Influence of non-linear chloride binding on the determination of apparent chloride diffusion coefficient for cement paste with mineral additives

The determination of apparent chloride diffusion coefficients (D_a) in standard codes employs the linear chloride binding assumption. However, this assumption is too simplified as chloride binding is actually non-linear. This paper intends to investigate the influence of non-linear chloride binding on the determination of D_a values in the bulk diffusion test. To achieve this aim, a semi-analytical method for the semi-infinite diffusion problem was firstly established. This method can be used in data fitting to obtain the D_a values considering non-linear chloride binding. Then, bulk diffusion test and mercury intrusion porosimetry were conducted on four different cement pastes with fly ash and blast furnace slag. Based on chloride profiles, the chloride binding isotherms were obtained through data fitting. Meanwhile, the D_a values considering the linear and non-linear chloride binding assumption were calculated respectively and compared. The results show that the chloride binding isotherms in the bulk diffusion test fit well with the Freundlich isotherm but are not so non-linear as expected. The proposed method can calculate the effective diffusion coefficient which cannot be obtained in standard codes. D_a values will vary along the chloride ion diffusion direction rather than be a constant once the non-linear chloride binding is considered. The comparison between different cement paste mixtures indicates that both the apparent diffusion coefficient calculated from free chloride profiles or total chloride profiles can only be regarded as a qualitative description of resistance to chloride diffusion for a certain mixture. The comparison between different mixtures based on these values is not reliable.