ARTICLE IN PRESS

0043-1354/$ - se

doi:10.1016/j.w

�Correspond

fax: +3214 58 0

E-mail addr1Present add

ment, Catholic

3001 Heverlee,

Water Research 39 (2005) 3531–3540

www.elsevier.com/locate/watres

Effect of humic acids on heavy metal removal by zero-valentiron in batch and continuous flow column systems

Jan Driesa,b, Leen Bastiaensa,�, Dirk Springaela,1, Stefaan Kuypersc,Spiros N. Agathosb, Ludo Dielsa

aDepartment of Environmental and Process Technology, Flemish Institute for Technological Research (Vito), Boeretang 200,

2400 Mol, BelgiumbUnit of Bioengineering, Catholic University of Louvain, Croix du Sud 2, boıte 19, 1348 Louvain-la-Neuve, Belgium

cDepartment of Material Technology, Flemish Institute for Technological Research (Vito), Boeretang 200, 2400 Mol, Belgium

Received 31 January 2005; received in revised form 31 May 2005; accepted 22 June 2005

Available online 10 August 2005

Abstract

The effects of Aldrich humic acids (HA) on the removal of Zero-valent iron (ZVI) was investigated in laboratory

systems. In batch, the removal rate of Zn and Ni (5mg l�1) was, respectively, 2.8 and 2.4 times lower in the presence of

HA (20mg l�1) than in the absence of HA, presumably due to the formation of HA–heavy metal complexes which

prevented the removal reactions at the ZVI surface. Chromate removal was not affected. In a column test, two parallel

systems were supplemented with a continuous input of simulated groundwater containing a mixture of the heavy metals

Zn, Ni and Cr(VI) (5mg l�1 each), with or without HA (at 20mg l�1). Initially, the two column systems efficiently

(490%) removed the heavy metals from the simulated groundwater. When the input heavy metal concentration was

increased to 8–10mg l�1, a significant breakthrough of Ni and Zn, up to 80%, occurred in the column system fed with

HA. Chromate and HA did not significantly break through. After 60 weeks, the effect of HA on leaching of the

accumulated metals (approx. 2mg g�1) was investigated. No significant leaching was observed. The results of this study

suggest that the impact of dissolved organic matter on the efficiency and lifetime of a ZVI barrier for in situ removal of

heavy metals should be considered in the design of the barrier.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Fe0; Iron oxide; Humic acid; Groundwater; Zinc; Nickel; Chromate

1. Introduction

Zero-valent iron (ZVI) is considered as a potential

remediation agent for the elimination of numerous

e front matter r 2005 Elsevier Ltd. All rights reserve

atres.2005.06.020

ing author. Tel.: +3214 33 51 79;

5 23.

ess: leen.bastiaens@vito.be (L. Bastiaens).

ress. Laboratory for Soil and Water Manage-

University of Leuven, Kasteelpark Arenberg 20,

Belgium.

heavy metals from contaminated groundwater. The

ZVI technology is very effective for the removal of

heavy metals with different chemical characteristics due

to multiple ZVI–metal interactions such as surface

complexation, reduction, (co)precipitation and cementa-

tion (Shokes and Moller, 1999). The basis for these

reactions is the corrosion of ZVI. The first corrosion

product is amorphous ferrous hydroxide, which is

predicted thermodynamically to convert to magnetite

(Fe3O4) (Odziemkowski et al., 1998). Mixed-valent iron

salts known as green rusts may also form. Their

d.

ARTICLE IN PRESSJ. Dries et al. / Water Research 39 (2005) 3531–35403532

subsequent oxidation can lead to the formation of

magnetite, maghemite, goethite and lepidocrocite (Roh

et al., 2000). As a result of these reactions, the ZVI

surface is coated by a layer of iron oxides and

oxyhydroxides, similar to natural oxide solid phases.

In the presence of water, the surfaces of iron oxides are

generally covered with surface hydroxyl groups (Stumm

and Morgan, 1996).

The three metals of interest in the present study, zinc

(Zn(II)), hexavalent chromium (Cr(VI)) and nickel

(Ni(II)), can be removed by ZVI through different

mechanisms. Bivalent Zn2+ can undergo several surface

complexation or adsorption reactions with the func-

tional groups at the ZVI surface. Zinc can form

unidentate or bidentate surface complexes. Metal–ligand

complexes such as hydrolyzed Zn (ZnOH+) can also be

adsorbed (Cornell and Schwertmann, 1996; Stumm and

Morgan, 1996).

Hexavalent chromium (Cr(VI), e.g., as chromate

CrO2�4 ) undergoes a reductive reaction in ZVI (Fe0)

systems where chromium serves as the oxidizing agent

(Eq. (1)) (Powell et al., 1995; Alowitz and Scherer, 2002):

Fe0 þ CrO42� þ 4H2O ! Fe3þ þ Cr3þ þ 8OH�: (1)

The removal of the reduced chromium species Cr(III)

can occur through precipitation of the sparingly soluble

Cr(OH)3 or precipitation of mixed iron (III)–chromiu-

m(III) (oxy)hydroxide solids (Blowes et al., 1997; Pratt

et al., 1997). Dissolved aqueous chromium concentra-

tions in equilibrium with these solid phases are below

10mg l�1 (Puls et al., 1999).

