slag paste

Construction and Building Materials 81 (2015) 303–312

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Reactivity and reaction products of alkali-activated, fly ash/slag paste N.K. Lee, H.K. Lee ⇑ Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology, Guseong-dong, Yuseong-gu, Daejeon 305-701, South Korea

h i g h l i g h t s  The amount of added slag affected the formation of reaction product.  As the amount of added slag increased, the amount of C-S-H gel increased.  The aluminosilicate gel with the Q4 units was similar to a Ca-based geopolymer.

a r t i c l e

i n f o

Article history: Received 23 August 2014 Received in revised form 14 January 2015 Accepted 18 February 2015

Keywords: Alkali activated fly ash/slag Reactivity Reaction product Ca-based geopolymer N-C-A-S-H

a b s t r a c t Few studies have described the reaction products of an alkali-activated, two-source binder and their characteristics due to the complexity of the mechanism. In this study, the microstructure, reaction products, and reactivity of alkali-activated, fly ash/slag binders synthesized at various mixture ratios of two raw materials were examined using various experimental techniques (NMR, ICP-OES, EDS, FT-IR and TGA) to systematically investigate the complex reaction mechanism of the binders. It was also intended to help assess durability of the binders. It was found that the amount of added slag primarily affected the amount of reaction product and its silicate structure, and as the amount of added slag increased, the amount of C-S-H gel increased and the amount of aluminosilicate gel decreased. Considering chemical composition and silicate structure, the aluminosilicate gel with the Q4 (nAl) units was found to be similar to a Ca-based geopolymer (N-C-A-S-H). The chemical composition ratios of the Ca-based geopolymer were nearly the same as those of C-S-H whereas their silicate structures were different. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Geopolymer materials are known as alternatives to ordinary Portland cement (OPC); the geopolymer binders can reduce CO2 emissions by 80% compared to the OPC [11]. The main sources of the geopolymers are metakaolin, fly ash, and blast furnace slag. Reaction products formed through an alkali-activation process are completely different depending on chemical composition of the raw sources [24]. The main reaction product of alkali-activated blast furnace slag, which is rich in Ca, is C-S-H gel with a low Ca/Si ratio (0.8–1.1) [24,16,17], whereas the main reaction product of alkali-activated metakaolin or fly ash, which is rich in Si and Al, is amorphous aluminosilicate gel with a three-dimensional framework of SiO4 and AlO4 tetrahedra linked through shared O atoms [35,41,10]. Several studies have reported the alkali-activation of twosource mixtures that consist of fly ash/slag or metakaolin/slag [37]; [47,31,3,38,5,19]. The addition of blast furnace slag in the ⇑ Corresponding author. Tel.: +82 42 350 3623; fax: +82 42 350 3610. E-mail address: [email protected] (H.K. Lee). http://dx.doi.org/10.1016/j.conbuildmat.2015.02.022 0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.

one-source mixture was effective for improving the mechanical properties of the geopolymers made with metakaolin [43,44] and fly ash [23]. The authors investigated mechanical properties such as compressive strength, elastic modulus, splitting tensile strength, and setting properties of alkali-activated, fly ash/slag concrete at various slag contents, and concluded that the slag content of 20– 30% relative to the total binder weight was optimal considering setting, flowability, and strength [27,28,20]. There are two different reaction products of the alkali-activated, two-source binder; both products had higher reactivity than that of the one-source binder. Puertas et al. [37] showed that two different reaction products existed in alkali-activated pastes. Correspondingly, Buchwald et al. [6] showed in an NMR study that alkali-activated metakaolin/slag binders had higher reactivity than the alkali-activated metakaolin binder, and established the types and quantities of the reaction products. The main reaction products were calcium silicate hydrate rich in Al that contains Na in its structure and alkaline aluminosilicate hydrate with a threedimensional structure [37]. Yip and Van Deventer [45] and Buchwald et al. [6] concluded, after SEM observations and NMR spectroscopic studies, that the C-S-H and aluminosilicate gel

304

N.K. Lee, H.K. Lee / Construction and Building Materials 81 (2015) 303–312

coexist in the alkali-activated metakaolin/slag binders. In addition, the chemical interaction may arise between the elements released by the dissolution of fly ash and slag particles [25], and the two phases (C-S-H and aluminosilicate) may not only be combined but also may chemically interact [6,25]. On the contrary, the authors stated that the discrete formation of C-S-H gel and aluminosilicate gel in an SEM length scale was not observed from their experiment, and the separation between calcium-rich and calcium-deficient regions was not identified [26]. Nevertheless, if the separation is observed, it may be due to inadequate mixing during the manufacturing process [26]. Recently, it was found that C-A-S-H gel and N-A-S-H gel are reaction products in a fly ash/slag geopolymer [19,5,29], and the N-C-A-S-H gel may be a hybrid-type phase of the N-A-S-H gel and the C-A-S-H gel [19]. The alkali-activation of a two-source mixture has a more complex mechanism than that of a one-source mixture. Few studies have described the reaction products of an alkali-activated, twosource binder and its characteristics due to the complexity of the mechanism [5,38,19]. Quantitatively evaluating the reaction products and reactivity of alkali-activated, fly ash/slag binder synthesized at various mixture ratios of two raw materials is important to figure out the complex mechanism. In the present study, the microstructure, reaction products, and reactivity of alkali-activated, fly ash/slag binders synthesized at various mixture ratios of two raw materials were examined using various experimental techniques (NMR, ICP-OES, EDS, FT-IR and TGA) in order to systematically investigate the complex reaction mechanism of the binders. The test results will help assess the durability of the alkali-activated, fly ash/slag binders with different amounts of the reaction products and different reactivity levels. 2. Experimental program 2.1. Materials Class F fly ash (containing more than 70% pozzolanic compounds (SiO2, Al2O3, and Fe2O3) in accordance with ASTM C618) and blast furnace slag were used as binder materials. The chemical compositions of these materials are listed in Table 1. Two types of alkali-activators were prepared: sodium silicate powder with a chemical composition of SiO2 (53.4%) and Al2O3 (25.2%), a bulk density of 0.62 g/ 3 cm and a molar ratio of 2.18 (Ms = SiO2/Na2O); and sodium silicate liquid (SiO2/Na2O = 1.0) mixed with 4 M NaOH and water glass (Korean industrial standards (KS) Grade-3; SiO2 (29%), Na2O (10%), H2O (61%), specific gravity 1.38 g/mL). Distilled water was used to dissolve the solid NaOH (98%); the latter alkali activator was prepared by mixing the NaOH solution with water glass at the SiO2/Na2O ratios developed by the authors [27].

