NUCLEAR FUELS B. R. T. Frost, B.Sc., Ph.D., F.I.M. Metallurgy Division, Atomic Energy Research Establishment, Harwell, Didcot, Berks.* Fuel manufacture . . The fuel cycle, 168 Ore to metal or ceramic fuel, 168 Metal fuels, 169 Uranium dioxide, 17 1 Carbides, 175 Other ceramic fuels, 176 Reprocessing, 177 . . . . . . . . . . . . . . 190 - Conclusion . . . . . . . . .. .. .. .. . . 204 References . . . . . . . . . . * . .. . . . . . . .. . . . . 168 Physical and chemical properties. . . . .. . . . . . . 181 . Physical properties, 18 1 Chemical properties, 185 Irradiation behaviour 2ggTh + In0 -+ 2iJ;Th + y 2$iTh -+ 0/3- + ";Pa ";Pa -+ 0/3- + 233U 92 It has taken only 27 years since the discovery of nuclear fission to establish the nuclear reactor as a cheaper source of electrical power than coal or oil (see Fig.I ) . 1 The unique feature of a reactor is its core in which a self-sustaining fission process is used as a convenient heat source. In the great majority of reactors the fissioning species is 235U which occurs in natural uranium at a concentration of 0.7 per cent, the remainder being 238U. The latter will undergo fission when bombarded with neutrons of high energy but more generally it absorbs neutrons to produce 239Pu by the reactions: 239Pu is in many respects a better fuel than 235U so that this breeding process enlarges our energy resources considerably by utilizing the otherwise useless 238U. The energy resources may be further enlarged by producing fissionable 233U from thorium: * Present address: Metallurgy Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439.163 .. .. . . 205 Frost 1.5 I I I I I I I I I I I I I Nuclear (plagnox) Coal-fired (remote from central coal fields) 2 Y 0, L n aJ aJ a C 0.5 / Oil-fired (including oil tax) I ! I975 0 0 0 Coal-fired (near central coal fields) I I I I I I I I I I I I970 - w I 1 Year of commissioning I965 Fig. I. Nuclear power stations now produce electricity at a lower cost than coal-fired plants serving t h e same parts of the country. The Magnox reactors, on which the first 5000 MW nuclear programme of the CEGB is based, use uranium metal fuel of natural enrichment and will produce considerable quantities of 239Pu as a by-product.This will have to be separated by chemical means from the uranium and fission products and will almost certainly be used as the fuel in fast breeder reactors, the prototype of which is being constructed at Dounreay. The Advanced Gas-Cooled Reactor (AGR), large-scale commercial versions of which are under construction at Dungeness and Hinkley Point, uses slightly enriched uranium in the form of UOZ pellets. The enrichment of the 235U isotope is performed in a process which is dependent on the slightly different rates of gaseous diffusion of 235U + nl + 95X + 139Y + 2nl I .o 0 238UFg and 235UFg. A fissionable atom, e.g. z35U, located on a crystal lattice in, say, a pellet made of sintered UOZ crystals undergoes fission when a neutron of suitable energy is absorbed.This occurs because the resulting nucleus is very unstable and splits into two parts of roughly equal mass, e.g. There is a discrepancy in mass between the two sides of this equation which corresponds to an energy release of about 200 MeV.* Most of this energy is imparted to the two fission fragments, which leave the site of fission very rapidly, travelling in straight lines of length about 10 pm before coming to rest. In doing so they impart their energy to the parent lattice, essentially as thermal vibrations, and incidentally causing considerable damage to the ~ ~~ ~~~ _ _ _ _ _ - ~ - _ _ _ ~ ~ _ _ _ _ _ J. * 1 MeV is approximately equal to 1.602 x R.I.C.Reviews 164 lattice. It is this energy, plus a smaller amount arising from other sources, which represents the heat source in the nuclear fuel which must be converted to a more useful form. Until now this has been done by using the heat to make high-temperature steam (typically 560 "C and 2400 lb in-2 pressure) which drives a turbine connected to an alternator (Fig. 2). In this method, nuclear fuel is being used as a direct alternative to coal or oil and there is a similar incentive to derive the maximum Carnot efficiency by working at high temperatures. The nuclear plant usually costs more than the fossil fuel plant to build (although there is recent evidence to suggest that this is no longer the case)2 and it is the cheapness of the nuclear fuel compared with coal or oil which makes its electricity generating costs competitive.It is possible to use nuclear heat in other ways. Two important examples are : 1. As a direct producer of dc electricity. If the fuel is a suitable electron emitting material, e.g. UC, it can be made to form the cathode of a diode and so produce electricity directly. The fuel surface must be at a temperature in excess of 1500 "C to produce an efficient diode. Experimental prototypes have operated successfully for many hours.3 2. As a source of process heat. An outstanding and fast-developing example of this application is in the desalination of sea water by flash distillation. A large nuclear plant might produce both fresh water and electricity on an economic basis in arid regions.4 The economic use of nuclear fuel depends on its ability to remain dimension- ally and chemically stable for a long period of time in a very unusual environ- ment. Firstly, the fuel limits its own life by the production of fission products which occupy a greater volume than the original uranium atoms and cause damage to the fuel material.Ten per cent of the fission product atoms are feed pump 165 Fig. 2. A schematic diagram t o show the conversion of nuclear-generated heat into energy via high-pressure st earn. Frost 12 noble gases (Xe and Kr) and if these congregate together to form bubbles they will cause an even greater volume change. At the same time the fuel must operate at a high surface temperature (-600 "C) to give a high thermodynamic efficiency and it must be reasonably inert chemically to the coolant, which may be a gas such as COz or a liquid such as sodium.Thus, the qualities required of a nuclear fuel are: (i) high melting point, (ii) high thermal conductivity, (iii) high density of uranium atoms, (iv) absence of highly absorbing species, e.g. boron or gadolinium, and (v) chemical inertness to oxidizing liquids and gases and to metal cladding materials. These conditions are most closely met by ceramic fuels, the properties of which are compared in Table 1. However, the first developments were con- cerned with metallic fuels. The earliest type of reactor to be developed in the UK on a large scale was gas-cooled and graphite moderated.To achieve criticality on a natural uranium fuel it was necessary to use the metallic form since this has the Table I. Comparison of ceramic fuel materials Fuel Metal uo2 PUOZ Cubic (f I uo r i t e) UOZ-~OPUOZ Solid uc PUC UC-2OPuC UN Pu N UN-20PuN PUP a At fuel operating temperatures us PUS UP 166 f 30 I660 2480b 2850 i 30 2600 (I atm) > 2500 2450 i 30 2330 2610 f 30 2600 (2 atm) Density density Melting expansion conductivity Structure (g cm-3) (g cm-3) point ("C) ( x 106 "C-1) (cal cm-1 s-1 "C-1) 9.66 10.96 Cubic (fluorite) 10.12 2750 It 40 2280 11.46 2740 9.76 I I .06 solution FCC 13.63 2350b 12.97 12.87 13.53 13.6 12.9 FCC Solid solution FCC 14.32 13.53 14.25 FCC 13.47 14.3 10.86 FCC 13.5 9.57 10.57 10.23 FCC FCC 9.35 9.07 9.89 FCC 8.77 10.8 0.018 (100 "C) (to 800 "C) 0.008 (700 "C) 9.3 10.3 9.7 11.9 9.2 (to 800 "C) 13 8.6 (to 980 oc) 9.7 b 2000 ppm oxygen.Coefficient of thermal Thermal - 0.005-0.007a 11.5 0.05 (to 870 "C) (I 00-700 "C) - 0.046" 0.04 (300 "C) 0.05 (800 "C) - 9.8 0.043" 11.8 0.028 (100 "C) (to 1000 "C) 0.041 (800 "C) 18.5 - 0.032 (100 "C) 0.039 (800 ocj - R. I.C. Reviews maximum density of uranium atoms. Uranium metal does not have a very high melting point (1 133 “C) and it exists in three different crystallographic forms in the solid; temperature excursions from one form to another cause a change in shape of the crystals.This phenomenon is known as ‘growth’ and has imposed limitations on the use of uranium metal. The metal is also chemically reactive, oxidizing readily in air, C02 and water. Finally, its low melting point together with the internal stresses set up by growth lead to a high fission-gas swelling rate. Notwithstanding these limitations the Magnox reactors have provided a reliable and increasingly competitive form of power, as Fig. I illustrates. The development of water-cooled reactors, initially as submarine propulsion units in the USA, led to the realization that U02 was, in many respects, a better fuel than uranium metal. Although the low fissile-atom density in U02 generally requires the use of enrichment, this is offset by the chemical inertness, very high melting point (2800°C) and lack of phase transformations in the solid state.5 Thus, it is possible to burn several times the number of 235U atoms in oxide to produce the same swelling as in metal.A further advantage of UOZ is that it is isomorphous with Pu02 and Tho2 and solid solutions of UO? Pu02 can be used as a fast reactor fuel and of UO2-ThO2 for thermal breeders. However, a major disadvantage of U02 is its low thermal conductivity which leads to high fuel operating temperatures-a temperature in the centre of the fuel pellets of 2000°C is not uncommon. Under these conditions a high proportion of the gaseous fission products move out of the fuel and pressurize the metal can which separates the fuel from the coolant, thereby complicating fuel element design.There are two answers to this problem. One is to disperse the oxide in a metallic matrix, e.g. stainless steel, which improves the thermal conductivity and restrains the swelling of individual oxide granules.6 These ‘cermet’ fuels are used in special applications, e.g. research and military reactors, where reliability is more important than economic operation. A second solution is to use uranium carbide which has nearly as high a melting point as U02 (2350°C) but a thermal conductivity five to 10 times higher. The carbide is more chemically reactive (comparable with uranium metal) but has a higher density than UOZ. Its use, therefore, has been confined to reactors using relatively inert coolants : sodium and terphenyls.A special application of carbide is as a dispersion of uranium-thorium dicarbide (U,Th)C2 in graphite. In the Dragon type of high temperature gas-cooled reactor the fuel element contains the fuel and moderator, thus minimizing graphite radiation- damage problems and avoiding the use of neutron-absorbing metal cladding. A unique feature of this type of fuel element is the coating of each dicarbide particle ( 4 0 0 pm diameter) with pyrolytic carbon to retain the fission products and minimize fission product release.’ From this brief discussion it can be seen that different reactor types demand the development of different types of fuel. The development of a fuel can be a lengthy and complex process, starting with the preparation of pure material from the ore and its fabrication into suitable shapes, followed by the evaluation of physical and chemical properties and a study of irradiation behaviour.Finally, when a fuel element design has been produced it must be tested Frost 167 rigorously on a statistical basis, generally in a prototype power-producing reactor such as the Calder Hall Magnox reactors and the Windscale AGR. These steps in the evolution of a fuel element will be traced in some detail in the remainder of this review. FUEL MANUFACTURE The fuel cycle Nuclear fuels have a high specific value, that is they are expensive to produce in terms of cost per gram but each gram is a potential source of much energy. Thus, a kilogram of 235U costs about E5000 but can theoretically produce 2.4 x 1010 kilowatts of heat when burned in a reactor.The conservation of fuel and its strategic use are important considerations in power reactor technology and the fuel cycle must be seen as one of the three important factors affecting the economics of nuclear power, the other two being the capital cost of the reactor plant and the operating costs. Figure 3 shows a schematic diagram of a fuel cycle. Ore to metal or ceramic fuel Uranium is most commonly found as uraninite, a form of pitchblende, which is basically a mixture of uranium oxides associated with lead, lanthanons and thorium.8 Commercial ores, as mined at present, contain at least 0.1 per cent u308. World reserves of uranium from this type of ore are estimated as several million tons.Lower grade ores are known to exist in large quantities and the dissolved salts of uranium in sea water, which are at the ppm level, represent a source of lo9 tonnes of uranium.9 However, extraction of uranium from these more dilute sources will involve more expensive extraction processes and will only be exploited when the supply of more concentrated ores is exhausted. Initially, the ores are crushed and concentrated physically or by roasting and then the uranium is dissolved by acid or sodium carbonate leaching. The crude liquor, containing about 5 g 1-1 of uranium, is purified by an ion exchange process, and then extracted into tributyl phosphate, the impurities being discarded in the aqueous phase.The pure uranyl nitrate solution which results is converted to uranium dioxide. In the UKAEA process,lO ammonia is added to Fig. 3. Schematic diagram of a fuel cycle. R.I. C. Reviews 168 Reactor Storage hoppers / ng t Spra To reduction reactor jacket heater ija; i pipe + Internal ' I1 heaters Lift pot 1 -% To scrubber From co m pressor Heat exchanger Fig. 4. Schematic diagram of the process for the denitration of uranyl nitrate. the nitrate to precipitate ammonium diuranate (ADU), (NH4)2U4013. This may be decomposed, by heating, to U03 or US08 which is reduced in hydro- gen to U02. The route chosen depends on the end product required. For the production of metal, ADU is decomposed to U03 below 350°C and reduced with hydrogen at about 700 "C.For ceramic oxide, ADU is decomposed above 450 "C to u308 which is reduced in hydrogen at -700 "C to give a powder with a lower reactivity or surface area per unit weight. In the USA, uranyl nitrate is evaporated to form molten uranyl nitrate hexahydrate which decomposes to U03 at 300 "C (Fig. 4). uo3 is then reduced with hydrogen at 600-800°C. These processes may be operated in a fluidized bed to give a continuous throughput. To produce uranium metal, U02 is converted to UF4 by treatment with HF: U02 + 4HF + UF4 + 2H20 -AH298 = 265.6 kJ mol-1 The UF4 is then reacted with high-purity calcium or with magnesium in a 'bomb', the efficiency of conversion being about 97 per cent. This is carried out by UKAEA at its Springfield Works on a 200 kg batch scale.The processes just described refer to the use of uranium at its natural isotopic composition. In many reactors it is necessary to enrich the z35U isotope con- centration to achieve criticality. This is done by gaseous diffusion of UF6 through a cascade of membranes. The UF6 is made by fluorination of UF4 between 450 and 550 "C (Fig. 5 ) and, after the diffusion process, is reconverted to UF4 by reaction with hydrogen in a fluidizing column or is hydrolysed to UO2F2 and reacted with aqueous ammonia to form ADU, which is then converted to U02 as described above. Metal fuels Enriched or natural uranium metal billets are melted by high-frequency induction heating under vacuum in magnesia-coated graphite crucibles and are Frost 169 I I L Fig.6. Cutaway view of a Mk 5 fuel element showing i t s method of construction. R. I.C. Reviews 170 Fig. 5. Flow diagram for uranium hexafluoride production. t F6 Stack cast into rods close in size to the final fuel rod. These are then machined to the final dimensions. For certain applications uranium is alloyed with other metals to produce fuels with improved performance or resistance to corrosion. For example, the ‘driver’ charge in the Dounreay fast reactor consists of highly enriched uranium alloyed with molybdenum. This allows it to operate to higher burn-ups than would otherwise be possible. The fuel elements in water- moderated research reactors of the DIDO type in this country and the MTR and ETR types in the USA are made from uranium-aluminium alloys.The two metals are melted together and cast into rectangular blocks which are then clad in aluminium and rolled into sheets which can be formed into desired shapes (Fig. 6). The uranium is in the form of small particles of UAl4 or UAl3 dispersed in a matrix of aluminium. In this way the swelling effects of the fission products are minimized and the uranium cannot react with the water coolant (vide infra). Uranium dioxide UOZ is mainly used in the form of dense, sintered cylindrical pellets, typically 1-2 cm in diameter. These are generally made by cold pressing and sintering UOZ powder, although other methods such as hot pressing are occasionally used. The ideal characteristics of UOz powder are:5 ( i ) a small particle size, (ii) regular shape, (iii) a ‘clean’ surface, i.e.a minimum of adsorbed gases, and (iv) a large surface area per unit mass. To some extent i and iv are incompatible with iii. The characteristics of the powders obviously depend upon the nature of the material from which they are formed.12 When U03 and u308 are reduced to U02 there is a large decrease in specific volume which leads to particle fracture and an increase in surface area. Careful control of the reduction temperature is needed to ensure that this effect is not counteracted by coalescence. Generally, UOZ powder from the reduction process is reduced in size by grinding in a rod or ball mill for several hours. Care must be taken not to pick up ceramic particles in this process.This can be done by proper choice of mill materials; rubber-lined mills are often used. In Fig. 7 the effects are shown of the powder preparation route, the precipitant and the milling procedure on UOz particle size. After milling, the powders are not generally free flowing and therefore will not fill dies with the same quantity of material each time. The powders are treated by the addition of 1-2 per cent of binder, generally an organic fluid such as polyethylene glycol, paraffin wax or polymethyl methacrylate ; this may be added dry or in solution. The powders are then granulated through sieves to produce a free flowing powder which is poured into tungsten carbide lined dies and compacted under a pressure of up to 150 ton in-2.To ensure uniform densification of the powder and for ease of loading into stainless- steel tubes, UOz is generally used in the form of right cylinders. The ‘green’ pellets are debonded (i.e. the binder is removed) by heating in an inert or 171 Frost 90 - A-High-pressure steam oxidation B-MCW PWR core-I UO, 8o - C-MCW Precipitated U02 - 60 50 70 - D-Fluidized-bed denitration, UO, reduction - .