Volcanic Tuff as Secondary Raw Material in the Production of Clay Bricks

The present work examines an innovative manufacturing technique for fired clay bricks, using tuff as a secondary raw material. Samples were made of clay and tuff (0–30 wt.%) fired at 900 to 1100 °C. The chemical and mineralogical compositions and physical and thermal analyses of raw materials were investigated by using SEM-EDS, RX and DTA-TG curves. The samples were analysed from the mineralogical, technological and mechanical points of view. The result show that the tuff’s presence in the clay mixtures considerably reduced the shrinkage of the product during the firing process, and the manufactured samples were of excellent quality. The compressive strength of the bricks varied from 5–35.3MPa, being influenced by the tuff content, clay matrix properties and firing temperatures. Finally, the heat demand for increasing the temperature from room to the firing temperature of the sample with 10% tuff content was 22%.


Introduction
The manufacturing of building materials and products is responsible for about 11% of CO 2 emission at the global level [1], impacting global warming and constantly contributing to climate change. Clay brick is one of the most used building materials in the world, in the construction of structural walls, arches and vaults and for infill panels, in the case of framed structures. The manufacturing process of such products is energy intensive, requiring raw materials preparation and the extrusion, drying and firing of green bricks at temperatures up to 950 • C, which are required for mineral phases' transformation to occur. The total CO 2 emissions generated during the production phase is influenced both by the fuel consumption used in all stages (the transport of raw materials from the quarry to the fabrication site, drying, firing etc.) and decomposition of minerals during the firing process. Even though the energy used for brick production has been reduced since 1990 [2], the role of industry in transitioning to climate neutrality [3] may be significant.
The potential in reducing the environmental impact and the depletion of raw materials may be achieved by using secondary materials and renewable energy systems to save conventional energy at the level of the brick industry and, consequently, in efficiently sustaining the growth of the global economy. The adoption of the circular economy model [3][4][5] is necessary in this approach, including innovative techniques to "close the loop" of products through their reuse or recycling at the end of their life cycles. The samples were a mixture of yellow/grey clay (70:30% by mass, containing mainly quartz, kaolinite, biotite and calcite / kaolinite, quartz, dolomite, calcite [26]) and volcanic tuff in 0% (S1), 5% (S2), 10% (S3), 20% (S4) and 30% (S5) of total mass. All raw materials were dried before use at 90 °C/1h. The specimens had a cylindrical shape with a diameter and height of 18 mm, pressed (40 N/mm 2 ) with a hydraulic press and then fired in an electrical oven at temperatures of 900 °C, 1000 °C and 1100 °C for each mixture. Ceramic materials were shaped by the dry pressing method to eliminate structural defects caused by deformation and shrinkage during the drying and firing processes, but also to reduce the consumption required for sample drying in the preliminary stage.
The firing temperature was gradually increased by 2°/min from room temperature up to a temperature of 600 °C and then, by 5 °C/min, to the final temperature. The samples were held at the final temperature for 2 h, and the heat treatments were conducted in air, as in [27]. Before and after firing, the samples were analysed for their structural, mechanical, and physical properties.

Analytical Methods
The morphologies of the particles and the local chemical composition were determined by SEM-EDS on a Jeol 5600 LV scanning electron microscope. Uncoated, fresh fractures of the fired samples were also subjected to SEM-EDS analysis. The EDS analysis used the ZAF correction standards, implemented in the AZTEC 4.0 software. The particle-size distribution of the raw materials was determined by intrusion mercury porosimetry (Pascal 140). By this method, the size range of the particles is calculated using the pressure needed to break the forces binding the particles together in the agglomerates in the lowpressure regime, according to the Mayer-Stowe theory. The thermal behaviour during the heating process of the press-ready mixture was evaluated using a combined DTA-TG analysis at up to 1000 °C, in air, with a heating rate of 10 °C/min (modified MOM Hungary). X-ray diffractions were recorded on an INEL-Equinox 3000 diffractometer using Co Ka1 radiation (λ = 0.17903 nm). Al samples were ground manually to a fine powder and treated with HF to decrease the SiO2 content. The samples were a mixture of yellow/grey clay (70:30% by mass, containing mainly quartz, kaolinite, biotite and calcite/kaolinite, quartz, dolomite, calcite [26]) and volcanic tuff in 0% (S1), 5% (S2), 10% (S3), 20% (S4) and 30% (S5) of total mass. All raw materials were dried before use at 90 • C/1h. The specimens had a cylindrical shape with a diameter and height of 18 mm, pressed (40 N/mm 2 ) with a hydraulic press and then fired in an electrical oven at temperatures of 900 • C, 1000 • C and 1100 • C for each mixture. Ceramic materials were shaped by the dry pressing method to eliminate structural defects caused by deformation and shrinkage during the drying and firing processes, but also to reduce the consumption required for sample drying in the preliminary stage.
The firing temperature was gradually increased by 2 • /min from room temperature up to a temperature of 600 • C and then, by 5 • C/min, to the final temperature. The samples were held at the final temperature for 2 h, and the heat treatments were conducted in air, as in [27]. Before and after firing, the samples were analysed for their structural, mechanical, and physical properties.

