Water Air Soil Pollut (2014) 225:1854
DOI 10.1007/s11270-013-1854-0
Effect of Competing Cations (Pb, Cd, Zn, and Ca)
in Fixed-Bed Column Biosorption and Desorption
from Citrus Peels
Abhijit Chatterjee & Silke Schiewer
Received: 6 September 2013 / Accepted: 18 December 2013 / Published online: 22 January 2014
# Springer Science+Business Media Dordrecht 2014
Abstract Citrus peels were utilized for dynamic
biosorption of Pb, Cd, and Zn from mono- and bi-
component solutions in fixed-bed columns at feed con-
centrations of 0.1 meq/L. Uptake at breakthrough and
saturation followed the order Zn<Cd<Pb in single-
metal biosorption. An overshoot of the Zn or Cd effluent
concentration was observed in Pb-Zn and Pb-Cd bimetal
systems, where Pb displaced initially bound Zn and Cd.
The desorption efficiency of the saturated column using
0.1 N nitric acid was 74 to 100 %, achieving a concen-
tration factor (CF) of 34 to 129, based on the average
desorbed metal concentration. The common practice of
defining the concentration factor based on the peak
concentration overestimates the CF value. Increasing
Ca concentrations in Pb-Ca and Cd-Ca systems reduced
target metal uptake, especially for Cd which had a lower
affinity than Pb. Actual mining effluent was treated
successfully. A sigmoid function was applied to de-
scribe experimental breakthrough data. A novel simple
two-parameter model was introduced to simulate over-
shoot and desorption curves.
Keywords Biosorption . Column . Binary systems .
Desorption . Calcium competition . Citrus peels
Electronic supplementary material The online version of this
article (doi:10.1007/s11270-013-1854-0) contains supplementary
material, which is available to authorized users.
A. Chatterjee : S. Schiewer (*)
Department of Civil & Environmental Engineering,
University of Alaska Fairbanks,
P.O. Box 755900, Fairbanks, AK 99775-5900, USA
e-mail: sschiewer@alaska.edu
1 Introduction
Mining and other industrial activities may lead to surface
water contamination with toxic metals. Decades of re-
search identified many low-cost biomass types functioning
as biosorbents for metal removal from waste streams (Arief
et al. 2008; Demirbas 2008; Vijayaraghavan and Yun
2008). However, biosorption research has primarily fo-
cused on two issues: (1) search for novel biosorbents and
the optimization of uptake in a batch reactor containing a
single target metal (Arief et al. 2008), (2) increasing the
uptake capacity by suitable processing (Vijayaraghavan
and Yun 2008) even by genetic modification (Bae et al.
2003) without considering the economic viability.
For commercial application, fixed-bed reactors, as
commonly used for ion exchange resins and activated
carbon sorption, are preferable due to better mass trans-
fer kinetics and higher effluent quality compared to
batch reactors. Also, mining effluents typically contain
several regulated and nonregulated metals. Therefore,
multimetal biosorption and desorption studies in packed
bed columns should be a priority; use of minimally
processed biosorbents is likely most cost efficient, es-
pecially if the saturated biomass can be desorbed and
reused, which also minimizes waste generation.
Past research of the present authors identified cheaply
and abundantly available citrus peels as promising
biosorbents for the removal of priority pollutants such
as cadmium (Schiewer and Iqbal 2010) and lead
(Balaria and Schiewer 2008) using batch reactors. Being
an underutilized waste material, a cheap and abundant
availability of the raw material can be ensured. For the
2012–2013 harvest season, total citrus production in the
1854, Page 2 of 13
US is estimated to be 11.4 million tons (USDA 2013),
approximately half of which constitutes waste material.
In prior batch experiments of the authors’ research
group, ion exchange was identified as the main mecha-
nism for heavy metal biosorption by citrus peels; the
amount of ions released in milliequivalents per gram
closely corresponded to that of the metal (Ni or Zn) taken
up (Iqbal et al. 2009). The relevant binding sites were
identified as carboxyl and hydroxyl functional groups:
blocking of carboxyl groups reduced Ni and Zn binding
by grapefruit peels by 79 and 73 % respectively, while
blocking hydroxyl groups reduced uptake by 23 and 29 %,
respectively (Iqbal et al. 2009). FTIR spectra showed a
shift in the wave number of the absorption peak corre-
sponding to carboxylic groups after lead adsorption from
1,636 to 1,645 cm−1 (Balaria and Schiewer 2008). For
protonated orange peels as used in the present study, batch
kinetic experiments showed that equilibrium was achieved
in approximately 1 h, with a maximum Cd uptake capacity
of 38 mg/g (Chatterjee and Schiewer 2013).
The main objective of the present research was to
advance this study to the next level and evaluate whether
citrus peel have a potential for actual industrial applica-
tion. For that purpose, a fixed-bed reactor was used to
remove target metals in the presence of several regulated
and nonregulated metals likely encountered in a real min-
ing effluent. A second objective was to provide a simple
model for describing desorption in packed bed columns,
including the overshoot phenomenon during adsorption.
2 Materials and Methods
2.1 “Protonation” of Citrus Peels
Dried and ground processed orange peels were obtained
from Alarma Consulting Corporation in Florida, USA.
After washing and air-drying, peels were “protonated”
by treating with 0.1 N nitric acid, as described by
Chatterjee and Schiewer (2011). The peel particles of
1–2 mm size were used for this study.
2.2 Continuous Flow Column System
Five grams of protonated peels were “wet packed” into
an acrylic column (length=30 cm, internal diameter=
1.3 cm), as described by Chatterjee and Schiewer (2011)
(Supplementary Fig. A1 and A2) yielding a bed height
of 24 cm with a void space of 70 %. Influent metal
Water Air Soil Pollut (2014) 225:1854
solution was pumped upward through the column at flow
rate (Q) 9 mL/min by a peristaltic pump (VWR medium
flow) from a reservoir (20 L) till the column became
saturated. These conditions were chosen since the same
flow rate and bed height yielded a reasonable degree of
sorbent usage (41 %) for Cd mono-metal column sorp-
tion, with an empty bed contact time of 8 min (Chatterjee
and Schiewer 2011). For comparison, approximately
50 % of saturation was achieved in batch experiments
after 8 min (Chatterjee and Schiewer 2013).
Four different types of influent solutions were used: (1)
single-metal solutions of 0.1 meq/L of Cd, Zn or Pb; (2) all
possible binary combinations of those metals, each with a
concentration of 0.1 meq/L in order to ensure a fair com-
parison; (3) single regulated metal (Pb or Cd) in presence
of varying amounts of Ca as a nonregulated alkali metal.
For systems with Pb, the influent target metal concentra-
tion was 20 mg/L (0.1 mmol/L) and Ca varied from 0 to
100 mg/L (0–2.5 mmol/L). For systems with Cd as the
target metal, its concentration was 5 mg/L (0.04 mmol/L)
and Ca varied from 0 to 20 mg/L (0–0.4 mmol/L); (4)
mining effluent collected from Wet Well 3, Greens Creek
Mine, Juneau, Alaska, USA. That effluent contained Zn
(2 mg/L, 2 × 10−2 meq/L), Cd (0.5 mg/L, 5 × 10−3 meq/L),
Pb (5 μg/L, 5 × 10−5 meq/L), Ca (180 mg/L, 1.74 meq/L)
and other nonregulated alkali and alkaline earth metals at
pH 4.0. Details of quantitative chemical analysis and sea-
sonal variation of Wet Well 3 effluent can be found else-
where (Department of Natural Resources 2013). The ef-
fluent sample was loaded to the column without any
pretreatment. Except for the mining effluent, all metal
solutions were prepared by diluting 10,000 mg/L stock
solution (VWR) and adjusted to pH 5.
After saturation, the column was prepared for desorp-
tion by passing distilled water through the column at
flow rate 9 ml/min for 45 min to expel any unbound
metal. 0.1 N nitric acid, which had proven effective for
metal desorption from citrus peels in batch studies
(Njikam and Schiewer 2012), was used as the desorp-
tion agent and pumped into the system at a flow rate of
9 ml/min until the desorption effluent concentration
became constant or lower than 0.1 mg/L.
All experiments were conducted at 25 °C. During sorp-
tion and desorption, effluent samples were collected inter-
mittently using a programmed fraction collector (Gilson
FC203B) attached to the column outlet (Supplementary
Fig. A1). Samples were then analyzed for the target metal
concentration (C, milligrams per liter or millimoles per
liter) using an atomic absorption spectrophotometer
Water Air Soil Pollut (2014) 225:1854 Page 3 of 13, 1854

