Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: In this work, the sorption capacity behavior of copper and cadmium divalent ions in single and binary
Received 13 October 2010 systems has been investigated. Batch metal sorption experiments to obtain the kinetic and equilibrium
Received in revised form 8 December 2010 data for the dry biomass of the Eichhornia crassipes species were performed under controlled temperature,
Accepted 9 December 2010
constant shaking and pH 5. An equilibrium time of 60 min was achieved for both Cu(II) and Cd(II) ions
in mono-component system as well as in binary one. It was found that the overall adsorption data were
Keywords:
best described by the pseudo second-order kinetic model. Six adsorption isotherms have been tested
Biosorption surface modeling
to fit the mono-component equilibrium data, obtaining the best description by the Langmuir-type one.
Multi-component
Metal adsorption capacity
A modified extended-to-multi-component Langmuir-type isotherm model was applied to predict the
binary adsorption data and its adjustable parameters were estimated by the PSO method. Finally, the
results have shown that the affinity of each metal ion onto the E. crassipes surface is influenced by the
presence of the other one.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction exist for the removal of heavy metals in low concentrations from
aqueous solution by macrophytes [11–16], activated carbon [17],
The pollution of aquatic ecosystem by heavy metals resulting fungi [18], orange peel [19–21], and higher plants [22], sequestering
from the release of industrial effluents is a matter of great con- and accumulating metallic ions.
cern. In aquatic environment, heavy metals such as cadmium and Although lot of works have used different biomaterials as low
copper are accumulated on biota and bio-magnified in the food cost biosorbent for single metal removal, while employing the
chain, resulting in harmful effects on human physiology and other biosorption technique, the presence of a large number of metals
biological systems [1,2]. Cadmium is one of the most toxic heavy in industrial wastewaters has increased the difficulty in studying
metals, even in low concentrations, causing renal dysfunction, bone the adsorption of multimetal systems [23]. In such systems, the
degeneration, lung inefficiency, liver damage and hypertension in metal ion biosorption is strongly depending on the solution pH,
humans, while copper can be deposited in the brain, skin, liver, initial metal ion concentration, surface properties of the adsor-
pancreas and myocardium causing serious toxicological effects on bent, the interaction between metal ion species among others
humans and animals [3,4]. [1]. So far, many biosorption researches have been mainly dedi-
In high metal concentration-contaminated wastewater treat- cated to test biomaterials as biosorbent, investigate their selectivity
ments, several conventional methods such as chemical precip- and adsorption capacity, and explain the interference and com-
itation, membrane filtration, ion exchange and electrochemical petition produced by metals for adsorption sites on the basis of
treatment have also been applied [5–9]. However, when the most of representative biosorption isotherms and kinetic models. In this
these methodologies are applied in low metal concentration efflu- context, literature survey also reveals that the application of dead
ent treatments, they could have high chemical costs or low removal biomass of aquatic macrophytes like Eichhornia crassipes [24,25],
efficiencies, low selectivities, high-energy requirements, and gen- Spirodela intermedia, Lemna minor and Pistia stratiotes [26], and Ege-
eration of secondary toxic slurries [10]. A variety of mechanisms ria densa [10–12] among others is preferred as efficient biosorbent
for removing heavy metal from aqueous system.
