Upload
jose-luis-monroy-rodriguez
View
216
Download
0
Embed Size (px)
Citation preview
8/12/2019 diversos sorbentes
1/12
Available at www.sciencedirect.com
jo ur na l ho me pa ge : ww w. el sevi er .com /l oc at e/ wa tr es
Simultaneous removal of As, Cd, Cr, Cu, Ni and Zn from
stormwater: Experimental comparison of 11 different
sorbents
Hulya Genc--Fuhrman, Peter S. Mikkelsen, Anna Ledin
Institute of Environment & Resources, Technical University of Denmark, Building 115, DK-2800 Kgs. Lyngby, Denmark
a r t i c l e i n f o
Article history:
Received 31 May 2006
Received in revised form
18 October 2006
Accepted 19 October 2006
Available online 14 December 2006
Keywords:
Heavy metals
Stormwater runoff
Secondary treatment
Sorption
Precipitation
Sorbents
A B S T R A C T
The potential of using alumina, activated bauxsol-coated sand (ABCS), bark, bauxsol-
coated sand (BCS), fly ash (FA), granulated activated carbon (GAC), granulated ferric
hydroxide (GFH), iron oxide-coated sand (IOCS), natural zeolite (NZ), sand, and spinel
(MgAl2O4) as sorbents for removing heavy metals from stormwater are investigated in the
present study. The ability of the sorbents to remove a mixture of As, Cd, Cr, Cu, Ni and Zn
from synthetic stormwater samples were evaluated in batch tests at a starting pH of 6.5.
The metal speciation and saturation data is obtained using the PHREEQ-C geochemical
model and used to elucidate the sorption data. It is found that BCS, FA, and spinel have
significantly higher affinity towards heavy metals mainly present as cationic or non-
charged species (i.e. Cd, Cu, Ni and Zn) compared to those present as anionic species (i.e. As
and Cr). However, IOCS, NZ and sand have higher affinity towards As and Cr, while alumina
has equally high affinity to all tested heavy metals. The Freundlich isotherm model isfound to fit the data in many cases, but ill fitted results are also observed, especially for FA,
BCS and GAC, possibly due to leaching of some metals from the sorbents (i.e. for FA) and
oversaturated conditions making precipitation the dominant removal mechanism over
sorption in batches with high heavy metal concentrations and pH. Calculated sorption
constants (i.e. Kd) are used to compare the overall heavy metal removal efficiency of the
sorbents, which in a decreasing order are found to be: alumina, BCS, GFH, FA, GAC, spinel,
ABCS, IOCS, NZ, bark, and sand. These findings are significant for future development of
secondary filters for removal of dissolved heavy metals from stormwater runoff under
realistic competitive conditions in terms of initial heavy metal concentrations, pH and
ionic strength.
&2006 Elsevier Ltd. All rights reserved.
1. Introduction
Urban stormwater runoff, especially road runoff, is rich
in heavy metals that, unlike organic pollutants, are not
degradable in the environment. The major sources of
heavy metals in stormwater runoff are building materials
(e.g. Cu from roofs and Zn from galvanized steel), and
traffic-related sources such as brake linings (Cu, Ni, Cr, Zn,
Pb), tire wear (Zn), and autocatalysts (Pt, Pd, Rh). Due to
short term (e.g. acute toxicity) and long-term (e.g. carcinoge-
nity and reproducing damages) adverse effects of
heavy metals in the aquatic environment, treatment of
stormwater runoff containing heavy metals receives in-
creased attention.
ARTICLE IN PRESS
0043-1354/$ - see front matter&
2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.watres.2006.10.024
Corresponding author. Tel.: +4545 2516 11; fax: +45 4593 2850.E-mail address:[email protected] (H. Genc--Fuhrman).
W A T E R R E S E A R C H 4 1 ( 2 0 0 7 ) 5 9 1 6 0 2
http://localhost/var/www/apps/conversion/tmp/scratch_10/dx.doi.org/10.1016/j.watres.2006.10.024mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_10/dx.doi.org/10.1016/j.watres.2006.10.0248/12/2019 diversos sorbentes
2/12
Stormwater best management practices (BMPs) are widely
used for handling of stormwater. Although most of the BMPs
are constructed to control the water flow in order to alleviate
peak flows and avoid flooding, more attention has recently
been directed to construct BMPs that also remove pollutants.
The most common BMPs in Denmark are ponds, basins,
wetlands and swales. Of these, some are built to infiltrate
stormwater, whereby pollutants are either accumulated in
the soil or transported to the groundwater. Others are built as
active scavengers for pollutants associated with particles. The
primary inadequacy of such systems is that the colloidal and
truly dissolved fractions cannot be removed via settling;
hence, there is a need for secondary treatment of stormwater,
after passing through a pond, basin or wetland. Filtration of
the stormwater, where both colloidal and truly dissolved
heavy metal fractions are primarily removed via sorption, is
one of the most promising technologies providing that
effective filtration media is used.
Although many studies of heavy metal removal from
stormwater using a variety of sorbents exist, they typically
only report removal percentages and do not document the
experimental conditions thus making the results impossible
to compare directly. Furthermore, the majority of previous
work on heavy metal removal using filtration/adsorption
deals with treatment in single heavy metal systems, and only
a few studies are carried out at realistic experimental
conditions in terms of stormwater (Liu et al., 2005; Pitcher
et al., 2004). However, it is important to understand and
quantify the heavy metal removal properties when they
coexist (Jang et al., 2005;Al-Asheh et al., 2000), and identical
experimental conditions are important when comparing
different sorbents. Consequently, the main objective of the
present study was to evaluate the efficiency of 11 pre-selected
sorbents/filtration materials in batch tests under well-con-
trolled experimental conditions to identify those most
promising for potential use in secondary stormwater treat-
ment systems. Co-existence of the most important heavy
metals in realistic concentrations as well as initial pH and
ionic strength are prioritized properties in this study, whereas
other properties such as the content of organic matter and
colloids are included in a later study focusing only on the
most promising sorbents.
