- Email: [email protected]
Tempering of dairy emulsions: Partial coalescence and whipping properties
International Dairy Journal 56 (2016) 92e100
Contents lists available at ScienceDirect
International Dairy Journal journal homepage: www.elsevier.com/locate/idairyj
Tempering of dairy emulsions: Partial coalescence and whipping properties Kim Moens*, A.K.M. Masum, Koen Dewettinck Laboratory of Food Technology and Engineering, Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, 9000 Ghent, Belgium
a r t i c l e i n f o
a b s t r a c t
Article history: Received 9 July 2015 Received in revised form 1 December 2015 Accepted 8 January 2016 Available online 28 January 2016
This study investigates the effect of applying a timeetemperature profile to natural and recombined cream to influence partial coalescence and, consequently, the whipping quality. To date, no clear relationship exists between the consequences of tempering on a microstructural level, partial coalescence, and whipping properties. Milk fat crystallisation was analysed using differential scanning calorimetry and the internal arrangement of fat crystals was visualised with cryo-scanning electron microscopy. Shear-induced partial coalescence and whipping properties were studied. Shear-induced partial coalescence was promoted, attributed to the observed changes in the fat crystal network. The effects on whipping properties were different for natural and recombined cream and thus dependent upon the interfacial composition. Consolidation of the partially coalesced fat droplet network by tempering increased the stability of whipped recombined cream during cold storage. Tempering is a promising tool to alter the susceptibility to partial coalescence by changing the internal fat crystal network, and influencing whippability. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Dairy cream is an emulsion of milk fat droplets in a continuous aqueous phase. An important characteristic of dairy cream is the ability of transforming the oil-in-water (O/W) emulsion into a ternary system by whipping, which is a combination of shear and air inclusion. A good quality whipped cream has a high overrun, a desirable firmness, and a high stability. The basic mechanism occurring during whipping is partial coalescence. Firstly, shearinduced partial coalescence occurs when fat droplets connect by protruding fat crystals. These fat crystals penetrate the interface of another fat droplet forming a connection (Fredrick, Walstra, & Dewettinck, 2010). Secondly, surface-mediated partial coalescence takes place when fat droplets enter the air bubbles during whipping, liquid oil is released from the fat crystal network and spread over the air bubble surface. If this occurs between two fat droplets that are close to each other at the air bubble surface, a junction is formed leading to partial coalescence (Hotrum, Stuart, van Vliet, Avino, & van Aken, 2005). The combination of both mechanisms results in a network of partially coalesced fat droplets,
* Corresponding author. Tel.: þ32 9 264 6198. E-mail address: [email protected] (K. Moens). http://dx.doi.org/10.1016/j.idairyj.2016.01.007 0958-6946/© 2016 Elsevier Ltd. All rights reserved.
giving structure to whipped cream by stabilising the air bubbles (Brooker, Anderson, & Andrews, 1986; Goff, 1997). Natural cream, obtained by the physical separation of a high-fat phase from milk, is considered the gold standard. It has exceptionally good whipping properties and a nice taste. However, for many reasons (amongst others, health, technological and practical reasons) the food industry aims to develop recombined creams obtained by emulsification of fat in an aqueous phase. Ingredients can vary greatly, using different types of fat, emulsifiers, hydrocolloids, sugars, sugar replacers, and flavours. Nevertheless, the whipping properties should always be optimised by ensuring an efficient envelopment of the air bubbles by partially coalesced fat droplets. Diverse strategies may be followed in adapting partial coalescence, such as the composition of the cream (proteins, emulsifiers, hydrocolloids and the amount and type of fat), and the processing conditions (temperature, shear rate, cooling rate, homogenisation pressure and tempering) (Fredrick et al., 2010; Hotrum et al., 2005; van Aken, 2001). The latter is the main subject of this study, and consists of the application of a specific timeetemperature profile in which the temperature is first increased to partly melt the fat, and then decreased again to induce heterogeneous crystallisation of the liquid fat (Boode-Boissevain, 1992; Drelon et al., 2006; Gravier, Drelon, Boisserie, Omari, & Leal-Calderon, 2006; Mutoh, Nakagawa, Noda, Shiinoki, &
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Matsumura, 2001; Mutoh, Kubouchi, Noda, Shiinoki, & Matsumura, 2007; Oortwijn & Walstra, 1982; Sugimoto et al., 2001; ThivilliersArvis, Laurichesse, Schmitt, & Leal-Calderon, 2010; Thivilliers, Drelon, Schmitt, & Leal-Calderon, 2006; Thivilliers, Laurichesse, Saadaoui, Leal-Calderon, & Schmitt, 2008). The pioneering research of Boode-Boissevain (1992) on tempering of emulsions suggests that tempering increases the susceptibility to partial coalescence due to repositioning of the fat crystals. Increasing the temperature leads to partial melting of the fat crystals by which the three dimensional fat crystal network is broken. The remaining fat crystals can move freely to the energetically more favourable interface. Decreasing the temperature induces growth of the crystals near the interface, resulting in larger crystals that can penetrate further into the aqueous phase, increasing the susceptibility to partial coalescence. Emulsions with a fat content lower than 25% are not prone to the effect of tempering (Boode-Boissevain, 1992), and a minimal amount of solid fat during tempering, estimated to be between 1.5% and 8%, is necessary (Boode-Boissevain, 1992; Drelon et al., 2006; Mutoh et al., 2007). Moens, De Clercq, Verstringe, and Dewettinck (2015) showed that tempering significantly influences fat crystallisation properties, depending upon the tempering temperature. Besides the change in melting properties and fat crystal polymorphism, the internal arrangement of the fat droplets is also influenced. Lamellar structures near the interface of the fat droplets were visualised in recombined cream tempered at 20 C. The latter observation adds to the hypothesis that crystals move to the interface. Although fat crystallisation properties largely determine the effect of tempering, the role of the interfacial composition may not be underestimated. Thivilliers et al. (2006) showed also that under quiescent conditions, partial coalescence may be promoted by tempering to such an extent that the emulsion forms a gel. Besides the fat crystallisation properties, the composition of the interface greatly influenced this gelling behaviour (Thivilliers et al., 2008). More specifically, tempering seems to have less effect on proteinstabilised O/W emulsions because of the thick interface. Conversely, O/W emulsions stabilised with low molecular surfactants have a thinner interface allowing crystals to pierce through, adding to the effect of tempering (Mutoh et al., 2007; Thivilliers et al., 2008). Mutoh et al. (2001) and Sugimoto et al. (2001) investigated the effect of tempering on protein-stabilised vegetable creams. Despite the thick interface, they observed increased flocculation after tempering which was not due to partial coalescence. It was suggested that the fat crystals near the interface could affect the adsorption of the proteins, consequently affecting interactions between the fat droplets. Tempering is likely to affect the whipping properties. It was demonstrated that tempering of natural cream at 30 C could improve its whipping properties by reducing the whipping time and serum loss, and increasing the overrun and firmness (Besner & Kessler, 1998); however, no link with partial coalescence or any other possible mechanism explaining the effects of tempering was provided in this study. Tempering the cream after it was whipped seemed to have similar effects. It was observed that for both natural and recombined cream, the elastic behaviour of the whipped cream increased when tempered at 25 C due to an increased connectivity between the fat droplets. The latter is explained by surfacemediated partial coalescence occurring at the air bubble surface (Drelon et al., 2006; Gravier et al., 2006; Riaublanc et al., 2005). In this current study, the relationship between tempering, partial coalescence kinetics, and whipping properties was studied. Two types of dairy cream, natural cream and recombined cream, were subjected to tempering at two temperatures (Tmax) of 20 C and 30 C.
