L ammonium nitrate solution

Cement and Concrete Research 53 (2013) 44–50

Contents lists available at ScienceDirect

Cement and Concrete Research journal homepage: http://ees.elsevier.com/CEMCON/default.asp

Solid–liquid equilibrium curve of calcium in 6 mol/L ammonium nitrate solution Keshu Wan a,b,⁎, Lin Li b, Wei Sun a,b a b

Jiangsu Key Laboratory of Construction Materials, Nanjing 211189, People's Republic of China School of Materials Science and Engineering, Southeast University, Nanjing 211189, People's Republic of China

a r t i c l e

i n f o

Article history: Received 17 February 2013 Accepted 3 June 2013 Keywords: Calcium leaching Acceleration (A) Cement paste (D) Nitrate (D)

a b s t r a c t Calcium leaching of cement-based materials is of concern for scientific and application significance. Solid–liquid equilibrium curve of calcium is crucial for understanding the calcium leaching mechanism and for calcium leaching modeling. Solid–liquid equilibrium curve of Portland cement system in 6 mol/L ammonium nitrate solution is experimentally obtained using the dissolving equilibrium between cement paste powders and simulated pore solutions. The obtained equilibrium curve in nitrate solution has a similar three-stage form to that in water, but the concentrations of dissolved calcium increase up to two orders of magnitude, which is the main acceleration mechanism for calcium leaching. Besides the solid–liquid equilibrium curve of calcium, the equilibrium curves between other solid elements (including sulfur, aluminum, manganese, iron) and dissolved calcium are presented. It is found that sulfur and manganese are also leached in three stages, but aluminum and iron are not leachable. © 2013 Elsevier Ltd. All rights reserved.

1. Introduction 1.1. Calcium leaching of cement-based materials Calcium (Ca) leaching of cement-based materials is of concern in the structures used for radioactive waste disposal containers, underground pipes, dams, and water tanks that are constantly composed to the low pH environment. Generally speaking, the concentration gradients between the pore solution and the environment water lead to diffusion of dissolved Ca from the pore solution to the surrounding water. Once the concentration of pore solution is reduced, the Ca in solid skeleton, mainly the calcium hydroxide (CH) and calcium–silicate–hydrate (C–S–H), will dissolve gradually. These processes will increase the porosity, consequently increase the permeability and decrease the strength of the cement matrix (such as in [1–10]). For modeling research of Ca leaching, most of the investigations use the solid–liquid equilibrium curve to describe the dissolution process, and use the diffusion to describe the transport process (such as in [2,4,6]). All these models strongly depend on the solid–liquid equilibrium curve, which describes the solubility of cement hydration product and controls the Ca dissolving and leaching processes. Experimentally, it was reported that the leaching front of concrete submerged in still field water for 100 years was only about 5 to 10 mm [11,12]. Considering the time consuming of the field water, ⁎ Corresponding author at: School of Materials Science and Engineering, Southeast University, Nanjing 211189, People's Republic of China. Tel.: + 86 25 52090670; fax: +86 25 52090667. E-mail address: [email protected] (K. Wan). 0008-8846/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.cemconres.2013.06.003

most of the laboratory tests applied accelerated leaching protocols with different accelerated methods, mainly 6 mol/L (6 M) ammonium nitrate solution [7–10,13–18], ammonium chloride solution [19], deionized water [13,20–24], and electrochemical method [25,26]. Among these methods, 6 M ammonium nitrate solution, which can accelerate the leaching speed two orders while still get the same end products [17], is the most effective method and has been intensively used to experimentally investigate Ca leaching behavior. 1.2. Solid–liquid equilibrium curve of Ca in water The solid–liquid equilibrium curve of Ca in water stems from the equilibrium in the CaO–SiO2–H2O system. Since Le Chatelier's classic studies on C–S–H in the 1880s [27], there have been numerous investigations on the dissolving equilibrium of C–S–H or cement paste at room temperature (such as in refs [28–32]). Because dissolution is much faster than diffusion in the leaching process of cement-based materials, Buil [2] firstly applied the solid– liquid equilibrium curve to model the leaching process in 1992, where the solid–liquid equilibrium curve was used to relate the Ca in solid skeleton s and the Ca dissolved in pore solution c. Berner [3] also raised the equilibrium curve to describe the dissolution behavior of hydrated cement in 1992. Since that, the equilibrium curve has been applied to build Ca leaching models or to explain the leaching phenomenon in numerous studies (such as in [6,16,19,21–23]). The original Buil's equilibrium curve is mathematically simple and with reasonable accuracy. The equilibrium curve has been slightly modified in later research. One typical modified equilibrium curve [16,19] is expressed by Eq. (1), and the corresponded equilibrium

K. Wan et al. / Cement and Concrete Research 53 (2013) 44–50

curve [16] was compared with the initial equilibrium curve by Buil [2] in Fig. 1. 8
  c 1=3  −2 3 3 2 > > > C c þ c CSH > > ceq x 3 x 2 > > <  1  c 11=3  cs ¼ CCSH eq > c > >  c 1=3  > C > 3 > > þ eq CH 3 ðc−x2 Þ C : CSH ceq ðc −x2 Þ

