Foamed bitumen stabilisation of recycled steel slag aggregates utilising geopolymer as a secondary binder Authors: Youli Lin, Yueji Bai, Dulanja Dayaratne, Farshid Maghool, Arul Arulrajah Published: 2026 ABSTRACT AbstractFoamed bitumen stabilisation is a sustainable technique for enhancing the mechanical properties of unbound granular materials and rehabilitating aged pavements. This process produces a flexible pavement base and subbase layers with improved mechanical performance and durability. This study investigated foamed bitumen stabilised (FBS) mixtures for pavements in which the aggregate matrix consists entirely of industrial by-products. Three aggregate compositions, including 100% ladle furnace slag (100LFS), 100% electric arc furnace slag (100EAFS) and a combination of 50% EAFS and 50% LFS by mass (50EAFS/50LFS) were used as parent materials for foamed bitumen stabilisation. Alkali-activated fly ash (FA) and slag (S)-based geopolymer was used as a secondary binder to replace conventional carbon-intensive binders such as hydrated lime or cement while enabling rapid strength gain. The mechanical performance of the FBS mixtures was evaluated through unconfined compressive strength (UCS), indirect tensile resilient modulus (ITM r ), repeated load triaxial (RLT) and four-point flexural bending tests, while reinforcement mechanisms were studied using scanning electron microscopy. The results showed that the FBS mixtures incorporating 10%(FA+S) geopolymer achieved higher UCS values than benchmark mixtures with 2% cement, ranged from 4.05 MPa to 20.1 MPa with respected to different aggregate compositions and curing conditions. Two FBS mixtures, 50EAFS/50LFS + 3%B + 10%(FA+S) and 100LFS + 3%B + 10%(FA+S) complied the stiffness requirement for pavement applications, with 50EAFS/50LFS + 3%B + 10%(FA+S) being the optimum mixture and achieved highest ITM r of 4135 MPa and 2685 MPa under unsoaked and soaked condition, respectively. RLT and four-point flexural bending results also showed acceptable performance of the resilient modulus and fatigue resistance of the FBS mixtures incorporating 10%(FA+S) geopolymer. The findings of this study demonstrated the technical feasibility of using steel slag aggregates and geopolymers in foamed bitumen stabilisation for future sustainable road construction, offering improved mechanical properties and durability and a reduced carbon footprint.Keywords: Foamed bitumen, steel slag, geopolymer, recycling, ground improvement Introduction Rapid population growth and infrastructure expansion in major cities have led to a surge in waste generation from the industrial, municipal and construction sectors. Industrial wastes, such as steel processing by-products, are often discarded because of their perceived lack of value (Maghool et al., 2019), resulting in significant landfill accumulation and increased environmental risks. Meanwhile, the growing demand for construction materials has driven the overexploitation of natural resources, causing the acquisition of virgin quarry materials increasingly difficult and expensive. To address these challenges, recycling and reusing waste materials has become a priority for governments and policymakers. Over the past decade, extensive research has examined the engineering properties of various recycled materials, such as recycled concrete aggregate (RCA) (Tuladhar et al., 2020), reclaimed asphalt pavement (RAP) (Baradaran et al., 2024a), crushed brick (CB) (Zhang et al., 2021) and recycled glass (RG) (Yaghoubi et al., 2021). These studies underscore the growing potential to replace virgin quarry resources with recycled alternatives in civil infrastructures, particularly in pavement applications such as subgrades, base and subbase layers and engineered fills (Hall et al., 2022a). Despite these advancements, several challenges continue to limit the broader adoption of recycled materials in pavement systems, including volumetric instability, moisture susceptibility in fines-rich mixtures, environmental compliance concerns related to pH and leachate behaviour, and source-dependent variability in material properties (Ma et al., 2023; Suer et al., 2009). Considering these drawbacks and the increasing interest in waste-derived resources, steel slags have gained attention as a relatively underutilised alternative for pavement construction. With their high skid resistance, angularity, and strong affinity for bitumen, steel slags can enhance surface durability and stiffness when incorporated with appropriate mixture designs (Maghool et al., 2019). Steel is a critical material in engineering and construction, playing a vital role in nearly every aspect of modern life. In 2022, global crude steel production exceeded 1800 million tonnes, a dramatic increase from the 189 million tonnes produced in the 1950s (World Steel Association, 2023). As with many industrial processes, steel manufacturing generates unique by-products, namely steel slag. Depending on the production technology and furnace type, the production of one tonne of steel generates approximately 200–400 kg of steel slag (Maghool et al., 2019). Based on the global steel production figures (approximately 1800 million tonnes in 2020) (World Steel Association, 2023), it is estimated that 360 to 720 million tonnes of steel slag are generated annually. Steel production primarily employs two methods: the basic oxygen furnace method and the electric arc furnace method, with the latter relying exclusively on scrap steel. This study focuses on the by-products of the electric arc furnace method. The operation of an electric arc furnace requires electricity, lime (fluxes), coke (fuel), magnesium oxide and steel scrap to produce steel (Heidrich & Woodhead, 2010). This method generates two main types of slag at different stages, namely, electric arc furnace slag (EAFS) and ladle furnace slag (LFS). EAFS is produced during the initial phase of steel manufacturing, where an electric arc is used to melt scrap and fluxes. This results in the formation of molten slag that needs to be removed from the steel (Heidrich & Woodhead, 2010). The molten steel is then transferred to a refining ladle furnace for further processing. The ladle furnace process, also known as secondary metallurgy, represents the final stage of steel manufacturing. During this stage, impurities are further removed from the molten steel, leading to the formation of LFS. On average, refining one tonne of scrap steel in an electric arc furnace yields approximately 120–150 kg of EAFS and 30 kg of LFS (Heidrich & Woodhead, 2010). In recent years, steel slags, particularly ground granulated blast furnace slag, has been widely used as fillers or supplementary cementitious materials in the cement and concrete industry. However, other types of steel slag, such as ladle furnace slag (LFS) and electric arc furnace slag (EAFS), remain largely underutilised and are rarely considered for construction applications. Their practical use in road engineering is often constrained by environmental concerns, including potential leaching and associated risks to soil and groundwater quality. The lack of consistent, widely accepted standards or specifications to guide the reutilisation of steel slags in road engineering further limits their adoption in mainstream infrastructure projects. In addition, economic factors, such as transportation, additional processing and pre-treatment required of steel slags to meet relevant engineering standards as construction materials in road engineering, have historically posed significant barrier to large-scale applications. Consequently, despite their potential, LFS and EAFS are only beginning to gain attention as viable construction materials, remaining primarily at the research stage, and systematic studies to optimise their utilisation remain limited. Various techniques are available for stabilising recycled materials to improve their properties in road construction, with ordinary Portland cement and hydrated lime traditionally used as binding agents. Lime was first used as a stabilising agent of soils in modern construction practice in 1924, while cement has been used as the binding agent in soil stabilisation technology since the 1960s (Hamzah et al., 2015). However, these conventional binders pose significant financial and environmental issues, including high greenhouse gas emissions, extensive natural resource consumption and substantial energy consumption during production (Mohamad et al., 2022). To mitigate these challenges, the construction industry is keen on seeking environmentally friendly alternatives to cement and lime in an effort to promote sustainable building practices. Geopolymerisation technology has emerged as a promising solution, involving a chemical reaction that integrates minerals rich in silica and aluminate in an alkaline environment to produce alkali-activated materials, known as geopolymers. These low-carbon binders offer performance comparable to cement, while providing additional benefits, such as high mechanical strength, low shrinkage, acid resistance and low thermal conductivity (Duxson et al., 2007). Geopolymers have been effectively used in stabilising a variety of unbound granular materials for road base/subbase and ground