Table 2 Mix proportion of alkali-activated fly ash/slag paste.

a

Sample

Water/ bindera

SiO2/Na2O ratio of sodium silicate

Slag/(fly ash + slag)a

Type of sodium silicate

PS10 PS30 PS50 LS10 LS30 LS50

0.4 0.4 0.4 0.4 0.4 0.4

2.12 2.12 2.12 0.94 0.94 0.94

0.1 0.3 0.5 0.1 0.3 0.5

Powder Powder Powder Liquid Liquid Liquid

All values are given as mass ratios.

use of liquid sodium silicate, fly ash, and slag were dry-mixed and then the liquid sodium silicate was added to the mixture. The AFS paste samples prepared by the process were immediately cast into 50 mm cubic molds. All of the samples were cured at 20 °C and at a relative humidity of 50% in a room with constant temperature and humidity levels. After one day, the samples were removed from their molds and were stored in a conditioning room until the day of testing. 2.3. Experimental details The AFS samples for the X-ray powder diffraction (XRD) test were prepared by mechanical grinding. The XRD data were recorded on a Rigaku D/MAX-2500 machine using Cu Ka radiation at a scanning rate of 2°/min from 2° to 160° in 2h. To analyze the polished surfaces of the samples on a NOVA 230 device (FEI Company), scanning electron microscopy (SEM) with back-scattered electron (BSE) images and energy-dispersive spectroscopy (EDS) analyses were performed. The samples were impregnated by using low-viscosity epoxy resin, polished with SiC paper, and coated with gold. Powdered AFS samples were dried at 80 °C in an oven for 1 day before the thermogravimetric analyses (TGA). The TGA measurements were applied to the powdered samples under N2 gas at 10 K/min up to 900 °C. Inductively coupled plasma optical emission spectroscopy (ICP-OES) was used to determine the water-soluble silicon, aluminum, calcium, and sodium contents of the AFS samples in distilled water. The powdered AFS samples were sonicated in distilled water for more than 2 h and the amounts of water-soluble elements were then measured by means of ICP-OES. The reactivity of the alkali-activator (sodium silicate powder or liquid) was evaluated indirectly through this quantitative analysis. 29 Si nuclear magnetic resonance (NMR) spectroscopy (Bruker AVANCE 400WB) was used to record the solid-state NMR spectra with the purpose of evaluating the reactivity and content of the reaction products (i.e., C-S-H gel and aluminosilicate gel. The 29Si resonance frequency was 79.42 MHz and the spinning rate was 5 kHz. All of the measurements were taken at room temperature with tetramethylsilane (TMS) as an external standard. The spectra were acquired using a pulse length of 1.5 ls, and a short repetition time of 20 s was chosen. The powdered samples were also analyzed by Fourier transform infrared (FT-IR) spectroscopy (Model FT-IR 4100, JASCO, Japan).

3. Results

2.2. Mixture proportions The mixture proportions are provided in Table 2 and are labeled with specific codes. The labels ‘P’, ‘L’ and ‘S’ represent powder, liquid, and slag, respectively. The numbers ‘10’, ‘30’, and ‘50’ refer to the percentage of slag to the total binder (fly ash + slag) by mass. The method for producing the alkali-activated, fly ash/slag (AFS) paste is as follows: In the case of use of powder sodium silicate, fly ash, slag, and powder sodium silicate were dry-mixed for 2 min to ensure the homogeneity of the mixture. Once water was added to the mixture, it was mixed for an additional 2 min. In the case of

3.1. X-ray powder diffraction (XRD) Fig. 1 shows the XRD patterns of the AFS samples, raw fly ash and raw slag. The raw fly ash shows peaks related to the presence

Q

C-S-H, Calcite M M

M

Oxide (wt.%)

Fly ash (FA)

Blast furnace slag (BFS)

CaO SiO2 Al2O3 Fe2O3 SO3 MgO Na2O K2O LOI

4.41 67.26 14.76 4.07 – 1.29 2.04 1.39 3.57

42.47 35.17 13.93 0.58 2.03 4.12 0.15 0.46 0.18

Q Slag

Intensity

Table 1 Chemical composition of the binder materials.

LS50 LS30 LS10 Fly ash 0

10

20

30

40

50

60

70

Two- Theta Angle (deg.) Fig. 1. XRD patterns of raw fly ash, raw slag and AFS samples activated by sodium silicate liquid.

N.K. Lee, H.K. Lee / Construction and Building Materials 81 (2015) 303–312

of quartz and mullite, and the raw slag clearly shows a diffuse band at approximately 30° in 2h due to its amorphous nature. The LS sample with the liquid activator shows a strong peak at approximately 29–30° in 2h regardless of the amount of slag in the mixture. It is similar to the peaks related to the presence of C-S-H or calcite, as observed in the previous study [29] using the powdered alkali-activator. The peaks corresponding to mullite and quartz were observed due to the presence of unreacted fly ash in the LS samples. As the amount of added slag increased, the width of the peak at 29–30° in 2h became larger, meaning that the LS sample is amorphous. The XRD patterns were slightly different at 20–30° in 2h between the LS10 (i.e., sample with 10% of slag) and the LS30 or LS50 (sample with 30% or 50% of slag) samples. The LS10 sample showed a larger diffuse peak at 20–25° in 2h as compared to the LS30 and LS50 samples due to the formation of aluminosilicate gel or silica as a reaction product [42,18]. The existence of the aluminosilicate gel will be explained later in this study.

3.2. Derivative thermal gravimetry (DTG) The results of the derivative thermal gravimetric (DTG) analysis of the LS samples are shown in Fig. 2(a). A reduction of weight of the sample between 150 and 400 °C was observed due to the evaporation of the physically absorbed water from the reaction products (i.e., C-S-H gel and aluminosilicate gel) [22,4]. The weight loss of the LS10 sample was lower than that of LS30 and LS50 samples over the entire temperature range; the weight loss of LS30 sample was similar to that of the LS50 sample between 150 and 300 °C and was slightly lower between 350 and 600 °C than that

0

DTG (%/min)

-0.05 -0.1 LS10 -0.15 LS30 -0.2 LS50 -0.25 -0.3 -0.35 0

100

200

300

400

500

600

700

800

900

1000

Temperature ( C) (a) Effect of slag content in AFS sample with liquid activator

0.05 0

DTG (%/min)

-0.05 -0.1 LS30 -0.15 PS30 -0.2 -0.25 -0.3 -0.35 0

100

200

300

400

500

600

700

800

900 1000 1100

Temperature ( C)

(b) Comparison between PS50 sample with powder activator and LS50 sample with liquid activator Fig. 2. Differential thermogravimetric analysis (DTG) results.

305

of the LS50 sample. In the LS samples, a peak at approximately 450–500 °C related to the presence of Ca(OH)2 was not identified, which is consistent with the results of the XRD analysis, as shown in Fig. 2(a). Fig. 2(b) shows the results of the DTG analysis for the LS and PS samples. The maximum weight loss peaks for the two types of samples are slightly different (100 °C for PS30 and 150 °C for LS30, respectively). Very weak peaks were observed between 500 and 700 °C in the LS samples, while significant peaks were observed in the PS samples due to the formation of silica gel from the decomposition of carbonate (564–658 °C) [39,33] or the presence of unreacted silica gel in the sodium silicate used as the powder activator [39,40,36]. In previous studies [2,4], the DTG results for metakaolin/slag blends showed that the intensity of the peaks at temperatures between 590 and 690 °C increased as the amount of added slag increased due to the presence of calcium carbonate. Thus, the weight loss peak between 500 and 700 °C may be due to the formation of silica gel from the decomposition of carbonate (564–658 °C). However, it is not clear whether the silica arises from the carbonation or unreacted silica gel from the sodium silicate powder. The cause will be discussed in the discussion section with the results of the FT-IR spectra.