93
2. Materials and methods 2.1. Materials Commercially available natural cream (NC; Debic, FrieslandCampina Professionals, Lummen, Belgium) with a fat content of 35% was used. Furthermore, recombined cream (RC) was produced by recombining anhydrous milk fat (FrieslandCampina butter, Noordwijk, The Netherlands), sweet cream butter milk powder (Westbury Dairies, Westbury, UK; containing 30.1 ± 0.1% proteins, 8.5 ± 0.1% fat and 51.5 ± 0.2% lactose), carrageenan (Satiagel ME4; Cargill Deutschland GmbH, Krefeld, Germany) and potable water. Sodium azide (Acros Organics, Geel, Belgium) was added to prevent microbial spoilage. Sweet cream butter milk powder (7.6%) was dissolved in potable water and kept overnight for full hydration. Carrageenan (0.01%) and sodium azide (0.01%) were added to this aqueous phase and pre-heated to 50 C. The anhydrous milk fat (35%) was melted and pre-heated to 50 C. The pre-emulsion was prepared by thoroughly mixing the aqueous phase with the fat phase using an Ultra-Turrax (10,000 rpm) for 10 min at 50 C. This pre-emulsion was then homogenised using a two-step laboratory scale homogeniser (APV cooling systems, Alberslund, Denmark) at 3e1 MPa. Finally, the emulsion was rapidly cooled to 5 C and stored in a thermostatic cabinet at 5 C for at least 7 d to complete fat crystallisation before tempering was applied. 2.2. Laser light diffraction The size of the fat droplets in natural cream and in recombined cream was measured with laser light diffraction according to the method described by Fredrick et al. (2013b). 2.3. Tempering procedure Cream (2.5 L) was heated to Tmax (20 or 30 C) using a Herbst HR-S3 Tempering Unit equipped with a double walled mixing bowl (Herbst Machinery Ltd, Omagh, Ireland) temperature-regulated €ltemaschinenbau with a Huber Thermostat (Peter Huber Ka GmbH, Offenburg, Germany). The sample was kept isothermally for 30 min at Tmax under continuous stirring before it was cooled to 5 C by passing the cream through tubing that was fitted in a water bath at 5 C. The cream was stored at 5 C for at least 7 d to complete crystallisation before analysis. 2.4. Differential scanning calorimetry Differential scanning calorimetry (DSC) with a refrigerated cooling system (model Q1000 DSC; TA Instruments, New Castle, DE, USA) was used to study the melting behaviour of (tempered) anhydrous milk fat (AMF), NC and RC. DSC was calibrated with indium (TA Instruments), azobenzene (SigmaeAldrich, Bornem, Belgium) and undecane (Acros Organics) prior to analysis. Nitrogen was used to purge the system. Samples (5e15 mg) were sealed in hermetic aluminium pans, and an empty pan was used as a reference. The melting profile was analysed by heating the sample at 5 C per min from 5 C to 55 C. Measurements were carried out in triplicate. 2.5. Nuclear magnetic resonance A Maran Ultra 23 MHz pulsed-field gradient nuclear magnetic resonance (NMR) instrument (Oxford Instruments, Tubney Woods, Abingdon, UK) was used to measure the solid fat content profile of NC and RC. The solid fat content was determined 1 week after
94
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
production in the range 5e50 C. An indirect solid fat content (SFC) measurement was carried out to eliminate the contribution of the aqueous phase to the free induction decay signal. A detailed description of this method is given in Fredrick et al. (2013b). 2.6. Light microscopy A Leica DM 2500 (Leica Microsystems Belgium BVBA, Diegem, Belgium) light microscope was used to visualise the fat droplets in cream. The cream was first diluted (1:10) in demineralised water. Thereafter, one drop was transferred to a microscopic slide and covered with a cover slip. The images were recorded with a Leica MC170HD camera (Leica Camera, Wetzlar, Germany). 2.7. Cryogenic scanning electron microscopy The microstructure of unwhipped and whipped cream was visualised using scanning electron microscopy with cryogenic sample preparation. Cream was mounted on an aluminium stub and quickly frozen in slushed nitrogen (210 C). The time between sample preparation and freezing was kept as short as possible to avoid crystal melting. Then, the sample was fractured at 150 C, sputter-coated with platinum, and finally observed with a Jeol JSM7100F scanning electron microscope (Jeol Europe B.V., Zaventem, Belgium). 2.8. Rotational viscosity analyses Rotational viscosity analyses were carried out to study the kinetics of partial coalescence without the inclusion of air, similar to the research of Fredrick et al. (2013a). An AR2000ex controlled stress rheometer (TA instruments, Brussels, Belgium) equipped with a starch pasting cell as measuring geometry was used. To follow partial coalescence, 30 mL of cream was transferred into the cup, and after an equilibration period at the measuring temperature (20 C), a constant shear rate was applied and the apparent viscosity of the cream was followed as a function of time. The time when the viscosity rises to a maximum was defined as the churning time (tch). Preliminary results showed that RC is much more sensitive to shear than NC. Therefore, a constant shear rate of 150 s1 was applied to NC, whereas a constant shear rate of 50 s1 was applied to RC. 2.9. Whipping properties Whipping was carried out with a Hobart N50 mixer (Illinois Tool Works, Glenview Chicago, IL, USA) equipped with a D-wire whip designed for maximum air inclusion in a low viscous product. The medium speed (position 2: 240 rpm) was applied at 5 C. All creams were whipped until a defined observable end point. The optimum whipping time is defined as the time at which the cream broke away from the wires and the bowl. Moreover, when the whipping process is stopped at this optimum whipping time (twh), the cream should not immediately start to flow around the wires of the impeller. These tests were always carried out by the same person who was trained to detect this optimal point. Whipping experiments were repeated at least twice for each cream (i.e., two whipping batches). For each whipping batch, the overrun, the serum loss, and the firmness were determined according to the methods described by Fredrick et al. (2013a). 2.10. Statistical analysis SPSS software (version 22, SPSS Inc., Chicago, IL, USA) was used for statistical comparison of the whipping properties. Analyses of
variance (one-way ANOVA) were carried out to determine significant differences between the results. A Levene test was used to test equality of variances. If this hypothesis was retained, the Tukey's post hoc test was executed for pairwise comparison. If the variances were not equal, the Dunnett's T3 post hoc test was used. All tests were carried out at a 95% significance level. 3. Results 3.1. Particle size distribution Fat crystallisation in emulsion droplets, as well as partial coalescence and the whipping properties, are greatly influenced by the droplet size. To eliminate this factor, the homogenisation process of recombined cream was adapted to obtain an emulsion of which the droplet size is in the same order of magnitude as natural cream. The Sauter diameter of natural and recombined cream was 2.95 ± 0.01 mm and 2.84 ± 0.01 mm, respectively. 3.2. Fat crystallisation The melting behaviour of pure AMF, NC and RC, whether tempered or not, is presented in Fig. 1. The untempered samples show the typical broad melting range resulting from the high diversity of fatty acids present in milk fat. This broad melting peak is separated into two distinguishable melting peaks after tempering at 20 C, of which the separation between both peaks is situated around 23 C. This separation is more clear in pure AMF because of the dilution effect in cream. Tempering at 30 C also results in a separation into two distinguishable peaks, but this time the peaks are separated at around 33 C. Milk fat has the tendency to crystallise in highly mixed crystals, i.e., crystals comprising different triacylglycerides (Walstra, 2003). At Tmax, a part of the crystals are melted and the released triacylglycerides are rearranged in such a way that purer crystals arise (Moens et al., 2015). This process can be compared with the fractionation process. Similar changes in melting behaviour are observed for pure AMF, as also for cream, indicating that this process occurs simultaneously in every individual fat droplet. Fig. 2A and B show NC after tempering at 20 C, and Fig. 2C and D show NC after tempering at 30 C. It is seen from Fig. 2A that 20 C tempering induces little or no aggregation, and that most of the fat droplets are still present as individual entities. It seems that partial coalescence does not occur during the tempering process. In contrast, Fig. 2C shows that during tempering at 30 C, some clustering occurs. Heating of those clusters resulted in complete merging indicating that the fat droplets are partially coalesced. The elongated shape of the clusters suggests that one fat droplet is connected to only two other fat droplets. The fat droplets in 20 C tempered cream show a more irregular surface (Fig. 2B) compared to fat droplets that were tempered at 30 C (Fig. 2D), suggesting differences in the fat crystal network of both creams. Likewise, Thivilliers et al. (2006) designates the rough and rippled surface of fat droplets to irregularly shaped/oriented crystals. These findings are validated by cryo-SEM images of RC obtained before and after tempering (Fig. 3). Fig. 3 shows the section of fat droplets exposing the fat crystal network in the droplets. A random distribution of fat crystals in the fat droplet was observed before tempering. Interestingly, well-defined lamellar regions near the interface were detected in fat droplets subjected to tempering at 20 C. This is probably the reason why irregular fat droplets were observed with light microscopy after tempering at 20 C. The latter also contributes to the hypothesis that the remaining fat crystals at Tmax move to the energetically more favourable interface where they grow during subsequent cooling, resulting in L- and M-type fat droplets
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
95
Fig. 1. Untempered (d), tempered at 20 C (e e e) and tempered at 30 C (- - -) melting curves of (A) anhydrous milk fat, (B) natural cream, and (C) recombined cream.
(Boode-Boissevain, 1992; Mutoh et al., 2001; Sugimoto et al., 2001; Thivilliers et al., 2008; Walstra, 1967). The same conclusions could be drawn for NC and RC. 3.3. Shear-induced partial coalescence Shear-induced partial coalescence occurs in cream when shear is applied in the absence of air. In this research, the samples were sheared at a constant shear rate until the time when the viscosity rises to a maximum, i.e., the churning time (tch), and relates to the susceptibility of the fat droplets to partial coalescence. Phase separation occurs at this point. During this experiment, the viscosity was measured as a function of time. The shape of the viscosity profile gives information about the progress of shear-induced partial coalescence. Fig. 4A shows the viscosity profile of NC and Fig. 4B shows the viscosity profile of RC, before and after tempering at 20 C and 30 C. Tempering clearly reduces tch of both types of cream and at both temperatures. Considering NC, the shape of the viscosity profile tempered at 20 C is similar to that of untempered cream (Fig. 4A). After a lag
phase, the viscosity rises suddenly, indicating that network formation occurs rapidly as soon as fat droplets are close enough to partially coalesce. Next, a small decrease in viscosity was observed that is related to rounding of the irregular aggregates (Fredrick et al., 2013a). Finally, the viscosity rose to a maximum value at tch. The maximum viscosity could be related to the strength of the partially coalesced fat droplets network, and is comparable for untempered NC and NC tempered at 20 C. In contrast, NC tempered at 30 C shows a shorter lag phase and a more gradual increase in viscosity. Thus, shear-induced partial coalescence started earlier but the network formation was slower. The latter effect could be explained by the formation of weaker bonds between the fat droplets that could be ruptured again during shearing. Additionally, a smaller number of possible contact points between the fat droplets could slow down the formation of large aggregates, but could favour the formation of smaller aggregates. The maximum viscosity at tch is lower, suggesting a weaker network of partially coalesced fat droplets. It was shown by Goff (1997) that the presence of proteins at the interface of RC tended to decrease the partial coalescence rate due
96
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Fig. 2. Light microscopic images of natural cream after tempering at 20 C (A and B) and at 30 C (C and D). The inserts are zoomed images of B and D (magnification factor 2). Scale bar ¼ 50 mm.
