It is shown that the modified pseudofirstorder (MPFO) kinetic equation proposed recently by Yang and AlDuri
simulates well the behavior of the kinetics governed by the rate of surface reaction and described by our general kinetic
equation, based on the statistical rate theory. The linear representation with respect to time, offered by the MPFO
equation seems to be a convenient tool for distinguishing between the surface reaction and the diffusional kinetics.
Also, a method of distinguishing between the surface reaction and the intraparticle diffusion model based on analyzing
the initial kinetic isotherms of sorption is proposed. The applicability of these procedures is demonstrated by the
analysis of adsorption kinetics of the reactive yellow dye onto an activated carbon.
believed to be a solid check that the adsorption kinetics is in the bulk phase (b) and the phase adsorbed on the surface (a).
controlled by the intraparticle diffusion. Further, k and T denote the Boltzmann constant and absolute
Meanwhile, recently it has been shown by Rudzinski and temperature, as usual. For the bulk phase the following expression
Plazinski that such a squareroot dependence in the initial region is accepted
of adsorption times may also be typical of kinetics controlled by
the rate of surface reaction. They arrived at this conclusion by b ) b + kT ln c (3)
accepting the popular Langmuir model of adsorption and applying
the new fundamental SRT approach to adsorption kinetics.4 While where c is the concentration of the solute in the bulk phase. The
applying that new SRT approach, Rudzinski and Plazinski5 have expression for a is derived from the equilibrium adsorption
also shown that the Lagergren equation is the limiting form of isotherm. This is because one fundamental assumption of the
the kinetic isotherm corresponding to the Langmuir model of SRT approach is that the adsorbed phase is in a quasiequilibrium.
adsorption, when the system is not far from equilibrium. It means that all the correlation functions are the same as they
Therefore, we may summarize the present situation as follows. would be at equilibrium at the same surface coverage.
Using the new fundamental SRT approach to adsorption kinetics, So let us consider the popular Langmuir model of adsorption
we have arrived at much more adequate and accurate kinetic and its corresponding adsorption isotherm:
equations to represent the kinetics which is governed by the rate of KLce
surface reactions. That more adequate but also more complicated e ) (4)
equation reduces to the squareroot dependence on time for short 1 + KLce
adsorption times and to the Lagergren equation for long times. These In eq 4 KL is the temperaturedependent Langmuir constant
simple equations are very attractive for being used to correlate whereas the subscript e refers to the equilibrium conditions.
experimental data, but their use is limited to short and long adsorption Assuming also that the bulk/surface exchange rate is proportional
times, respectively. Here we are going to show that the new modified to the frequency of the collisions of the solute molecules with
pseudofirstorder kinetic equation proposed recently on an empirical the available fraction of surface, one arrives at the following rate
basis by Yang and AlDuri6 reproduces very well the exact solution equation
of our kinetic equation in the whole region of adsorption times. It
means that this empirical equation reproduces well the behavior of
our theoretical kinetic isotherm in the whole region: from short
initial times to long times when the system is close to equilibrium.
d
dt
) Klsce(1  e) KLc
[
1

1
KLc 1  ] (5)
At the same time the simplicity of that empirical equation makes where Kls ) Klsce(1  e). One can also take into account the
it a very attractive and convenient tool to correlate experimental set of technical parameters which characterize the conditions
data. The theoretical background of this modified pseudofirstorder under which the kinetic experiment is carried out, i.e., the volume
equation will be discussed, and its applicability will be presented of solution (V), the solute initial concentration (cin), and the
by an appropriate analysis of some experimental data reported in monolayer capacity (Nm) which is connected with the applied
the literature. mass of adsorbent. Using the notation
Theory
The famous Lagergren kinetic equation is usually written as
{ ) cinV Nm
) K lst ) Klsce(1  e)t
(6)
[ ( ) ( )]
d
d b  a a  b lim ) (Kace + Kd)(e  ) (9)
) Kls exp  exp (2) fe dt
dt kT kT
whereas the SRT equation 7 leads to the following expression:
where Kls is the rate of adsorption/desorption at equilibrium,
whereas b and a are the chemical potentials of adsorbate (solute) d 2(  e2)
lim ) (  ) + O[e  ]2 (10)
fe d e(  e)(1  e) e
(4) Rudzinski, W.; Panczyk, T.; Plazinski, W. J. Phys. Chem. B 2005, 109,
21868. Integration of both eqs 9 and 10 with the boundary condition
(5) Rudzinski, W.; Plazinski, W. J. Phys. Chem. C 2007, 111, 15100.