Nickel participates in a cementation reaction in ZVI

systems. Cementation is a spontaneous electrochemical

process that involves the reduction of a more electro-

positive (noble) species such as Ni2+ by more electro-

negative (sacrificial) metals (e.g. Fe0) (Khudenko and

Gould, 1991). The product of the cementation reaction

is the noble species in a metallic form (Ni0).

The chemistry of metal ions in natural waters strongly

depends on the formation of complexes with inorganic

and organic ligands. Such complexes reduce the

concentration of the free metal species in solution and

affect the solubility, toxicity and mobility of the metal

(Snoeyink and Jenkins, 1980; Stevenson, 1994). Ground-

waters may contain relatively elevated concentrations of

natural organic polyelectrolytic ligands such as humic or

fulvic acids, which interact not only with the metal ions

in solution but also with oxide surfaces (Stevenson,

1994). At environmentally relevant pH values, which are

below the point of zero charge of the oxide, humic acids

(HA) and iron oxides have opposite charges (Vermeer

et al., 1999). Electrostatic attraction results in sorption

of the polyelectrolyte to the mineral, described as a

ligand exchange mechanism at the oxide surface (Gu

et al., 1994). The binding strength is considerable on

account of the multiple sites present on the HA and the

iron oxide.

Davis and Bhatnagar (1995) proposed three possible

scenarios in aqueous systems containing both HA and

heavy metals, in contact with an iron oxide. First, the

binding of HA to the oxide surface can block the

binding sites and compete with metal removal. Alter-

natively, complexes between metals and HA may remain

in solution and prevent the metal from adsorbing to the

oxide. Finally, metal adsorption can be enhanced by

the formation of ternary HA–metal surface complexes.

The enhanced metal binding can result from the

adsorption of metal–HA complexes, or through metal

complexation by the sorbed HA. The overall effect from

addition of HA on metal binding depends on the

stability of the numerous interactions involved.

The effect of HA on the behavior of heavy metals in

the presence of metal oxides has been investigated in

several studies (e.g., Davis and Bhatnagar, 1995;

Vermeer et al., 1999). Based on the considerations

above, it can be expected that HA also influence the

removal of metals by ZVI. However, few research

papers report on the impact of HA on the performance

of ZVI systems and, to the best of our knowledge, these

studies only considered the removal of organic con-

taminants (Klausen et al., 2001; Tratnyek et al., 2001;

Klausen et al., 2003; Dries et al., 2004). The objectives of

the present study were to investigate the effects of a

standardized commercial mixture of HA salts, i.c.

Aldrich HA, on the removal of zinc, nickel and

chromate in ZVI systems, both in short-term batch

kinetic experiments, and in a longer-term continuous

flow column test.

2. Materials and methods

2.1. Media

All experiments were performed with a dilute simu-

lated groundwater, prepared as follows. MilliQ water

supplied with 0.5mM CaCl2 and MgCl2, and humic

acids (HA, 20mg l�1, if applicable) was first deoxyge-

nated by flushing with nitrogen gas. In an anaerobic

glove box (Coy Laboratory Products; gas composition:

N2/CO2/H2:87.5/7.5/5), an anoxic bicarbonate solution

was added to obtain a final concentration of 0.5mM

each of NaHCO3 and KHCO3. The pH was set at 7.0.

The heavy metals (5mg l�1, if applicable) were added in

the final step, also in the glove box. The HA used were

commercially available sodium salts of HA, supplied by

Aldrich, and were used without pretreatment. The

content of organic carbon, Ni, Zn and Cr was 46%,

19, 12 and 13mgkg�1, respectively. The total acidity of

purified HA is 5.0mmol g�1 of which 3.7mmol g�1 are

carboxylic groups, and 1.3mmol g�1 phenolic groups

ARTICLE IN PRESSJ. Dries et al. / Water Research 39 (2005) 3531–3540 3533

(Vermeer and Koopal, 1998). The ZVI was cast iron

supplied by Gotthart Maier Metallpulver (Germany).

The iron filings had a size ranging from 0.25 to 2mm

and a specific surface area of 0.74570.007m2 g�1 (as

determined by N2 single point BET analysis).

2.2. Batch tests

A series of batch experiments were performed to

investigate the fate and effect of HA (at 20mg l�1) on the

removal of heavy metals (at 5mg l�1) by ZVI. The setup

of a typical test consisted of a number of 38ml glass

serum flasks completely filled with deoxygenated simu-

lated groundwater. The amount of ZVI in the flasks was

chosen to obtain useful data for kinetic analysis over the

experimental duration (7 days). The filled flasks were

crimp sealed with Teflon-lined rubber septa, and were

incubated at room temperature on a rotary shaker (at

10 rpm). After 1, 2, 4 and 7 days, selected flasks were

removed from the shaker, and sacrificed for analysis.

A pseudo first-order model (Eq. (2)) was applied to

describe the kinetics of heavy metal removal by ZVI:

C ¼ C0 e�kt, (2)

where C is the concentration at any time (mg l�1), C0 is

the initial concentration, k is the first-order rate constant

(d�1), and t is the reaction time. The natural logarithmic

transformation of Eq. (2) yields a linear equation with

the first-order rate constant k as slope. Values of k were

estimated by linear regression of the transformed data

versus time, using MS Excel’s data analysis tools

(Microsoft Corporation, Redmond, WA). The first-

order rate constant was normalized to a total ZVI

specific surface area of 1m2ml�1 to allow comparison

between rate constants determined with different ZVI

doses and surface areas.

