Cesium and Strontium Contamination of Nuclear Plant Stainless Steel: Implications for Decommissioning and Waste Minimization

Stainless steels can become contaminated with radionuclides at nuclear sites. Their disposal as radioactive waste would be costly. If the nature of steel contamination could be understood, effective decontamination strategies could be designed and implemented during nuclear site decommissioning in an effort to release the steels from regulatory control. Here, batch uptake experiments have been used to understand Sr and Cs (fission product radionuclides) uptake onto AISI Type 304 stainless steel under conditions representative of spent nuclear fuel storage (alkaline ponds) and PUREX nuclear fuel reprocessing (HNO3). Solution (ICP-MS) and surface measurements (GD-OES depth profiling, TOF-SIMS, and XPS) and kinetic modeling of Sr and Cs removal from solution were used to characterize their uptake onto the steel and define the chemical composition and structure of the passive layer formed on the steel surfaces. Under passivating conditions (when the steel was exposed to solutions representative of alkaline ponds and 3 and 6 M HNO3), Sr and Cs were maintained at the steel surface by sorption/selective incorporation into the Cr-rich passive film. In 12 M HNO3, corrosion and severe intergranular attack led to Sr diffusion into the passive layer and steel bulk. In HNO3, Sr and Cs accumulation was also commensurate with corrosion product (Fe and Cr) readsorption, and in the 12 M HNO3 system, XPS documented the presence of Sr and Cs chromates.


S1. Steel Composition and SEM Analysis of Steel Surface
The arrows in image (C) show the attacked ferrite stringers (scale bar = 50 μm).

S2. GD-OES Depth Profiling
For a quantitative assessment of the elemental depth distribution of steel samples, it is necessary to correlate the GD-OES measurement time with depth of analysis. In this work, the depth of a GD-OES crater generated after 20 seconds of sputtering was determined using laser confocal microscopy ( Figure S2). Assuming a constant sputtering rate, 1 time can then be converted into depth by an appropriate scaling calculation. However, as the sputtering rate of a sample could be affected by changes in the composition of a material, the depth equivalent data can only be considered as an estimate.
The crater depth was measured to be 761 ± 59 nm, corresponding to an average sputtering rate of 38 ± 3 nm s -1 . The presentation of GD-OES profiles as a function of sputtered depth ( Figures 1, 4, and S3) reveals the surface oxide thickness after acid and alkaline passivation treatment as 6 ± 1 nm and 12 ± 1 nm, respectively. This result is consistent with a 316L stainless steel passivation kinetic study which reported a film thickness of 4.8 nm after immersion in 6 M HNO3 for 24 hours. 2 An equilibrium film thickness was not obtained in this 24 hour study and therefore our reported value of ~6 nm after 720 hours is likely to be reasonable. passivating medium). The reduced Fe 2p signal after acid passivation treatment is due to an increased Fe solubility at low pH, whereas a similar effect occurs for Cr under basic conditions.
The Fe 2p1/2 and 2p3/2 peaks at 724.6 and 710.7 eV, respectively, are associated with Fe2O3. 3 Furthermore, the contributions at 719.8 and 706.7 eV are due to metallic Fe, 4 likely corresponding to photoemission from bulk material. It is important to note that the feature at ~ 720 eV in the spectrum from the alkaline sample may also be ascribed to the Fe 2p3/2 satellite, where the presence of the corresponding Fe 2p1/2 satellite at 733.1 eV supports this assignment. 3 In addition, the feature at ~ 742 eV may also be identified as a daughter peak of one of the Fe 2p peaks, although an exact assignment remains unclear. 5 The Cr 2p1/2 and Cr 2p3/2 peaks at 586.2 and 576.7 eV, respectively may be assigned as Cr2O3. 6 For a similar reason outlined for Fe, elemental Cr was also identified by the corresponding photoelectron lines at 583.4 and 574.1 eV. These results reveal that, in combination with the GD-OES data, a fundamental structure of the passive layer is a Cr2O3 layer underneath a Fe2O3 over layer.
The relative concentrations of these two components are highly sensitive to the solution pH, where Cr grows at the expense of Fe oxide under acidic conditions. The surface enrichment of Fe after alkaline pH treatment is also apparent, although this effect is more subtle owing to the high Fe oxide content in the passive film formed by atmospheric exposure. The increased Fe stability within the passive film under alkaline conditions has important ramifications for the identification of Cs present in the steel material as the Cs 3d photoelectron are likely to be masked by the more prominent Fe 2p peaks ( Figure S4). Thus, despite an increased amount of Cs accumulating on the steel surface at alkaline pH (see Table S3), no Cs could be detected by XPS on the steel surface after contamination under alkaline solution conditions. Figure S4. XPS high-resolution spectra of (A) Fe 2p, and (B) Cr 2p photoelectron peaks of 304 stainless steel as a function of passivation treatment.

S4. Sr and Cs Sorption and Kinetic Modelling
The Ho model pseudo-second order kinetic fits are shown in Figure S5. The model is described in the main paper. The rate of adsorption for the Lagergren pseudo-first order model is dependent on the sorption capacity of the substrate, which is expressed as: 7 where qe is the equilibrium uptake (g m -2 ) and k is the first order rate constant (hr -1 ). The integrated form over the boundary conditions t = 0 to t = t and qt = 0 to qt = qt is Therefore a plot of log(qe-qt) against t will yield a linear relationship of gradient -k and a yintercept of log(qe) /k2q 2 e is obtained. A fundamental disadvantage of this kinetic model is that S8 some knowledge of the equilibrium sorption capacity must be known. In this work, the maximum qt value measured for each individual sorption was taken as qe. The pseudo-first order kinetic plots are shown in Figure S6 for all for systems studied, where the pseudo-first order rate constant can be determined by the gradient of the fit. Another kinetic model tested was the Elovich model ( Figure S7). In the Elovich equation, the overall rate of analyte removal from solution is derived from competing adsorption and desorption processes, 8 which is expressed as: where qt is the amount sorbed at time t, α is the initial sorption rate (g m -2 hr -1 ) and β is a constant related to the rate of desorption (m 2 g -1 ). The integrated form over the boundary conditions t = 0 to t = t and qt = 0 to qt = qt is Rearranging into the linear form yields In order to simplify this kinetic model, it is often assumed that αβt > 1 i.e. the contribution of t0 is negligible. 9 The rate equation then becomes: When the qt is plotted against lnt, a linear plot of gradient 1/β and a y-intercept of ln(αβ)/β is obtained. The corresponding kinetic plots are shown in Figure S7 for all for systems studied, where the values of β and α can be calculated from the slope and intercept of the fits, respectively.

S10
The statistical results of the Ho, Lagergren, and Elovich kinetic fits are summarized in Table   S2. It can clearly be seen that the Lagergren and Elovich equations do not give reasonable R 2 values and in all cases Sr and Cs sorption behavior can be more accurately described by Ho pseudo-second order kinetics.