You are on page 1of 12

Chemosphere 41 (2000) 271282

Ionic strength eects in biosorption of metals by marine algae


S. Schiewer *, M.H. Wong
Department of Biology, Institute for Natural Resources and Waste Management, Hong Kong Baptist University, Kowloon Tong,
Hong Kong SAR, People's Republic of China

Abstract
Biosorption, the passive accumulation of metals by biomass, can be used as a cost-eective process for the treatment
of metal polluted industrial euents. The green alga Ulva fascia and the brown seaweeds Sargassum hemiphyllum,
Petalonia fascia, and Colpomenia sinuosa were characterized in terms of their number of binding sites, their charge
density and intrinsic proton binding constant (pKa ) using pH titrations at dierent ionic strengths. The determined
number of binding sites decreased in the order Petalonia P Sargassum > Colpomenia > Ulva. Due to their high number
of binding sites Sargassum and Petalonia are most promising for biosorption applications. The decrease of proton
binding with increasing ionic strength and pH as well as the increase of Cu and Ni binding with increasing pH and
decreasing ionic strength could be described by the Donnan model in conjunction with an ion exchange biosorption
isotherm. 2000 Elsevier Science Ltd. All rights reserved.

Glossary
C
CH
CM0:5
Ct
H
I
KH
KM
M
m
pKa
qH
qM

amount of free binding sites


(mequiv/g)
amount of protonated binding sites
(mequiv/g)
amount of metal binding site
complexes (mequiv/g)
total amount of binding sites
(mequiv/g)
proton
ionic strength (mmol/l)
equilibrium constant for the binding
of protons (l/mmol)
equilibrium constant for the binding
of metal ion (l/mmol)
metal ion (e.g., Cu, Ni)
dry weight of biomass (g)
log of acid dissociation constant
(dimensionless)
amount of protons bound to the
biosorbent (mequiv/g)
amount of metal bound to be
biosorbent (mequiv/g)

Corresponding author.

V
Vm
YV
[]
k
Indices
add
app
des
i
p

volume of solution (l)


specic cation binding volume per
mass of biosorbent (ml/g)
parameter for swelling of Vm
(ml/mequiv)
concentration of species in brackets
(mmol/l)
concentration factor intraparticle/
bulk (dimensionless)
added
apparent
desorbed
initial
intraparticle

1. Introduction
The passive (not metabolically mediated) binding of
metals by living or dead biomass is commonly referred
to as biosorption. This process can be used to remove
and recover toxic heavy metals from metal bearing industrial wastewaters (Volesky, 1990). Compared to
conventional techniques such as precipitation or
synthetic ion exchange resins, biosorption oers the

0045-6535/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 5 - 6 5 3 5 ( 9 9 ) 0 0 4 2 1 - X

272

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

advantage of rendering it possible to achieve drinking


water quality at low cost. This is the case because
cheaply available biomass, such as waste products from
other industries (e.g., food and fermentation industry)
or abundant marine algae, may be used as eective
biosorbents.
Algae possess a high metal binding capacity (Ramelow et al., 1992; Holan and Volesky, 1994) whereby the
cell wall plays an important role in metal binding (Crist
et al., 1988; Kuyucak and Volesky, 1989). This is due to
the presence of carboxyl and sulfate groups in algal cell
wall polysaccharides which can act as binding sites for
metals. Alginate, which is composed of mannuronic and
guluronic acids, is a major polysaccharide in brown algae and oers carboxyl groups (Lee, 1980). It constitutes
about 1040% of the brown algal dry weight (Percival
and McDowell, 1967). For Sargassum alginate contents
between 17% and 45% have been reported (Chapman,
1980; Fourest and Volesky, 1996) which corresponds to
0.850.25 mequiv/g of carboxyl groups per dry weight.
Between 5% and 14% alginate (0.250.7 mequiv/g
carboxyl groups) were determined in Colpomenia (Kalimuthu et al., 1991). Brown algae also contain about 5
20% of the sulfated matrix polysaccharide fucoidan
(Chapman, 1980) about 40% of which are sulfate esters.
0.27 mequiv/g of sulfate groups are reported for Sargassum (Fourest and Volesky, 1996). Alginate and fucoidan are known for their metal binding properties
whereby ion exchange between metal ions occurs (Haug
and Smidsrod, 1970). The cell wall matrix of green algae
contains complex heteropolysaccharides which also oer
carboxyl and sulfate groups (Lee, 1980). The extracted
polysaccharides from Ulva (12% of the algal dry weight)
contained 16% sulfate and 1519% uronic acids (McKinnell and Percival, 1962; Lahaye and Axelos, 1993).
Protein can constitute 1070% of the green algal cell wall
(Siegel and Siegel, 1973). A protein content of 1525% is
reported for Ulva whereby aspartic and glutamic acid
account for 12% of the protein which corresponds to
0.15 mequiv/g of carboxyl groups per dry weight
(Chapman, 1980; Lee, 1980; Guiry and Blunden, 1991).
Lysine and arginine make up 13% of the protein (0.08
mequiv/g of amine groups). In brown algae the protein
content is less than 30% (Siegel and Siegel, 1973). For
Sargassum 10% protein (0.17 mequiv/g carboxyl
groups) while lysine and arginine made up 11% of the
protein (0.07 mequiv/g amine groups) (Chan et al.,
1997). This means that in brown algae the carboxyl
groups of alginate are more abundant than either
carboxyl or amine groups of the protein and are therefore likely to be the main binding sites. Sulfate groups
appear to be of secondary importance.
Both carboxyl and sulfate groups are acidic. At low
pH they will be protonated and thereby become less
available from the binding of metals, which explains
why the binding of many metals increases with increas-

