Bioresource Technology: Yuyi Yang, Guan Wang, Bing Wang, Zeli Li, Xiaoming Jia, Qifa Zhou, Yuhua Zhao

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Bioresource Technology 102 (2011) 828834

Contents lists available at ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Biosorption of Acid Black 172 and Congo Red from aqueous solution by nonviable Penicillium YW 01: Kinetic study, equilibrium isotherm and articial neural network modeling
Yuyi Yang, Guan Wang, Bing Wang, Zeli Li, Xiaoming Jia, Qifa Zhou, Yuhua Zhao
College of Life Sciences, Zhejiang University, Hangzhou, Zhejiang Province 310058, PR China

a r t i c l e

i n f o

a b s t r a c t
The main objective of this work was to investigate the biosorption performance of nonviable Penicillium YW 01 biomass for removal of Acid Black 172 metal-complex dye (AB) and Congo Red (CR) in solutions. Maximum biosorption capacities of 225.38 and 411.53 mg g1 under initial dye concentration of 800 mg L1, pH 3.0 and 40 C conditions were observed for AB and CR, respectively. Biosorption data were successfully described with Langmuir isotherm and the pseudo-second-order kinetic model. The Weber-Morris model analysis indicated that intraparticle diffusion was the limiting step for biosorption of AB and CR onto biosorbent. Analysis based on the articial neural network and genetic algorithms hybrid model indicated that initial dye concentration and temperature appeared to be the most inuential parameters for biosorption process of AB and CR onto biosorbent, respectively. Characterization of the biosorbent and possible dye-biosorbent interaction were conrmed by Fourier transform infrared spectroscopy and scanning electron microscopy. 2010 Elsevier Ltd. All rights reserved.

Article history: Received 8 June 2010 Received in revised form 27 August 2010 Accepted 31 August 2010 Available online 6 September 2010 Keywords: Nonviable Penicillium YW 01 Congo Red Acid Black 172 Biosorption isotherms Articial neural network

1. Introduction Dyes are synthetic chemical compounds having complex aromatic structures which are extensively used in the textile, cosmetic, plastic, food, and pharmaceutical industries (Forgacs et al., 2004). The dye-containing wastewater discharged from the industries can adversely affect the aquatic environment by impeding light penetration. Moreover, most of the dyes are toxic, carcinogenic and harmful to human health. Even at low concentration (1 mg L1), dyes could be highly noticeable, and could cause an aesthetic pollution and disturbance to the ecosystem and water sources (Vimonses et al., 2010). Therefore, there is an increasing demand of efcient and economical technologies for removing dyes from water environment in the world. Biosorption has been found to be one of the prominent techniques for dye wastewater treatment in terms of cost and operation. Activated carbon is an effective adsorbent for dyes and has been widely used in wastewater treatment. However, this adsorbent has been limited in practice because of high cost and problems with its disposal (Xiong et al., 2010). Therefore, low-cost and effective materials used for dyes removal from large volumes of wastewater have been of great concerns for environmental scientists. Recently, chitin (Dolphen et al., 2007), jute stick powder

Corresponding author. Tel./fax: +86 571 88206995.


E-mail address: yhzhao1958@yahoo.com.cn (Y. Zhao). 0960-8524/$ - see front matter 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2010.08.125

(Panda et al., 2009), cattail root (Hu et al., 2010), various fungi (Arica and Bayramoglu, 2007; Binupriya et al., 2008; Akar et al., 2009), etc., have been investigated for removal of dye from wastewater. Penicillium is one of the most widespread fungi in the terrestrial environment. Penicillium has been widely used for dye removal via biodegradation or biosorption (Iscen et al., 2007; Shedbalkar et al., 2008; Gou et al., 2009). Dried Penicillium restrictum had been recently used for biosorption of Reactive Black 5 (Iscen et al., 2007). In this study, we investigated Penicillium YW 01 for biosorption capacities of AB (C.I. Acid Black 172, chemical formula: C40H20O14N6S2Na2Cr, molecular weight: 970) metal-complex dye and CR (C.I. Direct Red 28, chemical formula: C32H22N6Na2O6S2, molecular weight: 697) under different conditions, the biosorption kinetics for the two dyes and the characteristics of the fungus biomass related to the interaction between the dyes and the biomass, and the effects of the operational parameters on the biosorption capacity was also analyzed by using articial neural network (ANN) and Genetic Algorithms (GAs) hybrid model. AB having Cr (VI) in its structure was chosen in this study because it is one of the anionic metal-complex dyes and widely used in tanning and textile industries in China, as the metal-complex dyes discharged into the environment could cause more serious problem, and CR, a benzidine-based azo dye, was selected in this study as a model anionic dye because of its complex chemical structure, persistence and carcinogenicity, which has been widely used in textiles, paper, rubber and plastic industries.