2.3. Continuous flow column test

Two parallel continuous column systems were set up,

treating a continuous input of simulated groundwater

contaminated by a mixture of the heavy metals Zn, Ni

and Cr(VI). The first column system was the reference

Table 1

Operational periods in the continuous flow column test

Period (weeks) Nominal input metal concentration (mg l�1)

Zn Ni Cr(VI)

1 (0–27) 5 5 5

2 (28–60) 8 8 10

3 (61–92) 0 0 0

C: column, +or �: presence or absence of HA in the influent, respec

system fed with simulated groundwater devoid of HAs,

whereas the second column system was continuously

supplied with groundwater containing HA (at

20mg l�1). Deoxygenated simulated groundwater con-

taining heavy metals was pumped with a peristaltic

pump (Pharmacia Biotech pump P-1) into a 10ml glass

mixing vessel. The medium was then pumped upwards

through the columns. Each column system consisted of

three parallel glass columns (i.d.: 2.4 cm, height: 10 cm,

active volume: 37.8ml): one control column (labeled C1

for the reference system, and C4 for the system with

HA), and duplicate ZVI columns (labeled C2 and C3 for

the reference system, and C5 and C6 for the system with

HA). The control columns did not contain any filling

material in order to determine losses of metals and/or

HA to column material and tubing. The ZVI columns

were completely filled with approx. 10272 g of ZVI

filings in an anaerobic glove box. The resulting porosity

was about 5972% (determined gravimetrically). The

flow velocity in the ZVI columns was approx. 3.470.6

pore volumes per day, corresponding to a hydraulic

retention time of 7.171.2 h.

Table 1 descibes the three operational periods of the

test. During the first experimental period, the influent

heavy metal concentration was 5mg l�1 for each of the

metals. After 27 weeks, the influent metal concentration

was increased to 8mg l�1 (for Zn and Ni) and 10mg l�1

(for Cr(VI)). After 60 weeks, heavy metals were no

longer added to the influent to test the effect of HA on

the potential elution of the bound heavy metals. During

the elution phase (third period), one ZVI column, C3

and C6, respectively, out of each column system was

switched to the opposite regime.

Samples were taken with syringes from the mixing

vessel and the effluent of the columns at regular time

intervals for the determination of pH and concentra-

tions of heavy metals and HA. The flow rates were

routinely checked.

2.4. Reactivity and analysis of reacted ZVI

Column C2 (without HA) and column C5 (with HA)

were dismantled in an anaerobic glove box at the end of

Input of HA (20mg l�1)

Control columns ZVI columns

C1 C4 C2 C3 C5 C6

� + � � + +

� + � � + +

� + � + + �

tively.

ARTICLE IN PRESSJ. Dries et al. / Water Research 39 (2005) 3531–35403534

the third operational period. The reacted ZVI filings

originating from the lower and upper halves of each

column were collected separately, freeze dried, and

stored in closed serum vials under a N2 headspace. The

total metal content of the recovered ZVI filings was

determined in triplicate after acid digestion and extrac-

tion under high temperature using microwave destruc-

tion with HCl/HNO3/HF/H3BO3. The reacted ZVI

samples were ground by hand with a pestle and mortar

to remove the oxide coating. The oxide coating was then

dissolved for 1 h in dilute HCl (0.5M) for the

determination of the ferrous iron content.

A series of batch experiments were set up to

investigate the removal of Zn, Ni or Cr(VI) (at 5mg l�1)

by reacted ZVI in comparison with fresh ZVI. The setup

of each test consisted of a series of glass serum flasks as

described above (Section 2.2). The different experimen-

tal sets in each test were as follows. One set without ZVI

acted as control. The other sets contained equal

amounts of fresh ZVI, fresh freeze-dried ZVI, or reacted

freeze-dried ZVI from the lower or upper column halves

of the two dismantled columns. In the experiments with

Zn and Ni, the ZVI dose was 6 g l�1. In the test with

Cr(VI), the ZVI dose was 2 g l�1. Analyses of different

species were done as described below.

2.5. Resin equilibration method

The degree of metal–humate complexation for Zn and

Ni (at 5mg l�1) was estimated by the resin equilibration

approach, as described by Christensen and Christensen

(1999). The effect of complexation (EC) was determined

by comparing the equilibrium metal distribution in

simulated groundwater with and without HA, in contact

with a cation-exchange resin (Eq. (3)):

EC ¼ðM þMHAÞ

M¼

K�HA

KþHA, (3)

where M and MHA are the free and complexed aqueous

metal concentrations, respectively, corresponding to a

resin-sorbed concentration. K�HA and K+HA are the

resin–aqueous phase distribution ratios without and

with HA present, respectively. A large EC indicates that

a large fraction of the metal in solution is complexed by

HA. The resin used was Dowexs 50W-X4 (polystyrene

with sulfonic acid groups, 20–50 mesh, 5.2 equiv. kg�1

nominal exchange capacity, H+ form), purchased from

Bio-Rad. The resin was first converted to the Na+ form

with concentrated NaOH, and subsequently washed

several times with simulated groundwater. The con-

verted resins were dried at room temperature. Prelimin-

ary experiments showed that equilibrium concentrations

of zinc and nickel were reached within 3 days, and that

the HA did not interact with the resin. Duplicate 38ml

glass serum flasks were filled with simulated ground-

water with or without HA (at 20mg l�1), leaving no

headspace. The serum flasks contained 25mg of the

converted resin. The crimp-sealed reaction flasks were

incubated at room temperature on a rotary shaker (at

10 rpm) for 5 days before sampling of the aqueous

phase.

