Relation between Chemical Composition and Physical Properties of C-S-H Generated from Cementitious Materials

Transcription

1 Relation between Chemical Composition and Physical Properties of C-S-H Generated from Cementitious Materials Yuya Suda, Tats hik Saeki, Tsuyoshi Saito Journal of Advanced Concrete Technology, volume 13 ( 15 ), pp. 275 Nano-structural changes of C-S-H in Hardened Cement Paste during drying at 5 C Yoshimichi Aono, Fumiaki Matsushita, Sumio Shibata, Yukio Hama Journal of Advanced Concrete Technology, volume 5 ( 7 ), pp Cement reaction and resultant physical properties of cement paste Ippei Maruyama, Go Igarashi Journal of Advanced Concrete Technology, volume 12 ( 14 ), pp. -213

2 Journal of Advanced Concrete Technology Vol. 13, , May 15 / Copyright 15 Japan Concrete Institute 275 Scientific paper Relation between Chemical Composition and Physical Properties of C-S-H Generated from Cementitious Materials Yuya Suda 1*, Tatsuhiko Saeki 2 and Tsuyoshi Saito 3 Received 5 September 14, accepted 4 May 15 doi:.3151/jact Abstract Calcium silicate hydrate (C-S-H) is a dominant hydration product of cementitious materials. Therefore, its chemical composition and physical properties affect the performance of concrete. The purpose of this study is to investigate the chemical composition (CaO/SiO 2 molar ratio (Ca/Si ratio), and H 2 O/SiO 2 molar ratio (H 2 O/Si ratio)) and physical properties, such as density and specific surface area, of C-S-H. These factors are measured using synthesized C-S-H samples and C-S-H generated from various cementitious materials. Experimental results show that the H 2 O/Si ratio of C-S-H is proportional to the Ca/Si ratio independent of the mix proportion, curing temperature and type of binder. The density and specific surface area of C-S-H are affected by its Ca/Si ratio. A linear relationship is observed between the Ca/Si ratio and density of C-S-H independent of the mix proportion, curing temperature and type of binder. An inversely proportional relationship is found between the Ca/Si ratio and specific surface area of C-S-H. 1. Introduction 1 Assistant Professor, Department of Civil Engineering, National Institute of Technology, Toyota College, Toyota Japan. *Corresponding author, ysuda@toyota-ct.ac.jp 2 Professor, Department of Civil Engineering, Niigata University, Niigata, Japan. 3 Associate Professor, Department of Civil Engineering, Niigata University, Niigata, Japan. Calcium silicate hydrate (C-S-H) is a dominant hydration product of cementitious materials, hence its chemical composition and physical properties also affect the properties of concrete. However, the composition and physical properties of C-S-H, and the relationship between these, have not been entirely clarified. To accurately evaluate the performance of concrete, its physical properties should be determined with high accuracy. In particular, the strength, drying shrinkage and diffusion of gases and ions such as CO 2, O 2 and Cl - of concrete are controlled by its pore structure characteristics including pore volume, size distribution and tortuosity. Pores are formed in concrete by hydration products generated from cementitious materials. The structural characteristics of these pores are determined by the physical properties of hydration products, such as density and specific surface area. Therefore, it is important to clarify the chemical composition and physical properties of C-S-H, which is a dominant hydration product of cementitious materials. The purpose of this study is to estimate the chemical composition of C-S-H from its molar ratios of CaO/SiO 2 (Ca/Si ratio) and H 2 O/SiO 2 (H 2 O/Si ratio), and determine its physical properties, such as density and specific surface area. The relationship between the composition and properties of C-S-H will also be investigated using synthesized C-S-H and that generated from various cementitious materials. 2. Chemical composition of C-S-H This section discusses the relationship between the Ca/Si and H 2 O/Si ratios of C-S-H. It is known that the Ca/Si and H 2 O/Si ratios of C-S-H are influenced by environmental conditions such as temperature and humidity, curing age, mix proportion and type of materials. In general, the composition of C-S-H is expressed as C 1.7 SH 4. (Taylor 1997). The relationship between the Ca/Si and H 2 O/Si ratios of C-S-H has also been expressed as C x SH x+.63 (Fujii et al. 1983), C x SH x+1.7 (Taylor 1985), C x SH x+1.5 (Brouwers 4) and C x SH (19x-7)/17 (Richardson 14). In addition, this relationship has been evaluated by experiments and numerical calculation, such as molecular dynamics simulation and thermodynamic modeling (Kulik 11; Abdolhosseini et al. 14). However, the relationship between the Ca/Si and H 2 O/Si ratios varies among these studies. In addition, the effects of various conditions such as type of binder, mix proportion and curing conditions on the composition of C-S-H were not clarified. Therefore, the purpose of this section is to clarify the effect of various conditions on the Ca/Si and H 2 O/Si ratios of C-S-H. In this study, C-S-H and alite (C 3 S) were synthesized to evaluate the composition of C-S-H. Various cement pastes, consisting of ordinary portland cement (NC), moderate heat portland cement (MC), low heat portland cement (LC), blast furnace slag (BFS), fly ash (FA) and silica fume (SF), were also used. The effects of changing the water-to-binder ratio, replacement ratio of mineral admixture and curing temperature of cement paste were investigated.

