J. Chem. SOC., Faraday Trans. I, 1988, 84(2), 379-390 Characterisation of an RuO, xH,O Colloid and Evaluation of its Ability to Mediate the Oxidation of Water Andrew Mills* and Neil McMurray University College of Swansea, Department of Chemistry, Singleton Park, Swansea SA2 8PP A colloid of Ru0,-xH,O supported by polybrene was prepared, cha- racterised and assessed as an 0, catalyst. Thermal analysis (t.g.a. and d.t.g.a.) of RuO, .xH,O precipitated from the colloid indicated the presence of both weakly and tightly bound water. Dynamic light scattering indicated that the coagulated colloidal particles were large (d = 825 nm) and positively charged. Transmission electron microscopy demonstrated that the colloidal particles were themselves aggregates of crystallites too small (< 10 nm) to be clearly resolved.The colloid proved unstable towards coagulation under conditions of high electrolyte concentration (2 mol dm-3) even when the electrolyte was H,SO,. In the presence of CeIV ions the colloid did show some activity as an 0, catalyst (0, yield = 73 %) but also underwent some anodic corrosion to RuO, (27%). At low concentrations of CeIV ions (4.5 x lo5 mol dm-3) and colloid ([RuO,] = 0.02-0.001 25 g dm-3) the col- loid appeared to mediate the oxidation of polybrene over that of water by the Ce" ions. Kinetic studies performed under these conditions and in the presence of a high constant background concentration of polybrene (0.015 g dmW3) showed the kinetics to be biphasic with an initial fast step (associated with charging of the catalyst) followed by a second step which was proportional to the concentrations of both Ce'" ions and colloid.Under conditions where no extra polybrene was added to dilutions of the colloid some 0, evolution was observed (ca. 20%) and the kinetics of Ce" disappearance once again appeared biphasic, although more complicated and difficult to interpret. A major approach1 towards the conversion and storage of solar energy into chemical energy is the development of a photochemical system capable of cleaving water into hydrogen and oxygen efficiently, i.e. sunlight 2H20 )2H, + 0,. (1) photochemical system It is generally recognized' that an essential step towards achieving this objective is the discovery and development of catalysts which are stable, specific and fast-acting towards water reduction and oxidation, respectively.A great deal of progress has been made in the area of water reduction catalysts (H, catalysts), and there are now materials such as colloidal platinum which are stable, specific and able to act in the ps time domain., In contrast, progress in the area of water oxidation catalysts (0, catalysts) has been slow and, so far, most of the materials tested appear t o be either inactive as 0, catalysts or themselves undergo anodic corrosion when subjected to the strong oxidising conditions [E"(O,/H,O) = 1.23 V us. NHE) necessary to oxidise water to oxygen. For example, although ruthenium dioxide hydrate (RuO, - xH,O) has, for several years, found frequent use as a catalyst for the oxidation of ~ a t e r ~ - ~ by a strong oxidant such as CeIV ions (2) RuO,.zH ,O 2H20 + 4Ce4+----+4Ce3+ + 4H+ + 0, 379380 RuO, * xH,O Colloids and the Oxidation of Water we have7-' now established that for all commercial samples in the presence of CeIV ions some degree of anodic corrosion to ruthenium tetroxide (RuO,) occurs.Indeed, with the majority of samples tested, anodic corrosion, and not water oxidation, was the only process ob~erved.~ In addition we have shown that Ru0,-xH,O powder may be ' activated ' as a catalyst in reaction (2) and passivated towards anodic corrosion simply by heating the sample at ca. 145 "C in air for 5 h.' The 0, catalyst produced using this procedure is not, however, very fast-acting. For example, using 10mg of catalyst dispersed in a solution (100 cm3) of sulphuric acid (0.5 mol dm-3) containing CeTV ions (3.6 x In the search for faster-acting 0, catalysts several groups10-12 have produced colloids of RuO;xH,O; however, little effort has been made to characterise these colloids so as to help evaluate their potential as 0, catalysts.An ideal 0, catalyst colloid should be stable towards anodic corrosion, active towards water oxidation and stable towards coagulation at high ionic strength. In this paper we report the findings of a detailed characterisation of a real colloid (a colloid of