Emulsion stabilisation by complexes of oppositely charged synthetic polyelectrolytes.

We investigate the possibility of stabilising oil-water emulsions from the polyelectrolyte complexes (PEC) obtained in mixtures of a strong cationic polyelectrolyte (poly(diallyldimethylammonium chloride), PDADMAC) and a weak anionic one (poly(acrylic acid)sodium salt, PAANa). Unlike other previous work however, both polyelectrolytes (PEL) are chosen as they are completely water-soluble and possess no surface activity when present alone over nearly all the pH range. In water, the effects of PEL concentration, PEL mixing ratio and pH on the formation of PEC are studied in detail. At low pH where the anionic PEL is uncharged, complex coacervation occurs in which droplets rich in both polymers are dispersed in water. At intermediate pH, the PEC comprise a mixture of coacervate droplets and solid particles. At high pH where the anionic PEL is significantly charged, only complex coacervation is observed. On addition of dodecane followed by homogenisation, no stable emulsions arose from dispersions containing solid particle PEC due to either the large precursor particle aggregates or their inherent hydrophilicity. By contrast, oil-in-water emulsions stable to coalescence could be prepared from coacervate dispersions. We discuss the feasibility of the coacervate phase spreading at the oil-water interface in terms of the relevant spreading coefficients and compare the predictions with experiment for a range of oils. We encounter oils whose drops become engulfed by the coacervate phase as well as oils where no engulfing occurs.


INTRODUCTION
When mixing two polymer solutions, different types of phase separation can occur. If the polymers are incompatible (they repel each other), segregative phase separation takes place and each polymer is collected (predominantly) in a different phase. One example is the two phases formed by mixing dextran and methylcellulose in water. 1 Complete miscibility appears in exceptional cases and gives rise to a homogeneous solution as in the mixture of polystyrene and poly(2,6-dimethyl-1,4-phenylene oxide). 2 Finally, if the polymers show net attraction, usually through electrostatic interactions as with oppositely charged polyelectrolytes (PEL), associative phase separation is achieved. In this case, the two polymers are collected in one phase while the other phase consists almost entirely of solvent. This interaction can lead either to the formation of a complex coacervate (liquid-liquid type of phase separation) or a precipitate (solid-liquid type of phase separation). Both complexation mechanisms are driven by an increase of entropy due to the release of counter-ions initially bound to the PEL chains.
Although the factors that dictate the nature of each type of associative phase separation are not fully elucidated, it is generally assumed that strong interactions between PELs results in precipitates while in coacervation the interactions are relatively weak. 3,4 The distinction between these two types of phase separation is not clear in the literature and the general term polyelectrolyte complex (PEC) is adopted for both scenarios. However, optical microscopy is a simple and useful technique to distinguish between precipitation (amorphous solid particles) and complex coacervation (micron sized droplets). 3 The term coacervation was coined by Bungenberg de Jong and Kruyt in 1929 and derives from the Latin words "co" (together) and "acerv" (a heap or aggregate). 5 In coacervation, a dense polymer-rich phase (coacervate phase) and a very dilute polymerdeficient phase (aqueous phase) coexist. 6 Coacervates are subdivided into simple and complex.
In simple coacervation one macromolecule is present and the associative process is induced by the addition of a dehydrating agent such as salt or a change in temperature or pH of the medium.
