Implementation of a venturi photocatalytic reactor: Optimization of photodecolorization of an industrial azo dye

3.1 Chemicals and materials

The textile dye, Basic Red 46 (BR46, commercial name Astrazon Red FBL) was obtained from Aurassienne Spinning and Blankets (SAFILCO) Company, Algeria (molecular formula = C₁₈N₆H₂₁, color index number = 110,825, λ_max = 531 nm, Mw = 357.5 g/mol, azo group one, type cationic). Titanium dioxide powder used is TiO₂ Degussa P25 with greater than 99.5% purity (average primary particle size: 21 nm, BET surface area: 50 ± 15 m²/g). Initial pH of the aqueous solutions was adjusted by both sulfuric acid and sodium hydroxide.

3.2 The photocatalytic reactor

The first objective was to highlight the device ability for the implementation of photocatalytic reactions. The photocatalytic reactor is combined to an emulsion venturi operating in self-aspiration, on a recirculated boucle shown in Fig. 1 which consists of four main parts:

• –

  An emulsion venturi, consisting of a convergent and divergent separated by a throat diameter of 11 mm (Fig. 1). The diameters of the convergent and the divergent are similar and equal to 58 mm at the inlet and the outlet of the venturi. The device operates according to the Bernoulli energy conservation of a fluid, principle: In the constricted portion at the collar, the kinetic energy of fluid increases and the static pressure decreases. When the flow reaches a given value (between 400 and 500 L/h in our case see Fig. 2), the static pressure decreases below the atmospheric pressure. The external air is then self-aspired at the venturi throat through four symmetrically radial orifices. The liquid gas dispersion formed in the divergent causes a good oxygen transfer and a good homogenization of the liquid- solid–gas mixture (dye solution-TiO₂-air). Indeed, it is well known that the role of dissolved oxygen in photocatalytic process is to prevent the recombination of holes/electron pair formed in the semiconductor surface.

• Such a venturi is characterized by a ratio Q_G/Q_L < 1, this is suitable for the photocatalytic field, which does not require large quantities of oxygen. Such a device has been applied in aerobic biological processes. (Khellaf et al., 2014; Bouhelassa and Zoulalian, 1995a,b).

• –

  A cylindrical Pyrex Glass reactor (length: 700 mm, external diameter: 58 mm, thickness: 1.4 mm) made in Germany (Schott–Rohrglas GmbH Company) and delivered by SOMIVER (E.N.A.V.A) Company Algeria, The volume effectively irradiated was 1.584 L and the volume of colored water to be treated was 10 L. The reactor is irradiated by four UV-light lamps NARVA UV 010 (λ = 365 nm UV-A) to achieve a higher efficiency of photon, the distance between the lamps and the reactor wall was fixed at 35 mm The photoreactor and lamps were completely covered with an UV opaque cylinder in which the temperature is maintained constant by a controlled cooling air using adapted ventilators.

• –

  A recirculation loop includes a 25 l tank and a centrifuge pump fitted. The liquid recirculation flow rate can be controlled using a calibrated rotameter (CITEC DFM 200) placed between the pump outlet and the reactor inlet.

• –

  An air circuit mounted around of the venturi throat, perforated with four orifices through which, air is aspired, its flow rate is measured by a gas flowmeter (Rhone Duisburg GERMANY 07/96).

Close

Figure 1

Schematic representation of Emulsair photocatalytic reactor.

Export to PPT

Close

Figure 2

Variation of the flow rate ratio Q_G as a function of Q_L for different concentrations of BR46.

Export to PPT

3.3 Experimental design

The second main objective was to determine the optimum operating conditions for a given system that satisfies the specific operating conditions, the application of statistical experimental design reduced process variability combined with the requirement of less resources (time, reagents and experimental work), Response surface methodology (RSM) allows to solve multivariable equations and evaluate simultaneously the relative significance of several affecting factors even in the presence of complex interaction (Wu et al., 2010). RSM based on central composite design (CCD) was applied for optimization of photocatalytic decolorization process, consisting of 29 experiments for four variables (n = 4) and two levels low (−) and high (+). The total number of experiments was 29 determined by the expression: 2ⁿ (2⁴ = 16: factor points) + 2n (2 × 4 = 8: axial points) + 5 (center points: five replications), as shown in Table 3. The parameters in the design were studied at five different levels (−2, −1, 0, 1, 2), also the test factors were coded according to following Eq. (A.1):

The central composite design is based on a mathematical model which related the response Y to factors X₁, X₂, …, X_i, …, X_n, was employed for optimization of this process and the second-order polynomial response equation to correlate the dependent and independent variables: Eq. (A.2)

The statistical software MINITAB (16 Minitab Institute, USA) was used to fit the experimental data to the second-order polynomial equation.