ing pH (Greene et al., 1987; Ramelow et al., 1992). The


competition between protons and metal ions for the
same carboxyl and sulfate binding sites has been described by a pH sensitive biosorption isotherm model
which allows the prediction of pH eects on metal
binding and the amount of protons bound (Schiewer
and Volesky, 1995a).
Apart from the pH value, another important parameter in biosorption is the ionic strength. Sodium is
common in many wastewaters and high Na concentrations lead to high ionic strengths at which the amount of
heavy metals bound is reduced (Greene et al., 1987;
Ramelow et al., 1992). The eect of Na is more pronounced regarding the uptake of weakly bound metals
such as Zn or Ni. Strongly bound metals such as Cu are
less aected by the ionic strength. Sodium as a typical
ard'' ion (Stumm and Morgan, 1996) will not be bound
covalently and does therefore not compete directly with
the covalent binding of heavy metals by the biosorbent.
The eect of ionic strength can however be explained as
the result of competition of Na with the heavy metals for
electrostatic binding to the biomass. Since deprotonated
free carboxyl or sulfate groups are negatively charged,
they will electrostatically attract any cation. Algal biomass particles may have an overall negative charge the
magnitude of which increases with increasing pH as
more sites are deprotonated (Schiewer and Volesky,
1995a). A variety of models based on electrostatic eects
have been employed to describe the eect of ionic
strength on the binding of heavy metals and protons by
humic and fulvic acids (Stumm and Morgan, 1996).
Since algal particles are large compared to the thickness
of the electrical double layer, the negative charge of the
biomass will be completely balanced within the particle.
Therefore the relatively simple Donnan model can be
applied which assumes a homogenous concentration of
counterions throughout the polyelectrolyte phase. This
model in conjunction with isotherm equations for the
covalent binding of protons and metal ions has been
applied for predicting proton and metal binding by
Sargassum biomass (Schiewer and Volesky, 1997a,b).
Comparative biosorption studies have indicated that
Sargassum binds higher amounts of most metals than
other algae including Colpomenia and Ulva but failed to
explain what causes this superiority of Sargassum (Zhao
et al., 1994; Da Costa and de Franca, 1996).
In the present work the binding of protons and metal
ions by three brown seaweeds Sargassum hemiphyllum,
Colpomenia sinuosa and Petalonia fascia as well as the
marine green alga Ulva fascia will be investigated as a
function of metal concentration, pH and ionic strength.
Dierences in overall biosorption behavior are explained
as a result of dierent numbers of binding sites, anities
for metal complexation and charge density and predicted using the Donnan model combined with isotherms for covalent binding of metals and protons.

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

2. Materials and methods


2.1. Biomass
Three brown seaweeds (Sargassum hemiphyllum,
Colpomenia sinuosa and Petalonia fascia) as well as one
green alga (Ulva fascia) were employed as biosorbents.
Ulva and Sargassum are the most abundant green and
brown seaweeds in Hong Kong, respectively, and occur
globally widespread and in large quantities. Ulva and
Petalonia were collected in March 1998 in Big Wave Bay
(Hong Kong Island), Colpomenia and Sargassum were
harvested in April 1998 in Saikung and January 1998 in
High Island, respectively, both located at the eastern
coast of the New Territories. All samples were collected
from living material.
The seaweed was dried in an oven and ground in
a blender. About 20 g of the size fraction 12 mm
( average size after blending) were washed in 1 l DI
water to remove the major part of salt from the seawater. Then each biomass was washed in 1 l of 0.1 M
HNO3 for protonation, subsequently 20 times in 1 l of
DI water (to remove excess acid) and nally oven-dried
at 50C. The protonation was performed in order to
remove unknown quantities of light metal ions from the
seawater and to ensure that all binding sites were converted to their acidic form, i.e., saturated with protons.
2.2. Metal binding and pH titration experiments
The metal ions Cu and Ni were chosen because Cu is
known to bind strongly in many biomaterials whereas
Ni binds weakly through mostly electrostatic interactions. It can therefore be expected that these two metals
display very dierent binding behavior. The nitrate salts
CuNO3 2  2:5H2 O and NiNO3 2  6H2 O (Riedel de
Haen) were dissolved in DI water. Nitrate was chosen as
the anion because of its low tendency for complex formation with most metals.
0.1 or 1.0 g of biomass was contacted with 40 ml
metal and/or Na salt solution in 100 ml Erlenmeyer
asks on a Labline orbit shaker at 150 rpm. The pH
titrations at low ionic strength were performed using 0.1
g biomass (without the addition of Na salt), at medium
ionic strength 1 g biomass was used (no Na salt added)
and for high ionic strength 1 g biomass was employed in
a solution of 100 mM NaNO3 . In the metal (Cu or Ni)
binding experiments 0.1 g of biomass was used and the
NaNO3 concentration was zero for low ionic strength
and 60 mM for high ionic strength experiments. In both
types of experiments known amounts of 0.1 M NaOH or
HNO3 were added for adjustment of the pH value,
measured on a Orion 290A pH meter. The samples were
then left shaking overnight for complete equilibration
and the nal pH was recorded. Initial and nal metal
concentrations [M]i and [M] in the experiments involv-

