The relationship between microstructure and the reduction of elastic modulus in thermally and radiolytically corroded nuclear graphites

THE RELATIONSHIP BETWEEN MICROSTRUCTURE AND THE REDUCTION OF ELASTIC MODULUS IN THERMALLY AND RADIOLYTICALLY CORRODED NUCLEAR GRAPHITES T. D. BURCHELI., 1. M. PICKUP, B. MCEI\;ASEY and R. G. COOKE School

of Materials Science, University of Buth,

(Received

18Sepfembcr

Bath

BA2

7AY.

UK.

1085)

Abstract-Evidence for the different effects of thermal and radiolytic corrosion on the microstructure ;md porosity development of two needle-coke filler. pitch binder graphites are presented. An interprct;ltion of the effect5 of these microstructural changes upon elastic modulus is given by utilising km analytic:tl derivation of the empirical Knudsen equation. The more severe decrements of modulus found in thermally corroded graphites are ascribed to preferential development of mirrow. slit-shaped pores of high aspect ratio within the binder phase. The selective attach of these pores is attributed to the thermal rrnction ;misotropy of the graphite-CO, reaction. Radiolytically activated CO: ;ittacks graphite mlcroatructurc less selectively developing pores of iow aspect ratio which cause less tevere decrements in elastic modulus. Key Words-~ucle~lr

graphites.

e&tic

modulus.

microstructure.

I. INTRODUCTION It is well known[l. 21 that for a given weight loss, thermal corrosion of nuclear graphites causes a greater reduction in elastic modulus and strength than radiolqtic corrosion. Hawkins[3] investigated the properties of PGA graphite after radiolytic corrosion in C‘Oz and CO,;CO gases. and compared his results with those of Board and Syuires[il] and Rounthwaite IV ul. 151 for thermal c~~rr[~sion of PGA in CO, in the temperature range 7SO-950°C. Hawkins showed graphically that modulus was reduced by about 32% of the initial value for a 5% weight loss following thermal corrosion, but for radiolytic corrosion the reduction was about 18%. Hawkins also noted that the \,isual appearance of thermally and radiolyticall~ corroded graphites differed and proposed that the mode of attack for the two types of corrosion was different. Brocklehurst[h] found similar results to those of Hawkins for radiolytic corrosion of near-isotropic nuclear graphites and showed that relative changes in modulus and strength were the same for the different types of graphite examined. In a detailed study of the effect of the initial stages of thermal corrosion of PGA graphite by CO,/S%CO. Pickup[-/] showed the decrement in dynamic modulus. E. fitted an exponential relationship:

E = E’ exp( -7.0

s)

(1)

where E’ and .Yare the modulus of uncorroded graphitc and the fractional weight loss. respectively. Similarly. Kelly er ai.[81 reported that the effects on the n~odulus of radiolytic corrosion of a wide range of nuclear graphites could be described by: E = E’ exp(-3.61).

(2)

thermal

corrosion,

radiolytic

c(~rr(~si~)n.

Thus from eqns (1) and (2) for a 5% weight loss the modulus would be reduced by approximately 30% for thermal corrosion but only by lhc; for rndiolytic corrosion, in agreement with the earlier work of Hawkins[3]. The influence of thermal and radiolytic corrosion upon the strength of nuclear graphites can be fitted to similar equations which show that strength reductions are also more severe for thermal oxidation. Although the different effects of thermal and radiolytic corrosion on the modulus and strength of nuclear graphites are well-established. there is very little work reported on the associated changes in the microstructure of the graphite. Hoard and Squircs[JI and Rounthwaite et ul.[5] both report that during thermal oxidation the binder phase is preferentially corroded. For radiolytic corrosion of a Gilsocarbon graphite in CO,-based gas. Best and Stephen(91 reported that at low CO concentrations corrosion is uniform throughout the pore structure. but that closed porosity in the Gilsocarbon grist particles is opened up in the early stages of corrosion. in no previous paper has an attempt been made to relate the different effects of thermal and radiolytic corrosion upon the mechanical properties to the microstructural changes in the graphite. This paper presents (i) evidence for the different microstructural effects ofthermal and radiolytic corrosion. and (in) an interpretation of the effects of these microstructural changes upon elastic modulus using an analytical derivation of the Knudsen equation due to Buch[ 101.