Analytical Methods
The morphologies of the particles and the local chemical composition were determined by SEM-EDS on a Jeol 5600 LV scanning electron microscope. Uncoated, fresh fractures of the fired samples were also subjected to SEM-EDS analysis. The EDS analysis used the ZAF correction standards, implemented in the AZTEC 4.0 software. The particlesize distribution of the raw materials was determined by intrusion mercury porosimetry (Pascal 140). By this method, the size range of the particles is calculated using the pressure needed to break the forces binding the particles together in the agglomerates in the lowpressure regime, according to the Mayer-Stowe theory. The thermal behaviour during the heating process of the press-ready mixture was evaluated using a combined DTA-TG analysis at up to 1000 • C, in air, with a heating rate of 10 • C/min (modified MOM Hungary). X-ray diffractions were recorded on an INEL-Equinox 3000 diffractometer using Co Ka1 radiation (λ = 0.17903 nm). Al samples were ground manually to a fine powder and treated with HF to decrease the SiO 2 content. Compressive strength was measured on a Controls Advantest 9 hydraulic press with a load rate of 0.2 MPa/s on samples having a d/h ratio of~1. The apparent densities of the samples were determined according to SR EN 772-13:2001 [28].
The total shrinkages of the samples were calculated from the samples' dimensions measured using a calliper as (Di−Df) × 100/Di, where Di is a sample's initial diameter and Df its diameter after firing. Sample colour was estimated from sample photos as RGB hexadecimal values.

Results and Discussion
Scanning electron microscopy analysis of the raw materials was performed by using topographic contrast, due to the different distances travelled and the number of electrons emitted from the surface of the sample. All visible particles were agglomerated into coarser, irregular-shaped structures (Figure 2a-c). Compressive strength was measured on a Controls Advantest 9 hydraulic press with a load rate of 0.2 MPa/s on samples having a d/h ratio of ~1. The apparent densities of the samples were determined according to SR EN 772-13:2001 [28].
The total shrinkages of the samples were calculated from the samples' dimensions measured using a calliper as (Di−Df) × 100/Di, where Di is a sample's initial diameter and Df its diameter after firing. Sample colour was estimated from sample photos as RGB hexadecimal values.

