Influence of calcium and sodium nitrate on the strength and reaction products of the CaO-activated GGBFS system

Construction and Building Materials 215 (2019) 839–848

Contents lists available at ScienceDirect

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

Influence of calcium and sodium nitrate on the strength and reaction products of the CaO-activated GGBFS system Woo Sung Yum a, Jung-Il Suh a,b, Sungwon Sim a, Seyoon Yoon c, Yubin Jun a, Jae Eun Oh a,⇑ a

School of Urban and Environmental Engineering, Ulsan National Institute of Science and Technology (UNIST), UNIST-gil 50, Ulju-gun, Ulsan 44919, Republic of Korea Department of Architecture and Architectural Engineering, Seoul National University, 1, Gwanak-ro, Gwanak-Gu, Seoul 08826, Republic of Korea c Department of Civil Engineering, Kyonggi University, Suwon 16227, Republic of Korea b

h i g h l i g h t s  The influence of Ca(NO3)2 and NaNO3 on the CaO-activated GGBFS was investigated.  Both salts notably increased early strengths by forming more C-S-H.  NaNO3 increased pH while Ca(NO3)2 supplied additional Ca ions at early days.  Ca(NO3)2 continued increasing strength until 28 days while NaNO3 did not.  At 28 days of curing, the use of NaNO3 reduced C-S-H unlike that of Ca(NO3)2.

a r t i c l e

i n f o

Article history: Received 18 September 2018 Received in revised form 8 April 2019 Accepted 26 April 2019 Available online 3 May 2019 Keywords: CaO-activation GGBFS Calcium nitrate Sodium nitrate Cementless binder

a b s t r a c t The influence of the added nitrate salts (Ca(NO3)2 and NaNO3) on the CaO-activated ground granulated blast-furnace slag (GGBFS) system was investigated. Both salts significantly increased the early strength of the binder, and the Ca(NO3)2 maintained this tendency until 28 days of curing, whereas NaNO3 did not. At early days, the NaNO3 likely increased pH of samples, resulting in more dissolving GGBFS and formation of C-S-H, while the Ca(NO3)2 slightly decreased pH due to the common ion effect, but because the Ca (NO3)2 supplied additional Ca ions, it also increased the strength. At 28 days, the NaNO3 clearly induced the reduction in C-S-H, which was not the case for the Ca(NO3)2 samples. In addition, the pore size distribution was also significantly dependent on the cation type of nitrate salts. Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction A vast amount of carbon dioxide (CO2) is produced during the Portland cement (PC) manufacturing process, and it has been pointed out as a serious cause for accelerating global warming. For this reason, cementless binders have been developed to replace PC [1–6]. Ground granulated blast-furnace slag (GGBFS) is an industrial waste by-product from producing steel and often used as the main precursor material to produce cementless binders. The developed strength, durability, and production cost of the GGBFS-based cementless binders considerably depend on the type of chemical activators [1,2,7,8]. Alkaline activators (e.g., NaOH, KOH, sodium silicate, etc.) are chemical activators that have been extensively used to produce GGBFS-based cementless binders (alkali-activated GGBFS). However, although these activators yield ⇑ Corresponding author. E-mail address: [email protected] (J.E. Oh). https://doi.org/10.1016/j.conbuildmat.2019.04.240 0950-0618/Ó 2019 Elsevier Ltd. All rights reserved.

relatively high strength binders, their material costs are high, resulting in a low economic competence of cementless binders. In addition, these activators produce a very high pH (over 14) when dissolved, making it difficult to use them in construction fields due to safety concerns [9,10]. To avoid these major drawbacks of alkaline activators, Kim et al. [5] introduced a new type of cementless GGBBS binder using CaO as a main activator. CaO is more advantageous as it yields a relatively low pH (13) when dissolved and is less expensive than alkaline activators. Nevertheless, there have been concerns that CaO-activated GGBFS binders tend to produce significantly lower early strength compared to alkali-activated GGBFS binders. For this reason, it is essential to find add-on chemicals to increase the early strength of CaO-activated GGBFS binders. As such, Yum et al. [6] explored the use of CaCl2 as an addition for CaO to increase the early strength of this binder system. This study showed that the use of CaCl2 dissolved more of the glass phase of GGBFS and significantly enhanced the compressive strength at 3 and 28 days of curing.

840

W.S. Yum et al. / Construction and Building Materials 215 (2019) 839–848

Relative Intensity

However, because chloride ions released from dissolving CaCl2 may cause severe corrosion problems in reinforced concrete structures, particular attention should be placed on its use [11,12]. Therefore, it is desirable to find novel chemical additives without chloride to increase the early strength of CaO-activated GGBFS binders. Calcium nitrate (Ca(NO3)2) and sodium nitrate (NaNO3) are well-known accelerators along with CaCl2 for the hydration of PC [13–16]. Although the PC system is significantly different than the CaO-activated GGBFS system, the Ca(NO3)2 and NaNO3 nitrate salts may have potentially a beneficial effect on this cementless binder system because the main reaction product of CaOactivated GGBFS is also calcium silicate hydrate (C-S-H), which is similar to PC. Therefore, in this study, due to the potential beneficial effects, the nitrate salts Ca(NO3)2 and NaNO3 were investigated as chemical additives to increase the early strength of CaO-activated GGBFS binders, using the compressive strength test, measurement of the pH, powder X-ray diffraction (XRD), thermogravimetry (TG), and mercury intrusion porosimetry (MIP).