3.3. Fourier transform-infrared (FT-IR) spectroscopy Fig. 3(a) shows the results of the FT-IP spectra of the PS samples. The bands at 3440 cm1 and 1650 cm1 are related to the O–H stretching and bending vibrations of molecular water, respectively [24,12]. The band absorption for the PS10 sample was very weak at 1650 cm1, while that for the LS10 sample was stronger, as shown in Fig. 3(a) and (b). An absorption band at 790 cm1 appeared for all samples. In particular, as the amount of added slag decreased, the absorption band at 790 cm1 became strong, as shown in Fig. 3(a) and (b). This band is related to the symmetric stretching vibrations of the Si–O– Si (Al) bridges [24,12] due to the presence of the aluminosilicate gel and the glassy phase of the sample [12]. The level of carbonation in the AFS samples could be determined through the intensity of the absorption band at 1465 cm1. The PS30 and PS50 samples contained carbonate species characterized by a large absorption band near 1450 cm1 [24] related to the anti-symmetric stretching vibration of the CO2 ions [46,32]; the PS10 sample showed very weak band 3 absorption. In Fig. 2(a), as the amount of added slag increased in the mixture, the absorption band of the PS samples at 1450 cm1 associated with atmospheric carbonation became strong, while that of the LS samples was nearly constant. The absorption bands at approximately 1200, 1100, 950, 900, and 850 cm1 are associated with the Si–O–Si stretching vibrations of the SiQn units for n = 4, 3, 2, 1, and 0, respectively [7]. These values move into lower wavenumbers as the amount of aluminum substitution for silicon increases [24]. The wavenumbers corresponding to the maximum peaks of the PS10, PS30 and PS50 samples are 1079, 1018, and 991 cm1 in Fig. 3(c), respectively. The wavenumbers related to the maximum peaks of the LS10, LS30, and LS50 samples are 1023, 995, and 975 cm1, respectively, as shown in Fig. 3(c). In both the LS and PS samples, as the amount of slag increased, the maximum peak shifted toward a lower wavenumber. These results reveal the distribution of the Q1 and Q2 units for the PS30, LS30, PS50, and LS50 samples, while the shift toward a higher wave number for the PS10 and LS10 samples indicates the presence of a higher degree of polymerized units (Q3 and Q4 units) [24]. Accordingly, the more the amount of slag was added to the samples, the less the polymerized structure of the AFS phase was formed.

306

N.K. Lee, H.K. Lee / Construction and Building Materials 81 (2015) 303–312

0.9

795

Transmission (%)

1650

Al/Si ratio

PS10 PS30 PS50

0.8

S10

0.7

S20

0.6

S30 S40

0.5

S50

0.4 0.3

1465 465 0

0.2

960-1000 1000

2000

3450 3000

0.1 4000

5000

0 0

Wavenumbers (cm-1)

(a) AFS samples with powder activator

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Ca/Si ratio

1

1.1 1.2 1.3 1.4 1.5

(a) Al/Si ratio versus Ca/Si ratio

Transmission (%)

795

0.35

1650

Na/Si ratio

LS50

1465

0

960-1000 1000

2000

S20

0.25

LS30

465

S10

0.3

LS10

S30 S40

0.2

S50 0.15 0.1

3450

0.05 3000

4000

5000

0

Wavenumbers (cm-1)

0

(b) AFS samples with liquid activator

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Ca/Si ratio

1

1.1 1.2 1.3 1.4 1.5

(b) Na/Si ratio versus Ca/Si ratio

Transmission (%)

Fig. 4. EDS analysis of reaction products in PS samples.

PS10 AS10 PS30 AS30 PS50 AS50

800

900

1000

1100

1200

1300

Wavenumbers (cm-1) (c) 900~1200 cm-1 wavenumbers of AFS samples with powder activator and liquid activator Fig. 3. FT-IR spectra of AFS samples with different slag contents (10%, 30% and 50%).

3.4. Scanning electron microscopy (SEM) with energy-dispersive spectroscopy (EDS) Fig. 4 shows the atomic ratios of the reaction products in PS samples with the powder activator. The atomic ratio of each sample was measured for 20 times. Two apparent groups are shown in Fig. 4(a); N-A-S-H gel and C-(N)-A-S-H gel, as indicated by Ismail et al. [19]. As the amount of slag increased from 10% to 50% in the mixture, the main reaction product was N-C-A-S-H rather than N-A-S-H. In Fig. 4(b), there are three groups depending on the Ca/Si ratio and the Na/Si ratio. From the S10 sample in Fig. 4(b), it has been seen that as the Ca/Si ratio of the reaction products increased, the Na/Si ratio decreased slightly. The average of Na/Si ratios in the group (i.e., the dashed rectangle on the left-hand side of Fig. 4(b)) was higher than that in the middle of the groups. Meanwhile the

trend of the S20 sample was similar to that of the S10 sample. Thus, the reaction products measured by EDS are likely to be the N-A-S-H gel and the C-(N)-A-S-H gel in the S10 and S20 samples. The reaction products identified in the S30, S40 and S50 samples are mainly the C-N-A-S-H gel, the compositions of which are included in the dashed rectangle on the right-hand side in Fig. 4(b). Nevertheless, it was difficult to find the N-A-S-H gel in S30, S40 and S50 samples through an EDS analysis. The average of atomic ratio is given in Fig. 5. The atomic ratios of the LS sample with the liquid activator are not significantly different from that of the PS sample with the powder activator. The Al/Si ratios were 0.27 and 0.27 in the LS30 and PS30 samples, respectively; the Na/Si ratios were 0.09 and 0.17, and the Ca/Si ratios were 0.64 and 0.62. As the amount of added slag increased from 10% to 50%, the Ca/Si ratio increased from 0.21 to 0.81 in the LS and PS samples. The Al/Si ratio decreased from 0.38 to 0.25 as the amount of slag increased in the PS sample, but was nearly constant at 0.25 in the LS sample. The Al/Si atomic ratio of raw fly ash was 0.12 and that of raw slag was 0.22, as listed in Table 1. Although the Al/Si ratio of raw slag is higher than that of the raw fly ash, the Al/Si ratio of the PS sample was decreased by increasing the amount of slag. This has occurred since the degree of Al substitution for Si was decreased as the amount of added slag increased. The higher the degree of the Al substitution for the Si was, the higher the degree of polymerized units (Q3 and Q4 units) in the AFS samples occurred. This is also supported by the fact that the AFS sample with the low amount of added slag contains the reaction products with a highly polymerized structure, as shown in the NMR spectroscopy results listed in Table 4. In the PS samples, the amount of added slag did not remarkably affect the Na/Si ratio (0.08–0.17), and the Na/Si ratio was the highest when the amount of added slag was 30%. In the LS samples, the