colloidal repulsion, and the thickness and viscoelasticity of the interfacial layer. The extent of this effect depends upon the type of proteins present (Pelan, Watts, Campbell, & Lips, 1997). Visualisation of RC after production using light microscopy showed that bridging flocculation occurred (data not shown). During homogenisation, proteins are frequently positioned on the interface of two droplets resulting in the formation of clusters. It was shown that those clusters could be separated if the sample was first diluted with sodium dodecyl sulphate, pointing out the reversibility of the interactions (data not shown). The closer contact between the fat droplets because of bridging flocculation probably favours shearinduced partial coalescence in RC, nevertheless, similar effects of tempering were observed for RC as for NC. Tempering at 20 C promoted earlier partial coalescence but the course of the viscosity as a function of time, as well as the viscosity at tch, are similar to untempered RC. Tempering at 30 C also promoted earlier partial coalescence but the course of network formation was slower and the viscosity at tch was lower, suggesting a weaker network. 3.4. Whipping properties Table 1 lists the whipping properties of untempered and tempered NC and RC. Considering NC, tempering induced a considerable decrease in whipping time which seems a logical consequence of the increased susceptibility to shear-induced partial coalescence (Section 3.2.). Additionally, the whipping time was even shorter than the churning time because the capture efficiency was increased by the presence of air bubbles leading to surfacemediated partial coalescence (Hotrum et al., 2005). The overrun of tempered NC was significantly higher. van Aken (2001) reported that the overrun achieves a maximum value after stage one of the whipping process, after which it hardly changes towards the end of the process. In this first stage, the air bubbles were stabilised by the proteins. In the next phases, the proteins were replaced by partially coalesced fat droplets due to the occurrence of partial coalescence. It can be assumed that besides the shear-induced partial
coalescence, surface-mediated partial coalescence was favoured as a consequence of tempering. Consequently, the amount of air introduced in stage one seems to be higher, as during this stage, the fat droplets may already be able to partly stabilise the air bubble surface. The slightly increased serum loss after 24 h storage at 5 C could be the consequence of this higher overrun. More fat droplets are necessary to stabilise the air bubbles whereby less fat droplets are available to reinforce the network of partially coalesced fat droplets in between the air bubbles. Furthermore, the firmness after 24 h storage at 5 C was decreased due to the lower stability. Although faster shear-induced partial coalescence occurred for RC, it took significantly more time for RC to be whipped to an acceptable whipped cream than for NC. This is similar to the observations of Hotrum et al. (2005). As was observed for NC, surfacemediated partial coalescence could increase the capture efficiency mainly during the second phase of the whipping process resulting in a shorter whipping time (Hotrum et al., 2005). In contrast, it seems that for RC churning was delayed in the presence of air bubbles. Possibly, small aggregates were formed at the beginning of the whipping process due to fast shear-induced partial coalescence. If adsorption of aggregates at the air bubble surface is slower, the enveloping of the air bubble is less efficient. The latter observation is shown by cryo-SEM images of whipped NC and whipped RC (Fig. 5). In NC, the air bubble was covered with a monolayer of fat droplets whereas fragments of the air bubble surface in RC were not covered by fat droplets. In NC, the susceptibility of the fat droplets to shear-induced partial coalescence is lower, whereby the individual droplets or small aggregates seem to have enough time to adsorb to the air bubble surface and take part in surface-mediated partial coalescence. Tempering reduced the whipping time of RC, which is similar to the effect observed for NC. On the other hand, the overrun shows the opposite trend. An 8% reduction in overrun is noted for 20 Ctempered RC, and after tempering at 30 C, the overrun was decreased by 30% in comparison with the whipped untempered RC. It is likely that due to the very fast shear-induced partial
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
97
Fig. 3. Cryo-scanning electron microscopy images representing the fat network in droplets of recombined cream without tempering (A), after tempering at 20 C (B and C), and tempering at 30 C (D and E). Arrows are pointing at lamellar regions of fat crystals near the interface. Scale bar ¼ 1 mm.
coalescence, the overrun decreased during the second stage of whipping. The lower overrun for tempered RC led to a higher
firmness. For both tempering at 20 C and 30 C, a higher serum loss was observed during storage at 20 C. Fig. 6 shows fat droplets at
Fig. 4. Viscosity profiles of natural cream (A) and recombined cream (B) that were untempered (d), tempered at 20 C ($$$$), or tempered at 30 C (- - -). The viscosity profiles of natural cream are analysed at 150 s1 and of recombined cream at 50 s1.
98
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Table 1 Whipping properties of natural cream (NC) and recombined cream (RC): untempered and after tempering at 20 C and 30 C.a Cream
Tmax
twh (min)
Overrun (%)
Serum loss (%) 1 h at 20 C
NC
RC
UT T20 T30 UT T20 T30
9.0 4.2 4.2 11.0 7.1 6.8
± ± ± ± ± ±
a
0.4 0.1b 0.1b 0.6a 0.4b 0.6b
111.9 133.2 130.8 139.6 128.9 97.9
± ± ± ± ± ±
b
3.7 2.0a 0.3a 5.7a 1.0b 3.0c
0.5 ± 0.5 ND ND 11.1 ± 1.8c 17.6 ± 4.3b 23.4 ± 4.3a
Firmness (N) 24 h at 5 C 0.2 1.5 2.9 15.5 5.0 3.4
± ± ± ± ± ±
b
0.1 0.1a 2.4a,b 0.9a 1.7b 0.4b
1 h at 5 C 1.5 1.4 1.4 1.0 1.5 1.6
± ± ± ± ± ±
a
0.0 0.1a 0.0a 0.0b 0.1a 0.1a
24 h at 5 C 1.6 1.2 0.8 0.9 1.3 1.6
± ± ± ± ± ±
0.0a 0.2b 0.0c 0.1c 0.3b 0.1a
a Tmax gives the tempering process applied (UT, untempered; T20, tempered at 20 C; T30, tempered at 30 C) and twh gives the whipping time. Data the mean ± standard deviation; different letters indicate significant differences (P < 0.05) between UT, T20 and T30 within one type of cream.
Fig. 5. Cryo-scanning electron microscopy images representing the structure of, (A) whipped natural cream, and (B) whipped recombined cream. Scale bar image A ¼ 1 mm; scale bar image B ¼ 10 mm.
the interface of whipped RC that was tempered at 30 C. The fat droplets look more elongated compared to fat droplets observed in Fig. 5, suggesting more spreading. If the temperature is increased to 20 C, which is the case for the serum loss measurement, too much liquid oil may spread on the bubble surface, which is detrimental for stability (Hotrum, Stuart, van Vliet, & van Aken, 2004). As this was not seen for RC without tempering, this effect could be due to the effect of tempering. Interestingly, storage at 5 C decreased the amount of serum expelled from the whipped cream. The solid fat content (by NMR) of milk fat in RC amounts to 44.8 ± 1.2% at 5 C and 21.0 ± 1.8% at 20 C. The higher solid fat content at 5 C together with the increased spreading at the air bubble surface due to tempering could increase the stability of the air bubbles.
Fig. 6. Cryo-scanning electron microscopy image of whipped recombined cream tempered at 30 C. Scale bar ¼ 1 mm.
4. Discussion On the level of fat crystallisation, it can be concluded that at both 20 C and 30 C, a separation between different melting fractions was obtained after tempering. This confirms the research of Drelon et al. (2006) who investigated tempering of cream after whipping. Moens et al. (2015) studied the influence of tempering by X-ray diffraction. It was shown that this technique was useful in the attribution of polymorphic changes to a shift in fatty acid composition of the crystals. Based on the long spacings, it was demonstrated that a transition of highly mixed crystals to purer crystals occurred, and that by analysis of the domain size, an increase in crystal size was observed after tempering at 20 C or 30 C. The appearance of a new polymorphic form after tempering at 30 C suggests nucleation, in addition to crystal growth. Cryo-SEM images of the fat droplets suggest a change in internal structure. Lamellar regions near the interface can be observed, especially in fat droplets tempered at 20 C. Fig. 7 shows a schematic overview of the hypothesis based on the results of this research. The results obtained in our research confirm the assumptions made by Boode-Boissevain (1992). Fig. 7 is based on these assumptions. Increasing the temperature from 5 C to 20 C diminished the solid fat content of RC from 44.7 ± 1.2% to 21.0 ± 1.8%, and the solid fat content of NC from 45.7 ± 0.9% to 21.2 ± 0.3%. Possibly, the crystals could move to the energetically more favourable interface. Subsequent cooling induced growth of those crystals (Moens et al., 2015). Consequently, the fat crystals near the interface could stick out further into the serum phase, increasing the capture efficiency. Increasing the temperature from 5 C to 30 C diminished the solid fat content of RC from 44.7 ± 1.2% to 4.1 ± 2.0%, and the solid fat content of NC from 45.7 ± 0.9% to 5.9 ± 2.0%. The few crystals that are still present possibly moved to the energetically more favourable interface. As it was shown that subsequent cooling not only induced growth of those crystals, but also nucleation (Moens et al., 2015). It is expected
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Fig. 7. Schematic overview of proposed changes occurring in the fat crystal network of fat droplets during tempering. T ¼ temperature.
