Investigation of the formation mechanisms in high internal phase Pickering emulsions stabilized by cellulose nanocrystals

Abstract

Medium and high internal phase Pickering emulsions stabilized by cellulose nanocrystals (CNCs) have been prepared and the effects of CNC concentration and type of oil phase on the properties of emulsions were studied. The maximum oil phase volume that can be stabilized by CNCs is 87% when the CNC concentration is 0.6 wt.%; this slightly decreases to 83% when the CNC concentration is increased to 1.2 wt.% or higher. In addition, the oil droplets stabilized with 0.6 wt.% CNC suspensions have a larger size than those stabilized with higher concentration CNC suspensions. As evidenced by the change in oil droplet morphology and size, two different emulsion formation mechanisms are proposed. For a CNC concentration of 0.6 wt.%, the extra oil added into the emulsion is accommodated by the expansion of oil droplet size, whereas for CNC concentrations of 1.2 wt.% and higher, the oil is stabilized mainly by the formation of new oil droplets.

1. Introduction

Emulsions are heterogeneous systems normally consisting of two immiscible liquids, one of which is dispersed as droplets in the other one. Traditionally, the liquid droplets, or the dispersed phase, are stabilized by small molecule surfactants or surface-active polymers, which can be adsorbed onto the interface—thereby reducing the interfacial free energy. Emulsions can also be stabilized solely by fine solid particles to form solid-stabilized or Pickering emulsions [1,2]. The formation of oil-in-water (o/w) or water-in-oil (w/o) emulsions depends on the three-phase contact angle at the interface, which affects the bending of the fluid–fluid interface. If the three-phase contact angle is less than 90°, the particles will cause the fluid–fluid interface to bend towards the oil phase, leading to the formation of o/w emulsions; otherwise w/o emulsions will be formed [3]. If the particles are completely wetted by one of the phases, they may remain dispersed in that phase and no stable emulsion will form [4]. The stability of Pickering emulsions is related to the particle concentration and particle–particle interactions. At high particle concentrations, the irreversibly adsorbed particles at the interface and the formation of a three-dimensional particle network can prevent the coalescence of dispersed droplets [5,6]. At low particle concentration, however, the limited coalescence and bridging of droplets by a monolayer of particles become the major stabilization mechanism [6,7]. The emulsion stability can be dramatically improved when the stabilizing particles are weakly flocculated [8].

A wide variety of solid particles have been investigated to prepare either o/w or w/o Pickering emulsions, including inorganic particles such as silica [9], silicate [10], montmorillonite [11], calcium carbonate [12], carbon graphite [13] and polymer particles—for instance, polystyrene [14], poly(vinylidene difluoride) [15] and poly(tetrafluoroethylene) [15]. Since the first report of microcrystalline cellulose (MCC)-stabilized emulsions [16], cellulosic materials have been attracting more research interest, mainly due to their renewability and non-toxicity, which are beneficial for food, health and cosmetic applications where the use of surfactants can be undesirable. Cellulosic particles can be derived from various biomass sources and exhibit very different morphology depending on the source and the processing methods. Besides MCC, other cellulosic forms have been used as stabilizers in Pickering emulsions, including microfibrillated cellulose (MFC) [17,18], cellulose nanofibrils (CNFs) [19], bacterial cellulose [20] and cellulose nanocrystals (CNCs) [21–23].

Pristine cellulosic materials form o/w emulsions owing to their hydrophilic nature. Different surface treatments, such as silylation [19] and esterification [24], have been investigated to render cellulose hydrophobicity in order to prepare w/o emulsions or o/w/o (oil–water–oil) double emulsions [25]. Moreover, stimuli-responsive molecules have also been grafted onto cellulose surfaces to produce pH-responsive [26] or temperature-responsive [27] emulsions. Kalashnikova et al. [28] prepared o/w emulsions using CNCs derived from different sources with aspect ratios ranging from 13 to 160, and explored the stabilizing effects of non-sulfated CNCs. The electrostatic repulsion between the negatively charged sulfate groups on the surface of CNCs has been shown to affect the stability of emulsions, and the ionic strength of the aqueous phase typically controls these interactions [29]. Kalashnikova et al. [28,29] found that CNCs of small particle size tend to form a dense layer at the interface with a higher coverage (greater than 80%), whereas long nanocrystals are more likely to form an interconnected network on the oil droplet surfaces with a lower coverage (ca. 40%) and form bridges between droplets.