(6) Yang, X.; AlDuri, B. J. Colloid Interface Sci. 2005, 287, 25. (t ) 0) ) 0 yields the following integral form of the Lagergren
(7) Ward, C. A.; Findlay, R. D.; Rizk, M. J. Chem. Phys. 1982, 76, 5599. equation
The Modified PseudoFirstOrder Kinetic Equation Langmuir, Vol. 24, No. 10, 2008 5395
( )
Dn22 Its integration with the boundary condition (t ) 0) ) 0 yields
e
6
)1 2
1
n)1 n2
exp 
r2
t (14)
) 2Klsce(1  e)KLcint ) 2Klscinet (22)
where D is the intraparticle diffusivity and r is the particle radius.
Thus, at t f 0, tangent of this linear plot is always a linear
At long times, when f e, eq 14 reduces to the Lagergrenlike
function of the expression (cine)1/2. It also means that the square
expression
root dependence on time of the monitored surface coverage
(t1/2) cannot be treated as a proof that the kinetics of sorption
D2 6
ln(e  ) )  t + ln e + ln 2 (15) is governed by the rate of intraparticle diffusion.
r2 One should, however, remember that eq 21 does not apply at
which suggests that adjusting properly the value of e should extremely low times and surface concentrations when the
make the experimentally determined ln(e  ) data a linear predicted rate of adsorption is higher than the flux of solute
function of time. The tangent of such a linear plot will be equal molecules to the surface. For such extreme low times and surface
to D2/r2 and its abscissa to (ln e + ln (6/2)). For the purpose concentrations a more complicated procedure is to be applied.9
of our further consideration, it is essential to emphasize that the Nevertheless, drawing such doublelogarithmic plots may be a
value of e parameter determined from the abscissa should be source of interesting information concerning the first kinetic
equal to that making the plot ln(e  ) a linear function of time. process. Provided that it could be governed by the rate of
The important difference lies in the theoretical interpretation of intraparticle diffusion, as follows from eq 18, the tangent should
the abscissas in eq 15 (D) be a linearly increasing function of e. Thus, while considering
the first region of small adsorption times and surface coverages,
6 we face a similar situation to that when the system is close to
D ) ln e + ln (16) equilibrium. Namely, both the diffusional and the surface reaction
2
kinetics are represented by the same kind of theoretical expression,
and in eq 11 (L) and the difference lies only in the interpretation of their kinetic
coefficients. In the case of the diffusional kinetics the tangent
L ) ln e (17)
of the linear plot vs t1/2 will be a linear function of the equilibrium
It has been frequently emphasized in literature that for short surface coverage e, whereas in the case of the surface reaction
times, when the fractional uptake f e is less than around 0.3, model, it will be a more complicated function suggested by eq
eq 14 reduces practically to the following one 22.
Therefore, we may summarize the results of our considerations
Dt
6e as follows:
) (18) In the initial region of short adsorption times the squareroot
r
dependence of the adsorbed amount on time is the lumped
As we have already stated, the Lagergren equation is the expression valid for both the surface reaction and the diffusional
limiting form of both diffusional and the surface reaction kinetic kinetic models.
equations at long adsorption times when the system approaches For both kinetic models, the Lagergren expression is the
equilibrium. In the case of the surface reaction kinetic model, lumped form of appropriate, more adequate, and sophisticated
this conclusion was drawn by applying both the classical TAAD expressions, describing the rate of adsorption when the system
approach and the new fundamental SRT approach to adsorption approaches equilibrium.
kinetics. A somewhat different situation is faced when these two Distinguishing between these two kinetic models needs a
approaches are applied to predict the limiting form of the kinetic careful analysis of the behavior of the coefficients in these lumped
equations for short adsorption times, i.e., when t f 0. expressions.
