A Study on the Application of Recycled Concrete Powder in an Alkali-Activated Cementitious System

RCP has certain application potential in alkali-activated cementitious material in terms of improving strength and inhibiting rapid hydration process


Introduction
Environmental problems caused by construction and demolition waste (CDW) are becoming increasingly serious worldwide. It is estimated that more than 10 billion tons of CDW is generated globally each year [1], while waste concrete and clay bricks account for about 70-80 % of CDW [2,3]. The annual output of construction waste in China exceeds 2 billion tons, with an annual growth rate of 8%. In recent years, China has issued a series of policies to regulate the utilization of recycled aggregate (RA) and recycled fine powder (RP) [4][5][6][7], including "Recycled Coarse Aggregate for Concrete" (GB/T 25177-2010), "Recycled Fine Aggregate for Concrete and Mortar" (GB/T 25176-2010), "Technical Specification for Application of Recycled Aggregate" (JGJ/T 240-2011), and "Recycled Fine Powder for Concrete and Mortar" (JG/T 573-2020). However, the situation concerning CDW utilization is still not optimistic. By the end of 2021, in 35 pilot cities (districts) of China, about 50% of the CDW has been recycled, which is well below the average recycling rate of more than 90% in developed countries and regions [8,9]. Therefore, it is very meaningful to study and utilize CDW.
Recycled fine powder refers to the particles generated in the process of preparing recycled aggregate from construction wastes such as concrete, clay brick and tile, which normally exhibit potential pozzolanic reactivity [10,11], mainly involving recycled concrete powder (RCP), recycled brick powder (RBP) and some mixed powder. Some researchers [11][12][13][14][15] have used CDW powder instead of cement in concrete. The results show that the influence of an appropriate amount of CDW powder on the mechanical properties and microstructure of concrete is acceptable, but it is of great significance in cost saving and resource reuse. Zhao et al. [16] mixed 30% clay brick powder (CBP) of different fineness with cement to prepare mixed cement and found that CBP reduced the hydration heat and hydration heat release rate of paste at all stages, and CBP with small particle size could promote the early hydration of cement and reduce its setting time. Chen et al. [17] successfully prepared low-exothermic cement by mixing waste concrete recycled powder (WCRP) with ordinary Portland cement, and the compressive strength was improved by adding fly ash (FA) and silica fume (SF). RP is considered an inert substance before physical or chemical excitation. After mechanical grinding, some unhydrated cement particles and active substances of RCP are exposed, which can be more fully involved in the reaction, and the finer RP shows better pozzolanic activity and filling effect [10,18]. Some researchers used RCP to partly replace the cement-based cementitious materials used to prepare concrete. It was found that the optimal replacement rate of recycled concrete powder was about 30%. When the replacement rate of recycled concrete powder was 30%, the compressive strength of the RCP mortar reached 75.73% of cement mortar [19]. Using recycled powder instead of cement as a cementitious material reduces the chloride ion permeability of concrete [20,21]. Ma et al. [20] found that when the replacement of RCP was 30%, the chloride permeability of concrete was the lowest. Xiao et al. [15] studied the mechanical properties of concrete using different replacement amounts of recycled concrete and clay brick powder (RP). The results show that the cement mixture containing RP has a higher hydration rate than the cement mixture without RP.
Alkali-activated cementitious material refers to a kind of cementitious material produced by the reaction of aluminosilicate solid raw materials (e.g., slag, fly ash and metakaolin) with pozzolanic activity or potential hydraulicity with an alkaline activator [22]. The study on alkali-activated regenerated powder shows that the high pH environment brought by an alkaline activator can activate RP more effectively and generate abundant (N,C)-A-S-H gel [23][24][25], which makes the microstructure of the matrix denser and provides higher mechanical properties. The experiment showed that the activity effect was as follows: chemical activation > mechanical activation [26]. Robayo et al. [27] used NaOH + Na 2 SiO 3 to stimulate 100% red clay brick waste, and cured it at room temperature for 28 days; the strength of the specimens reached 54.38 MPa. Vásquez et al. [28] mixed 10% metakaolin and 90% CDW, activated by NaOH and Na 2 SiO 3 , and cured it at room temperature for 28 days; the strength reached 46.4 MPa.
RCP, as an auxiliary cementitious material, has been studied more in the OPC system. Compared with the production process of Portland cement, the process of alkali-activated cementitious material has the characteristics of low energy consumption, low carbon emission and low cost. In this work, the application potential of RCP as an alternative material in an alkali-activated cementitious system was systematically studied. Here, an RCP with basic material characterization was applied to alkali-activated slag in different proportions, and the fluidity, rheological properties and mechanical properties of the paste were characterized and discussed. The hydration products and their mechanism were characterized by XRD, SEM-EDS, thermogravimetric analysis and hydration thermal analysis. The action mechanism of RCP as an alternative cementitious material in an alkali-activated system was discussed. The results will help to improve the utilization of concrete waste. It provides more possibilities for recycled fine powder to replace cement as a cementitious material, and provides a strong reference for the study of recycled fine powder in alkali-activated cementitious systems.

