4732
Ind. Eng. Chem. Res. 2006, 45, 4732-4741
SEPARATIONS Adsorption of Reactive Dyes from a Textile Effluent Using Sawdust as the Adsorbent Sourja Chakraborty, Jayanta K. Basu, Sirshendu De, and Sunando DasGupta* Department of Chemical Engineering, Indian Institute of Technology, Kharagpur-721302, India
Adsorption is carried out for the removal of two reactive dyes from an industrial effluent. The specific dyes are reactive red and reactive black, commercially known as Cibacron Red RB and Cibacron Black B, respectively. An adsorbent developed from sawdust is used. Equilibrium and kinetic studies are carried out with the synthetic solutions of the dyes. For the mixture of two dyes, the bisolute Langmuir isotherm modified with an interaction factor is used. A mass transfer model including an external film mass transfer coefficient and an internal effective diffusivity is used to interpret the adsorption kinetic data. These parameters are obtained by fitting the experimental data to the model. An industrial effluent is successfully treated with the same adsorbent. The estimated model parameters are used for the prediction of the concentration profiles of dyes of the industrial effluent. Introduction Reactive dyes contribute to a major portion of the dyes used in the textile industry. The reactive groups of these dyes react with the ionized hydroxyl groups on the cellulosic fiber. As a result of the alkaline dyeing conditions, hydroxyl ions present in the dye bath compete with the cellulose substrate. As a result, an amount of hydrolyzed dyes are generated which can no longer react with the fiber.1 Thus depending on the degree of fixation, a highly colored effluent is generated which requires treatment before release. A number of inorganic and organic adsorbents are used for the removal of reactive dyes. This includes the use of the synthetic clays,2 chitosan,3,4 calcined alunite,5 activated carbon,1 and so forth. A number of biosorbents are applied for this purpose, with varying success for color removal. Robinson et al.6,7 used several low cost, renewable biosorbents such as apple pomace, wheat straw, and barley husk for the removal of five reactive dyes. It was observed that barley husk could be used as an efficient adsorbent for the removal of reactive textile dyes. In the present work, an adsorbent, developed from sawdust, is used for the removal of two reactive dyes, namely, Cibacron Black B (dye 1) and Cibacron Red RB (dye 2), which are used in the dyeing of 100% cotton knit wear in a winch machine at a hosiery dye house, Singhal Brothers’, located in Kolkata, India. Aqueous synthetic solutions of the dyes in both the singlecomponent systems and a mixture of the two dyes are used for the adsorption equilibrium and rate study. The Langmuir, Freundlich, Redlich-Peterson, and Fritz-Schlunder isotherm equations are used to fit the equilibrium data of the singlecomponent systems. For the dye mixture, an extended Langmuir isotherm equation modified with interaction factors8 is used. A two-resistance mass transfer model9 that includes an external film mass transfer coefficient and an internal effective diffusivity is used to predict the concentration profiles in batch adsorption * To whom correspondence should be addressed. E-mail: sunando@ che.iitkgp.ernet.in. Tel.: +91-3222-283922. Fax: +91-3222-255303.
Chart 1. Chemical Structures of Reactive Black and Reactive Red
of the two-component system. Experiments are also performed with the original industrial effluent stream using varying amounts of the adsorbent. The estimated model parameters are used to predict the concentration profiles during adsorption of dyes in the effluent. Sensitivity analysis is performed to observe the sensitivity of the model parameters. Materials and Methods Characterization of the Textile Effluent. The textile effluent is characterized in terms of dye concentrations, chemical oxygen demand (COD), total solid content (TS), conductivity, pH, and salt content and is presented in Table 1. In the subsequent sections, the two dyes (reactive black and reactive red) will be
10.1021/ie050302f CCC: $33.50 © 2006 American Chemical Society Published on Web 05/20/2006
Ind. Eng. Chem. Res., Vol. 45, No. 13, 2006 4733 Table 1. Characterization of the Effluent from the Textile Plant C0,1 (mg L-1)
C0,2 (mg L-1)
pH
conductivity ×10-3 (mΩ cm-1)
TS (mg L-1)
equivalent salt content (equivalent NaCl; g L-1)
COD (mg L-1)
173
118
6.86
116.3
1.05 × 10 5
56
6312
Table 2. Physical Properties of the Adsorbent and Activated Carbon property m-3)
solid density (kg moisture content (%) ash content (%) BET surface area (m2 g-1 of adsorbent)
CSD
granular activated carbon
1056 1.80 14.68 559
440 2.0 5.8 1100
denoted as dyes 1 and 2, respectively. The dyes are supplied by Ciba, India. Chemical structures of the dyes are given in Chart 1). The concentrations of the dyes in the aqueous solution are measured in a Genesys 2 spectrophotometer (Thermo Spectronic, U.S.A.) at wavelengths of 599 and 535 nm for dyes 1 and 2, respectively. For the dye mixture, the method by Vogel10 is used for the determination of the concentration of each dye in the solution. The COD and TS are measured by the standard method.11 Conductivity of the effluent is measured by an autoranging conductivitymeter Chemito 130, manufactured by Toshniwal Instruments, Ltd. (India). The solution pH is measured by a pH-meter, supplied by Toshniwal Instruments, Ltd. (India). The salt content of the effluent is determined by measuring the conductivity of the effluent stream and comparing it with the conductivity of the dye solution containing 1 equiv of sodium chloride. Various properties of the effluent are presented in Table 1. Preparation of the Adsorbent. The adsorbent is prepared from hardwood sawdust collected from a local sawmill. The material is washed thoroughly with water and completely dried in an oven. The