GGBFS

10

of 1.6 or higher. The GGBFS used in this study had a basicity of 1.76. The loss of ignition (LOI) of GGBFS was examined up to 1000 °C using a thermal analyzer (SDT Q600, TA Instruments, New Castle, DE, USA) with alumina pans and reported a 0.8 wt% (wt%) of LOI. After characterizing the commercial GGBFS, the mixture proportions of the paste samples with Ca(NO3)2 or NaNO3 were prepared as shown in Table 2. The amount of main activator (CaO) was fixed at 4 wt%, which was determined in preliminary experiments prior to this study, and the Ca(NO3)2 or NaNO3 were used as a 0.5, 1, 3, and 5 wt% substitution of the GGBFS. All samples were prepared with a water-to-binder weight ratio (w/b) of 0.35 (Table 2). Thus, as the solubilities of Ca(NO3)2 and NaNO3 are 121.2 g/100 g H2O and 91.2 g/100 g H2O in water at 25 °C, respectively [17], all nitrate salts were likely dissolved in the mixing water. It should be also noted that when Ca(NO3)2 and NaNO3 are dissolved in water by the same amount, they yield very similar moles of positive charges of the cations. For instance, when 100 g of these salts are dissolved in water, Ca(NO3)2 generates 0.609 mol of Ca2+ (i.e., 1.218 mol of positive charge), and NaNO3 produces 1.173 mol of Na+ (i.e., 1.173 mol of positive charge); in this case, the charge difference of the cations is merely 4.0%. Similarly, the moles of NO3 generated are also very similar. The raw powder materials GGBFS, CaO, Ca(NO3)2, and NaNO3 were dry mixed for 5 min, followed by the addition of deionized water and further mixing. All mixing procedures followed the guidelines of the American Society for Testing and Materials (ASTM) C305 [18]. The fresh pastes were cast in cubic molds (5  5  5 cm) and cylindrical molds (u2.54 cm  2.54 cm) for compressive strength tests and MIP tests, respectively. All samples were cured in humidity curing chambers at a constant temperature of 23 °C and a relative humidity of 99% until they were used for testing. Compressive strength tests were conducted at days 3, 7, and 28 days of curing, and the average values of triplicate samples determined. After the compressive strength tests, fractured samples were collected and finely ground for the TG and XRD tests. A hydration-stop procedure was carried out using a

Analytical grade Ca(NO3)2 (Sigma Aldrich, USA), NaNO3 (Sigma Aldrich, USA), and commercial GGBFS (Chunghae material Co., Ltd., Korea) were used in this study. To examine the characteristics of GGBFS, particle size distribution (HELOS (HI199) and RODOS, Sympatec, Clausthal-Zellerfeld, Germany), X-ray fluorescence (XRF, S8 Tiger wavelength dispersive WDXRF spectrometer, Bruker, Billerica, MA, USA), and XRD (Cu-Ka, k = 1.5418 Å, D/Max2500V/ PC, Rigaku, Japan) were examined. Fig. 1 shows the particle size distribution of the GGBFS. The median particle size of the GGBFS was approximately 11 mm. Then, the XRF and XRD measurements showed that the GGBFS consisted mostly of CaO, SiO2 (Table 1) and an amorphous phase (Fig. 2). The surface area and the basicity of GGBFS were 2,970 cm2/g and 1.79, respectively. In Korea, type-3 GGBFS is generally used in construction field and it has surface areas of 4000–6000 cm2/g. Therefore, the GGBFS used in this study had a considerably smaller value than that of type-3. In addition, the Korea Standard (KS) specifies that GGBFS must have basicity

100

1 Distribution

80

0.7

70

0.6

60

0.5

50

0.4

40

0.3

30

0.2

20

0.1

10

0 0.1

1

10

100

Cumulative distribution (%)

Density distribution

90

Cumulative

0.8

60

Fig. 2. XRD pattern of GGBFS.

2. Experimental procedures

0.9

20 30 40 50 Position [°2Theta] (Copper (Cu))

0 1000

Particle size (µm) Fig. 1. Particle size distribution of GGBFS.

Table 1 Chemical oxide composition, surface of area, and basicity of GGBFS. Oxide (wt%) CaO

SiO2

Al2O3

MgO

SO3

TiO2

Fe2O3

K2O

MnO

Na2O

Others

44.78

34.28

13.18

3.48

1.75

0.67

0.51

0.48

0.34

0.29

0.24

2

Note: surface area (g/cm ) of the GGBFS = 2970; basicity of the GGBFS = (CaO + MgO + Al2O3)/SiO2 = 1.79.

