Chemosphere 53 (2003) 655665 www.elsevier.

com/locate/chemosphere

Removal of PCBs from wastewater using y ash

Hendrik Nollet a, Murielle Roels a, Pierre Lutgen b, Paul Van der Meeren c, Willy Verstraete a,*
a

Laboratory of Microbial Ecology and Technology (LabMET), Faculty of Agricultural and Applied Biological Sciences, Ghent University, Coupure Links 653, B-9000 Ghent, Belgium b Umwelt Morgen, Rue Kommes, L-6988 Hostert, Luxembourg c Department of Applied Analytical and Physical Chemistry, Faculty of Agricultural and Applied Biological Sciences, Ghent University, Coupure Links 653, B-9000 Ghent, Belgium Received 3 July 2002; received in revised form 25 April 2003; accepted 25 April 2003

Abstract Liquids and sludges containing polychlorinated biphenyls (PCBs) can be treated to concentrate the PCBs in a solid residue. The latter can then be handled to destroy the PCBs. A study on sorption kinetics of PCBs on y ash was conducted in controlled batch systems. TCB and HeCB are removed at 25 C by adsorption on y ash up to 97% at pH 7, with an adsorbent dose of 5 g/l. An examination of the thermodynamic parameters shows that the adsorption of TCB and HeCB by y ash is a process occuring spontaneously at ambient conditions. Activation energies for the sorption process ranged between 5.6 and 49.1 kJ/mol. It was observed that the rate at which TCB and HeCB are adsorbed onto y ash showed a diusion limitation. The uptake rate of TCB and HeCB increases with increasing initial concentration and gradually tends to a constant value. A decrease in the adsorption of TCB and HeCB was observed when interfering ions and other PCB congeners were present. Changing the pH in the aqueous solution from 2 to 10 had no eect on the adsorption process. Overall, y ash can be used for an ecient removal of PCBs from several aqueous solutions. 2003 Elsevier Ltd. All rights reserved.
Keywords: Polychlorinated biphenyl; Fly ash; Adsorption; Kinetics; Activation energy

1. Introduction Polychlorinated biphenyls (PCBs) are widely acknowledged to be a group of the most persistent environmental contaminants which contains many congeners with contrasting environmental behaviour and toxicity. As a group PCBs pose high toxicological risks to wildlife, to agricultural ecosystems, and consequently to the human food-chain, because of their stable properties and ubiquitous distribution (World Health Organization, 1993). These considerations ultimately caused their

* Corresponding author. Tel.: +32-9-2645976; fax: +32-92646248. E-mail address: willy.verstraete@rug.ac.be (W. Verstraete). URL: http://welcome.to/LabMET

withdrawal from commercial production and use, but even 30 years later there is still much concern about the continued dispersion of PCBs into the wider environment from contaminated land and designated waste disposal sites. In case these substances have accumulated in highly concentrated and accessible sites, o site incineration of the contaminated material is the main and only technique used so far for their destruction (Van Agteren et al., 1998). Activated carbon adsorption is frequently used to remove apolar pollutants from wastewater. Hence, this adsorbent can be used as an eective remover of PCBs which are very apolar (Sotelo et al., 2002). In the present study, eorts have been made to investigate the use of y ash, a coal combustion byproduct, as a low cost adsorbent replacing activated carbon for the removal of PCBs.

0045-6535/$ - see front matter 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0045-6535(03)00517-4

656

H. Nollet et al. / Chemosphere 53 (2003) 655665

A literature survey revealed that y ash has been used as one of the ingredients for concrete and admixture used in the construction of roads, dams, and bridges (Roy et al., 1981). Fly ash can act as a catalyst during the formation or break down of persistent compounds. There exists evidence that catalytic reactions (the socalled de novo synthesis) within the range of 200350 C on the surface of y ash particles in the postcombustion zone of a municipal waste incinerator play an important role in the formation of highly noxious halogen-containing persistent organic pollutants such as PCBs and polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/F). Thermal treatment of y ash with trace amounts of PCDD/F (82116 ng/g) in oxygen-decient atmosphere is an eective method for detoxication of these pollutants, even at lower temperatures (Stach et al., 1999, 2000). Several investigators (Johnson et al., 1965; Peloso et al., 1983; Kumar et al., 1987; Mott and Weber, 1992; Viraraghavan and Alfaro, 1994; Banerjee et al., 1995, 1997; Akgerman and Zardkoohi, 1996; Gupta, 1998; Aksu and Yener, 1999; Gupta et al., 2000; Kao et al., 2000; Sakellarides and Albanis, 2000; Gupta and Ali, 2001; Ricou-Hoeer et al., 2001; Nollet et al., 2002) explored the use of y ash as an adsorbent for the treatment of wastewater to remove a variety of organic compounds and color. They concluded that y ash has a signicant adsorption capacity for the removal of organic compounds from aqueous solutions. However, little investigation has been conducted to determine the kinetics of the adsorption process using y ash. The objectives of this research were to develop a low cost adsorbent, to gain an understanding of the adsorption process kinetics and to determine the factors controlling the rate of adsorption.