3.4 Analysis

All the collected samples from the experiments were centrifuged at 4000 rpm for 25 min using a centrifuge (sigma 2–16) to remove the TiO₂ and color removal of the dye BR 46 was determined by UV absorption at λ_max = 365 nm using a UV–vis spectrophotometer (Shimadzu UV-160A) and calibration curve. The yield of decolorization (Y (%)) was expressed as the percentage ratio of decolorized dye concentration to that of the initial one.

The mineralization monitoring is carried out by the TOC measurement using a total carbon analyzer (V-CPH SHIMADZU).

4.1 Self-aspiration capacity

The influence of liquid flow on self-aspiration capacity of venturi was studied by varying the liquid flow (Q_L) between 500 and 1500 L/h. According to Bernoulli’s principle, when the liquid is pumped, at a high velocity, a low pressure is created in the venturi throat.

For the selected values of Q_L, the air was self-aspired in the venturi throat when the liquid pressure was lower than the atmospheric pressure (Table 1).

The effect of Q_L on self-aspiration capacity of the reactor was studied by representing in (Fig. 2) the variation of Q_G with respect to Q_L, in both clear water and charged with BR46 dye. The self-aspired air increased by increasing Q_L. Additionally, Experimental results presented show that the increase in air flow (Q_G) is more marked in the case when the recirculation liquid is charged with BR46 (15–45 mg/L). This phenomenon is probably due to the decrease in surface tension of the solution (Fig. 3) whose consequence is the decrease in bubble size due to the inhibition of the coalescence, which corresponds to the increase in the exchange surface-to-volume liquid ratio (Khellaf et al., 2014; Bouhelassa and Zoulalian, 1995a).

Close

Figure 3

Variation of the surface tension of solution with BR46 concentration (mg L⁻¹).

Export to PPT

The measurement of the surface tension (σ) of the solution for different concentrations of BR46 shown in (Fig. 3) has been performed using a tensiometer (KRUSS K8).

The variation of the surface tension (σ) as a function of dye concentration reveals that textile dyes reduce the surface tension of water, and this facilitates the formation of bubbles by reducing the energy required to create air–water contact surfaces, the obtained results are similar to those found by (Mazet et al., 1990) during a study concerning the removal of an industry textile dyes by wood sawdust.

On the other hand, it can be assumed that the friction loss created in contactor throat becomes lower in the presence of colored liquid and this effect leads to high flows of self-aspired air.

4.2 Photolysis and photocatalysis

To make a clear comparison in the efficiency of the photolysis under UV irradiation only and photocatalytic process (Fig. 4), two tests were carried out in the same operating conditions: pH (original) = 6.8, C₀ = 16 mg L⁻¹, [TiO₂]₀ = 0.5 g L⁻¹, Q_L = 1200 L h⁻¹, Q_G (self-aspired) = 365 L h⁻¹.

Close

Figure 4

Photodecolorization of BR46 as a function of time (pH (natural) = 6.8, C₀ = 15 mg L⁻¹, [TiO₂]₀ = 0.5 g L⁻¹, Q_L = 1200 L h⁻¹, Q_G (self-aspired) = 378 L h⁻¹, Q_G/Q_L = 0.315).

Export to PPT

It was observed that only 10% of decolorization was achieved after 150 min in the absence of the catalyst (photolysis). However, in the presence of TiO₂ Degussa P25 (0.5 g L⁻¹), it can be seen that 86% of BR46 is degraded after 150 min of irradiation. Note that for the experience of photocatalysis, the UV lamps were kept off during 30 min to achieve the adsorption equilibrium and then they were turned on.

4.3 Photocatalytic mineralization of BR46

Photocatalytic mineralization of the dye in UV/TiO₂ process was studied and monitored by TOC loss at the end of the photocatalytic process of BR46 solution (Fig. 5). TOC removal indicating the ultimate oxidation of BR46 was much slower than decolorization. The complete decolorization of BR46 occurred in 150 min, and thus it requires more than 400 min to reach the complete mineralization. This might be due to the formation of intermediates in BR46 solution with main dye molecules. TOC removal efficiency (Fig. 8) was determined as 93% at initial dye concentration of 15 mg/L, pH = 6.8, [TiO₂]₀ = 0.5 g L⁻¹, Q_L = 1200 L h⁻¹.

Close

Figure 5

TOC removal efficiency of BR46.

