K. L.
Thompson
*,
C. J.
Mable
,
A.
Cockram
,
N. J.
Warren
,
V. J.
Cunningham
,
E. R.
Jones
,
R.
Verber
and
S. P.
Armes
*
Department of Chemistry, The University of Sheffield, Dainton Building, Brook Hill, Sheffield, South Yorkshire S3 7HF, UK. E-mail: s.p.armes@sheffield.ac.uk
First published on 12th September 2014
RAFT-mediated polymerisation-induced self-assembly (PISA) is used to prepare six types of amphiphilic block copolymer nanoparticles which were subsequently evaluated as putative Pickering emulsifiers for the stabilisation of n-dodecane-in-water emulsions. It was found that linear poly(glycerol monomethacrylate)–poly(2-hydroxypropyl methacrylate) (PGMA–PHPMA) diblock copolymer spheres and worms do not survive the high shear homogenisation conditions used for emulsification. Stable emulsions are obtained, but the copolymer acts as a polymeric surfactant; individual chains rather than particles are adsorbed at the oil–water interface. Particle dissociation during emulsification is attributed to the weakly hydrophobic character of the PHPMA block. Covalent stabilisation of these copolymer spheres or worms can be readily achieved by addition of ethylene glycol dimethacrylate (EGDMA) during the PISA synthesis. TEM studies confirm that the resulting cross-linked spherical or worm-like nanoparticles survive emulsification and produce genuine Pickering emulsions. Alternatively, stabilisation can be achieved by either replacing or supplementing the PHPMA block with the more hydrophobic poly(benzyl methacrylate) (PBzMA). The resulting linear spheres or worms also survive emulsification and produce stable n-dodecane-in-water Pickering emulsions. The intrinsic advantages of anisotropic worms over isotropic spheres for the preparation of Pickering emulsions are highlighted. The former particles are more strongly adsorbed at similar efficiencies compared to spheres and also enable smaller oil droplets to be produced for a given copolymer concentration. The scalable nature of PISA formulations augurs well for potential applications of anisotropic block copolymer nanoparticles as Pickering emulsifiers.
Much of the growing literature on Pickering emulsions has focused on spherical particles. There are far fewer reports of the use of non-spherical Pickering emulsifiers based on rigid rods or flexible worms. Noble et al.15 prepared so-called ‘hairy’ colloidosomes using relatively large (10–70 μm) ‘microrods’ prepared from a photocurable epoxy resin as the droplet stabiliser. In this case the hot aqueous phase contained 1.5 wt% agarose that gelled upon cooling, trapping the ‘microrods’ at the interface. Vermant and co-workers16 used a multiple back-scattering technique to demonstrate that more stable Pickering emulsions are obtained when employing elongated polystyrene latexes with relatively high aspect ratios (with the original isotropic latex particles being used as a control). Similar results were also obtained with ellipsoidal hematite particles.16 Guevara et al.17 reported that 2D nano-sheets can produce highly stable foams. Cellulose fibres have also been used to stabilise both water-in-oil and oil-in-water emulsions.18–20 In this case the fibre aspect ratio strongly influenced the droplet surface coverage: short fibres led to densely-coated oil droplets (>80% coverage), whereas using longer fibres produced significantly lower surface coverages (∼40%).20
Recently, our group has reported the facile preparation of a range of diblock copolymer nanoparticles via polymerisation-induced self-assembly (PISA) using reversible addition–fragmentation chain transfer (RAFT) polymerisation in concentrated aqueous solution.21,22 In the prototypical formulation, a water-soluble poly(glycerol monomethacrylate) (PGMA) chain transfer agent is used to grow a water-insoluble poly(2-hydroxypropyl methacrylate) (PHPMA) block, leading to the in situ formation of PGMA–PHPMA diblock copolymer nanoparticles via RAFT aqueous dispersion polymerisation.23–25 In principle, the final particle morphology is simply dictated by the relative volume fractions of each block, although in practice the copolymer concentration can also play an important role.26 This versatile route to block copolymer nanoparticles enables the efficient synthesis of spheres, worms or vesicles directly in water at up to 25% solids. Moreover, the surface properties of these nanoparticles can be tuned by judicious selection of the stabiliser for a given core-forming block,27–31 which is likely to be of considerable interest when designing new bespoke Pickering emulsifiers.
