- Email: [email protected]
Volumetric deformation of gap-graded blended cement pastes with large amount of supplementary cementitious materials
Construction and Building Materials 54 (2014) 339–347
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Volumetric deformation of gap-graded blended cement pastes with large amount of supplementary cementitious materials Tongsheng Zhang, Peng Gao, Ruifeng Luo, Jiangxiong Wei, Qijun Yu ⇑ School of Materials Science and Engineering, South China University of Technology, 510640 Guangzhou, People’s Republic of China
h i g h l i g h t s Gap-graded blended cement pastes presented comparable chemical shrinkage with Portland cement paste. Most of chemical shrinkage occurred after the formation of skeleton structure, leading to reduced autogenous shrinkage. The microstructure of gap-graded blended cement paste was dense and homogeneous. Shrinkage stress was small and uniformly distributed in gap-graded blended cement paste. Gap-graded blended cement paste presented smaller volumetric deformation and superior resistance to cracking.
a r t i c l e
i n f o
Article history: Received 11 October 2013 Received in revised form 18 December 2013 Accepted 19 December 2013 Available online 19 January 2014 Keywords: Gap-graded blended cement Volumetric deformation Chemical shrinkage Hydration products
a b s t r a c t In this study, volumetric deformation of gap-graded blended cement pastes with large amount of supplementary cementitous materials was investigated and compared with those of Portland cement and reference blended cement pastes. The results show that the gap-graded blended cement pastes presented a higher initial packing density, therefore smaller amount of hydration products was needed to achieve dense microstructure, resulting in smaller chemical shrinkage. Notably, most of chemical shrinkage of gap-graded blended cements was occurred after the formation of skeleton structure, thus the autogenous shrinkage of gap-graded blended cements was reduced significantly due to restraint of the skeleton structure. Further, the microstructure of gap-graded blended cement pastes was dense and homogeneous due to grain size refinement and significant hydration of GBFS, and shrinkage stress was small and uniformly distributed. As a result, gap-graded blended cement pastes presented smaller volumetric deformation and superior resistance to cracking than Portland cement and reference cement pastes. Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.
1. Introduction The durability of concrete structures is increasing concerned during the past decades, and the early volumetric deformation of cement paste is generally considered as the key factor leading to early age cracking and, consequently, to the loss of durability of concrete structures. Tazawa and Bentz reported that finer cement usually resulted in larger autogenous shrinkage of cement-based materials, thus coarse cement was recommended in concrete manufacture if the mechanical properties of the concrete met the requirements [1,2]. Numerous literatures proved that the autogenous shrinkage of cement pastes with supplementary cementitious material (SCM) was significantly influenced by the hydraulic activity (or the type) and fineness of the incorporated SCMs [3–6]. For instance, the addition of silica fume increased the autogenous shrinkage of cement pastes remarkably [7]. When the specific ⇑ Corresponding author. Tel./fax: +86 020 87114233. E-mail address: [email protected] (Q. Yu).
surface area of granulated blast furnace slag (GBFS) was lower than 400 m2/kg, the addition of GBFS decreased the autogenous shrinkage of cement pastes. In contrast, for GBFS with specific surface area higher than 400 m2/kg, the autogenous shrinkage of cement pastes increased with the increase of GBFS replacement [8]. Low calcium fly ash presented low hydraulic activity, therefore the addition of low calcium fly ash generally reduced the autogenous shrinkage of cement pastes [9]. From the above literatures, it can be concluded that the volumetric deformation of cement-based materials mainly depends on the hydration process of cementitious materials and the composition of hydration products. Burrows and Ba proved that fast hydration of cementitious materials resulted in larger volumetric deformation, leading to a higher risk of cracking of cement-based materials [10,11]. Therefore optimization of hydration process of cementitious materials is an effective way to minimize the volumetric deformation of cement-based materials. Particle size distribution (PSD) has a significant influence on initial packing density and hydration process of cement paste,
0950-0618/$ - see front matter Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.conbuildmat.2013.12.053
340
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
eventually affecting the amount and composition of hydration products in cement paste. If cement paste had a higher initial packing density, smaller the amount of hydration products were needed to fill-up the voids in cement paste, therefore the volumetric deformation of cement paste would be improved dramatically. In previous study, a gap-graded PSD was proposed based on close packing theory, the initial packing density of blended cement paste was increased significantly, as voids between large size particles were filled in by fine particles grade by grade [12]. And the hydration process of gap-graded blended cements was optimized by arranging SCMs with high activity, cement clinker, and SCMs with low activity (or inert fillers) in fine, middle size and coarse fractions, respectively. The microstructure of gap-graded blended cement pastes was densified gradually at later ages due to hydration of GBFS. As a result, gap-graded blended cements with only 25% clinker by mass presented comparable mechanical properties with pure Portland cement [13,14]. Although with superior mechanical properties, it is more important to clarify the volumetric deformation of gap-graded blended cement pastes before application, as the durability of concrete structures is largely depended on the volumetric deformation of cement paste incorporated. In the present study, the volumetric deformation of gap-graded blended cement pastes, Portland cement paste and reference blended cement paste were comparatively studied, and the volumetric deformation mechanism of gap-graded blended cement pastes was discussed from the viewpoint of composition of hydration products and microstructure of hardened paste. The results will be very useful to improve the volumetric deformation of cement-based materials by reasonable utilization of cement clinker and SCMs. 2. Experimental procedures 2.1. Raw materials The chemical compositions of raw materials used in the experiment are given in Table 1. Cement clinker, GBFS, and low calcium fly ash (a Class F fly ash according to ASTM C 618 [15]) were ground and then classified by an air classifier. The PSDs of cementitious material fractions required by the gap-graded PSD are given in Fig. 1. 2.2. Preparation of gap-graded blended cements Gap-graded blended cements (BCF and BBCFF) were prepared by mixing cementitious material fractions homogeneously according to the mix proportions listed in Table 2, while reference cement and Portland cement were prepared by co-grinding the mixture. The Blaine specific surface areas of these two cements were controlled to be in the range of 350–360 m2/kg, which is seen to be equal to those of the gapgraded blended cements approximately. Fig. 2 shows that gap-graded blended cements presented wider PSDs than cements prepared by co-grinding (reference cement and Portland cement), which will leads to a higher initial packing density of cement paste. Although gap-graded blended cements had same mix proportion with reference cement, however, it should be noted that GBFS, cement clinker and fly ash were placed in the fine, middle size, and coarse fractions of gap-graded blended cements, respectively.
Fig. 1. Particle size distributions of cementitious material fractions used in the experiment.
procedures specified in Zhang’s research [13], to characterize the initial packing density of cement paste. In addition, water requirement for normal consistency of blended cements were determined according to EN 196-3 [16]. As the water requirements for normal consistency of Portland cement, reference cement and gap-graded blended cements were quite different from each other as shown in Table 3, therefore cement pastes and mortar with same fluidity were used for volumetric deformation measurements, in consideration that cement-based materials were mainly used under equal workability (fluidity). 2.3.2. Chemical shrinkage of blended cements Chemical shrinkage of the blended cements prepared, typical Portland cement and GBFS fractions was measured by volumetric method at 20 ± 1 °C, details of the method were specified in the literature [17]. To simulate the hydration of GBFS in blended cement, a mixture of 90% GBFS and 10% CaO by mass were mixed, and 0.2 mol/L NaOH solution was used as a simulated pore solution in chemical shrinkage measurement of GBFS [18,19]. Along with chemical shrinkage measurement, the hydration degrees of Portland cement and GBFS fractions was also determined by non-evaporation water method [20] and ethylene diamine tetraacetic acid disodium salt (EDTA) preferential solving method [18], respectively. 2.3.3. Autogenous shrinkage of cement pastes The autogenous shrinkage of blended cement pastes of normal consistency was measured using non-contact corrugated tube method at a 20 ± 1 °C environmental chamber [21]. Three parallel samples were performed and the average value of results was used as the autogenous shrinkage of cement pastes. 2.3.4. Restrained shrinkage cracking of gap-graded blended cement pastes Cement paste of normal consistency was cast into an ellipse-ring shaped mold, and a stainless steel ellipse in the center of the mold was used to restrain the shrinkage of the surrounding cement paste [22]. After being cured at 20 ± 1 °C and 90% relative humidity (RH) for 24 h, the outer molds of ring specimen were demolded, the top surface of cement paste was sealed using epoxy resin without hardener, and then the specimens were placed in an environmental chamber with 20 ± 1 °C and 50% RH. The initial cracking time and the width of cracks of cement pastes were recorded to characterize their cracking resistance.
