MgO-La2O3 mixed metal oxides heterostructure catalysts for photodegradation of dyes pollutant: synthesis, characterization and artificial intelligence modelling

doi:10.21203/rs.3.rs-1888369/v1

Download PDF

Research Article

MgO-La2O3 mixed metal oxides heterostructure catalysts for photodegradation of dyes pollutant: synthesis, characterization and artificial intelligence modelling

https://doi.org/10.21203/rs.3.rs-1888369/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

In the present work, we prepared MgO-La₂O₃-mixed-metal oxides (MMO) as efficient photocatalysts for degradation of organic pollutants. First, a series of MgAl-%La-CO₃ layered double hydroxide (LDH) precursors with different content of La (5, 10 and 20 wt%) were synthesized by the co-precipitation process and then calcined at 600°C. The prepared materials were characterized by XRD, SEM-EDX, FTIR, TGA, ICP and UV–vis diffuse reflectance spectroscopy. XRD indicated that MgO, La₂O₃ and MgAl₂O₄ phases were found to coexist in the calcined materials. Also, XRD confirms the orthorhombic-tetragonal phases of MgO-La₂O₃. The samples exhibited a small band gap of 3.0-3.22 eV based on DRS. The photocatalytic activity of the catalysts was assessed for the degradation of two dyes namely Tartrazine (TZ) and Patent Blue (PB) as model organic pollutants in aqueous mediums under UV-Visible light. Detailed photocatalytic tests that focused on the impacts of dopant amount of La, catalyst dose, initial pH of the solution, irradiation time, dye concentration, and reuse were carried out and discussed in this research. The experimental findings reveal that the highest photocatalytic activity was achieved with the MgO-La₂O₃-10% MMO with photocatalysts with a degradation efficiency of 97.4% and 93.87% for TZ and PB, respectively within 150 min of irradiation. The addition of La to the sample was responsible for its highest photocatalytic activity. Response surface methodology (RSM) and Gradient Boosting Regressor (GBR), as artificial intelligence techniques were employed to assess individual and interactive influences of initial dye concentration, catalyst dose, initial pH and irradiation time on the degradation performance. The GBR technique predicts the degradation efficiency results with R² = 0.98 for both TZ and PB. Moreover, ANOVA analysis employing CCD-RSM reveals a high agreement between the quadratic model predictions and the experimental results for TZ and PB (R² = 0.9327 and Adj-R² = 0.8699, R² = 0.9574 and Adj-R² = 0.8704, respectively). Optimization outcomes indicated that maximum degradation efficiency was attained under the optimum conditions: catalyst dose 0.3 g/L, initial dye concentration 20 mg/L, pH 4, and reaction time 150 min. On the whole, this studyconfirms that the proposed artificial intelligence (AI) techniques constituted reliable and robust computer techniques for monitoring and modeling the photodegradation of organic pollutants from aqueous mediums by MgO-La₂O₃-MMO heterostructure catalysts.

MgO-La2O3 mixed oxides

Photocatalytic activity

Dyes removal

artificial intelligence (AI)

Gradient boosting regressor

Water resources, especially surface water and groundwater, have become contaminated with organic substances as a result of increased population, urbanization, and a higher standard of living (Gijin et al. 2021; Abdi et al. 2022). Azo dyes are classified as a significant group of organic contaminants. These pollutants are known to be released into different aquatic matrices through multiple pathways, including textile industry, paper, dye, pharmaceutical industry, and bleaching industries etc (Su et al. 2020; Rizi et al. 2020; Quinto et al. 2020). Each year, around 8*10⁵ tons of synthetic dyes are produced, with azo dyes accounting for half of them (Mattos et al.2020). Approximately 20% of this amount occurs in industrial effluents creating considerable concern and loading for effluent treatment systems (Bhagat et al. 2021). The permanent release of these contaminants and their degradable precursors into the aquatic environment has harmful consequences for human health and deteriorates the whole ecosystem, exhibiting carcinogenic and mutagenic effects on living species (Balthazard-Accou et al. 2019; Han et al. 2016; Banu et al. 2019). These organic compounds are unable to be degraded by light or aerobic digestion, therefore they require treatment before being released into the natural water matrices (Su et al. 2020; Rizi et al. 2020).

Tartrazine (TZ) and patent blue (PB) anionic azo dyes are the most widely used compounds as an additive in textiles, cosmetics and food industries (Soufi et al. 2022). Although PB is easily degradable, it can provoke harmful impacts such as asthma, headaches and allergic reactions (Sadiq et al. 2021; Barka et al. 2011). Contrary to PB, TZ is considered a recalcitrant compound and has also adverse effects on the health of humans causing depression, anxiety, migraine, asthma attacks, sleep disorders and anaphylactic shock (Lacson et al. 2022). Therefore, it is crucially important to treat the wastewater containing these dyes before discharged into the aquatic environment (Zubair et al. 2022; Ali et al. 2022). In general, conventional treatment technologies for industrial effluents which contain dyes comprise mainly a combination of physico-chemical and biological approachs including adsorption and coagulation (Madondo et al. 2022; Yue et al. 2021). These processes are principally constrained by the non-biodegradable and toxic nature of the effluent, which reduces the efficiency of the process, and by the high quantities of solid waste produced, which needs the unnecessary contribution of other treatment processes.

Among them, AOPs has been considered an attractive process for the degradation of a variety of persistent chemical compounds such as dyes, including Fenton process, ultrasonic processes, UV/H₂O₂ treatments and electrochemical AOPs (Ismail et al. 2022; Bilinska et al. 2021; Souza et al. 2022). Mainly, AOPs implies the decomposition and mineralization of these organic compounds into less harmful compounds such as CO₂, H₂O and inorganic ions as a result of the generation of highly reactive oxygen species (ROS) that include O.₂⁻, .OH or HO.₂ and the resulting oxidation of organic compounds. One of the most efficient AOPs for water affected by dyes molecules is the application of the heterogeneous photocatalytic owing to its high oxidation capacity, strong stability, low toxicity, high flexibility and cost-effectiveness (Ye at al. 2016). This method employs semiconductor catalysts to produce ROS and decompose organic pollutants, which makes it an efficient process for removing dyes and other organic contaminants from water sources (Barka et al. 2013; Taoufik et al. 2022).

Currently, in the exploration of cost-effective semiconductor materials with outstanding performance, layered double hydroxide (LDH) ([M^2 + _1−xM³_x (OH)₂]^x.(Aⁿ⁻_x/n). mH₂O) is reportedly the best choice in terms of attractive features such as its unique structure, tunable bandgaps, ease of scaling, low cost, high surface area, tunable optical bandgap (2.7 eV) and excellent thermal and chemical stability. In the same vein, the calcinations of hydrotalcite-like materials produced mixed metal oxides which are reported to have also homogeneous dispersion of M²⁺ and M³⁺ at an atomic level, a high specific area, good photocatalytic activity and other fascinating physicals and chemicals properties (Elhalil et al. 2019; Ghemit et al. 2017; Cocheci et al. 2020). Consequently, mixed metal oxides from the calcination of hydrotalcites have been evaluated for different catalytic applications.

Another part of our work is devoted to modeling. In fact, one of the main principles in the design of any wastewater technology is the determination of the optimal conditions of the process. Photocatalysis is one of the water purification technologies that involves a series of experiments to learn more about the effects of input variables, which can be both time-consuming as well as expensive. It would therefore be interesting to develop general predictive techniques to analyze dye removal and to discover the relative importance of each variable and their interactions on removal capacity. One of the possible approaches to model processes of great complexity is the exploitation of artificial intelligence (AI) techniques. These empirical methods are characterized by total independence of the information about the process concerned which allows to map the non-linear behaviors between a group of input and output parameters. Various AI techniques have been used for estimation and modelling of removal of dyes, such as Response surface methodology (RSM) (Fetimi et al. 2021), Random Forest (RF) (Soares et al. 2020), Least square-support vector machine (LS-SVM) (Ghaedi et al. 2014), and Artificial Neurol Network (ANN) (Mossavi et al. 2022). For instance, Soares et al. 2017 reported the feasibility of RF and ANN to model the adsorption process of methylene blue, which is considered a well-known AI technique in this field. Based on the obtained results, both RF and ANN models exhibited similar performances.

RSM via central composite design (CCD) and Gradient Boosting Regressor (GBR) models are examples of AI tools that have been exploited in the present research. RSM is a commonly applied statistical tool for modeling and optimization of overall contributions from all variables. At the same time, it reduces the number of experiments while enhancing the performance of the multicomponent processes with the minimum of experimentation errors, and it provides scientists with a simple, convenient, and innovative procedure that can be rapidly implemented. Among the RSM designs employed by researchers for modeling the multicomponent systems which include the Box-Behnken Design (BBD), Central Composite Design (CCD), the, the Historical Data Design, and the User-Defined Design (Zhang et al. 2018; Kooh et al. 2022; Friedman et al. 2000).

On the other hand, GBR is considered a very attractive and robust AI tool explaining complex nonlinear problems between input and output sets, owing to its flexibility in determining relationships between factors in smaller data sets and at the same time its ability to manage a large number of input variables (Wei et al. 2019; Mazaheri et al. 2017). Furthermore, its capability has been verified as being capable of solving a variety of engineering datasets.

The evolutionary scalability of the GBR technique is determined by optimizing multiple critical models and algorithms, such as a robust new tree-based learning algorithm for processing sparse data and a reasonable weighted algorithm for data management.

