Resistance of concrete and mortar against combined attack of chloride and sodium sulphate

Resistance of concrete and mortar against combined attack of chloride and sodium sulphate

Cement & Concrete Composites 53 (2014) 59–72 Contents lists available at ScienceDirect Cement & Concrete Composites journal homepage: www.elsevier.c...

2MB Sizes 55 Downloads 128 Views

Cement & Concrete Composites 53 (2014) 59–72

Contents lists available at ScienceDirect

Cement & Concrete Composites journal homepage: www.elsevier.com/locate/cemconcomp

Resistance of concrete and mortar against combined attack of chloride and sodium sulphate Mathias Maes, Nele De Belie ⇑ Magnel Laboratory for Concrete Research, Faculty of Engineering and Architecture, Department of Structural Engineering, Ghent University, Technologiepark – Zwijnaarde 904, B-9052 Ghent, Belgium

a r t i c l e

i n f o

Article history: Received 5 August 2013 Received in revised form 10 June 2014 Accepted 23 June 2014 Available online 1 July 2014 Keywords: Concrete Mortar Chlorides Sulphates Combined attack Blast-Furnace Slag

a b s t r a c t Marine environments are typically aggressive to concrete structures, since sea water contains high concentrations of chlorides and sulphates. To improve predictions of concrete durability within such environments, it is important to understand the attack mechanisms of these ions in combination. In this research, the reciprocal influence of Cl and SO2 4 was investigated for four mixtures, namely with Ordinary Portland Cement, High Sulphate Resistant cement, and with Blast-Furnace Slag (50% and 70% cement replacement). Chloride penetration depths and diffusion coefficients were measured to  investigate the influence of SO2 4 on Cl attack. Besides, length and mass change measurements were per formed to examine the influence of Cl on SO2 4 attack. Since the formation of ettringite, gypsum and Friedel’s salt plays an important role, XRD-analyses were done additionally. It can be concluded that chloride penetration increases when the sulphate content increases at short immersion periods, except for HSR concrete. Concerning the sulphate attack, the presence of chlorides has a mitigating effect. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction A lot of damage is reported for constructions in marine environments [1–8]. Marine environments are very aggressive, since sea water consists mainly of chlorides and sulphates. Both ions can be very harmful for the durability of concrete structures. However, almost no literature is found about the reciprocal influence. Chlorides affect durability by initiating corrosion of the reinforcement steel, and sulphates by deteriorating the concrete itself. Concrete structures in marine environments mostly have a high economic impact (e.g. bridges, wharfs, piers, tunnels, etc.), so it is important to know the attack mechanisms in detail in order to predict concrete’s service life as exactly as possible. Concerning concrete deterioration due to chlorides, it is important to notice that corrosion will only be initiated by the free chlorides and not by the fraction that is chemically bound to the cement hydrates or physically adsorbed at the pore walls. So, chloride binding is a significant factor related to reinforced concrete durability for three reasons [9]: reduction of the free chloride concentration in the vicinity of the reinforcing steel will reduce the risk of corrosion; chloride binding will delay the chloride penetra-

⇑ Corresponding author. Tel.: +32 9 264 55 22; fax: +32 9 264 58 45. E-mail address: [email protected] (N. De Belie). http://dx.doi.org/10.1016/j.cemconcomp.2014.06.013 0958-9465/Ó 2014 Elsevier Ltd. All rights reserved.

tion; formation of Friedel’s salt results in a less porous structure and slows down the transport of Cl-ions. Friedel’s salt (3CaOAl2O3CaCl210H2O) is the result of chemical binding between chlorides and C3A. Chemical binding can also occur between chlorides and C4AF. Besides, physical binding occurs due to interaction with CSH. Factors influencing binding are [10–13]: chloride concentration, cement type, cement replacement, cation, alkalinity, temperature, water-binder factor, etc. Cement replacement by Blast-Furnace Slag (BFS) seems to have a positive influence on the resistance of concrete against chloride penetration. BFS concrete is already used in big marine structures because of the low hydration heat [14]. Besides, partial replacement of Ordinary Portland Cement (OPC) by BFS is considered as a promising way to improve concrete’s service life when the prediction is based on chloride penetration. In general, it is assumed that BFS concrete is able to bind more chlorides. This is attributed to increased Friedel’s salt formation. Replacement of 70% of the cement by BFS is the most suitable in view of chloride binding [15]. On the other hand, former research does not always show a higher chloride binding capacity of BFS concrete compared to OPC concrete [12,16]. Besides, when cured properly, BFS concrete possesses a finer pore structure resulting in a limited penetration depth of free chlorides and consequently a reduced risk for chloride initiated corrosion in comparison with

60

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

OPC concrete. The chloride diffusion coefficient of BFS concrete reaches lower values than the coefficient of OPC concrete [16–20]. The other main attack mechanism in marine environments is external sulphate attack. This occurs when water contaminated with sulphates penetrates into the concrete by means of diffusion or capillary suction. Sulphates are mostly found in the form of sodium sulphate (Na2SO4) or magnesium sulphate (MgSO4). The cation associated with SO2 4 has an influence on the attack mechanism and the resulting deterioration [21]. Sodium sulphate attack will result in expansive reaction products while magnesium sulphate attack will result in reduction in strength. In current paper, the influence of Na2SO4 and the combined attack of NaCl and Na2SO4 is examined. Because of the different attack mechanism, the influence of MgSO4 and the combined attack of NaCl and MgSO4 will be discussed in following papers. Once penetrated into the concrete, sulphates react with the lime formed during the hydration process of the Portland clinker. One of the reaction products is calcium sulphate or secondary gypsum. This calcium sulphate reacts with hydrated calcium aluminates, C3A, and forms ettringite (3CaOAl2O33CaSO432H2O). Gypsum leads to reduction of stiffness and strength, expansion and cracking and eventually to transformation of the material into a mushy and non-cohesive mass. Ettringite has the ability to swell strongly, which results in a densification of the microstructure followed by internal stresses that lead to cracking and destruction of the concrete [22]. The factors influencing the rate of external attack are: the quantity of sulphate ions, the possibility of the sulphates to penetrate into the concrete and the volume of C3A in the cement and the type of cement used to make the concrete. Temperature changes can have an influence on the sulphate attack mechanism as well. At low temperatures and in presence of soluble carbonate and reactive silicate, thaumasite (Ca3Si(CO3) (SO4)(OH)612H2O) can be formed. This is not expansive but lowers the strength and has a negative influence on the microstructure. It is generally assumed that thaumasite is only formed at temperatures below 15 °C. However, some researchers also found thaumasite at temperatures higher than 15 °C [23–25]. In the current paper, the influence of temperature fluctuations on the sulphate attack mechanism and on chloride penetration was not examined; the tests were performed at 20 °C. Nevertheless, in West-European marine environments sea water temperatures are often below 10 °C while in the Middle East and South East Asia sea water temperatures can be higher than 30 °C. According to Aköz et al. [26] raising temperatures of sodium sulphate solutions in the range of 20–40 °C have a beneficial influence on resistance of mortar against sulphate attack. However, they are not able to determine the dominant factor affecting the performance of mortar in a sodium sulphate solution at 40 °C. On the other hand they observed that raised temperatures until 40 °C could have negative effects on mortar resistance in magnesium sulphate solutions. They assume that this is due to decalcification of C–S–H to M–S– H, which leads to a porous structure. Concerning the influence of fluctuating temperatures on chloride diffusion, some researchers [7,27,28] found that chloride diffusion rises with increases in temperature. The use of cement replacement materials tends to improve the resistance against sulphate attack [29]. However, when BFS is used and the samples are exposed to magnesium sulphate, deterioration exceeds that observed in Portland cements [30]. In that case, the calcium hydroxide is consumed by the pozzolanic reaction, so the sulphates and the magnesium ions will react directly with the C–S–H due to the absence of Ca(OH)2, resulting in a cohesionless M–S–H [31]. Furthermore, BFS concrete which is partially immersed in a sulphate solution will show severe deterioration in the upper parts of the concrete in contact with air due to salt crystallization [32,33]. This is in contrast with the situation with