Export to PPT

4.4 Effect of emulsion venturi flow rate ratio

The oxygen resulting from the self-aspirated air by emulsion-venturi and the gas–liquid dispersion created in its divergent improve the photocatalysis process (Fig. 5); thus, the suspension homogenization (liquid–air–TiO₂) is favored and the transfer of oxygen is enhanced by the bubbles generated in the divergent section by the kinetic energy increase in the venturi throat. The recombination phenomenon of electron/hole pair is thus reduced. Otherwise oxygen captured by conduction band electron serves to form radicals such as O°₂ to give added photocatalytical effect. It has been shown that the degradation and mineralization rate constants increased with the dissolved oxygen concentration (Tseng et al., 2012). Several experiments also have shown that when all the oxygen in the system has been consumed, the process stops. However, their reinjection into medium has allowed restarting again the reaction (Piscopo et al., 2001).

The self-aspired gas flow rate is strongly influenced by the liquid flow rate; also it is more logical to consider the ratio (Q_G/Q_L). To evaluate the influence of this ratio on the rate of BR46 decolorization, several experiments were carried out at the same conditions of typical data for BR46 decolorization with initial concentration of [TiO₂]₀ = 0.5 g L⁻¹ at pH 6.8. The obtained results are summarized in Table 2.

Experimental results have proved that decolorization yield Y (%) of BR46 rises progressively as the flow rate ratio (Q_G/Q_L) increases at the same conditions and, could attain a yield of Y = 86.04% at high flow rate ratio.

4.5 Experimental design methodology and analysis of variance (ANOVA)

In this work, RSM was applied to assess individual and interactive effects of the four main independent parameters (initial dye concentration, concentration of TiO₂, flow rate and initial pH) on the decolorization efficiency Y (%) of BR46, the experimental ranges and the levels of the independent variables for BR46 removal are given in Table 3.

The four-level CCD design matrix and the experimental results obtained in the photocatalytic decolorization are presented in Table 4, which shows the statistical combinations of independent variables pH (X₁) Initial dye concentration (mg L⁻¹) (X₂) concentration of TiO₂ (g L⁻¹) (X₃) flow rate (L h⁻¹) (X₄) with the yield Y (%) (measured) was predicted over a period of 150 min. These predicted values are very close to the observed ones in all set of experiments.

Based on these results, an empirical relationship between the response and independent variables was attained and expressed by the following second–order polynomial equation (Eq. (C)):

Analysis of variance (ANOVA) study (Table 5) shows the results of the quadratic response surface model. ANOVA is required to test the significance and adequacy of the model. The ratio between the mean square of the model and the residual error is presented by the Fisher variation ratio F-value which is a statistically valid measure of how well the factors describe the variation in the data about its mean (Liu and Chiou, 2005). If the model is a good predictor of the experimental results, F-value should be greater than the tabulated value of F-distribution for a certain number of degrees of freedom in the model at a level of significance α. The high significant of the model is performed by (F-value = 24.02) which is much greater than the tabular F-value (F tabular = 3.48), indicating that the treatment differences are highly significant, F-value also indicates the variation from the model is significant or not when compared with the ones associated with residual error (Khataee et al., 2010). The quality of fitting the second-order equation was expressed by the coefficient of determination R². The R²-values provide a measure of how much variability in the observed response values can be explained by the experimental factors and their interactions. The R²-value is always between 0 and 1. The closer the R²-value is to 1, the better the model predicts the response (Liu and Chiou, 2005). It was found that the predicted values matched the experimental values reasonably well with R² = 0.96 and this implies that 96% of the variations for percent color removal are explained by the independent variables (Table 5). Adjusted R² (Adj-R²) is also a measure of goodness of a fit and corrects the determination coefficient R²-value for the sample size and the number of terms in the model. In this study the value of Adj-R² (0.9202) is very close to the corresponding R² value (Table 5).

Table 6 summarizes the estimates of regression coefficients accompanied with their standard errors, t-statistics and the corresponding p-values.

P-values were used as a tool to check the significance of each of the coefficients, which in turn, are necessary to understand the pattern of the mutual interactions between the test variables. The larger the magnitude of the Student’s t-test and smaller P-value, the more significant is the corresponding coefficient (Liu and Chiou, 2005).

