Manufacture of ceramics with high mechanical properties from red mud and granite waste

Red mud (bauxite residue) is an alkaline suspension that is the by-product of alumina production via the Bayer process. Its elevated annual production and the global inventory of red mud determine its valorisation. Granite can be used as a source of fluxing oxides for the ceramic industry, as can the flake-shaped waste generated during the flaming of granite. In this work, a set of ceramic pieces made of red mud and granite waste are prepared and characterised via X-ray diffraction, a hardness test, electron scanning microscopy, a leaching test, and determining open porosity, water absorption, bulk density and flexural strength of the samples. The main crystalline phases in the high-temperature fired products are hematite, pseudobrookite and anorthite; the presence of magnetite reveals their ferrimagnetic character. All samples present high mechanical properties. Leaching results are below critical levels established by regulations.


INTRODUCTION
Red mud is the residue generated during the Bayer process, which produces alumina from bauxite.Due to the way the alumina is extracted, the red mud generated constitutes an alkaline solution with a pH between 10 and 13.The annual residue production is estimated to be over 100 million tonnes (Mtpa) and the global inventory reached an estimated 3.5 billion tonnes (Bt) in 2014 (1).
Nowadays, discharging red mud into the sea is avoided (2), as the seabed near the point of discharge can be altered (3)(4)(5).Far from solving the problem, however, methods for storing red mud contribute to environmental and safety risks (6,7).
The great amount of mud generated during the process and its accumulated value determine the importance of its valorisation.Different solutions for using red mud have been presented (8)(9)(10)(11)(12)(13)(14), but its use in the fabrication of building elements is the most widespread solution, as this industry needs large quantities of raw materials.In this field, red mud is used as a starting material, along with traditional compounds like clay, silica or lime.There are several studies that endorse the use of red mud in concrete or cement (15) and in ceramic manufacture (bricks, tiles) (16,17), all of them meeting the required industrial standards.
The flaming of granite generates a flake-shaped residue which has barely been altered compositionally or morphologically, and thus these flakes could be used as a source of fluxing oxides, like feldspars, for the ceramic industry.The flakes are landfilled in the surroundings of the granite processing plant due to the perception that the risk of contamination is low.The production of granite waste can be estimated to be 22 Mtpa, taking in consideration that the amount of waste generated during industrial cutting and processing is 65% (18) of the total granite production (19).
Choosing this material provides an outlet for a residue that is rarely used.Most research on the utilisation of granite waste is limited to the use of the slurry from the cutting/polishing processes or crushed granite itself as aggregate in the production of concrete and ceramic tiles (20,21).
In this paper, the use of red mud and granite flakes from flaming (a treatment that confers a rustic finish to the granite surface) is presented as an option for changing an unused residue which is usually landfilled around transformation points into a raw material for producing ceramic tiles with high mechanical properties.These properties will provide a high added value to the final products.

MATERIALS AND METHODS
A set of samples (10 mm × 80 mm × 10 mm) were prepared using different quantities of the wastes taken as raw materials in this work: red mud from the Alcoa plant in San Ciprián (Lugo, Spain) and residues from the flaming of granite.The red mud was air dried and ground to obtain a fine dust with homogeneous granulometry.The granite flakes were milled in a Fritsch Pulverisette 7 ball mill with tungsten carbide jars, and later they were sieved in a 75 micron mesh.
For this work, the selected red mud/granite waste proportions (% w/w) were 50/50, 60/40 and 70/30.The green ceramic bodies were formed by uniaxially pressing the dry dust with a hydraulic Perkin-Elmer press of 15 Tn and under a load of 10 tons for 90 seconds.
The calcination treatment consisted of: a heating rate of 10 ºC/min, from room temperature to 573 ºC; a soaking time of 10 minutes; a heating rate of 10 ºC/ min to maximum firing temperature; a soaking time of 30 minutes; a cooling rate of 5 ºC/min to 573 ºC; a soaking time of 5 minutes; free cooling from 573 ºC to room temperature.Maximum firing temperatures were: 1100 ºC, 1150 ºC, 1160 ºC and 1170 ºC.For the hardness test and scanning electron microscopy (SEM), the ceramic pieces were polished to a specular finish using a Buehler Phoenix Beta automatic polisher at 500 rpm for 5 minutes with each cloth and diamond paste (9 micron to 1 micron).
The specimens were named according to the percentage of red mud in the paste: LG50, LG60 and LG70.

