Removal of Disperse Orange and Disperse Blue dyes present in textile mill effluent using zeolite synthesized from cenospheres

Dyes are one of the most hazardous contaminants released in to the environment and can be teratogenic, carcinogenic, and mutagenic to humans and other organisms if untreated dye-containing wastewater is discharged into water bodies (Markandeya et al. 2017d; Pandey et al. 2017). Dyes in wastewater cause color changes in the receiving water body, which is a sign of contamination due to dissolved metal ions, pigments, and other organic elements, and they also affect the aesthetics of the surrounding area (Dhiman et al. 2017b; Mohammed et al. 2020). Dyes can be naturally and synthetically derived and are commonly used in coloring of paper, electroplating materials, plastic, leather, cosmetic, food, pharmaceuticals, and textile mills (Singh et al. 2019; Ahmad et al. 2020). However, only textile mills produce very high levels of dye and floating solid materials containing effluent (Ojha & Markandeya 2016; Thirunavukkarasu & Nithya 2020). Therefore, removal of dye stuff from wastewater is necessary to ensure the safety of drinking water sources.

Disperse dyes are widely used for the coloring of polyesters, nylon, natural fibres, and acetate in the textile sector because of their bonding stability under different physical parameters (Kisku et al. 2015). Different chemical groups may be present in varying amounts, such as diazo dyes (10%), mono azo dyes (50%), styryl dyes (03%), anthraquinonoid dyes (25%), methine dyes (3%), acrylene benzimidazol (3%), amino naphthyl amide (1%), quinonaphthalon dyes (3%), and napthoquinone imine (1%) (Markandeya et al. 2017a). Azo dyes are the largest group of disperse dyes because they generate large numbers of molecular combinations, which results in a variety of dyes, and they also have a simple and low cost of production process compared to the expensive anthraquinone dyes (Markandeya et al. 2018). However, they have a complex aromatic fused structure that inhibits photosynthesis in aquatic plants (El-Shamy 2020). In many scientific studies, contact allergy has been reported after the use of synthetic textile dyes. In 1884 after the introduction of new synthetic organic dyes for dyeing fibres, numerous problems were reported such as skin eruptions caused by wearing dyed clothes (Markandeya et al. 2017b, 2017c). Further investigation revealed that dermatitis was associated with dyes and not by nylon (Markandeya et al. 2015b; Dhiman et al. 2017a). It is therefore imperative to develop methods of removing the dyes from wastewater in an efficient, ecofriendly, and economic way. Carvalho et al. (2011) studied adsorption of indigo carmine from the aqueous solutions using zeolite adsorbents. They developed zeolite using hydrothermal treatment with sodium hydroxide solution. The maximum adsorption capacity was found to be 1.23 mg/L. Lee et al. (2000) and Solanki et al. (2010) also used zeolite for the removal of wastewater impurities, but the adsorption capacity in zeolite was very low.

Statistical optimization of experimental conditions was carried out for the removal of dyes using zeolite derivatized from cenospheres. Response surface methodology (RSM) was used to reduce the total number of experiments and optimize the operating variables (Jawad et al. 2020e, 2021). RSM can enhance and optimize numerous process parameters and identify relative significances even in complex interactions (Markandeya et al. 2015a; Tiwari et al. 2015). It can establish favorable operational conditions for the process under investigation. RSM also offers many advantages like lowered process variability, increased percentage yield, and suitable confirmation of output response for both target and nominal achievements (Jawad et al. 2020a, 2020b, 2020c; Surip et al. 2020). The statistical design of the experiments determines which factors have the most effect on the response (percentage dye removal) (Jawad & Abdulhameed 2020a; Malek et al. 2020).

RSM is a sequential procedure that helps to quickly and efficiently identify optimum conditions (Tiwari et al. 2013; Reghioua et al. 2021b). Although statistical design of experiment has been mostly used in the optimization of industrial processes with multiple parameters (Haq et al. 2018), it has rarely been applied to adsorption processes (Jawad & Abdulhameed 2020b; Jawad et al. 2020d; Reghioua et al. 2021a). There is scope to determine the potential of statistical design in adsorption studies. The objective of this study was to determine a region of vector space in which operating requirements are satisfied for the removal of dyes using zeolite derivatized by cenospheres from wastewater in an efficient, ecofriendly, and economic manner.

