1 Introduction

The alkali–silica reaction (ASR) is one of the most severe durability issues in Portland cement-based materials. Its occurrence in concrete products can cause significant structural damage and shorten service life. ASR initializes when the reactive (also known as amorphous) silica from the aggregates is dissolved by the alkaline from cement [1]. Although using nonreactive aggregates can be the most effective strategy to prevent ASR, the transportation cost in some locations due to less availability of these aggregates can be significantly unprofitable. In addition, re-using waste materials as aggregates, such as crushed glass, which has a high content of amorphous silica, has been of great interest for decades because it provides an environmentally friendly solution to disposal of the wastes [2]. Therefore, an economic ASR control method is necessary to maximize the environmental benefit.

Adding supplementary cementitious materials (SCMs) is a feasible method of mitigating the ASR. Previous studies have reported the ASR mitigation effects attributed to SCMs, including metakaolin, coal fly ash (FA) and ground granulated blast-furnace slag (GGBS) [3, 4]. Alum sludge is a byproduct of drinking water treatment and because alum-based coagulant is usually added to raw water to remove insoluble particles such as sand and microorganisms, the primary chemical composition of alum sludge includes aluminum, silicon and organic compounds. Therefore, alum sludge could be a SCM. Previous studies [5,6,7,8,9,10] have demonstrated the successful utilization of alum sludge in the manufacture of mortar and concrete blocks. The results indicated that up to 10% cement replacement with calcined and milled alum sludge could improve mechanical performance, but a strength reduction would occur when higher proportions of cement were replaced. Because GGBS-incorporated concrete products exhibit considerable strength even with high volume cement replacement [11], blending GGBS with alum sludge could potentially improve mechanical performance further when more than 10% cement is replaced. In addition, a ternary blended system is expected to have satisfactory ASR resistance due to the high Al content of the sludge.

In the present study, mortars with ternary blended binders containing GGBS and alum sludge were developed to achieve considerable mechanical performance and ASR resistance. Compressive strength, the ASR mitigation performance and microstructural characteristics were investigated for different mixtures of GGBS and alum sludge.

2 Methods

2.1 Materials

General-purpose cement was used as the binder according to AS3972 [12] and clear crushed glass, which contains high levels of amorphous silica, was used as fine aggregate to accelerate the ASR. The alum sludge was supplied by a drinking water treatment plant located in South Australia. The raw sludge was calcined at 800 ℃ for 2 h, and then the alum sludge ash (ASA) was milled to pass a 75-µm sieve. The particle size distributions of the crushed glass, ASA and GGBS are shown in Fig. 1a. The chemical composition of the calcined alum sludge was analyzed using X-ray fluorescence (XRF), and the results were compared with those for cement and GGBS (Fig. 1b).

Fig. 1
A graph of volume in percentage versus particle size micrometers for cement, G G B S, A S A, and glass. A radar chart plots the chemical composition of C a O, S i O 2, A l 2 O 3, F e 2 O 3, and others in cement, G G B S, and A S A.

a Particle size of cement, alum sludge ash (ASA), granulated blast-furnace slag (GGBS) and glass aggregate; b chemical composition of cement, ASA and GGBS

2.2 Experimental Protocol

2.2.1 Sample Preparation

The mix proportions of mortars were designed according to AS1141.60.1 [13], and shown in Table 1. The water to cement and aggregate to cement ratios were 0.47 and 2.25, respectively. All mortar samples except those for the ASR tests were cured in a chamber with temperature and relative humidity controlled at 23 ℃ and 95%, respectively, for 28 days before testing.

Table 1 Mixture of the mortar samples (kg/m3)

2.2.2 Compressive Strength and Accelerated ASR Tests

The compressive strength was tested according to AS4456.4 [14], and the loading rate was 0.33 MPa/s. The ASR resistance of the mortar samples was evaluated according to AS1141.60.1 [13]. The demolded samples were cured in 80 ℃ water for 24 h. The initial length of the mortar beams was determined at the end of curing and then the samples were immersed in 1 M NaOH solution in a water bath with the temperature set to 80 ℃.

