AGR fuel corrosion: SIMFUEL-based method development in preparation for electrochemical studies of real spent AGR nuclear fuel electrode

Electrochemical corrosion of lower-activity spent fuel simulants, SIMFUELs, has been investigated in support of on-going analogous studies of a real AGR spent nuclear fuel electrode. Two electrode coupling methods have been investigated, to study the coupling of a sample of 25 GWd/tU SIMFUEL pellet to a sample of 20/25/Nb AGR cladding to replicate the real AGR fuel electrode setup. Method-1 involves coupling of individual standalone electrodes, while Method-2 involves coupling of the cladding and SIMFUEL within the same electrode. For both coupling methods, open circuit potentials and linear sweep voltammetry measurements were conducted in electrolyte solutions containing 30 μmol/m3 NaCl, dosed with NaOH to pH 8, 11.4 and 12.5. Resultant data were compared with single-component studies of SIMFUEL and cladding electrodes. Method-1 presents inconsistent electrochemical behaviour which we attribute to pitting/crevice corrosion effects on the cladding, whereas Method-2 shows a consistent behaviour that most resembles that of the SIMFUEL component.


Introduction
The majority of the UK's operating nuclear fleet are Advanced Gas-cooled Reactors (AGR). These use annular ceramic UO 2 fuel pellets, encased in a 20/25/Nb austenitic stainless-steel cladding to make up the fuel pins. Since 2018, the UK government has been transitioning from spent nuclear fuel (SNF) reprocessing to an open fuel cycle, initially involving long-term pond storage of SNF at Sellafield site, with an aim for final disposal in a geological disposal facility (GDF).
As of 2019, the UK's holdings of AGR SNF stood at ~ 2200 tHM (tonnes of Heavy Metal) with an estimated eventual inventory of > 4000 tHM by 2030, all ultimately destined for geological disposal. However, since GDF site construction has not started and with only rough estimates for planning purposes for waste package receipt by 2040s, it is likely that the UK's SNF holdings will remain in interim storage ponds for decades to come [1,2].
Currently, the UK's AGR SNF is stored in wet ponds in demineralised water dosed with caustic to a pH of 11.4, with chloride concentrations kept to < 1 ppm [3]. Undosed ponds have a pH of ~ 8; thus, the caustic dosing, which may be raised to pH 12.5 in the future deployments, is intended to inhibit the corrosion of the fuel pin cladding and thus maintain its structural integrity during the storage period. However, in the case of any fuel pins in which the integrity of the cladding has been compromised during its time in pond storage, water ingress may occur through cladding wall breaches to contact the UO 2 . This may lead to the formation of a corrosion cell at the interface between the pond water and the physically coupled fuel pellet and cladding. The consequences of this remain to be fully understood; thus, it is necessary to investigate the electrochemical corrosion processes taking place at this interface under relevant wet storage conditions.
Consequently, electrochemical studies on irradiated legacy UK AGR SNF have recently begun. For this work, a working electrode was fabricated from a small crosssectional piece of a reactor-discharged AGR fuel pin. Figure 1a shows the AGR fuel electrode, consisting of two components, a metallic stainless-steel cladding ring 1 3 and a semiconducting irradiated UO 2 fuel pellet. These are in close physical contact, see Fig. 1b, and may be bonded to each other due to fuel pellet expansion during in-reactor irradiation. The presence of a metal alloy and semiconducting ceramic with differing material properties within the same working electrode introduces complexities for electrochemical data acquisition and interpretation. Therefore, the use of non-active/lower-activity simulants has been explored to assist in the interpretation of real fuel data and thus support the addressing of this complex challenge.
In this study, we investigate two different simulant-based methods for replicating the real fuel-cladding coupling of the electrodes of Fig. 1a and b. Specifically, we study the electrochemical behaviour of a sample of unirradiated 20/25/ Nb stainless-steel AGR cladding coupled to a SIMFUEL that compositionally replicates a real fuel burn-up of 25 GWd/tU in a simplified simulant pond water (SSPW). These coupled studies are then compared to analogous studies of single-component electrodes comprised of the uncoupled SIMFUEL and cladding, in turn. The composite SIMFUEL/ cladding electrode experiments were conducted using two coupling methods: (a) Method-1 (M1)-consists of two separate resin cast "lollipop" electrodes of (i) a 25 GWd/tU SIMFUEL pellet, Fig. 1c, and (ii) a cross-sectional piece of 20/25/Nb AGR cladding, Fig. 1e. These are coupled together via a conducting copper cable to form a single working electrode, Fig. 1d. (b) Method-2 (M2)-consists of a single "lollipop" electrode containing both a 25 GWd/tU SIMFUEL pellet and a cross-sectional piece of 20/25/Nb AGR cladding, cast together in resin to form a single composite electrode wherein both components are physically as well as electrically coupled, Fig. 1f.
Comparison of these two methods will allow for determination of which mode(s) of SIMFUEL/cladding coupling best replicates the behaviour of the real fuel electrode system. Here, we present the preliminary findings of this campaign of work.