complete immersion. When the concrete is completely immersed in the sulphate solution, BFS concrete will have a higher resistance than OPC concrete. Generally, the C3A-content of the cement plays a major role in the attack mechanism of sulphates and in the binding behaviour of chlorides. When both ions penetrate the concrete together, C3A-binding will definitely influence this multi-ion transport. In literature a lot of papers are found on individual chloride and sulphate attack. Nevertheless, only limited literature is available concerning combined environmental attack of chlorides and sulphates. Two groups of papers are found: firstly, the papers investigating the effect of chlorides on sulphate attack and secondly the papers investigating the effect of sulphates on chloride attack. According to Al-Amoudi et al. [34] there are three possible schools of thought concerning the influence of chlorides on sulphate attack: (1) the sulphate attack mechanism is intensified, (2) sulphate attack is mitigated, and (3) the influence is insignificant. In their research, they examined the role of chloride ions on expansion and strength reduction due to sulphate attack by adding high volumes of sodium chloride, namely 15.7% Cl, to mixed sodium and magnesium sulphate solutions. The sulphate concentration amounted to 2.1% SO2 4 in which the sodium and magnesium sulphate were proportioned to provide 50% of the sulphate concentration from each of them. They found that the deterioration is more severe for specimens immersed in a pure sulphate solution than in a combined sulphate–chloride solution. Concerning the influence of the cement type, they concluded that the deterioration due to sulphate attack is not very different in cements with varying C3A contents in the range of 3.5–8.5%. Besides, Al-Amoudi et al. also found that replacement of OPC by BFS has only a marginal beneficial effect. Also Abdalkader et al. [35] investigated the influence of chloride on the performance of mortars subjected to sulphate exposure. Their tests were done at 5 °C with 6 g/l SO2 4 magnesium sulphate solutions diluted with 5 g/l Cl sodium chloride. They observed that sulphate attack is more severe when the samples are immersed in a combined solution compared to those stored in a pure sulphate solution, since the presence of chlorides accelerates damage caused by thaumasite. Based on the findings of Zuquan et al. [36] the influence of chlorides on the sulphate attack mechanism prolong the periods of the attack process, which results in a less severe deterioration. Their tests were performed at 20 °C with a combined solution of 5% Na2 SO4 and 3.5% NaCl. Concerning the influence of sulphates on chloride attack, De Weerdt and Geiker [37] concluded that chloride ions penetrate much deeper into the concrete compared to other elements originating from the seawater e.g. Mg2+, S2 and Na+ which have, approximately, a constant concentration from 20 mm inwards. This should mean that the binding competition between Cl and SO2 only occurs in the outermost layers. Nevertheless, due to 4 the binding of chlorides, more free sulphate ions will be present and they are able to penetrate deeper into the concrete. However, data found by Brown and Badger [22] indicate that Friedel’s salt converts to ettringite in the presence of sodium sulphate solution. Thus, ettringite is the stable phase under these conditions. This should mean that more free chlorides will be present in the concrete. Based on the findings of Zuquan et al. [36] the influence of sulphates (in the form of Na2SO4) on chloride diffusion is dependent on the exposure period, namely at early exposure periods the presence of sulphate decreases the concentration of chlorides and the chloride diffusion coefficient. But at later exposure periods, the presence of sulphates in combined solutions, cause an increased ingress of chloride. In this research, the influence of the multi-ion transport on the proper attack mechanisms was examined. Accelerated tests were

61

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

conducted in the laboratory by increasing the chloride and sulphate concentration of the test solution. On the one hand, the influence of sulphates on chloride attack was investigated by means of diffusion tests whereupon colour change boundaries and chloride diffusion coefficients were calculated. On the other hand, the influence of chlorides on sulphate attack was examined by measuring the length and mass change of specimens immersed in a combined solution. In addition, XRD-analyses supplemented with quantitative Rietveld analyses were performed.

2. Materials and methods 2.1. Materials To examine the influence of sulphates on chloride penetration, four different concrete mixtures were used: two Portland cement mixtures, namely one with Ordinary Portland Cement (OPC) and one with High-Sulphate Resistant cement (HSR), as well as two Blast-Furnace Slag (BFS) mixtures with Portland cement replacement levels of 50% (S50) and 70% (S70). Table 1 gives an overview of the compositions. The total binder content (cement + slag = B) was maintained at 350 kg/m3 and the water-to-binder factor (W/ B) at 0.45. These values are in accordance with EN 206–1 [38], when the concrete is applied in an XS2 environment. Superplasticizer (SP) based on polycarboxylic ethers was added to the mixture to obtain a slump between 100 mm and 150 mm (consistency class S3). Cubes with 150 mm side were cast and cured at 20 °C and a relative humidity (R.H.) higher than 95%. They were demoulded after 24 h and cured under the same conditions until the age of testing. Then a cylinder with diameter 100 mm was drilled from each cube and this cylinder was cut in three specimens with a thickness of 50 mm. The characteristics of the fresh concrete are tabulated in Table 1 as well. The slump was measured according to NBN EN 12350-2 [39] with indication of the consistency class and air content according to NBN EN 12350-7 [40]. To examine the influence of chlorides on the sulphate attack mechanism, four different mortar mixtures were used with the same binder types as used in the concrete mixes. The water-to-binder factor (W/B) was also maintained at 0.45. The water-to-sand factor was maintained at 0.15. Mortar cubes with a 20 mm side and mortar prisms 20  20  160 mm were prepared and cured at 20 °C and 95% R.H. until the age of testing. Besides, some cement paste samples were produced to perform XRD-analysis. These samples also had a W/B-ratio of 0.45. Prisms of 40  40  160 mm were produced and cured at 20 °C and 95% R.H. After 28 days curing, slices of 40  40  7 mm were cut off

Table 1 Concrete compositions and fresh concrete characteristics (slump according to NBN EN 12350–2).

Sand 0/4 (kg/m3) Aggregate 2/8 (kg/m3) Aggregate 8/16 (kg/m3) CEM I 52.5 N (kg/m3) CEM I 52.5 N HSR (kg/m3) BFS (kg/m3) Water (kg/m3) W/B-factor (–) BFS/B-factor (%) SP (ml/kg B) Slump Air content (%) Density (kg/m3)

OPC

HSR

S50

S70

781 619 480 350 – – 157.5 0.45 0 1.2 S3 3.2 2343

781 619 480 – 350 – 157.5 0.45 0 1.2 S3 1.4 2412

781 619 480 175 – 175 157.5 0.45 50 2.9 S3 3.0 2350

781 619 480 105 – 245 157.5 0.45 70 2.9 S3 1.9 2387

and these slices were vacuum saturated before immersing in the test solutions. In Table 2 the chemical composition of the Portland cements and the slag, determined in accordance with NBN EN 196-2 [41] and using Wavelength Dispersive X-ray Spectroscopy (WD-XRF). Blaine’s fineness and density of the cement and slag are shown as well. The slag meets all the requirements mentioned in NBN EN 15167-2 [42] and in the Belgian guidelines for a technical approval for BFS. Since HSR cement is used because of its low C3A-content, the C3A-content was calculated by using the Bogue equations. According to EN 197-1 [13], the C3A-content for HSR cement is limited to 3%. In current research, the C3A-content for HSR cement amounted to 2.50% and corresponds to the standard. For OPC the C3A-content was 7.92%. 2.2. Methods To determine the resistance against chloride penetration and the influence of sulphates hereon, a diffusion test was performed. Afterwards, Colour Change Boundaries (ccb), which give an indication of the chloride penetration depths, and chloride profiles, which provide the data to calculate a chloride diffusion coefficient (Dnssd), were obtained. The cylindrical concrete specimens were tested at the age of 28 days. To investigate sulphate attack and the effect of chlorides on this mechanism, length and mass change were measured. The mortar prisms and cubes were also tested at the age of 28 days. 2.2.1. Diffusion test The resistance of concrete to chlorides was evaluated by the diffusion test as described in NT Build 443 [43]. Using combined solutions next to the prescribed single chloride solution makes it possible to examine the influence of sulphates on chloride penetration. At the age of 28 days, the specimens were saturated in a 4 g/l Ca(OH)2 solution. After 10 days of immersion in this solution, the specimens were coated, except for the casting surface. The coated specimens were placed in the 4 g/l Ca(OH)2 solution for another 10 days. Afterwards the specimens were placed in the test solution, an aqueous NaCl solution with a concentration of 165 g NaCl per liter solution, equal to the one described in NT Build 443. To determine the influence of sulphates, two extra test solutions were made. These test solutions (Table 3) contained 165 g/l NaCl as well, but the sulphate content amounted to 27.5 g/l Na2SO4 (18.6 g/l 2 SO2 4 ) and 55 g/l Na2SO4 (37.2 g/l SO4 ). In the first combination, the Cl/SO2 ratio is equal to the ratio found in sea water from 4 the North Sea. The second combination has a higher Cl/SO2 4 ratio Table 2 Chemical composition, Blaine fineness, water content and density of the cement types and blast-furnace slag. Content (%)