2. EXPERIMENTAL

Thermal corrosion of an extruded nuclear graphite (PGA) c~~ntaining a needle coke filler and pitch binder wx carried out at 900°C in fOJSCC0 gas at at-

T. D. BURCHELL et al.

546

mospheric pressure; details of the methods have been published[7]. Radiolytically corroded samples of an extruded needle coke filler/pitch binder graphite were removed from a Civil Advanced Gas Cooled Reactor (CAGR) in the U.K. Typical AGR coolant gas compositions have been reported by Wood and Wickhamfll] as COJl% CO with additions of 165 vpm CH, and 200-300 vpm H,O. The mean coolant gas temperature for an AGR is approximately 723 K. Microstructural studies were made using inverted stage microscopes under polarised or UV light. In the latter case good contrast between pores and graphite at high magnification was achieved by impregnating the graphite with a resin containing a fluorescent dye. 3.

RESULTS

3.1 Pares in the Uncorroded Graphite Pickup et al.[7] identified three classes of pore in the uncorroded needle-coke graphites. (i) Type A are large cracks (Position A, Fig. 1) in the filter particles which are parallel to the basal planes and form by volumetric shrinkage during calcination of the filler particles; (ii) type I3 are gas entrapment pores (Position B, Fig. 1) within the binder phase which form during the mixing and baking stage of manufacture; (iii) type C pores are narrow, slit-shaped pores (Position C, Fig. 1) in the binder phase. These pores form as a result either of volumetric shrinkage on baking or of anisotropic contraction on cooling from graphitization temperatures. They are often found at domainimozaic boundaries in the binder phase and are frequently connected to gas entrapment pores. Most of the type B and C pores in the binder phase are open, whereas most of the type A pores in the filler particles are closed. 3.2 Pores in the Corroded Graphites Figure 2 shows pores photographed under UV light in an uncorroded graphite and the same graphite after thermal corrosion to 2.4% weight loss. Pickup et al.[7] showed that the only significant changes in the microstructure are the development of narrow, slit-shaped (type C) pores in the binder phase, typically l-5 p_rn in width. This observation was supported by quantitative image analysis which showed that the only pores developed by oxidation to about 2.5% weight loss are those with an area of less than about 100 +m2. The effects of radiolytic oxidation on microstructure are illustrated in Fig. 3 which compares an uncorroded graphite and the same graphite after corrosion to 4.2% weight loss. It can be seen that radiolytic corrosion is much less selective than thermal corrosion. Calcination cracks (type A pores) which are mainly closed in uncorroded graphite are opened and attacked in the early stages of oxidation. Type B gas entrapment pores are also attacked and new pores are created in the binder phase which are of lower

aspect ratio than the type C, slit-shaped pores developed by thermal corrosion. 4. 4.1 The mechanism

DKXUSSION

of thermaL and radio~yt~c

corrosion

Figures 2 and 3 illustrate the substantially different effects upon the microstructure due to thermal and radiolytic corrosion. In order to understand how these microstructural effects occur, it is necessary to consider the different corrosion mechanisms responsible for thermal and radiolytic corrosion of graphite. The selective development of the narrow type C cracks in the binder phase during thermal corrosion may be associated with the thermal reaction anisotropy of graphite. In thermally activated reactions with CO, and other oxidising gases carbon atoms in basal planes are many orders of magnitude less reactive than edge carbon atoms on prismatic planes{ 121. Pickup et al. [7] proposed that the formation of type C pores parallel to basal planes exposes prismatic carbon atoms at the edge of these pores, for example, at sites of disdinations. The selective nature of thermal corrosion will preferentially remove these atoms and so elongate this class of pore. Ionising radiation in nuclear reactors greatly enhances the chemical activity of the coolant gas compared to the thermally activated state, and consequently, it may be proposed that the oxidising species is much less selective in its attack upon the microstructure of the graphite than thermally activated CO,. The radiolysis of reactor coolants and the effects on gasification of graphite have been extensively studied[l3] and are very complex. Schematically the gasification reaction may be represented as CO? -J-L co: + e co; + c = 2CO