0≤c≤x1 x1 bc≤x2 x2 bc≤c

ð1Þ

eq

where x1 is the Ca concentration in pore solution when C–S–H dissolves quickly (mmol/L); x2 is the Ca concentration in pore solution when CH has completely dissolved, and C–S–H begins to dissolve (mmol/L); ceq is the saturated concentration of dissolved Ca in deionized water under normal temperature (mmol/L). Normally x1 is taken as 2 mmol/L [16,19], x2 is taken as (ceq − 3) mmol/L [16,19] or (ceq − 1) mmol/L [21–23], and ceq is taken as 20 mmol/L [16,19] or 22 mmol/L [21–23]. According to Eq. (1) and Fig. 1, the dissolving and leaching processes in water can be interpreted as the following three stages. In the first stage (x2 b c b ceq), CH is quickly dissolved and leached. In the second slow and central stage (x1 b c b x2), the Ca in C–S–H is partially dissolved and leached, resulted Ca/Si molar ratio (Ca/Si) in C–S–H from about 1.6 to about 0.8 [29]. It is still in this second period, the Ca in sulfoaluminates is dissolved and leached [7,21]. In the third stage (0 b c b x1), when the Ca concentration in pore solution is less than x1, about 2 mmol/L, the partially leached C–S–H is quickly and totally decalcified to be silica gel (such as in [3,6]). 1.3. Solubility in ammonium nitrate solution and unsettled questions The dissolution of CH in ammonium nitrate solution can be written according to the following chemical reaction [8]: CaðOHÞ2 þ 2NH4 NO3 →CaðNO3 Þ2 þ 2NH3 þ 2H2 O:

ð2Þ

For the dissolution of C–S–H, a similar reaction equation is followed. The gaseous reaction product NH3 and the high solubility of reaction product calcium nitrate favor the reaction process greatly. And the high Ca concentration due to high solubility of calcium nitrate changes the equilibrium curve and increases the concentration gradient. Several papers have investigated the accelerated Ca leaching of cement paste in 6 M ammonium nitrate solution. At room temperature, the saturated Ca concentration in the pore solution of cement paste is only about 20 mmol/L in water, while it is reported to be 2730 mmol/L [6] or 2900 mmol/L [8] in 6 M ammonium nitrate solution. For x1 and x2, Tognazzi [18] thinks that x1 does not change while x2 changes to a value close to 2730 mmol/L; Gerard et al. [6] think that both x1 and x2 do not change (still in the range of 0–20 mmol/L), and that the main acceleration effect is controlled by the magnifying factor of 60 on the

45

effective diffusion coefficient; Wan et al. [16] think that both x1 and x2 change and they take x1 as 273 mmol/L and x2 as 2320 mmol/L in their equilibrium curve in 6 M ammonium nitrate solution. Pichler et al. [7] propose a simple linear equilibrium curve in their acceleration model. Besides, debate exists on whether sulfoaluminates is leachable or not in 6 M ammonium nitrate solution (such as in [7,33–35]). Although 6 M ammonium nitrate solution is widely applied for Ca leaching research, several issues concerned the equilibrium curve in 6 M ammonium nitrate solution remain unsettled: (i) does the equilibrium curve in 6 M ammonium nitrate solution still keep the threestage form as that in water; (ii) if it keeps the same form as that in water, what are the values of x1, x2 and saturated concentration ceq; (iii) does the sulfoaluminates leach in 6 M ammonium nitrate solution? 1.4. Current study To clarify the accelerated Ca leaching process and the acceleration mechanism, this research focuses mainly on the solid–liquid equilibrium curve of Ca in 6 M ammonium nitrate solution. Special experiments are designed to obtain this equilibrium curve. Once the equilibrium curve is experimentally obtained, the several unsettled questions stated in Section 1.3 can be answered. 2. Experimental 2.1. Experimental design The solid–liquid equilibrium curve of Ca in water is one kind of phase diagram [1,2,17]. Considering that Ca leaching in 6 M ammonium nitrate solution is still a diffusion controlling process and a dynamic equilibrium between s and c still holds, the solid–liquid equilibrium curve of Ca in 6 M ammonium nitrate solution is also one kind of phase diagram. When cement-based materials are exposed in 6 M ammonium nitrate solution for a long time, dynamic equilibrium will be established, and concentration gradients of different ions will form in pore solution finally. Considering that the free K+ and Na+ leach very quickly, only after a short leach period, the influence of these free K+/Na+ can be neglected, so the main ions in pore solution are − only Ca2+, NH+ 4 and NO3 which keep ionic charge neutral. Just as the solid–liquid equilibrium curve in water describes the equilibrium of water (H+, OH−)-dissolved Ca-solid hydrated product, the solid–liquid equilibrium curve in 6 M ammonium nitrate solution describes the − equilibrium of 6 M ammonium nitrate solution (NH+ 4 , NO3 )-dissolved Ca-solid hydrated product. As illustrated in Fig. 2, from the leaching surface to the leaching front, the Ca2+ concentration will increase because of diffusion of Ca from the inner part to the surface, while the NH+ 4 concentration will decrease because of chemical reactions. Now let us consider the boundary conditions of the ion concentrations in pore solution. On