Fly ash is characterized by its neness, its large surface area (16 m2 /g) (Mancy et al., 1964) and its wide particle size distribution. A Malvern Mastersizer/S (Malvern Instruments Ltd.) equipped with a MS-17 wet sample dispersor unit and a 300RF lens was used to determine the particle size of the y ash. Before analysis the samples were diluted to prevent multiple scattering. The obscuration level was set at about 1516%. The refractive index model used for analysis by Mie theory was goethite in water (real refractive index 2.300, imaginary refractive index 1.000). The particle size of this material ranges from 7 to 120 lm. The major components of y ash are alumina, silica, iron oxide, calcium oxide and residual carbon. However, the constituents of y ash vary according to the type of coal used and the degree of combustion. The main chemical components of this y ash, expressed as lg/g y ash, were: K (497 218), Na (216 469), Ca (6035 841), Mg (5227 1053), Zn (11 0.13), Cu (7.72 0.91), Pb (3.79 0.14), Fe (1248 134), Cr (24.1 0.4), Cd (0.258 0.019), Mn (120 0.9), Co (4.95 0.32), Ag (0.1 0.025) and Ni (5.81 0.17). Fly ash was soxhlet extracted with acetone/hexane (1:1) in portions of 5 g for 20 h. No PCBs were detected in the concentrated extracts. 2.3. Adsorption batch experiments Kinetics and adsorption capacity of the y ash for TCB and HeCB were determined by means of controlled experiments in a batch system. Optimisation of adsorption was achieved by varying dierent parameters such as pH, initial concentration, dose of adsorbent, equilibration time, size of adsorbent particle and working temperature. 2.4. Adsorption studies The y ash used in this research was washed with distilled water for a period of 24 h in order to remove dust and ne particulate matter. The samples were dried overnight at 105 C before use. The stock solutions of the PCBs were made in methanol; the concentration of TCB and HeCB were 500 and 250 mg/l respectively. The adsorption experiments were performed in closed batch systems prepared in 120 ml serum bottles, capped with viton stops and sealed with an aluminium cap. The bottles were incubated in the dark at ambient temperature (22 1 C) and allowed to equilibrate on a rotary shaker. The isotherms were run by taking dierent concentrations of TCB (250, 500, 750, 1000, 2000 lg/l) and HeCB (125, 250, 375, 500, 1000 lg/l) at a desired temperature and pH. These concentrations were selected after preliminary investigations. Parallel experiments

2. Materials and methods 2.1. Chemicals 2,3,4-trichlorobiphenyl (PCB no. 21; TCB) and 2,20 ,3,30 ,4,5,6-heptachlorobiphenyl (PCB no. 173; HeCB) were purchased from Dr. Ehrenstorfer (Augsburg, Germany). The GC grade solvents methanol (highest purity) and hexane (for organic trace analysis) were purchased from Merck (Overijse, Belgium). The Bond Elut (C18, 3 ml, 200 mg, particle size 40) cartridges were obtained from Varian (St.-Katelijne-Waver, Belgium). All PCB stocks were prepared freshly in methanol before use. 2.2. Fly ash Fly ash from an electrical power generating station of Luxembourg was used in the adsorption experiments.

H. Nollet et al. / Chemosphere 53 (2003) 655665

657

were also performed without adding y ash particles (control). At a given time interval, 2 ml of the aqueous solution was withdrawn by a gastight syringe without opening the cap. After centrifugation the solutions were brought into a solid phase extraction (SPE) column with bond Elut C18. The concentrations of TCB and HeCB were determined by using gas chromatography (GC-ECD). 2.5. Kinetic studies The kinetics of the adsorption of TCB and HeCB onto y ash were evaluated through the use of controlled experiments in a batch system. A series of 120 ml serum bottles containing a dened volume of solutions of TCB and HeCB of known concentrations were kept in a thermostatic water bath placed on a shaker. After attaining the desired temperature, a known amount of the adsorbent was added to three of the four asks and the asks were allowed to agitate mechanically. The vials containing y ash were done in triplicate, and the vial without y ash was used as a control. The latter always showed negligible adsorption of TCB and HeCB onto the surface of the glass serum bottles. Samples were taken as described above and further analysed for TCB and HeCB concentration by GC-ECD. 2.6. Analytical techniques PCBs were extracted by using SPE. The cartridge used was a Bond Elut C18, with a column volume of 3 ml, a sorbent mass of 200 mg and a particle size of 40 lm (Nollet et al., 2002). After extraction, the PCBs were analysed with GC-ECD (Varian 3800). The extracted sample (1 ll) was injected with a split ratio of 1/10. The temperature was held for 1 min at 170 C and then increased upto 270 C in 10 min. Finally the temperature was held at 270 C for 2.5 min. The pressure on the column (CP-sil 8) was kept at 20 psi. Peak areas were quantied based on standard curves. 2.7. Kinetics (theory) The kinetics of sorption are described by the solute uptake rate which determines the residence time required for complete sorption. The sorption kinetics ultimately control the process eciency. The sorption of organic compounds from a liquid phase to a solid phase can be considered a reversible process with equilibrium being established between the two phases. The kinetic model that was used in the research to describe the reaction rates was based on the assumption that the adsorption of TCB and HeCB onto the y ash is a diusion controlled and rst-order reversible process. It can be expressed as:

AB

The rate equation for the reaction can be written as: dCB dCA dt dt k1 CA0 CA0 XA;t k2 CB0 CA0 XA;t

where CA0 initial concentration of solute in solution at time, t 0; CB0 initial concentration of solute on the sorbent at time, t 0; CA concentration of solute in solution at any time during the reaction; CB concentration of solute on the sorbent at any time during the reaction; XA;t fraction adsorbed onto the sorbent at time t; k1 rst-order adsorption rate constant; k2 rst-order desorption rate constant. At the equilibrium condition: dCB dCA 0 dt dt and XAe kc CB0 =CA0 1 kc 4 3

where XAe fraction of solute adsorbed at equilibrium condition, and kc equilibrium constant dened as: kc k1 CB0 CA0 XAe k2 CA0 CA0 XAe 5