2.6. Analyses

Samples for HA determination were filtered (0.45 mmMillipore membrane filter). Samples for metal analysis

were filtered (0.45 mm) and acidified (5 ml ml�1 65%

HNO3 Suprapur from Merck). Acidified samples

containing HA were filtered a second time to

remove precipitated HA. The pH was measured with a

Hamilton glass combination electrode and a WTW pH

325 pH meter. The total concentration of the heavy

metals was determined with a Perkin–Elmer optima

3000 DV ICP emission spectrometer. The concentration

of HA was determined with a UV/visible spectro-

photometer Ultro Spec 3000, at a wavelength of

254 nm (Stevenson, 1994). In solutions containing

chromate, the applied wavelength was 230 nm to avoid

interference with the spectrum of chromate. Ferrous

iron was quantified colorimetrically with cell test

1.14896.0001 from Merck. The morphology and ele-

mental composition of the fresh and reacted ZVI filings

were studied by means of scanning electron microscopy

(SEM, Jeol JSM-6340F) combined with energy disper-

sive spectrometry for the analysis of the characteristic X-

rays (EDS or EDXA, Princeton Gamma Tech Imix-PC

with ultra-thin window detector). Surfaces as well

as cross-sections were investigated. For observation

of the surfaces, ZVI particles were mounted on

SEM stubs using double-sided carbon tape. No further

sample preparation was required. Cross-section

samples were obtained by embedding the particles

in an electrically conducting resin and by subsequent

grinding and polishing. In case of platelet-like

particles, care was taken to embed the particles with

their basal plane perpendicular to the plane of polishing.

SEM observations of the surfaces were performed at

5 keV electron beam energy using the lower electron

detector of the SEM, while observations of the cross-

sections were performed at 20 keV beam energy using

the lower electron detector or the backscattered electron

detector (Centaurus). All EDXA work was done at

20 keV.

3. Results

3.1. Effect of HA on metal removal in batch systems

Fig. 1 shows the fate of HA in batch systems with

increasing ZVI doses. Increasing amounts of ZVI

resulted in a faster and more extended removal of the

ARTICLE IN PRESSJ. Dries et al. / Water Research 39 (2005) 3531–3540 3535

HA from the aqueous phase. At a ZVI dose of 2 or

3 g l�1, there was still approx. 90% of the HA in solution

after 7 days. At a ZVI loading of 6 g l�1, about 50% of

the HA were removed (Fig. 1). Table 2 shows that the

presence of HA negatively affected both the extent

(described by the cumulated amount of heavy metals q)

and the kinetics (described by the rate constant kN) of

Zn and Ni removal by ZVI in the batch experiments.

Chromate removal on the other hand was not signifi-

cantly impacted. The experimental metal removal

data were described quite well (R2X0:95) by a first-

order rate expression (Eq. (2)). The surface area

normalized rate constant kN was highest for Cr(VI),

and lowest for Zn. The normalized half-life t1/2-N of Ni

and Zn increased by a factor of 2.4 and 2.8, respectively,

in the presence of HA (Table 2). Using the resin

equilibration method, we estimated that about 10%

and 50%, respectively, of Ni and Zn were complexed by

the HA under experimental conditions comparable to

the batch ZVI experiments (e.g., ionic strength, pH and

medium composition).

0

0.8

0.6

0.4

0.2

1

0 14t (d)

HA

C/C

0 0

2

3

6

10

ZVI dose

(g l-1):

7

Fig. 1. Removal of HA by increasing ZVI doses in batch

systems.

Table 2

Removal of Zn, Ni and Cr(VI) by ZVI in batch experiments in the pr

(value7SD) for the effect of complexation (EC) and accumulated am

amount of ZVI

Metal

(5mg l�1)

HA

(20mg l�1)

ZVI dose

(g l�1)

R2 kNa(d�1)

Zn(II) � 6 0.98 (n ¼ 5) 6675

+ 6 0.95 (n ¼ 5) 2373

Ni(II) � 3 0.99 (n ¼ 7) 11776

+ 3 0.99 (n ¼ 7) 4872

Cr(VI) � 2 0.99 (n ¼ 3) 364734

+ 2 0.99 (n ¼ 3) 313729

aFirst-order rate constant kN and half-life t1/2-N norma1ized to a sbEstimated from data in Fig. 1.

