Modeling Suppression of Dioxin Formation During Coal
Combustion
Mamie A, Telfer1* and Brian K. Gullctt'
'U.S. Environmental Protection Agency, National Risk Management Research Laboratory, MD-
65, Research Triangle Park, NC 27711, USA.
'"joint Program with the Oak Ridge Institute for Science and Education Postdoctoral Program.
tel fer. marnie @ er»a. gov
Abstract
A homogeneous, gas-phase reaction mechanism has been developed to explain sulfur (S) and
chlorine (CI) interactions in an industrial, fire-tube boiler (North American Package Boiler,
NAPB) using #2 fuel oil (0.03% S) doped with copper naphthenate (CuNA) and 1,2-
dichlorobenzene (1,2-diClBz). The NAPB experiments were intended primarily for the
investigation of polychlori nated dibenzodioxin and furan (PCDD/F) formation. However,
significant reduction of sulfur dioxide (S02) from combustion of #2 fuel oil was also observed in
the NAPB upon injection of 1,2-diClBz. Interaction between S and CI has been of significant
interest due to potential of S to suppress the formation of PCDDs/Fs during cofired combustion
of high S coal with municipal waste (MW). A suggested mechanism is the gas-phase reaction
SO2+CI2+H2O = SO3+2HCI, which converts active molecular chlorine (Cl2) formed in the post-
flame zone by the metal-catalyzed Deacon process (2HC1 + 1/20.> = Cl2 + II?0) to inactive
hydrogen chloride (HC1). In the current study, this gas-phase suppression reaction is represented
by a series of elementary reaction steps, namely Cl+H20=HCl+OH and SO2+HO2—SO3+0II,
which were compiled from previously validated H/C1/0 (Proccacini, 1999) and H/S/O
(Zachariah and Smith, 1987) reaction mechanisms, respectively. The Chemkin 3.6 (Kee et al.,
2000) plug-flow reactor software program, Senkin (Lutz et al., 1987), was utilized to simulate
the 3-pass, post-flame zone of the NAPB, Pass 1 was held at a constant temperature of 665°C for
1.23 s, and Passes 2 and 3 declined from 665 to 350°C and 350 to 280°C, respectively, both in
0.69 s. The combustion products of #2 fuel oil were approximated by the theoretical products
based on complete combustion. 1,2-diClBz injection was represented by the addition of HQ, CI,
and/or Cl2 to the post-flame zone. The concentration of CI2 present in each pass was based on an
assumed and instantaneous conversion of HC1 via the Deacon reaction, as this heterogeneous
process could not yet be represented in the gas-phase mechanism. The model was able to
explain the observed rapid conversion of S02 to SO3 in the NAPB experiments upon injection of
1,2-diClBz, for a Pass I temperature of 665 °C and, assuming that the Deacon reaction has
occurred, to a HC1 conversion of 60% (i.e., 162 ppm of Cl2). These results present a possible
mechanism for gas-phase S and CI interactions and a potential means of elucidating the SO2
1

-------
suppression mechanisms of PCDD/F formation during cofired combustion of high S coal and
MW.
Introduction
Polychlorinated dibcnzodioxins (PCDDs) and polychlorinated dibcnzofurans (PCDFs) are trace,
toxic combustion byproducts that are chemically stable in post-combustion environments.
Consequently, it is more desirable to prevent their formation than to impose control or
destruction technologies. Research over the last 15 years has refined the possible formation
pathways of PCDDs/Fs to (i) low-temperature (250-350°C) reactions involving fly ash carbon
and organic or inorganic chlorine in the fly ash (i.e., de novo synthesis) (Stieglitz and Vogg,
1987), (ii) heterogeneous, catalytic reactions with the fly ash surface and gas-phase precursors
such as chlorinated phenols (CIPhs) and chlorinated benzenes (ClBzs) (Dickson and Karasck,
1987, Gullett et al., 1991), and (iii) high-temperature (> 500°C), homogeneous, gas-phase
reactions of precursor compounds (Ballschmiter et al., 1985). De novo synthesis has been the
most well-documented pathway and in the past has been considered the most predominant.
Current research, however, has encouraged more focus on mechanisms (ii) and (iii) due to recent
findings suggesting that gas-phase reactions forming PCDDs/Fs have been underestimated
(Huang and Buekens, 1999) because of oversimplified kinetics (Shaub and Tsang, 1983, Penner
et al., 1991).
PCDDs/Fs are not emitted in significant quantities from the combustion of coal in modern
boilers, despite sufficient fuel-borne chlorine (CI) and metal catalysts as well as the potential for
trace aromatic emissions. Similarly, cofiring of high sulfur (S) coal with municipal waste (MW)
has demonstrated significant suppression of PCDD/F emissions from normal waste firing
(Gullett et al., 1992, Scheidle et al., 1986, Lindbauer et al., 1992, Raghunathan and Gullett, 1996,
Gullett, 1998). The high S to CI ratio in coal and cofired coal/MW combustion systems
compared to that of MW combustion alone is considered the main reason for the difference in
PCDD/F emissions; hence, many studies have focused closely on the interaction between S and
CI. A mechanistic understanding of whether S inhibits gas-phase CI reactivity and, hence, the
formation of chlorinated organic precursor compounds, or inhibits the inorganic CI or catalyst
reactivity in the fly ash, can help to elucidate the predominant PCDD/F formation pathways and
develop preventive measures.