improvement, including RCA (Lin et al., 2024), waste glass aggregate (Xiao et al., 2020) and RAP (Avirneni et al., 2016). Foamed bitumen stabilisation is another promising approach adopted by road agencies. This technique stabilises granular materials while minimising the use of water or additives, thereby reducing costs and environmental impact. The foaming bitumen process was first explored by Csanyi (1957), who developed a specialised nozzle system for producing bitumen foam and applying it in paving materials. It was demonstrated that injecting steam into bitumen to create foam was efficient but impractical for in-situ applications because of the need for specialised equipment such as steam boilers (Australia, 1971). In 1968, Mobil Oil Australia adapted this process by using water instead of steam in hot bitumen (Australia, 1971). They redesigned the nozzle to enable transverse intersections of hot bitumen and water streams. This modification produced bitumen foam with an expansion ratio exceeding 10 times the original volume of unprocessed bitumen, offering greater repeatability and consistency for bulk production (Bowering & Martin, 1976). Moreover, this improved method was simpler for field implementation, eliminating the need for steam generators (Lee, 1980). However, owing to the patent rights of Mobil Oil Australia, the use of this technology was initially restricted to countries such as Australia, New Zealand, Japan and Germany (Ebels, 2008). After the patent expired in 1987, the technology expanded globally, leading to further advancements in stabilisation technologies for pavements. Unlike traditional stabilisation methods that rely on binders such as cement or hydrated lime, which are prone to shrinkage cracking, foamed bitumen stabilisation produces flexible pavement layers with enhanced stiffness and durability (Ramanujam et al., 2009; Saleh, 2007). This innovative method involves mixing a primary binder, foamed bitumen, with a secondary binder, typically hydrated lime or cement, along with a graded unbound granular material to improve the strength, modulus and moisture resistance of traditional unbound granular pavements (Wirtgen Cold Recycling Technology, 2012). During this process, hot bitumen (typically between 160 °C and 200 °C) is injected and comes in contact with cold water (15 °C to 25 °C), causing the bitumen to expand to over 10 times its initial volume and then gradually decay (Nataatmadja, 2001). In this expanded state, the foamed bitumen has a lower viscosity, enabling it to coat the fine particles within the granular material during mixing. This coating process creates a bitumen mastic that binds the larger particles, enhancing the material properties. During compaction, the bitumen mastic fills voids between the aggregate particles, creating non-continuous bonds. These bonds are often likened to a ‘spot-welding’ effect, significantly strengthen the structure of granular material (Fu, 2009). Several studies have evaluated the performance of foamed bitumen stabilised (FBS) mixes, focusing on the strength, stiffness, and permanent deformation characteristics of FBS materials (Chomicz-Kowalska & Maciejewski, 2020; He & Wong, 2006; Pitawala et al., 2022). In these studies, a secondary binder, usually cement or hydrated lime in limited quantities up to 4%, was used to enhance the workability and early strength properties of the mixes. He and Wong (2006) thoroughly investigated the decay properties of foamed bitumen, demonstrating that the water content significantly impacts the bitumen decay properties, whereas insufflated air pressure has a much smaller effect. Chomicz-Kowalska and Maciejewski (2020) evaluated the impact of the bitumen penetration grade on the viscoelastic and mechanical properties of FBS mixtures. Their study revealed that the penetration grade of the bituminous binder influenced the compatibility of cold recycled mixtures, with higher penetration grades leading to a reduced air void content. Additionally, as the penetration of the foamed bituminous binder increased, there was a corresponding decrease in indirect tensile strength, indirect tensile stiffness modulus, and creep stiffness modulus values (Chomicz-Kowalska & Maciejewski, 2020). More recently, Pitawala et al. (2022) investigated the flexural fatigue life performance of FBS beam specimens. Their experimental results showed that the flexural fatigue life increased with higher density and bitumen content of the FBS beams but decreased with increasing temperature and moisture content. Moreover, a negligible difference in flexural fatigue performance was observed between the FBS mixtures containing 2% and 4% hydrated lime. While the FBS mixtures containing 2% cement showed significantly higher flexural fatigue life compared to mixtures with the addition of 2% hydrated lime. EAFS and LFS have been investigated for their potential applications as unbound granular materials in road construction. Maghool et al. (2017) reported that the low organic content, high density, strength and durability of LFS and EAFS enable them for use as high-quality aggregates in road construction, exhibiting performance comparable to that of natural aggregates. In addition, the leaching test results demonstrated that both the LFS and the EAFS are classified as inert solid wastes under the Environmental Protection Agency (EPA) Victoria guidelines (EPA Victoria, 2010), confirming their environmental safety for use in road construction applications. Other studies have also demonstrated the potential applications of EAFS and LFS in road engineering, primarily focusing on their use as aggregate replacements in asphalt mixes and base/subbase layers (Medina et al., 2026; Terrones-Saeta et al., 2020). Despite these advantages, a systematic approach to evaluate the fatigue characteristics of FBS mixtures incorporating steel slags remains absent. Pavement materials are prone to structural degradation over time because of repeated cyclic loading and varying environmental conditions (Sudarsanan & Kim, 2022). This deterioration manifests as various forms of damage, impacting driver and passenger comfort and safety. Fatigue cracking is the most common form of pavement damage (Baradaran et al., 2024b), reflects the pavement's ability to withstand repeated loading without fracturing, referred to as the ‘fatigue life’ (Sudarsanan & Kim, 2022). Therefore, investigating the fatigue responses of FBS materials under different stress conditions is crucial for designing durable and safe pavement systems. Previous studies have predominantly focused on FBS mixtures containing natural aggregates or RAP, with cement or hydrated lime used as secondary binders (Halles et al., 2013, Iwański & Chomicz-Kowalska, 2013). However, the performance of FBS mixtures incorporating steel slag, as well as the comparative effectiveness of using low-carbon geopolymers and conventional cement as secondary binders for FBS mixtures, remains underexplored. A systematic investigation of the overall mechanical properties, fatigue resistance and reinforcing mechanism of FBS mixture incorporating steel slags has not been studied and reported in the literature. In addition, available studies that systematically investigated the bitumen foaming characteristics for foamed bitumen stabilisation also remains limited. The present study investigated the bitumen foaming characteristics, mechanical properties and flexural fatigue life performance of FBS mixtures incorporating EAFS and/or LFS mixtures for pavement base layers. The foamed bitumen characteristics were evaluated through half-life and expansion ratio measurements to identify optimal bitumen foaming parameters. The mechanical performance was quantified through unconfined compressive strength (UCS), indirect tensile resilient modulus (ITM r ), repeated load triaxial resilient modulus (Mr ) tests and microstructure analysis of the samples. The optimum FBS mixture was further characterised through modulus of rupture, flexural modulus, and flexural fatigue tests. Three parent material blends were examined, including 100% EAFS, 100% LFS, and a 50% EAFS + 50% LFS blend by mass, denoted as 100EAFS, 100LFS and 50EAFS/50LFS, respectively. Commercially available Class 170 bitumen, with a minimum penetration value of 62, was used as the primary binder for all the FBS mixtures, with a cement or geopolymer used as the secondary binder to promote early strength gain (Jameson, 2019). Industrial by-products, Class F low-calcium fly ash (FA) and ground granulated blast furnace slag (S), were used as aluminosilicate source for alkali activation. A binary precursor comprising FA and S (denoted as FA+S) in a FA:S mass ratio of 50:50 was added at a dosage of 10% by mass of the parent materials. The precursor (i.e. FA and S) was mixed with a liquid alkaline activator consisting of sodium hydroxide (NaOH) and sodium silicate (Na2SiO3) solutions to produce the geopolymer binder. The NaOH solution had a concentration of 8 mol, and the ratio of Na2SiO3:NaOH ratio was maintained at 70:30. The control group, consisting of parent materials stabilised with 3% foamed bitumen and 2% cement, was used as a benchmark to compare the performance of mixtures stabilised with 3% foamed bitumen and 