Hydration, mechanical, and environmental performance of blended cements incorporating ground granulated blast furnace slag, ground waste bottle glass, and metakaolin

This study investigates the effect of incorporating 20 wt% ground granulated blast furnace slag (GGBFS), ground waste bottle glass (GWBG) and metakaolin (MK) on the hydration behavior, mechanical and environmental performance of blended cements cured at 23 °C ± 2 °C for 7, 28, and 90 days. Mechanical performance was evaluated using the strength activity index (SAI), while environmental performance was assessed through heavy metal leaching and a preliminary sustainability screening. The tested mortars achieved SAI values greater than 80% at all curing ages, consistent with the observed development of hydration products and microstructural evolution. Heavy metal leaching tests showed Pb and Zn below the limit of detection, while Cr release ranged from 0.70 to 1.15 mg/kg, which is substantially below the non-hazardous waste acceptance limit. The preliminary sustainability screening showed reductions of 10.9%–17.2% in calculated binder-production embodied energy and 12.9%–18.9% in calculated binder-production CO 2 emissions compared to a plain Portland cement benchmark. These findings demonstrate the potential of these SCMs for producing blended cements with satisfactory mechanical performance, limited trace-metal release, and lower calculated binder-production CO 2 indicators compared with the plain Portland cement benchmark.

Cement is one of the most abundantly produced construction materials worldwide. In Kazakhstan, cement production has increased significantly due to the rapid development of the construction sector, reaching a historic record of 13.07 million tons in 2025 (Qazinform News Agency, 2026). The ever-increasing demand for cement has also intensified the issues related to energy consumption and environmental pollution. It has been reported that cement production is a major source of industrial CO 2 emissions, contributing about a 5%–8% of global emissions (Castillo et al., 2022). Thus, the development of sustainable alternatives to conventional cement has become a research priority. These challenges have stimulated extensive research into the use of supplementary cementitious materials (SCMs) as partial replacements for Portland cement (PC) (Kolawole et al., 2021). Depending on their chemical composition, SCMs can modify the hydrate assemblage formed during PC hydration (Lothenbach et al., 2011; Scrivener et al., 2018). Silica-rich SCMs generally decrease the Ca/Si ratio of calcium silicate hydrate (C-S-H) by consuming calcium hydroxide (CH), while alumina-rich SCMs promote Al incorporation into C-S-H and the formation of calcium aluminosilicate hydrate (C-A-S-H) and other alumina-bearing hydrates (Lothenbach et al., 2011; Juenger et al., 2019; Jing et al., 2023), with both cases potentially resulting in a denser microstructure and improved long-term mechanical properties.

The worldwide usage of SCMs is geographically very variable, depending both on local resource availability and also established practice in the construction sector of each country (or regions within countries). So, although some countries consider SCM blending to be commonplace (or even near-universal), there are many parts of the world where this is not the case, and so it is important to validate the use of locally sourced materials and commercially available SCMs in combination with the cements that are currently produced in those countries (and according to local standards), to provide the necessary insight and technical confidence underpinning broader deployment of lower-emissions blended cements and valorization of high-volume local wastes. At the industrial scale, data-driven process-monitoring approaches may also support production stability and quality control during the manufacture of BCs (Ali et al., 2025a; Ali et al., 2026; Ali et al., 2025b).

SCMs are commonly derived from industrial by-products, recycled waste materials, or naturally occurring minerals. Among the wide range of available SCMs, ground granulated blast furnace slag (GGBFS), ground waste bottle glass (GWBG) and metakaolin (MK) have attracted considerable attention as promising materials for the partial replacement of PC (Mughrabi et al., 2026; Zhang X. et al., 2024; Binici et al., 2007). GGBFS is an industrial by-product obtained during the production of pig iron in blast furnaces. About 500–550 million tonnes of blast furnace slag are produced worldwide each year, making it one of the most abundant industrial by-products available for utilization in the construction industry (Kunaev et al., 2023). Rapid water quenching followed by grinding converts molten slag into GGBFS (Crossin, 2015) with a predominantly amorphous glassy structure composed mainly of CaO, SiO 2, Al 2 O 3 and MgO (Jamil et al., 2020). In the highly alkaline environment of hydrating cement, the amorphous glass phase is progressively activated, promoting the formation of additional C-S-H and C-A-S-H gels, in which Al partially substitutes for Si (Zhao et al., 2021), thereby contributing to improved long-term mechanical performance and durability of cementitious materials (Binici et al., 2007; Sandybay et al., 2025). However, despite its high potential and extensive use in some regions such as western Europe, substantial quantities of blast furnace slag remain stockpiled worldwide. In Kazakhstan alone, more than 30 million tonnes of air-cooled blast furnace slag had accumulated in the slag dumps of ArcelorMittal Temirtau JSC by 2022 (Kadyrov et al., 2018), highlighting the broader regional need for research on slag valorization.

