Nanocomposites Photocatalysis Application for the Puriﬁcation of Phenols and Real Olive Mill Wastewater through a Sequential Process

: In this study, a synthetic phenol solution of water and raw olive mill wastewater (OMW) were considered to achieve puriﬁcation of the aqueous streams from pollutants. Only OMW was initially submitted to a coagulation / ﬂocculation process, to reduce the turbidity, phenols, and chemical oxygen demand (COD). This ﬁrst treatment appeared to be mandatory in order to remove solids from wastewater, allowing the successive use of laboratory-made core-shell nanocomposites. In detail, the optimal coagulant concentration, i.e., chitosan, was 500 mg / L, allowing a reduction of the turbidity and the COD value by 90% and 33%, respectively. After this, phenol wastewater was tested for photocatalysis and then OMW was treated by employing the laboratory-made nanocomposites in a photoreactor equipped with visible light sources and using optimal catalyst concentrations, which allowed for an additional 45% reduction of the COD of the OMW. In addition to this, the e ﬀ ect of the operating temperature was investigated on the photocatalytic process, and suitable kinetic models proposed.


Introduction
Nanotechnology is considered to be a "key technology" in the future and its contribution to water and wastewater treatment innovations has already been significantly high. The word nanotechnology includes a wide range of technologies performed on a nanometer scale for many applications [1]. Nanomaterials have been widely used for various environmental applications, such as the treatment of complex wastewaters, showing significant removal efficiency for both organic and inorganic contaminants [2,3].
Mediterranean countries are major producers of olive oil, and as a consequence, olive mill wastewater (OMW). The composition of OMW may vary much and depends on the type of processed olives, their ripeness, and the extraction process adopted (traditional press method or centrifugation method) [4]. The treatment of OMW is a challenging task due to the high pollutant load (in terms of chemical oxygen demand (COD)), high concentration of recalcitrant compounds (namely lignins and tannins), long-chain fatty acids, phenolic compounds, and biotoxicity [5]. Therefore, for its treatment and disposal, OMW requires a feasible solution from a technical and economical point of view [6]. Recently, a vast number of studies have been reported on the development of chemical and biological technologies for the degradation of organic and inorganic pollutants in aqueous matrices [7]. Efficient and novel methods such as advanced oxidation processes (AOPs) have been considered as one photocatalysts are promising materials for overcoming the two fundamental drawbacks of titanium dioxide: NPs that hinder TiO 2 application and development in the fields of environmental pollutants and energy sources. Recently, many techniques have been proposed to induce visible light activity and magnetic separability to TiO 2 NPs [32]. Very recently, large-scale synthesis of discrete and uniformly sized Fe-based nanoparticles and superparamagnetic Fe 3 O 4 /SiO 2 , based on the use of spinning disk reactor (SDR), has been reported [15]. Fe-based metallic glass technique was reported for remediation of industrially contaminated water with arsenic and nitrate by mass production [33].
Among all the mixing devices used for precipitation, SDR has many advantages when compared to other mixing devices, such as the rotational speed grater. Good results on controlling size and shape of the nanoparticles and NPs production of narrow particle distribution size (PDSs) at a specific target size and also through continuous operation, compatible to industrial practice, can be performed [34]. The proposed work is also economically effective; the catalyst is reusable and reduces operating cost. This approach has been considered in order to achieve economic feasibility of the process since immobilized systems may not be used due to the high capacity of the wastewater. However, in this process suspended photocatalytic particles allow for the operation of the system without any interruption, and the catalyst could be reused for several cycles [32,35]. Nanoparticles can be produced for industrial purposes; the scale-up has been discussed in the previous work of one of the authors [15].
Therefore, the present work reports the pretreatment of real olive mill wastewater through the coagulation-flocculation process and subsequent photocatalytic degradation by employing laboratory synthesized Ti-based nanocatalysts. The photocatalytic method showed efficient removal of targeted pollutants (COD and phenols). A comparison with synthetic phenol solution degradation is reported, as well as the kinetic data interpretation through suitable mathematical modeling.