2.3. Testing methods 2.3.1. Water requirement and packing density of cement pastes The specific gravity of cements was measured by Le Chatelier flask, and then the maximum volume concentration of solids of cement paste was tested according the
2.3.5. Drying shrinkage of cement mortars Mortar prisms of 25 25 280 mm were prepared at cement to sand mass ratio of 0.5, and the fluidity of the mortars was controlled in the range of 130– 140 mm by adjusting water addition. After being cured at 20 ± 1 °C and 90% RH
Table 1 Chemical compositions of Portland cement clinker, GBFS, and low calcium fly ash used in the experiment. Material
Density (g/cm3)
Portland cement clinker GBFS Low calcium fly ash
3.14 2.90 2.56
Note: LOI, loss on ignition.
Chemical composition (%) SiO2
Al2O3
Fe2O3
CaO
MgO
K2O
Na2O
SO3
LOI
21.6 35.22 45.43
4.35 12.15 24.36
2.95 0.25 6.70
63.81 37.08 7.53
1.76 11.25 1.51
0.51 0.49 1.23
0.16 0.25 0.36
1.06 1.19 1.03
1.19 0.36 7.88
341
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347 Table 2 Mixture proportions of gap-graded blended cements, Portland cement and reference cement. BBCFF
Cementitious material Content
<4 lm 25% GBFS
4–8 lm 11% GBFS
8–24 lm 25% Cement clinker
BCF
Cementitious material Content
<8 lm 36% GBFS
8–32 lm 25% Cement clinker
32–80 lm 39% Fly ash
Portland cement
100% Cement clinker
Reference cement
36% GBFS + 25% Cement clinker + 39% Fly ash
24–45 lm 19% Fly ash
45–80 lm 20% Fly ash
Note: 5% of gypsum dihydrate by mass of cementitious material was added for all the cements.
Fig. 4. The flexural load–flexure extension curve and location of maximum extension of the mortar specimen. Fig. 2. Particle size distributions of Portland cement, reference cement, and BCF and BBCFF cements.
Table 3 Water requirements and packing densities of Portland cement, reference cement, BBCFF and BCF cement pastes. Cement Id. Water requirement for normal consistency Specific density (g/cm3) Maximum wet density (g/cm3) Maximum solid volume concentration (%)
Portland cement
Reference cement
0.295
0.346
3.150 2.056 49.12
2.870 1.849 45.40
BCF 0.325 2.81 1.91 50.17
BBCFF 0.334 2.834 1.981 53.48
shrinkage was also determined by re-immersion of the prisms in lime-saturated water at 20 ± 1 °C for 14 days.
2.3.6. Fracture energy of gap-graded blended cement mortars Cement mortars were prepared at water: cement: river sand = 0.5:1:3 by mass and cast into 40 40 160 mm prisms. After being cured at 20 ± 1 °C and RH of 90% for 24 h, the specimens were demoulded and cured in lime-saturated water at 20 ± 1 °C for 28 days. A wedge-shaped crack with size of 1 15 mm was made in the center of the prisms as shown in Fig. 3, and then three-point bending test was performed on the prisms 2.4 kN/mm according to RILEM TC 50 [24]. The obtained flexural load-extension curve was corrected for eventual non-linearities at low loads as shown in Fig. 4, in which maximum extension (dmax) was selected when the flexural load equaled to 200 N. Thus the fracture energy (Gf) of the mortar prism can be calculated by Eq. (1) [25].
R dmax Gf ¼
0
PðdÞdd þ 12 mgdmax ðh aÞb
ð1Þ
Rd where 0 max PðdÞdd is the work done by the applied flexural load; 12 mgdmax is the work done by the gravity of the specimen; h is the height of the specimen (0.4 m); a is the depth of the wedge-shaped crack (0.15 mm); b is the width of the specimen (0.4 m); m is the weight of the specimen; P is the load applied on the specimen (N); d is the flexure extension of the specimen (m). As the repeatability of three-point bending test of cement mortar prisms was not so good, therefore six parallel specimens were used in each test, and the average of six results was taken as the fracture energy.
Fig. 3. The location and shape of the pre-cutting crack on the mortar specimen.
for 24 h, the specimens were demoulded and subsequently cured in lime-saturated water at 20 ± 1 °C for 2 days. Initial length of the mortar prisms was measured immediately after curing, then the specimens were exposed to a 20 ± 1 °C and 50% RH environmental chamber and the length change of three parallel prisms was measured at different ages according to ASTM C 596-96 [23]. Irreversible
2.3.7. Characterization of hydration products To characterize the outer-hydration products of Portland cement and GBFS, fine Portland cement fraction (D50 = 1.50 lm) and the mixture of 90% fine GBFS fraction (D50 = 1.67 lm) – 10% CaO were mixed with water and 0.2 mol/L NaOH solution into pastes of normal consistency, respectively. The pastes were cured in limesaturated water at 20 ± 1 °C for 28 days to insure completely hydrated, the hydration products were freeze-dried until constant mass was achieved, then hydration products of Portland cement and GBFS were characterized by transmission electron microscopy (TEM, Hitachi, 10 kV), X-ray energy dispersive spectroscopy (EDS, Oxford INCA X-Max, 20 eV) and thermogravimetric analysis (DSC-TG, Netzsch STA 409PC, 10 °C/min heating rate from 40 to 1000 °C in an N2 atmosphere).
342
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
3.2. Autogenous shrinkage of gap-graded blended cement pastes
Fig. 5. Chemical shrinkage of Portland cement, reference cement, BCF and BBCFF cements.
3. Results and analysis 3.1. Chemical shrinkage of gap-graded blended cements For Portland cement, reference cement and gap-graded blended cements, no obvious difference in chemical shrinkage was observed in the first 3 h, while significant difference in chemical shrinkage occurred after 3 h as shown in Fig. 5. The chemical shrinkage of Portland cement occurred mainly in the range of 3–24 h, while that of gap-graded blended cements and reference cement occurred mainly in the range of 6 h– 7 days. Generally, the chemical shrinkage of blended cement is smaller than that of Portland cement due to lower amount of hydration products. Such as, the chemical shrinkage of reference cement was about 3.7 mL/100 g after 28 days, which is much smaller than that of Portland cement. In contrast, the gap-graded blended cements presented much smaller chemical shrinkage at early ages, while the ultimate chemical shrinkage of gap-graded blended cements nearly equaled to that of Portland cement, indicating that large amount of hydration products generated in gapgraded blended cement pastes at late ages due to significant hydration of GBFS.
Fig. 6. Autogenous shrinkage of Portland cement, reference cement, BCF and BBCFF cement pastes.
The autogenous shrinkage of cement pastes mainly occurred in the first 24 h (before and during the formation of skeleton structure), and increased very slowly afterwards as shown in Fig. 6. For all cement pastes tested, there was a slightly decrease in autogenous shrinkage in the range of 500–1250 min, which can be attributed to the formation of AFt and AFm phases [4]. Compared with Portland cement paste, gap-graded blended cement pastes presented a larger decrease in autogenous shrinkage in the range of 500–1250 min due to the generation of larger amount of aluminium-incorporated hydration products (significant hydration of GBFS). Portland cement paste presented highest autogenous shrinkage, while blended cement pastes had a much smaller autogenous shrinkage. Although with same mix proportion, gap-graded blended cement pastes had a smaller autogenous shrinkage than reference cement paste, especially for BBCFF cement paste. The reason lies in that gap-graded blended cement pastes had much higher mechanical properties than reference cement paste especially at early ages, leading to a higher resistance to volumetric deformation. 3.3. Restrained shrinkage cracking of gap-graded blended cement pastes Initial cracking time of Portland cement paste was about 50 h as shown in Fig. 7, and that of reference cement paste was 62 h. In contrast, gap-graded blended cement pastes presented longer initial cracking time, for instance, the initial cracking times of BCF and BBCFF cement pastes were 81 h and 72 h, respectively. Notably, the width of initial cracks of cement pastes presented significant difference as shown in Figs. 7 and 8. Portland cement and reference cement pastes had only one wide crack, and the widths of cracks of these two cement pastes were 0.42 mm and 0.34 mm, respectively. In contrast, BCF and BBCFF cement pastes presented four fine cracks, and the widths of cracks were only 0.10 mm and 0.08 mm, respectively, indicating that shrinkage stress in gap-graded blended cement pastes was much smaller and uniformly distributed than those in Portland cement and reference cement pastes. Thus it can be concluded that gap-graded blended cement pastes had better resistance to cracking than Portland cement and reference cement pastes.