3.2. Effect of HA on metal removal in continuous flow

column systems

Fig. 2 shows the performance of the column systems

during the first two operational periods (with input of

heavy metals). The metals Zn, Ni and Cr(VI) were very

efficiently (490%) removed in the ZVI columns during

the first operational period (p27 weeks) of the experi-

ment. The effluent concentrations of Zn and Ni slightly

increased, while Cr(VI) was almost never detected in

the effluent (Fig. 2). The influent heavy metal concen-

trations were increased after 27 weeks of operation

(Table 1). Chromate removal generally remained un-

affected, although Cr was sometimes detected in the

effluent of column C5 especially towards the end of the

second period (Fig. 2(c)). The concentration of Zn and

Ni increased significantly in the effluent of columns fed

with HA and reached about 60–80% of the influent

concentration (Figs. 2(a) and (b)). In the third opera-

tional period, i.e. the elution phase, the sequestered

metals were not significantly mobilized by simulated

groundwater with or without HA (results not shown).

The removal of HA in the ZVI columns was variable but

high, with an average removal of 93% and 85% during

the first and second operational period, respectively

(Fig. 3(a)). There was no consistent breakthrough of

HA. Interestingly, a significant amount of dissolved iron

was released from the ZVI columns fed with HA

(Fig. 3(b)). Based on mass balances after 60 weeks of

operation, the accumulated amount of removed heavy

metals and HA by ZVI was approx. 2 and 4.5mg g�1,

respectively (results not shown).

3.3. Reactivity and analysis of reacted ZVI

Fig. 4 shows representative SEM micrographs of fresh

and reacted ZVI recovered from the continuous flow

column systems. EDS analysis revealed that the surface

esence or absence of HA: estimated first-order kinetic constants

ount of removed heavy metals (q) and HA (qHA) relative to the

t1/2-Na(min) EC q (mg g�1

ZVI)

qHAb(mg g�1

ZVI)

1571 — 0.83 —

4375 2.0 0.69 1.66

8.670.4 — 1.79 —

20.970.8 1.1 1.49 0.70

2.770.2 — 2.79 —

3.270.3 — 2.44 0.86

urface area concentration of 1m2ml�1.

ARTICLE IN PRESS

0

0.8

0.6

0.4

0.2

0

0.8

0.6

0.4

0.2

0

0.8

0.6

0.4

0.2

0 20 40 60

0 20 40 60

0 20 40 60

t (weeks)

t (weeks)

t (weeks)

[Zn]

C/C

i

2nd period1st period

2nd period1st period

2nd period1st period

[Ni]

C/C

i[C

r] C

/Ci

(a) Zn

(b) Ni

(c) Cr(VI)

Fig. 2. Effluent concentrations of the heavy metals Zn (a), Ni

(b) and Cr(VI) (c), relative to the influent concentrations (Ci) of

ZVI columns C2 (+) and C3 (� ) operated without the input of

HA, and of ZVI columns C5 (m) and C6 (n) fed with HA (at

20mg l�1) (Table 1 and the text describe the operational periods

divided by the dashed line).

0

5

10

15

20

25

0 50 100

t (weeks)

[HA

] (m

g l-1

)

2nd period

2nd period

1st 3rd

3rd

0

4

8

12

0 50 100

t (weeks)

[Fe]

(m

g l-1

)

1st

(a) HA

(b) Fe

Fig. 3. Evolution of the concentration of HA (a) and total

dissolved iron (b) in the influent (�), and in the effluent of ZVI

columns C2 (+) and C3 (� ) operated without input of HA,

and ZVI columns C5 (m) and C6 (n) fed with HA. Note that

C3 only received HA in the third operational period, and C6

only in the first and second operational periods (Table 1 and the

text describe the operational periods divided by the dashed

lines).

J. Dries et al. / Water Research 39 (2005) 3531–35403536

of reacted ZVI was covered with a layer which consisted

mainly of Fe and O, indicative of the presence of iron

oxides. The surface layer also contained Ni, Zn and Cr

(results not shown). The concentration of ferrous iron in

the oxide coating from the reacted ZVI samples was

determined after extraction and was about

1.670.9mg g�1 ZVI. Table 3 shows the total metal

content and reactivity of ZVI recovered from the

column systems. Reacted ZVI was considerably enriched

in metals (up to 40 times) in comparison with fresh ZVI

(Table 3). The metal content of reacted ZVI from

column C5 (fed with HA) was distributed more evenly

over the column, than was the case for ZVI recovered

from column C2 (operated without input of HA).

Finally, the concentration of Ni and Zn was, respec-

tively, 1.3 and 1.4 times higher in the bottom half of

column C2 than in the bottom half of column C5, while

the concentration of Cr did not differ significantly.

Freeze drying had no effect on the rate of metal

removal by fresh ZVI in batch systems (results not

shown). The removal rates of Cr(VI) and Zn in batch

systems were, respectively, four and 10 times higher by

reacted than by fresh ZVI (Table 3). The removal

kinetics of Zn and Cr(VI) did not differ significantly

among reacted ZVI samples recovered from the different

column systems or from different locations in each

system. In contrast, the rate of Ni removal was about

equal or slightly lower in reacted versus fresh ZVI

systems. Ni removal was about 20% slower with reacted

ZVI recovered from column C5 fed with HA (Table 3).

