Removal of two triphenylmethane dyes, Acid Fuchsin (AF) and Malachite green (MG), was studied by hybrid advanced oxidation processes of homogeneous (UV/Fe²⁺/H₂O₂) and heterogeneous (UV/TiO₂−SiO₂) photocatalysis. A comparison of various processes for removal of model pollutants was performed. The results showed that the utilizing hybrid photocatalytic processes in the presence of silica leads to rapid removal of pollutants, which may be ascribable to the synergistic influence of produced various radical species. The effects of operational variables were studied on the efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process. An artificial neural network (ANN) model was intended to predict the removal efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process under different operational conditions. The results indicated that there is a good concurrence between the ANN predicted values and experimental results with a correlation coefficient of 0.9873 and 0.9774 for removal of AF and MG dyes, respectively. The designed neural network model gives a dependable technique for modeling the removal efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process. Moreover, the relative significance of each variable was computed based on the input-hidden and hidden-output connection weights of the neural network model. The initial concentration of dyes was the most significant variable in the removal efficiency.

Abbreviations

AF: Acid Fuchsin
AOP: advanced oxidation process
ANN: artificial neural network
BET: Brunauer–Emmett–Teller
EDX: energy dispersive X-ray
MG: Malachite green
MSE: mean square error
TEM: transmission electron microscopy
XRD: X-ray diffraction

1 Introduction

Over the past years, advanced oxidation processes (AOPs) have been considered a most promising way for degradation of various pollutants, based upon the generation of hydroxyl radicals as the main reactive species, which attack most of the organic pollutants 1-4. AOPs have three major classifications: photolysis, photo-oxidation (like Fenton process, photo-Fenton process, and ozonation process), and heterogeneous photocatalysis 5. Heterogeneous photocatalysis involves light absorption by an oxide semiconductor such as TiO₂. When TiO₂ is exposed to light irradiation, that is more energetic than the band gap energy, the electrons are promoted from the valence to the conduction band, leaving positive holes 6-8. The resulting electron–hole pairs can either recombine and release heat energy or interact separately with O₂ and H₂O, as follows 9:

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0001$ (1)

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0002$ (2)

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0003$ (3)

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0004$ (4)

The hydroxyl radicals are very powerful oxidants and known to be responsible for the degradation of organic pollutants 10.

A combination of both adsorption and heterogeneous photocatalysis into a single process could offer an alternative method in wastewater treatment 11. Silica is widely used as adsorbent with excellent adsorption capacity due to their large specific surface area and porous nature. TiO₂–SiO₂ catalyst has been considered an advanced material to replace pure TiO₂, due to its different surface chemical properties, high catalytic activity, and low cost 12, 13. High adsorption capacity of SiO₂ improves the photocatalytic efficiency of TiO₂ by increasing the concentration of the target pollutant near the TiO₂ sites relative to the solution concentration of a pollutant. In the photocatalytic processes produced oxidizing species such as hydroxyl radicals does not migrate very far from the active centers of the TiO₂, thus the pollutant removal process mainly happens on the surface of TiO₂. Silica provides a synergistic effect by creating a common interface between the SiO₂ and TiO₂ particles 13-15.

On the other hand, the Fenton process as another AOP uses ferrous ions as catalysts that react with hydrogen peroxide to form hydroxyl radicals. Produced hydroxyl radicals may react with ferrous ions to form ferric ions or react with hydrogen peroxide to form other radical species, and it may also combine with each other to produce hydrogen peroxide, as follows 16, 17:

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0005$ (5)

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0006$ (6)

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0007$ (7)

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0008$ (8)

The removal efficiency of the Fenton process can be accelerated by irradiation with UV light (photo-Fenton process) 18. The effect of UV light is the creation of extra ^•OH radical species, as well as to regenerate Fe²⁺ catalyst from the photoreduction of Fe³⁺. The reduction of Fe³⁺ back to Fe²⁺ is critical for the progress of the Fenton process 19.

A literature review revealed that the combination of photo-Fenton and photocatalytic systems in the presence of silica for removal of organic pollutants was not investigated in detail. The aim of this study was utilizing hybrid homogeneous and heterogeneous photocatalytic processes in the presence of silica, to obtaining rapid removal of organic pollutants. In order to achieve the goal of creating an adsorbent-photocatalyst system, TiO₂–SiO₂ catalyst was prepared via the impregnation method and characterized by various techniques. Because of the chemical structure of the organic dyes which has a considerable effect on its removal rate 20, in the present work removal of two triphenylmethane dyes with different molecular structures, Acid Fuchsin (AF) as an anionic dye and Malachite green (MG) as a cationic dye, were studied by using individual and hybrid removal processes. A neural network model was developed to estimate the effect of operational variables on the removal efficiency of UV/Fe²⁺/H₂O₂/TiO₂–SiO₂ hybrid process. Recently, ANNs approach has been employed in various engineering and scientific applications. Because of the complexity of the reactions involved in the heterogeneous photocatalytic processes, application of ANN as a performance prediction tool is a useful technique 21.

2 Materials and methods

2.1 Materials

All chemicals were utilized as received without further purification. TiO₂ nanoparticles (Merck) having Brunauer–Emmett–Teller (BET) surface area 10 m² g⁻¹ and pure anatase phase. The silica used for this study was highly ordered mesoporous LUS-1 with BET surface area 900 m² g⁻¹. The preparation of silica LUS-1 has been detailed in a previous report 22. Hydrogen peroxide solution (30%), 6 M sodium hydroxide (NaOH), 0.1 M sulfuric acid (H₂SO₄), ferrous sulfate heptahydrate (FeSO₄ · 7 H₂O), and triphenylmethane dyes (AF and MG) were obtained from Merck. The molecular structures of AF and MG dyes are given in Supporting Information Fig. S1.

2.2 Preparation of TiO₂ impregnated SiO₂ photocatalyst

The preparation of TiO₂ nanoparticles supported on SiO₂ was performed via impregnation method 23. First, a powder mixture of 1 g SiO₂ and TiO₂, 50:50 w/w, was ground completely in an agate mortar. Then, the mixed powder was added to 100 mL boiling deionized water and dispersed for 15 min utilizing a probe sonicator (Bandelin HD 3200, 200 W). The suspension was stirred for 24 h and thereafter dried in an air oven at 80°C for 12 h. Then the dried solids were calcined at 450°C for 1 h.