RuO;xH,O supported by polybrene) similar to one previously reported as an 0, catalyst," and yet, from our work, apparently far from ideal as such a catalyst. mol dm-3), ti = 104 s. Experimental Materials Polybrene ( 1,5-dimethyl- 1,5-diazaundecamethylene polymethobromide, hexadimethrine bromide) and cerium(1v) sulphate tetrahydrate ( > 99 %) were obtained from Aldrich.Solutions containing Ce(SO4);4H,O were made up from the solid using 0.5 mol dm-3 H,SO, (AR; BDH) and standardised spectrophotometrically [e(Ce4+) = 5580 dm3 mol-1 cm-l at Amax = 320 nm].13 The solid ruthenium tetroxide (RuO,) used in the colloid preparation was made using a modified version of a procedure described by Connick and H~r1ey.l~ Thus anhydrous RuO, was prepared by distillation, in a stream of nitrogen, from a solution containing a dispersion of RuO, - xH,O (Johnson Matthey), potassium permanganate (AR; BDH) and dilute sulphuric acid. The volatile RuO, (m. pt 24.4-24.8 "C; literature value15 25.4 "C) was passed through a trap of anhydrous magnesium perchlorate and collected in a U-tube immersed in liquid nitrogen.In this work all the sdutions were made up using doubly distilled deionised water. Methods All u.v.-visible absorption spectra were recorded using a Perkin-Elmer Lambda 3 spectrophotometer. The gravimetric analyses (t.g.a. and d.t.g.a.) were performed using a Stanton Redcroft TG-750 instrument and a Linseis chart recorder (type LS4). Using this apparatus a sample (ca. 7 mg) was usually heated from ambient temperature (typically ca. 22 "C) to 900 "C at a rate of 20 "C min-' and the percentage weight loss was recorded as a function of time with an x / t chart recorder. In all cases N, was flowing at a rate of 25 cm3 min-l. Transmission electron micrographs of the colloid were recorded using an electron microscope (1 20 C TEM-SCAN) manufactured by JEOL.Copper grids, covered with a carbon support film, were loaded with sample as follows : a grid was placed on a droplet of colloid lying on a surface of dental wax; after ca. 30 min the grid was lifted off using a pair of tweezers, the excess liquid was drawn off with a piece of tissue paper and the grid was then left to dry (30 min) before being used in the TEM-SCAN. The dynamic light scattering experiments were performed by Dr J. R. Danvent and Ms Anne Lepre (Birkbeck College, London) using a Malvern Instruments Zeta-Nanosizer. A ' test system ' was devised to assess the ability of any sample of RuO, - xH,O, powder or colloid, to act as a catalyst or undergo corrosion when exposed to a strong oxidisingA .Mills and N. McMurray 38 1 agent such as CeIV ions. The 'test system' comprised a N, cylinder which provided a continuous flow (fz 180 cm3 min-l) of gas through two Dreschel bottles (each 125 cm3), and an oxygen membrane polarographic detector (0,-MPD) coupled in series. Of the Dreschel bottles, the first contained 100 cm3 of a CeIV solution (3.6 x rnol dm-3 in 0. I mol dm-3 H,SO,) and also had the additional feature of a rubber septum through which the stock colloidal RuO,.xH,O suspension (20 cm3; containing ca. 0.4 g dm-3 RuO, and 0.2 g dm-3 polybrene) were injected, and the second contained 100 cm3 0.1 mol dm-3 NaOCl in 1 mol dm-3 NaOH [used to trap, in the form of per-ruthenate (RuOJ any RuO, produced]. The percentage corrosion was calculated using the equation where N(Ru0;) is the number of moles of RuO, trapped in the hypochlorite solution (measured spectrophotometrically)16 and N(Ru0, * xH,O) is the number of moles of RuO, - xH,O injected.The percentage corrosion calculated using eqn (3) represented a minimum value, since RuO, also attacked the rubber septum and the glassware (as evidenced by blackening of both) before reacting in the hypochlorite trap. Owing to the high solubility of RuO, (ca. 20.3 g dm-3, at 20 OC),15 a long time (6-8 h) was required to flush out most of the RuO, produced (> 98 YO) following colloid injection, since the time taken to sweep out half of the RuO, produced was found to be ca. 70 min. The 0,-MPD has been described el~ewhere'~ and allowed the determination of the number of moles of 0, produced, following catalyst injection.The percentage 0, yield corrosion = [N(RuO,)/N(RuO, -xH,O)] x 100 YO (3) was calculated using the equation moles 0, produced x 400 moles CerV consumed 0, yield = YO (4) where the number of moles of CeIV consumed was determined spectrophotometrically from the drop in absorbance of the CeIV solution following catalyst injection. Catalyst Preparation The colloid of RuO, .xH,O supported by polybrene was prepared as follows : RuO, (200 mg) was dissolved in 200 cm3 water contained in a stoppered flask in an ice bath. This solution was stirred under these conditions for 30 min, during which time no change in the u.v.