Você está na página 1de 8

Available online at www.sciencedirect.

com
Proceedings
of the
Combustion
Institute
Proceedings of the Combustion Institute 34 (2013) 625632
www.elsevier.com/locate/proci

Experimental and kinetic modelling study


of H2S oxidation
Chenlai (Ryan) Zhou, Karina Sendt, Brian S. Haynes
School of Chemical and Biomolecular Engineering, The University of Sydney, Sydney, NSW 2006, Australia

Available online 21 June 2012

Abstract

Oxidation of H2S in an atmospheric pressure ow reactor at temperatures from 950 to 1150 K has been
studied under fuel-lean conditions (1.9 < [O2]0/[H2S]0 < 10). The reaction is strongly catalysed by silica sur-
face but this eect is suppressed with application of a B2O3 coating to the silica. A feature of the reaction is
the high selectivity (40%) to H2 formation, even in the presence of a large excess of oxygen.
A comprehensive chemical kinetic model for H2S oxidation has been developed. The model includes
detailed chemistry for disulfur interactions based on recent experimental and theoretical work. The model
shows extraordinary sensitivity to the rates of a signicant number of reactions that determine the radical
concentrations during the oxidation process. Minor variations (<factor 3) in the rate coecients of two of
the SH self-reaction channels allows accurate description of H2S consumption proles: on the one hand, the
reaction 2SH
H2S + S (followed by S + O2
SO + O and SO + O2
SO2 + O) is strongly branch-
ing whereas on the other, 2SH
HSSH (followed by HSSH + SH
H2S + HSS and HSS + SH

H2S + S2) is a powerful chain termination sequence, especially when the oxygen excess is reduced.
Disulfur interactions are extremely important in the ignition and propagation of H2S oxidation even
under fuel lean conditions.
2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Hydrogen sulde; Chemical kinetic modelling; Disulfur species; Oxidation mechanism

1. Introduction While there are extensive kinetic studies of the


impact of doped sulfur species on the combustion
Hydrogen sulde (H2S) is commonly present in radical pool (see [1] and references therein), the
some natural gases and is produced as a side prod- oxidation of H2S itself remains poorly understood.
uct in oil renery operations and fossil fuel gasi- Gargurevich [2] recently summarised the kinetics of
cation processes, often being further processed via reactions thought to be important in very fuel-rich
combustion in a Claus furnace. Sulfur chemistry Claus furnace combustion. Cerru et al. [3] recently
in these high-temperature processes is complex proposed a detailed sulfur mechanism validated
and poorly understood, leading to uncertainties against the limited available literature data but
in process design emissions control. noted that the role of SH radical remains uncertain,
in particular SH self-reaction involving HSS and
HSSH species, which are especially important in
Corresponding author. Fax: +61 2 9351 3471. stoichiometric and fuel-rich conditions.
E-mail address: brian.haynes@sydney.edu.au (B.S. We [4] have recently combined experimental
Haynes). and high-level theoretical studies of the H2/S2

1540-7489/$ - see front matter 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.proci.2012.05.083
626 C. Zhou et al. / Proceedings of the Combustion Institute 34 (2013) 625632

