Sei sulla pagina 1di 14

Available online at www.sciencedirect.

com

Separation and Purication Technology 61 (2008) 229242

Review

Biosorption isotherms, kinetics and thermodynamics


Yu Liu , Ya-Juan Liu
Division of Environmental and Water Resources Engineering, School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore Received 17 July 2007; received in revised form 27 September 2007; accepted 3 October 2007

Abstract Biosorption, as a cost-effective technology for the removal of soluble heavy metals and organics from aqueous solutions, has been extensively studied, and most biosorption research mainly focused on the process isotherms, kinetics and thermodynamics. Thus, this paper attempted to offer a better understating of representative biosorption isotherms, kinetics and thermodynamics with special focuses on theoretical approaches for derivation of combined LangmuirFreundlich isotherm as well as the pseudo-rst- and second-order kinetic equations and general rate law equation for biosorption. Meanwhile, some potential problems encountered in biosorption research were also discussed. 2007 Elsevier B.V. All rights reserved.
Keywords: Biosorption; Isotherm; Kinetics; Thermodynamics; Equilibrium; Rate law

Contents
1. 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biosorption isotherms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Langmuir isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Freundlich isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Sips isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1. Derivation from an equilibrium approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.2. Derivation from a thermodynamic approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. RedlichPeterson isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Khan isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. T th isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . o 2.7. RadkePrausnitz isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.8. DubininRadushkevich isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.9. Frumkin isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.10. FloryHuggins isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.11. BET isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.12. Temkin isotherm equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biosorption kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Pseudo-rst- and second-order equations for biosorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1. Pseudo-rst-order equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2. Pseudo-second-order equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Derivation of pseudo-rst- and second-order equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Approach by Boyd et al. [50] for derivation of pseudo-rst-order equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. Approach by Blanchard et al. [49] for derivation of pseudo-second-order equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. Approach by Liu et al. [46] for derivation of pseudo-rst-order equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 230 230 230 230 231 231 232 232 233 233 233 233 233 234 234 234 234 234 234 234 234 235 235

3.

Corresponding author. E-mail address: cyliu@ntu.edu.sg (Y. Liu).

1383-5866/$ see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.seppur.2007.10.002

230

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

4. 5.

3.2.4. Approach by Azizian [51] for derivation of pseudo-rst- and second-order equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.5. Approach by Rudzinski and Plazinski [48] for derivation of pseudo-rst- and second-order equations . . . . . . . . . . . . . . 3.3. A general rate law equation for biosorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1. Model development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2. Some considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Other useful kinetic equations for biosorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.1. Elovich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.2. WeberMorris equation or intraparticle diffusion equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biosorption thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Some concerns about biosorption research and application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Concern about kinetic and isotherm study of biosorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Concern about thermodynamic study of biosorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. Concern about application of biosorption technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

235 237 237 237 239 239 239 239 240 240 240 240 241 241

1. Introduction As heavy metals are toxic, nonbiodegradable and can be accumulated in living organisms, heavy metal pollution is becoming a serious environmental problem. To date, the amount of heavy metals discharged into the environment keeps on increasing. In the past decades, one has been looking for inexpensive technologies to control metal pollution. As a cost-effective mean, biosorption has been extensively studied for removing a wide variety of soluble heavy metals as well as some toxic organics from aqueous solutions, and many biomaterials have been tested as biosorbents including marine algae, fungal biomass, waste activated sludge, digested sludge, aerobic granules and so on [111]. So far, extensive research effort has been dedicated to a sound understanding of biosorption isotherm, kinetics and thermodynamics. Compared to biosorption isotherm, there is lack of a theoretical basis behind the kinetic description of biosorption data. In this regard, pseudo-rst- and second-order kinetic equations have been widely used to describe time evolution of biosorption under nonequilibrium conditions [10,1216]. Ones often choose pseudo-rst- and second-order kinetic equations without any rational. In such a circumstance, this paper reviewed the research state of the art of biosorption isotherms, kinetics and thermodynamics. 2. Biosorption isotherms In this section, some representative biosorption isotherm equations were discussed with a special focus on theoretical derivation of the combined LangmuirFreundlich isotherm equation, also known as the Sips isotherm. 2.1. Langmuir isotherm equation Langmuir [17] theoretically examined the adsorption of gases on solid surfaces, and considered sorption as a chemical phenomenon. Basically, the Langmuir isotherm equation has a hyperbolic form: Qe = Qe th Keq Ce 1 + Keq Ce (1)

in which Qe is the adsorption capacity by weight at equilibrium, Qe is the theoretical maximum adsorption capacity by weight, th and Keq represents the equilibrium constant of adsorption reaction, while Ce is concentration of adsorbate at equilibrium. This isotherm equation has been most frequently applied in equilibrium study of biosorption, however, it should be realized that the Langmuir isotherm offers no insights into the mechanism aspects of biosorption. 2.2. Freundlich isotherm equation Freundlich [18] proposed an empirical isotherm equation:
1/n Qe = kF Ce F

(2)

in which kF and nF are Freundlich constants. As the Freundlich isotherm equation is exponential, it can only be reasonably applied in the low to intermediate concentration ranges. Similar to the Langmuir isotherm equation, Eq. (2) has also been widely employed in biosorption research. 2.3. Sips isotherm equation In study of the distribution of adsorption energies of the sites of a catalyst surface, Sips [19] proposed an empirical isotherm equation which is often expressed as Qe = Qe th
n Keq Ce s n 1 + Keq Ce s

(3)

in which ns is the Sips constant. At the time of proposing the above empirical isotherm, Sips [19] noted that we do not know whether or not this form of isotherm actually represents any experimental results. Based on a comparison study of a number of isotherm equations applied to the sorption of methylene blue on lemon peel, Kumar and Porkodi [20] concluded that the Sips isotherm provided the best tting of experimental data, followed by the Langmuir and the Redlich Peterson isotherm equations. In the following sections, two approaches were developed for derivation of the empirical Sips equation.

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

231

2.3.1. Derivation from an equilibrium approach The overall biosorption reaction is often described by nA + B An B (4)

follows: Qe = Qth e and Qmax = Qth max (10)

in which A is the adsorbate, B the biosorbent, and An B represents A and B complex. According to Hammes [21], for Eq. (4), the equilibrium constant (Keq ) can be written as follows: Keq = [An B] [A]n [B] (5)

in which is an overall limitation coefcient of biosorption process. Hence, Eq. (9) becomes Qe = Qmax Keq [A]n 1 + Keq [A]n (11)

The denition of equilibrium constant shows that the terms [An B], [A] and [B] must be expressed as molar concentration. Thus, number of the moles of adsorbate bound to per mole of biosorbent (q) can be expressed in a way such that n[An B] q= [B] + [An B] If Eq. (5) is inserted into Eq. (6), we obtain q=n Keq [A]n 1 + Keq [A]n (7) (6)

Following common symbol used in study of biosorption equilibrium, [A] is placed by Ce representing molar concentration of adsorbate at equilibrium (mol l1 ), thus Eq. (11) turns into Qe = Qmax
n Keq Ce n 1 + Keq Ce

(12)

Obviously, Eq. (12) is the same as the Sips isotherm equation (Eq. (3)). 2.3.2. Derivation from a thermodynamic approach Liu et al. [26] developed the Sips equation from a thermodynamic approach. The overall biosorption reaction could be regarded as a simple change in the state of adsorbate [27], i.e., C Cads G (13)