Recently, we compared the Pickering emulsifier performance of linear and crosslinked PGMA–PHPMA block copolymer vesicles (also known as ‘polymersomes’) prepared using a PISA formulation.32 One important finding was that use of an ethylene glycol dimethacrylate (EGDMA) cross-linker was essential to preserve the relatively delicate vesicular morphology during the high-shear homogenisation required to generate the oil droplets. In the absence of any EGDMA, the linear vesicles simply dissociated during homogenisation and the resulting emulsion droplets became stabilised by individual copolymer chains, rather than vesicles. In contrast, EGDMA-crosslinked vesicles survived the emulsification conditions and were able to function as the desired ‘Pickering’ emulsifier. In the present work, we explore whether covalent cross-linking is also essential to prevent dissociation of PGMA-based block copolymer spheres and worms when employed as putative Pickering emulsifiers. In particular, we emphasise that the block copolymer worms described herein are much more readily accessible on a multi-gram scale than the various anisotropic nanoparticles that have been explored to date.15,16,18–20 Provided that conformal contact with the interface can be achieved, such worms are expected to be much more strongly adsorbed at the oil–water interface than the equivalent spherical nanoparticles. For example, if worm-like nanoparticles are sufficiently anisotropic (i.e. if their mean contour length L is much greater than the mean worm radius R), then their specific surface area, As, can be approximated by the equation: As ∼ 2/ρR, where ρ is the particle density and R is the mean worm radius (see ESI† for further details). In contrast, the spherical nanoparticles that undergo 1D fusion to form such worms during the PISA synthesis24,27 have a specific surface area given by As = 3/ρr, where r is the mean sphere radius and r ∼ R. Thus the reduction in specific surface area that occurs on fusing multiple spheres to form each individual worm is only around 33%. To a good first approximation, the energy of attachment of a worm of radius R comprising x fused spheres at the oil–water interface is expected to be x times higher than the energy of attachment of an individual spherical nanoparticle of radius r. Thus the worm morphology affords a specific surface area comparable to that of the original spheres, but the former particles are much more strongly adsorbed at the oil–water interface. Hence the present study is focused on addressing the following fundamental question: do worm-like nanoparticles offer significant advantages as Pickering emulsifiers over their spherical nanoparticle precursors?
For the sake of brevity, we introduce a shorthand notation to describe the various diblock and triblock copolymers synthesised in this study such that G, H, B and E stand for glycerol monomethacrylate, 2-hydroxypropyl methacrylate, benzyl methacrylate and ethylene glycol dimethacrylate, respectively. For example, Gx–Hy–Ez indicates a poly(glycerol monomethacrylate-block-2-hydroxypropyl methacrylate-block-ethylene glycol dimethacrylate) triblock copolymer, where x, y and z indicate the mean degrees of polymerisation (DP) of each block.
In the present work, we wished to investigate whether the problem of shear-induced particle disintegration was restricted to vesicles or also applied to spherical nanoparticles and worm-like micelles. It has been previously shown that linear PGMA–PHPMA vesicles are relatively delicate: they undergo partial collapse under ultrahigh vacuum conditions27 and do not tolerate the addition of ionic surfactants.33 Thus it was conceivable that the vesicular copolymer morphology alone was the primary reason for the disintegration observed during high-shear emulsification. On the other hand, it is also known that the PHPMA block is only weakly hydrophobic. For example, variable temperature 1H NMR studies indicate that significant plasticisation of the core-forming PHPMA block occurs on cooling an aqueous dispersion of PGMA–PHPMA worms from 25 °C to 5 °C, which is sufficient to induce a worm-to-sphere order–order transition.34 Thus the specific block composition (rather than the vesicular morphology per se) could be the main reason for the vesicle dissociation observed in the absence of chemical cross-linker. If this were the case, then linear PGMA–PHPMA spheres and worms might also be expected to break up when subjected to the same high shear conditions.