One example is the mixture of polyethylene glycol, potassium phosphate and water, in which the bottom phase is rich in salt and the top phase is rich in polymer. 1 In complex systems, two oppositely charged species are involved and two immiscible liquid phases arise. The upper phase (supernatant) consists almost entirely of solvent whilst the bottom phase is a dense clear liquid phase (coacervate) concentrated in both types of macromolecule. Before reaching this steady state, a turbid mixture composed of a metastable suspension of macroion-rich coacervate droplets dispersed in solvent is present. 6 Coacervates (droplets) hold higher amounts of water compared to precipitates (particles). Jha et al. determined the weight fraction of water in the coacervate system containing poly(acrylic acid) (PAA) as the polyacid and either poly(N,Ndimethylaminoethyl methacrylate, PDMAEMA) or poly(diallyldimethyl ammonium chloride), PDADMAC, as the polybase. 7 The weight fraction of water varied from 0.4 to 0.8 in these systems. In the mixture PAA-PDMAEMA, Spruijt et al. 8 measured water fractions > 0.7 which increased with salt concentration. Another special feature of the coacervate phase is that it displays a much higher viscosity compared to the initial PEL solutions. Liu and co-workers reported an increase by three orders of magnitude in the viscosity of the coacervate phase formed between PAA and PDADMAC compared to the individual PEL solutions. 4 Finally, low interfacial tensions (≤ 1 mN m -1 ) between the coacervate phase and the coexisting supernatant phase are common. [9][10][11] One of the most important applications of complex coacervation is in microencapsulation. 5 In this process, a substance (also called core material) which can be liquid or solid in nature, is encapsulated within a layer of coacervate phase. 12 Thus, oil drops or solid particles in water may be coated with this phase. For the encapsulation to be successful the adsorption of the complex onto the core material and its wettability are key parameters to consider. 13 The complex coacervate has to spread spontaneously over the surface of dispersed liquid droplets or particles and coat them to form a capsule. 14  where the subscripts 1, 2 and 3 refer to the core material (here oil), the aqueous phase and the coacervate, respectively. Thus, S1 is the spreading coefficient for oil at the water-coacervate interface etc. If we assume that 23 (water-coacervate) < 12 (oil-water) as mentioned above, then S1 is always negative and only three possible combinations of the spreading coefficients arise depending on the relative magnitudes of S2 and S3: 15 1 < 0; 2 < 0; 3 < 0 partial engulfing [5] 1 < 0; 2 > 0; 3 < 0 non-engulfing [6] A microcapsule containing a liquid core is similar to an emulsion droplet. As emulsions are thermodynamically unstable, a surface-active material or emulsifier should be added to protect the formed drops from coalescence. 16,17 In the past, both emulsions and microcapsules stabilised by protein (anionic)-polysaccharide (cationic) complexes have been extensively reported. Two different preparations are described. 18 The first consists of the addition of oil to an aqueous solution containing the protein-polysaccharide complexes followed by homogenisation. The second involves the formation first of a primary emulsion stabilised by protein, followed by the addition of the polysaccharide which adsorbs onto the protein layer forming a bilayer/multilayer. Jourdain et al. 19 found an improvement in emulsion stability by using the complexes prepared in water instead of adding the protein and the polysaccharide in two homogenisation steps. Bridging flocculation occurred with the second method while discrete dispersed oil droplets were achieved by using the complexes formed before homogenisation. They related these differences to the structure of the composite biopolymer at the interface that varies depending on the method employed. In all cases studied, the complex at the emulsion droplet surface enhances emulsion stability compared with emulsions of protein alone.
Despite the numerous examples of emulsions stabilised by protein-polysaccharide mixtures encountered in the literature, [19][20][21] in all cases the protein acts as a good emulsifier alone, i.e. is surface-active, and in some cases so does the polysaccharide. In this work we extend the idea introduced in our previous paper in which we investigated whether PEC formed between two oppositely charged strong polyelectrolytes (polystyrene sodium sulfonate, PSSNa and PDADMAC) were capable of stabilising oil-water emulsions. 22 The novelty of this system compared to those containing protein-polysaccharide complexes is that neither polyelectrolyte is surface-active alone and emulsion stabilisation was shown to be due to the adsorption of solid particles of PEC pre-formed in water on the surface of oil drops. Here, we study emulsion stabilisation in a different system of water-soluble polymers constituted by a strong (PDADMAC) and a weak (PAANa) polyelectrolyte in an attempt to establish the general pattern of behaviour in both aqueous PEC dispersions and emulsions containing them. For this system, unlike the first one, 22 the effect of pH is an important consideration as PAANa will be uncharged at low pH and fully charged at high pH. As a result, this will have an impact on the strength of the interaction between oppositely charged polymers; we know of no other study reporting the behaviour over a wide range of pH. We first study the behaviour of aqueous mixtures of the two polymers at different pH and mole fraction of each polyelectrolyte. The size and charge of the obtained complexes are measured and the type of associative phase separation is carefully appraised. Both solid precipitates and coacervate droplets form along the studied pH range (2)(3)(4)(5)(6)(7)(8)(9)(10), while for the case of two strong PELs only precipitates were detected. 22  , formamide (> 99%), -bromonaphthalene (97%) and n-hexadecane (99%). The latter three were purchased from Sigma-Aldrich.