In the first phase of this study, MgAl-CO₃ and a series of MgAl-%La-CO₃ LDHs with different La contents at Mg/(Al + La) molar ration of 3 were prepared through coprecipitation process and calcined at 500°C in a tubular furnace. The as-synthesized materials MgO-Al₂O₃-MMO and (MgO-La₂O₃- (5, 10, 20%)-MMO) were used as photocatalysts for the degradation of TZ and PB as a model of textile contaminants under the UV-visible irradiation. Many operating variables, including dye concentration, pH of the solution, catalysts doses, and reaction time, have been used to assess the as synthesized material's ability to degrade the dyes. The RSM (using CCD plane) and GBR for prediction and optimization of the performance of the photodegradation process using the prepared materials were constructed in the second stage of the work based on MATLAB and Design expert software.

2.1 Reagents

All reagents were used in the commercial form, without further purification. The chemicals employed for the synthesis of hydrotalcites were: Magnesium nitrate (Mg(NO₃)₂.6H₂O), aluminum nitrate (Al(NO₃)₂.9 H₂O), lanthanum nitrate (La(NO₃)₃.6H₂O), sodium carbonate (Na₂CO₃) were products of Sigma-Aldrich and NaOH used for pH adjustment was purchased from LaboChemie (Spain). The target pollutant compounds, that is TZ) C₁₆H₉N₄Na₃O₉S₂, 214.65 g/mol, pK_a = 9.4) and PB (C₇H₆O₃, 138.12 g/mol, pK_a =2.78) were purchased from Sigma-Aldrich.

2.2 Synthesis of LDHs precursors and MgO-La₂O₃-MMO photocatalyst

MgAl-%La-CO₃ LDHs were prepared through the co-precipitation procedure using a 3:1 molar ratio with La contents of 0, 5, 10, and 20% at a constant pH solution. Typically, in 150 mL of deionized water, a mixed solution of Mg (NO₃)₂ 6H₂O, Al(NO₃)₃ 9H2O, and La(NO₃)₃ with a total metal ion concentration of 2 mol/L was dissolved. Subsequently, the alkaline solution Na₂CO₃ (50 mL, 1 mol/L) was added to a stirred mixed nitrates solution in a beaker containing 50 mL of double distilled water. To maintain a constant pH value of about 8.7 ± 0.2, suitable amount of NaOH solution (2 M) was also added dropwise under vigorous stirring. The formed gel was kept under continuous stirring for 4 h and then placed in an autoclave submit to a hydrothermal treatment for 16 hours at 75°C to produce a structure for LDHs that is well-crystallized.

After that, the obtained precipitates were filtered and washed repeatedly using bi-distilled water in order to remove any ions that may have remained in the finished products, and then aged at 100°C for 16 h to obtain MgAl-La CO₃ LDHs with various content of La (0, 5, 10 and 20%), these precursors were denoted as MgAl-La-z-CO₃, where z represents the content wt% of La(III). Finally, MgAl-La CO₃ LDHs samples were calcined at 600°C for 6 h, to obtain MgAl-La mixed-metal oxides, which were named MgO-La₂O₃ MMO. In this research, the mixed-metal oxide with different content of La (5, 10 and 20 wt %) was denoted specifically as MgO-La₂O₃-5% MMO, MgO-La₂O₃-10% MMO and MgO-La₂O₃-20% MMO. A schematic diagram for the preparation of the mixed oxide photocatalysts is illustrated in Fig. 1.

2.3 Characterization techniques

The structural, textural and morphological characteristics of the as-synthesized products were evaluated by X-ray diffraction, scanning electron microscopy (SEM), Fourier transform infrared (FTIR), ICP and TGA. The X-ray diffraction (XRD) patterns of the LDH precursors and their calcined products were realized at room temperature on a D2 PHASER diffractometer coupled with Cu Kα radiation (λ = 1.5418 Å) at 30KV and 10 mA. The patterns were recorded in a 2h range from 5 to 70, in steps size of 0.01° (0.5 s/step). The shape and size of the as-prepared samples were observed using SEM coupled to the energy dispersive X-ray spectroscopy (SEM/EDX) on a FEI FEG 450 (TESCAN) instrument. FT-IR spectra were recorded with a PerkinElmer. FT-IR spectrophotometer, in a range of 4000 − 400 cm^− 1 and the products were pressed into KBr disks. Thermogravimetric analysis (TGA) was carried out by using a TGA-Q500 equipment. The samples were submitted to heating from room temperature to 900°C with a heating rate of 10°C/min using nitrogen as a sweep gas at a flow rate of 90 mL/min. Furthermore, the elemental composition of the synthesized particles was measured by using an inductively coupled plasma utilizing an ICP-MS (VARIAN 2011) spectrometer.

2.4 Adsorption/photocatalytic degradation

To perform the photodegradation experiments, 100 mL of TZ solution (20 mg/L) containing 50 mg of each catalyst was added to a 100 mL glass beaker, then, it was placed under ultraviolet and visible light (UV-Visible) irradiation by a 400 W halogen lamp for 180 min, located 10 cm above the cell. The lamp was positioned inside a quartz vessel with two compartments: one containing the lamp and the other transmitting the water in a recirculation channel for temperature control and cooling of the system. It is worth mentioning that for every experiment, the adsorption tests, without irradiation, were performed for 1 h in order to achieve the adsorption-desorption equilibrium between the surface of the catalyst and dye. Therefore, the influence of surface adsorption in the degradation of TZ and PB is eliminated. Furthermore, the photolytic degradation tests (direct photolysis), in the absence of photocatalysts, were also carried out under similar operating conditions.

The effect of dye concentration using 0.3 g/L of MgO-La₂O₃-10% MMO was assessed by varying its concentration from 10–20 mg/L for TZ and PB. Photocatalyst dose was changed from 0.1 to 0.7 g/L, keeping pH value at 4.05 and TZ and PB dye concentration at 20 and 10 mg/L, respectively. The effect of the pH was evaluated at pH values of 4, 6 and 9 (adjusting using NaOH (0.1N) and HCl (0.1N)).

The adsorbed quantity was calculated by applying the expression below:

$${q}_{ads}=\frac{({C}_{0}-{C}_{e})}{R}$$

where q_ads represent the adsorbed quantity (mg/g), Ce and C₀ correspond to the equilibrium and initial concentrations (mg/L), respectively. R represents the mass of catalyst per liter of aqueous solution (g/L).

At regular time intervals, 4 ml aliquots were taken and subsequently filtered to separate the catalysts. The residual concentration of TZ was defined by using a TOMOS V-1100 UV-vis spectrophotometer at its maximum adsorption wavelength of λ_max = 425 nm. The same procedure and concentrations were used for (PB) (λ_max=637 nm).

The rate equation and degradation efficiency can be determined by using the following:

$$ln \left( \frac{{C}_{t}}{{C}_{0}}\right)=kt$$

where C_t and C₀ (mg/L) represent the TZ concentration after the time t and before light irradiation (min), and k corresponds to the rate constant.

3.1 Design of experiments

To evaluate the influence of the efficient factors and also identify the optimal conditions for the photocatalytic degradation efficiency of both dyes TZ and PB using %La-MgO-Al₂O₃, the RSM design-CCD design was designed applying Design Expert Software, (version 11). The RSM is a statistical and mathematical tool used for analyzing the impact of independent factors on the response and developing the regression model equations, which are useful for reaching the best system efficiency. Additionally, RSM investigated the influence of interactions and relations between effective operating variables on the photocatalytic degradation efficiency of TZ and PB to achieve the best conditions. In this research, the effect of four variables and five levels, initial dye concentration (mg/L) (X₁), pH (X₂), photocatalyst dosage (g/L) (X₃), reaction time (min) (X₄). The observed response is the degradation efficiency of the dyes after 180 min (R, %).

While regarding all the possible combinations of the input variables, RSM analysis suggested 30 experiments in total for each dye, among which 6 replications to a central point aimed at the correct optimization of the data, and a quadratic model was used to explore the interaction between the independent and the dependent factors. The experimental intervals and different levels of experimental designs for both dyes removal envisaged in this research are illustrated in Table 1. To perform statistical calculations, the experimental variables X_i were coded as x_i based on the following transformation equation:

$${X}_{i}=\frac{{x}_{i}-{x}_{0}}{\varDelta x}$$

where X_i represents the uncoded value of the i^th independent parameters and x_i is the dimensionless coded value of the i^th independent parameters, x₀ is the value of x_i at the center point, and $\varDelta x$ presents the step change value of the actual parameter.

Table 1

Experimental range and levels of independent process factors
Variables	Ranges and Levels					𝛥X
Variables	-2	-1	0	1	2	𝛥X
Initial dye concentration (mg/L) (X₁)	5	10	15	20	25	5
pH (X₂)	3	4	5	6	7	1
Photocatalyst dosage (g/L) (X₃)	0.2	0.3	0.4	0.5	0.6	0.1
Reaction time (min) (X₄)	90	110	130	150	170	20

3.2 Gradient Boosting Regression modeling

The GBR is an ensemble learning algorithm which was developed further by Friedman (Friedman et al. 2000). The GBR is an optimization algorithm that is employed to solve classification and regression problems (Wei et al. 2019). It is able to generate a strong prediction model by combining weak prediction models, that is the decision tree. The key principle is that each calculation is performed by a base model, and the next calculation is taken to reduce the residual of the last model and make a new base model in the direction of the gradient with reduced residuals. A diagram of the GBR algorithm is represented in Fig. 2.

The tuning of the GBR model involves the setting of several hyperparameters including max_depth, min_samples_leaf and allowing to set it as n_estimators.

