During the hydration of Portland cement at temperatures around ambient, ettringite is formed at early ages, a consequence of the use of gypsum as a set regulator (Bensted, 2002a). Since the cement paste is still plastic, it is able to accommodate the volume changes associated with the conversion of precursor phases to ettringite and significant expansion is not observed unless excessive levels of sulfate are present in the mix materials. Most national and international standards therefore put restrictive limits on the sulfate contents of cements (and on C3A contents), as well as on the soluble sulfate contents derived from aggregates, etc., in order that the formation of ettringite is virtually completed at about the time of setting and does not generate stresses that could cause what is sometimes referred to as `internal sulfate attack’ in concrete subjected to curing at normal ambient temperatures (Skalny et al., 2002). However, if Portland cement with a sulfate content within the normal accepted limits is cured at elevated temperatures, ettringite is destabilised and does not form at early age. Instead `delayed ettringite formation’ (DEF) takes place on subsequent cooling of the concrete concerned and, in some circum-stances, this may cause significant expansion, which represents a particular form of `internal sulfate attack’. German guidance suggests that expansion from DEF can be avoided if concrete is not heated above 60ëC (Deutsche AusschuÃ?? fuÈr Stahlbeton, 1989). QXRD results for ettringite formation in a mortar cured at 20ëC and 100ëC (Yang et al., 1999) are shown Fig. 4.3(a), with corresponding expansion data in Fig. 4.3(b). Similar results were found by Scrivener and Lewis (1996). Whereas at 20ëC, substantial ettringite is produced within one day with no expansion, upon heat curing, ettringite formation is delayed until about 25 days and is followed by expansion over the period 25 days to 400 days. Cement hydrated at either room temperature or elevated temperatures contains similar amounts of ettringite after sufficiently long storage times. Thus, the presence of ettringite in concrete which has undergone expansion is not a sufficient indicator to diagnose DEF damage. During the heat curing regime, the components of ettringite must be stored, and it is suggested that excess sulfate sorbs onto C-S-H gel while aluminate either forms AFm type phases or is sorbed on the C-S-H gel. Note that much of the Al analysed in regions of C-S-H gel can be shown by NMR to be incorporated into the silicate chains.
The state of the art was reviewed comprehensively by Taylor et al. (2001) and Famy et al. (2002a) and there are significant sections as well as two case studies concerning DEF in the recent book by Skalny et al. (2002). There is some confusion in the terminology used by different authors, with the terms `delayed ettringite formation’, `secondary ettringite formation’ and `ettringite reformation’ all used. Skalny et al. (2002) have suggested that it would be more consistent to refer to DEF as `Heat induced internal sulfate attack’ but here we follow current convention and use `DEF expansion’ to refer to formation of ettringite, delayed because of heat curing, that leads to expansion of cement paste, mortar or concrete, in the absence of an external sulfate source. The environment of the cement paste plays a significant role in the likelihood of expansion following a heat cure (see, for example, the section on alkalis below), so the suggestion of the use of `internal’ is perhaps in some ways misleading. DEF described here only refers to internally generated ettringite.