J. MATER. CHEM., 1994, 4( 5), 729-734 Kinetic and Mechanistic Aspects of Iron(ii) Coordination to Bipyridyl- based Hydrogel Polymer Membranes Andrew L. Lewis* and J. David Miller The Speciality Materials Research Group, The Department of Chemical Engineering and Applied Chemistry, Aston University, Aston Triangle, Birmingham, UK B4 7ET A range of hydrophilic membranes composed of copolymers of 4-methyl-4’-vinyl-2,2’-bipyridyl with 2-hydroxyethyl methacrylate have been synthesized. These membranes readily coordinate iron(ii) from aqueous solution to form the tris(2,2’-bipyridyI)iron(ii)species, but at a rate very much slower than that of the free ligand in solution. Kinetic studies on the rate of development of these colour centres have shown the process to be anion-dominated and pseudo-first order for iron(ii) sulfate and second order for the chloride and perchlorate.The extent of coordination within the membrane is also dependent on the salt used, as the anion influences the concentration of Fe” partitioned within the gel matrix, and ultimately the position of the equilibrium established between the mono and tris complexes. Mechanisms are proposed incorporating the known water-structuring effects that anions impose on this type of hydrogel environment and accounting for the presence of ion pairs and their effect on the molecular reorganisations that are necessary in order for tris complexation to occur. 2,2’-Bipyridyl (bipy), its derivatives and structural relatives, are known to form complexes with a large number of metal ions.’-4 In 1888, Blau first observed and isolated the well known iron(1x) bipyridyl system.’ In aqueous solutions above pH 5, formation of the tris(2,2’-bipyridyl)iron(11)species is rapid, driven by a spin change accompanying the addition of the third ligand.This complex is intensely red in colour, due to a metal-to-ligand charge-transfer transition, which can easily be monitored spectrophotometrically to provide quanti- tative data. In solutions of pHG3, the rate is retarded by protonation of the bipyridyl nitrogens, and data for the rate of complex formation have been shown to fit fourth-order kinetics:2 This suggests a stepwise formation: Fe2++bipy K1 Fe( bipy)2 + (i) K2 Fe( bipy)2’ +bipy-Fe( bipy)?’ (ii) k3 Fe( bipy)? + +bipy CFe(bipy); + (iii) k-3 Steps (i) and (ii) are rapidly established equilibria and step (iii) is the rate-determining step in solutions of pH>,5.The values of the stability constants for the complex formation of the mono, bis and tris species [K1, K2 and K3 (k3)]are 104.3, 103.7and 109.5,re~pectively.~The consequence of these values is that, in solution, the concentration of Fe(bipy);’ is almost always very small compared to those of the mono and tris forms. The excellent ion-binding capability, the ability to stabilise unusual oxidation states and the potential to form complexes with catalytic activity or photoactivity make this ligand a desirable choice for use after its immobilisation into polymers.The preparation of polymers containing bipyridyl groups, and their application in metal-ion-specific absorption studies and as supports for catalytic complexes, are well documented in the literature.“12 However, the majority of hetero-and co-polymers described are hydrophobic and not ideally suited for use in aqueous solution. In earlier paper^'^'' we have described the preparation and some uses of hydrogel copolymer membranes composed of 2-hydroxyethyl methacrylate (HEMA) containing small percentages by weight of 4-methyl-4’-vinyl-2,2‘-bipyridyl (vbipy) and of a cross-linking agent. These materials swell when exposed to water, attain large equilibrium water contents and behave in ways comparable to those of aqueous solutions of the related monomeric organic molecules.We have already reported on studies of the complexes formed when these materials, and related membranes containing vinylpyridine, interact with solutions containing some transition-metal ~a1ts.I~The complexes are very similar to those found in true solution, but the rate at which equilibrium is attained is much slower. Here we report studies of the rate of formation of the [Fe( bi~y)~]” complex centre within membranes immersed in solutions of iron@) salts. The rates are unusually slow, but the main kinetic and mechanistic features of the process have been elucidated. These provide both an appreciation of how the polymer network influences the complexation process and of the nature of the interaction between the permeating ions and the hydrogel environment.Experimental Preparation of the Monomers, Complexes and Copolymers The 4-methyl-4’-vinyl-2,2’-bipyridyl(vbipy) used in this work was synthesized from 4,4’-dimethyl-2,2’-bipyridyl (Aldrich) uia the method of Abruna et and the monomeric Fe(vbipy)g+ complex was prepared by a method analogous to that reported for the tris(bipy) comp1ex.l’ The chelating membranes were fabricated using the glass plate membrane mould described previously.” The appropriate weight percent of vbipy was dissolved in the HEMA monomer, along with 0.5 wt.% azoisobutyronitrile (AIBN) initiator and an amount of ethylene glycol dimethacrylate (EGDM) cross-linker to ensure mechanical integrity. The solution was degassed with dry nitrogen for lOmin, after which it was injected into the membrane mould.The mould was placed horizontally in an oven at 60°C for 3 days, followed by 2 h at 90°C postcure. The resulting membrane was allowed to soak in distilled water for at least 7 days, with frequent changes of water. EGDM concentrations in the range 1-20 wt.% were investi- gated, the result being a reduction in the rate of complexation and the final absorbance value of the membrane as the cross- linked density was increased. We therefore standardised on 1 wt.% EGDM for all further studies. Coordination Studies using Visible Spectroscopy The preparation of copolymer samples as membranes is ideal for spectrophotometric investigation. While there are vari- ations in composition and thickness from one prepared batch to another, they are slight and data obtained from different specimens are reasonably reproducible.The random errors found in the measurement of rate constants are greater than those obtained for solution studies, as can be seen in the tabulated data, but their size is insufficient to mask the underlying trends that arise as experimental parameters are changed. In our estimation, rate data are accurate to f15%. The main experimental problems posed by the use of mem- branes reside in their lack of rigidity. We tried various ways of keeping a piece of membrane upright in a spectrophoto- meter lightpath, but none was completely satisfactory. For the present studies, in which observations need only be taken every few minutes, we found that the most convenient way of taking readings was to immerse the membrane sample in a capped thermostatted vessel containing reagent solution, quickly to remove the membrane when a reading was to be taken and to let it adhere by surface tension to the outer face of a clean spectrophotometric cuvette.Measurements were taken using an SP8-100Pye-Unicam spectrophotometer. This procedure has obvious drawbacks. Studies can only be carried out close to room temperature, with reagents that are not unduly air-sensitive. The advantages are that the technique is simple and that the reagent solution can permeate the mem- brane from both sides, thus reducing the period over which complex formation is controlled by the rate of diffusion of reactant solution into the membrane.In order to keep the visible absorbance within the spectro- photometric range it was necessary to restrict our work to thin membranes made up from a monomer mixture containing 0.5% by weight of the substituted bipyridyl monomer (see Fig. 1). In the swollen membrane this gives a concentration of 0.033 mol of 2,2’-bipyridyl per dm3 of membrane or 0.096 mol dm-3 in the imbibed water, if all the ligand groups are available in that phase. When a 4 cm2 piece of membrane was immersed in 20cm3 of an Fe2+ salt solution, even at Fig. 1 Structure of the poly( HEMA-vbipy) copolymer (m:n=300 :1 in 0.5 wt.