8630
Langmuir 2005, 21, 8630-8634
Interaction of Nonionic Surfactants with Copolymer Microgel Particles of NIPAM and Acrylic Acid Melanie Bradley and Brian Vincent* School of Chemistry, University of Bristol, Cantock’s Close, Bristol, BS8 1TS, United Kingdom Received February 8, 2005. In Final Form: July 11, 2005 Binding of the nonionic surfactants Triton X-100 and Triton X-405 onto linear copolymers of N-isopropylacrylamide (NIPAM) and acrylic acid and to cross-linked microgel particles of similar composition but differing in their cross-link densities has been studied. The binding capacities vary for each of these polymeric systems, being smallest for the linear copolymer. The binding is also significantly less in all cases for the more hydrophilic surfactant, namely, Triton X-405. By comparing estimates of the pore or “cage” size within the microgel particles with the dimensions of the free micelles in solution, it is concluded that micelles of Triton X-100 form within the microgel particles more readily for the lower cross-linked microgel particles. However, micelles do not form as easily inside either microgel for Triton X-405. The swelling/ deswelling behavior of each of the two microgels, in the presence of the surfactants, has been explained in terms of their relative binding behavior and how this contributes to the osmotic pressure difference inside and outside the microgel particles and also in terms of micelle “bridging” of the polymer network, causing shrinkage.
Introduction Polymers and surfactants are present together in many formulations, for example, in foods, surface coatings, pharmaceuticals, and cosmetics. Therefore, understanding the interaction between surfactants and polymers in solution is important from both a theoretical and an industrial viewpoint. The interaction of surfactants with water-soluble polymers has been extensively studied.1 At low concentrations of surfactant (for a given polymer concentration), beyond the critical aggregation concentration (cac), isolated surfactant molecules attach to each polymer chain. However, at higher surfactant concentrations the surfactant molecules bind as aggregates to the polymer chains. These aggregates are different to micellar aggregates in free solution (for example, the aggregation numbers are, in general, lower); they are often depicted as a “string of pearls” along the polymer backbone. The cac value is, in general, lower than the conventional critical micelle concentration (cmc) value for surfactants in bulk solution. The initial interaction of (isolated) surfactant molecules with a polymer chain in solution, i.e., just beyond the cac value, depends on the structures of both the surfactant and the polymer chain. If the polymer contains hydrophobic moieties, then clearly hydrophobic interactions between the two species play a role since all surfactants contain hydrophobic tails. If the surfactant headgroup is charged and if the polymer is a polyelectrolyte of opposite charge, then Coulombic attraction forces play a strong role. If the surfactant headgroup is nonionic, then binding will only occur if the polymer contains hydrophilic groups to which the surfactant headgroup can H-bond. The interaction between surfactants and microgel polymer particles has also attracted much attention in the past decade or so. A microgel particle is essentially a cross-linked polymer network which is swollen in a good solvent environment for the polymer concerned. The maximum extent of swelling in a good solvent depends on the fraction of cross-linking units present in the network. (1) Interactions of Surfactants with Polymers and Proteins; Goddard, E. D., Ananthabadmanabham, K. P., Eds.; CRC Press: Boca Raton, FL, 1993.
