Reduction of azo dyes with zero-valent iron

Article abstract of DOI:10.1016/S0043-1354(99)00331-0

The reduction of azo dyes by zero-valent iron metal (Fe⁰) at pH 7.0 in 10 mM HEPES buffer was studied in aqueous, anaerobic batch systems. Orange II was reduced by cleavage of the azo linkage, as evidenced by the production of sulfanilic acid (a substituted aniline). Adsorption of the dyes on iron particles was less than 4% of the initial concentration, and >90% mass balance was achieved by summing aqueous concentrations of dye and product amine. All of the 9 azo dyes tested were reduced with first-order kinetics. The kinetics of decolorization at the λ(max) of each dye were rapid: a typical k(obs) was 0.35 ± 0.01 min^-1 for Orange II at 130 rpm on an orbital shaker, corresponding to a surface area normalized rate constant (k(SA)) of 0.21 ± 0.01 L m^-2 min^-1. The rate of reduction of Crocein Orange G varied with initial dye concentration in a way that suggests saturation of surface sites on the Fe⁰, and varied with the square-root of mixing rate (rpm) in a manner indicative of mass transfer limited kinetics. Correlation analysis using k(obs) for all of the azo dyes, estimates of their diffusion coefficients, and calculated energies of their lowest unoccupied molecular orbitals (E(LUMO)), gave no strong trends that could be used to derive structure-activity relationships. Using an authentic sample of wastewater from a dye manufacturing operation and construction-grade granular Fe⁰, rapid decolorization was achieved that was consistent with reduction of azo dyes. (C) 2000 Elsevier Science Ltd. The reduction of azo dyes by zero-valent iron metal (Fe⁰) at pH 7.0 in 10 mM HEPES buffer was studied in aqueous, anaerobic batch systems. Orange II was reduced by cleavage of the azo linkage, as evidenced by the production of sulfanilic acid (a substituted aniline). Adsorption of the dyes on iron particles was less than 4% of the initial concentration, and >90% mass balance was achieved by summing aqueous concentrations of dye and product amine. All of the 9 azo dyes tested were reduced with first-order kinetics. The kinetics of decolorization at the λ_max of each dye were rapid: a typical k_obs was 0.35±0.01 min^-1 for Orange II at 130 rpm on an orbital shaker, corresponding to a surface area normalized rate constant (k_SA) of 0.21±0.01 L m^-2 min^-1. The rate of reduction of Crocein Orange G varied with initial dye concentration in a way that suggests saturation of surface sites on the Fe⁰, and varied with the square-root of mixing rate (rpm) in a manner indicative of mass transfer limited kinetics. Correlation analysis using k_obs for all of the azo dyes, estimates of their diffusion coefficients, and calculated energies of their lowest unoccupied molecular orbitals (E_LUMO), gave no strong trends that could be used to derive structure-activity relationships. Using an authentic sample of wastewater from a dye manufacturing operation and construction-grade granular Fe⁰, rapid decolorization was achieved that was consistent with reduction of azo dyes.

Full text of DOI:10.1016/S0043-1354(99)00331-0

Wat. Res. Vol. 34, No. 6, pp. 1837±1845, 2000
# 2000 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: S0043-1354(99)00331-0
0043-1354/00/$ - see front matter
www.elsevier.com/locate/watres
REDUCTION OF AZO DYES WITH ZERO-VALENT IRON
SANGKIL NAM* and PAUL G. TRATNYEK{M
Department of Environmental Science and Engineering, Oregon Graduate Institute of Science and
Technology, P.O. Box 91000, Portland, OR 97291-1000, USA
(First received 1 February 1999; accepted in revised form 1 July 1999)
AbstractÐThe reduction of azo dyes by zero-valent iron metal (Fe0) at pH 7.0 in 10 mM HEPES
bu€er was studied in aqueous, anaerobic batch systems. Orange II was reduced by cleavage of the azo
linkage, as evidenced by the production of sulfanilic acid (a substituted aniline). Adsorption of the dyes
on iron particles was less than 4% of the initial concentration, and >90% mass balance was achieved
by summing aqueous concentrations of dye and product amine. All of the 9 azo dyes tested were
reduced with ®rst-order kinetics. The kinetics of decolorization at the lmax of each dye were rapid: a
1
typical kobs was 0.3520.01 min for Orange II at 130 rpm on an orbital shaker, corresponding to a
surface area normalized rate constant (kSA) of 0.21 2 0.01 L m 2 min 1. The rate of reduction of
Crocein Orange G varied with initial dye concentration in a way that suggests saturation of surface
sites on the Fe0, and varied with the square-root of mixing rate (rpm) in a manner indicative of mass
transfer limited kinetics. Correlation analysis using kobs for all of the azo dyes, estimates of their
di€usion coecients, and calculated energies of their lowest unoccupied molecular orbitals (ELUMO),
gave no strong trends that could be used to derive structure-activity relationships. Using an authentic
sample of wastewater from a dye manufacturing operation and construction-grade granular Fe0, rapid
decolorization was achieved that was consistent with reduction of azo dyes. # 2000 Elsevier Science
Ltd. All rights reserved
Key wordsÐiron metal, decolorization, kinetics, mass transport, correlation analysis
NOMENCLATURE
INTRODUCTION
1
ra
iron surface area concentration (m2 L
)
The decolorization of azo dyes is one of many use-