High internal phase emulsions (HIPEs) are emulsion systems containing an internal, or dispersed, phase volume fraction greater than 74%, which is the maximum volume ratio of monodispersed non-deformable spheres when packed in the most efficient manner. Emulsions with internal phases of 50–74% and lower than 50% volume fractions are normally called medium internal phase emulsions (MIPEs) and low internal phase emulsions (LIPEs), respectively. Recently, cellulose-based HIPEs have been reported. Lee et al. [24] prepared w/o HIPEs using hydrophobic bacterial cellulose nanofibres (BCNs) as the stabilizer, which possessed an entangled network structure. Stable and uniform emulsions did not form after direct mixing of the two phases, even though the internal phase volume was only 50%. After 7 days of storage, however, the emulsions became stable and the maximum internal phase volume could increase up to 81%, which is in the HIPEs range. The authors ascribed this phenomenon to the disentanglement of BCN networks into small bundles and even individualized nanofibres. Capron & Cathala [21] produced o/w HIPEs using CNCs as the stabilizer and hexadecane as the oil phase through a special two-step procedure. In their work, an original low internal phase emulsion (10% oil) was prepared using an ultrasonicator. Extra oil was then added followed by high-speed homogenization to obtain HIPEs. By examining the evolution of oil droplet morphology through such a process, the authors proposed a formation mechanism of the HIPEs based on the critical coverage of dispersed droplets with CNCs. According to Capron & Cathala [21], the diameter of small oil droplets expanded following a limited coalescence of the droplets to maintain the minimum required coverage of CNCs to accommodate the added extra oil. However, the authors only examined one CNC concentration (0.5 wt.%) in their work.

In this paper, we have carried out an in-depth investigation of o/w HIPEs stabilized with CNCs. The effects of CNC concentration and type of oil phase on the formation of HIPEs and the properties of the emulsions have been systematically studied. The purpose of this work is to further reveal the formation mechanisms of HIPEs stabilized with CNCs by expanding Capron & Cathala's [21] work in order to offer a more comprehensive view.

2. Experimental set-up

(a). Materials

The CNCs were supplied by CelluForce Inc. in dry powder form. H₂SO₄-hydrolysed, spray-dried CNCs were dispersed in deionized (DI) water at 2 wt.% by stirring overnight using a mechanical mixer, followed by ultrasonication treatment for 10 min (Sonics, model VC 750, amplitude: 80%). The resulting CNC suspension was then purified by filtering the suspension through no. 4 and no. 42 Whatman filter papers sequentially. The CNCs prepared by acid hydrolysis of kraft wood pulp typically have dimensions of approximately 5 nm in cross direction and approximately 100 nm in length and have approximately five sulfate groups per 100 glucose units [30], as shown in figure 1. The ionic strength of the CNC suspension was controlled by adding 50 mM of NaCl (Fisher Scientific). The oil phases tested in this work were light mineral oil and hexane (99.9%), both of which were purchased from Sigma-Aldrich and used as received.

Figure 1.

A scanning transmission electron microscopy (STEM) image of CNCs.

