Photochemical degradation of dexamethasone by UV/Persulphate, UV/Hydrogen peroxide and UV/free chlorine processes in aqueous solution using response surface methodology (RSM)

ABSTRACT Dexamethasone (DEX) degradation was investigated using three UV-based advanced oxidation processes combined with sodium persulphate, hydrogen peroxide, and free chlorine in aqueous solutions in a bench-scale reactor. The goal was to determine which process would be the best option in terms of DEX degradation without the subsequent generation of toxic compounds. The process performance for DEX degradation followed the UV/S2O82−> UV/H2O2> UV/HOCl order under equal conditions. In the UV/S2O82− process, sulphate radicals led to a 70.4% DEX degradation. The experimental mineral inhibitory factors showed that nitrite and bicarbonate further decrease the efficiency of the UV/S2O82− process. In the optimal conditions, the UV/S2O8−2 process removed DEX by 100% and had a 74.1% mineralisation. Furthermore, the intermediates were detected using LC-MS/MS, and the cytotoxicity of the UV/S2O8−2 process effluent was evaluated using human embryonic kidney cultured cells. The treatment of DEX solution by the UV/S2O8−2 process significantly decreased the toxicity of the effluent. Therefore, it can be concluded that the UV/S2O8−2 process can be a reliable process for DEX degradation and detoxification in aqueous solutions.


Introduction
There has been an interest in recalcitrant organic contaminants such as pharmaceutical compounds that are extensively used in daily life.Due to inefficiency of conventional wastewater treatment process for their removal and their widespread use, the presence of these drugs in the environment is on the rise [1,2].The numerous pharmaceuticals present in drinking water are not easy to remove by conventional methods due to their physicochemical characteristics [3].Some of the pharmaceutical compounds persist against biological processes and physiologically affect humans and animals [4].Although they are in trace amount in aquatic systems, their continuous intrusion into the environment leads to the accumulations in aqueous media, posing adverse health effects in long term exposure [5].
Among pharmaceutical compounds, dexamethasone (DEX) is one of the most widely used glucocorticoid drug [6][7][8].Various concentrations have been reported for DEX in the effluent from wastewater treatment plants, hoggery wastewater, and surface water.DEX concentrations as high as 1050.90ng/L has been reported in wastewater samples [9,10].The presence of glucocorticoid hormone originating from DEX in water may lead to the toxicity in aquatic organisms [11][12][13].Accordingly, DEX presence in the aquatic environment is considered as an emerging environmental issue.
Various studies have been conducted for DEX removal from aqueous environment.Electric coagulation [14], adsorption [11], photocatalytical processes [15], etc. have been applied for DEX degradation.However, the investigated processes have been reported that they lead to the generation of secondary contaminants or presented relatively lower removal efficiencies.
Advanced oxidation processes (AOPs) are proposed as an effective technique to treat water containing non-biodegradable organic pollutants.AOPs are generally proposed as an alternative to the conventional treatment processes for the destruction of various types of organic contaminants from water.This is due to the ability of AOPs to produce highly reactive species such as hydroxyl radical (OH • ).UV-based AOPs have been recently applied for degradation of recalcitrant contaminant.Accordingly, various AOPs have been proposed such as Fenton and Fenton-like processes, persulphate-based processes and UV/H 2 O 2 process [16].In addition, UV/chlorine (UV/HOCl) process has also been investigated in recent years [17].
The traditional method for determining the optimum conditions of a process is costand time-consuming.The response surface methodology (RSM) helps examine the interaction effects of independent variables and makes it easy to optimise the investigated parameters with fewest possible experiments [18].
This study investigated the combined effect of various parameters, e.g. the initial DEX concentration, oxidant reagents' concentration, reaction time, radiation intensity, and pH on DEX oxidation from aqueous solutions by UV/S 2 O 8 −2 , UV/H 2 O 2 , and UV/ HOCl processes using the Box-Behnken design (BBD) of the RSM in Design Expert 10.0.1 (DOE) software.The efficiencies of DEX degradation at optimum conditions were also compared.The UV/S 2 O 8 −2 process with a 100% DEX removal efficiency was able to remove only 74.1% of the TOC due to the presence of DEX in the aqueous solution.Mineralisation results revealed that DEX decomposition leads to the production of intermediate products.Finally, the LC-MS analysis was adopted to investigate the generated degradation by-products and propose removal pathways of DEX photochemical degradation.

