The catalytic ozonation of diazinon using nano-MgO@CNT@Gr as a new heterogenous catalyst: the optimization of effective factors by response surface methodology

In this research, the degradation of the insecticide diazinon was studied using a new hybrid catalyst consisting of magnesium oxide nanoparticles (nano-MgO), carbon nanotubes (CNTs), and graphite (Gr), nano-MgO@CNT@Gr, under various experimental conditions. This study shows the optimization of the nano-MgO@CNT@Gr/O3 process for diazinon degradation in aqueous solutions. Box–Behnken experimental design (BBD) and response surface methodology (RSM) were used to assess and optimize the solo effects and interactions of four variables, pH, catalyst loading, reaction time, and initial diazinon concentration, during the nano-MgO@CNT@Gr/O3 process. Analysis of regression revealed an adequate fit of the experimental results with a quadratic model, with R2 > 0.91. Following the collection of analysis of variance (ANOVA) results, pH, catalyst loading, and reaction time were seen to have significant positive effects, whereas the concentration of diazinon had a considerable negative impact on diazinon removal via catalytic ozonation. The four variables for maximum diazinon removal were found to be optimum (82.43%) at the following levels: reaction time, 15 min; pH, 10; catalyst dosage, 1.5 g L−1; and diazinon concentration, 10 mg L−1. The degradation of diazinon gave six kinds of by-products. The mechanism of diazinon decomposition was considered on the basis of the identified by-products. According to these results, the nano-MgO@CNT@Gr/O3 process could be an applicable technique for the treatment of diazinon-containing wastewater.


Introduction
In recent years, the release of insecticides into the environment has become a major concern. 1 These toxic compounds have negative effects on the reproductive and central nervous systems, and they also have carcinogenic effects. 2 Diazinon ((CH 3 ) 2 CHC 4 N 2 H(CH 3 )OPS(OC 2 H 5 ) 2 ) is a non-selective organophosphorus insecticide. It is widely used on lettuce, turf, citrus fruits, almonds, cotton, alfalfa, and other crops and fruits. 3 The World Health Organization (WHO) has classied diazinon as a "moderately hazardous" class II pesticide. 4,5 The toxicity dose associated with diazinon to aquatic organisms and its fetal dose for human is 350 ng L À1 and 90-444 mg kg À1 respectively. 6 Diazinon is a non-polar and relatively water-soluble substance, and it shows soil resistance. 7 Hence, its presence in surface water and groundwater is worrying. Also, conventional systems are not able to remove this insecticide from water sources. 8 The vapor pressure of diazinon is 1.4 Â 10 À4 mm Hg at 20 C and its Henry's law constant is 1.4 Â 10 À6 atm m 3 mol À1 , demonstrating that it will not volatilize simply from soil and water resources. 6,7 Advanced oxidation processes (AOPs) have been widely used for organic pollutant removal using more reactive radicals, mainly free hydroxyl radicals. 9 AOPs can involve a combination of chemicals and UV radiation or chemicals and ozone; for example, O 3 10 The combination of ozone with an active catalyst in a catalytic ozonation process (COP) increases the capability for mineralization, through converting O 3 into very reactive hydroxyl radicals (cOH), and increases O 3 consumption, through enhancing the solubility of gaseous ozone. 11,12 Catalytic ozonation has attracted researchers due to its low cost, ease of operation, increased ozone solubility, and capability for catalyst separation from the environment and its subsequent reuse. 13 Catalytic ozonation is an acceptable technique for promoting ozonation, which is typically done in two ways. Firstly, the production of cOH is facilitated. 14 Hydroxyl radicals, nonselective radicals with an excellent oxidation potential (2.80 eV) derived from ozone decomposition, have a stronger oxidizing capacity than ozone (2.07 eV) and so have greater efficiency than ozone for the degradation of recalcitrant compounds. 14,15 Also, O 3 directly reacts with the intermediates derived from organic materials in catalytic reactions, therefore demonstrating higher selectivity and removal efficiency towards organic pollutants. 9,14 Catalytic ozonation can be separated into two categories: homogeneous and heterogeneous. 10 Transition metals have been used in homogeneous catalytic ozonation processes. 16 Many synthesized catalysts, such as metal oxides, 14,17,18 metal oxides coated on various supports, 14,19,20 and porous materials, 16,19,21 have been used in heterogeneous catalytic ozonation processes. Heterogeneous catalytic ozonation is considered to encourage advanced oxidation processes in water and wastewater treatment. 21,22 Heterogeneous catalytic ozonation shows better performance, presenting high oxidation efficiency, low O 3 consumption, and a high reaction rate compared with single ozonation. 19,23 Thus, herein, a new hybrid catalyst was synthesized via the combination of three materials in three steps: mixing, sintering, and granulation. In this work, nano-MgO, which was previously found to be effective in the degradation of refractory organic pollutants 19,20,24,25 due to its high activity, 26,27 chemical stability, 25,26 and environmental friendliness, 19,28,29 was selected as the main part of the new catalyst. Graphite (Gr) was used as a catalyst support and provided catalyst reinforcement and porosity. 28,30,31 Carbon nanotubes were used as a support, aiding the uniform distribution of active compounds, the creation of highly active sites, 21,32 and the enhancement of the mechanical properties of the composite. 21,33,34 All components have been used previously in catalytic ozonation. 30,35,36 Kaolinite was added to the mixture to form bonds between the catalyst components. 37,38 The purpose of this study is to probe the catalytic activity of nano-MgO@CNT@Gr as a catalyst for the ozonation of diazinon in aqueous solutions. Herein three aspects are focused on as follows: (1) The synthesis of a new hybrid catalyst using three materials (nano-MgO, CNT and graphite), to enhance the stability, reusability and easy separation of the catalyst.
(2) The investigation of the removal and mineralization efficiency of diazinon via a COP.
(3) The exploration of the diazinon degradation mechanism via a COP. RSM was used for the optimization study, and BBD was used to determine the optimum parameters for the catalytic ozonation of diazinon and also to explain the interactions between the studied parameters.