Congo red (Sigma-Aldrich) was used to stain some CNCs in order to observe their distribution within the emulsions. The staining was carried out by first mixing 200 g of 1.2 wt.% CNC suspension containing Congo red aqueous solution (Congo red versus CNC = 1 : 100 w/w) using an overhead stirrer. The flask with suspension was then submerged in an oil bath heated to 60°C for 24 h. The Congo red-stained CNC suspension was centrifuged at 4000 r.p.m. for 20 min at 25°C (Beckman J6-HC). The supernatant was decanted and the Congo red-stained CNCs were resuspended in 500 ml of 1 wt.% NaCl solution followed by centrifugation. This process was repeated twice and the supernatant from the last centrifugation was totally clear, suggesting the absence of free Congo red dye. The suspension was then dialysed against running DI water for 5 days in dialysis tubing (MWCO 12, −14 000 Da, Fisher Brand). The suspension, which appeared dark red (nearly purple), was finally concentrated to 1.2 wt.% using a rotary evaporator (Buchi Rotavapor R-200).

(b). Emulsion preparation

High internal phase Pickering emulsions were prepared using a homogenizer (Silverson L4RT-A) equipped with a square-hole small mixing head via a two-step procedure. In this method, a fixed amount of CNC suspension, 20 ml, was added into a 250 ml beaker, followed by direct addition of the oil phase (light mineral oil or hexane). The CNC concentrations investigated in this study were 0.6, 1.2 and 2.4 wt.% in the water phase for all w/o ratios used. The internal phase content was controlled by changing the volume of the oil added, ranging from 37 ml (65%) to 135 ml (87%) for light mineral oil and from 30 ml (6 0%) to 45 ml (69%) for hexane. The mixtures formed two clear separate layers naturally, as shown in figure 2a. The mixing head of the homogenizer was lowered to the bottom of the mixture and operated at 2000 r.p.m. for 1 min, during which only a small portion of the oil was mixed with the CNC suspension to form a low internal phase emulsion (first step, figure 2b). The homogenization speed was then increased to 10 000 r.p.m. for 1 min, and the mixture formed a gel-like emulsion instantly as shown in figure 2c and d (second step). Owing to the high viscosity of the emulsion, the beaker was manually moved around the mixing head during the high-speed homogenization process to ensure homogeneous mixing of the emulsion.

Figure 2.

A two-step process is used to prepare high internal phase Pickering emulsions. (a) Before mixing; (b) during low-speed mixing (2000 r.p.m. for 1 min); (c) after high-speed mixing (10 000 r.p.m. for 1 min); and (d) stable gel-like appearance of the final HIPE. (Online version in colour.)

(c). Characterization

The relative stability of the emulsions was evaluated by centrifugation. In this method, 20 g of emulsion was placed in a 50 ml graduated centrifuge tube. The samples were then centrifuged at 4000 r.p.m. for 10 min using an Eppendorf 5810R centrifuge. Using such a process, a part of the emulsion might break, whereby the oil and water phases travel to the top and bottom, respectively, as shown in the electronic supplementary material, figure S1. The stability of the emulsion was indicated by the volume percentage of the remaining emulsion phase of the total volume.

The oil droplet size in emulsions was measured with an optical microscope (Nikon Microphot-FX) equipped with a charge-coupled device digital camera (SPOT RT Slider; Diagnostic Instruments Inc.) and using a black-and-white model; the images were analysed using ImagePro software. The specimens for microscope observation were prepared by gently smearing a small amount of the creamy emulsion on a glass slide and placing a glass cover on top. In this process, no pressure was applied on the glass cover to avoid influencing the oil droplets in the emulsions. All images were taken on the same day as the emulsion was prepared. The images of emulsions containing Congo red-stained CNCs were taken using the same method, but using a colour model. To analyse the droplet size, an area was randomly selected on the image of each sample, whereby around 50 droplets were included. All the droplets in the selected area were measured along the same direction to exclude any possible bias caused by sample population and/or irregular droplet shape. An example of the droplet size analysis is demonstrated in the electronic supplementary material, figure S2. For samples with large droplets, when the number of droplets was less than 50 in any one image, multiple images were used to keep the total number of analysed droplets the same. The reported results are the number-averaged diameter and standard deviation.

