BEHAVIOUR AND CHEMICAL STATE OF IRRADIATED CERAMIC FUELS
INTERNATIONAL ATOMIC ENERGY AGENCY, VIENNA, 1974
BEHAVIOUR AND CHEMICAL STATE OF IRRADIATED CERAMIC FUELS The following States are Members of the International Atomic Energy Agency:
AFGHANISTAN HAITI PAKISTAN ALBANIA HOLY SEE PANAMA ALGERIA HUNGARY PARAGUAY ARGENTINA ICELAND PERU AUSTRALIA INDIA PHILIPPINES AUSTRIA INDONESIA POLAND BANGLADESH IRAN PORTUGAL -BELGIUM IRAQ ROMANIA BOLIVIA IRELAND SAUDI ARABIA BRAZIL ISRAEL SENEGAL BULGARIA ITALY SIERRA LEONE BURMA IVORY COAST SINGAPORE BYELORUSSIAN SOVIET JAMAICA SOUTH AFRICA SOCIALIST REPUBLIC JAPAN SPAIN CAMEROON JORDAN SRI LANKA CANADA KENYA SUDAN CHILE KHMER REPUBLIC . SWEDEN COLOMBIA KOREA, REPUBLIC OF SWITZERLAND COSTA RICA KUWAIT SYRIAN ARAB REPUBLIC CUBA LEBANON THAILAND CYPRUS LIBERIA TUNISIA CZECHOSLOVAK SOCIALIST LIBYAN ARAB REPUBLIC TURKEY REPUBLIC LIECHTENSTEIN UGANDA DENMARK LUXEMBOURG UKRAINIAN SOVIET SOCIALIST DOMINICAN REPUBLIC MADAGASCAR REPUBLIC ECUADOR MALAYSIA UNION OF SOVIET SOCIALIST EGYPT, ARAB REPUBLIC OF MALI REPUBLICS EL SALVADOR MEXICO UNITED KINGDOM OF GREAT ETHIOPIA MONACO BRITAIN AND NORTHERN FINLAND MONGOLIA IRELAND FRANCE MOROCCO UNITED STATES OF AMERICA GABON NETHERLANDS URUGUAY GERMAN DEMOCRATIC REPUBLIC NEW ZEALAND VENEZUELA GERMANY, FEDERAL REPUBLIC OF NIGER VIET-NAM GHANA NIGERIA YUGOSLAVIA GREECE NORWAY ZAIRE, REPUBLIC OF GUATEMALA ZAMBIA
The Agency's Statute was approved on 23 October 1956 by the Conference on die Statute of the IAEA held at United Nations Headquarters, Mew York; it entered into force on 29 Iuly 1957, The Headquarters of the Agency are situated in Vienna. Its principal objective is "to accelerate and enlarge the contribution of atom ic energy to peace, health and prosperity throughout the world".
© IAEA, 1974
Permission to reproduce or translate the information contained in this publication may be obtained by writing to the International Atomic Energy Agency, Kärntner Ring 11, P.O. Box 590, A-1011 Vienna, Austria.
Printed by the IAEA in Austria January 1974 PANEL PROCEEDINGS SERIES
BEHAVIOUR AND CHEMICAL STATE OF IRRADIATED CERAMIC FUELS
PROCEEDINGS OF A PANEL ON THE BEHAVIOUR AND CHEMICAL STATE OF IRRADIATED CERAMIC FUELS ORGANIZED BY THE INTERNATIONAL ATOMIC ENERGY AGENCY AND HELD IN VIENNA, 7-11.AUGUST 1972
INTERNATIONAL ATOMIC ENERGY AGENCY VIENNA, 1974 BEHAVIOUR AND CHEMICAL STATE OF IRRADIATED CERAMIC FUELS IAEA, VIENNA, 1974 STI/PUB/303 FOREWORD
The economic attractiveness of large nuclear power reactors is criti cally dependent on overall fuel performance and fuel life. The successful design of high performance, long-lived fuel elements is itself dependent on a thorough understanding of the special chemical and physical behaviour of irradiated fuel. The transformation of some of the original actinide elements into new species, the effects of fission products, the progressive alteration in the stoichiometry within the fuel, and the effects of oxygen on fuel and cladding are among the many important problems that must be in vestigated when developing a successful fuel. Extensive studies are under way on mixed urania-plutonia fuel systems, developed for fast breeder reactors, and on other advanced fuels. To review the progress and problems in this field, the International Atomic Energy Agency convened a panel of experts at its Headquarters in Vienna on 7 - 11 August 1972. The deliberations of the Panel, which was composed of 19 experts from seven countries and EURATOM, covered many aspects of transport properties, fuel/fission product-clad interaction, and the thermodynamics and phase equilibria of nuclear fuel m aterials. These Proceedings contain the texts of the 30 working papers con sidered by the Panel and the Panel's Summary and Recommendations.
CONTENTS
Stoichiometric effects in irradiated uranium-plutonium oxide fuel (IA E A -PL -463/1) ...... 1 C. E. Johnson, I. Johnson,. P. Blackburn, J .E . Battles, C .E . Crouthamel The composition and chemical state of irradiated oxide reactor fuel material (IAEA-PL-463/2) ...... 31 J. R. Findlay Redistribution of uranium and plutonium in mixed-oxide fuels during irradiation (IA E A -PL -463/3) ...... 41 D .R . O'Boyle, R .O . Meyer Determination du taux de combustion et de la composition en isotopes lourds et en produits de fission des combustibles des réacteurs thermiques et rapides (IA EA -PL-463/4a) ...... 67 R. Vidal, M. Robin, C. Devillers Désaccord entre les rendements de fission théorique et experimental des gaz nobles en neutrons rapides — Mise en évidence sur le combustible du réacteur RAPSODIE (IAEA- PL-463 / 4b) ...... V...... 77 B. K r y g e r Oxygen redistribution in LMFBR fuels (IA E A -P L -463/5) ...... 83 S. K. Evans, E .A . Aitken Modification du potentiel d'oxydation des oxydes mixtes (U, Pu)02, en fonction du taux de combustion (IA E A -PL-463/6a) ...... 87 F. Schmitz, J. Marticorena, G. Dean Evolution de l'activité du carbone en fonction du taux de combustion dans un combustible carbure (IAEA-PL-463/6b) ...... 99 Nicole L o r e n z e 11 i , J .P . Marcon Some phase equilibria and thermodynamic considerations for irradiated oxide nuclear fuels (IAEA-PL-463/7)...... 115 P. E. Potter Formation of phases and distribution of fission products in an oxide fuel (IA E A -P L -463/8) ...... 157 H. Kleykamp Microanalyse X d'oxyde (U, Pu)02+x simulant différents taux d'irradiation (IA E A -PL-463/9a) ...... 167 F. de Keroulas, D. Calais, F. Schmitz Migration des produits de fission radioactifs dans des combustibles (U, Pu)02 (IAEA-PL-463/9b) ...... 179 M. Mouchnino Distribution des produits de fission et localisation par spectrométrie gamma en cours d'irradiation et après irradiation (IAEA- PL-463 /9c) ...... 191 G. de Contenson, J. Monier, Nicole V i gn e s ou 11 Analyse à la microsonde de produits de fission métalliques dans les combustibles en géométrie eau ordinaire fonctionnant avec une zone fondue (IA E A -PL-463/9d) ...... 201 J. Bazin, M. Perrot, Nicole V i gn.e s о u lt The migration of fission products through reactor fuel materials (IA E A -PL-463/10) ...... 211 J .R . Findlay Contribution to the studies of the fission gases released from irradiated uranium oxide pellets (IAEA-PL-463/11 ) ...... 221 J. Klima, M. Podest, V. Vins Pu, U redistribution in (U, Pu)02 fuels by temperature gradients (IAEA-PL-463/12) ...... 225 M. G. Adamson, E. A. Ait ken Chemical interactions of fission products with stainless steel claddings (IAEA-PL-463/14) ...... 237 P. Hofmann, O. Gôtzmann Fuel/cladding compatibility of stainless steels with gas and sodium-bonded uranium plutonium carbide fuels (IAEA- PL -463/15) ...... 255 O. Götzmann, R. W. Ohse Out-of-pile investigations of fission product-cladding reactions in fast reactor fuel pins (IAEA-PL-463/16) ...... 269 E. A. Aitken, S. K. Evans, B. F. Rubin Réaction à l'interface gaine (acier inoxydable )/combustible (oxyde mixte (U,Pu)02±x) dans les éléments combustibles irradiés en neutrons rapides (1AEA-PL-463/17) ...... ' ...... 287 D. Calais, M. Conte, F. de Keroulas, R. Le Beuze The role of caesium in chemical interaction of austenitic stainless steels with uranium plutonium oxide fuels (IA E A -PL -463/18) ...... 299 R. W. Ohse, M. Schlechter Mise en évidence de la migration de l'oxygène sous irradiation: Irradiation L-l (IAEA-PL-463/l9a) ...... 315 M. Trotabas, F. de Keroulas, J. P. Gatesoupe Thermodiffusion et vaporisation dans les oxydes mixtes irradiés: Evolution du rapport Pu/U + Pu (IAEA-PL-463/19b) .... 325 M. Mouchnino Emission des gaz de fission par l'oxyde d'uranium dans les éléments combustibles (IAEA-PL-463/19c) ...... 337 P. Chènebault, R. Delmas Réactions entre le sodium et les oxydes mixtes (IAEA- P L -4 6 3 /l9 d )...... 349 M. Housseau, G. Dean, F. Perret Reaction behaviour of fission products in carbides (IA E A -PL-463/20) ...... 361 H. Holleck, E. Smailos Fission product distribution in fast reactor oxide fuels (IA EA -PL-463/21) ...... 379 H .J. Powell Reaction of sodium with uranium-plutonium oxide and uranium oxide fuels (IAEA-PL-463/22) ...... 393 P. E. Blackburn Thermodynamic studies of the sodium fuel reaction at General Electric (IAEA-PL-463/23) ...... 411 M. G. Adamson, E. A. Aitken, S. K. Evans Summary, conclusions and recommendations ...... 421 List of Participants’ ’ ...... 435 IAEA-PL-463/1 1
STOICHIOMETRIC EFFECTS IN IRRADIATED URANIUM- PLUTONIUM OXIDE FUEL
C .E . JOHNSON, I. JOHNSON, P. BLACKBURN, J.E. BATTLES, C.E. CROUTHAMEL Argonne National Laboratory, Argonne, 111., United States of America
Abstract
STOICHIOMETRIC EFFECTS IN IRRADIATED URANIUM-PLUTONIUM OXIDE FUEL. Stoichiometric effects in irradiated mixed urania-plutonia fuel are of major significance in determining overall fuel performance and fuel life. The paper discusses the results of extensive studies carried out to understand the various ways in which oxygen can interact with cladding, fuel and fission products under irradiation conditions.
I. INTRODUCTION
Most of the early work with oxide|fuels has been performed with enriched urania. А -major effort on the mixed urania-plutonia fuel began in about the middle 1960's. In comparison to urania the mixed urania-plutonia fuel is appreciably more oxidizing. The consequences and the importance of this fact are being fully recognized only very recently. Another relatively unique feature of the oxide fuel compared to other candidate fuels is its poor ther mal conductivity. This leads to unusually steep thermal gradients during fuel burnup and especially at the beginning of fuel life while fuel restructuring is taking place. The redistribution of oxygen during both the early restruc turing and over the long burnup period are believed to be significant in its effect on the overall fuel performance and fuel life. The fuel oxygen poten tials and temperatures which are set up in irradiated fuel are probably more strongly coupled with fuel performance and fuel life than any other parameters
Editor’s Note: The paper presented to the Panel for discussion is not presented in its entirety since almost all the material presented in it was published shortly after the panel was held (JOHNSON, C. E ., JOHNSON, J ., BLACKBURN, P. E ., CROUTHAMEL, C. E ., Effects of oxygen concentration on properties of fast reactor mixed-oxide fuel, Reactor Technol. 15^4 (1972)). The abstract of the paper is as follows: "This review assesses the current understanding of the effects of oxygen concentration on properties of fast reactor m ixed-oxide fuels. The differences in fission-product yields from eitherZ35U or 239Pu fission greatly affect the oxygen potential. For fast fission, 235U produces higher yields of oxide formers, whereas 239Pu produces higher yields of noble metals. The effects of oxygen and plutonium concentrations on thermal conductivity of ceramic fuels are examined. Increases in oxygen and plutonium concen trations appear to produce increases in the thermal conductivity. A mathematical model for predicting the oxygen potential is described and compared with mass-spectrometric measurements. Excellent comparisons are found for the urania and mixed urania-plutonia systems. The influence of oxygen potential on actinide, oxygen, and fission-product transport in irradiated fuels is discussed. A method utilizing the M o-M o02 indicator couple is given for evaluating the oxygen potential in the columnar and equiaxed regions of irradiated m ixed-oxide fuels. The contributions of oxygen and fission products, especially cesium, to cladding corrosion are evaluated. The cladding acts as an oxygen sink, thereby increasing corrosion during the irradiation of the fuel. The effect of oxygen potential on the sodium- fuel reaction, in the event of a cladding breach, is examined, "
1 2 JOHNSON et al.
The oxygen potential provides the chemical driving force for the attack of the stainless steel cladding, it controls the' chemical state of certain fission products, their interaction with the fuel, their contribution to fuel swelling, their volatility, and their redistribution. The oxygen potential controls the vaporization of the fuel components with the most prominent species being the uranium oxides in the mixed urania-plutonia fuels being irradiated (0/M 1.94-2.00) and thus to a major degree controls the radial redistribution of uranium in the fuel matrix. It also affects the potential for reaction of liquid sodium coolant with the fuel matrix to form low density sodium uranate in the event that a minor breach in the cladding occurs. The oxygen potential also affects the intrinsic thermal conductivity of the fuel, the rate and temperature for fuel restructuring, and the form of the radial oxygen potential gradient which is established in the operating fuel element. Thus, considerable emphasis is being placed on the determination and calcula tion of oxygen potentials in various regions of the fuel, expecially at the fuel-cladding interface. Also, attention has been given to the total oxygen available to the cladding, first, during the early restructuring period, and second, during the more prolonged fuel burnup period.
We also have tried to delineate the ways and the conditions under which oxygen can interact with the cladding, fuel, and fission products. Cladding oxidation can be promoted in various modes - internal oxidation, surface oxidation, and intergranular oxidation. There has been confusion and con siderable variation in the role assigned the cladding in controlling fuel stoichiometry during irradiation. Rand and MarkinJlJ■and Markin and Mclverl2] have determined that slightly hypostoichiometric mixed UO 2- 2O w/o Pu02 fuel has the potential for oxidizing stainless steel. Nevertheless, the visible oxide layers on the inner cladding surface in irradiated fuel did not seem to indicate any significant reaction. The conclusion drawnllH3] from these irradiated fuels has been that the clad and fuel were not in chemical equili brium and little control of the fuel stoichiometry was effected by the clad ding. This is at variance with our observations of prototypical fast test reactor fuel. Prototypic is defined approximately as follows: 316 stainless steel clad, UO 2- 2O w/o PUO2 fuel initially at 90 percent theoretical density operated at 400-500 w/cm, inner clad wall temperatures 500°C-700°C, and ini tial 0/M = 1.94-1.98.
We observe the cladding functioning as an oxygen sink by surface oxida tion, internal oxidation, and intergranular attack. The fuel cladding even in the plenum region is observed to function to some extent as an oxygen sink.
II. FISSION PRODUCT YIELDS AMD REACTOR NEUTRON SPECTRAL ! EFFECTS ON FUEL STOICHIOMETRY
The fission product yield distribution is important insofar as it varies the relative amounts of oxide forming fission products to effect changes in available oxygen as a function of the fuel burnup. Within the range of neutron energy spectra from thermal to the hardest fast reactor spectra, the fission product yields do not change appreciably, especially in the two peak yield regions, and no significant change in available oxygen can be attrib uted to neutron energy spectral shifts affecting fission product distribu tions.
However, a major shift in the fission yield distribution is evident with a change in fissile isotope from 23-3u or 235u to 239pu fission. Furthermore, because the fission yield distributions of all the1 fissile isotopes are rela tively constant in thermal to fast spectra, the isotope fission yield shift is similar in both thermal and fast spectra. Unfortunately, the available IAEA-PL-463/1 3
FIG. 1. Production of excess oxygen by fission, calculated for constant oxygen potential profile (14 kW/ft, 90% T . D. ).
oxygen increases in shifting from 233jj or 235jj to 239pu fission. Because of the difficulty in achieving prototypical power density in EBR-II irradiations, virtually all test fuel has been run with fully enriched urania-(20-30 w/o plutonia mixtures. However, future fuel in power production with large LMFBR's will be composed of natural or depleted urania-plutonia mixtures. Figure 1 shows the calculated production of excess oxygen in moles per centi meter of fuel as a function of burnup for fully enriched and natural urania fuels. This was calculated for a constant oxygen potential profile which was assumed to be set by the stainless steel cladding. It is evident that a sig nificant increase in available oxygen will be produced in future fuels due to the relative increase in plutonium fission.
A major concern is the ultimate fate of this excess oxygen in the reactor fuel pins. Our current assessment is that the stainless steel cladding will most likely react with the major part of it, and the fission product molyb denum in the metallic inclusions will react with a smaller part of it.
Reactor neutron spectrum effects will also affect fuel composition by changing the fuel Pu/U ratio. In the more moderated neutron spectra of large fast reactors, the competition between 238ц neutron capture to produce 239pu and. the depletion of 239pu by fission will change in different regions of the reactor.
A number of experimental reaction rates (оф) for selected fast reactor spectra have been measured directly by Dudeyl^] in EBR-II and in ZPR critical assemblies. Using these experimentally determined reaction rates, he has calculated 239ри/238ц rat±os as a function of the atom percent burnup. The data in Table I indicate that there are neutron spectral regions where the 239pU(/238u ratio in irradiated fuel can move in either direction. The more 4 JOHNSON et al.
T A B L E I
2 39 23 8 Calculated ■ Pu/ U vs. Burnup For Various LMFBR Conditions Based on Measured Reaction Rates In itia l Conditions* ^®U02“25 a/о ^^®Pu02
239 2 3 8 u R e a c t o r Pu C o n d i t i o n s L o c a t i o n 7. B .U . 239P u / 2 3 8 U (a to m s) (a to m s)
2 . 4 0 . 2 3 7 3 0 .9 3 6 2 3 .9 4 5 1 . ZPR-3 Assembly Core (Row 2) 5 . 7 0.2202 0 .8 5 0 9 3 . 8 6 4 60-Simulated 1 0 . 7 0 . 1 9 6 2 0 .7 3 2 4 3 . 7 3 3 Homogeneous 2 . 9 0 . 2 4 0 8 0 .9 5 0 2 3 . 9 4 6 EBR-II Core 4 . 1 0 . 2 3 2 4 0 .9 0 4 5 3 . 8 9 3 2. Blanket (Row 9) w i t h 2 3 8 U 7 . 8 0 .2 1 7 5 0 .8 2 4 1 3 . 7 8 8 9 . 9 5 0.2110 0 .7 8 8 6 3 . 7 3 7 3 . 5 0 . 2 3 1 3 0 .9 0 6 6 3 . 9 2 0 3. Core (Row 2) 6 . 7 0 .2 1 4 8 0 . 8 2 5 3 3 . 8 4 1 10.6 0 . 1 9 6 0 0 .7 3 2 8 3 . 7 3 9 ZPR-3 Assembly Core-Reflector 3 . 5 0 . 2 3 5 2 0 .9 1 9 1 3 .9 0 7 4. 6l-Sim ulated I n t e r f a c e 6 . 7 0.2222 0 .8 4 8 1 3 . 8 1 7 Homogeneous (Row 7) 1 0 . 7 0 .2 0 7 2 0 .7 6 6 5 - 3 .6 9 9 EBR-II Core 3 . 1 0 . 2 5 8 5 0 .9 9 5 1 3 . 8 5 0 ' w i t h a N i 4 . 1 0.2661 0 .9 8 5 9 3 .7 0 5 5. Reflector (Row 9) 8.2 0 . 2 7 0 7 0 . 9 7 7 6 3 . 6 1 1 1 0 . 3 0 . 2 7 5 0 0 .9 6 8 1 3 . 5 2 0 EBR-II Low 3 .6 0 . 2 3 5 3 0 .9 1 7 9 3 .9 0 2 6. Power (50 kW) Core Center (Row 1) 5 . 9 0 .2 2 6 4 0 .8 6 8 9 3 .8 3 7 Test Run 50 G 11.1 0 . 2 0 7 3 0 .7 6 3 1 3 . 6 8 1 C o re R a d i a l = 7 cm 3 . 4 0 . 2 2 8 3 0 . 8 9 8 1 3 .9 3 4 7. ЕВЗч- I I - F u l l C e n t e r A x i a l = 1 . 6 cm 6 . 4 0 .2 0 9 3 0 .8 0 9 7 3 . 8 6 9 Power Test Row 2 10.1 0 .1 8 7 5 0 .7 0 9 8 3. 785 R un 50H E d g e o f R a d i a l = 7 cm 3 . 1 0 . 2 6 9 3 1 . 0 2 8 3 .8 1 6 8. U p p e r A x i a l = 3 7 cm 6 . 3 0 .2 8 6 7 1 .0 4 4 3 . 6 4 1 B la n k e t 10.6 0 . 3 0 7 3 1 .0 5 1 3 .4 1 9 3 . 2 0 .2 4 1 9 0 .9 4 2 6 3 .8 9 7 4 . 3 0 .2 3 9 4 0 .9 2 4 9 3 . 8 6 3 3 9. Core Center 8.2 0 .2 3 0 4 0 .8 5 9 7 3 . 7 3 1 FFTF-EMC Mockup 10.0 0 .2 2 6 4 0 . 8 3 0 3 3 .6 6 7 i n Z P R -9 3 . 1 0 . 2 6 1 4 1 .0 0 5 3 .8 4 4 Assembly 27 C o r e 6.1 0 .2 7 1 7 1 .0 0 3 3 . 6 9 3 10. R e f l e c t o r 8.1 0 .2 7 7 9 0 .9 9 9 4 3 . 5 9 6 I n t e r f a c e 10.0 0 .2 8 3 7 0 .9 9 3 5 3 . 5 0 2 Demo-Mockup 1 1 . ZPR -6 A s s e m b ly Core Center - Same as 9 above 7
* 2 39 238 Calculated based on initial Pu/ U ratio of 0.250, initial atoms of 239pu = 2,0 and in itia l atoms of 238ц = 4.0.
moderated spectra favor increase in the plutonium content with burnup. In future large Liquid Metal Fast Breeder Reactors with more moderated neutron spectra, these reaction rates may play a significant role in increasing 2j“pu/23ou ratios in certain core regions and thus in increasing the oxygen potentials of irradiated fuel in these locations. In other core locations where the ^39pu/238y ratios remain relatively constant with burnup, the injec tion of fission products into the fuel matrix w ill produce higher oxygen po tentials than those regions where the 239ри/238ц rat ios are decreasing with b u r n u p . IAEA-PL-463/1 5
III. OXYGEN POTENTIALS IN URANIUM-PLUTONIUM FUEL
One of the most important parameters for the oxide fuel is the oxygen potential as a function of temperature, and concentration of oxygen and plu tonium. The experience with uranium oxide in thermal reactors has been that stoichiom etric uranium oxide, i.e., 0/U ratio = 2, is relatively stable with respect to cladding attack. Most of the early work with U-Pu oxide fueled irradiation experiments suggests that the differences between stoichiom etric UO 2 and Ui_yPUy 02, with 0/M = 2, were not fully appreciated. Thus, in the past LMFBR Program the result was to carry out irradiations of fuel with an oxygen-metal ratio of two or higher, inadvertently, and to set fuel specifi cations very near stoichiom etric composition.
There are two essential differences between urania and uranium-plutonium oxide involving the effects of oxygen concentration on fuel behavior. First, at equivalent 0/M ratio and temperature, the uranium-plutonium oxide has a much higher oxygen pressure or more positive oxygen potential, as compared with pure urania. Depending on the specific temperature and composition, the oxygen pressure may be many orders of magnitude higher than that for uranium oxide of the same oxygen concentration. For example, the m etal-rich boundary o f U O 2 at 1000°C is at an 0/U ratio of 1.998. At this oxygen-metal ratio, the oxygen pressure over UQ gPuo. 2O1.998 i s 10^ , 10? , a n d 10* times that over UO 1.998 at 1000°C, 2000°C, and 3000°C. These differences in the two systems are greatest for oxygen metal ratios just below 2 and at the lower tempera t u r e s .
The second difference between UO 2 and U-Pu oxide concerns the shape of the oxygen pressure or oxygen potential curve. For UÛ 2±X there is an extremely large change in oxygen pressure with small solid composition changes near the stoichiom etric composition (see Figure 2). At lower temperatures the change is so sharp for urania as to appear discontinuous. The effect of this is to keep the urania solid essentially at an 0/U ratio of 2 even though the oxygen potential may vary extensively. Consequently, since a ll postulated mechanisms for oxygen redistribution depend on the oxygen potential gradients, the oxygen
0 / M R A T IO
FIG. 2. Calculated oxygen potentials at 2240°C for U02ix, Pu02_x (fluorite single phase) andU0. boP u o .го°2±х. 6 JOHNSON et al. concentration in UO 2 under a thermal gradient is not likely to change radi cally from the center of the fuel to the cladding. Transport of very small quantities of oxygen is required to set up the oxygen potential gradient. Nor is there likely to be much capacity for reaction of oxide with cladding, since a very small amount of oxidation w ill reduce the oxygen potential of UO 2 to a value in equilibrium with the cladding. The effects of burnup are more complex, but enriched UO 2 w ill be less oxidizing as a function of burnup because of the more reducing nature of 235jj fission products as compared to those of 239pu<
For the mixed uranium-plutonium oxide, the oxygen pressure or oxygen potential versus 0/M curve is flatter and, of course, higher than that for ÜO 2 . The result is expected to be a broader oxygen potential gradient from the center of the fuel to the clad. Thus, more cladding oxidation is likely to occur with U-Pu oxide for a given oxygen potential change because the movement of larger quantities of oxygen at higher oxygen potentials is re quired. Because of these important differences, as well as other factors discussed throughout this paper, it is essential to have reliable oxygen potentials as a function of composition and temperature for the whole range of these two variables.
Several investigators have measured oxygen potentials of uranium- plutonium oxide, but the agreement between these measurements is poor. A review of experimental oxygen potentials appears in the paper by Johnson e t a l . [ 5] •
Because of the poor agreement among the various experimental measure ments, Blackburn 1^3 has devised a thermodynamic model for calculating oxygen potentials of U-Pu-oxide.
The model makes use of the observations by Markin and Mclver that the oxygen potential may be correlated with the average Pu valence (assuming U is tetravalent) for MÛ 2_X and with the average U valence Cassuming Pu is tetravalent) for MÛ 2+ X .
Instead of the em pirical relations derived by Markin and Mclver, however, Blackburn relates the cation concentrations to phase boundary and integral thermodynamic values for the oxide fluorite phase. Furthermore, the model covers a much broader composition and temperature range. Because of incom plete data for plutonium oxide phase boundaries and thermodynamic data, it is necessary to make an em pirical fit to oxygen pressure data for PuÛ 2_x -
The model is developed in a stepwise fashion starting with UO 2±x , t h e n adding plutonium, and finally incorporating fission products. Since UO 2 constitutes the bulk of the fuel, a viable model must predict accurate oxygen pressures over m etal-rich and oxygen-rich urania. The principles involved are demonstrated with the model for U 02±x in the paper by Johnson et al. [‘
Figure 3 shows calculated curves of oxygen potential versus temperature f o r U00.8Pu02°2±x at °/M ratios of 2.05, 2.01, 1.98, 1.95, a n d 1.91 w h e r e they are compared to the experimental measurements described above. Note that the curves are not straight lines, indicating, of course, that linear extrapolations are not valid.
The data used to derive oxygen pressures over U02±x give excellent agree ment with experiment• The values used for FuÛ 2_x have a greater uncertainty (±3 kcal) which is carried over to the calculated values for the U-Pu oxide. The curves in Figure 3 show that the oxygen potentials calculated with the IAEA-PL-463/1 7
TEMPERATURE ,»K
FIG .3. C alculated and observed atom ic oxygen pressures over U 0 2±x*
model are close to an average for the experimental values and hence are be lieved to be a reasonable substitute for the measured data. This is espe cially true for oxygen potentials at temperatures and compositions outside the range of the measurements. The model for calculating the oxygen potential is invaluable in calculating other fuel properties such as actinide and fis sion product redistribution, oxygen/metal ratios from oxygen potentials based on molybdenum distribution, oxygen content in fuel in equilibrium with the sodium-fuel reaction product, ИазМОд, etc.
IV. MATERIAL DISTRIBUTION IN A THERMAL GRADIENT
A. Distribution of Oxygen and Oxidized Fission Products in Thermal Gradient
1. Introduction
When a virgin fuel pin is brought to power, the radial and axial temperature gradient developed leads to a gradient in the oxygen potential (A G q 2 = RT ln PO 2) > *-ke Partial pressures of the various gaseous fuel oxide s p e c i e s (U 0 2 , U O 3 , PU O 2 , PuO) and the activities of the uranium, plutonium, and oxygen ions in the solid oxide. These gradients w ill lead to a tendency for the oxygen, uranium, and plutonium to redistribute. As the concentra tions of fission product elements increase due to burnup, some of these ele ments w ill also deviate from a uniform distribution and w ill participate in the transport processes in the fuel. Because of the gradual change in power level and overall composition due to burnup, a steady state is probably never 8 JOHNSON et al.
attained in the fuel pin. Since a combination of solid state diffusion and gaseous transport processes is involved, the concentration profiles at a given power level and time of irradiation w ill be influenced by the porosity of the oxide which depends on the in itia l density of the oxide. The clad can also react with oxygen or fission product elements and therefore must be considered. These processes can lead to a non-uniform distribution in the fuel pin of oxygen, uranium, plutonium, and some of the fission product ele ments. Of particular importance to the length of time of the fuel pin may remain at power in the reactor is the distribution of oxygen between the oxide and clad.
Theoretical estimates of the oxygen potential gradient in a radial temperature gradient for a urania-plutonia solid solution have been made by R a n d a n d R o b e r t s t 7 J [ 8 ] an V. FUEL CLADDING ATTACK The intergranular penetration of fission products into the austenitic stainless steel cladding of EBR-II irradiated mixed oxide fuel pins was first observed over four years ago by experimenters at Argonne National Laboratory.I ^ Since that time a number of irradiated fuels have been examined and an appreci able number have shown evidence of this phenomenon. A ll the facts concerning the correlation between intergranular attack, fuel fabrication, and reactor operating parameters are not known. However, studies have shown that the principal parameters initiating and controlling the cladding attack are the oxidizing potential, the amount of available oxygen, the fuel thermal gradi ents, and the temperature at the inner cladding wall. Unfortunately, these parameters either have not been accurately known or controlled with sufficient accuracy in many irradiated mixed oxide fuels to allow accurate correlations of them with cladding attack. Also, studies of irradiated cladding have shown that selected fission products (e.g., cesium and molybdenum) certainly accel erate oxidation of the cladding. Also, their transport to the clad wall within the fuel is controlled by both the local oxygen potentials and temperatures. Two types of cladding attack have been observed at the fuel clad interface. One is a recession of the cladding thickness by a uniform oxida tion (matrix attack) of the stainless steel. The second is an intergranular penetration of the cladding by oxygen and certain fission products along grain boundaries. Examination of high burnup irradiated pins indicated that the uniform oxidation is generally lim ited to less than 12% of the cladding thick ness, whereas intergranular corrosion, when it occurs, can be considerably deeper. Experimental results on cladding attack, together with a discussion of the mechanisms of attack, are presented in detail by Johnson et a l.[5]. IAEA-PL-463/1 9 VI. OUT-OF-PILE CORROSION EXPERIMENTS Out-of-pile experiments have been run to study the effects of fission products (Cs, Mo, Te, and I) and other potential fuel pin im purities (Cl- and OH-) on austenitic stainless steel. The object of these studies is to establish the reactants and reaction conditions that w ill simulate the type of corrosive attack observed in irradiated fuel pins. Two approaches have been followed in these experiments. First, differential thermal analysis has been used to examine compositions in the CS 2O-CS 2M 0O4-M 0O3 system to provide inform ation on temperatures and compositions of likely fused salt electrolytes which exist on the cladding surface. Second, capsule tests were initiated to determine the specific conditions under which different types of cladding corrosion occur. Examination of a ll m aterials were carried out using metallographic and microprobe techniques. Work was done in flowing ultra-high-purity helium atmosphere. These experiments are summarized in Table II. This data shows that liquid phases of cesium and oxygen or of cesium-molybdenum and oxygen exist in the temperature range above 500°C where in ter granular, attack is observed. Deep intergranular attack only occurred at high oxygen potentials in Run GS-12, shallow intergranular attack at low oxygen potential in the presence of a liquid phase. Because of the importance of oxidizing potential on the nature and extent of the corrosive attack, experiments were run in welded capsules. The closed system allows better control of the low oxygen potentials required even when run in ultra-high-purity helium. In these experiments CS 2O , C S 2O 2 + C S 2O 3 , CsOH, CsCl, CS 2M0O4 , M 0O 3 in various combinations were encapsulated in 20% cold-worked 316 stainless steel tubing (0.230 in. dia. with 0.015 in.-thick wall) and then heated to 650°C in helium for 144 hrs. After the heat treat ment, each capsule was sectioned and examined m etallographically; selected capsules were also sectioned and examined by the electron microprobe. Residual m aterial was selected from each capsule for X-ray diffraction analysis. The results are summarized in Table III. The types of corrosive attack observed m etallographically can be classi fied into three distinct categories: ( 1 ) shallow intergranular, ( 2) u n i f o r m surface oxidation, and (3) intergranular. The various forms of attack are described below. Shallow intergranular attack was observed in capsules C0-1, 2 , 3 , 8 , 11, 13, and 19. In these capsules the appearance of the residual m aterials in the capsules after sectioning indicated that they had been par tially or totally liquid during the heat treatment. Furthermore, liquid cesium metal flowed from a ll but CO-11 when opened indicating CS 2O h a d b e e n r e d u c e d by the stainless steel. Figure 4 shows a photomicrograph and electron micro probe X-ray images of a section of capsule CO- 8. The attack probably in iti ated along grain boundaries, but as reduction of CS 2O occurred attack by the liquid appears to have concentrated at triple points of grain boundaries. Shallow intergranular attack was confined to capsules that contained CS 2O alone, in combination with 50 mol % Cs metal, CsCl or CsOH and less than 5 0 m o l % M 0O 3. In a ll but one experiment-(CO-11; CS 2O + C s O H ) , t h e C S 2O w a s partially reduced to metal by the stainless steel and liquid cesium was observed when the capsules were opened. The extent of attack of the stainless steel was minimal for capsules containing CS 2O , C S 2O + C s , a n d C s 20 + CsCl. More extensive attack occurred w i t h C S 2O + M 0O3 and s till more extensive attack occurred with CS 2O + C sO H . This appears to follow the increase in oxygen potential; however, CsOH may undergo reaction by other mechanisms and further study of this compound is needed. The shallow intergranular attack appears to be associated with a liquid-solid reaction. T A B L E I I DTA and Capsule Experiments in Type 304 Stainless Steel R u n N o . M a t e r i a l 3 Experimental Conditions^ R e s u l t s Type of Attack C S - 2 CS 2O Five thermocycles ; maxi Breaks at 395 and 440°C in , Shallow intergranu- mum temp, of each cycle: first cycle; size of breaks l a r a t t a c k 730°C; holding temp, be diminished with each cycle; tween cycles: 650°C; X-ray diffraction showed a t im e : 120 h r major phase fee lattice, a Q = 8 . 3 7 A C M -1 50 mol % CS 2O - Eleven thermocycles ; Breaks at 560-570°C; X-ray Uniform oxidation and shallow intergranular 50 mol % CS 2M 0O 4 maximum temp. of each diffraction showed only al. et JOHNSON cycle varied from 670 to CS 2M 0O4 a t t a c k 925°C; time: 100 hr C M - 7 C s 2M o O ^ Six thermocycles; maxi Thermogram erratic at high Uniform oxidation mum temp, of each cycle: temperatures; no reaction 950°C; holding temp, be peaks at lower temperatures; tween cycles: 650°С ; X-ray diffraction showed t im e : 120 h r o n l y C S 2M 0O4 C M M - l 7 0 m o l % M 0O3- Eleven thermocycles; Eutectic about 455°C; liqu i- Uniform oxidation 30 mol % CS 2M 0O 4 maximum temp, of each dus about 500°C cycle varied from 670 to 925°C; time: 100 hr G S - 1 2 CS 2O 2-CS 2O 3-CS 2O ’ Capsule held at 690°C Major phage fee lattice Deep intergranular Commercial cesium f o r 1 1 7 h r a 0 = 8 . 3 7 A attack with cesium in o x i d e the grain boundaries, iron and nickel de pleted and chromium variable-depleted or e n r i c h e d . TABLE III Summary of Isotherm al Capsule Experiments Using Type 316 Stainless Steel (20% cold worked) at 650°C Run No.a Sample M aterial ,'5 m o l % X-ray Diffraction Results Metallographie Results CO-1 50 Cs20 + 50 Cs Liquid cesium flowed from capsule when Shallow intergranular attack. cut open. X-ray sample could not be o b t a i n e d . C 0 - 2 C s 2 0 Same as Run CO-1. Shallow intergranular attack. C0-3 75 Cs20 + 25 M 0O3 Same as Run CO-1. Shallow intergranular attack. More extensive than in Run C O - 2 . CO-4 50 Cs20 + 50 M 0O3 С з 2МоОд. Slight indication of uniform o x i d a t i o n . CO-5 25 Cs20 + 75 M 0O3 X-ray pattern could not be identified, Uniform oxidation. probably a Cs-Mo-0 compound. CO- 6 M 0O3 M 0O3 with one or more minor phases Uniform oxidation. unidentified. CO-7 50 "Cs 20 " + 5 0 C s 2M o 0 4 C s 2M o O ^ . Intergranular attack, about 2 - 3 m i l s . CO-8 C s 2 0 Liquid - cesium flowed from capsule when Shallow intergranular attack. cut open. X-ray sample could not be o b t a i n e d . CO-9 50 Cs20 + 50 Cs 2M o 0 4 Major phase not identified; Cs 2MoO¿, w a s Intergranular attack less a minor phase. severe than Run CO-7. CO-10 50 "Cs 20" + 50 CsOH AB 2X4, fee structure sim ilar to spinels. Severe intergranular attack, capsule ruptured; average pene tration 5-6 m ils, greater in some areas. CO-11 50 Cs20 + 50 CsOH X-ray diffraction pattern could not be Shallow intergranular attack, identified. more extensive than in Run CO-3. CO-12 50 "Cs 20" + 50 CsCl X-ray pattern not obtained. Intergranular attack, about 2 - 4 m i l s . TABLE III CCont'd) R u n N o . a Sample M aterial,^ mol % X-ray Diffraction Results Metallographie Results C O - 1 3 50 Cs20 + 50 CsCl Liquid Cesium flowed from capsule when Shallow intergranular attack. c u t o p e n . C O - 1 4 5 0 " C s 20" +•50.С , X-ray pattern could not be identified. Intergranular attack, about 3 - 4 m i l s . C O - 1 5 5 0 " C s 20 " + 5 0 M 0O3 C s 2M o 0 4 Uniform oxidation; uniform layer of oxide, thicker where sample m aterial con tacted capsule wall. al. et JOHNSON C O - 1 6 50 CsCl + 50 Cs 2M o 0 4 CsCl and CS 2M 0O 4 Isolated areas of uniform oxidation observed. C O - 1 7 50 CsCl + 50 CsOH Not enough m aterial remained in capsule Very severe intergranular to obtain an X-ray sample. attack, capsule ruptured; some other areas of nearly 1 total penetration. C O - 1 8 " C s 20 " АВ 2Х д , sa m e a s C 0 - 1 0 Intergranular attack, about 3 - 4 m i l s . C O - 1 9 70 Cs20 + 30 M 0O3 Same as CO-1. Shallow intergranular attack, same as Run CO-3. C 0 - 2 0 C s 2M 0O4 C s 2M o 04 No apparent attack. ^uns CO-1 through CO-6 were heated for 120 hr. A ll other runs were for 144 hr. k " C s 20" designates commercially supplied cesium oxide, which was shown by X-ray diffraction to be predominately CS 2O 2 a n d C s 203* IAEA-PL-463/1 13 - "■ О':. ", ■ -.■ ■..'■.■'-■.■'■'T! :'.'У . ■' ■ ■■■ i /■'vT1 :'I ■ ■ ‘ „ Л ■ ». '■ ■■ ^ .»-■ ■' и н м и н b \ , . г : -ü ..,. .'■ »mssmmm я— — fil a. PHOTOMICROGRAPH 200Х b. SPECIMEN CURRENT AS-POLISHED d. CHROMIUM Кп e. NICKEL K„ f. CAESIUM La FIG. 4. Photomicrograph and electron microprobe scanning images of shallow intergranular attack on Type 316 stainless steel by CszO in Run C O -8 (all X-ray im ages are 80 x 100 fim)- Uniform oxidation was observed in the capsules CO-4, 5, 6 , 1 5 , a n d 1 6 . In capsule CO-15, containing CS 2O2 + C S 2O3 a n d M 0O3 , a uniform oxide layer was formed on the interior surface of the stainless steel in contact with the sample m aterial. The photomicrograph and electron microprobe X-ray images of a typical area of CO-15 are shown in Figure 5. The electron microprobe X-ray images show chromium and nickel enrichment and iron deple tion have occurred in the oxidized area of the stainless steel. The small piece of sample m aterial adhering to the oxide layer is cesium and molybdenum (not shown) with traces of iron. X-ray diffraction analyses of the sample residue yielded Cs 2M o 04 only. In all capsules exhibiting uniform oxidation, examination of the sample residues indicated that the m aterials had remained 14 JOHNSON et al. a. PHOTOMICROGRAPH 200X b. SPECIMEN CURRENT AS-POLISHED c. IRON Ka d. CHROMIUM KQ FIG. 5. Photomicrograph and electron microprobe scanning images of uniform oxidation of Type 316 stainless steel by Cs20 2 + Cs20 3 and Mo03 in Run CO-15 (all X-ray images are 80 X 100 i¡m), solid' and that very little sintering had occurred. Also, very little attack occurred except in the areas where the solid sample m aterial was in direct contact with the stainless steel (i.e., the bottom section of the capsule). Although the uniform oxidation observed in capsule CO-15 appears some what sim ilar to the air oxidation of stainless steel, the depth of attack is much greater than the thickness of the scale formed on type 316 stainless steel oxidized in air at 650°C for 144 hrs. Also, the oxidation appears to lIt is possible that liquid could have been present early in the experiment but was consumed by reacting with the stainless steel or by interaction in the sample m aterial [e.g., Cs 20(l) + MoOjCs) ■* CsMoO^Cs) ] . IAEA-PL-463/1 15 FIG. 6. Photomicrograph of the ruptured section in Capsule CO-17 containing CsOH and CsCl. Severe intergranular attack and matrix attack propagating from grain boundaries is evident (600 x). have occurred by an inward reaction with islands of essentially unreacted alloy, whereas, in the air oxidation of stainless steel, the protective oxide layer of СГ 2О3 is formed by outward diffusion of chromium. Severe oxidation of stainless steels in air (1000-1200°C) results in localized scale breakdown and rapid inward oxidation to form stratified nodules.Г36-38] distribution of elements on the nodules is generally such that three layers are formed: ( 1 ) iron with some nickel outermost, ( 2) iron-chromium center, and (3) chromium at the oxide/alloy interface. It is apparent that uniform oxidation in Figure 5 is not a simple oxidation reaction, but is a more complex reaction involving cesium and oxygen. Also, the residual m aterial and location of attack indicate this is a solid-interaction with no liquidus phase involved. Intergranular attack was observed in seven of the capsule experiments in Table III; these were CO-7, 9, 10, 12, 14, 17, and 18. The most severe inter granular attack occurred in CO-10 and CO-17, in which the capsules ruptured. A photomicrograph of the ruptured section of capsule CO-17 is shown in Figure 6,. Also evident in the figure is the propagation of uniform oxidation in the grain boundaries. A photomicrograph and electron microprobe X-ray images for capsule CO-IO are shown in Figure 7. The photomicrograph shows extensive intergranular attack along the interior surface, with a crack extending through the capsule wall. The grain boundaries have been depleted of iron, chromium, and nickel and enriched with cesium. Chromium depletion is evident along some grain boundaries, whereas others show chromium enrichment. The chromium depletion occurs in those grain boundaries with the most cesium. 16 JOHNSON et al. a. PHOTOMICROGRAPH 128X b. SPECIMEN CURRENT e. NICKEL Ka f. CAESIUM L„ FIG. 7. Photomicrograph and electron microprobe scanning images of intergranular attack on Type 316 stainless steel by Cs20 2 + Cs20 3 and CsOH in Run C -10 (all X-ray images are 80 x 100 :im), In the capsules that experienced intergranular attack, the sample residues appear to have been partially liquid. In capsule CO-15 the sample residue had remained powdery. Also, intergranular attack occurred over the entire length of the capsule, as compared to uniform oxidation which was largely confined to the bottom section in contact with the solid sample m aterial. This behavior is consist ent with our belief that intergranular attack involves a liquid electrolyte- solid reaction. A series of capsule experiments, CO-21 through CO-40 were conducted to obtain a comparison of 304H, 304, 321, and 347 stainless steel to that of type IAEA -P L -463/ 1 17 E O â Ê a f c b . ... a. PHOTOMICROGRAPH 200X b. SPECIMEN CURRENT A S - POLI SHED ■BP ШВЩШИЯШ111 I— 11- ■ ■ .-'Ж - ■ ■■ ■ с. IRON Ka d. CHROMIUM Ka -I Ж ШЁЛЁШ w s t S B m e. NICKEL Ka f. CAESIUM La FIG. 8. Photomicrograph and electron microprobe scanning images for Capsule CO-28. 316 when subjected to the sim ilar environment'. Type 304H and 304 were annealed and large grain, while types 321 and 347 were 20% cold worked and fine grain. The 321 and 347 samples were supplied by B. F . Rubin of General Electric, and after annealing at 650°C for 144 hours, these m aterials appeared sensitized (i.e., when electrolytically etched in 10% o x a l i c a c i d f o r 10 seconds, carbide precipates were observed inter and intragranularly). The results of these experiments are summarized in Table IV. Generally the corrosion attack observed was sim ilar to 316. However, penetration was somewhat less in 321 and 341 and both 347 and 321 experienced a circum ferential internal separa tion about 4-6 m ils from the interior surface as shown in Figure 8 f o r c a p sule CO-28. This did not occur in 316 capsules. Capsules CO-29 through 32 TABLE IV со Summary of Isotherm al Capsule Experiment in Types 304H, 304, 321, and 347 Stainless Steels at 650°C, 144 hrs R un Sample M aterial Stainless Steel M etallography Results N o. m o l % Capsule M aterial Comments on M etallographie Examination C O -2 1 50(Cs202+Cs203)+ Type 304H-Annealed Intergranular corrosion attack, about 2-3 m ils pene 5 0 Cs2M oO^ tration. Attack was evident over entire length of c a p s u l e . C O -2 2 m Type 304-Annealed Same as CO-21, but less penetration (^1-2 m ils). n C O -2 3 Type 321-20% cold Intergranular corrosion attack with numerous loose w o rk e d grains in zone of attack. Penetration was about 1-2 m ils over entire length of capsule. и C O -2 4 3 Type 347-20% cold Same as CO-23. ONO e al. et JOHNSON w o rk e d C O -2 5 5 0 ( C s 2 0 2 + C s 2 0 3 ) + Type 304H-Annealed Intergranular corrosion attack. The attack was most CsOH evident at the interior surface and in a band about 4-6 mils from the interior surface. The zone in between was less effected. Numerous loose grains were lost from interior surface during polishing. и C O -2 6 Type 304-Annealed Intergranular corrosion attack with about 6-8 mils penetration. Numerous loose grains were lost from interior surface during polishing. и C O -2 7 Type 321-20% cold Same as CO-25 except penetration was slightly less. w o rk e d n C O -2 8 3 Type 347-20% cold Same as CO-25 and -27. In this capsule, circum fer w o rk e d ential separation occurred at the internal band of intergranular attack. C O -2 9 50 CsCl+50 CsOH Type 304H-Annealed Capsule ruptured. Slight intergranular corrosion attack on interior surface. Also, corrosion attack on exterior surface contacted by leaked sample m aterials. C O -3 0 и Type 304-Annealed Same as CO-29. C O -3 1 3 M Type 321-20% cold Same as CO-29, except intergranular penetration on w o rk e d interior surface was greater; about 1-2 m ils. This may indicate that rupture occurred later in the experiment. TABLE IV (Cont * d) R u n Sample M aterial Stainless Steel Metallographie Results N o . m o l % Capsule Material Comments on Metallographie Examination C O - 3 2 50 CsCl+50 CsOH Type 347-20% cold Same as CO-29 w o rk e d C O - 3 3 (CS 2O 2+ C S 2O 3) Type 304H-Annealed Same as CO-21, except intergranular penetration was slightly greater; about 3-4 m ils. C O - 3 4 и Type 304-Annealed Same as CO-21. C O - 3 5 и Type 321-20% cold Same as CO-23. w o rk e d C 0 - 3 6 a II Type 347-20% cold Same as CO-23. w o rk e d C O - 3 7 a 50 Cs20+50 CsOH Type 304H-Annealed Corrosion attack appearing as intergranular subsurface voids in same areas. Penetration was about 3-4 m ils. Attack was less near top of capsule. C O - 3 8 M Type 304-Annealed Same as CO-37. C O - 3 9 II Type 321-20% cold Same as CO-37 and -38 except corrosive attack less w o rk e d severe; penetration on about 1 m i l . C O - 4 0 II Type 347-20% cold Same as CO-37 and -38 except corrosive attack was w o rk e d less; about 1-2 m ils penetration. aThese capsules were examined by the electron microprobe and photo composites prepared. ^Each capsule was examined near the bottom, middle, and top regions. The penetration indicated for each capsule was that observed in the bottom section. Generally, the amount of penetration was about the same in a ll sections, although grain pull-out during polishing was less in the upper sections. с The residual m aterial in these capsules appear to have been partially or totally liquid during the experiments. Capsules CO-29 thru 32 must have been liquid since no residual m aterial remained in c a p s u l e . ONO e al. et JOHNSON a. PHOTOMICROGRAPH 300X b. SPECIMEN CURRENT c. IRON Ka AS-POLISHED d. CHROMIUM Ka e. NICKEL Ka f. CAESIUM La g. TELLURIUM L„ h. MOLYBDENUM La FIG. 9. Photomicrograph and electron probe scanning images of corrosive attack on Type 316 stainless steel by CSgOg + Cs20 3 and Te (Capsule CO-44; all X-ray images are 266 x 320 (im ). t o с о ONO e al. et JOHNSON g. MOLVßOENCiM La FIG. 10. Photomicrograph and electron microprobe scanning images of intergranular pitting attack Type 316 stainless steel by Mo03 and Te (Capsule CO-45; all X-ray images are 80 x 100 jjm). с о ONO e al. et JOHNSON d. CHROMIUM KQ e. NICKEL Ka f. CAESIUM La g. TELLURIUM t a h. M O LYBDENUM La FIG. 11, Photomicrograph and electron microprobe scanning images of intergranular attack on Type 316 stainless steel by CS2M0O4 and Te (Capsule CO-43, inner surface, capsule ruptured; all X-ray images are 160 x 200 jim ). M CTS ONO e al. et JOHNSON d. CHROMIUM Ko e. NICKEL Kq f. CAESIUM La " - . ** ; » J-s'-ra,~Ss¥'.'~y* *3S*pS ‘ A/* *«' ^ IAEA-PL-463/1 i ■ ■ ■ ; . \\\ - V g. TELLURIUM La h. MOLYBDENUM La FIG. 12. Photomicrograph and electron microprobe scanning images of intergranular attack on Type 316 stainless steel by CSzMo04 and Te (Capsule CO-43, outer surface at rupture; all X-ray images are 160 x 200 /im). t o -j TABLE V Summary of Isothermal Capsule Experiments in Type 316 Stainless Steel (20% cold worked) at 650°C, 144 hrs R u n Sample M aterial X-ray Diffraction Metallography Results N o . m o l % R e s u l t s C O - 4 1 50 Cs20+50 Cs 2M o 0 4 Unidentified compound. Corrosion attack appearing as intergranular subsur face voids in some areas through the dissolution and subsequent removal of m aterial. Penetration was about 1 m il. Attack was less near top of cap s u l e . C O -4 2 50 Cs20 + 50 Te Unidentified phase(s) Corrosive attack that appears as layers or bands containing Cs,and Те of reaction products separated by layers or bands of unreacted metal. C O - 4 3 a 5 0 C s 2M o 0 4+ 5 0 T e C s 2M o 04 o n l y Intergranular corrosion attack with m atrix attack propogating from grain boundaries. Capsule ruptured, but some residual m aterial remained in capsule. Corrosion attack occurred on exterior surface contacted by the leaked residual m aterial. . C O - 4 4 3 5 0 ( C s 2 ) 2+ С б 2 0 з ) + X-ray diffraction Same as CO-42, except the corrosion attack was 5 0 Те pattern was sim ilar more severe. to that for C O - 4 2 . C O - 4 5 3 5 0 M o 0 3 + 5 0 Т е Major phase was Mo0 2 Intergranular corrosion attack, about 1 m il penetration. C O - 4 6 3 50 CsOH+50 Cs 2M o 0 4 C s 2M o 04 o n l y Capsule ruptured, severe intergranular corrosion attack sim ilar to CO-IO and -17. C O -4 7 50 CsOH+50 Mo0 3 Unidentified phase(s) Uniform corrosion attack sim ilar to CO-15. Somewhat sim ilar to C O - 5 . C O - 4 8 5 0 M 0O3+ 5O С б 2М оОд - Slight amount of uniform corrosion attack. IAEA-PL-463/1 29 contained CsCl + CsOH with each of the sample types and a ll ruptured. This was sim ilar to the performance of 316 in capsule CO-17. A third group of experiments with 316 stainless steel is summarized in Table V. These experiments were conducted to observe the effect of tellu rium additions on the nature of the corrosive attack and to recheck and expand on the series CO-1 through 20. Note the results of CO-46 as compared, with CO-9, 10, 11, 16, and 17. Of particular interest in this group are the capsules CO-42, 43, 44, 45, and the usually severe corrosive attack observed in the presence of tellurium . The combinations of tellurium + CS 2O a n d C S 2M 0O4 in capsules CO-42 and 43 were more corrosive than CsOH + CS 2O in capsule CO-11. This would indicate that tellurium is capable of attack at the relatively low oxygen potentials in capsules 42 and 43 and the attack may possibly function independent of this parameter. In prototypical reactor fuels we have not observed telluri um in the grain boundaries of cladding showing intergranular attack, although the tellurium was observed on the outer cladding wall. In Figures 9, 10, 11,, and 12, the addition of tellurium to CS 2M 0O4 , M 0O 3 , C S 2O , C S 2O 2 + C S 2O3 caused corrosive attack even with the m aterials that previously had exhibited compatibly with 316 stainless steel. In an extension of some of the previous studies, capsule CO-46, C sO H + C s 2M o 0 4 showed severe attack while CO-47, CSOH-M 0O 3 , and CO-48, . M 0O 3 + C S 2M 0O 4 exhibited a relatively mild uniform corrosion. The latter two sample mixtures in CO-47 and 48 remained completely solid while in CO-46 liquid + CS 2M 0O4 solid was present. The data so far strongly supports the view that the dangerous rapid intergranular penetration of stainless steel requires a liquid electrolyte while uniform oxidation of the surface occurs with solid-solid and gas- solid interactions. A shallow intergranular attack occurs when the oxygen potential is reduced so that the liquid cesium salts are reduced to cesium metal. Cesium hydroxide and tellurium are very aggressive in promoting inter granular attack. The studies on these compounds are continuing. REFERENCES 1. M. H. Rand and T. L. Markin, "Some Thermodynamic Aspects of (U,Pu )02 Solid Solutions and Their Use as Nuclear Fuels," Thermodynamics of Nuclear M aterials, International Atomic Energy Agency, Vienna, 1968, pp. 637-650. 2. T. L. Markin and E. J. Mclver, "Thermodynamic and Phase Studies for Plutonium and Uranium-Plutonium Oxides with Application to Com patibility Calculations," Plutonium 1965, edited by A. E. Kay and W. B. Waldron, Chapman and H all, Ltd., 1967, pp. 845-857. 3. H. Holleck and H. Kleykamp, "The Stoichiometry Shift in an Oxide Fuel Element at High Bum up," Karlsruhe, August, 1970, EURFNR-836. 4. N. D. Dudey, Private Communication, Argonne National Laboratory, June, 1972. 5 . C. E. Johnson, J. Johnson, P. E. Blackburn, and C. E. Crouthamel, "Effects of Oxygen Concentration on Properties of Past Reactor Mixed-Oxide Fuel", Reactor Technology 1¿ 4 (1972). 30 JOHNSON et al. 6 . P.E. Blackburn, to be published. 7. M. H. Rand and L. E. J. Roberts, "Thermochemistry and Nuclear Engineering," Proc. Symposium on Thermodynamics, I.A .E.A ., Vienna, 1966, p. 12. 8 . M. H. Rand and T. L. Markin, "Some Thermodynamic Aspects of (и,Ри)0~ Solid Solutions and Their Use as Nuclear Fuels," AERE-R5560 (August, 1967). 9. E. A. Aitken, M. G. Adamson, S. K. Evans, and T. E. Ludlow, "A Thermodynamic Data Program Involving Plutonia and Urania at High Temperatures," Quarterly Report No. 17, GEAP-12254 (Oct., 1971), p. 3-14. 10. I. Johnson, C. E. Johnson, C. E. Crouthamel, and C. A. Seils, "Oxygen Potential of Irradiated Urania-Plutonia Fuel Pins," to be published. 11. N. R. Stalica, C. A. Seils, and C. E. Crouthamel, ANL Chemical Engineering Division Annual Report, ANL-7575, p. 102 (March, 1968). IAEA-PL-463/2 THE COMPOSITION AND CHEM ICAL STATE OF IRRADIATED OXIDE REACTOR FU EL M ATERIAL J.R. FINDLAY United Kingdom Atomic Energy Authority, Research Group, Applied Chemistry Division, AERE, H arw ell, Didcot, Berks., England Abstract THE COMPOSITION AND CHEMICAL STATE OF IRRADIATED OXIDE REACTOR FUEL MATERIAL. The composition of irradiated fuel may be calculated readily from the established fission yields and cross-section data. Examples of calculations undertaken by computer are given and differences due to the type of fission, capture cross-sections and irradiation times are considered. The calculated compositions are applied to oxide fuels to establish the change in oxygen balance occurring with burn-up. Some discussion of the uncertainties is given with reference to the behaviour of caesium, rubidium and niobium. The redistribution of oxygen under the radial temperature gradient of a fuel element is considered in relation to its effect on oxygen to metal ratios and oxygen potentials. 1. INTRODUCTION The chemical state of reactor fuel material during irradiation is influenced by the compositional changes resulting from the fission process and by the redistribution of both fuel and fission product phases. These effects are of concern to a ll reactor systems but they are particularly evident in fast reactors, where the build up of fission products is appreciable and where operating temperatures are high. The associated problems have been studied w ithin the U.K.A.E.A. by both practical and theoretical methods; the current status of this work is reviewed. 2. THE COMPOSITION OF IRRADIATED FUEL The primary need of a thermochemical assessment is to establish the composition of a fuel material at intervals during its irradiation history. Many computer programmes have been w ritten for thi 3 purpose, principally to establish inventories of short-lived or hazardous radioactive species for safety calculations ¡J , 2 j . In the latter, a ll members of fission product chains have to be considered whereas, for chemical considerations, considerable sim plification is possible because the short-lived activities are of relatively minor importance. The species that contribute principally on a mass basis are the stable and long-lived fission products. Other factors that are important are the composition changes in both heavy metal and fission product species arising from burn-out and transmutation effects. A computer programme has been w ritten at ABES, Harwell, specifically to calculate chemical compositions [_5] • It has been used to calculate fuel compositions after 10^2 burn-up in various idealised cases. The final concentrations are given in Table 1 expressed as a percentage of the in itia l heavy metal atom concentration. Cases considered are the thermal fission 31 32 FINDLAY TABLE 1. FISSION PRODUCT AND HEAVY METAL ATOM CONCENTRATIONS IN IRRADIATED FUEL AT 10% BURN-UP EXPRESSED AS PERCENTAGE OF INITIAL HEAVY METAL ATOM CONTENT Case No. 1 2 3 4 5 6 Therm al Therm al T herm al Fast Fast Fast Fission C o Initial heavy Z35U 100 100 100 100 0 0 .5 0 m etal content гзв^ 0 0 0 0 0 69. 02 23sPu 0 0 0 0 10 0 2 3 .9 4 24»Pu 0 0 0 0 0 5 .4 7 241 Pun 0 0 0 0 0 0 .9 3 242 pu 0 0 0 0 0 0. 14 Fuel rating (w • g"1) 10 1000 1 000 1000 200 200 Irradiation time (d) 10 150 102 102 86 514 476 Cooling time (d) 0 0 100 0 0 0 235u 88. 38 88. 22 88. 22 86. 89 0 0. 28 Final heavy 236U 1. 73 1 ,7 4 1. 74 3. 05 0 0 .4 7 metal content ,,, Np 1 x 10“3 1 x 10~3 1 X 1СГ3 4 .1 X 10~2 0 2. 6 X 10~3 238u 0 0 0 0 0 64. 50 239Pu 0 0 0 0 87. 25 17. 44 240Pu 0 0 0 0 2 .7 0 6. 36 241 Puri 0 0 0 0 4. 9 x 10‘ г 1. 08 241Am 0 0 0 0 1. 2 X 1 0 -3 6 .4 x 1 0 -2 242Pu 0 0 0 0 4 . 8 X Ю" 4 0. 22 Ge 4. 0 X 10^ 4. 0 x 10"4 4 X 10-4 3. 2 X 10-3 0 7 X 10-5 As 1 .2 X 1 0 '4 1. 4 x 1 0 '4 1. 2 X 10~4 1. 9 x 10~3 1 x 10-5 4 X 10-5 Se 4. 2 x i o “2 4. 2 X 1 0 'z 4. 2 X 10-2 8. 6 x 10-2 5, 0 X 10*5 4 .8 X 10-2 Br 1. 8 x l o '2 1. 8 X 1 0 '2 1. 8 x 10~2 3 .1 x 10‘ 2 1. 8 X 10-2 1. 8 X 10-2 Kr 0 .3 5 0 .3 7 0 .3 7 0.3 9 0 .1 9 0. 20 Rb 0. 37 0 .3 6 0. 36 0.3 8 0. 17 0. 17 Sr 0. 78 1, 20 1. 00 1. 18 0. 37 0, 39 Fission product Y 0 ,4 8 0, 57 0. 52 0 .5 6 0, 20 0, 21 concentration Zi 3. 26 3. 16 3. 14 3. 10 1 .9 9 2. 02 X Nb 3 .0 x 10'3 0. 14 0. 11 0. 14 СЛ о 5 x 10‘ 2 Mo 2 .4 8 1. 99 2. 25 1. 96 2. 07 2. 06 T c 0. 61 0. 60 0. 62 0,5 7 0. 58 0. 58 Ru 1. 10 1. 28 1. 15 1. 52 2. 12 2 .1 1 Rh 0. 29 0. 16 0. 27 0. 18 0. 52 0. 52 IAEA-PL-463/2 33 TABLE 1. (cont. ) Case No. 1 2 3 4 5 6 Fission Therm al Therm al Therm al Fast Fast Fast Co Pd 0. 16 0 .1 2 0.1 3 0. 23 1 .4 2 1. 37 A g 3. 0 x 10'3 3. 2 x 10-3 3. 0 x 10'3 1. 5 X 10~2 0. 16 0. 16 Cd 7 . 7 x 1 0 '3 7. 6 x 10"3 7. 7 x 10'3 1. 9 X Ю-2 8. 1 X 10’ z 8. 3 X 10-2 In 9. 6 x llT 4 9 . 2 x 10-4 9. 7 x 10'4 '2 . 8 x 10"3 8. 9 x IO"3 8. 6 x IO'3 Sn 8. 6 x 10'3 8. 9 x 10"3 8. 6 x 1 0 '3 3. 7 X IO-2 5. 3 x 10-2 5. 1 x 10"2 Sb 3 .3 x 10~3 6. 2 x 1 0 '3 4. 9 x 1 0 '3 5. 4 X 10~2 3. 6 x 1 0 '2 3 .4 x 10“2 T e 0. 26 0. 29 0.26 0.3 5 0 .3 4 0 .3 2 I » 0. 11 0 .1 4 0 .1 1 0 .2 1 0 .1 7 0. 16 X e 2. 16 2. 61 2 .6 1 2.1 0 2. 04 2 .0 5 Cs 1 .75 1 .4 2 1.47 1. 88 1.8 8 1. 87 Ba 0. 84 0. 78 0. 68 0,8 1 0. 65 0. 65 La 0. 65 0 .6 6 0. 65 0. 63 0 .5 3 0. 54 C e 1 .2 4 1. 84 1. 63 1, 78 1. 27 1. 29 Pr 0. 61 0 .4 8 0. 58 0 .4 8 0 .4 0 0 .4 2 Nd 2. 02 1 .49 1. 68 1. 48 1 .4 1 1 .4 4 Pm 2 . 9 x 10'J 0. 18 0. 19 0. 19 0. 16 0. 17 Sm 0 .37 0. 18 0. 20 0 .2 1 0 .3 6 0 .37 Eu 1. 8 x 10"2 1. 8 X l ( f 2 1. 8 X 10~2 2. 7 x 10"2 6. 9 x 10"2 6, 9 X 10"2 Gd 4. 3 x 1 0'3 3. 5 X IO“3 4. 3 X 10'3 4. 9 X 10"3 4. 5 X 10-2 4. 3 x 10-2 Tb 1. 0 X lO-4 1. 1 X 10-4 1. 1 x 10"4 3 .4 x 10"3 4. 5 x 10-3 4 .4 x 10"3 Dy 1. 0 X 1 0 '5 1. 0 x 1 0 '5 1. 0 x 10~5 4. 0 x 10"5 1. 5 x IQ"3 1 .5 x 10-3 o f u r a n iu m - 235 a t a r a t i n g i n o x i d e f u e l o f 10 and 1000 fast fission in uranium-235 and plutonium^239 at ratings, in oxide fuel, of 1000 w.g- "' and 200 w.g-1 respectively. A typical fast reactor irradiation of (u0.7 *4), j)02 at 200 w.g is included; in one instance (Case 3) the composition after 100 days cooling is given. The different compositions observed in Table 1 for nom inally the same burn-up condition are attributable to a combination of factors. The differences between thermal and fast fission are not large (cases 2 and 4 ) and occur principally in the low yield section of the fission yie.ld curve, giving higher fast fission yields for species before bromine and in the trough region between silver and antimony. Differences between the fission of uranium-235 and plutonium-239 are more marked due to a significant d is placement in the light element peak. For plutonium-239 fission, the yields of elements before zirconium are depressed and the yields between ruthenium and antimony, and to some extent in the lanthanide region, are enhanced. Superimposed on these differences are effects due to neutron capture processes, irradiation time and decay after irradiation. Capture processes exert their greatest effect in thermal fission at high neutron fluxes. 34 FINDLAY These differences are evident in oases 1 and 2 in Table 1 where the fuel rating and hence the neutron flux differs by a factor of 100. Yields affected by neutron capture are those of the noble metals ruthenium, rhodium, palladium, trough elements cadmium to antimony, high yield elements tellurium , iodine, xenon, and caesium and the rare earths. From the chemical view point, changes among classes of elements such as the noble metals and rare earths have little significance as the yield of a subsequent member is enhanced at the expense of a previous member. Sim ilarly, low yield elements are of little importance. The most significant m odifications to fuel chemistry occur in the 133 and 135 decay chains where neutron capture results in an enhanced yield of stable xenon at the expense of caesium. The relative yields of fission products are determined to some degree by the irradiation time. In general, for irradiation times of practical interest, the early members of decay chains w ill have reached radioactive equilibrium and w ill represent only a small proportion of the total yield of a given element. Some intermediate cases exist where chain members have half lives of sim ilar duration to irradiation times and variations in elemental yields can be expected. These effects are demonstrated by cases 1 and 2 in Table 1. In case 1, the irradiation time is very long and the majority of radioactive species w ill be at equilibrium ; in case 2 , t h i s may not always be so. This is evident in the yield of strontium which is enhanced in Case 2 by the higher concentration of °9gr-50 day half life. Other chains with members of significant half-life involve yttrium , zircon ium, niobium, and molybdenum yields, where yields of the lighter elements are enhanced at the expense of the molybdenum. Other examples exist with other elements but these are of less importance chemically, although it is of note that the yield of iodine is affected by the saturation of I in short irradiations. Changes of a sim ilar nature w ill be introduced by radioactive decay and the effect of a 100 day decay on the yields of Case 2 is shown in Case 3. Changes of concentration are noted again in elements at the maximum yield of the light element peak and Case 3 appears as an intermediate situation between the extremes of Cases 1 and 2. It is apparent from these calculations that the compositional changes induced by irradiation conditions are relatively minor and the major differences are attributed to the differences in the fissile species considered. Although this allows general cases to be established, when a computer programme is available, it is preferable to perform the calculation using the exact conditions of the irradiation being simulated to obtain the precise compositions. 3. OXYGEN REQUIREMENTS OF IRRADIATED OXIDE FUEL MATERIAL The oxygen requirements of irradiated fuel have been calculated by several workers with the object of establishing the change in oxygen balance of a system as bum-up proceeds The exercise is repeated here for some of the cases given in Table 1 using the classifications of oxide forming and m etallic species described by other workers; some discussion of the intermediate situations is included. From the partial molar free energyД& (02) of the various fission product metal/oxide systems, it is apparent that the stability of the rare earth, alkaline earth and zirconium oxides is such that these elements w ill exist as oxides in equilibrium with any UOg or (U,Pu )02 + -xfuel of practical interest. The oxygen potential for (U,Pu)02 - x °an become quite low when fuel of low in itia l oxygen to metal ratio is employed. The oxygen potential IAEA-PL-463/2 35 o f ( U , P u )02 has been shown to be defined by the plutonium valency £ 6 ] ; a value of -170k cal mole-"' corresponding to a Pu valency of 3.2 is seldom exceeded in a practical fuel element situation. The phases adopted by the stable oxide forming elements are considered to be as follows: strontium is assumed to exist as SrO, barium is thought to form BaZrOj, since the existence of this phase has been demonstrated by microprobe examination of irradiated fuel m aterial [_7j„ The rare earth elements and the zirconium uncombined with barium are considered to exist in solid solution and to adopt the mean valency of the fuel m aterial. The oxygen potential and valency of the solid solution is assumed to be controlled, in the case of (U,Pu) 02, by the plutonium valency. The assumption can be justified experimentally by X-ray measurements at the stoichiom etric condition; the lattice parameters of irradiated (U,Pu )02 were found to be consistent with the rare earths existing in the four valent state [_5l * At lower oxygen potentials there are insufficient data available to decide what effect fission product species exert. The hypothesis of control of oxygen potential by plutonium valency is like ly to be modified in some situations such as high burn-up irradiations of lew in itia l plutonium con tent and low in itia l oxygen to metal ratio. Here the concentration;of rare earth elements at the end of irradiation w ill be comparable to that of the plutonium, probably representing a significantly different system from that of the unirradiated fuel. The effect of fission product additions on the oxygen potential in the fuel system have been considered using data from C e 02-UC >2 systems but no inform ation is yet available for zirconium and other lanthanides \ ji \ . For present purposes, the data for the in itial fuel material is assumed to be applicable. The classes of elements that are considered to have no oxygen requirements except at high oxygen potentials of little technological interest are the non-metals I, Te, Br, Se, etc., and the noble metals Ru, Rh, Pd, Ag. Elements making demands depending on the oxygen potential of the system are: Mo, Cs, Rb, Tc, Nb. Neglecting the role of the intermediate elements, the oxygen balances for the yields in Table 1 are calculated in Table 2 and are expressed as the number of excess oxygen atoms at 1O/о b u r n - u p p e r 100 in itia l heavy metal atoms, Uranium fission, both fast and thermal, results in a very small liberation of oxygen which is essentially independent of irradiation or cooling conditions. Plut onium fission results in the liberation of a significant amount of oxygen as is w ell known and experim entally demonstrated L5j. Uncertainties arise when the division of oxygen between the species with oxidising potentials sim ilar to the fuel itse lf is considered. In the case of uranium fission, the small amount of excess oxygen w ill probably be dissolved in the caesium phase or possibly oxidise the niobium and the oxygen potential of the fuel system is like ly to be held at a low level by these reactions. The case of plutonium is more complicated, because the oxygen potential of the fuel can vary widely depending on the in itia l oxygen to metal ratio and this w ill be reflected in the extent of oxidation of the intermediate elements. Oxygen is known to dissolve in caesium to an extent depending on the oxygen potential, although quantita tive thermodynamic data on the extent of solution are not yet available. As a first approximation, it may be reasonable to assume the formation of C sO q 2 until the plutonium valency of the system reaches 4.00. Thereafter caesium is probably oxidised to CS 2O. Niobium is also oxidised at an oxygen potential corresponding to a plutonium valency of 3.9 but the yield of niobium in plutonium fission is relatively small and insignificant. со а> TABLE 2. OXYGEN BALANCE CALCULATIONS FOR IRRADIATED FUEL MATERIAL AT 10% BURN-UP 1 2 3 4 5 6 C ase No. M etal Oxygen M etal Oxygen M etal O xygen M etal O xygen M etal O xygen M etal O xygen Oxide formers BaO (in BaZr03) 0. 84 0. 84 0, 78 0. 78 0, 68 0. 68 0 .8 1 0 .8 1 0 .6 5 0 .6 5 0 .6 5 ' 0 .6 5 ZrO 0. 84 1. 68 0. 78 1. 56 0, 68 1 .3 6 0 .8 1 1. 62 0. 65 1. 30 0 .6 5 1. 30 SrO - 0. 78 0. 78 1. 20 1. 20 1,00 1. 00 1.18 1.18 0. 37 0 .3 7 0. 39 0. 39 FINDLAY T otal 2 .4 6 3 .3 0 2. 76 3. 54 2. 36 3. 04 2. 80 3. 61 1. 65 2 .3 2 1. 69 2 .3 4 Solid solution Rare earth 5. 42 5 .4 2 5 .4 8 5 .3 6 4. 45 4. 56 Zr (balance) 2 .4 2 2. 38 2 .4 6 2. 29 1. 34 1. 37 U + Pu 90. 11 8 9 .9 6 89. 96 89. 98 9 0 .0 0 90. 02 T ota l 9 7 .9 5 1 9 5 .9 97. 76 195 .5 97. 90 195. 8 97. 63. 195. 3 95. 79 1 9 1 .6 96. 25 1 9 2 .5 Oxygen о CO + 0, 80 + 0 .9 6 + 1. 16 + 1. 09 + + 5. 16 Balance MO2>00 IAEA-PL-463/2 37 Talcing caesium and rubidium together at MCU the excess oxygen available is reduced from 5.16 atoms in Case 6 for thé typical fast reactor conditions to 4.75. This excess oxygen is then taken up by the fuel m aterial increasing the oxygen to metal ratio. This process w ill continue until the molybdenum is oxidised. The oxidation of molybdenum occurs at an oxygen to metal r a t i o o f 2.005 if the molybdenum is as metal or at 2.012 if, as is more likely, molybdenum is alloyed in a m etallic inclusion phase. If the formation of CS 2O i s a n d M0O2 is assumed at an 0:M of 2.016, then the system is completely buffered at that point and no excess oxygen remains. The effect of these assumed conditions on the mean oxygen to metal ratio of a ( U , P u ) 0 2 _ fuel is shown diagramatically in Pig. 1 for various initial oxygen to metal ratios. FIG. 1. Change of 0:M ratio o fU .P u oxide fuel as a function of bum-up. 4. OXYGEN REDISTRIBUTION AND OXY&EN POTENTIALS IN A TEMPERATURE &RADIBJT The redistribution of oxygen tinder the influence of the radial temperature gradient is believed to occur via the CO/CO 2 atmosphere in a fuel element which is assumed to be in thermodynamic equilibrium with a ll regions of the fuel [_63 . The hypothesis is supported by experimental evidence which has been reviewed by Holleck In stoichiometric fuel, the oxygen to metal ratio is essentially constant across the radius but deviations in the direction of either hypo or hyper stoichiom etric fuel result in án increased radial composition gradient. The rim of the fuel remains close to the stoichiom etric condition and the divergence from stoichiom etry increases towards the fuel centre. The radial 0/M ratio gradient for various fuel systems is described by Rand 6^ and can be used in, for example, fuel modelling calculations to determine the effect of 0/M ratio on such properties as thermal conductivity, and creep. Because of the interest in fuel-clad interactions, it is also appropriate to establish the effect of burn-up and radial redistributions on oxygen potentials. The principle point of interest is at the fuel s u r fa c e , since it is the oxygen potential at this point that controls fuel- clad and other chemical reactions from a thermodynamic viewpoint. The 38 IAEA-PL-463/2 FIG. 2. Change of oxygen potential with burn-up fot U0 7Puo-3 oxide, 1. 98 initial 0:M ratio. oxygen potential is obtained quite simply, since at a particular point it is defined by the CO/CO 2 atmosphere within the fuel pin and the temperature of that position. The CO/CO 2 atmosphere is obtained from the mean oxygen to metal ratio of the fuel at an assumed mean fuel temperature of 1500° K ; t h e variation of mean 0/M ratio with burn-up is established from Pig. 1. The change of oxygen potential with burn-up for (U,Pu )02 f u e l o f 1.98 i n i t i a l oxygen to metal ratio is plotted in Fig. 2 for temperatures of 1000°K, 1500°K and 2000°K respectively representing centre, mean, and fuel rim condition. Due to the fam iliar ’S ' shaped curve for the dependence of oxygen potential on plutonium and uranium valencies in (U,Pu) 02, t h e oxygen potential remains low throughout the fuel as burn-up proceeds and passes through a rapid transition as the stoichiom etric state is traversed. The im plications of these curves on the incidence of fuel-clad interaction has yet to be established and is like ly to be complex, because it is probable that kinetic as w ell as thermodynamic factors control the reactions. However, the figure implies that considerable variations in the oxygen potential of the fuel can occur depending on fuel and irradiation conditions. 5. CONCLUSION The chemistry of irradiated fuel m aterial is determined in the first instance by the fuel composition which can be calculated, sufficiently accurately for current purposes. An understanding of the subsequent changes in the oxygen balance with bum-up has been developed with lim ited experimental confirmation. Effects in uranium fission are small« Significant changes in oxygen balance are expected for plutonium fission. The mechanisms of oxygen redistribution in a fuel element environment have been identified and are supported experimentally. Areas of uncertainty remain concerning the detailed behaviour of some fission products with regard to their phases FINDLAY 39 formad and th eir oxidation ch a ra cteristics. The current state o f knowledge is insufficient for the occurrence and extent of fuel-clad interactions to be predicted by theoretical means. Acknowledgements I wish to thank Mr. F.T. Ewart of AERE, Harwell fo r his help in preparing this paper and Dr. M.H. Rand for helpful discussions. REFERENCES FAIRCLOTH, R .L ., HOPPER, M.J. UKASA Rep. AEHE-R 6242 (1970) L23 CLARKE, R.H., UTTING, R.E. U.K. Rep. RD/b/N 1737 (1 970) И EWART, F.T. KAHEKO, H. UKAEA Rep. to be published (1972) HOLLECK, H., KLEYKAMP, H. GFK Rep. HFK-1181 (1970) 15] EWART, F.T., DAVIES, J.H. J. Nucl. Mat. 41_ (1971 ) 143 MARKIN, T .L ., RAND, M.H. Thermodynamics o f nuclear materials IAEA ( 1967) 637 BRADBURY, B .T ., DEMANT, J .T ., MARTIN, P.M., POOLE, D.M. J. Nucl. Mat. 17. (1965) 227 IAEA-PL-463/3 REDISTRIBUTION OF URANIUM AND PLUTONIUM IN M IXED-OXIDE FU ELS DURING IRRADIATION* D.R. O'BOYLE, R.O. MEYER Argonne National Laboratory, Materials Science Division, Argonne, 111., United States of America Abstract REDISTRIBUTION OF URANIUM AND PLUTONIUM IN MIXED-OXIDE FUELS DURING IRRADIATION. When mixed-oxide fuels are irradiated under normal operating conditions, radial and axial redistribution of the oxygen, uranium, plutonium and fission products occurs, which significantly affects the performance of the fuel element. The paper develops the theory for vapour-transport and solid-state-diffusion controlled redistribution and summarizes recent experimental evidence of actinide redistribution. The experimental data, which cover a range of oxygen-to-metal ratios from 1.90 to 2.00, suggest that radial actinide redistri bution is controlled by a vapour-transport process and is closely related to central-void formation. Slightly hypostoichiom etric fuels, irradiated at moderate linear power ratings (360 to 530 W/cm) show an enhanced plutonium concentration near the central void that is typically 35 to 50% greater than the initial concentration. Markedly hypostoichiometric fuels show significantly less redistribution,- and unrestructured fuels show no actinide segregation. For burn-ups between 1 and 11 at. no*appreciable effect of burn-up on redistribution is observed. The effect of actinide redistribution on allowable power rating is briefly discussed and several analytical models are described for predicting the extent of uranium and plutonium redistribution as a function of tim e and temperature. I. INTRODUCTION The radial and longitudinal distribution of fissile atoms in a mixed- oxide fuel element is important in the design and operation of fuels for fast breeder reactors. Both the major fuel constituents (uranium, plutonium, and oxygen) and the volatile fission products are redistributed both radially and axially in a fuel element during irradiation. The extent of the redistri bution is strongly dependent on the fuel temperature, temperature gradient and the initial oxygen-to-metal (O/M) ratio. Because actinide redistribution affects numerous fuel properties, such as thermal conductivity, fuel plas ticity and melting temperature, it is an important consideration for under standing fuel-element behaviour and for fuel-element modelling [ 1, 2] . Actinide redistribution influences two aspects of reactor operation that directly affect reactor safety. Increases in plutonium concentration, which are commonly observed near the central void of slightly hypostoichio metric fuel, raise the centreline temperature for a given power rating and reduce the melting temperature of the mixed-oxide fuel. Thus .whenmaximum safe-operating conditions are established, actinide redistribution must be taken into account. Sha et al. [3] have shown that for moderate amounts * Work performed under the auspices of the US Atomic Energy Commission. 41 42 O’BOYLE and MEYER TEMPERATURE (°K) FIG. 1. Effect of temperature and stoichiometry on the ratio of plutonium-to-uranium in the gas phase over (U0-8PuO0-2)O2.x (after Lackey et al. [ 12]). I of actinide redistribution the allowable linear power rating must be reduced by about 15 to 25 W/cm. When a large amount of actinide redistribution occurs, which is typical of fuels operated with centreline melting, the allowable linear power rating may be reduced as much as 105 W/cm, but Doppler broadening is not significantly affected in either case. Early attempts [4, 5] to determine the radial plutonium distribution in oxide fuels by means of alpha autoradiography and radiochemical analysis did not detect any extensive plutonium redistribution. Significant plutonium redistribution in fuel elements operated with centreline melting was reported by Lauritzen et al. [6], who examined several mixed-oxide fuel elements irradiated in a thermal flux using both radiographic and core-drilling techniques. These results showed a general increase in plutonium concen tration near the molten fuel zone, but the accuracy of the experimental methods was limited. The first study that used electron-microprobe techniques to determine the radial uranium and plutonium distribution in non-molten mixed-oxide fuel irradiated in a, fast-neutron flux was reported by O'Boyle et al. [7,8]. This work established that in non-rr}'olten stoichio m etric fuel the plutonium concentration increased significantly near the central void. IAEA-PL-463/3 43 For slightly hypostoichiometric mixed-oxide fuel Rand and Markin [ 9] calculated that the vapour phase would be uranium rich because of the high vapour pressures of U02 and U03 and they suggested that preferential vaporization of these species would result in a plutonium-rich solid near the centreline of the fuel element. Subsequent measurements by Battles et al. [ 10] and Ohse [ 11] have confirmed the high vapour pressures of U02 and U03 . Lackey et al. [ 12] have calculated the ratio of plutonium- to-uranium in the vapour phase in equilibrium with (U0_8Pu02)O2_x for various stoichiometries and temperatures, and these values are shown in Fig. 1. These calculations show that for a fuel with an O/M ratio of 1. 98 or greater the vapour phase will contain a lower fraction of plutonium than the solid and transfer of uranium to the cooler regions should occur, with subsequent plutonium enrichment near the central void. Conversely, for mixed-oxide fuel with an initial O/M ratio of s 1. 94 the vapour over the solid will contain a higher fraction of plutonium than the solid and plutonium transport to the cooler region should occur. Plutonium enrichment near the central void of slightly hypostoichio metric mixed-oxide fuels has been observed by many investigators and the results are in agreement with the predictions based on the vapour- pressure measurements. In recent measurements on irradiated mixed- oxide fuels with O/M ratios < 1. 95 a decrease in the plutonium concentration near the central void has been observed [13, 14]. Considering the large number of variables in fuel-element fabrica tion and operation, it is not surprising that prior to 1970 the available data on radial actinide distribution were not adequate to decide upon an analytical model for actinide migration. In this paper we present the results of a series of systematic measurements on mixed-oxide fuel elements irradiated in a fast-neutron flux and review the published literature on actinide redistri bution. The main emphasis of the review is on work reported after 1969 and, in particular, the work that is useful for establishing the redistribution limits and mechanisms. The experimental results for non-molten mixed- oxide fuels are discussed in terms of a vapour-transport mechanism for actinide migration, and several computer codes based on vapour transport are briefly reviewed. I I . TH EO RY During a vapour-transport process various molecular species that contain uranium and plutonium move through pores or cracks in the steep radial temperature gradient; evaporation occurs from hot surfaces and condensation occurs on cooler surfaces. The vapour pressures of the most abundant uranium and plutonium species (U03, U 02, PuO, Pu02 and UO) vary both with composition and stoichiometry of the solid phase and with temperature. Because, in general, the ratio of Pu/U of the vapour is different from that of the solid and the vapour pressure over the fuel near the central void is many orders of magnitude higher than over the fuel in the cooler region, radial composition changes are expected to occur when temperatures are sufficiently high for the vapour-transport process to operate. The rate of migration of the actinide molecules is limited by diffusion through the fission- or cover-gas atmosphere in pores and cracks in the 44 O'BOYLE and MEYER fuel. Phenomenologically, the following flux equation can be written for diffusion by any mechanism in a temperature gradient: VT ( 1 ) where the subscript i identifies the diffusing species, D is the diffusion coefficient, с is the concentration, Q* is the heat of transport, к is the Boltzmann constant, and T is the absolute temperature. Equation (1) gives the flux of atoms within the solid or g a s phase, depending on the operative mechanism. ^ An average flux Oj)> due to vapour-phase diffusion within voids in the fuel can be obtained by assuming that the average flux is given by the net flux in the vapour phase multiplied by the volume fraction of porosity P: Under this assumption, the average vapour-phase flux vanishes for fully dense fuel material. The average flux varies strongly with radial position because of the temperature dependence of the vapour pressures and the diffusion coefficients of the various species. A continuity equation can be obtained for this flux by utilizing the divergence theorem, with the result = - V • That is, a net influx of m aterial into a given region of the fuel results in a concentration change and this change accrues to the concentration c- in the solid phase, where the prime denotes the solid. If we restrict our attention to the radial dimension, the concentration per unit volume in the solid, as a function of radial position r and time t, is (4) о w h ere dc.[¡dt is given by Eq. (3), and c|(r, 0) is the initial homogeneous concentration of the solid. Equations (1) to (3) can be evaluated for the radial flux component for the vapour-phase mechanism. The radial component of Eq. (1) is simply dc¡ CjPjQ: dT (5) 1 dr kT^ dr Assuming ideal gas behaviour, the concentration of each species in the vapour phase is IAEA-PL-463/3 45 where is the equilibrium partial vapour pressure. Equations (1) to (6) can be written for each vapour species over the mixed-oxide fuel, but the subscript i can be used to represent all the uranium-bearing species or all the plutonium-bearing species. The partial pressurept depends on the temperature at a given point and also on the fuel composition and stoichio metry at that point. If an ideal solution is assumed so that Raoult's Law ap p lies, P? N j (7 ) ■ сCU "Ьс H?u where p¡ is the equilibrium pressure over the pure metal oxide, and N¡ is the atom fraction in the solid. The term p? can be written as о _ I A H , Pj ,= const exp^- (8) where ЛН is the heat of vaporization. Combining Eqs (6) to (8) and differentiating yields dc¡ _ Эс, Эс, dT dr Эг 3T dr _ p°t 9Nj p° N, AH, dT kT Эг k 2T s dr W after neglecting unity compared with AH¡/kT. For gas-phase diffusion, the heat of transport Q* is identically zero [ 15] therefore, the second term in Eq. (5) vanishes and Eq. (5) becomes Dip[aNi I),Pi NjA!I dT i kT Эг. k^Y3 dr 1 ; All the material parameters in Eq. (10) can be evaluated. Since Nj is the fractional composition of the solid, the first term yields the flux in the vapour due to a concentration gradient in the solid. The second term, which depends on the temperature gradient, is equivalent to the expression used by Nichols [16] to calculate fuel restructuring. The diffusion coefficient can be obtained from kinetic theory [ 17] and is given by w here a is the collision cross-section, cf and Mf are, respectively, the concentration and molecular weight of the fission gas (or cover gas) within the vapour phase, and R is the gas constant. It is interesting to note the sim ilarity between Eq. (10) and the more general Eqs (1) and (5). Although Q* has vanished, a sim ilar term has 46 O’BOYLE and MEYER arisen that contains an energy ДН. Thus the vapour-flux equation has the same form as the solid-state-flux equation, which is given by Eq. (5), if all quantities are understood to apply to the solid phase. Taking the divergence in Eq. (3) in cylindrical co-ordinates and combining Eqs (2), (3), (4) and (10) yields the following expression: _d_ rPD iPi 3Nj Ni ДН; dT Ci(r,t) = - -T- dt + c ¡ (r, 0) (12) dr kT Эг kT2 dr This equation can be solved numerically using finite difference approxi mations for the derivatives and integrating stepwise by summing. For com pleteness, the solid-state flux can be incorporated into this expression, although at high temperátures the additional terms are not significant. It can be seen from Eqs (5) and (10) that the relative importance of vapour-phase and solid-state transport can be estimated by comparing the concentration X diffusion coefficient for the vapour with the concentration X diffusion coefficient for the solid. The results of this calculation are plotted in Fig. 2, including a porosity correction to account for available diffusion paths. Figure 2 shows that vapour transport is the dominant m ass-transport mechanism at temperatures above 1800°C. ' FIG.2. Relative vapour-phase and solid-state transport fluxes assessed by comparing the concentration x diffusion coefficient for the vapour-phase and solid-state mechanisms. Dashed lines show the effect of the available diffusion paths for 90% dense fuel. IAEA-PL-463/3 47 FIG.3. Calculated radial temperature distributions before and after restructuring for the top and middle sections of a typical mixed-oxide fuel element irradiated in a fast neutron flux. Equation (12) gives the concentration of uranium and plutonium separately in the solid, so that the calculated cj,u/ (c¿ + Cpu) ratio can be compared with measurements made by microprobe analysis. In addition, if fission products in solution in the fluorite lattice are neglected, the sum of the uranium and plutonium concentrations is related to the fuel density p so that early in life p(r,t) = (l/NA)[Muc¿(r,t) + Mpuc'pu(r,t)] (13) where NA is Avagadro's number, and M¡ is the molecular weight of the metal oxide. Equation (13) gives the radial density distribution as a function of time and, if it is combined with mass conservation by summing inward from the fuel surface, the central void radius is obtained as a function of time. The close relationship between actinide redistribution and fuel restructuring found experimentally [18] is thus supported by the analytical description of the process. The time dependence of Eq. (12) arises principally through its dependence on temperature, which, at a fixed radial position, changes with time. Figure 3 shows calculated [ 19] temperature distributions for the top and middle sections of a 34. 3-cm long fuel element at start-up and after the central void has developed. The sharp reduction in temperature in the columnar-grain region that accompanies central void formation causes the concentration changes, as described by Eq. (12), to progress much more slowly. Thus, during the initial operation of a slightly hypostoichiometric 48 O’BOYLE and MEYER FIG.4. Simplified flow diagram to calculate actinide redistribution in mixed-oxide fuel. mixed-oxide fuel element (O/M = 1. 98 to 2. 00) the vapour-phase transport mechanism is expected to result in rapid actinide redistribution since: (a) the initial fuel geometry and relatively low fuel density favour high temperatures and temperature gradients; (b) the low fuel density relative to the restructured fuel provides adequate paths for vapour transport; and (c) because the radial oxygen gradient is initially flat, the total vapour pressures and the U/Pu ratio in the vapour over the hottest region of the fuel will be higher than after oxygen migration to the cooler regions has o cc u rre d . Several computer codes have been developed [ 12, 18, 20] to predict actinide redistribution and fuel restructuring based on the vapour-transport mechanism. In each code the fuel is divided into a number of radial seg ments and the flux of atoms and the composition of the vapour phase is calculated at each location based on published equilibrium thermodynamic data [ 9-11]. Equation (12) gives the concentration change of each actinide species as a function of temperature and vapour pressure. Meyer et al. [ 18] have used this expression to calculate the time dependence of the radial distri bution of plutonium and a simplified flow diagram for the calculation is shown in Fig. 4. Adamson and Aitken [ 20] have suggested a method that describes a cyclic flow of m aterial wherein uranium and plutonium move down the temperature gradient in the vapour phase and then a reverse flow occurs in the solid. This reverse flow occurs down the concentration gradient until a steady-state, concentration profile is achieved. Lackey et al. [ 12] follow-the amount of pore migration that occurs by vapour trans port and proportion the amount of uranium and plutonium transported, based IAEA-PL-463/3 49 on the relative vapour pressures [9], to predict the actinide redistribution. None of the models has been developed in sufficient detail to be incorporated into the fuel-element-modelling codes [ 1, 2] . III. EXPERIMENTAL RESULTS AND DISCUSSION III. 1. Experimental method A number of mixed-oxide fuel elements irradiated in EBR-II have been analysed at Argonne National Laboratory (ANL) for radial and longitudinal actinide redistribution. Longitudinal and transverse cross-sections were cut from full-length fuel elements and were mounted in electrically con ducting metallographic mounts. A fuel specimen was typically 6 mm long and had a gamma activity of about 50 R/h at 23 cm. All specimens were examined with a MAC-450 microprobe, which has been described elsewhere [21] . Chemical concentration profiles across each specimen were determined by measuring the intensity of characteristic X-rays, which were produced by a 1-jum dia. electron beam, and by comparing them' with the intensities measured from pure U02 and Pu02 standards. Two spectrometers were used simultaneously to record the intensities of the uranium M„ and plutonium Mg X-ray lines. A number of corrections were applied to the X-ray intensity values to obtain weight per cent compositions of uranium and plutonium. These corrections include effects such as gamma-induced background, beam- current drift, system deadtime, atomic number, X-ray absorption and fluorescence, all of which were corrected by means of the MAGRAM computer code [22]. Data were collected along radii of transverse fuel cross-sections, as shown in Fig. 5. The X-ray intensity ratios were measured at 14 to 20 evenly spaced locations from the edge of the central-void to the fuel- cladding interface. Inclusions and pores were avoided so that the results are representative of the uranium and plutonium concentrations in solid solution in the fuel màtrix. For each specimen, data were obtained along three radii spaced 120° apart. The accuracy of the concentration measure ments is estimated at ±3% of the value measured; i.e. the uncertainty is approximately ±2 wt.% for uranium and ±1/2 wt.% for plutonium. FIG. 5. Typical transverse cross-section through an irradiatèd mixed-oxide fuel element showing the radial locations where microprobe data were taken (fuel element NUMEC C -ll, top section). TABLE I. SUMMARY OF PELLET-TYPE MIXED-OXIDE FUEL SPECIMENS'1 EXAMINED AT ANL FOR URANIUM AND PLUTONIUM REDISTRIBUTION ______ Distance from Local Local Initial Initial Final centre Fuel-element Initial Meas. Max. % Increase Void radius bottom of core burn-up linear power fuel O/M Fuel typeb fuel density line temp. identification Pu /(U+Pu) Pu/(U+Pu) Pu/(U+Pu) (ц т ) (cm ) (at.<7o) (W/cm) ratio (% TD) (•C) PNL 1-14 Middle 15.2 • 0 .8 5 338 1.974 MB.E 9 2 .5 0 .2 5 0.250 0 0. 0 1820 PNL 17-10 Top 3 3 .0 2 .7 272 1.965 MB.U . ' 9 3 .3 0 .2 5 0.249 0 0 .0 1480 Middle 17.5 3 .5 358 1.965 MB.U 9 3 :3 0 .2 5 • 0.247 -1 15 1790 Bottom 2 .0 2 .9 . 