(AAS) (Perkin Elmer). All AAS analyses were triplicated. partially saturated. The MTZ moves through the column
Since the standard deviation was very low, mean values in the flow direction and breakthrough occurs when the
only are reported. The AAS calibration curve regression MTZ reaches the end of the column with length L.
 
coefficient was always equal or greater than 0.99. Samples tb
MTZ ¼ 1− L ð6Þ
were spiked intermittently and satisfactory spike recovery ts
was obtained. Calibration standards were checked after
every tenth sample.
2.4 Analysis of Desorption Curve
2.3 Analysis of Breakthrough Curve
The desorption curve is usually plotted as C vs. t and the
Experimentally determined breakthrough curves, i.e., total mass of metal desorbed at any time t (mde,t) can be
plots of the ratio of effluent (C, milligrams per liter) to found from Eq. (7).
influent (C0, milligrams per liter) concentration versus Zt
process time (t, hours), were compared in terms of break- mde;t ¼ Q  Cdt ð7Þ
through time, tb, saturation time, ts (defined as times 0
where C reached 5 and 95 % of C0, respectively), uptake
The desorption efficiency (E) is defined as the ratio of
capacity at breakthrough (qb, milligrams per gram) ca-
metal desorbed (mde,t) and the metal bound (mad,ts)
pacity at saturation (qe, milligrams per gram) and the
during the sorption cycle.
length of mass transfer zone (MTZ, centimeters).
mde;t
The metal adsorbed in the column (mad,t, milligrams) E¼  100% ð8Þ
at time t can be determined from Eq. (1) based on metal mad;ts
entering the column (min,t, milligrams, Eq. (2)) and To calculate the maximum possible desorption effi-
unadsorbed metal (mout,t, milligrams, Eq. (3)) exiting ciency Emax, the function mde,t was evaluated at the time
up to time t (hours). td, when the concentration in the effluent from desorp-
mad;t ¼ min;t −mout;t ð1Þ tion reaches less than 1 mg/L. For efficient use of the
desorption agent, desorption can be terminated at the
min;t ¼ C 0  Q  t ð2Þ time t0.9 when 90 % of the maximum desorption has
been achieved, i.e., E=0.9 Emax. The reagent require-
Zt  
C ment till t0.9 is denoted as V0.9.
mout;t ¼ C 0  Q  dt ð3Þ
C0 The average (from t=0 to t0.9) eluent concentration
0
(Cav,de) at the outlet is calculated as:
The saturation (maximum) uptake capacity (qs, mil- Zt0:9
ligrams per gram) is calculated as the total mass of metal C dt
adsorbed at saturation (mad,ts milligrams) to the mass of
0
adsorbent in column (M, grams): C av;de ¼ ð9Þ
t
Zts   The concentration factor (CF) is the ratio of metal
C
C0  Q  1− dt concentration in the eluent exiting the column to that in
C0
mad;ts 0 the influent stream during adsorption and can be calcu-
qs ¼ ¼ ð4Þ
M M lated according to Eq. (10). It is a measure for the overall
Similarly, uptake at breakthrough can be calculated as: effectiveness of the adsorption/desorption cycle, gener-
ating a concentrated metal solution from the dilute in-
Ztb  
C fluent stream.
C0  Q  1− dt
C0 C av;de
mad;tb
qb ¼ ¼
0
ð5Þ CF ¼ ð10Þ
M M C0
MTZ is defined as the mass transfer zone of the A higher CF implies a higher recovery ratio (R=Vs/
column where sorbate is actively adsorbed from fluid V0.9), i.e., lower required eluent volume (V0.9) for a
phase to solid phase, and where the sorbent is only given saturation volume (Vs) during adsorption and
1854, Page 4 of 13
therefore lower reagent cost. It is easy to prove that these
three key terms (E, R, CF) are related as follows, pro-
vided each of them is calculated for the same time period
(up to td or t0.9 or any other time t):
 