Waste glass is another promising SCM as it represents a considerable proportion of municipal solid waste (Delbari and Hof, 2024) and is still largely stockpiled and landfilled despite its high recycling potential (Khan et al., 2020). In Kazakhstan, the recycling rate of waste glass remains relatively low (Shon et al., 2022), at approximately 10%–11% per year (Montayev et al., 2025). As a non-biodegradable material, waste glass accumulates in landfills and creates long-term environmental concerns (Ahmed et al., 2023; Lai et al., 2026). Compared with conventional glass recycling, the use of finely ground waste glass in cementitious materials requires less stringent purity requirements, providing an attractive pathway for the valorization of mixed or non-recyclable glass waste (Mariaková et al., 2021). Finely ground waste glass contains a high proportion of amorphous silica (Shao et al., 2000), which reacts with CH released during cement hydration to produce additional C-S-H (Poudel et al., 2025). Moreover, its fine particles may act as fillers and nucleation sites, contributing to pore refinement and microstructural densification (Idir et al., 2011), while being sufficiently reactive to avoid issues with alkali-silica reaction.

MK is produced by the calcination of kaolin clay (Khatib et al., 2018) and is recognized as a highly reactive SCM (da Silva Andrade et al., 2019). Rich in reactive silica and alumina (Hosseinzadeh and Abdollahi, 2026), MK reacts with CH released during cement hydration, promoting the formation of additional C-A-S-H and hydrous aluminates (da Silva Andrade et al., 2019). The incorporation of Al into the hydrate structure results in more polymerized C-A-S-H with lower Ca/Si and higher Al/Ca ratios (da Silva Andrade et al., 2019). Furthermore, the filler effect of MK can accelerate early cement hydration, contributing to early strength development (El-Diadamony et al., 2018). As a result, MK improves the microstructure and mechanical performance of cementitious materials. Kaolin deposits are widely distributed in Kazakhstan (Nabokov et al., 1965), indicating the potential for future domestic MK production.

Although GGBFS, GWBG and MK have been extensively studied individually as SCMs, few studies have directly compared these three SCMs under identical replacement and curing conditions while considering hydration behavior, mechanical performance, heavy metal leaching, and calculated binder-production environmental indicators within the same experimental framework. It remains unclear whether higher mechanical performance is accompanied by lower embodied energy and CO 2. To examine this relationship, this study compares GGBFS, GWBG, and MK at a 20 wt% replacement level. The 20 wt% replacement level was selected as a moderate replacement level within the ranges previously investigated for GGBFS (El Sawda et al., 2026; Wu et al., 2017), GWBG (Harrison et al., 2020; Islam et al., 2017), and MK (Huang et al., 2025; Al-Hashem et al., 2022) in cement-based systems. The same replacement level was applied to all three SCMs to provide a consistent basis for comparison. These three materials were selected to represent different types of SCMs: GGBFS as an industrial by-product, GWBG as a recycled waste-derived material, and MK as an aluminosilicate material. GWBG, produced from post-consumer glass collected in Astana, represents a locally sourced waste-derived SCM, while GGBFS and MK were commercially supplied reference materials. The evaluation includes hydration behavior, strength development, microstructural evolution, heavy metal leaching, embodied energy (E-energy), and embodied CO 2 emissions (E-CO 2). Accordingly, this paper presents a comparative experimental assessment of three BC systems: BCGGBFS, BCGWBG, and BCMK.

Materials

Three SCMs, namely, GGBFS, GWBG and MK, were used in this study. The GGBFS was supplied as GreenCems® GGBS-450 (Mechel-Materials, Russia), while MK was supplied as Metaver® I (Newchem GmbH, Austria). Both GGBFS and MK were supplied in a pre-ground powdered form and were used as received without further grinding. In contrast, GWBG was produced from post-consumer soda-lime glass bottles and jars collected in Astana, Kazakhstan. The collected glass was crushed and subsequently ground using a laboratory ball mill. The PC used in this study was PC 400 D0 (CEM I 32.5N), manufactured by Jambyl Cement (Kazakhstan). GGBFS was characterized by a specific surface area (SSA) of approximately 522 m 2 /kg and a vitreous phase content of about 67%, whereas MK consisted mainly of SiO 2 (56.43 wt%) and Al 2 O 3 (38.28 wt%). The graded sand supplied by Alilnert Group (Kazakhstan) was used as the fine aggregate in all mortar mixtures.