3. Results

Two different types of oil were examined, light mineral oil and hexane. The oil content in the emulsions was calculated on volume basis, and the maximum oil content that could be stabilized at each CNC concentration was investigated by increasing the oil phase ratio in the system until a point where no stable emulsion could form. The results are illustrated as phase diagrams in figure 3. Since the purpose of this work was to prepare medium and high internal phase Pickering emulsions, the lowest oil ratio tested in this work was 65% and 60%, respectively, for the mineral oil/water emulsions and hexane/water emulsions. For both 2.4 and 1.2 wt.% CNC suspensions, the maximum oil content that could be stabilized by CNCs was 83%. However, the oil content increased to 87% when the CNC concentration was further lowered to 0.6 wt.%. When hexane was used as the oil phase, however, the maximum oil content was only 69%, yielding a MIPE, for all three tested CNC concentrations. This phenomenon can be ascribed to the difference in physical properties between light mineral oil and hexane, whose density and viscosity are 0.838 g ml⁻¹ and 23.7 cP (at 25°C) and 0.669 g ml⁻¹ and 0.3 cP (at 25°C), respectively. Since hexane has a lower density and viscosity than light mineral oil, it has a higher tendency to escape from the droplets in the emulsion, thus leading to a less stable emulsion at higher internal phase conditions.

Figure 3.

Phase diagrams of CNC-stabilized mineral oil (a) and hexane (b) based o/w Pickering emulsions. (Online version in colour.)

The prepared emulsions were observed using optical microscopy and figure 4 provides examples of emulsions with the two types of oil phase. Both emulsions demonstrated near spherical-shaped droplets and the oil droplets were packed very closely. Figure 5 shows a series of emulsions stabilized by the 0.6 wt.% CNC suspension at different mineral oil contents. The oil droplets were basically spherical when the oil content was lower than 83%. With increasing oil volume, however, the size of the oil droplets increased and the droplets started to deform into a polygonal shape.

Figure 4.

Optical microscopy images of (a) CNC–mineral oil and (b) CNC–hexane Pickering emulsions. The oil phase is 65% of total volume and the CNC concentration is 1.2 wt.% for both samples.

Figure 5.

Mineral oil/water Pickering emulsions stabilized by 0.6 wt.% CNC suspensions at (a) 65%, (b) 78%, (c) 83% and (d) 87% oil content.

The oil droplet sizes of both types of emulsions at various oil contents were analysed using image analysis and the results are plotted in figure 6. For the CNC–mineral oil system (figure 6a), the oil droplet sizes appear to be similar at 65% oil content regardless of CNC concentration. As the oil content increases, the droplet sizes remain practically unchanged for emulsions stabilized with 1.2 wt.% and 2.4 wt.% CNCs. This can be explained in terms of the coverage of CNCs at the oil droplet surface. At 1.2 wt.% and 2.4 wt.% CNC concentrations, CNC spindles form a dense network around each oil droplet, which restrains the expansion and deformation of the oil droplets when the oil content increases. Therefore, with increasing oil content, the oil droplet size in emulsions stabilized with 1.2 and 2.4 wt.% CNC suspensions remains unchanged, as do the oil droplet size distribution and their spherical shapes. When the CNC concentration is decreased to 0.6 wt.%, however, the surface coverage is reduced and a less dense CNC network surrounds the oil droplets, allowing them to expand to accommodate the extra oil. As a consequence, the oil droplet size of this emulsion increased gradually from 14 µm at 65% oil volume to approximately 30 µm at 87% oil volume. However, no stable emulsion could be formed at CNC concentrations of 1.2 and 2.4 wt.% for an 87% oil volume. For CNC–hexane emulsions (figure 6b), the oil droplet sizes of all emulsions were all similar at any specific oil content regardless of CNC concentration. This is due to the fact that the CNC–hexane combination could only form stable emulsions within a narrow range of oil content, namely 60–69%. The droplet size is similar to that observed in CNC–mineral oil emulsions at 65% of oil content, which can be ascribed to the saturated surface coverage.

Figure 6.

Mean oil droplet size of (a) CNC–mineral oil and (b) CNC–hexane emulsions. The raw data are available in the electronic supplementary material. (Online version in colour.)

