Rheological and physicochemical evaluation of dispersion systems based on enzymatically modified animal fat

Hemp Oil (Oleofarm) with the following fatty acid profile was used: C16:0 7.8%, C18:0 2.7%, C18:1 (9-cis) 10.7%, C18:1 (9-trans) 0.8%, C18:2 (all-cis) n-6 54.8%, C18:3 (all-cis) n-6 2.7%, C18:2 (all-cis) n-3 18.7%, C20:0 1.2%, C20:1 0.5%, other 0.2%. Mutton tallow was characterized by the following fatty acid profile: C14:0 6.4%, C16:0 27.8%, C16:1 1.2%, C17:0 2.0%, C18:0 28.2%, C18:1 (9-cis) 23.3%, C18:1 (9-trans) 2.7%, C18:2 (all-cis) n-6 1.9%, C18:2 (all-cis) n-3 0.6%, other 5.8%.

Lecithin was obtained from Lasenor. Synthesized diacylglycerols (DAG) were obtained from monoacylglycerols (MAG) using acid-catalyzed solvent-assisted conditions. Detailed description of a procedure and characteristics of MAG can be found in Pazdur et al. (2015)). Lipase, from Rhizomucor Miehei immobilized on immobead 150, ≥ 300 U/g was obtained from Sigma-Aldrich.

In the present work, Turbiscan test was used to assess the stability of emulsion systems (EM1–EM10). The aim of the analysis was to determine the profiles of backscattered light (BS) intensity for the each sample. According to Huck-Iriart et al. (2011) this determination is a reliable tool for evaluation of the changes, which are taking place in the dispersion system. Furthermore, the same authors stated that the profiles of BS light intensity can be considered as a specific macroscopic fingerprint of an emulsion. Therefore, it is possible to identify and observe changes of instabilities in the sample over time (Mengual et al. 1999).

Figure 1 shows backscattering profiles in the reference mode (∆BS (%)) for emulsions EM1–EM6 with respect to emulsion EM2 taken as a reference. Analyzing the obtained profiles, it was found that in samples EM1–EM6, several mechanisms of destabilization occurred at the same time. Local variations of ΔBS in the lower and upper part of the measuring cell as well as global variations in the entire sample were observed. The appearance of a negative peak in the lower part confirms the emulsion clarification at the bottom of the sample, while the positive peak in the upper part is attributed to the formation of more dense phase due to creaming process (Robins 2000). Creaming is related to the density difference between the bulk and the dispersed phases and occurs in O/W emulsions (Ghanbari et al. 2017). In EM1–EM6 samples, additional features in the middle part of the graph were also observed, indicating an increase of the particles’ diameter of the dispersed phase (flocculation or coalescence). According to Cerimedo et al. (2010), changes in the intensity of BS light result in changes in the size of the emulsion’s average particle size (D) and on the particle volume fraction (ø), i.e., BS = f (D, ø). Thus, observed changes in particle size can be attributed to the particles’ aggregation and coalescence, leading to an increase in the average particle size of an emulsion. As a consequence of these phenomena, creaming was observed, i.e., gravitational separation.

The beginning of phase separation of these systems (EM1–EM6) was dependent on time and total amount of emulsifiers in the system. The earliest phase separation was observed for the EM1 sample on day 2 (data in Appendix). The EM1 emulsion contained the smallest amount of emulsifiers formed during interesterification. At the latest, phase separation occurred in sample EM6 (data in Appendix). Analyzing the above results, it was found that the amount of emulsifiers produced during enzymatic interesterification in a fatty base of emulsions EM1–EM6 was not sufficient to stabilize the dispersion system.

In the case of EM7 and EM8 samples, containing the highest amount of emulsifiers produced in the interesterification process, the smallest changes of the ΔBS values were observed (Fig. 2). The high stability of these systems was confirmed by nearly unchanged profiles of BS light intensity (overlapped curves in Fig. 2) during the entire storage period. The ΔBS values in the middle part of the sample varied within 2%, which also indicated that there was no significant change in the particle size of these emulsions during storage. For these systems, no changes characteristic for creaming process, nor phase separation were observed.

Figures 3 and 4 present BS profiles in the reference mode (∆BS (%)) for emulsions EM9 and EM10, respectively. For these samples, significant changes were observed in the BS light intensity profiles at the entire sample height in subsequent scans during emulsion storage. Considerable distances were observed between the curves recorded for the subsequent measurements, which indicated an increase in the average droplet size of the system with time of storage. Definitely, the largest changes in the droplet size were observed for the EM10 emulsion, which contained synthetic diacylglycerols. According to earlier reports concerning measurements of droplet size distribution in emulsion systems with similar methods (Chauvierre et al. 2004; Palazolo et al. 2005), it can be assumed that our emulsions were also characterized by a wide particle size distribution, because the variations of BS light intensity were observed at the entire height of the samples for individual measurements. The confirmation of this assumption could be the particle size analysis using laser diffraction method.

