Evaluating the strength development of mortar using clinker fine aggregate with a combination of fly ash and its inhibitory effects on alkali–silica reaction and delayed ettringite formation

The binders used were Ordinary Portland Cement (OPC) manufactured by Taiheiyo Cement Corporation (density: 3.15 g/cm³, Blaine specific surface area: 3120 cm²/g) and FA (density: 2.27 g/cm³; Blaine specific surface area: 5190 cm²/g). Table 1 presents the chemical composition of OPC and FA. The fine aggregates used were ordinary Portland cement clinker (density in oven-dry condition: 2.72 g/cm³, percentage of absorption: 4.24%; CL), limestone crushed sand (density in oven-dry condition: 2.58 g/cm³, percentage of absorption: 1.85%; henceforth referred to as LS), and andesite (density in oven-dry condition: 2.57 g/cm³; percentage of absorption: 2.73%; henceforth referred to as An.) Here, An was used as the reactive aggregate for the expansion ratio test of ASR. Limestone may promote DEF-induced expansion [6]; therefore, in addition to LS, ISO standard sand conforming to ISO 679 (henceforth referred to as SS) manufactured by Japan Cement Association (density in oven-dry condition: 2.64 g/cm³, percentage of absorption: 0.42%, main components: SiO₂) was used for comparison in the DEF expansion ratio test. Table 2 presents the mineral composition of CL; Table 3 presents the chemical composition of CL, LS, and An; and Table 4 presents the mass fraction of the fine aggregates at each particle size range.

Three tests were conducted in this study: mortar compressive strength test, ASR expansion ratio test, and DEF expansion ratio test. Tables 5, 6, and 7, respectively, display the mix proportion of mortar in each test. Sample with CL as fine aggregate and FA substitution are expressed as “CL-FAXX”, where “XX” represents the substitution ratio; for example, the sample with CL and an FA substitution ratio of 30% is described as CL-FA30. Samples with LS or SS as fine aggregate and no FA substitution are expressed as “LS-OPC” or “SS-OPC”, respectively. Furthermore, the water/binder ratio (W/B) was uniformly set to 0.5; and fine aggregate/binder ratio (S/B) was set to 2.25 for CL and LS, and 3.0 for SS.

For the ASR expansion ratio test, with the purpose of promoting ASR, a 1 mol/L NaOH aqueous solution was prepared using a sodium hydroxide reagent (manufactured by Kanto Chemical Co., Inc., special grade) such that the alkali amount was adjusted to 1.20% Na₂Oeq with respect to the binder. Furthermore, the mixing ratio of An to fine aggregates was unified to a value of 30% based on a previous study [5] that reported a pessimum effect in which the expansion ratio was highest when the mixing ratio of An to the total aggregates was 30 mass%. For the DEF expansion ratio test, with the purpose of promoting DEF, potassium sulfate reagent (manufactured by Kanto Chemical Co., Inc., special grade) was added as well. The amount of potassium sulfate was referenced from previous studies at each level [9, 13] and set to 3% of the mass of OPC without FA substitution. In other words, a unit amount of 33 kg/m³ of potassium sulfate reagent was added to all samples in the DEF expansion ratio test.

The mixing method of mortar conformed to JIS R 5201 [14]. Because CL reacts with water, all fine aggregates were added in an oven-dry condition. Herein, the amount of water required for saturated surface condition was calculated from the percentage of absorption of each fine aggregate, and this amount of water was added to the mixing water. The mixed mortar was cast into a steel mould with internal dimensions of 40 × 40 × 160 mm. The mould for the cast mortar samples was removed after 24 ± 2 h for the compressive strength test and ASR expansion ratio test. For the compressive strength test, samples removed from the moulds were subjected to underwater curing for up to 91 days in a constant temperature room set to 20 °C. The materials and water required for mixing were kept in the constant temperature room for at least 12 h before mixing. For the ASR expansion ratio test, the samples were placed in a closed container and kept at 95% R.H. or higher, followed by storage in an isothermal bath set to 40 °C for up to 182 days, in accordance with the mortar bar method specified in [17] [15]. R.H. of 95% or higher was maintained by filling water at the bottom of the container used for storing the sample, and laying a drainboard such that the sample would not come in direct contact with the water. Herein, in case that Na⁺ and K⁺ may be leached out of mortar if the specimen exposed in a humid container. In this case, if the specimen size was small such as in this study, it is cautioned that the expansion rate of ASR may be underestimated [16, 17].

For the DEF expansion ratio test, the sample was referred using the method described in previous research [18]. The curing method was as follows. At first, mortar that cast into a mould was cured at 20 ± 2 °C for four hours. Next, it was heated to 80 °C over a 3-h period at a heating rate of 20 °C/h. Thereafter, the sample was subjected to high-temperature curing for 10 h in an environment of 100% R.H. and 80 °C, followed by natural cooling to approximately 20 °C. Lastly, after 24 h, the mould was removed, and the mortar was subjected to water curing in an isothermal room set to 20 °C.

