Optimization of TiO₂-P25 photocatalyst dose and H₂O₂ concentration for advanced photo-oxidation using smartphone-based colorimetry

Usually organic contaminants are removed from wastewater by coagulation/flocculation (Liu et al. 2018; Anastopoulos et al. 2020), reverse osmosis (Naushad 2014; Tatarchuk et al. 2019b), and adsorption (Naushad 2014; Danyliuk et al. 2020b). Another approach is advanced oxidation for deep destruction of organic pollutants. Advanced photocatalytic oxidation uses solar energy to form hydroxyl radicals from hydrogen peroxide (Mironyuk et al. 2020; Tatarchuk et al. 2020). Among the various photocatalysts, the most common are TiO₂ (Chen et al. 2020; Danyliuk et al. 2020a, 2020c), ZnO (Deebansok et al. 2020; Hoong et al. 2020), FeO (Sharma & Lee 2016), CdS (Liu et al. 2020), SnO₂ (Karthik et al. 2018; Wang et al. 2021), and ZrO₂ (Bailón-García et al. 2017; Abdi et al. 2019). These semiconductor materials have a moderate band gap and are able to form free electrons and holes under UV irradiation. The formed holes (h⁺_vb) have a high oxidizing potential, which allows direct oxidation of organic contaminants. Quite often the formed holes and electrons recombine without any reaction. Adverse recombination of charge carriers can be reduced by modifying the surface of the photocatalyst. Yadav Yadav et al. (2017) reported a combined strategy to slow down the surface recombination process by doping TiO₂ with tungsten and forming a nanocomposite with reduced graphene oxide (rGO). The TiO₂ sample doped with 1 wt.% of tungsten, provides the longest life of charge carriers and the highest photocatalytic activity in the degradation of the model contaminant p-nitrophenol in an aqueous solution. It is possible that tungsten atoms reduce interband recombination in the volume of the catalyst. Reduced graphene oxide strongly interacts with TiO₂. The result is rapid transfer of excited electrons to the tungsten atoms instead of surface recombination. The optimal ratio of rGO to TiO₂ is 1:1 (Yadav Yadav et al. 2017).

Pham & Shin (2019) synthesized a new photocatalyst composed of molybdenum (Mo)-doped NiTiO₃ and g-C₃N₄. The catalyst effectively inhibits recombination of electrons and holes by transferring the charge from NiTiO₃ to the g-C₃N₄ phase. The doping of the NiTiO₃ structure with Mo creates additional oxygen vacancies (V_O). At the same time, new bonds are formed between N and Ti atoms. The new composite photocatalyst promotes the degradation of the Methylene Blue dye under exposure to visible light. The degradation rate constant for the composite photocatalyst is increased 6.5 times compared to the constant with pure NiTiO₃. Photoluminescence spectra indicate that recombination in the composite photocatalyst is effectively slowed down by both doping with Mo and binding to g-C₃N₄.

The series of TiO₂ photocatalysts doped with Fe³⁺ has revealed the effect of doping degree (Xin et al. 2007). At a low degree of doping (Fe/Ti = 0.03% mol/mol), Fe³⁺ ions capture free electrons and inhibit the recombination of electron-hole pairs. As a result, the reaction rate is increased. When the Fe³⁺ doping exceeds 0.03% mol/mol, the formed Fe₂O₃ nanoparticles become active centers for the rapid recombination of electrons and holes. As a result, the reaction rate is decreased.

Ferrites are known as visible light active photocatalysts (Sundararajan et al. 2017; Tatarchuk et al. 2019a). Nanometer-sized ferrite particles with the general formula Mg_0.5Ni_xZn_0.5-xFe₂O₄ (x = 0.1–0.5) are active in the decomposition of Reactive Blue-19 (RB-19) dye. The Mg_0.5Ni_0.4Zn_0.1Fe₂O₄ sample has the highest activity. The degree of decomposition as much as 99.5% was recorded after 90 min of exposure to visible light. The photoluminescence data confirmed a significant reduction in charge carrier recombination. The modified structure of spinel ferrites probably ensures the formation of aggressive ·O²⁻ radicals resulting in rapid degradation of the RB-19 dye (Hayashi et al. 2019).

An effective way to reduce recombination of charge carriers is to add an external electron acceptor, such as H₂O₂ (Dionysiou et al. 2004). The reaction of H₂O₂ with free electrons causes the formation of highly aggressive hydroxyl radicals (Liu et al. 2012; Hayashi et al. 2019). The advantages of the H₂O₂ oxidizer are low cost, thermal stability, and good water solubility. However, an overdose of H₂O₂ can reduce the rate of photocatalytic oxidation. The excess H₂O₂ reacts with formed hydroxyl radicals producing less active molecular oxygen. Therefore, the optimal amount of H₂O₂ should be set separately for each photocatalyst and wastewater.