-visible absorption spectrum of the dissolved RuO, was observed. At the end of this period, with the solution still briskly stirred, an equal volume of polybrene solution (0.4 g dmP3) at 0 "C was added quickly.A reaction commenced immediately, with the solution darkening rapidly and taking on a greenish hue. The reaction appeared to be the reduction of RuO, (yellow/green) to colloidal Ru0,-xH,O (black) by the polybrene which also acted as a colloid support, preventing coagulation. This process was best observed by u.v.-visible spectrophotometry, and fig. 1 shows the observed change in absorption spectrum of an aliquot of a freshly mixed reaction solution placed directly into a sealed quartz spectrophotometer cell and thermostatted at 30 "C, as a function of time. From fig. 1, the reduction of RuO, (A,,, = 309 and 385 nm)" appeared to be complete within ca.7 h. In the catalyst preparation the reaction mixture was allowed to reach ambient temperature (ca. 20 "C) and a series of spectral changes identical to those illustrated in fig. 1 were observed but over a longer timescale, 48 h being required before the reaction went to completion. Using a similar procedure, Minero et al.ll reported that the resulting colloid is green with a maximum absorbance at ca. 400 nm. In our work, however, the maximum absorbance observed in this wavelength region (A,,, = 385 nm) was due to unreacted RuO, (see fig. 1) and the spectral profile of the colloid was always of the form illustrated in fig. 1 (e) after all the RuO, had reacted. This u.v.-visible spectrum remained unchanged over a period of382 RuO, * xH,O Colloids and the Oxidation of Water I I I I I I I 200 300 400 500 600 700 900 wavelength/ nm Fig.1. U.v.-visible spectra of a reaction solution thermostatted at 30 "C, containing RuO, (0.5 g drn-$) and polybrene (0.2 g dm-3) and recorded at 100 min intervals following mixing [for curve (a) t = 01. several months and there was no evidence of precipitation over this period. The appearance of the RuO;xH,O colloid was that of a black solution, transparent when diluted, with no trace of turbidity, but with a slight green coloration in transmitted light. Results and Discussion Catalyst Characterisation Determination of Ruthenium Metal Content The ruthenium metal content of the colloid was determined spectrophotometrically as the perruthenate ion (RuO,; A,,, = 385 nm, E~~~ = 2162 dm3 mol-1 cm-l) using the procedure developed by Larson and Thus, a 5 cm3 aliquot of colloid was made up to 50 cm3 in a volumetric flask with a solution containing sodium hypochlorite (NaOCl, 0.1 mol dm-3) plus NaOH (1 mol dm-3) and left ovenight to ensure complete oxidation of the RuO;xH,O to RuO; by the NaOCl.The final absorbance of the RuOh solution at 385 nm was measured relative to a blank consisting of 5 cm3 of 0.2 g dm-3 polybrene plus 45 cm3 of 0.1 mol dm-3 NaOCl in 1 mol dm-3 NaOH. This procedure was repeated several times, and from the appropriate Beer's law calculations the colloid was found to contain (2.8 +, 0.3) x mol dm-3 ruthenium metal, a figure which compared favourably with the value calculated (3 x mol dmW3) from the amount of RuO, added in the catalyst preparation.The concentration of RuO, in the colloid was calculated, from the Ru content, to be 0.37k0.04 g dm-3. Thermal Analysis of RuO, * xH,O Precipitated from the Colloid A 50 cm3 aliquot of the colloid was coagulated by the addition of an equal volume of 0.1 mol dm-3 H,SO,, precipitation was completed by centrifugation (126 g for 10 min) and the excess liquid was then decanted. The precipitate was then washed by dispersing it in 150 cm3 of 0.1 mol dm-3 H,SO, using stirring and ultrasound (4 min), followed by centrifugation and decantation of the resulting supernatant. The washing procedure, aimed at removal of the polybrene support, was repeated four times with 0.1 mol dme3 H,SO, and four times with 1 x mol dm-3 H,SO,.