N.K. Lee, H.K. Lee / Construction and Building Materials 81 (2015) 303–312

Element ratios

0.9 0.8

Al/Si

0.7

Na/Si

0.6

Ca/Si

0.5 0.4 0.3 0.2 0.1 0 0

10

20 30 Slag amount (%)

40

50

60

(a) PS sample with powder activator

307

with raw materials (i.e., fly ash and slag) to form aluminosilicate gel or C-S-H gel. The soluble Si or Na element content is defined as the alkali-activator content that did not react with raw materials, but dissolved in distilled water. In Table 3, the insoluble Si and Na contents did not significantly change regardless of the amount of slag when the liquid alkaliactivator was used. However, the insoluble Si and Na contents were slightly higher in the LS30 sample than in the LS10 and LS50 samples. The insoluble Si content of the alkali-activator was higher than the insoluble Na content of the alkali-activator in all samples except for the PS 10 sample. The high insolubility of Si (over 80%) can result in an increase in the Si/Al ratio of the reaction products by using sodium silicate instead of using sodium hydroxide as an alkali-activator. The high contents of insoluble Al and Ca (above 98%) indicate that the raw fly ash and slag particles were not soluble in distilled water, in contrast with sodium silicate.

0.8 Al/Si

0.7 0.6 Element ratios

3.6. Nuclear magnetic resonance (NMR) spectroscopy

Na/Si Ca/Si

0.5 0.4 0.3 0.2 0.1 0 0

10

20

30

40

50

60

Slag amount (%) (b) LS sample with liquid activator Fig. 5. Average element ratios of reaction product phases in AFS samples depending on the slag content.

Na/Si ratio was nearly constant at 0.08–0.12. The amount of added slag led to a significant change of the Ca/Si ratio, while it led to a slight change of the Al/Si ratio and the Na/Si ratio. 3.5. Insoluble elements of an alkali-activator An ICP-OES method was used to determine the reactivity of the alkali-activators (i.e., sodium silicate powder and liquid). The unreacted alkali-activator can be dissolved in distilled water, while the reacted alkali-activator cannot be dissolved in the water due to its incorporation in the reaction products. Hence, the reacted Si and Na contents of the sodium silicate can be indirectly calculated by measuring the amount of dissolved alkali activator. The reactivity of the alkali-activator was calculated and calibrated by measuring the soluble Si or Na content of the alkali-activator dissolved in distilled water, as shown in Eq. (1). It has been assumed that the reacted Si and Na in the alkali-activator are not dissolved in the distilled water. A powdered AFS sample was added to distilled water in order to determine the insoluble Si, Al, Ca, and Na contents by ICP-OES. After sonicating the powdered sample for 2 h at 25 °C, the components dissolved in the distilled water were quantified. The reactivity of the alkali-activator can be suggested as follows:

  Ew  100ð%Þ R¼ 1 Eo

ð1Þ

Here Ew is the soluble Si or Na content in the AFS sample and Eo is the total Si or Na content in the alkali-activator. Table 3 shows the insoluble Si or Na element contents of the sodium silicate powder and liquid. The insoluble Si or Na element content is defined as the alkali-activator content which reacted

Fig. 6(a) shows the 29Si NMR spectra of the raw materials (slag and fly ash). The spectrum of the raw fly ash contains peaks at 94, 99, 104, and 109 ppm, which were associated with the initial vitreous material [14,15], while most of the peaks appearing above 108 ppm were assigned to crystalline silica phases (Q4 (0Al) units), such as quartz (108 ppm) [13]. The spectrum of the raw slag contains a dominant peak at 77 ppm. Fig. 6(b) shows the spectrum of the PS samples with the powdered activator. As the amount of added slag increased, the intensity of the broad spectrum between 80 and 110 ppm increased. The PS10 and PS30 samples showed slightly higher intensity between 80 and 110 ppm compared to that of the raw fly ash, whereas the PS50 sample showed a strong broad peak between the maximum peaks of the raw slag and the raw fly ash. Fig. 6(c) presents the spectrum of the LS samples with the liquid activator. The spectra of the LS and PS samples were very different. The LS10 sample presented a very broad peak between 85 ppm and 106 ppm, which was similar to that of the PS50 sample. However, the LS30 and LS50 samples showed two predominant peaks at 82 ppm to 83 ppm and at 109 ppm. The peak intensity for the LS50 sample at 82 to 83 ppm was higher than that of the LS30 sample, while that of the LS50 sample at 109 ppm was smaller than that of the LS30 sample. The apparent difference among the maximum peaks indicates that the LS50 (or LS30) and LS10 have different silicate structures. Accordingly, it is clear that the silicate structure of the reaction product of the AFS binder is significantly affected by the amount of added slag. The NMR spectra of the AFS samples were deconvoluted with Origin software (OriginLab Corporation) to quantify the reaction products. The relative areas below the corresponding fitted peaks were calculated by using deconvolution peaks from the raw materials (fly ash and slag) and the AFS samples. The NMR peaks appearing at 88, 93, 99, 104 and 108 ± 1 ppm are attributed to silicate tetrahedra (Q4) surrounded by 0, 1, 2, 3 or 4 aluminum tetrahedra, respectively [13,21,30,34]. The deconvolution peaks were produced on the basis of these peaks associated with Q4 (nAl), where n = 1, 2, 3, and 4. Table 4 illustrates the relative areas obtained from deconvolution and the peak analysis results via Si NMR spectroscopy. The fly ash-reactivity and slag-reactivity of the AFS sample can be calculated as follows. First of all, the deconvolution peaks for the AFS sample corresponding to the peaks at 89, 93, 98, 104, and 109 ppm of raw fly ash are selected. Secondly, among the selected peaks, the peaks of which the intensity levels are higher than that of the corresponding deconvolution peaks for the raw fly ash are selected. The relative area is defined as an area below any deconvolution peak divided by the total area below all of the

308

N.K. Lee, H.K. Lee / Construction and Building Materials 81 (2015) 303–312

Table 3 Insoluble Si or Na element content of powder and liquid sodium silicate in distilled water (%). Insoluble content

LS10

LS30

LS50

PS10

PS30

PS50

Na Si Ala Caa

46.73 84.99 99.42 98.97

53.66 89.40 99.45 99.44

44.35 87.79 98.92 99.46

36.79 29.20 99.89 98.41

49.94 82.45 99.06 98.56

51.01 84.22 98.79 98.65

Slag Fly ash

a Insoluble Al and Ca contents (%) means the ratio of the insoluble Al and Ca contents into distilled water to the total Al and Ca contents in AFS sample, respectively.