Fig. 7. Initial cracking times of Portland cement, reference cement, BCF and BBCFF cement pastes.
343
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
Fig. 8. Restrained cracking images of Portland cement, reference cement and BBCFF cement pastes.
Fig. 10. Relative weight change of cement mortars at 50% RH and 20 ± 1 °C.
4 are the equilibrium values in Fig. 9a and b respectively, and reversible shrinkage is the difference between the two. Fig. 9a shows that reference cement mortar presented a high drying shrinkage at all tested ages, in contrast, gap-graded blended cement mortars had relative smaller drying shrinkage. Fig. 9b and Table 4 indicated that BBCFF cement mortars presented a smaller total drying and reversible shrinkage than Portland cement mortar. That is to say, most of shrinkage of the gap-graded blended cement mortars was irreversible shrinkage. Fig. 10 shows that reference cement mortar had higher weight loss compared with Portland cement mortar especially at early drying ages, leading to higher drying shrinkage. Gap-graded blended cement mortars presented comparable weight loss relative to Portland cement mortar, which can be attributed to dense microstructure of gap-graded blended cement pastes [26]. Thus gap-graded blended mortar showed comparable or even slightly smaller drying shrinkage compared with Portland cement mortar. 3.5. Fracture energy of gap-graded blended cement mortars Fig. 9. Total drying shrinkage and irreversible shrinkage of cement mortars at 50% RH and 20 ± 1 °C. (a) Total drying shrinkage and (b) irreversible shrinkage.
3.4. Drying shrinkage of gap-graded blended cement mortars The time of 0 day in Fig. 9b is equivalent to the final time in Fig. 9a. Total drying shrinkage and irreversible shrinkage in Table
Flexural load vs flexure extension curves of Portland cement, reference cement, and gap-graded blended cement mortars were plotted in Fig. 11. Compared with Portland cement mortar, gapgraded blended cement mortars had nearly equal anti-load ability and slightly larger extensibility (or toughness). Table 5 shows that the fracture energy of reference cement and Portland cement
Table 4 Drying shrinkage of cement mortars cured for 54 days at 50% RH and 20 ± 1 °C. Cement Id. Type of drying shrinkage
Total shrinkage (%) Irreversible shrinkage (%) Reversible shrinkage (%)
Portland cement
Reference cement
BCF
BBCFF
0.0648 0.0538 0.0110
0.0752 0.0672 0.0080
0.0680 0.0601 0.0079
0.0616 0.0547 0.0069
344
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
4.1. Effect of hydration products on the chemical shrinkage of cementitious materials
Fig. 11. Flexural load–flexure extension curves of mortar specimens of Portland cement, reference cement, BCF and BBCFF cements.
Table 5 Fracture energy of Portland cement, reference cement, BBCFF and BCF cement mortars. Cement Id.
Reference cement
Portland cement
BCF
BBCFF
Fracture energy (N/m)
46.4 ± 5.8
53.9 ± 3.7
72.3 ± 4.6
80.7 ± 4.1
mortars were 46.4 N/m and 53.9 N/m, respectively. In contrast, the fracture energy of BCF and BBCFF cement mortars were as high as 72.3 N/m and 80.7 N/m, respectively.
Fig. 12 indicates that the fiber-like gel products were generated during the hydration of Portland cement, while flake-like gel products were observed in GBFS paste, and the hydration products seemed like much denser than those in Portland cement paste. Fig. 13 shows that the Ca/Si ratio and Al/Si ratio of gel products of Portland cement were mainly in the range of 1.55–1.85 and 0.05–0.20, respectively, and the (K + Na)/Si ratio was very low or even undetected. In contrast, the Ca/Si ratio and Al/Si ratio of gel products of GBFS was in the range of 1.1–1.4 and 0.35–0.50, respectively, and the (K + Na)/Si ratio was as high as 0.1–0.2, indicating that larger amount of Al, K, Na were incorporated in the gel products (C–S–H and/or C–A–S–H gels) of GBFS. Many researchers have proved that C–A–S–H gel with high K+ and Na+ content had improved resistance to volume changes, as the inter-layer force of C–A–S–H gel was strengthen due to that K+ and Na+ mainly existed in the inter-layer of C–A–S–H gel sheets [27–30]. Compared with hydration products of Portland cement, the hydration products of GBFS had larger weight loss in the range of 40–130 °C as shown in Fig. 14, indicating that C–S–H and/or C–A–S–H gels in GBFS paste had larger amount of chemically bound water (including strongly absorbed water). Since the weight loss of Ca(OH)2 (300–425 °C), and CaCO3 (650–825 °C) can be obtained from the TG curve, the remaining weight loss was considered as an index of chemically bound water of the gel products. The total weight loss of gel products of Portland cement was 22.41%, while that of gel products of GBFS was as high as 24.26%. Generally, the transformation of free water into chemically bound water or strongly absorbed water during the hydration is the root cause of chemical shrinkage [31]. Thus it can be inferred that the fully hydrated GBFS results in larger chemical shrinkage than fully hydrated Portland cement.
4. Discussion It is generally accepted that the volumetric deformation of cement paste can be attributed to two main reasons. One is chemical shrinkage (internal factor) due to hydration of cementitious materials, which is closely related to the composition and structure of hydration products. The other one is drying shrinkage (external factor) due to moisture consumption or loss, which depends on the microstructure and mechanical properties of hardened cement paste. In addition, effect of GBFS hydration on the volumetric deformation of gap-graded blended cement pastes was mainly focused in terms of hydration products and microstructure of cement pastes, as the hydration degree of GBFS in gap-graded blended cement pastes can be as high as 60%, while the reaction degree of fly ash was lower than 10% [26].
4.2. Effect of GBFS on the chemical shrinkage and autogenous of gapgraded blended cements Fig. 15 shows that GBFS fractions presented a much higher chemical shrinkage than corresponding Portland cement fraction at equal hydration degree. For both GBFS and Portland cement, the variation of chemical shrinkage as a function of hydration degree presented a similar tendency. When the hydration degree was low (<0.2 for example), the chemical shrinkage increased linearly with the increase of hydration degree. However, when the hydration degree exceeded 30%, the chemical shrinkage varied significantly depending on the particle size of GBFS. Coarse GBFS fraction presented higher chemical shrinkage than fine GBFS fraction at a given hydration degree, which can be
Fig. 12. TEM images of the hydration products of Portland cement and GBFS.
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
345
Fig. 15. Effect of hydration degree on the chemical shrinkage of Portland cement and BFS.
Fig. 13. The composition of the gel products of Portland cement and GBFS. (a) Al/Si ratio vs. Ca/Si ratio and (b) (K + Na)/Si ratio vs. Ca/Si ratio.
Fig. 14. DSC–TG plots of hydration products of Portland cement and GBFS.
attributed to that inner-hydration products contribute to larger proportion of total hydration products during the hydration of coarse GBFS fraction, thus it can be concluded that outer-hydra-
Fig. 16. Chemical shrinkage of Portland cement and GBFS fractions.
tion products result in smaller chemical shrinkage compared with inner-hydration products, which is also confirmed by Richardson [32]. On one hand, fully hydrated GBFS resulted in larger chemical shrinkage than fully hydrated Portland cement. On other hand, more than 80% of fine GBFS (<8) and cement clinker (8–32 lm) were hydrated after 28 days [20], therefore outer-hydration products in gap-graded blended cement pastes occupied larger proportion, resulting in relative smaller chemical shrinkage per hydration products. To clarify the effect of fine GBFS on the chemical shrinkage of gap-graded blended cements, the chemical shrinkage of typical Portland cement and GBFS fractions was measured. Fig. 16 indicates that fine GBFS fractions (D50 = 1.67 and 6.81 lm) had nearly equal chemical shrinkage with corresponding Portland cement fractions at all tested ages, while coarse GBFS fractions (such as D50 = 16.08 lm) had relative smaller chemical shrinkage than Portland cement fractions. That is to say, replacement of fine Portland cement fractions (<10 lm) by corresponding GBFS fractions will not reduce the chemical shrinkage of blended cement. As a result, the 28 days chemical shrinkage of gap-graded blended cements nearly equaled that of Portland cement. More importantly, although the 28 days chemical shrinkage of gap-graded blended cements nearly equals to that of Portland
346
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
cement, gap-graded blended cements had much smaller chemical shrinkage before the formation of skeleton structure (finial setting time was about 3.5 h). That is to say, most of chemical shrinkage of gap-graded blended cements was occurred after the formation of skeleton structure, and the skeleton structure had a strong restraint on the volumetric deformation. Therefore, gap-graded blended cement pastes presented much smaller autogenous shrinkage than Portland cement paste.
