Reduction of N-nitrosodimethylamine with granular iron and nickel-enhanced iron. 1. Pathways and kinetics

Article abstract of DOI:10.1021/es9909778

Laboratory batch and column tests were conducted to examine the reduction pathways and kinetics of N-nitrosodimethylamine (NDMA) by iron (Fe) and nickel-enhanced iron (Ni/Fe). A decrease in NDMA concentration and increases in dimethylamine (DMA) and ammonium were observed in both Fe and Ni/Fe columns. In the Fe column, the transformation process of NDMA appeared to follow pseudo-first-order kinetics with respect to NDMA, with an average half-life of 13±2 h. A small amount of nickel (0.25%) plated onto the iron greatly enhanced NDMA transformation rates. At early time the NDMA half-life in the Ni/Fe column was 2 min but as time progressed the half-life increased to 4 min, and departures from first-order kinetics were observed. The mass balances of carbon in DMA and nitrogen in DMA and ammonium improved over time and reached 100% and 90%, respectively, after NDMA had passed through the column for more than 50 pore volumes (PV). No 1,1-dimethylhydrazine, nitrous oxide, or methane were detected. Based on the electrochemical properties of NDMA, the transformation mechanism of NDMA with Fe and Ni/Fe is postulated to be catalytic hydrogenation, resulting in N-N bond breakdown to form DMA and ammonium as final products. Nickel, being a much stronger catalyst than Fe for catalytic hydrogenation, resulted in a much faster reduction rate of NDMA. Of several methods tested, flushing the Ni/Fe column with 0.01 N sulfuric acid proved to be the most effective in restoring the Ni/Fe activity. The rapid transformation rate on Ni/Fe and the formation of nontoxic products indicate that this material may be applicable for treating NDMA contaminated water, both in-situ and above ground. Laboratory batch and column tests were conducted to examine the reduction pathways and kinetics of N-nitrosodimethylamine (NDMA) by iron (Fe) and nickel-enhanced iron (Ni/Fe). A decrease in NDMA concentration and increases in dimethylamine (DMA) and ammonium were observed in both Fe and Ni/Fe columns. In the Fe column, the transformation process of NDMA appeared to follow pseudo-first-order kinetics with respect to NDMA, with an average half-life of 13±2 h. A small amount of nickel (0.25%) plated onto the iron greatly enhanced NDMA transformation rates. At early time the NDMA half-life in the Ni/Fe column was 2 min but as time progressed the half-life increased to 4 min, and departures from first-order kinetics were observed. The mass balances of carbon in DMA and nitrogen in DMA and ammonium improved over time and reached 100% and 90%, respectively, after NDMA had passed through the column for more than 50 pore volumes (PV). No 1,1-dimethylhydrazine, nitrous oxide, or methane were detected. Based on the electrochemical properties of NDMA, the transformation mechanism of NDMA with Fe and Ni/Fe is postulated to be catalytic hydrogenation, resulting in N-N bond breakdown to form DMA and ammonium as final products. Nickel, being a much stronger catalyst than Fe for catalytic hydrogenation, resulted in a much faster reduction rate of NDMA. Of several methods tested, flushing the Ni/Fe column with 0.01 N sulfuric acid proved to be the most effective in restoring the Ni/Fe activity. The rapid transformation rate on Ni/Fe and the formation of nontoxic products indicate that this material may be applicable for treating NDMA contaminated water, both in-situ and above ground.

Full text of DOI:10.1021/es9909778

Environ. Sci. Technol. 2000, 34, 3489-3494
NDMA is very soluble and volatile and therefore can be readily
Reduction of
transported into various environmental compartments in-
cluding groundwater. Recently NDMA, at concentrations of
parts per billion, has been detected in groundwater samples
in several locations, for example, in Elmira, Ontario, Canada
N-Nitrosodimethylamine with
Granular Iron and Nickel-Enhanced
Iron. 1. Pathways and Kinetics
(
8), and in surface waters, seawaters, and soils (9).
To date research on NDMA has focused mainly on its
toxicity effects on humans and animals (10, 11) and its
occurrence in foodstuffs such as fruit juices and beer (2). In
natural environments or ambient conditions, NDMA appears
to be recalcitrant under both aerobic and anaerobic condi-
tions (7, 12) and is, therefore, difficult to remove from water.
Currently NDMA in wastewaters is treated using physical or
chemical methods, including UV destruction and adsorption
by granular activated carbon (4). These methods are either
very expensive or insufficient for achieving complete NDMA
removal, and thus there is a need for developing an efficient
and relatively inexpensive method that can destroy NDMA
to innocuous compounds.
