Production of lightweight expanded aggregates from smectite clay, palygorskite-rich sediment and phosphate sludge

Abstract

Lightweight expanded clay aggregates (LWAs) are porous materials with low density and high strength (EN-13055-1), and they are important in sustainable construction through their lightweight nature and ability to provide thermal or acoustic insulation. The objective of this work was therefore to evaluate the preparation of LWAs using a smectite clay (M1 formulation), whose application in common ceramic production is difficult. An alternative approach was proposed for the valorization of phosphate sludge and a palygorskite-rich sediment by mixing them with expanded clay (M2 formulation) for LWA production. This could result in economically cost-effective products with significant environmental benefits. Pellets were prepared and fired at various temperatures (1100°C, 1125°C and 1150°C), and relevant properties such as bloating index, density, water absorption and compressive strength were determined. Additionally, the microstructure, mineralogical transformations and phase compositions under various sintering temperatures were investigated. Increasing the temperature from 1000°C to 1150°C significantly improved the expansion properties of LWAs, and 1150°C seemed to be the optimal firing temperature at which the best expansion properties were achieved. In addition, the incorporation of the selected waste improved the properties of the final products, leading to lower density, greater strength and greater bloating with the development of the internal pore structure as compared to the LWAs without this addition. Because of their low density (0.6 g cm^–3) and sufficient compressive strength (0.86 MPa), the manufactured LWAs can be used in construction (as insulating panels or in lightweight concrete) and in green roofs.

Lightweight expanded clay aggregates (LWAs) are granular materials with well-developed pore structures and bulk densities of <1.2 g cm^–3 (González-Corrochano et al., 2011; Soltan et al., 2016). In general, the quality of these obtained products depends on the chemical and mineralogical characteristics of the raw materials and on the firing process. Usually, LWAs are obtained at temperatures of 900–1250°C and soaking times of 3–30 min depending on the kiln type (muffle furnaces or rotary kilns; González-Corrochano et al., 2011; Loutou et al., 2013; Volland & Brötz, 2015, Liu et al., 2017; Moreno-Maroto et al., 2017; Ayati et al., 2018; Sun et al., 2021, Li et al., 2021). Recent laboratory-scale studies report the use of microwave radiation for the production of LWAs (Franus et al., 2019). Clays are the most common raw materials used for LWA production, especially plastic clays such as smectite and vermiculite due to their plasticity (to facilitate pellet shaping) and chemical and mineralogical composition (the presence of decomposable components that generate gases upon firing, such as carbonates, organic matter and iron oxide; Ayati et al., 2018). In Tunisia, few studies have been conducted on the preparation and characterization of raw clays from the south of the country in LWA production (e.g. Fakhfakh et al., 2007). The results showed that some clays are suitable for producing low-density products with or without quartz addition depending on the sintering temperature.

Palygorskite is a naturally occurring fibrous clay mineral of sedimentary origin and is member of the sepiolite–palygorskite group of minerals. It is generally neoformed in soils and Palaeosols under dry or semi-dry climates (Da Silva et al., 2018). Chemically, palygorskite is a fibrous hydrated magnesium–aluminium silicate with a micro-channel structure and with a typical chemical formula of Si₈O₂₀(Mg,Al,Fe)₅(OH)₂(OH₂)₄.4H₂O (Zhao et al., 2006; Suárez & García-Romero, 2011). Palygorskite has a wide range of industrial applications, particularly for fibreglass, nanocatalysts, plasters and rubbers and in agriculture (Murray, 2000; Alvarez et al., 2011). Due to its sorptive, rheological and catalytic properties, palygorskite has been used as a supporting material for the preparation of a wide variety of nanocomposites, membranes and coatings (Zhang et al., 2016; Wang & Wang, 2016; Ezzatahmadi et al., 2019; Lakbita et al., 2019; Wei et al., 2019; Zhang et al., 2019; Chen et al., 2021, Fan et al., 2021). Recently, palygorskite has also been used in brick production (Loutou et al., 2019; Zhang et al., 2019; Fan et al., 2021; Wang et al., 2022). There are a few studies (Frayyeh et al., 2014; Moreno-Maroto et al., 2020) reporting the use of palygorskite in LWA manufacturing. Frayyeh et al. (2014) tested the production of LWAs based on an Iraqi palygorskite clay by firing it at 1100°C for 30 min, and this yielded materials with relatively low density (808 kg m^–3). In addition, Moreno-Maroto et al. (2020) suggested that a clay material having smectite and palygorskite as major components (Portuguese clay) has suitable bloating ability when sintered at 1210°C for 4 min. The obtained pellets showed good technological properties, namely low density and high compressive strength (CS).