1 1 C av;ad C 0 −C av;ad
¼E − ¼E ð11Þ
R CF C av;de C av;de
Here Cav,ad denotes the average concentration in the
effluent during the sorption cycle and can be calculated
using Eq. (12) up to ts or any other time t.
Zt
Cdt
0 tending this observation, the following simple equation
C av;ad ¼ ð12Þ
t with two empirical parameters A and B is introduced
For most practical situations, where sorption is ter-
minated at breakthrough, so that Cav,ad ≈0, Eq. (11) leads
to the following elegant relationship:
CF
E¼ ð13Þ
R
2.5 Modeling of Breakthrough and Desorption Curves
The dynamic behavior of fixed-bed multi-component
adsorption can be modeled by solving a system of
partial differential equations (mass balance, rate law)
coupled with an appropriate multi-component equilibri-
um relationship. This requires use of specialized soft-
ware, e.g., in MATLAB or COMSOL for numerical
modeling, as an analytical solution is only available
for a few idealized cases assuming no mass transfer
limitation. Still, the resulting model can often only sim-
ulate the breakthrough curve under specific conditions
and cannot be used for a priori prediction because opti-
mized model parameters are not necessarily intrinsic
ones but can be lumped parameters.
A number of semi-mechanistic and empirical models
are available for prediction of breakthrough curves ob-
tained from single-metal experiments utilizing sigmoid
functions (Chatterjee and Schiewer 2011; Chu 2004),
but none exist to simulate overshoot phenomena and
desorption curves. The following general sigmoid equa-
tion with A and B as two empirical parameters is used
here to represent breakthrough curves:
C 1
¼ ð14Þ
C 0 1 þ expð−αt þ βÞ
Water Air Soil Pollut (2014) 225:1854
As shown in the authors’ previous article (Chatterjee
and Schiewer 2011), this equation actually represents
the frequently used Thomas equation with Thomas pa-
rameters kTh =α/C0 and qTh =(β/α) (Q C0/M). One ob-
jective of this article was to introduce a simplified model
describing overshoot and desorption curves. It may be
noted that differentiation of any sigmoid curve results in
a bell shaped curve; for several distribution patterns
(Normal distribution, Maxwell–Boltzmann distribution,
Rayleigh distribution) the cumulative function shows
sigmoid curves (which have been used to simulate
breakthrough curves) whereas the probability density
function shows similarity with desorption curves. Ex-
here to model desorption data.
C ¼ A t 2 e−Bt
2
ð15Þ
Since overshoot is primarily caused by desorption of
the weaker metal by the stronger metal, this equation
was also applied to simulate the overshoot portion of
breakthrough curves. For curves with overshoot, the
sigmoid model Eq. (14) was used to describe data up
to the point that the effluent concentration reaches 95 %
of influent concentration (C/C0 =0.95). After that point,
the desorption model Eq. (15) was used to describe the
overshoot portion of the data.
3 Results and Discussion
3.1 Single-Metal Biosorption
Breakthrough curves of Cd, Zn, and Pb are shown in
Fig. 1a. It was found (Table 1, Run# 1–3) that break-
through times as well as uptake at breakthrough and
saturation in both molar and mass units followed the
same order (Zn<Cd<Pb). Molar uptake of Zn and Cd at
breakthrough was actually similar, being close to
0.2 mmol/g. Pb was clearly the preferred metal with
the best sorption performance. The two parameter sig-
moid model Eq. (14) adequately represented break-
through data as shown in Fig. 1a. Model parameters
are reported in Table 2.
Compared to the wealth of data available for
multimetal biosorption in batch reactors, only few studies
compare mono-metal capacities of these three metals in
fixed bed. For most biosorbents, metal uptake (on a mass
Water Air Soil Pollut (2014) 225:1854 Page 5 of 13, 1854

a b
1.0 Cd with Zn Cd with Pb Cd

0.9 Cd with Zn Mod Cd with Pb Mod Cd with Pb OS

Concentration Ratio C/Co (-)

2.0
0.8
1.8

Concentraion Ratio C/Co (-)

0.7
1.6
0.6 1.4
0.5 1.2
0.4 1.0
0.8
0.3
Zn Pb Cd 0.6
0.2
Data 0.4
0.1 0.2
Model
0.0 0.0
0 50 Time (h) 100 150 0 50 Time (h) 100 150

c d
Zn with Cd Zn with Pb Zn Pb with Zn Pb with Cd Pb
Zn with Cd Mod Zn with Pb Mod Zn with Pb OS Pb with Zn Mod Pb with Cd Mod
2.0 1.0
1.8
Concentration Ratio C/Co (-)

Concentration Ratio C/Co (-) 0.9

1.6 0.8
1.4 0.7
1.2 0.6
1.0 0.5
0.8 0.4
0.6 0.3
0.4 0.2
0.2 0.1
0.0 0.0
0 50 Time (h) 100 150 0 50 Time (h) 100 150

Fig. 1 Single and binary metal breakthrough curves. Feed con- a Single metals, b Cd alone or in presence of Pb or Zn, c Zn alone
centration 0.1 meq/L for each metal. Experimental data and pre- or in presence of Pb or Cd, d Pb alone or in presence of Cd or Zn
dictions of sigmoid model (mod) and overshoot model (OS).

basis) follows the order Pb>Cd>Zn, as in this study are converted from mass to molar units, no general trend
(Table 3). Though it is preferable to compare metal bind- in uptake could be discerned. In the present study the
ing at equal molar concentrations of each metal, as done in uptake sequence Pb>Cd>Zn was maintained also on a
the present study, most other studies compared metal molar basis, though uptakes for different metals were quite
uptake at equal mass concentration. When uptake values similar, as in some other studies.