Materials
The cementitious materials used involved a 42.5 ordinary Portland cement, a slag with a grade of S95 (GB/T 18046-2017) and an RCP. The RCP came from the construction concrete waste of Hebei Langfang Wanchen Environmental Protection Technology Co., Ltd. After crushing and grinding, the concrete waste was passed through a 100-mesh sieve to obtain a powder with a particle size less than 0.15 mm. The preparation process is shown in Figure 1. A NaOH solution with 9% Na 2 O content was used as the alkali-activator. An ISO sand, in accordance with the GB/T 17671 standard, and laboratory tap water were used to prepare the mixture. The chemical compositions of the cement, slag and RCP are shown in Table 1. The XRD pattern, particle size gradation curve of RCP, and SEM images of the RCP, are shown in      It can be found, from Table 1, that the chemical composition proportion of RCP is close to that of slag, with the main oxides of CaO, SiO 2 and Al 2 O 3 . As shown in Figure 2, RCP mainly contains SiO 2 , CaCO 3 , Ca(OH) 2 , C 2 S and aluminate minerals. It can be seen from Figure 3 that the particle size distribution of RCP is mainly between 0.6 µm and 400 µm. From Figure 4, after grinding, the RCP particles are observed with fewer edges and corners than slag, with finer particles and better gradation. The water requirement ratio of RCP is 110% (determined by the ratio of the water requirement of the sample containing RCP and the reference sample that reached the specified fluidity range), which is close to that of slag (108%).

Preparation of Specimens
The mortar specimens were prepared according to the mix proportions in Table 2 and in accordance with "Cement Test Methods-Determination of Strength" (GB/T 17671-1999). The binder-sand ratio was 1:3, the alkali-binder ratio was 0.5, and the water-binder ratio was 0.5. The content of the RCP in the cementitious materials was 0%, 10%, 20%, 30% and 40%. The size of the mortar specimen was 40 mm × 40 mm × 160 mm, curing for 3 days, 7 days, and 28 days at 20 • C ± 2 • C and relative humidity no less than 90%.  -2005). The mechanical properties were tested according to the "Method of Testing Cements-Determination of Strength" (GB/T17671-2021). The pictures of the compressive strength and flexural strength tests of the cement mortar are shown in Figure 5. The composition of the product was analyzed by XRF using a Bruker D8 ADVANCE (NASDAQ: BRKR, Billerica, Massachusetts, USA). The microstructure of the samples was analyzed by SEM-EDS using a JEOL JSM-7500F scanning electron microscope (Japan electron optics laboratory, Tokyo, Japan). Thermogravimetric analysis was carried out by a TAQ600 differential thermogravimetric synchronous thermal analyzer (WATERS TECHNOLOGY (CHINA) LIMITED, Hong Kong, China, 2014).

Hydration Heat Analysis
The hydration heat was tested by a TAM air 8-channel isothermal calorimeter (WA-TERS, Milford, DE, USA, 2019). The mix proportions of the paste sample were consistent with that of the cementitious material of specimens for the mechanical property test. For every sample, about 10 g of prepared paste was placed into the ampoule, sealed, and placed into the corresponding channel. The isothermal calorimeter collected the heat flow value automatically within 72 h.