sawdust is then treated with phosphoric acid (supplied by E-Merck, India) and charred in a muffle furnace at different temperatures, for example, 400, 500, and 600 °C. It is observed that the adsorbent charred at 500 °C shows the best adsorbing capacity for the dye. Hence, the temperature of charring is selected at 500 °C. This is continued for about 40 min to ensure complete charring. The charred carbonaceous material thus prepared is then cooled and washed with dilute ammonia solution and water successively to make it completely acid free. It is dried in an oven at 120 °C for 8 h.12,13 Different properties of the adsorbent (charred sawdust, or CSD) are determined by standard methods. Surface area of the adsorbents is measured by the BET method, and the values are reported in Table 2. A commercial granular activated carbon GAC-1240, supplied by NORIT, The Netherlands, is used for comparison of the performance of the prepared adsorbent. Table 2 also includes the properties of the commercial carbon (GAC) as well. Equilibrium Study. The adsorption equilibrium study is carried out with the synthetic solutions of the two dyes in both the single-component systems and the mixture (1:1) of the two dyes. Because acidic pH is favorable for adsorption of both the dyes, the pH of the solutions is adjusted to 2.0. The dye solutions of varying concentrations ranging from 50 to 250 mg L-1, with a volume of 0.2 L, are taken in stoppered conical flasks. About 0.1 g of adsorbent of particle size 0.044 mm is added in each solution. These solutions are equilibrated for a period of 8 h in a temperature controlled mechanical shaker. The temperature is maintained constant (298 K) with a variation of (1 K. The
equilibrium adsorption capacity Ye (mg g-1) for a single component is computed as follows:
Ye )
(C0 - Ce)V Ma
Yei )
(C0i - Cei)V Ma
For the dye mixture,
where, i denotes the number of the component, C0 and Ce are the initial and equilibrium dye concentrations (mg L-1), respectively, V is the volume of the solution (L), and Ma is the mass of the adsorbent used (g). Batch Kinetic Study. An agitated baffled vessel (internal diameter, 0.125 m; height, 0.26 m; number of baffles, 4; baffle clearance, 0.005 m; capacity, 1 L) made of Perspex and equipped with an impeller (diameter, 0.039 m) is used for this study. The temperature is maintained constant by a water bath. For singlecomponent systems, the adsorption rate is measured as a function of the solution pH (2.0-8.0), particle size of the adsorbent (0.044-0.226 mm), and initial concentration of the dyes (100300 mg L-1). For dye mixture, various combinations of the initial concentrations of the two dyes are selected. The pH of the solution is adjusted to 2.0, and 1 g of adsorbent is added in 1 L of dye solution of known concentration. All experiments are performed at a high stirrer speed (2500 rpm). Samples are collected at a definite interval of time and analyzed. Some experiments are also performed with the original industrial effluent, using varying amounts of the adsorbent (1.0-2.0 g L-1) at acidic pH (2.0). Theory Equilibrium Isotherms. Adsorption equilibrium data are fitted to a number of isotherms, namely, Langmuir, Freundlich, Redlich-Peterson, and Fritz-Schlunder isotherms, for the singlecomponent system. The Langmuir isotherm14 is given as
Ye )
YsCe YoKoCe ) 1 + KoCe 1 + KoCe
(1)
where Ye is the amount of dye adsorbed (mg g-1), Ys ) YoKo, Yo is the maximum amount of dye adsorbed per unit weight of adsorbent to form a complete monolayer on the surface at the equilibrium dye concentration Ce, and Ko is the Langmuir constant (L mg-1). The Freundlich isotherm,15 used for heterogeneous surface energy terms, is expressed as
Ye ) YfCe1/n
(2)
where Yf indicates the adsorption capacity and 1/n of the adsorption intensity. Redlich and Peterson16 combined the features of the Langmuir and Freundlich isotherms into a general isotherm equation given as
4734
Ind. Eng. Chem. Res., Vol. 45, No. 13, 2006
Ye )
KRCe 1 + KPCe
ν
(3)
where the exponent, ν, lies between 0 and 1. Another useful expression is the Fritz-Schlunder isotherm,17 which consists of four parameters. The isotherm is flexible in nature and generally shows better agreement with the experimental data. The Fritz-Schlunder isotherm is expressed as
Ye )
KF1Ceφ1 1 + KF2Ceφ2
(4)
For the two-component system, an extended Langmuir equation is used which was developed first by Butler and Ockrent.8 This bisolute Langmuir isotherm model assumes that all the adsorption sites are equally available to all adsorbed species and no interaction between the adsorbed species.16 For the two dyes reactive black and reactive red, denoted by 1 and 2, respectively, the extended Langmuir isotherm can be written as
Ye,1 )
Yo,1Ko,1Ce,1 1 + Ko,1Ce,1 + Ko,2Ce,2
Yo,2Ko,2Ce,2 Ye,2 ) 1 + Ko,1Ce,1 + Ko,2Ce,2
( )
Yo,1Ko,1 Ye,1 )
( ) ( ) ( ) ( ) ( )
1 + Ko,1
Ce,1 Ce,2 + Ko,2 η1,1 η2,1
Yo,2Ko,2 Ye,2 )
1 + Ko,1
Ce,2 η2,2
Ce,1 Ce,2 + Ko,2 η1,2 η2,2
N1(t) ) 4πR2kf1(C1t - Ce1t)
(9)
N2(t) ) 4πR2kf2(C2t - Ce2t)
(10)
The diffusion in pore liquid as per Fick’s law can be written as follows:
(5) N1(t) ) (6)
The above equations can be applied without significant error to a combination of various components, if the components are similar in nature and follow the Langmuir isotherm relation. However, the extended Langmuir isotherm sometimes shows deviation,18 when the maximum adsorption capacities (Yo) of the components differ from one another. Ho and Mckay19 modified the bisolute Langmuir isotherm with an interaction factor η and obtained an excellent fit of the adsorption data of Cu(II) and Ni(II) onto peat. Afterward, Mckay and Al-Duri20 used this interaction-factor isotherm model satisfactorily to represent the multicomponent adsorption equilibrium data of dyes on activated carbon. For a two-component system, the model proposed by these authors can be represented as follows.