10% (FA+S) geopolymer. The findings of this study demonstrate the potential of using foamed bitumen to stabilise steel slags for pavement applications, thereby contributing to more sustainable and durable road construction practices. Experimental program Materials EAFS and LFS used in this study were sourced from a local steel manufacturer in Victoria, Australia. The collected EAFS and LFS were dried in a temperature-controlled oven at 100 °C ± 5 °C for at least 24 h to remove initial moisture. The materials were then sieved through a 19 mm sieve to prepare graded representative samples with a grain size of 19 mm minus for further assessments. Figure 1 shows the particle size distribution (PSD) curves of 100LFS, 100EAFS and 50EAFS/50LFS blends. The gradation curve of 100LFS falls within the ideal grading envelopes of granular materials for foamed bitumen stabilsation as per Austroads guide AGPT04D-19 (Jameson, 2019). The gradation of 100EAFS was much coarser than LFS, and fell outside the ideal grading envelopes of FBS materials for particle size fractions of 4 mm minus. In addition, the gradation of 50EAFS/50LFS blend was generally complied with the ideal grading envelopes of FBS materials. According to the Unified Soil Classification System (ASTM, 2020), the 100LFS, 100EAFS and 50EAFS/50LFS are classified as well-graded sand (SW), well-graded gravel (GW) and SW, respectively. Particle size distribution of 100LFS, 100EAFS and 50EAFS/50LFS blends. FA (Class F, low calcium), S, and cement used in the present study were provided by a local cement supplier in Victoria, Australia. Commercially available Class 170 bitumen and foaming agent were used in this study to produce foamed bitumen per the standard mixture design specifications outlined in Austroads test method AGPT/T302 (Austroads, 2017). Table 1 shows the detailed experimental program. The blends investigated in this study were: 100EAFS, 100LFS and 50EAFS/50LFS, respectively, used as a parent material for foamed bitumen stabilisation. The bitumen content was kept constant at 3% for all the FBS mixtures as the primary binder. Cement is commonly used as a secondary binder in foamed bitumen stabilisation at low dosages (typically 2% by weight of the parent material) owing to its ability to provide an optimal balance between stiffness and flexibility (Graziani et al., 2016; Yahya et al., 2025). Accordingly, a fixed cement dosage of 2% was adopted as the benchmark condition in this study, following the dosage recommended by Zhalehjoo (2022). The addition of 2% cement as a secondary binder was effective for preparing FBS 100LFS and 50EAFS/50LFS samples. However, the same dosage of 2% cement was ineffective for FBS 100EAFS, as the compacted specimens collapsed after demoulding due to the poor gradation and cohesionless nature of EAFS. Table 1.Experimental program.Sample IDOMC (%)MDD (mg/m3)Test typeCuring age (h or day)Curing temperature (°C)100LFS + 3%B + 2%C9.12.65UCS3-d25 °C, 40 °C ITM r 3-h, 3-d25 °C, 40 °C RLT3-d25 °C100LFS + 3%B + 10%(FA+S)9.22.63UCS3-d25 °C, 40 °C ITM r 3-h, 3-d25 °C, 40 °C RLT3-d25 °C100EAFS + 3%B + 2%C a 3.12.46N/AN/AN/A100EAFS + 3%B + 10%(FA+S)5.32.55UCS3-d25 °C, 40 °C ITM r 3-h, 3-d25 °C, 40 °C RLT3-d25 °C50EAFS/50LFS + 3%B + 2%C 7.92.57UCS3-d25 °C, 40 °C ITM r 3-h, 3-d25 °C, 40 °C RLT3-d25 °C50EAFS/50LFS + 3%B + 10%(FA+S)7.82.57UCS3-d25 °C, 40 °C ITM r 3-h, 3-d25 °C, 40 °C RLT3-d25 °CFlexural beam test b 28-d25 °C a FBS mixture unable to prepare UCS, ITM r and RLT samples as compacted specimens slumped after demoulding. b Beam tests were conducted only on the optimum mixtures (i.e. 50EAFS/50LFS + 3%B + 10%(FA+S)).Note: Day denoted as d; hour denoted as h; N/A: not available. Experimental program. FBS mixture unable to prepare UCS, ITM r and RLT samples as compacted specimens slumped after demoulding. Beam tests were conducted only on the optimum mixtures (i.e. 50EAFS/50LFS + 3%B + 10%(FA+S)). Note: Day denoted as d; hour denoted as h; N/A: not available. Geopolymer was tested in this study to replace the cement as a more sustainable secondary binder for foamed bitumen stabilisation of 100LFS, 100EAFS and 50EAFS/50LFS mixtures. Replacing cement with an alkali-activated (FA+S) geopolymer under identical mix designs resulted in acceptable early-age stiffness and enhanced moisture tolerance (Fang & Zhang, 2020), achieving a more favourable strength flexibility balance while reducing binder-related carbon footprint (Turner & Collins, 2013). For geopolymer synthesis, a binary solid precursor (P) comprising FA and S at a fixed mass ratio of 10% to the parent material were alkali activated by blending with a mixed alkaline liquid activator (L) composed of sodium hydroxide (NaOH) and sodium silicate (Na2SiO3). The (FA+S) precursor dosage of 10% was adopted following recommendation from an earlier study on geopolymer stabilisation of steel slags with similar gradation (Maghool, 2019). A fixed FA:S ratio of 50:50 was adopted following the recommendation of an earlier study (Lin, 2024). Previous studies investigated the effect of various Na2SiO3:NaOH ratios, as well as the ratios of L to the total mass of P on the strength development of geopolymer. It was evident that increasing the Na2SiO3:NaOH ratio and L/P ratio up to a limit would result in the highest strengths being achieved (Aliabdo et al., 2016). According to an earlier study (Lin et al., 2023), it was determined that the optimal NaOH:Na2SiO3 ratio and L:P ratio were 70:30 and 0.4, respectively, in terms of strength gain, for the activation of FA- and S-based geopolymers. Hence, a Na2SiO3:NaOH of 70:30 and a L/P ratio of 0.4 were adopted in this study. The NaOH solution with a concentration of 8 mol was selected in the present study given the consideration to the economic and safety factors (Lin et al., 2023). Reagent-grade NaOH beads (20–40 mesh) with 97% purity were dissolved in distilled water to prepare the 8 mol NaOH solution. The NaOH solution was sealed in a plastic container and allowed to cool for 24 h before use. The Na2SiO3 solution used in this study was a commercially available D-grade liquid Na2SiO3 with a molar ratio of 2 (Ms = SiO2/Na2O, 29.4 wt% SiO2, 14.7 wt% Na2O and 55.9 wt% H2O). Expansion ratio and half-life The expansion ratio (ER) of foamed bitumen is defined as the ratio between the maximum increased volume of bitumen due to formation relative to the normalised bitumen volume. The transpired time in seconds for collapse the foam volume to 50% of its maximum volume is termed as half-life (HL) (Martinez-Arguelles et al., 2014). The ER deemed a correlation of bitumen viscosity and provides implications for the dispersibility of the binder, enabling a uniform coating on the granular particles. Conversely, the HL is an index used to assess foaming stability and provides insights into the foam collapse rate (Iwański et al., 2014). The characterisations of C170 bitumen in terms of its ER and HL were evaluated according to the Austroads test method AGPT/T301 (Austroads 2017). A measuring rod and a 20 L bitumen container were used for the ER and HL characterisation of the C170 bitumen. The test series collaborated with the varying test conditions to explore their impact on the foamed bitumen properties. Two different bitumen heating temperatures assessed were: 175 °C and 185 °C, respectively. The foaming moisture content ranged from 1% to 4%, and a foaming agent content of 0.5% were assessed. The air pressure and water pressure remained constant at 5 bar and 6 bar, respectively, throughout the experiments. 500 g quantity of bitumen was foamed at each trial and sprayed at a rate of 100 g/s. Mixing procedures Wirtgen WLB 10 S bitumen foaming unit (see Figure 2b) and the standard mixing procedure outlines in the Austroads test method AGPT/T302 (Austroads 2017) was followed to produce foamed bitumen. Figure 2a shows the schematic of manufacture foamed bitumen. The foamed bitumen is produced by injecting a controlled amount of nebulised water with compressed air onto the hot bitumen inside the expansion vessel. During this transition, the binding properties are enhanced, and low viscosity bitumen is added into the mixer pre-filled with granular materials to produce FBS materials. In terms of operating conditions, pre-heated bitumen (at between 105 °C and 110 °C) was poured into the apparatus vessel with pre-programmed temperatures (175 °C or 185 °C, respectively) maintained throughout the process. The desired amount of foaming agent (0.5% by mass of bitumen) was added to the bitumen to enhance the bitumen foaming characteristics. The foaming moisture content, air pressure and water pressure were pre-programmed to the desirable parameters to produce foamed bitumen. (a) Schematic of producing foamed bitumen (figure modified from (Jameson, 2019)), (b) Wirtgen WLB 10 S bitumen foaming unit and (c) Lab mixer. Initially, a calculated amount of dried parent material and a secondary binder (either a geopolymer or cement) were dry mixed in a Wirtgen WLM 30 Lab Mixer (see Figure 2c) for 2 mins to ensure uniform dispersion. For FBS mixtures incorporated geopolymer, the secondary binder comprised 10% precursor by mass of aggregate, denoted as 10%(FA+S), while the L consisted of 30% NaOH and 70% Na2SiO3 solutions. The