system (including H2, S2, HSSH, H2SS, 2SH, and Theoretical calculations were also performed
H + HSS) over a wide range of temperatures and for SH reactions with HO2, H2O2 and HSO spe-
pressures. This study revealed complexity in the cies and reported in elsewhere [12]. While substan-
kinetics that had not been included in any previ- tial barriers are present for the reaction SH +
ous attempts to model H2S oxidation for exam- H2O2 forming H2S + HO2 (R31) or HSOH + OH
ple, in the self-reaction of SH, formation of HSSH (R32), it was found that SH + HO2
HSO +
occurs in parallel with H2S + S production, with OH (R48) and SH + HSO
H2S + SO ( R36)
reaction to HSS + H also becoming important at are fast and barrierless.
higher temperatures (>1000 K). The chemistry of HSO interactions with H, O,
The objective of this paper is to develop a and OH (R85, R86, R88, R92 and R94) are
chemical kinetic model for fuel-lean H2S oxida- adopted from Rasmussen et al. [1]. Regarding
tion that incorporates detailed characterisation the reaction HSO + O2, our theoretical study
of the H2/S2 system along with additional new [12] revealed the formation of SO2 + OH (R72)
results for other reactions not included in earlier via the O2-insertion intermediate (HOOSO) to
models. In support of the model development, be several orders of magnitude faster than their
we have also obtained new data for oxidation of [1] assumed channel to HSO2 + O (R74) at tem-
H2S in a ow reactor in the critical temperature peratures below 800 K; at higher temperatures,
regime around 1000 K. H abstraction to form SO + HO2 (R70) becomes
dominant.
The mechanism of SO2-catalysed radical
2. Chemical kinetic model recombination primarily via HSO2 and HOSO
intermediates has been well studied, producing
The thermochemical data for the 41 species con- reliable kinetic parameters for relevant channels
sidered in H2S oxidation mechanism were mainly [1,20].
adopted from literature, including Leeds Univer- Except for the direct reactions H2S + SO2 and
sity (H2S, SO2, SO, S, HSOH, HSO2, HOSO, H2S + S2O [7], the rate coecients and products
HOSO2 and SO3) [5], Burcats Thermodynamic of the majority of sulfursulfur interactions are
Data (H2O, H2, H2O2, HO2, S2, S3, S4, S5, S6, S7, simply assumed. These include dummy processes
S8, O3, H, OH, O and O2) [6] and our own earlier that allow Sn (n = 38) species to form in partial
work (HSSH, HSS, S2O, H2S2O2, H2O. . .SO2, equilibrium with S2 (R259R264) in order to esti-
H2S3O and HSSSOH) [7,8]. Recent updates were mate the potential for these species to act as a sul-
used for the heats of formation of SH [9], HSO, fur sink.
HOS [10] and HSOO [11]. For the remaining spe- Chemical kinetic modelling of reaction condi-
cies (HSSO2, singlet SO (SO*), SSO2, HSSO and tions was carried out using CHEMKIN [21]. Plug
OSSO) we evaluated thermochemical data at the ow was assumed for ow reactors and the PRE-
G3 level [12,13] using statistical methods. Detailed MIX [21] ame code was employed for modelling
parameters in the 7-term polynomial form are tab- ame conditions reported in the literature.
ulated in the Supplementary data.
The mechanism (see Supplementary data) con-
sists of 277 reversible gas-phase elementary reac- 3. Experimental
tions, including the well established H2 oxidation
mechanism [14], SO and SO2 oxidation chemistry The experimental set-up comprises a silica ow
[1,15], reactions of SO3 with H, O, and OH radi- reactor (ID = 7 mm) housed vertically in a tem-
cals [16] and the H/S subset [8] with updated perature-controlled furnace (Fig. 1). The reactant
kinetic parameters [4]. Rate constants for reac- streams (1.0% H2S in N2 and O2 in N2) are fed
tions in the H/S/O system (R42R44, R58 and and preheated by owing downwards through
R78R83) are adopted from high-level theoretical annuli that surround the central reaction tube,
calculations [17]. mixing at its base and exiting at the top. The
With respect to SH oxidation channels, the H2S concentration entering the reactor was in
reaction SH + O2 was long considered to form the range 100520 ppm in order to eliminate reac-
HSO + O (R39) as the favoured products. How- tion heat release eects.
ever, our recent theoretical work [11] revealed that The countercurrent nature of the feed and
HSO + O formation is more than two orders of reaction streams ensured very precise temperature
magnitude slower than previous estimates [18] regulation (<1 K) of these streams, as conrmed
allowing the reaction via a four-membered cyclic by separate measurements and heat transfer mod-
transition state producing SO + OH (R40) to be elling. However, in passing through the insulation
competitive at temperatures >1000 K. Very surrounding the furnace box, the reaction gases
recently, Garrido et al. [19] concluded that the cooled only slowly, at about 1000 K s 1 this
route leading to H + SO2 (R41) via three-centre eect was measured accurately and accounted
ring structure HSO2 intermediate is actually the for in the kinetic modelling of the experimental
preferred channel for the reaction. reaction conditions.
C. Zhou et al. / Proceedings of the Combustion Institute 34 (2013) 625632 627