Materials
This research is an experimental study with an applied approach that has been performed on a laboratory and batch scale.The goal was to determine the efficiency of the combination of photochemical processes using UV/S 2 O 8 2− , UV/HClO, and UV/H 2 O 2 processes in aqueous solutions to remove DEX and examine the parameters affecting it.To this end, the variables of reaction time, oxidant dose, pH, and the initial concentration of DEX were defined, and the objectives were examined by performing experiments based on the RSM.

Applied materials
The chemical materials were sulphuric acid (H 2 SO 4 ) solution (0.  Germany).All the chemicals were used as received without further purification.All the solutions were prepared daily with de-ionised water purified by a Fomos system.

Experimental procedure
As shown in Figure 1 lamp to stabilise the UV emission intensity [19].For each experiment, a 10-mL sample was taken and regularly withdrawn from the photo-reactor for residual DEX concentration measurements.
To make samples with different concentrations of DEX as a contaminant, a stock solution of 100 mg.L −1 DEX was used.This solution was daily prepared using DEX powder and volume to 1 L of distilled water.To prevent the solution from exposure to direct light, it was covered with an aluminium foil and maintained at 4°C.

Analytical procedure
The DEX concentration was quantified by a Shimadzu Prominence Series highperformance liquid chromatography (HPLC) coupled with a Shim Pack VP-ODS C18 reversed-phase column (250 mm×4.6 mm, 5 μm) and an ultraviolet detector (Shimadzu UV-1600 spectrophotometer).Acetonitrile and water at a ratio of 40/60 (v/v) was selected as mobile phase with a flow rate of 1 mL/min.The column temperature was adjusted at 30°C and 20 μL of sample was injected for each test.UV detection at a wavelength of 213 nm was used to quantify the separated DEX [20].Based on the principles of this study, the percentage of DEX removal could be calculated using Equation 1: where C 0 and C t are the initial concentration of DEX and residual concentration of DEX, respectively.
Total organic carbon (TOC) was measured via a TOC analyser (TOC-VPCN, Shimadzu) equipped with an auto-sampler by catalytic oxidative combustion at 680°C, using an infrared detector to evaluate the samples' mineralisation of TOC values.The pH and temperature were determined by a pH metre Hack (HQ-USA) on site.The other parameters were determined based on the Standard Methods for the Examination of Water and Wastewater [21].
DEX oxidation by-products and the intermediates were identified by LC-MS/MS (Quattro Micro API micro mass Waters 2695, positive ion mode ESI mass spectra).The mobile phase was the same as the one used for HPLC with a flow rate of 0.2 mL.min −1 [22].Positive mode was selected for the ESI-S analysis.spray voltage of 3500 V was adopted for the full scan acquisition (150-400).other conditions of operation include: 380°C for vaporiser temperature, 5 L/min (N 2 ) for aux gas pressure of, 35 L/min (N 2 ) for sheath gas pressure of, 350°C for the capillary temperature, skimmer offset of −10 V, and tube lens offset of 110 V [22].
The experiments were conducted with different variables, e.g.solution pH, initial DEX concentration, oxidant reagents concentration, and reaction time.The influence of operating parameters was investigated.All the graphs were plotted and calculations were performed in Design-Expert version 10.0.3 and Microsoft Office Excel 2013.

Calibration curve validation, limit of detection (lod), and limit of quantification (LOQ)
The calibration curve validation for the HPLC equipment was conducted by injecting 200 μg mL −1 of the DEX standard solution, followed by preparing and injecting lower concentrations of nine DEX standard samples (170, 150, 120, 100, 70, 50, 20, 10, 5 μg mL −1 )  until the UV detector of the HPLC instrument did not detect concentrations < 5 μg mL −1 .Based on the linear regression analysis, the detected response was proportional to DEX samples' concentration (R 2 = 0.999).Equation ( 2) can be defined based on the standard deviation (σ) of the detected response and the slope of the calibration curve (S), where X is a proportional factor for the LOD (3.3) and the LOQ [20].
Based on the standard deviation (σ) of the detected response and the slope of the calibration curve (S), LOD and LOQ were 0.79 and 2.4 μg mL−1, respectively.

