Conditioning Influence of Kaolinite Matrices on Flexural Strength of Raw Pressed Slurry Collected from Ceramic Tile Production Wastewater

: Kaolinite is able to assure the high binding affinity of the filler particles of raw ceramic bodies. It acts as a matrix that strongly holds the other constituents’ particles in a compact structure. The slurry samples were characterized by XRD, mineralogical microscopy and SEM coupled with an EDX elemental analysis. The slurry collected from the ceramic tile production wastewaters had a significant amount of kaolinite (36%), mostly fine particles of 3 µ m, less surrounding quartz (37%) and mullite (19%) particles of 5–100 µ m in diameter and traces of lepidocrocite (8%). It is a dense paste with a relative moisture of 25%. The square bar of the slurry as received, pressed at a load of 350 N, had a flexural strength of 0.61 MPa. Increasing the moisture to 33% using regular water, followed by mechanical attrition at 2000 rpm for 5 min, resulted in a porous bar with a flexural strength of 0.09 MPa; by increasing the attrition speed to 6000 rpm, the microstructural homogenization was improved and the flexural strength was about 0.68 MPa. It seems that regular water does not assure an optimal moisture for the kaolinite matrix conditioning. Therefore, we used technological water at pH = 10, a moisture of 33% and attrition at 6000 rpm for 5 min, and the bar pressed at a load of 350 N had a flexural strength of 1.17 MPa. The results demonstrate that the bar moistened with technological water and an attrition regime assured a proper conditioning for the kaolinite matrix, achieving the optimal binding of the quartz and mullite particles under the pressing load. Bars with the optimal mixture were pressed at several loads, including 70, 140, 210 and 350 N, and the flexural strength was progressively increased from 0.56 MPa to 1.17 MPa. SEM fractography coupled with atomic force microscopy (AFM) revealed that the optimal moisture facilitated a proper kaolinite particle disposal regarding the quartz and mullite filler particles, and the progressive load assured the strong binding of the finest kaolinite platelets onto their surface.


Introduction
Kaolinite is one of the most versatile minerals among the phyllosilicate family due to its increased physicochemical stability and its binding abilities, which are due to its tetrahedral SiO 4 sheets linked by oxygen atoms, as well as an alumina sheet of octahedral AlO 6 , generating a multilayered structure [1,2].The data in the literature reveal that water or other molecules' abilities to penetrate kaolinite sheets increase their interplanar distance and subsequent binding ability [3,4].On the other hand, kaolinite presents interesting breaking behavior, being very resistant to the solicitation exerted perpendicularly to the silica sheets and easily breaking through the cleavage under the solicitation oriented directly to the sheet folds [5].The cleaved kaolinite sheets might be subjected to external forces that continue their fragmentation, resulting in fine and ultra-fine particles [6].There are several uses of ultra-fine kaolinite particles, including as a filler in washable paints [7,8], plasters and mortars [9,10].
Kaolinite is often found in the presence of other mineral particles, such as quartz or calcite, which might influence their integrity.When kaolinite particles are under intensive movement, admixture with other, harder, mineral particles like quartz might act as milling bodies generating very small fractions.Such a mechanism was observed in the quartz-kaolinite interaction in environmental street dust [11,12], but it is currently used in raw material preparation in the ceramic industry for porcelain pottery [13,14] and ceramic tiles [15,16].The mineral admixture is milled under a wet condition at a very well-controlled humidity level to obtain ceramic paste, which is further molded into dies.The used waters from ceramic facilities contain significant levels of dispersed particles, mainly quartz and kaolinite.These are further filtered during used-water treatment and become a pressed slurry.Iron presence does not allow for its recirculation in the same technological process because of the glaze stain risk [17,18], and therefore, it becomes a waste slurry.Its recycling poses a challenge for the reduction in dump site volume and in avoiding fine particle release into the atmosphere.The aim of sustainable recycling implies a lot of additional research regarding the microstructure and particle distribution within the resultant ceramic slurry to figure out its viable applications.
Our previous study not only showed the slurry formation from the wastewaters derived from wall and floor tile technological processes and, subsequently, the low amount of iron hydroxide crystallized as lepidocrocite, but it also revealed the presence of ultrafine kaolinite particles that seemed to be able to enhance the binding abilities through pressing [19].Our findings were in agreement with the literature [20,21].Thus, the composite science approach allowed us to better understand the pressed slurry behavior, where kaolinite acted as the matrix and the quartz particles represented the filler.
Usually, the coarse fractions found in the ceramic tile slurry are formed by quartz and mullite particles.The amount of quartz is assured by the genuine raw material feedstock, while the mullite particles are recirculated material obtained from the milling of the damaged fired biscuits.These particles disposed under the paste bulk are of great importance for assuring the desired homogeneity.Unfortunately, the particles are subjected to natural sedimentation in the used-water tanks prior to the pressing filtration.This further contributes to an uneven distribution of larger particles with respect to the finest fractions.The data in the literature mention composite failure due to filler non-homogeneities regarding the matrix [22,23].
Thus, micro-dispersion investigation plays a crucial role in the successful consolidation of the ceramic structure through the kaolinite matrix and requires advanced microscopic techniques for their characterization and to establish proper conditioning operations.Our first aim is to figure out a complex characterization of the ceramic tile wastewater pressed slurry using mineralogical optical microscopy (MOM) coupled with scanning electronic microscopy (SEM), along with an elemental analysis through energy dispersion spectroscopy (EDS) and mineral composition investigation by X-ray diffraction (XRD).It is expected that all of these observations will provide the proper parameters for the kaolinite matrix conditioning in order to achieve the optimal particle distribution within the composite structure so to achieve the best flexural strength.
Moisture is one of the most important conditioning parameters that must be followed because of its ability to activate kaolinite platelet adsorption onto the other mineral particles [4,24].Another important parameter is microstructure homogenization.The data in the literature indicate ball milling for high moisture levels [25] and blade milling for average moisture [26].It is expected that the moderate moisture will assure a proper mobility of the quartz and mullite particles towards the kaolinite matrix, which brings a homogeneous microstructure facilitating their optimal binding.The conditioning parameters and subsequent microstructure influence on sample flexural strength will be followed.The null hypothesis of the present research states that moisture and milling variation have no influence on the flexural strength of the pressed bars.

Samples Preparation
The current study used a ceramic slurry resulting from used water treatment from the ceramic tile fabrication technological process.We collected 20 kg of pressed slurry from the dump deposit of a ceramic tile facility in Cluj County, Romania (the company name is kept anonymous for economic reasons).The slurry was conditioned, and the compacted bars were prepared at room temperature, 23 • C, under atmospheric pressure.
The control sample was prepared by molding the slurry as received in the metallic die presented in Figure 1a using a compression load of 350 N. It results in a bar with rectangular section suitable for flexural testing, Figure 1b.The moisture of the received slurry was measured as being about 25%. the literature indicate ball milling for high moisture levels [25] and blade milling for average moisture [26].It is expected that the moderate moisture will assure a proper mobility of the quartz and mullite particles towards the kaolinite matrix, which brings a homogeneous microstructure facilitating their optimal binding.The conditioning parameters and subsequent microstructure influence on sample flexural strength will be followed.The null hypothesis of the present research states that moisture and milling variation have no influence on the flexural strength of the pressed bars.