required amounts of water and L were then gradually introduced to reach the predetermined optimum moisture content (OMC), followed by wet mixing for an additional 2 mins to obtain a homogeneous paste coating on the aggregate. For FBS mixtures incorporated cement, a precise amount of water was added to achieve OMC, and wet mixing was continued for 2 mins to ensure uniform consistency. Foamed bitumen was then sprayed into the mixer at a perpetual rate of 100 g/s through the orifice nozzle while mixing continued. The final mixing stage was extended for a further 2 mins to promote complete dispersion of the bitumen throughout the aggregate matrix. Upon completion of mixing, the materials were transferred into sealed containers, and immediate compaction was performed to inhibit moisture loss from the mixture. It should be noted that the total amount of water required to satisfy the OMC of each mixture was added in a single stage for mixing. This approach was adopted because of the higher water absorption of steel slag which requires more moisture to ensure adequate wetting and optimum mixing. In addition, the commercially available Na2SiO3 has relatively high viscosity; therefore, pre-wetting the aggregates to a relatively high moisture content improved the uniform dispersion of Na2SiO3 and prevented agglomeration of the stabilised material during mixing. The proposed mixing time and a mixer rotational speed of 70% were selected based on trials aimed at achieving the most homogenous mixing conditions before the wet mixtures began to harden. The same mixing procedures was applied to all the mixtures in this study to maintain consistency and reproducibility. Table 1 shows the material composition of each mixture investigated in the present study. Sample preparation and testing The PSD test was performed in accordance with AS 1141.11 (Australian Standard, 2020). The modified Proctor compaction was conducted to determine the corresponding optimum moisture content (OMC) and maximum dry density (MDD) of the foamed bitumen-stabilised mixtures as per the AS 1289.5.2.1 (Australian Standard, 2003). Unconfined compressive strength (UCS) of FBS mixtures were investigated in accordance with ASTM D5102 (ASTM, 2009). The UCS tests were performed on all the mixtures except 100EAFS + 3%B + 2%C (detailed in Table 1). The UCS samples were prepared in a steel mould with dimensions of 100 ± 1 mm in diameter and 200 ± 1 mm in height. A cylindrical sample was compacted into eight layers with twenty-five blows per layer by applying the modified compaction energy. After compaction, specimens were demoulded and sealed with impervious polythene films to control the internal moisture content before being placed in a humidity chamber (25 °C) or temperature-regulated oven (40 °C) for curing age of 3 d, as water is pivotal to facilitate pozzolanic reaction and geopolymerisation. The UCS tests were performed using the Universal Testing Machine (UTM-100), with an axial strain rate applied at a controlled rate of 0.5%/min until specimen fractured. A repeated loaded triaxial (RLT) test was implemented to evaluate the resilient modulus (Mr ) responses of the FBS mixtures in accordance with the AASHTO T307-99 standard (AASHTO, 2007). The RLT tests were conducted to simulate dynamic wheel loading on the pavement systems under traffic conditions. For each RLT test, a total of 15 load sequences with a combination of confining (from 20.7 kPa to 137.9 kPa) and axial stresses (from 20.7 kPa to 275.8 kPa) were applied to the FBS samples. The adopted multi-stage RLT tests allowed evaluating the Mr responses of the FBS samples under various stress levels with a large number of load cycles. A haversine-shaped load pulse was applied with a rate of 0.1 s loading time and 0.9 s resting time for 100 cycles at each loading stage. The same sample preparation procedures employed in the UCS tests were followed to prepare the standard-sized cylindrical sample (100 ± 1 mm diameter and 200 ± 1 mm height) that suit the standard triaxial cell for the RLT test. The RLT specimens were cured in a 25 °C humidity-controlled chamber for 3-d before the RLT test. In addition, the RLT test was not performed on 100EAFS + 3%B + 2%C mixture. The resilient modulus under tensile loading was evaluated by conducting an indirect tensile resilient modulus (ITM r ) test as per the AS 2891.13.1 (Australian Standard, 1995). The ITM r samples were compacted in a Marshall mold with 50 blows on each side to produce a standard-sized specimen of 100 ± 2 mm in diameter and 60 ± 2 mm in height. A haversine-shaped load pulse was applied with a constant rate of 40 ms rise time and 3000 ms repetition time using the UTM-100 loading machine until the target recovered horizontal strain of 50 ± 5 µε was attained. The horizontal displacements at the mid-height of the samples were measured by linear variable differential transducers (LVDTs). Each sample was assigned to 15 loading repetitions, which were distributed as 10 pre-conditioning repetitions followed by 5 testing repetitions. In this study, ITM r tests were conducted for all the FBS mixtures (see Table 1), except 100EAFS + 3%B + 2%C mixture. Whereas prepared specimens were cured at: (a) 3-h at ambient temperature (25 °C), (b) 3-d at 25 °C and (c) 3-d at 40 °C (unsoaked condition). Additionally, ITM r tests were also conducted on samples cured for 3-d at 40 °C, followed by vacuum soaking of the samples in a water bath for 10 min to understand the moisture damage to the FBS materials. Equation (1) (Australian Standard, 1995) was used to calculate the ITM r . where, ITM r denotes indirect tensile modulus (MPa), P reflects applied peak load (N), the Poisson's ratio (υ) of 0.4 was adopted for foamed bitumen stabilised materials following recommendation in Austroads specification (Zhalehjoo, 2022), H is average height of the specimen (mm) and hc is horizontal displacement (mm) measured by LVDTs after applied load. The four-point flexural beams were fabricated in rectangular steel mould with dimensions of 400 mm length, 100 mm wide and 100 mm height. The mass of the FBS material required to prepare the beam sample was calculated based on the pre-determined MDD of the FBS mixture. The beam samples were compacted into three layers by using a vibratory hammer until 98% MDD of the FBS mixture was achieved. To prevent moisture loss, compacted FBS beam samples were covered with impervious polythene sheet and kept in a humidity-controlled chamber for 48 h before demoulding. Subsequently, prepared beams were further cured at an ambient temperature in a humidity chamber until the testing age of 28-d. Four-point flexural bending tests were conducted only on the optimum mixture, 50EAFS/50LFS + 3%B + 10%(FA+S), which was determined based on the earlier UCS, RLT and ITM r results. The four-point flexural bending test consisted of three stages of testing to determine modulus of rupture, flexural modulus and fatigue life of the FBS mixture. The modulus of rupture test was conducted as per AS 1012.11 (Australian Standard, 2000). Monotonic loading was applied using a UTM-100 loading facility at the centre of customised loading clamps until the sample reached the peak load and fractured. The load span and support span were placed at 100 mm and 300 mm, respectively, along the beam, as shown in Figure 3. The mid-span vertical deflection during the test was recorded by a LVDT, which was installed through a metal frame and positioned in the centre of the loading surface of the beam specimen. Equation (2) (Australian Standard, 2000) was used to calculate the modulus of rupture. where fcf is modulus of rupture value of the beam (MPa), P max is maximum applied load (kN), L is total span length of the beam (mm) and B and H are the average width and height of the beam specimen (mm), respectively. Four-point flexural bending test. The flexural resilient modulus (FM) test was conducted using the same test setup employed for the modulus of rupture test in the present study. As recommended in an earlier study (Jallu et al., 2020), the flexural resilient modulus test was determined by dominating a stress-controlled approach at a stress level representing 40% of the applied maximum load measured during the modulus of rupture test. A pre-programmed stress value (40% of the peak load determined from the modulus of the rupture test) associated with a haversine-shaped load pulse was applied at a rate of 1000 ms, which was composed of 250 ms loading time and 750 ms rest time for 100 cycles. The LVDTs recorded corresponding values for the maximum and minimum deformation of the beam sample. The difference between the maximum and minimum deformations determined the mid-span vertical deformation under cyclic loading. Equation (3) was used to determine the flexural resilient modulus, assuming that the material was homogeneous and linearly elastic (Yeo, 2008). where S FM is the flexural resilient modulus (MPa), P FM is the applied load in a given load cycle of the FM (kN), δFM is the mid-span deflection in a given load cycle of the FM, L is the total span length of the beam (mm) and B and H are the mean width and height of the specimen (mm), respectively. Flexural fatigue life is described as the total number of load cycles