Photocatalytic Conversion of Gaseous Nitrogen Trichloride into

Available ChlorineExperimental and Modeling Study

F. Ge rardin,*
,
A. Cloteaux,
,
M. Guillemot,

M. Faure,
,
and J. C. Andre

Institut National de Recherche et de Se curite , Rue du Morvan, CS60027, 54519 Vanduvre Cedex, France

Laboratoire Re actions et Ge nie des Proce de s, UPR 3349 CNRS, 1 rue Grandville BP20451, 54001 Nancy Cedex, France
*S Supporting Information
ABSTRACT: In water, chlorine reacts with nitrogen-containing compounds to produce disinfection byproducts such as nitrogen
trichloride which induces ocular and respiratory irritations in swimming pool workers. A technical solution has been used to reduce NCl
3
exposure to acceptable levels, by adding a stripping step to the water recycling loop. The pollutants extracted are currently rejected into
the atmosphere without treatment. However, the physical properties of NCl
3
could be harnessed to induce its controlled degradation by
direct or indirect light. This paper describes the way to transform NCl
3
into oxidizing chlorine by photocatalysis under laboratory
conditions. Photocatalytic oxidation eciently degrades gaseous nitrogen trichloride, producing compounds such as HClO. About 60% of
NCl
3
decomposed was converted into HClO which could be used as a disinfection compound. A kinetic model is proposed for the
photocatalytic process based on a convection/diusion model. The LangmuirHinshelwood model was applied to the chemical part of
the mechanism. The apparent quantum yield was also estimated to assess the optimal irradiance for NCl
3
transformation. The results
show that photocatalysis performs much better than photolysis alone for NCl
3
removal, i.e. at least 25 times more ecient.

INTRODUCTION
Nitrogen-containing compounds (urea, ammonia, amino acids,
etc.) are provided in pools by swimmers and react with chlorine
which is widely used in water as a disinfection compound.
1
The nitrogen-containing molecules are progressively degraded
by chlorine giving rise to components such as haloforms,
aldehydes, and chloramines,
2
of which the most halogena-
ted form, nitrogen trichloride (trichloramine, NCl
3
), is
very volatile.
3
Nitrogen trichloride induces ocular and
respiratory irritations in lifeguards and other swimming pool
workers.
46
Received: February 5, 2013
Revised: April 3, 2013
Accepted: April 6, 2013
Published: April 8, 2013
Article
pubs.acs.org/est
2013 American Chemical Society 4628 dx.doi.org/10.1021/es400588m | Environ. Sci. Technol. 2013, 47, 46284635
To reduce NCl
3
exposure to acceptable levels a stripping step
was added to the recycling loop of water treatment.
7,8
This
process provides an interesting response in terms of both
occupational and public health, but in environmental terms it is
not entirely satisfactory. The nitrogen trichloride extracted by
this process should be treated before rejecting it into the
atmosphere, this would make the technique acceptable to
employees, facility owners, and citizens alike.
Alternatives to adsorption processes are possible thanks to
the physical properties of NCl
3
which should make it possible
to induce controlled degradation of this compound by
irradiation with direct or indirect light.
Photocatalysis is a recently developed, eective, relatively
cheap technique which is becoming more commonly used to
treat a large range of pollutants that can be decomposed into
mineral components.
9
Photocatalysis could provide an original
means to break down gaseous NCl
3
into nitrogen and chlorine,
thus contributing to the formation of available chlorine, i.e.,
gaseous chlorine, or hypochlorous acid, with its well-known
bactericidal properties.
Various authors have studied gaseous nitrogen trichloride
photodissociation.
10,11
Their results showed that NCl
3
could be
broken down using short-wave UV light irradiation. Indeed, the
absorption spectrum for this compound revealed an absorption
maximum at 220 nm. The free radical based photolytic
mechanism for NCl
3
breakdown suggested by Gilbert et al.
10
at 249 or 308 nm leads to the formation of Cl
2
, which itself
absorbs UV light irradiation to varying extents between 250 and
350 nm, producing free Cl

atoms. These radicals also can

enhance NCl
3
decomposition, leading to a quantum yield for
NCl
3
degradation greater than one.
12
The main drawback of this process is that NCl
3
only weakly
absorbs the wavelengths emitted by the most common
medium- or low-pressure mercury vapor lamps (Supporting
Information Figure S1).
13
The combination of the low energy
yield of these lamps with the low absorption by NCl
3
at the
wavelengths emitted makes this process not entirely econom-
ically viable for most swimming pools.
This paper describes the way to decompose NCl
3
by
heterogeneous photocatalysis in an annular reactor under
laboratory conditions. This process allows very ecient
conversion of gaseous nitrogen trichloride into recoverable
compounds such as gaseous chlorine and hypochlorous acid. A
kinetic model is proposed for the photocatalytic process based
on a convection/diusion model. The monomolecular/
monolayer LangmuirHinshelwood (LH) model was tested
for the chemical part
14
and the global quantum yield was also
estimated.

