Effects of slag characteristics on sulfate durability of Portland cement-slag blended systems

Construction and Building Materials 229 (2019) 116882

Contents lists available at ScienceDirect

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

Effects of slag characteristics on sulfate durability of Portland cement-slag blended systems Farzaneh Nosouhian a, Mustafa Fincan b, Natallia Shanahan a, Yuri P. Stetsko a, Kyle A. Riding c, A. Zayed a,⇑ a

Department of Civil and Environmental Engineering, University of South Florida, 4202 E Fowler Ave, ENB 118, Tampa, FL 33620, United States Department of Mechanical Engineering, University of South Florida, 4202 E Fowler Ave, ENB 118, Tampa, FL 33620, United States c Department of Civil and Coastal Engineering, University of Florida, 1949 Stadium Road, P.O. Box 116580, Gainesville, FL 32611, United States b

h i g h l i g h t s  Sulfate durability of GGBFS generally decreased with increasing its Al2O3 content.  Sulfate durability of GGBFS improved with increasing MgO/Al2O3 ratio of slag.  Solid volume increase of hydration products is linearly related to MgO/Al2O3 ratio.  Solid volume increase appears to be related to the length of the induction period.  Increasing slag fineness appears to decrease slag sulfate resistance.

a r t i c l e

i n f o

Article history: Received 31 October 2018 Received in revised form 10 July 2019 Accepted 3 September 2019

Keywords: Ground granulated blast-furnace slag Sodium sulfate Sulfate resistance Thermodynamic modelling

a b s t r a c t Sulfate resistance of Portland cement-slag mixtures was compared at 30, 50 and 70% cement replacement levels for slags with variable Al2O3 and MgO content and variable fineness in combination with moderate- and high-C3A cements. Sulfate resistance generally decreased with increasing slag Al2O3 content, although decreasing MgO also increased deterioration, particularly at 30% replacement. Increase in Al2O3 correlated with increased formation of secondary ettringite in the surface of the bars. A linear relationship was observed between the solid volume increase on exposure to sulfate solution predicted by GEMS and the MgO/Al2O3 ratio of slag at all cement replacement levels. There was also a linear relationship between the solid volume increase and the length of the induction period for the mortar bars exposed to sulfate solution. Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction Sulfate attack can occur when concrete is exposed to sulfate concentrations above 0.1% and is manifested by expansion and/or loss of strength and cohesion [1]. Expansion is typically attributed to secondary ettringite formation from the reaction of monosulfoaluminate present in concrete with the sulfate ions from the environment and possibly gypsum formation from the reaction of CH with sulfates, although the expansive nature of secondary gypsum formation is still debated in the literature [2–11]. Loss of strength and cohesion is attributed to decalcification of C-S-H; secondary gypsum formation has been suggested to play a role as well [7,8,12–14]. Cement replacement by ground granulated blast furnace slag (GGBFS) is generally reported to improve concrete performance in sodium sulfate environments through reduction of ⇑ Corresponding author. E-mail address: [email protected] (A. Zayed). https://doi.org/10.1016/j.conbuildmat.2019.116882 0950-0618/Ó 2019 Elsevier Ltd. All rights reserved.

the C3A content of the cementitious system, consumption of CH during slag hydration, and decrease in permeability [15,16]. However, ASTM C989 [16] cautions that sulfate resistance of OPC-slag blends may be dependent on the Al2O3 content of the slag. While the standard comments on the effect of low alumina (11% Al2O3) and high alumina (18% Al2O3) slags on sulfate durability, there is no comment on the effect of the slags with Al2O3 content of 11–18%. For a given ordinary Portland cement (OPC), two parameters affecting sulfate resistance of OPC-slag blends have been identified in the literature: cement replacement level and Al2O3 content of slag. Hooton [17] demonstrated that expansion in sodium sulfate solution decreased with increasing cement replacement by GGBFS for a slag with Al2O3 content of 8.4%. Higgins [18] examined performance of two slags with Al2O3 contents of 11.8 and 12.5%. While rapid expansion was observed after 30 and 12 months exposure to sodium sulfate solution, respectively, at 60% cement

2

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882

replacement, no expansion was observed when slag content was increased to 70%. A relationship between expansion and Al2O3 content of slag was observed by Hooton and Emery [19], who reported that expansion increased with increasing Al2O3 content from 8.4 to 11.4% at a constant cement replacement level of 50%. A similar range of slag Al2O3 contents (7.4–12.3%) was studied by Whittaker [20] at 40% cement replacement. Increase in susceptibility to sulfate attack of OPC-slag blends with increase in Al2O3 content of slag from 11.2 to 16% was observed by Gollop and Taylor [21]. Ogawa et al. [22] also reported a decrease in sulfate resistance with increase in slag Al2O3 content from 10.7 to 15.2% at 60% cement replacement. Generally, the chemical reactions during sulfate attack on OPCslag blends are similar to those occurring in plain OPC mixtures, except low quantities of gypsum are formed in the OPC-slag blends [23]. Other notable differences are formation of hydrotalcite (Ht) as well as lower Ca/Si and higher Al/Si ratio of C-S-H in OPC-slag mixtures [24]. In addition, hydration of slags with higher Al2O3 contents can also result in formation of monosulfoaluminate after hardening [25]. Generally, formation of monosulfate is considered undesirable as monosulfate will convert to ettringite on exposure to sulfates. While some authors attribute poor sulfate resistance of high-Al2O3 slags to secondary ettringite formation [21,24], this remains to be a point of debate. Whittaker et al. [26] concluded that although more monosulfate formed in the slag-containing mixes prior to sulfate exposure, secondary ettringite formation was not the cause of their observed deterioration since it was lower than that in the plain OPC mixes [26]. Addition of gypsum, or other forms of calcium sulfate, and limestone have been proposed to improve sulfate durability of higher-Al2O3 content slags [20–22,26–28]. However, it is not clear at what Al2O3 content addition of these materials is necessary. Additionally, the effect of other slag characteristics on sulfate resistance has not been considered in detail. Whittaker et al. [26] previously suggested that increase in MgO content of slag may be beneficial in terms of sulfate resistance as it would result in increasing formation of hydrotalcite and more alumina binding; however, this has not been examined in detail. Since there is a notable variation in MgO content in slags worldwide (3–14%) [29], the effect of MgO in combination with Al2O3, perhaps in terms of a MgO/Al2O3 ratio, on sulfate durability needs to be explored. The objective of this study was to compare sulfate resistance of Portland cement-slag blended systems prepared with slags having a range of Al2O3 and MgO content values at several cement replacement levels. 2. Materials and methods 2.1. Materials Two cements (Cement A and Cement B) and four slags (S1-S4) were used in this study. Cements A and B were ordinary Portland cement (OPC) with similar fineness and equivalent alkali (Na2Oeq) content, and differed mainly in their C3A and C3S content. Slags were selected based on their similar Ca/Si ratios and variable Al2O3 content (8–16%), but also varied in their fineness and MgO content. The characterization of these cements and slags has been previously published in [30] and is reproduced in Tables 1 and 2 for the reader’s convenience.

