Materials and Structures/Matériaux et Constructions, Vol. 35, March 2002, pp 80-84 SCIENTIFIC REPORTS Effects of microstructure on restrained autogenous shrinkage behavior in high strength concretes at early ages S. Igarashi and M. Kawamura Kanazawa University, Japan Paper received: March 12, 2001; Paper accepted: June 17, 2001 A B S T R A C T Fluorescence microscopic examinations were conducted to identify damages induced by restraining autogenous shrinkage. Characteristics of fluorescent areas and their correspondence to autogenous shrinkage behavior of high strength concretes were discussed. Silica fume concrete exhibited a greater creep potential when loaded at very early ages. The microstructure in sealed concretes with an extremely low water/binder ratio was porous. The vicinity of aggregate grains was more porous and weaker than the bulk matrix in sealed concretes. In addition, sealed silica fume concretes contained many unhydrated cement particles which were profiled by thin gaps between the core cement particles and the surrounding cement paste matrix. These features of microstructure were not observed in water ponded concretes. The detected fluorescent areas may be defects caused by selfdesiccation and autogenous shrinkage. The flaws had little effects on the development of strength. However, the presence of thin gaps around remnant cement particles may increase creep deformation to relieve internal stresses. R É S U M É Des analyses microscopiques fluorescentes ont été réalisées pour identifier les dommages causés par le retrait endogène. Les caractéristiques des surfaces fluorescentes et leurs liens avec le comportement du retrait endogène du béton à haute résistance, ont été débattus. Le béton fait de fumée de silice a montré une plus grande possibilité de fluage en étant chargé à de très jeunes âges. La microstructure dans les bétons obstrués ayant un rapport eau/liant extrêmement bas, était poreuse. Les environs des grains d agrégats étaient plus poreux et plus faibles que le volume du modèle des bétons obstrués. De plus, le béton obstrué à base de fumée de silice contenait plusieurs particules de ciment déshydraté qui ont été démontrées par les minces fentes entre le centre des particules de ciment et le contour de la pâte de ciment. Ces faces de la microstructure n ont pas été observées dans le béton submergé. Les surfaces fluorescentes détectées peuvent être défectueuses à cause de l auto-dessiccation et du retrait endogène. Les défauts ont peu d effets sur le développement de la résistance. De toutes façons, la présence de minces fentes autour des restes de particules de ciment peut augmenter la déformation du fluage et libérer les contraintes internes. 1. INTRODUCTION Autogenous shrinkage is a volume change which can be quite high in high strength concrete at early ages. The restraint against autogenous shrinkage can induce cracking since the strength of concrete at early ages is not high enough to resist stresses caused by the external and internal restraint [1]. Cracking and generation of stress are significantly influenced by the growth of internal solid structure at early ages. However, there are few studies to relate features of microstructure at early ages to the restrained shrinkage behavior in high strength concrete. In this study, the fluorescence microscopic examinations were conducted to identify damages induced by self-desiccation and restraining autogenous shrinkage. Characteristics of fluorescent areas detected in high strength concrete at early ages were discussed in relation to the viscoelastic behavior evaluated by the closed loop uniaxial restrained shrinkage testing apparatus. Effects of water ponding and silica fume on the shrinkage behavior were also discussed with emphasis on changes in microstructure observed by the fluorescence microscopy and microhardness measurements. Editorial Note Dr. Shin-ichi Igarashi and Prof. Dr. Mitsunori Kawamura are RILEM Senior Members. They participate in the work of RILEM TC 181-EAS: Early age shrinkage induced stresses and cracking in cementitious systems. 1359-5997/02 RILEM 80
Igarashi, Kawamura 2. EXPERIMENTAL 2.1 Materials and mix proportion of concretes The cement used was Ordinary Portland cement. A silica fume with specific surface area of 20 m 2 /g was used. The replacement of silica fume was 10%. River gravel with a maximum size of 10 mm was used as a coarse aggregate. The fine aggregate used was a river sand. Polycalboxylic acid type superplasticizer was used. Water/binder ratio of concretes was 0.25. Mix proportions of concrete are given in Table 1. Table 1 Mix proportion of concretes W/B Water Cement Silica Sand Gravel SP Fume (%wt. of binder) PC 0.25 145 581 0 559 1086 1.7 SF 0.25 142 510 57 559 1086 2.6 2.2 Restrained and free shrinkage tests The restrained and free shrinkage tests were carried out by the use of sealed specimens at 18 C. Concretes were directly cast into the molds of the restrained and free shrinkage testing