(a) Preparation and characterisation of aqueous PEC dispersions
A potentiometric titration for PAANa was performed with a calibrated pH meter (3510, Jenway) to determine the degree of ionisation as a function of pH. The natural pH of a 1 g L -1 PAANa solution was 9.7. The pH was increased to 12 with NaOH and the solution was titrated

(c) Determination of spreading coefficients
The three spreading coefficients for the system oil-water-coacervate were calculated from the measured interfacial tensions between the different phases ( 12 , 23 and 13 ). The oils under study are those included in Table 1. The interfacial tension between oil and water ( 12 ) was measured with a Krüss K11 tensiometer and the du Noüy ring method at 25 °C. The radius of the Pt-Ir ring was 9.545 mm and the wire diameter was 0.37 mm. The applied correction factors were taken from Harkins and Jordan. 25 The results of three separate measurements were averaged. After each measurement, the ring was rinsed with ethanol and heated to glowing in a blue Bunsen flame. The two remaining interfacial tensions ( 23 and 13 ) were calculated from the appropriate Young's equation (equations 7 and 8). For convenience the phases 1 (oil), 2 (water) and 3 (coacervate) have been re-named as o, w and c, respectively. Therefore, the two interfacial tensions to calculate are designated as and .
= + cos (water drop in air on coacervate) [7] = + cos (oil drop in air on coacervate) [8] where a signifies air. The surface energy of the coacervate ( ) was determined by using an indirect method in which the surface energy of a material is split into a polar ( ) and a nonpolar component ( ). As described earlier, 26 individual PEL solutions at pH = 10 was spread carefully on a clean glass slide and water was allowed to evaporate completely in an oven at 100 °C. After that, between 4 and 10 L of the test liquids were placed on the slide with a needle. The circle method was used to measure contact angles below 20º while the Young-Laplace fitting was applied for contact angles above 20º. When the contact angle between the liquid and the coacervate coated surface was too low to be measured accurately, it was taken to be < 5º. The contact angle values are given as the average of three independent measurements.
The least squares calculation was carried out to determine the best combination of and that fits all the data simultaneously. A 3-D surface energy diagram is obtained, which represents the goodness of fit (inverse of the sum of the squares of variances) to contact angle, set against a matrix of possible values of and . The values that best fit all the contact angles are read from the coordinates that define the peak. Finally, the surface tensions ( and ) and the contact angles ( and ) in equations 7 and 8 are measured with the du Noüy ring method and the Krüss DSA instrument, respectively. The procedure followed for these measurements is the same as stated above. This allows us to calculate the two remaining interfacial tensions in equations 7 and 8 ( and ) and hence the three S values defined in equations 1-3.