ful reduction reactions that can be e€ected by zero-
valent iron (Fe0). In the past, most investigations of
the environmental applications of reduction by Fe0
have focused on remediation of groundwater con-
taminated with chlorinated solvents (e.g. Gillham
and O'Hannesin, 1994; Matheson and Tratnyek,
1994), although there is now a substantial body of
literature on the potential for using Fe0 to treat ma-
terials contaminated with chlorinated aromatic
compounds (Chuang et al., 1995; Grittini et al.,
1995), nitro aromatic compounds (Agrawal and
Tratnyek, 1996; Burris et al., 1996; Singh et al.,
1998a), nitrate (Cheng et al., 1997; Chew and
Zhang, 1998; Huang et al., 1998), and various
metals (Cantrell et al., 1995; Blowes et al., 1997;
Pratt et al., 1997; Fiedor et al., 1998; Gu et al.,
1998). More recently, it has been suggested that Fe0
can be used to initiate remediation of more complex
anthropogenic chemicals by reduction of critical
functional groups. Pesticides that may be subject to
this treatment include alachlor and metolachlor
(Eykholt and Davenport, 1998); DDT, DDD and
DDE (Sayles et al., 1997); and atrazine (Monson et
al., 1998; Singh et al., 1998b). Dyes are another cat-
egory of complex chemicals that may be labile to
lmax
o
A
wavelength of maximum absorbance (nm)
rotation rate (rpm)
empirical coecient describing the kinetics of
1
mass transport (rpm1/2 min
)
C, C0
concentration of substrate at time t, and at
time 0 (mM)
1
D
ELUMO
di€usion coecient (cm2 s
)
energy of the lowest unoccupied molecular
orbital (eV)
pseudo ®rst order disappearance rate constant
kobs
1
(min
®rst order appearance rate constant (min )
)
1
kobs
kSA
'
surface-area
normalized
rate
constant
1
(L m 2 min
®rst order rate constant for chemical reaction
)
krxn
1
(min
)
concentration of substrate at half of maximum
K1/2
Vm
reaction rate (mM)
maximum reaction rate (mM min
1
)
*Present address: H. Lee Mott Cancer Center and
Research Institute, 12902 Magnolia Drive, Tampa, FL
33612, USA.
{Author to whom all correspondence should be addressed.
Fax: +1-503-690-1273; e-mail: tratnyek@ese.ogi.edu
1837

1838
Sangkil Nam and Paul G. Tratnyek
reduction by Fe0, as has recently been shown for
several azo dyes (Cao et al., 1999).
Puls, 1999). The sample of dye wastewater was e‚uent
from a manufacturing facility in South Carolina, and was
known to contain disperse azo dyes.
The reduction of azo dyes by Fe0 is of interest,
not only for its potential application in the decolori-
zation of wastewaters from dye use and manufac-
turing, but also because it provides a convenient
model system for investigating (i) the abiotic re-
duction of azo linkages and (ii) the kinetics of fast
reactions with granular iron metal. The abiotic re-
Batch experiments
All experiments were done in pH 7.0, 10 mM HEPES
[N-(2-hydroxyethyl)-piperazine-N'-(2-ethanesulfonic acid)]
bu€er. The bu€er was prepared with deionized and deoxy-
genated water. All individual dye reduction experiments
were done in 7 mL scintillation vials containing 1.000 2
0.002 g Fe0, resulting in an iron surface area concentration
duction of azo linkages in environmental media has (ra) equal to 1.42 m2 L 1. To decolorize the wastewater
sample, more of the construction grade iron was used:
been inferred from experiments done with anaerobic
2.5 g of iron from Master Builders and 2.0 g of the ma-
sediments (Weber and Wolfe, 1987; Yen et al.,
1991; Weber and Adams, 1995), but this system is
terial from Peerless. In an anaerobic glove box (H2/N2),
stock solutions of dye and HEPES bu€er were combined
with the Fe0, and the vials were sealed with Te¯on-lined
caps and Para®lm1. In all cases, the ®nal reaction volume
was 5 mL. The resulting initial concentrations were 10 mM
HEPES, and 60 mM, 300 mM, 1 mM, or 3 mM dye.
During the reaction, the vials were mixed on an orbital
shaker (New Brunswick Scienti®c) at room temperature.
so complex that it is dicult to fully characterize
the processes involved (Smolen et al., 1999). The
kinetics of fast reactions with granular iron in batch
experiments appear to be in¯uenced by mass trans-
port of the oxidant to the metal surface (Agrawal
and Tratnyek, 1996), although the signi®cance of The aqueous phase was sampled periodically and analyzed
by spectrophotometry or by high-performance liquid chro-
matography (HPLC).
this behavior under environmental conditions is still
not entirely clear (Scherer et al., 1997, 1999). To
advance our understanding of the reduction of azo
dyes in the environment and the reduction of en-
vironmental contaminants by iron metal, we have
investigated the reaction of 9 azo dyes with granu-
lar Fe0 in batch systems.
Analytical methods
The reduction of Orange II and appearance of reaction
products were determined by HPLC using a C-18 reverse
phase column (0.46 i.d. Â 25 cm, Separation Group,
Hesperia, CA) and detection of absorbance at 254 nm.
1
The mobile phase gradient was 1 mL min of 100 mM,
pH 7.0 phosphate bu€er for 5 min, followed by a 5 min
ramp to 100% methanol and 10 min at 100% methanol.
Under these conditions, retention times were 28.2 min for
Orange II and 3.8 min for sulfanilic acid.
Routine analysis of dye decolorization was done by
measuring absorbance at the lmax for each dye (Table 1)
using a UV-visible spectrophotometer (Model UV-265,
Shimadzu, Kyoto). Samples were diluted 2 to 10 fold with
deionized water so that readings never exceeded 1 AU.