Samples Preparation
The current study used a ceramic slurry resulting from used water treatment from the ceramic tile fabrication technological process.We collected 20 kg of pressed slurry from the dump deposit of a ceramic tile facility in Cluj County, Romania (the company name is kept anonymous for economic reasons).The slurry was conditioned, and the compacted bars were prepared at room temperature, 23 °C, under atmospheric pressure.
The control sample was prepared by molding the slurry as received in the metallic die presented in Figure 1a using a compression load of 350 N. It results in a bar with rectangular section suitable for flexural testing, Figure 1b.The moisture of the received slurry was measured as being about 25%.The slurry moisture and the microstructural uniformity was controlled by adjusting the water content and the milling speed.Several tests were performed, and the most relevant preparations are taken discussed in T0 to T3, Table 1.
Each composition was molded into the metallic die presented in Figure 1a at a loading pressure of 350 N for the test samples.Thus, the optimal conditioning regime was identified.The composition prepared in the optimal conditions was molded as bars at different loads: 70, 140 and 210 N.All molded bars, after pressing, were kept in the die 24 h to become stiffer prior to extraction.They were dried under atmospheric conditions at 23 °C for three months to assure a gentle water removal to preserve their obtained microstructure.Thereafter, the bars were subjected to flexural strength testing and to fractography analysis.The slurry moisture and the microstructural uniformity was controlled by adjusting the water content and the milling speed.Several tests were performed, and the most relevant preparations are taken discussed in T0 to T3, Table 1.
Each composition was molded into the metallic die presented in Figure 1a at a loading pressure of 350 N for the test samples.Thus, the optimal conditioning regime was identified.The composition prepared in the optimal conditions was molded as bars at different loads: 70, 140 and 210 N.All molded bars, after pressing, were kept in the die 24 h to become stiffer prior to extraction.
They were dried under atmospheric conditions at 23 • C for three months to assure a gentle water removal to preserve their obtained microstructure.Thereafter, the bars were subjected to flexural strength testing and to fractography analysis.

Investigation Methods
The mineralogical composition of the samples was investigated with X-ray diffraction (XRD) using a Brag-Brentano diffractometer, Bruker D8 Advance (Bruker Co., Karlsruhe, Germany) with a Cu K α radiation λ = 1.54056Å.The used 2 Theta angle range was between 10 and 80 degrees at a speed of 1 deg./minute.The diffraction peaks assignment was performed with Match 1.0 software with the PDF 2.0 database (Crystal Impact Company, Bonn, Germany).
The identified mineral correlations with the sample particles was established through mineralogical optical microscopy (MOM) performed in the cross-polarized light mode with a Laboval 2 (Carl Zeiss Company, Oberkochen, Germany).The MOM images were digitally acquired using a 14 MPx Sony photographical system (Sony Group Corporation, Minato, Japan).
Advanced microstructural observation, including its corresponding elemental analysis, was performed with a scanning electron microscope (SEM), the Hitachi SU8230 (Hitachi Company, Tokyo, Japan), and an elemental spectroscopy device, Energy Dispersive Spectroscopy (EDS) X-Max 1160 EDX (Oxford Instruments, Oxford, UK).The investigation was performed at an acceleration voltage of 30 kV in high vacuum mode.The samples' electrical conductivity was assured by a very thin Pt layer deposited by a mild metallization.Therefore, the Pt component from the elemental investigation was subtracted.
The samples' flexural strength was measured using a Lloyd LR5k Plus testing machine (Ametek Lloyd Instruments, Meerbusch, Germany).The measurements were conducted through the Ametek Lloyd software version 1.0 and further processed with Microcal Origin 6.0 software (Microcal Company, Northampton, MA, USA).The three point bending method was used with a span of 65 mm and a loading speed of 0.5 mm/min.The flexural strength results were statistically analyzed with the Microcal Origin Lab version 2018b software (Microcal Company, Northampton, MA, USA) using an ANOVA test followed by a Tukey post hoc test at a significance level p = 0.05.
Fractographic investigation was performed through SEM microscopy with an Inspect S Microscope (FEI Company, Hillsboro, OR, USA).The examination was performed in low vacuum mode at 25 kV acceleration voltages, without sample metallization.
Atomic force microscopy was performed in tapping mode with a Jeol JSPM 4210 Scanning Probe Microscope (JEOL, Tokyo, Japan) using NSC 15 Hard cantilevers produced by Mikro-Masch Company (Sofia, Bulgaria) with a resonant frequency of 325 kHz and a force constant of 40 N/m.The images were scanned at a rate of 1 Hz with a scanning area of 1 µm × 1 µm.At least three different macroscopic areas were scanned.All images were processed using Win SPM 2.0 software produced by JEOL Company (Tokyo, Japan).

Mineral Phase Characterization
The slurry collected from the ceramic tile production wastewater has a highly mineral characteristic due to the raw materials involved in the technological process and the specific procedures.Therefore, it requires a precise investigation using the adequate methods.X-ray diffraction (XRD) is one of the most important methods for crystalline phase identification and was successfully applied on the raw sample and on the conditioned one.
The initial sample, T0, has a well-developed XRD pattern presented in Figure 2a.Strong diffraction peaks are observed, having large intensity and a narrow aspect accompanied by several less intense peaks with slightly broadened width.This corresponds to a complex mixture of crystalline phases in which some of them have very small particles.The composition of the T0 sample is dominated by quartz, followed by kaolinite, mullite, and traces of lepidocrocite.