Materials
OMW samples were collected in the South of Lazio (Italy) from the local olive oil industry. In particular, the wastewater exits an oil extraction process based on the two-phase centrifugal method. The synthetic phenol solution procured was prepared at an initial concentration of 25 mg/L from phenol liquified 85% RE-Pure which was procured from Carlo Erba reagents. Five hundred mg of chitosan (from Sigma Aldrich-448877, with medium molecular weight and a degree of deacetylation at 75-85%) dissolved in 2.5 mL of 2 M HCl (from Sigma Aldrich 37%, ACS reagent grade) solution and 47.5 mL of deionized water. After 60 min, 50 mL of deionized water was added under mechanical stirring to produce 5 g/L chitosan stock solution [24]. The nanophotocatalyst was prepared according to a previous study where its characterization is also reported [32]. In brief, the nanoparticles are made of a magnetic core of Fe 3 O 4 , the first shell of SiO 2 , and the second shell of TiO 2 N-doped.
All the reagents were purchased from Sigma Aldrich (Milan, Italy) and Carlo Erba (Milan, Italy) and all the solutions were prepared with deionized water.

Experimental
The OMW was initially sieved at 0.3 mm to remove the coarse particles. Subsequently, six 600 mL aliquots were added in 1000 mL volume tanks for the coagulation/flocculation process, performed through a Jar test apparatus. The coagulant (chitosan) concentration was varied in the range 100-800 mg/L, the coagulation time was 120 min at fixed pH (4.6, that of raw OMW). The turbidity of the OMW was measured at of 5 min time steps. The sedimentation step (24 h) followed the coagulation process and the generated sludge was removed before starting with the following photodegradation step.
The photocatalysis of OMW was performed in a photoreactor with a 400 mL volume (OMW volume used was 300 mL). The equipment setup and procedure is reported elsewhere in detail [32,36]. According to equal time steps of 15 min, an aliquot of OMW solution was withdrawn and the COD was determined photometrically according to the Standard Methods for the Examination of Water and Wastewater, by standard procedure using a WTW CR4200 thermoreactor. Phenol concentrations were measured with a Merck Spectroquant ® Pharo 100 spectrophotometer. The kinetic experiments were conducted at four different temperature (25 • C, 35 • C, 40 • C, and 45 • C) for pretreated OMW and synthetic phenol solution of 25 mg/L initial concentration. The optimal catalyst concentration was assessed based solely on the sorption results in the dark phase, varying the catalyst concentration in the range 1-3 g/L, which is the typical investigated range [37].
The turbidity was measured using an HF Scientific™ Micro 100 turbidimeter and the pH with a Crison pH-meter. The COD was measured according to the closed reflux colorimetric method [38] and the phenol concentration was measured by a UV-Vis spectrophotometer at 270 nm by observing the change in the absorbance at the maximum absorption wavelength (T80+, PG Instruments, Ltd., Wibtoft, UK). The concentration was then calculated from a calibration curve. The OMW was characterized by an initial COD value of 12.24 g/L and initial turbidity of 3.89 g/L. The pH of zero-charge of the catalyst was measured as equal to 6.55, through the method reported elsewhere [39], which was very close to the values reported in other studies [40,41]. All the experiments were conducted in duplicate and the average values are reported (σ < 4%).