The present studies attempts to address whether S affects PCDD/F formation via a gas-phase
route. In doing so it is important to investigate the speciation of CI compounds since there have
not been many conclusive studies of the reactive form of chlorine or its concentration effect on
PCDD/F formation. The most predominant chlorine species in the combustion zone is hydrogen
chloride (HC1), a weak chlorinating agent (Gullett et al., 1990) likely due to the strength of the
H-Cl bond. As the temperature decreases in the post-combustion zone, HC1 is predicted by
2

-------
equilibrium to convert to molecular chlorine (Cl2), a much stronger oxidant and chlorinating
agent (Senior et al., 2000). Equilibrium calculations conducted on four different coals predicted
that, at 150 °C, 30 to 60% of total system chlorine would transform to CI2 (Senior et al., 2000).
Department of Energy (DOE)-sponsored field studies of emission levels of three utility boilers
reported CI2 levels as low as 4% of total chlorine (Miller et al., 1996). There is still not a lot of
confidence in measurement techniques for Cl2 (Ghorishi, 2001); however, if Cl2 is actually found
to be absent at low temperatures, it implies that the reaction transforming HQ is kinetically
limited. Detailed simulations of chlorocombustion systems (Booty et al., 1995, Senior et al.,
2000, Proccacini, 1999) have in fact revealed that for the high cooling rates in practical
combustion systems, the transformation of HC1, via CI radicals (Cl«) to CI2, is limited by the
kinetics of the CI- recombination to Cl2.
In the presence of catalysts, namely oxides and chlorides of transition metals such as copper (Cu)
and iron (Fe), HQ is converted to CI2 at temperatures around 400 "C via the Deacon Process
(Hisham and Benson, 1995):
2HC1 + '/2 02 = Cl2 + H20	(1)
CI2 produced from reaction (1) can then continue to chlorinate carbon in the fly ash [mechanism
(i)], or chlorinate aromatic hydrocarbons via substitution reactions [mechanism (ii)] and is
considered a critical step in the formation of PCDDs/Fs. In coal or MW combustors, iron or
copper oxide, respectively, can be present in the post-combustion zone either in wall deposits or
in fly ash particles. Thermodynamic studies report that the forward reaction of the Deacon
process dominates up to 600 °C under standard conditions (Liu et al., 2000). Liu et al. (2000)
found that, in bench-scale fluidized bed combustor (FBC) experiments with an HQ
concentration of 1500 ppm in air, conversion of HQ was favored at temperatures higher than
700 °C. Raghunathan and Gullett (1996) found reaction (1) to be independent of equilibrium,
implying that kinetics of the reaction are important. More recently, kinetics for a two-step
ch Iorination/dech 1 orination Deacon process involving a CuO catalyst has been proposed by
Nieken and Watsenberger (1999) and adapted and applied by Edwards et al. (2001) to
heterogeneous chlorination of mercury (Ghorishi, 1998).
Griffin (1986) postulated that the presence of S decreased levels of PCDDs/Fs by depleting the
available reactive chlorine gas (Cl2) via gas-phase reaction (2):
Cl2 + S02 + H20 = 2HC1 + S03	(2)
Experiments to study the homogeneous effect of S02 on available Cl2 were conducted by Gullett
et al. (1992) who reacted 200 ppm of CI2, 1000 ppm of S02, and 10% H20 in a continuous flow
reactor for 10 s at temperatures between 400 and 800 °C. Detectable traces of SO3 using a gas
chromatograph were observed only for higher inlet C\2 concentrations (-1000 ppm) and
temperatures around 800 °C. Measurements of other compounds (HC1, Cl2) were not reported.
3

-------
Equilibrium calculations revealed that reaction (2) is favored over the full temperature range
(400 - 800°C). Gullett et al. (1992) suggested that kinetics of reaction (2) may prevent
observation of measurable products until higher temperatures are obtained. Unfortunately, 800
°C is much higher than the temperature (430°C) at which the Deacon reaction produces
maximum amounts of CI2. Furthermore, the high concentration of CI2 (1000 ppm) required for
traceable amounts of SO3 is about 2 times higher than observed in MW combustors.
Subsequent experiments by Raghunathan and Gullett (1996) followed a mixture of 500 ppm CI2,
1000 ppm SO2, 3% H2O, and 10% O2 in a 10 s reactor zone at 400 °C over the course of 3 h and
measured CI2 conversion to HC1. Profiles of HC1 from a TECO HC1 analyzer were compared to
HC1 profiles from additional experiments employing exactly the same gas-phase reaetants along
with a copper oxide (CuO) catalyst. The outcome was aimed at determining whether reaction (2)
or an alternative hypothesis, S-poisoning of the Deacon catalyst [in reaction (1)], was responsible
for limiting available Ck The comparative experiments revealed that the reduction of CI2 by S
was due to the homogeneous gas-phase reaction (2) as opposed to deactivation of the catalyst
(i.e., conversion of CuO to a less reactive CuS04). Previously, Gullett et al. (1992) observed that
C11SO4 was still able to catalytically produce CI2 via the Deacon reaction; however, it shifted
maximum production of CI? 100 C° higher than that for CuO. Though this effect will likely
impact the peak formation of PCDDs/Fs, it is not likely to inhibit formation as effectively as
eliminating Cl2 altogether.