The aim of this work was to evaluate the adsorption capacity of
∗ Corresponding author. Tel.: +55 45 3379 7092; fax: +55 45 3379 7002. E. crassipes as well as to investigate on the adsorption equilibrium
E-mail addresses: modenes@unioeste.br, anmodenes@yahoo.com.br and kinetic of copper and cadmium divalent ions in both single and
(A.N. Módenes). binary component systems. Preliminary tests were also performed
1385-8947/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.cej.2010.12.029
A.N. Módenes et al. / Chemical Engineering Journal 168 (2011) 44–51 45
to optimize the conditions, viz. drying temperature and biosorbent metal solution was added to 250 mg dry plant in a 125-mL Erlen-
grain size for removal of copper and cadmium ions from single meyer flask. The solution pH was adjusted below the precipitation
and bi-metal systems. As the presence of one ion could influence threshold for each metal, approximately 5.0 for both metal ions
the maximum capacity or affinity of other ones in a Cu–Cd binary [12]. The mixtures were shaken on a rotary shaker for 120 min for
sorption system, correction parameters to the classical competi- each temperature setting, and then, the quantity of metal in each
tive Langmuir equation have been used in order to make the model sample was measured by AAS. The biosorbent grain size effect was
capable to depict the complexity of the adsorption process. The also studied, using three fractional grain sizes between 0.147 and
PSO method was used to predict all modified extended Langmuir 0.589 mm and their mixture as well. This test was carried out using
isotherm parameters, allowing fitting well the multi-component the same procedure for the temperature effect test, but in this case,
adsorption data in a 3-D representation. These studies may be a 30 ◦ C controlled temperature metal solution and 120 min con-
useful for low cost industrial wastewater treatment, where the tact time were used. Then the metallic solution was separated from
simultaneous removal of several heavy metals is needed. the biosorbent by using a vacuum filtration system and the metal
concentrations were measured by AAS. The amount of adsorbed
2. Materials and methods metal during contact time by E. crassipes biomass was calculated
according to the mass balance expressed by Eq. (1):
2.1. Chemicals V (C0 − Ct )
qt = (1)
m
The chemicals used were of analytical-reagent grade. Deionized
waters were also used as dilution media. Several stock solutions of where qt (in mequiv. g−1 ) is the amount of adsorbed metal by E.
Cu and Cd in deionized waters were prepared, ranging from 4.0 to crassipes, C0 and Ct (in mequiv. L−1 ) are the initial and final metal
8.0 mequiv. L−1 , from their water-soluble metallic salt (CuCl2 ·2H2 O concentrations in solutions, respectively, V is the solution volume
and CdCl2 ·H2 O, Merck Pro Analysis) and stored in 1-L volumet- and m is the adsorbent mass.
ric flasks for posterior metal adsorption experiments. Solutions of
NaOH (1 M) and HCl (1 M) were also used as a pH adjustment. Based 2.5. Kinetic test
on Cu and Cd standard solutions (1.0 g L−1 for AAS, Merck), different
diluted-metal concentrations ranging from 0.1 to 1.5 mg L−1 were In order to obtain the equilibrium time, single standard solutions
also prepared and used to obtain the calibration curve for atomic of Cu and Cd were prepared of approximately 4.0 mequiv. L−1 . For
absorption spectrometer (AAS). monocomponent kinetic experiments, a 250 mg dry biomass was
added to a 50 mL metal solution of Cu and Cd, at 5 initial pH, while
2.2. Biomass Sampling for a binary mixture only 25 mL of each metal solution was added
to 250 mg dry biomass. The mixtures were agitated on a shaker at
For biosorption experiments, aquatic macrophytes E. crassipes 30 ◦ C constantly controlled temperature. The solution for pH moni-
plants were collected from natural shallow ponds near Cascavel toring was carried out during the sorption experiments and the pH
city, which is located in the western region of the Brazilian Paraná adjustment to 5 was made, as required. The adsorption experimen-
state. After collection, aquatic macrophytes E. crassipes were thor- tal times were set up at several short contact times, ranging from 0
oughly cleaned in running tap water, in order to eliminate any to 120 min, in order to determine the equilibrium time. In this way,
remains of sediment and particles. In the laboratory, the biomasses a whole sample was taken at each setting time and the liquid phase
were dried using an electrical oven at 30 ◦ C and then grounded was separated from the adsorbent using a vacuum filtration system.
using the steel-knifes electrical mill and sieved into different par- Afterwards, initial and final cadmium concentrations in each liquid
ticle sizes. phase were determined by AAS and the amount of adsorbed metal
(qt ) during contact time by E. crassipes was calculated by Eq. (1).