2.2.3 Microstructural Analysis

The microstructural characteristics of the samples after the ASR attack were observed using backscattered electron (BSE) micrographs, and the elemental analysis of the cement matrix was evaluated using energy-dispersive X-ray spectroscopy (EDS) with accelerating voltage at 15 kV. X-ray diffraction (XRD) patterns were obtained using copper Kα radiation at 40 kV and 40 mA.

3 Results and Discussion

3.1 Compressive Strength

The 28-day compressive strength of the control sample was 34 MPa, and the percentage of strength for other samples relative to the control is shown in Fig. 2. For the binder with binary blends (ASA and GGBS group), 10% cement replacement improved the mechanical performance, but greater than this value, a strength reduction was observed. In contrast, for the ternary blended binders (containing both ASA and GGBS, labelled as AG group), the mortar with 30% cement replacement still had a considerable compressive strength compared with the reference, which was attributed to the extra pozzolanic reaction from the synergy of ASA and GGBS. The excessive portlandite (CH) from the GGBS could react with the silica species in the ASA, contributing to higher mechanical performance at a higher cement replacement level than with the binary blends.

Fig. 2
A triple-bar graph plots relative compressive strength in percentage for 10, 20, and 30 percent cement replacement in A S A, G G B S, and A G. The relative compressive strengths for 10, 20, and 30% cement replacement in A S A are 112, 94, and 84, in G G B S are 109, 98, and 90, and in A G are 108, 117, and 109.

Compressive strength relative to the control sample. AG, ASA and GGBS; ASA, alum sludge ash; GGBS, granulated blast-furnace slag

3.2 ASR-Induced Expansion

The results of ASR-induced expansion and the surface cracking observed in samples are shown in Fig. 3. Although 30% cement replacement with GGBS kept expansion less than 0.3% for 21 days, the 14-day expansion exceeded the threshold of 0.1%. As a comparison, 20% ASA content effectively prevented the ASR. GGBS exhibited a negligible ASR mitigation effect than ASA for the mortar with binary blended binders due to the higher Ca content in GGBS, which had the potential to promote ASR-induced expansion. In addition, high Al content in ASA was beneficial for binding alkaline. The analysis of the effect of Ca and Al on the ASR is discussed in Sect.  3.4. For the samples in the AG group, 20% cement replaced with a mix of ASA and GGBS suppressed ASR-induced expansion to less than 0.3%, and 30% cement replacement significantly mitigated the ASR. The expansion results were consistent with the surface visual observation of the samples shown in Fig. 3b, where tree-like cracks were found in the reference (R0) sample, and no obvious crack could be identified in AG30.

Fig. 3
A graph and an image. a. A stacked bar graph plots 21-day and 14-day expansions in percentage for R 0, G 10, G 20, G 30, A 10, A 20, A 30, A G 10, A G 20, and A G 30. It indicates G G B S, A S A, G G B S plus A S A, a 21-day threshold, and a 14-day threshold. b. Two surface visual observation images, R 0 and A G 30, indicate cracks.

a ASR-induced expansion; b surface visual observation of the mortars. ASA, alum sludge ash; ASR, alkali–silica reaction; GGBS, granulated blast-furnace slag

3.3 Phase Analysis of the Mortars

The XRD spectra of the samples before and after the ASR test are shown in Fig. 4. No CH peak can be found for sample A30 before the ASR test, which indicated that the CH from cement hydration participated in the pozzolanic reaction and was consumed by the excess ASA. However, the amount of CH may not be sufficient to generate considerable pozzolanic C-(A)-S–H. Thus, the cement dilution effect dominated the mechanical properties, causing a lower strength than the reference.

Fig. 4
Two X-ray diffraction spectra of the samples before and after the A S R test plot A G 30, A G 10, A 30, A 10, and R 0 with respect to 2 theta in degree. They indicate the peaks of portlandite, quartz, calcite, tricalcium silicate, and katoite in A G 30, A G 10, A 30, A 10, and R 0.

