Preparation of shrinkage-free alkali-activated slag material using MgO as both the activator and the expansive agent

2.1. Materials

In this study, ground granulated blast slag (GGBS) was provided by a local supplier with the grade S105 according to GB/T18046 ( 31 31. GB/T18046. (2017) Ground granulated blast furnace slag used for cement, mortar and concrete. Standardization Admination of the Peoples’s Republic of China, China. (in Chinese).
) and the BET surface area of 1535 m²/kg. The chemical composition of GGBS was obtained by X-ray fluorescence (XRF) (Rigaku Corporation), as shown in Table 1 . The main chemical components are SiO₂, CaO and Al₂O₃ with little Fe₂O₃, MgO, TiO₂, and K₂O. The particle size distribution was determined by a Mastersizer 2000 (UK Malvern Instruments Co. LTD, Malvern, UK) laser particle size analyzer, as shown in Figure 1a . The D₅₀ particle size of GGBS is about 20 μm. The mineral composition was analyzed by X-ray diffraction (XRD) technique using a Smart Lab X-ray diffractometer, as shown in Figure 1b . It can be seen that GGBS is vitreous with amorphous phases as its major content; and there also exists some melilite in GGBS.

Analytical reagent grade MgO was purchased from Tianjin Zhiyuan Chemical Reagent Co., LTD. According to DL/T 5296 ( 32 32. DL/T 5296. (2013) Technical specification of magnesium oxide expansive for use in hydraulic concrete. National Energy Administration, China. (in Chinese).
), the activity of MgO was measured as follows: ( 1 1. Fang, Y.; Wang, J.; Qian, X.; Wang, L.; Chen, P.; Qiao, P. (2022) A renewable admixture to enhance the performance of cement mortars through a pre-hydration method. J. Clean. Prod. 332, 130095. https://doi.org/10.1016/j.jclepro.2021.130095 .
) 1.7g MgO was added into 200mL citric acid solution with a concentration of 1.3%; ( 2 2. Xu, D.; Cui, Y.; Li, H.; Yang, K.; Xu, W.; Chen, Y. (2015) On the future of Chinese cement industry. Cem. Concr. Res. 78, 2-13. https://doi.org/10.1016/j.cemconres.2015.06.012 .
) phenolphthalein was added as the inductor to determine the time when the color of the solution turns into red. The activity index of MgO was measured as 75s, indicating the relatively high reactivity of MgO. Water without statement is tap water.

2.2. Mix proportion

Five AAS mixtures were considered to prepare one-part MgO-activated slag material. The dosages of MgO were 3%, 6%, 9%, 12%, and 15% of mass of GGBS, as shown in Table 2 . The mix ID “MSx” indicates the AAS was made with x% MgO activator by the weight of precursor (GGBS). The water/GGBS ratio of all mixtures was 0.48. To prepare the one-part AAS, the precursor materials and solid activators were dry-mixed for 5 mins, and then mixed another 5 mins after adding all the water. Samples with a size of 40 mm× 40 mm× 160 mm were prepared for the measurement of mechanical strength. The produced samples were cured at 23 °C and relative humidity higher than 95% until the designated test day.

2.3. Test methods

3.1. pH of pore solution

Figure 2 shows the pH of the pore solution of AAS over 24 h. It clearly shows that the pH of pore solution is generally higher in AAS, with more MgO at each time. A longer age facilitated the increment in the pH of the pore solution. The hydration of the AAS activated by MgO began with the hydration of MgO to form Mg(OH)₂, which is the activator for GGBS. The presence of Mg(OH)₂ is a competing result of the formation and consumption of Mg(OH)₂. Then, a higher dosage of MgO implies that more MgO is available for forming Mg(OH)₂, resulting in a higher pH pore solution.