required by the beam sample to reduce the flexural resilient modulus to half of its initial modulus (Gonzalez et al., 2010). The beam sample that underwent the flexural resilient modulus test was subsequently used for the flexural fatigue life test since it was a non-destructive evaluation. As recommended in earlier literature (Maghool, 2019), the test was implemented following a stress-controlled mode and at a stress intensity equivalent to 80% of the peak load determined from the modulus of the rupture test measured earlier. The haversine-shaped load pulse was applied at a rate of 500 ms, consisting of 250 ms loading time and 250 ms rest time for 100,000 cycles until the ultimate failure stage was attained. The mid-span deflection is constantly measured by LVDTs. The microstructural characterisation of the FBS samples was examined via scanning electron microscopy (SEM) (Gemini SUPRA 40VP). Some crushed fragments were collected from the FBS samples after the UCS tests and impregnated with isopropanol for 24 h. To enhance the electrical conductivity of the FBS specimens for SEM, the specimens were coated by a gold sputter (EMITECH K975X) before SEM tests. Results and discussion Bitumen foaming characteristics The moisture content required to achieve optimum foaming properties depends on the bitumen type, typically ranging from 2% to 2.5% for C170 bitumen (Nataatmadja, 2001). Figure 4 illustrates the relationships among the ER, HL, water content and bitumen heating temperature. According to the Queensland Transport and Main Roads guidelines (QTMR, 2017), the minimum requirements for ER and HL in foamed bitumen stabilisation are 10 times and 20 s, respectively. Based on the ER and HL results shown in Figure 4, an optimum moisture content of 2.2% was found to be effective for foaming C170 bitumen at 175 °C that complied with the minimum ER and HL requirements. Table 2 tabulates all the optimum parameters for foaming C170 bitumen. The optimum bitumen content of 3%, along with air and water pressures of 5 bars and 6 bars, respectively, were adopted in the present study, as recommended in the Austroads specification (Zhalehjoo, 2022). These optimum parameters were then used to produce the FBS samples. Effect of water content and bitumen heating temperature on C170 bitumen foaming characteristics. Table 2.Optimum parameters for foaming C170 bitumen.ParametersInputOptimum moisture content (%)2.2Optimum bitumen content (%)3.0Foaming agent (%)0.5Air pressure (bars)5Water pressure (bars)6Bitumen temperature (°C)175 Optimum parameters for foaming C170 bitumen. Compaction characteristics In the present study, the OMC and MDD values determined from modified Proctor compaction tests are relatively close to each other. Hence, for better illustration and comparison, the detailed OMC and MDD values for all the FBS mixtures are summarised in Table 1, while compaction curves of the unbound parent materials are presented in Figure 5. As shown in Figure 5, unbound 100LFS achieved higher OMC and MDD values of 9.1% and 2.65 Mg/m3, respectively, compared to unbound 100EAFS, which had OMC and MDD of 3.1% and 2.45 Mg/m3. The disparity in the OMC and MDD is attributed primarily to the gradations of 100LFS and 100EAFS. The unbound 100LFS contains more fine particles, which require additional water to facilitate particle mobility and achieve optimum compaction. Additionally, the better gradation of 100LFS allows for improved particle packing during compaction, and hence, resulting in a higher OMC and MDD. In contrast, the poorly graded 100EAFS tends to lose moisture during compaction because of a lack of fine and sand-sized particles that raise deficiencies on water absorption. However, the combination of 50EAFS/50LFS yielded OMC and MDD values of 7.9% and 2.57 Mg/m3, indicating an improved gradation after blending EAFS with LFS and thus a higher level of compaction, compared to 100EAFS blend. Compaction curves of unbound 100LFS, 100EAFS and 50EAFS/50LFS. Using unbound materials as the reference, an extended comparison of compaction results shown in Table 1 was conducted, considering the addition of 2% cement (hereafter denoted as 2%C) and 10%(FA+S) geopolymer as secondary binders into unbound mixtures at various stages. The effects of the addition of 2%C on the OMC and MDD of 100LFS, 100EAFS and 50EAFS/50LFS were barely noticeable due to the low dosage of additive content. Furthermore, 2%C proved ineffective as a secondary binder for the stabilisation of 100EAFS, as the compacted FBS 100EAFS sample collapsed after demoulding due to its poor gradation. Similarly, the addition of 10% (FA+S) geopolymer as a secondary binder had a minimal impact on the OMC and MDD of the 100LFS and 50EAFS/50LFS mixtures compared to their unbound counterparts (see Figure 5). Notably, the OMC and MDD of 100LFS and 50EAFS/50LFS with the addition of 10% (FA+S) geopolymer were identical to those of the FBS 100LFS and 50EAFS/50LFS mixtures with 2%C. However, the OMC and MDD values of 100EAFS increased noticeably with the addition of 10%(FA+S) geopolymer compared to its unbound counterpart and the FBS 100EAFS mixture with 2%C. Due to the poor gradation of 100EAFS, which lacks fine and sand-sized particles, the addition of fine particles from 10%(FA+S) geopolymer had a greater impact on the OMC and MDD of the FBS 100EAFS mixture. The increased fine content in the mixture required extra water to achieve optimum compaction and thus increased the OMC. Additionally, the fine particles in the 10%(FA+S) geopolymer filled the voids between the coarser 100EAFS aggregates during compaction, ultimately increasing the MDD of FBS 100EAFS compared to unbound 100EAFS. Unconfined compressive strength (UCS) The UCS results of the FBS samples cured for 3-d at 25 °C and 40 °C are depicted in Figure 6. For the FBS samples with 2%C as a secondary binder, the 50EAFS/50LFS mixture showed slightly higher UCS values than the 100LFS mixture, regardless of the curing temperature. This improvement can be attributed to the improved gradation of the 50EAFS/50LFS mixture compared to the pure 100EAFS and 100LFS. It should be noted that UCS tests were not conducted on the 100EAFS + 3%B + 2%C samples because of the coarse nature of 100EAFS and insufficient particle cohesion, which caused compacted samples to collapse after demoulding. As evident in Figure 1, 100LFS contained finer particles, while 100EAFS had a high portion of coarser aggregates. Blending EAFS and LFS together improved the gradation and particle interlock performance compared to pure 100EAFS and 100LFS mixtures. The finer LFS particles filled the voids between coarser EAFS particles, aiding the compaction of the FBS material into a denser matrix and ultimately improved the UCS of the samples. In addition, the FBS samples cured at 40 °C consistently yielded higher UCS values than those cured at 25 °C, indicating accelerated cement hydration and improved particle bonding at elevated temperature curing. However, using conventional foamed bitumen composition with the addition of 2%C resulted in lower UCS values for all the FBS mixtures with the highest UCS of 0.74 MPa recorded for 50EAFS/50LFS + 3%B + 2%C under a curing regime of 3-d at 40 °C. UCS results of FBS mixtures cured for 3-d at 25 °C and 40 °C. The UCS results of all FBS samples with 10%(FA+S) geopolymer as the secondary binder demonstrated significant strength improvements compared to FBS samples with 2%C (see Figure 6). The use of geopolymer as the secondary binder resulted in an approximate eight-fold increased UCS values for FBS 100LFS samples, rising from 0.53 MPa to 4.05 MPa under a 3-d, 25 °C curing condition. For the 50EAFS/50LFS mixture, the strength improvement of with geopolymer was even more pronounced, the UCS values increased from 0.59 MPa to 16.23 MPa under 3-d, 25 °C curing. Moreover, the addition of 10%(FA+S) geopolymer proved to be an effective secondary binder for stabilising 100EAFS mixture and achieved a UCS value of 3.25 MPa with 3-d, 25 °C curing. The primary role of secondary binders was to promote early strength development and reduce water susceptibility in FBS materials. It was observed that the addition of 10%(FA+S) geopolymer as a secondary binder showed significantly higher early strength compared to the corresponding FBS mixture with the addition of 2%C, regardless of types of parent material being used. This observation can be attributed to the higher mineral filler material content in the FBS mixtures containing 10%(FA+S) geopolymer compared to the conventional foamed bitumen composition with the addition of 2%C as a secondary binder. Specifically, foamed bitumen creates thin bitumen bubbles that act as viscoelastic phase within the FBS mixture (Iwański et al., 2018). During mixing, those bitumen bubbles burst, forming small splinters that selectively bind to finer particles, which are uniformly distributed throughout the mixture (Sabita 2020). The coating of fine FA and S particles subsequently formed a bitumen mastic. The mastic was pressed into larger aggregates under the compaction effect, creating non-continuous bonds that produced a ‘spot-welding’ effect, reinforcing the granular structure (Jenkins, 2000). The inclusion of this foamed bitumen mastic also effectively