SIMFUEL pellets with a simulated burn-up of 25 GWd/tU and cooling time of 100 years were used as best match available to the real fuel burn-up of ~ 12.2 GWd/tU. The SIMFUEL pellets were fabricated by Hiezl et al. [4]. Pellet compositions were calculated using FISPIN calculations. Details of this as well as the actual fabrication and characterisations are given in reference [4].
All electrochemical measurements were conducted in a SSPW electrolyte solution comprised of 30 μmol/m 3 NaCl in Ultrapure de-ionised water (Merck Millipore Direct Q3 Water Purification System). Solutions were dosed with NaOH to give pHs of 8, 11.4 and 12.5. SSPW is a simplified composition based on the Simulant Pond Water (SPW) composition reported by Howett et al. [5]. The composition of SPW is shown in Table S1 in the supplementary material. In order to best simulate the conditions to be used in the study of the real fuel sample, electrolyte solutions were not purged to remove oxygen.
SIMFUEL and steel cladding materials were fixed onto a brass base using a conducting silver epoxy and cast in an epoxy resin within a PTFE mould. The working surfaces of the resultant electrodes were polished before all measurements. 20/25/Nb electrodes were polished using sandpaper from 600 to 1200 grit, followed by 6, 3 and 1 μm diamond polishing paste. M2 composite and SIMFUEL electrodes were polished using sandpaper from 600 to 1200 grit.
All voltammetry measurements were carried out using a 3-electrode cell, containing a working, reference (a BASi MF-2052 silver/silver chloride electrode (SSCE), Alvatek Ltd) and counter electrode (Pt metal coil, Alvatek Ltd). Measurements were conducted using a Metrohm Autolab PGSTAT128N potentiostat with Metrohm Autolab NOVA software for control and data acquisition.
Linear sweep voltammetry (LSV) measurements were conducted at a scan rate of 1 mV/s between the potential range of − 0.8 to 1.5 V. Prior to each LSV experiment, electrodes were immersed in the electrolyte under study for 15 h in order to reach an equilibrium open circuit potential (OCP), during which the potential was recorded as a function of time. Figure 2a shows the measured OCP as a function of time for a series of experiments, recorded in accordance with M1 at pH 8, 11.4 and 12.5, conditions that are representative of, respectively, undosed pondwaters, pondwaters subject to the current caustic dosing strategy and pondwaters subject to a potential future dosing strategy. For comparison, Fig. 2b shows analogous OCP measurements conducted on standalone AGR stainless-steel cladding and 25 GWd/tU SIMFUEL electrodes. Figure 2c shows polarisation curves, again measured in accordance with M1, recorded in SSPW at pH 8, 11.4 and 12.5. For comparison, Fig. 2d shows analogous polarisation curves of standalone cladding and SIMFUEL electrodes.

Results and discussion
Comparison of Fig. 2a and b, initially, appears to present M1 OCP behaviour similar to that of the cladding, suggesting that the anodic process that controls the mixed potential of the OCP is carried by the cladding. Similarly, the initial polarisation curves of Fig. 2c appear to present a current response again dominated by the cladding component. However, subsequent multiple repeats showed a range of results wherein the OCP and LSV data exhibited a mixture of SIMFUEL-dominated or cladding-dominated electrochemical response (repeat datasets are provided in the supplementary material).
In any DC electrical system presenting more than one possible current path, the majority of the current will flow though the path with least total resistance. Thus, the voltammetric behaviour of the coupled system will be predominantly controlled by the route that provides the least overall electrical resistance. In the case of the experiments reported on here, that overall resistance will be derived from a combination of the in-series contact, electrode material, bulk electrolyte and polarisation resistances, the last reflecting the electrochemical processes at the surface of either the SIMFUEL or the cladding. On this basis, two possible current paths are possible in an M1 experiment, through the SIMFUEL or through the cladding. However, based on Fig. 2a, c and the repeat datasets in the supplementary material, the preferred current path is not clear. Figure 3a shows OCP data measured in accordance with M2 at pH 8, 11.4 and 12.5. For comparison, Fig. 3b shows OCP measurements of standalone cladding and SIMFUEL electrodes under the same measurement conditions. Comparison of Fig. 3a and b suggests that open circuit behaviour is dominated by the SIMFUEL component. Consistent with the data of Fig. 2, the OCP is driven more negative with increasing pH, thus moving the equilibrium potential to a region where dissolution of UO 2 to UO 2 2+ is avoided [5]. Figure 3c shows polarisation curves, measured in accordance with M2 at pH 8, 11.4 and 12.5, with Fig. 3d showing analogous polarisation curves of separate cladding and SIMFUEL electrodes for comparison. Comparison of Fig. 3c with d shows a current response dominated by the SIMFUEL component. Unlike the M1 data of Fig. 2c, this observation is consistent across multiple repeats, implying that M2 surface electrochemistry is reproducibly driven by the more favourable processes 1 3 taking place at the SIMFUEL surface, such as the oxidation of UO 2 to UO 2+x, rather than the cladding, where passivation of the steel surface is expected.