Results and Discussion
Scanning electron microscopy analysis of the raw materials was performed by using topographic contrast, due to the different distances travelled and the number of electrons emitted from the surface of the sample. All visible particles were agglomerated into coarser, irregular-shaped structures (Figure 2a-c ). The raw materials were analysed using the EDS probe (Table 1). The raw materials were analysed using the EDS probe (Table 1). From the EDS results it was found that all three raw materials had a high content of SiO 2 (61-74.6%), with significant amounts of Al 2 O 3 , Fe 2 O 3 and MgO and traces of Na 2 O. It may be observed that all raw materials have similar chemical compositions. From a compositional point of view, their alumina contents placed the materials in the category of semi-acidic clays with a significant chromophore oxide (iron oxide) content. An important aspect of the used tuff is its high fraction of glassy content, as suggested by the presence of amorphous, wide peak in the X-ray pattern.
Particle-size analyses of raw materials are presented in Table 2. The used clays can be considered "dusty", containing significant amounts of fine particles (<5 µm) and containing almost no particles with size >63 µm. As is well known from classical powder metallurgy, very fine particles cannot be conveniently formed into dense bodies, since extremely narrow channels inhibit air from being evacuated from the particles. In the present case, the resulting porosity can be of benefit, by reducing the future bricks' mass, provided their compression and green strengths can be maintained at acceptable limits. The forming pressure was chosen as the lowest pressure that gave sufficient green strength for the samples to be easily handled.
As can be seen in the compression curves of the mixtures (Figure 3), increasing pressure did not significantly increase density, a parameter that strongly influences the final mechanical properties on parts manufactured by powder metallurgy from the samples. These naturally agglomerated powder mixtures can be easily formed, by pressing, into a desired shape and have decent green strengths. This fact permits the easy handling of the samples; no spring back-related defects or cracks were visible in the samples, even after 48h of rest.
The X-ray diffractions on the samples fired at 900 • C are presented in Figure 4a. The presented peaks are indexed as originating from the quartz (PDF file 85-0865), feldspar (PDF file 70-1862) and hematite (PDF file 89-8104) Some traces of muscovite (PDF file 46-1409) were also present in samples with 5% tuff. The other chemical elements probably formed an amorphous matrix that bound everything together. Due to the high intensity of the quartz peaks the other phases with less intense peaks were difficult to identify. The macroscopically homogeneous matrix analysed on a mesoscopic level could be described as composed by a glassy matrix holding together the different crystalline phases (quartz, hematite) and containing the embedded porosity, similar to a particle-reinforced composite material.  The X-ray diffractions on the samples fired at 900 °C are presented in Figure 4a. The presented peaks are indexed as originating from the quartz (PDF file 85-0865), feldspar (PDF file 70-1862) and hematite (PDF file 89-8104) Some traces of muscovite (PDF file 46-1409) were also present in samples with 5% tuff. The other chemical elements probably formed an amorphous matrix that bound everything together. Due to the high intensity of the quartz peaks the other phases with less intense peaks were difficult to identify. The macroscopically homogeneous matrix analysed on a mesoscopic level could be described as composed by a glassy matrix holding together the different crystalline phases (quartz, hematite) and containing the embedded porosity, similar to a particle-reinforced composite material. During the firing process these raw materials underwent a series of transformations and decompositions that influenced the final properties of the samples. A DTA-TG curve is presented in Figure 4c. Although no water was added, the natural moisture from the samples evaporated in the low temperature regime and accounted for a mass loss of approximative 4%.
A second endothermic effect was present in the 250-500 • C and caused by the loss of the strongly associated water. At approximately 500 • C another event overlapped with the bonded water loss, the dehydroxylation of kaolinite which became metakaolinite. At temperatures over 700 • C carbonate decomposition occurs. The X-ray diffractions on the samples fired at 900 °C are presented in Figure 4a. The presented peaks are indexed as originating from the quartz (PDF file 85-0865), feldspar (PDF file 70-1862) and hematite (PDF file 89-8104) Some traces of muscovite (PDF file 46-1409) were also present in samples with 5% tuff. The other chemical elements probably formed an amorphous matrix that bound everything together. Due to the high intensity of the quartz peaks the other phases with less intense peaks were difficult to identify. The macroscopically homogeneous matrix analysed on a mesoscopic level could be described as composed by a glassy matrix holding together the different crystalline phases (quartz, hematite) and containing the embedded porosity, similar to a particle-reinforced composite material.

Density
The densities of the samples were determined before and after firing in the oven at 900 • C ( Figure 5). By adding tuff to the mixture, the samples' densities were slightly reduced from 1.71 g/cm 3 (reference sample) to 1.51 g/cm 3 (S5) ( Table 3). After the firing at a temperature of 900 • C, the samples densities were decreased by 10.9% (reference sample), 8.6% (S2), 9.8% (S3), 8.7 (S4) with 4.4% (S5), as compared with the samples' densities measured before firing ( Figure 5). 900 °C ( Figure 5). By adding tuff to the mixture, the samples' densities were slightly re-duced from 1.71 g/cm 3 (reference sample) to 1.51 g/cm 3 (S5) ( Table 3). After the firing at a temperature of 900 °C, the samples densities were decreased by 10.9% (reference sample), 8.6% (S2), 9.8% (S3), 8.7 (S4) with 4.4% (S5), as compared with the samples' densities measured before firing ( Figure 5).    The densities of the samples after firing at the different temperatures (900 • C, 1000 • C and 1100 • C) are shown in Figure 5. The densities of the samples S1 and S2 a decreased as firing temperature increased due to continuous mass loss. By increasing the tuff content, sample densities decreased due to the lower density of the used tuff (1.36 g/cm 3 ) compared with the other raw materials. Increasing the tuff content also modified the chemical composition of the samples and thus their densifications.
The tuffs used were rich in zeolites, which are aluminium tectosilicates that act as tetrahedral carcasses of silicon oxides with the role of molecular sieves [29], ion exchangers, humidity regulators and in the release of "zeolitic water" at temperatures above 100 • C. The use of tuffs with high a content of zeolites can contribute to the better sintering of the entire mixture with the appearance of a glassy mass at temperatures above 1000 • C, which may justify the increase of mechanical strength.
Based on the results obtained, the ceramic materials were of high density, according to SR EN 771-1:2015 [30], which can be used in structural and non-structural masonry elements, if all the requirements imposed by design codes are fulfilled [31,32]. In the context of sustainability, the density of materials may also play an important role; at the building level, the reduced density of bricks as masonry units can lead to the reduction structural elements' contributions and, consequently, the reduction of needed reinforcements in order for the future building to withstand the previsioned loads. The reduced need for materials directly impacts the carbon footprint of the building.

