Microstructure, hydration process, and compressive strength assessment of ternary mixtures containing Portland cement, recycled concrete powder, and metakaolin

EI检索SCI升级版 环境科学与生态学1区SCI基础版 环境科学与生态学1区IF 11.1
https://doi.org/10.1016/j.jclepro.2023.140085 Get rights and content

Highlights

  • Three different sources of RCP were compared.

  • The use of RCP and MK increases both yield stress and plastic viscosity.

  • The hydration of the pastes was studied using isothermal calorimetry, XRD, and TG.

  • The mechanical performance of RCP + MK pastes is equal or superior to the reference.

Abstract

The aim of this study is to evaluate the effect of recycled concrete powder (RCP) and metakaolin (MK) on the microstructure, hydration, and compressive strength of ternary mixtures containing Portland. The methodology involved producing cementitious pastes with different combinations of RCP + MK, where Portland cement was substituted by 30 %, and three sources of RCP were considered: construction, demolition, and laboratory. The materials were characterized physically, chemically, and mineralogically. The characterization of the fresh state properties was done using a rotational viscometer. The hydration of the pastes was evaluated by isothermal calorimetry, X-ray diffraction, and thermogravimetry. The compressive strength and elastic modulus were determined for 1, 7, 28, and 120 days. Finally, an analysis of the microstructure of the prepared pastes was conducted by scanning electron microscopy. The results show that RCP and MK have negative influence on the fresh state properties, increasing both the yield stress and plastic viscosity. The additions also promote early hydration of Portland cement (first hours), which decreases due to the dilution effect. The quantity of hydrated products tends to be comparable to the reference, except for calcium hydroxide, which is reduced due to the pozzolanic activity of MK. The compressive strength and elastic modulus of the RCP + MK mixtures can be equal and/or superior the reference, demonstrating a synergistic effect between both materials. The use of RCP + MK mixtures is presented as a technically viable alternative to replace Portland cement without compromising performance.

Keywords

Isothermal calorimetry
X-ray diffraction
Thermogravimetry
Compressive strength
Scanning electron microscopy
Sustainability

1. Introduction

Civil construction is one of the industries with the greatest impact on the environment, as it requires the consumption of natural resources and energy, generates construction and demolition waste (CDW), and emits considerable amounts of CO2 (Liu et al., 2019; Sun et al., 2020; Aslam et al., 2020). The reuse of CDWs is highly recommended as a sustainable strategy to mitigate these negative impacts, decreasing the amount of waste in landfills (Neves and De Brito, 2022). It is estimated that concrete represents 30 % of the total CDW, being one of the CDW's main components (Xiao, 2018). Concrete waste can be recycled as aggregates by crushing and sieving, and the classification depends on the particle size: Φ ≥ 4.75 mm - recycled coarse aggregate (RCA); between 4.75 mm–150 μm, recycled fine aggregate (RFA), and ≤150 μm, recycled concrete powder (RCP) (Xiao et al., 2018; Santos et al., 2020; Wu et al., 2021a). Although RCP can be produced during the crushing processes of recycled concrete, it is generally obtained through a grinding process (Tang et al., 2020; Rocha and Toledo Filho, 2023).

On the other hand, Portland cement is the most widely used material in the construction industry; however, its production emits approximately 7–8 % of the world's total CO2 emissions (Environment et al., 2018). To reduce greenhouse gases emissions, the use of supplementary cementitious materials (SCM) as partial substitutes for Portland cement has also been proposed (Yang et al., 2015). Among the wide variety of materials proposed as SCM, there has been extensive research into the use of RCP in recent years (Liu et al., 2019; Li et al., 2021).

In general, RCP is composed of calcium carbonate (CaCO₃), quartz (SiO2), hydrated products, and non-hydrated cement particles, suggesting a potential use as an alternative binder (Kaliyavaradhan et al., 2020a; Wang., 2022a). During the crushing or mechanical grinding process to produce RCP, the cement mineral phases, such as alite (C₃S) and belite (C₂S), are exposed and can participate in the new hydration process (Chen et al., 2022). Conversely, the physical effects of RCP, such as filling and nucleation in cement matrices, have also been highlighted (Ma et al., 2021). However, RCP decreases the rheological properties of cement-based materials, which is due to its irregular microstructure and high-water demand (Ma et al., 2020; Duan et al., 2020a). The use of RCP in high percentages and with a particle size larger than that of cement results in the decrease of mechanical strength and durability of cement-based materials. According to He et al. (2020), a considerable decrease in compressive strength was observed when RCP was utilized with a particle size of 142 μm (D50). On the other hand, for a D50 RCP of 2.3 μm, it was advised to use a maximum of 30 % replacement of cement by RCP without compromising the mechanical properties. In a similar study, Horsakulthai (2021) employed RCP with granulometry comparable to that of Portland cement. The author reported a significant decrease in physical and mechanical properties with increasing RCP contents and recommended limiting the percentage under 20 %. In another study, Cantero et al. (2021) observed that the durability of cement-based materials is negatively impacted by high amounts of RCP (∼30 %), attributing this to a decrease in hydration products.

In order to improve the performance of RCP in cement-based matrices, various treatment methods have been proposed, such as thermal, carbonation, and alkaline activation, as well as mechanical (grinding) (Tang et al., 2020; Zajac et al., 2020). The mechanical treatment involves reducing particle size through a grinding process to achieve a granulometry similar to Portland cement (Wu et al., 2021b; 2022b). Grinding can enhance the nucleation effect of RCP during the hydration reaction and promote a filler effect, increasing the compactness of cementitious matrices (Rocha and Toledo Filho, 2023). The literature also suggests the use of thermal treatment, as high temperatures can alter the physical and chemical properties of particles (Wu et al., 2021c). It is recommended to use temperatures above 500 °C to achieve smaller particle size and generate reactive elements in the RCP (Wu et al., 2021a; Sui et al., 2020). The use of CO2 treatment has been demonstrated to enhance the microstructure of RCP particles (denser and less porous), promoting RCP utilization (Wu et al., 2021d; Kaliyavaradhan et al., 2020b). Mehdizadeh et al. (2021a) suggested that carbonated RCP has a finer particle size than untreated RCP, which can be attributed to the formation of fine CaCO3 particles and an improvement in particle distribution. Kaliyavaradhan et al. (2022) have indicated that the formation of CaCO3 can homogenize RCP particles, leading to a positive impact on the performance of cementitious matrices. In terms of alkaline activation, Zhang et al. (2022) utilized chemical activators to enhance the hydration process and mechanical properties of cement-based materials containing RCP.

The hydration process of cement with RCP has been previously investigated and reported (Chen et al., 2021). Initially, Portland cement and the active particles of the RCP react with water to form hydrated products, such as ettringite (AFt), calcium hydroxide (CH), and calcium silicate hydrate (C–S–H). Depending on the characteristics of the RCP, amorphous SiO2 and Al2O3 particles may then react with CH to form C–S–H and C-A-S-H through a pozzolanic reaction (Singh et al., 2019). From a physical perspective, RCP has a nucleation effect that provides extra surface for the formation of hydrated products, which can accelerate cement hydration, initially leading to a higher heat of hydration (Sun et al., 2017; Bogas et al., 2019). Moreno-Juez et al. (2021) indicated that hydration kinetics is similar to the reference for 5 % and 10 % RCP, even presenting a higher cumulative heat for the first 6 h. Hou et al. (2021) pointed out that 10 % and 20 % RCP can provide a higher initial heat flow. On the other hand, Chen et al. (2022) stated that adding RCP up to 40 % reduced the cumulative heat released by 42.05 %. However, using fly ash (FA) and silica fume (SF) can attenuate this reduction, resulting in 23.13 % and 25.94 % less cumulative heat release, respectively.

The literature shows that RCP can improve the early hydration of cement. For this purpose, the use of different percentages and physical characteristics of the RCP, such as fineness and specific surface area (SSA), has been reported (Real et al., 2020; Liu et al., 2022a). Conversely, the utilization of RCP could potentially result in long-term adverse impacts on the hydration and mechanical strength of cement-based materials (Deng et al., 2021; Rocha et al., 2023). Therefore, in order to enhance these properties, some studies have suggested the incorporation of additional SCMs, such as FA, SF, and metakaolin (MK), to RCP (Chen et al., 2022; Wu et al., 2022a). However, there is limited research on cement-RCP hydration with the addition of SCMs. Previous studies have found that RCP possesses physical and chemical characteristics to be used as SCM, but the hydration of ternary mixtures between Portland cement, RCP, and other materials has been little explored. Therefore, this research proposes using MK as a pozzolan in Portland cement and RCP mixtures, considering that MK is also recycled from solid waste and has the potential to enhance the performance of cement-based materials (Homayoonmehr et al., 2021; Raheem et al., 2021). Thus, using MK and RCP to prepare cement-based materials could represent a technically viable and environmentally friendly alternative in construction industry (Singh et al., 2019; Wu et al., 2022a; Sun et al., 2022).

For these reasons, this study aims to evaluate the hydration and compressive strength of ternary Portland cement, RCP, and MK mixtures using cement pastes and three RCP sources (construction, demolition, and laboratory). The selection of various RCP sources is based on the assessment of material feasibility in relation to its availability. In the literature, RCP has been evaluated from both CDW (Deng et al., 2021; Hou et al., 2021) and laboratory-tested specimens (Cantero et al., 2022; Ma et al., 2021). While the chemical and physical properties of RCP differ based on the origin of the waste, the results have proven to be favorable when considering low replacement percentages and small particle size (Tang et al., 2020). Despite this, studies have not directly compared the performance of cementitious matrices using different sources of RCP, which suggests the potential for wider utilization of RCP. This article addresses this gap by examining three RCP sources in ternary mixtures, with the objective of determining whether the physical and chemical characteristics of each source impact the microstructure, hydration, and compressive strength of the ternary mixtures. The rheology of the mixtures was studied using a rotational viscometer. The hydration of the pastes was evaluated by isothermal calorimetry (up to seven days), X-ray diffraction (XRD) (1 and 28 days), and thermogravimetry (TG) (1, 7, 28, and 120 days). The compressive strength and elastic modulus were determined for the ages of 1, 7, 28, and 120 days. Finally, a study on the microstructure of the mixtures was carried out by scanning electron microscopy (SEM) for 1 and 28 days. The present investigation proposed the reuse of CDW with the use of other SCMs, specifically RCP and MK, presenting an alternative to reduce the consumption of Portland cement in the construction industry.