Eq. (7) shows that the theoretical maximum biosorption capacity is n (mol mol1 ). Multiplying both sides of Eq. (7) by the ratio of molar weight of adsorbate (MA ) to molar weight of biosorbent (MB ) results in 1000MA Keq [A]n 1000MA q=n MB MB 1 + Keq [A]n (8)

Let the term (1000MA /MB )q be Qth , and the term e n(1000MA /MB ) be Qth , then Eq. (8) becomes max Qth = Qth e max Keq [A]n 1 + Keq [A]n (9)

in which C and Cads is molar concentration of adsorbate in bulk solution, and that adsorbed at time t, while G is the effective free energy change of biosorption, which changes with the proceeding of the biosorption reaction. If the adsorbate concentration (C) in bulk solution increases, its adsorption is more favorable, i.e. G would decrease with the increase of the metal concentration [22]. According to Morel and Hering [27], G can be expressed as G = G nL RT ln C (14)

in which Qth is theoretical biosorption capacity by weight e (mg g1 ), and Qth is the theoretical maximum biosorption max capacity by weight (mg g1 ). Eq. (9) is derived purely according to the equilibrium law of a chemical process without any consideration of process limitation. It should be recognized that biosorption is a very complex process, while the binding sites of biosorbent are not all identical, i.e., as biosorption proceeds, the driving force of biosorption will be reduced, i.e., the resistance force of biosorption would increase [22]. These imply that interference between each binding site exists during biosorption [21]. Moreover, resistances due to restricted space (geometric factors) and afnity of adsorbate and biosorbent, external and internal mass transfer of soluble adsorbate from aqueous solution to biosorbent would also limit the biosorption capacity [5,2325]. As noted by Hammes [21], actually observed biosorption capacity would be much lower than the value predicted theoretically. Thus, for practical application, Eq. (9) needs to be corrected in accounting for the limitations that may be encountered in biosorption process. In this case, the observed biosorption capacity (Qe ) and actual maximum biosorption capacity (Qmax ) are dened as

in which nL is the positive coefcient. Evidence shows that the real driving force of biosorption is the difference between the amount adsorbed by unit biosorbent (Q) at a given adsorbate concentration and the theoretical amount that could be adsorbed by unit biosorbent at that concentration (Qth ), and this driving force is disappearing when the biosorption reaction gradually approaches its equilibrium state [22]. As biosorption proceeds, the driving force decreases and the adsorption resistance will increase. It is a reasonable consideration that the overall change of free energy of the biosorption process ( G) would increases with the increase of adsorption resistance, and decreases with the increase of the driving force of adsorption reaction. Liu et al. [26] proposed that the overall change of free energy of the biosorption reaction should be formulated as the function of the driving force and resistance of adsorption such that G= G + RT ln Resistance Driving force (15)

In a theoretical sense, Eq. (15) is indeed consistent with the expression for free energy change of an ideal gas and solution [27]. As pointed out earlier, the adsorption reaction becomes less favorable as the adsorption proceeds, i.e., G must increase

232

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

accordingly. These seem to imply that Q would reect the magnitude of adsorption resistance. On the other hand, the difference between Qth and Q represents the actual driving force of the biosorption process, a larger difference leads to a smaller value of G. Therefore, Eq. (15) can be written as follows: G= G + RT ln Q Qth Q (16)

Eq. (16) shows that when Q = 0.5Qth , Go is equal to G. This in turn implies that Go can be dened as the overall free energy change at Q = 0.5Qth , i.e. the driving force of biosorption is equal to the resistance force. As Q approaches Qth , G goes to innity and further adsorption becomes energetically impossible, this is indeed in agreement with that stated by Morel and Hering [27]. Substitution of Eq. (14) into Eq. (16) yields G= G nL RT ln C + RT ln Q Qth Q Qe Qe (17)

Fig. 1. Biosorption isotherm of uranium on Sargassum biomass. Eq. (20) prediction is shown by solid curve. Qe = 1.38 mmol g1 ; Kads = 0.24; nL = 0.73 th and R = 0.99, data from [74].

When the adsorption reaches its equilibrium, 0= G nL RT ln Ce + RT ln Qe th Qe th

G is zero. Hence (18)

Eq. (24) shows that the Langmuir adsorption isotherm is only a n particular case of Eq. (20) at nL = 1. When Ce L is much less than Kads , Eq. (20) is thus simplied to the Freundlich-type isotherm: Qe = Qe nL th C Kads e (25)

in which Ce , Qe and are the respective value of C, Q and Qth at equilibrium. Solving Eq. (18) for Qe gives Qe = Qe th e
Go /RT n Ce L n + Ce L

(19)

Eq. (19) can be rearranged as Qe = Qe th in which Kads = e


Go /RT n Ce L n Kads + Ce L

(20)

It appears that the Freundlich constant indeed is equal to Qe /Kads . It should be pointed out that the Freundlich isotherm th equation has been referred to as an empirical formula, and two constants involved have no clearly dened physical meanings. Eq. (25) indeed provides a theoretical basis for better interpreting the empirical Freundlich isotherm equation. Consequently, Eq. (20) can be regarded as a generalized form of the Langmuir and Freundlich models. 2.4. RedlichPeterson isotherm equation

(21)

In fact, Eq. (20) has the same formulation as the Sips isotherm equation (Eq. (3)). Analogue to a chemical reaction, the thermodynamic equilibrium constant of biosorption reaction (Keq ) can be dened as Keq = eG
o /RT

Similar to the Sips isotherm equation, Redlich and Peterson [28] proposed an isotherm compromising the features of the Langmuir and the Freundlich isotherms: Qe = Krp Ce 1 + rp Ce

(26)

(22)

Comparison of Eqs. (21) and (22) shows that Kads = 1 Keq (23)

Eq. (23) reveals the real physical meaning of Kads . It should be pointed out that substitution of Eq. (23) into Eq. (20) leads to Eq. (12). It is demonstrated here that the Sips isotherm equation can also be obtained from the thermodynamic approach as discussed above. An example of application of Eq. (20) was presented in Fig. 1. When nL equals 1, Eq. (20) is reduced to the Langmuir adsorption isotherm: Qe = Qe th Ce Kads + Ce (24)

in which Krp and rp are the RedlichPeterson constants, and is basically in the range of zero to one. If is equal to 1, Eq. (26) reduces to the Langmuir isotherm equation, while in case where the value of the term rp Ce is much bigger than one, the RedlichPeterson isotherm equation can be approximated by a Freundlich-type equation. It has been reported that the RedlichPeterson isotherm equation described Zn2+ biosorption by Rhizopus arrhizus very well [29]. 2.5. Khan isotherm equation Khan et al. [30] proposed an isotherm equation to describe adsorption of aromatics by activated carbon, which can be expressed in the following simplied form: Qe = Qmax bK Ce (1 + bK Ce )nK (27)

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

233

in which bK and nK are two constants. Similar to the Sips and RedlichPeterson isotherm equations, the Khan isotherm equation also reects the combined feature of the Langmuir and Freundlich isotherm equations. For example, if nK equals 1, Eq. (27) reduces to the Langmuir isotherm, whereas Eq. (27) can be simplied to a Freundlich-type isotherm when value of the term bK Ce is much greater than unity. 2.6. T th isotherm equation o This isotherm equation (Eq. (28)) was derived from the potential theory, and has been widely employed to describe biosorption on heterogeneous biosorbent surface [31]: Qe = Qmax bT Ce [1 + (bT Ce )1/nT ]nT (28)

adsorption energy (E) through E = 1/ 2KDR , and is the Polanyi potential and is further dened by = RT ln 1 + 1 Ce (34)