Various studies investigated the performance of GBR in the prediction of various pollutants. Mazaheri et al. 2017 evaluated the performance of AI algorithm GBR for modeling the adsorption of methylene blue and Cd (II) and then compared it with the neural networks (ANN) method. They found that both approaches have comparable performance, which means that both models have the potential to predict the response of the studied process. In the same vein, Hanandeh et al. 2021, employed six AI models that were developed and evaluated to predict heavy metal adsorption in single, binary and ternary environments. All models provided outstanding results, with the correlation coefficient between the predicted and expected values > 0.99 in all cases. The findings of this study also demonstrated that GBR model can be employed for the prediction of experimental data and can be useful in reducing the cost of the experiment. In another study, Bhagat et al. 2020 explored the prediction of the adsorption performance of Mn by GBR. They have proven the robustness and predictive ability of this methodology.

4.1 Characterization of catalysts

4.1.1 X-ray diffraction (XRD)

The XRD analysis was realized to recognize the phase structure of the prepared materials. Figure 3 illustrates the X-ray diffraction patterns of LDH precursors. It can be seen, that all LDH precursors exhibit identical, typical sharp and strong reflections, implying agood crystallinity. The XRD patterns of MgAl-%La-CO₃LDHs revealed that the recognized peaks were matched to MgAl-CO₃ LDH patterns and they have the appearance of a layered structure, which reveals that the presence of La in the samples does not cause the layered structure to be destroyed. Whereas the intensity of the peaks decreases proportionally with the Lanthanum (III) content.

The diffraction peaks at 11.42, 23.04, 34.29, 60.97, and 62.04° 2θ could be attributed to the (003), (006), (009), (110), and (113) reflections of LDH, respectively, characteristic of a layered structure, without any independent Mg or Al phases (JCPDS 70-2151). These findings reveal that MgAl-%La-CO₃-LDHs can be successfully synthesized with varied contents of La (5, 10 and 20 wt%). Almost similar typical structures were observed in the XRD profiles of the precursors MgAl- 5%La-CO₃, MgAl- 10%La-CO₃, and MgAl- 20%La-CO₃ and were in agreement with that of natural hydrotalcite (card JCPDS 40–0216) in this case, La³⁺ ions substitute for Al³⁺ ions. The characteristic XRD peaks of LaCO₃ OH started to appear as indicated by the peaks at 2θ = 15.87, 22.3, 30.23, 38.4, and 46.32° (JCPDS 36-1481).

The fineness and crystallinity as a measure of reflection intensity, decreased with increasing lanthanum content. Although, MgAl-20%La CO₃ still has the main characteristic of LDH, but the by-products have been produced. On the other hand, there were no diffraction bands for these lanthanum species in the XRD pattern of MgAl-5%La CO₃. This suggested that lanthanum species may be widely dispersed on the surface of catalysts in modest amounts that XRD could not detect. The possible reason may be related to the different chargedensities for La³⁺ and Al³⁺ ions. The La³⁺ has lower charge density compared to Al³⁺ owing to itslarger ionic radius, as a result, the probable insertion of La³⁺ into the hydroxide sheets would cause the octahedral structure to be distorted, resulting in the formation of a new phase. The high electric charge of lanthanum favored the formation of carbonate and oxy-hydroxide lanthanum species at the very early step of the co-precipitation; thus, its high ionic radius inhibited the intercalation of larger lanthanum species in the hydrotalcite structure (Guo et al. 2012).

The XRD patterns of the LDH precursors after the calcination at 600°C (see Fig. 4) indicated that the reflections associated with LDHs had completely disappeared and that some new ones had been observed after the calcination of MgAl-La-CO₃ LDHs, signifying the collapse of the lamellar structure. The characteristic XRD peaks of MgO oxide began to appear as illustrated by the reflections at 43.72 and 63.45° 2θ, whereas there is a detection of signals corresponding to MgAl₂O₄ phase at 36.93, 44.8, 59.4, and 65.2°2θ. This is also supported by the JCPDS data (Card No.21-1152).

By increasing the La content, the characteristic peaks are observed at 2θ of 29.73, 35.51, 51.68, 54.65 confirming the formation of La₂O₃ without an obvious new phase containing La³⁺ or Al³⁺ (JCPDS No. 05-0602). The phase La₂O₃ is produced from the thermal decomposition of the relative species of lanthanum carbonate and oxyhydroxide by calcination at 600 K (Sahnoun et al. 2018), thus the presence of peaks at 2θ angles of 13.23°, 22.97° and 43.18° confirming the formation of the monoclinic lanthanum dioxycarbonate La₂O₂CO₃ phase (JCPDS 48-1113) (Kiwaan et al. 2020; Elhalil et al. 2018). However, there is no detection of signals related to Al₂O₃ phase, which signifies that Al₂O₃ was amorphous.

4.1.2 Fourier transform infrared (FTIR) spectra

The functional groups of the prepared materials were determined by the FTIR spectra. Infrared spectroscopy of LDH precursors before and after calcination at 600°C are presented in Figs. 5 and 6. The absorption peak between 3314 and ~ 3682 cm^− 1 in the spectra of the LDH materials was attributed to the stretching vibration of the OH group with hydrogen bonding from interlayer water molecules. Another typical band for the LDH materials is observed at 1626 cm^− 1, which was assigned to the O-H bending vibration of water molecules. An absorption peak at 1372 cm^− 1 was ascribed to the stretching vibration of the carbonate anions ${CO}_{3}^{2-}$. In the low- frequency region, the peaks observed in the range of 500–870 cm^− 1 could be assigned as lattice vibration modes of metal-oxygen-metal vibration (M-O, M-OH). The difference in the infrared spectra compared to the non-calcined materials could be attributed to the difference in the structure of the compounds constituted with different amounts of La. It can be observed that after calcination (T = 600°C), the typical ${CO}_{3}^{2-}$ bands (1364 cm^− 1) are still present. This band is assigned to the formation of the La₂O₂CO₃ phase.

4.1.3 Thermogravimetric analysis (TGA)

Thermal analysis, such as thermogravimetric analysis (TGA), is a recognized technique for observing decomposition reactions or physical and chemical reactions of the prepared materials. The thermal stabilities of the MgAl-CO₃-LDH and MgAl-La-10%-LDH were performed as a function of the temperature by TGA and the findings are illustrated in Fig. 7. Accordingly, the TGA curve of MgAl-CO₃LDH exhibits three well-distinguished degradation parts. The first part is associated with the removal of water at the surface of the LDH interlayer (45–195°C) and a weight loss of 16.65% is displayed on the TG curve for this part. The second one (in the temperature range of 195–275°C) is associated with dehydration of the brucite-like layers, where TGA curve indicates a smaller weight loss of 2.31%. The lastweight loss of 9.2% also occurred at a temperature range of 275–505°C, which is attributable to the decomposition of interlayer carbonates ions and dehydroxylation of the MgAl-CO₃LDH layers. The total mass loss of the synthesized LDH sample was determined to be 28.16%. MgAl-10%La-CO₃LDH and MgAl-CO₃LDH display similar behavior. The temperature range for the first mass loss for MgAl-10%-La-CO₃LDH extended up to 185°C, is assigned to the dehydration of residual water molecules from the synthesized material surface.In the second degradation step at the central temperature range (185–245°C), a gentle mass loss was observed of about 2.57%, due to dehydroxylation of the metal hydroxide layers. A third step, which was observed between 405 and 505°C, is assigned to the de-carbonation that occurred.

4.1.4 SEM analysis

The surface morphologies and microstructure of the as-prepared samples MgAl-%La-CO₃LDHs with different content of La and the corresponding mixed oxides samples were examined through a SEM and corresponding EDX spectrum, as displayed in Fig. 8. It is clearly seen that the MgAl-CO₃ LDH showed a particles aggregation and the corresponding EDX spectrum showed the presence of Mg and Al also indicating that LDH was successfully synthesized. The SEM image of MgAl-%La-CO₃ LDHs (see Fig. b, c, and d) has a heterogeneous surface with clearly observable porosity. The La particles were randomly and highly dispersed on the surface of the MgAl-CO₃ LDH. The MgAl-CO₃LDH sample exhibits the same morphology as the MgAl-%La CO₃ LDH but with larger clusters. The EDX pattern further confirmed the presence of Mg, Al, and La in MgAl-%La-CO₃ LDHs. When MgAl-%La-CO₃ LDHs are calcined at 600°C, their agglomeration in the form of a sponge is more evident and crystallization of the oxide occurred forming large particles which are responsible for the increase of the catalyst surface heterogeneity (the heterogeneity of the surface). For the MgO-La₂O₃-(5,10,20%) MMO, the spectrum exhibits peaks associated with La along with the other component elements Mg, Al, and O.

4.1.5 Chemical composition analysis

The analysis of the chemical composition of the LDH precursors was performed by ICP-MS. The synthesized materials were dissolved in an acidic bath employing 63%-concentrated nitric acid. The obtained results are displayed in Table 2. Based on the analysis, as expected, MgAl-%La LDHs is composed essentially of magnesium, aluminum and lanthanum. The molar ratio of (La/Al) was found to be 0.051, 0.123, and 0.26 for MgAl-5%La-CO₃ LDH, MgAl-10%La-CO₃ LDH and MgAl-20%La-CO₃ LDH, respectively, which are slightly higher than the theoretical values (see Table 2).

Table.2 Elemental compositions of LDH precursors determined by ICP-MS analysis.

