- Email: [email protected]
Monitoring uranium corrosion in Magnox sludge using X-ray computed tomography: A direct analogue to “legacy” fuel storage ponds
Corrosion Science xxx (xxxx) xxxx
Contents lists available at ScienceDirect
Corrosion Science journal homepage: www.elsevier.com/locate/corsci
Monitoring uranium corrosion in Magnox sludge using X-ray computed tomography: A direct analogue to “legacy” fuel storage ponds C. Paraskevoulakosa,*, K.R. Hallama, A. Adamskab, T.B. Scotta a b
Interface Analysis Centre, School of Physics, University of Bristol, Bristol, BS8 1TL, UK Vertex, GF, Sir Christopher Harding House, North Shore Road, Whitehaven, CA28 7XY, UK
A R T I C LE I N FO
A B S T R A C T
Keywords: Interfaces (C) Pitting corrosion (C) Alloy (A)
During the early years of nuclear industry within the UK, the typical route followed to handle Magnox (Mg-Al alloy) casing after separation from the nuclear fuel has been immersion in water temporarily in specificallydesigned water ponds. Over the decades of storage, Magnox has been corroding in water, resulting in the formation of sludge within the ponds; known as Corroded Magnox Sludge. Uranium, which has been also contained within the ponds as a result of faults occurring during fuel-casing separation, is susceptible to corrosion in waterrich environments. Under certain circumstances, uranium corrosion products can be highly pyrophoric during oxidation in air. This raises serious concerns regarding the safety margins of nuclear pond decommissioning. The present study attempts to shed light on the behaviour of uranium in a Magnox sludge-surrogate environment, using X-ray computed tomography at distinct time intervals over an 18-month period. Focus was given onto the uranium-sludge interface to determine metallic corrosion onset and evolution.
1. Introduction
open air, implying that flammable hydrogen resulting from uranium and Magnox corrosion can be released into the gas phase above the pond. In addition, UH3 can be highly exothermic when oxidised in air (under certain circumstances related to its mass and surface area amongst others) [7, 8]. It is, therefore, of significant importance to determine the corrosion behaviour of uranium metal within a Magnox sludge environment. Corrosion rate monitoring will allow estimation of the extent of corrosion within the storage ponds, while potential formation and, particularly, persistence of UH3 within the ponds would raise concerns about fuel and skip’ lifts out of the pond during decommissioning.
Magnox nuclear reactors, using metallic uranium as fuel operated in the UK from the 1960′s until just a few years ago. The metallic fuel was encased in a magnesium alloy ‘Magnox’ cladding, which, along with the uranium metal, is susceptible to corrosion by water [1]. Until 1992, a significant volume of Magnox waste materials, including spent fuel assemblies, had been accumulated in the storage ponds at Sellafield in Cumbria [2]. These ponds were maintained at a high pH (above 10.5) to minimise corrosion [3]. However, considerable corrosion has still occurred over extended periods, raising concerns about the safe decommissioning of the facilities [4]. Corroded Magnox Sludge (CMS), arising from long-term corrosion of the cladding material, is a prevalent residual material amongst others (e.g. fuel fragments, wind-blown debris and concrete degradation products) [2]. Corrosion of Magnox forms brucite (Mg(OH)2) and liberates hydrogen [5]. Embedded uranium metal fuel is also expected to have been exposed to water throughout the storage period. This is validated by the presence of leached fission products (cesium and strontium) and traces of actinides, including uranium and plutonium, within these sludge beds [6]. Uranium reacts with water to form uranium hydride (UH3), uranium dioxide (UO2) and hydrogen, presenting potential uncontrolled thermal hazards during decommissioning. Most of the legacy ponds are
2. Background Uranium corrosion behaviour has been extensively studied over the last decades, predominantly in simplified environments including air, liquid water, hydrogen and water vapour [9–13]. Only recently, considerable research has been conducted to investigate uranium corrosion in more complicated and interacting environments such as grout, attempting to shed some light on the performance of intermediate level waste (ILW) drums [14–20]. Stitt et al. [21] demonstrated the feasibility of using X-rays to investigate uranium-grout systems over a prolonged time period. Thus, the fact that uranium corrosion could be
⁎
Corresponding author. E-mail addresses: [email protected] (C. Paraskevoulakos), [email protected] (K.R. Hallam), anna.m.adamska@sellafieldsites.com (A. Adamska), [email protected] (T.B. Scott). https://doi.org/10.1016/j.corsci.2020.108551 Received 22 October 2019; Received in revised form 23 January 2020; Accepted 17 February 2020 0010-938X/ © 2020 Elsevier Ltd. All rights reserved.
Please cite this article as: C. Paraskevoulakos, et al., Corrosion Science, https://doi.org/10.1016/j.corsci.2020.108551
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
encapsulate the uranium specimen. The mixture was allowed to cure in ambient conditions for three days before being hermetically sealed, using a second flange connected with the stainless-steel pot via a rubber gasket. No deliberate exchange of the sample inner atmosphere was allowed during the entire experimental period. The research plan focused on performing XCT scans (Zeiss Xradia 520 Versa) at distinct time intervals over a total period of approximately 18 months to determine the behaviour of the uranium within the CMS environment. A specifically-designed metal holder was used to host the sample within the XCT chamber, ensuring that the sample remained steady as the XCT stage rotated during scanning. Schematic illustration of the experimental set up is shown in Fig. 1.
monitored at distinct time intervals in a non-destructive manner was a considerable breakthrough compared to the traditional post-mortem analysis, where samples could be only investigated at a single time step after disrupting the system. So far, experimental studies on uranium corrosion in Magnox sludge have not been published to the best of our knowledge. Examination of CMS samples extracted directly from a legacy spent nuclear fuel storage pond located at Sellafield using scanning electron microscopy (SEM), energy dispersive X-ray (EDX) analysis and Raman spectroscopy revealed uranium traces, validating the dissolution of fuel parts into the CMS [6]. This paper presents the results of an initial feasibility timeresolved experimental project, where a uranium specimen is corroding in ambient conditions in CMS. X-ray computed tomography (XCT) has been used at distinct time intervals to primarily observe the extent of corrosion and the behaviour of the corrosion products forming.
4. XCT scanning parameters XCT scans were performed at four different time periods: 1) 20 days, 2) 50 days, 3) 400 days, and 4) 540 days after preparation. Scans were predominantly focused at the uranium-CMS interface to identify potential corrosion traces. A repeated process of fine adjustment was attempted with regards to scanning parameters to target compatibility between different scans, allowing direct comparison. Both low- and high-magnification scans were conducted, while the entire uranium specimen was probed. Since uranium is significantly dense (19.1 g/ cm3), a 140 kV beam was used to achieve adequate transmission levels. The resolution of the low-magnification scans, which enabled capturing a large sample volume, ranged between 30 μm/pixel and 35 μm/pixel. Selective high-magnification scans were also performed to investigate smaller volumes in greater detail. The resolution for these long duration scans (∼30 h) was between 2.8 μm/pixel and 2.9 μm/pixel.
3. Materials, sample preparation An experimental set-up was built to act as a simplified model of a legacy pond, where uranium fuel parts are located and corroding over the storage period. The goal was to monitor the behaviour of uranium when encapsulated in CMS over a considerable time period using XCT. A stainless-steel pot (25 mm diameter) with a flange on top, which could offer a sealed containment for the interacting materials, was used to host the uranium and the CMS. A Teflon® cup was placed around the inner surface to separate the steel containment from the CMS. A single uranium specimen (1.5 mm x 0.9 mm x 17.5 mm) was prepared after machining down from a mechanically-polished natural uranium disc. EDX was performed to determine the elemental composition. The results are presented in Table 1. The high content of impurities (C, N, O) is typical for this specific material, which originates from an unirradiated fuel rod used for Magnox reactors. Carbon and nitrogen peaks are mainly associated with the presence of carbide and carbo-nitride inclusion particles, while carbon deposition is also expected on the sample surface due to the electron beam (SEM). The oxygen peaks are related to the oxide layer formed vigorously on the surface of the uranium specimen before placed in the SEM chamber. Within only a few minutes of exposure in air, fresh uranium surfaces can tarnish, indicative of thin oxide layer formation. The uranium specimen was then immersed into the CMS, which was previously poured inside the cup. CMS powder, originally provided by industrial partners, was used as the core material to prepare the sludge. The powder was manufactured by corrosion of Magnox alloy type AL80 in water at 50 °C. The material was then sieved to remove metal and centrifuged to remove water. After sieving there was typically less than 1% w/w Magnox alloy remaining in the CMS powder. The final solids content was then measured and adjusted by oven drying. The solid, powderised material was mixed with distilled water (118 % w/s) to prepare the sludge, which was subsequently used to
5. XCT results 5.1. 20 days after preparation The first XCT scan was performed 20 days after preparation. It was expected that no signs of uranium corrosion would be observed at such a primary sample age and, therefore, this scan would act as a reference for subsequent scans. Post scanning, the data were reconstructed and processed using the X3DM viewer and Avizo software [22]. Characteristic 2D projections of three different planes, showing the interacting materials can be observed in Fig. 2. Fig. 3 also presents a single representative 3D model of the entire sample, where uranium can be observed in the central area. Detailed observation of the 2D projections and the reconstructed 3D model revealed the presence of features at the uranium-CMS interface. These features were only observed near the top of the uranium specimen. Fig. 4 illustrates two separate 3D views of the sample under different rotation angles with a couple of magnified screenshots attached next to each one, representing the top and the middle region of the uranium. It can be clearly observed that the top part of the uranium exhibits traces of morphology change (yellow arrows), while in the middle region uranium seems intact. The colour of these blisters appearing across the uranium surface in the greyscale is between the uranium and the background Magnox. Since tomographic images are practically density maps, the blisters’ densities lie between those of uranium (19.1 g/cm3) and CMS (1.5 g/cm3 – 2.5 g/cm3). Selective highmagnification scans were also performed, focusing on the top part of the uranium where features were observed. Reconstructed data validated their presence at the uranium-CMS interface.