- 100. M P 2 9 0 - W 2 80- 7 0 - C Q) 60 - - 50 a 40- 30 - - 20 - 10 0 Fig. 7. Uranium dioxide particle size distribution: (a) the effect of powder preparation, and (b) the effect of milling procedure. slightly oxidizing atmosphere at about 800°C. They are placed on molyb- denum trays and heated in a hydrogen atmosphere in molybdenum wound furnaces to promote sintering; several hours at 1500-1700°C are required to produce high-density pellets.13 Generally the aim is to produce pellets with about 5 per cent residual porosity ( i e .95 per cent of theoretical density). The porosity is then mostly within the grains and is ‘closed’, i.e. not accessible from the outside of the pellet. The reasons for choosing this density is discussed later when properties are considered. Figure 8 shows the effects of various process variables on the final sinter density of UOz. The most common form in which ceramic fuels are used is as right cylindrical pellets inside metal tubes. However, the quest for improved performance and 172 (a) - - - - - - - - - (b) - - - - - - - - 6-U rea-preci pitated diuranate UO, C-’B’ milled D-’D’ milled - I 1 I R.I.C. Reviews 74 72 90 'i 70 - t 88 86 68 Condition: as received - 66 U U f- + 1% PVA Ram diam., 64 E 9 a 0.41 I in Sintered for IOh 82 62 200 I Si ntered: 50 I 1800 a t 1725OC in H, I00 Compacting pressure (ton in-,) Compacted a t 100 ton in-, Sintered in H, I I 1700 1600 Si nteri ng temperature ("C) 5 s 25 86 I I t 1400 1500 Fig.8. Effects of processing variables on the sintered density of MCW UOz compacts. (a) As pressed and sintered density as a function of compacting pressure. (b) Sintered density as a function of compacting pressure and sintering time.(c) Effect of time and temperature on sintered density. (d) Sintering time as a function of compacting pressure and sintered density. economics has led to the development of other fuel forms, the most important being vibrocompacted fuel and coated particles. If fuel swelling considerations (see below) dictate the need for a low fuel density it may be uneconomic to make carefully sized pellets. It may be cheaper to make sintered spheres of about 500-1000 pm diameter and pack these into metal tubes;14 a single size of sphere gives a maximum packing density of about 74 per cent. By adding smaller particles, ideally reducing in diameter by multiples of seven, densities up to 90 per cent of theoretical can be made.Packing is achieved by vibrating the metal tube mechanically. 200 300 Compacting pressure (ton in- 2 ) Frost Sintered at 1725 OC t '& 90% dens^^^^ 60 74 ' I 40 as received + Condition: I yo PVA in H2 I 1 I l l 100 Compacting pressure (ton in-2) 100 200 250 - 5 I- I 92% dens: \ ' \ \ \ \ \ \ 194% \ dense -1 I 60 30 i 173 As explained later, improved fuel performance can be obtained by dis- persing fuel particles in an inert, conducting matrix. Stainless steel-UO2 cermets have given outstanding performance in small military reactors where fuel cycle economics are not of primary importance. In the Dragon type of high-temperatureereactor the fuel particles-oxide or carbide-are dispersed in graphite moderator.To restrict fission product release the particles are coated with carbon which is deposited pyrolytically from hydrocarbons in a fluidized bed (Fig. 9). The fabrication of oxide fuels enriched with plutonium instead of 235U is similar to that of UOZ. Plutonium, unlike uranium, can exist in the trivalent state so that a wider range of oxygen : metal ratios is possible than in UO2 and care must be taken in selecting the sintering atmosphere. The mixed oxide powder may be prepared by physically mixing PUOZ and U02 powders; the PuOz is made from nitrate solution by precipitation as the hydroxide followed by calcining. Alternatively, plutonium hydroxide and ADU may be coprecipi- tated from a mixed nitrate solution, calcined and reduced in hydrogen.The resulting powders are then pressed and sintered in the same way as U02, the sintering atmosphere being argon or helium-hydrogen mixtures or wet hydrogen.15 During these operations, and indeed in all operations on enriched fuel, precautions must be taken to limit the concentration of fissile material in localized areas to well below that required to form a critical mass. Generally this involves careful control of fuel movement. All steps in the fabrication of plutonium fuels must be carried out inside glove boxes to isolate the toxic &-active plutonium. In the future, when plutonium is recycled after irradiation to high burn-ups, appreciable levels of y-activity will arise and operations will have to take place in ~$7 shielded cells.Fig. 9. Photomicrograph of fuel particles, showing pyrolytically deposited carbon coating. R. I , C. Reviews 174 uoz + 3 c + u c + 2 c o 1' u02 + 4 c +uc2 + 2 c o 1' Carbides Of the three compounds in the uranium-carbon phase diagram, the mono- carbide is of interest in fast reactors and the dicarbide in high temperature gas- cooled (Dragon) type reactors. These carbides may be prepared by at least three methods : (i) direct reaction between uranium metal and carbon-this is difficult to control ; (ii) reaction of uranium, made reactive by hydriding, with hydrocarbons16 -this works at 600-800°C but it involves the use of uranium metal, introduc- ing an undesirable extra step in the fuel cycle; and (iii) UOz-carbon reaction-this is the most commonly used route and can be employed to make monocarbide, and dicarbide, To drive the reaction forward the CO must be removed rapidly; this is done either by reduction i n vacua with a high pumping speed17 or in a fluidized bed with a rapid flow of argon to entrain the CO.18 The reaction is generally carried out at 1400-1 500 "C.The product contains between 0.2 and 0.4 wt per cent oxygen and less than 100 ppm nitrogen, both in solid solution, and is very reactive to air and particularly to water vapour. From this stage onwards the fabrication technique is generally similar to that for UOZ except that all operations, whether with plutonium or not, must be carried out in dry atmospheres in glove boxes. In experiments at Harwell, Russell and Harrisonlg crushed reacted (U,Pu)C to - 30 mesh and milled the powder in a tungsten carbide mill with tungsten carbide balls. The powder was then pressed into pellets at 40 ton in-2 and sintered at 1550 "C for four hours in flowing argon.The final density was -94-95 per cent of theoretical; this was only achieved by taking precautions to eliminate surface oxidation of the powder which had been milled to about 2 pm particle size. High sinter densities can be achieved more easily by adding a small amount of nickel to the powder prior to pressing. In experiments at the United Nuclear Corporation in the USA 0.1 to 0.5 per cent nickel was added and densities as high as 98 per cent of theoretical could be obtained without difficulty;20 the mechanism involved has not yet been elucidated.UC has a lower melting point than U02 (-2350 "C as compared to 2800"C), is a good conductor, and does not have very high vapour pressure at its melting point. It is possible, therefore, to make carbide fuel by arc melting uranium and carbon together and casting into moulds. A considerable amount of development of this route was carried out by Atomics International in the USA for a UC fuelled thermal reactor of the sodium-cooled graphite- moderated type.21 Rods up to 1.0 in diameter were cast on a production scale for incorporation into long fuel elements. Arc-cast carbide differs from the sintered product in having a much larger grain size and a lower oxygen and nitrogen content.With plutonium fuels the loss of plutonium by vaporization is high and although this route has been studied it has been dropped in favour of the sintering method which has the advantage that it can be carried out in 175 Frost Final 3 so I evaporator column M (L___JI 1 s z Thermal denitration vessel I I g& Aqueous waste tank uo2 Sol feed tank n Other ceramic fuels Fig. 10. Schematic flowsheet for the production of (U,Pu)O2 spheres. equipment built for oxides, with the simple addition of the carbothermic reduction step. The coated carbide fuel particles used in Dragon-type reactors are generally not UC2 but (U,Th)Cz since the original objective was to convert thorium to 233U. Three methods have been used to make fuel particles. 1.Powder agglomeration. UOZ, Tho2 and carbon flour are mixed together with a binder and granulated to form green granules of -500 pm diameter. These are reacted and sintered either under vacuum at about 2000°C or in a fluidized bed at -1 700-1 800 "C. The latter is preferred since, after the reaction, hydrocarbons can be introduced to coat the particles with carbon. 2. Melting in an arc or plasma, or in a bed of carbon. The granules so produced contain shrinkage cavities and high internal stresses. 