apparatus (Fig. 1) [2]. This system, developed by Kovler [2] can be used to determine free shrinkage, restraining stresses and visco-elastic responses using the procedures outlined in [2]. The specimens were sealed immediately after casting. As for the water ponded specimens, the top surfaces of the specimens were fully covered by thick sponges saturated with water. Water was supplied everyday to ensure the saturated condition. The testing apparatus consisted of two identical specimens and the measuring devices. In the restrained shrinkage specimen, when shrinkage occurs and its level is beyond a strain of 10 10-6, the servo motor automatically starts working to pull the specimen back to the initial position. Length changes of both specimens and the restraining load generated by the restrained shrinkage specimen were continuously recorded. The restraint was given to the specimen after allowing to shrink for the first 12 hours. The test was conducted for 7 days after casting. 2.3 Strength tests Cylinder specimens of 50 mm in diameter by 100 mm in height were produced. They were cured under the same condition as the shrinkage tests. Splitting tensile and compressive strength tests were carried out at the ages of 1, 3 and 7 days. The specimens after the shrinkage tests were also used to obtain the splitting tensile strength. 2.4 Fluorescence microscopic examinations Slices with about 10 mm thickness were cut out from the middle of the specimens for the shrinkage test after the completion of shrinkage tests. The slices were immersed in ethanol, and then impregnated with the epoxy resin containing a fluorescence dye. After the resin hardened at a room temperature, the slices were polished with silicon carbide papers for the fluorescence microscopic examinations. 2.5 Microhardness measurements Specimens containing 10% of volume fraction of only coarse aggregate were produced. They were cured under the same conditions as the shrinkage tests. At the age of 7 days, slices were cut out from the specimens. The surfaces of the slices were polished. A microhardness tester with a Vickers indenter (Load = 0.1 N) was used to measure microhardness in the cement paste phase around aggregate grains. 3. RESULTS 3.1 Autogenous shrinkage behavior Fig. 2 shows free autogenous shrinkage for concretes with and without silica fume. There was little difference in free autogenous shrinkage strain between both. Autogenous shrinkage strain increased rapidly during the first 24 hours. However, thereafter, the rate of shrinkage was reduced. Fig. 3 shows the development of restraining stress Fig. 1 Restrained shrinkage testing apparatus. Fig. 2 Free autogenous shrinkage strain of high strength concretes. 81
Materials and Structures/Matériaux et Constructions, Vol. 35, March 2002 Fig. 3 Development of restraining stress. Fig. 5 Specific creep of restrained specimens. Fig. 4 Creep strain in restrained specimens. Fig. 6 Swelling strain in water ponded concretes. with time. The restraining stress generated in the silica fume-free concrete was greater than that in the silica fume concrete whereas free shrinkage strain was almost the same. It is found from Figs. 2 and 3 that the restraining stress cannot be evaluated from only the absolute values of free autogenous shrinkage. Fig. 4 shows increase of creep strains with time in the restrained specimens. Little differences in the strain were found between concretes with and without silica fume. Comparing Fig. 2 with Fig. 4, it is found that most of free autogenous shrinkage was compensated by creep. It appears that little difference in the shrinkage and the creep strains between concretes with and without silica fume (Figs. 2 and 4) are different from the results obtained by one of the present authors [3, 4]. It was reported in those studies [3, 4] that silica fume increased the free autogenous shrinkage and the total creep strains at an extremely low water/binder ratio. Effects of silica fume on the autogenous shrinkage in concrete may depend on the property of cement paste matrix and on other experimental factors such as water/binder ratios, curing temperatures and coarse aggregate content. Further investigation is necessary to understand the effect of silica fume on the free autogenous shrinkage in concrete. However, as shown in Fig. 5, there is a significant difference in the specific creep which is obtained by normalizing the total creep strains by the total restraining stress. Silica fume concrete exhibited a greater specific creep than the concretes without silica fume even if there was little difference in the total autogenous shrinkage and creep strain. The variations of the specific creep with time were relatively small after the first 24 hours. The specific creep of concretes slightly decreased with time. Autogenous shrinkage is usually explained by selfdesiccation giving rise to capillary tension. In order to avoid the risk of early age cracking due to self-desiccation, it was pointed out that water ponding at early ages was useful for reducing the adverse effects of self-desiccation [5]. Fig. 6 presents free autogenous shrinkage of the concretes which were continuously given sufficient water immediately after casting. Contrary to sealed concretes, they exhibited rapid swelling for the first 24 hours. Swelling strains in the silica fume-free concrete was greater than the silica fume concrete. Thereafter, however, they started to shrink so that differences in swelling strain between both decreased with time. It is confirmed from Fig. 6 that water ponding is effective to avoid autogenous shrinkage and the generation of restraining tensile stresses. 