Y. Yang et al. / Bioresource Technology 102 (2011) 828834

829

2. Methods 2.1. Preparation of biosorbent The Penicillium YW 01 (GenBank accession No. GU944770) used in this work was grown in the potato-dextrose broth (made in laboratory according to descriptions for ATCC media number 336 without incorporating agar) at pH 5.5, 30 C and shaken at 200 rpm for 5 days. The biomass was separated from the culture broth by ltration, and washed with generous amounts of distilled water. Then, the pretreated biomass was autoclaved at 121 C for 20 min and dried overnight at 50 C. The dried biomass was ground to power in a disintegrator and sieved through a No. 60 standard sieve to obtain uniform size for biosorption studies. The volume weight and the bulk porosity of the biomass powder were 0.32 g cm3 and 76.38%, respectively. 2.2. Preparation of dye solution and determination of dye concentrations The dyes AB and CR were obtained from Shanghai Sangon Biological Engineering Technology & Services Co., Ltd, China. Stock solutions (1000 mg L1) of dyes were prepared in deionized and double distilled water and diluted to get the desired concentration of dyes. Calibration curves for dyes were prepared by measuring the absorbance of different concentrations of the dyes. 570 and 497 nm were used to measure the absorbance of CR at pH 1.0 3.0 and pH 4.010.0, respectively. The AB was measured at 597 nm to determine the concentration in the solution. 2.3. Biosorption studies Biosorption experiments were carried out with 100 ml dye solution of desired concentration mixing 0.1 g adsorbent in a 250 ml Erlenmeyer ask. The mixture was agitated (200 rpm) at 30 C for 6 h unless otherwise stated. The inuence of hydrogen ion concentration on the biosorption process was studied over a pH range of 1.010.0, with adjustments being made using 0.1 mol L1 HCl or 0.1 mol L1 NaOH. The effect of dye concentration was studied in the range from 50 to 800 mg L1 at pH 3.0. The effect of temperature on the biosorption capacity of each adsorbent was investigated in the temperature range from 20 to 40 C at pH 3.0. The biosorption capacity, Qe (mg g1), was calculated as follows:

where, Co and Ce are the initial and nal concentrations (mg L1), respectively, M is the adsorbent dosage (g) and V the volume of solution (L). All the experiments were carried out at least three times and the data were analyzed by SigmaPlot software (Version 10.0, USA). MATLAB (Version R2009a, USA) software was used to model the biosorption process of dyes onto fungal biomass by articial neural network. 2.4. Characterization of the adsorbent Surface morphology of the biosorbent was determined using hanging drop method by SEM at 20 kV and 6000 magnication. FTIR spectra of virgin and dye-loaded biosorbent were recorded by PerkinElmer spectrum spectrophotometer in the region of 4004000 cm1. 3. Results and discussions 3.1. Effect of pH on biosorption capacity As shown in Fig. 1a, the biosorption capacity of Penicillium YW 01 increased from pH 1.0 to pH 3.0, and reached maximum at pH 3.0 (46.95 and 48.83 mg g1 for AB and CR, respectively), and then declined sharply with further increase in pH for both of the two dyes, indicating that the optimal pH for biosorption of Penicillium YW 01 is 3.0 for both of the two dyes under the experimental conditions. The change pattern of the biosorption capacity with pH could be associated with the effects of pH on both the activity of functional groups in the biosorbents and the chemical properties of dyes. Acidic conditions could be favorable for the biosorption between the two dyes and the fungal biomass, because a signicantly high electro-static attraction could exist between the positively charged surface of the adsorbent under acidic conditions and the anionic dyes (AB and CR are anionic dyes in solution for -SO3 group in their structure). At pH 10.0, the biosorption values were 22.56 and 21.67 mg g1 for AB and CR, which were 44% and 48% of maximum values, respectively. The low biosorption capacity under alkaline conditions could be mainly attributed to that the increasing number of negative charge on the surface of the fungal surface could result in electrostatic repulsion between the adsorbent and dye molecules (Aksu and Donmez, 2003) and that the existence of excess OH- ions may compete with the anionic dyes for the decreasing number of positively charged sites on the fungal surface as the pH increased. A similar trend was observed for the biosorp-

Qe

C o C e V M

Fig. 1. Effect of initial pH (a) and initial dye concentration and temperature (b) on biosorption capacity of the Penicillium YW 01 biomass for AB and CR.