4. Discussion

Sorption of metals to oxide surfaces is considered to

be a very fast process with equilibrium times of less than

24–72 h (Smith, 1996). We were not able to describe the

sorption of Zn by common equilibrium models such as

ARTICLE IN PRESS

Table 3

Heavy metal content (value7SD) of fresh and reacted ZVI recovered from the bottom and top of columns C2 (without HA) and C5

(with HA), and surface area normalized first-order kinetic constants for reacted ZVI expressed relative to the rate constant for fresh

ZVI (value7SD)

ZVI Metal content (mg g�1) Metal removal kinetics kN/kN,fresha

Zn Ni Cr Zn Ni Cr(VI)

Fresh 0.0670.03 0.5970.09 1.370.4 170.05 170.2 170.05

Bottom C2 2.470.1 3.070.2 3.970.5 972 0.970.1 4.170.9

Top C2 0.370.03 1.170.2 1.770.2 1173 0.970.1 3.970.4

Bottom C5 1.770.2 2.370.6 3.670.9 1073 0.7770.02 3.770.2

Top C5 1.470.2 1.870.2 2.470.2 1074 0.8170.03 3.670.6

akN,fresh: first-order rate constant for fresh ZVI (d�1) normalized to a specific surface area of 1m2ml�1. Zn: 3671; Ni: 130728;

Cr(VI): 330716.

Fig. 4. Representative SEM micrographs of the cross-section ((a) and (c)) and surface ((b) and (d) of fresh ((a) and (b)) and reacted ((c)

and (d)) ZVI recovered from the column test: EDS analysis showed that the light colored region on the cross-section of fresh particles

(a) consisted mainly of Fe0 and the embedded darker-colored nodules and flakes consisted of carbon. EDS analysis of the cross-section

of reacted ZVI (c) revealed the presence of a coating of iron oxides.

J. Dries et al. / Water Research 39 (2005) 3531–3540 3537

the Langmuir or Freundlich expressions (results not

shown), but a first-order rate model described the heavy

metal removal quite well (Table 2). This is apparently

due to the fact that the removal processes in ZVI systems

are kinetic rather than equilibrium phenomena, which

can be explained by the ongoing corrosion of the iron

(Johnson et al., 1996; Su and Puls, 2001; Alowitz and

Scherer, 2002; Miehr et al., 2004). As ZVI corrodes,

ARTICLE IN PRESSJ. Dries et al. / Water Research 39 (2005) 3531–35403538

more ferrous iron precipitates and forms new adsorption

sites for metal binding at the surface.

The presence of HA negatively affected the removal of

Zn and Ni by ZVI, both in batch and continuous flow

column systems. Two mechanisms may be responsible

for this effect: (i) formation of metal–humate complexes

remaining in solution and (ii) competition for reactive

surface sites between HA and heavy metals (Davis and

Bhatnagar, 1995).

The results from Fig. 1 indicate that, at ZVI doses

below 10 g l�1, a substantial amount of HA remained in

the aqueous phase where they could form non-reacting

metal–humate complexes (Davis and Bhatnagar, 1995).

The normalized rates of Ni and Zn removal in the batch

experiments decreased, respectively, by a factor of 2.4

and 2.8 in the presence of HA (Table 2), indicating that

HA had a more pronounced effect on Zn removal than

on Ni removal. EC was also about 1.8 times higher for

Zn than for Ni, indicating that more Zn than Ni was

complexed by HA. These findings are consistent with the

general observation that HA have a higher affinity for

Zn than for Ni (Stumm and Morgan, 1996; Milne et al.,

2003). The alternative explanation, i.e. competition for

surface sites between metals and HA, seems unlikely,

especially in the batch test with Ni (at a ZVI dose of

3 g l�1) where only a small portion (about 10%) of the

HA attached to the surface (Fig. 1).

In the column system fed with HA, a gradual

breakthrough of Ni and Zn was observed, but not of

HA (Figs. 2 and 3). The observed release of dissolved

iron can be explained by a ligand-promoted iron

dissolution mechanism (Stumm, 1997). There was no

correlation between the effluent concentrations of HA

and heavy metals (results not shown). Such a correlation

would be expected in the dissolved metal–humate

complexation scenario. The heavy metal breakthrough

in the columns fed with HA could thus not be attributed

directly to the formation of metal–humate complexes

remaining in solution, as was the case in the batch study.

However, the formation of dissolved metal–humate

complexes may have delayed the removal of the heavy

metals by ZVI in the column system receiving HA in

comparison with the reference system (without input of

HA). The delayed removal may indirectly have led to the

observed heavy metal breakthrough. Several lines of

evidence support this hypothesis. First, chromate, which

is not expected to interact with the negatively charged

HA, was removed just as well in both column systems

(Fig. 2(c)). This suggests that the oxide-covered surface

of ZVI was accessible for reaction, and was not

completely blocked by the sorbed HA. Second, a dark

brown front was gradually moving up in the columns

receiving HA. The observed brown color resembled the

color of the HA powder used in the present study, and

differed significantly from the typical black, gray or

rusty color of the different iron oxides (Odziemkowski et

al., 1998; Roh et al., 2000). This observation indicates

that the HA did migrate along the columns before they

were removed, potentially carrying the complexed

fraction of Zn and Ni along. Third, in batch experiments

set up with reacted ZVI recovered from the columns, Zn

was removed equally well by reacted ZVI collected from

both column systems, while Ni removal was only 20%

slower in reacted ZVI systems originating from the HA-

loaded columns (Table 3). The presence of HA on the

surface of reacted ZVI (approx. 4.5mg g�1) apparently

did not interfere with the sorption of Zn, and only

slightly with the reduction of Ni. The heavy metal

distribution in the ZVI columns (Table 3) supports the

first two arguments cited above. The top (effluent) part

of column C5 (fed with HA) was enriched 1.5 and 4.5

times in Ni and Zn, respectively, while the bottom half

contained 24% and 30% less Ni and Zn, respectively, in

comparison with column C2 (operated without input of

HA). These observations agree with the hypothesis that

both metals were transported through the column by the

HA. The distribution of Cr on the other hand was quite

similar in the two columns, because chromate does not

bind to the HA in solution.