2. Materials and methods

2.1. Materials

The cementitious materials used were CP II F-32 cement (LafargeHolcim), MK (Metacaulim do Brasil Indústria e Comércio Ltda.), and RCP. The first two were materials produced and sold locally in Brazil. The water used to make the paste was deionized.

For the production of RCP, three sources were used: construction (RCP-C), residues of hardened concrete from a precast plant; demolition (RCP-D), obtained from the demolition of a hospital in Rio de Janeiro, and laboratory (RCP-LAB), residues of concrete specimens tested in the laboratory. Initially, the material was collected after sieving with a maximum size of 150 μm. Subsequently, the material underwent a grinding process for 30 min in a ball mill in order to obtain a particle size distribution close to Portland cement. This was established based on the recommendations of the literature to use RCP as SCM (Tang et al., 2020; Rocha and Toledo Filho, 2023).

The cementitious materials were characterized physically, chemically, and mineralogically. The particle size distribution was determined by laser diffraction. Density was found using a helium picnometer. The SSA was determined using Blaine fineness, according to ASTM C204 (ASTM, 2019). The particles' morphology was examined by scanning electron microscopy, using a HITACHI-TM 3000 equipment. The chemical composition was found by X-ray Fluorescence (XRF) using a Shimadzu EDX-720 spectrometer. The loss on ignition was obtained following the recommendations of NBR NM18 (ABNT, 2012). The mineralogical composition was determined by X-ray diffraction (XRD) using a Bruker-AXS D4 Endeavor diffractometer.

2.2. Mix design

For the current investigation, 11 mixtures were prepared, including the reference cement paste (100 PC). Portland cement was replaced by 30 % (mass) of RCP and MK, with this percentage based on the literature (Tang et al., 2020; Rocha and Toledo Filho, 2023). The selection of this replacement percentage is based on both environmental and technical considerations. From an environmental perspective, the higher the replacement of Portland cement with RCP, the lower the resulting CO2 emissions (Oliveira et al., 2023). Technically, although research has shown that high levels of RCP can have a negative impact on the performance of cement-based materials, some authors have proposed an approximate range of 20 % (Tang et al., 2020). It should be noted that the 30 % replacement also includes another SCM, MK, which has been recommended for use in cementitious mixtures (Raheem et al., 2021; Weise et al., 2023), even when combined with RCP (Wu et al., 2022a). Therefore, the proposal aims to identify a suitable combination of both materials that can be practically applied. For each RCP source, three mixtures were considered: 30 RCP, 20 MK + 10 RCP, and 10 MK + 20 RCP. These percentages are derived from guidelines for the utilization of RCP and MK as reported in the literature, which fall within the range of 10 %–20 % (Siddique and Klaus, 2009; Tang et al., 2020). The w/c ratio of the cement pastes was 0.4. Table 1 summarizes the proportions by mass of the pastes.

Table 1. Proportions of the cement pastes (by mass).

MixturePortland cementMKRCP-CRCP-DRCP-LAB
100 PC100
70PC30MK7030
70 PC30C7030
70PC20 MK10C702010
70PC10MK20C701020
70PC30D7030
70PC20MK10D702010
70PC10MK20D701020
70PC30LAB7030
70PC20MK10LAB702010
70PC10MK20LAB701020

A Chandler Engineering™ paddle mixer was utilized to mix the pastes. Initially, the cementitious materials were blended with the water at a speed of 1000 ± 50 rpm for 2.5 min, followed by a 1-min stop time, to conclude at a high speed (3000 ± 50 rpm) for 2.5 min.

2.3. Methods

This section is organized into subsections defining the methods used for: a) rheological properties, b) isothermal calorimetry, c) XRD, d) TG, e) compressive strength, f) elastic modulus, and e) morphology analysis.

2.3.1. Rheological properties

The rheological properties were determined by a rotational viscometer, model HD DV-II Ultra (Brookfield) with two vanes, V-73 and V-75. Once the pastes were prepared, they were immediately placed in a 5.63 cm diameter beaker. The shear rate protocol developed by Tinoco et al. (2023) was applied to determine the flow curves (Fig. 1). The protocol consisted of an initial phase of pre-shearing for 200 s, varying the speed from 0 to 0.2 s−1. Subsequently, the speed was kept constant (0.2 s−1) for 70 s to guarantee the mixture's homogeneity. To obtain the flow curves, the rate was increased in 20 stages of 10 s until reaching a value of 42.5 s−1, referred to as the ascending ramp. Finally, the speed was reduced to 0.2 s−1 in 20 steps of 10 s, referred to as the descending ramp.

Fig. 1
  1. Download : Download high-res image (167KB)
  2. Download : Download full-size image

Fig. 1. Shear rate protocol. Based on Tinoco et al. (2023).(1)τ=τ0+μγ˙(2)τ=τ0+μγ˙+cγ˙2

Dynamic yield stress and plastic viscosity were obtained by regressing the data from the descending part of the flow curves. For the latter, the Bingham (Equation (1)) and modified Bingham (Equation (2)) rheological models were considered to determine the linear and non-linear behavior of the mixtures, respectively.Where τ is the shear stress (Pa), τ0 is the dynamic yield stress (Pa), μ is the plastic viscosity (Pa.s), γ˙ is the shear rate (s−1), and c is the second order parameter (Pa.s2).

2.3.2. Isothermal calorimetry

Isothermal calorimetry tests were performed to study the effects of RCP and MK on the hydration kinetics of Portland cement. Approximately 5 g of each paste was used in a glass ampoule, which was sealed and placed in a TAM Air isothermal calorimeter with 8 independent channels from TA Instruments. The test was conducted at 25 °C for 7 days (168 h). In the analysis of results, the pre-induction period was disregarded, as the pastes were prepared outside the calorimeter.

2.3.3. TG

TG was performed using a SD Instruments QT T600/DTA/DSC Simultaneous equipment from TA Instruments. A heating rate of 10 °C per minute from 35 °C to a final temperature of 1000 °C was considered. The test was conducted in a nitrogen atmosphere with a flow rate of 100 ml per minute. Ten milligrams of paste were used for the ages of 1, 7, 28, and 120 days. The TG test was also conducted for cementitious materials (Portland cement, MK, and RCP).

The content of portlandite (CH) was determined by calculating the amount of calcined base of the cement and/or binders at 1000 °C (excluding CaCO3) and the molecular mass ratio between CH and water. The quantity of chemically bound water was measured by the difference in mass lost at 45 °C and CH dehydration.

2.3.4. XRD

The mineralogical composition of the pastes was determined for 1 and 28 days using a Bruker-AXS D4 Endeavor diffractometer with Kβ-filtered Co Kα radiation. The measurements were made in the range of 5–70° 2θ in steps of 0.01°.

The process for preparing the samples for the XRD and TG tests consisted of placing 5 g of each paste into a sealed plastic bag. After the designated curing time for each test had passed, the hydration process of the pastes was stopped using an agate mortar and pestle with isopropanol alcohol, as recommended by Scrivener et al. (2018). The resulting solution was then filtered and stored in a vacuum desiccator until the day of testing.

2.3.5. Compressive strength and elastic modulus

The compressive strength of the pastes was determined for 1, 7, 28, and 120 days, in accordance with the recommendations of NBR 7215 (ABNT, 2019). Four cylindrical specimens were used per mix, with 2.5 cm of diameter and 5 cm of height, in a Shimadzu UH-F (100 kN) equipment. The axial deformation was calculated by means of the longitudinal variation of the specimens, using electric transducers (LVDT). The test displacement velocity was 0.1 mm/min, and the elastic modulus was calculated using ASTM C469 (ASTM, 2022a, ASTM, 2022b).

2.3.6. Morphology analysis

The microstructure of the pastes and the formation of hydrated products in the pastes for ages of 1 and 28 days were evaluated using a Hitachi TM3000 scanning electron microscope.

3. Results and discussion

This section presents the characterization of the materials, followed by the rheological properties, isothermal calorimetry, XRD, TG, compressive strength, elasticity modulus, and morphology analysis.

3.1. Characterization of materials

The particle size distribution of the materials used is presented in Fig. 2. It can be seen that RCP (C, D, and LAB) and Portland cement have similar granulometry; whereas the MK has a larger particle size. Table 2 verifies that the D50 of RCP and Portland cement is similar, close to 14 μm. RCP has finer particles (D10) than Portland cement and MK. However, RCP D90 is larger than Portland cement and comparable to MK D90 (∼60 μm).

Fig. 2
  1. Download : Download high-res image (226KB)
  2. Download : Download full-size image

Fig. 2. Distribution of the particle size of the materials.

Table 2. Physical properties of materials.

ParameterPortland cementMKRCP-CRCP-DRCP-LAB
D10 (μm)2.062.351.721.771.71
D50 (μm)14.1420.2814.4213.1913.82
D90 (μm)47.7959.0760.6757.2362.49
Specific gravity (g/cm³)3.052.712.682.662.61
Specific surface area (cm2/g)3762.771,2026.99252.228614.727988.73

In the literature, it is recommended that RCP have a fine particle size, similar to or smaller than that of Portland cement. This characteristic could improve the hydration and mechanical performance of cement-based materials (Wu et al., 2021e; Moreno-Juez et al., 2021). In the present investigation, with 30 min of grinding, the RCP (C, D, and LAB) achieved a similar D50 to that of Portland cement.

Table 2 also shows the density and SSA of the materials. Cement has the highest density of materials, followed by MK and RCP. Regarding the SSA, MK stands out above all materials, followed by RCP (C > D > LAB) and Portland cement. These data are consistent with previous works, since it is reported that RCP has a higher SSA than cement due to the irregular particles and hydration products present in RCP (Tang et al., 2020; Rocha and Toledo Filho, 2023).

Regarding the chemical composition, Portland cement has a high content of CaO, followed by SiO2, Al2O3, and Fe2O3 (Table 3). The cement meets the requirements for the CP II-F classification (Brazilian blended cement with carbonate material), since its loss on ignition (LOI) was 10.78 % (≤12.50 %) and the SO3 content was less than 4.50 %, as indicated by the NBR 16697 (ABNT, 2018). The MK presents a high content of Al2O3, followed by SiO2 and Fe2O3; this chemical composition is characteristic of the MK (Homayoonmehr et al., 2021; Raheem et al., 2021). The RCP is characterized by presenting the same oxides, but in different proportions. RCP-C has the highest CaO content, while RCP-D and RCP-LAB present a higher amount of SiO2. Subsequently, the Al2O3 and Fe2O3 compounds in the three RCP sources follow in quantity. The RCP cannot be classified as a pozzolanic material according to ASTM C618 (ASTM, 2022a, ASTM, 2022b), as the sum of the Fe2O3+Al2O3+SiO2 compounds is less than 70 % and the LOI is greater than 6 %.