841

W.S. Yum et al. / Construction and Building Materials 215 (2019) 839–848 Table 2 The mixture proportions and replacement percentage of each additional chemical of the paste samples (wt%). Group label

Sample label

Binder GGBFS

Control CN

SN

Control 0.5CN 1CN 3CN 5CN 0.5SN 1SN 3SN 5SN

96.0 95.5 95.0 93.0 91.0 95.5 95.0 93.0 91.0

CaO

4.0 4.0

Nitrate salts

Water

w/b

35.0

0.35

Sum

Ca(NO3)2

NaNO3

0.0 0.5 1.0 3.0 5.0 0.0

0.0 0.0

0.5 1.0 3.0 5.0

100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0

Note – CN: calcium nitrate, SN: sodium nitrate.

solvent-exchange method with isopropanol and vacuum drying to prevent further hydration [19]. Diluted paste samples were prepared to measure their pH values using the same mixture proportions of the control, 1CN (sample containing Ca(NO3)2), and 1SN (samples containing NaNO3) as shown in Table 2—with the exception that the w/b ratio was increased to 3. A pH meter (HI 3320, Hanna Instruments INC., Italy) was used to measure the pH of the diluted paste samples. The pH was measured 3 times for each sample every 24 h for the first 3 days after making the binder mixture, and then the average value was used. The diluted pastes were agitated with magnetic stirrers until the end of the experiment to prevent solidification. To verify the effect of Ca(NO3)2 and NaNO3 on the pastes, a Vicat needle test and flowability test were performed according to ASTM C187 and ASTM C230 [20,21], respectively. The auto Vicat apparatus (E004N; Matest, Italy) was used to measure the initial and final setting times of paste, and w/c was set at 0.35. The fresh pastes were cast into conical ring molds and measured 45 times at 15minute intervals. To measure the flowability, the fresh pastes were prepared with w/c = 0.35. The fresh paste was fully cast into a conical mold and then the mold was removed. After removing the conical mold, the table was tapped 25 times for 15 s, and then the average diameter was obtained after measuring it three times. The XRD patterns of the hardened samples were obtained using a high-power powder X-ray diffractometer (D/Max2500V/PC, Rigaku, Japan) with Cu-Ka radiation (k = 1.5418 Å) within a range of 5–60° at 2h. The measured XRD patterns were analyzed using the X’pert HighScore Plus software [22] with the International Centre for Diffraction Data (ICDD) PDF-2 database [23] and the Inorganic Crystal Structure Database (ICSD) [24]. The TG measurement was performed using a thermal analyzer (SDT Q600, TA Instruments, New Castle, DE, USA) with alumina pans. The range of the heating temperature was set from 30 °C to 1,000 °C with a heating rate of 10 °C/min in a nitrogen gas environment. The pore size distributions of hardened samples were measured with the mercury intrusion porosimetry (MIP (Auto pore IV 9500, Micrometrics Instrument Co., GA, USA)). The pressure was applied to 60,000 psi to measure the porosity of the hardened samples and the contact angle was 130°. A sample volume of 125 mm3 (5  5  5 mm) was used. Five identical samples were used to increase the reliability of the results and were immersed in isopropanol until testing. 3. Results and discussion 3.1. Compressive strength The compressive strength results are shown with the inset tables in Fig. 3. All samples containing Ca(NO3)2 (CN group)

showed an increase of strength for all days compared to the control ones (Fig. 3a). The 28-day curing samples showed an increase of 8.4–20% in strength depending on the amount of Ca(NO3)2 added. For the strength measurements at 3 days of curing, the strength was almost doubled for CN containing samples compared to the control. In particular, the compressive strength that was measured at 0.5CN was almost 2 times higher than that of the control sample at 3 days. Therefore, it would be adequate to use only 0.5 wt% Ca (NO3)2 for 3 days of strength enhancement. It is worth noting that despite the absence of dicalcium silicate and tricalcium aluminate [14], which are necessary in triggering the accelerating action of nitrate salts for PC, the use of Ca(NO3)2 still accelerated the strength gain in the CaO-activated GGBFS binder. In addition, the use of NaNO3 (SN group) was evaluated for its strength changing properties (Fig. 3b). The results showed that NaNO3 behaved with a more complex tendency compared to Ca (NO3)2 as an additive. Similar to the use of Ca(NO3)2, NaNO3 also acted as an accelerator given that the 3-day strengths were largely increased after the use of NaNO3 although the increasing effect started to decline when NaNO3 was used over 1 wt%. However, at 28 days, all the samples with NaNO3 produced lower strengths than the control; in particular, in proportion to the weight of NaNO3, the 28-day strength was smaller than that of the control. Thus, the effect of nitrate salts on the CaO-activated GGBFS system was significantly dependent on the type of cation (Ca(NO3)2 vs. NaNO3). Thus, the use of Ca(NO3)2 seemed to be more advantageous in the CaO-activated GGBFS system for improving the strength of the binder at all days compared to the addition of NaNO3. 3.2. Measurement results of the pH levels The pH is closely related to the development of strength in cementless binder systems (e.g., alkali- or CaO-activated GGBFSs). A higher pH dissolves more of the glass phase of raw materials (e.g., GGBFS and fly ash), resulting in the generation of more Ca, Si, or Al ions, which are necessary for developing strength by forming the reaction products such as C-S-H [25,26]. As such, the pH values of the control, 1CN, and 1SN were measured during the first 3 days using diluted paste samples (Fig. 4). The highest pH was obtained in 1SN. When NaNO3 is present, the solubility of Ca(OH)2 is generally increased in aqueous salt solutions [27]. As Ca(OH)2 is a hydrated form of CaO, the highest pH of 1SN was likely obtained due to the increased dissolution of CaO. As such, it is likely that the glass phase of GGBFS was more dissolved in 1SN samples within those 3 days due to the high pH. This results in the production of more reaction products, and consequently, the greater strength among samples at 3 days. However, it should be noted that the 28-day strengths of all SN samples were