Insert Eq. (34) into Eq. (33) and take natural log of both sides gives ln Qe = ln QDR KDR R2 T 2 ln2 1 + 1 Ce (35)

It is apparent that the DubininRadushkevich isotherm equation would not be simplied to the Langmuir or Freundlich type of isotherm. Furthermore, plot of ln Qe versus ln2 (1+(1/Ce )) should lead to a straight line, and its slope is given by Slope = KDR R2 T 2 (36)

in which bT and nT are two constants. Obviously, if nT = 1, Eq. (28) reduces to the Langmuir-type isotherm equation, but the T th isotherm equation could not reect the feature of the o Freundlich-type biosorption. It has been reported that Eq. (28) could reasonably describe the Ni2+ biosorption by Sargassum wihtii [32]. 2.7. RadkePrausnitz isotherm equation The RadkePrausnitz isotherm equation is often expresses as follows [33]: Qe = aR bR
Ce R aR + bR Ce R 1

Thus, the mean adsorption energy (E) can be calculated as follows: 1 RT E= (37) = 2 slope 2KDR Eq. (37) shows how E can be obtained from the Dubinin Radushkevich isotherm equation. In general, the Dubinin Radushkevich isotherm equation has been often used to determine the mean adsorption energy (E) that may provide useful information with regard to whether or not biosorption is subject to a chemical or physical process [32,3537]. 2.9. Frumkin isotherm equation The Frumkin isotherm equation was developed in taking into account of the interaction between the adsorbed species. According to Grchev et al. [38], The Frumkin isotherm can be expressed as follows: (38) exp(f) = KF Ce 1 in which is the coverage degree of adsorbent surface (0 1). Replacement of by its denition formula [35], Eq. (38) can be further translated to the following linearized form: ln Qe QF Q e 1 Qe = ln KF + 2f Ce QF (39)

(29)

in which aR , bR and R are all constants. Eq. (29) can be rearranged to Qe = aR


Ce R KR + Ce R 1

in which KR =

aR bR

(30)

For KR Ce R 1 , Eq. (30) becomes a Freundlich-type isotherm equation:

Qe =

aR R C KR e
Ce R 1 , Eq. (30) can simplied to

(31)

For KR

Qe = aR Ce

(32)

It was shown that the RadkePrausnitz isotherm equation (Eq. (29)) poorly described the Ni2+ biosorption [32]. 2.8. DubininRadushkevich isotherm equation Dubinin and Radushkevich [34] developed the following isotherm in accounting for the effect of the porous structure of an adsorbent: Qe = QDR exp(KDR 2 ) (33)

in which QF is the theoretical monolayer saturation capacity as given in the DubininRadushkevich isotherm equation (QDR ), f is the interaction coefcient and KF is the equilibrium constant. If f = 0, i.e., there is no interaction between adsorbate species, Eq. (39) reduces to the Langmuir-type isotherm. Eq. (39) has been applied for dyes adsorption on waste apricot-based activated carbon waste apricot [35]. 2.10. FloryHuggins isotherm equation The original FloryHuggins isotherm describes the behavior of a two-dimensional lattice of non-interacting particles of dif-

in which QDR is the DubininRadushkevich constant representing the theoretical monolayer saturation capacity [35], KDR is the constant of the adsorption energy which is related to mean

234

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

ferent sizes, i.e., it accounts for the effect of the surface coverage on adsorption [39,40]: exp(2nFH FH ) = KFH Ce nFH (1 )nFH (40)

in which KFH is the so-called equilibrium constant dened by FloryHuggins isotherm, nFH is constant, and FH is an effective constant indicating the interaction between adsorbed molecules [41]. It can be seen that Eq. (40) turns into the Frumkin isotherm if nFH = 1. Vijayaraghavan et al. [32] applied the FloryHuggins isotherm equation to determine the equilibrium constant (KFH ), which was further used to estimate the change in the Gibbs free energy during Ni2+ biosorption, while Horsfall and Spiff [42] also used this equation to study the equilibrium sorption of Al3+ , Co2+ and Ag+ by uted pumpkin waste biomass. 2.11. BET isotherm equation Brunauer et al. [43] developed an equation for multimolecular adsorption, so-called the BET equation: Qe = BCe Q0 (Cs Ce ) (1 + (B 1)Ce )/Cs (41)

kinetic studies, both pseudo-rst- and second- kinetic equations have been commonly employed in parallel, and one is often claimed to be better than another according to marginal difference in correlation coefcient. As noted by Rudzinski and Plazinski [48], in the past decades no attempts were made to clearly explain the theoretical origins of these two equations, i.e., current understanding of biosorption kinetics is much less than theoretical description of biosorption equilibrium. 3.1.1. Pseudo-rst-order equation The pseudo-rst-order kinetic equation or the so-called Lagergren equation has the following formulation: dQt = k1 (Qe Qt ) dt (43)

in which Qt is the amount of adsorbate adsorbed at time t, Qe is its value at equilibrium and k1 is a constant. The pseudo-rstorder Lagergren equation is indeed in line with the concept of linear driving force. 3.1.2. Pseudo-second-order equation The pseudo-second-order kinetic equation was rst proposed by Blanchard et al. [49], and since then it has been frequently employed to analyze biosorption data obtained from various experiments using different adsorbates and biosorbents as reviewed by Ho et al. [12]: dQt = k2 (Qe Qt )2 dt in which k2 is a constant. 3.2. Derivation of pseudo-rst- and second-order equations This section discussed some approaches in chronological order for derivation of the pseudo-rst- and second-order kinetic equations for biosorption (Eqs. (43) and (44)). 3.2.1. Approach by Boyd et al. [50] for derivation of pseudo-rst-order equation Boyd et al. [50] developed a rate equation, which was based on the exchange sorption of ions from aqueous solution by organic zeolites. For the case of two monovalent ions, the exchange reaction can be expressed as follows: A+ + BR B+ + AR (45) (44)

in which Q0 is the amount of solute adsorbed per unit weight of adsorbent in forming a complete monolayer on the surface, B is a constant relating to the energy of interaction with the surface and Cs is the saturation concentration of the solute. Preetha and Viruthagiri [29] applied the BET model to describe biosorption of Zn2+ by Rhizopus arrhizus. 2.12. Temkin isotherm equation This isotherm was rst developed by Temkin and Pyzhev [44], and it is based on the assumption that the heat of adsorption would decrease linearly with the increase of coverage of adsorbent [45]: Qe = RT ln(at Ce ) bt (42)

in which R is the gas constant, T the absolute temperature in Kelvin, bt the constant related to the heat of adsorption and at is the Temkin isotherm constant. Although the Temkin isotherm equation has been applied to describe adsorption on heterogeneous surface, Vijayaraghavan et al. [32] reported that Eq. (42) could not satisfactorily t the data of Ni2+ biosorption by Sargassum wightii. 3. Biosorption kinetics 3.1. Pseudo-rst- and second-order equations for biosorption Within the scope of the literature review, two kinetic models, namely pseudo-rst- and second-order equations, have been widely used to describe biosorption data obtained under nonequilibrium conditions [10,1216,47]. In most biosorption

in which A and B are free metal ion, and BR and AR represent metal ion-adsorbent complex. If mA+ and mB+ denote the concentrations of the ions A+ and B+ in solution, while nAR and nBR represent respective numbers of moles of A+ and B+ in the adsorbent, the reaction rate can be written as dnAR = kE1 mA+ nBR kE2 mB+ nAR dt = nAR (kE1 mA+ + kE2 mB+ ) + kE1 mA+ E