Table 3. Chemical composition of materials (in %).

OxidePortland cementMetakaolinRCP-CRCP-DRCP-LAB
CaO67.9436.7621.4532.44
SiO210.3139.4534.3242.9237.02
Fe2O33.987.285.226.854.41
Al2O32.9446.018.9413.6510.18
SO32.941.112.702.102.33
K2O0.370.001.763.502.95
SrO0.320.010.160.120.21
TiO20.301.841.100.870.53
MnO0.080.160.340.240.14
ZnO0.040.020.020.020.03
ZrO20.280.190.180.03
Rb2O0.030.040.02
BaO0.48
LOI10.783.368.478.049.70

As reported by other investigations, the chemical composition of RCP is heterogeneous and varies according to the source materials. The SiO2 and Al2O3 come from the fine aggregates (Prošek et al., 2019; Yang et al., 2020), while the CaO originates from the hydration products, calcium carbonate and coarse aggregate (Wu et al., 2021a; Zhang et al., 2022; Sui et al., 2020). The chemical composition results indicate that RCP-D and RCP-LAB have a higher proportion of fine aggregates, and RCP-C could possess a higher amount of cement paste, CaCO3, and coarse aggregate.

The mineralogical composition of the materials is presented in Fig. 3. Clinker phases such as C2S, C3S and Ferrite (C4AF) are distinguished in Portland cement, as well as calcite peaks attributed to the mineral addition of Portland cement (limestone filler). The MK presents quartz and kaolinite as main minerals. The mineralogical composition of the RCP is similar between the three sources. The quartz and muscovite peaks can be attributed to the fine aggregates of the concrete (Wu et al., 2021a; Ma et al., 2022). The calcite can be attributed to the coarse aggregate and carbonation of the cement paste (Wu et al., 2021a, 2022b; Zhang et al., 2021). Although hydration products such as portlandite (CH) are reported in the literature (Real et al., 2020; Caneda-Martínez et al., 2021; Ma et al., 2022), only RCP-LAB presents a slight CH peak. This situation can be attributed to the carbonation process of the three sources (Wu et al., 2021d). Other studies have also detected the presence of clinker phases in the RCP (Zhang et al., 2021; Fang et al., 2021a; Keppert et al., 2021); however, the results show the absence of non-hydrated cement grains in the RCP (C, D, and LAB).

Fig. 3
  1. Download : Download high-res image (351KB)
  2. Download : Download full-size image

Fig. 3. Mineralogical composition of the materials.

Fig. 4 shows the binders' mass loss and derived mass loss (DTG). In Portland cement, three stages of mass loss can be distinguished (Fig. 4a). The first stage, up to 350 °C (∼0.30 %), is attributed to the dehydration of: a) hydrated calcium sulfates, b) AFt, and c) some phases of C–S–H (Sha et al., 1999; Heap et al., 2013). The second stage shows a mass loss (∼0.65 %) in the range of 350–450 °C, corresponding to the dehydration of CH; however, this dehydration can be considered insignificant (Bezerra et al., 2021). Finally, the third stage shows that, from ∼600 °C, decarbonation of CaCO3 occurs (Taylor, 1997) together with the most significant mass loss (10.10 %), which is within the limit for the type of cement used (ABNT, 2018). The MK presents a mass loss of ∼1.25 % up to 350 °C (Fig. 4a), which is attributed to water loss. On the other hand, in the range of 350–580 °C, there is a mass loss of ∼1.24 % that can be associated with the dehydroxylation of kaolinite, transforming it into reactive metakaolin (Paiva et al., 2018; Caballero et al., 2019).

Fig. 4
  1. Download : Download high-res image (514KB)
  2. Download : Download full-size image

Fig. 4. Thermal analysis: a) Portland cement and MK, and c) RCP-C, RCP-D, and RCP-LAB.

In the case of the RCP (Fig. 4b), an initial loss of mass is observed up to approximately 200 °C, due to the evaporation of water, decomposition of AFt, and hydration products present in the RCP (C–S–H) (Lu et al., 2018; Mehdizadeh et al., 2021b). For 350–450 °C, there is the presence and dehydroxylation of CH (Fang et al., 2021b; Wu et al., 2022c), which is only perceptible for RCP-LAB (∼0.39 %), a result that is confirmed by XRD patterns (Fig. 3). The mass loss due to combined water was 1.60 %, 1.14 %, and 2.02 % for RCP-C, RCP-D, and RCP-LAB, respectively. From ∼550 °C, a considerable mass loss is observed, which is attributed to the presence of CaCO3 (Tu et al., 2016; Mehdizadeh et al., 2021a), 6.76 %, 6.43 %, and 7.97 % for RCP-C, RCP-D, and RCP-LAB, respectively.

The morphology of the particles of the materials is presented in Fig. 5. The MK particles are characterized as being irregular and mostly amorphous (Fig. 5b) (Tchakouté et al., 2017). RCP particles appear irregular and have a rough surface that, together with microporosity, could affect the rheological and mechanical performance of RCP mixtures (Wu et al., 2022b; Yang et al., 2022). In all three sources, crystalline SiO2 particles can be observed, mainly, in addition to hydration products adhered to the surface of larger particles, characteristics which have also been reported in the literature (Xiao et al., 2018; Wu et al., 2021e).

Fig. 5
  1. Download : Download high-res image (1013KB)
  2. Download : Download full-size image

Fig. 5. SEM images: a) Portland cement, b) MK, c) RCP-C, d) RCP-D, and e) RCP-LAB.

3.2. Rheological properties

Fig. 6 presents the rheological behavior of the pastes, shear stress versus shear rate. As a complement, Table 4 summarizes the main fit parameters for the linear Bingham model and the modified Bingham model (polynomial second degree). In all cases, there was an increase in the parameters τ0 and μ when compared to the reference. On the other hand, the modified Bingham model presented a better fit, with R2 values greater than 0.99 for all pastes, while for the Bingham linear model, the R2 values were below 0.90. In this sense, the modified Bingham model was considered for the analysis.

Fig. 6
  1. Download : Download high-res image (296KB)
  2. Download : Download full-size image

Fig. 6. Flow curves for the pastes studied.

Table 4. Rheological parameters for the Bingham model and the modified Bingham model.

MixtureBingham modelModified-Bingham model
τ0 (Pa)μ (Pa.s)R2τ0 (Pa)μ (Pa.s)c (Pa.s2)c/μR2
100 PC36.020.5800.94332.851.025−0.010−0.0010.983
70PC30MK414.102.9420.996416.632.5750.0080.0030.997
70 PC30C92.471.0620.91184.012.288−0.028−0.0120.998
70PC20 MK10C228.002.1150.956216.683.756−0.038−0.0100.998
70PC10MK20C118.921.2010.938134.222.727−0.035−0.0130.996
70PC30D79.350.7350.94479.881.944−0.021−0.0110.989
70PC20MK10D195.312.1610.977187.183.339−0.027−0.0080.998
70PC10MK20D144.761.2000.892105.793.227−0.032−0.0100.999
70PC30LAB86.461.0200.93574.991.350−0.014−0.0100.990
70PC20MK10LAB192.422.0040.986187.072.780−0.018−0.0060.996
70PC10MK20LAB115.401.8340.960111.232.316−0.026−0.0110.996

The rise in τ0 indicates an increase in adhesion and friction between the Portland cement, RCP, and MK particles. The increase in μ suggests a higher resistance to the flow of the utilized cementitious materials (Jiao et al., 2017). According to Duan et al. (2020a), the use of RCP leads to an increase in τ0 and μ due to greater water absorption and reduced presence of free water (workability reduction). Fine particles of RCP have a filling effect that enhances packing density and μ, revealing improved compatibility and segregation resistance. In addition, Hou et al. (2021) highlighted that the rise in τ0, μ, and thixotropy enables the use of RCP in cement-based materials for 3D printing, where increased demands for viscosity and thixotropy are required.

The rheological behavior results can be directly attributed to the replacement of Portland cement by MK and RCP, since no superplasticizer additives were used. In general, the literature reports incorporating RCP affects cement-based materials' workability due to its porosity, irregular microstructure, and high SSA; these characteristics increase the demand for water (Prošek et al., 2019; Li et al., 2021; Ma et al., 2021). Indeed, Table 2 and Fig. 5 confirm the irregular microstructure and higher SSA of RCP, regardless of the source, compared to the cement. Therefore, using RCP influences the increase in the parameters τ0 and μ. For example, for pastes 70 PC30C, 70PC30D, and 70PC30LAB, the τ0 was 155.74 %, 143.16 %, and 128.28 % higher than the reference, respectively. The μ of 70 PC30C, 70PC30D, and 70PC30LAB was 123.22 %, 89.66 %, and 31.70 % higher than the reference, respectively. These adverse effects of RCP on the rheology of cement pastes were also reported by Sun et al. (2020) and Dun et al. (2021).

Furthermore, it is noteworthy that MK has a predominant effect on the rheology of the pastes, as the rheological parameters increase with the MK content. For example, paste 70PC30MK has the highest τ0 of all pastes, 1168.28 % higher than the reference. In fact, the influence of MK on rheology is more significant than that of RCP, as the τ0 of 70PC30MK is 395.93 %, 421.57 %, and 455.58 % higher than the pastes that used only RCP: 70 PC30C, 70PC30D, and 70PC30LAB, respectively. The effect of MK on the rheological properties of pastes can be mainly attributed to two reasons: a) the high SSA of MK, which is the material with the highest SSA (Table 2), and b) the pozzolanic activity of MK, which would increase the water demand and decrease the workability of the pastes (Sfikas et al., 2014; Janotka et al., 2010).

In the case of the RCP + MK pastes, the rheological parameters are higher than the reference and the pastes with 30 % RCP, with the 20 MK + 10 RCP as the most prominent, followed by the 10MK + 20RCP pastes. This increase is attributed primarily to the SSA. To a lesser extent, the SSA and the irregular shape of the RCP particles increase the water demand and generate friction, respectively, contributing to the increase of τ0 and μ (Ge et al., 2012; Hou et al., 2021).

It can be observed that the flow curves are similar between the pastes that present the same proportions of RCP and MK. However, the RCP-C source presents the highest rheological parameters in all cases, followed by the RCP-D and RCP-LAB sources. These results confirm SSA's influence on the pastes' rheology, as SSA has the order RCP-C > RCP-D > RCP-LAB (Table 2).