The size (extent of swelling) of the microgel particles can be changed by varying the local thermodynamic conditions, e.g., solvent composition, temperature and (for polyelectrolyte-based microgels) pH, and ionic strength. The majority of the reports in the literature investigating the interaction between surfactants and microgel particles have focused on poly(N-isopropylacrylamide) (PNIPAM) microgels interacting with anionic, cationic, or nonionic surfactants.2-6 However, there are some reported studies with other systems, for example, (quaternized) polyvinylpyridine microgel particles with SDS.7 In general, it has been shown that anionic surfactants have a larger effect on PNIPAM microgel properties than cationic surfactants. In the presence of sodium dodecyl sulfate (SDS), the size of PNIPAM microgel particles was shown to increase due to SDS aggregate formation within the microgel network, causing internal electrostatic repulsion within and hence expansion of the polymer network.4,6 This also results in the conformational transition temperature of the microgel particles shifting to higher temperatures. At high concentrations of SDS the microgel particle size becomes nearly independent of temperature since electrostatic effects now dominate. The electrophoretic mobility of the particles increases with increasing SDS concentration and temperature. This is due to greater SDS binding to the microgel at higher temperatures. The cationic surfactants dodecylpyridine bromide and dodecyltrimethylammonium bromide were not observed to raise the conformation transition temperature of PNIPAM microgel particles in the way that SDS does.2,3 (2) Tam, K. C.; Ragaram, S.; Pelton, R. H. Langmuir 1994, 10, 418. (3) Wu, C.; Zhou, S. J. Polym. Sci., Part B: Polym. Phys. 1996, 34, 1597. (4) Mears, S. J.; Deng, Y.; Cosgrove, T.; Pelton, R. Langmuir 1997, 13, 1901. (5) Wang, G.; Pelton, R.; Zhang, J. J. Colloid Surf. A: Physiochem. Eng. Aspects 1999, 153, 335. (6) Woodward, N. C.; Chowdhry, B. Z.; Leharne, S. A.; Snowden, M. J. Eur. Polym. J. 2000, 36, 1355. (7) Crowther, H. M.; Morris, G. E.; Vincent, B.; Wright, N. G. The Role of Interfaces in Environmental Protection; Barany, S., Ed.; NATO Science Series; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2003.
10.1021/la0503597 CCC: $30.25 © 2005 American Chemical Society Published on Web 08/19/2005
Interaction of Nonionic Surfactants
Langmuir, Vol. 21, No. 19, 2005 8631
Table 1. Quantities Used in the Preparation of the Microgel Particles reagent
mass/g
water NIPAM (main monomer) BA (cross-linking monomer) AAc (comonomer) potassium persulfate (initiator)
500 2.5 0.15-0.25 0.25 0.25
It was speculated3 that the surfactant headgroup interacts with the nitrogen lone pairs on the polymer chains, and this reduces the H-bonding between the microgel and water, effectively making water a poorer solvent for the microgel network. This conclusion is supported by the observations that the cationic surfactant leads to actual shrinking of the microgel network at room temperature and that aggregation of the microgel particles occurs at high temperature. The influence of the nonionic surfactant Triton X-100 has been found not to alter either the extent of swelling or the transition temperature range of PNIPAM microgel particles.2 However, the addition of Triton X-100 was observed to cause a decrease in the size of poly(acrylic acid) (PAAc) microgel particles (through an observed decrease the intrinsic viscosity of the dispersion). In the present study the association of each of the two nonionic surfactants, Triton X-100 and Triton X-405, with PNIPAM-co-AAc microgel particles has been investigated. Changes in the size of the microgel particles, as a function of surfactant concentration, have been determined and the surfactant binding isotherms established for two sets of microgel particles with different cross-linker concentrations. A comparison is also made between the association of the surfactants with the microgel particles and with the corresponding, linear copolymer chains, i.e., having the same AAc content. Experimental Section Materials. N-Isopropylacrylamide (NIPAM) and N,N′-methylenebisacrylamide (BA) (both from Fisher) were recrystallized from hexane and methanol, respectively. Acrylic acid (AAc) and potassium persulfate (KPS) were used as received from Fisher without further purification. Milli-Q grade water was used throughout the work. Two octylphenyl ethoxylate (strictly, p-(1,1,3,3-tetramethylbutylphenyl ethoxylate) nonionic surfactants Triton X-100 (E ) 9.5) and Triton X-405 (E ) 35), where E is the average number of ethylene oxide repeat units per molecule, were used as received from Aldrich. Synthesis of the Microgel Particles. Single-step emulsion polymerizations were carried out in a 500 mL reaction vessel fitted with a reflux condenser, an overhead stirrer, a thermometer, and a glass nitrogen inlet tube. The vessel was immersed in an oil bath set at 70 °C. The quantities of the reactants and solvent used are given in Table 1. The reaction mixture was purged with nitrogen for 20 min, and once the reaction temperature had been reached, the initiator solution was added. The polymerization