2.2. Experimental methods The effect of slag characteristics on sulfate durability was evaluated by measuring length change of 25  25  285 mm mortar bars and compressive strength of 50 mm mortar cubes stored in 5% sodium sulfate solution. Mortar mixes were prepared at 0, 30, 50, and 70 wt% cement replacements with slag. Mortars were prepared in accordance with ASTM C109 [31] and ASTM C305 [32], with the exception of the water-to-cementitious materials (w/cm) ratio. Currently, ASTM C109 [31] specifies that, with the use of supplementary cementitious materials (SCMs), the amount of mixing water should be adjusted to maintain a constant flow of 110 ± 5. However, the w/cm is a major factor affecting compressive strength, permeability, and consequently sulfate resistance [12]. Since the objective of this study

was to compare compressive strength development of mortars prepared with different slags in lime and sulfate environments, the w/cm ratio was maintained constant at 0.485 in order to eliminate this parameter as a variable. Mortar bars were prepared and tested following ASTM C1012 [33], except the w/cm ratio was maintained at 0.485. Sulfate solution for both cubes and bars was changed at the intervals specified in ASTM C1012 [33]. Length change of mortar bars was recorded at 1, 2, 3, 4, 8, 13 and 15 weeks and also at 4, 6, 9, and 12 months after immersion in the sulfate solution. Additional length change measurements were taken when rapid expansion of the bars was observed, although the solution was not changed after these additional readings. In order to characterize phase content of OPC-slag mixes before sulfate exposure, pastes were prepared by hand mixing for 5 min and cured at the same conditions as the mortar bars, at 35 °C during the first 24 h and then at 23 °C until the age of immersion into sulfate solution. Phase composition of the paste samples and mortar bars after 1 year of sulfate exposure was studied using X-ray diffraction (XRD). The bars were separated into the surface that was largely deteriorated and the intact core. The paste and mortar bar samples were crushed and gently ground by hand with a mortar and pestle in order to minimize formation of additional x-ray amorphous content due to the grinding effects [34–36]. The material was then sieved through a 45 mm sieve to separate the sand in the case of mortar bars. The fraction passing 45 mm was then mixed with an internal standard in order to determine the amorphous/unidentified content of each sample [37,38]. Standard reference material (SRM) 676a obtained from the National Institute of Standards and Technology was used as an internal standard in this study. SRM 676a was mixed with the sample by hand in order to avoid increasing the amorphous content. No specific technique was used to stop the hydration and carbonation, as samples were loaded into the diffractometer immediately after preparation. XRD measurements were performed using a Phillips X’Pert PW3040 Pro diffractometer equipped with the X’Celerator Scientific detector and a CuKa X-ray source. Tension and current were set to 45 kV and 40 mA, respectively. Scans were performed in the range of 7–70° 2h, with a step size of 0.0167° 2h. Samples were then loaded into the sample holder using a back-loading technique in order to minimize preferred orientation, and placed onto a spinner stage that was rotating at 30 rpm in order to improve counting statistics [39]. Phase quantification was performed using the Rietveld refinement functionality of the Panalytical HighScore Plus 4.5 software. In addition, thermodynamic modeling was performed using the Gibbs free energy minimization software GEMS 3 [40–42]. Thermodynamic data was taken from the default Nagra-PSI database [43] and CEMDATA14 [44] for cementspecific compounds. GEMS models phase assemblage at equilibrium. Therefore, 100% and 70% hydration was assumed for ordinary Portland cement (OPC) and slag, respectively. This degree of slag hydration is widely used in the literature for GEMS modeling of OPC-slag blends [45,46] and can be expected in samples at the age of 1 year [47]. The 70% degree of slag hydration was also verified by subjecting some of the crushed and sieved core samples after 1 year of sulfate exposure to selective dissolution with ethylene diamine tetraacetic acid (EDTA) following the procedure initially described by Luke and Glasser [48].

3. Results and discussion 3.1. Length change measurements Based on the 8% C3A limit established by ASTM C150 [49] for moderate sulfate resistance, Cement A can be considered to have moderate sulfate resistance, while Cement B was expected to be susceptible to sulfate attack. The negative effect of C3A on cement sulfate resistance has been well-established in the literature [12,50,51]. Additionally, the C3S content of Cement B was somewhat higher compared to that of Cement A as was its C3A content. An increase in C3S has also been reported to have a negative effect on sulfate durability [3,4,6,52–55]. The higher sulfate resistance of Cement A was confirmed by length change measurements, which showed much faster deterioration for mortar bars prepared with Cement B (Fig. 1). Control B had a well-defined induction period of approximately 56 days, during which only small changes in length were recorded (Fig. 1b). After 56 days, the rate of expansion of Control B increased dramatically. Some deterioration and spalling at the corners as well as the first signs of warping were observed at 91 days. Increased warping and warping-induced cracking initiating at the bottom of the bars were observed at 105 days. At 120 days, expansion of Control B bars reached almost 1%, after which most of the bars broke and could no longer be measured. Although the induction period for Control A was not

3

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882 Table 1 Chemical oxide composition and physical properties of cements and slags. Analyte

Cements

SiO2 Al2O3 Fe2O3 CaO MgO Total SO3 S as sulfide (slag) SO3 as sulfate (slag) Na2O K2O Corrected L.O.I (950 °C) Total Na2Oeq Ca/Si Mean particle size (MPS) (mm) Blaine fineness (m2/kg)

Slags

A

B

S1

S2

S3

S4

21.20 5.15 3.61 63.91 0.70 2.59 – – 0.14 0.31 1.66 99.89 0.35 – 12.26 485

20.1 5.60 2.00 64.4 0.90 3.60 – – 0.08 0.47 1.80 99.56 0.39 – 13.81 474

38.59 8.09 0.51 38.11 10.83 2.21 0.89 0.02 0.30 0.38 1.15 99.89 0.55 0.99 9.16 642

35.67 10.82 0.54 41.93 7.90 1.91 0.68 0.22 0.20 0.37 0.53 99.85 0.44 1.18 8.36 680

35.44 14.25 0.45 41.06 5.25 1.99 0.67 0.31 0.20 0.30 1.06 99.84 0.4 1.16 11.15 574

32.86 16.29 0.36 37.98 8.88 2.61 0.952 0.23 0.37 0.44 0.4 100.41 0.66 1.16 11.80 466

Table 2 Mineralogical composition of cements and slags determined by Rietveld analysis. Phase

C3S C2S C3A Ferrite Gypsum Hemihydrate Anhydrite Calcite Portlandite Syngenite Quartz Aphthitalite Amorphous/unidentified

Cements

Phase

A

B

48.1 23.1 5.5 9.9 2.6 1.5 0.0 1.2 – 0.7 0.1 – 7.2

54.0 17.3 8.4 5.6 4.3 1.4 0.1 0.3 0.2 0.4 0.1 0.2 7.8

Slags

Calcite Melilite Merwinite Quartz Gypsum Amorphous/unidentified

as well-defined, it was approximated to be 180 days (Fig. 1a). The subsequent rate of expansion for this mix was significantly lower compared to Control B. For Control A, only minor spalling at the edges and corners was observed at 1 year. Significant improvement in sulfate durability for both cements was achieved with the lower-Al2O3 slag addition. Bars prepared with 30% of S1 slag did not show any expansion during the first year, while those prepared with 30% S2 started exhibiting slight expansion after 300 days with Cement A and after 270 days with Cement B (Fig. 1). Addition of S3 and S4 with higher Al2O3 resulted in slight improvement of sulfate durability for the high-C3A Cement B, although with Cement A durability was decreased. 30S4-A mortar had the shortest induction period of 120 days. This was followed by rapid expansion from 0.05% to 0.17% at 150 days, after which the bars broke. Although the induction period of the 30S3-A mixture was longer than that of Control A, its subsequent expansion was much more rapid and reached the value of 0.36% at 330 days. After this, 30S3-A bars broke and could not be measured. Control A bars reached the 0.36% expansion at 360 days; however, all the bars remained intact. With Cement B, addition of slags S4 and S3 extended the induction period from 56 days to 91 days. 30S4-B bars broke after 105 days, when their expansion was measured at only 0.11%. Expansion of 30S3-B bars, on the other hand, was significantly higher, 0.60% at 150 days. Increasing slag content to 50% for the most part did not result in significant improvement in durability compared to 30%. Mixes containing S1 again did not show any expansion during the first year.

