THE PATCHY STRUCTURE OF CEMENT PASTE IN CONVENTIONAL CONCRETES

Transcription

1 THE PATCHY STRUCTURE OF CEMENT PASTE IN CONVENTIONAL CONCRETES Sidney Diamond School of Civil Engineering Purdue University West Lafayette, IN U.S.A. Abstract It has been found that the hardened cement paste (hcp) in some conventionally-mixed concretes examined in backscatter-mode scanning electron microscopy (SEM) contains areas or patches that are uniformly dense and easily distinguished from nearby areas or patches that are coarsely porous. Such distinct dense and porous patches have been found both in laboratory mixed concretes and in commercial ready-mix concretes installed in various structures. Some of the porous hcp patches may run parallel to and be adjacent to some aggregate surfaces, but nearby dense patches without visible porosity generally abut other aggregate surfaces. The porous patches may contain both hollow shells derived from Hadley grain formation and irregular and interconnected capillary pores. The area proportion of porous to dense patches appears to be progressively higher for concretes of progressively higher w:c ratios, and dense areas are only rarely encountered in concretes of w:c ratios of ca Introduction Examination of the microstructure of concrete by backscatter-mode SEM has been a useful technique in investigations of specific deterioration problems and in studying the details of its internal structure. Summaries and illustrations of the latter applications have recently been provided by the writer [1] and by Scrivener [2] in a special issue of the journal Cement and Concrete Composites devoted to scanning electron microscopy. Recently the present writer [3] reported that the microstructure of certain laboratory mortars that had been cited as confirming the NIST model of the ITZ structure in concrete did not appear to be consistent with that model. Instead of displaying interfacial transition zones (ITZs) of high porosity surrounding the sand grains, SEM examination indicated that the hardened cement paste (hcp) in which the sand grains were embedded was made up of two types of contrasting patches: visibly dense patches showing little detectable porosity and exhibiting high backscatter coefficients, (i.e. bright appearance) and porous darker patches showing open and seemingly interconnected pores of sizes up to 15 μm. Both dense patches and porous patches were found to indifferently occupy the traditional ITZ space around aggregates, and both types of hcp area extended into the so-called bulk paste. The boundaries between dense and porous patches were found to be quite sharp.

2 In the present work it is reported that similar dense patches and highly porous patches are found in conventionally mixed laboratory and field concretes. Thus the patchy microstructure reported in the previous paper is not an artifact of the unusual mixing procedure that was used in preparing the mortars examined in that report, but rather appears to be a common characteristic of conventionally-mixed concretes as well. 2. The patchy microstructure of laboratory-mixed concretes A conventional concrete mix of water:cement (w:c) ratio 0.50 was prepared for this investigation. The cement was an ASTM Type I cement of normal characteristics; the coarse aggregate was a crushed dolomitic limestone of 25 mm maximum size; the sand was a local river sand of heterogeneous character. No admixtures were used. Mixing was done in an open pan counter-current Lancaster Type SKC mixer of rated capacity of 42 L, with the batch sized appropriately for the mixer. The mixing sequence was conventional: coarse aggregate, sand, and cement were added in that order, and dry mixed for 2 minutes. The water was then added while the mixer continued in operation. Small samples of the fresh concrete were secured at various times during mixing. These were placed in small containers, consolidated on a vibrating table and then sealed. After 24 hours at room temperature the sealed containers were opened and exposed to a fog room atmosphere maintained at ca. 23 o C for 27 additional days. The 28 day-old samples were then demolded, and thin prisms were sawn from the center portions of each sample using a precision diamond saw and a non-aqueous lubricant. The thin specimens were then dried, epoxy-impregnated, and polished in the usual manner for backscatter SEM examination. Examination revealed the existence of the patchy microstructure as previously described for certain mortars in the earlier publication [3]. Fig. 1 shows an area between and just below two large sand grains, which are uniformly gray in the picture. A porous patch of hcp is located between the sand grains, and an area of dense hcp occurs below and to the left of this porous area. The boundary between them, indicated by the dotted white line, is fairly sharp. The porous patch is significantly darker, and appears to be more completely hydrated. The two large unhydrated grains present within it are both primarily C 2 S, and have no hydration rims. In contrast, the dense patch exhibits a higher gray level, and displays a number of partly hydrated cement grains with relatively narrow inner product hydration shells. Fig. 1 was taken from a concrete specimen secured after only 1 minute of mixing. However, it was found that the separate areas of dense and porous microstructure persisted even after as much as 30 minutes of continuous mixing, the details of which will be published elsewhere. Thus it appears that the patchy feature is not an indication of inadequate mixing time. 3. Laboratory concrete: the effect of water:cement ratio Examination of various concretes suggests, not surprisingly, that the patchy microstructure is strongly influenced by the w:c ratio of the concrete. The writer has examined a number of concretes of w:c ratio less than about 0.4, especially high performance concretes, and in these he has not found evidence of the patchy structure illustrated above. Instead, it appears that essentially the entire hcp in such concretes is typically dense, with the only detectable pores of any size being occasional hollow shells, i.e. the residue of Hadley grains.