Probe 350
Sampling
300 B 2O3-Coating
Pump

H2S concentration (ppm)


Thermocouple 250 Bare Silica
Sampling
200

150

Fuel Oxidiser 100

50

Preheating 0
500 600 700 800 900 1000 1100 1200
annulus
Temperature (K)

Reaction Fig. 2. Comparison of H2S oxidation in the uncoated


tube silica reactor with oxidation in the same reactor coated
with B2O3. Inlet composition 325 ppm H2S, 600 ppm O2.
Mixing
zone
perature. While it cannot be demonstrated that
B2O3 itself neither promotes nor inhibits the reac-
tion signicantly, the apparent activation energy
Furnace
and overall reactivity of the system is less obvi-
Fig. 1. Schematic diagram of the experimental setup for ously catalytic in nature in the coated reactor.
H2S oxidation. All experimental proles reported here are there-
fore from measurements conducted in the B2O3-
coated reactor.
A total reactant ow rate of 420 cm3 min 1
(294 K, 1 bar) is controlled by mass ow control-
lers to achieve residence times around 0.2 s in the 4. Experimental results
isothermal hot zone and 0.3 s in the cooling zone,
depending on actual reaction temperatures The Figure 3 shows experimental results from the
reactor pressure was maintained at 1.05 bar. In ow reactor for three conditions, as a function
some experiments in which reaction was com- of reactor temperature. The data in Fig. 3a are
pleted inside the isothermal region, species were for low [H2S]0 (=100 ppm) and large oxygen
withdrawn rapidly (<5 ms) from specic points excess [O2]0/[H2S]0 = 10; those in Fig. 3b and c
along the centreline of the central reaction tube are for increasing values of [H2S]0 (=325 and
using a silica vacuum probe (3 mm OD  1 mm 520 ppm, respectively) and slight oxygen excess
ID with 30 lm orice at the tip). ([O2]0/[H2S]0  1.9). Based on the temperature at
The concentrations of H2S, H2 and SO2 were which 50% conversion of the H2S is achieved,
measured by micro-GC (MTI 200) with a preci- the very lean condition (a) is most reactive;
sion of 2%. Certied cylinder standards were used increasing [H2S]0 at constant [O2]0 (c) raises the
for direct calibration for H2S and H2, while SO2 conversion temperature by 40 K; relative to (c),
was calibrated using product gases from the com- decreasing [H2S]0 at constant [O2]0/[H2S]0 raises
plete oxidation of a known inlet concentration of the reaction temperature by a further 35 K. In
H2S. all cases, there is substantial formation of H2, typ-
Since silica surface was found to promote the ically 35% selectivity at the lowest temperature
oxidation of H2S [12], and since boric acid coating at which H2S conversion is substantially complete.
was found in early studies of H2 and hydrocarbon Qualitatively similar results, not reported here,
oxidation [2225] to be inert to the decomposition have been obtained over a wide range of condi-
of peroxy species, we investigated the use of this tions [12].
coating here in our analogous H2/S2/H2S system. The only sulfur components detected in the
Following [22], the reactor wall was rst coated product gases were H2S and SO2. As shown in
with boric acid that yielded a transparent impervi- Fig. 4 the results for the low ratio of [O2]0/
ous coating of B2O3 on baking at 773 K. A com- [H2S]0 show a signicant decit (low SO2 yield).
parison of experimental proles (for 310 ppm This decit peaks at the temperature at which
H2S oxidation with 600 ppm O2) measured in the H2S conversion is 50%, but decreases to
the B2O3-coated reactor and in an uncoated silica the noise level at higher temperatures. According
is presented in Fig. 2 it is apparent that the sys- to the simulations discussed in more detail below,
tem is much less reactive in the coated reactor, it is sulfur, mostly in the form of S2 with lesser
with >200 K increase in the 50% conversion tem- quantities of S4 and S8, that accounts for the def-
628 C. Zhou et al. / Proceedings of the Combustion Institute 34 (2013) 625632