-40

-50

-60

-70

-80

-90

-100

-110

-120

-130

-140

-150

-120

-130

-140

-150

-140

-150

Chemical shift (ppm)

deconvolution peaks. Afterward, the total sum of the relative areas (A) below the selected deconvolution peaks for the AFS sample was calculated, and the total sum of the relative areas (B) below the selected deconvolution peaks for the raw fly ash was also calculated. The total sum of the areas (C) below the newly-appearing peaks for the AFS samples, which do not match the deconvolution peaks for raw fly ash, was also calculated. Accordingly, the total relative area below the peaks associated with the presence of reaction product equals to (AB)+C. Finally, the amounts of reaction products and the reactivities of the raw materials could be calculated indirectly using the method proposed by Buchwald et al. [6]. The procedure for quantifying the reaction products and the reactivity is demonstrated in Fig. 7 and the deconvolution peaks are shown in Figs. 8, 9 and 10. Table 4 presents the relative area obtained from deconvolution results and the peak analysis by Si NMR spectroscopy. The fly ashreactivity, slag-reactivity and total reactivity were calculated by using the relative area, obtained as follows (cf. [6]):

Rslag ¼

Sislag  U slag ð%Þ Sislag

Rfly ash ¼

a b

0 2.44 3.45 0.40 1.15 14.58

-50

-60

-70

-80

-90

-100

-110

Chemical shift (ppm)

(b) AFS samples activated by powder activator

LS50

ð4Þ -40

-50

-60

-70

-80

-90

-100

-110

-120

-130

Chemical shift (ppm)

(c) AFS samples activated by liquid activator Fig. 6.

29

Si NMR spectra of the raw materials and ASF samples.

The reactivities are listed in Table 5. The total reactivities of the LS samples were 26.41% (LS10), 51.61% (LS30), and 61.77% (LS50); the values for the PS samples were 6.25% (PS10), 21.4% (PS30), and 44.98% (PS50). The total reactivity was even higher when mixed the liquid activator compared to the powder activator.

29

C-S-H

100

-40

LS30

Table 4 Relative area obtained from deconvolution results and peak analysis of

Slag Fly ash PS10 PS30 PS50 LS10 LS30 LS50

PS10

LS10

In these equations, Rslag and Rfly ash denote the fly ash-reactivity and slag-reactivity in the AFS sample, and RTotal indicates the total reactivity of the AFS sample. Sislag and Sifly ash are the silicon content (%) in the raw slag and fly ash, respectively. Fc and Sc are the raw fly ash and slag contents (%) in the AFS mixture, respectively. The peaks of the AFS sample which was identical to the deconvolution peaks of the raw fly ash and slag were selected. Uslag and Ufly ash are the sum of the relative areas below the selected deconvolution peaks of which the intensity is lower than the peak intensity of the raw fly ash and the raw slag. That is, these peaks appear due to the presence of the unreacted fly ash and slag particles remaining in the AFS sample.

77, etc.a

PS30

ð3Þ

RTotal ¼ F c  Rfly ash þ Sc  Rslag ð%Þ

Signal (ppm)

PS50

ð2Þ

Sifly ash  U fly ash ð%Þ Sifly ash

Sample

(a) Raw materials

Si NMR spectroscopy (%). Aluminosilicate (C-N-A-S-H)

Q1 79

Q2 (1Al) 81

Q2 86

Q4 (4Al) 89

Q4 (3Al) 94

Q4 (2Al) 98

Q4 (1Al) 104

Q4 (0Al) 109, etc.b

0 0 0.55 0.35 4.27 10.65

0 2.41 1.03 0.53 12.89 22.98

0 1.04 5.3 2.55 18.55 17.64

0.66 0.21 4.62 6.29 4.81 2.66 0.96

3.60 4.54 9.53 6.10 3.77 7.21 0.64

10.75 9.7 13.42 10.26 3.5 3.41 1.48

24.35 23.1 27.16 7.79 9.86 6.53 8.10

60.64 62.5 39.38 34.07 55.6 34.24 12.76

The peaks at 77, 75, 72, 70, 68, 65 ppm are attributed to the presence of raw slag in AFS sample. These are the peaks above 109 ppm. Most of the peaks appearing above 108 ppm were assigned to different crystalline phases of silica (Q4(0Al) signals) [13].

309

N.K. Lee, H.K. Lee / Construction and Building Materials 81 (2015) 303–312

NMR spectrum

PS10

By using Lorentz curve of Origin software and the literature study

Deconvolution of Raw materials and AFS sample

Fit

Selection of the same deconvolution peaks of AFS sample as that of raw fly ash

Selection of the peak of which the intensity is higher than that of raw fly ash

-60

-70

-80

-90

-100

-110

-120

-130

-140

-150

Chemical shift (ppm)

(a) PS10 AFS sample

Total relative area of the selected peaks for AFS sample

PS30 Fit

Total relative area of the selected peaks for raw fly ash

Total relative area of newly appearing peaks for AFS sample

-60

Total relative area of the peaks appearing due to reaction products

-70

-80

-90

-100

-110

-120

-130

-140

-150

Chemical shift (ppm)

(b) PS30 AFS sample

Using Eqs. 2, 3 and 4 Reactivity of raw material and total reactivity

PS50

Fig. 7. Procedure for quantifying the reaction products and the reactivity.

Fit

Fly ash Fit

-60

-70

-80

-90 -100 -110 -120 Chemical shift (ppm)

-130

-140

-150

(c) PS50 AFS sample Fig. 9. 29Si NMR spectra and deconvolution of ASF samples activated by powder activator.

-60

-70

-80

-90

-100

-110

-120

-130

-140

-150

Chemical shift (ppm)

4. Discussions

(a) Raw fly ash

4.1. Reactivity Slag Fit

-60

-70

-80

-90

-100

-110

-120

-130

-140

-150

Chemical shift (ppm)

(b) Raw slag Fig. 8.

29

Si NMR spectra and deconvolution of raw materials.

The fly ash-reactivity, slag-reactivity and total reactivity of the AFS paste (liquid type and powder type) were calculated by means of an NMR analysis and an ICP-OES analysis, respectively. In Table 5, the weight loss measured by DTG, the absorption extent of the band measured by FT-IR, and the reactivity obtained from the NMR spectra were compared. The weight loss up to 500 °C was related to the dehydration of the reaction product (i.e., aluminosilicate gel and C-S-H gel). Likewise, the absorption band at 3450 cm1 was related to the O-H stretching band of molecular water, revealing the degree of hydration. On the other hand, the reactivity was measured by using the NMR technique. In Table 5, the reactivity, weight loss and extent of absorption increased as the amount of slag increased. The total reactivity by NMR was twice as high in the LS30 sample as it was in the PS30

310

N.K. Lee, H.K. Lee / Construction and Building Materials 81 (2015) 303–312

Table 5 Fly ash-reactivity, slag-reactivity and total reactivity of AFS samples activated by sodium silicate powder and liquid (%). Reactivity

LS10

LS30

LS50

PS10

PS30

PS50

Fly ash Slag Total reactivity Weight loss by TGAa Absorption extent by FT-IRb

23.15 91.52 26.41 4.59 51.94

38.82 94.23 51.61 6.67 64.46

60.93 63.65 61.77 7.54 68.31

4.0* 100* 6.25 –** 23.50

1.57 87.81 21.4 6.76 59.56

22.97 88.88 44.98 –** 63.50

* In the case of PS10 sample, the fly ash-reactivity was minus 4.0, and the slagreactivity was 100%. Since the Si NMR spectrum assigned to raw slag was not observed in PS10 sample, all the raw slag particles seemed to react completely with alkali-activator, thus the slag-reactivity was determined to be 100%. The reasons why the fly ash-reactivity was below zero are (1) the actual fly ash-reactivity was very low (less than 5%), and (2) the noise of 29Si NMR spectrum occurred. Even though there are some difficulties in measuring the 29Si NMR spectra, it is certain that the fly ash-reactivity is very low when the amount of added slag is 10%, and the slag-reactivity is nearly 100%. ** TGA tests of PS10 and PS50 samples were not conducted. a Weight loss measured from TGA analysis. b Absorbed band percentage (%) related to O–H stretching bend at around 3400– 3500 cm1.