EXPERIMENTAL SECTION
Generation of Gaseous Nitrogen Trichloride. Nitrogen
trichloride is formed in aqueous solution by the action of chlorine,
in a +1 or 0 oxidation state, on nitrogen-containing materials. The
kinetics and reaction mechanism are known for the reaction of
sodium hypochlorite with ammonia or ammonium ion.
15,16
For this study, nitrogen trichloride was synthesized in an
open, continuous stirred-tank reactor (CSTR) (Figure 1). A
hypochlorite solution and a solution of ammonium sulfate
((NH
4
)
2
SO
4
) were supplied continuously to the reactor using a
peristaltic pump.
17
The hypochlorite solution was adjusted to
12 mol m
3
from a stock solution of bleach (NaClO, Merck,
614% active chlorine). The ammonium sulfate solution (4
10
1
mol m
3
), was prepared from ammonium sulfate crystals
(Merck, > 99.5%, A.C.S. reagent). The pH of these two
solutions was adjusted to 4 using phosphoric acid (Fluka, 85%).
This stabilized the NCl
3
produced, and enhanced the yield of
this compound compared to the other chloramine forms.
15
The
chosen reagent concentrations provide a Cl/N ratio of 15. The
outlet ow from the CSTR was carried toward a stripping
column fed with dry air. The Henrys constant for NCl
3
is 435 at
20 C.
3
About 95% of the nitrogen trichloride was transferred
from solution to the air. Residual species such as monochloramine,
dichloramine, or hypochlorous acid present in the stripping ow
were trapped in a solution of sulfamic acid.
5
The outlet air ow
from this step, loaded with nitrogen trichloride, was then mixed
with dry air which had been humidied in variable proportions to
adjust the relative humidity (RH) of the system to a target value.
RH ranging from 1 to 85% can be obtained.
Photoreactor. The experiments described here were
performed using a test bench equipped with an adaptable
photoreactor to study both the photolytic and photocatalytic
processes (Figure 1).
The annular plug-ow reactor consists of two concentric boro-
silicate glass cylinders (
internal
= 9.05 10
2
m and
external
=
0.1 m, L = 1.2 m, and V = 1.71 10
3
m
3
). The hydraulic
diameter d
h
is 9.5 10
3
m. The gaseous euent ows through
the interannular space and the length of the reactive area is L
r
=
1.0 m. Axial dispersion of NCl
3
in the reactor was veried by
Figure 1. Experimental setup.
Environmental Science & Technology Article
dx.doi.org/10.1021/es400588m | Environ. Sci. Technol. 2013, 47, 46284635 4629
calculating the specic Peclet number for each ow value. The
relation used to determine the Peclet number is dened for a
tubular reactor as described in 18.
=
*
P
U L
D
m r
(1)
where
* = + D D
U d
D 192
m
m
2
h
2
m
(2)
U
m
and D
m
respectively represent the mean ow velocity in the
reactor (m s
1
) and the molecular diusion coecient of NCl
3
(D
m
= 9.6 10
6
m
2
s
119
). D* is the axial dispersion coecient
(m
2
s
1
) calculated from the Taylor equation (eq 2). It takes into
account the dispersion of the compound under the simultaneous
action of molecular diusion and variation of the velocity of the
gaseous euent. Axial dispersion is considered negligible for
annular photoreactor with P > 100.
18
Illumination was provided by a UV-A lamp (Sylvania
Blacklight F40W/350 BL) placed at the center of the inner
cylinder. UV-A irradiance was measured with a light detector
(Gigahertz Optik radiometer model X11-XD-9511, range 315/
400 nm). The detector head is tted with the cosine diuser to
receive the incoming light signal. The detector allows
polychromatic measurements with uniform-wavelength-de-
pendence in its response to incident radiation. The maximum
wavelength for the lamp used was 350 nm, and 99% of the
power emitted ranged from 315 to 388 nm (Figure S1). The
UV-A unit photon ow of the lamp is F
0
= 3.7 10
5
E s
1
(or mol of photons s
1
), e.g. 12.8 W. The conversions between
E s
1
and W were made by integrating the spectrum of the
lamp. The irradiance measured at the outer surface of the inner
tube was I
0
= 34 W m
2
.
The reactor was tted with ten sampling openings to
monitor the concentrations of the dierent species over the
entire reactive area. The reactor temperature was monitored
and it remained constant during the experiment.
Some authors have shown that nitrogen trichloride breaks
down after UV light absorption.
1012
As photocatalysis with
TiO
2
is also due to UV-A light irradiation with wavelength
< 388 nm,
9
the photolytic and photocatalytic processes of
NCl
3
decomposition were studied separately. The rst part of
this work focuses on the photolytic eect, while the second part
describes the heterogeneous photocatalysis of nitrogen
trichloride. All the experiments were carried out in laboratory
conditions, i.e., with synthetic NCl