All of the Pickering emulsions prepared in this work remained stable without any phase separation over a 2 year period when stored at ambient conditions. This high stability can be ascribed to the irreversible adsorption of a layer of the CNC network at the oil/water interface, as well as the compact structure of oil droplets, which restricts their motion. In order to compare the relative stability of the emulsions, a centrifugation method was used, which is effective for mimicking stability under dynamic conditions. In this test, the freshly prepared emulsions were centrifuged at 4000 r.p.m. for 10 min, and the stability was evaluated by comparing the volume percentage of the unseparated emulsion phase. As shown in figure 7, for CNC–mineral oil emulsions, when the oil content is lower than 83%, all samples retain approximately 70–95% of the emulsion phase after centrifugation. At any specific oil content, the stability increased with increasing CNC concentration, which can be ascribed to the increasing coverage of CNCs at the oil droplet surfaces. Moreover, for the emulsions stabilized with the same CNC concentration, the stability increased with the increasing oil content—e.g. at [CNC] = 0.6 wt.%, the stability increases from 70% to 80% with increasing oil content from 65% to 83% (figure 7). We speculate that this is likely to be caused by the more compact packing of oil droplets in the emulsions. As discussed above, if the oil content is further increased to 87%, stable emulsions could only be formed at [CNC] = 0.6 wt.%. This emulsion, with its relatively large oil droplets, has a lower centrifugal stability than others, approximately 50%. This behaviour can be explained by the reduced surface coverage of the oil droplet by CNCs. Stability trends for CNC–hexane emulsions were similar to those discussed for CNC–mineral oil (figure 7)—i.e. stability increased with CNC concentration at any specific oil content. However, the overall stability of CNC–hexane emulsions was lower than that of the CNC–mineral system owing to the lower density and viscosity of hexane.

Figure 7.

Stability comparisons between (a) CNC–mineral oil and (b) CNC–hexane emulsions (based on centrifugation at 4000 r.p.m. for 10 min). (Online version in colour.)

It is noteworthy to emphasize that CNC-stabilized o/w high internal phase Pickering emulsions can only be formed through a two-step process, and direct mixing of the two phases results in phase separation. Using our protocol, a low internal phase o/w Pickering emulsion is, in effect, formed at the first low-speed homogenization step, where CNCs are irreversibly adsorbed at the oil droplet surface to form a network. The additional oil added at the second step is introduced into the oil phase by entering the existing oil droplets and/or forming new oil droplets. To investigate the evolution of the oil phase, emulsions stabilized with Congo red-stained CNCs (0.6 wt.% and 1.2 wt.%) were prepared and examined using optical microscopy (figure 8). Stained CNCs appear red to purple depending on their concentration, whereas the oil phase appears colourless, and the CNCs are dispersed in the water phase. LIPEs, which are free to flow, are formed during the first homogenization step (2000 r.p.m. for 1 min) (figure 8a,b). These emulsions are characterized with oil droplets having a wide size distribution, from a few micrometres to hundreds of micrometres, regardless of CNC concentration. The oil droplets in these emulsions are stabilized by the stained CNCs adsorbed at the o/w interface, which appear as purple coronas surrounding the oil droplets in the optical images of figure 8a,b. It is worth noting that, although the oil droplets in both emulsions have similar morphologies, the colour of the water phase from the emulsion stabilized with 1.2 wt.% CNCs is darker than that with 0.6% CNCs. That is to say, at this stage of emulsion evolution (LIPEs), CNCs are not completely consumed in stabilizing the oil droplets, and free CNCs are present in the water phase. Following the second step, high-speed homogenization (10 000 r.p.m. for 1 min), the structure of the emulsion drastically changes for both emulsions with 0.6 and 1.2 wt.% CNCs. The free (stained) CNCs disappear from the water phase and are only visible as thin purple coronas surrounding the oil droplets (figure 8c,d), and the oil droplets changed to a more compact structure with a much narrower size distribution than the LIPE stage. Moreover, the oil droplet size in the emulsion with 0.6 wt.% CNCs is slightly larger than the one with 1.2 wt.% CNCs, which agrees with the size analysis results (figure 6a). Figure 8d shows the formation of a HIPE.