273

ing heavy metal binding were measured by atomic absorption spectrometer (Varian AA-20). The initial
concentrations ranged from 10 to 1000 lg/ml.
2.3. Determination of the equilibrium cation binding
2.3.1. Protons
The mass balance for protons is
qH;i m Hadd VH;add OHadd VOH;add Hi Vi
qH m HV

mM;

where [H] is the concentration of protons in solution, qH


the amount of protons bound to the biomass, m the
mass of biosorbent and V is the solution volume. The
subscript i denotes initial values, otherwise nal values
are referred to. [H]add and [OH]add are the concentrations
of added acid or base respectively, with the volumes of
acid or base added being VH;add and VOH;add .
The amount of protons bound at equilibrium can be
obtained by solving Eq. (1) for qH
qH qH;i Hadd VH;add OHadd VOH;add Hi Vi
HV =m

mequiv=g:

For protonated biomass, the initial proton binding


qH;i equaled the total number of binding sites Ct except
for Ulva which contained 0.4 mequiv/g of excess
acid (which led to pH 3 when 0.1 g of Ulva were
equilibrated in 40 ml of DI water) so that qH;i Ct +
0.4 mequiv/g.
2.3.2. Metals
The mass balance for a divalent metal ion M is
0:5qM;i m Mi Vi 0:5qM m MV

mM;

where qM is the amount of metal bound to the biomass,


and [M] is the metal concentration in solution. Solving
this equation for qM and assuming that no metal is
bound initially (i.e., qM;i 0) yields
qM 2Mi Vi MV =m

mequiv=g:

The factor 2 is required to convert from molar to


equivalent concentration.
2.3.3. Desorption and swelling
Since Eq. (4) may lead to large errors in the calculation of qM if the dierence between [M]i and [M] is
small (Schiewer and Volesky, 1995b), the metal binding
qM was determined by desorption if [M] was larger that
50% of [M]i . For desorption and to measure the particle
swelling, the solution after adsorption was decanted or
ltered o and the wet metal laden biomass blotted on a
dry lter paper to remove excess liquid. The biomass
(without the lter paper) was then transferred onto
a weighing tray of a known mass and weighed

274

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

immediately (to avoid evaporation). The biomass on the


weighing trays was dried in an oven at 50C and weighed
again after being equilibrated at room atmosphere to
regain the ``normal'' moisture content.
To measure particle swelling, the specic wet particle
volume Vp (ml/g) was calculated as the wet weight mwet
(i.e., the volume, assuming a density of 1.0) divided by
the dry weight mdry both after adsorption and before
desorption.
For desorption, the biomass was then transferred into
40 ml of 0.1 M (or 0.9%) HNO3 in 100 ml Erlenmeyer
asks and equilibrated overnight on a Labline orbit
shaker at 150 rpm. The liquid after desorption was than
analyzed by AA to determine the desorbed metal concentration [M]des . The metal binding after adsorption
(and before desorption) was calculated as
qM 2Mdes Vdes Mmwet mdry
 1 ml=g=mdry Mdes Vdes

mequiv=g;

where Vdes is the volume of the desorbing solution with


the nal concentration [M]des after desorption equilibrium is reached. The term [M] (mwet )mdry ) is subtracted
in order to account for metal that is not bound to
binding sites but contained in the absorbed or adhering
water of the volume mwet mdry  1 ml/g. For an accurate determination of qM , it is desirable that this term,
i.e., mwet is kept as small as possible. This is the reason
for blotting the biomass dry before desorbing it and for
not drying and desorbing it together with the lter paper. The enumerator of Eq. (5) is the moles of metal
bound to the biomass after adsorption. The factor 2
converts this into mequiv. The denominator is the dry
weight of the desorbed biomass excluding the weight of
the metal bound to it. For Sargassum this value ranged
between 93% and 106% of the initial weight m of the
biomass before adsorption with the exception of one
data point of only 85% recovery.

3. Results and discussion


3.1. pH drop in DI water
When 0.1 g of the protonated biomass was equilibrated in 40 ml of deionized water, the nal pH ranged
between 5.1 (for Petalonia) and 3.0 (for Ulva). The release of protons responsible for this drop of pH must
have been due to the presence of excess acid which had
not been removed by washing with DI water after
treatment with HNO3 . It cannot have been caused by a
release of protons from dissociated acidic groups because no other cations were present to balance the
negative charge of dissociated acidic groups. Therefore,
even if protons were dissociated from the acidic sites,
they would be retained in the biomass by electrostatic

attraction to the negatively charged groups. From the


pH drop in DI water the quantities of excess acid in the
biomass were calculated as 0.025, 0.016, 0.0032 and
0.40 mmol/g for Sargassum, Colpomenia, Petalonia and
Ulva, respectively. This quantity is only for Ulva signicant compared to the total number of binding sites.
3.2. pH titration experiments
The pH titrations of protonated biomass for all four
algae are shown in Fig. 1. When NaOH was added to
the protonated biomass equilibrated in DI water or
NaNO3 solution, the weakly acidic (carboxyl) groups
dissociated with increasing pH and the proton binding
(Eq. (2)) decreased correspondingly. From the inection
points of the titration curves the total amount of binding
sites was determined as 2.6, 1.5, 2.9 and 1.1 mequiv/g for
Sargassum, Colpomenia, Petalonia and Ulva, respectively
(see Table 1, Part a). These quantities correspond to the
maximum metal binding observed for Sargassum (Fig. 3)
and the other algae (Schiewer and Wong, 1999). Assuming that the amount of protons initially bound
equaled the total amount of binding sites plus the
amount of excess acid leads to a representation of the
titrations in which the proton binding approaches zero
at the endpoint of the titration of the weakly acidic
groups. However, this only means that no more protons
are bound to the titrated weakly acidic carboxyl groups.
Other binding sites with higher pKa values (which may
occur in smaller quantities in the biomass) may still be
protonated.
In the titration of Petalonia at low ionic strength a
slight transient decrease of pH in spite of NaOH addition
was observed when the proton binding was reduced from
2 to 1 mequiv/g. Further addition of NaOH then lead to a
resumption of the pH increase. This pH decrease can be
explained by the disintegration of Petalonia biomass
which occurred when the proton binding was reduced to
1.5 mequiv/g. The release of acids from the cytoplasm
which lend Petalonia its characteristic acidic taste could
have caused the described decrease of pH.
In all four types of biomass a pronounced eect of
ionic strength was observed. While the ionic strength did
not inuence the number of acidic groups titrated, it
strongly aected their apparent pKa values (i.e., the log
of the proton binding constant)
pKa;app log KH pH log 1 f =f :