295 1.965 MB,U 9 3 .3 0 .2 5 0 .2 4 6 -2 0 .0 1420 ANL 007 MEYER andO'BOYLE Middle 1 6 .8 4 .7 512 1.98 CP,U 7 9 .0 0 .2 0 ‘ 0.295 48 1145 2210 ANL 012 Middle 1 6 .5 2 .9 525 1.98 CP,и 8 1 .1 0 .2 0 0.270 35 1015 2280 NUMEC C -6 Top 3 4 .5 7 .9 338 1.984 MB.E 9 1 .2 0 .2 0 0.248 24 430 1710 Middle 1 7 .3 10.5 446 1.980 MM, E 9 1 .3 0 .2 0 0.247 24 930 1810 Bottom 1 .3 8 .4 361 1.980 MM, E 9 1 .3 0 .2 0 0.242 21 435 1750 NUMEC C - l l Top 3 0 .5 8 .7 364 2.000 MB,E 9 0 .6 0 .2 0 0.292 46 585 1800 Middle 1 7 .8 1 0 .8 456 2 .0 0 0 MM, E 89 .0 0 .2 0 0.293 47 865 1970 Bottom 1 .0 8.7 367 2.000 MM, E 89.0 0 .2 0 0.292 46 510 1830 NUMEC C -15 Top 3 0 .0 8.5 341 1.998 MB, E 85.0 . 0 .2 0 0.330 65 720 1760 Middle 1 7 .3 10.5 420 1.993 CP,E 83.9 0 .2 0 0.297 48 960 1910 Bottom 1 .3 8.5 341 1.993 CP,E ' 8 3 .9 0 .2 0 0.292 46 560 1830 2 All fuel elements were irradiated in EBR-II. Ь CP = Coprecipitated, MM = Master Mix, MB = Mechanically Blended, E = Encapsulated, and U = Unencapsulated. IAEA-PL-463/3 51 DISTANCE FROM FUEL CENTRELINE, mm (a) FUEL ELEMENT NUMEC C-15. DISTANCE FROM FUEL CENTRELINE, mm О______05______10______15______ZQ 1 "1 " 1 1 3 0 - 25 __ TOP SECTION — n 0 Q • O g о о 20 ÿ j 3 0 ■ 2 3 25 MIDDLE SECTION о z i-, o i. 0 э 0 0 1 о о о— — о— о— о— о— о 0 5 о H 20 - -1 О О Û. , '^V O ID POSITION 3 0 - - BOTTOM SECTION 25 - - Л Г» Q 0 о 0 0 20 — lili. 1 ,1) 0 0.20 0.40 0.60 0.80 1.00 FRACTION OF FUEL RADIUS (b) FUEL ELEMENT PNL 17-10. FIG. 6. Radial distribution of plutonium in two pellet-type fuel elements irradiated in EBR-II. Fuel element NUMEC C-15 (a) was irradiated at a maximum linear power rating of 420 W/cm and developed a large central void and extensive actinide redistribution. Fuel element PNL 17-10 (b) was irradiated at a maximum power rating of 358 W/cm and no significant central void formed nor did actinide redistribution occur. The fuel type, irradiation conditions and maximum plutonium concentrations for both elements are listed in Table X. 52 O’BOYLE and MEYER III. 2. In-reactor results III.2.1. Radial actinide redistribution Seven pellet-type fuel elements irradiated in EBR-II at moderate power ratings were analysed in detail for radial actinide redistribution and charac teristic values that describe the fuel elements, calculated centreline temperatures and measured results are listed in Table I. The fuel elements covered a range of burn-ups from 1 to 11 at.% and were operated at power levels from 260 to 53 0 W/cm. The initial oxygen-to-metal ratio (O/M), fuel type, initial fuel pellet density and initial plutonium fraction listed in. Table I are the values specified during the fabrication of these elements. Initial fuel-pellet densities varied from about 79 to 93% TD and the fuels contained either 20 or 25 wt.% plutonia. Fifteen cross-sections from these fuels were examined with the shielded electron microprobe, and a data set from the top, middle, and bottom of a fuel element that showed extensive redistribution of plutonium (NUMEC C-15) and from an element that had negligible redistribution (PNL 17-10) is presented in Fig. 6. The straight line segments drawn in the figure are least-squares averages of the data points nearest the central void from the three data sets for that section. Local values of burn-up and linear power, corresponding to each location examined, were calculated from the average values for each fuel element. These calculations were made using the fission-rate profile for the first six rows of EBR-II [ 23] . Figure 7 shows a typical fission-rate profile and the values calculated for the top, middle, and bottom sections of fuel . element NUMEC C -ll. The measured maximum value of the weight ratio of plutonium-to- uranium + plutonium is listed in Table 1, and the per-cent increase value compares the measured maximum with the initial value. To obtain a specific DISTANCE, cm 15 20 25 T------1------"I------Г • INUMEC С-I I - 10.8 at.% < о: ''^ s r 456 w/cm Ld BOTTOM / TOP < SECTION / SECTION СГ I / - t Ld 1.00 -c m £ MIDDLE SECTION” i.7 ot.°/o < 0.90 О о 3 67 W/cm AVERAGE BURN-UP 9.7at.% AVERAGE LINEAR POWER 411 W/cm __]_____i_____I_____i_____I_____i_____L 2 4 6 8 10 12 DISTANCE FROM BOTTOM OF CORE, in. FIG. 7. Locations at the top, middle and bottom specimens from a typical fuel element, superimposed on the axial fission rate distribution in EBR-II. IAEA-PL-463/3 53 FRACTION OF FUEL RADIUS FIG.8. Typical plutonium-concentration profile in a slightly hypostoichiometric mixed-oxide fuel element irradiated at a linear power rating >360 W/cm to a burn-up >5 at,°jo. The temperatures given are end-of-life temperatures calculated for the fuel after restructuring has occurred. The fuel temperatures before restructuring are much higher and are close to those shown for the middle section of NUMEC С -11 in Fig. 3, value that characterizes the maximum actinide redistribution at the edge of the central void, a least-squares method was used. For the data points nearest the central void measured weight ratios of Pu/(U+Pu) were averaged for all data sets and extrapolated to the central-void position; these are listed in Table I as the measured maximum values. The standard deviation for each Pu/(U+Pu) result listed in Table I was found to be about 4% of the value tabulated. The general features of radial plutonium redistribution in stoichio metric and slightly hypostoichiometric fuels are shown in Fig. 8. In each case the maximum plutonium concentration observed near the central void was accompanied by a minimum plutonium concentration near the cool end of the columnar-grain zone. The flat portion of the curve indicates that the temperatures were too low (<1500°C) to permit any long-range migration of the actinides.. This conclusion is supported by the large residual in homogeneity observed in the Master M ix1 portion of elements NUMEC C-6 and C -ll even after 11 at.% burn-up. The typical radial plutonium distribution after irradiation'does not indicate whether the process is exclusively radial or includes some axial contribution, nor is it possible to decide from the shape of the curve whether the process is controlled by a solid-state diffusion or by a vapour-transport mechanism. The typical distribution shown in Fig. 8 would result from a divergent mass flux, regardless of which transport mechanism was responsible for the flux. A close relation between linear power, central void size and actinide redistribution is apparent, as shown in Table I. Fuel element PNL 1-14 operated with a peak linear power of 33 8 W/cm and did not form a central void nor did any measurable actinide redistribution occur. The middle 1 Master Mix [24] is a mechanically blended mixture of U02 and coprecipitated U02-50 wt.°¡o Pu02. 54 O'BOYLE and MEYER section of element PNL 17-20 that was irradiated at 358 W/cm (centreline temperature 1790°C) developed an irregual opening only s25 цш a c r o s s , which is the embryo of a central void, and radial redistribution of the actinides was insignificant. The other fuel elements listed in Table I formed large central voids even though some of the fuel sections operated at linear power ratings as low as 33 8 W/cm. In each case an increase in the plu tonium concentration near the central void was measured. Variations in the initial O/M ratio and fuel density as well as uncertainties in linear power values are believed responsible for the differences in fuel behaviour at linear power ratings close to 360 W/cm. As the central void is formed by the migration of the initial porosity up the temperature gradient, a large quantity of material is transported down the temperature gradient by means of the vapour-transport mechanism. When the maximum fuel temperature decreases because of geometry and densification effects, the vapour-transport process ceases to be active and, assuming that actinide redistribution occurs predominantly by this mechanism, the final actinide distribution is then relatively insensitive to burn-up. A 300°C temperature decrease, for example, reduces the total vapour pressure and hence the diffusion flux by two orders of magnitude [ 10] . The tempera tures listed in Table I are the centreline or central-void temperatures calculated from the tabulated values of linear power and central-void radius and are, therefore, the temperatures after restructuring of the fuel elem en t. It has been suggested by Bober et al. [ 25, 26] that volume diffusion in a temperature gradient is a long-term process wherein progressive redistri bution occurs after the vapour-transport process has ceased to be effective. The experimental evidence summarized in Table I does not support this theory. The high burn-up NUMEC C-6 fuel element with two to three times greater burn-up than the ANL fuel elements of the same stoichiometry did not show greater redistribution of the actinides. Also, thè two PNL fuel elements listed in the table experienced temperatures and temperature gradients sim ilar to some of the restructured fuels, while accumulating burn-ups to 3. 5 at.%, but in both fuel elements, where vapour transport did not occur, no actinide redistribution was detected. Additional evidence that solid-state diffusion does not contribute to actinide segregation is found in mixed-oxide irradiations reported in Ref. [ 27] (fuel pin No. E1H-30) and by Conte et al. [ 13] (fuel pin S-21). In each case the fuel elements had initial O/M ratios of 1. 98 or greater and were irradiated under conditions that normally result in substantial actinide segregation; i.e. at peak linear power ratings greater than 420 W/cm and burn-ups greater than 0.6 at.%. However, neither fuel element showed any significant actinide redistribution. The common characteristic of both fuel elements was that the pellet densities were >96. 5% and thus only limited pore migration by vapour-transport occurred during irradiation. If the solid-state diffusion mechanism played any significant role in actinide redistribution, substantial redistribution of uranium and plutonium would have occurred in both fuel elements. The results of surface-relaxation measurements by Maiya [28], which were designed to separate the effects of the various transport mecha nisms, showed that at temperatures above 1660°C vapour transport was responsible for the surface shape changes. These experiments also demon strated that solid-state mechanisms are not effective at high temperatures in producing long-range mass transport in U 02. Thus the experimental IAEA-PL-463/3 55 evidence indicates that over the normal range of operating conditions, actinide redistribution by vapour transport is dominant. Several effects occur early in the life of an operating fuel element that slow the rate of material transport via the vapour-transport process. Oxygen redistributes from the high- to the low-temperature region in hypo stoichiometric mixed-oxide fuels [9], thus lowering the O/M ratio at the fuel centreline, which affects the segregation process in two ways. First, a fuel composition exists between the O/M ratios of 1.94 and 1.97 where vaporization of uranium and plutonium species is congruent and segregation between the solid and vapour phase does not occur. Secondly, as the O/M ratio of the fuel near the central void is reduced, the total vapour pressure is decreased sharply. For example, if the O/M ratio decreases from 2. 00 to 1. 94, the total vapour pressure is lowered by an order of magnitude [ 10] thus reducing the rate of vapour transport by about an order of magnitude. The establishment of a concentration gradient also slows the segregation process. As fuel near the central void becomes enriched in plutonium, the vapour species becomes enriched in plutonium with the result that less uranium vapour is transported down the temperature gradient. It is apparent from Fig. 8 that true steady state has not been achieved at temperatures lower than -1700°C. Concentration changes that occur at radii greater than 0. 5, however slowly, will necessarily cause concen tration changes in the higher temperature regions, which prevent the use of equilibrium conditions to predict the concentration distribution. Therefore, even though temperature-gradient diffusion and back diffusion are in opposite directions in the high-temperature region of the fuel, the actinide concen tration distributions observed in fuel elements are not steady-state Sor et distributions. Rather, they are distributions that occur during the early high-temperature portion of fuel life and that have been quenched in place as restructuring lowers the radial temperature profile. III. 2. 2. Actinide redistribution in molten fuel Some of the early ANL fuel-element irradiations were designed to test the performance limits of vibration-compacted stoichiometric fuel at high power ratings. Results of the examination of these fuel elements are summarized in Table II. Because of the high power ratings — peak values in excess of 650 W/cm — all the elements operated with molten fuel during their life. Fuel operation and fabrication characteristics were sim ilar for all four elements, but the measured results varied from a small increase in plutonium content near the central void (+20% for SOV-3) to two cases where large increases (+115%) occurred. Some axial relocation of the molten fuel was also observed in these elements [ 29], so it is not sur prising that the maximum plutonium concentrations listed in Table II varied from element to element. It is clear from these measurements that increases in plutonium concentration in excess of 100% can occur in mixed-oxide fuel operated in the molten state. During cooling of mixed-oxide fuel from the liquid state the first solid to form will have a different composition from that of the liquid.' For moderately hypostoichiometric fuel the phase diagram [30] indicates that the first solid to freeze will be uranium-rich, leaving a plutonium-rich liquid. Therefore, the central region of a fuel element, which is the last to freeze on cooling, will be rich in plutonium. The exact shape of the сл Oî TABLE II. SUMMARY OF VIРАС MIXED-OXIDE FUEL SPECIMENS3 EXAMINED FOR URANIUM AND PLUTONIUM REDISTRIBUTION Local : •. Local Initial In itial Fuel-element In itial Meas. Max. °]o Increase burn-up.. linear power fuel O/M fuel density identification Pu/(U+Pu) Pu/(U+Pu) Pu/(U+Pu) ( a t. °jo) ■ (W/cm) ■ ratio (°}o TD) SOV-1 0 .2 0 Middle . 5 .0 - . 682 . '2 . 0 0 ■ • 7 9 .9 0 .2 8 40 MEYER andO'BOYLE Middle 5 .0 - 682 2 .0 0 , * 7 9 .9 0 .2 0 0.27- 35 Bottom 4 .3 ‘ 587 . • 2 .00 - 7 9 .9 ’ 0 .2 0 0. 37 85 ■ SOV-3 Middle ■ 3 .7 699 " 2 .0 0 - 8 3 .1 0 .2 0 , 0 .2 4 20 SOV-7 , - . ■ ' ~ 0 ; 0 : . Top 'r ' 2 .9 ■ . ' 568 8 5 .1 - 0 .2 0 0 .2 8 ' . ^ 40 HOV-15 " ' ' Middle ■ ‘ - 3 .5 699 2.00" 7 9 .8 0 .2 0 " 0 .4 3 . 115 Bottom ■3.0 /600 . .. - 2. 00 7 9 .8 0 ,2 0 0 .4 3 115 a All fuel elements were irradiated in EBR-II with some fuel melting. IAEA-PL-463/3 57 radial concentration profile will depend on the cooling rate and the degree of mixing that results from diffusion and thermal convection in the liquid [ 31] . Craig et al. [32] reported the radial plutonium concentration profiles in a series of mixed-oxide elements that operated with centre melting for 85 h. In each fuel element, with O/M ratios of 1. 94, 1. 97 and 2. 00, the maximum plutonium concentration was observed in the molten fuel region and the fuel with the lowest O/M ratio showed the greatest increase in plutonium concentration in the molten region (+28%). Chasanov [33] calcu lated the concentration of the solute in the liquid and in the solid phases for several values óf the distribution coefficient К and concluded that for (Uo.sPUo. 2)02, К =0. 7, based on the phase diagram reported in Ref. [ 34] . Thus for O/M ratios from 1. 90 to 2. 00 the central region of a mixed-oxide fuel element operated with molten fuel will be enriched in plutonium compared with the cooler, region. III. 2.3. Observations of vapour transport Several direct observations have been reported that support preferential evaporation of uranium-bearing species as the controlling mechanism for actinide redistribution in nearly stoichiometric mixed-oxide fuel. Lackey et al: [35] cite evidence of vapour transport from observations of sphere- packed fuels irradiated in ETR to 0. 7 at.% burn-up. In one element that contained (U0 15>(~,2 oo condensed material was found in the form of dendrites on the hot side of microspheres in a partially restructured region FIG.9. Dendrites of mixed-oxide fuel deposited on the hotter surface of microspheres located at the transition region from columnar to equiaxed grains.' Microprobe analysis showed the deposited fuel to be rich in uranium compared with the microspheres and the adjacent columnar grains (after Lackey èt al. [12]). 58 O'BOYLE and MEYER Л % 2 m LONGITUDINAL SECTION FIG. 10. Location of microprobe traces taken on dished-end-pellet interface region of a fuel element irradiated in EBR-II to 10.1 at.^o burn-up at a peak linear power rating of 427 W/cm. DISTANCE , microns FIG. 11. Plutonium distribution along the two vertical traces indicated in Fig. 10. The trace through the cooler region (location 5) showed a decrease in the plutonium concentration, and a corresponding increase in the uranium concentration, on moving toward the pellet interface. of the fuel, as shown in Fig. 9. Microprobe analysis showed that these dendrites had a higher uranium concentration than the bulk fuel and it was concluded that a vaporization-condensation mechanism is the primary restructuring process in fuels at temperatures above 1700°C. In recent work at ANL [36] the pellet-interface region of a mixed- oxide fuel element (NUMEC C -ll) was examined after irradiation to a burn up of 10. 1 at.% at 427 W/cm. The fuel element had an initial O/M ratio of 2. 00 and was fabricated from dished-end pellets such that vapour trans port through the 127-/um interface gap [24] was possible. Both radial and longitudinal concentration profiles were obtained along this interface region at the locations shown in Fig. 10. IAEA-PL-463/3 59 The results of the interface-gap examination along the two vertical traces are shown in Fig. 11. The minimum plutonium concentration at the gap interface was about 10 wt.% below the normal minimum (trace 5), whereas no longitudinal actinide variation was observed nearer the central void (trace 6). The decrease in plutonium concentration and corresponding increase in uranium concentration is due to preferential evaporation of uranium-rich species from the fuel near the central void and condensation of the uranium-rich vapour in the cooler portion of the pellet-interface gap. Bober et al. [25] investigated the effect of initial fuel stoichiometry- on uranium and plutonium redistribution in a series of out-of-reactor temperature-gradient anneals of mixed-oxide fuel that contained 15 mol.% Pu02 . Fuel specimens that had initial O/M ratios of 1. 96, 1. 97 and 2. 00 were annealed in a gradient furnace for 1 h in sealed tungsten capsules at maximum temperatures up to 2400°C. The specimens were fabricated . with an axial bore-hole at the hot end to provide a closed vapour space so that redistribution could occur by evaporation from the hotter surfaces and condensation on the cooler surfaces of the hole. Microprobe analysis of the m aterial in the bore-hole showed that in the fuel with an initial O/M ratio of 2. 00 (a) the condensate in the cooler region was uranium-rich compared with the average pellet composition, and (b) the bore-hole was completely filled with evaporated m aterial after 1 h, which indicates that mass transport by evaporation-condensation was extremely rapid. In contrast to this behaviour, the mixed-oxide fuel with an initial O/M ratio of 1. 96 had relatively little m aterial in the central cavity after the anneal and the m aterial in the cooler region of the cavity was rich in plutonium. These observations are consistent with a vapour-transport mechanism because for stoichiometric fuel the total vapour pressure (and hence mass transport rate) is much higher than for hypostoichiometric fuel and the vapour phase is richer in uranium than the solid. III. 2.4. Effect of oxygen-to-metal ratio on actinide redistribution All transport mechanisms in mixed-oxide fuel are strongly influenced by the initial fuel stoichiometry because of its effect on lattice defects and the equilibrium vapour pressures. If a vapour-transport process controls actinide redistribution, nearly stoichiometric mixed-oxide fuel will be enriched in plutonium near the central void and markedly hypostoichiometric fuel will be enriched in uranium. Recently Conte et al. [ 13] reported measurements on several mixed-oxide fuel elements that had initial O/M ratios of ~1.90 and 2.00. These elements contained synthetic fission- product additions and were irradiated in the EL 3 reactor to about 0. 6 at.% burn-up. Following irradiation, each element was sectioned and the radial distribution of plutonium was measured by an ingenious X-ray fluorescence technique [ 37] . Based on these measurements, the maximum change in plutonium concentration at the central void of two elements (S-23 and S-24) was calcu lated and these data, together with data from Table I on the maximum plu tonium concentration and some recent ANL measurements [ 14] on fuel element NUMEC C-19 are plotted in Fig. 12. All data are for (U0t 8Pu02)O2_x fuel irradiated at sufficiently high linear power ratings to form a central void but below the power rating for centreline melting. The error bars 60 O’BOYLE and MEYER FIG. 12. Effect of initial fuel stoichiometry on the maximum radial change in plutonium concentration for (U0>8Pu0-2)O2_x . All fuel elements were irradiated at linear power ratings above 360 W/cm but below that to cause centreline melting. The ANL data are from Table I and Ref. [ 2]. The CEN results were calculated from data reported in Ref. [ 5]. are estimates of the uncertainties in the initial O/M ratios (±0. 01) and the maximum measured plutonium concentrations based on the number of measurements that were made on each fuel element. Figure 12 shows that for hypostoichiometric mixed-oxide fuel with an initial O/M ratio >1. 96 the maximum plutonium concentration increases near the central void, whereas for fuels with an initial O/M <1. 96 a decrease in plutonium concentration occurs near the central void. The effect of initial O/M ratio on the maximum plutonium concentration also agrees with data reported in Ref. [38], which shows that for slightly hypostoichiometric fuels the maximum radial plutonium concentration decreases with a decrease in plutonium valence. These results confirm that actinide redistribution during irradiation of mixed-oxide fuels occurs predominantly by a vapour- transport mechanism, and that the initial fuel stoichiometry for congruent vaporization in-reactor is 1. 96 ± 0. 01. III. 2. 5. Longitudinal actinide redistribution In fuel elements that undergo extensive restructuring actinide migration can occur by a vapour-transport process that operates axially through the central void. If a temperature gradient exists along the wall of the central void, molecules will evaporate from the hotter surface and condense on the cooler surface of the void wall. The data listed in Table I for NUMEC fuel elements C-6 and C -ll show remarkable consistency between the measured maximum plutonium concentrations at the top, middle and bottom sections, even though the temperatures, burn-ups and linear power ratings were substantially different for these sections. IAEA-PL-463/3 61 To test for axial segregation, a mass-balance was performed on the data for the three NUMEC elements listed in Table I. The mass of plutonium per unit length of fuel within a region bounded by radii ^ and rg is given by 12 P u (m a ss) = J Pu(wt.%) 2irprdr (14) ti where Pu(wt.%) is the measured concentration of plutonium, which varies with radius, and p is the density of the mixed-oxide fuel. Since the m icro probe data were obtained on fully dense locations in the fuel (~l-/um dia. ), p is the theoretical density of the fuel. The mass of plutonium originally in a volume of fuel bounded by ^ and r2 is h P u (m a ss) =J (17.9) 27rprdr (15) ii where 17.9 is the wt.% of plutonium in (U0 8Pu0 2)02. If the ratio of Eq. (14) to Eq. (15) is greater than unity, a relative gain in the amount of plutonium in that region is indicated. Since the microprobe data were obtained on fully dense areas of the fuel, absolute values of the net gain or loss of material from the radial sections cannot be determined. A plutonium-mass ratio greater than unity can, therefore, be due to a net gain of plutonium, a net loss of uranium, or a combination of both. Mass ratios were evaluated from the microprobe data using suitable sums to replace the integrals in Eqs (14) and (15). The region over which the sums were performed was bounded by the central void and the cool end of the columnar grains, thus eliminating the uncertainties from measure ments in the colder regions where no significant actinide migration occurred. DISTANCE, cm DISTANCE FROM BOTTOM OF FUEL COLUMN, inches FIG. 13. Ratio of the total mass of plutonium in a radial section after irradiation compared with the initial total mass. Data are for three different axial locations in fuel element NUMEC C-15, which was irradiated to 10.5 at.^o burn-up at a peak linear power rating of 420 W/cm. 62 O’BOYLE and MEYER The results of this analysis for fuel elements NUMEC C-6 and C -ll did not indicate any significant axial redistribution. However, the results for NUMEC C-15, which are shown in Fig. 13, indicate a relative gain of plu tonium in the top section of the fuel element. Inasmuch as the top section of this element had the highest plutonium concentration of any non-melted fuel element examined, it is concluded that this section experienced some actinide segregation by axial transport, which did not occur in any of the other fuel elements. The high plutonium concentration in this section is unique, and the reasons for the increase are not understood. III. 3. Effect of actinide redistribution on allowable power rating Radial redistribution of uranium and plutonium in mixed-oxide fuel has an effect on two considerations in fast-reactor core design: allowable power rating and Doppler broadening. As a result of radial redistribution, the plutonium concentration near the fuel-element centreline changes with time, and the radial power generation profile is skewed accordingly. To determine the effect of redistribution on the maximum allowable power rating, calculations were performed by Sha, Huebotter and Lo [3] based on two 'limiting' cases2 of radial redistribution, assuming they occurred in a fuel element with 6.4-m m OD cladding, 3 80-jum-wall, thickness, and a fuel smeared density of 85% TD. The heat-transfer analysis used the VOID [39] computer program to locate the columnar, equiaxed and undisturbed fuel regions after restructuring. The radial heat-generation profile was determined by S8 transport theory [40] and was found to be closely proportional to the radial plutonium-mass profile. With the calcu lated radial heat-generation profile the THTB [41] computer program was used to calculate steady-state and transient temperature distributions. Figure 14 shows the results of these calculations as a function of the contact heat-transfer coefficient, which was varied from 0.3 to 1.2 W/cm2 • °C. Maximum allowable linear power rating is greatly assisted by the fuel restructuring process. The 130 to 160 W/cm benefit shown in Fig. 14 is, in roughly equal proportions, due to the geometric benefit from the central void and the improvement in thermal conductivity that results from higher fuel density after restructuring. Uncertainty in the (U,Pu)02 solidus temperature and its dependence on burn-up, stoichiometry and U/Pu ratio create a range of values that are taken as the design-limit temperature. This range is 2650 to 2800°C, which is worth 15 to 50 W/cm in allowable power rating. The penalty in allowable power rating for a moderate amount of redistribution (Case A) is about 15 to 25 W/cm, depending on the contact heat-transfer coefficient and the temperature that is assigned as the design limit. For a large amount of plutonium redistribution characteristic of centreline melting (Case B), the penalty ranges from 60 to 105 W/cm. The details of these calculations and the effect of actinide redistribution on Doppler broadening are given in Ref. [ 36] . 2 The two cases do not represent upper limits but rather 'typical' and 'high' degrees of redistribution, which afford a basis for interpolation. IAEA-PL-463/3 63 COEFFICIENT, W/(cm2 °C) 20 0.3 0 5 0.7 0.9 "I I I I г ------MELTING TEMP, 2800°C ------MELTING TEMP, 2650°C CDafter RESTRUCTURE, NO Pu-redistribution ([_® AFTER RESTRUCTURE, Ф ~ WITH Pu-REDISTRIBUTION CASE A — ® ОС. a. hi Ui 5 5 О о CL û. cc û: < < UJ Ui z 4 0 0 _J 0 BEFORE RESTRUCTURE, NO CENTRAL VOID 0 AFTER RESTRUCTURE, WITH Pu-REDISTRIBUTION CASE В I I I J- .1. 1 1 I J _ _L T 500 1000 1500 2000 CONTACT HEAT TRANSFER COEFFICIENT, BtU /(h f t 2 °F) FIG. 14. Effects of plutonium redistribution, fuel restructuring and contact heat-transfer coefficient on allowable power rating of mixed-oxide fuel elements. Case A corresponds to a moderate increase in plutonium concentration near the central void, characteristic of fuel elements irradiated in the 360-530 W/cm range. Case В corresponds to a large increase in plutonium concentration,near the central void, characteristic of fuels operated with centreline melting. IV. CONCLUSIONS Measurements of radial actinide distributions in mixed-oxide fuel after irradiation at linear power ratings greater than 360 W/cm have shown that for initial O/M ratios >1. 96 the plutonium concentration increases near the central void. In nearly stoichiometric restructured fuels plutonium concentrations near the central void are typically 35 to 50% higher than the original value, which reduces the maximum allowable power rating by 15 to 25 W/cm. For initial O/M ratios <1. 96 the uranium concentration increases near the central void because of preferential transport of plu tonium-rich vapour to the cooler regions of the fuel. In fuel elements operated with centreline melting the plutonium concentrion is higher near the fuel centreline for all oxygen-to-metal ratios from 1. 90 to 2. 00. Data on mixed-oxide fuels with burn-ups from about 1 to 11 at.% are reported, but no strong dependence of actinide segregation on burn-up was observed. Redistribution of uranium and plutonium is closely tied to central-void formation and for peak linear power ratings less than about 360 W/cm, 64 O'BOYLE and MEYER no central void is formed and no actinide segregation occurs. Thus radial actinide redistribution can -be avoided in mixed-oxide fuel elements by (a) choosing a fuel having an initial oxygen-to-metal ratio near 1. 96; (b) selecting a fuel with a high theoretical density (density >96%) so that the path for vapour transport is minimized; or (c) operating at a low fuel centreline temperature, which corresponds to a linear power rating <360 W/cm. - . A calculation of the fluxes due to vapour-transport and solid-state diffusion shows that at all temperatures above 1800°C vapour transport, which occurs early in life, is the dominant mass-transport mechanism. Direct evidence that supports a preferential vapour-transport mechanism was found in several stoichiometric fuel elements, where an enhanced uranium concentration occurred in the cool portion of the fuel near the equiaxed grain region. No indication of a solid-state-diffusion contribution to actinide segregation was seen in several fuel- elements where it might have been expected. A relative mass-balance analysis indicated that axial redistribution of the actinides occurred in one of the fuel elements examined at ANL. Flux equations for actinide transport in the vapour phase were presented and several computer codes, based on a vapour-transport mechanism, were mentioned. All these results are consistent with a vapour-transport mechanism in which uranium or plutonium is preferentially evaporated from the hot central region of the fuel and transported, by means of porosity or cracks, down the temperature gradient. The redistribution rate is greatly reduced when restructuring occurs and central-void formation lowers the centreline temperature. ACKNOWLEDGEMENTS Thè authors wish to express their sincere appreciation to B.R.T. F r o s t for guidance during the course of this work; to L. A. Neimark, J.D.B. Lambert and W.F. Murphy for providing information concerning the fuel elements and for their co-operation in the postirradiation analysis; to L.A. Pember for providing the PNL fuel specimens; to the ANL Irradiation Services Group for preparing the irradiated specimens; and to E.M. B u tle r for performing the electron-microprobe analyses. REFERENCES [ 1] JANKUS, V .Z ., WEEKS, R. W ., LIFE-I, a FORTRAN-IV Computer Code for the Prediction of Fast - reactor Fuel-element Behavior, Rep. ANL-7736 (1970). [2] JANKUS, V .Z ., WEEKS, R. W ., LIFE-II — a computer analysis of fast-reactor fuel-element behaviour as a function of reactor operating history, Nucl. Eng. Design 18 (1972) 83; '■ [ 3] SHA, W .T., HUEBOTTER, P.R., LO,.R.K., The effect of plutonium migration on allowable power ' - rating and Doppler-broadening, Trans. Amer. Nucl. Soc. 14 (1971) 183. [4} BATES, J.L ., CHRISTENSEN, J.A ., ROAK, W .E., Fission products and plutonium migrate in uranium- dioxide fuel, Nucleonics 2£ (1962) 88. [5] RUIZ, C .P., GERHART, J.M ., ALTER, H. W., EPSTEIN,-L. F ., The transport of plutonium in Pu02/U02 fuels: I, Trans. Amer. Nucl. Soc. 6 (1963) 82. [6] LAURITZEN, T. A ., NOVAK, P.E., DAVIES, J.H ., Experimental Studies of Plutonium and Fission Product Redistribution in Fast Reactor Fuels, Rep. GEAP-4466 (1966). !I AEA-PL-463/3 65 [7] O'BOYLE, D .R ., ’’Electron microprobe analysis of irradiated U02-20 w/o Pu02", Reactor Development' Program Progress Report, Rep. ANL-7382 (Sept. 1967) 76. ' ’ [8] O'BOYLE, D.R., BROWN, F.L ’, SANECKI, J.E ., Solid fission product'behavior in uranium-plutonium oxide fuel irradiated in a fast neutron flux, J. Nucl. Mater. 29_(1969) 27. [ 9 ] RAND, M .H ., MARKIN, T .L ., "Som e thermodynamic aspects of (U¿ Pu)02 solid solutions and their use as nuclear fuels", Thermodynamics of Nuclear Materials, 1967 (Proc. Symp. Vienna, 1967), IAEA, Vienna (1968) 637; also Rep. AERE R-5560 (1967). [1 0 ] BATTLES, J . E ., SHINN. W .A ., BLACKBURN, P .E .; EDWARDS, R .K . , '"A mass spectromet'ric investi gation of the volatilization behavior of (Uo.'8Puo.2)02-x"» Plutonium 1970 'and Other Actinides (Proc. Conf. Santa Fe, 1970), Nucl. M etall., Metall. Soc. AIME Г7 (1970) 733. [ il] OHSE, R. W .,’ "Evaporation behavior of substoichiometric (U,Pu)02", Plutonium 1970 an'd Other Actinides (Proc. Conf. Santa Fe, 1970), Nucl. M etall., Metall. Soc. AIME 17 (1970) 743. [12] LACKEY, W .J., HOMAN, F .J., OLSEN, A.R., "Porosity and actinide redistribution during irradiation in(U,Pu)02", Reactor Materials Performance (Proc. Conf. Richland, 1972), to be published in Nucl. Technol. [13] CONTE, M ., MOUCHNINO, M ., SCHMITZ, F ., "Postirradiation observation of mixed oxides with initial addition of fission product elements", Reactor Materials Performance (Proc. Conf. Richland, 1972), to be published in Nucl. Technol. [14] MEYER, R.O ., Unpublished results, Argonne National Laboratory (1972). [15] NICHOLS, F .A ., Kinetics of diffusional motion of pores in solids, J# Nucl. Mater. 30 (1969) 143. [16] NICHOLS, F.A ., Theory of columnar grain growth and central void formation in oxide fuel rods, J. Nucl. Mater. 22 (1967) 214. [ 17] JEANS, J.H ., An Introduction to the Kinetic Theory of Gases, MacMillan, New York (1940). [18] MEYER, R.O., BUTLER, E.M ., O'BOYLE, D .R., Actinide redistribution in mixed-oxide fuels irradiated in a fast flux, Trans. Amer. Nucl. Soc. 15 (1972) 216. [19] MEYER, R.O ., BUESCHER.'B. J ., A simple method of calculating the radial temperature distribution in a mixed-oxide fuel element, Nucl. Technol. 14 (1972) 153. [20] ADAMSON, M .G ., AITKEN, E. A ., A thermodynamic description of U/Pu segregation in mixed-oxide fuels, Trans. Amer. Nucl. Soc. 14 (1971) 179. [2 1 ] MACRES, V .G ., PRESTON, O ., YEW, N .C ., BUCHANAN, R ., "A shielded X -ray microprobe for the analysis of radioactive samples", Vth Int. Congr. X-Ray Optics and Microanalysis (Proc. Conf. Tubingen, 1968), Springer-Verlag, Berlin (1969) 248. [22] NATESH, R., BUTLER, E.M ., O'BOYLE, D .R., MAGRAM — A Computer Code for Quantitative Electron-microprobe Analysis of Radioactive Materials, Rep. ANL-7794 (1971). [23] KITTEL, J.H ., Guide for Irradiation Experiments in EBR-II, Unpublished report, Argonne National Laboratory (Feb. 1971). [24] NUCLEAR MATERIALS AND EQUIPMENT CORP., Fabrication of 20 w/o Pu02-U 0z Fast Breeder Fuels for Irradiation Testing in EBR-II, Rep. NUMEC-3524-74 (June 1970). [2 5 ] BOBER, M ., SARI, C ., SCHUMACHER, G ., Redistribution of uranium and plutonium during evaporation processes in mixed oxide fuel, J. Nucl. Mater. 40_(1971) 341. [2 6 ] BOBER, M ., SARI, C ., SCHUHMACHER, G ., Redistribution of plutonium and uranium in mixed (U, Pu) oxide fuel materials in a thermal gradient, J. Nucl. Mater. 39 (1971) 265. [27] RUBIN, B.F., Sodium Cooled Reactors, Fast Ceramic Reactor Development Program, Forty-first Quarterly Report, Rep. GEAP-10028-41 (1972) 5-8. [28] MAIYA, P.S., Surface relaxation kinetics of uranium dioxide, Acta Metall. 19 (1971) 255. [29] BROWN, F.L., NEIMARK, L. A., MURPHY, W.F., NATESH, R., Performance of mixed-oxide fuel elements with center melting, Trans. Amer. Nucl. Soc. 12 (1969) 107. [30] INTERNATIONAL ATOMIC ENERGY AGENCY, The Plutonium-Oxygen and Uranium-Plutonium-Oxygen Systems; A Thermochemical Assessment, Technical Reports Series No. 79, IAEA, Vienna (1967). [ 31] PFANN, W. G ., Zone Melting, John Wiley & Sons, New York (1958). [32] CRAIG, C.N. e ta l., "Steady-state performance of Pu02-U02 fast reactor fuels", Fast.Reactor Fuel Element Technology (Proc. Conf. New Orleans, 1971), Amer. Nucl. Soc., Hinsdale (1971) 555. [33] CHASANOV, M .G ., FISCHER, D .F., Out-of-pile Study of the Effects of Thermal Gradients in the Distribution of Plutonium in Fast-Reactor Fuel Materials, Rep. ANL-7703 (1970). [34] SNEDDON, B .J., Physical Properties of Some Plutonium Ceramic Compounds: A D.:tta Manual, TRG Rep. 1601 (R) (1968). [35] LACKEY, W .J., OLSEN, A.R., MILLER, Jr., J.L., BATES, D.K., Actinide redistribution in irradiated (U,Pu)02, Trans. Amer. Nucl. Soc. 14 (1971) 180. 66 O'BOYLE and MEYER [36] MEYER, R .О ., BUTLER, E .M ., O'BOYLE, D .R ., Actinide Redistribution in Mixed-oxide Fuels Irradiated in a Fast Flux, Rep. ANL-7929 (1972). [ 37] MOUCHNINO, M ., Measurement of the Pu/U Ratio in Irradiated Fuel by X-Ray Fluorescence, Unpublished report, CEA Cadarache (1972). [38] AITKEN, E .A ., ADAMSON, M .G ., DUTINA, D ., EVANS,-S.K ., LUDLOW, T .E ., A Thermodynamic Data Program Involving Plutonia and Urania at High Temperatures; Quarterly Rep. No. 19, GEAP Rep. to be published (1972). [39] THOMPSON, D .H ., private communication, Argonne National Laboratory (1969). [40] DUFFY, G .J., GREENSPAN, H ., SPARCK, S .D ., ZAPATKA, J. V ., BUTLER, M .K ., SNARG-ID, A One-Dimensional, Discrete-Ordinate, Transport-Theory Program for CDC-3600, Rep. ANL-7221 (1966). [41] STEPHENS, G .L ., CAMPBELL, D .J., Program THTB, For Analysis of General Transient Heat Transfer Systems, GE Rep. R6ÛFPD647 (April 1961). IAEA-PL-463/4а DETERMINATION DU TAUX DE COMBUSTION ET DE LA COMPOSITION EN ISOTOPES LOURDS ET EN PRODUITS DE FISSION DES COMBUSTIBLES DES REACTEURS THERMIQUES ET RAPIDES R. VIDAL, M. ROBIN, C. DEVILLERS CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Roses, France Abstract-Résumé DETERMINATION OF BURN-UP AND HEAVY ISOTOPE AND FISSION PRODUCT CONTENTS OF THERMAL AND FAST REACTOR FUELS. Burn-up is frequently the parameter which serves as a reference for purposes of analysing the numerous phenomena which, from a technological point of view, govern the behaviour of fuel elements under irradiation. It is therefore essential to have at one's disposal theoretical and experimental methods capable of determining, with adequate precision, the burn-up, the content of heavy isotopes, the percentage of fissions in each of the latter, and the fission-product content, irrespective of whether the sample to -be studied is taken from a fuel assembly irradiated in a power reactor or from elements located in irradiation loops in a research reactor. DETERMINATION DU TAUX DE COMBUSTION ET DE LA COMPOSITION EN ISOTOPES LOURDS ET EN PRODUITS DE FISSION DES COMBUSTIBLES DES REACTEURS THERMIQUES ET RAPIDES. Le taux de combustion est très souvent le paramètre qui sert de référence pour analyser les nombreux phénomènes qui régissent, du point de vue technologique,' le comportement des éléments combustibles sous irradiation. Il est donc indispensable de posséder des méthodes théoriques et expérimentales capables de déterminer, avec une précision suffisante, le taux de combustion, la composition en isotopes lourds, le pourcentage de fissions dans chacun d’eux et la composition en produits de fission, que l'échantillon à étudier soit prélevé dans un assemblage combustible irradié dans un réacteur de puissance ou dans des éléments placés dans les boucles d'irradiation d'un réacteur de recherche. 1. DEFINITION DU TAUX DE COMBUSTION Pour un élément combustible irradié dans un réacteur de puissance, le taux de combustion est défini par l'exploitant à partir du bilan thermique global de la Centrale et des courbes de distribution de puissance calculées à partir des distributions de flux mesurées périodiquement. Si le bilan thermique est en général très précis, les courbes de distribution de puissance le sont beaucoup moins, l'erreur sur l'énergie moyenne dégagée par un assemblage combustible peut atteindre plusieurs pour-cent et souvent plus si on veut atteindre une valeur absolue en un point particulier de l'assemblage. Pour les irradiations d'éléments isolés dans les boucles, en raison de la complexité des dispositifs d'irradiation, de la difficulté d'y introduire une instrumentation complexe, de l'évolution des distributions de puissance dans le réacteur lui-même, les erreurs sur le taux d'irradiation peuvent atteindre 1 0 % et parfois plus. 67 68 VIDAL et al. Du point de vue métallurgique, les phénomènes à étudier sont directe ment liés au nombre de fissions qui se sont produites dans l'échantillon; l'énergie produite n'en est que la conséquence. Aussi est-il indispensable d'exprimer le taux de combustion en nombre de fissions, grandeur qui, de plus, est directement mesurable sur l'échantillon. Exprimer le taux d'irradiation en énergie dégagée (MWj/t) est un moyen commode pour ^exploitant de réacteur, mais cette unité impose au métallurgiste et au physicien, pour calculer le nombre de fissions dans l'échantillon examiné, dé connaître l'énergie dégagée par fission. Ce coefficient variable d'un réacteur à l'autre, inaccessible à l'expérience, est assez mal connu et entraîne une incertitude supplémentaire sur le taux de fission pouvant attein d re 1 0 %. Le taux de combustion doit donc s'exprim er dans une unité qui traduit le nombre de fissions ou qui lui est proportionnelle. L'unité souvent employée est le pourcentage d'atomes disparus par fission, rapporté au nombre d'atomes lourds métalliques initial. 2. MESURE DU TAUX DE COMBUSTION 2.1. Méthode basée sur la mesure de l'appauvrissement en 235U [ 1 ] Pour la majorité des filières à neutrons thermiques le corps qui produit la plus grande part des fissions est le 235U. La mesure de l'appauvrissement en 235U, c’est-à-dire de la variation de la composition isotopique de l'uranium entre l'état initial et l'état final, donne donc directement le pourcentage d'atomes 235U disparu. La teneur en 236U donnant le nombre disparu par capture, on déduit le nombre de fissions dans le 235U directement par la relation n Y m V í — N6 \ Nu " \Nu J0 VNu + Nu J Pour connaître le nombre total de fissions, il faut tenir compte de celles produites dans le plutonium formé et aussi des fissions rapides dans le 238U. Pour les différentes filières de réacteurs à neutrons thermiques, les codes de calcul neutronique permettent de calculer avec une précision satisfaisante le pourcentage de fission dans chaque isotope du plutonium par rapport à l'appauvrissement en 235U. . Cette méthode relativement simple a été utilisée de façon satisfaisante pour les études sur les filières à uranium naturel où des précisions de 1 à 2 % ont été atteintes, grâce en particulier à la très bonne connaissance de la teneur initiale. 2. 2 .. Méthode basée sur le dosage du néodyme [ 2 ] Pour les combustibles de la filière à neutrons rapides où les fissions se font dans les différents isotopes fissiles (235U, M1Pu . . . ) ou pour ceux des réacteurs à eau légère où l'on ne connaît pas toujours avec suffisamment IAEA-PL-463/4а 69 de précision la teneur initiale et où le pourcentage des fissions dans le plutonium est important en raison des taux de combustion élevés, la méthode basée sur la mesure de l'appauvrissement en 235U n'est pas applicable. On détermine alors le taux de fission en dosant un ou plusieurs produits de fission particuliers. On utilise en.général le néodyme, parce qu'il ne migre pas dans le combustible, se prête bien aux mesures chimiques et isotopiques et, enfin, parce que certains de ses isotopes ont des propriétés nucléaires intéressantes. En particulier, le 148Nd est stable, il possède une faible section efficace de capture, un rendement de fission' élevé, peu d ifféren t pour le 235U et le 239Pu, que la fission soit produite par des neutrons th erm iq u es ou rap id es. Les méthodes de dosage actuellement utilisées, en particulier la double dilution isotopique, permettent d'atteindre une précision de l'ordre de 1% sur le rapport 148Nd/N238. Le 238U est souvent utilisé comme référence, TABLEAU I. COMPARAISON DES TAUX DE COMBUSTION Ecart Type de Taux de combustion {%) réacteur Appauvrissement U5 Néodyme-148 Nd Graphite-gaz 0, 056 0, 055 + 1, 1 (U naturel) 0, 070 0, 070 + 0,3 0, 074 0, 074 + 0, 2 0, 078 0, 078 + 0,3 0, 082 0, 081 + 0, 9 0, 083 0, 083 + 0,4 0, 133 0, 132 + 0, 3 0, 163 0, 162 + 0,4 0, 223_ 0, 222 + 0, 2 0,408 0,406 + 0,4 Graphite-gaz 0, 347 0,340 + 1, 9- (U naturel) 0, 347 0,347 -o , 1 0,383 0, 389 - 1,5 0,387 0,393 -1 ,7 Eau lourde 0, 111 0, 111 0,0 (U naturel) 0, 165 0, 161 + 2, 0 0, 332 0,329 + 0, 9 0,521 0, 526 - 1 ,0 0,812 0,818 - 0 ,8 0,914 0,926 - 1,3 1, 084 1,092 - 0, 8 70 VIDAL et al. car sa concentration initiale est en général bien connue et le nombre de noyaux de 238U évolue très peu au cours de l'irradiation. On déduit le nombre de fissions par atome lourd initial par la relation 1 148Nd WN2 38 F = — TTTTTri X ■ Ta N2 38 £ Ni Le rendement de fission ya est un rendement apparent qui tient compte du rendement pour chaque isotope fissile, et du taux de fission pour chacun d'eux. Ce rendement, qui doit être déterminé avec soin, varie de plusieurs pour-cent suivant le type de réacteur et la composition initiale du combustible. On adopte les rendements de fission suivants pour le 148Nd [3]. Fission thermique Fission rapide U-235 1, 65 1,72 Pu-2 39 1, 66 1, 74 Pu-241 1,87 U-238 - 2, 40 P u -2 4 0 La comparaison des deux méthodes de mesure du taux de fission effectuée sur des échantillons des filières à uranium naturel est présentée dans le tableau I [4]. 3. CALCUL DU TAUX DE FISSION DANS CHAQUE ISOTOPE LOURD L'évolution de la composition en isotopes lourds et donc le pourcentage de fission dans chacun d'eux varie au cours de l'irradiation suivant des lois qui ne sont pas linéaires. Cette évolution est aussi très différente suivant le type de réacteur. On a représenté sur les figures 1, 2, 3 et 4 la variation du nombre de fissions dans chacun des principaux isotopes fissiles pour les quatre types de réacteurs suivants: Réacteur à eau ordinaire pressurisée Réacteur à uranium naturel et graphite Réacteur à haute température Réacteur à neutrons rapides de 1000 MW(e). Les codes de calcul qui permettent d'obtenir ces résultats sont des codes de neutronique complexes puisqu'ils prennent en compte pour chaque étape les sections efficaces multigroupes des différents isotopes, et le spectre neutronique créé par la composition réelle du milieu à chaque étape (Coregraph [5], Everest [ 6 ], Apollo [7]). Ils sont vérifiés expérimentalement en comparant, pour différents taux d'irradiation, la composition en noyaux lourds (isotopes du plutonium) avec ceux obtenus par les analyses isotopiques et la mesure de la concentration en plutonium. La précision atteinte par ces codes pour la gamme d'irradiation usuelle varie de 2 à 5% suivant l'importance de l'isotope fissile [8,9]. Pour calculer la variation de la composition du combustible, le système d'équations qui décrit la concentration de tous les isotopes fissiles IAEA-PL-463/4а 71 MWj/t FIG. 2. Repartition des fissions ~ Réacteur graphite-gaz. 72 VIDAL et al. %>' 50 FIG. 3. Répartition des fissions — Réacteur a haute température (cycle U). IAEA-PL-463/4a 73 à chaque instant est simple, seul le calcul des sections efficaces est complexe. On peut donc définir, à partir des codes de neutronique, les sections efficaces moyennes à adopter. Puisqu'elles ne sont pas constantes, on peut se donner un jeu de sections efficaces correspondant pour chaque type de réacteur à des pas d'irradiation déterminée. 4. CALCUL DE LA COMPOSITION EN PRODUITS DE FISSION Si on connaît pour un échantillon le taux de combustion et l'histoire de l'irradiation en fonction du temps, le calcul précédent nous permet de calculer à chaque instant de l'irradiation le taux de fission de chacun des iso to p es. Le calcul de la composition en produits de fission, si ceux-ci étaient stables et non capturants, consisterait à pondérer chacun des taux de fission par les rendements de fission correspondants et à intégrer sur la durée de l'irradiation. En raison des constantes radioactives de la majorité des produits de fission, le calcul de la composition ne peut être défini avec précision que si on connaît l'histoire de l'irradiation; cette composition, évoluant constamment même après irradiation, ne peut être calculée qu'à un instant donné. Ce calcul est effectué à partir du programme Picfee [10], qui en plus de la composition en produits de fission pour différents temps d'irradiation donne aussi l'activité ß et y de l'échantillon. Si les taux de fission de chaque isotope ne sont pas donnés, le programme peut, en option, les calculer à partir d'un jeu de sections efficaces constantes moyennes sur l'ensemble de l'irradiation. A ce code de calcul est associée une bibliothèque qui contient les constantes radioactives de la plupart des nucléides (600) qui est régulièrement mise à jour pour tenir compte des nouvelles données. [ 1 1 ]. La vérification expérimentale du calcul de la composition en produits de fission n'a pas été faite à partir des dosages chimiques, car l'incertitude ne vient pas du formalisme qui est très simple, ni du taux de fission de chaque isotope qui est une donnée d'entrée, mais des constantes radioactives dont les incertitudes sont d'autant plus grandes que les temps de refroidisse ment sont plus courts. On a réalisé une vérification globale du code en utilisant une méthode calorimétrique. En effet, la puissance résiduelle dégagée par un combustible après irradiation provient essentiellement de l'énergie dégagée par l'émission ß et y des produits de fission. Pour les temps de refroidisse ment supérieurs à 1 mois, les énergies des ß et y sont suffisamment bien connues et la méthode calorimétrique est une vérification expérimentale précise des schémas de décroissance et des rendements qui définissent la composition en produits de fission du combustible étudié. Sur la figure 5 on a représênté la valeur calculée de la puissance résiduelle dégagée par une aiguille d'un élément combustible irradié dans Rapsodie, ainsi que les points de mesure pour des temps de refroidisse ment compris entre 1 et 3 mois [12]. En fonction de la précision des mesures et de celle du taux de combustion qui sert de base au calcul, on constate un très bon accord. 74 VIDAL et al. FIG, 5. Aiguille de combustible Rapsodie — Comparaison calcul-mesures calorimétriques. CONCLUSION La détermination par le calcul de la composition en produits de fission d'un combustible irradié peut être effectuée avec une bonne précision si on connaît le taux de combustion. Cependant, pour les taux de combustion élevés, en particulier pour les combustibles des filières à neutrons rapides et à haute température, le pourcentage de fission pour chaque isotope fissile varie dans de grandes proportions au cours de l'irradiation. Un soin particulier doit donc être apporté aux méthodes de calcul pour tenir compte de cette évolution, ainsi qu'aux valeurs de rendements par les neutrons thermiques et rapides des différents isotopes fissiles, en particulier du 24^-Pu. On peut conclure que les méthodes simples, qu'elles soient théoriques ou expérimentales, doivent être utilisées avec prudence et ne sont valables, pour les irradiations élevées, que dans la mesure où elles sont réajustées pour chaque gamme d'irradiation, car la composition en produits de fission n'est pas linéaire avec le taux d'irradiation. REFERENCES [1] LUCAS, M ;, Dosage du plutonium par dilution isotopique dans les combustibles irradiés - Traitement chimique — Mode opératoire, Rapport CEA-R-2564. [2] FREJA VILLE, G ., LUCAS, M ., Dosage du néodyme de fission dans les combustibles irradiés, Rapport CEA-R-4265. !AEA-PL-463/4a 75 [3] ROBIN, M ., BOUCHARD, J . , DARROUZET, M ., FREJA VILLE, G ., LUCAS, M ., PROST-MARECHAL, F. «Determination of the Nd148 yield in the fission ofU235, U238 and Pu239 by thermal and fast neutrons», Conférence BNES (Canterbury, septembre 1971). [4] ROBIN, M ., LUFFIN, M ., Communication personnelle. [5] COGNE, F ., HOFFMANN, A ., REUSS, P ., Coregraph 2 - Code de calculs de réseaux et d'évolution des réseaux à graphite, Note CEA-N-1344. [6] Calculs de cellule et évolution ponctuelle dans les réseaux a4 eau ordinaire “ Programme Everest, Rapport interne. [7] Code multigroupe de résolution de l'équation du transport pour les neutrons thermiques et rapides, Rapport CEA, en cours de publication, [8] Etude neutronique du plutonium dans les réacteurs á neutrons thermiques, Contrat CEA-EURATOM sur le recyclage du plutonium (Rapport final) 002-64-9 TRUF. [9] LAPONCHE, B. , BRUNET, M ., CHABRILLAC, M ., LUFFIN, J. , MORI ER, F ., VIDAL, R., RYCKELYNCK, J. , «-Etude neutronique du plutonium et de l'évolution des combustibles dans les réacteurs a4 neutrons thermiques», 4e Conf. Int. Util. Energie At. Fins Pacif. (Actes Conf. Genève) 9, ONU, New York, AIEA, Vienne (1972) 197. [10] ■ BARRE, B ., Programme d'intégration de courbes de fissions élémentaires tenant compte de l ’évolution des nuclides fissiles, Note CEA-N-1203, [ 11] BLACHOT, J . , de TOURREIL, R ., Bibliothèque des données nucléaires relatives aux produits de fission (36 édition)jNote CEA-N-1526. [ 12] COSTA, L. , RASTOIN, ] ., de TOURREIL, R ., Puissance résiduelle ß et y d'une fission de 235U et 239Pu, Résultats expérimentaux et théoriques pour un spectre neutronique «rap id e» (a4 paraître). IAEA-PL-463/4b DESACCORD ENTRE LES RENDEMENTS DE FISSION THEORIQUE ET EXPERIMENTAL DES GAZ NOBLES EN NEUTRONS RAPIDES Mise en évidence sur le combustible du réacteur Rapsodie B. KRYGER CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Roses, France Abstract-Résumé DISAGREEMENT BETWEEN THEORETICAL AND EXPERIMENTALLY-DETERMINED INERT GAS FISSION YIELDS IN A FAST-NEUTRON ENVIRONMENT - ANALYSIS PERFORMED ON RAPSODIE FUEL, Observations on (U, Pu)02 fuels irradiated in Rapsodie (experimental irradiations in the core blanket) which released large amounts of gas revealed that the actual volumes of gas released (83* 84* 85> 86Kr, 131, 132, 134, i36^e) were larger than the calculated volumes. The discrepancy was particularly large in the 11-11-0002 Fortissimo assembly, where all the pins exhibited the same phenomenon. At this stage of the study it was only possible to note the anomaly, which made it difficult to determine the true volume fractions released. To surmount this obstacle the authors have drawn up an experimental balance of true fission gas concentrations (gas released plus gas contained). DESACCORD ENTRE LES RENDEMENTS DE FISSION THEORIQUE ET EXPERIMENTAL DES GAZ NOBLES EN NEUTRONS RAPIDES - MISE EN EVIDENCE SUR LE COMBUSTIBLE DU REACTEUR RAPSODIE. Des observations portant sur des combustibles (U, Pu)02 irradiés dans Rapsodie (irradiations expérimentales ou cœur nourricier) et présentant un fort dégagement gazeux, ont fait apparaître que le volume des gaz dégagés (83* 84> 85» 86Kr, 131» l32> 134> 136Xe) était supérieur au volume calculé des gaz formés. Cet écart a été particulièrement mis en évidence dans l ’assemblage Fortissimo 11-11-0002 où toutes les aiguilles présentaient le même phénomène. A ce stade de l'étude on ne pouvait que constater l'anomalie contrariant la détermination de la fraction du volume effectivement dégagé. Pour lever cet obstacle les auteurs ont procédé à un bilan expérimental de la concentration des gaz de fission réellement existant (gaz dégagé + gaz occlus). 1 ) introduction Des observations portant sur des combustibles (U,Pu)C>2 irradiés dans Rapsodie (irradiations expérimentales ou coeur nourricier) et présentant un fort dégagement gazeux, ont fait apparaître que le volume des gaz dégagés (®3> ^®Kr, Xe) était supérieur au volume calculé des gaz formés. Cet écart a été particulièrement mis en évidence dans l'assemblage Fortissimo 11-11-0002 où toutes les ai guilles présentaient le même phénomène (tableau I). 77 78 KRYGER TABLEAU I. COMPARAISON DES VOLUMES DE GAZ DE FISSION CALCULES ET DEGAGES DANS L'ASSEMBLAGE FORTISSIMO 11-11-0002 Taux de combustion, Xe + Kr Xe + Kr Aiguille moyen calculés dégagés e n at. % cm 3 cm3 + 1, 5% 14 4,89 47,0 52, 6 15 4, 93 47; 6 52, 5 16 4 , 9 6 47,8 53, 6 22 4, 97 48, 0 52, 9 23 5, 00 48, 0 52, 6 24 5, 02 48,3 53, 1 29 4,96 47,8 52,1 30 5, 00 48,0 50,7 38 5, 02 48, 1 53,0 A ce stade de l'étude on ne pouvait que constater l’anomalie contrariant la détermination de la fraction du volume effectivement déga gé. Pour lever cet obstacle nous avons procédé à un bilan expérimental de la concentration des gaz de fission réellement existant (gaz dégagé + gaz occlus). 2) Techniques expérimentales La détermination volumétrique des gaz dégagés est faite à l'aide d'un capteur de pression^par comparaison avec des volumes gazeux étaIons_, puis un échantillon subit une analyse qualitative et quantitative par spectrométrie de masse. Par ailleurs, la récupération des gaz occlus se fait par évaporation ou fusion de l'oxyde mixte sous vide. Les gaz sont transportés au moyen d'un groupe de vide, une pompe Toëpler automatique permettant le stockage puis la détermination volumétrique des gaz souti rés. Un prélèvement est effectué ensuite pour analyse au spectromètre de m asse. 3) R ésu ltats : L'expérience a été menée sur l'aiguille 22 de l'assemblage précité où l'on a effectué la mesure des gaz dégagés et des mesures de IAEA-PL-463/4Ь 79 TABLEAU II. ANALYSE ISOTOPIQUE DU XENON ET DU KRYPTON DANS L1 AIGUILLE 22 DE L'ASSEMBLAGE FORTISSIMO 11-11-0002 Kryp on % X èr.on % § » > 83 84 85 86 131 132 134 136 Gaz dégagé 14, 50 16, 17 27,71 7,19 48, 93 15, 69 22, 28 33, 34 28, 65 Gaz occlus êch. 1 14,70 15, 5 27.6 7,5 49,4 14,12 21,04 34,47 30, 37 êch. 2 ' 14,81 15,8 27.7 7,4 49,1 15, 05 20,83 34,40 29,72 rétention gazeuse le long de l'axe de la colonne fissile. Ces dernières mesures varient, suivant la position du prélèvement, de 0, 048 à 0, 088 cm 3 Xe + Kr/g oxyde, soit une concentration résiduelle variant de 2, 2 à 4,1 cm3 Xe + Kr. L'intégration de nombreuses mesures suivant l'axe aurait permis de préciser cette rétention résiduelle ; cependant, comme sa contribution au volume total formé est faible, l'approximation est suffisante. Si l'on dresse, à partir de l'ensemble des mesures, le bilan expérimental des gaz de fission, on obtient ■ j Xe + Kr dégagé 52, 1 - 53, 7 cm3 96 - 93 % j (Г Xe + Kr retenu 2,2 - 4,1cm'3 4 - 7 % ) O \ ( Xe + Kr total 54, 3 - 57, 8 cm” 100 % j Les teneurs isotopiques sont consignées dans le tableau II. La comparaison des valeurs expérimentales et calculée s’établit ainsi : ( Xe + Kr calculé Xe + Kr mesuré : Ecart relatif ) ( Taux de combustion moyen i d'après le code Caprice = ( 4, 97 at % i rendemant de fission = 0, 25 ( cm 3, TPN cm3, TPN : ) ( 48, 1 54,3^V^57,8 : 16 % ) 80 KRYGER TABLEAU III. RENDEMENTS DE FISSION EN NEUTRONS RAPIDES DES XENONS ET KRYPTONS STABLES POUR 235U ET 239Pu Krypton % X ЁЛ10П % Xe + Kr Référence 83 84 85 86 131 132 134 136 % 235и 0, 614 1, 07 0, 298 i; 93 3, 15 4,45 7, 09 5, 93 24, 53 2 239pu 0, 366 0, 559 0, 134 0, 882 4, 06 5,42 7, 35 6, 92 25, 69 3 239р>ц 0, 380 0, 520 0, 138 0, 900 3,400 4, 550 6; 449 6,887 23, 22 (1 MeV) 235u 0, 382 0, 625 0, 253 1, 644 3,110 4, 550 7,810 6,791 25, 06 4 (1 MeV) 239т,Pu 0, 048 0, 200 0, 073 0, 530 4,899 6, 380 6, 300 6, 348 24,83 (2 MeV) 235и 0, 378 0, 619 0, 209 1, 682 3,081 4,408 7,736 6, 564 24, 68 5 (1 MeV) 239n Pu 0, 096 0, 195 0, 059 0, 530 4, 779 6, 222 6,143 5, 950 23, 97 . (2 MeV) Cet écart relatif de 16 % est bien supérieur à nos erreurs de mesure qui atteignent 3 %. On est donc amené à envisager une méconr naissance par défaut des paramètres qui participent au calcul des gaz de fission, à savoir : le taux de combustion et les rendements de fission. Le taux de combustion Nous utilisons, celui donné, par le code de calcul Caprice. Des contrôles effectués sur le combustible àu premier coeur de Rapsodie ont révélé un bon accord avec les déterminations expérimentales (analyse de 148Nd) [ 1 ]. L es rendem ents de fissio n Une compilation de la littérature ^2 à 5] a permis de rassembler les valeurs du tableau III. Si l'on compare les données des différents auteurs on observe souvent des désaccords importants sur les rendements individuels des isotopes alors qu'il y a unanimité sur le rende ment cumulé Xe + Kr qui oscille entre 0, 239 et 0, 257. De toute façon, ces données doivent être utilisées avec une certaine prudence. En effet, hor mis-les valeurs de Lisman et al. £2^ toutes les autres sont issues de cal- IAEA-PL-463/4Ь 81 TABLEAU IV. TAUX DE DEGAGEMENT DES GAZ DE FISSION DES AIGUILLES DE L'ASSEMBLAGE FORTISSIMO 11-11-0002 Taux de Xe + Kr Xe + Kr formé combustion dégagé Taux de (0, 29 at /-fission) dégagement Aiguilles moyen cm3 cm^ at. % +1, 5 % % 14 4,89 54, 5 52,0 S6, 5 15 4, 93 55,2 52, 5 95,1 16 4, 96 55, 5 53,6 96,6 22 4, 97 55,7 52, 9 94, 9 23 5, 00 55,7 52, 6 94, 5 24 5, 02 56,0 53,1 94, 9 29 4, 96 55,4 52,1 94,1 30 5, 00 55,6 50,7 91,1 38 5, 02 55,8 53,0 94, 9 culs théoriques. La relativement bonne concordance dé ces valeurs entre elles ne peut tenir lieu de garantie puisque les expériences de Lisman sont intervenues quatre ans après l'irradiation, l'auteur lui-même n'ex cluant pas la possibilité d'une certaine perte de gaz. Par ailleurs, les rendements de la littérature sont toujours donnés pour des énergies de neutron de 1 ou 2 MeV qui ne sont évidemment pas caractéristiques du spectre moyen de Rapsodie. П faut en effet utilisa* ces chiffres avec une certaine réserve quand on les emploie pour le com bustible de ce réacteur où l'énergie moyenne des neutrons se situe à 700 keV. Les quelques remarques émises ci-dessus montrent qu'entre le taux de combustion et les rendements de fission, c'est ce dernier paramètre qui est le plus suspect. Si donc on attribue l'origine de l'excès observé dans notre expérience au rendement de fission cumulé 83 84 85 86 131, 132, 134, 136, „ , , , ° °, °? ou Kr + ’ ’ Xe cela fait passer sa valeur moyenne de 0, 25 à 0, 29. Compte tenu de nos incertitudes expérimentales, cette valeur proposée se situerait entre 0, 28 et 0, 30. On donne dans le tableau IV le taux de dégagement des gaz de fission dès aiguilles de l'assemblage Fortissimo étudié en utilisant cette nouvelle valeur. 82 KRYGER REFERENCES [ 1] ANSELIN, F ., Communication privée. [2] LISMAN, F. L. , ABERNATHEY, R. M ., MAECK, W .J. , REIN,J, E ,, Fission yields of over 40 stable and long-lived fission products for thermal neutron fissioned 233U, 235 U, 2 39 Pu, and 241Pu and fastreactor fissioned 235U and239Pu, Nucl. Sei. Eng. 42 (1970) 191-214, [3] ANDERSON, C, A ., LA 3383 ( 1965). [4] BURRIS, L, , DILLON, I. G ., ANL 5742 (1957). [5] VARTERESSIAN, K .A ., BURRIS, L. , ANL 7678 ( 1970). IAEA-PL-463/5 OXYGEN REDISTRIBUTION IN LMFBR FUELS* S .K . EVANS, E .A. AITKEN General Electric Company, Vallecitos Nuclear Center, Pleasanton, C alif., United States of America Abstract OXYGEN REDISTRIBUTION IN LMFBR FUELS. The behaviour of oxygen in hypostoichiometric mixed oxide fuels in a temperature gradient is discussed from experiments performed out-of-pile. Oxygen migration toward the cold end of the temperature gradient has been consistently observed. However, the degree of distribution, characterized by the heat of transport, is a function of the over-all stoichiometry and the Pu content of the fuel. An empirical relation between the heat of transport and the over-all stoichiometry of typical LMFBR oxide fuels has been formulated and used to predict the O-to-M profile under a given set of irradiation conditions. Cladding oxidation imposes an additional constraint and must be properly accounted for in predicting O-to-M profiles. Other reactions involving fission products may influence the O-to-M profile under certain conditions. INTRODUCTION Stoichiom etry control in fast reactor fuel is of great importance due to the predicted effects of high oxygen activity on cladding integrity and fuel coolant reactions. The presence of temperature gradients in all fuels and the attendant gradient in oxygen potential require that the properties of the oxygen potential gradient be well established so that proper predictions regarding fuel behavior can' be made. Several aspects of the thermomigration problem require attention. Since oxygen migrates to the cooler regions of hypostoichiometric urania-plutonia fuel, it leaves the center deficient in oxygen when compared to the average fuel composition. The cladding may act as a sink for the excess oxygen which appears in its vicinity. In the presence of cesium fission products, the excess oxygen may enhance corrosion of the cladding materials. In the event of the failure of a fast reactor fuel pin in sodium coolant, fuel-sodium reaction products may form and cause swelling of the ruptured fuel pin with possible detrimental effects on coolant flow and fuel integrity. EXPERIMENTAL A series of out-of-pile experiments was conducted to determine the behavior of oxygen in (Ug 75 PU0 25Ю 2-Х subjected to a temperature gradient. The bulk of the experiments were performed in an axial geometry, when the temperature gradient was impressed across a linear array of fuel pellets. The axial experiments were followed by a similar experiment in a radial geometry (large pellets centrally heated by electric current through a molybdenum rod) to confirm the expected similarities between the two geometries. Both axial and radial temperature gradient experiments provided stoichiometry data as a function of the temperature gradient which could be described by the Arrhenius equation Kn X = Q7RT + С (1) where X is the deviation from stoichiometry, Q* is defined as a heat of transport, and С is a constant whose value is set by the initial stoichiometry. * Work performed under USAEC contracts A T(04-3)-189, PA-53 and A1704-3)-1S9, PA-10. 83 8 4 EVANS and AITKEN DISCUSSION Rand and Roberts[ 1 ] recognized that oxygen redistribution in mixed oxide fuel should take place in a temperature gradient provided a sufficient H 2 O or CO2 pressure was present in the system. It is easily shown that for a constant H2 :H 2 Û or C0 :CÛ 2 ratio in the mixed oxide system the equilibrium stoichiometry changes with temperature. The equilibrium is attained through transport of oxygen in the gas phase through the H 2 O and CO 2 species. Application of the Rand and Roberts mechanism to the Arrhenius equation would result in a heat of transport of -30 to -60 kcal/mole depending on the average stoichiometry. Since experimental results[2] in out-of-pile studies generally indicate far less negative values than mentioned above, A itk e n [3] proposed a mechanism for oxygen transport in the temperature gradient based on oxygen migration in both the gas and solid phases. The effect of the addition of the solid-phase migration component to the mechanism of oxygen transport is an alteration of the heat of transport term in the Arrhenius equation. The revised heat of transport can be written 0 * = Q * r r /(1 + Ls/Lg) (2) where ü*f¡R IS heat transport expected under conditions of constant H2:H20 or CO:CÛ 2 ratios, and Ls and Lg are phenomenological coefficients related to oxygen m obility in the solid and gas phases respectively. Equation (2) indicates than an increase in oxygen mobility in the solid with the gas phase mobility remaining constant results in a decrease in the magnitude of the heat of transport. Since the heat of transport describes the slope of the Sn X versus 1/T curve, a negative value for CL* would cause an increasing deviation from stoichiometry as the temperature is increased. Physically, this is observable as a transport of oxygen from high to lower temperatures. Thus, under gas phase oxygen transport, an oxygen concentration gradient would be created in the fuel due to the temperature gradient. If, then, solid state diffusion was allowed, this same concentration gradient would provide a driving force for oxygen ion diffusion which would tend to decrease the magnitude of the concentration gradient. Thus the mechanism of Aitken can be termed a cyclic mechanism with oxygen migrating down the temperature gradient in the gas phase and up the temperature gradient in the solid phase, A fter sufficient time then, a dynamic steady state is set up in which the magnitude of the heat of transport is decreased as the oxygen ion m obility in the solid is increased as is expected from Equation 2. Experiments[4] have shown the observed heat of transport is altered by the presence of artificial barriers in the fuel array. In a linear array of cylindrical pellets, the pellet interfaces act as barriers. For translating out-of-pile results to in-pile predictions, only those results pertaining to continuous fuel are considered. Experiments[5] have shown that the heat of transport is a function of the average deviation from stoichiometry of the form Q* = -22.5/X2 (3) as long as the average stoichiometry is less than 1.98 oxygen-to-metal ratio. For stoichiometries between 1.98 and 2.00 oxygen-to-metal ratio, the auxiliary H2 O and C O 2 gas pressures in the system are sufficiently large that the Rand-Roberts[1] mechanism of gas phase oxygen transport applies. The heat of transport in this stoichiometry region is approximately -30 kcal/mole. This empirical equation can be used to calculate predicted radial stoichiometry profiles in a hypothetical LM FBR fuel rod provided that the fuel surface stoichiometry and temperature profile are specified. The results of these calculations for a fuel surface temperature of 1100°С and a linear power rating of 15 kW/ft are shown in Figure 1. As can be seen from the figure, significant oxygen redistribution takes place even for fuel stoichiometries very close to 2.00. Calculations have been made to determine the effects of changes in linear power, and possible errors in experimentally determined quantities on the resulting profiles. As would be expected, a decrease in linear power results in a flattening of the stoichiometry profile. Changes in the profiles from errors in experimentally FIG. 1. Stoichiometry profiles for 15 kW/ft irradiation with 1100*0 fuel surface temperature using heat of of heat using temperature surface fuel 1100*0 with irradiation kW/ft 15 for profiles Stoichiometry 1. FIG. rnpr fo eprmn ihu riiil barriers. artificial without experiment from transport Oxygen-to-metal Ratio IAEA-PL-463/5 Relative radius, R/Ro radius, Relative 86 EVANS and AITKEN determined parameters appear to be rather small, of the order of ± 0.005 O'.M unitsfor an average stoichiometry of 1.98. The irradiation situation is complicated by volatile fission products, which may affect the oxygen redistribution, constantly being added to the system. The fission product most likely to act as an oxygen transport agent is cesium. An experiment was performed in the presence of cesium to determine the effects of this element on the behavior of the redistribution phenomenon. Results were not significantly different from experiments not containing cesium. One of the reasons that mixed oxide fuels are specified to be hypostoichiom etric is to avoid oxidation of the stainless steel cladding. The sequence of events which may be expected to occur in a hypostoichiometric fuel rod operating under neutron irradiation can be predicted w ith some degree of certainty. The rod begins its irradiation period with constant stoichiometry. As the temperature gradient is developed, oxygen redistributes toward the periphery of the fuel. This migration, as we have said before, can be expected to be completed very early in life. At this point, the fuel-cladding interface may or may not have a high enough oxygen potential to oxidize the chromium in the cladding. If it is assumed that a fuel surface stoichiometry of 1.998 or higher is sufficient to oxidize the chromium in the cladding, then use of fuel with an average stoichiometry about 1.96 will not result in cladding oxidation in the early stages of irradiation. As the irradiation progresses, the average oxygen activity in the fuel rod increases because of the effects of burnup and brings the fuel at the fuelxladding interface more closely to the stoichiometric composition. When the fuel surface stoichiometry reaches about 1.998, the chromium in the cladding will begin to be oxidized and form an adherent film on the cladding interior surface. Additional oxidation of the cladding will depend on the diffusion kinetics of chromium and iron through the chromium oxide layer or interactions between fission products and the oxide film which destroy the film and allow faster oxidation of the chromium than would be expected from diffusion processes alone. If the cladding retains its integrity and the oxidation process is sufficiently slow, the fuel will eventually reach an average stoichiometric composition. At this time, no measurable stoichiometry gradient will exist in the fuel. Before this change in the gradient can take place, the stoichiometry at the fuel center must increase. This increase gradually takes place as the average stoichiometry is raised. As soon as the fuel becomes slightly hyperstoichiom etric, the molybdenum fission products will begin to be oxidized to M0 O2 . This oxidation process buffers the oxygen activity in the fuel rod as long as some molybdenum metal is present in the system. If the cladding oxidation process is rapid compared to the increase in oxygen available with burnup, then the cladding would act as a sink for oxygen throughout the life of the fuel rod. If that were the case, the surface fuel stoichiometry would never rise above the point where cladding oxidation begins, and significant stoichiometry gradient would always be present. It is evident that redistribution of oxygen plays a strong role in determination of the lifetime of a liquid metal fast breeder reactor (LM FBR ) fuel pin. Future models of fuel pin performance must take into account such redistribution to provide a realistic picture of events taking place during the fuel lifetime concerning both mechanical and thermochemical properties of the pin during its irradiation life. REFERENCES [1] RAMD, M. H., and ROBERTS, L. H., Symposium on Thermodynamics of Nuclear Materials, 1 (IAEA, Vienna 1965) 3. ~ [2] EVANS, S. K., AITKEN, E. A., and CRAIG, C. V., "Effect of a Temperature Gradient on the Stoichiometry of Urania-Plutonia Fuel," J. Nucl. Matl., 30 (1969) 62-73. 13] AITKEN , E. A., "Thermal Diffusion in Closed Oxide Fuel Systems," J. Nucl. Matl. 30 (1969) 62-73. [4] AITKEN, E. A., ADAMSON, M. G., EVANS, S. K., and LUDLOW, T. E., "A Thermodynamic Data Program Involving Plutonia and Urania at High Temperatures, Quarterly Report 15," General Electric Company, April 1971 (GEAP-12210). [5] AITKEN, E. A., ADAMSON, M. G., EVANS, S. K., and LUDLOW, T. E., "A Thermodynamic Data Program Involving Plutonia and Urania at High Temperatures, Quarterly Report 17,” General Electric Company, October 1971 (GEAP I 2254). IAEA-PL-463/6a MODIFICATION DU POTENTIEL D'OXYDATION DES OXYDES MIXTES (U, Pu)02, EN FONCTION DU TAUX DE COMBUSTION F. SCHMITZ, J. MARTICORENA, G. DEAN CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Roses, France Abstract-Résu mé MODIFICATION OF THE OXIDATION POTENTIAL OF MIXED OXIDES (U, Pu)02 AS A FUNCTION OF BURN-UP. The oxidation potential of mixed oxide (represented by A G(q2)) is a key parameter of this nuclear fuel, and varies with the O/U + Pu ratio which develops during irradiation. It has been shown that in pure mixed oxides the variation of Д G(02) c a n be linked to the valencies of uranium and plutonium up to about 30°jo plutonium. In irradiated fuels this situation is changed by the dissolution of certain fission products in the (U, Pu)02 matrix. In order to avoid handling difficulties resulting from the radioactivity of irradiated fuels, the investigations are conducted mainly on non-radioactive samples simulating the chemical composition of a fuel after irradiation. It has been possible to confirm certain results on actually irradiated fuels. MODIFICATION DU POTENTIEL D'OXYDATION DES OXYDES MIXTES (U, Pu)Oz, EN FONCTION DU TAUX DE COMBUSTION. Le potentiel d'oxydation de l'oxyde mixte (représenté par Д G^q2)) est un paramètre clé de ce combustible nucléaire. Il varie avec le rapport O/U + Pu qui évolue en cours d’ irradiation. Il a été montré que dans les oxydes mixtes purs la variation de Д ^(02) Peut ^tre reliée aux valences de l ’ uranium et du plutonium jusqu' à environ 30% de plutonium. Dans les combustibles irradiés cette situation est changée par la mise en solution dans la matrice (U, Pu)02 de certains produits de fission* Pour éviter les difficultés de manipulation résultant dè la radioactivité des combustibles irradiés, les études sont essentiellement basées sur des échantillons non radioactifs simulant la composition chimique d'un combustible après irradiation. Certains résultats ont pu être confirmés sur des combustibles réellement irradiés. . INTRODUCTION Le potentiel d'oxydation de l'oxyde mixte (représenté par Д G(q j ) est un paramètre clé de ce combustible nucléaire. Il varie avec le rapport O/U + Pu qui évolue en cours d'irradiation. Il a été montré que dans les oxydes mixtes purs la variation de A G(Q,)peut Stre reliée aux valences de l'uranium et du plutonium jusqu'à environ 30 % de plutonium . Dans le s combustibles irradiés cette situation est changée par la mise'en solution dans la matrice (U,Pu)C>2 de certains produits de fission. Pour éviter les difficultés de manipulation résultant de la radio activité des combustibles irradiés, nos études sont essentiellement basées sur des échantillons non radioactifs simulant la composition chimique d'un combustible après irradiation / ”2 J. Certains résultats ont pu Stre confirmés sur des combustibles réellement irradiés. 87 со со TABLEAU I. COMPOSITION ET VALENCE MOYENNES DES PRODUITS DE FISSION (P. F. ) POUR DIVERS TAUX DE COMBUSTION DANS LE CAS DES REACTEURS RAPSODIE ET PHENIX a) RAPSODIE (Pu/U + Pu = 0, 26 — Uranium enrichi à 65% de 235U) b) PHENIX (Pu/U + Pu = 0, 20 — Uranium naturel) Groupes de P. F. t = 2,22% at TCF V = 6,46% at TCF 12,50% at TCF t- = 1,94 % at TCF Ъ = 5,43 % at TCF 2" = 10,90 % at TCF ppm at % ppm at % ppm at % ppm at % ppm at % ppm at % Z P. F. 44468 100 129356 100 250828 100 38726 100 108638 100 218039 100 Kr, Xe 5460 12,27 15650 12,09 31360 12,50 4493 11,60 12650 11,64 25400 11,64 Cs, Rb 4425 9,95 12810 9,96 25610 10,21 3618 9,34 10374 9,54 20870 9,57 Br, I, Se, Te 1168 2,62 3193 2,46 6351 2,53 1225 3 ,16 3197 2,94 6264 2,87 Métaux 123 87 27,85 36771 28,42 74685 29,77 13465 34,76 3 8805 35,71 78996 3 6,23 Oxydes non soluble 4243 9,54 12479 9,64 184 80 7,36 2702 6,97 6664 6,13 12991 5,95 Oxydes solubles 16785 3 7,74 4 83 73 3 7,39 94342 37,61 13223 34,14 36958 34,01 73518 33,71 Kr/Xe 0,1422 0Д423 0,1403 0,0723 0,0720 0,0717 Valences moyennes des P.F. 1,46 1,48 1,37 1,33 V1 1,55 1,31 v'i 1,75 1,66 1,6 8 1,55 1,55 1,51 2,00 2,05 1, 85 1,84 v2 ! 2,05 1, 85 4 2,25 2,19 2,2 5 2,07 2,08 2,04 3,55 3,52 3,52 3,54 3,51 3,50 Vsolub les Vj formation des oxydes de terres rares de ZiO^ et v 2 c°mme Plus formation de MoO^ V 1 comme v plus formation de Cs O et Rb O v' comme v plus formation de Cs O et Rb O 11' ¿ ¿ ¿j Í . Ù ¿ IAEA-PL-463/6; 89 I - EVOLUTION DES OXYDES MIXTES SOUS IRRADIATION 1-1 - Rappel sur le mécanisme de la fission La fission d'un atome lourd, dans une matrice d'oxyde mixte, se traduit par la formation de deux atomes de produits de fission ( P.F. ) et la libération de l'oxygène qui lui était précédemment lié. Ceci peut Stre sché matisé par la réaction : (U,Pu)02_x flfsl°? ( 1 - Z ) (U,Pu)02_x + 2 r(P .F.) +-г-(02-х) Cette réaction se poursuit par une redistribution de l'oxygène entre la matrice d'oxyde et les produits de fission : (1 - V ) [(U,Pu)02_ J + ÎT. 0 2_x + 2 2 -(p. F .)—» (1 - 2) [(U,Pu)0 2 _ x _ 2 1 + 2 ? ( P.F.)Oz ~ т ^ — i La répartition de l'oxygène est alors caractérisée par le potentiel d'oxydation û G(q 2) fixant le rapport O/U + Pu et la valence moyenne des produits de fission. Il est possible de déterminer la valence probable des divers produits de fission à une température donnée en comparant le's valeurs des AG(o 2) des oxydes de produits de fission à celui de la matrice (U,Pu)02_x 1-2- Valence moyenne des produits de fission La composition finale d'un combustible irradié est déterminée par le nombre de fissions et le rendement de fission de chacun des produits de fission, la durée de l'irradiation et le temps de refroidissement. Le programme PICFEE ¿_ 4.7 permet de calculer la concentration des différents produits de fission. Il est donc possible, en attribuant à chaque élément sa valence pro bable, v¿, en présence d'(U,Pu)02/ de calculer la valence moyenne v des pro duits de fission d'après la relation^/" 5 J : v = c i vi I C i Cette valence moyenne est inférieure à (2-x) et plus petite dans le cas de la fission du plutonium-239 que dans celui de l'uranium-235, la for mation de produits de fission nobles (Ru, Rh, Pd, Mo, etc. . ) étant favorisée pour le plutonium (tableau I). Il en résulte que la fission entrafhe une augmentation du rapport O/U + Pu de la m atrice selon la relation O _ s i ~ U + Pu 1 -ZT 90 SCHMITZ et al. OÙ - représente la stoechiométrie initiale de l'oxyde et Ъ le taux de com bustion en fraction d’atomes brûlés . Si l'on tient compte de la contribution des produits de fission solubles dans la matrice ( terres rares, Y2 O3 , ZrC^), de-valence moyenne - u, le rapport oxygène/métal de la matrice devient O í + O 2 - 2 “ (v-d u) 0//M = U + Pu + P.F. solubles = 1 - '2'+ 2 "S' d av ec 0 1 = oxygène lié à (U + Pu) C>2 = oxygène lié aux produits de fission solubles d = fraction de produits de fission solubles. La figure 1 schématise l'évolution de O/U + Pu et O/M en fonc tion du taux de combustion. \ Deux faits sont alors à retenir : - O/M e st toujours inférieur à O/U + Pu - Si O/U + Pu reste constant, O/M diminue régulièrement avec Ъ ". Nous admettons ici que l'existence d'ions Pu 3 + exclut la pré sence simultanée d'ions uranium de valence supérieure à 4 . Le rapport O/U+Pu définit alors le degré d'oxydation de la matrice, tandis que le rapport O/M définit le degré de SEtureticn en lacunes du réseau. Le rapport O/U + Pu croissant avec le taux de combustion, il existe une valeur de AG(Qg ) telle que la valence moyenne des produits de fission augmente à cause de l'oxydation du molybdène. Ce point atteint, le rapport O/U + Pu reste constant pendant une certaine période, l'oxygène libéré par fission étant consommé par le molybdène formé précédemment. Dans.le cas de la fission de l'uranium-23 5 seul, l'évolution de O/U est même définitive- . ment stoppée. D'autres composés stabilisateurs pourraient d'ailleurs être invoqués ( à la suite de la formation d'uranates ou de molybdates de césium ou de l'oxydation de la gaine par exemple ). 1-3- Me sure'du rapport O/U + Pu - La présence de produits de fission risquant de fausser les mesures gravimétriques après un traitement standard d'oxydo-réduction, nous avons développé une méthode basée sur la mesure du paramètre réticulaire de la matrice avant et après ce traitement standard . Cette méthode repose sur le fait que même en présence de produits de fission, le traitement préco - nisé pour les oxydes purs conduit à O/U + Pu = 2,00 et non O/M = 2, 00, ce que suggèrent certaines expériences /” 7 _ J . IAEA-PL-463/6a 91 combustibles: Type 2 — fissions U et Pu Type 3 — fission Pu seule La connaissance du rapport O/U + Pu d’un combustible irradié permet, connaissant le taux de combustion de calculer le rapport O/M d'après la relation (l-'f r ) (O/U + Pu) + d û O/M = 1 - 7r + 2 'à' d 92 SCHMITZ et al. II RESULTATS EXPERIMENTAUX La plupart des expériences ont été effectuées sur des échantillons de simulation obtenus par frittage à 1750° C. Le bilan d'oxygène,correspon dant à une stoechiométrie initiale et un taux de combustion donnés,est ajusté par addition d*U3 Og f équivalent à 3 UC>2 + O2 ) J. II—1 - Evolution du paramètre réticulaire avec le taux de combustion Dans deux compositions simulant des taux de combustion de 2 et 6 % at TCF, pour le domaine sous-stoechiométrique, nous avons pu vérifier la validité de nos hypothèses sur la valence moyenne des produits de fission. L'accord entre le calcul et les résultats est excellent pour le rapport O/U + Pu = 2,00. Dans ce cas le paramètre réticulaire est calculé . suivant l'équation /” 5 / : concentration des produits de fission considéré, en solution dans la matrice Vi volume moléculaire de cet oxyde en solution. Nous avons trouvé que le volume moléculaire moyen pondéré des produits de fission et le rapport Pu/U + Pu varient peu en fonction du taux FIG .2. Variation du paramètre en fonction du taux de combustion pour O/U+Pu - 2,00 Courbes calculées: 1 - Pu/U+Pu = 20% U naturel 2 — Pu/U+Pu = 25% U naturel 3 — Rapsodie: Pu/U+Pu = 26%, U enrichi à 65% d'uranium-235 Résultats expérimentaux: • — Simulation Pu/U+Pu = 20% ▼ *“ Rapsodie. IAEA-PL-463/6a 93 de combustion. La variation du paramètre réticulaire peut alors s'écrire sous la forme a3= i-гг + 2 d tr ^иоз^иог^РчОг vPu02) + 2 6 d v (p.f.)o2 Les résultats des calculs effectués pour les combustibles de Rapsodie et de Phénix sont reportés sur la figure 2. Les points expérimentaux obtenus lors des simulations, ou sur le combustible irradié de Rapsodie ZT6_7, montrent que l'accord avec nos prévisions est bon. Le léger décalage des points obtenus sur Rapsodie peut s'expliquer par le fait que le mélange CO/CO 2 utilisé pour ramener les échantillons à l'état standard ne correspondait pas tout à fait à 0 , 1 0 mais était légèrement plus oxydant. II—1 — 1 - RÔle_du molybdène Pour un essai simulant un taux de combustion de 10,2 % at TCF, nous avons établi un bilan d'oxygène tel que pour un rapport initial O/U + Pu = 1 ,94, la stoechiométrie finale de la matrice d'oxyde devait être soit de 2,015 soit de 1, 965 selon que le molybdène était à l'état métallique (valence des produits de fission 1,27 ) ou totalement oxydé ( valence des produits de fission 1, 72). FIG.3. Variation du paramètre cristallin en fonction de la température. 94 SCHMITZ et al. FIG.4. Schéma de la variation 8u potentiel d'oxydation des oxydes de simulation comparés aux oxydes purs et à l'équilibre Mo-Mo02. Après frittage à 1750° C, la mesure du paramètre cristallin conduit à un rapport'O/U + Pu = 1,98 intermédiaire, correspondant à une valence effec tive des produits de fission de 1, 59. Nous en déduisons donc qu'un état d’équi libre a été atteint entre la matrice et le système M 0 /M 0 O2 . L*oxydation du molybdène a d'ailleurs été mise en évidence par microanalyse X. Après un traitement standard d'oxydo-réduction à 800° C, nous observons simultanément une diminution du paramètre réticulaire et une augmen tation de poids, prouvant que la matrice en équilibre avec le système M 0/M 0 O2 a v ait un rapport O/U + Pu 4 2,00 . Sur ce même échantillon nous avons mesuré la variation du paramètre réticulaire en fonction de la température, à l'intérieur d'un capillaire scellé ( courbe a^(T) de la figure 3 ). Compte tenu de la dila tation pure de l'échantillon qui est mesurée au dilatomètre ( a 2 (T).d'ailleurs proche de celle d'un oxyde pur, la variation due au changement de O/U + Pu d'équilibre devrait correspondre à la courbe, аз (T). IAEA-PL-463/6a 95 Tous ces résultats prouvent donc qu'il existe un équilibre qui dépend de la température, tel que le rapport O/U + Pu décroît quand la température aug mente . La figure 4 donne schématiquement un modèle permettant d'expliquer l’ensemble de nos observations. Les conclusions sont alors les suivantes : Dans les combustibles irradiés la courbe Д ) = f ( T ) de l’équi libre Mo/Mo0 2 est déplacée vers des valeurs plus négatives. Ceci peut s'expli quer par une faible activité du molybdène dans la réaction Mo + O2 ------î ( M 0 O2 ) La pente des courbes A Gjq j = f ( T ) de la matrice doit être plus forte que celle du système M 0/M 0 O2 , tout au moins dans le domaine surstoe- chiométrique. I I —1—2 - Ipf^uence_ po_s_sible de la formation_dVon_ com posé du type UPdg Dans une simulation d'un taux de combustion de 16,25 % at TCF le bilan d’oxygène était tel que le rapport O/U + Pu final aurait dû être de 1,96. Or la mesure du paramètre cristallin conduit à une valeur de ce rapport de 1,973, correspondant à une valence effective des produits de fission de 1 ,2 0 au lieu de 1,26. Pour ces valeurs de stoechiométrie de la matrice, les produits de fission n'ayant aucune raison d'avoir une valence moyenne inférieure aux pré-- visions, une réaction libérant de l’oxygène à la suite de la formation d’un composé du type (U,Pu)Pdg permettrait d'expliquer cette différence. (U ,Pu)0 2 -y + 3x Pd—>(l-x) (U,Pu)0 2 -V + x (U,Pu)Pd 3 . 1 —x Il suffit alors de 0,6 % at de (U,Pu) réagissant avec 1, 8 at % de (Pd, Rh ou Ru ) pour provoquer l'écart observé. Or dans cette simulation, 5,4 % at de ces éléments étaient présents. La valeur O/U + Pu = 1,973 représente alors la stoechiométrie d’équi libre de la m atrice en p résen ce du com posé (U,Pu) (Ru, Pd, Rh)g à la tem pérature de frittage de 1750° C. II-2 - Mesures d'énergie libre molaire partielle AG^nj Pour mesurer directement l’influence des produits de fission sur le potentiel d'oxygène des composés mixtes, nous avons opposé dans une pile à électrolyte solide une oxyde mixte pur et un oxyde simulant un taux de com bustion de 16,2 5 at % TCF. Un four équipé d'une jauge à oxygène permet d'effec tuer des recuits sous atmosphère contrôlée de C0/Ç02 . Le processus expéri mental est le suivant : - Les deux électrodes de la pile sont d'abord équilibrées sous une atmosphère CO/CO 2 conduisant à un Д G(q 2 ). bien déterminé pour la température considérée.. 96 SCHMITZ et al. FIG. 5. Evolution en fonction de la température de 1' écart d [A G(Oz)3 entre la référence d ' oxyde mixte pur (Pu/U+Pu = 0,2) et un oxyde simulant 16,25% at TCF. ■ft Echantillon équilibré à 1187°C sous CO2/CO = 10"3 o • Echantillon équilibré à 1308°C sous CO2/CO = 10"3. - Une fois la f. e. m de la pile nulle, l'atmosphère est rapidement remplacée par de l'hélium. - Enfin le système est isolé, et la variation de la f.e.m de la pjle en fonction de la température permet de connaître directement l'écart de Д ) entre l'oxyde pur et celui contenant des produits de fission. La figure 5 représente les résultats obtenus sur des échantillons équilibrés sous une atmosphère de CO/CO 2 = 10-3, à des températures de 1187 et 1308° C . Une interprétation quantitative de ces phénomènes est rendue dif ficile par le fait que l'introduction de l'hélium ( certainement à cause de son p(C>2 ) différent de celui du mélange CO/CO 2 ) déséquilibre le système et IAEA-PL-463/6a 97 FIG. 6. Représentation schématique de l'évolution de Д G(o2) de l'oxyde de simulation (16,25 at % TCF) en fonction de la température. conduit à l'apparition d'une f.e.m. Toutefois, en descendant la température on retrouve un nouvel équilibre. Mais en continuant la descente en température la polarité de la pile ne change pas et la f.e.m croît de nouveau, dans le sens d'un Д G (Q 2 )Plus négatif que celui de l'oxydé de référence. L'intervention d'un mécanisme stabilisateur schématisé sur la figure 6 peut expliquer ce comportement. III - CONCLUSION L’influence de la présence de produits de fission sur le potentiel d’oxydation des oxydes mixtes a pu être mise en évidence sur des oxydes simulés soit directement par des mesures de f.e.m de piles à électrolyte Solide, soit indirectement par l'effet sur le paramètre cristallin de la. matrice. Il ressort de ces résultats que le potentiel d'oxygène des oxydes mixtes contenant des produits de fission est plus faible que celui des oxydes purs. En outre plusieurs mécanismes stabilisateurs de ce potentiel semblent intervenir et en particulier le système M 0/M 0 O2 . REFERENCES [1] RAND, M .H ., MARKIN, T .L ., in Thermodynamics of Nuclear Materials, 1967 (Compt. Rend. Coll. Vienne, 1967), AIEA, Vienne (1968) 637. [2] SCHMITZ, F ., Rapport CEA-R-3795 (1969). 98 SCHMITZ et al. [3] SCHMITZ, F ., DEAN, G ., HALACHMY, M ., J. Nucl. Mater. 