Mathematical Modeling
The kinetic data of turbidity variation over time of the OMW coagulation-flocculation process were interpreted according to Smoluchowski's theory (perikinetic model [42]): is the liquid volume temperature, Tur is the solid concentration (mg L −1 ), α P is the perikinetic probability factor, and ν (Pas) is the liquid dynamic viscosity. The following kinetic constant can be introduced: Equation (1) can be analytically solved using the initial condition Tur = Tur 0 for t = 0, where C0 (mg L −1 ) is the initial turbidity of the wastewater: From the value of k (L mg −1 min −1 ), Ani and coworkers [43] estimated the friction factor β to be equal to β = 2r, where r (kg min −1 ) is the coagulation process rate. In addition, the Brownian diffusion coefficient D (m 2 min −1 ), due to the existing concentration gradient driving force, can be estimated as: Since r decreases during the process, D undergoes an increase, up to reach the asymptotic r value. Regarding the photocatalytic kinetics interpretation, a pseudo-nth order kinetic model was used in a nonlinear regression of experimental data obtained at different operative temperature, based on COD removal for OMW treatment and phenol removal for synthetic phenol solution treatment: Appl. Sci. 2020, 10, 7329 5 of 12 where C (mg/L) represents the COD or phenol concentration, the c-index of reaction order n, and kinetic constant k C (L 1+n /min mg n−1 ) indicates the COD case, which will be substituted by the p-index for the phenol case. The regressed kinetic constant values obtained at different initial temperatures were fitted to the Arrhenius model, expressed by the following equation: where R is the gas constant, E (J mol −1 ) is the activation energy, k 0 (L 1+n /min mg n−1 ) is the pre-exponential constant, and T (K) is the temperature. The nonlinear regression of experimental data was accomplished in the Excel environment, using the nonlinear solver of Excel (Microsoft).  As reported in Figure 1a, the optimal chitosan concentration appears to be equal to 500 mg/L, very close to the optimal value found by other authors that treated similar OMW streams [44,45]. If one considers instead the optimal use of chitosan, it can be observed from Figure 1b that for COD removal, a better chitosan concentration would have been 100 mg/l, since a qCOD of 19.58 g/g was obtained, whereas for Tur removal, the optimal chitosan amount resulted to be again 500 mg/l (qTur = 112.93 g/g). The coagulation/flocculation step is fundamental for the subsequent AOPs technology As reported in Figure 1a, the optimal chitosan concentration appears to be equal to 500 mg/L, very close to the optimal value found by other authors that treated similar OMW streams [44,45]. If one considers instead the optimal use of chitosan, it can be observed from Figure 1b that for COD removal, a better chitosan concentration would have been 100 mg/L, since a q COD of 19.58 g/g was obtained, whereas for Tur removal, the optimal chitosan amount resulted to be again 500 mg/L (q Tur = 112.93 g/g). The coagulation/flocculation step is fundamental for the subsequent AOPs technology applications in order to reduce the turbidity of the effluent. This aspect is important for the photocatalysis process, allowing the light to penetrate into the liquid volume with higher intensities. Another important result was the observed pH increase, that reached 5.6. Therefore, the first pretreatment allowed for the reduction of 90% of the initial turbidity and 33% of the initial COD, providing also a partial pH neutralization of the OMW. The COD reduction was mainly given by the sorption of organic acids and other surface-charged contaminants onto precipitated flocks, removed in the subsequent sedimentation process (sedimentation time equal to 24 h). As already observed in previous studies, a chitosan concentration above the optimum value increases the turbidity of the stream due to the restabilization mechanism typical to these processes. The results achieved can be compared with those reported in the literature for classical coagulants, such as the one reported by Nieto and coworkers [37] who observed a turbidity removal of about 90% on a real OMW using a commercial Nalco-77171.