S1

S2

S3

S4

0.7 0.5 – – – 98.9

0.3 0.2 0.7 – – 98.8

0.6 0.5 – 1.5 0.4 97.0

0.3 0.7 – – – 99.0

Performance of 50S2-A was similar to that of 30S2-A, although 50S2-B bars had higher expansion compared to 30S2-B and broke after the age of 300 days (Fig. 2). Addition of S3 at 50% significantly reduced expansion in combination with Cement A, while with Cement B there was no improvement in performance when the slag content was increased from 30 to 50%. Both 30S3-B and 50S3-B mixes had the same length of the induction period, although 50S3-B bars broke before 120 days (0.10% length change) instead of expanding. Increasing S4 content from 30 to 50% did not improve durability with Cement A as both the length of the induction period and the age at breaking was unchanged, and 50S4-A bars broke at lower expansion values (0.09%) compared to 30S4A. With Cement B, increase in S4 content increased both the induction period and the age at breaking. Expansion prior to breaking, however, was unaffected. At 70% cement replacement, no breaks in the mortar bars were observed, and all bars appeared to be still in the induction period, with the exception of 70S2-A and 70S2-B (Fig. 3). The end of the induction period for these mixes was recorded at 300 and 270 days, respectively, and their expansion at 1 year was 0.16 and 0.15%. Deterioration in the S2-A mortars was observed in the form of spalling and was mostly confined to the ends of the bars (Fig. 4). Since slag S2 had the highest fineness compared to the other slags, increasing the content of high-fineness slag likely further reduced porosity, which would have reduced sulfate ion ingress leading to increased spalling of the surface layer. It appears that slag fineness is of importance for sulfate durability

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882

1

a)

0.9

Control-A 30S1-A 30S2-A 30S3-A 30S4-A

0.8 Expansion (%)

0.7 0.6 0.5 0.4

1

0.7 0.6 0.5 0.4

0.3

0.3

0.2

0.2

0.1

0.1 0 0

60

120

180 240 Time (days)

300

360

0

60

120

180 240 Time (days)

300

360

1

1 Control-B 30S1-B 30S2-B 30S3-B 30S4-B

0.8 0.7 0.6 0.5 0.4

0.8 0.7 0.6 0.5 0.4

0.3

0.3

0.2

0.2

0.1

0.1

0

Control-B 70S1-B 70S2-B 70S3-B 70S4-B

b)

0.9

Expansion (%)

b)

0.9

Expansion (%)

Control-A 70S1-A 70S2-A 70S3-A 70S4-A

0.8

0

0 0

60

120

180 240 Time (days)

300

360

Fig. 1. Length change of mortar bars prepared with 30% replacement by slag of a) Cement A and b) Cement B stored in 5% sodium sulfate solution. X indicates the last measurement, after which the set could not be measured.

1

a)

0.9

0

60

120

180 240 Time (days)

300

360

Fig. 3. Length change of mortar bars prepared with 70% replacement by slag of a) Cement A and b) Cement B stored in 5% sodium sulfate solution. X indicates the last measurement, after which the set could not be measured.

Control-A 50S1-A 50S2-A 50S3-A 50S4-A

0.8 0.7 Expansion (%)

a)

0.9

Expansion (%)

4

0.6 0.5 0.4 0.3

30S2-A

0.2

50S2-A

0.1

70S2-A

0 0 1

60

120

180 240 Time (days)

300

b)

0.9

Control-B 50S1-B 50S2-B 50S3-B 50S4-B

0.8 0.7 Expansion (%)

360

0.6 0.5 0.4 0.3 0.2 0.1 0 0

60

120

180 240 Time (days)

300

360

Fig. 2. Length change of mortar bars prepared with 50% replacement by slag of a) Cement A and b) Cement B stored in 5% sodium sulfate solution. X indicates the last measurement, after which the set could not be measured.

Fig. 4. S2-A bars at 1 year of exposure to 5% sodium sulfate solution.

of the OPC-slag mixes. Locher [56] reported a decrease in sulfate resistance with increasing slag fineness at cement replacement levels below 65%, while Wee et al. [57] reported ‘‘no consistent relationship” between slag fineness and sulfate durability. In general, the progress of deterioration for the OPC-slag mixes differed from that of the control mortars. Mortars containing slag broke at much lower expansion compared to the control mixtures, and their visual deterioration was characterized by spalling rather than warping. This is in line with the observations by Yu et al. [2,24] who also reported that failure in mortar bars with slag exposed to Na2SO4 solution occurred by cracking rather than expansion. They proposed that deterioration of slag mixtures in Na2SO4 environments was due to progressive spalling of the surface layer rather than expansion and cracking as with OPC mortars. The authors attributed this to pore refinement with slag addition since sulfate ions in the slag samples were concentrated within

5

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882

80.0 70.0

Compressive Strength (MPa)

1 mm from the surface. Whittaker et al. [26] also reported that the depth of sulfate ion penetration was lower for slag-containing mortars. The highest concentration was observed at 0.5–1 mm below the surface followed by a sharp drop in sulfate concentration in the slag-containing samples, while a concentration gradient existed in the plain OPC mortar. This explains why high expansion was observed in the control samples while slag-OPC mortars exhibited surface spalling and broke at low expansion values.

60.0

Control A 70S1-A 70S2-A 70S3-A 70S4-A

50.0 40.0 30.0 20.0 10.0

3.2. Compressive strength measurements

0.0

Although cement replacement with 30% slag improved compressive strengths by 28 days, all the mixes showed a drop in compressive strength when the strength at 1 year was compared to that at 90 days (Fig. 5). For the slag mixes prepared with Cement A, the largest drop in strength was observed for 30S3-A mix (16.7 MPa). Strength drop for 30S1-A and 30S2-A was very similar (10 and 10.4 MPa respectively), closely followed by 30S4-A (14 MPa). Unlike the length change measurements, compressive strength results showed a clear improvement in sulfate durability when slag content was increased from 30 to 50%. At 50% cement replacement, the strength drop was notably lower, and compressive strength of the 50S2-A, 50S3-A and 50S4-A was fairly similar at 1 year (Fig. 6). 50S1-A had the highest compressive strength of 59.4 MPa compared to approximately 53.6–56.7 MPa for the rest of the 50% slag mortars. The drop in strength was further reduced at 70% cement replacement with slag (Fig. 7). Again, slag S1 with the lowest Al2O3 content showed the best performance.