842

W.S. Yum et al. / Construction and Building Materials 215 (2019) 839–848

Compressive Strength (MPa)

35.00

30.00

25.00

20.00

15.00

10.00

5.00

0.00 3 days 7 days 28 days

Control 7.15 16.25 27.30

0.5CN 13.57 19.69 30.31

1CN 15.12 21.84 30.02

3CN 16.14 21.00 29.58

5CN 16.19 23.72 32.92

1SN 15.95 20.36 23.37

3SN 13.78 16.81 21.97

5SN 12.14 15.67 20.87

(a)

Compressive Strength (MPa)

35.00

30.00

25.00

20.00

15.00

10.00

5.00

0.00 3 days 7 days 28 days

Control 7.15 16.25 27.30

0.5SN 14.65 20.15 24.68

(b) Fig. 3. Compressive strength analysis of various binder mixtures with Ca(NO3)2 and NaNO3 additions. (a) Represents the Ca(NO3)2 (CN) group against the control one. (b) Represents the NaNO3 (SN) group against the control one.

lower than that of control. As to the possible reason for the lower strengths of the SN samples at 28 days will be discussed in the Section 3.4 TG/DTG. For the 1CN samples, the additional supply of calcium ions from the added Ca(NO3)2 probably induced a common ion effect [28], which suppressed the solubility of CaO, and thus, slightly reduced the pH. In general, a high pH is helpful to gain a high strength in cementless binders; however, although 1CN showed the lowest pH value for all days, it produced a significantly higher strength than the control; it is worth noting that Bellman et al. [29] demonstrated that GGBFS was still effectively activated at a lower pH (12) by simply increasing the concentration of calcium ions in the pore solutions by adding a small amount of soluble calcium

salts (e.g., 3 wt% of Ca(NO3)2), which resulted in a significant improvement in strength development. Therefore, the strength improvement of 1CN was likely achieved similarly compared to the study of Bellman et al. [29].

3.3. Setting times and flowability The setting times and flowabilities were measured for the mixture proportions of Table 2, and the results are given in Table 3. According to the ASTM C 191 [30], the initial setting time of Portland cement should be longer than 45 min and the final setting must take place within 345 min. Thus, Table 3 showed that all

W.S. Yum et al. / Construction and Building Materials 215 (2019) 839–848

843

13.1 Control 1CN

12.99

13

pH

1SN

12.9

12.86

12.8

12.78

12.90

12.85

12.79

12.83

12.74

12.73

12.7 1

2 Time (days)

3

Fig. 4. The change of pH of diluted paste samples during the first 3 days.

Table 3 The results of setting time and flowability of fresh pastes. Sample Labels

Control CN group

SN group

Setting times (min)

Flowability (mm)

Initial set

Final set

0.5CN 1CN 3CN 5CN

540 480 780 900 930

1140 1,110 1,260 1,320 1,320

16.5 17.8 18.8 19.0 20.8

0.5SN 1SN 3SN 5SN

450 300 390 390

1,110 1,080 1,200 1,230

18.0 18.8 19.5 19.7

samples in this study exhibited relatively long initial and final setting times compared to those of Portland cement. The dosages of Ca(NO3)2 and NaNO3 affected the setting times and flowability of sample pastes. Compared to the values of control, the setting times tended to be shortened relatively at small dosages, but as the dosage increased, the setting times have generally increased; on the other hand, the increased use of both Ca (NO3)2 and NaNO3 improved flowability. 3.4. XRD analysis The XRD patterns of hardened samples are provided with the reference patterns of the identified phases in Fig. 5. The type of reaction products analyzed with the XRD was very similar between the CN and SN group when the same weight of salts was used, although their compressive strengths were significantly different. Additionally, there was no difference in the type of reaction products for all samples between 3 and 28 days of curing. C-S-H [31] and akermanite (Ca2Mg(Si2O7)) were identified in all samples. Portlandite (Ca(OH)2) was observed in all samples with less than 3–5 wt% of Ca(NO3)2. All samples in this study showed carbonation to some extent in XRD, possibly due to the carbonation of portlandite with the atmospheric CO2 during the curing process [32–34]. An Al2O3-Fe2O3-mono (AFm) phase was also clearly identified in the samples containing any type of nitrate salts with more than 3 wt%. This seemed to occur mainly when the quantity of salts was at 5 wt%, causing the XRD peaks of this phase to increase significantly. The ICDD PDF-2 database indicates that this AFm phase in our study was nitrate (NO3) AFm (Ca4Al2(OH)12 (NO3)2
4H2O) [35]. According to other previous studies [36,37], this phase could also be a nitrite (NO2) AFm (Ca4Al2(OH)12(NO2)2
4H2O) phase or a mixture of NO3- and NO2-AFm phases. This could be possible because when NO3-AFm loses water during the drying processes (e.g., using a vacuum desiccator for sample preparation in this study), the XRD pattern dried, and the NO3-AFm phase may become similar to that of the NO2-AFm one [36].

Fig. 5. XRD patterns of hardened samples. The investigated samples were as follows: (a) the CN group on day 3, (b) the CN group on day 28, (c) the SN group on day 3, and (d) the SN group on day 28. The numbers in brackets indicate the reference numbers of identified phases in the ICDD PDF-2 database. Note that the reference pattern of C-S-H was obtained from [38] after removing the reflections of calcium hydroxide.