	Cu(%)	Mg( %)	Fe (%)	Zn (%)	Al (%)	La(%)	Ni (%)
MgAl-CO₃	0.0037	42.1160	0.0079	0.0001	12.3503	0.0006	0.0061
MgAl-5%La-CO₃	0.0017	40.1160	0.0030	0.0004	6.6928	0.3420	0.0004
MgAl-10%La-CO₃	0.0053	42.7840	0.0220	0.0001	5.2050	0.6422	0.0013
MgAl-20%La-CO₃	0.0030	43.2150	0.0062	0.0002	5.4814	1.4307	0.0024

4.1.6 UV–vis

The optical absorption spectra of the as-synthesized materials MgO-Al₂O₄-MMO, and MgO-La₂O₃-10% MMO were evaluated by UV–vis diffuse reflectance spectroscopy (UV-DRS), and the results are presented in Fig. 9. The sample MgO-Al₂O₄-MMO shows two absorption bands that are consistent with the data presented for MgAl₂O₄ prepared by another method (Mossavi et al. 2022). The MgO-La₂O₃-10% MMO sample exhibit a strong absorption band at 300 nm with its absorption edge extending to 550 nm, which gives a band gap of 3.05 eV. The absorption of the MgO-La₂O₃-10% MMO is apparently higher than that of the MgO-Al₂O₃-MMO in the range of 200 to 800 nm, which means that the absorbance band increases by the incorporation of La in the lattice. This remarkable photo adsorption enhancement of MgO-La₂O₃-10% MMO stems from the incorporation of La particles, confirming the existence of La species in the product. It was interesting to observe that MgO-La₂O₃-10% MMO might have higher photocatalytic activity under visible light irradiation. The rate of electron hole pair production on the surface of the catalyst rises with rising absorption intensity, resulting in enhanced photocatalytic performance. The band gap energies of the samples MgO-Al₂O₃-MMO, and MgO-La₂O₃-10% MMO were determined based on the start of the UV-Vis spectra of the absorption values (Eq. 4).

$${\left(ahv\right)}^{2}=k(hv-{E}_{g})$$

where the terms a, h, ν, K and E_g represent the absorption coefficient, Planck’s constant (J.s), is the light frequency (s^− 1), absorption constant and band gap of semiconductors (eV), respectively.

The average band gap energies for MgO-Al₂O₃-MMO, and MgO-La₂O₃-10% MMO were thus calculated to be 3.22 and 3.05 eV, respectively. It can be concluded that MgO-La₂O₃-10% MMO has lower band gap energy and is expected to present better photocatalytic activity under visible light compared with MgO-Al₂O₃-MMO. The lower band gap could be assigned to the composite oxides formed by calcination MgO, La₂O₃, MgAl₂O₄ and the coupling interaction among their phases, and the form of the heterojunction structure, resulting in an improvement in light utilization and photocatalytic performance.

4.2 Photocatalytic study

In an attempt to evaluate the photocatalytic activities of the catalysts (MgO-La₂O₃- (5, 10, 20%)-MMO), photocatalytic degradation of 20 ppm TZ and PB solutions by the catalysts was performed under ultraviolet-visible irradiation. Degradation of the dyes was determined by measuring the absorbance of the dye solutions in 30 min intervals for 3 h.

4.2.1 Effect of the La doping on the photocatalytic reaction

The photocatalytic performance of the MgO-La₂O₃ (5,10,20%) MMO with different La doping contents were examined by the photodegradation of two dyes TZ and PB under ultraviolet-visible irradiation in the same photocatalytic experiments. The photolytic degradation (direct photolysis) without photocatalysts was investigated under the same operating conditions as photocatalytic degradation. Figures 10 and 11 present the results of photocatalytic activities of different catalysts. It can be clearly seen that the degradation of TZ and PB is very insignificant in the absence of photocatalysts under ultraviolet-visible irradiation.

In the adsorption phase, MgO-La₂O₃-5% MMO and MgO-La₂O₃-10% MMO have hardly any adsorption of TZ after being stirred for 60 min. The adsorption efficiencies of TZ on MgO-Al₂O₃ MMO, MgO-La₂O₃-5% MMO, MgO-La₂O₃-10% MMO and MgO-La₂O₃-20% MMO are 9.57%, 5.85%, 4.65 and 6.85%, respectively. Clearly, MgO-Al₂O₃ MMO exhibits slightly higher adsorptive ability than other photocatalysts. While the adsorption process decreases in the presence of MgO-La₂O₃-5% MMO, MgO-La₂O₃-10% MMO catalysts respectively. On the other hand, Figs. 10 and 11 illustrates that there is a significant decrease in the concentration of both dyes during irradiation in the presence of all as-synthesized catalysts in comparison to photolysis. Also, these findings reveal that the doped catalysts display excellent photocatalytic degradation performance compared to the undoped ones for TZ and PB dyes. However, the photocatalytic activity clearly increases with an increase of the La content from 5 to 10% which could be related to the presence of the mixed-metal oxides containing MgO, La₂O₃ and MgAl₂O₄ that have relatively smaller band gap than MgO-Al₂O₃ MMO, thus enhancing the visible light absorption of mixed metal oxides. Based on the above, it can be deduced that the optimal content of lanthanum is 10% in our work. This composition is thought to be suitable for the separation of photogenerated e-and h⁺ pairs. Therefore, in the following experiments, the investigation of TZ and PB degradation was mainly focused on MgO-La₂O₃-10% MMO

4.2.2 Effect of the initial solution pH

The pH is a critical factor to understand the surface and charging properties of the catalyst. The effect of pH on the photodegradation of TZ and PB in an aqueous medium was studied at pH of 4.0, 6.0 and 9.0. The values of pH were adjusted using appropriate amounts of HCl or NaOH solutions. The experiments were performed to cover pH at an initial concentration of TZ (20 mg/L) and PB (10 mg/L) over MgO-La₂O₃-10% MMO keeping the same operating conditions, and under UV-Visible irradiation (see Figs. 12 and 13). As clarified in figures the photocatalytic performance was enhanced at pH of 4 and was decreased significantly at pH 9.0 for both dyes. Adsorption and thus degradation of the dye appear to be selected at a pH lower than the zero-point charge pH (pH_zpc), i.e. 7.51 for 10% La-MgO-Al₂O₃-MMO. Generally, the pH of point of zero charge (pH_pzc) is an important property of as prepared materials surface because it affects its ability to adsorb anionic or cationic compounds when they are dispersed into an aqueous solution. When the pH is below the pH_pzc value, the solid surface become protonated, and an anionic molecule could be adsorbed. Considering that TZ and PB are anionic dyes, the electrostatic aspects of the process can be discussed. In this regard, the effect of the electrostatic interaction of PB and TZ dyes can be analyzed in terms of their charges with the pH_ZPC of the catalysts. The pH_ZPC values of the catalysts MgO-La₂O₃-10% MMO is 7.51. Therefore, at pH > 7.51 the surface acquires negative charge, favoring the adsorption of cationic compounds, while at pH < 7.51, the surface of the catalyst acquires a positive charge, favoring the adsorption of anionic molecules.

Since the PB structure occurs as monovalent anions in the pH region studied (pK_a =2.78), the observed behavior could be uniquely caused by the change in photocatalyst properties. The increase in photocatalytic activity observed at pH 4 could be related to enhanced adsorption of PB onto the photocatalyst, favored by a pH between 3 and 7 (Fig. 12) and more efficient formation of hydroxyl radicals. On the other hand, TZ dye can be reformed to a fully anionic form (Na3Tar (S)🡪 3 Na+(aq) + Tar³⁻), which explains the maximum adsorption at pH = 4 at pH < 7.51. When the pH of the solution increased, OH⁻ ions raise in the solution, therefore, OH- ions competed effectively with TZ species for adsorption on the surface of the photocatalystcausing a decrease in the dye removal percentage. Therefore, the functional amine group positively charged will react with a positive charge surface of MgO-La₂O₃-10% MMO at pH < 7.51 (see Fig. 14).

The trend of the photocatalytic activity observed at pH of 4.0 could be due to the favorably enhanced adsorption of both TZ and PB on the photocatalyst at pH values between 4 and 7 and the more efficient formation of hydroxyl radicals. In the basic medium (pH = 9.0), the decrease in the degradation efficiency could be attributed to many phenomena simultaneously intervening: the non-favorable adsorption, and the dissolution of the photocatalysts. Additionally, in an acidic environment, the molecular structure of the dye can be transformed into a quinoid structure, which is unstable and can rapidly be decomposed (Zhang et al. 2018). Nevertheless, a further rise of the pH causes an enhancement of the coulombic repulsion between the negatively charged MgO-La₂O₃-10% MMO surface and the OH species involved in the photocatalytic oxidation mechanism, which implies a decrease in the degradation efficiency.

4.2.3 Effect of initial concentration of TZ

The influence of the initial dye concentration on its degradation has been investigated with varyingconcentrations of both dye from 10–20 mg/L, keeping the initial solution pH set at 4.08 and the optimum catalyst dose of 0.3 g/L (see Fig. 15). It is found that the degradation efficiency of the sample has been increased initially for dyeconcentrations up to 20 mg/L for TZ, whereas the efficiency of the degradation of PB tends to decrease with the increase of the initial concentration above 10 mg/L. As a result, a lower initial dye concentration could improve dye removal efficiency. This might be explained by the fact that as the initial dye concentration was increased, more dye molecules were adsorbed on the photocatalyst surface. The adsorption of O₂ and OH⁻ on the photocatalyst was decreased, as many active sites were filled by the dye molecules, resulting in a reduction of radical regeneration. Furthermore, photons were inhibited before presenting on the surface of the photocatalyst, resulting in a decrease in photon adsorption by the photocatalyst.