L A I G U I , R O B E R T W . G I L L H A M , * A N D
M A R E K S . O D Z I E M K O W S K I
Department of Earth Sciences, University of Waterloo,
Waterloo, Ontario, Canada N2L 3G1
Laboratory batch and column tests were conducted to
examine the reduction pathways and kinetics of N-
nitrosodimethylamine (NDMA) by iron (Fe) and nickel-
enhanced iron (Ni/Fe). A decrease in NDMA concentration
and increases in dimethylamine (DMA) and ammonium
were observed in both Fe and Ni/Fe columns. In the Fe
column, the transformation process of NDMA appeared to
follow pseudo-first-order kinetics with respect to NDMA,
with an average half-life of 13(2 h. A small amount of nickel
An innovative technology, developed by Gillham and
O’Hannesin, uses granular iron (Fe) as an electron source
for reductively degrading chlorinated solvents (13). This
technology has been widely accepted as a cost-effective
method for treating contaminated groundwater. Subsequent
research has extended the iron technology to include
reduction of nitroaromatics (14), azo dyes (15), and nitrate
(16) as well as the removal of chromium and uranium by
reductive precipitation (17, 18). Reduction of nitroaromatic
compounds proceeds stepwise, forming aromatic nitroso and
hydroxylamines as intermediates. Although nitro reduction
occurs much more easily on aromatic rings than on aliphatic
moieties (19), it nevertheless suggests that the nitroso group
in NDMA may be susceptible to reduction by iron, particularly
with nickel-enhanced iron (Ni/ Fe). Studies have shown that
Fe plated with a small amount of a more noble metal, such
as palladium or nickel, increases the dechlorination rates by
at least a factor of 10 (20, 21). Other studies have shown that
nitrosamines can be reduced by Raney-nickel and lithium
aluminum hydride to corresponding amines (22) and by some
low-valent titanium reagents to the corresponding hydrazines
(0.25%) plated onto the iron greatly enhanced NDMA
transformation rates. At early time the NDMA half-life in
the Ni/Fe column was 2 min but as time progressed the half-
life increased to 4 min, and departures from first-order
kinetics were observed. The mass balances of carbon in
DMA and nitrogen in DMA and ammonium improved over
time and reached 100% and 90%, respectively, after
NDMA had passed through the column for more than 50
pore volumes (PV). No 1,1-dimethylhydrazine, nitrous oxide,
or methane were detected. Based on the electrochemical
properties of NDMA, the transformation mechanism of
NDMA with Fe and Ni/Fe is postulated to be catalytic
hydrogenation, resulting in N-N bond breakdown to form
DMA and ammonium as final products. Nickel, being a
much stronger catalyst than Fe for catalytic hydrogenation,
resulted in a much faster reduction rate of NDMA. Of
several methods tested, flushing the Ni/Fe column with
(
23).
The objectives of the present study were (1) to determine
NDMA degradation kinetics and pathways with Fe and Ni/
Fe and (2) to examine factors affecting NDMA degradation
rates by Ni/ Fe. Column tests were conducted to compare
the effectiveness of Fe and Ni/ Fe in NDMA transformation.
Supplementary batch experiments were conducted to fa-
cilitate the examination of potential intermediates/ products
and adsorption of products.
0.01 N sulfuric acid proved to be the most effective in restoring
the Ni/Fe activity. The rapid transformation rate on Ni/Fe
and the formation of nontoxic products indicate that this
material may be applicable for treating NDMA contaminated
water, both in-situ and above ground.
Methods
Introduction
Chem icals and Materials. NDMA, DMA‚HCl (99%), and
dinitrofluorobenzene (DNFB, 99%) were purchased from
Sigma, and 1,1-dimethylhydrazine (UDMH‚2HCl, g99%) was
purchased from Fluka. All chemicals were used without
further purification.
N-Nitrosodimethylamine (NDMA) is one of the few proven
potent carcinogens (1, 2). Consequently, it is listed as one of
the U.S. EPA priority pollutants (3) and is a stringently
regulated compound. The U.S. EPA drinking water standard
for NDMA is set at 0.7 parts per trillion (4). NDMA was mainly
produced in rocket fuel production (4, 5) and has also been
used as an antioxidant, an additive for lubricants, and a
softener of copolymers (5). In addition, NDMA can be formed
naturally in the environment where precursors such as nitrite,
nitrous oxides, and dimethylamine (DMA) coexist (6, 7).
The granular iron used in the study was obtained from
Connelly-GPM Inc. (Chicago, IL). The material contains 89.8%
metallic iron, and the surfaces were covered with various
forms of iron oxides (data provided by Connelly GPM Inc.,
1998). The surface area, measured by the BET method, was
2
1.8 m / g. This material was used without prior cleaning.
The Ni/ Fe bimetal was produced by electroless plating in
an acid hypophosphite solution. The plating was performed
using the procedure given in ref 20. The iron was used as
*
Corresponding author phone: (519)888-4658; fax: (519)746-1829;
e-mail: rwgillha@sciborg.uwaterloo.ca.
1
0.1021/es9909778 CCC: $19.00

2000 Am erican Chem ical Society
VOL. 34, NO. 16, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3 4 8 9
Published on Web 07/14/2000

TABLE 1. Experimental Conditions and the Apparent NDMA Reduction Kinetics in the Ni/Fe Columnc
kinetic fita
rate constant
b
r2
note
total PV flow rate, mL/min initial concn, mg/L
7
3.2
0.21
8.64
8.30
8.30
9.01
8.85
10.05
11.3
8.7
3.20
2.40
4.47
10.65
5.3
6.2
first-order
first-order
first-order
zero-order
first-order
m ixed-order
zero-order
zero-order
zero-order
m ixed-order
first-order
m ixed-order
first-order
0.35 ( 0.05
0.25 ( 0.03
0.17 ( 0.02
0.60 ( 0.01
0.17 ( 0.03
0.956 before any treatm ent
0.972
0.938
0.999
1
2
3
3
5
5
7
8
9
9
9
02.5
72.8
07.2
12.4
65.2
92.5
88.8
12.7
12.4
21.8
92.1
26.3
27.0
25.1
14.7
28.4
21.4
23.4
25.2
25.6
24.2
24.0
0.959 5 PV after 1 wk shut off
0.92 ( 0.08
0.19 ( 0.01
0.61 ( 0.07
0.992
0.994
0.985
23 PV after 43 PV distilled-H2O flush
28 PV after 86 PV distilled-H2O flush
0.035 ( 0.002 0.997 6 PV after 13 d shut off
0.16 ( 0.01 0.971 29 PV after 70 PV acid (pH ) 2) wash
1
050
a
2
Plots of NDMA concentration vs residence tim e were fitted with either exponential or linear regression m odels. The best fit with highest r
b
(
>0.9) were chosen to describe pseudo-first-order and pseudo-zero-order kinetics. The dim ensions for pseudo-first-order rate constant and
pseudo-zero-order rate constant are m in and (m g/L) m in , respectively. Prior to 73 PV NDMA concentration decreased to below detection
lim it at the first sam pling port at a flow rate of 0.25 m L/m in and an initial concentration of 5 m g/L. Therefore it was not included in the table.