% copolymer) J. MATER. CHEM., 1994, VOL. 4 0.01 mol dm-3 concentration there was a large excess of the salt. The total meta1:ligand ratio was ca.26: 1 in that case. Therefore, once the salt concentration in the imbibed water is at equilibrium with that in the external solution, the concentration of Fe2+(aq) in the membrane will be constant. Initially we decided to investigate the rate of development of the colour of the Fe(bipy)g+ centre using a concentration range of 0.01 d[Fe2+]/mol dm-3 <0.48. Since there is ample evidence that the anion is often dominant in determining the properties of the salts in hydrogels, e.g. their permeability coefficients18 we extended our study to three different iron@) salts: the chloride (BDH), the perchlorate (Johnson Matthey) and the sulfate (BDH).Results and Discussion Equilibrium Studies The first parameter we consider here is the ‘infinity reading’, the absorbance value measured when the reaction appears to be essentially complete. In these studies the absorbance, and therefore the fraction of the 2,2’-bipyridyl groups converted into Fe(bipy)g+ centres, was found to vary with the concen- tration of Fe2+ used. It also varies with the anion used. The data are collected in Table 1, together with calculated values derived after making the following assumptions: (a) that the fraction of the 2,2’-bipyridyl groups in the copolymer available for coordination is independent of the Fe2+ concentration, but may vary from one anion to another; (b)that the absorp- tion coefficient at 534 nm for the Fe(bipy)i+ centres will be the same as that found for the copolymer we produced separately from HEMA and [Fe(vbipy),] SO4 monomers1’ (see Table2) and (c) that the only equilibrium affecting the Table 1 Observed and calculated ‘infinity readings’ for the develop- ment of colour at 534nm (the parameters used in the calculations are given at the foot of the columns) ~~~ ~ ~ we2+I/rnol dm-’ FeC12 absorbance obs.calc. Fe(c104)2absorbance obs. calc. FeSO, absorbance obs. 0.01 1.63 2.27 - 1.50 1.76 0.02 - 2.18 1.38 1.40 - 0.04 1.75 2.04 1.23 1.25 2.00 0.08 1.91 1.83 1.06 1.04 1.87 0.12 1.67 1.66 0.95 0.88 1.86 0.16 1.37 1.52 0.68 0.76 2.10 0.20 1.39 1.41 0.63 0.66 1.72 0.24 1.17 1.30 0.53 0.58 2.15 0.32 1.09 1.12 0.53 0.45 2.05 0.40 0.98 0.98 0.3 0.36 1.89 0.48 0.92 0.86 0.33 0.29 1.90 fraction converted: 0.83 0.57 0.66 & 0.05 Kiv 16 8.2 2-200 Table 2 Visible absorption spectral details for vbipy complex formation with Fe” ~~~~ precomplexed vbipy(wt.%) ~ absorbance (at 534 nm) El dm3 mol-’ cm-’ 0.05 0.282 6503 0.1 0.565 6514 0.25 1.420 6548 0.375 2.110 6487 0.5 2.800 6457 Membranes soaked in 0.25 rnol dmP3 FeS04 solution.coordinated from solution absorbance (at 534 nm) 0.206 0.431 1.110 1.700 2.260 €1 %vbipy involved dm3 rno1-I cm-’ in tris formation 6507 73 6539 76 6563 78 6453 81 6434 81 J. MATER. CHEM., 1994, VOL.4 observed absorbance value is 3[Febipy12+F==[Fe( bip~)~]~+ +2Fe2+ (iv) The stability constants for the equilibria involving Fe2+ and free bipyridyl in water are well known.4 They can be used to show that, at concentrations similar to those prevailing within the imbibed water of our swollen hydrogels, the concentrations of the free ligand and of the bis complex are negligible compared to those of the mono and tris complexes, while for the free ligand the equilibrium constant for equilib- rium (iv) is 4 x lo4. In the mechanism proposed below this constant is designated Kiv. When the copolymer is allowed to equilibrate in iron@) sulfate solution, the final recorded absorbance readings show some scatter, but no significant variation with the salt concen- tration is apparent.Therefore, 0.66 & 0.05 of the 2,2'-bipyridyl groups appear to be available for tris