Figure 16. Effect of initial concentration on the photocatalytic degradation of PB ([MgO-La₂O₃-10% MMO]₀ = 0.3 g/L; irradiation time:150 min and pH ~ 4.08).

4.2.4 Effect of photocatalyst dose

Optimal catalyst loading increases the rate of electron/hole pair generation and thus the formation of OH. radicals for water treatment. Therefore, the effect of catalyst dosage on the degradation efficiency was studied by varying the amount of the best photocatalyst (MgO-La₂O₃-10% MMO) in the range of 0.1–0.7 g/L by keeping the concentration of TZ and PB dyes at 20 and 10 mg/L, respectively (see Fig. 16). After 150 min of UV-Visible irradiation, the photocatalytic degradation efficiency (%) was evaluated. As can be seen, the photocatalyst amount of 0.3 g/L exhibited the best efficiency for the degradation of TZ and PB, and then a slight decrease in the degradation efficiency occurred with the rise of the catalyst dose.

As the amount of photocatalyst increases, more and more active sites are created on the surface of the photocatalyst, which leads to an increase in the formation of hydroxyl radicals. Increased amount of photocatalyst has a negative effect because aggregation of catalyst particles has a screening effect that prevents photons from reaching the inner catalyst surface.

However, an excessive dose of the photocatalyst decreases light penetration and scattering via the protective effect of the suspended particles and thus reduces the rate of photodegradation.

4.2.5 Efficiency of the regenerated photocatalyst

The reusability of photocatalyst is crucially important for industrial application due to the cost associated with the synthesis of the catalyst. Therefore, it’s important to study the repeating stability of the prepared catalysts. The catalyst was utilized multiple times to demonstrate its efficiency and cost-effectiveness. The catalyst was separated after each dye degradation experiment, washed with distilled water, dried at 100°C for 24 h and reused for further studies.

As seen in Fig. 17 the photocatalytic activity of the synthesized photocatalyst was nearly constant for three times cycling for both dyes. The MgO-La₂O₃-10% MMO reveals a high photostability and denotes excellent reusability in these experiments, with small deactivation, which can be attributed to residual dye adsorbed on the surface of the photocatalyst. The findings suggest that MgO-La₂O₃-10% MMO may have potential application in the environmental remediation issues as stable and effective material.

4.3 Analysis of response surface methodology for the TZ and PB photodegradation

The influence of different parameters on the degradation process efficiency of TZ and PB was investigated by Design Expert 11 software, and the experiments were planned through the CCD to optimize the main operating parameters in order to achieve a maximum photodegradation rate of both dyes in aqueous solutions. The CCD matrix and experimental results obtained in the photocatalytic degradation runs are given in Table 3.

The second order polynomial response equation (Eq. (5)) was employed to correlate the dependent and independent variables:

$$Y={\beta }_{0}+\sum _{i=1}^{k}{\beta }_{i}{x}_{i}+\sum _{i=1}^{k}{\beta }_{ii}{x}_{i}^{2} +\sum _{i<j}{\beta }_{ij}{x}_{i}{x}_{j}+\epsilon$$

In this expression, ${\beta }_{0}$ represents the interception term, x_i and x_j are the variable parameters of i and j with number k, ${\beta }_{i}, {\beta }_{ii} and {\beta }_{ij}$ represents the linear, quadratic and the interaction parameter coefficients, respectively. Lastly, the term "ɛ" indicates the random error.

Based on these outcomes, an empirical relationship between the response and independent parameters established and expressed by the proposed second-order polynomial equations 6 and 7 for TZ (Y₁) and PB (Y₂), respectively.

Y₁ = 95.41-0.0958${X}_{1}$-0.4962 ${X}_{2}-$1.4${X}_{3}$+ 0.4239${X}_{4}+0.0286{X}_{1}{X}_{2}-0.1876{X}_{1}{X}_{3}$+${0.0130 X}_{1}{X}_{4}-$ ${0.0728 X}_{2}{X}_{3}$+${0.0994 X}_{2}{X}_{4 }-0.0644{X}_{3}{X}_{4 }-0.0608{X}_{1}^{2 }+0.0560{X}_{2}^{2}-0.0139{X}_{3}^{2}+0.0703{X}_{4}^{2}$ (6)

Y₂ = 94.27–0.3570 ${X}_{1}$- 0.7385 ${X}_{2}-$0.2759 ${X}_{3}$+ 0.5996 ${X}_{4}-0.0396 {X}_{1}{X}_{2}-0.9727 {X}_{1}{X}_{3}$+${ 0.1551 X}_{1}{X}_{4}+$ ${0.0499 X}_{2}{X}_{3}$+${0.8827 X}_{2}{X}_{4 }+0.3066 {X}_{3}{X}_{4 }+1.22 {X}_{1}^{2 }+0.1063 {X}_{2}^{2}-0.2846 {X}_{3}^{2}-1.06{X}_{4}^{2}$. (7)

Photocatalytic efficiencies (R(%)) have been predicted by Eqs. (6) and (7) and presented in table 3. These results indicated good agreements between the experimental and predicted values of degradation efficiency. The predicted values were found to match the experimental values with a good fit with R² = 0.9327 and R² = 0.9574 for TZ and PB, respectively.

This suggests that 93.27% and 95.74% of the variations for TZ and PB removal are explained by the independent variables and this also signifies that the model does not explain only about remaining 6.73% and 4.26% of variation.

Table.3. The 4-factor central composite design matrix and the value of response function (degradation efficiency (%))

Run	Initial dye concentration (mg/L)	pH	Photocatalyst dosage (g/L)	Reaction time (min)	Degradation efficiency (%)
					Observed		Predicted
					TZ	PB	TZ	PB
1	15	5	0.4	130	95.413	94.629	95.45	94.27
2	15	7	0.4	130	94.322	93.182	94.45	93.22
3	20	4	0.3	150	97.257	93.642	97.67	94.34
4	15	3	0.4	130	97.174	96.457	96.44	96.17
5	10	6	0.3	110	96.223	92.657	96.02	93.35
6	20	6	0.5	150	93.457	92.403	93.88	92.67
7	20	4	0.5	110	93.915	94.485	94.03	94.4
8	10	4	0.5	150	95.128	94.175	95.07	94.63
9	20	6	0.3	150	97.175	94.402	96.68	94.45
10	15	5	0.4	130	95	93.772	95.45	94.27
11	5	5	0.4	130	95.214	99.573	95.64	99.87
12	15	5	0.6	130	92.481	92.914	92.65	92.58
13	20	6	0.5	110	93.221	90.28	93.03	91.18
14	15	5	0.4	130	94.645	94.798	95.45	94.27
15	10	6	0.5	150	94.245	95.011	94.07	95.1
16	15	5	0.4	90	94.364	91.222	94.6	91.24
17	15	5	0.4	130	95.525	93.646	95.45	94.27
18	10	4	0.3	110	96.741	97.117	97.01	96.61
19	20	6	0.3	110	95.662	94.885	95.83	94.19
20	15	5	0.4	130	95.525	94.788	95.45	94.27
21	10	6	0.3	150	96.665	93.147	96.87	92.99
22	25	5	0.4	130	95.348	98.987	95.26	98.44
23	15	5	0.4	170	97.247	89.117	96.3	88.84
24	15	5	0.2	130	98.456	93.598	98.24	93.68
25	20	4	0.5	150	94.217	92.55	94.88	92.36
26	10	4	0.5	110	94.785	96.842	94.22	97.29
27	15	5	0.4	130	96.328	93.98	95.45	94.27
28	10	4	0.3	150	97.248	93.122	97.86	92.72
29	10	6	0.5	110	93.437	95.17	93.23	94.23
30	20	4	0.3	110	97	97.197	96.82	97.61

The significance and adequacy of the quadratic RSM with regard to the interactions of the independent parameters was explained by analysis of variance (ANOVA). and the obtained results are displayed in Table 4. Degrees of freedom (D_f), sum of squares and mean squares were determined. Similarly, the F and p values were utilized to exhibit the significance of the quadratic model terms. The terms with high F-values and p-values less than 0.05 were considered significant and those with low F-value and p-values more than 0.05 were considered insignificant. The F-value obtained for TZ and PB 14.85 and 24.06. are clearly greater than the tabulated F (2.352 at 95% significance) confirming the adequacy of the model for processing the data. On the other hand, the proposed quadratic model can effectively analyze 95% of the data. Moreover. the F-value for the lack of-fit term (LOF) for TZ and PB were 0.801 and 1.77 that were smaller than the critical F0.05. 10.3 = 8.79 confirmed the goodness of the model from a statistical point of view. A high LOF indicates that the adjusted model is unable to fit the data adequately, whereas a non-significant term indicates the variation in the data according to the fitted model. Another evidence for the quality of the applied model is regression coefficient values presented in table that were obtained from the ANOVA. In this instance, the multiple coefficients of determination R², the adjusted one and the predicted one are close to the unity, confirming the accuracy of the suggested models and their efficiency in processing the data. Once the R² value approaches the adjusted R² value, it means that the proposed models contain the significant parameters and that all the selected parameters have an essential role in the response variable. Here, adjusted R² values for TZ and PB (0.8699. 0.9176) were very close to the corresponding R² value.