3.2. Setting time

Figure 3 presents the initial and final setting times of AAS with various dosages of MgO. With the increase of MgO, the hydration process of AAS was accelerated, as indicated by the reduction of the initial and final setting times. Compared with MS3, the initial setting times of MS6, MS9, MS12, and MS15 were decreased by 21.7%, 33.9%, 37.8%, and 44.3%, respectively, and the final setting times were accordingly decreased by 18.5%, 36.3%, 39.6%, and 41.5%, respectively. Additionally, saturated dosages of MgO seem to be determined at the MgO level between 12% and 15% since the final setting times of groups MS12 and MS15 reach at similar value. The activation mechanism of GGBS by MgO involves the following process. Highly reactive MgO reacts with water, producing a significant amount of expansive alkaline Mg(OH)₂. Alkali hydration of a GGBS corresponds to a complex process including the initial dissolution of the glass phase in the GGBS and consequent a later precipitation of the hydration products (C-A-S-H). Similar to the other pozzolans, the rate of dissolution of the glass in GGBS increases at a higher pH value. For the limited mixing water, excessive MgO could not generate a higher pH value due to the saturation of alkali compounds in the pore solutions. As a result, the dissolution of GGBS was slowed down, resulting in a similar hydration degree of GGBS, as indicated by the similar value of the final setting time of MS15 and MS12.

3.3. Mechanical strength

Figure 4 presents the average compressive strength of AAS activated by MgO at different curing ages along with the standard errors. As can be seen from this figure, the average compressive strengths of MS3 at 3d and 7d are less than 10 MPa. Compared with MS3, the 3d average compressive strengths of MS6, MS9, MS12, and MS15 were improved by 73.6%, 105.5%, 136.8%, and 131.1%, respectively. When the curing age went to 28d, the average compressive strengths of MS6, MS9, MS12, and MS15 were enhanced by 6.9%, 37.1%, 41.4%, and 30.8%, respectively compared with MS3. Additionally, MS12 achieved the highest average compressive strengths at all ages, with 136.8%, 80.4%, and 41.4% increments in the average compressive strength at 3d, 7d, and 28d, respectively, compared with MS3. More MgO can be expected to produce more hydrated Mg(OH)₂, bringing more alkali compounds into the AAS pore solution. In this manner, the hydrolysis of Al-O-Si bonds can be facilitated by more OH^- group, leading to the quicker breakdown of the depolymerized glass network and fast release of Si and Al ion in the pore solution. As a result, the hydration degree of GGBS and the formation of hydration products were promoted, contributing to higher average compressive strength. However, when the content of MgO was raised to 15%, AAS may suffer from excessive expansive stress and deformation due to more expansive hydration products like Mg(OH)₂ generated in the mixture, resulting in more microcracks in the mixtures and lower average compressive strength.

Beyond the average compressive strength, the average flexural strength development of AAS was also explored to fully evaluate the influence of MgO on the mechanical strength of AAS, as shown in Figure 5 . The optimum content for the highest average flexural strength was slightly different compared to that for average compressive strength, which was given in MS9. More specifically, compared with MS3, MS9 had 54.8%, 76.0%, and 39.7% increments in average compressive strength at 3d, 7d, and 28d, respectively. The difference between the average compressive strength and average flexural strength might be caused by the sensitivity of expansion and shrinkage deformation in these two tests, considering flexural strength is more sensitive to micro-cracks in samples. In summary, 9% was confirmed as the final optimal dosage for AAS activated by using MgO as the single activator.

3.4. Pore structure

The pore structure of AAS at 28d was measured using NMR, as shown in Figure 6a . The pore structure is dramatically refined by the increase in MgO, leading to the reduction of the total pore volume and the shift towards finer pores. The improvements in pore structure can be attributed to more expansive hydration products and the higher hydration degree of GGBS. To quantitatively analyze the pore structure, the pore sizes were divided into four zones: Z1 (<20nm), Z2 (20 - 100nm), Z3 (100 - 200nm), and Z4 (>200nm). The obtained result is concluded in Figure 6b . Normally, the pores in Z1 are related to gel pores in the hydration product. In other words, the volume of pores less than 20 nm can be regarded as an index of AAS’s hydration degree. In this study, the pore volumes located in zone Z1 increased with the increase of MgO, implying that the increase of MgO can promote the hydration degree of GGBS. In addition, the pores located in Z2 decreased with the content of MgO increasing, which can be attributed to the expansive hydration and denser microstructure caused by a higher hydration degree. This is well consistent with the results of mechanical strength, as shown in Figure 4 and Figure 5 .