where CO: is the activated species, thought to be ion clusters such as (COJ;. The rate of gasification is controlled by the flux of activated species reaching the pore wall, which in turn is affected by the extent of deactivation in the gas phase and at the surface. In British AGRs CO, CH, and H,O are added as inhibitors. Addition of methane involves the deposition of carbon at the pore wall which may be sacrificially gasified by the active species[l4]. Recent work by Best and Stephen[9] has indicated that the effectiveness of surface inhibitors such as methane is limited by their accessibility to the pores. They studied radiolytically corroded AGR modertor graphite (IMl-24), using quantitative image analysis, and found after oxidation a rapid increase in the proportion of pores with an area-to-perimeter ratio of 3-7 p.m. They proposed that corrosion causes dosed pores to be opened. Access to these freshly opened pores is via small entrances and diffusion of the surface inhibitors, methane and water, is severely re-

microstructure and reduction of elastic modulus

547

Fig. 1. Microstructure of a needle coke filler-pitch binder graphite illustrating three types of pore. Position A, filler particle porosity; position B, gas entrapment oorositv in the binder: oosition C. narrow slit shaped pores in the binder phase. tolariseh light microscoby. stricted until further corrosion opens up the pore entrances. When this occurs surface inhibition will improve rapidly and the corrosion rate of these pores decreases. Murdie et al. [ 151, using quantitative image analysis also demonstrated that radiolytic corrosion of AGR Gilsonite coke moderator graphite causes open porosity in the binder phase to interconnect with the previously closed porosity in the grist particle. In contrast to thermal corrosion there was no evidence to suggest preferential elongation of pores. Elsewhere Murdie[l6] shows that the aspect ratio of the pores developed by radiolytic corrosion decreases with progressive gasification.

Fig. 3. Needle coke graphite. (a) Uncorroded. (b) Radiolyticaliy corroded to 4.2 wt loss. Polarised light microscopy.

4.2 Models for the effects of thermal and rndiolytic corrosion upon elastic modulus

Clearly radiolytic and thermal corrosion have different effects upon the development of pore structure in graphites. An inte~retation of the effects of microstructural changes upon the mechanical properties is possible through an analytical derivation of the Knudsen equation due to Buch(lO]. The Knudsen relationship, between elastic modulus, E, and fractional porosity, P, is, E = E, exp( - bP)

(3)

where E, and 6 are constants (E, is effectively the value of E at zero porosity). By considering far field displacement effects produced by introducing a single

Fig. 2. Needle coke graphite. (a) Uncorroded. (b) Thermally corroded to 2.4% wt loss. Contrast enhancement between pores and graphite using UV fluorescent microscopy.

T. D. BURCHELL et al.

548

pore into an elastic continuum, Buch showed that the modulus decrement was a function of pore aspect ratio, ale, which is related to the constant b by, b = 1 +

0.594 (a/c).

(4)

Pickup ef al. [7] used the Knudsen equation to relate experimentally observed modulus changes on thermal corrosion to porosity for a needle coke graphite: E = E, exp( -7.5P).

(5)

To derive an analogous equation for radiolytic corrosion it is necessary to transform eqn (2). This can be done if it assumed that radiolytic oxidation is entirely non-selective, and the external volume of the graphite remains unaltered after corrosion. These assumptions apply to a model for radiolytic oxidation which is analogous to drilling small holes in the graphite. There is some justification for this model. (i) The decrement in modulus upon radiolytic corrosion of a wide range of nuclear graphites can be represented by one equation eqn (2), suggesting the porosity development is insensitive to graphite microstructure. (ii) Kelly et al.[8] have shown experimentally that property decrements, similar to those found for radiolytic oxidation, can be obtained by drilling small holes in graphite. (iii) In studies of the radiolytic corrosion of graphite single crystals Feates(l71 found no evidence for reaction anisotropy for the CO, reaction in the presence of methane. If W, and wf are the initial and final weights of graphite, then the fractional weight loss after radialytic corrosion, x, is x = (w, - wr)Iwz and, if initial and final volumes are equal, then

