Facile Synthesis of ZnO-CeO2 Heterojunction by Mixture Design and Its Application in Triclosan Degradation: Effect of Urea

In this study, simplex centroid mixture design was employed to determine the effect of urea on ZnO-CeO. The heterojunction materials were synthesized using a solid-state combustion method, and the physicochemical properties were evaluated using X-ray diffraction, nitrogen adsorption/desorption, and UV–Vis spectroscopy. Photocatalytic activity was determined by a triclosan degradation reaction under UV irradiation. According to the results, the crystal size of zinc oxide decreases in the presence of urea, whereas a reverse effect was observed for cerium oxide. A similar trend was observed for ternary samples, i.e., the higher the proportion of urea, the larger the crystallite cerium size. In brief, urea facilitated the co-existence of crystallites of CeO and ZnO. On the other hand, UV spectra indicate that urea shifts the absorption edge to a longer wavelength. Studies of the photocatalytic activity of TCS degradation show that the increase in the proportion of urea favorably influenced the percentage of mineralization.


Introduction
Triclosan (TCS) is a sterilizing agent commonly used in consumer products such as soaps, toothpaste, etc. [1,2], typically with a concentration between 0.1-0.3% [3]. Kim et al. conducted studies to determine the antimicrobial efficiency of TCS, and concluded that there was no difference between soap with TCS and soap without this compound [4]. They also showed that there is no other skin benefit from the molecule. On the contrary, frequent exposure to TCS can generate alterations in bacteria, making them resistant to the chemical [5,6]. Chen et al. corroborated the resistance of bacteria to TCS [7]. The TCS biodegradation cycle is long [8], and persists in wastewater [9]. TCS can be removed from wastewater by activated sludge, with 72-94% effectiveness. The remaining TCS will contact the environment, contaminating lakes, rivers, and groundwater; hence, the potential impact on snails, algae, fish, mammals, and even humans is unavoidable [10,11]. According to Dayan et al., breast milk contains between 100 to 2100 µg/kg of TCS [12]. China identified the presence of TCS in its population in quantifiable concentrations [13]. Elsewhere, (D = 0.9/Cosθ)

Diffuse Reflectance UV-Vis Spectroscopy (DRS UV-Vis)
The UV-vis diffuse reflectance spectra were performed on a Varían Cary 300 spectrophotometer (Varian Inc., Palo Alto, CA, USA), in the range of 800 to 200 nm, equipped with an integrating sphere. A BaSO 4 compound with 100% reflectivity was used as a reference. The bandgap energy (E g ) of the samples was estimated from the UV absorption spectra, considering that: α(E) where ∝ (E) ∝ E − E g m/2 is the absorption coefficient for a photon of energy E and m = 4 for an indirect band transition [47].

Catalytic Test
Photocatalytic degradation tests were carried out in a photochemical reactor provided with UV light irradiation (λ = 365 nm) using a mercury lamp (13 W). At a natural pH, the photocatalyst (0.5 g/L) was dispersed in 200 mL of a TCS aqueous solution (10 ppm). As an oxygen source, an airflow (3.2 L/min) was provided to dissolve 8.4 mg/L of oxygen. Before the lamp was turned on, the suspension was continuously stirred at 700 rpm for 60 min in darkness, to establish an adsorption-desorption equilibrium between photocatalyst and pollutant. The system was maintained by circulating water at room temperature, and confined in a dark crate with UV light protection. Approximately 3 mL of the suspension was sampled and filtered (nylon, 0.45 m) to determine the ACP residual concentration, using a UV-Vis spectrophotometer (Varían, Cary 300). The results obtained calculated the degradation percentage, the total organic carbon percentage, the molar ratio of mineralization, and catalytic activity. Conversion percentages were determined using the following equation: where [A • ] (ppm) is the concentration when the light is turned on to start the photodegradation process, and [A] (ppm) is the concentration after one hour under irradiation. Total organic carbon (TOC) was used. This is a global parameter to assess pollution elimination in water. Shimadzu TOC-VCHS analyzer equipment (Shimadzu Corp., Tokyo, Japan), the TOC-LCSN model, was employed to determine the TOC. The following equation determines the percentage of TOC: where [TOC • ] (ppm) is the total organic carbon measured when the light is turned on to start the photodegradation process, and [TOC] (ppm) is the total organic carbon measured after one hour under irradiation. The efficiency corresponds to the total organic carbon conversion in (mmol C /L) to UV degradation.

