Non-radioactive, representative metal samples with oxide-layer were used in the testing of NP CORD technology. The composition of the metal samples is presented in Table 1. The samples were 1.5 cm in length, 1.2 cm wide and 0.2 cm thick. The samples were oxidised using the setup shown in Figure 1. The setup used in this work was adapted from D. Horvath (Horvath, 2017) and R. Katona (Katona et al., 2022). The metal samples were inductively heated up to high temperature (600–800°C) and vapours of distilled water/boric acid solutions were passed over it for 8 h. The concentrations of boric acid (ACS reagent, ≥ 99.5%, Merck) solutions were 1 g/L and 15 g/L.

The oxide-layers were characterised by Scanning Electron Microscope with Energy Dispersive Spectroscopy (SEM-EDS) and X-ray diffraction (XRD).

D8 Advance, Bruker XRD and Zeiss Merlin SEM devices were used to characterise the formed oxide-layer. XRD measurements were recorded under the following conditions: 10°–100°2theta, a step size of 0.01°, and a scan speed of 4.5 sec/step. SEM analyses of samples were performed by placing them in a phenolic hot mounting resin with carbon filler (Polyfast, Struers). Samples were progressively polished to a 1 μm finish using diamond paste (ADS polycrystaline 1 μm diamond solution, Presi). For imaging, a EHT of 20 kV was used at 400 pA. For EDX analysis, a working distance of 8 mm was used and analysis was carried out at every 1 μm corresponding to the device resolution.

The efficiency of the chemicals were determined by the measurement of the amount of dissolved metals using Inductively coupled plasma mass spectrometry (X-Series II, Thermo Electron) and electrochemical tests.

Tafel-analyses were used to measure the rates of CORD decontamination at various concentrations. Voltalab PGZ201 potentiostat, Radelkis OP-0830P saturated calomel reference electrode (SCE) and stainless steel as a counter electrode were used in these tests. The working electrodes were SS 316Ti stainless steel discs with a diameter of 1.5 cm, which were ground by emery papers down to 180 grit. The polarisation range was ±200 mV with 1 mV/s sweep rate. The equation of the polarisation curve (Butler–Volmer equation) is as follows i = i 0 e β z a k R T Δ E − e − α z k F R T Δ E (1)

In Eq. 1 i is the current density (μA/cm²), i₀ is the equilibrium current density (μA/cm²), α/β are the factor of oxidation/reduction, z_a/z_k are the number of the change of the charge numbers and ΔE is the polarisation (V) (Li et al., 2018). The potentials and the equilibrium current densities were determined by Tafel extrapolation. Before the tests the open circuit potential (OCP) was measured for 5 min.

The XRD characterisation results are presented in Figure 2. The XRD profile of the surrogate samples showed the presence of various oxides of Ni, Fe and Cr such as Fe₃O₄, Fe₂O₃, NiO, Cr₂FeO₄ as seen in Figure 2A. The presence of boric acid appeared to have a small effect on the intensity of the peaks of the various oxides. From Figure 2B, the decrease in intensity of the oxide peaks is observed and the high intensity peaks are arising from base metal, showing effectiveness of the process in removal of the oxide layers.

The SEM images and EDX profiles for a representative sample (sample 3 in Table 2) are seen in Figure 3. Figure 3A shows a magnified image of a treated sample showing both the oxide layer and exposed bare metal. The red line splits the sample, one-half which was exposed to chemicals and the other half, protected by the silicon polymerised resin. An average oxide thickness of 5.0 ± 0.2 μm was seen. The EDX profile of this oxide is seen in Figure 3B. A multi-layered oxide thickness could therefore be seen on this oxide, with the inner layer having an enrichment in Cr and Ni oxide where the outer oxide layer was highly enriched in Fe oxides. These results were found to be coherent to the results obtained by J. Panter (Panter et al., 2006), A. Machet (Machet, 2004), and J. McGrady (McGrady, 2017) who also obtained a multilayer oxide layer, though under contact with high temperature water. Samples 1 and 3 from Table 2, which were oxidized in presence of boric acid, show a decrease in oxide thickness as compared to sample 2, as the concentration of boric acid is increased. This was also observed by Salih et al. (2014); Zhang et al. (2018).

Upon treatment, a disappearance of the majority of the oxide layer is seen in Figure 3C, with small remnants of the inner Cr oxide layer showing effectiveness of the treatment process. The presence of oxide can be attributed to the contribution of the resin.