Few other studies have been conducted on direct CI2 and S02 interactions. Liu et al. (2000)
monitored the effect of HCI concentration and temperature on SO2 conversion to SO3 in a
simulated fluidized bed combustor (FBC) environment. Gullett et al. (2000) observed an
immediate drop in S02 after injection of 1,2 dichlorobenzene (1,2-diClBz) into the feed of a 733
kW (2.5 x 106 Btu/h) North American Package Boiler (NAPB) fueled with #2 oil (0.03 wt% S)
and a copper (II) naphthenate (CuNA) mixture. Other investigations report formation
suppression of PCDD/F precursor compounds due to the introduction of S02. Xie et al. (2000)
observed that the addition of S02 at a S/Cl ratio of 2.5:1 decreased the ratio of chlorophenol to
phenol from 0.95 (S/Cl = 0) to 0.08. Banaee and Larson (1995) demonstrated the inhibition of
SO2 on chlorinated benzenes (PCDD/F precursors) during cocombustion of Cl-containing
polymer (Saran plastic wrap) with high S coals. Reactions (1) and (2) were considered the main
mechanism for the responses observed in Xie et al.'s (2000) and Liu et al.'s (2000) study. A
closer investigation, however, with particular focus on kinetics is required to understand the
viability of these mechanisms. The current study employs gas-phase chemical kinetic modeling
of sulfur and chlorine compounds to assess the feasibility of the homogeneous, gas-phase
inhibition pathway of S02 on PCDD/F formation [reaction (2)]. The model is applied to the
results observed in the NAPB runs where direct responses of gas-phase constituents, S02 and
HCI, were monitored. Ultimately, these investigations are designed to elucidate the viability of
the gas-phase PCDD/F suppression mechanism during the cofired combustion of high S coal and
MW.
4

-------
Modeling and Experimental Methodology
Chemkin 3.6 (Kee et al., 2000) simulations were conducted to deduce whether gas-phase
homogeneous reactions between S and CI could explain the S0?./HC1 interactions observed in the
NAPB runs (Gullett et al., 2000). Details of the boiler facility and experimental methodology
have been described previously (Gullett et al., 2000). The NAPB is a 3-pass fire-tube boiler.
Pass 1 consists of a burner enclosed in a single combustion chamber with an average temperature
of approximately 665 °C. Passes 2 and 3 consist of 24 and 20 firetubes, respectively, and
temperature profiles of 665 - 350 (Pass 2) and 350 - 280 °C (Pass 3) over a residence time of
0.69 s. All temperature readings were taken using type K thermocouples. The boiler was
connected to continuous emission monitors (CEMs) for oxygen, carbon dioxide, carbon
monoxide, nitrogen oxides, sulfur dioxide, hydrogen chloride, water (O2, CO2, CO, NOx, SO2,
HQ, H2O, respectively), and total hydrocarbons (THCs) such as methane.
Sulfur and chlorine transformations in each pass of the boiler were simulated using a I-D plug
flow reactor (PFR) software program called Senkin (Lutz et al., 1987) which can be assigned
with either linear and non-linear temperature profiles. Sulfur and chlorine reactions were
compiled from a H/S/O reaction set by Frenklach (1981) and a H/Cl/O reaction set from a
reduced C1-C2 chlorocombustion mechanism by Proccacini (1999). The combined S/Cl/H/O
mechanism is listed in Table 1. Note that CpC? combustion and chlorocombustion reactions
were not included in the mechanism because such a reaction set would not be sufficient to
simulate the combustion of #2 fuel oil. As a result, no carbon (C) and S reactions are presented
in this system. This can be justified, however, from the findings of a convincing review of sulfur
kinetics by Schofield (2001) who reported that, to a very good approximation, S speciation