100 14
SO2 (a) (b)
H2S 12
80

Sulfur Mass Deficit (%)


10
60
(c)
8

40
6
H2
20 4

0 2

300 0
H2S SO2 (b)
Concentration (ppm)

950 1000 1050 1100 1150


250 Temperature (K)
200
Fig. 4. Comparison between experimental data (sym-
150 bols) and modelling predictions using adjusted mecha-
nism (lines) for mass decit of sulfur calculated as 1 -
100 H2
50

0 350 1200

500 (c) 300


SO2 H2S SO2
H2S
Concentration (ppm)

1000
400 250

Temperature (K)
200 T
300
800
150 H2
200
100
H2 600
100
50

0
0 400
950 1000 1050 1100 1150 0 5 10 15 20 25 30
Temperature (K) Distance (cm)

Fig. 3. Comparison between experimental data (sym- Fig. 5. Comparison between experimental data (solid
bols) and modelling predictions (lines) for H2S oxidation symbols, probe sampling; open symbols, sampled at
in the B2O3 coated reactor. Dashed lines show the base- reactor exit) and modelling predictions (base model
case predictions for H2S; solid lines show predictions broken lines; adjusted model solid lines) for oxidation in
using the adjusted mechanism (dependent on actual B2O3 lined reactor at 1120 K. Reactor inlet conditions
experiments, see text). Inlet compositions: (a) [H2S] = [H2S] = 310 ppm, [O2] = 600 ppm. Thick dashed line
100 ppm, [O2] = 1000 ppm, (b) [H2S] = 325 ppm, [O2] = shows the temperature prole through the reactor and
600 ppm; (c) [H2S] = 520 ppm, [O2] = 1000 ppm. cooling section.

in the quench zone, leading to the reduction of


icit. Indeed, prolonged reaction under these condi- H2 selectivity to 21% at the reactor outlet.
tions gave rise to an observable yellow sulfur
deposit on the wall at the reactor outlet. This phe-
nomenon precluded study of lower (stoichiometric 5. Modelling and discussion
and fuel-rich) [O2]0/[H2S]0 ratios in the ow
reactor. 5.1. Mechanistic overview
Figure 5 shows results obtained as a function
of position in the ow reactor. The conditions in As shown by the dotted lines in Figs. 3 and 5,
the experiment are close to those in Fig. 3(b) at the base case model qualitatively predicts the
1120 K ([H2S]0 = 310 ppm, ([O2]0 = 600 ppm). overall trends of H2S consumption reaction; the
The lled symbols represent the results obtained solid lines in Figs. 35 show the predictions after
using the probe; the open symbols (at 28 cm from minor adjustments to key rate constants, as dis-
the reactor inlet) were obtained directly at the cussed below. Before proceeding to these adjust-
reactor outlet. The onset of reaction is detected ments, we rst describe the mechanism of
at 6 cm with complete consumption of H2S reaction as revealed by reaction path analysis, as
being achieved at 9 cm. The peak H2 selectivity summarised in Fig. 6 for propagating conditions
(41%) coincides with the removal of the last (50% H2S consumption) solid lines show the
traces of H2S. The oxidation of H2 occurs early most important pathways in the presence of a
C. Zhou et al. / Proceedings of the Combustion Institute 34 (2013) 625632 629