40 (1971) 325-337. [4] BARKE, B., Note CEA-N-1023 (1969). [5] ANSELIN, F., GEAP 5583 (1969). [6] TROTABAS, M ., Communication personnelle. [7 ] SCHMITZ, F . , et a l . , ANS Meeting on Fast Reactor and Fuel Elements (Karlsruhe, 1970). [8 ] DE KEROULAS, F . , et a l., «Microanalyse X d'oxyde (U, Pu)Oz + x simulant différents taux d'irradiation», ces comptes rendus, IAEA-PL-463/9a. IAEA-PL-463/6b EVOLUTION DE L'ACTIVITE DU CARBONE EN FONCTION DU TAUX DE COMBUSTION DANS UN COMBUSTIBLE CARBURE Nicole LORENZELLI, J.P . MARCON CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Roses, France Abstract-Résumé DEVELOPMENT OF CARBON ACTIVITY AS A FUNCTION OF BURN-UP IN A CARBIDE FUEL. The monocarbide UC and the mixed carbide (U, Pu)C can be used as nuclear fuels for fast reactors. Investigation of their thermodynamic properties is interesting from several points of view, including possible carburization of the stainless steel can by the fuel, release of volatile fission products (Cs, I, Te, Ba, Sr), and chemical behaviour in the presence of sodium. Although the properties of UC and (U, Pu)C are fairly well known, it is nevertheless important to consider the effect of fission products on them, since at a bum-up of 100 000 MWd/t the concentrations reach several per cent and are capable of modifying the UC matrix. One of the parameters which best characterizes the thermodynamic properties of a carbide is carbon activity. As the latter by definition depends to a considerable extent on the С/U ratio, its development as a function of burn-up is investigated from several points of view. EVOLUTION DE L' ACTIVITE DU CARBONE EN FONCTION DU TAUX DE COMBUSTION DANS UN COMBUSTIBLE CARBURE. Le monocarbure UC et le carbure mixte (U, Pu)C sont susceptibles d'être utilisés comme combustibles nucléaires pour les réacteurs rapides. L'étude de leurs propriétés thermodynamiques est intéressante à plusieurs points de vue: carburation éventuelle de la gaine d'acier inoxydable par le combustible, dégagement des produits de fission volatils (Cs, I, Te, Ba, Sr), comportement chimique en présence de sodium. Si les propriétés de UC et (U, Pu)C sont assez bien connues, il est important de considérer Г influence des produits de fission sur celles-ci, les concentrations pour un taux de combustion de 100 000 MWj/t atteignant plusieurs pourcent et pouvant modifier les propriétés de la matrice UC. Un des paramètres fixant le mieux les propriétés thermodynamiques d'un carbure est l'activité du carbone. C elle-ci dépendant à priori de façon considérable du rapport С/U, son évolution en fonction du taux de combustion est examinée sous différents aspects. INTRODUCTION Le monocarbure UC et le carbure mixte ( U, Pu )C sont suscep tibles d'être utilisés comme combustibles nucléaires pour les réacteurs rapides. L'étude de leurs propriétés thermodynamiques est intéressantes à plusieurs points de vue : carburation éventuelle de la gaine d'acier ino xydable par le combustible, dégagement des produits de fission volatils ( Cs, I, Te, Ba, Sr ), comportement chimique en présence de sodium. Si les propriétés de UC et ( U, Pu )C sont maintenant assez bien connues, il est important de considérer l'influence des produits de fission sur celles- ci, les concentrations,pour un taux de combustion de 100 000 MWj/t attei gnant plusieurs pour-cent et pouvant modifier les propriétés de la matrice UC. Un des paramètres fixant le mieux les propriétés thermodynamiques d'un carbure est l'activité du carbone. Celle-ci dépendant à priori de façon considérable du rapport С/U, son évolution en fonction du taux de combus tion devra être examinée sous deux aspects : - Calcul de la variation du rapport C en fonction du taux de, combustion. U 99 100 LORENZELLI et MARCON - A taux de comlgistion donné, mesure de l'activité du carbone en fonc tion du rapport _g_ , la combinaison de ces deux études pouvant seule donner l'évolution de l'activité du carbone en fonction du taux de combus tion. C I - CALCUL DE LA VARIATION DU RAPPORT — EN FONCTION DU TAUX DE COMBUSTION. Ce calcul nécessite à priori l'étude du comportement physico-chimique des produits de fission dans la matrice du monocarbure UC, c'est-à-dire une étude métallurgique du diagramme U - C - produits de fission (P. F. ) au voisinage de la composition UC, dont nous allons rappeler brièvement les principaux résultats. A - Etude métallurgique du diagramme U - C - Produits de fission(PF) L'étude des différents diagrammes ternaires U - C - PF ayant déjà fait l'objet de publications (1 , 2, ), nous parlerons simplement ici des résultats relatifs au carbure UC "simulant" un taux de combus tion donné, c’est-à-dire un composé contenant les différents produits de fission introduits sous forme inactive en proportion voulue, et mis en équilibre thermodynamique. 1 _ Techniques expérimental es Les différents P. F. ont été introduits soit sous forme de poudre métallique (Zr, Mo, Re, Ru, Rh, Pd) soit sous forme d'hydrure (terres rares), soit sous forme massive (Sr, Ba, Cs, I, Te). Etant dans l'impossibilité de faire diffuser simultanément tous les P. F ., nous avons opéré par étapes successives à savoir : - A partir d'un échantillon de UC préparé classiquement, nous faisons diffuser Zr et Mo sous vide à 1800° C. - Cet échantillon, rebroyé, est mélangé aux hydrures de terres rares (préalablement dissociés) et aux métaux nobles (Ru, Rh, Pd, Re), puis recuit à 1500° C dans un tube scellé de platine ou .d' iridium. Afin d'évi ter la formation possible de U2 Ir Cg, on intercale une feuille de tantale entre l'échantillon et le tube scellé. - Cs, Sr, Ba, I et Te sont introduits avec l'échantillon précédent dans un tube d'acier inoxydable scellé, puis recuit à 1000° C. - Une fois préparés les échantillons ont été étudiés par analyse chimique par diffraction de rayons X, et par la microsonde Castaing. 2 - Eê_sultats_ Rappelons d'abord que nous n'avons introduit que les P. F. suivants : Zr, Mo, Re (à la place de Te), Ru, Rh, Pd, Y, La, Ce (représentant Ce + Sm + Pm), Nd (Nd + Pr), Sr (Ba + Sr), Cs, I, Te. De plus, trois taux de combustion ont été étudiés : 70 000, 100 000 et 150 000 MWj/t). On a observé, outre la m atrice UC, un certain nombre de phases secon daires du type Ug Cg, U2 Ru Cg, UC^ Mo, Sr Cg, Cs I. IAEA-PL-463/6b 101 a - Matrice UC L'analyse à la microsonde s'étant révélée difficile, nous avons mesuré et calculé la variation du paramètre cristallin avec le taux de combustion et la teneur en carbone, en partant des hypothèses suivantes: - Zr et Y sont entièrement solubles'dans UC - Mo est partiellement soluble dans UC (1 % à 1600°C) - Les terres rares sont partiellement solubles, et la solubilité dépend du taux de carbone. - Tous les autres P. F. sont insolubles - A cause des faibles concentrations, il n'y a pas d'interaction entre P. F. pour le calcul des limites de solubilité et de la variation du para m ètre • ce dernier est d'ailleurs effectué à partir des courbes de varia tion de paramètre mesuré pour chaque P. F. La courbe 1 montre un excellent accord entre nos mesures et le calcul. On notera en p articu lier : - A 2 000° C, la solubilité du molybdène augmente beaucoup et on reste sur la partie 1 de la courbe jusquâ un taux de combustion important (/л/100 000 MWj/t. - Le taux et l'activité du carbone influent principalement sur les terres rares qui sont plus ou moins solubles : suivant l'activité du carbone, le paramètre varie entre la partie 3 de la courbe (ac /v^ 1), et la partie 2 pour laquelle les terres rares sont entièrement solubles (carbure sous-stoechio- m étrique). b - . Phases secondaires - UgCg : lorsque cette phase existe, elle ne contient pratiquement que de l'uranium, à l'exception de traces de terres rares (quelques pour-cent). - U„ Ru C2 : prévisible par l'étude du système U - Ru - C, cette phase très smble est toujours visible, sauf dans les composés sousrstoechiométri- ques ( C .¿1 1). Elle contient le rhodium (le composé U9 Eh C9 existe) ТГ ^ - UCgMo : cette phase est également très souvent présente mais dès que le rapport—^., devient inférieur à un "elle diminue en quantité en faveur de la phase voisine UC-^ y Mo. Le seul P. F. soluble dans cette phase est le rhénium, UCg Re ’ étant isom orphe de UCg Mo. - UPd3 : L'existence de cette phase est à priori prévisible (très grande stabilité (3) ) quoique jamais observée par analyse aux rayons X. Nous admettons donc l'existence d'une phase U (Pd, Rh, Ru)3 avec Pd^Rlr?Ru , mais de très faible importance. - Autres phases : nous noterons pour mémoire l 'existence des phases Sr Cg (Sr n'est pas soluble dans UC) et Cs I, toutes deux décelées par diffraction de rayons X. c - Influence du taux de carbone Il est évident que les différentes phases observées ne sont stables que pour une activité de carbone donnée, A savoir : - En présence de U9 C3 (ac^ 1), il existe U2 Ru C2 , UC2 Mo, Sr C2 Cs I, et pour un taux ae combustion suffisant, MC2 où M rep résen te les terres rares. A cause de la solubilité de ces dernières dans UC, la 102 LORENZELLI et MARCON F IG .l. Paramètre cristallin de UC en fonction du taux d'irradiation. IAEA-PL-463/ 6b 103 phase MCg n'apparaît-théoriquement qu'au-dessus de 55 000 MW j/t (figure 1 ), et elle est donc toujours en faible quantité. - Lorsque le taux de carbone diminue (en l'absence de Up Cg), il dis- • paraît d'abord la' phase SrCg, puis U2 Ru Со, avec apparition de UC-j^ y Mo. Pour un carbure sous-stoechiometrique(C < l)^ ц apparaît alorfe une phase métallique fondue contenant de l'uramum, du ruthénium, du molybdène et des terres rares. 3 - Çl^ssèment_des produits de_figsion . L'étude précédente nous a donc permis de classer chimique ment les différents produits de fission vis-à-vis du monocarbure UC, à savoir : a)- P. F. totalement solubles dans UC : Zr, Nb, Y. b) - P. F. partiellement solubles dans UC : Mo, terres rares c) - P. F. formant des carbures ternaires : UC2 M avec M = Mo, Re U 2 M'Ca avec M' = Ru, Rh, Pd, Mo, U 2 T eC 2. сЦ - P. F. formant des carbures insolubles MCg avec M = Ba, Sr M'C 2 avec M'= La, Ce, Nd, Pr (MC^ et M'Cg sont insolubles). 4 - P. F. formant des composés intermétalliques UMg avec M = Pd, Rh, Ru § - P. F. métalloïdes : Cs I, USb Se g)- P. F. restant métalliques Cs - Rb (en fait, ces deux métaux alcalins réagissent probablement avec l'oxygène dissous dans le carbure pour former un'uranate alcalin). B - Calcul de la variation de C en fonction du taux de combustion U Ce calcul nécessite â priori la connaissance des pourcen tages de produits de fissions formés après un certain taux de combustion, .ceci pour un spectre de neutrons déterminé et une puissance spécifique du combustible également donnée (ces deux derniers paramètres n'interve nant que faiblement). Le tableau I donne les quantités de P. F. libérés en p. p. m. atomiques, pour un combustible ayant atteint 100 000 MWj/t, avec une énergie moyenne de neutrons de 630 keV et une puissance spéci fique de 170 W/g (U + Pu). Si nous considérons donc un combustible comportant initialement de l'ura nium et du carbone (UC), après un certain taux de combustionT, la teneur en uranium sera passée à ( 1 -x), tandis qu'il sera apparu 2 xproduits de fission. - Si aucun de ces derniers ne réagissait avec le carbone, on libére ra it donc Tj atomes de carbone, donnant alors une composition UC]_ + En fait, nous avons vu que une partie des P. F. peut réagir avec le carbone, aussi, la quantité de carbone libérée sera inférieure àü , et elle peut même devenir négative. Nous voyons donc qu'il est indispensable de savoir quels sont les carbures ou alliages pouvant .se former, ceci en fonction de l'activité d'u carbone. • - ...... 0 LORENZELLI MARCON et 1 04 TABLEAU I. CALCUL DE LA QUANTITE DE CARBONE"CONSOMMEE ET LIBEREE AU COURS DE LA FISSION POUR UN TAUX DE COMBUSTION DE 100 000 MWj/t P . F . ppm atc)miques Composés probables Coefficients prévus C. consommé ppm at u R (U, Pu) C UC (u, Pu) c (U, Pu) C + + СО + (U, Pu)2 C3 i- u2 c 3 +(U, Pu)2 C3 со + (U, Pu)2 C3 Ru 14 050 U0 Ru Со U2R,\iC2 +P\iR u2 0 - 1 / 6 0 -2 3 5 0 Xe 21 250 gaz gaz 0 0 0 0 Zr 27 950 ZrC -ZrC 1 1 27 950 27 950 Cs 18 700 Cs ' Cs' 0 0 0 0 Mo 22 550 ]ÆoC + UCgMo MoC+(U, Pu)C'2Mo 1 1 22 550 22 550 T e rre s Ce 12 450 ' r a r e s : Nd 15 850 La 6 000 . P r 48 250/ 5 400 MC + £M 2 Cs MC + £ м 2 c 3 1 1 48 250 48 250 Y 4 400 Sm 2-350 Pm ^ 1 800 Pd 4 750 UPd3 (U, Pu)Pd3 -1/3 - 1/3 - 1600 - 1600 Ba 6 650 BaC 2 - BaC 2 2 2 13 300 13 300 Te ou Re 5 900 UReCg (U ,Pu) ReC 2 1 1 5 900 5 900 Rh 3 400 U2Rh C2 UoRhC9+Pu Rh9 0 - 1 / 6 0 - 550 Sr ' 8 100 SrC 2 ¿ Sr C2 ¿ 2 2 16 200 16 200 Kr 3 400 gaz gaz 0 0 0 0 I 1 600 C sl C sl 0 0 0 0 T e 2 950 U0 Te C9 U0 T e C9 0 0 0 0 Rb 2 300 ^ b 2 Rb ¿ 0 0 0 0 Nb 500 NbC NbC 1 1 500 500 Totaux 192 300 133 050 130 150 0. libéré : 96 150 IAEA-PL-463/6b 105 FIG. 2. Energie libre molaire partielle du carbone (RT log ac) pour différents systèmes m étal— carbone intervenant dans un combustible carbure irradié. A cet effet, nous avons porté les variations des énergies libres de formation des différents carbures en fonction de la température, en se ramenant à un atome de carbone (fig. 2). L'usage d'un tel diagramme doit être fait avec prudence, car, contrairement au cas des oxydes, il est ici nécessaire de tenir compte de l'entropie de mélange due à la mise en solution éventuelle d'un certain nombre de carbures dans UC, les dif férences d'énergie libre entre carbures étant assez faibles. 106 LORENZELLI et MARCON En fait, ce diagramme ne fait que confirmer les données obtenues à l'aide de la simulation : - Mise en solution de Zr C, Nb C et YC dans tous les cas, formation de Ba Cg et Sr Cg dans le cas où on est en présence de Ug Cg. - Dans le cas des terres rares (La, Ce, Nd, Pr) et du modybdène, nous voyons que l'on est situé entre les courbes U + UC- et UC + U2 C3 : l'état chimique de ces P. F. dépendra donc de l'activité du carbone de UC et de leur concentration : nous savons effectivement que pour un carbure très sous - stoechiométrique, une partie du Mo et des terres rares peu vent se trouver sous forme de composés intermétalliques. Ayant ainsi pu prévoir le comportement chimique de chaque P. F ., nous avons pu calculer l'évolution du taux de carbone en fonction du taux de combustion : pour cela, on affecte chaque P. F. d'un coefficient correspondant à sa consommation en carbone (tableau I). Sachant que l'on part à priori d'un combustible biphasé UC + Ug C3, ■ le calcul est aisé tant que l'on est en présence d' Ug Cg (a fixée, donc coefficients cons tants dans le tableau I). Nous constatons alors le fait très important suivant : - Le carbone consommé est supérieur au carbone libéré et la diminu tion du taux de carbone par fission n'est pas négligeable (tableau I ) . Ainsi, pour 1 0 0 0 0 0 MWj/t, on consomme 1 3 , 3 % at de carbone et on en libère 9, 6 %, ce qui correspond à une variation de composition de и с ^ о з ? à. U C ij 00 . En fait, la fission n'est pas la seule cause de diminution du taux de carbone : il est bien connu (4) qu’un combustible sous-stoechiom étrique (UC + U2 Cg) carbu re une gaine en a cier inoxyda ble; cette carburation est régie par deux facteurs : - La présence de Ug Cg - La cinétique de diffusion du carbone dans l'acier. La diminution du taux de carbone est liée à deux lois : - L'une linéaire en fonction du temps (fission "dêcarburante") - L'autre variant comme vHT~ (diffusion du carbone dans l'acier), ces deux phénomènes évoluant de façon continue tant que l'activité du carbone reste constante. Leur importance respective dépend à priori des conditions expérimentales d'irradiation (joint sodium, nature de l'acier, densité du carbure etc..) mais il est bon de constater qu'ils sont du même ordre de grandeur (diminution de quelques centièmes sur la teneur en carbone au bout d'un an). Lorsqu'il n'y a plus de UgC3, la diminution du taux de carbone s'accompagne alors d'une baisse de l'ac tivité du carbone et certains coefficients affectés aux différents P. F. (tableau I) changent,' un certain nombre de carbures devenant alors instables : Ba Cg, Sr Cg tout d'abord, puis MCg- (M = La, Ce, Nd, Pr, ) et Ug Ru Cg; Dans ces conditions,' la quantité de carbone consommé va diminuer fortement pour devenir égale à la quantité de carbone libéré. Nous atteignons alors un équilibre pour lequel le taux et l'activité du carbone sont tous les deux fixés (fig. 3 ) . Les valeurs correspondantes ■ IAEA-PL-463/6b 107 FIG.3, Variations’schématiques, en fonction du temps, du taux et de 1' activité du carbone dans un combustible de type carbure (UC ou (U, Pu)C) contenant initialement 10% en poids de M2 C3. (Les valeurs ne sont données qu'à titre indicatif.) sont respectivement UC^ qq et ac ru 0,005 à 1 000° K. On notera que cette dernière valeur^ sera alors probablement inférieure à l'activité du' carbone dans la gaine ( a 0, 01 à 1000° К pour un a cie r à 18 % Cr et 8 % de Ni contenant 3 000 ppm de carbone). Dans ce cas, la gaine cessera de consommer du carbone, l'équilibre thermodynamique étant alors atteint par l'ensemble gaine - combustible (fig 3). Rem arque Les résultats ci-dessus concernant la consommation de carbone par fission pour un combustible UC + Un C3 ont été confirmés dans le cas d'un com bustible (U, Pu)C + (U, Pu)«, C Q, Pu n „ MrY, ¿ ÜTPu “ °* Uü} 108 LORENZELLI et MARCON A l'aide d'une étude de simulation, nous avons observé le même comportement pour les différents P. F. que dans le cas de UC + Ug Cg, sauf le ruthénium et le rhodium qui forment une phase très minoritaire M (Ru, Rh)g avec M = Pu, La, Ce, Nd. A partir de ces résultats, nous avons calculé une évolution du taux de carbone pour (U, Pu) C + (U, Pu)g Cg en fonction du taux de combustion toutà fait comparable à UC + Ug C3 (tableau J} : au bout de 100 000 MWj/t,, la composition passe de MC^ 03 à MC^ q q . II - MESURE DE L'ACTIVITE DU CARBONE EN FONCTION DU RAPPO RT _C DANS UN COM BUSTIBLE "SIM ULE"OU IRRADIE U A - Technioues expérimentales Nous avons utilisé une méthode de vaporisation avec analyse par spectrométrie de masse qui nous permettait de déterminer les activi tés de tous les constituants. Le seul inconvénient de cette méthode est la température à laquelle s'effectue la mesure (T> 1500°C), sensiblement plus élevée que celle à laquelle le combustible est utilisé. L'appareil employé a été décrit par ailleurs ( 5 ). _ La mesure directe de la pression partielle de carbone LC (g) ou Cg (gjj étant très difficile, nous avons utilisé la présence de dicarbures gazeux MC2 pour déterminer l'activité du carbone en utilisant la relation : I FMC M(g) + 2 C (g) ------»• MC9 (g) avec a„ = K' J ¿ K m Cette formule nécessite la connaissance de K' (ou de l'énergie libre de formation de MCg (g) ) et un étalonnage préalable pour obtenir РМсг à partir de 1MC2 (courant ionique observé). En fait, une seule expérience est nécessaire si on effectue une mesure de M (g) et MC^ £f) au-dessus d'un mélange MC9 (s) + C (pour lequel ac =1). On aura alors : PiMC 2 PM ac ) X (------) p M obs p MC 2 M C 2 + C 1 MC? X ^ M pour une ( - — ) h ( ------) I M obs r Mc2 MC2 +c température donnée. Nous avons effectué l'étalonnage pour LaCg + C et UCg + C, en utilisant surtout la m esure de ^ ^LaJ ^UC / I 1j servant contrô-Le à très haute température (I-^ac est très supérieur à 1^ pour ac donné). IAEA“ PL-463/6b 109 В - Résultats Nous avons effectué deux séries d'expériences sur un com bustible simulant un taux de combustion de 100 000 MWj/t : - la première correspondait à un carbure UC légèrement sous - stoe- chiomêtrique (0, 95 - la seconde correspondait à un carbure sur - stoechiométrique (1, 05 < M/C< 1, 10) contenant environ 15 % de U 2 C 3, ain si que les phases secondaires de type UgRuCr,, UC^Mo et MC2 (M = La, Ce, Nd). Dans les deux cas, les quantités relatives de P. F. introduits sont les suivantes : Zr : 3, 1 % at Nd: 2, 1 % (correspondant à Nd + Pr) Mo : 2, 4 % at Ce : 1, 64 % ( " r Ce + Sm + Pm) Ru : 1, 4 % at La : 0, 6 % Rh : 0, 34 % at Y : 0, 44 % Pd : 0, 47 % at Re : 0, 59 % ( ” 1 1 Te) 1 - Activité du carbone et de l'uranium. La figure 4 montre la variation en fonction de 1/T de la pression partielle d'uranium pour deux types d'échantillons (indices 1 et 2) tandis que la figure 5 indique la variation de l'activité de carbone dans le cas du carbure sur-stoechiométrique (noté UC, ^¡. La mesure de l'activité du carbone pour UC- (cas n ]\ a été tr è s difficile à cause de la faible intensité de IT (a < 0, 02 à 2100“K). LaCg c^ ’ ' Nous l'avons cependant estimée égale à 0, 012 à 2100°K par comparaison avec les activités de carbone au-dessus des domaines U + UC (a = 0, 007 à 2100°K) (6) et Mo + MogC (a = 0, 015 à 2300°K), la présence de la phase Ug Mo impliquant une activité de carbone compa rable à celle observée dans ces domaines. Dans ces conditions, nous avons pu comparer les activités de carbone et d'uranium, â savoir : - Dans le cas UC^ , pour une activité de carbone très faible corres pondant à la présence de métal, on observe une activité d'uranium deux fois plus faible que celle du métal saturé en carbone (U + UC). Ce fait indique simplement que l'activité de l'uranium liquide est abaissée par la présence de molybdène, ce qui est très compatible avec la présence à froid de la phase Ug Mo. - Dans le cas de UC, , l'activité de carbone jusqu'à 1850°C est difficile à mesurer et voisine de 0, 7. Au-dessus de cette température on observe une variation importante de a avec la température, les valeurs obtenues étant strictement égales à celles mesurées par Storm s (6)'et Tetenbaunr (7) dans le domaine UC + UCg. Ce point est très important, car ceci indique que les activités de carbone dans les domaines UC + UC2 d'une part, MC + UCg + M'Cg (M' = La, Ce, Rh) + по LORENZELLI et MARCON FIG.4. Pressions partielles des différents éléments mesurés dans UC sous-stoechiométrique (1) et sur-stoechiométrique (2). + UCgMo + Ug Ru C2 d'autre part sont les mêmes, ce qui veut dire que la présence de produits de fission dissous dans MC (Zr, Mo et terres rares) ne modifient que faiblement la stabilité thermodynamique de cette phase. 2 - Activité des terres rares et de l'yttrium Nous nous sommes particulièrement intéressés à la mesure de l'activité des terres rares, qui,comme nous le verrons,dépend beaucoup IAEA-PL-463/6b 111 FIG. 5. Variation de 1' activité du carbone au-dessus de UC^ + X en fonction de 1/T. de l'activité du carbone. Ainsi, la figure 4 montre une variation énorme de l'activité de La, Ce et Nd entre UC., et UC1 ’ 1 -x 1 + x - Dans le cas de UC., , on observe, à basse température (T< 1500°C), pour La et Ce une activité voisine de celle du métal (aQe = 0, 7 et a-^a = 0, 25) correspondant à la présence d'une phase métallique liquide. Au-dessus de 1500°C, on observe pour La, Ce et Nd une chaleur de vaporisation très inférieure à celle du métal, qui peut s'expliquer par un appauvrissement de la surface de l'échantillon en terre rare au- dessus de 1500°C, la vitesse de diffusion étant alors inférieure à la vi tesse d'évaporation. - Dans le cas de UC]_+X-, on observe à t = 1850°C une brisure très forte dans la courbe log p = f 4 .) correspondant à la variation de l'activité de carbone déjà signaléeT Nous avons tenté de relier les activités observées de La, Ce et Nd- à la présence d'une phase MCg contenant principalement ces trois éléments, comme l'étude des diagrammes U-M-C (avec M = La, Ce, Nd) le laisse supposer (1). L'activité de chaque dicarbure solide dans cette phase peut être calculée à partir de la relation : ac= l log a L a C 2 = log PLa - log PLa + 2 log ac 3 . — 1 PT c étant mesurée au-dessus du domaine-LaC„+C. 1 1 2 LORENZELLI et MARCON Nous obtenons dans ces conditions (en utilisant pour a = 1 a =1 c c P„ et Path les données de la littérature (8)): Ce eL p Nd = 0,7) (ac со log 0 0 a L aC 2 == , log = 0 25 aC eC 2 = , log = 0 23 aNdCg , - pour t 1850°C (log a£ = 9700/T - 4,70), on calcule pour ‘ aLaCg, aCeCg et aNdCg les mêmes valeurs que ci-dessus, aux erreurs d'expérience près. Ce résultat est très intéressant, car il m ontre : - que la variation des activités de La, Ce et Nd en fonction de l'activité de carbone est uniquement compatible avec la vaporisation d'un diearbure. - que les activités de LaCg , CeCg et NdC^ dans la phase MC^ sont en bon accord avec les données des diagrammes U-M-C. Compte tenu des solubilités des terres rares dans UC, la phase MCg devrait théoriquement contenir 59 % de Nd, 35 % de Ce et 6 % de La, qui correspondraient donc à des activités respectives de 23, 25 et 8 %. Ces résultats confirment donc la présence d'une phase MCp (M = La, Ce, Nd) dans le carbure simulé. A partir de ces résultats, il paraît clairement que la mesure directe de l'activité d'une terre rare permet, connaissant le taux de combustion (c'est-à-dire la concentration de la dite terre rare), de cal culer approximativement l'activité du carbone. Or, pour l'instant, les mesures directes d'activité de carbone sur combustible irradié n'exis tent pas. Mais une mesure de vaporisation sur des combustibles irradiés par une méthode de Knudsen simple (détermination des masses évaporées par spectrométrie y sur les cibles recueillant la vapeur (9)) permet la mesure de Pcg , donc de ace grâce au comptage de l'isotopie 144q 6 . CONCLUSION Il ressort de cette étude deux conclusions importantes : - La fission d'un combustible de type carbure (UC ou (U, Pu) C) consom me du carbone tant qu'il existe un carbure supérieur (M2 C3). IAEA- PL-463/6b 1 - La dissolution des P. F. solubles (Zr, Nb, Mo, La, Ce, Nd, Y) dans la matrice UC ne modifie pas sensiblement la stabilité thermodynamique de celle-ci (activité de carbone et énergie libre de formation). On notera que ces deux résultats sont exactement opposés â ce que l'on observe dans les oxydes irradiés. REFERENCES [1] LORENZELLI, N ., MARCON, J.P ., J. Less-Common Met. 26 1 (1971) 71. [2 ] HOLLECK, H ., KLEYKAMP, H ., J. Nucl. Mater. 32 (1969) 1. [3] MARCON, J.P ., Rapport interne non publié. [4] LORENZELLI, R ., DELAROCHE, P ., HOUSSEAU, M ., PETIT, P ., 4e Conférence Internationale sur le Plutonium (Santa Fe, Oct. 1970). [5] LORENZELLI, N., MARCON, J.P., J. Nucl. Mater. 44 (1972). [6] STORMS, E .K ., The Refractory Carbides, Academic Press, New York (1967). [7 ] TETENBAUM, M ., HUNT, P .D ., J. Nucl. Mater. 40 (1970) 104. [8] FAIRCLOTH, R .L ., et a l . , J. ïnorg. Nucl. Chem. _30 (1968) 499. [9] HALACHMY, M ., Thèsé'de Doctorat, Université de Paris (1971). IAEA-PL-463/ 7 SOM E PHASE EQUILIBRIA AND THERM ODYNAM IC CONSIDERATIONS FOR IRRADIATED OXIDE NUCLEAR FUELS P.E. POTTER* European Institute for Transuranium Elements, EURATOM, Karlsruhe Abstract SOME PHASE EQUILIBRIA AND THERMODYNAMIC CONSIDERATIONS FOR IRRADIATED OXIDE NUCLEAR FUELS. The paper considers the recent published data on the phase equilibria and thermodynamics of the appropriate oxide systems. First, the binary systems, uranium-oxygen and plutonium-oxygen and the ternary system uranium-plutonium-oxygen and, secondly, the systems of the major fission products with uranium- oxygen and plutonium-oxygen are considered. 1. THE PHASE EQUILIBRIA AND THERMODYNAMICS OF THE URANIUM-PLUTONIUM-OXYGEN SYSTEMS The systems U-O, Pu-О and U-Pu-O will be discussed first. 1.1. The uranium-oxygen system Several papers reporting measurements of oxygen potentials of hypostoichiometric UO 2-X have appeared recently. These studies cover the temperature ranges of 1873 - 2173°K [ 1] using a transpiration technique with H2O/H2 mixtures; of 1800 - 2000°K [ 2 ] using an oxygen potential control technique by means of the reaction 2 С + O2 2 CO; of 2200 - 2 400°K [ 3 ] using a static system with controlled H2O/H2 ratios; and finally of 2080 - 2705°K [ 4] using a transpiration technique with H20/H2, mixtures. All the data for the oxygen potentials as a function of composition and temperature are shown in Fig. 1. Tetenbaum and Hunt [ 5] extended their measurements of oxygen potential in the hypostoichiometric U 02.x region to include measurements of the total pressure of U-bearing species in the same temperature range as their previous oxygen potential measurements. These pressures as a function of composition are shown in Fig. 2 and the isotherms show a minimum in the total pressure in agreement with the m ass-spectrom etric data; the position of the .minimum is 0 . 01 - 0 . 02 atomic ratio units higher than the composition calculated by Edwards et al. [ 6 ]. The values of Pattoret et al. [7] of O/U = 1. 987 at 2250°K and Ackermann et al. [ 8 ] of O/U = 1. 994 at 2000°K are in good agreement with the data of Tetenbaum and Hunt. * On leave of absence from the Atomic Energy Research Establishment, Research Group, UKAEA, Harwell, Didcot, Berks, England. 115 116 POTTER °/u RATIO i 1 i 1 I 1900 *C- 1800' » 1600"C I7 0 0 -C - (b) о 1700 “C I600°C • 1800“C о 1900°C — FROM PERRON11 — PRESENT STUDY Q> О E о Q. _ U-UOj /UOj .,, PHASE BOUNDARY О A FTER ACKERMANN el at [12] Ю < O/U IAEA-PL-463/.7 117 U02. M"UC2 □ jbo* •¿о ъ - ио2 -О ( с ! i ; ■¿So* cP О -о , 1 ■ UC2 U O 2 + C / U C 2 PHASE BOUNDARY isooK Soo* PRESENT WORK. 1900 К CLOSED SYMBOLS INDICATE UC2 2 0 0 0 К FORMATION. EXTRAPOLATED FROM TETENBAUM t H U N T 2.00 °/l RATIO F IG .l. Oxygen potentials for the uranium-oxygen system: (a) variation of log P^Q) with 0/U ratio for U02-x [3] ; (b) variation of relative partial molar free energy of oxygen with temperature and 0/U ratio for U02±x Cl] J (c ) variation of equilibrium oxygfen pressure (expressed as log P q ) with 0/U ratio at various temperatures [2 ]. 0/U ATOM RATIO FIG.2. Total pressure of uranium-bearing species as a function of temperature and urania composition [5 ]. 118 POTTER idus I ^ / и / / / j j f . . - 4 • / / l ^ l \ ' T o r / / X Upp*r ^ Soli us ^ 1/ 5 / V / ° / ■ 7 / ' / ' 0/ // / / Y / ' > / / JT \\ ' / / / / 0 ' / / £ / 4 Legend: / / ' / 0 0 W capsules / у / Û A Re capsules • ------Be mister — * • — Ba es 1 Present Work \ ^Lower y r Solidui --- ;------Ed wards and Martin \ u + u o 2_„ \ \ \ 1.4 1.5 1.7 1.8 1.9 2.0 2.2 2.3 Oxygen-to-Uranium Ratio FIG.3. Partial phase diagram for mania from UOU 6 to TO2.23 [1 2 ]. A knowledge of the gas-phase pressures for oxide systems at tempera tures above the melting point is of importance in reactor safety analysis and Reedy and Chasanov [ 9] have recently measured the total pressure of the uranium-bearing species using a transpiration technique with tungsten as the containment material up to a temperature of 3390°K. The data obtained log p (atm) = - 2-7^ 2 6- + 7. 373 are in good agreement with the extrapolated data of Tetenbaum and Hunt corrected for the liquid region using a recent value of 17. 7 kcal/mole [ 10] for the heat of fusion of UO 2, which is in good agreement with an earlier value of 18.2 kcal/ mole [ 1 1 ]. A condensed phase diagram for the region, O/U = 1. 5 to 2. 2, is shown in Fig. 3 [ 12]. TEMPERATURE^) I.. tnaie ltnu-xgnpaedarm 13]. 3 [1 diagram phase plutonium-oxygen tentative A FIG.4. I 5 A ltnu-xgn hs iga [ ]. 4 [1 diagram phase plutonium-oxygen A .5. FIG AAP-6/ 7 IAEA-PL-463/ -0 u P 119 120 POTTER 1.2. The plutonium-oxygen phase diagram A tentative phase diagram was published in 1966 as a result of the IAEA assessment of the available data [13]. This condensed phase diagram is shown in Fig. 4. The main features, which have been further considerated by Sari et al. [14], a re (a) The continuous solid solution between PuO^ei and Pu02.x above 650°C (b) The stoichiometry of the 'PuOi.6i' phase (c) The maximum oxygen content of the PuC>2-x phase in the presence of PuO i. 5 . The phase diagram suggested from this work in which the samples of different О/Pu ratios were quenched from different temperatures down to 5°C is shown in Fig. 5. The main features of this diagram are (a) P u 0 1-61 and P u 0 2 do not form a continuous solid solution, at least up to about 1000°C (b) The 'PuOi.ei' has a stoichiometry lower limit of О/Pu = 1. 62 5 and an upper lim it of О/Pu = 1. 69. The bcc structure of the C'-phase (Fig. 4) has been previously discussed [ 13] and it was concluded that the О/Pu ratio should be 1. 625 and not 1. 61, and 4 extra oxygen atoms would be statistically distributed on the 16 ordered oxygen vacancies of the bixbite unit cell of the bcc Pu0 1>5 low temperature, С-type rare earth oxide. Sari et al. stated that it was possible to have a statistical distribution on the 16 vacancies of the bixbite cell if one accommodates 2 more О atoms, giving О/Pu = 1. 6875, which is in good agreement with the experimental value for the upper phase limit of the C '-p h a se . The lattice parameters of the C'-phase increase with decreasing oxygen content and were in agreement with the appropriate values extrapolated from 700 - 900°C [ 13, 15]; for O/M = 1. 50, a = 11. 08 A. Finally, Sari et al. replotted the emf data of Markin and Rand [ 16], which gave the partial molal free energy of oxygen (AG0z) as a function of О/Pu ratio. These curves are shown in Fig. 6 and show the possible existence of a plateau indicating a univariant region (two condensed phases in a 2-component system) — C' + Pu02.Xz. The lattice parameter measure ments of Gardner et al. [ 15] also show irregularities at 700 and 900°C for О/Pu ratios of 1. 69 to 1. 72, i. e. in the univariant region. The two-phase region appears to become wider with increase in temperature. The upper lim it of the Pu 0 2 -Xl phase in the presence of С was found to be PuOi_g95, and not PuOj_98 as previously given. The lattice parameter of Pu0 2 is 5. 3952 ± 0.0005 A. Some observations by Marcon et al. [17] on the carbothermic reduction of PuC>2 in the region between Pu0 2 and P u 0 l t 5 have enabled some information on the Pu-О phase diagram to be obtained in the temperature range 1400 - 1800°C. The gas pressures, mainly CO, were measured as a function of the extent of reaction. If a phase diagram of the form shown in Fig. 4 were appropriate, then the С + Pu02_x (x = 0 0. 4) would be bivariant and the pressure of CO would fall continuously during the reduction, but two sharp breaks were observed in the curve (Fig. 7), which suggests a more complicated diagram of the type considered by Sari et al. IAEA-PL-463/ 7 121 ______0_ Pu FIG.6. AG (oxygen) versus О/Pu [1 3 ]. The region marked by C ’ gives the а -single phase region [1 4 ]. The various phase fields assumed to be present during the reduction process at 1400°C were: ( 1 ) Pu02.x (fee) — carbon with 0 S x S 0,28 b iv a ria n t (2 ) P uO i.72 (fee) — PuOi.69 (bee) — carbon monovariant (3) РиОг-у-carbon with 0. 31 S y S 0. 42 bivariant (4) P u 0 158 (bcc) — P u 0 1-5(i1ex )" c a r ':|on monovariant The limits appear to vary strongly with temperature above 1400°C and the bcc phase (C, Fig. 5, and cc, Fig. 8) disappears above 1700°C (Fig. 8 ). Dean et al. [ 18] measured the partial heat of solution of oxygen (AH0z) in Pu02.x using a Calvet microcalorimeter at 1100°C. A very sharp change in curvature was observed at О/Pu = 1. 70, and a very narrow 122 POTTER Pu FIG.7. The carbothermic reduction of Pu02 [1 7 ]. FIG.8. A plutonium-oxygen phase diagram [1 7 ]. IAEA-PL-463/ 7 123 2-phase region could be assumed to exist. The oxygen pressure measure ments of Riley [19] in the region 1. 6 < O/Pu <2.0 at temperatures > 700°C gave no information on the presence of a univariant region. 1.3. U-Pu-O sy stem The phase equilibria and thermodynamics of the system have been the subject of a previous IAEA Vienna Panel Meeting in 1966 [ 13]. It is , however, considered useful to survey briefly the additional work on these systems undertaken since the Panel Report was written. The studies can be divided into two sections, namely (1) phase diagrams (including melting points); and (2 ) thermodynamic measurements (excluding vaporization studies). Because of the high temperature gradients in the oxide fuels, thermodynamic data are required in the temperature range between 600 and 3000°K. It is essential that the thermodynamic functions of this system are well defined in order that models developed to describe the chemical behaviour of the complete multicomponent system during 'burn-up' are on a sound basis. 1.3.1. The phase diagrams Any consideration of the phase relationships of this system must be consistent with the binary phase diagrams. There have been two further phase diagram studies on the U-Pu-O system in the hypostoichiometric region of the diagram since the 1966 Vienna panel. Sari et al. [20] have studied the system for Pu/U + Pu ratios of 0. 05 - 0. 97 from stoichiometric to the fully reduced composition, i. e. until the appearance of a metal phase. Dean et al. [ 18] have also studied the hypostoichiometric region of the phase for Pu/U + Pu ratios between 0. 40 and 0. 76 and have suggested the presence of a M7O i2-type phase with a rhombohedral structure between Pu/U + Pu ratios of 0. 60 and 0. 90. Additional studies have also been reported [ 14, 21] for the hyper- stoichiometric region of the diagram. The results of Sari et al. [20] are shown on the room temperature isothermal section (Fig. 9). It can be seen that up to Pu/U + Pu ratios of about 0.20, there will be a range of FIG. 9. Isothermal section at room temperature of the U-Pu-O diagram between O/M = 1.5 and O/M = 2 .0 0 : • single phase fee; A single phase bec; X two fee phases; □ fee 4-bcc; fee tm etal; D bcc+m etal; ® hex. Pu20 3 [2 0 ]. 124 POTTER fluorite structure whose amount of oxygen vacancies depends on the Pu valency (from 4 to 3). The previous diagram [22] suggested that this single-phase fluorite region existed up to Pu/U + Pu ratios of about 0. 30 at room temperature, although this difference may simply be the result of different cooling rates. The area of single-phase fluorite structure de pends on the temperature, increasing with temperature [22] (see Fig. 10). The results shown in Fig. 9 were obtained with samples slowly cooled to room temperature, annealed at 200°C and then again slowly cooled. The hypostoichiometric mixed oxides were obtained by partial reduction with hydrogen containing known amounts of water. For complete reduction the experiments were carried out in pure hydrogen. For Pu/Pu+U ra tio s greater than about 0. 2 there are two fee phases, one containing 4-valent Pu and the second probably containing fully reduced 3-valent Pu. The work of Koizumi and Nakamura [23] confirmed the room-temperature observations of Sari et al. [20] that the two-phase area extends to Pu/U + Pu ratios of about 0 . 2 0 . A region of transition from fee to bcc exists between 45 and 50% Pu0 2 (the shaded area in Fig. 9). There is a large region of bcc single-phase oxide for the region with Pu/U + Pu ratios between 0. 5 and 0. 97. The mixed oxides reduced in pure hydrogen at 1800°C contained a metallic phase when allowed to cool in vacuum; under these conditions it would be expected • that the U4+ would be reduced, as in the binary system. The metallic phase is in equilibrium with a fee or a bcc phase and probably also with the hexagonal phase as in the Pu-О system. The lattice parameters of all the phases are shown in Fig. 11. The region in which metal and fully reduced oxide are present will be discussed later. The results reported by Dean at al. [ 18] for the Pu/U + Pu ratios of 0. 4 to 0. 7 indicate that up to 0. 6 for this ratio there are two fee phases present and above 0 . 6 the two phases present are a fee phase and a rhombohedral phase in place of the bcc phase found by Sari et al. The lattice parameters for the hypostoichiometric region of the U-Pu-O sy stem IAEA-PL-463/ .7 125 Pu U+Pu FIG. 11. Cubic lattice parameters plotted as a function of plutonium content. Bcc parameters are plotted at half their value, — — limit of cubic single phase region------Vegard lines [2 0 ]. given by Dean et al. are shown in Fig. 12. A super-lattice was present with a lattice parameter twice that shown in Fig. 12; the lattice parameter of this rhombohedral phase varied little between 80 and 90 at.% Pu and the angle a remained fair.ly constant and close to 89. 5°. For Pu/U + Pu ratios greater than 0. 9 it was assumed that a two-phase region exists containing a bcc and a fee phase. Figure 12 again shows a deviation from Vegard's law for the 126 POTTER Pu U+Pu □ H2 REDUCTION {□ fcc or bcc в RHOMBOHEDRAL) О МО2 * С a U О2* ^ Pu * U О2 x Ref. Г31 FIG. 12. Lattice parameters for the U-Pu-O system [ 18]. FIG .13. Limits of the two-phase region determined by DTA [2 0 ]. Pu/(U+Pu) ratios are: (1) 1; (2) 0.95; (3) 0 . 8 ; (4) 0.58 [22]; (5) 0.58; (6) 0.42 [22]. IAEA-PL-463/ 7 127 FIG. 15. Pseudo binary diagram U-Pu-O at Pu/U +Pu = 0.76 [1 8 ]. 1 2 8 POTTER fully reduced oxide in the presence of a carbide phase — the oxide was reduced with carbon. The carbide is probably a dicarbide or sesquicarbide. Some high temperature studies have been made using both DTA [ 20] and high temperature X-ray diffraction [18]. The limits of the immiscibility gap as a function of temperature and Pu concentration were determined between 400 and 650°C [20] and sim ilar behaviour found to that found earlier, namely a decrease in critical temperature with decreasing Pu content (Fig. 13). A section for Pu/U + Pu = 0. 9 7 obtained from the DTA measurements [ 20] is given in Fig. 14. However, the high temperature X -ra y dilatometry and electrical resistivity data [ 18] in the temperature range 400 - 700°C for Pu/Pu + U = 0. 76 suggest a rather more complicated section than that at Pu/Pu + U = 0. 97 (Fig. 15). The fact that for two-phase systems the lattice parameters of the more oxidized phase lie close to the Vegard's law line between U0 2 and P u 0 2 indicates that the Pu concentrations in both phases are little different. The region of the U-Pu-O system containing the two-phases U-Pu metal plus the fully reduced oxide, a solid solution between U0 2 and PuO i.5, is now considered. The equilibria in this two-phase field can be calculated by assuming that both the U-Pu liquid alloy and the oxide solid solution are both ideal solutions. F o r the region Ui-Xl PuXl (liquid) + Ui_X2 PuX20 2-o.5x2 (solid) the oxygen potential for the two-reactions U (dissolved in Pu) + 0 2 (gas) ^ U 0 2(dissolved in PuO i.5) and 3 Pu (dissolved in U) + — 0 2 (gas) ■=^Pu01 -5 (dissolved in U 02) must be the same. The oxygen potentials are given by where AGuq2 dissolved etc. are the partial molal free energies of the components, and for ideal solutions AGuoz - AGfuo2(soiid) + RT l n ( l- x 2) AGpu 0 l,5 = ACtf pu0l 6 (solid) + ln x 2 AG и = RT In (l-xj) AGPu = RT In x j where AG® are the standard free energies of formation, and equating the two expressions for RT In p0z 1-X2 4 AG?Pu0l, - 3 AGj 0 o 2 = 3 RT In + 4 RT In — ( 1 -Xl. x 2 IAEA-PL-463/ 7 129 U02 PuO,5 UO2 PuOl.5 FIG.16. Equilibria in the phase field Ui.Xj PuXl (liquid) + Ui_Xz PuX2O2- 0.5x2 so^ [23]. Pu (MOLE V.) FIG.17. Lattice parameters of ’fully-reduced* uranium-plutonium oxides [13]. 130 POTTER U UCi 5 Pu c1.5 FIG.18. Equilibria in the phase field PuX2 O2- ».sx2 (solid) + Ui-X3 PuXj C^s (solid) [24]. The results of some computer calculations [24] for 1600 and 2000°K are shown in Fig. 16. The metal phase will always be richer in uranium than the oxide solid solution. The phase structure of the fully reduced U0 2 - P u 0 1 5 solid solution and the lattice parameters of this phase were discussed previously at the Vienna Panel; there was a great deal of difference in the lattice param eters depending on the method of preparation (Fig. 17). Three methods of preparation gave different lattice parameters for solid solutions whose compositions were stated to lie on the line joining U0 2 and PuOx.5 . The alloys prepared by reduction in hydrogen were quite close to the values expected for a pseudo-binary section; however, when the reduction was carried out using metal or carbon the lattice parameters were considerably higher than those of the hydrogen reduction. The data of Sari et al. [20] in the presence of metal and those of Dean et al. [18] in the presence of, probably, sesquicarbide show a sim ilar deviation. In the presence of metal the oxide solid solution will contain a greater amount of Pu than the nominal composition and so the lattice param eters will be closer to the Vegard's law line. F o r the oxides reduced with carbon and described as 'in the presence' of carbon, and if the second phase is sesqui carbide, the sesquicarbide will be richer in uranium than the oxide phase and thus the same interpretation as given for the oxide in the presence of metal would be applicable to explain the high lattice param eters. The equilibria for the region U1.Xz Pux20 2 _0-5x2 (solid) + U j.^ P u ^ C ^ isolid ) is given by Some calculated equilibria for this region are shown in Fig. 18 for 1800°K. Finally, in considering the phase diagram, results for the hyper- stoichiometric region reported by Benedict and Sari [21] are shown together with the results for the hypostoichiometric region in Fig. 19. The liquidus-solidus of the U-Pu-O system. Because of the redistribution of both oxygen and uranium and plutonium in a fast reactor fuel operating under conditions where the centre temperatures are very close to the IAEA-PL-463/ 7 131 FIG.20. Solid-liquid phase diagram for the иОг-Pu02 system [25]. melting points of the oxide or indeed sometimes exceed the melting temperature, it is important to have information on the effect of stoichiometry and plutonium-concentrations on the solidus-liquidus tem peratures. The liquidus-solidus of the U-Pu-O system reported to the previous Vienna Panel for the stoichiometric solid solution [25] are re produced in Fig. 20. The addition of Pù lowers the liquidus and solidus temperatures. Some measurements of the effect of stoichiometry on the liquidus- solidus of the system have been reported by Aitken et al. [ 26] . The liquidus temperature changes little and the difference in solidus temperature when the O/M ratio changes from 2. 00 to 1. 92 is about 80°C for Pu/U+Pu ratios of 0 . 2 . 132 POTTER 1.95 2 ,0 0 2.05 2.10 °/M RATIO FIG .21. Partial pressures over U0_85 Pu0.i5 0 2±y at 2000°K (calculated) [28]. TABLE I. VAPORIZATION EXPERIMENTS ON U-Pu OXIDES Temperature Composition Reference range Method (initial) ( ° K ) [ 2 9 ] Uq.85 PUq.15 O 2-X 1 8 0 0 - 2 3 5 0 Knudsen effusion target collection and x = 0 to 0 . 06 mass-spectrometry [ 3 0 ] U q.8 P u 0.2 ^ 2 - X 1 9 0 5 - 2 4 1 1 Knudsen effusion mass loss, and mass- x " 0 to 0 , 0 0 8 spectrometry [ 1 8 ] U o.5 Рц о.5 O i-x 1 8 1 4 - 2 2 2 0 Knudsen effusion and target Uo.76 Pu0.24 O2-X collection [ 1 3 ] U o .l P u 0-9 O 2 - X 1 8 0 0 - 2 5 0 0 Knudsen effusion and target U 0.2 ^ u 0.8 O 2 -X collection IAEA-PL-463/ 7 133 1. 3. 2. The thermodynamic data for the U-Pu-О system Using the extrapolated oxygen potential data of Markin and Mclver [27] reported in the oxide panel meeting, together with the thermodynamic data for the gas phase molecules U03, U 02, UO, U and Pu02, PuO, Pu obtained from vapour pressure measurements of the metal or of the appropriate binary systems Rand and Markin [28] have calculated the individual species gas pressures and the total pressure above the U-Pu-O system as a function of O/U+Pu for a given Pu/U+Pu ratio. Some results of the calculations are shown in Fig, 21 for 2000°K and the total'pressure passes through a minimum at ~ M O it967; this, however, is not associated with a congruently vaporizing composition, nor with a constant vapour composition. These calculations and experimental measurements on this system, which will be discussed below, are of great significance to the redistribution of U and Pu via the gas phase in a fuel element. Experimental measurements on the vaporization of the U-Pu-O system have been reported by Ohse and Olson [ 29] , Battles et al. [ 30] and Dean et al, [1 8 ]. Some earlier work at Fontenay-aux-Roses on the vaporization of U-Pu-O was reported at the Vienna Panel [ 13]. The conditions of the various vaporization experiments are shown in Table I. Ohse and Olson [ 29] heated the samples of mixed oxides in tungsten effusion cells using an electron beam furnace. The technique employed allows the determination of partial pressures of the gas-phase components Time of heat treatement t (h) ------СП— FIG. 22. Rate of change of composition O/M of (U 0.g5 Pu0il 5)O 2±y at constant temperature in UHV [2 9 ]. 134 POTTER by simultaneous application of the effusion collection technique and m ass- spectrometric analysis. The relative abundances of PuO, Pu02, U03 and UO2 in the vapour above UQ-85 Puo.isOî-x» whose x ranged from 0 to 0.06, were determined. These workers took great care in avoiding the inherent problems of fragmentation, particularly that of UO3 , by using electron energies below the fragmentation potentials and by use of direct calibration and obtained the relationship (e.g. for U02) Iu o2 • T - = Puo2 ‘ C UOz where С is a specific ion sensitivity factor. All the specific ion sensitivity factors for the various gas-phase species were determined, and thus all the partial pressures could be determined at any composition by measure ment of the ion currents. Figure 22 shows the isothermal rate of the composition change of the solid solution U0 ,85Pu0. i 5 0 2±x as a function of time. A quasi-congruent evaporating composition close to O/M = 1. 97 is approached from both sides of the composition range. The partial pressures of Pu0 2 , PuO, UO3 , and UO2 over Uo.85Puo.15Ox.969 in the temperature range from 2000-2350°K are given in Fig. 23. From these relationships there were obtained the second-law partial enthalpy and entropy changes for this quasi- congruently evaporating composition of O/M = 1. 969, assuming the change in Cp to be negligible over the measured temperature range. Because of the rapid change of composition at both extremities of the substoichiometric single-phase region (Fig. 21), log(I + ,T) or logp¡ versus 1/T measurements cannot necessarily be interpreted as the partial enthalpy values. 1.94 1.96 1.98 2.00 0 / и FIG .2 3 . Partial pressure - composition diagram of U 0 3l U 02 , Pu02 , PuO and О over (U 0.g5 PLl0-i5)O2_y at 2103°K [ 29]. IAEA-PL-463/ 7 135 Finally, using the modified value for AGf(Pu0 j namely AGf° = - 102. 700 + 3. 16 T cal/mole may be compared with the previous result AGf = - 113. 100 + 4. 35 T cal/mole obtained from the earlier assessment of Ackermann et al. [ 31], whilst the new value was obtained from the more recent measurements of Ohse and Ciani [32] of the Pu-О system. Together with the pressure data for the gas species, Ohse and Olson obtained very good agreement for the oxygen potential of the system at 2108°K, namely - 117 to - 118 kcal • mole Og1 from both the reactions 2 (PuO) + (02) ^2(PuOz) and 2(UOz) + ( 0 2 ) ^ 2 ( U 0 3 ) The oxygen potential is also in good agreement with the predictions assuming ideal solid solution in the solid. The investigations of Battles et al. [ 30] are comparable, although the Pu concentration was slightly higher — the compositions examined were Uo.8 Puo.20 2-x with 1. 92 á O/M á 2. 00. The Knudsen effusion cells used in the study were made of W, Re and Ir. Vapour species typical of the vaporization of W or Re oxides were observed when the mixed oxide (O/M ~ 2. 0) was heated at about 1200°C or higher in W or Re effusion cells. Because of this incompatibility of the oxide with W and Re, iridium cells were used. The ion intensities of UO + , U 0 2, UO3 , PuO+ and P u 0 2 were measured as in the experiments of Ohse and Olson, and these were the only gas-phase species detected. The ion intensity of UO3 was determined using an ionizing electron energy of 15 eV, which is greater than the fragmentation potential of UO3 , given by Pattoret et al. [7] as 13. 6 ± 1 eV. UO+ was also found, which could again have been formed by the fragmentation of UO.J, the fragmentation potential being 13. 6 ± 0. 5 eV. The interpretation of the log(I+T) - 1/T plots can only be made if the composition of the solid for a series of measurem ents does not change markedly with time, i. e. one is close to a quasi-congruent vaporization, and that this point does not change much with temperature. The values for the heat of sublimation from the two studies are shown in Table II, which clearly shows the discrepancy in the data for U03. At 2241°K the vapour pressure was calculated for the system by determining the rate of effusion from the oxide and from the ion intensities, which showed U 0 2 to be the predominant species, the pressure of the gas was calculated. The total pressure decreased significantly as the O/M ratio decreased from 2. 0 to about 1. 94. The partial pressure was derived from the ion intensities by assuming that the ionization cross-sections and multiplier efficiencies are equal. The estimated partial pressures of the 136 POTTER TABLE II. ENTHALPIES OF VAPORIZATION OF THE GAS-PHASE SPECIES ABOVE U-Pu OXIDES Enthalpy of Composition Temperature vaporization Gas of î ange k ca l/mole species solid (•К) [3 0 ] [29] Uo.8 Р ц 0.2 O2.00 - 192 u o 3 1905-2411 [30] 158.5 ±2.0 115.8 ± 4.2 [30] U q.85 P^O.15 ^1.969 UOz 2000-2350 [29] 146.3 ±1.7 148.4 ±2.3 [29] Pu02 139.4 ± 1.8 141.9 ±5.0 PuO 123.4 ± 1.7 124.5 ± 1.3 0/M ATOM RATIO FIG.24. Partial pressures of the vapour species over (U 0,8 Pu0>2 )O2_x at 2241°K [30]. IAEA-PL-463/ 7 137 species are shown in Fig. 24. There appears to be no minimum as predicted in the calculations and found by Ohse and Olson in the range of O/M from 1. 92 to 2. 00. There is a discrepancy in the pressures of oxygen gas calculated from the two equilibria U 0 2 (g) + O(g) — U 0 3 (g) PuO(g) + O(g) — P u 0 2 (g) and the pressures are higher than predicted from the extrapolation of Rand and Markin data, even for curve A (Fig. 21). It is appropriate here to mention the earlier work reported by Pascard [13] on U-Pu oxides containing 10, 20 and 25% Pu, the data for which were reported only in terms of U and Pu-bearing species, indicating that the pressures of U-bearing species are greater than those of Pu-bearing species for O/M ratios between 1. 95 and 2. 00. The study of Dean et al. [ 18] extends the composition of Pu up to Pu/U+Pu ratios of 0. 50 and 0. 76. Free energy of formation data for PuO(g) and PuO 2(g) were determined from the evaporation of the binary system Pu-О and from the ternary system with Pu/U+Pu = 0. 76. Pu0 2 gas was found to be more stable than previously reported by Ohse and Ciani AGf(puo2) = - 117. 700 + 3. 05 cal/mole and this value of Pu0 2 suggests that the values for AG0z should be corrected slightly from those given by the extrapolation of the data of Rand and Markin [ 28]. Some specific heat data (Cp) have been reported in the temperature range between 1600°C and the melting points by Affortit and Marcon [ 33] Experiments on the thermal diffusion of oxygen in hyperstoichiometric urania — 15% plutonia solid solutions [ 34] also suggest that there may be a non-linear dependence of AG0¡, on temperature when T > 1200°C. Some data on the vaporization of Uo.5Puo.5O2.10 at 1814°K [ 18] are also available and, on the basis of the experimental results, suggest that the oxygen potentials should be modified; the presence of U O 3 was found to be less than that calculated, but some difficulty would be encountered in defining the composition of the solid for a given pressure measurement.- Measurements of the partial molal heat of solution (AHq2) for the mixed oxides with Pu/U+Pu = 0. 10 and 0. 15 have also been reported and are in semi-quantitative agreement with those of Rand and Markin. 2. FISSION PRODUCT PHASE DIAGRAMS Some aspects of the phase equilibria for the major fission products with uranium and plutonium oxides are discussed, together with the application of the principles of phase equilibria in predicting the chemical state of a 'burnt' fuel. Some attempts to predict the chemical state of burnt oxide fuels using thermochemical data were reported by Rand and Roberts [ 35]. At burn-ups of about 10% of all the U and Pu relatively large quantities of some of the fission product elements will be produced. The thermodynamic data for the oxides allow some predictions to be made concerning the chemical composition of this 'multicomponent' oxide system. IAEA-PL-463/ 7 139 The approach to predict the final chemical composition is simply to use an Ellingham diagram (Fig. 25) for the oxides of the fission product elements, together with that for the fuel oxide, so that the oxygen partial molal free energy (AGo2) of the system is a minimum. However, complications are whether the oxides formed are soluble in the fuel matrix or whether they are present as separate phases, and whether ternary phases can form. To answer these questions detailed observations on irradiated fuels must be made on extremely well-characterized materials; but some progress can be made by simply having some knowledge of the phase relationships of the fission product elements with the U-О and Pu-О systems. The earlier calculations for therm al fission [ 35] of 2 3 5UC>2 were extended by Rand and Markin [28] for a fuel composition Uo.85Puo.15O2 subjected to a fast neutron flux, and to a burn-up of 7% of the heavy atoms. The fission product elements can be divided into the following groups: (a) Those such as Ba, Sr, Zr, Y, and the rare earths that form very stable oxides (b) Nb, Mo, Tc with oxides whose oxygen potentials are quite close to that of the fuel, together with those that form very unstable oxides such as Rh, Pu and Pd (c) Cs, Rb, which will form a liquid phase at the operating temperature of the fuel elements and will most likely condense in the colder regions of the. fuel elements; I is also considered here (d) Te and Se. The rare gases will not of course form compounds under these conditions. Aspects of the phase equilibria will now be discussed. 2.1. The group of fission products Ba, Sr, Y and the rare earths 2. 1. 1. The Ba-U-O system Mclver [ 36] heated mixtures of U 0 2 and BaO together at 1500°C under a controlled oxygen potential of -110 kcal/mole O2 . For Ba concentrations up to a Ba/U+Ba ratio of 0. 1 no variation in the lattice parameter of U0 2 was found and it was concluded that under these conditions of temperature and oxygen potential Ba does not dissolve in UO2 . Ba forms ternary compounds with U and О, Ва1ГОз has a perovskite structure (cubic Pm3m) and there is a series of Ba uranates containing 6 -valent U [37]. The compounds are BaU 2 C>7 (tetragonal), Ba 2U3 O u , BaUC>4 (orthorhombic), and B a 3 UOg. These compounds with 6 -valent U would only be expected to form in an environment with a very high oxygen potential (fuel O/M > 2.00). 2.1.2. The Ba-Pu-O system Because of the slightly smaller ionic radius for a given valence for Pu compared with U, it is to be expected that Ba would be insoluble in P u 0 2 , and in the presence of Pu3+ the solubility may be higher. Again the perovskite BaPu0 3 compound exists with the ВаТЮ 3 structure [ 38]. Ba plutonates with higher plutonium valencies exist at correspondingly higher oxygen potentials [ 39]. 140 POTTER 2.1.3. The Sr-U-O system Mclver [ 36] found that the lattice parameter of U0 2 decreased in the presence of SrO at 1500°C and in the presence of oxygen at a potential of -110 kcal/mole 0 2 . The variation in param eter was given by a 0 = 5. 4700 - 0. 0046 X A where x is the mol. % of SrO (maximum about 1 2 mol. %). There are again a series of uranates (with 6 -valent U) SrU4 0 13, S r 2 U3Ou , SrU0 4 , Sr2 U0 5 and S r 3U06 . SrUOg has an a-rhombohedral form isomorphous with CaU04 and a ß -orthorhombic form isomorphous with BaU04 . Cordfunke and Loopstra [40] have measured the heat of solution of these compounds in nitric acid and have calculated the values for the heat of formation (ДН°298 ) as SrU 4 0 3 - 1424 kcal/mole (estimated), S r 2 U3Ou - 1242 kcal/mole, a-SrU0 4 - 469.6 kcal/mole, ß -SrU 0 4 - 469.9 kcal/mole, Sr 2 U0 5 - 617.2 kcal/mole and S r 3 U 0 6 - 760.2 kcal/mole. 2. 1.4. The Sr-Pu-O system A cubic perovskite SrPu0 3 exists, together with compounds containing 6 -valent Pu [ 38, 39]. 2 .1 .5 . The B a-Z r-O and S r - Z r - О system s As Zr is a major fission product as well as Sr and Ba, the possibility of the formation of Ba and Sr zirconates in irradiated fuels must be considered. Solid solutions of the form Baj-xS^ Z r0 3 with a perovskite structure may be formed under certain conditions. The formation of the Ba and Sr uranates, plutonates and zirconates or solid solutions of all components, namely Baj-xSrx (Ui_y.zPuy Zrz )0 3 and the equilibria with BaO and SrO and also with the mixed U, Pu oxides with some solubility of Zr in the oxide matrix, are very complex problems but it would be interesting to obtain correlations between the oxygen potential of the fuel matrix and the chemical form of the grey inclusions containing Ba. In some calculations of the chemical state of an irradiated fuel [41] the formation of BaZr0 3 has been assumed and there has been some evidence for this compound from studies by Bradbury et al. [42] on irradiated U0 2 . Ba and Zr were found together with some Sr by microprobe analysis. Schmitz [43] simulated a U0.gPu0.2 oxide fuel with 25 fission product elements with a bum-up of 2%; the O/M ratio was 1.978. A grey phase was found in the simulated oxide which was a (U-Pu) Ba-O phase. However, in a simulation experiment reported by Schmitz et al. [4 4 ], in which 16% burn-up was simulated in a Uq.sPuo. 2 oxide fuel, a grey phase containing Ba and Zr but with no U and Pu was found. Oi and Tanabe [ 45] found that Ba was segregated on the surface of slightly irradiated single crystals of UO2 . O'Boyle et al. [46] examined a U0 2 - 20 wt.% Pu0 2 with O/M = 2. 00 irradiated in a fast neutron flux to a burn-up of 2. 7%. M icro probe analysis revealed a grey phase that contained Ba, Sr and Ce. The Ce was believed to be a daughter product of 140Ba that originated from the decay chain: 16-s Xe -» 6 6 -s Cs -» 12.8-d 140Ba - 40. 2-h La - stable Ce IAEA-PL-463/ 7 141 14CBa is probably stable for a sufficient time to nucleate before decaying to Ce. Clearly the conditions under which these various phases containing Ba, Sr, Zr and sometimes U and Pu will require further assessm ent. The eutectic temperature for BaO and BaZrC >3 and for SrO and SrZrC>3 are given as 2000 and 2200°C at ~50 mol. % BaO and ~20 mol. % SrO [47]. 2.1.6. The U-Zr-O system Cohen and Schaner [48] investigated the phase relationships in this system and presented data for temperatures greater than 1000°C. The phase diagram presented by these authors is shown in Fig. 2 6 . A continuous solid solution was established in the temperature range 2300 to 2550°C. This solid solution has the fluorite structure. Pure ZrÛ 2 undergoes two transitions, monoclinic-tetragonal (1170 ± 20°C) and tetragonal-cubic FIG.26. TheUQ>-Zr02 temperature-composition diagram from 600 to 1200“C. The phase designations are: C, face-centred cubic; T, face-centred tetragonal; M, monoclinic [49]. 142 POTTER UC>2 mole fraction ZrOg FIG .27. Revised U 02-Zr02 phase equilibrium diagram. The phase designations are: L, liquid; C, face-centred cubic; T, face-centred tetragonal; M, monoclinic [49]. (2285 ± 15°C), before the melting point is reached (2710 ± 30°C). A two- phase region was found above 1660°C and extends to about 2300°C on the ZrC>2 -rich composition side. The phases in this region are fc cubic and fc tetragonal. The temperature at which the ZrC>2 -rich phase in the two- phase region transforms to the monoclinic structure was given as ~100°C and the solubility of Zr0 2 in UO2 was given as ~ 1 2 mol. %. Romberger et al. [49] extended these phase-equilibria studies to lower temperatures by the use of a molten fluoride ' flux' , which provided a means of readily obtaining the equilibrium phases at temperatures less than 1200°C. The revised phase diagram is shown in Fig. 27 and when compared with Fig. 26 it is seen that the later study indicates a more rapid decrease in the mutual solubility of the cubic and tetragonal phases below 1600°C. The eutectoid temperature is given as 1110°C, which is much higher than the previous estimate. The eutectoid composition was at 2 . 8 mol. % U02. IAEA-PL-463/ 7 143 FIG.28. Proposed room-temperature isothermal section Pu02-Pu01-5-Zr02 [51]. 2.1.7. The Pu-Zr-O system The earlier work on this system [ 50] has been recently extended by Mardon et al. [ 51] . These studies on samples with O/M ratios of 1. 61 to 2 . 00 showed that the.stabilization of cubic Z r 0 2 by P u 0 2 to room temperature as previously observed is related to the presence of reduced oxides, the compositions of which lie on the Pu0 i.6i-Z r 0 2 -X tie line. In the fully oxidized state tetragonal Z r 0 2 is stabilized over a wide range of РиОг contents at high temperatures, whilst at lower temperatures the solubility of РиОг in the tetragonal phase appears to pass through a maximum at ~1000°C and then decreases rapidly, leaving a wide two-phase region of РиОг-rich cubic (Pu, Z r)Ü 2 plus ZrO g-rich tetragonal (Pu, Z r )0 2. In РиОг-rich samples there is evidence of the existence of fluorite, С-type rare-earth oxide and pyrochlore structures at different oxygen potentials. The pyrochlore structure is based on the composition P u 2Z r 2 0 7 . A possible room temperature isothermal section for Pu2 -PuO i,5 -Z rO î is shown in Fig. 28. The lattice parameters for the fluorite Pu0 2 *Z r 0 2 solution do not deviate from Vegard's law. 2.1.8. A comparison of the U-Zr-O and Pu-Zr-O phase relationship The phase relationships show a number of regions of sim ilarity but also marked differences. Both systems have complete solubility in the cubic phase field at high temperatures, extensive solubility in tetragonal ZrÛ 2 and only limited solubility in monoclinic ZrÛ 2 . In the case of the Pu system stabilization of tetragonal Z rÛ 2 occurs over a much wider composition and temperature than in the U -system and there does not appear to be any eutectoidal decomposition of the tetragonal phase into cubic plus monoclinic structures as observed in the U 0 2 - Z r 0 2 system. The solubility limit of Z rÛ 2 i-n the fluorite cell of P u 0 2 is significantly higher at low temperatures than is the solubility of Z r 0 2 in U 0 2 [47]; this may reflect the slightly more favourable size difference between the РиОг and Z r 0 2 cells than between U 0 2 and Z r 0 2 cells, 5% and 7% respectively. 144 POTTER Romberger et al. [49], however, obtained very low figures for the mutual solubilities of UO2 and ZrC>2 at temperatures below 1200°C. F or the low and high temperature data to be consistent large deviations from ideality for the solid solutions must occur. The Zr formed in fission will be dissolved in the fluorite solid solution of the fuel matrix; this has, indeed, been found in the examination of irradiated oxides [46] as well as Y and the rare earths. There is an extensive region of fluorite solid solution at the actinide-rich end of these system s. 2.1.9. The U-Y-O system Bartram et al. [ 52] have described the phase relationships in the region UO3 -UO 2 -Y 2 O3 of this ternary system. A phase diagram for the temperature range 1000 - 1700°C is shown in Fig. 29. In the diagram the solid lines represent established phase boundaries and the long-dash lines represent probable boundaries, which are not well established. The short- dash lines denote the approximate composition line for three experimental conditions: (a) hydrogen - 40°C dew point, 1700°C; (b) air 1000°C; and (c) 10 СО 2/ CO at 1500°C. The phase diagram shows a U3 Og phase, a fee fluorite cubic solid solution phase, a bcc phase (rare earth С-type oxide) and two rhombohedral phases. The rhombohedral phases occur over a range of yttria compositions but at a constant oxygen to metal ratio of 1. 71 and 1. 87. The solubility of Y2Os in the fee fluorite solid solution ranges from 0 to 50 mol. % Y20 3 in dry hydrogen at 1700°C, and from 33 to 60 mol.% Y20 3 in air at 1000°C. As the yttria concentration is increased above Y j0 3 FIG.29. The UO2 -UO 3 -Y 2O 3 ternary diagram from 1000 to 1700eC [52]. IAEA-PL-463/ 7 145 18 mol. % the fluorite-U3Os phase boundary follows close to the M 0 2 com position line. The bcc solid solution extends from 80 to 100 mol. % Y2 O 3 in hydrogen; however, in air above 1000°C there is little or no solubility of UO3 . Low-temperature oxidation gives a maximum ratio for O/M of ~ 1. 56 for the bcc solid solution. O ■ B-UjOg • F к . F ¿ = R II . F • ■ RII X = R I . F ■ a RI at.% FIG.30. A section of the phase diagram for the ternary system U-La-0 at 1250cC [53]. 2.1.10. The U-La-O system Diehl and K eller [ 53] have recently published data on the U0 2 -U 0 3 ~La0 i.5 system. A section of the phase diagram at 1250°C is shown in Fig. 30. At 1250°C the following features were observed: (a) No La solubility in ß-UsOs (b) The fluorite (U, Ьа)Ог±х exists over a considerable range of composition (c) A rhombohedral phase 1, an ordered phase that occurs at the limiting compositions UO3 6 L aO i.5 ^ 7 0 1 2 ) with U6 + , and UO2 6 L a 0 i. 5 (M7 0 n) with U4+, and with a phase width corresponding to the composition UO2.5 5LaOi.s and UO2.5 7LaOi.s (d) A rhombohedral phase II; this phase extends from 71. 5 to 76. 5 mol. % LaO i,5 for U(VI) and shows only little phase width with respect to O/M ratio. Above 1310°C this ordered phase transforms reversibly to a disordered phase having a fluorite structure. 2.1.11. The U-Nd-O system The system U0 2 -U 0 3 -Nd0 i .5 has recently been described by Boroujerdi [ 54]. At 1250°C this system was found to consist of four single-phase regions and three two-phase regions; a section of the system is shown in Fig. 31. The single-phase regions are: (1) ß-U308 where no solubility of Nd01-5 could be detected; and (2) the fee fluorite phase (U, Nd)0 2 ±JU which covers a large area of the system. The limiting metal/ oxygen compositions 146 POTTER in this phase are MOi,6o ancl MO2.2 5. A rhombohedral phase occurs with limiting compositions UC>2 6 NdOi.5 and U0 3 6 Nd0 i. 5 . No variation in the U and Nd concentrations was found; this is a different behaviour to that of the La system. There is no solubility of U in the hexagonal A-type NdO^s lattice. In the quasi-binary section U 0 2+x-NdO},5 (p(02) = 1 atm) the phase width of the fluorite phase (U, Nd)Q2 ±x increases with rising temperature. 2.1.12. The U-Ce-O system Using high-temperature X-ray powder techniques Markin et al. [ 55] constructed a ternary phase diagram between UO2-U 3O8 and CeC>2-C eO i,8i for all concentrations of Ce and for temperatures between room temperature and 600°C. Reduction of oxides to a hypostoichiometric composition where z > 0 . 35 for U i.zCezC>2+x results in the formation of two phases MO2-00 and M 0 2-x in equilibrium at room temperature. Upon heating the two-phase product a single-phase is formed at a temperature dependent on the value of z. The O/M ratios of the M 0 2_x phase correspond only to a partial reduction of CeIV to Ce111 and represent an intermediate phase. In this respect the Ce-O [56] and the U-Ce-O systems are similar. Reduction of oxides with z < 0. 35 results in a single fee phase at all temperatures. P artial oxidation to a hyperstoichiometric composition when z < 0. 5 results in either a single-phase fee M 02+x, or M 0 2+x in equilibrium with an M4O 9 phase. Further oxidation causes the disappearance of the M 0 2+x phase; the M4O9 phase is then in equilibrium with а МзОв-у type phase. P artial oxidation to a hyperstoichiometric composition when z > 0. 5 results in a single-fcc M 02+x phase. Phase diagrams at room temperature, 200, 400 and 600°C are shown in Fig. 32. U 0 2 and C e 0 2 solid solutions obey Vegard's law. Markin and Crouch [57] used a gas equilibrium technique to obtain thermodynamic data at 800 - 9 50°C, that is oxygen potentials for hypo stoichiometric U-Ce-O (U1 .xCex0 2.y) where 0. 10 < z < 0. 75. Plots of partial molal enthalpy and entropy for the system show great similarity to plots for the (U, Pu) oxides but a very different behaviour to the pure oxide system, Ce02.x and Pu02-x. The similarity between enthalpy plots IAEA-PL-463/' 7 147 Lim it of reduction in present experiments FIG.32. U-Ce-O ternary phase diagram at (a) room temperature (b) 200, 400 and 600eC. for the mixed oxides supports the idea that О vacancies cannot locally order in the presence of moderately high concentrations of U (IV) ions. The relationships between Ce valency and A(jo2 are shown in Fig. 33, and do not lie on one plot for all concentrations as for the (U, Pu)C>2±x system. 2.1.13. The U-Gd-O system Some studies on this system have been described by Beale at al. [ 58]. Solid solutions between UO2 and Gd2Û3 prepared by sintering the oxides together at 1700°C in dry hydrogen exist up to Gd/U+Gd ratios of 0. 80; the solutions had a fee structure and the relation between lattice param eter and Gd concentration was linear. The O/U+Gd ratios were not determined. 148 POTTER -70 U, C e .О, 0.75 0.66 0.50 0.3¿ 0.15 0.10 -90 -130 3.7 З.Е 4.0 Cs V ALENCY FIG.33. AG(o 2) plotted against Ce valency at 800°C [57]. 2 . 1. 14. The Pu-C e-O system The only work reported on the rare earth-Pu-O system, i. e. for the rare earths that are present in reasonable quantities, is on the Pu-Ce-O. Mulford and Ellinger [ 59] have shown that the lattice param eters for solid solutions of СеОг-РиОг obey Vegard's law. 2 .2 . The fission product elements: Nb. Mo, Tc, Ru. Rh and Pd The oxides of the first three elements, namely Nb, Mo and Tc, form oxides that are more stable than Ru, Rh and Pd. Nb is considered to be present in an oxidized state as a separate phase. The transition metals Mo, Tc, Ru, Rh and Pd are present in the form of a single-phase alloy, which has been identified in irradiated oxide fuels [ 42, 46, 60-64]. The actual concentration of the elements in these inclusions depends on the temperature and position in the irradiated fuel. Bramman et al. [ 62] extracted these inclusions from the burnt oxide fuel of initial composition, Uo.85Puo.15 O 2, which has been irradiated in the Dounreay Fast Reactor to 8-8. 5% burn-up. The composition of the mechanically extracted inclusions was Mo 39. 1 wt. %, Tc 14. 2 wt. %, Ru 30. 4 wt. % and Rh 6. 7 wt. %, and this alloj^ had ahexagonal structure IAEA-PL-463/ 7 149 TABLE III. LATTICE PARAMETERS Rh Annealing temperature a o c 0 (at.% ) (•С) (A) 4 5 .2 1700 2 .7 5 9 7 4 .4 3 4 7 5 8 .7 1750 2 .7 4 3 5 4 .3 9 0 7 8 0 .9 1760 2 .7 1 9 9 4 .3 4 6 4 TABLE IV. COMPOSITION RANGE OF INCLUSIONS IN IRRADIATED PLUTONIUM OXIDES Inclusions Initial fuel composition Mo Tc Ru Rh Pd Pu01-7 8 .5 - 1 . 5 4 . 2 - 8 . 4 7 . 8 - 1 5 . 7 4 .6 - 6 .9 6 - 1 0 .8 ( U .7 ) ( 6 .1 ) ( 1 2 .1 ) ( 5 .8 ) (9 .5 ) Pu02.o 0 .3 - 1 . 5 1 . 2 - 3 . 1 5 . 9 - 1 5 . 3 1 . 3 - 6 . 6 0 .9 - 1 3 . 2 ( 0 .9 ) ( 1 .7 ) (8 .0 ) (3 .7 ) (7 .9 ) Average values in brackets. with a o = 2 .7 3 ± 0 .0 2 A and co = 4. 444 ± 0.04A . An attempt was made to prepare a synthetic alloy containing the same quantities of Mo, Tc, Ru and Rh by Bramman et a l ., but some Rh was lost during the preparation; the composition was (normalized to 100% total) Mo 43. 5 wt. %, Te 17. 7 wt. %, Ru 35. 5 wt. % and Rh 3. 3 wt. %. Although the X -ra y diffraction pattern indicated the presence of only a single-phase hexagonal structure with cell size ao = 2. 761 ± 0. 002 A and со = 4. 439 ± 0. 005 A, examination by microprobe analysis showed signs of a second-phase (5% volume) that had a composition close to MosRuTc. The melting point of the alloy was between 1800 and 1900°C. The hexagonal alloy has the same structure as the Mo-Ru alloys with the hexagonal close packed phase, e-phase [65]. This structure extends from 45 to 82 at. % Rh and the lattice parameters accurate to ± 0. 0002 A were as given in Table III. The chemically extracted inclusions in the studies of Bramman et al. [ 62] were found to have some Pd present. A composition was given as Mo 41.0 wt. %, T c 14. 9 wt. %, Ru 31. 9 wt. %, Rh 7. 1 wt. % and Pd 2. 0 wt. %.- One of the reasons why the composition of the alloy inclusion changes with radial position is that the alloy can oxidize to form Mo02, provided the 0 2 potential is high enough. This phenomenon is illustrated in the studies of Davies and Ewart [ 6 6 ], when the inclusions in two irradiated plutonium oxides of different initial . stoichiometry were examined. The composition of the inclusions is given in Table IV. Although there are Very large scatters on the results, it is apparent that for the oxide with the higher stoichiometry not only Mo but also Tc has been oxidized. 150 POTTER FIG.34. The effect of change of Mo activity on the oxygen potentials for the reaction Mo + Oz = Mo02. TcC>2 is less stable than M0 O 2 by more than 2 0 kcal/mole. Thus, when the alloys are oxidized one would expect M o02 to be formed and that not until all the Mo had been oxidized would Tc be oxidized. It is also worth noting that the reaction Mo) 1 + (O2 ) ^'хМ оО г^ may not buffer the oxygen potential in a burnt fuel. The Mo is not at unit activity, dissolved in the 5-component alloy (Mo-Tc-Ru-Rh-Pd), and immediately the alloy oxidizes the Mo concentration and thus its activity decreases and higher oxygen potentials are required to oxidize the alloy. If an ideal solution is assumed for Mo in the alloy, its atom fraction will be ~ 0 . 30. A difference of ~ 14 kcal at 2000°K is found for the oxygen potential for the above reaction when a Mo= 1 and aMo = 0..03. This is illustrated in Fig. 34. O'Boyle et al. [46] also extracted an inclusion from irradiated oxide m aterials with the composition 20. 0 wt. % Mo, 16. 6 wt. % Tc, 48. 6 wt. % Ru, 12. 9 wt. % Rh and 2. 0 wt. % Pd. U, Pu, С, N and О were not detected and there was a tendency for both Mo and Pd to migrate towards the colder regions of the fuel. An alloy was prepared with the same composition as the inclusion and was found to have a hexagonal structure with ao = 2. 735 ± 0. 001 and со = 4. 355 ± 0. 001 A.' In the irradiated oxides examined by Bramman et al. [62] a second cubic metallic phase adjacent to the Mo-Tc-Ru-Rh-Pd alloy was found. IAEA-PL-463/ 7 151 The composition of this inclusion was 2 5. 4 wt. % U, 13.5 wt. % Pu, 38. 4 wt. % Pd, 11.9 wt. % Rh and 2. 5 wt.o% Ru. The lattice parameter of this cubic phase was a о = 4. 127 ± 0. 002 A. Clearly this compound is of the СизАи type and sim ilar to URu3 [ 67] and URh3, which have lattice parameters ao = 3. 988 A and ao = 4. 063 A, respectively. It is to be expected that the U and Pu concentrations in this alloy would be different from the fuel matrix, namely Pu/U+Pu = 0. 15. Schmitz et al. [44] also observed this phase in a simulation of oxide fuel at 16% burn-up. Microprobe analysis showed the presence of a UPd3 phase in addition to the expected Mo-Tc-Ru-Rh-Pd alloy; in this simulation this alloy contained Re as a substitute for Te and also some Nb. The formation of these actinide-group VIII element alloys is curious at the stated initial oxygen concentrations of the fuel, which were close to 2 in the studies of Bramm an et al. and of Schmitz et al. The free energy of formation of URu3 has been measured by Holleck and Kleykamp [ 6 8 ] using an CaF 2 electrolyte galvanic cell in the temperature range 1000 - 1140°K. The value for the free energy of formation was AGf AGf The compounds of the U systems are most likely to be more stable than the corresponding alloys of the Pu system. Where the U potentials of the region Ru + URU3 are extrapolated to the temperatures appropriate to a fuel element, the equilibrium U oxide would be hypostoichiometric UOg-x. This is borne out by the observations that actinide group VIII element compounds can only be made by reaction of the oxide and metal in very reducing atmospheres (i.e. dry hydrogen) [70]. 2.3. The systems involving Cs. Rb and I Cs and Rb have very low melting points, 30 and 39°C, respectively, and, because of their high vapour pressures, aré expected to be found in the colder regions of the fuel element, namely, in the region between the fuel and the cladding. Some iodine will also be found in this liquid phase; this liquid phase will also dissolve oxygen, the amount depending on the oxygen potential of the fuel surface. A relation is required between the oxygen potential and oxygen concentration for these Cs-Rb-O-I solutions. The mode of reaction between this solution and the stainless steel cladding m aterials will depend markedly on the oxygen potential; the conditions under which compounds in the ternary systems Cs-Cr-O, Cs-Mo-O form are not yet known. The formation of Cs (and Rb) uranates and plutonates must also be considered, as also for a 'failed' fuel element pin situation where sodium primary coolant can come into direct contact with the oxides, and then the formation of sodium uranates and plutonates and the conditions under which they form must be considered [ 71]. Ohse and Schlechter [ 72] have considered several aspects of the role of caesium in fuel-cladding interactions. 152 POTTER 2.4. The Te and Se systems Te has been found associated with Ag in the region of the m etallic- transition element inclusions by Huber and Kleykamp [63] in irradiated U0. 1 5 Pu 0. 85O1 . 98i 0. 0 1 . The actual chemical form of both these fission products is not known, although because of their high vapour pressures they could migrate via the gas phase to the colder regions of the fuel elements. U and Pu oxytellurides and oxyselenides are known to exist but the actual oxygen potential at which these compounds form has not been determined. There are, of course, a series of U and Pu tellurides and selenides. 2.4.1. The U-Te-O system The ternary compound UOTe has been reported [ 73] to possess a tetragonal symmetry of theoPbFC l type with lattice param eters a 0 = 4. 004 A and с = 7. 491 A. A second ternary phase и 20 2Те has recently been identified [ 74] with a be tetragonal cell (a0 = 3. 964 and c 0 = 12. 346 A) and as isomorphous with the rare earth oxytellurides [75]. A tentative phase diagram for this system at 1200°C has been given by Breeze [ 76] and is shown in Fig. 35. The oxygen potential for the phase fields in which the oxytellurides are present would be required before any assessment as to their likely presence in a 'burnt' oxide fuel can be made. 1/2 0 , x x / \ Те U,Te,U7Tee FIG.35. A section of the U-Te-O system at 1200°C [76]. IAEA-PL-463/ 7 153 1/2 0 , FIG.36. A section of the U-Se-O system at 600°C [76]. 2.4.2. The Pu-Te-O system Pu¿02Te is the only reported ternary compound of the system, which is reported to be isomorphous with U20 2Te [ 77]. 2.4.3. The U-Se-O system Breeze [ 76] has also presented a possible phase diagram for this system at 600°C, see Fig. 36. Again the oxygen potentials of the phase fields involving UOSe, the only ternary compound of the system, are required. UOSe crystallizes with a tetragonal symmetry with a PbFCl type phase with lattice param terse ao = 3. 9005A and со = 6 . 9823 A. No evidence was obtained by Breeze for a U2 0 2Se compound. 2.4.4. The Pu-Se-O system No data is available for the ternary phase diagram but two ternary compounds exist, PuOSe with a tetragonal (Р4/mm) structure (ao = 4.151 A and co= 8 . 369 A) and Pu20 2 Se, which is not isomorphous with игОгТе but crystallizes with the A-type hexagonal L a 20 3 structure (a 0 = 3. 957 A and c 0 = 6 . 977 A) [ 78] . REFERENCES [1] JAVED, N.'A., J. Nucl. Mater. 43 (1972) 219. [2] WHEELER, V .J., J. Nucl. Mater. 39 (1971) 315. 154 POTTER [3 ] MARKIN, T. L ., WHEELER, V .J ., BONES, R .J., J. Inorg. Nucl. Chem. 30 (1968) 807. [4 ] TETENBAUM, M ., HUNT, P .D ., J. Chem. Phys. 49 (1968) 4739. [5] TETENBAUM, M., HUNT, P.D., J. Nucl. Mater. 34 (1970) 86. [6 ] EDWARDS, R .K ., CHANDRASEKHAR! AH, M .S ., DANIELSON, P .M ., J. High Temp. Sei. 1 (1969) 98. [7 ] PATTORET, A ., DROWERT, J . , SMOES, S ., Thermodynamics of Nuclear Materials, 1967 (Proc. Symp. Vienna, 1967), IAEA, Vienna (1968) 613. [8 ] ACKERMANN, R .J., RAUH, E .G ., CHANDRASEKHARIAH, M .S ., Rep. ANL-7048 (1965). [9 ] REEDY, G .T ., CHASANOV, M .G ., J. Nucl. Mater. 42 (1972) 3 4 1 .. [1 0 ] LEIBOWITZ, L ., CHASANOV, M .G ., MISHLER, L. W ., FISCHER, D .F ., J. Nucl. Mater. 39 (1971) 115. [1 1 ] HEIN, R .A ., FLAGELLA, P .N ., Rep. GEMP-578 (1968). [12] LATTA, R.E., FRYXELL, R. E ., J. Nucl. Mater. 35 (1970) 195. [1 3 ] INTERNATIONAL ATOMIC ENERGY AGENCY, The Plutonium-oxygen and Uranium-plutonium-oxygen Systems: A Thermochemical Assessment, Technical Reports Series No. 79, IAEA, Vienna (1967). [14] SARI, C ., BENEDICT, U ., BLANK, H., Thermodynamics of Nuclear Materials, 1967 (Proc. Symp. Vienna, 1967), IAEA, Vienna (1968) 587. [1 5 ] GARDNER, E .S ., MARKIN, T .L . , STREET, R. S ., J. Inorg. Nucl. Chem. 27 (1965) 541. [1 6 ] MARKIN, T. L ., RAND, M. H ., Thermodynamics (Proc. Symp. Vienna, 1965) 1, IAEA, Vienna (1966) 145. [1 7 ] MARCON, J . P . , POITREAU, J . , ROULLET, G ., Plutonium 1970 and Other Actinides, (Proc. Conf. Santa Fe, 1970), Nucl. Metall., Metall. Soc. AIME П pt. II (1970) 799. [1 8 ] DEAN, G ., BOIVINEAU, J . C . , CHEREAU, P ., MARCON, J .P ., Plutonium 1970 and Other Actinides (Proc. Conf. Santa Fe, 1970), Nucl. Metall., Metall. Soc. AIME 17 pt.II (1970) 753. [1 9 ] RILEY, B ., Sei. Ceram . 5 (1970) 83. [2 0 ] SARI, C ., BENEDICT, U ., BLANK, H ., J. Nucl. Mater. 35 (1970) 267. [2 1 ] BENEDICT, U ., SARI, C ., Rep. EUR 4136e (1970). [2 2 ] MARKIN, T .L ., STREET, R .S ., J. Inorg. Nucl. Chem. 29 (1967) 2265. [2 3 ] KOIZUMI, N .. NAKAMURA, Y ., Ceramic Nuclear Fuels (Proc. Symp. Columbus, 1969), Amer. Ceram. Soc. Spec. Pub. No. 2 (1369) 25. [2 4 ] POTTER, P .E ., Plutonium 1970 and Other Actinides, (Proc. Conf. Santa Fe, 1970), Nucl. M etall., Metall. Soc. AIME 17 pt.II (1970) 859. [25] LYON, W .L., BAILY, W.E., J. Nucl. Mater. 22 (1967) 332. [2 6 ] AITKEN, E. A ., EVANS, G. K ., ADAMSON, M .G ., LUDLOW, T . E ., Rep. GEAP-12229 ( 1971). [2 7 ] MARKIN, T .L ., McIVER, E .J ., Plutonium 1965 (Proc. Conf. London, 1965), Chapman and Hall, London (1967) 845. [28] RAND, M .H., MARKIN, T. L ., Thermodynamics of Nuclear Materials, 1967 (Proc. Symp. Vienna, 1967), IAEA, Vienna (1968) 637. [29] OH SE, R. W ., OLSON, W .M., Plutonium 1970 and Other Actinides, (Proc. Conf. Santa Fe, 1970), Nucl. Metall., Metall. Soc. AIME 17 pt.II (1970) 743. [3 0 ] BATTLES, J. E ., SHINN, W. A ., BLACKBURN, P .E ., EDWARDS, R. K ., Nucl. M etall., Soc. AIME 17 pt.II (1970) 733. [3 1 ] ACKERMANN, R .J ., FAIRCLOTH, R. L ., RAND, M. H ., J. Phys. Chem. 70 (1969) 3698. [32] OHSE, R. W ., CIANI, V., Thermodynamics of Nuclear Materials, 1967 (Proc. Symp. Vienna, 1967), IAEA, Vienna (1968) 545. [33] AFFORTIT, C., MARCON, J.P., Rec. Int. Haut. Temp. Ref. 1 (1970) 236. [3 4 ] ADAMSON, M. G ., CARNEY, R .F .A ., Trans. Amer. Nucl. Soc. 14 (1971) 179. [35] RAND, M .H., ROBERTS, L.E.J., Thermodynamics (Proc. Symp. Vienna, 1965) 1, IAEA, Vienna (1966) 3. [3 6 ] McIVER, E .J ., Rep. AERE M -1612 (1966). [3 7 ] ALLPRESS, G .G ., J. Inorg. Nucl. Chem. 26 (1964) 1847. [3 8 ] RUSSELL, L. E ., HARRISON, J .D .L ., BRETT, N. H ., J. Nucl. Mater. 2 (1960) 310. [3 9 ] KELLER, C .. Nucleonik 4 (1963) 271. [4 0 ] CORDFUNKE, C ., LOOPSTRA, B .O ., J. Inorg. Nucl. Chem. 29 (1967) 51. [4 1 ] ANSELIN, F . , Rep. BEAP-5583 (1969). [4 2 ] BRADBURY, В. T . , DEMANT, J .T ., MARTIN, P .M ., Rep. AERE-R 5149 (1966). [43] SCHMITZ, F., Ceramic Nuclear Fuels (Proc. Symp. Columbus, 1969), Amer. Ceram. Soc. Spec. Pub.2 (1969) 32. [4 4 ] SCHMITZ, F ., DEAN, G ., ROUSSEAU, M ., de KEROULAS, F ., van CRAEYNEST, J . C . , . Proc. Mtg Fast Reactor Fuel Elements, GfK Karlsruhe (1970) 396. [4 5 ] OI, N .. TANABE, I ., J. Nucl. Mater. 29 (1968) 288. [4 6 ] O 'BOYLE, D .R ., BROWN, F .L ., SENECKI, J .E ., J. Nucl. Mater. 29 (1969) 27. [47] von WASTENBERG, H., GURR, W., Z. Anorg. Allg. Chem. 196 (1931) 381. IAEA-PL-463/ 7 155 [4 8 ] COHEN, I ., SCHANER, B .E ., J. Nucl. Mater. 9 (1963) 18. [4 9 ] ROMBERGER, K ., BAES, C. F . , J r ., STONE, H .H ., J. Inorg. Nucl. Chem. 29 (1967) 1619. [5 0 ] CARROLL, D .F ., Rep. HW -69305 (1961). [5 1 ] MARDON, P. G ., HODKIN, D .J ., DALTON, J . T . , J. Nucl. Mater. 32 (1969) 126. [5 2 ] BARTRAM, S .F ., JUENKE, E. F . , AITKEN, E .A ., J. Amer. Ceram. Soc. 47 (1964) 171. [5 3 ] DIEHL, H .G ., KELLER, C ., J. Solid State Chem. 3 (1971) 621. [5 4 ] BOROUJERDI, A ., Rep. KFK 1330 (1971). [5 5 ] MARKIN, T. L ., STREET, R .S ., CROUCH, E .C ., J. Inorg. Nucl. Chem. 32 (1970) 59. [5 6 ] BEVAN, D .J ., KORDIS, J ., J. Inorg. Nucl. Chem. 26 (1964) 1509. [5 7 ] MARKIN, T .L ., CROUCH, E. C ., J. Inorg. Nucl. Chem. 32 (1970) 77. [5 8 ] BEALE, R .J ., HANDWERK, J .H ., WRONA, B .J ., J. Amer. Ceram. Soc. 52 (1969) 578. [5 9 ] MULFORD, R .N .R ., ELLINGER, F .H ., J. Phys. Chem. 62 (1958) 1466. [6 0 ] BRADBURY, B .T ., DEMANT, J .T ., MARTIN, P .M ., POOLE, D .M ., J. Nucl. Mater. 17 (1965) 227. [61] JEFFERY, B.M ., J. Nucl. Mater. 22 (1967) 33. [6 2 ] BRAMMAN, J .L ., SHARPE, R. M ., THOM, D ., YATES, G ., J. Nucl. Mater. 25 (1968) 201. [6 3 ] HUBER, H ., KLEYKAMP, H ., Rep. KFK 1324 (1972). [6 4 ] O'BOYLE, D .R ., BROWN, F .L ., DWIGHT, A .E ., J. Nucl. Mater. 35 (1970) 257. [65] ANDERSON, E., HUME-ROTHERY, W., J. Less-Common Met. 2 (1960) 19. [6 6 ] DAVIES, J .H ., EWART, F. T . , J. Nucl. Mater. 41 (1971) 143. [6 7 ] PARK, J .I ., J. Res. Natl. Bur. Stand. 72A (1968) 1. [6 8 ] HOLLECK, H ., KLEYKAMP, H ., J. Nucl. Mater. 35 (1970) 158. [6 9 ] CAMPBELL, G. M ., MULLINS, L. J . , LEARY, J .A ., Thermodynamics of Nuclear Materials, 1967 (Proc. Symp. Vienna, 1967), IAEA, Vienna (1968) 75. [7 0 ] ERDMANN, B ., KELLER, C ., J. Inorg. Nucl. Chem. 7 (1971) 675. [7 1 ] CORDFUNKE, E .H .P ., LOOPSTRA, B .O ., J. Inorg. Nucl. Chem. 22 (1971) 2427. [7 2 ] OHSE, R. W ., SCHLECHTER, M ., these Proceedings, IA EA -P-463/18. [7 3 ] KLEIN-HANEVELD, A .J ., JELLINEK, F ., J. Inorg. Nucl. Chem. 26 (1964) 1127. [7 4 ] BREEZE, E .W ., BRETT, N .H ., J. Nucl. Mater. 40 (1971) 113. [7 5 ] BALLESTRACCI, R ., Compt. Rend. 264 (1967) 1736. [7 6 ] BREEZE, E .W ., Ph. D. thesis, Dept. Ceramics, Sheffield U niv., 1971. [7 7 ] ALLBURT, M ., JUNKISON, A .R ., Rep. AERE-R 5541 (1967). [78] KELLER, C ., The Chemistry of the Transuranium Elements, Verlag Chemie (1971) 398. IAEA-PL-463/8 FORMATION OF PHASES AND DISTRIBUTION OF FISSION PRODUCTS IN AN OXIDE FUEL H. KLEYKAMP Institut für M aterial- und Festkörperforschung, Kernforschungszentrum Karlsruhe, Karlsruhe, Federal Republic of Germany Abstract FORMATION OF PHASES AND DISTRIBUTION OF FISSION PRODUCTS IN AN OXIDE FUEL. The composition and distribution of different metallic and ceramic fuel component-fission product, fission product-fission product and fission product-cladding m aterial inclusions in a uranium dioxide and a mixed oxide fuel investigated by microprobe analysis are given. The radial dependence of the metallic M o-Tc-Ru-Rh-Pd inclusions are reported, as are U-Pu-noble m etal and Pd-Sn-Te inclusions and Fe-rich precipitations together with Pd, Ni,. Mo or T c. Several ternary or quaternary oxide phases containing Cs, Sr, Ba, Zr, Mo, Ce or Cr occurring in the fuel and near the rim are classified. The technique of elecronprobe microanalysis on hot samples has been well developed in several research centres. This method has also been tested at the GfK in Karlsruhe with the shielded Cameca microprobe MS 46 over the last four years [1,2]. One of the main topics of the postirradiation program consists of the investigation of fission products in burnt oxide fuels and their radial distribution, the formation of phases with other fission products and with the fuel components and the cladding material as far as located in the fuel or in the gap. The interaction of the fuel and the fission products with the cladding is a second point of examination and will be given in a separate paper. I shall refer predominantly to the GfK irradiation test series on mixed oxide fuels in a thermal, epithermal and fast flux in the F R 2 in Karlsruhe, in the BR 2 in Mol and in the DFR, while comparing these results with those of other centres. 1. FISSION PRODUCT-FUEL COMPONENT PHASES The most surprising metallic fission product-fuel component inclusion found was a U, Pu, Ru, Rh and Pd-containing compound, which had already been reported by Bramman et al. [3]. In more recent investigations the Harwell group also observed a U, Pu and Pd-containing phase in a hypostoichiometric oxide in the fuel near the surface [4]. But this type of inclusion does not occur as a rule in burnt fuel, instead the noble metals form compounds with Tc and Mo, as described later. We have observed the fuel component-noble metal phases only occasionally in the ingots near the bottom of the central void of the epithermal neutron irradiated mixed oxide fuel elements [5]. They occur only together with the Mo-Tc-noble metal phases and in many cases adjacent to the oxide fuel matrix (Fig. 1). 157 158 KLEYKAMP Mo Tc Ru M ol 7 A - k - 1 3 Pos.1 ( 192 x ) FIG. 1. Two-phase metallic inclusion in the central void adjacent to the fuel U0>8PuQ' 2 0 2. Micro structure (130 X) : light grey phase : Mo, T c, Ru and small amounts of Rh; dark grey phase : U, Pu, Rh and Pd (see Fig. 2); black phase : fuel. Burn-up 5%, epithermal flux. AE -PL-463/8 EA IA 290 2S0 270 260 250 240 230 220 210 200 — - b [mm] FIG. 2. X-ray spectrum of the metallic fuel-noble metal inclusion in the central void of a UQ 8PuQ ,¿ü.¿ fuel after a burn-up of b°Jo, epithermal flux. 1 60 KLEYKAMP m m i ,'i s - % t *■ ШШШ SPECIMEN CURRENT MICROSTRUCTURE FIG.3. Two-phase metallic inclusion in the columnar grain zone of the fuel UQ % 85 PuQ, 15 0 2 . Above': T c, Ru, Rh, no Mo; below : Pd, Te, Sn; the distribution of Te and Sn is complementary. Burn-up 6%, thermal flux. IAEA-PL-463/8 161 An X -ray spectrum of the cubic face centred phase is given in Fig. 2. A quantitative analysis at this position gave 18 at. % U, 8 at. % Pu, 1 at. % Ru, 15 at. % Rh and 59 at. % Pd, which reveals that the structure should be of the type A B 3 with A = (U, Pu) and В = (Ru, Rh, Pd). Our measured U/Pu ratio of 30:70 is comparable with the Pu enrichment from 20% to approxi mately 30% at the central void. Further measurements on this phase in the centre of the ingot, i. e. a position not in contact with the fuel, show that the phase is much higher in U than the fuel at the central void (for example U:Pu = 94:6). Therefore, an enrichment of Pu in this phase is not specific. A second feature is the relatively small content of Ru, which is based on the missing PuRu3 phase in the Pu-Ru system and on the different stability of the actinide-noble metal compounds. We have measured the Gibbs energy of formation of the compounds URu3 [ 6 ] and URh3 [7] (AuCu3 -type), the latter much more stable with -63 kcal/mol at 1000°K. The AuCu3 -type UPd 4 will be stabilized as UPd 3 in this quinary compound. A possible explanation for the formation of the phase AB 3 is given by the equation (n + 1)A 0 2 + 3B = A B 3 + nA0 2 (i + i/n) , n » 1 (A 0 2 = (U, Pu)02ix ) with rAG = fAG< A B3> - fAG< A 0 2> + AG^ < О which can take place only at high temperature (T > 2000°K) and for a hypostoichiometric oxide [ 8]. Another fission product-fuel component phase consists of U, Pu and Te, probably with О as an oxitelluride, which is observed in the fuel near the surface. Only the minor part of the fission product Те will form this phase. But most of the Те compounds are formed with other fission products in the fuel or at the surface, for instance Cs-telluride, or with cladding materials in the clad or at the interface, which will be described later. The formation of Cs-uranate near the rim is possible. The existence of Na- uranate has been proven in failed fuel pins [5] . Microprobe analysis gave a significant diminution of Pu in the reaction zone. 2. FISSION PRODUCT-FISSION PRODUCT PHASES The most common inclusions that contain Mo, Tc, Ru, Rh and under certain conditions Pd, called 1 white inclusions' in early investigations, are well known in the literature [ 2,9-21]. Because of the low fission yield of Pd in thermal neutron irradiated urania, Pd is nearly undetectable in this type of inclusion. It is worth mentioning that our investigations revealed in most cases two phases in these inclusions with a complementary behaviour of Mo and Ru and an equal distribution of Tc and Rh [22]. This is probably based on the eutectic character of the binary Mo-Ru system. The Mo- rich part is often adjacent to the fuel matrix, while the Ru-rich part is near the pores. 162 KLEYKAMP Microstructure Я И М К ^ И р я Н я и ИНИВИВИЯ ¡¿Р^чИИИ ЗИ^шшмии • ? • = »мДжИвВи Ш и И > - '* • ■ .»?• - ч ■ ¡7 '■ . ' йг^м^ЯНкй М И И и Я г « M ö w I ^ д в и и la œ— I M НШЖМшЫ^^Ш^'« &ШВ£тШШмшШШШЁ1ЁШ Ш ы С * .. *v»..--f --]\\л\:Л :> г’ ш Ш Specimen current - Pu FIG. 4. Ceramic precipitations in the central void of a U0 _ 8 Pu0 _ 2 0 2 fuel after a burn-up of 6%, fast flux. Dendrites : fuel; matrix : two ceram ic phases containing oxides of Ba, Mo, Nb and small amounts of Zr, U, Pu; white inclusions : Mo, T c, Ru, Rh. IAEA-PL-463/8 163 The corresponding inclusions in the mixed oxide are monophase throughout and contain small amounts of Pd, but much less than the fission yield. Sometimes small amounts of Fe are observed as an impurity of the fuel. The radial dependence of the components has often been measured though the results of the different authors are contradictory. A decrease in the Mo and Tc concentration near the surface and the reverse behaviour of Ru and Rh was observed in a hypostoichiometric mixed oxide with O/M = 1.95 in an investigation in Karlsruhe [17]. .This result is in partial agreement with the observations of Crouthamel and Johnson [19] on a stoichiometric sample but in contrast to O' Boyle et al. [11], who found an increase in Mo and a decrease in Ru near the rim . The reason for the different results may be due to the different O/M ratio of the fuels and the irradiation conditions. Absolute concentration measurements on these inclusions in thermal neutron irradiated samples are not rich in meaning for radial distribution measurements because of the flux depression ensuing from an inhomogeneous distribution through fission. In more recent investigations we measured the Mo/Ru ratio over the radius of stoichiometric thermal and epithermal neutron irradiated mixed oxide fuels [22]. This ratio amounts to between 0. 3 and 0. 6 in the columnar grain zone (the theoretical value for therm al 239рц fission is 1 . 