70.0

Compressive Strength (MPa)

60.0 50.0

Control A 30S1-A 30S2-A 30S3-A 30S4-A

40.0 30.0 20.0 10.0

0.0 7

28

91 Time (days)

180

Compressive Strength (MPa)

70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0

28

91 Time (days)

91 Time (days)

180

360

Fig. 7. Compressive strength of mortar cubes prepared with 70% slag in combination with Cement A stored in 5% Na2SO4 solution.

Based on compressive strength results, the influence of slag characteristics appears to be most notable at 30% cement replacement. 30S4-A with highest alumina content showed the highest strength at 28 days among OPC-slag mixes while it showed the least rate of strength development between 28 and 91 days. After that, although all 30% mixes experienced strength drop, strength of 30S4-A and 30S3-A mixes with the highest alumina content (and formed ettringite) decreased more at one year of sulfate exposure. While an increase in the slag Al2O3 content appears to have increased the strength drop slightly, (30S4-A compared to 30S1A and 30S2-A), the MgO content of slag also appears to be of importance. S3 slag mean particle size (MPS) was very similar to that of S4 (11.15 and 11.80 lm respectively), the C/S ratios of these two slags were the same, and S3 Al2O3 content was lower compared to S4. However, S3 had the lowest MgO content of all the slags at approximately 5%, while for the rest of the slags it varied between 8 and 11%. Several studies have shown that alumina bound in hydrotalcite is not reactive and cannot participate in secondary ettringite formation [25,58]. Lower hydrotalcite formation would result in more alumina available for secondary ettringite formation, which can explain the poor performance of slag S3. Even at 50% replacement, S4 and S3 showed the lowest strength and highest ettringite at one year of exposure. The effect of fineness is noticeable at 70% cement replacement, as 70S2-A mortar had the highest compressive strengths by 180 days. 3.3. X-ray diffraction measurements

Control A 50S1-A 50S2-A 50S3-A 50S4-A

7

28

360

Fig. 5. Compressive strength of mortar cubes prepared with 30% slag in combination with Cement A stored in 5% Na2SO4 solution.

80.0

7

180

360

Fig. 6. Compressive strength of mortar cubes prepared with 50% slag in combination with Cement A stored in 5% Na2SO4 solution.

3.3.1. Phases analysis prior to sulfate immersion The starting phase assemblage immediately prior to sulfate immersion was determined using match-cured paste samples (Table 3). Cement replacement with slag significantly reduced the amount of CH at sulfate exposure regardless of slag composition. However, 50S1-A paste had the highest CH content, while CH was lowest for 50S3-A. These results are consistent with the heat evolution trends observed in [30]. The higher reactivity of the S3 slag is attributed to its lower MgO and higher Al2O3 content. Low MgO content appears to increase the early-age slag reactivity as previously reported by [30,59], which explains the lower CH content of the 50S3-A paste. The similar CH contents of 50S2-A and 50S4-A points to their similar reactivity at early ages, which is explained by low Al2O3 content but high fineness of S2 and high Al2O3 content but lower fineness of S4 slag. Formation of ettringite with slag addition was also reduced due to the reduction in cement content. However, a decrease in the amount of ettringite was observed with increasing slag Al2O3 content. Monosulfoaluminate was observed only with the higher Al2O3

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882

Table 3 Phase analysis of 50% OPC-slag pastes at the age of sulfate exposure*. Phase amount (wt%) Alite Belite Aluminate Ferrite Portlandite Ettringite Monosulfates Hemicarbonate Hydrotalcite Amorphous

Control A

50S1-A

50S2-A

50S3-A

50S4-A

8.5 10.1 1.2 2.1 8.2 4.2 0.0 0.0 0.0 65.5

1.1 5.0 0.0 0.4 3.9 2.1 0.0 1.2 0.0 86.2

1.0 4.6 0.0 0.3 3.3 1.0 0.0 0.5 0.1 89.2

0.6 5.3 0.0 0.5 3.0 0.8 0.5 0.9 0.0 88.1

0.6 4.2 0.0 0.0 3.4 0.4 1.1 0.8 0.1 89.4

* Control A was analyzed at the age of 1 day, while all the slag pastes were analyzed at 3 days to match sulfate exposure age.

content slags, S3 and S4. Formation of monosulfoaluminate OPCslag samples has been previously reported in the literature and is attributed to the release of alumina during slag hydration [23,24,26] and indicates potentially lower durability of these mixtures. Formation of hemicarboaluminate was also observed in all the OPC-slag pastes. The increase in hemicarboaluminate with increasing Al2O3 content of slag reported by Whittaker et al. [60] was not observed in this study, possibly due to the difference in age and curing temperatures of the pastes. Hemicarboaluminate has also been reported to convert to ettringite on sulfate exposure [61]. Virtually no hydrotalcite was observed in the OPC-slag samples at this age.

3.3.2. Phases analysis after 1 year of sulfate exposure Phase quantification results of the core and surface samples of the bars prepared with Cement A and 50% slag at 1 year are presented in Table 4. Data for Cement A mixes prepared with 30% and 70% slag are included in the Appendix (Tables A.1 and A.2). Comparison of the core and surface samples for Control A showed higher amount of CH in the core, while gypsum and ettringite were higher in the surface sample, which is expected during sulfate attack [8,51,62]. Cement replacement with slag significantly decreased the amount of CH and ettringite in the core samples both due to the reduction of cement content and the reaction of slag with CH. However, the amount of monosulfate increased and was highest for 50S3-A, which corresponded to the lowest amount of hydrotalcite. In the surface samples, no monosulfate was present, and the amount of ettringite was significantly higher than in the core. For Control A, a decrease in the CH amount in the surface corresponded to an increase in gypsum content. For the slag samples, virtually no CH was detected in the surface of the samples, and the amount of gypsum was also low, especially at higher Al2O3 contents of slag. This can be due to ettringite formation and/or leaching. Lower amounts of hydrotalcite at the surface

compared to the core of the samples also suggests surface leaching. In addition, ettringite amount also increased with increasing slag Al2O3 content. Although Whittaker et al. [26] reported that less ettringite formed in the slag samples at 1 year compared to the control, the alumina content of the slags used in that study were only 7 and 12%. The results for S1 and S2 are in agreement with these findings; however, it is clear that ettringite amounts comparable to that of the control can be found in slag mixes with high slag Al2O3 content as is the case with 50S4-A (Table 4). Since the amount of ettringite in the core of the bars, on the other hand, was significantly lower in the slag mixtures compared to the control, it appears that increased formation of secondary ettringite in the surface may have caused expansion of the surface layer, which led to spalling of the surface and cracking of the bars. No hemicarboaluminate was observed in any of the samples. Higher hemicarboaluminate content of the 50S1-A mixture prior to sulfate immersion did not appear to have a negative effect on its sulfate durability. Since deterioration of the slag-containing mortar bars was generally confined to the surface, ettringite amount was plotted as a function of the slag Al2O3 (Fig. 8). Good coefficients of determination were obtained at 30 and 50% cement replacement, which generally corresponds to the increased deterioration of the high-Al2O3 slag mixes. No deterioration of the mortars containing 70% slag was observed at 1 year due to lower levels of formed ettringite. This explains the low R2 value at this replacement level. 3.4. Thermodynamic modeling Thermodynamic modeling was also used to evaluate the effects of sulfate attack on Cement A mixes (Fig. 9). The left side of the plots represents the phases at the core of the bars, while the right

30 30% slag 25

R² = 0.9617

50% slag 70% slag

Ettringite (%)

6

20 R² = 0.8251

15

10 R² = 0.3329 5

0 0

5

10 Al 2 O3 content of slag (%)

15

20

Fig. 8. Ettringite content in the surface samples of the slag-containing mortars after 1 year of exposure to 5% Na2SO4 solution.

