The compatibility of stainless steels with low-sulphur UCS fuels

Journal of Nuclear Materials 60 (1976) 195-202 0 North-Holland Publishing Company

THE COMPATIBILITY

OF STAINLESS STEELS WITH LOW-SULPHUR UCS FUELS

W.E. STUMPF and S. VENTER Physical Metallurgy Division, Atomic Energy Board, Private Bag X.256, Pretoria, South Africa

Received 1 December 1975

Sodium-bonded and vacuum-bonded compatibility tests at 600 and 700°C between stainless steels and UCS fuels of differing sulphur contents and stoichiometries revealed that, over the range of compositions studied, the amount of carburisation of the cladding is a linear function of only the total non-metal content (C + 0 + N + S) of the fuel, and does not depend on the actual sulphur content of the fuel. A UC and a UCS fuel with the same non-metal content therefore cause equal amounts of carburisation of the cladding. X-ray and microstructural investigations of the fuels further showed that a singlephase region does not exist at non-metal contents greater than 50 at%, and that sulphur probably behaves as a carbon equivalent in that it enters the UC lattice in solid solution, but displaces a carbon atom which is freed to form a higher carbide. The conclusion is therefore drawn that the addition of sulphur to UC fuels has no direct advantage at non-metal contents greater than 50 at% as far as the compatibility behaviour of the fuel towards the cladding is concerned. Des essais de compatibilite lids au vide et his au sodium ‘a 600 et 700°C entre aciers inoxydables et combustibles UCS de diverses teneurs en soufre et de differents &arts a la sto~chiom~~ie ont revel6 que sur ~intervalle de compositjons Btudie, le taux de cementation du gainage est une fonction lineaire seulement de la teneur totale en additions non-m8talhques (C + 0 + N + S) du combustible et ne depend pas de la teneur en soufre du combustible. Un combustible UC et un combustible UCS avec la msme teneur en elements non metalliques conduisent done a des taux egaux de carburation de la gaine. Des etudes par rayons X et par examen micro~aphique des combustibles ont en outre montre’qu’une region monophasb n’existe pas pour des teneurs en elements non metaltiques superieures B 50 at% et que le soufre se comporte probablement comme un equivalent carbone en ce sens qu’il entre en solution solide dans le reseau de UC mais d&place un atome de carbone qui est lib& pour former un carbure de titre plus Bleve en carbone. On en tire done la conclusion que l’addition du soufre aux combustibles de UC ne prisente pas d’avantages directs pour des teneurs en non metalliques superieures &50 at% tant que le comportement de compat~b~it~ du combustible vis-&is du gainage est concern&. Vertrliglichkeitsuntersuchungen zwischen rostfreiem Stahl und UCS-Kernbrennstoff mit verschiedenen Schwefelgehalten und Stochiometrien bei 600 und 700°C unter Na- und Vakuum-Bindung haben ergeben, dass bei den betrachteten Zusammensetzungen die Au~ohlu~ der Hiille nur dem gesamten Nichtmetallgehalt des Brennstoffs (C + 0 + N + St proportional ist und nicht vom tatsachlichen Schwefelgehalt des Brennstoffs abhangt. Ein UC- und ein UCS-Brennstoff mit demselben Nichtmetallgehalt verursachen deshalb die gleiche Aufkohlung der Hiille. Rontgenographische und Geftigeuntersuchungen zeigen ferner, dass ein einphasiger Brennstoff bei Nichtmetallgehalten grosser als 50 At% nicht existiert und Schwefel sich wahrscheinlich wie ein Kohlenstoffaquivalent verhiilt, indem er in das UC-Gitter als feste Losung eingebaut wird, jedoch ein C-Atom verdmngt, das dann ein hoheres Carbid bildet. Daraus wird gefolgert, dass ein Schwefelzusatz zum UC-Brennstoff bei Nichtmetallgehalten grosser als 50 At% keinen unmittelbaren Vorteil fur das Vertrlglichkeitsverhalten des Brennstoffs mit der Hiille hat.