For the experiments with textile wastewater, decolorization
was determined by taking full scans of absorbance from
350 to 700 nm, vs a blank consisting of deionized water.
MATERIALS AND METHODS
Chemicals
The food dyes Allura Red, Tartrazine, and Sunset
Yellow FCF were obtained from Warner Jenkinson, St.
Louis, MO. Orange I was purchased from TCI America,
Portland, OR. Amaranth, Naphthol Blue Black, Crocein
Orange G, Orange II, and Acid Blue 113 were obtained
from Aldrich, Milwaukee, WI. Properties of the dyes, and
their structures, are summarized in Table 1 and Fig. 1. All
dyes were purchased in the highest purity that was com-
mercially available and used as received without further
puri®cation. Reagent grade, granular iron metal was
obtained from Fluka (>99.9%, Cat. No. 44905), sieved to
obtain the 16±32 mesh size fraction, and then used with-
RESULTS AND DISCUSSION
out any further treatment. The speci®c surface area of this
iron was determined to be 0.0071 m2 g
Pathway and kinetics
1
by BET gas
The reduction of azo dyes by Fe0 is illustrated in
Fig. 2 for Orange II. These data, obtained by
HPLC, show that degradation of Orange II pro-
duces stoichiometric amounts of sulfanilic acid. The
appearance of this product is consistent with reduc-
adsorption with Krypton (cf. previously reported values
summarized by Johnson et al., 1996). Decolorization of
the authentic dye wastewater was done with construction
grade, granular iron from Master Builders (Beachwood,
OH) and Peerless Metal Powders and Abrasives (Detroit,
MI). Typical speci®c surfaces areas for these metals have
been reported previously (Johnson et al., 1996; Su and tive cleavage of the azo group as shown in Eq. (1).
Table 1. Dyes and their properties
No.
Name
Synonym
C.I. No.
CAS No.
M.W.
lmax
1
2
3
4
5
6
7
8
9
Acid Blue 113
Allura Red
Amaranth
26360
16035
16185
15970
20470
14600
15510
15985
19140
3351-05-1
25956-17-6
915-67-3
1934-20-9
1064-48-8
523-44-4
633-96-5
2783-94-0
1934-21-0
637.68
452.45
538.52
328.34
574.54
328.34
328.34
408.40
468.41
566
500
521
484
618
476
484
482
424
FD&C Red #40
Acid Red 27
Acid Orange 12
Acid Black 1
Crocein Orange G
Naphthol Blue Black
Orange I
Orange II
Sunset Yellow FCF
Tartrazine
Acid Orange 7
FD&C Yellow #6
FD&C Yellow #5

Reduction of azo dyes
1839
ꢀ1†
Although this process presumably occurs with the to 1-amino-2-naphthol, it may have been dicult to
hydrazo (±N(H)±N(H)±) derivative as an intermedi- obtain a quantitative mass balance with respect to
ate, and produces 1-amino-2-naphthol as a product, this product due to its autoxidation (Kudlich et al.,
no standards were available for these compounds, 1999). Similar results were expected for all the azo
so they were not matched with product peaks on dyes, although the only additional product studies
the chromatograms. Even if a peak were assigned that were performed involved qualitative veri®ca-
Fig. 1. Structures of the nine dyes used in this study.

1840
Sangkil Nam and Paul G. Tratnyek
tion that sulfanilic acid was formed from the re- At the end of the experiment, extraction of the Fe0
duction of Orange I, Sunset Yellow FCF and (after removing the aqueous phase) with methanol
Tartrazine.
recovered <4% of the original dye, which appar-
Additional evidence that aromatic azo com- ently was adsorbed but not reduced. Previous stu-
pounds are reduced by Fe0 to give amines is avail- dies with chlorinated ethenes have found that
able from two previous studies: Weber identi®ed adsorption to nonreactive sites on Fe0 can in¯uence
the products formed from immobilized 4-aminoazo- the overall disappearance kinetics of substrate
benzene using HPLC (Weber, 1996) and Cao et al. (Burris et al., 1995; Allen-King et al., 1997), but
found shifts in UV absorbance spectra for Orange there is no indication of this complication in the
II that are consistent with Eq. (1) (Cao et al., data shown in Fig. 2. Qualitatively similar kinetics
1999). Reduction of Orange II by Eq. (1) can also for azo reduction by Fe0 were obtained by Weber
be e€ected by other reductants, as has recently been for 4-aminoazobenzene and its four-electron re-
shown with potassium borohydride (Laszlo, 1997). duction product aniline (Weber, 1996).
In contrast, Fe0 in the presence of H2O2 produces a
Fenton-like reaction that degrades azo dyes largely dye (at an appropriate lmax) was monitored for
by oxidation (Tang and Chen, 1996). most experiments in this study. In these exper-
In contrast to the above, only absorbance of the
Figure 2 also illustrates the kinetics of azo dye re- iments, kobs was determined from plots of the natu-
duction by Fe0. The disappearance of Orange II ral logarithm of absorbance vs time and linear
and appearance of sulfanilic acid ®t the ®rst order regression of these data to the integrated form of
kinetic models
Eq. (2). The results, summarized in Table 2, re¯ect
the e€ects of dye structure, initial concentration,
and mixing rate. All other experimental variables
were held constant, including the type and amount
of bu€er and iron. Bu€er e€ects are possible in
reactions with Fe0 (Schreier and Reinhard, 1995;
Johnson et al., 1998; Zawaideh and Zhang, 1998),
kobs
t
C . C0e
ꢀ2†
ꢀ3†
ko0bs
t
C . C0ꢀ1
e
†,
where C is the concentration of analyte at time t,
C0 is the initial or ®nal concentration, kobs is the
pseudo ®rst order disappearance rate constant and
kobs' is the ®rst order appearance rate constant.