Mineral Phase Characterization
The slurry collected from the ceramic tile production wastewater has a highly mineral characteristic due to the raw materials involved in the technological process and the specific procedures.Therefore, it requires a precise investigation using the adequate methods.X-ray diffraction (XRD) is one of the most important methods for crystalline phase identification and was successfully applied on the raw sample and on the conditioned one.
The initial sample, T0, has a well-developed XRD pattern presented in Figure 2a.Strong diffraction peaks are observed, having large intensity and a narrow aspect accompanied by several less intense peaks with slightly broadened width.This corresponds to a complex mixture of crystalline phases in which some of them have very small particles.The composition of the T0 sample is dominated by quartz, followed by kaolinite, mullite, and traces of lepidocrocite.The mass participation of each crystalline phase in the sample composition was determined by the RIR (Reference Intensity Ratio) method previously described in the literature [19], which is based on the relative intensities of each identified compound and its corundum factor.The obtained values are centralized in Table 2.
We found an interesting behavior of kaolinite where it generates some intense and narrow peaks and the other ones are less intense and broadened.This indicates the presence of bigger particles mixed up with very small fractions that requires further microscopical investigation.
The intensive preparation of the slurry sample during conditioning achieves a relative re-organization of the mineral particles by milling the bigger kaolinite particles, implying a better organization of the microstructure.This fact influences the relative intensities of the peaks in the XRD pattern, Figure 2b.Of course, the main distribution is the The mass participation of each crystalline phase in the sample composition was determined by the RIR (Reference Intensity Ratio) method previously described in the literature [19], which is based on the relative intensities of each identified compound and its corundum factor.The obtained values are centralized in Table 2.
We found an interesting behavior of kaolinite where it generates some intense and narrow peaks and the other ones are less intense and broadened.This indicates the presence of bigger particles mixed up with very small fractions that requires further microscopical investigation.
The intensive preparation of the slurry sample during conditioning achieves a relative re-organization of the mineral particles by milling the bigger kaolinite particles, implying a better organization of the microstructure.This fact influences the relative intensities of the peaks in the XRD pattern, Figure 2b.Of course, the main distribution is the same as the initial sample being dominated by quartz, but the kaolinite amount is significantly increased due to the better disposal of the particles in the three-dimensional network of the system matrix, Table 2.The lepidocrocite amount also slightly increases due to the better refinement of the microstructure, which has dispersed fine fractions.Mineralogical optical microscopy (MOM) uses cross-polarized light for the observation of sample particles, evidencing each mineral species in specific chromatic tones.The color intensity and tone vary with the particle position regarding the optical axis of the microscope featuring a range from the most intense tone to the extinction featuring a darkened color tone.It allows the association of the minerals identified by XRD to the particle's observation, bringing on correlations with their size and shapes, Figure 3. same as the initial sample being dominated by quartz, but the kaolinite amount is significantly increased due to the better disposal of the particles in the three-dimensional network of the system matrix, Table 2.The lepidocrocite amount also slightly increases due to the better refinement of the microstructure, which has dispersed fine fractions.Mineralogical optical microscopy (MOM) uses cross-polarized light for the observation of sample particles, evidencing each mineral species in specific chromatic tones.The color intensity and tone vary with the particle position regarding the optical axis of the microscope featuring a range from the most intense tone to the extinction featuring a darkened color tone.It allows the association of the minerals identified by XRD to the particleʹs observation, bringing on correlations with their size and shapes, Figure 3.The T0 sample has a heterogeneous distribution of the mineral particles as observed in Figure 3a.There are observed bigger mullite particles with boulder shapes colored in reddish brown tones accompanied with significant fractions of quartz particles colored in green-gray tones.These big fractions are embedded in a dense mixture of kaolinite particles: the bigger fractions present a lamellar-tabular shape having light blue to white tones while the smaller particles present a dense matrix having white aspects (the thicker areas have yellow shallow due to the small kaolinite particles overlapping).A closer look was taken at high magnification in Figure 3a', where the white kaolinite spots are more visible, The T0 sample has a heterogeneous distribution of the mineral particles as observed in Figure 3a.There are observed bigger mullite particles with boulder shapes colored in reddish brown tones accompanied with significant fractions of quartz particles colored in green-gray tones.These big fractions are embedded in a dense mixture of kaolinite particles: the bigger fractions present a lamellar-tabular shape having light blue to white tones while the smaller particles present a dense matrix having white aspects (the thicker areas have yellow shallow due to the small kaolinite particles overlapping).A closer look was taken at high magnification in Figure 3a', where the white kaolinite spots are more visible, and the smaller reddish dots belong to fine lepidocrocite particles.The overall measurements allow for size range centralization for each mineral species; see Table 2.
The intensive conditioning of the T3 sample causes a uniform distribution of the bigger quartz and mullite particles, which act as reference points of a network sustained by the dense mass of fine kaolinite particles acting like a composite matrix, Figure 3b.The kaolinite fine particle redistribution is evidenced in Figure 3b' by the obvious dissipation of the bright white clusters into a better dispersed paste.This fact is assured by the relative increase in sample moisture, allowing easier rearrangements within fine particles.This allows the lepidocrocite fraction to mobilize towards the kaolinite matrix.
Excessive moisture of the composition (over 40%) causes material cohesion failure and transforms it into wet slurry, Figure 3c.It allows better observation of the boulder shape of quartz and mullite particles along with the tabular shape of bigger kaolinite fractions.Figure 3c' has the advantage of showing a rounded red particle of lepidocrocite of 25 µm diameter.It is a slightly unusual finding due to the prevalence of smaller fractions.
The moisture influence on the kaolinite matrix is clearly evidenced by the MOM observations.It requires an additional investigation regarding water's influence on its crystalline structure.Therefore, the samples with controlled humidity were exposed to XRD investigation towards the 29.0 to 29.6 range of 2 theta angles to observe the position of the diffraction peak generated by the crystallographic plane having the Miller indices (−112) regarding PDF 02-0105.We observed peak position displacement to the lower angle position with the moisture increasing, Figure 4a.and the smaller reddish dots belong to fine lepidocrocite particles.The overall measurements allow for size range centralization for each mineral species; see Table 2.
The intensive conditioning of the T3 sample causes a uniform distribution of the bigger quartz and mullite particles, which act as reference points of a network sustained by the dense mass of fine kaolinite particles acting like a composite matrix, Figure 3b.The kaolinite fine particle redistribution is evidenced in Figure 3b' by the obvious dissipation of the bright white clusters into a better dispersed paste.This fact is assured by the relative increase in sample moisture, allowing easier rearrangements within fine particles.This allows the lepidocrocite fraction to mobilize towards the kaolinite matrix.
Excessive moisture of the composition (over 40%) causes material cohesion failure and transforms it into wet slurry, Figure 3c.It allows better observation of the boulder shape of quartz and mullite particles along with the tabular shape of bigger kaolinite fractions.Figure 3c' has the advantage of showing a rounded red particle of lepidocrocite of 25 µm diameter.It is a slightly unusual finding due to the prevalence of smaller fractions.
The moisture influence on the kaolinite matrix is clearly evidenced by the MOM observations.It requires an additional investigation regarding water's influence on its crystalline structure.Therefore, the samples with controlled humidity were exposed to XRD investigation towards the 29.0 to 29.6 range of 2 theta angles to observe the position of the diffraction peak generated by the crystallographic plane having the Miller indices (−112) regarding PDF 02-0105.We observed peak position displacement to the lower angle position with the moisture increasing, Figure 4a.This fact indicates that the water molecules progressively penetrate the (−112) crystallographic planes.The interplanar distance was calculated for each diffraction peak using Bragg's law and the obtained variation was plotted in Figure 4b.It clearly reveals the interplanar distance increase with the sampleʹs moisture.The lower humidity, 20%, assures an interplanar distance of 3.042223 Å that corresponds with the information in the PDF 02-0105 file.A small humidity increase to 25% determines the interplanar distance swelling to 3.044922 Å, a fact that confirms the initial assumption.The interplanar distance at 30% moisture is 3.046802 Å, representing a significant capacitating of the kaolinite particles binding effect.The interplanar distance continues its linear increasing with the moisture level up to 40%.The higher moisture leads to a lack of material cohesion and particle mobilization as free dispersed matter, a situation that is improper for XRD investigation.
Taking into consideration the moisture influence on the kaolinite interplanar distances, we increased the slurry moisture up to about 30% by adding water.The prepared slurry samples' effective moisture was measured with a thermo-balance and the obtained values are given in Table 1.This fact indicates that the water molecules progressively penetrate the (−112) crystallographic planes.The interplanar distance was calculated for each diffraction peak using Bragg's law and the obtained variation was plotted in Figure 4b.It clearly reveals the interplanar distance increase with the sample's moisture.The lower humidity, 20%, assures an interplanar distance of 3.042223 Å that corresponds with the information in the PDF 02-0105 file.A small humidity increase to 25% determines the interplanar distance swelling to 3.044922 Å, a fact that confirms the initial assumption.The interplanar distance at 30% moisture is 3.046802 Å, representing a significant capacitating of the kaolinite particles binding effect.The interplanar distance continues its linear increasing with the moisture level up to 40%.The higher moisture leads to a lack of material cohesion and particle mobilization as free dispersed matter, a situation that is improper for XRD investigation.
Taking into consideration the moisture influence on the kaolinite interplanar distances, we increased the slurry moisture up to about 30% by adding water.The prepared slurry samples' effective moisture was measured with a thermo-balance and the obtained values are given in Table 1.
The diffraction peaks in Figure 4a are less intense and slightly broadened, proving that they were generated by X-ray beam diffraction in the finest kaolinite particles, which are very numerous according to the MOM observation.Their diameter is difficult to measure directly on the MOM images due to the lack of magnification (a fact in agreement with Abbe's law).Thus, the fine kaolinite particle size was determined using the Full Width at the Half-height of the peak's Maximums (FWHM) using the Scherrer formula, resulting in an average size of 55 nm.

Sample Microstructure and Elemental Composition
The microstructure of the initial conditioning test samples, T series, was investigated with SEM imaging at low magnification (e.g., 100×) for the overall aspect and high magnification (e.g., 2000×) for the particle's distribution details.
The overall aspect of the initial sample, Figure 5a, has a heterogeneous aspect with irregular aspects induced by local particle segregations and presents significant pores.The larger ones (500-1000 µm) have oval and elongated shapes due to the local plastic deformation of the material during molding and several small, rounded pores of about 30-50 µm are observed as dark spots spread over the observation field in Figure 5a.Pore presence is unwanted because of their trait of flexural strength weakening.The high-magnification detail is focused on the compact material not over the pores to reveal its organization, Figure 5a'.The center of the observation field is dominated by the presence of large kaolinite tabular particles having polyhedral shapes and sizes in the range of 30-50 µm.The diffraction peaks in Figure 4a are less intense and slightly broadened, proving that they were generated by X-ray beam diffraction in the finest kaolinite particles, which are very numerous according to the MOM observation.Their diameter is difficult to measure directly on the MOM images due to the lack of magnification (a fact in agreement with Abbe's law).Thus, the fine kaolinite particle size was determined using the Full Width at the Half-height of the peak's Maximums (FWHM) using the Scherrer formula, resulting in an average size of 55 nm.