In terms of equilibrium conversion, the dierential form of the rate equation is: dXA;t k1 k2 XAe XA;t 6 dt Integration of Eq. (6), substituting k1 =kc for k2 , gives:     XA;t k1 k1 t ln 1 XAe kc   XA;t k 0 t 7 ln 1 XAe where k 0 overall rate constant k1 k1 k1 k2 kc 8

2.7.1. Particle diusion control process When the rate of a reaction is controlled by the intraparticle diusion process, the uptake curve for the solution is represented by the following familiar expression (Ruthven, 1984):   x XA;t 6 X 1 n2 p2 Dc t 9 1 2 exp XAe p n1 n2 r2 where XA;t =XAe fractional approach to equilibrium, Dc intracrystalline diusivity, r particle radius, and t reaction time.

658

H. Nollet et al. / Chemosphere 53 (2003) 655665

The lines representing the above expression converge asymptotically over time since the higher terms of the summation become small and are negligible. Therefore, Eq. (9) can be rewritten in the following simplied form:  2  XA;t 6 p Dc 1 2 exp 2 t 10 XAe p r A plot of ln1 XA;t =XAe versus t should be linear with a slope of p2 Dc 1=r2 , which is commonly known as the reaction rate constant or diusional time constant. The slope can be expressed as: k 0 p2 Dc =r2 11
Fig. 1. Adsorption of TCB on y ash (5 g/l) at 25 C for different initial concentrations of TCB ((- -j- -) 250 lg/l; (j) 500 lg/l; (- -m- -) 750 lg/l; (m) 1000 lg/l; (- - - -) 1500 lg/l; ( ) 2000 lg/l).

where k 0 overall rate constant. Therefore, in an intraparticle diusion controlled process, the relative rate is inversely proportional to the square of the particle radius. 2.7.2. Estimation of thermodynamic parameters The thermodynamic parameters such as free energy change (DG0 ), enthalpy change (DH 0 ), and entropy change (DS 0 ) for the process were estimated using the following equations: DG0 RT ln kc DG0 DH 0 T DS 0
0

12 13

where DG change in free energy, kJ/mol; DH 0 change in enthalpy, kJ/mol; DS 0 change in entropy, kJ/ (mol K); T absolute temperature, K; R gas constant 8.314 J/(mol K); kc equilibrium constant. From Eqs. (12) and (13) it can be written as: ln kc DS 0 DH 0 1 T R R 14
Fig. 2. Adsorption of HeCB on y ash (5 g/l) at 25 C for dierent initial concentrations of HeCB ((- -j- -) 125 lg/l; (j) 250 lg/l; (- -m- -) 375 lg/l; (m) 500 lg/l; (- - - -) 750 lg/l; ( ) 1000 lg/l).

Eq. (14) represents a mathematical relationship between kc and 1=T . A graph of ln kc versus 1=T would yield a straight line. DH 0 and DS 0 can be estimated from the slope and intercept of that line. 3. Results 3.1. Equilibration time Figs. 1 and 2 show the data from the adsorption studies of TCB and HeCB with y ash for dierent initial concentrations plotted as a function of time. The decreasing concentration of TCB and HeCB in the solution indicates that TCB and HeCB were adsorbed by the y ash, while losses in the controls never exceeded 6%. Initially TCB and HeCB were adsorbed rapidly from the solution, subsequently decreasing adsorption rates were observed and after a certain time period the remaining concentration of TCB and HeCB detected in the waterphase remained constant.

The adsorption experiments were carried out for dierent concentrations of TCB and HeCB with a xed adsorbent dose (5 g/l), pH (7.0) and temperature (25 C). The results (Figs. 1 and 2) indicate that a constant concentration for TCB and HeCB in the vessels was observed after about 3000 min (50 h) of contact time. Depending upon the initial concentration, a 8097% removal of TCB and HeCB was achieved during the rst 24 h of contact time, while only 25% of additional removal occurred in the following 24 h of contact time. 3.2. The eect of dose To achieve the maximum adsorption capacity of adsorbent for TCB and HeCB and thus the maximum

H. Nollet et al. / Chemosphere 53 (2003) 655665

659

removal of TCB and HeCB out of the waterphase, the dose of the adsorbent was varied from 1 to 10 g/l. Adding more than 5 g/l could not decrease the remaining concentration of TCB and the HeCB detected in the waterphase. These ndings are plotted in Fig. 3. The adsorbed amount decreases from 0.83 mg/g at dose 1 g/l to 0.16 mg/g at dose 6 g/l for TCB and from 0.45 mg/g at dose 1 g/l to 0.12 mg/g at dose 4 g/l for HeCB, subsequently stabilizing for both adsorbates. 3.3. Overall rate constant Eq. (10) permits to calculate the overall rate constant (k 0 ) from the plot of ln1 XA;t =XAe versus reaction time. So, the logarithmic values of the fraction of TCB and HeCB remaining in solution after 24 h (who still can be adsorbed) were plotted against the reaction time for different initial concentrations of TCB and HeCB (Fig. 4).

