Deep eutectic solvent in water pickering emulsions stabilised by cellulose nanofibrils

Deep eutectic solvent (menthol : dodecanoic acid) in water (30 : 70) emulsions stabilised with partially oxidised cellulose nanoparticles remained stable for 200 days at room temperature. Deep eutectic-based emulsions offer potential for non-aqueous reaction systems, chemical extraction, and controlled release. Pickering emulsions using polysaccharides are less toxic and more stable than surfactant-stabilised emulsions.


Introduction
Deep eutectic solvents (DESs) are a promising new type of nonvolatile solvent composed of hydrogen bond donors and acceptors. They are a highly tuneable subclass of ionic liquid (IL) but are oen cheaper, easier to make, and more environmentally friendly. [1][2][3][4] DESs and the broader ionic liquid (IL) solvent class have potential as reaction media and reactor systems. 2,[5][6][7][8][9] Their tuneable properties, including temperature stability and solvation of drug and catalytic molecules make ILs and DESs preferable to traditional oil/water emulsions. 2,10,11 There have been some reports of ionic liquid (IL) based emulsions (using IL as an additive, or as the oil, or as the polar phase, or even IL in IL emulsions). [10][11][12][13] There are also reports of DES in oil emulsions, stabilised with conventional surfactants. 14 However, to the authors' knowledge, the only report of a DES in water emulsion comes from the commercial EMLA® cream (EMLA stands for Eutectic Mixture of Local Anesthetics) which uses a eutectic mixture of lidocaine and prilocaine, stabilised by a non-ionic surfactant. 15 The DES in water pickering emulsion reported here has an advantage over these previously reported formulations because it does not use expensive and potentially toxic surfactants. In addition, as water is the major phase, it is more compatible with applications that require low toxicity and low cost, than DES in oil emulsions.
Menthol : dodecanoic acid (2 : 1) is a hydrophobic DES with a melting point around 7 C. 16 It has potential for extraction of materials from water, 17,18 separation of metals, 19,20 and extraction of pesticides from water. 21 TEMPO-oxidised cellulose nanobrils (OCNF) are negatively charged nanoparticles that can form Pickering-type emulsions. [22][23][24] Pickering emulsions stabilised by other polysaccharide-based nanoparticles have also been reported. [25][26][27] However, until now, their use has been limited to emulsions using model oils (such as hexadecane), or sunower oil.
Emulsions containing 30 vol% menthol : dodecanoic acid were stabilised using either OCNF, or hydrophobically modied OCNF (C8-OCNF). These emulsions, including their stability over time were assessed using laser diffraction, visual observation, and small-angle X-ray scattering.

Oxidised cellulose nanobrils
Oxidised cellulose nanobrils provided by Croda® as a ca. 8 wt% solids paste in water were prepared via TEMPO-mediated oxidation as previously described. 28,29 Previous work by this group determined the degree of oxidation for this particular batch of brils to be 25% i.e. number of carboxylate groups compared to total anhydroglucose units. 30,31 The brils have a large aspect ratio (hundreds of nm in length and a cross section of $5 nm) 32 and a high negative surface charge (À60 mV in z potential). [28][29][30] Residual salts and preservatives were removed from OCNF by dialysis against deionised water (18.2 MU cm) as previously described. 32 The OCNF was then freeze-dried and resuspended to 1.5 wt% in deionised water before being dispersed by sonication (Ultrasonic Processor, FB-505, Fisher -550 W), at 30% amplitude with 1 s on 1 s off pulses, for $1 h or until the dispersion became transparent.
The reaction was le overnight and then 50 wt% propan-2-ol (BDH Chemicals, 100%) in water was added to the reaction mixture in a 2 to 1 ratio which causes the C8-OCNF to aggregate. The mixture was centrifuged at 5000 RCF (relative centrifugal force) for ten minutes and the precipitate (C8-OCNF) collected. The precipitate was washed twice more with 50 wt% propan-2-ol and centrifuged.
The precipitate was re-suspended in deionised water and dialysed against deionised water for 24 h to remove the propan-2-ol. The C8-OCNF was dialysed for a further 24 h against deionised water with a pH of 3 to remove any ionically bound amine. Further dialysis was performed against neutral deionised water for 72 h with regular replacement of the water to remove any remaining salts.
As with OCNF, the C8-OCNF was then freeze-dried and redispersed to the desired wt% using sonication.
The z-potential of C8-OCNF was measured using a Malvern Zeta-sizer Nano ZSP® (Malvern, UK) in a folded capillary electrode cell using ultrapure Milli-Q water as the dispersant. The sample was equilibrated at 25 C for 120 s prior to testing and the results taken from an average of 3 measurements of 100 scans each.