LWAs might be successfully manufactured by using various industrial by-products and waste products as substitutes for natural raw materials. Examples include the use of automotive plastics (Liu et al., 2017), wastewater sludge (Mañosa et al., 2021), bauxite tailings (Yang et al., 2022), waste glass (Graziano et al., 2022; Liu et al., 2022), granite sawing waste (Soltan et al., 2016), petrochemical waste (Juan Daniel Martinez et al., 2018) and phosphate by-products or waste products (Loutou et al., 2013; Bayoussef et al., 2021a; Chao Ding et al., 2022). Phosphate sludge (PhS) in particular is a by-product of phosphate beneficiation, the production of which will increase in the coming years. The south of Tunisia is one of the regions where PhS has been stockpiled in huge quantities, such that new recycling solutions are critically required to reduce the economic and environmental impacts of this. Some studies report the reuse of PhS in geopolymers and in ceramic formulations (e.g. bricks; Dabbebi et al., 2018; Loutou et al., 2019; Moukannaa et al., 2019; Mabroum et al., 2020, Bayoussef et al., 2021b; Ettoumi et al., 2021). However, few studies have focused on the reuse of PhS for the production of LWAs (Loutou et al., 2013). These authors confirm that the incorporation of PhS is beneficial as it can act as a pore-forming or gas-releasing agent after the decomposition of some inherent compounds.

This paper investigates the production of low-density and porous LWAs using smectite clay (SmC) as the main raw material and palygorskite-rich sediment (PalS) and PhS additives, and the firing process was studied in an attempt to achieve maximum bloating and to develop a porous structure. The produced LWAs were characterized to inspect their mineralogical, physico-mechanical and microstructural properties, and the relevant formation mechanisms of the internal structure in the prepared LWAs were examined. The choice of using SmCs is due to their expandable characteristics during firing, and the addition of PhS and PalS could help to improve some properties of the LWAs, such as their porosity, density, water absorption and CS. In addition, the recycling of such waste products/by-products would have an obvious environmental impact, contributing to the circular economy.

Materials and methods

Materials

The raw materials used in this work were a SmC, a PalS and a PhS. The M1 formulation was formulated only with SmC and the M2 formulation was prepared by mixing 50% SmC, 25% PalS and 25% PhS. This latter formulation was chosen after several laboratory tests in which different mixtures with different proportions of PhS and PalS additions were used (Table 1).

Table 1. Composition of the tested formulations (wt.%).

The clay was collected on the El Hamma region in Gabes governorate, south-east Tunisia. It belongs to the Upper Cretaceous sedimentary clayey series (Coniacian–Santonian), exposed for >35 km in an approximately north-east to south-west direction along the Northern Chotts chain in southern Tunisia, which has significant reserves of smectitic clays (Boussen et al., 2015). These clays were used on phosphoric acid purification (Trabelsi & Tlili, 2017). The PalS is from part of the vast alluvial deposits also occurring in the Gabes region of south-east Tunisia. Despite these large reserves, its industrial use has never been developed to date due to its heterogeneous mineralogical and chemical composition and abundant impurities. The use of this raw material for LWAs might constitute an innovative approach. The PhS was collected from the Metlaoui phosphate plant located in Gafsa mining district, south Tunisia. During beneficiation, fluorapatite is separated from associated gangue minerals by crushing, screening, washing and flotation. This treatment generates large volumes of waste, mostly PhS, which is stockpiled in tailings ponds within the processing plant.

Characterization techniques

All of the characterization techniques were carried out in the laboratories of the Materials and Ceramics Engineering Department at the University of Aveiro in Portugal and in the Research Laboratory of Geo-systems, Geo-resources and Geo-environments in the Faculty of Sciences of Gabes, University of Gabes in Tunisia.

X-ray fluorescence (XRF) spectroscopy (Philips X'Pert Pro MPD) was used to determine the chemical composition of the raw materials and the loss on ignition (LOI), which corresponds to the weight loss of a sample after heating to 1000°C.

X-ray diffraction (XRD) analysis was used to determine the mineralogical composition of the raw materials. A Panalytical X'Pert Pro³ XRD device was used, with Cu-Kα radiation at 40 A and 45 kV. The mineralogical composition of the bulk materials was determined using powder XRD. The XRD traces were recorded between 3 and 70°2θ at a step size of 0.017°2θ. The mineralogical composition of the clay fraction was determined with oriented slides. Carbonates and organic matter were eliminated prior to the separation of the clay fraction by centrifugation. Suspensions were placed on glass slides, and the standard tests were conducted after air-drying, saturation with ethylene glycol and heating at 550°C. The XRD traces of the clay fractions were obtained from 2 to 30°2θ. Mineral-phase identification was carried out using the X'Pert HighScore Plus software.

The particle-size distribution was analysed by laser diffraction in the range 0.04–2000 mm using Coulter LS230 equipment. Plasticity was evaluated through determination of the Atterberg limit of the <400 μm fraction. The liquid limit (LL) was determined using the Casagrande method, whereas the plastic limit (PL) was estimated by repeated rolling of an ellipsoidal-sized sample mass by hand on a flat surface. The plasticity index (PI) was determined form the difference between the LL and PL values.