Acknowledgments This work was funded by National Natural Science Foundation of China (Nos. U1134008 and 51302090) and the China Postdoctoral Science Foundation (Nos. 2013T60802 and 2012M521599), their financial supports are gratefully acknowledged.
References 4.3. Effect of microstructure on the volumetric deformation of gapgraded blended cement pastes The micro-structural development of gap-graded blended cement pastes has been followed in our previous study [26]. Gap-graded blended cement pastes had higher initial packing density, therefore smaller amount of hydration products was needed to achieve dense microstructure [12,13], resulting in smaller chemical shrinkage. Although with only 25% cement clinker, the amounts of hydration products and un-hydrated phases, and porosity of gap-graded blended cement pastes were nearly equaled to those of Portland cement paste [26]. More importantly, the microstructure of gap-graded blended cement pastes was much more homogeneously, and pore size distribution was even finer than that of Portland cement paste, due to significant hydration of GBFS (large proportion of Ca(OH)2 were consumed) [26]. Therefore, gap-graded blended cement pastes also presented higher compressive strengths that reference cement paste, resulting in improved volumetric deformation resistance. In addition, the hydration heat of gap-graded blended cement pastes released slowly in the first 24 h and continuously at relative high rate afterward, resulting in small and uniformly distributed stresses in the hardened paste [13,14]. As a result, gap-graded blended cement pastes presented smaller autogenous shrinkage, higher fracture energy and superior cracking resistance than Portland cement and reference cement pastes.
5. Conclusions Main conclusions that can be drawn from the present study are summarized as follows: (a) Although fully hydrated GBFS results in larger chemical shrinkage, Outer-hydration products in gap-graded blended cement pastes occupied larger proportion than those in Portland cement paste, leading to relative smaller chemical shrinkage compared with inner-hydration products. Furthermore, most of chemical shrinkage of gap-graded blended cements was occurred after the formation of skeleton structure, therefore the autogenous shrinkage of gapgraded blended cements was reduced dramatically due to restraint of the skeleton structure. (b) Gap-graded blended cement pastes had a higher initial packing density due to grain size refinement, therefore small amount of hydration products was needed to achieve dense microstructure, the microstructure of gap-graded blended cement pastes was dense and homogeneously due to significant hydration of GBFS, and slowly released hydration heat resulted in small and uniformly distributed stresses in the hardened paste. (c) As a result, gap-graded blended cement pastes presented smaller volumetric deformation and superior resistance to cracking than Portland cement and reference cement pastes.
[1] Tazawa E, Miyazawa S. Influence of cement and admixture on autogenous shrinkage of cement paste. Cem Concr Res 1995;25(2):281–7. [2] Bentz DP, Jensen OM, Hansen KK, Olesen JF, Stang H, Haecker CJ. Influence of cement particle size distribution on early age autogenous strains and stresses in cement-based materials. J Am Ceram Soc 2001;84(1):129–35. [3] Wang Q, Miao M, Yan PY. The influence of high-temperature curing on the hydration characteristics of a cement–GGBS binder. Adv Cem Res 2012;24(1): 33–40. [4] Bentz DP, Jensen OM. Mitigation strategies for autogenous shrinkage cracking. Cem Concr Compos 2004;26(6):677–85. [5] Gleize PJP, Cyr M, Escadeillas G. Effects of metakaolin on autogenous shrinkage of cement pastes. Cem Concr Compos 2007;29(2):80–7. [6] Bentz DP, Sant G, Weiss J. Early-age properties of cement-based materials. I: Influence of cement fineness. J Mater Civ Eng 2008;20(7):502–8. [7] Rao GA. Long-term drying shrinkage of mortar: influence of silica fume and size of fine aggregate. Cem Concr Res 2001;31(2):171–5. [8] Tazawa EI, Miyazawa S. Influence of cement and admixture on autogenous shrinkage of cement paste. Cem Concr Res 1995;25(2):281–7. [9] Termkhajornkit P, Nawa T, Nakai M, Saito T. Effect of fly ash on autogenous shrinkage. Cem Concr Res 2005;35(3):473–82. [10] Burrows RW, Kepler WF, Hurcomb D, Schaffer J, Sellers JG. Three simple tests for selecting low-crack cement. Cem Concr Compos 2004;26(5):509–19. [11] Mehta PK, Burrows RW. Building durable structures in the 21st century. Concr Int 2001;23(3):57–63. [12] Zhang TS, Yu QJ, Wei JX, Zhang PP, Chen PX. A gap-graded particle size distribution for blended cements: analytical approach and experimental validation. Powder Technol 2011;214(2):259–68. [13] Zhang TS, Yu QJ, Wei JX, Zhang PP. A new gap-graded particle size distribution and resulting consequences on properties of blended cement. Cem Concr Compos 2011;33(5):543–50. [14] Zhang TS, Yu QJ, Wei JX, Zhang PP. Efficient utilization of cementitious materials to produce sustainable blended cement. Cem Concr Compos 2012;34(5):692–9. [15] ASTM C 618-03. Standard specification for coal fly ash and raw or calcined natural pozzolan for use as a mineral admixture in Portland cement concrete. New York: American Society for Testing and Materials; 2003. [16] EN 196-3. Methods of testing cement: determination of water requirement for normal consistency. London: British Standards Institution; 2005. [17] Zhang TS, Yu QJ, Wei JX, Zhang PP. Measurement of chemical shrinkage of cement paste: Comparison study of ASTM C1608 and an improved method. Constr Build Mater 2013;48:662–9. [18] Zhang TS, Yu QJ, Wei JX, Zhang PP, Li S. Effect of size fraction of ground granulated blast furnace slag on its strength contribution and hydraulic activity. Adv Sci Lett 2011;4(3):1286–91. [19] Zhang TS, Yu QJ, Wei JX, Zhang PP, Gong CC. Effect of simulative pore solution on the hydration kinetics of ground granulated blast furnace slag. Adv Cem Res 2012;24(5):283–90. [20] Wang Q, Feng JJ, Yan PY. The microstructure of 4-year-old hardened cementfly ash paste. Constr Build Mater 2012;29:114–9. [21] Gao P, Zhang TS, Luo RF, Wei JX, Yu QJ. Improvement of autogenous shrinkage measurement for cement paste at very early age: corrugated tube method using non-contact sensors. Constr Build Mater 2014;55:57–62. [22] He Z, Li ZJ. Influence of alkali on restrained shrinkage behavior of cementbased materials. Cem Concr Res 2005;35(3):457–63. [23] ASTM C 596-96. Standard test method for drying shrinkage of mortar containing hydraulic cement. New York: American Society for Testing and Materials; 2003. [24] RILEM TC 50-FMC. Fracture mechanics of concrete. Determination of fracture energy of mortar and concrete by means of three point bend tests on notched beams. Mater Struct 1985;18(106):285–96. [25] Mahmoud M, Reda T, Shrive NG. Enhancing fracture toughness of highperformance carbon fiber cement composites. ACI Mater J 2001;98(2): 168–78. [26] Zhang TS, Yu QJ, Wei JX, Gao P, Chen PX, Hu J. Micro-structural development of gap-graded blended cement paste containing high amount of supplementary cementitious materials. Cem Concr Compos 2012;34(9): 1024–32. [27] Chen W, Brouwers HJH. Alkali binding in hydrated Portland cement paste. Cem Concr Res 2010;40(5):716–22. [28] Hong SY, Glasser FP. Alkali sorption by C–S–H and C–A–S–H gels: Part II Role of alumina. Cem Concr Res 2002;32(7):1101–11.
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347 [29] Oh JE, Clark SM, Monteiro PJM. Does the Al substitution in C–S–H(I) change its mechanical property? Cem Concr Res 2011;41(1):102–6. [30] Wang Q, Yang JW, Yan PY. Influence of initial alkalinity on the hydration of steel slag. Sci China Technol Sci 2012;55(12):3378–87.
347
[31] Taylor HFW. Cement chemistry. 2nd ed. London: Thomas Telford; 1997. [32] Richardson IG. The calcium silicate hydrates. Cem Concr Res 2008;38(2): 137–58.