(Water could not be employed in this latter washing process since peptisation was found to occur at low ionic strengths.) The washed material (ca. 20 mg) was then allowed to dry in air at ambient temperature (ca. 20 "C) for a period of 10 days. Elemental analysis of the product revealed <4% C and <0.6 YO N by weight.J. Chem. SOC., Faraday Trans. I , vol. 84, part 2 Plate 1 (c 1 (dl Plate 1. Transmission electron micrographs of the colloid [(a) x 33 000, (b) x 2500001, the colloid coagulated by Na,SO, [(c) x 1000001 and a commercial sample (Johnson Matthey) of RuO;xH,O heated at 110 "C in air for 5 h [(d) x 1000001. A. Mills and N. McMurray (Facing p . 383)A. Mills and N. McMurray 100 90 383 - - B Fig. 2. T.g.a. (a) and d.t.g.a. (b) curves recorded for samples of RuO;xH,O obtained: (A) by precipitation of the colloid and (B) from a commercial source (Johnson Matthey).Fig. 2 shows the t.g.a. and d.t.g.a. profiles recorded for the precipitated Ru0,-xH,O [fig. 2(A)] and a commercial sample of RuO;xH,O (Johnson Matthey batch no. 061 174 [fig. 2(B)], and both are broadly similar. From fig. 2 both samples appear to be highly hydrated, and from their total percentage weight losses the values for x in the simple formula RuO;xH,O may be calculated as (A) x = 4.5 (colloidal) and (B) x = 3.2 (Johnson Matthey). The d.t.g.a. profiles for both samples are very similar and indicate the presence of two very distinct types of water in the samples, namely (i) loosely bound, probably physically absorbed water (d.t.g.a.Tmax = 80 "C) and (ii) tightly bound, probably chemically bonded water (d.t.g.a. Tmax x 220 "C). Transmission Electron Microscopy, Electron Diffraction and Energy Dispersive Analysis Transmission electron micrographs of the colloid (a) and (b), coagulated colloid (c) and a commercial sample of RuO, -xH,O (heated to 110 "C for 5 h, in air) ( d ) are illustrated in plate 1. The last transmission electron micrograph has been included for comparison with that of the coagulated colloid since it appeared to possess a similar stability towards anodic corrosion and activity as an 0, catalyst. From plate 1 and additional micrographs it appeared that all the samples of RuO;xH,O used were not composed of regular or well defined particles, but rather contained aggregates of crystallites too small ( c 10 nm) to be resolved readily. Electron diffraction experiments on both dispersed and coagulated specimens produced no discernible diffraction pattern, corroborating the t.e.m.evidence of their amorphous natures. Qualitative elemental analysis was performed by energy dispersive analysis on all the samples illustrated in plate 1, and confirmed that the subject material contained ruthenium and was not foreign matter or an artifact.384 RuO, - xH,O Colloids and the Oxidation of Water Dynamic Light Scattering It was found, using a Malvern Instruments Zetasizer 2c, that the stock colloid did not scatter light sufficiently to permit an estimation of the mean particle size or potential. Minero et al., however, estimated the hydrodynamic radii of their polybrene-stabilised RuO, colloid to be ca. 30 nm, using a dynamic light scattering technique.ll Interestingly the coagulated colloid, dispersed in water at pH 7, did scatter enough light to allow its mean particle size and potential to be estimated; these were determined to be 825 nm and 40 mV, respectively.The point of zero charge (P.z.c.) for RuO;xH,O powder is usually at pH < 3, therefore at pH 7, the Ru0,-xH,O colloid might be expected to exhibit a negative surface charge and thus a negative zeta potential. The observed positive zeta potential can be reconciled with the above if an overall superequivalent of cationic polybrene is adsorbed on the surfaces of the aggregate particles. Coagulation Studies In any chemical or photochemical study involving an 0, colloidal catalyst it may be necessary to subject the catalyst to conditions of appreciable electrolytic concentration (say > mol dm-3).For example, in order to evaluate the catalytic activity of an RuO, - xH,O colloid for water oxidation, Minero et a1.l' used a test system of CeIV ions in 0.1 mol dm-3 H,SO,, a high-ionic-strength medium. Ideally therefore, before any colloid is used as a catalyst, some assessment should be made of its stability towards coagulation in the presence of the intended high-ionic-strength medium. In our work two basic techniques were employed in order to determine (a) the maximum electrolyte concentration (for a variety of electrolytes) at which the colloid still possesses long-term (220 h) stability and (b) the timescale over which the coagulation occurs in a high-ionic-strength medium such as 0.1 mol dmd3 H,SO,.The results of this work are as follows. 