1. Introduction

of the methods suggested, in attempts to improve the compatibility behaviour in such fuel elements, has been the modification of the carbide fuel by suitable additions, usually of non-metallic constituents [l] . Ideally, these additions should eliminate the higher carbide phases in the fuel altogether, thus reducing

The compatibility interactions between hyperstoichiometric carbide fuels and their cladding materials still pose a potential limitation to the general use of carbide fuel elements in fast-breeder reactors. One 195

196

W.E. Stump& S. Venter / Compatibilityof stainlesssteels

the carburisation potential of the fuel towards the cladding. Wedemeyer [I ] has reported that US and UC, form a solid solution in which the tetragonai cell of the UC, is enlarged in the direction of the a axis, whereas the c axis remains virtually constant. Furthermore, he has reported that the melting of a mixture of hyperstoichiometric UC with US can eliminate the higher carbides UC, or UzC3 in the UC. He concludes that this could possibly improve the compatibility behaviour of the carbide fuel with cladding materials, as has been reported by Shalek and White [2] who have found that a slightly hyperstoichiometric UC produces significantly greater interaction with various metals at 1150°C than a fuel consisting of the same UC reacted with 10 wt% US. In this investigation the compatibility behaviour of a number of low-sulphur UCS fuels * with three stainless steels was determined by a technique 13-61 which yields quantitative embrittlement and carburisation results on the cladding, thus enabling small differences between slightly different fuels to be evaluated. In general, the sulphur contents of the UCS fuels were kept below about 0.5 wt%, which is just greater than the reported limit of solid solubility of US in UC [Z, 71.

2.Experimental

techniques

The basic experimental technique has been described elsewhere [3-61. Briefly, stainless steel (type 3 16) capsules containing tensile sheet specimens of stainless steel types 3 16 (solution-treated and 20% cold-worked), 32 1 and 347 (both solution-treated and aged for 5 h at SSO”C), were filled with equal amounts of dry fuel powder in a purified argon-filled glove box. The compositions, heat treatments and grain sizes of the steels have been reported elsewhere [6]. The capsules were either filled with -833 + 495 pm size fraction fuel particles (BET about 50 m”/kg) and then filled with sodium, or filled with ball-milled fuel powders (BET about 600 m2/kg) and evacuated before sealing by welding [6] . The liquid sodium was doubly * The term UCS, as used here, does not imply a stoichiometric single-phase fuel of the type UK, 91 .oo, but is used to describe the uranium-carbon-sulphur fuels in general.

filtered and had an oxygen content of about 5 ppm by weight. The preparation of the UCS fuels was carried out by the volatile metal process as described by Love11 et al. [7], with the difference that recent work [8] has shown that sulphur (from ZnS) will react directly with UC at about 65O*C to produce UCS plus free carbon, and not only with the metal phase in a (UC + U) cermet, as was suggested earlier. If the (UCS t C) mixture is then homogenised at elevated temperature, either a solid solution or a higher carbide is formed, depending on the exact composition of the fuel. For each UCS fuel, a 500 g batch of uranium carbide of the required composition was first prepared by the usual carbothermic reduction of uranium oxide at 1800°C under vacuum [5] . The uranium carbide was then ball-milled with tungsten-carbide spheres in a rubber-lined ball-mill to a BET specific surface area of about 1000 m2/kg, and then mixed for 5 h with the correct amount of ZnS in a mixing mill. The (UC t ZnS) mixture was pressed into pellets at a pressure of about 1.50 MPa, loaded into a tantalum-lined graphite boat and reacted at 650°C for 1 h under highly purified flowing argon. The temparature was then raised to 850°C for 1 h to drive off most of the zinc, which was collected on a cold finger in the furnace. These pellets were then re-crushed and screened to -90 pm size fraction, remixed for 5 h, pressed into pellets once again and then homogenised at 1850°C for 2 h under high vacuum, during which the zinc content was also reduced to less than 3 ppm by weight. This produced about 500 g of UCS pellets in each case, which could be crushed, screened or ballmilled to obtain the required size fractions for loading into the compatibility capsules. The complete preparation as described here was carried out in a purifiedargon glove-box system. Finally, the ZnS used in the process was purified beforehand under flowing pure CO at 116O’C for 16 h, which reduced the oxygen content to about 600 ppm by weight [9]. The homogenised UCS fuels were investigated thoroughly by chemical, X-ray and optical microstructural analyses. Chemical analyses of C, 0, N and S were carried out on commercially available LECO combustion or fusion apparatuses with the mean of at least two (but in many cases three or four) analyses reported here. X-ray analyses were carried out by using both powder-diffraction Debye-Scherrer and diffract*