(46)

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

235

in which kE1 and kE2 are the forward and reverse rate constants and E is a constant dened by E = nAR + nBR (47)

quantity of adsorbate removed per unit volume of solution. A mass balance on metal ions gives Cb = C0 C (54)

When the concentrations of A+ and B+ in solution are constant, integration of Eq. (46) leads to nAR = kE1 mA+ E (1 ekb t ) = Qt kE1 mA+ + kE2 mB+ (48)

in which C0 is initial concentration of adsorbate. When biosorption reaches its equilibrium, Eq. (54) becomes Cbe = C0 Ce At biosorption equilibrium, Eq. (53) reduces to k Cbe = a Ce kd (56) (55)

in which Qt is the adsorption capacity at time t, kb = kE1 mA+ + kE2 mB+ . Rearrangement of Eq. (48) yields Qt = Qe (1 e
kb t

(49)

in which Qe is the biosorption capacity at equilibrium. Apparently Eq. (49) is similar to the pseudo-rst-order kinetic equation (Eq. (43)). 3.2.2. Approach by Blanchard et al. [49] for derivation of pseudo-second-order equation In study of heavy metal removal by natural zeolites, Blanchard et al. [49] proposed that the overall exchange reaction of ammonium xed in zeolites by divalent metal ions (M2+ ) in solution can be described by Z(2NH+ ) + M2+ Z(M2+ ) + 2NH+ 4
4

in which Cbe and Ce are apparent concentration of adsorbed metal ions and concentration of free metal ions at biosorption equilibrium, respectively. Combination of Eqs. (53)(56) yield dC = (ka + kd )(C Ce ) dt (57)

Integration of Eq. (57) shows that C0 C e = ek1 t C Ce In Eq. (58), the term k1 is dened by k1 = k a + k b (59) (58)

(50)

In this approach, it was assumed that the metal ion concentration in solution varies very slightly during the rst hour and that the kinetic order is two with respect to the number (n0 nt ) of available sites for the exchange. Thus, the differential equation can be written as follows: dnt = k2 (n0 nt )2 dt (51)

in which represents the overall biosorption rate constant. Substitution of Eqs. (54) and (55) into Eq. (58) leads to Cb = Cbe ek1 t 1 ek1 t (60)

Dividing both sides of Eq. (60) by biosorbent concentration yields Qt = Qe (1 ek1 t ) (61)

in which nt is amount of M2+ exchanged at time t, no is exchange capacity and k2 is rate constant. Mathematically, Eq. (51) is the same as the well-known pseudo-second-order equation (Eq. (44)). 3.2.3. Approach by Liu et al. [46] for derivation of pseudo-rst-order equation Liu et al. [46] proposed that biosorption process is subject to the rst-order reversible kinetics that can be described as S+A
ka kd

In fact, Eq. (61) is the integrated form of the differential pseudo-rst-order equation as given in Eq. (43). It should be realized that in the sense of chemical reaction, the reaction orders of forward and reverse reactions involved in biosorption process cannot be simply preset to 1, and they need to be experimentally determined unless the complex mechanisms of adsorption process are known. 3.2.4. Approach by Azizian [51] for derivation of pseudo-rst- and second-order equations Based on a reversible adsorptiondesorption process, Azizian [51] proposed that the adsorption and desorption processes of soluble adsorbate can be depicted as follows: A + A(a) (62)

SA

(52)

in which S is the available site of biosorbent for biosorption, A is soluble adsorbate in solution, and SA represents adsorbatebiosorbent complex, while ka and kb are the respective rate constants for the adsorption and desorption processes. In this case, the overall biosorption rate is written as dC = ka C k d C b dt (53)

in which * represents the available site for adsorption. Based on Eq. (62), Azizian [51] further thought that the adsorption and desorption rates can be expressed as ra = ka C(1 t ) rd = kd t (63)

in which C is concentration of adsorbate at time t, Cb is apparent concentration of adsorbed adsorbate at time t in terms of the

236

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

in which ra and rd are adsorption and desorption rates, respectively, ka and kd are corresponding rate constants. Thus, the overall rate equation for adsorption is dt = ra rd = ka C(1 t ) kd t dt (64)

in which a = ka , = ka ( + C0 + 1/K)2 4C0 , K = ka /kd

= ( + C0 + 1/K)ka , = ( + C0 + 1/K)ka + , (74)

Change in concentration of adsorbate in solution can be described as follows: C = C0 t in which is further denes in a way such that = XQmax MA V (66) (65)

Azizian [51] thought that if the term t in Eq. (73) is very small, then the term exp(t) could be approximated by et 1 + t (75) Since t = Qt /Qmax and e = Qe /Qmax , Eq. (73) can be nally rearranged to t 1 1 = + t 2 Qt k 2 Qe Qe in which k2 = 2Qe (77) (76)

in which X is the mass of adsorbent used, Qmax is the maximum adsorption capacity, MA is molar mass of adsorbate and V is the volume of solution. Substitution of Eq. (65) into (64) gives dt = ka (C0 t )(1 t ) kd t dt (67)

Azizian [51] used Eq. (67) as a basis to derive the pseudo-rstand second-order equations as elaborated below. 3.2.4.1. Derivation of pseudo-rst-order equation. If initial adsorbate concentration is much higher than t , i.e., C0 t , Eq. (67) is reduced to dt = ka C0 (ka C0 + kd )t dt Integration of Eq. (68) leads to ln 1 t e = k1 t (69) (68)

In fact, Eq. (76) is the so-called pseudo-second-order equation for adsorption. 3.2.4.3. Some considerations. As concluded by Azizian [51], the sorption process obeys pseudo-rst-order kinetics at high initial concentration of solute, while it obeys pseudo-secondorder kinetic model at lower initial concentration of solute. However, such conclusions may be debatable as elaborated below. It should be realized that an important assumption indeed is hidden in Eq. (63), i.e., the adsorption was assumed to be the rst order with respect to the concentration of adsorbate (C) and available adsorption sites (1 t ), respectively, while desorption was assumed to be a rst order regarding the sites occupied ( t ). These seem to imply that the overall kinetic order of reversible adsorption process is restricted to a range of 12. More importantly, the reaction orders of forward and reverse reactions (Eq. (63)) cannot be simply preset to 1 unless the complex mechanisms of adsorption process are known. In fact, a fundamental challenge in chemical kinetics is the determination of the reaction order from experimental data. It is well known that the rate law is closely related to the reaction mechanism, and the reaction stoichiometry does not determine the reaction order except in the special case of an elementary reaction [52]. As shown in Azizians approach, Eq. (69) or Eq. (71) is obtained by assuming C0 t . However, under such a circumstance, Eq. (65) would become C = C0 . This implies that the pseudo-rst-order equation (Eq. (69) or Eq. (71)) derived by Azizian [51] would be valid only for a pure solution system, which is not common in adsorption study. Thus, it is apparent that the Azizians approach for derivation of the pseudo-rst-order kinetic equation for adsorption would be still debatable. In derivation of the pseudo-second-order kinetic equation (Eq. (76)), the term t was assumed to be very small, which led to Eq. (75). As is a constant dened by Eq. (74), the above assumption can be translated into another expression of that t should be very small. This seems to indicate that the pseudosecond-order kinetic equation (Eq. (76)) derived from Azizians

in which k1 is rate constant and is given by k1 = C0 ka + kd (70) Eq. (70) clearly shows that k1 is linearly related to initial adsorbate concentration in solution. It is known that t Qt = (71) e Qe Thus, Eq. (69) can be rewritten to ln 1 Qt Qe = k1 t (72)