This linearity test is employed to check the nature of the adsorption mechanism. In the whole concentration range, the ln1 XA;t =XAe versus t plot for TCB and HeCB was linear up to a certain point and passed through the origin indicating that the adsorption is limited by particle diusion. The straight line t through the origin was observed (R2 > 0:98) for TCB and HeCB. From these data, it can be concluded that the adsorption reaction can be approximated by rst-order reversible kinetics. The values of k 0 and other constants such as the forward reaction rates, k1 (adsorption of TCB and HeCB from solution onto y ash) and the reverse reaction rates, k2 (desorption of TCB and HeCB from the y ash to the solution) were calculated using Eqs. (5) and (8) and are summarized in Table 1. 3.4. Eect of solute concentration The adsorption experiments were carried out using dierent concentrations of TCB and HeCB in the range of 2502500 and 1251000 lg/l respectively for a dose of 5 g y ash per liter, a xed pH (7.0), temperature (25 C) and contact time (24 h). The results given in Fig. 5 reveal that the adsorbed amount after 24 h increases almost linearly with increasing concentrations of TCB and HeCB and that the amount adsorbed for HeCB is higher than for TCB. Table 1 shows that the overall rate constant values (k 0 ) increased slightly with increasing initial concentration of TCB and HeCB. In Figs. 6 and 7, the overall rate constant values are plotted as a function of initial concentration. This graph indicates that the uptake rate of TCB and HeCB increases no signicantly with increasing initial concentration and almost stay constant. An increase in uptake rate with increasing initial concentration was only observed for the experiments conducted at 55 C. 3.5. The eect of pH The adsorption of TCB and HeCB was studied over the pH range of 210. No signicant dierence for adsorption of TCB and HeCB for several pH could be observed (data not given). 3.6. The eect of interfering ions and other PCB congeners Natural waters can contain various ions such as sodium, potassium, calcium, magnesium, chloride, sulphate, nitrate and phosphate. The eect of these ions was studied on the adsorption of TCB and HeCB on y ash at various levels. The observed decrease in the removal of TCB and HeCB on y ash is presented in Table 2.

Fig. 3. The eect of adsorbent dose on the remaining concentration in the waterphase ((m) TCB and (j) HeCB) and the amount adsorbed of TCB and HeCB ((- -m- -) TCB and (- -j- -) HeCB).

Fig. 4. First-order reversible kinetic t for TCB and HeCB ((- -m- -) 500 lg TCB/l; (m) 1500 lg TCB/l; (- -j- -) 250 lg HeCB/l; (j) 1000 lg HeCB/l).

660

H. Nollet et al. / Chemosphere 53 (2003) 655665

Table 1 Values of rate constant (min1 ) at dierent initial concentrations, derived from the kinetic study at 55 C Initial concentration of solute (lg/l) TCB 250 500 750 1000 1500 2000 125 250 375 500 750 1000 k 0 (103 ) 5.78 0.21 6.20 0.17 6.62 0.24 6.86 0.31 7.54 0.27 8.02 0.16 5.15 0.19 5.39 0.23 5.79 0.14 6.18 0.17 6.56 0.26 7.18 0.20
0 k1 (103 ) 0 k2 (103 )

5.42 0.03 5.86 0.02 6.36 0.04 6.62 0.07 7.35 0.06 7.77 0.02 5.13 0.03 5.38 0.04 5.78 0.02 6.17 0.02 6.55 0.05 7.11 0.03

0.354 0.006 0.338 0.005 0.241 0.007 0.242 0.011 0.189 0.008 0.255 0.005 0.0243 0.006 0.0148 0.007 0.0093 0.006 0.0092 0.005 0.0053 0.008 0.065 0.006

HeCB

Fig. 5. The relation between the initial concentration of TCB and HeCB and the amount adsorbed of TCB (m) and HeCB (j) for 5 g y ash/l added.

Fig. 7. Eect of initial concentration of HeCB on overall rate constant ((j) 5 C; (N) 25 C; (d) 55 C).

Table 2 Eect of interfering ions on the decrease of PCB adsorption TCB (%) NaCl (102000 mg/l) NH4 Cl (102000 mg/l) CaCl2 2H2 O (10500 mg/l) CuCl2 2H2 O (10500 mg/l) Na2 SO4 (10500 mg/l) Na2 CO3 10H2 O (10500 mg/l) NaH2 PO4 2H2 O (10500 mg/l) 23.6 25.8 20.5 18.2 8.5 31.9 41.3 HeCB (%) 6.4 11.6 3.6 4.5 0 11.2 23.5

Fig. 6. Eect of initial concentration of TCB on overall rate constant ((j) 5 C; (m) 25 C; (d) 55 C).

PCBs are generally used as mixtures of 6080 dierent chlorinated congeneres. Hence when PCBs are de-

tected in the environment, not only one PCB congener but a variety of PCB congeners will be found. Therefore an experiment was conducted to investigate the inuence of other PCBs on the adsorption of one particular PCB. The observed decrease in the adsorption of TCB and HeCB on y ash when both compounds were present together was respectively 20.811.8% and 4.80.6% for 26 g/l of y ash added.