Plotting pKa;app versus the degree of dissociation


f C =Ct (plots not shown) can give a clear picture of
the variation of pKa;app with f and I. At low ionic
strength (0.1 g biomass, no NaNO3 added) the apparent
pKa values at half dissociation (i.e., the degree of dissociation f 0.5) ranged between 5.3 for Ulva and 6.4
for Colpomenia with the average apparent pKa value

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

275

Fig. 1. Titration of protonated algal biomass at dierent ionic strengths. Experimental data and predictions of the Donnan model for
rigid particles (DORI): (a) Sargassum, (b) Colpomenia, (c) Petalonia, (d) Ulva.

being 5.8 (Table 1, Part b). At high ionic strength (1 g


biomass, 100 mM NaNO3 ) the average pKa value was
3.8, i.e., 2 units lower than at low ionic strength. This
means that the apparent proton binding constant
changed in average by a factor of 100. This shift of the
apparent pKa values with ionic strength was largest for
Colpomenia (2.5 pH units) and lowest for Ulva (1.3 pH
units).
The eect of ionic strength on proton binding can be
explained by electrostatic eects: at high ionic strength
Na balances most of the negative charges in the biomass
whereas at low ionic strength an electrostatic attraction
of protons leads to intraparticle concentrations that are
higher than the bulk proton concentration. This higher
intraparticle concentration leads to an increased amount
of protons covalently bound to the binding sites, i.e., to
an apparently higher binding constant.
3.3. pH titration model
The total amount of protons bound qH equals the
sum of the covalently (CH) and electrostatically bound
protons.

qH CH Hp HVm

mequiv=g:

The intraparticle concentration is designated by the


subscript p (e.g., ([H]p ), otherwise the bulk concentrations are referred to. The amount of protons bound by
electrostatic attraction equals the concentration dierence between the particle and the bulk solution multiplied by the specic cation binding volume Vm in the
particle. When describing the amount covalently bound
protons (CH) mathematically, care has to be taken to
distinguish apparent and intrinsic binding constants.
The apparent proton binding constant is related to the
proton concentration [H] in the bulk of the solution and
dened as
KH;app CH=C H

l=mol:

The intrinsic binding constants relate to the higher


intraparticle concentration of the cations near the negatively charged binding site, i.e., in this case to [H]p
KH CH=C Hp l=mol:

The relation between the apparent and intrinsic


proton binding constants is therefore

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

0.17

k Hp =H Nap =Na M2 p =M2 0:5 :

11

0.14

One main factor determining k is the ionic strength. k


decreases with increasing ionic strength, approaching
the value 1.0. Note that even if at high ionic strengths
the concentrations of H and M in the particle approach
the respective concentrations in the bulk (i.e., if there is
negligible electrostatic binding) there will still be covalent binding. The concentration factor k is also strongly
dependent on the number of charges per unit volume in
the biomass since it is the negative charge of the biomass
that leads to the electrostatic accumulation of cations in
the biomass that is reected in k > 1. The number of
charges per unit volume (mol/l) equals the number of
free ionized binding sites C divided by the particle
volume Vm . From Eq. (11) and the condition for charge
neutrality one can derive

0.14

0.20
0.17
0.24
0.07
1.30
0.75
1.45
0.55
1.25
0.15
1.53
0.39

10

The concentration factor [H]p /[H] can be modeled


according to the Donnan equilibrium (Donnan, 1911)
whereby the concentration of any ion is assumed to be
homogeneous throughout the biomass particle and the
negative charge of the biomass is balanced by counterions such as protons (H), sodium (Na) or divalent metal
ions (M) whose concentration factor is

k C 2IVm C =2IVm 2 10:5 :

0.60
0.66
1.91
6.44

0.20
0.07
0.24
0.06

0.20
0.07
0.24
0.06

Dqb
(mequiv/g)
Vm Vm 0.5
Bt (ml/g)

l=mol:

12

The ionic strength in solution is

3.3
3.8
5.8
Average

MATLAB optimized.
Absolute mean square deviations of model predictions from data for qH .
b

2.6
1.5
2.9
1.1

5.9
6.4
5.4
5.3

3.5
3.9
3.6
4.0

2.7
3.6
3.1
3.9

I 2M 0:5Na 0:5H 0:5L M L  M Li

Sargassum
Colpomenia
Petalonia
Ulva

With pKa 3

Vm , opta
(ml/g)
pKapp
pKapp

pKa , opta

High I
Low I

Optimum pKa

Vm , opta
(ml/g)

d
Intrinsic pKa
Apparent pKa
Bt

c
b
a
Part

Table 1
DORI model parameters for pH titrations of four algae at dierent ionic strengths