Characterization of raw materials

The chemical composition of the materials was determined by X-ray fluorescence (XRF) using an Axios Max spectrometer (PANalytical, Malvern, United Kingdom) equipped with the Omnian software package. The samples were prepared as fused beads using an automatic XrFuse 6 fusion system (XRF Scientific, Brussels, Belgium). The crystalline phases were identified by X-ray diffraction (XRD) using a SmartLab XRD system (Rigaku, Tokyo, Japan), equipped with Cu Kα radiation. The instrument was operated at 40kV and 50mA, and diffraction data were collected over a 2θ range of 5-60 o. With a step size of 0.1 o and scan rate of 5 o 2θ/min (1.2 s per step) using a continuous scan mode. The particle size distribution (PSD) was determined using a Mastersizer 3000 (Malvern, United Kingdom), while the surface morphology of precursor materials was characterized by scanning electron microscopy (SEM) using a JEOL JSM-IT200 microscope (Tokyo, Japan). The total concentrations of heavy metals, including Hg, Cu, Pb, As, Ni, Cr and Zn in the raw materials were determined by inductively coupled plasma optical emission spectroscopy (ICP-OES, iCAP 6300, Thermo Fisher Scientific) following microwave-assisted acid digestion using a mixture of HNO 3, HCl, and HF, in accordance with EN 13656 (EN 13656:2020, 2020).

Preparation of specimens

Mix proportioning

Table 1 summarizes the mix proportions of the paste and mortar mixtures used in this study. A constant water-to-binder (w/b) ratio of 0.50 was adopted for all mixtures, and the sand-to-binder ratio for the mortar mixtures was fixed at 2.75. Each blended cement (BC) mix was designed with a 20 wt% replacement level of PC by the respective SCM.

Preparation of BC paste and mortars

Paste and mortar specimens were prepared using the mix proportions listed in Table 1. All mixtures were prepared using a 5 L laboratory mixer (Hobart, Troy, OH, USA) in accordance with ASTM C305-20 (ASTM C305-20, 2020). The fresh pastes were cast into plastic containers, while the mortars were cast into 50 × 50 × 50 mm steel cube molds. All specimens were compacted using tamping rod, sealed to prevent moisture loss, and stored at 23 °C ± 2 °C for 24 h. After demolding, the specimens were cured at 23 °C ± 2 °C under a relative humidity of ≥95% until the designated testing ages of 7, 28, and 90 days.

Characterization of the hydrated pastes

The reference paste (100 wt%) and the pastes containing 20 wt% GGBFS, GWBG, or MK, cured at 23 °C ± 2 °C for 7, 28, and 90 days, were used to investigate the microstructural evolution of BC pastes. Crystalline phases were identified by XRD analysis as described in Section 2.2. The FTIR spectra were recorded using an Alpha II ATR-FTIR spectrometer (Bruker, Germany) over the wavenumber range of 400–4000 cm -1. The microstructure was examined by SEM, while the chemical composition of hydration products was determined by energy-dispersive X-ray spectroscopy (EDS) using the equipment described in Section 2.2. At least 50 EDS spot analyses were performed for each sample at a magnification of 1000×. Thermogravimetric analysis (TGA) was performed using a Netzsch STA 449 F3 Jupiter system to evaluate the hydration products and determine CH content, which was calculated according to Equation 1: where ω C H is the mass loss during dehydration of CH as percentage of the ignited mass (%), M CH is the molar mass of CH (g/mol), M water is the molar mass of water (g/mol).

TGA tests were conducted from 30 °C to 1000 °C at a heating rate of 10 o C/min under an argon atmosphere.

Strength activity index

The mechanical performance of BC mortars was evaluated using the SAI method, with replacement conducted on a mass basis and a constant w/b ratio. The SAI represents the relative compressive strength of mortars containing 20 wt% GGBFS, GWBG or MK as a partial replacement for PC compared with the reference PC mortar, and was calculated according to Equation 2: where:

S B C = compressive strength of the mortar containing GGBFS, GWBG or MK.

S P C = compressive strength of the reference mortar.

The compressive strength of mortars was measured at 7, 28 and 90 days in accordance with ASTM C109/C109M (ASTM C109/C109M-20, 2020). A constant loading rate of 0.125 MPa/s was applied to all specimens. The results for each mortar are reported as the mean ± standard deviation of three specimens. Statistical differences among the mixtures at each curing age were assessed using one-way ANOVA followed by Tukey’s HSD test (p < 0.05).

Heavy metal leaching test

The leaching behavior of the hardened mortars was evaluated after 28 days of curing in accordance with EN 12457–2:2002 (CEN, 2002). The mortars were crushed to a particle size below 4 mm. The crushed samples were mixed with Milli-Q water at a liquid-to-solid ratio of 10 L/kg and agitated using a horizontal shaker for 24 ± 0.5 h at room temperature. The leachates were collected, filtered through 0.22 µm membrane, and acidified to pH < 2 with 1% HNO 3 prior to analysis. The concentrations of Zn, Pb and Cr in the leachates was measured using inductively coupled plasma mass spectrometry system (ICP-MS, iCAP RQ, Thermo Fisher Scientific). To ensure the reproducibility of the measurements, all tests were performed in triplicate.