Sample Microstructure and Elemental Composition
The microstructure of the initial conditioning test samples, T series, was investigated with SEM imaging at low magnification (e.g., 100×) for the overall aspect and high magnification (e.g., 2000×) for the particleʹs distribution details.
The overall aspect of the initial sample, Figure 5a, has a heterogeneous aspect with irregular aspects induced by local particle segregations and presents significant pores.The larger ones (500-1000 µm) have oval and elongated shapes due to the local plastic deformation of the material during molding and several small, rounded pores of about 30-50 µm are observed as dark spots spread over the observation field in Figure 5a.Pore presence is unwanted because of their trait of flexural strength weakening.The high-magnification detail is focused on the compact material not over the pores to reveal its organization, Figure 5a'.The center of the observation field is dominated by the presence of large kaolinite tabular particles having polyhedral shapes and sizes in the range of 30-50 µm.The left upper side of the image in Figure 5a' features a quartz particle having a boulder shape and a size of about 60 µm while the lower left corner features an equiaxed mullite particle of about 45 µm.These bigger particles take the filler particle role and are surrounded by a dense matrix of fine kaolinite particles in the range of 1-10 µm.Small ultra structural kaolinite particles cannot be observed by SEM and require more enhanced techniques that are described in a special subchapter.
The milling of the T1 sample at 2000 rpm, Figure 5b, generated a granulated overall mixture with rounded pellets of about 1 to 1.5 mm diameter relatively stacked each to another due to the molding load of 350 N. The spatial adhesion necks are observed The left upper side of the image in Figure 5a' features a quartz particle having a boulder shape and a size of about 60 µm while the lower left corner features an equiaxed mullite particle of about 45 µm.These bigger particles take the filler particle role and are surrounded by a dense matrix of fine kaolinite particles in the range of 1-10 µm.Small ultra structural kaolinite particles cannot be observed by SEM and require more enhanced techniques that are described in a special subchapter.
The milling of the T1 sample at 2000 rpm, Figure 5b, generated a granulated overall mixture with rounded pellets of about 1 to 1.5 mm diameter relatively stacked each to another due to the molding load of 350 N. The spatial adhesion necks are observed assuring sample mechanical cohesion.The high-magnification detail in Figure 5b' was taken on a single pellet.It is formed by a dense mixture containing boulder quartz and equiaxed mullite particles in the range of 25-35 µm surrounded by a dense matrix of kaolinite fine particles, most of them around 1 µm and below.The bigger kaolinite particles are triangularly shaped as a consequence of the intensive breaking during mixture milling.It seems that additional moisture at a pH of 6.9 facilitates milled slurry pelletizing.
The slurry microstructure overall aspect became uniform without pellets by increasing the milling speed at 6000 rpm, Figure 5b.Some rounded pores of about 300 µm occur occasionally, like the one observed on the upper left side of the image in Figure 5b.The grainy aspect of the sample surface is due to the intensive milling of the particles that causes their re-arrangements and refinement.This is shown by the high magnification microstructural detail in Figure 5b'.The observation field within the image is dominated by some freshly milled kaolinite particles with triangular shape and size of about 25 µm.It proves a strong reduction of the bigger kaolinite particles' diameter to half.These are accompanied by some quartz and mullite particles in the range of 25-40 µm being very well embedded into the fine kaolinite particle matrix.
The association between high-speed milling at 6000 rpm and alkaline moisture (technological water for the ceramic tile fabrication having pH = 10) in T3 assures a sustainable reticulation of the kaolinite fine particles that are organized into a dense matrix embedding bigger mineral fractions such as quartz and mullite, Figure 5d.The obtained material is uniform and without pores.The high-magnification investigation in Figure 5d' reveals an advanced milling of the bigger kaolinite particles that are chopped into small fractions ranging from 1 to 10 µm that surround the quartz and mullite particles in a dense manner.The advanced binding behavior is capacitated by the ultra-structural fractions of about 55 nm evidenced by the XRD determination.Therefore, T0 and T3 samples were subjected to an advanced SEM investigation with elemental analysis, Figure 6.assuring sample mechanical cohesion.The high-magnification detail in Figure 5b' was taken on a single pellet.It is formed by a dense mixture containing boulder quartz and equiaxed mullite particles in the range of 25-35 µm surrounded by a dense matrix of kaolinite fine particles, most of them around 1 µm and below.The bigger kaolinite particles are triangularly shaped as a consequence of the intensive breaking during mixture milling.It seems that additional moisture at a pH of 6.9 facilitates milled slurry pelletizing.The slurry microstructure overall aspect became uniform without pellets by increasing the milling speed at 6000 rpm, Figure 5b.Some rounded pores of about 300 µm occur occasionally, like the one observed on the upper left side of the image in Figure 5b.The grainy aspect of the sample surface is due to the intensive milling of the particles that causes their re-arrangements and refinement.This is shown by the high magnification microstructural detail in Figure 5b'.The observation field within the image is dominated by some freshly milled kaolinite particles with triangular shape and size of about 25 µm.It proves a strong reduction of the bigger kaolinite particles' diameter to half.These are accompanied by some quartz and mullite particles in the range of 25-40 µm being very well embedded into the fine kaolinite particle matrix.
The association between high-speed milling at 6000 rpm and alkaline moisture (technological water for the ceramic tile fabrication having pH = 10) in T3 assures a sustainable reticulation of the kaolinite fine particles that are organized into a dense matrix embedding bigger mineral fractions such as quartz and mullite, Figure 5d.The obtained material is uniform and without pores.The high-magnification investigation in Figure 5d' reveals an advanced milling of the bigger kaolinite particles that are chopped into small fractions ranging from 1 to 10 µm that surround the quartz and mullite particles in a dense manner.The advanced binding behavior is capacitated by the ultra-structural fractions of about 55 nm evidenced by the XRD determination.Therefore, T0 and T3 samples were subjected to an advanced SEM investigation with elemental analysis, Figure 6.Both samples T0 and T3 have similar elemental compositions dominated by oxygen, silicon and aluminum due to quartz, kaolinite and mullite presences, Table 3.The kaolinite particle refinement as a consequence of the milling process and their subsequent re-arrangement causes a slightly increased Al content in the T3 sample compared to T0.The Both samples T0 and T3 have similar elemental compositions dominated by oxygen, silicon and aluminum due to quartz, kaolinite and mullite presences, Table 3.The kaolinite particle refinement as a consequence of the milling process and their subsequent re-arrangement causes a slightly increased Al content in the T3 sample compared to T0.The kaolinite matrix structuring brings better Al characteristic X-ray beams, and the atomic percentage is situated around 7.5%.Figure 6a evidences the elements distribution map related to the microstructural features observed on the SEM-BSE images of the T0 sample.Quartz particles have an intense green aspect due to oxygen and silicon content while kaolinite and mullite appear as pale green due to the supplementary aluminum content.The particle shape makes a difference between kaolinite and mullite in the elemental map: mullite appears as rounded pale green particles with diameters in the range of 10-100 µm in good agreement with MOM investigation while kaolinite is mostly represented by fine grains surrounding bigger particles.Figure 6a also reveals some brown spots concentrated in randomly positioned areas indicating the lepidocrocite occurrences.Ca is labeled with blue tones and is related with glass particles introduced in the ceramic paste as a sintering mediator during tile firing.Na presence is also related to the glass particles (chemical formula 6SiO 2 •CaO•Na 2 O); thus, the pink dots are partly associated with the blue glass particles but also appear as individually spread spots that are most likely found as traces from the alkaline dispersion in the technological used water as previously observed [19].Overall, the elemental map of T0 sample sustains its heterogeneous distribution of the minerals.
The elemental map of the T3 sample reveals the microstructural uniformity of the milled mixture and advanced conditioning of the kaolinite fine particles network, which became a proper matrix that sustains the material cohesion.Quartz boulder-like particles are uniformly distributed in the elemental map in Figure 6b colored in intense green tones due to the oxygen (labeled in blue) and silicon (labeled in green).The mullite rounded particles also are uniformly distributed, having a pale green tone due to their aluminum content, besides Si and O.
The bigger kaolinite particles (with large tabular aspect) are completely missing due to their disintegration during the intensive milling at 6000 rpm.Thus, the kaolinite matrix appears pale yellow green because of the advanced refinement of fine particles and their mixing with iron brown spots.The disintegration of lepidocrocite clusters in fine fractions and their uniform distribution within the kaolinite matrix improves the sample homogeneity.It is noteworthy that the iron amount in the conditioned sample T3 remains constant at 0.3% and its re-distribution does not modify its participation in the elemental composition of the sample.
The particle re-arrangements within the T3 sample show the transformation of the pressed slurry into a composite material.This goal was achieved by the proper milling of the mixture in the presence of the alkaline moisture that assures the stability of kaolinite matrix and assures a proper behavior under the compaction load.