2.3 Characterization of TiO₂ impregnated SiO₂ photocatalyst

Titania particles supported on silica were characterized by a Philips X'pert MPD diffractometer using CuK_α radiation (λ = 0.15478 nm). The mean crystallite size of the TiO₂ particles was estimated from the line broadening of corresponding X-ray diffraction (XRD) reflections and as indicated by the Scherrer's equation 24:

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0009$ (9)

where D is the mean crystallite size (nm), k is a constant taken as 0.89, λ is the wavelength of the X-ray radiation, θ is the half diffraction angle, and β is the full width at half maximum intensity. The energy dispersive X-ray spectroscopy (EDX) method was used allowing to studying the chemical composition of the sample (MV2300). The size of the TiO₂ nanoparticles supported on silica was estimated by transmission electron microscopy (TEM, EM208 Philips, 100 kV).

2.4 Hybrid photo-Fenton and photocatalytic experiment

Hybrid photo-Fenton and photocatalytic removal experiments were performed in a batch quartz reactor. A 15 W (UV-C) mercury lamp (Philips, Holland, 254 nm), situated on top of the reactor was used as the light source. In each test, desired amount (0.1–0.6 g L⁻¹) of the TiO₂-impregnated SiO₂ catalyst was dispersed in 100 mL deionized water for 15 min using ultrasound irradiation to enhance the dispersion of particles in water 25. Then the prepared catalyst suspension and desired concentration of dye (5–30 mg L⁻¹), and Fenton reagents (H₂O₂ (0.2–0.7 g L⁻¹) and Fe²⁺ (0.05–0.3 mmol L⁻¹)) were transferred into the reactor. The process was carried out at pH 3.0 ± 0.1, adjusted by the addition of NaOH or H₂SO₄. Iron in its Fe²⁺ state acts as a photocatalyst in the Fenton process and requires a working pH < 4 26. The photocatalytic removal process was started by turning on the UV lamp and stirring with a magnetic bar to ensure proper transport of the reactants. After an irradiation time of 25 min, 5 mL samples were taken out from the reactor, centrifuged (Sigma 2-16P), and then the concentration of the dyes was estimated by UV-Vis spectrophotometry (Rayleigh UV-1600) using the maximum wavelength (λ_max) of the dyes (545 nm for AF and 617 nm for MG).

2.5 ANN modeling

All neural network analysis was done using Matlab software, 2011a version, according to the method described in a previous work 9. A three-layer neural network with a sigmoidal transfer function with back-propagation algorithm was designed in this study.

3 Results and discussion

3.1 Characterization results

Figure 1 depicts the XRD pattern of TiO₂ nanoparticles supported on silica. All XRD reflections can be indexed to the pure anatase phase of TiO₂ (JCPDS no. 21-1272). No rutile phase XRD reflections were detected for the TiO₂ particles. The mean crystallite size of TiO₂ particles supported on silica was calculated by using Eq. 9, from the (101) reflection of anatase at 2θ = 25.28° and it was found to be about 47 nm.

Details are in the caption following the image — **Figure 1**
Open in figure viewer PowerPoint

XRD pattern of TiO₂ nanoparticles supported on silica.

The TEM image of the nanosized TiO₂-impregnated SiO₂ photocatalyst is demonstrated in Fig. 2a. From the TEM image, the mean particle size of the TiO₂-impregnated SiO₂ photocatalyst is estimated to be about 50 nm. This result is in agreement with the mean particle size estimated from the XRD pattern. The composition of the TiO₂-impregnated SiO₂ photocatalyst was studied with EDX analysis. The EDX results (Fig. 2b) obviously affirm the presence of Si, Ti, and O elements. Also, the results revealed that the weight ratio of Ti to Si is about 1:1 for the TiO₂-impregnated SiO₂ sample.

3.2 Comparison of various removal processes

The removal of AF and MG dyes was carried out by using hybrid and individual removal processes (individual photo-Fenton process (UV/Fe²⁺/H₂O₂), individual photocatalytic process (UV/TiO₂ and UV/TiO₂−SiO₂), and hybrid homogeneous and heterogeneous photocatalytic processes (UV/Fe²⁺/H₂O₂/TiO₂−SiO₂, UV/Fe²⁺/H₂O₂/TiO₂, UV/Fe²⁺/H₂O₂/SiO₂). The same operational conditions were employed in various removal processes. Figure 3 shows the efficiency of various processes in removal of AF and MG, as the anionic and cationic dyes, respectively. Both dyes are quickly degraded during the first 5 min of the removal processes, particularly in the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ and UV/Fe²⁺/H₂O₂/TiO₂ processes. This behavior can be related to the higher initial concentration of dye molecules in the initial times of the removal processes. The higher initial concentration provides an increased driving force enhancing the rate of mass transfer of the dye molecules between the aqueous and solid phases 27. It can be observed that the efficiency of various removal processes was found to be in the order of UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ > UV/Fe²⁺/H₂O₂/TiO₂ > UV/TiO₂−SiO₂ > UV/TiO₂ > UV/Fe²⁺/H₂O₂/SiO₂ > UV/Fe²⁺/H₂O₂, for both AF and MG dyes. The results showed that the hybrid homogeneous and heterogeneous photocatalytic processes using titania impregnated silica (UV/Fe²⁺/H₂O₂/TiO₂−SiO₂) was the most efficient process for the removal of cationic and anionic dyes than the individuals or other hybrid removal processes. The comparison of removal efficiency between UV/TiO₂−SiO₂ and UV/TiO₂ processes shows that the TiO₂ nanoparticles impregnated onto SiO₂ catalyst act more efficient in the removal process than the pure TiO₂ nanoparticles. This behavior is due to the fact that the SiO₂ with large surface area and porous structure enhances the adsorption rate of dye molecules, which results in more electron transfer to the conduction band of TiO₂ 28-30, resulting in the rapid removal of organic pollutants. Also, the comparison of removal efficiency between UV/Fe²⁺/H₂O₂/SiO₂ and UV/Fe²⁺/H₂O₂ processes implies that the high adsorption ability of SiO₂ assists in the improvement of the photo-Fenton process. The comparison of the removal efficiency of the UV/Fe²⁺/H₂O₂/TiO₂ process with individual UV/Fe²⁺/H₂O₂ and UV/TiO₂ processes shows that the hybrid homogeneous and heterogeneous photocatalysis provides more desirable results in comparison to the individual processes, which may be because of the synergistic effect of produced various highly reactive radicals. The hybrid homogeneous and heterogeneous photocatalysis led to a reaction rate constant (k_{(UV/Fe2+/H2O2 + UV/TiO2−SiO2)} = 0.079), which was greater than the sum of the reaction rate constants of the individual UV/Fe²⁺/H₂O₂ (k_{(UV/Fe2+/H2O2)} = 0.017) and UV/TiO₂−SiO₂ (k_{(UV/TiO2−SiO2)} = 0.036) processes. The synergistic index (f) for the combined process was calculated as follows 31:

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0010$ (10)

Thus, a synergistic effect is observed for the hybrid system.

3.3 Effects of operational variables on removal efficiency

To study the efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid processes in the removal of AF and MG dyes under various conditions, effects of six operational parameters including TiO₂−SiO₂ dosage, initial dye concentration, initial Fe²⁺ concentration, initial H₂O₂ concentration, initial pH of the solution, and UV-light intensity, were investigated.

3.3.1 The effect of initial dye concentration

To study the influence of the initial dye concentration on the removal rate of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process, the dye concentration varied from 5 to 30 mg L⁻¹, whereas the other variables were kept constant. According to the results given in Tab. 1, the removal rate of both AF and MG dyes decreased as the initial dye concentration increased. An increase in the concentration of dyes from 5 to 30 mg L⁻¹ decreased the removal rate of both AF and MG dyes from 71 and 97% to 17 and 50%, respectively. The higher removal efficiency of MG than AF could be because of the positively charged structure of MG molecules that interacted with the photogenerated electrons of TiO₂−SiO₂ catalyst and this prompts the quick oxidation of MG by free holes 10. AF molecules with negatively charged sulfonic groups (–SO₃⁻) repel the photogenerated electrons of the TiO₂−SiO₂ catalyst. This phenomenon enhances the probability of charge-carrier recombination and prompts the end of the photocatalytic oxidation 10. With increasing the initial dye concentration, more dye molecules are adsorbed on the TiO₂−SiO₂ surface, and consequently the chance of forming reactive species such as ^•OH radicals will be diminished 23, 32, 33. On the other hand, with increases in the initial concentration, the dye solution gets impermeable to UV irradiation because the molar extinction coefficient of the dyes at 254 nm (the wavelength of UV lamp) is very high. In that case, the UV photons get intercepted before reaching the photocatalysts surface 23, 34, 35.

Table 1. The effects of the operational parameters on the removal rate of UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process

Initial dye concentration (mg L⁻¹)	TiO₂−SiO₂ dosage (g L⁻¹)	Initial Fe²⁺ concentration (mmol L⁻¹)	Initial H₂O₂ concentration (g L⁻¹)	Initial solution pH	UV light intensity (W m⁻²)	AF	MG
						Removal rate (%)
5	0.4	0.05	0.4	3	56.5	71	97
10	0.4	0.05	0.4	3	56.5	59	89
15	0.4	0.05	0.4	3	56.5	48	80
20	0.4	0.05	0.4	3	56.5	40	67
25	0.4	0.05	0.4	3	56.5	29	56
30	0.4	0.05	0.4	3	56.5	17	50
10	0.1	0.05	0.4	3	56.5	19	42
10	0.2	0.05	0.4	3	56.5	38	61
10	0.3	0.05	0.4	3	56.5	50	78
10	0.5	0.05	0.4	3	56.5	58	85
10	0.6	0.05	0.4	3	56.5	55	82
10	0.4	0.1	0.4	3	56.5	64	92
10	0.4	0.15	0.4	3	56.5	69	93
10	0.4	0.2	0.4	3	56.5	66	95
10	0.4	0.25	0.4	3	56.5	65	91
10	0.4	0.3	0.4	3	56.5	59	85
10	0.4	0.05	0.2	3	56.5	41	63
10	0.4	0.05	0.3	3	56.5	48	73
10	0.4	0.05	0.5	3	56.5	65	93
10	0.4	0.05	0.6	3	56.5	61	91
10	0.4	0.05	0.7	3	56.5	57	85
10	0.4	0.05	0.4	5.1	56.5	55	85
10	0.4	0.05	0.4	6.3	56.5	51	82
10	0.4	0.05	0.4	7.6	56.5	30	76
10	0.4	0.05	0.4	8.4	56.5	21	69
10	0.4	0.05	0.4	9.2	56.5	11	67
10	0.4	0.05	0.4	3	48.5	50	82
10	0.4	0.05	0.4	3	43	43	71
10	0.4	0.05	0.4	3	36	32	60
10	0.4	0.05	0.4	3	29.5	25	54
10	0.4	0.05	0.4	3	24	21	49

3.3.2 The effect of TiO₂−SiO₂ dosage

To study the effect of TiO₂−SiO₂ dosage on the removal efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process, the TiO₂−SiO₂ dosage varied from 0.1 to 0.6 g L⁻¹, whereas the other variables were kept constant. Tab. 1 shows the variation of the removal rate for AF and MG dyes as a function of the TiO₂−SiO₂ dosage. According to the results given in Tab. 1, the increase in the TiO₂−SiO₂ dosage until about 0.4 g L⁻¹ enhanced the removal rate of both AF and MG dyes. By varying the TiO₂−SiO₂ dosage from 0.1 to 0.4 g L⁻¹, the removal rate of AF and MG dyes improved from 19 and 42% to 59 and 89%, respectively. This can be due to an increase in the number of accessible catalytic and adsorption sites on the TiO₂−SiO₂ surface 23, 36. An improvement on the removal efficiency of the hybrid process with an increase in the TiO₂−SiO₂ dosage is not remarkable for concentrations of >0.4 g L⁻¹. High turbidity of the solution caused by high catalyst loading leads to a reduction in UV light penetration into the solution 23, 37, 38. A further increase in the TiO₂−SiO₂ dosage from 0.4 to 0.6 g L⁻¹ decreased the removal rate of AF and MG dyes from 59 and 89% to 55 and 82%, respectively.