Simplex-Centroid Mixture
The simplex-centroid mixture is an experimental design used to understand the effects of the different components of a mixture with a minimum of experiments and an efficient mapping of the experimental space. The experimental space consists of different points distributed over an equilateral triangle, representing the proportions of the components of a mixture. The sum of the ratio of all the components of the mixture is equal to 1, or 100% [48]. This was the experimental design used to determine the effect of the proportions of ZnO, CeO 2 , and urea on the photodegradation of TCS. With the presence of constraints (i.e., x i = x + y + z = 100%), the degree of freedom of the system was reduced from 3 to 2. Out of 10 samples studied, samples 1-3 represent the vertices in the ternary diagram, 4-6 represent the edges of the diagram with binary mixtures, while 7-10 are the ternary Nanomaterials 2022, 12,1969 4 of 17 experimental space: see Figure 1. The experimental variation of the same sampled points was determined by synthesizing sample 10 three times: see Table 1. ternary diagram, 4-6 represent the edges of the diagram with binary mixtures, while 7-10 are the ternary experimental space: see Figure 1. The experimental variation of the same sampled points was determined by synthesizing sample 10 three times: see Table 1.

Figure 1.
Three-factor simplex-centroid mixture design with three levels corresponding to urea, zinc nitrate, and cerium nitrate. The procedure consisted of dissolving the required amount in 15 mL of ethanol. The solutions were mixed and remained agitated for 1 h at 500 rpm. Subsequently, the solvent was removed in an oven at 80 °C for 12 h. The product obtained was solid, and received heat treatment at 500 °C for 3 h at 2 °C/min. Calculations were made to obtain 5 g of each sample. The amounts of each compound are shown in Table 1.

Temperature Programmed Oxidation (TPO)
The TPO study was carried out in a BELCAT-3000 apparatus (Bel-Japan, Tokyo, Japan) using a thermal conductivity detector (TCD), and 0.1 g of catalyst. In these experiments, the flow rate of the 5%/O2/95% He mixture was 10 mL/min, and the heating rate was 10 °C/min. Finally, the spectra were recorded from room temperature to 500 °C.  The procedure consisted of dissolving the required amount in 15 mL of ethanol. The solutions were mixed and remained agitated for 1 h at 500 rpm. Subsequently, the solvent was removed in an oven at 80 • C for 12 h. The product obtained was solid, and received heat treatment at 500 • C for 3 h at 2 • C/min. Calculations were made to obtain 5 g of each sample. The amounts of each compound are shown in Table 1.

