First results on the thermodynamic databases and reactive transport models for steel-cement interfaces at high temperature
Résumé
The interface between carbon steel and cementitious materials is a key issue in the concepts of disposal cell for vitrified high-level waste (HLW) in argillaceous sedimentary formations for the Belgium, Dutch and French national programs. The first two programs rely upon supercontainers containing the HLW encased in prefabricated cylindrical concrete buffer material. The concrete is made from CEM I cement and limestone aggregates. The pH has to be kept at high values during the thermal phase, and much longer beyond, in order to keep the carbon steel overpack passivated, to limit corrosion and ultimately radionuclide release. In the French concept, the annular gap between a carbon steel sleeve and the host rock is filled with a bentonite/cement grout that imposes corrosion-limiting environmental condition during the thermal phase only. The alkalinity of the grout is moderate (pH ~ 11) and should have been neutralized readily afterwards to prevent the dissolution of the nuclear glass and radionuclide release under alkaline pH values. The different national concepts of disposal cell for intermediate level waste (ILW) present a wider diversity of configurations and materials compared to the HLW disposal concepts. However, in most concepts carbon steel is used for the containers of the ILW packages that are the immobilized in a concrete or steel container with a cement-based material. The subtask 2.2 of the ACED WP is dedicated to identify and model the main reactive processes at the cement/steel interface and their consequences in terms of transfers and microstructures evolutions. All data (from modelling and experiments) have to describe geochemical behaviors in function to the environmental constraints, representative of conditions planned to be encountered in geological facilities. The thermodynamics of Fe-containing cement phases represents the first main input of this report, based on a broad bibliographical review and first results. A second major issue is to investigate the kinetics of the steel/cement interactions in the geologic disposal, from a literature review and first results obtained in the BACUCE project. A last and third objective is to gather the chemical and physical properties of all the materials considered in Task 2.2 and to model their initial states. The first part provides an overview on the thermodynamics of Fe-containing cement phases, corrosion products, and iron pore solution speciation. The use of non-ideal solid-solutions to model some of the cement hydrate phases is detailed, together with preliminary results on the iron uptake in C-S-H. These are essential for assessing the fate of iron that results from corrosion and serve as inputs and constraints for improving the reactive transport modelling at the interface and waste package level. CEMDATA18 database for cement phases was improved from Cemdata07 mainly with the extension of aluminum and iron containing phases as well as the changes in the C-S-H models concerning the volume and the alkali uptake. These changes can lead to considerably different modeling results between the two database versions. Thermoddem database contains data for modelling cement phases, iron corrosion products such as iron carbonates, oxyhydroxides, silicates and sulfides, these can be supplemented with data for iron corrosion products or clayey phases can be selected from ThermoChimie. Fe-siliceous hydrogarnet is the most stable iron bearing cement hydrate phase from ambient to elevated temperatures. This is replaced by ferrihydrite in low pH cements. Generally, Fe(III)-bearing cement phases, Fe oxides and hydroxides have low solubility. Their stability increases with temperature increasing and this keeps the iron present in the pore solution at low concentrations. This and the uptake of iron in C-S-H, a main cement phase, will result in the retention of iron near the cement/metal interface. The most stable corrosion products in highly alkaline conditions (pH>11), maintained by the cement system, is magnetite and goethite under anoxic and oxic conditions, respectively. The formation of the stable iron oxides/hydroxides can be inhibited, depending on kinetics, temperature and local composition, resulting in the formation of different intermediate metastable phases such as amorphous oxide/hydroxide phases, hydroxychloride phases, green rust GR1(Cl-). These phases are more soluble, and their presence will enhance the mobility of iron. Because of the high pH, iron in the pore solution will be in its hydrolyzed form. The presence of chloride ions and other ligands could lead to an increased amount of dissolved iron and the stabilization soluble iron phases for instance green rust GR1(Cl-). EURAD Deliverable D2.8 – Thermodynamic database and model of steel-cement interfaces, 1st results EURAD (Deliverable n° 2.8) Dissemination level: PU Date of issue of this report: 18/02/2021 Page 5 Some of the Fe containing cement phases like the (Al, Fe)-monosulfate and (Al, Fe)-ettringite are modeled as non-ideal solid solutions. For simplification in reactive transport codes, they can be treated ideal solid solutions. This simplification should produce results for the solid and aqueous composition within the experimental uncertainity and will be further investigated. In the case of C-S-H, a non-ideal mixing model is necessary to accurately describe the experimental data on the uptake of iron and other elements. Preliminary results show that the model can be successfully parameterized against experimental data for iron uptake in C-S-H. After additional estimates necessary for the iron in aluminum bearing C-S-H, the model will be used to investigate the impact on the iron at the cement/steel interface. For codes that do not use complex solid solution models, we investigate the possibility of using a smart discretization