Preliminary sustainability screening assessment

To provide a preliminary evaluation of the environmental benefits of replacing PC with SCMs, a simplified environmental assessment was conducted. The embodied energy (E-energy) and embodied CO 2 emissions (E-CO 2) associated with binder production were selected as the environmental indicators. The evaluation considered only the production stage of the constituent materials, while mixing, curing, use, and end-of-life processes were excluded.

The E-energy and E-CO 2 factors of PC, GGBFS, GWBG, and MK were adopted from previous studies and are summarized in Table 2. The values for the BCs were calculated based on the replacement ratio of each SCM. The strength normalized indicator E-CO 2 /f c,28 was calculated by dividing the binder production E-CO 2 value of each mortar by its corresponding mean 28-day compressive strength (f c,28), following the performance normalization approach reported by Choi et al. (Choi et al., 2026).

Characterization of raw materials

The chemical compositions of PC, GGBFS, GWBG and MK presented in Table 3, while their normalized CaO-SiO 2 -Al 2 O 3 compositions were plotted in the ternary diagram in Figure 1. As shown in Figure 1, SiO 2 is the predominant component of all the tested SCMs. GGBFS is characterized by a relatively high CaO content, while GWBG is rich in SiO 2, indicating its potential pozzolanic reactivity (Barbhuiya et al., 2025). In contrast, MK contains high proportions of both SiO 2 and Al 2 O 3, which are essential for its chemical reactivity and ability to promote the formation of alumina-bearing hydration products (Mohammed, 2017).

As shown in Figure 2, XRD was used to identify the mineralogical phases of GGBFS, GWBG and MK. The diffraction pattern of GGBFS revealed the presence of several crystalline phases, including melilite, Ca 2 (Al, Mg)[(Al, Si)SiO 7 ], and calcite (CaCO 3). Melilite represents a solid solution between gehlenite (Ca 2 Al 2 SiO 7) and åkermanite (Ca 2 MgSi 2 O 7) (Sun et al., 2020). In contrast, GWBG exhibited a broad amorphous halo between 15° and 40° 2θ, indicating a predominantly amorphous glassy structure consistent with the silica-rich composition illustrated in the CaO-SiO 2 -Al 2 O 3 ternary diagram (Figure 1). The amorphous and silica-rich nature of GWBG is consistent with its potential pozzolanic reactivity when finely ground (Shi and Zheng, 2007). A small peak corresponding to quartz (SiO 2) (Xiao et al., 2021) was detected. The XRD pattern of MK showed both amorphous and crystalline characteristics. Unlike GWBG, distinct diffraction peaks were superimposed on the amorphous background. The observed peaks were ascribed to a minor quartz impurity.

In addition to the chemical and mineralogical characteristics, the particle size distribution (PSD) is one of the main factors influencing the reactivity of SCMs in cementitious systems (Ndahirwa et al., 2022). As shown in Figure 3, GGBFS and MK exhibited finer particle size distributions than PC and GWBG. The finer particle size of GGBFS and MK provides a larger reactive surface area, which is expected to enhance their reactivity in cementitious systems (Tural et al., 2024). It has also been reported that glass particles with a particle size lower than 75 µm undergo a pozzolanic reaction (Ahmad and Zhou, 2023).

SEM images were used to characterize the morphology of GGBFS, GWBG, and MK (Figure 4). The SEM micrograph of GGBFS (Figure 4A) revealed irregular angular particles with rough surfaces and a broad particle size distribution. The particles exhibited numerous edges and fractured surfaces that are typical of GGBFS (Yu et al., 2023). In contrast, GWBG (Figure 4B) consisted predominantly of angular particles with relatively smooth surfaces and sharp edges, characteristic of fractured glass. Only a limited amount of adhered fine particles was observed on the particle surfaces. The SEM micrograph of MK (Figure 4C) revealed numerous fine flake-like particles forming loosely agglomerated clusters. Such particle morphology has been reported to influence the packing behavior of MK powders (Keskin-Topan et al., 2024).

Considering the potential environmental and human health risks associated with Hg, Cu, Pb, As, Ni, Cr and Zn (Cheng et al., 2023), their concentrations were determined in the raw materials. Among these elements, Pb, Cr and Zn were detected (Table 4), whereas Hg, Cu, As and Ni were below the limit of detection (LOD) in all materials tested. The relatively high Pb concentration in GWBG may be related to its heterogeneous post-consumer origin and the possible presence of Pb-containing enameled glass (Turner, 2019). Therefore, Pb, Cr and Zn were selected as the target elements for subsequent leaching assessment of the hardened mortars.

Characterization of hydration products

The hydration products formed in cement pastes play a key role in determining their microstructure and mechanical performance (Qian et al., 2024). Therefore, XRD analysis was performed to investigate the phase evolution of the BC pastes, and the corresponding diffraction patterns are presented in Figure 5. At 7 days (Figure 5A), the main crystalline hydration products identified in all pastes were calcium hydroxide (CH), ettringite (AFt), calcium monocarboaluminate (Mc), and calcite. Residual clinker phases, including alite (C 3 S) and belite (C 2 S), were also detected at 32 o -33 o and 52 o (Zhao et al., 2021). These clinker phases remained detectable throughout the curing period up to 90 days (Figures 5A–C), although their diffraction intensities gradually decreased with curing age, which is consistent with the continued development of hydration products.

