J . Chem. SOC., Faraday Trans. 1, 1989, 85(1), 55-64 Effect of Spinel Oxide Composition on Rate of Carbon Deposition Geoffrey C. Allen* and Josephine A. Jutson Central Electricity Generating Board, Berkeley Nuclear Laboratories, Berkeley, Gloucestershire GL13 9PB The deposition of carbon on fuel cladding and other steels results in a reduction in heat-transfer efficiency. Methane and carbon monoxide are added to the gaseous coolant in power reactors to reduce the radiolytic oxidation of the graphite moderator and this is known to increase the rate of carbon deposition. However, the composition of oxides formed on steel surfaces within the reactor may also influence deposition. In this investigation carefully characterised spinel-type oxides of varying com- position have been subjected to y-irradiation under conditions of tem- perature, pressure and atmosphere similar to those experienced in the reactor.The rate of carbon deposition has been studied using scanning electron microscopy (SEM) and energy-dispersive X-ray analysis (EDX). Carbon deposition has been observed in power reactors on surfaces used in reheater and superheater pipework under operating conditions. Such deposition results in a reduction in heat-transfer efficiency and may require ‘ down-grading ’ of the reactor to prevent overheating of the fuel. Two probable sources of the carbon deposit are methane and carbon monoxide which are added to the carbon dioxide gas coolant to reduce the radiolytic oxidation of the graphite moderator. Much research has concentrated on the identification of an optimum gas coolant composition to prevent corrosion of the graphite moderator while reducing the rate of carbonaceous deposit on fuel cladding and other surfaces.l Recent observations have indicated that the surface morphology of fuel cladding also affects the rate of carbon deposition. If this is the case it is important to know how this occurs and to identify accurately the nature of the surface responsible for producing such surface accumulations. The steels used form ‘protective’ oxide layers composed of two major phases, a rhombohedra1 phase consisting of Cr,O,, Fe,O, or a solid solution of these binary oxides together with a cubic spinel phase.Oxides formed in fuel regions consist mainly of Cr,O, and a manganese-iron-chromium spinel,2 while those formed on reheater and superheater alloys consist mainly of a chromium-nickel-iron spinel with Fe,O, and Cr203., The temperature of these surfaces varies according to their position in the reactor.Since the composition of the oxide formed on a steel also varies with temperature, a range of oxide compositions would be expected to form on each type of steel. It has been suggested that iron and nickel may be active in catalysing the deposition of To verify these suggestions a programme of research has been initiated to investigate the effect of iron, nickel, manganese and chromium on the rate of carbon deposition. A range of standard spinels has been prepared by solid-state reaction6 in atmospheres similar to those of the Advanced Gas Cooled Reactor.Each series of spinels synthesised involved two or more of the component metallic elements of the substrate alloy in varying proportions. Each of the prepared spinels was subjected to radiation in the y cell 5556 Carbon Deposition on Spinels facility at Berkeley Nuclear Laboratories under conditions similar to those experienced by fuel cladding and other steels in reactors. Results from this study are given below. Experimental Spinels in the series Ni,Fe,-,Fe,O,, Mn,Fe,-,Fe,O,, FeFe,-,Cr,O, and Mn,Fe,-,Cr,O, were prepared by solid-state reaction at 950 "C in an atmosphere of C0,/2 O h C0.6 The structure and composition of each spinel was confirmed by X-ray diffraction (XRD) and energy-dispersive X-ray analysis (EDX). ' X-Ray Diffraction Lattice parameters were recorded using a Philips PW 1050 vertical X-ray diffractometer with unfiltered Cu radiation.A graphite crystal monochromator in the diffracted beam removed Cu K,, fluorescent and incoherently scattered radiation. The powdered samples, which consisted of small randomly orientated