Table 1 Breakthrough curve parameters for Pb, Zn and Cd removal from single and binary systems, with a feed concentration of 0.1 meq/L
for each metal

Run # M1 M2 C0, M1 (mg/L) C0, M2 (mg/L) tb1 (h) tb2 (h) ts1 (h) ts2 (h) qb1 (mg/g) qb2 (mg/g) qs1 (mg/g) qs2 (mg/g)

1 Pb – 10.36 – 50 – 120 – 57 – 85 –
2 Cd – 5.62 – 37 – 120 – 22 – 44 –
3 Zn – 3.27 – 30 – 110 – 11 – 20 –
4 Pb Cd 10.36 5.62 50 15 114 108 49 9 71 0
5 Pb Zn 10.36 3.27 48 12 113 95 57 5 88 6
6 Cd Zn 5.62 3.27 10 11 77 80 6 4 20 11
1854, Page 6 of 13 Water Air Soil Pollut (2014) 225:1854

Table 2 Breakthrough curve modeling parameters metal concentrations used here were much lower than
Run # Metal α (h−1) β A (min−2) B (min−2) those used in most other studies.

1 Pb 0.13 0.54 3.2 Binary Metal Biosorption

2 Cd 0.07 0.32
3 Zn 0.09 0.30 3.2.1 Effect on Individual Uptake
4 Pb 0.12 0.51
4 Cd 0.36 0.55 0.028 0.0022 Effect of Presence of Other Metals (Pb, Zn) on Cd
5 Pb 0.10 0.40 Biosorption A comparison of run# 2, 4, and 6 in Table 1
5 Zn 0.21 0.35 0.011 0.0018 and Fig. 1b shows that the effluent concentration
6 Cd 0.10 0.17 approached the influent concentration more quickly
6 Zn 0.12 0.20 (higher C/C0) for a bimetal system, than for the single-
metal system. Uptake at breakthrough and saturation
was lower for bimetal systems than for mono-metal
Selectivity of lead over other metals can be explained systems. This means that the uptake of Cd was nega-
by its favorable hard-hard interaction with hard ligands tively influenced by the addition of a second metal. In
such as carboxylate and hydroxyl groups and other the presence of Pb, the breakthrough curve of Cd first
relevant properties (smaller hydrated ion radius, lower intercepted the C/Co=1 line when Cd uptake was
hydrolysis constant and higher covalent character of 72 mg/g. Due to the higher affinity of Pb, it then dis-
bonding) (Table 4). pelled previously bound Cd from the sorbent. The pres-
Since there are few studies comparing sorption ence of this desorbed Cd in the effluent caused the
of these three metals using citrus peels or other effluent concentration to exceed the influent concentra-
fruit residues in a fixed-bed reactor, plant-based tion, which is commonly termed “overshoot”. The re-
sorbent materials with comparable chemical com- placement of Cd ions by Pb ions reached its peak rate at
position are listed in Table 3. The metal uptake of the breakthrough of Pb (Fig. 1b, d, Supplementary
citrus peels for each of these three metals is com- Fig. A3a), when the Cd overshoot reached 1.76 times
parable or higher than for other established the feed concentration. By the time that the Pb concen-
biosorbents and commercially available ion exchange tration approached that in the feed, all sites were satu-
resins (Table 3). This is especially remarkable since the rated with Pb and almost all of the previously adsorbed

Table 3 Comparison of mono-component saturation uptake for Pb, Cd, and Zn biosorption in fixed-bed columns

Sorbent Saturation uptake mg/g (mmol/g)a Feed concentration Order of uptake Ref
(mg/L) (on mass basis)
Pb Cd Zn