Figure 5 shows the percentage increase of average droplet size of emulsions EM1–EM10 during storage (32 days, 30 °C). In general, a smaller increase of average droplet size in emulsions containing emulsifiers formed during the interesterification process was observed. The smallest increase was detected for EM7 and EM8, 9% and 13%, respectively. The largest increase, almost double compared to the initial value, was noted for EM10, which contained synthetic DAG as emulsifiers. Also, a significant increase (49%) occurred for EM9, in which the lecithin was used as an emulsifier.

Taking into account the colorimetric evaluation, carried out 1 month after the manufacturing of emulsion systems, as well as a visual assessment, it can be concluded that the investigated emulsions were characterized by green and yellow tones (Table 2, Fig. 6). The darkest color was shown by the EM10 sample containing synthetic diacylglycerols. The remaining emulsions were much brighter compared to the EM10 emulsion, and their brightness was determined in the range (41.35–59.66). The b * parameter has been characterized for nearly all emulsions as a shade of yellow. The highest values for this shade of color were determined for emulsions EM2, EM8, EM9. In turn, for the parameter a * determining the share of green and red, negative values of this feature were noted, indicating a significant share of green in the general tone of this emulsion. Considering raw materials from which fatty mixtures were prepared, and then emulsions, it can be concluded that the green color came from hemp oil, which contains high amount of chlorophyll pigments (Teh and Birch 2013). Emulsion EM5 revealed low L * values and high b * values at the same time, which could be caused by destabilization processes. According to McClements (2007), coalescence may lead to color changes of an emulsion, due to difference in light scattering by larger droplets. Based on that, it can be concluded that emulsion color becomes deeper when coalescence occurs.

In the microscopic image of the prepared emulsions, the complexity of their structure is clearly visible. In most of them, there were differences in the size of droplets constituting a dispersed phase. The smallest droplets of a similar size were observed for the emulsion EM7 and EM8 (Fig. 7). A fairly homogeneous size of droplets was also characteristic of emulsion EM9 (Fig. 7), although the presence of single larger droplets in it affected the reduced assessment and ultimately allowed to qualify these changes as a heterogeneity of the system. It also affected the total mean droplet size of the system, which consequently had an effect on changing the type of the physical system.

Emulsions containing emulsifiers formed during interesterification with an addition of water in a range of 0.70–1.25%, i.e., EM5, EM6, EM7, and EM8, revealed the narrowest particle size distribution (Fig. 8). The diameter of the droplets after 30-day storage period was within a range of approximately 0.5–14.0 μm. Remaining emulsions with emulsifiers formed during interesterification, with water addition to the reaction system below 0.70% (EM1–EM4), were characterized by multimodal particle size distribution, which was correlated with the instabilities occurring in these samples, confirmed also by Turbiscan test. Moreover, for these emulsions, the range of droplet diameter was wider and the maximum droplet diameter was approximately 20.0–25.0 μm. Emulsion EM9, containing as an emulsifier lecithin, revealed a formation of additional new peaks (around 22 and 24 μm). Appearance of new fractions can suggest that the process of emulsion separation started. While EM10 with synthetic DAG was characterized as highly polydisperse system with multimodal particle size distribution.