Design of experiment (DOE) for the optimisation of DEX degradation processes parameters
For optimisation of the process Statistical DOE and RSM have been widely used in recent years.It is used for investigation of the effect of independent variables and their interactions [23][24][25].Reduction in the number of experimental runs leading to less time and resource consumption makes the RSM design favourable for AOPs [26,27].
In the present study, the BBD statistic model was used to survey the effect of independent variables on the measured responses.Herein, various key parameters were optimised by using the BBD of RSM, and four independent variables were taken into account to investigate the removal efficiency of the DEX.Two level ranges for the initial DEX concentration, oxidant reagents' concentration, reaction time, and solution pH selected for the BBD were 2 to 20 mg.L −1 , 0.5 to 5 mmol.L −1 , 10 to 60 minutes, and 3 to 12, respectively.Table 1 is illustrating ranges of the independent variables along with their coded levels.The values in this table have been selected based on our preliminary experiments and similar studies reported by other researchers.The response factor was the percent removal of DEX after 60 min of reaction time.Details of BBD are given in Table S1.As shown in Table S1, there were 27 experimental runs.They were consisted of 16 factorial, eight axial, and two replicates at the central point [28].Assessing the pure error, controlling the model fitting adequacy and the repeatability of the results was investigated by replication of the central point.Furthermore, to reach the lowest possible systematic errors, all the experiments were conducted in a random.To achieve the optimal conditions for DEX removal using UV/persulphate, UV/H 2 O 2 , and UV/HClO, the experimental design was performed as a function of the main parameters.The behaviour of the system can be explained using the quadratic equation (Equation ( 3)).Least-squares regression method was adopted for data analysis to predict the process response and coefficients using the second-order equation: where Y is the predicted response by the model; β 0 is the model constant (intercept term); xi and xj are the coded values of the independent variables; and β i , β ii and β ij are the linear, quadratic, and interaction coefficients, respectively.
The statistical software Design-Expert 10.0.3 was used in the present study for designing of the experiments.Multiple regression analysis was selected for the polynomial model.Validation of the predicted model equation was conducted by ANOVA test.A p-value of < 0.01 at a 99% confidence level was adopted for the statistical significance of the model.Correlation coefficient (R 2 ) was used for the validity and fit quality of the polynomial model.Three-dimensional response surface and twodimensional contour plots which were obtained from the fitted quadratic equation was used for the investigation of the simultaneous interaction of the two factors on the process response.The validation of the model was investigated by measured and predicted values of DEX percent removal.Finally, for maximising the DEX oxidation level, the critical variables optimum values were obtained by numerical techniques in the software.

DXE increases both serum and urine albumin (ALB) due to various pathological causes.
Clinically, acute kidney injury is the reason for high serum albumin (sALB) [29,30].In addition, DEX may inhibit the cell proliferation [31].For investigation of the cytotoxicity of the DEX solution, before and after treatment by UV/PS process, the cultured human embryonic kidney (HEK) 293-cell were used in the MTT assay.The cells were obtained from National Cell Bank of Iran (NCBI).They were cultured at 37°C and 5% CO 2 in Dulbecco's Modified Eagle medium (DMEM) supplemented with 10% heat-inactivated foetal bovine serum.
The cell viability was investigated by exposing them to DEX solution before and after treatment by UV/PS process.A total of 7,000 viable cells with 100 μl of 10% of the FBS medium were transferred to each 96-well plate and incubated for 24 h (37°C and 5% CO 2 ) for stabilisation.To several wells of each plate only the culture medium was added as a control group.The tests were conducted by exposure of the cells with various concentrations of DEX solutions (2, 11, and 20 μg/mL) for 24 and 48 h.The cells were incubated with 100 μL of 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyl-tetrazolium bromide at the final concentration of 5 mg/mL for 4 h.The yellowish tetrazolium was transferred to the mitochondria of viable cells and converted by the enzyme to the purple crystals of formazan.The intensity of the purple colour indicated the number of viable cells.After 4 hours with the formation of formazan crystals, the culture medium containing tetrazolium salt was removed, and then 100 μL of DMSO was added to each plate.An Elisa reader (BioTek model) microplate reader at the wavelengths of 540 and 570 nm was used to measure the absorbance of the prepared solution.Then, the mean uptake of each treated group was divided by the mean uptake of the control group and multiplied by 100, demonstrating the percentage of survival of each group compared to the control group.The results are reported as means ± standard deviation (SD)., and HClO, it was observed that the removal efficiency of DEX at all test times in the combined state is higher than the application of UV rays alone.The results showed that < 10% DEX oxidation is observed by direct UV photolysis after 60 min of reaction time in the absence of oxidants.However, the removal efficiency of DEX increased in the presence of sodium persulphate, sodium hydroxide, and hypochlorite ions when the experiments were performed based on DOE (Table S1 oxidation potential (E 0 SO4˙− = 2.5-3.1 V) compared to HO˙ (E 0 HO˙ = 1.9-2.7 V), it has strong affinity to electron-rich contaminants leading to higher reaction rate constants.Furthermore, the quantum yield of persulphate is 40% higher than that of hydrogen peroxide under 254 nm radiations [17].According to the obtained results, the present study proceeded by investigating the UV/S 2 O 8 −2 process as the optimal process for DEX removal from aqueous solutions.