10000

Ca in solid [mmol/L]

Ref[16] 8000 6000 4000 2000 0 0

2

4

6

8

10 12 14 16 18 20

Ca dissolved [mmol/L] Fig. 1. Ref [2]: the collected dissolution data and the proposed solid–liquid equilibrium curve in water taken from reference [2]; Ref [16]: the modeling solid–liquid equilibrium curve in water taken from reference [16].

46

K. Wan et al. / Cement and Concrete Research 53 (2013) 44–50

the leaching surfaces, the concentration of NO− 3 is 6000 mmol/L, the concentration of Ca2+ is close to 0, and the concentration of NH+ 4 is also close to 6000 mmol/L. On the leaching front, we suppose that the concentration of NO− 3 is still 6000 mmol/L, the concentration of NH+ 4 is close to 0 because of chemical reaction, and the concentration of Ca2+ is about 3000 mmol/L. To determine this phase diagram, the environmental conditions need to be fixed. The temperature is fixed at room temperature (about 20 °C), and the pressure is fixed at 1 atm. The solid–liquid equilibrium curve of Ca is the equilibrium Ca in solid skeleton s and the equilibrium Ca concentration in the pore solution c, both of which are spatially distributed and hard to measure directly. In this research, simulated pore solutions are applied. To express the different dissolution degrees at different spatial positions, different concentrations of the simulated pore solutions are prepared using ammonium nitrate and calcium nitrate. Besides, from the trial experiments using one gram cement powder in 100 mL 6 M ammonium nitrate solution, it is found that low concentrations of dissolved Ca are hard to obtain, so more solution volumes are applied to obtain low concentrations of dissolved Ca. The solution volumes and concentrations of different main ions applied for solubility experiments are listed in Table 1. Only 1 g cement paste powders solute in 100–500 mL solutions, which guarantees enough ammonium nitrate in the equilibrium systems. The equilibrium concentrations s and c will be experimentally determined. Accordingly, the solid–liquid equilibrium curve of Ca in 6 M ammonium nitrate solution can be experimentally obtained. 2.2. Sample preparations The specimens were prepared with PI 52.5 Portland cement from Wuhan Huaxin factory, which were produced by the pure cement clinker mixed with 5 wt.% gypsum. The chemical composition of the cement is listed in Table 2. To rule out the influence of the unhydrated clinker, the specimen should be as mature as possible. Cement pastes with w/c ratios of 0.53 were used in this research, and the cement pastes were cured in a standard curing room (temperature 20 ± 3 °C, relative humidity over 95%) for 24 months. Before further solubility experiments, the powder specimens were prepared using the following procedures. First, the cement pastes were dried at 60 °C for ten days till constant weight. Second, to avoid the influence of carbonation, only the center part of the sample was taken and crushed. Third, the crushed powders were passed through a 0.075 mm mesh sieve. 2.3. Solubility experiments All solubility experiments were performed at room temperature (about 20 °C). According to the experimental design in Section 2.1, mixed solutions with the ion concentrations listed in Table 2 were prepared firstly. One gram solid cement paste powders were equilibrated in 100–500 mL of the different solutions with different concentrations (Table 1). The cement powder and solution systems were magnetically stirred in sealed beakers. After a certain period, the concentration of dissolved Ca was analyzed using EDTA method. Once the Ca concentration was stabilized, the solid powders were

Table 1 The solution volumes and concentrations of different main ions applied for solubility experiments on 1 g cement paste powder. Number

1

2

3

4

5

6

7

8

Volume (mL) CCa (mmol/L) CNH4 (mmol/L) CNO3 (mmol/L) Number Volume (mL) CCa (mmol/L) CNH4 (mmol/L) CNO3 (mmol/L) Number Volume (mL) CCa (mmol/L) CNH4 (mmol/L) CNO3 (mmol/L)