Figure 8.

Optical microscopy images of mineral o/w emulsions stabilized with Congo red-stained CNCs at 78% oil content. These images depict the evolution of CNC stabilization from LIPEs to HIPEs. (a) A LIPE formed using 0.6 wt.% CNCs and 2000 r.p.m. homogenization for 1 min; (b) a LIPE formed using 1.2 wt.% CNCs and 2000 r.p.m. homogenization for 1 min; (c) a HIPE formed from the 0.6 wt.% CNC LIPE following a second homogenization using 10 000 r.p.m. for 1 min; and (d) a HIPE formed from the 1.2 wt.% CNC LIPE using the same method as above.

4. Discussion

Capron & Cathala [21] reported a similar two-step procedure to prepare HIPEs using hexadecane as the oil phase and 0.5 wt.% CNC suspensions, hydrolysed from cotton using sulfuric acid, as the water phase. In their study, a low internal phase emulsion (10% oil content) was first prepared using an ultrasonicator. Then extra oil was added into the already formed emulsion, followed by shearing with a double cylinder-type homogenizer. The resulting final emulsions had a gel-like appearance similar to ours. In the emulsion formation mechanism they proposed, all CNC spindles were adsorbed at the oil droplet surface at the first step, which was supported by the sugar profile analysis of the water phase. However, from the detailed colour microscopy images of our emulsions (figure 8a,b), it is clear that free CNC spindles are present in the water phase after the first-step homogenization. This was particularly evident for the 1.2 wt.% CNC case. This can be explained from the different mixing techniques used at the first step, where an ultrasonicator and a homogenizer were employed in Capron and Cathala's study and our experiments, respectively. The ultrasonicator supplied a stronger shearing force than our first-step homogenizer (2000 r.p.m. for 1 min); consequently, the oil droplets in Capron and Cathala's case had a much smaller size than the ones from the low-speed homogenization, 4 µm versus 130 µm diameter. Since the smaller oil droplets possess larger specific surface area than the larger ones, more CNCs could be adsorbed at the oil droplet surface in Capron and Cathala's case. Despite the significant difference in morphology at the first mixing step in these two studies, the oil droplets in the final HIPEs were very similar. For example, in our emulsion with 87% oil volume content and 0.6 wt.% CNCs (figure 5d) and Capron and Cathala's emulsion with 86% internal phase and 0.5 wt.% CNCs (fig. 2c in their paper), the oil droplets exhibited deformed polygonal shapes in both emulsions, suggesting that the oil droplets were stabilized on the basis of a similar mechanism.

According to Capron and Cathala's emulsion formation mechanism, during the homogenization process at the second step, the initial oil droplets expanded to accommodate the extra oil, which resulted in a decrease in surface coverage. When the surface coverage was not high enough to stabilize the emulsion, a limited coalescence process had to occur, through which the oil droplets could coalesce to reduce the surface area to accommodate more oil by re-arrangement of CNC particles at the oil droplet surface [7]. The oil droplets would finally deform to accommodate optimum CNC distribution. We postulate that this is likely to be the case for the small oil droplets observed in our emulsions shown in figure 8a,b. Based on our observations, we further propose that there are two other processes coinciding with CNC re-arrangement: (i) the large oil droplets are broken down to small droplets by the shearing force from high-speed homogenization at the second step and (ii) the free CNCs in the water phase can form new oil droplets and/or migrate to existing oil droplets. Consequently, the free CNCs would be completely consumed during the second homogenization step and the re-arrangement of CNCs would occur afterwards, which is supported by the microscopy images in figure 8c,d, where stained CNCs were only observed at the oil droplet surface but not in the continuous phase.