Materials
All the chemicals used in this study were obtained from Merck (Germany). Only analytical grade chemicals were used. The MgO nanoparticles were prepared via a precipitation method. 39,40 Diazinon was purchased from Sigma-Aldrich (Germany) and was of analytical grade. All solutions were made with ultra-pure water (with an electrical resistance of 18 mU cm at 25 C). Diazinon stock solution (40 mg L À1 ) was prepared and stored in a refrigerator for further use. The CNTs (>95% carbon) were obtained from the Neutrino Company (Tehran, Iran). Graphite ($97 wt%) was purchased from Sinochem (Tehran, Iran). Puried kaolinite (Al 2 Si 2 ) 5 (OH) 4 ($90%) was purchased from Iran China Clay Industries Co. Real sewage samples were taken from the storage tanks of a pesticide manufacturing plant and used aer 10-fold dilution and ltration using membranes with a pore diameter of 0.45 mm (PTFE, Membrane Solutions).

Catalyst preparation
For synthesizing the magnesium nanoparticles, 0.2 M magnesium nitrate (MgNO 3 $6H 2 O) (98 wt%) was mixed into 100 mL of deionized water in a 1 L beaker. Aerward, 0.4 M sodium hydroxide was mixed into 100 mL of de-ionized water and this was then added to the prepared magnesium nitrate solution in a dropwise fashion. The solution was stirred for 4 h at 250 rpm to separate the magnesium hydroxide. The obtained substance was centrifuged at 3500 rpm for 10 min, and Mg(OH) 2 gel was obtained. The obtained gel was washed several times with methanol and de-ionized water, dried in an oven (FG-BF400E) at 90 C for 48 h, and then calcied at 450 C for 4 h. 41,42 For preparing the proposed composite, the raw materials (MgO nanoparticles, CNTs, and graphite: 81.5%, 14% and 4.5% by weight, respectively), kaolinite (up to 10% by weight of the raw materials), and distilled water (as needed) were mixed in a stainless steel cup using a planetary mill (NARYA-MPM 4*250) to create a homogeneous mixture. The obtained slurry was dried at 120 C for 48 h. The resulting composite was sintered in an electrical furnace with a corundum furnace tube (60 mm in diameter, 1000 mm in length; AZAR Furnaces, TF5/40-1720) under a hydrogen-and argon-controlled atmosphere for 2 h at 1350 C to create a solid compound. The prepared composite was crushed into ne granules (40-30 mesh). The obtained granules were washed two times with de-ionized water, dried at 90 C for 48 h, and stored in a desiccator for use in further experiments.