In order to further investigate the possible loss of copolymer morphology during high shear emulsification, six examples of linear and crosslinked block copolymer worms and spheres were prepared using RAFT-mediated PISA formulations. Table 1 summarises the copolymer morphologies, molecular weight data and DLS diameters obtained for the various G–H, G–H–E, G–Bz and G–H–B nanoparticles used in this investigation. It is emphasised that DLS reports a spherical-average hydrodynamic diameter based on the Stokes–Einstein equation; thus this parameter should be treated with some caution when considering highly anisotropic worm-like particles.
Target block copolymer composition | Copolymer morphology | Linear or crosslinked? | M n b/g mol−1 | M w/Mn | DLS diameter (PDI)/nm | Average worm contour length (nm) | Droplet diameter at 0.06% w/w (μm) | Pickering adsorption efficiencya (%) | Droplet surface coverage at maximum efficiencya |
---|---|---|---|---|---|---|---|---|---|
a Calculated at 0.06% w/w copolymer concentration. | |||||||||
G100–H200 | Spheres | Linear | 82![]() |
1.20 | 50 (0.14) | — | 48 ± 14 | — | — |
G100–H200–E20 | Spheres | Crosslinked | — | — | 47 (0.13) | — | 237 ± 129 | 90 | 0.93 |
G45–H140 | Worms | Linear | 39![]() |
1.08 | 315 (0.34) | 176 ± 115 | 46 ± 14 | — | — |
G45–H100–E10 | Worms | Crosslinked | — | — | 122 (0.28) | 172 ± 117 | 179 ± 50 | 90 | 0.47 |
G51–B250 | Spheres | Linear | 51![]() |
1.19 | 81 (0.10) | — | 418 ± 117 | 80 | 0.91 |
G37–H60–B30 | Worms | Linear | 21000 | 1.16 | 72 (0.21) | 120 ± 90 | 219 ± 73 | 100 | 0.71 |
Representative TEM images obtained for the six types of block copolymer nanoparticles summarised in Table 1 are shown in Fig. 2. The synthesis of crosslinked spherical nanoparticles (G100–H200–E20) was relatively straightforward, with 10 mol% EGDMA (based on HPMA monomer) simply being added at the end of the HPMA polymerisation.23 In this case, the cross-linker has minimal effect on both the particle size and morphology, with uniform spheres of around 50 nm diameter being produced both in the presence and absence of EGDMA. In contrast, the preparation of crosslinked worms was rather more problematic. This is presumably because the pure worm phase occupies a relatively narrow region of the phase diagram.27 Highly anisotropic flexible worms were obtained when targeting a G45–H140 diblock (see Fig. 2C). However, replacing 10 units of HPMA with the same number of EGDMA units (i.e. targeting G45–H130–E10) only resulted in a mixed vesicle/worm phase (see Fig. S1 in the ESI†). It seems that the addition of cross-linker shifts the vesicle/worm phase boundary, at least for this particular formulation. In view of this problem, a shorter PHPMA block was targeted to afford an overall triblock composition of G45–H100–E10. Fortunately, this formulation yielded a pure phase comprising crosslinked worms (Fig. 2D). Recently, we have found that RAFT aqueous emulsion polymerisation of BzMA using a PGMA macro-CTA invariably gives a purely spherical morphology, regardless of the target block DPs or copolymer concentration.35 This also proved to be the case in the present work, with G51–B250 forming well-defined, near-monodisperse spherical nanoparticles with a DLS diameter of 67 nm. In striking contrast, growing a relatively short PHPMA block from the PGMA macro-CTA prior to BzMA polymerisation leads to well-defined worms when targeting a G37–H60–B30 composition. This particular triblock copolymer formulation is perhaps best considered as an example of a RAFT aqueous emulsion polymerisation of BzMA using a G37–H60 diblock copolymer precursor. DLS analysis of this precursor indicates the presence of ill-defined, weakly scattering nano-objects with a sphere-equivalent diameter of 141 nm. This suggests that this diblock precursor has not actually undergone micellar nucleation prior to BzMA addition and is therefore likely to comprise weakly interacting molecularly-dissolved copolymer chains. This unusual RAFT PISA formulation clearly warrants further work, but in the present study it is simply exploited as a convenient route to linear G37–H60–B30 worms that comprise a strongly hydrophobic PBzMA block in addition to the weakly hydrophobic PHPMA block.