Mechanical characterisation methods and techniques
A flexural strength test was conducted following the method described by ISO standard 10545-4 (22).The equipment used was an MTS 809 test system with a displacement speed of 0.005 mm/s.
Vickers hardness measurements were obtained in accordance with the UNE-EN 843-4:2005 standard (23), employing a Matsuzawa MXT70 microhardness tester with an incorporated optical microscope.An adequate number of indentations were made under a load of 1 kgf (9.8 MPa) for 40 s.

Compositional and morphological characterisation techniques
The chemical composition of the raw materials was determined by x-ray fluorescence (XRF), using a 4 kW sequential wavelength-dispersive X-ray spectrometer (Philips Magix PRO PW-2440).
In the case of the red mud and the ceramic products, powder x-ray diffraction (XRD) was performed on a Bruker D8 diffractometer, using MoKa1 radiation (0.907300 Å) and a LYNXEYE XE detector.The recorded diffraction angle was from 1º to 35º (2q).For the granite residue, a PANalytical X'Pert PRO MPD diffractometer was used, with CuKα1 radiation (1.5406 Å) and an X'Celerator RTMS detector.The recording was done with a diffraction angle of 10° to 80° (2θ).
Water absorption, open porosity and bulk density were evaluated according to the UNE-EN ISO 10545-3 standard (24).Five samples were used for this determination.
The morphology of the raw materials and the final products was analysed via SEM using a JEOL JSM-6490 LV, with a Si(Li) Oxford Inca X-Sight detector for energy dispersive x-ray spectroscopy (EDX).Prior to analysis, the samples were goldsputtered in order to avoid surface charging.
The leaching test was performed according to the UNE-EN 12457-4 standard (25).Distilled water was used as the leaching liquid.The conductivity and the pH of the leachate were measured with an Oakton PC 510 tester.The leachate analysis was performed with a Nexion 300D inductively coupled plasma mass spectrometer (ICP-MS) by Perkin Elmer.

Raw materials characterisation
The raw material compositions as determined by XRF were presented in Table 1.While the red mud presented high iron oxide (47.35%) and alumina (19.99%) content, as well as significant titanium oxide content (9.81%), the granite waste was rich in silica (67.73%) and alumina (14.19%) with a relatively high content of fluxing oxides that promote sintering.
The XRD analysis in Figure 1 shows that red mud mainly consisted of hematite, goethite, boehmite, calcium iron aluminium silicate and rutile.Quartz, calcite, natrodavyne, sodium aluminium silicate and magnetite were present in lesser amounts.In the case of granite, the main components were quartz, albite, anorthoclase, biotite and muscovite; other phases occurring in a lesser amount were orthoclase and sodium aluminium silicate.

