MOLYBDENUM IN GLASSES CONTAINING VITRIFIED NUCLEAR WASTE

Transcription

1 MOLYBDENUM IN GLASSES CONTAINING VITRIFIED NUCLEAR WASTE R.J. Hand, R.J. Short, S. Morgan, N.C. Hyatt, G. Möbus and W.E. Lee Immobilisation Science Laboratory, Department of Engineering Materials, University of Sheffield, Sir Robert Hadfield Building, Mappin Street, Sheffield, S1 3JD, UK Immobilisation of molybdenum in high level nuclear waste glasses melted under oxidising, neutral and reducing conditions has been studied using X-ray diffraction (XRD) and extended X-ray absorption fine structure (EXAFS). Molybdenum is present in all cases in the hexavalent oxidation state as the tetrahedral MoO 2 4 molybdate species, with a Mo-O contact distance of ca Å, contrary to suggestions that melting under reducing conditions can lower the Mo oxidation state. Additional studies on the early stages of incorporation of calcined high level waste into the borosilicate base glass show that a low melting point CsLiMoO 4 phase, in which molybdenum is also present as MoO 2 4, is readily formed. The MoO 2 4 groupings are not immobilised within the polymeric borosilicate network, instead they are likely to be located in extra-framework cavities, together with network modifier cations. The MoO 2 4 units are therefore readily available for incorporation into molybdatecontaining crystalline material, known as yellow phase, which is observed in vitrified high level waste. The glasses melted under oxidising, neutral and reducing conditions have been heat treated and the formation of powellite molybdates analogous to yellow phase was observed. Interestingly, samples prepared from glasses melted under reducing conditions showed increased powellite nucleation. STEM indicates that this may be because the

2 powellite molybdates are nucleating on metallic silver particles which are present in the reduced melts. Introduction The conventional technique for immobilising high level nuclear waste is to vitrify the waste by combining it with a borosilicate base glass. High level nuclear waste can contain corrosion products, a wide range of fission products, as well as lanthanides and actinides; some of the species present can have low solubility limits in borosilicate glass and this can put quite low limits on the amount of waste that can be incorporated into the glass thereby limiting the reduction in waste volume that can be achieved by vitrification. One particular element that causes difficulties is molybdenum [1,2]. The current paper details studies carried out on a high molybdenum waste stream (stream A) that may arise from future reprocessing technologies together with some results obtained from studies on the early stages of melting of a waste stream (stream B) that is currently being vitrified at Sellafield in the UK by British Nuclear Fuels Ltd (BNFL). The compositions of the two waste streams are given in table 1. Experimental Stream A The necessary chemicals for stream A and its associated base glass (in the weight ratios 20:80 and 35:65) were weighed to an accuracy of ±0.01g and placed in a stainless steel vessel. They

3 were then intimately mixed using a PVC spatula and transferred to an alumina crucible for preheating to 1000 o C at 1 o C/min in a muffle furnace. After transfer to the melting furnace the melt was allowed one hour batch-free time before a gas sparge was introduced into the melt. It has been suggested that reducing the oxidation state of molybdenum may lead to increased molybdenum solubility in borosilicate glasses [1] and thus the sparge was used to control the redox conditions within the melt. To create a strongly reducing atmosphere, N 2 gas containing 5% H 2 (BOC 99.5% pure) was used (reducing sparge) whereas compressed air (BOC) was used for a neutral atmosphere (air sparge). One end of a hollow pure alumina rod was connected to the gas cylinder via a plastic tube. The other end of the alumina rod was positioned in the melt so that it was approx. 1cm from the bottom of the crucible, and the gas control valve set at a level where the melt was gently bubbling giving a flow rate between 2-5 l/min depending upon the melt viscosity. This process was continued for 4.5 hours to give a total melting time of 5.5 hours. The melts were poured into a preheated steel block mould and annealed at 550 o C for one hour in a muffle furnace, before cooling to room temperature at 1 o C/min cm sections cut from the glass samples were placed in an alumina boat and heat treated for 48 hours at 600ºC in a tube furnace. Heating and cooling rates were 5ºCmin 1. The resultant materials were examined using X-ray diffraction (XRD), extended x-ray absorption fine structure (EXAFS) and scanning mode transmission electron microscopy (STEM). XRD was carried out using a Philips PW1410 X-ray Powder Diffractometer with CuKα radiation between 10º to 60º2θ. The EXAFS spectra were collected on Station 16.5 of the Synchrotron Radiation Source, Daresbury, UK. The spectra were recorded at ambient temperature in transmission mode using ion chambers filled with a mixture of Ar (698 mbar) / balance He (incident beam, I o ) and Kr 368 mbar / balance He (transmitted beam, I t ). The samples, in the form of fine powders, diluted where necessary with BN (in a ca. 1:10 volume