Characterisation of aqueous PEC dispersions
Aqueous PEC dispersions and oil-water emulsions were studied in detail for mixtures of PDADMAC and PAANa. Since PDADMAC is a strong polybase, it is fully ionised at all pH. On the contrary, PAANa is a weak polyacid. Therefore, its degree of ionisation varies with pH. A potentiometric titration of a 1 g L -1 PAANa solution was carried out in triplicate to determine the degree of ionisation as a function of pH and consequently the pKa of the polyacid ( Figure S1). The pKa value for our sample of PAANa was found to be 6.6, in reasonable agreement with the literature whose values range from 5.5 to 6.8. 28,29 It is worth noting the increased pKa of a polyacid compared to the value of a low molecular weight analogue (pKa = 4.2). The behaviour of weak polyelectrolytes upon increasing or decreasing the charge density is not equivalent to the case of low molecular weight electrolytes. In a polyelectrolyte, not all the counterions will dissociate from the polymer chain. In fact, as the magnitude of the charge on the chain increases (by dissociation), it becomes progressively more difficult to remove the next proton due to the close proximity of charged groups. When the polyelectrolyte is strongly charged, some of the counterions remain bound to the polymer chain due to the large electrostatic potential on the chain, thereby reducing the effective charge of the polyion. This is known as counterion condensation and is described by Manning's theory. 30

(a) Low [PEL]
As mentioned above, the pH is a key parameter to consider in the study of PEC formation in this polyelectrolyte mixture. Aqueous PEC dispersions at low concentration were first characterised in terms of their average diameter and zeta potential. The appearance of aqueous PEC dispersions prepared from 0.1 g L -1 PEL solutions at different xPAANa and pH is shown in Figure 1.
Here, x refers to the mole fraction using the values of Mw given in Table 1 (131. At the other mole fractions, the samples were too polydisperse for cumulant analysis and in some cases the size distribution showed more than one peak. Therefore, from the results obtained at extreme pH values, one can state that the interactions in the system PDADMAC + PAANa cannot be ruled by electrostatics alone. This associative phase separation has to be also mediated through additional inter-molecular interactions, such as van der Waals, hydrogen bonding or hydrophobic and dipole interactions. 34 This agrees with previous work from Alonso et al. 35 They detected aggregates at pH = 3 where PAA is not ionised. Their occurrence was explained by the formation of H-bonds between different chains of PAA. 35 The carboxylic acid moiety is a highly polar functional group due to the strongly polarized carbonyl and hydroxyl groups which can interact through hydrogen bonding with other carboxylic acids or water molecules. 35 At low pH, the authors also pointed out the possibility that PDADMAC could form complexes with single chains or small aggregates of PAA. At high pH, no complex was detected at pH = 13. By increasing the pH, the charge density of PAA increases. As a result, there is an increase in the amount of water linked to the polymer through hydrogen bonding. 35 Conversely, the charges in PDADMAC are given by a quaternary amine surrounded by an organic environment, which makes it less favourable for structured water. 35 At intermediate pH these quaternary amines can destroy the water shell that surrounds the carboxylic acid groups of PAA and electrostatic interaction takes place. 35 However, with increasing pH, the number of water molecules associated with the carboxylate group increases and the breaking of the water shell by the quaternary amines becomes less likely. 35 As a result, no complexation is achieved. The authors proved the hindering effect of water molecules since, after the addition of a hydrogen bonding breaker, the complexation at high pH took place. 35 However, it is worth mentioning that other parameters such as the chain rigidity and the ionic strength can also affect the strength of the interaction between charged polymers. 3,4,36 The zeta potential of each mixture at various pH was also measured (  Figure S3 for three extents of ionisation of PAANa. Although the trends are the same, the data does not collapse onto a single curve and we prefer to use xPAANa.