Although a wide range of sorbents are available as possible
filtration media the following 11 are selected based on a
literature survey: activated bauxsol-coated sand (ABCS),
alumina, bauxsol-coated sand (BCS), bark, fly ash (FA),
granulated activated carbon (GAC), granulated ferric hydro-
xide (GFH), iron oxide-coated sand (IOCS), natural zeolite (NZ),
sand, and spinel (MgAl2O4). A mixture of As, Cd, Cr, Cu, Ni
and Zn is chosen, as dissolved fractions of these metals
reportedly pose environmental concerns (Makepeace et al.,
1995). Pb is excluded in this study, as Pb in stormwater is
primarily associated with suspended solids, which are
expected to be satisfactorily removed during pre-filtration
treatment (i.e. by settling). Furthermore, due to increasing
environmental concerns, Pb usage is decreased, which in turn
has significantly reduced Pb concentrations in stormwater
(Furumai et al., 2002). Other metals such as Pt, Pd and Rh were
excluded as they are available in very low concentrations of
less concern.
Among the sorbents tested, alumina and IOCS are com-
monly used metal-oxide-based sorbents for heavy metal
removal from aqueous solutions (Baumgarten and Kirchhau-
sen-Dusing, 1997;Khaoidhiar et al., 2000). Similarly, BCS and
ABCS are previously tested as cost-effective sorbents espe-
cially for As (Genc--Fuhrman et al., 2005). FA is an effective
heavy metal sorbent possibly due to its carbon, silica,
alumina and iron oxides content (Bailey et al., 1999). NZ is
widely available at low cost and has high ion exchange
capacity, thus it may have the potential to be used for
stormwater treatment (Pitcher et al., 2004). Activated carbon
(AC) and GAC are often employed for heavy metal removal
(Chen and Wang, 2000). On the other hand, GFH is effectively
used for arsenate removal (Sperlich et al., 2005) and included
here to observe whether the sorbent is capable of removing
other heavy metals. Bark is a by-product of the timber
industry and its high tannin content, especially polyhydroxy
polyphenol groups, gives bark an ion exchange ability ( Bailey
et al., 1999). Moderate heavy metal removal using bark is
reported (Palma et al., 2003), and bark is suggested as a cost-
effective alternative to AC (Al-Asheh et al., 2000). Sand is also
included in the batch tests for two main reasons: first, sand
filtration is commonly used for stormwater treatment (San-
salone, 1999), and the current study may provide data for
comparison; second, because IOCS, ABCS, and BCS are
prepared by coating a sorbent onto sand, having the removal
data for sand itself is useful for comparative purposes (i.e. to
see the effect of the coating). To the authors knowledge, no
study has yet been published using spinel for heavy metal
removal from water.
2. Materials and methods
2.1. Sorbents
The supplier information and physical characteristics of the
sorbents are provided inTable 1. Specific surface areas of the
samples were measured using the single-point-N2-BET meth-
od. Further details about the various sorbents, their physical
and chemical properties and preparation can be found from
the references cited inTable 1. Similar particle sizes of the
sorbent materials were used, as particle size may have an
effect on the removal (Smith, 1998). All sorbents were sieved
(or crushed if required) to the desired particle size (0.61 mm)and dried at 40 1C for 3 h without additional treatment before
being used in batch experiments, except for sand which was
first acid washed for 24 h using 10% HNO3. Alumina, bark, FA,
GAC, GFH, NZ, sand and spinel were commercially supplied,
while ABCS, BCS and IOCS were prepared in the laboratory.
BCS and ABCS were prepared from the alumina production
industry by-product red mud (Genc--Fuhrman et al., 2005),
while IOCS was prepared in the laboratory by coating goethite
onto sand (Lo et al., 1997). FA is a waste incineration by-
product supplied from a municipal waste incineration plant
(MSWI) in Denmark. There are several NZs but here clinopti-
lolite, the most abundant naturally occurring zeolite, was
used.
ARTICLE IN PRESS
W AT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 5 9 1 6 0 2592
8/12/2019 diversos sorbentes
3/12
2.2. Heavy metal concentrations
It is documented in the literature that the concentrations of
heavy metals in stormwater vary several orders of magnitude
(Makepeace et al., 1995;Walker et al., 1999). In this study, an
attempt was made to use realistic concentration ranges,
where the lowest values were below the Danish emission
limit values for discharge into fresh surface waters (DELV; see
Table 2) (DME, 1996); while, the highest values were signifi-
cantly higher representing extreme conditions such as
snowmelt or first flush concentrations after long dry weather
periods. The purpose was to evaluate and compare the
sorbents at both highly polluted and slightly polluted condi-
tions, since it is well known that some sorbents are effective
at rather high concentrations, but are less efficient and may
leach specific contaminants at low concentrations.