Ce,1 η1,1
However, even for such systems, the above assumption leads to substantial errors beyond the first few minutes if the agitation is high. Hence, both the resistances are important for kinetic study.21,22 The kinetic model used in the present work is based on the unreacted shrinking core23 and is an extension of the model for the single-component system.9 It is assumed that the pore diffusivity is independent of concentration and that the adsorption isotherm is irreversible. The model is already developed for a two-component system having a Freundlich isotherm for the mixture.24 In the present work, the same model is modified using the mixture isotherms presented by eqs 7 and 8. Considering the spherical adsorbent particles, the following rate equations are written for components 1 and 2, respectively.
N2(t) )
4πDp1Ce1t 1 1 Rf1 R
[
]
4πDp2Ce2t 1 1 Rf2 R
[
]
(11)
(12)
The mass balance on a spherical element of the adsorbent particle can be written as
N1(t) ) -4πRf12Ye1tF N2(t) ) -4πRf22Ye2tF
[ ] [ ] dRf1 dt
(13)
dRf2 dt
(14)
The average concentration on the adsorbent particle can be written as
(7)
(8)
where ηi,j is the interaction factor of component i for the adsorption of component j. This interaction factor is specific to each component in a given system and depends on the other components present. The isotherm models, given by eqs 7 and 8, are used in this study for the two-component mixture. Kinetic Model. In adsorption, mainly two resistances prevail: the external liquid film resistance and the resistance in the adsorbent particles. The intraparticle diffusion resistance may be neglected for solutes that exhibit strong solid to liquid phase equilibrium solute distribution in the initial period of operation.
[ ( )] [ ( )]
Y1t ) Ye1t 1 -
Rf1 R
3
Y2t ) Ye2t 1 -
Rf2 R
3
(15) (16)
The differential mass balance over the system by equating the decrease in the adsorbate concentrations in the solution with the accumulation of the adsorbates in the adsorbent can be written as
N1(t) ) -V N2(t) ) -V
( ) ( ) ( ) ( ) dC1t dY1t ) Ma dt dt
(17)
dC2t dY2t ) Ma dt dt
(18)
The dimensionless terms used for simplification are as follows:
Ind. Eng. Chem. Res., Vol. 45, No. 13, 2006 4735
Cit* )
Cit Rfi kfiR Ma , Ri* ) , Bii ) , Xi ) , C0i R Dpi VC0i Ceit Dp1t , and τ ) 2 Ceit* ) C0i R
(i denotes the ith component). Combining eqs 9 and 11, the following expression is obtained.
Ce1t* )
Bi1(1 - R1*)C1t* R1* + Bi1(1 - R1*)
) g11(C1t*, R1*)
(19)
Similarly, combining eqs 10 and 12, the following expression is obtained.
Ce2t* )
Bi2(1 - R2*)C2t*
) g12(C2t*, R2*)
used, the time derivatives of Ye1,t and Ye2,t are obtained as follows.
dYe,1t dC1t* dC2t* dR1* ) h11A11 - h12A12 - h11B11 + dτ dτ dτ dτ dR2* h12B12 (27) dτ dYe,2t dC1t* dC2t* dR1* ) -h21A11 + h22A12 + h21B11 dτ dτ dτ dτ dR2* h22B12 (28) dτ where
h11 )
Yes,1 η2,1 + Ko,2*Ce,2* η1,1η2,1 [1 + (K *C */η ) + (K *C */η )]2 o,1 e,1 1,1 o,2 e,2 2,1
h12 )
Yes,1 Ko,2*Ce,1* η,11η2,1 [1 + (K *C */η ) + (K *C */η )]2 o,1 e,1 1,1 o,2 e,2 2,2
dR1* 2 dτ [R1* + Bi1(1 - R1*)]
h21 )
Yes,2 η1,2 + Ko,1*Ce,1* η1,2η2,2 [1 + (K *C */η ) + (K *C */η )]2 o,1 e,1 1,2 o,2 e,2 2,2
(21)
h22 )
Yes,2 Ko,1*Ce,2* η2,2η1,2 [1 + (K *C */η ) + (K *C */η )]2 o,1 e,1 1,2 o,2 e,2 2,2
R2* + Bi2(1 - R2*)
(20)
Differentiation of eqs 19 and 20 with respect to τ gives the following expressions for components 1 and 2, respectively.