Previously, Blanazs et al. reported that G54–H140 linear worms exhibit thermo-responsive behaviour in aqueous solution at 10% w/w solids, undergoing a reversible morphological transition to form spheres on cooling from 25 °C to 5 °C as judged by TEM, SAXS and rheology studies.34 Variable temperature DLS studies confirmed that the G45–H140 linear worms prepared in this work undergo a similar thermal transition (see Fig. 3A). Thus the apparent spherical-average hydrodynamic diameter decreases from 315 nm at 25 °C to just 25 nm at 4 °C, with a concomitant reduction in scattered light from approximately 3 × 105 to 6 × 103 kilocounts per second (kcps). These observations are fully consistent with the worm-to-sphere thermal transition reported previously.34 Interestingly, further cooling from 4 to 2 °C led to a further reduction in size from 25 nm to 12 nm and a further drop in count rate to 1 × 103 kcps, which suggests further disintegration of the spherical nanoparticles to give either molecularly dissolved or very weakly interacting copolymer chains. A similar second transition has been inferred for a G49–H130 diblock copolymer by Kocik et al. on the basis of small-angle X-ray scattering studies.36 Moreover, it is worth emphasising that the corresponding sphere-to-worm transition does not occur on returning to 25 °C for this 0.25% w/w aqueous copolymer solution. Such irreversible behaviour is presumably observed because the highly cooperative 1D fusion of multiple spheres to form worms becomes infinitely slow at such high dilution. In contrast, the dissociation of worms to give spheres is not a cooperative process, so it is relatively unaffected at such low copolymer concentrations. Fig. S2† indicates that copolymer concentrations of at least 1.0% w/w are required for the linear G45–H145 worms to regain their original sphere-equivalent diameter after a cooling/warming cycle.
![]() | ||
Fig. 3 Variable temperature DLS studies obtained for 0.25% w/w aqueous dispersions of linear (A) G45–H140 worms, and (B) crosslinked G45–H100–E10 worms and linear G37–H60–B30 worms (see Table 1). The blue and red data points indicate the change in sphere-equivalent diameter on cooling and heating, respectively. Note the irreversible thermo-responsive behaviour observed for the linear G45–H140 worms at this relatively low copolymer concentration, and also the lack of any thermo-responsive behaviour found for the other two types of worms. Error bars represent the standard deviation determined for three separate measurements. |
In contrast, the crosslinked G45–H100–E10 worms exhibit qualitatively different behaviour on cooling a 0.25% w/w aqueous dispersion from 25 °C to 5 °C, see Fig. 3B. There is essentially no change in the apparent sphere-equivalent diameter on either cooling or heating, indicating that no worm-to-sphere transition occurs in this case. These negative observations are of course fully consistent with the additional covalent stabilisation conferred by the EGDMA cross-linker. Similarly, there is no discernible change in particle size during the same thermal cycle for the G37–H60–B30 worms. Thus it appears that the PBzMA block is sufficiently hydrophobic to suppress the thermo-responsive nature of the PHPMA block, with the former conferring additional physical stabilisation via enhanced hydrophobic interactions between the core-forming blocks. In principle, the greater stability indicated by the lack of thermal response observed for the chemically or physically crosslinked worms is likely to enhance the ability of such nanoparticles to survive high-shear emulsification (see later).
Initially, the linear G100–H200 and crosslinked G100–H200–E20 spherical nanoparticles were compared as putative Pickering emulsifiers.