Nonlinear regression using Eq. (2) and the data for
decolorization of Orange II gives kobs=0.35 2
Table 2. Summary of decolorization experiments
1
)
No.
Dye
C0 (mM) o (rpm)
kobs (min
1
2
3
4
5
6
7
8
Acid Blue 113
Allura Red
Allura Red
Allura Red
Allura Red
Amaranth
Amaranth
Amaranth
Amaranth
300
300
300
300
300
300
300
300
300
60
100
100
100
120
140
100
100
120
140
90
100
120
140
164
90
100
100
120
140
160
120
140
160
100
100
124
158
100
100
120
140
100
120
130
140
100
120
140
100
0.06420.002
0.14420.004
0.14120.027
0.15920.007
0.22620.017
0.14920.004
0.13720.003
0.19420.008
0.27920.000
0.17420.012
0.19720.000
0.29120.011
0.35620.010
0.53020.015
0.16720.016
0.20720.006
0.17720.007
0.27620.007
0.38020.007
0.45820.002
0.23920.009
0.30320.006
0.39320.010
0.11920.014
0.10420.006
0.17020.013
0.22520.015
0.02320.002
0.13820.005
0.21820.002
0.27420.013
0.16120.003
0.24520.011
0.30620.009
0.38020.011
0.14820.006
0.24120.004
0.30420.018
0.11620.038
1
0.01 min and C0=25723 mM, whereas ®tting Eq.
(3) to the data for formation of sulfanilic acid gives
1
kobs'=0.602 0.03 min and C0=2352 3 mM. The
two-fold di€erence between kobs and kobs' produces
a small excess mass balance at two minutes, but
after 2 min the mass balance is remarkably good.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Crocein Orange G
Naphthol Blue Black
Orange I
60
60
60
60
300
300
300
300
300
300
1000
1000
1000
3000
3000
3000
3000
300
300
300
300
300
300
300
300
300
300
300
300
Orange I
Orange I
Orange II
Orange II
Orange II
Orange II
Fig. 2. Time course for disappearance of the azo dye
Orange II and appearance of the product sulfanilic acid.
Sunset Yellow FCF
Sunset Yellow FCF
Sunset Yellow FCF
Tartrazine
Reaction was with 200 g L (ra=1.42 m2 L 1) of unpre-
1
treated, 16±32 mesh Fluka Fe0 in pH 7.0,10 mM HEPES
bu€er. C0 of the dye was 300 mM, mixing rate was
130 rpm, and the temperature was ambient.

Reduction of azo dyes
1841
but we decided it was more important to control increasing C0, which presumably is due to satur-
pH for this particular study. We also chose not to ation of reactive surface sites.
investigate the e€ect of iron surface area concen-
This site-saturation behavior was ®rst modeled
tration (ra) in this study, because Cao et al. (1999) using an empirical formulation analogous to the
had shown that the rate of Orange II reduction is Michaelis±Menton equation (Johnson et al., 1996,
linearly dependent on ra, and this relationship has 1998), and subsequently described more rigorously
been found for many other compounds that are in terms surface site concentrations, reaction rate
subject to reduction by Fe0, including nitrobenzene constants and complexation constants (Arnold and
(Agrawal and Tratnyek, 1996) and various chlori- Roberts, 1997; Scherer et al., 1998b). Since the two
nated solvents (e.g. Matheson and Tratnyek, 1994; formulations are mathematically equivalent, and the
Johnson et al., 1996). Surface area normalized ®rst former requires fewer explicit assumptions about
order rate constants (kSA) can be estimated from the system, we have applied it to the data obtained
any of the values of kobs reported in this study, by in this study. The model is
dividing by ra (which is 1.42 m2 L for all data in
1
dC
dt
VmC
Table 2). The data in Fig. 2, for example, give
ˆ
,
ꢀ4†
2
1
K1=2 ‡ C
kSA=0.2120.01 L m min for the disappearance
of Orange II. This value is greater than would be
expected for any chlorinated solvent (Scherer et al.,
1998a) and is about 10 times larger than values
reported previously for nitro aromatic compounds
(Agrawal and Tratnyek, 1996).
where Vm is the maximum reaction rate for a par-
ticular experiment (type and amount of iron, ro-
tation rate, temperature, etc.) and K1/2 (C at Vm/2)
re¯ects the anity of the metal surface for the or-
ganic reactant. It has been shown previously that
K1/2 is roughly constant for a particular substrate
(presumably because all iron tends to be coated
with similar oxides (Scherer et al., 1998b)), but Vm
varies (due to di€erent concentrations of reactive
sites, etc. (Johnson et al., 1996, 1998; Scherer et al.,
1998b)). Therefore, we have ®t Eq. (4) to all the
data in Fig. 3 by treating K1/2 as a global parameter
(one value common to all four rotation rates) and
Vm as a local parameter (one value for each ro-
tation rate). The resulting values are K1/2=3.1 2
0.2 mM, and Vm=0.56 2 0.08, 1.02 2 0.04, 1.24 2
E€ect of initial concentration
The e€ect of initial dye concentration on the kin-
etics of dye reduction was investigated with Crocein
Orange G, and is shown in Fig. 3. Over the range
of C0 studied (60 mM to 3 mM), decolorization
rates (kobs  C0) increased monotonically but nonli-
nearly with C0 (at each mixing rate). Similar trends
have been observed for the reduction of carbon tet-
rachloride (Scherer and Tratnyek, 1995; Johnson et
al., 1996; Tratnyek et al., 1997; Scherer et al.,
1998b) and tetrachloroethene (Arnold and Roberts,
1997) in batch systems containing granular Fe0. All
of these systems appear to give a transition from
pseudo ®rst-order to zero-order kinetics with
1
0.06 and 1.5620.05 mM min for 90, 120, 140 and
160 rpm, respectively. The agreement between the
®tted curves and the data is apparent from Fig. 3.