(b) High [PEL]
In terms of visual inspection at this low concentration, this system is comparable to the system containing PDADMAC + PSSNa (two strong polyelectrolytes) reported previously. 22 However, things change upon increasing the concentration of the initial PEL solutions. This study was completed at four different pH values covering the entire range, i.e. pH = 2, 4, 6 and 10. Figure 4 shows In order to further evaluate the difference between these two types of associative phase separation (precipitation and coacervation) centrifugation of 1 mL of the aqueous PEC dispersions (xPAANa = 0.5) prepared at two different pH was carried out. The initial concentrations of the PEL solutions used were 5 g L -1 and 10 g L -1 for pH = 4 and 10, respectively. At pH = 4, after centrifugation a white precipitate is collected on the bottom of the Eppendorf tube as seen in Figure 5(2), consistent with the solid-liquid type of phase separation. The appearance of the aqueous PEC dispersion at pH = 10 is shown before and after centrifugation in Figure 5(1). The dispersion was initially whitish which after centrifugation yields a phase separated system consistent with the liquid-liquid type of phase separation of the complex coacervate. Three different phases were identified and designated supernatant (phase A), interface (phase B) and coacervate phase (phase C). The supernatant was mostly water depleted in coacervate droplets ( Figure 5(A)). Some spherical entities were still evident around the edges of the drop as soon as the water began to evaporate. However, they were hardly visible at the outset. The interface consisted of an aqueous phase rich in coacervate droplets (Figure 4(B)). Images of the same drop placed on a glass slide were taken at different times. Coalescence of the coacervate droplets occurred as after 17 min a substantial increase in their size was detected. After 22 min droplets were no longer visible as they had coalesced to a bulk film of coacervate phase. Finally, the coacervate phase after centrifugation was a transparent and viscous phase concentrated in both polymers ( Figure 5(C)). The water content was 63 ± 3% which is close (~ 66%) to the value reported for the same system at pH = 7. 7 The entire pH range was studied at a high PEL concentration (5 g L -1 ) for a mixture containing an equal mole fraction of both polymers (xPAANa = 0.5). form. As pointed out in the introduction, it is generally assumed that precipitation rather than coacervation is obtained for strongly interacting polymer pairs. Therefore, one would expect precipitate formation at high pH values which is not the case. This may be related to the enhanced ability of PAANa to form hydrogen bonds at high pH, as explained earlier. Due to the dipoles present in the carboxylic acid moieties, water is associated to them through hydrogen bonding leading to partial screening of the negative charge of the ionic groups. 35 Therefore, the electrostatic interaction with PDADMAC is weakened and this results in the formation of a complex coacervate containing water instead of a precipitate. The volume fraction of coacervate droplets is higher at pH 5 and 6 cf. higher pH and these samples display maximum turbidity initially ( Figure 6). One year after preparation the coacervate droplets in these dispersions fully coalesced forming a transparent coacervate phase on the bottom leaving a transparent supernatant solution where few coacervate droplets were still detected. Despite a low interfacial tension between the coacervate droplets and water, [9][10][11] since the droplets are very small the Laplace pressure is high which is consistent with the occurrence of coalescence at pH = 5 and 6. From pH = 7 to 10, a white dispersion concentrated in coacervate droplets sediments slowly but complete coalescence is not achieved. Moreover, it easily re-disperses recovering the white dispersion obtained after preparation. Therefore the stability to coalescence of the coacervate droplets at high pH is considerably improved. It may be that the coacervate droplets at high pH possess excess of charged polymer at their interfaces which act to prevent their coalescence.
Intrigued by the high stability of coacervate droplets at pH = 10, we wondered if they were homogenous from their core to the surface or if a distinct surface layer of mixed polymers formed around the core. TEM images of the coacervate droplets obtained from a fresh aqueous PEC dispersion of composition xPAANa ≈ 0.5, [PEL] = 5 g L -1 at pH = 10 are shown at different magnifications. Figure 7 shows different areas of the unstained sample, whereas Figure S5 corresponds to the sample negatively stained with uranyl acetate. The uranyl acetate crystals surround the structures giving a 3D appearance. In all cases, near spherical droplets ranging in diameter between 200 nm and 500 nm are observed, although some are distinctly non-spherical.
Since water evaporates from the droplets during preparation, the size of the remaining concentrated polymer droplet will be less than that in the original dispersion (fully hydrated).
The droplets are homogeneous in texture and exhibit no internal ordering. For the unstained case, the droplets appear to be surrounded by very small particulates (black dots in Figure 7).