2 ). We found increasing values with increasing distance from the central void. This behaviour may be due to a higher oxygen potential in the inner region of the fuel corresponding to the occurrence of Ba-molybdate and other Mo-containing phases in this region. Sometimes Mo is completely absent in these m etallic precipitations in the colùmnar grain zone (Fig. 3). However, it is dangerous tc infer from the Mo/Ru or Тс/Ru ratios in these inclusions near the central void of stoichiometric or hyperstoichio- metric fuels that Mo, or particularly Tc, has been oxidized because it should be taken into consideration that the fission products are formed independently in the first step before alloying and can diffuse in the thermal gradient with different rates and even in different directions. For Pd another type of inclusions was detected, predominantly in the unrestructured zone consisting of Pd between60 and 80%, together with Te and. Sn and probably Sb [22]. The distribution of Te and Sn'in the inclusions is in many cases complementary (Fig. 3). The existence of this is curious because of the relatively high vapour pressures of the components. Pd4Te melts peritectically at 1000°C and Pd3Sn melts congruently at 1322°C. This phase could be formed in the colder part and transported as a liquid by thermal diffusion or by evaporation-condensation processes along the thermal gradient. In most cases, however, Pd was found together with Mo, Tc or the cladding components Fe and Ni near the fuel surface. An asymptotically décreasing frequency of these inclusions was stated with increasing distance from the surface up to a depth of 0. 5 mm [22]. The ceram ic fission product phases in the fuel contain Mo in nearly all cases, in part adjacent to Mo-containing metallic phases, which have been mentioned before [22]. The ceramic phases consist of Ba-molybdate, zirconate, cerate or ferrate. There is no evidence of a systematic concentration dependence of the radius [11-14]. In a few cases pure Z r 0 2 has been detected in the fuel without any accompanying element. Zr should be solved in the fuel and so it is possible that the observed Zr is a daughter product of the decay Sr-Y-Zr [22]. In a fast neutron irradiated 164 KLEYKAMP FIG. 5. M etallic vapour deposited Fe-N i-Cr cladding m aterial precipitations in cracks and ceram ic Cs and Ba-molybdate phases near the fuel surface and in the gap. Burn-up 6.5% , thermal flux. IAEA-PL-463/8 165 TABLE I. CLASSIFICATION OF FISSION PRODUCT PHASES FORMED IN A BURNT OXIDE FUEL Group Components Main constituents Position I U, Pu, Rh, Pd Pd' Central void U, Pu, Те, О (?) Te Near surface II Mo, Tc, Ru, Rh, Pd Mo, Ru Entire fuel Tc, Ru, Rh Ru Columnar grains Pd, Te, Sn, Sb Pd Unrestructured zone Pd, Mo, Sn Pd Unrestructured zone Ba, Sr, Mo, Zr, 0 Ba, Mo Entire fuel Ba, Mo, 0 Gap Ba, Ce, 0 Entire fuel Ba, F.e, 0 Entire Fuel Ba, Mo, Nb, 0 Ba • Central void Cs, Mo, 0 Near surface, gap III Pd, Mo Near surface Pd, Fe Near surface Pd, Fe, Ni Near surface Pd, Fe, Mo Pd, Mo Near surface Pd, Fe, Mo, Tc Pd, Mo Near surface Fe, Cr, Ni Fe (> 75%) Near surface Ba, Mo, Fe, 0 Near surface Cs, Te, 0 (?) Near surface Cr, Te, 0 (?) Near surface Cs, Cr, 0 At surface, gap fuel we detected a two-phase ceram ic ingot at the lower (hot) end of the central void that had not been detected before (Fig. 4) [23]. It is possible that the liquid ceramic phase was transported by thermal diffusion to the bottom. The fuel has been precipitated in a dendritic form. In addition to Ba-molybdate, Cs-molybdate was found in the gap between the fuel and the cladding (Fig. 5), more rarely in the fuel just below the surface [22]. 3. FISSION PRODUCT-CLADDING MATERIAL INCLUSIONS Components of the cladding material are transported into the fuel in different ranges and form compounds with the volatile fission products. The transport mechanisms have been discussed, for instance, the iodine-iodide cycle [24], but are not yet certain. The penetration depth of Fe is higher than that of Ni. Pd-rich precipitations appear together 166 KLEYKAMP with Fe and Ni only in a small region of the unrestructured zone up to 0. 5 mm from the fuel surface because of the high Pd vapour pressure. Pd has sometimes been found together with Mo up to 0.1 mm from the surface, but never with Cr. A vapour-deposited cladding material phase is often observed in cracks near, the fuel surface, sometimes together with Pd. An example of a fission product-free precipitation is given in Fig. 5. Pd-free phases found are Mo-Fe alloys and Cr and Cs tellurides, the latter probably in an oxidized form. 4. CONCLUSIONS The multicomponent system fuel-fission products-cladding components entails an immense number of phases and an attempt has been made to classify them into three groups (Table I). It is not easy to analyse and to compare all the pubished microprobe investigations on burnt fuels in detail because important specifications and the irradiation conditions are often lacking. Samples with a higher burn-up and more refined metallo- graphic preparation and electron probe techniques will elucidate further hidden problems. REFERENCES [1] GIACCHETTI, G ., RÄNSCH, J . , Proc. 5th Int. Congr. X-ray Optics and Microanalysis, Tübingen, 1968, Berlin (1969) 250. [2] KEGEL, B. , Kerntechnik 11 (1969) 631. [3] BRAMMAN, J .I ., SHARPE, R .M ., THOM, D ., Y ATES, G ., J. Nucl. Mater. 25 (1968) 201. [4] EWART, F. T ., HORSPOOL, J .M ., JAMES, G ., private communication. [5] KLEYKAMP, H ., Mol-7A experiment, unpublished results. [6] HOLLECK, H ., KLEYKAMP, H ., J. Nucl. Mater. 35 (1970) 158. [7] HOLLECK, H ., KLEYKAMP, H ., J. Nucl. Mater, (in press). [8] HOLLECK, H ., KLEYKAMP, H ., Rep. KFK-1181 (1970). [9] BRADBURY, В. T ., DEMANT, J .T ., MARTIN, P .M ., POOLE, D .M ., J. Nucl. Mater. 17 (1965) 227. [10] JEFFERY, В. M ., J. Nucl. Mater. 22_ (1967) 33. [11] O'BOYLE, D. R., BROWN, F. L. , SANECKI, J. E. , J. Nucl. Mater. 29 (1967) 27. [12] STALICA, N.R. , SEILS, C. A ., Rep. ANL-7575 (1969). [13] STALICA, N.R., SEILS, C. A., Ceramic Nuclear Fuels, Proc. Symp. Washington (1969). [14] FROST, B.R. T ., Ceramic Nuclear Fuels, Proc. Symp. Washington (1969). [15] DAVIES, J .H ., EWART, F. T . , TAYLOR, R .G ., Rep. AERE-R 6264 (1969). [16] DAVIES, J .H ., EWART, F .T .., Rep. AERE-R 6310 (1970). [17] KEGEL, B ., Mikrochim. Acta, Suppl. IV (1970) 179. [18] JOHNSON, C .E ., STALICA, N.R. , SEILS, C. A ., ANDERSON, K. E ., Rep. ANL-7675 (1970). [19] CROUTHAMEL, C. E. , JOHNSON, C. E ., Rep. ANL-7753 (1970). [20] O’ BOYLE, D. R ., BROWN, F. L ., DWIGHT, A. E ., J. Nucl. Mater. 35 (1970) 257. [21] DAVIES. J .H ., EWART, F. T ., J. Nucl. Mater. 41 (1971) 143. [22] HUBER, H ., KLEYKAMP, H ., Rep. K FK -1324 (1972). [23] KLEYKAMP, H ., DFR-304 experiment, unpublished results. [24] JOHNSON, C. E ., CROUTHAMEL, C. E. , J. Nucl. M ater. 34 (1970) 101. IAEA-PL-4 6 3 /9а MICROANALYSE X D'OXYDE (U,Pu)02+x SIMULANT DIFFERENTS TAUX D'IRRADIATION F. de KEROULAS, D. CALAIS, F. SCHMITZ CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Roses, France Abstract-Résumé X MICROANALYSIS OF OXIDE (U,Pu)Oz + x SIMULATING VARIOUS IRRADIATION LEVELS. Metallography and X microanalysis reveal that fission products are distributed basically in the following three ways: (a) In the form of two types of m etallic inclusions. The first type (termed 0j) has the following mean composition: Ru (55 ± 5%), Mo (35 ± 5%), Rh = 10%, Pd =* 3%, Re =* 3%. The second type (termed 02) is composed of pure rhenium, (b) In solution in the m atrix, (c) In the form of a grey phase containing barium and zirconium (B a/Z r ratio « 2) but no other fission product nor uranium or plutonium. This phase is of the barium zirconate (BaO)2ZrOz type. As regards the 0j phase, an increase in the 0 /(U + Pu + soluble F. P. ) ratio or in temperature results in oxidation of molybdenum. In addition, annealings at over 1600°C cause demixing with the appearance of a m etallic phase of the type (U,Pu)Pd3 through reaction of palladium with the mixed oxide. MICROANALYSE X D’OXYDE (U ,Pu)02 + x SIMULANT DIFFERENTS TAUX D’IRRADIATION. La métallographie et la microanalyse X permettent de conclure que les produits de fission se répartissent pour l'essentiel detrois façons différentes: a) Sous forme de deux types d'inclusions métalliques. Le premier type (appelé 0 t) a la composition moyenne suivante: Ru (55 ± 5%), Mo (35 ± 5%), Rh « io%, Pd =* 3%, Re =* 3°Jo. Le second type (appelé 0 2) est composé de rhénium pur. b) En solution dans la m atrice, c) Sous forme de phase grise contenant du baryum et du zirconium (rapport Ba/Zr =* 2) mais aucun autre produit de fission, ni uranium, ni plutonium. Cette phase est du type zirconate de baryum (BaO)2 Z r02. En ce qui concerne la phase 0lf une augmentation du rapport 0 /(U + Pu + P. F. solubles) ou de la température conduit a' une oxydation du molybdène. En outre, des recuits effectués au dessous de 1600°C provoquent une démixion avec apparition d’une phase métallique, du type (U,Pu)Pd3 par réaction du palladium avec l'oxyde mixte. 1 - INTRODUCTION. Dans le cadre d’un programme de simulation de l'état chimique du combustible oxyde mixte (ü?Pu)o + x après irradiation [l], nous avons été conduits à analyser à la microsonde électronique des oxydes correspondant à divers taux de combustion. Un combustible nucléaire (U,Pu)o + x contenant des produits de fission sous forme inactive est facilement utilisable pour les examens et mesures physiques qui sont rendus difficiles sur un oxyde irradié. Il s'agit de connaître par microanalyse X la répartition des produits de fission et la nature des phases révélées par micrographie, en fonction de divers paramè tres : taux de combustion, traitements thermiques, écarts à la stoechiométrie. Une microsonde САЙЕСА est u tilisée. 167 1 6 8 de KEROULAS et al. • On a analysé : a) des oxydes mixtes simulant 'divers taux de combustion : - 17 500 MWj/t - 50 000 MV/j/t - 100 000 wï/j/t - 150 000 KVfj/t b ) des oxydes renfermant certains produits de fission choisis de façon à analyser leur comportement spécifique (simulations partielles) : - (U,Pu)0 + Ho (3,65°/° en poids), - (U,Pu)02 + Ru (3,85°/° en poids), - (u,Pu)o + Ce (3,55°/° en poids) + Ba (0,85°/° en poids) + Zr (1,10°/° en poids). Pour le s deux premiers types d 'é c h a n tillo n s, l e s teneurs de moly bdène et de ruthénium correspondent à la somme des produits de fission métal liq u es d'une sim ulation de 150 000 MWj/t. Dans le troisièm e type d 'é c h a n til lons, le cérium représente les terres rares et 1'yttrium d'une part et le baryumrep résente le baryum et l e strontium d 'au tre p a rt, pour иле sim ulation de 150 000 MWj/t. c) des oxydes mixtes avec du zirconate, de baryum (simulations partielles) : - (u?Pu)o^ + zirco n ate de baryum ( 100/0 en poids), — UO^ + zirconate de baryum (l0°/° en poids). Ces expériences sont faites pour étudier des mécanismes de réaction. Les quantités de zirconate de baryum introduites dans les échantillons sont arbitraires. On doit souligner que le mode d'introduction de Ba et Zr n'est pas le même dans c) (ZrO^.BaO) et dans b) (Zr et Ba métalliques). 2 - PREPARATION DES ECHANTILLONS. a) Les pastilles de combustibles simulés, de forte densité, sont préparées par frittage durant 100 heures au voisinage de 1 700°C dans des ampoules Pu02 en iridium scellées. Pour les oxydes mixtes, le rapport ------e st U02 + Pu02 0,20 et le rapport oxygène sur métal varie de 1,96 à 2,00 après fabrica tio n . Nous appelons rapport ^ de l'oxyde simulé le rapport du nombre to tal d'atomes d'oxygène liés dans la matrice à des atomes métalliques au nom bre total d'atomes métalliques de cette matrice (U + Pu + produits de fission) IAEA-PL-463/9a 169 A un rapport “ donné, correspondant à un taux de combustion choisi, est lié un rapport —— ^ — bien déterminé» U + Pu b) Les échantillons correspondant aux simulations partielles sont fabriqués sous atmosphère d1 argon-hydrogène (10°/°J à 1 75&°C (12 heures). c) Les traitements thermiques effectués sur tous ces échantillons se font dans des fours à vide où la pression d'oxygène croît avec la température. Ceci a pour effet, par rapport aux conditions initiales, d'augmenter le rapport ^ . 3 - REPARTITION DES PRODUITS DE FISSION . La micrographie met en évidence dans la matrice d'oxyde mixte des inclusions blanches métalliques de produits de fission et une phase grise (f igure 1). La microanalyse X montre : ï) Deux types d'inclusions métallicues. - les unes (^ ) contenant le molybdène et le ruthénium comme éléments majo ritaires avec du rhodium, palladium et rhénium en quantité moindre (figure 2 et 7) ; il arrive que (0 ) renferme de l'uranium et du plutonium. FIG. 1. Aspect micrographique d’un oxyde mixte (U,Pu)02 simulant un taux de combustion de 150 000 MWj/t. On observe la matrice oxyde; une phase brillante qui peut, suivant la température du traitement de frittage, présenter une démixion; une phase grise. 170 de KEROULAS et al. FIG. 2, Microanalyse X d'une inclusion métallique 0^ FIG. 3. Microanalyse X de la phase 02. - les autres (fi ) constituées de rhénium quasi-pur avec parfois des traces d'autres produits de fission; le rhénium remplace le technétium dans les expériences de simulation (figure 3). La proportion des produits de fission dans les inclusions métalliques varie d'un échantillon à l'autre et d'une inclusion à l'autre dans un échantillon donné, en fonction des caractéristiques de chaque oxyde. IAEA-PL-463/9a 171 2) Des produits de f is s io n en s o lu tio n . Dans la matrice d'oxyde on révèle par microsonde la présence de zirconium et de molybdène ainsi que celle de quelques terres rares (La, Pr, Ce) qui y sont solubles. Les teneurs obtenues sont en général inférieures aux valeurs nominales. 3) Des inclusions non métalliques (oxydes mixtes). Le baryum et une p a rtie du zirconium se trouvent rassem blés en une phase grise dont la composition correspond à celle d'un zirconate de baryum. Cette phase peut contenir de l'uranium et du plutonium. Exemple de ré p a r titio n des produits de f is s io n : oxyde ( u ,P u )o ; 17 500 MWj/t ; “ = 1 ,93 ; ---° p- = 1,96 ; 1 700°C, 48 heures. Inclusions métalliques. Л- Ru 55 - 5°/° Ko 35 - 5°/° Rh = 10°/° Pd = 3°/o Re = 3°/° fi 2 P1137 Micrographiquement on ne distingue pas ^ de . - En so lu tio n dans la m atrice : Zr 0, 30°/° Mo 0,05°/ ° La 0 ,5 °/° Ce 0 ,5 °/° Pr 0 ,5 °/° Bs - Phase grise contenant du baryum et du zirconium (rapport — = 2) mais aucun autre produit de fission. On n'y décèle ni uranium, ni plutonium. C ette phase est du type zirco n ate de baryum (B a 0 )2Z r 0 2. 4 - INFLUENCE DE LA TEMPERATURE ET DU RAPPORT ^ . Toutes les microanalyses X faites sur des combustibles recuits en tr e 1 750°C et 2 400?C, montrent une t r è s importante m igration du molybdène des inclusions vers la matrice oxyde. Dans ces inclusions la teneur en molybdène tombe de 35°/° à des valeurs in fé rie u r e s à 5°/°. S i la durée et la température du traitement sont suffisamment élevées, la phase fi est alors essentiellement constituée de ruthénium avec des traces de rhodium, le palla dium ayant disparu en raison de sa tension de vapeur relativement importante. 172 de KEROULAS et al. Ces analyses ont porté sur des oxydes simulant des taux de combus tio n de 17 500 MWj/t (“ i n i t i a l = 1,957 —— initial = 1 , 96 ) et de 50 000 MWj/t initial = 1,97 -■ " initial = 1 ,98). Si dans les expériences de simulation on est maître du rapport ^ 0 (donc du rapport ^ ^ ) lors de la fabrication, il n'en est plus de même au cours des divers traitements thermiques ultérieurs. En effet, ces recuits sont effectués dans des fours à vide ou l'écart à la stoechiométrie de l'oxyde est fixé par la pression partielle d'oxygène, qui croît avec la température. Il s'ensuit que l'évolution du molybdène dans les oxydes est gouvernée non plus par le rapport ^ initial mais par cette pression d'oxygène. Dans nos conditions expérimentales de recuit, les rôles respectifs de la température et du rapport — sont donc extrêmement l i é s „L'oxyde de moly- M bdène a une énergie lib r e vo isin e do c e l l e de (U;Pu)o.> [ s ] . A une température donnée, on peut par conséquent s'attendre à une oxydation du molybdène pour des valeurs de — suffisamment élevées suivant la ré a c tio n : (U,Pu)02 + x + ^ Mo - ^ Mo02 + (ü,Pu)02 + x _ y U = 0) (1 ) Dans le cas de la simulation où le seul produit de fission intro duit est le molybdène (3>66°/° de Mo, 1 700°C, 100 h ), le rapport ^ ^1 • obtenu lors de la fabrication est 2 et la réaction (l) peut se produire. Les inclusions métalliques de molybdène pur sont alors entourées d'un liseré FIG. 4. Inclusion métallique entourée d'un liseré d'oxyde de molybdène. IAEA-PL-463/9а 173 d'oxyde MoO^ (figure 4). En outre, on constate alors que la matrice contient nettement plus de Mo dissous que dans les simulations globales : 5 000 et 500 ppm respectivement. Cette dernière valeur correspond à la microanalyse X de l'oxyde mixte simulant 150 000 MV/j/t de rapport ^ = 1,975. Dans aucun échantillon il n'a été constaté de variation de composi tion des inclusions métalliques ф très riches en rhénium. 5 - FORMATION DE PHASE DU TYPE (u,Pu)Me^. L'analyse .de deux oxydes simulant 50 000 et 150 000 MWj/t, élaborés par frittage à 1 750°C Í 50°C et ayant subi un traitement d'oxydo-réduction à 800°C durant quatre heures pour ramener le rapport J à 2, montre un dédou blement des inclusions 0^ initialement monophasées (figure 5 : faible gros sissement et figure 6 : fort grossissement). Cette demixion peut aussi s'ob server directement après frittage (figure 1) ; ce fut le cas de la simulation correspondant à 100 000 MWj/t. Les deux phases issues de ji ont les compositions suivantes : - la première, située au centre des inclusions,contient le molybdène, du rho dium, le ruthénium, du palladium et le rhénium (figure 7) ( c o lo r a tion blanche); FIG, 5. Démixion des Inclusions 0, après traitement d'oxydo-réduction. (Faible grossissement. ) 174 de KEROULAS et al. FIG. 6. Démixion des inclusions 0j après traitement d’oxydo-réduction. (Faible grossissement. ) FIG. 7. Microañalyse X de la phase 0j après démixion (phase blanche). IA EA -PL-463/9а 175 (RhLß, {Pd Lot, INCLUSION BIPHASEE PHASE SOMBRE U Ld, MATRICE 20kVMICA(002) (UPu)O 4 0 0 * c/m in 4*10* c/m¡n 25 kV 11011) INCLUSION BIPHASEE PHASE BLANCHE PHASE SOMBRE 10* c/m in 25 kV (1011) 25 kV (1011) 4«10‘ c/min 7°50 PdLp, U ila . PuLd., RhL<*. К \J 7 15 l°10 7°10 8° 7°20 14° 13° 12° FIG. 8. Microanalyse X de la phase 0, après démixion (phase sombre). - la seconde, placés en périphérie, est constituée de palladium, de rhodium, d'uranium et de plutonium (figure 8) (coloration foncée). La for mule de c e tte dernière phase e st (U,Pu) (PdPôi)^. 6 - ANALYSE DE LA PHASE GRISE. Dans les échantillons circulant 17 500, 50 000 et 150 000 MWj/t, on observe une phase grise constituée de zirconate de baryum de formule (BaO) ZrO (rapport ~ déterminé par microanalyse X équivalent à 2). Cette phase est inerte vis-à-vis de l'oxyde (U,Pu)0 + c'est-à-dire qu'elle ne contient ni uranium, ni plutonium (figure 9) . " Par contre,' dans les simulations partielles * : - (u Pu)o + zirconate de baryum (l0°/° en poids), 1 750°C, 4 heures, - U02 + zirconate de baryum (l0°/° en poids), 1 750°C, 4 heures, la phase grise contient des proportions importantes d'uranium et de plutonium (de l'ordre de 10 à 15°/° en poids) (figure 10). Il s'agit d'une solution so- Ba. lid e du type (Ba o) (U,Pu Zr)0„ (le rapport ------déterminé par rai- ’ ' 2 Zr+Pu+U croanalyse X est équivalent à i). Ces expériences ont été faites pour étudier d'une manière plus simple les mécanismes de réaction. 176 de KEROULAS et al. 35kV 4*104 c/min э —I CL Э t t 7 ° 3 0 8 °______ FIG. 9. Microanalyse X de la phase grise supposée être (Ba0)2Z r02. FIG. 10. Microanalyse X de la phase grise supposée être (BaO) (U, Pu, Zr) 0 2. IAEA-PL-463/9a 177 C'est-à-dire que la réaction 1 Ba 0 Zr 02 + (u,Pu) ~ (Ba o) [(ü ,P u ,Z r)o ] + (ü,P u,Z r)02 (2) serait possible suivant 1, alors qu'il n'en est pas de même pour la réaction 1 (Ba 0)2 Zr02 + (u,Pu)02 ç (Ba o) (u,Pu,Zr)c>2 (3 ) REFERENCES [1] SCHMITZ, F., Rapport CEA-3795 (l969) . GEAP 5753 (i969). KEROULAS (de), F ., SCHMITZ, F ., Communication au "The winter meeting of the American Nuelear Society"^ Oct. 1971 - Miami-Beách - Floride. [2] DAVIES, J.H ., EWART, F.T., et TAYLOR, R.G., AERE-R-6264 (l969). MARTIN, T.L., et Me IVER, E .J., Third International Conference on Plutonium - -Hovembre 1965. E d ité p a r KAY, A .E ., etWALDRON, M.E., In s titu te of M etals, СKAPMANH and HALL - LOKDON. IAEA-PL-46379b MIGRATION DES PRODUITS DE FISSION RADIOACTIFS DANS DES COMBUSTIBLES (U, Pu)C>2 M. MOUCHNINO CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Rosès, France Abstract-Résumé MIGRATION OF RADIOACTIVE FISSION PRODÚCTS IN (U ,P u )02 ГО ELS. The radial spectrometry technique has made it possible to study the distribution of radioactive fission products in various types of mixed oxides irradiated under widely differing conditions. The aim of applying this form of spectrometry is to derive criteria for temperature and stoichiometry and to furnish data additional to those yielded by the post-irradiation examination of the fuel. MIGRATION DES PRODUITS DE FISSION RADIOACTIFS DANS DES COMBUSTIBLES (U ,P u )02. La technique de specttométrie radiale a permis d'étudier la répartition des produits de fission radioactifs dans différents types d'oxydes mixtes irradiés dans des conditions très diverses. Le but visé est d'obtenir, â partir de la spectrométrie, des critères concernant la température et la stoechiométrie et de fournir des renseignements complémentaires aux examens faits sur le combustible après irradiation. I - INTRODUCTION : . - .. Les produits de fission formés en cours d'irradiation peuvent être classés selon leur état d'oxydation dans une matrice d'oxyde mixte (U, Pu)C>2 + x en 3 groupes distincts : . - groupe des éléments ne s'oxydant normalement pas (Ru - Rh - C s.. ) - groupe des éléments's'oxydant (Zr, Ce, Pr, Ba, La, S r ..) : - groupe intermédiaire des éléments dont le-degré d'oxy dation est fonction de la stoechiométrie et de la tem pérature (Mo, Nb). • ■ Les études faites par micro-analyse X des produits de fission / Г - _27 donnent des résultats conformes aux prévisions thermodynamiques. Ainsi on a bien observé par,exemple,: . , - des inclusions métalliques contenant les éléments Ru, Rh - des inclusions d' oxydes "contenant Ba, Ce, Sr. La technique de spectrométrie radiale nous a permis d'étudier la répartition des produits de fission radioactifs dans différents types d'oxydes.mixtes irradiés dans des .conditions très diverses. Le but visé est d'obtenir, à partir de la spectrométrie, des critères concernant la température et la stoechiométrie et de fournir des renséignements complémentaires aux examens faits sur le combustible après irradiation. : 179 1 80 MOUCHNINO II - TECHNIQUE EXPERIMENTALE La mesure de l'intensité des raies gamma, émises par un combustible irradié, au moyen d'une diode Ge-Li, permet d'obtenir la répartition radiale (ou longitudinale) de certains P. F. Les caractéristiques de la technique sont les suivantes : 1 - Les P. F. dont l'activité résiduelle est suffisante sont seuls susceptibles d'être détectés. Pour des temps de refroidissement de 1 mois on peut détecter les nuclides suivants : 1131 Ba La 140 Ce 141 Zr 95 Pour des temps de refroidissement de 3 mois, on peut encore trouver les nuclides suivants : Ce Pr 144 Zr Nb 95 Ru 103 Cs 137 Pour des temps de refroidissement de 1, 5 an on ne peut analyser que les nuclides suivants : Ce Pr 144 Ru Rh 106 Cs 137 2 - La résolution énergétique de la diode utilisée est de 3 keV. 3 - La résolution géométrique, fonction de l'intensité des P. F. émis, est en moyenne de 0, 3 x 0, 5 mm , la surface du colli mateur utilisé. La résolution limite est de 0,1 x 0,1 mm . Toute l'émis sion mesurée se rapporte à une telle surface et en particulier les P. F. précipités ne peuvent pas être isolés. 4 - Les valeurs directement obtenues sont relatives. On peut les étalonner en assignant à un des nuclides, le zirconium par exemple, la valeur déterminée par le calcul. Le code de calcul utilisé, PICFEE¿ 3 J implique la connaissance des puissances linéaires en fonc tion de la température. Les courbes absolues tracées en tirets sur les figures servent de référence pour déterminer s'il y a ou non transfert de matière (enrichissement ou appauvrissement local). 5 - La détermination de la répartition des nuclides est faite sur un diamètre. 6 - La méthode se prête à-l'étude d'ensemble des P. F. contenus à l'intérieur d'un crayon expérimental. IAEA-PL-463/9b 181 III - ECHANTILLONS EXAMINES La liste des échantillons examinés est portée sur le tableau I avec leurs caractéristiques d'origine (O/M, densité), le taux de combustion, la puissance linéaire moyenne. Ils ont été séparés en 2 groupes en fonction des réacteurs où s'est déroulée l'irradiation : Rapsodie (neutrons rapides), EL3 (neutrons thermiques). Notons que les échantillons M appartiennent à un lot d'oxyde avec addition de P. F. non radioactifs et soumis à une faible irradiation dans EL3 j \ ] . I V - RESULTATS IV. 1- Irradiation en neutrons thermiques : échantillons M Le creusement de flux à l'intérieur des échantillons se traduit par des courbes décroissant de la périphérie vers le centre. IV. 1-1 Zirconium-95 D'après la méthode choisie (II - 2) les courbes de répar tition expérimentale et calculée coincident, au moins pour les échantil lons à fortes densitfe initiales (S 13 et S 21). Sur'les échantillons de faible densité (S 14 - S 24) l'effet de densification se traduit par un décalage entre les 2 courbes, expérimentale et calculé^ croissant vers le centre de l'échantillon ig. 17. IV. 1-2 Praséodyme-141 Les courbes de répartition ne présentent pas d'anomalies ¿Fig. 27. IV. 1-3 Ruthénium-103 /Fig. 3/ On obtient 3 allures de courbes : a) courbe décroissante de la périphérie vers le centre (S 21) et coïncidant avec la courbe théorique calculée. b) courbe présentant 2 pics - un pic périphérique sans enrichissement par rapport à la courbe de référence. - un pic, enrichi, situé dans la zone des grands grains colonnair'es. Sur les autoradiographies £\/ ce genre de répartition se traduit par 2 anneaux.le ruthénium-103 est en effet le nuclide dont l'ém is sion est la plus forte et il est détecté directement par cette méthode. TABLEAU Г. LISTE DES ECHANTILLONS EXAMINES Eile O/M d % dth MWj/t Puissance Rem àrques - d’irradiation linéaire W/cm "e l з Echantillons M ' S 11 2,00 93 ^ 6 000 : 450 addition Ba-Zr-Ce : s 12 1,-90 93 ' 1 f . 470 f f Neutrons ■ S 13 2, 00 97 I T Í 520 addition Ru Therm iques S 14 ' 2, 00 85 - T t : 480 Г T S 21 ■ 2, 00 98- I I , 425 addition Mo. S 22 1, 90 96, 5 f r ■ 445 addition Ce, Zr ' S 23 2, 00 96, 5 : 525 It. S 24 1, 90 82 •; 11 : 450 T T ï Rapsodie A ' •• 1,99 ' 86,4 12.000 ' 540 - 540- 572 r t 1 T T T ' A' '430 - 430- 460 ■ 3 cycles ; B ' 1,97 ; 94 . r î 510 - 510- 530 Neutrons ■B' f 1 î T r T 410 - 410- 425 Rapides . C 1,96 • .94 - M 480'- 480- 510 combustible annulaire ■ ; IAEA-PL-463/9b 183 S 11 O/M * 2, 00 ■ R (mm) S 24 Ç /M ~ l ,9 0 R(mrri) FIG. 2. Répartition îadîale du 141Pr dans deux échantillons M, 184 MOUCHNINO S 11 S 12 0/M= 2, 00 0 / M ~ 1, 90 S 21 S 22 FIG. 3. Répartition radiale du 103 Ru dans les échantillons M. c) Courbe présentant 3 pics (ou 3 anneaux à l'autoradio- graphie) : - Un pic périphérique sans enrichissement - Un pic situé à la base des grains colonnaires, enrichi. - Un pic, en général enrichi, situé . soit au bord du trou centraI(S 22 - S24) . soit à proximité du trou central (S 12) en un point intermédiaire de la zone à grains colon naires. Les courbes du type c appartiennent aux échantillons sous-stoechiométriques, les courbes du type b aux échantillons stoe- chiométriques. IV. 1 - 4 Ce sium-137 Le césium est uniquement trouvé à la périphérie des échantillons où ce nuclide a été détecté (S 22 - S 23 - S 24) après un temps de refroidissement de 150 j. Les valeurs expérimentales trouvées coincident avec celles calculées montrant qu'il n'y a pas d'enrichisse ment en périphérie.. IAEA-PL-463/9b 185 s 11 О/М = 2, 00 S 13 О/М = 2, 00 FIG. 4. Répartition radiale du 95Nb. IV 1 - 5 Niobium-95 /Fig. 47 Les allures de la répartition de ce nuclide se classent en 2 groupes selon qu'il y a ou non. un enrichissement en niobium au centre de l'échantillon (ou près du trou central). Font partie du groupe présentant un enrichissement au centre S 11 - S 13 - S 14 - S 21 (très faible enrichissement)^ 23. Ce sont donc tous les échantillons initiale ment stoechiométriques. Sur la fig.4, on montre sur le cas typique de S 13 que l 'enrichissement au centre est corrélatif d'un appauvrissement à l'intérieur de l'échantillon. IV . 2 Irradiation en neutrons rapides : échantillons A, В, C Les temps de refroidissement très longs (620 j ) de ces échantillons au moment des mesures n'ont permis de mesurer que les activités du praséodyme-144, du rhodium-106, du cesium-137. 186 MtXJCHNINO Rappelons qu'ils ont été irradiés en neutrons rapides et que l'allure des courbes de répartition calculée ne présentent pas de creusement vers le centre. IV . 2 -1 Praseodyme-144 Sur les échantillons, et en particulier sur B, le praséo- dyme est constant le long des diamètres examinés. IV . 2 - 2 Rhodium-106 On retrouve les courbes caractéristiques à 2 ou 3 pics / F ig .ÿ • a) Sur l'échantillon A, situé au flux maximum de l'aiguil le, on a l'allure décrite en IV. 1-3, à 2 pics : l'un en périphérie non en richi, l'autre, en deçà du trou central, enrichi. Sur l'échantillon A1 situé à 140 mm du flux maximum, on a un seul pic d'enrichissement, au centre de l'échantillon. b) Sur l'échantillon B, situé au flux maximum,on a l'allure décrite en IV . 1-3, à 3 pics : - Un 1er pic non enrichi en périphérie, - Un 2e pic enrichi à la base des grains colonnaires, - Un 3e pic enrichi en deçà du trou central. A - Flux max. . A' - 80 % Flux max. O/M = 1, 99 o/M = 1, 99 O/M =1,97 O/M = 1,97 * O/M = 1,96 FIG. 5. Répartition radiale du i06Rh dans les échantillons A, B, C. IAEA-PL-463/9b 187 A' --F lu x max. A' - ' 80 % Flux max. FIG. 6. Répartition radiale du 13,Cs dans les échantillons A, B, C. SürT échantillon" B' situé à 140 mm du flux maximum on a un seul pic d'enrichissement au trou 'central. .... c) Sur l-1 échantillon C on note un appauvrissement à proximité du trou central, avec une remontée, à l'activité calculée, au trou central. , . , IV . 2 - 3 Césium-137 / F i g .i/ ' Les pics périphériques dé césium montrent un étroit palier situé du côté de l'oxyde et au niveau de l'activité calculée. Le sommet du pic n'excède pas de plus de-50 % celui du palier, sauf sur l'échantillon A où .'l'enrichissement maximum atteint 2, 5 fois celui de l'activité, prévue (au palier). V - INTERPRETATION V - 1 P. F. solubles : Zr - Pr On ne constate aucune migration, en particulier sur les échantillons S. 11 - S 24 - B où une.diffusion de plutonium a été mise, en évidence .£§J . 188 MOUCHNINO V - 2 Migration des P. F. ne s'oxydant pas Ru Rh, Cs V - 2-1 Cas du ruthénium et du rhodium Bien que ces nuclides ne s'oxydent pas, l'influence de la stoechiométrie de l'oxyde est prépondérante sur leur migration. Il y a 2 allures de courbes de répartitions : à 2 pics pour les oxydes stoechio- métriques, à 3 pics pour les oxydes sous-stoechiométriqués. a) 2ème pic (à la base des grains colonnaires) La première vallée qui sépare le premier pic,non enrichi, du deuxième pic enrichi est appauvrie par rapport à la courbe de référen ce. Elle est caractéristique d'une migration, dans le sens du gradient thermique, du rhodium (ou du ruthénium). /~6_7 donne aussi l'exemple d'accumulation des P .F . à cet endroit dans un oxyde sous-stoechiométrique et l'attribue à l'état de non oxydation de ces P. F. Dans le cas général, ce point de vue est valable. Dans le cas du Ru - Rh, l'influence de la sous-stoechiométrie ne joue pas. L'enrichissement du 2ème pic pourrait s'expliquer par migration du rho dium de la 2ème vallée, appauvrie. £ l _/constate un phénomène analogue sur un oxyde légèrement surstoe- chiométrique et l'explique par une évaporation - condensation du rhodium à l'intérieur des bulles migrant, qui conduit à un enrichissement en rho dium à l'opposé du déplacement des bulles, c'est-à-dire vers les parties froides de la zone à grains colonnaires. Un tel type d'explication étant indépendant de la stoechiométrie de l'oxyde devrait donner naissance au 2ème pic de rhodium dans les oxydes stoechio- métriques. b) 3ème pic d'enrichissement (oxydes stoechiométriques et non stoechiométriques). Nous constatons par contre que ce 3ème pic situé à l'intérieur des grands grains colonnaires se retrouve pour les 2 types d'oxyde. On peut expliquer sa formation par une migration du rhodium dans le sens de gradient thermique, liée au déplacement des bulles se déplaçant dans les grains colonnaires. c) Vaporisation - condensation axiale Le flanc du 3ème pic situé vers le trou central peut s'appauvrir et le 3ème pic disparaître complètement par vaporisation du rhodium dans la cheminée centrale. Inversement, au bord du trou central il peut y avoir enrichissement en Rh par condensation. Ainsi, sur l'échantillon B, on constate un appauvrissement sur le flanc du 3ème pic; sur l'échantillon B', il y a enrichissement en rhodium le long du trou central par condensation, l'oxyde étant plus froid à ce niveau. V. 2 - 2 Césium On met en évidence sur le pic périphérique du césium un palier dont l'activité est celle prévue par le calcul. Ceci est certai- IAEA-PL-463/9b 189 nement un phénomène général que la spectrométrie radiale ne peut pas toujours détecter du fait de sa résolution géométrique limitée. On trouve une zone en périphérie de l'oxyde où le césium n'a pas migré et par rap port à laquelle on peut dire qu'il y a une zone extérieure, vers la gaine, enrichie ou non en césium, et une zone intérieure vers l'oxyde qui est fortement appauvrie et dans laquelle le césium migre vraisemblablement comme les gaz rares vers le centre. V. 3 'Migration des P. F. pouvant s'oxyder : cas du Nb On constate sur les oxydes stoechiométriques une accu mulation de niobium au centre de l'échantillon (expérience M). Il est vraisemblable que le niobium ne migre qu'à l'état métallique. C'est donc que la température est suffisante pour que le potentiel d'oxygène de l'oxyde mixte d'uranium et de plutonium soit inférieur à celui de l'oxyde de niobium, c'est-à-d ire comprise entre 2 200 °C et 2 300° C, ce qui est tout à fait vraisemblable pour les puissances linéaires des expérien ces où le phénomène a été constaté. VI - CONCLUSIONS Nous avons mis en évidence de nombreux phénomènes de migration deproduits de fission radioactifs qu'ils soient métalliques ou qu'ils deviennent métalliques à haute température. Dans tous les cas, la migration semble s'effectuer dans le sens du gradient de tempé rature. Elle est vraisemblablement liée au déplacement des bulles, soit dans la partie périphérique de l'oxyde, avec possibilité de blocage de cette migration à la base des colonnaires dans le cas des oxydes sous- stoechiométriques (rhodium), soit dans la partie située à l'intérieur de la zone des grains colonnaires (Rh - Nb). La migration dépend simultanément des conditions initiales de stoechio- métrie et des températures atteintes en cours d'irradiation. Des phéno mènes de vaporisation - condensation ont été aussi mis en évidence dans la cheminée centrale. Le césium se redistribue entre l'oxyde et la gaine après avoir migré à l'intérieur de l'oxyde dans le sens du gradient ther mique. REFERENCES [1 ] O'BOYLE, D .R ., BROWN, F. L ., SANEKI, J. E ., JMN 29 (1969) 27-42. [2] DEKEROULAS, F.. SCHMITZ, F.. ANS Transact 14 2 (1971). [3] BARRE, H ., Note CEA-N-1023 (1969). [4 ] CONTE, M ., MOUCHNINO, M .. SCHMITZ, F . , Congrès ANS (Richland, avril 1972).- [5 ] MOUCHNINO, M ., «Thermodiffusion et vaporisation dans les oxydes mixtes irradiés - Evolution du rapport Pu/U + P u » , ces comptes rendus, IA EA -PL-463/19b. [6 ] NEIMARK, L. A ., LAMBERT, J .P .B ., MURPHY, W .F .. RENFRO, C . W ., Congrès ANS (Richland, avril 1972). [7 ] BAHL, J .K ., FRESHLEY, M. D ., BNWL SA -3441 (1971). IAEA-PL-463/9C DISTRIBUTION DES PRODUITS DE FISSION ET LOCALISATION PAR SPECTROMETRIE GAMMA EN COURS D'IRRADIATION ET APRES IRRADIATION G. de CONTENSON, J. MONIER, Nicole VIGNESOULT CEA, Centre d'études nucléaires de Saclay, . Gif-sur-Yvette, France Abstract-Résumé DISTRIBUTION OF FISSION PRODUCTS AND LOCALIZATION BY GAMMA SPECTROMETRY DURING AND AFTER IRRADIATION. Gamma spectrometry makes it possible to identify and enumerate gamma-emitting radioactive atoms. It is used during and after irradiation to determine the longitudinal and radial movement of the fission products. Examinations during irradiation are carried out directly through the irradiation device after .each reactor cycle or, in the case of accident, immediately after unloading. These examinations reveal accu mulations of fission products accompanying movements of matter, for example, in the case of fuel exhibiting core-melt characteristics. The post-irradiation examinations, performed in a hot cell, are carried out longitudinally and transversally on the pins before any destructive investigation is made, and radially on micrographie sections. These sensitive analyses make it possible to localize, after irradiation, fission products both in the U 0 2 and in the fuel element structures. Examples of localization of fission products are given for the case of an oxide which has undergone limited melting. DISTRIBUTION DES PRODUITS DE FISSION ET LOCALISATION PAR SPECTROMETRIE GAM M A EN COURS D ' IRRADIATION ET APRES IRRADIATION. La spectrométrie gamma permet d'identifier et de dénombrer les atomes radioactifs émetteurs gamma; elle est utilisée en cours d'irradiation et après irradiation pour la détermination des évolutions longitudinales et radiales des produits de fission. Les examens en cours d'irradiation sont réalisés directement à travers le dispositif d'irradiation après chaque cycle de pile ou, en cas d'incident, immédiatement après le déchargement de la manipulation. Ces analyses permettent de déceler les rassemblements de produits de fission accompagnant les mouvements de matière, par exemple, dans le cas de combustible présentant une fusion à cœur. Les examens après irradiation, réalisés en cellule chaude, sont effectués longitudinalement et transversalement sur les crayons avant tout examen destructif, et radialement sur les coupes micrographiques. Ces analyses fines permettent de localiser, en fin d* irradiation, les produits de fission aussi bien dans l'U O j que dans les structures des éléments combustibles. Des exemples de localisation de produits de fission seront présentés dans le cas d ’ oxyde ayant subi une fusion limitée. I. INTRODUCTION L'analyse par spectrométrie gamma, permettant de déceler les rassemblements de produits de fission en cours d'irradiation et après irradiation, apporte.des renseignements qualitatifs sur la vie du combustible. Dans ce texte, nous nous sommes surtout attachés à illustrer cet aspect par des exemples pratiques dans le cas des combustibles oxyde d'uranium : éléments combustibles fonctionnant à coeur fondu ou à température élevée mais sans fusion, crayons ayant subi des ruptures de gaine. 191 BAS H A U T e OTNO e al, et CONTENSONde F I G .l. Evolution du y scanning à EL3 de l'essai TOF 7. IAEA-PL-463/Эс 193 II. EXAMENS EN COURS D'IRRADIATION Afin d'etudier et de suivre les mouvements de combustible (UO2) et des produits de fission en cours d'irradiation, de détecter des anomalies et de confirmer les ruptures de gaines, les containers placés dans la pile EL 3 subissent systématiquement une scrutation gamma après chaque cycle de pile (environ tous les mois et demi). Cette opération est réalisée immédiatement après la fin du cycle ou, en cas d'incident, après le déchargement de la manipulation. Pour cela, on utilise le canal de radiographie situé dans les installations de servitude du réacteur EL 3 du Centre d'études nucléaires de Saclay. L'intensité gamma recueillie est proportionnelle: — d'une part, aux quantités d'U0 2 rencontrées localement, — d'autre part, aux rassemblements de produits de fission particulièrement à vie courte (lanthane, xénon, iode). Cette méthode permet tout particulièrement d'arrêter un essai lorsqu'une trop grande activité gamma (ponts d'U0 2 ou de produits de fission) est observée sur un point vital de l'élément combustible. La figure 1 donne un exemple de l'intérêt de cette méthode. La première courbe de scrutation gamma nous montre un profil caractéristique de combustible ayant fonctionné avec fusion à coeur. En effet, nous avons lors du refroidissement une coulée d'UOä vers le bas, et à la solidification une formation de ponts d'U0 2 et de trous très importants. Ces rassemblements d'U0 2 se différencient des rassemblements de produits de fission, d'une part par des irrégularités nombreuses dans la courbe de spectrométrie gamma, d'autre part, un pont d'U0 2 est toujours accolé à un trou comme on peut l'observer dans la partie droite de la courbe. Cette courbe est surtout représentative de la répartition du 140La dans l'élément combustible. Les deux autres courbes indiquent une restructuration normale de l'U 0 2 , la puissance ayant décru durant le deuxième cycle, et la présence d'un pic de produit de fission dans la partie froide du crayon (près du volume libre). Le pic observé est dû à des produits de fission volatils (Cs et I) qui se sont accumulés dans la pastille de graphite supérieure. La présence de ces produits de fission a été confirmée par des examens fins en laboratoire chaud. De même, des rassemblements de produits de fission solides ont été observés et retrouvés par scrutation gamma fine et examens micrographiques. III. EXAMENS APRES IRRADIATION 1. Dispositif expérimental Cette technique, qui permet d'identifier et de dénombrer les atomes radioactifs émetteurs gamma, estutilisée couramment en laboratoire chaud pour la détermination des évolutions longitudinales et radiales des produits de fission. Les équipements ayant déjà été décrits [1 ], nous rappelons seulement qu'ils se divisent en deux parties: — les dispositifs de positionnement et de focalisation qui doivent être adaptés à chaque cas d'analyse (dispositif destiné aux analyses longitudinales pour 194 de CONTENSON et a l. les éléments combustibles de grande longueur et dispositif destiné à l'analyse radiale des éléments combustibles); — les dispositifs de détection et de traitement de l'information qui peuvent être identiques pour plusieurs types d'analyses. Les analyses effectuées peuvent être qualitatives, semi-quantitatives ou quantitatives. Le traitement de l'information demande des moyens importants; en effet, la précision de la technique dépend du nombre de points examinés. Pour améliorer l'acquisition et pour traiter rapidement les informations, le LECI s'est équipé d'un système dont l'élément de base est un calculateur permettant d'effectuer automatiquement les opérations suivantes ; a) acquisition simultanée des données sur 2 chaînes d'analyse, b) stockage en mémoire de spectres enregistrés, c) calcul et présentation des résultats sous forme d'histogrammes. Rappelons que ces examens étant effectués après un temps de refroidissement assez long, les éléments à vie courte sont rarement observables. Les résultats permettent de mettre en évidence la répartition longitudinale ou radiale de l'activité, soit totale, soit d'un ou de plusieurs P. F. ayant un rôle particulier. Des exemples d'application sont présentés dans le cas d'éléments oxydes fonctionnant avec fusion limitée, à température élevée mais sans fusion, et de crayons rompus. 2. Eléments fonctionnant avec fusion limitée de l'oxyde Sur un crayon qui a fonctionné constamment avec fusion limitée à fort taux de combustion (60 000 MWj/t d'U), la répartition d'activité longitudinale permet de mettre en évidence les points particuliers suivants (figure 2 ): — une répartition d'activité très perturbée avec des pics et dépressions caractéristiques d'un élément ayant fonctionné avec fusion de l'oxyde; — la présence de pics de Cs au droit des pastilles de graphite qui servent de séparateur (point froid) des colonnes combustibles en fusion; la dégradation du graphite observée pour des taux de combustion élevés est probablement liée à la présence des Cs, le graphite se comportant comme un bon piège à Cs; — des accumulations importantes de ruthénium et de rhodium en deux zones bien déterminées de l'étage haut; des analyses à la microsonde ont précisé les compositions de ces produits de fission. Sur ce même crayon, une analyse radiale de la répartition des produits de fission (137Cs, 103Ru + 106Rh, 144Ce) met en évidence trois zones de concentration (figure 3) ; a) Le césium est principalement à la périphérie, ce qui a été confirmé lors de la microanalyse X. b) Les ruthénium sont situés principalement sur une couronne à 3, 4 mm du centre. c) Le cérium est réparti dans tout le combustible. Il faut insister sur le fait que cette analyse a été faite avec une collimation de 2 mm, ce qui explique -la forme de la répartition du cérium. Le système de collimation ayant été récemment amélioré, des analyses plus fines peuvent maintenant être effectuées et un exemple est IAEA-PL- 463/9c 195 FIG .2. Analyse longitudinal de produits de fission dans un oxyde fondu. montre sur la figure 3. On remarque que l'augmentation de la concentration en Z r est limitée à la zone fondue dans le cas de refroidissements lents; ceci a été confirmé par des analyses à la microsonde. Des analyses détaillées au Ge-Li, effectuées sur des coupes longitu dinales d'oxyde partiellement fondu séparées par des pastilles de graphite, ont permis de confirmer la présence de 134Cs et de 137Cs dans le graphite, les éléments 95Z r, 144P r, 95Nb ne pénétrant pas du tout dans ce matériau (figure 4). L'importance de cette diffusion du césium est liée au taux de combustion et à l'état du combustible en contact avec le graphite jusqu'à environ 25 000 MWj/t d'uranium; cette diffusion est sans importance sur le comportement de ces pastilles de graphite. Par exemple, dans un des cas présentés, la voûte d'UC>2 s'est effondrée au cours de l'irradiation. Une explication possible est que le flux des atomes de césium se dirigeant vers les zones les plus froides serait venu directement en contact avec la pastille de graphite et aurait engendré une dégradation de cette dernière. 3. Eléments fonctionnant à des températures inférieures à la température de fusion Outre le cas particulier des oxydes fondus à coeur, la répartition des césium, dans des crayons fonctionnant à des températures maximales de l'ordre de 2200°C - 2500°C, présente un intérêt notable. 196 de CONTENSON et al. a ) FIG.3, a et b. Analyse radiale des produits de fission par spectrométrie y. 9 5 Z r + 9 5 N b G e L i Collimation 1mm TO6RI, G e L i Collimation 1mm 1 3 7 с , - G e Li Collimation 1mm 198 de CONTENSON et al. FIG.4. Répartition des activités longitudinales. Tronçons TOF 14-B1- TOF 15-B2. IAEA-PL-463/9C 199 2 10 0 5-10‘ Zr Е з = 0,724 MeV 210 0 -ю3 '" P r E ÿ = 2,18 MeV 2-10 о FIG. 5. ELP-9. Tronçon prélevé à 163 mm du début de 1' U 02 (côté bas). Longueur 100 mm, détecteur G e-Li. Des analyses globales faites à des temps de refroidissement successifs mettent en évidence: — après un temps de refroidissement de l'ordre de quatre mois, l'image de la répartition de Zr qui est en bon accord avec le flux intégré lors de la fin de l'irradiation, — après des temps de refroidissement de l'ordre de un an, l'image d'une concentration variable du 137Cs tout le long du combustible. Le césium migrant vers les points les plus froids est représentatif des irrégularités de température existant le long du crayon (variations de puissances spécifiques, des variations du coefficient de transfert thermique et de l'existence de couches isolantes au contact combustible - gaine). Ce point a été confirmé par des analyses fines au détecteur G e-Li et est illustré par la figure 5. IV. CONCLUSION La spectrométrie gamma en cours d'irradiation permet, sans dommages pour les éléments combustibles, de suivre facilement les mouvements de 200 de CONTENSON e t'a l. matière (U0 2 ou produits de fission) occasionnes par la fusion limitée de l'oxyde. Cette technique est actuellement en cours de mise au point dans d'autres installations du CEA. La spectrométrie gamma après irradiation permet surtout de localiser qualitativement les rassemblements de produits de fission et de mieux choisir les zones où seront réalisés des examens micrographiques ou des examens à la microsonde. REFERENCES [1 ] HULOT, J . P . , MANSARD, B . , MONIER, J.', ROUSSEL, E ., La spectrométrie gamma au L E C I- Possibilités offertes par l'intégration d'un petit caclulateur dans une chafne de mesure, B .I.S .T . — Publication CEA (à paraître). IAEA-PL-463/9d ANALYSE A LA MICROSONDE DE PRODUITS DE FISSION METALLIQUES DANS LES COMBUSTIBLES EN GEOMETRIE EAU ORDINAIRE FONCTIONNANT AVEC UNE ZONE FONDUE J. BAZIN, M. PERROT, Nicole VIGNESOULT CEA, Centre d'études nucléaires de Saclay, Gif-sur-Yvette, France Abstract-Résumé MICROWAVE ANALYSIS OF METALLIC FISSION PRODUCTS IN FUEL ARRANGED IN ORDINARY WATER GEOMETRY AND OPERATING WITH A FUSED ZONE. The distribution of m etallic fission products was studied in experimental fuel elements in ordinary water geometry (UOz clad with Zircaloy). The elements in question had been taken to burn-ups of 30 000 to 60 000 MWd/t U with a continuous lim ited fused zone. Some of the elem ents had been cooled rapidly, others slowly. The microwave analyses were aimed at determining the qualitative and quantitative composition of precipitated fission products (globular fission products, slugs collected in the central stacks, grey phases) and the radial distribution of fission products in solution in the U 02 matrix, principally zirconium. The results of these analyses are correlated with metallographic observations (by microscopy and auto radiography). The inspections carried out on the reaction layers at the oxide-clad interface are not dealt with. ANALYSE A LA MICROSONDE DE PRODUITS DE FISSION METALLIQUES DANS LES COMBUSTIBLES EN GEOMETRIE EAU ORDINAIRE FONCTIONNANT AVEC UNE ZONE FONDUE. La répartition des produits de fission m étalliques a été étudiée dans des élém ents combustibles expérimentaux en géométrie eau ordinaire (U02 gainé Zircaloy). Ces éléments ont fonctionné jusqu’à des taux de combustion compris entre 30 000 et 60 000 MWj/t d'U ave en permanence une zone fondue limitée. Ces éléments ont subi soit un refroidissement rapide, soit un refroidissement lent. Les analyses á la microsonde ont porté sur la composition qualitative et quantitative de produits de fission précipités (P. F. globulaires, lingots rassemblés dans les cheminées centrales, phases grises), et sur la répartition radiale de produits de fission en solution dans la matrice dTJOg, principalement le zirconium. La corrélation est faite entre les résultats de ces analyses et les faciès observés en métallographie (microscopie - autoradiographie). Il ne sera pas fait mention des examens effectués dans les couches de réaction au contact oxyde-gaine. I. INTRODUCTION Le CEA a effectué une série d'irradiations expérimentales dont le but est de définir le comportement en pile, en géométrie eau ordinaire, d'éléments combustibles fonctionnant en régime permanent avec une zone fondue limitée. Le combustible est formé d'UC>2 enrichi à 4, 5 ou 9,6% en 235U gainé en Zircaloy 2. Les puissances atteintes sont comprises entre 1100 et 1300 W/cm. Le taux d'irradiation maximal atteint est de 60 000 MWj/t d'U. La répartition des produits de fission métalliques a été étudiée par examen à la microsonde blindée. Pour la rédaction de ce texte, nous avons choisi comme exemple deux types d'éléments combustibles: a) Le crayon TOF 4 qui, après avoir atteint un taux de combustion de 60 000 MWj/t d'U, a été trempé en fin d'irradiation (chute de barre après détection de rupture de gaine). Les figures 1 et 2 montrent l'aspect macrographique de cet élément après irradiation. La zone centrale très 201 202 BAZIN et al. FIG. 1. TOF 4 — Maciographie de l'échantillon. poreuse, visible sur la figure 1 , correspond à la zone fondue en fin d'irradiation. La zone densifiée située entre les grains colonnaires et les grains equiaxes correspond à une zone fondue plus importante ayant existé en cours d'irradiation. b) Les crayons TOF 15 et TOF 11, après avoir atteint un taux de combustion d'environ 40 000 MWj/t d'U, ont été refroidis lentement (pas de rupture de gaine). La figure 3 montre l'aspect macrographique du crayon TOF 15, la zone centrale à haute densité correspond à la zone fondue en fin d'irradiation. L'autoradiographie ßy de cette section montre deux anneaux appauvris séparés par un anneau enrichi correspondant aux précipités métalliques dans les grains colonnaires. IAEA-PL-463/9d 203 ШЁЯаЁШlililí! ТШЁяШ ШШШШ / ■, ¿ * _ Л' * - , > ;?• / * 1 ИЯ 'Ifî'wÊkШ. ? • » - -Ш * Ш : . ■ ■ V* ’•■ */* V ■> ..- - • H i m ииняа i зяр* 3TWF-«^‘¡«SfiU V - '»*»#«ы V.»**. - ' .' - ' Я 1 | | р 1 Ш 1 И И и М Я » ' V ^ w ' - j f ■ к к ^ ч*.- ИНШп / ■■. ■Н а ш а; ./=. ■ ШШЛт Ш Я Ё Ё Ш ё ь , -, /■-■.":■/ : ï FIG. 2. TOF 4 — Alphagraphie de l'échantillon. Le plutonium apparaît en blanc. II. COMPORTEMENT DES PRODUITS DE FISSION 1. Produits de fission métalliques Dans ces éléments combustibles les produits de fission métalliques existent dans trois zones: a) Des lingots de forte dimension (50 mm3 vers 60 000 MWj/t d’U) sont trouvés dans la cheminée centrale du combustible. Ces lingots sont biphasés. Une analyse a été effectuée sur un lingot extrait de TOF 11. La composition moyenne est la suivante: Mo 50, 7%, Ru 28, 9%, Tc 20, 4%. Le rhodium et le palladium sont présents à l'état de traces (<0, 1%). b) Des nodules de 5 à 10 ium de diamètre sont situés dans la zone fondue de haute densité des combustibles refroidis lentement (TOF 15, TOF 11); ils sont composés presque exclusivement de molybdène (97 à 98%); les autres éléments sont Ru, Tc, Rh et Pd. 204 AI e al. et BAZIN FIG.3. TOF 15 - Structure de l'échantillon. IAEA-PL-463/9d 205 1 “/.Mo x---- * MOLYBDENE -50 x ^ x—x“ * X 1 V* Ru RUTHENIUM X -50 * - x - X---- X _ °/0 Tc TECHNETIUM - 30 * * X X X 1 °/o Rh RHODIUM - 5 X X x----- x № e/o Pd , PALLADIUM -1 X 0,5 mm X X—X—X-- X x----- x GRAINS I GRAINS COLONNAIRES ZONE FONDUE EQUIAXES I 1 2 3 4 5 I 6 7 • i i i i i i l FIG. 4. Composition des P. F. globulaires en fonction du rayon. c) Des précipités de 2 à 5 /ял. de diamètre sont situés dans les grains basaltiques. Us sont composés de Mo, Ru, Tc, Rh et Pd. Leur composition varie en fonction du rayon. Sur la figure 4 on voit que dans l'anneau où sont rassemblées la plupart des inclusions, elle varie relativement peu: de 36 à 42% pour le molybdène, de 29 à 39% pour le ruthénium, de 19 à 26% pour le technétium; le rhodium et le palladium sont toujours inférieurs à quelques pour-cent. A l'extérieur de cette couronne (point n°l) la variation est beaucoup plus importante pour le Mo (12%) et le Ru (59%), le technétium 206 URANIUM MICROGRAPHIE AI e al. et BAZIN ZIRCONIUM ■ X 264 FIG. 5. Phase grise dans la zone fondue. IAEA-PL-463/9d 207 est voisin des précédents (24%), ainsi que Rh et Pd. Les valeurs que nous venons de citer ont été obtenues sur TOF 15. L'examen de TOF 11 donne des résultats semblables. Remarque: Lors de l'exament de TOF 4 (refroidi rapidement), il a été trouvé, dans une fissure proche du centre, quelques globules métalliques contenant environ 50% de Pd, 30% d'U et 15% de Sn. Ces globules sont les seuls dans lesquels on a trouvé de l'uranium. 2. Phases grises Phase grise au centre du combustible Dès 8000 MWj/t d'U, il a été observé au centre de la zone fondue, autour des cheminées centrales de faible diamètre, des combustibles refroidis lentement, une phase grise dont les aspects sont représentés sur la figure 5. Elle se présente sous forme lamellaire ou dentritique près des fissures, ou sous forme allongée au droit des porosités, des inclusions métalliques ou des microfissures. La composition de ces phases est: uranium ~ 40% zirconium ~ 15% baryum ~ 30% oxygène ~ 13% On a également détecté quelques pour-cent de molybdène, ainsi que des traces de Cs, Ce, Sr, Nd, La, etc. Lorsque la cheminée est importante, il n'y a plus aucune trace de phase grise; en effet, à ce niveau, il n'y a pas d'UÛ2 liquide dans lequel sont dissous les éléments à base de zirconate de baryum. Remarque: Par suite d'un défaut de témoin, la concentration en baryum est surestimée et celle d'oxygène, déterminée par complément à 1 0 0 , sous-estimée. Phase grise en périphérie Ce type de phase a été remarqué en périphérie de l'oxyde très irradié (60 000 MWj/t d'U) dans les grains équiaxes. Des spectres effectués à la sonde électronique ont permis de détecter la présence de: uranium en forte proportion, baryum, cérium, néodyme et zirconium, ces éléments étant cités par ordre décroissant de concentrations estimées. La figure 6 représente l'analyse d'une de ces phases obtenue lors de l'examen du crayon TOF 4. 3. Produits de fission solubles dans l'U 0 2 Les produits de fission pouvant être en solution dans la matrice d'U02 sont les oxydes de terres rares (Nd, Ce, La, Pr etc. , cités par ordre décroissant de rendement de fission), l'oxyde de zirconium, l'oxyde de baryum et’ l'oxyde de molbydène. 208 BAZIN et al. ДДЕ h t- IMAGE ELECTRONIQUE URANIUM BARYUM CERIUM FIG. 6. Phase grise dans les grains équiaxes. La répartition radiale de l'oxyde de zirconium a été particulièrement étudiée dans le cas des éléments combustibles ayant subi un refroidisse ment lent. Le diagramme d'équilibre UC>2 -Z r 0 2 permet d'espérer, dans ces conditions de refroidissement, une ségrégation de la solution avec enrichissement en zirconium dans le sens du refroidissement. Cet enrichisse ment doit nous permettre de localiser avec précision la limite de zone fondue au moment du refroidissement. Des analyses ont été effectuées sur les crayons TOF 15 et TOF 11. La courbe de la figure 7 (TOF 15) montre effectivement une augmentation de concentration du Zr à partir d'un rayon de 1,8 mm pour TOF 15. Cette valeur correspond parfaitement aux dimensions de la zone fondue déterminée à partir des microstructures. Pour TOF 11, la correspondance est aussi bonne. IAEA -PL-463/9d 209 FIG. 7. TOF 15 Repartition radiale du zirconium dans 1*U02. FIG. 8. TOF 11 ~ Repartition radiale du baryum dans Г и 0 2. 210 BAZIN et al. D'autre part, on note également, sur la figure 7, des variations radiales de concentrations qui coïncident très bien avec la succession des anneaux vus sur l'autoradiographie ßy. En ce qui concerne les terres rares, une mesure de la concentration radiale du néodyme a été effectuée sur le crayon TOF 4. Nous n'avons pas constaté de variation dans la répartition de cet élément en dehors de la zone de réaction au contact oxyde-gaine. L'étude du baryum a permis de constater deux comportements différents de ce produit de fission: a) Dans le cas du crayon TOF 4 à refroidissement rapide la concentra tion du baryum est très faible et pratiquement constante de long du rayon. Une seule augmentation notable est constatée à la limite externe de la zone très poreuse correspondante au volume d'oxyde en fusion en fin de vie. b) Dans le cas du crayon TOF l i a refroidissement lent on constate une série de variations de concentration dans la partie solide de l'oxyde et une importante augmentation dans la zone en fusion (fig. 8 ). Il faut noter que les mesures de concentration du baryum ont été effectuées en dehors des plages de phases grises riches en cet élément. Dans les différents combustibles fondus analysés, nous n'avons jamais détecté de molybdène en solution dans la matrice d'U02. III. CONCLUSIONS Les points principaux que l'on peut retirer de ces examens sont: a) En ce qui concerne les précipités métalliques, la présence de nodules de molybdène pratiquement pur dans la zone fondue, alors que dans la cheminée centrale les lingots de produits de fission contiennent principale ment Ru, Mo, Tc, le rhodium étant à l'état de traces. b) En ce qui concerne les précipités d'oxyde, ils ont été trouvés dans les grains équiaxes, mais également dans la zone fondue lorsque celle-ci est refroidie lentement; leur composition dans cette zone est: uranium ~ 40% baryum ' 30% zirconium ~ 15% oxygène ~ 13% c) La variation de la teneur en Zr dissous dans la matrice a permis, pour les combustibles à refroidissement lent, de retrouver la limite de la zone fondue. d) Dans les combustibles examinés, nous n'avons pas trouvé de molybdène en solution dans la m atrice. IAEA-PL-463/10 THE MIGRATION OF FISSION PRODUCTS THROUGH REACTOR FUEL MATERIALS J. R. FINDLAY United Kingdom Atomic Energy Authority, Research Group, Applied Chemistry Division, AERE, Harwell, Didcot, Berks., England Abstract THE MIGRATION OF FISSION PRODUCTS THROUGH REACTOR FUEL MATERIALS. The mechanisms for the migration of fission products in nuclear fuel materials, principally oxides, are discussed. Fission products are released from surface regions by fission recoil; further release of fission products and fuel material is induced by a 'knock-out' process in which atoms are released by interaction with fission product tracks that traverse the surface. This process imparts mobility also to fission product atoms in the bulk of the fuel matrix. Insoluble fission products segregate into second phases such as gas bubbles or inclusion phases. Fission product gas bubbles are found intra-granularly and at grain boundaries. Release of gas occurs principally from the grain boundaries by mechanisms involving interlinkage of gas bubbles or by boundary sweeping. The release characteristics of short-lived fission products are determined by their short life-tim e in the fuel material, which allows description by simple diffusion models. The behaviour of iodine tellurium, and probably the alkali metals, is closely analogous to that of the noble gases. Most fission products are expected to be freely mobile at elevated temperatures, or, at low temperatures, by radiation- induced diffusion, although many are insufficiently volatile for transfer in the gas phase. 1. INTRODUCTION The behaviour and migration of fission products in reactor materials has been studied extensively because of the many aspects of fission product behaviour that are important to fuel performance. A wide spectrum of investigatory techniques has been used and there have been several discussions and meetings devoted to the top ic {j , 2\. Much information has been derived from fuel element irradiations either by post irradiation examination or by measurement during the irradiation. This has been supported by experiments specifically designed to study migration and other processes by precise control of conditions and by varying conditions to determine their effect on the processes involved. The observed behaviour has proved to be highly complex, although in some limited instances simple models describe the processes adequately. In this paper the current knowledge of some of the more important processes is reviewed. 2. FISSION BECOIL EFFECTS A fission fragment possesses considerable kinetic energy on its formation which is then dissipated within the material in its path. The distance travelled depends upon the stopping power o f the material traversed; generally this is the fuel material itself, where stopping distances of a few (im are typical. A proportion of those fission fragments born within a few microns of an external surface w ill be released to be slowed down within the external atmosphere, or, more likely, to re-embed in any adjacent solid 211 212 FINDLAY phase such as fuel cladding or containment. Fission products are released in direct proportion to their fission yield and the fraction released depends on the fuel surface to volume ratio which, in the case of particulate fuel, can be quite high. In addition to movement and release of the newly born fission fragment, mobility is induced in other atoms by the passage of the fission fragment track. When the track traverses the surface, atoms within its sphere of influence can be ejected by this 'knock-out' process. Release of gaseous fission products from oxide fuel under irradiation has been observed by this mechanism, ¿¡mission, was detected from specimens irradiated in configurations where release by r e c o il would be lo s t by absorption in the surrounding media [3 ,4 ] . By determining the dependence o f this emission upon neutron flux, Carroll has deduced that the emission occurs with a very high efficiency from a surface layer of only a few nm thickness H * . It is likely that this release is associated with the vapouriszation of uranium atoms from the surface o f both uranium metal and UO2 induced by the passage of a fission fragment. The effect was first observed by Rogers (_63 • ^n equilibrium is believed to be set up where the surface vapourization is balanced by re deposition of uranium atoms from neighbouring surfaces. However there are considerable discrepancies between different workers on the number of atoms volatilised per fission fragment track [7,8]. Although the amount of fission product release by recoil induced processes is seldom of great importance technologically, recoil processes have a considerable influence on fission product mobility particularly at low temperatures. At a typical fast reactor fuel rating of 200 w.g-^ statistically, every atom in an oxide fuel material comes within the influence of a fission fragment track approximately once every minuteThis frequency of interaction is sufficiently high for some re-distribution and segregation of fission products to be expected at temperatures that would otherwise be prohibitively low. This possibility is discussed later in rela tion to the formation o f fis s io n product inclusions and to the enhancement of diffusional processes by irradiation. 3. DIFFUSION PROCESSES The primary mechanisms of fission product migration, once the fission products have come to rest in the fuel material is expected to be determined by a diffusion process, either thermally activated or radiation induced. In the first instance models to describe fission product release were based on simple cla s s ica l d iffu sion theory where_ the diffu sin g species is presumed to be in complete solution in the matrix l_10,1l3. These have been shown to oversimplify a complex situation. The complexity arises principally from the highly insoluble nature of many of the fission products and the artificial state of supersaturation that exists when the fission product comes to rest in the lattice. Several differing patterns of behaviour are observed and these are now discussed in different categories. 3.1 F ission product gas behaviour The behaviour of the fis s io n product gas has been studied exten sively, because of their high fission yields and high specific volume at fuel element temperature. Their behaviour can be observed directly by optical and electron microscopy; releases can be measured readily by gas handling techniques. The formation of bubbles of fission product gas in fuel materials is well known and has been observed intra-granularly, at grain boundaries IAEA-PL-463/10 213 and on other defect structures. There has been considerable discussion in the literature about the modes of precipitation of the intragranular gas bubbles in order to explain the numbers of gas bubbles observed in irradiated fuel and the influence of temperature on both size and concentration [12,13, 14^» These discussions and other experimental evidence have led to the postulation of a mechanism for the re-solution of gas bubbles during irradiation such that a dynamic equilibrium exists in which some of the gas exists in bubbles and the balance is dissolved in the fuel matrix. The practical impact of this situation is that the fuel matrix contains a fract ion of the total fission product gas inventory effectively in solution, thus providing a concentration of diffusing species for the operation of solute diffusion. Bubble migration, which is probably a comparatively slow process is not, therefore, a fundamental requirement for gas mobility. Due to the re-solution process, the bubbles within the grains are in a dynamic condition which controls their size and number. The bubbles are observed to remain small and the gas they contain is stored at very high effective pressures and occupies little more volume than as single atoms homogeneously dispersed in the matrix £l 5~]. The effect of re-solution and other processes on the magnitude of gas migration rates has been considered by Speight Û O an<^ Whapham 1.12]. Speight has shown that the solution of the diffusion equation including a description of the re-solution-precipitation process simplifies to the simple diffusion model of Booth in which the diffusion coefficient D is replaced by a modified value to account for the reduced concentration of atoms in solution. Diffusion coefficients are, therefore, lower than expected for diffusion of the free atoms in true solution; as the re-precipitation mechanism is temperature dependent, the apparent activation energy is modified also due to concentration of re-dissolved atoms being a function of temperature. Additional migration mechanisms are considered by Whapham and again a decreased diffusion coefficient and a changed activation energy is pre dicted [1 2 ]. At the present time, a thorough substantiation of these predictions experimentally has not been achieved. Some measurements of the diffusion coefficients for the release of long-lived fission product gas from stoichio metric single crystal UO2 are in progress at AEEE. Preliminary work has indicated that the release of occurs with simple diffusion kinetics with a diffusion coefficient of 2.6 x 101if cm2 sec” "1 at 1 5 8 0 °C 073« Specimens were irradiated at 7 w.g“ 1 heat rating and 1025 fissions m“3. This diffusion coefficient is in good agreement with values obtained by post irradiation measurements on the emission of and suggests that the concentration of mobile species in the lattice was similar in both experi ments. The burn-up of the reactor experiment was lew and may have been insufficient for a representative bubble structure to develop. The technologically more important phenomena are concerned with the behaviour at grain boundaries, since the principle mechanism for the release of gas from fuel materials is via the gas bubbles accumulated at grain boundaries TIS»]. When a sufficient concentration is established, the bubbles interlink forming a connecting path to the exterior. The gas collects at grain boundaries by diffusion from the grain or by the sweeping of the grain by growth or by boundary movement in a temperature gradient. Parameters determining gas release are therefore the rate of accumulation at the boundaries, the capacity of the boundary, or its rate of movement. The rate of gas accumulation at the grain boundaries has been considered by Speight taking account of the fact that a boundary containing gas does not represent a perfect sink, because of re-solution back into the matrix from the grain 1 FINDLAY 214 x3500 1.2 X 1025 fissions m FIG. 2. Scanning election micrographs o f irradiated UO2 grain faces. IAEA-PL-463/10 215 boundary bubbles £16] This process can be observed experimentally. Scanning electron microscopy observations of fracture surfaces of UO2 irradiated to 1600°C and doses below 2 x 1025 fissions m“3 are shown in Fig. 1. The increase of size and concentration of the gas bubbles at the boundaries can be used quantitatively to determine the kinetics of the accumulation process. Preliminary observations show that gas accumulates considerably more slowly than expected from the gas atom diffusion co efficient indicating the influence of re-solution from the gas boundary bubbles The concentration of gas atoms required to saturate the boundary and effect interlinkage can be calculated with suitable assumptions of surface energy and restraint conditions. When bubble size is small and controlled by surface energy, 1.4 x 101 5 gas atoms m-2 can be accommodated. Under restrained conditions more gas can be held in larger bubbles but significant fuel swelling is produced. In a high burn-up irradiation it is significant that the capacity of the boundaries in a fuel of, say, 10 цт grain size, is not large representing only 2^5 of the total gas inventory at 1CÇS burn-up. The majority of gas in a high burn-up irradiation must be stored either atomically, or in small bubbles in the matrix, or it must be released. The rate of movement of a boundary is important primarily at high temperatures where movement of a pore or boundary up a temperature gradient causes the formation of large columnar grains. ’Wi e n a pore reaches a free surface such as the central void of a fuel element, gas is released without the necessity of saturation of grain boundaries. This effect has been demonstrated by Keller to be the principle mechanism of gas release in UO2 fuel at high temperature £20]. 3.2 Behaviour of short-lived fission products The behaviour of short-lived fission products is important to many aspects of reactor design. It has become customary to express the emission of short-lived species using the ratio of release rate to birth or genera tion rate B/B which may then be translated into a diffusion coefficient using the version of the Booth model for radioactive species^lcTj. The mechanisms of migration described for long lived and stable species are expected to apply equally to short-lived species but the relative importance of the various behavioural factors will differ. The behaviour at grain boundaries is likely to be less important as short-lived species will decay in the boundary bubbles before being released, although some exceptions are possible for the longer lived species whose half lives may be comparable with the time required to saturate and vent the boundary. This results in the emission being principally controlled by the emission from the external surface which, for the application of a diffusion equation, can be considered as a perfect sink. It is expected therefore that the Booth diffusion model can be applied with the qualification that the concentration of mobile species will be reduced by the proportion held in intragranular bubbles. A further conclusion is that the emission seen will represent principally the behaviour of species born close to the surface. The simple diffusion model has been used successfully by several workers to interpret measurements of gas release. Experimental results from different workers for the emission of from uo2 during irradiation are shewn in Fig. 2 ^21-24З. There is fair agreement and the results can be described by a single activation energy of 3.0 eV between 1000°C and 2000°C [)iQ* These values apply to low burn-up low rating experiments and agree well with laboratory gas.release measure ments using post irradiation heating methods [isQ. Irradiation under these conditions appears to have little influence on either the absolute rates of migration or upon the activation energy. At higher fuel ratings, as 216 FINDLAY ! 1 -U . o ' * и и О x Jackson iow density sinter 0.9 w.g“* _ e e О 0 0 -o О 0 ® Jackson high density sinter 0.9 w.g" 1 -O О О О 0 О Jackson high density sinter 5.8 w.g" 1 О £ \§ N * -OI 1600 ° с * Melehan high density sinter 0.3 w.g" 1 i > \ 0 Souhlier sinter ft single crystalO.S-l^w.g “ 1 л ч ©Findlay h ¡gh density sinter 15.7 w.g" 1 : к < о F in d la y h igh density sinter 5.8 w.g" 1 I и \ о V <> \ О О \ 0 О \ 1N 01 0 > « - \ \ <, \ N <\ ° \ < Ul ► \ \ , N / ? 3 kcal mole -1 r I \ 3 .0 e .v) \ y I < . 1 1 f\ \ > . V ы i Ч ----- 1l v \ \ - : V 1 i i 4 5 6 7 в Reciprocal temperature x Ю 4 °K FIG.2.- The emission o f85mKr from UO2 0 during irradiation. demonstrated by the results of Carroll shown in Figure 3 [25j i the emission decreases due, it is presumed, to the formation of trapping sites which reduce the effective concentration of mobile species. In the range of fuel ratings appropriate to fast reactors, the emissions increase possibly due to the restoration of the concentration in solution Ъу extensive re-solution from gas bubbles. In this region, considerably lower activation energies were found upto 1б00°С from (U,Pu)C>2 with higher values' upto 2000°C. These results are shown in Fig,4 [4>22] . This situation is explained possibly by the domination of thermal effects above 1бОО°С and the existence of radiation enhanced diffusion at lower temperatures. A range of fuel burn-ups was studied and the emission of IAEA-PL-463/10 217 FIG. 3. The effect of fuel rating o f the emission of 85mKr from U 02. short-lived activities was found to be independent of burn-up once the physical changes in the releasing area of the fuel had been taken into accountfV]. It appears that high burn-up fuel is rendered porous by the release of stable gas and remains in that condition provided the rate of gas generation remains sufficiently high to prevent pore closure. The behaviour of short-lived species can be affected by precursor effects a n d it has been shown that much of the emission of species such as 133xe of 5 day half life occurs as the 21 hour ргесигзог\_22, . In some recent experiments, comparative measurements of halogen, tellurium and gas emissions have been made and, to a first approximation, the emission characteristics of these individual species are the same and the effect of radioactive half-life on emission behaviour is in agreement with that predicted by the diffusion model C"*íQ. The precise relationship between the behaviour of parent and daughter has yet to be conclusively resolved, although it appears that the behaviour of noble gas and halogen species is closely analogous. 3.3 Behaviour of other fission products The behaviour of fission products other than the gases and short-lived species has received less study but some conclusions can be drawn. Caesium and to a less degree rubidium form an important category of fission products, because of their high yield and mobility. Extensive mobility of caesium is observed by Y scans of irradiated fuel elements and in other irradiation experiments\27J. Post irradiation laboratory experiments have indicated that caesium is released from oxide fuels at least to the same extent as the fission product gases t o •, In some instances caesium releases exceed gas release notably when the fuel is high density. This may indicate that 218 FINDLAY Reciprocal temperature x 7— г С°к) FIG.4. The emission of 85rtlKr from (U, Pu)O2 0 during irradiation. caesium is able to migrate along grain boundaries at a significant rate, and be released whilst the gas is released subsequently by interlirikage of gas bubbles. Many other species are appreciably mobile at fuel element temperatures but they are not observed as released species as their vapour pressures are too low for appreciable transfer to the gas phase. Some special cases exist where non-volatile fission products have gaseous or volatile short lived precursors. These include °®Sr, ®^Sr, ^9sr> 90gr ^ 91 уs 137cs, ^ 3 c s , 135cs, 138ßa, 139ва, 140ва. However, once the fission product precursor has decayed, the minimum mobility expected from a non-volatile species would be that describing the self diffusion of the heavy metal cation in the matrix phase. Above 1400° C this is sufficiently high for significant movement to occur within.the life time of a fuel material. Evidence of mobility at lcwer temperatures has been obtained in oxide fuel at high fuel ratings (lOCQw.g- ) and bulk fuel temperatures as low as 400° C. Extensive fission IAEA-PL-463/10 219 product inclusion phases were found that can only be explained by the assumption of a relatively rapid radiation induced diffusion process ^28^] . It is thus evident that there is considerable mobility within the matrix and that fission products segregate into separate phases representing their thermodynamically preferred state. The. accommodation of rare earth elements, Sr and Zr within the fluorite lattice has been demonstrated indirectly by X-ray crystallographic measurements on irradiated fuel Í . 2 8 ] , Metal and oxide inclusionsphases have been observed extensively and the development and composition of these phases will be discussed in other sections -of the panel discussion. 4. CONCLUSIONS The behaviour of fission products in oxide fuel materials has proved to be complex and affects the chemical and physical properties of the fuel element system. The fission process initially deposits the fission products in a unique situation of supersaturation and subsequent redistributions occur by a series of complex mechanisms. These include the formation of gas bubbles, inclusion phases and solid solution with the matrix. In many instances radiation has been shown to influence these redistributions either by enhancing diffusion processes at lew temperatures or controlling the processes of gas bubble nucléation and growth by re-solution mechanisms. Some of the processes can be described quantitatively by model calculations but in many instances the important parameters, whilst identified, are not known with sufficient accuracy for predictive purposes. REFERENCES L11 See Nucl Applns 2 (1966) 117 et seq. | _ 2 ] Int. Congress on diffusion of fission products. Nov. 19б9. Saclay. French. Soc. Radioprotection. [ f ] CARROLL, R.M., SISMAN, 0., J. Nucl. Mat. 17. (1965) 305. [4] FINDLAY, J.R., WATERMAN, M.J., BROOKS, R.H., TAYLOR, R.G., J. Nucl. Mat. 3¿ (1970) 24 Г5~1 CARROLL, R.M. MORGAN, J.G., PEREZ, R.B. USAEC Rep ORNL 4200 р.бО (1968) 16] ROGERS, M.D., J. Nucl. Mat. 1¿ (1 965), 65. NILSSON, G., J. Nucl. Mat. 20 (1966) 215. [8] FERRARI, S. SEGRE, G.J., J. Nucl. Mat. 24 (1967) 150 V/HAPHAM, A.D., MAKIN, M.J. Phil. Mag. 7, 81, (1962), 1441 fcÓ] BOOTH, A.H. AECL Rep DCI 27, AECL 496 (1957) j j l ] CUBICCIOTTI, D. U.S.A. Rep. NAA-SR-194 (1952) V/HAPHAM, A.D. UKAEA Rep AERE-R 6770 (1 971 ) [13] TURNBULL, J.A. J. Nucl. Mat. ¿8 (1971) 203 [14J MANLEY, A.J. UKAEA Rep. TRG l68l(Tff) (1 ?68) 220 FINDLAY [ 15] FINDLAY, J.R. Chemical Nuclear Data B.N.E.S. (1971 ) [l6j SPEIGHT, M. Nucl. S ei. & Engng, 3 7 (1969) , 180 [ i f ] JOHNSON, F .A ., M0EET0N-SMITH, M.J. Unpublished work at AERE (1972) [lé ] DAVIES, D., LONG, G., UKAEA Rep ASRE-R 4-347 (1963) [ 19] GREENOUGH, &.B., NAIRN, J .S ., SAYERS, J.B. Int. Cong, peaceful uses Atom. Energy (Proc. conf. Geneva 1971) A/Conf 49 P .501 U.M. [20l HILBERT, R .F., STROHOK, V.W., CHUBB, !7., KELLER, D.L. J. Nucl. Mat. 38 (1971) 26 [ 21] JACKSON, G., DAVIES, D., BIDDLE, P ., WATERMAN, M.J. UKASA Rep AERE -R 4714 (1964) A3RE-M 1607 (1965) ]_22] FINDLAY, J.R. WATERMAN, M .J., Unpublished work at AERE. Г23~l M E L E H A N , J .B ., BARNES, R .H ., GATES, J .S ., R O U G H , F.A. J U.S. A.B.C. Rep. BMI - 1 623 (1963) ^24] SCHURENKAMPER, A ., SOUHLIER, R. CEA, France Rep. CEA 2588 (1964) CARSOLL, R.M., MORGAN, J.& ., PEREZ, R.B., SISMAN, 0 ., Nucl. S ei. & Engng. 38 ( 1969) 143 \26^ CHENEBAULT, P ., DALMAS, R. Ceramic Nucl. Fuel Symp. Knoyer and Kaznoff, Am. Ceram. Soc. Nucl. Section (1969) \2Í\ ROBERTS, L.S.J. et al. Int. Conf. Peaceful uses Atom. En. (Proc. Conf. Geneva 1965) 11, U.N. New York (1965) 464 \28Д DAVIES, J.H ., EWART, F .T ., J. Nucl. Mat. (1971 ) 143 IAEA-P L -463/11 CONTRIBUTION TO THE STUDIES OF THE FISSION GASES RELEASED FROM IRRADIATED URANIUM OXIDE PELLETS J. KLÍMA, M. PODEST, V. VINS Nuclear Research Institute, fteá near Prague, Czechoslovakia Abstract CONTRIBUTION TO THE STUDIES OF THE FISSION GASES RELEASED FROM IRRADIATED URANIUM OXIDE PELLETS. The postirradiation diffusion coefficient of fission-gases released during annealing in the temperature region 1200 - 1600°C was measured on samples of thermally processed dense U02. The characteristic value of the material, which was sintered to practically the same density (99Уо T .D .), was the grain size, which ranged from 10 to 100 fim. A clear dependence of 133Xe release on the regime of thermal processing of the original pellets before irradiation was observed. In the mentioned region of the grain size of the original material the value of the diffusion coefficient ranged from 5 X 10"u to 1 x 10_7cm2/s. For samples in the grain size range of 20 -40 pm the same activation energy was estimated (Ед - S0 000kcal/mol), An interrelation exists between the D0 and the average grain size, which has not yet been defined more exactly. 1. INTRODUCTION Studies on the migration of gaseous fission products in irradiated UO2 have recently been started at the Nuclear Research Institute of the Czechoslovak Commission for Atomic Energy. A part of these endeavours has been dedicated to a study of the fission gases released during post irradiation annealing in dependence on the changes of the structure in UO2 pellets that had been irradiated to low burn-up (about 1015 fissions/cm 3 ) at temperatures not exceeding 100°C. The present report discusses the results obtained in the region of annealing temperatures from 1 2 0 0 to 1600°C. In in-pile experiments the contribution of individual structural regions of UO2 - produced as a consequence of irradiation - cannot be evaluated from the total balance of released fission gas. For this reason UO2 that had been prepared with varying average grain size (1 0 - 1 0 0 /um) was used in our experiments. In such a way the process that takes place during irradiation and during which the original grain size changes can be simulated. It is not possible to affect the contribution to the total mass transport produced by the potential resulting from the thermal gradient. The results of the experiments presented in this report will be used for model calculations of fuel element behaviour. It is expected that on this basis it will be possible to affect the contribution of other processes during irradiation to the release of gases. 2. EXPERIMENTAL PART The kinetics of fission gas release was measured on UO2 pellets of varying grain size. The material of all the pellets was of the same origin 221 222 KLIMA et al. (powdered UO2 , specific surface 6.5 m2/g and U/О ratio 2.0). The pellets were prepared by sintering in hydrogen and maintaining at 1600°C for 4 hours. They were then processed in the same atmosphere for 0.5 - 8 hours in the temperature region of 1700 - 2200°C. Samples that had practically the same density were chosen for the experiments (~ 98% T.D.). The original metallographic structure of the unirradiated pellets was compared with the structure of the pellets after irradiation and after postirradiation annealing. The proper annealing of the irradiated pellets was performed in a furnace, which worked up to 2000°C in a dynamic inert atmosphere (the carrier gas being purified He). The gaseous radionuclide ( 133Xe) was detected in the carrier gas by means of a well-type Nal(Tl) crystal. The burn-up of the irradiated pellets was estimated radiochemically from the activity of Zr-Nb, Ce and Cs. The amount of fission gas generated was calculated from the tabulated value of the 133Xe fission yield (6.62%). Experiments were performed only in the lower region of annealing temperatures, i.e. 1200 - 1600°C. It was confirmed that in this temperature region no significant grain growth took place that would be outside the erro r limits of the evaluation method [ 1 ]. 3. RESULTS AND DISCUSSION The diffusion coefficient was calculated by means of the expression of Innthoff and Zimen [2]: where F is the relative amount of gas released X is the decay constant of 133Xe t is the time of annealing (seconds) S is the surface of the pellet (cm2) V is the volume of the pellet (cm3) The dependence of the average values of the diffusion coefficient on the average grain diameter is given in Table I. It can be seen that the absolute value of the diffusion coefficient is rather sensitive to the grain size. In the region of 10 to 100 д т the value of the diffusion coefficient varies within the limits of approx. 5 X 10~n to 1 X 10' 7 cm2 /s . At the same time, from samples described in Table II it was estimated that the activation energy of this process did not depend on the grain size, while the value Do did show some change. No evaluation is, however, possible on the basis of the limited amount of data. From Table II it can be seen that Do depends strongly on the thermal annealing of the pellets before irradiation: e.g. for samples maintained at 1900°C for 0.5 hours Do is approximately 6 times higher than for samples maintained for the same period at 2200°C. On the other hand, the sample maintained for 8 hours at 1800°C did not reach the Do values for UO2 pellets annealed at 2200°C for a short period of time (0.5 hours). At a difference of only 100 degC the Dj700 = 1/9 of the Dq800 approximately. IAEA-PL-463/11 223 TABLE I. DEPENDENCE OF THE DIFFUSION COEFFICIENT ON THE DIAMETER OF GRAIN Grain diameter D Temperature Series Thermal processing (lira) (cm 2/s) (°C) 11 20 4 .9 7 9 x 10‘ n 1225 1700°C/8 h 12 20 1. 65 X 10"9 1445 1700°C/8 h 35 30 5.0 x 10'10 1225 1900°C/0. 5 h 36 30 1 .4 3 x 1 0 "8 1445 1900°C/0. 5 h 27 40 1.30 x 10"9 1225 l900°C/2 h ' 97 40 1.35 X 10"9 1225 1800°C/8 h 98 40 3.71 X 10'* 1445 1800“C/8h * 19 30 5. 08 x 10'10 1225 1800°C/0. 5 h 20 30 1.40 x 10'8 1445 1800°C/0. 5 h 80 45 1 .2 1 x 1 0 " 9 1225 2200°C/0. 5 h 81 45 3. 626 X 1 0 '8 1445 2200°C/0. 5 h 45 60 1.56 x 10"8 1225 1900-C/8 h 107 70 1 .0 x 1 0 "7 1225 2200"C/2 h 88 90 4. 54 X 10'7 1225 2200°C /8h TABLE II. DEPENDENCE OF THE DIFFUSION COEFFICIENT ON TEMPERATURE Average diameter Activation energy Do Series of pellets of grains Remarks (kcal/mol) (cm z/s) (pm) 11 .1 2 20 56 393 4.'93 x 10-1 Heat treatment at 1700°C/8 h; 98. 7°lo T .D . 35. 36 30 52180 ■ 9.97 x 10-1 Heat treatment at 1900°C/0. 5 h; 98.3% T .D . 9 7 .9 8 40 53 241 3 .7 2 Heat treatment at 1800°C/8 h; 97e#) T. D, 80.81 45 54716 6 .0 5 Heat treatment at 2200°C/0. 5 h; 9 7 .9 °jo T .D . 224 KLIMA et al. This statement is supported by Miekeley and Felix [3], who estimated that in the first phase of annealing the apparent coefficient D is affected by the particle size and annealing temperature. It is necessary to emphasize that the data presented in this report were obtained in very dilute solutions of gases and that they might accordingly be valid only in the initial stages of the irradiation of the fuel element. The authors suggest that the interrelation of Do and the average grain diameter can be explained by.the annealing of some more mobile defects during thermal processing. The removal of areas where the gas can be bound in a relatively stable way could be a direct consequence of this effect. This hypothesis needs, however, further experimental work, especially for the correlation with thermally unprocessed material. 4. CONCLUSION In the process of measuring the kinetics of fission gas released from irradiated pellets of thermally processed UO2 the dependence of the diffusion coefficient on the average diameter of the grain was found. For some of the samples it was indicated that the activation energy was mutually identical and did not depend on the grain size. An interrelation exists between the grain size and D0, a more detailed explanation of which needs further experimental work. REFERENCES [1 ] HILLIARD, M ., ’ Met. Progr. (1964) 99. [2] INNTHOFF, W., ZIMEN, K. E., Trans. Chalmers Univ. Tech. 116 (1956) 6. [3] MIEKELEY, W., FELIX, F.W ., J. Nucl. Mater. 42 (1972) 291. IAEA-PL-463/12 Pu, U REDISTRIBUTION IN (U,.Pu)02 FUELS BY TEMPERATURE GRADIENTS* M.G. ADAMSON, E.A. AITKEN General Electric Company, Vallecitos Nuclear Center, Pleasanton, Calif., United States of America Abstract Pu ,U REDISTRIBUTION IN (U,Pu)02 FUELS BY TEMPERATURE GRADIENTS. A predictive model of the time-dependent evolution of radial Pu/U + Pu gradients in operating non- molten mixed oxide fuels is being developed. This model is based on preferential evaporation-condensation and vapour transport of metal-bearing species either along cracks and porosity channels within the fuel or inside closed pores migrating up the temperature gradient; equilibrium thermodynamics are applied to calculate the continually changing vapour composition over the hot fuel. Predictions of the model are used in combination with recent in-pile experimental data (from non-molten mixed oxide fuel pins irradiated in EBR-II from 0. 2 to about 11 at.% burn-up at varying powers) to delineate the key parameters in radial actinide redistribution. A correlation between fuel stoichiometry (O-to-Pu ratio) and the degree of redistribution (described by an enrichment factor ypu) has been identified; however, the degree of restructuring of the fuel — as determined by the peak linear power rating, time at this power, and the-initial fuel density — also has a strong influence on the final value of ypu. Certain kinetic features of the process and intended modifications to the present model are discussed, and evidence for axial actinide vapour transport in an irradiated annular fuel is also presented. 1. I N T R O D U C T IO N Last year we critically reviewed the principal mechanisms (and variants) that have been proposed to account for radial actinide element redistribution in irradiated mixed oxide fuels and also outlined a possible analytical description of this phenomenon based on a preferential evaporation - condensation mechanism (i.e., vapor transport). [1] This paper brings up to date our theoretical studies in this area and also incorporates some recent GE irradiated fuel pin data in a comparison between theoretical predictions and measured end-of-life radial Pu/U+Pu profiles at a variety -of fuel power ratings, burnups, initial fuel densities and 0/M ratios. In this way some of the key parameters characterizing radial and axial actinide redistribution have been delineated. Our earlier conclusion that vapor transport would be the most important mechanism for actinide redistribution in fuels which do not undergo center melting [ 1 ] has since received convincing confirmation both from out-of-pile and in-pile experiments. Bober and co-workers at Karlsruhe [2] in some carefully designed out-of-pile axial temperature-gradient tests essentially confirmed Rand and Markin's original calculated "cross-over" 0/M value at 2400°C (i.e., the 0/M value at which the vapor changes from U-rich to Pu-rich as 0/M decreases; 1.96 to 1.97). These workers also showed that the over-all vapor transport rate in a temperature gradient of 2400° to 1900°C falls off as the over-all fuel 0/M drops from 2.00 to 1.96. Lackey et al., [3] found evidence in irradiated SPH EREPAC mixed oxide fuel pins for vapor deposited U-rich dendrites in cooler regions of the fuel (on the hot side of intact microspheres) and, more recently, Argonne workers [4] report evidence of radial vapor phase preferential evaporation - condensation effects along the interface between originally dish-ended mixed oxide fuel pellets. Work performed under USAEC Contracts AT(04-3)-189, PA 53 and AT(04-3)-189, PA 10. 