The molecular weight (g/mol) of Disperse Orange 25 (DO, C₁₇H₁₇N₅O₂) and Disperse Blue 79:1 (DB, C₂₃H₂₅BrN₆O₁₀) dyes was 323.35 and 625.38, respectively, and λmax (nm) was 470 and 570, respectively.

Zeolite was prepared from modified cenospheres to enhance the adsorption capacity. Coal fly ash was modified as per the method described by Markandeya et al. (2017c) and (USEPA 1990) to obtain the modified cenospheres. The modified cenospheres were mixed with alumina and calcium hydroxide in the ratio of 1:6:1 (w/w). To make slurry (A) with a concentration of 2 weight%, the mixture was mixed with Milli-Q water. Then the slurry (A) was mixed with sodium bicarbonate as the binding mixture in different proportions in four separate columns:

• –

  slurry (A);

• –

  slurry (A) and sodium bicarbonate (10:1 w/w);

• –

  slurry (A) and sodium bicarbonate (10:2 w/w);

• –

  slurry (A) and sodium bicarbonate (10:3 w/w).

Sodium bicarbonate was used as a binding material to prevent percolation and the ratio 10:2 (w/w) of slurry (A) and sodium bicarbonate produced the best result. The zeolite produced from modified cenospheres was used as an adsorbent for this study (Figure 1).

RSM is an empirical technique used to identify operating variables for the optimum removal of contaminants or impurities from wastewater. The variables affecting the percentage removal of dyes were investigated using the Box–Behnken design (BBD). The experimental design was carried out using Design-Expert software, v 11 (Stat-Ease Inc., USA). All experiments were performed in batch system using a 250-mL reactor containing 100 mL of DO or DB dye solution. The variable input parameters were contact time from 60 to 140 min for DO and 80 to 160 min for DB, dye concentration from 20 to 60 mg/L, agitation speed from 100 to 180 rpm, pH values from 4 to 8 for DO and 4 to 10 for DB dyes, and adsorbent dosage values from 0.2 to 1.0 g/L for DO and 0.4 to 1.2 g/L for DB dyes. Selection of process parameters was done on the basis of batch experiments conducted previously (Markandeya et al. 2017d). Five independent parameters for optimization were contact time (A), dye concentration (B), agitation speed (C), pH (D), and adsorbent dosage (E). The main processes involved in the optimization of multiple variables included statistically designing the experiment, estimating the coefficient, analyzing the response, and checking the suitability of the model (Box & Behnken 1960; Jaafari et al. 2020).

The model coefficients (β_i) are determined and used for predicting response values for different combinations of coded values of the variables.

This study analyzed and quantified the characteristics and morphological properties of the zeolite adsorbent. The, roughness/smoothness of the surface and other different properties of the cenospheres and zeolite are shown in Figure 2a and 2b, respectively. Cenospheres are uniform spherical aggregates with a rough and porous texture which provides high surface area to mass ratio, thereby facilitating the adsorption of dye molecules. The insets in Figure 2b (scaled at 10 μm) show scanning electron microscopy (SEM) images of zeolite at higher magnification and represents aggregates of zeolite crystals with defined grain boundaries (Dhiman et al. 2017b). Energy-dispersive X-ray (EDX) spectroscopy plots in Figure 3 shows that zeolite (b) contained more aluminum, calcium, and silica than the metal-leached-out-cenospheres (a).

Brunauer–Emmett–Teller analysis showed that the surface area was 2,832.08 m²/g and pore volume was 8.37 cm³/g at pores less than 2,153 A° at P/Po, 0.9910, whereas pore size was 698.32 A° Barrett–Joyner–Halenda average adsorption pore diameter of 4 V/A (Kacan 2016).

Spectral analyses of cenospheres and zeolite using attenuated total reflection–Fourier transform infrared (ATR–FTIR) spectroscopy are shown in Figure 4. Cenospheres (a) displayed two peaks at 1,053 and 798 cm⁻¹ analogous to an asymmetric and symmetric marking of M–O–M (M = Si/Al). However, some intense peaks were observed due to desired modification on cenospheres. Jain et al. (2013) also found such peaks after modification of coal fly ash, which matches this study. Moreover, a peak at 1,463 cm⁻¹ in zeolite spectrum (Figure 4b) demonstrated the presence of sodium bicarbonate. Mason et al. (2006) also found sodium bicarbonate at the same peak length. The presence of the weak band at 3,463 cm⁻¹ is related to O–H stretching of water molecules.