Fluidity
As shown in Figure 7, the fluidity of the mortar decreases with the increase in RCP content. Compared with A0, the RCP content in the mortars of series A, from 10% to 40%, reduces the fluidity of the mortar by 1.5-3.9%; however, the fluidity of the mortar of series A is higher than that of series B, generally. Compared with Portland cement particles, the shape of slag is complex and has a poor grain, the surface is rough and porous, which increases the water requirement and reduces the fluidity of the cement mortar. For the alkali-activated slag system, the fluidity of the mortar first increased and then decreased with the content of RCP. The fluidity of the mortar with 10~40 % RCP content was 2~8.5 % higher than that without RCP. This is mainly because the passivated RCP particles are more rounded than the slag particles and play a ball effect. When the content of RCP exceeds 30%, the fluidity begins to decline, which indicates that the increase in water requirement brought by excessive RCP has a significant impact on the fluidity of the mortar. Therefore, for workability, a content of less than 30% RCP is appropriate.

Rheological Properties
The rheological properties curve fitting results are shown in Figure 8, in which all the correlation coefficients R2 are greater than 0.99. It can be found from Figure 8c that the addition of RCP reduces the plastic viscosity of the paste in both series A and B, but the influence on series B is greater than that of series A. With the increase in RCP content, the hydration products decreased, the cohesion of the slurry decreased, and the plastic viscosity decreased. However, the hydration rate of slag in an alkaline environment was higher than that of cement slurry, and the influence of RCP on internal structure flocculation was greater than that of cement slurry. The yield stress results are shown in Figure 8b. The yield stress of series B first decreased and then increased and decreased again. The RCP particles look more rounded than slag particles, and a small amount of RCP can reduce the internal friction between mortars. Meanwhile, the RCP particles in an alkali-activated slag system play a disintegrating effect on the flocculation structure formed in the hydration process [29], thus reducing the yield stress. With the increase in RCP content, the microaggregate effect of RCP is obvious, which enhances the cohesion between the slurry and increases the yield stress. However, when the content exceeds 30%, excessive inactive particles destroy the connection between the products [30], reduce the adhesion, and lead to a reduction in yield stress. In Portland cement paste, the addition of RCP increases the yield stress, and the cement hydration is slower than that of the alkali-activated system in the early stage. The RCP absorbs part of the free water, and the distance between the particles becomes smaller, increasing the friction and adhesion between the particles [31].

Mechanical Properties
The influence of RCP content on mechanical properties is shown in Figure 9. It can be seen from Figure 9a,b that the increase in RCP content reduces the compressive and flexural strength of cement mortar. With the RCP content from 10% to 40%, the compressive strength of A0 at 28 days decreased by 15%, 25%, 40% and 46%, while the compressive strength of A0 at 3 days decreased by 6%, 17%, 28% and 44%. This implies that RCP has little effect on the early strength of cement mortar. Oliveira et al. [11] believed that the existing hydration products in RCP become the crystal nucleus of the product in the early hydration process of cement. However, because the activity of RCP is lower than that of cement particles, excessive RCP dilutes C 2 S and C 3 A in the cementitious system, leading to the reduction in hydration products and the later strength. Some researchers [15,32,33] believe that the replacement of Portland cement by RCP should not exceed 30%.
It can be found from Figure 9c,d that the addition of RCP first increases and then decreases the compressive strength and flexural strength of alkali-activated slag mortar. Due to its round grain shape and gradation, RCP plays a filling role in the early hydration, and the existing hydration product particles become the growth point of new gel substances. The free Ca 2 + brought by RCP promotes the generation of C-S-H gel and C-A-S-H gel [34]. In an alkaline environment, the glassy substance on the surface of slag particles can be rapidly depolymerized by OH -, and the presence of Ca 2+ accelerates dissolution and hardening [35], combining with uniformly dispersed RCP particles to form a dense structure. Liu [36] believed that RCP could be used as an ultra-fine aggregate in cementation materials to improve particle gradation, inhibit crack propagation, and to form a better microstructure. However, when the content of RCP exceeds 30%, the filling effect of RCP on the microstructure is not enough to make up for the strength loss caused by the reduction in hydration products. Meanwhile, the increase in water requirement also affects the integrity of the hydration process of cementing materials, leading to strength decline. It is worth noting that the activity index of RCP from the standard curing of 28 days (the ratio of 28 days for the compressive strength of mortar containing 30% SCM to 28 days for the compressive strength of mortar without SCM) reaches 91%, which in accordance with the requirements of the SCM activity index regulation in GB/T1596, should exceed 70%. This indicates that an alkaline environment contributes to activating the activity of RCP.