crystals, were dispersed in methanol and mounted on a non-reflecting silicon crystal wafer to improve the signal- to-noise ratio. Evaporation of the methanol gave an even dispersion of the specimen. Energy-dispersive X-Ray Analysis SEM and EDX analysis was carried out using a Cambridge Instruments Model S150 Mk I1 fitted with a Kevex windowless detector analysis system. Qualitative analysis was obtained by measurement of the peak energy in the characteristic spectrum of a sample.Relative amounts of carbon on each surface were compared by measuring the intensity of the carbon K, line in the spectrum obtained from each of the samples examined. Radiation Experiments Each spinel was pressed into a disc of 8 mm diameter, 1-2 mm thick and loaded into a siiica tube with silica spacers designed to allow unimpeded gas flow over the discs (fig. 1). The silica tube was then loaded into a stainless-steel capsule and placed in the y cell facility at Berkeley Nuclear Laboratories (fig. 2). A 6oCo source is used in this facilitys and the general arrangement of the source within the y cell is indicated in fig. 3 at the end of the guide tube. The capsule was placed in an outer irradiation position (fig. 4) where a dose rate of 8.20-8.34mW g-l was obtained Samples were exposed to two separate temperature regimes within the y cell and examined by SEM and EDX after these exposures.Experiment 1 : y Cell Exposure at 200 "C Spinels in the series FeFe,-,Cr,O, and MnFe,_,Cr,O, were maintained at a temperature of 200 "C for 30 days in the y cell facility. Gas of composition CO,/l O h CO with 800 vpm CH, and 15 vpm C2H6 flowed through the capsule at a rate of 2-3 cm3 min-l and 40 bar pressure in a single-pass experiment. Water was added to the gas mixture to the extent of 200-300 vpm. Experiment 2: y Cell Exposure 550 "C Spinels in the series Ni,Fe,-,Fe,O, and Mn,Fe,-,Fe,O, were maintained at 550 "C for 15 days using the gas composition and flow conditions of experiment 1.J . Chem. SOC., Faraday Trans. I , Vol.85, part 1 Plate 1 Plate 1. Electron micrographs showing Fe,O, spinel surface before (a) and after (b) exposure in the y cell for 30 days at 200 "C: G. C. Allen and J. A. Jutson (Facing p . 56)J . Chem. SOC., Faraday Trans. I , Vol. 85, part 1 Plate 2 Plate 2. Electron micrographs showing Fe2,BCr0,404 spinel surface before (a) and after (b) exposure in the y cell for 30 days at 200 "C. G. C. Allen and J. A. JutsonPlate 3. Electron micrographs showing carbon spheres on the surface of ( 1 1 ) FeCr,O, and ( h ) Mn,,7,Fe,,,,Cr10, after expobure in the ;I cell for 30 days at 200 "C. G. C . .411en and J. A. JutsvnG . C. Allen and J . A . Jutson 57 / f f / 0 \ . \ 1 cam part m en t \ . f f , . . .. . silica silica spinel tube spacer disc gas flow vent - 10 mm Fig.1. Arrangement of spinel discs in y-cell capsule. double doors 1 gamma [cell - rack housing L source B- flask concrete - shielding Plug door Fig. 2. Plan of y cell. Results y Cell Exposure at 200 "C Visual examination of the discs indicated a blue interference film at the surface of the spinels with high iron content in the series FeFe,-,Cr,O,. Subsequent examination by SEM and analysis using EDX showed that this was due to carbon. In fact, all of the spinels in both series had some carbonaceous deposit on the surface. Micrographs of back-scattered images showed what appeared to be a general layer of carbon across the spinel surface. Secondary electron micrographs of this layer are shown in plates 1 and58 Carbon Deposition on Spinels y//A irradiation posit ion 1000 mm t 1 Fig.3. General arrangement of source. irradiation positions W \source carrier 10 cm I I Fig. 4. Arrangement of irradiation position. 