appears independent of the source of fuel S. It was also deduced that the presence of C-S species
such as carbonyl sulfide (COS), carbon monosulfide (CS), and carbon disulfide (CS2), played a
minimal role in S speciation. Similarly, potential S and CI species reactions where not included
in the mechanism due to lack of kinetic and mechanistic data available in the temperature region
applicable to combustion processes. Most kinetic information presented in the NIST Chemical
Kinetics Database (1998) for such reactions applies to stratospheric chemistry investigations
where temperatures are around 300 K. Furthermore, in reaction (2), it is clear that S and CI do
not combine directly to form the final products, and it is also highly thermodynamicaliy unlikely
that S and CI combine as intermediate compounds. From these assumptions, it can be deduced
that global reaction (2) is likely to occur via a series of elementary steps involving reaction of CI2
with H20 (or products of 1120 such as OH and H) followed by subsequent oxidation of SO2 to
SO3. These elementary steps are present in the S/H/O and Cl/H/O submechanism and are
therefore included in the S/Cl/H/O compiled mechanism in Table 1 and are likely to be:
H20 + 02 = H02 + OH (reverse 12, Table 1)
Cl2 + OH = CI + HOC1 (reverse 63, Table 1)
Cl2 + M = CI + CI + M (reverse 60, Table 1)
5

-------
CI + HiO = HQ + OH
H20 + CIO = OH + HOC1
S02 + 02 = S03 + O
(reverse 69, Table 1)
(reverse 72, Table 1)
(reverse 30, Table 1)
Due to the lack of hydrocarbon reactions in the model, theoretical estimations of the flame
products H2O, 02, Ha, and S02, based on complete combustion of #2 fuel oil, were employed as
the feed into the post-flame zone of the NAPB. HC1, CI, and Cb addition to the feed inlet
represented injection of 2,8 wt % 1,2-diClBz into the burner. HC1 was measured from the outlet
of the NAPB at 540 ppm and was set to this concentration for the simulations. There is still more
work required to incorporate the 2-step heterogeneous Deacon reaction scheme of Nieken and
Watsenberger (1999) with refitted kinetic parameters by Edwards et al. (2001) into the S/Cl/H/O
mechanism. Hence, for the preliminary simulations, Cb concentrations were based on an
assumed conversion of HC1 (540 ppm) via the Deacon reaction.
Results and Discussion
#2 fuel oil contains a low concentration of S (0.03 wt%) which immediately converts to
approximately 45 ppm of S02 in the flue gas. Figure 1 reveals that, when 1,2-diClBz is injected
into the burner feed (after 20 minutes of run time), S02 drops immediately to zero and 540 ppm
of HC1 is rapidly formed. When the injection of 1,2-diClBz ceases (around 280 minutes of run
time), SO2 steadily rises again.
The experiments show an immediate response of S02 upon injection of the CI compound.
Chemkin simulations of the post-flame zone of the NAPB were conducted to investigate whether
the interaction of S and CI was due to the gas-phase homogeneous reaction (2). Figure 2
presents the response of 45 ppm of S02 in Pass 1 of the NAPB with the inclusion of (i) 540 ppm
of HC1, and (ii) 440 ppm of HC1 with 100 ppm of CI radicals, representing the injection of 1,2-
diClBz. HQ is the most predominant CI compound in the high temperature environment of a
flame. It is expected that, as 1,2-diClBz is injected into the flame, chlorine will dissociate from
the aromatic ring and abstracted hydrogen from other hydrocarbons to form HQ. Super-
equilibrium concentrations of CI radicals may also exist due to reactions of HQ and oxidizing
radicals (OH, H02, and O) (Proccacini, 1999). Figure 2 reveals that in the presence of HC1 only,
SO2 remains constant, and conversion to SO3 does not occur. When CI radicals are introduced to