Since Eq. (4) is an empirical model, the absolute
values of the ®tted parameters are of little general
Fig. 3. E€ect of initial dye concentration on initial deco-
lorization rate (kobs  C0) of Crocein Orange G by Fe0
Fig. 4. E€ect of mixing rate on initial decolorization rate
(unpretreated Fluka 16-32 mesh, ra=1.42 m2 L 1) at var- (kobs  C0) of Crocein Orange G by Fe0 (unpretreated
ious rotation rates on an orbital shaker. Data for 10 mM,
pH 7.0 HEPES bu€er at room temperature. Curves are
from ®ts to Eq. (4).
Fluka 16-32 mesh, ra=1.42 m2 L 1) at various initial con-
centrations. Data for 10 mM, pH 7.0 HEPES bu€er at
room temperature. Curves are from ®ts to Eq. (5).

1842
Sangkil Nam and Paul G. Tratnyek
signi®cance, but the qualitative agreement with pre- cates that the observed rate should approach zero
viously reported trends is remarkable. Again, the as o approaches zero, but our data indicate that
anity of organic reactant for the metal surface this limit occurs around o1/2=8 (Fig. 4). This
(K1/2) appears to be relatively una€ected by exper- anomaly appears to be due to inecient mixing by
imental conditions, and the controlling variables the table shaker used in this study because visual
apparently are represented by Vm. It has been observation revealed a breakpoint around 60 rpm
recognized previously that Vm depends primarily on below which the Fe0 particles are not suspended.
the concentration of reactive surface sites and the An ad hoc correction for this can be made by sub-
intrinsic rate constant for reaction at these sites tracting a constant o€set from o 1/2. Global ®tting
(Johnson et al., 1996; Scherer et al., 1998b), but the of the corrected empirical model to the data that
apparent increase in Vm with rpm suggests that Vm are not confounded with site saturation e€ects (60
is confounded with e€ects of mass transport under and 300 mM) gives the lines shown in Fig. 4. The
the conditions of this study.
®tted value for krxn is not well constrained, indicat-
ing that these results are controlled entirely by mass
E€ect of mixing
transport. The other ®tting parameters are: A =
To clarify the in¯uence that mass transport had 0.0920.04 rpm1/2 min and the o€set in o1/2 is 7.9
on the kinetics observed in this study, the data for 20.5. Extrapolation with these ®tted values to mM
Crocein Orange G have been plotted vs mixing rate concentrations of dye shows that the observed deco-
in Fig. 4. The rate of decolorization clearly lorization rates deviate from the rates predicted, as
increases with the rate of mixing, and, at micromo- the site saturation e€ect becomes increasingly im-
lar values of C0, the trend is linear with the square portant (top two sets of data in Fig. 4).
root of rpm. At higher C0, the data suggest some
1
E€ect of structure
tailing, presumably due to the saturation e€ect that
was discussed in the previous section. Qualitatively,
Another way to assess the relative signi®cance of
these results indicate that azo dye reduction in this mass transport and chemical reaction is through
system was limited by mass transport at low dye correlation analysis (Tratnyek, 1998). The kinetics
concentrations and by the saturation of surface sites of mass transport limited reactions often correlate
at high dye concentrations.
to di€usion coecients of the substrates (D), and
Evidence for mixed kinetic control has been the kinetics of reactions limited by chemical re-
found for other contaminants and other model sys- duction typically correlate to descriptors of elec-
tems containing granular Fe0 (e.g. Agrawal and tronic properties such as the energy of the lowest
Tratnyek, 1996), but the relative importance of unoccupied molecular orbital (ELUMO). For the
mass transport vs chemical reaction in these systems nine dyes used in this study, di€usion coecients
has not been quanti®ed. Scherer et al. (1997) were were estimated using the method of Hayduk and
able to do this for the reduction of carbon tetra- Laudie (Lyman et al., 1982), and values of ELUMO
chloride at an Fe0 rotating disk electrode, because were
calculated
with
MOPAC
(CAChe
the rotating disk electrode provides a well charac- WorkSystem software, Oxford Molecular Group,
terized ¯ow regime. The electrochemical model they Beaverton, OR) using AM1 and PM3 parameters,
used can be simpli®ed to the semi-empirical model:
with and without correction for solvation using the
conductor-like screening model (COSMO). The
values of these ®ve descriptors are summarized in
Table 3.
1
1
1
ˆ
‡
ꢀ5†
kobs
C
krxn
C
Ao1=2C
Fig. 5 shows a matrix of scatter plots illustrating
the correlations between kobs, D and two calcu-
lations of ELUMO. Only values of kobs measured at
o=100 rpm and C0=300 mM are included, because
they make up the largest group of experiments
where krxn is a ®rst order rate constant for reaction
at the surface, A is an empirical constant that
depends in part on the di€usion coecient of the
dye and o is the mixing rate in rpm. Eq. (5) indi-
Table 3. Dye descriptors
ELUMO ELUMO
No.