In the CN group, regardless of curing duration, weaker peaks of portlandite were identified when more Ca(NO3)2 was used as a substitution. Indeed, portlandite was not identified in the 3CN and 5CN samples. However, unlike the CN group, portlandite was clearly identified in all SN group samples for entire curing duration.

3.5. TG/DTG The results from the TG/DTG tests are illustrated in Fig. 6. For all samples, the decomposition of C-S-H, NO3- and/or NO2-AFm,

844

W.S. Yum et al. / Construction and Building Materials 215 (2019) 839–848

Fig. 6. TG/DTG results of hardened samples during various curing days. The following conditions were investigated: (a) the CN group on day 3, (b) the CN group on day 28, (c) the SN group at day 3, and (d) the SN group that shows an enlarged window of the 0–350 °C range on day 28.

Ca(OH)2, and calcite was identified [37,39]. Overall, the TG/DTG results were consistent with the XRD results regarding the phase identification of reaction products. In Fig. 6, each figure illustrates that for each group (CN or SN) on the same day of curing, the total

weight loss until 1000 °C was noticeably increased with TG. This occurred because a higher amount of nitrate salts was used. Additionally, given that the raw GGBFS in this study barely showed a weight loss until 1000 °C with the TG (i.e., LOI = 0.8 wt%), the

845

W.S. Yum et al. / Construction and Building Materials 215 (2019) 839–848 Table 4 Total porosity and average pore size of hardened samples. Group label

Sample label

Curing days (3 days)

Curing days (28 days)

Total porosity (%)

Average pore size (nm)

Total porosity (%)

Average pore size (nm)

CN

Control 0.5CN 1CN 3CN 5CN

39.45 39.29 34.62 34.86 34.73

27.5 26.6 23.9 23.3 19.8

38.51 34.23 34.54 33.06 32.58

16.2 17.0 15.9 15.8 14.4

SN

Control 0.5SN 1SN 3SN 5SN

39.45 34.41 33.84 35.75 36.48

27.5 29.1 23.9 24.0 24.9

38.51 36.58 33.87 34.25 37.88

16.2 19.5 20.7 21.6 20.0

increase of the total weight loss indirectly suggests that the increased use of nitrate salts resulted in more dissolution of GGBFS because the additional formation of reaction products required an additional supply of vital elements (e.g., Si, Al), which were generated only from GGBFS in this study. In both the CN and SN groups (Fig. 6), the weight losses below 200 °C were mainly due to thermal decompositions of C-S-H and AFm phases (NO3- and/or NO2-AFm); C-S-H is a core phase to develop strength of the hardened sample. However, given that very little or no NO3/NO2-AFm phases were found in XRD for all days, when Ca(NO3)2 or NaNO3 was replaced up to 1 wt%, in the samples with 0–1 wt% of nitrate salts (i.e., 0.5CN, 1CN, 0.5SN, and 1SN), the DTG peaks below 200 °C were mainly due to the dehydration of C-S-H. At 3 days of curing, the weight loss of C-S-H in the samples of both groups clearly increased as the quantity of nitrate salts increased. These results support that the use of nitrate salts—regardless of the type of cation of the salts—was clearly beneficial to produce C-S-H when, at 28 days, the quantity of nitrate salts increased (up to 1 wt%). Here, the CN group samples did not show any decrease of the DTG peak that indicates C-S-H levels; however, the SN group samples did clearly display a reduction of C-S-H based on the DTG peak (see the area with the shading lines in the enlarged figure in Fig. 6). Thus, the effect of using NaNO3 was changed from being advantageous (at 3 days) to detrimental (at 28 days) for developing strength as the curing process progressed. In general, NO3- and NO2-AFm phases have their thermal decompositions in similar temperature ranges of 0–150 °C, 200– 300 °C, 400–600 °C, and near 700 °C in TG [36,37]; the weight losses occur through the progressive reduction of water (H2O), hydroxyl (OH) groups, nitrate (NO3), and nitrite (NO2) or combination of these reductions. In this study, the DTG peak was around 250 °C, which was an isolated DTG peak of NO3- and NO2-AFm phases from the DTG peaks of other reaction products, and gradually increased as more nitrate salts were used. For this reason, the substitution of nitrate salts was likely the main cause for the formation of the NO3- and/or NO2-AFm phases. With temperatures of 400–600 °C during the TG, along with dehydroxylation of the OH groups, the reduction of NO2 was the main cause of weight loss for the NO3- and NO2-AFm phases [36,37]. To be more exact, the major weight loss of NO2-AFm occurred near 450 °C (see the d in Fig. 6) and near 580 °C for NO3-AFm (see the r in Fig. 6), which was found previously as well [36]. Additionally, unlike the samples with nitrate salts of up to 3 wt%, the 5 wt% samples showed a significant increase of the DTG peak near 450 °C, regardless of the nitrate salt type or days of curing. This was especially the case for 5SN on day 28 of curing, showing a large DTG peak (the gray area in the 400–580 °C range, see Fig. 6d). Next, three identically repeated samples were executed to rule out a potential experimental error to confirm this