CEM I 52.2 N

CEM I 52.2 N HSR

BFS

CaO SiO2 Al2O3 Fe2O3 SO3 MgO K2O CO2 Na2O Cl Sulphide content Insoluble residue Loss on ignition Blaine fineness (m2/kg) Density (kg/m3)

63.37 18.90 5.74 4.31 3.34 0.89 0.73 0.50 0.47 – – 0.41 1.51 353 3122

63.90 21.62 3.53 4.05 2.40 1.82 0.51 0.34 0.15 0.026 0.05 0.48 0.95 – –

41.24 36.37 9.83 0.26 1.62 7.41 0.41 0.90 0.28 0.02 0.79 0.43 1.30 394 2830

62

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

Table 3 Overview of the test solutions.   Influence of SO2 4 on Cl -attack (Ccb + Cl -profiles) Ref. (Cl) 165 g/l NaCl Comb. 1 (Cl) 165 g/l NaCl + 27.5 g/l Na2SO4  Comb. 2 (Cl ) 165 g/l NaCl + 55 g/l Na2SO4

Influence of Cl on SO2 4 -attack (Mass + Length change) Ref. (SO2 50 g/l Na2SO4 4 ) 2 Comb. 1 (SO4 ) 50 g/l Na2SO4 + 50 g/l NaCl

in order to clarify the influence of the sulphates more in detail. The tests took place at 20 °C and the period of immersion lasted for 7 weeks (ccb + Dnssd) or for 14 weeks (ccb). After storage in the test solutions, the ccb and chloride profiles were obtained. The ccb was determined by means of the colorimetric method, more specifically by spraying a 0.1 M AgNO3 solution onto both halves of split specimens. This results in a visible white deposit of AgCl2, where free chlorides have penetrated into the concrete. For analysis, photographs of the split specimens were taken and analysed by using ImageJ Software. For each half specimen the penetration depth was measured at 6–9 places with an interval of 10 mm. To obtain a chloride profile, chloride concentrations had to be measured at different depths. For practical reasons, chloride profiles and the ccb could not be obtained from the same specimens. Both were determined on different specimens from the same batch. Powder was collected from the cylindrical specimens up to a depth of 20 mm, using a profile grinder. Layers of 2 mm thickness were ground. Acid-soluble as well as water-soluble chlorides were extracted from the powder. The acid-soluble chloride content gives an indication of the total chloride content and the water-soluble chloride content is used to estimate the free chloride content. The extraction and titration method is similar to the method described by Maes et al. [20]. Although, the titration solution was slightly changed and had a total volume of 50 ml which consists of 20 ml 0.3 mol/l HNO3 supplemented with 10 ml acid soluble chloride extraction solution and 20 ml distilled water on the one hand or with 5 ml water soluble chloride extraction solution and 25 ml distilled water on the other hand. The titration was executed with 0.01 M AgNO3. For the analysis, namely a potentiometric titration, a Metrohm MET 702 automatic titrator was used. Repeatability of the used extraction and titration method in this paper was checked. Triplicate tests were conducted on four homogenized samples. The results indicate that the used water extraction method is quite repeatable, since the coefficient of variation ranged from 0.8% to 4.8%. This corresponds to the repeatability results obtained by Yuan [12]. Chloride contents, in wt% of concrete, were calculated using Eqs. (1) and (2) (ct = total chloride content, cw = water-soluble chloride content).

ct ð%Þ ¼

10  100  35:45  0:01  Vol: AgNO3 ðmlÞ 1000  2

10  100  35:45  0:01  Vol: AgNO3 ðmlÞ cw ð%Þ ¼ 1000  2:5

ð1Þ

chloride content cf is assumed to be 80% of the water-soluble chloride content, in accordance with the findings of Yuan [12]. He compared the water-soluble chloride content and the free chloride content which was measured by means of a pore solution extraction. Furthermore, the bound chloride content is the difference between total and free chloride content. The non-steady state diffusion coefficient Dnssd was calculated in accordance to the method described in NT Build 443 [43]. Non-steady-state diffusion coefficients and chloride surface concentrations were obtained by fitting Eq. (3) to the measured chloride profiles, using a non-linear regression analysis in accordance with the least squares method.

  x cðx; tÞ ¼ cs  ðcs  ci Þerf pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4Dnssd t

where c(x, t) is the chloride concentration at depth x and time t (mass% concrete), ci the initial chloride concentration (mass% of concrete), cs the chloride concentration at the surface (mass% of concrete), Dnssd the non-steady-state diffusion coefficient (m2/s), x the distance from the surface until the middle of the considered layer (m) and t the exposure time (s). Compared to the mass of concrete, it was reasonable to assume that the initial chloride concentration ci in Eq. (3) equalled 0% (in reality, this value can differ from 0%). The first layer was excluded from the regression analysis, since the measured chloride concentration in the first layer is generally considered not representative. The performed method exhibits a minor adjustment compared to the method prescribed in NT Build 443 [43]. The straightforward method prescribed in there is not suitable for detecting small changes in diffusion coefficients [44]. Thus, first an average surface chloride concentration was calculated per concrete mixture, regardless the used test solution. This surface concentration was used to fit Eq. (3) to the measured chloride profiles again. This time only the non-steady state diffusion coefficient was estimated. This way, by using a constant chloride surface concentration per mixture, the small differences between the diffusion coefficients become more visible. 2.2.2. Mass change To examine the influence of chlorides on the sulphate attack mechanism, mass change measurements were done. At the age of 28 days, mortar cubes with a 20 mm side were immersed in the test solutions, namely a 50 g/l Na2SO4 solution and a 50 g/l Na2SO4 + 50 g/l NaCl solution (Table 3). The Na2SO4 concentration is conform the ASTM C 1012-4 standard [45], the concentration of the added NaCl is equal to make a clear distinction between the single-ion solution and the multi-ion solution. Firstly, the specimens were vacuum saturated with a 4 g/l Ca(OH)2 solution. The mass of the specimen (surface dry) was measured just before immersion in the test solution, after saturation, mref and at different time intervals mx: every 2 weeks until 8 weeks immersion, then every month until 6 months immersion and then every 4– 5 months until 20 months immersion. Consequently, the mass change was calculated as follows:

Mass change ð%Þ ¼ ð2Þ

where 10 represents the dilution factor; 35.45 is the atomic mass of chlorides (g/mol); 0.01 is the concentration of the titration solution (mol/l); and 2 (or 2.5 for water-soluble chloride content) is the mass of the concrete powder in the extraction solution. The factor 0.01 can be replaced by the exact concentration of the AgNO3-solution, resulting from the calibration and with unit mol/l. The water-soluble chloride concentration gives an indication of the free chloride content in the concrete. In this research, the free

ð3Þ

mx  mref  100 mref

ð4Þ

2.2.3. Length change Another method to examine the influence of chlorides on the sulphate attack mechanism is to measure length change. Mortar prisms 20  20  160 mm were prepared with metal studs embedded in the mortar, in order to measure more accurately, see Fig. 1. The end of the stud is used as measuring point during the test. To analyze the results, the total length of the studs was subtracted from the total measured length.

63

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

Table 4 Free chloride surface concentrations for specimen immersed at the age of 28 days and at the age of 84 days, respectively. Cs,0 (wt%/concrete)

Fig. 1. Metal studs placed in the moulds to be embedded in the mortar specimens. The end of the stud is used as fixed measuring point.