Physical and mechanical properties of the specimens
The generation of open porosity depended on both composition and temperature, as Figure 2 shows.Thus, up to 1150 ºC open porosity decreased with temperature via densification due to the higher formation of glassy phase.Above this temperature, open porosity increased with temperature due to the generation of oxygen during the reduction of iron from red mud, confirmed by a density decrease that is more pronounced as the red mud content increased.As this process was happening, some pores were closed by the formation of the glassy phase; however, as the red mud content increased, the glassy phase content was limited (a smaller amount of granite) and could not close all of the pores that the gas opened when being released, so this favoured better pore interconnection.In the LG50 sample, this porosity effect got masked by the densification produced by the formation of larger amounts of glassy phase provided by the granite, closing pores and their interconnections.The increase of glassy phase was favoured by the presence of alkali-feldspars (albite, anorthoclase, orthoclase, sodium aluminium silicate) in the granite.Figure 2 also shows the vitrification temperatures of the samples, marked as the points where the water absorption and linear shrinkage lines cross.As the figure shows, the temperature at which vitrification began increased with the amount of red mud content in the paste: LG50 begins at 1135ºC, LG60 at 1140ºC and LG70 at 1150ºC.
Water absorption was directly related to the porosity developed during sintering.It can, therefore, be considered a simple way to predict the technological properties of the final products.These results match those for bulk density, except for the case of LG50, as shown in Figure 3.It has been pointed out that the formation of closed pores is responsible for this behaviour (26).Dense phases like hematite, which comes from red mud, were responsible for the higher bulk density in samples rich in bauxite residue.The bulk density values varied between 2.00 g/cm 3 and 2.25 g/cm 3 .
Soaking times were introduced in the firing process at 573 ºC to 1) facilitate the polymorphic transformation of quartz from phase α to the more reactive phase β during the heating process and 2) minimise the stress originating during the cooling process from the reduction of volume due to the transformation of quartz from phase β to phase α.Thus, the presence of a more reactive phase of quartz led to the best densification of the specimens by generating a greater amount of glassy phase, while mechanical behaviour was improved by the reduction of stress during the cooling process.
One of the most representative properties of the mechanical behaviour of a ceramic material is flexural strength.The densification of samples led to better mechanical behaviour, with higher flexural strength (Figure 4); for samples LG60 and LG70, a decrease in density and the formation of iron aluminium silicates (augite, hornblende) led to lower flexural strength (27).Amphiboles (hornblende) and pyroxenes (augite) developed from biotite during the firing process, as described by Bowen's reaction series.

