Optimization of Mullite-Based Chamotte Production as a Result of Calcination of a Kaolin-Boron Waste Mixture by Response Surface Method and its Usability in Ceramic

This study aims to use kaolin and boron waste in the production of chamotte material. The usability of the produced chamotte in the FFC (fine fire clay) body in sanitary wares, instead of the chamotte supplied from abroad, was also investigated. The studies were carried out using a statistical experimental design. The percentage of boron waste, calcination temperature and grinding time were examined as independent variables, and the effects of these factors on the percentage of mullite phase were examined. The interaction of the parameters both individually and with each other was revealed, and the process was modeled with a second-order mathematical model that is a function of independent variables. With an optimization study that provides the desired product feature with the least possible cost, the properties of the vitrified product obtained with both the produced chamotte and the reference chamotte were compared. When the chamotte, produced by optimizing the experimental design, was added to the vitrified body, this product achieved an improvement of 32.73% in strength, 5.88% in shrinkage, and 4.67% in water absorption compared to the reference factory chamotte. As a result, with the current study, it has been determined that when both kaolin and boron waste are calcined together, fireclay with better properties can be obtained instead of imported fireclay and can be used industrially.


Introduction
Turkey is an important producer and exporter of ceramic sanitary ware in Europe. It produces nearly 10% of the world's production [1]. The main products of ceramic sanitary ware are sinks, shower trays, toilet bowls, and urinals. Two different slip is used for the products produced in the ceramic sanitary ware sector. These slips are vitreous china (VC) and fine fire clay (FFC) slips. FFC is used in the production of large-size products in ceramic sanitary ware products, while vitreous china is used in the production of other products such as small sinks and toilet bowls [2][3][4]. In recent years, it is seen that in today's modern life, vitrified products have become an important visual element in the bathroom, and as a result, design studies have been given importance in the sector [4]. For this reason, the demand for special design products in bathrooms is gradually increasing and large-sized, straight and sharp-line products are generally preferred. In the production of such large-sized products, FFC slips with a high content of fired clay are used due to their low deformation characteristics. The main raw materials used in the production of FFC are clay, kaolin, quartz, and chamotte [4][5][6].
Chamotte is the fired state of clay until it loses its binding properties [2]. The FFC body in ceramic sanitary ware production typically contains 30% chamotte material [7]. Kaolin is calcined and used as fireclay, especially since its firing color is white. In hard and soft porcelain as well as bone porcelain, kaolin's calcination behavior is crucial. [5]. Since chamotte is a fired raw material, it provides strength to the product and minimizes deformation when it is in semi-product and fired products in FFC bodies. In addition to these important features provided by chamotte material, Turkey is entirely dependent on foreign suppliers for the supply of chamotte material due to the high investment requirements of the kaolin enrichment and grinding processes. The use of exported chamotte also increases the production costs of ceramic sanitary ware products and causes economic instability [6,7]. Therefore, the purpose of this research is to obtain low-cost chamotte by synthesizing kaolin and boron wastes as an alternative to the exported chamotte material and to investigate its usability in the sector.
The assessment of boron waste in ceramic bodies has been the subject of researchers. In a study on adding boron waste to floor tile masses, it was stated that products with low porosity, high strength values and low water absorption were obtained [8]. Sağlam and Emrullahoğlu [9] investigated the usability of boron waste in hard porcelain as a result of calcination by mixing it with various raw materials. When the experimental results were examined according to the reference sample, it was seen that results close to the reference sample were obtained by using chamottecontaining boron waste. Recently, studies have been carried out using boron mineral wastes in the composition of tiles to reduce the sintering temperature [10][11][12]. Boron waste helps to minimize energy consumption for tile production.
Çiçek et al., [11], in their study, determined that the sintering temperature of samples containing 5 to 6% by weight boron wastes decreased by 60 to 70°C. The use of boron tailing in the production of porcelain stoneware tiles was investigated by Karadagli and Cicek [12]. They added 3-10 wt.% of tailings to formula that was used to make ceramic tiles and the sintering temperature was lowered by 38°C.