Bi1(1 - R1*) dC1t* dCe1t* ) dτ [R1* + Bi1(1 - R1*)] dτ Bi1C1t*
dC1t* dR1* - B11 dτ dτ
) A11 and
Combining eqs 9 and 13 and after nondimensionalization, an expression is obtained expressing the rate of shrinkage of the adsorbate particle due to component 1,
Bi2(1 - R2*) dC2t* dCe2t* ) dτ [R2* + Bi2(1 - R2*)] dτ Bi2C2t*
dR2* [R2* + Bi2(1 - R2*)]2 dτ dC2t* dR2* ) A12 - B12 dτ dτ
Yes,1Ce,1*/η1,1 1 + Ko,1*Ce,1*/η1,1 + Ko,2*Ce,2*/η2,1
Yes,2Ce,2*/η2,2 Ye,2t ) 1 + Ko,1*Ce,1*/η1,2 + Ko,2*Ce,2*/η2,2
(23)
(29)
R1*2
Similarly, when eqs 10 and 14 are combined and after nondimensionalization, the rate of shrinkage of the adsorbate particle due to component 2 is obtained,
dR2* ) dτ
-Bi2Dp2
( )
C02 (C * - Ce2t*) FYe,2t 2t Dp1R2*2
(30)
(24)
where Yes,1 ) Ys,1C0,1; Yes,2 ) Ys,2C0,2; Ko,1* ) Ko,1C0,1; and Ko,2* ) Ko,2C0,2. Substituting the expressions of Ce1t* and Ce2t* from eqs 19 and 20 into eqs 23 and 24, Ye1,t and Ye2,t can be expressed in terms of C1t*, C2t*, R1*, and R2*.
Ye1t )
C01 (C * - Ce1t*) FYe,1t 1t
(22)
In terms of nondimensional concentrations, eqs 7 and 8 can be expressed as
Ye,1t )
dR1* ) dτ
( )
-Bi1
Yes,1g11/η1,1 ) S1 1 + Ko,1*g11/η1,1 + Ko,2*g12/η2,1 (C1t*,C2t*, R1*, R2*) (25)
Yes,2g12/η2,2 ) S2 Ye2t ) 1 + Ko,1*g11/η1,2 + Ko,2*g12/η2,2 (C1t*, C2t*, R1*, R2*) (26) When eqs 25 and 26 are differentiated and eqs 19 and 22 are
When eqs 19 and 20 are used for Ce1t* and Ce2t* and eqs 23 and 24 are used for Ye1,t and Ye2,t, eqs 29 and 30 can be expressed as
dR1* ) d11(C1t*, C2t*, R1*, R2*) dτ dR2* ) d22(C1t*, C2t*, R1*, R2*) dτ
(32)
When eqs 15-18 are combined and after nondimensionalization, the following expressions are obtained for components 1 and 2, respectively.
( )
( )
dC1t* dYe,1t dR1* + X1(1 - R1*3) ) 3X1Ye,1tR1*2 dτ dτ dτ
4736
Ind. Eng. Chem. Res., Vol. 45, No. 13, 2006
( )
( )
dC2t* dYe,2t dR2* ) 3X2Ye,2tR2*2 (34) + X2(1 - R2*3) dτ dτ dτ When eqs 27, 28, 31, and 32 are used, the following two governing equations of bulk concentration are obtained.
dC1t* dC2t* - X12 ) Z11 X11 dτ dτ
(35)
dC1t* dC2t* -X21 + X22 ) Z22 dτ dτ
(36)
where
X11 ) 1 + X1(1 - R1*3)h11A11 X12 ) X1(1 - R1*3)h12A12
Figure 1. Equilibrium adsorption of the reactive dyes in single-component systems and in the 1:1 mixture.
Z11 ) [X1(1 - R1*3)h11B11 + 3X1Ye,1tR1*2]d11 -
Table 3. Langmuir Constants of Reactive Dye Systemsa
X1(1 - R1* )h12B12d22
dye
Yo, mg g-1
Ko, L mg-1
r2
reactive black reactive red
541.96 423.68
0.02 0.06
0.99 0.99
3
X21 ) X2(1 - R2* )h21A11 3
a
X22 ) 1 + X2(1 - R2* )h22A12 3
The temperature was 298 K, and the adsorbent size was 0.044 nm.
Table 4. Freundlich Constants for Reactive Dye Systemsa
Z22 ) [X2(1 - R2*3)h22B12 + 3X2Ye,2tR2*2]d22 X2(1 - R2*3)h21B11d11 a
Equations 35 and 36 can be simplified as
dye
Yf, L g-1
n
r2
reactive black reactive red
36.58 84.47
0.46 0.31
0.97 0.93
The temperature was 298 K, and the adsorbent size was 0.044 mm.