H. Nollet et al. / Chemosphere 53 (2003) 655665

661

3.7. The adsorption isotherm The adsorption data of TCB and HeCB were tted in the Freundlich and Langmuir models. Freundlich isotherm. The logarithmic form of the Freundlich model is given by the following equation: log X log kf 1=n log Ce 15

ln k 0 ln A

Ea RT

17

where X is the amount adsorbed (mg/g); Ce is the equilibrium concentration of the adsorbate (mg/l); and kf and n are the Freundlich constants related to adsorption capacity and the sorbent anity for the adsorbate, respectively (Weber, 1972). When log X is plotted against log Ce , a straight line with slope 1=n is obtained which shows that the adsorption of TCB and HeCB follows the Freundlich isotherm (R2 > 0:99). The Freundlich constants kf and n were calculated and their temperature dependence is presented in Table 3. Langmuir isotherm. The model assumes uniform energies of adsorption onto the surface and no transmigration of adsorbate in the plane of the surface. The linear form of the Langmuir isotherm is given by Eq. (16): 1=X 1=Xm 1=aXm Ce 16

where A is the pre-exponential factor or the frequency factor, k 0 is a specic rate constant, and Ea is called the activation energy, representing the minimum energy that reactants must have for the reaction to proceed. The experimental activation energy was determined using dierent values of k 0 with their corresponding temperature values from the Vant HoArrhenius equation, as follows: ln
0 k1 E T 1 T 2 0 RT1 T2 k2

18

where X is the amount adsorbed (mg/g); Ce is the equilibrium concentration of the adsorbate (mg/l). Xm and a are the Langmuir constants related to adsorption capacity and energy of adsorption, respectively. When 1=X is plotted against 1=Ce , a straight line was tted to the data, but the regression coecient values were lower than 0.5 for TCB, while for HeCB better ts (R2 > 0:99) were obtained. These results indicate that the adsorption of TCB by y ash did not follow the Langmuir adsorption isotherm, while HeCB did. 3.8. Eect of temperature on rate of adsorption Kinetic experiments were conducted at 5, 25 and 55 C. These experiments were done to investigate the thermodynamic characterictics of the rate of TCB and HeCB adsorption by y ash. The results are summarized in Figs. 6 and 7. They show that the adsorption rate increases with increasing temperature. The increase in adsorption rate is described by the Arrhenius equation:
Table 3 Freundlich and Langmuir constants for TCB and HeCB Adsorbates Langmuir constants a (l/mg) 5 C TCB HeCB 71.52 25 C 21.79 55 C Xm (mg/g) 5 C 0.118 25 C 0.149 55 C

In the above equation, R is the gas constant and is equal to 8.31 J/(mol K), T1 and T2 are temperatures (K) at 0 0 condition 1 and 2, and k1 and k2 are the reaction rate constant values at conditions 1 and 2. The activation energy (Ea ) values are summarized in Table 4. The thermodynamic parameters such as free energy change (DG0 ), enthalpy (DH 0 ) and entropy change (DS 0 ) were determined using Eqs. (13) and (14). The values of the equilibrium constant (kc ) at dierent temperatures were calculated using Eq. (5). The slope and intercept which were obtained from plotting the logarithmic values of equilibrium constant (kc ) versus inverse temperature (1=T ), revealed the values of DH 0 and DS 0 . The values of the thermodynamic parameters are summarized in Table 5.
Table 4 Activation energy values Activation energy E (kJ/mol) TCB 525 C 2555 C 6.724 14.03 HeCB 49.08 5.573

Table 5 Thermodynamic parameters TCB DH 0 (kJ/mol) DS 0 (kJ/(mol K)) DG0 (kJ/mol) )15.697 )0.0195 )9.847 HeCB 40.468 0.163 )8.432

Freundlich constants n 5 C 1.21 1.32 25 C 1.03 1.37 55 C 0.76 0.72 kf (mg/g) 5 C 0.93 0.41 25 C 1.06 0.38 55 C 4.06 1.52

662

H. Nollet et al. / Chemosphere 53 (2003) 655665

4. Discussion 4.1. Adsorption equilibrium Out of the results discussed above it is clear that y ash can be used to remove PCBs from an aqueous solution. Initially the rate of adsorption of TCB and HeCB onto y ash was very high (Figs. 1 and 2). This was then followed by a slower rate, and gradually approached a plateau. Equilibrium was reached after a rather long contact time of 50 h. Sotelo et al. (2002) observed an adsorption equilibrium for 2-monochlorobiphenyl and 2,20 ,5,50 -tetrachlorobiphenyl of 25 and 75 h respectively onto granular activated carbon which corresponds to the results discussed in this paper. The kinetic model that was used in the research to describe the reaction rates was based on the assumption that the adsorption of TCB and HeCB onto y ash is a diusion controlled rst-order reversible process. The model was also used by Banerjee et al. (1997) to describe the adsorption of o-xyleen by y ash. When the remaining concentrations of TCB and HeCB in the waterphase for dierent initial concentrations ranging between 2502000 and 1251000 lg/l respectively were plotted against t0:5 , a straight line in the initial phase of the adsorption reaction was obtained (results not shown). In all the cases, the regression coefcient values (R2 > 0:98) suggest that the uptake of TCB and HeCB varies almost linearly with the square root of time. This functional relationship corresponds to the characteristic of intraparticle diusion (Ruthven, 1984). Out of Fig. 3 it is clear that the adsorbed amount of the y ash decreases when the dose of y ash was increased, but very important here is that with a higher dose of y ash added the remaining concentration in de waterphase decreased to become stable at a dose of 5 g/l. To apply y ash as an adsorbent in industry two conditions must be fullled. First of all the operational cost must be as low as possible, so the dose of y ash must be optimal. Secondly, the removal eciencies must be sufcient. A dose of 5 g/l meets those two conditions. Therefore, during all the experiments discussed in this paper a dose of 5 g y ash per liter was applied. As revealed by the water solubility data, the polarity of PCBs, and thus their anity towards the aqueous phase, decreases with increasing chlorine substitution. One may expect that the amount adsorbed from water for a given equilibrium concentration should increase in the order of TCB to HeCB. This was conrmed in our experiments (Fig. 5). This is in contradiction with the results observed elsewhere (Pirbazari and Weber, 1981; Sotelo et al., 2002), but obvious according to Traubes rule. Traubes rule can be considered as a conrmation of the Polanyi adsorption potential theory, which is only applicable to physical adsorption (Manes and Hofer, 1969). These results indicate that the PCBs with very low