Among the presented results, it can be established that the substitution of Portland cement by MK and RCP increases both τ0 and μ parameters of the cement pastes. MK has a greater influence on the rheology of the pastes due to a double effect: SSA and pozzolanic activity. Whereas RCP, regardless of the source, mostly suggests a physical effect dependent on the SSA of the material.

3.3. Isothermal calorimetry

The results of the heat flow and cumulative released heat are presented in Fig. 7. Regardless of the RCP source, the heat flow and cumulative released heat exhibit similar behavior in all pastes, displaying the typical periods of induction, acceleration, and deceleration.

Fig. 7
  1. Download : Download high-res image (749KB)
  2. Download : Download full-size image

Fig. 7. a) Specific heat flow for RCP-C, b) cumulative heat for RCP-C, c) specific heat flow for RCP-D, d) cumulative heat for RCP-D, e) specific heat flow for RCP-LAB, and f) cumulative heat for RCP-LAB.

The results show that replacing Portland cement with RCP and MK reduces the induction period and advances the acceleration period compared to the reference. During the induction period, the kinetics were in the order of 70PC30MK > 70PC20MK10RCP> 70PC10MK20RCP>70PC30RCP>100 PC, regardless of the source of RCP. It can be seen that the increase in kinetics (induction period) largely depends on the MK content; the higher the MK content, the greater the heat flow. This behavior can also be observed for RCP, but to a lesser extent. The heat flow of the pastes with 30 % RCP was above the reference but below the mixtures with MK. These results indicate that both MK and RCP influence the early hydration of PC, which can be attributed to the fine particles (D10) and the nucleation effect associated with the SSA of the materials used (Yang et al., 2020; Liu et al., 2022a). The SSA of MK is 3.45 times greater than that of Portland cement (Table 2), which explains its effect in increasing the kinetics and reducing the induction period. Furthermore, the SSA of RCP-C, RCP-D, and RCP-LAB is 2.47, 2.45, and 2.14 times higher than the SSA of Portland cement, demonstrating a minor influence. These results are also in agreement with Gencel et al. (2020), Li et al. (2022), and Wu et al. (2022c), who additionally attribute the nucleation effect at early ages to the CaCO3 particles in the RCP.

During the acceleration period, the kinetics of the mixtures with MK and RCP are higher than the reference, regardless of the origin of RCP. The fine particles of RCP and MK disperse cement grains and accelerate hydration, providing nucleation sites for hydration products (Xiao et al., 2018; Moreno-Juez et al., 2021). An advance of the first heat peak (shift to the left) is also observed, which is associated with the hydration of C3S (Yang et al., 2020; Liu et al., 2022a). The mixtures with the highest MK content present a more accentuated shift to the left, followed by the pastes with 30 % RCP. However, the maximum value of the first heat peak is reduced when Portland cement is replaced by RCP and MK, which is mainly attributed to clinker dilution (He et al., 2020). Due to the reduction of the first heat peak, it can be established that, during the acceleration period, the dilution effect has a major influence on hydration compared to the nucleation effect, the latter having a greater impact on the induction period.

However, there is not only a physical effect on the part of MK and RCP; the chemical effect of both materials generates secondary phases. The pozzolanic reaction of the amorphous particles of SiO2 and Al2O3 in MK with the CH of the cement paste forms secondary C–S–H, in addition to calcium aluminate hydrates and aluminosilicate hydrates (C4AH13, C2ASH8, and C3AH6) (Sabir et al., 2001; Siddique and Klaus, 2009; Homayoonmehr et al., 2021). Regarding the RCP, some authors have pointed out a pozzolanic potential (Xiao et al., 2018; Prošek et al., 2019). The possible formation of hemi or monocarboaluminates has even been mentioned due to the reaction of CaCO3 and the aluminous phases in the mixture (Puerta-Falla et al., 2015; Medjigbodo et al., 2018; Moreno-Juez et al., 2021). However, since the heat flow of 30 % RCP pastes is lower than MK pastes, and considering the chemical composition of RCP, it could be stated that, regardless of the source, RCP has a predominantly physical effect (filling and nucleation) on the early cement hydration (induction and acceleration).

Zhang et al. (2021) and Wang et al. (2022a) indicate that using RCP accelerates the early hydration process, and that the first heat flow peak depends on the substitution percentage. In this case, the substitution of 30 % Portland cement reduced the first peak, regardless of the materials' proportions (MK and RCP). Conversely, the intensity of the heat flow differs from those of Topič and Prošek (2017), Wang et al. (2020), and Chen et al. (2021), who reported an increase in the intensity of the first peak; however, these authors used lower substitution percentages and finer RCP particles. In this case, the higher substitution content must have caused a decrease in intensity due to the dilution effect.

During the deceleration period, a second heat flow peak can be noticed, the intensity and position of which depend on the MK content. The higher the MK proportion, the greater the heat flow intensity and displacement to the left. This peak is associated with sulfate depletion, that is attributed to the dissolution of C3A and the final formation of ettringite, which occurs when the calcium sulfate phase is exhausted (Lerch, 2008; Hesse et al., 2011). Lagier and Kurtis (2007) indicated that this second peak would depend on the SSA of the MK and the alkaline content of the cement, factors that would favor the dissolution of the MK and increase the solubilized aluminum. In this sense, the second peak could be attributed to the acceleration of C3A hydration due to the increase in MK's dissolution rate and the MK reaction itself (2018). On the other hand, in the mixtures with 30 % RCP, the second peak does not appear, which would confirm its formation specifically for the use of MK.

From the results presented, it can be seen that RCP and MK accelerate the initial hydration of cement pastes, which is reflected in the cumulative heat curves. During the first hours, the pastes with RCP and MK have a higher cumulative heat than the reference. These results are in agreement with the literature, which indicates that RCP with a particle size equal to or smaller than Portland cement could improve the cumulative released heat at early ages (Xiao et al., 2018; Duan et al., 2020b; Wang et al., 2022b). In this case, the use of 30 % RCP, regardless of the source, presented a higher cumulative heat than the reference up to approximately 3 h, while, with the use of MK, the cumulative heat was higher than the reference up to about 7 h. Li et al. (2023) also noted that for 30 % RCP and a D50 of approximately 13 μm, the cumulative heat is higher than the reference during the initial hours. However, at later ages, the samples with RCP demonstrate lower levels of cumulative heat, resulting in a decrease in compressive resistance. Zhang et al. (2021) reported that only with the use of ultrafine RCP (D50 = 0.249 μm) there was a higher cumulative heat up to 72 h, considering 2 % and 4 % substitution.

By the end of the test (7 days), the cumulative heat of 70PC30MK had reduced by 4.97 % compared to the reference. Table 5 presents an overview of the released heat data for different time periods. Pastes with 30% RCP showed the greatest reduction compared to the reference, at 21.64 %, 20.95 %, and 23.35 % for sources C, D, and LAB, respectively. Prošek et al. (2019) observed that after 5 days, the samples with RCP released more heat than the reference, suggesting a reactivation of the old clinker. However, this phenomenon was not observed in the analyzed pastes, indicating the absence of old clinker in the RCP material. These findings were further supported by the XRD results (Fig. 3). The difference between MK pastes and RCP pastes can be attributed to the pozzolanic activity of MK, which contributes to the formation of more hydration products, therefore, shows more released heat. The 10 MK + 20 RCP pastes had reductions of 10.58 %, 11.53 %, and 12.37 % for RCP-C, RCP-D, and RCP-LAB, respectively. Meanwhile, the 20 MK + 10 RCP pastes presented reductions comparable to the 70PC30MK mixture, at 7.52 %, 7.36 %, and 7.85 % for RCP-C, RCP-D, and RCP-LAB, respectively.

Table 5. Cumulative heat for different time periods.

MixtureReleased heat (J/g cement)
3 h24 h72 h168 h
100 PC13.002173.294233.081265.618
70PC30MK20.168139.173218.402252.417
70 PC30C13.625133.311179.949208.111
70PC20 MK10C17.664136.854208.446245.645
70PC10MK20C16.416136.601198.359237.506
70PC30D13.973133.798181.987209.976
70PC20MK10D18.171137.951209.635246.075
70PC10MK20D15.649135.131196.284235.001
70PC30LAB14.030134.720179.914203.587
70PC20MK10LAB18.723139.070209.119244.759
70PC10MK20LAB16.563136.662195.356232.765

Although no significant differences were observed between the heat flow and cumulative heat between the three RCP sources, the LAB source presented the lowest values in all cases, which can be attributed to its lower SSA among the other RCP sources. The reduction of the cumulative heat of the pastes could mean a reduction of the mechanical properties, especially for 30 % of RCP, that presents the lowest cumulative heat of all the pastes; however, these values are only referential (Chen et al., 2022; Jiang et al., 2022). As stated by Oliveira et al. (2023), altering the grinding time (resulting in a finer RCP) has an impact on the kinetics of hydration, RCP ground for only 30 min exhibited lower levels of hydration products compared to RCP ground for 2 h. These findings suggest that the hydration process may benefit from using RCP with a smaller D50, even smaller than that of the Portland cement employed (CP II F-32).

It should be noted that the use of RCP leads to an increase in kinetics during the induction period (early hydration), even with a 30 % substitution rate. Despite the different chemical compositions of each RCP source, the behavior of the heat flow curves, and the cumulative heat values remains similar. It is worth noting that RCP primarily has a physical effect on hydration (nucleation and filling), as the physical properties of the RCPs are similar. While the accumulated heat may decrease with RCP use, the addition of MK can increase this value, making it a viable strategy for enhancing the performance of cement-based materials with RCP.

3.4. XRD

The pastes were characterized by X-ray diffraction to determine their mineralogical composition during the hydration process. Fig. 8a and b presents the mineralogical analysis of the pastes for 1 and 28 days, respectively.

Fig. 8
  1. Download : Download high-res image (1MB)
  2. Download : Download full-size image

Fig. 8. Mineralogical composition of the pastes for a) 1 and b) 28 days.