Flexural Behavior of the Samples and Fractography
The test sample bars (T series) were subjected to the three-point bending test for assessing their flexural behavior.The test was conducted on three distinct specimens for each sample, and the obtained mean values are centralized in Figure 7; the standard deviation is represented as error bars.

Flexural Behavior of the Samples and Fractography
The test sample bars (T series) were subjected to the three-point bending test for assessing their flexural behavior.The test was conducted on three distinct specimens for each sample, and the obtained mean values are centralized in Figure 7; the standard deviation is represented as error bars.The flexural strength of the test samples, Figure 7a, strongly depends on their preparation and on the subsequent microstructure.A lower flexural strength was obtained on T1 sample, while the most resistant was T3.The statistical analysis shows significant differences for all samples p value being less than 0.05.However, a statistical similarity was observed individually between T0 and T2 (p > 0.05), but the microstructural differences do not allow their association as a relevant statistical group, proving that the flexural strength depends mainly on the conjugated effect of mechanical milling of the mixture and strict pH control.
The bending elasticity module variation, Figure 7b, proves that it is strongly dependent on the milling speed and microstructural refinement.Both samples milled at 6000 rpm present a high bending modulus, forming a relevant statistical group (p > 0.05).The T1 sample milled at 2000 rpm has lower bending elasticity, while T0 has an intermediary value with significant statistical differences (p < 0.05).
The bar's deflection, Figure 7c, on the bending load is relatively similar for all tested samples and has no correlation with the microstructural aspects and conditioning parameters.It, rather, can be correlated with the fractography aspects in Figure 8.The fracture surface was investigated with SEM.
The fracture surface of the T0 sample is very corrugated due to the inter-granular failure promoted within the kaolinite binder, Figure 8a.The presence of bigger kaolinite particles with large tabular shapes acts as tension concentrators that tend to section the binding layer over the tangential solicitation induced by the bending load.This fact reflects a horizontally oriented compression effort in the upper layers of the bar and a tensile effort on the lower side of the bar.This asymmetric solicitation pushes the bigger kaolinite particles onto the fine binder layer and causes its failure.It is sustained by the development of collateral cracks propagated from the main fracture surface within the sample's bulk.Such a fracture is clearly observed in the high-magnification detail in Figure 8a'.The fine kaolinite particles rupture can be observed on the crack sides detaching bigger quartz and mullite particles.
The failure mechanism in the T1 sample is different because of the microstructural properties.The asymmetric horizontal tensions induced by the main bending effort (compression in the upper layers and tensile in the lower layers of the bar) act as tensors on the cohesion necks between the microstructural pellets, Figure 8b.Thus, the failure is promoted by crack appearances on the lower side of the bar that propagate on the upper layer without the development of collateral cracks.The adhesion neck failure is inter-granular The flexural strength of the test samples, Figure 7a, strongly depends on their preparation and on the subsequent microstructure.A lower flexural strength was obtained on T1 sample, while the most resistant was T3.The statistical analysis shows significant differences for all samples p value being less than 0.05.However, a statistical similarity was observed individually between T0 and T2 (p > 0.05), but the microstructural differences do not allow their association as a relevant statistical group, proving that the flexural strength depends mainly on the conjugated effect of mechanical milling of the mixture and strict pH control.
The bending elasticity module variation, Figure 7b, proves that it is strongly dependent on the milling speed and microstructural refinement.Both samples milled at 6000 rpm present a high bending modulus, forming a relevant statistical group (p > 0.05).The T1 sample milled at 2000 rpm has lower bending elasticity, while T0 has an intermediary value with significant statistical differences (p < 0.05).
The bar's deflection, Figure 7c, on the bending load is relatively similar for all tested samples and has no correlation with the microstructural aspects and conditioning parameters.It, rather, can be correlated with the fractography aspects in Figure 8.The fracture surface was investigated with SEM.The fracture surface of sample T2 has a compact surface free of collateral cracks, proving the increased homogeneity of the material after milling at 6000 rpm, Figure 8c.The advanced mechanical conditioning associated with a molding load of 350 N assures a compact structure without internal pores, explaining the increase in flexural strength.The general aspect is slightly grainy with a uniform distribution due to the bigger particles like quartz and mullite being immobilized in the kaolinite matrix.Its fine particles are The fracture surface of the T0 sample is very corrugated due to the inter-granular failure promoted within the kaolinite binder, Figure 8a.The presence of bigger kaolinite particles with large tabular shapes acts as tension concentrators that tend to section the binding layer over the tangential solicitation induced by the bending load.This fact reflects a horizontally oriented compression effort in the upper layers of the bar and a tensile effort on the lower side of the bar.This asymmetric solicitation pushes the bigger kaolinite particles onto the fine binder layer and causes its failure.It is sustained by the development of collateral cracks propagated from the main fracture surface within the sample's bulk.Such a fracture is clearly observed in the high-magnification detail in Figure 8a'.The fine kaolinite particles rupture can be observed on the crack sides detaching bigger quartz and mullite particles.
The failure mechanism in the T1 sample is different because of the microstructural properties.The asymmetric horizontal tensions induced by the main bending effort (compression in the upper layers and tensile in the lower layers of the bar) act as tensors on the cohesion necks between the microstructural pellets, Figure 8b.Thus, the failure is promoted by crack appearances on the lower side of the bar that propagate on the upper layer without the development of collateral cracks.The adhesion neck failure is inter-granular due to the fine kaolinite particles plucking off from the binding layer along with the tensile effort.
The fracture surface of sample T2 has a compact surface free of collateral cracks, proving the increased homogeneity of the material after milling at 6000 rpm, Figure 8c.The advanced mechanical conditioning associated with a molding load of 350 N assures a compact structure without internal pores, explaining the increase in flexural strength.The general aspect is slightly grainy with a uniform distribution due to the bigger particles like quartz and mullite being immobilized in the kaolinite matrix.Its fine particles are uniformly distributed as observed in the high-magnification detail, Figure 8c', and assure an advanced binding of the quartz and mullite filler particles.Thus, the bending solicitation compresses the upper side of the microstructure, which resists the solicitation well, while the lower layers are subjected to the horizontal tensile solicitation, which stresses the fine kaolinite binder cohesion.Increasing the solicitation, the fine binder particles are finally split via inter-granular failure, explaining the plucked aspect of the fine microstructure in Figure 8c'.The initiated crack is faster propagated on the bending solicitation direction, generating a brittle failure of the specimen.This fact is sustained by the lower bar deflection, Figure 7c.