solubility tend to be chemisorbed on the surface of the adsorbent. The low solubility favors irreversible adsorption, since the solutesolvent interaction is very week. As we can conclude from the partition co ecients, the adsorption increases as the y ash content and the lipophilic character of the compounds increase. One of the objectives was to investigate the eects of the solute concentration on the adsorption rate, in a series of experiments in which the initial concentrations of TCB and HeCB ranged between 250 and 2000 lg/l and 125 and 1000 lg/l respectively. The overall rate constant values (k 0 ) remained almost constant or increased slightly at higher temperature (55 C) with an increase in initial concentration (Figs. 6 and 7). In most of the cases the uptake rate increases with a higher initial concentration of the solute, and gradually approaches a plateau (Banerjee et al., 1997). Those results were observed for a dose of adsorbent for which the maximum adsorption capacity was reached. In the experiments conducted in this paper the adsorbent is always present in excess, the maximum adsorption capacity will never be reached. The adsorption capacity of y ash (Fig. 5) increases with increasing concentrations of TCB and HeCB, but never reached a plateau within the concentration range tested. The resistance to the uptake of solute from the solution decreases with the increasing adsorbent concentration. So the adorbent dose is in no case the inhibiting factor. The observed adsorption capacity of GAC, compared to y ash, was a factor 1.3 higher for TCB and a factor 1.1 lower for HeCB within the same concentration range of TCB and HeCB (results not published). The results presented in Figs. 6 and 7 indicate that the adsorption rate increases with increasing temperature. Some of the properties of closely related molecules are dependent on the dierences in the arrangement of atoms in space. In some cases, the substitution of molecular fragments can aect the planarity of the molecules; e.g., in case of PCBs, it is largely determined by degree of substitution on the ortho positions. In the absence of ortho substitution, the two phenyl rings are more or less in the same plane. The increasing substitution at the ortho position leads to increased crowding and results in twisting of the rings. Since the larger the planar surface, the larger the adsorption, the non-ortho substituted PCB isomers show greater adsorptivity; e.g. KOC value of 2,20 -dichlorobiphenyl is 5 times less than KOC value for 2,4-dichlorobiphenyl. TCB has one ortho substituent while HeCB has two ortho substituents, but during the adsorption process the eect of the number of chlorines (lipophilic character) is probably more important than the number of ortho substituents. This was conrmed by the experiment where the adsorption of TCB and HeCB were studied when both compounds were present together. The adsorption of TCB decreased

H. Nollet et al. / Chemosphere 53 (2003) 655665

663

with 11.820.8% when HeCB was present for dierent concentrations of y ash added. The inuence of TCB on the adsorption of HeCB was less pronounced, a slight decrease of 0.64.8% was observed. The results also indicate that the eect of interference of other PCBs present on the adsorption of one particular PCB decreases with increasing concentrations of y ash added. The observed decrease in the adsorption of TCB and HeCB in the experiments conducted to investigate the inuence of interfering ions is in the range of 031.9% (Table 2). These ndings suggest a non-selective nature of the adsorbent during the uptake of these two PCBs in the presence of various ions. Over a broad pH range, no signicant dierences in adsorption of TCB and HeCB on y ash was observed. This was probably due to the relatively high dose of y ash added. However, Gupta (1998), Gupta and Ali (2001) and Ricou-Hoeer et al. (2001) observed an optimal removal of the compounds studied on the adsorbate at a pH ranging from 5 to 7. 4.2. Adsorption kinetics The parameters of the kinetic model proposed above were estimated for each compound by means of curve tting. A set of kinetic experiments was carried out. Experiments without y ash were also performed to ensure that solute volatilization was not important in the adsorption period studied. No signicant change in TCB and HeCB concentration in the waterphase was detected. The model used is capable to reproduce the experimental data. The kinetics of the adsorption of TCB and HeCB by y ash can be approximated by rst-order reversible kinetics. The values of k 0 ranged between 5.78 103 and 8.02 103 per min for TCB and between 5.15 103 and 7.18 103 per min for HeCB. The forward rate constant values are 7.5 to 48.1 times for TCB and 3.1 to 61.8 times for HeCB greater than the reverse rate constant values at 25 C. The values of k 0 are a factor 5 lower than those observed for TCB (24.23 103 26.22 103 per min) and HeCB (19.12 103 24.12 103 per min) on GAC (results not published) and a factor 10 lower than those described by Banerjee et al. (1997) for o-xylene probably due to the longer equilibration time causing a at slope of the curve. While in the latter study the forward rate constant values for o-xylene were 34 times higher than the reverse rate constant values. The lower solubility of PCBs compared to o-xylene (factor 10) favors irreversible adsorption, since the solutesolvent interaction is very weak. The results of adsorption equilibrium for TCB and HeCB are shown in Table 3. The experimental data were tted with the Langmuir and Freundlich isotherm. The Langmuir equation did not reproduce equilibrium data satisfactorily for TCB. This can be attributed to (i) the

