Cet article présente une brève revue de la littérature concernant les essais ayant été effectués par le passé visant à déterminer la contribution en alcalis des granulats à la solution interstitielle du béton. Par la suite, une analyse du contenu en alcalis d’échantillons de béton et mortier scellés (de façon à éliminer tout échange avec l’extérieur) confectionnés à l’aide de différents granulats est effectuée de façon à déterminer la contribution réelle en alcalis de ces derniers à travers le temps (série d’échantillons scellés).
Six différents granulats, dont le contenu en alcalis diffère les uns des autres, ont été utilisés dans la fabrication d’échantillons de mortier et de béton. Ces derniers ont par la suite été scellés de façon à éviter tout lessivage d’alcalis et ont été placés à des conditions de température contrôlées (38°C et 60°C). Après 28, 91, 168 et 365 jours, les échantillons de béton et de mortier ont été soumis à deux essais d’extraction de façon à déterminer le contenu en ions sodium et potassium de la solution interstitielle. Parallèlement, les granulats étudiés ont également été immergés dans des solutions alcalines (0,7N KOH pour mesurer la libération du sodium et 0,7N NaOH pour mesurer la libération du potassium). Après 14, 42, 91, 182 et 365 jours, les solutions ont été prélevées et analysées de façon à mesurer la quantité d’alcalis lessivés par les granulats en milieu fortement basique. Les résultats ont par la suite été comparés entre eux afin de déterminer la contribution réelle en alcalis des granulats à la solution interstitielle du béton.
4.2 Article 2:
Effect of alkali release by aggregates on alkali-silica reactionCe deuxième article traite également de la contribution en alcalis des granulats, mais cette fois en faisant intervenir le développement de la réaction alcalis-silice. Ainsi, des
échantillons de béton et de mortier contenant des granulats réactifs ont été placés en expansion libre et leur contenu en alcalis a été mesuré à travers le temps de façon à vérifier l’effet d’une libération d’alcalis des granulats à la solution interstitielle sur le développement de l’expansion dans le temps, ainsi que le potentiel de liaison de ces ions dans la formation du gel de silice.
Deux granulats réactifs dont les teneurs en alcalis diffèrent et un granulat de référence (également utilisés lors de la réalisation du premier article) ont été utilisés dans la fabrication d’échantillons de mortier et de béton. Les échantillons ont été placés à deux différentes températures (38°C et 60°C), au-dessus d’une petite quantité bien déterminée d’eau distillée, et ont été mesurés dans le temps de façon à déterminer de façon précise l’évolution de l’expansion. Le contenu en alcalis des différents échantillons a été déterminé après 28, 91, 168 et 364 jours. Les résultats obtenus des échantillons fabriqués avec le granulat de référence, étant non réactif et ne contenant pas d’alcalis, ont ensuite été comparés avec ceux obtenus des échantillons contenant les granulats réactifs afin d’établir un bilan d’alcalis complet dans la solution interstitielle, c’est-à-dire la quantité d’alcalis libérée par le ciment dans la solution, la quantité d’alcalis lessivée lors de la conservation des échantillons dans des conditions d’humidité élevées ainsi que la quantité d’alcalis chimiquement liée dans la formation du gel de silice. Des liens ont par la suite été réalisés avec l’expansion des échantillons de façon à déterminer si la libération d’alcalis par les granulats dans le temps influence la progression de l’expansion.
5
Article 1:
VALIDATION OF THE ALKALI CONTRIBUTION BY AGGREGATES TO THE CONCRETE PORE SOLUTIONCédric Drolet, Josée Duchesne, Benoît Fournier Department of Geology and Geological Engineering
Laval University, Québec, Canada, G1V 0A6
ABSTRACT
Mortar and concrete specimens were made with six reactive and non-reactive aggregates from Eastern Canada, with different alkali contents, to determine their alkali contribution to the pore solution. Two methods were used to determine the alkali content of samples over time: the high pressure extraction method and the espresso extraction method. Also, an alkali leaching test was conducted directly on the same aggregate materials. One aggregate was clearly identified as a potential source of Na+ to the pore solution. Results obtained after one year at 60°C showed that rich alkali-bearing aggregates, which can be identified in an alkali leaching test by immersion in alkaline solutions, can release up to 1.75 kg Na2O/m3 of concrete if used as fine fraction and up to 0.24 kg Na2O/m3 of concrete if used as coarse fraction. However, no clear K+ contribution from aggregate was observed.