3 In a recent publication Sahu et al. [4] reported on a new method for determining w:c ratio of mature concretes by backscatter-mode SEM. In this method, randomly selected small areas ( 109 μm by 109 μm) of the bulk hcp (taken well away from any aggregates) were imaged at 800x. A binary segmentation was used to delineate pore pixels, and the percentage of pore pixels in each sampled area was determined. It was found that if enough areas were averaged (ca. 40 images per sample) a good linear relationship was found between the overall average % of pore pixels and the w:c ratio of reference concretes of different w:c ratios, ranging from Fig. 1. Illustration of the distinction between dense and porous patches found in 28-day old laboratory mixed concrete. The boundary between the porous and dense areas is indicated by the dotted line to The reference concretes used by these authors were laboratory mixed and cured underwater for 28 days prior to examination. The set of images obtained for these concretes was kindly provided to the present writer by Sahu and co-workers. It appears that the images of w:c 0.45 concretes were essentially all

4 dense, and showed little detectable porosity. In concretes of w:c ratio higher than 0.45, distinct porous and dense microstructures could be seen to exist in different areas. The w:c 0.75 and 0.85 image sets showed mostly porous microstructure, but even the w:c 0.85 image set showed a few dense regions. Fig. 2 shows two images taken from the w:c 0.55 specimen showing the clear distinction between dense areas and porous areas in the same concrete. In addition to the obvious difference in the amount of visible pore space present, the gray level of the dense area imaged Fig. 2. Images selected from w:c 0.55 concrete to illustrate the existence of porous and dense areas, respectively, in different area of the concrete. on the right appears higher than the gray level of the porous area imaged on the left. This feature has been noted previously [3]. The average content of detectable pore space for the 40 small areas taken from this w:c 0.55 specimen was close to 3%. However, the individual values ranged from as little as 1% to as much as 7%. A histogram compiled from the original data showed a clear indication of bimodality, with a larger group centered at abut 1% porosity and a smaller group clustered around 5% porosity. The magnification used in the Sahu et al. study, 800x, is relatively high and the area depicted is correspondingly small. Accordingly, such magnification is not appropriate for mapping the sizes and areal extents of individual dense and porous patches. For that purpose the present writer has found that a much lower magnification, of the order of 100 to 150x, is generally more appropriate. Further, the contacts between the hcp and the adjacent aggregates is an important concern, and the wider areas examined at the lower magnification range usually includes a number of these as well. 4. Field concretes Illustrations are provided below of the co-existence of dense and porous patches in various field concretes examined by the writer and his colleagues. The images were secured from core

5 samples taken from slabs and shallow footings. Some contained fly ash. The concretes were all mixed in commercial batch plants and delivered by ready-mix trucks to various residential sites. The typical w:c ratio ranged between 0.6 and 0.8, although it is difficult to establish the exact mix design corresponding to any given core. The coarse aggregates are all dense, igneous rock, and the fine aggregates are all manufactured sands. Generally no air entraining agents or other chemical admixtures were used, and it appears that the concretes in question were cured only minimally, or not at all. The structures sampled were typically 8 to 10 years old and fully hydrated. The concrete specimens were prepared either as thin sections or as polished surface specimens and were dried, epoxy impregnated, and polished in the usual manner for backscatter SEM examination. The images shown below were all selected to display the extent and distribution of dense and porous patches in the portions of the concretes imaged. The magnification in all but the last image was kept uniform at 100x. Fig. 3 shows an area in which the central part of the field contains a strip of porous hcp interposed between dense hcp at the top and bottom of the field. Fig. 3. Field concrete showing elongated porous patch between dense areas of hcp above and below the porous patch.

6 The porous hcp area displayed is irregular and varies in width from less than 200μm to about 500 μm. It is easily distinguished from the dense areas above and below it. The large light-gray colored aggregate in the upper right part of the field is seen to be in contact with both dense and porous hcp along different surfaces. The upper surface and most of the left-hand surface of this aggregate abut the upper dense patch, with no sign of extra porosity (at this magnification) along the region of contact. The lower surface and the adjacent lower part of the left hand surface of the aggregate are clearly in contact with porous hcp. The other aggregates in the lower half of the field are similarly bounded by the two kinds of hcp on different surfaces. However, the smaller sand grains in the upper left corner are entirely bounded by dense hcp. Sometimes the individual dense and porous patches are smaller and more intimately intermingled. Fig. 4 shows an area in a specimen taken from a different concrete in which the patches of porous hcp are smaller than those in Fig. 3 and locally more intertwined with (but still distinguishable from) dense areas of hcp. In some areas there may be a tendency for the narrower porous patches to follow the outlines of adjacent aggregates, as seen near the bottom of Fig. 4, or almost surround small sand grains, as in the center of the field. Fig. 4. An area in a different field concrete showing intermingled porous and dense patch areas, with the porous patches running along some of the sand grain surfaces.

7 Fig. 5. Large area in a specimen from a concrete core that is entirely in porous hcp. The area depicted is about 900 μm x 900 μm. Fig. 6. Large area in the same concrete as Fig. 5 that is entirely in dense hcp. Microcracking is characteristically found in such areas.