the system, S02 converts rapidly (within 200 ms) to SO3, but the conversion is not complete,
leaving approximately 40 ppm of S02 and forming around 5 ppm of SO3. It is likely that there
are not enough CI radicals present and/or their persistence at 665°C is not sufficient to
completely deplete the 45 ppm of S02.
6

-------
Table 1. List of Reactions Compiled for the S/Cl/H/O Mec!
(k = A T**b exp(-F,/RT))
REACTIONS
A
b
E
i. h+o2=oh+o
1.23E+I7
-0.9
16619.0
2. 0+H2=0H+H
2.19E+14
0.0
13682.0
3. OH+H2=H20+H
3.16E+07
1.2
2641.0
4, 0H+0H=H20+0
3.39E+13
0.0
4964.0
5. H+02+M=H02+M
2.51E+15
0.0
0.0
6. H02+H=0H+0H
2.29E+14
0.0
1849.0
7. H02+H=H2+02
1.58E+13
0.0
357.0
8. H02+H=H20+0
2.00E+13
0.0
1849.0
9. HO2+0=0H+02
3.98E+13
0.0
357.0
10. H02+H20=Hj02+0H
2.82E+13
0.0
32765.0
11. ho2+ho2=h2o2+o2
2.00E+12
0.0
0.0
12. H02+0H=H20+02
4.90E+12
0.0
357.0
13. H202+M=0H+0H+M
1.20E+17
0.0
47460.0
14. H202+H=H20+0H
2.19E+15
0.0
11696,0
15. H202+0=H02+0H
2.82E+13
0.0
6354.0
16. O2+M=0+O+M
1.86E+11
0.5
95714.0
17. H2+M=H+H+M
2.24E+12
0.5
92537.0
18. 03+M=02+0+M
2.45E+14
0.0
22697.0
19. 03+0=02+02
5.25E+12
0.0
4150.0
20. 03+H=02+0H
1.20E+13
0.0
0.0
21. 03+0H=II02+0,
7.76E+11
0.0
1908.0
22. 03+H02=02+0vf0H
1.20E+11
0.0
3098.0
23. S+02=S0+0
1.29E+12
0.0
0.0
24. 0+S2=S0+S
3.98E+12
0.0
0.0
25. S0+0.=SO,+O
3.47E+11
0.0
6493.0
26. SO+SO=S02+S
5.01E+11
0.0
3296.0
27. SO+O+M=S02+M
1.15E+22
-1.8
0.0
28. S02+0+M=S03+M
1.00E+I5
0.0
0.0
29. S0+02+M=S03+M
1.00E+00
0.0
0.0
30. S03+0=S02+02
1.00E+12
0.0
0.0
31. S03+S0=S02+S02
1.00E+13
0.0
0.0
32. OH+SO=S02+H
3.16E+13
0.0
0.0
33. OH+S=SO+H
1.00E+13
0.0
0.0
34. H02+S=SH+02
1.00E+12
0.0
0.0
35. H02+S02=S03+0H
1.00E+12
0.0
0.0
36. SO3+H=S02+OH
1.00E+12
0.0
0.0
37. H2S+M=SH+H+M
2.00E+14
0.0
74049.0
38. H+H2S=H2+SH
2.29B+11
0.0
0.0
39. SH+SH=H2S+S
7.76E+12
0.0

(k = A T**b cxp(-
A	b
E/RT))
E
40. SH+H=H2+S
2.00E+13
0.0
0.0
41.SH+S=H+S,
2.69E+13
0.0
0.0
42. H2S+0=OH+SH
4.37E+12
0.0
3277.0
43. H,S+0=SO+H2
1.00E+0Q
0.0
0.0
44. H2S+0=HSO+H
1.00E+00
0.0
0.0
45. H2S+0H=H20+SH
3.16E+12
0.0
0.0
46. SH+H202=H02+H2S
1.00E+11
0.0
0.0
47. SH+02=SO+OH
1.00E+00
0.0
0.0
48. SH+0,=S02+H
1.00E+00
0.0
0.0
49. SH+H02=H202+S
1.00E+11
0.0
0.0
50. SH+H02=H2S+02
1.00E+00
0.0
0.0
51. SH+0=OH+S
2.29E+11
0.7
1906.0
52. SH+0=SO+H
9.55E+I3
0,0
0.0
53. SH+OH=H20+S
1.00E+12
0.0
0.0
54. SH+HSO=H2S+SO
1.00E+12
0.0
0.0
55. H+SO+M=HSO+M
1.00E+15
0.0
0.0
56. HS0+0=SH+02
1.00E+12
0.0
0.0
57. HS0+0,=H02+S0
1.00E+12
0.0
0.0
58. C1+II+M=HC1+M
7.20E+21
-2.0
0.0
59. C1+H2<=>HC1+H
4.80E+13
0.0
5000.0
60. C1+C1+M=C12+iM
5.75E+14
0.0
-1600.0
61. Cl+H02=HCl+02
1.08E+13
0.0
100.0
62. C1+H02=C10+0H
5.47E+13
0.0
894.0
63. Cl+HOCl=Cl2+OH
1.81E+13
0.0
3360.0
64. CI+HOC!=HC!+C10
1.8IE+13
0.0
258.0
65. C1+H202=HC1+H0,
1.02E+12
0.0
800.0
66. C12+0=C10+C1
1.26E+13
0.0
2800.0
67. C12+H=HC1+C1
7.49E+13
0.0
1200.0
68. HOCl=Cl+OH
1.76E+20
-3.0
56720.0
69. HC1+0H=:H20+C1
2.20E+I2
0.0
1000.0
70. HC1+0=OH+C1
5.25E+12
0.0
6400.0
71. HOCI+M=Cl+OH+M
1.00E+18
0.0
55000.0
72. H0CI+0H=C10+Hj0
1.80E+12
0.0
3000.0
73. H0Cl+0=ClO+OH
5.00E+I3
0.0
1500.0
74. HOCl+H=HCl+OH
1.00E+13
0.0
1000.0
75. C10+0=Cl+02
5.75E+13
0.0
400.0
76. C10+H2=HOCl+H
1.00E+13
0.0
13500.0
77. C10+H202=H0C1+H02
5.00E+12
0.0
2000.0

-------
HCI -o- S02
1200
60
50
1000
E
S- 40
800
30
600
20
400
r! 10
200
0
-10
0	40	80	120 160 200 240 280
20	60	100 140 180 220 260	300
Run Time (min)
Figure 1. Continuous monitoring of S02 and HCI at the end of Pass 3 from the North American
Package Boiler with combustion of #2 fuel oil doped with copper (II) naphthenate and injected
with 1,2 dichlorobenzene at 20 minutes of run time.
£
a.
a.
E
o
o
to
m
o
s
50
40
30
20
10
O"
¦ -Q ¦
-m—S02
-©—$03
540ppm HCI
440ppm HCI/
1OOppm CI
. -o. -o - •
-a -
	o.