The size of the particles was in the micrometer range and their diameter was heterogeneous (Figure 5(a)). This variation can be explained by non-uniform distribution of raw cenospheres (Figure 6(a)). The smallest particle size (70–300 nm) determined by TEM is shown in Figure 5(b).

Particle size distribution plot shown in Figure 6(a) demonstrate that the most common granular size was 30–60 μm, whereas Figure 6(b) of formed zeolite shows that the most common granule size was in the range of 10–50 μm.

We applied three isotherms – Langmuir, Freundlich, and Temkin–and found that Langmuir isotherm better correlated compared to the others for the adsorption of dye onto the zeolite surface. The monolayer maximum adsorption capacity was found to be 125 mg/g and 109.8 mg/g for DO and DB dyes, respectively. The values of R_L = 0.0909 and 0.1656 also revealed a favorable isotherm for DO and DB dyes, respectively. Based on R² values of pseudo-first-order and second-order kinetic models, the adsorption process in this study follows the pseudo-second-order kinetic model as it gives a maximum value of R² (Markandeya et al. 2017b).

RSM-based BBD was used to optimize conditions for maximum uptake of dyes onto the zeolite surface. A statistical model was developed using quantitative data obtained from experiments using Equation (4) and transforming this equation in terms of coded and actual factors as shown in Equations (5)–(8):

F-test and correlation coefficient (R²) were used to check significance level. Model terms were evaluated with 95% confidence level using p-values. 3D plots were developed for removal of DO and DB dyes.

The F-value 12.67 implies that the model for the removal of DO dye is significant. There is only a 0.01% chance that large F-values could occur due to noise. The values of Prob > F<0.05 confirmed that the model fits significantly. In this case, the significant model terms are A, B, C, D, E, A², B², D², and E² and values greater than 0.1 indicate that the model terms were not significant (Table 1). If there are a number of insignificant model terms (without including those required to support hierarchy), the model can be reduced.

The model F-values of 11.75 imply that the model developed for DB dye fitted significantly. The values of Prob > F<0.05 clearly indicate that the model terms were significant. The significant model terms are A, B, C, D, E, DE, B², D², and E² and values greater than 0.1 indicate that these model terms are not significant (Table 2). Moreover, the lack-of-fit F-value of 1,397.58 specifies that the lack-of-fit was significant. There is a very small probability (only a 0.01%) that a large lack-of-fit F-value could occur due to noise. Table 3 presents the model fit summary of the quadratic models developed for both dyes.

For DO dye, the value Pred R² was 0.6409, which is better correlated with the value Adj R² of 0. 8384. If the ratio of signal (observation) to noise (error) is greater than 4, then it is desirable. The present ratio of 14.084 indicates an adequate signal, and therefore this model can be used to navigate the design space successfully.

For DB, the value Pred R² was 0.6154, which is in reasonable agreement with the value Adj R² of 0.8269. Adeq precision also measures the signal (observation) to noise (error) ratio and, in this case, the value 13.768 indicates adequacy of signal. Therefore, this model can be effectively used to navigate the design space.

Validation of model was performed using different combinations of five variables: contact time, pH, agitation speed, adsorbent dosage, and dye concentration for DO and DB dyes. The predicated and experimental values of the model were compared to determine validity of the models (Tables 4 and 5).

The optimization process was accomplished using numerical node under optimized conditions. During optimization, five adsorption governing factors (pH, contact time, agitation speed, dosage ,and initial dye concentration) were adjusted and maximum removals were reported for both DO and DB dyes. The Design-Expert software integrated individual factors at tunable levels.

To achieve further understanding regarding effects of the independent parameters and their interactions with dependent parameters (removal), 3D response surface plots for the measured responses were made based on the quadratic models. 3D response surfaces indicate that the interaction between the corresponding parameters are negligible. An elliptical or saddle shape of the contour plots indicates a significant interaction between two variables (Liu & Chiou 2005). In Figure 7, the upper horizontal axis represents the dependent variable and the almost parallel contour lines suggest that there is no significant interaction between the variables.