Mullite is formed by heating kaolin-type minerals and is obtained as a result of the reaction of alumina-silica mixtures [13]. Michel et al., [14] in their research on aluminasilica systems, determined that the mullite phase provides strength in the ceramic body. In his study on ceramic mullites, Schneider [15] stated that mullite can be used in hightemperature applications with its properties such as low thermal expansion, high melting point, good frictional strength. When studies on the use of chamotte and/or mullite-based chamotte in the FFC body are examined, it is understood that the mullite phase positively affects the properties of the ceramic body. Kong et al. [16] obtained the mullite phase from an alumina-silica mixture at low temperatures such as 1200°C by high-energy grinding. Although the mullite phase is often formed at temperatures above 1500°C, it has been revealed that the grinding effect and the influence of impurities may produce the mullite phase at low-temperature values.
The world's kaolin reserves are rapidly depleting due to the demand for ceramic building materials. Hence, researches have been conducted to examine the production of mullite from low-cost kaolin wastes [17][18][19][20][21][22][23]. The use of kaolin waste as an alternative raw material for the manufacture of mullite ceramics was studied by Brasileiro et al. [18]. According to the authors, the mixtures heated above 1500°C only produced mullite and quartz or mullite and alumina as the only crystalline phases, depending on their original composition. While samples with greater alumina contents displayed higher bulk density values, those with a higher proportion of kaolin waste showed lower water absorption values. Menezes et al. [19] investigated the use of kaolin waste as an alternative raw material for the production of dense mullite bodies and ceramic tiles. The authors demonstrated that kaolin waste can be used to make ceramic tiles.
In recent years, the use of low-cost alternative raw materials or wastes in the production of ceramic sanitary ware has gained importance [13][14][15][16][17]. However, there are limited research articles using different wastes as alternative raw materials in sanitary ware production such as stone cutting waste [24], wall tile waste [25], galvanized waste [26], waste glass [27] and quartzite waste [28]. In these researchers, it has been revealed that improvements in physicomechanical properties (strength, water absorption, firing temperature, etc.) of new sanitary ware products with waste addition are generally achieved. Additionally, no research has been conducted on the use of kaolin and boron wastes together as raw materials in the production of sanitary ware. Therefore, the study is significant in terms of recycling kaolin and boron wastes and lowering ceramic sanitary ware production costs by obtaining low-cost chamotte from wastes through optimization.
In this study, the production of mullite-based chamotte as a result of the calcination of kaolin and boron waste instead of imported chamotte used in the production of sanitary ware products was investigated. Boron waste was added to reduce the calcination temperature of kaolin and to support mullite formation. The obtained kaolin and boron waste chamotte were used in the FFC body. The interactions of the parameters affecting the synthesis method, both individually and with each other, were determined by statistical experimental design. The process is modeled with a second order mathematical model that is a function of independent variables. In addition, the properties of the vitrified product produced with the reference chamotte and the chamotte produced with the optimization study were compared.

Materials and experimental procedures
Boron and kaolin waste materials used in recipe composition were supplied by Ilksem Mining. For the characterization of raw materials, samples of approximately 50 g were prepared by applying sample reduction methods and their chemical analyzes were carried out using an XRF device. Analysis results are given in Table 1.
The DTA-TG graph of kaolin used as the main raw material in the study is given in Fig. 1, and the image analysis made by scanning electron microscope is given in Fig. 2. In Fig. 1, it is seen that endothermic reactions occur in the kaolin sample due to the crystal water moving away from the body of the kaolin at 553°C. A second endothermic peak is seen at 749°C. This peak is due to the decay of other carbonate and sulfated compounds contained in kaolin. Other peaks of kaolin are exothermic. Among them, the first exothermic peak is caused by the formation of the primary mullite phase and is observed to occur at 1037°C in all three samples. The second exothermic peak resulting from the secondary mullite phase is observed at a temperature of 1139°C. In the SEM examinations in Fig. 2, images of typical kaolinite morphology were obtained. It is seen that kaolinite crystals are irregularly scattered and unrelated to each other. The kaolinite particles present themselves as hexagonal or euhedral crystals whose size varies between 0.5 and 3 μm. The crystals are typically thin and flexible plates.
In the study, the workflow chart in which the calcination method and mullite-based chamotte are used in the FFC body is given in Fig. 3. In Fig. 3a, boron waste was added to the kaolin raw material in the proportions of 0-50% in accordance with the experimental design in the preparation of chamotte and the percentage of the amount of kaolin + boron waste was ensured to be 100%. The prepared mixture was ground in a wet ball mill between 30-120 min in accordance with the experimental design. After grinding, the mixture was dried and granulated by passing through a 1 mm sieve. The prepared mixtures were placed in crucibles and calcined at 700-1200°C. Phase analysis of the fireclay samples formed as a result of calcination was performed with the XRD device. The resulting  phases were evaluated by Richter analysis and their % distribution was determined. As a result of the optimization of the experimental design results, the mixture in which the maximum mullite phase is obtained at the lowest temperature was selected for use in a vitrified body.