0.2	0.4	0.6	0.8
Residence Time in Pass 1 (s)
1.2
Figure 2, Senkin simulations of the conversion of 45 ppm SO2 to SO3 in Pass 1 of NAPB after
injection of 540 ppm HCI (solid line) and 440 ppm HC1/100 ppm CI radical (broken line). Pass I
was held at an average temperature of 665 °C for 1.23 s.
8

-------
«— S03
..o	°*"°
a 25 i
o
<0
JU
o
_. .o.. - - -o - - - - -o o:
\ O
X(HCI) = 40%
X{HCI)= 100%
a ¦ o—e	o-o-
-08
0 ¦ -
0.2	0.4	0.6	0.8	1
Residence Time in Pass 2 and 3 (s)
1.2	1.4
Figure 3. Conversion of remaining 40 ppm of SCb (from Pass 1) to SO3 in Passes 2 and 3 of the
NAPB where HC1 was converted to CI 2 by 40 (solid lines) and 100% (broken lines) via the
Deacon process. Passes 2 and 3 were held at 665-350 °C and 350-280 °C for 0.69 s, respectively.
45 f.
40 -i\
. . Q ¦
. O • ¦
. o - •
.G-'
35 J \
a.
CL
cz
o
o
CO
o
30
25
20
15
10
\ 
-------
During the NAPB run, CuNA is present which can serve as a catalyst to the Deacon reaction
[reaction (1)] converting HC1 to Cb- Bench-scale experiments show that the Deacon reaction
produces optimum concentrations of CI2 around 400°C (Gullett et al., 1992); hence, simulations
were conducted for Passes 2 and 3 of the boiler with CI2 present to represent the instantaneous
conversion of the Deacon reaction. Figure 3 displays the conversion of the remaining 40 ppm of
S02 to SO3 as a result of assumed conversion of 40% of HC1 to CI2 (i.e., 108 ppm Cl2) and
assumed 100% conversion of HQ to CI2 (i.e., 270 ppm Cl2) via the Deacon reaction. Figure 3
reveals that for the temperature environments of Passes 2 and 3, no less than 100% conversion of
HC1 is sufficient to completely convert SO2 to SO3 via reaction (2). The elementary steps
leading to the global reaction (2) are obviously dependent on temperature and, at a CI2
concentration of 108 ppm (i.e., 40% HC1 conversion), require a temperature range higher than
665 °C - 280 °C to reach completion.
Subsequent simulations were conducted in Pass 1 at 665 °C, assuming that the Deacon reaction
occurred instantaneously to 20, 40, and 60% conversion of HC1. Only when the conversion was
set to 60% did S()2 completely convert to SO3 within the allowed residence time in Pass 1 (1.23
s). Concentrations of CI2 for 20 and 40% conversion of the Deacon reaction were not high
enough to reduce SO2 completely. Figure 4 displays the SO2/SO3 conversion in Pass 1 for the
60% conversion simulation along with simulations conducted at Pass 1 temperatures of 500 and
800 °C. When the temperature increased to 800 °C, SO3 forms more rapidly but not completely,
and when the temperature is decreased to a constant 500 °C, conversion is virtually negligible.
Exploration of different temperatures in Pass 1 is helpful in elucidating what happens in a more
realistic environment where heat from the burner flame and heat losses to the chamber walls
provide axial and radial temperature gradients. Due to lack of information about the effect of
temperature on the Deacon reaction, it is difficult to know whether the assumption of 60%
conversion of HQ to Cl2 is valid at the simulated temperatures of 500-800 "C. While fixed-bed
experiments of the Deacon reaction reveal that at 600 °C only 15% conversion of HQ is possible
(Gullet et al., 1992), Liu et al. (2000) report that, for a bench-scale fluidized bed combustor
system, conversion of HQ to Cl2 is favored at temperatures higher than 700 °C. Note that none
of the reactor conditions of these experiments resemble the plug-flow reactor conditions
employed in the Chemkin simulations or the practical post-flame conditions of the North
American Package Boiler, where intense temperature gradients exist in all three passes. It is
possible that, in the practical environment where sufficient mixing and temperature gradients do
exist, low temperature (400 °C) regions near the chamber walls encourage the Deacon reaction
and formation of Cl2, and high-temperature regions (650 °C) exist to enable reaction (2) to
proceed to completion. In addition, the absence of any C combustion reactions may have
significant effect on the radical pool, which in turn may affect S and €1 reactions. This needs to
be investigated in the future. These simplified simulations, however, are helpful in showing that,
if certain conditions are employed, it is possible, using the current S/Cl/H/O mechanism, to
10
-------
simulate a rapid reduction of S02 by injecting a chlorinated compound such as 1,2-diClBz in the
NAPB system.
A sensitivity analysis using the Senkin software program was conducted to determine which
reactions most significantly affect the major species. The predominant reactions in the S/Cl/H/O
mechanism presented in Table 1 responsible for the formation of HC1 and SO3 are:
CI + H20 = HC1 + OH (reverse. 69)	S02 + H02 = S03 + OH	(35)
An investigation of the intermediate compounds (CI, HO2, OH) suggests that reactions (69) and
(35) are likely to be initiated by reactions (12), (64), (63), (61), and (62) of the S/CI/I I/O
mechanism:
02 + Ii20 = H02 + OH (reverse. 12)	CI + HOC1 = HC1 + CIO	(64)
CI2 + OH = CI + HOC1 (reverse. 63)	CI + H02 = HC1 + 02	(61)
CIO + OH = CI + HO, (reverse. 62)
The above reactions provide a surplus of H02 radicals, enabling the conversion of S02 to SO3.