3.2 Fluorescence microscopic examinations Fig. 7 shows the fluorescence micrographs for the polished surfaces of concretes without silica fume. Regardless of the condition of restraint, the whole mortar matrix in concretes cured under a sealed condition was found to be quite bright from the fluorescence microscopic examinations for concrete specimens. Furthermore, the vicinity of aggregate grains was brighter than the bulk cement paste matrix. The brighter areas in the fluorescence micrographs indicate the more porous areas. It is found that the cement paste matrix in concretes cured without supply of water from external sources, especially in the vicinity of aggregate grains in the sealed concrete was so porous in nature. On the 82
Igarashi, Kawamura Fig. 7 Fluorescence micrographs for concretes without silica fume: (a) Sealed (free), (b) Sealed (restrained), (c) Water ponded. Fig. 8 Fluorescence micrographs for silica fume concretes: (a) Sealed (restrained), (b) Sealed (restrained) [taken under long exposure time], (c) Water ponded. contrary, the matrix in the water ponded concrete specimens was much darker than that in sealed concretes. Water ponding at very early ages resulted in the formation of dense microstructure. The fluorescence micrographs for silica fume concretes are given in Fig. 8. It is found that silica fume concretes had dense microstructures compared to the concrete without silica fume. However, as shown in Fig. 8(b), brighter regions than bulk matrix were also observed around aggregate grains in the sealed silica fume concretes. It should be noted that there are many remnant cores of cement particles of which peripheries were profiled by thin porous areas in the sealed silica fume concretes. They were distributed in the whole cement matrix. However, the bright peripheral regions of the cement particles were not observed in concretes in which enough water was supplied at the initial curing period (Fig. 8(c)). Taking into account that the bright fluorescent regions around aggregate grains and the bright peripheral regions around remnant cement particles were not found in the ponded concretes, the formation of such porous areas could be attributed to selfdesiccation in silica fume concretes with an extremely low water/binder ratio. Fig. 9 Microhardness distribution in the vicinity of aggregate. 3.3 Microhardness distribution patterns around aggregate grains Measured microhardness values around aggregate grains are plotted in Fig. 9. In the sealed specimens, values of microhardness in the vicinity of aggregate were smaller than in the bulk cement paste matrix. It is found from the microhardness measurements that weak microstructures are formed around aggregate grains in the sealed specimens with an extremely low water/binder ratio. However, such weak regions around aggregate grains were not present in the ponded concrete specimens. 4. DISCUSSION 4.1 Damages induced by self-desiccation and restraining stresses As shown in Fig. 3, internal stresses are generated by the restraint against autogenous shrinkage. Generally, it is considered that microcracks occur if those stresses exceed the tensile strength. However, as shown in Fig. 7, visible cracks were not observed under the fluorescence microscope. Ratios of the restraining stress to tensile strength were always less than 0.20, as shown in Fig. 10. The ratios are found to be far less than the critical values above which microcracks are formed. Therefore, no serious damages by the restraining stress were induced in the concretes in this study. Splitting tensile strengths after the restrained shrinkage tests are given in Table 2. Only a little decrease in the strength was found for 83