For the stained sample, these particulates are not visible as uranyl acetate crystals surround the droplets. In general, the coacervate droplets have the same appearance as those observed with optical microscopy (Figure 4(c)) despite the latter being of larger size. The optical microscope image was obtained around 1 hr after sample preparation where it is likely that some droplet coalescence has occurred. Additional TEM images of coacervate droplets are shown in Figure   S6. Using EDX analysis, no special element was detected. In fact the polyelectrolytes are composed mainly of C and in lower proportion N, O, Na and Cl. Since the TEM grid is carbon coated, carbon was one of the main peaks displayed in the spectra (data not shown). The small particulates have not been identified. Using this technique, we conclude that the coacervate droplets are homogeneous and do not appear to possess any internal structure or surface shell.
We confirm that their size falls in the nm range and that they are spherical in shape. However, the reason for their coalescence stability at high pH remains to be established. Few TEM images of the coacervate phase in mixed polymer systems exist in the literature. 39,40 They also display a homogeneous texture without internal ordering or evidence of a surface shell or membrane.

Preparation and characterisation of oil-in-water emulsions
Our interest here is whether the PEC prepared in water are surface-active enough to adsorb to an oil-water interface created on emulsifying the aqueous phase with the non-polar alkane dodecane. In our earlier work in the system PDADMAC + PSSNa, 22 the oil-in-water emulsion stability was enhanced by increasing the initial concentration of PEL. Moreover, on varying xPSSNa, the emulsion droplet size and the extent of coalescence were reduced upon approaching the conditions leading to charge neutrality. In that case, the oil was added in one step during homogenisation. As identified in the previous section, for the system PDADMAC + PAANa, pH has a dramatic influence on the associative phase separation that takes place. Therefore, this parameter will be a key one in the study of emulsion stabilisation. Figure S7 shows the appearance of the emulsions obtained from the aqueous PEC dispersions prepared at pH values from 2 to 6 from 5 g L -1 PEL solutions at xPAANa = 0.5. At pH = 2 (coacervate), an o/w emulsion was formed after mixing which creamed rapidly and completely phase separated within 11 days after preparation. An air-in-water foam was also formed during mixing which was only stable for a few min. This surface activity is most likely connected with the presence of fully protonated poly(acrylic acid). In order to verify this assumption, emulsions containing the individual PEL solutions at the same concentration and pH were prepared with dodecane. That with PDADMAC coalesced rapidly after preparation until complete phase separation whilst emulsion formation with PAANa produces considerable amount of foam and the emulsion was stable for a longer period of time compared to that formed from the aqueous PEC dispersion. This is consistent with the findings of Ishimuro and Ueberreiter 41 who showed that the airwater surface tension was lowered at low pH on adding PAA (uncharged) whereas PAA was not surface-active at higher pH (charged). Further investigation of the ability of the PAANa solution at pH = 2 to stabilise emulsions was carried out. Solutions of PAANa at different concentrations (0.05 to 5 g L -1 ) were prepared and emulsions were obtained following the standard procedure. As seen in Figure S8, o/w emulsions prepared at 0.05 and 0.10 g L -1 coalesced completely in less than 10 min after preparation and no air bubbles persisted during high-shear homogenisation. On the other hand, emulsions prepared from solutions at higher concentration (> 0.50 g L -1 ) generated bubbles while mixing and a stable emulsion cream was achieved for a longer period of time despite some coalescence of oil. The concentration of PAANa required to display surface activity therefore must be > 0.10 g L -1 .
At pH values of 3 and 4 where coacervate droplets and precipitate particles co-exist in water, no stable emulsions were obtained either ( Figure S7). The reasons may be that the number of particles and coacervate drops is not sufficient for stabilisation or that particles/aggregates are too large to remain attached to droplet interfaces or that they are too hydrophilic in nature. In order to increase the overall particle concentration, aqueous PEC dispersions were prepared from the PEL solutions at 10 g L -1 . However, this was not sufficient as the emulsion completely coalesced after homogenisation (data not shown). At pH 5 and 6 (coacervate), an unexpected situation was encountered. At these pH aqueous PEC dispersions are the most turbid. However, after homogenisation of oil, the aqueous phase became almost transparent at pH = 6 and no emulsion was obtained. It was noticed that the coacervate droplets coalesced on the head of the homogeniser and did not take part in emulsion stabilisation. At pH = 5, an initial viscous emulsion collected on the top of the vial; it was not stable for a long period of time however. In this case the turbidity of the separated aqueous solution decreased as well due to the coalescence of some of the coacervate phase on the head of the homogeniser.