Especially, the correlation between the pore volumes and MgO content could be linearly fitted for zones Z1 and Z2, as shown in Figure 7 . It can be found that the gel pores (less than 20 nm) in the hydration products of AAS increase linearly (R² = 0.905) with the content of activators, while the capillary pores between 20-100 nm decrease linearly (R² = 0.989) with the amount of MgO. This indicates that the amount of expansive hydration products generated by MgO is the key factor in the refinement of pore structures. These expansive alkali compounds involved in hydration effectively fill the pores and refine the pore structure, which is the direct reason for the improvement of the compressive strength of AAS within a certain range.

3.5. Hydration products analyzed XRD

Figure 8 shows the XRD patterns of MgO-activated slag pastes at the ages 7d and 28d. Major crystalline minerals found in the pastes include typical hydration products such as hydrogarnet (Ht), C-A-S-H, gehlenite, MgO, and Mg(OH)₂, identified according to references and their JCDPF files ( 38-41 38. Suescum-Morales, D.; Bravo, M.; Silva, R.V.; Jiménez, J.R.; Fernandez-Rodriguez, J.M.; Brito, J. de (2022) Effect of reactive magnesium oxide in alkali-activated fly ash mortars exposed to accelerated CO₂ curing. Constr. Build. Mater. 342, 127999. https://doi.org/10.1016/j.conbuildmat.2022.127999 .
39. Zhang, J.; Lv, T.; Han, Q.; Zhu, Y.; Hou, D.; Dong, B. (2022) Effects of fly ash on MgO-based shrinkage-compensating cement: microstructure and properties. Constr. Build. Mater. 339, 127648. https://doi.org/10.1016/j.conbuildmat.2022.127648 .
40. Kuenzel, C.; Zhang, F.; Ferrandiz-Mas, V.; Cheeseman, C.; Gartner, E. (2018) The mechanism of hydration of MgO-hydromagnesite blends. Cem. Concr. Res. 103, 123-129. https://doi.org/10.1016/j.cemconres.2017.10.003 .
41. Rodríguez-Navarro, C.; Hansen, E.; Ginell, W.S. (1998) Calcium hydroxide crystal evolution upon aging of lime putty. J. Am. Ceram. Soc. 81 [11], 3032-3034. https://doi.org/10.1111/j.1151-2916.1998.tb02735.x .
). All samples of AAS possess similar phase compositions. With the increase of MgO, more Mg(OH)₂ appears to be present in AAS pastes. Nevertheless, no other apparent differences can be found among AAS mixtures at different ages.

3.6. Autogenous shrinkage

Figure 9 shows the results of autogenous shrinkage of five AAS mixtures over 56d. The development of autogenous shrinkage of AAS can be generally classified into three stages based on the tendency as marked in the figure. The first stage is located between the final setting and about 4d. Most of the AAS mixtures developed substantial expansive deformations during that stage. For MS3 with the lowest MgO dosage, its deformation during the first stage is immeasurable, indicating a low activated degree of GGBS due to insufficient activator. During stage I, the expansive deformations of AAS except for MS3 developed in almost linear tendencies with the curing age. In addition, the expansive deformation was sensitive to the content of MgO. With the higher dosage of MgO, there were higher expansive deformations in the mixtures. As a result, the highest autogenous expansion was recorded at 1713.5 με in MS15. Unlike traditional AAS activated by NaOH or water glass, which develops excessive autogenous shrinkage from the beginning of the final setting. The expansive properties of AAS activated by MgO can be attributed to the following two mechanisms. During the setting and hardening processes of AAS, the hydration heat released from the activation of GGBS induces the first part of expansive deformation. However, this part of expansive deformation lasts for short periods due to the quick setting of AAS (< 4 hours). The predominant mechanism for the expansion is exclusively the expansive nature of the hydration of MgO. During the dissolution process of MgO-activated slag, MgO could react with water and cause the formation of Mg(OH)₂. This part of the reaction produces high alkaline conditions for activating GGBS and contributes to expansive Mg(OH)₂. Ye et al. ( 28 28. Ye, Q.; Yu, K.; Zhang, Z. (2015) Expansion of ordinary Portland cement paste varied with nano-MgO. Constr. Build. Mater. 78, 189-193. https://doi.org/10.1016/j.conbuildmat.2014.12.113 .
) determined that during the transformation of MgO into Mg(OH)₂, the bulk volume of material could be increased at least by 2 times.