At 28 days, the prisms were vacuum saturated and immersed in the test solutions, cf. mass change measurements (Table 3). The length of the specimen (‘length between the ends of the metal studs’ – ‘length of the metal studs’) was measured just before immersion lref and at different time intervals lx: every 2 weeks until 10 weeks immersion, then every 11 weeks until 10 months immersion. So, the length change was calculated as follows:

Length change ð%Þ ¼

lx  lref  100 lref

ð5Þ

2.2.4. XRD-analysis To do the XRD-analysis OPC, S50 and S70 cement paste samples were immersed in different test solutions. After a certain immersion time, the samples were crushed to pass a 500 lm sieve. Since the internal standard approach was selected for absolute phase quantification and estimation of the amorphous or non-identified phase content by XRD analysis [46–48], a 10 wt% ZnO internal standard was added to the obtained powder. Finally, the powders were side-loaded into sample holders to reduce preferred orientation effects. The XRD data were collected on a Thermo Scientific ARL X’tra diffractometer equipped with a Peltier cooled detector. Samples were measured in h/2h geometry over an angular range of 5–70° 2h (Cu Ka radiation) using a 0.02° 2h step size and 1 s/step counting time. Afterwards, every XRD-measurement was quantitatively analysed by means of the Rietveld method for whole-powder pattern fitting to investigate the reaction product formation. Topas Academic V4.1 software was used for Rietveld refinement [48,49].

Mean

St. dev.

28 Days OPC HSR S50 S70

0.517 0.504 0.876 0.814

0.044 0.046 0.127 0.134

84 Days OPC HSR S50 S70

0.423 0.485 0.589 –

0.014 0.069 0.080 –

environments, since the degradation is accelerated by increasing the chloride content. Table 5 shows the non-steady state diffusion coefficients Dnssd (1012 m2/s), for OPC and HSR concrete as well as for BFS concrete at different sulphate contents at 28 days and at 84 days. On the one hand, the influence of increased sulphate concentration is not very clear. Generally a small increase in diffusion coefficient is noticed at 28 days, especially in case of a high sulphate concentration of 55 g/l Na2SO4. For the specimen tested at the age of 84 days, it rather seems that Dnssd decreases when sulphates are added to the 165 g/l NaCl reference solution. Although, the diffusion coefficient decrease due to the addition of 55 g/l Na2SO4 for concrete with an age of 84 days is smaller (range 13–19%) than the increase for concrete with an age of 28 days (range 21–28%). On the other hand, the diffusion coefficients clearly decrease when the OPC is replaced by BFS and increase when the OPC is replaced by HSR cement, regardless the age and the composition of the test solution. Fig. 2 shows the chloride penetration depths (ccb) for OPC, HSR, S50 and S70 concrete immersed in the chloride solutions as mentioned in Table 3. Fig. 2a and b shows the penetration depth for specimens immersed at the age of 28 days for a period of 7 weeks and 14 weeks, respectively. Next, Fig. 2c and d shows the penetration depth for specimens immersed at the age of 84 days, also for a period of 7 weeks and 14 week, respectively. From Fig. 2a and b it can be seen that the penetration depth in HSR concrete immersed at 28 days decreases significantly (Oneway ANOVA and Dunnett’s T3 Post Hoc test, level of significance = 0.05) when the Na2SO4 content in the 165 g/l NaCl solution increases from 0 g/l (Ref. (Cl)) to 55 g/l (Comb. 2 (Cl)). The sulphate increase from 0 g/l to 27.5 g/l (Comb. 1 (Cl)) has no statistically significant influence on the chloride penetration depth. Besides, the penetration depth increases for OPC and BFS concrete immersed at 28 days when the sulphates are added to the chloride solution. This increase is significant for both mixtures, regardless

3. Results 3.1. Influence of sulphates on chloride attack The free chloride concentrations at the surface are calculated using Eq. (3) and the measured free chloride profiles. Next to the surface concentrations, also diffusion coefficients are estimated during the first regression analysis. However, these coefficients are not used for further calculations considering the reasons given in Section 2.2. Table 4 gives the mean surface concentrations and standard deviations on the individual values, based on five measurements per mixture, regardless the test solution. These results are used to recalculate the diffusion coefficients. It should be noted that these experimental surface concentrations are not comparable to realistic surface concentrations in marine

Table 5 Non-steady state diffusion coefficients for specimen immersed at the age of 28 days and at the age of 84 days respectively. Dnssd (1012 m2/s) OPC

HSR

S50

S70

28 Days Ref. (Cl) Comb. 1 (Cl) Comb. 2 (Cl)

6.98 6.27 8.42

8.52 9.16 10.55

2.49 2.98 3.19

1.69 3.19 2.09

84 Days Ref. (Cl) Comb. 1 (Cl) Comb. 2 (Cl)

5.27 3.72 4.54

5.97 6.11 5.11

2.94 2.55 2.33

– – –

64

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

Fig. 2. Chloride penetration depth after 7 and 14 weeks immersion starting at an age of 28 days (a and b) or at an age of 84 days (c and d), in three different solutions (cf. Table 3). The error bars represent standard errors on the average.

the immersion time (One-way ANOVA and Dunnett’s T3 Post Hoc test, level of significance = 0.05). After an immersion period of 7 weeks it seems that OPC concrete has the worst resistance against chlorides when 55 g/l Na2SO4 is present in the combined solution. Notwithstanding, after an immersion period of 14 weeks it is clear that HSR concrete has the lowest resistance against chlorides, regardless the sulphate content. Also from Fig. 2c and d, it can be seen that the chloride penetration depth in HSR concrete decreases significantly after 7 weeks immersion when the Na2SO4 content in the 165 g/l NaCl solution increases from 0 g/l to 27.5 g/l and 55 g/l (One-way ANOVA and Dunnett’s T3 Post Hoc test, level of significance = 0.05). After 14 weeks immersion, no significant differences in chloride penetration were measured for HSR. Oppositely, the penetration depth increases for OPC and BFS concrete when the sulphates are added to the chloride solution. However, after 7 weeks immersion, the chloride penetration increase for both BFS mixtures was not statistically significant. Similar to the concrete immersed at an age of 28 days, OPC concrete has the worst resistance against chlorides after an immersion period of 7 weeks when 55 g/l Na2SO4 is present in the combined solution. However, after an immersion period of 14 weeks HSR concrete has the lowest resistance against chlorides, regardless the sulphate content. Overall, the penetration depth for concrete immersed at an age of 84 days is comparable or even slightly higher than the penetration depth for concrete immersed at an age of 28 days. Nevertheless, the penetration depth is expected to decrease in time.

3.2.1. Mass change The results of the mass change measurements are shown in Figs. 3–5. In the graphs, the mass change is shown in percent change compared to the initial mass, cf. Eq. (4). The first graph, Fig. 3, shows the results obtained for all four mortar mixtures after immersion in a 50 g/l Na2SO4 solution without addition of chlorides. Standard errors on the average are shown for the last measurement. The standard errors on the average values of S50 and S70 are too small to be visible in the graph, 0.11% and 0.07% respectively. From Fig. 3, it is clear that OPC mortar specimens lose a larger part of their mass compared to HSR and BFS mortar specimens when permanently immersed in a sodium sulphate solution. Only the mass loss for OPC after 616 days in the solution is significant (t-test, level of significance = 0.05). Until 320 days, the differences between the four mixtures are rather small. It is clear that the mass of the OPC specimens starts decreasing slowly around 100 days in

3.2. Influence of chlorides on sulphate attack The influence of chlorides on the attack mechanism of sodium sulphate is shown by means of mass and length change of mortar specimen. For the mass results, the values shown in the graphs are the average of at least five specimens. The length results are the average of at least three specimens.

Fig. 3. Mass change as a function of the immersion time in a 50 g/l Na2SO4 solution.

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

Fig. 4. Mass change as a function of the immersion time in a combined 50 g/l Na2SO4 + 50 g/l NaCl solution.

the solution, while the mass of the specimens of the other mixtures does not change obviously during this immersion period. After the first 300 days of the immersion period, 2% mass decrease was measured for OPC. Until 300 days of immersion, no statistically significant mass losses were measured. At the end of the testing period, after 616 days in the solution, the mass loss for OPC amounted to 26.5% compared to the initial mass. For the HSR mixture, the mass loss was around 2% and for the BFS mixtures around 0%. It should be noted that the mass of the HSR mortar cubes was increased with 1.2% after 450 days immersion in the 50 g/l Na2SO4 solution. Nevertheless, from that point on the mass of the HSR cubes started decreasing until 2.0% after 616 days storage in the test solution, which is a significant decrease (t-test, level of significance = 0.05). In contrast, the mass of the specimens made with BFS did not change significantly during the whole test period. Fig. 4 shows the results of the mass change measurement for the specimens stored in the combined test solution, namely 50 g/ l Na2SO4 + 50 g/l NaCl. Standard errors on the average are shown for the last measurement.