Simultaneous differential scanning calorimetry–thermogravimetric analysis (DSC-TGA) was used to investigate the thermal behaviour of the raw materials. This technique allows for measuring of both heat flow and weight changes in a material as a function of temperature in a controlled atmosphere. The obtained results allow for differentiation between endothermic events that involve a weight loss and exothermic events without an associated weight loss. The DSC-TGA analyses were performed using a HITACHI STA300 coupled device, reaching a maximum temperature of 1200°C, in an inert atmosphere at a heating rate of 5°C min^–1 and using α-Al₂O₃ as an inert material.

Preparation of pellets and the firing process

The production of LWAs involved three main steps, namely the preparation of powder and mixture formulations, pelletization and firing (Fig. 1). The first step involved drying the constituents at 105°C for 24 h to ensure removal of free water. Then, the materials were crushed in a jaw crusher and sieved through a 200 μm sieve. The preparation of the raw material was carried out according to Moreno-Maroto et al. (2020). After weighing, powders were moistened and hand mixed to obtain a homogeneous sample that could be shaped into small pellets with diameters of between 12 and 15 mm. The pellets were dried at room temperature for 72 h, then at 115°C for 48 h until a constant weight was achieved. The firing process, conducted in a laboratory-scale muffle furnace (Fig. 1), was critically studied to determine the optimum heating rate, maximum temperature and soaking time. After several trials, the following conditions were defined: (1) firing temperatures of 1100°C, 1125°C and 1150°C; (2) heating rate of 5°C min^–1; and (3) soaking time of 15 min followed by fast cooling.

Figure 1. The protocol followed for the manufacture of the LWAs.

Characterization of LWA properties

The crystalline phases formed upon firing were identified by XRD under the same conditions as previously described. The microstructure was examined by scanning electron microscopy (SEM; Hitachi S4100 and Quanta 250 equipped with an energy-dispersive spectroscopy (EDS) device). The shell part of the LWAs and their interior structures were observed and compared. The EDS analysis was conducted to analyse the elemental composition and distribution (EDS mapping) of the LWA matrix.

The bloating index (BI) expresses the volume change after sintering of the LWAs. It was determined according to González-Corrochano et al. (2011) using Equation 1:

(1)$${\rm BI\;}( {\rm \% } ) = \displaystyle{{D2-D1} \over {D1}} \times 100$$

where D1 and D2 are the diameters of the pellets before and after sintering, respectively.

The bulk density was determined using the sand dislocation method (De Jesus Fernandes Pinto, 2005). Each test was repeated and the average value was determined.

The water absorption of the pellets after 24 h of immersion in water was determined using the standard EN-1097-6 (2013) according to Equation 2:

(2)$${\rm WA\;}( {\rm \% } ) = \displaystyle{{M2-M1} \over {M1}} \times 100$$

where M1 is the mass (g) of the saturated pellets after immersion in water and M2 is the mass (g) of the dry pellets before immersion in water.

The CS of sintered LWAs was evaluated based on the load of rupture measured by pressing with a maximum pressure of 5 kN and a loading rate of 0.5 mm min^–1 using a sphere uniaxial compressive machine. The CS was calculated using the determined rupture load values according to Equation 3 (Moreno-Maroto et al., 2020):

(3)$${\rm \sigma c\;}( {{\rm MPa}} ) = \displaystyle{{2.8{\rm Fc}} \over {{\rm \pi }D^2}}$$

where Fc is the rupture load (N) and D is the diameter of the sintered pellets (mm).

Results and discussion

Raw materials characterization

Chemical composition

The chemical compositions of the used raw materials (SmC, PalS and PhS; Table 2) show that the SmC is characterized by high contents of SiO₂ (51.55%), Al₂O₃ (13.28%) and Fe₂O₃ (6.67%). The CaO, MgO, K₂O, Na₂O and SO₃ amounts are relatively low and may be related to the presence of minor quantities of calcite, dolomite, feldspars and gypsum. This composition is comparable to that of some other clays used for the production of LWAs (Anan & Abd El-Wahed, 2017; Liu et al., 2017; Moreno-Maroto et al., 2020; Abdelfattah et al., 2022). The main oxides in the PalS sample were SiO₂ (34.11%), Al₂O₃ (8.94%) and CaO (22.04%), with Fe₂O₃ (4.03%), MgO (3.39%) and SO₃ (8.43%) also being abundant. The PhS is mainly composed of CaO (31.29%), SiO₂ (21.88%) and P₂O₅ (16.72%), and these are related to the raw phosphate rock composition, which is rich in carbonate–fluorapatite (Ettoumi et al., 2020). Other oxides, including Al₂O₃ (3.8%), SO₃ (3.73%), MgO (2.01%) and F (1.34%), are present in minor amounts. All studied materials show high LOI, namely 16.80%, 15.84% and 15.62% for SmC, PalS and PhS, respectively. The LOI is mainly due to the dehydroxylation of clay minerals, organic matter oxidation and decomposition of carbonates and hydroxides with the release of water vapour and various gases during firing (Abdelmalek et al., 2017).