Name
D Â 106
ELUMO
(eV) AM1/COMSO
ELUMO
(eV) PM3/COMSO
1
)
(cm2
s
(eV) AM1
(eV) PM3
1
2
3
4
5
6
7
8
9
Acid Blue 113
Allura Red
Amaranth
3.035
3.876
3.646
4.537
3.391
4.537
4.537
4.188
3.878
2.139
3.189
5.045
0.999
2.339
0.967
0.891
3.066
5.526
2.083
3.039
4.932
0.874
2.291
0.890
0.766
2.918
5.448
1.702
1.462
1.701
1.180
1.857
1.503
1.327
1.504
1.327
1.493
1.365
1.533
1.307
1.631
1.208
1.424
1.391
1.174
Crocein Orange G
Naphthol Blue Black
Orange I
Orange II
Sunset Yellow FCF
Tartrazine

Reduction of azo dyes
1843
utility of Fe0 as a reductant in authentic waste-
water.
For this purpose, a sample of wastewater was
obtained from
a dye manufacturing facility in
South Carolina, U.S.A., and aliquots of the waste-
water were exposed to two types of granular Fe0
that are currently used in full scale installations of
permeable reactive barriers. Since the wastewater
sample was believed to contain a variety of disperse
azo dyes (as well as other dyes), the e€ect of treat-
ment with Fe0 was determined by periodically col-
lecting absorbance spectra (from 300 to 800 nm).
The spectra obtained with Peerless (Fig. 6(a)) and
Master Builder (not shown) Fe0 show decoloriza-
tion that is somewhat variable with wavelength.
This selectivity is more apparent from the di€erence
spectra shown in Fig. 6(b), which were calculated
by subtracting the scan at each elapsed time from
the ®rst scan. Initially, decolorization around 500 to
600 nm is the most notable change, presumably due
to reduction of labile dyes that have lmax in this
range (cf. Table 1). After about 45 min, the samples
were nearly decolorized, and the di€erence spectra
Fig. 5. Correlation matrix of rate constants for decoloriza-
tion of azo dyes (C0=0.3 mM and o=100 rpm), vs
selected descriptor variables: ELUMO (AM1 and PM3,
both with COSMO) and D. Coecients of correlation, r2,
are given on each plot. Points marked with crosses are for
dyes with two azo linkages.
where the type of dye is the only variable (Table 2).
Two smaller groups of data, corresponding to 120
and 140 rpm, are not shown because they exhibit
less distinctive but similar trends to those shown in
the ®rst row of Fig. 5. Correlations to the values of
ELUMO calculated without correction for solvation
are also not shown in Fig. 5 (because correlations
to the corrected descriptors are better). No individ-
ual data for azo dyes have been excluded. The data
in Fig. 5 reveal only weak correlations, and they
are strongly leveraged by the two points for dyes
containing two azo groups (Acid Blue 113 and
Naphthol Blue Black). Correcting kobs or ELUMO
for the possible formation of hydrazone tautomers
(in dyes with phenolic hydroxyl groups ortho to azo
linkages) is unlikely to improve the correlation
because all of the dyes used are susceptible to this
except for Orange I and Tartrazine. The ambiguity
in the trends shown in Fig. 5 does not allow us to
further resolve the role of mass transport and reac-
tion rate control by correlation analysis, but, at the
same time, it is not inconsistent the conclusions
drawn from the data in Figs 3 and 4.
Wastewater decolorization
All of the experiments described above were con-
ducted with solutions prepared in the laboratory
from single dyes and dilute aqueous bu€ers,
whereas wastewaters from dye manufacturing and
application operations are known to be very com-
plex mixtures of the colorants, conditioners, and
salts (Reife and Freeman, 1996; Carliell et al.,
1998). Since reduction of contaminants by Fe0 is
known to be very sensitive to solution chemistry
(Johnson et al., 1998), it was of interest to test the
Fig. 6. Absorbance (a) and di€erence (b) spectra for deco-
lorization of wastewater from a dye manufacturing facility
due to reaction with construction grade Fe0 (Peerless, 8±
16 mesh, 400 g L 1). Data are for 160 rpm and room tem-
perature.

1844
Sangkil Nam and Paul G. Tratnyek
Blowes D. W., Ptacek C. J. and Jambor J. L. (1997) In
situ remediation of Cr(VI)-contaminated groundwater
using permeable reactive walls: Laboratory studies.
Environ. Sci. Technol. 31(12), 3348±3357.
Burris D. R., Campbell T. J. and Manoranjan V. S. (1995)
Sorption of trichloroethylene and tetrachloroethylene in
a batch reactive metallic iron-water system. Environ. Sci.
Technol. 29(11), 2850±2855.
appear to stabilize in the visible region. Below
400 nm, however, the di€erence spectra suggest a
more sustained decrease in absorption, perhaps due
to ¯occulation of dissolved organics by dissolving
iron and associated adsorption to the metal-oxide
coated particles of Fe0. These side e€ects deserve
further investigation because they may contribute
signi®cant practical advantages over other chemical
reductants (e.g. dithionite, formamidine sul®nate,
Burris D. R., Hat®eld K. and Wolfe N. L. (1996)
Laboratory experiments with heterogeneous reactions in
mixed porous media. J. Environ. Eng. 122(8), 685±691.
formaldehydesulfoxylate, and borohydride (Dubrow Cantrell K. J., Kaplan D. I. and Wietsma T. W. (1995)
Zero-valent iron for the in situ remediation of selected
metals in groundwater. J. Haz. Mat. 42(2), 201±212.
Cao J., Wei L., Huang Q., Wang L. and Han S. (1999)
Reducing degradation of azo dyes by zero-valent iron in
aqueous solution. Chemosphere 38(3), 565±571.