2.3. Batch experiments
Sorption experiments were carried out using synthetic
stormwater samples in 50 mL conical PE batches at room
temperature (2271 1C). The required concentrations of As, Cd,
ARTICLE IN PRESS
Table 1 Overview and physical characteristics of the sorbents used in the study
Sorbent Supplier pHpzca Par. size, mm Surface areab, m2/g Literature
Aluminac Haldor Topse, Denmark 9.1d 0.61.0 238.9 (Baumgarten and Kirchhausen-Dusing, 1997)
ABCS Prepared in the laboratory 8.3e 00.5 42.2 (Genc--Fuhrman et al., 2005)
Bark Zugol, Sweden 0.61.0 0.32f (Vasconcelos and Bec-a, 1997)
BCS P repared in the l aboratory
8.3
e
00.5 4.6 (Genc-
-Fuhrman et al., 2005)FA MSWI, Denmark 6.2 00.5 2.5 (Wang et al., 2004)
GAC Kemira, Denmark 6.1 0.61.0 784.5 (Marzal et al., 1996)
GFH IPSg, Germany 7.5-8 0.61.0 172.6 (Sperlich et al., 2005)
IOCS Prepared in the laboratory 9 0.61.0 1.5 (Lo et al., 1997)
NZh Euremica, UK 5.6i 01.0 24.2 (Matheickal and Yu, 1997)
Sand Dansand, Denmark 0.7 0.61.0 0.1 (Urbonas, 1999)
Spinelc,j Haldor Topse, Denmark 11k 0.61.0 11.8
a Point of zero charge.b BET-N2single point analysis.c Developed commercially as catalysts.d From (Stumm, 1992).e pHpzc of red mud (Apak et al., 1998).f From the cited literature.g
Ingenieurgesellschaft Prof. Dr. Sieker MBH.h Clinoptilolite ((Ca, Na2, K2)3[Al6Si30O72].24H20).i pHpzc is reported at the presence of cations (Feng et al., 2000).j MgAl2O4.k From (Anderson, 2005).
Table 2 Quantification limits, Danish emission limit values, runoff and initial concentrations in the batch experiments(C0) of the heavy metals.
Heavy metals and concentrations (mg/L)
As Cd Cr Cu Ni Zn
Quantif. limit 1.0 0.05 0.5 1.0 0.5 1.0DCRa 1210 0.0513,730 12300 0.061410 149,000 0.722,000
ACRb 1210 0.311 10230 6.5150 6150 16.6580
DELVc 4 5 7 12 160 110
Batch 1 o1 0.1 o0.5 2.6 0.6 22.5
Batch 2 2.6 3.0 3.5 7.6 12.3 67.6
Batch 3 14 10.4 18 36.1 58.9 340
Batch 4 23 23.1 34.2 20.4 89.9 543
Batch 5 52 221 376 588 178 1350
Batch 6 128 160 212 246 545 3400
Batch 7 396 735 784 1250 2220 13,700
Batch 8 1000 2670 2830 1820 8640 52,300
a Detected.b Average heavy metal concentration ranges reported in the literature for stormwater runoff (Makepeace, 1995).
c Danish emission limit values (DELV) (DME, 1996).
W AT E R R E S E A R C H 41 (2007) 591 602 593
http://-/?-http://-/?-8/12/2019 diversos sorbentes
4/12
Cr, Cu, Ni and Zn were obtained by step-by-step diluting their
commercial stock solutions at 1000mg/L in HNO3 to the
desired concentrations. The ionic strength of the water
samples was controlled using 0.01M NaCl, and all samples
had 0.003 M NaHCO3 to minimize the pH changes during
experiments. It was decided not to use a stronger buffer since
this would influence the heavy metal speciation in a way not
realistic for stormwater. After preparing the solutions with
the required heavy metal concentrations, ionic strength, and
buffer; the pH was adjusted to 6.5 using strong acid or base
solutions to resemble the lowest realistic pH observed in
stormwater runoff. Afterwards, 50 mL of the heavy metal
solutions with pH 6.5 were added to the PE flasks with 1g
sorbent to obtain a 20 g/L sorbent dosage. Then, the solutions
and the sorbents were mixed by gently shaking the batches,
and the pH was re-measured and re-adjusted to 6.5 in each
batch within a few minutes. Following the pH adjustment the
batches were capped tightly and shaken for 48h. The pH
values were checked after the completion of the shaking time
to identify possible pH variations from the starting values.
The temporal evolution of the solution pH due to sorbent-
sorbate interactions was monitored in a separate set of
experiments for 48h. The used 48h shaking time and 20g/L
solid/solution ratio may overestimate the actual removal of
metals occurring in real flow-through stormwater filters but
were used here to compare the 11 sorbents under identical
conditions. It is noted that further studies on the effects of
sorption kinetics and filter surface loading are in progress.
Later, the batches were taken from the shaker and
centrifuged for 20min at 2000rpm (894g), and finally the
samples were carefully dispensed to 50 mL PE sample cups,
acidified to pH 1.52 using strong HNO3 and stored at 4 1C
until the heavy metal measurements.
Two sets of control batches were also run. The first set
included 11 batches with different sorbents without any
heavy metal addition, following the same experimental
procedure as described previously to determine whether any
of the sorbents would release heavy metals. The second set of
control batches also followed the same experimental proce-
dure, but without adding any sorbents. This trial included 8
batches with the initial heavy metal concentrations given in
Table 2in order to quantify any losses due to sorption to the
PE flasks or additions due to contamination during the
handling of the samples. No significant metal sorption on to
the flask walls was observed and all concentrations measured
from the second set of control experiments were within75%
of the initial metal concentrations.
All chemicals used were reagent grade and used without
any further purification.
2.4. Heavy metal quantification
All the samples were sent to a certified commercial laboratory
(Analytica, Sweden) for quantification of heavy metal con-
centrations, where inductively coupled plasma atomic emis-
sion spectrometry (ICP-AES) or inductively coupled plasma
sector field mass spectrometry (ICP-SFMS) was used, accord-
ing to USEPA Methods 200.7 (modified) and 200.8 (modified),
respectively. The methods quantification limits are given in
Table 2.
2.5. Heavy metal speciation calculations
Speciation calculations for the heavy metals were carried out
using a computer model, PHREEQ-C version 2 with a
thermodynamic database (i.e. LLNL.DAT) (Parkhurst and
Appelo, 1999), at a wide pH range employing the initial heavy
metal concentrations used in the experiments. The model
was also used for identification of oversaturated conditions of
any relevant metal mineral/salt during the experiments.