In the second stage of the study, the fireclay, which was found suitable in the first stage, was used in the FFC body recipe. Recipe compositions were created by mixing FFC raw materials and semi-finished chamotte in different proportions (Fig. 3b). The materials weighed in accordance with the recipe compositions were dispersed in the mechanical mixer by adding water and casting sludge was obtained. After the appropriate sieving process, the sludge remaining under the sieve is poured into plaster molds and shaping is achieved. The shaped samples were sintered at temperatures ranging from 1150°C to 1220°C after drying at room temperature and then in an oven at 100°C.
The strength, deformation, shrinkage and water absorption values of the sintered samples were determined.
An experimental design was made by using the Central Composite Design (CCD) method, which is one of the statistical experimental design methods, to reduce the number of experiments and determine recipe compositions. With the experiments to be carried out according to this design, the individual and interaction of the parameters on the dependent variable are determined. In addition, what are the most effective parameters and the order of influence of the parameters, as well as a second or if necessary, a third-order statistical model that can predict dependent variables are revealed. The percentage of boron waste amount, calcination temperature and grinding time factors were examined as independent variables. The applied CCD design and the level values of the selected variables are given in Table 2.   1 3

Results
The phase percentage values obtained under the experimental conditions according to the three-factor central composite design are presented in Table 3. In the products obtained under each experimental condition, not only the mullite phase but also the other phases formed were evaluated and reflected in Table 3.
As can be seen, in addition to the formation of some mullite phases, anorthite, quartz, sodium aluminum silicate and other phases can also be formed within the limits of the independent parameters tested. Although the aim of this research was mainly concentrated on the mullite phase, the perturbation plots of all detected phases are generated and are shown in Fig. 4 to compare the main effects of tested factors on these phases. In these plots, we can see the individual effects of three factors while keeping the other two factors constant at their middle values. In particular, the presence of mullite and anorthite phases in the chamotte positively affects the product quality. The point to be mentioned here is that while the percentage of the amount of boron waste has a positive effect on the formation of anorthite (Fig. 4b), its effect on the mullite and quartz phases (Figs. 4a and 4c) is negative while no significant effect on NaAl silicate phase (Fig. 4d). Increasing calcination temperature increases the formation of mullite, anorthite phases (Figs. 4a and 4b). In contrast, it has a negative effect on quartz and NaAl phases (Figs. 4c and 4d). The grinding time factor is the least significant effect on all factors while the NaAl silicate phase increases with decreasing the grinding time (Fig. 4d).
XRD analyses of all products obtained in Table 3 were performed and the phase percentages were determined from here. XRD analyses of some examples of how the different conditions tested affect phase formation are given in Figs. 5 and 6 for experiments 11 and 3, respectively. As shown in Fig. 5, the calcination temperature of 700 0 C was not sufficient for the formation of both mullite and anorthite. Only significant quartz phases (56.68%) and sodium aluminum silicate phases (4.71%) have occurred.
According to the XRD graph of the product obtained in experimental conditions no. 3 given in Fig. 6, both the mullite phase (9.66%) and anorthite phase (12.12%) can be obtained if the temperature is increased to 1100 0 C.
Another purpose of applying the experimental design method is to determine the minimum grinding time and the condition or conditions that give the maximum mullite phase at the minimum calcination temperature. At the end of the evaluation of the CCD method applied an optimization study that gives these conditions was also carried out. The composition of the recipe, which was prepared under the condition that gave the most appropriate result, was used within the sanitary ware body, and the result obtained with the vitrified ware of the enterprise was compared.

Model estimating mullite phase and its evaluation
With the CCD method applied in the study, a quadratic mathematical model has been obtained that can predict the mullite phase as a function of the investigated parameters. The mathematical model found in terms of coded units is given below. In Eq. 1, A: The amount of boron waste (%), B: The calcination temperature ( 0 C) and C: The grinding time (minutes) are expressed. Equality 1 has a high coefficient of certainty (R 2 = 0.8816). That is, this model is able  Table 4. In the given ANOVA table, the owner parameters that are statistically important at the 95% confidence level are also indicated (p < 0.05). As can be seen, the model predicting mullite formation is statistically significant (p = 0.0322 < 0.05). Looking at the F probability values, it is seen that the boron waste percentage is an important first and second-order parameter. In addition, it is seen that the first-order effect of the calcination temperature and the bilateral interactions of the boron waste content and the calcination temperature have a statistical significance at the level of 95% in the mullite formation.
