Characterization of chloride-induced corrosion products that form in steel-reinforced cementitious materials

Abstract

The goals of this investigation were to identify the chloride-induced corrosion products that form from steel reinforcement in concrete, determine where they form, the corresponding corrosion rate, and relate these observations to the performance of steel-reinforced concrete in the field. This information is intended to be incorporated into structural service life models such that more accurate predictions of the remaining lifetime of a structure can be made. To accomplish these goals, experiments were conducted on various cementitious materials that ranged from steel in simulated pore solution to concrete of various types.

In situ electrochemical and Raman spectroscopy observations were performed on steel immersed in either a Type 10 or Type 50 white cement simulated pore solution that contained varying levels of chlorides up to 1 M NaCl. The composition of these pore solutions was developed from pore solutions expressed from 6 month old cement paste cylinders. The surface of the steel was either ground with 240 grit SiC paper or left in the as-received condition with a mill scale predominately composed of magnetite (Fe3O4). The effects of this type of pore solution, surface condition of the steel, and varying chloride exposure on the type and distribution of corrosion products were studied. The results indicated that the critical chloride/hydroxide ratios at which corrosion initiates depend upon the pH of the simulated pore solution, even within the narrow range of 12.9 to 13.4. Corrosion initiated with lower chloride levels and at lower applied potentials in the lower pH level of the white cement simulated pore solution than in the relatively higher pH Type 10 cement simulated pore solution. In addition, mill scale was observed to provide some enhanced corrosion rates of the as-received steel were similar tot he ground steel surfaces. The corrosion products observed included magnetite, maghemite, Green Rust I, and haematite.

The coordinate with the aforementioned simulated pore solution-steel experiments, steel plates were cast into a 0.45 w/c cement paste which has been stabilized with 10% by mass of silica sand (henceforth referred to as modified cement paste). As before, both Type 10 and white cement pastes were studied as well as the effect of the surface finish of the steel, either ground or as-received. These steel-reinforced cement paste prisms were cured in their moulds for 3 months and then partially immersed in a their respective simulated pore solutions with sufficient chlorides, added as NaCl, to make a 1M solution. Potential mapping in a manner similar to ASTM C876 and linear polarization resistance measurements were performed regularly to determine the corrosion state of the prisms. Once corrosion was considered to be initiated, the prisms were sectioned and examined ex situ using chemical, macro- and microstructural techniques to determine the influence of cement type, surface finish of the steel, surface cracks, and the confinement of the modified cement paste cover on the formation of corrosion products. A range of corrosion products was observed to form primarily within shrinkage cracks: magnetite, goethite, haematite, and possibly Green Rust I. The formation of these products could not be correlated to the measured corrosion rates because the area of steel actually corroding could not be determined non-destructively. Furthermore, some corrosion products such as magnetite, haematite, and goethite were observed to form within the Type 10 modified cement paste cover but not within the white modified cement paste because of the relatively more open pore structure in the former.

The final experimental program involved the study of four different types of concrete: a low quality concrete (o.54 w/cm), an industrial standard concrete (0.41 w/cm), a high performance concrete (0.27 w/cm), and a high performance concrete with 10% by mass of cement of silica fume (0.25 w/cm). Concrete prisms (500 x 100 x 100 mm) had been cast commercially from these four mix designs with an embedded five element corrosion probe. Channels had been sawn into half of the prisms to position ~0.3 mm cracks subsequently induced by three-point bending. The prisms were exposed to simulated sea water (ASTM D1141) for up to four years. During this time, the corrosion state of the prisms was regularly assessed using open circuit potential and linear polarization resistance measurements. After at least three years exposure, the average corrosion rates (i.e., not corrected for area of steel actually corroding) indicated that the cracked low quality prisms had the highest corrosion rate (~0.04 A/m^2) while the steel in the industrial standard and high performance concretes were corroding at lower rates (~0.01 A/m^2). The same trends were noted in the uncracked specimens except that the rates were approximately one order of magnitude lower for only the high performance concretes (~0.002 A/m^2). Those prisms with the highest corrosion rates were then sectioned and studied in a manner similar to the modified cement paste prisms with particular attention paid to the effect of concrete type and position of the induced crack, on the type and location of any corrosion products that formed. Within the industrial standard and high performance concrete with silica fume, a dense layer of magnetite formed at the steel/concrete interface and within the induced crack. Magnetite also formed in the other two concretes but additional corrosion products such as akaganeite, goethite, and haematite were also observed within the concrete cover and were attributed to the higher connectivity of their pore networks from a higher w/cm ratio or microcracks. The removal of the corrosion products from the steel permitted the area of the steel that actually corroded to be estimated. Corrected corrosion rates indicated that the steel in the cracked high performance concretes was corroding at a rate almost an order of magnitude higher than the steel in either the industrial standard or low quality concretes with the higher electrical resistivity of the former concretes confining corrosion to a localized area at the root of the crack.

Overall, all experiments indicated that there is not any correlation between the corrosion rate measurements, steel surface finish, or chloride ion source on the types of location of corrosion products. Products which, int heir pure state, have specific volumes not more than 3.5 were observed within the concrete specimens which suggests that the typically assumed range of 6 to 7 must be revised to reflect the service environment when used in theoretical service life models.