The surfaces of samples immersed in phosphoric acid were photographed for visual inspection after 150 days of immersion. Samples were first taken out from acid, and the excessive solution on the surface was slightly wiped out, then the samples were dried at room temperature for 1 hour prior to visual inspections.

The mass of cylindrical samples was measured after 3, 7, 14, 21, 28, 35, 44, 60, 75, 90, 120, and 150 days of submersion in phosphoric acid solution or water. Three samples were used for each mortar mix at each age, and the averages were obtained. Before each measurement, samples were taken out from the phosphoric acid solution or water and wiped gently with a wet cloth. Each measurement was completed within 30 s followed by replacing the samples into the containers in the same orientation.

The capillary sorptivity test was conducted prior to the acid or water immersion. The rate of water ingress was measured by monitoring the mass gains of pre-dried samples over time with one surface contacting with water (ASTMC1585-04, 2004). The results can reflect the tendency of specimens to absorb and transmit water or acid solution via capillarity actions. It is worth noting that capillary sorptivity is not only dependent on porosity but also associated with pore size distributions and connectivity of pores (Pavía and Condren, 2008; Lee and Lee, 2016). The capillary sorptivity test was performed based on ASTM C1585-04 (ASTMC1585-04, 20) using three disk-shaped specimens (50 mm diameter and 10 mm height) for each mortar mixture. Detailed testing procedures can be referred to other studies (Albitar et al., 2017; Ren et al., 2019).

The permeable porosity of the mixtures before and after immersion was measured in compliance with ASTM C642-06 (CASTM 642 (2006), Standa, 2006). Three cylindrical samples (Ф 27.5 mm × H 55 mm) were used for each mix and each exposure condition. The samples were first oven-dried at 60°C until a constant mass was achieved. Drying at 110°C was not adopted due to the possible damages on the pore structures of AASF binders (Ismail et al., 2013). It is worth pointing out that due to the severe damages on the samples in phosphoric acid with pH at 2, the permeable porosity test was not performed on those samples. The volume of permeable voids (VPV) can be quantified based on the calculation methods described in previous studies (Ren et al., 2019; Guo et al., 2021) based on different mass values of specimens measured under various conditions.

The compressive strength of all mortar mixes after 56 days of curing prior to submersion in phosphoric acid solutions or water was determined on cubic samples according to ASTM C109M (AASTM, 1999). After 150 days of immersion in different pH environments, the compressive strength of all mortar mixes was measured again. Triplicates of each mix after exposure were tested to obtain average readings.

The degradation depth of the mixes due to acid ingress was measured after 3, 7, 11, 14, 21, 28, 35, 44, 60, 75, 90, 120, and 150 days of immersion. Immersion in water did not cause any apparent degradation of the sample; therefore, total degradation depth of the mixtures in water was not measured. Six cylindrical samples of each mortar mix were used for each pH condition throughout the immersion period. A slice of about 5 mm was cut off from each cylindrical sample using a diamond saw. After that, the surface of the slice was cleaned with distilled water and dried naturally in the ambient environment; then a phenolphthalein solution was sprayed onto the plane surface and the color change was the criterion used to distinguish the degraded region surrounding the undegraded core (delineated by reaction front in Figure 2). The pink/violet color indicates a pH higher than 8.3, with dark and light violet corresponding to pH > 10.0 and 8.3 < pH < 10, respectively. The uncolored area is considered as the degraded part, with the thickness expressed as neutralized depth (ND). Sample surfaces were first observed using an optical video microscope, and the images were processed with the MATLAB image processing toolbox so that ND can be determined with relatively high precision. For each sample slice, the ND value was obtained from five points randomly distributed around the circumference.

In addition, the dissolved and/or detached depth (DD) was also taken into account due to the partial dissolution of the degraded part, also performed by Larreur-Cayol et al. (2011). The DD value was obtained by measuring the initial diameter and the current diameters at different time intervals. Then the DD was deduced from the formula: D D = D 0 − D i 2 ( i = 1 , 2 , 3 ... ) , (1) where DD is the dissolved/detached depth, D ₀ is the initial diameter of the sample which was 27.5 mm in this study. D _i is the diameter after each immersion period. For example, D ₁ is the averaged diameter when the first diameter measurement was performed after 3 days of immersion. DD was measured before the corresponding ND measurements, expressed as the average of six samples for each mix at each pH condition.

The total degradation depth (TDD), denoted as “degradation depth,” was the sum of two parts: TDD = DD + ND, as illustrated in Figure 2. According to different mechanisms, a greater ND reflects a higher alkalinity loss, whereas a greater DD indicates more structural mass loss caused by dissolution and/or decomposition of mortar samples due to acid immersion.