energetic heterogeneity of the y ash used or to (ii) the model of Langmuir. A better reproduction was obtained for HeCB, indicating that the range of adsorption energy for this compound is smaller than that for the TCB. Similar results were obtained by Sotelo et al. (2002) in their study of the kinetics of four chlorinated organics from water onto GAC. They found that the Langmuir equation did not represent the equilibrium data well for a monochlorobiphenyl while a better representation was found for a tetrachlorobiphenyl. The Freundlich isotherm tted well for TCB and HeCB. However, due to the empirical nature of the Freundlich equation it is dicult to discuss the estimated values of the parameters. The kf values found are rather low compared to the values reported by Sotelo et al. (2002), 387 and 10 mg/g for respectively 2-monochlorobiphenyl and 2,20 ,5,50 -tetrachlorobiphenyl while the n values on the other hand are higher 0.386 and 0.181 respectively. The validity of both Langmuir and Freundlich equations for HeCB may indicate that the apparent monolayer coverage is not reached in the concentration range studied. In order to elucidate the possible mechanism of chemisorption of HeCB, an additional study is needed to determine the interacting surface groups. As chlorosubstituents make aromatic compounds more electrophobic it is reasonable to propose that HeCB can adsorb at specic sites of the y ash. The activation energy (Ea ) values summarized in Table 4 for the two dierent temperature ranges indicate that these values are small for TCB and HeCB respectively in the temperature range of 525 C and 2555 C, but are high for HeCB in the temperature range of 525 C. Those low activation energies are characteristic for a diusion controlled process (Banerjee et al., 1997). Low activation energies (540 kJ/mol) are characterictic for physisorption, while higher activation energies (40800 kJ/mol) suggest chemisorption. The thermodynamic parameters such as free energy change (DG0 ), enthalpy change (DH 0 ) and entropy change (DS 0 ) were determined using Eqs. (13) and (14). The negative value of heat of adsorption (DH 0 ) suggests that the adsorption of TCB onto y ash is an exothermic process. The negative values of free energy change (DG0 ) indicate the spontaneous nature of the adsorption process. However, the negative DG0 value decreased with an increase in temperature, indicating that the spontaneous nature of adsorption is inversely proportional to the temperature. While the whole story for HeCB is vice versa. The negative values of DH 0 and DS 0 for TCB support the fact that the reaction is spontaneous at low temperature while the positive values of DH 0 and DS 0 for HeCB indicate spontaneous reaction at high temperatures where exothermicity is relatively unimportant (Zumdahl, 1992).

664

H. Nollet et al. / Chemosphere 53 (2003) 655665

The results obtained in this work indicate that the adsorption mechanism of each PCB congener must be taken into account to model the adsorption kinetics of the system, and agree with the complex behaviour of PCB mixtures observed in literature. The industry produces every year a huge amount of mineral byproducts. These byproducts rapidly accumulate and cause enormous problems of disposal unless a way can be found to utilize these byproducts through resource recovery programs. Nowadays a lot of eorts are made to limit, recycle or valorize those product streams. Fly ash has an annual worldwide production of about 600 million tonnes. Of this, 10% is used in the construction industry. The results discussed in this paper indicate that y ash can be used for the succesful and ecient removal of PCBs from wastewater. The main drawback of using y ash as an adsorbent for the removal of hydrophobic compounds may be the release of heavy metals from the particles when used in an aqueous medium. Leaching tests of metals from mixtures of y ash were performed by Albanis et al. (1998). As the y ash content and pH values increased, the concentrations of Mg, Mn and Fe decreased in the aqueous solutions by metal precipitation. The concentrations of Ni, Cu and Zn remained at low levels 0.010.07 mg/l without signicant changes with the y ash content. In contrast, Cr concentrations in leachates decreased with the y ash content due to the formation of more soluble metal species at high pH values. Although heavy metals are released from the y ash particles, the concentrations released remain low. Prewashing of the y ash in alkaline conditions may easily release metals as Cr and Cu. In conclusion, the use of y ash as an adsorbent could be an alternative for commercially available carbon due to its low cost and good eciency.

for TCB. The results indicate that next to physisorption for TCB and HeCB, chemisorption may play an important role in the adsorption process for HeCB. Negative values for DH 0 and DS 0 for TCB were found. This indicates that the reaction is spontaneous at low temperature while the positive values of DH 0 and DS 0 for HeCB means that the reaction is spontaneous at high temperatures.

Acknowledgements Useful discussions with Lieven De Kempeneer, Stefaan Dewildeman and Joris Roels are acknowledged. This work was nancially supported by the Flemisch government institution IWT (Institute for the Promotion of Innovation by Science and Technology in Flanders).