830

Y. Yang et al. / Bioresource Technology 102 (2011) 828834

Table 1 Adsorption capacities of Congo red and other dyes onto various adsorbent reported. Adsorbent Cattail root Chitosan hydrogel beads impregnated with carbon nanotubes Glycidyl Methacrylate-g-Poly (ethylene terephthalate) Fiber Modied saccharomyces cerevisiae subsp. uvarum Trametes versicolor Aspergillus niger Penicillium restrictum Penicillium YW 01 Dyes Congo Red Congo Red Congo Red Congo Red Congo Red Congo Red Reactive Black 5 Congo Red Acid Black 172 pH 7.0 5.0 2.0 4.6 2.0 6.0 1.0 3.0 3.0 Adsorption capacity (mg g1) 38.79 450.40 16.60 93.1 51.81 14.16 142.04 411.53 225.38 References (Hu et al., 2010) (Chatterjee et al., 2010) (Arslan, 2010) (Safarikova et al., 2005) (Binupriya et al., 2008) (Fu and Viraraghavan, 2002) (Iscen et al., 2007) Present work

Fig. 2. Effect of contact time on the biosorption of AB (a) and CR (b) onto the Penicillium YW 01 biomass with different initial dye concentration.

tion of CR on Trametes versicolor (Binupriya et al., 2008) and coir pith carbon (Namasivayam and Kavitha, 2002). Hence, acidic conditions were more favorable for the application of biosorption process of anionic dyes onto fungal biomass. 3.2. Effect of dye concentration and temperature on biosorption capacity The effect of dye concentration and temperature on biosorption process was shown in Fig. 1b. The biosorption capacity for AB increased rapidly with initial dye concentration from 100 to 200 mg L1, and tended to keep stable when C0 was above 200 mg L1 under all the three temperature conditions. However, a linear increase in the biosorption capacity for CR was observed as the C0 increased up to 800 mg L1 under all the three temperature conditions. The biosorption capacity tended to increase with temperatures from 20 to 40 C for all the initial concentrations of the two dyes. The effects of the temperature could be attributed to the decrease in the viscosity of the solution with the temperature increasing. Compared with the adsorption capacity of other adsorbents for CR and other dyes (Table 1), the biosorption capacity of CR by Penicillium YW 01 was the highest among the fungus adsorbent, although lower than chitosan hydrogel beads impregnated with carbon nanotubes (Chatterjee et al., 2010), indicating that Penicillium YW 01 was an efcient adsorbent for CR removal from the solution. 3.3. Biosorption kinetics

pesudo-rst-order Lagergen (Lagergren, 1898), the pesudosecond-order (Ho and McKay, 1998) and Weber-Morris (Weber and Morris, 1963) models have been tested to describe the biosorption kinetics. The pesudo-second-order and Weber-Morris model performed well in describing the biosorption kinetic for the dye-biosorption although the pesudo-rst-order model of biosorption of AB and CR onto biomass, which had yielded a determined coefcient (R2) less than 0.90, could not be used to well describe the biosorption kinetics.The equation of pesudo-secondorder kinetic model was expressed as follows:

t 1 1 t Q t K 2 Q 2 max Q max

where k2 is the equilibrium rate constant of pseudo-second-order biosorption (g mg1 min1), and Qmax is the maximum biosorption capacity (mg g1) for the pseudo-second-order biosorption. According to the R2 in Table 2, the pseudo-second-order model showed satisfactory ts (R2 > 0.99 for all the tested conditions). The Qmax values estimated from the pesudo-second-order kinetic model were also well accord with the experimental data (Qeq,exp values in
Table 2 Parameters of pseudo-second-order kinetic model for adsorption of AB and CR onto Penicillium YW01 biomass. Dye C0 (mg L1) 50 100 50 100 Qeq,exp (mg g1) 49.10 97.88 49.16 98.16 Pseudo-second-order kinetic model k2 102(g mg1min1) 2.38 0.37 2.07 0.41 Qe,cal (mg g1) 49.01 98.04 49.26 99.01 R2 1.000 0.999 1.000 0.999

AB

From Fig. 2, it can be observed that the biosorption rate of the two dyes at all the tested initial dye concentrations was initially fast and then gradually decreased to reach equilibrium. The

CR

Y. Yang et al. / Bioresource Technology 102 (2011) 828834

831

Fig. 3. Weber-Morris plots for biosorption of AB and CR onto the Penicillium YW 01 biomass at 50 mg L1 (a) and 100 mg L1 (b).