Table 4 Phase quantification of the core and surface samples for 50% slag mortar bars stored in 5% Na2SO4 solution for 1 year. Sample ID Control A

50S1-A

50S2-A

50S3-A

50S4-A

Phase amount (wt %)

Core

Surface

Core

Surface

Core

Surface

Core

Surface

Core

Surface

Belite Calcite CH Gypsum Ettringite Monosulfate Hydrotalcite Amorphous

1.5 1.3 17.6 1.8 14.6 0.2 0.0 63.0

1.0 9.3 4.9 8.8 21.4 0.0 0.0 54.5

1.3 1.1 5.9 1.3 4.0 1.3 1.7 83.5

0.5 15.9 0.2 4.1 9.5 0.0 0.0 70.0

1.2 1.0 3.8 2.2 5.7 1.7 1.3 82.6

0.0 17.5 0.0 3.4 11.3 0.0 0.6 67.1

1.4 0.6 3.7 1.1 4.6 2.7 0.4 85.4

0.0 15.5 0.0 2.6 13.2 0.0 0.3 68.4

1.1 0.6 3.1 1.2 8.0 1.8 1.7 82.5

0.0 11.9 0.0 0.6 20.0 0.0 0.8 66.7

7

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882

Gypsum

Monocarbonate Hydrogarnet

Ferrihydrite-mc

Calcite

Hydrotalcite Ettringite

Portlandite

CSH

a) Control A Monocarbonate

Ferrihydrite-mc

Hydrogarnet Calcite

Portlandite

Hydrotalcite

Portlandite

Hemicarbonate Monocarbonate

Calcite

Ferrihydrite-mc Hydrotalcite

Hydrogarnet Ettringite

Ettringite Monosulfate CSH

Slag

Slag

b) 50S1-A Hemicarbonate

Hydrotalcite

Monocarbonate Stratl

CSH

c) 50S2-A Ferrihydrite-mc

Monocarbonate Hydrogarnet

Calcite

Hydrogarnet

Hydrotalcite

Ferrihydrite-mc

Calcite

Stratl Ettringite Ettringite Monosulfate

Monosulfate Slag

CSH

d) 50S3-A

Slag

CSH

e) 50S4-A

Fig. 9. Predicted phase assemblage for a) Control A, b) 50S1-A, c) 50S2-A, d) 50S3-A and e) 50S4-A mixtures immersed in 5% Na2SO4 solution.

side of the plots represents the surface of the bars in contact with the sulfate solution. First, phase composition predicted by thermodynamic modeling at exposure to very low volumes of sodium sulfate solution, simulating the conditions at the core of the mortar bars, was compared for all the OPC-slag mixtures. For Control A, the phases predicted at the core were CSH, CH, ettringite, monocarboaluminate (Mc), hydrogarnet (Hg) and hydrotalcite

(Ht). Generally, addition of slag increased the predicted volumes of C-S-H and Ht, while the volumes of ettringite, CH and Hg were reduced. This is in line with the GEMS modeling results recently reported by Fernández et al. [63] and consistent with the XRD results for the core samples. Among the slag mixes, phase assemblage varied with the level of cement replacement as well as with the chemical composition of anhydrous slag. At 30% replacement,

8

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882

although CH was predicted for all the slag mixes, its amount decreased with increasing Al2O3 content of the slag. Only ettringite was predicted for 30S1-A and 30S2-A, while both monosulfate and ettringite were present in 30S3-A and 30S4-A. Monosulfate conversion to secondary ettringite is known to cause deterioration in sulfate environments, so the results of thermodynamic modeling are in agreement with the observed deterioration of the 30S3-A and 30S4-A bars in sodium sulfate solution. Hemicarboaluminate (Hc) and Mc have also been reported to convert to ettringite on sulfate exposure [61], and the volume of these phases as well as the volume of Hc specifically increased with increasing Al2O3 content of slag. Increase in Hc with increasing Al2O3 content of slag was previously reported by Whittaker et al. [60]. At 50% replacement, CH and ettringite were only predicted for 50S1-A and 50S2-A mixes. Monosulfate was predicted for all the slag mixes except for 50S1-A, and its volume was largest in 50S3-A, even though more Al was present in 50S4-A. This corresponded to the lowest volume of Ht due to the low MgO content of S3 slag. Stratlingite was predicted for 50S3-A and 50S4-A, which indicates that these mixes had more Al than could be incorporated into C-S-H [64]. It should be noted that presence of stratlingite was not detected by XRD, possibly due to its low amounts. At 70% replacement, stratlingite was predicted for all the slag mixes, and its volume generally increased with the slag Al2O3 content. Ettringite was predicted only for 70S1-A, while the rest of the slag mixes contained monosulfate. Based on the predicted phase assemblage, S1-A mixes are expected to have good durability regardless of the cement replacement level, which is in line with expansion and compressive strength measurements. With increase in the Na2SO4 solution volume, decomposition of CH and formation of gypsum was predicted for Control A as well as decomposition of Mc. These phases are generally consistent with the XRD results at 1 year. A notable increase in the volume of ettringite was predicted as well. The maximum total solid volume was calculated to be 88 cm3/100 g of binder, which is a 28 cm3/100 g increase compared to the initial solid volume prior to sulfate exposure and is also in excess of the total initial volume, which includes the pore volume. In addition, this modeling approach is able to predict the effects of leaching [65], which can be observed as a decrease in the volumes of C-S-H, ettringite and gypsum with increasing Na2SO4 solution volume. While thermodynamic modeling was carried out for all the mixes, only the results for 50% cement replacement are presented

here as similar trends were observed at other replacement levels. Cement replacement with slag S1 resulted in an increase in C-SH volume and a decrease in ettringite compared to the Control A mixture (Fig. 9a). Addition of slag S1 also significantly reduced the overall predicted solid volume increase, which is consistent with the observed length changes (Fig. 2). While no monosulfate was predicted with S1 addition, this phase was predicted for the rest of the slag mixtures at low sulfate solution amounts (Fig. 9). It is interesting to note that for 50S3-A and 50S4-A mortars, GEMS modeling predicted a decrease in the solid volume of C-S-H due to leaching at a solution to binder mass ratio of 1, while a later decrease in C-S-H solid volume was predicted for other mixtures. Constantinides and Ulm [66] reported an increase in C-S-H porosity due to decalcification, which substantiates the predicted decrease in C-S-H solid volume. This decrease in C-S-H solid volume for 50S3-A and 50S4-A coincides with a notable increase in predicted volume of ettringite. Gollop and Taylor [23] stated that in slag-OPC mixes some calcium ions necessary for formation of ettringite are obtained through decalcification of C-S-H. The higher overall alumina content of 50S3-A and 50S4-A mixes may, therefore, be responsible for this decrease in C-S-H solid volume. Additionally, Whittaker et al. [26] reported that decalcification of C-S-H in OPC-slag samples occurred at earlier ages compared to the plain cement paste, which is consistent with the results from thermodynamic modeling. Volume changes for all the mixes prepared with Cement A are presented in Table 5. Volume increase was calculated as the difference between the maximum and the initial solid volumes of the system. Comparing the maximum solid volumes for all the mixtures to their respective initial total volumes, which include the volume of pores, it can be noted that the maximum solid volumes for Control A, 30S3-A, 30S4-A and 50S4-A mixtures exceeded their initial total volumes. It is not surprising, therefore, that all the slag mixtures broke before the age of 1 year, and an expansion of 0.36% was recorded for Control A. At all replacement levels, the maximum solid volume and the volume increase became larger with increasing alumina content of slag, except for S3. At 30 and 70% replacement, the volume increase for S3 mixes was equal or larger than for S4 mixes. This is likely due to the low amount of MgO in slag S3. The solid volume increase on exposure to sulfate solution is generated by the increase in ettringite volume (Fig. 9). At higher MgO contents (as in S1, S2, and S4 slags), some of the Al2O3 reacts with MgO to form