The T3 sample presents a very smooth and uniform fracture surface without collateral cracks, Figure 8d.The general grainy aspect is mostly attenuated due to the advanced refining of the kaolinite matrix, which embeds better the bigger quartz and mullite particles.These mineral filler particles are hard to observe in the high-magnification detail in Figure 8d' due to their optimal coverage with ultra-fine kaolinite particles that assures an advanced compaction between the matrix and granular material.The bending solicitation compresses the upper layers of the bar, which resist well under horizontal stress and the lower layers are subjected to horizontal tensile solicitation.It requires a relatively high stress to generate kaolinite binding layer failure.The plucked kaolinite particles are finer than observed on T2 and the microstructure is more compact.Thus, the failure brittleness is increased, and the crack is faster propagated from the specimen bottom to the top in the direction of the bending solicitation.The bar deflection is significantly increased compared to the T2 sample due to the optimal binding assured by the properly conditioned kaolinite matrix.
The flexural behavior of the test samples (T series) shows that the kaolinite matrix conditioning parameters for T3 are optimal for the raw pressed slurry sample preparation.The experiment was further continued using the raw slurry prepared in the T3 conditions and the bar specimens were molded at different pressing loads to observe the microstructure compaction on the flexural strength.Therefore, we prepared three specimens for each sample at different loads: 70, 140, 210 and 350 N, which were subjected to the three-point bending test (S series).The obtained results are presented in Figure 9.
conditioning parameters for T3 are optimal for the raw pressed slurry sample preparation.The experiment was further continued using the raw slurry prepared in the T3 conditions and the bar specimens were molded at different pressing loads to observe the microstructure compaction on the flexural strength.Therefore, we prepared three specimens for each sample at different loads: 70, 140, 210 and 350 N, which were subjected to the three-point bending test (S series).The obtained results are presented in Figure 9.The S series' flexural strength strongly depends on the compaction load as observed in Figure 9a.A linear increase in the flexural strength along with the pressing load.The S1 sample has the minimum mean value of the flexural strength of about 0.56 MPa while S4 has the maximum mean value situated at 1.17 MPa.The bending Youngʹs modulus presents an exponential decrease with the compaction load, Figure 9b.The S1 sample exhibits the higher bending elastic modulus (mean value 275.96MPa) due to the relative flexible bonding within the kaolinite matrix particles.It decreases exponentially due to the kaolinite matrix binding enhancement as a consequence of pressing compaction.Thus, S4 has the lower bending elastic modulus, having a mean value of about 137.05 MPa.The bar deflection under bending effort has a linear decrease tendency, proving that higher compaction grade generates a stiffer material.
The S series sample fractography was also investigated with SEM, and the obtained images are presented in Figure 10.The lower pressing load, S1 sample, generates compaction deficiencies such as a combination of rounded pores with diameters in the range of 50-250 µm and large elongated pores of 2 mm length and 250 µm widths, Figure 10a.The high-magnification detail in Figure 10a' reveals a smooth surface of the elongated pore, proving its formation as an air inclusion in the ceramic paste that was prolonged during the molding process.Unfortunately, the compaction load was not enough to eliminate these pores.The mixture presents a kaolinite matrix with relatively loose particles that assure a relatively poor binding.The pores act as failure promoters under bending efforts, causing progressive de-structuring of the specimen.
The fracture surface of the S2 sample presents only smaller pores with rounded shapes in the range of 50-250 µm, Figure 10b.The bending effort induces local stress within these pores that acts as local tension concentrators that initiate failure on the bending direction.This is proven by the surface irregularities that start from the pore margins and are propagated in the material bulk.The fine microstructure aspect is better observed at high magnification in Figure 10b', revealing the better compaction of the fine kaolinite particles within the matrix.The S series' flexural strength strongly depends on the compaction load as observed in Figure 9a.A linear increase in the flexural strength along with the pressing load.The S1 sample has the minimum mean value of the flexural strength of about 0.56 MPa while S4 has the maximum mean value situated at 1.17 MPa.The bending Young's modulus presents an exponential decrease with the compaction load, Figure 9b.The S1 sample exhibits the higher bending elastic modulus (mean value 275.96MPa) due to the relative flexible bonding within the kaolinite matrix particles.It decreases exponentially due to the kaolinite matrix binding enhancement as a consequence of pressing compaction.Thus, S4 has the lower bending elastic modulus, having a mean value of about 137.05 MPa.The bar deflection under bending effort has a linear decrease tendency, proving that higher compaction grade generates a stiffer material.
The S series sample fractography was also investigated with SEM, and the obtained images are presented in Figure 10.The lower pressing load, S1 sample, generates compaction deficiencies such as a combination of rounded pores with diameters in the range of 50-250 µm and large elongated pores of 2 mm length and 250 µm widths, Figure 10a.The high-magnification detail in Figure 10a' reveals a smooth surface of the elongated pore, proving its formation as an air inclusion in the ceramic paste that was prolonged during the molding process.Unfortunately, the compaction load was not enough to eliminate these pores.The mixture presents a kaolinite matrix with relatively loose particles that assure a relatively poor binding.The pores act as failure promoters under bending efforts, causing progressive de-structuring of the specimen.
The fracture surface of the S2 sample presents only smaller pores with rounded shapes in the range of 50-250 µm, Figure 10b.The bending effort induces local stress within these pores that acts as local tension concentrators that initiate failure on the bending direction.This is proven by the surface irregularities that start from the pore margins and are propagated in the material bulk.The fine microstructure aspect is better observed at high magnification in Figure 10b', revealing the better compaction of the fine kaolinite particles within the matrix.
The pressed slurry pores completely disappear after compaction at the load of 210 N as seen in the fracture surface of the sample S3, Figure 10c.It is compact and smooth, and only some relative bumps occur.The high-magnification detail in Figure 10c' reveals that the bumps are some quartz particles very well coated with a dense kaolinite matrix layer proving the enhanced binding.Therefore, the upper layers of the bar subjected to local horizontal compression solicitation resist very well.The elongation solicitation induced in the lower layers of the bar causes the kaolinite matrix failure by inter-plucking of the fine kaolinite particles.Once the crack is initiated, it is faster propagated to the upper layers, causing bar breaking.The advanced compaction of the kaolinite matrix on the filler particles assures an increased flexural strength but a rather brittle failure.Increasing the compaction load to 350 N, the kaolinite matrix is conditioned to assure optimal binding, and the microstructure and failure mode observed in Figure 10d for the S4 sample are like those observed for the T3 sample and have the same behavior.The pressed slurry pores completely disappear after compaction at the load of 210 N as seen in the fracture surface of the sample S3, Figure 10c.It is compact and smooth, and only some relative bumps occur.The high-magnification detail in Figure 10c' reveals that the bumps are some quartz particles very well coated with a dense kaolinite matrix layer proving the enhanced binding.Therefore, the upper layers of the bar subjected to local horizontal compression solicitation resist very well.The elongation solicitation induced in the lower layers of the bar causes the kaolinite matrix failure by inter-plucking of the fine kaolinite particles.Once the crack is initiated, it is faster propagated to the upper layers, causing bar breaking.The advanced compaction of the kaolinite matrix on the filler particles assures an increased flexural strength but a rather brittle failure.Increasing the compaction load to 350 N, the kaolinite matrix is conditioned to assure optimal binding, and the microstructure and failure mode observed in Figure 10d for the S4 sample are like those observed for the T3 sample and have the same behavior.