4 ratio), were packed into aluminium sample holders. At least two data sets were acquired per sample, over the energy range ev. The Mo K absorption edge was calibrated by measuring the K-edge from a Mo foil at ev and the data was analysed using EXCURV98. TEM analysis was performed using a Jeol JEM-2010F FEG-TEM. The samples were prepared by polishing on both sides alternately using 120, 240, 600, and 1200 grit papers until they were approximately 30µm in thickness. They were then cleaned using acetone, broken into small pieces approximately 3mm across, and glued using epoxy resin to a 3.05mm diameter copper ring with a 1000µm diameter hole in the middle. The samples were then further ground using a Gatan ion beam miller until perforation and were then carbon coated. An accelerating voltage of 200kV was used, and a Gatan Digiscan imager was used to acquire STEM images of samples. Stream B Stream B was supplied by BNFL as a full-scale simulant calcine. Borosilicate glass frit (also supplied by BNFL) was placed into an Inconel 601 crucible (75ml capacity) and heated to 1050ºC in a muffle furnace. The crucible was removed from the furnace and 25 wt% of stream B (taking into account the loss-on-ignition figures) was batched onto the melt surface. The crucible was then replaced in the furnace for 5, 10 and 16 minutes after which the entire crucible was removed and quenched in water. The solidified melts were photographed and then filled with cold-setting epoxy resin before sectioning using a diamond saw. One section of each sample was polished to 1µm using an oil-based lubricant. These specimens were then carbon coated and analysed using a JEOL JSM 6400 scanning electron microscope equipped with an Oxford Analytical Link EDX

5 system. For elemental dot mapping a preliminary EDX trace was taken to determine which elements to map. This EDX trace was also used for background subtraction from the final dot maps. The dot maps were obtained with an accelerating voltage of 20 kev and each element was mapped for 200 frames. Samples taken from the other sections were ground and sieved to <100µm prior to XRD analysis. Results Stream A Both air sparged samples were homogeneous and, as shown by figure 1, amorphous to XRD (peaks labelled Al were due to the sample holder). In comparison both reducing sparged samples showed evidence of phase separation and evidence of Ag was seen in the XRD traces for these samples (see figure 1). A model of 4 oxygens tetrahedrally arranged around a Mo 2 centre i.e. a MoO 4 grouping fits with the EXAFS results which show that only 1 coordination shell at 1.76(2) Å was present around the Mo. The lack of a second coordination shell suggests that charge balance is achieved by a random distribution of cations within borosilicate glass. There was no evidence of Mo being present in anything other than a hexavalent (VI) oxidation state. The 20wt% sample that had been melted in air was amorphous to XRD after heat treatment (figure 3). XRD patterns obtained on all other samples revealed silver metal and powellite molybdates (peaks labelled with in figure 3). No exact match could be made although AgNd(MoO 4 ) 2 (ICDD card [49-380]); Nd 2 (MoO 4 ) 3 (card [ ]) and NaCe(MoO 4 ) 2 (card [ ]) are all possible matches. Powellite molybdates can potentially incorporate a range