3.3.3 The effect of initial Fe²⁺ concentration

The initial concentration of Fe²⁺ is an important parameter in the Fenton and photo-Fenton processes because Fe²⁺ ion acts as catalyst that initiates the decomposition of H₂O₂ to produce the highly reactive hydroxyl radicals (Eq. 5) 16, 17. The influence of the initial concentration of Fe²⁺ on the removal rate of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process was studied with concentrations in the range of 0.05–0.3 mmol L⁻¹, while the other variables were kept constant. Table 1 shows the influence of the initial concentration of Fe²⁺ on the removal efficiency of the hybrid process. The results showed that the removal rate of both AF and MG dyes increased as the initial Fe²⁺ concentration increased until about 0.15–0.2 mmol L⁻¹. An increase in the initial concentration of Fe²⁺ from 0.05 to 0.15 mmol L⁻¹ increased the removal rate of AF dye from 59 to 69% and an increase in the initial concentration of Fe²⁺ from 0.05 to 0.2 mmol L⁻¹ increased the removal rate of MG dye from 89 to 95%, because the catalytic effect of Fe²⁺ increases with an increase in its initial concentration. Addition of Fe²⁺ of >0.15–0.2 mmol L⁻¹ in the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid procedure did not influence the enhancement in the removal efficiency. The removal rate of both AF and MG dyes started to decrease, accompanied by an increase in the initial concentration of Fe²⁺ catalyst of >0.15–0.2 mmol L⁻¹. At higher Fe²⁺ concentrations, according to Eq. 5, a large number of Fe³⁺ ions was created 19. Ferric ion undergoes a reaction with ⁻OH ions to produce Fe(OH)²⁺ complex in solution (Eq. 11) 19, which has a strong UV absorption from 290 to 400 nm, and consequently the strength of UV light would decrease 39, 40.

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0011$ (11)

3.3.4 The effect of initial H₂O₂ concentration

To study the relationship between the initial H₂O₂ concentration and removal efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process, the initial H₂O₂ concentration varied from 0.2 to 0.7 g L⁻¹, while the other variables were kept constant. Table 1 indicates the influence of the initial concentration of H₂O₂ on the removal rate of AF and MG dyes. The results indicated that the removal rate of both AF and MG dyes increased as the initial H₂O₂ concentration increased. An increase in the initial concentration of H₂O₂ from 0.2 to 0.5 g L⁻¹ increased the removal rate of both AF and MG dyes from 41 and 63% to 65 and 93%, respectively. With increasing the H₂O₂ concentration, additionally hydroxyl radicals are produced. An improvement on the removal efficiency of the hybrid process with an increase in the H₂O₂ concentration is not remarkable for concentration of >0.5 g L⁻¹. An increase in the initial concentration of H₂O₂ from 0.5 to 0.7 g L⁻¹ decreased the removal rate of AF and MG dyes from 65 and 93% to 57 and 85%, respectively. This could be because of the recombination and scavenging of ^•OH radicals by hydrogen peroxide, according to Eqs. 7 and 8 19, 41-43.

3.3.5 The effect of initial solution pH

The initial solution pH is an important parameter in Fenton reactions for maintaining the effectiveness of the process because the pH value affects the ^•OH radical production and accordingly the performance of the Fenton process 19. The effect of the initial solution pH on the removal efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process was studied for pH values between 3 and 9.2, while the other variables were kept constant. Table 1 shows the variation of the removal rate for AF and MG dyes as a function of the solution pH. It can be seen from Tab. 1 that the increase in the initial pH decreased the removal rate of both AF and MG dyes. These observations are in agreement with the results of previous studies 44-46, since the maximum hydroxyl radical production is expected from Fenton's reaction around pH 2.8. An increase in the initial pH from 3 to 9.2 decreased the removal rate of AF and MG dyes from 59 and 89% to 11 and 67%, respectively.

On the other hand, the initial pH of the solution can affect the surface charge of the TiO₂−SiO₂ catalyst and also shifts the potentials of catalytic reactions 47. Thus, the initial pH affects the adsorption of dye molecules onto the TiO₂−SiO₂ surface, and consequent change in the photocatalytic reaction rate. Point of zero charge value, pH_ZPC, of the TiO₂−SiO₂ catalyst was found to be 6.2. The TiO₂−SiO₂ catalyst surface is positively charged in acidic media (pH < 6.2), whereas it is negatively charged under alkaline conditions (pH > 6.2), according to Eqs. 12 and 13 48, 49:

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0012$ (12)

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0013$ (13)

Under acidic conditions, the adsorption of anionic molecules mainly occurs on the surface of the catalyst because the TiO₂−SiO₂ surface is positively charged by adsorption of H⁺. The increase in the removal rate at acidic pH may be because of the high adsorption rate of anionic molecules on the TiO₂−SiO₂ surface. Unlike acidic conditions, under alkaline conditions, adsorption of cationic molecules mainly occurs on the surface of the catalyst. The increase in the removal rate at alkaline pH is probably related to the high rate of cationic molecules adsorption on the TiO₂−SiO₂ surface. Under acidic conditions maximum hydroxyl radicals produced from Fenton's reaction and high adsorption rate of AF molecules occurred on the catalyst surface, thus hydroxyl radical production is along with adsorption of AF molecules. While in the case of the MG dye, hydroxyl radical production is not along with the adsorption of MG molecules. As a result, we are faced with a drastic reduction in the removal rate of AF dye with an increase in pH.

