Modeling of reactive orange 12 (RO 12) adsorption onto gold nanoparticle-activated carbon using artificial neural network optimization based on an imperialist competitive algorithm
Graphical abstract
Introduction
Dyes and pigments have toxic nature and aromatic complex structure that generate more hazard problem for all ecosystems. Generally most of these compounds cause mutagenic, teratogenic and carcinogenic effects which subsequently lead to the generation of health disorders such as dysfunction of the kidney, reproductive system, liver, brain, and central nervous system [63]. Dyes according to their charge and nature classified to categories including anionic (direct, acid, and reactive dyes), cationic (basic dyes), and nonionic (disperse dyes). Among the anionic dyes are the brightest class of soluble dyes and generate serious environmental and health problems [29], [66]. These hazards and difficulties encourage the researchers to design novel approaches or material for their safe removal and subsequent achievement of clean aquatic environment [39], [40], [41], [61]. One of the best promising approaches that is greener toward others named adsorption. Adsorption with unique advantages such as simple and safe operation, lower toxicity and cost compared to other protocol is strongly recommended for this purpose [26], [27], [28], [31], [34], [37], [39], [44], [51], [56], [57], [60], [70]. Application of the nanoparticles due their high ordered structure, high mechanical strength and high specific surface area leads to a significant improvement in the performance of adsorption system. The presence of reactive and unsaturated atoms on their surfaces makes them potential for efficient binding the other atoms. Recently a growing interest for application of nanoparticles as sorbents for the removal of organic or inorganic pollutants devoted that may be assigned to points such as their safe and high efficiency synthesis with low cost, their high adsorption capacity that leads to a reduced consumption of material and large surface area, and their ability to incorporate and synthesise various nature and polarity nanoparticles by changing the nature and content of various atoms in their structure [14], [30], [32], [54]. Reactive orange 12 (RO-12) (Fig. 1; anionic dye) can be utilized for coloring silk, wool, leather, jute and cotton and can be applied as biological staining, dermatology, veterinary medicine, and green ink manufacture due to its stability toward both acidic and alkali solutions, being an ethylated product [33], [36], [46], [50]. RO-12 is toxic for humans and animals by causing permanent injuries to their eyes [31], [34], [38], [42], [45], [64], while limited and rare reports were guided on this dye removal. Most of the dyes are resistant to be decolorized by chemicals, heat, and light due to their complex chemical structures [19], [20], [35], [56], [57], [58], [68], [76]. The gold nanoparticles are widely utilized as photosensitive, surface enhanced Raman spectroscopy, as well as in chemical analysis due to its antibacterial properties and unique optical properties and was candidate as good replacement instead of antibiotics as antimicrobial agent. Because of its high specific surface area at the nanometer scale it suggests and candidates gold nanoparticles as a unique adsorbent for pollutant removal [9], [22], [65], [77]. The relationships between multi-input variables (pH, adsorbent dosage, agitation period, temperature, etc.) and output (removal percentage) in the adsorption process help the researchers to design an efficient and cheep protocol in wastewater treatment. Solving and modeling the complex relation between input and output variables can be simply performed by an artificial neural network (ANN) model imitated by biological neuron processing [11]. Multiple linear regression (MLR) [62] and ANNs have been widely applied for modeling and simulation of experimental data and real behavior of adsorption process [1], [3], [4], [5], [10], [12], [21], [48], [78]. Karimi and Ghaedi [48] used an artificial neural network model to predict the removal efficiency of methylene blue on the activated carbon (AC) prepared from peanut sticks and applied genetic algorithm (GA) for the optimization of effective variables. They modeled the adsorption kinetics via the trained ANN as fitness function with acceptable accuracy of ADD = 1.65% and R2 = 0.998. Badday et al. (2013) [81] applied an artificial neural network approach for modeling of ultrasound-assisted transesterification process of crude Jatropha oil catalyzed by a heteropolyacid based catalyst. The comparison of the results of the ANN model with regression analysis shows superiority of ANN models for successful and repeatable prediction of results. Celekli et al. [12] used an ANN model to predict the adsorption of Lanaset Red G (LR G) onto the lentil straw (LS) following optimization of variables such as adsorbent particle size, pH, initial dye concentration and contact time as inputs to construct a neural network for prediction of dye uptake as output. It was found that the ANN model is able for efficient prediction of adsorption data with low error and high coefficient values.
The meta-heuristic optimization algorithms are flexible with and/or without modification to assort specific problem. Different heuristic optimization algorithms are imitated by biological, physical and social processing [2], [7]. Recently, the imperialist competitive algorithm (ICA) as a meta-heuristic optimization method and a new socio-politically motivated strategy has been proposed [8], [74]. Taghavifar et al. [73] applied a hybridized artificial neural network and imperialist competitive algorithm optimization method for the prediction of soil compaction in the soil bin facility. Their results indicated that the combined ICA–ANN further succeeded to denote lower modeling error among [73]. To our knowledge, the adsorption of reactive orange 12 using gold nanoparticle loaded on activated carbon has not modeled with ANN and optimized with ICA in the literature. Therefore, the aim of the present study is to develop a batch system for the removal of RO-12 and to model and optimize the system characteristics using the ANN and ICA, respectively. From this point of view, the potential usage of gold nanoparticles loaded on activated carbon (Au-NP-AC) as environmentally friendly absorbent for the removal of RO-12 was investigated. The Au-NP-AC is applicable for the removal of the high amount of RO-12 in a short contact time with high adsorption capacity.
Section snippets
Materials
All chemicals, including NaOH, HCl, AC, RO-12 and other reagent with the highest purity available were purchased from Merck, Darmstadt Germany. Reactive orange 12 (Fig. 1), was used without any further purification and 20 mg of this pure solid was dissolved completely in 100 mL distilled water. The working solutions were prepared by successive dilutions of the abovementioned solution. The RO-12 concentration was determined at 416 nm using a Jusco UV–Visible spectrophotometer model V-530 (Tokyo,
Characterization of the Au-NP-AC
Determination of specific surface area by N2/77 K adsorption developed by Brunner–Emmett–Teller (BET) at equilibrium is always used to determine the adsorbent surface area to attain information about porosity and its diversity [22], [77]. Table 1 and Fig. 3(A)–(F) reveal the presence of various sizes in this porous material and the presence of high surface area which is accessible for more compounds to adsorb or diffuse into them. On the other hand, the most of pore volume belongs to size lower
Conclusion
This investigation shows that the efficiency of Au-NP-AC as a good, green and low-cost with high adsorption capacity (714.3 mg g− 1) for the removal of RO-12 from aqueous solutions in a short time (< 20 min) is usable. In this study, the effective pH was 1 and the optimum adsorbent dose was found to be 0.02 g. Langmuir isotherm gave a better fit to adsorption isotherms than Freundlich isotherm using linear and nonlinear methods. The Langmuir is the best model for fitting experimental data that may be
References (81)
- et al.
J. Hazard. Mater.
(2009) - et al.
Expert Syst. Appl.
(2013) - et al.
J. Hazard. Mater.
(2011) - et al.
Chem. Eng. J.
(2008) - et al.
Bioresour. Technol.
(2009) Process Biochem.
(2005)Expert Syst. Appl.
(2012)- et al.
Chem. Eng. J.
(2011) - et al.
Bioresour. Technol.
(2011) - et al.
Bioresour. Technol.
(2012)