After 150 days immersion in phosphoric acid with pH at 2.0, sample surfaces were examined by using environmental scanning electron microscopy (ESEM), using an FEI Quanta instrument at a 15 kV accelerating voltage. Polished samples were evaluated in a low vacuum mode using a backscatter detector to avoid the need for carbon coating of the samples. A LINK-ISIS (Oxford Instruments) energy-dispersive X-ray (EDX) detector was used to determine the chemical compositions of both the degraded and undegraded part of the sample. From each region, 20 points were selected within the binder matrix excluding unreacted slag, FA, and OPC particles as well as sand.

The appearances of samples after 150 days of immersion in the phosphoric acid and water are shown in Figure 3. After immersion in water, as shown in Figure 3A, all binders remained intact with a smooth surface and original sharp edges.

When the pH of the acid is 2.0, as presented in Figure 3B, AASF mortar samples seemed to have no obvious signs of degradation, such as discolorations, compared to the control samples submerged in water. In contrast, it is apparent that the OPC-based mortar mixes were severely degraded evidenced by their rough and soft surfaces (particularly around the edges of the samples), due to mortar spalling and less mechanical sound binder. In particular, some shallow holes were visible on the surface of the 100OPC sample. The surface appeared to be more yellowish/brown than the sample after immersion in water, consistent with the findings in other studies (Bertron et al., 2004; Aiken et al., 2017; Muthu and Santhanam, 2018). This color can be attributed to a higher relative content of ferrite in the surface layer or the formation of new iron-containing precipitation since Fe is a favorable element for resistance toward acid attacks (Bertron et al., 2005a; Muthu and Santhanam, 2018). Furthermore, OPC-based samples were found softer in touch than AASF binders, indicating lower resistance of OPC against strong acid.

When the acid pH was 3.0 (Figure 3C), white compounds on the surfaces of all binder mixes were observed, which was likely due to the formation of calcium phosphates, such as brushite (CaHPO₄·2H₂O) (Aiken et al., 2017). In addition, some “pin-hole”-like voids on the surface of 100OPC appeared due to binder dissolution after exposing to the acid. Furthermore, an increased roughness of edges of 65Slag_35OPC was also observable. In comparison, AASF mortar samples did not present obvious damage.

When the pH of the acid was four (Figure 3D), more white compounds were noticed on the surfaces of samples than when the scenario of pH at 3, especially for AASF and 100OPC samples. No apparent damage was observed on AASF samples. The edges of the OPC-based samples were more intact than the ones in acid with higher aggressivity.

The results in Figure 3 show two types of visible performance changes due to exposure to the phosphoric acid solutions, which is the precipitation of white reaction products and rough surfaces and edges. These two effects seemed to occur concurrently. For AASF binders, no obvious signs of damage were observed, and the precipitation effect was more noticeable when the pH value increased. For OPC-based binders, the degradation was more prominent, but the precipitation was less distinguishable when exposed to the acid with lower pH values. One probable reason for this phenomenon is that the calcium phosphates at pH 2 tend to dissolve in the acid solution rather than to deposit on the sample surfaces. By visual observations, it is evident that AASF-based samples present fewer deteriorations in phosphoric acid than the OPC-based binders as consistent in other studies, in which different acids were involved (Aiken et al., 2017; Albitar et al., 2017; Aiken et al., 2018).

Figures 4A–D show the mass changes of the mortar mixes over time in the phosphoric acid solutions and water. Within the first 3 days, all samples obtained mass gains because of water ingress. Although the samples were cured under sealed condition rather than drying condition before immersion, self-desiccation would induce a decreasing saturation degree of the sample with time; thus, water absorption occurs when exposed to solutions.

After the first 3 days, the mass of mortars in water continuously increased before reaching a plateau after about 44 days (Figure 4A). By contrast, the mass of all mixtures in the acid of pH at 2 and 3 started to decrease after 3 days (Figures 4B,C). In general, the mass loss of all samples was higher in pH at 2. In both solutions, 100OPC showed the most rapid mass loss, which was nearly linear with time. A similar trend was observed for OPC concrete exposed to sulfuric acid (Albitar et al., 2017). After 150 days, the mass loss of 100OOPC reached around 5 and 2% in pH at 2 and 3, respectively. The significant mass loss of 100OPC is consistent with its visual damage as shown in Figures 3B,C. In contrast, the mass loss for the other three types of binders decelerated gradually with time. The lower mass loss of AASF binders can be attributed to the formation of an insoluble or a less soluble degraded layer acting as a barrier toward the ingress of acid solutions as reported by Bernal et al. (2012). For 65slag_35OPC, it is possible that some reaction products such as brushite (Ren et al., 2022) blocked the tortuous capillary pores, and thus, the mass loss caused by the phosphoric acid attack became less gradual. The results in Figure 4 suggest that AASF binders displayed higher phosphoric acid resistance than OPC-based binders when pH was at 2–3.