Temperature Programmed Oxidation (TPO)
The TPO study was carried out in a BELCAT-3000 apparatus (Bel-Japan, Tokyo, Japan) using a thermal conductivity detector (TCD), and 0.1 g of catalyst. In these experiments, the flow rate of the 5%/O 2 /95% He mixture was 10 mL/min, and the heating rate was 10 • C/min. Finally, the spectra were recorded from room temperature to 500 • C.  Figure 2 represents the results of the nitrogen adsorption/desorption process. All samples present type IV isotherms, characteristic of the adsorbent/adsorbate interaction between molecules in a condensed state. The hysteresis cycle is unique in each sample; it allowed us to identify the predominant pore types. The CN sample exhibits an H3−type hysteresis [49,50]. It presents two distinctive features: (i) the adsorption branch resembles a type II isotherm, and (ii) the lower boundary of the desorption branch is usually located at the cavitation-induced p/p0. Loops of this type are characteristic of non-rigid aggregates of plate-like particles. The sample Ce and Zn [51] have an H2(b)−type loop, and pores with large neck width. Binary (ZnCN, ZnCe, and CeCNN) and ternary material (Zn 16 Figure 2 represents the results of the nitrogen adsorption/desorption process. All samples present type IV isotherms, characteristic of the adsorbent/adsorbate interaction between molecules in a condensed state. The hysteresis cycle is unique in each sample; it allowed us to identify the predominant pore types. The CN sample exhibits an H3−type hysteresis [49,50]. It presents two distinctive features: (i) the adsorption branch resembles a type II isotherm, and (ii) the lower boundary of the desorption branch is usually located at the cavitation-induced p/p0. Loops of this type are characteristic of non-rigid aggregates of plate-like particles. The sample Ce and Zn [51] have an H2(b)−type loop, and pores with large neck width. Binary (ZnCN, ZnCe, and CeCNN) and ternary material (Zn16Ce16CN66, Zn16Ce66CN16, and Zn66Ce33CN33) showed an H2(b)−type loop with the possible presence of surface blocked pores. Finally, the replicas (Zn33Ce33CN33) also present an H2(b)−type loop. Loop type and characteristics are taken from Tommes, M. et al. [52]. Surface areas were discussed to determine the influence of urea: see Table 2. The CN sample obtained the highest surface area, followed by Ce and Zn. As for the mixtures, zinc has the largest influence on surface area. It is probably the crystallinity of zinc that affects this property. On the other hand, there were similar areas across all of the replicate samples; therefore, there is a good reproducibility. Surface areas were discussed to determine the influence of urea: see Table 2. The CN sample obtained the highest surface area, followed by Ce and Zn. As for the mixtures, zinc has the largest influence on surface area. It is probably the crystallinity of zinc that affects this property. On the other hand, there were similar areas across all of the replicate samples; therefore, there is a good reproducibility.     ZnCN has diffractions corresponding to the hexagonal structure of zinc oxide. The presence of a graphite phase is discarded; it is probably in an amorphous form. The diffraction intensity is lower and broader, compared to Zn. These two characteristics are typical of the formation of ZnO/C [54]. This product shows excellent activity in the visible region [55]. On the other hand, the CeCN sample exhibited the cubic structure of cerium oxide. The mixture of urea and cerium nitrate is explosive in the presence of an energy source [56]. Urea in contact with cerium nitrate decomposes to biuret and ammonia, then to isocyanic acid ((HNCO)3) at high temperatures [57]. On the other hand, urea promotes the nucleation of cerium oxide at low temperatures [58]. Therefore, the CeCN sample ZnCN has diffractions corresponding to the hexagonal structure of zinc oxide. The presence of a graphite phase is discarded; it is probably in an amorphous form. The diffraction intensity is lower and broader, compared to Zn. These two characteristics are typical of the formation of ZnO/C [54]. This product shows excellent activity in the visible region [55]. On the other hand, the CeCN sample exhibited the cubic structure of cerium oxide. The mixture of urea and cerium nitrate is explosive in the presence of an energy source [56]. Urea in contact with cerium nitrate decomposes to biuret and ammonia, then to isocyanic acid ((HNCO) 3 ) at high temperatures [57]. On the other hand, urea promotes the nucleation of cerium oxide at low temperatures [58]. Therefore, the CeCN sample shows higher intensity diffractions concerning Ce. As for the ZnCe sample, it showed similar patterns to zinc oxide. However, it cannot be attributed to the hexagonal wurtzite structure of zinc oxide because it presents a left shift of 0.5 • . According to the literature, the shift is caused by bond formation between -Ce-O-and -Zn-O [59], resulting in the ZnO-CeO 2 heterojunction [60]. The results of the ternary samples are shown in Figure 4. In sample Zn 66 Ce 16 CN 16 , the hexagonal, cubic, and polymorphic structures of zinc oxide, cerium, and heterojunction (Zn-Ce) coexist. Sample Zn 16 Ce 66 CN 16 is dominated by cerium oxide (cubic) and traces of Zn-Ce. Sample Zn 16 Ce 16 CN 66 did not show the presence of carbon. However, the cubic structure of cerium oxide stands out. On the other hand, for Zn 33 Ce 33 CN 33 , the cubic structure of the cerium oxide and the heterojunction coexist. It is concluded that the displacement diffraction peak is a function of urea concentration, and may indicate a better interaction between ZnO and CeO.