Table 5 Predicted volume changes for mixtures immersed in 5% Na2SO4 solution. Mix ID Control A 30S1-A 30S2-A 30S3-A 30S4-A 50S1-A 50S2-A 50S3-A 50S4-A 70S1-A 70S2-A 70S3-A 70S4-A 50S1-B 50S2-B 50S3-B 50S4-B

Initial total volume (cm3/100 g binder)

Initial solid volume (cm3/100 g binder)

Max solid volume (cm3/100 g binder)

Solid volume increase (cm3/100 g binder)

80 81 81 81 81 82 82 82 82 82 82 82 82 82 82 82 82

60 59 59 60 62 58 59 60 62 56 57 58 58 60 60 60 62

88 77 81 85 83 71 76 81 84 63 73 81 81 70 76 82 82

28 18 22 25 21 13 17 21 22 7 16 23 23 11 16 22 20

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882

mixes prepared with Cement B. For Cement A-slag mixes, the coefficient of determination was notably lower. Comparing the expansion behavior of the 50% slag mixes prepared with Cements A and B, the induction period in the Cement B mixes was more clearly defined, while in the Cement A mixes the expansion was more gradual, except for 50S4-A.

Volume increase (cm3 /100 g binder)

30 25

R² = 0.7117 20 R² = 0.9014

15

10

R² = 0.9713

30% slag

4. Conclusions

50% slag

5

70% slag

0 0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

MgO/Al2 O3 ratio Fig. 10. Predicted volume increase for Cement A mixes as a function of the MgO/ Al2O3 content of slag.

Volume increase (cm3 /100 g binder)

25 20 R² = 0.7033 R² = 0.9698

15

10 Slag-Cement A

5

Slag-Cement B 0 0

50

100

150 200 Induction period (days)

250

300

9

350

Fig. 11. Relationship between the predicted volume increase and the induction period of the mortar bars (50% replacement mixes) exposed to 5% Na2SO4 solution.

hydrotalcite, making the alumina bound in hydrotalcite unavailable for ettringite formation. Since MgO content of S3 was notably lower, more alumina was available to form ettringite on exposure to sulfates, which explains the higher maximum solid volumes of the S3 mixtures. In order to explore this relationship, predicted solid volume increase was plotted against the MgO/Al2O3 ratio of the slags (Fig. 10). High R2 values were observed at all cement replacement levels. Kunther et al. [46] had previously attempted to relate the increase in the solid volume predicted by GEMS for OPC and OPC +SCM mortars exposed to a variety of sulfate solutions. The authors reported that there was no relationship between the predicted solid volume increase and the measured expansion for OPC in different sulfate solutions. The authors also compared the predicted volume change and expansion for all the mixes exposed to sodium sulfate, but again there was no clear relationship. In the current study, it was not possible to establish a relationship between the predicted volume increase and expansion either. Deterioration of slag-containing mortars was mainly manifested through spalling followed by cracking at low expansion values. It appears that expansion measurements alone may not be sufficient to evaluate deterioration of OPC-slag blends. Therefore, a different metric, besides expansion values, is needed to probe the relationship between the predicted volume increase and sulfate durability. In order to explore the use of the length of the induction period as such a metric, predicted increase in the solid volume was plotted against the induction period for the 50% slag mixes (Fig. 11). A linear relationship was observed between these two parameters for

It was observed in this study that addition of slag changed the failure mode of the mortar bars exposed to sodium sulfate solution compared to the plain OPC mixes. Instead of expansion observed for the control mixtures, deterioration of the mortar bars containing slag was confined to the surface and manifested in the form of spalling. Generally, addition of slag, regardless of its characteristics, improved sulfate resistance of the high-C3A Cement B by extending the induction period. However, the extent of the improvement in durability was related to the alumina content of slag and appeared to decrease with increasing slag Al2O3 content, as expected. For Cement A with the lower C3A content, improvement in sulfate resistance was only observed with low-Al2O3 slags S1 and S2, while addition of higher Al2O3 slags S3 and S4 had a negative effect on durability at 30% and 30–50%, respectively. At 70% cement replacement, the highest expansion among the slag mixes was observed for S2 slag, which had the highest fineness. It appears that slag fineness may be of importance for sulfate durability of the OPC-slag mixes, and more research is needed in this area. XRD analysis showed that amount of ettringite formed in the surface of the mortar bars after 1 year of sulfate exposure increased with increasing slag Al2O3 content, with 50S4-A (16% Al2O3) having comparable amounts to that of Control A mixture. The amount of ettringite in the core, on the other hand, was notably lower in the slag mortars. Therefore, it appears that increased formation of secondary ettringite in the surface is responsible for the spalling of the surface and failure of the slag mortars. A linear relationship was observed between the solid volume increase on exposure to sulfate solution predicted by GEMS and the MgO/Al2O3 ratio of slag at all cement replacement levels. Thermodynamic calculations also predicted that solid volume for a number of mixes (Control A, 30S3-A, 30S4-A and 50S4-A) will exceed the initial total volume. This corresponded to mortar bar failure observed within 1 year of sulfate exposure. Additionally, since cracking rather than continued expansion was observed for slag mixtures, it was proposed to use the induction period as a possible metric to compare sulfate resistance of mixtures containing slag. A linear relationship was observed between the induction period and solid volume increase predicted by GEMS, particularly for Cement B. Declaration of Competing Interest There is no conflict of interest to report. Acknowledgements The authors would like to thank the Florida Department of Transportation and the Federal Highway Administration for providing partial funding for this work under contract number BDV25-977-28. The opinions, findings, and conclusions expressed in this publication are those of the authors and not necessarily those of the Florida Department of Transportation or the US Department of Transportation.