2.6. Analyzing sorption data
The solid phase heavy metal concentrations, qe (mg/g), were
determined for each sorbent by analyzing the corresponding
heavy metal concentration before and after the treatment
using the equation below:
qe C0 Ce
X , (1)
whereC0 and Ceare the initial and equilibrium heavy metal
concentrations in the solution (mg/L), and X is the sorbentdosage (g/L).
Freundlich isotherm,
logqe logK 1n
logCe, (2)
whereK is correlated with the quantity of sorbate associated
with the sorbent, and n is the Freundlich isotherm constant
related to the strength of the sorption, was used to investigate
the sorption data further.
The overall efficiency of each sorbent in terms of heavy metal
removal was determined using sorption constants Kd(L/g)
Kd qeCe
, (3)
whereKdvalues were calculated corresponding to all the initial
concentrations used, and average Kd values were calculated
and used to rank the sorbents for each metal.
3. Results
3.1. pH variations during the batch experiments
The pH increased with time for all sorbents except bark,
where the pH decreased (seeFig. 1). It was accentuated that
the pH increases were mainly due to dissolution of the
sorbent minerals in the process of shaking, while for bark the
observed pH decrease was primarily attributed to the release
of natural organic matter containing both acids and phenols
(confirmed by a yellow coloring of the samples during the
batch tests).
In most cases the pH increase was moderate, e.g. from pH
6.5 to 6.8 for ABSC, while for FA a major pH shift was observed
from pH 6.5 to 10.5. It is noted that the reason that the pH is
not changed significantly for ABCS is that ABCS is prepared
from activated bauxsol (Genc--Fuhrman et al., 2004), which
was refluxed by HCl previously, and this process probably
already neutralized the surface. A final pH of 7.7 was observed
in the NZ batches due to hydrolysis of zeolites as well as
cationic exchange. Similar results are reported elsewhere
with a remark that the pH increase is almost unavoidable
ARTICLE IN PRESS
W AT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 5 9 1 6 0 2594
8/12/2019 diversos sorbentes
5/12
in a NZ heavy metal system (Ouki and Kavannagh, 1997;
Pitcher et al., 2004).
3.2. Heavy metal speciation and saturation conditions
The pH-dependent distributions between the dissolved metal
species (as calculated using PHREEQ-C) are presented inFig. 2
for stormwater composition corresponding to batch 5 (Table2). The results illustrate that both As and Cr are present as
anionic species; and the dominating As specie is HAsO42 in
the pH range 6.510, while Cr is mainly present as CrO42 at pH
values above 6.5. The other metals show somewhat more
complicated patterns, for example the most important Cd
species are CdCl+ and Cd2+ below pH 9.5, whereas Cd(CO3)22
starts dominating above pH 10. The non-charged CuCO3is the
dominating Cu-specie in the pH range 6.59.0, while the
negatively charged species such as Cu(CO3)22 and
CuCO3(OH)22 dominate at higher pH values. The dominating
Ni-specie is Ni2+ until about pH 9.5, succeeded by the
hydroxide species Ni(OH)2and Ni(OH)3 at higher pH. Zinc is
mainly present as Zn2+ below pH 8.5 and as Zn(OH)2 atpH49.0.
It is evident from the results presented in Table 3that the
number of oversaturated minerals/salts is increasing with
increasing pH and increasing initial heavy metal concentra-
tion. Therefore, sorption may no longer be the primary
removal process in those batches, since precipitation of
minerals/salts on both the sorbents and the walls of the PE
flasks may also occur.
3.3. Equilibrium concentrations and sorption isotherms
The results of simultaneous removal of Cd, Cu, Ni, and Zn
using the tested sorbents are presented in Fig. 3on a double
logarithmic scale, while the results for As and Cr are
presented inFig. 4on a double or semi logarithmic scale. In
these figures, the equilibrium heavy metal concentrations (Ce)
are depicted on the horizontal axis vs. the solid phase metal
concentrations (qe) on the vertical axis. As expected the
amount of heavy metal removed increased with increasing
initial heavy metal concentration in nearly all of the batches,
except a few cases where leaching was observed for As and Cr
(hence the semi-logarithmic depiction). Leaching of As was
observed when using bark, BCS, GAC, FA, NZ and sand, while
leaching of Cr was observed when using ABCS, BCS, FA and
spinel. It is highlighted that leaching of As and Cr was also
observed in the control batches involving sorbents but no
heavy metal addition. For example, 7, 114, 10 and 9 mg/L As
leaching was observed from Bauxsol, FA, GAC and NZ,
respectively. In the case of Cr, leaching was 137, 124, 40, 8
and 20mg/L from ABCS, FA, NZ, sand and spinel, respectively.
The calculated Freundlich isotherm parameters (Kandn, cf.
Eq. (2)) are shown inTable 4. Although a high correlation was
observed in some cases, the r2 values were low especially for
bark, BCS and FA. Fitting Langmuir isotherms to the data did
not improve the correlation. It is noted that here although
r2X0:5 signifies a statistically significant correlation, using
the estimated Freundlich isotherms for prediction is consid-
ered highly uncertain unless r2 close to 1.0 is reached.
3.4. Sorbent ranking
The calculated averageKdvalues for each batch are also given
in Table 4, while the overall results are presented in Fig. 5.
Each tested sorbent demonstrates different removal effi-
ciency for each heavy metal, for example, while IOCS
effectively retains As, it fails to repeat its performance for
Cd and Ni. Similarly, FA has very high affinity for Cd and Ni,
ARTICLE IN PRESS
Fig. 1 The variation of pH values as a function of time during the sorption experiments with starting pH of 6.5, initial As, Cd,
Cr, Cu, Ni and Zn concentrations of 1 mg/L, sorbent dosage of 20 g/L, ionic strength of 0.01 M NaCl and 0.003 M HCO3 buffer.