References
Akgerman, A., Zardkoohi, M., 1996. Adsorption of phenolic compounds on y ash. J. Chem. Eng. Data 41, 185187. Aksu, Z., Yener, J., 1999. The usage of dried activated sludge and y ash wastes in phenol biosorption/adsorption: Comparison with granular activated carbon. J. Environ. Sci. Health, Part A 34, 17771796. Albanis, T.A., Danis, T.G., Kourgia, M.G., 1998. Adsorption desorption studies of selected chlorophenols and herbicides and metal release in soil mixtures with y ash. Environ. Technol. 19, 2534. Banerjee, K., Cheremisino, P.N., Cheng, S.L., 1995. Sorption of organic contaminants by yash in a single solute system. Environ. Sci. Technol. 29, 22432251. Banerjee, K., Cheremisino, P.N., Cheng, S.L., 1997. Adsorption kinetics of o-xylene by yash. Water Res. 31, 249261. Gupta, V.K., 1998. Equilibrium uptake, sorption dynamics, process development, and column operations for the removal of copper and nickel from aqueous solution and wastewater using activated slag, a low-cost adsorbent. Ind. Eng. Chem. Res. 37, 192202. Gupta, V.K., Mohan, D., Sharma, S., Sharma, M., 2000. Removal of basic dyes (rhodamine B and methylene blue) from aqueous solutions using bagasse y ash. Separ. Sci. Technol. 35, 20972113. Gupta, V.K., Ali, I., 2001. Removal of DDD and DDE from wastewater using bagasse y ash, a sugar industry waste. Water Res. 35, 3340. Johnson, G.E., Kunka, L.M., Field, J.E., 1965. Use of coal and yash as adsorbents for removing organic contaminants from secondary municipal euents. I & EC Process Des. Develop. 4, 323327. Kao, P.C., Tzeng, J.H., Huang, T.L., 2000. Removal of chlorophenols from aqueous solution by y ash. J. Hazard. Mater. 79, 237249. Kumar, S., Upadhyay, S.N., Upadhya, Y.D., 1987. Removal of phenols by adsorption on y ash. J. Chem. Technol. Biotechnol. 37, 281290.

5. Conclusions Based upon the results of this research, the following conclusions can be drawn: Fly ash is found to be eective for the removal of PCBs from an aqueous solution. The kinetics of the adsorption of TCB and HeCB by y ash can be approximated by rst-order reversible kinetics. The eect of the initial concentration of the solute on the rate of reaction was found to be of considerable signicance. The uptake rate constant of TCB and HeCB increases non-linearly with the increasing initial concentration and gradually approaches a plateau. The adsorption capacity for the higher chlorinated PCB, HeCB, is higher than the adsorption capacity

H. Nollet et al. / Chemosphere 53 (2003) 655665 Mancy, K.H., Gates, W.E., Eye, J.D., Deb, P.K., 1964. The adsorption kinetics of ABS on yash. In: Bloodgood, D.E. (Ed.), Proceedings 19th Industrial Waste Conference. Purdue University, W. Lafayette, IN, pp. 146 160. Manes, M., Hofer, L.J.E., 1969. Application of the Polanyi adsorption potential theory to adsorption from solution on activated carbon. J. Phys. Chem. 73, 584590. Mott, H.V., Weber, W.J., 1992. Sorption of low molecular weight organic contaminants by yash: consideration of enhancement of cut-o barrier performance. Environ. Sci. Technol. 26, 12341242. Nollet, H., Lutgen, P., Verstraete, W., 2002. Chemical removal of PCBs from water samples under ambient conditions. J. Chem. Technol. Biotechnol. 77, 517524. Pirbazari, M., Weber, W.J., 1981. In: Cooper, N.J. (Ed.), Chemistry Water Reuse, vol. 2. Ann Arbor Science, USA, pp. 309339. Peloso, A., Rovatti, M., Ferraiolo, G., 1983. Fly-ash as adsorbent material for toluene vapors. Resour. Conserv. 10, 211220. Ricou-Hoeer, P., Lecuyer, I., Le Cloirec, P., 2001. Experimental design methodology applied to adsorption of metallic ions onto y ash. Water Res. 35, 965976. Roy, W.R., Thiery, R.G., Schuller, R.M., Suloway, J.J., 1981. Coal yash: a review of the literature and proposed classication system with emphasis on environmental impact. Environ. Geol. Notes 16, 143. Ruthven, D.M., 1984. Principles of Adsorption and Adsorption Process. John Wiley and Sons, New York.

665

Sakellarides, T.M., Albanis, T.A., 2000. A new organophosphorus insecticides removal process using y ash. Int. J. Environ. Anal. Chem. 78, 249262. Sotelo, J.L., Ovejero, G., Delgado, J.A., Martinez, I., 2002. Comparison of adsorption equilibrium and kinetics of four chlorinated organics from water onto GAC. Water Res. 36, 599608. Stach, J., Pekarek, V., Endrst, R., Hetejs, V., 1999. Dechlorination of hexachlorobenzene on MWI y ash. Chemosphere 39, 23912399. Stach, J., Pekarek, V., Grabic, R., Lojkasek, M., Pacakova, V., 2000. Dechlorination of polychlorinated biphenyls, dibenzop-dioxins and dibenzofurans on y ash. Chemosphere 41, 18811887. Van Agteren, M.H., Keuning, S., Janssen, D.B., 1998. Handbook on Biodegradation and Biological Treatment of Hazardous Organic Compounds. Kluwer Academic Publishers. Viraraghavan, T., Alfaro, F., 1994. Removal of phenol from wastewater by adsorption on peat, yash and bentonite. In: Huang, C.P. (Ed.), Proceedings 26th Mid Atlantic Industrial Waste Conference, Newark, Delaware. Technomic Publishing Co., Lancaster, PA, pp. 311317. Weber, W.J., 1972. Physicochemical Processes for Water Quality Control. John Wiley and Sons, New York. World Health Organization, 1993. Polychlorinated biphenyls and terphenyls. Environmental Health Criteria 140, second ed. World Health Organization, Geneva, Switzerland. Zumdahl, S.S., 1992. Energy, enthalpy and thermochemistry. In: Chemical Principles. D.C. Heath and Company, Lexington, MA, Toronto, pp. 325424.