The present writer has recently attempted to call attention to a feature observed in the microstructure of many concretes that appears to have been generally overlooked,46±48 although it had been mentioned previously.43 Some of the cement paste in many archetypical concretes consists of distinct patches or local areas that are visibly highly porous and contain few if any large cement grains; these are intermingled with other areas locally rich in large cement grains and exhibiting only limited pore space. An example is provided as Fig. 2.10, taken from a 28-day old laboratory mixed w:c 0.45 concrete. Such areas (or `patches’) of sharply differing local porosity were found by the writer46 in laboratory mortars prepared so as to duplicate those studied by Winslow et al.,49 but not examined microscopically by them. MIP results from these mortars had previously been interpreted as indicating percolation of ITZs for mortars of high sand content. The present writer46 suggested that in fact percolation of porous patches, rather than of ITZs, might explain the earlier MIP results. The sizes of the individual porous patches vary, but are often of the order of 200 um across; a roughly spherical patch of such dimensions in concrete would incorporate a paste volume of ca. 4 x 10^6um.
In many cases the boundaries between adjacent dense and porous patches are seen to be surprisingly sharp. Figure 2.11 shows such a boundary, in a mortar specimen hydrated in limewater for about 7 years. The mortar depicted was actually one of those originally prepared by Winslow et al.;49 part of it was sampled at 28 days for examination in MIP, and the remainder was stored in limewater. In 2003 the mortar was again sampled and examined in SEM.46 It is seen that, while long-term hydration products have infilled the dense area to the right leaving few visible pores, the porous patch to the left has remained very noticeably porous despite the long underwater storage.
While no quantitative study has been made, in the writer’s experience the proportion of the porous patches is clearly higher for concretes with higher w:c ratios. For concretes of w:c ratios around 0.4, i.e. at the lower end of the archetypical concrete w:c ratio range, porous patches are relatively few in number and tend to be generally isolated from each other. In contrast, in many very high w:c ratio concretes examined by the writer, porous areas tend to predominate and appear to `percolate’; local areas of dense paste, similar in appearance to the dense area in the right side of Fig. 2.11, are also found, but appear to be isolated from each other.
Especially in lower w:c ratio concretes, many of the porous patches border on sand grains, as in the example shown in Fig. 2.10; however some apparently do not. Porous patches that border on sand grains extend far beyond even the widest estimate of the extension of conventional ITZs. It should be noted that sizes of such porous patches are of the same general order of magnitude as the spacings between many sand grains in concretes. It should be mentioned that recently Wong and Buenfeld50 questioned the existence of dense and porous patches, attributing their appearance in back- scatter SEM to an artifact associated with incomplete penetration of the epoxy resin used in backscatter SEM. The present writer51 and Diamond and Thaulow52 have demonstrated that this suggestion was incorrect. While the existence of complementary porous and dense patches have been observed by the present writer primarily in backscatter SEM, they are also readily observable in secondary electron SEM examination of polished surfaces, and in fluorescent optical microscopy of thin sections.
Furthermore, several years ago, Landis53 prepared a conventionally mixed w:c 0.6 mortar using a fine concrete sand (maximum size of about 0.4 mm). A portion of the mortar was cast as a 4 mm diameter cylinder and moist cured for about 30 days, and then air-dried for a like period. A small chip was broken off and exposed to a synchrotron beam line at the National Synchrotron Light Source at Brookhaven National Laboratory, so as to generate data for computed tomography imaging. The results were kindly provided to the present writer in the form of a series of approximately 500 successive resolved images, each 1.2 um per pixel. Figure 2.12 shows one of these slices. The larger, mostly uniform gray features are sand grains. The individual smaller white particles are clearly residual unhydrated cement cores, surrounded by groundmass. It is seen that the lower left corner of the mortar chip (marked `A’) constitutes a dense patch containing closely-spaced unhydrated cement grain cores. Above and slightly to the right of this patch, and sharply delineated from it, is a region of more visibly porous paste containing almost no unhydrated cement. This delineation of a dense patch ± porous patch structure in Landis’s mortar specimen is of particular interest since the examination involved no specimen preparation, and thus no specimen preparation artifacts are possible. So far as the writer knows, perhaps the first clear published recognition of the existence and potential importance of such local patches in field concrete was provided by Idorn,54 on the basis of observations made in fluorescent thin section examinations of a deteriorating marine concrete. The concrete appeared to have an overall w:c ratio of ca. 0.45, but Idorn observed distinct porous and dense patches of wildly different apparent local w:c ratios, as indicated by local intensity of the fluorescence. He estimated that the porous patches exhibited local w:c ratios tending toward 1.0; in contrast, the dense patches appeared to exhibit local w:c ratios of only about 0.20. Idorn54 considered that the patchy structure derived from porous and dense `micropatches’ that had previously existed in the fresh concrete. Close examination of backscatter SEM images of the freshly mixed mortars prepared by Kjellsen,27 one of which was shown as Fig. 2.2, appears to support this idea. Examination of Fig. 2.2 suggests the grouping of large cement grains into local areas, leaving what appear to be other areas of higher local water content that are almost devoid of such cement grains. Such considerations imply the possibility that the local patches may simply be due to inadequate mixing. However, in a trial to specifically investigate this hypothesis, it was found48 that long-continued mixing of fresh concrete in an efficient pan mixer apparently did not succeed in eliminating the local patches. Furthermore, a recent trial55 indicated that, surprisingly, complete dispersion of fresh concrete induced by an extremely heavy dose of superplasticizer did not eliminate them either. The possible implications of these findings with respect to concrete permea- tion capacity and durability concerns remain to be established. Idorn54 specific- ally attributed the deterioration of the marine concrete he examined to the easy entry of external ions through the porous patches. The degree to which local porous patches exist and might connect in the three dimensional structure of ordinary concretes is certainly worth exploring.