Table 2. Chemical composition (wt.%) of the studied materials and other clays mentioned in the literature that were used in the LWA production.

Mineralogical composition

Figure 2a presents the XRD traces of the SmC, PalS and PhS bulk samples and Fig. 2b displays the XRD traces of the oriented clay samples. The semi-quantitative estimation of the main crystalline phases (Table 3) shows that the main mineral phases identified in the SmC sample are smectite (montmorillonite) as a major clay mineral and quartz as a major associate mineral with minor calcite and gypsum. Similar XRD data were reported by Boussen et al. (2015), confirming the presence of the (001) diffraction plane (at 14.5–15.0 Å) of montmorillonite.

Figure 2. (a) XRD traces of the studied materials (bulk samples) of the SmC (green trace), PalS (blue trace) and PhS (red trace) samples. (b) XRD traces of the oriented aggregates of the SmC and PalS samples. Cal= calcite; Dol = dolomite; Fap = fluoroapatite; Gp = gypsum; Ilt = illite; Kln = kaolinite; Plg = palygorskite; Qz = quartz; Sme = smectite.

Table 3. Mineralogical composition of the starting raw materials.

+ = 1–5%; ++ = 6–25%; +++ = 26–50%; ++++ = 51–75%; +++++ = 76–100%.

Cal = calcite; Dol = dolomite; Gp = gypsum; Fap = fluoroapatite; Plg = palygorskite; Qz = quartz; Sme = smectite.

The PalS consists mainly of palygorskite in addition to quartz, calcite, dolomite and gypsum as associated minerals. Palygorskite is often present in some Tunisian clay-rich sediments as a fibrous clay mineral characterized by its (001) diffraction peak at 10.5 Å (Tlili et al., 2010). Similar results have been reported in the recent work of Saadaoui & Eloussaief (2017) and Allouche et al. (2020).

The main mineral phases of PhS are fluorapatite, quartz, calcite, dolomite, gypsum and smectite, which are in accordance with the results obtained by Ettoumi et al. (2020). The presence of these phases is associated to the fine and undesirable fractions being eliminated after the washing of the phosphates.

The XRD traces of the clay fractions (Fig. 2b) demonstrate that the dominant mineral is smectite in the SmC sample and palygorskite in the PalS sample, with minor amounts of illite (9.99 Å) and kaolinite (7.16 Å). Therefore, the reflection intensity of 14.48 Å in the air-drying test is associated with the presence of a 2:1-type mineral. This peak shifted to 17.8 Å after saturation with ethylene glycol and to 10.0 Å after heating at 550°C, both of which being typical of smectite. The peak at 10.48 Å in the air-drying preparations represents the first-order peak of palygorskite, and the peak at 6.37 Å is associated with the second-order peak of this mineral. The peaks were not affected by ethylene glycol treatment, confirming the presence of palygorskite.

Particle-size distribution and plasticity

The particle-size distributions of the raw materials (Fig. 3) and the related uniformity coefficient (Table 4) reveal that SmC and PalS exhibit narrow distributions with particle sizes of between 0.1 and 50 μm. The PhS shows a wider size distribution, with particle sizes reaching 110 μm. The clay fraction content (<2 μm) was ~40% in SmC, 8% in PalS and ~10% in PhS. These results agree with the nature and mineralogy of the materials. The silt fractions (2–63 μm) were 60%, 92% and 42% in SmC, PalS, and PhS, respectively. Both PalS and SmC lack a sand fraction (ø > 63 μm), but PhS contains a sand fraction of ~48%.

Figure 3. Thermal analyses (DSC-TGA) of the raw materials: (a) SmC, (b) PalS and (c) PhS.

Table 4. Particle-size distribution (%) and Atterberg limits (%) of the raw materials.

The particle size plays an important role in plasticity, as this property is associated with the clay fraction of the material (Moreno-Maroto et al., 2020). The results regarding the Atterberg limits (Table 3) show that the studied materials have moderate to high LLs: 64% for SmC and 52% for PalS, with PIs of 32% and 39%, respectively. These moderate to high values are usually attributed to the presence of montmorillonite in SmC and palygorskite in PalS (Daoudi et al., 2015; Abbaslou et al., 2016).