-1 -1 -1 c
received without acid wash. The surface area of the Ni/ Fe,
measured by the BET method, was 3.1 m / g.
Colum n Experim ents. Four columns were prepared using
Plexiglas tubing (50 cm × 3.81 cm ID) which were constructed
according to Gillham and O’Hannesin (13). Seven sampling
ports were located along each column at distances of 2.5, 5,
The second batch experiment was prepared in a similar
manner using UDMH as the starting material. Each bottle
(60 mL), containing 10 g of Fe or Ni/ Fe, was filled with a 20
mg/ L UDMH solution. Periodically duplicate bottles were
sacrificed, and aqueous samples were removed for UDMH,
DMA, and ammonium analyses.
2
1
0, 15, 20, 30, and 40 cm from the influent end. Sampling
Analytical Methods. Analyses for NDMA, DMA, and
UDMH were performed using a series 1100 Hewlett-Packard
high performance liquid chromatograph (HPLC). The HPLC
consisted of a diode array detector, a quaternary pump, an
autosampler, and a degasser. A C18 ODS Hypersil column (5
µm particles, 125 × 4 mm) equipped with a LiChrospher 100
RP-8 guard column (5 µm, 4 × 4 mm) was used. Aqueous
samples collected from sampling ports along the column or
from the batch bottles were filtered through 0.2 µm syringe
filters (Gelman Sciences, Ann Arbor, MI). For NDMA analysis,
20 µL of filtrant was injected onto the HPLC, and the
absorbance of NDMA was measured at a wavelength of 225
nm. The mobile phase composition was 80/ 20 water/
methanol, and the flow rate was 1 mL/ min. For DMA and
UDMH analyses, a precolumn derivatization method was
developed using DNFB (Sanger’s reagent) as the derivati-
zation reagent. DNFB reagent was prepared daily in aceto-
nitrile. The molar ratio of DNFB to DMA or UDMH was greater
than 50 to ensure complete derivatization. In 1.5 mL GC
vials, 325 µL of filtered aqueous samples were combined
with 25 µL of DNFB, 100 µL of acetonitrile, and 50 µL of 0.1
M NaOH (adjusting pH to 12). The GC vials were then vortexed
for a few seconds. The derivatization reactions were com-
pleted in less than 2 min, and the derivatives were stable for
at least 36 h at room temperature. The absorbance of the
derivatives was measured at a wavelength of 380 nm. The
mobile phase consisted of 40/ 60 0.2% acetic acid (pH 4)/
methanol. The flow rate was 1.0 mL/ min, and the injection
volume was 20 µL.
ports, mounted along the side of the column, consisted of
a nylon Swageloc fitting (0.16 cm O.D.) with a 16 gauge luer
lock syringe needle. Columns were packed with either 100%
Fe or 100% Ni/ Fe in 2 cm increments to a porosity of 0.59.
2
NDMA solution prepared with distilled water (DI-H O)
without deoxygenation was pumped through the columns
from the bottom using a variable speed multichannel
peristaltic pump. The average flow rate was 0.21 mL/ min for
the Fe columns and was varied between 0.2 mL/ min and 10
mL/ min for the Ni/ Fe columns (Table 1). Flow rates were
determined by measuring the discharge, and the variation
in flow rates was accounted for in calculating residence time
of solution in the columns. The initial concentration of NDMA
solutions varied between 5 mg/ L and 28 mg/ L. The NDMA
solution was stored in a 30-L collapsable Teflon bag to
minimize volatile loss and was kept in the dark to prevent
possible photolysis. One pair of Fe and Ni/ Fe columns
2
received only DI-H O to serve as controls. The columns
were sampled by clamping the effluent line to allow water
to flow from a sampling port directly into a glass syringe.
Batch Experim ents. Two sets of batch experiments were
2
conducted. One was to examine the formation of N O during
NDMA degradation with Fe and Ni/ Fe as well as to determine
NH
bottles, 50 g of Fe or Ni/ Fe and 100 mL of 50 mg/ L NDMA
solution were added, resulting in a headspace-to-solution
ratio of 1:2. Two sets of controls were prepared. One set
+
4
adsorption on Fe and Ni/ Fe materials. In 160 mL serum
2
contained Fe or Ni/ Fe and was filled with DI-H O, while the
second set contained only NDMA solution without metals.