Ozonation procedure
Ozone was produced using an ozone generator (ARDA, MOG-5G-H) with a pure oxygen ($99.999%) gas feed. The ozone was continually introduced into a Pyrex cylindrical impinger, acting as the reactor, with a volume of 500 mL, which was operated in semi-batch mode with a constant O 3 gas ow rate of 3 L min À1 (1 g h À1 z 16.7 mg min À1 ). The reactor exhaust gas was treated with 2% KI solution (see Fig. S1 †). The diazinon removal experiments were conducted with 500 mL of reaction solution at the desired concentration; the preset amount of catalyst was added, and ozonation was then carried out for the specied time. The solution pH was adjusted using NaOH or HCl normal solutions at the desired level. All experiments were performed at 23 AE 3 C (room temperature). The catalyst was separated from solution aer 1 min of simple sedimentation. Residual dissolved ozone was quenched using Na 2 S 2 O 2 . The sample was ltered using a 0.2 mm PTFE syringe lter to separate probable catalyst particles, then the ltrate was analyzed for diazinon and total organic carbon (TOC). Also, control experiments were carried out utilizing single ozonation with no catalyst (SOP). The efficiencies of single ozonation and catalytic ozonation were compared under the desired conditions. The diazinon degradation efficiencies from the single ozonation and catalytic ozonation process experiments were calculated using the following equation: 43 where C 0 and C t indicate the initial and ultimate diazinon concentrations, respectively, of the solution from a specic experiment.
The degree of mineralization of diazinon in the SOP and COP experiments was calculated according to eqn (2): 44 where TOC 0 and TOC t indicate the initial and ultimate TOC, respectively, of the solution from a target experiment. The catalytic activity of the synthesized catalyst was obtained as follows: 9 Catalytic activity (%) ¼ [RE COP À (RE SOP + RE ads )] where RE COP , RE SOP , and RE ads are the diazinon removal efficiencies using nano-MgO@CNT@Gr/O 3 , the single ozonation process, and adsorption via nano-MgO@CNT@Gr without O 3 .

Response surface methodology
Design-Expert soware (version 7.0.0) was used for statistical analysis. Box-Behnken experimental design with RSM was utilized to specify the independent parameters and to optimize the experimental parameters for diazinon removal via the catalytic ozonation process. 45 Four independent variables, pH (X 1 ; 3-10), catalyst loading (X 2 ; 0.5-1.5 g L À1 ), ozonation time (X 3 ; 5-15 min), and diazinon concentration (X 4 ; 5-15 mg L À1 ), were xed with low, medium, and high levels, which were coded as À1, 0 and +1 (Table 1). Each independent parameter level was determined according to eqn (4) as follows: 44,45 where x i is the coded value, X 0 is the real value of the independent parameter at the center point, X i is the real value of the independent parameter, and DX is the step change value between low and high levels. According to BBD, the experimental design included ve replicates of the central points and 29 factorial points (see Table  2).
Experiments were performed based on the design points and data were assessed to determine the contributions of the four parameters, which were then correlated using a second-order polynomial equation as follows: 44,45 where y signies the predicted response; x i and x j represent the coded independent parameters, b 0 is a constant coefficient, b i , b ii and b ij are regression coefficients for linear, quadratic and interaction effects, respectively, 3 is the random error, and k is the number of desired independent variables. For analysis of the results, Design-Expert soware was used and ANOVA was used for quadratic and interaction term determination. 46

Characterization of the optimum catalyst
The chemical composition of nano-MgO/CNT/graphite was analyzed via X-ray diffraction (XRD) (PW1730 -Philips), scanning over 2q values between 10 and 80 . The functional groups on the surface of the catalyst were analyzed via FTIR spectroscopy (Avatar, Thermo) over wavenumbers ranging from 4000 to 400 cm À1 . The results of EDX microanalysis were used to show the existence of primary components in the synthesized catalyst. Microstructural evaluation was performed using FESEM (FE-SEM, Mira III, Czech Republic) equipped with energy dispersive X-ray (EDX) microanalysis to characterize the surface structure of the catalyst. Specic surface area (SSA) and pore size measurements of the catalyst particles were carried out via the Brunauer-Emmett-Teller (BET) method using N 2 adsorption/desorption data obtained using BELSORP mini II apparatus (Japan) at 77 K according to the BET and Barrett-Joyner-Halenda (BJH) isotherm models. The pH of the point of zero charge (pH PZC ) was determined via the conventional method. 39

