Mineral admixtures, such as slag, fly ash, and natural pozzolans, have been used in the concrete mix proportions to mitigate or eliminate the AAR reaction. The following main theories have been proposed: 1) the blending agents act as a diluent because they are usually less reactive than portland cement and liberate alkali at smaller rates; therefore, the blended mixtures are able to maintain a higher effective water-cement ratio than the corresponding mixtures containing only cement, causing a further dilution of the alkali content. Some mineral admixtures, however, can have a higher soluble-alkalis content than cement. So, the beneficial effect may not occur when such admixtures are used; 2) mineral admixtures cause pore-size and grain-size refinement leading to a lower permeability and, consequently, a reduction of the migration of alkalis towards the reactive aggregate; 3) the blending agents lower the Ca(OH)2 content of the cement paste, reducing its pH; 4) the hydration products of blended cement systems have a higher binding power for alkalis than those of portland cement. The typical Ca/Si ratio of C-S-H in portland cements is about 1.8, but reduced ratios are observed when mineral admixtures are used. The surface charge on C-S-H depends on its Ca/Si ratio: when the Ca/Si ratio is high, the surface charge of C-S-H is positive and anions are adsorbed on the C-S-H fibers (but not cations as Na+ and K+, which remain in the pore solution); when the Ca/Si ratio is lower than about 1.2-1.3, the surface charge of C-S-H becomes negative and alkali cations are incorporated on the C-S-H.

The figure below compares the efficiency of different levels of slag in reducing the expansion in specimens made with cement 1. Only limited improvement was observed when 50% slag-replacement was used, but greater reductions in expansion were noted for higher replacement levels. Slag replacements up to 45% did not effectively reduce the mortar bar expansion. However, remarkable reductions in expansion were observed for higher slag replacement.

The micrographs below show the difference in the structure when 45% and 55% of slag are used. Note that the lack of gel around the aggregate for the higher level of slag replacement.

From Wang et al. Cement and Concrete Research, V27 N12:1899, (1997).