Amberlite IRC 718 20 (0.10) 119 (1.06) 68 (1.04) Pb 250, Cd, Zn 600 Cd>Zn>Pb (Malla et al. 2002)
Chelex 100 (Bio-Rad) 23 (0.11) 112 (1.00) 83 (1.27) Pb 375, Zn 400, Cd>Zn>Pb (Malla et al. 2002)
Cd 350 or 600
Wetland sediment 97 (0.47) 18 (0.16) 9 (0.14) 50 each Pb>Cd>Zn (Seo et al. 2008)
Grape stalk waste 56 (0.27) 33 (0.29) – Cd 60, Pb 120 Pb>Cd (Miralles et al. 2010)
Tea waste 46 (0.22) – – 100 – (Miralles et al. 2010)
Orange waste – – 29 (0.44) 20–40 – (Marin et al. 2010)
Wheat straw – 17 (0.15) – 100 – (Muhamad et al. 2010)
Green coconut shell 55 (0.27) 38 (0.34) 17 (0.26) 100 each Pb>Cd>Zn (Sousa et al. 2010)
Waste from boron – 86–138 80–110 150 each Cd>Zn (Atar et al. 2012)
enrichment plant (0.76–1.23) (1.22–1.68)
Citrus Peels 85 (0.41) 44 (0.39) 20 (0.31) Pb 10.36 Cd Pb>Cd>Zn This study
5.62, Zn 3.27
a
Values within parenthesis show uptake in molar units (millimoles per gram)
Water Air Soil Pollut (2014) 225:1854
Table 4 Physicochemical properties characterizing the binding
strength of metals (Wulfsberg 1992)
Parameter Zn Cd Pb
Crystal radius (r) (pm) 88 95 119
Hydrated radius (pm) 430 426 401
Hydrolysis constant (−log K) 8.96 10.08 7.7
Covalent Binding χ2 (r +85) (pm) 536 551 718
Cd was lost. Overshoot of Cd in presence of Pb confirms
the higher selectivity of Pb observed in single-metal
systems.
The model adequately described the overshoot data
though there is a “break” at the transition point between
Eqs. (14) and (15) and a shortcoming of the model in
that it does not predict the asymptotic value approached.
The same binary system was later studied (Vilar et al.
2008) using immobilized algal biosorbent but at a higher
Cd/Pb ratio of 1.8:1 in the feed. Still 58 % of the
biosorbed Cd was found to be displaced by Pb, causing
a 69 % drop in saturation uptake.
From a commercial aspect, the overshoot problem
can be eliminated by terminating column operation at
the breakthrough of weaker metal, i.e., in this case when
the Cd uptake was 9 mg/g and Pb uptake was 15 mg/g.
The trade-off for doing that is that less of the stronger
bound metal is taken up. Nevertheless, this strategy is
necessary if both metals have to be removed.
Effect of Presence of Other Metals (Cd, Pb) on Zn
Biosorption The biosorption of Zn was also strongly
reduced by Pb (Table 1, Run# 3, 5, 6 and Fig. 1c). While
the breakthrough capacity was almost equally affected by
the presence of either Cd (64 % reduction) or Pb (55 %
reduction), the final uptake was drastically reduced by Pb.
Like Cd in the Pb-Cd system, the exit concentration of Zn
also showed an overshoot in the presence of Pb (Fig. 1c,
Supplementary Fig. A3b) with a peak of C/C0 =1.57,
confirming the same sequence of affinity (Pb>Zn) as in
the single-metal study (section 3.1).
A seaweed-derived biosorbent also showed a higher
overshoot of Cd than of Zn in the presence of Cu, which
was the preferred metal in that study (Kratochvil and
Volesky 2000).
Effect of Presence of Other Metals (Cd, Zn) on Pb
Biosorption Compared to both Cd and Zn, the
biosorption of Pb was least influenced by the presence
Page 7 of 13, 1854
of competing metals (Table 1, Run# 1, 4, 5; Fig. 1d,
Supplementary Fig. A3c). Figure 1d shows that the three
breakthrough curves of Pb were very closely positioned,
i.e., neither Zn nor Cd affected Pb binding, due to its
higher affinity towards the biosorbent.
Modeling Multimetal Biosorption For all three cases
(Pb-Cd, Pb-Zn, Zn-Cd), the sigmoid function (Eq. 14)
simulated experimental breakthrough data and over-
shoot was described using the bell curve function
(Eq. 15). Model parameters were reported in Table 2.
As shown in Fig. 1b, c, d, the model described bimetal
systems well.
3.2.2 Comparison of the Binary System
with its Mono-Component Counterparts
Comparison with Respect to Individual Uptake To com-
pare the biosorption efficiency between an M1-M2 bi-
nary system and the corresponding single-metal sys-
tems, the fraction function F (Xi) was defined as
X i;b
F ðX i Þ ¼ 100 ð16Þ
X i;s
Whereby Xi,b is any variable (e.g., uptake, or break-
through time) for metal i in the binary system and Xi,s is
the same value for the respective single-metal system. F
values reported in Table 5 show that for the Cd-Zn
binary system, both Cd and Zn retained similar percent-
ages of their mono-metal uptake both at breakthrough
(33 and 36 %) and at saturation (45 and 55 %), which is
in keeping with their similar affinities for the sorbent. In
contrast, for the Pb-Cd and Pb-Zn systems, the high
affinity metal Pb was always able to retain its full
capacity but at the expense of hampering its co-ion
adsorption. Saturation uptake of Cd and Zn in presence
of Pb was critically hampered due to overshoot as ex-
plained in the section 3.2.1.
Comparison with Respect to Total Uptake A compari-
son of the combined saturation capacity of the binary
system (qs1 +qs2) with the mono-metal saturation capac-
ity of the stronger metal of the pair, each expressed in
molar units, reveals two different scenarios (Table 5).
The Pb-Zn system showed a minor increase in the total
uptake compared to mono-metal Pb uptake, which is
expected as the combined influent concentration was
doubled in the binary system. On the other hand, the
1854, Page 8 of 13
combined saturated capacity of the Pb-Cd and Cd-Zn
systems was a little lower than the uptake capacity of Pb
or Cd alone. This may be due to the scatter in the
experimental data (especially for Pb, Fig. 1a).
3.3 Effect of Calcium Ions
A total of 11 fixed-bed experiments were conducted
(Table 6, Run# 7–12 for Pb, 13–17 for Cd) with gradu-
ally increasing Ca concentration in the influent. Com-
parison of Fig. 2a and b shows that calcium affects
cadmium biosorption much more than lead biosorption.
While 5 mg/L of calcium had no discernible effect on
the Pb breakthrough curve, the same Ca concentration
had a huge impact on the Cd breakthrough curve, re-
ducing the breakthrough time from 36 to 10 h. This is
expected as Pb has a higher affinity than Cd. For both Pb
and Cd, the mass transfer zone length was found to
increase linearly with increasing Ca concentration, with
concurrent decrease in breakthrough and saturation
capacity.
for Pb: MTZ/cm=7.99+0.15 C0/(mg/L) (R2 =0.94);
for Cd: MTZ/cm=13.46+0.41 C0/(mg/L) (R2 =0.96).
Unlike the influence of other heavy metals, the in-
hibitory effect of Ca for Pb binding is only significant at
50 mg/L or higher concentration. In contrast, Cd uptake
was already affected at a relatively low Ca concentration
of 5 mg/L. Comparing run# 6 (Table 1) and run# 14
(Table 6), 0.125 mmol/L (5 mg/L) of Ca had a similar
effect as 0.05 mmol/L of Zn on Cd breakthrough.
One study (de Franca et al. 2002) revealed that cal-
cium affects zinc biosorption onto a Sargassum species
in fixed-bed reactor, but there are no published articles
focusing on the effect of alkali metals on fixed-bed
biosorption of Pb or Cd. This extent of alkali metal
inhibition is still much lower compared to commercial
ion exchange materials; for example, at a moderate
aqueous concentration of Pb (37–49 mg/L), a cation
Table 5 Removal of Pb, Zn, and Cd comparing single and binary metal mixtures
Run # M1 M2 Fa (qb1) (%) Fa (qb2) (%) Fa (qs1) (%)
4 Pb Cd 86 40 84
5 Pb Zn 100 45 104
6 Cd Zn 33 36 45
a
The function F is defined according to Eq. (16)
Water Air Soil Pollut (2014) 225:1854
exchanger IRA 120 Amberlite was found to lose 94 %
of its Pb sorption capacity in the presence of Ca co-ions
(Wilke et al. 2006).
In conclusion, Pb could still be effectively be re-
moved in column applications even if the molar con-
centration of Ca was much higher than that of Pb
(2.5 mmol/L of Ca vs. 0.1 mmol/L Pb), while Cd uptake
was severely impacted even by low Ca concentrations.
3.4 Desorption
Metals, unlike organic matter, cannot be “degraded” but
can only be concentrated and reused. Hence, desorption
is essential for a meaningful metal sequestration tech-
nology. Figure 3 shows metal recovery using 0.1 M
nitric acid as the desorption agent. Desorption was faster
than biosorption; the curve reached at its peak concen-
tration (Cp) within 10–15 min (tp) (Table 7). In the
present study, desorption was achieved in similar
timeframes both in mono-metal (Fig. 3a) and binary
systems (Fig. 3b–d), with 90 % of Emax achieved in
t0.9 =18–54 min, except for the Cd mono-metal system,
where 100 min were necessary. Compared to that, it was
reported that the amount of acid required for desorption
of Pb increased with the addition of a second metal in
the Pb-Ni and Pb-Zn system using black gram husk as
biosorbent (Saeed et al. 2005).
The Pb concentration peak was much higher than that
for the other co-cations in Pb-Cd and Pb-Zn desorption
(Fig. 3c, d) which confirms that hardly any of the
weaker bound Cd or Zn remained on the sorbent after
being displaced by Pb during the overshoot. For the Pb-
Cd system, the amount of remaining Cd was insignifi-
cant and was not incorporated within Table 7.
Maximum desorption efficiency (Emax) was moder-
ately high (>85 %) except for Zn in the Pb-Zn system
(Table 7). In some cases, Emax values exceed 100 % due
to the measurement of concentrations near the detection
limit and error associated with the numerical integration
of experimental data.
Fa (qs2) (%) qs1 +qs2 (mmol/g) Mono-metal qs1 (mmol/g)
0 0.34 0.41
30 0.52 0.41
55 0.35 0.39
Water Air Soil Pollut (2014) 225:1854 Page 9 of 13, 1854