Table 2) at all the dye concentrations of AB and CR. The results suggested the boundary layer resistance was not the rate limiting step since the dye-biosorption follows pseudo-second order kinetics (Xiong et al., 2010). The Weber-Morris model could be used to investigate the mass transfer mechanism in the dye-fungus system and is expressed as

Q t K w t 1 =2 I
1/2

where Kw is the intraparticle rate constant, t is the square root of time, Qt is the amount of adsorbed dye per unit weight of adsorbent at time t (mg g1) and I is the value of intercept. The intercept gives an idea about the thickness of the boundary layer, i.e. the larger is the intercept, the greater is the boundary layer effect (zer et al., 2006). The Weber-Morris plots were presented in Fig. 3 and the parameters were summarized in Table 3. The Weber-Morris plots for biosorption of AB and CR at 50 mg L1 were linear and did not pass the origin (Fig. 3a), indicating the signicance of intraparticle diffusion existed in the biosorption of dyes onto the fungal biomass (zer et al., 2006). Higher intercepts indicated that the boundary layer effect could not be ignored for biosorption of AB and CR at initial concentration of 50 mg L1 (Table 3). Fig. 3b showed that the intraparticle diffusion of AB

and CR within fungus biomass at initial concentration of 100 mg L1 occurred in two stages. The rst straight portion could be attributed to macropore diffusion (stage I), i.e. transport of dye molecules from bulk solution to the surface of the adsorbent, and the second linear portion could be attributed to micropore diffusion (stage II), i.e. the binding of the dye molecules on the active sites of biosorption. Since the kw,2 values were smaller than the kw,1 values (Table 3), the intraparticle diffusion should be the limiting step for the biosorption of AB and CR onto fungal biomass. Similar trend was observed with biosorption of CR and rhodamine B onto jute stick power (Panda et al., 2009). All the intercepts calculated at 100 mg L1 were higher than that of 50 mg L1 (Table 3), indicating that the boundary layer effect had more inuence on the biosorption process at higher concentrations. 3.4. Biosorption isotherms To investigate the biosorption mechanisms, the surface properties and the capacity or afnity of nonviable Penicillium YW 01 for AB and CR, the Langmuir, Freundlich and Dubinin-Radushkevich adsorption isotherms were selected to explicate the dye-fungus system in this study. 3.4.1. Langmuir isotherm The Langmuir theory assumes a homogeneous type of adsorption. That is, once a dye molecule occupies a binding site, no further adsorption can occur at that site (Langmuir, 1918). The linearized equation is given as

Table 3 Parameters of Weber-Morris model for adsorption of AB and CR onto fungal biomass. Dye Concentration (mg L1) 50 100 50 100 Initial linear portion Kw,1 0.1113 3.9091 0.1319 3.6818 I1 47.23 62.214 46.97 65.593 R2 0.887 0.905 0.962 0.936 Second linear portion Kw,2 0.1349 0.0994 I2 95.477 92.716 R2 0.829 0.949

AB CR

Ce 1 Ce Q max Omax K L Q max

where Ce is the equilibrium dye concentration in the solution (mg L1), Qe is the equilibrium dye uptake on the biosorbent

Table 4 Biosorption isotherm parameters for the adsorption of AB and CR onto Penicillium YW 01 at various temperatures. Dye T (C) Qexp (mg g1) Langmuir constants Qmax (mg g AB 20 30 40 20 30 40 182.31 220.26 225.38 355.43 383.09 411.53 185.19 222.22 227.27 357.14 384.62 416.67
1