20 h Coagulation Experiments. The method employed was based on that developed by Furlong et al.,' The colloid studied contained 0.02 g dm-3 RuO,, 0.015 g dm-3 polybrene, in order to simulate the catalyst and polymer concentrations reported by Minero et a1.l' in their CeIV kinetic experiments. The extent of coagulation induced over a 20 h period by the addition of H,SO,, Na,SO, and MgSO, was measured as a function of electrolyte concentration. The experiments with the last two electrolytes were conducted at pH 7. Solutions containing colloid and electrolyte at various concentrations were prepared in 20 cm3 glass sample bottles and allowed to stand at 20 "C for 20 h.Each solution was then transferred to a 15 cm3 centrifuge tube and spun at 126 g for 20 min before the top 3 cm of liquid was drawn off andits absorbance (A,) measured at A = 300 nm. A Stokes law calculation showed that aggregates with a radius > 120 nm should have been swept out of the liquid sample taken following centrifugation." A measure of the stability of the colloid towards coagulation was calculated using the equation stability = (AJA,) x 100% ( 5 ) where A , = absorbance of colloid with electrolyte present and A , = absorbance of colloid with no electrolyte present. Fig. 3 shows the variation of the observed colloid stability us. the negative logarithm of the cation concentration, from which the following is seen.(a) The colloid was completely coagulated even by very low concentrations of electrolyte (< 1 x mol dmP3). This suggests that the polybrene present was rather ineffectual as a steric stabiliser and that a large part of the colloid's stability was electrostatic in origin. (b) Coagulation385 A. Mills and N. McMurray 120, .tL 0 2.6 -0-- 100 x .- 60 n 0 20 0 2.8 3.0 3.2 3.4 3.6 3.8 4.0 - 4 .O -log ([cation]/mol dmd3) Fig. 3. Plot of the colloid stability (YO) us. the negative logarithm of the cation concentration of the coagulating electrolyte. The cations used were H' (O), Na' (a) and Mg2+ (0). of the colloid by the neutral salts (Na,SO, and MgSO,) occurred in almost the same region of ionic strength. Given that the colloid was most likely stabilised electrostatically, this result implies that, in accordance with the Schultz-Hardy rule, counter-ions surrounding the colloid were predominately anionic and that the colloid particles were, therefore, cationic.The measured positive zero potential of the colloidal material at pH 7 using dynamic light scattering supports this conclusion. (c) Coagulation of the colloid by H,SO, occurred at a slightly higher (ca. 1.5 times) ionic strength than that found for the neutral salts Na,SO, and MgSO, at pH 7. This enhancement in colloid stability towards coagulation was most likely due to the potential-determining action21 of the protons present on the colloid particles. At a pH below the P.Z.C. of the RuO, .xH,O, as in this work, the surface of the particles would be highly protonated and, therefore, electrostatically more stable towards coagulation.Rate of Colloid Coagulation. Several groups, including Minero et al.," employed CerV ions in H,SO, (30.1 mol dm-3) to test [uia reaction (2)] the activity of their RuO;xH,O colloids. It was decided, therefore, to study the variation of the rate of coagulation of our colloid in 0.1 mol dm-3 H,SO, as a function of its initial concentration. The RuO;xH,O colloid stock solution was used to prepare a range of colloid concentrations (0.02-0.001 25 g dm-3 RuO, with 0.01 5 g dm-3 polybrene). A fixed aliquot (2.5 cm3) of each solution was stirred, at a constant rate, in a 1 cm quartz spectrophotometric cell thermostatted at 30 "C. The process of coagulation was initiated by an injection of 5 x lo-, cm3 of 5 mol dm-3 H2S0, solution which gave (within 10 s) a concentration of H,SO, in the cell of 0.098 mol dm-3.Observation of the progress of coagulation was achieved by following the change in absorbance (at 1 = 320 nm) of the colloid as a function of time and a typical trace is shown in fig. 4. This method of monitoring the progress of coagulation exploits the fact that the absorption/scattering spectrum of the colloid decayed as flocculation proceeded such that dA/dA, over the