Table 1 Elemental composition of natural uranium used for preparation of the ILW drum – simulant packages as determined via energy dispersive Xray analysis, (20 kV acceleration voltage). Natural uranium Element
Wt %
C
7.78 ± 0.67 4.51 ± 0.27 5.08 ± 1.52 0.23 ± 0.05 0.11 ± 0.04 82.50 ± 1.87
N O Al Si U
5.2. 50 days after preparation XCT scans were repeated with the sample aged 50 days. The goal was to confirm that the features were still present and, if so, monitor their size. Potential growth would indicate that uranium continues to react in the sludge environment. Data reconstruction and subsequent 2
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 1. Schematic illustration of the experimental set up used to investigate the behaviour of uranium in a CMS environment. Dimensions are given in mm.
initiated. The third set of analyses would enable the drawing of more conclusions about the system’s long-term behaviour with regards to corrosion evolution. The configuration of the scan was adjusted so that the data of the third set were directly comparable to the first two analysis cycles. The system’s status more than a year after sample preparation can be evaluated based on Fig. 6. Areas near the top of the uranium are significantly different than the corresponding areas at 20 days and 50 days. The reaction front seems to have propagated both in the core metal and across its length. Feature volumes have increased considerably compared to the early stages of sample monitoring. It is also interesting to note that in the top region coalescence of the blisters has occurred across the surface, forming a consistent layer of varying thickness. The first two datasets exhibited a slightly different behaviour since isolated pits/blisters were observed across the uranium surface instead of a coherent layer. Migrated particles within the sludge volume are also apparent (Fig. 6, blue arrows) in agreement with observations drawn from the second set of analyses (50 days).
3D visualisation proved that the feature volume had increased. Several pits were observed in the uranium surface, not only at the top part of the metal but also in the middle height region. Fig. 5 illustrates representative 3D views and some allocated magnified screenshots confirming the presence of features at the uranium-CMS interface. It is also obvious that, in contrast to data collected 20 days after preparation, areas on top of the uranium specimen within the sludge volume are occupied by particles resembling those formed on the uranium surface. This behaviour probably suggests a migration mechanism of the particles from the uranium surface towards the sludge volume over this period of 30 days. These areas are highlighted in Fig. 5 using blue arrows. 5.3. 400 days after preparation The third set of measurements was performed over a year after sample preparation (400 days). The first two datasets proved that corrosion of uranium within the sludge environment had probably
Fig. 2. 2D projections associated with three different plane orientations, probing the interacting materials of the sample. U, S and CMS stand for uranium, steel and corroded Magnox sludge respectively. 3
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 3. Reconstructed 3D model observed under two different rotation angles.
5.4. 540 days after preparation
no visible change.
The fourth and final set of measurements was performed almost 1.5 years after the sample was prepared (540 days). After demonstrating corrosion propagation over the first year of sample aging, this scan was performed to reveal if there was any further growth of the blister’ volumes or if the reaction had slowed down. The data collection process followed was compatible with the previous sets so that the results are comparable. Fig. 7 illustrates selective screenshots of the data collected for this last, fourth scan. It can be observed that 540 days after the experiment initiated the geometry of features at the uranium-CMS interface is similar to the corresponding profile 400-day profile. Comparing the same areas in both datasets reveals no difference while the bottom area of the uranium rod, which was intact during the first year of scanning showed
5.5. Comparing the datasets A direct comparison between the four data sets corresponding to different sample ages follows. Image segmentation was performed to isolate uranium metal from the rest of the materials (including the blisters at the uranium-CMS interface), allowing investigation of the metal surface. The results are grouped together and presented in Fig. 8. Craters around the uranium surface formed within 20 days of preparation. At later stages, it is evident that additional craters have formed while those which were observed at the early stages have grown both inwards and across the metal’s surface. It is interesting to note that the craters formed and developed only on the top half of the uranium specimen. The bottom half seems intact even at the latest stage of
Fig. 4. Sample views under two different rotation angles, with attached magnified screenshots representing the top and the middle uranium regions. Yellow arrows demonstrate the presence of features at the uranium-CMS interface. Views correspond to 20 days after sample preparation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 4
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 5. Sample views under two different rotation angles, with attached magnified screenshots representing the top and the middle uranium regions. Yellow arrows demonstrate the presence of features at the uranium-CMS interface, while blue arrows point at migrated particles within the CMS volume. Views correspond to data obtained 50 days after sample preparation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
part of those features within the sludge during the 30-day period between the two scans is confirmed. This behaviour was previously (Sections 5.1-5.3) commented upon based on the low-magnification scan dataset. However, Fig. 9 shows in greater detail that the features morphology has changed between the two first scans. Based on the data acquired it can be safely stated that the bulges observed to form and grow at the uranium-CMS interface are related to the rapid corrosion of the uranium metal in the sludge.
investigation (540 days after preparation). This observation was consistent over the entire dataset volume. No noticeable changes occurred between the last two scans (400 days to 540 days). In addition to these low-magnification scans which allowed probing the entire sample volume, high-magnification scans were also performed to investigate the morphology of the interfaces in greater detail. Only the top part of the uranium specimen was investigated. The same process was followed with regards to scan configuration and data processing. High-magnification scans were performed only at the two early monitoring stages (20 days and 50 days post sample preparation), where low-magnification scans exhibited slight differences even though progress of the reaction front could be undeniably stated to have occurred. Fig. 9 illustrates results from the high-magnification scans. Characteristic 2D projections under three different planes, both at 20 days and 50 days after sample preparation, can be observed. Volumetric increase of the features that appeared at the uranium-CMS interface after the first scan (20 days) is evident. Also, migration of a
6. Quantitative analysis Data retrieved from the low-magnification scans were used to perform quantitative analysis. Avizo [22] software was used for material segmentation and, predominantly, volume calculations. Segmenting uranium metal from the corrosion products (assumed to be UH3) was feasible even if the scan resolution was low (30 μm/pixel - 35 μm/pixel) considering the size of the uranium specimen. A progressive increase of
Fig. 6. Sample views under two different rotation angles, with attached magnified screenshots representing the top and the middle uranium regions. Yellow arrows demonstrate the presence of features at the uranium-CMS interface, while blue arrows point at migrated particles within the CMS volume. Views correspond to data obtained 400 days after sample preparation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 5
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 7. Sample views under two different rotation angles, with attached magnified screenshots representing the top and the middle uranium regions. Yellow arrows demonstrate the presence of features at the uranium-CMS interface. Views correspond to data obtained 540 days after sample preparation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 8. Segmented uranium volume sets at different time periods under three different rotation angles. Each set contains four uranium volume profiles corresponding to growing sample ages starting from the left side (20 days) to the right side (540 days). 6
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 9. Characteristic 2D projections associated with the high resolution XCT scans under different 2D planes. Images presented correspond to data collected 20 days and 50 days after sample preparation. Progress of the reaction front can be observed.
showing the best correlation results among alternative trends. Even though just four data points limits the ability to draw reliable conclusions, it seems that there is a rapid growth of the corrosion product volume over the first 100 days, which subsequently slows down. Based on the fitting formula provided to correlate the corrosion product volume evolution with time, it seems that blister-type corrosion initiated about 17 days after sample preparation. However, it is recognised that
the corrosion product volume was determined, associated with reduction in the uranium volume. Based on the calculated volumes of the corrosion products and the uranium at the three different time periods, the evolution of the corrosion percentage was tracked. The evolution of the corrosion percentage and the corrosion product volume over time are given in Fig. 10. Logarithmic fitting was applied to the data points in both cases, 7
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 10. Evolution of the corrosion percentage (left) and the corrosion product volume (right) with time.
intervals, revealed the formation, development and volumetric growth of features at the uranium-CMS interface. XCT findings indicate the onset and progress of uranium corrosion within the sludge. Uranium metal reacted with water present in the sludge to liberate hydrogen via the following chemical formula.
the number of available data points is not adequate for reliable use of the curve fitting formula. Based on the volume of material loss, corrosion rate was also calculated to determine quantitatively the drop in the growth rate of the corrosion products. Non-corroded uranium metal volume was measured at each different time interval based on the Avizo material segmentation process. Consequently, the mass of uranium reacted between two consecutive XCT scans could be calculated. The initial surface area of the uranium specimen was also known, since the specimen was thoroughly measured before it was encapsulated in the CMS. As a result, the following formula was implemented to determine the corrosion rate at the specific timings when XCT scans were conducted:
Corrosion rate (gU /m2 /day ) =
U mass loss (g ) ISA (m2) × Time Period (days )
U + 2H2 O → UO2 + 2H2
(2)
It has been recognised that, under certain circumstances, uranium can also react with hydrogen to form uranium hydride (Eq. 3).