CH was identified by its characteristic peaks (Stepkowska et al., 2004). At all curing ages, the intensity of the relatively isolated peak at 18.1 o was lower in the BC pastes than in PC and gradually decreased with increasing curing age up to 90 days (Figures 5A–C), with the most pronounced reduction observed in BCMK. This reduction may reflect both clinker dilution and CH consumption during hydration and may be associated with the formation of additional C-S-H and C-A-S-H (Chang et al., 2022; Zhang Y. et al., 2024).

The peak centered at 29.5 o became progressively more pronounced with increasing curing age in the BC pastes, particularly in BCGWBG and BCMK, with the latter exhibiting the highest intensity. This peak could contain overlapping contributions from poorly crystalline C-(A)-S-H and calcite (Martins Torres et al., 2020). Due to this overlap, the sharp component of this peak was indexed only as calcite, with the C-(A)-S-H potentially contributing to the broad signal underneath the sharper calcite peak.

Differences were also observed in the AFt peaks among the BC pastes. From 7 to 90 days of curing, AFt peaks were slightly more intense in BCGGBFS and BCGWBG than in PC, while BCMK showed the highest intensity. A similar trend was observed for Mc, with BCMK exhibiting the most pronounced peaks. The higher intensities of AFt and Mc in BCMK are consistent with the contribution of reactive alumina from MK, in agreement with its Al 2 O 3 content determined by XRF (Table 3), to the formation of these Al-containing hydration products (Song et al., 2023). Overall, the most pronounced changes in the hydration products were observed in BCMK, as suggested by its lower CH and higher AFt and Mc peaks intensities.

The hydration products identified by XRD were further investigated by FTIR analysis (Figure 6). The band observed at around 946 cm -1 is assigned to the asymmetric Si-O stretching vibration of C-S-H (Skocek et al., 2020). The band shifted slightly toward higher wavenumbers from 7 to 28 days in all pastes, with little further change at 90 days (Figures 6A–C). The most pronounced shift was observed in BCMK. Such a shift has been associated with a lower Ca/Si ratio and a higher degree of silicate chain polymerization in C-S-H (Hou et al., 2014). Changes in the intensity of the band at 452 cm -1, assigned to the in-plane Si-O bending vibration of SiO 4 tetrahedra, are also consistent with possible structural reorganization and increased polymerization of silicate units during hydration (Mollah et al., 2000). Only minor changes were observed in the intensities of the bands at 1110 cm -1 and 3670 cm -1 during curing. These bands are assigned to AFt (Wang et al., 2026) and Mc (Fernández-Carrasco and Vázquez, 2009), respectively. The band at 3640 cm -1, assigned to the symmetric O-H stretching vibration of CH (Moreno-Reyes et al., 2026), gradually decreased in intensity in all BC pastes with increasing curing age up to 90 days. This decrease suggests the progressive consumption of CH during hydration. The findings are in good agreement with the XRD results, which also showed lower CH peak intensities in the BC pastes, particularly in BCMK. The absorption bands at 872 and 1417 cm -1 are assigned to the carbonate vibrations of calcite (Yılmaz and Olgun, 2008). The higher intensities of these bands in BCGGBFS at 7 days (Figure 6A) are consistent with the presence of calcite in the raw GGBFS (Section 3.1; Figure 2). At 28 and 90 days (Figures 6B,C), these bands became less intense, which may be associated with the formation of Mc during hydration (Deng et al., 2026). This interpretation is consistent with the slightly higher Mc peak intensity observed in BCGGBFS by XRD at 28 days. Finally, the absorption peaks centered at 1645 cm -1 and 3390 cm -1 correspond to the H-O-H bending vibration of water associated with sulfate-containing phases (Kunaev et al., 2023) and O-H stretching vibration of bound water (Choudhary et al., 2015), respectively.

To further examine the morphology of the hydration products and the microstructural evolution of the BC pastes, SEM analysis was performed. Representative SEM images of PC and BC pastes cured for 7, 28 and 90 days are presented in Figures 7–9.