Analytical methods
The potassium iodide (KI) method was used for determining the utilized ozone dosage. 47 The diazinon concentration was measured via HPLC (Agilent 1260, Santa Clara, CA) using a C18 column (10 cm length and an inside diameter of 4.6 mm, Eclipse Plus, Agilent) with a UV detector at 254 nm. A 65 : 35 (v/ v) mixture of acetonitrile and water was used as the mobile phase at a ow rate of 1.0 mL min À1 . TOC measurements were carried out using a TOC analyzer (Shimadzu, TOC analyzer, VCSH model). The COD concentration measurements were performed using the closed reex method. 47 Intermediate analysis was performed using GC/MS apparatus (Agilent 6890-5973N) equipped with a HP-5 MS capillary column (30 m length, 0.25 mm I.D., 0.25 mm thickness). Helium at a constant ow rate of 1 mL min À1 was used as the carrier gas. The MS interface temperature was 280 C and the ion source temperature was 230 C. The oven temperature program was as follows: 1.2 min at 40 C; 4.2 C min À1 to 50 C; 16.7 C min À1 to 80 C (1.2 min).

Catalyst characterization
An FESEM image of the nano-MgO@CNT@Gr composite is shown in Fig. 1a. In the FESEM image of the composite, it is evident that the shapes and sizes of the nano-MgO particles and CNTs, as seen in Fig. 1b and c, were preserved. Furthermore, it is quite clear that the composite has a heterogeneous and irregular shape and a porous-like surface. Using the method we describe, all components were successfully combined uniformly with each other.
The synthesized nano-MgO@CNT@Gr composite was characterized via X-ray diffraction (XRD). The XRD pattern is shown in Fig. 2 15 The mean crystallite size (D) of nano-MgO@CNT@Gr particles was calculated using the Debye-Scherrer formula 26,39,41 given by eqn (6): where D is the particle size, l is the X-ray wavelength of Cu Ka radiation (1.5418Å), k is the shape factor (0.9), b is the full width at half maximum (FWHM), and q is the angle of reection. The average size of the nano-MgO@CNT@Gr crystals was calculated to be 22.65 nm. The crystalline size of nano-MgO@CNT@Gr prepared in this study is close to the crystalline size of MgO reported in the literature. 15,39,48 Based on the results of XRD analysis, MgO nanoparticles were shown to be successfully coated on the CNTs and graphite. EDX analysis (Fig. 3) conrmed that nano-MgO@CNT@Gr is made up of magnesium (Mg; 27.55%), carbon (C; 35.04%), oxygen (O; 32.63%), and small amounts of aluminum (Al; 2.13%) and silica (Si; 2.65%). Therefore, all the components of nano-MgO@CNT@Gr have been preserved during the synthesis process.
From the FTIR spectrum of the nano-MgO@CNT@Gr composite (Fig. 4), the strong broad band at 3430 cm À1 can  This journal is © The Royal Society of Chemistry 2020 be attributed to the stretching mode of surface hydroxyl groups (O-H) present on the surface due to the physical adsorption of molecular water. The peaks at 1630 and 1735 cm À1 are related to the bending mode of these hydroxyl groups. 48 The wide band at 1457 cm À1 and band at 600 cm À1 can be assigned to the Mg-O stretching mode. The peak at 1171 cm À1 is associated with H À ions. 39 Therefore, it was deduced that the surface of nano-MgO@CNT@Gr is covered mainly in hydroxyl groups.
Based on the BET surface area and total pore volume results, the surface area of nano-MgO@CNT@Gr was calculated to be 221.631 m 2 g À1 . The total pore volume of nano-MgO@CNT@Gr (at P/P 0 ¼ 0.99) was found to be 0.155 cm 3 g À1 , with an average pore size of 3.65 nm. This shows, therefore, that nano-MgO@CNT@Gr prepared via the selected procedure is a mesoporous material (as per IUPAC classications) with a high SSA. The SSA is one of the main parameters affecting the catalytic activity of a substance, due to the provision of available sites for the adsorption of pollutants and/or the production of reactive species. Hence, preparing catalysts with a high SSA is one of the main points being focused on by researchers in this eld. 39 Thus, it was expected that the synthesized nano-MgO@CNT@Gr would have high catalytic activity.