It can be observed that the pastes present a mineralogical composition derived from the source of materials and products resulting from hydration. Muscovite (2θ = 8.7°) appears in all pastes that contain RCP, which intensity depends on the amount of RCP, regardless of the source (C, D, and LAB). Quartz also occurs in pastes that contain RCP, standing out for 2θ of 20.9° and 26.6°. Both minerals can mostly be attributed to the presence of SiO2, coming from the sand (Qin and Gao, 2019; Liang et al., 2021a; Ma et al., 2021). Calcite is found in all pastes, including the reference and 70PC30MK, indicating the presence of carbonate material from Portland cement. The calcite peak for 2θ of 29.5° stands out from all pastes, followed by minor peaks of 23.1°, 39.5° and 43.2°. Calcite is due to both the RCP (coarse aggregate and carbonation process) and the cement (limestone filler) (Caneda-Martínez et al., 2021; Mehdizadeh et al., 2021b; Wang et al., 2022b). It can be seen that the pastes with the highest MK content present slightly lower calcite peaks, especially 70PC30MK. This is due to the fact that the 70PC30MK paste does not have another source of CaCO3, as is the case with RCP pastes. These results can be verified by the mineralogical composition (Fig. 3) and TG (Fig. 4) of the materials used. The intensity of each mineral is similar in the mixtures: 70 PC30C–70PC30D∼70PC30LAB, 70PC10MK20C–70PC10MK20D∼70PC10MK20LAB, and 70PC20 MK10C–70PC20MK10D∼70PC20MK10LAB, indicating that the RCP source has no significant influence on the mineral composition, once the mineralogical composition of each RCP source was also similar (Fig. 3).

As expected, non-hydrated cement particles such as C2S, C3S, and C4AF appear in the XRD pattern of the 1-day-old pastes, indicating incomplete hydration (Deng et al., 2021; Liu et al., 2022b; Wang et al., 2022a). It is observed that the 100 PC paste has the highest C3S peak, while the pastes with RCP + MK presented a decrease in the intensity of C3S, which is attributed to a) the dilution effect of Portland cement, lower clinker content, and b) the acceleration of early hydration of Portland cement by part of the RCP and MK (Qin and Gao, 2019; Xu et al., 2021; Chen et al., 2021), as presented in Subsection 3.3. It is important to note that C2S and C3S present similar intensities in pastes containing RCP, regardless of the source, which would indicate that there would be no difference in the effects of RCP on Portland cement hydration.

Regarding the hydration products, portlandite is present in all pastes, with an intensity peak at 18.1° and to a lesser extent at 28.7°, 34.1°, and 47.1°. As expected, the reference paste presented the highest Portlandite peaks, followed by the 30 % RCP pastes, even at one day of age. The decrease in CH content in the pastes containing RCP is a result of dilution, as there is a lesser quantity of Portland cement present in the mixtures (Gao et al., 2022). The intensity of the portlandite decreases with the increase in MK content, which is attributed to the pozzolanic reaction, with the 70PC30MK paste having the lowest Portlandite content. Additionally, a slight AFt peak is observed at 9.1°, corroborating the results presented from isothermal calorimetry (Subsection 3.3).

For 28 days, the XRD results showed minerals from the RCP with few non-hydrated cement particles and a higher amount of hydration products. Muscovite and quartz were present in the RCP pastes, with intensities similar to the 1-day results, highlighting their crystalline structure. Calcite was detected in all pastes, presenting a similar intensity as the 1-day results. When the pastes of the three RCP sources were compared, the minerals found had similar intensities: 70 PC30C–70PC30D∼70PC30LAB, 70PC10MK20C–70PC10MK20D∼70PC10MK20LAB, and 70PC20 MK10C–70PC20MK10D∼70PC20MK10LAB.

The non-hydrated cement particles C2S and C3S appear with lower intensity than the 1-day pastes (Wang et al., 2022a), and the presence of C4AF is absent, indicating a more advanced hydration process. These results also demonstrate that the RCP source has no particular influence on the hydration, as they all presented similar intensities (C ∼ D ∼ LAB).

As hydration products, peaks of CH, AFt, and monocarboaluminate (Mc) were detected. The highest intensities of portlandite occurred in the reference, followed by pastes with 30 % RCP, regardless of the source. The increase in the amount of MK already reduced the intensities of the portlandite, which is almost imperceptible for the 70PC30MK mixture. When comparing the portlandite intensities with the 1-day pastes, the reduction was significant, confirming the pozzolanic activity of MK. AFt could also be detected, but at a lower intensity than the 1-day pastes. Finally, Mc appeared for 2Ɵ = 11.7° together with the disappearance of C4AF; the Mc peak was present in the RCP pastes but to a lesser extent for 70PC30MK. The presence of Mc in the pastes can be attributed: a) in reference to the presence of limestone filler in Portland cement (Lothenbach et al., 2008); b) in 70PC30MK mixture to the carbonaceous material of Portland cement and the incorporation of MK (aluminum phases) (Puerta-Falla et al., 2015), although to a lesser extent due to the reduction of Portland cement content; c) in pastes with 30 % RCP to the CaCO3 of the RCP with aluminum (Al) phases of Portland cement, a situation previous reported in the literature (Lu et al., 2018; Wu et al., 2022c); and d) in RCP + MK pastes, to the role of MK as an aluminizing agent in the reactivity of CaCO3 in RCP and Portland cement systems (Puerta-Falla et al., 2015; Weise et al., 2023).

Finally, it can be observed that the RCP source does not have a particular effect on the XRD results, neither on the intensity or generation of new phases. The pastes presented similar intensities and the same mineralogical composition. Although Chen et al. (2021) indicate that the RCP does not change the mineral composition of the pastes, only the intensity of the peaks, this investigation demonstrated that the presence of the Mc is related to the presence of the CaCO3 of the RCP, regardless of the source. Additionally, the use of MK also contributes to the formation of Mc, which is an alternative source of Al, apart from C3A and C4AF from Portland cement. The findings indicate that the mineral composition of the MK + RCP mixtures is comparable, attesting to the pozzolanic reactivity of MK and the inclusion of inert minerals from RCP, such as quartz and muscovite.

3.5. TG

Fig. 9, Fig. 10, Fig. 11 show the DTG curves for the pastes from sources C, D, and LAB, respectively, considering the four test ages (1, 7, 28, and 120 days). An initial mass loss can be observed in all cases, between the range of 35 and 200 °C. This loss is produced by the evaporation of free water and chemically bound water in C–S–H, as well as AFt and Mc (Gruskovnjak et al., 2011; Bezerra et al., 2021; Chen et al., 2022). The mass loss peak in the 60–100 °C range can be attributed to the free water and decomposition of C–S–H and AFt (Neves Junior et al., 2012; Leklou et al., 2017; Fernández et al., 2018); while the peak (DTG) in the range 130–150 °C is due to Mc (Lothenbach et al., 2007; Fernández et al., 2018; Lu et al., 2018). The mass loss in the range of 350 and 450 °C corresponds to the dehydroxylation of CH (Wu et al., 2021e). On the other hand, a mass loss can be perceived in the range of 200 and 350 °C, which is attributed to the dehydration of aluminum products (Das et al., 1996). Finally, from ∼600 °C the decarbonation of CaCO3 occurs (Wu et al., 2022a).

Fig. 9
  1. Download : Download high-res image (490KB)
  2. Download : Download full-size image

Fig. 9. DTG curves for RCP-C at ages a) 1, b) 7, c) 28, and d) 120 days.

Fig. 10
  1. Download : Download high-res image (515KB)
  2. Download : Download full-size image

Fig. 10. DTG curves for RCP-D at ages a) 1, b) 7, c) 28, and d) 120 days.

Fig. 11
  1. Download : Download high-res image (488KB)
  2. Download : Download full-size image

Fig. 11. DTG curves for RCP-LAB at ages a) 1, b) 7, c) 28, and d) 120 days.

It can be observed that the pastes with RCP, regardless of the source, present a similar behavior at the ages studied. For one day, the reference presents the highest peak of C–S–H and AFt, followed by the 70PC30MK mixture, indicating a higher degree of hydration than the other pastes, as presented in the isothermal calorimetry (Subsection 3.3). Additionally, a slight Mc peak is observed, which is more noticeable for 70PC30MK, followed by the MK + RCP pastes. The Mc peak is not perceptible in the reference, demonstrating that MK could be an additional source of Al to react with CaCO3 from RCP and even Portland cement. On the other hand, in the range of 200 and 350 °C, the mass loss of the reference and the pastes with 30 % RCP is similar. In contrast, there is a more significant mass loss for the mixtures with MK, especially 70PC30MK, indicating the production of phases with Al, which is specifically attributed to MK. In the range of 350 and 450 °C, the reference has the highest CH content, followed by the pastes with 30 % RCP, confirming the XRD results (Subsection 3.4). The higher the MK content, the lower the mass loss in this range, validating the consumption of CH by the pozzolanic reaction of MK. In the case of CaCO3, at temperatures above 600 °C, the reference presents the highest CaCO3 content due to the presence of carbonate material (limestone), followed by the mixtures with 30 % RCP, RCP + MK, and 70PC30MK. As previously mentioned, this behavior is because the 70PC30MK paste does not have another source of CaCO3, as in the case of RCP pastes. The elevated CaCO3 content is anticipated, as the CP II F-32 cement utilized can have a range of 11 %–25 % carbonate material according to NBR 16697 (ABNT, 2018).

For seven days, the peak associated with AFt was slightly reduced and stabilized at later ages, corroborating the stabilization of AFt in the presence of CaCO3 (Kuzel and Pöllmann, 1991; Schmidt et al., 2008; Bentz et al., 2017). In the other pastes, there was an increase in the range of C–S–H and AFt, especially those with MK, indicating an increase in hydration products. It is important to note that the Mc-associated peak was not only noticeable in the MK mixes but also in the reference and 30 % RCP pastes. These results would be in agreement with Lothenbach et al. (2008), who discussed the influence of limestone on cement hydration (Mc formation), and with Bonavetti et al. (2001), who indicated that, in cements with limestone, the Mc is detected after the first few days. In the present investigation, it was not possible to detect the Mc at one day of age in DTG curves and XRD results (Fig. 8a). In addition, it was verified that the RCP would only provide a physical effect (nucleation) in the early hydration of Portland cement, as was commented in the isothermal calorimetry (Subsection 3.3). On the other hand, the formation of Mc in the pastes with 30 % RCP indicates that the CaCO3 source of the RCP could offset some effects of cement reduction (CP II F-32). Finally, it is observed that the CaCO3 peaks of all the pastes decreased compared to the 1-day age, confirming the CaCO3 reaction for the formation of Mc.