In a kinetic and mechanistic study of HC1 oxidation to CI2 by Gavriliv et al. (1975), the HO?
radical was suggested to be important in initializing the conversion of HC1 to Cl2 via CI radicals.
Thus, S02's affinity for the H02 radical is likely to aid the reverse process (Cl2 back to HC1),
possibly explaining in more detail the mechanistic effect of S02 on available Cl2 concentrations.
Conclusions
The current S/Cl/H/O model reveals that it is possible to explain the rapid decrease of 45 ppm of
S02 due to 1,2-diClBz injection at 2.8 wt% in the NAPB system via a series of elementary gas-
phase reaction steps, resulting in overall global reaction (2). Reaction of CI radicals with H20,
and Cl2 with OH, providing a surplus of H02 radicals that initiates the reaction S02 + H02 = SO3
+ OH, was found to be the underlying mechanism describing the rapid reduction of S02.
Simplified Senkin simulations of the NAPB system reveal that the conversion of S02 to SO3 was
dependent on temperature and Cl2 concentration, displaying complete conversion for Pass 1
temperatures around 665 °C and (for Deacon reaction conversions) no less than 60% (which
provides around 162 ppm of Cl2). Continued development of the S/Cl/H/O mechanism to
incorporate C combustion reactions and Deacon reaction kinetics plus adequate experimentation,
is required to further validate the elementary steps leading to the global gas-phase reaction
S02+C12+H20 = SO3+2HCL The current investigation, nonetheless, demonstrates the potential
viability of understanding and modeling gas-phase interactions between S and CI for elucidating
predominant mechanisms of PCDD/F formation suppression by S during cofired combustion of
high S coal and municipal waste.
11
-------
Acknowledgements
This research was supported in part by an appointment to the Postdoctoral Research Program at
the National Risk Management Research Laboratory, administered by the Oak Ridge Institute for
Science and Education through Interagency Agreement No. DW89938167 between the U.S.
Department of Energy and the U.S. Environmental Protection Agency.
References
Ballschmiter, K., Zoller, W„ Buchert, H„ Class, Th. (1985) Fres. Z. Anal. Chem., 322, 587.
Banaec, J., Larson, R.A. (1995) Waste Management, 15, 8, 609-614.
Booty, M.R., Bozzelli, J.W., Ho, W., Magee, R.S. (1995) Environmental Science and
Technology, 29, 12, 3059-3063.
Dickson, L.C., Karasek, F.W, (1987) J. Chromatog389, 127-137.
Edwards, J., Srivastava, R.K., Lee, C.W., Kilgroe, J.D., Ghorishi, B. (2001) The A&WMA
Specialty Conference on Mercury Emissions: Fate, Effects, and Control, Chicago, August 21-23,
pp 238.
Frenklach, M., Lee, J.H. (1981) Combustion and Flame, 41, 1-16.
Gavriliv, A.P., Kochubei, V.F., Moin, F.B. (1975) Kinet. Katal, 16, 666.
Ghorishi, B. (1998) EPA-600/R-98-014 (NTIS PB98-127095), Air Pollution Prevention and
Control Division, February.
Ghorishi, B. (2001) Personal communication, October 15.
Griffin, R.D. (1986) Chemosphere, 15, 9-12, 1987-1990.
Gullelt, B.K., Bruce, K.R., Beach, L.O. (1990) Waste Management Research, 8, 203.
Gullett, B.K., Bruce, K.R., Beach, L.O. (1991) Presented at the 1991 Conference on Thermal
Treatment of Radioactive Hazardous Chemical Mixed and Medical Waste, Knoxville, TN, May
13-17.
Gullett, B.K., Bruce, K.R., Beach, L.O. (1992) Environmental Science and Technology, 26, 10,
1938-1943.
Gullett, B.K., (1998) Proceedings of the 2nd Asia Pacific Conference on Sustainable Energy and
Environmental Technologies, 1998, June 14-17, Queensland, Australia, 115-124.
Gullett, B.K., Touati, A., Wai Lee, C. (2000) Environmental Science and Technology, 34, 2,
2069-2074.
Hisharn,	Benson, S.W. (1995) J. Phys. Chem., 99, 6194-6198.
Huang, H., Buekens, A. (1999) Chemosphere, 38, 1595.
Kee, R.J., Rupley, F.M., Miller, M.A., Coltrin, M.E., Grcar, J.F., Meeks, E„ Mofffat, H.K., Lutz,
A.E., Dixon-Lewis, G., Smooke, M.D., Wamatz, J., Evans, G.H., Larson, R.S., Mitchell, R.E.,
Petzold, L.R., Reynolds, W.C., Caracotsios, M.. Stewart, W.E., Glarborg, P., Wang, C, Adigun,
O., CHEMKIN Collection, Release 3.6, Reaction Design, Inc., San Diego, CA (2000).