225 226 ADAMSON and AITKEN 2. THEORETICAL (VAPOR TRANSPORT) MODEL Our approach to developing an authentic model for actinide element redistribution in noncenter-melted irradiated mixed oxide fuels has been to start with a simple yet realistic description of the basic phenomenon (i.e., U, Pu redistribution by preferential evaporation-condensation) and then to take account, step by step, of modifying factors such as solid state back-diffusion and fuel restructuring (plus concomittant effects). In the basic fuel pin "m odel" a transverse fuel section is divided into 30 concentric equal-width annuli (radial segments) and each is assigned starting values of О/Pu ratio, temperature, mass of fuel and open pore volume. For initial calculations the radial temperature gradient and О/Pu gradients (see Figure 1) are assumed to be time-independent; the heat of transport used to calculate the О/Pu gradients is - 10 kcal/mole. [1] Recent work at this laboratory indicates that use of this "over-all" heat of'transport represents an oversimplification; however, at this stage of development of the model, errors incurred by its use are trivial.) The initial equilibrium vapor pressures of the metal bearing species PuÛ 2 , PuO, Pu, UO 3 , U O 2 , UO, and U above each fuel position are calculated using "V A P R E S" a FORTRAN IV computer program which has been written as a subroutine. From these equilibrium partial pressures the radial partial pressure gradient of each gas specie is computed at time zero; at subsequent times new equilibrium partial pressures are calculated by application of Raoult’s Law (partial pressure of species i mole fraction of i in solid). The metal-bearing species are allowed to flow (by molecular diffusion) between the fuel segments for a time interval St, i.e., down the temperature and respective partial pressure gradients - the area available for vapor transport between consecutive fuel segments is assumed to be a fixed fraction of the open porosity assigned to those particular segments. The amount of each vapor specie lost or gained by each radial segment in the time interval 6 t is determined by solving the appropriate diffusion equation, - which is Fick's Second Law expressed in cylindrical coordinates: D¡ (a2c¡/sr2 + 1/r • ac¡/3r) - эс/at = 0 (I) where C¡ is the concentration of gaseous species i, r is the fuel radius, t is time, and D¡ is the simple kinetic theory diffusion coefficient for species i ( = 3/8o2n V RT Equation (1) is solved numerically using the appropriate difference equations for C(r,t + 5t)¡, etc. The total mass of each segment and the absolute amounts of U and Pu contained in the solid part of each segment are adjusted after each gas phase transfer interval (§t) in preparation for the next cycle of calculations; these values are either stored in the computer or printed out as they are calculated. One assumption implicit in this representation of vapor phase mass transport is that the rate of vapor transport of a particular specie at a particular radial location is slower than its rates of evaporation or condensation, in other words at a particular location fuel and vapor are always assumed to be in thermodynamic equilibrium. A flow diagram of the calculation sequence is shown in Figure 2. "Initial input" contains such information as fuel dimensions, initial O/M and.Pu/U+Pu ratios, fuel density, linear power level, fuel surface temperature, etc. In its present form the computer program generates radial Pu/U+Pu profiles with the same general shape as those measured in irradiated fuels. Some of these calculated profiles are illustrated in Figure 3. Except for one unique value of O/M the calculated degree of plutonium enrichment (or depletion - depending on over-all О/M) at the fuel center increases continually with time until, just before the mass of the innermost fuel segments reaches zero, the fuel composition approaches pure plutonia (or urania). Clearly this is unrealistic since available data from fuel pin irradiations (see below) show that (i) radial Pu/U+Pu gradients are established early in life (concomitant with restructuring), (ii) the (maximum) degree of Pu enrichment at the fuel center rarely exceeds ~ 5 0 % and (iii) once established the radial Pu/U+Pu gradients are relatively insensitive to further irradiation. This is taken as evidence that another process or combination of processes is operating which causes the system to attain, or at least approach, a steady-state condition. Such processes are believed to be either back diffusion of uranium and plutonium ions through the solid phase (i.e., down the concentration gradients created by vapor transport) or a combination of ionic "back-diffusion" and decreased vapor transfer rates as a result of fuel restructuring (i.e., both porosity and temperature decrease with time at a particular radial position). The next step in the development of our calculational model will be to include solid state (back) diffusion of U and Pu to see whether this process alone can bring about a steady-state condition; an attempt will then be made to simulate fuel restructuring by allowing the open porosity in each radial segment to decrease as the segment gains mass through vapor transport. Inclusion of this last process should make preferential evaporation - condensation in migrating closed pores and vapor transport through open porosity equivalent. In this context it may be noted Lackey,et al., have related actinide redistribution to restructuring kinetics by tying vapor phase fractionation of U and Pu to radial pore migration rates; [5] in this way a significant slowing down of the rate of actinide redistribution with time was obtained. IAEA-PL-463/12 227 FIG. 1. Calculated radial oxygen (O/M) gradients for several fuel compositions (OS = -10 kcal/mol). 3. PREDICTIONS OF THE VAPOR TRANSPORT MODEL AND COMPARISONS WITH FUEL PIN DATA 3.1 EFFECT OF STOICHIOMETRY A prediction of the present vapor transport model is that the degree of plutonium enrichment (or depletion) at the center of fuels operated at similar power levels (and hence with similar radial temperature gradients) will be a function of fuel stoichiometry. [1] According to equilibrium thermodynamic partial pressure calculations there will be no tendency for evaporative Pu enrichment or depletion at the hot center of fuel with (0/Pu)center =3.45 and T q = 2500°C; in a radial temperature gradient of 2500 - 920°C this corresponds to an over-all fuel composition Lfo.80*au0.20*^ 1.973 (assuming QJ = -10 kcal/mole). The calculated Pu/U+Pu profiles 228 ADAMSON and AITKEN FIG. 2. Flow diagrams for PUGRAD, shown in Figure 3 illustrate that Pu enrichment is expected when the over-all 0/M >1.973 and that Pu depletion should occur when 0/M <1.973. In Figure 4 an attempt has been made to correlate the degree of Pu enrichment (or depletion) at the center of a number of high-rated fuels (T q = 1500°C) with over-all plutonium valency, using the calculated quasi-congruently evaporating composition as an anchor point. The degree of actinide redistribution has been expressed as a "Pu enrichment factor," 7 pu, which is simply the ratio of the Pu mole fraction measured at the inner surface (or center) of the fuel to that measured at the outer surface of the fuel at end-of-life. The U and Pu mole fractions were determined, via chemical concentrations, from measurements of the intensity of- the uranium Ma and plutonium Mß characteristic x-rays in a MAC-450 shielded electron IAEA-PL-46 3/12 229 r/ro FIG. 3. Calculated Pu profiles for MO 1%fi , MOb 988 and MO2 00B (J indicates number of iterations). microprobe analyzer (~ 1 micrometer diameter electron beam). Data'were collected across the radii of transverse fuel cross sections (from at least twenty different fuel matrix locations) and reduced by M A G IC IV , a computer program for microprobe data reduction; [6 ] the values of the Pu mole fraction at the inner and outer surfaces of the fuel were determined by linear extrapolation of the data from contiguous fuel positions. The error bars draw n in Figure 4 represent the total uncertainty in the calculated Pu enrichment factor 7 pu. All the fuels plotted on this diagram underwent appreciable restructuring during irradiation in EBR-2; their vital statistics [7] are summarized in Table 1. In determining the final over-all Pu valency of the fuels the O/M change with burnup was accounted for by using Д(0/М) % 0.002 per atom percent burnup, which refers to fully U-235 enriched mixed oxide fuel in a fast neutron flux at O/M ratios less than 2.00. It appears there is a reasonably good linear 230 ADAMSON and AITKEN / F 3 B / ( + A N L D< / / / / / / lp“'p“ + UW / • -■w / / Л / т / / Calcu lated (VAPRES) / V / v Pu or ul°ver-al11 / / / / “ 25 0 0 С in ali cases / / / FIG. 4. Correlation between degree of plutonium enrichment (ypu) and stoichiometry (О/Pu) in irradiated mixed oxide fuels having undergone appreciable restructuring. correlation between 7 pu and Vpu for fuels with O/M (20% Pu) >1.98 however, thus far, data for low O/M fuels are lacking. A number of fuel pins recently examined at A N L [4] also show Pu enrichments which can be fitted to the correlation shown in Figure 4;.these pins (Numec С -series) either finished or started near the stoichiometric composition (Vpu = 4.00)-and were sufficiently high-rated that appreciable restructuring and central void formation occurred. , The evaporation-condensation mechanism for radial actinide redistribution predicts that low O/M fuels «1.97.) should show Pu depletion at the center provided that center melting does not occur. It has been suggested [1,3] that this may never take place because, under such conditions, the proposed solid-state Soret effect, which would be expected to move Pu preferentially up the thermal gradient in significantly IAEA-PL-46 3/12 231 hypostoichiometric fuel, could become important. Further, the much reduced vapor pressures of metal-bearing species over fuel at low 0/M ratios may reduce the rate of U-Pu partitioning via the evaporation-condensation process to such an extent that solid state back-diffusion of U and Pu ions may prevent the development of significant Pu/U+Pu gradients. The combined effect of these various factors may be to cause a lower degree of Pu depletion than predicted by extrapolation of the linear 7 p u-0 /M correlation shown in Figure 4 to low 0/M values. Uncertainties in some of the high temperature thermodynamic data used to make the vapor pressure calculations, and in the high temperature/low 0/M value of Q£, may also influence the actual value of the "cross-over" O/M corresponding to a particular linear power rating. We are hopeful that some of these uncertainties will be resolved as we continue to refine our calculational model, and evaluate new irradiated fuel pin data. In concluding this action it is worth noting that the first report of Pu depletion at the center of an irradiated mixed oxide fuel was made recently by F. Schmitz and coworkers from CEA Euratom; [8 ] this effect was observed in a fuel with starting O/M = 1.90 which underwent appreciable restructuring during irradiation. Further results of this type are awaited with interest. 3.2 EFFECT OF BURMUP ON RADIAL ACTINIDE REDISTRIBUTION Accumulation of data from post-irradiation examinations of a number of adequately characterized GE mixed oxide fuel pins has reached the point where we can begin to search for correlations between 7 p u and various suspected primary parameters. The effect of fuel stoichiometry has already been described, however, it is evident that interacting with this are the degree of restructuring of the fuel and its maximum temperature (which to a large extent are determined by the initial fuel density) and burnup or irradiation time. Table 2 summarizes data from four fuel sections which were selected to illustrate the effect of burnup at approximately constant stoichiometry. Further data on these fuel pins are given in Table 1. Samples F2G-G and F5Q-H1, which started at the same O/M, attained markedly different burnups at similar (high) power levels. Appreciable restructuring occurred in each case. The difference in 7 pu for these two pins is small and the direction of this change may simply reflect the difference in the final O/M ratios, i.e., ypu is higher for the higher O/M fuel. Although they showed appreciable restructuring the two intermediate burnup samples operated with significantly lower calculated center temperatures which may account for the somewhat low 7 pu values obtained. At any event it appears from these preliminary observations that the effect of burnup on actinide redistribution is small and most probably derives from the O/M change with burnup. 3.3 EFFECT OF FUEL DENSITY Table 3 summarizes the results of two short term irradiations performed with low and high density U0.80^u0.20Ol.98 fuel Pellets- It is judged significant that the high density fuel, which had the highest linear power rating, suffered less Pu enrichment than the relatively low density fuel. The photomicrographs of these fuel sections showed considerably less restructuring in the high density fuel than in the low density fuel; thus clearly the degree of actinide redistribution which can occur in a given fuel is predicated by the extent to which that fuel can restructure. Since fuel restructuring occurs primarily via evaporation-condensation across pores and bubbles migrating up the temperature gradient the connection between actinide redistribution and restructuring becom es obvious. In Table 4 a comparison is made between three fuel samples, all with high densities but each having a different starting O/M. The irradiations were short-term (low burn-up) and each fuel operated at the same (high) power rating. From the results an O/M effect whose direction agrees with the prediction of the vapor transport model, is apparent - however the degree of Pu enrichment is substantially less than obtained from analogous low density fuels (see Figure 4). These results provide additional support for the correlation between degree of actinide redistribution and available volume for restructuring (and hence vapor transport) mentioned above. 3.4 AN EXAMPLE OF AXIAL ACTINIDE MIGRATION To date our theoretical studies of actinide redistribution have been focussed on radial effects, as it was considered important to understand this basic fuel pin phenomenon before any attempt was made to model three dimensional actinide redistribution. However, during examinations of the radial Pu/Pu+U profiles in transverse sections from different axial locations of fuel pin F3B-3 (which began irradiation with a 40-mil diameter central void) [9] it became apparent that some axial transport of fuel had occurred and that this had produced a marked effect on the "radial" results. Here an attempt is made to explain the observed results in terms of the vapor transport "model." 232 TABLE 1. Irradiated Fuel Pin Data In itia l^ Section Heiqht Peak Linear Power T q (calod ,] Composition (inches from Density (% T.D.) Annulus diam (k w /ftl °C Burnup (at. %) Sample N o. O /M %Pu b ottom ) Smeared Pellet {in mils) SOL EOL SOL EOL Anal. Calc. F3B-3M 2.00 25 12.9 90.8 41 16.36 15.85 2163* 4.9 F3B-3G 2.00 25 6.8 90.8 41 16.36 15.85 256 3* 5 .3 ~ 5 F2G-G 1.975 20 6.2 92.9 95.5 - 14.86 13.81 2570 2319 10.8 12.7 F5Q-H1 1.974 25 ■ 9.8 90.85 94.4 - 13.92 13.38 2519 4.9 5.7 E 1H-30F 1.98 25 7.5 89.9 93.0 - 13.1 11.6 2175 1975 7.7 '■'■9 E 1H -23K 1.974 25 11.0 - 92.4 41 ~ 1 3 1800 1600 7.0 ~ 9 F1E 1.98 20 6.5 86.6 89.2 _ 14.8 ~ 2 5 0 0 0.17 0.20 DMO ad AITKEN andADAMSON F1D 1.98 20 6.5 92.8 95.3 - 16.2 0.17 0.20 F1B 2.00 20 6.5 93.9 96.2 - 16.4 0.17 0.20 F1C 2.03 20 6.5 93.8 95.6 - 16.8 0 .1 6 0.21 tA ll fuel made from coprecipitated mixed oxide *V\iith high power driver TABLE 2. Effect of Burnup Average Power Burnup Sample O/M (kW/ft) (at. %) 7Pu F2G-G 1.975 (1.992) 14.2 10.8 1.23 E1H-30F 1.98 (1.992) 12.4 7.7 ~1.1 E1H -23K 1.974 (1.985) 13.0 7.0 ~1.1 F5Q-H1 1.974 (1.982) 13.6 4.9 1.18 IAEA-PL-463/12 233 TABLE 3. Effect of Density Pellet Density Average Power Pin No. O/M (% th.) ...... kW/ft TPu F1E 1.98 89.2 14.8 -1 .1 FID 1.98 95.3 16.2 1.0 (Burnup = 0.17 at. %) TABLE 4. Effect of О/M at High Density Pellet Density Average Power Pin No. O/M (% th.) kW/ft УРи F1D 1.98 95.3 16.2 1.00 FIB 2.00 96.2 16.4 1.07 F1C 2.03 95.6 16.8 1.18 (Burnup = 0.17 at. %) “‘-i/-* ™ 1 ШЁЁЩШШШ Distance from Cladding Wall (mil) FIG. 5. Photomicrograph of specimen F3B-3M and corresponding radial distributions of Pu/U + Pu and oxygen. 234 ADAMSON and AITKEN jß ~ r - . i." Distance from Cladding Wall (mil) FIG. 6. Photomicrograph of specimen F3B-3G and corresponding radial distributions of Pu/U + Pu and oxygen. Figure 5 shows a photomicrograph of specimen F3B-3M and corresponding radial distributions of Pu/U+Pu and oxygen; Figure 6 shows specimen F3B-3G. Sections 3M and 3G were taken from positions 12.9 inches and 6 . 8 inches from the bottom of the fuel column, respectively; the 6 . 8 inch position corresponds to the reactor core mid-plane and it ran 300-400°C hotter at the center line that the upper section. The simultaneous observations of a decrease in the central void diameter of the upper (cooler) fuel section and a particularly high value of 7 p u together with an increase in the central void diameter of the central section, initially suggested that vapor transport of Pu-rich material up the central void had occurred (i.e., evaporation from the central (hot) section and condensation at the upper (cool) section); however, the starting composition of this fuel (O/Ш = 2 .0 0 ) was such that preferential evaporation of Pu could not occur (unless gross reduction of the fuel occurred during irradiation) and- mass balance calculations performed on these two fuel sections essentially eliminated this possibility. Qualitative radial oxygen profiles for these two fuel sections, taken with the shielded electron microprobe, indicated that the fuel in the upper (cool) section contained more oxygen (O/M > 2.00) than the central section (O/M < 2.00). Partial pressure calculations for the various metal-bearing species (of the type originally performed by Rand and Markin [10]) indicate that uranium should move preferentially (as UO 3 ) to cooler regions inside the central void. This migration of U O 3 would also account for the observed axial relocation of oxygen. The resulting increase in O/M of the upper fuel section would then, because of the O/M - 7 p u correlation discussed previously, tend to bring about a greater degree of Pu enrichment due to radial actinide redistribution than at the central fuel section. Although it is probably over-simplified this explanation is fully consistent with the predictions of our current vapor transport models for both oxygen and actinide redistribution. IAEA-PL-46 3/12 235 SUMMARY AND CONCLUSIONS: DELINEATION OF KEY PARAMETERS IN RADIAL ACTINIDE REDISTRIBUTION Three important parameters that have been identified as contributing to the occurrence of this phenomenon are: fuel O/M (over-all value), maximum (center-line) temperature and time of operation at this temperature, and initial fuel density. Radial actinide redistribution is intimately associated with fuel restructuring and thus, when over-all O/M is favorable, the degree to which Pu depletion or enrichment takes place is directly dependent on fuel linear power rating and the consequent degree of fuel restructuring. The latter is a function of initial fuel density; above 96% theoretical density fuel, even at very high power ratings, shows very little actinide redistribution when compared with fuel of ~ 9 0 % theoretical density. In terms of the vapor transport model this is consistent with a drastic reduction in the available volume for vapor flow as fuel density increases. Fuel which undergoes appreciable restructuring without center melting shows Pu enrichment at the center line when its over-all O/M (20% Pu) >1.97; the degree of enrichment appears to increase linearly with O/M up to —2.00. Theoretically, a fuel should not show any actinide redistribution when its over-all O/M has reached — 1.97, however this remains to be confirmed. Because the center line temperature falls as restructuring proceeds (at constant power rating) it is not possible to give an exact figure for a "redistribution" temperature threshold. Such a determination is feasible however from the out-of-pile radial temperature gradient tests currently underway at GE. The uncertainties presently inherent in calculating center-line temperatures also tend to make an in-pile determination a rather futile exercise. However, based on vapor pressure calculations out-of-pile experiments it is considered unlikely that preferential evaporation-condensation is an effective mechanism for actinide redistribution at temperatures below 2000°C. ACKNOWLEDGMENTS This work was sponsored by the United States Atomic Energy Commission under Contracts AT(04-3)-189, PA 53 and AT(04-3)-189, PA 10. The author gratefully acknowledges the assistance of G. F. Melde, T. E. Lannin, and S. K, Evans in several aspects of this study. REFERENCES [1] ADAM SON, M. G., and AITKEN, E. A., "A Thermodynamic Data Program Involving Plutonia and Urania at High Temperatures, Quarterly Report 16," General Electric Company July 1971 (GEAP-12229) and Trans. ANS 14 No. 1,179 (1971). [2] BOBER, M., SARI, C., and SCH UM AKER, G., J. Nucl. Matl. 40 (1971) 341. [3] LACKEY, W. J., OLSEN, A. R„ M ILLER, J. L.,and BATES, D. K., Trans. NAS 14 (1971) 180. [4] MEYER, R. 0., BUTLER, E. M., and O'BOYLE, D. R., ANL-7929, February 1972 and Trans ANS 15 No.1,216 (1972). [5] LACKEY, W. J., HOMAN, F. J., and OLSEN, A. R., paper presented at the AN S Topical Meeting, "Reactor Materials Performance," Richland, Washington, April 23-26,1972. [6] See, for example, LANNIN, T. E., MELDE, G. F., and ROSENBAUM, H. S., Electron Probe Society of America, 5th National Meeting, July 1970, New York, paper 25. [71 For further details, see PA 10 series of reports: GEAP-10028-37 to 42, inclusive. [8] COMTE, M. MOUCHININO, M., and SCHMITZ, F., paper presented at the ANS Topical Meeting "Reactor Materials Performance," Richland, Washington, April 23-26,1972. [9] For further details about F3B; see GEAP-10028-38 and GEAP-10028-41 PA 10 Quarterlies. [10] M ARKIN, T. L., and RAND,M . H„ Proc. Symp. Thermodynamics Nucl. Matl., Vienna, 1967 IAEA (1968) 637. IAEA-PL-463/14 CHEMICAL INTERACTIONS OF FISSION PRODUCTS WITH STAINLESS STEEL CLADDINGS P. HOFMANN, O. GÖTZMANN Institut für M aterial- und Festkörperforschung, Kernforschungszentrum Karlsruhe, Karlsruhe, Federal Republic of Germany Abstract CHEMICAL INTERACTIONS OF FISSION PRODUCTS WITH STAINLESS STEEL CLADDINGS. As shown in many postirradiation investigations, fission product enhanced oxidation is the main cause of cladding attack in oxide fuel pins. Cladding attack becomes significant at temperatures above 500°C. To obtain a better understanding of the reaction possibilities of the various fission products, the dependence of those reactions on the clad temperature and the oxygen potential in the system, we have undertaken an out-of-pile test program to investigate the compatibility behaviour of fission product elements with various types of stainless steels. The results show that reactions between caesium and stainless steel occur only above a certain oxygen potential. The reactions of tellurium and selenium and aggressive fission product elements do not depend on the oxygen potential in the same way as do those of caesium. Iodine and bromine in the elementary state cause heavy attack on steel claddings. However, when bonded to caesium they are not dangerous. With surplus oxygen or moisture in the system, though, reactions may occur even with Csl and CsBr. Other fission products, like antimony, cadmium, indium and tin, also react with steels. The addition of oxygen-consuming materials to the fuel reduces most of the reactions, especially the dangerous ones with caesium. INTRODUCTION It was previously assumed that stainless steels would not raise any major compatibility problems since it had been concluded from thermo dynamic considerations that stoichiometric U0 2 and (U, Pu)02-x [x> 0 . 01] were compatible with steels. The oxygen potential of the fuel is too low to oxidize steel. An increase in the O/M ratio above 2. 00 for U 0 2 and above 1. 99 for (U, P u )0 2 leads to thermodynamic instability of the fuel/ cladding material system. This may result in oxidation reactions with the steel, particularly with chromium and iron. A number of out-of-pile studies of U 0 2 and (U, P u )0 2 with steels have confirmed this phenomenon [1-4]. However, examination of postirradiated fuel pins has shown that chemical reaction occurs in the fuel/clad system even if the system had been initially thermodynamically stable. In addition to oxidation reactions fission product reactions have been observed. At the phase boundary of nuclear fuel and cladding, in the reaction zone and in the grain boundaries of the cladding material Cs and/or Mo, and sometimes also oxygen, have been clearly detected with the microprobe [5-12]. Tellurium has been observed at the phase boundary [5,8-11] and in the grain boundaries [5,7,10], together with Cs and/or Mo. Iodine has been detected at the phase boundary or in the grain boundaries of the cladding [ 8], sometimes together with Cs [8,11]. 237 238 HOFMANN and GÖTZMANN Since the average affinity to oxygen of the fission products generated is not as high as that of uranium .and plutonium atoms, the oxygen potential in the fuel is raised in the course of burn-up. This rise in the oxygen potential is the most critical change with respect to the compatibility behaviour of the system because many fission product reactions with the cladding are only possible at a definite O/M ratio. In addition to the O/M ratio of the fuel, a variety of factors can exert an influence on the type and extent of attack on the. cladding material. In this context the burn- up and temperature play a decisive part. The burn-up opens the way to reactions with the cladding material (fission products, increase of O/M), while the temperature determines the reaction kinetics. Below a certain temperature the reactions will take place at such a slow rate that there is no significant attack on the cladding material. INVESTIGATION OF IRRADIATED FUEL PINS GfK Karlsruhe conducted fuel pin irradiation tests in thermal (BR2, FR2) and fast (DFR) neutron fluxes, using (U, P u )0 2 pellet fuel of different O/M ratios and sinter densities and different 235U enrichment. Burn-ups varied between 10 and 90 MW-d/kg. In most cases 1.4988 steel was used as cladding material (Table I). Investigations of irradiated samples did not yield noticeable reactions between fuel and cladding below 500°C, regardless of burn-up. The strongest-reactions with the cladding were always observed at the hottest point of the fuel pins. The maximum reaction depth was about 100/jm at maximum cladding temperatures between 560 and about 700°C (Fig. 1) [12, 13]. In the reaction zone and the grain boundaries of the cladding material an oxide phase rich in chromium was found to prevail. Although fission products must have participated in the strong reactions, caesium and some molybdenum were detectable in only one DFR irradiation sample in the reaction zone (Fig. 2) [14]. A metallic iron-nickel layer was formed immediately at the fuel surface (Fig. 2). In all the other samples examined with the microprobe only molybdenum, barium and sometimes caesium and tellurium were found at the fuel/cladding interface [ 1 2 ]. The various results obtained from postirradiation examinations can be compared only to a limited extent since the 235u enrichment varied and 235U fission - due to a change in the fission product yield - leads to a lesser oxidation of the fuel than does 239Pu fission. OUT-OF-PILE COMPATIBILITY STUDIES To study the reaction behaviour of the individual fission products as a function of various parameters (O/M ratio, temperature, burn-up, cladding material), comprehensive out-of-pile compatibility studies were made with pure fission products and mixtures of oxide nuclear fuel and fission products (Table I). Thermodynamic evaluations, the results of postirradiation examinations [5-14] and experiments with metallic [15,16] and non-metallic [16-18] fission products showed that Cs(+Rb), Te(+Se) and I(+Br) are primarily responsible for the chemical interaction with the IAEA-PL-463/14 239 TABLE I. EXPERIMENTS WITH SIMULATED FISSION PRODUCTS Stainless steel cladding materials for the compatibility tests with simulated fission products (wt. %) 1.4401 1.4541 1 .4970 1 .4981 1.4988 (SS 316) (SS 321) Fe balance balance balance balance balance Cr 1 6.8 18.1 1-6.1 1 7 .2 1 7 .4 Ni 12.3 9 .3 1 4.9 1 6.7 1 2 .8 Mo 2 .2 0 - 1.1 1 .77 1 .4 V - - 0. 04 - 0 .8 4 T i - 0 .4 7 0. 57 - - Nb - - - 0 .7 0 0. 89 Si 0 .5 6 0 .5 5 0. 52 0 .4 0 0 .4 8 Mn 1 .2 1 .1 1 .9 1 .2 1 .2 С 0 .0 4 4 0 .0 6 0.0 5 5 0 .0 9 4 0 .0 8 SS + U 02, (U .P u )02 SS + Se, T e, Cs, I, Br, Csl, CsBr, CszC 0 3, CsOH,'Cs2C r04 , Cs2Cr2 0 7 , Cd, In, Sn, Sb, M o03 SS + U 02 , (U, pu)02 . Se, T e, Cs, I (stabilized with Mo, Ni, Cr, Nb, V, T í, Zr) SS + U 02 + Se, Te, I, Csl, CsBr, Cs. Mo SS + U 02 + Se + Te (+ Mo) 10 and burn-up SS + UOg + Se + Te + Csl (+ Mo) 20°jo SS + U 02 , (U, Pu)02 + Se + Te + Csl + Cs (+ Mo + Ru) Annealing temperature: 300-1000eC Annealing time : 500-3000 h U02 : O/M = 2.00-2.08 (U. 7Pu0 . 3)Oz : O/M = 1 .9 6 -2 .0 8 cladding materials. Therefore, the reaction behaviour of these fission products with respect to various austenitic stainless steels was investigated first (Table I). The experiments were conducted under isothermal temperature condi tions. These simulating experiments do not seem to require a temperature gradient. Disregarding attacks to the cladding due to a van Arkel de Boer process, the temperature gradient only enhances the activity gradient of the reactive and highly volatile fission products towards the cladding. However, for reactive elements an activity gradient is reached also under isothermal conditions since the cladding represents a sink for these elements [ 19]. The fission products and mixtures of fission products and fuel, respectively, were pressed into little stainless steel cups with cylindrical holes, closed gas-tight with a conical plug (cold welding) and placed into 240 HOFMANN and GÖTZMANN FIG .l. Fuel cladding reactions in DFR 304/1 fuel pin: 1.4988 + (U0 8 PuQ 2)0 1 98, burn-up 49 462 MW-d/t, ЭЗ'Уо Z35U, En > 0.1 MeV, 650-700”C. FIG.2. Electron microprobe X-ray images of reaction phases of DFR 304/1 fuel pin (see Fig. 1). 242 HOFMANN and GOTZMANN 10mm i------1 Fuel-Fission Product Mixture 4 . и d -t - t ; л .. — ' V Я % « FIG.3. Annealing capsule with compatibility sample. annealing capsules provided with a gas-tight screw (Fig. 3). The fabrication ,of the fuel/fission product mixtures and the preparation of the cups and annealing capsules were done in glove boxes under a high-purity argon atmosphere (H20 and 0 2 content < 5 ppm). RESULTS The out-of-pile compatibility studies with pure fission products and simulated burn-up showed that above all tellurium, selenium and iodine - even in the case of substoichiometric fuel - react with the cladding material, while Cs and Csl (CsBr) undergo reactions above a definite oxygen potential only. Reactions with Te, Se, I and Br start at about 400°C. UO^ Stoichiometric U 0 2 is compatible with the steels up to at least 1000°C. UO2. 08 reacts only above 600°C. After 1000 h at 700°C the oxide layers of the cladding were only 1 /uzn thick. IAEA-PL-463/14 243 Cs, Cs compounds Oxygen-free caesium is compatible with the steels up to 3000 h at 1000°C. Even minor impurities of caesium with oxygen or water cause a reaction with the cladding. Figure 4 shows the dependence of the caesium reaction on the O/M ratio of the fuel. The oxidation of the cladding by hyperstoichio m etric fuel is considerably accelerated in the presence of Cs. While after 1000 h reaction zones of less than 5 ,um were observed in contact with UO2.08 800°C, 100 д т were attained after the addition of Cs (simulated burn-up 10 at. %) with all the other conditions unchanged (Fig. 4). If molybdenum is added to the mixture of U 0 2 08 + Cs, the reactions with the cladding become weaker (Fig. 5). The Cs compounds CsOH, Cs2C 03 and Cs2C r04 react very violently with the steels at 800°C and 1000 h and grain boundary reactions take place, penetrating more than 1000 jim , While Cs2C 03 is compatible with the steels for 1000 h at 400°C, reactions were found to occur between 30 and 100 ^m when CsOH was present and the temperature remained the same. Cs2C r20 7 also reacts with the cladding. The chemical interactions cover some 50 д т at 800°C and 1000 h. I, Br, I and Br compounds Investigations with I and Br reveal that reactions with the steels already take place at 400°C. Iodine and bromine react predominantly with Cr. There are preferential reactions along the grain boundaries of. the cladding material, yielding a type of reaction similar to that of pitting corrosion. However, the halogens are not present in their elemental forms in the fuel of a spent fuel pin but as the compounds Csl and CsBr, which are thermodynamically very stable. As high-purity compounds Csl and CsBr do not even react with the steels at 800° С for 1000 h. Nevertheless, above a critical oxygen potential or/and a certain impurity of the water reactions take place with the cladding. In experiments carried out in the presence of Csl for 1000 h at 800°C (simulated burn-up 20 at. %) there was no reaction for stoichiometric U 02, although marked reactions took place (about 20 цт) forU O 2 08 (Fig. 6). On the other hand, U 02 og reactions are much reduced in the absence of Csl (Fig. 6). The addition to U 0 2 08 °f free iodine without Cs (-simulated burn-up 20 at. %) caused reactions up to 50 ß m of depth in the cladding material after 1000 h at 800°C. For stoichiometric U02 the maximum depth attained by the reaction was about 10 pm at some points, the other experimental conditions remaining unchanged (Fig. 7). Moreover, in this casé the oxidation of the cladding by hyperstoichiometric fuel is considerably accelerated in the presence of Csl and I. Se and Te Pure selenium and tellurium cause the strongest chemical interactions with all cladding materials. At temperatures below 800°C Se proves to be more aggressive to steels than Te (Figs 8 and 9). However, much smaller amounts of Se are generated during irradiation than of Te so that it does not constitute a specific problem. The maximum penetration depths 244 HOFMANN and GOTZMANN FIG. 4. Dependence of caesium reactions on O/M ratio of the fuel, 800eC/1000 h, simulated burn-up 10 at.^o: (a) 1.4988 + U02 û4 + Cs; (b) 1. 4988 + U02 _ 08 + Cs; (c) 1.4988 + U02, 08 ; (d) 1.4988 + U02 e 00 + Cs. IAEA-PL-46 3/14 245 (b) FIG. 5. Influence of Mo on UOa + Cs reactions with stainless steel, 800°C/1000 h: (a) 1.4988 +U 02 08 + Cs, simulated burn-up 10 at.°/o; (b) 1.4988 + U02. m + Cs, simulated burn-up 20 at.‘fo; (c) 1.4988 + UO., ю + Cs + Mo, simulated burn-up 20 at.°fo. 246 HOFMANN and GÖTZMANN FIG. 6. Dependence of Csl reactions on O/M ratio of the fuel: 800"C/1000 h, simulated burn-up 20 at.^o; '(a) 1.4988 + U02 _ „„ + Csl; (b) 1.4988 + UO2-08 + Csl; (c) 1.4988 + U02 . 08. IAEA-PL-463/14 247 FIG. 7. Dependence of I reactionson O/M ratio of the fuel: 800°C/1000 h, simulated burn-up 20 at.'/a; (a) 1.4988 + UO2-00 +1; (b) etched; (c) 1.4988 + U02 _ os +1; (d) etched. 248 HOFMANN and GOTZMANN FIG. 8. Depth of penetration of (a) Te and (b) Se in 1.4988 steel, 700°C/1000 h. lAEA-PL-463/14 249 ТСС) FIG. 9. Comparison of the penetration depths of Те and Se in 1.4988 steel. of Te into the steels 1.4970 and 1.4988 as a function of the temperature and annealing period are shown in Fig. 10. Steel 1.4988 shows a slightly better behaviour with respect to the Te reactions than steel 1.4970; the other types of steel examined (1.4401, 1.4541) range in between. Beginning at 500 and 700°C the attack along the grain boundaries of the cladding is predominantly from Se and Te, respectively. Below these temperatures a uniform attack is observed, which migrates into the cladding (inter facial reaction). Above 700°C the depths of penetration of Se and Te are dependent on the grain size of the cladding material. For coarse grains the maximum depths of penetration of Se and Te are greater than for fine grains [18]. Due to the solution of the cladding in Se and Te the cladding components diffuse into the fuel even at a temperature as low as 400°C, especially Cr; Te and Se diffuse into the cladding material and are found at the reaction front together with Cr [19]. The reactions of tellurium and selenium with the cladding do not depend on the oxygen potential of the fuel in the same manner as do those of Cs. There is merely an overlapping of the reactions. The reactions with Te and Se are clearly a function of burn-up and temperature. This fact is demonstrated in Fig. 11. Also in this, case the addition of metallic molybdenum to U 0 2 00 + Se + Te - similar to the simulated burn-up - reduces the chemical interactions with the cladding at 800°C and after 1000 h (Fig. 12). Simulated system: UO?, + Te + Se + Cs + I If Cs and I are added to the partly simulated system of U 02 + Se + Te, the reaction with the cladding is considerably intensified, especially in combination with hyperstoichiometric fuel. For example, reactions up to about 200/um in depth were observed in the cladding material with UO„ Z, 08 (simulated burn-up 10 at. %) at 800 С and after 1000 h (Fig. 13). 250 HOFMANN and GÖTZMANN T(°C) T (°C) FIG. 10. Depths of penetration of Te in 1. 4970 and 1.4988 steels after various annealing times. While Cs and Te were detected in the reaction zone of the cladding, the grain boundaries yielded only Cs together with Cr. In this system the reactions depend not only on the simulated burn-up and the temperature but, in addition, on the oxygen potential of the fuel. Cs proves to be a hazardous fission product, above all with- hyperstoichiometric fuel. Also in this case Mo did not adversely affect the reaction behaviour of the partly simulated system with respect to the cladding; sometimes a minor reduction in the reactions was also noticeable here. IAEA-PL-463/14 251 -, ■ I------1 (a) ¿C pm t...... -1 FIG. 11. (U02 < 00 + Те + Se) reactions and influence of Mo on clad attack, 800*C/1000 h: (a) 1.4988 + U0210Q + Te + Se, simulated burn-up 10 at.^o; (b) 1.4988 + UO2>00 + Te + Se, simulated burn-up 20 at.%; (c) 1.4988 + UO2>00 + Te + Se + Mo, simulated burn-up 20 at.^o. 252 HOFMANN and GÖTZMANN ? V •* 't' ■л X ^ Я ?í - -,,íí‘!"'í с/ ■ ■'•••с .' ' иииииимииииииии!■ '* . ' ' . 4 / - ••*, ; Л - ".Ä ' ~” -Î"* ' ’ ' ■"• l il il í i»ü -йХ.:ь1 (a) .♦■-•■ •“•■ -'.V,-'- •:-:;--;it!';v':j^ y > :. ’.^ . A *. *'!;í^ ‘ : ; >:\v ■: r>.,•■;-■%*<’ V \ '':V > .‘ ’" * 'Л ^^ДШ ^^Д1^Ш М1ДДЁ1НИМямш§Д|181^М ш^^Д|НМитИ1 -- <■ •“ *.- . i ! ~ Г.-f iv • . ,.'-:¿a (Ь) FIG. 12. Chemical interactions of Mo03 with 1.4988 steel: (a) 600“C/1000 h; (b) 800eC/1000 h. Sb, Cd, In and Sn The metals antimony, cadmium, indium and tin, whose melting point is low, also react with the steels. Antimony and cadmium react pre ferentially with Cr in reaction zones of some lOOium of depth and after 1000 h at 800°C. The reactions with indium and tin are stronger. Reaction IAEA-PL-463/14 253 (a) i. . i------1 (b) FIG. 13. Comparison of fuel/cladding reactions out-of-pile and in-pile: (a) 1.4988 + UO2>08 + Te + Se + Csl + Cs, 800eC/1000 h, simulated burn-up 10 at. (b) 1.4 9 8 8 + (UOQ ^ 8 PuQ _ 2 )Ol 98 , 650 -7 00°C/5000 h, burn-up 49 462 M W d/t, 93% 235 U, En > 0 .1 MeV (DFR 304). zones with depths of about 120 pim (In) -and 350/um (Sn) are observed at 800° С and 1000 h. The preferred reaction partners are nickel and iron. Obviously, we do not anticipate the same reactivity in the fuel pin for these fission products. They are produced in only small amounts during burn-up so that the reactions will be reduced and, probably, even become negligible. 254 HOFMANN and GÔTZMANN Metallic fission products All the other fission products generated in nuclear fission - primarily m etallic fission products - do not show specific incompatibility reactions with respect to the types of steel used. In these examinations Mo did not prove to be a reactive fission product, even at an O/M ratio of the fuel of 2. 08. Pure M 0 O 3 reacts with the steels only above 500°C. At 600°C the Cr-bearing oxide phase in steel 1.4988 was about 5 цтп thick after 1000 h (Fig. 12). At the same time iron and Cr diffused into M o03, penetrating to about 50 (jm. COMPARISON OF OUT-OF-PILE AND IN-PILE CONDITIONS Figure 13 gives a comparison of the reaction zones obtained out-of-pile (UO 2 08 + Te + Se + Cs + I; simulated burn-up 10 at. %, 800°C for 1000 h) and in-pile (burn-up about 5 at. %, (U, Pu)0198, 6 5 0 - 7 0 0°C for approximately 5000 h). The reaction zones are sim ilar to each other and have nearly identical compositions. A metallic phase containing iron and nickel is formed immediately at the fuel surface; then comes a caesium-rich oxide phase followed by grain boundary reaction products (Fig. 2). In both cases the cladding materials was steel 1.4988. REDUCTION OF THE REACTIONS The addition of an oxygen-binding material (getter: Zr, Ti, Nb) to the fuel or fuel pin remarkably reduced or even prevented the reactions with the cladding [17, 20]. This applies in particular to the dangerous caesium reactions. Some of the getters studied (Zr, Nb) also bind the reactive fission products Te and Se. Studies relating to the optimum location of such getters and their effects are being performed at present. REFERENCES [1] GÔTZMANN, O ., THÜMMLER, F., Rep. KFK-1081 (1969). [2] PRICE, D .E. et a l . , Rep. BM I-1900 (1968) VI. [3] LAURITZEN, T .. Rep. GEAP-5633 (1968). [4] SMITH, F.M ., Rep. BNWL-1101 (1969). [5] RUBIN, B . F . e t a l . , Rep. G EA P-10028-38 (1971). [6] JOHNSON, C .E ., JOHNSON, I . , CROUTHAMEL, C .E ., Rep. ANL 7822 (1971) 25. [7] PERRY, K .J ., BAILY, W. E ., Rep. GEAP-13729 (1971). [8] JOHNSON, C .E . e t a l . , Rep. ANL-7775 (1971) 89. • [9] RUBIN, B .F . et a l . , Rep. GEA P-10028-36 (1970) 48. [10] RUBIN, B .F. et a l . , Rep. G EA P-10028-37 (1971) 24. [11] JOHNSON, C .E., CROUTHAMEL, C.E., J. Nucl. Mater. 24 (1970) 101. [12] HUBER, H ., KLEYKAMP, H ., Rep. KFK 1324 (1972). [13] GÔTZMANN, О., unpublished report. [14] KLEYKAMP, H ., unpublished report. [15] HOFMANN, P ., THÜMMLER, F ., WEDEMEYER, H ., Rep. KFK 979 (1969) [EURFNR-685]. [16] HOFMANN, P ., THÜMMLER, F ., WEDEMEYER, H ., Reaktortagung Berlin 1970, Rep. AED-CONF. 70-116, p. 579. [17] HOFMANN, P., GÔTZMANN, О., Reaktortagung Hamburg 1972, p. 367. [18] GÔTZMANN, О ., HOFMANN, P ., Rep. KFK 1619 (1972). [19] GÔTZMANN, О ., HOFMANN, P ., Rep. KFK 1271/4 (1972) 112-20. [20] HOFMANN, P.. GÔTZMANN, О., Rep. KFK 1272/1 (1972). IAEA-PL-463/15 FUEL/CLADDING COMPATIBILITY OF STAINLESS STEELS WITH GAS AND SODIUM-BONDED URANIUM PLUTONIUM CARBIDE FUELS O. GÖTZMANN Institut fur M aterial- und Festkörperforschung, Kernforschungszentrum Karlsruhe R .W . OHSE European Institute for Transuranium Elements, EURATOM, Karlsruhe, Federal Republic of Germany Abstract FUEL/CLADDING COMPATIBILITY OF STAINLESS STEELS WITH GAS AND SODIUM-BONDED URANIUM PLUTONIUM CARBIDE FUELS. An assessment is made of the compatibility of stainless steels with gas and sodium-bonded uranium plutonium carbide fuels by considering compatibility results, thermodynamic data and phase relationships. The fuel stabilization is considered both on the basis of the carbon potential of the fuel and the buffering carbides and on their location in the appropriate phase diagram. The influence of the radial temperature gradient and burn-up on the carbon potential of the fuel is discussed. Finally, clad carburization, carbon activity and solubility in steel and sodium and its effect on the mechanical properties of the steel are considered with special attention to carbon transfer and the effect of oxygen dissolved in sodium. 1. INTRODUCTION Clad compatibility, as one of the main life-lim iting factors of a fuel pin, is clearly of importance when considering the applicability of a fuel m aterial on the basis of economic and safety considerations. A considerable number of compatibility tests within the various advanced fuel programs [ 1 - 8 ] on uranium and uranium plutonium carbide fuels have clearly outlined the main problems. One of the main features of the carbides is the narrow homo geneity range of the single-phase region of UC and (U, Pu)C below 1000°C. Even a slight change in composition leads into the adjacent two-phase fields M+MC or MC+M 2Cg. Although, according to the thermodynamic data, even single-phase carbides are unstable in the presence of austenitic stainless steels and even more so with nickel- or vanadium-base alloys, good com patibility has been observed with stainless steels up to 800°C. The main compatibility problem arises in the two-phase fields. The carbides produced by the carbothermic reduction in an industrial fabrication process always contain small amounts of sesquicarbides or even dicarbides [9], which for kinetic reasons are less compatible than the sesquicarbides. The rate of carburization in the presence'of sesquicarbides in the fuel increases with increasing carbon activity. Hypostoichiometric fuels, containing free m etals, are excluded because of possible eutectic reactions with steel components at low temperatures. 255 256 GÔTZMANN and OHSE Even though these eutectic reactions have not been observed below 900°C, the formation of intermetallic compounds [10, 11] of the (U,Pu)Fe2 or (U, Pu)Ni5 type is likely to take place. Attempts have, therefore, been made to stabilize the single-phase product [11-14] by suitable additives. The main topics, such as fuel stabilization, carbon potential as a function of temperature gradient and burn-up, carbon transfer in sodium bonding, clad carburization, and the role of oxygen con tamination are discussed in the following review of the literature on the basis of thermodynamics, phase relationships and the available experimental re s u lts . 2. THERMODYNAMICS AND FUEL STABILIZATION Fuel compositions of both two-phase fields M+MC and MC+M2C3 were found to be incompatible with austenitic stainless steels. A minimum of chemical interaction was found in contact with monocarbides. Compatibility tests with hyperstoichiometric fuels containing sesquicarbides and eventually dicarbides led to the carburization ofthe stainless steelby forming (Cr, Fe)23Ce precipitates [1, 15] to an extent depending on the carbon potential of the fuel. Tests with hypostoichiometric fuels, containing free metals, show alloy formation such as (U, Pu)Fe2 [11] in iron-rich alloys or (U, Pu)Nis in nickel- rich alloys. Since carbide fuels, produced by the carbothermic reduction, contain small amounts of sesquicarbides, their chemical interaction with steel cladding up to 800°C consists almost entirely of clad carburization. The formation of intermetallic compounds with components of the cladding m aterial plays a major role only with alloys of a higher nickel content (>25 wt. %) [15] and with stainless steels in case of hypostoichiometric fuel compositions. The carbon potentials for the various reactions of significance are given in Fig. 1. The major components of the austenitic steels, Fe, Cr and Ni show a different reaction behaviour with UC or (U, Pu)C. Chromium is likely to react with carbon to Cr23C6 or Cr7C3, whereas iron and nickel form alloys of the (U, Pu)(Fe, Ni)2 or (U, Pu)(Ni, Fe)6 type, depending on the nickel content of the steel. Iron alone is not capable of reacting with UC or (U, Pu)C below the eutectic temperature of approximately 1100°C since UFe2 is less stable than UC. Nickel, however, is likely to react with UC or (U, Pu)C since UNi5 is a rather stable compound. Nickel-base alloys are therefore not compatible with carbide fuels, even at temperatures as low as 700°C. In view of the complex reaction [16] forming chromium carbide and an intermetallic compound of the UFe2type, stainless steels are also not thermo dynamically stable, even with single-phase carbide fuels. It has been shown, however, in many out-of-pile tests that this instability does not lead to any appreciable reaction below 800°C. Because of the narrow homogeneity range of the monocarbide single phase region, even a slight change in its carbon content leads to the adjacent two-phase fields. Apart from technological difficulties during the production step to keep the carbon content within tolerable limits on a commercial basis, changes during handling and filling of the pins due to uptake of oxygen and nitrogen, which act as carbon equivalents, are unavoidable. To avoid the highly reactive second phase, attempts are being made to stabilize the single-phase fuel. IAEA-PL-463/15 257. FIG.l. Temperature dependence of carbon potential. In principle this can be achieved in three ways: A surplus of carbon in the case of hyperstoichiometric fuels can be consumed by adding metals that form binary carbides with a carbon activity less than that of MC in equilibrium with M2C3. Likewise, for hypostoichiometic fuels the free metal is consumed by adding carbides whose carbon activity is greater than that of MC in equilib rium with free metal. This means that the buffering phase should have a carbon potential within these boundaries. The condition of remaining within the carbon potential field of the single-phase carbide fuel leads to binary end members in the ternary-phase diagram, which form a three-phase field discussed further below. A stabilization effect can also be achieved in the case of a hypostoichio m etric composition by adding a metal to the fuel that forms an interm et a lli с compound with uranium and plutonium. The stability of this compound should not be greater than that of UC. It can also be achieved by adding an inter- metallic compound to the hyperstoichiometric fuel to consume the surplus carbon by forming carbides. The intermetallic compound should again be less stable than the carbides at the upper phase boundary of the single-phase 258 GÖTZMANN and OHSE field in equilibrium with the sesquicarbide. This second method of fuel stabilization has been successfully performed by using Fe and UFe2[ 10, 13] . The disadvantage of using Fe and UFe2 as buffering agents is a low melting temperature in the UC-Fe system of approximately 1100°C and, in addition, the comparatively high swelling rate of Fe-stabilized hypostoichiometric fuels under neutron irradiation. In the case of a ternary compound formation such as UMoC2, as a third possibility, the buffering ability should not only be checked by a comparison of carbon potentials but also by its position in the appropriate ternary-phase diagram [12]. According to the phase rule, the activities of all three components remain constant within a three-phase field. Therefore, the buffering condition in the sense of keeping the carbon activity constant requires the presence of three phases, which in the case of adding Mo are given by UC, UMoC2 and Mo [12, 17] in Fig. 2, Stabilization tests with UC containing 10 at. % Mo show a variation tolerance in carbon content from 45 to 50 at.% without the appearance of free metal or a sesquiphase. Similar results could be obtained by adding only 5 at.% of tungsten forming UWC2[12]. No embrittlement of stainless steel was observed after 3000 h at 800°C. С The carbon potentials given in Fig. 1 permit a preliminary selection of buffering agents where carbides are formed by carbon exchange reactions. To judge the stabilization effect attainable by adding metals such as chromium that form binary carbides, the carbon potential of the ternary fuel (U,,Pu)C [18, 19] has to be discussed more closely. Carbon activities were obtained by measuring the ion intensity ratio UC2/UC+ [20, 21] over the composition range from C/U= 0. 9 to 1. 9. The carbon activity at 2100°K changes from 0. 006 at C/U= 0. 9 in equilibrium with uranium .metal to 0. 1 at C/U= 1. O.and 0. 65 at C/U= 1. 1 in equilibrium with the uranium dicarbide. The carbon potentials in the ternary systems were, calculated from the binary data under the assumption of ideal solution. The carbon potentials of the lower and upper phase boundary of the monocarbide single-phase field of the binary systems UC and PuC are given in Fig. 1 by the reactions M+C = MC IAEA-PL-463/15 259 and 2 MC+C=2 MC-j 5. One must bear in mind that the carbon potential showing a bivariant behaviour in the single-phase field by forming a band between these boundary lines consists of a constant line in the adjacent two-phase fields, whereas in the ternary U-Pu-C system the two-phase region is represented by a band as shown in Fig. 3. According to Fig. 3, the carbon potential of the upper limit of the ternary monocarbide homogeneity range at 700°C is approximately -4 kcal/g • atom C, whereas the lower limit will be around -15 kcal/g - atom C. Compared to the ternary fuel, the carbon potential of Сг23С6/Сг7С3 at 700°C will be slightly more positive, according to to Fig. 1 around -13 kcal/g • atom C, than the lower-phase boundary of the ternary monocarbide. Compatibility tests made with Cr23C6 confirmed the thermodynamic prediction by showing good buffering ability. According to Fig. 1, Zr, Nb, Ti and Та are not suited since they form stable carbides, which would reduce the single-phase UC. Temperature (°K) FIG.3. Carbon potential of the ternary U -P u -C monocarbide phase field for Pu/(U + Pu) = 0.2. 3. THE EFFECT OF OXYGEN IMPURITIES, TEMPERATURE GRADIENT AND BURN-UP ON THE CARBON POTENTIAL OF THE FUEL The production of carbide fuels by carbothermic reduction leads to oxygen impurity levels of the order of 0. 3 to 0. 5 wt. % and to sm all amounts of sesquicarbides. According to the U-Pu-C-O [18, 23] phase diagram, the corresponding fuel compositions will be at the lower phase boundary of the (U, PuJCj 5 monoxycarbide two-phase field. The calculated CO equilibrium pressure at a centre temperature of 1400°C is of the order of 10'3 atm [22-25] . 260 GÖTZMANN and OHSE If we assume a constant pressure ratio of P FIG .4. Carbon potential for various P ç 0 /t>co ratios according to reaction 2 C 0 .-'C + C 0 2 compared with carbide fuels. 2 A change in C/(U+Pu) ratio as a function of burn-up largely depends on the chemical state of the rare earth elements La, Ce, Sm and Nd, which could be dissolved in the (U,Pu) monocarbides. The carbon potential could either be buffered by Ba and Sr, which form dicarbides (Sr, Ba)C2, or by Mo and Tc, which form ternary compounds of the (U,Pu)MC2 type. Zr and Nb are present as stable monocarbides. Ru and Rh form the ternary U2RuC2 and UgRhCg compounds [2 7 ]. With the assumption that the rare earth elements are forming mono- carbides, an increase in carbon potential as a function of burn-up is not to be expected. Fission yield calculations for a fast neutron flux result in a IAEA-PL-463/15 261 small decrease in the C/M ratio of the fuel. A decrease in carbon potential in the fuel during burn-up is also suggested by postirradiation examinations of vanadium stabilized (U, Pu)C. It was found that the initial VC/V2C ratio had changed in favour of an increase in the V2C phase [28] . The chemical interaction of a burned oxide fuel with stainless steel is mainly caused by fission products, especially caesium [29-32] . In carbide fuels most of the fission products, as discussed above, form stable carbides and are, therefore, not migrating towards the clad but will be tied up in the fuel. In the case of caesium no stable carbides are to be expected. Apart from small amounts of Csl, it is likely to remain in its metallic state and will, therefore, migrate towards the clad. Since the oxygen potential of the oxycarbides is far too small, the possible amounts of oxygen dissolved in caesium, as shown in Fig. 5, are far below 1 ppm. It should, therefore, show good compatibility with stainless steels. In the case of sodium-bonded carbides the influence of caesium can certainly be neglected compared to sodium since the oxygen potential of oxygen containing sodium or even N a^ is sufficiently negative and, therefore, of the order of the oxycarbide fuel. Sodium should, therefore, be more aggressive to stainless steel than caesiu m . Iodine will be fixed by the Csl compounds [33, 34] . Tellurium is reported to form stable compounds with the carbide fuel [35]. No fission product reactions with cladding m aterials in carbide fuels have been reported so fa r. Temperature (°C) FIG. 5. Oxygen potential of oxygen dissolved in caesium and sodium. 4. THE EFFECT OF SODIUM BONDING The main advantage of liquid sodium bonding in carbide fuel pins is the reduction of the fuel surface temperature because of the high heat transfer of sodium and the greater gap tolerance to accommodate for fuel swelling. 262 GÔTZMANN and OHSE The disadvantages of the sodium, bonding were found to be higher rates of carbon transfer, a possible solubility of steel components in sodium, and an increase in carburization of the clad by carbon released as a result of fuel oxidation by oxygen dissolved in sodium. Even oxygen contents of 5-10 ppm not only increased the solubility of the steel components by oxygen exchange reactions but clearly oxidized uranium monocarbide [2, 36] forming oxide layers, and by increasing the carbon potential in the sodium accelerating the carbon transfer to the clad. Cr20 3 layers on stainless steels were dissolved in oxygen containing sodium to form sodium chromite N a C r0 2 [37, 38] . A surface depletion of both chromium and nickel could be observed by microprobe analysis. Except for the higher rate of carbon transfer, oxygen dissolved in sodium will not markedly influence the reaction behaviour with the clad since the amount of sodium in the pin is rather small and, therefore, like wise the amount of oxygen dissolved in sodium, which is responsible for these exchange reactions. Therefore, the carburization possibility of the clad due to exchange reactions with the oxygen dissolved in sodium must remain small since the calculation for a fuel pin with a pellet diameter of 8 mm and a diagonal gap of 1 mm with a wall thickness of 0. 35 mm shows that 1 0 0 ppm oxygen dissolved in the sodium bonding yields an increase in the overall carbon content of the steel clad of only 0. 0011 wt. %. This is ■small compared with the normal carbon content of austenitic steels of 0. 07 to 0. 1 wt. %. Even a local carburization of the cladding caused by an inhomo geneous distribution of the carbon content due to the oxygen exchange reac- tion of the order of 1 0 0 : 1 would not cause severe embrittlement at the steel surface since only carbon contents in austenitic steels of 0 . 6 wt. % are con sidered to be too high with regard to embrittlement. 1 0 0 ppm of oxygen dissolved in sodium, on the other hand, corresponds to an increase in the overall equivalent carbon content in the fuel of only 0. 00013 wt. %. Besides this, the specification of the carbon content of the fuel is possible to no more than 0. 01 wt.%. So far as long-term carburization is considered, as will be in the case of reactor operation, the carburization of the clad due to oxygen . impurities in the sodium bonding can be neglected. Similarly, the dissolution of cladding components due to oxidation by the oxygen dissolved in sodium can be neglected since the high affinity of carbide fuels to oxygen keeps the oxygen potential of the system too low. This, of course, does not hold true for the sodium loop where 'the oxygen impurities are practically kept constant at the given flow rate of the sodium. 5. CLAD CARBURIZATION The long-term carburization behaviour of austenitic stainless steels presents the main compatibility problem and requires special attention because of its effect on the mechanical properties of the steel clad. The carbon transfer via a sodium bond is believed to be more rapid and carbur ization of the clad, therefore, more pronounced than in gas-bonded pins. Carbon activity and solubility in austenitic steels with differing chromium contents are given in Figs 6 and 7 [39]. The carbon solubility decreases with increasing chromium content and decreasing temperature. If the solu bility limit is exceeded, a second phase, М 2зС6(М = Сг, Fe), is precipitated in .the temperature range from. 500 to 800°C and at a chromium content of IAEA-PL-463/15 263 T °c «Зд (° r 1) FIG.6 . Temperature dependence of carbon activity at they-carbide phase boundary in F e-C r-8 wt.^oNi alloys with various Cr concentrations. T °C 103/T (°Г1 ) FIG. 7. Temperature dependence of carbon solubility in Fe-Cr- 8 wt,°]o Ni alloys. 264 GÔTZMANN and OHSE 15 wt. % and above. Carbon contents of austenitic steels are of the order of 0. 07 to 0. 1 wt. %. If the carbon content reaches values above 0. 6 wt. %, embrittlement of the austenitic steel takes place. Stainless steels exposed to sodium containing only 2 ppm carbon at 700°C exhibit a surface carbon concentration of practically 0. 6 wt. %. As shown in Fig. 8 [39] , sodium containing 0. 03 ppm carbon leads to M 23C6 precipitations at 700°C, corres ponding to a carbon potential of -18 kcal/g. atom C. These values are only of significance for a flowing sodium loop, whereas in the case of sodium bonding with a low absolute amount of carbon its depletion would prevent further precipitation, unless the carbon level can be kept constant by a carbon transfer from the fuel. As shown in Fig. 3, the carbon potential range of the ternary monocarbides at 700°C reaches from -15 kcal/g - atom С at the lower boundary to -4 kcal/g- atom С at the upper phase boundary and would indeed allow Cr 23C6 precipitation. T °C FIG. 8 . Temperature dependence of the carbon concentration in sodium in equilibrium with carbon saturated F e-C r- 8 wt.% Ni alloys with various Cr contents. Carburization of austenitic steels with nickel contents up to 25 wt.% remained small in contact with either stoichiometric or hyperstoichiometric sesquicarbide-containing carbide fuels at temperatures up to 800°C [1,40]. Sodium bonding does not bring about a more severe effect under the same temperature and compositional conditions [40] . More carbide precipitation was found in contact with hyperstoichiometric fuels containing dicarbide as second phase [15]. When comparing the in-pile and isothermal out-of pile compatibility behaviour, the difference in carbon potential at the fuel surface and the kinetic of carbon transfer has to be considered. IAEA-PL-463/15 265 • Vacuum 600 °C •^*4^ • UC " ------I 1 1 ?--- 700 °C Vacuum Ü C ~ *~ 1 1 1 1 Vacuum 800 °C .. UC i i i i 0 200 400 600 800 1000 TltC h F IG .9. Ductility of 1.4988 steel at room temperature, solution treated, aged 3 h at 750eC, and then annealed in vacuum (O) in contact with UC ( • ) . TIME h FIG.10. Ductility of 1.4988 steel at room temperature, solution treated, aged 3 h at 750°C, and then annealed in vacuum (O) in contact with U C 1 + X ( • ) . 266 GÔTZMANN and OHSE TIKE h FIG. 11. Ductility of 316 ss at room temperature, solution treated, aged 3 h at 750 °C, and then annealed in vacuum (O) in contact with UC1 + X (« ). Assuming a radial gas-phase connection in the fuei by either an inter connected porosity or radial cracks, a carbon transfer by CO, as shown in Fig. 4, leads to a higher carbon potential at the fuel surface. However, not only the carbon potential from the thermodynamic point of view, but also the kinetics of carbon transport via the gas phase has to be accounted for. The high carbon release of UC 2 would tend to give a better isothermal approach of the compatibility behaviour under a temperature gradient. In tests with dicarbide as a second phase in the fuel precipitation zones caused by inter actions with the fuel were already observed in the cladding m aterials at 600°C [15]. There is an increase in precipitation in the clad with tempera ture. At 800°C, where most precipitation was observed, precipitation zones reached depths of more than 500 (im after annealing for 500 h. The pre cipitation behaviour of the various steels depended greatly on their compo sition and stabilizing or modifying additives. Steel of type 1.4981 showed the least and type AISI 316 SS the most precipitation. At temperatures above 900°C a strong reaction between steels and carbide fuels of all compositions with intermetallic compounds and chromium carbide'can be observed [15,41]. Postirradiation investigations of hypostoichiometric fuel pins showed more clad carburization than was expected from out-of-pile tests with fuels containing sesquicarbide. This carburization, however, was not considered as detrimental. No fission product reactions with the clad were observed [42]. The room-temperature mechanical properties of type 1.4988 steel were not affected after annealing at temperatures from 600 to 800°C in contact with practically stoichiometric UC (equiv. C~4.85 wt. %) as compared with spe cimens annealed under vacuum (Fig..9) [43]. Similar annealing in contact with hyperstoichiometric UC (equiv. C~4. 95 wt.%) with UC2 as second.- IAEA-PL-463/15 267 phase at the same temperatures showed a marked decrease in room- temperature ductility (Figs 10 and 11) [44]. However, ductility is still maintained at an acceptable level. The various test results suggest that compatibility will not become an insurmountable problem for carbide fuels. REFERENCES [1] LATIMER, T.W ., Rep.ANL 7827 (1971). [2] ELKINS, P.E., Rep. NAA-SR-7502 (1964). [3] BATEY, W ., BOLTON, K ., DONALDSON, D .M ., YATES, G ., Carbides in Nuclear Energy 1_, M acm illan and C o ., London (1964) 392. [4] STRASSER, A.A ., KITTEL, J.H ., Proc. Symp. Plutonium Fuels Technology 1967, Nucl. Metall., Metall Soc.AIME 13 (1968) 469. [5] HAYES, W.C., Rep.TID-7676, 35. [6] Rep. WARD - 3791 -4 5 (1970). [7] GÔTZMANN, O ., HOFMANN, P ., STUMPF, W ., Fast Reactor Fuel and Fuel Elements, GfK, Conf. - 700920 (1970). [8] Reps TUSR-10, 11 (1970, 1971). [9] HODKIN D .J ., EVANS, J .P ., MARDON, P .G ., Rep.AERE-R 5173 (1966). [10] JACOBY, W.R., PALM, R.G., LATIMER, T.W .. Trans. Amer.Nucl. Soc. 10 1 (1967). [11] GÔTZMANN, О., SCHERBL, D ., Rep. KFK 1213 (1970). [12] CHUBB, W ., N u cl.S ei.E n g . 29 (1967) 176. [13] JORDAN, K .R ., HERBST, R .J ., PALM, R .G ., Rep.ANL-7120 (1965) 301. [14] GORLE, F ., TIMMERMANNS, W ., CASTEELS, F ., VANPEEL, J .. BRABERS, M ., Thermodynamics of Nuclear Materials, 1967 (Proc.Symp. Vienna, 1967), IAEA, Vienna (1968) 481. [15] GÔTZMANN, О., HOFMANN, P ., KFK Rep., to be published. [16] GÔTZMANN, О., Rep., KFK 1111, EUR-4315.d, VIII (1969). [17] HOLLECK, H ., KLEYKAMP, H ., J. N ucl. M ater. 32 (1969) 1. [18] BLANK, H ., Reps KFK 1111, EU R-4315.d, 11 (1969). [19] MARDON, P .G ., POTTER, P .E ., Plutonium 1970 and Other Actinides (Proc.C onf.Santa Fe, 1970), Nucl. M etall., Metall. Soc. AIME 1Л pt. II (1970) &42. [20] STORMS, E.K., Thermodynamics (Proc.Symp. Vienna, 1965) 1, IAEA, Vienna (1966) 309. [21] STORMS, E.K., IAEA Panel Thermodynamics of the UC-and PuC-Systems, IAEA, Vienna, 1968. [22] POTTER, P.E., Rep.AERE-R 6890 (1971). [23] POTTER, P.E., Plutonium 1970 and Other Actinides (Proc. Conf. Santa Fe, 1970), Nucl. M etall., Metall., Soc.AIME 17 pt.II (1970) 859. [24] POTTER, P.E., Thermodynamics of Nuclear Materials, 1967 (Proc.Symp. Vienna, 1967), IAEA, Vienna (1968) 337. [25] OHSE, R .W ., SCHLECHTER, M ., ZAMORANI, E ., TUSR 9_ (1970) 36. [26] STEELE, B.C.H., JA VED, N.A., ALCOCK, C.B., J.Nucl. Mater. 35 (1970) 1. [27] HOLLECK, H ., SMAILOS, M ., Rep. KFK 1272Д , 112 -4 0 . [28] DELBRASSINE, A ., COOUERELLE, M ., VERSTAPPEN, G ., presented GCFR Specialist Mtg AERE, Harwell, 1971. [29] RUBIN, D .F ., e t a l . , Rep. GEAP 10028-38 (1971). [30] JOHNSON, C .E ., JOHNSON, I ., CROUTHAMEL, C .E ., Rep. ANL 7822 (1971) 25. [31] OHSE, R .W ., SCHLECHTER, M ., these Proceedings, IAEA-PL-463/18. [32] HOFMANN, P ., GÔTZMANN, О .; these Proceedings, IAEA-PL-463/14. [33] OHSE, R .W ., SCHLECHTER, M ., TUBR 10 (1970). [34] FITTS, R.B., LONG, E.L., LEITNAKER J.M ., Rep. ORNL-TM-3385 (1971)/ [35] BREESE, E .W ., Ph.D . Thesis, Sheffield U n iv ., 1971. [36] WATANABE, H., KIKUCHI, T ., FURUKAWA, K., J.Nucl. Mater. 43 (1972) 321. [37] BORGSTEDT, H.U., FREES, G., Werkst. Korros. 21 (1970) 435. [38] CASTEELS, F ., COOLS, A ., DRESSELAERS, J . , Fast Reactor Fuel and Fuel Elements, GfK, CONF- 700920 (1970). [39] NATESAN, K., KASSNER, T .F., TMS Fall Meeting, AIME, Detroit, 1971, AED-CONF 430-010 (1971) ; LONGSON, B ., THORLEY, A. W ., Rep.TRG -1906 (1969). 268 GOTZMANN and OHSE [40] POWERS, R ., STRASSER, A ., Rep. UNC-5229 (1968). [41] MOUCHNINO, J . , Rep.CEA-R 3353 (1967). [42] STRASSER, A., private communication. [43] STUMPF, W., GÔTZMANN, О., Rep. KFK 1384 (1971). [44] DIETRICH, R., private communication. IAEA-PL-463/16 OUT-OF-PILE INVESTIGATIONS OF FISSION PRODUCT- CLADDING REACTIONS IN FAST REACTOR FUEL PINS* E.A. AITKEN, S.K. EVANS, B.F. RUBIN General Electric Company, Vallecitos Nuclear Center, Pleasanton, Calif., United States of America Abstract . . OUT-OF-PILE INVESTIGATIONS OF FISSION PRODUCT-CLADDING REACTIONS IN FAST REACTOR FUEL PINS. Type-316 stainless steel fuel pins containing fuel at a known stoichiometry and additives o f caesium oxide, tellurium, iodine and molybdenum were heat treated in a temperature gradient between 550 and 1100°C for 100 hours. The distribution o f each additive along the temperature profile was measured with the aid of tracer isotopes. The reactions of the fission products with the fuel and cladding were determined by metallographic examination. Caesium was found to react with iodine, tellurium and fuel. The caesium-fuel reaction was dependent on the stoichiometry o f the fúel. Tellurium reacted with the fuel in the presence or absence o f caesium. However, the distribution o f tellurium was affected by Cs-to-Te ratio. Intergranular attack of the stainless steel occurred between 750 and 950°C when tellurium was present and the fuel was stoichiometric. Caesium oxide caused intergranular attack when the oxygen potential was higher than.the potential characteristic o f stoichiometric fuel. Iodine and molybdenum additions with or without caesium did not induce intergranular attack. If the fuel was hypostoichiometric, none o f the additives produced inter granular attack. The implication o f these data on the nature and mechanism of cladding attack in irradiated mixed oxide fuels is discussed. I. IN T R O D U C T IO N Post-Irradiation examinations of numerous mixed oxide fuel pins have revealed regions where the internal cladding surfaces have reacted with fission products. These reactions appear as general surface attack or as intergranular attack. General corrosive attack has been observed at depths as great as 3 mils while intergranular attack has penetrated as much as 9 mils. [1] Examples of each are shown in Figure 1 . Electron microprobe examination has shown that cesium, molybdenum, and oxygen are generally present within the grain boundaries of the reacted cladding while tellurium is usually confined to the fuel-cladding interface. This attack can reduce the effective cladding thickness of an operating fuel element, particularly at the high cladding temperatures anticipated in commercial LM FB R s and may lessen restraint at high burnup. Accordingly, an out-of-pile test program has been instituted at General Electric Company to investigate, individually and in combination, the "volatile fission products participating in the reaction. Isothermal stress rupture tests were originally performed to screen the corrosively aggressive fission products. Small quantities of potential reactants l2 , H IO 3 , Csl, Cs, CsCI, CS2 O, and CsOH were loaded into individual capsules of Type 316 stainless steel and heat treated at 1200°F for time periods to 400 hours. Internal pressures ranged from atmospheric to 600 psi, including the vapor pressure of the volatile additive, at the test temperature, yielding effective hoop stresses of from 300 to 11,000 psi. The latter pressure would be expected to require approximately 10,000 hours to rupture the capsules. Failure occurred within 400 hours in the capsules loaded with CS2 O and CsOH. None of the other additives caused any significant dimensional change. This evidence of corrosive behavior of cesium oxide was the starting point of the base test program which utilized a thermal gradient heat treatment, mixed oxide fuel, and various volatile fission products. This test environment enabled the oxygen activity at the fuel-cladding interface Work performed under USAEC Contracts AT(04-3)-189, PA 53 and AT(04-3)-189, PA-10 269 5 0 0 X AS POLISHED SAMPLES FIG. 1. Examples of general and intergianular attack in irradiated mixed oxide fuel pins. IAEA-PL-463/16 271 THERMAL GRADIENT EXPERIMENTAL CAPSULES (0.25 x 5.5 INCHES LONG) FIG. 2. Temperature profile and capsule description for thermal gradient capsules. to be varied, as is the case in an operating element when oxygen migrates radially to the cool region of the fuel pellet or is released during fissioning. This method also simulates the migration of volatile fission products down a thermal gradient. II. EXPERIMENTAL PROCEDURE The pellets were coprecipitated (U—25 w t% Pu>2 ±x of 90 to 95% TD. Stoichiometry was controlled for the specific test by regulating the I\l2-6%(H2+H20) atmosphere of the furnace during the 1 B00°C sintering treatment. Control 0:M determinations before and following specific experiments were made using the thermogravametric method developed by Lyon. [2] Fission product additions were made to the hot end of the capsule and these consisted of cesium oxide, iodine, tellurium, and molybdenum; usually one additive was used, but sometimes two or more were combined in a single capsule. The cesium oxide was maintained dry; in several cases, radioactive Cs—137 was added to the cesia so that migration or interactions of the isotope could be followed with the gamma counter. Since the experimental aim was to examine the behavior of cesium in the oxygen environment established by the fuel, the cesia addition was modest. In most cases, these additions amounted to ~50 mg Cs, ~ 0 .1% of the total fuel weight of 37 grams. This was equivalent to a fuel burnup of about 2 atom percent and affected the overall fuel 0:M by less than 0.002 units. In a few experiments, 2, 4, or 10 times this amount was added; in the latter case, the cladding oxidized so drastically that the additive corroded through the capsule wall; The iodine was added as radioactive Pdl2-131. The tellurium addition was irradiated in the General Electric Test Reactor to produce a mixture of Te—128+129 isotopes that could be traced by the gamma scanner. Molybdenum was added as a mixture of the metal and molybdenum trioxide that was expected to be reduced to equimolar mixtures of molybdenum and molybdenum dioxide. Radioactive M o-99 was included in the mixture for tracing purposes. Each test capsule utilized about 20 pellets with a total height of about 5.5 inches. The.pellets and additives were loaded within an argon-filled glove box into 6 -inch lengths of Type 316 stainless steel tubing, all from the same heat and were then sealed by welding with end caps of the same alloy.* The tubular cross section *The capsules of Test Series 1 and 2 operated over the temperature range of 550 to 1500°C. Because the upper temperature exceeded the melting range of stainless steel, molybdenum tubing was used; a one-inch long stainless steel insert was added at the cold end so that its reactions with the fission product additives could be observed. 272 AITKEN et al. T A BLE 1. Summary of Thermal Gradient Capsule Experiments (Fueled capsules heated 100 h in thermal gradient. Additives marked with an asterisk contained tracer.) Temperature Series 0 :M Range, ° C Container Additive Results 1 1.967 660 - 1550 M o Cs2M o 0 4/U Cs20 increased 0:M redistribution rate Q ' = —7 kcal/mole 2 -A 1.967 700 - 1550 M o Cs2M o 0 4 /U Confirmed the results of Series 1. 2 - В 1.967 550 - 1550 M o CS2M 0O 4/U Diffusion-controlled reaction along 2 — С 1.967 660 - 1550 M o CS2M 0O 4/U cladding surface between 600 and 750°C. 3 - A 1.967 550 - 1150 Cs2M o 0 4/U Cs caused diffusion-controlled reaction 3 - В 1.967 5 5 0 - 1150 Cs2 0 noted above. Cs20 in either form reacted. 3 - С 1.967 5 5 0 - 1150 None Stainless capsules suitable for testing. 4 - A 1.94 550 - 1150 Cs20 ‘ Amount of Cs20 retained at high 4 - В 1.967 550 - 1150 Cs2 t r temperature end increased with 0:M . 4 - С 2.00“ 550 - 1150 Cs20 ' Diffusion-controlled reaction observed 4 - D 2.01 550 - 1150 Cs2 ° * again. Oxidation of cladding with 0:M ^2.00 4 - E 2.00 550 - 1150 CS2O Excess (500 mg) Excess Cs20 ruptured Capsule E. Severe intergranular attack. 5 - A 1.96 550 - 1150(a) Cs20 + Те|ь|" Те reduced migration of Cs20 toward low tempeiature end. Cs and Те appear to cohabit in fuel phase. 5 - B 1.96 5 5 0 - 1150(a) Cs20 * + 1|с|- No reaction of cladding with I and Cs added. 5 - C 2.00- 550- 1150(al Cs20 ' + 1|с | ' I content did not alter Cs20 distribution. 5 - D 2.00“ 550 - 1150(a) C î20 ' + Те*ь * ’ Cs and Т е appear to cohabit; Те lessens migration of Cs20 to cold end. Intergranular attack near 800°C. 5 - E 2.00- 550 - 1150(a) Cs20 ‘ Excess {100 mgl Excess of Cs20 produced drastic intergranular attack. 5 - F 2.00- 550- 1150(al Cs20 ' Excess (200 mg) Cs2U 0 4 and C s2U 207 identified at cold 1 end in E and F. 6 -A 2.00“ 550 - 11 50 Cs20 * + T e (cl Cs and Те cohabited. Excess Cs deposited 0 along length, migrating to cold end. but n inhibited by Те and less migrated than in D. .6 - 8 2.00- 550 - 1150 Cs20 * + T e*b) Cs and Те cohabited at about 725°C. Те > restricted Cs migration to cold end. Intergranular attack between 750 and 950°C as in 5 - D . 6 - C 2.00- 5 5 0 - 1150 Te Intergranular attack between 750 and 950°C. 6 — D 2.00- 550 - 1150 Cs20 * Cs deposited along length, more concentrated at hot end. 6 - E 2.00- 550 - 1150 Cs20 * + |lcl- Cs + 1 concentrated at cold end; excess Cs deposited along capsule length. £l U О + 6 -r F 2.00- 550 - 1150 Cs and 1 concentrated at cold end as Csl. 7 - A 2.00- 550 - 1150 Cs20* + Mo(b) + M o:0 3 Mo remained at hot end; Cs20 deposited along length. 7 - B 2.00- 550 - 11 50 Cs20* + M o^ + Мо*Оз 7 - C 2.00- 550 - 11 50 М о О ‘ з + Mo Mo remained at hot end - oxide film on upper temperature in half of cladding. 7 - D 2.00- 550 - 1150 МоО*з + Mo(d) Mo remained at cold end - oxide film on lower temperature in half of cladding. 7 - E 1.96 550 - 11 50 М о О * з + M o Mo remained at hot end — no oxide film. 7 - F 2.00“ 550 - 11 50 Cs20* (Reheat of No significant change in Cs20 distribution Capsule 6 — D) after additional 100 hours of heat treatment. 8 -A 1.97 5 5 0 - 1100 Т е* (12 mg) No intergranular attack. Те remained at high temperature end. 8 - B 2.00“ 5 5 0 - 1100 Cs20 ‘ + Те1Ы * 1 mil intergranular attack. 8 — C 2.00“ 5 5 0 - 1100 Cs20 ’ + Te (a) Temperature drifted gradually during the heat treatment due to deterioration of the heater windings; (b) 1:1 ratio of Cs to Te or Cs to I or Cs to Mo + M 0O 3; (c) 4:1 ratio of Cs to Te or Cs to I; (d) 2:1 ratio of Cs to Mo; (e) Additive at cold end. IAEA-PL-46 3/16 273 FIG. 3. Arrhenius plot o f the deviation from stoichiometry against reciprocal absolute temperature o f fuel pellets tested in a temperature gradient (Series 1 and 2). dimensions were 0.25 inch outer diameter and 0.015 inch wall. The capsules were heat treated for 100 hours in a molybdenum-wound furnace in which a thermal gradient of about 550°C was maintained over the 6 inch long capsules. The capsules usually operated with temperature extremes of 550 and 1100°C. The capsule and operating conditions are illustrated in Figure 2. A total of six capsules was positioned within the furnace in most experimental runs. The furnace temperatures along the gradient were monitored by a chromel-alumel thermocouple that was withdrawn in one-fourth inch increments to establish an axial profile of temperature. After heat treatment, the capsules were gamma counted at 1/8-inch intervals with a germanium-lithium detector coupled with a Nuclear Data Analyzer. In some cases the rods were investigated for intergranular attack by pulsed eddy current testing; slight indications were found and were used to guide sample selection in subsequent metallographic examination. The capsules were visually examined, dimensioned with a micrometer at one-inch intervals, then slit longitudinally so that the two halves could be separated and the pellets and cladding surfaces could be examined. Pellet examination consisted of viewing, 0:M measurement in a number of cases, and occasionally an x-ray diffraction analysis to identify a deposit or compound. The cladding examination consisted primarily of visual studies or metallographic analyses with one inch longitudinal segments mounted, then polished and ground with kerosene. Initial viewing was accomplished in the as-polished state, since intergranular penetration of cesium could be detected best in samples without etching. The samples were then electrolytically etched to reveal grain structure. At intervals, transverse sections of cladding were mounted for viewing the structure in the radial direction. Selected samples exhibiting intergranular attack were examined by electron microprobe to identify reacting species. A total of eight tests, generally consisting of six capsules per test have been treated and examined. A description of the tests, the additives, and major findings are presented in Table 1. During Test 5 the furnace heating element degraded and the temperature profile of Figure 2 did not prevail throughout the 100 hour test period. The temperature of the hot end fluctuated between 1100 and 1250°C during the test while the temperature at the low end dropped from 500 to 250°C. 274 AITKEN et al. CESIUM INTENSITY (cpm) INCHES FROM COLD END FIG. 4. Tracer distribution of caesium after heat treatment in a temperature gradient (Series 4) as a function o f О : M ratio. III. RESULTS The results described in this section will focus on the behavior of specific fission product element or groups of elements with emphasis on migration and interaction tendencies between other fission products, fuel, or cladding constituents. Other features such as behavior and migration of interstitial impurities (carbon and nitrogen) will be discussed at another time. A. EFFECTS OF CESIUM OXIDE ADDITIONS 1. 0:M Redistribution In the first two series listed in Table 1 the cesium compound was heated to 1500°C and allowed to decompose and vaporize to the cold end. At the cold end, noticeable buildup of cesium in the fuel-cladding gap and in the fuel was observed. On the other hand, the majority of the pellets above about 870°C appeared free of cesium contamination. Analysis of the 0:M of these pellets showed that oxygen redistribution had occurred. The data shown in Figure 3 as an Arrhenius plot gives an apparent heat of transport of about -7 kcal/mole. Oxygen redistribution under identical conditions in the absence of cesium does not occur as rapidly and gives a slightly different heat of transport. [3] It appears therefore, that cesium influences the oxygen redistribtuion mechanism. The pellets below 870° showed evidence of cesium in the grain boundaries and absorbed moisture on storage in air. This result suggests the existence of a cesium-fuel compound analogous to fuel-sodium compounds. [4] The threshold for formation of this compound from Figure 3 appears to occur at an 0:M of about 1.985. 2. Effect of 0:M on Cesium Migration In Series 4 the effect of initial 0:M on the migration of cesium in the temperature range of 550 to 1100°C was measured. Radioactive cesium -137 was included with the stable cesium oxide additive in order to trace the cesium movement down the thermal gradient. The capsules were prepared with fuel at 0:M ratios of 1.94,1.965, 2.00, and 2.01. The results of the gamma counting illustrated in Figure 4 show that vvith stoichiometric and hyperstoichiometric fuel the bulk of the cesium remained at the hot end of the capsule. More cesium migrated to the cold end in the capsule containing fuel with the 0:M ratio of 1.965 and virtually all of the cesium migrated to LAEA-PL-463/16 275 the cold end when the fuel had a ratio of 1.94. This type of cesium distribution is believed to result from a cesium-fuel compound which as noted previously forms at an 0:M greater than 1.985. All of the cesium oxide combined with the hottest pellet if 0:M was greater than 1.985. Only small amounts were free to migrate to the cold end. When the fuel had an 0:M of 1.965 the oxygen associated with the cesium oxide was sufficient to raise the hottest pellet to an 0:M above 1.985 but this dissociation allowed only part of the cesium to be combined with the fuel at the high temperature end; the remainder of the cesium migrated as the metal vapor to the cooler region. In the capsule with the lowest stoichiometry, 1.94, the oxygen associated with the cesium oxide was insufficient to raise the 0:M of the hottest pellet to 1.985 so that all of the cesium was free to vaporize as metal to the cold end. This behavior can also account for the recent A N L observation that more fission product Cs migrates to the colder blanket region in irradiated pins as the average 0:M of the fuel is decreased. [5] 3. Cesium-Fuel Compound Formation The compounds which form by the interaction of cesium oxide with fuel were not determined in this study since the quantity of additive was usually too small. By analogy with sodium-fuel reaction compounds, however, it is likely that either C S M O 3 or CS3 MO 4 would be the reaction product in the 0:M region between 1.985 and about 2.00. In capsules where excess amounts of cesium oxide were present (Capsules 4E, 5E, and 5F) and the overall oxygen environment was characteristic of hyperstoichiometric fuel, evidence of preferential vaporization of uranium oxide' was detected. Capsule 4E showed a deposit on cladding part way down the temperature gradient that consisted of uranium oxide but no plutonium upon examination with the microprobe. This capsule failed so that most of the cesium was lost sometime during the test. In addition, capsules 5E and F, which did not fail, exhibited a yellow mass of crystals between the pellets at the cold end. These crystals had a crystal structure of CS2 UO 4 and CS2 U 2 O 7 by x-ray diffraction. It is surmised that in hyperstoichiometric fuel with cesium, part of the vapor is cesium uranate which leads to preferential volatilization of U from the fuel at high temperatures and deposition at low temperatures. In capsules containing hypostoichiometric fuel and cesium, where part or all of the cesium migrated to the cold end, the deposit on the capsule walls was usually a brownish or brownish yellow compound of cesium containing an unknown amount of oxygen. This material would absorb moisture readily on exposure to air. 4. Reactions with Cladding As Figure 4 showed, when cesium oxide was added to stoichiometric fuel (capsule 4C) only small amounts of the additive migrated to the cool end where it could react with the cladding. No intergranu/ar reactions were observed. When cesium oxide was added to the cold end rather than to the hot end of the capsule (capsule not- listed in Table 1) containing stoichiometric fuel, intergranular attack was observed. A photomicrograph of this type of attack is seen on the left side of Figure 5. It is believed that the cesium oxide at the cold end did not react with the fuel at the low temperature (550°C) as it would at high temperatures (1100°C) so that locally high oxygen activity prevailed causing intergranular attack by the cesium. Intergranular and cladding oxidation was observed in the hot region of the capsules loaded with stoichiometric fuel and excess amounts of oxygen (capsules 4E, 5E, and 5F). Copious quantities of a green chromium oxide film were formed on the surface. Excess quantities of CS2 O (>50 mg) probably increased the oxygen activity noticeably so that the threshold for intergranular attack was exceeded. For stainless steel, oxidation occurs at A G 0 2 = -120 kcal/mole at 800°C. Stoichiometric fuel made by sintering in wet hydrogen has an oxygen potential, AG Û 2 a -95 kcal/mole. Maiya and Busch [6 ] have reported that the intergranular attack by cesium-oxygen occurs at A G 0 2 = -8 4 kcal/mole. Therefore, the threshold for attack by cesium-oxygen appears to occur at a higher oxygen potential than that for oxidation of stainless steel. B. EFFECTS OF TELLURIUM ADDITIONS 1. M igration Behavior in Fuel Gamma scans of capsules containing tellurium tracer (capsules 6 C, 8 A , and 8 D) added to the fuel with 0:M values of 1.97 and 2.00 are presented in Figure 6 . These tracers show that tellurium combines with the fuel in both oxygen environments and that very little migrates toward the cold end. Visual examination of the fuel does not reveal any prominent fuel-tellurium compound at the hot end, however, UOTe is known to form by direct combination of uranium oxide and tellurium. [7] I NT ERG RAN U LAR ATTACK IN STAINLESS STEEL * * '9:.-s * ■ * ■ . sîi • ■ ■ м и Ш И М м Cs Те Cs — Те FIG. 5. Examples o f intergranular attack on Type 316 stainless steel by caesium oxide, tellurium and caesium oxide-tellurium mixtures. IAEA-PL-46 3/16 277 Inches fro m Cold End FIG. 6 . Tracer distribution o f tellurium after heat treatment in a temperature gradient (Series 8) at two О : M values. 2. Effect on Cladding Intergranular attack of the cladding to a depth of about 1 mil was observed over the temperature range of 750 to 950°C in the samples loaded with stoichiometric fuel. This is shown in the middle section of Figure 5. Tellurium additions of 12 milligrams or 50 milligrams did not result in any differences in the degree of penetration but additions in the presence o f hypostoichiometric fuel (0:M = 1.97) eliminated intergranular attack; thus, there appears to be a critical level of oxygen activity required for the attack to occur. C. EFFECTS OF CESIUM AND TELLURIUM ADDITIONS 1. Migration Behavior in Fuel Capsule 5D, which contained stoichiometric fuel with additions of cesium oxide and tellruium at a Cs:Te= 1, gave the tracer distribution shown in Figure 7. The tw o elements showed a common distribution pattern and for the most part were confined to the hotter half of the capsule. A gamma scan of capsule 5A containing the same additives with hypostoichiometric fuel, had a similar pattern as in Figure 7 except the peaks of cesium and tellurium concentration extended about 1/2-irich further toward the cold end. As a comparison, 278 AITKEN et al. BEHAVIOR WITH STOICHIOMETRIC FUEL Cs INTENSITY (cpm) FIG. 7. Tracer distribution o f caesium alone and caesium with tellurium after heat treatment in a temperature gradient. (Series 5 and 6). when cesium alone was included in the capsule with stoichiometric fuel (capsule 6 B), the majority was retained at the hot end hut a significant fraction of the element was present at the cold end. The identical distribution pattern of tellurium and cesium suggests that there is a strong association between the two when dissolved in the fuel. Color changes in the fuel pellets in the hot end were observed after exposure to air, however x-ray diffraction lines other than the fuel could not be identified with known compounds. 2. The Effect of Cesium:Tellurium Ratio The tracer distribution of capsules containing stoichiometric fuel with cesium and tellurium at three atom ratios is shown in Figure 8 . In all cases, the majority of the cesium and tellurium remained at the hot end, although the relative peak heights changed systematically from Cs:Te = 1 to 4. At Cs:Te = 4 slightly more cesium migrated to the cold end. The peak at the extreme hot end is located at a position similar to the peaks obtained in capsules containing tellurium alone (Figure 6 ). If this peak is associated with the excess tellurium, then its tendency to diminish with increasing cesium content suggests that the cesiunv.tellurium ratio may be a key factor controlling the migration and activity of tellurium. The cesium to tellurium ratio from fission is about 4, however migration of cesium axially to the UO 2 blanket fuel in fast reactor fuel pins has been reported. [5] This axial migration was more extensive as the initial 0 : M of the fuel was decreased. This type of behavior suggests that the Cs:Te ratio may differ locally from the fission ratio. If the law of mass action applies, then a decreasing local concentration of cesium would increase the local tellurium activity and hence, intergranular attack. 3. Effect on Cladding As with capsules containing tellurium alone, intergranular attack occurred at a Cs:Te = 1 and 2 as long as the fuel was stoichiometric. At Cs:Te = 4 one test showed intergranular attack but the other did not. This lack of reproducibility may indicate that the threshold for intergranular attack exists at the Cs:Te about 4. A summary of the degree of intergranular reactions in the cladding by cesium and tellurium is shown in Table 2. Although the degree of attack increased from 0 at 750°C to 1 or 2 mils at 950°C, the rate of attack did not appear to be sensitive to a particular Cs:Te ratio from the results obtained so far. An illustration of the intergranular attack observed in capsule 5D where the Cs:Te = 1 is shown at the right hand side of Figure 5. The region of intergranular attack was associatecf visually with a black band which formed on the inside surface of the capsule in the temperature region of 750 and 950°C. No intergranular reaction was t three at FIG. FIG. 8 . Cs or Te Intensity (counts/min) asu-otluim ratios. caesium-to-tellurium Tracer distribution of caesium with tellurium after heat treatment in a temperature gradient (Series (Series gradient temperature a in treatment heat after tellurium with caesium of distribution Tracer IAEA-PL-463/16 279 8 ) 2 8 0 AITKEN et al. T A BLE 2. Cesium—Tellurium Behavior in Type 316 Cladding Thermal Gradient Capsules C s--Те Capsule Fuel 0 : M A to m Ratio W eight, mg Results 8 A 1.97 0:1 0 :1 2 No intergranular attack 6C 2.00 0:1 0 :1 2 1 mil intergranular attack 8 D 2 .00 0:1 0 :5 0 1 mil intergranular attack 5 A 1.97 1:1 5 0 :4 8 No intergranular attack 6 B 2.00 1:1 5 0 :4 8 2 mils intergranular attack 5 D 2.00 1:1 5 0 :4 8 2 mils intergranular attack 8 B 2.00 1:1 1 5:12 1 mil intergranular attack 8 E 2 .00 2:1 3 0 :1 5 1 mil intergranular attack 6 A 2.0 0 4:1 5 0 :1 5 No intergranular attack 8C 2 .00 4:1 5 0 :1 5 1 mil intergranular attack seen at higher temperatures. No reaction was detected at lower temperatures in the 100-hour heat treatment. The region of imergranular attack was associated with a globular intergranular precipitate which was shown to be chromium rich by electron microprobe examination. These precipitates which may be Cr 2 3 Cg appear to be preferentially dissolved by the action of cesium and tellurium at the appropriate oxygen activity. A microprobe scan of the chromium, cesium, and tellurium and oxygen content of the intergranular reaction zone of capsule 50 is shown on the left side of Figure 9. As a comparison, a similar reaction zone from the intergranular reacted region of a fuel pin irradiated E B R -2 (F3B3) is shown on the right hand side of Figure 9. In both cases the cesium and oxygen penetrated intergranularly whereas tellurium remained close to or at the cladding surface. It is not understood why tellurium did not penetrate with cesium and oxygen into the intergranular regions. Tellurium nevertheless appears to be an important constituent in triggering attack in the presence of stoichiometric fuel. D. EFFECT OF MOLYBDENUM AND MOLYBDENUM OXIDE ADDITIONS 1. Migration Behavior with Fuel Capsule 7C and D in Table 1 contained additions of molybdenum and molybdenum trioxide in such a proportion that equal molar mixtures of Mo and M 0 O 2 would exist after equilibration. Tracer distribution in these capsules showed no migration of molybdenum in both stoichiometric and hypostoichiometric fuel. In capsule 7D, where the molybdenum and molybdenum oxide had been added to the cold end, no migration of the M o -9 9 tracer was noted either even though the tracer was present in the M 0 O 3 additive. 