DOE
As mentioned before, the effects of four different independent variables on the process performance were investigated by BBD.Second-order fitting equation model was used to express the response functions for the process performance by analysing the experimental data.It was found that some of the terms were not significant at the 99% confidence level.Therefore, they were removed from model to increase the prediction capabilities in the study.Equation ( 4) is giving the reduced model based on the coded values.
where X1, X2, and X3 are independent variables of oxidant concentration (mmol.L −1 ), initial DEX concentration (mg.L −1 ), and reaction time, respectively.The adequacy of the model was determined by diagnostic plots.For the data normality detection the normal probability plot of the studentized residuals could be used which is a good graphical representation.As it is illustrated in Figure 2(a), it is observed that the data are normally distributed in the model response.Figure 2(b) is illustrating the actual values in front of the response predicted by the model.Evidently, it is found that the experimental results show a good correlation with the predicted values by the statistical model which is validating the predicted model [25].Consequently, it can be concluded that the secondorder polynomial model is quite satisfactory for the prediction of DEX removal in the investigated process.In addition, the ANOVA test was also performed to further verify the adequacy and significance of the predicted model.S2.
The significance of the regression model could be demonstrated by Fisher's F-value with a low probability value (P).The results show that (table S2), Fisher's F-value is 37.07 implying the high significant for the prediction of DEX removal by the model.It is also further confirmed by the low probability value of <0.0001 at a significance level of 0.01.For better comparison of the residual error with pure error, the LOF (Lack-of-Fit) test was performed.It was found that the P-value for LOF is also <0.05.The adequacy of the model was also evaluated by the residual error.Residual error is showing the difference between observed and predicted response values.The obtained LOF value indicates that the deficiency of the model to estimate the response factor is not significant.Therefore, model's adequacy is well confirmed.Furthermore, the adequate precision (AP) ratio was calculated to be 22.83 (> 4).It is also impling an adequate signal for the quadratic model.Hence, it can be deduced the model can be used for navigation of the design space.p-values are also used for the expression of the significance of variables and their interactions.If the F value is lower than 0.05 and more than Prob (95% confidence level), the model terms could be considered significant [25].Accordingly, X1, X2, X3, X1X2, X2X3, X1 2 , X2 2 , and X3 2 were significant terms with p-values < 0.05.While, the X0, X0X2, X0X3, and X0 2 were insignificant terms with p-values > 0.01 (X0, X1, X2, X3) which are removed from the model.Finally, the coefficient of determination (R 2 ) was applied for verification of the suitability and validity of the quadratic model [32].The calculations showed that the R 2 and adjusted R 2 values for the reduced model were 0.9498 and 0.8913, respectively, displaying satisfactory agreement.

Effect of initial persulphate concentration
The interactive effects of operational parameters on DEX removal have been illustrated by 3D response surface and 2D contour plots.The simultaneous effect of two factors have been investigated on the response at various ranges, while the other factors have been kept constant [33].
The effect of initial persulphate concentration as the oxidant, initial DEX concentration as the contaminant, reaction time, and pH on DEX removal efficiency is depicted in Figure 3(a-d).Based on Figure 3(a,b), the DEX oxidation has shown an increasing trend as the persulphate concentration increases from 0.5 to 5 mmol.L −1 .
The higher initial persulphate concentration, the higher sulphate radial will be generated.Consequently, DEX removal efficiency will be elevated.However, further increase of persulphate concentration (> 2.75 mmol.L −1 ) shows a slight increase in the DEX oxidation.Although higher persulphate concentrations produce more sulphate radicals, the removal efficiency does not rise significantly.Besides, many studies have reported that the removal efficiency decreases at excess concentrations of persulphate.This reduction in removal efficiency could be due the self-scavenging of sulphate radicals.Furthermore, presence of redundant persulphate in the solution may also lead to the scavenging of radical species with persulphate molecules in the solution [34].