that new crystals were formed in the centre of the fat globule that push the fat crystals near the interface further into the serum phase. The capture efficiency probably increased due to a few larger crystals that protruded further out from the fat droplet surface; however, if there are only a few crystals, the amount of possible reactive places for partial coalescence to occur is decreased. This research has shown that changes in the internal fat microstructure influence shear-induced partial coalescence. It was confirmed that tempering of cream at a temperature where part of the fat is still solid increases the susceptibility of the fat droplets to partial coalescence (Boode-Boissevain, 1992; Mutoh et al., 2001; Thivilliers et al., 2006). After tempering at 20 C, shear-induced partial coalescence occurs sooner, probably because the crystals protrude further out from the fat globule into the serum phase. The amount of possible reactive places may be high because many crystals persist at 20 C that could be situated near the interface. Consequently, as soon as partial coalescence started, the aggregates become large very fast. The latter is shown by the high slope of the first rise in the viscosity profile (Fig. 4). The viscosity at phase separation could be related to the strength of the network, suggesting that more connections between fat droplets induce a stronger network and thus a higher viscosity at phase separation. Tempering at 30 C resulted in an earlier start to shear-induced partial coalescence compared to 20 C-tempering. This may be related to the hypothesis that fat crystals protrude even further into the serum phase. However, if less possible reactive places are present, the formation of a partially coalesced fat droplet network occurs more slowly, which is indicated by a lower slope of the first rise in the viscosity profile (Fig. 4). The viscosity at phase separation was lower, suggesting a weaker network. Although shear-induced partial coalescence occurred much faster in RC, the same effects of tempering were observed, suggesting that effects on the fat crystals were similar for both creams and thus not influenced by the interfacial composition. The latter is in contrast with the findings of Thivilliers et al. (2008) who concluded that proteins at the interface protect against the effects of tempering. However, in their research, tempering was carried out without shearing for emulsions with 70% of fat, showing that the effects of tempering are more clearly shown under shearing conditions. Whereas the effect of tempering on the fat crystal network and on shear-induced partial coalescence are the same for both creams, clear differences in whipping properties were observed for NC and RC. Air bubbles seem to promote partial coalescence in NC. The larger penetration distance of the fat crystals possibly promoted surface-mediated partial coalescence by which more air could be incorporated during the first phase of whipping, resulting in a
99
higher overrun. The latter observations are in accordance with the study of Besner and Kessler (1998), however, in contrast, it was observed in the present study that stability was not improved, and softening of the whipped cream occurred during storage. The differences between tempering at 20 C and 30 C are clear for shearinduced partial coalescence, but much less pronounced in terms of whipping behaviour. In contrast to NC, the air bubbles in RC did not seem to favour partial coalescence. Too many aggregates were likely formed in the first phase of the whipping process by which an efficient enveloping of the air bubbles became impossible. Nevertheless, tempering seemed to improve the stability of whipped cream during longer storage at 5 C, which could be attributed to a higher tendency of oil spreading. The whipping time for RC is relatively longer compared with NC, although it could be concluded from the tch values that shear-induced partial coalescence occurred much faster in RC. In conclusion, the churning time cannot, in all cases, be used to predict the whipping time, as whipping is a combination of shear-induced partial coalescence and surface-mediated partial coalescence. 5. Conclusions The effect of tempering on fat crystallisation, shear-induced partial coalescence, and whipping properties of natural and recombined cream was investigated. Tempering led to an increase in susceptibility to shear-induced partial coalescence which could be due to the presence of larger fat crystals near the interface of the fat droplets. The progress of the viscosity during shearing helped in understanding the difference between tempering at 20 C and 30 C. Tempering at 20 C probably resulted in larger crystals at the interface, whereas for 30 C-tempered cream, the gradual progress of shear-induced partial coalescence could be related to less contact points. The latter is explained by the fewer large crystals that persist at 30 C and which were probably pushed further into the serum phase due to nucleation and crystal growth in the centre of the fat globule. Tempering of natural cream caused a shorter whipping time and a higher overrun, but a softer product with a lower stability. In recombined cream, a shorter whipping time was also noted, but in contrast with natural cream, the overrun was lower. Furthermore, a firmer whipped cream was obtained with increased stability upon storage at 5 C. This paper studied the consequences of tempering on the fat crystal network of the fat droplets. It was shown that this greatly affects shear-induced partial coalescence and whipping behaviour. It must be stated that the last two mentioned mechanisms are very sensitive to both compositional and processing factors. Type of fat, emulsifiers, and particle size distribution are amongst others factors that could affect the results noted for the cream analysed in this research. In conclusion, tempering may be an interesting technique to promote partial coalescence in those emulsions where partial coalescence is hampered due to product formulation. Acknowledgements Hercules Foundation is acknowledged for its financial support in the acquisition of the scanning electron microscope JEOL JSM-7100F equipped with cryo-transfer system Quorum PP3000T and Oxford Instruments Aztec EDS (grant number AUGE-09-029). Benny Lewille is greatly acknowledged for his assistance with the experiments. References van Aken, G. A. (2001). Aeration of emulsions by whipping. Colloids and Surfaces APhysicochemical and Engineering Aspects, 190, 333e354.
100
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Besner, H., & Kessler, H. G. (1998). Interfacial interaction during the foaming of nonhomogenized cream. Milchwissenschaft, 53, 682e686. Boode-Boissevain, K. (1992). Partial coalescence in oil-in-water emulsions (Ph.D thesis). Wageningen, The Netherlands: Wageningen Agricultural University. Brooker, B. E., Anderson, M., & Andrews, A. T. (1986). The development of structure in whipped cream. Food Microstructure, 5, 277e285. Drelon, N., Gravier, E., Daheron, L., Boisserie, L., Omari, A., & Leal-Calderon, F. (2006). Influence of tempering on the mechanical properties of whipped dairy creams. International Dairy Journal, 16, 1454e1463. Fredrick, E., Heyman, B., Moens, K., Fischer, S., Verwijlen, T., Moldenaers, P., et al. (2013a). Monoacylglycerols in dairy recombined cream: II. The effect on partial coalescence and whipping properties. Food Research International, 51, 936e945. Fredrick, E., Moens, K., Heyman, B., Fischer, S., Van der Meeren, P., & Dewettinck, K. (2013b). Monoacylglycerols in dairy recombined cream: I. The effect on milk fat crystallization. Food Research International, 51, 892e898. Fredrick, E., Walstra, P., & Dewettinck, K. (2010). Factors governing partial coalescence in oil-in-water emulsions. Advances in Colloid and Interface Science, 153, 30e42. Goff, H. D. (1997). Instability and partial coalescence in whippable dairy emulsions. Journal of Dairy Science, 80, 2620e2630. Gravier, E., Drelon, N., Boisserie, L., Omari, A., & Leal-Calderon, F. (2006). Consolidation of foams deriving from emulsions by temperature cycling (“tempering”). Colloids and Surfaces A-Physicochemical and Engineering Aspects, 282, 360e368. Hotrum, N. E., Stuart, M. A. C., van Vliet, T., Avino, S. F., & van Aken, G. A. (2005). Elucidating the relationship between the spreading coefficient, surfacemediated partial coalescence and the whipping time of artificial cream. Colloids and Surfaces A-Physicochemical and Engineering Aspects, 260, 71e78. Hotrum, N. E., Stuart, M. A. C., van Vliet, T., & van Aken, G. A. (2004). Spreading of partially crystallized oil droplets on an air/water interface. Colloids and Surfaces A-Physicochemical and Engineering Aspects, 240, 83e92. Moens, K., De Clercq, N., Verstringe, S., & Dewettinck, K. (2015). Revealing the influence of tempering on polymorphism and crystal arrangement in semicrystalline oil-in-water emulsions. Crystal Growth and Design. http://dx.doi.org/ 10.1021/acs.cgd.5b00665.
Mutoh, T. A., Kubouchi, H., Noda, M., Shiinoki, Y., & Matsumura, Y. (2007). Effect of oil-soluble emulsifiers on solidification of thermally treated creams. International Dairy Journal, 17, 24e28. Mutoh, T. A., Nakagawa, S., Noda, M., Shiinoki, Y., & Matsumura, Y. (2001). Relationship between characteristics of oil droplets and solidification of thermally treated creams. Journal of the American Oil Chemists' Society, 78, 177e182. Oortwijn, H., & Walstra, P. (1982). The membranes of recombined fat globules. 4. Effects on properties of the recombined milks. Netherlands Milk and Dairy Journal, 36, 279e290. Pelan, B. M. C., Watts, K. M., Campbell, I. J., & Lips, A. (1997). The stability of aerated milk protein emulsions in the presence of small molecule surfactants. Journal of Dairy Science, 80, 2631e2638. Riaublanc, A., Anton, M., Mariette, F., Georges, C., Gravier, E., Drelon, N., et al. (2005). Impact of fat crystals on the foaming capacity and stability of whipped creams. Sciences des Aliments, 25, 427e441. Sugimoto, T., Mori, T., Mano, J., Mutoh, T. A., Shiinoki, Y., & Matsumura, Y. (2001). Effects of fat crystallization on the behavior of proteins and lipids at oil droplet surfaces. Journal of the American Oil Chemists' Society, 78, 183e188. Thivilliers-Arvis, F., Laurichesse, E., Schmitt, V., & Leal-Calderon, F. (2010). Shearinduced instabilities in oil-in-water emulsions comprising partially crystallized droplets. Langmuir, 26, 16782e16790. Thivilliers, F., Drelon, N., Schmitt, V., & Leal-Calderon, F. (2006). Bicontinuous emulsion gels induced by partial coalescence: kinetics and mechanism. Europhysics Letters, 76, 332e338. Thivilliers, F., Laurichesse, E., Saadaoui, H., Leal-Calderon, F., & Schmitt, V. (2008). Thermally induced gelling of oil-in-water emulsions comprising partially crystallized droplets: the impact of interfacial crystals. Langmuir, 24, 13364e13375. Walstra, P. (1967). On crystallization habit in fat globules. Netherlands Milk and Dairy Journal, 21, 166e191. Walstra, P. (2003). Physical chemistry of foods. New York, NY, USA: Marcel Dekker, Inc.