500 0 6000 6000 9 100 200 5600 6000 17 100 2300 1400 6000

330 0 6000 6000 10 100 300 5400 6000 18 100 2400 1200 6000

250 0 6000 6000 11 100 800 4400 6000 19 100 2500 1000 6000

170 0 6000 6000 12 100 1000 4000 6000 20 100 2600 800 6000

130 0 6000 6000 13 100 1500 3000 6000 21 100 2700 600 6000

100 0 6000 6000 14 100 2000 2000 6000 22 100 2800 400 6000

100 50 5900 6000 15 100 2100 1800 6000 23 100 2900 200 6000

100 150 5700 6000 16 100 2200 1600 6000 24 100 3000 0 6000

collected by centrifugating and vacuum suction filtrating. Then the Ca in solid was determined using X-Ray Fluorescence (XRF). Besides the elemental analysis, X-Ray Diffraction (XRD) was performed for phase analysis for several specimens with different leaching degrees. 2.4. Analysis dissolved Ca concentration using EDTA titration The equilibrium concentrations of dissolved Ca were determined using EDTA method. Firstly, dilute the test solution to be less than 100 mmol/L. Secondly put 5.0 mL test solution into a 250 mL beaker flask, and then add 1 mL Triethanolamine to minimize the interference of iron and manganese, and add 1–3 mL buffer solution and 3 drops of Eriochrome Black T. Thirdly, titrate the solution with the standard EDTA solution until blue or purple swirls begin to show. Finally, the end point is reached when all traces of red and purple have disappeared and the solution is pure blue in color. For each dissolved Ca concentration, three titration tests were performed and averaged with errors less than 3%. 2.5. Elemental analysis using XRF The elemental concentrations of the solid powders were analyzed using the ARL QUANT'X Energy-Dispersive X-Ray Fluorescence (EDXRF), which was coupled to a Si (Li) EDS detector. The dried powder specimen was placed in sample cups and irradiated for 30 s with 4 kV, 8 kV, 16 kV and 28 kV X-rays respectively. Quantitative elemental concentrations were calculated by comparison with the cement composition shown in Table 1. Besides the two main elementals Ca and silicon (Si), sulfur (S), aluminum (Al), manganese (Mg), and iron (Fe) elementals were also analyzed. Si does not leach in high Ca/Si ratio level, and even at Ca/Si ratio as low as 0.3, Si leaching can be ignored compared to Ca leaching [34]. So the concentrations of other solid elementals are expressed by their molar ratios with Si. 2.6. Phase analysis using XRD XRD powder scans of several samples collected from the solubility experiments were analyzed with Bruke X-ray diffractometer. The diffraction patterns were acquired with Cu Kα radiation operating at 40 kV and 30 mA. Scans were collected at 0.15 s per step and a step

NO3-:6000mmol/L Ca2+:0mmol/L NH4+:6000mmol/L

Constant 3000mmol/L 0mmol/L

Leaching front

Fig. 2. Illustration of the main ion concentration gradients in the pore solution of a partly leached cement paste.

Table 2 Chemical composition of the cement/wt.%. CaO

SiO2

Al2O3

Fe2O3

MgO

SO3

K2O

Na2O

others

LOI

Total

62.60

21.35

4.64

3.31

3.08

2.25

0.54

0.21

1.07

0.95

100.00

Others include TiO2, P2O5, BaO, ZnO, MnO, SrO, PbO, and Cl.

K. Wan et al. / Cement and Concrete Research 53 (2013) 44–50

size of 0.02° over the range from 5°to 80°2θ. A LynxEye array detector with high efficiency was applied for X-ray detection.

(a)

Counts

2.7. Measurements of degree of hydration using thermogravimetric (TG) analysis The powder samples prepared in Section 2.2 were taken for TG analysis directly. The NETZSCH STA 449 F3 thermal analyzer was used for TG analysis in this study. 0.3 gram powder sample was heated from 25 °C to 1000 °C with a 10 °C per minute heating rate under nitrogen gas protection. The non-evaporable water was calculated using the weight losses in the range of 105 °C–1000 °C. Then the degree of hydration can be calculated using Wn/Wn(∞), where Wn is the non-evaporable water of the specimen, and Wn(∞) is the non-evaporable water of fully hydrated cement pastes. Fully hydrated cement pastes typically contain about 23% of non-evaporable water [30]. Three TG analysis were performed and averaged for degree of hydration.

47

1500 1000 500 0 1500 1000 500 0 1500 1000 500 0 1500 1000 500 0 1500 1000 500 0

Cement clinker

Ca/Si=3.14 Cement paste

Ca/Si=1.85 Number 22

Ca/Si=1.17 Number 19

Ca/Si=0.32 Number 5

10

20

30

40

50

60

70

80

(b)

3. Results

900

3.1. Equilibrium process

600 300

2880 2820

Number 22

Ca dissolved [mmol/L]

2760 2700 1560

Number 13

1500

900 600 300

Counts

One key point to experimentally obtain the equilibrium curve is that the equilibrium state should be reached in the cement powder and ammonium nitrate solution systems. The concentration of dissolved Ca in the simulated pore solution is used as an indicator for equilibrium. The time evolutions of Ca concentration for three typical systems (numbers 6, 13, and 22) with different leaching degrees are shown in Fig. 3, from which the equilibrium process in 6 M ammonium nitrate solution can be clearly observed. The equilibrium states are reached within one or two days. Considering the powder specimens and the high ammonium nitrate concentration, this short equilibrium time is reasonable. To have more accuracy, both c and s were measured after 9 days of equilibrium. To further show the composition of the equilibrium product, the XRD results of the cement clinker, the cement paste without Ca leaching and three partly leached specimens with different Ca/Si ratios are shown in Fig. 4. Several information can be obtained. First, belite peaks between 30° and 35° 2θ are always observed in all specimens, which means that the cement paste was not fully hydrated yet. This is consistent with the experimental results of hydration degree, 0.93 ± 0.03. Second, very sharp peaks of portlandite are found in the cement paste without Ca leaching, and the portlandite