3. Sol-gel process.22 This has been developed primarily for oxide fuel but can be used for carbides. Oxide is dispersed in an electrolyte to form a sol. Droplets of the sol are dispersed in an immiscible liquid and are hardened or gelled by reaction with a suitable reagent, e.g.2-ethylhexanol. The spheres are then dried and calcined. A flowsheet developed at Oak Ridge National Laboratory for the production of (U,Pu)Oz spheres is shown in Fig. 20. To make carbides, carbon black is added at the sol stage and the final calcining temperature is higher. The advantages of this process are that it produces spheres continuously in simple columns, the product being very regular in size and shape and possessing a very fine grain size which facilitates sintering. Laboratory studies have been made and are continuing on uranium nitride, phosphide and sulphide which are similar in structure to UC.23 Generally the G3-l I Reject 1 *duct Roundometer 176 '.LJ -9 Drier 1 1 1 Furnace i j Screens I R.I. C. Reviews starting point for their preparation is uranium, prepared in a reactive form by first forming UH3 at 250-270 "C. The temperature is raised to 400-600 "C and a suitable gas admitted (nitrogen, phosphine or HzS). The hydride decomposes to UN + N2 at 1400°C in argon. With phosphine, U3P4 is formed and is and a compound is formed. With nitrogen, UN1.7 is formed and is decomposed decomposed to UP at 1400°C. HzS forms ,%US2 which decomposes to a mixture of US2, u2s3 and U3S5 at 1400°C and to US at 1800°C. The powders may be pressed and sintered in a similar manner to UC. Reprocessing The life of a nuclear fuel in a reactor core is generally dictated by the nett burn-out of fissile material, the production of neutron-absorbing fission products, and dimensional changes in the fuel due to the conversion of heavy fissile atoms to less heavy fission products.An additional factor which may over-ride these considerations is the production of plutonium either for military purposes, in which case the need is to keep the higher isotopes of plutonium to a low level, or in the blankets of fast breeder reactors where the objective is to produce plutonium rather than power. With fuels irradiated in thermal reactors the objective of reprocessing is to separate the uranium, the plutonium (or z33U if thorium is presbnt) and the fission products. The separated uranium is either made up to the required enrichment and recycled through the thermal system or it may become the breeder of a fast reactor. The plutonium is probably used in a fast reactor although, until their technology is firmly established, plutonium may be used to enrich thermal-reactor fuel.This is certainly the intention in the USA; the UK, however, plans to introduce fast reactors as early as possible and will reserve its plutonium for this purpose. In fast reactor fuels the main requirement is to remove the fission products. Since the fuel contains both uranium and plutonium these need not be separated in reprocessing although the plutonium concentration will need adjustment. There are basically two types of reprocessing : hydrometallurgical-which operates below -100 "C-and pyrometallurgical or high-temperature pro- cessing.The first stage in hydrometallurgical reprocessing is to dissolve the fuel in nitric acid. Ideally the fuel should be separated from its cladding and in the Magnox reactors which use metal fuel this is relatively easy; with ceramic fuels, particularly those in dispersed forms, it is necessary to chop up or crush the whole assembly and to leach the fuel from the structural material. The separation of uranium and plutonium from the fission products is based on the solubility of the two actinides in their four-valent state in a number of organic compounds such as diethyl ether, tributyl phosphate and Butex (/3,/3'- dibutoxydiethyl ether). Plutonium may then be separated from uranium by reducing it to a lower valency, e.g. by ferrous sulphamate. The process is engineered by means of countercurrent contactors which may be packed columns, pulsed columns or mixer-settlers.24 The process is shown diagram- matically in Fig.11. The aqueous (nitrate) solution is fed into the middle of the columns and flows upwards. This phase, rich in U and Pu, flows to the plutonium stripping column and finally to a uranium stripping column; the Frost 171 Demin. Demin. Demin. Sodium water carbonate Demin. Water HNO, I Nitric a c i d 0 Plutonium Demin. water ~~~~~~ Solvent Solvent Uranium solution solution water plutonium reductant Metal fuel elements l r I r rei3 Scrub . I i Solvent: tributyl phosphate *ea Flow sensing device (orfice or remote indicating rotameter) Flow control motor solution To To To waste disposal uranium processing plutonium processing Fig.1 I. Processing flowsheet for separating Pu, U and fission products by solvent extraction. material to prevent accidental build-up of insolubles, the prevention of excessive radiation decomposition of the organic solvent, and the disposal of the radioactive wastes. A view of part of the Windscale plant is shown in Fig. 12. This account deals only with general principles and scarcely does justice to an important and complex subject; the interested reader should see references 24 and 25. Falls in the USA.26 The simplest variant of this type of processing is melt refining where a metallic fuel is melted under vacuum in oxide crucibles. The more volatile fission products vaporize and are trapped elsewhere while the rare earths and yttrium form a slag or dross with the crucible material.The remaining melt, which comprises uranium, plutonium and the noble fission products such as ruthenium and molybdenum, may be filtered or bottom- poured into moulds. There is generally a remelting and casting step where the 179 Fig. 12. Part of the Windscale, UK, recovery plant. important product in each case is an aqueous solution of the actinide and the organic phase is sent for recovery and recycle. The uranium and plutonium solutions are concentrated, after adequate decontamination, and converted to ADU and hence to U02 or to plutonium hydroxide and Pu02. Important considerations in the design and operation of such plants are the prevention of the formation of critical masses of solutions, accurate accounting of the fissile Aqueous reprocessing has become the standard method world-wide.However, a considerable amount of effort has been devoted to alternative processes, most of which are pyrometallurgical in character. Indeed, one such process has operated as an integral part of the EBR-I1 fast reactor at Idaho Frost plutonium content is adjusted. The product is a U,Pu-fissium alloy which is used as the fuel for the EBR-I1 reactor. This fuel does not perform very well from an irradiation standpoint : oxide and carbide give better performance, but there is a problem of whether they can be accommodated in a high- temperature process. In one process developed in the USA the oxide fuel is reduced to metal with magnesium at -750°C and dissolved in a liquid metal such as zinc.By a fractional crystallization process the uranium is separated (as U-Zn compounds) from the fission products and the zinc is finally removed by vaporization. Alternatively the molten U-Pu alloy may be con- tacted with molten salts to remove the fission products and eventually con- verted back to oxide or made into carbide. A notional scheme for such a process is shown in Fig. 13. The other type of process which has received considerable attention is the ‘fluoride volatility process’ in which oxide or carbide fuel is converted to hexafluoride in a fluidized bed.27 This separates the U,Pu and volatile fission- product fluorides from the non-volatile fluorides.UF6 and PUF6 are then separated from the fission-product fluorides by fractional distillation and finally converted to the oxide or to the tetrafluoride and hence to metal and carbide. A flow sheet for a carbide process is shown in Fig. 14. Economic assessments show that where aqueous processing is firmly established, as in the UK and the USA, these high-temperature processes have little to offer in the way of improved fuel cycle economics although they may offer an advantage in reducing fuel cooling times and hence reducing fissile inventories. The waste products from chemical reprocessing, i.e. the fission products, present a problem in their storage and disposal. At present the common Fig.13. Scheme for pyrochemical reprccessing of oxide and carbide fuels. R.I.C. Reviews 180 practice is to reduce their bulk by evaporation and store them indefinitely in large tanks. It has been shown that they can be adsorbed on clays which are then fired to produce a glassy product. The rate of leaching from this material is low and it can probably be dumped safely in ocean deeps. The fission products represent an increasingly important source of radio- active isotopes. 90Sr is extracted and converted to strontium titanate-a stable compound which is used as an energy source in thermoelectric generators, such as the Ripple 5 W generator which powers navigation lights in remote places.28 The design of fuel elements cannot proceed without a detailed knowledge of the physical and chemical properties of the fuel, the latter being important also in the development of fuel fabrication and reprocessing routes.The problems may be illustrated best by confining the discussion to those proper- ties of importance to the design of ceramic fuelled elements of the type used in the AGR and the SGHWR. Their design is a compromise between a number of factors. In particular the cladding must remain intact to burn-up levels at which the decrease in fissile atom concentration and the increase in fission product absorption cause a drop in power output. Neutron economy dictates that the cladding must be as thin as possible; therefore, it must be stressed as lightly as possible, i.e. fission gas release and swelling must be low, and corrosion losses should be small.These aspects are considered in more detail in the remainder of this review. Frost Fig. 14. Simplified flow diagram for fluid-bed fluoride-volatility processing of carbide core and blanket from a fast reactor. PHYSICAL AND CHEMICAL PROPERTIES Physical properties Gas release and swelling are temperature-dependent phenomena and, hence, a detailed knowledge of the thermal conductivity of fuel is essential in order to 181 13 h - I W M -0 '6 0.07 .