: E. Dickinson (Ed.), Food Emulsions and Foams, Royal Society of Chemistry, London, 1987, p. 100. [2] B.E. Brooker, M. Anderson, A.T. Andrews, Food Microstruct. 5 (1986) 277. [3] D.F. Darling, J. Dairy Res. 49 (1982) 695.
[4] H.D. Goff, J. Dairy Sci. 80 (1997) 2620. [5] E. Graf, H.R. M¨uller, Milchwissenschaft 20 (1965) 302. [6] H. Mulder, P. Walstra, The Milk Fat Globule, Pudoc, Wageningen, 1974. [7] D.G. Schmidt, A.C.M. van Hooydonk, Scanning Electron Microsc. III (1980) 653. [8] G.A. van Aken, Colloids Surf. A 190 (2001) 333. [9] E.C. Needs, A. Huitson, Food Struct. 10 (1991) 353. [10] K. Boode, P. Walstra, Colloids Surf. A 81 (1993) 121. [11] M.A.J.S. van Boekel, P. Walstra, Colloids Surf. 3 (1981) 109. [12] P. Walstra, Physical Chemistry of Foods, Marcel Dekker, New York, 2003. [13] J.H. Labuschagne, Churning in the absence of air, Thesis, Wageningen Agricultural University, The Netherlands, 1963. [14] N.E. Hotrum, T. van Vliet, M.A. Cohen Stuart, G.A. van Aken, J. Colloid Interface Sci. 247 (2002) 125. [15] N.E. Hotrum, M.A. Cohen Stuart, T. van Vliet, G.A. van Aken, Langmuir 19 (2003) 10210. [16] N.E. Hotrum, M.A. Cohen Stuart, T. van Vliet, G.A. van Aken, Colloids Surf. A 240 (2004) 83. [17] N.E. Hotrum, Emulsion droplet spreading at air/water interfaces: mechanisms and relevance to the whipping of cream, Thesis, Wageningen University, The Netherlands, 2004. [18] W.D. Harkins, A. Feldman, J. Am. Chem. Soc. 44 (1922) 2665. [19] E.P. Schokker, M.A. Bos, A.J. Kuijpers, M.E. Wijnen, P. Walstra, Colloids Surf. B 26 (2003) 315. [20] N.E. Hotrum, M.A. Cohen Stuart, T. van Vliet, G.A. van Aken, in: E. Dickinson, T. van Vliet (Eds.), Food Colloids, Biopolymers and Materials, Royal Society of Chemistry, Cambridge, 2003, p. 192. [21] J. van Camp, S. van Calenberg, P. van Oostveldt, A. Huyghebaert, Milchwissenschaft 51 (1996) 310. [22] J.G. Zadow, F.G. Kieseker, Aust. J. Dairy Technol. 30 (1975) 114. [23] B.M.C. Pelan, K.M. Watts, I.J. Campbell, A. Lips, J. Dairy Sci. 80 (1997) 2631. [24] H.D. Goff, W.K. Jordan, J. Dairy Sci. 72 (1989) 18. [25] D.F. Darling, D.W. Butcher, J. Dairy Res. 45 (1978) 197. [26] N.E. Hotrum, M.A. Cohen Stuart, T. van Vliet, G.A. van Aken, in: E. Dickinson (Ed.), Food Colloids, Interactions, Microstructure and Processing, Royal Society of Chemistry, Cambridge, 2005, p. 317.