900 600 300 900 600 300 900 600 300 10

15

20

25

30

35

Fig. 4. XRD patterns of the cement clinker, the cement paste without Ca leaching, and three typical equilibrium solid products with different Ca/Si ratios, panel (b) is the enlarge figure of the dotted rectangular parts in panel (a).

peaks disappear in the partly leached specimen with Ca/Si ratios of 1.85, 1.17, and 0.32. Third, there is no CSH peaks visible in all the specimens, generally CSH cannot be seen or only very poorly be seen by XRD. Fourth, calcite peak near 30° 2θ is observed on two partly leached specimen with Ca/Si ratios of 1.85 and 1.17, which may arise from the carbonation of the dissolved Ca in solution. Fifth, small peaks of ettringite at 9°, 15° and 22° 2θ are observed on the partly leached specimen with Ca/Si ratios of 1.85, and small peaks of AFm (possibly monocarbonate) are observed in the partly leached specimen with Ca/Si ratios of 1.17 and 0.32.

3.2. Experimentally obtained equilibrium curves

1440 180

Number 6

120 60 0

0

2

4

6

8

10

Leaching time [days] Fig. 3. The time evolutions or equilibrium process of dissolved Ca concentrations for three typical systems.

Using the equilibrium concentration of dissolved Ca in the simulated pore solution c as the x-coordinate, and the equilibrium Ca in solid skeleton s (Ca/Si ratios are applied here) as the y-coordinate, the solid–liquid equilibrium curve of Ca in 6 M ammonium nitrate solution can be experimentally obtained. Besides the solid–liquid equilibrium curve of Ca [Fig. 5(a)], the relation between solid Mg–dissolved Ca [Fig. 5(b)], the solid S–dissolved Ca [Fig. 5(c)], the solid Al–dissolved Ca [Fig. 5(d)], and the solid Fe–dissolved Ca [Fig. 5(e)] are presented. It is found that S and Mg are also leached in three stages, but Al and Fe are not leachable.

K. Wan et al. / Cement and Concrete Research 53 (2013) 44–50

(a)

10000

2.0

Ca/Si molar ratio

Considering that the solid–liquid equilibrium curve of Ca in Fig. 5(a) keeps similar three-stage form to that in water (Fig. 1), same equations as Eq. (1) is anticipated. In Eq. (1), the equilibrium curve in water can be expressed by the concentration of C–S–H, CH, x1, x2 and ceq. Similarly, we try to use these parameters and Eq. (1) to fit the solid–liquid equilibrium curve of Ca in 6 M ammonium nitrate solution. The cement paste applied for this research has been cured for 24 months. The experimental determined degree of hydration is 0.93 ± 0.03, which is close to the calculated maximum hydration degree, 0.92, using α = 0.239 + 0.745 tanh[3.62(w/c − 0.095)] [22]. According to the initial composition data in Table 1 and hydrated degree of 0.92, the hydrated products of the cement paste can be calculated. The quantity of C–S–H and CH are calculated to be 5846 mmol/L and 3759 mmol/L respectively. Using CCSH as 5846 mmol/L, CCH as 3759 mmol/L, and taking x1 as 130 mmol/L, x2 as 2950 mmol/L, and ceq as 3000 mmol/L, the equilibrium curve (c and s) is calculated and shown in Fig. 6, which fits well with the experimental equilibrium curve (c and Ca/Si ratio). Not only the experimentally obtained equilibrium curve of Ca in ammonium nitrite solution is different from that in water (such as in [2,3,6,16,19]), but also is it different to any supposed equilibrium curve in ammonium nitrate solution [6–8,16,18]. Although the

8000 1.5 6000 1.0 4000 0.5

0.0

2000 0 0

500

1500

2000

2500

3000

Fig. 6. Comparison between the experimental equilibrium curve and calculated equilibrium curve of Ca using Eq. (1) (the applied parameters: CCSH as 5846 mmol/L, CCH as 3759 mmol/L, x1 as 130 mmol/L, x2 as 2950 mmol/L, and ceq as 3000 mmol/L).

equilibrium curve is still a three-stage form as that in water, the inflection points of the three stages are different with reference's values. At room temperature, the saturated concentration of Ca2+ in deionized water ceq changes from about 20 mmol/L in water to

(b) 3.0

0.30

2.5

0.25

2.0 1.5 1.0

0.0

0.20 0.15 0.10 0.05

0.5

0.00 0

500

1000

1500

2000

2500

3000

0

500

Ca dissolved [mmol/L]

1000

1500

2000

2500

3000

2500

3000

Ca dissolved [mmol/L]

(c)

(d)

0.30

0.30

0.25

0.25

Al/Si molar ratio

S/Si molar ratio

1000

Ca dissolved [mmol/L]