Rheological properties of emulsion systems are crucial for determining their functional properties and thus affect their quality, as flow phenomena are involved both in the emulsion preparation and in their use (Barnes 1994; Gallegos et al. 2004; Brümmer 2006). Over few decades, various rheological models were constructed that can be used for fitting experimental data of diluted, semi-diluted, and dense emulsions. Both theoretical and empirical approaches were employed to correlate the emulsion viscosity, texture, and viscoelasticity with volume fraction of droplets present in emulsion systems (Schowalter et al. 1968; Palierne 1990). First report concerning the influence of droplet size distribution on emulsion rheology has been raised by Pal (1996)), who noted more pronounced shear thinning behavior of emulsions containing smaller droplets, and more importantly, a strong dependence of rheological properties on the droplet size distribution at the same volume fraction of dispersed phase. Further studies on this problem, carried by Salager et al., have shown that the lowest apparent viscosities were observed for emulsions with the highest polydispersity indexes (Salager et al. 2001). Intense studies on emulsion rheology allowed later for formulation of several constitutive models describing rheological behavior of monodisperse emulsions (Pal 2001, 2003). An important progress in rheological modeling of dense emulsion systems composed of polydisperse droplets can be found in the report of Baldyga et al. (2016), who considered the effect of droplet polydispersity on the rheological properties of emulsion viscosity by combining models proposed by Palierne (1990)) and Pal (2003)). Taking the above into account, to gain more information on rheological properties of our emulsions and compare them with other systems published in the literature, we have employed two fitting approaches. First one, involving fitting a general rheological model. The best fits were obtained with a Cross equation: \( \eta ={\eta}_{\infty }+\frac{\eta_0-{\eta}_{\infty }}{1+{\left(C\dot{\gamma}\right)}^m} \) , where η₀ is the zero shear viscosity, corresponding to the viscosity of an emulsion at extremely small shear rates (lower Newtonian plateau) (Macosko and Larson 1994). The coefficient η_∞ is the infinite shear viscosity indicating a viscous behavior at very high shear rates (upper Newtonian plateau). The exponent m, known as the Cross rate constant, reflects the steepness of the slope of a viscosity curve in the shear-thinning region; m = 0 corresponds to the Newtonian flow, while m close to 1 characterizes the shear thinning behavior of the investigated system. The parameter C is understood as a consistency (or Cross time constant). The onset of a critical shear rate, at which shear thinning behavior begins, can be estimated from reciprocal of C. Parameters of fits with Cross model, together with goodness of fit (R²) coefficients, are collected in Table 3. Secondly, another fitting with a more detailed model that involves polydispersity of emulsion droplets in calculations of viscosity, proposed by Baldyga et al. in their report (Baldyga et al. 2016). Their model (denoted as JB- model further in the text) enables calculations of emulsion viscosity with regards to several parameters characterizing an emulsion system: volume fraction of droplets, their size distribution, viscosity of the dispersed phase and viscosity of dispersing medium, interfacial tension and the shear rate. Details of the model, together with all the equations, can be found in the original work (Baldyga et al. 2016). Here, main equations being a core of JB model are given. Emulsion viscosity can be calculated as

\( \eta ={\eta}_c\left(1+5{\int}_0^{\infty }{f}_{\mathrm{v}\left(\mathrm{Ri}\right)}{\mathrm{I}}_{\left(\mathrm{Ri}\right)}\mathrm{dRi}\right) \), where η_c-viscosity of the dispersing medium, f_v(Ri)I_(Ri)dRi-is a volume fraction of droplets, which size falls in the range Ri to Ri + dRi, \( {I}_{(Ri)}=\frac{\left(4/{\mathrm{N}}_{\mathrm{Ca},\mathrm{i}}\right)\left(2+5\mathrm{K}\right)+\left(\mathrm{K}-1\right)\left(16+19\mathrm{K}\right)}{\left(40/{\mathrm{N}}_{\mathrm{Ca},\mathrm{i}}\right)\left(1+\mathrm{K}\right)+\left(2\mathrm{K}+3\right)\left(16+19\mathrm{K}\right)} \); capillary number N_{Ca, i} is calculated according to the following expression: \( {N}_{Ca,i}={R}_i{\upeta}_c\dot{\gamma}/\sigma \), where σ is the interfacial tension, and K coefficient determines the value of the viscosity ratio of the dispersed phase to the viscosity of the dispersing phase. It has to be noted, however, that description of JB model involves only Newtonian media or as continuous phases and disregards non-Newtonian liquids, which in fact are present in many emulsion systems. In our case, the dispersing phase is a shear thinning fluid containing thickening agent (dissolved carboxymethylcellulose); therefore, its shear dependent viscosity was also taken into account in fitting procedure. For fitting purposes, we have modified slightly the original description of JB model by using shear dependent viscosity measured for the continuous phase in equation. Volume fractions of the dispersed phases estimated on the basis of electrical conductivity measurements according to the method described by McClements (2005)). The obtained values for the investigated systems are listed in Table 3. Volume fractions were used for viscosity calculations based on JB model. Results of fits with both models are shown in Fig. 9. The applied fitting procedures resulted in both cases similar goodness of fit coefficients; however, the JB model seems to work better in the low shear rates range. For shear rates above 10 s⁻¹ in emulsions EM1–EM8, the JB model does not hold well in contrast to the Cross equation, which fits well the higher shear rate range. Fitting procedure performed by assumption of constant viscosity of the continuous phase gave much worse results, especially at higher shear rates (results of fitting not shown in the paper). Weaker correlation of the JB model with experimental data at higher shear rates can be attributed to enhanced hydrodynamic interactions between droplets deformed during flow and orientation of thickener macromolecules in the shear direction.

All the emulsions prepared with interesterified fat blends are characterized by relatively high values of η₀ viscosities followed by pronounced shear thinning behavior with increasing shear rate. It is noticeable that emulsions EM7 and EM8, prepared from fat blends with the highest polar fraction of acylglycerols, exhibit also higher zero shear viscosities, which clearly shows that this modification offers a convenient way for adjustment of the consistency of the prepared emulsions. Both lecithin-based material and physical mixture of fats exhibited few times larger η₀ viscosities in comparison to emulsions containing interesterified fat blends with lower content of polar fraction.