Freundlich constants
1

Dubinin-Radushkevich (D-R) constants R


2

Kl (L mg 0.057 0.058 0.060 0.042 0.046 0.054

n 9.93 7.78 7.02 4.94 5.14 4.30

KF (L g

Qmax (mg g1) 162.40 192.48 200.34 281.46 304.91 340.35

b (mol2 KJ2) 0.0004 0.0006 0.0008 0.001 0.0008 0.0011

R2 0.868 0.903 0.946 0.899 0.904 0.918

E (kJ mol1) 35.35 28.87 25.00 22.36 25.00 21.32

CR

0.997 0.995 0.993 0.977 0.975 0.989

93.69 97.51 92.76 100.48 112.16 104.59

0.997 0.986 0.911 0.918 0.883 0.901

832

Y. Yang et al. / Bioresource Technology 102 (2011) 828834

(mg g1), Qmax is the maximum biosorption capacity (mg g1), and KL is the Langmuir constant (L mg1). The Langmuir model has a high R2 values (Table 4), indicating the formation of a monolayer of AB and CR covering the Penicillium YW 01 surface. Qmax and KL reached maximum values of 227.27 mg g1 and 0.060 L mg1 at 40 C for AB, respectively (Table 4). Qmax and KL reached maximum values of 416.67 mg g1 and 0.054 L mg1 at 40 C for CR, respectively (Table 4). These values indicated that the dye molecules exhibited the highest afnity for the adsorbent at 40 C and agreed with the experimental results (Qexp values in Table 4). The suitability of the adsorbent for the sorbate can be expressed using the Hall separation factor (RL, dimensionless), which can be calculated in the following equation (Hall et al., 1966)

(Table 4), which is not within the energy range of ion-exchange reactions, 816 kJ mol1. This implied that physical adsorption may be one of the mechanisms for the biosorption for AB and CR. Similar results were observed for the adsorption of uranium onto amberlite IR-118 resin (Kilislioglu and Bilgin, 2003). 3.5. Sample characterization It can be found from the SEM image in Fig. 4 that the surface of the fungus biomass was heterogeneous, smooth and porous. The porous characteristics of the biomass with a bulk porosity of 76.38% indicated the biomass has a large surface area for dye interaction and a great capacity for dye holding. The band positions in the FTIR spectra of the biosorbent before and after AB and CR biosorption were presented in Table 5. The bands at 3415 and 3418 observed after biosorption were corresponded to the stretching vibration of NH in the structure of CR (Wang and Wang, 2008) and OH formation in the structure of AB, respectively, attributing to the biosorption of AB and CR onto biomass. The strong biosorption bands of OH and/or NH groups were observed at 3275 cm1 for virgin biosorbent, but this biosorption peak could not be observed for the dye-loaded biomass. The results, with acid condition enhancing the biosorption, implicated that NH groups may be responsible for the dye-biosorption in the biomass. The changes observed in the band 2920, 2925 and 2851 cm1 indicated that ion-exchange between protons of symmetric or asymmetric CH and the symmetric stretching vibration of CH2, respectively, of aliphatic acids (Wahab et al., 2010). These changes observed between 1660 and 1500 cm1 might be due to the electrostatic forces of attraction between the negative charge of carboxylate anion and positive group of dyes (Farinella et al.,

RL

1 1 K LC0

where KL and C0 are the Langmuir constant (L mg1) and initial dye concentration (mg L1), respectively. All the RL values (data were not shown) for AB and CR were in the range 01, indicating that the biosorption process was favorable. 3.4.2. Freundlich isotherm The Freundlich isotherm assumes a heterogeneous surface with a nonuniform distribution of heat of adsorption (Freundlich, 1906). The equation of Freundlich isotherm is given as below:

1 InQ e InK F lnC e n

where KF (L g1) and n (dimensionless) are characteristic constants that indicate the extent of the biosorption, the degree of nonlinearity between solution concentration and biosorption, respectively. The values of Freundlich constant n for AB and CR (Table 4) were all in range 210, indicating that good biosorption occurred. In addition, n values were greater than unity, indicating that the dyes were favorably adsorbed by the fungus biomass. The high values of KF (Table 4) for AB and CR implied ready uptake of the dyes from the solution with high adsorptive capacities of these biosorbents. Similar trend were observed for biosorption of Remazol Blue reactive dye onto dried yeasts (Aksu and Donmez, 2003), Bromophenol Blue Dye onto Phizopus stolonifer, Fusarium sp. Geotrichum sp. and Aspergillus fumigatus (Zeroual et al., 2006). 3.4.3. Dubinin-Radushkevich isotherm The Dubinin-Radushkevich (D-R) isotherm model is more general than Langmuir isotherm, because it does not assume a homogeneous surface or constant sorption potential (Dubinin and Radushkevich, 1947). The D-R model is postulated within adsorption space close to the adsorbent surface (Akar et al., 2009), which was applied to distinguish the nature of biosorption as physical or chemical. The model can be linearized as follows:

Fig. 4. SEM image of the Penicillium YW 01 biomass.