Keywords: C: Alkali-aggregate reaction; D: Alkalis; B: Pore solution
RÉSUMÉ
Des échantillons de mortier et de béton ont été créés avec six granulats de différentes teneurs en alcalis, réactifs et non-réactifs, provenant de l’est du Canada, en vue de déterminer la contribution en alcalis de ces derniers à la solution interstitielle de bétons et de mortiers. Deux méthodes ont été utilisées pour déterminer la teneur en alcalis des échantillons à travers le temps : l’extraction sous haute pression et l’extraction « Espresso ». Aussi, un essai d’extraction d’alcalis par lessivage a également été conduit directement sur les granulats. Un granulat a été clairement identifié comme source potentielle en Na+ à la solution interstitielle. Les résultats ont montré que certains granulats riches en alcalis, qui peuvent être identifiés par l’essai de lessivage, peuvent relâcher jusqu’à 1,75 kg Na2O/m3 de béton lorsqu’utilisés comme fraction fine et jusqu’à
0,24 kg Na2O/m3 de béton lorsqu’utilisés comme fraction grossière. Par contre, aucune contribution claire en ion K+ des granulats à la solution interstitielle n’a pu être observée.
1 INTRODUCTION
Alkali-silica reaction (ASR) is a deleterious reaction affecting many concrete structures around the world, leading to excessive cracking, serviceability issues and reduction of the structure durability. This reaction, which involves unstable siliceous phases contained in fine and/or coarse aggregates, needs three essentials conditions to take place: (1), the aggregate is potentially reactive, (2), the concrete pore solution has a high alkali concentration, and (3), the concrete is exposed to high humidity conditions (Thomas et al., 2013). Generally, the higher the alkali concentration in the concrete pore solution, the greater the expansion due to ASR; in fact, (OH)- anionsare released into the pore solution from portlandite (Ca(OH)2) to establish equilibrium with Na+ and K+ cations. The high pH of the pore solution leads to the dissolution of metastable, amorphous, poorly crystalline or micro/crypto-crystalline forms of silica contained in some types of aggregates. The silica now in solution reacts with alkalis and calcium cations to form a gel, which imbibes water and increases in volume, leading to cracking of concrete (Thomas et al., 2013).
Stanton (1940), who identified the reaction in the late 30’s, demonstrated that using low alkali cement can prevent ASR. However, some later studies showed that this preventive method has not been effective in reducing excessive expansion with some types of aggregates (Stark, 1978 ; Macdonald et al., 2012). Although it was found that the alkali threshold above which deleterious ASR expansion occurs indeed varies from one aggregate to another, these observations also suggested that the cement was likely not the only source of alkalis to the pore solution to promote ASR in concrete.
2 CEMENT ALKALI CONTRIBUTION AND EFFECT OF HYDRATION ON THE COMPOSITION OF THE PORE SOLUTION
The most important source of alkalis to the pore solution of concrete is the portland cement. Indeed, alkalis occur in clinker in various phases. If sulfates are available in the
kiln, alkalis tend to form potassium sulfate (K2SO4), sodium-potassium sulfate ((K, Na)2SO4) and calcium-potassium sulfate (K2Ca2(SO4)3). If the amount of sulfate is not sufficient to bind all alkalis, the remainders are distributed between the C3A, C2S and, in a smaller proportion, in C3S and C4AF (Jawed & Skalny, 1977). However, not all alkalis in the cement are readily soluble. McCoy & Eshenour (1968) found that the amount of immediately water soluble sodium and potassium vary from 10 to 60% of the total cement alkali content and that even after one year, a considerable proportion of alkalis might not dissolve in the concrete pore solution. The amount of soluble alkalis from the cement strongly depends on the constituents of the clinker. Indeed, alkali sulfates are considered to be readily soluble, while alkalis incorporated in calcium silicate and aluminate clinker phases are considered poorly soluble and might only reach the pore solution when these phases hydrate (Diamond, 1975). On the other hand, when cement hydrates, some alkalis can be bound in the formation of C-S-H gel, depending on the Ca/Si ratio and the alumina content of the C-S-H. Indeed, a lower Ca/Si ratio fosters alkali binding (Hooton et al., 2010 ; Bérubé et al., 1995). Vollpracht et al. (2016) made a review of a great number of experiments aiming at analysing the chemistry of the pore solution of cement pastes conducted over the past few decades; the authors found that, for pastes with a water-to-cement ratio of 0.5, typical alkali concentration varies from 100 to 200 mmol Na+/L and from 400 to 450 mmol K+/L for low alkali cement (0.20-0.31% Na2O, 0.48-0.75% K2O) and from 250 to 350 mmol Na+/L and from 400 to 800 mmol K+/L for high alkali cements (0.32-0.43% Na2O, 1.04-1.30% K2O).