Table 6 Effect of Ca concentra-

tion on biosorption of Pb or Cd Run # Target metal C0 of metal C0 of Ca tb ts qb qs MTZ
mg/L mg/L h h mg/g mg/g cm

7 Pb 20 0 21 56 45 73 9
8 Pb 20 5 23 49 51 72 7
9 Pb 20 20 22 66 51 100 12
10 Pb 20 50 10 58 24 60 15
11 Pb 20 75 3 58 7 58 21
12 Pb 20 100 2 58 5 51 22
13 Cd 5 0 36 146 18 44 14
14 Cd 5 5 10 62 5 14 15
15 Cd 5 10 12 93 6 21 17
16 Cd 5 15 2 36 1 7 20
17 Cd 5 20 2 46 1 10 21

The ratio of t0.9/td (Table 7) shows that for Pb desorp-

a 1 tion (single and binary system), 90 % of Emax was
0.9
achieved within less than 20 % of the total desorption
time (td), while the remaining 80 % of the time were
0.8
Concentration Ratio C/C 0 (-)

utilized only to accomplish the remaining (10 % of

0.7 Ca=100 Emax) desorption. The long tail (i.e., long td) in Pb
0.6 Ca=75 desorption curves can be explained by the higher cova-
0.5 Ca=50 lent index of Pb (Table 4) implying a higher proportion
0.4 of covalent bonding with Pb. For practical purposes, this
Ca=20
0.3
means that in industrial applications it would be advan-
Ca=5
tageous to terminate desorption at t0.9 to achieve a
0.2
Ca=0 mostly complete desorption using a much lower volume
0.1
of eluent than required for complete desorption.
0 For single metal desorption, the CF and recovery
0 20 40 60 80
Time (h)
ratio (R) followed the order Cd<Pb<Zn. Most pub-
lished studies (Barron-Zambrano et al. 2010; Martin-
b 1
Lara et al. 2012) define CF (i.e., CFp in Table 7) on the
basis of the peak desorbed metal concentration (Cp)
0.9
instead of Cav. However, CFp is of little practical rele-
Concentration Ratio C/C 0 (-)

0.8
vance since for any industrial desorption process the
0.7 entire volume of desorbent used has to be considered,
0.6 i.e., an average concentration within this solution should
Ca=20
0.5 be the basis for calculating CF as proposed in Eq. (10).
Ca=15 As shown in Table 7, using Cp to calculate CF may lead
0.4

0.3
Ca=10 to an overestimated CF depending on the extent of
Ca=5 dilution occurring between this peak time and the time
0.2
when desorption was terminated.
0.1 Ca=0
Within binary systems, CF of Pb was decreased from
0
113 to 70 in the presence of Zn, but was marginally
0 50 100 150 200
Time (h)
increased in presence of Cd. Similarly, CF of Zn was
Fig. 2 Breakthrough curves for Pb and Cd in presence of different
decreased with the addition of a second metal (Cd or Pb).
Ca concentrations (mg/L). a Pb with feed concentration of In practical applications, where column operation
0.1 mmol/L, b Cd with feed concentration of 0.04 mmol/L would be terminated at breakthrough, the amount sorbed
1854, Page 10 of 13 Water Air Soil Pollut (2014) 225:1854

a b
Zn Cd Pb 1200 Cd Zn
Zn Mod Cd Mod Pb Mod
3000 Cd Mod Zn Mod
1000

Effluent Conc. C (mg/L)

2500
Effluent Conc. C (mg/L)

800
2000
600
1500

400
1000

500 200

0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Time (h) Time (h)

c d
3000 Pb Zn 4000 Pb Cd
Pb Mod Pb Mod
3500
2500
Effluent Conc. C (mg/L)

Effluent Conc. C (mg/L)

3000
2000
2500

1500 2000

1500
1000
1000
500
500

0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Time (h) Time (h)

Fig. 3 Desorption for single and binary systems using 0.1 N nitric acid. Experimental data and predictions of two-parameter model. a
Mono-metal system, b Zn-Cd system, c Pb-Zn system, d Pb-Cd system

qb would be somewhat lower than for operation Nevertheless, the overall high-concentration factors
till exhaustion qs, which would result in a reduced achieved confirm the usefulness of biosorption/de-
amount to be desorbed, and thereby in lower con- sorption using citrus peels to concentrate heavy
centration factors, depending on the ratio of qb/qs. metals from wastewater.

