SERA
United States
Environmental Protection
Agency
National Risk Management
Research Laboratory
Cincinnati, OH 45268
Research and Development
EPA/600/SR-01/054 August 2001
Project Summary
Chlorine Absorption in S(IV)
Solutions
Sharmistha Roy and Gary T. Rochelle
The rate of chlorine (CI2) absorption
into aqueous sulfite/bisulfite [S(IV)] so-
lutions was measured at ambient tem-
perature using a highly characterized
stirred-cell reactor. The reactor media
were 0 to 10 mM S(IV) with pH ranging
from 3.5 to 8.5. Experiments were per-
formed using 20 to 300 ppm CI2in nitro-
gen (N2) or in air. Chlorine absorption
was modeled using the theory of mass
transfer with chemical reaction. Chlo-
rine reacts quickly with S(IV) to form
chloride and sulfate. Chlorine absorp-
tion is enhanced by increasing pH and
S(IV) concentration. The rate constant
for the reaction of CI2 with S(IV) was too
rapid to be precisely measured using
the existing stirred-cell reactor, due to
mass transfer limitations. However, the
most probable value of the rate con-
stant was determined to be 2 x 109 L/
mol$s.
These results are relevant in the
simultaneous removal of CI2, sulfur di-
oxide (SO2), and elemental mercury (Hg)
from flue gas. The developed model
shows that good removal of both CI2
and Hg should be possible with the in-
jection of 1 to 10 ppm CI2 to an existing
limestone slurry scrubber. These re-
sults may also be applicable to scrub-
ber design for removal of CI2 in the pulp
and paper and other industries.
This Project Summary was developed
by the National Risk Management Re-
search Laboratory's Air Pollution Pre-
vention and Control Division, Research
Triangle Park, NC, to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
Mercury (Hg) pollution is an important
problem because of its behavior in the
environment (bioaccumulation) and the
potential for deleterious health effects.
Roughly 85% of anthropogenic Hg emis-
sions are from combustion sources. The
flue gas from these sources contains sul-
fur dioxide (SO2) and hydrogen chloride
(HCI) at much higher concentrations than
the Hg compounds. Aqueous scrubbing
is currently used to remove SO2 and HCI
from these flue gases. It should be pos-
sible to remove Hg by conventional scrub-
bing technologies with the addition of
reagents to produce chlorine (CI2), which
will oxidize the Hg to a more soluble form
through reaction in the mass transfer
boundary layer.
Mercury reacts with CI2 to form mercu-
ric chloride (HgCI2) which is very soluble
and can thus be easily removed through
aqueous scrubbing. The Hg must react
with the CI2 before the CI2 gets reduced
by the dissolved SO2 present as aqueous
sulfite/bisulfite [S(IV)]. S(IV) represents
sulfurs in the +4 oxidation state (sulfite
and bisulfite). Therefore, the kinetics of
the reaction between CI2 and S(IV) needs
to be quantified to ensure that the CI2 will
be available to react with the Hg.
Experimental Apparatus and
Methods
All CI2 absorption experiments were
performed at ambient temperature in the
well-characterized stirred-cell contactor
-------
with Teflon surfaces shown in Figure 1.
Teflon tubing, fittings, and valves were
used for all the connections. Mass flow
controllers are labeled "FC."
Stirred-cell Reactor Apparatus
The stirred-cell contactor allowed gas/
liquid contact, for which mass transfer
properties were known or measured, at a
known interfacial area (A) of 8.1 x 10'3 m2.
The cylindrical reactor had a 0.01 m in-
ner diameter and 0.016 m height. The
reactor vessel consisted of a thick glass
cylinder with Teflon-coated 316 stainless
steel plates sealed to the top and bottom
by thick gasket clamps. Four equally
spaced, Teflon-coated, 316 stainless steel
baffles were welded to the bottom plate.
The baffles were long enough to extend
to the main body of the gas phase. The
bottom plate contained ports for liquid
inlet and outlet. The top plate contained
ports for the gas inlet and outlet, solution
injection, and pH probe. The total volume
of the reactor was 1.295 x 10'3 m3.
The stirred-cell contactor was equipped
with independently controlled Teflon-
coated agitators for gas- and liquid-phase
mixing. The gas inlet was near the center of
the top plate, directly above the gas agi-
tator blade, to ensure that the inlet gas was
properly mixed. Gas and liquid agitation
speeds were measured using a tachom-
eter. The mass transfer coefficients (kg, l\ )
were a function of the agitation rates.