Model tting
Based on four-factor three-level BBD, 29 experiments were carried out to obtain the efficient removal of diazinon and determine the interactions between experimental factors, including ve replications at the center point. Modeling of the quadratic polynomial equation between the response and independent variables was applied, which includes pH (X 1 ), catalyst loading (X 2 ), ozonation time (X 3 ), and diazinon concentration (X 4 ). For optimized RSM, a sufficient model t should be gained to obtain rich and clear results; this is necessary to ensure the adequacy of the model. Table 1 presents the RSM factors and levels. The RSM experimental design and results are presented in Table 2. Variance analysis (ANOVA) of the model regression parameters for the removal of diazinon via a COP using the results from the carried-out experiments are shown in Table 3. The ANOVA results for the second-order quadratic regression indicate the higher signicance of the model. The model F-value is 10.21, with a p-value of less than 0.0001. This p-value of less than 0.05 indicates that the model terms for diazinon removal are signicant; it shows that only 0.01% of the total variation could not be explained by the model due to noise. The lack of t is the variation of the data around the tted model and tests the adequacy of a model t. The "lack of t F-value" of 2.07 shows that the lack of t was not meaningful relative to the pure error, and this model successfully tted the data.
The R 2 value of 0.9108 suggests that the empirical model could not describe only 8.92% of the total variation. This demonstrated the sufficiency of the quadratic t to navigate the design space. Therefore, the RSM developed in the present study for diazinon removal efficiency prediction was considered to be appropriate.
A small gap between R adj 2 and R 2 is acceptable for the judgment of the model correctness. The calculated R adj 2 of 0.8215 means that the independent variables did not describe  17.85% of the total variation. The variation coefficient (CV) indicates the model reproducibility and is represented as the percentage ratio between the standard error of the estimate and the mean observed response. A model CV of less than 10% can be considered as reasonably reproducible. 49 The model "Adequate Precision" measures the range of predicted values and their related error. The model CV was 2.88 (less than 10%), and the adequate precision value was 13.074 (higher than 4), implying that the model tness was desirable.
The ANOVA of the quadratic model coefficients for diazinon removal are shown in Table 4. The table shows that six terms (X 1 , X 2 , X 3 , X 4 , X 1 2 , and X 4 2 ) were judged to be statistically meaningful for diazinon removal. The model independent variables pH (X 1 ), catalyst loading (X 2 ), reaction time (X 3 ), and diazinon concentration (X 4 ), and the quadratic terms X 1 2 and X 4 2 were highly considerable based on a 95% condence level (p < 0.05). The order of the F-values of the four factors is as follows: time (X 3 ) > catalyst loading (X 2 ) > pH (X 1 ) > diazinon concentration (X 4 ), and it appears that time (X 3 ) has most considerable impact on diazinon decomposition. The actual values and predicted values of diazinon removal efficiency were close to the same straight line (see Fig. S2 †). This denotes that the predicted values attained from the model were in acceptable accordance with the obtained experimental data. As anticipated, pH, catalyst loading, and reaction time had signicant positive effects, while the concentration of diazinon had a considerable negative impact.

Analysis of the response surface
Three-dimensional curves of the response surfaces and twodimensional contour plots were obtained to determine the optimal values of the variables. The model response surfaces and contour plots were obtained with two variables held at the central level and the other two variables varied in the experimental ranges. The results are shown in Fig. S3-S8. † In Fig. S3, † the response surface and contour plot were obtained as a function of pH and catalyst loading. At the same time, the time and diazinon concentration were kept constant at 10 min and 10 mg L À1 , respectively. As shown in Fig. S3, † a higher diazinon removal efficiency (76%) was realized at 10 min and 10 mg L À1 diazinon concentration when the pH and catalyst loading variables were at higher values (10 and 1.5 g L À1 , respectively).
The impact of the pH and time variables on the diazinon removal efficiency is demonstrated in Fig. S4. † When the diazinon concentration and catalyst loading were kept xed at 10 mg L À1 and 1 g L À1 , respectively, the diazinon removal efficiency increased with increasing time and pH. The relationship between the factors, hence, was signicant in terms of optimization; however, the time and pH had a more signicant effect on the efficiency of diazinon removal.
To investigate the impact of pH and diazinon concentration on diazinon removal, tests were carried out at diazinon concentrations varying from 5 to 15 mg L À1 at various pH levels at a xed catalyst loading of 1 g L À1 and time of 10 min (see Fig. S5 †). The gure displays that the removal efficiency of diazinon increased with rising pH at the central level of diazinon concentration.
The impact of time and catalyst loading on the removal efficiency of diazinon is presented in Fig. S6. † When pH and diazinon concentration was kept xed at 7 and 10 mg L À1 , respectively, the diazinon removal efficiency increased with increasing time and catalyst loading. Therefore, the time and catalyst loading (due to an increased surface area) have a more considerable effect on the diazinon removal efficiency.
As shown in Fig. S7, † the catalyst loading and diazinon concentration were also varied. At the same time, the time and pH were kept constant at central levels. The diazinon removal efficiency increased with increased catalyst loading at the central level of diazinon concentration.
As shown in Fig. S8, † 73% diazinon removal efficiency was realized at the highest application time and with the diazinon concentration at the central level, while the pH and catalyst loading levels were kept xed at 7 and 1 g L À1 , respectively.

