Keywords

1 Introduction

The dissolution of the reactive silica components of an aggregate can result in a concrete durability issue known as the alkali–silica reaction (ASR). The reactive silica components of an aggregate, which dissolve when the concrete pore solution’s alkali concentration is sufficiently high, can react with calcium and alkali ions in the pore solution to form the ASR products that lead to expansion and deleterious cracking of the concrete [1]. Reactive silica minerals are those that are amorphous, have poor crystallinity, highly strained, or contain high amount of defects making them susceptible to alkali dissolution.

Two test methods are widely used to assess ASR potential: the accelerated mortar bar test (AMBT) and the concrete prism test (CPT). AMBT AS1141.60.1 makes use of 1 M NaOH at 80 °C to accelerate the reaction and an expansion of <0.1% at 21 days indicates that the aggregate is nonreactive [2]. CPT AS1141.60.2, on the other hand, involves boosting the cement alkali content to 1.25% Na2Oeq and storing the concrete prisms in a sealed, high humidity environment at 38 °C. The CPT is a longer test method than the AMBT and has an expansion limit of 0.03% at 1 year for an aggregate to be considered nonreactive [3]. The Australian AMBT and CPT methods are based on ASTM C1260 and ASTM C1293 respectively, with slightly modified test limits.

Both the AMBT and CPT remain widely used although heavily criticized owing to their limitations. The use of high temperature as well as excessive alkali supply in the AMBT can result in false positives (i.e., identifying an aggregate as reactive even if it is not). Moreover, the influence of cement alkalinity, which significantly contributes to the ASR, is hard to determine [4]. CPT on the other hand is prone to alkali leaching, which can lead to underestimation of expansion and is reportedly ≈25–35% in 1 year [5, 6]. Because of the limitations of these test methods, an alternative ASR test, the simulated pore solution method (SPSM), has been developed at the Laboratory of Construction Materials, EPFL Switzerland. This test method involves immersing the concrete prisms in a solution based on the pore solution’s alkali concentration for the concrete being assessed for ASR potential and so far has shown promising potential as an alternative ASR test method [7, 8].

Due to the availability of various test methods for assessing the risk of an aggregate for ASR, it is therefore important to assess the effect of these test methods on the type of ASR products. Moreover, the comparison of the composition of the ASR products formed using these methods to those in actual structures affected by ASR is also important to determine whether ASR products formed during accelerated testing resemble those in the field. Therefore, we investigated the composition of the ASR products in both the AMBT, and SPSM and compared them to ASR products in a 25-year-old bridge in Australia, decommissioned because of ASR.

2 Methods

2.1 ASR Testing as Per AMBT and SPSM

Expansion tests using AMBT and SPSM were carried out using Australian reactive aggregates and Australian cement. The cement complied with the 0.6% Na2Oeq alkali limit.

For the AMBT, mortar bars composed of 1 part cement to 2.25 parts graded aggregate by mass (440 g cement per 990 g of aggregate) and a water to cementitious materials ratio equal to 0.47 were prepared in accordance with AS1141.60.1. The mortar specimens were prepared in 25 × 25 × 285 mm molds with a gauge length of 250 mm, then cured in a high humidity environment at room temperature (23 ± 2 °C) for 24 h. Next, the specimens were demolded and placed in a water-filled container, before being placed in an oven at 80 °C for another 24 h to allow the specimens to slowly equilibrate to 80 °C. Horizontal comparator was used to obtain zero-hour length measurements before immersing the specimens in 1 M NaOH solution at 80 °C for 28 days. Succeeding expansion measurements were obtained at 1, 3, 7, 10, 14, 21, and 28 days. Three readings were taken per mortar specimen at each age. Total expansion incurred by the aggregate after 10 and 21 days of NaOH immersion was used to classify its ASR potential when used in the field in accordance with AS1141.60.1.

For the SPSM, concrete prisms (70 × 70 × 280 mm) were prepared with a cement content of 410 kg/m3 and water-to-cement ratio of 0.46. The concretes were cured for 28 days in a high humidity chamber (>90% relative humidity) before storage in simulated pore solution at 60 °C. Expansion measurements were taken before storage and every month thereafter using a vertical comparator. The simulated pore solution was prepared based on the extracted pore solution of an equivalent binder system at 28 days.

