Materials and Experimental Methods

A full description of the sample preparation, measuring technique and method for calculation of the diffusion coefficients for the cement pastes is given elsewhere [8,22]. Only a brief description is given here.

The cement used was a white Portland cement with a Blaine fineness of 420 m²/kg, whose Bogue calculated potential phase composition is provided in Table 1. Silica fume was added as a dry powder. The silica fume employed in these experiments had a specific surface of 20.5 m²/g and a chemical composition of (mass fraction, %): SiO₂: 90.8, Fe₂O₃: 0.94, Al₂O₃: 0.54, MgO: 1.32, and SO₃: 0.57. Superplasticizer was added at a rate of 1.0 % by mass of cement + silica fume. The superplasticizer was a napthalene-based dry powder. Demineralized, freshly boiled water was used for all mixes.

 

Table 1:
Bogue Potential Phase Composition of Cement

Phase	mass %
C3S	66.1
C2S	21.2
C3A	4.3
C4AF	1.1
C$C\bar{S}$	3.5
free CaO	1.96
Na2O equivalent	0.17

Mixing was performed for 5 minutes in a 5 liter epicyclic laboratory mixer. The water was added in two steps during mixing. This procedure ensures the homogeneity of the paste and the proper dispersion of the silica fume.

After mixing, cylindrical samples 44 mm in diameter and 21 mm in length were cast from the cement paste. These samples were then stored in lime-saturated water for approximately 100 days for prehardening and water saturation. Before the chloride exposure, approximately 1 mm was ground off of each sample base. The bottom and cylindrical surfaces of the sample were then sealed with a 1 mm layer of polyurethane, with the newly ground surface being subsequently exposed to chlorides. For this, the samples were exposed to a simulated pore solution also containing 3 % NaCl (as in typical seawater) for one month at 20 ºC. In this way, leaching of the cement pastes during the chloride exposure was avoided. Leaching can have a serious influence on the ingress of chlorides, with calculated diffusion coefficients being up to 2.5 times greater under leaching than under non-leaching conditions [22].

After chloride exposure, the samples were vacuum dried and the chloride profiles were measured by electron probe micro analysis (EPMA). Detailed studies of various drying regimes have indicated that the chloride profiles are unaffected by the vacuum drying process [22]. Typical profiles are provided in Fig. 1. The modelled chloride profiles have been calculated with a computer program that assumes chloride transport to follow Fick's law and takes into account chloride binding by the cement paste components [8,22].

  

Figure 1: Influence of silica fume addition on chloride ingress. Chloride profiles for five different pastes are shown. Except for silica fume addition, marked on the curves, the pastes are identical. The pastes have a w/c ratio (mass basis) of 0.3 and have been exposed to a 3 % NaCl solution for one month at 20 ºC. Leaching was minimized throughout the chloride exposure. The smooth curves are calculated profiles with diffusion coefficients as given in the table inset within the plot. The fluctuating curves are based on the EPMA mapping [22].