U+
3 H → UH3 2 2
(3)
XCT scans demonstrated the presence of corrosion products at the uranium-CMS interface, resulting from uranium corrosion in water. However, the dataset obtained cannot be used to characterise the corrosion products and determine whether UH3 has formed and persisted. Potential dissection of the system and subsequent surface analysis would not provide any reliable results since UH3 oxidises very rapidly in air. Based on the literature available, it is possible to predict the identity of the corrosion products formed based on their morphology. Stitt et al. [14,21], performed synchrotron X-ray tomography experiments on systems where a single uranium specimen was encapsulated in grout, mimicking the ILW drums in Sellafield storage facilities [19, 23]. The results showed that a coherent layer of relatively constant thickness had formed at the uranium-grout interface of one of the samples, which allowed to cure in ambient conditions. In contrast, a second sample, which was exposed to hydrogen to deliberately form UH3 on the uranium surface exhibited a completely different morphology. Blisters were observed on the uranium surface at different locations, instead of a smooth uniform layer of constant thickness. Synchrotron X-ray powder diffraction (XRPD) analysis was also conducted to determine the phases present in both samples. The results revealed a UO2 layer of uniform thickness across the first sample, while UH3 blisters were detected on the second one. Same findings have been also reported on similar uranium-grout systems [16]. A selection of characteristic images available in the literature, showing the different morphologies of corrosion products (UO2 and UH3) in uranium-grout systems, is given in Fig. 12. One representative image related to the uranium-CMS system presented in the previous sections is also included to allow comparison. Figs. 12a and 12b illustrate 3D models of isolated uranium specimens where the corrosion products are attached and presented using different colours. Fig. 12a corresponds to a uranium-grout system which evolved in ambient conditions and, therefore, only a coherent UO2 layer had formed, whereas, Fig. 12b illustrates the state of a uranium-grout system subjected to hydrogen to deliberately form UH3. Craters appear on the isolated uranium surface where blister-type corrosion products (typical of UH3 presence) are also observed. Figs. 12a and 12b are reproduced
(1)
where ISA stands for the initial surface area of the uranium specimen and (since the specific day of corrosion onset is unknown), time period refers to the sample age when the XCT scans were acquired. The results are shown in Fig. 11. Since material segmentation is a visual process, several intensity threshold values were tested in the form of a sensitivity analysis. The calculated corrosion rate 20 days after sample preparation was considerably high. However, after the end of the second scan (50 days from sample preparation) it had already dropped to almost half. The corrosion rate at the last two sets of measurement (400 days and 540 days) was found to be low and almost constant. 7. Discussion Both low- and high-resolution scans were performed to investigate the behaviour of uranium within a CMS environment, over a period of approximately 18 months. All scans performed at distinct time
Fig. 11. Evolution of the corrosion rate over time. 8
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 12. Screenshots reproduced from literature [14,16,21], illustrating corrosion products formed at uranium-grout interfaces are compared with the behaviour of a uranium-CMS interface investigated in this study.
possibly contributed to maintain this elevated hydrogen pressure. Post-formation persistence of UH3 is often questioned and cannot be validated or safely predicted based on the morphology of the corrosion products. Oxidation of UH3 (Eq. 4), which in air can raise serious concerns since it is highly exothermic, is one possible scenario to have occurred within the CMS environment.
and adapted from [21]. Fig. 12c is a radiography shot displaying the uranium-grout interface of a sample subjected to hydrogen online on the I12 beamline at Diamond Light Source [24]. The UO2 layer formed during sample curing has been detached from the uranium metal, while UH3 blisters (confirmed via XRPD analysis) can be observed between the uranium metal and the detached oxide layer. This image was reproduced and adapted from [14]. Fig. 12d (reproduced and adapted from [16]) shows a similar situation, where uranium encapsulated in grout and stored within a stainless steel mini drum was exposed to hydrogen. The formation of blisters across the uranium surface at the early reaction stages, associated with UH3 presence, is also evident. In compliance with the UH3 blister-type morphology shown in all previous examples in uranium-grout systems, which differs significantly from the coherent UO2 layer of relatively uniform thickness, it can be assumed with considerable safety that the corrosion products observed to grow in the uranium-CMS system resemble the UH3 morphology. In case this assumption is valid, it is interesting to note that in contrast with all the corresponding cases mentioned in uranium-grout hydrided systems, the uranium-CMS system was not exposed to hydrogen. This means that there was no deliberate attempt to form UH3. Instead, it formed and developed naturally in ambient conditions. Natural formation of UH3 in a uranium-bearing system is a rare scenario based on the available literature [25]. A combination of elevated hydrogen pressure and high temperature is usually required to activate a uranium-hydrogen reaction. Therefore, reaction initiation and progress over a one-year period in ambient conditions is an interesting finding and worthy of further study. With uranium encapsulation in CMS, hydrogen is expected to have been liberated as part of a uranium-water reaction. It is also possible that hydrogen evolution could have occurred from magnesium reaction with water, in the case of any pure metal swarf being present in the CMS powder mixture. It seems that the generated pressure was sufficient to allow hydrogen to permeate through the sludge, reach the metal surface and favour a uranium-hydrogen reaction, which consequently led to the formation and growth of UH3 particles. The hermetically-sealed containment has
UH3 +
7 3 O → UO2 + H2 O 2 2 2
(4)
Stitt et al. [26] performed synchrotron experiments, specifically targeted on the persistence of UH3, using uranium-grout systems previously hydrided artificially and subsequently embedded in water. The XRPD analysis revealed UH3 peaks even 10 months after sample preparation, implying that UH3 had only partially or not at all oxidised during this period. Decommissioning of nuclear waste facilities, including the legacy ponds, has only recently started and will proceed over the following decades. XCT time-resolved analysis showed that uranium corrodes rapidly while encapsulated in CMS. Based on the morphology of the corrosion products probed, UH3 is highly possible to form, and potentially, persist over time. It is important to note that UH3 formation in a uranium-sludge system in ambient conditions has not been previously demonstrated, to the best of our knowledge. This also applies for other uranium-containing systems, including uranium-grout packs. The sealed containment of the sample investigated for this study has probably favoured corrosion initiation and progress, since hydrogen generated from uranium-water (and, potentially, magnesium-water) reaction was probably confined within the sample’s core. If the results drawn from this study were applicable to the real case of legacy ponds, it would be safe to claim that a considerable volume of the uranium immersed into the legacy ponds decades ago has been significantly corroded. It is also highly possible that UH3 has formed as part of uranium-water corrosion, but it is questionable if this has persisted over the storage period. Lifting out solid components or sludge from the ponds is essential for decommissioning. The presence of UH3, either on the solid matrix or in the sludge mixture could reasonably be identified 9
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
water vapour corrosion regime. The corrosion rates derived at different time intervals during this experimental study are plotted against time in Fig. 14. Relative data found in the literature are also included, as in Fig. 13. It seems that uranium is significantly more susceptible to corrosion within a CMS environment than in grout. The comparison between the two systems (uranium-CMS and uranium-grout) is valid since most of the experimental parameters, including uranium elemental composition, geometry and temperature are the same. 8. Conclusions The primary results of a feasibility study using XCT to investigate the corrosion of metallic uranium in a CMS environment are presented in this paper. Scans were performed at distinct time intervals for a period of more than a year. The results revealed that uranium corrosion had already initiated within the first 20 days after sample preparation and progressed rapidly during the following months. Based on the morphology of the corrosion products, it is believed that UH3 has formed in ambient conditions and potentially persisted.
Fig. 13. Long-term uranium corrosion rates in various environments as found in the literature compared with the corrosion rates determined for different time periods for the uranium-CMS sample investigated for the present study.
Data Availability The raw/processed data required to reproduce these findings cannot be shared at this time as the data also forms part of an ongoing study. CRediT authorship contribution statement C. Paraskevoulakos: Conceptualization, Methodology, Software, Formal analysis, Investigation, Writing - original draft, Writing - review & editing. K.R. Hallam: Validation, Writing - review & editing. A. Adamska: Resources, Conceptualization. T.B. Scott: Conceptualization, Project administration, Supervision.