Thermal analysis

Figure 4 shows the thermal analysis curves (DSC-TGA) of the studied materials. The DSC-TGA curve of SmC indicates the presence of three endothermic and one exothermic events, with a total weight loss of ~18.5% up to 1150°C (Fig. 4a). This value is in accordance with the LOI determination. The first endothermic event occurs at between 50°C and 250°C and involves an significant weight loss of ~14%, mainly due to the removal of adsorbed and interlayer water, which is in agreement with Alabarse et al. (2011) and Arab et al. (2015).

Figure 4. Particle-size distribution of the raw materials.

The second weight loss event was observed in the 400–600°C range and is mainly due to the dehydroxylation of the clay minerals, with an endothermic peak at >500°C and a weight loss of ~4%. The obtained results are in accordance with Fontaine et al. (2020). This confirms that the studied clay (SmC sample) is a Fe-smectite, as shown by the chemical analysis. According to Arab et al. (2015), dehydroxylation occurs at between 400°C and 700°C, with Fe-rich smectites showing an endothermic peak at ~500°C, whereas Fe-poor smectites show the same peak at ~700°C. The third step occurred at ~850°C with a small weight loss of <1% due to carbonate decomposition. The single exothermic event occurs near 1000°C without weight loss and is related to recrystallization (Mahmoudi et al., 2017; Chalouati et al., 2020).

The DSC-TGA curves of the PalS sample (Fig. 4b) show four events, similar to those observed by De Souza et al. (2021) and Wang et al. (2021). The first event involves a strong endothermic peak, centred at ~120°C with a weight loss of ~10% due to the removal of adsorbed water and zeolitic water from the palygorskite channels. The second event is observed at between 400°C and 600°C with an endothermic peak at ~ 500°C and a weight loss of 3.5% due to the removal of zeolitic water. The third event occurs at between 700°C and 900°C with an endothermic peak at 750°C and a relative weight loss of 8.5%. This is mainly due to carbonate decomposition (Karunadasa et al., 2019) and dehydroxylation of palygorskite, which occurs at between 700°C and 900°C (Frost & Ding, 2003; Wang et al., 2022), and both carbonate and palygorskite were confirmed to be present in the PalS sample in the XRD analysis. This endothermic peak is immediately followed by a broad exothermic peak coinciding with the structural collapse of palygorskite as well as possible phase recrystallization.

The DSC-TGA curves of PhS exhibit four successive endothermic events due to the presence of clay minerals, carbonates and phosphates in the sample (Fig. 4c). The first event extends from room temperature to 200°C with a weight loss of 3.5%, which corresponds to the removal of adsorbed water. The second event, due to dehydration, occurs at between 200°C and 500°C with a weight loss of 3%. The third event is observed at between 400°C and 650°C and is assigned to the dehydroxylation of clay minerals. The last loss occurs at between 650°C and 900°C, with a large endothermic peak centred near 750°C and a total weight loss of 9% due to the decomposition of calcite. Finally, an exothermic event without mass loss is observed at ~1000°C and is assigned to Ca–Mg-silicate formation (Harech et al., 2021, 2022).

Characterization of the LWAs

XRD analysis

The XRD traces of the LWAs sintered at 1100°C, 1125°C and 1150°C show significant phase changes (Fig. 5). As expected, the firing caused changes in the initial crystalline phases. The clay minerals (smectite and palygorskite) underwent transformations, whereas the carbonates, mainly calcite and dolomite, decomposed, which is in agreement with the thermal analyses. Quartz, which was initially present in the raw clays, was identified in the M1 and M2 pellets. The peak intensity of quartz decreases with increasing temperature, which is due to its partial melting. Fluorapatite remained present, and it was identified only in the M2 XRD trace, in which phosphate was added. New crystalline phases, namely anorthite, akermanite, diopside, cristobalite, hematite and magnetite, were formed in both sintered samples. Anorthite formation results from the reaction between free CaO released upon decarbonation with aluminosilicates from the dehydroxylation of clay minerals (Liao & Huang, 2011; Loutou, 2013). Hence, this newly crystalline phase was more abundant in the M2 sample, which is richer in calcium carbonate than the M1 sample. The intensity of the anorthite peaks was strongest in the M2 sample fired at 1100°C and 1125°C, but the peak intensities decreased slightly at 1150°C, probably due to partial melting. Akermanite and diopside were identified in both samples after firing at 1150°C. They resulted from the reaction between Ca, Mg and the Al and Si released from the decomposition of clayey materials upon firing. The peak intensities were slightly larger in the M2 LWA than that in the M1 LWA. This can be explained by the contribution of magnesium oxide (MgO) resulting from the decomposition of dolomite present in the additives. Cristobalite peak intensities (4.05 and 2.51 Å) are observed in all of the pellets and are more pronounced in the M1 pellet fired at 1150°C. The peak intensities of newly formed cristobalite slightly increased with increasing temperature from 1100°C to 1150°C, whereas the quartz reflection intensities decreased. Such mineralogical changes suggest that some cristobalite might have been derived from the transformation of quartz with increasing firing temperatures. Indeed, quartz first begins to melt at 1100°C, and part of the molten quartz is transformed into cristobalite. These results are consistent with previous studies (e.g. Aras & Kristaly, 2019), which reported that cristobalite is mainly formed from SiO₂ released by the decomposition of clay minerals at ~1100°C and from the transformation of quartz with increasing firing temperatures up to 1200°C, and this is demonstrated by the gradual decrease in quartz peak intensities and the gradual increase in cristobalite peak intensities. Hematite was recorded in all of the XRD traces from 1100°C, and its reflection intensity increased with increasing temperature. The hematite in the M1 sample formed from the oxidation of Fe that was initially present in the crystal lattice of clay minerals (i.e. Fe-smectite) after structural decomposition. The M2 XRD traces contain magnetite. The dark red colour of the M1 LWA might be attributed to hematite enrichment. However, the M1 LWA shows a slightly darker colour in comparison to the M2 LWA, which can be related to magnetite formation after the probable transformation of hematite. Magnetite might result from the reduction of hematite in the presence of carbonates (Ayati et al., 2018; Pei et al., 2022).