complex formation and a lower limit of Kiv>200 can be deduced. When the counter- ion is either chloride or perchlorate a definite variation of absorbance with [Fe"] is seen, which means that Ki, must be considerably smaller in these cases and least-squares fitting gives values of 8.2 and 16 for the equilibrium constant and 0.83 and 0.57 for the available ligand fractions. Given the scatter of the experimental results, the agreement between observed and calculated absorbance values is very good (Fig. 2). The only doubt concerns the values at very low concentrations of FeCl,, where the fit is poor. We believe that the rate of diffusion of salts into the membranes becomes dominant at low concentrations, and unexpectedly low values are then observed because the final equilibrium would only be attained after a longer period has elapsed than we were able to allow in our studies.That is, the final absorbance values had not been reached when we eventually took our readings for experiments at low iron concentrations. This point is amplified below when the kinetic observations are discussed. The nature of the anions involved is known to have a marked impact on the properties of hydrogels exposed to salt solutions and this is attributed to their interaction with the imbibed water, i.e. their structure-making or structure-breaking properties." We note that iron@) salts of the two structure-breaking, singly charged anions exhibit much lower values of Kiv than does the structure-making sulfate.The 3.0 T 2.5 -Fig. 2 Observed [x, Fe(ClO&; 0, FeCl,] and calculated (-) variations of the infinity absorbance for membranes soaked in iron(n) salt solutions 731 variations in the fractions of ligand groups utilised by the salts may correlate with the positions of the anions in the Hofmeister series and may also vary inversely with the changes in the equilibrium water contents of poly-HEMA samples when immersed in salts of these anions." Howerfer, more than three points are needed to prove such relationships, and so the need for further studies with other anions is indicated. If the variation in the final absorbance readings can be attributed correctly to an [Fe2+]-dependent equilibri um, then the attribution can be verified in extra experiments.A change in the Fe2+ concentration made after a reaction system has been left to stand should cause a change in the system's visible absorbance. We carried out experiments in which such changes were used to demonstrate qualitatively the involvement of equilibrium (iv). We did not use them to obtain quantitative support, as the achievement of re-equilibration requires much longer elapsed times than we felt able to use (weeks rather than single days). However, we observed the changes summar- ised in Table 3, in which the absorbance readings obtained for experimental sequences in which membrane samples were immersed in 20cm3 aliquots of Fe(C104)2 solutions for 24 h before their visible absorbances were measured.The mem- branes were then washed and immersed in solutions of different concentrations until the following day. The whole operation was repeated five times. As the concentration of aquated Fe2+ ions is altered, so the concentration of tris- bipyridyl complexes changes in the sense that equilibrium (iv) requires. The absorbances of the specimens on day 5 were observed to change further when the systems were left unchanged in the same salt solution for a further period, showing that the move towards re-equlibrium had not been completed in the 1 day allowed per measurement. The time scales involved prevented us from making further attempts to continue this aspect of the investigations.Kinetic Studies The kinetic behaviour of our copolymer in its reactions with solutions of iron@) salts varies with the anion used, as also does its equilibrium behaviour. Therefore we discuss our results with different salts separately. First we consider the development of the red colour of the