Substituting the exponential factors from eqn (5) for thermal corrosion and eqn (8) for radiolytic corrosion into eqn (4) gives pore aspect radius of a/c = 11 for thermal corrosion and a/c = 6 for radiolytic corrosion. These pore aspect ratios must be regarded as notional since they are obtained using (i) a simple model for the effect of a single pore on the modulus of an elastic continuum and (ii) a simple model for radiolytic corrosion, However, these estimates are in qualitative agreement with the microstructural evidence (Fig. 2) for the development of high aspect ratio pores on thermal corrosion, and supports the observation that radiolytically corroded pores tend to be more equiaxed than pore developed by thermal corrosion (Fig. 3). 5. CONCLUSIONS The principal microstructural change on thermal corrosion of nuclear graphites is the development of fine pores of high aspect ratio in the binder phase. This selective attack is ascribed to the thermal reaction anisotropy of the graphite. The more energetic gaseous species participating in radiolytic corrosion are much less selective than thermally activated CO*, thus pores of lower aspect ratio are developed. Using a simple model which gives a theoretical derivation of the Knudsen equation and a simple model for radiolytic corrosion, the more severe effects of thermal corrosion on elastic modulus are related to the selective development of pores of high aspect ratio. Acknowledgements-We

thank the SERC and CEGB for financial support and Mr. T. J. Mays for useful discussions.

REFERENCES

where pI and pr are the initial and final bulk densities of the graphites; thus x = 1 - (P,lP,).

(6)

If pc is the crystal density of graphite (2.26 gicm3), then initial porosity, P,, is defined as

1. B. T. Kelly, in Physics of Graphite. Applied Science Publishers, London, 135-136 (1981). 2. J. E. Brocklehurst, in Chemistry a&Physics of Carbon. (Edited bv P. L. Walker. Jr. and P. A. Thrower). Vol. 13, Mar&I Dekker, New York, 145-295 (1977):’ 3. N. Hawkins, in Proc. Second Conf. on Industrial Carbon and Graphite. Sot. Chem. Ind., London, 355-366 (1966). 4. J. A. Board and R. L. Squires, in Proc. Second Conf. on Industrial Carbon and Graphite. Sot. Chem. Ind., London, 289-297 (1966). 5. C. Rounthwaite, G. A. Lyons and R. A. Snowdon, in Proc. Second Conf. on In&trial Carbon and Graphite. Sot. Chem. Ind., London, 299-318 (1966).

P, = 1 - (PI/PC>

6. J. E. Brocklehurst,

and P, is similarly defined. Thus eqn (6) becomes: 7.

x = 1 - [(l - P,)l(l

- Pi)] 8.

and re-arranging 9. Pr =

P, + x(1 - PJ.

(7)

10.

Equation (7) may be used to transform eqn (2) to an equation relating elastic modulus, E, to fractional porosity, P,

11.

E = E, exp(-4.5P).

(8)

R. G. Brown, K. E. Gilchrist and V. Y. Labaton, .I. Nucl. Marl. 35, 183 (1970). I. M. Pickuo, Ph.D. thesis. Universitv of Bath, England (1984); I. M. Pickup, B. McEnaneyand R. G. Cooke, submitted to Carbon 24,535 (1986). B. T. Kelly, P. A. V. Johnson, P. Schofield, J. E. Brocklehurst and M. Birch, Carbon 21, 441 (1983). J. V. Best and W. J. Stephen, in Carbon ‘82. Sot. Chem. Ind., London, 341-343 (1982). J. D. Buch, in Extended Abstracts, 16th Carbon Conference. San Diego, American Carbon Society, 400401 (1983). C. J. Wood and A. J. Wickham, Nucl. Energy 19, 277 (1980).

12. J. B. Lewis, in ModernAspects of Graphite Technology. (Edited by L. C. F. Blackman). Academic Press, London, 129-199 (1970).

Microstructure

and reduction of elastic modulus

13. A. J. Wickham and J. V. Headley, Publication No. TPRDIBi0247iN38, Central Electricity Generating Board, England (1984). 14. J. V. Sherman, in Gas Chemistry in Nuclear Reactors and Large Industrial Plant. (Edited by A. Dyer), Hey den, London, 99-110 (1980).

549

15. N. Murdie, I. A. S. Edwards and H. Marsh, In Carbon ‘82. Sot. Chem. Ind.,. London, 350-352 (1982). 16. N. Murdie, Ph.D. thesis, University of Newcastle upon Tyne, England (1985). 17. F. S. Feates, Trans. Faraday Sot. 64, 3093 (1968).