Table 5. Redlich-Peterson Constants for Reactive Dye Systemsa
dC1t* Z11X22 + Z22X12 ) dτ X11X22 - X12X21
(37)
dC2t* Z11X21 + Z22X11 ) dτ X11X22 - X12X21
(38)
Numerical Solution. The four equations, namely, eqs 31, 32, 37, and 38 are solved simultaneously to find the concentrations of dye 1 and 2 (C1t and C2t) at any time using the RungeKutta method (using subroutine IVPRK from IMSL math library in Microsoft FORTRAN PowerStation). The initial conditions used are C1t* ) C2t* ) 1.0 and R1* ) R2* ) 1.0 at time τ ) 0.0. The four unknown process parameters, the external mass transfer coefficients (kf1 and kf2) and the internal effective diffusivities (Dp1 and Dp2), are estimated by minimizing the sum of the square of the errors between the experimental and the calculated concentration data of the two dyes, for some selected experiments. For this, the optimization subroutine BCPOL from IMSL math library of FORTRAN 90, which uses a direct complex search algorithm, is used. These parameter values are then used to predict the concentration profiles for the rest of the experiments, using the synthetic solutions as well as the industrial effluent. Results and Discussions Equilibrium Adsorption. Figure 1 shows the equilibrium adsorption of the dyes, where the solid symbols represent a single-component system and the hollow symbols are for the 1:1 mixture. The present adsorbent (CSD) shows a good adsorption capacity for the single-component systems of both the dyes. From the figure, it is observed that the equilibrium adsorption of both reactive black (dye 1) and reactive red (dye
dye
KR, L g-1
KP, (mg L-1)-ν
n
r2
reactive black reactive red
7.75 23.6
0.01 0.05
1.10 1.04
0.99 0.96
a
The temperature was 298 K, and the adsorbent size was 0.044 mm.
Table 6. Fritz-Schlunder Constants for Reactive Dye Systemsa dye reactive black reactive red a
KF2, KF1, (mg g-1)(mg L-1)-φ1 (mg L-1)-φ2 3.81 0.93
0.02 0.01
φ1
φ2
r2
1.319 1.22 0.99 2.588 2.46 0.99
The temperature was 298 K, and the adsorbent size was 0.044 mm.
2) decreases considerably in their mixture compared to the single-component system. This is probably due to mutual interaction effects. Adsorption equilibrium data for the single-component systems of the reactive dyes are fitted to a number of isotherms. Tables 3-6 show the isotherm constants and the correlation coefficient (r2) values for the different isotherms. It is observed that both the Langmuir and the Fritz-Schlunder isotherms fit well for reactive red, as the average value of the correlation coefficient (r2) is higher than 0.98. For reactive black, all the three isotherms, that is, the Langmuir, Redlich-Peterson, and FritzSchlunder isotherms, fit well (r2 is higher than 0.99). The maximum adsorption capacity of reactive black is about 542 mg g-1, and that of reactive red is about 424 mg g-1 at 298 K (from the Langmuir isotherm). Two-Component System. The comparison between the experimental and the calculated values of Ye from the bisolute Langmuir isotherm is shown in Figures 2 and 3 for reactive black (dye 1) and reactive red (dye 2), respectively. It is observed that although for reactive red the estimated values
Ind. Eng. Chem. Res., Vol. 45, No. 13, 2006 4737
Figure 2. Comparison between the experimental and the estimated Ye values for dye 1.
Figure 4. Experimental and model fitted concentration profiles of a synthetic mixture (C0,1/C0,2 ) 70.5:70 mg L-1; Ma ) 1.0 g).
Figure 5. Experimental and model fitted concentration profiles of a synthetic mixture (C0,1/C0,2 ) 54.18:92.26 mg L-1; Ma ) 1.0 g). Figure 3. Comparison between the experimental and the estimated Ye values for dye 2. Table 7. Interaction Factors η1,1
η1,2
η2,1
η2,2
0.88
0.71
1.32
0.83
(hollow circles) lie within (15% of the experimental values (Figure 3), there is appreciable variation between the experimental and the estimated Ye values (hollow squares in Figure 2) for reactive black. The interaction effects of two dissimilar components have been reported by many researchers to be the cause for the deviation.17 Specifically, the differences in the equilibrium adsorption capacity (Yo) signify components that are not similar in nature. As mentioned earlier, Yo for reactive red is 424 mg g-1 and that for reactive black is 542 mg g-1 (1.28 times greater than the previous). There have been further reports18 of deviation from the extended Langmuir isotherms in the literature under similar circumstances of differences in the adsorption capacity. Hence the extended Langmuir equation fails to predict the equilibrium adsorption capacity of reactive black.