In the last decade, red, green, blue (RGB) measurements have often been used to determine dye concentration. The RGB color model is an additive color model in which primary red, green, and blue colors are added together to form a broad variety of other intermediate colors. RGB color is obtained by mixing red, green, and blue in different proportions: each hue can be described by three numbers, denoting the brightness of the three primary colors. Any electronic image created by digital cameras (pictures on a computer monitor, photographs on a phone screen) are based on the RGB model. The accuracy of RGB camera color measurements is acceptable for analytical purposes. The analysis of digital images taken by a smartphone can be used in analytical chemistry, including water quality monitoring (Rezazadeh et al. 2019; Golicz et al. 2020). The advantage of our simplified approach is the ability to instantly obtain the concentration of contaminant using an image from smartphone, without wasting time with solution sampling (Danyliuk et al. 2020a, 2020c). Safarik et al. (2019) described the direct determination of dye concentration with a smartphone. Image analysis was carried out using the ON Color Measure and Color Lab applications providing numerical data in the RGB and hue, saturation, lightness (HSV) color coordinates, respectively. Özdemir et al. (2017) described the determination of dyes in the 0 to 5 ppm range using a smartphone-based spectrometer and a commercial spectrophotometer.

In this work, the optimal doses of photocatalyst and oxidant in the advanced oxidation reaction were quickly estimated using a color analysis application on a smartphone. The model substance was the Direct Red 23 azo dye (DR23). The molecular structure of the DR23 dye contains the chromophore –N = N–, which is susceptible to attack by radicals. The degradation of the azo-group leads to the fading of the red color recorded in color images. The degradation rate was measured with a smartphone camera without sampling. The combined effect of photocatalyst and oxidant amounts was described quantitatively and the optimal parameters were found. The novelty and advantages of this study is the possibility to use smartphones for fast and easy measurement of the rate of dye photodegradation directly in the solutions without sampling. In recent years, smartphones are widely used for colorimetric measurements in analytical chemistry (Rezazadeh et al. 2019) and in water testing (Golicz et al. 2020). However, as far as we know, there are no studies on the use of smartphone to register rate of dye photocatalytic degradation.

The photocatalyst was commercial Aeroxide TiO₂ P25 from Evonik (Shayegan et al. 2019). Reagent grade H₂O₂ (31.5%) was obtained from SferaSim (Ukraine). Technical grade Direct Red 23 dye (CAS 3441-14-3) was obtained from Boruta (Poland).

Phase composition of the photocatalyst was analysed using X-ray diffraction (XRD) at λ_Kα1 = 0.154 nm (STOE STADI P diffractometer, Germany). The pore size distribution in the P25 sample was measured by N₂ adsorption/desorption at 77 K using a Quantachrome Autosorb Nova 2200e surface area analyzer.

The microphotoreactor has been described in a previous paper (Danyliuk et al. 2020a, 2020c). The UV irradiation source was a UV light-emitting diode (LED) (3 W). The LED brand was UV LED Epistar BIN 1. The operating paremeters were as follows: 360–365 nm, 700 mA, 8–15 lm. The model dye Direct Red 23 has the minimal value of UV absorbance at a wavelength of 360 nm. Therefore, UV irradiation with a wavelength of 365 nm causes minimal direct photolytic degradation of the dye. In this a way, the effects of the photocatalyst and oxidant are more pronounced. The reaction cuvette was filled with the exact volume (20 mL) of DR23 solution (25 mg/L). Hydrogen peroxide concentrations ranged from 0 to 10 mM. P25 photocatalyst doses ranged from 0.5 to 1.5 g/L. The reaction mixture was stirred in the dark for 20 min to reach equilibrium for the adsorption of DR23 on the TiO₂ surface. Subsequently, UV irradiation was used for 30 min. A blank experiment (without UV irradiation) showed no adsorption of DR23 dye on the TiO₂ surface. After the centrifugation, no undegraded DR23 dye was observed on the TiO₂ surface. The TiO₂ precipitate was white.

The kinetic lines of DR23 dye degradation were recorded with a Samsung Galaxy A6 smartphone (SM-A600FN). At specified intervals, the microreactor was opened and images of the reaction mixture were recorded (Figure 1). Due to the white color of TiO₂, the photocatalyst suspension was a good background for the DR23 dye solutions. The measurement procedure is described elsewhere (Danyliuk et al. 2020a, 2020c). To ensure better reproducibility of the color value measurements, the background of the scene was also black. The front light was a 7-W LED lamp with a color temperature of 6,500 K (Figure 1). As a result of degradation of the DR23 dye, the reaction mixtures changed color from bright red to off-white. RGB values of the registered images were obtained using the application Spectrum (available on Play Market).

