The effects of chemical transformation processes on the frost and frost-deicing salt resistance of concrete are much less significant than the physical effects, but they are nevertheless significant. Our investigations showed that monosulfate (AFm phase) is particularly instable and will transform to ettringite (AFt phase) under frost and also under frost-deicing salt attack. This delayed formation of ettringite, which is supported by thermodynamic conditions at low temperatures, may reduce considerably the frost and frost deicing salt resistance of concretes without air-entrainment.

During the past few years delayed ettringite formation (DEF) has probably received more attention, and been involved in more controversy, than any other concrete deterioration mechanism. Even its name has been subject to dispute. Our extensive experience on the investigation of many occurrences of DEF is presented here as a series of questions, with some answers. Where answers have become available, they have explained phenomena that have greatly bothered us and other investigators. Where answers are not available, the questions will provide directions for needed research.

The formation of ettringite in hardened concrete is not only a problem of heat treatment. Ettringite also occurs in no heat-treated concrete, which is exposed only to normal climatic conditions. In some cases the mechanism of damage in concrete pavements correlates with this ettringite formation in the hardened concrete. Structural changes by ettringite formation were caused above all by varying moisture conditions and, as a result, by transportation of moisture and substances within the concrete structure, which also lowers the pH value of the pore solution. The primary ettringite from the paste is microcrystallin at normal pH of 13.5 to 14 in the pore liquid. Thus ettringite may dissolve in the pore liquid and recrystalize at a lower pH in larger spaces, where the capillary transportation is interrupted. This recrystallized ettringite in the air voids was stable up to 60°C. But the mechanism of this ettringite formation is supported and accelerated by higher temperatures (e.g. 60°C) because of the intensive drying. Microstuctural defects like microcracks may be created by alternating temperatures and later on filled and may be widened by ettringite crystals. In concrete pavements no indications were found for recrystallized ettringite itself to be the primary cause of crack formation. The expansion of concrete is reduced by introducing artificial air voids, because there is more available space for accumulation of ettringite. But the combined action of freezing and thawing and de-icing salt after filling the artificially entrained air voids with ettringite crystals may causedamages.

In a number of isolated cases in the U.K., expansion and cracking has occurred in precast concrete elements subject to early heat treatment followed by wet or moist exposure, and in some in-situ concretes of large section and high cement content, again subject to wet or moist exposure. The present paper discusses some of the work relating to the cracking of these field concretes and to the laboratory expansion testing of concretes and mortars subject to high early temperatures which has been carried out by the British Cement Association. It is shown that expansion is caused by expansion of the cement paste fraction. It is shown that concretes and mortars made using a number of Portland cements can be induced to show abnormal expansion when subject to wet curing after a severe early cure and that the parameters influencing expansion are probably total sulfate as SO, magnesium oxide and alkali contents of the cement and cement fineness. Given our present understanding, it does not seem possible to eliminate the expansion by a suitable choice of portland cement composition. However, it is shown that limiting the concrete temperature at all times to below 70 C in precast concrete and limiting the cement content in in-situ concrete of large section are probably secure ways of avoiding late expansions associated with ‘delayed ettringite formation’ in concretes subject to prolonged wet or moist exposure.

The occurrence of premature concrete deterioration found in Texas Department of Transportation (TxDOT) precast concrete elements and other types of concrete structures found in Texas has prompted TxDOT to conduct an investigation into the cause of this deterioration. TxDOT’s investigation of 69 prestressed concrete box beams, of which 56 are exhibiting various degrees of deterioration, includes a historical documentation review, chemical analysis and petrographic examination with both optical and scanning electron microprob microscopy. The cement mill that supplied the cement for the 56 beams that are deteriorating conducted an independent investigation. An overview of both investigations is presented. The investigation conducted by TxDOT revealed distress associated with pasted expansion. Ettringite was found throughout the concrete samples in the voids, cracks, gaps around aggregate particles and concentrations or nests exclusively within the cement paste. The presence of alkali-silica reaction (ASR) was also observed but little if any distress could be conclusively attributed to the ASR. The cement manufacturer’s investigation concluded that the main cause of distress was ASR and that delayed ettringite formation occurred as a secondary mechanism most likely resulting from poor construction practices. The varying conclusions of TxDOT’s and the cement manufacturer’s investigations typify the polarization of industry nation wide if not world wide over similar case studies involving this phenomenon described herein as premature concrete deterioration. Accordingly, the conclusion of this paper addresses some of the ramifications of not attaining a resolution to this problem.