X-ray Diffraction
shows higher intensity diffractions concerning Ce. As for the ZnCe sample, it showed similar patterns to zinc oxide. However, it cannot be attributed to the hexagonal wurtzite structure of zinc oxide because it presents a left shift of 0.5°. According to the literature, the shift is caused by bond formation between -Ce-O-and -Zn-O [59], resulting in the ZnO-CeO2 heterojunction [60].
The results of the ternary samples are shown in Figure 4. In sample Zn66Ce16CN16, the hexagonal, cubic, and polymorphic structures of zinc oxide, cerium, and heterojunction (Zn-Ce) coexist. Sample Zn16Ce66CN16 is dominated by cerium oxide (cubic) and traces of Zn-Ce. Sample Zn16Ce16CN66 did not show the presence of carbon. However, the cubic structure of cerium oxide stands out. On the other hand, for Zn33Ce33CN33, the cubic structure of the cerium oxide and the heterojunction coexist. It is concluded that the displacement diffraction peak is a function of urea concentration, and may indicate a better interaction between ZnO and CeO.  Table 2. In the binary samples, the crystal size of zinc oxide is reduced; urea favors the crystal size of cerium. In ternary samples, higher urea concentrations increase the crystal size of cerium, and cause a reverse effect on zinc.
In Table 3, we analyze various literature sources related to heterojunction. Research articles 1, 6, 8, and 10 present X-ray diffractions with similar effects to the ternary samples. Additionally, paper 4 evidences displacement due to heterojunction. Moreover, the degree of heterojunction essentially depends on two factors: the precursors and the additives. If both precursors have the same organic or inorganic part, this favors the formation of a heterojunction.  The crystal sizes (D) are shown in Table 2. In the binary samples, the crystal size of zinc oxide is reduced; urea favors the crystal size of cerium. In ternary samples, higher urea concentrations increase the crystal size of cerium, and cause a reverse effect on zinc.
In Table 3, we analyze various literature sources related to heterojunction. Research articles 1, 6, 8, and 10 present X-ray diffractions with similar effects to the ternary samples. Additionally, paper 4 evidences displacement due to heterojunction. Moreover, the degree of heterojunction essentially depends on two factors: the precursors and the additives. If both precursors have the same organic or inorganic part, this favors the formation of a heterojunction.

Diffuse Reflectance UV-Vis Spectroscopy
The results of the UV absorption spectra are shown in Figure 5. In agreement with the literature, the Zn sample presents an absorption edge of 446 nm [71,72]. It also presents an absorbance at wavenumber 344 nm related to the photon absorption of Zn 2+ [73]. Ce and CN show an absorbance edge at 428 and 460 nm, respectively [74,75]. The Ce sample presented two absorbances at wavenumbers of ≈ 290 and 260 nm, corresponding to the absorption of Ce 4+ and Ce 3+ oxidation states [76]. The CN sample presented an intense absorbance at 400 nm, about the degree of exfoliation of carbon nitride [77].

Diffuse Reflectance UV-Vis Spectroscopy
The results of the UV absorption spectra are shown in Figure 5. In agreement with the literature, the Zn sample presents an absorption edge of 446 nm [71,72]. It also presents an absorbance at wavenumber 344 nm related to the photon absorption of Zn 2+ [73]. Ce and CN show an absorbance edge at 428 and 460 nm, respectively [74,75]. The Ce sample presented two absorbances at wavenumbers of ≈ 290 and 260 nm, corresponding to the absorption of Ce 4+ and Ce 3+ oxidation states [76]. The CN sample presented an intense absorbance at 400 nm, about the degree of exfoliation of carbon nitride [77]. The ZnCN sample shows an absorption edge at 700 nm in the red region of the visible spectrum. According to the literature, carbon-doped zinc oxide shows an absorption edge in the red [78]. This might be induced by carbonaceous materials, which lead to better photocatalytic performance [55]. On the other hand, the ZnCe sample shows activation at 500 nm. According to Xiong et al., the interaction between cerium and zinc has a photosensitizing effect of absorbing the red [79]. Meanwhile, CeCN shows an absorption gap near 473 nm. This gap is slightly more significant than that of the Ce sample (461 nm). This slight enhancement is due to the interaction with nitrogen [80]. Thus, reduction reactions occur in ZnO, and oxidation reactions occur in CeO2.
The results of the UV spectroscopy analysis of the ternary samples are shown in Figure 6. Zn16Ce66CN16 shows no absorption at wavelengths longer than 500 nm. The spectrum of Zn66Ce16CN16 showed absorptions at 459 nm and 700 nm. Zn16Ce16CN66 also showed two absorptions, at 542 nm and 700 nm. The sample Zn33Ce33CN33 showed activation at wavelength 446 nm, close to the values of the pure samples, while the replicates showed activation at 481 nm. Therefore, the reproducibility has a difference of 7.8%. The ternary samples possess high absorption in both the UV and visible regions. This indicates that the modified samples could benefit from both visible and UV responses. The ZnCN sample shows an absorption edge at 700 nm in the red region of the visible spectrum. According to the literature, carbon-doped zinc oxide shows an absorption edge in the red [78]. This might be induced by carbonaceous materials, which lead to better photocatalytic performance [55]. On the other hand, the ZnCe sample shows activation at 500 nm. According to Xiong et al., the interaction between cerium and zinc has a photosensitizing effect of absorbing the red [79]. Meanwhile, CeCN shows an absorption gap near 473 nm. This gap is slightly more significant than that of the Ce sample (461 nm). This slight enhancement is due to the interaction with nitrogen [80]. Thus, reduction reactions occur in ZnO, and oxidation reactions occur in CeO 2 .
The results of the UV spectroscopy analysis of the ternary samples are shown in Figure 6. Zn 16 Ce 66 CN 16 shows no absorption at wavelengths longer than 500 nm. The spectrum of Zn 66 Ce 16 CN 16 showed absorptions at 459 nm and 700 nm. Zn 16 Ce 16 CN 66 also showed two absorptions, at 542 nm and 700 nm. The sample Zn 33 Ce 33 CN 33 showed activation at wavelength 446 nm, close to the values of the pure samples, while the replicates showed activation at 481 nm. Therefore, the reproducibility has a difference of 7.8%. The ternary samples possess high absorption in both the UV and visible regions. This indicates that the modified samples could benefit from both visible and UV responses. Eg was estimated using the Tauc equation [81]. The Eg energies are given in Table 3. The ZnO sample presents an Eg in agreement with the literature [82]. Ce presents an Eg close to that reported in the literature [83,84]. Additionally, the Eg of CN agrees [85]. The Eg was estimated using the Tauc equation [81]. The Eg energies are given in Table 3. The ZnO sample presents an Eg in agreement with the literature [82]. Ce presents an Eg close to that reported in the literature [83,84]. Additionally, the Eg of CN agrees [85]. The heterojunction (ZnCe) obtained the lowest Eg value of the binary samples. Concerning the ternary samples, if the proportion of Zn is high, the Eg decreases. The Eg value is lowest in ternary mixtures with a higher proportion of urea (Zn 16 Ce 16 CN 66 ). However, Eg increases with higher cerium ratios.