Thirty-five experiments were carried out with the central composite experimental design. The two independent variables were the concentration of H₂O₂ (0; 0.5; 1; 2.5; 5; 7.5; 10 mM) and the dose of the TiO₂ P25 photocatalyst (0.5; 0.75; 1.0; 1.25; 1.5 g/L). The dependent variables were the degradation rate constant and the final degree of degradation. The computations were made in the Design-Expert V.8.0.6 software.

The results of the XRD analysis of the P25 photocatalyst sample are shown in Figure 2(a). The diffractogram indicates that the sample is composed of anatase (87.9%) and rutile (12.1%) phases. This combination of TiO₂ polymorphs provides the high activity of the P25 photocatalyst. The peaks at 25.3°, 37.8°, 47.6°, 53.8°, and 62.1° correspond to planes (101), (004), (200), (105), and (213) of the anatase crystal structure. The peak at 27.6° corresponds to the plane (110) of the rutile phase. The lattice parameters for anatase are a = 3.7855 Å and c = 9.5064 Å, while for rutile they are a = 4.5928 Å and c = 2.9588 Å. The average crystallite size was calculated according to the Scherrer formula:⁠, where β₍₃₁₁₎ is the full width at half maximum of the (3 1 1) reflection, and θ is Bragg's angle of reflection. The Brunauer–Emmett–Teller analysis data of the P25 sample (N₂ adsorption/desorption isotherms and pore size distribution) are presented in Figure 2(b) and 2(c). The specific surface area of the P25 sample is 31.72 m²/g. This small surface area suggests rather low adsorption properties. This is consistent with the study by Bouanimba et al. (2018) which showed that the P25 photocatalyst is not able to adsorb the Bromothymol Blue dye and it provides much faster photodegradation compared to other TiO₂ samples. Moreover, previous studies have shown that the P25 photocatalyst provides rapid degradation of many organic compounds (Katal et al. 2020).

Our previous study (Danyliuk et al. 2021) substantiates the established concentrations of the DR23 dye, H₂O₂, and TiO₂ P25. The concentrations of dyes in wastewater from the textile industry usually range from 10 to 50 mg/L (Sohrabi & Ghavami 2008). The model solution of DR23 dye had a concentration of 25 mg/L, which is in this range. Our previous study (Danyliuk et al. 2021) showed that the optimal concentration of H₂O₂ is 10–25 mM for the Rhodamine B dye degradation. For the dose of TiO₂, it is a wider range from 0.2–3 g/L, depending on photoreactor geometry and pollutant concentration (Torkaman et al. 2016). Our previous study (Danyliuk et al. 2021) also showed that the optimal concentration of the P25 photocatalyst is 1.5 g/L for the microphotoreactor used.

Calibration lines for smartphone measurements were constructed by stepwise adding the dye stock solution directly into the reaction vial and capturing color images. The registered colors and corresponding RGB values are presented in Table 1. Among the RGB values, the R value turned out to be the most useful for determining the concentration of the DR23 dye (Figure 3). Dye DR23 has a red color and a maximum absorption at 510 nm. For this reason, the intensity of red can be used as an analytical signal during image analysis. The calibration equations for the red color component (Figure 3(a)–3(c) and Table 2) showed the highest coefficient of determination (R² is in the range from 0.979 to 0.994). This is why the R component was chosen to determine the dye DR23. The calibration coefficients are dependent on the dose of TiO₂. Generally, the higher the dose of TiO₂, the larger the coefficients of the equation (Table 2).

Various approaches to RGB data analysis have been reported in the literature (Colzani et al. 2017; Neto et al. 2019). The use of R, G, or B values alone provides acceptable calibrations. However, in our previous study (Danyliuk et al. 2020a, 2020c) we proposed a new normalized component R′ which is calculated as follows: ⁠. This normalized component is less susceptible to variations of the external light. The normalized component provides higher values of correlation coefficient (R²) in a wider concentration range.

The kinetic lines of degradation of the DR23 dye are presented in Figure 4 (both linear and semi-logarithmic forms). The semi-logarithmic lines (Figure 4(b), 4(d), 4(f), 4(h), 4(j)) are rather linear. This suggests that a first-order kinetic equation can be used to describe the obtained experimental data. The calculated values of the reaction rate constants are presented in Table 3. Corresponding values of the coefficient of determination (R²) range from 0.923 to 0.999 (Table 3). The high values of the coefficient of determination confirm that the reaction kinetics is well described by the first-order model.