W AT E R R E S E A R C H 41 (2007) 591 602 595
8/12/2019 diversos sorbentes
6/12
but it is ineffective for As and Cr (leaching). However, alumina
is consistently effective for all heavy metals tested. Conse-
quently, the overall heavy metal removal efficiency of the
sorbents at the experimental conditions in a decreasing order
is, alumina4BCS4GFH4FA4GAC4spinel4ABCS4IOCS4
NZ4bark4sand.
4. Discussion
4.1. Alumina
At the experimental conditions used alumina is capable of
removing all heavy metals to concentrations below the DELV
except for the batches with the highest concentrations of Cd,
Cu and Zn. This high efficiency is probably due to aluminas
high surface area (Table 1) and favorable heavy metal surface
interactions.
The dominant heavy metal species in the alumina system
(420% at pH 7.4, cf.Fig. 2) are HAsO42, CdCl+ and Cd2+, CrO4
2,
CuCO3, Ni2+ and Zn2+. The alumina surface is expected to be
positively charged at the experimental pH (note that the pHpzcof alumina is about 9, but decreases when forming hydro-
xides), which enhances electrostatic removal of the nega-
tively charged HAsO42 and CrO4
2 species. On the other hand,
alumina performed equally well for the heavy metal cation
species, which can be explained by other removal mechan-
isms like surface complexation (hydroxide groups), pore
ARTICLE IN PRESS
Fig. 2 Speciation of As, Cd, Cr, Cu, Ni and Zn in water obtained using the PHREEQ-C model with ionic strength of 0.01 M NaCl,
0.003M NaHCO3buffer and heavy metal concentrations according to batch 5 (seeTable 2).
W AT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 5 9 1 6 0 2596
8/12/2019 diversos sorbentes
7/12
diffusion and specific adsorption. For batch numbers 58
precipitation of Cu, Cd and Zn minerals/salt may also be
counted as a relevant removal process (oversaturation, cf.
Table 3).
4.2. Activated bauxsol-coated sand and bauxsol-coatedsand
ABCS efficiently retains As to values below the DELV
independent of the initial As concentration, and it is also
one of the most efficient sorbents for Cu. However, the
sorbent fails to remove Cr, Cd, Ni, and Zn as efficiently as As
and Cu.BCS performs better than ABCS for removing all metals
except for As. This can be explained by the negatively charged
BSC-surface at the experimental pH (8.6) compared to that of
ABCS (6.8). Consequently, the electrical attraction increases
the removal of positively charged metal ions. Furthermore,
the surface complexation between dissolved species and
oxide- and hydroxide groups on the surface will increase with
increasing pH. Oversaturation occurs only in batches 7 and 8
for ABSC (Table 3), thus sorption may be the dominant
mechanism for ABCS. On the other hand, for BCS it is
expected that e.g. Zn minerals/salts can precipitate in all
batches except for batch 1, and that Cd and Cu minerals/salts
may precipitate for batches 38. It can, therefore, besuspected that precipitation and not sorption is the major
removal process for Zn, Cd, and Cu in the BCS batches.
4.3. Bark
Bark suffered from low heavy metal removal efficiency (see
Fig. 5), and minor As and Cr leaching is observed especially at
low initial heavy metal concentrations (see Fig. 4). Further-
more, water discoloration is observed most likely due to the
release of organic acids and phenols. Among the metals,
especially Cu is known to form strong complexes with natural
organic acids, which can reduce the sorption (Duker et al.,
1995;Dario and Ledin, 1997), and the low averageKdobserved
for Cu is, therefore, probably attributed to this phenomenon.
Considering the low affinity of the bark to the tested heavy
metals and the leaching of organic matter as well as As and
Cr, bark may be considered unsuitable as a secondary
treatment filtration media.
4.4. Fly ash
FA has high affinity for Ni, Zn, Cu and Cd, which at the high
pH during the experiments with FA mainly are present non-
charged or anionic species (i.e. Ni(OH)2, Zn(OH)2, Cd(CO3)22,
Cu(CO3)22). In addition, oversaturated conditions and conse-
quently precipitation of Cd, Cu, Ni and Zn minerals/salts (seeTable 3) will also contribute to the overall removal. The results
indicate that FA has low affinity to Cr and As species.
However, this is due to the considerable leaching from the
sorbent (seeFig. 4b). Despite the high removal efficiency of FA
towards Cd, Cu, Ni and Zn, FA is unsuitable for secondary
treatment of stormwater due to the leaching of As and Cr and
the high effluent pH.
4.5. Granulated activated carbon
GAC is the most effective sorbent for Cr and it has moderate
to high efficiency for Cd, Cu, Ni and Zn. On the other hand, it
has a significantly lower efficiency for As compared to that ofthe other heavy metals (Fig. 5). Oversaturated conditions of
Zn-, Cd- and Cu- minerals/salts may also contribute to the
removal efficiency in the GAC batches in the same manner as
in the BCS batches (Table 3).