Fig. 14. Time-resolved evolution of uranium corrosion rates in various environments as found in the literature compared with the corrosion rates determined for the uranium-CMS system presented in this study.
as a hazard, since it can be highly exothermic when oxidised. The XCT scans enabled precise calculation of the uranium corrosion product volume at different time intervals, as well as of the volume of the remainder uranium metal as corrosion progressed. The accumulated corrosion rates at four distinct moments when scans were performed (20 days; 50 days; 400 days; 540 days) were also determined based on the relative volume calculations, as shown in Fig. 11. Comparison of the datasets corresponding to the last two scans (at 400 days and 540 days) revealed almost negligible change. It is highly probable that the availability of hydrogen produced from Magnox corrosion dropped significantly and that, therefore, a uranium-hydrogen reaction could not take place. However, even if hydrogen was still abundant within the cell, potential diffusion though the corrosion layer to reach the underlying metal surface and, subsequently, activate reaction would be questionable. The results were used to compare with relevant available literature data where uranium corroded in different environments including oxygen, water vapour, liquid water and grout. Such comparison is given in Fig. 13 where a variety of uranium corrosion rates reported in the literature [10,15,20], corresponding to different environments are shown. To enhance consistency for appropriate comparison, the corrosion rate units reported in Fig. 11 were converted from gU/m2/day to gU/cm2/min. Samples where uranium was encapsulated in grout and stored in both water vapour and liquid water at 25 °C [15,20], exhibited longterm corrosion rates which lay between the ranges suggested by Haschke et al. [10] for uranium oxidation and uranium corrosion in water vapour regimes. Cumulative corrosion rates calculated for the uranium-CMS sample discussed in this paper are also presented in Fig. 13. During the early stages of uranium corrosion, considerably high rates can be observed lying above the uranium-water vapour corrosion regime. However, the cumulative corrosion rates when the longer interaction periods are considered (0 days - 400 days and 0 days - 540 days) are lower compared to the early stages and lie below the uranium-
Declaration of Competing Interest None. Acknowledgements The presented work is part of the TRANSCEND programme which is funded by UK Research and Innovation. The TRANSCEND consortium comprises key industry partners and leading academic researchers from 11 research intensive universities with significant expertise in nuclear research and development (EPSRC Reference: EP/S01019X/1). The authors would also like to thank Sellafield Ltd (CoE in uranium and reactive metals) for supporting this research. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.corsci.2020.108551. References [1] K.R. Hallam, P.C. Minshall, P.J. Heard, P.E.J. Flewitt, Corrosion of the alloys Magnox AL80, Magnox ZR55 and pure magnesium in air containing water vapour, Corros. Sci. vol. 112, (2016) 347–363. [2] S.F. Jackson, S.D. Monk, Z. Riaz, An investigation towards real time dose rate monitoring, and fuel rod detection in a First Generation Magnox Storage Pond (FGMSP), Appl. Radiat. Isot. vol. 94, (2014) 254–259 Dec. [3] N.C. Collier, N.B. Milestone, The encapsulation of Mg(OH)2 sludge in composite cement, Cem. Concr. Res. vol. 40, (no. 3) (2010) 452–459 Mar. [4] National Decommissioning Authority, “Annual report and accounts 2012-2013”, https://tools.nda.gov.uk/publication/annual-report-and-accounts-2012-2013/ [Accessed: 15-November-2018]. [5] J. Cronin, N. Collier, Corrosion and expansion of grouted Magnox, Mineral. Mag. vol. 76, (no. December) (2012) 2901–2909. [6] C.R. Gregson, D.T. Goddard, M.J. Sarsfield, R.J. Taylor, Combined electron microscopy and vibrational spectroscopy study of corroded Magnox sludge from a
10
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
[7]
[8] [9]
[10] [11] [12] [13] [14]
[15]
[16]
[17]
legacy spent nuclear fuel storage pond, J. Nucl. Mater. vol. 412, (no. 1) (2011) 145–156 May. F. Le Guyadec, X. Génin, J.P. Bayle, O. Dugne, A. Duhart-Barone, C. Ablitzer, Pyrophoric behaviour of uranium hydride and uranium powders, J. Nucl. Mater. vol. 396, (no. 2) (2010) 294–302. T.C. Totemeier, R.G. Pahl, S.L. Hayes, S.M. Frank, Characterization of corroded metallic uranium fuel plates, J. Nucl. Mater. vol. 256, (1998) 87–95. A. Banos, K.R. Hallam, T.B. Scott, Corrosion of uranium in liquid water under contained conditions with a headspace deuterium overpressure. Part 2: the ternary U + H2O(l) + D2 system, Corros. Sci. vol. 152, (2019) 261–270 May. J.M. Haschke, Corrosion of uranium in air and water vapor: consequences for environmental dispersal, J. Alloys. Compd. vol. 278, (no. 1) (1998) 149–160. M.M. Baker, L.N. Less, S. Orman, Uranium+water reaction: Part 1. - Kinetics, products and mechanism, Trans. Faraday Soc. vol. 62, (1966) 2513–2524. J. Bloch, M.H. Mintz, Kinetics and mechanisms of metal hydrides formation—a review, J. Alloys. Compd. vol. 253–254, (1997) 529–541. N.J. Harker, The Corrosion of Uranium in Sealed Environments Containing Oxygen and Water Vapour, PhD Thesis University of Bristol, 2012. C.A. Stitt, C. Paraskevoulakos, N.J. Harker, A. Banos, K.R. Hallam, C.P. Jones, T.B. Scott, Real time, in-situ deuteriding of uranium encapsulated in grout; effects of temperature on the uranium-deuterium reaction, Corros. Sci. vol. 127, (2017) 270–279 Oct.. C.A. Stitt, C. Paraskevoulakos, A. Banos, N.J. Harker, K.R. Hallam, A. Davenport, S. Street, T.B. Scott, In-situ, time resolved monitoring of uranium in BFS:OPC grout. Part 1: Corrosion in water vapour, Sci. Rep. vol. 7, (no. 1) (2017) 7999 Dec. C. Paraskevoulakos, C.A. Stitt, K.R. Hallam, A. Banos, M. Leal Olloqui, C.P. Jones, G. Griffiths, A.M. Adamska, J. Jowsey, T.B. Scott, Monitoring the degradation of nuclear waste packages induced by interior metallic corrosion using synchrotron Xray tomography, Constr. Build. Mater. vol. 215, (August) (2019) 90–103. C.P. Jones, C.M. Brenner, C.A. Stitt, C. Armstrong, D.R. Rusby, S.R. Mirfayzi, L.A. Wilson, A. Alejo, H. Ahmed, R. Allott, N.M.H. Butler, R.J. Clarke, D. Haddock,
[18]
[19]
[20]
[21]
[22] [23] [24]
[25] [26]
11
C. Hernandez-Gomez, A. Higginson, C. Murphy, M. Notely, C. Paraskevoulakos, T.B. Scott, Evaluating laser-driven Bremsstrahlung radiation sources for imaging and analysis of nuclear waste packages, J. Hazard. Mater. vol. 318, (2016) 694–701. C. Paraskevoulakos, K.R. Hallam, T.B. Scott, Grout durability within miniaturised Intermediate Level Waste drums at early stages of interior volume expansion induced by encapsulated metallic corrosion, J. Nucl. Mater. vol. 510, (2018) 348–359 Nov. ILW Treatment and Storage | Sellafield Ltd., http://www.sellafieldsites.com/ solution/waste-management/ilw-treatment-and-storage/ [Accessed: 13-May2018]. C.A. Stitt, C. Paraskevoulakos, A. Banos, N.J. Harker, K.R. Hallam, H. Pullin, A. Davenport, S. Street, T.B. Scott, In-situ, time resolved monitoring of uranium in BFS:OPC grout. Part 2: Corrosion in water, Sci. Rep. vol. 8, (no. 1) (2018) 9282. C.A. Stitt, M. Hart, N.J. Harker, K.R. Hallam, J. MacFarlane, A. Banos, C. Paraskevoulakos, E. Butcher, C. Padovani, T.B. Scott, Nuclear waste viewed in a new light; a synchrotron study of uranium encapsulated in grout, J. Hazard. Mater. vol. 285, (2015) 221–227. User’s Guide - Avizo, [Accessed: 27-November-2018]. J.D. Palmer, G.A. Fairhall, Properties of cement systems containing intermediate level wastes, Cem. Concr. Res. vol.22, (no. 2–3) (1992) 325–330. M. Drakopoulos, T. Connolley, C. Reinhard, R. Atwood, O. Magdysyuk, N. Vo, M. Hart, L. Connor, B. Humphreys, G. Howell, S. Davies, T. Hill, G. Wilkin, U. Pedersen, A. Foster, N. De Maio, M. Basham, F. Yuan, K. Wanelik, I12: the joint engineering, environment and processing (JEEP) beamline at diamond light source, J. Synchrotron Radiat. vol. 22, (no. 3) (2015) 828–838 May. T. Totemeier, Characterization of uranium corrosion products involved in a uranium hydride pyrophoric event, J. Nucl. Mater. vol. 278, (no. 2) (2000) 301–311. C.A. Stitt, N.J. Harker, K.R. Hallam, C. Paraskevoulakos, A. Banos, S. Rennie, J. Jowsey, T.B. Scott, An investigation on the persistence of uranium hydride during storage of simulant nuclear waste packages, PLoS One vol. 10, (no. 7) (2015).