The bottles were sealed immediately with Teflon-faced
silicone septa and aluminum seals and mixed daily by hand
shaking. They were stored in the dark at room temperature
for 7 days. Samples of 2 mL headspace from each bottle were
2
Analysis for N O gas was performed on a GOW-MAC series
350GP gas chromatograph (GC) equipped with a thermal
conductivity detector and a Porapak Q (100/ 120) mesh
column (2 m × 0.16 cm). The carrier gas was helium at a flow
rate of 20 mL/ min. The detector temperature was 140 °C,
and the oven temperature was isothermal at room temper-
ature. Two mL gas samples were injected onto the GC after
shaking at 250 rpm for 2 h to allow equilibration between the
2
injected onto a GC for N O analysis. Thereafter the serum
bottles were centrifuged at 2000 rpm for 15 min. The
supernatant was carefully removed for NH
+
4
and organic
analyses. Bottles were weighed to determine the amount of
2
aqueous and gas phases. The concentration of N O in samples
solution remaining. Then 50 mL of DI-H
2
O was added to
from metal particles by shaking
was calculated based on gas-phase concentrations using an
Oswald Coefficient of 0.5937 at 25 °C (24).
+
each bottle to wash off NH
4
at 250 rpm for 30 min. The bottles were centrifuged again,
Analyses for methane and other hydrocarbons were
performed on a Hewlett-Packard Model 5790A GC equipped
with a flame ionization detector and a Megabore GS-Q
capillary column. The carrier gas was air at a flow rate of 10
+
and the supernatant was collected for NH
4
analysis. The
+
NH
4
concentrations in the wash water were calculated to
account for the amount of unremoved solution.
3
4 9 0
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 16, 2000

a
FIGURE 1. Decreases in NDMA concentration with time in Fe and
Ni/Fe columns. In the Fe column, the flow rate was 0.2 mL/min, and
the initial NDMA concentration was 5 mg/L. In the Ni/Fe column,
the flow rate was 8.7 mL/min, and the initial NDMA concentration
was 25 mg/L.
b
mL/ min. The detector temperature was 120 °C, and the
injector temperature was 60 °C. The oven temperature was
held at 60 °C for 5 min and then ramped up to 120 °C at 15
°C/ min. A 2-mL aqueous sample collected with a glass syringe
was transferred to a 4-mL vial. Samples were shaken for 15
min at 250 rpm to allow organic partitioning between the
aqueous and gas phases. Headspace samples of 250 µL were
injected onto the GC using a gastight syringe. Aqueous
concentrations of hydrocarbons were calculated using the
Henry’s law constants.
Ammonium was determined by Philip Analytical Services
Co. (Mississauga, Ontario), using the automated phenate
method. Total aqueous concentrations of Ni and Fe were
analyzed by Atomic Absorption Analysis by the Water Quality
Laboratory at the University of Waterloo.
FIGURE 2. Decreases in NDMA concentration, increases in DMA
and ammonium concentrations, and mass balances for (a) the Fe
column after 73 PV and (b) the Ni/Fe column after 273 PV.
Results
cm). From the first sampling port to the effluent end Eh
fluctuated slightly with a lowest observed value of -151 mV.
Again the changes in pH and Eh were similar in the control
column which received only DI-H O.
2
NDMA Transform ation Products. In both the Fe and Ni/
Fe columns, the NDMA transformation products detected
NDMA Transform ation Using Fe and Ni/Fe. For tests
conducted on the Fe column, NDMA concentration profiles
along the column were measured after every 6 to 7 pore
volumes (PV). After 30 PV of NDMA solution passed through
the column, the concentration profiles appeared to reach a
quasi steady state. At an initial concentration of 5 mg/ L, the
average NDMA removal was 83.0 ( 4.8% (n ) 11) by the
effluent end. Figure 1 includes an example of NDMA
disappearance along the column (73 PV). Using a pseudo-
first-order kinetic model, the average rate constant between
were DMA and ammonium. As NDMA disappeared, DMA
+
and NH
4
formed accordingly. The carbon mass balance was
about 80% at low pore volumes and increased gradually.
After 56 PV in the Fe column and 38 PV in the Ni/ Fe column,
a carbon mass balance near 100% was obtained as DMA.
Examples of mass balances in the Fe column and the Ni/ Fe
column are shown in Figure 2 where the carbon mass balance
reached 98 ( 2% in the Fe column (73 PV, Figure 2a) and 93
( 7% (273 PV, Figure 2b). This suggests that at early time a
portion of the DMA was adsorbed onto the particle surfaces.
Analyses of aqueous samples collected from columns and
batch bottles showed formation of very low amounts (sub
µg/ L, data not shown) of methane and other hydrocarbons.
Similarly low levels of hydrocarbons were detected in the Fe
and Ni/ Fe controls. Thus the hydrocarbons detected are
unlikely a result of the presence of NDMA in the solution.
-
2
-1
3
0 PV and 100 PV was (5.7 ( 0.7) × 10
h
(n ) 11), hence
the average half-life of NDMA in the Fe column was 13 ( 2
h. The pH and Eh profiles were relatively stable after 25 PV.
From the influent end to the effluent end, the pH gradually
increased from 6.3 to 9.2 and the Eh decreased from 400 mV
to -126 mV. The changes in pH and Eh were similar in the
2
control column which received only DI-H O.