procedure, for producing the optimal number of discrete phases that satisfactorily approximate the non-ideal solid solution model. In terms of material composition, low carbon or weakly alloyed steels are mostly made of metallic iron. They are never in equilibrium with water even under anoxic conditions and will be subjected to several types of corrosion processes, such as generalized (uniform) corrosion or localized corrosion (e.g. pitting corrosion). The present report will focus on the generalized corrosion of iron under anoxic and fully water saturated conditions: The effect of temperature, of water chemistry and pH (from neutral pH to highly alkaline pH) and of the configuration (steel in contact with aqueous solutions vs. close contact between solid materials). At 80°C and neutral pH in argillaceous media, the corrosion rate ranges from a few µm/y in synthetic clayey solution up to 30 µm/y when steel is embedded in the argillaceous rock. The corrosion decreases exponentially with time over one year approximately. Layered zonation of corrosion products (oxyhydroxide, chukanovite and siderite, iron silicates) generally takes place in an embedded configuration. At 80°C and high-pH conditions, the long-term corrosion rates (beyond one year) is 0.1 µm/y or lower due to passivation by the formation of compact magnetite film or layer uniformly over the whole surface of iron. The passivation of the steel by magnetite may become inactive when the alkaline pH decreases below 10.5. However, the corrosion rates under moderately alkaline pH (10.5 – 11.5) are not well characterized yet, and will be a major input of the Subtask 2.2. The very few papers indicate that corrosion rates one order of magnitude higher than in classical cement environment. The activation energy is commonly low for all the pH range and temperature increase slightly enhance corrosion in most cases. The modeling of experiments in the Subtask 2.2 will be performed with reactive transport modeling (RTM), in particular HYTEC. The main mathematical features of RTM are presented with as special emphasis on temperature dependency for thermodynamics as well as kinetics and diffusion (Arrheniustype laws). The possibility of chemical feedback on porosity and diffusion could be important for modeling the time evolution of the steel/ cement interface. HYTEC is interfaced with the Thermoddem database and will benefit from the development of the thermodynamics of iron-cement phase performed in Subtask 2.2. The kinetics of the formation of the corrosion products is also introduced in the report. The BACUCE experiments are taking place in the IRSN underground research laboratory located in an argillaceous formation of Toarcian argillite at Tournemire (France). The in-situ experimental set-up aims to characterize the interface between carbon steel and two cementitious buffers (a bentonite/cement grout and a CEM I paste) until three years of interaction at the ambient temperature of 15°C and at 80°C with a heating system. One specificity is to study the influence of imperfect contacts or voids that are filled with grout porewater solution mixed with the argillite one. Experiments at the laboratory are also conducted in support of the in situ experiments at 20°C. The first lab experiment concerns the porewater chemistry of the bentonite/cement grout. The second experiment assesses of the efficiency of the slightly alkaline conditions to protect the steel against corrosion. Complementary experiments at 80°C are under progress in Subtask 2.2. The third lab experiments are planned to assess the effect of mill scale on the corrosion rate of low carbon steel in similar conditions as the BACUCE experiments. These experiments will also allow to evaluate the synthetic porewater evolution in contact with the two buffer materials (CEM I/MREA) at 80°C, as well as the cement and steel corrosion products evolution. The geochemical, mineralogical and transport parameters of all materials (CEM I paste, grout, argillite) have been compiled and adapted as HYTEC input files. The initial chemistry and mineral assembly have been modeled, first at ambient temperature, and after a temperature increase up to 80°C. In hydrated CEM I cement pastes, temperature affects the balance between AFT/AFm phases, decreases pH (but the hydroxide content remains high) and rises the sulfate dissolved concentration. Compared to CEM I, the bentonite/cement grout is still poorly known. Modeling has been done on batch tests with hydrated grout and water at 20°C. The model has then been extrapolated to the porewater and mineralogy of the hydrated grout at 20°C and 80°C. Two sets of thermodynamic data have been tested, improving the description of the C-A-S-H phases. The implementation of a kinetic approach for the hydrations of slag and silica fume, especially at 20°C, will be done as an improvement of this preliminary model. Eventually, the modeling of steel corrosion is on-going (corrosion products and intrinsic rate constant). An alternative set of lab experiments are also in progress in Subtask 2.2 to gain information on corrosion intensity at the interfaces between carbon steel and a CEM II/B -based concrete. A temperature of 80°C is also considered for the sake of consistency and comparison with the other experiments of Subtask 2.2. Synthetic groundwater of the Boda claystone is used for hydrating the cement and as boundary conditions. The physical and chemical properties of the concrete and the Boda water chemistry have been compiled to further define the initial states of the reactive transport model. The report ends with a series of planned continuations and improvements. Subtask 2.2 will also pass information from its relatively isolated small-scale processes investigated on interface scale to more complex systems at waste package scale (task 3) and disposal scale (task 4).
Domaines
| Origine | Fichiers éditeurs autorisés sur une archive ouverte |
|---|---|
| Licence |