2.2 Analysis of the Composition of the ASR Products

Polished sections of mortar/concrete that underwent expansion tests using the AMBT, and SPSM, as well as the concrete from the demolished bridge were prepared and subjected to scanning electron microscopy–energy-dispersive spectroscopy (SEM–EDS) analysis. The AMBT samples were sectioned after 28 days in the AMBT bath while the concrete prisms were sectioned after 6 months in the simulated pore solution at 60 °C.

The mortars and concretes were cut to fit a 25-mm diameter mold, then vacuum impregnated with epoxy resin and polished, first with silicon carbide paper until the sample surface had been fully uncovered from the resin, followed by automated polishing using MD Largo Struers discs lubricated with petrol and diamond spray as the polishing agent (9 µm, 3 µm and 1 µm particle sizes). After polishing, the samples were cleaned in an ultrasonic bath for 2 min and then stored in a vacuum desiccator for at least 2 days to dry. The samples were coated with carbon to prevent charging during SEM imaging.

Imaging and elemental analysis of the carbon-coated polished sections were carried out using an FEI Quanta 200 with Bruker XFlash 4030 EDS detector. The microscope was operated in backscattered electron (BSE) mode, 15 kV accelerating voltage and 12.5 mm working distance in a high vacuum.

3 Results and Discussion

The AMBT and SPSM expansion plots are shown in Fig. 1, confirming the high reactivity of the aggregates as indicated by the significant degree of expansion. The AMBT mortars (dacite and greywacke) both exceeded the 0.1% limit at 10 days and 0.3% at 21 days, making them reactive as per AS1141.60.1. There is currently no established limit for the SPSM but the high degree of expansion of the dacite concrete confirmed the reactivity of the aggregate.

Fig. 1
Two line graphs of expansion versus days and months. A, greywacke mortar is high at (30, 0.50) and Dacite mortar is (30, 0.45) percent. B, dacite concrete extends between (0, 0.00) and (6, 0.13) percent. The values are approximate.

Expansion plots of the aggregates subjected to a accelerated mortar bar test and b simulated pore solution immersion test

Full size image

Figure 2 presents images of the 25-year-old bridge before it was decommissioned, showing extensive damage due to ASR. The cracks observed have a map crack appearance, which is characteristic of ASR [9].

Fig. 2
Two photographs. A group of people investing in the bridge is at the left and the cracks observed on the bridge are at the right.

A 25-year-old bridge in New South Wales, Australia, suffering from alkali–silica reaction

Full size image

Figure 3 shows the ASR damage observed in the mortar and concrete that underwent AMBT and SPSM respectively. In both cases, the cracks were concentrated within the aggregate and extend towards the paste, which indicated that deleterious ASR damage originated within the aggregate and explains the characteristic map crack appearance of ASR damage in affected structures. Figure 4 shows the ASR products observed in the AMBT sample, and Fig. 5 shows the ASR products observed in the concrete sample subjected to SPSM with their corresponding EDS maps. For both cases, the presence of calcium, silicon and alkalis are notable confirming the composition of the ASR product (alkali-calcium silicate hydrate). It is however notable that whereas the alkali present in the AMBT is only sodium, the ASR product in the concrete prism has both sodium and potassium. This indicates that the type of alkali in the ASR product is strongly affected by the dominant alkalis in the pore solution.

Fig. 3
Two S E M images of concrete. The regions representing aggregate and paste are highlighted.

Scanning electron microscopy images showing extensive damage in the mortar and concrete after a accelerated mortar bar test and b simulated pore solution immersion test

Full size image
Fig. 4
Five S E M images. A, aggregate, B S E, and A S R product is marked with an arrow. B, Calcium. C, silicon. D, sodium. E, no potassium is marked with an arrow in Potassium.

Alkali–silica reaction (ASR) product observed in the accelerated mortar bar tested mortar with EDS maps showing strong presence of calcium (Ca), silicon (Si) and sodium (Na). BSE, backscattered electron mode; EDS, energy-dispersive spectroscopy

Full size image
Fig. 5
Five S E M images. A, aggregate, paste, B S E, and A S R product is marked with an arrow. B, Calcium. C, silicon. D, sodium. E, potassium.