Validation of the model
According to the results obtained from the response surface model, the maximum theoretical removal efficiency of diazinon was attained under the optimal conditions as follows: pH, 10; catalyst loading, 1.5 g L À1 ; time, 15 min; and diazinon concentration: 10 mg L À1 . To investigate the reliability of the RSM, a series of additional experiments were performed in triplicate under the optimum conditions. An average diazinon removal efficiency of 82.43% was obtained from the experiments. The experimental value was in good agreement with the predicted maximum diazinon removal efficiency from the quadratic model, which indicates that the model can better express the experimental conditions. Through using a response surface model to optimize the experimental factors for diazinon removal, the accuracy and validity of the model approach and the reliability of RSM could be conrmed.

Catalytic mechanism
The pH of the solution plays an essential role in catalytic ozonation, as it can affect the catalyst surface properties, the dissociation constant (pK a ), and the production of active radicals. 25 The pH PZC value of synthesized nano-MgO@CNT@Gr in these experiments was 9.3 (data not shown). Thus nano-MgO@CNT@Gr is considered a strongly basic oxide substance with a high isoelectric point. 39 Therefore, it was expected that the surface of nano-MgO@CNT@Gr would be covered broadly by hydroxyl groups. 39 This is in close agreement with the strong broad band from OH observed in the FTIR spectrum. Surface hydroxyl groups are active sites for hydroxyl radical production. 50 Therefore, the larger the number of active sites, the higher the production of hydroxyl radicals; thereby, the better removal of organic matter is achieved. The predominance of hydroxyl groups on the surface of the catalyst can be related to surface hydroxylation due to the dissociative chemisorption of water molecules. 51 In summary, the synthesized nano-MgO@CNT@Gr compound consists of mesoporous nanoparticles with a very high specic surface area and a high density of basic surface functional groups. These features are rmly favorable for an ozonation catalyst.
Also, the O 3 concentrations in the inlet gas and off-gas were measured during the reaction (see Fig. 5). The former was controlled at 5.5 AE 0.2 mg L À1 . The off-gas O 3 concentration in the control group experiment (deionized water without diazinon or catalyst) immediately increased to 5.52 mg L À1 aer 4 min, which indicated that O 3 quickly saturated the solution.
In the single ozonation process, the O 3 concentration in the off-gas gradually increased and then it reached a steady-state within 30 min of about 4.6 mg L À1 . This result shows that the aqueous O 3 was consumed constantly. Following Henry's law, due to the consumption of O 3 , the solution remained unsaturated, so O 3 was continuously transferred from the gas phase to the liquid phase. In the COP, at rst, the off-gas O 3 concentration remained at zero, but aer 35 min it started to rise. The absence of O 3 in the off-gas indicated that O 3 was consumed immediately aer being transferred to the liquid phase. Similar to the single ozonation process, the ozone concentration in the effluent gradually increased in the presence of nano-MgO@CNT@Gr. These results demonstrate that nano-MgO@CNT@Gr can enhance O 3 degradation.
Organic substance reactions directly with O 3 and indirectly with cOH are typically observed during heterogeneous catalytic ozonation. 19 tert-Butanol (TBA) as a conventional scavenger can react with hydroxyl radicals more selectively with a higher reaction rate constant of 6 Â 10 8 M À1 s À1 while reacting with O 3 at a rate constant of 3 Â 10 À3 M À1 s À1 . 19 Therefore, to identify the presence of cOH in the reactions, TBA was broadly applied as a probe molecule. To study the effects of nano-MgO@CNT@Gr on the generation of cOH, 2 g (4 g L À1 ) of TBA was added to the reactor. As seen in Fig. 6, TBA addition to  This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 7718-7731 | 7725 the reaction decreased diazinon removal in both the single ozonation and nano-MgO@CNT@Gr/O 3 processes. This shows the presence of cOH radicals in the nano-MgO@CNT@Gr/O 3 process. Also, the effects of nano-MgO@CNT@Gr on O 3 decomposition and cOH generation were investigated.
As stated above, the addition of nano-MgO@CNT@Gr to the ozonation process catalyzes the oxidation of diazinon via the enhancement of ozone decomposition, so leading to the production of oxidative radical species in the form of cOH. Catalytic ozonation involves two mechanisms in reactions with organic substances: direct reactions with ozone and indirect reactions with produced radicals. Reactions can occur both at the surface of the catalyst and in solution. The proposed mechanisms can be described as follows: O 3 direct oxidation in solution: Radical oxidation in solution: O 3 direct oxidation on the catalyst surface: Radical oxidation on the catalyst surface: (nano-MgO@CNT@Gr)-S (cOH) 2 + diazinon / CO 2 + H 2 O + intermediate (15) (nano-MgO@CNT@Gr) Àdiazinon + cOH / CO 2 + H 2 O + intermediate (16) where S in (nano-MgO@CNT@Gr)-S represents Lewis acid sites on the nano-MgO@CNT@Gr particle surface. According to the results from TBA experiments, it can be concluded that cOH indirect reactions are the major mechanism involved in diazinon degradation in the COP, as shown in the reaction presented in eqn (16). pH analysis demonstrated that the surface hydroxyl groups played a signicant role in the diazinon catalytic ozonation process. The results showing the dissolved ozone variations demonstrated that the catalyst could enhance the production of hydroxyl radicals. Due to the low diazinon adsorption (around 2.77%) on the catalyst, another probable mechanism based on eqn (16) could be somewhat involved.
The removal of diazinon during single nano-MgO@CNT@Gr adsorption was about 2.77%, which is negligible and indicates that using nano-MgO@CNT@Gr as an adsorbent for the adsorption removal of diazinon is not viable. As shown in Fig. 6, about 88% of diazinon and 26% of the TOC was removed using nano-MgO@CNT@Gr/O 3 aer 30 min, whereas the diazinon and TOC removal values via single ozonation were about 37% and 12% aer the same time period, respectively. These results indicate that nano-MgO@CNT@Gr is highly effective for the COP. To better explain the catalytic activity of nano-MgO@CNT@Gr in the catalytic ozonation process, the kinetics of diazinon removal were compared according to the following reaction model (eqn (17) and (18)): 52 in which C 0 and C t indicate the initial and the residual amounts of diazinon, respectively, and k obs is the PFO reaction rate constant. According to the results, the PFO model shows a desirable t (R 2 > 0.93) with the diazinon removal efficiency. The pseudo-rst-order rate constant of nano-MgO@CNT@Gr/ O 3 (0.0079 min À1 ) was almost 4.6 times higher than that obtained via single ozonation (0.0017 min À1 ) (see Table 5).