After 7 days (Figure 7), microstructural features consistent with C-S-H, C-A-S-H, CH and ettringite (AFt) were observed in the pastes, in agreement with the XRD and FTIR results. Among these products, C-S-H, C-A-S-H and AFt are primarily responsible for the early strength development of cementitious materials (Li et al., 2019; Chand et al., 2021). In BCGGBFS (Figure 7B), the unreacted slag particles were coated with a relatively thick layer of hydration products, suggesting the initiation of slag hydration. In contrast, the unreacted GWBG particles were either covered by a thin layer of hydration products or remained largely uncoated (Figure 7C), suggesting their lower early-age reactivity. This observation agrees well with previous studies reporting that the pozzolanic reaction of glass proceeds slowly under ambient curing conditions, with significant pozzolanic activity becoming evident only after approximately 28 days (Mohan et al., 2026). At the same time, hydration products were observed to precipitate around the glass particles, suggesting that the glass surfaces may have provided nucleation sites for their precipitation. In the representative SEM image, BCMK appeared more compact than the other pastes (Figure 7D), with plate-like features consistent with AFm-type phases appearing more frequently than in other pastes. This interpretation is supported by the identification of Mc by XRD and FTIR. Needle-like features consistent with AFt were also more abundant in BCMK, which agrees well with the more intense AFt peaks observed by XRD. Together, these observations are consistent with the contribution of reactive alumina from MK to the early development of Al-containing hydration products.

At 28 days (Figure 8), the hydration products surrounding GGBFS particles became considerably more abundant, and most slag particles in the examined region appeared almost surrounded by a continuous hydrate matrix (Figure 8B). The visible areas between particles in the examined region appeared increasingly filled with hydration products, consistent with continued GGBFS hydration. Many partially reacted GWBG particles were still clearly observed in BCGWBG paste (Figure 8C), suggesting that the reaction of GWBG remained incomplete at this curing age. Several cracks and micropores were also observed. Their presence could be associated with interfacial heterogeneity around the partially reacted glass particles, although some cracks may also have formed during sample preparation. In contrast, BCMK appeared relatively compact in the representative SEM image, with only isolated micropores in the examined region (Figure 8D). This observation is consistent with the continued development of Al-containing hydrate matrix.

After 90 days, many GWBG particles appeared more closely integrated into the surrounding hydration products (Figure 9C), and the investigated region appeared visually more compact than that observed at 28 days. However, the representative BCGWBG image showed more visible pores than the corresponding BCGGBFS and BCMK images, giving the examined region a less compact appearance (Figures 9B,D). In BCGGBFS, the further integration of residual slag particles into the hydration matrix suggests continued slag reaction, whereas BCMK showed a relatively compact appearance in the representative SEM image, with only isolated visible pores in the examined region.

To further investigate the effect of SCM incorporation on the composition of the hydration products, SEM-EDS point analysis was performed on PC and BC pastes cured for 28 days Figure 10 represents the relationship between the Al/Ca and Si/Ca atomic ratios obtained from the analyzed points. Because the EDS analyses were performed on unpolished surfaces, some scatter is expected. Atomic ratios are therefore used to compare compositional trends among the analyzed points. The compositional regions shown in Figure 10 are treated as approximate and are interpreted together with the XRD and FTIR results.

As AFm represents a family of hydrated calcium aluminate phases (Yang et al., 2025), the AFm compositional region shown in Figure 10 may be associated with the Mc phase identified by XRD and FTIR. Compared with PC, the BC pastes exhibited generally higher Si/Ca ratios among the points tentatively associated with C-S-H and C-A-S-H. This difference may partly reflect the lower CaO contents of the SCMs relative to PC, as determined by XRF (Table 3). Similar reductions in the Ca/Si ratio have been reported for cement systems incorporating MK and GGBFS (Adu-Amankwah et al., 2025; Zhao and Khoshnazar, 2020). The Al/Ca ratio generally increased in the order BCGWBG < BCGGBFS < BCMK, following Al 2 O 3 contents of the corresponding SCMs determined by XRF (Table 3). The slight increase in the Al/Ca ratio observed for BCGWBG is consistent with previous studies on waste glass-blended PC systems (Zheng, 2016). Several points with high Si/Ca ratios (>0.7) were observed in BCGWBG and were interpreted as Si-rich regions that may be associated with partially reacted or unreacted glass particles, as also observed in the SEM image (Figure 8C). Overall, the higher Si/Ca and Al/Ca ratios observed in the BC pastes are consistent with more Si-rich C-S-H compositions, while the higher Al/Ca ratios, particularly in BCMK, are consistent with greater Al incorporation in C-A-S-H. This interpretation of more Si-rich silicate hydrates also agrees with the shift the Si-O band toward higher wavenumbers observed by FTIR.

Figure 11 presents the DTG curves of PC and BC pastes cured for 7, 28 and 90 days. Four characteristic mass loss peaks were observed in all tested samples. The most pronounced mass loss peak, located at about 100 °C, is mainly associated with the dehydration of C-S-H, C-A-S-H and AFt, as well as the loss of physically bound water (Jing et al., 2023). Throughout the curing period (Figures 11A–C), BCMK exhibited the highest mass loss in this temperature region, followed by PC, BCGWBG and BCGGBFS. Such behavior could be attributed to the formation of a larger amount of hydration products.