65

Looking at Fig. 4, it seems that all the specimens undergo a mass increase during the immersion period of 616 days. However, after 450 days, the mass of the OPC samples is no longer increasing. After 616 days in the combined solution, the mass of the HSR specimens has increased with 2.5%, which was significantly more (One-way ANOVA and Dunnett’s T3 Post Hoc test, level of significance = 0.05) than the mass change of the OPC, S50 and S70 mixtures, which amounted to 1.4%, 1.0% and 0.5% respectively. Fig. 5 gives an overview of the mass change for the individual mortar mixtures after immersion in the Ref. (SO2 4 ) and Comb. 1 (SO2 4 ) solutions, with indication of the standard errors on the average for the last measurement. The mass of the OPC mortar cubes is clearly influenced by the composition of the test solution. A decrease in mass of 26.5% was measured after 616 days immersion in a single sodium sulphate solution, while an increase of 1% is measured after immersion in the sodium sulphate solution with addition of sodium chloride. For the other mixtures, the differences are not that distinct. After 616 days, HSR mortar underwent a small mass decrease when immersed in the single sulphate solution and a small, but statistically significant, mass increase when immersed in the combined sulphate + chloride solution. 3.2.2. Length change The results of the length change measurements are shown in Figs. 6–8. In these graphs, the length change is shown in percent compared to the initial length, cf. Eq. (5). Fig. 6 shows the results obtained for all four mortar mixtures after 497 days immersion in a 50 g/l Na2SO4 solution without addition of chlorides, with indication of the standard errors on the average for the last measurement. As can be seen from the graph in Fig. 6, the length of the OPC specimen was influenced the most. The significant length increase (t-test, level of significance = 0.05) amounted to 0.59% after 497 days in the test solution. Until 140 days of immersion, the differences between the mixtures were negligible. From that

Fig. 5. Mass change per mixture as a function of the immersion time in a 50 g/l Na2SO4 and in a combined 50 g/l Na2SO4 + 50 g/l NaCl solution. (Remark: the scale of the ‘Mass change’-axis is different for the OPC-graph.)

66

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

combined solution, the length change is stabilised for HSR while it still increases in the single sulphate solution, which means that the length change was exactly the same for the HSR specimen in both solutions.

4. Discussion 4.1. Influence of sulphates on chloride attack

Fig. 6. Length change as a function of the immersion time in a 50 g/l Na2SO4 solution.

immersion period on, a clear length increase was found for OPC and HSR mortar, with a big increase for OPC between 300 and 497 days. After 497 days, the length change for S50 and S70 stayed stable, 0.04% and 0.02%, while it slightly increased for HSR, 0.13%. The graph in Fig. 7 shows the length change results after immersion in a combined solution with 50 g/l Na2SO4 and 50 g/l NaCl, with indication of the standard errors on the average for the last measurement. The results for the length change measurements after immersion in the combined sulphate and chloride solution are comparable to the results found after immersion in a single sulphate solution. Here also the OPC mortar underwent the highest length increase, namely 0.42%, followed by the HSR mortar and the BFS mixtures. Also after immersion in a sulphate solution with addition of chlorides, the BFS samples underwent almost no length change. Fig. 8 gives an overview of the mass change for the individual mortar mixtures after immersion in the Ref. (SO2 4 ) and Comb. 1 (SO2 4 ) solutions, with indication of the standard errors on the average for the last measurement. The length increases for the OPC prisms after immersion in the combined solution were in the same order of magnitude as in the single sulphate solution until 300 days. After 497 days of immersion, the length increase in the combined solution was clearly smaller than in the single sulphate solution. Besides, it seemed that the HSR concrete is less resistant to expansion when it is immersed in a 50 g/l Na2SO4 + 50 g/l NaCl solution. The length increase after immersion in the combined solution amounted to 0.15% after 300 days of immersion compared to 0.10% in the single sulphate solution. Notwithstanding, between 300 days and 497 days in the

Fig. 7. Length change as a function of the immersion time in a combined 50 g/l Na2SO4 + 50 g/l NaCl solution.

In general, BFS concrete has the highest resistance against chloride penetration. The chloride penetration fronts as well as the diffusion coefficients are the smallest for these mixtures, regardless the sulphate concentration in the combined solutions. This is in accordance with the general assumption, namely that the replacement of Ordinary Portland Cement by BFS results in an increasing resistance against chloride penetration [20]. Concerning the influence of sulphates on chloride penetration in concrete, the main trends obtained by measuring the chloride penetration depth are in accordance with the main trends from the diffusion coefficients. Based on the diffusion coefficients obtained at the age of 28 days (start of the immersion period), it seems that the free chloride diffusion increases when sulphates are added to the 165 g/l NaCl solution, especially when the sulphate content amounts to 55 g/l Na2SO4. Nevertheless, when the specimens were immersed at the age of 84 days, the diffusion coefficients remained constant or decreased slightly when the sulphate content increased. A possible explanation for this phenomenon could be the on-going hydration and the formation of expansive reaction products (e.g. ettringite) leading to a densification of the matrix which makes it more difficult for the chloride and sulphate ions to penetrate. Because of this, the influence of sulphates on chloride penetration becomes inferior compared to the densification of the matrix. On the other hand, the trends found for the chloride penetration depths, measured by means of the colorimetric method, are quite similar at the different testing ages. At the age of 28 days as well as at the age of 84 days, after 7 and 14 weeks of immersion, the chloride penetration depth increased significantly for OPC and BFS concrete (except at 84 days after 7 weeks immersion) when Na2SO4 was added to a NaCl solution. On the other hand, for HSR concrete, the chloride penetration depth slightly decreased when the sulphate content increased. So, it can be determined that a sulphate content of 27.5 g/l Na2SO4 added to a 165 g/l NaCl solution has almost no influence on the chloride penetration depth. Meanwhile a sulphate content of 55 g/l Na2SO4 influences the chloride penetration depth significantly when considering the effect of the binder type. At low sulphate concentrations it is clear that HSR shows the highest chloride penetration depth. However, after a 7 weeks immersion period in the combined solution with 55 g/l Na2SO4, it is remarkable that the chloride penetration in the OPC concrete is slightly higher than in HSR concrete. Since the C3A content of HSR is already very low from the beginning, there is no competition between Cl and SO2 4 to bind the C3A. So, chlorides can penetrate the concrete unhindered. The same phenomenon is found for OPC when high amounts of sulphates are added to the chloride solution. A part of the C3A binds with the sulphates and the amount of free chlorides increases. This means that the adverse/harmful effect of HSR compared to OPC, in chloride containing environments can disappear when high sulphate contents are present. Nevertheless, this phenomenon is not observed when the concrete is immersed for 14 weeks. Overall, the diffusion coefficients found at the age of 28 days and the chloride penetration depths at 28 days as well as at 84 days confirm the findings of Al-Amoudi et al. [50]. They found that the concentration of free chlorides in the pore solution

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

67

Fig. 8. Length change per mixture as a function of the immersion time in a 50 g/l Na2SO4 solution and a combined 50 g/l Na2SO4 + 50 g/l NaCl solution.