3.3.6 The effect of light intensity

To study the influence of UV light intensity on the removal efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process, UV light intensity was varied in the range of 24–56.5 W m⁻², while the other variables were kept constant. Table 1 shows the variation of the removal rate for AF and MG dyes as a function of UV light intensity and solution. According to the results given in Tab. 1, decreasing the UV light intensity increased the removal rate of both AF and MG dyes. A decrease in the light intensity from 56.5 to 24 W m⁻² decreased the removal rate of AF and MG dyes from 59 and 89% to 21 and 49%, respectively. In the photo-Fenton process, UV light is used for photoreduction of Fe³⁺ to Fe²⁺ and photolysis of H₂O₂. The photolysis rate of H₂O₂ directly depends on the UV light intensity 19. Therefore, improvement of the removal rate at high UV light intensity is because of the increase in generation of highly reactive ^•OH radicals from the photo-dissociation of hydrogen peroxide 19, 50. On the other hand, in the photocatalytic process, UV light irradiation provides the photons required for the generation of electron–hole pairs. In the presence of low intensity of UV light, the electron–hole separation rivals with recombination and reduces the generation of ^•OH radicals 10, 51. When more radiation falls on the TiO₂−SiO₂ surface (at high light intensity), more hydroxyl radicals are produced and consequently the removal rate increased 52.

3.4 ANN modeling

The effects of operational variables on the removal efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process were analyzed by utilizing a multilayer feed-forward ANN model. Operational variables (including TiO₂−SiO₂ dosage, initial dye concentration, initial Fe²⁺ concentration, UV-light intensity, initial H₂O₂ concentration, and initial pH of the solution) were chosen as the input of the network and removal efficiency (%) of dyes was chosen as an output of the network.

The experimental results were divided into test (one fourth), validation (one fourth), and training (one half) sets. Since the used transfer function in the hidden layer was sigmoid, all experimental results (Y_i) were scaled to a new value x_i, as follows:

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0014$ (14)

where Y_i,min and Y_i,max are the extreme values of the variable Y_i 53, 54.

The neuron number of a hidden layer influences the efficiency of the neural network model. In order to assess the best configuration of hidden nodes, experiments were performed by altering the number of neurons from 1 to 20. To avoid any accidental correlation due to the random initialization of the weights, each topology was evaluated three times. Figure 4 demonstrates the correlation between the neuron number in the hidden layer and the mean square error (MSE). The lowest MSE is resulted when 14 neurons were used. Hence, 14 neurons were chosen as the best number of neurons.

Figure 5 demonstrates a correlation between experimental removal efficiency (%) results and ANN-predicted values for the test set. The correlation coefficients (R²) of the lines are 0.9873 and 0.9774 for the removal of AF and MG dyes, respectively. The R²-value infers an agreeable representation of the hybrid removal process by the model. According to Fig. 5, there is an acceptable concurrence between the ANN-predicted values and the experimental results. Therefore, the designed ANN model is able to predict adequately the removal rate of both anionic and cationic dyes for the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process.

The designed ANN model provided the weights listed in Supporting Information Tabs. S1 and S2. In order to assess the relative significance of the input variables on the removal efficiency of the hybrid process the neural weight matrix was used 9. The following equation was proposed to estimate the relative significance of each input variable:

$urn:x-wiley:14381656:media:clen201400449:clen201400449-math-0015$ (15)

where I_j is the relative importance of the jth input variable on the output variable, N_i and N_h are the number of inputs and hidden neurons, respectively; W is the connection weight, the superscripts “i,” “h,” and “o,” respectively, refer to input, hidden, and output layers; and subscripts “k,” “m,” and “n,” respectively, refer to input, hidden, and output neurons 21, 54, 55.

According to the results given in Tab. 1, among the operational variables, the initial dye concentration has greater influence on the removal rate of both AF and MG dyes. While the initial solution pH was identified as the parameter with great effect on the removal of AF, it was identified as the fourth efficient parameter on the removal of MG. As can be seen in Tab. 2, the initial concentrations of Fe²⁺ and H₂O₂ were identified as the parameters with weak effect on the removal of both AF and MG. Nevertheless, the difference in the relative significance of the operation variables is not that significant. The high sensitivity of the network response reveals low flexibility in the process control and the necessity of exact control of all operational parameters at desired values.

Table 2. Relative significance (%) of the input variables on the removal rate of AF and MG by UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process

Variable	AF	MG
	Relative importance (%)
Initial dye concentration	24	20
TiO₂−SiO₂ dosage	15	17
Initial Fe²⁺ concentration	4	6
Initial H₂O₂ concentration	8	11
Irradiation time	13	15
Initial solution pH	20	12
UV light intensity	16	19

4 Concluding remarks

The removal of AF and MG from aqueous solutions has been studied by hybrid and individual homogeneous and heterogeneous photocatalytic processes using titania-impregnated silica. The removal efficiency of different processes was found to be in the order of: UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ > UV/Fe²⁺/H₂O₂/TiO₂ > UV/TiO₂−SiO₂ > UV/TiO₂ > UV/Fe²⁺/H₂O₂/SiO₂ > UV/Fe²⁺/H₂O₂. The best performance of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ hybrid process was observed for operational conditions with optimal values: TiO₂−SiO₂ dosage of 0.4 g L⁻¹, dye concentration of 5 mg L⁻¹, initial Fe²⁺ concentration of 0.15−0.2 mmol L⁻¹, H₂O₂ concentration of 0.4 g L⁻¹, irradiation time of 25 min, initial solution pH of 3, and UV-light intensity of 56.5 W m⁻². The removal efficiency of the UV/Fe²⁺/H₂O₂/TiO₂−SiO₂ process under different operational conditions was successfully predicted by utilizing a three-layered back-propagation neural network with 14 neurons in the hidden layer. There was an acceptable concurrence between the ANN-predicted values and the experimental results with an R² of 0.9873 and 0.9774 for removal of AF and MG dyes, respectively.

Acknowledgment

The authors thank University of Tehran for the support of this work.

The authors have declared no conflicts of interest.