A smaller mass loss in the temperature range 120–200 °C was assigned to the dehydration of the Mc phase (Bakolas et al., 2006), while the CH dehydration peak was observed between 400 and 500 °C. As an isolated peak, it was used to quantify the CH content in the hydrated PC and BC pastes based on the mass loss within this temperature range. The calculated CH contents at different curing ages are presented in Figure 12.

To distinguish CH consumption from the effect of replacing 20% of PC, the theoretical CH contents were calculated as 80% of the corresponding PC values. These values were 11.84, 11.20, and 11.68% at 7, 28, and 90 days, respectively. All BC pastes showed measured CH contents below the corresponding theoretical values.

The differences between the measured and theoretical values suggest that SCM reactions contributed to CH consumption in addition to the dilution effect (Jia et al., 2019; Wild and Khatib, 1997; Mejdi et al., 2019). Among the tested BC pastes, BCMK exhibited the lowest CH content throughout the curing period. Relative to the theoretical dilution levels, BCMK showed the largest reductions in CH content at all curing ages. This suggests higher apparent CH consumption for BCMK than for the other tested SCMs. The lower CH content of BCMK is also consistent with the lower CH peak intensity observed by XRD and the weaker O-H stretching band at 3634 cm -1 in the FTIR spectra.

BCGGBFS showed only a slight increase in CH content with curing time, whereas BCGWBG exhibited a slight decrease. These trends suggest that Ca released during reaction of GGBFS may partly compensate for CH consumption, while the trend observed for GWBG is consistent with the gradual development of its pozzolanic reaction up to 90 days. The partially reacted GGBFS and GWBG particles observed by SEM at 28 days are consistent with incomplete reaction at this curing age.

The final endothermic peak between 700 and 800 o C corresponds to the release of CO 2 from the decomposition of calcite. The slightly higher calcite content observed in BCGGBFS compared with PC may be partly associated with the calcite originally present in the raw GGBFS (Section 3.1).

The main hydration products identified in the investigated pastes and their evolution with curing age are summarized in Table 5.

SAI

Figure 13 shows the SAI values of mortars incorporating 20 wt% GGBFS, GWBG and MK after 7, 28 and 90 days of curing. All BC mortars exhibited SAI values above 80% throughout the curing period.

The observed strength development is discussed below in relation to the hydration evolution summarized in Table 5. No statistically significant differences were found among the mortars at 7 and 90 days (p > 0.05). At 28 days, only BCGWBG showed significantly lower compressive strength than PC (p = 0.033). The observed SAI trends are consistent with the combined influence of PSD, chemical composition, amorphous content, and reaction behavior.

The SAI of BCGGBFS was numerically slightly lower than that of the control mortar at 7 and 28 days. At later curing ages, the progressive hydration of GGBFS was accompanied by an increasingly compact appearance in the representative SEM images and a numerical increase in SAI.

At 7 days, BCGWBG showed slightly higher mean SAI than PC, but the difference was not statistically significant. This early strength response may be partly associated with the “filler effect” of the fine GWBG particles (Baikerikar et al., 2023). Although the pozzolanic reaction of GWBG was still limited at this age, as evidenced by the partially unreacted glass particles observed in the SEM image (Figure 7C), the precipitation of hydration products around these particles suggests that their surfaces may have acted as nucleation sites. At 28 days, the SAI of the BCGWBG mortar decreased to a value lower than that of the control mortar. SEM images still revealed large unreacted GWBG particles after 28 days (Figure 8C), indicating that the reaction of the glass powder remained incomplete. This interpretation is further supported by the TGA results, which showed only a slight change in CH content between 7 and 28 days (Figure 12), suggesting that the pozzolanic reaction was still insufficiently developed to fully compensate for the reduced PC content. Consequently, the limited formation of additional C-S-H at this stage may have contributed to the lower SAI observed relative to the control mortar. At 90 days, the continued pozzolanic reaction of GWBG likely contributed to the gradual formation of additional C-S-H, resulting in the subsequent strength development.

In comparison, BCMK maintained a high SAI throughout the entire curing period, exceeding 94% at all ages and reaching approximately 104% after 90 days. This performance is consistent with combined effects of the chemical composition, amorphous content, PSD, and reaction behavior. The formation of additional hydration products was supported by the XRD, FTIR, TGA, and SEM results.

Heavy metal leachability

The hardened mortars were subjected to the EN 12457–2 (CEN, 2002) leaching test to evaluate the potential release of Pb, Cr and Zn detected in the raw materials (Section 3.1; Table 3). The concentrations of these metals in the leachates of the BC mortars (80% PC and 20% of GGBFS, GWBG or MK) are presented in Table 6. The waste acceptance limits specified in the European Council Decision 2003/33/EC (Council of the European Union, 2003) are also presented for comparison. All samples produced alkaline leachates, with final pH values ranging from 12 to 12.5, and no visible precipitates were observed.