4.6. Granulated ferric hydroxide
GFH was the third most effective sorbent when ranked
according to average Kd values (Fig. 5). Previously GFH has
been advocated as a highly efficient sorbent especially for
arsenate removal (Sperlich et al., 2005), but here GFH
demonstrated moderate As removal efficiency and is out-
performed by alumina, IOCS, ABCS, and spinel. This may be
ARTICLE IN PRESS
Table 3 Oversaturated heavy metal minerals/salts, e.g. SI40, where SI is defined as log(IAP/Ks) in the batch experiments,estimated by PHREEQ-C modelling with ionic strength of 0.01M NaCl and 0.03 M NaHCO3buffer
pH Batch 1 Batch 2 Batch 3 Batch 4 Batch 5 Batch 6 Batch 7 Batch 8
5.8 (Bark) Zn, Cd, Cu
6.8 (ABCS) Zn, Cd, Cu Zn,Cd, Cu
7.2 (Sand) Cu, Cd, Zn Zn, Cd, Cu Zn, Cd, Cu7.4 (Alumina) Cu, Cd Zn, Cd Zn, Cd, Cu Zn, Cd, Cu
7.5 (IOCS) Zn, Cd, Cu Zn, Cd, Cu Zn, Cu, Cd Zn, Cd, Cu
7.7 (NZ) Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu Zn,Cd, Cu
7.8 (GFH) Zn, Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu
8.6 (BCS, GAC) Zn Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu, As
8.7 (Spinel) Zn Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu Zn, Cd, Cu,
Ni
Zn, Cd, Cu, Ni, As
10.6 (FA) Cu Ni, Cd, Cu Zn, Ni, Cd,
Cu
Zn, Ni, Cd,
Cu
Zn, Ni, Cd,
Cu
Zn, Ni, Cd,
Cu
Zn, Ni, Cd,
Cu
Zn, Ni, Cd, Cu, As
W AT E R R E S E A R C H 41 (2007) 591 602 597
8/12/2019 diversos sorbentes
8/12
due to the pH differences, as the literature reports higher
arsenate removal efficiencies using GFH at the pH values
below 7.5, e.g. below pHpzc(seeTable 1) while the equilibrium
pH here was 7.8.
Despite its medium As removal efficiency, GFH is the
second most effective sorbent for Cr and it is also effec-
tive for Cd, Ni and Zn. The dominant heavy metal species
at pH 7.8 are HAsO42, CdCl+ and Cd2+, CrO4
2, CuCO3, Ni2+,
and Zn2+ (Fig. 1). The pH is close to pHpzc, e.g. both posi-
tively and negatively charged groups are available at the
sorbent surface. This indicates that the electrostatic
attraction as well as surface complexation can be impor-
tant removal mechanisms. Moreover, the saturation calcula-
tions indicate that Cd, Cu and Zn could also be removed
due to precipitation of their minerals/salt in batches 58 (see
Table 3).
4.7. Iron oxide-coated sand
IOCS has moderate to poor heavy metal removal efficiency
compared to that of the other sorbents, except from bark,
sand and NZ (Fig. 5). Interestingly, IOCS is the second most
ARTICLE IN PRESS
Fig. 3 Cd, Cu, Ni and Zn removal using (a) sorbents with high heavy metal removal efficiency, (b) sorbents with moderate or
low heavy metal removal efficiency; with starting pH of 6.5, sorbent dosage of 20 g/L, ionic strength of 0.01M NaCl, and
0.003M NaHCO3buffer. DELV is Danish emission limit values for fresh surface waters.
W AT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 5 9 1 6 0 2598
8/12/2019 diversos sorbentes
9/12
8/12/2019 diversos sorbentes
10/12
4.11. Adsorption isotherms
Freundlich and Langmuir isotherms are commonly used in
the literature to study heavy metal sorption. Here, it was
observed that the Freundlich isotherm fitted the data better
suggesting heterogeneous surface characteristics. When
evaluating FA, spinel, and BCS (sorbents with high end pH)
sorption isotherms gave lower r2 values especially for Zn, Ni,
Cu and Cd (seeTable 4). In these systems, precipitation may
play a role as removal mechanism besides adsorption, which
in turn may have decreased the capability of the Freundlich
isotherms to predict the removal.
Leaching of e.g. As from FA and bark could also provide
explanation on why the isotherm did not explain the data
very well.
5. Conclusions
In this study, 11 sorbents are examined in batch tests as
potential filter media for secondary treatment of stormwater.
The sorbents are tested in for the combined removal of As,
Cd, Cr, Cu, Ni and Zn removal from synthetic stormwater
samples at a starting pH of 6.5. However, the pH values are
changed to 5.8 (bark), 6.8 (ABCS), 7.2 (sand), 7.4 (alumina), 7.5
(IOCS), 7.7 (NZ), 7.8 (GFH), 8.6 (BCS, GAC), 8.7 (spinel) and 10.6
(FA) during the experiments, suggesting that in a secondary
stormwater treatment system with sufficient reaction time,
the pH of the discharged water is controlled by the sorbent
material rather than the inlet water composition.
It is found that the magnitude of the heavy metal removal
depends on the initial heavy metal concentration and varies
to a great extent among the sorbents. For example, while
sand and bark fail to remove all heavy metals to meet the
DELV; alumina, BCS, GFH and GAC can efficiently remove the
heavy metals down to the DELV. The results from the present
study also show that the heavy metal speciation and the
surface charge are pH dependent, and that several possible
removal mechanisms interact in a complex manner.
The Freundlich isotherm is applied to the sorption data and
the constantsK and n are presented. Although the isotherm
fit the data in many cases, ill fitted results are also observed
especially for FA, BCS and GAC possibly due to leaching of
some metals from the sorbents and oversaturated conditions
making precipitation the dominant removal mechanism over
sorption in batches with high heavy metal concentrations
and pH.
The sorbents are ranked using average Kd values, and
it is found that when overall heavy metal removal is
considered alumina is the most effective followed by BCS,
GFH, FA, GAC, spinel, ABCS, IOCS, NZ, bark, and sand.
It is highlighted that despite the high affinity of FA, NZ
and bark to some heavy metals leaching of especially As and/
or Cr limits their usage. The current study provides important
new information on the efficiency of a range of potential
sorbents in a multiple sorbate system using experimental
conditions, which resemble those of real life stormwater
runoff in terms of initial heavy metal concentration, pH, and
ionic strength.