In the Ni/ Fe column, the rate of NDMA disappearance
was much faster than in the Fe column (Figure 1). During
the initial 273 PV, the decrease in NDMA concentration
appeared to follow pseudo-first-order kinetics over an
influent concentration range between 5 and 28 mg/ L. The
rate constant, however, decreased over time, for instance,
The total nitrogen mass balance in the Fe column was 90
( 4% in which nitrogen as DMA accounted for 49 ( 1% and
-
1
-1
+
from 0.35 min after 73 PV to 0.25 min after 102 PV and
as NH
4
accounted for 41 ( 3% (73 PV, Figure 2a). In the
-
1
to 0.17 min after 273 PV (Table 1). Thereafter, the kinetics
appeared to change with both zero-order and mixed-order
kinetics being observed (Table 1). The pH values along the
column ranged between 6.5 and 8.1, with no clear trend along
the column. The Eh dropped sharply from 300 mV in the
influent to around zero mV at the first sampling port (2.5
Ni/ Fe column nitrogen as DMA accounted for 46 ( 4% and
+
as NH
4
accounted for 42 ( 5%, resulting in a total nitrogen
mass balance of 88 ( 9% (273 PV, Figure 2b). No other
nitrogen containing intermediates or products of NDMA
reduction were detected. UDMH was not detected in aqueous
samples taken from batch and column experiments with
VOL. 34, NO. 16, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3 4 9 1

either Fe or Ni/ Fe. A three-week batch test showed an
insignificant decrease in UDMH concentration in the pres-
ence of Fe or Ni/ Fe. Thus UDMH is ruled out as an
intermediate of NDMA reduction in our experimental
2
systems. N O, which was examined in detail in a batch
experiment, was not detected in the headspace samples with
either Fe or Ni/ Fe at a metal to solution-to-headspace ratio
of 1:2:1 (wt:vol:vol).
Ammonium adsorption on metal surfaces was examined
in a batch experiment in which the iron particles were washed
with DI-H
2
O at the end of the experiment. A net recovery
+
+
of 0.09 µg NH
4
/ g Fe and 0.26 µg NH / g Ni/ Fe was obtained
4
in the batch experiment, respectively, which accounted for
+
1
7.4% and 20.2% of the total NH
4
/ N mass balances,
respectively. This suggests that some nitrogen loss unac-
counted for in the column experiment may be attributed to
ammonium adsorption. Furthermore, some ammonium may
not be detected due to its complexation with dissolved iron.
Im provem ent of Ni/Fe Colum n Perform ance. A loss of
reactivity in the Ni/ Fe column was observed as time
progressed (Table 1). Consequently several experiments were
conducted to examine ways that may help to regain the
reactivity. These experiments included shutting off the
column for a few days, flushing the column with distilled
water, and flushing with acid solution (H
results of these sequential experiments, summarized in Table
, indicate that first, the recovery of Ni/ Fe reactivity (i.e.
2 4
SO , pH 2). The
1
NDMA reduction rates) was achieved after each treatment.
The extent of the recovery, however, decreased after each
treatment with the exception of acid wash. Second, the rate
of reactivity loss increased over time. Detailed studies showed
that shutting the column down not only recovered the Ni/ Fe
reactivity but also changed NDMA reduction kinetics from
zero- or mixed-order to pseudo-first-order kinetics, as initially
observed (Figure 3a, Table 1). Compared to the initial pseudo-
-1
first-order rate constants after 73 PV (0.35 min ), the pseudo-
first-order rate constants after a 7-day shut off at 307 PV and
-
1
-1
a 13-day shut off at 912 PV were 0.17 min and 0.031 min
,
FIGURE 3. Recovery of lost reactivity of Ni/Fe after (a) shutting the
column off for 7 days after 307 PV and 13 days after 912 PV and (b)
flushing with distilled water for 86 PV after 789 PV and with pH 2
H SO solution for 67 PV after 1021 PV. The symbols are measured
concentrations, and the dash lines are calculated from kinetic
models.
respectively, suggesting loss of recoverable active sites by
this means. In addition, it was noticed that by shutting the
column off, the brownish rust that had formed at the very
bottom part of the column disappeared after a few days.
2
4
2
Flushing with DI-H O also regained some activity of the
Ni/ Fe for NDMA reduction but to a lesser extent (Figure 3b).
The NDMA reduction appeared to follow zero-order-kinetics
after a 43 PV flush at 595 PV and an 86 PV flush at 789 PV
(
Table 1) with half-lives of 16.3 and 19.2 min, respectively.
In contrast, flushing with pH 2 H SO for 67 PV dramatically
2
4
increased the Ni/ Fe reactivity, and NDMA transformation
kinetics appeared to be pseudo-first-order again (Figure 3b).
After 29 PV following the acid wash, the pseudo-first-order
-
1
rate constant increased to 0.16 min , which was similar to
the values measured at 273 PV (before any column activation
treatment) and after one week shutting off for the first time
(
Table 1). Analysis of effluent samples collected during flushes
with water and acid showed that NDMA concentration
dropped rapidly and complete removal was achieved after
3
PV (Figure 4), noticing that the decrease in NDMA
concentrations was due not only to water replacement but
also concurrent degradation. DMA was also flushed out by
water and acid solution, however, with some retardation
FIGURE 4. Effluent concentrations of NDMA, DMA, total dissolved
Fe and Ni, and changes in pH during Ni/Fe column wash with pH
(
Figure 4). A relatively fast removal of DMA occurred during
the initial 10 PV, and a slow but continuous removal was
observed thereafter. During the acid wash, the pH of the
effluent dropped from 8.5 to approximately 5 during the first
2
2 4
H SO solution. The flow rate used was 10.3 mL/min for the initial
21 PV and was reduced to 3.5 mL/min thereafter.
3
.6 PV then remained at this level with small fluctuations
some fluctuation, while the total Ni concentrations declined
gradually to 0.7 mg/ L at the end of the acid wash. The
cumulative amount of Ni that was washed out of the column
was less than 1% of the total plated Ni in the column. No Ni
was detected in the effluent during the water flush.