To encounter this deviation, the modified bisolute Langmuir isotherm18 is used. The model equations are presented in eqs 7 and 8, where ηi,j is the interaction factor of component i for the adsorption of component j. The values of ηi,j are evaluated by minimizing the sum of the square of the errors between the experimental and the estimated Ye data, using an optimization algorithm. The values are as shown in Table 7. In Figures 2 and 3, the solid symbols represent the Ye values estimated from the modified bisolute Langmuir isotherm. It is evident from Figure 2 that there is a marked improvement in the predicted values for dye 1. The calculated values lie within (15% (solid squares). In Figure 3, there is marginal improvement in the predictions compared to the previous approach (see solid and hollow circles). Kinetics Study. An adsorption kinetics study is performed with both the synthetic solutions of the dyes and the industrial effluent having a mixture of dyes. It is observed that acidic pH favors the adsorption of both the dyes. Hence, all the experiments are carried out at an acidic pH (2.0). Model Predictions of the Kinetic Data. As discussed in the theory section, the model equations, that is, eqs 31 and 32 and eqs 37 and 38 are numerically solved to predict the concentration
4738
Ind. Eng. Chem. Res., Vol. 45, No. 13, 2006
Figure 6. Experimental and model fitted concentration profiles in a synthetic mixture (C0,1/C0,2 ) 100.63:46.84 mg L-1; Ma ) 1.0 g).
Figure 7. Prediction of the concentration profiles of dyes in a synthetic mixture (C0,1/C0,2 ) 100:100 mg L-1; Ma ) 1.0 g). Table 8. Model Parameters kf1 (m s-1) 9.46 ×
10-4
kf2 (m s-1) 1.67 ×
10-4
Dp1 (m2 s-1)
Dp2 (m2 s-1)
10-11
1.02 × 10-11
1.12 ×
profiles for both the dyes. The four unknown process parameters, that is, the external mass transfer coefficients (kf1 and kf2) and the internal effective diffusivities (Dp1 and Dp2), are estimated by optimizing the fitting of the experimental concentration profiles with those predicted from the model. For this, the following three experimental data points at three concentration ratios are selected: (i) C0,1 ) 70 mg L-1 and C0,2 ) 70.53 mg L-1; (ii) C0,1 ) 54.18 mg L-1 and C0,2 ) 92.26 mg L-1; and (iii) C0,1 ) 100.63 mg L-1 and C0,2 ) 46.84 mg L-1 for reactive black and reactive red, respectively. In Figures 4-6, the concentration profiles of both the dyes as functions of time are presented. The symbols represent the experimental data, and the solid lines are the model-fitted values. The values of the model parameters are given in Table 8. The estimated values of the parameters are used for the prediction of the concentration profiles with different initial concentrations (Figures 7-9). It is observed from the figures that the predicted values are close to the experimental data in
Figure 8. Prediction of the concentration profiles of dyes in a synthetic mixture (C0,1/C0,2 ) 75:100 mg L-1; Ma ) 1.0 g).
Figure 9. Prediction of the concentration profiles of dyes in a synthetic mixture (C0,1/C0,2 ) 150:100 mg L-1; Ma ) 1.0 g).
most of the cases. The average percentage deviation between the experimental and the predicted values is 8.5%. Thus the present model is successful in predicting the concentration profiles for the two-component mixture of dyes. As can be seen from Table 8, the values of DP1 and DP2 are approximately equal to each other, whereas kf1 is quite large compared to kf2. Therefore, it is expected that the contribution of intraparticle diffusion to the kinetics of the process is approximately equal in both cases. But the large values of kf1 denote a faster decay for dye 1 when other conditions (e.g., initial concentration and adsorbent amount) are kept the same. This is evident in Figure 4, which shows relatively rapid degeneration of dye 1 with respect to dye 2. Sensitivity Analysis of the Model Parameters. A sensitivity analysis of the model parameters is performed next. The accuracy of the model predicted concentration of both the dyes is insensitive to the estimated values of kf1 and kf2, which are not shown here. Figures 10 and 11 show the effect of change in the parameters Dp1 and Dp2 on the predicted concentration profiles of reactive black (dye 1) and reactive red (dye 2), respectively, at an initial concentration of 100:100 mg L-1 and adsorbent loading of 1.0 g L-1. A variation of (50% is introduced in the value of Dp1 and Dp2, and the concentration
Ind. Eng. Chem. Res., Vol. 45, No. 13, 2006 4739
Figure 10. Effect of variations in Dp1 on prediction of the concentration profile of dye 1 (C0,1/C0,2 ) 150:100 mg L-1; Ma ) 1.0 g).
Figure 12. Concentration decay of dye 1 in the industrial effluent using varying amounts of the adsorbent (pH ) 2.0; dp ) 0.044 mm; T ) 298 K).
Figure 13. Concentration decay of dye 2 in the industrial effluent using varying amounts of the adsorbent (pH ) 2.0; dp ) 0.044 mm; T ) 298 K). Figure 11. Effect of variations in Dp2 on prediction of the concentration profile of dye 2 (C0,1/C0,2 ) 100:100 mg L-1; Ma ) 1.0 g).