3 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , Density Blaine LOI Table 1 Chemical composition of binders. Chemical composition (%) (g/cm 3 ) (cm 2 /g) (%) SiO 2 Al 2 O 3 Fe 2 O 3 CaO MgO SO 3 Na 2 O K 2 O C 3 S NC MC LC BFS FA-A FA-B SF-A SF-B Table 2 Mineral composition of each (a) cement and (b) FA. (a) C 3 S C 2 S C 3 A C 4 AF Gypsum Bassanite Calcite Periclase NC MC LC (b) Mulite Quartz Magnetite Gypsum Calcite Glass FA-A FA-B Table 3 Chemical composition (%) of glass in FA. SiO 2 Al 2 O 3 Fe 2 O 3 CaO MgO Na 2 O K 2 O FA-A FA-B Experimental procedure Materials Four C-S-H samples with a wide range of Ca/Si ratios were synthesized. The target Ca/Si ratios were.8, 1., 1.25 and 1.5. C-S-H samples were synthesized by reacting calcium hydroxide, ethyl silicate and water for 2 days at 4 C. Solid and aqueous phases were then separated by filtration. The concentrations of Ca and Si in the aqueous phase were measured by EDTA titration and spectrophotometer. Therefore, Ca/Si ratios were determined from the concentrations of Ca and Si in the aqueous phase before and after synthesis of C-S-H. Alite (C 3 S) was synthesized by reference to the chemical composition reported by Yamaguchi and Takagi (Yamaguchi and Takagi 1969). The chemical formula of alite (Ca 6 Mg 2 (Na 1/4 K 1/4 Fe 1/2 )O 36 (Al 2 Si 34 O 144 )) reported by Yamaguchi and Takagi is shown in Table 1. Alite was synthesized by conducting calcination 3 times in an electric furnace at the temperature of 16 C for 3 h. The amount of CaO in the synthesized alite was measured by the ethylene glycol method. The quantitative value of CaO was.47%. In addition, it was confirmed through measurement by powder X-ray diffraction (XRD) that the synthesized alite was pure tricalcium silicate, and it was evaluated to have been properly synthesized in this study. The physical properties (density and Blaine fineness) are shown in Table 1. NC, MC and LC were used as cement paste samples. In addition, NC was also replaced by BFS, two types of FA (FA-A and FA-B) and two types of SF (SF-A and SF-B). The physical properties and chemical compositions of each binder are listed in Table 1. The mineral compositions of NC, MC, LC and FA were measured by the XRD/Rietveld method (Table 2). The chemical composition of glass in FA was calculated from the chemical and mineral compositions of FA (Table 3) Mix proportion and curing conditions The experimental conditions used for the cement paste samples are shown in Table 4. Each sample was mixed for 3 min. Samples were remixed regularly to prevent bleeding. After mixing, each sample was placed in a plastic bottle (φ32 65 mm), and then cured at a specified temperature (5, or 4 C) for 1, 3, 7, 28 or 91 days for C 3 S, and 1, 3, 7, 28, 91, 182 or 365 days for cement Pretreatment and drying condition After each curing period, C 3 S paste and cement paste samples were broken into small pieces, and immersed in acetone for 2 days to prevent further hydration. Synthesized C-S-H was also immersed in acetone for 2 days. Samples were dried under decompression by aspirator at.1 kpa for 6 h to vaporize acetone. Moreover, these samples placed in a vacuum desiccator in the presence of silica gel at.1 kpa for 3 days to completely dry acetone. The dried C 3 S paste and cement paste samples were crushed into particles smaller than 9 μm. Samples were then dried at C- (using saturated LiCl solu-

4 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , Table 4 Experimental conditions and measured properties. Experimental conditions Measured properties Binder W/B (%) Replacement ratio (%) Temp. ( C) Composition Density Specific surface area C 3 S NC MC LC NC-BFS 45 5, , 5, , 7 NC-FA-A 45 15, 55 15, 65 15, NC-FA-B 55 15, NC-SF-A 45 4, 8 NC-SF-B 45 4 tion) in a vacuum desiccator for 1-2 weeks at.1 kpa or oven-drying at 1 C for 1-3 days until the mass no longer changed. In order to prevent carbonation, soda lime was placed in each vacuum desiccator. In this study, drying at 1 C is defined as % RH Analysis (1) Thermogravimetric analysis The amount of combined water in the dried samples was measured by thermogravimetric analysis (TGA). The amount of combined water in the samples dried at 11% RH was calculated from their weight reduction upon heating from 5 to C, while that in the sample dried at % RH was calculated from the weight reduction from 1 to C. To prevent the oxidation of sulfide in BFS, the highest temperature investigated for the hydrated samples containing BFS was 9 C. The amount of calcium hydroxide in each sample was determined from the mass loss between 415 and 5 C. (2) XRD/Rietveld method The amounts of unhydrated clinker and hydration products were measured by the XRD/Rietveld method. Samples dried at were used in XRD experiments. α-al 2 O 3 ( mass%) was mixed with each sample as an internal standard. XRD patterns were measured using a tube voltage of 4 kv, tube current of 4 ma, scan range of 2θ=5 to 7, step width of.2, and scan rate of 2 /min. The software used for Rietveld analysis was Topas4.2 (Bruker AXS). The target substances for quantification were alite (C 3 S), belite (β-c 2 S), aluminate phase (cubic-c 3 A and orthorhombic-c 3 A), ferrite phase (C 4 AF), portlandite (Ca(OH) 2 ), ettringite (C 6 A 3SO 3 H 32 ), monosulfate (C 4 A SO 3 H 12 ), hydrogarnet (C 3 AH 6 ), calcite (CaCO 3 ), periclase (MgO), lime (CaO), gypsum (CaSO 4 2H 2 O), bassanite (CaSO 4.5H 2 O), and corundum (α-al 2 O 3 ). In NC-BFS samples, hydrotalcite (M 6 A CO 3 H 12 ) was added as a target substance for quantification. Moreover, for NC-FA samples, quartz (SiO 2 ), mullite (Al 6 O 13 Si 2 ) and magnetite (Fe 3 O 4 ) were added as target substances for quantification. The amount of amorphous phase was calculated using Eq. (1) from the quantitative value of the internal standard (α-al 2 O 3 ). ( A R) G = A R (1) where G is the amount of amorphous phase (%), R is the amount of α-al 2 O 3 (%), and A is the value measured for α-al 2 O 3 (%). Previous results noted that the amount of calcium hydroxide measured by TGA was higher than that measured by XRD using an internal standard. This is because calcium hydroxide contains the amorphous phase (Bentur et al. 1979; Ramachandran 1979; Escalante-Garcia et al. 1999). From these studies, the amount of calcium hydroxide in each sample was measured by TGA in this study.