Gas Source and Flow Path
Gas feed was prepared by quantita-
tively mixing 0.1% CI2 (1000 ppm in N2)
with N2. The CI2 cylinder was supplied by
Air Products. The flow rates of all gas
streams were controlled by Brooks mass
flow controllers. The synthesized gas
stream, typically at a flow rate of 1.2 L/
min, was continuously fed to the reactor.
After exiting the reactor, the gas stream
was diluted with house air and continu-
ously analyzed for CI2. An empty 125-mL
Erlenmeyer flask was connected after the
reactor outlet to capture any water vapor
or liquid. Since this flask stayed empty
throughout an experiment, no liquid ex-
ited the reactor through the gas outlet.
When the CI2 concentration to the reactor
was less than 30 ppm, approximately 3 L/
min of dilution air was used. When the
CI2 concentration was greater, 36 L/min
of dilution air was used. The CI2 analyzer
output was connected to a strip chart re-
corder. The flux of CI2 was calculated
from the gas-phase material balance. An
analyzer with an electrochemical sensor
(NOVA Model 540P) was initially used.
Later experiments used ion mobility spec-
trometry (IMS) (Molecular Analytics
AirSentry 10-CI2).
Hood
Figure 1. Stirred-cell Reactor Apparatus.
Analyzer Calibration
The CI2 analyzer was calibrated at the
beginning and end of each experimental
series to check for analyzer drift. There
was essentially no drift for the IMS ana-
lyzer. During calibration, the gas flow rate
was identical to that in an experiment.
Other than bypassing the reactor, the gas
flow path during calibration was the same
as during an experimental run. To cali-
brate the analyzer zero, N2 (without CI2)
was supplied and diluted with house air.
To calibrate the span, the gas flow rates
were adjusted to give different CI2 concen-
trations spanning the range of interest.
Reactor Solution and Analysis
The reactor contained the aqueous
S(IV) solution, ranging from 0 to 10 mM,
used in absorbing CI2. The reactor fluid
volume in a typical experiment was 1.06
x 10'3 m3. Distilled water was first added
to the reactor. For experiments at pH« 4,
the reactor solution was buffered by in-
jecting a stock solution of equimolar suc-
cinic acid/sodium succinate. The buffer
concentration in the reactor ranged from
5 to 50 mM total succinate. The S(IV)
solution was obtained by injecting a stock
solution containing equimolar sodium
sulfite and sodium bisulfite. For experi-
ments at pH > 7, stock solutions of only
sodium sulfite were used. Liquid samples
were periodically taken from the bulk of
the reactor and analyzed for S(IV) con-
centration by iodometric titration, and some
samples were analyzed for chloride us-
ing ion chromatography. The pH of the
bulk reactor solution was continuously
monitored and recorded using a strip
chart recorder. In the buffered experi-
ments, essentially all of the S(IV) was
present as bisulfite since the pH was much
lower than the pKa of the sulfite/bisulfite
reaction.
Iodometric Titration for S(IV)
After the S(IV) sample was withdrawn
from the reactor, it was directly injected
into excess iodine solution to avoid air
oxidation to sulfate. The S(IV) reduced
the iodine to iodide. The excess iodine
was titrated with sodium thiosulfate. When
the yellow color of the iodide started to
fade (as the iodine was reduced to iodide
by the thiosulfate), a couple drops of
starch indicator were added to enhance
the endpoint detection. The endpoint was
reached when the blue solution turned
clear.
The S(IV) concentration was deter-
mined from the difference between the
amount of thiosulfate used to titrate the
excess iodine and the amount needed if
no S(IV) were added to the iodine. The
difference indicates how much of the io-
dine reacted with S(IV).
Discussion of Results
The rate constant for the CI2/S(IV) reac-
tion was extracted from CI2 absorption data
using the model for mass transfer with
-------
fast irreversible reaction in the boundary
layer. However, a precise value for the
rate constant could not be determined
due to the rapid reaction rate. The data
were used to quantify an approximate
value for the rate. The implications of the
effect of this rate constant on Hg removal
in a typical limestone slurry scrubber are
examined.
Rate of Reaction for CI2 with
S(IV)
Figure 2 depicts the chlorine flux as a
function of the S(IV) concentration. The
curves are calculated using the model for
mass transfer with fast irreversible reac-
tion in the boundary layer.
NCL =-
ci
CL
(1)
2, buf
The partial pressure of CI2 at the inter-
face and the S(IV) concentration at the
interface can be related to the bulk CI2
and S(IV), respectively.
F = O,
2,30V)
N,
CI
*k
S(IV)
2,buf tbufferl
and
CL
CL,b
(2)
The partial pressure of C2 in the bulk
(which is equivalent to the CI2 exiting the
reactor) can be written in terms of the
inlet CI2 concentration through a gas-
phase material balance. Thus, Equation
3 displays the model used to calculate
the curves, and Table 1 lists the param-
eters that were supplied to the model.