The degradation pathway
The high degradation of diazinon has been observed during ozonation processes. Ohashi et al. (1994) identied several diazinon intermediates, including 2-isopropyl-6hydroxylmethylpyrimidine (IMP), diazoxon, and triethyl phosphate, during the ozonation process. 53 As has been illustrated by other authors, the degradation of organic compounds in the COP does not happen right away to form CO 2 , but rather it occurs through the formation of long-lived by-products. 54 In this study, for the determination of organic intermediates, a relatively slow catalytic ozonation process (16.7 mg O 3 per L, diazinon concentration: 10 mg L À1 , catalyst loading: 1.5 g L À1 ) was used, which allowed us to obtain slower kinetics and establish desirable conditions for the identication of by-products. GC/ Fig. 7 A schematic diagram showing the proposed diazinon degradation pathways during the nano-MgO@CNT@Gr/O 3 process. The numbers match those given in Table 6.
This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 7718-7731 | 7727 MS was used for the identication of the diazinon degradation by-products from the COP. Analysis of samples obtained from the nano-MgO@CNT@Gr/O 3 process permitted the identication of six intermediate products that were considered transformation compounds. Data relating to the six by-products are presented in Table 6. As shown in Fig. 7, two probable degradation routes were considered. The rst route includes the cleavage of the P]O (pyrimidine group) bond, yielding IMP (compound 1). 53,54 Studies have shown that IMP is much less toxic than its parent diazinon. 7,55 The oxidation of the P]S bond of diazinon via the substitution of sulfur by oxygen leads to the formation of diazoxon (compound 2). The double oxidation of diazinon results in the production of hydroxydiazoxon (compound 3). Besides, cOH attack can lead to the production of a hydroxy derivative of diazoxon. In the nano-MgO@CNT@Gr/O 3 system, the loss of the pyrimidine group was probably due to oxidative desulfuration as a result of hydroxyl radical attack at the thiono group (-P]S-O). Diazoxon was produced following hydrolysis or via an oxidative mechanism affecting diazinon directly. 56,57 These oxidative mechanisms have previously been observed in metabolic studies of most thiophosphates. 57 In the second route, the thiophosphoric portion of diazinon was preserved. Hydroxydiazinon (compound 4) or 2-hydroxydiazinon (compound 6) were produced via the hydroxylation of the primary or secondary carbon atoms of the isopropyl group. Upon further oxidation, diazinon methyl ketone (compound 5) and diazinon aldehyde were formed. Aer the oxidation of the P]S bond of diazinon to a P]O bond, the oxygen analogue diazoxon was formed, leading to the production of the hydroxylated derivatives. In a previous study it was reported that hydroxydiazinon (4) generated, via a dehydration reaction involving the 1-hydroxy-isopropyl group, an isopropenyl-substituted compound, which was subsequently oxidized and decarboxylated to the hydroxyethyl derivative. 57 Other degradation by-products were probably present in the catalytic ozonation system but were not identied via GC/MS because of its limited sensitivity and their low concentrations.