2.1 Alkali contribution by aggregates
Many studies were conducted over the past few decades for evaluating the potential alkali contribution to the concrete pore solution from different types of aggregates. In many cases, aggregates were ground and placed in various types of solutions, at different temperatures and for various periods of time. The solutions were sampled over time and analysed for Na+ and K+ concentrations to monitor the alkali release from aggregates. Table 1 summarizes the experimental conditions, types of aggregates selected and main results obtained by several researchers since the mid 1970’s.
The majority of the tests to assess the alkali contribution by aggregates were conducted using a saturated lime solution Ca(OH)2 (Van Aardt & Visser, 1977a ; Stark & Bhatty, 1986 ; Kawamura et al. 1989 ; Le Roux et al. 1997 ; Lu et al., 2006 ; Wang et al., 2008 ; Soares et al., 2016). However, such a solution causes the precipitation of secondary reaction products over time, when reactive aggregates are selected, thus leading to a reduction in the calcium hydroxide concentration of the immersion solution. This phenomenon can be controlled by adding some solid lime in excess and by agitating the samples (Bérubé & Fournier, 2004). On the other hand, secondary reaction products can incorporate alkalis, thus leading to an underestimation of the potential alkali release by aggregates (Bérubé & Duchesne, 1996).
It is also important to note that the pH of the saturated lime solution has a value of 12.45 at 25ºC, while the pH of the concrete pore solution varies generally from 13.2 to 13.9. In order to obtain a pH similar to the pore solution’s pH, some authors used alkaline solutions, i.e. 0.7N NaOH to measure K release and 0.7N KOH to measure Na release (Bérubé & Duchesne, 1996 ; Ouali, 1997 ; Wang et al., 2008 ; Locati et al., 2010 ; Berra et al., 2014 ; Soares et al., 2016 ; Menéndez et al., 2016). Comparing the results obtained by several authors on a large selection of aggregates using different extraction solutions (Table 1), Bérubé & Fournier (2004) suggested that experiments be conducted on aggregates with particle size similar to that used in concrete, along with an aggregate-to- solution ratio of 4.