10

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882

Appendix

Table A1 Phase quantification of the core and surface samples for 30% slag mortar bars stored in 5% Na2SO4 solution for 1 year. Sample ID 30S1-A

30S2-A

30S3-A

30S4-A

Phase amount (wt %)

Core

Surface

Core

Surface

Core

Surface

Core

Surface

Belite Calcite CH Gypsum Ettringite Monosulfate Hydrotalcite Amorphous

1.1 0.3 10.1 1.6 6.2 2.2 0.9 77.6

0.5 15.6 1.3 5.1 13.9 0.0 0.0 63.7

1.0 0.4 8.8 0.7 4.6 3.2 0.6 80.6

0.0 13.2 1.3 4.6 17.3 0.0 0.1 63.6

1.5 1.4 8.0 2.4 5.9 4.6 0.2 75.9

0.0 11.2 0.0 6.7 21.1 0.0 0.4 60.6

1.6 0.5 6.5 1.7 8.5 4.2 1.1 75. 4

0.0 10.7 0.0 0.7 26.6 0.0 0.5 61.4

Table A2 Phase quantification of the core and surface samples for 70% slag mortar bars stored in 5% Na2SO4 solution for 1 year. Sample ID 70S1-A

70S2-A

70S3-A

70S4-A

Phase Amount (wt %)

Core

Surface

Core

Surface

Core

Surface

Core

Surface

Belite Calcite CH Gypsum Ettringite Monosulfate Hydrotalcite Amorphous

0.6 1.2 2.1 4.0 3.0 0.5 1.6 86.8

0.0 23.3 0.0 1.1 2.7 0.0 0.7 72.3

0.7 3.3 1.2 2.6 4.9 0.9 1.6 84.6

0.2 20.4 0.0 1.6 4.6 0.0 0.8 72.5

0.9 0.7 1.4 2.1 4.4 1.2 0.3 89.1

0.3 17.3 0.0 0.8 3.6 0.0 0.4 77.5

1.0 0.7 1.1 0.8 4.6 1.0 1.6 89. 8

0.5 11.7 0.0 0.9 4.4 0.0 1.2 81.4

References [1] A.M. Neville, Properties of Concrete, fourth ed., Pearson Education Limited, Harlow, England, 2006. [2] C. Yu, W. Sun, K. Scrivener, Mechanism of expansion of mortars immersed in sodium sulfate solutions, Cem. Concr. Res. 43 (1) (2013) 105–111. [3] M. Gonzalez, E. Irassar, Ettringite formation in low C3A portland cement exposed to sodium sulfate solution, Cem. Concr. Res. 27 (7) (1997) 1061–1072. [4] B. Tian, M.D. Cohen, Does gypsum formation during sulfate attack on concrete lead to expansion?, Cem Concr. Res. 30 (2000) 117–123. [5] M. Santhanam, M.D. Cohen, J. Olek, Effects of gypsum formation on the performance of cement mortars during external sulphate attack, Cem. Concr. Res. 33 (2003) 325–332. [6] N. Shanahan, A. Zayed, Role of tricalcium silicate in sulfate resistance, Adv. Cem. Res. 27 (7) (2015) 409–416. [7] M.D. Cohen, B. Mather, Sulfate attack on concrete – research needs, Mater. J. 88 (1991) 62–69. [8] R.S. Gollop, H.F.W. Taylor, Microstructural and microanalytical studies of sulfate attack. I. Ordinary portland cement paste, Cem. Concr. Res. 22 (6) (1992) 1027–1038. [9] W. Müllauer, R.E. Beddoe, D. Heinz, Sulfate attack expansion mechanisms, Cem. Concr. Res. 52 (2013) 208–215. [10] A. Zayed, N. Shanahan, K. Brown, A. Hanhan, Cement Composition and Structural Durability in Florida, University of South Florida, Tampa, FL, 2003. [11] F. Nosouhian, D. Mostofinejad, H. Hasheminejad, Concrete durability improvement in a sulfate environment using bacteria, J. Mater. Civ. Eng. 28 (1) (2016). [12] S. Mindess, J.F. Young, D. Darwin, Concrete, Prentice Hall, 2003, p. 644. [13] F. Nosouhian, D. Mostofinejad, H. Hasheminejad, Influence of biodeposition treatment on concrete durability in a sulphate environment, Biosyst. Eng. 133 (2015) 141–152. [14] F. Nosouhian, D. Mostofinejad, Reducing permeability of concrete by bacterial mediation on surface using treatment gel, ACI Mater. J. 113 (3) (2016) 287– 293. [15] E. Özbay, M. Erdemir, H.I. Durmusß, Utilization and efficiency of ground granulated blast furnace slag on concrete properties – a review, Constr. Build. Mater. 105 (2016) 423–434. [16] ASTM C989, Standard Specification for Slag Cement for Use in Concrete and Mortars, ASTM Standards, 2013, pp. 1–8. [17] R.D. Hooton, Canadian use of ground granulated blast-furnace slag as a supplementary cementing material for enhanced performance of concrete, Can. J. Civ. Eng. 27 (4) (2000) 754–760.

[18] D.D. Higgins, Increased sulfate resistance of ggbs concrete in the presence of carbonate, Cem. Concr. Compos. 25 (8) (2003) 913–919. [19] R.D. Hooton, J. Emery, Sulfate resistance of a Candian slag cement, ACI Mater. J. 87 (1990) 547–555. [20] M.J. Whittaker, The Impact of Slag Composition on the Microstructure of Composite Slag Cements Exposed to Sulfate Attack, 2014. [21] R.S. Gollop, H.F.W. Taylor, Microstructural and microanalytical studies of sulfate Attack. V. Comparison of different slag blends, Cem. Concr. Res. 26 (7) (1996) 1029–1044. [22] S. Ogawa, T. Nozaki, K. Yamada, H. Hirao, R.D. Hooton, Improvement on sulfate resistance of blended cement with high alumina slag, Cem. Concr. Res. 42 (2) (2012) 244–251. [23] R.S. Gollop, H.F.W. Taylor, Microstructural and microanalytical studies of sulfate attack. IV. Reactions of a slag cement paste with sodium and magnesium sulfate solutions, Cem. Concr. Res. 26 (7) (1996) 1013–1028. [24] C. Yu, W. Sun, K. Scrivener, Degradation mechanism of slag blended mortars immersed in sodium sulfate solution, Cem. Concr. Res. 72 (2013) 37–47. [25] F. Fernàndez-Altable, A.Y. Leinekugel-le-Cocq-Errien, G. Le Saoût, K. Scrivener, Effect of slag on aluminate distribution and sulphate resistance, in: Proceedings of the 17th International Conference on Building Materials (IBAUSIL), 2009, pp. 691–696. [26] M. Whittaker, M. Zajac, M. Ben Haha, L. Black, The impact of alumina availability on sulfate resistance of slag composite cements, Constr. Build. Mater. 119 (2016) 356–369. [27] C. Yu, W. Sun, K. Scrivener, Cement and concrete research degradation mechanism of slag blended mortars immersed in sodium sulfate solution, Cem. Concr. Res. 72 (2015) 37–47. [28] D. Mostofinejad, F. Nosouhian, H. Nazari-Monfared, Influence of magnesium sulphate concentration on durability of concrete containing micro-silica, slag and limestone powder using durability index, Constr. Build. Mater. 117 (2016) 107–120. [29] M. Juenger et al., Supplementary cementitious materials for concrete: characterization needs, MRS Proc. 1488 (2012). pp. imrc12-1488-7b-026. [30] A. Markandeya, N. Shanahan, D.M. Gunatilake, K.A. Riding, A. Zayed, Influence of slag composition on cracking potential of slag-portland cement concrete, Constr. Build. Mater. 164 (2018) 820–829. [31] ASTM C109/C109M-16a, Standard Test Method for Compressive Strength of Hydraulic Cement Mortars (Using 2-in. or [50-mm] Cube Specimens), ASTM International, West Conshohocken, PA, 2016. [32] ASTM C305-14, Standard Practice for Mechanical Mixing of Hydraulic Cement Pastes and Mortars of Plastic Consistency, ASTM International, West Conshohocken, PA, 2014.