2.3. AAS measurements
2.6. Equilibrium tests
For metal concentration measurements, after each test and
biosorption experiment, the liquid phase was separated from Several batch single metal sorption experiments using the E.
the adsorbent by a vacuum filtration system using 0.45 m- crassipes biomass as biosorbent were carried out using a con-
membranes. An absorption atomic spectrometer, model AA stant volume of 50 mL metal-supplied solution (4.0 mequiv. L−1 ) in
932-GBC (Analitica) was used. For cadmium analysis, eight diluted contact with dry biomass, ranging from 20 to 550 mg. For batch
standard solutions ranging from 0.004 to 1.8 mg L−1 were used in Cu–Cd sorption experiments, only 25 mL of each metal solution
order to calibrate the absorption atomic spectrometer (AAS), while with the same initial concentrations (4.0 mequiv. L−1 ) was added
for copper, the AAS calibration was made by using eight diluted to dry biomass, ranging from 20 to 550 mg. All experiments were
standard solutions ranging from 0.01 to 4 mg L−1 , respectively. The performed in duplicates, with constant shaking and controlled tem-
calibration coefficients (r2 ) obtained from four calibration stan- perature of 30 ◦ C, and the pH adjustment to 5 was made, as required,
dards for all analyses were 0.995 or better. The wavelengths (nm) during the 120 min contact time. Afterwards, the whole sample was
used for the metals were: Cu, 224.7 and Cd, 228.8. After the AAS cal- taken at each setting time and the liquid phase was separated from
ibration, some samples were diluted to fit in the calibration range. the adsorbent by using a vacuum filtration system, which deter-
Confidence intervals of 95% were calculated for the results. mined the initial and final metal concentrations for each liquid
phase by AAS and the amount of adsorbed metal (qt ) during contact
2.4. Preliminary tests time by E. crassipes was calculated by Eq. (1).
In order to improve the biosorption kinetic and experimental 2.7. Search analysis of kinetic and equilibrium adsorption
equilibrium conditions, the temperature effect on the drying plant parameters
was tested using a 30–50 ◦ C temperature range, while the sorption
temperature effect was evaluated at four controlled temperatures, In this work, a stochastically optimized global method, called
ranging from 25 ◦ C to 45 ◦ C. For these tests, a volume of 50 mL for a particle swarm optimization (PSO) proposed by Kennedy and
46 A.N. Módenes et al. / Chemical Engineering Journal 168 (2011) 44–51
Table 1
Metal percent removal in preliminary biosorption tests at different values of drying temperature, solution temperature and grain size, using the E. crassipes biomass.
Copper 52 ± 2 45 ± 2 41 ± 2 49 ± 2 45 ± 2 40 ± 2 44 ± 2 43 ± 2 44 ± 2 45 ± 2
Cadmium 41 ± 2 35 ± 2 53 ± 2 63 ± 2 56 ± 2 57 ± 2 52 ± 2 51 ± 2 52 ± 2 53 ± 2
Eberhard [27], has been applied to search for the globally opti- previous results, the 30 ◦ C solution temperature was set up for sin-
mized parameter values, which correspond to a set of non-linear gle and binary metal sorption experiments. On the other hand, the
models that represent the kinetic and equilibrium adsorption data. metal-sorption experiments based on fractional biomass grain sizes
The PSO method is similar to the evolutionary algorithms (EAs) (0.147–0.589 mm) have not shown significant differences on the
and is based on a population of individuals, called particle swarm. metal ion removal as compared to their mixture and different grain
However, the only difference between EA and PSO is in the space sizes tested (see Table 1). For practical purpose, the grain size effect
localization of individuals (particles) recognized by the operator might be considered insignificant, and consequently, grains, with-
speed. In genetic algorithms GA, the individuals pass through the out a previous size separation after the grounding process, were
operator reproduction, mutation and election [13]. PSO method has chosen, obtaining thus a 45% and 53% removal rate for copper and
been recently applied in several studies to solve nonconstrained cadmium, respectively.
optimization problems [28–33]. In contrast to the linearization
methods, the search procedure was found to reach convergence 3.2. Kinetic test
in a few iterations for the given numbers of parameters and objec-
tive functions. The obtained results were comparable with those The E. crassipes biomass-based kinetic data using copper and
obtained via an application of a hybrid genetic algorithm. In early cadmium in single and binary biosorption systems, at pH 5 and
works [28–30], the parameter identification procedure based on 30 ◦ C solution temperature are shown in Fig. 1. Logistic-type func-
the PSO method was reported, having empirically estimated the tions are used to fit the adsorption process of single metal (Fig. 1a)
essential PSO parameter values and their influence on the search achieving a good correlation coefficient (r2 = 0.981 for Cu and
space for the estimation of modeling parameters, in order to obtain
a quick and good convergence and reliable results.