Alkali–silica reaction (ASR) product observed in the concrete subjected to the simulated pore solution method (SPSM) sample with EDS maps showing strong presence of calcium (Ca), silicon (Si), sodium (Na) and potassium (K). BSE, backscattered electron mode; EDS, energy-dispersive spectroscopy

Full size image

Figure 6 shows the ASR products forming around the aggregate and in the paste, which has a notably darker color than the ASR products inside the aggregate, suggesting a difference in composition. Table 1 tabulates the EDS results of the ASR products inside the aggregate (SPSM sample) and the ASR products around the aggregate and near the paste (SPSM sample). As can be seen, the ASR products outside the aggregate have a composition closer to C-S–H and a much higher Ca/Si ratio and lower Na + K/Si than the ASR products inside the aggregate. In general, the silicon content of the ASR product decreased and calcium content increased as the product came in closer contact with the cement paste  [10,11,12,13,14]. The role of calcium, however, remains controversial. Although higher calcium content in the ASR product results in higher stiffness [15], as the ASR product becomes more rigid, it also has decreased swelling potential [1, 16]. The substitution of alkalis with calcium suggests there is a competitive reaction between calcium and alkalis and that calcium is always preferentially absorbed, which supports the alkali recycling theory [1].

Fig. 6
Two S E M images. A S R product is marked with an arrow, aggregate, and paste areas are labeled.

Alkali–silica reaction (ASR) products observed in the simulated pore solution method sample surrounding the aggregate and located in the cement paste

Full size image
Table 1 Energy-dispersive spectroscopy results for the Alkali–silica reaction (ASR) product observed in the simulated pore solution method (SPSM)sample (normalized without oxygen)
Full size table

Table 2 tabulates the EDS results of the AMBT sample (inside the aggregate). As can be observed, consistent with the EDS mapping results, almost no potassium can be detected in the ASR products. The total alkali content (Na + K) is, however, comparable to the SPSM samples (≈20%), as well as the Ca/Si and (Na + K)/Si. This indicates that the composition of the ASR products inside an aggregate has a similar stoichiometric ratio of calcium, silicon and alkali regardless of the ASR test method. The type of alkali, however, varies depending on the dominant alkali/s in the pore solution.

Table 2 Energy-dispersive spectroscopy results for alkali–silica reaction (ASR) products observed in the accelerated mortar bar test (AMBT) sample (normalized without oxygen)
Full size table

Figure 7 shows the ASR product in the demolished concrete bridge with corresponding EDS maps. As can be observed, the ASR product was also concentrated inside the aggregate and also contained calcium, silicon and alkali similar to the ASR products in the AMBT and SPSM samples. The ASR product in the bridge, however, showed the presence of both sodium and potassium and hence its composition was closer to the SPSM sample than the AMBT sample.

Fig. 7
Five microscopic images. A, S E and A S R product are marked with arrows. B, calcium. C, silicon. D, sodium. E, potassium.

Alkali–silica reaction (ASR) product in the demolished bridge (NSW, Australia) with energy-dispersive spectroscopy mapping

Full size image

4 Conclusions

This study showed a difference in composition between ASR products found in a mortar subjected to AMBT, concrete subjected to SPSM and a 25-year-old bridge demolished because of ASR. Our results showed the following.

  1. 1.

    The ASR products in the AMBT, SPSM and demolished bridge samples all contained calcium, silicon and alkali. The ASR products in the AMBT sample, however, only contained sodium whereas those in the SPSM and bridge samples contained both sodium and potassium, which indicated that the type of alkali in the ASR product is significantly affected by the dominant alkali in the pore solution.

  2. 2.

    Although the type of alkali in the ASR product was affected by the test method, the total Na + K in the ASR products found inside the aggregate remained the same regardless of the test method, which suggests that the total alkali plays a more significant role in ASR expansion than the type of alkali.

  3. 3.

    The composition of the ASR products varied according to the location in the concrete. ASR products within the aggregate had a lower Si/Ca ratio and higher (Na + K)/Si ratio than ASR products in the cement paste and those surrounding the aggregates. Thus, the closer the ASR product is to the cement paste, the higher the calcium content and the lower the alkali content, which confirms the theory of alkali recycling where Ca2+ substitutes for alkali. This finding also suggests that the composition of the ASR products affects the rate of expansion and that ASR products with lower Si/Ca ratio and higher alkali content (observed inside the aggregate) may be more deleterious.