W.E. Stump.f S. Venter { Compatibility

197

of sfuidess steels

3. Results and discussion

meter methods. Unfortunately, optical microstructures could not be prepared from the very porous original UCS pellets after homogenisation at 1850°C. The original UCS pellets were, therefore, melted in an argonarc furnace which changed the chemical compositions of the fuels in all cases, probably on account of a loss of both C and S in the form of CS, gas (as identified by mass spectrometry [9] ), as well as causing a general decrease in oxygen content. In contrast, arc-melting of low-oxygen.UC or US on their own produced no compositional changes in either fuel [ lo]. After completion of the compatibility tests for 1000 h at 600 and 7OO”C, the room-temperature tensile properties of the steel specimens were determined for each test at a strain rate of 1.3 X 1O-3/s on two separate specimens, the average of the two being reported here. Selected specimens were also examined by means of microhardness gradient measurements, and the total carbon and sulphur content of the sheet specimens (0.65 mm thick) were determined for each test. Carbon and sulphur analyses of the UCS powders were also carried out on the vacuum-bonded tests after completion of the compatibility anneal.

3. I. Fuel compositions The chemical and X-ray analyses of all the fuels are given in table 1 and are also illustrated on a portion of a ternary U-C-S phase diagram in fig. 1. In this figure the carbon is taken as the sum of the actual carbon contents plus the oxygen and nitrogen contents in solution in the fuels. Also shown in table 1 and fig* 1 are the compositions and changes in composition of three arc-melted fuels (obtained from fuels B, D and G respectively) while in figs. 2(a)-(c) their respective optical microstructures are shbwn. These (and other arc-melted specimens) showed the presence of a second phase, very similar in appearance to the UC, phase usually found in UC fuels, for all stoichiometries greater than 50 at% (C + 0 + N + S) and only at stoichiometries less than SO at% (fig. 2(a)) was a singlephase fuel found. The volume fraction of second phase in the fuel generally increased as the stoichiometry increased above 50,at% (figs. 2(b) and (c)). It is also interesting to note that, although the fuel shown in fig.

Table 1

Initial chemical and X-ray compositions Fuel

C wt%

0

wt%

A B c D E F G H

4.65 4.56 4.69 4.67 4.67 4.77 4.74 4.93

0.05 0.02 0.07 0.02 0.04 0.01

J K L M

4.67 4.80 4.79 5.02

0.03 0.08 0.06

Arc-melted

R _____.

specimens: 4.47 4.65 4.70 ..~

0.08

0.10

0.07

(from 0.03 0.01 0.02

N wt%

0.04 0.02 0.02 0.03 0.03 0.05 0.03 0.05

of UCS and UC fuels used

s wt%

0.23 0.45 0.28 0.38 0.42 0.31 0.47 0.26

0.05 0.03 0.05 0.04

CC+ 0 at%

+N) S

49.27 48.40 49.18 48.85 49.01 49.63 49.20 50.51

at%

0.90 1.75 1.09 1.48 1.62 1.19 1.80 0.99

49.75 50.25 50.47 51.59

Total nonmetal content at%

Phases

a0 nm

50.15 SD,16 SO.27 50.33 50.63 50.82 51.01 51.50

UC UC UC UC UC UC UC UC + UC2

0.4975 * 0.0003 0.4968 0.4971 0.4978 0.4973 0.4970 0.4974 0.4981

-c 0.0002 i 0.0006 2 0.0003 + 0.0003 * 0.0003 +- 0.0004 -c 0.0002

49.75 50.25 50.47 51.59

UC UC UC

0.4962 0.4965 0.4964

f 0.0002 f 0.0004 * 0.0005

fuels B, D and G respectively) 0.02 0.03 0.03

0.37 0.38 0.35

47.73 48.68 49.07 -__-__

X-ray analyses

Metallographic 1.47 1.48 1.36

49.20 50.17 50.42

single phase two phases two phases -~ _.

analyses

198

W.E. StumpA S. Venter / Compatibility

of stainlesssteels

s at. % Fig. 1. Chemical compositions of UCS and UC fuels in the ternary U-C-S phase diagram used for compatibility testing (o) and arc-meited specimens (0) used for optical microstructural examinations.