5 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , (3) Heat-treated XRD/Rietveld method The amount of unhydrated BFS was measured by the heat-treated XRD/Rietveld method (Sagawa et al. 6). The hydrated sample dried at was fired at 9 C for min to crystallize the amorphous phase of unhydrated BFS. The amount of unhydrated BFS was then measured by the XRD/Rietveld method. The target substances for quantification were gehlenite (Ca 2 Al 2 SiO 2 ), akermanite (Ca 2 MgSi 2 O 7 ), merwinite (Ca 3 MgSi 2 O 8 ) and α -C 2 S. The reaction ratio of BFS was calculated using Eq. (2): a+ b+ c Fa = 1 M (1 Ig ) (2) where F a is the reaction ratio of BFS (%), a, b and c are the quantitative values of gehlenite, akermanite and merwinite, respectively, M is the replacement ratio of BFS, and I g is the loss on ignition of unhydrated BFS. (4) Selective dissolution method The amount of unhydrated FA and SF was measured by the selective dissolution method (Ohsawa et al. 1999). A hydrated sample (1 g) dried at was added to a centrifuge tube containing aqueous HCl (2N, ml). The hydrated sample was shaken for 15 min in a water bath at 6 C. The sample was separated by centrifugation, and then the sample was washed with hot water. The tube containing the sample was filled with aqueous Na 2 CO 3 (5wt%, ml), and then shaken for min in a water bath at 8 C. The sample was separated by centrifugation and then washed with hot water. The residue was dried at 1 C. The reaction ratio of FA or SF was calculated from Eq. (3): X'(1 Ig ')/ k 2 Fb = 1 (1 Ig ) k1 (3) where F b is the reaction ratio of FA or SF, k 1 is the replacement ratio of FA or SF, k 2 is the residue extracted of unhydrated FA or SF, X is the residue extracted of hydrated FA or SF, I g is the loss on ignition of the hydrated sample, and I g is the loss on ignition of the extracted residue. Ca(OH) 2 Ca(OH) 2 Ca(OH) Silica gel θ/degree Fig. 1 XRD patterns of synthesized C-S-H. Table 5 Composition of synthesized C-S-H. Target Ca/Si ratio Obtained Ca/Si ratio Obtained H 2 O/Si ratio *1 Obtained H 2 O/Si ratio * * *1 Drying conditions: % RH (1 C dry) *2 Drying conditions: (saturated LiCl solution) *3 Quantitative value of calcium hydroxide is 4.81wt. %. 2.2 Experimental results Composition of synthesized C-S-H Figure 1 shows the XRD patterns of four synthesized C-S-H samples and silica gel synthesized from ethyl silicate. A peak from silica gel at 21 (2θ) was not observed in the patterns of C-S-H. When the Ca/Si ratio was 1.5, a peak was observed at 18 (2θ) from calcium hydroxide. Therefore, the amount of calcium hydroxide in C-S-H with a Ca/Si ratio of 1.5 was measured by TGA, as shown in Table Relation between the Ca/Si and H 2 O/Si ratios of C-S-H Figure 2 show the relationship between the reaction ratio and amount of combined water in C 3 S paste dried at RH of % and 11%. The amount of combined water in cement paste is generally used to indicate the development of hydration (Nagamatsu et al. 1986). Linear relationships are observed between the reaction ratio and combined water of C 3 S pastes independent of the water-to-binder ratio and curing temperature. Therefore, the chemical equation of hydration of C 3 S can be written as: ( ) ( ) acs + bh a CSH a CSH + cch (4) 3 3 n p q where n is calculated from the slope of the graph in Fig. 2 (n=b/a). In this study, n was 2.65 for % RH and 3.47 for. The Ca/Si ratio (p) and H 2 O/Si ratio (q) of C-S-H were calculated from the reaction ratio and amount of calcium hydroxide in C 3 S paste. In addition, the H 2 O/Si ratio (q) can be formulated as a function of the Ca/Si ratio from p=3-c/a and q=n-c/a, hence the relationship between the Ca/Si and H 2 O/Si ratios can be written as: q = p + ( n 3) (5) The composition of C-S-H obtained from these chemical equations is the temporal and spatial mean for C-S-H generated from the beginning of hydration until a specified age. Figure 3 shows the relationships between the Ca/Si and H 2 O/Si ratios of synthesized C-S-H generated from C 3 S under RH of % and 11%, and the previous results (Cong et al. 1996; Thomas et al. 3; Richardson 14). The relationships between the Ca/Si and H 2 O/Si ratios calculated from Eq. (5) are consistent with the results obtained for C-S-H synthesized at RH of % and 11%.

6 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , H 2 O/Si ratio (mol/mol) Synthesized C-S-H (% RH) Synthesized C-S-H () X. Cong et al. Thomas et al. (D-dry) Eq. (5) Eq. (5) % RH Richardson Ca/Si ratio (mol/mol) Fig. 3 Relationship between Ca/Si and H 2O/Si ratios of C-S-H. Thus, there is a linear relationship between the Ca/Si and H 2 O/Si ratios of C-S-H. In addition, the slopes obtained from Eq. (5) show the same tendency as the results of previous studies, especially research on C-S-H dried at % RH (Thomas et al. 3; Richardson 14). In particular, the relation between the Ca/Si ratio and H 2 O/Si ratio reported by Thomas et al. is the result of D-dried C-S-H, and that reported by Richardson is the results of C-S-H based on structure model. The structure model is based on C-S-H dried to the greatest possible extent without collapsing the tobermorite-like structure. It can probably be agreed that the relation between Ca/Si ratio and H 2 O/Si ratio of C-S-H with tobermorite-like structure was properly assessed in this study Relationship between Ca/Si and H 2 O/Si ratios in C-S-H The accuracy of the linear relationship between the Ca/Si and H 2 O/Si ratios of C-S-H obtained using synthesized C-S-H and C 3 S paste was evaluated by the composition of C-S-H generated from various cement pastes (Table 4). To evaluate the amount of hydration products in cement paste and quantify unmeasured hydration products such as C 2 AH 8, C 4 AH 13 and C 3 FH 6, the amount of hydration products in various cement pastes was calculated by mass balance in accordance with previous research Table 6 Composition of hydration products under different drying conditions. C-S-H CH AFt AFm HG CAH CFH HT (BFS) Drying conditions C % RH C % RH C % RH C % RH C % RH C % RH C % RH C % RH Composition C x SH C x SH x-.35 C x SH x+.47 CaO Ca(OH) 2 Ca(OH) 2 C 6 A 3SO 3 H C 6 A 3SO 3 H 7 C 6 A 3SO 3 H 32 C 4 A SO 3 H C 4 A SO 3 H 9 C 4 A SO 3 H 12 C 3 AH C 3 AH 6 C 3 AH 6 C 4 AH C 4 AH C 4 AH 13 C 3 FH C 3 FH 6 C 3 FH 6 M 6 A CO 3 H M 6 A CO 3 H 12 Molar mass (g/mol) 56.1x x (x-.35) 56.1x (x+.47 ) M 6 A CO 3 H 12 * CH: Calcium hydroxide, AFt: Ettringite, AFm: Monosulfate, HG: Hydrogarnet, CAH: C 4 AH 13, CFH: C 3 FH 6, HT: Hydrotalcite (Maruyama and Igarashi 14). A flowchart of the method used to calculate the amount of hydration products from the mass balance is presented in Fig. 4. Table 6 shows the composition of each hydration product considered as a target substance for quantification of the mass balance. The hydration products measured by Rietveld analysis and TGA were converted to anhydrites, and then the mass balance between starting materials and hydration products was calculated. The amounts of CAH and CFH were then Amount of combined water (% of binder at % RH) W/B: % ºC W/B: 4% ºC W/B: 5% ºC W/B: 4% 4 ºC W/B: 4% 4 ºC y=.1x 5 R 2 :.971 n: Reaction ratio of C3S (% of binder) (a) % RH Amount of combined water (% of binder at ) y=.273x 5 R 2 :.984 n: Reaction ratio of C3S (% of binder) (b) Fig. 2 Relationship between reaction ratio of C 3S and amount of combined water. (a) % RH, dry sample, (b), dry sample.