large DTG peak; however, all the repeated samples further confirmed the initial result. For this reason, the substitution of nitrate salts was likely the main cause for the formation of the NO3- and/or NO2-AFm phases. In this study, although it might be speculated that a significant increase of NO3- and NO2-AFm phases might affect the formation of C-S-H and the compressive strength of samples, it was difficult to find a clear correlation; furthermore, no study has shown that nitrate salt phases affect compressive strength and precipitation of C-S-H in cementitious binders. 3.6. MIP The MIP results of hardened samples with the total porosities are shown in Table 4 and Fig. 7. The actual pore size distributions of hardened paste samples might be different the MIP results; however, earlier studies demonstrated that the MIP results were still helpful in comparing pore size distributions between samples [40–42]. In general, the strength of a hardened cementitious paste is inversely proportional to its total porosity and overall pore size. Generally, it is known that the effect of pores on the properties of binder differs depending on whether the size of the pores is larger or smaller than 0.05 lm. If the pore size is larger than 0.05 lm, it affects the strength and permeability of the binder. On the other hand, if the pore size is smaller than 0.05 lm, it affects the drying shrinkage and creep. Therefore, voids smaller than 0.05 lm have relatively little effect on the strength of the binder [43]. In this study, as curing days passed, all samples in each group largely lost their pores larger than 0.05 lm, resulting in the reduction of the average pore size, from 3 to 28 days. As a result, the strength of all samples was largely improved with additional curing days. Further investigation of the pore size distribution showed that it was significantly dependent on the cation type of the nitrate salts. For instance, in the CN group, regardless of the curing days, the shape of the distribution curve was differentiated into two distinctly different curves depending on the dosage of Ca(NO3)2 (see the inset figures in Fig. 7a). These results show that when Ca (NO3)2 increased from 0 to 1 wt%, the curve gradually contracted in a similar form of the curve shape. When the dosage was over 3 to 5 wt%, the distribution curve was significantly altered due to the considerable reduction of pores smaller than 0.05 lm. As opposed to the CN group, regardless of curing days, the curve shape of the SN group was not significantly changed by the dosage of NaNO3, and changes in the overall pore size and total porosity were minimal. Taken together, it was not appropriate to simply compare the values of total porosity or average pore size between the CN and SN groups to explain the different strengths of them. In the CN group, although the curve shape of the pore size distribution was largely affected by the quantity of Ca(NO3)2, the total porosity was not notably affected regardless of curing period

846

W.S. Yum et al. / Construction and Building Materials 215 (2019) 839–848

Fig. 7. The MIP results of hardened pastes during various curing days. The following conditions were investigated: (a) the CN group at 3 days, (b) the CN group at 28 days, (c) the SN group at 3 days, and (d) the SN group at 28 days.

(except 0.5CN). For example, the 1CN and 3CN samples showed similar values (see Table 4) for the total porosity on all days of curing but noticeably different ones for their size distributions. Consequently, given that the strength values were very similar between all samples at every same day of the curing, the relationship between strength and pore size distribution in the CN group was difficult to interpret accurately. In the SN group, at 3 days of curing, the 1SN sample showed the lowest level of porosity with an average pore size among samples. This was in accordance with its results concerning the strength, which was the strongest among the SN group. Nonetheless, at 28 days of curing, the strengths of the SN group samples were contradictorily lower than that of the control, despite showing a smaller value for the total porosities. This cause can be seen in the curves by the shift in distribution to the right, which represents a smaller pore size, with higher weights of NaNO3. In sum, these results show that the average pore size was more important than the total porosity in determining the strength of the SN group at 28 days of curing.

As mentioned earlier, in this study, the compressive strength was generally inversely related to its overall size of pores, which is represented by the average pore size, as shown in Fig. 8. In Fig. 8a, the CN group samples at 3 days showed an exact inverse relationship between strength and pore size although its tendency was rather reduced at 28 days (Fig. 8b). In Fig. 8c, although the 0.5SN sample at 3 days had a higher compressive strength than the control at 3 days despite its higher average pore size, the other SN samples generally followed the inverse relationship at all days. 4. Conclusions In this study, the influences of the nitrate salts Ca(NO3)2 and NaNO3 on the CaO-activated GGBFS system were investigated.  The use of both nitrate salts was clearly beneficial in developing strength in the early days of curing; the 3-day strengths were almost doubled in all samples regardless of the type of cation in the nitrate salts. However, at 28 days, while the use of

847

3-day pore size

25

30 25 20

20

15 10

15

5 0 Control

10 0.5CN

1CN

3CN

5CN

40 35

28-day pore size

30

25

25 20

20

15 10

15

5 0

Control

Sample label

30 28-day strength

0.5CN

25 20

20 15

15

10 3-day strength 3-day pore size

0.5SN

1SN

3SN

5SN

10

Compressive strength (MPa)

25

30

Average pore size (nm)

Compressive strength (MPa)

30

35

Control

10

(b)

40

0

5CN

3CN

Sample label

(a)

5

1CN

30

40 28-day strength

35

28-day pore size

25

30 25

20

20 15

15

10 5 0

Control

0.5SN

1SN

3SN

5SN

Average pore size (nm)

35

Average pore size (nm)

3-day strength

Compressive strength (MPa)

30

40

Average pore size (nm)

Compressive strength (MPa)

W.S. Yum et al. / Construction and Building Materials 215 (2019) 839–848

10

Sample label

Sample label

(c)

(d)

Fig. 8. The tendency of average pore size and compressive strength varying with the curing period: (a) the CN group at 3 days, (b) the CN group at 28 days, (c) the SN group at 3 days, and (d) the SN group at 28 days.