Contents lists available at ScienceDirect
International Dairy Journal journal homepage: www.elsevier.com/locate/idairyj
Tempering of dairy emulsions: Partial coalescence and whipping properties Kim Moens*, A.K.M. Masum, Koen Dewettinck Laboratory of Food Technology and Engineering, Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, 9000 Ghent, Belgium
a r t i c l e i n f o
a b s t r a c t
Article history: Received 9 July 2015 Received in revised form 1 December 2015 Accepted 8 January 2016 Available online 28 January 2016
This study investigates the effect of applying a timeetemperature profile to natural and recombined cream to influence partial coalescence and, consequently, the whipping quality. To date, no clear relationship exists between the consequences of tempering on a microstructural level, partial coalescence, and whipping properties. Milk fat crystallisation was analysed using differential scanning calorimetry and the internal arrangement of fat crystals was visualised with cryo-scanning electron microscopy. Shear-induced partial coalescence and whipping properties were studied. Shear-induced partial coalescence was promoted, attributed to the observed changes in the fat crystal network. The effects on whipping properties were different for natural and recombined cream and thus dependent upon the interfacial composition. Consolidation of the partially coalesced fat droplet network by tempering increased the stability of whipped recombined cream during cold storage. Tempering is a promising tool to alter the susceptibility to partial coalescence by changing the internal fat crystal network, and influencing whippability. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Dairy cream is an emulsion of milk fat droplets in a continuous aqueous phase. An important characteristic of dairy cream is the ability of transforming the oil-in-water (O/W) emulsion into a ternary system by whipping, which is a combination of shear and air inclusion. A good quality whipped cream has a high overrun, a desirable firmness, and a high stability. The basic mechanism occurring during whipping is partial coalescence. Firstly, shearinduced partial coalescence occurs when fat droplets connect by protruding fat crystals. These fat crystals penetrate the interface of another fat droplet forming a connection (Fredrick, Walstra, & Dewettinck, 2010). Secondly, surface-mediated partial coalescence takes place when fat droplets enter the air bubbles during whipping, liquid oil is released from the fat crystal network and spread over the air bubble surface. If this occurs between two fat droplets that are close to each other at the air bubble surface, a junction is formed leading to partial coalescence (Hotrum, Stuart, van Vliet, Avino, & van Aken, 2005). The combination of both mechanisms results in a network of partially coalesced fat droplets,
* Corresponding author. Tel.: þ32 9 264 6198. E-mail address: [email protected] (K. Moens). http://dx.doi.org/10.1016/j.idairyj.2016.01.007 0958-6946/© 2016 Elsevier Ltd. All rights reserved.
giving structure to whipped cream by stabilising the air bubbles (Brooker, Anderson, & Andrews, 1986; Goff, 1997). Natural cream, obtained by the physical separation of a high-fat phase from milk, is considered the gold standard. It has exceptionally good whipping properties and a nice taste. However, for many reasons (amongst others, health, technological and practical reasons) the food industry aims to develop recombined creams obtained by emulsification of fat in an aqueous phase. Ingredients can vary greatly, using different types of fat, emulsifiers, hydrocolloids, sugars, sugar replacers, and flavours. Nevertheless, the whipping properties should always be optimised by ensuring an efficient envelopment of the air bubbles by partially coalesced fat droplets. Diverse strategies may be followed in adapting partial coalescence, such as the composition of the cream (proteins, emulsifiers, hydrocolloids and the amount and type of fat), and the processing conditions (temperature, shear rate, cooling rate, homogenisation pressure and tempering) (Fredrick et al., 2010; Hotrum et al., 2005; van Aken, 2001). The latter is the main subject of this study, and consists of the application of a specific timeetemperature profile in which the temperature is first increased to partly melt the fat, and then decreased again to induce heterogeneous crystallisation of the liquid fat (Boode-Boissevain, 1992; Drelon et al., 2006; Gravier, Drelon, Boisserie, Omari, & Leal-Calderon, 2006; Mutoh, Nakagawa, Noda, Shiinoki, &
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Matsumura, 2001; Mutoh, Kubouchi, Noda, Shiinoki, & Matsumura, 2007; Oortwijn & Walstra, 1982; Sugimoto et al., 2001; ThivilliersArvis, Laurichesse, Schmitt, & Leal-Calderon, 2010; Thivilliers, Drelon, Schmitt, & Leal-Calderon, 2006; Thivilliers, Laurichesse, Saadaoui, Leal-Calderon, & Schmitt, 2008). The pioneering research of Boode-Boissevain (1992) on tempering of emulsions suggests that tempering increases the susceptibility to partial coalescence due to repositioning of the fat crystals. Increasing the temperature leads to partial melting of the fat crystals by which the three dimensional fat crystal network is broken. The remaining fat crystals can move freely to the energetically more favourable interface. Decreasing the temperature induces growth of the crystals near the interface, resulting in larger crystals that can penetrate further into the aqueous phase, increasing the susceptibility to partial coalescence. Emulsions with a fat content lower than 25% are not prone to the effect of tempering (Boode-Boissevain, 1992), and a minimal amount of solid fat during tempering, estimated to be between 1.5% and 8%, is necessary (Boode-Boissevain, 1992; Drelon et al., 2006; Mutoh et al., 2007). Moens, De Clercq, Verstringe, and Dewettinck (2015) showed that tempering significantly influences fat crystallisation properties, depending upon the tempering temperature. Besides the change in melting properties and fat crystal polymorphism, the internal arrangement of the fat droplets is also influenced. Lamellar structures near the interface of the fat droplets were visualised in recombined cream tempered at 20 C. The latter observation adds to the hypothesis that crystals move to the interface. Although fat crystallisation properties largely determine the effect of tempering, the role of the interfacial composition may not be underestimated. Thivilliers et al. (2006) showed also that under quiescent conditions, partial coalescence may be promoted by tempering to such an extent that the emulsion forms a gel. Besides the fat crystallisation properties, the composition of the interface greatly influenced this gelling behaviour (Thivilliers et al., 2008). More specifically, tempering seems to have less effect on proteinstabilised O/W emulsions because of the thick interface. Conversely, O/W emulsions stabilised with low molecular surfactants have a thinner interface allowing crystals to pierce through, adding to the effect of tempering (Mutoh et al., 2007; Thivilliers et al., 2008). Mutoh et al. (2001) and Sugimoto et al. (2001) investigated the effect of tempering on protein-stabilised vegetable creams. Despite the thick interface, they observed increased flocculation after tempering which was not due to partial coalescence. It was suggested that the fat crystals near the interface could affect the adsorption of the proteins, consequently affecting interactions between the fat droplets. Tempering is likely to affect the whipping properties. It was demonstrated that tempering of natural cream at 30 C could improve its whipping properties by reducing the whipping time and serum loss, and increasing the overrun and firmness (Besner & Kessler, 1998); however, no link with partial coalescence or any other possible mechanism explaining the effects of tempering was provided in this study. Tempering the cream after it was whipped seemed to have similar effects. It was observed that for both natural and recombined cream, the elastic behaviour of the whipped cream increased when tempered at 25 C due to an increased connectivity between the fat droplets. The latter is explained by surfacemediated partial coalescence occurring at the air bubble surface (Drelon et al., 2006; Gravier et al., 2006; Riaublanc et al., 2005). In this current study, the relationship between tempering, partial coalescence kinetics, and whipping properties was studied. Two types of dairy cream, natural cream and recombined cream, were subjected to tempering at two temperatures (Tmax) of 20 C and 30 C.