For 28 days, there has been the presence of C–S–H, AFt and Mc in all pastes, the latter being more prominent in pastes with MK, specifically 70PC30MK. Regarding the mass loss in the range of 350 and 450 °C, as expected, there has been a greater consumption of CH with the increase of MK content. The samples with 30 % RCP showed a slight reduction in CH, which could indicate a possible pozzolanic potential, albeit limited. Regarding CaCO3 (>600 °C), no significant changes were observed concerning the age of 7 days. Finally, for 120 days, the peaks associated with the Mc remained constant compared to 28 days, while the temperature range for C–S–H and ettringite increased, which could indicate a significant increase in hydration products due to the pozzolanic reaction, especially for 70PC30MK and MK + RCP pastes. In the case of CaCO3, no significant changes were observed. For a more complete analysis of the TG results, Fig. 12, Fig. 13, Fig. 14 summarize the results of the CH content, free water, and CaCO3 versus the age of the samples, respectively.

Fig. 12
  1. Download : Download high-res image (305KB)
  2. Download : Download full-size image

Fig. 12. CH content in the analyzed pastes.

Fig. 13
  1. Download : Download high-res image (328KB)
  2. Download : Download full-size image

Fig. 13. Chemically bound water content.

Fig. 14
  1. Download : Download high-res image (314KB)
  2. Download : Download full-size image

Fig. 14. CaCO3 content in the analyzed pastes.

It is important to mention that the error of the TG can be in the order of 5 % due to equipment measurement errors, sample heterogeneity, and operator manipulation (Deschner et al., 2012). This suggests that not all differences are significant. Regarding the CH content (Fig. 12), a gradual increase can be observed in the reference, mainly in the first days, corresponding to the hydration of the cement (Scrivener et al., 2015). In the case of pastes with 30 % RCP, the CH content is lower than the reference, a situation attributed to the inert materials in the RCP (dilution effect) (Wu et al., 2021a; Liu et al., 2022a). Although a decrease in the CH of 7 for 28 days can be observed in the three sources of RCP, it can be considered negligible. In the case of pastes with MK, the CH content is significantly lower than the reference, confirming the pozzolanic reaction. Additionally, the results show no significant difference in the CH content by the RCP source. It is important to emphasize that the pastes with RCP-LAB have a higher CH content (not significant), which can be attributed to the CH of the residue composition (Fig. 4b).

Regarding the chemically bound water content (Fig. 13), it is observed that, for the age of one day, there is a difference between the reference and the other pastes, indicating a dilution effect; however, this difference decreases as the curing age increases. Although significant differences between pastes cannot be distinguished, there is a trend regarding the amount of MK and RCP. For example, pastes with higher RCP content tend to have lower chemically bound water content than pastes with MK, which can even exceed the reference for 120 days. These results suggest that, although the RCP decreases the hydration products with respect to the reference, the MK + RCP pastes could generate secondary hydration products (mainly due to the pozzolanic reaction) that would compensate for the dilution effect. Wu et al. (2022a) also stated that the use of RCP reduces the quantity of hydrated products, particularly CH and C–S–H, which affects the formation of mechanical strength in the pastes.

It can be verified that the CaCO3 content decreases from 1 to 7 days (Fig. 14), confirming a partial consumption of CaCO3 with Al phases (Portland cement + MK) for the formation of Mc. However, it is essential to note that the reduction of CaCO3 in the first days is ∼5 %, which suggests that the amount of Mc is low, confirming XRD patterns for 28 days (Fig. 8b). Therefore, the Mc could have a minor impact on the total hydrated products and mechanical strength.

3.6. Compressive strength

The compressive strength and its variation with respect to the reference (black lines) and the 70PC30MK paste (red lines) is presented in Fig. 15 for the three RCP sources. It can be initially observed that the replacement of 30 % of Portland cement by MK generates an increase in compressive strength, except for the age of 1 day, where it is statistically the same (p value˃0.05). For 7, 28, and 120 days, the increase was 31.00 %, 34.80 %, and 22.82 %, respectively. This behavior can be attributed to the pozzolanic activity and the filler effect of the MK, which improves the microstructure and, thus, the mechanical properties of the cement-based materials (Raheem et al., 2021; Wu et al., 2022a).

Fig. 15
  1. Download : Download high-res image (1MB)
  2. Download : Download full-size image

Fig. 15. Compressive strength for: a) RCP-C, b) RCP-C variation, c) RCP-D, d) RCP-D variation, e) RCP-LAB, and f) RCP-LAB variation.

Regardless of age and source, the mixes with 30 % RCP have the most significant reduction in compressive strength compared to the reference. On the first day, a decrease in compressive strength of 42.98%, 33.77%, and 34.17% was observed for RCP-C, RCP-D, and RCP-LAB, respectively. Li et al. (2023) noted a 50% decrease in compressive strength for mixtures with 30% RCP and similar particle size (D50 ∼13 μm) on the first day. After 28 days, the decrease in compressive strength was 25.56%, 20.77%, and 23.98% for RCP-C, RCP-D, and RCP-LAB, respectively. This reduction is higher than that of the 70PC30MK paste, whose range is 35–50 %, considering the three RCP sources. These results indicate the negative effect of substituting high percentages of RCP (30 %) on compressive strength, as reported by other investigations (Wu et al., 2021e). This reduction is explained by the predominance of clinker dilution over other effects, such as nucleation (first hours) and filler, indicating a considerable amount of inert material (Liang et al., 2021b). On the other hand, it can be established that there are no significant differences in the compressive strength means when comparing the three sources of RCP by age (p value > 0.05). This indicates that the source of the RCP does not have a significant effect on the compressive strength of the pastes (70PC30RCP). These results can be compared to the isothermal calorimetry and TG (3 Results and discussion, 3.3 Isothermal calorimetry.5), where pastes with 30 % RCP present the lowest cumulative heat and combined water content regardless of the source. It should be noted that mixtures containing 30% RCP exhibit a reduced growth rate of compressive strength at a prolonged age of curing. For instance, there was an 11.59 %, 6.96 %, and 9.80 % increase in compressive strength from 28 to 120 days for 70 PC30C, 70PC30D, and 70PC30LAB, respectively. On the other hand, the increase in compressive strength from 7 to 28 days was 27.08 %, 29.82 %, and 22.55 % for the same mixtures. This decrease in growth rate can be mainly attributed to the dilution effect as these pastes have a higher RCP content. As a result, the active components of Portland cement are lower and primarily react in the early stages (Rocha et al., 2023).

The 20 MK + 10 RCP pastes present a reduction of 21.28 % (RCP-C), 12.70 % (RCP-D), and 15.95 % (RCP-LAB) compared to the reference after one day. Subsequent ages show a considerable increase in compressive strength; for example, after 28 days the increase is 27.31 %, 36.18 %, and 35.92 % above the reference for RCP-C, RCP-D, and RCP-LAB, respectively. However, for 120 days this increase tends to be lower, with 18.71 %, 16.84 %, and 17.30 % for RCP-C, RCP-D, and RCP-LAB, respectively. When the mixtures are compared with the 70PC30MK paste, the reduction is in the range of 20–35 % after one day. Later ages (7, 28, and 120 days) show minor reductions (∼5 %). It can be established that there is no significant difference between the means of 70PC30MK and 70PC20MK10RCP (p value > 0.05), regardless of the source of RCP. These results indicate that the mechanical performance of 70PC30MK is comparable to the 20MK + 10RCP combination. Finally, when the compressive strength values are compared between the three sources of RCP by age, it can be concluded that they are statistically similar (p value > 0.05). Therefore, the use of the RCP source can be considered indistinct in this combination (20 MK + 10 RCP).

These results indicate that the pozzolanic activity of MK can be benefited by RCP since the results of the 20 MK + 10 RCP pastes are equal in mechanical strength to the 70PC30MK mixture. This can be attributed to the nucleation effect of RCP and the formation of the hydration products by the pozzolanic reaction of MK (Prošek et al., 2019; Yang et al., 2020; Raheem et al., 2021). On the other hand, the filling effect of smaller RCP particles (less than 10 μm) can contribute to improving the compactness of the microstructure (Chen et et al., 2021), and, to a lesser extent, the reaction of some RCP particles can form hydration products, for example, CaCO3 to produce Mc (3 Results and discussion, 3.4 XRD.5) (Lu et al., 2018; Wu et al., 2021e). Oliveira et al. (2023) suggested that smaller RCP particles with a high SSA have the tendency to enhance compressive strength at lower substitution levels, specifically 7 % and 15 %.

In the 10 MK + 20 RCP pastes, the compressive strength values are higher than the reference for all ages except for one day. There is a 29.96 %, 23.27 %, and 26.58 % reduction compared to the reference for RCP-C, RCP-D, and RCP-LAB, respectively. However, an increase in compressive strength is observed for later ages; for example, for 28 days, there is a positive variation of 14.67 %, 15.00 %, and 18.52 % for RCP-C, RCP-D, and RCP-LAB, respectively. In the case of 120 days, an attenuation in the increase is perceived, with ranges of 9.22 %, 7.38 %, and 4.92 % for RCP-C, RCP-D, and RCP-LAB, respectively. When the results are compared to the 70PC30MK paste, there is a reduction for all ages and sources of RCP, ranging from 24 to 31 %, 13–16 %, 12–15 %, and 11–14 % for 1, 7, 28, and 120 days, respectively. These results indicate a growing approximation to the 70PC30MK paste; however, the differences are notable for a p value of less than 0.05. Lastly, when the results of the three RCP sources by age are compared, there are no significant differences (p value˃0.05), indicating that the use of the RCP source for 10 MK + 20 RCP pastes is indistinct. Wu et al. (2022a) also reported comparable findings, where a 20RCP+10MK combination showed a significant increase (38.10 %) in mechanical strength compared to the mixture with 30 % RCP. This indicates that incorporating MK improves the microstructure of cement-based matrices and enhances their mechanical properties.

The results of the 10 MK + 20 RCP combination are presented as a viable alternative, since it requires minimal use of pozzolans and a higher RCP content, resulting in better mechanical strength than the reference. Like the 20 MK + 10 RCP combination, the RCP provides a nucleation effect for the cement's hydration and MK's pozzolanic reaction. However, with a lower MK content (10 %), the hydrated products of the pozzolanic reaction are reduced. Similarly, by increasing the RCP content (20 %), inert and large particles are introduced (less filler effect), which reduces the compressive strength when compared to 70PC30MK and 20 MK + 10 RCP.