- c, x .e 0.06 5 3 0 8 C.05 (d 0.04 - E E 182 calculate temperature distributions.29 U02 is an extrinsic semiconductor up to 1100-1200 "C, heat being conducted by phonon-phonon interactions. At high temperatures the electronic contribution becomes more important.The conductivity of U02 is low, typically 0.03 W cm-1OC-l at 1000°C, and this produces large temperature gradients between the surface and centre of fuel pellets which have a dominant effect on fuel performance. The conductivity varies with temperature (Fig. 25), and it is affected by irradiation below 500 "C. It has become an accepted practice, therefore, to relate fuel performance to the integrated thermal conductivity, rather than to attempt to calculate a temperature profile. The value of the integral is equal to the linear heat output of a fuel pin in W em-l divided by 4~ and its value where TI = 500°C and T2 = 2800°C (the melting point) is about 70 W cm-1 (see Fig. 26).30 The thermal conductivity of UC is between five and 10 times higher than that of U02 since it is an electronic conductor.31 The conductivity varies much less with temperature (Fig.17) and the Jk. d0 concept is therefore less import- ant. For similar heat outputs UC fuel rods experience much lower centre temperatures and thermal gradients and hence have different gas release and swelling characteristics. The effects of fabrication variables on the conductivity of UOz have been studied. The effect of the fractional porosityp is to decrease the value of k to k p according to the relation: Fig. 15. Thermal conductivity of irradiated stoicheiometric polycrystalline UOa. 0.09 Curve: 0.08 \ I. Saclay 2. Chalk river 3. AERE 4. AERE unirradiated 5.GE San JosC (based on f i t toORNL curve for unirradiate UOz at lower temperatures) 0.03 0.02 1600 1400 1200 1800 600 800 200 400 1000 Temperature ("C) R.I.C. Reviews I 80 70 60 I- 50 3 40 $ 5; In 30 20 10 0 500 I000 Fig. 16. Comparison-of integrated thermal conductivity out-of-reactor and in-reactor. where n has values between 1.5 and 4.0 depending on the shape and distribu- tion of the porosity. Oxygen in excess of an oxygen to metal ratio of two reduces the conductivity since the excess oxygen atoms act as impurity scattering centres. Single crystals are more conducting than polycrystalline material but they are usually metal rich and the effect of free uranium as a grain boundary phase is difficult to assess.The effect of irradiation on thermal conductivity has been studied by direct in-pile measurements in which thermocouples were placed in the fuel centre and along radii,32 by the comparison of pre- and post-irradiation measure- ments, and by the use of ‘markers’ in the fuel microstructure. Markers are grain growth and melting which occur at well-defined temperatures. Thus, from the irradiation of a pin with a known heat output in a known tempera- k, = k(l - np) Frost I I I T / Revised / ~ A E C L / / AProbzble range of ou t-of- dI I I I I :ztor E- San josC 3000 2500 2000 I500 Temperature (“C) 183 Sin tered (4.85% C 91% T.D.) 3 Cast 4.76% C (irradiated) 50.4/, 0.5-2 wt%O a) Arc cast 50.8% C (d) Oxycarbides b) Arc cast 51.7% C @) Arc cast 52% C @ A.C.cast 52% C @Cast UC (a) 52.5%C (bJ @ Cast 52.5% C + Absolute values 100 200 300 400 500 600 703 800 900 1000 1100 1200 1300 Temperature ("C) Fig, 17. Thermal conductivity of UC (close to stoicheiometric composition). ture environment (e.g. in a pressurized water loop at 300°C) the markers can be used to derive a thermal conductivity value. Below 500 "C the irradiation-induced decrease in conductivity is dependent on the temperature and burn-up; the damage becomes more stable as the burn-up increases. A major uncertainty in determining fuel temperatures is the thermal gradient through the filling gas in the gap between the fuel and the cladding. Often the gas is pure helium which becomes contaminated with fission gases during the irradiation, decreasing in conductivity and causing the fuel tem- perature to rise.Later (p. 198), it is seen that fission gases are retained within the fuel to a large extent at temperatures up to 1500"C, and form bubbles which cause the fuel to swell. In these lower temperature regions the fuel is fairly strong. Thus, although porosity may be present, the fuel will not deform by plastic flow into the porosity but will swell outwards and strain the cladding. The desire to remedy this situation has led to an intensive study of the mechanical properties of fuels out-of-pile and in-pile, with particular emphasis on creep processes.33 Both hypo- and hyperstoicheiometric U02 flow more readily than U02.000 and this has led to attempts to influence creep strength by the addition of 3- or 5-valent oxides to UO2-without marked success to date.The basis for this is the concept that vacancies play an important role in diffusion-type creep processes. This suggests that irradiation of fuel, which produces many vacancies, will produce an enhancement of the creep rate in those temperature regions where thermal effects are small, i.e. below 1000-1200 "C. In uranium carbide the variation of creep rate with composition under fixed load and temperature conditions is rather different.34 An increase in carbon R.I. C. Reviews 184 content leads to an increase in strength; since UC has a very narrow range of existence an increase in carbon content leads to the formation of a second phase of u2c3 or UC2 the particles of which may impede dislocation move- ment or lock grain boundaries.At low carbon concentrations uranium precipitates in the grain boundaries, improving grain boundary sliding, and within the grains. Quite small departures from stoicheiometry lead to changes in creep rate of one or two orders of magnitude. Chemical properties Lack of space prevents a full discussion of chemical aspects of fuel behaviour but some important aspects of the interaction of a fuel with its environment will be discussed. This is important first of all in fabrication processes. The flow and sintering characteristics of powders are affected by the nature of their surfaces and hence by the adsorption of water vapour and other gases; for example, oxygen is chemisorbed onto UOz at temperatures above -195°C with a maximum enthalpy of adsorption of 230 kJ These atoms are mobile at room temperature and above, and surface layers of higher oxides form readily. Uranium carbide oxidizes even more readily and, while it is possible to handle U02 powders at room temperature in fairly oxidizing atmospheres, UC must be handled in glove boxes containing an inert gas with a low water-vapour level.Under normal conditions the fuel inside a fuel pin is separated from coolants by the cladding. Commonly, this is zirconium in water-cooled reactors and stainless steel in COZ- or sodium-cooled reactors. Only when the cladding fails is the fuel exposed to the coolant; it is important that this does not then result in large quantities of fuel becoming entrained in the coolant.Under these circumstances we can rule out UC for reactors with water or C02 as coolant since clad failure will result in rapid oxidation of the fuel to u308 with a large volume increase and progressive splitting of the can until the us08 powder falls from the can. UOZ, on the other hand oxidizes more slowly. In fact, in water reactors the reaction rate at -350OC is so low that reactor operation can often continue after a fuel element failure has been detected, although this is highly dependent on the oxygen level in the water.5 In sodium, U02 and U O Z - ~ are stable. With U02+$ the excess oxygen atoms dissolve in the sodium; this can often lead to the disintegration of UOefz pellets in a sodium environment.Similarly UC and UC1-% are stable in sodium with a low oxygen content (typically 10 ppm in fast reactor circuits) but in UCl+$ the UC2 needles are decomposed by sodium to give UC and carbon. If a carbon ‘sink’ is available (e.g. stainless steel or zirconium) the carbon is transferred via the sodium to that sink (Fig. 18).36 The compatibility of fuel with its cladding is, of course, a question of thermo- dynamics and kinetics. Thus UOZ is thermodynamically less stable than zirconium but this combination is regularly used in water reactors, including the UKAEA’S Steam Generating Heavy Water Reactor at Winfrith, because the kinetics at normal operating temperatures are very slow.Experience shows that it is difficult to design fuel-cladding combinations from first principles and it is usually necessary to study reaction rates with a number of fuel-metal Frost 185 I300 h U 0, 1200 2 c U a I- 6 I loo I000 Fig. 18. Carbon contents of Type 304 and Type 410 stainless steels in equilibrium with carbon- saturated sodium. combinations. Generally a fuel specimen is sandwiched under pressure between two pieces of cladding material, held at a constant temperature and examined by sectioning followed by optical microscopy and by an electron microprobe, to determine reaction rates and the nature of the reaction products. In this field and in the development of improved