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Volumetric deformation of gap-graded blended cement pastes with large amount of supplementary cementitious materials Tongsheng Zhang, Peng Gao, Ruifeng Luo, Jiangxiong Wei, Qijun Yu ⇑ School of Materials Science and Engineering, South China University of Technology, 510640 Guangzhou, People’s Republic of China
h i g h l i g h t s Gap-graded blended cement pastes presented comparable chemical shrinkage with Portland cement paste. Most of chemical shrinkage occurred after the formation of skeleton structure, leading to reduced autogenous shrinkage. The microstructure of gap-graded blended cement paste was dense and homogeneous. Shrinkage stress was small and uniformly distributed in gap-graded blended cement paste. Gap-graded blended cement paste presented smaller volumetric deformation and superior resistance to cracking.
a r t i c l e
i n f o
Article history: Received 11 October 2013 Received in revised form 18 December 2013 Accepted 19 December 2013 Available online 19 January 2014 Keywords: Gap-graded blended cement Volumetric deformation Chemical shrinkage Hydration products
a b s t r a c t In this study, volumetric deformation of gap-graded blended cement pastes with large amount of supplementary cementitous materials was investigated and compared with those of Portland cement and reference blended cement pastes. The results show that the gap-graded blended cement pastes presented a higher initial packing density, therefore smaller amount of hydration products was needed to achieve dense microstructure, resulting in smaller chemical shrinkage. Notably, most of chemical shrinkage of gap-graded blended cements was occurred after the formation of skeleton structure, thus the autogenous shrinkage of gap-graded blended cements was reduced significantly due to restraint of the skeleton structure. Further, the microstructure of gap-graded blended cement pastes was dense and homogeneous due to grain size refinement and significant hydration of GBFS, and shrinkage stress was small and uniformly distributed. As a result, gap-graded blended cement pastes presented smaller volumetric deformation and superior resistance to cracking than Portland cement and reference cement pastes. Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.
1. Introduction The durability of concrete structures is increasing concerned during the past decades, and the early volumetric deformation of cement paste is generally considered as the key factor leading to early age cracking and, consequently, to the loss of durability of concrete structures. Tazawa and Bentz reported that finer cement usually resulted in larger autogenous shrinkage of cement-based materials, thus coarse cement was recommended in concrete manufacture if the mechanical properties of the concrete met the requirements [1,2]. Numerous literatures proved that the autogenous shrinkage of cement pastes with supplementary cementitious material (SCM) was significantly influenced by the hydraulic activity (or the type) and fineness of the incorporated SCMs [3–6]. For instance, the addition of silica fume increased the autogenous shrinkage of cement pastes remarkably [7]. When the specific ⇑ Corresponding author. Tel./fax: +86 020 87114233. E-mail address: [email protected] (Q. Yu).
surface area of granulated blast furnace slag (GBFS) was lower than 400 m2/kg, the addition of GBFS decreased the autogenous shrinkage of cement pastes. In contrast, for GBFS with specific surface area higher than 400 m2/kg, the autogenous shrinkage of cement pastes increased with the increase of GBFS replacement [8]. Low calcium fly ash presented low hydraulic activity, therefore the addition of low calcium fly ash generally reduced the autogenous shrinkage of cement pastes [9]. From the above literatures, it can be concluded that the volumetric deformation of cement-based materials mainly depends on the hydration process of cementitious materials and the composition of hydration products. Burrows and Ba proved that fast hydration of cementitious materials resulted in larger volumetric deformation, leading to a higher risk of cracking of cement-based materials [10,11]. Therefore optimization of hydration process of cementitious materials is an effective way to minimize the volumetric deformation of cement-based materials. Particle size distribution (PSD) has a significant influence on initial packing density and hydration process of cement paste,
0950-0618/$ - see front matter Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.conbuildmat.2013.12.053
340
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
eventually affecting the amount and composition of hydration products in cement paste. If cement paste had a higher initial packing density, smaller the amount of hydration products were needed to fill-up the voids in cement paste, therefore the volumetric deformation of cement paste would be improved dramatically. In previous study, a gap-graded PSD was proposed based on close packing theory, the initial packing density of blended cement paste was increased significantly, as voids between large size particles were filled in by fine particles grade by grade [12]. And the hydration process of gap-graded blended cements was optimized by arranging SCMs with high activity, cement clinker, and SCMs with low activity (or inert fillers) in fine, middle size and coarse fractions, respectively. The microstructure of gap-graded blended cement pastes was densified gradually at later ages due to hydration of GBFS. As a result, gap-graded blended cements with only 25% clinker by mass presented comparable mechanical properties with pure Portland cement [13,14]. Although with superior mechanical properties, it is more important to clarify the volumetric deformation of gap-graded blended cement pastes before application, as the durability of concrete structures is largely depended on the volumetric deformation of cement paste incorporated. In the present study, the volumetric deformation of gap-graded blended cement pastes, Portland cement paste and reference blended cement paste were comparatively studied, and the volumetric deformation mechanism of gap-graded blended cement pastes was discussed from the viewpoint of composition of hydration products and microstructure of hardened paste. The results will be very useful to improve the volumetric deformation of cement-based materials by reasonable utilization of cement clinker and SCMs. 2. Experimental procedures 2.1. Raw materials The chemical compositions of raw materials used in the experiment are given in Table 1. Cement clinker, GBFS, and low calcium fly ash (a Class F fly ash according to ASTM C 618 [15]) were ground and then classified by an air classifier. The PSDs of cementitious material fractions required by the gap-graded PSD are given in Fig. 1. 2.2. Preparation of gap-graded blended cements Gap-graded blended cements (BCF and BBCFF) were prepared by mixing cementitious material fractions homogeneously according to the mix proportions listed in Table 2, while reference cement and Portland cement were prepared by co-grinding the mixture. The Blaine specific surface areas of these two cements were controlled to be in the range of 350–360 m2/kg, which is seen to be equal to those of the gapgraded blended cements approximately. Fig. 2 shows that gap-graded blended cements presented wider PSDs than cements prepared by co-grinding (reference cement and Portland cement), which will leads to a higher initial packing density of cement paste. Although gap-graded blended cements had same mix proportion with reference cement, however, it should be noted that GBFS, cement clinker and fly ash were placed in the fine, middle size, and coarse fractions of gap-graded blended cements, respectively.
Fig. 1. Particle size distributions of cementitious material fractions used in the experiment.
procedures specified in Zhang’s research [13], to characterize the initial packing density of cement paste. In addition, water requirement for normal consistency of blended cements were determined according to EN 196-3 [16]. As the water requirements for normal consistency of Portland cement, reference cement and gap-graded blended cements were quite different from each other as shown in Table 3, therefore cement pastes and mortar with same fluidity were used for volumetric deformation measurements, in consideration that cement-based materials were mainly used under equal workability (fluidity). 2.3.2. Chemical shrinkage of blended cements Chemical shrinkage of the blended cements prepared, typical Portland cement and GBFS fractions was measured by volumetric method at 20 ± 1 °C, details of the method were specified in the literature [17]. To simulate the hydration of GBFS in blended cement, a mixture of 90% GBFS and 10% CaO by mass were mixed, and 0.2 mol/L NaOH solution was used as a simulated pore solution in chemical shrinkage measurement of GBFS [18,19]. Along with chemical shrinkage measurement, the hydration degrees of Portland cement and GBFS fractions was also determined by non-evaporation water method [20] and ethylene diamine tetraacetic acid disodium salt (EDTA) preferential solving method [18], respectively. 2.3.3. Autogenous shrinkage of cement pastes The autogenous shrinkage of blended cement pastes of normal consistency was measured using non-contact corrugated tube method at a 20 ± 1 °C environmental chamber [21]. Three parallel samples were performed and the average value of results was used as the autogenous shrinkage of cement pastes. 2.3.4. Restrained shrinkage cracking of gap-graded blended cement pastes Cement paste of normal consistency was cast into an ellipse-ring shaped mold, and a stainless steel ellipse in the center of the mold was used to restrain the shrinkage of the surrounding cement paste [22]. After being cured at 20 ± 1 °C and 90% relative humidity (RH) for 24 h, the outer molds of ring specimen were demolded, the top surface of cement paste was sealed using epoxy resin without hardener, and then the specimens were placed in an environmental chamber with 20 ± 1 °C and 50% RH. The initial cracking time and the width of cracks of cement pastes were recorded to characterize their cracking resistance.