3
.
Reactor Conguration for Photolysis. Photolysis of
nitrogen trichloride was studied in the basic reactor
conguration, i.e., in homogeneous phase, with the following
operating conditions:
Inlet ow rate: Q
0
= 1.35 10
4
m
3
s
1
Mean velocity: U
m
= 0.095 m s
1
, D* = 4.33 10
4
m
2
s
1
,
and specic Peclet number P = 219
Residence time: = 10.5 s
Concentration at the reactor inlet: C
0
= 2 10
4
mol m
3
Temperature in the reactor: T = 27 C
Total pressure: 10
5
Pa
Relative humidity: RH = 80%
Irradiance: 9.2, 22.8, and 34 W m
2
.
The nitrogen trichloride concentration and relative humidity
used in these experiments were based on the average values
measured at the outlets of existing stripping devices.
7,8
Reactor Conguration for Photocatalysis. The 250-m
thick photocatalytic support used in this study consisted of cellulose
bers coated with 18 g m
2
TiO
2
(Ahlstrom 1048 PC 500
Millennium, 350 m
2
g
1
specic surface, anatase form; 20 g m
2
SiO
2
, 2 g m
2
zeoliths). The physical adsorption of gas molecules
on the catalyst surface is characterized by the determination of
BrunauerEmmettTeller (BET) surface area. For TiO
2
, It was
measured by N
2
adsorption and was found to be 317 m
2
g
1
; the
median diameter of deposited pellets on the medium is 1.4 m.
20
The outer side of the inner tube was covered with the photo-
catalytic medium to produce a reactive zone of length 1.0 m,
with a catalytic surface of 0.286 m
2
and a total of 5.1 g of TiO
2
.
Photocatalysis of nitrogen trichloride was performed in the
same conditions as the photolysis study, but three NCl
3
concentrations were tested. These concentrations at the reactor
inlet were C
0
4.5 10
5
/10
4
/2 10
4
mol m
3
.
Sampling Methods and Analyses. Samples of nitrogen
trichloride and chlorinated compounds were taken at the inlet
and outlet of the reactor, and at each sampling point along the
reactors length. Sampling and analyses were performed by the
specic method described by He ry et al.
6
The sampling device
was composed of two successive parts, i.e. a tube containing
silica gel coated with sulfamic acid and a cassette containing
quartz lter soaked with a solution of sodium carbonate and
diarsenic trioxide. This allows separate assessment of nitrogen
trichloride and all oxidizing chlorine species without distinction,
including hypochlorous acid, dichlorine, monochloramine, and
dichloramine. The sampling time was 15 min. Sampling was
started 90 min after the experiment was initiated on the test bench.
Byproducts of degradation were identied using a range of
analytical techniques. The rst of these relies on samples taken
using multibed supports (Carbotrap 300). These sampling
supports contain several successive adsorbent layers. Sampling
was performed over approximately 120 min at 100 mL min
1
.
Compounds trapped on these tubes were analyzed by thermal
desorption (Perkin-Elmer ATD 400) coupled to gas phase
chromatography (Perkin-Elmer Clarus 500) and mass spec-
trometry (Perkin-Elmer Turbo mass Gold). This technique
allows identication fo hydrochloric acid and many other
compounds. The analytical conditions were
Thermal desorption: Tube heated at 320 C during 20 min
with 50 mL min
1
of helium.
Gas chromatography:
Column: GS-Gaspro plot, 60 m diameters 0.32
mm/0.53 mm)
Helium ow rate =1.4 mL min
1
Oven: 15 min at 30 C, ramp at 10 C min
1
,
5 min at 220 C.
Mass spectrometry:
Scan mod with 20 < m/z < 100
Source temperature: 200 C.
Monochloramine and dichloramine compounds were
trapped by passing the outow gas through a bubbler
containing 40 mL of carbon tetrachloride (CCl
4
) with a ow
rate of 2 L min
1
for 60 min. The contents were then analyzed
by UV spectrometry (Perkin-Elmer Lambda 950) according to
the protocol proposed by Czech et al.
21
Hypochlorous acid and/or dichlorine contained in the
reactor exhaust were identied by absorption in two bubblers in
series, each containing 60 mL of ultrapure water. Sampling was
performed over 45 min with a ow rate of 300 mL min
1
. The
solution was then analyzed by N,N-diethyl-paraphenylenediamine
Environmental Science & Technology Article
dx.doi.org/10.1021/es400588m | Environ. Sci. Technol. 2013, 47, 46284635 4630
(DPD) colorimetry.
22
As this technique cannot distinguish
between hypochlorous acid and Cl
2
, the results obtained represent
the total concentration for both compounds.
Assays for chloride ions that may be present at the surface of
the photocatalytic support were performed after photocatalytic
degradation of NCl
3
. Ions were desorbed from samples of the
photocatalytic medium by washing with 20 mL of ultrapure
water. Chloride ions were identied in an aliquot of wash
solution using a colorimetric method.
23