sample; however, the weight loss, 6.67% of the LS30 sample by DTG and the extent of absorption, 64.46% by FT-IR were fairly similar to those of the PS30 sample (i.e., 6.76% and 59.56%, respectively). These differences mean that the type of alkali-activator (liquid or powder) did affect the reactivity of each raw material (i.e., fly ash and slag) measured by the NMR analysis. The LS30 and PS30 samples were similar in the degree of the hydration measured by the DTG regardless of the type of the alkali-activator. Equally, both of the samples were comparable in the reactivity of the alkali-activator regardless of the amount of slag as listed in Table 3. It seems that the total reactivity of AFS binder is strongly dependent upon the mixture ratio of raw materials as well as the type of alkaliactivator, while the degree of hydration is associated with the reactivity of alkali-activator. Table 5 shows the fly ash-reactivity and slag-reactivity, and total reactivity of AFS sample. The NMR peak of the PS10 sample overlapped with the noise peaks during the measurements by decreasing the reliability of the results so that the peak intensity of the PS10 sample was very low, as shown in Fig. 9(a). The slag-reactivity in all of the samples was higher than the fly ash-reactivity since the raw slag had a latent hydraulic property unlike fly ash, and was primarily composed of amorphous phases. In addition, the fly ash-reactivity was much higher when mixed with the liquid activator compared to the powder activator, while the slag-reactivity was roughly equal regardless of alkali activator. Therefore, using the liquid activator resulted in the relatively higher fly ashreactivity, which can contribute to the higher compressive strength of the AFS binder compared to the results with the powder activator. 4.2. Reaction products The results from XRD, FT-IR, EDS, and NMR analysis for the AFS paste provide important new insight into the structural characteristics of the reaction products. A highly polymerized silicate structure with Q4 (i.e., 103 to 115 ppm) and Q3 (i.e., 95 to 100 ppm) units was identified when the amount of slag was 10% in the LS sample with the liquid activator, whereas reaction products with Q1 (i.e., 74 to 78 ppm) and Q2 (i.e., 83 to 88 ppm) units, but not with Q4 and Q3 units, were observed when the amounts of slag was 30% and 50% in the LS sample. Fig. 6(c) clearly shows the differences among the maximum NMR peaks of the LS samples. As mentioned in the previous research

LS10 Fit

-60

-70

-80

-90

-100

-110

-120

-130

-140

-150

Chemical shift (ppm)

(a) LS10 AFS sample

LS30 Fit

-60

-70

-80

-90

-100

-110

-120

-130

-140

-150

Chemical shift (ppm)

(b) LS30 AFS sample

LS50 Fit

-60

-70

-80

-90

-100

-110

-120

-130

-140

-150

Chemical shift (ppm)

(c) LS50 AFS sample 29

Fig. 10. Si NMR spectra and deconvolution of ASF samples activated by liquid activator.

[24], the Q1 and Q2 units are similar to the silicate structure of C-S-H in alkali-activated slag or OPC, and the silicate structure of aluminosilicate gel in a geopolymer is similar to the Q3 and Q4 structures. Based on this, it was found that both the amount of reaction product and its silicate structure were remarkably varied depending on the amount of slag added, as listed in Table 4. The microstructural characteristics can generally affect the durability of the AFS paste. It was reported that aluminosilicate gel was resistant to acids, while C-S-H gel was vulnerable to acids [1]. Hence, future research on the acid resistance of AFS paste based on the amounts of C-S-H gel and aluminosilicate gel needs to be conducted. The amounts of the aluminosilicate gel and C-S-H gel in the AFS sample were evaluated quantitatively by using the NMR results. The chemical composition ratios of these reaction products were measured by using the EDS results. Fig. 4 shows the distribution of the atomic ratios for the reaction products in AFS samples measured by using EDS. It was difficult to distinguish between

311

N.K. Lee, H.K. Lee / Construction and Building Materials 81 (2015) 303–312

the C-S-H gel and aluminosilicate gel in spite of applying the EDS point analysis results. As shown in Fig. 5, although the Ca/Si ratios were varied depending on the amount of added slag, a substantial amount of Ca (i.e., Ca/Si ratio of 0.2–0.8) was incorporated into the reaction products. The silicate structures of the reaction products are shown in Table 4. The C-S-H consists of Q1 (chain end groups) and Q2 (middle groups in chains) units, while the aluminosilicate gel consists of Q4 (4Al), Q4 (3Al), Q4 (2Al), and Q4 (1Al) units; these units are known to be three-dimensional cross-linked sites with the different amounts of Al substitution for Si. Fig. 5 and Table 4 show that the aluminosilicate gel with Q4 (nAl) units with n between 0 and 4 contains the certain amount of Ca. From the fact that the geopolymer is a reaction product formed by polycondensation according to Davidovits [9], it can be concluded that the aluminosilicate gel in the present study is likely to be a Ca-based geopolymer, and the C-S-H gel is similar to C-S-H gel in OPC or alkali-activated slag. Therefore, the Ca-based geopolymer and C-S-H were only slightly different in terms of their chemical compositions but were quite different in terms of their silicate structures; both formed simultaneously in the AFS sample. The different types of silicate tetrahedra in the Ca-based geopolymer and C-S-H are quantitatively assessed in Table 4. The results from the DTG and FT-IR analyses indicate that the reaction product in the AFS sample may have been carbonated. In the FT-IR data, the absorption bands at 1465 cm1 related to carbonation were observed in all of the samples; however, this signal was very weak in the PS10 sample. There is no obvious relationship between the degree of carbonation and the amount of added slag. The LS sample with the liquid activator had the higher degree of carbonation than the PS sample with the powder activator due to the larger amount of C-S-H in the LS sample, as listed in Table 4. In Fig. 2(b), the PS30 sample with the powder activator showed widely distributed peaks at a range of 500–700 °C due to the removal of the water absorbed by the unreacted silica gel in the powder activator or silica gel resulting from the decomposition of carbonate. The FT-IR results show that both LS and PS samples were affected by carbonation. Therefore, the high peak of the PS 30 sample at 500–700 °C is more likely to be due to the dehydration of absorbed water from the unreacted silica gel in the powder activator rather than the dehydration of silica gel resulting from the decomposition of carbonate. As the amount of added slag increased, the maximum absorption band shifted toward lower wavenumbers, as shown in Fig. 3(c). The wavenumbers for the LS10, LS30 and LS50 samples at their maximum intensity levels were 1023, 995, and 975 cm1, while those of the PS10, PS30, and PS50 samples were 1079, 1018 and 991 cm1, respectively. Since the amount of C-S-H with Q1 and Q2 units in the AFS sample increased with the amount of added slag, the absorption band shifted toward lower wavenumbers, which was in agreement with the results of the previous research [24]. In addition, when the same amount of slag was added in the LS and PS samples, the maximum absorption band in the LS sample was lower than that in the PS sample since the amount of C-S-H in the LS sample was higher than that in the PS sample, as listed in Table 4. The Si/Al ratios from the EDS analysis were compared with the Si/Al ratios obtained from Engelhard’s equation [13] and the NMR results in Table 6. The Si/Al ratios measured by the EDS analysis were not similar to any of those of the C-S-H gel and Ca-based aluminosilicate gel. Since it was difficult to distinguish between C-S-H and Ca-based aluminosilicate by using only a BSE image on SEM, the Si/Al ratio determined by the EDS analysis has been the average of either the C-S-H or Ca-based aluminosilicate Si/Al ratios randomly measured by the EDS analysis; although it might be possible to distinguish between the C-S-H and Ca-based aluminosilicate by using elemental maps for calcium via SEM, the discrete calcium-