Fig. 4 shows how the mean droplet diameter varies with copolymer concentration for these two dispersions. For the chemically crosslinked G100–H200–E20 nanoparticles, smaller oil droplets are obtained at higher copolymer concentrations. This is because there are more nanoparticles available to coat (and hence) stabilise the oil droplet surface. A minimum mean droplet diameter of approximately 50 μm is obtained at between 0.5 and 1.0% w/w copolymer concentration, which corresponds to the optimum conditions required for emulsification (i.e. minimum droplet diameter and maximum adsorption efficiency). This behaviour is typical of a Pickering emulsifier and has been reported for both PGMA-stabilised polystyrene latexes prepared by conventional aqueous emulsion polymerisation37 and also for the crosslinked G53–H350–E20 vesicles discussed earlier.32 The maximum adsorption efficiency for these G100–H200–E20 spheres was calculated to be 90% based on turbidimetry studies of the lower aqueous phase of the emulsion after creaming of the oil droplets. This makes them rather more efficient than the previously reported crosslinked G53–H350–E20 vesicles,32 but less efficient than the (larger) PGMA50–PS latex particles.37
In contrast, no significant change in the mean oil droplet diameter occurs when varying the concentration of the linear G100–H200 spherical nanoparticles. Taken at face value, these data suggest that the linear spheres can stabilise smaller oil droplets than the crosslinked spheres at a given copolymer concentration (<1.0% w/w). Given that these nanoparticles are essentially the same size, this should not be the case if they adsorb with similar packing densities at the surface of the oil droplets. In principle, the surface coverage (Cs) for these spherical nanoparticles packed around large spherical oil droplets can be calculated using eqn (1) below, as reported previously.11
![]() | (1) |
This equation is simply the surface area of the adsorbed particles divided by the surface area of the droplets, where Cs is the surface coverage of the droplets by the spherical nanoparticles, D is the mean droplet diameter (as determined by laser diffraction), mp is the mean nanoparticle mass, ρp is the nanoparticle density, dp is the mean nanoparticle radius (as determined by DLS) and Vd is the total volume of the oil droplet phase. In the case of the crosslinked G100–H200–E20 nanoparticles of 47 nm diameter, oil droplets with a mean diameter of 237 μm are produced at a copolymer concentration of 0.10% w/w (assuming an adsorption efficiency of 90%, see Table 1). This suggests a Cs value of approximately 0.93, indicating that the entire oil droplet surface is uniformly coated with close-packed nanoparticles. This is certainly consistent with a stable emulsion, since such a dense nanoparticle layer should prevent droplet coalescence. Indeed, TEM provides good evidence for a close-packed monolayer of spherical nanoparticles for this particular system (see later). In principle, P should not exceed 0.84 for small monodisperse spheres packed as a monolayer on larger spheres.38 In practice, this modest discrepancy simply reflects the various approximations that are inherent in such calculations.
In the case of the linear G100–H200 spherical nanoparticles, the same 0.10% w/w copolymer concentration affords stable oil droplets of approximately 48 μm diameter. Using eqn (1), we calculate Cs = 0.19 under these conditions, which suggests that only 19% of the droplet surface is coated by the nanoparticles. Given an oil volume fraction of 0.50, this Cs value seems to be rather low to account for the excellent long-term droplet stability that is observed experimentally, particularly given that the almost identical crosslinked G100–H200–E20 nanoparticles exhibit such a comparatively high surface coverage. Instead, a more likely scenario is that these oil droplets are actually stabilised by individual G100–H200 copolymer chains, which are generated via dissociation of the linear G100–H200 spherical nanoparticles during high shear homogenisation. Thus, as found for the crosslinked G53–H350–E20 vesicles discussed earlier,32 the EGDMA comonomer is essential to preserve the original copolymer morphology generated during the PISA synthesis and hence ensure that a true Pickering emulsion is obtained (as opposed to emulsion droplets stabilised by a molecularly-dissolved G100–H200 diblock copolymer surfactant).