As can be seen in Table 4, the order of effectiveness of the parameters in the formation of mullite is the first order of boron waste percentage (A) > second order interaction between boron waste percentage and calcination temperature (AB) > second order of boron waste percentage (A 2 ) > the first order of calcination temperature (B). It is seen that effect of the grinding time parameter is not significant however its second-order effect is important at an 87% confidence level (p = 0.1373).
The graph showing the mullite phase values predicted by the model given in Eq. 1 and the actual mullite phase values, and the normal distribution graph of the residual values with the difference between the estimated and actual values are given in Fig. 7. As can be seen, the line drawn depending on the actual and predicted values is linear (Fig. 7a), and the distribution of residual values shows a normal distribution (Fig. 7b).
As a result of the experiments carried out with the CCD experimental design, the individual effects of boron waste additive percentage, calcination temperature and grinding time on the formation of the mullite phase were determined. The percentage of boron waste content individual effect plot (Fig. 8) is presented for the conditions where  the calcination temperature is at 950 0 C and the grinding time is 75 min. As can be seen, the increasing percentage of boron waste has a negative effect on the formation of the mullite phase, and as the amount of boron waste in the recipe decreases, the formation of the mullite phase increases. The mullite phase is not formed in 40% boron waste addition, and the mullite phase is maximized in 10% boron waste addition. It is also clear from the graph that the effect is not linear, but curvilinear, that is, quadratic. The effect of calcination temperature on the fireclay production recipe is presented in Fig. 9 for conditions where the amount of boron waste is 25% and the grinding time is 75 min. When Fig. 9 is examined, it is seen that increasing calcination temperature increases the formation of the mullite phase in the chamotte and reaches a maximum at 1100 0 C. For conditions where the calcination temperature is 950 0 C and the amount of Boron waste is 25%, the effect of the grinding time on the formation of the mullite phase in the chamotte is given in Fig. 10. It is observed that the grinding time is not as important as other parameters, and the increased grinding time does not make a significant contribution to the formation of mullite. This situation is clearly understood from the statistical evaluation in the ANOVA table given in Table 4.

Optimization study
In the study, experimental conditions were determined that gave the maximum formation of the mullite phase with minimum calcination temperature and grinding time. The percentage of boron waste amount was kept within the applied range. It was also desired that other phase formations be in the range.
In general, the mullite phase is increasing, while the anorthite phase is decreasing. In particular, the amount of boron waste increases the formation of anorthite, while the mullite phase decreases. Accordingly, among the options presented in Table 5, option 12 was preferred, which also has a high formation of anorthite phase along with the maximum mullite phase. However, in this option, the temperature should be at 995 0 C. It is understood from the optimization study that 8.12% mullite and 16.84% anorthite phase can be obtained under these conditions.
The perturbation plot showing the individual effects of three factors at optimized conditions is presented in Fig. 11. As seen any deviation of optimized values of factors affects the mullite phase formation significantly. Any increase of calcination temperature from 995 0 C and grinding time from 48 min increases the amount of mullite phase. In contrast, increased boron waste content from 10% and decreased calcination temperature from the optimized level cause a negative effect.

Interaction of parameters in the formation of mullite phase at optimized condition
In order to investigate the dual interactions of three tested factors between them the mullite phase formation, 3D response surface plots are generated at optimized levels of the factors. They are presented in Figs. 12, 13 and 14 for the boron waste percentage-calcination temperature (AB), boron   Table 4 that the interaction of the amount of boron waste (A) and the calcination temperature (B) is the second most important model parameter in the formation of the mullite phase. This interaction is presented in Fig. 12 at an optimized grinding time of 48 min. As can be seen, the change in temperature and boron waste content affects the formation of the mullite phase, but this effect changes depending on the temperature. It is understood that the best result is obtained at high temperatures with low boron waste content. It is seen that a lower mullite phase is obtained at 800 0 C and the increasing amount of boron waste is almost not effective in the phase formation at this temperature. At 1100 0 C, the best result is achieved with an amount of 10% boron waste, while increasing the amount of boron waste at this temperature rapidly reduces the formation of mullite nonlinearly. The 3D response surface graph showing the interaction of the amount of boron waste (A) and the grinding time (C) at an optimized calcination temperature of 995 0 C in the formation of mullite is given in Fig. 13. As can be seen, in the interaction graph, the high and low grinding time curves are almost at the upper exponent, so the effect of the grinding time is not much at the highest boron waste content of 40%. At optimized boron waste content of 10%, it is seen that an increase in mullite phase formation up to 12% was possible with increasing grinding time.