Utilization of emulsions in everyday use is closely related to their viscoelastic behavior that characterizes consistency of an emulsion at quasi-stationary state. To reveal the viscoelasticity of our emulsions and correlate their rheological behavior with the content of polar fractions in the interesterified fats used as the oil phase, we have additionally performed oscillatory tests. Figure 10 shows dependences of G′ and G″ on imposed stress amplitude in the range 0.01–100 Pa, obtained at constant angular frequency 1 rad/s for all the investigated materials.

The amplitude sweeps of (G′) and loss (G″) moduli presented in Fig. 10 revealed strong correlation between consistency and composition (in particular on the content of polar fraction in the fat blends) of the investigated materials. Emulsions containing larger content of polar fractions (EM6, EM7, EM8) exhibit more extended linear viscoelastic range (LVE) and a bit higher values of G′ modulus in comparison to emulsions EM1–EM5. Moreover, for emulsions EM7 and E8, G′ modulus is approximately two times higher in comparison to the other emulsions containing interesterified fats, although for both of them, these values are far from those obtained for the reference emulsion EM9 stabilized by lecithin. On the other hand, G′ modulus of emulsion EM10 prepared with synthetic DAG as a stabilizer is very high, although the LVE range is pretty narrow, which suggests moderate stability of this system caused, i.e., by creaming.

Dense emulsions are considered as a deformation-dependent systems characterized by visible dependence of dynamic moduli on the applied frequency and smaller differences between corresponding G′ and G″ values. To study these relations, we have performed the corresponding frequency tests and their results are shown in Fig. 11. The frequency sweeps show clearly that emulsions containing interesterified fats (EM1–EM8) are more frequency dependent than EM9 and EM10, although the observed changes are rather moderate, as illustrated by relatively small slope of G′ at low-frequency regime. This can be considered as evidence of enhanced stability of the investigated materials in the prolonged time period. However, closer look to the low-frequency regimes, one can see that the G′ modulus of systems with lower amounts of polar fractions (EM1–EM5) appears to be more frequency dependent, than for emulsions EM6–EM8 that contained higher amounts of polar fractions in the oil phase. Small strain/stress oscillatory tests performed in wide frequency range enable analysis of materials in a time domain. From analysis of G′ dependency on frequency at various regimes, one can estimate the time-driven changes of the investigated system, provided that the whole investigation is performed at proper conditions, namely in isothermal conditions under constant strain or stress taken from LVE range.

To estimate the time-dependent structure reformation after shearing, we have performed a three-interval test combining observation of viscosity change before and after applying the high shear rates. The measurements were conducted in three steps. In every step, constant shear rate was applied and the viscosity change in time was recorded. First, the material was equilibrated in the measuring cell over 10 min; then, the reference viscosity was measured at shear rate equal to 0.25 s⁻¹ by collecting few points within 15 s. Next, high shear rate (500 s⁻¹) was applied for 1 s to disturb the emulsion microstructure; after high shearing period, the shear rate 0.25 s⁻¹ was restored and viscosity was measured within longer period of time. As a result, the information about the percentage of microstructure recovery related to the reference viscosity value is obtained, which enables estimation of emulsion microstructure reconstruction after shearing and can be used for testing the relative changes in consistency. The estimated percentages of microstructure recovery after 1 s and after 60 s are shown in Table 4.

For emulsion EM10 prepared with synthetic DAG, the viscosity overshoot was observed revealing the microstructure reconstruction higher than 100% (Fig. 12). This phenomenon is quite often observed in materials capable for formation of stable microstructures resulting from strong interactions between droplets and dispersing media. In such systems, the application of high stresses results in partial microstructure destruction and in the sample volume besides individual droplets larger aggregates appear that may block the flow of the material causing temporary thickening of the measured fluid, which is visible as higher viscosity. With the progress of time of the measurement at constant low shear rate, these aggregates reform and finally the system reaches equilibrium with the viscosity value close to the reference one.

Typical cosmetic emulsions are usually prepared, stored, transported, and utilized at temperatures between 0 and 50 °C; therefore, to study the thermal breakdown of our systems, we have investigated our materials in a rotational mode in this temperature regime. The results are depicted in Fig. 13.

Emulsions containing interesterified fat blends show similar trends in their thermal behavior with visible signs of emulsion destabilization, evidenced by slight increase in viscosity around this temperature followed by its drastic drop above 33 °C. Materials prepared with lecithin and synthetic DAG show similar behavior, although the initial increase in viscosity is less pronounced.