The description of the PSO algorithm is basically as follows: sin-
gle particle in a chaotic movement in an N-particles swarm, namely
solution search space, is kinematically described as reported in
previous works [13,28–30], where its position vector and the
corresponding velocity one are iteratively assigned, based on a
few parameters or variables that resemble swarm inertia, chaotic
movement of each particle in the swarm and their geographical
space constraints. For the initialization of the population, the fol-
lowing parameter values are predefined: particle number, number
of iterations, number of parameters, inertia factors, acceleration
constants and the search limit. The fitness of each particle is
measured in accordance with statistical criterion (the last square
method as objective function). When the specified number of iter-
ations is reached, the best search solution is used as information
in the process of restriction of the search space, where the most
sensible parameters are considered. This evaluation of the param-
eter sensitivity is possible in a graphical dialog, by using Maple®
software. The end of the search is determined by the better global
position of the swarm. For the PSO method application to equilib-
rium adsorption data, a swarm of 1000 particles with 30 iterations
was considered and a built-in objective function based on the least
square statistical method was used.
r2 = 0.996 for Cd) and 2 (1.889 for Cu and 0.574 for Cd), allowing to
extract further information such as equilibrium time. The biosorp-
tion of each metal has increased sharply at short contact and slowed
gradually as equilibrium was approached around 60 min contact
time and achieving approximately 52% and 58% removal for Cu
and Cd ions, respectively. This kind of behavior is typical for metal
biosorption characterized by no energy exchange reaction, and
thus, the metal removal is governed by a purely physical–chemical
interaction between the biosorbent and the metal-supplied liquid
phase.
For binary adsorption process (Fig. 1b), logistic-type functions
that allow extract further information from kinetic data are also
used, achieving a good correlation coefficient (r2 = 0.984 for Cu and
r2 = 0.994 for Cd) and 2 (0.583 for Cu and 0.142 for Cd). As can be
observed in Fig. 1b the same equilibrium time for binary adsorption
system was also achieved, with removal values of 28 and 25% for
copper and cadmium ions, respectively.
In binary system, the addition of simultaneously observed
removal values for copper and cadmium (approximately 53%)
resembles the 52–58% removal range that was observed in single
adsorption processes. In single adsorption systems, it was observed
that the cadmium removal is higher than copper one, whereas
for a Cu–Cd binary system the preference was inverted, having a
higher copper removal than cadmium one when both metallic ions
are simultaneously being adsorbed onto the E. crassipes biosorbent
surface. This phenomenon might be associated with the different
physical/chemical properties existing between copper and cad-
mium ions in an adsorption process or other ones on the biosorbent
surface.
The kinetic results in the present work were similar in order to
apply this kind of biosorbent in effluent treatment system, when
they are compared with other studied biosorbents in previous
works. Pietrobelli et al. [12], under the same experimental con-
dition, have reported that around 70% uptake of Cu(II) and Cd(II) Fig. 2. The kinetic data of (a) single and (b) binary sorption of copper and cadmium
was observed in single metal adsorption experiments using the ions for E. crassipes, at pH 5, 120 min and 30 ◦ C. Standard deviations are less than the
symbol size for copper and cadmium. All metal adsorption kinetic data were also
E. densa as biosorbent, with equilibrium times around 45–60 min. fitted by linear-type functions, according to the pseudo second order model.
Besides, in another work, Lodeiro et al. [34] have reported that
the cadmium removal and equilibrium time have been found as
tion of the sorption rate constant, namely K2 (g mequiv.−1 min−1 ),
being around 50% and 3 h, respectively, using Bifurcaria bifurcata,
in a simple way.
Saccorhiza polyschides, Pelvetia caniculata, Ascoplyllum nodosum,
Laminaria ochroleuca as biosorbents. Taking into account its low t 1 1
= 2
+ t (3)
equilibrium time, the E. crassipes as biosorbent could reduce the qt K2 qeq qeq
mass transfer zone in a continuous system for lower dimensions
For single and binary metal sorption data, obtained at pH 5 and
than other biosorbents or take the biosorption-based effluent treat-
30 ◦ C solution temperature, the uptake metal concentration in time,
ment systems for a long using time.
labeled as qt , by the E. crassipes biomass was calculated using Eq.