(
Figure 4). Accordingly dissolution of Fe and Ni was observed
in the effluent and quickly reached the maximum concen-
trations of 240 mg/ L and 3.5 mg/ L, respectively (Figure 4).
The total Fe concentration remained above 200 mg/ L with
3
4 9 2
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 16, 2000

distribution since methyl groups are inductively electron
donating. The amine nitrogen might have a partial positive
charge, while the nitroso nitrogen might have a partial
negative charge. Consequently it is suggested that the initial
addition of hydrogen adsorbed on the iron surface would
occur at the nitroso nitrogen as shown in eq 1:
Discussion
NDMA can be reduced by both Fe and Ni/ Fe to form DMA
and ammonium, which are nontoxic. The NDMA reduction
rate is significantly enhanced by plating a small amount of
Ni (0.25%) onto the Fe surface. For instance, at 73 PV NDMA
reduction by Ni/ Fe was 340 times faster than that by Fe
(
Figure 1). The much faster reduction rate of NDMA observed
with Ni/ Fe than with Fe can be explained by the proposed
catalytic hydrogenation reaction mechanism since Ni is a
stronger hydrogenation catalyst than Fe (25). As discussed
in detail in Part 2 of this paper (26), this mechanism is
proposed based on comparison of the results obtained from
column and batch experiments to the electrochemical
experiments. In this mechanism, the charge-transfer reaction
does not take place between Fe or Ni and the organic
contaminant but rather between metal and water molecules.
This surface bound intermediate might react further with
another surface adsorbed hydrogen to give N-N bond
cleavage as shown in eq 2 and consequently to produce DMA
and ammonia as final products as observed in this study.
This produces adsorbed atomic hydrogen (FeHads and NiHads
)
at local cathodes (20, 26) which in turn takes part in catalytic
hydrogenation reactions, resulting in NDMA reduction.
Furthermore, atomic H can be absorbed onto Ni and then
diffuse into the metal lattice to form NiHabs in the presence
of P (20). The adsorbed or absorbed atomic H is a very
powerful reducing agent leading to fast NDMA reduction as
evidenced in Figure 1. The use of strong hydrogenation
catalysts such as Pd have also been used to enhance reductive
dehalogenation reactions (21, 27).
Most previous studies of bimetals have been conducted
in batch experiments over short periods of time (hours to
days, e.g. 21, 34). The limited information that is available
indicates a loss of reactivity over time (35, 36). On the other
hand, a gain of reactivity has also been reported (20). Such
a discrepancy may be due to differences in the method of
catalyst preparation, for example the use of acid washed (21)
versus unwashed (20) iron material or to differences in the
The proposed catalytic hydrogenation reaction mecha-
nism can also be used to postulate the NDMA transformation
pathway. Previous studies have proposed several NDMA
transformation pathways (28-31). The first NDMA trans-
formation pathway involves stepwise reductions to form
UDMH and then methane and N (31). The reduction of
2
UDMH across the C-N bond was observed in the presence
of transition metals in UDMH vapor phase at above room
catalyst support such as Al
2 3 3 4 2 3
O (27) versus Fe O or Fe O
(
20). In the 5-month column experiment with Ni/ Fe, we
temperature. Our results showed that no UDMH and CH
4
observed that while NDMA degradation rates with Ni/ Fe are
consistently faster than with Fe, the Ni/ Fe reactivity decreased
over time and the reduction kinetics changed from pseudo-
first-order to mixed-order or zero-order. The loss of reactivity
above background level were detected, indicating that this
pathway did not occur under our experimental conditions.
The formation of a small amount of background hydrocar-
bons can be attributed to the presence of iron carbides and
surface carbon (32).
may have been caused by (1) adsorption of NDMA reduction
+
products (DMA and NH
4
), (2) surface coverage by corrosion
The second NDMA reduction pathway usually occurs
under alkaline conditions and involves two electron transfer
products, and (3) possible loss of galvanic coupling (20).
+
Washing with DI-H
2
O removes adsorbed DMA and NH
4
,
to form DMA and N
2
O without production of ammonia (24,
resulting in a partial recovery of Ni/ Fe reactivity. The adsorbed
2
9). The lack of N O formation indicates that this pathway
2
+
DMA and NH
4
may compete with NDMA for active sites on
did not occur under our experimental conditions. A more
convincing argument can be made using results obtained
from electrochemical reduction experiments (see Part 2 for
details) which showed that reduction of NDMA to DMA and
the metal surface, as suggested by a competition batch
experiment which showed that with an addition of 3 mg/ L
DMA to 25 mg/ L NDMA solution, the reduction rate of NDMA
decreased with both Fe and Ni/ Fe (data not shown). Being
a weak base, DMA is considered to be an inhibitor in catalytic
hydrogenation processes (19). Therefore, despite the small
amount of DMA that was removed during flushing, a
significant portion of the reactivity was restored.