In the continuous flow column test, no heavy metal

breakthrough was apparent after a passage of about

1500 pore volumes of simulated groundwater without

HA (Figs. 2(a)–(c)). The results of the column test,

therefore, support the observation of other researchers

that ZVI is a promising technology for the removal of

mixed heavy metals (Shokes and Moller, 1999). More-

over, the ‘bound’ metals were not mobilized again by

simulated groundwater with or without HA, irrespective

of the column history.

In batch experiments set up with ZVI recovered from

the columns, the reacted ZVI was significantly more

reactive towards Zn and Cr(VI) than fresh ZVI, while

removal of Ni was somewhat slower (Table 3). Although

care was taken to avoid air contact with the wet ZVI

particles, oxidation of the iron oxide layer on reacted

ZVI during freeze drying and storage could not be

excluded, and may have influenced the heavy metal

removal kinetics. The reacted ZVI was recovered from

the column systems at the end of the third operational

period that lasted about 30 weeks during which no

heavy metals were introduced (Table 1). The enhanced

Zn removal might be explained by the presence of a high

number of sorption sites on the oxide surface resulting

from ZVI corrosion. The latter is clearly illustrated by

the thicker oxide coating found on the surface of the

reacted ZVI (Fig. 4). The enhanced removal of Cr(VI)

by reacted ZVI is probably an effect of additional

reaction mechanisms operating in reacted ZVI systems.

Besides reductive precipitation by ZVI metal (Eq. (1)),

which may or may not have been hindered by the oxide

coating, ligand exchange and, more probably, reduction

by surface-bound ferrous iron could have occurred in

ARTICLE IN PRESSJ. Dries et al. / Water Research 39 (2005) 3531–3540 3539

the presence of the iron oxide coating (Cornell and

Schwertmann, 1996; Fendorf and Li, 1996; Buerge and

Hug, 1999). Finally, the thick oxide coating observed on

the surface of reacted ZVI (Fig. 4) probably hindered the

electron transfer necessary for the reduction reaction of

Ni, resulting in lower Ni removal rates in the batch series

with reacted ZVI (Table 3). Nickel sorption to the iron

oxide coating apparently was not such an important

process as was the case for Zn, which is consistent with

the observation that iron oxides have a higher sorption

affinity for Zn than for Ni (Cornell and Schwertmann,

1996; Stumm and Morgan, 1996).

5. Conclusions

Our results indicate that ZVI is a promising reactive

agent for the in situ removal of mixtures of the heavy

metals Zn, Ni and Cr(VI) from contaminated ground-

water. We found, however, that the formation of

metal–humate complexes in solution prevented the

removal of Zn and Ni in batch experiments and delayed

the removal of both metals in a long-term continuous

flow column test. Humic substances bind heavy metals

and influence their mobility, their physical and chemical

characteristics and their fate (Stevenson, 1994). More-

over, HA interact with the iron oxide coating present on

the surface of corroded ZVI, which may affect the

reduction capacity of ZVI (Klausen et al., 2003). The

presence of dissolved organic matter in groundwater

might thus impact the efficiency and lifetime of a

reactive ZVI barrier for in situ removal of heavy metals,

and should therefore be considered in the design of the

permeable barrier. In view of the results presented here,

more research is needed to predict the impact of HA in

order to improve the design of future barriers.

Acknowledgements

The authors thank An Kemps and Josef Trogl for

their assistance with the monitoring of the column test,

and Valere Corthouts, Gust Nuyts and Raymond

Kemps for their support with the metal and SEM

anlyses. This research was funded in part by a Vito

Ph.D. grant, and by EC Project no. QLK3-CT-2000-

00163.

References

Alowitz, M.J., Scherer, M.M., 2002. Kinetics of nitrate, nitrite,

and Cr(VI) reduction by iron metal. Environ. Sci. Technol.

36, 299–306.

Blowes, D.W., Ptacek, C.J., Jambor, J.L., 1997. In-situ

remediation of Cr(VI)-contaminated groundwater using

permeable reactive walls: laboratory studies. Environ. Sci.

Technol. 31, 3348–3357.

Buerge, I.J., Hug, S.J., 1999. Influence of mineral surfaces on

chromium(VI) reduction by iron(II). Environ. Sci. Technol.

33, 4285–4291.

Christensen, J.B., Christensen, T.H., 1999. Complexation of

Cd, Ni, and Zn by DOC in polluted groundwater: a

comparison of approaches using resin exchange, aquifer

material sorption, and computer speciation models

(WHAM and MINTEQA2). Environ. Sci. Technol. 33,

3857–3863.

Cornell, R.M., Schwertmann, U., 1996. In: Bock, B. (Ed.), The

Iron Oxides: Structure, Properties, Reactions, Occurrence

and Uses. VCH Press, Weinheim, Germany.