The effect of initial DEX concentration process performance
Figure 3 is illustrating the effect of initial DEX concentration on the process performance.As shown in the figure, the higher initial concentration of DEX, the lower DEX removal efficiency is observed.In the same conditions, it is assumed that the constant amounts of sulphate and hydroxyl radicals are generated in the solution [35].Increase in the initial concentration of organic contaminant in the solution leads to the higher ratio of contaminant to the oxidising agents and therefore would decrease the probability of desired reaction to remove DEX.Moreover, the generated oxidation by-products in the solution may also cause side reactions that consume oxidising radical species [36].These byproducts diminish the DEX removal efficiency by competing with the initial DEX as a contaminant in response to radicals.Another adverse effect of these by-products on DEX removal is to reduce the probability of collisions and reactions between DEX and radical species [7].Accordingly, higher initial concentration of DEX deteriorated the UV/S 2 O 8 −2 process performance.

The effect of reaction time on DEX degradation
The effect of reaction time and pH on DEX removal efficiency is illustrated in Figure 3(c,d).
The DEX removal efficiency significantly increased as the reaction time increased in the solution.As expected, increase of the reaction time improves the generation of reactive sulphate radicals for DEX removal.The removal slope decreased after 30 minutes, such that there was no significant difference between 30 and 45 minutes.The removal efficiency increased with prolonging the reaction time, such that in the first 30 minutes, the maximum removal efficiency occurred, and then the slope of the graph (which is the same as the decomposition speed) decreased over time.This reduction in the removal rate is due to a decline in the concentration of contaminants and oxidants.

The effect of initial pH on DEX degradation
To identify the impact of pH on the degradation of DEX, the initial pH values were varied as 3, 7.5, and 12.The results showed that the DEX degradation rate increased with the decrease in pH.As shown in Figure 3(c,d), increasing the pH from 3 to 12 results in the removal of a lower percentage of DEX.With a further increase in pH, the removal efficiency of DEX decreased.However, the DEX degradation rate was almost constant after 35 min.It has been reported that sulphate radical could be scavenged by hydroxyl radical at alkaline conditions [37,38].Furthermore, SO 4 −• is converted into OH˙ under alkaline conditions.On the other hand, the oxidation potential of OH˙ reduces with increasing the pH values [39].In general, the reduction of removal efficiency at a high pH can be justified by reducing the production of sulphate radicals and decreasing the oxidation potential due to the change of oxidising species from sulphate radicals to hydroxyl radicals [40,41].Nevertheless, no significant effect was observed for pH on the degradation efficiency.The results showed that the final DEX removal efficiency of 100% is possible at all three pH levels (7.5, 3.5, and 12).It is quite clear that by altering other independent variables of the reaction; especially the concentration of persulphate, the initial pH does not affect the degradation efficiency of DEX in the UV/S 2 O 8 −2 system from pH 3 to 12. Our results are in accordance with the result of HOU et al. [36].Based on the ANOVA and the simultaneous effect of other independent parameters, e.g. the initial contaminant concentration, oxidant content, and process time, pH with P = 0.464 and F = 0.65 had no significant effect on DEX removal efficiency.

Determining the optimum conditions based on the DOE
To achieve optimal test conditions, the statistical DOE and RSM were used.The BBD statistical model, which is an RSM model, was adopted.Using the proposed quadratic model, the program searched the available design space to find the optimal value for the maximum DEX removal percentage.As given in Table 2, the model predicted the optimal conditions of the test as 2.2 mg/L, 3.8 mmol.L −1 , 43 minutes, and 8.9 for the initial DEX, oxidant, time, and pH, respectively.It has been reported that the sulphate radical activation energy is minimum at neutral pH [39]; hydroxyl radical is the dominant one at pH 12; both sulphate and hydroxyl radicals are present at near-neutral pH; and sulphate radical prevails at pH < 7.
According to the predicted model, pH had no significant effect on other parameters.The experiments were thus repeated at pH = 7.According to the optimum conditions, to  4.
Based on Table 4, the removal efficiency of DEX increased with raising the persulphate concentration, and this caused the same removal rate to occur faster.Almost complete destruction of DEX molecules was reached under optimum conditions.This is in complete agreement with the predicted values by the proposed model.