Eq. (72) indeed is the integrated form of the pseudo-rst-order equation (Eq. (44)). 3.2.4.2. Derivation of pseudo-second-order equation. Azizian [51] further proposed that if initial adsorbate concentration in solution is not too high as compared to the term t , the term t cannot be negligible. In this case, Eq. (67) is directly integrated for derivation of the kinetic equation for adsorption. After integration, following expression was obtained by Azizian [51]: t = (et 1) 2a( et ) (73)

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

237

approach would be valid only for initial stage of adsorption at which t can be reasonably assumed to be small. Therefore, Eq. (76) is applicable only for initial stage of adsorption. In addition, Eq. (76) was obtained by directly integrating the basic equation (Eq. (67)) under the sole assumption that t is very small, thus the assumption that initial concentration of solute was not too high for the term t is not necessary for derivation of Eq. (76). Consequently, these seem to indicate that the conclusion of the sorption process obeys the pseudo-second-order model at lower initial concentration of solute drawn by Azizian [51] is still debatable. Eq. (67) is the basic equation in Azizians approach, and it can be rearranged as follows: dt = ka t2 (ka C0 + ka + kd )t + ka C0 dt (78)

To simplify Eq. (82), it is reasonable to assume that at equilibrium, te would be close 1 with regard to the total available site for adsorption under given conditions, thus Eq. (82) reduces to 1 1 dt = 2(1 t ) + 3 (1 t )3 + 5 (1 t )5 + . . . dt 3 60 (84) Rudzinski and Plazinski [48] thought that if values of the second and third terms on the right-hand side of Eq. (84) are much smaller than that of the rst term, Eq. (84) is reduced to dt = 2(1 t ) dt Substitution of Eqs. (83)(80) into Eqs. (85)(82) leads to dQt = k1 (Qe Qt ) dt (86) (85)

As 0 < t < 1, t2 is smaller than t , while value of the term ka is much less than value of the combined term (ka C0 + ka + kd ) in Eq. (78). It is thus apparent that Eq. (78) would be reasonably approximated by dt ka C0 (ka C0 + ka + kd )t dt (79)

This equation is the same as the pseudo-rst-order kinetic equation (Eq. (43)), in which k1 = 2. If value of the third term on the right-hand side of Eq. (84) is negligible compared to those of the rst two terms, Eq. (84) can be approximated by dt 1 = 2(1 t ) + 3 (1 t )3 dt 3 (87)

Eq. (79) is truly valid especially at small t or t. Now it becomes clear that the basic equation (Eq. (67)) proposed by Azizian [51] indeed is much closer to the rst-order kinetics rather than other orders. Without any further assumptions, direct integration of Eq. (79) also yields the following pseudo-rstorder equation: ln 1 Qt Qe = k1 t (80)

in which k1 is the rst-order rate constant, but dened as k1 = ka C0 + ka + kd (81)

Eq. (87) shows a hybrid-order between the rst and the third power. Rudzinski and Plazinski [48] thought that the secondorder kinetic equation (Eq. (44)) might simulate such a hybrid behavior, and they believed that this would explain the theoretical background of the pseudo-second-order kinetic equation for biosorption. Nevertheless, mathematically Eq. (87) cannot be readily attributed to a second-order kinetic model. To further look into this point, the term 1 t in Eq. (84) is dened as t = 1 t Substituting Eq. (88) into Eq. (87) gives 1 dt = 2t + 3 3 t dt 3 (89) (88)

Eq. (81) represents the general formulation of the rst-order rate constant for adsorption. In a special case where C0 , Eq. (81) would reduce to Eq. (70). 3.2.5. Approach by Rudzinski and Plazinski [48] for derivation of pseudo-rst- and second-order equations In order to understand the possible theoretical foundation of the pseudo-rst- and second-order kinetic model for adsorption, Rudzinski and Plazinski [48] proposed following equation by introducing the statistical rate theory of interfacial transport: dt 1 1 = 2(te t ) + 3 (te t )3 + 5 (te t )5 + . . . dt 3 60 (82) in which can be regarded as constant coefcient, t and te are the fraction of surface sites occupied by adsorbate at time t and equilibrium, respectively. Furthermore, in Eq. (82), t is dened as t = Qt Qe (83)

As shown in Eq. (88), t is than 1, while in biosorption process, t decreases over time. Thus, compared to t , 3 would t be negligible, i.e., Eq. (89) indeed is very close to the rstorder kinetics as described by Eq. (85). It is apparent that the approach by Rudzinski and Plazinski [48] can explain the theoretical origin of the pseudo-rst-order kinetic equation, but may not offer insights into the possible theoretical basis of the pseudo-second-order kinetic equation. 3.3. A general rate law equation for biosorption 3.3.1. Model development It should be pointed out that both pseudo-rst- and secondorder equations have a common feature of the preset reaction order. However, the order of a chemical reaction must be determined from experiments, and cannot be simply preset to the

238

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

rst- or second-order unless the reaction mechanisms are well known. It is apparent that there is not any theoretical basis for biosorption reaction to be restricted to rst- or secondorder. In order to establish a general rate law equation for biosorption, biosorption reaction on the surface of biosorbent is assumed to be rate-controlling step. In this case, attention is turned from adsorbate concentration in bulk solution to change in the effective number of adsorption sites at the surface of biosorbent over time. Hence, the overall biosorption reaction on the surface of biosorbent can be regarded as a simple change in the adsorption state of biosorbent: Bt Bt+1 (90)
Fig. 2. Copper biosorption by dried activated sludge. Data from [75], and Eq. (92) prediction is shown in solid curve. x = 1.4, kx = 0.08 min1 , Qe = 75 mg g1 and correlation coefcient = 1.000.

in which Bt and Bt+1 represent the respective states of biosorbent at time t and t + 1 along with processing of biosorption. The effective concentration (t ) of number of the adsorption sites on the surface of biosorbent available for biosorption at time t can be quantied as follows: t = 1 Qt Qe (91)

becomes dt = k2 2 t dt (95)

In fact Eq. (91) is identical to Eq. (88). Obviously, for a virgin biosorbent t equals 1, and it tends to decreases over time. When biosorption process reaches its equilibrium, Qt equals Qe , and t becomes zero. Liu and Sheng [53] thought that if the reaction rate law is applied to Eq. (90), following rate expression for biosorption can be obtained: d = kx x t dt (92)

Combination of Eqs. (91), (94) and (95) yields following formulation: dQt = k2 (Qe Qt )2 dt in which k2 is dened as k2 = k2 Qe (97) (96)

in which kx is the rate constant with a unit of time inverse and x is the biosorption reaction order with regard to the effective concentration (t ) of the adsorption sites available on the surface of biosorbent. Obviously, Eq. (92) is the result of application of the universal rate law to a biosorption process, and can be used without any further assumption. According to the denition of the reaction order by IUPAC [52], the exponent x in Eq. (92) can be integral or rational nonintegral numbers. Examples of the description of biosorption data by Eq. (92) were presented in Figs. 2 and 3. When x = 1, Eq. (92) is reduced to dt = k1 t dt (93)

Basically Eq. (96) is the so-called the pseudo-second-order kinetic model for biosorption as shown in Eq. (44). It has been considered that k2 and Qe in Eq. (44) or Eq. (97) are two independent constants, however, Eq. (97) reveals that in the pseudo-second-order kinetic equation k2 and Qe indeed are interrelated.