Supporting Information

References

1 M. A. Sheikh, A. Kumar, M. Paliwal, R. Ameta, R. C. Khandelwal, Degradation of Organic Effluents Containing Wastewater by Photo-Fenton Oxidation Process, Indian J. Chem. Sect. A 2008, 47, 1681−1684.
Web of Science® Google Scholar
2 H. Eskandarloo, M. Hashempour, A. Vicenzo, S. Franz, A. Badiei, M. A. Behnajady, M. Bestetti, High-Temperature Stable Anatase-Type TiO₂ Nanotube Arrays: A Study of the Structure-Activity Relationship, Appl. Catal. B 2016, 185, 119–132.
10.1016/j.apcatb.2015.11.048
CAS Web of Science® Google Scholar
3 P. R. Gogate, A. B. Pandit, A Review of Imperative Technologies for Wastewater Treatment, I: Oxidation Technologies at Ambient Conditions, Adv. Environ. Res. 2004, 8, 501−551.
10.1016/S1093-0191(03)00032-7
CAS Web of Science® Google Scholar
4 P. R. Gogate, A. B. Pandit, A Review of Imperative Technologies for Wastewater Treatment, II: Hybrid Methods, Adv. Environ. Res. 2004, 8, 553−597.
10.1016/S1093-0191(03)00031-5
CAS Web of Science® Google Scholar
5 D. Bahnemann, Photocatalytic Water Treatment: Solar Energy Applications, Sol. Energ. 2004, 77, 445−459.
10.1016/j.solener.2004.03.031
CAS Web of Science® Google Scholar
6 V. K. Gupta, R. Jain, A. Mittal, T. A. Saleh, A. Nayak, S. Agarwal, S. Sikarwar, Photo-Catalytic Degradation of Toxic Dye Amaranth on TiO₂/UV in Aqueous Suspensions, Mater. Sci. Eng. C 2012, 32, 12−17.
10.1016/j.msec.2011.08.018
CAS PubMed Web of Science® Google Scholar
7 H. Eskandarloo, A. Badiei, M. A. Behnajady, G. M. Ziarani, UV–LEDs Assisted Preparation of Silver Deposited TiO₂ Catalyst Bed Inside Microchannels as a High Efficiency Microphotoreactor for Cleaning Polluted Water, Chem. Eng. J. 2015, 270, 158–167.
10.1016/j.cej.2015.01.117
CAS Web of Science® Google Scholar
8 H. Eskandarloo, A. Badiei, C. Haug, Enhanced Photocatalytic Degradation of an Azo Textile Dye by Using TiO₂/NiO Coupled Nanoparticles: Optimization of Synthesis and Operational Key Factors, Mater. Sci. Semicond. Process. 2014, 27, 240–253.
10.1016/j.mssp.2014.06.029
CAS Web of Science® Google Scholar
9 H. Eskandarloo, A. Badiei, M. A. Behnajady, Study of the Effect of Additives on the Photocatalytic Degradation of a Triphenylmethane Dye in the Presence of Immobilized TiO₂/NiO Nanoparticles: Artificial Neural Network Modeling, Ind. Eng. Chem. Res. 2014, 53, 6881–6895.
10.1021/ie500253q
CAS Web of Science® Google Scholar
10 H. Eskandarloo, A. Badiei, Fabrication of an Inexpensive and High Efficiency Microphotoreactor Using CO₂ Laser Technique for Photocatalytic Water Treatment Applications, Environ. Technol. 2015, 36, 1063–1073.
10.1080/09593330.2014.974681
CAS PubMed Web of Science® Google Scholar
11 S. K. Das, M. K. Bhunia, A. Bhaumik, Highly Ordered Ti-SBA-15: Efficient H₂ Adsorbent and Photocatalyst for Eco-Toxic Dye Degradation, J. Solid State Chem. 2010, 183, 1326–1333.
10.1016/j.jssc.2010.04.015
CAS Web of Science® Google Scholar
12 Ş. Neaţu, V. I. Pârvulescu, G. Epure, N. Petrea, V. Şomoghi, G. Ricchiardi, S. Bordiga, et al., M/TiO₂/SiO₂(M = Fe, Mn, and V) Catalysts in Photo–Decomposition of Sulfur Mustard, Appl. Catal. B 2009, 91, 546–553.
10.1016/j.apcatb.2009.06.026
CAS Web of Science® Google Scholar
13 X. Gao, I. E. Wachs, Titania-Silica as Catalysts: Molecular Structural Characteristics and Physico-Chemical Properties, Catal. Today 1999, 51, 233–254.
10.1016/S0920-5861(99)00048-6
CAS Web of Science® Google Scholar
14 C. Anderson, A. J. Bard, An Improved Photocatalyst of TiO₂/SiO₂ Prepared by a Sol-Gel Synthesis, J. Phys. Chem. 1995, 99, 9882–9885.
10.1021/j100024a033
CAS Web of Science® Google Scholar
15 C. Anderson, A. J. Bard, Improved Photocatalytic Activity and Characterization of Mixed TiO₂/SiO₂ and TiO₂/Al₂O₃ Materials, J. Phys. Chem. B 1997, 101, 2611–2616.
10.1021/jp9626982
CAS Web of Science® Google Scholar
16 E. Chamarro, A. Marco, S. Esplugas, Use of Fenton Reagent to Improve Organic Chemical Biodegradability, Water Res. 2001, 35, 1047–1051.
10.1016/S0043-1354(00)00342-0
CAS PubMed Web of Science® Google Scholar
17 M. A. Behnajady, N. Modirshahla, F. Ghanbary, A Kinetic Model for the Decolorization of CI Acid Yellow 23 by Fenton Process, J. Hazard. Mater. 2007, 148, 98−102.
10.1016/j.jhazmat.2007.02.003
CAS PubMed Web of Science® Google Scholar
18 J. García-Montano, F. Torrades, J. A. García-Hortal, X. Domenech, J. Peral, Combining Photo-Fenton Process with Aerobic Sequencing Batch Reactor for Commercial Hetero-Bireactive Dye Removal, Appl. Catal. B 2006, 67, 86–92.
10.1016/j.apcatb.2006.04.007
CAS Web of Science® Google Scholar