2. Effect on Cladding No intergranular attack was observed with capsules containing molybdenum and molybdenum trioxide. A uniform green oxide film was formed along the capsule inner surface in capsule 7C and D to a depth of about 2.5 micrometers in thickness. This oxidation behavior is in agreement with thermodynamic expectations. The oxygen activity for the Mo and M 0 O 2 reaction is higher than for oxidation with stainless steel so that a chromium oxide-rich scale should form in the presence of stoichiometric fuel. In the presence of hypostoichiometric fuel, however, the oxygen in the Mo and M 0 O 2 should be absorbed by the fuel instead of the cladding and no oxide scale was found in this case (capsule 7E). FIG. 9. Microprobe scans of intergranularly attacked region of out-of-pile thermal gradient capsule containing caesium and tellurium and o f cladding from an irradiated fuel pin. CO CO 282 AITKEN et al. EFFECTS OF CESIUM/IODINE RATIO 5000 4000 3000 ÍNTENSITY (cpm) 2000 1000 0 01 234560123456 INCHES FROM COLD END FIG. 10. Tracer distribution o f caesium and iodine after heat treatment in a temperature gradient (Series 5 and 6) at two caesium-to-iodine ratios. E. EFFECT OF CESIUM OXIDE WITH MOLYBDENUM AND MOLYBDENUM OXIDE ADDITIONS 1. Migration Behavior in Fuel The addition of cesium oxide did not affect the migration behavior of molybdenum. The molybdenum tracer peaks.remained at the hot end in all cases. The cesium tracer distribution was essentially the same as if cesium alone was present; that is, most of the cesium was at the hot end if the fuel was stoichiometric and at the cold end if it was hypostoichiometric. There was no indication that a cesium-molybdenum-oxygen compound was present from the tracer results although no direct identification was attempted. 2. Effect on Cladding No intergranular attack was observed in capsules containing cesium oxide, with a Mo and M 0 O 3 mixture. The reaction behavior was similar to capsules containing cesium oxide alone or Mo and M 0 O3 alone when placed at the hot end where equilibration with the fuel could occur. In the presence of cesium oxide, a green oxide was seen visually in capsule 7A but diminished noticeably in capsule 7B where the Mo and M 0 O 3 content was reduced to one half the amount used in capsule 7 A. In both cases the amount of cesium oxide was unchanged. The oxide scale was too thin however to be detected after polishing. This behavior suggests that most of the oxygen associated with the M 0 O3 oxidized the cladding and little or none combined with the fuel or the cesium oxide. F. EFFECT OF CESIUM OXIDE AND IODINE ADDITIONS 1. Migration Behavior In capsule 6 F, which contained stoichiometric fuel, the cesium oxide and palladium iodide additions'were made at a Cs:I = 1. The tracer distribution (Figure 10) after heat treatment showed that all of the cesium and iodine were at the cold end as if the reaction product was cesium iodide. In capsules 6 E and 5C, where Cs:I = 4, only part of the cesium and essentially all of the iodine moved to the cold end. In capsule 5B, with hypostoichiometric fuel and Cs:I = 4, essentially all of the iodine moved to the cold end along with most but not all of the cesium. IAEA-PL-463/16 283 BEHAVIOR OF Cs20, Mo, Te, AND l2 FIG. 11. Temperature profile and tracer distribution o f Cs, Mo, Te and I after heat treatment in a temperature gradient (Series 8). These observations indicate that iodine combined stoichiometrically with cesium as cesium iodide which migrated to the cold end. In contrast to the cesium and tellurium interaction, cesium iodide did not appear to associate with the fuel. Cesium in excess of that needed to form cesium iodide behaved as if cesium oxide alone were added to the capsule. 2. Effect on Cladding No intergranular attack was observed when cesium and iodine was added to the capsules containing stoichiometric fuel, regardless of the Cs:I ratio. G. EFFECT OF CESIUM WITH MOLYBDENUM, IODINE, AND TELLURIUM AT FISSION YIELD RATIOS 1. M igration Behavior In capsule 8 F containing stoichiometric fuel, additions of cesium oxide, Mo and M 0 O 3 , P d l2 , and Te were made in the approximate ratio of their fission yields. The combined tracer distribution of these fission products is shown in Figure 11 along with the temperature profile. The relative distribution of the cesium appears to follow qualitatively the combined results obtained from the paired combinations of cesium-tellurium, cesium-iodine, and cesium-fuel compounds. The molybdenum tracer remained at the high temperature end and all of the iodine moved to the cold end. The small cesium and tellurium peaks in the region 2 to 3 inches from the cold end appear to be strongly coupled. 2. Effect on Cladding Intergranular attack up to a depth of about 1 mil was observed in the region 750 to 950°C identical to that seen with other capsules in Series 8 which contained stoichiometric fuel and tellurium. It appears that the combined heat, treatment of these volatile fission product additives give the same results as with cesium-tellurium additives alone. It appears that cesium-tellurium is uniquely responsible for this type of intergranular attack. 2 8 4 AITKEN et al. OXIDATION REACTIONS OF (U-25 wt % Pu) 0 2 ± x AT 800° c 160 150 140 130 120 _ A G (02) 110 kcal/mole 10Q 90 80 70 60 50 40 OXYGEN/METAL RATIO FIG. 12. Oxygen potential plotted against О : M o f mixed oxide fuel at 800°C relative to oxidation couples for various elements. SUMMARY AND DISCUSSION The volatile fission products, cesium, tellurium, iodine, and molybdenum, when heated in a temperature gradient in sealed Type 316 stainless steel capsules containing mixed oxide fuel at various 0:M levels, show marked interactions between the fuel phase and cesium and tellurium. Iodine and molybdenum did not show any evidence of reaction with the fuel phase. Iodine appears to combine with cesium and so does tellurium, except that the Cs-Т е compound is strongly associated with the fuel phase. For hypostoichiometric compositions, cesium has less tendency to combine with the fuel. In particular, no compound appears to form at an 0:M ratio below 1.985. At the stoichiometric composition, cesium appears to combine with the other constituents in the following order: I > Te > Fuel > Mo. With hypostoichiometric fuel, no oxidation or intergranular attack is observed in keeping with its thermodynamic stability with respect to oxidation of the chromium in the cladding. With stoichiometric fuel, intergranular attack is observed between 750 and 950°C when tellurium is present with or without cesium. Oxidation is also observed when cesium oxide and molybdenum oxide or both are present in sufficient quantities to raise the oxygen activity above that of the stoichiometric fuel. The oxidative attack by cesium oxide is intergranular but its aggressiveness depends on the oxygen activity. The distinctive features between cladding attacked oxygen activity are illustrated in Figure 12. In this curve the relationship between the oxygen potential, A G O 2 and 0:M is shown at 800°C as determined by Markin and Mclver. [8 ] Superimposed on this curve are values for several oxidation couples at 800°C. The exact location of these couples depends on the cladding and fuel temperatures since the fuel system is not isothermal near the fuel-cladding gap. The value for chromium oxidation in stainless steel is estimated from its activity in the stainless steel. The approximate oxygen potential for UO 2 .0 0 as defined in this paper is shown also at approximately -9 5 kcal/mole, characteristic of sintering in wet hydrogen atmospheres. It is seen that the oxidation of the chromium in stainless steel should not occur for A G Û 2 < -125 to -130 kcal/mole. This value is the oxidation 1 threshold regardless of whether the attack is a uniform oxidation or intergranular oxidation. The results in this paper support this threshold for oxidation and intergranular attack IAEA-PL-46 3/16 285 since hypostoichiometric fuel, (0:M = 1.97, A G 02 = — 150_kcal/mole) where no oxidation or intergranular attack is observed, is above the threshold. Stoichiometric fuel (AGÛ 2 = -9 5 kcal/mole) where intergranular attack with tellurium has been observed is below this threshold. Preliminary experiments [9] with tellurium and cesium heated isothermally at 750°C in stainless steel capsules with the chromium and chromium oxide buffer did not produce intergranular attack whereas without the buffer (presumably higher oxygen activity) intergranular attack occurred. If these experiments are correct, then the experimentally derived oxidation threshold is brought closer to the expected theoretical value for oxidation of chromium in stainless steel. _ In experiments with cesium alone, no intergranular attack was observed with stoichiometric fuel, i.e., (AGO 2 = -9 5 kcal/mole). Maiya [6] has reported that oxygen potential in cesium and cesium oxide has to be greater than -8 4 kcal/mole to induce intergranular attack. This interesting result suggests that intergranular attack may occur by the presence of cesium or tellurium but tellurium appears to be the more sensitive as far as the oxygen potential is concerned even though both induce attack at essentially a stoichiometric composition. Since the oxygen to metal ratio increases during burnup in most fast reactor fuel irradiations, then a hypostoichiometric fuel would encounter intergranular attack threshold caused by tellurium first if a sufficient amount of tellurium is present during the burnup. On the other hand, if the fuel were hyperstoichiometric, intergranular attack could be initiated as soon as either the cesium or the tellurium were sufficiently abundant to be aggressive. If oxidation of the cladding by hyperstoichiometric fuel occurs rapidly early in life before the fission products build up, then the oxide scale may have a lessening effect on the nature of intergranular attack also. These finer points of the mechanism need further study and analysis before identification of intergranular attack mechanism can be delineated. REFERENCES [1] M c C a r t h y , w . H., PERRY, k . J„ BENNETT, W. j ., and HULL, G. R., ANS Topical Meeting Reactor Materials Performance, Hanford, Washington, April 23—26,1972 (to be published). [2] LYON,W. L., GEAP-4271, May 1963. [3] AITKEN, E. A., ADAMSON, M. G. EVANS, S. K„ and LUDLOW, T. E., GEAP-12210, April 1971. [4] BARTRAM, S. F., GEMP-733, April 1970. [5] ANL-7900, Reactor Development Program Progress Report (December 1971) 5.19. [6] ANL-7833, Reactor Development Program Progress Report (June 1971) 5.11. [7] HANEVELD, A. J. KLEIN, JELLINEK, F., J. Inorg. Chem. 26 (1964)1127. [8] M ARKIN , T. L., M clVER, E. J., Proceedings of the Third International Conference on Plutonium, London, A. E. Kay and M. B. Weldron (ed.) Barnes and Noble, Inc., New York (1967) 845-57. [9] AITKEN, E. A., unpublished results. IAEA-PL-463/17 REACTION A L 'IN T E R F A C E GAINE (ACIER INOXYDABLE)/COMBUSTIBLE (OXYDE MIXTE. (U,Pu)02±x) DANS LES ELEMENTS COMBUSTIBLES IRRADIES EN NEUTRONS RAPIDES D. CALAIS, M, CONTE, F. de KEROULAS, R. LE BEUZE CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Roses, France Abstract-Résumé REACTIONS AT THE CLAD (STAINLESS STEEL)/FUEL (MIXED OXIDE (U,Pu)02 ) INTERFACE IN FUEL ELEMENTS IRRADIATED W ITH FAST NEUTRONS. The problems of compatibility between cladding (stainless steel) and fuel (mixed oxide, (U ,Pu)02 ± x) that arise in fast reactors are associated largely with the development of a clearly defined reaction zone at relatively high burn-ups (50 000 MWd/t). The evolution of this reaction zone can shorten the lifetime o f a fuel element. The authors concentrate on the purely chemical aspects o f the ”oxide-clad" reaction and suggest guide lines for out-of-pile simulations designed to provide a good interpretation o f the mechanisms involved. The analysis bears exclusively on irradiations with fast neutrons. No influence o f neutron energy on the development o f the oxide-clad reaction is evident. However, micrographie observations indicate that, other things being equal, the thickness of the reaction zone is considerably greater when irradiation is with fast neutrons than otherwise. In laying down these guide lines for simulation experiments the authors made use o f micrographie examinations and X -ray microanalyses carried out on the reaction zones observed at the fuel- clad interface. REACTION A L'INTERFACE GAINE (ACIER INOXYDABLE)/COMBUSTIBLE (OXYDE MIXTE (U,Pu)02 ± ) DANS LES ELEMENTS COMBUSTIBLES IRRADIES EN NEUTRONS RAPIDES. Dans les réacteurs rapides, les problèmes de compatibilité entre gaine (acier inoxydable) et combustible (oxyde mixte (U,Pu)0 2 ± x ) sont liés essentiellement à l ’apparition d’une zone de réaction très bien mise en évidence pour des taux d'irradiation relativement élevés (50 000 M W j/t). Le développement de cette zone de réaction peut nuire à la durée de vie d'un élément combustible. Les auteurs exposent l'aspect purement chimique permettant de comprendre le développement de la réaction «o x y d e -g a in e » et de suggérer un schéma pour orienter des expériences de simulation hors pile réalisées afin d'obtenir une bonne interprétation des mécanismes mis en jeu. L’analyse concerne exclusivement des irradiations en neutrons rapides. L’influence de l ’énergie des neutrons sur l'évolution de la réaction oxyde- gaine n'est pas évidente. Cependent des observations micrographiques montrent que, toutes choses égales par ailleurs, l'épaisseur de la zone de réaction est nettement plus importante dans le cas d’une irradiation en neutrons rapides. Pour l'orientation des expériences de simulation les auteurs se sont inspirés des examens micrographiques et des microanalyses X effectuées sur les zones de réaction à l ’interface gaine-combustible observées sur les éléments combustibles irradiés. I * INTRODUCTION Dans les réacteurs rapides, les problèmes de com patibilité entre gaine (acier inoxydable) et combustible (oxyde mixte (UjPuÎO^ ± ) sont liés essentiellem ent à l ’apparition d’une zone de réaction très bien mise en évidence pour des taux d’irradiation relativem ent élevés (50 000 MWj/t) (fig .l). Le développement de cette zone de réaction peut nuire à la durée de vie d*un élément combustible. 287 2 8 8 CALAIS et al. ACIER f ^30ß COMBUSTIBLE ' ^ 35/i FI G. 1. Aspect micrographique de la zon e de réa ction observée à l'interface gaine-combustible. La gaine est en acier inoxydable X 18 M. Le combustible est un oxyde mixte à 20% de PuOz . Le taux de combustion est voisin de 50 000 MWj/t. LAEA-PL-463/17 289 Les paramètres susceptibles de jouer un rôle dans l ’évolution de cette réaction sont très nombreux : température intérieure de la gaine, température de surface du combustible, nuance de l'acier u tilisé, écart à la stoechiom étrie de l'oxyde mixte ( U , P u )02 * de départ, taux et durée d'irradiations, puissance linéaire, nature et proportions des différents produits de fission (ceci est lié au rapport Pu239/|j235 dans le com bustible), densité du combustible, énergie des neutrons (que l ’on reliera à une accélé ration possible de la diffusion sous irradiation). Aussi les recoupements entre les m ultiples expériences relatives à cette réaction et décrites dans la bibliographie sont-ils d ifficiles à faire. Notre but est d’exposer ic i l'aspect purement chimique permettant de comprendre le développement de la réaction "oxyde-gaine” et de suggérer un schéma pour orienter des expériences de sim ulation hors pile réalisées afin d’obtenir une bonne interprétation des mécanismes mis en jeu. Notre analyse concerne exclusivement des irradiations en neutrons rapides. L'influence de l ’énergie des neutrons sur l'évolution de la réaction oxyde-gaine n'est pas évidente. Cependant des observations micrographiques montrent que, toutes choses égales par ailleurs, l'épaisseur de la zone de réaction e s t n e t t e m e n t plus importante dans le cas d'une irradiation en neutrons r a p i d e s . Les hypothèses sont la plupart du temps très sim plificatrices. Mais, vue la complexité du phénomène à étudier, il nous est apparu nécessaire d'avoir un fil directeur. Nous nous sommes bien entendu inspirés pour l'o rientation des expériences de sim ulation des examens micrographiques et des microanalyses X effectuées sur les zones de réaction à l'interface gaine-com bustible 0bservé3s sur les éléments combustibles irrad iés. I I - EXAMEN METALLOGRAPHIQUE DES INTERACTIONS GAINE-COMBUSTIBLES SE PRODUISANT DANS LES ELEMENTS COMBUSTIBLES DES REACTEURS RAPIDES. Les micrographies faites sur des sections d'éléments combustibles fortement irradiés mettent en évidence une zone de réaction à l'interface oxyde-acier. La fig. 1 représente la mise en évidence la plus nette de la réaction oxyde-gaine correspondant à un taux d 'irradiation de 50 000 MWj/t (neutrons rapides). On distingue entre l'a cie r et l'oxyde mixte 2 zones bien stratifiées (a et b) côté acier et une 3ème zone grise foncée à la périphérie du combustible et qui se prolonge dans les joints de grainsde l ’oxyde (zone c). L'examen m icrographique de nombreux échantillons permet d'affirm er que la zone c est la première à se former en fonction du temps d’irradiation et du taux de combustion. , 280 CALAIS et al. I I I - tlICROANALYSE X DE LA ZONE DE REACTION □es microanalyses X ont été faites à l'aid e d'une micro sonde САИЕСА MS 35 sur des échantillons amincis afin de réduire leur activité. Celle-ci est de l'ordre de 200 à 300 m illirad /heure à 30 cm. Ces échantillons sont décontaminés dans des cellules étanches et puis introduits dans le m icroanalyseur sans précautions biologiques particulières. Un écran de plomb placé entre l ’échan tillo n et le détecteur de rayonnement permet d'ebaisser le bruit de fond au dessous de 2 0 0 0 0 coups par minute, dans les cas les plus défavorables, par exemple, u tilisation des raies d’émission X ULa. et Pu La. 1 1 Dans la zone a on retrouve les constituants de l'acier inoxydable ; Fe, Ni, Crj des produits de fission : tellure, césium, baryum, palladium, molybdène; de l'uranium et du plutonium en faible q u a n t i t é . La zone b contient : de l'uranium et du plutonium, du fer, du chrome et du nickel et du césium. Dans la zone c on trouve : de l'uranium et du plutonium en concentrations plus faibles que celles correspondant à l ’oxyde mixte de départ, du césium et du chrome. Les teneurs en U et Pu croissent de a vers c, c'est l'inverse en ce qui concerne les constituants de l'acier. Dans le combustible on trouve des précipités riches en chrome et en fer sur une profondeur de 1 0 0 ,um au delà de la zone c. Nos méthodes de préparation et d'analyse ne nous permettent pas actuellement de déceler la présence éventuelle d'iode dans la zone de réaction. IV - PRINCIPES DIRECTEURS DE NOTRE ANALYSE Pour .donner une vue d'ensemble de nos idées, nous dirons tout de suite que,très grossièrement : - la zone a correspond à une déchromisation de l'acier due à l'action des produits de fission vo latils ou de leurs combinaisons (Te,ICs, I 2f1o) (fig. 2). - la zone b provient de la réaction directe du combustible tU^uJO^ ± ( x = 0 en périphérie) sur l'acier inoxydable "déchrcmisé" [zone a) x conduisant à la formation de ferrite (Cr Feî^O^. - la zone c est constituée d'un mélange d’urano-plutonate de césium, (provenant de l ’action directe du Cs sur (U,Pu)0 t v o i s i n de la stoechiom étrie), et de chromate de césium Tdû a la réaction fe rrite + urano-plutonate de césiun! et d'oxyde (U,Pu)Ü 2 t Des mesures de tension de vapeur du césium effectuées sur des combustibles irradiés confirment la présence d'urano-plutonate de césium à la périphérie du combustible. Ce schéma réactionnel s'appuie sur les résultats obtenus par m icroanalyse X de la zone de réaction observée sur les combustibles irradiés, des mesures de tension de vapeur effectuées sur ces derniers et sur des considérations thermodynamiques et des expériences de sim ulation que nous allons exposer maintenant. IAEA-PL-463/17 29 1 FIG. 2. Enregistrement radial de la teneur en chrome effectué par microanalyse X sur l'échantillon correspondant à la figure 1. On note la déchromisation de la zone a. V - DONNEES THERMODYNAMIQUES ET CINETIQUES A première vue la réaction observée à l'in terface gaine- combustible pourrait être due à l'action directe de l ’oxyde sur l'acier. L'énergie libre des oxydes mixtes (U,Pu)0 t non irradiés peut être considérée comme connue en fonction de 1 écart à la stoe- chiom étrie. L’acier inoxydable est susceptible de réduire les oxydes mixtes surstoechiom étriques. Ceci est fondé sur des considérations thermodynamiques et provient essentiellem ent du fa it que (Cr Fe^Og est plus stable que (U,Pu) 0 2 . Cependant ”1’inoxydabilité” de l'a cie r est due non pas à des raisons thermodynamiques mais à des raisons cinétiques : la formation d’une barrière anti-diffusion de ferrite (Fe CrJ^O^ empêche l'oxydation de l'a cie r. A cet effet des expériences de com patibilité directe ont été effectuées sur des couples de diffusion (L^PuJC^ t x _ a c i e r X 18 11 : x est maintenu constant par contrôle de l'atm osphère du recuit. Les traitem ents sont effectués en présence de témoins (oxyde et acier) sans contact entre eux. L'acier est préalablement recuit à 70D°C pour augmenter sa réactivité (précipitation des carbures de chrome). Entre 700°C et 90Q°C l ’épaisseur de la zone de réaction est insignifiante, 2 9 2 CALAIS et al. équivalente à celle observée sur le témoin a'acier et ceci pour X = + 0,1. Cet écart à la stoechiom étrie n'est pas atteint en pile. Il nous faut donc chercher pour expliquer l'apparition d'une zone de réaction dans les éléments combustibles irradiés d’autres raisons qu’une simple réaction directe entre oxyde et acier. V I - FACTEURS SUSCEPTIBLES DE DEVELOPPER LA "REACTION OXYuE-GAINE" EN P I L E Pour observer une réaction (U,Pu)0„ . -acier dans les combus- 2 1 X tibies irradiés on peut alors penser que plusieurs conditions sont nécessaires. a) L’oxyde mixte irradié à l ’interface gaine-combustible doit avoir un écart à la stoechiom étrie qui lui permette, d’un point de vue thermodynamique,de réagir avec l ’acier. Nous ne connaissons pas la valeur de cet écart à la stoechiom étrie (la présence de produits de fission en solution dans l ’oxyde m odifie l ’énergie libre de celu i-ci), mais nous pensons qu’elle est voisine de zéro. Les combustibles bruts de fabrication sont sous-stoechiométriques et par conséquent ne devraient, pas réagir avec l ’acier inoxydable. Cependant la "fission est oxydante”. Le rapport oxygène/métal en solution croît avec le taux de combustion. Il en résulte, qu’au bout d’un certain temps de fonctionnement du réacteur, x peut atteindre la valeur critique permettant à la réaction de se produire. En outre, dans un gradient thermique en assiste à une redistribution de l ’oxygène te lle que.même pour un oxyde mixte globalement sous-- stoechiom étrique,le rapport 0 est plus élevé à basse température (c’est-à-dire au voisinage de la^gaine) qu’à haute température. b) La diffusion peut être accélérée dans (Fe СгЭ^СЦ sous irradiation (neutrons rapides) et ainsi fa cilite r l ’extension de la couche de ferrite qui cesse d’être une barrière anti-diffusion. c) Il est bien connu qu’une déchromisation de l ’acier peut - diminuer sa résistance à l ’oxydation. L’action de certains produits de fission pourrait entraîner un appauvrissement en chrome de la gaine au voisinage du combustible ce qui fa cilite ra it la réaction (la teneur nominale en chrome de 1‘acier X 18 PI u tilisé est de 17 %). Compte tenu de nos résultats obtenus par m icroanalyse X de la zone de réaction observée sur les éléments combustibles irradiés, nous allons chercher à définir le rôle de certains produits de fission ou de leur combinaison. V II - ROLE DU TELLURE Le principe de nos expériences est de recuire du tellure dans des conteneurs en acier inoxydable. La quantité de Te est te lle que la réaction se fa it simultanément sur le bouchon avec le tellure gazeux et dans le fond du conteneur avec le tellure liquide. Les traitem ents ont lieu entre 700 et 1000°C, A 900 et ■\000°C, on observe une réaction aussi importante que le Te soit liquide ou gazeux : précipitation d’une phase 0 riche en Te, Cr, Mn, induisant une déchromisation locale de l ’acier très supérieure à celle provoquée par la précipitation des carbures de chrome. IAEA-PL-463/17 2 9 3 La m icroanalyse X conduit aux résultats suivants (concen tration en poids) : - p h a s e 0 : Te = 74,6 * 1 %, Cr = 19 1 0,3 %, Mn = 6,1 * 1 %. - acier (m atrice austénitique en présence de précipités de phase 0) i on note l ’absence totale de précipités de carbures du type (Cr Fe) C ) : Fe = 70 * 1 %, Cr = 13 * 0,5 %, Ni = 15,2 1 0 , 5 % , Mo * 2 ,2 * 0 , 5 % , Mn = 0 . - acier hors zone de diffusion où l ’on observe une précipitation de carbure de chrome du type (Cr Fe) C : Fe = 65,2 * 1 %, C r = 17 1 0,5 %, Ni = 13,5 t 0,5 Í. Mo 2,2 * 0,5 %, Mn = 1,7 1 0 . 5 % . Si le tellure, à haute température,est susceptible de provoquer une déchromisation de l'a cie r, celui-ci reste austénitique. Certes sa résistance à l ’oxydation diminue, mais elle peut partiellem ent être compensée par un apport de chrome dû à la diffusion thermique et provenant du matériau de gainage non affecté par la présence de tellure. Il n'en est plus de même à basse température. Les essais effectués à 700 et 800°C montrent l'apparition d'une zone ferritique (en volume) et une déchromisation préférentielle de l'a cie r le long des joints de grains (700°C) ou de part et d'autre des joints de grains (800°C). Nos expériences de sim ulation semblent montrer que dans lencas d’un élément combustible type (U,Pu )02 ± x, fonctionnant à une température intérieure gaine de 600°C, il y a risque de déchromisation sup erficielle de la gaine accompagnée de l ’apparition d'une zone ferritique par action du tellure. Ceci est très cohérent avec nos analyses à la microsonde faites sur des sections d’éléments combustibles irradiés. Le tellure se retrouve en quantités importantes dans la zone a et seulement dans cette zone. V III - ROLE DES I0DURES DE CESIUM ET DE MOLYBDENE (COMBINAISON DE PRODUITS DE FISSION) Le fait de retrouver sur des échantillons irradiés du césium à la périphérie de l ’oxyde et en contact avec la gaine, et des précipités riches en fer et en chrome à l'in térieu r du combustible, suggère un phénomène de transport en phase vapeur. (Cette hypothèse est très fréquemment avancée par les chercheurs américains.-i L'iodure de césium, formé par combinaison de produits de fission, réagirait avec la gaine en y abandonnant le césium pour former un mélange d'iodures volatils du type ^C r, IßCr, ^Fe, l 3 F e . Ceux-ci se décomposeraient dans les parties chaudes du combustible en formant des précipités riches en Fe et en Cr et en libérant l'iode qui est ainsi recyclé continuellement. La réaction décrite ici est en outre fa cilité e par la formation d'urano-plutonate de césium à la périphérie du combustible : 2 Cs I * Ее * fU,Pu)0 2 x + (U,Pu)04Cs + (Fe) 1 2 ( 1 ) 3 qui piège le Cs libéré ou en excès par rapport à l'iode (dans un élément combustible irradié le rapport £s est voisin de 1 0 en régime permanent), I Des expériences de sim ulation corroborent cette hypothèse. Des pastilles d'acier inoxydable sont recuites dans un gradient de température (600 à 1000°C) en présence d'iodure de césium. 2 9 4 CALAIS et al. Les traitem ents ont lieu dans des ampoules de silice scellées. A 70Q°C l ’acier subit une corrosion intergranulaire importante s'accompagnant d’une perte de poids appréciable. La m icroanalyse X dans la zone corrodée révèle une augmentation de la teneur en nickel et une diminu tion de la teneur en chrome et en fer. Oans la zone de l'ampoule de silice portée à 900°C on observe un dépôt m étallique sur les parois de la capsule. La m icroanalyse X montre que ce dépôt est constitué quasi exclusivement de Fe (70 %) et de Cr (30 %), La pastille d’acier placée à cette température accuse un gain de poids. Il y a donc bien eu transport du fer et du chrome à partir des zones froides vers les zones chaudes de l'acier par l ’interm édiaire de l'iodure de césium. Cette fois encore les expériences de sim ulation sont en accord avec notre modèle et les résultats de la m icroanalyse X des zones de réaction observées sur les éléments combustibles irradiés : - déchromisation de l'acier, sous l ’action de l'iodure de césium, d’où possibilité de réaction (U,Pu)0 - acier inoxydable. - pic de concentration en césium dans la zone a et la zone c. - présence probable d'urano-plutcnate de césium dans la zone c provenant en partie du piégeage par le combustible du césium libéré par l'interm édiaire de la réaction (1). La teneur en U et Pu de la zone c est inférieure à celle observée dans ]a partie centrale du combustible et des mesures de tension de vapeur de césium sur combustible irradié impliquent la présence de Cs 2 ( (U,Pu)0^ ) au contact de la gaine. Cependant les données thermodynamiques recu eillies dans la bibliographie semblent devoir écarter un phénomène de transport du fer et du chrome dans les éléments combustibles irradiés par l'interm édiaire de l ’iodure de césium. En effet, la tension de vapeur de l'iode si elle est contrôlée par la décom position de I Cs (composé très stable, l ’énergie libre de formation correspondant à la réaction Cs (liquide) + I (gazeux) -*■ Csl (liquide) est de (- 87 000 * 19 T) calories/m ole) serait trop faible (5,6-10~15 aj-m g 1 0 0 0 °M pour permettre un transport macroscopique de fer et de chrome du type VAN ARKEL même si la form ation d'urano- plutonate ds césium à la périphérie du combustible fa cilite à priori ce déplacement. Comme celui-ci a été réalisé hors pile, on peut souhaiter que de nouvelles mesures soient faites pour déterm iner de façon plus sure les constantes thermodynamiques de I Cs. Par contre, d’après les données actuelles, l'iodure de molybdène l 2 Mo aurait une énergie libre de formation à l'état gazeux suffisamment im portante, (- 25 000 + 5,5 T) calories/inole, pour permettre un transport en phase vapeur ou chrome et du fer de la gaine vers le combustible par l'interm édiaire de la réaction I- Mo + Fe Cr (acier) -*• Mo + Fe 1 ? + C r I ? . 2 13 *3 On explique alors très bien la présence de molybdène dans la zone a. A notre avis l'iodure de molybdène et l'iodure de césium agissent de façon sim ultanée pour dégrader la gaine en produisant un effet analogue : déchromisation de l'a cie r au contact du combus t i b l e . Notons enfin que les images X obtenues à partir d’échan tillo n s irradiés ne montrent aucune superposition dans la répartition des teneurs en césium et molybdène. La form ation de molybdate de césium dans la zone de réaction semble par conséquent devoir être écartée à moins qu’il ne soit pas stable dans les conditions de préparation m étallographique. 295 IA EA-PL-46 3/17 REACTION FIG. 3. Aspect micrographique de la réaction uranate de césium - acier (recuit par paliers entre 840cC et 1200°C pendant 48 h). 2 9 6 CALAIS et al. IX - COMPATIBILITE URANATE-CESIUM-ACIER INOXYDABLE (fig . 3) Des expériences de com patibilité uranate rie césium-césiurn- acier inoxydable effectuées de 600 à 12Q0°C ont montré que le chrome de l'a cie r inoxydable était susceptible de réduire l ’uranate de césium C s 2 (U0/])pour former un chromate de césium et de l'U 02, Dans le cas des échantillons irradiés', la présence de Cr dans la гопе c peut alors s'expliquer de la façon suivante : le chrome sous forme de ferrite (Cr FeijOg ou sous forme m étallique (acier) présent dans la zone b réagit avec 1 'urano-plutonate de césium se trouvant dans la zone c : Cr (acier) + (U,Pu)0 3 ( ( C s 2 0))->- ( C r 2 0 g ) ( ( C s 2 0)) + (U,Pu)0 2 t x C r 2 0 3 + (U,Pu)0~ ((Csz0) -> (Cr 2 ü 6 ) ' ( ( C s 2 0) ) + (U,Pu)0 2 ± X ~ ROLE DU PALLADIUM Le palladium a une tension rie vapeur élevée, ce qui peut expliquer sa présence au contact de la gaine. Cependant cette expli cation ne nous satisfait pas pleinement et nous entreprenons actuel lement des études à ce sujet. X I - R O L E DU BARYUM Le rôle du Ba nous semble de peu d'im portance. La majeure partie du baryum produit par fission est fixé dans le combustible sous forme de zirconate de baryum. Nous avons pu montrer ceci par microanalyse d’échantillons (U, Pu)0^ simulant divers taux d'irradiation par introduction de produits de fission inactifs. Il est probable que le baryum que l ’on retrouve au contact de la gaine provient de la décroissance de l'iode, radioactif. X II - RECAPITULATION DES RESULTATS La m icroanalyse X, effectuée sur des lames minces prélevées dans des combustibles irradiés, les expériences de sim ulation visant à reconstituer hors pile la réaction oxyde-gaine et des considéra tions thermodynamiques nous amènent aux conclusions suivantes : - Zone a. La zone a correspond à un acier déchromisé sous l'actio n des produits de fission vo latils ou de leurs combinaisons (Te, I C s , I 2 Mo). La résisteince à l'oxydation du matériau de gainage est diminuée, ce qui favorise une réaction directe de ce dernier avec l'oxyde mixte (U,Pu)0 2 t x . - Zone b. La présence en fortes proportions dans la zone b d'uranium et de plutonium et de chrome et de fer, nous amène à proposer le mécanisme de réaction suivant : acier zone a (déchrom isé.c'est-à- dire oxydable) + oxyde mixte d'uranium et de plutonium voisin de la stoechiom étrie -*■ fe rrite (FeCr) 2 0 3 . La croissance de cette zone de réaction (b) obéit à une cinétique complexe. Elle est essentiel lement gouvernée par la diffusion en phase solide, le potentiel d*oxygène à l'in terface gaine-combustible et la déchromisation superficielle de l'acier (zone a), - Zone c. En périphérie l'oxyde mixte (U,Pu)0 2 ± x est voisin de la stoechiom étrie. Le césium non combiné à l'iode et celui libéré par la réaction de I Cs sur l ’acier peut alors donner lieu à la form ation d'un urano-plutonate de césium (U,Pu)0¿| Cs2. IAEA-PL-463/П 29 7 X III - PROPQSITIGN D’UM SCHEMA PERMETTANT DE SUIVRE LE DEVELOPPEMENT DES REACTIONS OBSERVEES AUX INTERFACES GAINE COMBUSTIBLE EN COURS D'IRRADIATION SOUS NEUTRONS RAPIDES 1) Le combustible est initialem ent sous-stoechiométrique (cas classique). On assiste à une redistribution rapide de l ’oxygène. En périphérie l ’écart à la staechiom étrie est pratiquement nul. 2 ) Formation de produits de fission : I, Cs, Mo, Te. Très rapidement la quantité de molybdène et de césium est suffisante pour fixer la totalité de l ’iode. 3) Il y a du césium libre. Celui-ci se dirige vers la gaine (zone froide) ■ et réagit avec l ’oxyde stoechiom étrique pour forme de l ’urano- plutonate de césium. La zone c apparaît la première,comme nous l'fw ons observé, 4) Les produits de fission combinés ou non (Te, I Cs, IgMo) attaquent la gaine et provoquent sa déchromisation (zone a). 5) La surface intérieure de la gaine voit alors sa résistance à l ’oxydation diminuer. Il s'ensuit que la réaction directe combustible-gaine est désormais possible avec formation de fe rrite ( F e C r ) , , 0 3 ( z o n e b ) . XIV - CONCLUSIONS Nos résultats expérimentaux montrent que les produits de fission volatils (tellure, iodures de césium, etc) peuvent jouer un rôle important dons le développement de la réaction oxyde-gaine observée sur des éléments combustibles des réacteurs rapides, ceci en raison de leur action déchromisante vis-à-vis de l ’acier (action directe et phénomènes de transport en phase vapeur du type VAN ARK.EL). Nous pensons également que l'écart à la stoechiornétrie du combus tib le est un paramètre essentiel. En outre, il est probable que la température à l'in térieu r de la gaine et l ’accélération de la diffusion sous irradiation ont une grande influence. Le mécanisme de la réaction que nous avons présenté n'est qu'un schéma directeur. Il est probable que les phénomènes sont beaucoup plus complexes. Une analyse plus approfondie est nécessaire pour cerner complètement le problème. IAEA-PL-463/18 THE ROLE OF CAESIUM IN CHEMICAL INTERACTION OF AUSTENITIC STAINLESS STEELS WITH URANIUM PLUTONIUM OXIDE FUELS R.W. OHSE, M. SCHLECHTER European Institute for Transuranium Elements, EURATOM, Karlsruhe, Federal Republic of Germany Abstract THE ROLE OF CAESIUM IN CHEMICAL INTERACTION OF AUSTENITIC STAINLESS STEELS W ITH URANIUM PLUTONIUM OXIDE FUELS. The role of caesium in fuel/cladding interaction of austenitic stainless steels with mixed oxide fuels is discussed by considering the phase relationships, oxygen potentials and defect structures of the oxide layers and the oxygen-containing liquid multicomponent caesium phase. The possible reaction mechanisms of the various kindsofattack, the formation of a layered structure, the surface ablation of steel or matrix attack, and the intergranular attack are discussed. The initial alloy oxidation state of the stainless steels, obtained by the formation o f a passivating oxide scale, is considered on the basis of the chromium content, the constitution diagram and the oxygen partial pressure at the fuel surface. Dissolution o f the oxide components at high burn-up in the oxygen-containing liquid caesium phase by oxygen exchange reactions gives a reasonable explanation for changes in oxidation rate caused by changes in defect concentration. The scale formation, stress development, breakdown of the oxide layers, and penetration o f the oxygen-containing caesium in between the oxide film and the alloy explain the strong increase in corrosion. 1. INTRODUCTION Swelling and clad corrosion are the main life-lim iting factors of a fuel element at high burn-up. In view of the latest irradiation results [1, 2], clad corrosion is being given increasing priority. The high burn-up system under consideration is a multicomponent system formed under the irreversible conditions of radial temperature and composition gradients and fission product distribution. Considerable uncertainties exist with regard to chemical state and concentration of fission products. Practically un controllable crack formation during fuel cycling provides radial gas-phase connections under high temperature gradients, thus permitting gas trans port, which can lead to remarkable changes in the fission product concen trations along the clad. Postirradiation examination [3,4] of irradiated fuel pins reveals various kinds of chemical attacks, mainly depending on the . oxygen potential, the temperature and burn-up, resulting in the formation of a layered structure, the dissolution of passivated oxide films, ablation of the steel or matrix attack, and finally intergranular attack. For a successful approach it seems unavoidable to simplify the multi- component system by restricting it to the main reaction products formed at the interface fuel cladding. The role of caesium, regarded as the most aggressive fission product [5, 6 ] in clad interaction, and the reaction 2 9 9 3 0 0 OHSE and SCHLECHTER mechanisms are discussed by comparing the initial state of oxidation of the austenitic stainless steel and its defect mechanism with the final reaction layers found at high burn-up. 2. INITIAL OXIDATION STATE OF AUSTENITIC STAINLESS STEELS AT THE OXYGEN POTENTIAL OF THE URANIUM PLUTONIUM O XID E F U E L 2.1. Phase relationships and oxygen potentials of the protective oxide scales and the fuel The chemical state and structure of oxide scales on the cladding of austenitic stainless steels [7] are given by the oxygen potential and tem perature at the fuel surface and the phase relationships within the phase fields of significance of the Fe-Cr-O [8-11], Ni-Cr-O [12] and Fe-Cr-Ni-O [12] diagrams. The composition of the oxide scales depends strongly on the Cr concen tration of the alloy, as shown in the Fe-Cr-O and Ni-Cr-O phase diagrams given in Figs 1 and 2. The principal oxides formed on pure iron oxidized on air [11, 13] above 570°C are FeO, Fe 3 0 4 and а-ГегОз . Above 700°C 95% of the oxide consists of wüstite FeO. Below 570°C only F e 3Û4 and F e 2 0 3 are found. Small additions of Cr up to 2% increase the oxidation rate [14, 15]. From 2 - 13% Fe-Cr alloys are in equilibrium with spinel-type (F e , C r ) 3 0 4. The oxidation rate decreases with increasing Cr content. Above 13% Cr the Fe-Cr alloys are in equilibrium with СГ 2О 3 . The range of Fe-C r alloy solid solutions coexisting with the spinel-type oxide (Fe, Cr) 3 0 4 seems rather uncertain [10, 11]. The same applies to FeO coexisting with chromite intermediate compositions of magnetite Fe 3 O4 - F e C r 2 O4 . Both composition limits marked by A and В in Fig. 1 are subject to further investigation. In addition, the range of solid solutions of Fe-Cr that coexist with wüstite FeO as a function of temperature is also open to further 0 //tfyr / / ) I I I ' 4 / / s j t r y ' KS К/ -J______¿ ______J______* I ______k _ ,r 10 A 20 30 40 50 60 70 80 90 at* Cr FIG. 1. Fe-Cr-O phase diagram at 1300°C (1000°C) [8] IAEA-PL-463/18 3 0 1 0 FIG.2 . Ni-Cr-0 phase diagram at 1000°C [12]. N1 FIG.3. Plane section o f the quaternary phase diagram F e -C r-N i-0 at 1000°C [12]. investigation. According to the phase diagram given by Seybolt [ 8 ], solid solutions of Fe-C r below 13 wt. % are not compatible with the (Fe, Cr )2 0 3 type oxide. An ironchromite (Fe-Cr) 3 0 4 spinel [16], i. e. a solid solution of F e 3 0 4 with F e C r 2 0 4 as an intermediate layer, is to be expected. According to Fig. 2 [12], the oxide scale in equilibrium with Ni-Cr alloys at 1000°C is of the spinel type [16] below 10 at. % Cr and consists of C r 2 0 3 above 10 at. % Cr. Microprobe analysis at low Cr contents (< 10 at. % Cr) reveals an oxide layer of NiO above a second oxide layer presumably consisting of a spinel in an NiO matrix. Above 10 at. % Cr the nickel content in the Cr 2 0 3 layer was found to be as low as 0 . 7 at. %. 3 0 2 OHSE and SCHLECHTER Temperature °C FIG .4. Oxygen potentials o f compounds involved in fuel/cladding interaction and of uranium plutonium oxides as a function of valency. The influence of nickel has been investigated by various authors [12, 17-20]. Figure 3 shows a plane section (in reality a three-dimensional surface) of the quaternary system Fe-Cr-Ni-O [12], based on the four ternary systems, at an oxygen concentration corresponding to the oxygen solubility limit of the ternary alloy Fe-Cr-N i at 1000°C. The minimum chromium concentration at which the alloys are in equilibrium with Cr 2 0 3 , starting from 13 at. % Cr in the Fe-Cr system, passes through a minimum of 7 at. % Cr at 70 at. % Ni and 23 at. % Fe to 10 at. % Cr in the Ni-Cr system. The chromium content of the alloy must, of course, be higher to compensate for the depletion caused by the formation of the Cr 2 0 3 sc a le . A surface depletion as low as 10. 8 at. % Cr at the metal-oxide interface was observed at a bulk concentration of 21. 9 at. % Cr. Figure 4 plots the oxygen potential against temperature of all compounds possibly involved in the initial pure system, which is assumed to be still free of fission products, compared to the potential of the oxide fuel at various valencies [21-24]. As shown in Fig. 4, the oxygen potential of the oxide fuel is fixed by its temperature and valency. The valency is fixed by the fuel composition O/M and Pu/(U + Pu). To determine whether a Fe-C r alloy with a given Cr/(Fe + Cr) mole fraction at the oxygen partial pressure [21, 25, 26] of the fuel surface is in equilibrium with the sesqui- oxide or the spinel type oxide, its constitution diagram, i. e. log po ver sus Cr/(Fe + Cr), must be known. The constitution diagram in Fig. 5 [ 8 ] is given for 1300°C. The oxygen partial pressure of a four-valent ternary oxide fuel at 1300°C is 10 ~13 atm, corresponding to an oxygen potential of - 93 kcal/mole [21-23]. According to Fig. 5, alloy compositions above 13%Cr would be in equilibrium with the sesquioxide. According to Fig. 4, the oxygen partial pressure at 1300°C above pure Cr 2 0 3 in equilibrium with chromium metal is 1 0 '16 atm, which is in agreement with the constitution IAEA-PL-463/18 3 0 3 Cr (wt.%) Cr»Fe F IG .5. F e -C r-0 constitution diagram, logpo2 versus mole fraction Cr/(Fe+Cr) [ 8]. diagram. According to the diagram calculated by Schmalzried [27] from the values of Katsura and Muan [28] alloy compositions above only 10% Cr would already be in equilibrium with the sesquioxide under the conditions of an oxygen pressure of 10 ' 15 atm and a temperature of 1300°C. The reason for these variations in the oxygen pressure at which the alloy is in equi librium with the sesquioxide as a function of chromium content seems to be the difficulty of establishing equilibrium under conditions of relative small changes in oxygen pressure. According to Katsura and Muan [28], the oxygen pressure at which the alloys FeCr 2 0 4 and C r 2 0 3 coexist, i. e. within the three-phase field of the phase diagram in Fig. 1, was determined to be P o 2 = 2. 8 X 1 0 " 14 atm. A valency increase in the fuel to 4. 001 would increase the oxygen partial pressure to approximately 10~12 atm, i. e. into the two- phase field where the alloy is already in equilibrium with the spinel. If we. assume the singular point at which the alloy is in equilibrium with the sesquioxide and the spinel to be around 13% Cr, as indicated by Seybolt and others [ 8 , 9, 29], a depletion of Cr in the alloy to below 13% by the formation of Cr 2 0 3 would lead to the formation of an intermediate spinel scale. The depletion or possibly variation of the Cr content in the alloy depends on the component diffusion coefficients [11, 14, 16, 30] in the alloy compared with those in the sesquioxide and spinel. 2.2. Defect structure of the protective layers To understand the role of fission product reactions on the protective, passivating properties of the (Fe, Cr) 20 3 oxide layer, it is necessary to investigate its defect structure [31, 32] at minimum oxidation rate. A mini mum in oxidation rate assumes low rates of ionic transport, which can only 3 0 4 OHSE and SCHLECHTER be expected at low defect concentrations. The increase in oxidation rate by adding up to 2% Cr can be explained by an increase in cation defects. A Cr content of 2 - 13% results in the formation of iron chromites of the spinel type (Fe, Cr) 3 0 4, i. e. in a solid solution of Fe 3 0 4 and F e C r 2 0 4 . At higher Cr concentrations (13 - 30%) a minimum in oxidation rate is observed around 18 wt. % Cr. Cr 2 0 3 and F e 2 0 3 form a continuous series of solid solutions [33]. Cr 2 0 3 is considered to be cation deficient [14]. For energetic reasons oxygen interstitials are very unlikely [17]. Fe 2 0 3 is oxygen deficient [14]. Transport rates via oxygen vacancies and iron interstitials are comparable and depend on temperature. This would mean an equilibrium between Schottky [33] and Frenkel [34] disorder. Some authors [14, 30, 35-37] consider iron interstitials as slightly more mobile than the larger oxygen vacancies. Since the end members of the solid solution range are of opposite defect structure, it seems reasonable to assume a minimum in defect concentration, i. e. low oxidation rates at more stoichiometric compositions, which itself depends on temperature and oxygen pressure [14]. The oxidation rates are near to parabolic [38]. Cation diffusion through the M 2 0 3 type oxide seems to be the predominant mechanism in the early stage. A minimum in ionic defect concentration was found at 4 mol. % Fe 2 0 3 [17, 39]. Cr 2 0 3 containing less than 4% Fe 2 0 3 behaved as a p-type semiconductor, whereas Cr 2 0 3 containing more than 4% F e 2 0 3 was shown to be a n-type semiconductor. Any disturbances of this 'defect concentration equilibrium1, whether by chemical interaction or under irradiation at a high neutron flux [40] , will tend to increase the oxidation rate. The aim of further investigations should therefore be to explain a reduction of the protective properties of the oxide layers [41,42], apart from possible irradiation influence, by reaction with fission product phases, possibly by Cr depletion, which may then lead to ruptures and cracks because of the formation of stresses and voids at the steel/oxide interface, since the cation diffusion through the M 2 O 3 type oxide seems to be the predominant mechanism. The possibility of forming intermediate layers by chromium depletion [ 1 2 , 18] and its consequences on stress development [9, 11, 15, 29,43] by the various oxide to metal volume ratios [11,44] has to be taken into account. Crack forma tion could, of course, also be due to differential contractions during temperature cycling under reactor operating conditions. Stress and crack formation [15, 17] could finally explain why these scales are lifteä, giving rise to a greatly increased corrosion rate. 3. FUEL/CLAD INTERFACE AT HIGH BURN-UP 3. 1. Determination of reaction products at the interface stainless steel- oxide layers-fission product phases Since a purely thermodynamic approach to the irreversible interaction behaviour of the multicomponent system stainless steel/fission product/fuel necessarily contains many assumptions and uncertainties, a preliminary selection among the possible reactions should be obtained by microprobe and X-ray analysis of the reaction products formed under reactor conditions. As a main result of the various postirradiation examinations of irradiated fuels made in the various laboratories, caesium is in general accepted [ 6 ] IAEA-PL-463/ 18 305 to be one of the most aggressive elements in oxide fuel/cladding interactions. The investigation of the role of caesium requires microprobe analysis of the reaction products of preoxidized stainless steel samples, brought into contact with liquid oxygen-containing caesium phases and (U, Ри)Ог mixed oxides under a temperature gradient at 500 to 600°C clad temperature. Both the composition of the various reaction zones formed between the clad and the fuel and, in the case of intergranular attack, the reaction products formed within the grain boundaries must be known as a function of oxygen potential and temperature in order to establish the general reaction and finally the reaction mechanism. Special attention should be paid to adjacent fission product phases [45], which possibly exist as liquid multi- component electrolytes. Postirradiation examination [5,46-49] of irradiated fuels with austenitic stainless steel cladding reveals three kinds of attack, the form ation of a layered structure in the case of initially near to stoichiometric fuels, the dissolution of thin passivating layers, and the subsequent surface ablation of the steel in the presence of a liquid phase, and finally the inter granular attack of sensitized steels within the grain boundaries. All three types of attack are usually of a rather local and less uniform appearance, depending on the local concentration of fission products. In the case of a layered structure it is assumed that the passivating oxide layers, mainly consisting of Сг2 Оз, are formed before an appreciable amount of fission products has reached the fuel/clad interface. Surface ablation is observed on the non-protected steel surface after a thin protective oxide film has been dissolved in the liquid phase. Intergranular attack is observed especially in sensitized steels, where Cr 2 3 C 6 precipitations along the grain boundaries are oxidized. To give a complete treatment of the various kinds of chemical attack, the reaction mechanism leading to a layered structure is discussed. F ig u re 6 summarizes the general features of analytical results reported by the various laboratories [2, 3, 50-54], giving the composition of the scale formation at the interface fuel clad after high burn-up. As shown in Fig. 6 , the original M 20 3 oxide layer is separated from the steel by a liquid FIG.6 . Schematic summary of fission products analysed within the layered structure at the fuel/cladding interface after high burn-up. 306 TABLE I. PHYSICAL CONSTANTS AND THERMODYNAMIC DATA ON CAESIUM AND IODINE COMPOUNDS M.P. B.P. A G f(c ) л н 2 Э8(с) CC) CC) (kcal/ mole) (kcal/ mole) Cs 28.40 ±0.01 [63] 678.4 [63] I 113.5 [63] 184.35 [63] SCHLECHTER andOHSE CszO 490 (in N2 ) [63] Csl 621 [63] 1280 [63] - 80.5 [65] CsFeOg - 137.8 [5] Cs2Cr0 4 (954) [64] Cs2 U 0 4 -478 [69] Cs2 Mo0 4 (936) [64] Fel2 609 [65] 935 [ 66] 45.770+ 0.0069 Tlog T - 0 0512.T -3 0 .0 ± 2 .0 [67] (298 - 867°K) [67] Crl2 856 [63] 1248 [65] -26.000(lOOO'K) [ 68] ' - 3 7 . 8 ± 1.4 [67] N il2 797 [63] - 2 . 0 0 0 ( 1000°K) [ 68] -2 3 .0 ± 2 .0 [67] / IAEA-PL-463/18 3 0 7 multicomponent phase of sim ilar fission product composition to the outer phase at the СГ 2О3 fuel oxide interface. In addition to steel components in both intermediate layers, precipitations of Fe, Cr and Ni are found at the fuel surface. The assumption of ionic transport across the liquid phase, however, requires the existence of a gradient in chemical potential, possibly realized by the precipitation of compounds at the oxygen-rich interface towards the fuel. Further information can be obtained from the chemical state of the steel compounds found at the fuel surface. Any possible reaction mechanism must, therefore, be able to explain the initial increase in rate of oxidation, which by surface depletion of steel components may lead to intermediate layers, causing cracks and the final lifting of the layers, thus allowing the liquid phase to penetrate into the interface steel/M 20 3, leading to a rapid increase in corrosion. 3. 2. Thermodynamics and phase relationships Once the reaction products are known, it is necessary to determine under what conditions they are formed. The main parameters are tem perature and oxygen potential at the interface fuel/clad given by the tem perature profile and the radial O/M composition gradient, which itself is determined by the initial bulk composition, burn-up, the chemical state of fission products and their distribution. Therefore, the phase relationships of significance for these reaction compounds and their oxygen potential should be investigated in the temperature range of the fuel/ clad interface. The following phase diagrams are of direct significance: U-Pu-O [22, 55], Cs-O [56-58], Cs-Cr(SS)-0, Cs-Mo-O [51, 59], Fe-Cr(Ni)-0 [8-12] and Cs-(U, Pu)-0 [55, 60]. Since the high Cs oxides from Cs20 upwards are not to be expected because of the oxygen potential of the fuel surface, special attention should be paid to the oxygen solubility in caesium and its partial molar free energy compared with that of the fuel. It should then be possible to predict under what conditions caesium-fuel compounds such as Cs 2 U 0 4 or caesium-fission product compounds such as Cs 2 M 0 O 4 are formed and whether caesium could be stabilized [61, 62] in its radial position. The available thermodynamic data and phase relationships on caesium compounds are summarized in Fig. 7, Fig. 5 of IAEA-PL-463/15 (these Proceedings) and Table I for further discussion. Figure 7 shows the binary Cs-O phase diagram [57, 58]. Figure 5 of IAEA-PL-463/15 [70] gives the oxygen potentials of Cs20 and oxygen dis solved in caesium [56, 71], together with the data for Na20 [72, 73] and oxygen dissolved in sodium and, for comparison, the oxygen potential of the ternary (U, Pu)-oxides at the fuel surface for the valency range of technological interest at high burn-up. According to the latter figure, Cs20 at the oxygen potential of a stoichiometric fuel with a valency of 4 can only be expected at temperatures below 450°C. At a fuel/clad interface tem perature of approximately 600°C, -Cs20 can only be formed at the oxygen potential of a hyperstoichiometric fuel with a valency of 4. 001, corres ponding to a O/M composition of 2. 0004 at 20 mol. % of P u02. In addition, the melting point of Cs20 of 490°C (Table I) lies below the temperature range of 500 to 650°C expected at the fuel/clad interface. Therefore a liquid, . most likely a multicomponent, thin layer of caesium, containing oxygen, and the more volatile fission products such as tellurium, selenium and rubidium is to be expected. The amount of caesium or other, volatile fission 3 0 8 OHSE and SCHLECHTER Cs a t.2 0 Cs-0 F IG .7. C s-0 phase diagram [57, 58]. products available at the fuel/ clad interface depends on the stability of compounds suchas Csl, Cs 2 (U, Pu)04, Cs 2 M o 0 4 and R b 2 M o0 4 [59, 64, 7 4 -7 8 ] as a function of the oxygen potential, and the density of the fuel with regard to interconnected porosity and radial crack formation. Simulation experiments on fission product transport by Evans and Aitken [62] and Crouthamel and Johnson [61] confirm transport towards the clad almost up to stoichiometric fuel compositions, whereas in hyper stoichiometric fuels caesium appeared to be fixed in its radial distribution by the formation of stable compounds. Postirradiation examination clearly shows that molten layers are not observed all along the fuel/clad interface but seem in general to be rather localized in areas of high radial crack formation. Local concentration changes of these volatile phases seem to be mainly due to gas transport facilities along interconnected radial gas phases, such as cracks, formed during temperature cycling. A dense fuel will show no appreciable migration of caesium towards the clad. According to the estimated values [59, 61, 62,64, 74-78], Na 2U 0 4 and C s 2 U 0 4 compounds are only expected at the oxygen potential of a slightly hyperstoichiometric fuel. Because of the stable Csl compound, the chemical state of caesium is of importance for chemical gas transport reactions of stainless steel components by the van Arkel-de Boer mass transfer mechanism. According to fission yield calculations for a fast neutron flux [79-82], the ratio of Cs to I is of IAEA-PL-463/18 3 0 9 the order of 10 to 1. Free iodine will therefore only be available if caesium can form compounds with either fuel, fission product or cladding compo nents, which are more stable than Csl. 3. 3. Reaction mechanism Determination of the various possible reaction mechanisms requires a detailed investigation of the composition of the various reaction zones, i. e. the possible depletion of the steel components within the clad and oxide layer such as Cr, copipared with the initial state of passivation. This should give evidence on to what extent the adjacent liquid phase [29, 73, 83-85] can alter the composition of the initial oxide layer by dissolving components and, as a consequence of this, changing the defect concentration or even defect structure and thus the rate of oxidation. A further possibility to be noted here is the introduction of fission products into the protective oxide film, such as molybdenum, which may increase the number of cation vacancies and hence also increase the rate of oxidation. In all considera- tions special attention has to be paid to questions of ionic transport, i. e. the gradient of the chemical potentials [ 8 6 ] of the various components, and their diffusions coefficients. In the following an attempt is made to point out possible reaction mechanisms. 3. 3. 1. Disturbance of the anion-cation defect concentration balance within the passivating oxide layer According to section 2. 2, the end members of the solid solution range F e 20 3 and C r 2 0 3 are of opposite defect structure [14]. A minimum vacancy concentration and thus a minimum oxidation rate is obtained by approaching a stoichiometric composition. The minimum defect concentration was shown to be obtained at 4% Fe 2 0 3 [17, 39] in the solid solution. A reduction on either side would consequently again increase the defect concentration and thus the rate of oxidation. The rate of oxidation in the case of spinel formation below 13% Cr depends mainly on the phase width and stability of wüstite FeO, which changes with temperature and alloy composition. Since the austenitic stainless steels 304, 316, 1.4988 and 1.4981, which are considered as cladding materials, are above 15% Cr, the influence of FeO will not be considered. There is experimental evidence for a lowering of the Cr concentration in the (Fe, Cr)2 0 3 layer. It seems therefore reasonable to develop possible reaction mechanisms, which, in agreement with the thermodynamic data and phase relationships of the simplified system, can explain Cr dissolution into a liquid Cs phase, containing a sufficiently high concentration of oxygen, on behalf of an oxygen exchange mechanism, and subsequent com pound formation, possibly as a chromite C s-C r0 2 [87, 8 8 ], to guarantee the necessary gradients of the chemical potentials. F ig u re 8 gives a schematic display of the possible reactions at the various phase boundaries. The chromium-rich, M 2 0 3-type oxide (aCr 0 n e a r to 1 ) is assumed to be dissolved in caesium containing oxygen according to C r 2 0 3 +[0]£ =2[Cr02]-Cs (1) 3 1 0 OHSE and SCHLECHTER oxide fuel at the sensitizing ' I \ ' \ temperature ^ ------' \ disturbance h ------: у I \of defect 1 intergranular / surface i spinel \balance reaction mechanism of gas transport along 1 attack / ablation J formation \ СГ2О3 dissolution radial cracks FIG. 8 . Schematic display of possible reaction mechanism of chemical interaction in oxide fuels at the fuel/cladding interface. Of course, the ionic chromium complex formed and its chromium valency depends on the oxygen potential at the fuel surface. Since the oxygen potential decreases with increasing temperature, the solubility limit de creases and precipitation from the supersaturated solution will occur at the high temperature side of the liquid multicomponent caesium layer according to' [CrOz]-Cs +[Cs]+ =[CsCr02] Cs (2) Because the activity of Fe 2 Oa in the M 2 0 3 oxide is rather low and the oxygen potential of C sFe0 2 calculated from the estimated values given by Fitts and others [5, 89] is above the oxygen potential of the ternary stoichiometric oxide fuel (Uo.so Puo.2o Ю 2.00 at a valency of 4, the formation of C s F e 0 2 can only be expected at the chromium-depleted, iron-rich surface of the oxide layers at one oxygen potential of a hyperstoichiometric oxide fueL Crack formation within the fuel, as schematically indicated in Fig. 8 , would permit Cs vapour transport as either metal or iodide and the neces sary oxygen transfer via the C0/C0 2 gas-phase mechanism. The gas diffusion current [90] is mainly determined by the pressure, the tempera ture gradient and the diffusion coefficient. According to Hall measurements [58], caesium and oxygen exist to a great extent in their ionic state within the liquid caesium. The selective dissolution of the M 2 0 3 oxide layer will disturb the defect concentration balance and increase the rate of chromium diffusion to the liquid oxygen containing Cs phase. As a result, chromium depletion will occur at the metal surface, leading to the formation of an intermediate layer of the spinel-type oxide. It is assumed that stress developments IAEA-PL-463/18 3 1 1 within these layers during temperature cycling under reactor operating conditions will lead to cracks and finally to the lifting of the oxide scale [18]. The penetration of the liquid multicomponent phase between the oxide layer and the metal would then give rise to increased corrosion by surface ablation and intergranular attack. 3. 3. 2. Surface ablation and intergranular attack Though there are cases where a transition from a thin protective film to a thick non-protective scale occurs, in general thin oxide films of the M 2 0 3 type are assumed to be dissolved before spinel formation can take place. Surface ablation of the unprotected alloy may consist of simple dissolution [91] up to the solubility limit of the steel component within the liquid phase. With increasing oxygen dissolved in caesium and increasing temperature, the rate of corrosion increases greatly and finally leads to intergranular attack. The concentration of caesium ions in liquid caesium increases with increasing oxygen content. It is assumed that the metal components of the steel alloy go into solution by reducing C s+ to Cs according to Fe + 2 Cs+ = Fe2+ +2 Cs (3) Cs+ is formed again at the oxygen potential of the fuel by reducing oxygen according to 2 C s + l/2 0 2 = 2Cs+ +02" ’ (4) The dissolution mechanism through oxidizing iron and reducing the caesium ions according to reaction (3) can likewise be formulated by the iron reacting with the active oxygen ion O2' in the diffusion layer at the phase boundary alloy-liquid phase to Fe +02' = FeO + 2e (5) 2 C s + + 2 e = 2 C s, where the caesium is again oxidized to the caesium ion according to reaction (4) by dissolving oxygen in the liquid fission product layer at the fuel s u rfa ce . The formation of FeO at the chromium-depleted iron-rich, unprotected alloy surface depends on the oxygen potential and the temperature and should be possible above 570°C and an oxygen potential of - 100 kcal/mole. Since the oxygen potential of FeO increases with temperature far more rapidly than that of the four-valent (U, Pu)02, FeO would become unstable at the high-temperature side of the liquid layer towards the fuel. Apart from the proposed van Arkel-de Boer gas transport mechanism [5, 24, 92-96], this could give a possible explanation for the Fe, Cr and Ni transfer in a liquid fission product phase towards the fuel surface [97, 98]. The possible reaction mechanism is shown schematically in Fig. 8 . As shown in Fig. 6 , Cs, Mo, О and I have been found in the grain boundaries of austenitic stainless steel. Various assumptions [59, 99] have been made on the formation of low melting eutectic compositions such 3 1 2 OHSE and SCHLECHTER as offered in the Cs 2 MoC>4 - M 0 O 3 phase diagram at 450°C [100]. The formation of such compounds would, however, require a rather high oxygen potential. Besides this, they do not explain the Fe, Cr and Ni depletion reported along the attacked grain boundaries. Though the presence of Mo could possibly be explained by a chemical gas transport reaction involving M0 O 2 I 2 sim ilar to the WO2 I 2 oxyiodide gas transport reported by Dettingmeiyer et al. [101], Cs-Cr-O compounds, such as those discussed above, are necessary to explain the Cr depletion. The formation of molybdenum compounds such as Cs 2 M o 0 2 could possibly liberate iodine for further gas transport reactions where Fe is transported as F e l 2 towards the fuel surface. The intergranular oxidation attack observed in sensitized steels [2, 70, 102] can be explained by the oxidation of chromium carbide phases, such as С г 2зС6, precipitated along the grain boundaries of the stainless steel. Since the formation of Сг 2зС 6 phases in the grain boundaries leads to a chromium depletion at the grain surface, intergranular attack by the oxygen-containing liquid caesium phase, penetrating into the grain boundaries [46, 52,74, 103-105], occurs more easily. The intercrystalline corrosion proceeds along the grain boundaries by dissolving the steel components as described above and leads to an appreciable reduction in strength and elongation. The formation of circular cracks and fracture observed within the area of grain boundary attack can now be explained by differential contractions during thermal cycling. REFERENCES [1] AMERICAN NUCLEAR SOCIETY, Fast Reactor Fuel Element Technology (Proc. Conf. New Orleans, 1971), Trans. Amer. Nucl. Soc. 14 Suppl. 1 (1971). [2 ] JOHNSON, C . E ., CROUTHAM EL,~C.E., Ann. Meeting Boston, 1971, Trans. Amer. Nucl. Soc. 141 (1971). [3 ] JOHNSON, C .E ., CROUTHAMEL, C .E ., J. Nucl. Mater. 34 (1970) 101. [4 ] PERRY, K .J ., CRAIG, C .N ., Trans. Amer. Nucl. Soc. 12 (1969) 564. [5 ] FITTS, R.B., LONG, E .L ., LEITNAKER, I. M ., Rep. ORNL-TM-3385 (1971). [ 6] JOHNSON, C .E ., CROUTHAMEL, C .E ., Fast Reactor Fuel Element Technology (Proc. Conf. New Orleans, 1971), Trans. Amer. Nucl. Soc. 14 Suppl. 1 (1971) 17. [7] ROSENBERG, S.J., US NBS Monograph 106 (1968)” 116. [ 8] SEYBOLT, A .U ., J. Electrochem. Soc. 10]_ (1960) 147. [9 ] LAI, D ., BORG, R.J., BRABERS, M .J ., MACKENZIE, J .D ., BIRCHENALL, C .E ., Corrosion 17 (1961) 109. [10] WOODHOUSE, D ., WHITE, J ., Trans. Br. Ceram. Soc. 54 (1955) 333. [11] BIRCHENALL, C .E ., Z . Elektrochem. 63 (1959) 790. [12] CROLL, J.E., WALLWORK, G.R., Oxid. Metals 1 (1969) 55. [13] DAVIES, M .H ., SIMNAD, M . T ., BIRCHENALL, C .E ., J. Met. 3 (1951) 889. [14] HAY, K .A ., HICKS, F .G ., HOLMES, D .R ., Korrosion 23 (1971) 33. [15] WOOD, G .C., HODGKIESS, T., WHITTLE, D.P., Corros. Sei. 6 (1966) 129. [16] SCHMALZRIED, H ., Werkst. Kotros. 22 (1971) 371. [17] SHARP, W .B .A ., Corros. Sei. 10 (1970) 283. [18] HOBBY, M .G ., WOOD, G .C ., Oxidat. Metals 2 (1969) 23. [19] WOOD, G .C ., WRIGHT, I .G ., HODGKIESS, T ., WHITTLE, D .P ., Korrosion 23 (1971) 16. [20] PFEIFFER, H ., HAUFFE, K ., Z . Metallk. 43 (1952) 364. [21] RAND, M .H., MARKIN, T.L., Thermodynamics of Nuclear Materials 1967 (Proc. Symp. Vienna, 1967), IAEA, Vienna (1968) 637. [22] INTERNATIONAL ATOMIC ENERGY AGENCY, The Plutonium-oxygen and Uranium-plutonium-oxygen Systems: A Thermochemical Assessment, Technical Reports Series N o .79, IAEA, Vienna (1967). [23] OHSE, R. W ., CIANI, C.', Thermodynamics of Nuclear Materials 1967 (Proc. Symp. Vienna, 1967), IAEA, Vienna (1968) 545. IAEA-PL-463/18 3 1 3 [24] SPEAR, K .E ., OLSEN, A . R. , LEITNAKER, J .M ., Rep. ORNL-TM-2494 (1969). [25] OHSE, R. W ., OLSON, W .M ., Plutonium 1970 and Other Actinides (Proc. Conf. Santa Fe, 1970), Nucl. Metall., Metall. Soc. AIME 17 pt.II (1970). [26] OHSE, R. W ., OLSON, W. M ., Rep. EUR 4633 e (1971). [27] SCHMALZRIED, H., private communication, 1972. [28] KATSURA, T., MUAN, A., Trans. AIME 230 (1964) 77. [29] WOOD, G ., Corros. Sei. 2 (1961) 173. [30] LINDNER, R., AkeRSTROM, k . , Z . Phys. Chem ., N .F . 6 (1956) 162. [31] WAGNER, C ., ZIMENS, K .-E ., Acta Chem. Scand. 1_ (1947) 547. [32] WAGNER, C . , Trans.Faraday Soc. 34 (1938) 851. [33] SCHOTTKY, W ., Thermodynamik, Springer-Verlag, Berlin (1929). [34] FRENKEL, J. , Z . Phys. 35 (1926) 652. [35] LINDNER, R., Arch. Kem. 4 (1952) 381. [36] IZVEKOV, V .l., GOBUNEV, N.S., BADAD-ZAKHRAPIN, A .A., Fiz. Metal. Metalloved. 14 (1962) 195. [37] LINDNER, R., Z. Naturforsch. 10 (1955) 1027. [38] WAGNER, C ., Korrosion 23 (1971) 2. [39] FOOTNER, P .K ., HOMES, D .R ., MORTIMER, D ., Nature, Lond. 216 (1967) 54. [40] STIEGLER, J .O ., BLOOM, E .E ., J. Nucl. Mater. 41 (1971) 341. [41] JAENICKE, W ., Passivierungs-und Anlaufvorgänge an Metalloberflächen, Springer-Verlag, Berlin (1956). [42] OHSE, R .W ., Dissertation Univ. Erlangen-Nürnberg, 1958. [43] BIRCHENALL, C .E ., Corrosion 17 (1961) 109. [44] PILLING, N .B ., BEDWORTH, R.E., J. Inst. Met. 29 (1923) 529. [45] POTTER, P .E ., submitted to IAEA Panel Behaviour and Chemical State of Fission Products in Irradiated Fuels, Vienna, 1962. [46] WEBER, J .W ., JENSEN, E .D ., Ann. Meeting Boston, 1971, Trans. Amer. Nucl. Soc. 14 (1971) 175. [47] FITTS, R. B ., LONG, F .L ., LEITNAKER, J .M ., Fast Reactor Fuel Element Technology (Proc. Conf. New Orleans, 1971), Trans. Amer. Nucl. Soc. 14 Suppl. 1 (1971) 18. [48] COQUERELLE, M ., ANDRIESSEN, H ., HOPPE, Ñ7, Reaktortagung Bonn, 1971, Deutsches Atomforum. [49] GARZAROLLI, F., TRINKL, A ., FRANCKE, K .P ., Proc. Reaktortagung Berlin, 1970, Deutsches Atomforum, p .525. [50] JOHNSON, C .E ., JOHNSON, I ., CROUTHAMEL, C .E ., Nucl. Sei. Eng. (1971). [51] HUBER, H ., KLEYKAMP, H ., Rep. KFK 1324 (1972). [52] PERRY, K .J ., MELDE, G .E ., D UN CAN, R .N ., Fast Reactor Fuel Element Technology (Proc. Conf. . New Orleans, 1971), Trans. Amer. Nucl. Soc. 14 Suppl. 1 (1971) 17. [53] STALICA, N.R. , SEILS, C . A . , CROUTHAMEL, C .E ., Rep. ANL-7550 (1968). [54] COQUERELLE, M . et a l . , Reps TUSR-10 (1971); TUSR-13 (1972). [55] SARI, C ., BENEDICT, U ., BLANK, H ., J. Nucl. Mater. 35 (1970) 267. [56] BRAUER, G ., Z . Anorg. Chem. 255 (1947) 101. _ [57] ELLIOTT, R .P., Constitution o f Binary Alloys, First S uppl., M cGraw-Hill, New York (1965) 368, [58] KENDALL, P. W ., J. Nucl. M ater., 35 (1970) 41. [59] SPITSYN, v . l . , KULESHOV, I .M ., J. Gen. Chem. USSR 21 (1951) 445. [60] BLACKBURN, P.E., JOHNSON, P.E., BATTLES, C. E., JOHNSON, J.E., MARTIN, A. E., TETENBAUM, M ., CROUTHAMEL, C.E., TEVEBAUGH, A.D ., VOGEL, R.C., Rep. ANL-7822 (1971). [61] CROUTHAMEL, C .E ., JOHNSON, C .E ., Rep. ANL-7833 (1971). [62] EVANS, S.K., AITKEN, E.A., 74th Ann. Meeting Amer. Ceram. Soc. Washington, D.C., 1972. [63] CHEMICAL RUBBER PUBLISHING C O ., Handbook o f Chemistry and Physics (1971-1972) 52nd Ed. [64] SPITSYN, V .l., KULESHOV, I.M., J. Gen. Chem. USSR 21 (1951) 1717; BELYAEV, I.M., CHIKOVA, N . N . , J. Inorg. Chem. USSR 9 (1964) 1483. [65] ROLSTEN, R.F., Iodide Metals and Metal Iodides, John W iley, New York, London (1961). [ 66] SCHAFER, H ., HÖNES, W .J ., Z . Anorg. A llg. Chem. 288 (1956) 62. [67] KUBASCHEWSKI, О., EVANS, E.L., ALCOCK, C.B., Metallurgical Thermodynamics, Pergamon Press, Oxford (1967). [ 68] WICKS, C .E ., BLOCK, F .E ., Bur. Mines Bull. 605 (1963) 36, 85. [69] SPITSYN, V . l., Ed., ANL-Trans. 33, Publishing House Moscow Univ. (1961). [70] G Ö TZM A N N , O ., OHSE, R .W ., "Fuel/cladding compatibility o f stainless steels with gas and sodium-bonded uranium plutonium carbide fuels", these Proceedings, IAEA-PL-463/15. [71] M A IY A , P .S ., Rep. ANL-7825 (1971). [72] AITKEN, E .A ., Rep. GEAP-5683 (1968). 3 1 4 OHSE and SCHLECHTER [73] WEEKS, J.R ., KLAMUT, C .J., Corrosion of Reactor Materials (Proc. Conf. Salzburg, 1962) ^ IAEA, Vienna (1962) 105. [74] AITKEN, E .A ., EVANS, S . K ., ROSENBAUM, H .S ., RUBIN, B .E ., Ann. ANS Meeting Boston, 1971, Trans. Amer. Nucl. Soc. 14 (1971) 176. [75] KELLER, C ., KOCH, L., WALTER, K.H., J. Inorg. Nucl. Chem. 27 (1965) 1225. [76] AITKEN, E.A., EVANS, S .K ., Rep. GEAP-12099 (1970). [77] CORDFUNKE, E.H .P., LOOPSTRA, B.O., J. Inorg. Nucl. Chem. 33 (1971) 2427. [78] BLACKBURN, P .E ., MARTIN, A .E ., BATTLES, J . E . , O'HARE, P .A .G ., HUBBARD, W .N ., Fast Reactor Fuel Element Technology (Proc. Conf. New Orleans, 1971), Trans. Amer. Nucl. Soc. 14 (1971) 20. [79] О'BOYLE, D.R., BROWN, F.L., SANECKI, J.E., J. Nucl. Mater. 29 (1969) 27. [80] BEDFORD, R.G., JACKSON, D.D., Rep. UCRL-12314 (1965). [81] OLSEN, A.R., FITTS, R.B., СОХ, C.M ., Rep. ORNL-TM-2716 (1969). [82] DAVIES, J.H ., EWART, F.T., J. Nucl. Mater. 41 (1971) 143. [83] EDELEANU, C ., LITTLEWOOD, R., Electrochim. Acta 3 (1960) 195, [84] KOENIG, R .F ., VANDENBERG, S . R ., M et. Progr. 61 (1952) 71. [85] MENZIES, I.A ., Werkst. Korros. 19 (1968) 1050. [86] OHSE, R.W ., Z. Elektrochem., Ber. Bunsenges. physik. Chem. 64 (1960) 1171; 64 (1960) 1171. [87] CHASANOV, M.G., Nucl. Sei. Eng. 30 (1967) 310. _ ' [88] BORGSTEDT, H.U., FREES, G., Werkst. Korros. 21 (1970) 435. [89] LEITNAKER, J .M ., SPEAR, K .E ., Rep. ORNL-1140 (1969). [90] OHSE, ‘ R. W ., OLSON, W .M ., Proc. Reaktortagung Berlin, 1970, Deutsches Atomforum, p .570. [91] ZOTOV, V .V ., NEVZOROV, B.A ., UMNYASHKIN, Ye.V ., KIRILLOV, P.L., in Liquid Metals, Rep. NASA-TTF-522 (1969). [92] SCHAFER, H., Chemische Transportreaktionen, Verlag Chemie, Weinheim/Bergstrasse (1962). [93] OHSE, R.W ., SCHLECHTER, M ., ZA MORA NI, E., TUSR 10 (1970) 26. [94] OHSE, R.W ., POTTER, P.E., SCHLECHTER, M., TUSR 12 (1972) 30. [95] ZAUGG, W .E ., GREGORY, N .W ., J. Phys. Chem . 70 (1966) 490. [96] ZAUGG, W .E., GREGORY, N.W., J. Phys. Chem. 70^(1966) 486. [97] SIMON, R .H ., LINDGREN, J .R ., SILTANF.N, J .N ., FIT TS, R .B ., Rep. G A -10262 (1970). [98] LEITNAKER, J .M ., FITTS, R .B ., CUNEO, D .R ., LONG, E .L ., SPEAR, K .E ., Rep. ORNL-452 (1970). [99] JAVED, N.A., ROBERTS, J.T .A ., Rep. ANL-7901 (1972). [100] LEVIN, E.M ., McMURDIE, H .F., HALL, F.P., Phase Diagrams for Ceramists, Amer. Cer. Soc. (1956) 36. [101] DETTINGMEIYER, J.H ., TILLACK, J., SCHÄFER, H., Z. Anorg. Allg. Chem. 369 (1969) 161. [102] TEDMON, C .S., VERMILYEA, D.A., ROSOLOWSKI, J.H ., J. Electrochem. Soc. И8 (1971) 192. [103] KEROULAS, F., LE BEUZE, R., CALAIS, D., VAN CRAEYNEST, A., CONTE, M., J. Nucl. Mater. 43 (1972) 313. [104] MAIYA, P.S., BUSCH, D .E., Reps ANL-7833 (1971); ANL-7854 (1971). [105] MAIYA, P.S., BUSCH, D.E., SANECKI, J.E ., Rep. ANL-RDP-3, p.5.13. IAEA-PL-46 3/19á MISE EN EVIDENCE DE LA MIGRATION DE L ' OXYGENE SOUS IRRADIATION: IRRADIATION L -1 M. TROTABAS, F. de KEROULAS, J.P. GATESOUPE CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Roses, France Abstract-Résumé ANALYSIS OF IRRADIATION-INDUCED OXYGEN MIGRATION IN THE L -l EXPERIMENT. One of the most important factors in the choice of a nuclear fuel is the O/U + Pu ratio, which governs most of the physico-chemical properties, such as thermal conductivity, vapour tension, oxygen potential, compatibility with cladding and so on. Rand and Markin have proposed a model in which the oxygen gradient in the fuel under irradiation is controlled by a constant CO/COZ or H2/H20 ratio within the fuel pin. This model was supplemented by Aitken, who also took oxygen transport in the solid and by the vapour phase above the mixed oxide into account. The purpose of the L-l irradiation was to analyse the radial redistribution of oxygen resulting from the thermal gradient set up during irradiation and to check the validity of these models. MISE EN EVIDENCE DE LA MIGRATION DE L'OXYGENE SOUS IRRADIATION: IRRADIATION L -l. L'un des facteurs les plus importants dans le choix d'un combustible nucléaire est lè rapport O/U + Pu, qui conditionne la plupart des propriétés physico-chimiques (conductibilité thermique, tension de vapeur, potentiel d'oxygène, compatibilité avec la gaine, etc.). Rand et Markin ont proposé un modèle dans lequel le gradient d'oxygène dans le combustible sous irradiation est contrôlé par un rapport C 0/ C 02 ou H2/H20 constant à l'intérieur de l'aiguille combustible. Ce modèle a été complété par Aitken en tenant compte en plus d'un transport de l ’oxygène dans le solide et par la phase vapeur au-dessus de l'oxyde m ixte. Le but de l'irradiation L-l était d'essayer de mettre en évidence la redistribution radiale de l'oxygène sous l'effet du gradient thermique en cours d'irradiation, et de vérifier la validité de ces modèles. I - INTRODUCTION L'un des facteurs les plus importants dans le choix d'un combus tible nucléaire est le rapport O/U + Pu qui conditionne la plupart des pro priétés physico-chimiques ( conductibilité thermique, tension de vapeur, potentiel d’oxygène, compatibilité avec la gaine, e tc.. .). RAND et MARKIN £ l J ont proposé un modèle dans lequel sous irradiation le gradient d'oxygène dans le combustible est contrôlé par un rapport CO/CO2 ou H2 /H 2 O constant à l'intérieur de l'aiguille combustible. Depuis ce modèle a été complété par AITKEN en tenant compte en plus d'un transport de l'oxygène dans le solide et par la phase vapeur au- dessus de l'oxyde mixte. Le but de l'irradiation L -l était d'essayer de mettre en évidence la redistribution radiale de l'oxygène sous l'effet du gradient thermique en cours d'irradiation, et de vérifier la validité de ces modèles . 315 3 1 6 TROTABAS et al. TABLEAU I. CARACTERISTIQUES DES DIVERS CRAYONS ET DE L 1 IRRADIATION. (Diamètre des pastilles: 5, 50 mm; jeu oxyde-gaine: 0, 1 mm sur le diamètre) N° crayon % densité O/U + Pu Puissance linéaire théorique en paiier (W/cm ) L -ll 96, 8 2,03 + 0,01 290 L-12 95,9 1,93 + 0,01 330 L-13 86,2 2,04 + 0,01 360 L-14 83,3 1,93 +0,01 360 2 - METHODES EXPERIMENTALES 2-1 - Conditions d'irradiation Le combustible choisi est un oxyde mixte de rapport Pu/U + Pu=0,2 dont les caractéristiques sont rassemblées dans le tableau I . L'irradiation a été effectuée dans le réacteur EL-3^ en neutrons ther miques. Les conditions ont été choisies de manière à éliminer tout effet ris quant de masquer la redistribution de l'oxygène : - La durée de l'irradiation est courte ( 40 heures ) de manière à ne pas devoir tenir compte de l'influence des produits de fission formés sous irradiation. - L'irradiation a été arrêtée par chute de barres ( effet de trempe ) afin d'éviter une éventuelle homogénéisation de l'oxygène en cours de refroi dissement . Le déroulement de l'irradiation a été le suivant : - une montée rapide à 200 W /cm , - puis une montée lente à la vitesse de 30 W/cm par heure, jusqu'à la puissance maximale . 2-2 - Examens après irradiation Après irradiation les crayons ont subi les examens suivants : - une neutrographie', - des coupes micrographiques, - des examens à la microsonde de Castaing, - des examens par diffraction X sur des microprélèvements en fonction du rayon, afin de déternlner, à partir du paramètre, réticulaire, le rapport O/U + Pu local. IAEA-PL-46 3/19a 3 1 7 3 - RESULTATS 3-1 - Echantillons surstoechiométriques 3-1-1 Craypn L-П La coupe micrographique effectuée sur ce crayon d'oxyde dense et surstoechiométrique ne met en évidence qu'un léger grossissement de grains au centre. Une autoradiographie a de cette coupe ne montre aucune hétéro généité de plutonium. Les résultats des examens par diffraction de rayons X, sur des microprélèvements pris sur deux coupes radiales, mettent en évidence un faible gradient d'oxygène ( tableau. II ) . 3 -1 - 2_ - J^ra_Y_qn_ L -13 Le combustible de ce crayon est un oxyde surstoechiométrique de faible densité, dont la température à coeur en cours d'irradiation a été plus forte que pour le crayon précédent. La neutrographie met en évidence l'existence d'un trou central de section irrégulière qui va en s'élargissant vers les extrémités des pas tilles. Ce trou est bordé d'un liseré, signe d'un enrichissement en plutonium. Les coupes micrographiques longitudinales et transversales confir ment l'irrégularité du trou central. FIG. 1. Oxyde Danielle L-13 n° 2. TABLEAU II. RESULTATS DES EXAMENS PAR DIFFRACTION X EFFECTUES SUR MICROPRELEVEMENTS. PARAMETRES RETICULAIRES ET O/U + Pu DEDUITS (a2 + x = a2 00- 0,12 x) POUR LES ECHANTILLONS S U RS T OE CHIOME T RIQ UE S. (R rayon de la pastille) № crayoi 1 Echantillon Position des Paramètre a e n ï O/M déduit Température en °C prélèvements (r/R) ( calculée avec O/U+Pu moyen ) centre (0*05-0,21) 5,4511+0,0006 2,031 1500/1480 intermédiaire 1 5,4521+0,0007 2,023 1500/1440 (0,09-0,34 ) 11 1 intermédiaire 2 5,4527+0,0008 2,018 1370/1170 (0,47-0,68 ) bord (0,60-0,81) 5,4532+0,0005 2,013 1250/1000 échantillon à moyenne 5,4548+0,0005 2,094 0/U+Pu=2,00 centre (0,12-0,17,) 5,4435+0,001 2,000 2130/2120 1 Intermédiaire 5,4534+0,0008 2,010 2080/1960 (0,24-0,52) bord (0,72-1) 5,4547+0,001 2,000 1660/850 centre (0,12-0,18) 5,4478+0,001 2,058 2130/2120 13 2 intermédiaire 5,4533+0,0008 2,013 2070/1830 (0,35-0,64) bord (0,71-1) 5,4534+0,001 2,012 1690/850 centre (0,12-0,15) 5,4598+0,001 2,128 2130/2120 3 intermédiaire 5,44 82+0,001 2,055 2030/1690 (0,43-0,71) bord (0,75-1) 5,453 8+0,001 2,008 1580/850 échantillon global 5,4548+0,001 2 ,00 ramené à 2 , OC IAEA-PL-463/19a 3 1 9 Un examen par microanalyse X met en évidence un enrichissement en plut onium sur une largeur d*environ 300 jum autour du trou central (fig. 1). Les résultats des mesures de diffraction X prouvent l'existence d'un gradient du rapport Q/U + Pu,variant d'environ 2,00 à 2,13 du bord au centre des échantillons, sans tenir compte de la migration du plutonium ( tableau II). Il convient de remarquer que les deux phénomènes causant une diminution du paramètre réticulaire , les valeurs ainsi trouvées pour O/U + Pu le sont par excès„ 3-2- Echantillons sous-stoechiométriques D'une façon générale la mise en évidence du gradient d'oxygène sur ces échantillons a été rendue difficile du fait de l'oxydation lors des microprélèvements et des examens. 3 - 2 -1 Crayon L - 12_ L'oxyde de ce crayon de forte densité avait une stoechiométrie initiale de 1, 93 . Les examens micrographiques montrent un faible grossissement de grains ainsi que la précipitation fine d'une seconde phase à coeur. La nature de ces précipités n'a pu Stre déterminée. Aucun gradient de plutonium n'a été trouvé parautoradiographie a . Les résultats des analysés par diffraction X ( tableau III ) prouvent, malgré une oxydation en cours d'examen et un biphasage pouvant Stre dû à cette oxydation,ou à la taille importante des microprélèvements (~ 1 mm), de la périphérie au centre des échantillons^ 1'existence d'une variation du rapport O/U + Pu d'environ 1,90 à 1 ,9 5 . 3_-2_-2 - Crayon_L-14 Comme dans le cas de l'aiguille L -13, la température atteinte à coeur de ce crayon d'oxyde de faible densité est supérieure à celle du crayon L -1 2 . Les examens micrographiques mettent en évidence des précipités métalliques vers le bord du trou central. Une analyse à la microsonde de Castaing prouve que ces précipités sont uniquement constitués d'uranium et de plutonium et ont le même rapport Pu/U + Pu que la matrice. Par ailleurs aucune migration de plutonium ne semble avoir eu lieu. Les mesures de paramètre effectuées sur des microprélèvements montrent que, maigre une réoxydation certaine, il subsiste un gradient d'oxy gène dans l'échantillon, le rapport O/U + Pu ne dépassant jamais 1,98 en périphérie. TABLEAU III. RESULTATS DES EXAMENS PAR DIFFRACTION X EFFECTUES SUR MICROPRELEVEMENTS. PARAMETRES RETICULAIRES ET O/U + Pu DEDUITS (a2. x = a 2,oo+ 0. 345 x) POUR LES ECHANTILLONS SOUS-STOECHIOMETRIQUES. (R = rayon de la pastille) № crayon Température en °C Echantillon Position des ч о prélèvement (r/R) Param étré a en A O/M déduit ( calculée avec O/U+Pu moyen ) t 5,4887+0,001 ( 1,903 centre (0-0,15) 1900/1880- 1 V 5,4598*0,001 (. 1,987 bord (0,84-0,9 5) 5,4729TO,001 1,949 1190/940 centre (0-0>15 5,4 886+0,001 1,903 1900/1880 2 intermédiaire 12 5,4895+0,001 1,901 1840/1630 (0,25-0,55) ( 5,4886+0,01 1,903 3 global < 5,4782+0,01 1,954 1900/730 1 5,4640^,01 1,975 ramené à2,00 global 5,4553+0,000 7 2,00 centre (0,30-0,50) f 5,4766+0,001 ( 1,950 2410/2290 j 5,4638T0,001 j 1,977 1 intermédiaire 5,4627+0,001 1,981 2290/2030 (0,50-0,70 bord (0,76-1) 5,4644+0,001 1,976 1860/1000 13 centre (0,28-0,65) ( 5,4 863+0,002 f 1,912 2429/2110 \5,4646+0,001 l 1,975 2 intermédiaire r 5,4696+0,001 ( 1,961 2290/1540 (0,57-0,85 ) \5,4616+0,001 l 1,984 bord (0,65-1) 5^4627+0,001 1,981 1610/1000 ramené à 2,00 global 5,4560+0,001 2,00 IAEA-PL-46 3/19a 321 \ FIG. 2. Variation radiale de température et de stoechiométrie dans les crayons L-12 et L-14. a Température calculée avec le O/M moyen. b Température calculée compte tenu de la redistribution de l'oxygène. + Points expérimentaux. 4 - DISCUSSION 4-1 - Profils de températures (fig. 2 et 3 ) La détermination de la redistribution de l'oxygène en fonction de la température nécessite la connaissance du profil radial des températures dans l'oxyde mixte. Le calcul de ce profil a été fait à partir des données du tableau I, en tenant compte du creusement de flux, du coefficient .de transfert oxyde- gaine ( h = 0, 5 ^ 3 de la conductibilité thermique de l'oxyde ( fonction de O/U + Pu); la densité et la température et en prenant comme tempé rature de début des grains colonnaires 1950° C ¿~5 J . 322 TROTABAS et al. FIG. 3. Variation radiale de température et de stoechiométrie dans les crayons L-13 et L -ll. — Température calculée avec le O/M moyen. - - O/M expérim ental. Il a été effectué, d'une part sans tenir compte de la redistribu- tion d'oxygène, d'autre part en tenant compte de cette redistribution et en adoptant le modèle proposé par AITKEN £ 2J„ dans le cas des échantillons L-12 et L-14. Selon cet auteur,dans les échantillons sôus-stoechiométriques ( U, Pu) C>2 _x/ à l’état stationnaire, l'écart local à la stoechiométrie, x, est donné par la formule : où : Q = chaleur de transport de l'oxygène en calories R =. constante des gaz parfaits T = température au point considéré en °K. Pour nos échantillons nous avons pris : Q* = -5,6 kcal A = 0,4 IAEA-PL-46 3/19a 3 2 3 4-2 - Interprétation des résultats 4 -2 -l_ - Echantillons_squ^stoechiométrique_s Si l*on considère que les valeurs les plus probables des para mètres réticulaires après irradiation correspondent aux valeurs les plus éle- vées; on voit sur la figure 2 que dans le cas de l'échantillon L-12. les résul tats expérimentaux sont en assez bon accord avec les prévisions d'AITKEN. . Dans le cas de l'échantillon de faible densité L-14, ce type de calcul amenant à un rapport O/U + Pu à coeur de l'ordre de 1, 85 explique ; la précipitation de métal observée par micrographie, la limite d'existence de la phase ( U,Pu) Û2 -x à température ambiante étant voisine de 0/U +Pu=l,90. Il est regrettable que les échantillons aient subi une oxydation lors des micro prélèvements, car sinon nous aurions pu obtenir des précisions intéressantes sur le profil d' oxygène et la limite d'existence de la phase (U,Pu)0 2 _x en fonction de la température. 4-2-2 - Echantillons_surstoechiométriques Dans ce cas nous n'avons pas pu effectuer un calcul des tempé ratures tenant compte de la redistribution de l'oxygène, la variation de la conductibilité thermique et la chaleur de transport de l'oxygène, Q*, étant moins bien connues. Nous avons simplement admis que la conductibilité thermique pour un rapport O/U + Pu = 2 ,0 3 -2 ,0 4 , était équivalente à celle d'un oxyde de rapport O/U + Pu = 1 ,9 8 . Dans le cas de l’échantillon L -ll, dont les températures à coeur et en surface n'ont pas dépassé respective ment 1500 et 700° C, la migration de l'oxygène est très faible et bien infé rieure à celle prévue par RAND et MARKIN £ l J . Pour l'échantillon L-13 de faible densité, où la migration de plu tonium observée conduit à des valeurs par excès pour les teneurs en oxygène au centre, le gradient d'oxygène est nettement visible, mais toujours inférieur aux prévisions des précédents auteurs. Ces résultats sont à rapprocher de ceux de JEFFS 6_7 qui a observé un effet de redistribution de l'oxygène sensiblement du mSme ordre que le nStre pour un échantillon ( U,Pu) O2 94 . 5 - CONCLUSION Malgré les difficultés rencontrées pour effectuer des prélèvements en fonction du rayon et leur taille importante (- 1 mm de diamètre:); nous avons pu vérifier la migration de l'oxygène sous irradiation. L'allure observée est celle prévue par RAND et MARKIN, mais les valeurs absolues sont moins fortes que celles annoncées. Dans le cas des oxydes fortement sous-stoechiométriques nous pensons que le modèle proposé par AITKEN représente assez bien le phénomène observé. 3 2 4 TROTABAS et al. Par contre dans le cas des oxydes surstoechiométriques il devient nécessaire de tenir compte des phénomènes de vaporisation-condensation au. delà de 2000° C . Il faut enfin signaler que la méthode de mesure du rapport O/U + Pu à partir de la valeur du paramètre réticulaire est relativement imprécise et peut expliquer des écarts avec les modèles théoriques. Ceci est surtout vrai dans le cas des oxydes surstoechiométriques où la variation du paramètre réticu laire en fonction de la teneur en oxygène est faible. Il conviendrait donc à l'avenir d'effectuer des mesures directes, par thermogravimétrie, sur des pré lèvements plus petits. REFERENCES [ 1 ] RAND, M .H ,, MARKIN, T . L. , AERE R-5560 (1967). [2] AITKEN, E.A., J. Nucl. Mater. 30 (1969) 62-73, [ 3] MIKAILOFF, H ., Communication personnelle. [ 4 ] WEILBACHER, J. C. , Thèse, Paris (1972). [ 5] MIKAILOFF, H ., Communication personnelle. [ 6] JEFFS, A. T. , AECL 3690 (1970). IAEA-PL-463/19b THERMODIFFUSION ET VAPORISATION DANS LES OXYDES M IXTES IRRADIES - EVOLUTION DU RAPPORT Pu/U+ Pu M. MOUCHNINO CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Roses, France Abstract-Résu mé THERMODIFFUSION AND VAPORIZATION IN IRRADIATED MIXED OXIDES - EVOLUTION OF THE Pu/U+Pu RATIO. The distribution of the Pu/U + Pu ratio in nuclear fuels is a determining factor for important parameters such as the spatial power distribution and the Doppler coefficient of reactivity. X-ray autofluorescence, a method of spectrographic analysis, makes it possible to measure the Pu/U+Pu ratio in favourable conditions on a large number of oxide samples irradiated in very different conditions. THERMODIFFUSION ET VAPORISATION DANS LES OXYDES MIXTES IRRADIES - EVOLUTION DU RAPPORT Pu/U+Pu. La répartition du rapport Pu/U + Pu dans les combustibles nucléaires est déterm inante pour des paramètres aussi importants que la distribution spatiale de la puissance et le coefficient Doppler de réactivité. Une méthode d’analyse spectrographique, l’autofluorescence X, permet de mesurer dans de bonnes conditions c e rapport Pu/U + Pu sur un grand nombre d'échantillons d'oxydes, irradiés dans des conditions très différentes. INTRODUCTION La répartition du rapport Pu/U + Pu dans les combustibles nuclé aires est déterminante pour des paramètres aussi importants que la distribu tion spatiale de la puissance et le coefficient Doppler de réactivité. Une méthode d'analyse spectrographique, l'autofluorescence X , permet de mesurer dans de bonnes conditions ce rapport Pu/U + Pu sur un grand nombre d'échantillons d'oxydes, irradiés dans des conditions très différentes £ _ \ J . Pour exploiter quantitativement les résultats, trois types de modèles sont actuellement proposés £ , l j \ - la diffusion thermique où une seule phase solide est considérée; - l'évaporation condensation différentielle où deux phases, solide et gazeuse,sont considérées, la deuxième phase étant constituée de pores ou de bulles ouverts ou fermés ; - la fusion non congruente où deux phases, solide et liquide, sont considérées. 