Based on the studies by Siebert et al. and Shoesmith et al. [6,7], the forms of the SIMFUEL polarisation curves shown in Figs. 2d and 3d can be explained as follows: (i) under pH 8 conditions, at applied overpotentials < 0.2 V, the current reflects the oxidation of UO 2 to UO 2+x at the SIMFUEL surface. As applied overpotential increases above 0.2 V, oxidation of the surface to UO 2 2+ occurs which, at this pH, reacts to form a hydrated layer that part passivates the surface [8][9][10]. (ii) At the higher alkalinity of pH 12.5, the electrogenerated U(VI) reacts to form the soluble hydroxyanion, UO 2 (OH) − . The current inhibiting hydrated layer is thus not formed, resulting in a higher current being passed at high overpotentials [11,12]. Behaviour intermediate between these two extremes is seen at pH 11.4. As the LSV data of Fig. 3c are dominated by SIMFUEL behaviour, a similar interpretation can be applied. In Method-2, the 25 GWd/tU SIMFUEL component dominates the electrochemical behaviour at open circuit conditions and when under an applied potential. This result is consistent across experiment repeats and provides confidence that M2 electrode will behave predictably in further measurements. Initially, the current would be expected to be carried by the cladding, since its bulk material resistance will be less than that of the SIMFUEL. However, under low-chloride, alkaline pH conditions, the cladding surface would be protected at OCP and potentials positive of OCP by a passivating oxide layer [13], potentially inhibiting current flow at the surface of the cladding, whereas under high-chloride conditions, passive layer breakdown is accelerated due to increased adsorption   of Cl − onto the passive layer surface [14,15]. In contrast, the SIMFUEL component, under the same conditions, would be at overpotentials conductive to oxidation and dissolution, especially at pHs 11.4 and 12.5, therefore allowing the passage of current across the SIMFUEL-electrolyte interface.
In the M1 data of Fig. 2a and c, expanded upon in supplementary material, OCP and LSV results indicate that the preferred current path is unclear. Initially, based on bulk electrode resistances, the expected current path should be through the steel cladding. However, this is not always the case, suggesting that in some measurements, the combination of electrode + polarisation resistance of the steel cladding is greater than that of the SIMFUEL. We hypothesise that under the conditions applied here, due to the aforementioned passivating oxide layer-which might be expected to form over the potential range indicated by the shaded area of Fig. 4c-the polarisation resistance of the steel must sometimes be greater than that of the combined bulk electrode and polarisation resistance of the SIMFUEL, such that the preferred current path is through the SIMFUEL.
As described above, under the applied potential conditions employed here, the dominant electrochemical process at the SIMFUEL surface is the oxidation of UO 2 to UO 2+x and consequently UO 2 2+ , while on the steel surface, growth of passive oxide layer is preferred. The process at the SIM-FUEL is facile and may readily generate appreciable current through the electrode. While the process at the steel cladding may passivate the electrode surface, impeding current flow. This, of course, implies that current will always preferentially pass through the SIMFUEL. However, the passivation afforded by the oxide layer may be compromised by adventitious corrosion processes such as pitting and crevice corrosion, therefore allowing current to pass through the cladding. The intermittency of current passage through the steel under OCP and LSV measurement conditions may also reflect the highly surface preparation-dependent nature of such corrosion processes. This explanation may be confirmed by quantification of the respective electrode bulk and polarisation resistances by electrochemical impedance spectroscopy (EIS) and this is the subject of ongoing work.

Conclusions
We demonstrate the use of simulant-based methods for replicating electrochemical corrosion behaviour of a real AGR SNF electrode. Two mechanisms of electrode coupling were investigated, differing only by the mode of physical contact between the cladding and SIMFUEL. The two coupling methods show differing electrochemical behaviour in the simulant systems. Data from Method-2 experiments have shown that the SIMFUEL component dominates the electrochemical behaviour over all experiment types conducted. Thus, during M2, the expected current path is initially through the cladding bulk, it then switches materials at the electrode-electrolyte interface to be carried by the lower polarisation resistance associated with the SIMFUEL oxidation processes.
Data from Method-1 coupling studies show a variation in experiment-to-experiment electrochemical behaviour with no preference towards the SIMFUEL or claddingdominated behaviour during either OCP or LSV measurements. The uncertainty could possibly be attributed to the competing current paths offered by the polarisation resistances associated with passivating oxide film formation on the cladding surface versus the more facile electrochemical processes on the SIMFUEL surface, complicated by inconsistently observed loss of cladding passivation afforded by adventitious corrosion processes such as pitting and crevice corrosion. The above-mentioned variation in whether the SIMFUEL or cladding dominates the electrochemistry in M1 is then consistent with the favoured current path being through the SIMFUEL in the absence of cladding pitting/crevice corrosion, the opposite being true in its presence.
Future work will include the use of EIS to quantify the respective bulk and polarisation resistances and their roles in controlling the behaviour observed in Method-1. Additionally, these data, especially that obtained in Method-2, will be used to support the campaign of work on irradiated legacy spent AGR nuclear fuel.