7 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , [No. 1] Amount of hydration products, unhydrated clinker and admixture measured by XRD/Rietveld, TGA and selective dissolution methods [No. 5] Calculate mass balance of Al 2O 3 and Fe 2O 3 between hydration clinker and admixture and hydration products [C-A-H and C-F-H were assumed to be C 4AH 13 and C 3FH 6] [No. 2] Convert hydration product into anhydrite [E.g. Ca(OH) 2 CaO] [No. 6] Calculate mass balance of CaO and SiO 2 between hydration clinker, admixture and hydration products [No. 3] Calculate amount of hydration clinker and admixture [C 3S, C 2S, C 3A, C 4AF, BFS, FA, SF] [No. 7] Calculate amount of C-S-H and Ca/Si ratio [No. 4] Calculate amount of reacted oxides in hydration clinker and admixture (CaO, SiO 2, Al 2O 3, Fe 2O 3) and amount of consumed oxides in hydration products (CaO, Al 2O 3) [No. 8] Calculate mass balance between hydration clinker, admixture and hydration products [No. 9] Determine amount of combined water at % and from composition of hydration products in Table 6 [No. ] Determine amount of hydration products at % and 11% Fig. 4 Flow chart showing method used to calculate mass balance. determined from the amount of residual oxides (Al 2 O 3 and Fe 2 O 3 ) that were not consumed as hydration products (CH, AFt, AFm, C 3 AH 6 and HT (BFS)). The amount of C-S-H and the Ca/Si ratio was determined from the residual CaO and SiO 2 that were not consumed as hydration products (CH, AFt, AFm, C 3 AH 6, HT (BFS), CAH and CFH). The amount of combined water in each hydration product was determined from the composition of each hydration product (Table 6). In this study, the chemical compositions of CAH and CFH were assumed to be C 4 AH 13 and C 3 FH 6, respectively. Therefore, the amount of combined water in each cement paste sample calculated from the composition of hydration products was compared with the amount of combined water determined by TGA. The amount of combined water in C-S-H was calculated from Eq. (5) by considering the effect of the Ca/Si ratio. It is reported that the Ca/Si ratio of C-S-H generated from BFS and FA is lower than that generated from Portland cement. The Ca/Si ratio generated from Portland cement is approximately 1.7 (Taylor, 1997). The Ca/Si ratio generated from BFS is approximately 1. (Richardson and Groves, 1992; Thomas et al. 12). Measurements using a scanning electron microscope (SEM) and transmission electron microscope (TEM) revealed that C-S-H with various Ca/Si ratios are distributed in blended cement paste (Taylor et al. ). However, the chemical composition of C-S-H generated from NC-BFS, NC-FA and NC-SF is expressed as a mean value in this study. For this reason, the C-S-H generated from each material, such as NC, BFS and FA, cannot be distinguished with this calculation method of phase composition. On the other hand, Richardson and Groves found that the mean value of Si/Ca ratio and Al/Ca ratio of C-S-H in NC-BFS paste measured by TEM is a linear relation (Richardson and Groves 1993). Sasaki and Saeki reported that the relation between Ca/Si and adsorbed chloride ion of synthesized C-S-H shows the same tendency as that of NC paste, NC-BFS paste, and NC-FA paste (Sasaki and Saeki 6). Moreover, a linear relation between the Ca/Si ratio and H 2 O/Si ratio was obtained from this study. This allows the water content of C-S-H with various Ca/Si ratios in cement paste to be evaluated from the mean Ca/Si ratio. Even though the chemical composition of C-S-H is evaluated by mean value, this calculation method of phase composition can reflect the difference of chemical composition of C-S-H. The evaluation of mean value is simple and easily handled in engineering. Therefore, in subsequent studies, the chemical composition of C-S-H is expressed by the mean value. Figure 5 shows the changes in phase composition of the various cement pastes at each age. In these figures, C-S-H can be seen to be the most abundant of the hydration products. In addition, when mineral admixtures were used, the amount of calcium aluminate hydrates increased in the hydration products. The Ca/Si ratio of C-S-H in various cement pastes is illustrated in Fig. 6. The Ca/Si ratio of C-S-H in NC-BFS and NC-FA pastes is also affected by the water-to-binder ratio and curing age. In addition, when the same admixture was used, the Ca/Si ratio in each sample decreased as the replacement ratio increased. The tendencies of the NC-BFS and NC-FA pastes are the same as those reported previously (Richardson and Groves 1992; Sakai et al. 4; Girão et al. ). Figure 7 compares the calculated and experimental values for the amount of combined water in cement pastes containing NC, MC, LC, BFS, FA and SF. The calculated values agree well with the experimental values.