2, and a corresponding layer with scattered spherules of carbon is shown in plate 3. EDX analysis was carried out using 20 and 10 kV electron-beam exciting voltages. The intensity of the carbon K, X-ray signal measured in counts (100 s)-' was much higher for analysis of a given sample using a 10 kV rather than 20 kV electron-beam energy and in every case the carbon intensity measured from spherule regions by EDX was higher than that recorded from the general surface (tables 1 and 2). y Cell Exposure at 550 "C No spherules of carbon were identified on the spinel surface following exposure to the same gas conditions but higher temperature.A layer of carbon was apparent only over small areas of the surface of the spinel samples with high iron content. The measured intensity of the carbon K, X-ray line for each spinel is listed in tables 3 and 4.G. C. Allen and J. A. Jutson 59 Table 1. Carbon counts per 100 s for spinel series FeCr,Fe,-,O, maintained in the cell for 30 days at 200 "C spinel general surface sp herule 10 kV 20 kV 10kV 20kV Fe3+(Fe2+Fe3+)0," - 5583 4266 19 989 1776 1159 1506 990 3138 2026 672 205 3386 1467 1025 985 322 1 1059 928 478 15341 - - - Inverse (a) - normal (b) spinel transition, degree of inversion estimated for intermediate spinels. Octahedral site occupancy indicated by brackets. Table 2. Carbon counts per 100 s for spinel series Mn,Fe,-,Cr,O, maintained in the cell for 30 days at 200 "C spinel general surface spherule 10kV 20kV 10 kV 20 v 928 478 15341 - - 447 533 71 1 1430 303 - 4392 - - 506 - Normal-normal spinel transition, octahedral site occupancy indicated by brackets.Table 3. Carbon counts per 100 s for spinel series Mn,Fe,-,Fe,O, maintained in the cell for 15 days at 550 "C carbon counts (general surface) spinel 10 kV 20 kV Fe3+(Fe2+Fe3+)Oda - 2165 /MnOb - 1157 1081 I805 433 263 36 1 526 Inverse (a) - normal (b) spinel transition, degree of inversion estimated for intermediate spinels. Octahedral site occupancy indicated by brackets.60 Carbon Deposition on Spinels 0. Table 4. Carbon counts per 100 s for spinel series Ni,Fe,-,Fe,O, maintained in the cell for 15 days at 550 "C b I I I I spinel carbon counts (general surface) 20 kV Fe3+( Fe2+Fe3+)0, 2165 487 390 473 Fig.inverse-inverse spinel transition, octahedral site occupancy indicated by brackets. 6000 A -- loo0 t - -A - - -- -A- - - -- -A - 0.5 1.0 1.5 2.0 X Fe C r 0 5. Carbon deposition on spinels in the series FeFe,-,Cr,O, after 30 days in the y cell at 200 "C (EDX). A, 10 kV; 0, 20 kV. For both experiments the intensity of the recorded signal for a given sample was plotted against its composition. For a given series of spinels the relative carbon count was considered to be proportional to the relative amount of carbon deposition. Discussion y Cell Exposure at 200 "C The results for the series FeFe,-,Cr,O, showed a decrease in the amount of carbon deposited with decreasing iron(II1) content of the spinel-type oxide and increasing chromium concentration (fig.9, indicating iron to be active in promoting carbon deposition. The increased penetrating power of the 20 kV electron beam compared with that of the 10 kV beam, gave lower intensity carbon K, signals suggesting that this element was present as a surface layer on the spinel rather than incorporated within the oxide matrix. This is in agreement with the visual evidence from SEM micrographsG. C. Alien and J . A . Jutson 61 2 8 I I I 1 0.25 0.50 0.75 1 Fe C r20, X MnCr204 Fig. 6. Carbon deposition on spinels in the series Mn,Fe,-,Cr,O, after 30 days in the y cell at 200 "C (EDX). 20 kV electron beam (plates 1 and 2). A comparison of the disc surface before and after exposure in the y cell showed an almost transparent layer surrounding the oxide particles which was attributed to carbon.The spherule shapes identified at the spinel surface which gave rise to more intense energy-dispersive X-ray signals appeared to consist solely of carbon. By contrast, the intensity of the carbon K , signal recorded from the Mn,Fe,_,Cr,O, series showed little variation with change in iron(I1) or manganese(I1) content (fig. 6). A careful consideration of the results from these two series indicated that while iron@) in the absence of iron(m) had little effect on carbon deposition, a variation in the iron(m) concentration when iron(I1) was present in constant concentration markedly affected the rate of deposition. y Cell Exposure at 550 "C In general the intensity of the carbon K, line recorded for the series Mn,Fe,-,Fe,O, and Ni,Fe,,Fe,O, was lower than that measured at 200 "C.The exposure time was also lower, but previously it had been supposed that, at a higher temperature, deposition would occur at a faster rate since methane destruction and vinyl radical (C&) production, which is thought to be a precursor to deposition, is known to increase with increasing temperat~re.