KH;app KCH Hp =H

Dqb
(mequiv/g)

Dqb
(mequiv/g)

276

M 2Mi Nai Hi Hexcess m Hadd VH;add


OHadd VOH;add =V

mM=l:

13

3.4. Modeling of pH titration data


For each algal biomass the intrinsic proton binding
constant KH and the specic cation binding volume Vm
of the particle were determined by nonlinear tting using
the program MATLAB to minimize the mean square
deviations between model and experimental data. The
results are listed in Table 1 (Part c). The average of the
optimized pKa values was 3.33. This is quite close to the
pKa values 3.65 and 3.38 of guluronic and mannuronic
acids, respectively, which are the components of alginate
(Haug, 1961).
In order to minimize the number of constants to be
determined and to ensure that the dierences in metal
binding constants reect dierences in the amount of
metal bound, it is desirable to use the same pKa value
for all algae if possible. Since the binding sites in all
algae employed in this study are mainly carboxyl groups
this was feasible. The optimum pKa value to be used for
all algae was 3.0. However, the exact choice of the pKa

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

had little eect on the model t since dierent anities


for protons could still be reected in dierent cation
binding volumes Vm . Choosing pKa 3.0 for all algae lead
only to a negligible increase in modeling error compared
to using the optimum pKa values individually determined for each alga. It would be similarly possible to use
other nearby pKa values (e.g., 3.3) for all algae with little
eect on the modeling outcome.
The optimized Vm values for a common pKa 3.0 are
also listed in Table 1 (Part d). The parameters for
Sargassum are similar to those obtained in earlier work
for biomass of this genus (Ct 2.1, pKa 2.8, Vm 1.4)
(Schiewer and Volesky, 1997a). The optimum cation
binding volume Vm decreased in the order Petalonia > Sargassum > Ulva > Colpomenia. This order can be
partially explained by the fact that Petalonia and Sargassum are the algae with the highest number of binding
sites. It is therefore only to be expected that a larger
volume in these algae will be involved in cation binding.
The larger optimized Vm for Ulva compared to Colpomenia can not be explained by the number of binding
sites. However, since the matrix polysaccharides in green
algae like Ulva are heteropolysaccharides, their carboxyl
groups are not as concentrated but distributed over a
relatively larger volume. This can explain that the cation
binding volume in Ulva is disproportionally large.
To further reduce the number of tting parameters, it
was investigated whether it is possible to assume that Vm
is proportional to Ct , e.g.
Vm 0:5 ml=mequivCt

ml=g:

14

For Sargassum and Petalonia this leads to Vm values


similar (less than 10% dierence) to the optimized ones
(Table 1, Part e). For Colpomenia Vm is much higher
than the optimized value (i.e., using Eq. (14) leads to an
underestimation of the ionic strength eects as shown in
Fig. 1) while for Ulva Vm is 40% lower than the optimized value (probably due to the fact that heteropolysaccharides occur in green algae as indicated above).
Nevertheless, the average modeling error was only 0.03
mequiv/g larger than when individually tted Vm parameters were used (Table 1, Part d). The overall t for
the model using the same pKa 3 for all algae and assuming Vm 0.5Ct is quite good as depicted in Fig. 1.
To model the 11 titration curves only 6 parameters were
necessary: the number of binding sites Ct in each of the
four algae and the common values for KH and Vm /Ct .
3.5. Metal binding model
When the Donnan model is applied for metal binding, the concentration factor k can be calculated from
the equation for charge neutrality in the particle, expressing all intraparticle concentrations in terms of their

277

bulk concentrations and the concentration factor k


which yields (Schiewer and Volesky, 1997b)
2
2
k I 3M=4M I 3M =16M

I M C =Vm =2M0:5 :

15

The calculation of the metal binding is iterative


starting with evaluating Eq. (15) with an estimated
number of free carboxyl sites C . The obtained value for
k is then used to calculate the intraparticle concentrations according to Eq. (11).
Regarding covalent binding, the following reactions
are considered:
H C $ CH

KH CH=Hp C ;

M2 2C $ 2CM0:5

16

KM CM0:5 2 =Mp C 2 :
17

Divalent metal ions are assumed to bind to two


monovalent C . The mass balance for the binding sites
which are either free (ionized) or occupied by protons or
metal ions is
Ct C CH CM0:5

mequiv=g:

18

From these equations a pH sensitive isotherm equation can be derived which allows for the calculation of
the amount of metal and protons bound covalently
(Schiewer and Volesky, 1995a).
CH Ct KH Hp =1 KH Hp
KM Mp 0:5 mequiv=g:

19

CM0:5 Ct KM Mp 0:5 =1 KH Hp
KM Mp 0:5 mequiv=g:

20

Subsequently the isotherm equations (19) and (20)


are evaluated. This renders the calculation of C according to Eq. (18) possible and the above routine is
repeated for an iterative calculation of C till stable
values are obtained, unless C /Vm is a constant (Eq.
(22)), in which case no iterations are necessary. Finally
the eletrostatic binding is added to the covalent binding
for protons (Eq. (7)) and metal ions
qM CM0:5 2Mp MVm

mequiv=g:

21

3.6. Modeling metal binding data


For the constant cation binding volume (no particle
swelling) proportional to the number of binding sites
(Eq. (14)) and pKa 3 as determined from the pH titrations, the metal binding constants KM were determined using MATLAB to minimize the mean square
deviations between model and experiments for Cu

