Many of the chemical and mineralogical changes in pastes made with portland cement (PC) and with portland cement plus silica fume (PC-SF) subjected to magnesium sulfate attack occurred in a zone which was referred to as "transition zone". This zone extended from the surface double-layer inwards where only few changes could be detected. The width of the transition zone varied according to the character of the paste, being greater for the PC paste. Among the characteristic features of the transition zone were the continuous increases in (1) the Ca/Si ratio of the C-S-H gel, and (2) the bulk CaO/SiO2 ratio of the paste from the surface inwards. In addition, the transition zone had a smaller amount of unhydrated cement particles and exhibited a strong differentiation in the relative amount of the residual phases. The mechanism of the magnesium sulfate attack involved the formation of two counter diffusion patterns. Hydroxide ions, (OH)-, diffused outwards from the paste, whereas SO4(2-) ions diffused inwards from the surface. Due to the lower Ca/Si ratio of the C-S-H gel in the PC-SF paste and its alteration products, the PC-SF paste was more susceptible to the magnesium sulfate attack than the PC paste. A schematic model is presented which is based on the microstructure (presented in Part I) and the spatial chemical and mineralogical variations. The model attempts to provide explanation and trace the progression of magnesium sulfate attack.