The probability value (P) of coefficients was greater than 0.05, indicating the term did not have a significant effect on the predicted response (Jiang et al., 2013). Based on obtained results Table 6, P-value is relatively low, indicating the significance of the model, we can affirm that initial dye concentration (mg L⁻¹) (X₂), concentration of TiO₂ (g L⁻¹) (X₃), and flow rate (L h⁻¹) (X₄) with P values smaller than 0.0001 for 95% confidence level implied that terms were significant. Also pH (X₁) had a slightly negative effect with (p-value < 0.031). Moreover quadratic effects, shows that dye concentration presents a negligible P-value (p-value 0.002), indeed the mutual between dye concentration and TiO₂ loading (p-value < 0.0001) and pH solution with TiO₂ concentration (p-value < 0.007) are significant comparing to others factors.

The major diagnostic plots (Fig. 6) are used also to evaluate the residual analysis (difference between the observed and the predicted response value) of the response surface design, ensuring that the statistical assumptions fit the analysis data. Normal probability plots of the residuals are a suitable graphical method to verify whether the standard deviations between the actual and the predicted response values follow a normal distribution. The results illustrated in (Fig. 6) convey the general impression of a normal distribution of underlying errors, since the residuals fall near to a straight line; thus, there is no clear indication of non-normality of experimental results. Based on this plot, the residuals appear to be randomly scattered; thus, the model proposed is adequate and the constant variance assumption is confirmed.

Close

Figure 6

Variation of RB46 concentration as a function of time in photocatalysis with and without Venturi (PH = 6.8, C₀ = 15 mg L⁻¹, [TiO₂]₀ = 0.5 g L⁻¹, Q_L = 1440 L h⁻¹, Q_G self-aspired = 385 L h⁻¹, Q_G/Q_L = 0.267).

Export to PPT

4.6 Effect of variables as response surface

The main objective of the optimization is to determine the optimum values of variables for photocatalytic decolorization process, from the model obtained using experimental data.

Response surfaces shown in (Fig. 7) provide the results of the interaction between factors which are more significant.

Close

Figure 7

Residual plots for photocatalytic decolorization efficiency of BR46.

Export to PPT

Fig. 8(1) shows the effects of catalyst concentration and initial dye concentration for flow rate of 970 L h⁻¹ and initial pH = 7 at high levels of dye concentration in parallel with TiO₂ loading, the decolorization efficiency of the dye decreases as the concentrations increases in this case the photons got intercepted before they could reach the catalyst surface which decreases the color removal of BR46, likewise at very low concentrations of dye and catalyst, the response surface does not present any significant photodegradation due to the adherence of catalyst particles to the system walls, including to some non-illuminated sections.

Close

Figure 8

The response surface plots of photocatalytic decolorization efficiency Y (%) by the interaction between, (1) Initial dye concentration BR46 (mg L⁻¹) and TiO₂ concentration (g L⁻¹), (2) Initial dye concentration BR46 (mg L⁻¹) and flow rate Q_L (L h⁻¹) and (3) pH of solution and TiO₂ concentration (g L⁻¹).

Export to PPT

Additionally, the optimum catalyst loading for photodegradation is varied, and mainly depends on the dimension of the photoreactor (De Lasa et al., 2005).

In contrast, it should be noted that with decreasing the initial concentration of dye, the photocatalytic decolorization efficiency provides very significance results.

Fig. 8(2) represents the photocatalytic decolorization efficiency as a function of flow rate and initial dye concentration at catalyst concentration of 0.55 g L⁻¹ and initial pH of 7. The figure shows that the increase in decolorization efficiency Y (%) of BR46 was caused by an increase in flow rate at the higher values of the initial dye concentration and by a decrease in recirculated liquid flow at the lower values of the initial dye concentration. The presumed reason is that when the recirculated liquid flow is increased, the turbulence in the system is enhanced which ensures better dispersion of particles (BR46 and TiO₂) in the solution inside the reactor. It may lead to decompose more and more adsorbed dye molecules on the surface of TiO₂ and thus photocatalytic decolorization efficiency increases. The higher decolorization efficiencies at higher flow rates were also attributed to the increase in the mass transfer coefficient. (Dijkstraa et al., 2002) In the case of low flow the solution is more exposed to lamp irradiation which improves the formation of hydroxyl radicals (^•OH) and this leads to increasing in degradation efficiency.

The degradation efficiencies of dyes were observed in high levels at acidic solution, and it was also effective under alkaline condition as a function of increasing on TiO₂ loading (Fig. 8(3). BR46 is a cationic dye thus, in acidic medium the repulse force inhibiting the adsorption of dye molecule in the catalyst surface. But the formation of hydroxyl radicals remains possible which react with dye molecule, in basic medium, hydroxyl radicals were also produced on the catalyst surface by the following reaction (Eq. (D)) (Gozmen et al., 2009):