Catalytic Test
The photocatalytic activity of the samples was tested with a TCS conversion reaction under UV irradiation. The change in the normalized concentration as a function of time is shown in Figure 7a,b. All samples exhibited photocatalytic activity. The Zn, Ce, and CN samples showed conversions of more than 50%. However, they failed to mineralize the by-products: see Figure 7c. The binary samples obtained higher conversion values concerning CeCN, the removal of by-products being favored by cerium oxide. The ternary samples presented conversion values equal to or higher than Zn. In these samples, the degradation of by-products was efficient. The ternary samples are identified by the presence of the heterojunction of ZnO and CeO, and the increase in the amount of urea increased the mineralization. According to the reaction constant values, Ce obtained the highest value of the pure samples, CeCN of the binaries, and Zn 16 Ce 16 CN 66 of the ternaries. The results of the TPO analysis show that Zn and CN present deposition of organic material on the surface, whereas Ce obtained the lowest deposition. The binary samples obtained lower deposited amounts compared to Zn. The deposition of organic matter in the ternary samples depends on two factors: urea and cerium. The results were compared with articles related to the degradation of triclosan: see Table 4. From the table, it is understood that the pH and catalyst concentration influence the conversion and TOC percentage. The Zn and Ce samples of the current study presented an activity within the range reported in the literature. The binary combinations showed lower conversion but better TOC percent, even though the power source is of lower wattage in the current work. The ternary samples show similar conversions to those The results were compared with articles related to the degradation of triclosan: see Table 4. From the table, it is understood that the pH and catalyst concentration influence the conversion and TOC percentage. The Zn and Ce samples of the current study presented an activity within the range reported in the literature. The binary combinations showed lower conversion but better TOC percent, even though the power source is of lower wattage in the current work. The ternary samples show similar conversions to those reported in Table 4, but higher TOC conversions than the binary and pure samples. The samples with heterojunction were shown to be highly efficient photocatalysts with high redox capacity [79]. The photocatalytic mechanism considered as a semiconducting heterojunction was discussed according to the catalytic activity and characterization results. A possible Z-scheme electron transfer mechanism of the urea-assisted ZnO 2 −CeO 2 composite catalyst ZnO 2 −CeO 2 is proposed in Figure 8.  Linear, quadratic, and cubic mathematical models [48,91] were fitted to the reaction efficiency (TOC) data using Statistica 12.0 software (Tulsa, OK, USA) [92]. The criteria for model choice were provided by the following statistical data: correlation coefficient, standard deviation, mean square F-test, and p-value. The p-value tests whether the model is significant or makes a significant additional contribution to explaining the response (TOC) when comparing mathematical models. The p-value criterion usually allows the Linear, quadratic, and cubic mathematical models [48,91] were fitted to the reaction efficiency (TOC) data using Statistica 12.0 software (Tulsa, OK, USA) [92]. The criteria for model choice were provided by the following statistical data: correlation coefficient, standard deviation, mean square F-test, and p-value. The p-value tests whether the model is significant or makes a significant additional contribution to explaining the response (TOC) when comparing mathematical models. The p-value criterion usually allows the model to be chosen if it is less than 0.05. Using this criterion, only the cubic model is suitable for modeling the response: see Table 5. If the p-value values of the three models were close, the correlation coefficient (R 2 ) criterion would be chosen to select the model. The coefficients of the selected model are shown in Table 6. It is inferred that the product obtained from precursor A achieves the most considerable response value (TOC) compared to the other two products (B and C): see Table 6. However, the combination of all three species has the most significant influence on the response. This suggests that efficiency is the synergistic result of the interaction between the species. The projected response surface in the contour plot of the experimental triangular space estimated by the mathematical model is shown in Figure 9. The contour plot illustrates the variations of the interactions in the response (TOC). The dark green regions represent low efficiencies, and the dark red regions represent high efficiencies. The optimal mixture covers a small area and is positioned just above the Zn 33 Ce 33 CN 33 sample. Figure 9 shows a contour plot projection of the response surface; b and c are contour plots of Eg and surface area responses.
The statistical analysis of Eg response and surface area is shown in Appendix A. The area with the lowest Eg (dark green color) covers only a fraction of the surface area. The region with the lowest Eg is found in the binary mixtures between Zn-Ce, heterojunction being the sample with the lowest Eg. The surface area values are homogeneous in all combinations. The optimum area value (dark red) is in and around the CN sample.
The projected response surface in the contour plot of the experimental triangular space estimated by the mathematical model is shown in Figure 9. The contour plot illustrates the variations of the interactions in the response (TOC). The dark green regions represent low efficiencies, and the dark red regions represent high efficiencies. The optimal mixture covers a small area and is positioned just above the Zn33Ce33CN33 sample.