Differentiating both the sides of Eq. (91) gives dt 1 dQt = dt Qe dt (94)

Substituting Eqs. (91) and (94) into Eq. (93) leads to dQt /dt = k1 (Qe Qt ) which is the same as the pseudo-rst-order kinetic equation (Eq. (43)) in which k1 = k1 . This means that the pseudo-rst-order kinetic equation for biosorption is only a special case of the proposed general rate law equation (Eq. (92)) for biosorption. On the other hand, in case where x = 2, Eq. (92)

Fig. 3. Cadmium biosorption by Enterobacter sp. Data from [76], and Eq. (92) prediction is shown in solid curve. x = 2.1, kx = 0.032 min1 , Qe = 14 mg g1 and correlation coefcient = 0.995.

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

239

3.3.2. Some considerations Eq. (92) shows that biosorption kinetics indeed obeys the universal rate law for a chemical reaction. It can be seen in Figs. 2 and 3 that the general rate law equation can provide a satisfactory description for the biosorption data obtained. In the sense of chemistry, reaction order cannot be predicted theoretically, and there exist mixed-order reactions with a fractional order for their rate [52]. In almost all previous studies of biosorption kinetics, Eqs. (43) and (44) had been directly chosen to t biosorption data without any explanation about the rational behind. As shown in Figs. 2 and 3, now it becomes clear that indeed there is no reason and need to preset biosorption kinetics to be the rst- or second-order unless biosorption mechanisms are known. In study of biosorption kinetics, one often knows nearly nothing about the reaction mechanisms due to the fact that biosorption process is extremely complex and various mechanisms would be involved [54]. Therefore, theoretically the order of biosorption process must be determined by the general rate law equation (Eq. (92)) rather than the preset-order kinetic equations. Rudzinski and Plazinski [48] thought that the form of Eq. (43) seemed to suggest a one-site-occupancy adsorption when the adsorbing molecule reacts with one adsorption site, while in the case of two-site occupancy adsorption, Eq. (44) would be naturally obtained. Such assumptions sound reasonable only for elementary reaction, but even if biosorption process can be regarded as an elementary reaction, these two assumptions could not lead to the establishment of the pseudo-second-order equation as pointed out earlier. In fact, due to the complexity of biosorbent surface, biosorption process cannot be readily assumed to be an elementary reaction. As shown in Eq. (92), the biosorption rate varies with the effective concentration of available sites of biosorbent, this means that in Eq. (92), biosorption is x-order with respect to the concentration of available sites of biosorbent () rather than the concentration of soluble adsorbate in bulk solution, as the result, it is unnecessary to link the biosorption kinetics to the possible biosorption mechanisms. Two constants (k2 and Qe ) in the pseudo-second-order equation for biosorption have been considered two independent parameters as shown earlier. However, Eq. (97) clearly shows that these two constants are interrelated. The integrated Eq. (44) for biosorption has following form: Qt = tQ2 k2 e tk2 Qe + 1 (98)

to the well-known Langmuir isotherm in the sense of model structure. Based on the assumption that the rate of adsorption solely depends on the fraction of sites available at time t, Ritchie [56] proposed a model for the kinetics of adsorption of gases on solids: dt (100) = (1 t )n dt in which t is the fraction of surface sites occupied by adsorbed gas at time t, n the number of surface sites occupied by each molecule of adsorbed gas, theoretically n should be a positive integer, and is the rate constant. Apparently, Eq. (100) does not match with the rate law of chemical reaction. In addition, Ritchie [56] also assumed that adsorption of gas on solids would be subject to elementary reaction, which may no longer be valid in the case of biosorption as discussed above. By introducing the concept of t , Eq. (100) can be written as follows: dt (101) = n t dt Eq. (101) has a similar formulation to Eq. (92). This in turn indicates that (i) the kinetics of adsorption of gas on solids is also subject to the universal rate law; (ii) the assumption of an elementary adsorption reaction is not necessary in Ritchies approach; (iii) in Ritchies approach n in Eq. (100) is an integer, but this constrain is no longer necessary in Eq. (101) because n in Eq. (100) indeed describes the reaction order of adsorption with respect to effective concentration of adsorption sites dened by the term .

3.4. Other useful kinetic equations for biosorption Apart from those rate law-based kinetic equations for biosorption as discussed earlier, some different types of kinetic equations have also been employed in description of biosorption data under nonequilibrium conditions. 3.4.1. Elovich equation The equation rst proposed by Roginsky and Zeldovich in 1934, and now generally known as the Elovich equation has been extensively applied to biosorption data [5760]. According to McLimtock [61], the differential form of this equation is often expressed as dQt = aE eE Qt (102) dt in which aE and E are two constants. It had been reported that biosorption data could be reasonably described by the Elovich equation, which considers that the rate-controlling step is the diffusion of the dye molecules [57]. In fact, the Elovich equation reveals the behaviors of chemisorption. 3.4.2. WeberMorris equation or intraparticle diffusion equation The WeberMorris equation is given by Qt = ki t 1/2 (103)

As shown above, k2 is related to Qe by Eq. (97). Replacing k2 in Eq. (98) by Eq. (97) gives Q t = Qe t t + tr (99)

in which tr = 1/k2 . According to Roels [55], the relaxation time of a chemical reaction is dened as the inverse of its reaction rate constant. Now it becomes clear that tr in Eq. (99) indeed represents the relaxation time of biosorption process. The terms tr and Qe in Eq. (99) are two independent constants with clearly dened physical meanings. It should be realized that Eq. (99) has a mathematic nature of hyperbolic equation, which is similar

240

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

in which ki is the intraparticle diffusion rate constant. In this equation, it is assumed that intraparticle diffusion is rate-limiting step in overall biosorption process, and Eq. (103) has been used to describe biosorption data obtained under such a circumstance [60,62]. 4. Biosorption thermodynamics In the sense of the reaction thermodynamics, change in free energy ( G ) of biosorption can be calculated in a way such that G = RT ln Keq (104)

in which T is absolute temperature and R is the gas constant. Using the Keq values determined from the biosorption isotherm equations, the corresponding values of G of biosorption can be determined at different experimental temperatures. It is known that G is the function of change in enthalpy of biosorption ( H ) as well as change in standard entropy ( S ): G = H T S (105)

tion data in the literature. Due to the complexity of biosorption mechanisms, it is rather difcult for researchers to choose isotherm and kinetic models according to the known mechanisms. So far, the criteria for choosing a isotherm or kinetic equation for biosorption data is mainly based on the goodness of curve tting which is often evaluated by statistical analysis. However, a good curve tting in the sense of statistical evaluation may not necessarily imply that this curve tting has true physical meaning, i.e., if a set of biosorption data is analyzed by different isotherm or kinetic equations, the best t equation may not be the one reecting the biosorption mechanisms. In this regard, it seems that most isotherm and kinetic studies of biosorption would be attributed to a simple mathematical exercise. It should be realized, as noted by Ho et al. [12], that selection of kinetic equation should be based on the mechanisms. Consequently, to formulate a mathematical expression of biosorption, one needs to seek for the models with strong theoretical characteristics rather than simple curve tting.