In an excellent model in RSM, to obtain a good design space the value of the adequate precision (or S/N) must be higher than 4 (Karimifard et al. 2018). In this study, the adequate precision was 14.89 and 23.96 for TZ and PB. respectively. Also, the coefficient of variance (C.V. %) value for both dyes (see Table 5) were very low which indicates the plausibility and reproducibility of the model. Thus, the proposed model can be applied to predict the degradation efficiency of TZ and PB by MgO-La₂O₃ MMO heterostructured catalysts.

Table 4

Analysis of variance (ANOVA) for fit of degradation efficiency from central composite design
Source of variation	Sum of square		D_f		Mean square		F-value		P-value
Source of variation	TZ	PB	TZ	PB	TZ	PB	TZ	PB	TZ	PB
Model	58.51	142.55	14	14	4.18	10.18	14.85	24.06	< 0.0001	< 0.0001
X₁	0.2204	3.06	1	1	0.2204	3.06	0.7832	7.23	0.3901	0.0168
X₂	5.91	13.09	1	1	5.91	13.09	21	30.93	0.0004	< 0.0001
X₃	46.81	1.83	1	1	46.81	1.83	166.31	4.32	< 0.0001	0.0554
X₄	4.31	8.63	1	1	4.31	8.63	15.33	20.39	0.0014	0.0004
X₁X₂	0.0131	0.025	1	1	0.0131	0.025	0.0466	0.0592	0.832	0.8111
X₁X₃	0.5633	15.14	1	1	0.5633	15.14	2	35.76	0.1776	< 0.0001
X₁X₄	0.0027	0.3847	1	1	0.0027	0.3847	0.0096	0.9089	0.9232	0.3555
X₂X₃	0.0847	0.0399	1	1	0.0847	0.0399	0.3009	0.0943	0.5914	0.763
X₂X₄	0.158	12.47	1	1	0.158	12.47	0.5614	29.45	0.4653	< 0.0001
X₃X₄	0.0663	1.5	1	1	0.0663	1.5	0.2356	3.55	0.6344	0.079
${X}_{1}^{2}$	0.1014	40.92	1	1	0.1014	40.92	0.3602	96.67	0.5574	< 0.0001
${X}_{2}^{2}$	0.0859	0.3097	1	1	0.0859	0.3097	0.3052	0.7317	0.5888	0.4058
${X}_{3}^{2}$	0.0053	2.22	1	1	0.0053	2.22	0.0189	5.25	0.8926	0.0369
${X}_{4}^{2}$	0.1357	30.6	1	1	0.1357	30.6	0.4821	72.29	0.4981	< 0.0001
Residual	4.22	6.35	15	15	0.2814	0.4233
Lack of Fit	2.6	4.95	10	10	0.2599	0.4952	0.801	1.77	0.6433	0.2741
Pure Error	1.62	1.4	5	5	0.3245	0.2795
Cor Total	62.73	148.9	29	29

Table 5

Statistical parameters for the proposed quadratic model by ANOVA.
R²		Adjusted R²		Predicted R²		Adeq Precision		C.V. %
TZ	PB	TZ	PB	TZ	PB	TZ	PB	TZ	PB
0.9327	0.9574	0.8699	0.9176	0.7241	0.7949	14.8912	23.9623	0.5558	0.6902

Another evidence for the validity of the applied model in the RSM approach is influencing plots which are presented in Fig. 18. As can be seen from the figure the predicted data were randomly distributed around their experimental values and no tendency was observed for TZ and PB.

3D response surface plots are a useful tool for visualizing the depiction of coexistent interaction effects between binary influence test variables. The response surface plots for TZ and PB degradation obtained in this study from Eqs. (6) and (7) are displayed in Figs. 19, 20 and 21. Responses were represented as a function of two factors, with the other variablebeing constant. The findings proved that the interactions of the four mentioned parameters were significant. Figure 19 illustrates the effect of pH and initial dye concentration on photocatalytic degradation efficiency of TZ and PB for an initial dye concentration of 0.3 g/L and reaction time of 150 min. As it is obvious from the figure, the best response was obtained at a pH of about 4 through the whole range of the initial dye concentration (10–20 mg/L). This can be attributed to various factors. As mentioned in Section 4.2.2. the pH_pzc of the catalyst was estimated at 7.8. At pHs < 7.8. the protonated dye molecules can be repelled by the positively charged surface of the catalyst. At pH greater than 7.8, the catalyst surface has a negative charge, which creates an attractive force between it and both dyes, which have been deprotonated and behave like hydrophobic molecules (Mossavi et al. 2022).

The interaction effects between the dose of the catalyst and initial dye concentration are displayed in Fig. 20. As it is obvious. the degradation efficiency was diminished by raising the dose of catalyst, whereas the best degradation efficiency was achieved in dose of catalyst around 3 g/L. When the pH and reaction time are constant, the active sites on the photocatalyst surface diminish with the enhancement of the dyes concentration, which reduces the degradation efficiency.

The interaction effects between the initial dyes concentration and reaction time in Fig. 21 show the best degradation efficiency of TZ in the moderate reaction time (about 150 min) thorough the used range for the initial concentration of TZ. At the same time. Figure 20 shows that at a moderate concentration of PB (10 mg/L) the best degradation of PB was obtained through the whole range of the reaction time.

In summary, the optimal RSM run includes pH 4, catalyst dosage 0.3 g/L, initial concentration of TZ 20 mg/L and reaction time 150 min for the maximum degradation efficiency of TZ (97.31%). On the other hand, the optimum values of the factors for the maximum degradation efficiency of PB were 10 mg/L, 4, 0.3 g/L and 150 min for initial dyes concentration (X₁), pH (X₂), photocatalyst dose (X₃) and reaction time (X₄), respectively (94.46%).

4.4 The GBR model performance

Based on the same data selection as employed in the RSM model, GBR analysis was conducted to establish a nonlinear algorithm for predicting the photocatalytic degradation of the tartrazine and patent blue. The initial dyes concentration, pH, photocatalyst dose and reaction time were considered as input data for the model, and degradation efficiency was described as output. The experimental datasets were randomly subdivided into three subsets, including the training, testing, and validation (80%, 10%, and 10% of the set points were applied, respectively) in the aim of predicting the capabilities of the RBG for the unobserved data set. The mean square error (MSE) and regression analysis coefficient (R²) were used to evaluate its performance.

It's worthy to note, that before starting the training of the GBR model, novel non-linear features were added consisting of polynomial combinations (degree 3) of the input data in order to improve the predictive quality of the model. Subsequently, the influence factor data as inputs were first normalized by employing Eq. (8) before building the machine learning model.

$${x}_{i}^{*}=\frac{{x}_{i}-\mu }{\sigma }$$

where ${x}_{i}^{*}$and x_i correspond to the normalized and original values of input parameters, respectively; µ and $\sigma$ represents the mean value and standard deviation for each parameter , respectively. The procedure for regularizing the GBR algorithm generally consists of optimizing two parameters that are max_depth and allowing to set it as n_estimators, which corresponds to the number of regression trees constructed based on a bootstrap stages that will be performed and max_depth is the limits of the number of nodes in the tree. GBR is effective enough to avoid overfitting, therefore a high number generally gives better performance.

The GBR model was fitted with variable values for n_estimators (100–5000). The findings suggest that the optimal tuning parameters for GBR model of TZ and PB are achieved based on the n_tree of 750 and 700, respectively. The max_depth parameter determines the maximum depth of each tree. The default value for max_depth are 18 for TZ and 12 for PB.

Regarding the training data set, the coefficient of determination R² values of 0.9882 and 0.9868 for TZ and PB, respectively, while the MSE values for TZ and PB (0.0228,0.0629) were very low, for GBR model are obtained. This implies that 98.82% and 98.68% of the variations for photocatalytic decolorization of TZ and PB are attributed to the independent variables and this also implies that the model does not explain only about 1.18% and 1.32% of the variation, which confirms the high accuracy of the GBR model for modelling and prediction of the dyes elimination.

4.5 Possible photocatalytic mechanism

The photocatalytic degradation performance depends on many factors including the band gap energy and electron-hole pair recombination rate of the photocatalyst. The photocatalysts elaborated in this research are mainly based on the modification of active MgO by uniform distribution of La through the LDH structure. In general, both MgO and La₂O₃ are irradiated under UV visible as results in the propagation of an electron from valence band (VB) to the conduction band (CB) and the generation of hole or electron deficiency in the VB as illustrated in Fig. 22.