19 N. Modirshahla, M. A. Behnajady, F. Ghanbary, Decolorization and Mineralization of CI Acid Yellow 23 by Fenton and Photo-Fenton Processes, Dyes Pigm. 2007, 73, 305−310.
10.1016/j.dyepig.2006.01.002
CAS Web of Science® Google Scholar
20 A. R. Khataee, M. B. Kasiri, Photocatalytic Degradation of Organic Dyes in the Presence of Nanostructured Titanium Dioxide: Influence of the Chemical Structure of Dyes, J. Mol. Catal. A Chem. 2010, 328, 8–26.
10.1016/j.molcata.2010.05.023
CAS Web of Science® Google Scholar
21 R. Mohammadi, H. Eskandarloo, M. Mohammadi, Application of Artificial Neural Network (ANN) for Modeling of Dyes Decolorization by Sn/Zn-TiO₂ Nanoparticles, Desalin. Water Treat. 2015, 55(7), 1922–1933.
10.1080/19443994.2014.928237
CAS Web of Science® Google Scholar
22 L. Bonneviot, M. Morin, A. Badiei, 2001, Patent WO 01/55031 A1.
Google Scholar
23 H. Eskandarloo, A. Badiei, M. A. Behnajady, Application of Response Surface Methodology for Optimization of Operational Variables in Photodegradation of Phenazopyridine Drug Using TiO₂/CeO₂ Hybrid Nanoparticles, Desalin. Water Treat. 2015, 54, 3300–3310.
10.1080/19443994.2014.911705
CAS Web of Science® Google Scholar
24 A. L. Patterson, The Scherrer Formula for X-ray Particle Size Determination, Phys. Rev. 1939, 56, 978–982.
10.1103/PhysRev.56.978
CAS Web of Science® Google Scholar
25 P. R. Gogate, V. S. Sutkar, A. B. Pandit, Sonochemical Reactors: Important Design and Scale Up Considerations with a Special Emphasis on Heterogeneous Systems, Chem. Eng. J. 2011, 166, 1066−1082.
10.1016/j.cej.2010.11.069
CAS Web of Science® Google Scholar
26 M. Y. Ghaly, G. Härtel, R. Mayer, R. Haseneder, Photochemical Oxidation of p-Chlorophenol by UV/H₂O₂ and Photo−Fenton Process: A Comparative Study, Waste Manage. 2001, 21, 41−47.
10.1016/S0956-053X(00)00070-2
CAS PubMed Web of Science® Google Scholar
27 V. K. Gupta, R. Jain, A. Mittal, T. A. Saleh, A. Nayak, S. Agarwal, S. Sikarwar, Photo-Catalytic Degradation of Toxic Dye Amaranth on TiO₂/UV in Aqueous Suspensions, Mater. Sci. Eng. C 2012, 32, 12−17.
10.1016/j.msec.2011.08.018
CAS PubMed Web of Science® Google Scholar
28 W. Zhao, L. Feng, R. Yang, J. Zheng, X. Li, Synthesis, Characterization, and Photocatalytic Properties of Ag Modified Hollow SiO₂/TiO₂ Hybrid Microspheres, Appl. Catal. B 2011, 103, 181–189.
10.1016/j.apcatb.2011.01.025
CAS Web of Science® Google Scholar
29 L. Pinho, M. J. Mosquera, Photocatalytic Activity of TiO₂-SiO₂ Nanocomposites Applied to Buildings: Influence of Particle Size and Loading, Appl. Catal. B 2013, 134–135, 205–221.
10.1016/j.apcatb.2013.01.021
CAS Web of Science® Google Scholar
30 X.-J. Wang, F.-T. Li, Y.-J. Hao, S.-J. Liu, M.-L. Yang, TiO₂/SBA-15 Composites Prepared Using H₂TiO₃ by Hydrothermal Method and Its Photocatalytic Activity, Mater. Lett. 2013, 99, 38–41.
10.1016/j.matlet.2013.02.060
CAS Web of Science® Google Scholar
31 M. V. Bagal, P. R. Gogate, Degradation of 2,4-Dinitrophenol Using a Combination of Hydrodynamic Cavitation, Chemical and Advanced Oxidation Processes, Ultrason. Sonochem. 2013, 20, 1226–1235.
10.1016/j.ultsonch.2013.02.004
CAS PubMed Web of Science® Google Scholar
32 M. A. Behnajady, N. Modirshahla, R. Hamzavi, Kinetic Study on Photocatalytic Degradation of CI Acid Yellow 23 by ZnO Photocatalyst, J. Hazard. Mater. 2006, 133, 226−232.
10.1016/j.jhazmat.2005.10.022
CAS PubMed Web of Science® Google Scholar
33 L. Yin, J. Gao, J. Wang, X. Luan, P. Kang, Y. Li, K. Li, et al., Synthesis of Er³⁺: Y₃Al₅O₁₂ and its Effects on the Solar Light Photocatalytic Activity of TiO₂–ZrO₂ Composite, Res. Chem. Intermed. 2012, 38, 523−536.
10.1007/s11164-011-0368-x
CAS Web of Science® Google Scholar
34 V. K. Gupta, R. Jain, S. Agarwal, A. Nayak, M. Shrivastava, Photodegradation of Hazardous Dye Quinoline Yellow Catalyzed by TiO₂, J. Colloid Interface Sci. 2012, 366, 135−140.
10.1016/j.jcis.2011.08.059
CAS PubMed Web of Science® Google Scholar
35 M. B. Kasiri, A. R. Khataee, Photooxidative Decolorization of Two Organic Dyes with Different Chemical Structures by UV/H₂O₂ Process: Experimental Design, Desalination 2011, 270, 151−159.
10.1016/j.desal.2010.11.039
CAS Web of Science® Google Scholar
36 R. Jain, S. Sikarwar, Semiconductor-Mediated Photocatalyzed Degradation of Erythrosine Dye from Wastewater Using TiO₂ Catalyst, Environ. Technol. 2010, 31, 1403−1410.
10.1080/09593331003758789
CAS PubMed Web of Science® Google Scholar
37 M. Tanveer, G. Tezcanli Guyer, Solar Assisted Photo Degradation of Wastewater by Compound Parabolic Collectors: Review of Design and Operational Parameters, Renew. Sust. Energ. Rev. 2013, 24, 534−543.
10.1016/j.rser.2013.03.053
CAS Web of Science® Google Scholar