If Eq. (104) is inserted into Eq. (105), it becomes ln Keq = H 1 S (106) R R T It appears from Eq. (106) that both S and H of biosorption can be determined from the plot of ln Keq to 1/T. Liu and Xu [63] reported the negative values of G of Ni2+ biosorption by aerobic granules, indicating that the Ni2+ biosorption process could occur spontaneously, while the value of H was estimated as 63.8 kJ mol1 , and 0.26 kJ mol1 K1 for S . Basically, the heat evolved during physical adsorption is of the same order of magnitude as the heat of condensation, i.e., 2.120.9 kJ mol1 [77], while the heats of chemisorption generally falls into a range of 80200 kJ mol1 [64]. Therefore, it seems that Ni2+ biosorption by aerobic granules would be attributed to a physico-chemical adsorption process rather than a pure physical or chemical adsorption process. The positive value of H indicates that Ni2+ biosorption is an endothermic process. The low value of S may imply that no remarkable change in entropy occurred during the Ni2+ biosorption by aerobic granules. In addition, the positive value of S reects the increased randomness at the solidsolution interface during biosorption [5], and it also indicates that ion replacement reactions occurred [65,66]. Similar results were also found in on the Ni2+ biosorption by Chlorella vulgaris and polyporous versicolor and Fe3+ , Cr6+ and Pb2+ biosorption by Zoogloea ramigera [5,25,77].

5.2. Concern about thermodynamic study of biosorption In study of biosorption thermodynamics, it appears from Eqs. (105) and (106) that determination of value of the equilibrium constant is a key towards estimates of G , H and S . As shown earlier, determination of the equilibrium constant is closely related to the isotherm equation employed. Value of the equilibrium constant has been often calculated from the Langmuir isotherm equation [62,67,68]. Basar [35] reported the equilibrium constant determined by Frumkin isotherm equation, while FloryHuggins isotherm equation was also employed in determination of the equilibrium constant of Ni2+ biosorption [32]. Meanwhile, in biosorption of heavy metals, the equilibrium constant was also dened in a way such that Keq = Cad,e Ce (107)

5. Some concerns about biosorption research and application 5.1. Concern about kinetic and isotherm study of biosorption As discussed earlier, a very large number of isotherm and kinetic equations can be found for description of various biosorp-

in which Cad,e is the concentration of adsorbed metal ion on biosorbent at equilibrium. Aksu [5] and Han et al. [69] used Eq. (107) to determine the equilibrium constants of lead biosorption by Chlorella vulgaris and chaff, respectively. It should be pointed out that Eq. (107) is incomplete or even invalid as one is not able to apply the equilibrium law to a biosorption process unless the reaction stoichiometry is known. It is obvious that determination of the equilibrium constant closely depends on the isotherm equation used. For a given set of biosorption data, it can be expected that value of the equilibrium constant determined by different ways would vary signicantly, and such a variation in the equilibrium constant would in turn lead to inadequate estimates of G , H and S . These would make comparison of reported G , H and S values difcult or even unrealistic, and subsequently some G -, H - and S based conclusions on biosorption thermodynamics should be re-examined.

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242

241

5.3. Concern about application of biosorption technology To date, biosorption has been regarded as an effective technology for the removal of soluble heavy metals from aqueous solution. Most biosorbents currently used are in the form of suspended biomass, which are not effective and durable in repeated long-term application, and also make the post-separation of suspended biomass from the treated efuent extremely difcult [3]. These would cause problems associated with maintenance of biosorbent stability and regeneration of used biosorbents, and also limit application of conventional biosorbents in the form of suspended biomass in the removal of soluble metals from industrial wastewater. As Eccles [1] noted, a small number of pilot-plant studies have been carried out to investigate the potential of microorganisms to remove metals from liquid wastes but only one system in the past 15 years has been commercialized. Unfortunately, compared to basic research of biosorption, today application of biosorption technology falls far behind and there is a long journey ahead. In order to overcome the drawbacks associated with biosorbents in the forms of dispersed microorganisms, some attention has been turned to developing immobilized-type biosorbents, e.g. immobilized blue green microalgae and fungal biomass were used to remove cadmium and chromium [70,71], while entrapped fungal hyphae in structural brous network of papaya wood was also explored for the removal of heavy metals [72]. In addition, Zhang et al. [59] studied the biosorption of Cu2+ ions onto biolm under various experimental conditions. When selecting appropriate biosorbents for the removal of heavy metals from industrial wastewater, three criteria need to be taken into account, i.e., effectiveness in terms of effective separation of biosorbent from the treated water, adsorption capacity and rate, robustness to harsh operating conditions and reliability without release of harmful materials from adsorbent. Previous research showed that aerobic granules have the advantages of compact microbial structure, and excellent settling ability. The settling velocity of aerobic granules is as high as 71 m h1 , which is 58 times higher than that of microbial ocs, and aerobic granules can be completely separated out of the treated efuent by gravity in 1 min [73]. It is apparent that the characteristics of aerobic granules could satisfy the basic requirements for biosorbents. Recent study showed that aerobic granule-based biosorption process is an efcient and cost-effective technology for the removal of heavy metals from industrial wastewater streams [6,10,11,63]. References
[1] [2] [3] [4] [5] [6] H. Eccles, Trend Biotechnol. 17 (1999) 462. A.A. Hamdy, Curr. Microbiol. 41 (2000) 232. R. Vieira, B. Volesky, Int. Microbiol. 3 (2000) 17. S. Singh, B.N. Rai, L.C. Rai, Proc. Biochem. 36 (2001) 1205. A. Aksu, Proc. Biochem. 38 (2002) 89. Y.Y. Liu, S.F. Yang, S.F. Tan, Y.M. Lin, J.H. Tay, Lett. Appl. Microbiol. 35 (2002) 548. [7] M.X. Loukidou, A.I. Zouboulis, T.D. Karapantsios, K.A. Matis, Colloids surf. A: Physicochem. Eng. Aspects 242 (2004) 93.