In general, a photocatalyst dose equal to 1.0 g/L (20 mg) provides higher values of the reaction rate constant (Table 3). The maximum value of the reaction rate constant (k = 0.0889 min⁻¹) was recorded at TiO₂ dose equal to 1.0 g/L (20 mg) and H₂O₂ concentration equal to 2.5 mM.

The reference data were the final dye degradation values determined by spectrophotometric measurements at 510 nm. The numerical values of final degree of degradation are presented in Table 3. Figure 5 shows the 3D graph of the dye degradation degree as a function of the photocatalyst dose (A) and the concentration of H₂O₂ (B). The maximum degradation efficiency of DR23 was achieved at a concentration of H₂O₂ of 2.5 mM and a photocatalyst dose of 1.0 g/L. Higher concentrations of H₂O₂ and photocatalyst doses lead to a lower degree of degradation. The probable causes are reduced penetration of UV radiation into the reaction volume and a side reaction of H₂O₂ with hydroxyl radicals. Figure 6 shows the surface plot of the final degree of degradation (%) as a function of the photocatalyst dose (A) and the concentration of H₂O₂ (B).

We observed this phenomenon during experiments.

The central composite experimental design used a two-factor matrix. The independent variables were the photocatalyst dose (in mg) and the concentration of H₂O₂ (in mM) decoded as A and B, respectively. The dependent parameters were the degradation rate constant (Y) and the final degree of degradation (Z). Polynomial functions for both parameters were computed on the basis of the experimental data presented in Table 4.

Using the obtained Equation (1), the predicted values of the degradation rate constants were calculated. The predicted values were plotted against the actual experimental values (Figure 7). The coefficient of determination R² is 0.917. This indicates that the obtained equation is quite reliable.

Using the obtained Equation (2), the predicted values of the degradation rate constants were calculated. The predicted values were plotted against the actual experimental values (Figure 8). The coefficient of determination R² is 0.675. This indicates that the obtained equation is quite reliable. The summary statistics of the models for the degree of degradation is presented in Table 5.

Validity of the obtained quartic models was also assessed using analysis of variance (ANOVA). Tables 6 and 7 show the results of evaluation. ANOVA divides the full variance of the results into two parts: one related to the model and the other related to the experimental errors. The goal is to determine whether the deviation from the model is significant or not. This is done by calculating the value of F, expressed as the ratio of the quartic error to the residual error of the mean model. If the calculated value of F exceeds the value of the tabular value of F, then the model has a strong deviation from the experimental data.

The obtained value of F = 15.80 proves the usefulness of the model. It was found that A, B, A², B³, A²B², A⁴, B⁴ are important parameters of the model, while AB, B², A²B, AB², A³, A³B, and AB³ are insignificant parameters for the correct description of the reaction rate constant. The F values in Table 6 indicate that variable B (concentration of H₂O₂) has a more significant effect on the reaction rate constant compared to variable A (photocatalyst dose).

The obtained value of F = 14.31 proves the usefulness of the model. The model parameters B, A², B², A³, B³, A⁴, and B⁴ are significant ones, while the parameters A, AB, AB², and AB³ are insignificant ones. The F values in Table 7 indicate that variable B (concentration of H₂O₂) has a more significant effect on the degree of degradation compared to variable A (photocatalyst dose).

In this work, the optimal doses of photocatalyst TiO₂ (P25, Degussa) and oxidant H₂O₂ in the advanced oxidation process were quickly estimated using a color analysis application on a smartphone. Photocatalytic degradation of the DR23 dye was used as a model reaction. The Samsung Galaxy A6 smartphone showed good repeatability for the DR23 dye determination via color measurements. Thus, the rate of the advanced oxidation reaction was measured with the smartphone camera without sampling. Standard spectrophotometric measurements were used as reference data. The two measuring techniques produce very similar results. Surface response graphs plotted in the Design-Expert software are convenient for comparison purposes. The optimal doses of TiO₂ and H₂O₂ were determined. It was concluded that the maximum degradation efficiency of DR23 was achieved at H₂O₂ initial concentration equal to 2.5 mM and photocatalyst dose equal to 1.0 g/L. The higher H₂O₂ concentrations lead to degradation reduction due to the side reaction of H₂O₂ with OH radicals and decreased penetration of UV radiation into the reaction volume. Smartphone measurements are useful for fine-tuning the oxidation parameters. The proposed approach can be applied under field conditions in wastewater treatment plants.

The authors thank the Ministry of Education and Science of Ukraine for financial support in the framework of projects 0120U104158 and 0120U102035.

All relevant data are included in the paper or its Supplementary Information.