The same experiments were performed on the linear G45–H140 and crosslinked G45–H100–E10 worms, see Fig. 5. An aqueous dispersion of the linear worms was diluted immediately prior to emulsion preparation at 20 °C in order to avoid inadvertently triggering the worm-to-sphere transition. Thus such nanoparticles should be present in their original worm morphology (and this was confirmed by TEM studies). As a comparison, emulsions were also prepared by conducting high-shear homogenisation at 0 °C with the aid of an ice bath. The crosslinked worms displayed similar behaviour to the crosslinked spherical nanoparticles discussed above, and also the crosslinked vesicles reported previously.32 The mean size of the oil droplets is gradually reduced at higher copolymer concentrations until a limiting minimum diameter of around 49 μm is attained. The maximum adsorption efficiency is again approximately 90%, making the affinity of these crosslinked worms for the n-dodecane–water surface comparable to that of the crosslinked spheres. This is perfectly reasonable given that they are both Gx–Hy–Ez copolymers, which should hence exhibit similar particle wettabilities at the oil–water interface. However, it is worth emphasising that, although the same Pickering adsorption efficiency (90%) is observed in both cases, the worms produce significantly smaller oil droplets at approximately half the surface coverage (0.47 vs. 0.93; see Table 1). For example, when using an initial copolymer concentration of 0.10% w/w (for which the final supernatant concentration of non-adsorbed copolymer is ∼0.01% w/w in both cases), the mean droplet diameter is 237 μm for the crosslinked spheres, but only 131 μm for the crosslinked worms. Hence the total surface area of the latter oil droplets is significantly higher. This is because the high aspect ratio of the worms allows the droplet surface to become sufficiently coated to prevent coalescence at a somewhat lower packing fraction than that of the spherical particles, particularly when working at such low copolymer concentrations. It is hypothesised that the nanoparticle packing fraction on the oil droplet surface gradually increases when using higher concentrations of the crosslinked worms, instead of remaining essentially constant as found for the spherical nanoparticles.
Using the linear worms produced oil droplet diameters of approximately 50 μm, regardless of the copolymer concentration. This indicates that the linear worm morphology is not sufficiently robust to survive the high-shear homogenisation conditions, thus the oil droplets formed in this case are actually stabilised by individual copolymer chains. This interpretation is supported by the data obtained using the linear worms for emulsifications conducted at around 0 °C. Under these conditions, DLS studies suggest that linear worms dissociate to produce essentially molecularly-dissolved copolymer chains, see Fig. 3. Hence this is consistent with the almost identical droplet diameter vs. copolymer concentration data sets obtained when using the linear worms at 25 °C and 0 °C, see Fig. 5A.
PGMA-stabilised polystyrene latex particles have been previously employed by Thompson et al. for the preparation of covalently crosslinked colloidosome microcapsules.37 The hydroxyl groups on the PGMA steric stabiliser chains are readily amenable to crosslinking using an oil-soluble polymeric diisocyanate (PPG-TDI), which leads to the formation of colloidosomes. Importantly, these covalently-stabilised latex superstructures can survive an ethanol challenge intact, even after complete removal of the oil droplet phase. In the present work, we used the same approach to cross-link the various linear and crosslinked PGMA–PHPMA nanoparticles (or linear copolymer chains) at the oil–water interface in order to study the copolymer morphology via TEM. However, ethanol is a good solvent for the core-forming PHPMA block and is thus likely to swell or even dissolve the copolymer nanoparticles. Therefore an alternative approach was required: n-hexane was employed instead of n-dodecane to allow convenient removal of the oil phase via evaporation, rather than via an ethanol challenge. The PPG-TDI crosslinker was dissolved in n-hexane prior to homogenisation and the resulting emulsion was allowed to stand at 20 °C for 1 h to enable cross-linking to occur. The n-hexane was then evaporated by magnetically stirring the diluted aqueous colloidosome suspension exposed to the atmosphere at 20 °C (at the back of a fume hood). Intact microcapsules were observed by optical microscopy for both worms and spheres prepared either with or without the EGDMA crosslinker. These microcapsules were imaged by TEM to assess the copolymer morphology. Fig. 6 shows the microcapsules prepared using the linear G100–H200 spheres and the linear G45–H140 worms. The microcapsule surface is completely smooth and featureless in each case, with no evidence for any adsorbed copolymer particles. Thus the microcapsule shell appears to comprise a molecular film of PPG-TDI crosslinked PGMA–PHPMA copolymer chains. These observations are fully consistent with the concentration-independent droplet diameters observed when employing such linear nanoparticles as putative Pickering emulsifiers. Moreover, similar TEM observations were reported for linear PGMAx–PHPMAy vesicles in our earlier study.32 Thus all the experimental evidence suggests that linear PGMA–PHPMA nanoparticles are not sufficiently robust to survive high shear homogenisation.