Mg/Si molar ratio

Ca/Si molar ratios

12000

2.5

3.3. Fitted equilibrium curve of Ca using material compositions

Ca in solid[mmol/L]

48

0.20 0.15 0.10 0.05

0.20 0.15 0.10 0.05

0.00 0

500

1000

1500

2000

2500

3000

0.00

0

500

Ca dissolved[mmol/L]

1000

1500

2000

Ca dissolved [mmol/L]

(e) 0.30

Fe/Si molar ratio

0.25 0.20 0.15 0.10 0.05 0.00

0

500

1000

1500

2000

2500

3000

Ca dissolved [mmol/L] Fig. 5. The obtained equilibrium curves in 6 M ammonium nitrate solution: (a) the solid–liquid equilibrium curve of Ca; (b) the solid Mg–liquid Ca equilibrium curve; (c) the solid S–liquid Ca equilibrium curve; (d) the solid Al–liquid Ca equilibrium curve; (d) the solid Fe–liquid Ca equilibrium curve.

K. Wan et al. / Cement and Concrete Research 53 (2013) 44–50

about 3000 mmol/L in 6 M ammonium nitrate solution. x1 is the concentration of dissolved Ca in pore solution when C–S–H dissolved quickly, which changes from 2 mmol/L in water to 130 mmol/L in 6 M ammonium nitrate solution. x2 is the concentration of dissolved Ca in pore solution when CH had completed dissolved and C–S–H began to dissolve, which changes from about 17–20 mmol/L in water to about 2950 mmol/L in 6 M ammonium nitrate solution, that is about ceq − 50. 4. Discussion on the solid–liquid equilibrium curves

Both the equilibrium curves of Ca in water and that in 6 M ammonium nitrite solution can be described by Eq. (1), which means that the dissolution and leaching mechanisms are similar in both conditions. According to Eq. (1), Figs. 5 and 6, the leaching process still can be divided into three stages as that in water. In the first stage (x2 b c b ceq), CH is quickly dissolved and leached. In the second slow and central stage (x1 b c b x2), the Ca in C–S–H is partially dissolved and leached, resulted Ca/Si ratio in C–S–H from about 1.3 to about 0.5. The Ca/Si ratios on the second stage is lower than that of the equilibrium curve in water, which implies deep dissolution of C–S–H in 6 M ammonium nitrite solution. In the third stage (0 b c b x1), when the dissolved Ca concentration in pore solution is less than x1 (about 130 mmol/L), the partially leached C–S–H is quickly and totally decalcified to be silica gel. Ca leaching includes the Ca dissolution process and the diffusion transport process. The dissolution process is a thermodynamic equilibrium process described by the solid–liquid equilibrium curve of Ca. On the one hand, the equilibrium curve under accelerated condition keeps same three-stage form as that in water, and the meaning of each stage is similar to that in water either. On the other hand, the greatly increased Ca concentration and high concentration gradient in pore solution will accelerate the diffusion transport process, which is believed to be the main acceleration mechanism. So 6 M ammonium nitrate solution is undoubtedly an ideal accelerated Ca leaching protocol. The experimentally obtained equilibrium curve, which describes the quantitative concentration distribution in pore solution, is crucial for modeling research of accelerated Ca leaching. Besides, from Fig. 5(a) and (b) it is found that both solid–liquid equilibrium curve of Ca and solid Mg-dissolved Ca equilibrium curve have similar form, which means Mg will be leached out with Ca simultaneously. 4.2. The dissolution processes of sulfoaluminates Hydrated Al, Fe, and S phases or named sulfoaluminates (mainly Aft and AFm phases) are one of the main phases besides CH and C–S–H in hydrated cement pastes. For the Ca leaching in water, it is thought that the Ca in sulfoaluminates is dissolved and leached at the second stage [7,21]. For the Ca leaching in ammonium nitrate solution, it is normally thought that the Ca in sulfoaluminates is hard to leach [7,33,34]. In a recent research, Puertas et al. conclude that sulfoaluminates are dissolvable in ammonium nitrate solution based on IR results [34], and they expressed the sulfoaluminate dissolution by the following chemical reaction [35]

3CaO  Al2 O3  3CaSO4  32H2 O þ NH4 NO3 →ðNH4 Þ2 SO4 þ CaðNO3 Þ2 þ Al2 O3  xH2 O þ ð32−xÞH2 O

be reasonably extended to other hydrated Al, Fe, and S phases using following chemical reactions 3CaO  ðAl; FeÞ2 O3  CaSO4  mH2 O þ NH4 NO3 →ðNH4 Þ2 SO4 þ CaðNO3 Þ2 þ Al2 O3  xH2 O þ Fe2 O3  yH2 O þ zH2 O soluble