Among the investigated metals, only Cr was detected in the leachates, while Pb and Zn remained below the LOD in all mortars. The low release of Pb and Zn under the selected test conditions may be associated with their retention within cementitious matrix through several possible mechanisms, including sorption onto hydration products, precipitation of low-solubility phases, association with C-S-H or other hydrate phases (Moulin et al., 2000). The measured Cr concentrations ranged from 0.7 to 1.15 mg/kg. These values exceeded the inert waste acceptance limit of 0.5 mg/kg but remained substantially below the non-hazardous waste limit of 10 mg/kg specified in Council Decision 2003/33/EC. Overall, under the EN 12457–2 batch-leaching conditions applied to the 28-day mortars, the incorporation of 20 wt% GGBFS, GWBG or MK resulted in only limited release of the examined metals. All measured values complied with the non-hazardous waste acceptance limits specified in Council Decision 2003/33/EC under these specific test conditions.

Preliminary screening assessment of environmental sustainability

The environmental evaluation considers both trace-metal behavior and binder-production indicators. The trace-element contents of the raw materials (Table 4) are related to their subsequent release from the hardened mortars (Table 6). The preliminary sustainability screening additionally considers binder-production E-energy and E-CO 2, with the results summarized in Table 7. The assessment considers only literature-derived binder-production factors and excludes mixing, curing, use and end-of-life stages. The GWBG factor reported in (Hossain et al., 2017) includes the collection, transport, crushing, grinding and sieving of waste glass, which is broadly similar to the preparation route used in this study. For GGBFS and MK, which were commercially sourced in this study, the environmental factors were adopted from published literature data (Kanagaraj et al., 2023; Jones et al., 2011). As expected, the incorporation of SCMs reduced both E-energy and E-CO 2 compared to PC owing to the lower clinker content of the blended systems.

Among the tested SCMs, the incorporation of GGBFS and GWBG resulted in comparable calculated binder-production indicators, with both blended systems exhibiting substantially lower E-energy and E-CO 2 than PC. BCMK showed the highest E-energy among the BCs due to the additional thermal treatment required during MK production. Nevertheless, its calculated indicators remained lower than those of PC. Overall, within the assessed binder-production scope, the partial replacement of PC with the three selected SCMs resulted in lower estimated E-energy and E-CO 2 values than PC.

To account for differences in mechanical performance, E-CO 2 was normalized by the mean 28-day compressive strength (E-CO 2 /f c,28). BCMK showed the lowest E-CO 2 /f c,28 value, followed by BCGGBFS, while BCGWBG was comparable to PC.

This study evaluated the effects of GGBFS, GWBG, and MK on the hydration, mechanical performance, heavy metal leaching, and calculated binder-production indicators of blended cements. GWBG represented a locally sourced waste-derived SCM, whereas GGBFS and MK were commercially supplied. Based on the experimental results, the following conclusions can be drawn: XRD, FTIR, SEM-EDS, and TGA demonstrated that the incorporation of GGBFS, GWBG, and MK influenced the hydration behavior and microstructural development of blended cements. BCMK exhibited the lowest CH content throughout the curing period, consistent with the more pronounced formation of Al-containing hydration products, while BCGGBFS and BCGWBG showed continued hydration throughout the curing period.

All blended cement mortars showed satisfactory mechanical performance, with SAI values exceeding 80% throughout the curing period. BCGGBFS showed gradual strength development with curing age, while BCMK maintained high mechanical performance at all curing ages, reaching SAI of about 104% at 90 days. At 28 days, only BCGWBG showed significantly lower compressive strength than PC (p = 0.033).

The preliminary screening assessment showed that the incorporation of GGBFS, GWBG, and MK resulted in lower calculated binder-production E-CO 2 and E-energy values than PC. The reductions in E-energy and E-CO 2 emissions ranged from 10.9% to 17.2% and 12.9%–18.9%, respectively. GGBFS and GWBG provided the greatest reduction in embodied energy, while all blended cements showed lower calculated binder-production E-CO 2 values than PC.

Under the EN 12457–2 batch-leaching conditions applied to the 28-day mortars, the concentrations of the investigated trace elements in the leachates remained below the non-hazardous waste acceptance limits, indicating limited metal release under these specific test conditions.

The present study focused on three SCMs at a single 20 wt% replacement level and a constant w/b ratio of 0.50. Hydration and microstructural characterization were performed on cement pastes, while mechanical performance and leaching were evaluated using mortars. Further studies considering different replacement levels would extend the present findings.

Overall, the results demonstrate that GGBFS, GWBG, and MK are promising supplementary cementitious materials with satisfactory mechanical performance and lower calculated binder-production indicators than PC. Building on these findings, future research may focus on long-term durability under relevant exposure conditions, particularly the potential alkali-silica reaction behavior of GWBG-containing systems, as well as mixture optimization, scale-up, and performance-based quality control to support broader practical application.