8 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , CSH.8 CFH CAH.6 HT HG.4 AFm AFt CH.2 FA, BFS Cement Age (days) (a) Age (days) (b) Age (days) (c) Age (days) (d) Fig. 5 Change in the phase compositions at each age (a) NC, W/B=55%, Curing temperature ºC (b) LC W/B=55%, Curing temperature ºC (c) NC+FA, W/B=55%, Replacement ratio of FA=15%, Curing temperature ºC (d) NC+BFS, W/B=55%, Replacement ratio of BFS=5%, Curing temperature ºC. Phase composition (g/g) This confirms that the H 2 O/Si ratio of C-S-H is proportional to the Ca/Si ratio independent of mix proportion, curing temperature and type of binder. Although, in the case of the use of mineral admixture such as BFS, FA or metakaolin, it is known that Si in C-S-H was substituted for Al (Richardson and Groves 1992). In this study, the effect of Al substituted C-S-H (named as C-A-S-H) was not taken into consideration. Accordingly, in order to evaluation the effect of Si-Al substitution in C-A-S-H for phase composition, a part of Si in C-S-H was made to substitute Al. For the calculation method, after SiO 2 (molar) was calculated by mass balance in the flow chart [No. 6] in Fig. 4, Al was added to Si. Once more, the phase compositions were determined by mass balance. The additive rate of Al (molar) was % (BFS) or 7 % (NC, MC and LC) of Si (molar), based on previous research results (Richardson and Groves. 1992; Girão et al. 7). Figure 8 shows the relation between the Al/Ca molar ratio and Si/Ca molar ratio. The Al/Ca molar ratio increases with increasing Si/Ca molar ratio. The relation in this study shows the same tendency as the results of previous research (Richardson and Groves, 1993). Figure 9 shows the Ca/Si ratio (mol/mol) d 91d 365d NC BFS 5% 1.6 BFS 7% FAA15% FAA% Replacement ratio W/B (%) W/B (%) W/B (%) Fig. 6 Ca/Si ratio of C-S-H in different cement pastes. comparison between the calculated values as Al substituted C-S-H and calculated values as C-S-H for the amount of combined water in cement pastes. As shown in Fig. 9, the correlation factor is.968. The calculated value as C-A-S-H is in good agreement with the calculated value as C-S-H. Hence, it is thought that the phase compositions are seldom affected by Al substituted C-S-H. Combined water (%) -test- 4 N4 ºC N55 ºC M4 ºC M55 ºC L4 ºC L55 ºC N55 5 ºC N55 4 ºC W/B Temp. Combined water (%) -test- 4 BFS35% BFS5% BFS7% FAA15% FAA% FAB15% FAB% R:.977 R:.94 Replacement 4 4 ratio Combined water (%) -calculated- Combined water (%) -calculated- (a) (b) BFS, FA, SF Fig. 7 Comparison of calculated and experimental values of combined water in each cement paste. (a) NC, MC and LC paste dried at, (b) NC-BFS, NC-FA and NC-SF paste dried at. SFA4% SFA8% SFB4%

9 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , Al/Ca ratio (mol/mol) Richardson NC-BFS (Al/Si: %) NC, MC, LC (Al/Si: 7%) Si/Ca ratio (mol/mol) Fig. 8 Relationship between Si/Ca and Al/Ca ratios of C-S-H. 3. Physical properties of C-S-H In this section, the density and specific surface area of C-S-H are evaluated. Previous research reported that the density of C-S-H is influenced by its composition (Beaudoin et al. 1978). However, the density of C-S-H reported in the literature varies widely because of different drying conditions (Skalny et al. 198; Taylor 1997; Bentz et al. 1997). Moreover, the relationship between the composition and density of C-S-H is not fully understood. With regard to the specific surface area of C-S-H, it is known that a lower Ca/Si ratio tends to lead to a higher specific surface area (Beaudoin et al. 1978; Sasaki et al. 6). However, these tendencies were observed in results obtained for synthesized C-S-H studied by nitrogen adsorption. The properties of C-S-H in cement paste have not been studied sufficiently. In addition, it has been reported that specific surface areas obtained with the nitrogen absorption method differ from those obtained with the water adsorption method (Daimon et al. 1976; Jennings ; Odler 3). Therefore, the purpose of this study is to obtain fundamental data about the density and specific surface area of C-S-H. The samples for measurement were synthesized hydration products, C 3 S paste and various cement pastes. 3.1 Experimental procedure Materials, mixture and curing conditions C-S-H, ettringite, monosulfate, C 3 AH 6 and C 4 AH 13 were synthesized for subsequent measurement. The method used to synthesize C-S-H is described in Section Ettringite and monosulfate were synthesized by reaction of C 3 A and CaSO 4 2H 2 O. C 3 A was first prepared by mixing CaCO 3 and Al 2 O 3 in the molar ratio of 3:1, and then firing at 14 C for 3 h in an electric furnace. Ettringite and monosulfate were then synthesized by mixing C 3 A and CaSO 4 2H 2 O in the molar ratios of 1:3 and 1:1, respectively, with a water-to-solid ratio of. Samples were cured for 14 days at C. C 3 AH 6 was synthesized by reaction of C 3 A and water with a water-to-solid ratio of 1.. Samples were cured for 28 days at 4 C. Combined water (%) -calculated for C-A-S-H- 4 NC, MC, LC, BFS R: Combined water (%) -calculated for C-S-H- Fig. 9 Comparison of calculated (as C-S-H or C-A-S-H) values of combined water in each cement paste. C 4 AH 13 was synthesized according to a reported method (Werner 1967) by mixing NaAlO 2 and CaO with the molar ratio of 1:2 in water with the water-to-solid ratio of.. Samples were cured for 28 days at C. The density and specific surface area of commercial calcium hydroxide (Kanto Chemical Co., Inc.) and hydrotalcite (Wako Pure Chemical Industries, Ltd.) were also measured. Table 4 shows the combination of measurement items of density and specific surface area of the C 3 S paste and cement paste samples. Hydrated samples for density measurement were cured for 28 and 91 days, whereas samples for specific surface area measurement were cured for 1, 3, 7, 28, 91, 182 and 365 days Analysis (1) Density of synthesized hydration products and cement pastes The synthesized hydration products and cement pastes samples were dried at and then crushed into particles smaller than 9 μm. The crushed cement paste samples were also finely milled, and dried again at 11% RH before their density was measured using a gas pycnometer with helium gas. (2) Porosity of cement pastes Hardened cement paste samples cured for 1, 3, 7, 28, 91, 182 and 365 days were broken into pieces of about 5 mm to measure porosity. Porosity was calculated from the weight difference between saturated surface-dried samples and those dried at 5 C for 3 days. These hydrated samples were saturated by pouring water over them under reduced pressure. After saturation, the surface water was removed from saturated samples by wiping. (3) Specific surface area of synthesized hydration products and cement pastes The specific surface areas of the synthesized hydration products and cement paste specimens were calculated from the amount of water adsorbed at RH of 11%, 22% and 33%, as determined by gravimetric analysis. Samples were prepared by crushing them into particles smaller than 9 μm. Samples were dried at % RH until