Figure 5. XRD traces of (a) M1 and (b) M2 pellets fired at 1100°C, 1125°C and 1150°C. Ak = akermanite; An = anorthite; Crs = cristobalite; Di = diopside; Fap = fluorapatite; Hem = hematite; Mag = magnetite; Qz = quartz.

Morphology and microstructure of the LWAs

Figure 6 shows the effects of the sintering temperature on the characteristics of the LWAs based on SmC. The expansion of the fired pellets gradually enhanced with increasing sintering temperature from 1100°C to 1150°C (Fig. 6). Both M1 and M2 formulations show maximum expansion at 1150°C (Fig. 6). Some trials conducted at higher temperatures (not shown) showed excessive melting and deformation, hence 1150°C can be considered as the optimal firing temperature. Fired pellets show a stable shape and noticeable bloating. An internal porous structure was developed, but some samples show unvitrified surfaces with pores and cracks (Fig. 6). The M1 pellets are more prone to such defects. These cracks were probably caused by the excessive pressure of gases entrapped in the inner pores. Fakhfakh et al. (2007) also reported crack formation in the external surfaces of LWAs produced from SmC and sintered in the same temperature range. These authors suggest that this is a commonly encountered problem when plastic clays are used due to their excessive shrinkage upon firing. This is particularly critical when large amounts of gas are released almost simultaneously due to carbonate decomposition (Liu et al., 2017). The desirable bloating is favoured, but there are risks of crack formation (Fig. 6). The addition of PhS and PalS minimized this problem. The morphology of such samples shows a vitrified outer surface, denoting the better fluxing characteristics of the M2 mixture (Fig. 6). This ensures greater deformation and a greater capacity to contain the formed gas without crack formation. However, several individual surface pinholes are visible to the naked eye (Fig. 6).

Figure 6. Images of M1 and M2 pellets (the external aspect and the inner part) fired at various temperatures.

SEM images of the LWAs produced from the M1 (100% SmC) and M2 (50% SmC, 25% PhS and 25% PalS) formulations and sintered at 1100°C, 1125°C and 1150°C are shown in Figure 7. For both M1 and M2 LWAs fired at 1100°C, the SEM images indicate low porosity characterized by several small and closed pores scattered in a liquid-rich environment that arose during the sintering process and in which unreacted particles are visible due to lack of sufficient sintering. At 1125°C, the SEM images of the M1 sample without addition are noticeably different from those of the M2 sample containing PhS and palygorskite. The M1 LWA is characterized by a weak porous structure, predominantly with a large number of small pores and with few oval-shaped and sometimes elongated large pores. The development of few large pores ranging in size from 500 μm to 3 mm is clearly observed in the inner core; however, the shell part shows a dense microstructure containing some macropores of 100–350 μm in size and abundant small pores of 20–100 μm in size.

Figure 7. SEM images of the shell and inner core of M1 (100% SmC) and M2 (50% SmC + 25% PalS + 25% PhS) pellets sintered at 1100°C, 1125°C and 1150°C.

Unlike the M1 LWA, the M2 LWA is characterized by a compact structure. The addition of PhS and PalS increased the formation of molten phases. Therefore, with continuous heating at 1125°C, the unreacted solid phases observed at 1100°C began to dissolve, yielding a molten phase that covered the remaining particles. The liquid phase filled the voids around the remaining particles and created a dense microstructure.

The SEM images of the M1 and M2 formulations fired at 1150°C revealed the development of a highly porous network for both samples but particularly for the M2 sample with additives. For both pellets, a relatively dense shell covering the internal core developed (Fig. 7). The outer shell was much less porous than the interior core.