InQ e InQ max be2

Table 5 The FTIR spectral characteristics of nonviable Penicillium YW 01 biomass before and after biosorption of AB and CR. Suggested assignment Band positions (cm1) Unloaded biomass -OH and/or NH stretching -OH and/or NH stretching -CH symmetric stretching -CH asymmetric stretching Amid-I Amid-II -CH bending vibrations -CH bending vibrations -SO3 stretching -C-O stretching -CH bending vibrations (aromatic) 3275 2920 2851 1649 1541 1456 1398 1260 1027 931692 AB-loaded biomass 3415 2925 1660 1549 1410 1378 1230 1038 CR-loaded biomass 3418 2925 1633 1548 1411 1378 1041

where e is Polanyi potential, Qmax is the theoretical saturation capacity of biomass (mg g1), b is a constant related to the energy of transfer of the solute from bulk solution to solid adsorbent. The constant b gives us the information about the mean energy of biosorption (E), which can be calculated from the following equation (Akar et al., 2009):

1 2b1=2

The magnitude of E value is useful for estimating the type of biosorption process. It was found the estimated values of E for AB and CR were in the range between 21.32 and 35.35 kJ mol1

Y. Yang et al. / Bioresource Technology 102 (2011) 828834

833

2007). The band at 1260 cm1 observed for the unloaded biomass, indicating the existence of SO3 stretching in the biosorbent. This band disappeared after CR-loaded biomass, responding for the biosorption of CR onto biomass. Similar results were observed for CR onto surfactant-modied montmorillonite (Wang and Wang, 2008). This bond slightly shifted to a lower frequency (1230 cm1) after AB biosorption. Bands at 931692 cm1, assigned to the biosorption characteristics of aromatic skeletal groups, have been diminished after biosorption. The changes observed in the FTIR spectrum between the unloaded and the loaded biomass indicated that functional groups including amine, amide, SO3 and carboxyl groups could be responsible for the biosorption of AB and CR onto biosorbent. 3.6. ANN modeling Articial neural networks (ANN) have been used in many literatures to investigate the relative importance of parameters during adsorption or degradation (Aleboyeh et al., 2008; Aber et al., 2009; Khataee et al., 2010), and have also been employed in our research. In this study, 129 experimental sets of each dye were used to feed the ANN structure. The range of variables studied is summarized in Table 6. The data sets were divided into training, validation and test subsets, each of that included 77, 26, and 26 samples, respectively. All samples were normalized in the range 0.10.9 using the following equation:

gation neural network ANN, which was calculated by the following equation:

MSE 1=N

iN X yi; pred yi; exp i1

10

Ai 0:8

X i minX i 0:1 maxX i minX i

where min(Xi) and max(Xi) are the extreme values of variable Xi (Khataee et al., 2010). Firstly, we used mean square error (MSE) to determine the structure of three-layered feed forward back propa-

Table 6 Model variables and their ranges. Variables Input layer Temperature (C) pH Initial dye concentration (mg L1) Time (min) Output layer Adsorption capacity (mg g1) Range 2040 110 50800 5360 21.45225.38 (AB) 22.42411.53 (CR)

where yi,pred and yi,exp are the values predicated by the neural network and obtained by experiments, respectively. N is the number of data point and i is an index of data. A series of topologies, in which the number of nodes was varied from 3 to 10, were used to determine the optimum number of hidden nodes. Each topology was repeated three times to avoid random correction due to random initialization of the weights. Then, the Genetic Algorithms (GAs), which mimic biological evolution and progressively seek superior solutions using the guide of selective function (tness) and partial combination of candidate solutions (Ooba et al., 2006), were applied to optimize the weights and bias for the selected structures of ANN, and the optimized model was noted as ANNGA hybrid model. Fig. 5a showed the network error versus the number of neurons in the hidden layer. It could be found that neural network performance of 5 and 7 nodes on the hidden layer were smallest during the tested topologies and could be used to model the biosorption process of AB and CR, respectively. However, the neural network performance of 5 and 7 nodes on the hidden layer using ANN-GA hybrid model were 1.12 104 and 0.10 104, respectively, which were better than that of the ANN model (1.54 104and 0.17 104, respectively). Hence, the ANN-GA model was more reasonable that just ANN to get the net weights, which was not considered in many reported literatures(Aleboyeh et al., 2008; Aber et al., 2009; Khataee et al., 2010). Determination coefcients (R2) of 0.9979 and 0.9996 were obtained between the experimental data and the predicted results using ANN-GA hybrid model for AB and CR, respectively (Fig. 5b).The network weights give information about relative importance of input variables and are calculated as following:

  W ih  jm W ho PkNi ih mn m1 W km j j k1  Ij    W ih PkNi PmNh PNi jm ih W ho mn m1 k 1 W km j j k1 PmNh

11

where Ij is the relative importance of the jth input variable on the output variable, Ni and Nh are the numbers of input and hidden neurons, respectively, W is connection weight, the superscripts i, h and o refer to input, hidden and output layers, respectively, and subscripts k, m and n refer to input, hidden and output neuron

Fig. 5. The performance of ANN model (a) and comparison between the experimental values and the predicted values using ANN-GA model (b).