XRD Analysis
After 28 days of standard curing, the samples were analyzed by XRD, as shown in Figure 10. By comparing A1 with B0-B4, it can be observed that there is an obvious Ca(OH) 2 diffraction peak in A1, and there are SiO 2 crystals and CaCO 3 crystals formed via carbonation in all of the six series of samples. The SiO 2 in the sample comes from sand particles in RCP, and the diffraction peak increases with the increase in RCP content, indicating that the low activity SiO 2 particles in RCP cannot participate in the alkaliactivation reaction, even after 28 days of curing. Yang [37] pointed out that SiO 2 cannot react with other substances in the hydration process, which only play a micro-aggregate effect in the hydration process. Additionally, the hydrotalcite phase is also found in B0-B4. At 2 θ of 20-40 • , there is a long mantle-shaped diffraction peak, indicating that there are more amorphous substances in the product. With the blend of RCP, the peak pattern and number of B1-B4 are basically the same as that of B0, indicating that the RCP has little effect on the type of hydration products. However, 10% of RCP causes a slight increase in the C-S-H peak value near 29 • , indicating that the alkaline environment would depolymerize some active substances in RCP and promote the formation of gel substances. Some researches [38,39] show that the small particles in RCP can fully react and dissolve in an alkaline environment, while the large particles only partially react. This also proves that when RCP content exceeds 30%, excessive inactive particles will destroy the gel network [30,39], leading to a significant decline in strength.

Thermogravimetric Analysis
The TG and DTG curves of the samples after standard curing for 28 days are shown in Figure 11. From room temperature to 300 • C, the mass loss in A1 is related to the adsorption water removal of C-S-H, AFt and AFm, while the mass loss in B0-B4 was due to the dehydration of C-S-H and C-A-S-H gel [40,41]. A weak endothermic peak can be found in the B series from 200 • C to 400 • C due to the decomposition of the hydrotalcite phase [42], which can be observed in the XRD patterns in Figure 7, while AFt and AFm are not found. The mass loss of B0-B4 is 18.46%, 21.61%, 20.76%, 19.47% and 24.17%, respectively. The increase in RCP content leads to more chemically bound water, which indicates that RCP plays a pozzolanic effect. On the one hand, the active SiO 2 and Al 2 O 3 in RCP react directly with lye to form C-S-H and C-A-S-H gels. On the other hand, the Ca 2+ in RCP promotes the formation of C-A-S-H [17]. The difference between mass loss is smaller than the difference between strength decrease, for which Kim et al. [43] believed might be related to the different degrees of C-S-H generated, resulting in the difference between water losses. At about 450 • C, there is an obvious endothermic peak in A1, which is from the dehydration decomposition of Ca(OH) 2 , while there is no obvious endothermic peak in the B0-B4 sample. This indicates that Ca(OH) 2 is not generated from alkali-activated slag. The endothermic peaks of all samples appear between 600 • C and 800 • C, and the mass loss varies from 0.68% to 2.7%; this mainly comes from the decomposition and transformation of C-S-H and C-A-S-H and the decomposition of carbonate [40].

SEM-EDS Analysis
The SEM images and EDS results of the samples are shown in Figure 12 and Table 3. They show that the strengthening of the matrix is obvious, especially in B1 and B2, in which there are no large numbers of exposed unreacted RCP particles. With the gel phase precipitating and hardening, it eventually binds non-reactive RCP particles to strengthen the matrix. The strengthening of the matrix may also be attributed to the decomposition of calcite in RCP. Bassani et al. [44] noted that fine-grained calcite can be reprecipitated in an alkaline condition, which has a strengthening effect. Lamellar C-A-S-H gel and amorphous gel (C-S-H) can be observed. In contrast, B4 is obviously looser in structure than the others, and the connection between the gels is loose, which is caused by the reduction of hydration products in the microstructure. Wu et al. [45] found that the SEM images of RCP can capture more microcracks and pores, especially when the content of recycled concrete powder is high. The RCP particles were encapsulated by C-S-H due to the nucleation and filler effect of RCP. It has also been mentioned in XRD analysis that too many unreacted particles will destroy the gel network, which leads to a decrease in mechanical strength [31,40]. The EDS results in Table 3 also show that the Ca/Si of sample B4 is higher than that of B0-B3, and the increase in Ca/Si will increase the defect degree of the silicon chain in C-S-H [46,47], resulting in a decreased stability of the layered structure and a decline in mechanical strength. In general, it is proved that with the increase in RCP content, the microstructure of the matrix will first become dense and then loose.