Investigation of DEX mineralisation rate
The mineralisation efficiency is almost constant at the pH values of 3-7.Still, at neutral conditions, the mineralisation rate was faster.This could be due to the scavenging effect of H + on radicals such as HO • became less significant with the increase in pH [37].According to the time point of 100% DEX removal using all three processes of UV/S 2 , UV/HClO, and UV/H 2 O 2 in 45 minutes, the mineralisation rate of DEX with an initial concentration of 2.2 mg.L −1 at pH 7 and in the presence of 3.8 mmol.L −1 of oxidants persulphate, hydrogen peroxide, and hypochlorite ions was investigated.TOC removal was employed to assess the mineralisation of DEX.In the UV/S 2 O 8 −2 process, the DEX degradation rate of 100% was attained when the reaction time was 45 min; however, only 74.1% of the TOC was removed.This demonstrated that DEX may be decomposed into other relatively stable substances.The results of TOC removal efficiency due to the presence of DEX are listed in Table 5.
As Table 5 shows, the UV/S 2 O 8 −2 process with a 74.1% TOC removal was the most efficient in the degradation and mineralisation of DEX, whereas the TOC removal efficiency for UV/H 2 O 2 and UV/HClO processes was 52% and 23%, respectively.However, the UV/S 2 O 8 −2 process with a 100% DEX removal efficiency manage to remove only 74.1% of the TOC due to the presence of DEX in the aqueous solution.The results of the mineralisation study of UV/S 2 O 8 −2 showed that the decomposition of DEX produced intermediate products.

Effect of inorganic anions on DEX degradation
Inorganic anions are naturally found in water, which may lead to the effect of AOPs by changing the content of reactive oxidising agents in the solution [7].Accordingly, DEX removal in the presence of anions such as nitrite, nitrate, bicarbonate, sulphate, and chloride was investigated by UV/S 2 O 8 −2 process.In this study, 1 mmol.L , and Cl − were added into the DEX-laden solution to investigate their impact on DEX degradation.
In this process, the pH, initial DEX concentration, persulphate concentration, reaction time, and concentration of different ions equalled 7, 2.2 mg.L −1 , 3.8 mmol.L −1 , 45 min, and 1 mmol.L −1 , respectively.DEX removal efficiency in the solution prepared with deionised water was higher compared to those in the presence of NO 3 − , NO 2 − , HCO 3 − , SO 4 −2 , and Cl − .The experimental results are shown, denoting that all of the ions inhibited DEX degradation efficiency.According to the results, the removal efficiency in the presence of 1 mmol.L −1 of nitrite, nitrate, bicarbonate, sulphate, and chloride was 65.3, 91.9, 80.2, 96.5, and 98.4%, respectively.However, the removal efficiency in the control sample was 100%.This is due to the scavenging effect of NO 3 − , NO 2 − , HCO 3 − , SO 4 −2, and Cl − for reactive species present in the solution.Adding nitrate to the system has less effect on the removal efficiency of DEX.Some authors reported that NO 3 − can generate hydroxyl radicals in aqueous solutions under UV irradiation, and this production of hydroxyl , it had no significant inhibitory effect on removal efficiency [42].The scavenging effect of NO 2 − towards SO 4 •− and OH • inevitably reduced the chance of collision with DEX molecules, decreasing the removal efficiency [7].As observed in the results, there was a negligible effect of sulphate ion on the process performance.The effect of SO 4 −2 with SO 4 •− and OH • was much lower than that of other inorganic anions such as chloride and bicarbonate.Therefore, SO 4 −2 is not a robust scavenger for hydroxyl and sulphate radicals.This is due to the fact that there might be reactions of SO 4 −2 with hydroxyl radicals that could generate highly reactive sulphate radicals [42].As demonstrated, bicarbonate (HCO

The effect of radical scavengers on process performance
SO 4 •− and OH • were generally considered as the main oxidants in organic compounds destruction [36].To evaluate the contribution of the oxidising species and their interaction with SO The obvious decrease in DEX degradation has been mainly attributed to the loss of reactive species [7].The lower DEX removal in the presence of ethanol than that of humic acid verifies that the SO 4 • − plays the main role in oxidation of DEX.Besides, the reaction between ethanol and SO 4 • − affected the decline in DEX removal efficiency.Based on the results, the oxidation process was significantly diminished by the addition of 1 g/L ethanol, which reduced the removal efficiency to 70%.It is proved that SO 4 • − and OH • were the main radicals for DEX degradation in this study.