When an acid solution passed through the column, it
dissolved and thus removed some of the corrosion products
that covered the metal surfaces. The high concentration of
total dissolved Fe in the effluent during the acid wash,
particularly at early time (Figure 4) may be attributed, to a
large extent, to the dissolution of the surface films, thereby
making iron and nickel more accessible to the NDMA
molecules in the solution. As a result, we observed a dramatic
improvement in Ni/ Fe reactivity and a reappearance of the
pseudo-first-order kinetics after the acid wash. Muftikian et
al. (37), using X-ray photoelectron spectroscopy, showed a
progressive growth of hydroxylated oxide films on the Pd/ Fe
surface in the course of TCE reduction in aqueous solution.
Upon acid wash using 3 M HCl, the thick surface film was
removed, and the original activity of the Pd/ Fe was nearly
restored. In contrast to Muftikian et al. (37) who reported no
loss of Pd upon acid wash, we observed a small removal of
N
2
O only occurred at potentials as low as -1.3 V (SHE).
The third pathway, which usually occurs under acidic
conditions, involves initial reduction across the NdO bond
to form UDMH (28, 29). Subsequently, UDMH is reduced to
DMA and ammonia (28). Although such a reaction pathway
may occur for a number of N-nitrosamines such as N-
nitrosodiphenylamine (30), UDMH was not detected in our
experiments despite our various efforts to do so. Furthermore,
the batch test using UDMH as the starting material in the
presence of Fe and Ni/ Fe did not result in a significant
reduction of UDMH over a period of three weeks. Alterna-
tively, the initial reduction can occur across the N-N bond
-
-
to form DMA and NO (12). NO may be subsequently
reduced to ammonia through sequential electron transfer
steps (33). Hence both reduction processes ultimately result
in the formation of DMA and ammonia. Formation of the
same products makes it difficult to distinguish these two
reduction processes. Both of the above proposed pathways,
however, do not provide satisfactory answers when the NDMA
structure is carefully inspected. The two methyl groups
attached to the amine nitrogen might result in a shift in charge
VOL. 34, NO. 16, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3 4 9 3

Ni (1%) in the effluent. In the present study, Ni was plated
on the oxide-covered granular iron without prior cleaning.
It is possible that a small percent of Ni was still in cation
form and was incorporated in the surface oxides. Further-
more, we cannot rule out the possibility that Ni would lose
contact with the catalyst support which may undergo
dissolution by acid. This may partially explain the lack of full
recovery of Ni/ Fe reactivity after the acid wash. Further study
is needed to determine the long term use of Ni/ Fe materials.
Shutting the column off for several days resulted in partial
recovery of the Ni/ Fe activity. NDMA solution that was
pumped through the column contained dissolved oxygen,
resulting in the formation of brownish patches at the bottom
of the column. The brownish patches associated with Fe(III)
disappeared during the shut down period, indicating reduc-
tion of the three-valent film on the metal. The reduced iron
products were mainly magnetite as we observed by its
distinctive black color. Magnetite will not prevent hydrogen
evolution (38, 39). There may be other reasons for the
observed gain of reactivity of the Ni/ Fe; however, further
studies are needed to examine the surface changes of the
Ni/ Fe during NDMA reduction in aqueous solution. An
understanding of the gain of reactivity may provide a simple
but useful means to regenerate Ni/ Fe, for example, in
canisters for above ground water treatment.
(15) Weber, E. J. Environ. Sci. Technol. 1996, 30, 716-719.
16) Cheng, I. F.; Muftikian; R.; Fernando; Q.; Korte, N. In Abstracts,
(
2
1
13th National Meeting of the American Chemical Society, April
3-17, Division of Environmental Chemistry, San Francisco,
CA; American Chemical Society: Washington, DC, 1997; pp 165-
66.
1
(17) Gu, B.; Liang, L.; Dickey, M. J.; Yi, X.; Dai, S. Environ. Sci. Technol.
1998, 32, 3366-3373.
(
18) Fiedor, J. N.; Bostick, W. D.; Jarabek, R. J.; Farrell, J. Environ. Sci.
Technol. 1998, 32, 1466-1473.
19) Freifelder, M. In Catalytic hydrogenation in organic synthesis
procedures and commentary; John Wiley and Sons: NY, 1978;
Chapter 5, pp 26-42.
(
(20) Odziemkowski, M. S.; Gillham, R. W.; Focht, R. In Environmental
issues in the electronics/semiconductor industries and electro-
chemical/photochemical methods for pollution abatement; Sim-
pson, C. R., Mendicino, L., Rajeshwar, K., Fenton, J. M., Eds.;
The Electrochemical Society, Inc.: Pennington, NJ, 1998; Vol.
9
8-5, 91-102.
(
21) Korte, N.; Liong, L.; Muftikian, R.; Grittini, C.; Fernando, Q.
Platinum Metals Rev. 1997, 41, 2-7.
(22) Seebach, D.; Wykypiel, W. Synthesis 1979, 423-424.
(
(
(
23) Entwistle, I. D.; Johnstone, R. A. W.; Wilby, A. H. Tetrahedron
982, 38, 419-423.
24) Wilhelm, E.; Battino, R.; Wilcock, R. J. Chem. Rev. 1977, 77,
19-262.
