complex branching mechanisms may occur following the
oxidation of UDMH, we cannot conclude with certainty that
the proposed UDMH reaction scheme is the only pathway
that could account for NDMA formation during chlorami-
nation of dimethylamine. Such conclusions await detailed
mechanistic studies. However, we conclude that the pre-
ponderance of several lines of research outlined below
indicate a strong correspondence between NDMA, dimeth-
ylcyanamide, and dimethylformamide formed by UDMH
oxidation and that formed by chloramination of dimethyl-
amine. We propose that the following scheme may provide
a more useful framework for understanding NDMA formation
during chlorination of water and wastewater than the current
focus on reactions involving nitrite.
We hypothesize that the formation of UDMH is the rate-
limiting step within the whole scheme at pH values less than
12. The observation that the NDMA formation rate is
positively correlated with monochloramine concentration
is consistent with a UDMH formation step being rate-limiting.
In contrast, the oxidation of UDMH by Cl[I], the next step
in the reaction sequence, was observed to be complete in
less than 10 min under conditions used in our experiments.
Had later oxidation steps been rate-limiting, an early drop
in Cl[I] concentrations would have been observed when
monochloramine was mixed with dimethylamine due to Cl-
[I] consumption during the rapid formation of UDMH; such
a decrease was not observed. The observation that increasing
bicarbonate buffer concentration does not diminish NDMA
formation indicates that the rate-limiting step does not
involve the NH• radical typically observed during one-
electron-transfer reactions; this evidence permits a two-
electron-transfer step to be rate-limiting, such as the step
involved in UDMH formation.
Evidence obtained in this study rules out significant
NDMA formation during chlorination via two potential
formation pathways from dimethylamine: a pathway in-
volving reactions occurring during breakpoint chlorination
and
a pathway involving nitrite, dim ethylam ine, and
After UDMH formation, previous research indicates that
the next step is the oxidation of UDMH to the ionic
intermediate dimethyldiazene (DMD) (15). This oxidation
step, unlike UDMH formation, can occur with oxidants other
than monochloramine or chlorinated dimethylamine. Both
the molecular and ionic forms of UDMH can be oxidized by
monochloramine [Figure 2, K7 ) 10-7.1, k8 ) 3.1 × 10-2 M-1
s-1, k9 ) 8.72 × 103 M-1 s-1 (15)] or oxidants including cupric
halide (23) and hydrogen peroxide (14). In our experiments,
oxidation of UDMH by hypochlorous acid and hydrogen
peroxide both resulted in NDMA formation. Because oxida-
tion of protonated UDMH is much faster than molecular
UDMH, UDMH is more stable at high pH values.
Prior papers indicate that further oxidation of the diazene
intermediate leads to various products favored under different
conditions (Figure 2). Under conditions of high UDMH
concentration, the diazene could dimerize to form tetra-
methyltetrazene (TMT) in a base-catalyzed reaction (15). As
the pH increases, formaldehyde monomethylhydrazone
(FMMH) could form (24). At high pH and high UDMH
concentrations, formaldehyde dimethylhydrazone (FDMH)
formation would be favored (24). Similar to the situation
with UDMH, FMMH, and FDMH were not analyzed during
these experiments (14). Finally, NDMA formation should be
maximized where other products are not favored: at low
UDMH concentrations and lower pH, but not so low that
UDMH formation is hindered.
The observation of NDMA formation from monochloram-
ine and dimethylamine being maximized near a pH value of
8.0 is consistent with the UDMH pathway. At pH values below
8.0, a slower rate of UDMH formation would limit NDMA
production, while at higher pH values, other products of
UDMH oxidation, such as formaldehyde dimethylhydrazone,
may be favored. Indeed, when dimethylamine is reacted with
monochloramine at pH ) 11.6 (where UDMH forms rapidly
and stably) and then lowered to pH ) 6.8 (where UDMH is
rapidly oxidized to a variety of products including NDMA)
just prior to quenching the reaction, the NDMA yield is higher
than for the same reaction carried out continuously at either
pH ) 6.8 or pH ) 11.6 (Table 1, experiment 4).
The formation rates of NDMA and dimethylcyanamide
are positively correlated with monochloramine concentration
during reaction with dimethylamine but remain fairly
constant with increasing dimethylamine concentrations. The
formation rate of dimethylformamide is not strongly cor-
related with monochloramine or dimethylamine. In all of
our experiments, NDMA, dimethylcyanamide, and dimeth-
ylformamide were minor products (<20% yield with respect
to total Cl[I] loss). Similar to the situation with changes in
pH, changes in monochloramine and dimethylamine con-
centrations may favor other products due to competing
monochloramine. NDMA formation rates measured at
chlorine doses in excess of the breakpoint were similar to
those observed during chlorination of dimethylamine in the
absence of ammonia; the breakpoint simply removed am-
monia from solution and NDMA formation occurred through
reaction of hypochlorite and dimethylamine. Furthermore,
addition of nitrite to a solution of hypochlorite, ammonia,
and dimethylamine resulted in a decrease in NDMA forma-
tion rate (Table 1, experiment 1G), probably because Cl[I]
rapidly oxidized nitrite (10). The opposite would be expected
if nitrite were involved in NDMA formation during chlorina-
tion as proposed by previous researchers (8, 9).
The relative NDMA formation rates from different pre-
cursors support formation via the UDMH pathway. The ideal
precursors for UDMH formation, ammonia, and dimethyl-
amine, resulted in far higher concentrations of NDMA than
did chlorination of other potential precursors. NDMA was
formed at an order of magnitude lower rate when mono-
chloramine and chlorinated dimethylamine were mixed, as
these two compounds would not react via nucleophilic
substitution. NDMA was not observed when ammonia was
chlorinated in the absence of amine or when hydrogen
peroxide was added to a mixture of ammonia and dimethyl-
amine. Therefore, chlorine, rather than a general oxidant, is
necessary for NDMA formation. Hypochlorite addition to
dimethylamine alone and monochloramine addition to
trimethylamine resulted in over an order of magnitude slower
rate of NDMA formation relative to chlorination with the
ideal precursors. Formation of NDMA from alternate pre-
cursors is further addressed below.
Prior papers have provided details of UDMH formation
rates under a variety of conditions. Following formation of
monochloramine, UDMH can be produced through a two-
electron-transfer nucleophilic substitution and oxidation
reaction [Figure 1, k6 ) 0.081 M-1 s-1 (21)]. Alternatively,
chlorinated dimethylamine (CDMA) can react with ammonia
to form UDMH, but at a slower rate (21). In the present
research, the formation rate of NDMA from the reaction of
chlorinated dimethylamine with ammonia was approxi-
mately half that measured from the reaction of dimethyl-
amine with monochloramine. UDMH formation from reac-
tion of dim ethylam ine and m onochloram ine is base-
catalyzed above a pH of 12 (21). At lower pH values, the
UDMH formation rate is even slower than the noncatalyzed
rate due to the reduction in the fraction of amine or ammonia
in the active, deprotonated form [Figure 1, K3 ) 10-10.72 (5)].
Additionally, at pH values below 7, monochloramine reacts
with protonated dimethylamine to form chlorinated di-
methylamine [Figure 1, k5 ) 0.21 M-1 s-1 (22)]. As noted
above, chlorinated dimethylamine could then react with
ammonia to form UDMH, but at a much slower rate.
9
VOL. 36, NO. 4, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 5 9 3