Ca(NO3)2 was still effective in increasing strength, that of NaNO3 was not advantageous in improving strength compared to the control sample without any nitrate salts.  The analysis of pH showed that the use of NaNO3 increased the pH of the paste samples because it increased the solubility of CaO, resulting in the increase of strength due to the increase of dissolving GGBFS by the high pH during the early days (e.g., before 3 days). In turn, the use of Ca(NO3)2 caused a decrease in the sample’s pH due to the common ion effect of calcium; however, as the addition of Ca(NO3)2 also provided an additional supply of calcium ions, Ca(NO3)2 also improved the strength of the samples during the early days of curing.  The types of reaction products were very similar in all samples when the same weight of salts was used, regardless of the type of nitrate salt and the length of curing days.  C-S-H was the main phase that improved the binder’s strength development in this study. On day 3 of curing, the use of nitrate salts was clearly beneficial to produce C-S-H, regardless of the cation type in the salts. On day 28 of curing, however, the samples with NaNO3 clearly displayed a reduction of C-S-H when the weight of the salt increased, which was not the case for the Ca(NO3)2 samples. This changed the advantageous use of NaNO3 on day 3 of curing for the developing strength during the curing process into a detrimental one on day 28.

 The NO3- or/and NO2-AFm phases were generated in considerable amounts when nitrate salts were replaced by 3–5 wt%. At 3 wt% of both nitrate salts, the NO3-AFm phase was likely to be produced more, while at 5 wt%, NO2-AFm seemed to form more.  In this study, all samples mostly lost their pores larger than 0.05 lm as curing days passed (from 3 to 28 days), resulting in the reduction of the average pore size and the improvement of their strength. However, the pore size distribution was significantly dependent on the cation type of nitrate salts. The shapes of the distribution curves of the samples with Ca(NO3)2 were roughly differentiated into two distinctly different curve shapes depending on the quantity of Ca(NO3)2, while those of the NaNO3 samples were not significantly changed by the dosage of NaNO3. Conflict of interest None. Acknowledgments This study was supported by the Basic Science Research Programs (NRF-2016R1D1A1B03932908) through the National

848

W.S. Yum et al. / Construction and Building Materials 215 (2019) 839–848

Research Foundation of Korea (NRF), and also by a grant (144132) from the Technology Business Innovation Program (TBIP) funded by the Ministry of Land, Infrastructure and Transport of Korea. References [1] W. Jiang, Alkali-activated Cementitious Materials: Mechanisms, Microstructure and Properties Ph.D. Thesis, The Pennsylvania State University, Pennsylvania, US, 1997. [2] T. Bakharev, J.G. Sanjayan, Y.-B. Cheng, Alkali activation of Australian slag cements, Cem. Concr. Res. 29 (1) (1999) 113–120. [3] D. Li, Y. Chen, J. Shen, J. Su, X. Wu, The influence of alkalinity on activation and microstructure of fly ash, Cem. Concr. Res. 30 (6) (2000) 881–886. [4] S. Alonso, A. Palomo, Alkaline activation of metakaolin and calcium hydroxide mixtures: influence of temperature, activator concentration and solids ratio, Mater. Lett. 47 (1–2) (2001) 55–62. [5] M.S. Kim, Y. Jun, C. Lee, J.E. Oh, Use of CaO as an activator for producing a pricecompetitive non-cement structural binder using ground granulated blast furnace slag, Cem. Concr. Res. 54 (2013) 208–214. [6] W.S. Yum, Y. Jeong, S. Yoon, D. Jeon, Y. Jun, J.E. Oh, Effects of CaCl2 on hydration and properties of lime (CaO)-activated slag/fly ash binder, Cem. Concr. Compos. 84 (2017) 111–123. [7] E. Altan, S.T. Erdog˘an, Alkali activation of a slag at ambient and elevated temperatures, Cem. Concr. Compos. 34 (2) (2012) 131–139. [8] Y. Jeong, W.S. Yum, D. Jeon, J.E. Oh, Strength development and microstructural characteristics of barium hydroxide-activated ground granulated blast furnace slag, Cem. Concr. Compos. 79 (2017) 34–44. [9] D. Ravikumar, N. Neithalath, Effects of activator characteristics on the reaction product formation in slag binders activated using alkali silicate powder and NaOH, Cem. Concr. Compos. 34 (7) (2012) 809–818. [10] K.-H. Yang, A.-R. Cho, J.-K. Song, S.-H. Nam, Hydration products and strength development of calcium hydroxide-based alkali-activated slag mortars, Constr. Build. Mater. 29 (2012) 410–419. [11] ACI 318-14, Building Code Requirements for Structural Concrete American concrete Institute Farmington Hills, Michigan, 2014. [12] ACI Committee 212 reports ‘‘Admixture for Concrete, ACI 212.1 R-81” and ‘‘Guide for Use of Admixtures in Concrete, ACI 212.2 R-81,” Concrete International, 1981, pp. 64-65. [13] A.-M. Paillere, Application of Admixtures in Concrete, CRC Press, 2014. [14] H. Justnes, E.C. Nygaard, Technical calcium nitrate as set accelerator for cement at low temperatures, Cem. Concr. Res. 25 (8) (1995) 1766–1774. [15] P. Rosskopf, F. Linton, R. Peppler, Effect of various accelerating chemical admixtures on setting and strength development of concrete, J. Testing Eval. 3 (4) (1975) 322–330. [16] R. Myrdal, Accelerating admixtures for concrete. State of the art, (2007). [17] I.L. Organization, International Labour Organization. https://www.ilo.org/. [18] ASTM C 305, Standard Practice for Mechanical Mixing of Hydraulic Cement Pastes and Mortars of Plastic Consistency, ASTM International, West Conshohocken, PA, 1999. [19] J. Zhang, G.W. Scherer, Comparison of methods for arresting hydration of cement, Cem. Concr. Res. 41 (10) (2011) 1024–1036. [20] ASTM C 187, Standard Test Method for Normal Consistency of Hydraulic Cement,, ASTM International, West Conshohocken, PA, 1998. [21] ASTM C230, Flow Table for Use in Tests of Hydraulic Cement, ASTM International, West Conshohocken, PA, 2003.