It can be observed that the combinations of 20 MK + 10 RCP and 10 MK + 20 RCP presented better mechanical performance than the reference. This can be attributed to the synergistic effect between pozzolan (MK) and RCP (Yang et al., 2020), wherein small RCP particles contribute to the pozzolanic reaction of CH with the amorphous SiO2 and Al2O3 of MK, in addition to the filling effect of both materials (MK + RCP), which results in the increase in compressive strength (Sun et al., 2020). However, the synergistic effect was more efficient in the combination of 20 MK + 10 RCP, with no significant difference with 30 % MK. These results agree with the TG (Subsection 3.5), which showed that the pastes with the highest percentage of MK had a higher combined water content, indicating the presence of more hydration products.

Regarding age, although MK + RCP has positive results in compressive strength (7, 28, and 120 days), in all cases, there was a one-day reduction compared to the reference. This decrease of one day can be explained by the lower amount of hydrated products (Fig. 13), less heat cumulative up to the first day (Fig. 7), and the non-formation of additional hydration products, such as Mc (Fig. 8a). These results suggest that the synergistic effect of RCP + MK does not influence initial ages, with the best results occurring at 7 and 28 days.

Other authors have indicated that the incorporating mineral additives with RCP can maintain or improve the properties of cement-based materials (Yang et al., 2020; Chen et al., 2021; Liang et al., 2021a). Wu et al., 2021e, 2022a found that using 10–20 % SCMs (MK and SF) with RCP can improve compressive strength compared to a mix of 30 % RCP and may even be slightly higher than the reference mortar. Chen et al. (2021) noted that FA and SF can provide complete hydration and improve the compactness of cement-based materials. Cantero et al. (2022) and Sun et al. (2022) reported that there is an increase in compressive strength of mixtures of RCP with other SCM (SCGP and FA), a phenomenon more evident at later ages (300–365 days). Prošek et al. (2019) also highlighted that the use of SCM (such as FA) can improve the compressive strength of cement-based materials with RCP (28 and 90 days), attributing this result mainly to the pozzolanic reaction. Hence, the utilization of additional SCMs, particularly pozzolans, may compensate for the detrimental impact of incorporating RCP on the mechanical performance of cementitious matrices. This is attributed to the production of hydration products (C–S–H and C-A-S-H) and the refinement of pores in the cement matrix (Sun et al., 2021; Yang et al., 2022). Nevertheless, it is important to determine appropriate proportions when combining SCM and RCP to attain a synergistic effect between the materials.

3.7. Elastic modulus

Fig. 16 shows the results of the elastic modulus for the three RCP sources, in addition to its variation with respect to the reference (black lines) and the 70PC30MK paste (red lines). It can be seen that the values of the elastic modulus are similar for the reference and 70PC30MK, although there may be a slight decrease due to the addition of MK. Statistically, these values are similar for all ages (p value˃0.05). Therefore, it can be established that the use of 30 % MK does not significantly influence the elastic modulus, maintaining the value for all ages tested. Qian and Li (2001) and Dinakar et al. (2013) also pointed out that the elastic modulus presents small variations in cement-based materials with the increase of MK, confirming the non-significant influence of MK on the elastic modulus.

Fig. 16
  1. Download : Download high-res image (1MB)
  2. Download : Download full-size image

Fig. 16. Elastic modulus for: a) RCP-C, b) RCP-C variation, c) RCP-D, d) RCP-D variation, e) RCP-LAB, and f) RCP-LAB variation.

Mixtures containing 30 % RCP exhibited the lowest values of elastic modulus at all ages. For 1 day of age, the reduction was 45.01 %, 44.99 %, and 42.54 % compared to the reference for RCP-C, RCP-D, and RCP-LAB, respectively. Although there was an improvement with age, for 120 days, the reduction was in the order of 30% for the three sources. In comparison with 70PC30MK paste, the lowest values were also presented for 1 day of age, reductions of 43.41 %, 43.40 %, and 40.88 % for RCP-C, RCP-D, and RCP-LAB, respectively. There were also improvements with curing age; for 120 days, the decrease was in the order of 28 %. Finally, comparing the three sources by age, it can be concluded that there are no differences for 30 % of RCP (p value > 0.05), which indicates that regardless of the source of RCP, the same influence is applied to the elastic modulus. In the literature, it has been reported that the elastic modulus decreases with the increase in RCP. This situation is attributed to the dilution effect and the low density and rigidity of the RCP particles (Bogas et al., 2019; He et al., 2020; Cantero et al., 2022). Therefore, 30 % of RCP is adverse for this property, where the decrease is significant at all ages of curing.

For the combination of 20MK + 10RCP, the results are comparable to the reference, except for 1 day of age. The decrease in the elastic modulus for 1 day of age with respect to the reference is 24.12 %, 18.98 %, and 19.67 % for RCP-C, RCP-D, and RCP-LAB, respectively, while at 120 days, the reduction is only 7.58 %, 7.60 %, and 7.13 % for RCP-C, RCP-D, and RCP-LAB, respectively. Regarding the 70PC30MK paste, there is a similar trend; for 1 day, the reduction is 21.92 %, 16.63 %, and 17.34 % for RCP-C, RCP-D, and RCP-LAB, respectively, while for 120 days, it is only 6.17 %, 6.20 %, and 5.71 % for RCP-C, RCP-D, and RCP-LAB, respectively. It is possible to establish that the elastic modulus of the 20 MK + 10 RCP pastes is statistically equal to the reference for 7, 28, and 120 days, independent of the source. The results prove the synergistic effect between MK and RCP, although MK does not have a significant influence, nor does it contribute to a considerable reduction, since the filling effect and pozzolanic reaction of MK compensate for the dilution effect of RCP.

The results of the 10MK + 20RCP combination show an increase compared to the mixtures with 30 % RCP, and a reduction compared to the reference and the 70PC30MK paste. The greatest decrease in the elastic modulus occurs at 1 day of age, and the lowest decrease is found at 120 days, 12.95 %, 11.94 %, and 11.67 % for RCP-C, RCP-D, and RCP-LAB, respectively. It is important to note that the results are statistically similar to the reference (120 days) and the 20 MK + 10 RCP pastes (28 and 120 days). When the results are compared to the 70PC30MK paste, the reduction percentages are comparable to the reference at all ages; for example, for 120 days, the decrease is 11.63 %, 10.60 % and 11.34 % for RCP-C, RCP-D, and RCP-LAB, respectively. When comparing the results of the three sources of RCP by age, it can be concluded that the values are similar, and there is no significant difference (p value > 0.05). The synergistic effect between MK + RCP can also be verified in this combination; however, to a lesser extent. This is due to a higher RCP content, where the formation of hydrated products of the pozzolanic reaction is limited, only comparable to the reference's elastic modulus at 120 days. Cantero et al. (2022) assessed the dynamic elastic modulus of RCP mixtures containing FA over a period of 365 days. The results show that the values increase with curing age, attributed to a higher degree of hydration and the reactivity of FA in the presence of RCP. These findings are correlated with compressive strength.

The combination of 20 MK + 10 RCP presents the best results compared to the MK + RCP pastes, exhibiting similar values (significant) to the reference after 7 days, confirming the results of mechanical strength. Additionally, the 10MK + 20RCP combination also exhibits values similar to the reference at 28 days, but all remain significant for 120 days.

The study has demonstrated the negative impact of RCP on compressive strength and elastic modulus. Although it was found to improve early hydration (Subsection 3.3), it has no significant effect on long-term compressive strength and elastic modulus. The use of a combination of MK and RCP has shown the potential to enhance mechanical performance by leveraging the chemical and physical properties of these materials. Thus, incorporating MK and RCP in concrete can be a feasible alternative with the potential to replace up to 30 % of Portland cement. The mixture of 20 MK + 10 RCP outperforms the reference in terms of mechanical performance. The mixture of 10 MK + 20 RCP presents comparable mechanical performance to the reference, utilizing a higher quantity of RCP and presenting a more sustainable solution. In this context, both combinations demonstrate superior technical and environmental performance as compared to the reference mixture.

3.8. Microstructure

Fig. 17, Fig. 18, Fig. 19, Fig. 20 show the SEM images of the references (100 PC and 70PC30MK), RCP-C, RCP-D, and RCP-LAB, respectively. For the ages 1 and 28 days. As with other investigations into the use of RCP in cementitious matrices, it was possible to detect different hydration products (Sui et al., 2021; Dun et al., 2021; Wu et al., 2022d): AFt (particles with elongated shape), CH (hexagonal distribution), and the C–S–H gel formation.

Fig. 17
  1. Download : Download high-res image (965KB)
  2. Download : Download full-size image

Fig. 17. Microstructure of the pastes: a) 100 PC (1 day), b) 70PC30MK (1 day), c) 100 PC (28 days), and d) 70PC30MK (28 days).

Fig. 18
  1. Download : Download high-res image (1MB)
  2. Download : Download full-size image

Fig. 18. Microstructure of the pastes: a) 70 PC30C (1 day), b) 70PC20 MK10C (1 day), c) 70PC10MK20C (1 day), d) 70 PC30C (28 days), e) 70PC20 MK10C (28 days), and f) 70PC10MK20C (28 days).

Fig. 19
  1. Download : Download high-res image (1MB)
  2. Download : Download full-size image

Fig. 19. Microstructure of the pastes: a) 70PC30D (1 day), b) 70PC20MK10D (1 day), c) 70PC10MK20D (1 day), d) 70PC30D (28 days), e) 70PC20MK10D (28 days) and f) 70PC10MK20D (28 days).

Fig. 20
  1. Download : Download high-res image (1MB)
  2. Download : Download full-size image

Fig. 20. Microstructure of the pastes: a) 70PC30LAB (1 day), b) 70PC20MK10LAB (1 day), c) 70PC10MK20LAB (1 day), d) 70PC30LAB (28 days), e) 70PC20MK10LAB (28 days) and f) 70PC10MK20LAB (28 days).

To confirm the presence of hydration products, a SEM-EDS analysis was conducted (Fig. 21). The points for analysis were determined based on the morphology of the hydration products. The results confirm the presence of CH, AFt, and C–S–H, and also reveal the presence of quartz particles from RCP. This corresponds with findings from previous studies (Franus et al., 2015; Gallucci et al., 2010; He et al., 2022) that indicate the elemental chemical composition is indicative of the presence of hydration products.

Fig. 21
  1. Download : Download high-res image (715KB)
  2. Download : Download full-size image

Fig. 21. SEM-EDS analysis: a) 70PC10MK20C (28 days) and b) 70PC20MK10D (28 days).