range 250-400 nm, tended to zero. It was assumed that when the absorbance of the suspension had fallen by half the value it was to fall over the complete coagulation process (AA), a constant degree of coagulation had occurred. The time taken, in s, for this constant degree of coagulation [t(AA/2)] was plotted as a function of the negative logarithm of the colloid concentration and the result is shown in fig.5. From the value of the gradient (=0.99 kO.01) of the386 RuO, - xH,O Colloids and the Oxidation of Water I 1 I I 100 200 300 40 0 ti s Fig. 4. Plot of the absorbance (A = 320 nm) of the colloid (0.01 g dm-3 [RuO,], 0.015 g dm-3 [polybrene]) us. time following injection (at t = 0) of H,SO, (final [H,SO,] = 0.098 mol dm-3). '3 2 1 -log([R~O~]/gdm-~) Fig. 5. Plot of the logarithm of the time taken for the absorbance of the colloid (at 320 nm) to diminish by half, i.e. log [t(AA/2)] following addition of H,SO, (0.098 rnol dm-3) us. the negative logarithm of the colloid concentration, expressed in g dm-3. straight line shown in fig.5, it would appear that coagulation of the colloid is a second- order process. This result is in good agreement with the theory of diffusion-controlled flocculation developed by Smoluchowski,2' which predicts that the coagulation of primary colloidal particles will be second order provided sufficient electrolyte is present to remove the repulsive energy barrier. Interestingly, although it is apparent from fig. 4 and 5 that coagulation in 0.1 mol dm-3 H,SO, can be rapid at high concentrations of colloid, the time for coagulation was always found to be at least 100 times that for the colloid to mediate the reduction of any CeIV ions present. Thus, although not ideal, it does appear that for this colloid a system comprised of CeLV ions in 0.1 mol dm-3 H,SO, can be used to help evaluate, via reaction (2), its ability to mediate the oxidation of water.Corrosion of the Colloid by CeIV ions and Stoichiometry of Oxygen Release Using the 'test system ' described in the Experimental section, 20 cm3 of the stock colloid (0.37 g dm-3 RuO,, 0.2 g dm-3 polybrene) were purged with N, to remove 0, and then injected into the first Dreschel bottle of the test system containing 3.6 x mol dm-3A . Mills and N. McMurray 387 Celv ions in 0.1 mol dm-3 H,SO,. Any 0, (due to catalysis) or RuO, (due to corrosion) produced following catalyst addition was swept out of the reaction vessel by a continuous stream of N, and then analysed (see Experimental section for details). The percentage corrosion and percentage 0, yield exhibited by the colloid were determined, using eqn (3) and (4), to be 27 YO and 73 %, respectively.The resistance of the catalyst towards corrosion was greater than might have been expected in the light of its apparent high degree of hydration. This is because it has been established that the percentage corrosion of RuO, - xH,O increases with increasing degree of hydrati~n.~ Although the colloid exhibited a high degree of catalytic activity towards 0, evolution, the appreciable degree of corrosion which occurred concomitantly makes it far from ideal as an 0, catalyst. Kinetics of eelv Reduction A study of the kinetics of CeIV reduction mediated by the colloid was undertaken using a stopped-flow technique. The stock colloid solution (0.4 g dm-3 RuO,, 0.2 g dm-3 polybrene) was diluted with water to produce a variety of dilute colloid concentrations (0.02-0.001 25 g dm-3 RuO,) which allowed the reduction of CeIV ions to be monitored spectrophotometrically.[Note that with these dilute colloids the RuOJpolybrene ratio (w/w) was constant (= 2).] Previous workers5~ l1 have suggested that a high [Ce"]/[RuO,] ratio favours corrosion. In the previous section we established with the test system that 0, yields of ca. 73 O h are achieved using a [Ce4+]/[Ru0,] ratio = (3.6 x lop3 mol dm-3)/(0.08 g dm-3). Thus, in order to favour 0, evolution, rather than catalyst corrosion, in our kinetic studies we employed a sufficiently small CeIV concentration (4.5 x mol dm-3) to ensure that for any of the dilute colloids used the [Ce4+]/[Ru0,] ratio was always less than or equal to that used in the test system.The observed kinetics of the reduction of the CeIV ions appeared to be biphasic. The initial phase was a fast, high-order (> 1) process, possibly corresponding to the oxidation or charging of the colloidal particles. The second phase