According to the mechanism described above, the energy band diagram of a semiconductor photocatalyst illustrates the role of light absorption, oxidative and reductive capabilities of photogenerated charge carriers, and ultimately determines photocatalytic activity. The electrons-holes charges are photogenerated through excitation process on the surface of MgO/La₂O₃ by ultraviolet-visible light. According to the literature, the valence band potential of La₂O₃ (5.2 ev) is more negative when compared with MgO (7.8 ev) (Kumari et al. 2019; Pathak et al. 2016). In fact, based on the results of the UV–vis diffuse reflectance spectroscopy, the combination of the two catalysts resulted in the formation of a smaller band gap (3.05 eV), thereby enhancing the photocatalytic activity in the visible light zone. As a result, the MgO-Li₂0₃-MMO needs less energy to form electron-hole pairs, allowing for more photons to be used, leading to improved photocatalytic activity. A large number of photogenerated charges accumulated on the surface of catalyst, permitted the photodegradation of dye based on the following reactions:

MgO/La₂O₃ + hv → ${h}_{VB}^{+}+{e}_{CB}^{-}$ (9)

Reduction processes: ${e}^{-}+{O}_{2}\to { O}_{2}^{.-}$ (10)

${ O}_{2}^{.-}$ + ${H}^{+}$ →${HOO}^{.}$ (11)

${HOO}^{.}$+ ${HOO}^{.}$→${H}_{2}{O}_{2}$ + ${O}_{2}$ (12)

${H}_{2}{O}_{2}$+ ${ O}_{2}^{.-}$→${ OH}^{-}$ + ${ OH}^{.}$ + ${O}_{2}$ (13)

${H}_{2}{O}_{2}$+ $h$→${ OH}^{.}$ (14)

Oxidation processes: ${H}_{2}O+{{La}_{2}{O}_{3}( h}^{+})\to {H}^{+}+{OH}^{.}$ (15)

$${ OH}^{-}+{ {La}_{2}{O}_{3}( h}^{+})\to {OH}^{.}$$

$Dye$ + ${HO}^{.}$→ Dye (${HO}^{.}$) → Intermediates +${HO}^{.}$→ CO₂ + H₂O (17)

In the present paper, a series of MgAl-%La-CO₃ LDHs with various contents of La (5, 10 and 20 wt%) have been successfully prepared by co-precipitation process under constant pH and employed as precursors for the synthesis of mixed oxides MgO-La₂O₃ MMO. The synthesized heterostructure catalysts were characterized employing various physicochemical techniques including XRD, SEM/EDX, FTIR, TGA and ICP-AES. The incorporation of La into the lattice was confirmed through FTIR and Raman characterization. The MgAl-%La-CO₃ LDHs sample exhibits a typical hydrotalcite structure, whose crystallinity decreases as the amount of La increases as a result of partial Al substitution. The photocatalytic performance of the as-synthesized was assessed for the degradation of textile dyes, TZ and PB as a model of organic pollutants in an aqueous medium under UV-visible irradiation. The effect of the La³⁺, initial concentration of dye and initial pH was investigated. It was found that the presence of La absolutely improved the photocatalytic activity of the MgAl-MMO for the degradation of both studied pollutants under UV-Visible irradiation. All catalysts exhibited improvedeffects for both dyes, but the best performance was reached with the MgO-La₂O₃-10% MMO catalyst. The second interest of this study was to assess the performance of two AI techniques, namely RSM and GBR, in predicting the removal of two model pollutants and estimating the values of new observations. The MSE and R² statistical criteria were explored to evaluate the predictive performance of the two AI approaches. The GBR model developed from the reduced experimental data performed significantly better in the validation experiments, meaning that this model was able to capture the nonlinearities in the experimental data more effectively than the RSM technique, with a regression coefficient greater than 98% for the photocatalytic removal efficiency of both dyes. From this study, it can be seen that this photocatalyst presents a exceptional performance to be applied for the degradation of recalcitrant organic pollutants such as dyes. Also, this paper can be considered as a stimulus to encourage and inspire other scientists to conceive and optimize technologies such as the photodegradation process for efficient wastewater treatment based on AI techniques.

Ethics approval and consent to participate

Not applicable

Consent for publication

Not applicable

Competing interests

The authors declare that they have no competing interests.

Funding

Not applicable

Availability of data and materials

All data and materials generated or analyzed during this study were included in 661 this published article.

Authors' contributions

N. Taoufik: Conceptualization, Investigation, Formal analysis, Writing – original draft preparation; F.Z Janani: Formal analysis, Writing – original draft; H. Khiar: Formal analysis, Writing – original draft. M. Sadiq: Writing – original draft; M. Abdennouri: Analyzed and interpreted the data; Writing e review & editing, Data curation. M. Sillanpää; Analyzed and interpreted the data; Writing e review & editing, Data curation. M. Achak: Writing – original draft; N. Barka: Supervision, Analyzed and interpreted the data; Data curation, Writing e review & editing, Conceptualization, Validation.