ð3Þ

From the equilibrium curve of S, Al and Fe in Fig. 5(c), (d) and (e), it is found that solid S is decreasing greatly but Al and Fe only change slightly during the leaching process, which can be explained by the solubility of the chemical reaction products. Accordingly, Eq. (3) can

soluble

insoluble

3CaO  ðAl; FeÞ2 O3  3CaSO4  nH2 O þ NH4 NO3 soluble

insoluble

ð4Þ

insoluble

→ðNH4 Þ2 SO4 þ CaðNO3 Þ2 þ Al2 O3  xH2 O þ Fe2 O3  yH2 O þ zH2 O soluble

4.1. Leaching process of CH/C–S–H and the acceleration mechanism

49

ð5Þ

insoluble

Besides, the equilibrium curve of S in Fig. 5(c) has similar three-stage form to that of Ca, so the dissolution process of hydrated Al, Fe, and S phases should also be in three stages, and the chemical reactions of Eqs. (4) and (5) may be divided into three steps. What is more interesting is to find that over 80% S is dissolved in the first stage. 4.3. The universality of the equilibrium curves Only one case study was applied to obtain the equilibrium curve in 6 M ammonium nitrate solution, can the obtained equilibrium be applied to other cement-based materials? The equilibrium curve of Ca in water is obtained from the solubility data of several decades, which is not sensitive to cement compositions. Similarly, the equilibrium curve of Ca in ammonium nitrate solution should also be insensitive to cement compositions. Water cement ratios mainly influence on the hydration degrees and capillary porosities. The capillary porosity only influences the equilibrium time but not equilibrium product. Furthermore, powder specimens are applied to solubility experiments, so the influence of capillary porosity can be ignored. Because the cement paste applied in this research has a high degree of hydration, the influence of anhydrated clinker can be ruled out. So the obtained equilibrium curves in Fig. 5 or expressed by Eq. (1) can be directly applied for ordinary Portland cement. When different admixtures are applied, the quantity of CH and C–S–H changes accordingly, and then the form of the equilibrium curve will change according to Eq. (1). 5. Conclusions To clarify the accelerated leaching mechanism in 6 M ammonium nitrate solution, the equilibrium curve of Ca in 6 M ammonium nitrate solution is experimentally obtained in this research. The obtained equilibrium curve can be expressed using Eq. (1), where ceq is 3000 mmol/L in the ideal case, x2 is (ceq − 50) mmol/L, and x1 is about 130 mmol/L. The three-stage equilibrium curve keeps similar form to that in water, which implies the similar dissolving and leaching mechanism. The greatly increased Ca concentration and high concentration gradient in pore solution are believed to be the main acceleration mechanism. The obtained equilibrium curve can be used for modeling research of the accelerated Ca leaching in 6 M ammonium nitrate solution. Besides the solid–liquid equilibrium curve of Ca, the other equilibrium curves between other solid elements (including S, Mg, Al, Fe) and dissolved Ca are presented. It is found that S and Mg are also leached in three stages, but Al and Fe are not leachable. So sulfoaluminates and Mg are leachable in ammonium nitrate solution, and they are also leached in three stages. Acknowledgments The authors greatly appreciate reviewers' helpful comments in improving the quality of this paper. In addition, the financial support from the foundation of National Basic Research Program of China (No. 2009CB623203) and National Natural Science Foundation of China (No. 51008072) are acknowledged.

50

K. Wan et al. / Cement and Concrete Research 53 (2013) 44–50

References [1] U.R. Berner, Modelling the incongruent dissolution of hydrated cement minerals, Radiochim. Acta 44 (45) (1988) 387–393. [2] M. Buil, A model of the attack of pure water or undersaturated lime solutions on cement, ASTM STP 1123 (1992) 227–241. [3] U.R. Berner, Evolution of pore water chemistry during degradation of cement in a radioactive waste repository environment, Waste Manage. 12 (1992) 201–219. [4] C. Carde, R. Francois, J.M. Torrenti, Leaching of both calcium hydrate and C–S–H from paste: modeling the mechanical behavior, Cem. Concr. Res. 26 (1996) 1257–1268. [5] C. Carde, R. Francois, Modelling the loss strength and porosity increase due to the leaching of cement paste, Cem. Concr. Compos. 21 (1999) 181–188. [6] B. Gerard, C. Le Bellego, O. Bernard, Simplified modelling of calcium leaching of concrete in various environments, Mater. Struct. 35 (2002) 632–640. [7] C. Pichler, A. Saxer, R. Lackner, Differential-scheme based dissolution diffusion model for calcium leaching in cement-based materials accounting for mix design and binder composition, Cem. Concr. Res. 42 (2012) 686–699. [8] F.H. Heukamp, F.J. Ulm, J.T. Germaine, Mechanical properties of calcium-leached cement pastes Triaxial stress states and the influences of the pore pressures, Cem. Concr. Res. 31 (2001) 767–774. [9] F.J. Ulm, E. Lemarchand, H.F. Heukamp, Elements of chemomechanics of calcium leaching of cement-based materials at different scales, Eng. Fract. Mech. 70 (2003) 871–889. [10] V.H. Nguyen, B. Nedjar, H. Colina, J.M. Torrenti, A separation of scales homogenization analysis for the modelling of calcium leaching in concrete, Comput. Methods Appl. Mech. Eng. 195 (2006) 7196–7210. [11] J.S. Ryu, N. Otsuki, H. Minagawa, Long-term forecast of Ca leaching from mortar and associated degeneration, Cem. Concr. Res. 32 (2002) 1539–1544. [12] C. Alonso, M. Castellote, I. Llorente, C. Anderade, Ground water leaching resistance of high and ultra high performance concretes in relation to the testing convection regime, Cem. Concr. Res. 36 (2006) 1583–1594. [13] S. Kamali, B. Gerard, M. Moranville, Material and environmental parameter effects on the leaching of cement pastes: experiments and modeling, Cem. Concr. Res. 38 (2008) 575–585. [14] N. Burlion, D. Bernard, D. Chen, X-ray microtomography: application to microstructure analysis of a cementitious material during leaching process, Cem. Concr. Res. 36 (2006) 346–357. [15] G. Constantinides, F.J. Ulm, The effect of two types of C–S–H on the elasticity of cement-based materials: results from nanoindentation and micromechanical modeling, Cem. Concr. Res. 34 (2004) 67–80. [16] K.S. Wan, Y. Li, W. Sun, Experimental and modelling research of the accelerated calcium leaching of cement paste in ammonium nitrate solution, Constr. Build. Mater. 40 (2013) 832–846.