(1). The Ho’s second-order rate equation has fitted better the single
3.3. Adsorption kinetic
metal sorption data as well as the Cu–Cd binary one, according to
the good correlation coefficient values (r2 ≈ 0.999) and as shown in
In order to analyze the metal sorption kinetics in single and
Fig. 2. The sorption rate constant (K2 ) and the uptake metal capac-
binary systems, the pseudo-first order and pseudo-second order
ity at equilibrium (qeq ), which are summarized in Table 2, were
kinetic models were applied to the data. A simple pseudo first-order
estimated from the adjustable straight line parameters, accord-
equation, which was an early proposal by Lagergren, is described
ing to Eq. (3). Fig. 2a shows the single sorption data behavior for
by Eq. (2) [35,36]. From the linearization of the pseudo-first order
copper and cadmium ions and their linear fits, where the slope
kinetic equation, plotted as log(qeq − qt ) versus t, the adsorption
obtained from the linearity of the copper kinetic data is observed
rate constant (K1 ) can be calculated [37]:
K
1 Table 2
log (qeq − qt ) = log (qeq ) − t (2)
2303 The equilibrium uptake concentration (qeq ) and sorption rate constant (K2 ) of Cu
and Cd ions for E. crassipes, at pH 5 and 30 ◦ C, obtained from the linearity of the
where K1 is the rate constant of the pseudo first-order adsorp- adsorption kinetic experimental data in single and binary systems, according to the
tion process (min−1 ), qeq and qt (in mequiv. g−1 ) are the adsorption pseudo second-order model.
amounts at equilibrium and in time, respectively.
System Metal qeq (mequiv. g−1 ) K2 (g mequiv−1 min−1 )
On the other hand, the Ho’s second-order rate equation, which
Copper 0.323 ± 0.002 2.344 ± 0.032
has been called a pseudo-second order kinetic expression, has also Single
Cadmium 0.507 ± 0.005 0.717 ± 0.132
been applied widely [38,39] and described by Eq. (3). For this case, it
was convenient to plot the experimental data as t/qt against t, which Copper 0.245 ± 0.007 0.752 ± 0.150
Binary
Cadmium 0.214 ± 0.023 1.558 ± 0.016
shows a linear tendency of the data and allows for the determina-
48 A.N. Módenes et al. / Chemical Engineering Journal 168 (2011) 44–51
Table 3
Isotherm models used to fit the single metal adsorption data.
Langmuir qeq = qmax [(bL Ceq )/(1 + bL Ceq )] Mono layer Homogeneous surfaces Two theoretical parameters
Freundlich qeq = kF (C)nF Multi layer Heterogeneous surfaces Two empirical parameters
Temkin qeq = B ln(kT Ceq ) Mono layer Chemisorption Two theoretical parameters
Single n 1/n
Toth qeq = qmax (bT Ceq /(1 + (bT Ceq ) T ) T ) Multi layer Heterogeneous surfaces Three empirical parameters
Sips qeq = qmax ((bS Ceq )m )/(1 + (bS Ceq )m ) Combined Langmuir–Freundlich Three semi-empirical parameters
Redlich–Paterson qeq = kRP Ceq /1 + aRP Ceq g Heterogeneous surfaces Three semi-empirical parameters
A.N. Módenes et al. / Chemical Engineering Journal 168 (2011) 44–51 49
Table 4
Single metal adsorption modeling parameters for Langmuir, Freundlich, Toth, Sips,
Temkin and Redlich–Peterson isotherm models and their respective fitting statistical
parameters, for Cu and Cd adsorption data obtained at pH 5, 30 ◦ C and 120 min
equilibrium time.
where q∗exp and q∗pred (in mequiv. g−1 ) are the quantities of equi-
librium adsorbed metal by E. crassipes biosorbent, which were
Table 5
experimentally obtained in binary metallic system and predictable Binary metal adsorption modeling parameters for extended modified Langmuir
by modeling, respectively. isotherm model obtained from the Cu–Cd adsorption data (pH 5, 30 ◦ C and 120 min
In the case of a mixture of Cu(II) and Cd(II) ions, it can be seen equilibrium time).