was a slower process which appeared to obey first-order kinetics. A plot of the logarithm of the first-order rate constant for this second phase (log k,) us. -log [RuO,] produced a good straight line with a gradient of 1.67. This result implied that for the second phase the rate of reaction ( - d[Ce4+]/dt) was proportional to ([RuO,)"[polybrene]"), where n + m = 1.67. How- ever, if polybrene was not involved in the second phase of the reaction (i.e. rn = 0), then it is likely that a major reaction product would have been 0,.In order to check this, 10 cm3 of a Ce'" solution (9 x mol dm-3) in 0.1 mol dmP3 H,SO, were placed in a cylindrical quartz reaction cell, incorporating in its base an 0,-MPD for dissolved 0, mesurernent.,, This solution was purged with N, for 20 min, sealed and, after a few minutes, 10 cm3 of stock colloid (0.0025 g dmP3 RuO,) also purged with N,, were added to produce a final colloid concentration of 0.001 25 g dm-3 RuO,. Fig. 6 shows the observed changes in current from the 0,-MPD [curve (b)] as a function of time as well as the decay of the CeIV absorbance [fig. 6(a)] observed with the stopped-flow apparatus under identical reaction conditions. From the 0,-MPD, it was found that the maximum amount of 0, observed under these reaction conditions corresponded to an 0, yield of only 20 %.This value for the 0, yield is in marked contrast to that of 73 O/O determined above using the test system (in which the CeIV, RuO, and polybrene concentrations were ca. 70 times greater). The reasons for these two very different 0, yields remain, as yet, unclear. However, it is clear that in our stopped-flow study of the reduction of CeIV ions, the major reaction is not water oxidation but, most probably, polybrene oxidation (i.e. rn # 0). Colloidal RuO, appears to be able to mediate both these oxidation reactions since, in the absence of RuO,, the oxidation of either polybrene or water by CeIV ions was found to be very slow. Since the oxidation of polybrene appears in the rate law we388 RuO; xH20 Colloids and the Oxidation of Water Fig.6. Reduction of CeTV ions (4.5 x mol dm-3) in 0.1 mol dm-3 H,SO,, mediated by the colloid ([RuO,] = 0.001 25 g dmP3, [polybrene] = 0.00063 g dm-3). The decay of the CeIV absorbance (a) was monitored using a stopped-flow technique. The concomitant evolution of 0, (b) was monitored using an 0,-MPD. The maximum change in current [Ai(O,-MPD)] observed corresponded to a 20% 0, yield. I I I I I t l s Fig. 7. Reduction of CerV ions (4.5 x mol dmP3 H,SO,, mediated by the colloid ([RuO,] = 0.01 g dm-3) with added polybrene ([polybrene] = 0.01 54 g dm-3. Plot of the negative logarithm of the observed change in absorbance of the Ce'" ions us. time. 0.5 1.0 1.5 2 .o must assume that either its coverage of the surface of the colloidal particles is low or that the polybrene counterions (Br-) are involved in the oxidation process.Although it is unclear what happens to the polybrene during oxidation, we did not find any evidence for bromine formation due to the oxidation of the Br- counterions. Minero et u1.l' reported the details of a kinetic study of Ce'" reduction mediated by an RuO, - xH,O/polybrene colloid, under reaction conditions in which ' dioxygenA . Mills and N . McMurray 389 -log ( [ RuO, 1 /g dm - 3 ) Fig. 8. Plot of the logarithm of the first-order rate constant for the second phase of the CeIv reduction kinetics us. -log [RuO,] ; [polybrene] was fixed at 0.01 5 g dm-3. evolution should be the predominant pathway '. These reaction conditions are similar to those employed in our initial kinetic study reported above (i.e.[Ce'"] = 4.5 x loe5 mol dm-3 in 0.1 mol dm-3 H,SO,, [RuO,] = 0.02-0.0005 g dmP3) with the exception that Minero et al. employed a fixed polybrene concentration (0.015 g dm-3) for all values of We repeated the kinetic study of Minero et a/." using our colloid and found, once again, that the kinetics of the Cel" reduction were biphasic (see fig. 7). The initial phase was fast, with no simple order. In addition it appeared very distinct from the second phase, and this allowed us to determine readily the quantity of Ce'" reduced in the initial phase (A[Ce4+]). This quantity was found to be directly proportional to the amount of RuO, present. The second phase was a slower process, which obeyed first-order kinetics almost perfectly (correlation coefficients were usually 2 0.9999).The