278

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

binding at pH 4 at low ionic strength (0.1 g biomass, no


NaNO3 added). The values are listed in Table 2. For Vm
from the titration experiments the model ts about
equally well as when Vm is optimized for the metal
binding data along with KM . The metal binding constants decrease in the order Sargassum > Petalonia > Colpomenia  Ulva  0. Model predictions and
data are shown for Sargassum only in Fig. 3 (concerning
data for the other algae the reader is referred to
(Schiewer and Wong, 1998)). The much smaller binding
constant in Ulva may be caused by the fact that in green
algal heteropolysaccharides two carboxyl binding sites
rarely occur at a suitably near distance to allow for
bridging of a metal ion between two carboxyl groups as
it occurs in metal binding to alginate. In fact, metal
binding by Ulva can be modeled with KM 0 (i.e., no
covalent binding occurs) which leads to the error
Dq 0.15 mequiv/g which is negligibly larger than
Dq 0.14 mequiv/g for the optimum KM .
It is, however, to be questioned whether a model
assuming constant cation binding volume Vm should be
used: swelling of the biomass particles occurred, especially at high ionic strength. While at low ionic strength
the volume Vp of Sargassum particles remains between 2
and 3 ml/g, (data not shown), swelling of Sargassum
occurs at high ionic strength (0.1 g biomass, 60 mM
NaNO3 ) and low metal concentration as shown in Fig. 2.
Both for Ni and Cu the particle volume increases by a
factor of 5 with decreasing metal concentration.
This can be explained by the fact that divalent metals
are able to form crosslinks between dierent alginate
chains and thereby reduce swelling and prevent solubilization (Kohn, 1975). While alginic acid and its salts
with divalent metal ions (except Mg) are insoluble, alginates of monovalent cations such as Na are soluble and
leaching of alginate could occur at high ionic strength if
the sites are not saturated with protons or divalent
metals (Percival and McDowell, 1967).
The other types of biomass swell more easily than
Sargassum, i.e., they are less stable and therefore less
suitable for biosorption applications. Ulva displayed a
larger particle volume > 4 ml/g even for low ionic
strength and high metal concentration. Petalonia had a
similar particle volume to Sargassum at low I and high

metal concentration but swelling already took place at


low ionic strength when the metal concentrations were
low to intermediate (data not shown).
Since Sargassum combines the advantages of natural
occurrence in large quantities, a high number of binding
sites, a high metal binding constant and a higher stability than e.g., Petalonia, Sargassum was chosen for
further studies of ionic strength eects in the binding of
Cu and Ni.
The data for the binding of heavy metals and protons
by Sargassum at high and low ionic strength for Cu and
Ni are shown in Fig. 3. The optimum Vm and KM to t
these data were determined using MATLAB and are
listed along with the resulting model deviations in Table
3. If the same Vm /Bt 0.5 as for the pH titrations is used
and only KM is optimized, the average model deviations
increase esp. for Ni as also indicated in Table 3.
Better model ts may be obtained by taking the observed swelling into account. Especially at high ionic
strength, swelling increases with decreasing metal concentration because more free sites which are not bound
to metal ions occur. A linear increase of Vm with the
number of free carboxyl sites C was assumed, as proposed in previous work (Schiewer and Volesky, 1997b),
i.e.,
Vm YV C

ml=g;

22

where YV is a tting parameter and C is the number of


free carboxyl groups (that do not covalently bind to any
cation). With Eq. (22), i.e., for the Donnan model with
particle swelling (DOSW), the following equations can
be derived by combining the Donnan model Eq. (11)
with the isotherm equations (19), (20) (Schiewer and
Volesky, 1997b).
qH Ct KH H HYV 1 1=k=1=k
KH H KM M0:5 mequiv=g;
qM Ct KM M0:5 2MYV k 1=k=1=k
KH H KM M0:5 mequiv=g:

Vm , tita (ml/g)

KM , optb (l/mM)

Dqc (mequiv/g)

1.30
0.75
1.45
0.55

0.23
0.064
0.17
0.000063

0.31
0.20
0.29
0.14

As determined in titration experiments.


MATLAB optimized.
c
Absolute mean square deviations of model predictions from data for qH and qM .
b

24

These equations for the Donnan model which take


swelling into account (DOSW model) were used in the

Table 2
DORI model parameters for Cu binding by dierent algae at pH 4 and low ionic strength, for pKa 3.0
Sargassum
Colpomenia
Petalonia
Ulva

23

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

Fig. 2. Swelling of Sargassum: particle volume Vp at high ionic


strength.

279

modeling of the experimental data in Fig. 3. The model


parameters YV and KM as determined by MATLAB
tting are listed in Table 3. If YV was individually tted
for Cu and Ni, respectively, the t was slightly better
than when YV 5 was used to model the binding of both
metals (shown in Fig. 3) which however had the advantage of requiring one less tting parameter.
Electrostatic eects are important in metal biosorption in two respects. The rst is an indirect multiplicative
eect: due to the higher intraparticle concentrations the
covalent binding of metal ions or protons will be higher.
This is reected in the modeling by the factor 1/k in the
denominator of Eqs. (23) and (24) (i.e., if denominator
and enumerator were both multiplied by k this would
imply that due to electrostatic attraction all binding
constants are multiplied by k). The second is a direct
additive eect: a signicant fraction of the total amount
of metal bound may be bound not covalently but simply
by electrostatic attraction (for protons this direct eect
was negligible under the conditions of this study). This is
reected in the model e.g., in Eqs. (23) and (24) by the

Fig. 3. Ionic strength eects in the binding of protons and metals at pH 4 by protonated Sargassum biomass. Experimental data and
Donnan model for swelling particles (DOSW) with common pKa and YV as well as modeling taking into account: only electrostatic
eects (``elst'', with KH KM 0) or only covalent binding (``cov'', with k 1.0). (a) Proton binding in the presence of Cu, (b) Cu
binding, (c) proton binding in the presence of Ni, (d) Ni binding.

