Pore structures surveyed so far include (1) air voids, typically tens or hundreds of um in size; (2) hollow shell pores, usually roughly 3 to 15 um sizes to sizes well below the resolution of backscatter SEM, and (4) extremely fine, ca. 10 nm-sized pores within inner hydration products. In this section a discussion is provided of presumably even finer `gel pores’ that have long been considered to be present in hardened cement pastes. The existence of gel pores in cement pastes was postulated by Powers and Brownyard26 primarily on the basis of water vapor adsorption isotherms measured after drying thin paste specimens over hydrated magnesium per- chlorate (`P-drying’); in each case they calculated BET surface areas from the adsorption isotherms. These experiments were carried out for cement pastes of different w:c ratios and of various degrees of hydration. Their results led Powers and Brownyard to infer that the product of hydration was a characteristic `cement gel’ of high surface area. The volume of water absorbed by this cement gel at high RH values was always found to be at least equivalent to four surface monolayers of condensed water. The spaces into which this 4-monolayer thick- ness of water was adsorbed were considered to constitute a set of characteristic `gel pores’ inherent in all normally cured (not steam cured) cement hydration products. Water adsorbed in excess of four monolayers was believed to occupy empty space outside the boundary of the cement gel, and was termed `capillary water’; the space occupied was termed `capillary pore space’. Powers and Brownyard went on to provide the classic gel pore/capillary pore model that has been almost universally employed ever since. These authors estimated gel pores to be ca. 1 nm in size, based on hydraulic radius considerations; they considered that such pores occupy ca. 28% of the volume of the cement gel. Gel pore sizes have been re-assigned variously by succeeding authors. Some years later Feldman and Sereda35,36 carried out extensive water vapor sorption studies in compacts of dry cement which had been hydrated after being pressed together. Rather than drying their specimens first, they initially established desorption isotherms from the wet state, drying to various end points before beginning adsorption measurement. Various drying and wetting scanning loops were measured, and simultaneously these authors measured a number of physical and dimensional changes taking place in the hydrated compacts. Their results led them to the conclusion that the ultimate structure of the C-S- H hydration products formed in cement hydration contained deformed nano- meter-scale layer structures with interlayer spaces initially occupied by water. The interlayer spaces were thought to be only one or at most a few water monolayers thick. Water vapor could desorb from these deformed layer structures readily, but the structures were said to collapse when dried below 11% RH, i.e. as had been done as a preliminary step in the studies of Powers and Brownyard.26 The collapsed layer structures were considered to be difficult or impossible to reopen in subsequent adsorption cycles, and Feldman and Sereda considered that water vapor surface areas measured by water vapor adsorption measurements after drying were incorrect. These concepts would seem to negate those of Powers and Brownyard,26 and indeed in their various publications Feldman and Sereda systematically declined to use the term `gel pores’; the collapsible interlayer spaces were not really considered by them to be `pores’ at all. Nevertheless, subsequent authors somehow have succeeded in fusing the two concepts. The cartoon originally published by Feldman and Sereda36 to illustrate their concepts has been republished ad infinitum, but usually with the Powers and Brownyard model and its associated calculation of gel porosity also invoked in the same treatment. It should also be remarked that these gel pores or collapsible interlayer structures, both thought to be 1 nm or less in size, are an order of magnitude smaller than the ca. 10 nm pores reported within inner hydration product by Richardson.30 Whatever the ultimate resolution, it seems that neither gel pores, nor alternatively, deformed nanometer-scale layer structures, are likely to be significantly involved in fluid transmission in concrete.