ARTICLE IN PRESS
Table4Freundlichisotherm
constan
ts(Kandn,cf.Eq.(2))andaverageKd
valuesa
Sorbents
Heavymetals
As
Cd
Cr
Cu
Ni
Zn
K,
L/mg
n
r2
Kd
a,
L/g
K,
L/mg
n
r2
Kd
a,
L/g
K,
L/mg
n
r2
Kd
a,
L/g
K,
L/mg
n
r2
Kd
a,
L/g
K,
L/mg
n
r2
Kd
a,
L/g
K,
L/m
g
n
r2
Kd
a,
L/g
Alumina
1.6
108
0.0
1
0.9
3
5.1
1
6.3
5
0.8
6
0.9
1
6.8
4
0.0
5
0.6
0
0.9
3
0.0
8
0.0
6
0.5
3
0.80
0.8
9
1.4
6
0.6
2
0.7
3
7.4
7
0.05
0.5
4
0.5
7
3.1
5
ABCS
0.3
9
0.2
0
0.6
3
2.2
6
0.3
5
1.0
6
0.6
0
1.1
5
0.0
1
0.9
3
0.8
6
0.0
1
1.0
2
0.7
6
0.58
3.3
6
0.6
1.5
7
0.9
1
0.1
5
0.99
1.3
2
0.6
7
0.6
1
Bark
0.3
8
8.7
4
0.0
9
0.0
0
0.2
6
1.1
5
0.7
8
0.2
2
0.0
6
1.0
4
0.9
6
0.0
2
0.0
9
0.4
2
0.93
0.0
3
0.1
8
1.3
1
0.8
0
0.2
0
0.0
01
0.6
3
0.8
7
0.0
7
BCS
0.0
6
0.7
8
0.5
9
0.0
9
11.1
7
0.8
6
0.5
3
34.7
4
0.1
2
2.8
1
0.3
8
0.0
0
0.3
3
0.4
7
0.32
5.4
4.4
5
1.1
2
0.6
9
7.5
8
3.59
1.2
1
0.2
0
8.3
7
FA
4.3
104
0.4
7
0.5
3
0.1
1
5.0
1
0.6
4
0.5
3
12.3
5
0.0
3
0.1
0.4
6
0.20
2.9
6
0.0
4
0.3
0
0.5
5
8.5
4
0.04
0.4
2
0.8
5
6.3
3
GAC
0.0
9
1.1
8
0.5
5
0.0
2
1.4
5
1.1
0.8
3
2.3
8
0.3
3
0.9
3
0.8
0.7
3
0.3
7
0.8
0.76
1.2
8
0.9
7
0.9
4
0.7
4
2.0
0
0.91
0.9
9
0.5
4
1.3
7
GFH
0.1
3
0.6
5
0.8
7
0.4
9
2.7
4
0.5
3
0.8
9
5.6
4
0.0
3
0.6
0.7
6
0.3
9
0.2
6
0.7
8
0.61
0.8
9
0.1
4
0.5
3
0.5
3
1.8
1
1.21
0.8
5
0.8
1
2.8
7
IO
CS
0.9
3
0.1
9
0.3
8
3.1
6
0.0
7
1.2
5
0.8
9
0.0
6
0.0
4
1.1
6
0.9
6
0.0
5
0.4
1
1.2
9
0.64
0.3
2
0.1
7
2.1
3
0.5
7
0.0
1
0.24
1.3
4
0.9
1
0.0
7
NZ
0.0
4
0.7
9
0.8
0.0
5
0.2
1
1.0
7
0.9
4
0.2
4
0.0
4
1.6
2
0.8
0
0.0
2
0.1
3
0.6
8
0.76
0.7
9
0.5
0
1.4
3
0.9
7
0.1
1
0.36
1.1
5
0.9
8
0.1
9
Sand
0.0
4
1.0
9
0.8
3
0.0
4
0.4
2
3.5
7
0.2
7
0.0
0
0.0
4
1.7
3
0.7
9
0.0
1
0.1
0
1.5
7
0.59
0.0
5
0.1
4
2.9
8
0.4
2
0.0
0
0.03
1.3
7
0.8
5
0.0
0
Spinel
0.2
9
0.5
5
0.5
3
1.0
4
1.7
6
1.1
7
0.6
5
5.1
7
0.0
5
1.2
4
0.6
2
0.0
1
0.0
8
0.6
3
0.71
0.7
9
1.6
7
1.3
3
0.5
2
3.9
3
2.55
1.2
2
0.6
3.2
1
a
AverageofKd
values.
W AT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 5 9 1 6 0 2600
8/12/2019 diversos sorbentes
11/12
Acknowledgements
This work is funded by Danish Technical Research Council
(Grant no.: 26-03-0326). Laboratory technician Susanne Kruseand the M.Sc. students Yushan Zhou and Peng Wu are
acknowledged for their contribution during the experiments.
Thanks are also to Haldor Topse A/S, Kemira Denmark and
Ingenieurgesellschaft Prof. Dr. Sieker MBH for supplying the
alumina, GAC, and GFH, respectively.
R E F E R E N C E S
Anderson, M.A., 2005. Personal communication.Al-Asheh, S., Banat, F., Al-Omari, R., Duvnjak, Z., 2000. Prediction of
binary sorption isotherms for the sorption of heavy metals by
pine bark using single isotherm data. Chemosphere 41, 659665.Apak, R., Guc-lu, K., Turgut, M.H., 1998. Modelling of Copper(II),
Cadmium(II), and Lead(II) adsorption on red mud. J. ColloidInterface Sci. 230, 122130.
Bailey, S.E., Olin, T.J., Bricka, R.M., Adrian, D.D., 1999. A review ofpotentially low-cost sorbents for heavy metals. Water Res. 33,24692479.
Baumgarten, E., Kirchhausen-Dusing, U., 1997. Sorption of metalions on alumina. J. Colloid Interface Sci. 194, 19.