834

Y. Yang et al. / Bioresource Technology 102 (2011) 828834 Dolphen, R., Sakkayawong, N., Thiravetyan, P., Nakbanpote, W., 2007. Adsorption of reactive red 141 from wastewater onto modied chitin. J. Hazard. Mater. 145, 250255. Dubinin, M.M., Radushkevich, L.V., 1947. Proc. Acad. Sci. U.S.S.R. Phys. Chem. Sect. 55, 331333. Farinella, N.V., Matos, G.D., Arruda, M.A.Z., 2007. Grape bagasse as a potential biosorbent of metals in efuent treatments. Bioresour. Technol. 98, 19401946. Forgacs, E., Cserhati, T., Oros, G., 2004. Removal of synthetic dyes from wastewaters: a review. Environ. Int. 30, 953971. Freundlich, H.M.F., 1906. ber die adsorption in lsungen. Z. Phys. Chem. 57, 385470. Fu, Y.Z., Viraraghavan, T., 2002. Removal of Congo Red from an aqueous solution by fungus Aspergillus niger. Adv. Environ. Res. 7, 239247. Gou, M., Qu, Y., Zhou, J., Ma, F., Tan, L., 2009. Azo dye decolorization by a new fungal isolate, Penicillium sp. QQ and fungal-bacterial cocultures. J. Hazard. Mater. 170, 314319. Hall, K.R., Eagleton, L.C., Acrivos, A., Vermeule, T., 1966. Pore- and solid-diffusion kinetics in xed-bed adsorption under constant-pattern conditions. Ind. Eng. Chem. Fund 5, 212223. Ho, Y.S., McKay, G., 1998. Kinetic models for the sorption of dye from aqueous solution by wood. Process Saf. Environ. Prot. 76, 183191. Hu, Z.H., Chen, H., Ji, F., Yuan, S.J., 2010. Removal of Congo Red from aqueous solution by cattail root. J. Hazard. Mater. 173, 292297. Iscen, C.F., Kiran, I., Ilhan, S., 2007. Biosorption of Reactive Black 5 dye by Penicillium restrictum: the kinetic study. J. Hazard. Mater. 143, 335340. Khataee, A.R., Dehghan, G., Ebadi, A., Zarei, M., Pourhassan, M., 2010. Biological treatment of a dye solution by Macroalgae Chara sp.: Effect of operational parameters, intermediates identication and articial neural network modeling. Bioresour. Technol. 101, 22522258. Kilislioglu, A., Bilgin, B., 2003. Thermodynamic and kinetic investigations of uranium adsorption on amberlite IR-118H resin. Appl. Radiat. Isot. 58, 155160. Lagergren, S., 1898. Zur theorie der sogenannten adsorption gelster stoffe. Handlingar 24, 139. Langmuir, I., 1918. The adsorption of gases on plane surfaces of glass, mica and platinum. J. Am. Chem. Soc. 40, 13611403. Ooba, M., Hirano, T., Mogami, J.-I., Hirata, R., Fujinuma, Y., 2006. Comparisons of gap-lling methods for carbon ux dataset: a combination of a genetic algorithm and an articial neural network. Ecol. Model. 198, 473486. zer, A., Akkaya, G., Turabik, M., 2006. Biosorption of Acid Blue 290 (AB 290) and Acid Blue 324 (AB 324) dyes on Spirogyra rhizopus. J. Hazard. Mater. 135, 355364. Panda, G.C., Das, S.K., Guha, A.K., 2009. Jute stick powder as a potential biomass for the removal of congo red and rhodamine B from their aqueous solution. J. Hazard. Mater. 164, 374379. Safarikova, M., Ptackova, L., Kibrikova, I., Safarik, I., 2005. Biosorption of watersoluble dyes on magnetically modied Saccharomyces cerevisiae subsp. uvarum cells. Chemosphere 59, 831835. Shedbalkar, U., Dhanve, R., Jadhav, J., 2008. Biodegradation of triphenylmethane dye cotton blue by Penicillium ochrochloron MTCC 517. J. Hazard. Mater. 157, 472479. Vimonses, V., Jin, B., Chow, C.W.K., 2010. Insight into removal kinetic and mechanisms of anionic dye by calcined clay materials and lime. J. Hazard. Mater. 177, 420427. Wahab, M.A., Jellali, S., Jedidi, N., 2010. Ammonium biosorption onto sawdust: FTIR analysis, kinetics and adsorption isotherms modeling. Bioresour. Technol. 101, 50705075. Wang, L., Wang, A., 2008. Adsorption properties of Congo Red from aqueous solution onto surfactant-modied montmorillonite. J. Hazard. Mater. 160, 173180. Weber, W.J., Morris, J.C., 1963. Equilibrium and capacities for adsorption on carbon. J. Sanit. Eng. Div. ASCE 89, 31. Xiong, X.J., Meng, X.J., Zheng, T.L., 2010. Biosorption of C.I. Direct Blue 199 from aqueous solution by nonviable Aspergillus niger. J. Hazard. Mater. 175, 241 246. Zeroual, Y., Kim, B.S., Kim, C.S., Blaghen, M., Lee, K.M., 2006. Biosorption of bromophenol blue from aqueous solutions by Rhizopus stolonifer biomass. Water Air Soil Pollut. 177, 135146.