When pH was at four (Figure 4D), the mass loss of AASF after the first 3 days was not significant. 100OPC even experienced a slight increase in mass, which can be attributed to the continuous hydration of cement given a limited damage of the sample by acid (Figure 3C). The formation of white compounds on the surface may also contribute to the mass gain.

Figure 5 shows the water absorption and VPV of the mixtures before and after 150 days of immersion. The water absorption of AASF binders did not change significantly despite the type of solutions. Although some white deposits formed on the surface of AASF immersed in acid of pH at three and four (Figure 3), their influences on the water absorption seem minor. 100OPC mixture showed lower water absorption after immersed in water due to the continuous hydration, and thus, the pore structure was more densified. Acid ingress damaged the structure of OPC; hence, the decrease in water absorption during immersion was counterbalanced.

As shown in Figure 5B, the two AASF binders displayed a decrease in VPV irrespective of the pH values, indicating the possible deposition of reaction products such as calcium phosphates. This corresponds well with the white compounds observed on the surface of the AASF samples after immersion (Figure 3). It is possible that the white compound was a chemical reaction product involved in the precipitation on the surface pores, thus leading to reduced VPV values. Similar phenomenon was reported in another study on the sulfuric acid attacks, which confirmed that the reaction mainly occurs on the sample surface with products blocking pores (Kawai et al., 2005). The inconsistent results between water absorption and VPV indicate that VPV is more sensitive to the precipitation effect. In comparison, 100OPC experienced a smaller decrease or even increment after submersion in acid, indicating a hindered densification or coarsened porosity due to acid attack. It is reasonable since OPC contains more reactive calcium than AASF binders so that more products were dissolved at a high concentration of hydroniums.

The compressive strengths of different mortar mixes after 150 days of immersion in different conditions are shown in Figure 6. For AASF-based binders, acid of pH at 2 and 3 caused reduction in the compressive strength, while pH at 4 seemed to be not so aggressive. These results are in line with the mass changes, which show that little change in mass can be observed for pH at 4, whereas pH at 2 and 3 resulted in obvious mass losses after 150 days of exposure. Immersing in water, interestingly, also caused strength reduction of AASF. This is probably due to the leaching of ions from the pore solution or even solid products. The reduction of around 10% in strength due to immersion in water is similar to what had been reported in another literature (Huang et al., 2018). The reason why pH at 4 was even better than water in terms of strength has not been clearly understood, which is probably associated with the reaction products after degradation clogging the pores and thus impeding the ion leaching process.

For 100OPC, strength increase was observed except for pH at 2. The increase of the compressive strength in water is probably due to further hydration during 150 days of immersion. The higher strength of 100OPC in pH at 3 than at 4 is consistent with the lower VPV of the mixture in pH at 3, as shown in Figure 5, which might be due to the formation of more phosphates precipitated in the specimens such as brushite (Ren et al., 2022). The acid with pH at 2 seemed too aggressive to maintain the structural integrity of OPC. The large standard deviation when pH was at 4 was possibly related to the unevenly distributed degradation reaction products on specimens.

Figure 6 shows that OPC had higher strength than AASF in all conditions, although the acid induced more visual damage (Figure 3) and higher mass loss (Figure 4) for OPC. These seeming contradictory results are due to the fact that visual damage and mass loss occurred mainly on the surface of the samples, while the strength was dependent also on the core part which was relatively intact. This point will be further verified in the next section.

Figure 7 shows total degradation depth (TDD) evolution within 150 days of immersion in phosphoric acid. In general, the stronger the acid (lower pH values) was, the larger degradation depth was observed for all mixtures. The results are in accordance with the findings from other studies that lower pH values of acids would usually result in greater degradation depths (Bertron et al., 2005b; Lloyd et al., 2012). Within the first 7 days of immersion, very small degradation depth was seen, whereas a significant increase was observed soon after, followed by a gradually decreasing rate in the development of degradation depth until the end of immersion. This phenomenon was also reported in a previous study on paste mixtures (Ren et al., 2020b). Within 150 days, total degradation depth of all mixes was within 3 mm, indicating that the degradation basically occurred in the surface, rather than the whole sample.