Figure 4 shows the results of the treatment of the samples under different combinations using the CORD process. The results represent the sum of total concentrations of Cr, Fe and Ni after 3 cycles of the CORD process. Figure 4A represents the 3 different combinations of concentrations of permanganate and oxalic acid. Samples 1, 2 and 3 from Table 2 were utilised for these tests.The effects of higher efficiency at higher concentrations are evident with the total dissolved metals generally increasing with increase in concentrations across both potassium permanganate and oxalic acid. The selectiveness of the potassium permanganate towards Cr dissolution is also evident with significant higher Cr contents detected after the permanganate step of the process as compared to Fe and Ni. On the other hand, oxalic acid shows a much higher efficiency in dissolution of Fe and Ni whereas no increase in Cr dissolution was noticed across the 3 samples after the oxalic acid step in the process. This confirms the need for an oxidative step in the CORD process for the removal of Cr oxides as found in the literature and a reductive step with oxalic acid for the Fe and Ni oxides (Ocken, 1999; Balaji et al., 2018; Wille and Bertholdt, 1998). The first two concentration combinations required up to 6 cycles for complete removal of the visual oxide layer, though the sample was more or less clean after 5 cycles. Whereas for the combination with the highest concentration required only 3 cycles for complete removal of the visual oxide layer. A summary of the number of cycles applied to various samples is presented in Table 2.

Figure 4B shows the effects of a fixed (high) concentration of oxalic acid with the potassium permanganate concentrations varying as before. Samples 3, 6, and 7 from Table 2 were used for these tests. The higher oxalic acid concentration appeared to affect the number of cycles required, as seen in Table 2 with the surface appearing much cleaner after 3 cycles, as seen in Figure 5A than at lower oxalic acid concentration in Figure 5B.

Similarly, Figure 4C shows the effects of a fixed (high) concentration of potassium permanganate and varying oxalic acid. Samples 3, 4, and 5 from Table 2 were used for these tests. The results here showed less promise at lower oxalic acid concentrations for total dissolution of the oxide layer mainly due to the lower efficiency in dissolution of Fe which is the major component of the oxide layer in stainless steels. The samples did appear to be visually cleaner than those at low concentrations of potassium permanganate as in the case of Figure 4A but this could be due to the smaller size of the oxide layers as seen in Table 2. The results obtained here suggest significant contribution of oxalic acid concentration to the dissolution of the oxide layer for stainless steel arising from its efficacy in dissolution of Fe oxides. Higher overall dissolution of the oxide is observed at higher dissolutions of Fe oxide.

The results presented here do not make use of any UV technology for the oxalic acid stage as suggested by Wille and Bertholdt, (1998) and Ocken, (1999). The results depict the CORD process is effective at all tested concentrations with fewer cycles required at higher concentrations of the acids. This also means higher concentrations of metals found in solution leading to easier precipitation discussed further in Section 3.4. This could suggest the application of the CORD process at a higher concentration without use of UV is possible.

The polarisation curves of stainless steel 316Ti in 3 mmol/L HNO₃ and 2, 6, 15 mmol/L KMnO₄ are presented in Figure 6A. The results of the Tafel-extrapolation are summarised in Table 3.

In Figure 6A the passivisation effects of the strong oxidising agents can be observed. The current densities in the cathodic polarisation curves were higher, when the concentrations of the KMnO₄ were increased and the amount of the current in the anodic polarisation decreased by larger KMnO₄ concentrations. Further, the anodic polarisation curves showed passive regions. Similar characteristics of polarisation curves were observed with previous studies (Tian et al., 2019). According to the results of Tafel-extrapolation, passivation effects of KMnO₄ were significantly higher above 6 mM concentration. The equilibrium current density in 15 mM concentration was more than 4 times higher than in 6 mM and more than 5 times than in 2 mM.

The dissolutions of the oxide-layer from stainless steel samples after treatment with oxalic acid can be seen in Figure 6B. Table 3 shows the results of Tafel-extrapolations. In this case the equilibrium current density continuously increased by dosage of the acid and it reached the maximum at the highest concentration (18.5 mM).

The modelling of the solubilities of the different metal hydroxides as function of the pH and metallic concentration in solution is shown in Figure 7.

According to the modelling results, the solubilities of metal hydroxides are highly dependent on the pH of the solution. Indeed, it is not possible to fully precipitate all the metals present for a single pH value. It is therefore necessary to carry out several successive precipitations in order to obtain hydroxide slurries containing every metal. These solubility curves correspond to those found in the literature (Cravotta, 2008; Oncel et al., 2013). Their model also shows that the pH of 8.5 and 12 are ideal to precipitate the metals of interest. From the solubility curves in Figure 7, it is possible to group the hydroxides according to two precipitation pH values. The first can be carried out at pH 8.5 to precipitate Fe, Cr, and Zn, while the second, at pH 12, removes Mn, Ni and Co. To support this hypothesis, according to an IAEA report, the nuclear research centre of Marcoule in France also adjusts the pH to 8.5 to precipitate iron during the waste treatment (IAEA, 1992).