0.23
0.0077
5
5
0.24
0.14
0.23
0.0077
b

MATLAB optimized.
Absolute mean square deviations of model predictions from data for qH and qM .
c
As determined in titration experiments.

4.9
10.3
0.33
0.28
0.18
0.0041
1.3
1.3
0.29
0.15

KM,
opta
(l/mM)
Vm , titc
(ml/g)
Dqb
(mequiv/g)
KM , opta
(l/mM)

0.22
0.0024
4.0
6.7
Cu
Ni

KM , opta
(l/mM)
YV
(ml/g)
Dqb
(mequiv/g)
KM , opta
(l/mM)
DOSW

YV , opta
(ml/g)
Vm , opta
(ml/g)

Dqb
(mequiv/g)
DORI

Table 3
Donnan model parameters for Cu and Ni binding by Sargassum at pH 4 for dierent ionic strengths, with pKa 3.0

0.24
0.16

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

Dqb
(mequiv/g)

280

terms [H] YV (1)1/k) and 2[M] YV (k)1/k) in the enumerator for protons and metals, respectively.
The importance of electrostatic eects might be best
visualized by considering the hypothetical cases that:
(a) only electrostatic but no covalent binding occurs i.e., the covalent binding constants
KH KM 0;
(b) only covalent binding but no electrostatic attraction occurs i.e., the concentration factor
k 1.0, which means [H] [H]p , [M] [M]p .
The results for modeling these two limiting cases with
the same constants as for the complete DOSW model
are also shown in Fig. 3.
Case a: When only electrostatic attraction is taken
into account, the binding of protons in the presence of
either Cu or Ni is reduced to almost zero. This means
the amount of protons bound directly by electrostatic
binding is negligibly small. The binding of Cu is significantly reduced, especially at high ionic strength, but
nevertheless a substantial amount of Cu is bound by
electrostatic attraction alone. The binding of Ni is reduced at high I but actually increased at low I because of
the reduced competition by protons.
Case b: If only covalent binding is considered, obviously no ionic strength eects occur. The binding of
protons is reduced to 0.15 mequiv/g both in the presence of Cu and Ni. Comparing this proton binding
(which is due purely to the intrinsic binding constants)
with the complete model reveals the importance of the
indirect multiplicative eect of electrostatic attraction.
The binding of Cu is intermediate to that obtained by
modeling only electrostatic eects at low and high I,
respectively. This means electrostatic and covalent
binding are of comparable magnitude whereby electrostatic eects are dominant at low I and covalent binding
at high I. The binding of Ni is for the most part lower
than considering electrostatic eects alone. This shows
the predominant importance of electrostatic eects in
the binding of Ni.
The larger importance of covalent binding for Cu
compared to Ni is due to the fact that while electrostatic
attraction is equally strong for both metals, the covalent
binding constants are 3090 times higher for Cu than for
Ni (Table 3). This corresponds to the fact that Cu has a
higher Nieboer class ``b'' index than Ni: the higher
electronegativity to Cu leads to more equal (covalent)
sharing of electrons with a ligand atom like oxygen as it
occurs in carboxyl groups (Stumm and Morgan, 1996).
4. Conclusions
The number of titrated weakly acidic carboxyl
groups was 2.9 for Petalonia, 2.6 for Sargassum, 1.5 for
Colpomenia and 1.1 for Ulva. The titration curves for all
algae showed a marked eect of ionic strength: the

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

apparent proton binding constant at low ionic strength


was in average 100 times higher than at high ionic
strength. This can be explained as a result of electrostatic attraction of protons to negatively charged carboxyl sites which results in intraparticle proton
concentrations that are higher than the bulk concentrations in solution such that proton binding sites so that
the intraparticle proton concentration approaches the
concentration of protons in the bulk of the solution.
The Donnan model was successfully used to account
for the ionic strength eects in pH titrations and in metal
binding. The same pKa value 3.0 could be used for all
algae and it was assumed that the cation binding volume
is proportional to the number of binding sites. The Cu
binding constants decreased in the order Sargassum > Petalonia > Colpomenia > Ulva  0. Virtually no
covalent metal binding occurred in Ulva possibly because green algae, which lack alginate, do not oer
carboxyl groups spaced at a suitable distance for
bridging of one metal ion between two binding sites. The
results indicate that brown algae are more suited for
biosorption applications than Ulva because of their high
metal binding capacity and anity. Sargassum excels in
also being more stable (less swelling and leaching) and is
therefore the most promising among the seaweeds
studied.
In metal binding experiments at high ionic strength
swelling of the biomass particles was observed. The
model t improved compared to the Donnan model for
rigid particles when particle swelling proportional to the
number of free binding sites was assumed. Electrostatic
attraction to the negatively charged binding sites increased the binding of cations as a combination of two
eects. First, the covalent binding of protons and metals
is increased due to the elevated concentrations of these
ions in the vicinity of the binding sites. Second, a signicant amount of divalent metals can be bound electrostatically in the biomass. For protons this quantity is
negligible compared to the amount covalently bound
because protons as monovalent ions experience less
electrostatic attraction, their binding is mostly covalent.
The intrinsic binding constant for Cu was 3090
times higher than that for Ni. Correspondingly, covalent
binding was more important for Cu than for Ni which
was bound predominantly by electrostatic attraction.