For all pH conditions (pH at 2, 3, and 4) at the end of exposure, the increasing order of total degradation depth is the same: 50Slag_50FA ≈ 60Slag_40FA < 65Slag_35OPC <100OPC, corresponding well with the mass losses when pH was at 2 and 3 (Figure 4). The results indicate that AASF mortars exhibited better resistance than OPC mortars against the ingress of phosphoric acid, while Figure 6 shows that the strength of OPC samples was actually higher. When it comes to the evaluation of durability of AASF and OPC, an overall consideration of different performances is needed. However, as to the role of concrete as the protection of steel rebar, it seems the AASF system is a better choice than the OPC with comparable 56-day strength.

The mass losses and corresponding degradation depths for the four mortar mixes exposed to the phosphoric acid (pH = 2) are plotted in Figure 8A. It is apparent that total degradation depth becomes greater as the mass loss increases, regardless of mortar types. This is reasonable because mass losses due to dissolution and decomposition can be reflected by the dissolved depth (DD), as a part of degradation depth. Moreover, at the same degradation depth (e.g., shown as the horizontal dash line in Figure 8A), it is apparent that 100OPC sample suffers the greatest mass loss (marked “a” value on x-axis). Thus, it is possible to conclude that the 100OPC sample is more likely to experience great structural disintegrations, which can be attributed to successive decomposition and/or dissolution of main hydration products in OPC binders such as Portlandite, ettringite, and C-S-H gels (Oueslati and Duchesne, 2014). In comparison, 65Slag_35OPC and AASF-based binders tend to obtain much smaller mass losses. It is possible to associate this difference to the different relative aluminum contents in these systems. It is reported that gels with higher aluminum are more resistant to acid attacks because of the intensive cross-linked networks (Bernal et al., 2012; Varga et al., 2015). According to Bernal et al. (2012), the degraded layer of alkali-activated slag mortars, which is mainly composed of aluminosilicates, is still relatively compact after the acid attacks instead of the highly porous silica-rich gel for pure OPC mortars. Another study confirmed that AASF binders displayed less mass loss than OPC under silage effluent attack due to their higher Al/Si and lower Ca/Si ratios (Aiken et al., 2017).

To confirm the relationship between chemical composition and degradation depth, the Al/Si and Ca/Si ratios in both the undegraded and degraded layers of the samples immersed for 150 days with pH at 2 were measured by EDX, and the results are listed in Table 4. Moreover, the slope k values of the linearly fitted degradation depth-mass loss curves (Figure 8B) are also listed in Table 4 for the convenience of comparison. The Al/Si ratios for the two AASF-based binders are higher than those of the 100OPC binder. The opposite is true for the Ca/Si ratios. The Al/Si ratio shows a slight decrease for AASF mortar samples (i.e., from 0.52 to 0.48 for 60Slag_40FA) and an increase for OPC (from 0.24 to 0.30) when the area transforms from the undegraded to degraded part of the samples. By contrast, Ca/Si had a significant decrease for all mixes, indicating a much higher vulnerability of Ca than Al when exposed to phosphoric acid attacks. From Table 4, it can also be seen that the k values of the fitted lines correspond well with Al/Si and Ca/Si ratios: a higher Al/Si or a lower Ca/Si corresponds to a larger k regardless of the region (degraded or undegraded).

The correlation between k values and the Al/Si or Ca/Si ratios in the two areas is further fitted linearly, as shown in Figure 9. It is interesting to find that despite the changes in the chemical composition due to acid ingress, the Al/Si and Ca/Si ratios always have linear relationship with k. This suggests a way to roughly assess the relative degradation resistance of different binders against phosphoric acid, that is, total degradation depth against mass loss, by the chemical compositions of the binder. A higher Al/Si ratio and a lower Ca/Si ratio would lead to larger degradation depth under the same mass loss, or a lower mass loss when total degradation depth is the same. This point was also supported by the results in the study by Aiken et al. (2017). The comparison in Figure 9 shows that the R ² value for Ca/Si of the undegraded layer (R ² = 0.972) is higher than that of the degraded layer (R ² = 0.964), whereas the R ² value for Al/Si shows an opposite result (0.960 and 0.968, respectively). This implies that Ca/Si in the undegraded layer and Al/Si in the degraded layer are more representative in reflecting the resistance to structural disintegrations.

The results in this study provide a possible method to predict the degradation kinetics when only mass loss or degradation depth could be obtained, especially after a long immersion period. In addition, this result shows that the resistance of binders against mass loss when suffering from acid attacks seems to be determined by Al/Si (degraded layer) and Ca/Si (undegraded layer) ratios irrespective of the binder types.