reduces the brittleness associated with high cementitious contents. In addition, the dissolution of FA and S during alkali activation produced a highly cross-linked binder structure comprising both calcium aluminosilicate hydrate (C–A–S–H) and sodium aluminosilicate hydrate (N–A–S–H) gels (Ismail et al., 2014). The coexistence of these gels promotes the development of a denser and less porous matrix (Humad et al., 2019). Therefore, the FBS mixtures with geopolymer demonstrated higher strength than that of 2%C mixtures. This synergistic benefit of the hybrid binder system comprising 3%B and 10%(FA+S) was further evidenced through comparison with previous studies. Maghool et al. (2019) reported that the 50EAFS/50LFS mixture stabilised solely with 10%(FA+S) geopolymer binder yielded a UCS of approximately 12 MPa UCS under 7-d, 25 °C curing. In contrast, the 50EAFS/50LFS mixture in the present study, stabilised using a binary system of 3%B and 10%(FA+S), achieved a UCS value of 16.23 MPa after only 3-d of curing at 25 °C. This observation indicated that the inclusion of foamed bitumen played a critical role in terms of early strength gain while reducing the curing duration. Similar to the FBS mixtures containing of 2%C, the FBS mixtures incorporating 10%(FA+S) geopolymer cured at 40 °C consistently achieved higher UCS values than those cured at 25 °C. Previous study has noted that the strong temperature sensitivity of (FA+S) geopolymers achieve higher UCS values under elevated temperature curing (Lin et al., 2023). This behaviour could be attributed to the low reactivity of FA at ambient temperature (25 °C) due to its limited content of reactive glassy phases relative to S. Higher curing temperatures promote the dissolution of these glassy phases, facilitating the formation of reaction products that contribute to improved compressive strength (Chithiraputhiran & Neithalath, 2013). Similarly, increasing the curing temperature from 25 °C to 40 °C accelerated the reaction rate of S and the precipitation of gel products, as additional thermal energy enables a more reactant particles to overcome the activation energy required for gel product formation (Chithiraputhiran & Neithalath, 2013). During geopolymerisation, both the dissolution and gelation stages depend on surpassing this energy threshold, and moderate heating allows a larger proportion of particles to fully participate in the reaction (Chithiraputhiran & Neithalath, 2013). The degree of strength improvement varies significantly by parent material type. The 50EAFS/50LFS + 3%B + 10%(FA+S) mixture demonstrated the greatest sensitivity to elevated temperature curing, with the UCS increasing by 23.8% (from 16.23 MPa to 20.1 MPa) when the curing temperature was elevated from 25 °C to 40 °C. Benchmarking this mixture against similar FBS systems incorporating recycled aggregates further highlights its excellent load-bearing performance. For example, Maghool et al. (2022) reported UCS values of approximately 1.5 MPa for a 90RCA/10RG blend and 1.2 MPa for a 90RCA/10RAP blend, both of which were stabilised with 3% foamed bitumen and 2% hydrated lime. Under the same curing regime (3-d at 40 °C), the 50EAFS/50LFS + 3%B + 10%(FA+S) mixture achieved a strength of 20.1 MPa, outperforming these conventional systems by an order of magnitude. The UCS results indicate that FBS mixtures utilising the 10% (FA+S) geopolymer as the secondary binder could be a suitable design for pavement applications requiring higher strength performance and moisture durability, such as airport pavement base/subbase layers. According to the Civil Aviation Administration of China specification, the civil aviation pavement base layer requires a UCS of 3.5–5 MPa, while the subbase layer requires a UCS of 3.0–4.5 MPa (Civil Aviation Administration of China, 2017). Based on these criteria, the 100EAFS + 3%B + 10%(FA+S) mixture complied with the subbase minimum UCS requirement, while the 100LFS + 3%B + 10%(FA+S) and 50EAFS/50LFS + 3%B + 10%(FA+S) mixtures complied with both base and subbase minimum UCS requirements. Indirect tensile resilient modulus (ITM r ) Figure 7 presents the ITM r results of the FBS samples with respect to different secondary binder types and curing conditions. Both FBS 100LFS and 50EAFS/50LFS samples incorporating 2%C as the secondary binder showed significantly lower ITM r values compared to their counterparts containing 10%(FA+S) geopolymer. This trend reflects the distinct roles of the secondary binders. In the 2%C system, the cement served primarily as an active filler, producing only a limited hydrate network during the short curing period, which was insufficient to form a stiff load-bearing matrix. In contrast, the 10%(FA+S) geopolymer rapidly developed a hybrid gel network comprising C-A–S–H and N–A–S–H that densified interfacial transition zones along the aggregate surfaces, hence improving early-age stiffness (Chithiraputhiran & Neithalath, 2013; Fang & Zhang, 2020). ITM r of the FBS mixtures under different curing conditions. Furthermore, the FBS samples containing 10%(FA+S) geopolymer inherently possessed a higher fraction of fine filler than the conventional foamed bitumen composition with 2%C. During mixing and compaction, this increased fines content improved the chances of FA and S particles being uniformly coated by foamed bitumen, promoting more continuous and evenly distributed bitumen mastic across the aggregate matrix. This mastic filled the intergranular voids and provided a moisture-resistant viscoelastic matrix, while the geopolymer gels supplied rigid bridges at the particle contacts. The synergistic effect of a continuous bitumen mastic and a percolated gel network accounted for the consistently higher ITM r measured in the 10% (FA+S) mixtures compared with 2%C mixtures. As shown in Figure 7, FBS samples subjected to 3-h curing showed the lowest ITM r for all the tested FBS mixtures. Notably, extending the curing age to 3-d at 25 °C substantially increased the ITM r of all unsoaked FBS samples. In particular, the ITM r of the 50EAFS/50LFS + 3%B + 10%(FA+S) sample increased from 1619 MPa (3-h curing) to 3059 MPa with an extended curing age of 3-d curing at 25 °C. In addition, elevating the curing temperature to 40 °C further enhanced the ITM r for all the FBS samples, regardless of different parent material blends. The increased stiffness of the FBS samples at higher curing temperature can be attributed to changes in the bitumen properties within the specimens caused by binder aging, as higher curing temperature typically resulted in greater mechanical performance of material stabilised with bitumen-based binders (Wirtgen Cold Recycling Technology 2012). Under 3-d, 40 °C curing regime, the 50EAFS/50LFS + 3%B + 10%(FA+S) sample recorded the highest ITM r of 4135 MPa, corresponding to the dense microstructure presented in Figure 12e. Meanwhile, the ITM r values of the other FBS samples were relatively comparable, ranging from 1824 MPa to 2714 MPa under the same curing regime (i.e. 3-d at 40 °C). Figure 7 also depicts the effects of moisture damage to the FBS samples, as demonstrated by the discrepancy in the ITM r values for the samples tested under unsoaked and soaked conditions with the same curing regime. The ITM r of 3-d soaked FBS samples were lower than those of their corresponding unsoaked counterparts, demonstrating that moisture damage negatively impacted the stiffness of the FBS samples (Muthen, 1998). It should be noted that FBS materials typically exhibit high void ratios and permeability, allowing water to infiltrate from external sources or wet subgrade layers (Muthen, 1998). The soaked performance of FBS materials is critical to be investigated. The variance observed in ITM r of FBS samples between the unsoaked and soaked conditions can be attributed to the humid environments, which negatively affected the particle-to-particle bonding. Specifically, the bond between bitumen and aggregate may weaken or fully rupture when exposed to water (Vinet-Cantot et al., 2019). Strongly adsorbed functional groups in bitumen are easily displaced by water, making water displacement chemically favourable (Vinet-Cantot et al., 2019). The breaking of adhesive bonds between bitumen and aggregate in compacted mixtures, known as stripping, is a form of passive adhesion that can reduce ITM r . According to the Austroads guide (Jameson, 2019), the minimum ITM r requirement for pavement base applications is 4000 MPa for FBS materials cured for 3-d at 40 °C and tested under unsoaked condition. For the FBS materials cured under the same condition but tested after the soaking process, the requirement is 2000 MPa. While the subbase application requires a minimum ITM r of 2500 MPa (unsoaked) and 1500 MPa (soaked). For both the pavement base and subbase layers, a minimum retained modulus ratio of 0.5 is required for roads carrying an average daily traffic exceeding 1000 ESAs (equivalent standards axles), corresponding to an 8 tonnes single axle with dual tyres (Jameson, 2019, Moffatt 2025). The retained modulus ratio was calculated as the ratio of the ITM r values of 3-d soaked samples to those of the corresponding 3-d unsoaked samples. As illustrated in Figures 7 and 8, the 50EAFS/50LFS + 3%B + 