Table 1 - Alkali release by aggregates stored in various extraction solutions (adapted/updated from Bérubé & Fournier, 2004)
Authors Material Duration Temperature
(ᵒC)
sample shaking
Aggregate:solution
ratio Mass (g) material size Extraction solution
soluble alkalis (% Na2Oeq)
kg Na2Oeq/m3
of concrete(1) Van-Aardt & Visser (1977) 6 feldspars, 2 clay minerals 250 days 39 no 1:50 0.5 powder 25 ml saturated lime(2) 0.01-2.32 0.37-44.03
Stark & Bhatty (1986)
3 feldspars, 3 sands, 3
gravels, 1 andesite 90 days 38 yes 1:5 5 <80µm 25 ml saturated lime 0.26-1.99 4.81-36.82 3 feldspars, 1 andesite 28 days 80 no 1:5 5 <80µm 25 ml saturated lime 0.83-5.03 15.36-93.06
25 ml distilled water 0.03-0.13 0.56-2.41 Kawamura et al. (1989) 2 andesites 180 days 40 yes 1:2 100 0.15-5 mm 200 ml saturated lime 0.02-0.07 0.37-1.30 Bérubé & Duchesne (1996) 17 different aggregates 578 days 38 yes 1:1 40 1.25-5 mm
40 ml distilled water 0.001-0.151 0.02-2.79 40 ml saturated lime(2) 0.002-0.11 0.037-2.04 40 ml alkaline solution(3) >0.004-0.684 0.07-12.65
Le Roux et al. (1997) 8 different aggregates 7 hours 100 no 1:1 500 0.08-20 mm 500 ml saturated lime(2) 0.0085-0.21 0.16-3.89 Ouali (1997) 7 different aggregates 1000 hours 60 no 1:4,2 120 0.08-0.63 mm 500 ml alkaline solution(3) 0.008-0.47 0.15-8.69
Lu et al. (2006) gneiss, granite, K-feldspar 28 days 80 yes 1:2 25
10-12 mm 50 ml saturated lime(2) 0.006-0.016 0.11-0.296 5-10 mm 50 ml saturated lime(2) 0.007-0.029 0.13-0.54 1.25-5 mm 50 ml saturated lime(2) 0.011-0.037 0.20-0.68 0.63-1.25 mm 50 ml saturated lime(2) 0.015-0.046 0.28-0.85 0.15-0.63 mm 50 ml saturated lime(2) 0.022-0.064 0.41-1.18 0.08-0.15 mm 50 ml saturated lime(2) 0.051-0.112 0.94-2.07 Wang et al. (2008) Nepheline, Alaskite 24 hours 150 no 1:1 unknown 0.15-0.65 mm saturated lime - 0.093-0.118
(4)
alkaline solution(3) - 0.236-0.490(4)
Locati et al. (2010) 2 K-feldspars, 2 Na-
feldspars 24 hours 80 no 1:10? 25 0.15-0.30 mm alkaline solution
(5)
0.009-0.016 0.17-0.30 Berra et al. (2014) 2 aggregates 180 days 80 no 1:1 unknown 0-4 mm alkaline solution
(3)
0.037-0.041 0.68-0.76 4-20 mm alkaline solution(3) 0.011-0.015 0.20-0.28
Soares et al. (2016) 6 granitic aggregates 455 days 38 yes 1:4 unknown
0-2 mm saturated lime 0.024-0.045 0.45-0.83 4.75-9.5 mm saturated lime 0.007-0.010 0.14-0.18 12.5-20 mm saturated lime 0.005-0.007 0.09-0.13 0-2 mm alkaline solution(5) 0.190-0.206 3.51-3.81 4.75-9.5 mm alkaline solution(5) 0.131-0.153 2.43-2.84 12.5-20 mm alkaline solution(5) 0.104-0.144 1.92-2.66 Menéndez et al. (2016) Granodiorite 91 days 60 unknown 1:4 unknown 0-4 mm alkaline solution(3) - 6.29
1. For a concrete containing 1850 kg/m3 of aggregates 2. Saturated lime solution with solid lime in excess
3. 0,7M KOH to measure Na release and 0,7M NaOH to measure K release
4. For a concrete containing 0,1 m3 of interstitial solution of a 0,7 mol/L Na2Oeq concentration by m3 of concrete
Results reported in Table 1 showed without a doubt that alkalis can be released from certain types of aggregates in a high pH environment. However, different experimental conditions induce high variability in the rate and extent of alkali release by aggregates. For instance, increasing alkali release from aggregates is obtained with increasing pH (Bérubé & Duchesne, 1996) and temperature (Ouali, 1997) of the attack solution, as well as increasing specific surface area of the aggregates (Lu et al., 2006). Also, poorly crystalline and micro/crypto-crystalline mineral grains as well as cleavage planes and micro-cracks due to metamorphism or processing of the aggregates lead to a greater alkali release (Broekmans, 2012).
Constantiner & Diamond (2003) studied the alkali release in the pore solution of mortars incorporating an almost pure limestone with partial replacement of ground feldspars of different types (microcline, oligoclase and labradorite), along with the presence or not of alkali-silica reactive fine aggregates. The authors found that feldspars release alkali ions into the pore solution over time, and that the releasing process was continuing steadily even towards the end of the experimental period (9 month). However, when alkali-rich feldspars are combined with ASR reactive aggregates, the alkali release into the pore solution may be masked by incorporation of alkali ions into ASR products.