Nc,A
N,
CL
CLin
(3)
The rate constant of the CI2/S(IV) reac-
tion (k2S(IV)) was chosen to best fit the
data. All the data plotted below were ob-
tained through IMS analysis. The data at
0.01 mM S(IV) are actually in succinate
buffer with no S(IV). The inverted triangles
in Figure 2 represent points in which 15
to 20% oxygen was added. All the data
are in 50 mM buffer except for the points
represented by the squares with diago-
nal lines which are in 5 mM buffer.
10
O
265 ppm
265 ppm
21 ppm
oxygen
H 5 mM buffer
21 ppm
0.01
0.1
10
(mM)
Figure 2. Chlorine Absorption in Buffered S(IV), k 2S(|V) =2x10 L/mol$s.
Table 1. Values of parameters in global model.
Parameter Units Value
Dc,2
G
*
m2/s
m3/s
1.5x1 0"9
0.07083
1
Since HOCI reacts with sulfite, one pos-
sible mechanism for CI2 reaction with
S(IV) is that the CI2 first hydrolyzes in
water to form HOCI, and then the HOCI
(not CI2 directly) reacts with S(IV). These
overall reactions are:
Hc, atm$m /kmol 16.7
k,, kmol/s$atm$m2 0.00075
CI2+
HOCI
Cf
HOCI +
so2 + H++ cr
L/mol$s
2.45 x 10"
2x 109
,-5
1.18 L/min
Figure 2 shows that, at high S(IV), the
CI2 flux does not depend on the S(IV)
concentration since the limit of gas film
resistance is approached. At low S(IV),
the flux is limited (in some cases inhib-
ited) by the buffer-enhanced CI2 hydroly-
sis reaction. In this region, the flux
depends only on the buffer reaction rate.
However, the data show that, when very
little S(IV) was injected, the CI2 flux was
less than what it was initially in buffer
alone. Thus, S(IV) inhibited CI2 absorp-
tion at very low S(IV) concentrations. As
the S(IV) increased, the chlorine flux in-
creased until the limit of gas film resis-
tance was reached.
(4)
(5)
If this were the case, the rate of CI2
absorption in S(IV) would be equivalent
to the rate of CI2 hydrolysis to form HOCI
since CI2 hydrolysis is the rate-limiting
step. Then, the HOCI would react with
S(IV). However, since the addition of S(IV)
results in a greater CI2 removal rate than
the CI2 hydrolysis rate, it must not depend
on HOCI formation. Thus, CI2 itself reacts
with S(IV) directly, and it is not necessary
for HOCI to form before CI2 reaction with
S(IV) occurs.
In the intermediate region of Figure 2,
the flux is limited by S(IV) diffusion to the
interface [depicted by flux increasing lin-
early with S(IV)] and/or kinetics (depicted
by curvature). Looking at the 265 ppm
data, for S(IV) between 0.5 and 0.8 mM,
the flux increases linearly with S(IV),
which is consistent with the model of S(IV)
depletion. At the lower inlet concentration
of 21 ppm, the data do not fall on the
model curve at S(IV) concentrations be-
low 0.1 mM. These deviations result from
-------
the experimental uncertainty in the S(IV)
concentration measurements at low S(IV).
For example, if the iodometric analysis
yielded a S(IV) concentration of 0.07 mM,
the actual value could be 0.03 mM due to
the inaccuracies of analysis at low S(IV)
concentration. Figure 2 shows that there
is only a very small range (depicted by
curvature) where the CI2 flux should be
limited by the kinetics of the CI2/S(IV) re-
action.
Even though it was difficult to obtain an
exact value for the rate constant because
of the mass transfer limitations of the re-
actor, an approximate value can easily
be determined. Model curves were calcu-
lated for various rate constants to deter-
mine the value that best fitted the data.
Instead of plotting flux as a function of
bulk S(IV) as was done in Figure 2, plot-
ting CI2 penetration (CI2out/CI2in) as a func-
tion of [S(IV)]b/Pcl2in magnifies the errors
in the data. This allows better observa-
tion of which value for k2S(IV) best fits the
data. Figure 3 plots the same data, with-
out separately labeling the points with
oxygen, and uses the same model as in
Figure 2. The points on the y-axis do not
contain any S(IV).