Abdi J, Sisi AJ, Hadipoor M, Khatae A (2022) State of the art on the ultrasonic-assisted removal of environmental pollutants using metal-organic frameworks. J. Hazard. Mater 424:127558. 2022
Ali MA, Mubarak MF, Keshawy M, Zayed MA, Ataalla M (2022) Adsorption of Tartrazine anionic dye by novel fixed bed Core-Shell-polystyrene Divinylbenzene/Magnetite nanocomposite. Alexandria Eng J 61:1335–1352
Bhagat SK, Tung TM, Yaseen ZM (2021) Heavy metal contamination prediction using ensemble model: case study of bay sedimentation, Australia. J Hazard Mater 403:123492
Balthazard-Accou K, Emmanuel E, Agnamey P, Raccurt C (2019) Pollution of water resources and environmental impacts in urban areas of developing countries: case of the city of Les cayes (Haiti). Environmental Health - Management and Prevention Practices, IntechOpen
Banu V, Mohamed H, Arunprasath R (2019) Biological treatment of azo dyes on effluent by Neurospora sp isolated and adopted from dye contaminated site. J Text Inst 111:1–7
Barka N, Qourzal S, Assabbane A, Nounah A, Ait-Ichou Y (2011) Photocatalytic degradation of patent blue V by supported TiO2: Kinetics, mineralization, and reaction pathway. Chem Eng Commun 198:10
Bilinska L, Gmurek M (2021) Novel trends in AOPs for textile wastewater treatment. Enhanced dye by-products removal by catalytic and synergistic actions. Water Resource and Industry 26:100160
Barka N, Bakas I, Qourzal S, Assabbane A, Ait-Ichou Y (2013) Degradation of phenol in water by titanium dioxide photocatalysis. Orient J Chem 29:1055–1060
Bhagat SK, Tiyasha T, Tung TM, Mostafa RR, Yaseen ZM (2020) Manganese (Mn) removal prediction using extreme gradient model. Ecotoxicol Environ Saf 204:111059
Birjega R, Pavel OD, Costentin G, Che M, Angelescu E (2005) Rare-earth elements modified hydrotalcites and corresponding mesoporous mixed oxides as basic solid catalyst. Appl Catal A 288:185–193
Cocheci L, Lupa L, Olea NST, Muntean C, Negrea P (2020) Sequential use of ionic liquid functionalized Zn-Al layered double hydroxide as adsorbent and photocatalyst. Separ Purif Technol 250:117104
De Mattos NR, De Oliveira CR, Camargo LGB, da Silva RSR, Lavall RL (2019) Azo dye adsorption on anthracite: A view of thermodynamics. kinetics and cosmotropic effects. Sep Purif Technol 209:806–814
Elhalil A, Elmoubarki R, Farnane M, Machrouhi A, Mahjoubi FZ, Sadiq M, Qourzal S, Abdennouri M, Barka N (2019) Novel Ag-ZnO-La₂O₂CO₃ photocatalysts derived from the Layered Double Hydroxide structure with excellent photocatalytic performance for the degradation of pharmaceutical compounds. J Sci Adv Mater Devices 4:34–46
Elhalil A, Elmoubarki R, Farnane M, Machrouhi A, Mahjoubi FZ, Sadiq M, Qourzal S, Barka N (2018) Photocatalytic degradation of caffeine as a model pharmaceutical pollutant Mg doped ZnO-Al₂O₃ heterostructure, Environmental Nanotechnology, Monitoring & Management. 10:63-72.7
El Hanandeh A, Mahdi Z, Imtiaz MS, Modelling of the adsorption of Pb (2021) Cu and Ni ions from single and multi-component aqueous solutions by date seed derived biochar: Comparison of six machine learning approaches. Environ Res 192:110338
Fetimi A, Daas A, Benguerba Y, Merouani S, Hamachi M, Kebiche-Senhadji O, Hamdaoui O (2021) Optimization and prediction of safranin-O cationic dye removal from aqueous solution by emulsion liquid membrane (ELM) using artificial neural network-particle swarm optimization (ANN-PSO) hybrid model and response surface methodology (RSM). J Environ Chem Eng 9:105837
Friedman J, Hastie T, Tibshirani R (2000) Additive logistic regression: A statistical view of boosting (With discussion and a rejoinder by the authors). Ann Stat 28:337–407
Guo Y, Zhu Z, Qiu Y, Zhao J (2012) Adsorption of arsenate on Cu/Mg/Fe/La layered double hydroxide from aqueous solutions. J Hazard Mater 239–240:279–288
Gijin Kv, Chen YL, Oudheusden Bv, Gong S, de Wilt HA, Rijnaarts HHM, Langenhoff AAM (2021) Optimizing biological effluent organic matter removal for subsequent micropollutant removal. J Environ Chem Eng 9(5):106247
Ghaedi M, Ghaedi AM, Hossainpour M, Ansari A, Habibi MH, Asghari AR (2014) Least square-support vector (LS-SVM) method for modeling of methylene blue dye adsorption using copper oxide loaded on activated carbon: Kinetic and isotherm study. J Ind Eng Chem 20:1641–1649
Ghaedi AM, Vafaei A (2017) Applications of artificial neural networks for adsorption removal of dyes from aqueous solution: a review. Adv Colloid Interface Sci 245:20–39
Ghemit R, Boutahala M, Kahoul A (2017) Removal of diclofenac from water with calcined ZnAlFe-CO₃ layered double hydroxides: effect of contact time, concentration, pH and temperature. Desalin Water Treat 83:75–85
Han J, Zeng HY, Xu S, Chen CR, Liu XJ (2016) Catalytic properties of CuMgAlO catalyst and degradation mechanism in CWPO of methyl orange. Appl Catal Gen 527:72–80
Ismail GA, Sakai H (2022) Review on effect of different type of dyes on advanced oxidation processes (AOPs) for textile color removal. Chemosphere 291:132906
Kiwaan HA, Atwee TM, Azab EA, El-Bindary AA (2020) Photocatalytic degradation of organic dyes in the presence of nanostructured titanium dioxide. J Mol Struct 1200:127115
Kumari V, Kumar N, Yadav S, Mittal A, Sharma S (2019) Novel mixed metal oxide (ZnO.La₂O₃.CeO₂) synthesized via hydrothermal and solution combustion process – A comparative study and their photocatalytic properties. Materials Today:Proceedings. 19: 650–657
Karimifard S, Moghaddam MRA (2018) Application of response surface methodology in physicochemical removal of dyes from wastewater: A critical review. Sci Total Environ 640–641:772–797
Kooh MRR, Thotagamuge R, Chau YFC, Mahadi AH, Lim CM (2022) Machine learning approaches to predict adsorption capacity of Azolla pinnata in the removal of methylene blue. J Taiwan Inst Chem Eng 132:104134
Lacson CFZ, Lu MC, Huang YH (2022) Calcium-based seeded precipitation for simultaneous removal of fluoride and phosphate: Its optimization using BBD-RSM and defluoridation mechanism. J Water Process Eng 47:102658
Mossavi E, Hosseini Sabzevari M, Ghaedi M, Ahmadi Azqhandi MH (2022) Adsorption of the azo dyes from wastewater media by a renewable nanocomposite based on the graphene sheets and hydroxyapatite/ ZnO nanoparticles. J Mol Liq 350:118568
Mazaheri H, Ghaedi M, Ahmadi Azqhandi MH, Asfaram A (2017) Application of machine/statistical learning, artificial intelligence and statistical experimental design for the modeling and optimization of methylene blue and Cd(ii) removal from a binary aqueous solution by natural walnut carbon. Phys Chem Chem Phys 19:11299
Madondo NI, Chetty M (2022) Anaerobic co-digestion of sewage sludge and bio-based glycerol: Optimisation of process variables using one-factor-at-a-time (OFAT) and Box-Behnken Design (BBD) techniques. S Afr J Chem Eng 40:87–99
Pathak N, Ghosh PS, Gupta SK, Kadam RM, Arya A (2016) Defects induced changes in the electronic structures of MgO and their correlation with the optical properties: a special case of electron–hole recombination from the conduction band. RSC Adv 6:96398–96415
Quinto ML, Khan S, Picasso G, Sotomayor MDPT (2020) Synthesis, characterization, and evaluation of a selective molecularly imprinted polymer for quantification of the textile dye acid violet 19 in real water samples. J Hazard Mater 384:121374
Rizi NS, Shahzeydi A, Ghiaci M, Zhang L (2020) Photocatalytic degradation of cationic and anionic organic pollutants in water via Fe-g-C₃N₄/CF as a macroscopic photo-Fenton catalyst under visible light irradiation. J Environ Chem Eng 8:104219
Sadiq M, Elhalil A, Abdennouri M, Barka N, Bensitel M, Lamonier C (2021) Synthesis, characterization, and evaluation of a selective molecularly imprinted polymer for quantification of the textile dye acid violet 19 in real water samples. Bull Mater Sci 44:1–10
Sahnoun S, Boutahala M, Tiar C, Kahoul A (2018) Adsorption of tartrazine from an aqueous solution by octadecyltrimethylammonium bromide-modified bentonite: Kinetics and isotherm modeling. C R Chim 21:391–398
Soares APde MR, Carvalho FO, Silva CEF, Gonçalves AHS, Abud AKS (2020) Random Forest as a promising application to predict basic-dye biosorption process using orange waste.J. Environ. Chem. Eng103952
Soufi A, Hajjaoui H, Elmoubarki R, Abdennouri M, Qourzal S, Barka N (2022) Heterogeneous Fenton-like degradation of tartrazine using CuFe₂O₄ nanoparticles synthesized by sol-gel combustion. Appl Surf Sci Adv 9:100251
Su M, Xu R, Chen Z, Tang J, Ji C, Yang M, Liu Y, Zhang H, Chen D (2020) Heterostructured Bi₂O₂CO₃/rGO/PDA photocatalysts with superior activity for organic pollutant degradation: Structural characterization, reaction mechanism and economic assessment. Ecotoxicol Environ Saf 204:111112
Srivastava V, Maydannik P, Sharma YC, Sillanpää M (2015) Synthesis and application of polypyrrole coated tenorite nanoparticles (PPY@TN) for the removal of the anionic food dye’tartrazine’ and divalent metallic ions viz.Pb(II). Cd (II) Zn (II). Co(II). Mn (II) from synthetic wastewater. RSC Adv 5:80829–80843
Taoufik N, Sadiq M, Abdennouri M, Qourzal S, Khataee A, Sillanpää M, Barka N (2022) Recent advances in the synthesis and environmental catalytic applications of layered double hydroxides-based materials for degradation of emerging pollutants through advanced oxidation processes. Mater Res Bull 154:111924
Wang F, Li Y, Liang S, Min J, Chen Z, Cai Y, Fu X, Jiang X (2015) Preparation and photoluminescence properties of uniform LaCO₃OH triangular nanoplates. Mater Lett 148:114–117
Wei L, Yuan Z, Zhong Y, Yang L, Hu X, Zhang Y (2019) An improved gradient boosting regression tree estimation model for soil heavy metal (Arsenic) pollution monitoring using hyperspectral remote sensing. Appl Sci 9:1943
Ye Z, Li J, Zhou M, Wang H, Ma Y, Huo P, Yu L, Yan Y (2016) Well-dispersed nebula-like ZnO/CeO₂@HNTs heterostructure for efficient photocatalytic degradation of tetracycline. Chem Eng J 304:917–933
Yue R, Chen T, Ye Z, Benoit, Rahaman MS (2021) photo-Fenton graphene oxide membrane with improved perm-selectivity and self-cleaning ability for efficient dye removal under visible light irradiation. J Water Process Eng 44:102443
Zhang D, Qian L, Mao B, Huang C, Huang B, Si Y (2018) A data-driven design for fault detection of wind turbines using random forest and XGboost. IEEE Access 6:21020–21031. 2018
Zubair M, Aziz HA, Ihsanullah I, Ahmad MA, Al-Harthi MA (2022) Enhanced removal of Eriochrome Black T from water using biochar/layered double hydroxide/chitosan hybrid composite: Performance evaluation and optimization using BBD-RSM approach. Environ Res 209:112861

Download PDF

Editorial decision: Major Revision
31 Aug, 2022
Reviewers agreed at journal
13 Aug, 2022
Reviewers invited by journal
13 Aug, 2022
Editor invited by journal
12 Aug, 2022
Editor assigned by journal
10 Aug, 2022
First submitted to journal
23 Jul, 2022

You are reading this latest preprint version

MgO-La2O3 mixed metal oxides heterostructure catalysts for photodegradation of dyes pollutant: synthesis, characterization and artificial intelligence modelling

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials And Methods

2.1 Reagents

2.2 Synthesis of LDHs precursors and MgO-La₂O₃-MMO photocatalyst

2.3 Characterization techniques

2.4 Adsorption/photocatalytic degradation

3. Computational Details

3.1 Design of experiments

3.2 Gradient Boosting Regression modeling

4. Results And Discussion

4.1 Characterization of catalysts

4.1.1 X-ray diffraction (XRD)

4.1.2 Fourier transform infrared (FTIR) spectra

4.1.3 Thermogravimetric analysis (TGA)

4.1.4 SEM analysis

4.1.5 Chemical composition analysis

4.1.6 UV–vis

4.2 Photocatalytic study

4.2.1 Effect of the La doping on the photocatalytic reaction

4.2.2 Effect of the initial solution pH

4.2.3 Effect of initial concentration of TZ

4.2.4 Effect of photocatalyst dose

4.2.5 Efficiency of the regenerated photocatalyst

4.3 Analysis of response surface methodology for the TZ and PB photodegradation

4.4 The GBR model performance

4.5 Possible photocatalytic mechanism

5. Conclusion

Declarations

References

Status:

Version 1

MgO-La2O3 mixed metal oxides heterostructure catalysts for photodegradation of dyes pollutant: synthesis, characterization and artificial intelligence modelling

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials And Methods

2.1 Reagents

2.2 Synthesis of LDHs precursors and MgO-La2O3-MMO photocatalyst

2.3 Characterization techniques

2.4 Adsorption/photocatalytic degradation

3. Computational Details

3.1 Design of experiments

3.2 Gradient Boosting Regression modeling

4. Results And Discussion

4.1 Characterization of catalysts

4.1.1 X-ray diffraction (XRD)

4.1.2 Fourier transform infrared (FTIR) spectra

4.1.3 Thermogravimetric analysis (TGA)

4.1.4 SEM analysis

4.1.5 Chemical composition analysis

4.1.6 UV–vis

4.2 Photocatalytic study

4.2.1 Effect of the La doping on the photocatalytic reaction

4.2.2 Effect of the initial solution pH

4.2.3 Effect of initial concentration of TZ

4.2.4 Effect of photocatalyst dose

4.2.5 Efficiency of the regenerated photocatalyst

4.3 Analysis of response surface methodology for the TZ and PB photodegradation

4.4 The GBR model performance

4.5 Possible photocatalytic mechanism

5. Conclusion

Declarations

References

Status:

Version 1

2.2 Synthesis of LDHs precursors and MgO-La₂O₃-MMO photocatalyst