Shrinkage
One way of comparing the degree to which samples' were sintered is their total shrinkage during the solid-stage sintering process [33]. At the initial stage, the dimensional variation is minimal; no significant shrinkage should be visible. In the second stage of sintering, considerable densification occurs, leading to important shrinkage. The ultimate goal of the sintering process, in the present study, was to obtain good compressive strength, yet, at the same time, to have minimal densification in order to preserve as much porosity as possible.
In Figure 6 the dimensional changes of the samples after the firing process are presented. In the graph, the results of firing at 1100 • C indicate a stronger, more intense sintering. In the ternary phase diagram of the K 2 O-Al 2 O 3 -SiO 2 system [34] no liquid phase was formed in the samples fired up to 1000 • C. This is in accordance with the DTA analysis presented in Figure 4. In the first two heating regimes, the samples were subjected to solid-state sintering, so the mass transport in this case is rather limited and the shrinkage is low. The low sintering degree suggests the samples were in the end of the first stage of sintering (see Figure 7). At this stage, the dominant mass transport mechanism is surface diffusion. The samples mainly consist of small particles, so their specific surface areas are high. Even if the samples are in the initial stage of sintering, with no major impact on the porosity their mechanical strength is high. The linear shrinkage of samples increased as more tuff was added, but even so, it was considerably lower than in the case of the control sample (S0) (Figure 7). At temperatures < 1000 • C the shrinkage was less than 1% for all samples, lower, by a factor of three, than other authors findings [20]. The numerical value of the measured shrinkages is presented in Table 4.   Tuff content (%) Figure 7. SEM: sample with 5% tuff content, unfired (a); fired at 900 • C (b); 1000 • C (c) and 1100 • C (d); samples with 30% tuff content unfired (e); fired at 900 • C (f); 1000 • C (g) and 1100 • C (h). In the secondary electron micrographs, it was evident that vitrification was more extensive as the tuff content increased. Additionally, the sintering necks (examples marked by the black arrows) were stronger as the temperature increased, also confirming stronger sintering. Increasing the temperature to 1100 • C, some liquid phase formed, and the liquidphase sintering changed the microstructure; the initial particles were hard to see, having merged into a continuous matrix circumventing the pores. The shrinkage was high; 5.6% at 30% tuff, but even so it was 30% lower than in the case of the control sample.
The different sintering was also visible, macroscopically, in the samples by a colour change from reddish (#908579), at 900 • C, to a light-brownish appearance (#89786e) at 1000 • C, turning to grey (#454f5b) at 1100 • C. No significant colour variation was observed as a function of increasing tuff content.