Identification of photochemical oxidation intermediates
A major concern of organic contaminants oxidation in the AOPs is the generated byproducts.In addition, the identification of oxidation by-products during the treatment process is critical to establishing the degradation pathway of different AOPs [44].In order to propose the degradation pathways of sulphate and hydroxyl radicals, oxidation intermediates were identified using LC/MS analysis.The DEX degradation efficiency in the UV/S 2 O 8 −2 process was examined at optimum conditions (pH 7, 2.2 mg.L −1 initial DEX concentration, and 3.8 mmol.L −1 persulphate concentration) at 5, 15, 30, and 45 min; potential intermediates were detected through LC/MS; and the following DEX degradation pathways were proposed accordingly.
According to the LC/MS analysis and calculation of the quantum chemical, the degradation intermediates were identified in the present study.It was concluded that DEX removal in the UV/S 2 O 8 −2 process was mainly via sulphate radical and also direct destruction of DEX molecule by UV irradiation.To understand the degradation pathway of DEX, the LC-MS (ESI+) analysis of the products was performed.Potential oxidation by-products in the DEX degradation by the UV/persulphate processes for the reaction times of 5, 15, 30, and 45 min and an initial persulphate concentration of 3.8 mmol.L −1 were identified.The principal characteristics of the detected intermediates by LC-MS are summarised in Table S3.In this Table, the molecular weight ratio (m/z) and the supposed structures of the fragments are briefly shown.Figure 4 is illustrating the removal pathway of DEX degradation in the investigated process.This pathway is proposed based on the degradation intermediates of DEX (Table S3) and quantum chemical calculations.Two pathways were involved in the photochemical oxidation system based on the structural elucidation of 10 intermediates.
The proposed pathway shows that the DEX molecule was oxidised mainly via two pathways.The decomposition reaction mainly involved hydroxylation, de-fluorination, and the attack of OH• radical on the five atomic rings.Firstly, the displacement of the fluorine atom by OH • and the substitution of aromatic rings by hydroxyl radicals led to the formation of hydroxylated DEX [7].The hydroxylation reaction elevated the electron cloud density on the benzene ring [45].This reaction was more favourable to the electrophilic species' attack and led to the removal of fluorine from DEX and the formation of multihydroxylated products [46].
Five minutes after starting the degradation process the main fragment ions identified a molecular ions peak at m/z of 391.4,357.1, and 306.3 that were related to the intermediates (D1), (D2), and (D5).Then, because of the instability of the dihydroxyacetone chain, the five atomic rings were attacked by hydroxyl radicals and the side chain disappeared from the molecule, leading to the formation of smaller intermediates (D3), (D4), and (D6) at m/z of 308.3, 307, and 286.3, respectively [7].Finally, after 45 min of degradation of the DEX molecule, the cleavages of the rings and the bonds occurred successively, leading to the formation of intermediates (D7)-(D12).The LC-MS result suggested that the DEX removal was mainly attributed to a series of reactions, e.g.hydroxylation, decarboxylation, and aromatic ring cleavage.In summary, DEX photochemical oxidation was mainly due to: OH • radical oxidation, OH • radical substitution, and direct decomposition of DEX by UV radiation.

Effect of treatment in the UV/S 2 O 8 −2 process on DEX cytotoxicity
The cytotoxicity of the process influent and effluent of DEX solution in the UV/S 2 O 8 −2 process was investigated by HEK cell viability.It was found that the viability of HEK cells was significantly inhibited in the presence of DEC containing solution (process influent).Investigation of various DEX concentration exposures to the cells also confirmed that there is significant cell viability in higher DEX concentrations.In addition, exposure times of 24 and 48 h hours also confirmed that reduction in cell viability within 48 h is higher than that of 24 h.Exposure of the cells to influent DEX solution at three different concentrations of 2, 11, and 20 μg/mL for 48 h, the viable cells decreased to 67 ± 3.16%, 46 ± 3.28%, and 40 ± 2.86%, respectively, while it was 100% for the control cells.Results additionally indicated that the cytotoxicity of the PS/UV-treated 20 μg/mL DEX solution decreased the viability of cells to 96.63 ± 2.87% at 48 h.While, the viable cells exposed to the process effluent with various initial DEX concentrations of 2 and 11 μg/mL did not significantly change (P > 0.05).Accordingly, it was found that the viability of cells exposed to the process effluent was considerably higher than that of influent DEX solution.Therefore, it can be deduced that oxidation of pharmaceutical compounds in the UV/S 2 O 8 −2 process leads to the considerable reduction in cytotoxicity solutions [47].

Conclusion
The main objective of this study was to assess the performance of UV/persulphate, UV/hydrogen peroxide, and UV/free chlorine processes in DEX degradation in aqueous solutions.The DEX removal efficiency was studied using an experimental design methodology.The BBD based on RSM was successfully adopted for modelling and prediction of the effects of operational parameters.