reaction was allowed to continue for 24 h. The microgel dispersion was then allowed to cool and subsequently dialyzed against Milli-Q water to remove any unreacted monomer and other impurities, changing the dialysate twice daily for a week. The linear copolymer was synthesized using the apparatus and method above. However, the cross-linking monomer was omitted from the reactants listed in Table 1. Characterization of the Microgel Particles. The solids content and, hence, yield of the various microgel preparations were determined gravimetrically. When required, the pH was adjusted by the addition of small quantities of 0.1 M HCl or NaOH. The diffusion coefficients of the various microgel particles were determined by photon correlation spectroscopy (PCS) using a Brookhaven Instruments Zeta Plus dynamic light scattering apparatus fitted with a 15 mW laser (λ ) 678 nm) and the detector set at 90°. The Stokes-Einstein equation was then used to
calculate the hydrodynamic diameter of the particles. This required input of the viscosity of the PEO solution used at the appropriate concentration and temperature. These values were obtained using a standard capillary viscometer. The size of the microgel particles is expressed here in terms of the swelling ratio, S, defined as S ) (d/do)3, where do is the diameter of the microgel particles in water at 20 °C. The number of carboxyl groups per microgel particle was calculated from potentiometric titration data. These titrations were carried out at room temperature in a vessel that supported a pH electrode. The microgel dispersion was stirred at low speed while the titrant (1 mM NaOH) was added. The binding isotherm for each surfactant to the microgel particles was determined using the following (solution depletion) method. A given quantity of microgel dispersion was added to a known amount of surfactant solution, and the system was left to equilibrate for at least 7 days. The supernatant was then separated by centrifugation at 10 000 rpm for 20 min using a Sorvall RC 5b Plus centrifuge. The equilibrium surfactant concentration was then determined from a calibration graph of absorbance, at 275 nm, versus surfactant concentration.
Results Titration Data. The number of carboxyl groups on the linear copolymer and microgel particles was calculated from the potentiometric titration curves shown in Figure 1. There are approximately 8 × 1020 COOH groups/g and 7 × 1020 COOH groups/g for the copolymer and microgel, respectively. Swelling Ratio of the Microgel Particles. Two batches of microgel particles were prepared which differed in their cross-linker concentration, 6 and 10 wt %, but had the same AAc content, 10 wt %, based on the total monomer concentration. The yields determined gravimetrically of the 6 and 10 wt % cross-linked microgel particles were 96% and 98%, respectively. The particle number concentration (N) was determined to be 2.58 × 1011 and 1.80 × 1011 mL-1, respectively, for the 6 and 10 wt % cross-linked microgel dispersions. The effect of the nonionic surfactants Triton X-100 and Triton X-405 on the hydrodynamic diameter of each batch of microgel particles was investigated. The results for Triton X-100 may be seen in Figure 2, where the swelling ratio is shown as a function of the surfactant concentration. Figure 2a shows the data, at pH 3, for the two microgels in the region of the cmc for Triton X-100 (cmc ≈ 0.1 g L-1 8). The swelling appears to be greater at a given concentration of surfactant for the lower cross-linked microgel particles, as would be expected. Also, for both microgels there appears to be a minimum surfactant concentration, only above which any increase in swelling ratio is observed; this would correspond to the cac, which is around or perhaps just below the cmc value. Figure 2b shows data over a much larger concentration range (up to 200× the cmc value) for the two microgels at pH 3 and also additional data for the 10% cross-linked microgel particles at pH 8. At pH 3 the swelling ratio goes through a maximum with increasing surfactant concentration, with the maximum occurring at ∼10-20× the cmc value. Thereafter, the swelling ratio decreases, eventually to values less than the swelling ratio in water alone. For the microgel at pH 8 the swelling ratio just decreases steadily with increasing surfactant concentration over the whole concentration range. Figure 3 shows similar plots to those in Figure 2 for the two microgels at pH 3 but for Triton X-405 rather than Triton X-100. The cmc for Triton X-405 is ∼1.6 g L-1,8 i.e., significantly higher than that for Triton X-100. As with (8) Wolszczak, M.; Miller, J. J. Photochem. Photobiol. A: Chem. 2002, 147, 45.
8632
Langmuir, Vol. 21, No. 19, 2005
Bradley and Vincent
Figure 1. Potentiometric titration curves for (9) water and (0) linear PNIPAM-co-AAc (10 wt %) polymer and PNIPAM-co-AAc (10 wt % AAc) microgels at different cross-linker concentrations ((2) 6 and (b) 10 wt %).
Figure 3. Effect of cross-linker concentration on the swelling ratio as a function of Triton X-405 concentration, at pH 3 and 20 °C, for PNIPAM-co-AAc (10 wt % AAc) microgels at different cross-linker concentrations ((2) 6 and (b) 10 wt %).