Lindbauer, R.L., Wurst, F„ Prey, T. (1992) Chemosphere, 25, 7-10, 1409-1414.
12
-------
Liu, K., Pan, W.P., Riley, J.T. (2000) Fuel, 79, 1115-1124.
Lutz, A.E., Kee, R.J., Miller, J,A. SENKIN: A Fortran Program for Predicting Homogeneous
Gas Phase Chemical Kinetics with Sensitivity Analyziz, Sandia Report SAND87-8209; Sandia
National Laboratories: Livermore, CA, 1987.
Miller, S.M., Ness, S.R., Weber, G.F., Erickson, T.A., Hassett, D.J., Hawthorne, S.B., Katrinak,
K.A., Louie, P.K.K. (1996) Final Report on Contract DE-FC21-93MC30097, September.
Nieken, U., Watsenberger, O. (1999) Chem. Eng. ScL, 54, 2619-2626.
NIST Chemical Kinetics Database (1998), NIST Standard Reference Database 17 - 2Q98,
National Institute of Standards and Technology, Gaithersburg, MD.
Penner, S.S., Li, C.P., Richards, M.B., Wiesenhahn, D.F. (1991) ScL Tot. Environ., 104, 35.
Proccacini, C. (1999) Ph.D. Thesis, Massachusetts Institute of Technology, Cambridge, MA ,
September.
Raghunathan, K.. Gullett, B.K. (1996) Environmental Science and Technology, 30, 6, 1827-
1834.
Scheidle, K„ Wurst, F„ Kuna, R.P. (1986) Chemosphcre, 17, 2089.
Schofield, K. (2001) Combustion and Flame, 124, 137-155.
Senior, C.L., Sarofim, A.F., Zeng, T., Helbie, J.J., Mamani-Paco, R. (2000) Fuel Processing
Technology, 63, 197-213.
Shaub, W.M., Tsang, W. (1983) Environ. ScL TechnoL, 17, 721.
Stieglitz, L„ Vogg, H.(1987) Chemosphere, 16, 1917-1922.
Xie, Y., Xie, W„ Liu, K„ Dicken, L„ Pan, W.P., Riley, J.T. (2000) Energy and Fuel, 14, 597-
602.
Zachariah, M.R., Smith, O.I. (1987) Combustion and Flame, 69, 125-139.
13
-------
kt™™ r.™ rw/n TECHNICAL REPORT DATA
NRMKL-KIr-r-DHU (Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA/600/A-01/117
3. RECI
4. TITLE ANO SUBTITLE
Modeling Suppression of Dioxin Formation During Coal
Combustion
S. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Mamie A. Telfer and Brian K. Gullett
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
See Block 12
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
EPA/DOE IAG DW89938167
12. SPONSORING AGENCV NAME ANO ADDRESS
EPA, Office of Research and Development
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Published paper;
14. SPONSORING AGENCV CODE
EPA/600/13
is.supplementary notes Appa) project officer M.A. Telfer (ORISE Postdoctoral Program),
Mail Drop 65), 919/541-0734). For presentation at Pittsburgh Coal Conference, New-
castle. New South Wales, Australia, 12/4-7/01,
is.as ract rj^e paper discusses a homogeneous, gas-phase reaction mechanism that has
been developed to explain sulfur (S) and chlorine (CI) interactions in an industrial,
fire-tube boiler, using No. 2 fuel oil (0.03% S) doped with copper naphthenate (CuNA)
and 1,2-dichlorobenzene (1,2-diClBz). The experiments were intended primarily for the
investigation of polychlorinated dibenzodioxin and furan (PCDD/F) formation. However,
significant reduction of sulfur dioxide (S02) from combustion of No. 2 fuel oil was
also observed upon injection of 1,2-diClBz. Interaction between S and CI has been of
significant interest due to the potential of S to suppress the formation of PCDDs/Fs
during cofirea combustion of high-S coal with municipal waste (MW). A suggested mech-
anism is the gas-phase reaction S02 + C12 + H20 = S03 + 2HC1 that converts active
molecular chlorine (C12)—formed in the post-flame zone by the metal-catalyzed Dea-
con process—to inactive hydrogen chloride (HCl). In this study, the gas-phase sup-
pression reaction is represented by a series of elementary reaction steps compiled
from previously validated reaction mechanisms. The model was able to explain the ob-
served rapid conversion of S02 to S03 in the boiler experiments upon injection of
1,2-diClBz to a HCl conversion of 60%. These results present a possible mechanism for
gas-phase S and CI interactions and a potential means of elucidating the S02 sup-
pression mechanisms of PCDD/F formation during cofired combustion of high-S coal and
MW.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution
Mathematical Models
Furans
Halohydrocarbons
Coal
Combustion
Boilers
Pollution Control
Stationary Sources
Dioxins
13B
12A
07C
21D
21B
13A
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report J
Unclassified
21. NO. OF PAGES
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
-------