10 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , Table 7 Physical properties of synthesized hydration products. Density (g/cm 3 ) Specific surface area (m 2 /g) This study Taylor (1997) Balonis (9) This study C-S-H Calcium hydroxide Ettringite Monosulfate Hydrogarnet C 4 AH Hydrotalcite their water vapor sorption was measured. LiCl (), CH 3 COOK (22% RH) and MgCl 2 6H 2 O (33% RH) were used as saturated salts. The specific surface area of each sample was calculated by the BET method (Brunauer et al. 1938). To evaluate the specific surface area of C-S-H, the total amount of combined and adsorbed water in C 3 S paste at each RH was calculated by adding the amount of water adsorbed at RH of 11%, 22% and 33% to the amount of combined water in each sample at % RH. A model of the combined water + adsorbed water in a sample under each RH is shown in Fig Experimental results Density of synthesized hydration products The density of synthesized hydration products are presented in Table 7. When the Ca/Si ratio of synthesized C-S-H decreased, the density of C-S-H also decreased. The densities of ettringite, monosulfate, hydrogarnet and C 4 AH 13 are consistent with those reported previously (Taylor 1997; Balonis et al. 9) Relationship between Ca/Si ratio and density of C-S-H generated from cementitious materials As decreases in the Ca/Si ratio of synthesized C-S-H lead to decreases in density, the density of C-S-H in cement paste was evaluated. The density of C-S-H generated by various cementitious materials was calculated from Eq. (6): ρ C S H = mc S H 1 mi P ρ i (6) where ρ C-S-H is the density of C-S-H (g/cm 3 ), m C-S-H is the amount of C-S-H (g/g of sample dried at ), P is the density of cement paste dried at (g/cm 3 ), m i is the amount of each hydration product (except C-S-H), unhydrated clinker mineral and unhydrated admixture (g/g of sample dried ), and ρ i is the density of each hydration product (except C-S-H), unhydrated clinker mineral and unhydrated admixture (g/cm 3 ). The densities of hydration products except C-S-H are listed in Table 7. The amounts of unhydrated clinkers, admixtures and hydration products in cement paste were calculated from the mass balance described in Section The density of C 3 FH 6 was assumed to be 2.81 g/cm 3 (Balonis et al. 9). The relationships between the density and Ca/Si ratio of C-S-H generated from various cementitious materials are presented in Fig. 11. Linear relationships are observed independent of mix proportion, curing temperature and type of binder. A linear equation for this data was obtained using regression analysis, which can be written as: ρc S H=.459 ( Ca Si ) (7) This allows the density of C-S-H with various Ca/Si ratios in hardened cement paste to be evaluated from the mean Ca/Si ratio, because the relationship is expressed linearly. The correlation factor is.844. Figures 12 (a), Dry at 1 C TGA Combined water Water adsorption i% RH Adsorbed water Combined water + i% RH Adsorbed water i% RH: 11%, 22%, 33% Fig. Model of combined water at different relative humidity. 2.6 y=.459x R 2 = Cement paste 1.8 NC, MC, LC NC-BFS 1.6 NC-FA 1.4 NC-SF C3S paste 1.2 C-S-H Ca/Si ratio (mol/mol) Fig. 11 Relationship between density and Ca/Si ratio of C-S-H generated from various cementitious materials. Density of C-S-H (g/cm 3 )

11 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , Density of C-S-H (g/cm 3 ) Eq.(7): This study Density of C-S-H (g/cm 3 ) nm Tobermorite y=.242x Eq.(7): T his study 1.4nm Tobermorite Metajenite Richardson Ca/Si ratio (mol/mol) (a) Various H2O/Si ratio Ca/Si ratio (mol/mol) (b) Normalized H2O/Si ratio Fig. 12 Relationship between density and Ca/Si ratio of C-S-H reported by previous research. (a) Various H 2O/Si ratio, (b) Normalized H 2O/Si ratio. Jennite Allen et al. 7 Thomas et al. Beaudoin et al Muller et al. 13 Pellenq et al. 9 Manzano et al. 12 Taylor 1997 Young and Hansen 1987 Benz 1997 (b) show the relation between the density and Ca/Si ratio of C-S-H reported in previous studies (Allen et al. 7; Thomas et al. ; Beaudoin et al. 1998, Muller et al. 13; Pellenq et al. 9; Manzano et al. 12; Taylor 1997; Young and Hansen 1987; Benz 1997; Richardson 14). As shown in Fig. 12 (a), at the same Ca/Si ratio, the density of C-S-H reported by researchers differs. This is because the hydrated samples were dried under various condition, such as 1ºC, d-drying, and saturated condition. A major factor was also the fact that the H 2 O/Si ratio of each C-S-H sample was not standardized. In this study, in order to standardize the H 2 O/Si ratio of each C-S-H sample, the H 2 O/Si ratio of each C-S-H was normalized by Eq. (5). Figure 12 (b) shows the relation between the density and Ca/Si ratio of C-S-H normalized by Eq. (5). Similar to this study, the densities reported in former studies decreased with decreasing Ca/Si ratios (Fig. 11). However, the relation between the density and Ca/Si ratio in this study differs from past research results. When the Ca/Si ratio is the same value, densities in this study are smaller than past research results, possibly because small pores and closing pores were not evaluated in this study. Therefore, to estimate the engineering adequacy of Eq. (7), the porosity measured after drying at 5 C was compared with the theoretical porosity calculated from the amount and density of hydration products, unhydrated binder and unhydrated admixture. Densities were measured at. In this case, the porosity after drying at differed from that of samples dried at 5 C. However, as shown in Fig. 13, the amount of combined water in the sample dried at was very nearly equal to that in the sample dried at 5 C. Therefore, it was assumed that the samples dried at and 5 C were of equal porosity. The theoretical porosity of each cement paste dried at was calculated from the amount and density of hydration products, unhydrated clinkers and unhydrated mineral admixtures. The densities of the hydration products shown in Table 7 were used, while the density of C-S-H was calculated using Eq. (7). This means that Combined water (% of binder dried at 5 ºC) 4 NC BFS FA R: Combined water (% of binder dried at ) Fig. 13 Comparison of amount of combined water at 11% RH and after drying at 5 C. the density of C-S-H was calculated considering the change in Ca/Si ratio shown in Fig. 6. Figure 14 compares the theoretical porosity calculated from the amount and density of hydration products dried at and that measured following drying at 5 C. Figures 14 (a) and (b) show the porosity evaluated by Eq. (7) (y=.459x+1.392). Figures 14 (c) and (d) show the porosity evaluated based on previous research results (y=.242x+1.978). As shown in Figs. 14 (a) and (b), the change in theoretical porosity with Ca/Si ratio agrees with experimental values. The correlation factors are.947 (a) and.927 (b). The slopes of the regression equation are 1.1 (a) and 1. (b). Therefore, the linear relationship between the Ca/Si ratio and density of C-S-H is adequately described by Eq. (7). As shown in Figs. 14 (c) and (d), the correlation factors are.95 (c) and.97 (d). The slopes of the regression equation are.94 (c) and.956 (d). The values calculated based on previous research results (y=.242x+1.978) were larger than the experimental value due to these results. This is because the density values in the earlier research results are larger than the densities measured in this study. The differences between this study and previous research on