Contents lists available at ScienceDirect
Corrosion Science journal homepage: www.elsevier.com/locate/corsci
Monitoring uranium corrosion in Magnox sludge using X-ray computed tomography: A direct analogue to “legacy” fuel storage ponds C. Paraskevoulakosa,*, K.R. Hallama, A. Adamskab, T.B. Scotta a b
Interface Analysis Centre, School of Physics, University of Bristol, Bristol, BS8 1TL, UK Vertex, GF, Sir Christopher Harding House, North Shore Road, Whitehaven, CA28 7XY, UK
A R T I C LE I N FO
A B S T R A C T
Keywords: Interfaces (C) Pitting corrosion (C) Alloy (A)
During the early years of nuclear industry within the UK, the typical route followed to handle Magnox (Mg-Al alloy) casing after separation from the nuclear fuel has been immersion in water temporarily in specificallydesigned water ponds. Over the decades of storage, Magnox has been corroding in water, resulting in the formation of sludge within the ponds; known as Corroded Magnox Sludge. Uranium, which has been also contained within the ponds as a result of faults occurring during fuel-casing separation, is susceptible to corrosion in waterrich environments. Under certain circumstances, uranium corrosion products can be highly pyrophoric during oxidation in air. This raises serious concerns regarding the safety margins of nuclear pond decommissioning. The present study attempts to shed light on the behaviour of uranium in a Magnox sludge-surrogate environment, using X-ray computed tomography at distinct time intervals over an 18-month period. Focus was given onto the uranium-sludge interface to determine metallic corrosion onset and evolution.
1. Introduction
open air, implying that flammable hydrogen resulting from uranium and Magnox corrosion can be released into the gas phase above the pond. In addition, UH3 can be highly exothermic when oxidised in air (under certain circumstances related to its mass and surface area amongst others) [7, 8]. It is, therefore, of significant importance to determine the corrosion behaviour of uranium metal within a Magnox sludge environment. Corrosion rate monitoring will allow estimation of the extent of corrosion within the storage ponds, while potential formation and, particularly, persistence of UH3 within the ponds would raise concerns about fuel and skip’ lifts out of the pond during decommissioning.
Magnox nuclear reactors, using metallic uranium as fuel operated in the UK from the 1960′s until just a few years ago. The metallic fuel was encased in a magnesium alloy ‘Magnox’ cladding, which, along with the uranium metal, is susceptible to corrosion by water [1]. Until 1992, a significant volume of Magnox waste materials, including spent fuel assemblies, had been accumulated in the storage ponds at Sellafield in Cumbria [2]. These ponds were maintained at a high pH (above 10.5) to minimise corrosion [3]. However, considerable corrosion has still occurred over extended periods, raising concerns about the safe decommissioning of the facilities [4]. Corroded Magnox Sludge (CMS), arising from long-term corrosion of the cladding material, is a prevalent residual material amongst others (e.g. fuel fragments, wind-blown debris and concrete degradation products) [2]. Corrosion of Magnox forms brucite (Mg(OH)2) and liberates hydrogen [5]. Embedded uranium metal fuel is also expected to have been exposed to water throughout the storage period. This is validated by the presence of leached fission products (cesium and strontium) and traces of actinides, including uranium and plutonium, within these sludge beds [6]. Uranium reacts with water to form uranium hydride (UH3), uranium dioxide (UO2) and hydrogen, presenting potential uncontrolled thermal hazards during decommissioning. Most of the legacy ponds are
2. Background Uranium corrosion behaviour has been extensively studied over the last decades, predominantly in simplified environments including air, liquid water, hydrogen and water vapour [9–13]. Only recently, considerable research has been conducted to investigate uranium corrosion in more complicated and interacting environments such as grout, attempting to shed some light on the performance of intermediate level waste (ILW) drums [14–20]. Stitt et al. [21] demonstrated the feasibility of using X-rays to investigate uranium-grout systems over a prolonged time period. Thus, the fact that uranium corrosion could be
⁎
Corresponding author. E-mail addresses: [email protected] (C. Paraskevoulakos), [email protected] (K.R. Hallam), anna.m.adamska@sellafieldsites.com (A. Adamska), [email protected] (T.B. Scott). https://doi.org/10.1016/j.corsci.2020.108551 Received 22 October 2019; Received in revised form 23 January 2020; Accepted 17 February 2020 0010-938X/ © 2020 Elsevier Ltd. All rights reserved.
Please cite this article as: C. Paraskevoulakos, et al., Corrosion Science, https://doi.org/10.1016/j.corsci.2020.108551
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
encapsulate the uranium specimen. The mixture was allowed to cure in ambient conditions for three days before being hermetically sealed, using a second flange connected with the stainless-steel pot via a rubber gasket. No deliberate exchange of the sample inner atmosphere was allowed during the entire experimental period. The research plan focused on performing XCT scans (Zeiss Xradia 520 Versa) at distinct time intervals over a total period of approximately 18 months to determine the behaviour of the uranium within the CMS environment. A specifically-designed metal holder was used to host the sample within the XCT chamber, ensuring that the sample remained steady as the XCT stage rotated during scanning. Schematic illustration of the experimental set up is shown in Fig. 1.
monitored at distinct time intervals in a non-destructive manner was a considerable breakthrough compared to the traditional post-mortem analysis, where samples could be only investigated at a single time step after disrupting the system. So far, experimental studies on uranium corrosion in Magnox sludge have not been published to the best of our knowledge. Examination of CMS samples extracted directly from a legacy spent nuclear fuel storage pond located at Sellafield using scanning electron microscopy (SEM), energy dispersive X-ray (EDX) analysis and Raman spectroscopy revealed uranium traces, validating the dissolution of fuel parts into the CMS [6]. This paper presents the results of an initial feasibility timeresolved experimental project, where a uranium specimen is corroding in ambient conditions in CMS. X-ray computed tomography (XCT) has been used at distinct time intervals to primarily observe the extent of corrosion and the behaviour of the corrosion products forming.
4. XCT scanning parameters XCT scans were performed at four different time periods: 1) 20 days, 2) 50 days, 3) 400 days, and 4) 540 days after preparation. Scans were predominantly focused at the uranium-CMS interface to identify potential corrosion traces. A repeated process of fine adjustment was attempted with regards to scanning parameters to target compatibility between different scans, allowing direct comparison. Both low- and high-magnification scans were conducted, while the entire uranium specimen was probed. Since uranium is significantly dense (19.1 g/ cm3), a 140 kV beam was used to achieve adequate transmission levels. The resolution of the low-magnification scans, which enabled capturing a large sample volume, ranged between 30 μm/pixel and 35 μm/pixel. Selective high-magnification scans were also performed to investigate smaller volumes in greater detail. The resolution for these long duration scans (∼30 h) was between 2.8 μm/pixel and 2.9 μm/pixel.
3. Materials, sample preparation An experimental set-up was built to act as a simplified model of a legacy pond, where uranium fuel parts are located and corroding over the storage period. The goal was to monitor the behaviour of uranium when encapsulated in CMS over a considerable time period using XCT. A stainless-steel pot (25 mm diameter) with a flange on top, which could offer a sealed containment for the interacting materials, was used to host the uranium and the CMS. A Teflon® cup was placed around the inner surface to separate the steel containment from the CMS. A single uranium specimen (1.5 mm x 0.9 mm x 17.5 mm) was prepared after machining down from a mechanically-polished natural uranium disc. EDX was performed to determine the elemental composition. The results are presented in Table 1. The high content of impurities (C, N, O) is typical for this specific material, which originates from an unirradiated fuel rod used for Magnox reactors. Carbon and nitrogen peaks are mainly associated with the presence of carbide and carbo-nitride inclusion particles, while carbon deposition is also expected on the sample surface due to the electron beam (SEM). The oxygen peaks are related to the oxide layer formed vigorously on the surface of the uranium specimen before placed in the SEM chamber. Within only a few minutes of exposure in air, fresh uranium surfaces can tarnish, indicative of thin oxide layer formation. The uranium specimen was then immersed into the CMS, which was previously poured inside the cup. CMS powder, originally provided by industrial partners, was used as the core material to prepare the sludge. The powder was manufactured by corrosion of Magnox alloy type AL80 in water at 50 °C. The material was then sieved to remove metal and centrifuged to remove water. After sieving there was typically less than 1% w/w Magnox alloy remaining in the CMS powder. The final solids content was then measured and adjusted by oven drying. The solid, powderised material was mixed with distilled water (118 % w/s) to prepare the sludge, which was subsequently used to
5. XCT results 5.1. 20 days after preparation The first XCT scan was performed 20 days after preparation. It was expected that no signs of uranium corrosion would be observed at such a primary sample age and, therefore, this scan would act as a reference for subsequent scans. Post scanning, the data were reconstructed and processed using the X3DM viewer and Avizo software [22]. Characteristic 2D projections of three different planes, showing the interacting materials can be observed in Fig. 2. Fig. 3 also presents a single representative 3D model of the entire sample, where uranium can be observed in the central area. Detailed observation of the 2D projections and the reconstructed 3D model revealed the presence of features at the uranium-CMS interface. These features were only observed near the top of the uranium specimen. Fig. 4 illustrates two separate 3D views of the sample under different rotation angles with a couple of magnified screenshots attached next to each one, representing the top and the middle region of the uranium. It can be clearly observed that the top part of the uranium exhibits traces of morphology change (yellow arrows), while in the middle region uranium seems intact. The colour of these blisters appearing across the uranium surface in the greyscale is between the uranium and the background Magnox. Since tomographic images are practically density maps, the blisters’ densities lie between those of uranium (19.1 g/cm3) and CMS (1.5 g/cm3 – 2.5 g/cm3). Selective highmagnification scans were also performed, focusing on the top part of the uranium where features were observed. Reconstructed data validated their presence at the uranium-CMS interface.