3 2 5 3 2 6 MOUCHNINO Le premier modèle de thermodiffusion nécessite seulement la connais sance de trois paramètres définissant la diffusion chimique et thermique. C'est le plus simple. Il est vraisemblablement le plus précis pour le domaine de température compris entre 2200° C et la température de fusion. Le second opère entre 1600° et 2300° C; au-dessus les données thermodynamiques sont difficiles,à mesurer hors pile. Le troisième modèle n'intéresse que la classe des combustibles fonctionnant à coeur fondu. Nous donnons la description du modèle de thermodiffusion. L'application de ce modèle ( 1ère partie des résul tats ) n'est possible que lorsque les températures de surface et à coeur de l'oxyde sont connues en fonction du temps d'irradiation. Ce n'est pas toujours le cas; l'irradiation pour des taux de combustion élevés étant fort complexe à cause des cyclages et le contact oxyde-gaine variant, on ne peut faire que des hypothèses sur les coefficient d'échange oxyde-gaine et sur les valeurs cherchées de la température. Aussi dans la deuxième partie des résultats ne donnerons-nous que lesvaleurs expérimentales du rapport Pu/U + Pu mesu rées directement par la méthode d'autofluorescence X. II - MESURE RADIALE DU RAPPORT Pu/U + Pu PAR LA METHODE D'AUTOFLUORESCENCE X La mesure de l'intensité des raies Ka émises spontanément par l'uranium et le plutonium d'un combustible irradié permet la détermination précise du rapport Pu/U + Pu. La détection des photons X est faite à l'aide d'un détecteur solide Ge Li à fenStre mince et à préamplificateur refroidi. L'échantillon est une section droite, d'épaisseur constante, préalablement polie. Le rayonnement est limité par un collimateur de surface 0,3x0, 5 mm (ou 0, 5 x 0,5 mm ). Un système mécanique permet le déplacement de l'échan tillon par pas de 0, 1 mm. On obtient , le long d'un diamètre arbitraire de l'échantillon, les valeurs relatives de la concentration en uranium et en plu tonium. Le rapport de ces valeurs relatives donne la valeur absolue du rap port Pu/U + Pu. La comparaison des résultats ainsi obtenus sur des échantillons homogènes avec les valeurs globales données par l'analyse chimique conduit à une précision de la méthode de 5 %. Cette méthode est limitée par l'émis sion spontanée assez faible qui entraîne des erreurs statistiques de comptage importantes dès que le niveau de comptage est bas. III - ECHANTILLONS UTILISES La liste des échantillons utilisés est portée sur le tableau I avec leurs caractéristiques d'origine ( les valeurs du rapport moyen O/M, la densité initiale), le taux de combustion, la puissance linéaire moyenne. Ils ont été séparés en trois groupes en fonction des réacteurs où’s'est déroulée l'irradiation : Rapsodie (neutrons rapide^, EL 3-(neutrons. thermiques), Osiris (neutrons épithermiques ). Il y a un large échantillonnage, selon les caractéristiques propres des échantillons d'une part, selon l'irradiation elle-même d'autre part. Notons que les échantillons M appartiennent à un lot d'oxydes avec addition de produits de fission non radioactifs et scumis à une faible irradiation dans EL 3 ¿f3 J. IAEA-PL-463/19b 3 2 7 TABLEAU I. TABLEAU DES PRINCIPALES CARACTERISTIQUES D'ECHANTILLONS D'OXYDE O/M d Z Puissance Remarques % dth MWj/t W/cm A 1,99 86,4 12 000 540 540 572 Rapsodi& B 1,97 94 12 000 508 508 534 3 cycles C 1,96 94 12 000 480 480 513 -GO-61 1,93 96 50 000 450 GO-63 ' 1,93 94 50 000 3 94 Entrée sodium après rupture Osiris gaine à mi-irra diation GO-21 2,00 87 100 000 429 - GO-23 2,00 93,5 100 000 440 S -ll 2,00 93 6 000 450 EL 3 S-23 2,00 96,5 6 000 525 Noter la sous- S-24 1,90 82 6 000 450 stoechiométrie très basse. IV - RESULTATS IV - 1 - Echantillons irradiés dans Rapsodie IV-1-1 - Description qualitative des résultats Les trois échantillons montrent trois allures différentes de la répar tition du plutonium en fonction du rayon du combustible ( fig. 1 ). - A montre un décrochement du rapport Pu/U + Pu, sur une couronne entourant le trou central , de 0,5 mm d'épaisseur. L'augmentation du rapport Pu/U + Pu est égale à 0,05.. - C montre une distribution constante. - Sur B on observe un profil caractéristiques de la migration du plutonium. La courbe a un maximum situé près du trou central, avec une valeur de Pu/U + Pu égale à 0 ,3 0 0 . Le minimum du rapport Pu/U+ Pu vaut MOUCHNINO FIG. 1. Répartition diam étrale de Pu/U+Pu sur les échantillons A, В, C. IAE A-PL-4 6 3 /19Ь 3 2 9 0,220 . En périphérie le rapport vaut 0 ,2 4 0 . Sur le diamètre de l'échantillon on peut noter une répartition non symétrique. En particulier, le gradient Pu/U + Pu est plus élevé sur la partie gauche. IV-1-2 - Description qualitative du profil de B. Application du modèle de thermodiffusion ( Effet Sorret ) /~4J a donné le principe du calcul qu'on rappelle brièvement. Le flux de plutonium est composé de deux termes; — * ______> p D , . _____ > JPu = - DPu (grad CPu + v grad T) RT2 DPu coefficient de diffusion chimique CPu concentration molaire du plutonium Q+ chaleur de transport du plutonium et le bilan matière en un point s'écrit : ^ CPu = - div JPu Ъ t Connaissant la concentration initiale et les conditions aux limites ( flux nuls aux extrémités ) on calcule CPu ( R, T ). Les données sont celles de Z~5_7 pour DPu = D„exp - Q/R T . Do = 0,34 cm^/sec Q =97 kcal/ mole et de /~4_7 pour Q+ qui vaut 3 7 kcal/mole. A partir de la puissance et des coefficient,de transfert oxyde-gaine les températures â coeur et en surface ont été évaluées /~ 6_7 en fonction du temps: 1 T max = 2700° C Tmin = 1150° C temps de maintien = 10 jours 2 T max = 2600° C Tmin = 1075° C temps de maintien = 30 jours 3 T max = 2500° C Tmin = 1000° C temps de maintien =40 jours Sur la figure 2 on donne à la fois les courbes calculées et mesurées qui sont en bon accord. IV - 2 - Résultats qualitatifs en neutrons épithermiques et thermiques IV-2-1 - Effet du creusement de flux Dans le cas des irradiations faites à Osiris,le creusement de flux à l'intérieur de l'échantillon fait que de manière indépendante de toute migra tion du plutonium le rapport Pu/U + Pu est maximum au centre de l'échantillon 330 MOUCHNINO TABLEAU II. VALEURS DE Pu/U + Pu APRES IRRADIATION Taux de combustion Milieu I Milieu 2 MWj/t 0 <. R <• 1, 5 mm 1,5 < R <2,5 mm 0 0,200 0,200 50 000 0,183 0,160 100 000 0,138 0,119 ( ou à proximité du trou central ). Il est nécessaire, pour mettre en évidence une éventuelle migration du plutonium,de calculer les valeurs radiales de Pu/U + Pu,en fin d'irradiation, résultant du creusement de flux. Le programme Rapace /~7_7 fournit deux points de la c ourbe cher chée, qu'on assimile ici, en première approximation, à une droite passant par ces deux points. Les résultats sont donnés sur le tableau II et sur la figure 3 . IAEA-PL-463/19b 331 FIG. 3. Calcul Rapace valeurs PuAJ +Pu après irradiation — calcul à 2 groupes. Cr O ô i Q 0-6 3 FIG. 4. Répartition diamétrale de Pu/U + Pu sur les échantillons GO-61, GO-63. 3 3 2 MOUCHNINO Connaissant la valeur initiale du rapport Pu/U + Pu et la valeur mesurée au bord de l'échantillon, on obtient d'après les résultats précé dents la valeur de Pu/U + Pu au centre de l'échantillon . IV-2-2 - Résultats qualitatifs a ) Echantillons GO-61, GO-63 (fig . 4 ) Sur l'échantillon GO-61, partie droite du diamètre balayé on obtient l'effet pur du creusement de flux. Sur la partie gauche il y a de forts écarts par rapport à la courbe prévue : - à la périphérie, avec un enrichissement en plutonium - en F un appauvrissement . Sur l'échantillon GO-63 on ipet en évidence un effet marqué de la migration du plutonium : rappelons ( tableau I ) que l'échantillon a subi la même irradiation que GO-61 quoique à puissance linéaire moyenne infé rieure. Le crayon GO-63 s'est rompu à mi-irradiation. Micrographiquement J on n'observe pas de zone de réaction oxyde-sodium, mais on peut sup poser son existence d'après la valeur résiduelle du jeu oxyde-gaine de 100/um supérieure à la valeur initiale de 80/nm, la zone de réaction ayant disparu lors du polissage de l'oxyde. L'effet obtenu sur la répartition du plutonium est donc vraisemblablement lié à cette zone de réaction à la suite d'un échauf- fement anormal. IAEA-PL-463/19b 3 3 3 FIG. 6. Répartition diamétrale de PuAJ +Pu sur les échantillons M:S-11, S-23, S-24. 334 MOUCHNINO b ) Echantillons G-21, G-23 (fig. 5 ) Les profils de migration du plutonium sont plus accentués sur l'échan tillon GO-21. c ) Echantillons S-ll, S-23, S-24 (fig,. 6 ) Les répartitions sur S -ll et S-23 montrent une forte remontée de la teneur en plutonium près du trou central. On n'observe pas le minimum corres pondant des courbe précédentes. Pour S-24, très sous-stoechiométrique, on obtient un appauvrissement marqué vers le trou central et un maximum inter médiaire. C'est donc une répartition inverse des précédentes. V - INTERPRETATION L'ensemble des résultats montre l'effet des trois mécanismes envisagés pour la migration du plutonium. Le cas 1 de la thermodiffusion est celui le plus fréquemment rencon tré ( échantillon B, GO-63, GO-21, GO-23 ) quoiqu'on ne puisse pas affirmer qu'il soit toujours pur, c'est à dire indépendant du cas 2 de vaporisation con densation. Le modèle de thermodiffusion donne une bonne description des résultats obtenus sur l'échantillon B. Cette expérience confirme la validité des températures calculées ainsi que les valeurs D et Q+ qui interviennent dans le modèle. Un cas de thermodiffusion pur semble Ôtre celui de GO-63 . On a en effet un minimum bien accentué, comme sur la courbe théorique de la figure 2 . L'évaluation d'une température moyenne à coeur donne 2600° C pour GO-63 ; cette température moyenne n'excède pas 2200° C pour GO-61. Sur les courbes de GO-21 et GO-23, le minimum est par contre très peu accentué. Dans ces cas il est possible qu'un phénomène de vaporisation condensation soit intervenu pour enrichir en plutonium préférentiellement les zones les plus chaudes et appauvrir simultanément en plutonium les zones froides ( bas descolonnaires) . Seule une analyse quantitative de courbes permettra de trancher. La superposition des mécanismes de vaporisation et de thermodiffu sion est mise en évidence sur l'échantillon M -24. Dans ce cas l'oxyde est très sous-stoechiométrique ( O/M ^1,90 ) et la température de l'échantillon est certainement très élevée pendant la durée relativement courte de l'irradia tion . Oft a vu que la répartition de Pu/U + Pu était inversée par rapport aux précédentes. Dans œ cas,en effet, le plutonium £9 J est bien vaporisé pré férentiellement dans les parties les plus chaudes et donne lieu à une diminution de Pu/U + Pu. Cet effet est rattrapé par la thermodiffusion dans les parties inter médiaires du rayon et permet d'obtenir le maximum observé. Le cas de la vaporisation "pure" est observé sur S -ll et S-13. Au-dessus des oxydes stoechiométriques, l'oxyde d'uranium UOg majoritaire dans la phase vapeur donne lieu- près du trou central à un enrichissement en plutonium. Le IAEA-PL-463/19b 3 3 5 minimum correspondant n'est pas observé sur la courbe expérimentale. Il n'y a donc pas de thermodiffusion associée. Enfin le cas 3 de fusion non congruente est observé sur l'échantillon A avec un palier enrichi caractéristique. CONCLUSIONS Les mesures du rapport Pu/U + Pu mettent en évidence les trois méca nismes de transport du plutonium dans les oxydes irradiés. Seuls les calculs de thermodiffusion ont été entièrement exécutés. Ils ont permis de décrire quanti tativement la courbe expérimentale obtenue sur un des échantillons. On observe qualitativement sur les autres échantillons : - l'effet de vaporistion pure, - l'effet de vaporisation et de thermodiffusion simultanées, - l'effet de fusion. A l'avenir,des irradiations en neutrons rapides et une connaissance plus précise des phases vapeur au-dessus des oxydes irradiés devrait permettre de mieux faire la part respective des divers processus de redistribution du plutonium. REFERENCES [ 1] MOUCHNINO, M. , C om ité des Laboratoires Chauds d'Euratom, Petten (1972). [2 ] AITKEN, E.A . , ADAMSON, M .G ., EVANS, S .K .. LUDLOW, T..E. , GEAP 12 229 (1971). [3] CONTE, M ., MOUCHNINO, M ., SCHMITZ, F., Congrès ANS (Richland, 1972). [4] SCHUMACHER, G ., INR 4. 70. 27 ( 1970). [5 ] SCHMITZ, F ., LINDNER, R ., Radiochim. Acta 1 (1963) 218. [6] CARTERET, Y ., Communication personnelle. [7] ROUSSEAU, Communication personnelle. [8]. CONTE, M ., Communication personnelle. [9] RAND, M .H., MARKIN, T .L ., in Thermodynamics of Nuclear Materials, 1967 (Compt. Rend. Coll. Vienne, 1967) AIEA, Vienne (1968) 637. IAEA-PL-463/19с EMISSION DES GAZ DE FISSION PAR L'O XYD E D'URANIUM DANS LES ELEMENTS COMBUSTIBLES P. CHENEBAULT, R. DELMAS CEA, Centre d'études nucléaires de Grenoble, Grenoble, France Abstract RELEASE OF FISSION GASES BY URANIUM OXIDE IN FUEL ELEMENTS. Two main groups of methods have been used to evaluate the amounts o f fission gases released by uranium oxide fuel elements in "light water” and "heavy water - gas"' reactors: analysis of the gases released instantaneously during irradiation, and analysis of local concentrations in the irradiated elements. A synthesis of the results obtained in these two ways reveals correlations between the amounts of gas released and the main irradiation parameters: temperature, geometry and tim e. EMISSION DES GAZ DE FISSION PAR L’OXYDE D'URANIUM DANS LES ELEMENTS COMBUSTIBLES. Pour évaluer les quantités de gaz de fission émises par l ’oxyde d'uranium dans les conditions ou ce combustible est utilisé dans les réacteurs des filières «eau légère» et «eau lourde - gaz», deux groupes de méthodes ont été employés: l'analyse des gaz émis instantanément pendant l'irradiation, et l'analyse des concentrations locales dans les combustibles irradiés. La synthèse des résultats obtenus par ces deux voies fait apparaître des corrélations entre les quantités de gaz émises et les paramètres principaux de l’irradiation: températures, géométrie, temps. Une analyse de plus en plus rigoureuse du comportement dans les conditions d'utilisations en pile de l'oxyde d!uranium fritte autorise aujourd'hui l'élaboration de codes de calcul prenant en compte les phé«. nomènes qui conditionnent les performances et la durée de vie des éléments combustibles* Des contributions à cet objectif de quelques équipes américaines, о publiées récemment (1), montrent à la fois les progrès faits en quelques années et les difficultés du problème posé. En ce qui concerne la migration des gaz de fission, les chercheurs de Bettis Laboratory ont perfectionné les modèles décrivant le mouvement des bulles dans les gradients thermiques. Ainsi, H.R. Warner et F.A . Nichols (2), tenant compte d*observátions confirmées, introduisent des conditions d*interaction entre les bulles et les dislocations, jusqu*à une taille de 500 A environ et entre les joints de grains et les bulles de 3 3 7 3 3 8 CHENEBAULT et DELMAS о tailles comprises entre 500 et 5000 A. Le code BUBL-1, présenté par ces auteurs, fournit à la fois la fraction de gaz échappée du combustible et le taux de gonflement de celui-ci. Des confirmations expérimentales sont présentées. Le plus souvent, le modèle de diffusion simple de Booth (1957) continue d*®tre préconisé, malgré ses imperfections. Pour notre part, nous avons analysé et confronté plusieurs groupes de résultats expérimentaux obtenus par les équipes du CEA, notamment celles de Grenoble, au cours de la mise au point des éléments combustibles du réacteur EL, : nous avons ainsi obtenu les bases d*un 4 mode de calcul des quantités de gaz de fission sortis du combustible irradié dans des conditions géométriques et thermiques proches de celles qui sont utilisées dans les réacteurs modérés à l*eau. Les expériences exploitées sont de trois types : a) Des échantillons de petites dimensions sont maintenus dans un flux de neutrons à une température constante repérée par un thermocou ple; les gaz de fission dégagés sont entrafhés sur des pièges et analysés. Ces expériences (1963-1966) fournissent des informations sur lfeffet de la température, sur le dégagement de divers isotopes radioactifs (3) et la migration propre des halogènes précurseurs des gaz rares (4). b) A puissance nucléaire stable, les gaz de fission libérés dans un crayon combustible sont recueillis quantitativement et analysés (1965-1967) : on a étudié dTune part les proportions relatives des gaz de période courte en équilibre radioactif, d*autre part l*accumulation en fonction du temps des isotopes stables. L*utilisation des résultats obtenus n*est possible que grâce à la connaissance précise des températures qui ont été repérées par des thermocouples internes. Les sept expériences effectuées ont ainsi fourni des données thermiques précises sur l*oxyde d*uranium en géométrie de crayon et des corrélations entre la fraction globale de gaz dégagés, le temps et un état thermique donné (5) (6)» IAEA-PL-463/19с 3 3 9 85 с) Des profils de concentration des gaz stables, ou de Kr, bon traceur de ces gaz* ont été déterminés dans des échantillons de combus tible provenant d*une quarantaine de crayons irradiés à des taux de combustion compris entre 600 et 12 000 MWj/t d*U et des tempérâtures centrales estimées entre 1700 et 2600*C (7). Bien que dévaluation de ces températures soit souvent délicate, le très grand nombre dlanalyses pratiquées permet de tirer des conclusions directement utiles. 1 - IRRADIATION ISOTHERME DE PETITS ECHANTILLONS Rappelons les conditions expérimentales, décrites en détail dans une publication ancienne (3) : un échantillon d*UO naturel est placé en pile dans un dispositif chauffant maintenant une température fixe (limitée à 1850*C environ) ; les gaz de fission libérés sont entralhés par un cou rant d’hélium purifié vers une rampe de piègeage, puis analysés quanti tativement par spectrométrie y ; chaque expérience, comportant généra lement deux phases isothermes, dure 120 heures. Parmi les isotopes analysés systématiquement, nous discuterons particulièrement ici le 85m cas de Kr (période radioactive 4,4 heures), 1.1 - La fraction libérée : R ______Nb d'atomes dégagés par seconde______B Nb d*atomes créés par seconde à l*équilibre radioactif croft d*abord, puis se stabilise assez rapidement à une valeur d*équilibre F. La figure 1 présente les résultats relatifs à divers échantillons d*U02 fritté irradiés à des températures variables de 340 à 1850*C. 1.2 - Au cours du traitement, les échantillons sont soumis à une puissance 3 nucléaire faible, de quelques Watts par cm , insuffisante pour induire des tensions mécaniques amenant la rupture. Les groupements de points mis en évidence sur la figure 1 résultent des caractéristiques géométriques des classes d*échantillons considérés ; plus précisément une analyse 340 CHENEBAULT et DELMAS -тА- т О - о о L. * £ xtn к Ü- d (J) 0 h -2 g cm'3 n mm mm 10 г O 10,6 50 1,5 3 8 Л 9,8 1000 1,5 3 □ 10,6 10 8 8 о о о 8 о -3 о 1 0 -- % О о □ □ 10" 4- - □ Température (°С ) _|_ «00 600 800 1000 1200 1-400 1600 1800 FIG. 1. Echantillons dTJC^ frittes irradiés à des températures variables. détaillée montre que le paramètre responsable est le rapport S /v de leur surface réelle à leur volume : en première approximation, F est propor~ tionnel à S/V. Cet effet de la géométrie a des conséquences importantes dans la pratique» car, dans un élément combustible, la taille moyenne du frag° ment émetteur de gaz de fission dépend : IAEA-PL-463/19с 3 4 1 - de la valeur de la porosité ouverte, - pendant l'irradiation, des fragmentations qui résultent du gra dient thermique et des propriétés mécaniques locales, - à long term e, de la décohésion progressive des joints de grains due à l’accumulation de bulles de gaz (8), qui amène des régions de moins en moins chaudes du combustible à participer efficace ment à l’effet global. 1.3 - F ne dépend pas de la température pour T < 900*C environ. Le mécanisme principalement responsable de l’émission de gaz de fission à basse température est Rejection ou knock-out (9), également sensible au rapport S /v des échantillons. Compte tenu de l’état de fragmentation moyen du combustible dans les conditions dtulisation en pile, nous admettrons que la fraction 85m de Kr dégagée par éjection, Féj, dans le cas d’un combustible de densité supérieure à 93 % de la densité théorique, dont la porosité est _ 4 entièrement isolée de la surface, est égale à 10 . 1.4 - Aux températures supérieures à 900°C, F dépend de la température (au-delà de 1800*C les valeurs expérimentales indiquées sont surestimées à cause de l’évaporation du combustible, voir Réf. (3))t 2 - IRRADIATION DES CRAYONS A PUISSANCE CONSTANTE Les gaz de fission libérés par le combustible et rassemblés dans le volume libre du crayon sont recueillis pendant l’irradiation à puissance stable et analysés par spectrométrie y . 2.1 - A équilibre radioactif, rapidement obtenu pour certains isotopes 87 85m ( Kr, Kr), la fraction F est stable à l’intérieur d’un intervalle de temps de quelques jours, comme dans le cas des expériences isothermes (î 1.1) 2.2 » Ce rapport croît très lentement, au moins au-delà d’un certain taux de combustion ; ainsi le crayon "Compère 1" a été irradié, dans 3 4 2 CHENEBAULT et DELMAS TABLEAU I. FRACTIONS F RELATIVES A 85mKr T cen trale T aux de combustion T gaine (° C) 1C kdTTc estim ée F (M W j/t U) T s(W cm ^ ) Ce) -3 1970 570 3 1 .4 2080 5 .1 0 2880 530 2 9 ,5 1925 2 . 5 . 1 0 " 3 5490 530 2 9 ,5 1925 2 , 5 . 1 O-3 27230 530 2 8 ,4 1900 ' io'2 FIG. 2. Courbes expérimentales obtenues dans l’expérience Compère I aux taux de combustion inférieurs a 6000 MWj/t d*U. IAEA-PL-463/19C 3 4 3 un dispositif Cyrano comportant le repérage par thermocouple de la température centrale et un système calorimétrique de mesure de la puissance (10) ; les fractioreF relatives à ^ й г, mesurées à différentes phases de la vie du combustible, sont indiquées au tableau I. L’accroissement de F avec le taux de combustion est du même ordre pour les autres isotopes mesurés (^^Se, ^^Xe et ^K r), 2,3 Dans chaque prélèvement effectué on a observé que les fractions libérées F pour différents isotopes radioactifs étaient sensiblement pro portionnelles à leur période radioactive. La figure 2 présente différentes courbes expérimentales obtenues dans l’expérience Compère I (11) aux taux de combustion inférieurs à 6000 MWj/t d'U. Ainsi on peut é c r ir e ; A Fi = T7 [1] Xi = cte de désintégration de l'isotope i A dépendant de l’état thermique du combustible, de sa fracturation et de la durée de l’irradiation. 3 - GAZ DE FISSION DANS UO^ IRRADIE 3.1 - La figure 3 présente les corrélations existant entre la concentra. 85 tion locale en Kr, traceur des gaz stables, et le taux de combustion. Les échantillons dans lesquels ont été effectués les prélèvements et les analyses proviennent de crayons combustibles irradiés dans des conditions thermiques représentatives des réacteurs de puissance ; les points représentatifs ont été individualisés suivant 1 a microstructure : 3,1.1. « Les concentrations locales sont à peu près corrélées aux m icrostructures, ce qui traduit probablement le fait qu’elles dépen dent surtout de la température. 3 4 4 CHENEBAULT et DELMAS Taux de combustion (10 M W j/tU ) FIG. 3. Corrélations entre la concentration locale en 85Kr et le taux de combustion. 3,1.2 - Le domaine relatif aux zones de microstructures basalti ques est limité supérieurement par une concentration de l*ordre de 8.10^ 3 atomes par cm d'UO^ qui peut être attribuée à la température limite inférieure de la recristallisation basaltique ,soit environ 1800*C. IAEA-PL-463/19C 3 4 5 >j) ( Dimension Dimension apparent« des grains FIG. 4. Distribution radiale de la concentration. Dans le bas du domaine, les valeurs expérimentales les plus faibles mesurées relatives à des combustibles ayant approché la tem - 15 3 pérature de fusion au centre sont d’environ 5.10 atomes par cm . 3.2 - La distribution radiale de la concentration est représentée par les figures 4a et 4b. 3.2.1 - Les gaz de fission sont quantitativement retenus dans les zones du combustible dont la température n*excède pas 1300*C. Des essais d'irradiation poussés à 35000 MWj/t d'U environ paraly sent montrer qu'en deçà de ce taux de combustion, la fraction dégagée des gaz de fission formés reste très inférieure à 1 % tant qu'aucune modification microstructurale de l’oxyde n'est visible (12). 3..2.2 - Dans la partie centrale du combustible, les concentrations mesurées sont d'autant plus faibles que la température est plus élevée 3 4 6 CHENEBAULT et DELMAS et le faciès de recristallisation plus marqué ; elles sont pourtant toujours mesurables. Lorsque la température est suffisante et les conditions de distribution des volumes libres convenables pour qu’un mouvement cen tripète de bulles de gaz soit possible (transport par évaporation-conden sation), la concentration des gaz de fission stables est à peu près cons tante dans la zone recristallisée et la transition de concentration avec la région "froide" est abrupte (figure 4a). Par contre si le faciès de recristallisation de la région centrale est de type équiaxe ou peu orienté, le profil présente une décroissance progressive des concentrations vers le centre (figure 4b). CONCLUSION La fraction de gaz de fission radioactifs dégagée, à l’état d’équilibre, dans un crayon combustible contenant de l’UO^ fritté, est inversement proportionnelle à la constante de désintégration de chaque isotope. Elle croît lentement avec le taux de combustion à puissance con stan te. L’analyse des gaz de fission retenus dans les pastilles d’UO^ irradiées montre que les concentrations locales des gaz stables attei gnent une limite dans la zone à recristallisation basaltique et permet de chiffrer approximativement ces limites en fonction de la température. REFERENCES (1) Nuc.Appl. Techn., Vol 9, N* 1 et 2 (1970). Symposium on theoretical models for predicting in-reactor performan of fuel and cladding m aterial. (2) H.R. Warner, F.A , Nichols ibid. (1970) p. 148, A statistical fuel swelling and fission gas release model. IAEA-PL-463/19c 3 4 7 (3) R. Soulhier, A. Schürenkämpe r. Resultats nouveaux sur le comportement des gaz de fission à haute tem pérature dans l’UO^ en pile. CEA R 2588 (1964). (4) P. Chènebault, J.P . Hairion. Observations sur le dégagement du brome et de l’iode hors de l’U02 durant l’irradiation. CEA R 3446 (1968). (5) J.P . Stora, P. Chènebault. ^ Programme Cyrano - Mesure de l’intégrale de conductibilité thermique de l’U02 fritté jusqu’à 2300* C - Evolution des gaz de fission à puissance constante. CEA R 3618 (1 9 6 8 ). (6) P. Chènebault, R. Delmas, Ceramic Nucl. Fuels, et. O .L. Kruger and A .I. Kaznoff, Am. Cer. Soc. (1969). In-pile mobility of fission gases in UO^ fuel rods. (7) P. Chènebault, G. Kurka, E. Le Boulbin - Nucl. Inst, and Methods 65 (1968) p. 163-168, North-Holland Publ. Co. Détermination des gaz de fission présents dans le bioxyde d’uranium ir ra d ié . (8) J.C . Janvier, B. De Bernardy de Sigoyer, R. Delmas. Irradiation d’oxyde d’uranium en gaine résistante - Effets du jeu diamétral initial sur le comportement global. CEA R 3358 (1 9 6 7 ). (9) J.P . Hairion, R. Soulhier Etude de l’éjection en tant que mode d’émission des produits de fission par l’oxyde d’uranium, Radiation Damage in Reactor Materials (Compt.Rend.C o ll. Vienne, 1969) 2, ATEA, Vienne (1969) 297. (10) J.P . Stora, P. Chènebault, non publié (1970). 3 4 8 CHENEBAULT et DELMAS (il) P. Chènebault, Congres Inter. Diff. Produits Fission, Commun, N* 5, Saclay (4~6 Nov. 69). Dégagement instantané des gaz de fission de période courte dans les aiguilles combustibles. (12) J. Delafosse et coll., CEA Saclay, communie, personnelle. IAEA-PL-463/19d REACTIONS EN TRE LE SODIUM ET LES OXYDES MIXTES M. HOUSSEAU Section d'études sur les céramiques à base de plutonium G. DEAN, F. PERRET CEA, Centre d'études nucléaires de Fontenay-aux-Roses, Fontenay-aux-Roses, France Abstract-Résu mé REACTIONS BETWEEN SODIUM AND MIXED OXIDES. In order to predict the consequences of in-pile cladding failures, reactions between sodium and mixed oxides have been studied outside the reactor with various fuels, some irradiated and some not. The possibilities of reaction in the U -Pu-O-Na system were studied some years ago by Aitken on the basis of thermodynamic data relating to (U,Pu)-0 and Na-0 compounds. Here the authors study the factors involved in these reactions, particularly the effects of oxide composition and sodium purity. REACTIONS ENTRE LE SODIUM ET LES OXYDES MIXTES. Afin de prévoir les conséquences possibles d'une rupture de gaine en pile, les réactions entre le sodium et les oxydes m ixtes ont été étudiées hors pile sur des combustibles préalablem ent irradiés ou non. Les possibilités de réaction dans le système U-Pu-O-Na ont été examinées, voici quelques années, par. Aitken à partir des données thermodynamiques sur les composés (U, Pu)-0 et Na-О. Les auteurs étudient les facteurs intervenant dans ces réactions, et plus particulièrement les effets de la composition de l ’oxyde et de la pureté du sodium. 1. - Introduction Afin de prévoir les conséquences possibles d'une rupture de gaine en pile, les réactions entre le sodium et les oxydes mixtes ont été étudiées hors pile, sur des combustibles, préalablement irradiés ou non. Les possibilités de réaction dans le système U-Pu-O-Na ont été examinées, voici quelques années, par Aitken, à partir des données thermodynamiques sur les composés (U, Pu) - O et Na - O ( 1 ). Dans ce travail nous étudions les facteurs interve nant dans ces réactions, et plus particulièrement les effets de la composition de l'oxyde et de la pureté du sodium. 2. - Méthodes expérimentales : 2. 1 - Echantillons La majorité des essais a été effectuée sur des pastilles d'oxydes mixtes simulant la composition chimique corres pondant à un taux de combustion déterminé. 3 4 9 3 5 0 HOUSSEAU et al. Ces échantillons, obtenus par frittage, avaient une densité variant entre 90 et 95 % de la valeur théorique ( 2 ). Ils ont été scellés dans un tube d'acier inoxydable de type 316 avec des proportions pondérales sodium / oxyde variant entre 0 , 1 et 1 . Un type de capsule à deux compartiments permet des études avec du sodium en phase vapeur ou purifié "in situ" par des copeaux de titane - zirconium. Quelques essais ont été effectués sur des échantillons irradiés, contenus dans leur gaine d'acier inoxydable. Dans ce cas les proportions pondérales sodium / oxyde sont de l'ordre de 0 , 8 . Le plus généralement la teneur initiale en oxygène du sodium utilisé est de l'ord re de 1 0 0 ppm. Les traitements thermiques ont lieu à des températures variant entre 300 et 1000° C. 2. 2 - Méthodes d'examen : 2. 2. 1 - Evolution de la réaction Dans le cas des oxydes non irradiés, pour les essais en "gonflement libre", la réaction est suivie par radiographie en mesu rant sur les films les dimensions des pastilles à ± 0, 05 mm. La valeur absolue du gonflement est obtenue en comparant les mesures, effectuées au palmer, sur les échantillons avant et après réaction. Dans le cas des expériences en "gonflement lié", (échantillons contenus à l'intérieur d'une gaine avec un faible jeu), la cinétique de la réaction est obtenue en suivant la variation diamétrale de la gaine, avant et après traitement dans le sodium, au moyen d'un palmer. 2. 2. 2. - Caractérisation des phases La détermination des phases, en présence, après réac tion, est effectuée à l'aide d'un diffractomètre Philips. La valeur du paramètre réticulaire de la matrice d'oxyde mixte permet de déterminer les rapports O/U + Pu et O/U + Pu + produits de fission solubles ( 3 ). Des examens micrographiques permettent d'avoir une idée de l'étendue de la zone de réaction. 3. - Résultats 3. 1.- Cas des oxydes purs (représentatifs du combustible en début de vie) Nous donnerons rapidement les principaux résultats qui seront publiés en détail ultérieurement ( 4 ). LAEA-PL-463/19d 351 ду V FIG. 1. Variation du gonflement des oxydes mixtes (U, Pu)^ en fonction de l'écart à la stoechiométrie d’équilibre (Vpu “ 3,4) (gonflement libre). 3. 1. 1. - Cinétique, seuil et amplitude de la réaction Au dessous de 500° C on n'observe aucun signe de réaction. Les essais effectués ont montré, par extrapolation à gonflement nul, qu'il existe pour le rapport O/U + Pu un seuil au dessous duquel il n'y a pas de réaction entre les oxydes mixtes et le sodium. Pour des températures comprises entre 600 et 900 ° C, il correspond à une valence du plutonium voisine de 3, 4, indépendem- ment de la teneur en plutonium et de la température, valeur assez pro che de celle observée par Blackburn et al.( 5 ) pour 25 % de plutonium. A l'intérieur de ces limites? la température influe sur la vitesse de réaction sans modifier l'amplitude finale du gonflement ou la stoechio métrie d'équilibre. Le gonflement en volume, G = ^r-, observé à l'équilibre, est proportionnel à A x,, l'écart entre le rapport O/U + Pu initial et celui correspondant à la valence 3, 4 du plutonium. Pour les teneurs en plutonium inférieures à 30 %, il peut s'exprim er par la formule G x ( fig, 1 ). 3. 1. 2. - Produit de réaction ( 6 ) ( 7 ) Le produit de la réaction entre les oxydes mixtes et le sodium est un composé du type Na3 (U, Pu) O4 . La structure est cubique à faces centrées, ou quadratique selon la teneur en plutonium, mais le volume de la maille varie de façonÇ sensiblement linéaire entre o les composés purs d'uranium (a = 4, 78 A) ou de plutonium (a = 4,88 A) qui sont cubiques. 3 5 2 HOUSSEAU et al. Les principales propriétés mesurées sur NagU0 4 sont : - un point de fusion de 1420 - 25° C - Une conductibilité thermique à 540° C, environ 6 fois plus faible que celle d'UOg. 3. 2. - Oxydes mixtes contenant des produits de fission solides A part les études en "gonflement lié" (combustible à l'intérieur d'une gaine avec un faible jeu), tous les essais ont été effec tués sur des oxydes mixtes de rapports Pu / U + Pu = 0, 20 et O/U + Pu = 2, 00, simulant des taux de combustion de 2 - 6 - 10, 5 - 16, 2 at % TCF. . 3. 2. 1.' - Influence du taux de combustion Jusqu'à 200° C on n'observe pas de réaction avec le sodium. Par contre dès 300° C une réaction importante a lieu, et l'amplitude du gonflement en volume, G = ^ y V , est d'autant plus forte que le taux de combustion est élevé ( fig. 2 ) et, pour un temps de traite ment de 36 heures, G est sensiblement proportionnel à 2 V. En outre, les cinétiques et le gonflement final sont très supérieurs à ceux observés sur des oxydes purs de mêmes teneurs en Pu et en oxygène. Ainsi, dans le cas d'une simulation de 16, 2 at % TC F, au bout de 144 heures de ' traitement à 300° C, le gonflement en volume est 14 fois plus grand que celui observé sur un oxyde pur après 2 000 heures de traitement à 900°C. 3. 2. 2. - Influence spécifique de certains produits de fission solides. Le tableau I indique les gonflements observés pour une simulation contenant tous les produits de fission solides correspondant à un taux de combustion de 16, 2 at % TC F, et pour des simulations correspondant au même taux mais ne contenant que certains produits de fission. On voit que : - Les produits de fission insolubles dans la matrice, (ruthénium et molybdène) sont sans effet sur la réaction : on retrouve à 900 ° C des modifications comparables à celles de l'oxyde mixte pur de même composition (teneur en Pu et O/U + Pu). - Les produits de fission solubles, comme les terres rares (représentées ici par le Ce) ou le zirconium sont responsables de la presque totalité du gonflement. - A un moindre titre, les oxydes contenant du baryum et en plus du cérium et du zirconium., participent à la réaction. La différence peut s'expliquer par le fait qu'une partie du zirconium est piégée par le baryum pour former la "phase grise" du type Ba Zr 0 3, ce dernier composé réagissant peu avec le sodium. IAEA-PL-463/19d 3 5 3 FIG. 2. Evolution du gonflement de l’oxyde mixte en fonction du taux de combustion simulé (gonflement libre). TABLEAU I. GONFLEMENTS EN VOLUMES OBSERVES SUR UNE SIMULATION DE 16% AT TCF, TOTALE OU AVEC CERTAINS PRODUITS DE FISSION SPECIFIQUES Simulation Composition 200°C 300 ° C 900°C 408 h 36 h 81 h 144 h 2 0 0 0 h TOTALE TCF 16 % 0 31 % 4 0 % 56% Ru 0 0 0 0 4 % Mo 0 0 0 0 4 % PARTIELLE Zr + Ce £ 6 % 30 % 50 % Ba + Zr f Ce 33 % 33 % 42 % 42 % Ba Zr Og 0 0 0 9 % L'accélération considérable de la réaction avec le sodium des oxydes contenant des produits de fission, par rapport aux oxydes purs, pourrait s'expliquer par une modification des propriétés thermo dynamiques de l'oxyde, ou par un rôle catalytique du cérium ou du zirconium. 3. 2. 3. - Influence de la teneur en oxygène dans le sodium Ainsi que l'avait signalé Aitken ( 1 ), selon sa pureté en oxygène, le sodium pourra plus ou moins favoriser la formation d'un uranate de sodium. Nous avons vérifié expérimentalement cette influence 3 5 4 HOUSSEAU et al. sur un oxyde mixte simulant un taux de combustion de 10, 5 at % TC F, dans des aiguilles à deux compartiments. Le compartiment A des aiguilles est rempli de sodium seul pour une première aiguille, et de sodium avec des copeaux de titane zirconium pour une seconde. Dans le compartiment B de chacune d'elles sont placés des échantillons d'oxyde de même composition. Dans un premier temps les aiguilles sont chauffées à 750° C pendant 50 heures, en évitant tout contact entre le sodium liquide et les échantillons, tout en purifiant le sodium de la seconde. Ensuite elles sont retournées puis chauffées à 300° C. La différence de comportement des deux échantillons est nettement mise en évidence, après 144 heures de trait ement, par radiographie. Alors que l'oxyde recuit dans le sodium normal (contenant 1 0 0 ppm d'oxygène) s'est complètement désagrégé, celui traité dans le sodium purifié (devant contenir initialement de 1 à 2 ppm d'oxygène), bien qu'ayant gonflé, a conservé sa forme. Ceci est la preuve d'un ralentissement de la réaction en présence d'un sodium de plus faible teneur en oxygène. Le gonflement observé après 36 heures de traitement est d'ailleurs probablement dû à la faible quantité de sodium mise dans le dispositif. Au début, la simple réduction de l'oxyde par le sodium entraîne une pollution en oxygène favorisant ainsi la réaction. 3. 2 .4 . - Action de la vapeur Dans un réacteur à neutrons rapides refroidi au sodium la température de celu i-ci est de l'ordre de 500° C. En cas de rupture dVétanchêité d’une aiguille, le combustible étant généralement au dessus de 900° C, dans une zone importante de ce dernier,le sodium ne peut exister que sous forme de vapeur. Aussi avons nous examiné si celle- ci était susceptible de réagir de la même manière que le sodium liquide. Nous avons utilisé pour cela une aiguille à deux compartiments avec un oxyde simulant un taux de combustion de 10, 5 at % TCF. Contrairement aux observations faites par Aitken ( 8 ) sous gradient thermique, en isotherme, nous n'avons pas trouvé de réaction dans les conditions du tableau II, après 440 heures de traitement. Il faut noter qu'en phase vapeur le sodium est pratique ment exempt d'oxygène, puisqu'à 1 000 °K, les données thermodynami ques permettent de calculer un rapport de pression : p (Na) gaz ^ 1¿J14 "p (NaO) gaz pour du sodium liquide saturé en Na^ O. Il se peut également que l'apport de sodium soit insuf fisant pour permettre la poursuite de la réaction au delà de la surface de l'échantillon. 3.2. 5. - Interaction mécanique oxyde - gaine ("gonflement lié") Dans tout ce qui précède, nous n'avons considéré que le cas où les échantillons pouvaient gonfler librement. Il n'en n'est évidem ment pas de même dans une aiguille de combustible où le jeu oxyde-gaine IAEA-PL-463/19d 3 TABLEAU II. TEMPERATURES DES ESSAIS AVEC DU SODIUM EN PHASE VAPEUR SUR UN ENCHANTILLON SIMULANT 10, 5 AT % TCF Température Température Pression de vapeur de l'oxyde (°C) du sodium (°C) du sodium (mm Hg' 1 0 0 0 400 0 , 8 970 550 9 900 650 . 90 850 400 0 , 8 550 850 650 est faible. Dans ces conditions, en cas de pénétration de sodium, une poussée radiale sur la gaine s'exercera,risquant de provoquer la rup ture. Dans un but technologique nous avons effectué des essais sur des oxydes simulés ainsi que sur des oxydes pré-irradiés dans le réacteur Rapsodie. 3. 2. 5.1. - Essais sur des échantillons simulés Un oxyde simulant un taux de combustion de 16, 2 at % TCF a été encapsulé dans une gaine d'acier inoxydable de'O, 35 mm d'épaisseur, avec un jeu diamétral initial de 0 , 1 mm. д w La fig. 3 représente la variation du diamètre de la gaine, ( 0 ) %, en fonction de la température de traitem ent. On rem ar que que la déformation de l'aiguille s'accélère fortement à partir de 500° C. - 3. 2 5. 2. - Essais sur des échantillons préalablement irradiés Des aiguilles du prem ier coeur de Rapsodie dont le taux de combustion variait entre environ 0, 8 et 7 at % TCF, ont été recuites en présence de sodium statique. L'influence de la température et du taux de combustion sur la réaction oxyde - sodium a été suivie en mesurant le gonflement diamétral maximum hors rupture de la gaine. 3. 2. 5. 2.1. - Influence de la température sur la cinétique Les échantillons étudiés avaient un taux de combustion de 6 at % TCF. 356 HOUSSEAU et al. FIG. 3. Evolution du gonflement diamétral de la gaine en fonction de la température dans le cas d'éxperiences en gonflem ent lié. TABLEAU III. CINETIQUES DE DEFORMATION D'AIGUILLES RAPSODIE IRRADIEES A 6 AT % TCF, PUIS RECUITES DANS DU SODIUM STATIQUE Température Temps (°c ) ( heures ) 650 2 1 0, 5 34 0 , 8 80 1 , 2 1 aiguilles 146 1 j rompues 500 90 0, 5 400 230 0 , 1 Ainsi que le prouve le tableau III, comme pour les oxydes simulés, la poussée sur la gaine ne débute de façon notable qu'au-delà de 400° C, pour s'accélérer fortement après 500° C. Ces résultats sont à rapprocher de ceux obtenus par Greenberg et al. (9) ou Aitken et al. (10) .au cours d'expériences en sodium dynamique. IAEA-PL-4|3/19d 3 TABLEAU VI. VALEURS PROBABLES DES ENERGIES LIBRES MOLAIRES PARTIELLES AG(0¡) ET DES TENEURS EN OXYGENE DU SODIUM, A L'EQUILIBRE DANS LE SYSTEME OXYDE MIXTE - SODIUM - URANATE Taux de combustion en at % TCF ) % maxi hors 0 ) rupture 1 , 2 0 0 , 2 3, 05 0 , 2 6 ,8 0 1 3. 2. 5. 2. 2. - Influence du taux de combustion . Les résultats obtenus sur des aiguilles Rapsodie à 650° C. après 70 à 80 heures de traitement dans le sodium, sont rassemblés dans le tableau IV. On remarque qu$ comme dans le cas des oxydes simulés, le gonflement de l'oxyde , répercuté ici sur la gaine, croît avec le taux de combustion. 3. 2. 6 . - Examens sur le combustible après réaction Dans le cas des oxydes simulés, nous n'avons pu mettre en évidence d'autre composé que celui du type Na3 MO4 . Ici encore le rapport O/U + Pu d'équilibre de l'oxyde mixte correspond à une valence du Pu voisine de 3, 4 à 3, 5. Dans le cas des oxydes irradiés, nous n'avons jamais pu mettre en évidence le composé de réaction. Toutefois, le rapport O/U + Pu de la matrice évolue dans le sens d'une réduction, sans toutefois jamais atteindre la valeur obtenue pour les oxydes purs ou de simulation. Ceci peut d'ailleurs être simplement dû à une oxydation des aiguilles dans les cellules d'examen ou en cours de stockage. Par ailleurs, les examens effectués tant sur des aiguilles rompues en pile que sur les essais hors pile montrent une forte migration'du césium avec une perte préférentielle de celui-ci pour le combustible, les pertes étant toujours très faibles. Enfin, une comparaison effectuée entre une aiguille rompue et sa voisine non rompue montre que la rétention Xe + Kr est environ 2 à 3 fois plus faible dans le cas de l'aiguille rompue ( 11 ). 3 5 8 HOUSSEAU et al. 4. - CONCLUSION Ainsi que le prouvent les résultats expérimentaux que nous venons de décrire, les facteurs intervenant dans la réaction oxyde- sodium sont de deux types : - D'une part des facteurs potentiels, concernant les conditions d'apparition de la réaction, tels que le rapport O/U + Pu de l'oxyde mixte, la teneur en oxygène du sodium, la présence de produite de fission. - D'autre part, des facteurs relevant de considérations de cinétique et concernant le développement de la réaction, tels que la température de l'oxyde, les possibilités d'apport des constituants (U, Pu), O, Na. Le premier groupe de facteurs, et en particulier les énergies libres molaires partielles respectives des divers composés en présence (oxyde, sodium, uranate) est très important car il condi tionne les conditions d'apparition de la réaction. Or nous avons vu TABLEAU V. DIVERS RESULTATS CONCERNANT L'ENERGIE LIBRE MOLAIRE PARTIELLE AG(o2) D'EQUILIBRE AVEC LE SODIUM, L'OXYDE ET L ’URANATE DE SODIUM POUR 0 Blackburn Gai Electric Nos résultats â Blackburn (mesuK mesure mesure l'équilibre VpUü¿3 , t de O/yi équilibre directe directe ^pu — 3, 6 v/aporisatic i f .ejn (O) Ag (o 2) AG(o ,) Température A W WoocSeÿ de dans Markin Woodley ° С Markin * * pile Na 500 - 174 - 198 - 170 - 199 600 - 189 650 - 185 -180 700 ‘ - 161 - 181 - 154 - 176 - 160 750 - 178 - 176 900 - 153 - 167 - 146 - 162 - 148 ■jf Valeurs extrapolées IAEA-PL-463/19d 359 TABLEAU IV. INFLUENCE DU TAUX DE COMBUSTION SUR LA DEFORMATION D'AIGUILLES RAPSODIE RECUITES 70-80 HEURES A 650° DANS DU SODIUM STATIQUE Temperature _^ G ( O2 ) Teneur en O dans Na (ppm) 500° C 195 ^ 0 , 0 1 700° C 180 0 , 0 2 900° C 165 0 , 3 précédemment, que dans les oxydes purs ou simulés, quelles que soient la température ou la teneur en plutonium de l'oxyde considéré, la compo sition d'équilibre correspond à une valence du plutonium voisine de 3, 4. Ceci entraîne pour l'énergie molaire partielle/^ G(ç>g)à l'équilibre les valeurs reportées dans le tableau V en utilisant les valeurs de Rand et al,( 12 ) ou Woodley ( 13 ). En supposant que la pression partielle d'oxygène varie peu avec la température, nous avons comparé nos résultats avec ceux de la littérature (5 , 15 ). Si, comme nous le pensons, les valeurs les plus proba bles sont les plus basses (les valeurs les plus hautes étant déduites de méthodes plus discutables), les teneurs en oxygène correspondantes du sodium sont celles du tableau VI. On voit donc que l'oxyde mixte et le sodium pourront être oxydant ou réducteur l'un vis-à-vis de l'autre, et que selon leur potentiel respectif d'oxygène ils favoriseront ou non la réaction, et que., pour éviter toute réaction,il faut un oxyde mixte très réduit et un sodium très propre. REFERENCES ( 1 ) E. A. Aitken, G. E. A. P 5683 (1968) (2 ) F. Schmitz, rapport C.E. A R. 3795 (1969) ( 3 ) F. Schmitz, G. Dean, M. Halachmy, J. Nucl. Mat. 40 (1971) 325 - 337 ( 4 ) M. Housseau, G. Dean, J .P . Marcon, J. F . Marin, rapport CEA à paraître. 3 6 0 HOUSSEAU et al. ( 5 ) P. E. Blackburn, ANL 7854 (1971) (6 ) J. P. Marcon, 0. Pesme, M. de Franco, à paraître dans le journal des Hautes Températures et des Rêfractaires. ( 7 ) O. Pesme, J. C. Boivineau : communication privée ( 8 ) E. A. Aitken, S. K. Evans, G. E. A. P - 12 133 (1970) ( 9 ) S. Greenberg, W .E. Ruther, J .R . Honekamp, D. A. Donahue A.N. S. Trans. 14 1 (1971) 178 (10) E. A. Aitken, S. K. Evans, G. F . Melde, B. F . Rubin Fast Reactor Fuel Element Technology (Ruth Far makes Ed.) Am. Nucl. Sty. (1971) 459-478. (11) B. Kryger : communication privée (12) M. H. Rand, T. L. Markin, A .E .R . E - R - 5560 (1967) (13) R. E. Woodley, HEDL - SA - 269 (1970) (14) P.E. Blackburn, A. E. Martin, J. E. Battles, P. A. G. O'Hare, W. N. Hubbard, F ast Reactor Fuel Element Technology (Ruth Far makes Ed.) Am. Nucl. Sty. (1971) 479 - 493 (15) E. A. Aitken : communication privée IAEA-PL-463/20 REACTION BEHAVIOUR OF FISSION PRODUCTS IN CARBIDES H. HOLLECK, E. SMAILOS Kernforschungszentrum Karlsruhe, Institut für Material- und Festkörperforschung, Karlsruhe, Federal Republic of Germany Abstract REACTION BEHAVIOUR OF FISSION PRODUCTS IN CARBIDES. Phase diagrams of the systems (Th, U, Pu)-fission product-carbon are given. These ternary sections are partially estimated. Investigations of UC with a simulated high burn-up (10, 20, 30%) show the occurrence o f four different phases as inclusions. I. PHASE DIAGRAMS The phase diagrams of the fuel with fission products can be given for the following components: Y, La, Ce, Nd, Zr, Mo, Ru, Rh, Pd, Se, Te. No phase diagrams are known for the alkaline.and the alkaline earth metals. Isothermal sections are shown in Figs 1 to 8 . In some cases they are estimated from only a few experimental results. These phase diagrams give only data for the individual behaviour of single fission products. II. INVESTIGATIONS WITH SIMULATED BURN-UP (10, 20, 30%) IN UC The joint reaction behaviour of more than one fission product present in the fuel is sometimes different from the behaviour of only one fission product. Investigations with the high-melting fission products Zr, Mo, Ru, Rh and Pd showed the occurrence of three phases: (U, Zr)C, UMoC2 and U2(Ru, Rh, Pd)C2. The lattice parameters of (U, Zr)C are: at 10% burn-up a = 4. 952 A at 20% burn-up a = 4. 943 A at 30% burn-up a = 4. 930 A Figure 9 shows the distribution of the elements in a sample with a fission product concentration according to 10% burn-up. The platinum metals Ru, Rh and Pd show the same behaviour in this case [ 15] . Simulated experiments including the rare earths and Sr [16] show that the behaviour of these is somewhat different from that one would expect. The rare earth metals are soluble in UC only in very low concen trations and the platinum metals Ru, Rh and Pd mutually show a different behaviour (see Fig. 10). 361 362 HOLLECK and SMAILOS С FIG. 1. Isothermal sections in the systems Th-Ce-C,. U -C e-C and Pu-Ce-C. IAEA-PL-463/20 363 С FIG. 2. Isothermal sections in the systems U-La-C, U-Ce-C and U-Nd-C. 364 HOLLECK and SMAILOS FIG. 3. Isothermal sections in the systems Th-Zr-C, U-Zr-C and Pu-Zr-C. IAEA-PL-463/20 365 С FIG.4. Isothermal sections in the systems Th-Mo-C, U-Mo-C and Pu-Mo-C. 366 HOLLECK and SMAILOS С FIG. 5. Isothermal sections in the systems Th-Ru-C, U-Ru-C and Pu-Ru-C. I.5. re nhlis omto o rnu ad ltnu-iso pout n -ue cmons[17] 7 1 [ compounds el -fu and product plutonium-fission and uranium of formation f o enthalpies Free 5a. FIG. ÛG° [kcal / mol] IAEA-PL-463/20 [K] T 367 368 HOLLECK and SMAILOS С PuçRh* Pu3Rh¿ PuRh3 FIG. 6. Isothermal sections in the systems Th-Rh-C, U-Rh-C and Pu-Rh-C. IAEA-PL-46 3/20 3 6 9 с Pu3Pdt FIG. 7. Isothermal sections in the systems Th-Pd-C, U-Pd-C and Pu-Pd-C. 3 7 0 HOLLECK and SMAILOS С С U 3Te4 U 2 Te3 и 7 Т е 12 FIG. 8. Isothermal sections in the systems U-Se-C and U-Te-C. IAEA-PL-463/20 3 7 1 ELECTRON - MICROSCOPE PICTURE (NEGATIVE) FIG. 9. Relative distribution of the elements U , Ru, Rh, Pd, C, Zr, Mo in a UC + (Zr, Mo, Ru, Rh, Pd) sample (arc-melted and homogenized for 145 h at 1500°C) as at lO^o burn-up: Phase observed: (U,Zr)C, U2(Ru,Rh,Pd)C2 UMoC2. 3 7 2 HOLLECK and SMAILOS FIG.9 (cont.) Zr IAEA-PL-463/20 373 t«' MICROGRAPH SHOWING THE AREA TESTED 400 : 1 FIG.10. Relative distribution of the elements U, Zr, C, Ru, Rh, Pd, Mo, La, Ce, Pr, Nd, in a UC + fission products sample (arc-melted and homogenized for 145 h at 1500°C) as at 20% burn-up (512:1). 3 7 4 HOLLECK and SMAILOS FIG. 10 (cont. ) ■r« ELECTRON - MICROSCOPE MICROSCOPE - ELECTRON ITR (NEGATIVE) PICTURE aas# aas# тшжшш . . ft. 'i-j: IAF.A-PL-463/20 IAF.A-PL-463/20 4 FIG. 10 (con t. ) t. (con 10 FIG. Ru 5 7 3 376 HOLLECK and SMAILOS ELECTRON - MICROSCOPE Mo PICTURE (NEGATIVE) Pr Nd FIG.10 (cont.) IAEA-PL-463/20 377 TABLE I. FUEL - FISSION PRODUCT INTERACTIONS Phases in alloys corresponding to 10, 20, 30% burn-up: UC (substoich.) + (Sr, Y, Zr, Mo, Ru, Rh, Pd, La, Ce, Pr, Nd, Sm) Phases detected: low content 1. (U, Zr, rare earth) С 2. UM oQ 3. U2RuCz 4. (U, La, Ce,Pr,Nd) (Ru,Rh) *i ' Ji 5. (U, La, Ce,Pr,Nd) (Ru.Rh.Pd) x2 У2 Xi > x2 ; y, < y2 Phases 4 and 5 probably carbon-containing lattice parameters of the fuel matrix: Burn-up - 0 2 * (%) ' (ppm) 0 4 .9 6 0 600 10 4.957 1000 20 4.944 800 30 4 .9 3 5 2200 The following phases could be identified by microprobe analysis (Table I): (a) (U, Zr)C with about 1 wt.% rare earths in solution. The lattice para meters are: at 0% burn-up 4.96 0 A at 10% burn-up 4.957 A at 20% burn-up 4.944 A at 30% burn-up 4.935 A (b) UMoC2 with no other elements in solution. The lattice parameters are: a = 5.642 A; b= 3.250 h с = 11. 027 A. (c) U2RuC2 with about 0. 5 wt. % Rh in solution (a = 3.442 A, с = 12.559 A). (d) Two phases containing rare earth metals, platinum metals, a low uranium content (~ 3 wt.%) and a low carbon content (1 -2 wt. %). One phase contains a high content of rare earth and mainly ruthenium as platinum metal, while the other phase shows a low content of rare earth metals and a high content of rhodium and palladium and also about 3 wt.% Ru. This phase is perhaps a lanthanide-platinum metal perovskite-carbide [13]. These results are obtained with substoichio- metric UC. Tests with stoichiometric or hyperstoichiometric carbide are in progress. 378 HOLLECK and SMAILOS III. CONCLUSIONS Investigations into the chemical state of the fission products in a carbide fuel with simulated burn-up can give some understanding of the effect in highly irradiated fuels. It will be more difficult to examine fuel-fission product reactions in irradiated carbide fuels than in an oxide fuel. This is because the temperature and the temperature gradient are lower and the phases formed are in part very refractory. This means that grain growth of fission product phases is prevented. It is not possible at the moment to predict the change in stoichiometry because the carbon content of the rare earth phases is not yet known. Nevertheless, an increase in the carbon potential with increasing burn-up seems to be probable in hypostoichiometric carbides. It is also not possible to predict the reaction behaviour of the fission products in a carbide fuel from the results in systems with a limited number of components. REFERENCES [1 ] STECHER, P., NECKEL, A., BENESOVSKY, F., NOWOTNY, H., Planseeber. Pulvermet. 12 (1964) 181. [2] HOLLECK, H., estimated. [3] HAINES, H.R. , POTTER, P.E.; Rep. AERE-R-6512 (1970). [4 ] HOLLECK, H., unpublished data. [5] BENESOVSKY, F., RUDY, E., Planseeber. Pulvermet. £(1961) 65: [ 6] RUDY, E ., in BENESOVSKY, F. , (Ed.) Pulvermetallurgie in der Atomkerntechnik, 4. Plansee Seminar Reutte/Tirol, 1962. [ 7 ] CHUBB, W ., KELLER, D. L. , Rep. BMI 1685 (1964). [8] HOLLECK, H., KLEYKAMP, H., J. Nucl. Mater. 32 (1969) 1. [9 ] HOLLECK, H., KLEYKAMP, H.', J. Nucl. Mater. 35(1970)158. [10] HOLLECK, H., KLEYKAMP, H., I. Nucl. Mater, (in press). [11] HOLLECK, H., Monatsh. Chem. 102 (1971) 1699. [ 1 2 ] BREEZE, E. W ., Thesis Univ. Sheffield, 1971; BREEZE, E.W ., BRETT, N.H. , WHITE, J ., J. Nucl. Mater. 39 (1971) 157. [13] HOLLECK, H., J. Nucl. Mater. 42 (1972) 278. [ 1 4 ] LORENZELLI, N ., MARCON, J. P ., I. Less-C om mon M et. 26 (1972) 71. [15] HOLLECK, H., SMAILOS, E., KFK-Ber. 1271/4, pp. 112-43. [16] HOLLECK, H., SMAILOS, E., KFK-Ber. 1272/1. [17] HOLLECK, H., KLEYKAMP, H., THÜMMLER, F., Proc. Reaktortagung Bonn, 1971, Deutsches Atom forum, p. 582. lAEA-PL-463/21 FISSION PRODUCT DISTRIBUTION IN FAST REACTOR OXIDE FUELS H .J. POWELL UKAEA, Dounreay Experimental Research Establishment, Dounreay, Scotland, United Kingdom Abstract FISSION PRODUCT DISTRIBUTION IN FAST REACTOR OXIDE FUELS. The paper describes results obtained at Dounreay by postirradiation examination of fuel pins irradiated in the Dounreay Fast Reactor (DFR) as part of the PFR fuel development program. 1. INTRODUCTION Operating conditions in fast reactor fuel elements result in an environment of unique severity. The necessarily high fast neutron flux produces high fission rates and corresponding intense fluxes of energetic fission fragments, beta particles and gamma photons, which ensure that every atom in a fuel element is violently disturbed many times during irradiation. Moreover, economic considerations require higher burn-up levels than in other types of reactors so that the fuel elements must withstand these severe conditions for considerable periods of time during which a substantial fraction of the original actinide atoms is transformed into new species. Because of these requirements, most fuel development programs for fast reactors have concentrated on refactory ceramic compounds of uranium and plutonium, with the major emphasis on the mixed oxides. One of the more important characteristics of mixed (U, Pu) oxide, from the fuel-element designer's viewpoint, is its low thermal conduc tivity, which leads to high centre temperatures and steep temperature gradients even in small diameter fuel pins. For example, the reference design of fuel pin for the Prototype Fast Reactor (PFR), currently under construction at Dounreay, has an internal diameter of only 5.08 mm, but centre temperatures in excess of 2000°C and average radial temperature gradients of almost 1000°C/mm are to be expected. These steep temperature gradients provide large deriving forces for migration of the various species present and the high temperature of much of the fuel should lead to high mobilities for many species. Hence extensive redistribution of some fission products is to be expected. 2. EXPERIMENTAL DETAILS 2.1. Fuel pins The fuel pins consisted of a sealed tube of M316 stainless steel containing a column of mixed (U, Pu) oxide between two natural UO2 379 380 POWELL insulator pellets. The cladding diameter was generally 5.08/5.84 mm and the plutonium concentration 15%, but a few pins with diameter 7.01/8.25 mm and a few containing (U, 25% Pu) oxide were irradiated. A plenum was provided, generally at the upper, cooler end, for the accommodation of released fission-product gases. In addition, the fuelled length of the pin contained only 70-85% of the possible weight of fuel, in the form of vibro-compacted powder or annular pellets, the remaining space being provided for accommodation of fuel swelling resulting from the accumulation of fission products. The fuel pins were irradiated singly, in groups of three or in sub-assemblies of 77 pins. In each case they were cooled by downward flowing reactor NaK so that the clad temperature increased from 230°C at the upper end to a value, generally about 650°C, at the bottom end that was determined by the heat output of the pin and the coolant flowrate. Further details of the fuel and fuel pin manufacturing route have been published elsewhere [ 1 , 2 ]. 2.2. Post-irradiation examination After discharge from the reactor and cooling for periods that varied between 1 and 12 weeks, the fuel pins were subjected to a standard examination sequence consisting of the following stages; (a) X-radiography of the irradiation vehicle to determine the position and condition of its components (b) Vehicle breakdown and extraction of the fuel pins (c) Pin radiography to determine the size and position of any central hole in the fuel (d) Measurement of pin length and diameter (e) Determination by g a m m a scanning of the axial distribution of a number of fission products. At this point, some pins were replaced in new vehicles for further irradiation, while others were pierced to extract released fission product gases prior to section for metallography and chemical analysis. Descriptions of these examination procedures have been published elsewhere [1-4] but brief accounts of the techniques that are particularly relevant to the present studies are given below. 2.2.1. Gamma scanning Two techniques were used in this non-destructive procedure for determining fission product distribution. All pins were examined by simply moving the pin, on a motorized trolley, past a collimating slit about 0.33 mm wide. The emergent gamma-ray beam entered a sodium iodide phosphor coupled to a photomultiplier, the output from which was further amplified and sorted in a single-channel analyser [4]. Pulses corresponding to gamma photons within ±40 keV of a selected energy were recorded. Gamma scans were normally carried out at nominal energies of 520, 760 and 1600 keV, corresponding to 103Ru + 106Rh, 95Zr and 95Nb and 140La gamma rays, respectively. Although quick results were obtainable in this way, it suffered one grave disadvantage. The gamma photons recorded at the highest energy level (1.6 MeV) were IAEA-PL-46 3/21 381 derived almost entirely from disintegration of 140La, but at the lower energies the photons consisted of those in the photopeak of interest together with minor peaks from other nuclides and a background continuum derived from Compton scattering of initially higher energy photons. The relative contribution of this background increased with decreasing photopeak energy so that the 520-keV scan was very strongly influenced by varia tions in distribution of high energy gamma emitters and misleading results were frequently obtained. An alternative, but more time-consuming, procedure consisted in the collection of complete gamma spectra, using a multichannel analyser, at intervals along the fuel pins, which permitted the determination of photo peak activities with reasonable precision. More recently a lithium-drifted germanium detector has been used, which offers considerably better energy resolution. Even with this improvement, misleading results were occasionally obtained on short-cooled fuels. 2.2.2. Fission product gas release Gases were extracted through a hole drilled at or near the top of the fuel pin plenum into a closed system of known volume. Measurements of the pressure after piercing, together with the results of m ass-spectrom etrie analysis of samples of the gas, gave estimates of the volumes of fission- product and other gases in the pin. Fractional gas release was calculated on the assumption of a production rate of 213 mm3 /g-% burn-up in oxide fuel. 2.2.3. Chemical analysis Samples consisting of about 10 mm lengths of the fuel pin, containing about 2 g of fuel, were digested with aqua regia, followed by concentrated nitric acid and then dissolved in concentrated nitric acid containing a little hydrofluoric acid. The resulting solution normally contained all the actinides, part of the cladding and most of the fission products. (Incom plete dissolution of the noble metals ruthenium, rhodium and palladium generally occurred and part of the total inventory of molybdenum and technetium also remained in the insoluble residue.) Radiochemical separations were carried out on aliquots of the solution prior to deter mination of 95Zr, 144Ce, 137Cs (and occasionally 106Ru, 134Cs, ®9Sr and 90Sr) by ß - or 7 -counting. Plutonium, uranium, neodymium (and occasionally molybdenum) were determined m ass-spectrom etrically by isotope dilution techniques. Further details of these procedures are available elsewhere [5]. Investigations into the' radial distribution of some fission products were undertaken by a microdrilling technique [6 ]. Samples weighing about 10 ß g were obtained from known positions and dissolved in nitric acid. The solutions were analysed for fission products by gamma- spectrometry, as for the larger samples. More recently an electron probe micro-analysis unit has been acquired. Investigations into the composition of metallic inclusions have been reported already [7] and results should shortly be available for complete cross-sections of some fuel pins. 382 POWELL To obtain data on the axial distribution of retained fission-product gases, a number of samples were dissolved in a boiling mixture of sulphuric and phosphoric acids in a helium-purged system [8 ]. Xenon and krypton were removed from the helium stream by traps containing active charcoal and were subsequently recovered for analysis by warming the traps. The solution was analysed for fuel and fission-product elements by the normal techniques after it had been shown that concordant results were obtained by the two dissolution methods. 3. RESULTS AND DISCUSSION 3.1. Fission product gases Data on fission-product gas release have been obtained from a considerable number of experimental fuel pins irradiated in the DFR. Most of these pins were of PFR reference size (5.08 mm i.d.) and contained U, 15% Pu) oxide at a smear density of about 80% of theoretical, but data have also been obtained from a few pins in which one or more of these parameters had alternative values. Results obtained by piercing vibro-compacted and annular pellet fuel pins are presented in Figs 1 and 2 respectively, and measurements of the gas retention profiles in one pin of each type are given in Fig.3. Graphical integration of the gas content data for the samples indicated average release fractions of 0.62 and 0.38 for the vibro-compacted and annular pellet fuel respectively, in reasonable agreement with the values of 0.70 and 0.40 obtained by piercing the pins. Fission-product gas release in5.08-m m diameter pins containing vibro-compacted (U, 15% Pu) oxide appears to be independent of heat rating (in the range 32-44 kW/m) and smear density (78-85% of theoretical). The gas release fraction R is essentially dependent only on the mean burn-up B% 2 (see F ig .l). The empirical equation R = 1 - | e x p (-B / 6 ) predicts the gas release fraction in the above-mentioned conditions to within ±0.1 and generally within ±0.05. Thus gas release increases pro gressively to about 90% at 10% burn-up. During the early stages of irra diation (burn-up below 6 %) the mean gas content of the fuel also increases to a maximum value of about 300 mm3 /g . The empirical equation above indicates a decrease in retained gas concentration at higher burn-ups, but the data are equally consistent with the maintenance of a saturation level of about 300 mm3/g until at least 9% burn-up. Most of the changes in fuel or irradiation parameters that have been tested have led to in creases in the already high gas release values; note the enhanced gas release in 7.0 mm pins (linear ratings about 75 kW/m), in pins containing 70% dense fuel and in pins containing (U, 25% Pu) oxide. Gas release values for annular pelleted fuel pins are generally lower than for corresponding vibro-compacted fuels (Fig.2). At heat ratings below about 200 W/g in 5.08-mm pins restructuring is limited to equi- axed grain growth; in these conditions gas release is dependent on fuel- clad conductance (determined mainly by the width of the as-fabricated gap between fuel and clad) and heat rating. At higher heat ratings columnar 1 The burn-up units used throughout this paper are % FIMA (Fission per Initial actinide Metal Atom). IAEA-PL-463/21 383 Pin i.d. Smear density mm '/.TD О 5.1 0.15 80 • 5.1 0.25 80 □ 5.1 0.15 70 0 2 - Л 7.0 0.15 8 0 J ______L _L 10 Mean burn up. В \ FIG. 1. Fission-product gas release from vibro-compacied fuel in DFR. grain zones develop and gas release values approach those for vibro- compacted fuels. There are indications that the initial fuel-clad gap width is of little importance in these conditions, probably because of rapid gap closure during irradiation. Results obtained by piercing tests and from isothermal irradiations of small fuel samples [9] clearly indicate the importance of fuel temperature to gas release behaviour. At the high temperatures prevailing in the columnar grain zone essentially complete gas release undoubtedly occurs. However, the assumption of rather low gas release fractions from much of the cooler fuel used in some fuel pin modelling codes is of doubtful validity. For example, a pin irradiated to 8 .8 % burn-up in conditions such that only about 40% of the fuel formed columnar grains, released 85% of the fission product gas; the average gas release from the equiaxed grains must have been at least 75%. 3.2. Solid and liquid fission products The fission process in uranium or plutonium results in the produc tion of atoms of more than 30 elements, represented by about 120 isotopes with half-lives of at least a few days and therefore likely to be present 384 POWELL 0.8 - ï 0.6 ° 0.4 0.2 5 10 Mean burn-up % FIG. 2. Fission-product gas release from annular pellet fuel in DFR. during post-irradiation examination. Only a few of these isotopes have properties that permit detailed study of their distribution by simple techniques such as gamma-spectrometry and the behaviour of the majority must be inferred from these few.. Since fission-product behaviour will obviously be determined mainly by the chemical and physical properties of the species present, any attempt to predict distributions must start with an assessment of the chemical state of the fission products. In practice, the most-important factor is the fate of the two oxygen atoms liberated by each fission event. The results of X-ray examination of simulated fuel-fission product mixtures and published or estimated thermodynamic data on fission product oxides have been used by several workers to assess the chemical state of high burn-up fuels. The salient features are summarized below. (1) The stable oxide-forming elements, barium, strontium, zirconium, niobium, yttrium and the rare earths, should be present as oxides. Barium and (probably) strontium oxide are not soluble in (U, Pu) oxide and should therefore exist as a separate phase: they are generally lAEA-PL-463/21 385 Distance from core centre, mm FIG. 3. Fission-product gas retention. believed to be present as zirconates (Ba, Sr)ZrC>3 . The remainder of the zirconium and the rare earths should be present as solid solution in the (U, Pu) oxide matrix. (2) The elements with atomic numbers 43 - 52 (i.e. technetium to tellurium) should be present as elements. (3) The halogens should form caesium compounds and the remaining caesium should be present in elemental form, ‘except in markedly hyper stoichiometric fuel, .in which caesium oxide can form. An additional 386 POWELL possibility is the formation of double oxides - reactions with fuel, fission products or cladding constituents have all been suggested, leading to the formation of caesium uranate, molybdate or chromate. (4) The state of molybdenum, the only high yield fission-product not considered so far, should be determined by the prevailing stoichiometry of the fuel because the oxygen potential of the system M0 /M 0 O2 is very close to that of stoichiometric (U, Pu) oxide. Thus molybdenum should remain in elemental form in hypostoichiometric fuel and be oxidized in hyperstoichiometric material. In high burn-up fuel the state of molybdenum is expected to be influenced by the initial stoichiometry, the burri-up and the identity of the fissioning species (which determines the fission-product spectrum'and therefore the magnitude of the burn-up dependence of stoichiometry). Experimental fuel pins irradiated in both DFR and EBRII have contained highly enriched uranium in order to operate at high heat ratings with plutonium concentrations representative of power-producing fast reactors and so the majority of the fission have been of uranium atoms. The initial oxygen/metal ratios have generally been in the range 1.98 - 2.01 and oxygen-balance calculations indicate that the terminal condition in most pins was a near-stoichiometric (U, Pu) oxide matrix with partial oxidation of the molybdenum. One further factor that must be considered in assessing the distribu tion of fission-product species is the effects of short-lived precursors. None of the fission-product species nuclides studied during post-irradiation examination is produced directly in fission, the primary fission products are radioactive species that undergo several stages of beta decay. It is possible that some isotopes of non-migratory fission products may have mobile precursors with half-lives sufficiently long for migration to occur. Thus demonstration of the immobility of one isotope of an element does not necessarily provide evidence that all isotopes of the element are immobile. Several instances of such precursor effects will be cited during the following discussion in which the fission-product elements are considered in groups suggested by the classification scheme outlined in the previous paragraph. 3.2.1. Alkali metals Since thé alkali metals should be present mainly in highly volatile, elemental form, extensive migration is to be expected and is generally observed. The salient features revealed by numerous analyses are outlined below. (1) Gross radial migration to the cool periphery occurs even in relatively mild irradiation condition; fuel pins in which only equiaxed grain growth has occurred have been found to have caesium concentra tions near the periphery that are a factor of 1 0 0 or more greater than those near the centre. Most of the data have been obtained by the m icro drilling procedure and refer to 131Cs only, but a crude separation into control and peripheral regions was made for a few samples by ejecting most of the fuel with a screw-driven plunger. The isotopic composition of the caesium in the two regions was almost always virtually identical and similar to that calculated from the fission yields, apart from slight depletion in 133Cs. IAEA-PL-463/21 387 (2) Axial migration of caesium towards the cooler end of the fuel pin is frequently observed (see, for example, Fig.4). The isotopic composition of caesium in the fuel, column is independent of axial loca tion except in a small region near the cool end (Fig.5). (3) Very little caesium is found in the plenum of any pin and the small quantity that is present generally differs markedly from that in most of the fuel (Fig.5), being highly enriched in 133Cs, with minor quantities of 135Cs and only traces of 137Cs. (4) Samples from failed pins contain only about 20 - 25% of the calculated amount of caesium, the remainder having been leached out by exposure to the reactor coolant (NaK), which enters the pins after failure. 2 c ------1------1------1------1------] 700 500 u • a Ê 0) I— *0 _0 и 300 -200 -Ю0 0 100 Distance from D.F.R. Core Centre, mm FIG. 4, Axial distribution of 137 Cs in vibro-compacted fuel. (U.Pu)O-. UO?*+1 Plenum 1 0 0 ------1------1— 2------1------1------i------1------ a a 135Cs a b # Bulk fuel ■ □ Cs > • 0 1î7Cs д D ° Per'P^erQl *uel -200 0 +200 Distance from core centre, mm FIG. 5, Axial distribution of caesium isotopes. 388 POWELL This suggests that most of the caesium has effectively been released from the fuel matrix and implies comparable release fractions for the alkali metals and inert gases. These observations indicate that caesium is, as expected, quite mobile within the fuel, and that most of the small quantities of caesium found in the plena are produced therein by decay of xenon precursors. Isothermal irradiation experiments by Findlay et ál. [9] have shown that the release fractions for the short-lived xenon isotopes increase with increasing half-life, so that the release of 133Xe (5.3 days) is greater than that of 135Xe (9.2 hours) and very little release of 137Xe (4 minutes) occurs. Thus these data are consistent with the observations on caesium in the plena. However, more detailed examination reveals that the iso topic composition of caesium in the plena is not markedly dependent on heat rating, which largely determines fuel temperatures, and is affected by plenum volume, which influences the volume of xenon transferred from the fuel. It appears that delays between release from the fuel and arrival in the plenum are also important in determining the isotopic composition of caesium in the plena. Comparison of the observed compositions with the fission yields indicates a mean age of 20 - 30 hours for xenon reaching the plena. Analysis of samples taken from the top uranium dioxide insulator pellet and the adjacent enriched fuel in DFR fuel pins has revealed not only sharp increases in the total caesium content (Fig.4), but also iso topic compositions more like those in the plena than in the bulk of the fuel. These observations most probably result from transfer of caesium from the plenum and are consistent with observations at Argonne of the forma tion of caesium uranate. 3.2.2. Halogens Analytical difficulties have impeded studies of halogen distribution in fuel pins since there are no reasonable long-lived radioactive isotopes of iodine or bromine (virtually complete decay of 131I occurs before destructive examination can begin) and the stable species are produced in rather low yield. Postirradiation annealing experiments have shown that release fractions for iodine are similar to those for the inert gases and extensive migration is therefore likely. Gamma-scanning data and analysis of fuel pin samples have revealed migration of 131I to the cooler end of a few pins and Davies et al. [10] has reported a similar effect from analysis of cladding samples. These observations of caesium and iodine concentrations at similar locations have generally been assumed to provide confirmation of the predicted formation of caesium iodide, but the extent of migration is somewhat surprising in view of the low vapour pressure of the compound at fuel pin cladding temperature. 3.2.3. Refractory oxide-forming elements The rare-earth elements and yttrium, zirconium, barium and strontium are all powerful oxide formers and their oxides are refractory, non-volatile materials. Thus little migration would be expected and the available data generally support this view. For example, no radial IAEA-PL-46 3/21 389 V o id I Large Equiaxed G roins | Columnar G ra in s' | ICIod □ 0 0.2 0.4 0j6 0.8 1.0 Fraction of fuel radius FIG. 6. Radial distribution of fission products in a 7-mm pin. migration of 144Ce was found even in a 7.0-mm pin irradiated in D F R at about 75 kW/m (Fig.6 ). Similar observations have been obtained by other workers [11] for 141 Ce and electron-beam microanalysis has indicated no radial migration of lanthanum, cerium, praseodymium, neodymium or samarium [12]. On the other hand, short-range migration of 140Ba/La has been shown (by gamma-scanning techniques) to occur and implies a non-uniform distribution of the daughter isotope 140Ce, and evidence of radial migration of stable barium (mainly 138Ba) has been obtained by micr'oprobe techniques. Release and migration of xenon and caesium pre cursors undoubtedly accounts for these observations. The few data that have been obtained for strontium isotopes indicate that redistribution of both 89Sr and 90Sr can occur, as a result of release and migration of the krypton and rubidium precursors. Indications of more extensive re distribution of 89Sr than 90Sr are consistent with the markedly longer half-lives of the mobile precursors of the lighter isotope. Axial gamma- scans of short-cooled elements occasionally indicate migration of 95Zr and/or 95Nb. Since subsequent chemical analysis (for 95Zr only) has never confirmed the gamma scans and since also radial scans on longer-cooled elements do not show any migration (see Fig.6 ), it is believed that niobium migration, probably as a volatile oxide, may be responsible. 3.2.4. Noble metals The noble metals ruthenium, rhodium and palladium, which are expected, on thermodynamic grounds, to be present in elemental form, are frequently found as metallic inclusions in oxide fuels. The chemistry of these inclusions, which normally also contain molybdenum and technetium, has been described elsewhere by Bramman et al. [7]. 390 POWELL Evidence from beta-gamma autoradiography and from analysis of micro drilled samples indicates that considerable migration of ruthenium can occur. In one frequently observed pattern high concentrations are found at the fuel centre and the periphery of the columnar grain zone with low concentrations between these regions and intermediate concentrations in the equiaxed grain zone. However, departures from this pattern are too numerous to be regarded as exceptions. For example, in a 7.0-mm dia. fuel pin ruthenium concentration was high throughout most of the columnar grains and low and remarkably constant in the central, large grain core (Fig.6 ). The latter effect may be a result of operation with molten fuel. It is clear from these and other observations that the ruthenium-rich regions are also those with high concentrations of inclusions. Gamma scans of the highly rated 7.0-mm diameter pins reveal considerable downward movement of ruthenium; this is believed to be a further consequence of operating in central melting conditions. Little evidence of such movement is normally found in 5.08-mm diameter pins and so the ruthenium distribution should correspond to the burn-up profile in the fuel. However, dissolution and chemical analysis of fuel pin samples gives lower than expected concentrations of both ruthenium and molybdenum (Fig.7). There is a reasonably clear trend to increasing molybdenum content with increasing distance from the bottom of the pin (i.e. decreasing clad temperature) but the variations in concentration do not appear to result from axial migration of molybdenum since there are no enriched regions to compensate for the depleted parts of the fuel. No systematic trend is revealed by the ruthenium data. The most probable explanation for these low recoveries of molybdenum and ruthenium is incomplete dissolution of the metallic inclusions. The semi-metallic fission-product elements antimony and tellurium should also be present in elemental form and since their vapour pressures are high at moderate temperatures, radial migration is expected to occur. Microdrilling data generally indicate that this is so for 125Sb, which concentrates near the cladding, and Stalica and Seils [12] have recently observed similar effects on tellurium and tin. -2 00 0 +200 Distance from core centre, mm FIG. 7, Axial distribution of 95Zr, 106Ru, 144Ce and Mo in a 5-m m pin. IAEA-PL-463/21 391 4. INTERNAL CORROSION OF OXIDE FUEL PINS Several fission-product elements, particularly iodine, tellurium and caesium (in the presence of oxygen) are known to cause corrosion of stainless steels and concentration of these elements at the fuel periphery may have deleterious effects on fuel-element endurance. Although there is no evidence that internal corrosion has had a major influence in any recorded fuel pin failure, it is probably the most important technological aspect of fission product redistribution. Metallographie examination of sections from a large number of mixed oxide fuel pins irradiated in DFR has revealed that about one third of the pins has suffered intergranular attack at the hotter end of depths of 25 - 75 /um. Microprobe examination of corroded samples and considera tion of the fabrication and irradiation conditions of both corroded and corrosion-free pins have not so far led to unequivocal conclusions on the mechanism of corrosion or the species responsible. The cyclic, van Arkel process proposed by Johnson et al. 113], which involved the formation of volatile ferrous iodide at the cladding surface and its subsequent de composition in the fuel, is not believed to be important in DFR fuel pins. The reasons for rejecting this proposal are: (1) High iodine concentrations and internal corrosion are observed at opposite ends of the pin (2) Microprobe studies have not revealed iodine to be a major constituent of the corrosion product (3) There is no evidence of a significant increase in the iron content of the fuel. Evidence from out-of-pile studies indicates that an oxidation process involving fission-product caesium is the most probable cause of internal corrosion. Support for this conclusion is provided by the observation that corrosion occurs most frequently in high burn-up, high rating pins since these pins should release the largest amounts of fission products. Somewhat surprisingly, however, there is no evidence that the initial oxygen/metal ratio of the fuel (in the range 1.97 - 2.03) has any effect on corrosion. General Electric workers have recently reported a similar conclusion [ 14]. 5. CONCLUSIONS. Fission-product gas release increases with increase in fuel tempera ture and burn-up. The gas release fraction, R, for 5.08-mm dia. pins containing vibro-compacted (U, 15% Pu) oxide fuel is related to the mean burn-up, B% bu: F = 1 - f exp (- В / 6 ) Variations in maximum heat rating (in the range 30 - 45 kW/m) and in smear density (78 - 85% TD) have little effect on gas release in these pins. Further increases in heat rating (larger diameter pins) or plutonium 392 POWELL content and decreases in smear density to about 70% TD each produce higher gas-release fractions. Gas release from annual pellets is generally lower than from comparable vibro-compacted fuel but is less predictable because of uncertainties in fuel clad conductance. The migratory tendencies of the non-gaseous fission products are reasonably consistent with predictions based on oxygen-balance calcula tions. Fission products with high volatilities (alkali metals, halogens, antimony and tellurium) are most susceptible to redistribution, whilst those that form refractory oxides (alkaline earths, rare earths and zirconium) do not migrate unless they have reasonably long-lives mobile precursors. Detailed analysis of caesium distribution in a number of fuel pins indicates that this element is very mobile within the fuelled region, but reaches the plena only as its gaseous precursor, xenon. The noble metal elements and molybdenum and technetium are some what mobile, particularly in highly rated fuel pins, probably as liquid metallic inclusions. The major technological problem caused by fission-product migra tion is likely to be internal corrosion of the cladding. Intergranular attack, to depths of 25 - 75 /jm, has been observed in about one third of the pins irradiation in DFR. Corrosion susceptibility is greatest in high burn-up, highly rated fuel pins, but does not appear to be markedly sensitive to fuel stoichiometry for oxygen/metal ratios of 1.97 - 2.03. REFERENCES [1] LAWTON, H., BAGLEY, K. Q ., EDMONDS, E., TILBE, H .E., Fast Breeder Reactors, Pergamon, Oxford (1967) paper 4B/4. [2 ] BAGLEY, K .Q . , DONALDSON, D .M ., in Plutonium 1960. [3 ] SWANSON, K .M ., KERSWELL, A ., TRG Rep. 246 (D). [4 ] DIGGLE, W .R ., BLACKADDER, W ., TRG Rep. 603 (D). [5] DIGGLE, W .R., GAUNT, A.J. , MELHUISH, K. R., NEWMAN, E. E., TRG Rep. 1494 (D). [6 ] TENNIS ON, W ., CLARK, M .L . , TRG Rep. 1 1 63(D ). [7 ] BRAMMAN, J . I . , SHARPE, R .M ., THOM, D ., YATES, G ., J. Nucl. Mater. 25 (1968) 201, [ 8 ] IENNISON, W ., CLARK, M. L. , TRG Rep. 1382 (D). [ 9 ] FINDLAY, J. R. , WATERMAN, M .J ., BROOKS, R .H ., TAYLOR, R. G. , J. Nucl. M ater. 35 (1970) 24. [1 0 ] DAVIES, J .H ., BOYLE, R. F. , HANUS, J .F ., Rep. C EA P-5100-4. [1 1 ] ARGONNE NATIONAL LABORATORY, Rep. ANL 7357. [12] STALICA, N. R ., SEILS, C .A ., Ceramic Nuclear Fuels (Proc. Symp.), American Ceramic Society (1969). [1 3 ] JOHNSON, C ., CROUTHAMEL, C ., CLEN, H ., BLACKBURN, P ., Trans. Amer. N ucl. Soc. 12 (1969) 565. [1 4 ] GENERAL ELECTRIC, Rep. CEAP 12008-40. IAEA-PL-46 3/22 REACTION OF SODIUM WITH URANIUM-PLUTONIUM OXIDE AND URANIUM OXIDE FU E LS P .E . BLACKBURN Atgonne National Laboratory, Argonne, 111., • United States of America Abstract REACTION OF SODIUM WITH URANIUM-PLUTONIUM OXIDE AND URANIUM OXIDE FUELS. The interaction between sodium and uranium-plutonium oxide fuel has been investigated at Argonne National Laboratory. These investigations include: identification of the reaction product; thermodynamic properties of the reaction product; phase and equilibration studies. The product of the reaction between UO2 + x and sodium is Na3U04 and between U-Pu oxide and sodium is Na3U]_yPuy04. The thermodynamic studies of Na3U04were used to calculate the O/M ratio of the fuel and oxygen concentration of the sodium. These calculations are compared to the results of measured fuel O/M ratios and sodium oxygen concentrations in equilibrium with Na3L']__yPUy04. In a liquid metal fast breeder reactor, the chemical interactions of the sodium coolant with the uranium-plutonium oxide fuel in the event of a breach in the stainless steel cladding can cause fuel swelling and cladding rupture. The severity of the problem may determine how long failed fuel elements can be left .in the reactor core and also whether or not failed fuel elements can be stored ,in the reactor sodium coolant. Studies of the sodium-fuel reaction have been in progress at Argonne National Laboratory for the past three years. These studies have included identification of the reaction product, measurements of thermo dynamic properties of the reaction product, equilibration of fuel and sodium to establish equilibrium oxygen contents in the fuel and in the sodium, and investigation of the reaction mechanism. This paper summarizes the work that has been published, reinterprets some of the earlier results with new data, and reports recent unpublished results. 1. Reaction Product Although no reaction occurs between stoichiometric UO 2 and oxygen- free sodium at normal reactor temperatures, the presence of excess oxygen, i.e., as UO 2+X or oxygen dissolved in the sodium, leads to the formation of a voluminous compound of ЫазШд (theoretical density = 5.6 g/cm3). A number of other products (i.e., Na 2Û, NaUO^, Na 2ÜÛ 4 ) have been postulated by others as the reaction product, but only ЫазШд can be in equilibrium with sodium and UO 2 . Pepper, Stubbles, and Tottle [1] (PST) were the first to identify NaßUO^ by chemical and mass-balance analysis as the phase which can exist in equilibrium with sodium and UO 2 . These authors also established the structure and lattice parameter of Na 3ÜC>4 as face-centered cubic (fee) with a = 4.79 to- 4.80 A. Bertram and Fryxel [2] found the structure of the reaction product to be more complex than that proposed by PST. In addition to the strong lines found by PST, Bartrum and Fryxell found 393 3 9 4 BLACKBURN FIG. 1. Isothermal section of the N a-U-0 phase diagram at about 800° C. many additional lines which they indexed as a cubic cell with the space group P4232- On the basis of the X-ray data alone Bartram and Fryxell proposed the formula NanUsOjg for the compound. However, because of the presence of other phases they were unable to confirm this composition by chemical analysis." Martin and Mrazek [3,4] (MM) have obtained X-ray data for the reaction product which agrees with those of Bartram and Fryxell, but chemical analysis of the compound by MM shows it to have the composition NaßUO^. Furthermore, MM found that the reaction product must have a fairly narrow composition range because slight variations in overall concentration always produced at least one other phase. Blackburn, Martin, Battles, O'Hare and Hubbard [6 ] established the phase relations among UO 2 , Na,- N a ß U O ^ , N a 2U 20y, a n d N 341105 for the Na-U-0 system which showed in agree ment with PST that N 33UO 4 is the phase in equilibrium with UO 2 . T h e phase relations are shown in Fig. 1. Martin and Mrazek [3,4] have also equilibrated U0.8Pu0.2^2-x with sodium to find by X-ray analysis that the reaction product for the mixed oxide is Na 3U ] _ „ P u „ 0 4 (ШазМОд). T h e compound is isomorphous with НазТЮд and has a Pu/U ratio which is nearly the same as the U-Pu-oxide from which it is formed. X-ray analysis of N a 3M 04 showed it to have a fee structure with a lattice parameter of 4 . 7 7 X. IAEA-PL-463/22 395 2. Equilibrium Conditions The reaction of sodium and fuel results in three phases (sodium, fuel, and NaßMO^) in local equilibrium. The important variable is the pressure of the1oxygen in equilibrium with these three phases. If both the oxygen pressure produced by the fuel and that produced by oxygen dissolved in sodium are less than the pressure in equilibrium with the three phases, no reaction will occur. Oxygen pressures greater than the equilibrium value will cause the reaction to proceed until the oxygen pressure is reduced to the equilibrium value. Recently, measurements have been made to acquire thermodynamic data f o r N a 3Ü 0^. These data include enthalpy of formation by two methods: mass-spectrometric measurements of sodium and oxygen pressures over the Na-U-0 system [5,6] and solution-calorimetric experiments on ШазИОд [3]. Heat-capacity data for ЫазШд were obtained from 5 to 350°K with an adiabatic calorimeter [7]. In addition, the enthalpy of ШазШд w a s measured between 500 and 1200 К by drop calorimetry [8]. N o Ut-No FIG. 2. Na-U-0 phase diagram. The three-phase regions studied by mass-spectrometry are labelled A to E. 396 BLACKBURN The mass-spectrometric experiments have been carried out to determine the oxygen pressure in equilibrium with the three-phase system Na-U02- NA 3UO 4 . In these investigations, samples (93.4 wt. % ЫазШ^-б.б wt. % UO 2 and 80 wt. % ЫазШд-2 0 w t % U O 2 ) were contained in an iridium-lined tungsten Rnudsen cell. The vapor effusing from the heated cell was analyzed with a Bendix time-of-flight mass spectrometer. In the initial experiments, it was found that the oxygen pressure in equilibrium with N a - N a 3-UÛ 4- ü 02 (region E in Fig. 2) could not be measured directly. It was also demonstrated that sodium was the only vapor over .N33110д w i t h a pressure high, enough to measure with the mass spectrometer. Thus, only sodium pressures in three adjacent three-phase regions (В, C, and D in Fig. 2) could be measured before reaching a region A where both sodium and oxygen pressures could be measured. The oxygen pressures in each of the three-phase regions may be calcu lated stepwise from the measured oxygen pressure in region A, the measured sodium pressures in regions A through D, and literature data for the sodium pressure over liquid sodium for region E. The steps for this calculation are given below. Regions A and В have phases Na 2Ü 0¿, a n d N a U 03 in common. Thus, the equilibrium constant K4 for the reaction 2 N a 2U 0 4 (s) = 2NaU0 3 (s) + 2Na(g) + 0 2 (g) (1) applies to both regions A and B. Since the compounds Na 2Ü 0^ a n d N a U 03 are at unit activity, the equilibrium constant may be written as K1 ■ (V > 0 2>A - (Р«а ^ (Р02>В <2> (The subscripts A through E refer to the regions A through E i n F i g . 2) The oxygen pressure in region В may be found by rearranging Eq. (2): = (PNa^A(P02^A 1 ^Na^B The oxygen pressures in the other regions may be calculated from measured quantities using the following equations: = ^ P 0 2 ^A(PN a ^ PNa^B ¡ ( 4 ) ^P02^D = (P0 2 ^A(Na)A%a'B^PNa)C ¡ (PNa^D (5^ ^P02^E = ^O^A^Na^A^Wß^Na^C ^ ^NaM^a^E ^ Three series of weight-loss and temperature-dependency measurements were conducted, and an accurate account was kept of the total weight loss of the sample. These measurements were carried out between 800 and 1300 C. Also, samples were taken for X-ray diffraction analysis at the conclusion of each experiment. The X-ray data were used to establish phases present in vaporizing sample and to aid in construction of the phase relations s h o w n i n F i g . 1. IAEA -PL-463/22 ;- 3 9 7 Tahle I. Oxygen and Sodium Pressures in U-O-Na System at 1200°K Three Phase R e g i o n i n F i g . 2 A в С D E P h a s e s N a 2U 20 7 _ N 3 oU 0 . " N a . U O ” N a . U O . " N a oU 0 . _ 2 4 4 5 3 4 3 4 N a U 0 3 " N a U 0 3 “ N a U 0 3 _ N a U 0 3 " u o 2 N a „ U 0 . N a , U 0 _ N a , U 0 , d o 2 N a 2 4 4 5 3 4 2 x 10- 7 2 x 10- 1 1 2 x 1 0 " 5 1 x 10~4 1 .2a PN a (atm) 00 1 1—1 0 - 2 3 P (atm) X 2 x l O " 1 1 1 x 10- 1 2 3 x 10 3 x 10" 27 2 3 Pressure over liquid Na (JANAF Tables, Dow Chemical Corp., Midland, M i c h i g a n ) Because the ionization cross sections of sodium relative to oxygen are not well known, the pressure of sodium and oxygen measured in region A were established by the following procedure. The sodium pressure was found from the measured Na+ ion intensity and from a calibration of ion intensity versus pressure. The calibration was established when sodium was vaporized from regions В and С (Fig. 2). This sodium pressure (in region A), converted to effusion rate, was subtracted from the total effusion rate (determined by weight loss) to give the oxygen effusion rate which was used to calculate oxygen pressure in region A. Measured and calculated pressures in the Na—U—0 system at 1200°K are g i v e n i n T a b l e I. These data were used with auxiliary data to calculate an enthalpy of formation of N 33UO 4 (Table II).. The samples used for solution calori metry, heat capacity and enthalpy measurements were taken from a single b a t c h o f N 331104 prepared at Argonne. The thermodynamic properties of N 33X104 at 298.15°K зге summsrized in Table II. Note the excellent agreement for the enthalpy of formstion of N 33UO 4 by two independent methods. These dsta, toegther with auxiliary data, are used to calculste the equilibrium oxygen pressure over N 3-U 02-N 33Ü 04 for the reaction: N a 3U 0 4 = 0 2 + 3 N a ( 1 ) + U 0 2 (s) (7) The calculated oxygen pressures are those in equilibrium with stoichio m e t r i c U O 2 , confirming the observation that only UÛ 2+X resets with sodium. To calculate oxygen pressures for НазМОд and fuel, the difference in the free energy of formation of N 33MO 4 snd (U,Pu)0 2 is assumed to be the same as that for the difference between N 33404 a n d U O 2 . A correction is also 398 BLACKBURN Table II. Thermodynamic Properties of Na^üO^ at 298.15 - 1 - 1 41.35 c a l m o l d e g К CP - 1 - 1 s 47.37 c a l m o l d e g К дн° -477.7 k c a l m o l " 1 (Solution Calorimetry) дн° -477.8 k c a l m o l " 1 (Mass Spectrometry) made for the oxygen deficiency in the fuel. That is, the oxygen pressure is calculated for the reaction, Na.U. Pu 0. = + 3Na + U. Pu 0. (8) 3 1-y y 4 2 2 1-y у 2-x The oxygen potential for reaction (2) may be calculated from Лб0 2 = ifc [AGf(Na 3U04) - (1 - f + f ) A G f (UO^ + f i G f(PuO,) - З Д С ^] (9) w h e r e t h e AGf's are free energies of formation and AG's are oxygen and sodium potentials. The sodium potential in the case of sodium-fuel contact is zero. If the fuel surface is hotter than the sodium, the sodium potential will be negative. The calculated equilibrium oxygen content of the fuel is uncertain because experimental measurements of oxygen potential data for the fuel are in poor agreement. The measured oxygen potentials [9—11] of gPug 2^2-x at oxygen-to-metal (0/M) ratios of 1.91, 1.95, 1.98, 2.01 Ü,0 . and 2.05 are plotted in Fig. 3. Woodley's measurements of U-25% Pu oxide were normalized to 20% Pu to produce the points in Fig. 3. In addition to the measured points, calculated oxygen potential curves based on Blackburn's thermodynamic model for the fuel are also given [12]. These same oxygen-potential data (measured and calculated) were extrapolated to lower temperatures to calculate the oxygen content of the U-20% Pu oxide fuel in equilibrium with sodium and Na^MO^. The results are shown in Fig. 4 together with measured fuel oxygen contents after equilibration [4]. The wide scatter in the experimental oxygen potentials and their variation with temperature are reflected in the divergence in the cal culated fuel oxygen concentration in equilibrium with sodium and ИазМ0 4 . The fuel oxygen contents calculated from the oxygen-potential thermo dynamic model and equation (9) are in better agreement with the measured oxygen concentrations than are those determined from the extrapolated measured oxygen potentials. Table III lists calculated oxygen concentrations for the 0, 10, 20, and 30% Pu oxide fuel using the model and equation (9). Measured oxygen con centrations are also listed for 0, 10, and 20% Pu oxide [4]. FIG. 4 . Calculated and measured measured and Calculated . 