38 P. Esparza, M. E. Borges, L. Díaz, M. C. Alvarez-Galván, J. L. G. Fierro, Photodegradation of Dye Pollutants Using New Nanostructured Titania Supported on Volcanic ashes, Appl. Catal. A 2010, 388, 7−14.
10.1016/j.apcata.2010.07.058
CAS Web of Science® Google Scholar
39 X. K. Zhao, G. P. Yang, Y. J. Wang, X. C. Gao, Photochemical Degradation of Dimethyl Phthalate by Fenton Reagent, J. Photochem. Photobiol. A 2004, 161, 215−220.
10.1016/S1010-6030(03)00344-7
CAS Web of Science® Google Scholar
40 M. P. Titus, V. G. Molina, M. A. Banos, J. Gimenez, S. Esplugas, Degradation of Chlorophenols by Means of Advanced Oxidation Processes: A General Review, Appl. Catal. B 2004, 47, 219−256.
10.1016/j.apcatb.2003.09.010
CAS Web of Science® Google Scholar
41 A. K. Abdessalem, N. Bellakhal, N. Oturan, M. Dachraoui, M. A. Oturan, Treatment of a Mixture of Three Pesticides by Photo– and Electro–Fenton Processes, Desalination 2010, 250, 450−455.
10.1016/j.desal.2009.09.072
CAS Web of Science® Google Scholar
42 M. Y. Ghaly, G. Hartel, R. Mayer, R. Haseneder, Photochemical Oxidation of p–Chlorophenol by UV/H₂O₂ and Photo–Fenton Process: A Comparative Study, Waste Manage. 2001, 21, 41−47.
10.1016/S0956-053X(00)00070-2
CAS PubMed Web of Science® Google Scholar
43 N. Golash, P. R. Gogate, Degradation of Dichlorvos Containing Wastewaters Using Sonochemical Reactors, Ultrason. Sonochem. 2012, 19, 1051−1060.
10.1016/j.ultsonch.2012.02.011
CAS PubMed Web of Science® Google Scholar
44 C. A. Martínez-Huitle, E. Brillas, Decontamination of Wastewaters Containing Synthetic Organic Dyes by Electrochemical Methods: A General Review, Appl. Catal. B 2009, 87, 105−145.
10.1016/j.apcatb.2008.09.017
CAS Web of Science® Google Scholar
45 M. Diagne, N. Oturan, M. A. Oturan, Removal of Methyl Parathion from Water by Electrochemically Generated Fenton's Reagent, Chemosphere 2007, 66, 841−848.
10.1016/j.chemosphere.2006.06.033
CAS PubMed Web of Science® Google Scholar
46 J. H. Sun, S. P. Sun, J. Y. Sun, R. X. Sun, L. P. Qiao, H. Q. Guo, M. H. Fan, Degradation of Azo Dye Acid Black 1 Using Low Concentration Iron of Fenton Process Facilitated by Ultrasonic Irradiation, Ultrason. Sonochem. 2007, 14, 761–766.
10.1016/j.ultsonch.2006.12.010
CAS PubMed Web of Science® Google Scholar
47 M. Pelaez, A. A. de la Cruz, K. O'Shea, P. Falaras, D. D. Dionysiou, Effects of Water Parameters on the Degradation of Microcystin-LR under Visible Light-Activated TiO₂ Photocatalyst, Water Res. 2011, 45, 3787−3796.
10.1016/j.watres.2011.04.036
CAS PubMed Web of Science® Google Scholar
48 N. Daneshvar, M. Rabbani, N. Modirshahla, M. A. Behnajady, Kinetic Modeling of Photocatalytic Degradation of Acid Red 27 in UV/TiO₂ Process, J. Photochem. Photobiol. A 2004, 168, 39−45.
10.1016/j.jphotochem.2004.05.011
CAS Web of Science® Google Scholar
49 T. A. Gad-Allah, M. E. Ali, M. I. Badawy, Photocatalytic Oxidation of Ciprofloxacin Under Simulated Sunlight, J. Hazard. Mater. 2011, 186, 751−755.
10.1016/j.jhazmat.2010.11.066
CAS PubMed Web of Science® Google Scholar
50 Z. Zhang, Y. Feng, Y. Liu, Q. Sun, P. Gao, N. Ren, Kinetic Degradation Model and Estrogenicity Changes of EE₂ (17α–Ethinylestradiol) in Aqueous Solution by UV and UV/H₂O₂ Technology, J. Hazard. Mater. 2010, 181, 1127−1133.
10.1016/j.jhazmat.2010.05.132
CAS PubMed Web of Science® Google Scholar
51 D. Dong, P. Li, X. Li, Q. Zhao, Y. Zhang, C. Jia, P. Li, Investigation on the Photocatalytic Degradation of Pyrene on Soil Surfaces Using Nanometer Anatase TiO₂ Under UV Irradiation, J. Hazard. Mater. 2010, 174, 859−863.
10.1016/j.jhazmat.2009.09.132
CAS PubMed Web of Science® Google Scholar
52 S. Aber, H. Mehrizade, A. R. Khataee, Preparation of ZnS Nanocrystal and Investigation of Its Photocatalytic Activity in Removal of CI Acid Blue 9 from Contaminated Water, Desalin. Water Treat. 2011, 28, 92−96.
10.5004/dwt.2011.2207
CAS Web of Science® Google Scholar
53 A. Aleboyeh, M. B. Kasiri, M. E. Olya, H. Aleboyeh, Prediction of Azo Dye Decolorization by UV/H₂O₂ Using Artificial Neural Networks, Dyes Pigm. 2008, 77, 288–294.
10.1016/j.dyepig.2007.05.014
CAS Web of Science® Google Scholar
54 M. B. Kasiri, H. Aleboyeh, A. Aleboyeh, Modeling and Optimization of Heterogeneous Photo-Fenton Process with Response Surface Methodology and Artificial Neural Networks, Environ. Sci. Technol. 2008, 42, 7970–7975.
10.1021/es801372q
CAS PubMed Web of Science® Google Scholar
55 G. D. Garson, Interpreting Neural-Network Connection Weights, AI Expert 1991, 6, 47–51.
Google Scholar

Citing Literature

Volume44, Issue7

July 2016

Pages 809-817

Hybrid Homogeneous and Heterogeneous Photocatalytic Processes for Removal of Triphenylmethane Dyes: Artificial Neural Network Modeling