The highly viscous nature of these particular coacervate systems makes it difficult with respect to emulsification.
For pH values between 7 and 10 (coacervate), the behaviour was similar so the discussion will be focused on the emulsion prepared at pH = 10. Emulsions prepared with the individual PEL were not stable as complete phase separation was obtained after homogenisation ( Figure S9). For the PEL mixture, two different preparations were evaluated (o = 0.2): addition of oil in one step and addition of oil stepwise summarised in Figure 8. For the emulsion prepared with the addition of oil stepwise, the oil was added in aliquots of 200 L and homogenised with the Ultra-turrax homogeniser each time. Between each addition, the emulsion was left to stand for a few hours to assess its stability. In the case of the addition of oil in one step, the emulsion was virtually completely phase separated in less than 1 day. The extent of coalescence was high but there was a residual volume of emulsion cream around the walls of the vial. For the emulsion prepared stepwise, the emulsion cream was stable for a longer period of time. The final emulsion containing ϕo = 0.20 liberated around 10% of oil by coalescence but was subsequently stable for at least 6 months (Figure 8(b)). The viscosity of the emulsion increased with an increase in o as expected. Since the average drop diameter was independent of o (between 24 and 28 m), it implies that the additional oil does not swell existing drops stabilised by coacervate but is used in creating new drops which become stabilised. In order to compare the one step and multistep protocols more correctly, an extra experiment was performed. Oil was added in one step and then five homogenisations were carried out leaving the emulsion to stand for 1 hr between each step. No stable emulsion could be prepared either with this modified procedure. This strongly highlights the importance of the addition of oil stepwise in obtaining a stable emulsion.
Forgiarini et al. 42 studied the system water/nonionic surfactant/decane to obtain nanoemulsions by three different emulsification methods: (i) stepwise addition of oil to a watersurfactant mixture, (ii) stepwise addition of water to an oil-surfactant mixture and (iii) mixing of all components simultaneously. Although the nature of the emulsifier is different to the one studied here, nanoemulsions with high kinetic stability were only obtained by method (ii). In our case, the reason why a stable emulsion can only be prepared with the addition of oil stepwise is not fully understood. On the one hand, the kinetics of adsorption of the coacervate phase onto the oil-water interface could be relatively slow compared to that of surfactant molecules. Therefore, the subsequent addition of oil followed by homogenisation may enable a more homogeneous coating around oil drops to be formed. In fact it is common during industrial manufacture of emulsions that the dispersed phase (here oil) is drip fed into the reactor during mixing as this allows good dispersion and stabilisation in a more controlled manner.
Optical microscope images were obtained by placing a sample of fresh emulsion on a glass slide without a coverslip (Figure 9(a)). As seen in the left hand image, the central group of droplets are enveloped by a film which is most likely that of coacervate. On the right hand image, concentric spheres are visible and in some cases the inner sphere possesses nodules or horns. These kind of structures were reported in a study of microencapsulation in which oil droplets were surrounded by the coacervate complex prepared between a protein and a polysaccharide. 43 Cryo-SEM images of a selected emulsion (Figure 9(b)) show the curvature of the oil-water interface with the interior of the oil droplet on the left and the adjacent continuous phase on the right. Both the interface and the aqueous phase contain frozen, In order to enhance the emulsion stability, an aqueous PEC dispersion prepared from 30 g L -1 PEL solutions at pH = 10 was prepared ( Figure S10(a)). The dispersion exhibits similar turbidity to the one prepared from 5 g L -1 PEL solutions. However, a viscous coacervate phase formed on the bottom of the vial, around the walls and on the magnetic stirrer bar as seen in the inverted vial ( Figure S10(d)). Spherical coacervate droplets were visible in the dispersion ( Figure S10(c)). However, upon addition of oil and homogenisation, the head of the homogeniser was covered by the viscous coacervate phase and the oil on the top was mixed with it without being emulsified ( Figure S10(b)). The aqueous phase after homogenisation was significantly depleted of coacervate droplets ( Figure S10(e)). Therefore, systems exhibiting complex coacervation are very sensitive and the stabilisation of emulsions from them has to be optimised for each individual pair of polyelectrolytes.