12 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , Porosity (%) -test- 6 N4 ºC Density: Eq. (7) N55 ºC 5 M4 ºC M55 ºC 4 y=1.1x R: Porosity (%) -calculation- (a) NC, MC, LC paste L4 ºC L4 ºC N55 5 ºC N55 4 ºC Porosity (%) -test- 6 5 Density: Eq. (7) BFS35% BFS5% BFS7% FAA15% 4 y=1.x R: Porosity (%) -calculation- (b) BFS, FA, SF FAA% FAB15% FAB% SFA4% SFA8% SFB4% Porosity (%) -test- 6 N4 ºC Density: Reference value N55 ºC 5 M4 ºC M55 ºC 4 y=.94x R: Porosity (%) -calculation- (c) NC, MC, LC paste L4 ºC L4 ºC N55 5 ºC N55 4 ºC Porosity (%) -test- 6 5 Density: Reference value BFS35% BFS5% BFS7% FAA15% 4 y=.956x R: Porosity (%) -calculation- (d) BFS, FA, SF FAA% FAB15% FAB% Fig. 14 Comparison of calculated and experimental porosity. (a) NC, MC and LC pastes with Eq. (7), (b) NC-BFS, NC-FA and NC-SF pastes with Eq. (7), (c) NC, MC and LC pastes with earlier research, (d) NC-BFS, NC-FA and NC-SF paste with earlier research. Specific suraface area (m 2 /g) 15 5 W/B: 35% Replacement ratio: 15% Replacement ratio: 35% W/B: 45% Replacement ratio: % Replacement ratio: 5% W/B: 55% Replacement ratio: 7% Age (days) Age (days) Age (days) (a) NC (b) NC-FA (c) NC-BFS Fig. 15 Change of specific surface area at each age (a) NC paste, Curing temperature: ºC (b) NC-FA paste W/B: 55%, Curing temperature: ºC (c) NC-BFS paste, W/B: 55%, Curing temperature: ºC. SFA4% SFA8% SFB4% the density of C-S-H is considered to be the result of differences among the various experimental methods employed, including the Archimedes method, helium pycnometer method, molecular dynamics method and small-angle neutron scattering method. However, the porosity of hardened cement paste can be evaluated from Eq. (7) in engineering Specific surface area of cement paste Figures 15 (a), (b) and (c) show the change of specific surface area of cement paste samples at each age. The specific surface of each cement paste increases with curing age. As shown in Fig. 15 (a), even though the W/B ratios differed, the specific surface area of the samples were the same value until 7 days. After 7 days of curing, increases in W/B tend to lead to higher specific surface area. The specific surface area at each age was affected by the replacement ratio of mineral admixture in Figs. 15 (b), (c). For pretreatment of specific surface area measurements, samples were oven dried at 1ºC for 1-3days after immersion in acetone (Section 2.1.3). Previous studies have pointed out that the capillary pressure due to surface tension of the removed water caused by