In the introduction, it was concerned that high volumes of FA substitution with OPC may reduce mortar strength. Therefore, this section examines the strength development of mortar in which high volumes of FA is substituted with OPC and CL is used.

Figure 1 shows the results of the compressive strength test up to the age of 28 days. As per the figure, the compressive strength decreased as the substitution ratio of FA with respect to OPC increased. However, CL-FA60 had a higher compressive strength than LS-OPC at an age of 28 days. Meanwhile, despite using CL, compressive strength of CL-FA70 and CL-FA80 were lower than that of LS-OPC at an age of 28 days. However, comparing CL-FA70, CL-FA80, and LS-FA30 revealed a similar strength development in all three samples at an age of 28 days.

Figure 2 shows the results of the compressive strength test up to an age of 91 days for samples using CL as fine aggregate and an FA substitution ratio of 70% or 80%. As per the figure, the compressive strengths of both CL-FA70 and CL-FA80 were lower than that of LS-OPC at the age of 28 days: however, their compressive strength leveled to that of LS-OPC at an age of 91 days.

As per the above results, even at an FA substitution ratio of 60%, samples using CL as a fine aggregate exhibited better strength development than LS-OPC. In addition, even at an FA substitution ratio of 70% or 80%, setting a sufficient curing period resulted in a strength development that was equivalent to that of LS-OPC.

Figure 3 shows the results of the ASR expansion ratio test, and Fig. 4 shows the appearances of LS-OPC with expansion and CL-FA70 without expansion as representative samples at the age of 182 days after the test. As per Fig. 3, samples without FA substitution expanded more than 0.7% regardless of the aggregate type. In addition, as per Fig. 4, the LS-OPC with expansion leaked out a gel that was considered to be caused by ASR. Due to the influence of this phenomenon, cracks were observed more clearly. These results indicate that the expansion observed in this study was due to ASR. Furthermore, previously reported data [5] stated that expansion was certain suppressed in samples where 30% FA was substituted by cement, although it did not reach the level of “harmless” specified in [17] [15]. Therefore, it was confirmed that substituting FA with cement can suppress ASR-induced expansion.

As per the previous report [5], ASR-induced expansion ratio was smaller for CL than that for LS when there was no FA substitution; however, this tendency was reversed when the FA substitution ratio was 30%. As mentioned above, Miyamoto et al. discussed that this was due to the increased pH in the mortar due to the leaching of alkali from CL, which hindered the ASR expansion suppression effect of FA. Therefore, it was hypothesized that substituting more FA for cement could lower the pH of the mortar and achieve sufficient an ASR suppression effect [5].

Based on this hypothesis of high-volume substitution FA to cement written in introduction, it was conducted that an ASR expansion ratio test using samples where the FA substitution ratio with respect to cement was 70% and 80%. As per the results, mortars with an FA substitution ratio of 70% or more did not expand clearly. In addition, as per Fig. 4, these mortars did not exhibit any gel leakage, which is thought to be caused by ASR. Therefore, an FA substitution ratio of 70% or more with respect to cement can obtain sufficient ASR expansion suppression effect even when CL was used. Furthermore, as confirmed previously, these mortars had the same strength development as LS-OPC at an age of 91 days; therefore, with careful curing, an FA substitution ratio of 70% or 80% can provide a sufficiently practical countermeasure to ASR even if CL and reactive aggregate are mixed and used.

Figure 5 shows the DEF expansion ratio test results, and Fig. 6 shows the appearances of LS-OPC with swelling and CL-OPC without swelling as representative examples at the age of 182 days after the DEF expansion ratio test. As per Fig. 5, mortars that used SS and LS as fine aggregates started to expand after an age of 91 days, with an expansion of approximately 0.9% at an age of 182 days. As per Fig. 6, many fine cracks were generated on the LS-OPC surface.

Meanwhile, at the age of 182 days, no clear expansion behavior was observed in samples with 30% FA substitution that used SS, LS, and CL, and samples that used CL with only OPC as the binder. Furthermore, as per Fig. 6, CL-OPC did not exhibit the fine cracks. Kawabata et al. reported that the DEF-induced expansion can be suppressed by substituting 20% or more FA or 40% or more blast furnace slag fine powder with cement [19]. Correspondingly, in this study, although the observed age was only 182 days, the results indicate that the expansion caused by DEF can be suppressed by substituting 30% FA with cement. Additionally, the expansion caused by DEF may be suppressed at an age of 182 days even when CL is used.