The graph which shows the interaction between calcination temperature (B) and the grinding time (C) at optimized boron waste content of 10%, is given in Fig. 14. At low and high levels of grinding time, increased calcination temperature gives the higher mullite phase formation in chamotte. At the low level of calcination temperature, increased grinding time had almost no effect on the mullite phase while the mullite phase formation increased at the high level of calcination temperature.  (Table 6). Accordingly; the casting slip contains 32% chamotte, while the remaining 68% consists of other compositions. According to these compositions, clay group raw materials were opened in the mixer and a casting slip was obtained to be poured into gypsum molds. Viscosity measurement was also carried out for pouring casting slip into gypsum molds. The prepared casting slips were shaped from gypsum molds by the slip casting method. The formed samples were first dried at room temperature, then at 100 0 C in the oven, and sintered at 1220 0 C in the operating oven.
The strength, % water absorption, % deformation and % shrinkage of the sintered samples were determined. The properties of vitrified products produced using the chamotte obtained by optimization and the reference sample of the factory are given in Table 7. It has been determined that the product produced with optimized chamotte has provided an improvement of 32.73% in strength, 5.88% in shrinkage and 4.67% in water absorption compared to the reference factory sample. The deformation of the optimized product was found to be 14.28% worse than the reference product. This is due to the anorthite phase formed, as well as mullite in the composition of the chamotte. In literature, better-vitrified product properties have been reported by using various waste materials in ceramic sanitary ware. For instance, Öztürk et al. [24] found that new sanitary ware samples with waste had higher water absorption and shrinkage values than the standard sample in a study using stone-cutting waste. Tarhan [25] concluded that wall tile waste (up to 10% by weight) reduces thermal expansion, water absorption, temperature deformation, and increases fireclay strength. In study by Güngör et al. [26], in which they examined the effect of adding galvanized waste (~ 65% ZnO) instead of kaolin and albite to the sanitary ware composition, a lower firing temperature, greater whiteness, less water absorption and faster densification values have been obtained. In another study, Boulaiche et al., [27] investigated the effect of the substitution of feldspar by soda-lime glass waste (20%) on rheological behavior, thermal, physical-mechanical and structural properties of sanitary-ware vitreous china bodies. The use glass waste the composition of bodies reduced water absorption and increased flexural strength. De Medeiros et al., [28] evaluated the substitution of quartzite waste (15%) for feldspar in ceramic bodies. The physical and mechanical properties of the product are within the recommended range for sanitary ware, with mechanical strength values exceeding 35 MPa and water absorption values close to 0.5%.

Conclusion
The use of mullite-based chamotte produced from kaolin and boron waste instead of imported chamotte in the production of sanitary ware products was investigated. The statistical experimental design was used to determine the   effects of the parameters (boron waste percentage, calcination temperature and grinding time) on mullite formation. A mathematical model that predicts the mullite phase and the conditions that provide the highest mullite phase formation by optimization were determined. The properties of the factory product produced with the imported chamotte were compared to the product properties produced with the chamotte obtained from the optimization study. The important results obtained are as follows: • According to the CCD experimental design, with increasing boron waste percentage, the formation of the mullite phase increased. While the mullite phase does not occur with the addition of 40% boron waste, it is at the maximum in a 10% addition. • The increasing of calcination temperature increased the formation of the mullite phase in the chamotte, reaching a maximum at 1100 0 C. • The grinding time was not as crucial as other factors, and the increased grinding time didn't make a significant contribution to the formation of mullite. • The optimum conditions for obtaining the maximum mullite phase (8.12% mullite) were determined at 10% for boron waste percentage, calcination temperature of 995 0 C and grinding time of 48 min. • The new sample outperformed the production sample in terms of strength (34.47 MPa), shrinkage (5.4%), and water adsorption (10.61%) when the features of FFC products made with the optimized chamotte and those produced with the imported chamotte were compared.