Figure 2 .
Figure 2. Design expert plot: (a) normal probability plot for DEX removal, and (b) actual values in front of response predicted by BBD.

Figure 3 .
Figure 3. 3D response surface and 2D contour plots for DEX oxidation (%).(a.b): the effect of initial DEX as the contaminant and persulphate as the oxidant on DEX oxidation and (c.d): the effect of time and pH on DEX oxidation.

− 2 process
significantly promoted the DEX removal from aqueous solutions.Furthermore, due to the maximum efficiency (100%) of DEX in 15 minutes and the high oxidation ability of sulphate radicals (SO 4• − ), the UV/S 2 O 8 −2 process was chosen as the optimal process for DEX removal from aqueous solutions.The results revealed that decreasing the initial pH or initial DEX concentration and increasing the reaction time and S 2 O 8 concentration enhance the degradation rate.The anion effect study showed that anions such as Cl − and SO 4 −2 have the least effect on DEX removal efficiency, while NO 3-, HCO 3-, and NO 2 − decrease its removal.The presence of ethanol and other scavengers suppresses the degradation of DEX.Results of the mineralisation study of the UV/S 2 O 8 −2 process indicated that the decomposition of DEX leads to the production of intermediate products.DEX also induces cell death and reduces cell viability in a concentration-dependent manner.Significant DEX detoxification was achieved after treatment in the PS/UV process.In conclusion, the UV/S 2 O 8 −2 process is a reliable technique for the complete degradation, mineralisation, and detoxification of DEX-contaminated solutions.
of the German Merck Company.DEX powder with 99% purity was procured from Alfa Aesar Company.Acetonitrile and KH 2 PO 4 were of analytical grade and purchased from Merck.All the other used chemical were of reagent grades (Merck,

Table 1 .
Experimental levels of values of independent variables.

3.1. Comparison of the efficiency of UV, UV/S 2 O 8 −2 , UV/H 2 O 2, and UV/HClO processes on the DEX removal efficiency
The performance of UV/S 2 O 8 −2 , UV/H 2 O 2 , and UV/HCLO processes for DEX removal was thoroughly investigated by performing numerous experiments with DEX solutions at an initial reaction pH, oxidants concentration, and reaction time based on the DOE (Table S1), where SO 4 ˙−, HO˙ and Cl˙ are expected to serve as the dominant radicals for DEX destruction in the UV/S 2 O 8 −2 , UV/H 2 O 2 , and UV/HClO oxidation processes, respec- It helps predict significant and nonsignificant factors based on P and F values.Large F values confirm the adequacy of model fit.P values <0.01 indicate that the model terms are significant.The results of ANOVA for the reduced quadratic model is given in table

Table 2 .
Proposed solutions of optimal conditions with the BBD statistical model.was selected as the optimal process for DEX removal from aqueous solutions.Therefore, the present study continued with the investigation of UV/S 2 O 8 −2 as the optimal process for DEX removal from aqueous solutions.Finally, experiments were performed under optimised conditions to validate the model's prediction for the UV/S 2 O 8 −2 process.The experiments were repeated at pH 7 and in the presence of different concentrations of persulphate at different times with a constant value of 3.8 mg.L −1 for the initial concentration of DEX.The results of these experiments are presented in Table

Table 3 .
DEX removal efficiency results at optimum conditions and different times.

Table 4 .
Results of DEX removal under optimal conditions using the UV/S 2 O 8 −2 process.

Table 5 .
TOC removal efficiency due to the presence of 2.2 mg.L −1 initial DEX in 45 minutes, at pH 7, and in the presence of 3.8 mmol.L −1 of persulphate, hydrogen peroxide, and hypochlorite.
The performances of UV/S 2 O 8 −2 , UV/H 2 O 2 , and UV/HOCl were evaluated and compared using both comprehensive process modelling and experimental investigation.To prove the reliability of the proposed quadratic model high correlation coefficients (R 2 = 0.9498) was calculated.It indicates the success of RSM for optimisation of the operating parameters to predict process behaviour for DEX degradation.The optimum operational parameters for DEX removal were determined in this study (2.2 mg/L, 3.8 mmol/L, 43 min, and 8.9 for initial DEX concentration, persulphate, time, and pH, respectively).Based on the findings, the degradation efficiency of DEX generally followed the UV/S 2 O 8 2− > UV/H 2 O 2 > UV/HOCl order under equal conditions, and the use of the UV/S 2 O 8