10%(FA+S) sample complied the stiffness requirements for both base and subbase applications. Meanwhile, the 100LFS + 3%B + 10%(FA+S) sample complied with the minimum stiffness requirement for subbase application. Retained resilient modulus (ITMretained) FBS mixtures. Repeated load triaxial (RLT) The RLT tests were performed to simulate real-world dynamic traffic conditions on FBS mixtures. The Mr of the FBS layer in a road structure has a significant impact on both the bending deformation and fatigue life of the pavement surface layer, as well as on the distribution of traffic loads across weaker underlying layers (Fu, 2009). As presented in Table 1, excepted for the 100EAFS + 3%B + 2%C mixture, all other FBS mixtures with the addition of either 2%C or 10%(FA+S) geopolymer were prepared for the M r test. The FBS samples prepared for the Mr test were cured for 3-d at 25 °C. The Mr of the FBS mixtures under various test sequence are presented in Figure 9. It was evident that FBS mixtures clearly yielded comparable responses in terms of Mr (see Figure 9). The Mr of all the FBS mixtures increased as the confining pressure increased. Additionally, under the same applied confining pressures of 20.7 kPa (test sequence 1 to 3), 34.5 kPa (test sequence 4 to 6), 68.9 kPa (test sequence 7 to 9), 103.4 kPa (test sequence 10 to 12) and 137.9 kPa (test sequence 13 to 15), the Mr values for all the FBS mixtures increased as the axial stress increased. Since Mr represents the measure of recoverable axial strain under applied axial stress, a higher Mr value indicates a greater capacity of the samples to recover from deformation. Mr of the FBS mixtures under various test sequence. Figure 9 also shows that the FBS mixtures with the addition of 2%C generally had lower Mr values compared to their 10%(FA+S) geopolymer counterparts. This trend is consistent with the lower UCS and ITM r results observed in the same FBS mixtures with the addition of 2%C. For instance, the 50EAFS/50LFS + 3%B + 2%C mixture yielded a Mr range of 142.4 MPa to 480.1 MPa, whilst the 50EAFS/50LFS + 3%B + 10%(FA+S) mixture achieved a higher Mr range of 222.5 MPa to 583.1 MPa, corresponding to an approximate increase between 21% and 56%. The enhanced Mr of the FBS mixture with the addition of 10%(FA+S) geopolymer can be attributed to the higher dosages of mineral fillers, which improved the formation of the bitumen mastic. This enhanced bitumen mastic formation contributes to beneficial properties such as homogenous dispersion of bitumen, improved adhesion of bitumen to the aggregate, and reduced moisture sensitivity (Sabita, 2020). It should be noted that the Mr values of the FBS mixtures measured from the RLT test were significantly lower than those obtained from the ITM r test. Earlier studies have suggested that the ITM r test may overestimate the M r of FBS materials, whereas the RLT test is considered more reliable and better reflects field stress conditions (Fu, 2009). The stress calculations in the ITMr test are based on the continuum mechanics assumption, which may not fully apply to FBS materials (Fu, 2009). In contrast, the RLT test uniformly applies confining and deviatoric stresses, with stress calculations based solely on the principle that internal stresses must balance the applied loads. Additionally, Poisson's ratio, which is required in stress calculations for the ITM r test, is not needed in the RLT test, highlighting the fundamental differences between the two methods. Since the concept of Poisson's ratio assumes a continuous medium, the Mr derived from the ITM r test may be overestimated compared to those obtained from the RLT test. Four-point bending test The modulus of rupture, flexural modulus and fatigue life results of the FBS beam specimen are presented in Table 3. In this study, four-point flexural bending tests were conducted only on the optimum 50EAFS/50LFS + 3%B + 10%(FA+S) mixture, which yielded the highest strength and resilient modulus determined from UCS, ITM r and RLT tests. To better evaluate the performance of the optimum FBS mixture under a four-point flexural bending test, a benchmark mixture from a previous study was used for comparison (Disfani et al., 2014). This benchmark mixture was tested under the same testing setup and conditions and comprised recycled concrete aggregate (RCA) and crushed brick (CB) at a blending ratio of 50:50 as the parent material and was stabilised with 3% cement (by mass of the parent material). RCA and CB are well-known high-quality recycled material suitable for pavement construction (Department of Transport and Planning Technical Note, 2023) and are recognised for their comparable performance as to the virgin aggregates (Hall et al., 2022b). The mixture used for comparison of the four-point bending test results was denoted as 50RCA/50CB + 3%C, as shown in Table 3. The selected mixture 50RCA/50CB + 3%C is accepted in practice for pavement construction in Victoria, with a binder content of 3% cement for stabilisation (Sustainability Victoria, 2013). Therefore, using these practically accepted mixtures as a benchmark allows for a meaningful comparison of the fatigue performance of the FBS mixtures investigated in the current study against locally accepted and utilised cement-stabilised recycled material blends in real-world pavement applications. Table 3.Comparison of four-point bending test results between optimum FBS 50EAFS/50LFS + 3%B + 10%(FA+S) mixture and cement-stabilised RCA and CB aggregates.Specimen50EAFS/50LFS + 3%B + 10%(FA+S)50RCA/50CB + 3%C (Disfani et al., 2014)Width (mm)97.7100.6Height (mm)102.0100.7Moisture content (% of OMC)99.1100.3Density (% of MDD)96.395.4Modulus of rupture test Peak load, P (kN)18.63.7Modulus of rupture, σt (MPa)5.31.1Flexural modulus test Applied load (% of peak load)4040Mean flexural modulus a (MPa)1297.011780.8Tensile stress for modulus a (kPa)2171.0432.9Beam flexural fatigue test Applied load (% of peak)7080Tensile stress for fatigue test b (kPa)3797.9758.2Tensile strain b (microstrain)2751.370.9Initial modulus b (MPa)1381.910700.0Fatigue life (cycles to half initial modulus)3484729401 a Cycles 50 to 100. b Mean of 1st 50 cycles. Comparison of four-point bending test results between optimum FBS 50EAFS/50LFS + 3%B + 10%(FA+S) mixture and cement-stabilised RCA and CB aggregates. Cycles 50 to 100. Mean of 1st 50 cycles. The 50EAFS/50LFS + 3%B + 10%(FA+S) mixture in this study shows the modulus of rupture value of 5.3 MPa, corresponding to the 18.6 kN peak load. Meanwhile, the comparison mixture 50RCA/50CB + 3%C had a lower modulus of rupture of 1.1 MPa, corresponding to the 3.7 kN peak load. The higher modulus of rupture observed in the 50EAFS/50LFS + 3%B + 10%(FA+S) mixture can be attributed to the improved bonding between the coarser aggregate, facilitated by the incorporation of foamed bitumen and the 10%(FA+S) geopolymer. Additionally, the lower secondary binder content in 50RCA/50CB + 3%C may have contributed to its lower strength. The mean flexural modulus was computed using data from the 51st to 100th loading cycles to minimise potential errors associated with the initial 1st to 50thles, as recommended in earlier study (Maghool, 2019). The 50EAFS/50LFS + 3%B + 10%(FA+S) mixture showed a lower flexural modulus value of 1297 MPa, compared to 50RCA/50CB + 3%C mixture, which had flexural modulus value of 11,780.8 MPa. This observation indicated greater deformation in the FBS beam specimen during the flexural modulus test (Arulrajah et al., 2015) compared to the benchmark 50RCA/50CB + 3%C mixture. The reason can be attributed to the incorporation of foamed bitumen enhanced the flexibility and ductility of the FBS material compared to pozzolanic binder-stabilised RCA and CB aggregates. In addition, the presence of crystallised calcium in the RCA likely contributes to the improved flexural modulus in 50RCA/50CB + 3%C mixture (Mohammadinia et al., 2019). Figure 10a and b show the initial testing setup for the fatigue life test and failure stage of the FBS beam specimen after reaching the total number of loading cycles corresponding to half of the initial modulus of the FBS beam specimen. The applied load for the 50EAFS/50LFS + 3%B + 10%(FA+S) mixture was set at 70% of its modulus of rupture peak load, whereas the comparison mixture, 50RCA/50CB + 3%C, was tested under an applied load equal to 80% of the peak load as reported. The selection of a 70% load rate in the present study was governed by the capacity and safety limitations of the testing facility. In the comparison study, the mixture had a lower peak load, hence, an applied load of 80% of its peak load is feasible for fatigue testing. Despite this difference, the 70% load applied in the present study remains appropriate for fatigue assessment and falls within the loading range commonly used as per Austroads specification (Zhalehjoo, 2022). (a) Testing setup for flexural modulus and fatigue life tests and (b) Failure stage of FBS beam specimen after fatigue life test. Figure 11 illustrates the flexural fatigue life behaviour of the 50EAFS/50LFS + 3%B + 10%(FA+S) mixture. At the earlier stage of testing, the average resilient modulus value was recorded as 1381.9 MPa over the first 50 cycles. With continued loading at 70% of the peak