RESULTS AND DISCUSSION

Photolytic Degradation. Tests were performed to assess
the removal rate (X) for NCl
3
at dierent irradiances.
=

X
C C
C
100
in out
in
(3)
C
in
and C
out
are, respectively, the inlet and outlet NCl
3
concentration (mol m
3
). Irradiances measured at the outer
side of the inner tube in the direction of the lamp of 9.2, 22.8,
and 34 W m
2
were used. The results of these tests are shown
in Figure 2, illustrating how nitrogen trichloride breaks down
after absorbing UV-A irradiation.
A nonlinear relationship is observed between the NCl
3
removal
rate and the irradiance, I. Some authors have suggested reaction
mechanisms that may explain these experimental observations.
1012
The decomposition reaction may involve a step by step
mechanism, or even a more complex process involving branching
chain decomposition.
24
Additional studies will be necessary to
determine the precise nature of the reaction mechanism observed.
However, based on our experimental results, it is possible to suggest
the following reaction mechanism.
The absorption of a photon by NCl
3
leads to the formation
of NCl
2

and Cl

radicals
+ +

h NCl NCl Cl
3 2
(4)
Thermodynamically, if the rst step corresponds to the
creation of an NCl
2

radical and an atom of chlorine (eq 4), it

is possible that the latter can react with NCl
3
to generate a
second NCl
2

radical (eq 5) according to the mechanism

suggested by Briggs and Norrish:
12
+ +

Cl NCl NCl Cl
3 2 2
(5)
+ + +

NCl NCl N 2Cl Cl
2 3 2 2
(6)
+ + +

Cl NCl X NCl X
3 4
(7)
+

2NCl N 4Cl
4 2 2
(8)
X and X are unidentied products.
The photon can thus be absorbed by a single NCl
3
molecule
but will ultimately lead to degradation of six NCl
3
molecules. It
is, however, possible to suggest a branching chain decom-
position, complementary to steps 5, 6, 7, and 8:
24
+ +

2NCl N Cl 2Cl
2 2 2
(9)
+ +

Cl NCl NCl Cl
3 2 2
(10)
It must be remembered that once the reaction has proceeded
for some removal rate, the Cl
2
concentration is no longer
negligible in the reactor. This Cl
2
can then play a part in the
process. The absorption of photons by gaseous chlorine leads
to the formation of free Cl

atoms.
+

h Cl 2Cl
2
(11)
These reactive species then react with NCl
3
according to
reactions 5 and 7 to increase NCl
3
decomposition through the
mechanism proposed by Briggs and Norrish.
12
With regard to byproducts, the dierent analytical techniques
presented above did not reveal the presence of other chlorine
species. Dichlorine is trapped, in part, in bubblers containing
ultrapure water as hypochlorous acid.
Photocatalytic Degradation. The annular tubular photo-
catalytic reactor can be modeled based on three main
assumptions. The rst is to consider the reactor as a steady-
state reactor, i.e. inlet and outlet parameters are not time-
dependent. The second is that the radial velocity prole in
the annular space is fully developed in laminar conditions.
All experiments were performed at Reynolds number Re =
60 with d
h
as characteristic length. The third assumption
is that the internal mass transfer into the catalyst is
negligible.
20
From these assumptions, a two-dimensional (2D) model is
proposed. This model is based on a convection/radial diusion
principle in the annular space of the reactor. The mass balance
in the gas phase can be written as eq 12:

u r
C
z
D
C
r r
C
r
( )
1
m
2
2
(12)
where r is radial coordinate (m), z is axial coordinate, u is
ow velocity (m s
1
), and C is the concentration of NCl
3
(mol m
3
).
The model takes the velocity prole developed in the annular
space into account, as expressed by the following relationship:
25

=
+
+

u r
U r
r
r
r
( )
2
1
1 ln
m
2
e
2
e
2
(13)
with
=
r
r
i
e
(14)