~ However, it is likely that more than one process leading to carbon deposition was involved. At the lower temperature, deposition may have occurred via the Boudouard reaction : CO,(g) + CO(ads) (1) CO(ads)+ C(ads) + O(ads) (2) (3) while at higher temperatures a deposition mechanism in which the radical products of radiolysis were involved was likely to have been important. O(ads) + CO(g) or CO(ads) - CO,(g) CO;(g) - cow (4) COi(g) ----+ C0,CO; or CO,CO,CO;t polymerisation ( 5 )62 Carbon Deposition on Spinels 0.25 0.5 X 0.75 1 Mn Fe 0, Fig.7. Carbon deposition on spinels in the series Mn,Fe,-,Fe,O, after 15 days in the y cell at 550 "C (EDX). A, 10 kV; .,20 kV. positive ions C2H&3ds) + 3CHj(g)- 2C(ads) + 3CH,(g) . (10) At this stage in the investigation, however, it is not possible to identify either as being the major pathway to deposition. For the spinel series Mn,Fe,-,Fe,O, the intensity of the measured carbon K, signal decreased very rapidly at first and slowly thereafter with decreasing iron(I1) content (fig. 7). The exception to this general observation is MnFe,O,, where the presence of manganese appeared once more to encourage deposition.The preparation of MnFe,O, in C0,/2% CO in fact produced a mixture of a spinel which appeared to be depleted in Mn" and enriched in Fe" with a lattice parameter corresponding to that for Mno,,Fe,,,Fe,O, together with a small amount of MnO. Therefore manganese ferrite was specially prepared in CO, but this product again decomposed to give a manganese depleted spinel and MnO after exposure in the y cell. Manganese oxide is known to promote iron as a catalyst for CO decomposition," and this could have been responsible for the increased carbon count. More than one solid-state property may be responsible for promoting the deposition of carbon, although the activation energy for the thermal outer-sphere Mn"/Mn"' electron transfer between ions in aqueous solution is close to that for the Fe"/Fe"' electron transfer under similar conditions" and both types of intervalence charge- transfer transition have, for example, been identified in the mineral yoderite (Al, Mg, Fe, Mn), Si,01,(OH)2.12 The question here is whether the manganese ion occupies tetrahedral or octahedral positions and whether it has an oxidation state of two or three.Conductivity measurements on manganese-iron-oxygen spinels indicate the following equilibrium between octahedral ions for the temperature range - 180 to 300 "C. Fe3+ + Mn2+' Fez+ + Mn3+; + 0.30 eV.G. C. Allen and J. A . Jutson 0 63 I I I I Fig. 8. Carbon deposition on spinels in the series Ni,Fe,-,Fe,O, after 15 days in the y cell at 550 "C (EDX).20 kV electron beam. This equilibrium lies well to the left and, according to Rieck and Driessens,13 the ground- state formula for manganese ferrite may be written and the equilibrium constant for the above distribution is ca. 36.14 In view of this it is difficult to attribute the enhanced deposition behaviour of our sample to Mn2+ + Fe3+ charge transfer when the Mn2+ occupancy of octahedral sites in the spinel lattice is so small. Moreover, it is highly unlikely that a heteronuclear Mn2+ + Fe3+ intervalence interaction would occur more readily than the homonuclear Fez+ + Fe3+ transfer. A possible explanation is that the presence of manganese and the reducing conditions of the experiment enhance the Fez+ occupancy of octahedral sites, thereby increasing the Fe2+/Fe3+ ratio in these positions and hence the catalytic activity of the oxide substrate.For the series Ni,Fe,-,Fe,O, the recorded carbon Ka signal decreased with increasing nickel content (fig. 8). However, X-ray diffraction showed that exposure in the y cell in the reducing gas atmosphere produced partial decomposition of nickel-containing spinels, the products being nickel and an iron(I1)-enriched spinel. This reduction process has