profiles are recalculated. The variations in the predicted concentration of dye 1 due to variations in Dp1 are found to be in the range of -32.75% to +15.73%. For dye 2, the variations in the predicted concentration due to variations in Dp2 are in the range of -13.18% to +4.09%. Hence, the variations in the value of effective diffusivity have a substantial effect on the prediction of the concentration profile for dye 1, and the effect is relatively less for dye 2. Adsorption Studies Using Industrial Effluent. The textile effluent, which contains a mixture of the two reactive dyes (reactive black and reactive red), is treated using varying amounts of the adsorbent (1.0-2.0 g L-1). The initial concentrations of dye 1 and dye 2 in the effluent are 173 and 115 mg L-1, respectively. The pH of the solution is maintained at 2.0, and smaller particles of adsorbent (0.044 mm) are used. The concentration decay has been calculated using the same shrinking core model with previously estimated model parameters. The concentration decay of the dyes is shown in Figures 12 and 13 for dye 1 and dye 2, respectively. In these figures, the symbols are the experimental points and the solid lines are the concentration profiles predicted from the model. It can be
observed from Figure 12 that, using an adsorbent dose of 1.0 g L-1, about 80% removal of dye 1 is possible in 1 h and the removal rate increases as the adsorbent dosing increases, due to the increase in the adsorption sites. When an adsorbent dose of 2 g L-1 is used, about 89% dye removal is possible. The decline in the concentration of dye 2 in the effluent is shown in Figure 13. It is observed from the figure that, by using 1.0 g L-1 adsorbent, about 79% removal of dye 2 is possible. When an adsorbent dose of 2.0 g L-1 is used, about 92% removal is achieved. From the figures, it is observed that the experimental points are reasonably close to the predicted lines in most of the cases. Overall, it can be concluded that the present twocomponent model is able to predict the transient concentration profile of the dyes in the adsorption of an industrial effluent. Taking 1 h as the basis, it is found that about 85% COD removal is achieved using a 1.0 g L-1 adsorbent dose. As expected, removal of COD increases with increase in mass of adsorbent (93% removal with 2 g L-1 of adsorbent), as a result of the increased adsorption of the dyes. Conclusions A waste carbonaceous material, sawdust, is used to develop an effective adsorbent and is used successfully for the removal
4740
Ind. Eng. Chem. Res., Vol. 45, No. 13, 2006
of two reactive dyes present in a textile effluent. Adsorption equilibrium data for the single-component systems fit reasonably well to the Langmuir and the Fritz-Schlunder isotherms for both the dyes. The extended Langmuir isotherm modified with the interaction factors agrees well with the equilibrium data of dye mixtures. The generalized shrinking core model is used to interpret the kinetic data of the two-component system. The proposed two-resistance mass transfer model is effective to predict the concentration profiles of the two-component dye system. Internal effective diffusivities and mass transfer coefficients are estimated. With these estimated parameter values, the concentration decay of the actual effluent is predicted. Adsorption of the textile effluent shows that, by using an adsorbent dose of 2 g L-1, the dye removal is about 93% and 89% in 1 h for reactive black and reactive red, respectively, while the COD removal is about 93%. Nomenclature A11 ) defined in eq 21 A12 ) defined in eq 22 Bi ) Biot number defined after eq 18 B11 ) defined in eq 21 B12 ) defined in eq 22 CSD ) charred sawdust Ce ) equilibrium liquid-phase concentration, mg L-1 Cet ) equilibrium liquid-phase concentration at time t, mg L-1 Cet* ) nondimensional equilibrium liquid-phase concentration at time t C0 ) initial liquid-phase concentration, mg L-1 Ct ) liquid-phase concentration at time t, mg L-1 Ct* ) nondimensional liquid-phase concentration at time t C0,1 ) initial liquid-phase concentration of dye 1 in the twocomponent mixture, mg L-1 C0,2 ) initial liquid-phase concentration of dye 2 in the twocomponent mixture, mg L-1 r2 ) correlation Coefficient Dp ) pore diffusion coefficient, m2 s-1 dp ) particle diameter, mm d11 ) function used in eq 31 d22 ) function used in eq 32 g11 ) defined in eq 19 g12 ) defined in eq 20 h11 ) defined after eq 27 h12 ) defined after eq 27 h21 ) defined after eq 28 h22 ) defined after eq 28 kf ) liquid-phase mass transfer coefficient, m s-1 Ko ) Langmuir isotherm constant, L g-1 Ko* ) defined after eqs 23 and 24 KF1 ) Fritz-Schlunder isotherm constant, (mg g-1)(mg L-1)-φ1 KF2 ) Fritz-Schlunder isotherm constant, (mg L-1)-φ2 KR ) Redlich-Peterson isotherm constant, L g-1 KP ) Redlich-Peterson isotherm constant, (mg L-1)-ν Ma ) mass of the adsorbent, g n ) Freundlich isotherm constant N(t) ) adsorption rate at time t, mg s-1 R ) adsorbent particle radius, m R* ) nondimensional concentration front Rf ) radius of concentration front, m S1 ) defined in eq 25 S2 ) defined in eq 26 T ) temperature, K