Figure 2. Effect of cross-linker concentration on the swelling ratio as a function of Triton X-100 concentration, at 20 °C, for PNIPAM-co-AAc (10 wt % AAc) microgels at different crosslinker concentrations (wt %) and pH values ((2) 6 wt %, pH 3; (b) 10 wt %, pH 3; (O) 10 wt % and pH 8): (a) low surfactant concentrations; (b) higher surfactant concentrations.
Triton X-100 there is very little (if any, within the experimental error) increase in the swelling ratio above that for water alone up to around the cmc. Thereafter, within experimental error the swelling ratio essentially only decreases, that is, there is no obvious swelling maximum in this case with increasing surfactant concentration, as is the case for Triton X-100.
Binding Isotherms. The binding isotherms for Triton X-100 onto each of the two types of microgel particles, that is, with different cross-linker concentrations, are shown in Figure 4. In Figure 4a the binding isotherms for the microgels at pH 3 are shown at Triton X-100 concentrations in the region of the cmc (∼0.1 g L-1). Below the cmc there is only a very small amount of binding if any within experimental error of the surfactant molecules to the microgel particles. Above the cmc the bound amount increases gradually. In Figure 4b the binding isotherms, over a much wider concentration range of Triton X-100, are shown for the two microgels at pH 3. Data are also shown for the linear NIPAM-AAc (10 wt %) copolymer. Clearly, the binding of the surfactant is greater to the 6% cross-linked microgel particles than to the 10% cross-linked microgels particles, but the binding is significantly less to the linear copolymer than to either microgel. The corresponding binding isotherms for Triton X-405 to the two microgels and to the linear copolymer are shown in Figure 5. By comparing the results shown here with those in Figure 4b it can be seen that the binding of this surfactant to the microgel particles and the linear copolymer is now very much lower than that for Triton
Interaction of Nonionic Surfactants
Figure 4. Surfactant binding isotherms for Triton X-100, as a function of equilibrium surfactant concentration, at pH 3 and 20 °C, onto the PNIPAM-co-AAc (10 wt % AAc) microgels with different cross-linker concentrations ((2) 6 and (b) 10 wt %) and (0) onto a linear PNIPAM-co-AAc (10 wt % AAc) copolymer: (a) low surfactant concentration; (b) higher surfactant concentration.
Figure 5. Surfactant binding isotherms for Triton X-405, as a function of equilibrium surfactant concentration, at pH 3 and 20 °C, onto the PNIPAM-co-AAc (10 wt % AAc) microgels with different cross-linker concentrations ((2) 6 and (b) 10 wt %) and (0) onto a linear PNIPAM-co-AAc (10 wt % AAc) copolymer.
X-100. Also, in contrast to Triton X-100, the binding of this surfactant to the 6% and 10% cross-linked microgel particles is broadly similar given the scatter in the data points. The binding of Triton X-405 to the linear copolymer is again lower than to either microgel. Discussion The aqueous solution behavior of Triton X-100 has been extensively studied.8-10 The phase diagram of the binary Triton X-100/water system, at 25 °C, shows two isotropic solution phases L1 (micelle phase) and L2 (reversed
Langmuir, Vol. 21, No. 19, 2005 8633
micelles) separated by a hexagonal liquid crystalline phase (38.5-53.5 wt % surfactant).9 Over the concentration regime investigated in this study (up to 20 g L-1 or 2 wt % surfactant) the solutions are in the L1 micellar solution phase. Similar phase information for Triton X-405 is not available in the literature, as far as we are aware. However, it seems reasonable to assume that the concentration range of Triton X-405 studied here also corresponds to the micellar solution phase. As shown in the Results section, interaction of the PNIPAM-co-AAc microgel particles with Triton X-100 and Triton X-405 gives rise to very different swelling and binding behavior. These two systems are therefore discussed separately below. Triton X-100. As stated in the Introduction, nonionic surfactants do not, in general, interact with polymers except when H-bonding between them can occur. From the swelling results shown in Figure 2b it can be concluded that this would appear to be the case here: swelling of the microgels occurs at pH 3 but not at pH 8. At pH 3 H-bonding is known to occur between the ether oxygens in the headgroup of the surfactant and the protons on the carboxylic acid group of the AAc moieties in the polymer chains.10-12 At pH 8 the AAc moieties are dissociated, preventing this H-bonding. Figure 4b shows that binding of Triton X-100, at pH 3, to the different