The developed porous structure involves a large proportion of macropores with sizes between 200 and 700 μm in the shell and between 200 μm and 3 mm in the inner core, particularly for the M2 sample. The pore-size distribution is inhomogeneous as much smaller pores are also present. The development of large pores is mainly governed by the internal pressure of the trapped gas during sintering (González et al., 2010).

Most of the pores are not interconnected and only a few interior pores are well interconnected or separated by a thin wall, particularly for the M2 pellets (Fig. 7). These enclosed pores are observed inside the large spherical or ovoid pores in the inner core or close to the surfaces of the LWAs. However, the pores of the shell are more isolated and the walls between the pores are thicker.

The development of pores in the M1 LWA is due to thermal expansion of SmC at high temperature, whereas in the M2 LWA gases liberated from the decomposition of PhS and PalS can create internal pressure, increasing the volume of the pores initially developed at 1125°C. At this stage, the released gas could be more easily trapped by the liquid phase and bloating is more noticeable.

This microstructure has a significant effect on water absorption. Pore structures with large numbers of connected pores in the LWAs might provide spaces to accommodate water and improve water absorption. However, a pore structure containing only non-connected or enclosed pores leads to low water absorption (Moreno-Maroto et al., 2017).

Figure 8 displays the EDS analysis results for randomly selected points on the surfaces and cores of both M1 and M2 formulations at various sintering temperatures. Chemical elemental analysis of the LWAs using EDS reveals the presence of Si, Al, Fe, Na, K and O, elements that result from the decomposition of the raw minerals present in the clayey materials, especially the SmC and PalS samples. Therefore, sintering reactions formed new crystalline phases, which is in agreement with the XRD results that show the Si-rich, Al-rich and K,Ca-Na-rich mineral phases composing the LWAs matrix, as was confirmed by EDS mapping (Fig. 9a,b). The elemental composition of the M2 LWA has changed due to the composition of the additives, with a particular enrichment of Ca, P, S, F and minor Cl, which derive from the decomposition of the carbonate–fluorapatite and gypsum present in the PhS. The presence of iron in both M1 and M2 formulations is related to the starting raw material composition, whereby the sintering at high temperatures promotes the formation of ferrous oxides such as hematite, as detected by XRD. Additionally, EDS mapping of the M2 formulation (Fig. 9b) clearly shows Ca,P-rich sectors homogeneously distributed in the LWA matrix. However, S, F and Cl show a heterogeneous distribution, and their concentrations are probably too low to show clear variations. Non-volatile, potentially hazardous species tend to be immobilized/become inert in the matrix as a result of firing (Wei, 2015; Franus et al., 2016). This behaviour is common in the formulation of ceramics.

Figure 8. EDS spectra of the M1 and M2 LWAs at 1100°C, 1125°C and 1150°C.

Figure 9. (a) EDS elemental mapping of the M1 LWAs at 1100°C, 1125°C and 1150°C. (b) EDS elemental mapping of the M2 LWAs at 1100°C, 1125°C and 1150°C.

Characterization of LWA properties

The results of the expansion properties of the pellets fired at different temperatures, including BI, bulk density, water absorption, and CS, as a function of the firing temperature are shown in Fig. 10 & Table 5. The values obtained for each property are the averages of six measurements obtained from each aggregate variety. The BI increases with firing temperature, and more gradually for the M1 pellets (Fig. 10 & Table 5). The BI of these samples reached 20% at 1150°C, demonstrating the expansive behaviour of the used SmC. For the M2 samples, the increase of BI with temperature is non-linear, and a sharp increase was registered from 1125°C to 1150°C. At this temperature, the expansion is almost double that of the M1 samples. Bloating is a complex phenomenon that causes volume increase upon firing due to gas release (CO₂, CO, O₂, SO₃) from certain components (organic matter, carbonates, Fe₂O₃, etc.). The gases are trapped by the fairly viscous liquid phase and generate inner porosity (González-Corrochano et al., 2010, 2011; Dondi et al., 2016; Molinari et al., 2020; Moreno-Maroto et al., 2020). The pressure of trapped gases increases with temperature, causing bloating or expansion (González-Corrochano et al., 2010). PalS and PhS promote gas formation due to their relatively high contents of carbonaceous compounds. This effect is more evident on PhS, whereas PalS accelerates the formation of fluxing phases, which is crucial for avoiding gas release into the exterior of the samples. Wang et al. (2022) studied the influence of palygorskite on the firing processes of clay mixtures and confirmed its fluxing characteristics.

Figure 10. Physical and mechanical properties of sintered LWAs as a function of temperature.

Table 5. Physical properties of the studied LWAs and the commercial LWA.