6 of +1, +2 and +3 cations [3-5] and there are a variety of such species present in stream A. Typical STEM micrographs of the crystals seen in heat treated samples are shown in figure 4. EDX analysis indicates that features labelled Den. and Mo are molybdenum rich and are presumably the powellite phases observed by XRD. There is some evidence that these molybdenum-rich crystals may be associated with Ag particles although further work is required to confirm this observation. Stream B The samples that were quenched 5 and 10 minutes after placing the waste calcine on the melt surface had a yellow green surface layer and little glassy material was apparent (figures 5a and b). The sample quenched after 16 minutes contained only localised regions of the yellowgreen surface layer (figure 5c). XRD of the surface layer showed it to contain CsLiMoO 4 ; other experiments [6] showed that this phase was also obtained if the calcine was heated with no base glass present (figure 6). SEM EDS dot maps of the sectioned samples confirmed the presence of increased amounts of Cs and Mo in the surface layer (figure 7). Discussion During the early stages of melting stream B, a cesium lithium molybdate (CsLiMoO 4 ) that again contains MoO 2 4 tetrahedra is formed on the melt surface. As melting advances this molybdate is incorporated into the glass melt (compare figures 5a and 5c). However examination of stream A indicates that molybdenum in borosilicate wasteforms is present as MoO 2 4, probably associated with a random array of modifying cations, and therefore is not well bonded into the glass matrix; similar results have been obtained by Calas et al. on the

7 French SON68 simulant nuclear waste glass [7]. Given that the MoO 2 4 groupings are not well bonded into the glass the yellow phase found in full scale simulant melts could, in principle, be material that has not been fully incorporated into the melt or material that has crystallised on cooling; in at least some cases observation of the yellow phase found makes the former seem to be more likely. This work contradicts the suggestion of Lutze [1] that melting in a reducing atmosphere could reduce the tendency for yellow phase formation by reducing the oxidation state of Mo in the glasses. Instead reducing conditions led to the formation of metallic silver within the melt and there is some evidence that the presence of such silver particles may have acted as nucleation sites for the powellite molybdates. Whether or not powellite molybdates are nucleated on silver metal particles, it is clear that if the melt is conducted under reducing conditions then 2 the MoO 4 groups are readily removed from the matrix as powellite molybdates. Powellite molybdates can potentially incorporate a number of the different +1, +2 and +3 cations present in the waste stream hence, even though the MoO 2 4 groupings within the glass appear to be associated with a random array of cations, it is reasonable to suppose that this does not impede crystallisation of powellite structured molybdates from the glass. Further support for this supposition is given by the observed lack of any second coordination shell in EXAFS results obtained on a heat treated sample of stream A glass containing 35wt% waste [5]. The formation of powellite molybdate precipitates within the surface layers formed during long term durability experiments on nuclear waste glasses has also been reported [8] which again tends to suggest that the MoO 4 2 groupings and associated cations are relatively easy to remove from the glassy matrix.

8 The association of MoO 2 4 groupings with cations within the glassy matrix means that species such as 137 Cs and 90 Sr can potentially be incorporated into molybdate phase that is formed within the glass; indeed elsewhere we have identified Sr within molybdate crystals formed in these glasses [5]. The incorporation of specific radionuclides within the low durability yellow phase is potentially of concern, however the radionuclides in question are relatively short lived (half lives ~30 years), and thus within any proposed storage/disposal scenario the levels of these radioisotopes will have become negligible before very long term corrosion processes affect the integrity of the stainless steel canister into which the glass is sealed. Conclusions 2 Molybdenum in vitrified high level wasteforms is present in the +6 oxidation state in MoO 4 groupings. Although these groups can be incorporated into the glass matrix they will tend to be associated with modifier cations and are thus not well bonded into the glass. It is therefore relatively easy to remove these grouping from the glass and incorporate them into crystalline phases such as the powellite molybdates observed here as well as in long term durability experiments on nuclear waste glasses. Acknowledgements We thank BNFL and the EPSRC for funding this work. We also thank B Bilsborrow of CLRC Daresbury Laboratory and Prof F Livens of the University of Manchester for their help with the EXAFS analysis. References