polymer systems decreases in the order 6% microgel > 10% microgel > linear copolymer. If H-bonding is responsible for the surfactant binding to the polymer, then the amount of surfactant binding should be related to the number of carboxylic acid sites available. The number of carboxyl sites for each system was calculated from the potentiometric titration curves shown in Figure 1. There are ∼8 × 1020 COOH groups/g for the copolymer and ∼7 × 1020 COOH groups/g for each of the microgels. For the linear copolymer the maximum number of bound Triton X-100 molecules, calculated from the plateau concentration in the binding isotherm, is ∼1 × 1020 molecules/g. If the surfactant is binding to the polymer chains as micelles, then this means that only a small fraction of the available -COOH sites are being utilized as binding sites. This could simply be due, in part, to steric repulsion between the bound micelles along the chain. At the same time it has to be remembered that previous studies of the interaction of surfactants, both with polyelectrolytes and microgels, have concluded that, in general, the aggregation numbers of bound micelles are reduced compared to those in free solution.4,12,13 The question arises as to why the surfactant binding is greater to the microgels than the copolymer? At the same time one has to consider why the swelling ratios of both sets of microgel particles exhibit a sharp maximum at a Triton X-100 concentration of ∼1-2 g L-1 (particularly for the 6% cross-linked microgel), as shown in Figure 2b. The most likely explanation is that the micelles are absorbed into pores or “cages” within the cross-linked polymer network which forms the structure of each microgel particle.14-16 This would imply that each micelle (9) Galatanu, A. N.; Chronakis, I. S.; Anghel, D. F.; Khan, A. Langmuir 2000, 16, 4922. (10) Saito, S. J. Colloid Interface Sci. 1993, 158, 77. (11) Anghel, D. F.; Saito, S.; Iovescu, A.; Baran, A. Colloids Surf. A: Physicochem. Eng. Aspects 1994, 90, 89. (12) Vasilescu, M.; Anghel, D. F.; Almgren, M.; Hansson, P.; Saito, S. Langmuir 1997, 13, 6951. (13) Voisin, D.; Vincent, B. Adv. Colloid Interface Sci. 2003, 106, 1. (14) Murase, Y.; Tsujii, K.; Tanaka, T. Langmuir 2000, 16, 6385. (15) Mounir El Sayed, A.; Darwish, N.; Shehata, A. J. Appl. Polym. Sci. 2004, 91, 3921. (16) Mounir El Sayed, A.; Kawasaki, H.; Maeda, H. J. Appl. Polym. Sci. 2004, 93, 2001.
8634
Langmuir, Vol. 21, No. 19, 2005
Bradley and Vincent
is bound to several COOH sites rather than to single sites, as is likely to be the case with the linear copolymer chains. This would also explain the shrinkage of the microgel particles at higher surfactant concentrations since the micelles are essentially bridging the chain network together within each microgel particle. In this connection we have attempted to calculate the relative sizes of the cages and compare these values to the known sizes of Triton X-100 micelles (although in free solution). According to Robson and Dennis the Triton X-100 micelle in dilute aqueous solution has an oblate ellipsoid shape with semiaxes of 5.2 and 2.7 nm;17 the aggregation number has been quoted as 111.8 An approximate value of the pore size was estimated for both microgels from the known percentage of cross-linking monomer in each one (assuming that all of the monomer molecules present in the reaction are incorporated) and assuming, for simplicity, that cubic pores are formed within the polymer network. These pore size calculations for the 6 and 10 wt % crosslinker concentrations gave 4.1 and 2.4 nm, respectively, for the pore lengths. It must be remembered that the pore size is unlikely to be uniform across the microgel particles. However, these calculations would lend credence to the fact that micelles can just fit within the pores and that this process is easier for the lower (i.e., the 6%) cross-linked microgel particles compared to the corresponding 10% ones. This is the most likely explanation as to why the binding of Triton X-100 is greater for the 6% cross-linked microgel particles, as shown in Figure 4b. It remains to explain why the microgel particles initially swell at low surfactant concentrations of Triton X-100, i.e., prior to the maximum (Figure 2). One has to consider the osmotic pressure balance inside and outside the microgel particles. At low surfactant concentrations (but beyond the cac) most of the surfactant becomes bound to the polymer chains inside the microgel particles and the osmotic pressure inside the microgel particles increases, inducing an influx of water molecules, leading to swelling. However, eventually the concentration of