increases significantly when sulphates were present in the chloride solutions. This finding can be easily explained by the fact that a part of the C3A will preferentially bind with sulphates. Also the fact that Friedel’s salt will convert to ettringite when sodium sulphate is present [22] leads to an increase in free chlorides (and a higher chloride diffusion coefficient). Another factor enhancing the free chloride diffusion, and as a consequence the chloride penetration depth, is the increase in alkalinity of the pore solution due to the addition of sodium sulphate to the chloride solution, which inhibits chloride binding [35 cited in 34]. However, the obtained results are opposite to the conclusions of Zuquan et al. [36]. Although, it should be noted that the test results of Zuquan et al. [36] are obtained after 90 days, 250 days and 400 days immersion, which is longer than in current paper (49 days to 98 days). They conclude that the presence of sulphates reduces the chloride diffusion coefficient and the chloride concentration by 30–60% at early exposure periods. They also attribute this to gradual formation of ettringite which leads to a compacted microstructure and decreases the ingress of chlorides. Overall, according to the diffusion coefficients obtained in this paper, the conclusions of Zuquan et al. [36] should be nuanced since after very short exposure periods, namely 49 days (7 weeks) to 98 days (14 weeks), the presence of sulphates lead to an increase in chloride penetration. In addition, a XRD analysis was performed on cement paste samples immersed in a single 165 g/l NaCl solution to identify the reaction products. The XRD profile, as can be seen in Fig. 9, shows peaks referring to the presence of ettringite, as well as Friedel’s salt. The observed ettringite formation is not caused by sulphate attack since there was no external sulphate source present. The ettringite is formed during the hydration. Notwithstanding the fact that ettringite is an expansive reaction product, it will not cause deterioration during hydration. The ettringite fraction is also much smaller than when sulphate attack is considered. More important in this case, immersion in a single chloride solution is the presence

of Friedel’s salt. This indicates the binding of penetrated chlorides. Since only free chlorides will initiate corrosion, chloride binding is an important issue. From Fig. 9 it is clear that an increase in BFS content results in a (small) increase in Friedel’s salt formation/ chloride binding. However, this is not in accordance with former results described in Maes et al. [20], where it was found that chloride binding in BFS concrete is lower than in OPC concrete. Although, in that paper, chloride binding was investigated by measuring chloride profiles only. XRD- and Rietveld results concerning combined attack are shown in Section 4.2 of this paper. 4.2. Influence of chlorides on sulphate attack Overall, OPC mortar shows the lowest resistance against sodium sulphate attack. The mass loss as well as the length increase was the highest for OPC. This is in accordance with previous research [29]. Next, from the obtained results it seems that the performance of BFS mortar with 50% and 70% cement replacement in sodium sulphate rich environments is better than that of HSR mortar. Nevertheless, the resistance of HSR mortar in sulphate containing environments is much better than that of OPC mixtures. The influence of chlorides on sulphate attack is the most clear for the mass changes of the OPC mortar. When OPC mortar is immersed for 616 days in a single sodium sulphate solution, more than a quarter of the initial mass of the specimen is lost due to cracking and spalling of the outermost layers. When OPC samples are immersed in a combined solution of sodium sulphate and sodium chloride for the same period, no mass is lost. Notwithstanding, the mass of OPC mortar stored in the combined solution tends to decrease again after 450 days of immersion. However, not in the same order of magnitude as the decrease in the single sulphate solution. In general, it can be concluded that the presence of chloride ions in a sulphate rich environment has a mitigating effect on the mass loss of OPC mortar due to sodium sulphate

68

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

OPC Ettringite

5

10

S50 Friedel's salt

15

20

S70 Gypsum

25

30

35

°2θ Reaction product Gypsum Ettringite Friedel’s salt

OPC 1.22 15.63 14.17

Amount [%] S50 2.9 10.32 17.65

S70 8.85 18.94

Fig. 9. XRD-profiles measured for OPC, S50 and S70 after 6 months immersion in a 165 g/l NaCl solution + Quantitative Rietveld analysis after XRD-measurements. The amount of reaction products is compared to the internal standard (%).

attack. This finding is in accordance with the findings of Al-Amoudi [34]. Nevertheless, the mitigating effect is rather a delaying effect since it seems that the deterioration will occur at a later time. This statement is in accordance with the conclusions of Zuquan et al. [36]. After 616 days in the solution the specimens were also visually analysed, as tabulated in Table 6. The visual inspection of the specimens confirms the measurements described in Section 3.2 of this paper. The OPC samples immersed in the 50 g/l Na2SO4 solution are clearly deteriorated.

The visual deterioration of the OPC specimens stored in the combined solutions as well as HSR mortar stored in the single sulphate solutions is rather negligible, however some small cracks were observed. Besides, neither the HSR samples stored in the combined solution nor the BFS samples show deterioration. Concerning the length change, it seems that there is no obvious effect of the chlorides on the sulphate attack mechanism until 300 days of immersion, regardless the mortar type. However, after more than 300 days of immersion, namely 497 days, it becomes clear that the length of the OPC specimens in the single sulphate

Table 6 Deterioration at the end surfaces of mortar cubes, used for mass change measurements, exposed to sulphate and sulphate + chloride solutions for 616 days. Mix OPC

HSR

S50

S70

50 g/l Na2SO4

50 g/l Na2SO4 + 50 g/l NaCl

69

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

Table 7 Deterioration at the end surfaces of mortar prisms, used for length change measurements, exposed to sulphate and sulphate + chloride solutions for 300 days and 497 days. Mix

300 days immersion 50 g/l Na2SO4

497 days immersion 50 g/l Na2SO4 + 50 g/l NaCl

50 g/l Na2SO4

50 g/l Na2SO4 + 50 g/l NaCl

OPC

HSR

S50

S70

Cracking + crumbling

Longitudinal cracks

Transverse cracks Fig. 10. Crack types observed in OPC specimen after 497 days of immersion in a 50 g/l Na2SO4 solution or in a 50 g/l Na2SO4 + 50 g/l NaCl solution.

solution increases due to sulphate attack and that chlorides have a mitigating effect on this. These findings are also translated into more cracks in the specimen, as can be seen in Table 7. The OPC samples are already clearly cracked at the edges after a storage period of 300 days in the single sulphate solution as well as in the combined solution. However, no big differences were measured by length change nor observed by visual inspection. After 497 days of immersion, the amount of cracks is increased and transverse and longitudinal cracks are observed along the whole length, as can be seen in Fig. 10. Furthermore, the edges start to crumble. The samples stored in the single sodium sulphate solutions show more cracks compared to the samples stored in the combined solution. Besides, also the HSR samples show some small deterioration at the edges. Nevertheless, the observed

deterioration does not extend after 300 days of immersion, it stays quite stable. This observation is in accordance with the length change measurements. In contrast, the BFS samples do not show any deterioration after any immersion time. Based on these findings the length change measurements are in accordance with the mass change measurements. In both cases OPC specimens show a bad resistance to sulphate attack when stored for more than 300 days. Besides, the influence of the chlorides on the mass loss and length change due to sodium sulphate attack becomes visible for OPC after more than 300 days immersion. The chlorides have no influence on sulphate attack in BFS mortar since no deterioration was observed after immersion in the single sulphate solution nor after immersion in the combined solution.

70

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

OPC Ettringite

5

10

S50 Friedel's salt

15

20

S70 Gypsum

25

30

35

°2θ Reaction product Gypsum Ettringite Friedel’s salt

Amount [%] S50 4.07 25.85 -

OPC 18.38 33.99 -

S70 3.23 (2.62) -

Fig. 11. XRD-profiles measured for OPC, S50 and S70 after 6 months immersion in a 50 g/l Na2SO4 – solution. + Quantitative Rietveld analysis after XRD-measurements. The amount of reaction products is compared to the internal standard (%).

OPC Ettringite

5

10

S50 Friedel's salt

15

20

S70 Gypsum

25

30

35

°2θ Reaction product Gypsum Ettringite Friedel’s salt

OPC 1.8 17.9 15.2

Amount [%] S50 4.6 22.7 19.2

S70 5.9 26.2 22.8

Fig. 12. XRD-profiles measured for OPC, S50 and S70 after 6 months immersion in a combined solution. + Quantitative Rietveld analysis after XRD-measurements for OPC, S50 and S70 specimen immersed in a combined Na2SO4 and NaCl solution for 6 months. The amount of reaction products is compared to the internal standard (%).

In order to explain these findings by means of formation of reaction products, some XRD-measurements were performed on OPC, S50 and S70 cement paste samples. The cement paste samples were stored for 6 months in solutions similar to the test solutions, namely a 50 g/l Na2SO4 solution and a combined solution of 50 g/l Na2SO4 and 50 g/l NaCl. The XRD-profiles are shown in Figs. 11 and

12. The peaks that indicate ettringite, gypsum and Friedel’s salt are highlighted in the profiles since these phases are assumed to be the main reaction products in sulphate and chloride rich environments, together with gypsum. The amounts, quantified by Rietveld analysis for three phases in particular, namely ettringite, Friedel’s salt and gypsum are shown in a table underneath the graphs.