13 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , oven-drying at high temperature causes the collapse of small pores such as gel pores in C-S-H (Juenger et al. 1; Korpa et al. 6; Snoeck et al. 14). Compared with other drying methods, the oven drying method causes the specific surface area decrease owing to the collapse of small pores as the result of capillary pressure. On the other hand, the solvent exchange method, using solvents such as methanol and isopropanol, preserves the microstructure due to the reduction of capillary pressure on drying by the decrease of surface tension (Juenger et al. 1; Collier et al. 8; Snoeck et al. 14). However, in this study, the values of specific surface area (NC, W/B=45% and 55%, Age: 6 months) were 133 m 2 /g (W/B=45%) and 139 m 2 /g (W/B=55%). These values are in agreement with the value (131 m 2 /g) of the saturated sample (NC, W/C=5%, Age: 6 months) reported in a prior study (Snoeck et al. 14). Additionally, other features of specific surface area, such as differences by W/B and age, are the same as reported previously (Maruyama et al. 14). These results might be due to the preservation of the microstructure by acetone exchange. Therefore, in this study, it seems possible to investigate evaluation methods and the relative effect of various features with the specific surface area Specific surface area of hydration products The BET specific surface areas of synthesized hydration products are presented in Table 7, and increase as the Ca/Si ratio decreases. The specific surface area of calcium hydroxide and hydrogarnet are lower than those of other hydration products. The specific surface area of ettringite, monosulfate and C 4 AH 13 are similar Specific surface area of C-S-H Figure 16 depicts the relationship between the reaction ratio and amount of combined water + adsorbed water in C 3 S paste at different RH. At the same RH, there is a linear relationship between the reaction ratio and amount Combined water + Adsorbed water (%) 4 % ºC 4% ºC 5% ºC 4% 5 ºC 4% 4 ºC y=.244x R 2 :.977 y=.1x R 2 :.971 % RH y=.232x R 2 :.975 y=.254x R 2 : % RH 33% RH Reaction ratio (%) Reaction ratio (%) Fig. 16 Relationship between the reaction ratio of C 3S and combined water at different relative humidities. of combined water + adsorbed water in C 3 S. Therefore, the H 2 O/Si ratio of C-S-H at each RH can be expressed as a function of the Ca/Si ratio from Eq. (5). The values of n are shown in Fig. 16. The amount of water adsorbed by C-S-H can be calculated from the composition of C-S-H at each RH using the equation: A C S H ( q q ) 18. i% RH %RH = 56.1 p q %RH (8) where A C-S-H is the amount of water adsorbed by C-S-H (g/g of C-S-H dried at % RH), q i%rh is the H 2 O/Si ratio at each RH (11%, 22% and 33%), q %RH is the H 2 O/Si ratio at % RH, and p is the Ca/Si ratio. Equation (8) indicates that the amount of water adsorbed by C-S-H can by expressed as a function of the Ca/Si ratio. Figure 17 shows the relationship between the Ca/Si ratio and amount of adsorbed water in C-S-H. The amount of water adsorbed by C-S-H at each RH increases with decreasing Ca/Si ratio. This relationship indicates that a lower Ca/Si ratio increases the specific surface area of C-S-H. The specific surface area of samples with different Ca/Si ratios was calculated using BET plots. The relationship between the Ca/Si ratio and specific surface area of each sample can be formulated using regression analysis. In this study, the logarithm function (y=a-b log(x)) was used as the regression expression because it gave the highest correlation factor of the expressions investigated. The relationship between Ca/Si ratio and specific surface area can be written as: SC S H= log( Ca Si) (9) where S C-S-H is the specific surface area of C-S-H (m 2 /g). The correlation factor is.998. The relationship between the Ca/Si ratio and specific surface area calculated from Eq. (9) is shown in Fig. 18. The relationship between the Ca/Si ratio and density of C-S-H is linear, whereas an inversely proportional relationship is observed between the Ca/Si ratio and specific surface area of C-S-H. The specific surface area calculated from Eq. (9) gives the temporal and spatial mean of C-S-H generated from the Amount of adsorbed water (g/g of C-S-H dried at % RH) % RH 33% RH Ca/Si ratio (mol/mol) Fig. 17 Relationship between Ca/Si ratio and amount of adsorbed water at different relative humidities.

14 Y. Suda, T. Saeki and T. Saito / Journal of Advanced Concrete Technology Vol. 13, , Table 8 Distribution range of Ca/Si ratio and mean Ca/Si ratio from the literature. Author(s) Materials, mixture and curing age Range of Ca/Si Mean Ca/Si Measurement method Stucke and Majumdar Pure C 3 S, W/C:.6, days SEM-EDS Diamond Pure C 3 S, W/C:.4, 3 1/2 years SEM-EDAX Williams et al. FA-CH, 5:5 CH/FA, W/C:.8, 78 days SEM-EDS C 3 S, W/C:.4, 3 days C 3 S, W/C:.4, 91 days Ogawa et al. C 3 S-white clay, 6:4 C 3 S/clay, W/C:.4, 3 days FESEM-EDX C 3 S-white clay, 6:4 C 3 S/clay, W/C:.4, 91 days Girão et al. White cement-fa-koh, 7: cement/fa, W/C:.5, 28 days TEM-EDX OPC-GGBS, 9: OPC/GGBS, W/C:.4, years Taylor et al. OPC-GGBS, 5:5 OPC/GGBS, W/C:.4, years TEM-EDX OPC-GGBS, 25:75 OPC/GGBS, W/C:.4, years beginning of hydration until a specified curing time. In addition, the relationship between the Ca/Si ratio and specific surface area in Eq. (9) cannot calculate the local specific surface area of C-S-H with various Ca/Si ratios. This is because the relationship was expressed as a nonlinear equation. However, the specific surface area calculated from Eq. (9) agrees well with the experimental results obtained for synthesized C-S-H shown in Fig. 18. To investigate the engineering applicability of Eq. (9), S C-S-H calculated using Eq. (9) from the mean Ca/Si ratio of C-S-H was compared with S C-S-H D calculated using Eq. () from the distribution of various Ca/Si ratios in C-S-H as follows: S 1 = S () n C S H D C S H i S C-S-H i is calculated using Eq. (9) from the local Ca/Si ratio in C-S-H, and n is the number of local Ca/Si ratios of C-S-H. The distribution of local Ca/Si ratios in C-S-H in cement paste was compared with reported results (Stucke et al. 1977; Diamond 1976; Williams et al. 2; Ogawa et al. 198; Girão et al. ; Taylor et al. ), Specific surface area of C-S-H (m 2 /g of C-S-H dried at % RH) 5 4 Eq. (9) Synthesized C-S-H Ca/Si ratio (mol/mol) Fig. 18 Relationship between Ca/Si ratio and specific surface area of C-S-H. as shown in Table 8. The results in this table were chosen to represent different combinations of materials, mix proportions and curing times. Figure 19 compares S C-S-H calculated using Eq. (9) with the mean Ca/Si ratio of C-S-H and S C-S-H calculated using Eq. (9) from the distribution of Ca/Si ratios in C-S-H. As shown in this figure, S C-S-H calculated using Eq. (9) from the mean Ca/Si ratios of C-S-H agreed well with that considering the distribution of Ca/Si ratio in C-S-H. Therefore, the specific surface area of C-S-H can be evaluated from the mean Ca/Si ratio in engineering Evaluation of specific surface area of cement paste To verify the applicability of Eq. (9) and the accuracy of the specific surface area of each hydration product, the specific surface areas of cement paste were compared with calculated values, which were determined from the integrated values of the amount and specific surface area of various hydration products in cement paste. The S C-S-H D calculated by distribution of Ca/Si ratio (m 2 /g) Stucke and Majundar Diamond Williams et al. Ogawa et al. Girao et al. Taylor et al S C-S-H calculated by average Ca/Si ratio (m 2 /g) Fig. 19 Comparison of SC-S-H calculated from mean Ca/Si ratio of C-S-H and that calculated from distribution of Ca/Si ratio of C-S-H.