Starch
Starch dispersions were prepared by dissolving 1.5 wt% soluble starch (Sigma-Aldrich, S9765) in deionised water at 80 C with stirring for 45 minutes, then allowing it to return to room temperature.

Characterisation
Droplet measurements were made using a Mastersizer 3000E laser diffraction particle size analyser (Malvern, UK). Samples were added dropwise to the dispersion unit until the obscuration was within the acceptable range (7-20%). Five repeat measurements were made for each time point to ensure sample stability. Droplet size is reported as Sauter diameter D [3,2] for easy comparison.
SAXS measurements were performed on an Anton-Paar SAXSpoint 2.0 provided by the Material and Chemical Characterisation Facility (MC 2 ) 33 equipped with a copper source (Cu Ka, l ¼ 1.542Å) and a 2D EIGER R series Hybrid Photon Counting (HPC) detector. The sample detector distance was 556.9 mm covering a q range of about 0.008-0.4Å À1 . The emulsions were loaded into 1 mm quartz capillaries and measured at 25 C (AE0.1 C Peltier unit). Data was collected in one frame, with 900 s exposure, then processed. Fitting was performed using SASView (Version 4.2.1, see http:// www.sasview.org/ for more information). The model for elliptical cylinders was used from SASView 4.2.1 without modication. 34 Fitting parameters are given in the ESI. ‡ Viscosity of 1.5 wt% OCNF or C8-OCNF in water was measured using a stress-controlled Discovery Hybrid Rheometer, Model HR-3 (TA Instruments) with a sand-blasted 40 mm parallel plate geometry with a gap of $800 mm. Flow sweeps were performed at shear rates from 0.01 to 100 s À1 with ten points per decade, at 25 C (Peltier unit, AE0.1 C).