Analysis of selected high-strength specimens
From the results, it can be deduced that the specimens with better mechanical properties (high flexural strength and low water absorption) were those that were sintered at 1150 °C and above.Moreover, Figure 2 shows that vitrification began near this temperature for the three compositions under study.That is why further study was limited to these samples.
Vickers hardness values for all specimens were 440-550 kgf/mm 2 (4.3-5.4GPa), as shown in Table 2.These results depended on densification and on crystallography, and as a higher bulk density and some phases like quartz also increased hardness, the granite content in the paste played an important role in the hardness results.
The mineralogical composition was determined by X-ray diffraction.The main phases for the three studied compositions were: hematite, pseudobrookite, magnetite, sodium anorthite, rutile, nepheline and quartz.A semiquantitative analysis of the fired samples is presented in Table 3. Rutile and nepheline XRD signals became more intense with higher red mud content, whereas granite supplied quartz.Quartz intensity decreased with temperature since it transformed into silicates, providing the formation of the glassy phase and thus the densification and strength of the specimens.This explains the increasing strength in sample LG50.The silicates formed were anorthite, nepheline and augite.The formation of these phases could have been as follows: Albite + Calcite + Quartz = Anorthite Na-feldspar + K-feldspar = Nepheline Biotite + Calcite + Quartz = Augite + Alkali-feldspar Biotite + Calcium iron aluminium silicate + Quartz = Augite + Alkali-feldspar The presence of magnetite in the final products was evidence of their ferrimagnetic character, as magnetite is a ferrimagnetic mineral itself.Even though magnetic properties were not an object of study in this work, the samples were subject to an empirical test, proving that they were attracted by a magnet.The content of magnetite in the fired products varied from 5.1% to 12.3%, as presented in Table 3.Even in the case of the smallest value, the magnetite content was enough to justify the ferrimagnetic character of the ceramics.
Although it could not be identified in the diffractions, ilmenite might have been formed during the firing.The development of ilmenite in the process is essential to explaining the presence of pseudobrookite in the fired samples, as other researchers have found previously (28).The possibility that cations like magnesium entered the crystal lattice of pseudobrookite produced a thermal stabilisation of this phase, since the eutectoid of these solid solutions is located at lower temperatures than the pure aluminium titanate solution (28,29).Furthermore, the formation of magnetite can be similarly explained through the oxidation of wustite from the decomposition reaction of ilmenite.Wustite in the oxidising furnace atmosphere during the cooling stage leads to the formation of magnetite and hematite (30).
The formation of ilmenite from hematite and rutile happened during the heating stage, where a local reducing environment around the samples was created by the diffusion of air to the upper part of the furnace and helped by the release of CO 2 from the decomposition of carbonated phases (calcite and natrodavyne).When the temperature increased, ilmenite decomposed.During the cooling stage, the environment around the samples became oxidising and magnetite and hematite were generated from wustite.Scanning electron microscopy shows the morphological and compositional heterogeneity of the samples.Figure 5 shows that all the specimens were highly porous, with a pore size of about 20-200 mm.The number and average size of the pores increased with the granite content in the paste.Although most of porosity was closed porosity, the number of open pores was not negligible, particularly in the LG70 specimens.Open porosity decreased as the temperature rose in the case of the LG50 samples, as the glassy phase was formed.On the contrary, in the specimens with more red mud content the formation of the glassy phase was not enough to offset the effect of the oxygen generated by the reduction of iron.The release of oxygen was produced during the process of reducing Fe 3+ to Fe 2+ , and with the increase of temperature the iron-rich phases in red mud reacted and gas was emitted.Dehydration of goethite to hematite presented magnetite as an intermediate compound, so part of the Fe 3+ reduced to Fe 2+ (31).Furthermore, hematite transformation to magnetite contributed to this oxygen release as well (32).
These results can be related to the water absorption and bulk density results.The shape of the pores varied with composition and temperature, and thus the more granite there was in the paste or the higher the firing temperature was, the more rounded the pores were.The presence of different phases was demonstrated by the different tones of the backscattered electron compositional images (BEC) in Figure 5, and these can be identified by energydispersive x-ray spectroscopy (EDX) of the selected points marked in the images (Table 4).
According to the EDX and XRD data, in specimen LG50 fired at 1150 ºC, area 1 was compatible with an aluminium-hematite or hematite supported in aluminium silicate matrix.However, area 2 was characterised by a high content of silicon, much higher than aluminium, so it was composed of a vitreous matrix and some quartz.At 1160 ºC, areas 1 and 2 corresponded to hematite particles with different structures, area 3 seemed to be an aluminium silicate (anorthite) with the presence of hematite, and area 4 presented a composition similar to that of pseudobrookite.When this composition was fired at 1170 ºC, area 1 was rich in iron and the amount of titanium could guide the presence of pseudobrookite and rutile.Area 2 was constituted by silicon and oxygen, which corresponds to quartz.In LG70 at 1150 ºC, area 1 was constituted by aluminium silicates (anorthite and nepheline) and was enriched in silicon, which may indicate the presence of quartz.The Ti/Fe relation in area 2 indicated the presence of pseudobrookite in an aluminium silicate matrix.At 1170 ºC, area 1 for LG70 showed hematite supported in a nepheline/anorthite matrix; while area 2 corresponded to pseudobrookite with a presence of anorthite.
The leaching test determines material stability under ordinary environmental conditions and the release of hazardous elements.The most restrictive standard applicable for this case, the "Guideline for drinking-water quality" published by WHO (33), was used as a reference, along with other considerations where applicable.The pH and conductivity values were within the range to consider the leachate as drinking-water, except in the case of pieces sintered at 1170 °C, where conductivity was above the standard limit values for specimens LG60 and LG70.This could be due to the higher open porosity that these samples presented, which favoured the dissolution of phases inside the piece.Even so, levels in leachates for all the specimens were far below the established limits, so the leaching of samples does not involve any health risk and the structural stability of the material is guaranteed.LG50-1150°C LG50-1170°C LG60-1150°C LG70-1150°C LG70-1170°C

Figure 2 .
Figure 2. Water absorption (solid line) and linear shrinkage (dashed line) according to the final firing temperature.Inset: Open porosity according to the final firing temperature.Black line: LG50; Dark grey line: LG60; Light grey line: LG70.