fuels phase diagram studies have some importance.Since we are concerned with ceramic fuels with two species of atom, the ternary diagram is usually the simplest one studied-four- and five-component systems often require investigation. The melting points of U02 and UC are so high that the preparation of homogeneous alloys and the determination of liquidus and solidus temperatures demands the use of special high temperature techniques including refractory metal resistance furnaces, arc-image and electron-beam heating devices, tungsten/tungsten-rhenium thermocouples and optical pyrometry. X-ray diffraction studies, both of lattice parameters to determine phase boundaries and of crystal structures, play a vital role in this work. Finally, we are interested in the stability of the situation within an operating fuel; will the large thermal gradient cause changes in the distribution of the components? Markin and Rand37 have recently examined this situation for U02 and (U,Pu)Oz where a thermal gradient of 103°C cm-1 is not unusual.They showed that the presence of small quantities of carbon within the fuel (as little as 10 ppm, which is below the normal level) leads to the formation of a CO2/CO gas which is then able to transport material along the thermal gradient. Thus it is assumed that a constant CO&O ratio exists across the 186 Carbon content of steel (yo) R.I.C. Reviews 0 200 z h 3 4 0 0 IU- P 600 800 I .90 To vacuum system Constriction sealed before equil i brat ton I .95 Furnace Reference metal metal oxide Cooling water Heat shield Uranium oxide in crucible R.F.coil Drain Optical flat 117L-__ Pyrometer “Black body” hole 0:U ratio Fig. 19. Variation of ACoa with 0 : U ratio at 2200 K. Fig. 20. (a) Gas equilibration apparatus. (b) Oxygen distribution across a U0.85Puo.1502fy fuel pin for various COz : CO ratios. 2.05 O 2.00 I .95 I .92 Centre 2.00 1930 1765 1575 2.10 Frost I I ? .- U 0 e 1 I 1 I 2.05 Temperature (K) 1345 I I r2x 1,02(crn)2 cpo2 - I ratio I .982 I 1 2 3 4 5 6 Sukace 187 fuel. The oxide composition in equilibrium with a given CO2/CO ratio at any given temperature can be obtained from an Ellingham diagram; for a particular temperature distribution, the oxygen to metal (0 : M) ratio throughout the fuel can then be plotted.Results for a ‘mixed’ oxide fuel containing 15 per cent Pu and 85 per cent U are plotted in Fig. 19. An oxide with O:M fi 2.00 shows a negligible oxygen gradient, whereas deviations from stoicheiometry produce large oxygen gradients. In turn this alters the U:Pu ratio across the fuel; material with a high initial 0: M ratio will build up a high oxygen level towards its centre, vapour of composition uo3 will transfer uranium down the gradient leaving the plutonium enriched at the centre, possibly resulting in still higher temperatures. Autoradiography of irradiated fuel cross sections qualitatively confirms these predictions. To carry out this type of calculation it is necessary to know the value of the oxygen potential ACh2 and its tempera- ture coefficient.These have been measured by a galvanic cell technique using a solid electrolyte, reversible to oxygen ions only, e.g. Tho2 + Y203, or by a gas equilibration technique (H20/H 2 or C02/CO) using a reference metal/ metal oxide mixture to fix the oxygen potential (Fig. 188 2 Secondary knock-on path 3 Tertiary knock-on path H)I Intense ionization X Interstitial 0 Thermal or displacement Fig. 22. Five mechanisms of radiation effects. Represented are intense ionization vacancies, interstitials, impurity atoms and thermal or displacement spikes. Grid-line intersections are equilibrium positions for atoms. Fig. 23. Fission fragment tracks in vacuum-deposited UOz film 23 nm thick irradiated with 5x 1015 slow neutrons cm-2. The mean grain size is about 3 nm diameter.Frost 189 I o3 I o2 IRRADIATION BEHAVIOUR The irradiation behaviour of a fuel determines its performance in the reactor and is therefore of paramount importance. When a uranium or plutonium atom undergoes fission it splits into two fission fragments having atomic weights between 72 and 161, each atom having an initial energy of nearly 100 MeV. In addition, between two and three neutrons are produced, each with an initial energy of several MeV (Fig. 21). Both fission fragments and neutrons move rapidly through the lattice, exchanging their energy with the lattice atoms until they come to rest (Fig. 22). The energy carried by the fission fragments predominates and within the fuel we can probably neglect the neutrons.Each fission fragment is highly ionized (about 20+) and has a range of about 7-10 pm in fuel materials (Fig. 23). Over this range it excites atoms up to 100 A radius from the track centre, equivalent to raising their temperature to thousands of degrees locally; this is known as a 'fission spike'. Within each spike a number of atoms are displaced from their lattice positions to produce single vacancies and interstitials. If the temperature is high these recombine. If it is low they remain as single defects and have a large effect on transport processes such as thermal and electrical conductivities. At intermediate temperatures they may cluster and collapse into loops.Such effects are ob- servable in a number of ways, e.g. by x-ray lattice parameter measurements, by transmission electron microscopy and by resistivity measurements. However, they only have intrinsic practical importance in relation to their effect on thermal conductivity. They are also significant in relation to gaseous fission M h . 10' . E" v I .; a c L 0' n 2 to-' lo-; to-' 190 Zr Fig. 24. Mass distribution of fission products from thermal fission of 235U. R. I. C. Re views I00 I 80 60 40 30 .- : z 20 g - aJ x c, : 10 ? 3 m al .? 6 - !! .- aJ x $ 4 .- C 3 U L (d w .- E TI $ 2 U In I 0.8 0.6 0.4 1 I I 80 90 Fig. 25. Relative fission yields for 2351) and z39Pu.0.3 0.2 70 product behaviour and to diffusion processes. In the latter context, diffusion is controlled by vacancies and fission produces an abundant supply of these. It is possible therefore that such processes as creep and sintering may be enhanced while fission is proceeding. Once the fission fragments have come to rest. thev are more usuallv described in Fig. 24. Furthermore, they are radioactive and undergo decay processes, and Frost I I I I I _----- 1 Pu-239 thermal neutron' spectrum U-235 thermal neutron spectrum 150 --- I 100 130 160 110 140 5 I20 Mass number 191 they may capture neutrons so that the prediction of their concentration at some arbitrary point during irradiation is difficult.There are some significant differences between the yields from uranium and plutonium fission (Fig. 25). Studies of fission products have tended to concentrate on the stable species, that is those present in fuels some weeks after fission has ceased, since it is generally inconvenient to study fuel while it is in a reactor. An exception is the ‘swept capsule’ type of experiment in which an inert gas is used to entrain the volatile fission products from a fuel sample in pile to suitable detection equipment. The abundant fission products may be grouped conveniently according to their physical and chemical characteristics. Noble metals Mo, Tc, Rh, Ru, Pd. At high oxygen potentials Mo may exist as Moo2 or Moo3 Cs, I, Br, Te. Some of these may be Readily oxidized but insoluble in U02 BaO, SrO Readily oxidized but soluble in UOZ Zr, Ce, Nd Volatiles present as caesium halides Xe, Kr Stable gases The detailed study of fission products is important for several reasons: (a) they must be removed and stored or else used as tracers or heat sources, (b) they cause the fuel to swell, (c) the gases if released create a pressure, and ( d ) some may react with cladding materials.In terms of fuel element behaviour, swelling and gas release are the most important effects. The ‘basic’ swelling of a fuel is the nett volume increase incurred when all the fission products are atomically dispersed but are able to attain their equili- brium oxidation states. In an oxide fuel, solution of zirconium and the rare earths in the lattice causes a volume decrease while major increases arise from caesium, molybdenum, the noble metals, barium and strontium.Davies39 and Wait40 have derived a basic swelling rate of 0.5 per cent volume increase per 1 per cent burn-up while Anselin and Baily4l using different assumptions derive a value of 0.30 per cent AV/V. Davies’ corresponding figure for carbide is 0.9 per cent due to the large contributions from carbide-forming Mo and Ru and from Cs. To derive these figures it was obviously necessary to determine the oxidation states of the fission products and this was done by referring to the published thermodynamic data. The results are sensitive to the oxygen potential of the fuel, i.e. they will differ as the O:M ratio varies.The general conclusion is that the O:M of the fuel will rise as burn-up proceeds. In irradiated fuel elements the distribution of fission products is not random; the volatile fp’s (particularly Cs) migrate down the temperature gradient, the noble fp’s agglomerate to form a precipitate of second phase visible in the optical microscope (Fig. 26) and the gases, being insoluble and fairly mobile, agglomerate to form bubbles. A great deal of work has been done and continues to be done to study the distribution of fp’s in irradiated fuel elements and a number of techniques is available : 1. Autoradiographs of fuel sections. 