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

From the profiles obtained by analysing the cement paste from a single sulphate solution, as shown in Fig. 11, it is clear that ettringite is present. The peaks referring to ettringite in the OPC samples are more distinct than in the BFS samples. Quantitative Rietveld analysis shows that, next to ettringite, also a large fraction of gypsum is explicitly present in the OPC paste. Besides, in the S50 sample a major fraction of ettringite is found while almost no gypsum is present. In the S70 sample (almost) none of both phases are found. Significantly more gypsum and ettringite have formed in the OPC mixture compared to the BFS mixtures. According to the small peaks in the profile, small fractions of Friedel’s salt seem to be present as well. However, after the quantitative Rietveld analysis, no Friedel’s salt was found. The total fraction of expansive reaction products amounts to 52.37% for OPC, 29.92% for S50 and 5.85% for S70. The large fraction of expansive reaction products for OPC can cause the mass and length increase. After formation of these reaction products cracks appear, which results in spalling and crumbling leading to mass loss (see Tables 6 and 7). No Friedel’s salt is formed in these cement pastes, which is obvious since they were not exposed to chlorides. These findings provide an explanation for the mass and length change results and for the observed deterioration after immersion in a 50 g/l Na2SO4. In the end, some cement paste samples were subjected to a XRD-analysis after immersion in combined test solutions. In the profiles, clear peaks belonging to ettringite as well as to Friedel’s salt were found, as shown in Fig. 12. Ettringite as well as Friedel’s salt represent a large fraction of the crystalline phases. The amount of gypsum of the samples immersed in the combined solution is rather small compared to the amount of gypsum found in the OPC samples after immersion in a single sulphate solution. In general, the amount of Friedel’s salt increases when the slag content increases. This was also found for the samples immersed in a single chloride solution, even the amounts are in the same order of magnitude. According to these results, it can be stated that sulphates have only limited influence on chloride binding. However, based on the diffusion test and the chloride profiles it was clear that chloride penetration slightly increases when sulphates are present in the solution. Based on the results described in Section 3.2 it can be concluded that chlorides will bind less when sulphates are present since the free chloride concentrations increases in combined solutions (see results ccb). On the other hand, the ettringite fraction in OPC samples is clearly influenced by the presence of chlorides in the solution compared to a pure sulphate solution. The formation of ettringite and gypsum in OPC samples decreases, namely from 33.99% and 18.38% in the 50 g/l Na2SO4 solution to 17.9% and 1.8% in the 50 g/l Na2SO4 + 50 g/l NaCl solution. The amount of ettringite in the S50 samples immersed in the single sulphate and the combined sulphate chloride solution are comparable. Overall, the reciprocal influence of sulphate and chloride ions on the sulphate binding behaviour in BFS concrete is negligible. Nevertheless, for the S70 samples a large increase is measured in the ettringite content compared to the content measured in the samples immersed in the single sulphate. It is not clear how this increase can be explained. Probably, the value found after immersion in the single sulphate solution should be neglected. The link between the amount of reaction products and the deterioration, which was not observed, in BFS concrete is not immediately clear since it is generally assumed that a higher amount of ettringite results in a higher degree of deterioration. However, based on the results in this paper the opposite is found, the high ettringite fraction does not lead to deterioration. As

71

described by Kunther et al. [51], deterioration due to expansion of the reaction products will only occur when there is a coexistence of ettringite and gypsum. Next to the volume increase, which is the prerequisite for expansion, also supersaturation in the solution and force exerted by the formed minerals are decisive. In this paper, it is clear that most deterioration is observed for the samples where ettringite and gypsum coexist, namely the OPC samples in the single sulphate solution. In all the samples immersed in the combined solution the amounts of ettringite are rather high, although, the gypsum content is small. This could explain why almost no deterioration is found for samples immersed in the combined sulphate and chloride solution. 5. Conclusions  Free chloride penetration in Ordinary Portland Cement concrete increases when the sodium sulphate content in the chloride containing environment increases, regardless the age of the concrete. So, the presence of sodium sulphate in a chloride solution aggravates the penetration of free chlorides. Nevertheless, this conclusion is only valid for immersion periods between 7 weeks and 14 weeks. In high sulphate resistant concrete the chloride penetration depth stays stable or decreases slightly when sodium sulphate and sodium chloride are both present in a solution. After a hort immersion period (7 weeks) in a chloride solution with high sodium sulphate content, the chloride penetration depth is even lower than in Ordinary Portland Cement concrete. This beneficial effect of high sulphate resistant concrete disappears when the concrete is immersed for a longer period (14 weeks). Concerning the influence of sodium sulphate on the chloride diffusion coefficient, it can be concluded that the diffusion coefficient increases when the sodium sulphate content of the chloride solution increases. Nevertheless, when immersion starts at 84 days, the diffusion coefficient does not change significantly.  Chlorides have a mitigating effect on sodium sulphate attack. Sulphate deterioration is delayed by the presence of chlorides. Both, sulphates and chlorides will bind with the C3A hydration products to form ettringite and gypsum on the one hand and Friedel’s salt on the other hand. However, chlorides’ reaction product, Friedel’s salt, is not stable in the presence of sodium sulphate solution [22]. So, it is assumed that over time Friedel’s salt will disappear, more ettringite and gypsum will form and deterioration will occur.  When mortar is exposed to a sodium sulphate and chloride containing environment, a large fraction of ettringite is present, however, almost no deterioration is observed. This can be attributed to the small amount of gypsum. Deterioration due to sodium sulphate attack only occurs when ettringite and gypsum are found together. So, the finding of Kunther et al. [51] is confirmed by the results in this paper. In specimens exposed to a single sodium sulphate environment, where deterioration is observed, both ettringite and gypsum are present in large amounts.  In general, replacement of Ordinary Portland Cement by BlastFurnace Slag improves the resistance of concrete/mortar against chloride penetration and sodium sulphate attack. Concerning the influence of sulphates on chloride attack in BFS concrete, the trends are similar to the trends found for ordinary Portland cement concrete, namely an increase in chloride penetration when the sodium sulphate content in the combined solution increases. Besides, chlorides have no influence (neither positive nor negative) on sodium sulphate attack in BFS mortar. BFS mortar already possesses a high resistance against sodium sulphate

72

M. Maes, N. De Belie / Cement & Concrete Composites 53 (2014) 59–72

attack, since no significant mass change or expansion is observed when immersed in a pure sodium sulphate solution.