load, the resilient modulus remained stable until it reached half of its initial modulus at 34847 cycles, except for certain loading cycles where fluctuation of resilient modulus was observed. As shown in Figure 11, at around 30000 cycles, the resilient modulus began to fluctuate more steeply, ranging between 1450 MPa and 1550 MPa for over 3750 cycles. This behaviour likely indicates the onset of crack propagation within the beam specimen, leading to a reduction in the resilience modulus (Arulrajah et al., 2015). Additionally, minor fluctuation at about 10000 cycles could be attributed to noise in the testing environment. As shown in Table 3, the optimum 50EAFS/50LFS + 3%B + 10%(FA+S) mixture identified in this study showed a higher fatigue life (34847 cycles) compared to the benchmark 50RCA/50CB + 3%C mixture (29401 cycles), indicating the acceptable durability and longevity of the proposed optimum FBS mixture. Fatigue life test result of optimum FBS 50EAFS/50LFS + 3%B + 10%(FA+S) mixture. SEM analysis Figure 12a and b illustrate the morphologies of the geopolymer precursors. FA appears predominantly spherical, whereas S exhibits angular particles with sharp edges. Figure 12c and d show the microstructure of the unbound EAFS and LFS particles, both of which are irregularly shaped, with the EAFS appearing coarser than the LFS. The unbound materials appear loosely structured and lack particle packing. In contrast, the 50EAFS/50LFS + 3%B + 10%(FA+S) sample shows a substantially denser microstructure (see Figure 12e). Two complementary binder phases are observed including, a nearly continuous foamed bitumen mastic forming smooth films that coat fines and bridge adjacent grains and rigid alkali-activated gels with a rougher appearance that develop a gel-enriched interfacial transition zone around EAFS/LFS granular particles and form intergranular bridges at particle contacts (Jain & Adhikary, 2025). These phases fill intergranular voids, reduce connected porosity, and generate a more continuous load transfer path, with only microcracks observed. SEM images of: (a) FA, (b) S, (c) unbound EAFS, (d) unbound LFS and (e) 50EAFS/50LFS + 3%B + 10%(FA+S) mixture (cured for 3-d at 40 °C). Figure 12e also shows that after 3-d, 40 °C curing, most of the geopolymer precursors (i.e. FA and S) dissolved in the alkaline environment, forming strong geopolymer gels that bonded adjacent particles and filled voids. As reported in previous studies, the binary precursor (FA+S) was highly sensitive to the curing temperature; increasing the curing temperature enhanced the reactivity of FA and S, thus promoting the formation of geopolymer gels (Lin et al., 2023). In addition, bitumen binders with a relatively smooth and glossy surface were also captured in Figure 12e, in contrast to geopolymer gel products which had a coarser and duller surface. During compaction, the bitumen particles in the mastic were physically pressed against larger aggregates, forming localised non-continuous bonds that acted as ‘spot welding’ to link coarser aggregates together (Jenkins, 2000). The coexistence of bitumen films and geopolymer gels indicates interpenetrating binding mechanisms; the viscoelastic bitumen mastic enhances coating uniformity and moisture resistance, while the geopolymer gel network stiffens contact zones and locks fines within the aggregate matrix. From an aggregate perspective, EAFS and LFS exhibit rough surfaces and a uniformly angular particle structure. These characteristics, combined with their relatively high specific surface area, increase the interaction with foamed bitumen and consequently enhance adhesion and bitumen bonding (Chen et al., 2022). Conclusion This study investigated the viability of using steel slag aggregates as a full replacement of conventional virgin quarry material in foamed bitumen stabilisation for pavements. FA and S-based geopolymer was trailed as an alternative secondary binder to reduce the carbon footprint of the final product. The FBS 100EAFS, 100LFS and 50EAFS/50LFS mixtures were systemically evaluated for their mechanical properties and flexural fatigue performance. Based on the experimental results, the following conclusions can be drawn: The FBS 100EAFS, 100LFS and 50EAFS/50LFS mixtures incorporating 10%(FA+S) geopolymer as a secondary binder yielded significantly higher UCS values compared to benchmark mixtures containing 2%C. In particular, the 50EAFS/50LFS + 3%B + 10%(FA+S) mixture achieved the highest UCS value of 16.23 MPa and 20.1 MPa after 3-d curing at 25 °C and 40 °C, respectively. The FBS 100EAFS, 100LFS and 50EAFS/50LFS mixtures incorporating 10%(FA+S) geopolymer showed strong potential for applications requiring higher strength, such as airport pavement base and subbase layers. According to the Civil Aviation Administration of China specification, the 100EAFS + 3%B + 10%(FA+S) mixture met the UCS requirement for the subbase layer, while the 100LFS + 3%B + 10%(FA+S) and 50EAFS/50LFS + 3%B + 10%(FA+S) mixtures complied with both base and subbase minimum UCS requirements. The ITM r values of the FBS 100EAFS, 100LFS and 50EAFS/50LFS mixtures incorporating 10%(FA+S) were consistently outperformed those of the benchmark mixtures containing 2%C. The ITM r value of the 50EAFS/50LFS + 3%B + 10%(FA+S) mixture successfully complied with the stiffness requirements for both base and subbase applications, while the 100LFS + 3%B + 10%(FA+S) mixture complied with the minimum stiffness requirement for subbase application only. However, water exposure negatively impacted the ITM r values of all the FBS mixtures tested after the soaking process, indicating a loss of adhesion between the aggregates and bitumen due to moisture damage. Additionally, the curing duration and temperature played a pivotal role in strength development, with higher UCS and ITM r values achieved under extended curing duration and elevated curing temperature. The resilient modulus values measured from RLT tests followed a similar trend to the UCS and ITM r results. The FBS mixtures incorporating 10%(FA+S) geopolymer yielded higher Mr values compared to those of benchmark mixtures containing 2%C. The modulus of rupture of the selected optimum 50EAFS/50LFS + 3%B + 10%(FA+S) mixture outperformed that of the conventional cement-treated granular material mixture reported in earlier study under the same curing and testing conditions. While the flexural modulus was lower, which was attributed to the enhanced flexibility and ductility of the FBS material compared to pozzolanic binder-stabilised granular materials. The flexural fatigue life of the 50EAFS/50LFS + 3%B + 10%(FA+S) mixture was 34847, demonstrating an acceptable fatigue resistance. The FBS 100LFS and 50EAFS/50LFS mixtures incorporating geopolymer as a secondary binder successfully met road authority requirements for pavement base and subbase application. The findings of this study have significant implications for developing future sustainable roads with enhanced mechanical properties and durability. Although the inclusion of 10%(FA+S) geopolymer as a secondary binder in foamed bitumen stabilisation has demonstrated acceptable mechanical performance, several practical limitations remain regarding its production and field application. The use of 10%(FA+S) geopolymer binder involves additional steps to measure accurately and incorporate alkaline activator solutions for mixing compared with the use of conventional cement or hydrated lime as secondary binder in FBS materials. Appropriate training and safety measures are essential for handling alkaline activators on site, whether within an in-plant or mobile mixing facility. Nevertheless, the adoption of such low-carbon binders for pavement construction aligns with Victoria's net-zero emissions targets, highlighting the importance of addressing these operational constraints. To address these challenges, future field trials are recommended to identify the most suitable mixing methods, including the use of a stabiliser machine or a pugmill. In addition, segmented construction and robust quality assurance should be implemented to ensure the uniformity and consistent performance of the foamed bitumen-stabilised layers, thereby enabling reliable and sustainable pavement construction. Additionally, future studies on the effects of different geopolymer precursors and their dosages are needed to broaden the use of geopolymers as secondary binders in foamed bitumen stabilisation. Acknowledgements This research was supported by the Australia Research Council Linkage Project funding scheme (project number LP220200548). The authors gratefully acknowledge the Repurpose IT Pty Ltd. (Australia) and Downer Group (Australia) for providing the bitumen products used in this research. The authors also would like to acknowledge the Independent Cement and Lime Pty Ltd. (Australia) for providing fly ash and slag used in this research. Funding Statement This research was supported by the Australian Research Council (Grant no. LP220200548). Disclosure statement No potential conflict of interest was reported by the author(s). Data availability statement Data will be made available on request. References Associated Data Data Availability Statement Data will be made available on request.