( )
1
2ln
2
1
(15)
where r
i
is the internal radius (m) and r
e
is the external radius of
reactor (m).
Figure 2. Photolytic removal rate at 1 m vs irradiance.
Environmental Science & Technology Article
dx.doi.org/10.1021/es400588m | Environ. Sci. Technol. 2013, 47, 46284635 4631
Boundary conditions are as follows:
at the inner surface of the external cylinder, r = r
e
:

D
C
r
0
r
m
e
(16)
at the catalyst surface, r = r
i
,
=

r D
C
r
r
p m
i
(17)
where r
p
is the photocatalytic rate (mol m
2
s
1
).
The kinetics of NCl
3
degradation at the surface of the catalyst
relies on a monomolecular and monolayer LangmuirHinshel-
wood model.
14
Initially, the internal material transport of NCl
3
into the
catalyst is considered to be nonlimiting with respect to the
reaction process. Some authors have shown that the internal
material transport into the pores of each TiO
2
grain in the
medium used in tests can be neglected.
20
The irradiance is
taken into account in the kinetic law expression (eq 18) and
varies between 0.09 and 9.2 W m
2
to limit NCl
3
photolysis.
Indeed, with an irradiance of 9.2 W m
2
, the rate of photolytic
NCl
3
decomposition between the inlet and the outlet of
the reactor is less than 2% for the three concentrations tested.
NCl
3
photolysis in the photocatalytic medium was therefore
neglected.
With relatively constant oxygen concentration and relative
humidity, considering that the products formed are only
partially and noncompetitively absorbed by the catalyst, the
monomolecular/monolayer LH model describing the kinetics
of NCl
3
decomposition is
=
+
r I k
K C
K C 1
n
p p
p
p (18)
where I is the irradiance (W m
2
), n is the order with respect
to I, and k
p
and K
p
are the LH rate constant (mol s
1
W
1
)
and the adsorption constant (m
3
mol
1
), respectively.
Figure 3 shows how NCl
3
degradation evolves in the reactor with
an irradiance of 9.2 W m
2
and initial NCl
3
concentrations of
2 10
4
(case 1), 1.1 10
4
(case 2), and 4.5 10
5
molm
3
(case 3). The compound almost completely disappears with a
reactor length of around 0.6 m or longer. Photocatalysis is very
eective compared to photolysis (Figure 2). It must be
remembered that, with an irradiance of 9.2 W m
2
, the
contribution of photolysis to the degradation process is negligible
in the cases presented here. In addition, the results indicate that
the convection/diusion/reaction model presented is quite well
adapted to the process for all the concentrations tested. The
removal rate (Figure 3b) seems not really dependent on the initial
concentration of NCl
3
.
Tests were performed to assess how irradiation of the catalyst
aects the NCl
3
removal rate (Figure 4a and b). The NCl
3
concentration used for these tests was 2 10
4
mol m
3
, while
the irradiance varied between 0.09 and 9.2 W m
2
. The kinetics
of nitrogen trichloride decomposition depends on the
irradiance emitted, and how many photons are absorbed by
the catalyst. Figure 4c represents the experimental and modeled
rates at 0.25 m for dierent irradiances and shows that two
kinetic phenomena are apparent. Both of these are well predicted
Figure 3. Evolution of NCl
3
concentration in the annular space of the reactor (a) and the corresponding removal rate (b).
Environmental Science & Technology Article
dx.doi.org/10.1021/es400588m | Environ. Sci. Technol. 2013, 47, 46284635 4632
by the model. The results indicate that for an irradiance of greater
than about 0.02I
0
, e.g. 0.6 W m
2
, the kinetics is only aected by
radial diusion. The ux of photons absorbed by the catalyst does
not limit NCl
3
photocatalysis. For this range of light intensities, the
process is considered to be of 0 order relative to I. However,
chemical reactivity is apparent from an irradiance below 0.2
Wm
2
. The constants of the LH model described above were
determined from these experiments by the parametric optimiza-
tion based on the least-squares method. Thus, n = 1, k
p
= 2.6
10
6
mol s
1
W
1
, and K
p
= 3 10
3
m
3
mol
1
. The LH model
constants determined here only apply to the Ahlstrom 1048
photocatalytic medium. These results show that NCl
3
photo-
catalysis is very eective relative to photolysis. For example, for an
irradiance of 0.6 W m
2
, the degradation yield is over 95% in the
presence of the photocatalytic medium with a reactor length of 0.5
m. In these conditions, photolysis does not function at all.
The eect of humidity on the photocatalytic degradation of
NCl
3
was studied with the same ow-rate as in the studies above,
with an irradiance, I, of 9.2 W m
2
. The NCl
3
concentration at the
reactor inlet was approximately 2 10
4
mol m
3
. A relative
humidity of 24% was used in this experiment. In this case,
degradation was signicantly less ecient than with a RH = 78%
(Figure 5). Indeed, the relative humidity is a key parameter in the
photocatalytic process as it is the main source of OH