X-ray diffraction spectra of the samples before and after ASR testing. ASR, alkali–silica reaction

Incorporating GGBS into the ASA provided an additional CH source and ensured a high degree of pozzolanic reaction. The cement dilution effect could be overwhelmed by the generation of pozzolanic products, which contributed to the better mechanical performance of the mortars in the AG group than those in the ASA and GGBS groups. After the ASR test, kaoite (C3AH6) was detected as a new phase in the samples with ASA content. Kaoite is usually precipitated at a high Al/Si ratio, accompanying C-A-S-H formation, especially in high-alkaline solution [15]. The later formed C-A-S–H gels were beneficial for mitigating ASR, attributed to their alkaline binding ability.

3.4 Microstructural Characteristics and Elemental Analysis

Figure 5 shows the BSE images of the R0, A10 and AG30 samples after the ASR attack. ASR gels could be identified in two locations: interior glass aggregate labeled as T1 and surrounding the glass aggregate labeled as T2. Both T1 and T2 were found in R0 and A10, but only T2 was detected in AG30. The major cracks in R0 were collinear to the T1 gels, indicating that the growth of the ASR gels’ interior aggregates may be the main factor in cement matrix damage and sample expansion.

Fig. 5
Three B S E-SEM images of R 0, A 10, and A G 30 samples. The cracks, T 1, T 2, A S R gel type 1, A S R gel type 2, and glass are indicated in R 0. T 1, T 2, and glass are indicated in A 10. T 2 and glass are indicated in A G 30.

BSE-SEM images of a R0, b A10 and c AG30 samples. BSE, backscattered electron; SEM, scanning electron microscopy

The elemental analysis of T1 and T2 was performed using EDS technology to obtain the atomic ratios of Ca, Al, Si and Na in the ASR gels. The results listed in Table 2 show that the Ca/Si ratio of the ASR gels surrounding the aggregates (T2) was significantly higher than that inside the aggregates (T1) for the R0 and A10 samples. Although the initially formed ASR gels had a low Ca content, the gels surrounding the aggregates could be more able to absorb Ca2+ than the interior ASR gels and then transform into Ca-rich ASR gels [16]. Incorporating Ca into ASR gels could increase their viscosity, making them difficult to transport in pores. In addition, further taking up Ca from the pore solution could transform ASR gels into C-S-H [17]. The Ca-rich ASR gels and C-S-H layer surrounding the aggregates will limit the extrusion of the gels inside the aggregates but cannot prevent alkaline transportation into T1 and ASR gel precipitation in T1. The growth of the constrained T1 ASR gels finally cracked the aggregates and cement matrix.

Table 2 Atomic ratios in the ASR gel of samples R0, A10 and AG30

The Ca/Si ratios of the T2 ASR gels in AG20 were much lower than in R0 and A10 and similar to those of T1 in R0 and A10. ASR gels in AG30 were more flowable due to their lower Ca content. The possible reason for this result is that the Al from the ASA prohibited Ca absorption by the ASR gels, thus eliminating the constrain effect. The Al layer surrounding the aggregate observed in the EDS mapping shown in Fig. 1 could be evidence for this explanation (Fig. 6).

Fig. 6
Two energy-dispersive spectrometry maps of the sample A G 30 indicate the A l layer surrounding the aggregate.

Energy-dispersive spectrometry mapping of sample AG30

4 Conclusions

We developed a mortar with a ternary blended binder containing ASA and GGBS. This newly developed mortar exhibited satisfactory mechanical performance and excellent ASR resistance even with high reactive waste glass aggregates. Based on the results, the following conclusion can be drawn:

  • The mechanical performance of mortars can be improved using a mix of ASA and GGBS as a cement replacement. The mortar with 20% cement replaced in the AG group exhibited the highest compressive strength of 40 MPa, and 30% cement replacement resulted in a considerable strength compared with the reference.

  • ASA had a more substantial ASR mitigating effect than GGBS when used as the SCM and could effectively mitigate ASR. 10% ASA content in the binder could restrain ASR-induced expansion in mortars within 0.3% in the 21-day test. For the mortars with ternary blended binder, 30% cement replacement could successfully suppress ASR.

  • The Al content of ASA played a significant role in mitigating ASR. The XRD result indicated the incorporation of Al in the C-S–H structure, which would limit silica dissolution. The microstructural and elemental analyses also suggested that ASA could prevent ASR gels absorbing Ca, eliminate the restriction effect of ASR gels surrounding the aggregate and impede cracking caused by the growth of ASR gels inside aggregates.