While applying the classical TAAD approach to adsorption The squareroot dependence on time or the Lagergren plots
kinetics, from eq 8, one arrives at the following limiting form have been frequently used to correlate experimental kinetic
of the surface reaction kinetic equation: isotherms. However, as the theoretical background of these
lumped relations has remained unclear until recently, both their
d
lim ) Kac (19) successful applications and the observed deviations could not
tf0 dt receive solid interpretation. It seems that the main source of their
Its integration with the boundary condition (t ) 0) ) 0 yields popularity and common use were the simple correlations of
(8) Suzuki, M. Adsorption Engineering; Kodansha: Tokyo, 1990. (9) Panczyk, T. Phys. Chem. Chem. Phys. 2006, 8, 3782.
5396 Langmuir, Vol. 24, No. 10, 2008 Rudzinski and Plazinski
d e
) KM (e  ) (23)
dt
where KM is a constant. We will call it the MPFO (modified
pseudofirstorder) equation. Its integration with the boundary
condition (t ) 0) ) 0 yields
ln(e  ) + ) ln e  KMt (24)
e
Let us remark that eq 23 is a hybrid between the squareroot
dependence when f 0 and the Lagergren dependence when
f e. This is because
d 1
lim ) KMe2 (25)
tf0 dt
Figure 1. The comparison of the theoretical kinetic adsorption
When compared to eq 21 this would imply that in the limit isotherms calculated from the SRT equation (7) for three different
f 0 the coefficient KM has the following interpretation values of the equilibrium surface coverage: e ) 0.25 ()), e ) 0.5 (0),
and e ) 0.75 (O) with their approximate values calculated from the
Klscin MPFO equation (24) (s). In the case of panel (B) the linear representation,
KM ) (26) suggested by eq 24, was used. All calculations were performed for
e ) 1.1e.
Then in the limit f e, from eq 23 we obtain
d
lim ) KM(e  ) + O[e  ]2 (27)
fe dt
By comparison with eq 10 in the limit f e, the interpretation
of KM is identical with that of k1 coefficient given by eq 13.
The MPFO kinetic equation offers a very attractive simple
way of correlating experimental data. Now we are going to show
that this hybrid between the squareroot and the Lagergren
dependence simulates fairly well the behavior of the fully exact
function (t) in eq 7 also in the mediate region between short
and long adsorption times. We performed some simple model
investigations, plotting kinetic isotherms calculated from eqs 7
and 24 as functions of the reduced time /0.9, where 0.9 is the
time when the surface coverage reaches the value ) 0.9e.
The dimensionless time for the MPFO equation is defined as )
KMt. The results are shown in Figures 1 and 2.
Looking through Figures 1 and 2, we can see a quite good
agreement between the theoretical kinetics isotherms predicted
by the MPFO equation (24) and the SRT expression (7). The
accuracy of MPFO equation (24) (treated as an approximation
of the exact equation (7)) is increasing while the parameter is
increasing. Then we have also found that when the value of
Figure 2. The comparison of the theoretical kinetic adsorption isotherms
is very high in comparison to e, one can observe the reverse calculated from the SRT equation (7) and from the MPFO equation (24)
behavior, i.e., a decreasing accuracy of approximation (24). (s). All calculations were performed for ) 2e. All the notations are
However, the situation that . e is highly nontypical when the same as in Figure 1.
considering the real kinetic experiments. This is because such
high values of parameters mean that no significant changes in drawn that the MPFO equation reproduces the SRT expression
the solute concentration will be observed in the course of (7) best for the low values of e.