Davis, A.P., Bhatnagar, V., 1995. Adsorption of cadmium and

humic acid onto hematite. Chemosphere 30, 243–256.

Dries, J., Bastiaens, L., Springael, D., Agathos, S.N., Diels, L.,

2004. Competition for sorption and degradation reactions

in batch zero valent iron systems. Environ. Sci. Technol. 38,

2879–2884.

Fendorf, S.E., Li, G., 1996. Kinetics of chromate reduction by

ferrous iron. Environ. Sci. Technol. 30, 1614–1617.

Gu, B., Schmitt, J., Chen, Z., Liang, L., McCarthy, J.F., 1994.

Adsorption and desorption of natural organic matter on

iron oxide: mechanisms and models. Environ. Sci. Technol.

28, 38–46.

Johnson, T.L., Scherer, M.M., Tratnyek, P.G., 1996. Kinetics

of halogenated organic compound degradation by iron

metal. Environ. Sci. Technol. 30, 2634–2640.

Khudenko, B.M., Gould, J.P., 1991. Specifics of cementation

processes for metals removal. Water Sci. Technol. 24, 235–246.

Klausen, J., Ranke, J., Schwarzenbach, R.P., 2001. Influence of

solution composition and column aging on the reduction of

nitroaromatic compounds by zero-valent iron. Chemo-

sphere 44, 511–517.

Klausen, J., Vikesland, P.J., Kohn, T., Burris, D.R., Ball, W.P.,

Roberts, A.L., 2003. Longevity of granular iron in ground-

water treatment processes: solution composition effects on

reduction of organohalides and nitroaromatic compounds.

Environ. Sci. Technol. 37, 1208–1218.

Miehr, R., Tratnyek, P.G., Bandstra, J.Z., Scherer, M.M.,

Alowitz, M.J., Bylaska, E.J., 2004. Diversity of contaminant

reduction reactions by zerovalent iron: role of the reductate.

Environ. Sci. Technol. 38, 139–147.

Milne, C.J., Kinniburgh, D.G., Van Riemsdijk, W.H., Tipping,

E., 2003. Generic NICA–Donnan model parameters for

metal-ion binding by humic substances. Environ. Sci.

Technol. 37, 958–971.

Odziemkowski, M.S., Schumacher, T.T., Gillham, R.W.,

Reardon, E.J., 1998. Mechanism of oxide film formation

on iron in simulating groundwater solutions: Raman

spectroscopic studies. Corrosion Sci. 40, 371–389.

Powell, R.M., Puls, R.W., Hightower, S.K., Sabatini, D.A.,

1995. Coupled iron corrosion and chromate reduction:

mechanisms for subsurface remediation. Environ. Sci.

Technol. 29, 1913–1922.

Pratt, A.R., Blowes, D.W., Ptacek, C.J., 1997. Products of

chromate reduction on proposed subsurface remediation

material. Environ. Sci. Technol. 31, 2492–2498.

Puls, R.W., Paul, C.J., Powell, R.M., 1999. The application of

in situ permeable reactive (zero-valent iron) barrier technol-

ARTICLE IN PRESSJ. Dries et al. / Water Research 39 (2005) 3531–35403540

ogy for the remediation of chromate contaminated ground-

water: a field test. Appl. Geochem. 14, 989–1000.

Roh, Y., Lee, S.Y., Elless, M.P., 2000. Characterization of

corrosion products in the permeable reactive barriers.

Environ. Geol. 40, 184–194.

Shokes, T.E., Moller, G., 1999. Removal of dissolved heavy

metals from acid rock drainage using iron metal. Environ.

Sci. Technol. 33, 282–287.

Smith, E., 1996. Uptake of heavy metals in batch systems by a

recycled iron-bearing material. Water Res. 30, 2424–2434.

Snoeyink, V.L., Jenkins, D., 1980. Water Chemistry. Wiley,

New York, USA.

Stevenson, F.J., 1994. Humus Chemistry: Genesis, Composi-

tion, Reactions. Wiley, New York, USA.

Stumm, W., 1997. Reactivity at the mineral–water interface:

dissolution and inhibition. Colloids Surfaces A 120,

143–166.

Stumm, W., Morgan, W., 1996. In: Schnoor, J.L., Zehnder, A.

(Eds.), Aquatic Chemistry. Wiley, New York, USA.

Su, C., Puls, R.W., 2001. Arsenate and arsenite removal by

zerovalent iron: kinetics, redox transformation, and im-

plications for in situ groundwater remediation. Environ. Sci.

Technol. 35, 1487–1492.

Tratnyek, P.G., Scherer, M.M., Deng, B., Hu, S., 2001. Effects

of natural organic matter, anthropogenic surfactants, and

model quinines on the reduction of contaminants by zero-

valent iron. Water Res. 35, 4435–4443.

Vermeer, A.W.P., Koopal, L.K., 1998. Adsorption of humic

acids to mineral particles. 2. Polydispersity effects with

polyelectrolyte adsorption. Langmuir 14, 4210–4216.

Vermeer, A.W.P., McCulloch, J.K., Van Riemsdijk, W.H.,

Koopal, L.K., 1999. Metal ion adsorption to complexes of

humic acid and metal oxides: deviations from the additivity

rule. Environ. Sci. Technol. 33, 3892–3897.