Capron & Cathala [21] tested only one CNC concentration, 0.5 wt.%. However, we discovered that at higher CNC concentrations, 1.2 wt.% and 2.4 wt.%, different trends were observed. First, the oil droplet size in the emulsions stabilized by CNC suspensions at these two concentrations remained basically constant as oil content was changed (figure 6). Second, the maximum oil contents that could be stabilized by 1.2 wt.% and 2.4 wt.% CNC suspensions were both 83%, lower than that stabilized with a 0.6 wt.% CNC suspension. Third, and finally, the oil droplets remained spherical in shape in all emulsions stabilized by 1.2 wt.% and 2.4 wt.% CNC suspensions regardless of the oil volume content. All these results indicated that the formation of Pickering HIPEs at higher CNC concentrations was following a mechanism different from the one proposed by Capron & Cathala [21].

Based on the evidence presented in figure 8b, there was a large amount of free CNCs in the water phase after the first homogenization step for emulsions containing 1.2 wt.% CNCs because the CNCs had reached saturation at the oil droplet surfaces. When such a system was further homogenized with extra oil (second step), the formation of new oil droplets became dominant rather than expansion of existing oil droplets. Since the oil surface coverage was high due to CNC abundance, coalescence was unlikely to happen. Therefore, the oil droplets had a similar size in all of the emulsions stabilized with 1.2 wt.% and 2.4 wt.% CNC suspensions (figure 6a). The maximum oil content for 1.2 wt.% and 2.4 wt.% CNC suspensions was 83%, where the oil droplets in the emulsions reach the maximum packing density, and further increase in oil content breaks up the balance leading to collapse of the emulsion. This mechanism, together with the one discussed for 0.6 wt.% CNC suspensions, is schematically illustrated in figure 9.

Figure 9.

Schematic of the formation mechanisms of CNC-stabilized oil/water HIPEs at different CNC concentrations in the water phase. (Online version in colour.)

It is important to note that, when the oil phase was switched to hexane, the formed emulsions had a lower maximum oil content and stability than those composed of mineral oil. This is because it is more difficult to stabilize hexane than mineral oil in Pickering emulsions owing to the fact that both the density and viscosity of hexane are lower than those for the mineral oil used in this work. This is also the reason why the CNC–hexane combination can only form medium internal phase emulsions that are less stable than the CNC–mineral oil ones.

5. Conclusion

Medium and high internal phase Pickering emulsions stabilized with CNCs were prepared through a scalable, two-step process. We used homogenization at low shear (approx. 2000 r.p.m. for 1 min) at first, followed by high shear (10 000 r.p.m. for 1 min). The effect of CNC concentration in the water phase on the properties of the emulsions, the maximum oil content, the oil droplet size and the emulsion stability, were studied in depth. It was found that the concentration of CNCs in the water phase plays a very important role on the formation of HIPEs, especially during the second step of the homogenization process. On the basis of our experimental results, we proposed a new comprehensive CNC-stabilized HIPE formation mechanism. At 0.6 wt.% concentration, CNC spindles stabilize an increasing volume of oil, mainly by expanding the size of the oil droplets through the limited coalescence process. Therefore, the emulsions have larger oil droplet size, higher maximum oil content and relatively lower stability under centrifugation. When the CNC concentration is increased to 1.2 wt.%, the formation of new oil droplets becomes dominant. As a consequence, these emulsions have a smaller oil droplet size, lower maximum oil content and higher stability under centrifugation. Under ambient storage conditions, all emulsions have remained stable over a 2 year period. The property of the oil phase also has a significant effect on the formation and stability of Pickering emulsions. Owing to the characteristic rheological properties of CNC-stabilized HIPEs and the non-toxic nature of CNCs, this type of emulsion has potential applications in a variety of industries, such as food, cosmetic, pharmaceutical and oil/gas drilling fluids.

Data accessibility

Additional data can be found in the electronic supplementary material.

Competing interests

We declare we have no competing interests.

Funding

The authors gratefully acknowledge financial support from Natural Resources Canada under the Transformative Technologies Program, and NSERC.