Table 1 Elemental composition of natural uranium used for preparation of the ILW drum – simulant packages as determined via energy dispersive Xray analysis, (20 kV acceleration voltage). Natural uranium Element
Wt %
C
7.78 ± 0.67 4.51 ± 0.27 5.08 ± 1.52 0.23 ± 0.05 0.11 ± 0.04 82.50 ± 1.87
N O Al Si U
5.2. 50 days after preparation XCT scans were repeated with the sample aged 50 days. The goal was to confirm that the features were still present and, if so, monitor their size. Potential growth would indicate that uranium continues to react in the sludge environment. Data reconstruction and subsequent 2
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 1. Schematic illustration of the experimental set up used to investigate the behaviour of uranium in a CMS environment. Dimensions are given in mm.
initiated. The third set of analyses would enable the drawing of more conclusions about the system’s long-term behaviour with regards to corrosion evolution. The configuration of the scan was adjusted so that the data of the third set were directly comparable to the first two analysis cycles. The system’s status more than a year after sample preparation can be evaluated based on Fig. 6. Areas near the top of the uranium are significantly different than the corresponding areas at 20 days and 50 days. The reaction front seems to have propagated both in the core metal and across its length. Feature volumes have increased considerably compared to the early stages of sample monitoring. It is also interesting to note that in the top region coalescence of the blisters has occurred across the surface, forming a consistent layer of varying thickness. The first two datasets exhibited a slightly different behaviour since isolated pits/blisters were observed across the uranium surface instead of a coherent layer. Migrated particles within the sludge volume are also apparent (Fig. 6, blue arrows) in agreement with observations drawn from the second set of analyses (50 days).
3D visualisation proved that the feature volume had increased. Several pits were observed in the uranium surface, not only at the top part of the metal but also in the middle height region. Fig. 5 illustrates representative 3D views and some allocated magnified screenshots confirming the presence of features at the uranium-CMS interface. It is also obvious that, in contrast to data collected 20 days after preparation, areas on top of the uranium specimen within the sludge volume are occupied by particles resembling those formed on the uranium surface. This behaviour probably suggests a migration mechanism of the particles from the uranium surface towards the sludge volume over this period of 30 days. These areas are highlighted in Fig. 5 using blue arrows. 5.3. 400 days after preparation The third set of measurements was performed over a year after sample preparation (400 days). The first two datasets proved that corrosion of uranium within the sludge environment had probably
Fig. 2. 2D projections associated with three different plane orientations, probing the interacting materials of the sample. U, S and CMS stand for uranium, steel and corroded Magnox sludge respectively. 3
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 3. Reconstructed 3D model observed under two different rotation angles.
5.4. 540 days after preparation
no visible change.
The fourth and final set of measurements was performed almost 1.5 years after the sample was prepared (540 days). After demonstrating corrosion propagation over the first year of sample aging, this scan was performed to reveal if there was any further growth of the blister’ volumes or if the reaction had slowed down. The data collection process followed was compatible with the previous sets so that the results are comparable. Fig. 7 illustrates selective screenshots of the data collected for this last, fourth scan. It can be observed that 540 days after the experiment initiated the geometry of features at the uranium-CMS interface is similar to the corresponding profile 400-day profile. Comparing the same areas in both datasets reveals no difference while the bottom area of the uranium rod, which was intact during the first year of scanning showed
5.5. Comparing the datasets A direct comparison between the four data sets corresponding to different sample ages follows. Image segmentation was performed to isolate uranium metal from the rest of the materials (including the blisters at the uranium-CMS interface), allowing investigation of the metal surface. The results are grouped together and presented in Fig. 8. Craters around the uranium surface formed within 20 days of preparation. At later stages, it is evident that additional craters have formed while those which were observed at the early stages have grown both inwards and across the metal’s surface. It is interesting to note that the craters formed and developed only on the top half of the uranium specimen. The bottom half seems intact even at the latest stage of
Fig. 4. Sample views under two different rotation angles, with attached magnified screenshots representing the top and the middle uranium regions. Yellow arrows demonstrate the presence of features at the uranium-CMS interface. Views correspond to 20 days after sample preparation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 4
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 5. Sample views under two different rotation angles, with attached magnified screenshots representing the top and the middle uranium regions. Yellow arrows demonstrate the presence of features at the uranium-CMS interface, while blue arrows point at migrated particles within the CMS volume. Views correspond to data obtained 50 days after sample preparation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
part of those features within the sludge during the 30-day period between the two scans is confirmed. This behaviour was previously (Sections 5.1-5.3) commented upon based on the low-magnification scan dataset. However, Fig. 9 shows in greater detail that the features morphology has changed between the two first scans. Based on the data acquired it can be safely stated that the bulges observed to form and grow at the uranium-CMS interface are related to the rapid corrosion of the uranium metal in the sludge.
investigation (540 days after preparation). This observation was consistent over the entire dataset volume. No noticeable changes occurred between the last two scans (400 days to 540 days). In addition to these low-magnification scans which allowed probing the entire sample volume, high-magnification scans were also performed to investigate the morphology of the interfaces in greater detail. Only the top part of the uranium specimen was investigated. The same process was followed with regards to scan configuration and data processing. High-magnification scans were performed only at the two early monitoring stages (20 days and 50 days post sample preparation), where low-magnification scans exhibited slight differences even though progress of the reaction front could be undeniably stated to have occurred. Fig. 9 illustrates results from the high-magnification scans. Characteristic 2D projections under three different planes, both at 20 days and 50 days after sample preparation, can be observed. Volumetric increase of the features that appeared at the uranium-CMS interface after the first scan (20 days) is evident. Also, migration of a
6. Quantitative analysis Data retrieved from the low-magnification scans were used to perform quantitative analysis. Avizo [22] software was used for material segmentation and, predominantly, volume calculations. Segmenting uranium metal from the corrosion products (assumed to be UH3) was feasible even if the scan resolution was low (30 μm/pixel - 35 μm/pixel) considering the size of the uranium specimen. A progressive increase of
Fig. 6. Sample views under two different rotation angles, with attached magnified screenshots representing the top and the middle uranium regions. Yellow arrows demonstrate the presence of features at the uranium-CMS interface, while blue arrows point at migrated particles within the CMS volume. Views correspond to data obtained 400 days after sample preparation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 5
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 7. Sample views under two different rotation angles, with attached magnified screenshots representing the top and the middle uranium regions. Yellow arrows demonstrate the presence of features at the uranium-CMS interface. Views correspond to data obtained 540 days after sample preparation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 8. Segmented uranium volume sets at different time periods under three different rotation angles. Each set contains four uranium volume profiles corresponding to growing sample ages starting from the left side (20 days) to the right side (540 days). 6
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 9. Characteristic 2D projections associated with the high resolution XCT scans under different 2D planes. Images presented correspond to data collected 20 days and 50 days after sample preparation. Progress of the reaction front can be observed.
showing the best correlation results among alternative trends. Even though just four data points limits the ability to draw reliable conclusions, it seems that there is a rapid growth of the corrosion product volume over the first 100 days, which subsequently slows down. Based on the fitting formula provided to correlate the corrosion product volume evolution with time, it seems that blister-type corrosion initiated about 17 days after sample preparation. However, it is recognised that
the corrosion product volume was determined, associated with reduction in the uranium volume. Based on the calculated volumes of the corrosion products and the uranium at the three different time periods, the evolution of the corrosion percentage was tracked. The evolution of the corrosion percentage and the corrosion product volume over time are given in Fig. 10. Logarithmic fitting was applied to the data points in both cases, 7
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 10. Evolution of the corrosion percentage (left) and the corrosion product volume (right) with time.
intervals, revealed the formation, development and volumetric growth of features at the uranium-CMS interface. XCT findings indicate the onset and progress of uranium corrosion within the sludge. Uranium metal reacted with water present in the sludge to liberate hydrogen via the following chemical formula.
the number of available data points is not adequate for reliable use of the curve fitting formula. Based on the volume of material loss, corrosion rate was also calculated to determine quantitatively the drop in the growth rate of the corrosion products. Non-corroded uranium metal volume was measured at each different time interval based on the Avizo material segmentation process. Consequently, the mass of uranium reacted between two consecutive XCT scans could be calculated. The initial surface area of the uranium specimen was also known, since the specimen was thoroughly measured before it was encapsulated in the CMS. As a result, the following formula was implemented to determine the corrosion rate at the specific timings when XCT scans were conducted:
Corrosion rate (gU /m2 /day ) =
U mass loss (g ) ISA (m2) × Time Period (days )
U + 2H2 O → UO2 + 2H2
(2)
It has been recognised that, under certain circumstances, uranium can also react with hydrogen to form uranium hydride (Eq. 3).