Ultra-Structural Aspects
The ultra-structural aspects of the kaolinite matrix binding on the quartz and mullite filler particles were followed closely by atomic force microscopy (AFM).The effect of the compaction load on the mixture conditioned at 6000 rpm and pH = 10 was investigated, Figure 11.
The topography of the S1 sample reveals some sub-micron particles with boulderlike shapes belonging to quartz having a size range between 200 and 800 nm partly covered with randomly oriented kaolinite nano-platelets in the range of 40-60 nm, Figure 11a.These are in good agreement with the average crystallite size determined by the Scherrer formula from the XRD pattern in Figure 2a.The diffuse positioning of kaolinite matrix ultra-structural elements partly generates the binding lack and influences the flexural behavior of the specimen.However, this loose distribution of the ultra-structure leads to a relatively smooth surface.The submicron particle unevenness is covered by the randomly oriented kaolinite nanoparticles reducing the local asperities.
Increasing the compaction load at 140 N in sample S2 begins to generate kaolinite nano particles that are textured over the quartz and mullite submicron particles, Figure 11b.They are more evident on the surface topography, some of them being attached with the lateral side to the submicron feature.Their diameter is around 50 nm.This not only

Ultra-Structural Aspects
The ultra-structural aspects of the kaolinite matrix binding on the quartz and mullite filler particles were followed closely by atomic force microscopy (AFM).The effect of the compaction load on the mixture conditioned at 6000 rpm and pH = 10 was investigated, Figure 11.The submicron features within S3 sample topography are more evident, Figure 11c, due to the organized disposal mode of the kaolinite nanoparticles.They are about 40-50 nm and are very well attached to the quartz particle surfaces in an ordered manner like fish scales.This nanoparticles disposal increases the cohesion of the kaolinite matrix and assures the pressed slurry cohesion and subsequently helps increase the flexural strength.The intensive attachment of the matrix to the filler particles leads to the increasing of the surface corrugation as observed in the three-dimensional profile presented below the topographic image in Figure 11c.
The compaction load of 350 N used for the S4 sample determines a more enhanced adhesion of the kaolinite nanoparticles to the quartz and mullite submicron particles that The topography of the S1 sample reveals some sub-micron particles with boulder-like shapes belonging to quartz having a size range between 200 and 800 nm partly covered with randomly oriented kaolinite nano-platelets in the range of 40-60 nm, Figure 11a.These are in good agreement with the average crystallite size determined by the Scherrer formula from the XRD pattern in Figure 2a.The diffuse positioning of kaolinite matrix ultra-structural elements partly generates the binding lack and influences the flexural behavior of the specimen.However, this loose distribution of the ultra-structure leads to a relatively smooth surface.The submicron particle unevenness is covered by the randomly oriented kaolinite nanoparticles reducing the local asperities.
Increasing the compaction load at 140 N in sample S2 begins to generate kaolinite nano particles that are textured over the quartz and mullite submicron particles, Figure 11b.They are more evident on the surface topography, some of them being attached with the lateral side to the submicron feature.Their diameter is around 50 nm.This not only certainly increases the surface asperities, but also facilitates the binding ability of the matrix, a fact that explains the flexural strength increase.
The submicron features within S3 sample topography are more evident, Figure 11c, due to the organized disposal mode of the kaolinite nanoparticles.They are about 40-50 nm and are very well attached to the quartz particle surfaces in an ordered manner like fish scales.This nanoparticles disposal increases the cohesion of the kaolinite matrix and assures the pressed slurry cohesion and subsequently helps increase the flexural strength.The intensive attachment of the matrix to the filler particles leads to the increasing of the surface corrugation as observed in the three-dimensional profile presented below the topographic image in Figure 11c.
The compaction load of 350 N used for the S4 sample determines a more enhanced adhesion of the kaolinite nanoparticles to the quartz and mullite submicron particles that become almost the same topographical feature.The kaolinite particles are very well attached to the surface but are still visible; in Figure 11d, they have a tabular shape and diameter of about 55 nm, in good agreement with the XRD observation for the pattern in Figure 2b.