tris centre observed when membranes are immersed in solutions of iron@) sulfate. Kinetic observations were made by taking a membrane specimen, immersing it in a thermostatted solution of the chosen iron(1r) salt from which air was excluded to minimise unwanted oxidation. At known time intervals the membrane was removed from the solution and its visible absorbance at 534nm was quickly measured.The specimen wau then returned to the salt solution. In our earliest studies we began by using low-concentration iron(i1) solutions, but with vol- umes sufficient to ensure a large excess of Fe2+ over ligand, but we were unable to find a simple rate equation for the plot of absorbance against time. Up to concentrations of ca. Table 3 Changes in the absorbance measured at 534 nm of membranes after immersion in solutions of Fe( C10,)2 of various concentrations concentrations concentrations decreasing daily increasing dailj day CFe''l/mol dmP3 absorbance w+1/mol dmP3 absorbance ~~~ 0.40 0.31 0.01 1.59 0.20 0.38 0.05 1.50 0.10 0.45 0.10 1.37 0.05 0.54 0.20 1.26 0.01 0.61 0.40 1.( j7 0.01 mol dm-3, the profile indicated the occurrence of two distinct processes.Only when we used an excess of Fe2+ at a concentration 3 0.04 mol dm-3 could we successfully describe the profile from 5 min onwards by a single-termed rate equation. At these concentrations a first-order rate equation provides a good description to at least 80% achievement of the final absorbance, i.e. more than two half-lives. No other rate equation that we investigated offered a comparable standard of curve fitting; a point of note, see below. The results are typically reproducible to f 15% for membrane specimens coming from the same prepared sheet, but less so for specimens from different batches. This level of reproduc-ibility is less than that typifying kinetic studies in homo- geneous solutions, but is sufficient to warrant the deductions made below.In Table4 we quote the rate constants deter- mined for membrane samples immersed in FeSO, solutions over a 12-fold range of concentrations, from 0.48 down to 0.04 mol dm-3. We also quote the datum for 0.01 mol dmP3 Fez+, up to 50% completion. Even with the large potential errors imposed by the limitations of our technique, it is clear that a pseudo-first-order rate equation can be used to describe the data, and that the rate constant shows no clear dependence on the Fez+ concentration in the salt solution. This fact helps us to explain the change in the reaction profile at low concentrations. When a piece of an iron-free membrane is immersed in a solution of an iron@) salt, the salt must first diffuse into the imbibed water of the membrane before it can react with the ligand groups of the copolymer.The rate of the diffusion J. MATER. CHEM., 1994, VOL. 4 process is concentration dependent, unlike the dependence found for complex formation. Therefore at sufficiently low FeS04 concentrations the diffusion of ions into the membrane will become the only rate-determining process. At intermediate concentrations (0.01 mol dm-3) both processes will be import- ant, but at higher concentrations (20.04 mol dm-3) complex formation will be the slowest step. Elsewhere we have already reported detailed studies of the permeation of divalent trans- ition-metal salts through copolymer membranes which exhibit rates of permeation consistent with this explanation.” In Table 4 we also quote the rate constants determined for the formation of [Fe( bipy),12+ centres when iron-free copoly- mer membrane specimens are immersed in solutions of FeC1, or Fe(ClO,),.Most of the points made above still apply, with two notable exceptions. The value of the visible absorbance finally achieved by the membrane is [Fe2+]-dependent, see above, while the rate equation which describes the colour development up to at least 80% completion is now a second- order equation. A first-order equation cannot be used to describe these experimental data. Again