surfactant inside the micelles reaches a maximum, and then the concentration of the free surfactant outside increases relative to that inside. The surfactant contribution to the osmotic pressure difference, i.e., between the inside and outside of the microgel particles, decreases, and deswelling will commence. This has been discussed before by us18 in the context of the absorption of homopolymers of poly(ethylene oxide) (PEO) into similar microgel particles. However, in the present work the maximum in the swelling ratio (Figure 2b) occurs at only ∼1-2 g L-1 Triton X-100 and, as can be seen from Figure 4b, the binding of the surfactant is still increasing beyond 8 g L-1. Hence, it must be that the micellar “bridging” mechanism, referred to earlier, is in this case the dominant mechanism for microgel shrinkage beyond the maximum in the swelling ratio (Figure 2b). At pH 8 (Figure 2b), where there is very little if any binding of the surfactant, the concentration of free surfactant just increases steadily and so, therefore, does the osmotic pressure of the solution. Hence, there will be net movement of water molecules out of the microgel particles, leading to monotonic deswelling of the microgel particles.19,20
Triton X-405. Comparing Figure 5 with Figure 4b, it can be seen that the binding of Triton X-405 is much less than that for Triton X-100 for both microgels and for the linear copolymer at similar wt/vol. concentrations of surfactant. The lower binding of Triton X-405 to the linear copolymer, compared to Triton X-100, simply reflects the greater hydrophilicity of the former molecule (as is also seen in their relative cmc values, see earlier). To the best of our knowledge the micelle dimensions of Triton X-405 have not been investigated. However, Triton X-405 is a much larger molecule than Triton X-100 (3 times as many EO units per molecule in the headgroup). It could well be that micelles of Triton X-405 are not able to form easily within the cross-linked polymer network of the microgel particles (except perhaps nearer the periphery of the particles, where the pores are likely to be larger). In that case most of the binding of Triton X-405 within the particles could predominantly be in the form of surfactant monomers. As mentioned previously, given the large scatter in the data points in Figure 5, one could conclude that there is actually little difference in the absorption of Triton X-405 onto the two microgels. This again would lend some support to the argument that the Triton X-405 is absorbing largely as monomers inside the microgel particles. Such bound monomers would not contribute much to the internal osmotic pressure inside the microgel particles. Hence, as the concentration of free surfactant increases on the outside of the microgel particles, so the osmotic pressure of the free solution tends to increase relative to that inside, resulting in movement of water molecules from the inside to the outside, i.e., deswelling occurs predominantly over the whole surfactant concentration range, as shown in Figure 3.
(17) Robson, R. J.; Dennis, E. A. J. Phys. Chem. 1977, 81, 1075. (18) Bradley, M.; Ramos, J.; Vincent, B. Langmuir 2005, 21, 12091215.
(19) Saunders, B.; Vincent, B. J. Chem. Soc. Faraday Trans. 1996, 92, 3385. (20) Saunders, B. R.; Vincent, B. Colloid Polym. Sci. 1997, 275, 9.
Conclusions The nonionic surfactants Triton X-100 and Triton X-405 both bind to linear copolymers of N-isopropylacrylamide and acrylic acid and to cross-linked microgel particles of similar composition but differing in their cross-link densities. However, the binding capacities vary for each of these polymeric systems, being smallest for the linear copolymer. The binding is also significantly less, in all cases, for the more hydrophilic surfactant, namely, Triton X-405. By comparing estimates of the pore or “cage” size within the microgel particles with the dimensions of the free micelles in solution, it is concluded that micelles of Triton X-100 form within the microgel particles more readily for the lower cross-linked microgel particles, but micelles do not form as easily inside either microgel for Triton X-405. The swelling/deswelling behavior of each of the two microgels, in the presence of the surfactants, is explained in terms of their relative binding behavior and how this contributes to the osmotic pressure difference inside and outside the microgel particles and also in terms of micelle bridging of the polymer network, causing shrinkage. Acknowledgment. We gratefully acknowledge financial support from the EPSRC through the IMPACT Faraday Partnership (GR/R90086/01). LA0503597