Discussion
Ceramic tiles are used all over the world for floor and wall pavement and are produced in the hundred billions of tons per year [27].Regarding quality and quantity, these aspects peaked in 2022, with about 12.19 billion square meters of ceramic tile production reported from Asia; consistent production was reported from the European Union and United States [28].This huge annual production implies significant amounts of ceramic slurry recovered from wastewater that cannot be re-circulated into the technological process due to iron hydroxides content that strains the final glaze.Otherwise, it is a very balanced material, containing well-selected and processed minerals that are worth being utilized in some useful applications instead of being dumped.The dumped wastewater ceramic slurry might release very dangerous atmospheric emissions such as PM1, PM2.5 and PM10 if they are not properly wetted and supervised [29,30].Thus, dumping is costly and potentially dangerous for the environment; in fact, we are pleading once more for re-utilization of certain manufacturing sub-products or other applications that might be identified by a close investigation of the wastewater ceramic slurry composition and its behavior.
One of our preliminary studies was focused on slurry formation by mineral particulate matter sedimentation from wastewater [19], revealing that quartz is generally the dominant mineral, followed by kaolinite and mullite, followed by moderate traces of lepidocrocite (iron hydroxide).We found that the mineral proportion strongly depends on the slurry humidity and sedimentation grade.Some local appreciation of the kaolinite amount regarding the quartz proportion was noticed [19].The general microstructural aspects and compression strength of the raw pressed samples were also discussed in another one of our previous studies [16], revealing a better consistency of the mineral composition of the slurry in the solid state (e.g., dense paste with relative low humidity), the dominant mineral being quartz followed by kaolinite and mullite and a low amount of lepidocrocite.Our preliminary results show that kaolinite presents both bigger particles and very small ones, even nanoparticles in the range of 40-60 nm [16,19], that are responsible for the microstructural cohesion.Thus, we wonder if such a mixture might be conditioned as a composite matrix to enhance the filler particles of quartz and mullite and to follow the conditioning effect on the flexural behavior of the raw pressed samples.
The data in the literature show that the kaolinite crystal structure is sensitive to the presence of water because of H 2 O molecule penetration within the pseudo-hexagonal sheets of silica tetrahedrons [3,31].It causes interplanar distance to increase.Our performed test shows that the interplanar distance of the (−112) crystallographic plane linearly increases with the water moisture of the slurry from 304.22 pm at 20% to 305.25 pm at 40%.It further shows that the kaolinite binding ability for quartz particles increases with the moisture level until its liquefaction, implying liquid dispersion of the constituent particles [32].Thus, the proper adjustment of the ceramic slurry moisture would help the proper conditioning of kaolinite particles to become a structural matrix.The MOM investigation shows that moisture of about 40% water causes particle dispersion like that observed in Figure 3c, which would be counterproductive, besides the large increase of the kaolinite interplanar distance.Therefore, we aimed to increase the slurry moisture by around 30%; the effective measurement performed after water addition followed by homogenization was 33%.
The mineral composition of the initial slurry is dominated by quartz (40% wt.) followed by kaolinite (32% wt.) and mullite (22% wt.) and small traces of lepidocrocite (6% wt.), which corresponds to a filler/matrix ratio of 2.12.The composite approaching regarding the ceramic slurry supposed that quartz and mullite play the filler particles role because of their predominant size range of about 30-100 µm (fewer fractions between 10-30 µm are observed).Kaolinite is a natural particle binder if it is properly moistened, taking the matrix role.Unfortunately, the slurry as received, the T0 sample, reveals a significant amount of bigger kaolinite particles that have less binding behavior but act as tension concentrators under bending stress, causing poor flexural strength.Their size must be reduced by proper milling.The data in the literature indicate that the bigger kaolinite particles are weak under the mechanical shock oriented on the sheet folds, causing their cleavage in thin and thinner slices that subsequently might be broken in very fine fractions [5].The properly moistened slurry was milled in a blade mill under the conditions described in Table 1.Quartz and mullite particles act as milling bodies along with the mill blades, causing intense fragmentation of the bigger kaolinite particles.Milling at 2000 rpm enhances the kaolinite matrix microstructure by refining the particle size but does not achieve the optimum as observed in Figure 5b.Increasing the milling speed at 6000 rpm the kaolinite particles is significantly enhanced as observed in Figure 5c,d.The upper limit of the kaolinite particle size decreases from 80 µm to 20 µm; this case is isolated because most of the particles are situated below 5 µm.A lot of kaolinite nanoparticles are formed due to the intensive fragmentation during milling at 6000 rpm, sustained by the crystallite size of 55 nm determined from the XRD patterns using the Scherer formula.These nanoparticles were further confirmed by the ultra -structure investigation performed with AFM.
The fine particle re-arrangements after the intensive milling of T2 and T3 samples induce a relative increase in kaolinite amount at 36% and incorporate the finest fractions resulting from lepidocrocite cluster disaggregation as observed in the elemental map in Figure 6b, in good agreement with observations in the literature [33,34].Thus, the filler/matrix ratio decreases to 1.27, assuring an optimal level for the microstructure binding.SEM observation revealed a better refinement of the microstructure of the T3 sample regarding T2 due to the alkaline moisture based on the small sodium content induced by the technological water moisture added to the ceramic paste.The data in the literature confirm that a high pH value stabilizes fine kaolinite particle dispersion and facilitates their reticulation within the solid phase [35,36].This fact helps achieve a better refined and compact fine microstructure in the T3 sample compared to T2.Adjusting the moisture and milling speed, the samples in the T series were prepared.The results of neutral pH moisture combined with low milling speed, T1, are a pelletized microstructure with the lowest flexural strength even compared with un-conditioned slurry, T0.The usage of high milling speed assures good refinement of the kaolinite matrix; it increases the flexural strength of the pressed bars.Increasing the moisture pH to 10 (by the addition of technological water) ensures optimal conditioning of the kaolinite matrix, resulting in the greatest value of the flexural strength of the pressed bar.
The compaction load influence of the flexural strength was followed on the slurry with kaolinite matrix conditioned under optimal conditions 6000 rpm and 33% moisture with pH = 10.The lower compaction load in the S1 and S2 samples generates pores and molding faults that facilitate the failure propagation through the bending solicitation direction.The pores and molding faults were eliminated by increasing the compaction loads to 210 N and 350 N, respectively, assuring a uniform and compact microstructure.On one hand, the lack of kaolinite particle orientation within the S1 and S2 samples affects the filler particle binding, but on the other hand, it assures a more elastic structure that has an increased bending Young's modulus and significantly higher bending deflection.The increased compaction load forces the wetted fine kaolinite particles to stick on the filler particles surfaces (e.g., quartz and mullite), which remain optimally consolidated after the samples' drying.The AFM investigation in Figure 11 evidences the kaolinite nanoparticles sticking on the filler particles as a consequence of compaction load.The advanced compaction of the kaolinite matrix presses the matrix on the filler particles, assuring a structural rigidity that significantly reduces the bending elastic modulus and sample deflection.Thus, the S3 and S4 samples' failure surface is smooth and free of collateral cracks assuring increased values of flexural strength.The greatest value was obtained for the S4 sample due to the optimal conditioning of the kaolinite matrix.The obtained values are in good agreement with the literature regarding the flexural strength obtained for the raw compacts of kaolinite composites [37].However, it might be increased by the addition of glass fibers [37] or short vegetal fibers to achieve a better reticulation thorough the composite [38].This additional reinforcement might be useful for developing ecological bricks using low thermal treatment or for chemical activation via geopolymer consolidation.The flexural strength obtained for S4 is compatible with the compressive strength achieved at the same compaction load observed in our previous study [16] that makes our composite material suitable as a plaster for ecological houses that are desired to minimize the carbon footprint.This is the path for our further research regarding the rich kaolinite matrix composite containing quartz and mullite particles as filler.
The observed behavior of the samples investigated in this study strongly depends on the mineralogical composition of the used ceramic slurry.Thus, the established optimal conditioning parameters of the kaolinite matrix could be generalized for the ceramic slurries containing quartz, kaolinite and mullite.It is difficult to predict the behavior of the other ceramic slurries that might contain other minerals or additives.However, kaolinite particle fragmentation through attrition will increase its binding ability regardless of filler admixture composition.An optimal moisture of around 33% is recommended for all ceramic composite pastes that use kaolinite as a matrix.

Conclusions
The investigated raw pressed slurry collected from the ceramic tile production technological wastewater has a stable mineral composition dominated by quartz (40% wt.) followed by kaolinite (32% wt.) and mullite (22% wt.) and small traces of lepidocrocite (6% wt.) that correspond to a filler/matrix ratio of 2.12.The optimal conditioning of the kaolinite matrix increases its amount at 36% wt.It induces a filler/matrix ratio of 1.27 due to the association with the fine dispersion of the lepidocrocite clusters.The lower milling speed (2000 rpm) associated with the neutral moisture generates a pelletized microstructure exhibiting the lowest flexural strength.The high milling speed (6000 rpm) associated with the alkaline pH assures the best conditioning of the kaolinite matrix, which induces the proper reticulation of the composite material, sustaining well the quartz and mullite particles ensuring the higher flexural strength.The bending elastic modulus and the samples bending deflection decrease with the flexural strength increasing due to microstructural rigidity.The advanced conditioning of the kaolinite matrix allows optimal positioning of the ultra-structural nanoparticles regarding the quartz and mullite surface that hold these

Figure 1 .
Figure 1.Sample physical characteristics: (a) sample design and (b) resulting bar.

Figure 1 .
Figure 1.Sample physical characteristics: (a) sample design and (b) resulting bar.

Figure 4 .
Figure 4. XRD kaolinite interplanar distance variation with the sample humidity: (a) detailed XRD pattern for the kaolinite (−112) peak PDF 02-0105 and (b) interplanar distance variation with humidity and its linear fit (the red line drawn between the interplanar distance markers).

Figure 4 .
Figure 4. XRD kaolinite interplanar distance variation with the sample humidity: (a) detailed XRD pattern for the kaolinite (−112) peak PDF 02-0105 and (b) interplanar distance variation with humidity and its linear fit (the red line drawn between the interplanar distance markers).

Figure 6 .
Figure 6.SEM images and corresponding elemental analysis for the samples: (a) T0 and (b) T3.The EDS elemental map and spectrum are given on the right side of each SEM image.

Figure 6 .
Figure 6.SEM images and corresponding elemental analysis for the samples: (a) T0 and (b) T3.The EDS elemental map and spectrum are given on the right side of each SEM image.

J
. Compos.Sci.2024, 8, x FOR PEER REVIEW 12 of 20 due to the fine kaolinite particles plucking off from the binding layer along with the tensile effort.

Figure 9 .
Figure 9. Mechanical properties for the S series: (a) flexural strength, (b) bending Young modulus, (c) bending deflection.The error bars represent the standard deviation, and the red lines are the variation fit.

Figure 9 .
Figure 9. Mechanical properties for the S series: (a) flexural strength, (b) bending Young modulus, (c) bending deflection.The error bars represent the standard deviation, and the red lines are the variation fit.

J
. Compos.Sci.2024, 8, x FOR PEER REVIEW 15 of 20 certainly increases the surface asperities, but also facilitates the binding ability of the matrix, a fact that explains the flexural strength increase.

Table 1 .
Sample preparation and matrix conditioning parameters.

Table 2 .
Mineral composition of ceramic slurry samples.

Table 2 .
Mineral composition of ceramic slurry samples.

Table 3 .
Elemental composition of ceramic slurry samples.