Table 6 Si/Al ratios and chain length of the reaction products measured from spectroscopy and EDS analyses. Sample

PS10 PS30 PS50 LS10 LS30 LS50

Al-substituted C-S-H

Ca-aluminosilicate

Chain length

Si/Al

Si/Al

– – 26.72 21.34 19.74 11.79

– 2.87 13.36 12.98 5.54 4.46

4.93 2.89 2.08 2.81 2.57 4.26

29

Si NMR

Si/Al by EDS

2.57 3.62 3.83 4.11 3.57 4.02

rich and calcium-deficient phases were not identified in the SEM length scale [26]. Accordingly, each of the Si/Al ratios in Fig. 4 can be either for the C-S-H or Ca-based aluminosilicate. The chain length (11.79) of the C-S-H phase for the LS50 sample in Table 6 was similar to the chain length (13.4) given in Buchwald et al. [6] for the same mixture ratio. However, the Si/Al ratios (i.e., 4.46 and 4.26) of the Ca-based aluminosilicate and the C-S-H phase were quite different from the values (i.e., 1.19 and 2.8) of the aluminosilicate and the C-S-H measured by Buchwald et al. [6] as the different types of alkali-activators were used; in the present study, the liquid sodium silicate containing a substantial amount of silicon was used, whereas, in the their study, the liquid sodium hydroxide was used as an alkali-activator. The silicate ion in the liquid sodium silicate condensed to form Si-O-Si (Al) bonds [8], resulting in the higher Si/Al ratio in the reaction products. Therefore, the Si/Al ratios of the C-S-H gel and Ca-based aluminosilicate gel were much higher than the results of Buchwald et al. [6]. 5. Concluding remarks The results of an experimental study conducted to evaluate the reactivity and reaction products of alkali-activated, fly ash/slag paste have been summarized. The following conclusions can be drawn from the results presented in this paper. 1. The total reactivity of the AFS paste calculated by an NMR peak analysis was much higher when mixed with the liquid activator compared to the powder activator. The reactivity of the liquid activator was slightly higher than that of the powder activator. 2. The fly ash-reactivity was significantly higher when mixed with the liquid activator compared to the powder activator; the slagreactivity was comparable regardless of the activator types. The use of the liquid activator resulted in an increase in the fly ashreactivity and the total reactivity. 3. The amount of added slag primarily affected the amount of reaction product and its silicate structure. As the amount of slag increased, the amount of C-S-H with Q1 and Q2 units increased, whereas the amount of aluminosilicate gel with Q4 units decreased. A highly polymerized silicate structure with a Q4 unit was identified in all of the LS samples with the liquid activator, while the reaction products with Q1 and Q2 units were identified when the amounts of added slag were 30% and 50% in the LS samples. 4. The EDS analysis and the 29Si NMR spectra showed that the aluminosilicate gel with Q4 (nAl) units contained a certain amount of Ca. This indicates that considering chemical composition and silicate structure, the aluminosilicate gel was similar to a Cabased geopolymer. The chemical compositions of the Ca-based geopolymer were nearly the same as those of C-S-H whereas their silicate structures were different. The results of this research will help assess the effects of the amounts of the reaction products and reactivity levels on the durability of alkali-activated, fly ash/slag binders. Future works on the

312

N.K. Lee, H.K. Lee / Construction and Building Materials 81 (2015) 303–312

acid and chloride resistance of AFS paste will be followed in the future on the basis of the amounts of C-S-H gel and aluminosilicate gel. Acknowledgements This research was supported by a Grant (Code 11-Technology Innovation-F04) from Construction Technology Innovation Program (CTIP) funded by Ministry of Land, Transportation and Maritime Affairs (MLTM) of Korean government. References [1] Allahverdi A, Skvara F. Nitric acid attack on hardened paste of geopolymeric cements. Part 1. Ceramics 2001;45(3):81–8. [2] Bernal SA, de Gutierrez RM, Provis JL, Rose V. Effect of silicate modulus and metakaolin incorporation on the carbonation of alkali silicate-activated slags. Cem Concr Res 2010;40(6):898–907. [3] Bernal SA, Provis JL, Rose V, Mejía de Gutierrez R. Evolution of binder structure in sodium silicate-activated slag-metakaolin blends. Cement Concr Compos 2011;33(1):46–54. [4] Bernal SA, Rodríguez ED, de Gutiérrez RM, Gordillo M, Provis JL. Mechanical and thermal characterisation of geopolymers based on silicate-activated metakaolin/slag blends. J Mater Sci 2011;46(16):5477–86. [5] Bernal SA, Provis JL, Walkley B, San Nicolas R, Gehman JD, Brice DG, et al. Gel nanostructure in alkali-activated binders based on slag and fly ash, and effects of accelerated carbonation. Cem Concr Res 2013;53:127–44. [6] Buchwald A, Hilbig H, Kaps C. Alkali-activated metakaolin-slag blends— performance and structure in dependence of their composition. J Mater Sci 2007;42(9):3024–32. [7] Clayden NJ, Esposito S, Aronne A, Pernice P. Solid state 27 Al NMR and FTIR study of lanthanum aluminosilicate glasses. J Non-Cryst Solids 1999;258(1):11–9. [8] Criado M, Fernández-Jiménez A, Palomo A, Sobrados I, Sanz J. Effect of the SiO2/Na2O ratio on the alkali activation of fly ash. Part II: 29Si MAS-NMR Survey. Microporous Mesoporous Mater 2008;109(1):525–34. [9] Davidovits J. Chemistry of geopolymeric systems, terminology. In: Geopolymer, 99; (1999). p. 9–40. [10] Davidovits J. Geopolymers: inorganic polymeric new materials. J Mater Educ 1994;16:91–138. [11] Duxson P, Provis JL, Lukey GC, Van Deventer JS. The role of inorganic polymer technology in the development of ‘green concrete’. Cem Concr Res 2007;37(12):1590–7. [12] El-Didamony H, Amer AA, El-Sokkary TM, Abd-El-Aziz H. Effect of substitution of granulated slag by air-cooled slag on the properties of alkali activated slag. Ceram Int 2013;39(1):171–81. [13] Engelhardt G, Michel D. High-Resolution Solid-State NMR of Silicates and Zeolites. New York: Wiley; 1987. [14] Fernández-Jiménez A, Palomo A. Characterisation of fly ashes. Potential reactivity as alkaline cements. Fuel 2003;82(18):2259–65. [15] Fernández-Jimenez A, De La Torre AG, Palomo A, López-Olmo G, Alonso MM, Aranda MAG. Quantitative determination of phases in the alkali activation of fly ash. Part I. Potential ash reactivity. Fuel 2006;85(5):625–34. [16] Gruskovnjak A, Lothenbach B, Holzer L, Figi R, Winnefeld F. Hydration of alkaliactivated slag: comparison with ordinary Portland cement. Adv. Cem. Res. 2006;18(3):119–28. [17] Haha MB, Le Saout G, Winnefeld F, Lothenbach B. Influence of activator type on hydration kinetics, hydrate assemblage and microstructural development of alkali activated blast-furnace slags. Cem Concr Res 2011;41(3):301–10. [18] Hou X, Kirkpatrick RJ, Struble LJ, Monteiro PJ. Structural investigations of alkali silicate gels. J Am Ceram Soc 2005;88(4):943–9. [19] Ismail I, Bernal SA, Provis JL, San Nicolas R, Hamdan S, van Deventer JS. Modification of phase evolution in alkali-activated blast furnace slag by the incorporation of fly ash. Cement Concr Compos 2014;45:125–35. [20] Jang JG, Lee NK, Lee HK. Fresh and hardened properties of alkali-activated fly ash/slag pastes with superplasticizers. Constr Build Mater 2014;50:169–76.