2. Gamma spectrometry; the fuel is scanned by a counter enclosed in a R. I . C. Reviews 192 Fig.26. Agglomeration of noble fission products as seen through a microscope. Fig. 27. Caesium movement in (U, Pu)Oz. Frost 193 heavy-metal collimator and connected to a multi-channel analyser. It is possible to scan the cross-section of fuel pin in -1.0 mm steps for various isotopes, the results being printed out as a contour map or a coded colour pattern (Fig. 27). 3. Small diameter corings may be drilled from across the radius of a fuel pellet and subjected to radiochemical analysis. 4. The electron microprobe analyser has been used to an increasing extent to study segregated fission products. Either a shielded instrument is used or, as in the pioneering work of Bradbury et ~ 1 . ~ 4 2 the specimen activity is minimized by using a thin polished section.The majority of the inclusions in oxide fuel contain the ‘noble’ metals Mo, Ru, Rh, Tc and Pd (Fig. 28). A minority contain Ba, Zr, Sr and Ce; the presence of zirconium is unexpected since it dissolves in the oxide lattice. It is possible that it is present as barium zirconate. High burn-up oxide fuels in which the centre has been molten during irradiation usually contain ingots of the noble fission products at the bottom 250 I x .- n - v) v v) k‘ c U - C aJ 0 16’02’ Fig. 28. Electron probe microanalysis shows the inclusion of ‘noble’ metals in oxide fuel elements. 194 Ru I2O50’ Bragg angle R.I. C. Reviews of the shrinkage void which forms on freezing. The melting point of these ingots has been found to be about 1850°C.However, it is the gaseous fission products which dominate fuel element per- formance. Ten per cent by weight of the stable fission products are accounted for by xenon and krypton. Stated in another form, 24 cm3 of gas at NTP are formed in each gram of UOZ after 1 per cent burn-up. It can easily be seen that if all of this is released into the fuel element can a high pressure will build up unless precautions, such as adding a large plenum to the end of the element, are taken. The importance of predicting fission gas release has led to an intensive study of the kinetics and mechanisms of gas release as a function of temperature, composition and burn-up. In rather oversimplified terms, the presently understood position is that at temperatures below 1000 "C gas atom migration rates are very slow and any gas release occurs by 'recoil' (a fission event near to the surface produces a gas atom or its daughter and this escapes from the surface) or by 'knock-out' when a fission product collides with a static gas atom and knocks it out of the fuel.Between 1000 "C and 1600 "C gas atom mobility increases to significant values and some gas is able to diffuse to surfaces during the lifetime of the fuel element. Gas diffusion is controlled by the usual diffusion equation: D = DO exp (- Q/kT) During irradiation the fractional release of gas, F, is given by9 where t is the irradiation time and S/V the surface area:volume ratio of a particle of radius a, known as the equivalent sphere, and usually found by BET surface area measurement.A common method of studying gas release has been to measure the gas evolved from an irradiated fuel sample when heated in the laboratory. Under these conditions44 F = 6 d $ - ; ; i 3 Dt - 3D't where D' = - D a2 D' is the apparent diffusion coefficient. For U02 of 95 per cent theoretical density D has a value at 1400°C of -5 x lO-I5 cm2 s-1 and the activation energy Q, derived from measurements over a range of temperature, has a generally accepted value of 293 21 kJ mol-l. For UC D is about an order of magnitude smaller (-5 x 10-16 cmz s-l) and the activation energy is about the same (Fig. 29). This type of analysis has had some success but anomalies have appeared which, in general, indicate some hindrance to diffusion.Often called 'trap- ping',46 this is probably due to two processes acting in addition to atmoic diffusion. Firstly, atoms will agglomerate to form stable bubbles (Fig. 30), the motion of which will differ from that of atoms and, secondly, gases become trapped at grain boundaries and when cracking occurs-as at a reactor shut- 195 Frost - \ 3 ( - 14 - 15 - 1 6 - N n E - ? 0 - 17 - t8 - 19 - 20 9 8 5 6 7 Temperature I / T ( K x lo4) Fig. 29. Diffusion of lssXe from uranium carbide. Fig. 30. Aggtomeration of atoms to form stable bubbles. R.I.C. Reviews 196 10 I00 h v - G 1.0 0. I c s 2 E vl W 0.0 I 900""l'OOO down or start-up-these are suddenly released.This latter phenomenon has been called the 'burst' effect in some experiments. Recent work by Whapham47 and others has shown that even this picture is too simple since the fission process can cause bubbles to break up and the gas disperse into the lattice. Under steady powder conditions a large release of grain boundary gas is delayed until the bubbles touch one another. Above 1600°C grain growth begins to be significant in U02; up to 1800- 2000°C grain boundary sweeping will enhance the rate of gas release. Above 2000°C the bubbles will grow rapidly and move up the temperature gradient by evaporation from the hot face and condensation on the cold. We can simplify prediction by assuming that all the gas is released above 1600°C (Fig. 31).Analogous information on carbide is lacking at present. The fact that the fission gases form bubbles is significant in relation to fuel swelling. Below 1000°C the contribution of bubbles to swelling is small and above 1600 "C most of the gas has been released. In the intermediate region bubbles become important. The swelling of a fuel sample after burn-up b is: Fig. 3 I . Gas release from U02 as a function of fuel-cell temperature. Frost 14 I 100 I200 I300 1400 1500 1600 1700 1800 1900 2000 2100 2200 Fuel centre temperature ("C) -- *'- Sb + 4 p r - r3 3 V 197 where S is the solid fp swelling and p the number of bubbles of radius r. We need to evaluate p and r. Greenwood et aL48 derived an expression for = ( 8n'u2r~Dg) 3p l I 2 where /3 is the gas generation rate, Dg the gas diffusion coefficient, ro the radius of bubble nucleus (a few lattice spacings) and a the lattice parameter.When a gas atom comes to rest in the fuel lattice, it strains the lattice. This strain can be relieved if vacancies flow to the gas atom. During irradiation vacancies are plentiful and the internal gas pressure p is balanced by the surface tension force y in a bubble of radius r such that 2Y P", p must also satisfy the modified gas law: p v = nc.- kT N where n is the number of gas atoms per bubble and N is Avogadro's number. Thus we should be able, for any given temperature and irradiation condition, to calculate p and r.49 In practice the temperature varies across the fuel section and this has an effect on both p and r ; more bubbles are nucleated near the centre and their growth rate there is greater.Bubbles can migrate by a variety of mechanisms depending on their size- small ones may exhibit a Brownian motion or they may move by surface or volume diffusion until trapped by a dislocation or a grain boundary. When the bubbles grow larger (>65 nm radius) they can break away from dislocations and move up the temperature gradient until they meet a grain boundary. Above 450 nm radius the temperature gradient driving force exceeds the grain boundary force and either the bubble escapes or it will drag the boundary with it, giving rise to long columnar grains. The situation in an operating fuel pin of UOz is illustrated in Figs 32a & b which show schematically a typical microstructure together with a plot of bubble velocity versus radius. Up to -1500°C the original grain structure remains.At the cooler rim of this region the gas bubbles are small and at the inner edge they are larger and more numerous but probably less than 65 nm diameter. Above -1500°C grain growth begins to be measurable and gas bubbles reach grain boundaries fairly rapidly, where they link up and form a path for rapid release. Above about 1750 "C the bubbles have exceeded 450 nm radius and they move up the gradient forming columnar grains and removing all volatile fission products rapidly, together with the initial sinter pores which form a cavity at the centre of the fuel. Below 1500°C the bubbles persist and contribute to the fuel swelling.The larger bubbles at the centre contribute more than those at the edge and a stress gradient is set up-compressive at the centre and tensile at the edge. The net swelling is determined by the bubbles at the region of zero stress which is about two-thirds of the distance from the centre of the region (Fig. 33).50 198 R.I.C. Reviews Expressed more quantitatively, the contribution to swelling in ceramic fuels of the fission gases is of the order 0.5-1.0 per cent AV/Vper 1 per cent burn-up (Fig. 34). This must be added to the solid fission product value of between 0.25 and 0.5 per cent A V/V to give a maximum swelling of 1.5 per cent A V/ Y per 1 per cent burn-up for oxide and about 2.0 per cent AV/V for carbide.Experimental values derived from diameter changes in fuel pins are generally Centre temperature Onset of columnar grain growth Onset of equiaxed grain growth Original sintered grain structure persists Gas bubble behaviour in pellet Schematic view of highly rated UO, pellet (b) Fig. 32. (a) Cross-section of irradiated UOz pellet (courtesy 1. nucl. Mater.30); (b) schematic view, showing gas bubble behaviour. Frost 199 14§ 0.0 I 0 - 0.01 - 0.02 - 0.03 a h - a I m - 0.04 t -0.05 -0.06 v v -0.07 - 0.08 I - 0.09 -0.1 -0.1 I t -0.12 Fig. 33. Non-uniform swelling stress in a fuel pellet. *: 6 1 . 