The bulk density of pellets is inversely proportional to the BI, as it decreases as the firing temperature increases from 1100°C to 1150°C. The bulk density of both LWAs slightly decreased when the firing temperature increased from 1000°C to 1125°C, although significant bloating was not recorded within this temperature range. When the temperature increased to 1150°C, the pellets showed excessive bloating, which resulted in a significant decrease in bulk density. Bloating is associated with pore formation, which decreases sample weight. The M2 pellets tend to be lighter than the M1 pellets at the same firing temperature, and the bulk density of the pellets is inversely proportional to the BI. The lowest values of bulk density (0.6 and 0.8 g cm^–1) were obtained by firing at 1150°C, and it is clear that the bulk density of the M2 sample was lower than that of the M1 sample. According to European standard EN-13055-1, a good-quality LWA must show a density of <1.2 g cm^–3. Dondi et al. (2016), in a systematic review of LWAs, established the following classification according to density: group I = very low density (0.3–0.6 g cm^–1); group II = low density (0.61–0.99 g cm^–3); group III = medium density (1.0–1.4 g cm^–3); and group IV = high density (>1.4 g cm^–3). SmC pellets (M1 sample) sintered at 1150°C belong to group II (0.8 g cm^–3), whereas M2 pellets fired at the same temperature belong to group I (0.6 g cm^–3). Most of our results are fully consistent with the literature, where the effects of temperature and composition on the bloating of LWAs are detailed (Soltan et al., 2016; Liu et al., 2017, 2018; Yun Cao et al., 2019; Moreno-Maroto et al., 2020; Li et al., 2021; Graziano et al., 2022).

The water absorption of pellets (Fig. 10 & Table 5) can be used to estimate the formation of vitreous phases and the abundance of open pores. As more pores are formed upon heating, there will be a tendency for an increase in water absorption unless abundant vitreous phases form to seal the pores and to create an impermeable shell. In the M1 pellets, the incipient fluxing character of the SmC does not generate sufficient vitreous phases, so the water absorption tends to increase with increasing firing temperature. In the M2 samples, this increment is only visible at 1150°C due to the rapid formation of porosity. In any case, the water absorption values are not very high, indicating that a significant fraction of closed pores was formed and rendering the LWAs less accessible to water.

The CS values of the pellets are expected to show an inverse relationship with water absorption and density because porous samples tend to be mechanically weaker. In general, these trends were observed in this study (Fig. 10 & Table 5). In the M1 pellets, the change in firing temperature causes a gradual variation in the CS (similar to the remaining properties; i.e. the bulk density, BI and water absorption), whereas in M2 sample, change occurs rapidly in the 1125–1150°C range. With the addition of PhS and PalS, the CS increased substantially whilst the density did not increase significantly. The formation of a large amount of a vitreous phase, which fills the pores, might be responsible for the greater strength of the M2 aggregates. However, with increasing temperature, the strength of the LWAs would inevitably decrease because the developed large pores would render the LWAs significantly weaker. Despite the lower density of these M2 pellets in comparison to their M1 counterparts, the CS of the former is higher due to the extended formation of vitreous phases. The addition of PalS and PhS to the SmC seems to ensure a better balance of properties was achieved, namely lower density but greater mechanical strength. These properties are comparable to those of light commercial expanded clay aggregates (Table 5), suggesting the possibility of producing LWAs from phosphate waste and PalS.

Conclusion

In this work, laboratory-prepared pellets based on a SmC and formulated with the addition of PalS and PhS (25 wt.% each) expanded when fired at 1100–1150°C. The expansive potential of the tested formulations (M1 and M2) allows for their use in the fabrication of LWAs. The optimal firing temperature was 1150°C, ensuring maximum expansion of both formulations. The fired pellets had low density (0.8 and 0.6 g cm^–3 for M1 and M2, respectively), and the CS is adequate to ensure their integrity in further formulations.

In the pure SmC pellets, the evolution of studied properties (density, CS, water absorption) with increasing firing temperature was gradual and monotonous. The addition of PalS and PhS changed that pattern, with slower progress occurring at between 1100°C and 1125°C and then with a sudden change arising at 1150°C. This is due to the stronger fluxing characteristics of the PalS above the intermediate tested temperature, whereas PhS enhances gas formation simultaneously with the development of an adequately viscous liquid phase that traps the released gases. This would cause considerable expansion with a developed internal porous structure. Total porosity, pore connectivity and pore-size distribution define the pore structure of the LWAs, and consequently they affect their bulk density, water absorption and CS. The addition of PalS and PhS seemed to ensure a better balance of LWA properties was achieved, namely lower density and greater mechanical strength. The recycling of such waste products/by-products has an obvious environmental impact, contributing to the circular economy. The S, F and Cl occurred in too low concentrations in the LWA matrix to cause environmental issues. However, the emission of gases after the decomposition of PhS would require precautions to be taken at the industrial scale.

The production of LWAs from clay and the used additives could be industrially feasible on the condition that the manufacturing process, with specific regard to the expansion phenomenon, is controlled to produce LWAs with a good porous structure and low density.