Hydration Heat Analysis
The influence of RCP content on the hydration exothermic behavior of samples is shown in Figure 13, in which Figure 13a,b are the cumulative exothermic curves and the hydration exothermic rates of the samples within 72 h, respectively; Figure 13c-f are the local amplification diagrams of the samples in Figure 13a,b, respectively. The hydration heat release process of alkali-activated slag can be divided into the initial reaction stage, induction stage, acceleration stage, and deceleration stage, which is similar to the hydration heat release of cement [48]. It can be seen from Figure 13a,b that the cumulative heat release of A1 is greater than that of B0-B4 after about 40 h, and the hydration heat release takes a longer time to reach the induction and acceleration stages, which implies that alkaliactivated slag exhibits the characteristics of lower heat and quicker setting than Portland cement. It can be found from Figure 13c,d that the cumulative heat release decreases with the increase in RCP content. When the content increased from 0% to 40%, the total cumulative heat release was 220.98 J/g, 220.74 J/g, 199.52 J/g, 195.40 J/g and183.11 J/g, respectively, which is due to the reduction of reactive substances in the hydration system of series B. The initial stage of the alkali-activated slag reaction is short and can be completed in a few minutes. At this stage, lye dissolves the active substances in the slag and RCP, which produces C-S-H gel and C-A-S-H gel. The amorphous gel products generated by the reaction quickly wrap the active particles, blocking the contact between OHand the active substance. This means the reaction entered the induction period at about 0.8-1.5 h, during which the hydration rate gradually decreased, and the cumulative heat release decreased. The second reaction peak is shown in Figure 13f. At this stage, numerous hydration products were generated, and the hydration heat release rate increased. The inactive particles in RCP become the growth sites of new hydration products, making the hydration products cross-connect to form a dense microstructure. However, with the increase in the content of RCP, the hydration heat release rate decreases gradually. Obviously, a 10-40% RCP content delayed the induction period of the reaction, and the accelerated reaction was delayed by 0.1 h-0.5 h. The peak rate of hydration heat release during the accelerated reaction decreased by 0.78 mW/g, 2.26 mW/g, 3.59 mW/g and 4.40 mW/g, respectively. From just entering the acceleration stage to the emergence of the peak of the acceleration stage, the duration of B0-B4 was about 0.8 h, 1 h, 1 h, 1.5 h and 3 h, respectively, indicating that RCP had an obvious effect on inhibiting the rapid hardening of the alkali-activated slag. The delayed effect of RCP on the second hydration heat release during the peak of the alkali-activated slag is of positive significance to inhibit the quick setting of alkali-activated cementitious materials.

Conclusions
RCP can effectively improve the fluidity of alkali-activated slag to a certain extent, which is due to the slowing down effect of RCP on the hydration process of alkali-activated cementitious materials and the "ball bearing effect" of the RCP particles.
The addition of RCP leads to the compressive strength of alkali-activated slag cementitious material first increasing and then decreasing; the compressive strength increased by 13% and 3% with an RCP content of 10% and 20%, respectively, after curing for 28 days.
The incorporation of RCP does not change the types of hydration products in alkaliactivated slag; the main hydration products are C-S-H gels and C-A-S-H gels. The active particles in RCP are depolymerized in an alkaline environment, and the inactive particles become the nucleation sites of the generated products, which combine with the gel material to form a dense microstructure.
RCP has the potential to slow down the quick reaction between the alkali solution and slag.
In addition, the activation methods of RCP include thermal activation and mechanical activation, as well as chemical activation. In this work, the activity of RCP was activated by grinding and alkali-activation; however, thermal activation and the effect of grinding time on the properties of RCP as an alternative cementitious material requires further study.