The formation of secondary ettringite within the entrained air voids reflects a relatively high degree of concrete saturation causing the smaller voids to be filled with pore solution when the concrete freezes. Freeze-concentration of dissolved solids in the pore solution and temperature changes may result in a condition contributory to precipitation of secondary ettringite. Over time, the cyclic freezing and thawing accompanied by secondary ettringite precipitation would result in filling of some of the entrained air voids.
The source of material for ettringite formation is probably from dissolution of calcium monosulfoaluminate within the cement paste. The dissolution of these phases (less than 10 % by volume) and subsequent precipitation as ettringite in the air voids may increase permeability of the paste facilitating the deterioration. Calculations on potential volume of ettringite based upon mix design and cement composition estimates indicate that more than sufficient ettringite (8 % to 10%) may be produced as a result of cement hydration to fill the entrained air void system. Therefore, losses of 2 % and less, by volume of the entrained air void system due to secondary ettringite formation, should not be considered to indicate excessive ettringite formation as proposed in some studies .
The occurrence of ettringite in concretes is not new, nor unique, and in many cases its presence in voids is attributed to the specimen being from "old concrete". Terzaghi  observed secondary ettringite crystallized in air voids in cores extracted from a gate structure of a shipway in Virginia. She considered this to reflect the presence of seawater in the pores of the concrete that provided the conditions necessary for formation of ettringite. Idorn  noted that ettringite is a common occurrence in sound, older concretes and will often be found to accumulate in voids and cracks. Trägårdh and Lagerblad  examined concretes damaged by a combined alkali-silica reactivity and freeze-thaw attack. They commonly observed calcium hydroxide and ettringite in air voids and attributed their occurrence to dissolution and leaching caused by water movements during recurrent cycles of freezing and thawing. The dissolved constituents precipitate in cracks and air voids as secondary products.
Taylor  examined microstructural and chemical aspects of sulfate reactions in concrete and concluded that the ettringite occurring in cracks and voids in both sound and deteriorated concretes is not necessarily the cause of the cracking but a product of recrystallization in a pre-existing crack or void of some other origin. He states that in early stages of hydration, ettringite is replaced by monosulfate as a result of continued hydration of aluminate. Ettringite is frequently observed in mature concretes, whereas monosulfate is not. Studies using backscattered electron imaging and X-ray diffraction (XRD) at NIST do find monosulfate in mature cement pastes but the XRD data do not exhibit very strong diffraction peaks. This may indicate a poorly-crystalline monosulfate, or changes due to dehydration or relative humidity during powder preparation. Taylor  feels that the most likely cause for the presence of ettringite in concretes is from carbonation, resulting in the monosulfate being replaced by a mixture of ettringite and hemicarbonate. He considered the carbonate could come from small amounts of calcite in the gypsum and notes that only 2.4 % (by mass of monosulfate) CO2 is necessary.
Atkins et al.  performed solubility measurements of ettringite (C6A3 H32 ), hydrogarnet (C 3AH6), monosulfate (C4AH12 ), and tetracalcium aluminate hydrate (C4AH13 ) and found that dissolution of pure monosulfate lead to the precipitation of ettringite with sucessive dispersions. These findings were in agreement with those of Brown  that the tetra-calcium aluminate hydrates are thermodynamically metastable relative to C3AH 6 and ettringite. C4 AH13 was also found to be metastable relative to C3AH 6.
Ludwig, Stark, and Eckart [37, 38, 39, 40] exposed a suspension of monosulfate in water to freezing and thawing cycles and found monosulfate will transform to ettringite. They considered a change in thermodynamic stability at low temperatures to account for the transformations where formation of ettringite is favored and felt the sulfate required for this reaction was derived through partial decomposition of the monosulfate. They also found that freeze-thaw cycling in 3 % NaCl solution resulted in transformation of monosulfate to Friedelís salt and ettringite. In these experiments, they observed no indications of carbonation, and thought the SO3 required for ettringite formation may come from the conversion of monosulfate to monochloride. The apparent uniform distribution of Friedel's salt in the Iowa cores would indicate, for these pavements, that either the chloride was present upon mixing or, less likely, that they have been permeated by a road salt solution.
Skalny et al.  discuss ettringite formation in concrete as a consequence of an aging via Ostwald ripening. They consider a high moisture content to facilitate this process of dissolution, transport, and recrystallization into the available empty spaces and, that this mode of ettringite formation is non-expansive. Neville  also recognizes the role of excess moisture in this process as hydration products may only be formed in water. The formation of ettringite in the air voids therefore must indicate they are at least partially filled with water.
Detwiler and Powers-Couche  examined the effects of sulfates in concrete on its resistance to freezing and thawing through a series of laboratory freeze-thaw tests using both 3 % NaCl and 3 % NaCl and gypsum solutions. They concluded that the freeze-thaw damage precedes the filling of voids with ettringite with deposition occurring in cracks and voids after damage occurred. These data do not explain the occurrence of filled voids in concretes from this study where filled voids may be observed in concretes exhibiting little or no damage, but they do show how easily the formation of secondary ettringite in voids can occur in concretes exposed to cyclic freezing and thawing.