Materials and Structures/Matériaux et Constructions, Vol. 35, March 2002 Fig. 10 Ratios of the restraining stress / splitting tensile strength. Table 2 Splitting tensile strength of concretes after the shrinkage test (N/mm 2 ) Free Specimen Restrained Specimen PC 4.99 4.24 SF 5.03 4.45 concretes subjected to the restraining stress. On the other hands, as shown in Fig. 7, sealed concretes had porous microstructures compared to the water ponded specimens. Such a porous microstructure resulted from the lack in water at the early ages. Insufficient supply of water must have interrupted further hydration of cement. Generally, the interfacial transition zone is considered as a weak region around aggregate. However, the weak region was not detected in the water ponded concrete. Furthermore, taking into account that the concretes with an extremely low water/binder ratio contain silica fume, the bright areas around aggregate cannot be attributed to the formation of the interfacial transition zone. Therefore, the bright regions around aggregate suggest the presence of invisible microcracks. Such damages may be induced by the restraint of rigid aggregate grains against autogenous shrinkage of the surrounding matrix. In silica fume-bearing concretes, the whole mortar matrix was darker even in the sealed specimens. This fact suggests that the dense microstructure has already been formed in the silica fume-bearing concretes during early ages. 4.2 Creep of silica fume concrete at early ages It is well known that the mature high strength concrete with silica fume exhibits smaller creep deformation than the concrete without silica fume. This lower creep potential in silica fume concrete results from the higher strength and the dense microstructure at the time of loading. However, in this study, the specific creep of silica fume concrete was greater than the concrete without silica fume. The greater creep deformation under tensile loading was also reported by Bissonnette and Pigeon [6]. From Fig. 8, it can be seen that silica fume concretes at early ages had many bright fluorescence areas along the peripheries of remnant cement particles over the cement paste matrix. The SEM examinations at a higher magnification revealed that there were small gaps between the core cement particles and the surrounding cement paste matrix. They were very similar to hollow shell hydration of cement particles [7]. Greater chemical shrinkage due to the pozzolanic reaction of silica fume and the acceleration of hydration of cement under the environment with less water available may be related to the insufficient filling around remnant cement particles. Unhydrated cement particles function as inclusions in the cement paste matrix at microscopic scale. However, those cement particles separated from the matrix by gaps must be less effective for confining the tensile creep deformation of the cement paste matrix. Furthermore, silica fume reacts with Ca(OH) 2 at early stages of cement hydration so that relatively large amounts of colloidal CSH gel is produced at early ages [8]. Namely, the incorporation of silica fume at extremely low water/binder ratios leads to the formation of much CSH gel as a source of creep deformation. However, functions of unhydrated cement particles as deformation arresters are less as stated above. The greater creep potential to relieve the restraining stress in silica fume concrete at early ages reflects the characteristics of microstructure formed in concretes with extremely low water/binder ratios. 5. CONCLUSIONS (1) Sealed concretes were more porous than the concretes cured in water. (2) Water ponding at early ages was effective to prohibit autogenous shrinkage of high strength concretes. Dense microstructure was formed in the water ponded concrete. (3) Weak regions were formed around aggregate grains in sealed concretes. The restraint provided by rigid aggregate grains against autogenous shrinkage caused invisible microcracks in the vicinity of aggregate grains. (4) There were many remnant core cement particles surrounded by fine gaps in silica fume concrete at early ages. (5) Less effective function of the cement particles with the gaps as an inclusion may account for the greater creep potential of silica fume concrete at early ages. REFERENCES [1] Paillère, A. M., Buil, M. and Serrano, J. J., Effect of fiber addition on the autogenous shrinkage of silica fume concrete, ACI Mat. J. 86 (2) (1989) 139-44. [2] Kovler, K., Testing system for determining the mechanical behavior of early-age concrete under restrained and free uniaxial shrinkage, Mater. Struct. 27 (170) (1994) 324-30. [3] Igarashi, S., Bentur, A. and Kovler, K., Autogenous shrinkage and induced restraining stresses in high strength concrete, Cem. Concr. Res. 30 (11) (2000) 1701-1707. [4] Kovler, K., Igarashi, S. and Bentur, A., Tensile creep behavior of high strength concretes at early ages, Mater. Struct. 32 (219) (1999) 383-87. [5] Bentur, A., Igarashi, S. and Kovler, K., Control of autogenous shrinkage stresses and cracking in high strength concretes, in Utilization of High Strength /High Performance Concrete, Proc. of 5th Intl. Symp., Sandefjord, (June 1999) 1017-26. [6] Bissonnette, B. and Pigeon, M., Tensile creep at early ages of ordinary, silica fume and fiber reinforced concretes, Cem. Concr. Res. 25 (5) (1995) 1075-85. [7] Diamond, S., Aspects of concrete porosity revisited, Cem. Concr. Res. 29 (8) (1999) 1181-88. [8] Wu, Z.-Q. and Young, J. F., The hydration of tricalcium silicate in the presence of colloidal silica, J. Mat. Sci. 19 (1984) 3477-86. 84