Coagulation/Flocculation Results and Modeling
Once the optimal coagulant concentration was fixed, the subsequent kinetic experiments were performed. Figure 2 reports the kinetic data results and modeling.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 6 of 12 sorption of organic acids and other surface-charged contaminants onto precipitated flocks, removed in the subsequent sedimentation process (sedimentation time equal to 24 h). As already observed in previous studies, a chitosan concentration above the optimum value increases the turbidity of the stream due to the restabilization mechanism typical to these processes. The results achieved can be compared with those reported in the literature for classical coagulants, such as the one reported by Nieto and coworkers [37] who observed a turbidity removal of about 90% on a real OMW using a commercial Nalco-77171. Once the optimal coagulant concentration was fixed, the subsequent kinetic experiments were performed. Figure 2 reports the kinetic data results and modeling. The model was able to fit the overall experimental data well, including the asymptotic trend reached after 35-40 min. The second order of reaction was found to be suitable for this kind of mechanism, as demonstrated also by the high correlation coefficient, R2, obtained (0.98). The regressed kinetic constant was equal to 6.68×10 -4 L/mg min, which result is close to the regressed value obtained by Kumar and coauthors, who modeled the coagulation kinetic of tannery wastewater by a pseudo-nth order kinetic model, obtaining a reaction order of 1.96 and a kinetic constant of 8.62×10 -4 L/mg min [46]. From the kinetic constant value and rate values, the Brownian diffusion coefficient trend was estimated (Figure 3). The model was able to fit the overall experimental data well, including the asymptotic trend reached after 35-40 min. The second order of reaction was found to be suitable for this kind of mechanism, as demonstrated also by the high correlation coefficient, R2, obtained (0.98). The regressed kinetic constant was equal to 6.68 × 10 −4 L/mg min, which result is close to the regressed value obtained by Kumar and coauthors, who modeled the coagulation kinetic of tannery wastewater by a pseudo-nth order kinetic model, obtaining a reaction order of 1.96 and a kinetic constant of 8.62 × 10 −4 L/mg min [46]. From the kinetic constant value and rate values, the Brownian diffusion coefficient trend was estimated ( Figure 3). mechanism, as demonstrated also by the high correlation coefficient, R2, obtained (0.98). The regressed kinetic constant was equal to 6.68×10 -4 L/mg min, which result is close to the regressed value obtained by Kumar and coauthors, who modeled the coagulation kinetic of tannery wastewater by a pseudo-nth order kinetic model, obtaining a reaction order of 1.96 and a kinetic constant of 8.62×10 -4 L/mg min [46]. From the kinetic constant value and rate values, the Brownian diffusion coefficient trend was estimated (Figure 3).   The initial D value was 2.85×10 -18 m 2 /min and reached the maximum value of about 9.2×10 -17 m 2 /min at the end of the process. The reported values were characterized by the same order of magnitude as the D coefficients values calculated by Okolo and coworkers [47].  It is possible to observe that the optimal photocatalyst concentration was equal to 2 g L −1 (a similar value was obtained in previous work using the same catalyst but without N-doping [48]) and 1.5 g L −1 for OMW and synthetic phenol solution, respectively. The same experiments were conducted at T = 45 °C, the maximum value chosen for the experimental campaign, and it was observed that the optimal concentration values were the same as those used at lower temperature values, whereas the q values underwent a slight reduction (about 7% for OMW and 10% for phenols). Another important result was that the pH of the pretreated OMW, 5.6, was lower in comparison to the pH of zero-charge of the photocatalyst, which resulted as positively charged and able to adsorb deprotonated organic acids and other negatively charged organic species present in the OMW. Figure 5 reports the kinetic experimental and modeled data of OMW and synthetic phenol It is possible to observe that the optimal photocatalyst concentration was equal to 2 g L −1 (a similar value was obtained in previous work using the same catalyst but without N-doping [48]) and 1.5 g L −1 for OMW and synthetic phenol solution, respectively. The same experiments were conducted at T = 45 • C, the maximum value chosen for the experimental campaign, and it was observed that the optimal concentration values were the same as those used at lower temperature values, whereas the q values underwent a slight reduction (about 7% for OMW and 10% for phenols). Another important result was that the pH of the pretreated OMW, 5.6, was lower in comparison to the pH of zero-charge of the photocatalyst, which resulted as positively charged and able to adsorb deprotonated organic acids and other negatively charged organic species present in the OMW. Figure 5 reports the kinetic experimental and modeled data of OMW and synthetic phenol solution photodegradation.