Results and discussion
Emulsions made with OCNF remained visually stable for more than 200 days at room temperature with no evidence of aggregation or creaming (Fig. 1). Emulsions made with C8-OCNF creamed aer 23 days, but the creamed layer remained stable (see Fig. S1 ‡). Creaming is due to the lower density of the DES droplets, compared to the bulk water.
Visual observations were supported by laser diffraction measurements of droplet size. As shown in Fig. 2, droplets stabilised by OCNF remained stable in excess of 200 days with only a minor increase in droplet size. Droplets stabilized with C8-OCNF were smaller and appeared to be more stable, with almost no change even aer 100 days.
The hydrophobic chains on C8-OCNF will allow greater interaction with the hydrophobic DES, compared to unmodied OCNF, and reduce the energy necessary to create a new interface, thus improving homogenization. This could allow tighter packing around the droplets, leading to smaller, more stable droplets. Similar effects have been observed for cellulose nanocrystal-stabilised emulsions using standard oils. 35,36 While larger droplets may be expected to cream faster due to having a lower density, this was not the case here. The difference in creaming between OCNF and C8-OCNF is due to the higher viscosity of the continuous phase of OCNF-over C8-OCNF-stabilised emulsions which works to prevent creaming (Fig. S2 ‡). 22 Increased viscosity comes from the higher surface charge of OCNF compared to C8-OCNF, which increases the excluded volume.
Previous emulsion research demonstrated that cellulose nanobrils with higher charge provided greater stability than lower charged brils. 23 The method of hydrophobisation used here necessitates the replacement of some of the negative carboxylate groups with hydrophobic chains, thus reducing the z potential from À60 mV to À43 mV. This would reduce repulsion between brils, and allow movement of droplets, ultimately leading to creaming.
In comparison, emulsions stabilised with soluble starch polymers, rather than cellulose particles, which have an even smaller z potential (À14 AE 1 mV) 37 creamed on day 2. By day 23 there was precipitation and clumping and the mean droplet size was over 20 mm (Fig. S3 ‡).
Small angle X-ray scattering (SAXS) was used to investigate the structure of these emulsions. Fig. 3 shows the scattering patterns for OCNF and C8-OCNF in water and for the DES in water emulsions.
The scattering pattern of OCNF in water can be t with an elliptical cylinder model with a minor radius of 12.4 AE 2Å and a major radius of 58 AE 9Å, in line with previous research. 32 The same model can be used to t the majority of the data for the emulsion sample, with a minor radius (13.4 AE 2Å) and a major radius (66 AE 9Å) that are within the error for OCNF brils in water, suggesting that the OCNF brils are dominating the SAXS signal.
The scattering pattern for C8-OCNF in water can be t with the same parameters as OCNF except that the length is significantly shorter (300 AE 20 compared to >1000Å). Given the benign nature of the modication procedure, this apparent change in length is most likely a result of bril ocs. The length reported by SAXS is actually the distance between intersection points rather than an actual shortening of the brils (see Fig. S5 ‡). Similar aggregation has been observed for nanocrystalline cellulose upon addition of salt to screen repulsive charges. 38 As with OCNF, the SAXS data from the emulsion stabilised by C8-OCNF can be t with a similar model.
As shown in Fig. 3, the scattering from both emulsion samples have an upturn at low q which cannot be tted with the elliptical cylinder model. This is probably from the emulsion droplets. Based on the laser diffraction data, these droplets are far outside of the probed range of this SAXS instrument and so cannot be tted. The upturn appears to happen at higher q for the C8-OCNF stabilised emulsions, indicative of smaller droplets, which is consistent with the laser diffraction results. Also, due to the smaller surface charge, C8-OCNF is prone to selfaggregation, which could also contribute to the low-q signal.
The SAXS results demonstrate that for both OCNF and C8-OCNF the cellulose nanobril structure is unchanged in the presence of the hydrophobic DES, compared to in water dispersions.
For the purposes of completeness, DES in water emulsions were also made with more traditional surfactants (dioctyl sodium sulfosuccinate (AOT) or Tween20). This produced emulsions with even smaller droplets ($0.2-0.6 mm) that remained stable for more than 60 days (see ESI ‡). However, the emulsion stabilised with AOT broke down by day 100, with separate oil and water layers. The emulsion stabilised with  Tween20 creamed within two weeks but otherwise remained visually stable throughout the observation period (see ESI ‡). While the surfactant-stabilised emulsion appeared somewhat stable, the cellulose-stabilised pickering emulsions are of greater interest, not only due to prolonged stability (compared to AOT-stabilised) but also due to bioavailability and renewability of this stabiliser. Furthermore, particle-stabilised pickering emulsions are superior to surfactant-stabilised emulsions e.g. in terms of stability (e.g. of very large droplets, a wide range of oil/water ratios, and against Ostwald ripening) and the capacity for phase inversion. 39 Particles can enhance emulsion stability not only by coating droplets like surfactants but also by forming a three dimensional network which immobilises droplets. [39][40][41] The OCNF offers the added benet of being a rheological modier on its own. 42,43 Thus, stabilization and rheological control can be achieved using the same additive.
Preliminary research by our group suggests that other polysaccharide nanoparticles (e.g. cellulose nanocrystals) are also capable of stabilising DES in water emulsions.
Future research will focus on using other hydrophobic DESs, or even making DES in DES emulsions, as well as exploring applications of DES-based emulsions.

Conclusions
This research presents for the rst time a DES in water emulsion stabilised by polysaccharides. DES in water emulsions offer new possibilities e.g. for chemical extraction and separation, controlled release (e.g. of drugs), as reactor systems, and for template-based synthesis e.g. of metal organic frameworks.

Conflicts of interest
There are no conicts to declare.