radicals.
9
Although an excess of humidity can, in some cases, inhibit the
reaction process by competing for adsorption with the pollutant to
be degraded, it contributes in the tests presented here to increasing
the reaction speed. Thus, the OH

concentration, which is
considered constant, was integrated into the kinetic constant k
p
of
the LH model presented above.
The overall apparent quantum yield for photocatalysis is
dened by
=
r
F
p
a
(19)
where is the overall quantum yield for photocatalysis, r
p
is the
average photocatalytic rate (mol m
2
s
1
), and F
a
(E m
2
s
1
) is the
light ux absorbed by the catalyst, with

=

F F d
a
300
388
a
(20)
where F
a
is the light ux absorbed by the catalyst (E m
2
s
1
nm
1
)
for a given wavelength, .
=

F
I
N hc
a
a
A
(21)
Here, I
a
is the absorbed ux (W m
2
) for a given wavelength, , N
A
is Avogadros number (mol
1
), h is Plancks constant (j s
1
), and c is
the speed of light (m s
1
). I
a
was calculated from light irradiation
and optical properties of materials.
26,27
About 10% of the light
emitted was absorbed by the Pyrex tube; the TiO
2
on the medium
absorbed 62% of photons emitted.
26,27
These values were integrated
for 300 nm < < 388 nm.
Figure 4. Inuence of irradiance on NCl
3
decomposition case 1: I = 9.2 W m
2
; case 2: I = 0.09 W m
2
.
Figure 5. Inuence of relative humidity on NCl
3
decomposition.
Environmental Science & Technology Article
dx.doi.org/10.1021/es400588m | Environ. Sci. Technol. 2013, 47, 46284635 4633
Given the thickness of the photocatalytic medium, we
considered photons to be homogeneously absorbed in the volume
of medium. The quantum yield for the photocatalytic reaction
indicates that a chain mechanism is not involved. However, the
maximum quantum yield for the process as a whole can be
determined. The maximum quantum yield can be determined
based on the minimum number of photons that must be absorbed
by the photocatalytic support to cause the desired rate of removal.
Table 1 shows the relationship between the overall quantum yield
of NCl
3
degradation and the light ux absorbed by the catalyst.
It appears, in this case, that the maximal overall quantum yield of the
process is close to 0.9. This value was obtained in conditions where
the chemical processes produced in each nanoparticle of TiO
2
,
linked to multiphotonic eects, are negligible.
The results presented in Table 1 suggest that the kinetics of
photocatalysis can be maintained even with a signicantly
reduced irradiation of the light source tted to the reactor, close
to 0.6 W m
2
. However, this observation is applicable only for
the photocatalytic medium tested (Ahlstrom 1048).
Based on the results obtained in these studies, a reaction
mechanism for NCl
3
decomposition can be suggested. The
mechanism of catalyst activation and OH

radical production
described in the literature, involving a single photon of adequate
energy, is as follows:
9
+ +
+
h TiO TiO (e h )
2 2
(22)
+
+
H O OH H
2
(23)
+
+
OH h OH
(24)
+

O e O
2 2
(25)
+
+
O H HOO
2
(26)
+

2HOO O H O
2 2 2
(27)

H O 2OH
2 2
(28)
For the nitrogen trichloride degradation mechanism, we
suggest the following steps:
+ +

NCl OH NCl HClO
3 2
(29)
+

2NCl N 2Cl
2 2 2
(30)
According to this mechanism, the main product is chlorine gas,
although this can be partially transformed in a wet medium into
HClO and HCl, in line with eq 31. The stoichiometry obtained
is three moles of Cl
2
produced (or equivalent) for every two
moles of NCl
3
degraded.
+ + Cl H O HClO HCl
2 2
(31)
The analytical techniques described in Section 2.5 indicate
that the main products of the reaction are HClO (around 60%)
and/or Cl
2
. HClO can be partially broken down by photolysis,
even in the near UV range, according to the mechanism
suggested by Vogt and Schindler:
28
+ +

h HClO OH Cl
(32)
Cl

and OH

radicals can then react with NCl

3
as part of
reactions 10 and 29, or they can form Cl
2
and increase the rate
of NCl
3
elimination.
The presence of gaseous HCl is also conrmed by mass
spectrometry; analysis of the contents of the CCl
4
-containing
bubbler indicates that monochloramine represents about 10%
(molar). Chloride was detected on the lters, representing around
20% (molar) of the initial nitrogen trichloride concentration.
With regard to nitrogen-containing byproducts, the theoreti-
cal mechanism indicates the presence of N
2
. It is, however,
likely that nitrogen oxides (NO and NO
2
) are formed by the
reaction of HClO and OH