experiment. From the point of view of a successful application Then, we also decided to compare the behavior of theoretical
of an adsorption process, the highly desirable situation is that kinetic isotherms calculated from the full form of the IDM
e, i.e., when almost all the sorbate is removed from the equation (14) with those calculated from the MPFO equation
solution. An inspection into literature seems to suggest that the (24). Now, the dimensionless time for the IDM equation is defined
most typical cases are these when lies between these two values as ) D2t/r2. The results are shown in Figure 3. One can see
) 1.1e and ) 2e, accepted in our model calculations there a somewhat systematically worse agreement than that in
presented in Figures 1 and 2. A general conclusion can also be Figure 2. It seems, however, that the MPFO empirical equation
The Modified PseudoFirstOrder Kinetic Equation Langmuir, Vol. 24, No. 10, 2008 5397
Table 2. The Values of the Tangents (KM, D2/r2) and of the Abscissas of Their Linear Plots (Data Regressions) Presented in Figure 8
MPFO plot Lagergren plot
cin (mg/L) ln Ne (SRT) ln Ne + ln(6/ ) (IDM)
2
KM (1/min) abscissa (exptl) 2 2
D /r (1/min) abscissa (exptl)
35.4 4.43 3.93 1.81 103 4.45 2.78 103 3.87
60.3 4.98 4.48 9.40 103 4.99 2.23 103 4.97
87.2 5.35 4.86 6.15 103 5.36 1.10 103 5.00
131 5.84 5.34 5.30 103 5.78 9.54 103 5.49
diffusional kinetic mechanism. We have faced a similar situation be combined yet with a kind of a film resistance equation. Such
in our previous works, where such switching occurred at about still complicated model of adsorption kinetics seems to be
80% of the equilibrium surface coverage in the systems suggested by the data presented in Figure 4. There, best linear
investigated by us.5 Then, as the remaining parts of the kinetic regressions will not originate at zero. Combining the SRT equation
isotherms are closer to equilibrium, they should be well correlated with a film resistance model would result into a system of two
by the Lagergren plots (15). Thus, assuming that switching in differential equations. We will present a more complicated model
our present system occurs at 66% of the equilibrium surface in a future publication.
coverage, we decided to construct the following graphical
illustration. Conclusions
Every kinetic isotherm was divided into two parts: the first
part containing N(t) values smaller than 0.66Ne and the second The MPFO equation can reproduce very well the behavior of
part the values N(t) > 0.66Ne. the kinetic eqs 5, 7, 11, and 22 based on the SRT approach by
The first part was correlated by using the linear MPFO assuming the Langmuir model of adsorption and a homogeneous
representation (24) whereas the second part was correlated by solid surface. The accuracy of this approximation is the highest
the linear Lagergren plot (15). for the values of parameter, corresponding to the most common
Looking into Figure 8, we can see fairly good linear experimental conditions. Thus, the empirical MPFO equation
correlations, predicted by the assumption of switching from the can be treated as an approximation of our exact SRT equation.
surface reaction to the diffusional sorption kinetics. So, it seems The linear representation of the MPFO equation (24) can be
interesting to analyze the values of the coefficients (tangents, applied only when the true experimental values of Ne are known.
abscissas) of their linear regressions. They are collected in This is because treating Ne as the bestfit parameter, leading to
Table 2. the best linearity of ln(Ne  N(t)) + N(t)/Ne vs t plots, enables
One can see when looking into Table 2 that the values of obtaining unphysical values of Ne as well as ce when the mass
abscissas in the MPFO plots are in very good agreement with balance relation is used.
their theoretically predicted values, i.e., with ln Ne. However, in One can propose a simple method, based on the analysis of
the case of the Lagergren plots, their abscissas are generally not kinetic adsorption isotherms measured at very short, initial times
equal to ln Ne as suggested by eq 11. Moreover, their values are of experiment, to distinguish between the SRT and the IDM
not equal to ln Ne + ln (6/2) as predicted by the diffusional kinetic models. This, however, would require measurements
Lagergrenlike expression (15). This is more evidence for the carried out with an exceptionally high accuracy.
fact that switching takes place from one kinetic mechanism to
Such possibility is offered by applying the linear MPFO
another.
representation (24). This has led us to discover switching from
An explanation as to why such switching takes place may not
the surface reaction to diffusional kinetics when the surface
be a trivial problem. At present we may launch the following
coverage reaches about twothirds of the equilibrium surface
hypothesis.
coverage.
There may exist a gradient of concentration at the entrance
to pores of the adsorbent. So, in general the SRT equations should LA8000448