Acknowledgements Research funded by a Ph.D. grant of the Agency for Innovation by Science and Technology (IWT). The authors would like to thank Philip Van den Heede and Hugo Eguez Alava for critically reading the manuscript. References [1] Song H-W, Pack S-W, Ann KY. Probabilistic assessment to predict the time to corrosion of steel in reinforced concrete tunnel box exposed to sea water. Constr Build Mater 2009;23(10):3270–8. [2] Kwon SJ, Na UJ, Park SS, Jung SH. Service life prediction of concrete wharves with early-aged crack: probabilistic approach for chloride diffusion. Struct Saf 2009;31(1):75–83. [3] Sibbick T, Fenn D, Crammond N. The occurrence of thaumasite as a product of seawater attack. Cem Concr Compos 2003;25(8):1059–66. [4] Costa A, Appleton J. Chloride penetration into concrete in marine environment – Part II: Prediction of long term chloride penetration. Mater Struct 1999;32(219):354–9. [5] Costa A, Appleton J. Chloride penetration into concrete in marine environment – Part I: Main parameters affecting chloride penetration. Mater Struct 1999;32(218):252–9. [6] Pack SW, Jung MS, Song HW, Kim SH, Ann KY. Prediction of time dependent chloride transport in concrete structures exposed to a marine environment. Cem Concr Res 2010;40(2):302–12. [7] Lindvall A. Chloride ingress data from field and laboratory exposure – influence of salinity and temperature. Cem Concr Compos 2007;29(2):88–93. [8] Castro P, De Rincon OT, Pazini EJ. Interpretation of chloride profiles from concrete exposed to tropical marine environments. Cem Concr Res 2001;31(4):529–37. [9] Glasser FP, Marchand J, Samson E. Durability of concrete – degradation phenomena involving detrimental chemical reactions. Cem Concr Res 2008;38(2):226–46. [10] Delagrave A, Marchand J, Ollivier JP, Julien S, Hazrati K. Chloride binding capacity of various hydrated cement paste systems. Adv Cem Based Mater 1997;6(1):28–35. [11] Izquierdo D, Alonso C, Andrade C, Castellote M. Potentiostatic determination of chloride threshold values for rebar depassivation – experimental and statistical study. Electrochim Acta 2004;49(17–18):2731–9. [12] Yuan Q. Fundamental studies on test methods for the transport of chloride ions in cementitious materials doctoral thesis. Ghent: Ghent University; 2009. [13] Tang L. Chloride transport in concrete – measurement and prediction doctoral thesis. Göteborg: Chalmers University of Technology; 1996. [14] De Schutter G. Hydration and temperature development of concrete made with blast-furnace slag cement. Cem Concr Res 1999;29(1):143–9. [15] Luo R, Cai YB, Wang CY, Huang XM. Study of chloride binding and diffusion in GGBS concrete. Cem Concr Res 2003;33(1):1–7. [16] Bleszynski R, Hooton RD, Thomas MDA, Rogers CA. Durability of ternary blend concretes with silica fume and blast-furnace slag: laboratory and outdoor exposure site studies. ACI Mater J 2002;99(5):499–508. [17] Leng FG, Feng NQ, Lu XY. An experimental study on the properties of resistance to diffusion of chloride ions of fly ash and blast furnace slag concrete. Cem Concr Res 2000;30(6):989–92. [18] Angst U, Elsener B, Larsen CK, Vennesland O. Critical chloride content in reinforced concrete – a review. Cem Concr Res 2009;39(12):1122–38. [19] Rajamane NP, Peter JA, Dattatreya JK, Neelamegam M, Gopalakrishnan S. Improvement in properties of high performance concrete with partial replacement of cement by ground granulated blast furnace slag. J Inst Eng India Civil Eng Div 2003;84(Mai):38–42. [20] Maes M, Gruyaert E, De Belie N. Resistance of concrete with blast-furnace slag against chlorides, investigated by comparing chloride profiles after migration and diffusion. Mater Struct 2013;46(1–2):89–103. [21] Al-Amoudi OSB. Sulfate attack and reinforcement corrosion in plain and blended cements exposed to sulfate environments. Build Environ 1998;33(1):53–61. [22] Brown PW, Badger S. The distributions of bound sulfates and chlorides in concrete subjected to mixed NaCl, MgSO4, Na2SO4 attack. Cem Concr Res 2000;30(10):1535–42. [23] Brown PW, Hooton RD, Clark BA. The co-existence of thaumasite and ettringite in concrete exposed to magnesium sulfate, at room temperature and the influence of blast-furnace slag substitution on sulfate resistance. Cem Concr Compos 2003;25(8):939–45.

[24] Sahu S, Badger S, Thaulow N. Evidence of thaumasite formation in Southern California concrete. Cem Concr Compos 2002;24(3–4):379–84. [25] Sahu S, Badger S, Thaulow N. Mechanism of thaumasite formation in concrete slabs on grade in Southern California. Cem Concr Compos 2003;25(8):889–97. [26] Akoz F, Turker F, Koral S, Yuzer N. Effects of raised temperature of sulfate solutions on the sulfate resistance of mortars with and without silica fume. Cem Concr Res 1999;29(4):537–44. [27] Matsumura T, Shirai K, Saegusa T. Verification method for durability of reinforced concrete structures subjected to salt attack under high temperature conditions. Nucl Eng Des 2008;238(5):1181–8. [28] Otsuki N, Madlangbayan MS, Nishida T, Saito T, Baccay MA. Temperature dependency of chloride induced corrosion in concrete. J Adv Concr Technol 2009;7(1):41–50. [29] Al-Dulaijan SU, Maslehuddin M, Al-Zahrani MM, Sharif AM, Shameem M, Ibrahim M. Sulfate resistance of plain and blended cements exposed to varying concentrations of sodium sulfate. Cem Concr Compos 2003;25(4–5):429–37. [30] Rasheeduzzafar, Alamoudi OSB, Abduljauwad SN, Maslehuddin M. Magnesium–sodium sulfate attack in plain and blended cements. J Mater Civil Eng 1994;6(2):201–22. [31] Alamoudi OSB, Rasheeduzzafar, Maslehuddin M, Abduljauwad SN. Influence of chloride-ions on sulfate deterioration in plain and blended cements. Mag Concr Res 1994;46(167):113–23. [32] Gruyaert E, Van den Heede P, Maes M, De Belie N. Investigation of the influence of blast-furnace slag on the resistance of concrete against organic acid or sulphate attack by means of accelerated degradation tests. Cem Concr Res 2012;42(1):173–85. [33] Liu Z. Study of the basic mechanisms of sulfate attack on cementitious materials doctoral thesis. Ghent: Ghent University; 2010. [34] Al Amoudi OSB OSB, Maslehuddin M, Abdul-Al YAB. Role of chloride ions on expansion and strength reduction in plain and blended cements in sulfate environments. Constr Build Mater 1995;9(1):25–33. [35] Abdalkader AHM, Lynsdale CJ, Cripps JC. The effect of chloride on performance of cement mortars subjected to sulfate exposure at low temperature. In: SCMT 2013–3rd international conference on sustainable construction materials and technology. Kyoto, Japan 2013. [36] Zuquan J, Wei S, Yunsheng Z, Jinyang J, Jianzhong L. Interaction between sulfate and chloride solution attack of concretes with and without fly ash. Cem Concr Res 2007;37(8):1223–32. [37] De Weerdt K, Geiker MR. Changes in the phase assemblage of concrete exposed to seawater – case study. In: Jacobsen S, Justnes H, editors. Proceeding of the proceedings of the international congress on durability of concrete (ICDC 2012), 18–21 June 2012, Trondheim, Norway. Trondheim, Norway: NTNU Trondheim; 2012. [38] 206-1 NE. Concrete – Part 1. Specification, Performance, Production and Conformity. Brussels, Belgium: European committee for standardization; 2000. [39] NBN EN 12350-2. Slump-test. Testing fresh concrete. Brussels, Belgium: European committee for standardization; 2009. [40] NBN EN 12350-7. Air content – pressure methods. Testing fresh concrete. Brussels, Belgium: European committee for standardization; 2009. [41] NBN EN 196–2. Chemical analysis of cement. Methods of testing cement Brussels, Belgium: European committee for standardization; 2005. [42] NBN EN 15167-1. Ground granulated blast furnace slag for use in concrete, mortar and grout. Part 1: Definitions, specifications and conformity criteria. Brussels, Belgium: EUROPEAN COMMITTEE FOR STANDARDIZATION; 2007. [43] NT Build 443. Accelerated Chloride Penetration. Concrete Hardened. Espoo, Finland: NordTest; 1995. [44] Maes M, Caspeele R, Van den Heede P, De Belie N. Influence of sulphates on chloride diffusion and the effect of this on service life prediction of concrete in a submerged marine environment. In: Strauss A, Frangopol DM, Bergmeister K, editors. 3rd International symposium on Life-cycle Civil Engineering: Lifecycle and sustainability of civil infrastructure systems. Vienna, Austria; 2012. [45] International A. C 1012-04: Standard test method for length change of hydraulic-cement mortars exposed to a sulfate solution; 2004. p. 6. [46] Bish DL, Howard SA. Quantitative phase-analysis using the Rietveld method. J Appl Crystallogr 1988;21:86–91. [47] Martin-Marquez J, De la Torre AG, Aranda MAG, Rincon JM, Romero M. Evolution with temperature of crystalline and amorphous phases in porcelain stoneware. J Am Ceram Soc 2009;92(1):229–34. [48] Snellings R, De Schepper M, De Buysser K, Van Driessche I, De Belie N. Clinkering reactions during firing of recyclable concrete. J Am Ceram Soc 2012;95(5):1741–9. [49] Coelho AA. Topas Academic Version 4.1, 2007. , March 16; 2012. [50] Al-Amoudi OSB, Rasheeduzzafar, Maslehuddin M, Abduljauwad SN. Influence of sulfate ions on chloride-induced reinforcement corrosion in portland and blended cement concretes. Cem Concr Aggr 1994;16(1):3–11. [51] Kunther W, Lothenbach B, Scrivener K. Influence of bicarbonate ions on the deterioration of mortar bars in sulfate solutions. Cem Concr Res 2013;44:77–86.