Carliell C. M., Barclay S. J., Shaw C., Wheatley A. D.
and Buckley C. A. (1998) The e€ect of salts used in tex-
tile dyeing on microbial decolourisation of a reactive
azo dye. Environ. Technol. 19(11), 1133±1137.
et al., 1996)) that have been used to decolorize
wastewaters containing dyes.
CONCLUSIONS
All of the nine dyes tested were rapidly decolor-
ized by Fe0. Decolorization of the azo dyes was due
to reduction of the azo groups, which resulted in
formation of aromatics amines as products. The
kinetics of decolorization were ®rst order, and var-
ied with initial dye concentration, rate of mixing,
and dye structure in a manner that indicates that
the rates of reaction were controlled by mass trans-
fer. These results with azo dyes are consistent with
previous results obtained with other contaminants
(chlorinated solvents, nitro aromatics, and various
inorganics), suggesting that highly reactive com-
pounds studied in batch model systems with gentle
mixing are prone to mass transport in¯uenced kin-
etics, even though reaction rate controlled kinetics
are the norm for most classes of contaminants and
most experimental designs. With respect to the po-
tential for application of Fe0 in the treatment of
wastewaters from the dye industries, the impli-
cations of this study are mixed: although decolori-
zation of azo dyes is very rapid, the resulting
products are aromatic amines that may be of regu-
latory concern and require further treatment.
Cheng I. F., Muftikian R., Fernando Q. and Korte N.
(1997) Reduction of nitrate to ammonia by zero-valent
iron. Chemosphere 35(11), 2689±2695.
Chew C. F. and Zhang T. C. (1998) In situ remediation of
nitrate-contaminated ground water by electrokinetics/
iron wall processes. Water Sci. Technol. 38(7), 135±142.
Chuang F.-W., Larson R. A. and Scully Wessman M.
(1995) Zero-valent iron-promoted dechlorination of
polychlorinated biphenyls. Environ. Sci. Technol. 29(9),
2460±2463.
Dubrow S. F., Boardman G. D. and Michelsen D. L.
(1996) Chemical pretreatment and aerobic±anaerobic
degradation of textile dye wastewater. In Environmental
Chemistry of Dyes and Pigments, eds A. Reife and H. S.
Freeman, pp. 75±104. Wiley, New York.
Eykholt G. R. and Davenport D. T. (1998) Dechlorination
of the chloroacetanilide herbicides alachlor and metola-
chlor by iron metal. Environ. Sci. Technol. 32(10), 1482±
1487.
Fiedor J. N., Bostick W. D., Jarabek R. J. and Farrell J.
(1998) Understanding the mechanism of uranium
removal from groundwater by zero-valent iron using X-
ray photoelectron spectroscopy. Environ. Sci. Technol.
32(10), 1466±1473.
Gillham R. W. and O'Hannesin S. F. (1994) Enhanced
degradation of halogenated aliphatics by zero-valent
iron. Ground Water 32(6), 958±967.
AcknowledgementsÐThe authors would like to recognize
V. Renganathan (Oregon Graduate Institute), who was S.
N.'s major advisor during this and other studies of the en-
vironmental fate of azo dyes; Cikui Liang and David
Gallagher (Oxford Molecular Group), who helped to per-
form molecular modeling calculations on the largest dyes;
and Michelle Scherer (University of Iowa) and Joel
Bandstra (OGI) for their input on the role of mass trans-
port in this system.
Grittini C., Malcomson M., Fernando Q. and Korte N.
(1995) Rapid dechlorination of polychlorinated biphe-
nyls on the surface of a Pd/Fe bimetallic system.
Environ. Sci. Technol. 29(11), 2898±2900.
Gu B., Liang L., Dickey M. J., Yin X. and Dai S. (1998)
Reductive precipitation of uranium(VI) by zero-valent
iron. Environ. Sci. Technol. 32(21), 3366±3373.
Huang C.-P., Wang H.-W. and Chiu P.-C. (1998) Nitrate
reduction by metallic iron. Water Res. 32(8), 2257±2264.
Johnson T. L., Scherer M. M. and Tratnyek P. G. (1996)
Kinetics of halogenated organic compound degradation
by iron metal. Environ. Sci. Technol. 30(8), 2634±2640.
Johnson T. L., Fish W., Gorby Y. A. and Tratnyek P. G.
(1998) Degradation of carbon tetrachloride by iron
metal: Complexation e€ects on the oxide surface. J.
Contam. Hydrol. 29(4), 377±396.
Kudlich M., Hetheridge M. J., Knackmuss H.-J. and Stolz
A. (1999) Autoxidation reactions of di€erent aromatic
o-aminohydroxynaphthalenes that are formed during
the anaerobic reduction of sulfonated azo dyes. Environ.
Sci. Technol. 33(6), 896±901.
REFERENCES
Agrawal A. and Tratnyek P. G. (1996) Reduction of nitro
aromatic compounds by zero-valent iron metal. Environ.
Sci. Technol. 30(1), 153±160.
Allen-King R. M., Halket R. M. and Burris D. R. (1997)
Reductive transformation and sorption of cis- and
trans-1,2-dichloroethene in a metallic iron-water system.
Environ. Toxicol. Chem. 16(3), 424±429.
Arnold W. A. and Roberts A. L. (1997) Development of a
quantative model for chlorinated ethylene reduction by
zero-valent metals. Preprint Extended Abstracts, Div.
Environ. Chem., American Chemical Society, 213th
National Meeting, San Francisco, CA 37(1), 76±77.
Laszlo J. A. (1997) Regeneration of dye-saturated quater-
nized cellulose by bisul®te-mediated borohydride re-
duction of dye azo groups: An improved process for

Reduction of azo dyes
1845
decolorization of textile wastewaters. Environ. Sci.