1 0 1 . 1 2. I 3.2 r4 Com press ion 4.3 6 Z 5.3 .a 0, 6.4 x ul 7.5 ; 8.5 9.6 10.6 11.7 12.7 10 9 I 7 ! 8 1 lower than this, which is not surprising as the claddingand the external coolant pressure tend to exert a restraining influence on bubble growth.Fast reactor fuels should be capable of attaining a burn-up in excess of 5 per cent heavy atoms and preferably 10 per cent to give an acceptable fuel cycle cost. To achieve this it must be possible to contain fuel swelling of the order of 10 per cent or more within the cladding without failure. If voidage is provided within the fuel, whatever the form (central hole, sinter pores, cavities in packed powder) it is unable to accommodate the swelling efficiently because the fuel cannot deform sufficiently in the cold regions below 1000-1200 "C.Thecladding, which is generally stainless steel, is itself impaired by the fast neutron irradia- tion being unable to sustain more than about 1 per cent creep strain.51 There is, therefore, a considerable effort being applied to the problem of altering the fuel properties or its geometry relative to the can, in order to accommodate swelling more effectively. As an example, one approach is to leave a generous gap between the fuel and its can, and to fill the gap with sodium. This acts as a good conductor of heat, keeping the fuel cool, and may be displaced to a plenum or reservoir as the fuel swells. Since the fuel cycle cost is a major component of the power cost in fast reactors the potential cost benefit of such developments is very considerable.However, it should be recognized that fuel and fuel element irradiation studies are very expensive. There are many steps in the development of a fuel element to the stage where it is acceptable to reactor constructors and their customers. The first steps are Fig. 35. Very high burn-up fuel particle testing rig. Frost 20 1 concerned with evolving a fabrication method and measuring important properties out of pile, as discussed earlier. Then two types of in-pile test may be used; first, short lengths of fuel element, typically 5-15 cm long, are made with realistic radial dimensions to establish the correct temperature profiles and are irradiated in a materials testing reactor (MTR), such as DIDO and Fig. 36. Detailed examination of irradiated pins.R . I.C. Reviews 202 PLUTO at Harwell, under carefully controlled and monitored conditions (Fig. 35).52953 This establishes whether there are any important problem areas and gives preliminary data on swelling and gas release. Detailed examination of the irradiated pins is performed in shielded cells, usually with a cell devoted to each type of measurement (mensuration, fission gas release, optical and electron microscopy, fission product distribution etc.) (Fig. 36). Secondly, the kinetics of fission product release are measured in a swept capsule rig in an MTR.54 A small sample of fuel is held inside an electric heater at a fixed temperature while helium flows over it; the gases are entrained to a Fig. 37. In-pile assembly. Frost 203 counter and the volatiles to a metal surface which is subsequently removed for analysis (Fig.37). Having established the fuel performance on a small scale, full-size prototype fuel elements are made into clusters and, where possible, tested in prototype power reactors either as part of the main ‘driver’ charge or in special loops to test failure characteristics. The essential point is that statistical information must be amassed because in the first place there must always be some latitude in manufacturing specifications and in the second place conditions vary considerably from point to point in a reactor core in terms of temperature and neutron flux and spectrum. Consequently, small scale testing may give misleading information.The examination of irradiated fuel pins on a statistical scale plus the careful analysis of the results is a demanding and costly operation but nevertheless it is vital to the successful exploitation of any reactor type. CONCLUSION Much of the engineering and plant aspects of nuclear reactors is based on conventional engineering and chemical engineering practice. The fuel cycle is a unique but vital part of the system demanding intensive research and develop- ment at every stage from the ore to reprocessing of the irradiated fuel. The nature of the work makes it expensive, particularly in?erms of capital facilities such as test reactors, hot cells, glove boxes and remote operations. However, the rewards are potentially high and may be measured in terms of savings of hundreds of millions of pounds in power programmes.The approach in the UKAEA to these problems is broadly based with the preliminary exploratory work and the basic research on mechanisms being centred on the Research Group at Harwell. The development of fuel element designs and statistical testing is the responsibility of the Reactor Group in a number of laboratories, and the production of fuel elements and their re- processing is the responsibility of the Production Group. This review has been concerned primarily with expounding the general principles behind this work rather than with descriptions of detail. The interested reader is referred to other works for this detail, and particularly to the annual reports of the UKAEA.REFERENCES 1 Report of the Select Committee on Science and Technology, The UK nuclear reactor programme, xxxvii, London: HMSO, 1967. 2 G. F. Tape, 3rd International Conference on the Peaceful Uses of Atomic Energy (ICPUAE-3), Geneva 1964, 1, paper P192, 69. 3 P. D. Dunn et al., Nature, Lond., 1962, 195, 65. 4 J. T. Ramey et al. ICPUAE-3, Geneva, 6,428. 5. J. Belle (ed.) Uranium dioxide USAEC, 1961. 6 B. R. T. Frost et al. ICPUAE-3, Geneva 1964. Paper P153. 7 R. A. U. Huddle and L. R. Shepherd, IAEA Conference on New Nuclear Fuels, Prague 1963, 2,467. 8 W. D. Wilkinson, Uranium metallurgy. New York: Interscience, 1962. 9 UKAEA Ann. Rept No. 11, para. 202; No. 13, para. 252; and No. 14, para. 223. 10 ICPUAE-1, Geneva 1955. Vol. 8. 11 S. A. Cottrell et al., ICPUAE-3, Geneva 1964. Paper P150. 12 S. Naymark and C . N. Spalaris, ICPUAE-3, Geneva 1964. Paper P233. 13 ICPUAE-2, Geneva 1958, 6, 569-629. 14 J. E. Ayer and F. E. Soppet, J. am. ceram. Soc., 1965, 48, 180; 49, 207. R.Z. C. Reviews 204 15 L. E. Russell and J. D. L. Harrison, ENEA Symposium on Reactor Materials. Stockholm, 1959. 16 F. Brown et al., Carbides in nuclear energy (ed. L. E. Russell), vol. 10, 692. London: Macmillan, 1963. 17 R. Ainsley et al., ibid., 540. 18 J. D. L. Harrison and J. W. Isaacs, ibid., 556. 19 J. D. L. Harrison et al., ibid., 629. 20 K. M. Taylor et al., ibid., 668. 21 H. Pearlman and R. F. Dickerson, ICPUAE-3, Geneva 1964. Paper P234. 22 J. P. McBride et al. 0 ~ ~ ~ - 3 8 7 4 , 1966. 23 M. Allbutt and R. M. Dell, J. nucl. Muter., 1967, 24, 1. 24 G. R. Howells et al., ICPUAE-2, Geneva 1958, 17, paper P307, 3. 25 ICPUAE-3, Geneva 1964,lO. 26 L. Burris et al., ICPUAE-2, Geneva 1958, 17, paper P538, 401. 27 S. Lawroski, Chem. Engng. Prog., 1955’51,461. 28 F. W. Yeats, Atom, Lond., December 1966, no. 122,282. 29 IAEA Technical Reports Series No. 59, Thermal conductivity of uranium dioxide, Vienna 1966. 30 J. A. L. Robertson, A. M. Ross, M. J. F. Notley and J. R. MacEwan, J. nucl. Muter., 1962, 7, 225. 31 J. A. Leary, R. L. Thomas, A. E. Ogard and G. C. Wonn, Carbides in nuclear energy (ed. L. E. Russell), vol. 1, 365. London: Macmillan, 1963. 32 D. J. Clough and J. B. Sayers. ~~1~3-R.4690, 1964. 33 W. M. Armstrong and W. R. Irvine, J. nucl. Muter., 1963, 9, 121. 34 J. J. Noreys, Carbides in nuclear energy (ed. L. E. Russell),vol. 1,435. London : Macmillan, 1963. 35 J. D. M. McConnell and L. E. J. Roberts in Chemisorption (ed. W. E. Garner), 218. London: Butterworths, 1957. 36 B. A. Webb, North American Aviation Report NAA-SR-6246, 1962. 37 M. H. Rand and T. L. Markin, ‘Some thermodynamic aspects of (U,Pu)Oz solid solu- tions and their use as nuclear fuels’, A E R E - R . ~ ~ ~ ~ , 1967. 38 T. L. Markin, ‘Thermodynamic data for U02 and (U,Pu)Oz applied to fuel preparation problems’, mrn-R.5538, 1967. 39 J. H. Davies, unpublished work. 40 E. Wait and B. R. T. Frost, IAEA Conference ‘Plutonium as a reactor fuel’, Brussels 1967, paper SM-88/25, Proceedings, 469. 41 F. Anselin and W. E. Baily, Trans. Am. nucl. Soc., 1967, 10, 103. 42 B. T. Bradbury, J. T. Demant, P. M. Martin and D. M. Poole, J. nucl. Muter., 1965,17, 227. 43 J. I. Bramman, R. N. Sharpe, D. Thorn and G. Yates, J. nucl. Muter., 1968, 25, 201. 44 A. H. Booth, Canadian report CRDC-721, 1957. 45 D. Davies and G. Long, AERE-R.4347, 1963. 46 J. R. MacEwan and W. H. Stevens, J. nucl. Muter., 1964, 11, 77. 47 A. D. Whapham and B. E. Sheldon, 6th International Conference on Electron Micro- scopy, Kyoto, Japan, 1966. Proceedings, 375. 48 G. W. Greenwood, A. J. E. Foreman and D. E. Rimmer, J. nucl. Muter., 1959, 1, 305. 49 R. S. Barnes and R. S. Nelson, AErn-R.4952, 1965. 50 B. L. Eyre and R. Bullough, J. nucl. Muter., 1968,26, 249. 51 P. T. Nettley et al., British Nuclear Energy Society Conference on Fast Reactors, London 1966. 52 0. S. Plail, NucE. Pwr, December 1960. 53 N. H. Hancock, ~~m-R.4156, 1966. 54 G. Jackson, D. Davies and P. Biddle, A E R E - R . ~ ~ ~ ~ , 1966. Frost 205