As the most promising results on emulsion stabilisation were achieved with the aqueous PEC dispersion prepared from 5 g L -1 PEL solutions (xPAANa = 0.5) by stepwise addition of oil, we evaluated the emulsion stability in the case of different oils. We observed different behaviour depending on the type of oil. Emulsions of squalane and isopropyl myristate completely phase separate in less than 1 hr after preparation ( Figure S11). That of PDMS also coalesces 1 day after preparation but the optical microscope images once creaming had halted reveal the same kind of behaviour as with dodecane: oil droplets surrounded by a coacervate phase ( Figure S12). For these three oils, a viscous phase remained at the oil-water interface and around the walls after the majority of oil had coalesced. For toluene, the emulsion was much more stable to coalescence compared with the other oils and optical microscopy reveals a different morphology of the oil droplets ( Figure S13). Here, the droplets are more deformed and stretched and a large number of them become encased by the coacervate phase.

Determination of spreading coefficients
In microencapsulation, complex coacervates are used to encapsulate hydrophobic core materials (oil drops or solid particles). In order to form core/shell drops (as here), the coacervate In order to evaluate whether our experimental findings are in line with theoretical predictions, S1, S2 and S3 have been calculated from knowledge of the respective interfacial tensions as explained earlier. For these, was 71.9 ± 0.2 mN m -1 at 25 °C, in good agreement with the best literature value. 44 The value of for each oil is shown in  Table 3 together with the air-liquid contact angles on a glass slide covered with the coacervate phase. The least squares calculation was carried out to determine the best combination of and that fits all the data simultaneously. The 3-D surface energy diagram so obtained is shown in Figure 10. , the values of S1-3 can be calculated for each oil and these are also given in Table 2. For dodecane and toluene, the combination of the three spreading coefficients fulfil the condition for complete engulfing observed experimentally. However, for isopropyl myristate and squalane the predicted morphology is complete engulfing although the emulsion was destabilised in less than 2 hr. For PDMS, even if the optical microscope images of the freshly prepared emulsion showed oil droplets surrounded by the coacervate phase, the emulsion was broken 1 day after preparation. In these cases the prediction does not reflect the observed morphology. We note that agreement between predicted and observed morphology is not always fulfilled. 15,47,48 Torza and Mason 15 found discrepancies in systems when S2 or S3 were quite small (related to the errors in measuring interfacial tensions). Tasker et al. 48 also noted the discrepancy between the predicted and observed morphologies when the S2 values were close to 0. We accept that our analysis uses as substrate a dried coacervate phase which may possess a surface energy different to that of a hydrated one formed in situ in water.

CONCLUSIONS
In aqueous mixtures of a strong (PDADMAC) and a weak ( This could be a result of their relatively low amount, their considerable size or their intrinsic hydrophilicity. By contrast, at higher pH between 7 and 10, stable dodecane-in-water emulsions could be formed from the coacervate phase of near neutral charge if oil is added sequentially. Oil droplets coated by the coacervate phase are observed. The morphology of the oil droplets coated by the coacervate phase is compared with theoretical predictions using equilibrium spreading coefficients for a range of oils. Despite the agreement for dodecane and toluene (complete engulfing), a discrepancy is found for the other oils. It is suggested that kinetic aspects linked to the viscosity of the coacervate phase play a role in the encapsulation process which are not accounted for in equilibrium spreading coefficients. From these results and those in our earlier work, 22