Table 7 Desorption curve parameters for breakthrough curves 1–6

Run # Metal(s) td t0.9 tp (t0.9/td)×100 Cp Cav,de CFP CF V0.9 Emax R A B

min min min % mg/L mg/L ml (%) min−2 min−2

1 Pb 241 40 12 17 2,725 1,167 263 113 360 108 180 47 0.012

2 Cd 200 100 15 50 1,146 191 204 34 900 86 75 14 0.011
3 Zn 78 24 10 31 1,007 421 308 129 216 99 275 17 0.015
4a Pb 555 34 12 6 3,647 1,219 352 118 306 115 201 52 0.013
5 Pb 420 54 14 13 2,580 728 249 70 486 86 126 23 0.008
5 Zn 65 18 12 28 340 123 104 38 162 73 317
6 Cd 65 20 10 31 1,141 456 203 81 180 90 231 22 0.019
6 Zn 56 20 10 36 667 306 204 93 180 109 240 14 0.018
a
Negligible amount of Cd present
Water Air Soil Pollut (2014) 225:1854
It was found that Eq. (15) can describe desorption
curves generally well but under-predicted the desorption
peak height. One reason could be the unequal distribution
of data points as there are more data points after the peak
and the standard optimization routine (Solver in MS
Excel) weighed each data point equally. In addition to
that, the equation used is not skewed and therefore unable
to describe a highly asymmetric curve. To account for
asymmetry, a third model parameter would have to be
introduced, which would however reduce stability of the
model fitting procedure. Nevertheless, since no simple
models for describing desorption in packed bed columns
are available in biosorption literature, the model present-
ed here constitutes a step forward.
3.5 Treatment of Mining Wastewater
Figure 4 shows that 5 g of sorbent could treat 5.40 L
(i.e., 85 times the bed volume) of wastewater without
any breakthrough. The breakthrough times of Zn and Cd
were 10 h and 12 h respectively. None of them showed
any overshoot. Since a limited quantity of effluent was
collected, the experiment was terminated after 21 h.
Even though the concentrations of target metals in the
wastewater were lower than in previous experiments
(section 2.2), the breakthrough time was shorter than
in the single-metal solutions due to the presence of other
cations such as Ca at a concentration (180 mg/L) which
was higher than the experimental range (0 to 100 mg/L).
Another study (Oliveira et al. 2010) reported a similar
decrease (around 50 %) in capacity for the treatment of
electroplating effluent using peanut hulls.
0.7
Zn Cd
Concentration Ratio C/C 0 (-)
0.6
0.5
0.4
0.3
0.2
0.1
0 rate of the mining effluent and inlet concentration.
0 5 10 15
Time (h)
Fig. 4 Breakthrough curve for mining effluent
Page 11 of 13, 1854
3.6 Scale-Up for Industrial Application
Different approaches to scale-up a dynamic adsorption
system are elaborately discussed by Tien (1995). Firstly,
for a given flow rate in terms of bed volumes per unit
time, the number of bed volumes treated up to break-
through (or any given ratio of C/C0) remains constant
before and after scale-up (Tien 1995). Secondly, the
present authors obtained Cd breakthrough curves using
the same biosorbent and found that (Chatterjee and
Schiewer 2011) three apparently different breakthrough
curves for different bed heights, determined at
similar influent concentration and flow rate as in
this study, became almost superimposed when
plotted as a function of number of bed volumes
treated instead of time. This indicates a constant-
pattern situation, where the length of unused bed
(LUB) method can be applied. LUB represents the
fraction of unused capacity of the column which is
converted into an equivalent length of the bed:
 
q
LUB ¼ L 1− b ð17Þ
qs
The breakthrough time for the longer column is cal-
culated as:
 