4 FIG. U-20% Pu OXIDE O/M RATIO ‘ x ± 2 2^ gPuQ Uq over potentials oxygen measured and lated calcu f o Comparison 3. FIG. 2.00 1.91 1.92 1.94 1.93 1.99 1.95 1.96 1.97 1.98 00 1100 0 00 1 0 0 9 0 0 8 0 0 7 0 0 6 0 0 5 0 0 4 0 0 3 / I I / OOO 1500 2000 2500 2000 0 0 2 0 0 5 2 0 0 0 2 0 0 5 1 O O IO ^martin &nd —V 3 ...... JÜO >!LLr— J____L U-2CP/o and ------IAEA-PL-463/22 _ mraz u xd OM ais n qiiru wt N ad a) 04, Na:)M and Na with equilibrium in ratios O/M oxide Pu T E M P E R A T U R E , °C , E R U T A R E P M E T T E M P E R A T U R E ,°K E R U T A R E P M E T |& 0 £ J ____L ALS S L IA T N E T O P N E G Y X O ALS L IA T N E T O P N E G Y X O D E T A L O P A R T X E D E R U S A E M NEARLY L R A E IN L O/M D E R U S A E M _L Table III Calculated and Experimental Isothermal Oxygen Contents of Fuel and Sodium for Fuel-Sodlum Reaction u o 2 p p m U0.7Pu0.3°2 U 0 . 8 P u 0 . 2°2 U 0 . 9 P U 0 . 1 ° 2 i n N a O/M O/M O/M O/M °2 b b 1 b e T e m p ,°K Temp,° F c a l c a c a l c a m e a s c a l c a m e a s c a l c ‘ m e a s c a l c c c a l c ^ m e a s 6 0 0 6 2 1 1 . 9 3 j 1 . 9 5 / - 1 9 8 0 - 2 . 0 0 - .01 .01 - 70 0 8 0 1 1 . 9 3 7 1 . 9 6 0 - 1 9 8 г - 2 . 0 0 - .20 . 10 - 8 0 0 9 8 1 1 . 9 4 2 I .9 6 3 ( 1 . 9 5 5 ) f 1 9 82 - 2 . 0 0 - 1.6 .54 - 9 0 0 116 1 1 . 9 4 5 1 . 9 6 4 - 1 ( 1 . 9 8 ) f 2 . 0 0 2 . 0 0 8 . 0 2 . 0 - 9 8 3 1 0 0 0 13 4 1 1 . 9 4 ? 1 . 9 6 6 - 1 9 8 , ( 1 . 9 8 ) f 2 . 0 0 2 . 0 0 30 5 . 9 0 . 2 1 1 0 0 152 1 1. 9 4 1 . 9 6 ? 1 . 9 5 8 1 1 . 9 8 2 . 0 0 2 . 0 0 95 15 • 0 . 3 ” 4 1 2 0 0 170 1 1 . 9 5 0 ! . 9 6 8 . 1 . 9 5 9 1 1 . 9 8 2 . 0 0 2 . 0 0 170 22 0 . 5 * 8 4 a Using Blackburn oxygen potential model and Na^UO^ thermodynamic data [12]. ^ A. E. Martin and F. C. Mrazek [4]. C Using oxygen solubility data of T. F. Kassner and D. L. Smith [13] ^ Using oxygen solubility data of K. T. Claxton [14]. e D. L. Smith [15] May be kinetically controlled. IAEA-PL-463/22 401 An estimated uncertainty in the oxygen potential of ±3 kcal results in O/M uncertainties from ±0.004 (10% Pu, 1200°K) to ±0.02 (30% Pu, 600°K). At a given temperature, the equilibrium oxygen content of the sodium for fuel-sodium-Na^O^ equilibrium is practically constant for either U0 2 or U-Pu oxide fuel. However, as shown in Table III, the equilibrium oxygen content of the sodium is strongly dependent on the temperature at the sodium-fuel interface. These calculations are based on solubility data for Na20 in Na, [13,14] Henry's Law, and oxygen potentials from equation (9). That is, AG02/2 - A G f (N a 2 0) p p m 0 = p p m 0 (s) • e x p (------— ------) ( 10 ) w h e r e p p m 0 is concentration of 0 in parts per million by_weight, ppm 0 (s) is sodium oxygen concentration saturated with Na 20, A G 0 2 is t h e equilibrium oxygen potential and AGf(Na 20) is the free energy of form ation of Na 20. The two sets of calculated oxygen concentrations in Table III are based on two sets of Na20 solubility data. The calculated oxygen concentration of sodium is completely independent of the fuel oxygen potentials so that uncertainties in the fuel oxygen potential will not effect the sodium oxygen concentration. Table III also lists Smith's J15J experimentally measured oxygen concentrations in equilibrium with sodium, fuel, and product. In these measurements, uranium-plutonium oxide pellets with known initial O/M ratios were equilibrated with sodium and a vanadium wire. A measurement of the internal friction of the vanadium wire and a calibration curve established the vanadium oxygen content. The vanadium oxygen content and a calibration curve for oxygen dissolved in vanadium versus that dissolved in sodium established the sodium oxygen content. The measured sodium oxygen concentrations are much lower than those calculated from equation (10) using oxygen potentials calculated with equation (9). Much better agreement is obtained if oxygen potentials measured by Javed and Roberts [11] for O/M = 1.96 are used. However, Javed and Roberts measurements are 10 to 20 kcal lower than all other measurements. In addition, Javed [16] has used the same technique to measure oxygen potentials of U 0 2_x , finding values which are about 10 kcal more negative than other measured and calculated oxygen potentials, suggesting a systematic negative bias in his oxygen potentials or positive bias in his oxygen analysis. Another difficulty in these calculations is the large temperature gap between the oxygen-potential measurements of thé fuel and the measurements of Na20 solubility in sodium. The minimum temperatures for the fuel oxygen-potential measurements were 800 [7], 950 [8], and 1000 [9]°C. The maximum temperature for measurement of Na20 solubility in sodium is 550 С [14]. Thus either set of measurements must be extrapolated several hundred degrees to calculate the oxygen concentrations in sodium and fuel at equilibrium. The poor agreement of these extrapolations (see Fig. 4) is caused by the large differences in the temperature dependence of A G 0 2 as well as by the differences in the measured values. The data in Table III refer to the conditions prevailing at the sodium- fuel interface. That is, for the temperature at the interface and for a fuel of given plutonium content, the O/M ratio of an oxygen-rich fuel will be decreased to the level indicated. Other calculations, based on densities of the fuel and the product, indicate a volume increase of 402 BLACKBURN about 1.0% for each 0.01 O/M decrease in the fuel oxygen concentration as the fuel reacts to form ЫазМОд. One of the problems still to be resolved is the distribution of oxygen expected in the rest of the fuel— away from the sodium-fuel interface and toward higher temperatures. Further work is required to establish the oxygen distribution in fuel under normal operating conditions. Work at Argonne [17] involving measurements using irradiated fuels, has not yet included fuel with an O/M ratio as low as that expected after reaction with sodium. At GE, oxygen distribution in out-of-pile, thermal-gradient experiments is being measured with and without sodium [18,19]. 3. Reaction Between Sodium and Fuel Sodium has been reacted with fuel under isothermal conditions [3-6] and in a thermal gradient [18,19]. An example of the reaction between sodium and U q gPuo 2^2 (initial O/M = 1.995) pellets [4] at 9 0 0 ° C (2 days) is shown in Fig. 5. The dark grey material is НазМОд, which forms a surface layer'and surrounds the lighter fuel-oxide grains. The dark circles are fuel porosity. The growth pattern appears to consist of the formation of a surface layer and also grain-boundary penetration. The initial grain-boundary penetration proceeds by formation of a very thin intergranular layer (as indicated by the thin lines). This is followed by thickening of the ЫазМОд l a y e r leading to cracking and further intergranular attack. In extreme cases the oxide crumbles into a powder. An objective of the isothermal study was to establish the equilibrium 0/M value of the mixed oxide as a function of temperature and of the Pu/(U + Pu) ratio of the mixed oxide. A lattice-parameter method was used. The reaction products were treated with ethyl alcohol to remove the free sodium and examined by X-ray diffraction. The lattice parameter of the mixed oxide was used to establish the O/M composition of the mixed— oxide phase. The measured O/M results are given in Table III. Both the 0/M ratio measured after equilibration, used with the X-ray lattice parameter, and calculations using the thermodynamic data for NaßUOz, and the oxygen-potential model show the equilibrium fuel O/M ratio for U 02~ 20% P u 0 2 is "vl.96. A further test of this composition was made with pellets of lower initial O/M ratios [20]. Mixed-oxide pellets with O/M compositions of 1.960 and 1.970 were prod uced for kinetic studies by hydrogen reduction at 1590 and 1530°C, respec tively. A comparison of the reaction of pellets of these compositions with sodium is of particular interest because the equilibration study had indicated that pellets with about the 1.960 0/M composition should not react with sodium at temperatures from 800 to 1100°C (see Fig. 3 and Table II). Figures 6 and 7 compare the reaction of pellets of O/M com positions 1.970 and 1.960 with sodium at 900°C for two days. Figures 8 and 9 make a similar comparison for reaction for two days at 700 C. It is apparent that the pellets'with O/M = 1.960 did not react at either tem perature in the two-day reaction time. On the other hand, the pellets with O/M = 1.970 showed evidence of surface reaction, with a greater amount of reaction at 900 than at 700 C. " A reaction layer is evident in the reaction at 900 С but only grain-boundary attack at 700 C. Thus, these tests confirm the findings of the equilibration study insofar as a two- day test at these temperatures is concerned. FIG. 6. Surface region of mixed-oxide pellet, Pu/(U +Pu) = 0. 20, O/M = 1.970, after exposure to sodium for 2 days at 900°C; 27OX, as polished. IAEA-PL-463/22 IAEA-PL-463/22 4 0 5 FIG. 7. Surface region of mixed-oxide pellet, Pu/(U + Pu) = 0.20, O/M = 1.960, after exposure to sodium for 2 days at 900°C; 270X , as polished. 406 BLACKBURN FIG. 8. Surface region of mixed-oxide pellet, Pu/(U + Pu) = 0. 20, O/M = 1. 970, after exposure to sodium for 2 days at 700°C; 270X , as polished. : AAP-6/2 7 0 4 IAEA-PL-463/22 FIG. 9. Surface region of mixed-oxide pellet, Pu/(U+Pu) = 0.20, O/M = 1.960, after exposure to sodium for 2 days at 700°C; 270X , as polished. 408 BLACKBURN Data from the thermal-gradient experiments £18, 19] are in general agreement with those obtained in the isothermal experiments. That is, U-20% Pu oxide is reduced to an O/M ratio of slightly less than 1.96 for U-20% Pu oxide. Aitken has discussed out-of-pile and in-pile experiments on the sodium-fuel reaction [18]. He assumes that the oxygen concentration threshold for reaction in sodium is 10-100 ppm. This corresponds to calculated concentrations for the higher fuel-sodium temperatures in Table III. Smith's measured oxygen concentrations are much lower than those calculated. Because of the assumed high equilibrium oxygen con centrations in the sodium, it is concluded that the sodium-fuel reaction will be limited or even reversed by sodium removal of oxygen. However, since the oxygen-content specifications for sodium are between 1 and 5 ppm, onlg those reactions in which the sodium-fuel interface is in excess of 600 С (or higher based on the measured values) can involve oxygen removal. Where the removal can occur, its removal rate must be faster than the rate of reaction of sodium with fuel to keep the reaction from causing the maximum possible destruction of the fuel pin. To establish if the reaction can be limited by sodium carrying away oxygen, one needs data for the oxygen-removal rate and for the sodium- fuel reaction rate. No data exist for the former. For reaction rates of sodium with fuel, the data are still incomplete. Martin and Mrazek have shown that the reaction rate is temperature dependent [4]. For fine powder, the isothermal reaction reaches com pletion in <2 days at 900 C, <7 days at 800°C, and >8 days at 600°C. Some in-pile reaction rates of defected pins with sodium appear to vary as the fourth power of the time. The mechanism for the reaction has been proposed to be one in which the rate is controlled by sodium dif fusion from a cladding point defect through the product layer [17]. The product is assumed to form a hemisphere with its center at the point defect. However, the defected pins showing a one-fourth-power time dependence had slit defects rather than point defects, and even an initial point defect will form a larger opening in time. Furthermore, in deriving an equation for the rate of increase of fuel-pin diameter with time, an inappropriate expression was used for an assumed sodium diffusion-controlled reaction. It was assumed that the increase in mass of product with time is inversely proportional to the radius of the product hemisphere growing from the pin hole cladding defect (dm/dt = k/r). This leads to cladding swelling which varies as t 1/4. However, the assumed mass increase with time applies to slab growth, not spherical growth. For the spherical case, as in other geometries for diffusion-controlled growth, the rate of increasing thickness of the product varies inversely with the total thickness (dx/dt = k/x). This leads to parabolic growth or a t1/2 dependence for both the growth of product and fuel-pin diameter. Since the observed increase is not parabolic, it must not be controlled by diffusion of sodium. In fact, because of the surface attack, inter-granular attack, and fuel cracking, it is unlikely that any simple mechanism can be proposed for this apparently complex reaction. Another possible mechanism involves a rate controlled by diffusion of oxygen to the fuel-product interface. If the rate-limiting mechanism is diffusion of sodium through an impervious product layer, one would expect the product to form a layer of near-uniform thickness on the IAEA-PL-46 3/22 409 surface of the fuel. Since intergranular attack is observed, sodium- diffusion control appears unlikely. On the other hand, if diffusion of sodium to the fuel-product interface is faster than oxygen diffusion within the fuel, then intergranular attack could be explained as follows. Presumably sodium can move rapidly to the fuel-product interface, i.e., it is limited only by the rate of reaction at the interface. The movement of sodium may be by diffusion or vapor transport depending on the nature of the ЫазМОд film. The slow step is assumed to be oxygen diffusion from the fuel to the fuel-product interface. Since oxygen is believed to diffuse more rapidly along grain boundaries, more reaction occurs’ along intergranular paths than normal to fuel surfaces or grain surfaces. Under these conditions, the length of sodium-diffusion paths can show considerable variation (compare the surface-layer thickness to the circuitous length of the intergranular material in Fig. 5). Presumably sodium diffuses rapidly through these paths whereas oxygen moves slowly to the grain boundaries. These arguments refer to the case where the oxygen for the reaction is furnished by the fuel. Further work is in progress to establish the mechanism and rate of reaction between sodium and fuel. REFERENCES [1] PEPPER, R. T., STUBBLES, J. R., TOTTLE, R. C., Appl. Mater. Res. 3_ (1964) 203. [2] BARTRAM, S. F., FRYXELL, R. E., J. Inorg. Nucl. Chem. 32_ (1970) 3701. I3J O'HARE, P. A. G., SHINN, W. A., MRAZEK, F. C., MARTIN, A. E., J. Chem. Thermodynamics 4_ (1972) 401. J4] BLACKBURN, P. E., et al. Chemical Engineering Division Fuels and Materials Chemistry Semiannual Reports, ANL-7822 (1971) and ANL-7877 (1972) [5] BATTLES, J. E., SHINN, W. A., BLACKBURN, P. E., J. Chem. Thermodynamics 4_ (1972) 425. 16] BLACKBURN, P. E., MARTIN, A. E., BATTLES, J. E., O'HARE, P. A. G., HUBBARD, W. N., Proceedings of the Conference on Fast Reactor Fuel Element Technology (R. Farmakes, ed.) Amer. Nucl. Soc., Hinsdale, 111. (1972) 479. |7] OSBORNE, D. W., FLOTOW, H. E., J. Chem. Thermodynamics 4_ (1972) 411. [8] FREDRICKSON, D. R., CHASANOV, M. G. ibid 4_ (1972) 419. [9] MARKIN, T. L., McIVER, E. J., "Plutonium 1965" (Kay, A. E. and Waldron, M. B. eds.) Chapman and Hall, London (1967) 845. [10] WOODLEY, R. E., American Ceramic Society Meeting, Anaheim, Calif. Conf.-711030-5, Nov. (1971). [11] JAVED, N. A., and ROBERTS, J. T. A., ANL-7901 (1972). [12] BLACKBURN, P. E., to be published. 410 BLACKBURN [13] KASSNER, T. F., SMITH, D. L., ANL-7335 (1967). [14] CLAXTON, K. T., J. Nuclear Energy, (1965) 849. [15] SMITH, D. L., Private communication (July 1972). [16] JAVED, N. A., J. Nucl. Mater. 43 3 (1972) 219. [17] JOHNSON, C. E., JOHNSON, t., BLACKBURN, P. E., CROUTHAMEL, C. E., "Stoichiometric Effects on U-Pu Oxide Fuel," IAEA (1972). 118] AITKEN, E. A., EVANS, J. K., MELDE, G. F., RUBIN, B. F., Proceedings of the Conference on Fast Reactor Fuel Element Technology (R. Farmakes ed.) Amer. Nucl. Soc., Hinsdale, 111. (1972) 459. [19] AITKEN, E. A., ADAMSON, M. J., EVANS, S. K., LUDLOW, T. E., GEAP 12210 (April 1971). £20] MARTIN, A. E., SCHILB, J. D., to be published in ANL Semiannual Report, January-July (1972). IAEA-PL-463/23 THERMODYNAMIC STUDIES OF THE SODIUM FUEL REACTION AT GENERAL ELECTRIC* M .G. ADAMSON. E. A. AITKEN, S.K. EVANS General Electric Company, Vallecitos Nuclear Center, Pleasanton, C alif., United States of America Abstract THERMODYNAMIC STUDIES OF THE SODIUM FUEL REACTION AT GENERAL ELECTRIC. Available thermodynamic information for the equilibrium system sodium (Í) + sodium-fuel compound + unreacted fuel is critically reviewed. Results are interpreted primarily in terms of the equilibrium oxygen concentration in sodium (C0) and the oxygen potential (ДG^) of the remaining fuel phase. Results of series of isothermal and non-isothermal capsule tests performed at General Electric are presented and an alternative experim ental approach utilizing a high temperature EMF ce ll is also described. Vanadium wires were used to determine C0 in the isothermal tests. Although there is significant scatter in the vanadium wire results, practically all the C0 values fall within the range 0.1 to 1 ppm (oxygen by weight) between 700 and 850°C. When expressed as oxygen potentials these data are considerably more negative than values obtained either from determinations of the O/M ratio of the unreacted fuel phase or calculations based on the known free energy of formation of Na3U04. Unusual kinetic features of the sodium-fuel reaction are suspected as a possible cause of this discrepancy, but there is also some doubt concerning the validity of the thermodynamic approximations used to describe dilute solutions of oxygen in sodium. 1. INTRODUCTION The principal manifestation of reaction between sodium coolant and mixed (U,Pu) oxide fuel in a defected LM FB R fuel pin is swelling [1] which may ultimately change the coolant flow and temperature distribution near the affected area. Previous work at General Electric [1] and at Argonne National Laboratory [2] has established the general conditions under which sodium will react with U-|.2puz02-x* The reaction product has been characterized as face-centered cubic Na3M Û 4 (where M = U].zPuz and a0 = 4.78 A) and it is formed according to re a c tio n ( 1 ): + 21^ jl\la( + which takes place in the temperature range 600 to 1100QCwhen x < 0 .0 4 (M = Ug 8qPu().2o)- Thermodynamic property measurements on the analogous/isomorphous compound 1Маз1104 have been made recently and these data used to calculate the corresponding oxygen potential (AGc^) and equilibrium oxygen level in sodium (C0) in the 3-phase (univariant) region Na - <М02.у> - <1\1азМ 04>. [3] In performing this calculation two assumptions are implicit: (i) the thermodynamics of Na3(U ,Pu)0^ are the same as those of isostructural М азиОд and (ii) dilute solutions of oxygen in sodium continue to obey Henry's law at 600 to 1100°C. To avoid the uncertainties inherent in this calculational approach we are attempting to measure directly both Д Б 02 and C0 for the (isothermal) reaction 2 ±v 3 {Na} + * Work-performed under USAEC Contracts AT(04-3)-189, PA-53 and AT(04-3)-189, PA-10. 'I'Througnout this paper the phase of a compound or element will be specified by the following parenthetical symbols: <> pure solid phase, j| liquid phase, () gas phase, [] dissolved phase (solid or liquid). 411 412 ADAMSON et al. A second objective of our present program is the determination of Co in the nonisothermal situation corresponding to a failed fuel element operating in LM FB R. It is the purpose of this paper to present our results to date and, in discussing these data, to review what is presently known about the thermodynamics of this intriguing chemical system. 2. EXPERIMENTAL Values for y in reaction (1) under a near-typical thermal gradient have been determined by heating sealed molybdenum capsules containing sodium and mixed oxide fuel pellets of various over-all stoichiometries for various times up to 1000 hours. The sodium was loaded into the cool end of each capsule which was then maintained at some temperature between 550 and 1100°C; the hot end of these capsules was usually ca. 1450°C. After completion of a thermal gradient anneal the capsules were unloaded and visual observations made to determine where reaction bands (roughening of the fuel surface near the low temperature sides of the pellets) began to appear and also where actual fuel pellet swelling began. Oxygen-to-metal ratio determinations were made on all the pellets which did not contain excess sodium. Isothermal anneals of closed capsules containing oxide fuel, excess sodium and vanadium wires are being performed in an attempt to determine the oxygen concentration in liquid sodium (C0) when oxide fuel is in equilibrium with its reaction product. Thus far the majority of tests have been run with urania or urania-ceria ^ 0 .8 5 ^ e0.15^2+x) as substitutes for mixed (U,Pu) oxide, it being our intention to use mixed oxide samples as soon as the method appeared reliable. High purity nickel (carbon < 0.002 wt%) is the usual capsule material; each thick-walled capsule being carefully cleaned and vacuum annealed at 1000°C prior to loading. Occasionally tests are performed in either stainless steel or molybdenum capsules. The capsules are charged with 2 to 5 g of oxide fuel, in the form of shards or powder, and approximately 5 g of sodium under purified helium. A vanadium wire is suspended from each capsule lid so that three inches of wire are immersed in the liquid metal at temperature. Arc-welding in a helium atmosphere is used to close the capsules after which they are annealed in an upright position at the required temperature (usually 750°C) in a stream of argon. At the end of the equilibration the system is quenched by dropping the capsule(s) into oil. The cooling rate achieved using this method is sufficiently high to prevent any significant back-equilibration of oxygen between the vanadium wire and liquid sodium during cooling. After removal from the capsules each vanadium wire is carefully cleaned and then analyzed for oxygen using inert-gas fusion; the procedure for this analysis is that recommended by the Argonne National Laboratory. [4] Smith and Lee's tables relating the oxygen content of vanadium to the oxygen content of sodium [5] are used to obtain the value of C0 in each experiment. Inert-gas fusion analyses run on oxygen standards and on blanks (i.e., unreacted vanadium wires) indicate that the standard deviation on a single measurement of the concentration of oxygen in vanadium, [0] y, is.± 14% and that the limit of detectability in the present equipment is equivalent to 0.02 wppm oxygen in sodium (at 750°C). The relative amounts of sodium, M 0 2 +x (M = U, U-|.2Cez or Ui_zPuz), and vanadium in these tests were chosen so that the presence of vanadium would cause only an imperceptible shift in the calculated М 0 2 -у/1МазМ04 equilibrium ratio (usually about 4:1, according to Equation (1). In converting values of [0] у to C0 it is assumed that the equilibrium described by Reaction (2) is reversible and, if perturbed, is rapidly re-established. Generally the capsules were static during the isothermal anneals but toward the end of the present series of tests some attempts were made to stir the reactants, either by periodic inversion of the capsule or by swirling only (no inversion). Some preliminary experiments have been performed with a high temperature EM F (oxygen-concentration) cell incorporated in a stirred stainless steel vessel containing sodium. This cell was designed to operate at 550 to 650°C, with the vessel containing two ports for solid oxide-electrolyte oxygen sensor tubes, gas inlet and outlet lines, and an additional port for sample introduction or withdrawal (i.e., uranium metal, or a U O 2 specimen, can be dipped into or withdrawn from the liquid sodium during operation). In this approach sodium, which constitutes the sample electrode of the electrochemical cell Me, Me Oxide (Pt)//Th02 ~6 wt % Y 2 O3// [0] |\|a , Reference Electrode Electrolyte Sample Electrode is titrated potentiometrically with oxygen in the presence of oxide fuel while its oxygen activity is continuously monitored as "cell E M F " (Note: for a cell incorporating coexistent Me, MeO as the reference electrode A G û 2 (sample) =4 F [EMF(volts)] + A G o2 (Me,MeO). At constant temperature a buffering effect should occur (provided the kinetics of Reaction (2) are rapid in comparison with the rate of addition of oxygen) as soon as the IAEA-PL-463/23 413 univariant 3-phase field j Naf — The oxygen sensors are made up from high purity thoria— 6 wt % yttria electrolyte tubest which are loaded with a mixture of copper, cuprous oxide and thoria— 6 wt % yttria (300-m esh powders) together with a directly bonded platinum current collector. These tubes are sealed into the reaction vessel using a frozen sodium seal/teflon sleeve combination. The reference electrode compartment of the oxygen sensor is connected to an external vacuum and argon back-fill system and is isolated from the atmosphere inside the reaction vessel; during cell operation this compartment contains argon at approximately one atmosphere pressure. Uranium foil contained in a small basket, which could be lowered into or raised out of the liquid sodium during cell operation, was used as an oxygen getter whenever it was desired to reduce the oxygen content of the sodium to less than 0.1 wppm. Oxygen was admitted into the cell quantitatively by bubbling a 0.5% oxygen in argon mixture through the stirred sodium via a calibrated flow gauge. After each addition the bubbler tube was flushed with argon and sodium was forced part way up the tube to remove possible sodium oxide deposits. 3. RESULTS AND DISCUSSION 3.1 Thermal Gradient Tests The results and experimental conditions for the thermal gradient series of experiments are summarized in Table 1. Preliminary tests (Capsule 1-5) indicated that the temperature of formation of the reaction compound TABLE 1. SUMMARY OF EXPERIMENTAL CONDITIONS AND RESULTS FROM THERMAL GRADIENT CAPSULES HEATED IN A TEMPERATURE GRADIENT Internal Heat Initial 2-y r Q* Na (mg) 0 :M T R (°C ) at T R Capsule Temperature (°C ) Treatm ent (h) T N <°C) (kcal/mole) Remarks 1 500-1150 100 35 1.965 500 750 1.965 NM* Not at steady state 2 900-1450 300 32 1.965 900 975 1.970 -5.7 Reaction bands 1000°C 3 900-1450 300 85 1.965 900 980 1.975 -5.7 Reaction bands < 1000°C 4 1105-1475 300 37 1.933 1110 1110 1.94-1.95 -2.4 Reaction at coldest pellet o nly 5 1240-1500 300 36 1.964 1240 NM -15.7 No reaction product, no visual evidence Na Capsule may have failed. 6 1105-1475 100 39 1.958 1060 ~ 1 1 0 0 NM — 1 inch long pellet at cold end. Not at steady state 7 1105-1475 1000 49 1.966 1110 - NM -15.4 No reaction product Capsule may have failed 8 1105-1475 300 138 1.80 1110 1190 1.832 -4.81 Some pellet swelling < 1 190°C 100% PuO-j 8 9 595-1330 1072 98 1.939 595 780 1.962 -5.1 Pellet surfaces dulled *NN1 = not measured TR = temperature where smelling occurs due to reaction of fuel with sodium Tfyj = temperature of sodium reservoir Q* = heat of transport tlsostatically pressed from 99.9(+)% purity material and fired at ca. 1900°C under vacuum; prepared at the G ER& D Center, Schenectady. 414 ADAMSON et al. depends upon the lowest temperature in the capsule (i.e., the sodium metal reservoir). The temperature of the cold reservoir, T|\|, also appears to influence the threshold oxygen activity or oxygen-to-metal ratio, (0 / M ) r , fo r formation of the reaction compound which suggests that the sodium activity is determining (О/M) p¡. When there is an excess of sodium in the system the sodium fugacity at temperature T r ( / r ) is determined by the fugacity at T|\| tf|\|) an£l the sodium activity at the reaction position can be expressed as / n a N = T (4) ■fa w h e re / r is the fugacity of sodium vapor over liquid sodium at T r . Thus we can write for the free energy change in Reaction (2) A G T ( = A /°( N a 3M04)T -AG /O (M02_y)T) = -RT Sn{aNa3M04 y / (y f°)3 . / T 7 | (5) where AG^denotes standard free energy of formation and /02 is the fugacity of oxygen. If a jj^M o ^is set equal to unity and the fugacity of oxygen is equated with its partial pressure a simplified expansion results which, in principle, allows A G ^ I^ M O ^ y to be calculated from knowledge of (О/M) r , T r and the fugacity-temperature relations for sodium vapor. There was evidence for the presence of free sodium (or Na2 0) in the pellets at the lower temperature end (below T r) of all the capsules except capsule 11. Generally it was found that reaction bands occurred at higher temperatures than actual fuel swelling and furthermore the corresponding (0/M )r values were not significantly dependent on the over-all oxygen activity of the fuel (initial 0:M ). The results of the present experiments are shown in Figure 1 together with recent Argonne National Laboratory data [6] from isothermal capsule tests. There is reasonable agreement with the Argonne data when reaction bands are considered to be the point at which the reaction proceeds in the low stoichiom etry capsules. Where swelling is considered, however, all the data seem to lie higher in stoichiometry. Since sodium was placed in the high-temperature ends of these capsules at the beginning of these experiments, sodium vapor would have passed over all the pellets while migrating toward the cold end of the capsules before condensation. In the capsules with 1.964 initial stoichiometry, the sodium would have had an opportunity to react with these pellets before the establishment of the stoichiometry gradient, resulting in indications of reaction or swelling early in the experiment which have not been indicative of the steady-state situation. However, in the low stoichiometry case, the average oxygen-to-metal ratio in those pins was low enough that no reaction with sodium should have occurred early in life. Once the stoichiometry gradient was established, reaction with sodium vapor would have occurred at the proper location. Therefore, on the basis of comparisons with the Argonne data, it appears that the assignment of the threshold reaction temperature and stoichiometry at the point where reaction bands begin to be observed is proper. A limitation of this thermal gradient capsule technique is that the determined values of (0/M )r are insufficiently precise to warrant application of Equation (5) in an attempt to confirm the predicted sodium activity effect. Strictly speaking, this effect would be expected to produce somewhat different (O/M) p values to the Argonne results since the latter all refer to sodium at unit activity. It may be significant that the greatest discrepancies between these two sets of results occur at the highest temperatures, where (in the case of the thermal gradient tests) the sodium activity is significantly reduced below unity. The heats of transport of oxygen (Q*) determined in this investigation (see Table 1) indicate that sodium is participating in the over-all oxygen transport process, most probably as a gas phase Na- 0 species; this and other oxygen thermal diffusion phenomena related to mixed oxide fuel form the subject matter of a companion paper. [7] 3.2 Isothermal Capsule Tests The results and conditions of tests performed to date are summarized in Table 2. In some of the initial tests Type-304 stainless steel was included in the capsules to see whether the j Na[ - [Cr] s s - Tr (“ci FIG. 1. Comparison of thermal gradient sodium-fuel reaction data with ANL isothermal data. Generally, as written in Table 2, the C0 values have an associated error of ± 14%, which is the mean standard deviation measured at the oxygen levels in question. Where larger errors, or value ranges, are presented it indicates that duplicate analyses were performed on the vanadium wire in that experiment - with significantly different results. In later tests only single determinations were made on the vanadium wire. An x-ray powder pattern was taken of the product from test 4 after separation of excess sodium by vacuum distillation. The pattern showed the product was a mixture of the two face-centered cubic phases N831)04 (aQ = 4.796 ± 0.001Â) and U 0 2+x (a0 = 5.469 ± 0.001Â) in the approximate ratio 1:1 which corresponds with expectations. The absence of other phases and the highly crystalline character of the N831104 product'lends confidence to our assumption that thermodynamic equilibrium is attained within the time of these tests. According to the available thermodynamic data for Reaction (2), when M = U [3] the final O/U ratio of unreacted urania should be 2.000 which, within the uncertainty of the determination, is confirmed by the measured и 0 2+х lattice parameter. In certain capsules (e.g., tests 4, 20, and 21) a small am ount of sodium oxide was added to the sodium to give an initially high oxygen level - approximately 1000 wppm. In others (e.g., tests 16 and 17) a large amount of l\la20 was used, together with stoichiometric urania, in an attempt to approach equilibrium from another direction, i.e., with all the available oxygen initially in the sodium. In these latter two experiments the vanadium wire was coiled at the top of the capsule so that it did not initially contact the sodium; after 24 hours at 850°C the capsule was inverted to bring the sodium/fuel system into contact with the wire. On opening both these capsules no vanadium wire could be found which indicates that the wires had oxidized to the point of disintegration (> 9 0 wppm 0 in Na) at some point during the anneal. Presumably 24 hours at 850°C was not sufficient time for all the sodium oxide to react with the UC^.oo- It is evident from Table 2 that there is considerable scatter in the measured equilibrium oxygen levels (C0) for the sodium-fuel reaction at a given temperature. In fact the variability is so great that it has not been possible to draw any conclusions about the effect of temperature on C0 on the basis of measurements over 100° temperature ranges. Generally the measured C0 values for U 02+ x, Uo.85^e0.15^2+x ап[* U0.75PU0.25Û2 at 700 to 850°C lie between 0.1 a n d 1 wppm. Subsequent discussion will be based upon these figures yet it is clearly desirable to inquire further into the reason for the observed irreproducibility of results. TABLE 2. ISOTHERMAL (NICKEL) CAPSULE TESTS (ALL TESTS PERFORMED AT 750° ± 2°C IN Ni CAPSULES EXCEPT WHERE OTHERWISE NOTED) Test No. Tim e [О ] у C o (Capsule No.) (hours) Oxide Sample Additions. Comments (wt % ) (ppm ) 1 115.0 UgOg; 1.5 gm powder M o Capsule 0 .5 6 ± 0 .0 8 0 .6 6± 0 .2 6 2 (A 1 ) 80 .0 UgOg; 2.5 gm powder 0.1 4 ± 0 .0 6 0 .0 3 ± 0 .1 0 3 (A 2 ) 80.0 UO 2 1 9 ; 4 gm shards 0.07±0.01 0.025±0.005 4 91.5 U 3 O 3 ; 2.5 gm powder № 2 0 (2 0 mg) 0.3 6 ± 0 .0 5 0 .2 6± 0 .0 8 5 91 .5 UO 2 jg; 3.8 gm shards 304 Stainless (2 gm) 2.38 > 1 5 6 91 .5 - № 2 0 (20 mg); 304 Stainless (2 gm) 0 .06±0.01 0.021±0.003 7 91 .5 - Na, as loaded 1.3 5± 0 .1 9 1 0±4 8 (A 6 ) 90 .5 UgOg; 2.5 gm powder 0 .0 8 -0 .6 0 0 .0 3 -0 .7 9 9 (A 7 ) 90.5 изОд; 2.5 gm powder Type 304 Stainless (2 gm) 1 .31±0.19 9 ± 4 1 0 (A 8 ) 90 .5 11зОд; 2.5 gm powder Type 304 Stainless (2 gm) 0 .4 0 ± 0 .1 4 0 .1 5 -0 .6 2 1 K A 9 ) 90.5 U0.85Ce0.15°2+x; 5 am Pow der 0 .3 5 ± 0 .1 0 0.14-0.41 1 2 (A 1 0 ) 90.5 u0.85Ce0.15°2+X' 5 9 m pow der 0 .5 4 ± 0 .0 8 0 .6 H 0.2 1 1 3 ( M 0 1 ) 4 8 .0 u0.75Pu0.25°2.00; ^ 9m shards 0 .5 0± 0 .0 7 0.5 1± 0 .1 6 1 4 ( M 0 3 ) 1 1 2 . 0 u0.75Pu0.25°2.00' 5 9 m shards (T = 700°C) 0.21 ±0 .0 3 0.04±0.01 0 .7 9 ± 0 .1 1 1 5 ( M 0 4 ) 4 8 .0 u0.75Pu0.25°2.00' 5 9 m shards 1 .6 ± 0 .6 1 6 IA 1 1 ) 53.0 UO 2 .0 0 ' 2.5 gm powder Na20 (0.375 gm );850°C* Wire disintegrated 1 7 ( A 1 2 ) • 53.0 UO 2 0 O'’ 2-5 gm pow der Na20 (0.375 gm );850°C* Wire disintegrated 1 8 (A 1 3 ) 53.0 UO 2 1 9 ; 5 gm powder (outgassed) Na20 (50 mgm); 850°C* 0 .8 7 ± 0 .1 3 - 1 0 1 9 (A 1 4 ) 53.0 l^O g; 2.5 gm powder 8 5 0 ° C * 0 .1 4± 0 .2 ' 0.7 8± 0 .1 2 2 0 ( M 0 5 ) 2 2 . 0 U0.75PU0.25P2.00; 5 9 m pow der Na20 (20mgm);800°C* 0.08±0.01 0 .1 8±0.02 21 ( M 0 6 ) 6 5 .5 u0.75pu0.25°2.00' 5 9 m pow der Na20 (20 mgm); 800°C* 2.2±0.3 > 4 0 2 2 (A 1 5 ) 9 6 .0 U 3 O 8 ; 5 gm powder Large Stainless Steel Capsule; 800°C ‘t' 0.1 1 ± 0 . 0 2 0.2 3± 0 .0 7 23 (A 1 6) 1 0 0 .0 UßOg; 5 gm powder Large Stainless Steel Capsule; 800°C^ 0.04±0.01 0 .0 8± 0 .0 2 2 4 (A 1 7 ) 50.0 UO 2 .0 5 Pellet (~4 gm) Mo Capsule ; 800°C^ 0.41 ± 0 .0 6 1 ,5±0.6 *Capsufe contents stirred by periodic inversion (every 4 hours), tCapsule contents periodically stirred by swirling only (no inversion). IAEA-PL-463/23 417 The technique used to analyze the vanadium wires has been shown to be reliable so we are forced to consider the reaction system itself for an explanation. Periodic agitation of the capsules at temperature - to assist equilibration between the vanadium wire, sodium and the fuel phase - has not resulted thus far in any improvement in the reproducibility of results. Indeed it appears that if the fuel phase comes into contact with the wire (as a result of agitation) high values of C0 result. This and other observations suggest that there may be a kinetic barrier effect in the static capsule equilibrations. For example, oncea film of reaction product (N aßM O4 ) forms over the fuel phase sluggish diffusion of oxygen through 1\1азМ 04 could prevent further communication between oxygen in МОз+у and oxygen in the sodium. In this situation, which may only be relieved by continuous comminution of the reacting solid phase, the vanadium wire would simply equilibrate with the oxygen remaining in the sodium and (assuming this remaining oxygen reflects C0) yield a low answer. (This last conclusion takes into account the relative oxygen capacities of the vanadium wire and the sodium present in tests 1-2 1, inclusive.) The results of tests 4 and 20 indicate that up to some limiting initial oxygen level there is a preference, at least initially, for the fuel to react with oxygen in the sodium (as opposed to oxygen in the fuel); if this, and the preceding hypothesis, is true then such factors as the (i) initial oxygen level in sodium, (ii) the surface area of the fuel phase, and (iii) the rate of heating of the reaction capsule, may determine the final (apparent) value of C0 -a n d hence its variability. Tests 22 and 23 were designed to eliminate some of the uncertainty mentioned above, the capsules being loaded with approximately ten times the amount of sodium (and the same amount of vanadium wire) used in tests 1 to 21. The results are low (0.1-1.25 wppm) which together with the additional information that the initial oxygen level in the sodium used to fill these capsules was 60-70 wppm (i.e., more than ample to saturate the vanadium if this oxygen does not react with fuel) indicates that the equilibrium value C0 must indeed be less than 1 ppm. Although the absolute diff usivities of sodium and oxygen in Na3 U 04 have not been measured, it may be inferred from the measured growth of reaction product phase on Ug ggPug 20Û2-y [8 ] that sodium diffuses faster than oxygen and that the diffusion of oxygen ions in N331104 is slower than in hypostoichiometric mixed oxide. This qualitative picture tends to corroborate the "stuck kinetics" explanation for the variability of static capsule isothermal equilibration results. It is clearly desirable to perform similar experiments with continuously agitated solid (fuel) phase. 3.3 Electrochemical Measurements The purpose of preliminary experiments performed in the stainless steel-contained EM F cell was to calibrate the oxygen sensors by means of controlled additions of oxygen to the sodium and also to characterize the chemical behavior of stainless steel in the resulting solutions. Reports that stainless steel can interfere when the oxygen concentration in sodium is monitored using EM F cell oxygen meters have been made by several workers. [9,10,11] It is usually assumed that this interference is due to participation of chrom ium in the reaction 2 Thus far reliable thermodynamic data for the Cr— 0 — (excess)Na system are lacking yet, together with associated kinetic information, such data are required to determine the relative effectiveness of mixed oxide fuel and stainless steel cladding as getters for oxygen in the event of a fuel pin failure in an LM FB R . Previous attempts to measure oxygen activities in the N a -O -C r system using EM F cells [9,10,11] were all made under conditions (e.g., low temperature, impure electrolyte, air reference electrode) which would induce a significant degree of electronic conduction through the electrolyte. In the present investigation the cell design, materials and operating conditions were chosen so that extrinsic electronic conduction and concentration polarization effects in the electrolyte would be virtually negligible - hence measured cell EMFs should be close to theoretical 'therm odynam ic' values. Incremental additions of oxygen to sodium at 625°C, after extended gettering with uranium foil, yielded the following cell equation: EMF(volts) = 1.328 - 0.078 log-|Q С (wppm) (7) where С is the concentration of oxygen in sodium. Although this cell equation differs somewhat from the "theoretical" cell equation* we do not regard this as valid evidence of cell nontheoreticalness since some of the data used in the derivation of the so-called theoretical cell equation are only applicable up to 500°C and certain ^Derived by assuming the solution of oxygen (Na2 0 ) in sodium obeys Henry's law [13] and using published solubility and thermodynamic data [5] for Na2 Û. 418 ADAMSON et al. assumptions are questionable (see later). Each time the uranium foil oxygen getter was raised from the sodium a perceptible decrease in cell voltage occurred over a period of several hours. Since no oxygen was admitted to the reaction vessel during this time, this decrease in voltage is believed to be the result of re-establishment of oxygen equilibrium between oxygen-containing compounds on the inner surface of the stainless steel vessel and oxygen dissolved in the sodium. The final "plateau" value of the EM F, 1.462 to 1.467 volts at 625°C, corresponds to an oxygen potential of about -18 6 kcal/mole'^ if the cell is assumed "theoretical." This value may refer to the univariant 3-phase field j Na f + [Cr] gg + 3.4 General Discussion In Figure 2 an attempt has been made to bring together all the available oxÿgen potential data for the sodium-fuel system. Free energy calculations for various levels of oxygen in sodium (0.1,1, 2, 5,10, 20, 50, and 100 ppm pxygen by weight) were performed using the usual assumption that such dilute solutions obey Henry's law; values of the free energy of formation of Na2Û were taken from the latest JA N A F Thermochemical Tables (second edition, June 1971) and, to describe the solubility of Na2 Û in liquid sodium, Smith and Lee's [5] recent oxygen solubility correlation was used (log C°0 (Na) (ppm by wt) =6.907 — 28Q9/T(deg K )"'). The resulting relationship between oxygen potential, oxygen concentration (C0) and temperature is: AGq2 = 2AG/°(Na20) + (9.152 log10Co(Na) - 63.762)T + 25,708 (kcal/mole 02) (8) Straight lines were drawn through the calculated oxygen potentials for each oxygen level; the resulting isopleths are dashed above 550°C, in acknowledgement of the fact that oxygen solubilities have only been measured up to this temperature. The data points in Table 2 were converted to oxygen potentials using Equation (8 ) and then plotted on Figure 2 together with their associated error bars. It can be seen that the scatter of data is much greater than the experimental limits of error on individual datum points. Dashed vertical lines have been drawn between unusually high or low measurements and the main cluster of values (0.1-1 ppm). The cross-hatched area in Figure 2 indicates the region of (equilibrium) oxygen potentials determined for the sodium and mixed oxide fuel reaction by measurements of the final O/M ratio in unreacted fuel; the data were taken from the results of isothermal capsule teste at AN L [6] and temperature gradient capsule tests at GE (see Section 3.). Also drawn on this diagram are A G [|2 versus T relationships for four hypostoichiometric mixed oxide compositions (O/M = 1.92, 1.94, 1.96, and 1.98; 20% Pu) [12] and equilibrium in Reaction (2) (i.e., the 3-phase re g io n { Na} - <М0 2 +у)-Ш азМ 04»; this last line (indicated as A -A ); was calculated using (i) the free energy of formation data recently obtained at A N L for Na3 UÛ4 and (ii) the same approximations for solutions of oxygen in sodium (and thermodynamic data) as were used in the derivation of Equation (8). Figure 2 highlights the fact that the equilibrium Д Е 02 values derived from measurements on the liquid sodium phase all lie below the 1 ppm isopleth. If these measurements are authentic then they represent a considerable deviation from the predicted CD values (i.e., line A -A ); in the same temperature range calculated C 0 values range from 30 to 125 ppm. This discrepancy is particularly puzzling because, as illustrated in Figure 2, O/M measurements on the unreacted oxide fuel phase in equilibrium with №314104 and sodium generally give higher A G ib values than predicted. These latter values were obtained using the A G o 2 ~ 0 /M relationships established for mixed oxide fuel by Markin and Mclver [12]; since the galvanic cell technique by which these oxygen potential data were measured was applied in the temperature range 700 to 1140°C (i.e., overlapping the temperatures of the present experiments) uncertainty in these data should probably be discounted asa possible source of the discrepancy. Other possible 'thermodynamic' sources of error are (i) the assumption that at temperatures in the range 600 to 1100°C dilute solutions of oxygen in sodium continue to obey Henry's law (is the activity coefficient of sodium oxide in liquid sodium indeed independent of concentration?), (ii) the limiting concentration of Na20 in sodium at high temperatures follow the low temperature solubility relationship of Smith and Lee [5] (is the heat of solution of oxygen invariant with temperature even as the melting point of Na20 is approached?), and (iii) the free energy of formation of Na3(U,Pu)04 is identical to that for N331)04 IAEA-PL-463/23 419 T <°K) FIG.2. Oxygen potential diagram for the Na-(U,Pu)-0 system. (and independent of plutonium content). In the absence of information on any one of these last three points it is clearly premature to decide whether the observed discrepancy between (A G 0 2 ) data from the fuel phase and (C0) data from the liquid sodium phase is real (spurious results caused by unusual kinetic features of the vanadium wire equilibration experiments) or illusory (invalid assumptions in the equations used to convert C0 to A G c^ K It is hoped that the experiments currently underway at General Electric - isothermal equilibrations incorporating vanadium wires and EM F cell oxygen sensors - will soon provide the answer to this question. 420 ADAMSON et al. REFERENCES [1 ] AITKEN , E. A., EVAN S, S. K., M ELD E, G. F., and RUBIN, В. F., Trans. ANS, Suppl. No. 1, Vol. 14, April 1 9 7 1 , 1 9 . [2] BLACKBURN, P. E., M ARTIN, A. E., BATTLES, J. E., O'H ARE, P. A. G., HUBBARD, W. K., ibid 20. [3] ANL-RDP-1, January 1972 5.18-5.23. [4] KASSN ER, T. F., and SM ITH, D. L., A N L ST-6 (January 1971) 35; see also SM ITH, D. L., Met. Trans. 2, (1971) 579-583. [5] SM ITH, D. L. and LEE, R. H., ANL-7891 (January 1972). [61 See ANL-7854, (August 1971) 5.8-5.Э and ANL-RDP-3 (March 1972) 5.9-5.10. [7] AITKEN , E. A. and EVAN S, S. K., (this Vienna meeting). [8] ANL-7900 (December 1971) Б.7-5.8. [9] ISAACS, H. S., M IN USH KIN , B., and SALZANO , F. J., Proc. of the Int. Conf. "Sodium Technology and Large Fast Reactor Design," November 7-9, 1968 at Argonne National Laboratory, ANL-7520, Pt 1,460. [101 M INUSHKIN, B. and KISSEL, G., "Corrosion in Liquid Metals," Ed. Draley and Weeks, Plenum Press, 1 9 7 0 , 5 1 5 . [11] JANSSON, S. V. and BERKEY, E., ibid, 479. [12] M A R K IN , T. L. and Mel VER, E. J., Plutonium 1965, 845. [13] For an example of this derivation see KOLODNEY, M., MINUSHKIN, B. and STEINMETZ, H., Electrochemical Technology 3 No. 9-10 (1965) 244. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 1. SUMMARY 1.1. Introduction The chemistry of irradiated fuel is a very broad subject covering one of the most critical aspects of nuclear power generation. Irradiation of reactor fuel results in a chemistry of unique complexity. The transformation of some fraction of the original actinide elements into new species, the generation of both soluble and insoluble fission products in the fuel matrix, the redistribution of the mobile species, as well as their chemical reactions with fuel, cladding and coolant are just some of the changes that must be studied and understood to provide for safe and reliable operation of nuclear power plants. Although the basic approach to studying these changes in various types of fuel and reactor systems remains the same, with many common areas of interest, each reactor system, and indeed in certain instances each fuel system, has unique properties requiring separate study. The Panel discussions gave principal emphasis to sodium-cooled fast reactor fuels for various reasons: (1) The multinational nature of the LMFBR effort (2) The timeliness of LMFBR fuel studies in the light of the initial operation of various large LMFBRs in the near future (3) The absence of commercial proprietary restrictions that might normally be found at more advanced stages of exploitation of the field. Fission product control represents a very significant part of the cost of nuclear power stations. It is necessary that fission product release from the reactor system be less than a very small fraction of the generation rate of some isotopes; moreover, circuit contamination has a large influence on maintenance costs. While such control will include many measures external to the fuel element itself, the most vital aspect involves the integrity of the fuel cladding. It may be expected, therefore, that attention should be focused on the chemical reactions occurring between the fuel/fission products and the cladding,' since these interactions are recognized as potential failure mechanisms. A knowledge of fuel chemistry is important also in consideration of the mechanical and physical properties of the fuel, core physics, shut-down and trip management, fuel transport and reprocessing. In the future reactor licensing requirements are likely to demand a more detailed and quantitative description of fission products and fuel behaviour. The Panel's discussions focused mainly on the stainless steel-mixed oxide system, which at present is the preferred LMFBR fuel. In addition, carbide fuels were discussed to a limited extent. The discussions brought out the great complexity of chemical interactions between fuel, fission products, cladding and coolant. With regard to cladding integrity for mixed oxide fuel rods, an important phenomenon is that of internal corrosion, for which the oxygen potential at the fuel-clad interface at high burn-up is a key variable. To date difficulties have precluded experimental measure ments of the oxygen potential in irradiated fuel. Calculations of this quantity are also difficult because, for example, of the uncertainty about the oxidation 421 422 SUMMARY states of the various fuel constituents. Clearly, this information, which has a bearing on the oxygen-to-metal ratio (O/M) specified in production fuel and on the possible incorporation of oxygen activity control in fuel pins, is required. In addition to oxygen potential, fission products such as Te and Cs have been shown to play a role in clad oxidation. These effects have not been clearly defined, but possible potential kinetic effects were discussed. Knowledge of such effects may offer alternative methods for oxygen activity control. A considerable body of data on inert fission product gas behaviour has been accumulated and gas release is now believed to be reasonably well understood, particularly for oxide fuels. In addition, although it appears that volatile fission products will also be released and are therefore likely to be leached into the sodium coolant, little experimental work has been performed in this important area. The behaviour of fission products in the event of clad failure was not discussed. To do so would involve consider ations of particular circuit design. However, several papers on the sodium- fuel reaction evaluated the threshold oxygen potential for reaction product formátion. Swelling, due to the low density of the reaction product, and its adverse effect on promoting the consequences of clad failure, was also discussed. The Panel's discussions covered three main areas: transport properties, fuel/fission product — cladding interaction, and thermodynamics and phase equilibria. The following sections summarize the highlights of these discussions 1.2. Transport properties Transport arises from the existence of a gradient in the state variables in the irradiated fuel pins. For fission gases and other volatile constituents the driving force in the gas phase is the pressure differential; for solid constituents it is the chemical potential gradient. These processes move toward thermodynamic equilibrium as defined by the thermodynamic proper ties of the constituents in the fuel. In irradiated fuels an additional driving force is the temperature gradient which tends to perturb thermodynamic equilibrium and establish instead a steady state condition determined by a balance of competing processes. To determine the steady state condition, a knowledge of the dominant process (es) is necessary. Transport phenomena are dependent on mobility (diffusivity) in addition to a driving force. Clearly for some constituents mobility is limited and the rate of movement is negligible. Specific transport processes that influence fuel behaviour are the fission gases, the volatile fission products and solid fission products. Moreover, for two (UC,U02) and three (U,Pu02) component fuels one or more compositional variations (O/M, C/M, Pu/U) are possible in the presence of a temperature gradient. Each of these phenomena is discussed individually. 1.2.1. О/М variation The variation of the O/M ratio of oxide fuels in a temperature gradient affects significantly the cladding threshold for oxidation and other fuel properties and has been demonstrated both in-pile and out-of-pile. Such an occurrence was predicted by Rand and Roberts based on a concept of SUMMARY 423 constant overpressure of oxygen- carrying g ases that are in local equilibrium with the fuel. C 0 2 /CO was proposed as the relevant gas system because of the unavoidable level of carbon impurities present in oxide fuels. This concept successfully predicts that oxygen would concentrate near the cold region of the fuel in hypostoichiometric oxides and near the hot regions in hyperstoichiometric oxides. The Rand-Roberts concept has broader appli cability than C 0 2 /C0 since H2 0/H2, CsO(CsOH)/Cs and Mo03/Mo are alternate choices for major gas phase constituents in irradiated fuels under certain conditions. The degree of redistribution predicted by the C0 2 /C 0 concept has been successful for U0 2+J[ at low temperatures but not at high temperatures. In hypostoichiometric mixed oxides the degree of redistri bution has been usually less than predicted. This result has led to a variatior of the concept in which both gas and solid phase diffusion occur simultaneously in the fuel and establish a steady state redistribution. A detailed understanding of the redistribution mechanism derived from basic principles of irreversible thermodynamics is not possible. Proposed mechanistic concepts need to be assessed and guided by experimental results such as diffusion measurements. Results to date have characterized oxygen redistribution by H2 0 /H 2 and C 0 2 /C0. These gas systems are likely to be dominant in U0 2 fuel pins of light water reactors and at the beginning of life of fast reactor pins. In fast reactors the stainless steel cladding is permeable to hydrogen and reacts with carbon under certain conditions so that their influence as oxygen carriers may diminish with burn-up. With burn-up other fission products increase in concentration, such that they may become dominant oxygen carriers along the temperature gradient. These constituents must be characterized and identified as controlling oxygen redistribution in future studies. Furthermore, the role of actinide oxide gaseous transport on oxygen redistribution in the central core region of the fuel must also be assessed. Heretofore confirmation of oxygen redistribution in irradiated fuels has been limited to methods that give quantitative relations with limited accuracy. At this Panel Meeting O/M ratios of core-drilled samples and oxygen profiles by microprobe examination have been shown to offer direct evidence and confirmation of predictions of oxygen redistribution in lightly irradiated fuels that were rapidly quenched to retain the O/M profile during operation. In addition, an indirect method based on the measurement of Mo content in the fuel phase and in the metal inclusion phase was used to indicate oxygen activity (O/M) variations in highly irradiated fuel's. More studies involving these tools and other new measurement methods are needed to properly characterize and understand oxygen redistribution behaviour in irradiated fuel systems. Mechanistic concepts and models of g a s and solid phase transport that predict rate of response and steady state conditions are needed to provide a basis for evaluating special situations in fuels that limit performance or that might prejudice safety margins and licensing constraints. Some typical transport models needed are oxygen transport rate in axial directions as well as radial directions in thermal and fast reactors and carbon-oxygen transport in carbide fuels and in carbon-coated oxide fuels (Amoeba effect). 1.2.2. U/Pu redistribution The radial redistribution of U and Pu in a non-centre melted mixéd > oxide takes place by the evaporation-condensation mechanism in the central 424 SUMMARY void and by migrating pores during restructuring. This effect may be superposed by thermal diffusion as a typical long-term effect, which can account for alterations in the Pu/U profiles. Both effects may lead to Pu enrichment near the central void .if the O/M ratio is greater than 1. 97. The enrichment will be influenced by the stoichiometry, the porosity, the central temperature and the temperature gradient; the effect of burn-up is negligible. In addition to the radial redistribution there is an axial depletion of U via U 0 3 vaporization from the hot part of the central void and therefore a relative Pu enrichment, which can cause a local melting of the surface firstly by an increase of the central temperature of the fuel and secondly by a rapid decrease of the solidus temperature in the quasi binary U 0 2 -P u 0 2 system with increasing Pu content. These effects can lead to a Pu02-enriched melt at the bottom of the central void. Future studies should include an assessment of the relative rate of transport by solid and gas phases in the radial and axial directions. For engineering design purposes a model predicting the degree of axial and radial Pu enrichment and temperature change for a range of operating conditions is needed to allow for adequate safety margins in the reactor design. 1.2.3. Fission-product gas transport An understanding of the transport mechanism of the fission-product gases is being obtained from fuel element irradiation studies and from more specific experiments using in-reactor and postirradiation techniques. An equilibrium is set up within the fuel grains between gas in solution and gas in small bubbles. The release of gas to the free volume occurs by motion of the grain boundaries or by escape following boundary saturation. Short lived gases behave in the manner expected from their half-life. Tritium is analogous to hydrogen and migrates rapidly through fuel at all operating fuel temperatures. The configuration of gas bubbles developed in carbide and nitride systems is not well known and has not received sufficient study. A different behaviour from that of oxide fuel can be expected due to a modified rate of radiation- induced re-solution possibly with the development of larger gas bubbles. 1.2.4. Fission-product gas release For most purposes the release of stable and long-lived fission-product gases can be estimated sufficiently accurately for fuel element design, although gas release in some thermal reactor systems remains an important factor. Detection of short-lived fission products is used extensively for determining the occurrence of failed fuel. More knowledge, particularly on rates of axial migration of gaseous species, would enable failure signals to be specified with greater confidence and possibly enhance detection sensitivity. Halogen behaviour is of concern to all reactor systems because of the implications for reactor safety and siting; further study is needed in many areas. 1.2.5. Fission-product gas retention Fission-product gases have the potential to produce fuel swelling, the extent of which depends strongly on particular fuel configuration. The SUMMARY 425 generally recognized plastic nature of oxide fuels in fast reactor irradiations is believed to allow the stresses imposed by the growth of gas bubbles to be relieved to some degree. In other fuel configurations with flatter tempera ture gradients (fast reactor carbides and lower rated oxide fuel irradiations) the development of gas bubbles in the outer regions of fuel may be of more significance because of the increased strength of the fuel. The relationship between clad stress* and gas bubble swelling is an area of considerable uncertainty where further study is needed. 1.2.6. Volatile fission-product transport The transport mechanism of volatile species within the fuel grains of oxide fuel is expected to be similar to that of the fission-product gases, although in carbides there is evidence that caesium migration is less than that of the gases. The retention of volatile fission products is of concern only because of their small but inexorable contribution to swelling, which is removed by their release. Once escaped from the fuel grains, the volatile species are expected to be transported in the gas or liquid phase. Subject to kinetic factors such as the movement along pores and cracks, they will distribute according to their vapour pressure. This distribution will be influenced by chemical reactions with clad, fuel or other fission products with the formation of the compounds that are thermodynamically preferred. The knowledge of the kinetic factors in the fuel environment and the thermo dynamically preferred phases is little known and requires extensive further study with particular reference to kinetic factors limiting both radial and axial transport. 1.2.7. Solid fission-product transport The well-known formation of the inclusion phases in oxide fuels is dependent upon the existence of an initial transport process. Solid state diffusion is sufficiently rapid under irradiation with vapour transport contri buting where vapour pressures allow. A radial distribution of fission product phases produced is determined by the temperature gradient. The rare earths and the Zr dissolved in the matrix seem to be unaffected by the temperature gradient in non-melted fuels; zirconium, barium and molybdenum concentrations are seen in melted regions. At present it is not clear whether the metallic inclusions move to the centre or to the surface. For fuels close to stoichiometry there is evidence from beta-gamma auto radiography that enrichment of highly active fission products (for instance Ru) appears in the cold part of the columnar grain zone. Further motion down the temperature gradient may be prevented kinetically. Mo, Tc, Ru, Rh and Pd contained in the inclusion may be concentrated in this region also. The factors determining the relative concentrations of the constitutents are not known, although a dependence upon the O/M distribution for several species has been seen. Radial variations of Mo/Ru ratios have been found. While this observation has been used to infer the radial O/M gradient in the fuel, the interpretation must be treated with caution since there will be an independent movement of the constitutents before alloying. Most Pd-containing phases are segregated near the surface. Pd-containing phases with cladding materials are immediately adjacent to the fuel surface. Pd-containing phases with Sn, Sb and Te are located at greater depths. 426 SUMMARY The composition of ceramic phases is not believed to depend on radial positions. The Ba-containing phases appear to segregate at temperatures just below the melting point of the oxide, although Ba molybdates have been observed near the surface. Other fission-product distributions can be seen when the compositions diverge from stoichiometry. In carbide fuels migration of fission products along the temperature gradient has been observed to form inclusions in a segregated ring. Further study of this mechanism and the phases formed are needed for the evaluation of the shift of carbon potential with burn-up. Carbon transport to the colder regions in a carbide fuel is in principle possible but is not likely to occur under the temperature conditions employed in a practical irradiation. 1.3. Fuel/fission product/clad interaction 1.3.1. Mixed uranium plutonium oxide fuels with stainless steels 1. 3. 1. 1. Initial oxidation of austenitic stainless steels at the oxygen potential of the mixed oxide fuel The chemical state and structure of oxide layer on the cladding of austenitic stainless steels are given by the oxygen potential and temperature at the fuel surface and the phase relationships in the Fe-C r-N i-O system. The composition of the oxide layers depends strongly on the Cr concen tration of the alloy and the oxygen potential. Below the threshold concen tration of 13% Cr alloys are in equilibrium with M30 4 spinel type; above this critical value, the alloys are in equilibrium with M2 0 3 . To understand the role of fission-product reactions on the protective, passivating properties of the (Fe,Cr)2 0 3 oxide layer, it is necessary to consider its defect structure (Cr20 3 is considered as cation deficient, while Fez0 3 is considered oxygen deficient). Any disturbances of the 'defect concentration equilibrium1, be it by chemical interaction or under irradiation at a high neutron flux, will tend to increase the oxidation rate. The aim of further investigations should therefore be to explain a reduction of the protective properties of the oxide layers by reaction with fission-product phases.. The possibility of forming intermediate layers by chromium depletion and its consequences on stress development by the various oxide to metal volume ratios has to be taken into account. As a result of differential contractions during temperature cycling under reactor operating conditions crack formation could occur. Stress and crack formation could finally explain why these scales are lifted, giving rise to a strongly increased corrosion rate. 1.3. 1.2. Fuel/clad interface at high burn-ups Postirradiation results. Postirradiation examinations of a large number of mixed oxide fuel pins with austenitic cladding reveal that, whatever the pin configuration or the neutron spectrum (thermal or fast), internal cor rosion may occur at the fuel/clad interface. The formation and subsequent growth of the reaction zone observed are not uniform along the fuel column or along the fuel circumference (local attack). The higher corroded zones are mostly associated with higher clad temperatures. SUMMARY 427 With high-density pellet fuel the reaction zone often has a layered structure. At low burn-up a C s-rich zone sometimes appears at the surface of the fuel and in the fuel grain boundaries. At higher burn-ups the accumu lation of fission products, mostly Cs and Te with some Mo, I, Ba and Pd, results in the dissolution or lifting of the protective layer at the inner surface of the cladding. This promotes subsequent clad corrosion, either with uniform attack, a layered structure, or by intergranular penetration. With low-density vibro fuel no special feature is observed to occur at the edge of the,fuel and fission product migration to the fuel/clad interface results in intergranular corrosion .of the cladding. It will be noted that when intergranular attack occurs, high local concen trations of O, Cs and Cr are observed in grain boundaries, with Te, I and Mo also being found. Out-of-pile compatibility studies. Results here have shown that reaction between Cs and stainless steel occurs only above a certain oxygen potential. The extent of cladding attack by Cs appears to follow the increase in oxygen potential in the system; the reactions result in grain boundary attack. Purified Cs does not react with steel at temperatures up to at least 1000°C. On the other hand, CsOH reacts very strongly with austenitic steels; even at 400°C, grain boundary attack takes place. The reactions of Te do not seem to depend on the oxygen potential in the same way as those of Cs. Reactions of Te with the cladding take place even at 400°C (uniform attack). Above 700°C the attack along the grain boundaries of the cladding is predominant. The Cs/Те ratio appears to have an influence on the reactivity of Te. Elemental I and B r cause heavy attack on steel claddings, when combined with Cs they are not as corrosive. Nevertheless, above a critical oxygen potential and/or a certain moisture level reactions may occur even with Csl and CsBr and attack of the cladding by a van Arkel-de Boer process may take place. Molybdenum and Mo oxide do not enhance the fission products reactions with the cladding; however, sometimes a minor reduction of the reaction is noticed. By addition of an oxygen-binding material (Zr, Ti, Nb, Cr) to the fuel or cladding reactions with the cladding are remarkably reduced or even avoided. This applies in particular to intergranular attack induced by Cs and Te reactions. 1.3. 1.3. Reaction mechanisms Postirradiation examinations of irradiated fuels as well as simulation experiments with austenitic stainless steel cladding reveal several types of attack; the formation of a layered structure, the dissolution of thin passivating layers, and the subsequent surface ablation of the steel probably in the presence of a liquid phase, and finally the intergranular attack of sensitized steels within the grain boundaries. All types of attack are usually of a rather local and less uniform appearance, depending on the local concentration of fission products. In the case of a layered structure it is assumed that the passivating oxide layers, mainly consisting of Cr2 0 3, are formed before an appreciable amount of fission products has reached the fuel/clad interface. 428 SUMMARY At the temperature of the fuel/clad interface of approximately 600°C Cs20 can only be formed at the oxygen potential of a hyperstoichiometric fuel. Moreover, the melting point of Cs20 lies below the temperature range expected at the fuel/clad interface. Therefore, a liquid (most likely a multicomponent thin layer of caesium) containing oxygen and the more volatile fission products, such as tellurium, selenium and rubidium, is to be expected. It seems reasonable to expect that the chromium-rich M2 Os layer is dissolved in the case of thin films, or the passivating properties (given by a minimum in defect concentration) are changed by selective dissolution reactions with oxygen dissolved in caesium. 1.3.2. Mixed uranium plutonium carbide fuels with stainless steels The main topics, such as fuel stabilization, carbon potential as a function of temperature gradient and burn-up, carbon transfer in sodium bonding, clad carburization, and the role of oxygen contamination are to be considered on the basis of thermodynamics, phase relationships and the available experimental results. Fuel compositions of the two-phase fields M+MC and MC+M2 C3 (M=U, Pu) are incompatible with austenitic stainless steels. A minimum of chemical interaction is found in contact with monocarbides. Compatibility tests with hyperstoichiometric fuels containing sesquicarbides and dicarbides lead to the carburization of the stainless steel with the formation of (Cr, Fe)23C6 precipitates to an extent depending on the carbon potential of the fuel. Tests with hypostoichiometric fuels containing free metals show the formation of intermetallic compounds such as MFe2 or MNi5. As a result of the narrow homogeneity range of the monocarbide single phase region a slight change in its carbon content leads to the adjacent two-phase fields. Problems associated with the use of carbide fuels include: (1) Stabilization of the monocarbide region (2) Effect of sodium bonding, which accelerates carbon transfer to the cladding (3) Carburization of the cladding, which is recognized as a potential problem since with a carbon content beyond the critical value of 0 . 6 wt.% embrittlement of austenitic steel takes place. Results of carbide fuel studies indicate that: (a) Ni-base alloys are not compatible with carbide fuels at 700°C (b) Austenitic steels are compatible even with slightly hyperstoichio metric fuel up to 800°C, except for alloys with higher N i-contents (e.g. Incoloy 800) (c) A few types of steel behave well even at 900°C (however, there is a danger of more pronounced reaction in steel stabilized with Ti) (d) At 1000°C there are strong reactions with all steels (e) Reactions resulting in the formation of U, Pu/intermetallic compounds are more dangerous than carbide reactions because they accelerate the overall rate of reaction. Postirradiation investigations of hyperstoichiometric fuel pins showed more clad carburization than was expected from out-of-pile tests with fuels containing sesquicarbide. No fission product reactions with the clad were observed. SUMMARY 429 1.4. Thermodynamics and phase equilibria 1.4. 1. Oxide fuels The oxygen potential of the fuel matrix and its variation with burn-up is perhaps the most important factor that determines its performance. Direct measurements of oxygen potential during operation are practically impossible, but measurements during postirradiation examinations could be and should be undertaken. In the absence of direct measurements, a knowledge of the likely chemical state of 'burnt oxides' has been built up using the existing knowledge of thermodynamic data and phase equilibria of the appropriate systems. The identification of phases formed during irradiation, together with the knowledge of the relevant phase equilibria, may help as an indicator of local oxygen potential variations within the fuel. A very important contribution to this problem would be to examine the change of oxygen potential with burn-up in carbon-coated particle fuels where CO pressure measurements with burn-up would give a direct measure of oxygen potential. This system is much better thermodynamically defined with a 'cladding' of unit activity carbon instead of a cladding of stainless steel, which can act as an ill-defined oxygen sink. It should, however, be noted that Dragon Project studies have already demonstrated a large discrepancy between observed and calculated oxygen yields for 239Pu thermal fission in a U 0 2 matrix. 1.4. 1.1. U-Pu-O ternary system For the region of interest for oxide fuel specifications knowledge of the solid phase relationships is adequate. The existence of the fluorite lattice over the broad temperature range from 500 to about 2700°C for both hypo- and hyperstoichiometric systems is well established. An aspect of the phase relationships that must be considered, however, is the effect of both stoichiometry (O/M ratio) and the Pu/U+Pu ratio on the solidus and liquidus temperatures of the system. In view of the “tendency to-go to higher power ratings and the probable occurrence of both oxygen and fissile atom redistributions, a study of the effect of these parameters on the solidus- liquidus temperatures would be desirable, particularly in the region of low O/M ratios. The successful application of models on redistribution of oxygen and of plutonium and uranium in both the gas and solid state phases requires a knowledge of the variation of the chemical potential of oxygen, uranium and plutonium in the solid state as a function of temperature and their concentration, especially at temperatures above 2000°C. The oxygen potential data used so far have been based on extrapolation to operational temperature levels and a need exists for oxygen potential measurements under fuel operating temperatures. Measurements made at temperatures up to 1700°C (on U0 8Pu0 2^ 2 -^ suggest that the low tempera ture data may be extrapolated up to 1700°C, provided that the O/M ratio is >1. 96; below 1. 96 the measured values of oxygen potential are lower than the extrapolated values. These measurements should be continued and extended to include different Pu/U+Pu ratios. Work should also continue on development of models to describe the system. 430 SUMMARY Studies of the vaporization of the mixed oxides indicate that for U0 sPuq^Oi 97 a very important change takes place in the predominancy of Pu- or U-containing species in the gas phase, as found both in experimental measurements and in the examination of irradiated fuels. The oxygen potential calculated from equilibria of the various gas phase species should of course be consistent with direct measurements. After further oxygen potential measurements the consistency of the mass-spectrometric data should again be reexamined. A future requirement will be the need for thermodynamic data at temperatures in excess of 3000°C for fast reactor safety assessments. Some data on U oxide have already appeared. 1.4. 1.2. Oxide-fission product system Oxygen potential of the fuel. The variation of the oxygen potential of the fuel matrix as a function of burn-up is dependent on the following: (1) Fission-released oxygen is only partly bound by the fission products; and increase in the mean valency of uranium and plutonium cations results, producing a less negative oxygen potential; (2) Dissolution of soluble fission products in the fuel matrix. The burnt fuel atoms are partially replaced by a mixture of soluble rare earth and zirconium ions with a mean valency lower than that of the fuel matrix. To maintain charge neutrality, oxygen vacancies are formed that can be filled by oxidation of uranium from the tetravalent to the pentavalent state. Experimental data are lacking and appropriate measurements should be made to determine the influence of soluble fission products on the oxygen potential of mixed oxides. While a part of the zirconium could contribute to the formation of a second ceram ic phase BaZrOa (this highly stable phase having been shown to exist in irradiated oxide fuel), volatile precursors of Ba (and Zr) might prevent the appreciable formation of this phase; (3) Overall variation in the Pu/U+Pu ratio as a function of burn-up, together with local changes in this ratio as a result of migration, lead to a variation of the mean valency of the actinide ions; (4) During burn-up the increasing level in oxygen potential eventually reaches the value where oxidation of Mo occurs. In most fuel system the Mo concentration is sufficient to buffer the excess oxygen. With buffering the variation in oxygen potential is controlled by the effect of the dissolution of soluble fission products only; (5) The possibility that depassivation of the cladding may occur and that the oxygen activity in the closed fuel-stainless steel system may be controlled by a clad-metal-metal oxide equilibrium should be considered. The cladding can thus be considered as an oxygen sink, binding excess oxygen and controlling the oxidation state of the fuel and fission products. This undesirable effect could be eliminated by the introduction into the fuel element of an appropriate getter to bind excess oxygen and fix the oxygen activity at a sufficiently low value so that no appreciable clad corrosion could occur. 1.4. 1.3. Insoluble fission products Insoluble fission products are found as metallic inclusions, as oxide phases, arid as phases containing cladding material components. The SUMMARY 431 knowledge of the stability of the fission product phases is required to determine the oxygen potential in specific regions of the fuel and to under stand the behaviour of such volatile fission products as Cs, Te, Se and I. Metallic inclusions found in the high temperature regions contain Mo, Tc, Ru, Rh and Pd in different proportions. The Mo content of this phase in equilibrium with the Mo02 phase provides a possible means of determining the local oxygen potentials. To use this method, an exact understanding of the equilibria and the thermodynamic data are required. The occurrence of the (U, Pu)Me3-type phage provides another means of obtaining information about the oxygen potential, if the free energy of for mation of this phase and the composition and stability of the matrix in contact with this phase are known. Formation of these compounds, which can only occur at high temperatures and in hypostoichiometric fuels, leads to an increase in the available oxygen. It would be worthwhile to obtain data on these equilibria. With regard to the other metallic or ceramic fission product inclusions, at the present time there is no apparent system atic dependence of their composition on parameters such as stoichiometry, temperature or Pu/U+Pu ratio. It is important to know the stabilities of complex oxides of the volatile fission products Cs, Te, Se and I with fuel components and cladding com ponents. Cs and Te may react with the fuel and clad to form complex com pounds. Also, Cs-Cr-O compounds may form. Iodine shows no evidence of reaction with the fuel phase. The reaction behaviour of the volatile elements depends strongly on the O/M ratio of the fuel. Some further specified measurements of thermodynamic stabilities of the relevant com pounds and determination of phase relationships in the complex systems including cladding material components could give a better understanding of the mode of reaction. Studies of constitution of Cs-containing systems could be very helpful in explaining the reaction behaviour. It would be desirable to know also the critical oxygen activity leading to a severe cladding attack by oxygen-containing Cs. 1. 4. 1. 4. Sodium-mixed oxide fuel interactions In LMFBR fuel pins with cladding defects chemical interactions of the sodium coolant with the uranium/plutonium oxides could cause swelling and subsequent failure propagation. Information on such reactions is therefore of importance for reactor safety. In the presence of M 02 (M=U, Pu) the product of the sodium-oxide fuel reaction has been identified as Na3 M04, isostructural with Na3U04. Between 800 and 1000°C the O/U+Pu of the mixed oxide phase in equilibrium with this compound corresponds to a plutonium valency of about 3.5. For lower O/U+Pu ratios mixed oxides do not react with sodium. The presence of fission products appears to enhance the reaction rate without changing the reaction product. However, more knowledge is required on the Mo2+x- fission product-sodium systems. The most important reaction variable is the partial pressure of oxygen in equilibrium with the three phases Na3 M04, МОг-х and Na. The variation of this value with temperature has been deduced from: (a) The O/U+Pu ratio of the mixed oxide in equilibrium with Na3 M04 (b) M ass-spectrom etric measurements of sodium and oxygen pressures over the Na-U-O system SUMMARY (c) e.m.f. measurements (d) Determination of the oxygen concentration in sodium at equilibrium. Discrepancies in the data are apparent. For example, direct measure ments on sodium lead to lower values than the other methods. Unusual kinetic features of the sodium-fuel reaction may be a possible cause of the discrepancy, but there is also some doubt about the validity of the use of Henry’s law for dilute solutions of oxygen in sodium. 1.4.2. Carbides and nitrides With regard to carbide and nitride advanced fuels for fast reactors, the present state of our knowledge on the thermodynamics and phase equilibria for the appropriate regions of the U-Pu-C and U-Pu-N systems, together with those of the fission-product systems, is believed to be suf ficiently adequate to enable a prediction of the chemical state of burnt carbides and nitrides. Providing there are only small quantities of sesqui- carbide present together with the monocarbide, the carbon potential of the carbide fuel should decrease during burn-up. However, in the case of the nitride, which is easier to fabricate to a single-phase mononitride, the nitrogen potential is expected to increase with burn-up. Experimental measurements and detailed observations of irradiated fuels should be undertaken to verify these predictions. 2. CONCLUSIONS There are important areas of fuel and fission-product chemistry that must be investigated. Work on description of the chemistry of fuels under actual operating conditions, including questions relating to power cycling, design basis accidents, reprocessing and ultimate disposal, is essential to the introduction of any reactor system. The Panel's discussions, which centred on fast reactor fuels, in particular oxide fuels, led to the following identification of important problem areas where further work is required: 2.1. Transport properties (1) Because of the importance of oxygen redistribution to the threshold for cladding attack and reactions with fission products, oxygen redistribution should be measured directly and evaluated in terms of the relevant oxygen- carrying species moving under the influence of a temperature gradient. The role of volatile fission products such as caesium as well as impurity substances such as carbon and hydrogen should be identified. (2) Mechanistic concepts and models for both O/M, C/M and Pu/U redistribution under radial and axial temperature gradient are needed to predict the degree of redistribution and the rate of transport for a variety of design and off-design conditions in which the fuel may be subjected. (3) Because of the potential role of molybdenum as an oxygen sink for fission-released oxygen, both its transport behaviour and its thermodynamic behaviour should be evaluated and understood, in particular in out-of-pile experiments; SUMMARY 433 (4) There is a general need for understanding the kinetic factors of oxygen and volatile fission-product transport in a fuel pin tô enable pre- ' dictions about the local concentrations of (1 ) retained fission gases which influence swelling, or (2 ) condensed reactive constituents, which influence cladding attack. ■ 2. 2. Fuel/fission product/clad interaction • ■ (1) To understand fully the conditions that can initiate and are necessary for each of the cladding attacks identified by the Panel, the thermodynamics and phase relationships of the various reactants and products must be studied in greater detail. (2) A study of the reductions of the protection properties of the oxide layers by fission product attack is required to explain the various types . of clad attack. - (3) The role of caesium and tellurium in. fuel-clad interaction, especially with regard to intergranular attack, has to be investigated more closely. 2.3. Thermodynamics and phase equilibria (1) Efforts should be supported to measure the oxygen potential of fuel during postirradiation examination. (2) Measurements of solidus and liquidus temperature in the U-Pu-O system as a function of Pu/U+Pu and O/M ratios in the relevant regions should be undertaken. (3) Measurements of the oxygen potential of the mixed U-Pu-O system should be made at temperature levels extending to the Upper fuel operating temperatures, with data above 3000°C required for safety analysis purposes. (4) Measurements should be made to determine the influence of the rare earth ions on the oxygen potential of the U-Pu-O system. (5) The effect of getters to prevent oxidation of the cladding should be evaluated. (6 ) A determination should be made of the oxygen potential at which Mo-containing inclusions in the irradiated fuels oxidizes. (Y) Measurements should be made of the free energy of formation of the Cu3Au-type phases containing U, Pu, Ru, Rh and Pd, and of the oxygen potential at which these compounds are formed in the presence of U-Pu oxides. (8 ) Phase diagram studies on Cs-containing systems with fuel and clad components should be made. (9) More knowledge on the U-Pu-Na-O-fission product systems is required. (10) Measurements of the oxygen potential for the formation of Na-(U,Pu)-0 compounds should be extended. (11) A determination of the effect of burn-up on carbon and nitrogen potentials in carbide and nitride fuels should be undertaken. 3. RECOMMENDATIONS (1) The Panel does not make, any recommendation for another panel meeting a few years hence or for any other specific information exchange 434 SUMMARY meeting. It does feel that it was only able to discuss a small part of a field that is under active development and is of critical importance in the successful implementation of a safe, economical advanced reactor system, such as the LMFBR. It believes that the Agency should actively follow developments in the field, giving consideration to a possible future meeting of wider scope, including other reactor systems and possible extension to non-commercial aspects such as permissible discharges, circuit mainte nance, and behaviour of various species in circuit and engineered safeguards. (2) Since accurate predictions of the chemical composition of irradiated fuels, corresponding to a given initial composition and irradiation conditions, is fundamental to assessments of chemical interaction in fuel-clad systems, the Panel agreed that, as a first step in deciding whether the basic input data and computational methods in use at different establishments are giving chemical inventories in sufficient agreement for such assessments, the results from several laboratories of a test calculation for a given fuel irradiation situation should be intercompared. The Panel recommends that the Agency give consideration to the desirability and feasability of a suitable international intercomparison of analytical predictive techniques. (3) The Panel recommends publication of the Proceedings of this meeting. (4) For future panel meetings the Panel recommends that papers be distributed for review by prospective panel participants prior to the meeting. This would permit additional time for detailed discussion of salient points and for evaluation of results. LIST OF PARTICIPANTS Aitken, E. A Plutonium Research Laboratory, Nuclear Energy Division, Vallecitos Nuclear Center, Pleasanton, Calif. 94566, United States'of America Anselin, F. CEN de Fontenay-aux-Roses, B. P. № 6 , 71 Fontenay-aux-Roses, France Conti, U. CNEN—Fast Reactor Programme, Via dell'Arroveggio 56/23, 40129 Bologna, Italy Crouthamel, K. E. Argonne National Laboratory, Chemical Engineering Division, 9700 South Cass Avenue, Argonne, 111. 60439, United States of America Cubicciotti, D. Nuclear Energy Division, General Electric Company, Vallecitos Nuclear Center, Pleasanton, Calif. 94566, United States of America Dean, M. Département d'études des combustibles à base de plutonium, Division de métallurgie et d'étude des combustibles nucléaires, 71 Fontenay-aux-Roses, France Findlay, J. I. Atomic Energy Research Establishment, Applied Chemistry Division, Building 429, Harwell; Didcot, Berks, United Kingdom 435 436 LIST OF PARTICIPANTS Flowers, R. H, Atomic Energy Research Establishment, Applied Chemistry Division, Building 10. 5, Harwell, Didcot, Berks, United Kingdom Hofmann, P. Institut für Material- und Festkörper forschung, Kernforschungszentrum Karlsruhe, Postfach 3640, D-75 Karlsruhe, Federal Republic of Germany Holleck, H. Institut für Material- und Festkörper forschung, Kernforschungszentrum Karlsruhe, Postfach 3640, D-7 5 Karlsruhe, Federal Republic of Germany Kleykamp, H. Institut für Material- und Festkörper forschung, Kernforschungszentrum Karlsruhe, Postfach 3640, D-75 Karlsruhe, Federal Republic of Germany Ohse, R. W.; (EURATOM) Europäisches Institut für Transurane, Postfach 2266, D-7 5 Karlsruhe, Federal Republic of Germany Podest, M. Nuclear Research Institute, Rez near Prague, Czechoslovakia Potter, P.E. (EURATOM). Europäisches Institut für Transurane, Postfach 2266, D-75 Karlsruhe, Federal Republic of Germany Powell, H. J. Dounreay Experimental Reactor "Establishment, Dounreay, Scotland, Ünited' Kingdom Randl, R. Bundesministerium für Bildung und . Wissenschaft, Postfach 120124, D-53 Bonn 12, Federal Republic of Germany LIST OF PARTICIPANTS 437 Rosen, S. Fuels and Materials Branch, Division of Reactor Development and Technology, US Atomic Energy Commission, Washington, D. C. 20545, United States of America Sârbu, I. Institutul de Fizica Atómica, Casuta Póstala no. 35, . Bucuresti, Romania Schmitz, F. CEN de Fontenay-aux-Roses, B. P. № 6 , 71 Fontenay-aux-Roses, France HOW TO ORDER IAEA PUBLICATIONS Exclusive sales agents for IAEA publications, to whom all orders and inquiries should be addressed, have been appointed in the following countries: ■ UNITED KINGDOM Her Majesty's Stationery Office, P.O. Box 569, London SE1 9NH UNITED STATES OF AMERICA UNIPUB, Inc., P.O. Box 433, New York, N.Y. 10016 In the following countries IAEA publications may be purchased from the sales agents or booksellers listed or through your major local booksellers. Payment can be made in local ■ currency or with UNESCO coupons. ARGENTINA Comisión Nacional de Energía Atómica, Avenida del Libertador 8250, Buenos Aires AUSTRALIA Hunter Publications, 58 A Gipps Street, Collingwood, Victoria 3066 BELGIUM Office International de Librairie, 30, avenueMarnix, B-1050 Brussels CANADA Inform ation Canada, 171 Slater Street, Ottawa, Ont. К 1 AOS 9 C.S.S.R. S.N.T.L., Spálená 51, CS-11000 Prague Alfa, Publishers, Hurbanovo námestie 6, CS-80000 Bratislava FRANCE Office International de Documentation et Librairie, 48, rue Gay-Lussac, F-75005 Paris HUNGARY Kultura, Hungarian Trading Company for Books and Newspapers, P.O. Box 149, H-1011 Budapest 62 INDIA Oxford Book and Stationery Comp., 17, Park Street, Calcutta 16 ISRAEL Heiliger and Co., 3, Nathan Strauss Str., Jerusalem ITALY Librería Scientifica, Dott. de Biasio Lucio "aeiou". Via Meravigli 16,1-20123 Milan JAPAN Maruzen Company, Ltd., P.O.Box 5050, 100-31 Tokyo International NETHERLANDS Marinus N ijhoff N.V., Lange Voorhout 9-11, P.O. Box 269, The Hague PAKISTAN Mirza Book Agency, 65, The Mall, P.O.Box 729, Lahore-3 POLAND Ars Polona, Céntrala Handlu Zagranicznego, Krakowskie Przedmiescie 7, Warsaw ROMANIA Cartimex, 3-5 13 DecembrieStreet, P.O.Box 134-135, Bucarest SOUTH AFRICA Van Schaik's Bookstore, P.O.Box 724, Pretoria Universitas Books (Pty) Ltd., P.O.Box 1557, Pretoria SPAIN Nautrónica, S.A., Pérez Ayuso 16, Madrid-2 SWEDEN C.E. Kritzes Kungl. Hovbokhandel, Fredsgatan 2, S-10307 Stockholm U.S.S.R. Mezhdunarodnaya Kniga, Smolenskaya-Sennaya 32-34, Moscow G-200 YUGOSLAVIA Jugoslovenska Knjiga, Terazije 27, Y U -11000 Belgrade Orders from countries where sales agents have not yet been appointed and requests for information should be addressed directly to: вОч Publishing Section, ■ lnternat¡onal Atomic Energy Agency, Kärntner Ring 11, P.O.Box 590, A-1011 Vienna, Austria INTERNATIONAL ATOMIC ENERGY AGENCY VIENNA, 1974 PRICE: US $20.00 SUBJECT GROUP: V Austrian Schillings 386,- Reactors and Nuclear Power/ (£8.50; F.Fr. 90,-,- DM 52,-) Reactor Fuels and Fuel Cycles