U+
3 H → UH3 2 2
(3)
XCT scans demonstrated the presence of corrosion products at the uranium-CMS interface, resulting from uranium corrosion in water. However, the dataset obtained cannot be used to characterise the corrosion products and determine whether UH3 has formed and persisted. Potential dissection of the system and subsequent surface analysis would not provide any reliable results since UH3 oxidises very rapidly in air. Based on the literature available, it is possible to predict the identity of the corrosion products formed based on their morphology. Stitt et al. [14,21], performed synchrotron X-ray tomography experiments on systems where a single uranium specimen was encapsulated in grout, mimicking the ILW drums in Sellafield storage facilities [19, 23]. The results showed that a coherent layer of relatively constant thickness had formed at the uranium-grout interface of one of the samples, which allowed to cure in ambient conditions. In contrast, a second sample, which was exposed to hydrogen to deliberately form UH3 on the uranium surface exhibited a completely different morphology. Blisters were observed on the uranium surface at different locations, instead of a smooth uniform layer of constant thickness. Synchrotron X-ray powder diffraction (XRPD) analysis was also conducted to determine the phases present in both samples. The results revealed a UO2 layer of uniform thickness across the first sample, while UH3 blisters were detected on the second one. Same findings have been also reported on similar uranium-grout systems [16]. A selection of characteristic images available in the literature, showing the different morphologies of corrosion products (UO2 and UH3) in uranium-grout systems, is given in Fig. 12. One representative image related to the uranium-CMS system presented in the previous sections is also included to allow comparison. Figs. 12a and 12b illustrate 3D models of isolated uranium specimens where the corrosion products are attached and presented using different colours. Fig. 12a corresponds to a uranium-grout system which evolved in ambient conditions and, therefore, only a coherent UO2 layer had formed, whereas, Fig. 12b illustrates the state of a uranium-grout system subjected to hydrogen to deliberately form UH3. Craters appear on the isolated uranium surface where blister-type corrosion products (typical of UH3 presence) are also observed. Figs. 12a and 12b are reproduced
(1)
where ISA stands for the initial surface area of the uranium specimen and (since the specific day of corrosion onset is unknown), time period refers to the sample age when the XCT scans were acquired. The results are shown in Fig. 11. Since material segmentation is a visual process, several intensity threshold values were tested in the form of a sensitivity analysis. The calculated corrosion rate 20 days after sample preparation was considerably high. However, after the end of the second scan (50 days from sample preparation) it had already dropped to almost half. The corrosion rate at the last two sets of measurement (400 days and 540 days) was found to be low and almost constant. 7. Discussion Both low- and high-resolution scans were performed to investigate the behaviour of uranium within a CMS environment, over a period of approximately 18 months. All scans performed at distinct time
Fig. 11. Evolution of the corrosion rate over time. 8
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
Fig. 12. Screenshots reproduced from literature [14,16,21], illustrating corrosion products formed at uranium-grout interfaces are compared with the behaviour of a uranium-CMS interface investigated in this study.
possibly contributed to maintain this elevated hydrogen pressure. Post-formation persistence of UH3 is often questioned and cannot be validated or safely predicted based on the morphology of the corrosion products. Oxidation of UH3 (Eq. 4), which in air can raise serious concerns since it is highly exothermic, is one possible scenario to have occurred within the CMS environment.
and adapted from [21]. Fig. 12c is a radiography shot displaying the uranium-grout interface of a sample subjected to hydrogen online on the I12 beamline at Diamond Light Source [24]. The UO2 layer formed during sample curing has been detached from the uranium metal, while UH3 blisters (confirmed via XRPD analysis) can be observed between the uranium metal and the detached oxide layer. This image was reproduced and adapted from [14]. Fig. 12d (reproduced and adapted from [16]) shows a similar situation, where uranium encapsulated in grout and stored within a stainless steel mini drum was exposed to hydrogen. The formation of blisters across the uranium surface at the early reaction stages, associated with UH3 presence, is also evident. In compliance with the UH3 blister-type morphology shown in all previous examples in uranium-grout systems, which differs significantly from the coherent UO2 layer of relatively uniform thickness, it can be assumed with considerable safety that the corrosion products observed to grow in the uranium-CMS system resemble the UH3 morphology. In case this assumption is valid, it is interesting to note that in contrast with all the corresponding cases mentioned in uranium-grout hydrided systems, the uranium-CMS system was not exposed to hydrogen. This means that there was no deliberate attempt to form UH3. Instead, it formed and developed naturally in ambient conditions. Natural formation of UH3 in a uranium-bearing system is a rare scenario based on the available literature [25]. A combination of elevated hydrogen pressure and high temperature is usually required to activate a uranium-hydrogen reaction. Therefore, reaction initiation and progress over a one-year period in ambient conditions is an interesting finding and worthy of further study. With uranium encapsulation in CMS, hydrogen is expected to have been liberated as part of a uranium-water reaction. It is also possible that hydrogen evolution could have occurred from magnesium reaction with water, in the case of any pure metal swarf being present in the CMS powder mixture. It seems that the generated pressure was sufficient to allow hydrogen to permeate through the sludge, reach the metal surface and favour a uranium-hydrogen reaction, which consequently led to the formation and growth of UH3 particles. The hermetically-sealed containment has
UH3 +
7 3 O → UO2 + H2 O 2 2 2
(4)
Stitt et al. [26] performed synchrotron experiments, specifically targeted on the persistence of UH3, using uranium-grout systems previously hydrided artificially and subsequently embedded in water. The XRPD analysis revealed UH3 peaks even 10 months after sample preparation, implying that UH3 had only partially or not at all oxidised during this period. Decommissioning of nuclear waste facilities, including the legacy ponds, has only recently started and will proceed over the following decades. XCT time-resolved analysis showed that uranium corrodes rapidly while encapsulated in CMS. Based on the morphology of the corrosion products probed, UH3 is highly possible to form, and potentially, persist over time. It is important to note that UH3 formation in a uranium-sludge system in ambient conditions has not been previously demonstrated, to the best of our knowledge. This also applies for other uranium-containing systems, including uranium-grout packs. The sealed containment of the sample investigated for this study has probably favoured corrosion initiation and progress, since hydrogen generated from uranium-water (and, potentially, magnesium-water) reaction was probably confined within the sample’s core. If the results drawn from this study were applicable to the real case of legacy ponds, it would be safe to claim that a considerable volume of the uranium immersed into the legacy ponds decades ago has been significantly corroded. It is also highly possible that UH3 has formed as part of uranium-water corrosion, but it is questionable if this has persisted over the storage period. Lifting out solid components or sludge from the ponds is essential for decommissioning. The presence of UH3, either on the solid matrix or in the sludge mixture could reasonably be identified 9
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
water vapour corrosion regime. The corrosion rates derived at different time intervals during this experimental study are plotted against time in Fig. 14. Relative data found in the literature are also included, as in Fig. 13. It seems that uranium is significantly more susceptible to corrosion within a CMS environment than in grout. The comparison between the two systems (uranium-CMS and uranium-grout) is valid since most of the experimental parameters, including uranium elemental composition, geometry and temperature are the same. 8. Conclusions The primary results of a feasibility study using XCT to investigate the corrosion of metallic uranium in a CMS environment are presented in this paper. Scans were performed at distinct time intervals for a period of more than a year. The results revealed that uranium corrosion had already initiated within the first 20 days after sample preparation and progressed rapidly during the following months. Based on the morphology of the corrosion products, it is believed that UH3 has formed in ambient conditions and potentially persisted.
Fig. 13. Long-term uranium corrosion rates in various environments as found in the literature compared with the corrosion rates determined for different time periods for the uranium-CMS sample investigated for the present study.
Data Availability The raw/processed data required to reproduce these findings cannot be shared at this time as the data also forms part of an ongoing study. CRediT authorship contribution statement C. Paraskevoulakos: Conceptualization, Methodology, Software, Formal analysis, Investigation, Writing - original draft, Writing - review & editing. K.R. Hallam: Validation, Writing - review & editing. A. Adamska: Resources, Conceptualization. T.B. Scott: Conceptualization, Project administration, Supervision.