with NCl
2

. Given the
wavelengths emitted by the light sources, it is also possible
that traces of ozone are formed from nitrogen dioxide.
The results presented in this study conrm that nitrogen
trichloride can be readily degraded by a photocatalytic mechanism.
Although the precise reaction mechanism has not been fully
elucidated, the photocatalytic process, by improving photon
capture, signicantly increases NCl
3
decomposition compared to
photolysis alone. The convection/diusion/reaction model
proposed in this study predicts the evolution of nitrogen trichloride
concentrations in an annular photocatalytic reactor working under
laboratory conditions. This model could be easily adapted to
design a full scale reactor. The constants of the Langmuir
Hinshelwood kinetic model were also determined for the
photocatalytic support used in the tests described here.
Beyond the ecacy of the process, this work demonstrates
that nitrogen trichloride can largely be transformed into
oxidizing chlorine, mainly in the form of gaseous chlorine
and hypochlorous acid. Nitrogen-containing byproducts were
not extensively characterized, and compounds such as ozone
were not sought in this study.
Thus, taken together, the results presented in this study show
that it is possible to design a device to treat the gaseous euent
from the buer tank of swimming pools. This type of procedure
would make it possible to recycle the treated euent after the
stripping step, by reintroducing the chlorinated byproducts
with known bactericidal properties into the water treatment
circuit. However, complementary studies must be performed to
assess the incidence of potential copollutants such as
chloroform. Further study is also required to characterize the
impact of chloride deposits on catalyst activity.

ASSOCIATED CONTENT
*S Supporting Information
Figure S1. Extinction coecient of gaseous NCl
3
and emission
spectrum of the experimental lamp between 230 and 400 nm.
This material is available free of charge via the Internet at
http://pubs.acs.org.

AUTHOR INFORMATION
Corresponding Author
*E-mail: fabien.gerardin@inrs.fr; phone: +33 383 509 820; fax:
+33 383 502 184.
Notes
The authors declare no competing nancial interest.
Table 1. Overall Apparent Quantum Yield at 0.25 m for Each
Irradiance Tested
irradiance I
(W m
2
)
absorbed light ux F
a
(E m
2
s
1
)
r
p
(mol m
2
s
1
)
( 10
7
)
overall apparent
quantum yield
0 0 0 -
0.1 1.8 10
7
1.7 0.9
0.7 1.3 10
6
3.0 0.2
1.9 3.4 10
6
3.9 0.1
6.3 1.1 10
5
3.3 0.03
9.2 1.6 10
5
3.1 0.02
Environmental Science & Technology Article
dx.doi.org/10.1021/es400588m | Environ. Sci. Technol. 2013, 47, 46284635 4634