3 SCOPE AND OBJECTIVES OF WORK
Even if the studies presented previously tried to reproduce the internal environment of concrete, the real behavior of aggregates in concrete is still not known. Indeed, although national standards such as CSA A23.2-27A-2014, RILEM AAR-7.1-2015, ASTM C 1778- 2015 recommend to limit the alkali contributed by the portland cement for preventing ASR in concrete incorporating potentially reactive aggregates, no recommendation is made concerning the potential alkali release by aggregates to the concrete pore solution.
The objective of the present study is to determine the real alkali contribution by aggregates to the pore solution in concrete and mortar specimens. In order to do so, the alkali content of the pore solution was measured over time for mortar and concrete specimens incorporating a selection of reactive and non-reactive aggregates of different alkali
contents. The comparison between the alkalis concentrations measured in the pore solution of the different combinations should permit to isolate the aggregate alkali contribution.
4 MATERIALS AND EXPERIMENTAL PROCEDURES 4.1 Materials
Six reactive and non-reactive aggregates from Eastern Canada, with different alkali contents were chosen and are presented in Table 2. The main minerals forming each aggregate (Table 2), were determined by a Siemens D5000 X-ray diffractometer using CuKα radiation generated at 30 mA and 40 kV. Specimens were step-scanned as random powder mounts from 5±65ᵒ 2θ at 0.02ᵒ 2θ steps integrated at 1.2 s step-1.
Aggregates were used in the production of low and high-alkali mortar and concrete specimens. A high purity limestone (HP), non-reactive and almost alkali-free, was used as a control to determine the alkali contribution from the portland cement. Three non-reactive aggregates of moderate- to high-alkali content, i.e. greywacke (RS), granite/gneiss (GG), and phonolite (PH), were used to measure “pure” alkali contribution. In parallel, two alkali- silica reactive aggregates, i.e. siliceous limestone (SPT) and greywacke (SPH), were used to determine the influence of the formation of alkali-silica gel on alkali concentration in pore solution. The results of the accelerated mortar bar (CSA A23.2-25A) and concrete prism (CSA A23.2-14A) tests (CSA A23.1-14/A23.2-14, 2014) are given in Table 2.
Table 2 - Aggregates studied
Sample
Principal minerals and chemical formula
Alkali-reactivity Alkali content
Name location Degree 1 Testing 2 Na2O
(%) K2O (%) Na2Oeq (%) HP High Purity limestone, Newfoundland Calcite (CaCO3), Dolomite (CaMg(CO3)2) NR 0.010% (AMBT) 0.002% (CPT) <0.01 0.04 <0.04 RS Greywacke, Quebec Quartz (SiO2),
Plagioclase Feldspar (NaAlSi3O8),
Microcline (KAlSi3O8),
Chlorite ((Mg,Fe)6(Si,Al)4O10(OH)8),
Calcite (CaCO3),
Phlogopite (KMg3(Si3Al)O10(OH)2
NR 0.339% (AMBT) <0.01% (CPT)4,5 2.37 2.87 4.26 GG Granitic- Gneiss, Quebec
Plagioclase Feldspar (NaAlSi3O8),
Quartz (SiO2),
Microcline (KAlSi3O8),
Amphibole (Ca2(Mg,Fe,Al)5(Al, Si)8
O22(OH)2),
Magnetite (Fe2+Fe3+2O4),
Biotite (KFeMg2(AlSi3O10)(OH)2
NR 0.067% (AMBT)
0,031% (CPT)4 5.01 3.62 7.39
PH Phonolite, Quebec
Plagioclase Feldspar (NaAlSi3O8),
Microcline (KAlSi3O8),
Rhodesite (Ca,K,Na)8Si16O40 • 11(H2O)
Nepheline (Na,K)AlSiO4
Aegirine (NaFe3+[Si2O6])
NR 0.010% (AMBT) 0.015% (CPT) 9.71 5.26 13.17 SPT Spratt siliceous limestone, Ontario Calcite (CaCO3), Quartz (SiO2), Dolomite (CaMg(CO3)2) HR 0.315% (AMBT) 0.184% (CPT) 3 0.03 0.10 0.10 SPH Springhill greywacke, New- Brunswick Quartz (SiO2), Calcite (CaCO3),
Muscovite (KAl2(Si3Al)O10(OH,F)2,
Chlorite ((Mg,Fe)6(Si,Al)4O10(OH)8),
Plagioclase Feldspar (NaAlSi3O8),