been observed previously during the preparation of nickel spinels.6 It is possible that for these spinels the rate of deposition depends on both the iron and nickel content. In all the experiments carried out so far the greatest rate of deposition has occurred on magnetite (Fe,O,) which, as an inverse spinel, contains equal amounts of Fe" and Fe"' in octahedral sites.Electron exchange between these ions stabilises the spinel structure, offering a catalytic surface through the process of mixed-valence interaction. l5 When Cr'l' replaces Fe"' in the series Fe Cr,Fe,-,O,, Fe" is displaced to the tetrahedral position, thereby reducing the capability for intervalence electron exchange in the octahedral lattice. Similarly, when Fe" is replaced by Ni" or up to 50% Mn" the capability for intervalence electron exchange is reduced and in each experiment this reduction was accompanied by a reduction in carbon deposition. Previous work on the catalytic effect of iron and iron oxides on carbon deposition16 has shown that Fe,O, is a more active catalyst for deposition than Fe,O,, FeO or Fe, giving further support to the view that the Fe"/Fe"' couple may be the most effective of 3 F A R 8564 Carbon Deposition on Spinels the ion species.Rethwisch and Dumesicl' have also noted a high activity of magnetite for water-gas shift relative to other oxides. This they attribute to the variable oxidation state of iron which facilitates surface oxygen transfer. A number of processes involving iron, manganese and nickel may be important in deposition and the present studies indicate that of these Fe"/Fe"' exchange could be the most important. Conclusions Preliminary investigations into the effect of surface oxides on carbon deposition have shown that the iron content of the spinel-type oxide plays a major role? at least in the initial stages. The effect of the presence of nickel and manganese is not yet clear, but future experiments involving exposure of oxides for much longer periods in the y cell should clarify their role in deposition.Future work will also include the quantification of deposits, the deposition behaviour of other spinel surfaces, the effect of exposure to reactor-type gas environment without y radiation and the variation in deposition behaviour with temperature. This work was carried out at the Berkeley Nuclear Laboratories of the Research Division and the paper is published with permission of the Central Electricity Generating Board. The authors thank Mr S. M. Underwood for his work on the scanning electron microscope, also Mr J. V. Best and Mr R. M. Parfitt for their assistance with the irradiation experiments in the y cell. References 1 G. R. Marsh, D. J. Norfolk and R. F. Skinner, CEGB Report TPRD/B/0592/N85 (1985). 2 P. A. Tempest and R. K. Wild, J. Nucl. Mat., 1981, 102, 183. 3 A. F. Smith, Werkstofle Korrosion, 1981, 32, 1. 4 A. M. Brown, A. M. Emsley and M. P. Hill, CEGB Report No. RD/L/M/160/80 (1981). 5 R. T. K. Baker and J. J. Chludzinski, J. Catal., 1980, 64, 464. 6 G. C. Allen, J. A. Jutson and P. A. Tempest, J. Nucl. Mat., in press, 7 J. T. Buswell and G. K. Rickards, CEGB Report RD/B/N3365 (1975). 8 J. V. Best, CEGB Report RD/B/N2431 (1972). 9 D. J. Norfolk, R. F. Skinner and W. J. Williams, Gas Chemistry in Nuclear Reactors and Large Zndustrial Plant, Conference Proeedings, University of Salford, April 1980. 10 K. M. Kreitman, M. Baerns and J. B. Butt, J. Catal., 1987, 105, 319. 1 1 N. S. Hus, Electrochim. Acta, 1968, 13, 1005. 12 R. M. Abu-Eid, K. Langer and F. Siefert, Phys. Chem. Min., 1978, 3, 271. 13 G. D. Rieck and F. C. M. Driessens, Acta. Crystallogr., 1966, 20, 521. 14 A. Miller, J. Appl. Phys., 1960, 31, 2615. 15 G. C. Allen, P. M. Tucker and R. K. Wild, Philos. Mag., 1982, 46, 41 1. 16 R. T. K. Baker, R. J. Alonso, J. A. Dumesic and D. J. C. Yates, J. Catal., 1982, 77, 74. 17 D. G. Rethwisch and J. A. Dumesic, Appl. Catal., 1986, 21, 97. 18 G. C. Allen, J. A. Jutson and P. A. Tempest, CEGB Report TPRD/B/1028/R88 (1988). 19 H. J. Yearian, J. M. Kortvight and R. H. Langenheim, J. Chem. Phys., 1954, 22, 1 196. Paper 8/0047 1 D ; Received 29th January, 1988