at low concentrations diffusion becomes important, and again the computed rate constants appear to be independent of the concentrations of Fe2+ used.When once an iron@) salt, in the concentration range we have used, has diffused into one of our hydrogel membranes essentially all the 2,2’-bipyridyl groups will be present initially as mono complexes. Fig. 3 shows the visible absorption spectra of a membrane exposed to iron(I1) per- chlorate, taken at various intervals of elapsed time. Early on, when the absorbance due to the tris complex is not too Table 4 Rate constants at 25 “C for the formation of the [Fe( bi~y)~]’+ colour centre within the swollen copolymer membranes [Fe2+]/mol cm-3 FeSO,k/w4 min-l FeCl,k/dm3 mol-I min-l Fe(C10412k/dm3 mol -min- 0.01 63 - - 0.04 64 29.6 16.8 0.08 93 28.9 14.4 0.12 101 29.6 6.5 0.16 118 22.9 10.4 0.20 106 23.6 7.4 0.24 66 25.8 11.2 0.32 76 24.9 9.8 0.40 86 23.0 32.2 0.48 69 20.9 19.3 0.3 0.2 8 a % v)13 (d 3.1 500 400 300 600 500 460 wavelengthhm 3.0300 Fig.3 (a)Visible absorption spectra for mono- and tris-(2,2’-bipyridyl) iron@) (after Krumholz2’): (-) Fe(bipy),+; (---) Fe( bipy)<+ (value of E to be multiplied by 20). (b) Visible spectra of early kinetic stages for the complexation of 0.48 mol dmP3 Fe(C104), to 0.5 wt.% vbipy-HEMA copolymer membrane: (i) 10; (ii)15; (iii) 40; (iv) 120; and (v) 210 min. J. MATER. CHEM., 1994,VOL. 4 P-bipy +Fe2+-SO,'-+P-bipy .Fe2+-SO,'-P-bipy -Fez+ * SO,'-+( SO,'-)P-bipy -Fe2+ 2(SOi-)P-bipy Fe2+(S0,'-)( P- bipyj2*Fe2+ +Fez+.SO',-(SO:-)( P-bipy), -Fe2++P-bipy .Fez+* SO,'-+(SO:-)(P-bipyj, Fe2+ +Fez+* SO:-Scheme 1 P-bipy +Fe2+.(X 3, P-bipy Fe2+ (X-j, P- bipy * Fez+ * (X-),+( X -)P-bipy * Fe2+ X -+2(X -)P-bipy Fe2 X -+((X -)P -bipy} * Fez+ +Fe2+ -(X 3, {(X -)P- bipy) 2* Fe2++(X-)P- bipy -Fe2+.X2+ {(X-)P- bipy}, -Fe2++ Fe2+* (X 7, Sche'me2 733 fast (v) rate-determining step (vi) fast (vii) fast (viii) fast (v') fast (ix) rate-determing step (x) fast (xi) intense, that there is an additional absorbance at ca.400 nm which can be attributed to the combination of the spectra arising from the [Fe( bi~y)~]~' and [Fe( bipy)12+ species (literature2' La,z 410 nm, 1=320 dm3 mol-I cm- '). There-fore we have independent evidence supporting the argument for the predominant initial mono complex formation.The argument that the concentration of the bis complex will be negligible should still hold. Therefore the order of the rate constant for the formation of the [Fe( bipy),12+ centre will correspond to the number of moles of the mono complex taking part in the rate-determining step. The involvement of the bis complex in a rate equation would require an inverse dependence upon [Fe2+] to occur, which is not observed. Therefore we can write the rate expression for the sulfate as: and that for X=Cl or ClO, as dCFe(bipy)3X21=kx[Fe(bipy)2+]2 (3)dt The rate constants quoted in Table4 are many orders of magnitude smaller than those describing complex formation between Fe2+ and 2,2'-bipyridine in aqueous s~lution.~ We believe that difference to be an intrinsic property of ligand- containing copolymers.In order to form the tris complex it is necessary to bring different segments of polymer chain into close proximity, a process which will demand that some rearrangement occurs to present them in suitable relative configurations. If we are correct in attributing the very slow rates of reaction to the need for some polymer-chain reorganis- ation to occur, then this reorganisation need occur for only 1 mol of the mono complex when the anion is sulfate, but for 2 mol when the anion is chloride or perchlorate. We attribute this difference to the differing charges on the anions. When a salt diffuses into the swollen hydrogel it must do so as ion pairs for electrostatic reasons, and ion-pairing will continue to exist after complexation has occurred or a build-up of positive charge would occur. Thus the mono complex initially formed is probably best written as