[17] J.J. Thomas, J.J. Chen, A.J. Allen, H.M. Jennings, Effects of decalcification on the microstructure and surface area of cement and tricalcium silicate pastes, Cem. Concr. Res. 34 (2004) 2297–2307. [18] C. Tognazzi, Cracking and Leaching of Cement-Based Materials, Experimentations. Ph.D. Thesis INSA Toulouse, 1998. (in French). [19] K. Nakarai, T. Ishida, K. Maekawa, Modeling of calcium leaching from cement hydrates coupled with micro-pore formation, J. Adv. Concr. Technol. 4 (2006) 395–407. [20] K. Haga, S. Sutou, M. Hironaga, S. Tanaka, S. Nagasaki, Effects of porosity on leaching of Ca from hardened ordinary Portland cement paste, Cem. Concr. Res. 35 (2005) 1764–1775. [21] M. Mainguy, O. Coussy, Propagation fronts during calcium leaching and chloride penetration, J. Eng. Mech. (ASCE) 126 (2000) 250–257. [22] S. Bejaoui, B. Bary, Modeling of the link between microstructure and effective diffusivity of cement pastes using a simplified composite model, Cem. Concr. Res. 37 (2007) 469–480. [23] J. Jain, N. Neithalath, Analysis of calcium leaching behavior of plain and modified cement pastes in pure water, Cem. Concr. Compos. 31 (2009) 176–185. [24] H. Saito, S. Nakane, Comparison between diffusion test and electrochemical test for leaching degradation of cement hydration products, ACI Mater. J. 96 (1999) 208–212. [25] H. Saito, A. Deguchi, Leaching tests on different mortars using accelerated electrochemical method, Cem. Concr. Res. 30 (2000) 1815–1825. [26] H.L. Chatelier, Experimental researches on the constitution of hydraulic mortars (translation by J.L. Mack), Ann. Mines 11 (1887) 345–465. [27] H.H. Steinour, The system CaO–SiO2–H2O and the hydration of calcium silicates, Chem. Rev. 40 (1947) 391–460. [28] S.A. Greenberg, T.N. Chang, Investigation of the colloidal hydrated calcium silicates. II. Solubility relationships in the calcium oxide–silica–water system at 25 °C, J. Phys. Chem. 69 (1965) 182–188. [29] H.M. Jennings, Aqueous solubility relationships for two types of calcium silicate hydrate, J. Am. Ceram. Soc. 69 (1986) 614–618. [30] H.F.W. Taylor, Cement Chemistry, 2nd ed. Thomas Telford, London, 1997. [31] J.J. Chen, J.J. Thomas, H.F.W. Taylor, H.M. Jennings, Solubility and structure of calcium silicate hydrate, Cem. Concr. Res. 34 (2004) 1499–1519. [32] D. Damidot, B. Lothenbach, D. Herfort, F.P. Glasser, Thermodynamics and cement science, Cem. Concr. Res. 41 (2011) 679–695. [33] C. Carde, G. Escadeillas, R. Franqois, Use of ammonium nitrate solution to simulate and accelerate the leaching of cement pastes due to deionized water, Mag. Concr. Res. 49 (1997) 295–301. [34] J.J. Chen, J.J. Thomas, H.M. Jennings, Decalcification shrinkage of cement paste, Cem. Concr. Res. 36 (2006) 801–809. [35] F. Puertas, S. Goni, M.S. Hernandez, C. Varga, A. Guerrero, Comparative study of accelerated decalcification process among C3S, grey and white cement pastes, Cem. Concr. Compos. 34 (2012) 384–391.