t ) time, s V ) volume of batch reactor, L X ) dimensionless group defined after eq 18 X11, X12 ) defined after eq 35 X21, X22 ) defined after eq 36 Yo ) maximum monolayer adsorption capacity, mg g-1 Ye ) solid-phase concentration, mg g-1 Yet ) solid-phase concentration at time t, mg g-1 Yes ) defined after eqs 23 and 24 Yf ) Freundlich isotherm constant, mg g-1 Ys ) Langmuir isotherm constant, mg g-1 Yt ) average solid-phase concentration at time t, mg g-1 Z11 ) defined after eq 35 Z22 ) defined after eq 36 Greek Symbols F ) adsorbent density, kg m-3 ν ) Redlich-Peterson isotherm constant φ1, φ2 ) Fritz-Schlunder isotherm constants τ ) nondimensional time Superscripts * ) dimensionless Subscripts 1 ) component 1 2 ) component 2 i ) ith component t ) at time “t” Literature Cited (1) Al-Degs, Y.; Khraisheh, M. A. M.; Allen, S. J.; Ahmad, M. N. A. Sorption Behavior of Cationic and Anionic Dyes from Aqueous Solution on Different types of Activated Carbons. Sep. Sci. Technol. 2001, 36, 91. (2) Orthman, J.; Zhu, H. Y.; Lu, G. Q. Use of Anion Clay Hydrotalcite to Remove Colored Organics from Aqueous Solutions. Sep. Purif. Technol. 2003, 31, 53. (3) Chiou, M.-S.; Li, H.-Y. Equilibrium and Kinetic modeling of Adsorption of Reactive Dye on Cross-linked Chitosan Beads. J. Hazard. Mater. 2002, 93, 233. (4) Gibbs, G.; Tobin, M. J.; Guibal, E. Influence of Chitosan Preprotonation on Reactive Black 5 Sorption Isotherms and Kinetics. Ind. Eng. Chem. Res. 2004, 43, 1. (5) Ozacar, M.; Ayhan, S. I. Adsorption of Reactive Dyes on Calcined Alunite from Aqueous Solutions. J. Hazard. Mater. 2003, 98, 211. (6) Robinson, T.; Chandran, B.; Nigam, P. Removal of Dyes from an Artificial Textile Dye Effluent by Two Agricultural Waste Residues, Corncob and Barley Husk. EnViron. Int. 2002, 28, 29. (7) Robinson, T.; Chandran, B.; Naidu, S.; Nigam, P. Studies on the Removal of Dyes from a Synthetic Textile Effluent using Barley Husk in Static-batch mode and in a Continuous Flow, Packed-bed Reactor. Bioresour. Technol. 2002, 85, 43. (8) Butler, J. A. V.; Ockrent, C. Studies in Electrocapillarity. Part III. The Surface Tensions of Solutions Containing Two Surface-Active Solutes. J. Phys. Chem. 1930, 34, 2841. (9) Jena, P. R.; De, S.; Basu, J. K. A Generalized Shrinking Core Model Applied to Batch Adsorption. Chem. Eng. J. 2003, 95, 143. (10) Vogel, A. I. Textbook of Practical Organic Chemistry; Longmans: London, 1970. (11) Trivedi, R. K.; Goel, P. K. Chemical and Biological Methods for Water Pollution Studies, 2nd ed.; Environmental Publication: Aligarh, 1986. (12) Datta, S.; Basu, J. K.; Ghar, R. N. Studies on Adsorption of p-nitro Phenol on Charred Sawdust. Sep. Purif. Technol. 2001, 21, 227. (13) Methods and sampling and test for actiVated carbons, powdered and granular; Bureau of Indian Standards: New Delhi, India, 1977 (IS: 877). (14) Langmuir, I. Adsorption of Gases on Plain Surfaces of Glass Mica Platinum. J. Am. Chem. Soc. 1918, 40, 1361. (15) Freundlich, H. M. F. Over the Adsorption in Solution. J. Phys. Chem. 1906, 57, 385.
Ind. Eng. Chem. Res., Vol. 45, No. 13, 2006 4741 (16) Choy, K. K. H.; Mckay, G.; Porter, J. F. Sorption of Acid Dyes from Effluents using Activated Carbon. Resour. ConserV. Recycl. 1999, 27, 57. (17) Suzuki, M. Adsorption Engineering; Elsevier: Tokyo, 1990. (18) Leyva-Ramos, R.; Bernal-Jacome, L. A.; Guerrero-Coronado, R. M.; Fuentes-Rubio, L. Competitive Adsorption of Cd(II) and Zn(II) from Aqueous Solution onto Activated Carbon. Sep. Sci. Technol. 2001, 36, 3673. (19) Ho, Y. S.; Mckay, G. Competitive Sorption of Copper and Nickel Ions from Aqueous Solution using Peat. Adsorption 1999, 5, 409. (20) Mckay, G.; Al-Duri, B. Prediction of Multicomponent Adsorption Equilibrium data using Empirical Correlations. Chem. Eng. J. 1989, 41, 9. (21) Dimitrios, C.; Verma, A.; Irvine, R. L. Activated Carbon Adsorption and Desorption of Toluene in the Aqueous Phase. AIChE J. 1993, 39, 2027.
(22) Mckay, G. Analytical Solution using a Pore Diffusion Model for the Adsorption of Basic Dye on Silica. AIChE J. 1984, 30, 692. (23) Levenspiel, O. Chemical Reaction Engineering, 2nd ed.; Wiley: New York, 1972. (24) Jena, P. R.; Basu, J. K.; De, S. A. Generalized Shrinking Core Model for Multicomponent Batch Adsorption Process. Chem. Eng. J. 2004, 102, 267.
ReceiVed for reView March 3, 2005 ReVised manuscript receiVed March 14, 2006 Accepted March 18, 2006 IE050302F