93
2. Materials and methods 2.1. Materials Commercially available natural cream (NC; Debic, FrieslandCampina Professionals, Lummen, Belgium) with a fat content of 35% was used. Furthermore, recombined cream (RC) was produced by recombining anhydrous milk fat (FrieslandCampina butter, Noordwijk, The Netherlands), sweet cream butter milk powder (Westbury Dairies, Westbury, UK; containing 30.1 ± 0.1% proteins, 8.5 ± 0.1% fat and 51.5 ± 0.2% lactose), carrageenan (Satiagel ME4; Cargill Deutschland GmbH, Krefeld, Germany) and potable water. Sodium azide (Acros Organics, Geel, Belgium) was added to prevent microbial spoilage. Sweet cream butter milk powder (7.6%) was dissolved in potable water and kept overnight for full hydration. Carrageenan (0.01%) and sodium azide (0.01%) were added to this aqueous phase and pre-heated to 50 C. The anhydrous milk fat (35%) was melted and pre-heated to 50 C. The pre-emulsion was prepared by thoroughly mixing the aqueous phase with the fat phase using an Ultra-Turrax (10,000 rpm) for 10 min at 50 C. This pre-emulsion was then homogenised using a two-step laboratory scale homogeniser (APV cooling systems, Alberslund, Denmark) at 3e1 MPa. Finally, the emulsion was rapidly cooled to 5 C and stored in a thermostatic cabinet at 5 C for at least 7 d to complete fat crystallisation before tempering was applied. 2.2. Laser light diffraction The size of the fat droplets in natural cream and in recombined cream was measured with laser light diffraction according to the method described by Fredrick et al. (2013b). 2.3. Tempering procedure Cream (2.5 L) was heated to Tmax (20 or 30 C) using a Herbst HR-S3 Tempering Unit equipped with a double walled mixing bowl (Herbst Machinery Ltd, Omagh, Ireland) temperature-regulated €ltemaschinenbau with a Huber Thermostat (Peter Huber Ka GmbH, Offenburg, Germany). The sample was kept isothermally for 30 min at Tmax under continuous stirring before it was cooled to 5 C by passing the cream through tubing that was fitted in a water bath at 5 C. The cream was stored at 5 C for at least 7 d to complete crystallisation before analysis. 2.4. Differential scanning calorimetry Differential scanning calorimetry (DSC) with a refrigerated cooling system (model Q1000 DSC; TA Instruments, New Castle, DE, USA) was used to study the melting behaviour of (tempered) anhydrous milk fat (AMF), NC and RC. DSC was calibrated with indium (TA Instruments), azobenzene (SigmaeAldrich, Bornem, Belgium) and undecane (Acros Organics) prior to analysis. Nitrogen was used to purge the system. Samples (5e15 mg) were sealed in hermetic aluminium pans, and an empty pan was used as a reference. The melting profile was analysed by heating the sample at 5 C per min from 5 C to 55 C. Measurements were carried out in triplicate. 2.5. Nuclear magnetic resonance A Maran Ultra 23 MHz pulsed-field gradient nuclear magnetic resonance (NMR) instrument (Oxford Instruments, Tubney Woods, Abingdon, UK) was used to measure the solid fat content profile of NC and RC. The solid fat content was determined 1 week after
94
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
production in the range 5e50 C. An indirect solid fat content (SFC) measurement was carried out to eliminate the contribution of the aqueous phase to the free induction decay signal. A detailed description of this method is given in Fredrick et al. (2013b). 2.6. Light microscopy A Leica DM 2500 (Leica Microsystems Belgium BVBA, Diegem, Belgium) light microscope was used to visualise the fat droplets in cream. The cream was first diluted (1:10) in demineralised water. Thereafter, one drop was transferred to a microscopic slide and covered with a cover slip. The images were recorded with a Leica MC170HD camera (Leica Camera, Wetzlar, Germany). 2.7. Cryogenic scanning electron microscopy The microstructure of unwhipped and whipped cream was visualised using scanning electron microscopy with cryogenic sample preparation. Cream was mounted on an aluminium stub and quickly frozen in slushed nitrogen (210 C). The time between sample preparation and freezing was kept as short as possible to avoid crystal melting. Then, the sample was fractured at 150 C, sputter-coated with platinum, and finally observed with a Jeol JSM7100F scanning electron microscope (Jeol Europe B.V., Zaventem, Belgium). 2.8. Rotational viscosity analyses Rotational viscosity analyses were carried out to study the kinetics of partial coalescence without the inclusion of air, similar to the research of Fredrick et al. (2013a). An AR2000ex controlled stress rheometer (TA instruments, Brussels, Belgium) equipped with a starch pasting cell as measuring geometry was used. To follow partial coalescence, 30 mL of cream was transferred into the cup, and after an equilibration period at the measuring temperature (20 C), a constant shear rate was applied and the apparent viscosity of the cream was followed as a function of time. The time when the viscosity rises to a maximum was defined as the churning time (tch). Preliminary results showed that RC is much more sensitive to shear than NC. Therefore, a constant shear rate of 150 s1 was applied to NC, whereas a constant shear rate of 50 s1 was applied to RC. 2.9. Whipping properties Whipping was carried out with a Hobart N50 mixer (Illinois Tool Works, Glenview Chicago, IL, USA) equipped with a D-wire whip designed for maximum air inclusion in a low viscous product. The medium speed (position 2: 240 rpm) was applied at 5 C. All creams were whipped until a defined observable end point. The optimum whipping time is defined as the time at which the cream broke away from the wires and the bowl. Moreover, when the whipping process is stopped at this optimum whipping time (twh), the cream should not immediately start to flow around the wires of the impeller. These tests were always carried out by the same person who was trained to detect this optimal point. Whipping experiments were repeated at least twice for each cream (i.e., two whipping batches). For each whipping batch, the overrun, the serum loss, and the firmness were determined according to the methods described by Fredrick et al. (2013a). 2.10. Statistical analysis SPSS software (version 22, SPSS Inc., Chicago, IL, USA) was used for statistical comparison of the whipping properties. Analyses of
variance (one-way ANOVA) were carried out to determine significant differences between the results. A Levene test was used to test equality of variances. If this hypothesis was retained, the Tukey's post hoc test was executed for pairwise comparison. If the variances were not equal, the Dunnett's T3 post hoc test was used. All tests were carried out at a 95% significance level. 3. Results 3.1. Particle size distribution Fat crystallisation in emulsion droplets, as well as partial coalescence and the whipping properties, are greatly influenced by the droplet size. To eliminate this factor, the homogenisation process of recombined cream was adapted to obtain an emulsion of which the droplet size is in the same order of magnitude as natural cream. The Sauter diameter of natural and recombined cream was 2.95 ± 0.01 mm and 2.84 ± 0.01 mm, respectively. 3.2. Fat crystallisation The melting behaviour of pure AMF, NC and RC, whether tempered or not, is presented in Fig. 1. The untempered samples show the typical broad melting range resulting from the high diversity of fatty acids present in milk fat. This broad melting peak is separated into two distinguishable melting peaks after tempering at 20 C, of which the separation between both peaks is situated around 23 C. This separation is more clear in pure AMF because of the dilution effect in cream. Tempering at 30 C also results in a separation into two distinguishable peaks, but this time the peaks are separated at around 33 C. Milk fat has the tendency to crystallise in highly mixed crystals, i.e., crystals comprising different triacylglycerides (Walstra, 2003). At Tmax, a part of the crystals are melted and the released triacylglycerides are rearranged in such a way that purer crystals arise (Moens et al., 2015). This process can be compared with the fractionation process. Similar changes in melting behaviour are observed for pure AMF, as also for cream, indicating that this process occurs simultaneously in every individual fat droplet. Fig. 2A and B show NC after tempering at 20 C, and Fig. 2C and D show NC after tempering at 30 C. It is seen from Fig. 2A that 20 C tempering induces little or no aggregation, and that most of the fat droplets are still present as individual entities. It seems that partial coalescence does not occur during the tempering process. In contrast, Fig. 2C shows that during tempering at 30 C, some clustering occurs. Heating of those clusters resulted in complete merging indicating that the fat droplets are partially coalesced. The elongated shape of the clusters suggests that one fat droplet is connected to only two other fat droplets. The fat droplets in 20 C tempered cream show a more irregular surface (Fig. 2B) compared to fat droplets that were tempered at 30 C (Fig. 2D), suggesting differences in the fat crystal network of both creams. Likewise, Thivilliers et al. (2006) designates the rough and rippled surface of fat droplets to irregularly shaped/oriented crystals. These findings are validated by cryo-SEM images of RC obtained before and after tempering (Fig. 3). Fig. 3 shows the section of fat droplets exposing the fat crystal network in the droplets. A random distribution of fat crystals in the fat droplet was observed before tempering. Interestingly, well-defined lamellar regions near the interface were detected in fat droplets subjected to tempering at 20 C. This is probably the reason why irregular fat droplets were observed with light microscopy after tempering at 20 C. The latter also contributes to the hypothesis that the remaining fat crystals at Tmax move to the energetically more favourable interface where they grow during subsequent cooling, resulting in L- and M-type fat droplets
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
95
Fig. 1. Untempered (d), tempered at 20 C (e e e) and tempered at 30 C (- - -) melting curves of (A) anhydrous milk fat, (B) natural cream, and (C) recombined cream.
(Boode-Boissevain, 1992; Mutoh et al., 2001; Sugimoto et al., 2001; Thivilliers et al., 2008; Walstra, 1967). The same conclusions could be drawn for NC and RC. 3.3. Shear-induced partial coalescence Shear-induced partial coalescence occurs in cream when shear is applied in the absence of air. In this research, the samples were sheared at a constant shear rate until the time when the viscosity rises to a maximum, i.e., the churning time (tch), and relates to the susceptibility of the fat droplets to partial coalescence. Phase separation occurs at this point. During this experiment, the viscosity was measured as a function of time. The shape of the viscosity profile gives information about the progress of shear-induced partial coalescence. Fig. 4A shows the viscosity profile of NC and Fig. 4B shows the viscosity profile of RC, before and after tempering at 20 C and 30 C. Tempering clearly reduces tch of both types of cream and at both temperatures. Considering NC, the shape of the viscosity profile tempered at 20 C is similar to that of untempered cream (Fig. 4A). After a lag
phase, the viscosity rises suddenly, indicating that network formation occurs rapidly as soon as fat droplets are close enough to partially coalesce. Next, a small decrease in viscosity was observed that is related to rounding of the irregular aggregates (Fredrick et al., 2013a). Finally, the viscosity rose to a maximum value at tch. The maximum viscosity could be related to the strength of the partially coalesced fat droplets network, and is comparable for untempered NC and NC tempered at 20 C. In contrast, NC tempered at 30 C shows a shorter lag phase and a more gradual increase in viscosity. Thus, shear-induced partial coalescence started earlier but the network formation was slower. The latter effect could be explained by the formation of weaker bonds between the fat droplets that could be ruptured again during shearing. Additionally, a smaller number of possible contact points between the fat droplets could slow down the formation of large aggregates, but could favour the formation of smaller aggregates. The maximum viscosity at tch is lower, suggesting a weaker network of partially coalesced fat droplets. It was shown by Goff (1997) that the presence of proteins at the interface of RC tended to decrease the partial coalescence rate due
96
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Fig. 2. Light microscopic images of natural cream after tempering at 20 C (A and B) and at 30 C (C and D). The inserts are zoomed images of B and D (magnification factor 2). Scale bar ¼ 50 mm.