REFERENCES
(1) Guidelines for the Safe Recreational Water Environments, Vol. 2,
Swimming Pools and Similar Environments; World Health Organization
(WHO), 2006.
(2) Weaver, W. A.; Li, J.; Wen, Y.; Johnston, J.; Blatchley, M. R.;
Blatchley, E. R., III Volatile disinfection by-product analysis from
chlorinated indoor swimming pools. Water Res. 2009, 43, 33083318.
(3) Holzwartz, G.; Balmer, R. G.; Sony, L. The rate of chlorine and
chloramines in cooling towers. Henrys law constants for flashoff.
Water Res. 1984, 18, 14211427.
(4) He ry, M.; Hecht, G.; Gerber, J. M.; Gendre, J. C.; Hubert, G.;
Blache re, V.; Rebuffaud, J.; Dorotte, M. Exposition aux chloramines
dans les atmosphe res des halls de piscine. INRS - Cahiers de Notes
Documentaires 1994, 156, 285292.
(5) He ry, M.; Hecht, G.; Gerber, J. M.; Gendre, J. C.; Hubert, G.;
Rebuffaud, J. Exposure to chloramines in the atmosphere of indoor
swimming pools. Ann. Occup. Hyg. 1995, 39, 427439.
(6) He ry, M.; Gerber, J. M.; Hecht, G.; Subra, I.; Possoz, C.;
Dieudonne , M.; Andre , J. C. Exposure to chloramines in a green salad
processing plant. Ann. Occup. Hyg. 1998, 42, 437451.
(7) Ge rardin, F.; Gerber, J. M.; He ry, M.; Quenis, B. Extraction de
chloramines par contact gaz/liquide dans les eaux de piscines. INRS -
Cahiers de Notes Documentaires 1999, 177, 2129.
(8) Ge rardin, F.; Gagnaire, F.; Hecht, G.; He ry, M.; Hubert, G.;
Massin, N.; Subra, I. Re duction de lexposition des travailleurs au
trichlorure dazote dans deux secteurs dactivite s par action sur les
proce de s synthe se. INRS - Cahiers de Notes Documentaires 2005,
201, 918.
(9) Alberici, R. M.; Jardim, W. F. Photocatalytic destruction of VOCs
in the gas-phase using titanium dioxide. Appl. Catal., B 1997, 14, 55
68.
(10) Gilbert, J. V.; Wu, X. L.; Stedman, D. H.; Coombe, R. D.
Photolysis of nitrogen trichloride. J. Phys. Chem. 1987, 91 (16), 4265
4269.
(11) Coombe, R. D.; Gilbert, J. V.; Beaton, S. S.; Mateljevic, N.
Photodissociation of gaseous NCl
3
at 193 and 249 nm. J. Phys. Chem.
2002, 106 (36), 84228426.
(12) Briggs, A. G.; Norrish, R. G. The decomposition of nitrogen
trichloride photosensitized by chlorine. Proc. R. Soc. London 1964, 278,
2734.
(13) Clark, T. C.; Clyne, M. A. A. Kinetic mechanisms in nitrogen-
chlorine radical systems. Part 1.- The formation and detection of the
NCl
2
and N
3
free radicals using time-resolved absorption spectropho-
tometry. Trans. Faraday Soc. 1969, 65, 29943004.
(14) Land, E. M. Photocatalytic degradation of NO
X
, VOCs, and
chloramines by TiO
2
impregnated surfaces. Ph.D. Dissertation,
Georgia Institute of Technology, 2010.
(15) Jafvert, C. T.; Valentine, R. L. Reactional scheme for the
chlorination of ammoniacal water. Environ. Sci. Technol. 1992, 26,
577586.
(16) Ozekin, K.; Valentine, R. L.; Vikesland, P. J. Modeling the
decomposition of disinfecting residuals of chloramines. In Water
Disinfection and Natural Organic Matter: Characterization and Control;
Minear, R., Amy, G., Eds.; American Chemical Society: Washington,
DC, 1996; pp 115125.
(17) Ge rardin, F.; Subra, I. Mise au point dune me thode de
pre le vement et danalyse du trichlorure dazote en phase aqueuse.
INRS - Cahiers de Notes Documentaires 2004, 194, 3950.
(18) Villermaux, J. Ge nie de la re action chimique Conception et
fonctionnement des re acteurs, 2nd ed.; Tec & Doc Lavoisier: Paris,
France, 1993.
(19) Perry, R. H.; Green, D. W. Perrys Chemical Engineers Handbook,
7th ed.; McGraw-Hill: New York, 1997.
(20) Assadi, A. A.; Bouzaza, A.; Wolbert, D. Photocatalytic oxidation
of trimethylamine and isovaleraldehyde in an annular reactor:
Influence of the mass transfer and the relative humidity. J. Photochem.
Photobiol., A 2012, 236, 6169.
(21) Czech, F. W.; Fuchs, R. J.; Henry, F. A. Determination of
mono-, di-, and trichloramine by ultraviolet absorption spectrometry.
Anal. Chem. 1961, 33 (6), 705707.
(22) Standard Methods for the Examination of Water and Wastewater,
9th ed.; APHA-AWWA-WEF: Washington, DC, 1995.
(23) Zall, D. M.; Fisher, D.; Garner, M. Q. Photometric
determination of chloride in water. Anal. Chem. 1956, 28, 16651668.
(24) Rubtsov, N. M. Kinetic mechanism and chemical oscillations in
the branching chain decomposition of nitrogen trichloride. Mendeleev
Commun. 2001, 8 (5), 173175.
(25) Van Dam, M. H. H.; Corriou, J. P.; Midoux, N.; Lamine, A. S.;
Roizard, C. Modeling and measurement of sulfur dioxide absorption
rate in a laminar falling film reactor. Chem. Eng. Sci. 1999, 54, 5311
5318.
(26) Faure, M.; Ge rardin, F.; Andre , J. C.; Pons, M. N.; Zahraa, O.
Study of photocatalytic damages induced on E.coli by different
photocatalytic supports (various types and TiO
2
configurations). J.
Photochem. Photobiol., A 2011, 222 (23), 323329.
(27) Andre , J. C.; DAllest, J. F. Industrial photochemistry. VIII: Light
repartition in heterogeneous cylindrical photoreactors. J. Photochem.
1987, 36 (2), 221234.
(28) Vogt, R.; Schindler, R. N. Product channels in the photolysis of
HClO. J. Photochem. Photobiol., A 1992, 66, 133140.
Environmental Science & Technology Article
dx.doi.org/10.1021/es400588m | Environ. Sci. Technol. 2013, 47, 46284635 4635