The model curves were calculated with
rate constants ranging from 2.5 x 105 L/
mol$s to infinity. The rate constant used in
Figure 3 fits the data better than the other
rate constants used. Thus, the most prob-
able value of the rate constant is 2 x 109
L/mol$s, although it could be an order of
magnitude smaller or larger. Many of the
low S(IV) points, especially for the 21
ppm data, do not fall on the curve, but
that could be due to the inability to accu-
rately measure low S(IV) concentrations.
In order to get a more precise rate constant,
an apparatus with higher mass transfer
coefficients is needed so that the absorp-
tion falls in a region controlled by reac-
tion kinetics instead of mass transfer.
Hg Removal in a Typical
Limestone Slurry Scrubber
The expected Hg removal in a lime-
stone slurry scrubber can be predicted
using the extracted rate constant for the
CI2/S(IV) reaction, a preliminary rate con-
stant for Hg/CI2, and typical mass transfer
characteristics for a scrubber. Table 2
tabulates the parameters used in the
model. The value for k2S(IV)at 55° C was
estimated from the value at 25° C. The
model must be supplied with a given CI2
inlet and a constant S(IV) concentration.
The model accounts for the two simulta-
neous reactions occurring at the gas/liq-
uid interface: the depletion of CI2 through
reaction with S(IV) (k2S(IV)) and the reac-
tion of Hg with CI2 (k2Hg). Figure 4 shows
o
o_
C
o
'ra
-------
reaction will always be difficult in the ex-
isting apparatus. On one end, absorption
is limited by the CI2 hydrolysis reaction,
and on the other end, it is limited by gas-
film control in the stirred-cell contactor.
Chloride does not affect CI2 absorption
in S(IV) since the CI2/S(IV) reaction is
irreversible. Oxygen does not affect CI2
absorption in S(IV) either, nor does it seem
to catalyze S(IV) depletion at the ranges
investigated.
Recommendations
In order to accurately predict Hg and
CI2 removal in a scrubber, a better model
with precise kinetics is needed. The CI2/
S(IV) reaction rate needs to be precisely
measured in a gas/liquid contactor with
higher mass transfer coefficients. Further-
more, this reaction rate should be mea-
sured at 55° C to simulate a typical
limestone slurry scrubber.
Simultaneous absorption of Hg and CI2
must be measured and modeled to ob-
tain a precise value for k2Hg. These ex-
periments should also be done at 55° C.
Simultaneous absorption of Hg, CI2, and
SO2 should also be studied. Furthermore,
in order to completely simulate flue gas,
CO2, nitrogen oxides, and oxygen should
be added to the inlet gas.
Results have shown that chloride does
not affect CI2 absorption. However, ex-
periments were not done in sodium chlo-
ride solutions higher than 0.02 M.
Limestone slurry may have 1 M Cf. Thus,
absorption into 1 M chloride must be quan-
tified.
Nomenclature
A
Dc
D
Hg
gas/liquid contact area (m2)
diffusion coefficient for CI2 in water
(m2/s)
diffusion coefficient for Hg in water
(m2/s)
10
Figure 4. Predicted Hg Penetration.
(J> reactant stoichiometric coefficient
(dimensionless)
FC Mass flow controller
G gas flow rate to reactor (m3/s)
Hc, Henry's law constant for CI2 (atm-
2 m3/kmol)
H|Hg Henry's law constant for Hg (atm-
m3/kmol)
IMS ion mobility spectrometry
k individual gas-film mass transfer co-
efficient (kmol/s$atm$m2)
k° individual physical liquid-film mass
transfer coefficient (m/s)
k2 buf second order rate constant for C\J
buffer reaction (L/mol$s)
N
C|
k2 Hg second order rate constant for Hg/
CI2 reaction (L/mol$s)
second order rate constant for C\J
S(IV) reaction (L/mol$s)
flux of CI2 (kmol/m2$s)
partial pressure of CI2 (atm)
negative logarithm of acid disso-
ciation constant
[S(IV)] concentration of S(IV) in liquid (M)
Subscripts
b in bulk
at gas/liquid interface
rci,
PKi
i
in
out
inlet
outlet
-------
S. Roy and G. Rochelle are with the Department of Chemical Engineering, The
University of Texas at Austin, Austin, Tx 78712.
Theodore G. Brna is the EPA Project Officer (see below).
The complete report, entitled "Chlorine Absorption in S(IV) Solutions," (Order No.
PB2001-107826; Cost: $27.00, subject to change) will be available from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161-0001
Telephone: (703) 605-6000
(800) 553-6847(U.S. only)
The EPA Project Officer can be contacted at:
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711-0001
United States
Environmental Protection Agency
CenterforEnvironmental Research Information
Cincinnati, OH 45268
PRESORTED STANDARD
POSTAGES FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use
$300
EPA/600/SR-01/054
------- |