8 On the other hand, in some concretes the sizes of individual dense and porous hcp patches are not intermingled, but very large areas are seen to be either one or the other. Figs. 5 and 6 are both taken from different areas of the same 50 mm concrete specimen. The total area imaged in Fig. 5, about 900 μm on a side, is entirely included within a porous hcp zone. In contrast, in Fig. 6, an area of the same size is entirely within a dense hcp zone. It will be noticed that some microcracking is evident in Fig. 6, but is absent from Fig. 5. This does not appear to be accidental; the writer and his colleagues have noted that a much greater tendency toward microcracking is often characteristic of the larger dense hcp patches. 5. The pores in porous patches Most of the images shown previously were taken at magnifications too low to see the details of the pores of the porous patches. However, the high-magnification left-hand image of Fig. 2, for a w:c 0.55 concrete, shows that most of the pores depicted there are primarily remnants of hollow shell hydration grains. This is not true of the porous patches found by the writer in the field concretes examined, most of which have substantially higher w:c ratios than Fig. 7, taken at high magnification, Fig. 7. Details of the pore spaces in a porous patch of a high w:c ratio field concrete. Several of the hollow shell grains present are encircled for easy recognition; however most of the pores are interconnected capillary pores.

9 shows an area that is reasonably typical of the characteristics of porous patches found in most of the concretes examined by the writer. Some of the pores are clearly hollow shells resulting from the Hadley grain hydration of smaller cement grains. However, most of the pores imaged in this porous patch are highly irregular in outline and do not have recognizable shells. They represent true capillary porosity i.e. remnants of originally water-filled space. As such, it is likely that they are highly interconnected in three dimensional space, and the porous patches in these concretes are zones of exceedingly high local permeability 6. Discussion All of the concretes examined in this work are conventional concretes of high w:c ratios, and so far as is known, none have been batched with superplasticizers. This places them at the other end of the spectrum from high performance concretes, which are usually batched with silica fume and superplasticizers, and at w:cm ratios usually less than 0.4. The writer has examined various concretes falling into the high performance concrete category, and has generally found that the hcp is entirely or almost entirely dense; detectable porous patches were negligible in extent or entirely absent. Published explanations for the favorable strength and impermeability characteristics of high performance concretes generally rest on the assumption that the quality of the ITZs surrounding the aggregates is responsible for these characteristics. It appears to the writer that a more likely explanation may be the virtual absence of bulk porous patches, rather than the character of the ITZs. The cause or causes of the differentiation of hcp in dense and porous areas in the higher w:c concretes examined here are certainly not obvious. In the earlier study on mortars [3] it was hypothesized that the cause may have been the unusual method of reciprocating mixing action used to prepare the specific mortars studied. This is obviously not the cause, since dense and porous patches are here seen to occur in both conventional laboratory mixed and conventional commercially-mixed concretes as well. It seems likely that the origin of discrete dense and porous patches may be traceable to the original fresh concrete. It would seem that the paste in conventional as-mixed fresh concrete of w:c greater than about 0.45 must ordinarily be composed of clusters or droplets of relatively low w:c and relatively high w:c, which are ineffectively homogenized in the mixing process. While conventional concrete mixing is successful in mixing cement paste and aggregate and coating all of the aggregate surfaces with a layer of fresh cement paste, insuring that the fresh paste itself is homogeneous on a fine scale may not be achievable, especially when extra water is present. 7. Conclusions Backscatter-mode SEM investigations of conventional laboratory-mixed and commercially mixed concretes of w:c higher than about 0.45 have confirmed the co-existence in them of distinctly dense and distinctly coarsely-porous patches of hardened cement paste. Individual aggregates may be entirely enclosed within dense patches, within porous patches, or may be in contact with dense patches on some surfaces and porous patches on others. Contacts with dense patches appears to be intimate, and lack obvious porous ITZs. The individual pores within the porous patches consist of both hollow shells produced by Hadley grain mode of hydration of smaller cement grains and of interconnected irregular capillary pores.

10 Acknowledgments The writer would like to express his great pleasure in being able to contribute to this special volume commemorating the 60 th birthday of his long-term friend and colleague, Professor Arnon Bentur. He is grateful to Sadananda Sahu for providing the collection of images of the reference concretes cited in Section 3 of this report, and for assistance in the examinations of of field concretes, much of which was carried out at the RJ Lee Group facility. References 1. Diamond, S., The microstructure of cement paste and concrete a visual primer, Cem. Concr. Composites 24 (2004) In press. 2. Scrivener, K. L., Backscattered electron imaging of cementitious microstructures: understanding and quantification Cem Concr Composites, 24 (2004) In press. 3. Diamond, S., 'Percolation due to overlapping ITZs in laboratory mortar? A microstructural evaluation, Cem. Concr. Res. 33 (7) (2003) Sahu, S, Badger, S., Thaulow, N, and Lee, R.J., Determination of water-cement ratio of hardened concrete by scanning electron microscopy, Cem. Concr. Composites. 24 (2004) In press.