2(a) had a stoichiometry well below 50 at%, no second phase similar in appearance to U-metal on the grain boundaries could be found anywhere in the sample. The observation by Wedemeyer [l] that higher carbides are eliminated by the melting of a hyperstoic~ometric UC with US, is therefore confirmed by this work; the reason here, however, is that the stoichiometry was moved towards lower values by compositional changes during melting, and was apparently not due to the existence of a single-phase field at a nonmetal content greater than 50 at%. In general, the X-ray phase analyses were not sensitive enough to indicate and identify the second phases which apparently should be present in all the UCS fuels used for the compatibility tests. Finally, except for the observation that the unit cell dimensions of the UCS fuels are generally larger than those of the UC fuels (table I), no systematic variation in a0 with that of either sulphur content or stoichiometry could be observed. The unit cell dimensions for the UCS fuels are, however, very similar to those reported by Shalek and White [2] and Love11 et al. [7] for their UC-US solid solutions, indicating that the UCS fuels used here probably also consisted of a solid solution UCS phase, but with a second phase present at stoichiometries greater than 50 at%. 3.2. Sodium-bonded compatibility tests In fig. 3(a) and (b) the tensile elongations and total carbon contents of the cladding sheet specimens are

Fig. 2. Optical microstructures of arc-melted UCS specimens with non-metal contents of (a) fuel P, 49.20 at%, (b) fuel Q, 50.17 at% and (e) fuel R, 50.42 at%. Magn. 500 X.

shown as a function of the stoichiometry of the UCS fuels annealed in sodium-bonded contact with the UCS fuels for 1000 h at 600 and 700°C respectively. A few results obtained previously with sulphur-free UC at 700°C have also been included [6]. From fig. 3(a) it

S. Venter / Compatibility of stainless steels

W.E. Stumpf

a

-&Ls-b --o-g ,

-

60OY 700~C

-

-g%

y.

-o-.-

s

ucs UCS

-x- UC

-g_&600~C 2

-

:I:::.,::?:

700*t

Steel 321

_

at 7oooc at 6OO’C at 7OOT

_

:- -: -0s : s ! : :“: Steel 347

:

. Y

0. ---*50

51 at.%(GO+N+SI in fuel

52

\’ v

t 0: 3 _

Steel 316 x Y)cJ+-: I : : : : ; : : : : E o4OOT E _ -o-UCS at 700°C / -4JCS at 6OOT .c / al?,6 %Q -X-UC at 700°C z $2. k x ZQ+-

2 5

! : f :

uuL. Sd 4

:

I : 0’

Steel 321

0

: : : ,* ~700°C.

b “2.

B

I? a-’ x

’

50

’

a ’

c ’

51

3 * 8 ’

52

at.%(C+O+N+SJ in fuel Fig. 3. (a) Room-temperature elongations and (b) total carbon contents of the steel specimens after annealing for 1000 h at 600 and 700°C in sodium-bonded contact with UCS and UC fuels of different stoichiometries. The initial values (+) of elongation and carbon content after annealing for 1000 h in vacuum only, are also given.

is apparent that embrittlement of the cladding is quite substantial at both temperatures, even for stoichiometries as low as 50.2 at% (C + 0 + N + S) and that at stoichiometries greater than about 50.5 at%, the elongation has reached such a low value that it is insensitive