F. Nosouhian et al. / Construction and Building Materials 229 (2019) 116882 [33]

[34]

[35]

[36]

[37]

[38]

[39]

[40] [41]

[42]

[43] [44] [45]

[46]

[47]

[48] [49] [50]

ASTM C1012/C1012M-15, Standard test method for length change of hydraulic-cement mortars exposed to a sulfate solution, ASTM International, West Conshohocken, PA, 2015, pp. 5–9. D. Jansen, F. Goetz-Neunhoeffer, C. Stabler, J. Neubauer, A remastered external standard method applied to the quantification of early OPC hydration, Cem. Concr. Res. 41 (6) (2011) 602–608. D. Jansen, S.T. Bergold, F. Goetz-Neunhoeffer, J. Neubauer, The hydration of alite: a time-resolved quantitative XRD approach using the G-factor method compared with heat release, J. Appl. Crystallogr. 44 (5) (2011) 895–901. D. Jansen, C. Stabler, S. Dittrich, J. Neubauer, ‘‘Does Ordinary Portland Cement contain amorphous phase ? A quantitative study using an external standard method,” no. January, pp. 31–38, 2011. R. Snellings, A. Bazzoni, K. Scrivener, Cement and concrete research the existence of amorphous phase in portland cements: physical factors affecting Rietveld quantitative phase analysis, Cem. Concr. Res. 59 (2014) 139–146. I.C. Madsen, N.V.Y. Scarlett, A. Kern, Description and survey of methodologies for the determination of amorphous content via X-ray powder diffraction, Zeitschrift für Krist. 226 (12) (2011) 944–955. D. Bish, R.J. Reynolds, Sample preparation for X-ray diffraction, in: D. Bish, J. Post (Eds.), Modern Powder Diffraction, The Mineralogical Society of America, Washington, DC, 1989, pp. 73–99. Paul Scherrer Institut (PSI), GEMS 3 [Software]. D.A. Kulik et al., GEM-Selektor geochemical modeling package: revised algorithm and GEMS3K numerical kernel for coupled simulation codes, Comput. Geosci. 17 (2013) 1–24. T. Wagner, D.A. Kulik, F.F. Hingerl, S.V. Dmytrieva, GEM-Selektor geochemical modeling package: TSolMod library and data interface for multicomponent phase models, Can. Mineral. 50 (2012) 1173–1195. W. Hummel, U. Berner, E. Curti, F.J. Pearson, T. Thoenen, Nagra/PSI chemical thermodynamic data base 01/01, Radiochim. Acta 90 (9–11) (2002) 805–813. EMPA, CEMDATA14. A. Schöler, B. Lothenbach, F. Winnefeld, M. Zajac, Hydration of quaternary Portland cement blends containing blast-furnace slag, siliceous fly ash and limestone powder, Cem. Concr. Compos. 55 (2015) 374–382. W. Kunther, B. Lothenbach, K.L. Scrivener, On the relevance of volume increase for the length changes of mortar bars in sulfate solutions, Cem. Concr. Res. 46 (2013) 23–29. V. Kocaba, E. Gallucci, K.L. Scrivener, Cement and concrete research methods for determination of degree of reaction of slag in blended cement pastes, Cem. Concr. Res. 42 (3) (2012) 511–525. K. Luke, F.P. Glasser, Selective dissolution of hydrated blast furnace slag cements, Cem. Concr. Res. 17 (2) (1987) 273–282. ASTM C150/C150M-16, Standard Specification for Portland Cement, ASTM International, West Conshohocken, PA, 2016. P.K. Mehta, P.J.M. Monteiro, Concrete: Microstructure, Properties and Materials, third ed., McGraw-Hill, New York, NY, 2006.

11

[51] N. Shanahan, A. Zayed, Cement composition and sulfate attack, Cem. Concr. Res. 37 (4) (2007) 618–623. [52] H.T. Cao, L. Bucea, A. Ray, S. Yozghatlian, The effect of cement composition and pH of environment on sulfate resistance of Portland cements and blended cements, Cem. Concr. Compos. 19 (2) (1997) 161–171. [53] P.J.M. Monteiro, K.E. Kurtis, Time to failure for concrete exposed to severe sulfate attack, Cem. Concr. Res. 33 (7) (2003) 987–993. [54] P.K. Mehta, D. Pirtz, M. Polivka, Properties of alite cements, Cem. Concr. Res. 9 (4) (Jul. 1979) 439–450. [55] T. Meagher, N. Shanahan, D. Buidens, K.A. Riding, A. Zayed, Effects of chloride and chloride-free accelerators combined with typical admixtures on the earlyage cracking risk of concrete repair slabs, Constr. Build. Mater. 94 (2015) 270– 279. [56] F.W. Locher, The problem of the sulfate resistance of slag cement, no. 9. 1966, pp. 395–401. [57] T.H. Wee, A.K. Suryavanshi, S.F. Wong, A.K.M.A. Rahman, Sulfate resistance of concrete containing mineral admixtures, Aci Mater. J. 97 (5) (2000) 536–549. [58] V. Fernández-Altable, C. Yu, G. Le Saout, K. Scrivener, Characterisation of slag blends and correlation with their sulphate resistance in static and semidynamic conditions, in: 13th International Congress on the Chemistry of Cement, 2011, pp. 434–440. [59] M. Ben Haha, B. Lothenbach, G. Le Saout, F. Winnefeld, Cement and concrete research influence of slag chemistry on the hydration of alkali-activated blastfurnace slag—Part I: Effect of MgO, Cem. Concr. Res. 41 (9) (2011) 955–963. [60] M. Whittaker, M. Zajac, M. Ben Haha, F. Bullerjahn, L. Black, The role of the alumina content of slag, plus the presence of additional sulfate on the hydration and microstructure of Portland cement-slag blends, Cem. Concr. Res. 66 (2014) 91–101. [61] E.F. Irassar, Sulfate attack on cementitious materials containing limestone filler – a review, Cem. Concr. Res. 39 (3) (2009) 241–254. [62] E.F. Irassar, V.L. Bonavetti, M. González, Microstructural study of sulfate attack on ordinary and limestone Portland cements at ambient temperature, Cem. Concr. Res. 33 (1) (2003) 31–41. [63] Á. Fernández, B. Lothenbach, M.C. Alonso, J.L. García Calvo, Thermodynamic modelling of short and long term hydration of ternary binders. Influence of Portland cement composition and blast furnace slag content, Constr. Build. Mater. 166 (2018) 510–521. [64] M.C.G. Juenger, R. Siddique, Recent advances in understanding the role of supplementary cementitious materials in concrete, Cem. Concr. Res. 78 (2015) 71–80. [65] B. Lothenbach, B. Bary, P. Le Bescop, T. Schmidt, N. Leterrier, Sulfate ingress in Portland cement, Cem. Concr. Res. 40 (8) (2010) 1211–1225. [66] 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 (1) (2004) 67–80.