[22] X‘pert Highscore Plus software, version 3.0e, PANalytical, Almelo: Netherlands, 2012. [23] A. Belsky, M. Hellenbrandt, V.L. Karen, P. Luksch, New developments in the Inorganic Crystal Structure Database (ICSD): accessibility in support of materials research and design, Acta Crystallogr. Sect. B: Struct. Sci. 58 (3) (2002) 364–369. [24] R. Allmann, R. Hinek, The introduction of structure types into the Inorganic Crystal Structure Database ICSD, Acta Crystallogr. Sect. A: Found. Crystallogr. 63 (5) (2007) 412–417. [25] S. Song, H.M. Jennings, Pore solution chemistry of alkali-activated ground granulated blast-furnace slag1, Cem. Concr. Res. 29 (2) (1999) 159–170. [26] H. Park, Y. Jeong, J.-H. Jeong, J.E. Oh, Strength Development and Hydration Behavior of Self-Activation of Commercial Ground Granulated Blast-Furnace Slag Mixed with Purified Water, Materials 9 (3) (2016) 185. [27] J. Johnston, C. Grove, The solubility of calcium hydroxide in aqueous salt solutions, J. Am. Chem. Soc. 53 (11) (1931) 3976–3991. [28] D.W. Oxtoby, H.P. Gillis, L.J. Butler, Principles of modern chemistry, Cengage learning, 2015. [29] F. Bellmann, J. Stark, Activation of blast furnace slag by a new method, Cem. Concr. Res. 39 (8) (2009) 644–650. [30] ASTM C 191, Standard Test Method For Time of Setting of Hydraulic Cement by Vicat Needle, C 191, ASTM International, West Conshohocken, PA, 2004. [31] H.F. Taylor, Cement chemistry, Thomas Telford, 1997. [32] K. Gu, F. Jin, A. Al-Tabbaa, B. Shi, J. Liu, Mechanical and hydration properties of ground granulated blastfurnace slag pastes activated with MgO–CaO mixtures, Constr. Build. Mater. 69 (2014) 101–108. [33] O. Burciaga-Díaz, I. Betancourt-Castillo, Characterization of novel blastfurnace slag cement pastes and mortars activated with a reactive mixture of MgO-NaOH, Cem. Concr. Res. 105 (2018) 54–63. [34] C. Shi, D. Roy, P. Krivenko, Alkali-activated Cements and Concretes, CRC Press, 2003. [35] G. Renaudin, M. Francois, The lamellar double-hydroxide (LDH) compound with composition 3CaO. Al2O3. Ca (NO3) 2.10 H2O, Acta Crystallogr. Section C 55 (6) (1999) 835–838. [36] M. Balonis, M. Medala, F.P. Glasser, Influence of calcium nitrate and nitrite on the constitution of AFm and AFt cement hydrates, Adv. Cem. Res. 23 (3) (2011) 129–143. [37] G. Renaudin, J.-P. Rapin, B. Humbert, M. François, Thermal behaviour of the nitrated AFm phase Ca4Al2 (OH) 12 (NO3) 2 4H2O and structure determination of the intermediate hydrate Ca4Al2 (OH) 12 (NO3) 2 2H2O, Cem. Concr. Res. 30 (2) (2000) 307–314. [38] K. Mohan, H. Taylor, Analytical electron microscopy of cement pastes: IV, bdicalcium silicate pastes, J. Am. Ceram. Soc. 64 (12) (1981) 717–719. [39] H. Song, Y. Jeong, S. Bae, Y. Jun, S. Yoon, J.E. Oh, A study of thermal decomposition of phases in cementitious systems using HT-XRD and TG, Constr. Build. Mater. 169 (2018) 648–661. [40] S. Diamond, Mercury porosimetry: an inappropriate method for the measurement of pore size distributions in cement-based materials, Cem. Concr. Res. 30 (10) (2000) 1517–1525. [41] D. Shi, D.N. Winslow, Contact angle and damage during mercury intrusion into cement paste, Cem. Concr. Res. 15 (4) (1985) 645–654. [42] H. Park, Y. Jeong, Y. Jun, J.-H. Jeong, J.E. Oh, Strength enhancement and poresize refinement in clinker-free CaO-activated GGBFS systems through substitution with gypsum, Cem. Concr. Compos. 68 (2016) 57–65. [43] K. Metha, P. Monteiro, Concrete: Microstructure, Properties, and Materials, third ed., McGraw-Hill Publishing, New York, 2006.