25) In Hydrogen effects in catalysis. Fundamentals and Practical
Applications; Pa a` l, Z., Menon, P. G., Eds.; Marcel Dekker: NY,
1988.
1
2
(
(
26) Odziemkowski, M.; Gui, L.; Gillham, R. W. Environ. Sci. Technol.
2
000, 34, 3495-3500.
Acknowledgments
This research was funded by the NSERC/ Motorola/ ETI
Industrial Research Chair in Groundwater Remediation held
by Dr. R. W. Gillham.
27) Munakata, N.; McNab, W.; Haag, W.; Roberts, P. V.; Reinhard,
M. In Abstracts, 213th National Meeting of the American Chemical
Society, April 13-17, Division of Environmental Chemistry, San
Francisco, CA; American Chemical Society: Washington, DC,
1
997; pp 168-170.
28) Lunn, G.; Sansone, E. B.; Keefer, L. K. Synthesis 1985, 1104-
108.
(
Literature Cited
1
(
1) Haley, T. J. In Handbook of Carcinogens and Hazardous
Substances: Chemical and Trace Analysis; Bowman, M. C., Ed.;
Marcel Dekker: NY, 1982; Chapter 1, pp 1-18.
2) Massey, R. C. In Nitrosamines: Toxicology and Microbiology;
Hill, M. J., Ed.; Ellis Horwood Series in Food Science and
Technology. VCH Publishers: NY, 1988; Chapter 2, pp 16-47.
3) National Primary and Secondary Drinking Water Regulation.
Fed. Regist. 1989, 54, 22062.
(
(
(
(
29) Lund, H. Acta Chem. Scand. 1957, 11, 990.
30) Hayes, B. T.; Stevens, S. T. J. Chem. Soc. (C) 1970, 1088-1089.
31) Logan, S. R. J. Catalysis 1991, 129, 25-30.
32) Deng, B.; Campbell, T. J.; Burris, D. R. Environ. Sci. Technol.
(
1
997, 31, 1185-1190.
(
33) Gootzen, J. F. E.; Peeters, P. G. J. M.; Dukers, J. M. B.; Lefferts,
(
(
L.; Visscher, W.; van Veer, J. A. R. J. Electroanal. Chem. 1997,
4
34, 171-183.
34) Wang, C. B.; Zhang, W. X. Environ. Sci. Technol. 1997, 31, 2154-
156.
35) Sivavec, T. M.; Mackenzie, P. D.; Horney, D. P. In Abstracts,
4) Fleming, E. C.; Pennington, J. C.; Wachob, B. G.; Howe, R. A.;
Hill, D. O. J. Hazard. Mater. 1996, 51, 151-164.
(
(
2
(
(
(
(
5) The Merck Index, 12th ed.; Budavari, S., Ed.; 1996.
6) Mills, A. L.; Alexander, M. J. Environ. Qual. 1976, 5, 437-440.
7) Mosier, A. R.; Torbit, S. J. Environ. Qual. 1976, 5, 465-468.
8) Sen, N. P.; Baddoo, P. A.; Water, D.; Boyle, M. Inter. J. Environ.
Anal. Chem. 1994, 56, 149-163.
2
1
13th National Meeting of the American Chemical Society, April
3-17, Division of Environmental Chemistry, San Francisco,
CA; American Chemical Society: Washington, DC, 1997; pp 83-
5.
8
(
9) Jenkins, S. W. D.; Koester, C. J.; Taguchi, V. Y.; Wang, D. T.;
Palmentier, J.-P. F. P.; Hong, K. P. Environ. Sci., Pollut. Res.
(
(
(
(
36) Korte, N.; West, O.; Liang, L.; Pelfrey, M. J.; Houk, T. C. Federal
Facilities Environ. J. Autumn 1997, pp 105-114.
37) Muftikian, R.; Nebesny, K.; Fernando, Q.; Korte, N. Environ. Sci.
Technol. 1996, 30, 3593-3596.
38) Pourbaix, M. In Lectures on Electrochemical Corrosion; Plenum
Press: NY, 1973.
39) Marcus, P.; Protopopoff, E. In Electrochemical Surface Science
of Hydrogen Adsorption and Absorption; Jerkiewicz G., Marcus
P., Eds.; The Electrochemical Society Inc.: Pennington, NJ, 1997;
Vol. 97-16, pp 211-224.
1
995, 2, 207-210.
(
(
(
10) Guttenplan, J. B. In Genotoxicology of N-nitroso compounds;
Rao, T. K., Lijinsky, W., Epler, J. L., Eds.; Plenum Press: NY,
1
984; Chapter 4, pp 59-90.
11) Leach, S. In Nitrosamines: Toxicology and Microbiology; Hill,
M. J., Ed.; Ellis Horwood Series in Food Science and Technology.
VCH Publishers: NY, 1988; Chapter 4, pp 69-87.
12) Larson, R. A.; Weber, E. J. In Reaction Mechanisms in Envi-
ronmental Organic Chemistry; Lewis Publishers: 1994; Chapter
3
, pp 169-215.
13) Gillham, R. W.; O’Hannesin, S. F. Ground Water 1994, 32, 958-
67.
14) Matheson, L. J.; Tratnyek, P. G. Environ. Sci. Technol. 1994, 28,
045-2053.
(
(
Received for review August 18, 1999. Revised manuscript
received May 10, 2000. Accepted May 17, 2000.
9
2
ES9909778
3
4 9 4
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 16, 2000