However, the modelling highlights one of the limitations of the precipitation phenomenon, which is dependent on the concentration of the metals. Indeed, it is not possible to precipitate 100% of the metals in solution because of the solubility limit of the generated hydroxides.

Figure 8 shows the results of the precipitation experiments of a synthetic stainless steel solution to which other metals have been added. Metal concentrations were analysed after each step of the test. These data show the importance of carrying out two separate precipitations for two different pH values. Indeed, some metals such as iron, chromium, zinc seem to precipitate mainly at pH 8.5 but others metals like manganese, nickel and cobalt require a second precipitation at pH 12 to be have higher removal efficiency. The first precipitation removes up to 99.99% of Cr, 57.20% of Mn, 99.96% of Fe, 90.21% of Co, 95.23% of Ni and 99.30% of Zn. This also corresponds to the removal efficiency observed by Oncel et al. (2013) who found 99.96% for Fe, 99.50% for Cr and 99.59% for Zn for a precipitation at pH 8, therefore portraying the reliability of the method. After the second precipitation, a total of 99.99% of Cr, Mn, Fe, Zn and 99.98% of Co, and 99.23% of Ni have been removed. During this step, manganese is almost totally removed from the solution compared to the first precipitation at pH 8.5. This behaviour was also observed by Balintova and Petrilakova (2011) where 16% of the manganese was removed at pH 8.2 and 89.49% at pH 9.98. The results are also coherent with those obtained from modelling the solubility of hydroxides in Figure 7.

Some fission products like Sr, Cs, Pu, Am, and Ce, could also be found in solution. These elements were not modelled because of their very low concentrations compared to the ones selected in Figure 7. According to Lehto and Hou (2010), Am precipitates as Am(OH)₃ between pH 11 and 12, Pu is expected to precipitate as Pu(OH)₄ for a pH > 6 as concurred by Stroud et al. (2006). For Ce, it precipitates as Ce(OH)₃ from pH 8.5 according to Bouchaud et al. (2012). Whereas for Sr, 99% of the hydroxide complex Sr(OH)₂ remains in solution at pH ≈ 13 and in the pH range (6–12) targeted in this work, Cs remains in solution as Cs⁺ as seen by Bull (1974). The removal of these soluble fission products such as Cs and Sr along with other traces would require extraction using ion exchange resins.

The results in Figure 8 were obtained from a synthetic sample generated with dissolved metal salts. At the end, the final concentrations obtained at the end of the protocol are between 0.05 and 60 μg/L for initial concentrations between 1 and 800 mg/L. Based on the information of AmberLite IRN97 H Ion Exchange Resin (Dupont Water Solutions), which is a potential nuclear grade strong acid cation resin used in extracting of the dissolved metals, calculations were done to simulate the amount of resin saved by using the precipitation protocol. According to the total exchange capacity of the resin (≥2.1 eq/L), the initial volume of the solution (250 ml) and the concentrations of metals in Figure 8 and an assumed efficiency of 20%, a theoretical estimate of 30 g of resin would be required without the precipitation protocol as opposed to just 2.5 mg with. This represents a reduction of 99.99% in the amount of resin required. The estimate here doesn’t account for the added Na⁺ content in solution nor the K⁺ in solution and only estimates the requirements for the dissolved metals. Additional volumes of the resin would therefore be required to retain the K⁺ and Na⁺ from the solution. Further tests are required for the precipitation protocol using KOH instead of NaOH, which could avoid the addition of Na⁺ into the solution at the cost of increased K⁺ concentration. The resin, Amberlite IRN97 H can also be used in the removal of K⁺ (Dupont Water Solutions, 2019).

The combined mass of the sludge across the two precipitation steps on the synthetic sample in Figure 8 was approximately 14.5 g. Upon drying, the total precipitate weighed approximately 0.7 g, suggesting about 95% of weight is lost in water during drying. Cements and geopolymers are potential candidates for the final conditioning of these precipitates but these require evaluation in terms of mechanical resistance, and radio-chemical resistance and durability.

However, the CORD process uses oxalic acid, which is found in the liquid waste. It is known that some metals can form a complex M(C₂O₄) which can be soluble in solution like chromium oxalate, which was also seen in a literature study (Remoundaki et al., 2007). Preliminary tests were carried out on the surrogate metal samples without destruction of oxalic acid, and the initial results portrayed lower precipitation levels mainly arising from Cr. This data and the pictures of the solution at different pH can be found in the supplementary data in Supplementary Figures S2, S3, and S4. Therefore, it is necessary to destroy the remaining oxalic acid before the precipitation process. Preliminary tests have shown the use of H₂O₂ under heating is a viable option to destroy oxalic acid, and the results can be found in supplementary data in Supplementary Figure S5.