Assessment of the evolution of the redox conditions in a low and intermediate level nuclear waste repository (SFR1, Sweden)

Abstract The evaluation of the redox conditions in an intermediate and low level radioactive waste repository such as SFR1 (Sweden) is of high relevance in the assessment of its future performance. The SFR1 repository contains heterogeneous types of wastes, of different activity levels and with very different materials, both in the waste itself and as immobilisation matrices and packaging. The level of complexity also applies to the different reactivity of the materials, so that an assessment of the uncertainties in the study of how the redox conditions would evolve must consider different processes, materials and parameters. This paper provides an assessment of the evolution of the redox conditions in the SFR1. The approach followed is based on the evaluation of the evolution of the redox conditions and the reducing capacity in 15 individual waste package types, selected as being representative of most of the different waste package types present or planned to be deposited in the SFR1. The model considers different geochemical processes of redox relevance in the system. The assessment of the redox evolution of the different vaults of the repository is obtained by combining the results of the modelled individual waste package types. According to the model results, corrosion of the steel-based material present in the repository keeps the system under reducing conditions for long time periods. The simulations have considered both the presence and the absence of microbial activity. In the initial step after the repository closure, the microbial mediated oxidation of organic matter rapidly causes the depletion of oxygen in the system. The system is afterwards kept under reducing conditions, and hydrogen is generated due to the anoxic corrosion of steel. The times for exhaustion of the steel contained in the vaults vary from 5 ky to more than 60 ky in the different vaults, depending on the amount and the surface area of steel. After the complete corrosion of steel, the system still keeps a high reducing capacity, due to the magnetite formed as steel corrosion product. The redox potential in the vaults is calculated to evolve from oxidising at very short times, due the initial oxygen content, to very reducing at times shorter than 5 years after repository closure. The redox potential imposed by the anoxic corrosion of steel and hydrogen production is on the order of −0.75 V at pH 12.5. In case of assuming that the system responds to the Fe(III)/magnetite system, and considering the uncertainty in the pH due to the degradation of the concrete barriers, the redox potential would be in the range −0.7 to −0.01 V. A Monte-Carlo probabilistic analysis on the rate of corrosion of steel shows that the reducing capacity of the system provided by magnetite is not exhausted at the end of the assessment period, even assuming the highest corrosion rates for steel. Simulations assuming presence of oxic water due to glacial melting, intruding the system 60 ky after repository closure, indicate that magnetite is progressively oxidised, forming Fe(III) oxides. The time at which magnetite is completely oxidised varies depending on the amount of steel initially present in the waste package. The behaviour of Np, Pu, Tc and Se under the conditions foreseen for this repository is discussed.

The SFR1 repository contains heterogeneous types of wastes, of different activity levels 23 and with very different materials, both in the waste itself and as immobilisation matrices 24 and packaging. The level of complexity also applies to the different reactivity of the 25 materials, so that an assessment of the uncertainties in the study of how the redox 26 conditions would evolve must consider different processes, materials and parameters.

28
This paper provides an assessment of the evolution of the redox conditions in the SFR1. 29 The approach followed is based on the evaluation of the evolution of the redox simultaneously. The assessment of redox conditions can only be approximated by 88 adequately discussed assumptions (Wanner, 2007). Uncertainties concerning processes, 89 parameters and materials must be also assessed and evaluated. 90 The most usual way to define the redox state of a system is based on the measure or 91 determination of the redox potential (Eh), which can be related to the concentration of 92 the redox active species in the system. Nevertheless, redox reactions are known to be 93 slow and in many cases they need the action of catalysts to proceed, despite being 94 thermodynamically favoured. Thermodynamic equilibrium among redox species can not 95 be granted and this precludes the achievement of conclusions on the redox state of a 96 system simply based on the determination of redox active species.

98
This implies that the simple monitoring and calculation of redox potentials is sometimes 99 not a very good indicator of the redox state of all redox couples in the system, as it has 100 been proven from many groundwater analyses that different redox potentials can be 101 measured depending on the redox couple considered to be in equilibrium in the system 102 (Lindberg and Runnells, 1984).      About fifty different type of waste categories are foreseen to be deposited in the SFR1 164 until its closure (approx. at year 2050). They can be represented by 15 generic waste 165 packages (see Table 1) differing in the type and materials of the container, type and 166 amount of immobilising matrix and type and amount of waste.  In the light of the amount of materials present in the repository, many redox processes 173 can be identified as being of relevance in the SFR system: metal corrosion, degradation 174 of organic matter, radiolysis, sulphate reduction, gas generation and fermentation,...

175
The level of radiation that may cause generation of oxidants in the repository is very 176 low and previous assessments on the influence that radiolysis can have on the oxidant  Table 1 179 Suprimit: Figure 1 balance of the repository, such as the one in Moreno et al. (2001) have shown that this 181 effect is expected to be minimal. Therefore, processes that a priori can have a larger influence in the short and long-term  Just after repository closure, there will be oxidants in the system due to repository 200 construction and operation periods, and under non-disturbed conditions it is foreseen 201 that these oxidants will be rapidly consumed and the system will reach anoxia and later 202 develop reducing conditions. Groundwaters reaching the repository at long-term are not 203 expected to contain appreciable concentrations of oxidising species, due to their prior 204 interaction with soils and minerals that will cause oxidant consumption. According to 205 the expected evolution of the SFR1 environment (SKB, 2008), the only possibility of 206 oxidising conditions to reach repository depths arises from an inflow of glacial oxygen-207 rich and diluted water. This situation would, if at all, happen during ice melting events.