3D response surface plots and 3D contour plots (Figure 7(a) and 7(b)) for the removal of both dyes show that zeolite (adsorbent) was a function of dye concentration and contact time. The percentage removals of DO and DB dyes initially increased rapidly as the contact time increased and decreased as the contact time decreased. As the dye concentrations increased, the percentage removal of both dyes decreased within the respective experimental ranges. However, DO dye was found to be more prominent compared to the DB dye on contact time and dye concentration (Figure 7(a) and 7(b)). This may be due to the presence of more unstable chemicals in the matrix of the DO dye (Nandi et al. 2009). Shukla et al. (2014) also found the same results. It has also been observed that the removal of both dyes increases with an initial increase in dye concentrations (10 to 40 mg/L), but an increase from 40 to 60 mg/L casued the removal of dye to decrease. This may be because of the relatively limited adsorption sites for higher dye concentrations, i.e., >50 mg/L.

Figure 7(c) and 7(d) presents the 3D response surface plots for the percentage removal of DO and DB dyes as a function of agitation speed and contact time. It is clear that the removal percentage increased with the increasing contact time, but there was no significant effect for agitation speed (Wang 2013). The curved portion of the Figure 7(c) and 7(d) indicates that there is almost no interactive influence of these two parameters on decolorization.

Figure 7(e) and 7(f) shows 3D response surface plots for decolorization as a function of pH and contact time. In this case, the removal percentage increased with increasing pH value (from pH 2 to 6) but decreased with a further increase in pH up to 8 (Yahia & Sellaoui 2020). The curved portion in Figure 7(e) and 7(f) indicates that for the removal of the dyes, there was good interaction between pH and contact time.

Figure 7(g) and 7(h) shows 3D response surface plots for decolorization as a function of adsorbent dosage and contact time. It can be seen that the percentage removal of both dyes increased with increasing adsorbent dosage. However, the removal of DO dye was greater compared to DB dye, as shown in Figure 7(g). The percentage removal increased as the adsorbent dosage increased from 0.2 to 1.0 g/L; however, at a lower dosage of 0.2 g/L, fewer dye molecules could reach the limited number of available adsorption sites, resulting in lower removal. Increasing adsorbent dosage to 1.0 g/L resulted in the increased removal due to more sites available for the uptake of both dyes (Zhang et al. 2020).

Cubical representations of modeling for the removal of both DO and DB dyes were determined and the maximum values of dye removal were 96.00% and 95.23%, respectively, at the optimized parameters for DO dye and DB dye.

The maximum adsorption compacity of the present adsorbent was compared with other zeolite (Table 6). It is clear that the present zeolite adsorbent has good potential for the removal of both dyes compared to other zeolites prepared using different materials. In this study, the adsorption capacities for DO and DB dyes were 125.0 mg/L and 109.8 mg/L, respectively.

This investigation was based on the removal of disperse dyes from colored wastewater using an optimization technique. The optimization technique was based on the batch process and focused on the interactions between various parameters: pH, contact time, agitation speed, adsorbent dosage, and dye concentration. Related R² values of 0.9102 and 0.9038 proved that predicted data fitted well with actual data for both dyes. Based on the optimized conditions, the maximum percentage removal of DO was 96.00% (contact time: 119 min, adsorbent dosage: 0.67 g/L, shaking speed: 158 rpm, pH 6.10), and for DB dye it was 95.23% (contact time: 122 min, adsorbent dosage: 1.05 g/L, shaking speed: 145 rpm, pH 5.68). It was also observed that zeolite can be used for efficient removal of disperse dyes. ANOVA and regression modeling showed that all the process parameters have values of Prob. >F< 0.05, which indicates that the RSM-based BBD model was significant in the present study. The study also concluded that modified adsorbent zeolite has a high potential for use in the removal of dyes from colored wastewater.

The first author is highly grateful to Vibhuti Mishra for language correction. The work was financially supported by TEQIP II (TEQIP-II 22).

There is no conflict of interest.

All authors have read and approved the manuscript.

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