[21] Klinowski J. Nuclear magnetic resonance studies of zeolites. Prog Nucl Magn Reson Spectrosc 1984;16:237–309. [22] Kong DL, Sanjayan JG. Effect of elevated temperatures on geopolymer paste, mortar and concrete. Cem Concr Res 2010;40(2):334–9. [23] Kumar S, Kumar R, Mehrotra SP. Influence of granulated blast furnace slag on the reaction, structure and properties of fly ash based geopolymer. J Mater Sci 2010;45(3):607–15. [24] Lecomte I, Henrist C, Liegeois M, Maseri F, Rulmont A, Cloots R. (Micro)structural comparison between geopolymers, alkali-activated slag cement and Portland cement. J Eur Ceram Soc 2006;26(16):3789–97. [25] Lloyd RR, Provis JL, van Deventer JS. Microscopy and microanalysis of inorganic polymer cements. 2: the gel binder. J Mater Sci 2009;44(2):620–31. [26] Lloyd RR, Provis JL, van Deventer JS. Microscopy and microanalysis of inorganic polymer cements. 1: remnant fly ash particles. J Mater Sci 2009;44(2):608–19. [27] Lee NK, Lee HK. Setting and mechanical properties of alkali-activated fly ash/ slag concrete manufactured at room temperature. Constr Build Mater 2013;47:1201–9. [28] Lee NK, Kim HK, Park IS, Lee HK. Alkali-activated, cementless, controlled lowstrength materials (CLSM) utilizing industrial by-products. Constr Build Mater 2013;49:738–46. [29] Lee NK, Jang JG, Lee HK. Shrinkage characteristics of alkali-activated fly ash/ slag paste and mortar at early ages. Cem Concr Compos 2014;53:239–48. [30] Lippmaa E, Mägi M, Samoson A, Tarmak M, Engelhardt G. Investigation of the structure of zeolites by solid-state high-resolution silicon-29NMR spectroscopy. J Am Chem Soc 1981;103(17):4992–6. [31] Luo X, Xu J, Bai E, Li W. Systematic study on the basic characteristics of alkaliactivated slag-fly ash cementitious material system. Constr Build Mater 2012;29:482–6. [32] Mollah MYA, Lu F, Cocke DL. An X-ray diffraction (XRD) and Fourier transform infrared spectroscopic (FT-IR) characterization of the speciation of arsenic (V) in Portland cement type-V. Sci Total Environ 1998;224(1):57–68. [33] Mostafa NY, El-Hemaly SAS, Al-Wakeel EI, El-Korashy SA, Brown PW. Characterization and evaluation of the hydraulic activity of water-cooled slag and air-cooled slag. Cem Concr Res 2001;31(6):899–904. [34] Neuhoff PS, Stebbins JF, Bird DK. Si–Al disorder and solid solutions in analcime, chabazite, and wairakite. Am Mineral 2003;88(2–3):410–23. [35] Palomo A, Glasser FP. Chemically-bonded cementitious materials based on metakaolin. Br Ceram Trans J 1992;91(4):107–12. [36] Peng GF, Huang ZS. Change in microstructure of hardened cement paste subjected to elevated temperatures. Constr Build Mater 2008;22(4):593–9. [37] Puertas F, Fernández-Jiménez A. Mineralogical and microstructural characterisation of alkali-activated fly ash/slag pastes. Cement Concr Compos 2003;25(3):287–92. [38] Puligilla S, Mondal P. Role of slag in microstructural development and hardening of fly ash-slag geopolymer. Cem Concr Res 2013;43:70–80. [39] Thiery M, Villain G, Dangla P, Platret G. Investigation of the carbonation front shape on cementitious materials: effects of the chemical kinetics. Cem Concr Res 2007;37(7):1047–58. [40] Villain G, Thiery M, Platret G. Measurement methods of carbonation profiles in concrete: thermogravimetry, chemical analysis and gammadensimetry. Cem Concr Res 2007;37(8):1182–92. [41] van Jaarsveld JGS, Van Deventer JSJ, Schwartzman A. The potential use of geopolymeric materials to immobilise toxic metals: Part II. Material and leaching characteristics. Miner Eng 1999;12(1):75–91. [42] van Jaarsveld JGS, Van Deventer JSJ, Lukey GC. The characterisation of source materials in fly ash-based geopolymers. Mater Lett 2003;57(7):1272–80. [43] Xu H, van Deventer JSJ. The geopolymerisation of aluminosilicate minerals. Int J Miner Process 2000;59(3):247–66. [44] Yip CK, Lukey GC, van Deventer JSJ. The coexistence of geopolymeric gel and calcium silicate hydrate at the early stage of alkaline activation. Cem Concr Res 2005;35(9):1688–97. [45] Yip CK, Van Deventer JSJ. Microanalysis of calcium silicate hydrate gel formed within a geopolymeric binder. J Mater Sci 2003;38(18):3851–60. [46] Yu P, Kirkpatrick RJ, Poe B, McMillan PF, Cong X. Structure of calcium silicate hydrate (C-S-H): near-, mid-, and far-infrared spectroscopy. J Am Ceram Soc 1999;82(3):742–8. [47] Zhao FQ, Ni W, Wang HJ, Liu HJ. Activated fly ash/slag blended cement. Resour Conserv Recycl 2007;52(2):303–13.