Compressive Strength
The compressive strength of materials is the most important parameter of masonry units, according to which their usage in construction is established. The mechanical strength of the analysed samples was highly influenced by the tuff content and firing temperature. Sample S5, fired at a temperature of 900 • C, had the compressive strength of 5MPa, 68% lower than the reference sample (S1 , Table 5). At 900 • C and 1000 • C the raw material particles were less sintered, due to their high SiO 2 content, which sinters at higher temperature. When the firing temperature was increased to 1100 • C, compressive strength was increased due to the good bonding between the powder particles, achieved by better sintering. All samples fired at this temperature had compressive strengths between 25 MPa (S5) and 35 MPa (S3); compressive strength tended to be maximized for samples containing 10% tuff (Figure 8). The maximum value of compressive strength was reached because the highest density among all samples sintered at 1100 • C was obtained from this composition.   The high compressive strength, obtained at 1100 °C, allowed the incorporation o further waste material with a double role: to act as a space holder to further decrease den sity and, more importantly, to act as supplementary heat input to reduce the energ needed for firing the bricks [11]. Saw dust or other vegetal residues were successfull added to clay mixtures to act as pore-forming agents and to reduce heat input.
The obtained data suggests that the higher the tuff content, the lower the sample compressive strength. This can be linked to the quartz content, since the quartz undergoe The high compressive strength, obtained at 1100 • C, allowed the incorporation of further waste material with a double role: to act as a space holder to further decrease density and, more importantly, to act as supplementary heat input to reduce the energy needed for firing the bricks [11]. Saw dust or other vegetal residues were successfully added to clay mixtures to act as pore-forming agents and to reduce heat input.
The obtained data suggests that the higher the tuff content, the lower the sample's compressive strength. This can be linked to the quartz content, since the quartz undergoes a sudden, important volume change during the alpha-beta transformation. Although the heating rate was significantly reduced in the transformation temperature range, microcracks eventually appeared between the quartz and the surrounding matrix, especially in the case of larger SiO 2 particles. These cracks can easily propagate through the glassy matrix. So, the importance of reducing compressive strength is not only concerned with increasing porosity but also with the matrix's reduced capacity to block the propagation of cracks. Similar trends of increased tuff percentage with decreased density and compressive strength have been observed in other studies [20,22,35] as evidenced in Table 6. Gencel et al. [22] and Cay et al. [35], from samples of very similar densities and manufactured by the semi-dry pressing process, obtained values of compressive strength, for samples with up to 30% tuff, greater than the minimum value required by national regulations. Vakalova et al. [20] also observed a decreasing trend in the value of the compressive strength of clay-tuff samples manufactured by extrusion at temperatures up to 1050 • C, obtaining compressive strengths up to 23MPa. In their case, the shrinkage was extremely high, up to 14%; tuff had been added in their clay mixture as a humidity stabilizer with favourable effect or for reducing cracks originating from the drying process.
An increase in firing temperature can lead to increase of CO 2 emissions and costs related to the firing process, however, this increase may be within acceptable limits. For a simplified approximation of the energy increase needed, we assumed a similar volumetric heat capacity for the clay minerals and the volcanic tuff [36,37]; the heat difference necessary for increasing the temperature from room temperature to the firing temperature depends on the sample's mass. By using a high quantity of tuff with a low density (1.36 compared to 2.2 g/cm 3 for the clays) the heat demand to increase the temperature of the sample S5 containing 30% tuff to 1100 • C was 116% of that needed for the sample S0 to reach 900 • C. In the case of sample S3, which had the best mechanical properties, the increase of heat demand for firing was approximative 22%. This can lead to the reduction of the environmental impact of these materials by allowing the manufacture of hollow bricks from a stronger material that allows a high void fraction. Using the above-mentioned simplification, we can conclude that the incorporation of a void fraction of over 20% can lead to more benign environmental impact, compared with the reference sample; yet, at the same time, the compressive strength should be above the minimal imposed value of 20 MPa.

Conclusions, Contributions, and Further Work
Herein, we have investigated the potential for using tuff as a secondary raw material in clay brick production. We analysed the optimal percentage of tuff that can be incorporated in a clay matrix. The results showed the following: The compressive strength of tuff-clay samples depended strongly on sintering temperature. When heated to 900 • C, the values of compressive strength were lower than 15.7 MPa for all samples (S2-S5). On the other hand, samples fired at 1100 • C presented higher compressive strengths, an increase of up to three-fold as compared with the reference bricks. This increase was due to the forming of stronger bonds between the particles. The increase in compression strength was also due to the formation of an important volume fraction of liquid that further accelerate the sintering process.
The added volcanic tuff further reduced the firing shrinkage. The formation of the liquid phase accelerated the sintering and, therefore, the firing shrinkage. By correctly choosing the sintering conditions we managed to reduce shrinkage by almost one magnitude lower, when comparing sample S2 to the reference sample S1 at 1100 • C. In the other cases the shrinkage reduction was also significant, though the difference was lower as the tuff content increased. Increasing the firing temperature increased the embodied energy of the final product; however, by increasing the mechanical strength, one can add pore formers that improves thermal performance in energy-efficient buildings.
The present work showed that tuff can be successfully used as a secondary raw material in the fabrication of fired clay bricks, in ratios of up to 30%, thus, contributing to the circular economy and the EU's zero-waste target. The results obtained in the present work will contribute to future research on the optimization of ceramic products based on tuff and other secondary raw materials, incorporated in a clay matrix as pore forming agent, in terms of thermal, mechanical and environmental performance.