colloidal repulsion, and the thickness and viscoelasticity of the interfacial layer. The extent of this effect depends upon the type of proteins present (Pelan, Watts, Campbell, & Lips, 1997). Visualisation of RC after production using light microscopy showed that bridging flocculation occurred (data not shown). During homogenisation, proteins are frequently positioned on the interface of two droplets resulting in the formation of clusters. It was shown that those clusters could be separated if the sample was first diluted with sodium dodecyl sulphate, pointing out the reversibility of the interactions (data not shown). The closer contact between the fat droplets because of bridging flocculation probably favours shearinduced partial coalescence in RC, nevertheless, similar effects of tempering were observed for RC as for NC. Tempering at 20 C promoted earlier partial coalescence but the course of the viscosity as a function of time, as well as the viscosity at tch, are similar to untempered RC. Tempering at 30 C also promoted earlier partial coalescence but the course of network formation was slower and the viscosity at tch was lower, suggesting a weaker network. 3.4. Whipping properties Table 1 lists the whipping properties of untempered and tempered NC and RC. Considering NC, tempering induced a considerable decrease in whipping time which seems a logical consequence of the increased susceptibility to shear-induced partial coalescence (Section 3.2.). Additionally, the whipping time was even shorter than the churning time because the capture efficiency was increased by the presence of air bubbles leading to surfacemediated partial coalescence (Hotrum et al., 2005). The overrun of tempered NC was significantly higher. van Aken (2001) reported that the overrun achieves a maximum value after stage one of the whipping process, after which it hardly changes towards the end of the process. In this first stage, the air bubbles were stabilised by the proteins. In the next phases, the proteins were replaced by partially coalesced fat droplets due to the occurrence of partial coalescence. It can be assumed that besides the shear-induced partial
coalescence, surface-mediated partial coalescence was favoured as a consequence of tempering. Consequently, the amount of air introduced in stage one seems to be higher, as during this stage, the fat droplets may already be able to partly stabilise the air bubble surface. The slightly increased serum loss after 24 h storage at 5 C could be the consequence of this higher overrun. More fat droplets are necessary to stabilise the air bubbles whereby less fat droplets are available to reinforce the network of partially coalesced fat droplets in between the air bubbles. Furthermore, the firmness after 24 h storage at 5 C was decreased due to the lower stability. Although faster shear-induced partial coalescence occurred for RC, it took significantly more time for RC to be whipped to an acceptable whipped cream than for NC. This is similar to the observations of Hotrum et al. (2005). As was observed for NC, surfacemediated partial coalescence could increase the capture efficiency mainly during the second phase of the whipping process resulting in a shorter whipping time (Hotrum et al., 2005). In contrast, it seems that for RC churning was delayed in the presence of air bubbles. Possibly, small aggregates were formed at the beginning of the whipping process due to fast shear-induced partial coalescence. If adsorption of aggregates at the air bubble surface is slower, the enveloping of the air bubble is less efficient. The latter observation is shown by cryo-SEM images of whipped NC and whipped RC (Fig. 5). In NC, the air bubble was covered with a monolayer of fat droplets whereas fragments of the air bubble surface in RC were not covered by fat droplets. In NC, the susceptibility of the fat droplets to shear-induced partial coalescence is lower, whereby the individual droplets or small aggregates seem to have enough time to adsorb to the air bubble surface and take part in surface-mediated partial coalescence. Tempering reduced the whipping time of RC, which is similar to the effect observed for NC. On the other hand, the overrun shows the opposite trend. An 8% reduction in overrun is noted for 20 Ctempered RC, and after tempering at 30 C, the overrun was decreased by 30% in comparison with the whipped untempered RC. It is likely that due to the very fast shear-induced partial
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
97
Fig. 3. Cryo-scanning electron microscopy images representing the fat network in droplets of recombined cream without tempering (A), after tempering at 20 C (B and C), and tempering at 30 C (D and E). Arrows are pointing at lamellar regions of fat crystals near the interface. Scale bar ¼ 1 mm.
coalescence, the overrun decreased during the second stage of whipping. The lower overrun for tempered RC led to a higher
firmness. For both tempering at 20 C and 30 C, a higher serum loss was observed during storage at 20 C. Fig. 6 shows fat droplets at
Fig. 4. Viscosity profiles of natural cream (A) and recombined cream (B) that were untempered (d), tempered at 20 C ($$$$), or tempered at 30 C (- - -). The viscosity profiles of natural cream are analysed at 150 s1 and of recombined cream at 50 s1.
98
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Table 1 Whipping properties of natural cream (NC) and recombined cream (RC): untempered and after tempering at 20 C and 30 C.a Cream
Tmax
twh (min)
Overrun (%)
Serum loss (%) 1 h at 20 C
NC
RC
UT T20 T30 UT T20 T30
9.0 4.2 4.2 11.0 7.1 6.8
± ± ± ± ± ±
a
0.4 0.1b 0.1b 0.6a 0.4b 0.6b
111.9 133.2 130.8 139.6 128.9 97.9
± ± ± ± ± ±
b
3.7 2.0a 0.3a 5.7a 1.0b 3.0c
0.5 ± 0.5 ND ND 11.1 ± 1.8c 17.6 ± 4.3b 23.4 ± 4.3a
Firmness (N) 24 h at 5 C 0.2 1.5 2.9 15.5 5.0 3.4
± ± ± ± ± ±
b
0.1 0.1a 2.4a,b 0.9a 1.7b 0.4b
1 h at 5 C 1.5 1.4 1.4 1.0 1.5 1.6
± ± ± ± ± ±
a
0.0 0.1a 0.0a 0.0b 0.1a 0.1a
24 h at 5 C 1.6 1.2 0.8 0.9 1.3 1.6
± ± ± ± ± ±
0.0a 0.2b 0.0c 0.1c 0.3b 0.1a
a Tmax gives the tempering process applied (UT, untempered; T20, tempered at 20 C; T30, tempered at 30 C) and twh gives the whipping time. Data the mean ± standard deviation; different letters indicate significant differences (P < 0.05) between UT, T20 and T30 within one type of cream.
Fig. 5. Cryo-scanning electron microscopy images representing the structure of, (A) whipped natural cream, and (B) whipped recombined cream. Scale bar image A ¼ 1 mm; scale bar image B ¼ 10 mm.
the interface of whipped RC that was tempered at 30 C. The fat droplets look more elongated compared to fat droplets observed in Fig. 5, suggesting more spreading. If the temperature is increased to 20 C, which is the case for the serum loss measurement, too much liquid oil may spread on the bubble surface, which is detrimental for stability (Hotrum, Stuart, van Vliet, & van Aken, 2004). As this was not seen for RC without tempering, this effect could be due to the effect of tempering. Interestingly, storage at 5 C decreased the amount of serum expelled from the whipped cream. The solid fat content (by NMR) of milk fat in RC amounts to 44.8 ± 1.2% at 5 C and 21.0 ± 1.8% at 20 C. The higher solid fat content at 5 C together with the increased spreading at the air bubble surface due to tempering could increase the stability of the air bubbles.
Fig. 6. Cryo-scanning electron microscopy image of whipped recombined cream tempered at 30 C. Scale bar ¼ 1 mm.
4. Discussion On the level of fat crystallisation, it can be concluded that at both 20 C and 30 C, a separation between different melting fractions was obtained after tempering. This confirms the research of Drelon et al. (2006) who investigated tempering of cream after whipping. Moens et al. (2015) studied the influence of tempering by X-ray diffraction. It was shown that this technique was useful in the attribution of polymorphic changes to a shift in fatty acid composition of the crystals. Based on the long spacings, it was demonstrated that a transition of highly mixed crystals to purer crystals occurred, and that by analysis of the domain size, an increase in crystal size was observed after tempering at 20 C or 30 C. The appearance of a new polymorphic form after tempering at 30 C suggests nucleation, in addition to crystal growth. Cryo-SEM images of the fat droplets suggest a change in internal structure. Lamellar regions near the interface can be observed, especially in fat droplets tempered at 20 C. Fig. 7 shows a schematic overview of the hypothesis based on the results of this research. The results obtained in our research confirm the assumptions made by Boode-Boissevain (1992). Fig. 7 is based on these assumptions. Increasing the temperature from 5 C to 20 C diminished the solid fat content of RC from 44.7 ± 1.2% to 21.0 ± 1.8%, and the solid fat content of NC from 45.7 ± 0.9% to 21.2 ± 0.3%. Possibly, the crystals could move to the energetically more favourable interface. Subsequent cooling induced growth of those crystals (Moens et al., 2015). Consequently, the fat crystals near the interface could stick out further into the serum phase, increasing the capture efficiency. Increasing the temperature from 5 C to 30 C diminished the solid fat content of RC from 44.7 ± 1.2% to 4.1 ± 2.0%, and the solid fat content of NC from 45.7 ± 0.9% to 5.9 ± 2.0%. The few crystals that are still present possibly moved to the energetically more favourable interface. As it was shown that subsequent cooling not only induced growth of those crystals, but also nucleation (Moens et al., 2015). It is expected
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Fig. 7. Schematic overview of proposed changes occurring in the fat crystal network of fat droplets during tempering. T ¼ temperature.