Previous reports [20, 21] have indicated that the suppression of ASR-induced expansion by FA generally involves reducing the amount of alkali contained in the cement matrix and lowering the pH of the pore solution. In this mechanism, dissolution of FA in the liquid phase supplies in a highly large amount of silicate ions but a small amount of Ca ions. Therefore, Ca ions derived from cement are used for C–S–H precipitation, and the concentration of Ca ions in the liquid phase decreases. In generally, Ca ions, which typically have a strong electrostatic interaction, are electrically adsorbed on the silanol group of C–S–H. However, in case that the concentration of Ca ions in the liquid phase is very low, Na and K ions are electrically adsorbed on the silanol group of C–S–H instead of Ca ions. As a result, the concentration of OH ions, which exists as a counter for Na and K ions, decreases as well, thereby decreasing the pH in the liquid phase [20], which in turn suppresses ASR. In other words, the adsorption of Na and K ions to C–S–H is inhibited when the concentration of Ca ion in the liquid phase is kept high, leading to insufficient ASR expansion suppression effect by FA. CL increases the Ca ion concentration in the liquid, which may reduce the suppression effect of FA on ASR-induced expansion.

Herein, Fig. 7 shows a backscattered electron image of the transition zone between CL and the cement matrix of CL-FA60 at an age of 91 days. In Fig. 7, it could be confirmed that the transition zone is dense, although the hydrated zone of CL is only the surface layer around CL and the large unhydrated zone remains in the interior. Therefore, it was considered that the supply of Ca ions from CL was to be stagnant after the age of 91 days. In addition to this, based on the results of the amount of leaching of Na and K ions from the clinker in case that CL was immersed in saturated calcium hydroxide solution [5], it was considered that the leaching of Na ion from CL almost converges after 91 days and the leaching of K ion from CL almost converges only few days. Therefore, it was considered that the influence of leaching of Na and K ions from CL was to be small after the age of 91 days.

Based on the above aspects, the following discussion is provided: if FA is substituted at a high ratio of 70% or more with cement, then the total amount of Na and K ions derived from cement decreases even after the age of 91 days. Furthermore, it was considered that compared with the increase in the amount of Ca ions by using CL as the fine aggregate, the decrease in the amount of Ca ions due to a decrease in the amount of cement was effective. Therefore, ASR may be sufficiently suppressed because the pH in the system did not increase even when CL was used due to these factors. However, the actual pore volume and pH of the pore solution for mortar specimens were not measured in this experiment. Therefore, it was also considered important to measure these values directly.

It was shown in this study that using CL could suppress DEF-induced expansion as in the case of using FA. Although it was a qualitative evaluation, the molar ratio of the binder SO₃ and Al₂O₃ (SO₃/Al₂O₃ molar ratio) might be used as an indicator for evaluating the risk of DEF-induced expansion [18]. For the present study as well, the relationship between the SO₃/Al₂O₃ molar ratio and expansion ratio is summarized in Fig. 8 along with references to previously reported data [22, 23].

It can be seen from Fig. 8 that expansion occurred in the samples that did not use CL in the range of 1.0–1.3 SO₃/Al₂O₃ molar ratio, including the previously reported data. However, CL-OPC, which had an SO₃/Al₂O₃ molar ratio in the same range, did not exhibit a clear expansion. These results indicate that the mechanism of suppressing DEF-induced expansion using CL might be different from the mechanism of ettringite production suppression due to decreased SO₃/Al₂O₃.

Kawabata et al. indicated that the DEF-induced expansion suppression effect of FA might be due to the improved mass transfer resistance in the system [19]. In addition, Miyamoto et al. reported that using CL improved the mass transfer resistance in mortar [3]. In addition, as shown in Fig. 7, When CL used for fine aggregate, the transition zone between the CL and the cement matrix is denser. Densification of the transition zone is effective in improving mass transfer resistance. Therefore, these results were considered that the suppression effect of CL on DEF-induced expansion was partly due to the improved mass transfer resistance in the mortar.

CL supplies Na and K ions into the system [5]. Exposing mortar to water results in the transfer of substances inside and outside the system. For example, exposing mortar to general water, such as tap water, results in a decrease the pH of the pore solution because Na and K ions are leached from the mortar. This phenomenon promotes secondary precipitation of ettringite because it approaches an environment where ettringite is more likely to be precipitated. Herein, Na and K ions are supplied from CL to the pore solution even when Na and K ions are leached from the mortar system. Therefore, the pH in the pore solution might remain high. In the present study, it was considered that this effect might contribute to the suppression of DEF-induced expansion as well. However, this is currently only a hypothesis; the expansion of the specimen must be observed over a longer period of time to prove it. This is because DEF-induced expansion is known to progress over the course of months to years. The present study set the age period to up to 182 days, which might have been a short period for investigating the expansion. As mentioned in Sect. 4.1, the influence of leaching of Na and K ions from CL might be small after the age of 91 days. Furthermore, as mentioned in Sect. 4.1, it is difficult to accurately determine the actual behavior occurring in the pore solution because the actual pore volume and pH of the pore solution for mortar specimens were not measured in this experiment. Therefore, investigations need to be continued comparisons must be made with results of FA-substituted samples to evaluate whether DEF-induced expansion suppression can occur over a long period of time when CL was used.