Dolomite (CaMg(CO3)2),
HR 0.463% (AMBT)
0.217% (CPT) 3 1.72 2.66 3.47
1. NR = Non reactive; HR = highly-reactive
2. AMBT: Accelerated mortar bar test (CSA A23.2-25A); CPT: Concrete Prism Test (CSA A23.2-14A) 3. According to Fournier et al. (B. Fournier et al., 2016)
4. According to the tests carried out by the quarry
5. The quarry from which RS originates is now closed. The lasts CPT tests results available for this aggregate were obtained using a previous version of CSA A23.2-14A in which the quantity of cement used was 310 kg/m3. In-field inspection of concrete elements made with this aggregate showed no sign of deterioration, confirming its non-reactive character (Bérubé, Fournier, et Frenette, 1989).
4.2 Fabrication of mortar and concrete specimens
The alkali contribution from fine aggregates to the concrete pore solution was determined on mortar specimens using fine particles (size ranging from 0 to 4 mm, as presented in Table 3) produced from the coarse aggregates listed in Table 2. A water-to-cement ratio of 0.5 and cement-to-aggregate ratio of 1:2.25 were used. Distilled water was used in the mixes to prevent any alkali contribution from that source, along with a local high alkali (HA) portland cement (0.33% Na2O, 1.13% K2O, for a total of 1.07% Na2Oeq). Mortar
in a fog chamber at 23ᵒC, samples were sealed in their molds with an appropriate lid and insulation tape (Figure 1) and placed in temperature-controlled cabinets at 38 and 60ᵒC.
Table 3 - Particle size distribution of aggregates used in mortar specimens.
Particle size Mass (%) Passing Retained 4 mm 2 mm 10 2 mm 1 mm 20 1 mm 500 µm 20 500 µm 250 µm 25 250 µm 125 µm 15 < 125 µm 10
Concrete samples were made using aggregate particle sizes ranging from 5 to 20 mm, a water-to-binder ratio of 0.5 and 410 kg/m3 of portland cement; the general procedure described in CSA A23.1-14A-2014 was used for concrete production. As a volumetric approach was employed, the mass of fine and coarse aggregate used in mixes are shown in Table 4. The samples were made with the same cement that was used for the mortar. In order to limit the alkali contribution from the fine aggregate, the sand used in the concrete specimens was produced from the high purity (HP) limestone. In addition, some samples were also made using a low alkali (LA) cement (0.22% Na2O, 0.62% K2O, for a total of 0.63% Na2Oeq) to evaluate the influence of the pore solution alkali content on the aggregate alkali release kinetics. Concrete samples were molded in regular 100 X 200 mm cylindrical plastic molds. After one day in a fog chamber at 23ᵒC, samples were sealed in their molds using the same sealing method used for mortars (Figure 1) and placed in temperature- controlled cabinets at 38 and 60ᵒC.
As all specimens were stored in sealed containers, alkali leaching was avoided. However, this procedure also limits external humidity to reach the samples, thus limiting the progress of ASR.
Table 4 - Mass of fine and coarse aggregate used in concrete mixes.
Constituent (kg/m3) Concrete mixes SPH SPT HP PH GG RS Fine Aggregate 796 772 742 753 763 772 Coarse Aggregate 943 952 988 917 954 945
Figure 1 – 70 and 150-mL cylindrical plastic bottles used for conservation of mortar specimens, and 100 X 200 mm plastic molds used for conservation of concrete specimens.
4.3 Pore solution alkali measurement methods
After 28, 91, 168 and 365 days, samples were taken out of the conditioning cabinets. Two methods were used to determine Na+ and K+ concentration in the pore solution.
The first method is the pore solution extraction method, also known as the high pressure