199

to any further carburisation at higher stoichiometries. Fig. 3(b) shows, however, that carburisation indeed increases even further at higher stoichiometries and that a linear relationship is found between carbon increase in the steel cladding and the stoichiometry of the fuel at 7OO”C, while the linearity of this relationship is slightly less certain at 600°C. This figure also indicates that a pronounced temperature effect does exist for carburisation of the steels in sodium-bonded contact with the UCS fuels. Furthermore, it is apparent that there is little difference between the three types of steels used here. This indicates that the presence of the rather strong carbide. forming elements Nb and Ti in the steels 347 and 321 respectively, do not alter the carburisation behaviour significantly if compared to steel 3 16 which contains only the milder-carbide-forming element MO. Secondly, no difference in carburisation behaviour is apparent between the cold-worked microstructure of steel 3 16 and the well-annealed coarser-grained microstructures of the other two steels. This probably proves that cladding-alloy content and/or microstructure do not have a significant thermodynamic effect if eventual levels of total carburisation are considered after lengthy annealing times, although kinetic effects could still be possible during the early stages of carburisation. So, it is clear that the amount of carburisation of the cladding is a function only of the total non-metal contents of the fuel, and that UCS and UC fuels of the same stoichiometry produce similar amounts of carburisation of the cladding specimens, irrespective of the actual sulphur or carbon content of the fuel. In fig. 4 the total sulphur analyses of the steel cladding specimens are shown as a function of the at% sulphur in the UCS (and UC) fuels used in the compatibility test. (A few results of the vacuum-bonded tests have been included here.) It is clear that there is absolutely no increase in sulphur content of the steels during the compatibility anneal, indicating that embrittlement is caused only by carburisation due to carbon released by the fuel. Finally, fig. 5 shows some microstructures and microhardness gradient measurements of steel 347 annealed at 700°C for 1000 h in sodium-bonded contact with UC (fuel L) and UCS (fuel E) of about the same stoichiometry. Once again it is obvious that carburisation of the cladding specimens is a function only of the stoichiometry of the fuel.

200

W.E. Stumpf S. Venter / Compatibilityof stainlesssteels

6

so

I

I,,,,,,.,.1 0 0.4

BEFDRE INKAIING

0.8 12 1.6 at.% Sulphur in fuel

2.0

Fig. 4. Total sulphur content of the 0.65 mm thick steel specimens after annealing in contact with UC and UCS fuels in both sodium- and vacuum-bonded tests as a function of the sulphur content of the fuel.

Both the optical microstructures and these carburisation results do seem to indicate that these UCS fuels were not single-phase, but that the S merely displaced a carbon atom from the lattice, which in turn produced a higher carbide which caused the observed carburisation of the steels. In this respect the behaviour of the S in the UCS seems to be very similar to that of an oxygen or nitrogen atom in ordinary UC, i.e. these do not participate directly in any of the compatibility interactions of UC with steels, apart from the fact that they influence the stoichiometry of the UC by producing higher-carbide phases. The addition of sulphur to UC fuels has, therefore, no direct advantage at non-metal contents greater than 50 at% as far as the compatibility behaviour of the fuel towards the cladding is concerned. This conclusion is not in agreement with the preliminary observation of Shalek and White [2] as mentioned earlier. Unfortunately these authors did not provide chemical analyses of the UC-US mixtures after homogenisation, making it uncertain whether compositional changes from a higher stoichiometry towards a lower stoichiometry also occurred during homogenisation, as was found here during melting. 3.3. Vacuum-bonded

tests

Vacuum-bonded tests were carried out with fine ball-milled UCS powders at 600 and 700°C. The ball-

Fig. 5. Optical microstructures and microhardness gradients in steel 347 annealed for 1000 h at 700°C in sodium-bonded contact with (a) UCS-E and (b) UC-L fuels with similar stoichiometries.

milling caused an increase in oxygen content in all the UCS powders of between 500 and 700 ppm above those oxygen contents in solution reported in table 1. It is interesting to note that if UC is ball-milled under identical conditions, the oxygen increase is always at least 1500 ppm [5;6]. It does, therefore, appear that UCS fuels may possibly have a greater oxidation resistance than UC fuels. A similar observation has been made by Shalek and White [2]. The compatibility results with the ball-milled UCS powders at 600 and 700°C are shown in figs. 6(a) and (b) where the elongation and total carbon contents of the cladding-sheet specimens, respectively, are given as a function of the stoichiometry of the fuel. Once

W.E.Stumpf S. Venter / Compatibility of stainless steels

Steel _+GEZ 7oooc-P-__,_, 600%-8-L,_,_ 0. : ’ : : -60 at 700*c 8

Lo-;

:

: :

%.

Z”

-o- 600 “C

50

316

.

:

:

,

- --.: :‘7-

:.