Reusability of the catalyst
Reusability and stability are major properties of a catalyst for its practical application in the degradation of pollutants. To evaluate the reusability of nano-MgO@CNT@Gr and the permanence of its catalytic activity, ve sequential COP experiments were carried out under optimal experimental conditions: pH ¼ 10; catalyst loading ¼ 1.5 g L À1 ; diazinon concentration ¼ 10 mg L À1 ; and time of reaction ¼ 15 min. The catalyst was separated via simple sedimentation aer each run. Then, the separated catalyst was washed with distilled water and dried for 24 h at 110 C in an oven and reused for further COP experiments. The diazinon removal efficiency was calculated aer each run, and the results are shown in Fig. S9. † The results showed that the diazinon removal efficiency for fresh nano-MgO@CNT@Gr was about 82.43% under the dened conditions. The diazinon removal efficiency was not signicantly affected aer ve successive cycles of nano-MgO@CNT@Gr reuse, remaining at more than 78%. These results indicate that nano-MgO@CNT@Gr could maintain its catalytic capacity as well as its stability aer ve successive applications. Furthermore, the excellent chemical stability of nano-MgO@CNT@Gr was further conrmed via FESEM analysis of used and fresh catalyst samples (see Fig. 8). No considerable changes in the FESEM image were observed aer ve sequential COP experiments using the catalyst. This nding indicates that nano-MgO@CNT@Gr not only shows excellent catalytic activity but also acceptable stability.

Conclusions
This study investigated the effectiveness of a new hybrid catalyst that combines three materials, nano-MgO, CNTs, and graphite, for diazinon degradation during catalytic ozonation.
Box-Behnken experimental design and RSM were performed to specify the independent variables and optimize the experimental parameters of pH, catalyst loading, reaction time, and initial diazinon concentration. Regression analysis indicated the desirable t of the experimental data with a second-order model with an R 2 coefficient of >0.91. ANOVA of the coefficients for each independent variable showed that pH, catalyst loading, and reaction time had signicant positive effects; in contrast, the diazinon concentration had a meaningful adverse impact on diazinon removal during catalytic ozonation. The optimum independent variable levels were found for maximum diazinon degradation (i.e., 82.43% diazinon degradation): reaction time, 15 min; pH, 10; catalyst dose, 1.5 g L À1 ; and diazinon concentration, 10 mg L À1 . The results showed that the nano-MgO@CNT@Gr/O 3 process is a suitable technique for diazinon degradation. Six kinds of diazinon by-products were identied during the nano-MgO@CNT@Gr/O 3 process. The diazinon degradation route was considered based on the distinguished intermediate products under the given set of reaction conditions. The advantages of the nano-MgO@CNT@Gr/O 3 process as an oxidative treatment are its simple handling and rapid degradation. Thus, this catalytic ozonation process could be used for wastewater treatment as a newly developed method to reduce the levels of other pesticides and endocrine disrupting chemicals.

Conflicts of interest
The authors declare no conicts of interest.