2.3. Testing methods 2.3.1. Water requirement and packing density of cement pastes The specific gravity of cements was measured by Le Chatelier flask, and then the maximum volume concentration of solids of cement paste was tested according the
2.3.5. Drying shrinkage of cement mortars Mortar prisms of 25 25 280 mm were prepared at cement to sand mass ratio of 0.5, and the fluidity of the mortars was controlled in the range of 130– 140 mm by adjusting water addition. After being cured at 20 ± 1 °C and 90% RH
Table 1 Chemical compositions of Portland cement clinker, GBFS, and low calcium fly ash used in the experiment. Material
Density (g/cm3)
Portland cement clinker GBFS Low calcium fly ash
3.14 2.90 2.56
Note: LOI, loss on ignition.
Chemical composition (%) SiO2
Al2O3
Fe2O3
CaO
MgO
K2O
Na2O
SO3
LOI
21.6 35.22 45.43
4.35 12.15 24.36
2.95 0.25 6.70
63.81 37.08 7.53
1.76 11.25 1.51
0.51 0.49 1.23
0.16 0.25 0.36
1.06 1.19 1.03
1.19 0.36 7.88
341
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347 Table 2 Mixture proportions of gap-graded blended cements, Portland cement and reference cement. BBCFF
Cementitious material Content
<4 lm 25% GBFS
4–8 lm 11% GBFS
8–24 lm 25% Cement clinker
BCF
Cementitious material Content
<8 lm 36% GBFS
8–32 lm 25% Cement clinker
32–80 lm 39% Fly ash
Portland cement
100% Cement clinker
Reference cement
36% GBFS + 25% Cement clinker + 39% Fly ash
24–45 lm 19% Fly ash
45–80 lm 20% Fly ash
Note: 5% of gypsum dihydrate by mass of cementitious material was added for all the cements.
Fig. 4. The flexural load–flexure extension curve and location of maximum extension of the mortar specimen. Fig. 2. Particle size distributions of Portland cement, reference cement, and BCF and BBCFF cements.
Table 3 Water requirements and packing densities of Portland cement, reference cement, BBCFF and BCF cement pastes. Cement Id. Water requirement for normal consistency Specific density (g/cm3) Maximum wet density (g/cm3) Maximum solid volume concentration (%)
Portland cement
Reference cement
0.295
0.346
3.150 2.056 49.12
2.870 1.849 45.40
BCF 0.325 2.81 1.91 50.17
BBCFF 0.334 2.834 1.981 53.48
shrinkage was also determined by re-immersion of the prisms in lime-saturated water at 20 ± 1 °C for 14 days.
2.3.6. Fracture energy of gap-graded blended cement mortars Cement mortars were prepared at water: cement: river sand = 0.5:1:3 by mass and cast into 40 40 160 mm prisms. After being cured at 20 ± 1 °C and RH of 90% for 24 h, the specimens were demoulded and cured in lime-saturated water at 20 ± 1 °C for 28 days. A wedge-shaped crack with size of 1 15 mm was made in the center of the prisms as shown in Fig. 3, and then three-point bending test was performed on the prisms 2.4 kN/mm according to RILEM TC 50 [24]. The obtained flexural load-extension curve was corrected for eventual non-linearities at low loads as shown in Fig. 4, in which maximum extension (dmax) was selected when the flexural load equaled to 200 N. Thus the fracture energy (Gf) of the mortar prism can be calculated by Eq. (1) [25].
R dmax Gf ¼
0
PðdÞdd þ 12 mgdmax ðh aÞb
ð1Þ
Rd where 0 max PðdÞdd is the work done by the applied flexural load; 12 mgdmax is the work done by the gravity of the specimen; h is the height of the specimen (0.4 m); a is the depth of the wedge-shaped crack (0.15 mm); b is the width of the specimen (0.4 m); m is the weight of the specimen; P is the load applied on the specimen (N); d is the flexure extension of the specimen (m). As the repeatability of three-point bending test of cement mortar prisms was not so good, therefore six parallel specimens were used in each test, and the average of six results was taken as the fracture energy.
Fig. 3. The location and shape of the pre-cutting crack on the mortar specimen.
for 24 h, the specimens were demoulded and subsequently cured in lime-saturated water at 20 ± 1 °C for 2 days. Initial length of the mortar prisms was measured immediately after curing, then the specimens were exposed to a 20 ± 1 °C and 50% RH environmental chamber and the length change of three parallel prisms was measured at different ages according to ASTM C 596-96 [23]. Irreversible
2.3.7. Characterization of hydration products To characterize the outer-hydration products of Portland cement and GBFS, fine Portland cement fraction (D50 = 1.50 lm) and the mixture of 90% fine GBFS fraction (D50 = 1.67 lm) – 10% CaO were mixed with water and 0.2 mol/L NaOH solution into pastes of normal consistency, respectively. The pastes were cured in limesaturated water at 20 ± 1 °C for 28 days to insure completely hydrated, the hydration products were freeze-dried until constant mass was achieved, then hydration products of Portland cement and GBFS were characterized by transmission electron microscopy (TEM, Hitachi, 10 kV), X-ray energy dispersive spectroscopy (EDS, Oxford INCA X-Max, 20 eV) and thermogravimetric analysis (DSC-TG, Netzsch STA 409PC, 10 °C/min heating rate from 40 to 1000 °C in an N2 atmosphere).
342
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
3.2. Autogenous shrinkage of gap-graded blended cement pastes
Fig. 5. Chemical shrinkage of Portland cement, reference cement, BCF and BBCFF cements.
3. Results and analysis 3.1. Chemical shrinkage of gap-graded blended cements For Portland cement, reference cement and gap-graded blended cements, no obvious difference in chemical shrinkage was observed in the first 3 h, while significant difference in chemical shrinkage occurred after 3 h as shown in Fig. 5. The chemical shrinkage of Portland cement occurred mainly in the range of 3–24 h, while that of gap-graded blended cements and reference cement occurred mainly in the range of 6 h– 7 days. Generally, the chemical shrinkage of blended cement is smaller than that of Portland cement due to lower amount of hydration products. Such as, the chemical shrinkage of reference cement was about 3.7 mL/100 g after 28 days, which is much smaller than that of Portland cement. In contrast, the gap-graded blended cements presented much smaller chemical shrinkage at early ages, while the ultimate chemical shrinkage of gap-graded blended cements nearly equaled to that of Portland cement, indicating that large amount of hydration products generated in gapgraded blended cement pastes at late ages due to significant hydration of GBFS.
Fig. 6. Autogenous shrinkage of Portland cement, reference cement, BCF and BBCFF cement pastes.
The autogenous shrinkage of cement pastes mainly occurred in the first 24 h (before and during the formation of skeleton structure), and increased very slowly afterwards as shown in Fig. 6. For all cement pastes tested, there was a slightly decrease in autogenous shrinkage in the range of 500–1250 min, which can be attributed to the formation of AFt and AFm phases [4]. Compared with Portland cement paste, gap-graded blended cement pastes presented a larger decrease in autogenous shrinkage in the range of 500–1250 min due to the generation of larger amount of aluminium-incorporated hydration products (significant hydration of GBFS). Portland cement paste presented highest autogenous shrinkage, while blended cement pastes had a much smaller autogenous shrinkage. Although with same mix proportion, gap-graded blended cement pastes had a smaller autogenous shrinkage than reference cement paste, especially for BBCFF cement paste. The reason lies in that gap-graded blended cement pastes had much higher mechanical properties than reference cement paste especially at early ages, leading to a higher resistance to volumetric deformation. 3.3. Restrained shrinkage cracking of gap-graded blended cement pastes Initial cracking time of Portland cement paste was about 50 h as shown in Fig. 7, and that of reference cement paste was 62 h. In contrast, gap-graded blended cement pastes presented longer initial cracking time, for instance, the initial cracking times of BCF and BBCFF cement pastes were 81 h and 72 h, respectively. Notably, the width of initial cracks of cement pastes presented significant difference as shown in Figs. 7 and 8. Portland cement and reference cement pastes had only one wide crack, and the widths of cracks of these two cement pastes were 0.42 mm and 0.34 mm, respectively. In contrast, BCF and BBCFF cement pastes presented four fine cracks, and the widths of cracks were only 0.10 mm and 0.08 mm, respectively, indicating that shrinkage stress in gap-graded blended cement pastes was much smaller and uniformly distributed than those in Portland cement and reference cement pastes. Thus it can be concluded that gap-graded blended cement pastes had better resistance to cracking than Portland cement and reference cement pastes.