[8] P.X. Sheng, Y.P. Ting, J.P. Chen, L. Hong, J. Colloid Interface Sci. 275 (2004) 131. [9] N.N. Fathima, R. Aravindhan, J.R. Rao, B.U. Nair, Environ. Sci. Technol. 39 (2005) 2804. [10] H. Xu, J.H. Tay, S.K. Foo, S.F. Yang, Y. Liu, Water Sci. Technol. 50 (2004) 155. [11] H. Xu, Y. Liu, J.H. Tay, Bioresour. Technol. 97 (2005) 359. [12] Y.S. Ho, J.C.Y. Ng, G. McKay, Sep. Purif. Methods 29 (2000) 189. [13] M. Erdem, A. Ozverdi, Sep. Purif. Technol. 51 (2006) 240. [14] I. Kiran, T. Akar, A.S. Ozcan, A. Ozcan, S. Tunali, Biochem. Eng. J. 31 (2006) 197. [15] E. Rubin, P. Rodriguez, R. Herrero, M.E.S. de Vicente, J. Chem. Technol. Biotechnol. 81 (2006) 1093. [16] S.W. Won, H.J. Kim, S.H. Choi, B.W. Chung, K.J. Kim, Y.S. Yun, Chem. Eng. J. 121 (2006) 37. [17] I. Langmuir, J. Am. Chem. Soc. 40 (1918) 1361. [18] H. Freundlich, Z. Physik. Chem. 57 (1907) 385470. [19] R. Sips, J. Chem. Phys. 16 (1948) 490. [20] K.V. Kumar, K. Porkodi, J. Hazard. Mater. 138 (2006) 633. [21] G.G. Hammes, Thermodynamics and Kinetics for the Biological Sciences, Wiley-Interscience, New York, 2000. [22] Metcalf, Eddy, Wastewater Engineering, 3rd ed., McGraw-Hill, Singapore, 2003. [23] M.S. Alhakawati, C.J. Banks, S. Smallman, Water Sci. Technol. 27 (2003) 143. [24] A. Selatnia, M.Z. Bakhti, A. Madani, L. Kertous, Y. Mansouri, Hydrometallurgy 75 (2004) 11. [25] E.D. van Hullebusch, M.H. Zandvoort, P.N.L. Lens, J. Chem. Technol. Biotechnol. 79 (2004) 1219. [26] Y. Liu, H. Xu, S.F. Yang, J.H. Tay, J. Biotechnol. 102 (2003) 233. [27] F.M.M. Morel, J.G. Hering, Principles and Applications of Aquatic Chemistry, Wiley, New York, 1993. [28] O. Redlich, D.L. Peterson, J. Phys. Chem. 63 (1959) 1024. [29] B. Preetha, T. Viruthagiri, Afr. J. Biotechnol. 4 (2005) 506. [30] A.R. Khan, R. Ataullah, A. AlHaddad, J. Colloid Interface Sci. 194 (1997) 154. [31] J. T th, Acta Chem. Acad. Hung. 69 (1971) 311. o [32] K. Vijayaraghavan, T.V.N. Padmesh, K. Palanivelu, M. Velan, J. Hazard. Mater. 133 (2006) 304. [33] C.J. Radke, J.M. Prausnitz, J. AIChE 18 (1972) 761. [34] M.M. Dubimin, L.V. Radushkevich, Chem. Zentralbl. 1 (1947) 875. [35] C.A. Basar, J. Hazard. Mater. 135 (2006) 232. [36] W. Riemam, H. Walton, Ion Exchange in Analytical Chemistry, Pergamon Press, Oxford, 1970. [37] S. Tunali, T. Akar, A.S. Ozcan, S. Kiran, A. Ozcan, Sep. Purif. Technol. 47 (2006) 105. [38] T. Grchev, M. Cvetkovska, T. Stalov, J.W. Schultze, Electrochim. Acta 36 (1991) 1315. [39] P.J. Flory, J. Chem. Phys. 10 (1942) 51. [40] M.L. Huggins, J. Phys. Chem. 10 (1942) 151. [41] M.K. Kaisheva, G. Saraivanov, Langmuir 7 (1991) 2380. [42] M. Horsfall, A. Spiff, Acta Chim. Slov. 52 (2005) 174. [43] S. Brunauer, P.H. Emmet, E. Teller, J. Am. Chem. Soc. 60 (1938) 309. [44] M.J. Temkin, V. Pyzhev, Acta Physiochim. URSS 12 (1940) 217. [45] C. Aharoni, M. Ungarish, J. Chem. Soc., Faraday Trans. 73 (1977) 456. [46] Y. Liu, S.F. Yang, H. Xu, K.H. Woon, Y.M. Lin, J.H. Tay, Process Biochem. 38 (2003) 997. [47] K.V. Kumar, S. Sivanesan, V. Ramamurthi, Process Biochem. 40 (2005) 2865. [48] W. Rudzinski, W. Plazinski, J. Phys. Chem. B 110 (2006) 16514. [49] G. Blanchard, M. Maunaye, G. Martin, Water Res. 18 (1984) 1501. [50] G.E. Boyd, A.W. Adamson Jr., L.S. Myers, J. Am. Chem. Soc. 69 (1947) 2836. [51] S. Azizian, J. Colloid Interface Sci. 276 (2004) 4752. [52] IUPAC, Compendium of Chemical Terminology, 2nd ed., IUPAC, 1997. [53] Y. Liu, L. Sheng, Biochem. Eng. J. 38 (2008) 390394. [54] H. Xu, Nanyang Technological University, Singapore, PhD Thesis, 2007.

242

Y. Liu, Y.-J. Liu / Separation and Purication Technology 61 (2008) 229242 [66] S.K. Tangkawanit, A. Rangsriwatananon, Micropor. Mesopor. Mater. 79 (2005) 171. [67] E.A. Oliveira, S.F. Montanher, A.D. Andrade, J.A. Nobrega, M.C. Rollemberg, Proc. Biochem. 40 (2005) 3485. [68] X.S. Wang, Y. Qin, Z.F. Li, Sep. Sci. Technol. 41 (2006) 747. [69] R.P. Han, J.H. Zhang, W.H. Zou, H. Shi, H.M. Liu, J. Hazard. Mater. 125 (2005) 266. [70] A. Saeed, M. Iqbal, World J. Microbiol. Biotechnol. 22 (2006) 775. [71] R.S. Bai, T.E. Abraham, Bioresour. Technol. 87 (2003) 17. [72] M. Iqbal, A. Saeed, Enzyme Microb. Technol. 39 (2006) 996. [73] Y. Liu, J.H. Tay, Biotechnol. Adv. 22 (2004) 533. [74] J.B. Yang, B. Volesky, Water Res. 33 (1999) 3357. [75] O. Gulnaz, S. Saygideger, E. Kusvuran, J. Hazard. Mater. 120 (2005) 193. [76] W.B. Lu, J.J. Shi, C.H. Wang, J.S. Chang, J. Hazard. Mater. 134 (2006) 80. [77] Y. Sag, T. Kutsal, Biochem. Eng. J. 6 (2000) 145.

[55] J.A. Roels, Energetics and Kinetics in Biotechnology, Elsevier, New York, 1983. [56] A.G. Ritchie, J. Chem. Soc., Faraday Trans. 73 (1977) 1650. [57] A. Aretxaga, S. Romero, M. Sarra, T. Vicent, Biotechnol. Prog. 17 (2001) 664. [58] Y. Sag, Y. Aktay, Biochem. Eng. J. 12 (2002) 143. [59] J. Zhang, B. Jiang, X.G. Li, R.X. Liu, Y.L. Sun, Chin. J. Chem. Eng. 13 (2005) 135. [60] A.B. Perez-Marin, V.M. Zapata, J.F. Ortuno, M. Aguilar, J. Saez, M. Llorens, J. Hazard. Mater. 139 (2007) 122. [61] I.S. McLintocl, Nature 216 (1967) 1204. [62] G. Akkaya, A. Ozer, Process Biochem. 40 (2005) 3559. [63] Y. Liu, H. Xu, Biochem. Eng. J. 35 (2007) 1742. [64] D.O. Hayward, B.M.W. Trapnell, Chemisorption, 2nd ed., Butterworth, London, 1964. [65] S.I. Lyubchik, A.I. Lyubchik, O.L. Galushko, L.P. Tikhonova, J. Vital, I.M. Fonseca, S.B. Lyubchik, Colloids Surf. A: Physicochem. Eng. Aspects 242 (2004) 151.

Potrebbero piacerti anche