Technol. 31(12), 3647±3653.
Lyman W. J., Reehl W. F. and Rosenblatt D. H. (eds)
(1982) Handbook of Chemical Property Estimation
Methods. McGraw-Hill, New York.
Matheson L. J. and Tratnyek P. G. (1994) Reductive
dehalogenation of chlorinated methanes by iron metal.
Environ. Sci. Technol. 28(12), 2045±2053.
Monson S. J., Ma L., Cassanda D. A. and Spalding R. F.
(1998) Con®rmation and method development for
dechlorinated atrazine from reductive dehalogenation of
atrazine with Fe0. Anal. Chim. Acta 373, 153±160.
Pratt A. R., Blowes D. W. and Ptacek C. J. (1997)
Products of chromate reduction on proposed subsurface
remediation material. Environ. Sci. Technol. 31(9), 2492.
Reife A. and Freeman H. S. (eds) (1996) Environmental
Chemistry of Dyes and Pigments. Wiley, New York.
Sayles G. D., You G., Wang M. and Kupferle M. J.
(1997) DDT, DDD, and DDE dechlorination by zero-
valent iron. Environ. Sci. Technol. 31(12), 3448±3454.
Scherer M. M. and Tratnyek P. G. (1995) Dechlorination
of carbon tetrachloride by iron metal: E€ect of reactant
concentrations. Preprint Extented Abstracts, Div.
Environ. Chem., American Chemical Society, 209th
National Meeting, Anaheim, CA 35(1), 805±806.
bu€er. Preprint Extended Abstracts, Div. Environ. Chem.,
American Chemical Society, 209th National Meeting,
Anaheim, CA 35(1), 833±835.
Singh J., Comfort S. D. and Shea P. J. (1998a)
Remediating RDX-contaminated water and soil using
zero-valent iron. J. Environ. Qual. 27(5), 1240±1245.
Singh J., Shea P. J., Hundal L. S., Comfort S. D., Zhang
T. D. and Hage D. S. (1998b) Iron-enhanced remedia-
tion of water and soil containing atrazine. Weed Sci. 46,
381±388.
Smolen J. M., Weber E. J. and Tratnyek P. G. (1999)
Molecular probe techniques for reductant identi®cation
in reducing sediments. Environ. Sci. Technol. 33(3), 440±
445.
Su C. and Puls R. W. (1999) Kinetics of trichloroethene
reduction by zerovalent iron and tin: Pretreatment
e€ect, apparent activation energy, and intermediate pro-
ducts. Environ. Sci. Technol. 33(1), 163±168.
Tang W. Z. and Chen R. Z. (1996) Decolorization kinetics
and mechanisms of commercial dyes by H2O2/iron pow-
der system. Chemosphere 32(5), 947±958.
Tratnyek P. G., Johnson T. L., Scherer M. M. and
Eykholt G. R. (1997) Remediating groundwater with
zero-valent metals: kinetic considerations in barrier de-
sign. Ground Water Monit. Rem. 17(4), 108±114.
Tratnyek P. G. (1998) Correlation analysis of the environ-
mental reactivity of organic substances. In Perspectives
in Environmental Chemistry, ed. D. L. Macalady, pp.
167±194. Oxford Univ Press, New York.
Scherer M. M., Balko B. A., Gallagher D. A. and
Tratnyek P. G. (1998a) Correlation analysis of rate con-
stants for dechlorination by zero-valent iron. Environ.
Sci. Technol. 32(19), 3026±3033.
Scherer M. M., Balko B. A. and Tratnyek P. G. (1998b)
The role of oxides in reduction reactions at the metal±
water interface. In Mineral±Water Interfacial Reactions:
Kinetics and Mechanisms, eds D. L. Sparks and T. J.
Grundl, ACS Symposium Series, No. 715. American
Chemical Society, Washington, DC, pp. 301±322.
Scherer M. M., Westall J. C., Ziomek-Moroz M. and
Tratnyek P. G. (1997) Kinetics of carbon tetrachloride
reduction at an oxide-free iron electrode. Environ. Sci.
Technol. 31(8), 2385±2391.
Scherer M. M., Westall J. C. and Tratnyek P. G. (1999)
The kinetics of nitro reduction by iron metal: A case of
mixed control. Preprint Extended Abstracts, Div.
Environ. Chem., American Chemical Society, 217th
National Meeting, Anaheim, CA 39(1), 346±348.
Weber E. J. and Wolfe N. L. (1987) Kinetic studies of the
reduction of aromatic azo compounds in anaerobic sedi-
ment/water systems. Environ. Toxicol. Chem. 6, 911±
919.
Weber E. J. and Adams R. L. (1995) Chemical- and sedi-
ment mediated reduction of the azo dye disperse blue
79. Environ. Sci. Technol. 29(5), 1163±1170.
Weber E. J. (1996) Iron-mediated reductive transform-
ations: Investigation of reaction mechanism. Environ.
Sci. Technol. 30(2), 716±719.
Yen C.-P. C., Perenich T. A. and Baughman G. L. (1991)
Fate of commercial disperse dyes in sediments. Environ.
Toxicol. Chem. 10(8), 1009±1017.
Zawaideh L. I. and Zhang T. C. (1998) The e€ects of pH
and addition of an organic bu€er (HEPES) on nitrate
transformation in Fe0±water systems. Water Sci.
Technol. 38(7), 107±115.
Schreier C. G. and Reinhard M. (1995) Transformation of
chlorinated ethylenes by iron powder in 4-(2-
Hydroxyethyl)-1-Piperazineethanesulfonic Acid (HEPES)