In contrast, Fig. 6C and D shows TEM images obtained for colloidosomes prepared using crosslinked G100–H200–E20 spherical nanoparticles and crosslinked G45–H100–E10 worms, respectively. Clearly, the former emulsifier leads to the formation of colloidosomes that comprise densely-packed spherical nanoparticles. This image is consistent with the high packing efficiency calculated above, but is in striking contrast to the smooth, featureless microcapsule surface obtained when using the corresponding linear spherical nanoparticles. Moreover, it is perhaps worth noting that the inter-particle separation distance within the colloidosome wall is significantly smaller than in our previous work,37 which in principle could lead to enhanced retention of macromolecules or nanoparticles encapsulated within such colloidosomes.39
A densely-packed layer of nanoparticles is also observed on the surface of colloidosomes prepared using the crosslinked G45–H100–E10 worms (see Fig. 6D). In the literature, it is generally assumed that highly anisotropic rod-like particles such as ellipsoidal polystyrene particles lie predominately flat at both the oil–water and air–water interfaces.16,40,41 Similar behaviour is also observed for cellulose nanofibers.18,19 In the present study, this also seems to be the case for the majority of worms, although some worms do appear to protrude out from the edge of the colloidosome surface. However, it is not clear whether this is actually the case in the ‘wet’ emulsion, or if this is simply a drying artefact that arises during TEM grid preparation. In principle, it may be feasible for some fraction of a relatively long worm to become adsorbed within the plane of the droplet surface, with the remainder of the worm extending out into the aqueous phase. Given the relatively flexible nature of these worms, this hypothesis is not unreasonable (particularly in the limit of high worm coverage).
Overall, it is clear that linear PGMA–PHPMA diblock copolymers in the form of either spherical nanoparticles or worms dissociate during homogenisation (just like the analogous vesicles reported earlier32) to produce individual copolymer chains. Under such high-shear emulsification conditions, the EGDMA cross-linker is essential to preserve the original copolymer morphology and hence ensure genuine Pickering stabilisation of the oil droplets, as opposed to emulsion stabilisation via molecularly-dissolved block copolymer surfactant. However, PHPMA is known to be only weakly hydrophobic,42 so we decided to examine whether a more hydrophobic core-forming block could enhance the stability of such nano-objects when evaluated as putative Pickering emulsifiers.
To address this important question, linear G51–B200 spherical nanoparticles of approximately 67 nm diameter were prepared via RAFT aqueous emulsion polymerisation for assessment as a Pickering emulsifier. Accordingly, these nanoparticles were homogenised with n-dodecane at various copolymer concentrations and the resulting emulsions were characterised using optical microscopy and laser diffraction (see Fig. 7).43 Using the G51–B200 spheres leads to a strongly concentration-dependent droplet diameter, unlike the earlier data set obtained for the linear G100–H200 spherical nanoparticles (which is included in Fig. 7 to aid direct comparison). The increase in mean droplet diameter observed on lowering the concentration of the linear G51–B250 spheres is fully consistent with the behaviour of the EGDMA crosslinked nanoparticles discussed above, as well as the PGMA–PS latexes reported earlier.37 TEM imaging of the colloidosome surface prepared with these G51–B250 spheres also shows fully intact spheres adsorbed at the droplet interface. Thus it appears that chemical cross-linking is not a pre-requisite to preserve the original copolymer morphology during homogenisation provided that the core-forming block is sufficiently hydrophobic to stabilise the nanoparticles with respect to their shear-induced dissociation.