208
According to the foreseen climatic evolution of the system, the earliest time at which 209 melting water would reach the repository is 60 ky after its closure. This means that, 210 once the system has reached reducing conditions, its capability to buffer an oxidising 211 intrusion will basically remain the same until 60 ky and that after this time, it will be 212 jeopardised only in the case of melting ice water inflowing the system.        The following approach has been implemented in the model.  Table 1 by implementing the relevant redox processes into 266 a conceptual and numerical model.  The rate and extent of the different redox processes occurring in the system is discussed In the presence of oxygen, iron corrosion produces Fe(III) oxides and hydroxides.

291
Hematite is the thermodynamically stable Fe(III) oxide in the stability field of water 292 followed by goethite and ferrihydrite or hydrous ferric oxides. However, despite their 293 lower stability with respect to hematite, goethite and hydrous ferric oxides can prevail 294 Suprimit: Table 1 295 Suprimit: 1BTF, 2BTF   Table 2 and 337 have been selected after an extensive literature review.  Table 2 It is relevant to point out that fast corrosion of Al and Zn has been pessimistically has been attributed a generic composition of CH2O. The same mechanism is considered 369 for both "generic organic matter" and bitumen degradation (eq.4).   Table   382 2. Electron acceptors in the process are oxygen, ferric iron and sulphate. In order to 383 implement the bacterial activity and growth in the system, a threshold concentration of  Under abiotic conditions, that is, when electron acceptor concentrations are too low to 392 keep microbes active, generic organic matter and bitumen are assumed to chemically 393 degrade at a constant rate, equal to 10 -12 mol dm -3 s -1 , two orders of magnitude below 394 the biotic rate. Table 2 and Table 3 show the values of the parameters for these 395 equations.

397
In the model it is assumed that acetate generated during organic matter degradation 398 microbially degrades to carbonate according to eq.5. As in the case of bitumen and 399 generic organic matter, the acetate biotic degradation rate is controlled by the growth 400 rate of aerobes, IRB and SRB and the availability of O2, Fe(III) or S(VI) in solution.

401
In agreement with the available information in the literature, it is assumed that acetate Degrading cellulose is considered from a monomer of cellulose already hydrolysed 409 (C6H12O6 in eq.6).      Table 4.  can be oxidised by strong oxidants (e.g., O2) to a preselected equivalence point, which is 523 the Electron Reference Level (ERL). The RDC of a system gives, thereof, an estimation 524 of its capacity to accept oxidants, i.e., is a chemical sum of the maximum amount of 525 oxidants that the system is able to buffer.      Given the relevance that the iron system has in this environment, we have defined as 554 electron reference level the most oxidised form of iron, Fe(III). By considering this, 555 eq.11 and eq.12 are used to calculate both terms of eq.10.

581
In all package types except O.12 in vault BLA pH is buffered to 12.5 due to the 582 equilibrium with portlandite, which also controls Ca 2+ aqueous concentration.

584
A comparison of the time evolution towards reducing conditions depending on the 585 different redox processes considered is shown in Figure 5. As can be seen, low redox   The anoxic corrosion of steel has, as a consequence, the production of H2(g) at expenses 682 of water reduction. H2(g) is the most abundant gas generated in the system. According    The initial RDC in the system is given by steel and the different type of organic matter 712 considered (generic organic matter, bitumen and cellulose).  Formatat ... Suprimit: Table 5 751 Suprimit:      The results of the calculation indicate that at the end of the assessment period of interest 825 (100 ky) the system has not been depleted of its RDC, which is equivalent in this 826 simulation to say that not all magnetite has been transformed into goethite and, 827 therefore, the system is still able to buffer an oxidant intrusion ( Figure 13). After 40ky 828 of infiltration of meltwater, the amount of magnetite oxidised into goethite is 95%, and 829 5% of the magnetite remains unoxidised.

831
The amount of magnetite in the system provides enough RDC as to counteract an O2 832 intrusion for more than 50 ky. After this period Eh increases.   to be present in the system.

885
As shown in Figure 14, vaults inventories are lower than calculated solubilities except 886 for the BLA, where the formation of metallic Se could account for a more effective 887 retardation mechanism than for the other vaults. Therefore, the most likely retention 888 mechanism for Se is not expected to be precipitation, but interaction with solid surfaces 889 of the system. In the BLA vault, the Tc inventory is below the calculated solubility limit in all cases, 900 implying that Tc will not be solubility limited in this vault ( Figure 14). In the case of 901 1BTF, the inventory is either below of very close to the solubility limit (6.7×10 -9 M),

902
indicating that solubility will neither be an important retardation process for Tc in this   presumably cement and its degradation products as well as iron oxides (Figure 14).      The redox potential in the vaults is calculated to evolve from oxidising at very short 991 times, due the initial oxygen content, to very reducing at times shorter than 5 years after 992 repository closure. The redox potential imposed by the anoxic corrosion of steel and 993 hydrogen production is on the order of -0.75 V at pH 12.5 in case of assuming that the   the system to a given redox pair combined with the uncertainty regarding the pH