After this initial period, autogenous shrinkage became more obvious since the dissolution of MgO almost finished in Stage I. The length of the MgO-activated slag started to reduce, and Stage II occurred. As shown in Figure 9 , the autogenous expansion of AAS decreased quickly in about 10 days, despite the dosage of MgO. Compared with Portland cement, AAS normally possesses finer and denser pore structures, which inevitably cause high autogenous shrinkage due to the high pore pressure. Besides the self-desiccation of AAS, the consumption of ions during hydration also induces shrinkage. For the MgO-activated slag, another mechanism for the chemical shrinkage would be the consumption of the expansive Mg(OH)₂.

Stage III occurred in around 14 days, which could be assumed as a steady period of MgO-activated slag. In this period, the deformation of AAS tends to be stable except for low dosages such as MS3 and MS6. This can be attributed to the benefit of expansive Mg(OH)₂. Especially, the final shrinkage of the group MS9 almost reached zero. In the contrast, for AAS with lower dosages of MgO (MS3 and MS6), the autogenous shrinkage continuously developed over time. As a result, the final deformations were recorded as -1546.4με and -1138.1με for MS3 and MS6, respectively.

3.7. Drying shrinkage

Figure 10 shows the drying shrinkage of AAS activated by MgO over 56d. As shown in the figure, the addition of MgO could also significantly mitigate the drying shrinkage of AAS. In addition to the low dosage of MgO like in MS3, the drying shrinkage developed continuously to 2700με at 56d. The drying shrinkage of AAS was significantly altered when the activator MgO is used with more than 9 wt.% of precursors, with little drying shrinkage in the mixtures for the group MS9, while MS12 and MS15were even in expansive conditions at 56d, indicating that AAS shrinkage can be efficiently controlled by MgO activation. Considering the autogenous shrinkage of AAS shown in Figure 9 , MS9 might be the optimum group to achieve zero shrinkage including autogenous and drying for the application of alkali-activated slag.

It has been reported that the large drying shrinkage in AAS is caused by the structural incorporation of alkali cations in C-A-S-H. As a result, the stacking regularity of C-A-S-H layers is reduced, causing the C-A-S-H easier to collapse and redistribute during the drying process ( 16 16. Ye, H.; Radlińska, A. (2016) Shrinkage mechanisms of alkali-activated slag. Cem. Concr. Res. 88, 126-135. https://doi.org/10.1016/j.cemconres.2016.07.001 .
, 38 38. Suescum-Morales, D.; Bravo, M.; Silva, R.V.; Jiménez, J.R.; Fernandez-Rodriguez, J.M.; Brito, J. de (2022) Effect of reactive magnesium oxide in alkali-activated fly ash mortars exposed to accelerated CO₂ curing. Constr. Build. Mater. 342, 127999. https://doi.org/10.1016/j.conbuildmat.2022.127999 .
). A slight expansion is expected for an engineered AAS for practical application. The benefits of MgO on drying shrinkage mitigation can be summarized in two aspects. Firstly, the formation of expansive hydration products of Mg(OH)₂ contributes to the major expansive values. Secondly, with the increased amount of MgO, there are more alkali compounds to accelerate the dissolution of GGBS, thus increasing the hydration degree of the slag and generating more hydration products. As a result, there is a denser microstructure in the AAS mixtures activated by a high amount of MgO, which will reduce moisture losses during the drying process. This can be confirmed by the water loss during the drying process until 56d, as shown in Figure 11 .

It can be found that AAS lost most of its moisture during the first week, regardless of MgO content. After 7d, the moisture in the samples with a higher amount of MgO reduced slower compared with MS3, possibly because the AAS activated by higher dosages of MgO developed a denser microstructure, as indicated by the pore structure analysis. In addition, the presence of hydrophilic, un-hydrated MgO may also be beneficial to retain moisture during the drying process.