that new crystals were formed in the centre of the fat globule that push the fat crystals near the interface further into the serum phase. The capture efficiency probably increased due to a few larger crystals that protruded further out from the fat droplet surface; however, if there are only a few crystals, the amount of possible reactive places for partial coalescence to occur is decreased. This research has shown that changes in the internal fat microstructure influence shear-induced partial coalescence. It was confirmed that tempering of cream at a temperature where part of the fat is still solid increases the susceptibility of the fat droplets to partial coalescence (Boode-Boissevain, 1992; Mutoh et al., 2001; Thivilliers et al., 2006). After tempering at 20 C, shear-induced partial coalescence occurs sooner, probably because the crystals protrude further out from the fat globule into the serum phase. The amount of possible reactive places may be high because many crystals persist at 20 C that could be situated near the interface. Consequently, as soon as partial coalescence started, the aggregates become large very fast. The latter is shown by the high slope of the first rise in the viscosity profile (Fig. 4). The viscosity at phase separation could be related to the strength of the network, suggesting that more connections between fat droplets induce a stronger network and thus a higher viscosity at phase separation. Tempering at 30 C resulted in an earlier start to shear-induced partial coalescence compared to 20 C-tempering. This may be related to the hypothesis that fat crystals protrude even further into the serum phase. However, if less possible reactive places are present, the formation of a partially coalesced fat droplet network occurs more slowly, which is indicated by a lower slope of the first rise in the viscosity profile (Fig. 4). The viscosity at phase separation was lower, suggesting a weaker network. Although shear-induced partial coalescence occurred much faster in RC, the same effects of tempering were observed, suggesting that effects on the fat crystals were similar for both creams and thus not influenced by the interfacial composition. The latter is in contrast with the findings of Thivilliers et al. (2008) who concluded that proteins at the interface protect against the effects of tempering. However, in their research, tempering was carried out without shearing for emulsions with 70% of fat, showing that the effects of tempering are more clearly shown under shearing conditions. Whereas the effect of tempering on the fat crystal network and on shear-induced partial coalescence are the same for both creams, clear differences in whipping properties were observed for NC and RC. Air bubbles seem to promote partial coalescence in NC. The larger penetration distance of the fat crystals possibly promoted surface-mediated partial coalescence by which more air could be incorporated during the first phase of whipping, resulting in a
99
higher overrun. The latter observations are in accordance with the study of Besner and Kessler (1998), however, in contrast, it was observed in the present study that stability was not improved, and softening of the whipped cream occurred during storage. The differences between tempering at 20 C and 30 C are clear for shearinduced partial coalescence, but much less pronounced in terms of whipping behaviour. In contrast to NC, the air bubbles in RC did not seem to favour partial coalescence. Too many aggregates were likely formed in the first phase of the whipping process by which an efficient enveloping of the air bubbles became impossible. Nevertheless, tempering seemed to improve the stability of whipped cream during longer storage at 5 C, which could be attributed to a higher tendency of oil spreading. The whipping time for RC is relatively longer compared with NC, although it could be concluded from the tch values that shear-induced partial coalescence occurred much faster in RC. In conclusion, the churning time cannot, in all cases, be used to predict the whipping time, as whipping is a combination of shear-induced partial coalescence and surface-mediated partial coalescence. 5. Conclusions The effect of tempering on fat crystallisation, shear-induced partial coalescence, and whipping properties of natural and recombined cream was investigated. Tempering led to an increase in susceptibility to shear-induced partial coalescence which could be due to the presence of larger fat crystals near the interface of the fat droplets. The progress of the viscosity during shearing helped in understanding the difference between tempering at 20 C and 30 C. Tempering at 20 C probably resulted in larger crystals at the interface, whereas for 30 C-tempered cream, the gradual progress of shear-induced partial coalescence could be related to less contact points. The latter is explained by the fewer large crystals that persist at 30 C and which were probably pushed further into the serum phase due to nucleation and crystal growth in the centre of the fat globule. Tempering of natural cream caused a shorter whipping time and a higher overrun, but a softer product with a lower stability. In recombined cream, a shorter whipping time was also noted, but in contrast with natural cream, the overrun was lower. Furthermore, a firmer whipped cream was obtained with increased stability upon storage at 5 C. This paper studied the consequences of tempering on the fat crystal network of the fat droplets. It was shown that this greatly affects shear-induced partial coalescence and whipping behaviour. It must be stated that the last two mentioned mechanisms are very sensitive to both compositional and processing factors. Type of fat, emulsifiers, and particle size distribution are amongst others factors that could affect the results noted for the cream analysed in this research. In conclusion, tempering may be an interesting technique to promote partial coalescence in those emulsions where partial coalescence is hampered due to product formulation. Acknowledgements Hercules Foundation is acknowledged for its financial support in the acquisition of the scanning electron microscope JEOL JSM-7100F equipped with cryo-transfer system Quorum PP3000T and Oxford Instruments Aztec EDS (grant number AUGE-09-029). Benny Lewille is greatly acknowledged for his assistance with the experiments. References van Aken, G. A. (2001). Aeration of emulsions by whipping. Colloids and Surfaces APhysicochemical and Engineering Aspects, 190, 333e354.
100
K. Moens et al. / International Dairy Journal 56 (2016) 92e100
Besner, H., & Kessler, H. G. (1998). Interfacial interaction during the foaming of nonhomogenized cream. Milchwissenschaft, 53, 682e686. Boode-Boissevain, K. (1992). Partial coalescence in oil-in-water emulsions (Ph.D thesis). Wageningen, The Netherlands: Wageningen Agricultural University. Brooker, B. E., Anderson, M., & Andrews, A. T. (1986). The development of structure in whipped cream. Food Microstructure, 5, 277e285. Drelon, N., Gravier, E., Daheron, L., Boisserie, L., Omari, A., & Leal-Calderon, F. (2006). Influence of tempering on the mechanical properties of whipped dairy creams. International Dairy Journal, 16, 1454e1463. Fredrick, E., Heyman, B., Moens, K., Fischer, S., Verwijlen, T., Moldenaers, P., et al. (2013a). Monoacylglycerols in dairy recombined cream: II. The effect on partial coalescence and whipping properties. Food Research International, 51, 936e945. Fredrick, E., Moens, K., Heyman, B., Fischer, S., Van der Meeren, P., & Dewettinck, K. (2013b). Monoacylglycerols in dairy recombined cream: I. The effect on milk fat crystallization. Food Research International, 51, 892e898. Fredrick, E., Walstra, P., & Dewettinck, K. (2010). Factors governing partial coalescence in oil-in-water emulsions. Advances in Colloid and Interface Science, 153, 30e42. Goff, H. D. (1997). Instability and partial coalescence in whippable dairy emulsions. Journal of Dairy Science, 80, 2620e2630. Gravier, E., Drelon, N., Boisserie, L., Omari, A., & Leal-Calderon, F. (2006). Consolidation of foams deriving from emulsions by temperature cycling (“tempering”). Colloids and Surfaces A-Physicochemical and Engineering Aspects, 282, 360e368. Hotrum, N. E., Stuart, M. A. C., van Vliet, T., Avino, S. F., & van Aken, G. A. (2005). Elucidating the relationship between the spreading coefficient, surfacemediated partial coalescence and the whipping time of artificial cream. Colloids and Surfaces A-Physicochemical and Engineering Aspects, 260, 71e78. Hotrum, N. E., Stuart, M. A. C., van Vliet, T., & van Aken, G. A. (2004). Spreading of partially crystallized oil droplets on an air/water interface. Colloids and Surfaces A-Physicochemical and Engineering Aspects, 240, 83e92. Moens, K., De Clercq, N., Verstringe, S., & Dewettinck, K. (2015). Revealing the influence of tempering on polymorphism and crystal arrangement in semicrystalline oil-in-water emulsions. Crystal Growth and Design. http://dx.doi.org/ 10.1021/acs.cgd.5b00665.
Mutoh, T. A., Kubouchi, H., Noda, M., Shiinoki, Y., & Matsumura, Y. (2007). Effect of oil-soluble emulsifiers on solidification of thermally treated creams. International Dairy Journal, 17, 24e28. Mutoh, T. A., Nakagawa, S., Noda, M., Shiinoki, Y., & Matsumura, Y. (2001). Relationship between characteristics of oil droplets and solidification of thermally treated creams. Journal of the American Oil Chemists' Society, 78, 177e182. Oortwijn, H., & Walstra, P. (1982). The membranes of recombined fat globules. 4. Effects on properties of the recombined milks. Netherlands Milk and Dairy Journal, 36, 279e290. Pelan, B. M. C., Watts, K. M., Campbell, I. J., & Lips, A. (1997). The stability of aerated milk protein emulsions in the presence of small molecule surfactants. Journal of Dairy Science, 80, 2631e2638. Riaublanc, A., Anton, M., Mariette, F., Georges, C., Gravier, E., Drelon, N., et al. (2005). Impact of fat crystals on the foaming capacity and stability of whipped creams. Sciences des Aliments, 25, 427e441. Sugimoto, T., Mori, T., Mano, J., Mutoh, T. A., Shiinoki, Y., & Matsumura, Y. (2001). Effects of fat crystallization on the behavior of proteins and lipids at oil droplet surfaces. Journal of the American Oil Chemists' Society, 78, 183e188. Thivilliers-Arvis, F., Laurichesse, E., Schmitt, V., & Leal-Calderon, F. (2010). Shearinduced instabilities in oil-in-water emulsions comprising partially crystallized droplets. Langmuir, 26, 16782e16790. Thivilliers, F., Drelon, N., Schmitt, V., & Leal-Calderon, F. (2006). Bicontinuous emulsion gels induced by partial coalescence: kinetics and mechanism. Europhysics Letters, 76, 332e338. Thivilliers, F., Laurichesse, E., Saadaoui, H., Leal-Calderon, F., & Schmitt, V. (2008). Thermally induced gelling of oil-in-water emulsions comprising partially crystallized droplets: the impact of interfacial crystals. Langmuir, 24, 13364e13375. Walstra, P. (1967). On crystallization habit in fat globules. Netherlands Milk and Dairy Journal, 21, 166e191. Walstra, P. (2003). Physical chemistry of foods. New York, NY, USA: Marcel Dekker, Inc.