Photocatalysis Results and Modeling
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 12 In both cases, a T increase led to faster kinetics of the photodegradation process. In detail, the OMW COD removal efficiency at 180 min was equal to 45.8, 46.1, 47, and 47.3% for T = 25, 35, 40, and 45 °C, respectively, showing that the temperature increase did not modify substantially the asymptotic value, whereas it influenced the apparent kinetic constant kC, which increased from 4.94×10 -4 L 1+n /min mg n−1 at 25 °C to 6.91×10 -4 L 1+n /min mg n−1 at 45 °C, with a reaction order always in the range 1.24-1.25. The COD removal efficiency was higher than that reported in a previous study where the same catalyst without N-doping was used in similar operative conditions [48]. In a recent study of OMW, the photo-Fenton treatment at neutral pH was active only for a few minutes due to stability issues, and 80% microcontaminants degradation was reported. However, the photo-Fenton efficiency was tested in diluted OMW, and it would be not possible to reuse the catalyst [21]. In this study, the catalyst could be reused and there were no stability issues observed. In both cases, a T increase led to faster kinetics of the photodegradation process. In detail, the OMW COD removal efficiency at 180 min was equal to 45.8, 46.1, 47, and 47.3% for T = 25, 35, 40, and 45 • C, respectively, showing that the temperature increase did not modify substantially the asymptotic value, whereas it influenced the apparent kinetic constant k C , which increased from 4.94 × 10 −4 L 1+n /min mg n−1 at 25 • C to 6.91 × 10 −4 L 1+n /min mg n−1 at 45 • C, with a reaction order always in the range 1.24-1.25. The COD removal efficiency was higher than that reported in a previous study where the same catalyst without N-doping was used in similar operative conditions [48]. In a recent study of OMW, the photo-Fenton treatment at neutral pH was active only for a few minutes due to stability issues, and 80% microcontaminants degradation was reported. However, the photo-Fenton efficiency was tested in diluted OMW, and it would be not possible to reuse the catalyst [21]. In this study, the catalyst could be reused and there were no stability issues observed.
Regarding the results of synthetic phenol solution photodegradation, the phenols removal efficiency at 180 min was equal to 89.4, 89.9, 90.1, and 90.5% for T = 25, 35, 40, and 45 • C, respectively, whereas the apparent kinetic constant k P values were 5.8 × 10 −3 , 6.3 × 10 −3 , 6.5 × 10 −3 , and 6.8 × 10 −3 at 25, 35, 40, and 45 • C, respectively, with a reaction order always in the range 1.4-1.42. Therefore, the synthetic phenol solution photodegradation was characterized by faster kinetics in comparison with that observed for OMW phototreatment. Even if a proper comparison cannot be made, OMW is very complex wastewater, containing various recalcitrant pollutants in addition to phenol compounds. Guo and coworkers [49] reported an optimal pH equal to 6 for phenol's photodegradation, below the pH of zero-charge of the photocatalyst (TiO 2 -SiO 2 ) used here and achieving a total removal above 85% in 140 min.
Finally, the apparent kinetic constant variation with T has been modeled using the Arrhenius model ( Figure 6).
Appl. Sci. 2020, 10, x FOR PEER REVIEW 9 of 12 phenol solution photodegradation was characterized by faster kinetics in comparison with that observed for OMW phototreatment. Even if a proper comparison cannot be made, OMW is very complex wastewater, containing various recalcitrant pollutants in addition to phenol compounds. Guo and coworkers [49] reported an optimal pH equal to 6 for phenol's photodegradation, below the pH of zero-charge of the photocatalyst (TiO2-SiO2) used here and achieving a total removal above 85% in 140 min. Finally, the apparent kinetic constant variation with T has been modeled using the Arrhenius model ( Figure 6). The regressed parameters are reported in Table 1. The pre-exponential factor of phenol's kinetics was higher in comparison to that of OMW, as well as the apparent activation energy was lower. The E value for OMW cannot be properly considered an activation energy, since this wastewater is a complex of several organic and inorganic compounds, whereas E can be defined only for a single compound; therefore, the results obtained cannot be directly compared, but can be useful to better characterize the photodegradation kinetics of this particular two-phase OMW.

Conclusions
In this study, a sequential batch process was proposed and applied to treat real olive mill wastewater streams. The first step was a coagulation/flocculation process, using chitosan at 500 mg L −1 as a coagulant. The results obtained showed a reduction of turbidity and chemical oxygen demand of 90% and 33%, respectively. The kinetic of the process was interpreted according to Smoluchowski's The regressed parameters are reported in Table 1. Table 1. Regressed parameters of photocatalysis kinetic modelling.

OMW
Synthetic Phenol k 0 (L 1+n /min mg n−1 ) 0.024 0.7 E (kJ mol −1 ) 9.57 6.17 The pre-exponential factor of phenol's kinetics was higher in comparison to that of OMW, as well as the apparent activation energy was lower. The E value for OMW cannot be properly considered an activation energy, since this wastewater is a complex of several organic and inorganic compounds, whereas E can be defined only for a single compound; therefore, the results obtained cannot be directly compared, but can be useful to better characterize the photodegradation kinetics of this particular two-phase OMW.

Conclusions
In this study, a sequential batch process was proposed and applied to treat real olive mill wastewater streams. The first step was a coagulation/flocculation process, using chitosan at 500 mg L −1 as a coagulant. The results obtained showed a reduction of turbidity and chemical oxygen demand of 90% and 33%, respectively. The kinetic of the process was interpreted according to Smoluchowski's theory (perikinetic model) and a kinetic constant of 6.68 × 10 −4 L 1+n /min mg n−1 was estimated. Subsequently, laboratory-made nanophotocatalyst particles were employed using pretreated wastewater and a synthetic phenol solution. The photocatalysis kinetic process was analyzed at different temperature values. In detail, employing the optimal catalyst concentration for the two solutions, that is 1.5 and 2 g L −1 , for synthetic phenol and olive mill wastewater, respectively, a phenol removal of 90% and a COD removal efficiency of 45.8% were reached. The Arrhenius model fit the data well for describing the kinetic constant-temperature trends for both reactions.