9 [1] Lutze W. in Radioactive Wasteforms for the Future (eds. W. Lutze and R.C. Ewing), North Holland, Amsterdam, 1988, pp [2] Schiewer E., Rabe H. and Weisenberger Scientific Basis for Nuclear Waste Management V (ed. Lutze W.) MRS Proc. 11, 1982, [3] Stedman N.J., Cheetham A.K. and Battle P.D. J. Mater. Chem. 1994, [4] Teller R.G. Acta Cryst., 1992, C48, [5] Short R.J. Incorporation of molybdenum in nuclear waste glasses PhD thesis, University of Sheffield (2004) [6] Morgan S., Rose P.B., Hand R.J., Hyatt N.C., Lee W.E. and Scales C.R. Ceramic Trans., 2004, 155, [7] Calas G., Le Grand M., Galoisy L., Ghaleb D., J. Nucl. Mater., 2003, 322, [8] Gong W.L., Wang L.M., Ewing R.C., Vernaz E., Bates J.K. and Ebert W.L., J. Nucl. Mater., 1998, 254,

10 Tables Table 1: Composition of the two simulant waste streams studied Stream A Stream B Oxide Weight % Mole % Oxide Weight % AgO Al 2 O BaO BaO 0.71 CeO CeO Cr 2 O Cr 2 O Cs 2 O Cs 2 O 4.86 Fe 2 O Fe 2 O Gd 2 O Gd 2 O La 2 O HfO MoO La 2 O Nd 2 O Li 2 O 5.53 NiO MgO 4.02 Rb 2 O MoO RuO Nd 2 O Sm 2 O NiO 0.40 SrO P 2 O TeO Pr 6 O TiO SiO Y 2 O Sm 2 O

11 ZrO SrO 1.60 Total TeO TiO Y 2 O ZrO LOI Table 2: Base glass compositions the slight differences between the two base glasses arises from the fact that stream B has an elevated Li 2 O content and thus the Li 2 O content of the base glass has been reduced Oxide Stream A /wt% Stream B /wt% B 2 O ± 0.8 Li 2 O ± 0.4 Na 2 O ± 0.5 SiO ± 1.0

12 Figures Ag Al Ag Al 35wt% waste reducing sparge Intensity 35wt% waste air sparge 20wt% waste reducing sparge 20wt% waste air sparge Degrees 2! Figure 1: XRD traces of the as cast stream A glasses Experiment Theoretical model FT amplitude 35wt% waste reducing sparge 35wt% waste air sparge 20wt% waste reducing sparge 20wt% waste air sparge Radial distance /Å Figure 2: Fourier transforms of the EXAFS data obtained on the as-cast stream A glasses

13 Ag Ag 35wt% waste reducing sparge Intensity 35wt% waste air sparge?? 20wt% waste reducing sparge 20wt% waste air sparge Degrees 2! Figure 3: XRD traces of heat treated stream A glasses a) b) Figure 4: STEM annular dark field images of heat treated stream A glasses prepared using a reducing sparge: a) 20wt% waste loading and b) 35wt% waste loading

14 c) b) 5 cm a Figure 5: Quenched samples of stream B glasses after a) 5 minutes; b) 10 minutes and c) 16 minutes melting Intensity Surface layer of sample quenched after 10 minutes CsLiMoO Degrees 2! Figure 6: XRD trace of a) stream B glass quenched after 10 minutes melting and b) CsLiMoO4 60

15 a) Cs 1 mm b) c) Mo Figure 7: Stream B glass melted for 5 minutes: a) Back scattered electron SEM image; b) Cs EDS dot map and c) Mo EDS dot map