P-bipy.Fe2+.SOz-. We deliberately show the polymer chain and the anion at opposite sides of the cation to indicate that, in this configuration, the anion will block the access of other ligands to the cation.Before a bis complex can be formed the cation must be made more accessible by rearranging the structure. We choose to write such a rearranged structure as (SOi-)P-bipy.Fe2+. A plausible mechanism for the first-order formation of the sulfate product can now be written as in Scheme I.In this scheme the back reactions are assumed to be slow and unimportant, and are not shown.The formation of products with singly charged anions in a second-order process can be explained by the slightly modified Scheme 2. The occurrence of this mechanism, rather than a first-order alternative, requires an encounter between two (X-)P-bipy.Fe2'.X-species to be more probable than the formation and sub- sequent reaction of the ion-paired configuration (X-)2P-bipy.Fe2+, i.e.there must be a structural factor which militates against the latter alternative. This does not seem too unlikely. Conclusions We have demonstrated that hydrogel membranes made by the copolymerisation of 2-hydroxyethyl methacrylate with 4-methyl-4'-vinyl-2,2'-bipyridyl are capable of coordinating iron (11) in aqueous solution.When hydrated, typically 40% of the membrane's weight is due to imbibed water, which provides a medium through which metal salts can difl'use and coordinate with the appended 2,2'-bipyridyl ligands. Both the permeating anion type and the three-dimensional polymer network contribute to produce rates of complexation that are very much slower than the analogous reactions of the free species in solution. At low iron concentrations this can be attributed in part to a diffusion-controlled process, but at higher iron concentrations there is no doubt that the nature of the anion becomes the dominant factor. Within the polymer, in excess of 80% utilisation of irnmobi- lised ligands in tris complexes is possible, a phenomenon that reflects the high degree of polymer chain flexibility, despite the fact that every complex centre is essentially a three-centred cross-link.14 However, the extent of coordination is once again shown to be influenced by the type of salt used, which also determines the position of the equilibrium between the mono and tris complexes.We wish to thank SERC for their financial support of this work through the provision of a research studentship to A.L.L. References 1 W. W. Brandt, F. P. Dwyer and E. C. Gyarfas, Chem. Re?>., 1954, 54,959. 2 H.Irving and D. H. Mellor, J. Chem. Soc., 1962, 5222. 3 L. F. Lindoy and S. E. Livingstone, Coord. Chern. Rev., 1967, 2, 173. 4 J.D.Miller and W.R. McWhinnie, Adv. Inorg. Chem. Radiat. Chem., 1969,12,135. 734 J. MATER. CHEM., 1994, VOL. 4 F. Blau, Berichte., 1888,21, 1077. 13 A. L. Lewis and J. D. Miller, J. Chem. Soc., Chem. Commun., R. J. Card and D. C. Neckers, J. Am. Chem. Soc., 1977,99,7733. 1992,1029. R. J. Card and D. C. Neckers, Inorg. Chem., 1978,17,2345. 14 A. L. Lewis and J. D. Miller, Polymer, 1993,34,2453. C. E. Carraher, J. E. Sheats and C. U. Pittman Jr., Metal-15 A. L. Lewis and J. D. Miller, J. Mater. Chem., 1993,3, 897. containing Polymeric Systems, Plenum Press, New York, 1985, 16 H. D. Abruna, A. 1. Breikss and D. B. Collum, Inorg. Chem., 1985, p.385. 24, 987. 9 D. C. Neckers, J. Macromol. Sci., Polym. Chem. Ed., 1983,21,3115. 17 F. W. Cagle Jr. and G. F. Smith, Anal. Chem., 1947,19,384. 10 D. C. Neckers and K. Zhang, J. Polym. Sci., Polym. Chem. Ed., 18 C. J. Hamilton, S. M. Murphy, N. D. Atherton and B. J. Tighe, 1983,21,3115. Polymer, 1988,29, 1879. 11 J. D. Miller and D. S. Morton, J. Chem. Soc., Dalton Trans., 19 A. L. Lewis and J. D. Miller, Polymer, in the press. 1983,1511. 20 P. Krumholz, J. Am. Chem. Soc., 1949,71,3654. 12 C. M. EIIiott, C. J. Baldy, L. M. Nuwaysir and C. L. Wilkins, Inorg. Chem., 1990,29, 389. Paper 3/05994D; Received 7th October, 1993