Fig. 14. Time-resolved evolution of uranium corrosion rates in various environments as found in the literature compared with the corrosion rates determined for the uranium-CMS system presented in this study.
as a hazard, since it can be highly exothermic when oxidised. The XCT scans enabled precise calculation of the uranium corrosion product volume at different time intervals, as well as of the volume of the remainder uranium metal as corrosion progressed. The accumulated corrosion rates at four distinct moments when scans were performed (20 days; 50 days; 400 days; 540 days) were also determined based on the relative volume calculations, as shown in Fig. 11. Comparison of the datasets corresponding to the last two scans (at 400 days and 540 days) revealed almost negligible change. It is highly probable that the availability of hydrogen produced from Magnox corrosion dropped significantly and that, therefore, a uranium-hydrogen reaction could not take place. However, even if hydrogen was still abundant within the cell, potential diffusion though the corrosion layer to reach the underlying metal surface and, subsequently, activate reaction would be questionable. The results were used to compare with relevant available literature data where uranium corroded in different environments including oxygen, water vapour, liquid water and grout. Such comparison is given in Fig. 13 where a variety of uranium corrosion rates reported in the literature [10,15,20], corresponding to different environments are shown. To enhance consistency for appropriate comparison, the corrosion rate units reported in Fig. 11 were converted from gU/m2/day to gU/cm2/min. Samples where uranium was encapsulated in grout and stored in both water vapour and liquid water at 25 °C [15,20], exhibited longterm corrosion rates which lay between the ranges suggested by Haschke et al. [10] for uranium oxidation and uranium corrosion in water vapour regimes. Cumulative corrosion rates calculated for the uranium-CMS sample discussed in this paper are also presented in Fig. 13. During the early stages of uranium corrosion, considerably high rates can be observed lying above the uranium-water vapour corrosion regime. However, the cumulative corrosion rates when the longer interaction periods are considered (0 days - 400 days and 0 days - 540 days) are lower compared to the early stages and lie below the uranium-
Declaration of Competing Interest None. Acknowledgements The presented work is part of the TRANSCEND programme which is funded by UK Research and Innovation. The TRANSCEND consortium comprises key industry partners and leading academic researchers from 11 research intensive universities with significant expertise in nuclear research and development (EPSRC Reference: EP/S01019X/1). The authors would also like to thank Sellafield Ltd (CoE in uranium and reactive metals) for supporting this research. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.corsci.2020.108551. References [1] K.R. Hallam, P.C. Minshall, P.J. Heard, P.E.J. Flewitt, Corrosion of the alloys Magnox AL80, Magnox ZR55 and pure magnesium in air containing water vapour, Corros. Sci. vol. 112, (2016) 347–363. [2] S.F. Jackson, S.D. Monk, Z. Riaz, An investigation towards real time dose rate monitoring, and fuel rod detection in a First Generation Magnox Storage Pond (FGMSP), Appl. Radiat. Isot. vol. 94, (2014) 254–259 Dec. [3] N.C. Collier, N.B. Milestone, The encapsulation of Mg(OH)2 sludge in composite cement, Cem. Concr. Res. vol. 40, (no. 3) (2010) 452–459 Mar. [4] National Decommissioning Authority, “Annual report and accounts 2012-2013”, https://tools.nda.gov.uk/publication/annual-report-and-accounts-2012-2013/ [Accessed: 15-November-2018]. [5] J. Cronin, N. Collier, Corrosion and expansion of grouted Magnox, Mineral. Mag. vol. 76, (no. December) (2012) 2901–2909. [6] C.R. Gregson, D.T. Goddard, M.J. Sarsfield, R.J. Taylor, Combined electron microscopy and vibrational spectroscopy study of corroded Magnox sludge from a
10
Corrosion Science xxx (xxxx) xxxx
C. Paraskevoulakos, et al.
[7]
[8] [9]
[10] [11] [12] [13] [14]
[15]
[16]
[17]
legacy spent nuclear fuel storage pond, J. Nucl. Mater. vol. 412, (no. 1) (2011) 145–156 May. F. Le Guyadec, X. Génin, J.P. Bayle, O. Dugne, A. Duhart-Barone, C. Ablitzer, Pyrophoric behaviour of uranium hydride and uranium powders, J. Nucl. Mater. vol. 396, (no. 2) (2010) 294–302. T.C. Totemeier, R.G. Pahl, S.L. Hayes, S.M. Frank, Characterization of corroded metallic uranium fuel plates, J. Nucl. Mater. vol. 256, (1998) 87–95. A. Banos, K.R. Hallam, T.B. Scott, Corrosion of uranium in liquid water under contained conditions with a headspace deuterium overpressure. Part 2: the ternary U + H2O(l) + D2 system, Corros. Sci. vol. 152, (2019) 261–270 May. J.M. Haschke, Corrosion of uranium in air and water vapor: consequences for environmental dispersal, J. Alloys. Compd. vol. 278, (no. 1) (1998) 149–160. M.M. Baker, L.N. Less, S. Orman, Uranium+water reaction: Part 1. - Kinetics, products and mechanism, Trans. Faraday Soc. vol. 62, (1966) 2513–2524. J. Bloch, M.H. Mintz, Kinetics and mechanisms of metal hydrides formation—a review, J. Alloys. Compd. vol. 253–254, (1997) 529–541. N.J. Harker, The Corrosion of Uranium in Sealed Environments Containing Oxygen and Water Vapour, PhD Thesis University of Bristol, 2012. C.A. Stitt, C. Paraskevoulakos, N.J. Harker, A. Banos, K.R. Hallam, C.P. Jones, T.B. Scott, Real time, in-situ deuteriding of uranium encapsulated in grout; effects of temperature on the uranium-deuterium reaction, Corros. Sci. vol. 127, (2017) 270–279 Oct.. C.A. Stitt, C. Paraskevoulakos, A. Banos, N.J. Harker, K.R. Hallam, A. Davenport, S. Street, T.B. Scott, In-situ, time resolved monitoring of uranium in BFS:OPC grout. Part 1: Corrosion in water vapour, Sci. Rep. vol. 7, (no. 1) (2017) 7999 Dec. C. Paraskevoulakos, C.A. Stitt, K.R. Hallam, A. Banos, M. Leal Olloqui, C.P. Jones, G. Griffiths, A.M. Adamska, J. Jowsey, T.B. Scott, Monitoring the degradation of nuclear waste packages induced by interior metallic corrosion using synchrotron Xray tomography, Constr. Build. Mater. vol. 215, (August) (2019) 90–103. C.P. Jones, C.M. Brenner, C.A. Stitt, C. Armstrong, D.R. Rusby, S.R. Mirfayzi, L.A. Wilson, A. Alejo, H. Ahmed, R. Allott, N.M.H. Butler, R.J. Clarke, D. Haddock,
[18]
[19]
[20]
[21]
[22] [23] [24]
[25] [26]
11
C. Hernandez-Gomez, A. Higginson, C. Murphy, M. Notely, C. Paraskevoulakos, T.B. Scott, Evaluating laser-driven Bremsstrahlung radiation sources for imaging and analysis of nuclear waste packages, J. Hazard. Mater. vol. 318, (2016) 694–701. C. Paraskevoulakos, K.R. Hallam, T.B. Scott, Grout durability within miniaturised Intermediate Level Waste drums at early stages of interior volume expansion induced by encapsulated metallic corrosion, J. Nucl. Mater. vol. 510, (2018) 348–359 Nov. ILW Treatment and Storage | Sellafield Ltd., http://www.sellafieldsites.com/ solution/waste-management/ilw-treatment-and-storage/ [Accessed: 13-May2018]. C.A. Stitt, C. Paraskevoulakos, A. Banos, N.J. Harker, K.R. Hallam, H. Pullin, A. Davenport, S. Street, T.B. Scott, In-situ, time resolved monitoring of uranium in BFS:OPC grout. Part 2: Corrosion in water, Sci. Rep. vol. 8, (no. 1) (2018) 9282. C.A. Stitt, M. Hart, N.J. Harker, K.R. Hallam, J. MacFarlane, A. Banos, C. Paraskevoulakos, E. Butcher, C. Padovani, T.B. Scott, Nuclear waste viewed in a new light; a synchrotron study of uranium encapsulated in grout, J. Hazard. Mater. vol. 285, (2015) 221–227. User’s Guide - Avizo, [Accessed: 27-November-2018]. J.D. Palmer, G.A. Fairhall, Properties of cement systems containing intermediate level wastes, Cem. Concr. Res. vol.22, (no. 2–3) (1992) 325–330. M. Drakopoulos, T. Connolley, C. Reinhard, R. Atwood, O. Magdysyuk, N. Vo, M. Hart, L. Connor, B. Humphreys, G. Howell, S. Davies, T. Hill, G. Wilkin, U. Pedersen, A. Foster, N. De Maio, M. Basham, F. Yuan, K. Wanelik, I12: the joint engineering, environment and processing (JEEP) beamline at diamond light source, J. Synchrotron Radiat. vol. 22, (no. 3) (2015) 828–838 May. T. Totemeier, Characterization of uranium corrosion products involved in a uranium hydride pyrophoric event, J. Nucl. Mater. vol. 278, (no. 2) (2000) 301–311. C.A. Stitt, N.J. Harker, K.R. Hallam, C. Paraskevoulakos, A. Banos, S. Rennie, J. Jowsey, T.B. Scott, An investigation on the persistence of uranium hydride during storage of simulant nuclear waste packages, PLoS One vol. 10, (no. 7) (2015).