kinetics reported by Minero et al. appear to be concerned solely with this second phase and their findings agree very well with our own." Fig. 8 shows a plot of log k, us. -log [RuO,]. A similar plot has been reported by Minero et d.," and in both plots it appeared that at high [polybrene]/[RuO,] ratios (i.e. % 2), the rate of Ce'" reduction was proportional to Identical reaction conditions as those employed in the above kinetic study were used to investigate the variation of the 0, yield as a function of [RuO,]. An 0,-MPD was used, as before, to monitor the amount of 0, produced; however, in no case was there found to be any evidence for 0, evolution during the reduction of the CeI" ions, mediated by the variety of colloid concentrations employed in the presence of the fixed high polybrene concentration (0.01 5 g dmP3).From this work it would appear, therefore, that the kinetics for CeIV reduction reported by Minero et al." and repeated by ourselves (see above and fig. 7 and 8) are not associated with the oxidation of water, but, more likely, with the oxidation of polybrene. [RuO,l* [RuO,I. Conclusion The results of the catalyst characterisation demonstrated that a true colloidal dispersion of RuO;xH,O supported by polybrene was prepared and that it possessed some catalytic activity for the oxidation of water. However, the stability of the colloid towards: (i) coagulation (including H,SO,), and (ii) anodic corrosion by CeI" ions390 RuO, - xH,O Colloids and the Oxidation of Water appeared poor.In addition, at the low concentrations of CerV ions and colloid necessary to carry out a kinetic study using a stopped-flow technique the colloid appeared to mediate the oxidation of polybrene rather than water. Indeed, at ratios of [polybrene]/ [RuO,] $ 2 no 0, evolution was observed during the reduction of the CerV ions. As a result of this work it is apparent that the colloid is unsuitable for use as an 0, catalyst or incorporation as such in a photochemical system capable of splitting water. Ideally, such a catalyst should be fast-acting, stable towards corrosion and coagulation and specific to the oxidation of water, i.e. not able to mediate more readily the oxidation of the support. Further work is now in progress to produce colloids of RuO;xH,O which utilise different supports and which are generated via reaction pathways other than that reported here.The colloids produced will be characterised using the techniques described above and through such research it is hoped that an ‘ideal’ 0, catalyst may be found. We thank the S.E.R.C. and the Royal Society for financial support of this work. References 1 Energy Sources through Photochemistry and Catalysis, ed. M. Gratzel (Academic Press, New York, 2 A. Demortier, M. De Backer and G. Lepoute, Nouv. J. Chim., 1983, 7, 421. 3 J. Kiwi and M. Gratzel, Chemia, 1979, 33, 289. 4 J. Kiwi, M. Gratzel and G. Blondeel, J. Chem. SOC., Dalton Trans., 1983, 2215. 5 G. Blondeel, A. Harriman, G. Porter, D. Urwin and J. Kiwi, J. Phys. Chem., 1983, 87, 2629. 6 A. Harriman, G. Porter and P. Walters, J. Chem. SOC., Faraday Trans. 2, 1981, 77, 2373. 7 A. Mills and M. L. Zeeman, J. Chem. SOC., Chem. Commun., 1981, 948. 8 A. Mills, J. Chem. SOC., Dalton Trans., 1982, 1213. 9 A. Mills, C. Lawrence and R. Enos, J. Chem. SOC., Chem. Commun., 1984, 1436. 1983). 10 J. Kiwi, J. Chem. SOC., Faraday Trans. 2, 1982, 78, 339. 11 C. Minero, E. Lorenzi, E. Pramauro and E. Pelizzetti, Inorg. Chim. Acta, 1984, 91, 30 L. 12 P. A. Christensen, A. Harriman, G. Porter and P. Neta, J. Chem. SOC., Faraday Trans. 2, 1984, 80, 13 T. J. Sworski, J. Am. Chem. SOC., 1958, 79, 3655. 14 R. E. Connick and C. R. Hurley, J. Am. Chem. SOC., 1952, 74, 5012. 15 W. P. Griffith, in The Chemistry of the Rarer Platinum Metals (Interscience, London, 1967). 16 J. L. Woodhead and J. M. Fletcher, J. Chem. SOC., 1961, 5039. 17 A. Mills and C. Lawrence, Analyst (London), 1984, 109, 1549. 18 P. Wehner and J. C. Hindman, J. Am. Chem. SOC., 1950, 72, 3911. 19 R. P. Larsen and L. E. Ross, Anal. Chem., 1959, 31, 176. 20 D. N. Furlong, A. Launikonis, W. H. F. Sasse and J. V. Sanders, J. Chem. SOC., Faraday Trans. I , 21 D. J. Shaw, in Introduction to Colloidal and Surface Chemistry (Butterworths, London, 3rd edn, 1980), 22 A. Mills, A. Harriman and G. Porter, Anal. Chem., 1981, 53, 1254. 145 1. 1984, 80, 57 1. p. 150. Paper 710’73; Received 14th January, 1987