Chen, J.P., Wang, X., 2000. Removing copper, zinc, and lead ion bygranular activated carbon in pretreated fixed-bed columns.Sep. Purif. Technol. 19, 157167.
Dario, M., Ledin, A., 1997. Sorption of Cd to colloidal ferrichydroxides-impact of pH and organic acids. Chem. Spec.Bioavailab. 9, 314.
Danish Ministry of Environment (DME), 1996. Departmental orderon quality standards for surface waters and requirements fordischarge of some hazardous substances to streams, lakes andsea. BEK no. 921 of 08/10/1996.
Duker, A., Ledin, A., Karlsson, S., Allards, B., 1995. Adsorption ofzinc on colloidal (hydr)oxides of Si, Al, and Fe in the presenceof fulvic acid. Appl. Geochem. 10, 197205.
Feng, D., Aldrich, C., Tan, H., 2000. Removal of heavy metal ions bycarrier magnetic separation of adsorptive particulates. Hy-drometallurgy 56, 358369.
Furumai, H., Balmer, H., Boller, M., 2002. Dynamic behavior ofsuspended pollutants and particle size distribution in highwayrunoff. Water Sci. Technol. 46, 413418.
Genc--Fuhrman, H., Tjell, J.C., Mcconchie, D., 2004. increasing thearsenate adsorption capacity of neutralized red mud (bauxsol).
J. Colloid Interface Sci. 271, 313320.
Genc--Fuhrman, H., Bregnhj, H., McConchie, D., 2005. Arsenateremoval from water using sand-red mud columns. Water Res.39, 29442954.
Jang, A., Seo, Y., Bishop, P.L., 2005. The removal of heavy metals in
urban runoff by sorption on mulch. Environ. Pollut. 133,117127.
Khaoidhiar, S., Azizizan, M.F., Osathaphan, K., Nelson, P., 2000.Copper, chromium, and arsenic adsorption and equilibriummodeling in an iron-oxide-coated-sand, background electro-lyte system. Water Air Soil Pollut. 119, 105120.
Liu, D., Sansalone, J.J., Cartledge, F.K., 2005. Comparison ofsorptive filter media for treatment of metals in runoff. J.Environ. Eng. 131, 11781186.
Lo, S., Jeng, H., Lai, C., 1997. Characteristics and adsorp-tion properties of iron-coated-sand. Water Sci. Technol. 35,6370.
Marzal, P., Seco, A., Gabaldon, C., 1996. Cadmium and zincadsorption onto activated carbon: influence of temperature,pH, and metal/carbon ratio. J. Chem. Technol. Biotechnol. 66,
279285.Makepeace, D.K., Smith, D.W., Stanley, S.J., 1995. Urban storm-
water quality: summary of contaminated data. Crit. Rev.Environ. Sci. Technol. 25, 93139.
Matheickal, J.T., Yu, Q., 1997. Biosorption of lead(II) fromaqueous solutions by phellinus badius. Miner. Eng. 10 (9),947957.
Ouki, S.K., Kavannagh, M., 1997. Performance of natural zeolitesfor the treatment of mixed metal-contaminated effluents.Waste Manage. Res. 15, 383394.
Palma, G., Freer, J., Baeza, J., 2003. Removal of metal ions bymodifiedPinus radiatebark and tannins from water solutions.Water Res. 37, 49744980.
Parkhurst, D.L., Appelo, C.A.J., 1999. Users guide to PHREEQC(Version 2): a computer program for speciation, batch reac-
tions, one dimensional transport and inverse geochemicalcalculations. Water Resources Investigations Report 99-4259,US Geological Survey.
Pitcher, S.K., Slade, R.C.T., Wards, N.I., 2004. Heavy metal removalfrom motorway stormwater using zeolites. Sci. Total Environ.334335, 161166.
Sansalone, J.J., 1997. Infiltration as an urban source control formetal elements and solids. In: Proceedings of the 27thCongress of the International Association of Hydraulic Re-search, IAHR, Part A.
Sansalone, J.J., 1999. Adsorptive infiltration of metals inurban drainagemedia characteristics. Sci. Total Environ.235, 188.
Smith, E.H., 1998. Modeling batch kinetic studies of cadmiumremoval by a recycled iron adsorbent. Sep. Sci. Technol. 33,
149168.
ARTICLE IN PRESS
Fig. 5 Ranking of the sorbents based on average Kdvalues (sorbents with lower rank are more effective).
W AT E R R E S E A R C H 41 (2007) 591 602 601
8/12/2019 diversos sorbentes
12/12
Sperlich, A., Werner, A., Genz, A., Amy, G., Worch, E., Jekel, M.,2005. Breakthrough behaviors of granular ferric hydroxide(GFH) fixed-bed adsorption filters: modeling and experimentalapproaches. Water Res. 39, 11901198.
Stumm, W., 1992. Chemistry of the Solid-water Interface. Wiley,Canada, 20pp.
Urbonas, R., 1999. Design of a sand filter for stormwater qualityenhancement. Water Environ. Res. 71 (1), 102113.
Vasconcelos, L.A.T., Bec-a, C.G.G., 1997. A study of a continuous-flow process for the removal of Pb(II) ions in aqueous solution
using fixed-bed columns packed with natural and activatedpine bark. Eur. Water Pollut. Control 7 (3), 718.
Walker, W.J., McNutt, R.P., Maslanka, C.K., 1999. The potentialcontribution of urban runoff to surface sediments of thePassaic river: sources and chemical characteristics. Chemo-sphere 38, 363377.
Wang, J., Teng, X., Wang, H., Ban, H., 2004. Characterizing themetals adsorption capabilities of a class F coal fly ash. Environ.Sci. Technol. 38 (24), 67106715.
ARTICLE IN PRESS
W AT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 5 9 1 6 0 2602