The 100 PC and 70PC30MK pastes' microstructures appear denser than the RCP + MK mixtures. For 1 day, several hydration products are distinguished in the 100 PC mix; on the other hand, for 70PC30MK, the presence of C–S–H and CH gel is observed. For 28 days, in the reference mix, the presence of C–S–H, CH, and, in a lesser amount, AFt can be noted, a situation verified with the results of XRD (Subsection 3.4) and TG (Subsection 3.5). In the case of 70PC30MK, C–S–H gel is observed, but not the presence of CH. This is attributed to the pozzolanic reaction of MK with CH to form C–S–H gel, as well as C-A-S-H (a secondary hydration reaction) (Chen et al., 2021; Liang et al., 2021a). Although CH was not detected, the amount of CH is reduced for this curing age, as presented in the XRD (Fig. 8b) and TG (Fig. 12) results.

Pastes with RCP present an open microstructure, especially for 70PC30RCP mixes, regardless of the source. This is because RCP reduces the amount of hydratable components and produces a loose microstructure, which leads to a decline in the mechanical properties of cement-based materials (Sun et al., 2020; Sui et al., 2021; Ma et al., 2022). These findings can be confirmed in 3 Results and discussion, 3.6 Compressive strength.7, where mixtures containing 30% RCP, regardless of its source, presented the greatest decrease in compressive strength and elastic modulus. The RCP and MK mixtures already improve the microstructure of the pastes compared to the 70PC30RCP pastes once the CH reacts to form C–S–H, compensating the amount of hydrated products. The latter can be verified in the SEM images of 1 and 28 days, where it is possible to detect CH at the age of 1 day; whereas for 28 days, CH detection is not possible. This enhancement in the microstructure of the RCP + MK mixtures can be confirmed through the results of compressive strength and elastic modulus, as the mixtures with RCP + MK equal or exceed the reference a with greater curing time. Other authors have used SCM as FA and SF, which increased C–S–H gel due to the nucleation and filler effect, thereby improving the compactness of cement matrices with RCP (Chen et al., 2022). The SEM images at 1 and 28 days for 100 PC and 70PC30MK exhibit a larger quantity of hydration products in comparison to the other mixtures. A loose and porous microstructure is noticeable in mixtures with a higher proportion of RCP, along with several quartz particles (RCP). Nevertheless, as the MK content and age increase, the paste microstructure becomes more compact, with the mixtures of 70PC20MK10RPC showing similar characteristics to the references (100 PC and 70PC30MK) at 28 days.

It is important to mention that the RCP particles, mainly quartz, are surrounded by hydration products, even as the hydration progresses (28 days), indicating the filler and nucleation effect of the RCP (Xiao et al., 2018; Ma et al., 2022; Yang et al., 2022; Liu et al., 2022b), a result attributed to: a) the first hours of the hydration reaction (Subsection 3.3) and b) the synergistic effect of RCP + MK, demonstrated in the TG and compressive strength results (3 Results and discussion, 3.5 TG.6). Although the RCP results in low activity due to a higher content of inert components, it could have a filling effect in the pores and fissures (Sun et al., 2020; Liu et al., 2022a; Wu et al., 2022b). Yang et al. (2022) pointed out that the products adhered to the RCP particles benefit the bond strength between the RCP particles. However, this effect could only be effective for lower percentages (Li et al., 2021; Chen et al., 2022; Wu et al., 2022d), while higher substitutions (30 % RCP) result in a significant drop in compressive strength (Fig. 15). According to the results of TG, compressive strength, and elastic modulus, the use of RCP + MK has a synergistic effect. In this sense, the RCP would increase the contact ratio of water and clinker and, on the other hand, the nucleation effect of the RCP would help with the second hydration reaction (Liu et al., 2022b; Zhang et al., 2022). Additionally, the formation of hydrated products from the reaction of CaCO3 with aluminum phases of the mixture, products that, to a lesser extent, could contribute to improving mechanical properties of pastes with RCP. The previously mentioned effects are observed in the MK + RCP pastes, which presented a mechanical performance equal to or higher than the reference, even compensating for the negative effects of using only RCP (Xiao et al., 2018).

It is observed that the RCP particles can represent the weak part in the Portland cement and MK matrix, reducing the compressive strength. But other authors (He et al., 2020; Wang et al., 2020) pointed out that a finer RCP can generate a denser and more refined microstructure, even similar to the reference, due to the nucleation and filler effect of the RCP. Also, Oliveira et al. (2020) reported that using RCP does not negatively affect the microstructure. Therefore, the use of finer RCP particles would be recommended to generate positive effects on hydration and compressive strength in pastes with only RCP, or it is desired to increase the RCP content in order to propose more sustainable alternatives to the use of Portland cement.

The current study aims to propose a reduction in the amount of Portland cement by 30 %, specifically by incorporating sustainable materials such as RCP and MK. The proposal presents a low amount of cement, even with the inclusion of 20 % carbonate material in CP II F-32 (Fig. 14). In existing literature, RCP has been proven as a sustainable material, as it is a waste product and results in a decrease in both CO2 emissions and energy consumption (Cantero et al., 2022). For instance, He et al. (2022) reported that non-renewable energy consumption (NREC) for Portland cement is 5.75 MJ/kg and the CO2 equivalent emission (CO2-e) is 0.476. In comparison, RCP has an NREC of 0.1 MJ/kg and a CO2-e of 0.019. Another study by He et al. (2020) noted that CO2 emissions from cement can reach up to 930 kg/ton at a cost of 600 RMB/t binder. In contrast, the use of RCP results in a lower range of CO2 emissions, between 744 and 767.6 kg/t, and a cost range of 480 and 510 RMB/t binder, depending on the RCP granulometry. Oliveira et al. (2023) stated that replacing cement with RCP leads to a decrease in CO2 emissions, even for RCP with extended grinding periods. For example, a RCP with a grinding process of up to 6 h and with 25 % substitution still results in lower CO2 emissions compared to the reference, reducing it by up to 25 %.

Various studies conducted on the use of MK have shown that it can effectively replace Portland cement. This not only guarantees its performance, but it also significantly reduces CO2 emissions (Heath et al., 2014; Abdellatief et al., 2023). Additionally, MK is readily available in many parts of the world, particularly in Brazil, making it a favorable substitute (Rashad, 2013; Assi et al., 2020). While these findings support the environmental feasibility of the proposal (MK + RCP), further investigation is necessary to not only suggest low-impact materials, but also to estimate the reduction in CO2 emissions and propose environmental performance indicators. To achieve this, an in-depth environmental assessment of the proposed mixtures must be conducted using methodologies, such as Life Cycle Assessment (LCA), that allow for the determination and quantification of potential environmental impacts.

This study was focused on investigating the hydration and compressive strength of ternary mixtures containing Portland cement, RCP, and MK. The results indicate that RCP can be used regardless of its source, as its effect is primarily physical, involving nucleation and filling. However, further research is needed to assess the mechanical performance and durability of these mixtures. While only compressive strength was examined, complementing the results with other properties like shrinkage, tensile strength, among others, can enhance our understanding of these materials and promote their use in the construction industry.

4. Conclusions

In this investigation, the hydration and compressive strength of ternary mixtures of Portland cement, RCP, and MK were evaluated, with a partial substitution of 30 % of Portland cement and three sources of RCP. Based on the results obtained, the following conclusions can be drawn.

  • The physical, chemical, and mineralogical characterization of the materials established that there are no notable differences between the three RCP sources, in addition to a potential use of these residues as a mineral additive. The use of RCP can be indistinguishable despite having a different source, demonstrating that the material is mostly inert. On the MK side, its high SSA and high SiO2 and Al2O3 content are noteworthy, making it suitable for use as a pozzolan.

  • Both RCP and MK have a significant effect on the rheological properties of the pastes, increasing the parameters τ0 and μ. The MK has a greater influence than the RCP, which is attributed to its high SSA and pozzolanic activity, whereas the physical effect (SSA) is considered the primary influence of RCP.

  • The use of RCP and MK can accelerate the early hydration of Portland cement; however, due to the dilution effect, the cumulative heat (up to seven days) decreased in all cases. The use of only RCP presented the lowest heat of hydration, regardless of the source, suggesting a high content of inert particles in its composition. Still, the use of RCP + MK demonstrated a synergistic effect between both materials, increasing the cumulative heat of the paste.

  • RCP and MK did not change the types of hydration products, but they did modify their amounts. The pozzolanic potential of MK was confirmed, as the amount of CH decreased with the curing age of the pastes and the amount of hydrated products increased. Additionally, Mc was detected, attributed to the reaction of CaCO3 with the aluminum phases of the mix, both from Portland cement and MK.

  • Compressive strength and elastic modulus exhibited a similar behavior. The pastes with RCP + MK showed a mechanical performance equal to or higher than the reference, particularly for the 20 MK + 10 RCP combination, confirming the synergistic effect between RCP and MK.

The results show that, although 30 % of RCP can be negative on the hydration and mechanical strength, the combination with other SCMs, in this case MK, can improve the performance of cement-based materials. A combination between the physical effect of RCP and the pozzolanic activity of MK is presented, as well as a reaction between both materials, although to a lesser extent. Therefore, using RCP + MK for a 30 % replacement of Portland cement is feasible. In order to increase the substitution rate of Portland cement, future studies could carry out mechanical treatments on the RCP, reduction of the particle size and increase of the SSA to potentiate its physical effect. On the other hand, the use of other pozzolans with a high Al2O3 content could be considered to improve the reactivity of the mixtures between RCP and SCM. Also, bio-based pozzolans, such as rice husk ash and bagasse ash, could further contribute to the environmental performance of cement-RCP mixtures.

A detailed study of the environmental impacts of the use of RCP and MK is necessary. Although they can be considered waste and reduce CO2 emissions by substituting Portland cement, the best technical-environmental alternative of the combination RCP + MK should be evaluated by a proper LCA study.

The approach proposed in the current investigation was limited to a fixed substitution percentage of 30 % between RCP and MK, while also taking into consideration physical characteristics of the materials such as D50 and SSA. Although the results show the proposal to be both technically and environmentally feasible, they are constrained by the specific parameters of the experimental design. It is recommended to explore other variables, such as additional treatments, particle size, and expanding the substitution percentages.

CRediT authorship contribution statement

Joaquin Humberto Aquino Rocha: Conceptualization, Methodology, Investigation, Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Romildo Dias Toledo Filho: Supervision, Investigation, Writing – original draft.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors would like to acknowledge CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior), Brazil, for the financial support. Also, to Matheus Pimentel Tinoco and Andréia Arenari de Siqueira for reviewing the first draft.

Data availability

Data will be made available on request.

References

Cited by (5)

View Abstract