numbers, respectively. The initial concentration of AB with a relative importance of 41.43% appeared to be the most inuential parameter in the biosorption process for AB, followed by pH (23.45%), temperature (19.17%) and time (15.95%). Temperature with a relative importance of 36.76% appeared to be the most inuential parameter in the biosorption process for CR, followed by pH (29.83%), time (24.30%) and initial concentration (9.11%). The results also indicated that all of the variables have strong effects on the biosorption capacity for AB and CR. 4. Conclusions The inactive Penicillium YW 01 biomass was efcient as an adsorbent for the removal of AB and CR form aqueous solutions. The maximum biosorption capacity of CR was 411.53 mg g1, which was best among the reported fungi. Kinetic study revealed that intraparticle diffusion was the limiting step for biosorption of AB and CR. The performance of ANN-GA appeared to be better than that of ANN to get the net weights which were used to assess the relative importance of operational parameters. The dye-biosorbent interactions were conrmed by FTIR and heterogeneous, smooth and porous structure were observed by SEM technique. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.biortech.2010.08.125. References
Aber, S., Amani-Ghadim, A.R., Mirzajani, V., 2009. Removal of Cr(VI) from polluted solutions by electrocoagulation: modeling of experimental results using articial neural network. J. Hazard. Mater. 171, 484490. Akar, T., Tosun, I., Kaynak, Z., Kavas, E., Incirkus, G., Akar, S.T., 2009. Assessment of the biosorption characteristics of a macro-fungus for the decolorization of acid red 44 (AR44) dye. J. Hazard. Mater. 171, 865871. Aksu, Z., Donmez, G., 2003. A comparative study on the biosorption characteristics of some yeasts for remazol blue reactive dye. Chemosphere 50, 10751083. Aleboyeh, A., Kasiri, M.B., Olya, M.E., Aleboyeh, H., 2008. Prediction of azo dye decolorization by UV/H2O2 using articial neural networks. Dyes Pigments 77, 288294. Arica, M.Y., Bayramoglu, G., 2007. Biosorption of Reactive Red-120 dye from aqueous solution by native and modied fungus biomass preparations of Lentinus sajor-caju. J. Hazard. Mater. 149, 499507. Arslan, M., 2010. Use of 1, 6-Diaminohexane-functionalized Glycidyl Methacrylateg-Poly(ethylene terephthalate) ber for removal of acidic dye from aqueous solution. Fiber. Polym. 11, 177184. Binupriya, A.R., Sathishkumar, M., Swaminathan, K., Ku, C.S., Yun, S.E., 2008. Comparative studies on removal of Congo red by native and modied mycelial pellets of Trametes versicolor in various reactor modes. Bioresour. Technol. 99, 10801088. Namasivayam, C., Kavitha, D., 2002. Removal of congo red from water by adsorption onto activated carbon prepared from coir pith, and agricultural solid waste. Dyes Pigments 54, 4748. Chatterjee, S., Chatterjee, T., Woo, S.H., 2010. A new type of chitosan hydrogel sorbent generated by anionic surfactant gelation. Bioresour. Technol. 101, 38533858.

You might also like