Acknowledgements
The authors would like to thank The Croucher
Foundation for nancial support. The second author
received the Senior Research Fellowship when the study
was carried out.

281

References
Chan, J.C.C., Cheung, P.C.K., Ang, P.O., 1997. Comparative
studies on the eect of three drying methods on the
nutritional composition of seaweed Sargassum hemiphyllum
(Turn.) C. Ag., J. Agricult. Food Chem. 45, 30563059.
Chapman, V.J., 1980. Seaweeds and Their Uses. Chapman and
Hall, London, UK, pp. 89, 194240.
Crist, R.H., Oberholser, K., Schwartz, D., Marzo, J., Ryder,
D., Crist, D.R., 1988. Interactions of metals and protons
with algae. Environ. Sci. Technol. 22, 755760.
da Costa, C.C.A., de Franca, F.P., 1996. Cadmium uptake by
biosorbent seaweeds: adsorption isotherms and some process conditions. Separation Sci. Technol. 31, 23732393.
Donnan, F.G., 1911. Theorie der membrangleichgewichte und
membranpotentiale bei vorhandensein von nicht dialysierenden elektrolyten. Z. Elektroch 17, 572581.
Fourest, E., Volesky, B., 1996. Contribution of sulphonate
groups and alginate to heavy metal biosorption by the dry
biomass of Sargassum uitans. Environ. Sci. Technol. 30,
277282.
Greene, B., McPherson, R., Darnall, D., 1987. Algal sorbents
for selective metal ion recovery. In: Patterson, J.W., Pasino,
R. (Eds.), Metals Speciation Separation and Recovery.
Lewis, Chelsea, MI, pp. 315338.
Guiry, M.D., Blunden, G., 1991. Seaweed Resources in Europe:
Uses and Potential. Wiley, Chichester, pp. 2425, 49.
Haug, A., 1961. Dissociation of alginic acid. Acta Chem.
Scand. 15, 950952.
Haug, A., Smidsrod, O., 1970. Selectivity of some anionic
polymers for divalent metal ions. Acta Chem. Scand. 24,
843854.
Holan, Z.R., Volesky, B., 1994. Biosorption of lead and nickel
by biomass of marine algae. Biotechnol. Bioengrg. 43, 1001
1009.
Kalimuthu, S., Kaliaperumal, N., Ramalingam, J.R., 1991.
Standing crop, algin and mannitol of some alginophytes of
Mandapam coast. J. Mar. Biol. Ass. India 33, 170174.
Kohn, R., 1975. Ion binding on polyuronates alginate and
pectin. Pure Appl. Chem. 42, 371397.
Kuyucak, N., Volesky, B., 1989. The mechanism of cobalt
biosorption. Biotechnol. Bioengrg. 33, 823831.
Lahaye, M., Axelos, M.A.V., 1993. Gelling properties of watersoluble polysaccharides from proliferating marine green
seaweeds (Ulva spp.). Carbohyd. Polym. 22, 261265.
Lee, R.E., 1980. Phycology. Cambridge University Press,
Cambridge, UK, pp. 711, 381.
McKinnell, J.P., Percival, E., 1962. The acid polysaccharide
from the green seaweed Ulva lactuca. J. Chem. Soc. Part II,
pp. 20822083.
Percival, E., McDowell, R.H., 1967. Chemistry and Enzymology of Marine Algal Polysaccharides. Academic Press,
London, UK, pp. 99126.
Ramelow, G.J., Fralick, D., Zhao, Y., 1992. Factors aecting
the uptake of aqueous metal ions by dried seaweed biomass.
Microbios 72, 8193.
Schiewer, S., Volesky, B., 1995a. Modeling of the proton-metal
ion exchange in biosorption. Environ. Sci. Technol. 29,
30493058.

282

S. Schiewer, M.H. Wong / Chemosphere 41 (2000) 271282

Schiewer, S., Volesky, B., 1995b. Mathematical evaluation of


the experimental and modeling errors in biosorption.
Biotechnol. Techniques 9, 843848.
Schiewer, S., Volesky, B., 1997a. Ionic strength and electrostatic eects in biosorption of protons. Environ. Sci.
Technol. 31, 18631871.
Schiewer, S., Volesky, B., 1997b. Ionic strength and electrostatic eects in biosorption of divalent metal ions and
protons. Environ. Sci. Technol. 31, 24782485.
Schiewer, S., Wong, M.H., 1999. Metal binding stoichiometry
and isotherm choice in biosorption. Environ. Sci. Technol.,
in press.

Siegel, B.Z., Siegel, S.M., 1973. The chemical composition of


algal cell walls. CRC Critical Reviews in Microbiology 3,
126.
Stumm, W., Morgan, J., 1996. Aquatic Chemistry, third ed.
Wiley, New York, pp. 283287, 301304, 581586.
Volesky, B., 1990. Removal and recovery of heavy metals by
biosorption. In: Volesky, B. (Ed.), Biosorption of Heavy
Metals. CRC Press, Boca Raton, FL, pp. 743.
Zhao, Y., Hao, Y., Ramelow, G.J., 1994. Evaluation of
treatment techniques for increasing the uptake of metal
ions from solution by nonliving seaweed algal biomass.
Environ. Monit. Assess. 33, 6170.

You might also like