that the modified extended Langmuir model provides adequate Metallic ion Parameters Estimated values by the PSO method
prediction of the binary data, including the single one. Fig. 4 shows
qmax1 0.615
a wide visual perspective of adsorption behavior as a mixture of
bL1 0.444
Cu(II) and Cd(II) ions in different concentration values is consid- Cu(II)
nL1 0.7044
ered. Regarding the modeling of binary adsorption data, the values r2 0.9864
of the maximum adsorption capacities (qmax,i ) and affinity param- qmax2 0.667
eter (bLi ) for Cu(II) and Cd(II) ions were kept the same that those bL2 0.566
obtained from the fits of the single adsorption data by Langmuir Cd(II) nL2 1.3649
isotherm model. Adjustable parameters were estimated by the PSO r2 0.9894
OF 0.062002
method and their values and indicators of the goodness of fit (r2 and
OF) are summarized in Table 5. In Fig. 5, the obtained simulation
results by using the PSO method have been compared to the equi-
50 A.N. Módenes et al. / Chemical Engineering Journal 168 (2011) 44–51
4. Conclusion
The results of this study suggest that the dry E. crassipes biosor-
bent shows a significant capacity to absorb of Cu(II) and Cd(II)
ions in single adsorption system as well as in the binary ones.
Batch tests have shown that the best experimental condition to
adsorb metal divalent ions onto sorbent surface has been achieved
by working at pH 5 and 30 ◦ C metallic solution temperature. It
was observed that there is no significant influence on the metal-
lic ion removal performance when a specific grain size range or a
mixture of them was used. From the kinetic tests, a short equilib-
rium time around 60 min was obtained, indicating that the dry E.
Fig. 5. Linear correlation between the experimental maximum adsorption capacity
values and PSO-adjusted, Langmuir one for (a) copper and (b) cadmium ions in their
crassipes biomass could be used in adsorption-based continuous
mixtures for E. crassipes at pH 5, 120 min equilibrium time and 30 ◦ C. treatment systems. In addition, the pseudo-second order model
has fitted better the single metal sorption data as well as the
Cu–Cd binary one. The equilibrium data obtained from single metal
adsorption systems were interpreted on the basis of a set of six
librium experimental data of the binary system. The quality of the isotherm models, resulting that the best description was assigned
predicted model results was assessed by the sum of the squared to the Langmuir one. It was found that the maximum uptake of
errors between the equilibrium experimental data points and the Cd(II) ions (0.667 mequiv. g−1 ) by the E. crassipes is higher than
theoretically predicted ones (0.01134 for Cu and 0.01590 for Cd). that of Cu(II) one (0.615 mequiv. g−1 ), revealing the following metal
According to the modified extended Langmuir sorption model, affinity order Cd(II) > Cu(II). On the other hand, the equilibrium
the presence of other metals in the solutions affects the shape of the data of a mixtures of Cd(II) and Cu(II) ions were graphically rep-
isotherm curve, due to changes in apparent metal affinity for the resented by 3D adsorption surface and modeled by an modified
active sites. Regarding the value of correction parameter of species extended-to-multi-component Langmuir-type isotherm, obtaining
i (nLi ), if the biosorbent has different affinity between both metal good fits. The PSO method has allowed the prediction of all mod-
ions then different nLi values are expected to be predicted by the ified extended Langmuir parameters. Under iso-concentration of
proposed model, instead of nLi ones equals to one. In the present metallic ions in binary solutions, the predicted maximum capacity
work, the estimated values of correction parameter of species i of Cu(II) was estimated to be 0.418 mequiv. g−1 , while the pre-
were 0.7044 and 1.3649 for Cu and Cd, respectively. Performing the dicted qmax for Cd(II) was estimated to be 0.207 mequiv. g−1 . In
evaluation of this result (see Table 5), it is observed that the ratio addition, a sum of these predicted maximum metal capacities is
of affinity to correction parameters (bLi /nLi ) are 0.6303 and 0.4147 very near to those obtained in single metal adsorption systems.
for Cu and Cd ions, respectively, indicating that the corrected affin- Finally, the results have shown that the affinity of each metal ion
ity parameter of Cu(II) by the biosorbent E. crassipes is higher than onto the E. crassipes surface is influenced by the presence of the
Cd(II) in the binary adsorption system. other one.
A.N. Módenes et al. / Chemical Engineering Journal 168 (2011) 44–51 51