Fig. 7. Initial cracking times of Portland cement, reference cement, BCF and BBCFF cement pastes.
343
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
Fig. 8. Restrained cracking images of Portland cement, reference cement and BBCFF cement pastes.
Fig. 10. Relative weight change of cement mortars at 50% RH and 20 ± 1 °C.
4 are the equilibrium values in Fig. 9a and b respectively, and reversible shrinkage is the difference between the two. Fig. 9a shows that reference cement mortar presented a high drying shrinkage at all tested ages, in contrast, gap-graded blended cement mortars had relative smaller drying shrinkage. Fig. 9b and Table 4 indicated that BBCFF cement mortars presented a smaller total drying and reversible shrinkage than Portland cement mortar. That is to say, most of shrinkage of the gap-graded blended cement mortars was irreversible shrinkage. Fig. 10 shows that reference cement mortar had higher weight loss compared with Portland cement mortar especially at early drying ages, leading to higher drying shrinkage. Gap-graded blended cement mortars presented comparable weight loss relative to Portland cement mortar, which can be attributed to dense microstructure of gap-graded blended cement pastes [26]. Thus gap-graded blended mortar showed comparable or even slightly smaller drying shrinkage compared with Portland cement mortar. 3.5. Fracture energy of gap-graded blended cement mortars Fig. 9. Total drying shrinkage and irreversible shrinkage of cement mortars at 50% RH and 20 ± 1 °C. (a) Total drying shrinkage and (b) irreversible shrinkage.
3.4. Drying shrinkage of gap-graded blended cement mortars The time of 0 day in Fig. 9b is equivalent to the final time in Fig. 9a. Total drying shrinkage and irreversible shrinkage in Table
Flexural load vs flexure extension curves of Portland cement, reference cement, and gap-graded blended cement mortars were plotted in Fig. 11. Compared with Portland cement mortar, gapgraded blended cement mortars had nearly equal anti-load ability and slightly larger extensibility (or toughness). Table 5 shows that the fracture energy of reference cement and Portland cement
Table 4 Drying shrinkage of cement mortars cured for 54 days at 50% RH and 20 ± 1 °C. Cement Id. Type of drying shrinkage
Total shrinkage (%) Irreversible shrinkage (%) Reversible shrinkage (%)
Portland cement
Reference cement
BCF
BBCFF
0.0648 0.0538 0.0110
0.0752 0.0672 0.0080
0.0680 0.0601 0.0079
0.0616 0.0547 0.0069
344
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
4.1. Effect of hydration products on the chemical shrinkage of cementitious materials
Fig. 11. Flexural load–flexure extension curves of mortar specimens of Portland cement, reference cement, BCF and BBCFF cements.
Table 5 Fracture energy of Portland cement, reference cement, BBCFF and BCF cement mortars. Cement Id.
Reference cement
Portland cement
BCF
BBCFF
Fracture energy (N/m)
46.4 ± 5.8
53.9 ± 3.7
72.3 ± 4.6
80.7 ± 4.1
mortars were 46.4 N/m and 53.9 N/m, respectively. In contrast, the fracture energy of BCF and BBCFF cement mortars were as high as 72.3 N/m and 80.7 N/m, respectively.
Fig. 12 indicates that the fiber-like gel products were generated during the hydration of Portland cement, while flake-like gel products were observed in GBFS paste, and the hydration products seemed like much denser than those in Portland cement paste. Fig. 13 shows that the Ca/Si ratio and Al/Si ratio of gel products of Portland cement were mainly in the range of 1.55–1.85 and 0.05–0.20, respectively, and the (K + Na)/Si ratio was very low or even undetected. In contrast, the Ca/Si ratio and Al/Si ratio of gel products of GBFS was in the range of 1.1–1.4 and 0.35–0.50, respectively, and the (K + Na)/Si ratio was as high as 0.1–0.2, indicating that larger amount of Al, K, Na were incorporated in the gel products (C–S–H and/or C–A–S–H gels) of GBFS. Many researchers have proved that C–A–S–H gel with high K+ and Na+ content had improved resistance to volume changes, as the inter-layer force of C–A–S–H gel was strengthen due to that K+ and Na+ mainly existed in the inter-layer of C–A–S–H gel sheets [27–30]. Compared with hydration products of Portland cement, the hydration products of GBFS had larger weight loss in the range of 40–130 °C as shown in Fig. 14, indicating that C–S–H and/or C–A–S–H gels in GBFS paste had larger amount of chemically bound water (including strongly absorbed water). Since the weight loss of Ca(OH)2 (300–425 °C), and CaCO3 (650–825 °C) can be obtained from the TG curve, the remaining weight loss was considered as an index of chemically bound water of the gel products. The total weight loss of gel products of Portland cement was 22.41%, while that of gel products of GBFS was as high as 24.26%. Generally, the transformation of free water into chemically bound water or strongly absorbed water during the hydration is the root cause of chemical shrinkage [31]. Thus it can be inferred that the fully hydrated GBFS results in larger chemical shrinkage than fully hydrated Portland cement.
4. Discussion It is generally accepted that the volumetric deformation of cement paste can be attributed to two main reasons. One is chemical shrinkage (internal factor) due to hydration of cementitious materials, which is closely related to the composition and structure of hydration products. The other one is drying shrinkage (external factor) due to moisture consumption or loss, which depends on the microstructure and mechanical properties of hardened cement paste. In addition, effect of GBFS hydration on the volumetric deformation of gap-graded blended cement pastes was mainly focused in terms of hydration products and microstructure of cement pastes, as the hydration degree of GBFS in gap-graded blended cement pastes can be as high as 60%, while the reaction degree of fly ash was lower than 10% [26].
4.2. Effect of GBFS on the chemical shrinkage and autogenous of gapgraded blended cements Fig. 15 shows that GBFS fractions presented a much higher chemical shrinkage than corresponding Portland cement fraction at equal hydration degree. For both GBFS and Portland cement, the variation of chemical shrinkage as a function of hydration degree presented a similar tendency. When the hydration degree was low (<0.2 for example), the chemical shrinkage increased linearly with the increase of hydration degree. However, when the hydration degree exceeded 30%, the chemical shrinkage varied significantly depending on the particle size of GBFS. Coarse GBFS fraction presented higher chemical shrinkage than fine GBFS fraction at a given hydration degree, which can be
Fig. 12. TEM images of the hydration products of Portland cement and GBFS.
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
345
Fig. 15. Effect of hydration degree on the chemical shrinkage of Portland cement and BFS.
Fig. 13. The composition of the gel products of Portland cement and GBFS. (a) Al/Si ratio vs. Ca/Si ratio and (b) (K + Na)/Si ratio vs. Ca/Si ratio.
Fig. 14. DSC–TG plots of hydration products of Portland cement and GBFS.
attributed to that inner-hydration products contribute to larger proportion of total hydration products during the hydration of coarse GBFS fraction, thus it can be concluded that outer-hydra-
Fig. 16. Chemical shrinkage of Portland cement and GBFS fractions.
tion products result in smaller chemical shrinkage compared with inner-hydration products, which is also confirmed by Richardson [32]. On one hand, fully hydrated GBFS resulted in larger chemical shrinkage than fully hydrated Portland cement. On other hand, more than 80% of fine GBFS (<8) and cement clinker (8–32 lm) were hydrated after 28 days [20], therefore outer-hydration products in gap-graded blended cement pastes occupied larger proportion, resulting in relative smaller chemical shrinkage per hydration products. To clarify the effect of fine GBFS on the chemical shrinkage of gap-graded blended cements, the chemical shrinkage of typical Portland cement and GBFS fractions was measured. Fig. 16 indicates that fine GBFS fractions (D50 = 1.67 and 6.81 lm) had nearly equal chemical shrinkage with corresponding Portland cement fractions at all tested ages, while coarse GBFS fractions (such as D50 = 16.08 lm) had relative smaller chemical shrinkage than Portland cement fractions. That is to say, replacement of fine Portland cement fractions (<10 lm) by corresponding GBFS fractions will not reduce the chemical shrinkage of blended cement. As a result, the 28 days chemical shrinkage of gap-graded blended cements nearly equaled that of Portland cement. More importantly, although the 28 days chemical shrinkage of gap-graded blended cements nearly equals to that of Portland
346
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347
cement, gap-graded blended cements had much smaller chemical shrinkage before the formation of skeleton structure (finial setting time was about 3.5 h). That is to say, most of chemical shrinkage of gap-graded blended cements was occurred after the formation of skeleton structure, and the skeleton structure had a strong restraint on the volumetric deformation. Therefore, gap-graded blended cement pastes presented much smaller autogenous shrinkage than Portland cement paste.