Steel 321

; : : :

:

Steel

321

Steel

347

51 at.% (C+O%SI in fuel

:

i

52

Fig. 6. (a) Room-temperature elongations and (b) total contents of steel specimens after annealing for 1000 h and 700°C in vacuum-bonded contact with ball-milled powders of different stoichiometries. The initial values the control specimens after annealing for 1000 h under um only, are also given.

carbon at 600 UCS (+) of vacu-

again embrittlement of the steels is found for all the UCS fuels used here, and the amount of carbon pickup by the steels is also a linear function of the stoichiometry of the fuel. It should be kept in mind that there could be a very small contribution towards the embrittlement and carburisation due to free carbon released by the 500 to 700 ppm surface oxygen picked

201

up by the fuel during ball-milling [6]. In general, however, the embrittlement and carburisation of the steels is much less than that found with sodium-bonding. It is also apparent from figs. 6(a) and (b) that the elongation shows a decided temperature dependence, whereas the total amount of carbon in the steels does not reveal such a dependence. (This is exactly the opposite to what was found in the sodium-bonded tests.) The absence of a temperature effect in the carbon pickup by the steels probably indicates that the ratelimiting step in the process of the carbon transfer from the fuel to the cladding lies in the actual transfer of carbon from the fuel surface to the cladding surface through a gas phase [5,6], and that diffusion in the fuel particles or the cladding is probably not ratelimiting. The temperature dependence of the elongation in fig. 6(a) is, therefore, probably only of a structural nature, i.e. at lower stoichiometries of the UCS fuel the total carbon content of the steel, which is similar at both 600 and 700°C has a different effect on the room-temperature mechanical properties after annealing at 700 and at 6OO”C, possibly due to differences in coarsening rates of the carbides. At higher fuel stoichiometries, however, structural differences will become less well defined as the carbon content increases to higher levels, and the apparent temperature differences should disappear. Finally, analyses of the fuel after these tests revealed that the sulphur content of the fuel had not changed, but that the carbon content had decreased significantly for all the fuels used here. These analyses, however, proved that the stoichiometry of the fuel had not decreased to 50 at% during these vacuumbonded compatibility tests, indicating that further carburisation could still have been possible at longer times.

4. Conclusions (a) Arc-melting of UCS with stoichiometries greater than 50 at% changes the stoichiometry towards lower values. (b) A single-phase region in the U-C-S system does not exist at stoichiometries greater than 50 at% (at low sulphur contents), but a second phase, similar in appearance and carburisation behaviour to UC,, is found in the solid-solution UCS phase.

202

W.E. Stumpf

S. Venter / Compatibility of stainless steels

(c) The carburisation of cladding specimens in both vacuum-bonded and sodium-bonded contact with UCS fuels of stoichiometry greater than 50 at%, is a function only of the total non-metal content (C + 0 t N + S) of the fuel, and is not a function of the actual sulphur content. A UC and a UCS fuel of similar stoichiometry thus produce the same amount of carburisation. (d) Sulphur is not transferred from the fuels to the cladding. (e) UCS fuels appear to have a superior oxidation resistance as compared to UC. (f) There is apparently a difference in the ratelimiting process for carbon transfer from the fuel to the cladding between vacuum-bonded and sodiumbonded tests.

and Dr A. van Tets for some helpful discussions, and the South African Atomic Energy Board for permission to publish this work.

References [l]

[ 21

[3] [4] [5] [6]

Acknowledgements We would like to thank Mr P.W. Steyn, Mr N.J.S. Grobler, Mr P.J. Willems and Mrs G.A.G. Schwarzer for their experimental assistance, Dr G.H.B. Love11

[ 71 [8] [9]

[ 101

H. Wedemeyer, Karlsruhe Report, KFK 1111 (1969) Section XIV. P.D. Shalek and G.D. White, Carbides in Nuclear Energy, Proc. Sym. Harwell(1963) (McMillan, London, 1964) Vol 1, p. 266. W.E. Stumpf and 0. Gotzmann, Karlsruhe Report KFK 1385 (1971). W.E. Stumpf and S. Venter, Pelindaba (South Africa) Reports PEL 242 and 243 (1975). W.E. Stumpf and S. Venter, J. Nucl. Mater. 58 (1975) 211. S. Venter and W.E. Stumpf, J. Nucl. Mater. 59 (1976) 221. G.H.B. Lovell, A. van Tets and E. J. Britz, J. Nucl. Mater. 48 (1973) 74. G.H.B. Lovell, A. van Tets and E.J. Britz, J. Nucl. Mater. 51(1974) 337. G.H.B. Love11 and A. van Tets, unpublished work. W.E. Stumpf and S. Venter, unpublished work.