Conclusions
In this work, the ZnO−CeO2/urea system was synthesized by the solid-state combustion method to identify the role of urea in shaping heterojunction properties. According to the X-ray diffraction results, no additive, such as urea, is necessary for the formation of the heterojunction. The function of urea in the mixtures is more conducive to the formation of cerium oxide, while disfavoring heterojunction materials. The photocatalytic degradation of the TCS degradation reaction was tested. Zinc oxide and cerium oxide

Conclusions
In this work, the ZnO−CeO 2 /urea system was synthesized by the solid-state combustion method to identify the role of urea in shaping heterojunction properties. According to the X-ray diffraction results, no additive, such as urea, is necessary for the formation of the heterojunction. The function of urea in the mixtures is more conducive to the formation of cerium oxide, while disfavoring heterojunction materials. The photocatalytic degradation of the TCS degradation reaction was tested. Zinc oxide and cerium oxide showed conversions higher than 50%. It is worth noting the activity of carbon nitride in the degradation; it presented a higher TOC conversion than zinc oxide and cerium oxide. The binary samples presented lower conversions with respect to their pure counterparts, but with greater degradation of by-products. In the ternary samples, the conversion of less than 50% persisted, although the degradation of by-products was greater than in the binary and pure samples. The mixture design helps us to understand the influence of the species. It also provides the most suitable composition for TCS mineralization.   The coefficient values of the selected model are presented in Table A2. Of the pure compounds, C has a more significant influence on Eg. AB has a positive effect in the binaries. The BC combination has the opposite effect, increasing the value of Eg. Undoubtedly, the interaction of the three species (ABC) has a strong influence on Eg, as indicated by the value of the coefficient. The surface area is modeled with the cubic fit. This is indicated by the p-values and correlation coefficient, see Table A3. The product obtained from urea calcination dramatically influences the response. This is indicated by the values of the coefficients of the sectioned model. It is worth mentioning that binary combinations negatively influence the area. Again, the interaction between species benefits this property: see Table A4.