Acknowledgments This work was funded by National Natural Science Foundation of China (Nos. U1134008 and 51302090) and the China Postdoctoral Science Foundation (Nos. 2013T60802 and 2012M521599), their financial supports are gratefully acknowledged.
References 4.3. Effect of microstructure on the volumetric deformation of gapgraded blended cement pastes The micro-structural development of gap-graded blended cement pastes has been followed in our previous study [26]. Gap-graded blended cement pastes had higher initial packing density, therefore smaller amount of hydration products was needed to achieve dense microstructure [12,13], resulting in smaller chemical shrinkage. Although with only 25% cement clinker, the amounts of hydration products and un-hydrated phases, and porosity of gap-graded blended cement pastes were nearly equaled to those of Portland cement paste [26]. More importantly, the microstructure of gap-graded blended cement pastes was much more homogeneously, and pore size distribution was even finer than that of Portland cement paste, due to significant hydration of GBFS (large proportion of Ca(OH)2 were consumed) [26]. Therefore, gap-graded blended cement pastes also presented higher compressive strengths that reference cement paste, resulting in improved volumetric deformation resistance. In addition, the hydration heat of gap-graded blended cement pastes released slowly in the first 24 h and continuously at relative high rate afterward, resulting in small and uniformly distributed stresses in the hardened paste [13,14]. As a result, gap-graded blended cement pastes presented smaller autogenous shrinkage, higher fracture energy and superior cracking resistance than Portland cement and reference cement pastes.
5. Conclusions Main conclusions that can be drawn from the present study are summarized as follows: (a) Although fully hydrated GBFS results in larger chemical shrinkage, Outer-hydration products in gap-graded blended cement pastes occupied larger proportion than those in Portland cement paste, leading to relative smaller chemical shrinkage compared with inner-hydration products. Furthermore, most of chemical shrinkage of gap-graded blended cements was occurred after the formation of skeleton structure, therefore the autogenous shrinkage of gapgraded blended cements was reduced dramatically due to restraint of the skeleton structure. (b) Gap-graded blended cement pastes had a higher initial packing density due to grain size refinement, therefore small amount of hydration products was needed to achieve dense microstructure, the microstructure of gap-graded blended cement pastes was dense and homogeneously due to significant hydration of GBFS, and slowly released hydration heat resulted in small and uniformly distributed stresses in the hardened paste. (c) As a result, gap-graded blended cement pastes presented smaller volumetric deformation and superior resistance to cracking than Portland cement and reference cement pastes.
[1] Tazawa E, Miyazawa S. Influence of cement and admixture on autogenous shrinkage of cement paste. Cem Concr Res 1995;25(2):281–7. [2] Bentz DP, Jensen OM, Hansen KK, Olesen JF, Stang H, Haecker CJ. Influence of cement particle size distribution on early age autogenous strains and stresses in cement-based materials. J Am Ceram Soc 2001;84(1):129–35. [3] Wang Q, Miao M, Yan PY. The influence of high-temperature curing on the hydration characteristics of a cement–GGBS binder. Adv Cem Res 2012;24(1): 33–40. [4] Bentz DP, Jensen OM. Mitigation strategies for autogenous shrinkage cracking. Cem Concr Compos 2004;26(6):677–85. [5] Gleize PJP, Cyr M, Escadeillas G. Effects of metakaolin on autogenous shrinkage of cement pastes. Cem Concr Compos 2007;29(2):80–7. [6] Bentz DP, Sant G, Weiss J. Early-age properties of cement-based materials. I: Influence of cement fineness. J Mater Civ Eng 2008;20(7):502–8. [7] Rao GA. Long-term drying shrinkage of mortar: influence of silica fume and size of fine aggregate. Cem Concr Res 2001;31(2):171–5. [8] Tazawa EI, Miyazawa S. Influence of cement and admixture on autogenous shrinkage of cement paste. Cem Concr Res 1995;25(2):281–7. [9] Termkhajornkit P, Nawa T, Nakai M, Saito T. Effect of fly ash on autogenous shrinkage. Cem Concr Res 2005;35(3):473–82. [10] Burrows RW, Kepler WF, Hurcomb D, Schaffer J, Sellers JG. Three simple tests for selecting low-crack cement. Cem Concr Compos 2004;26(5):509–19. [11] Mehta PK, Burrows RW. Building durable structures in the 21st century. Concr Int 2001;23(3):57–63. [12] Zhang TS, Yu QJ, Wei JX, Zhang PP, Chen PX. A gap-graded particle size distribution for blended cements: analytical approach and experimental validation. Powder Technol 2011;214(2):259–68. [13] Zhang TS, Yu QJ, Wei JX, Zhang PP. A new gap-graded particle size distribution and resulting consequences on properties of blended cement. Cem Concr Compos 2011;33(5):543–50. [14] Zhang TS, Yu QJ, Wei JX, Zhang PP. Efficient utilization of cementitious materials to produce sustainable blended cement. Cem Concr Compos 2012;34(5):692–9. [15] ASTM C 618-03. Standard specification for coal fly ash and raw or calcined natural pozzolan for use as a mineral admixture in Portland cement concrete. New York: American Society for Testing and Materials; 2003. [16] EN 196-3. Methods of testing cement: determination of water requirement for normal consistency. London: British Standards Institution; 2005. [17] Zhang TS, Yu QJ, Wei JX, Zhang PP. Measurement of chemical shrinkage of cement paste: Comparison study of ASTM C1608 and an improved method. Constr Build Mater 2013;48:662–9. [18] Zhang TS, Yu QJ, Wei JX, Zhang PP, Li S. Effect of size fraction of ground granulated blast furnace slag on its strength contribution and hydraulic activity. Adv Sci Lett 2011;4(3):1286–91. [19] Zhang TS, Yu QJ, Wei JX, Zhang PP, Gong CC. Effect of simulative pore solution on the hydration kinetics of ground granulated blast furnace slag. Adv Cem Res 2012;24(5):283–90. [20] Wang Q, Feng JJ, Yan PY. The microstructure of 4-year-old hardened cementfly ash paste. Constr Build Mater 2012;29:114–9. [21] Gao P, Zhang TS, Luo RF, Wei JX, Yu QJ. Improvement of autogenous shrinkage measurement for cement paste at very early age: corrugated tube method using non-contact sensors. Constr Build Mater 2014;55:57–62. [22] He Z, Li ZJ. Influence of alkali on restrained shrinkage behavior of cementbased materials. Cem Concr Res 2005;35(3):457–63. [23] ASTM C 596-96. Standard test method for drying shrinkage of mortar containing hydraulic cement. New York: American Society for Testing and Materials; 2003. [24] RILEM TC 50-FMC. Fracture mechanics of concrete. Determination of fracture energy of mortar and concrete by means of three point bend tests on notched beams. Mater Struct 1985;18(106):285–96. [25] Mahmoud M, Reda T, Shrive NG. Enhancing fracture toughness of highperformance carbon fiber cement composites. ACI Mater J 2001;98(2): 168–78. [26] Zhang TS, Yu QJ, Wei JX, Gao P, Chen PX, Hu J. Micro-structural development of gap-graded blended cement paste containing high amount of supplementary cementitious materials. Cem Concr Compos 2012;34(9): 1024–32. [27] Chen W, Brouwers HJH. Alkali binding in hydrated Portland cement paste. Cem Concr Res 2010;40(5):716–22. [28] Hong SY, Glasser FP. Alkali sorption by C–S–H and C–A–S–H gels: Part II Role of alumina. Cem Concr Res 2002;32(7):1101–11.
T. Zhang et al. / Construction and Building Materials 54 (2014) 339–347 [29] Oh JE, Clark SM, Monteiro PJM. Does the Al substitution in C–S–H(I) change its mechanical property? Cem Concr Res 2011;41(1):102–6. [30] Wang Q, Yang JW, Yan PY. Influence of initial alkalinity on the hydration of steel slag. Sci China Technol Sci 2012;55(12):3378–87.
347
[31] Taylor HFW. Cement chemistry. 2nd ed. London: Thomas Telford; 1997. [32] Richardson IG. The calcium silicate hydrates. Cem Concr Res 2008;38(2): 137–58.