Finally, high shear emulsification of n-dodecane was conducted in the presence of the linear G37–H60–B30 worms. The important question here is: does a relatively short PBzMA block confer sufficient stability to the worms so as to enable the formation of a genuine Pickering emulsion? Fig. 8 confirms that larger droplets are formed as the worm concentration is gradually reduced. In view of the data sets discussed above, this strongly suggests that these anisotropic nanoparticles adsorb intact at the oil–water interface, in contrast to the molecularly-dissolved copolymer chains obtained when examining the linear G45–H140 worms. As indicated by the absence of a worm-to-sphere transition for G37–H60–B30 (see Fig. 3), even a relatively short PBzMA block is sufficient to stabilise these worms during high-shear emulsification; in this case nanoparticle dissociation is prevented as a result of the significantly stronger hydrophobic interactions between the core-forming blocks. This is further confirmed by TEM analysis of the colloidosome surface in Fig. 6F, which clearly shows that the original worm-like morphology is preserved at the droplet surface. It is also emphasised that these G37–H60–B30 worms proved to be extremely efficient Pickering emulsifiers: no worms could be detected in the supernatant phase after droplet creaming, suggesting essentially 100% adsorption at the oil–water interface. We speculate that this enhanced efficiency compared to the EGDMA crosslinked worms may be a result of the more hydrophobic PBzMA block increasing the particle contact angle at the oil–water interface, hence leading to stronger adsorption.
It is worth emphasising that, in general, the worms can be considered to be more effective Pickering emulsifiers than the equivalent spherical nanoparticles. More specifically, they are at least as efficiently (and almost certainly rather more strongly) adsorbed at the droplet interface. Moreover, smaller oil droplets are consistently produced when worms are used as the Pickering emulsifier when compared to an equal mass of spherical nanoparticles. The surface coverage (Cw) of the worms adsorbed on the surface of the oil droplets can be estimated using a modified version of eqn (1), which was derived for spherical nanoparticles.
![]() | (2) |
The main difference between eqn (1) and (2) is that hp now represents the mean worm thickness (estimated from TEM) rather than the sphere diameter. The surface coverages calculated for the lower worm concentrations are shown in Table 1. Worm surface coverages, Cw, of 0.47 and 0.63, are calculated for the EGDMA and BzMA stabilised worms respectively. These values are significantly lower than those obtained for oil droplets stabilised using spherical nanoparticles. The reciprocal of the mean oil droplet diameter, D, is plotted against the mass of adsorbed worms, mp, in Fig. 9a. According to eqn (2), the gradient should be inversely proportional to the worm surface coverage (Cw).19,20
Thus two distinct surface coverage regimes can be obtained, depending on the copolymer concentration used to prepare the Pickering emulsions. Very similar behaviour has been recently reported by Kalashnikova et al.19 for anisotropic Pickering emulsifiers based on bacterial cellulose nanofibres. At a relatively high copolymer concentration (e.g. 1.0% w/w) a high surface coverage is obtained, while at a relatively low copolymer concentration (e.g. 0.10% w/w) a significantly lower surface coverage is observed. [In contrast, a linear plot – indicating a single ‘high surface coverage’ regime - is obtained for the crosslinked spheres, see Fig. S4 in the ESI.†] This was confirmed by TEM analysis of dried Pickering emulsion droplets (prepared using n-hexane as the oil phase and in the absence of any PPG-TDI crosslinker) for both copolymer concentrations (i.e. 0.10 and 1.0% w/w) of the G37–H60–B30 triblock copolymer worms (see representative images shown in Fig. 9b and c, respectively). Clearly, the Pickering emulsion droplet surface is much more densely packed with worms when using the higher copolymer concentration. At the lower copolymer concentration, the worms appear to adjust their surface packing density by forming ‘islands’, rather than individual worms becoming more evenly spaced. This cooperative behaviour allows them to stabilise somewhat smaller oil droplets than the equivalent spherical nanoparticles, while at the same time being much more strongly adsorbed at the oil–water interface. It is worth noting that these triblock copolymer linear worms are much more flexible than the bacterial cellulose nanofibres reported by Kalashnikova et al.19 Thus it is the particle anisotropy, rather than the particle stiffness, that appears to be responsible for their strikingly similar Pickering emulsifier behaviour. In summary, the highly anisotropic character of these copolymer nanoparticles directly enhances their ability to act as an efficient Pickering emulsifier.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4sm01724b |
This journal is © The Royal Society of Chemistry 2014 |