 LUB
tb ¼ t 1− ð18Þ
Lnew
Where t* represents stoichiometric time calcu-
lated assuming an ideal (vertical step function)
breakthrough curve, as it would occur if sorption
happened instantaneously without any mass trans-
fer limitations. A numerical example was provided
by Tien (1995). Thirdly, Thomas model parameters
are typically assumed to be valid both for labora-
tory scale and full scale sorption systems in indus-
trial scale-up. As discussed in section 2.5, Thomas
model parameters can be the determined from the
sigmoid model. Assuming a desired operation time
(i.e., tb) of 1 week, the amount of citrus peels
required and bed volume can be calculated by
applying the sigmoid model so that breakthrough
does not occur before 7 days for a known flow
20 25 From a given bed volume, the column height can
be calculated keeping the same aspect ratio as
used for the laboratory experiment.
1854, Page 12 of 13
4 Conclusions
Protonated citrus peels can be used in packed bed col-
umns for the removal of Cd, Zn and Pb but the presence
of co-ions affects target metals biosorption causing an
overshoot of the weaker cation being displaced by the
stronger one. Alkali metals such as Ca also inhibit target
metal biosorption. Nevertheless, 5 g of protonated citrus
peels purified 5.40 L of mining effluent containing inter-
fering regulated and nonregulated metals. Concentration
factors calculated based on the peak (commonly done)
and average (more practically relevant) desorption efflu-
ent metal concentrations, were respectively 104–352 and
34–128, whereby mono-metal CF decreased in the order
Zn>Pb>Cd. Breakthrough curves can be simulated by
the Thomas model. A two-parameter empirical model
was newly introduced for simulation of desorption curve
and overshoot phenomena.
Acknowledgments This research was supported in part by the
National Research Initiative of the US Department of Agriculture
Cooperative State Research, Education, and Extension Service,
grant number 2005-35504-16092 and by a United States Geolog-
ical Survey National Institutes for Water Resources grant. Orange
peels were provided by Alarma Consulting Corporation in Florida.
We thank the geochemist Pete Condon for collecting mining
effluent. None of the above participated in design, execution or
interpretation of this study.
References
Arief, V. O., Trilestari, K., Sunarso, J., Indraswati, N., & Ismadji,
S. (2008). Recent progress on biosorption of heavy metals
from liquids using low cost biosorbents: characterization,
biosorption parameters and mechanism studies. Clean-Soil
Air Water, 36, 937–962.
Atar, N., Olgun, A., & Wang, S. (2012). Adsorption of cadmium
(II) and zinc (II) on boron enrichment process waste in
aqueous solutions: batch and fixed-bed system studies.
Chemical Engineering Journal, 192, 1–7.
Bae, W., Wu, C. H., Kostal, J., Mulchandani, A., & Chen, W.
(2003). Enhanced mercury biosorption by bacterial cells with
surface-displayed MerR. Applied and Environmental
Microbiology, 69, 3176–3180.
Balaria, A., & Schiewer, S. (2008). Assessment of biosorption
mechanism for Pb binding by citrus pectin. Separation and
Purification Technology, 63, 577–581.
Barron-Zambrano, J., Szygula, A., Ruiz, M., Sastre, A. M.,
& Guibal, E. (2010). Biosorption of Reactive Black 5
from aqueous solutions by chitosan: column studies.
Journal of Environmental Management, 91, 2669–
2675.
Chatterjee, A., & Schiewer, S. (2011). Biosorption of cadmium(II)
ions by citrus peels in a packed bed column: effect of process
Water Air Soil Pollut (2014) 225:1854
parameters and comparison of different breakthrough curve
models. Clean-Soil Air Water, 39, 874–881.
Chatterjee, A. & Schiewer, S. (2013). Multi-resistance kinetic
models for biosorption of Cd by raw and immobilized citrus
peels in batch and packed bed columns. Chemical
Engineering Journal, in press.
Chu, K. H. (2004). Improved fixed bed models for metal
biosorption. Chemical Engineering Journal, 97, 233–239.
de Franca, F. P., Tavares, A. P. M., & da Costa, A. C. A. (2002).
Calcium interference with continuous biosorption of zinc by
Sargassum sp. (Phaeophyceae) in tubular laboratory reactors.
Bioresource Technology, 83, 159–163.
Demirbas, A. (2008). Heavy metal adsorption onto agro-based
waste materials: a review. Journal of Hazardous Materials,
157, 220–229.
Department of Natural Resources. (2013). http://dnr.alaska.gov/
mlw/mining/largemine/greenscreek/. Accessed 2 Sept
2013.
Iqbal, M., Schiewer, S., & Cameron, R. (2009). Mechanistic eluci-
dation and evaluation of biosorption of metal ions by grapefruit
peel using FTIR spectroscopy, kinetics and isotherms model-
ing, cations displacement and EDX analysis. Journal of
Chemical Technology and Biotechnology, 84, 1516–1526.
Kratochvil, D., & Volesky, B. (2000). Multicomponent
biosorption in fixed beds. Water Research, 34, 3186–3196.
Malla, M. E., Alvarez, M. B., & Batistoni, D. A. (2002).
Evaluation of sorption and desorption characteristics of cad-
mium, lead and zinc on Amberlite IRC-718 iminodiacetate
chelating ion exchanger. Talanta, 57, 277–287.
Marin, A. B. P., Aguilar, M. I., Ortuno, J. F., Meseguer, V. F., Saez,
J., & Llorens, A. (2010). Biosorption of Zn(II) by orange
waste in batch and packed-bed systems. Journal of Chemical
Technology and Biotechnology, 85, 1310–1318.
Martin-Lara, M. A., Blazquez, G., Ronda, A., Rodriguez, I. L., &
Calero, M. (2012). Multiple biosorption-desorption cycles in
a fixed-bed column for Pb(II) removal by acid-treated olive
stone. Journal of Industrial and Engineering Chemistry, 18,
1006–1012.
Miralles, N., Valderrama, C., Casas, I., Martinez, M., &
Florido, A. (2010). Cadmium and lead removal from
aqueous solution by grape stalk wastes: modeling of a
fixed-bed column. Journal of Chemical and Engineering
Data, 55, 3548–3554.
Muhamad, H., Doan, H., & Lohi, A. (2010). Batch and continuous
fixed-bed column biosorption of Cd2+ and Cu2+. Chemical
Engineering Journal, 158, 369–377.
Njikam, E., & Schiewer, S. (2012). Optimization and kinetic
modeling of cadmium desorption from citrus peels: a process
for biosorbent regeneration. Journal of Hazardous Materials,
213–214, 242–248.
Oliveira, F. D., Soares, A. C., Freitas, O. M., & Figueiredo,
S. A. (2010). Copper, nickel and zinc removal by
peanut hulls: batch and column studies in mono, tri-
component systems and with real effluent. Global Nest
Journal, 12, 206–214.
Saeed, A., Iqbal, M., & Akhtar, M. W. (2005). Removal and
recovery of lead(II) from single and multimetal (Cd, Cu,
Ni, Zn) solutions by crop milling waste (black gram husk).
Journal of Hazardous Materials, 117, 65–73.
Schiewer, S., & Iqbal, M. (2010). The role of pectin in Cd binding
by orange peel biosorbents: a comparison of peels,
Water Air Soil Pollut (2014) 225:1854
depectinated peels and pectic acid. Journal of Hazardous
Materials, 177, 899–907.
Seo, D. C., Yu, K., & DeLaune, R. D. (2008). Comparison of
monometal and multimetal adsorption in Mississippi River
alluvial wetland sediment: batch and column experiments.
Chemosphere, 73, 1757–1764.
Sousa, F. W., Oliveira, A. G., Ribeiro, J. P., Rosa, M. F.,
Keukeleire, D., & Nascimento, R. F. (2010). Green coconut
shells applied as adsorbent for removal of toxic metal ions
using fixed-bed column technology. Journal of
Environmental Management, 91, 1634–1640.
Tien, C. (1995). Adsorption calculations and modeling. Boston:
Butterworth-Heinemann.
USDA. (2013). Citrus production forecast 2013 http://
www.ers.usda.gov/publications/fts-fruit-and-tree-nuts-
Page 13 of 13, 1854
outlook/fts-355.aspx#.UYQ9e8rih48. Accessed 6 Sept
2013.
Vijayaraghavan, K., & Yun, Y. S. (2008). Bacterial
biosorbents and biosorption. Biotechnology Advances,
26, 266–291.
Vilar, V. J. P., Loureiro, J. M., Botelho, C. M. S., & Boaventura, R. A.
R. (2008). Continuous biosorption of Pb/Cu and Pb/Cd in
fixed-bed column using algae Gelidium and granulated agar
extraction algal waste. Journal of Hazardous Materials, 154,
1173–1182.
Wilke, A., Buchholz, R., & Bunke, G. (2006). Selective
biosorption of heavy metals by algae. Environmental
Biotechnology, 2(2), 47–56.
Wulfsberg, G. (1992). Principles of descriptive inorganic chemis-
try. Sausalito: Univ Science Books.