large excess of oxygen ([O2]0/[H2S]0 = 10, Fig. 3a) In considering the mechanism of ignition, it is
while the other lines show the additional pathways the direct reaction of H2S with O2 ( R25) that
that become increasingly competitive as [O2]0/ creates the rst radical centres, as rst postulated
[H2S]0 is reduced, rst to 4 (dashed lines) and by Merryman and Levy [26]. Rapid branching
then to 1.9 (as in Fig. 3b and c; dotted lines in only occurs once SH has accumulated suciently
Fig. 6). for its self-reaction to generate S ( R3, R4) for
Under all conditions, H2S is largely consumed the oxidation of S and then SO to occur at a sig-
by H (R2), OH (R27R28), and O (R29), forming nicant rate.
the SH radical. Consumption of SH occurs both
via its self-reaction forming S radical ( R3 and 5.2. H2S consumption
R4) and via reaction with O forming SO and
H (R44). At high [O2]0/[H2S]0, S is predominantly Figure 7 shows the logarithmic sensitivity coef-
converted to SO (R59) which is further oxidised to cients for [H2S] at the reactor exit for most sen-
SO2. (R116), these processes being strongly sitive reactions under the conditions in Fig. 3a
branching through their production of O radical. and b, in each case at the predicted temperature
The H + O2 recombination reaction (R221) plays for 50% H2S consumption. It is immediately
an important role in opposing chain branching. apparent that the model predictions are extremely
With decreasing [O2]0/[H2S]0, the concentration sensitive to the rate parameters of a signicant
of S increases and S2 emerges as an important number of reactions. In particular, the reaction
intermediate through reaction of S with SH (R6); SH + SH
H2S + S ( R4 and R3) has a sen-
the concentration of S2 rises because its consump- sitivity coecient near 2 with both high (a) and
tion occurs principally via its reaction (R174) with low (b) oxygen excess. From the discussion of
O which is now less available. Reaction rates of reaction pathways above, it is apparent that this
channels directly involving O2 (R59, R116 and sensitivity arises from the fact that the formation
R221) also become less eective in competing with of S atoms in turn promotes the chain branching
other channels, particularly those that consume reactions S + O2
SO + O (R59) and
the radicals S, SO and H (e.g. SH + S
S2 + H SO + O2
SO2 + O (R116). The sensitivity to
(R6), SO + SO
SO2 + S ( R133) and H2S + O2
SH + HO2 ( R25) arises from the
SH + H
S + H2 ( R5)). There is now signi- role of this reaction in determining the ignition
cant formation of SO2 (10%) and H2 (15%) temperature. Among channels inhibiting H2S oxi-
via these secondary channels. dation, pathways involving SH (R6, R9, R10
Approaching the stoichiometric condition and R44) have the greatest sensitivity, especially
(Figs. 3b, 3c and 5), availability of O is further under the low O2 excess condition.
reduced and the proportion of SH consumption The rates of some of the reactions appearing in
via reaction with O (R44) declines relative to reac- Fig. 7 are better known than others: for example,
tions with HSS and HSSH (R11 and R20, respec- there is relatively little uncertainty about the rate
tively). These chain-terminating reactions together of H + O2 + M
HO2 + M (R221); nor about
produce more S2 which accumulates to signicant the rates of reactions S and SO with O2 (R59
levels (Fig. 4). The H + O2 recombination now and R116, respectively) which are well known
plays only a minor part in determining the state from combustion and atmospheric chemistry
of the radical pool. studies. There are few experimental rate determi-

+H -H2 +OH -H2O


R2
H2S R27, R28 O2 in large excess
+O O2 in moderate excess
R29
-OH O2 in low excess

SH
+H +SH -H2S +S -H +SH +SH + M
-H2 -R3, -R4 +O R6 -R9 R10
R44 -H
-H +SH +SH
-R5
+O2 -O +O -S -H2S -H2S
S SO S2 HSS HSSH
R59 R174 R11 R20
+O2 +SO
R116 -R133
-O -S

SO2

Fig. 6. Schematic mechanism for fuel lean H2S oxidation showing main consumption pathways. As the initial [O2]/[H2S]
is decreased, the additional pathways shown as dashed and dotted lines become competitive.
630 C. Zhou et al. / Proceedings of the Combustion Institute 34 (2013) 625632

well into the cooling zone when the furnace is at


S + SH = S2 + H (R6)
lower temperature due to the accumulation of
SH + SH = HSS + H (-R9)
SH + SH + M = HSSH + M (-R10) SO and S2 (at lower values of [O2]0/[H2S]0) which
SH + O = SO + H (R44) continue to react even down to a few hundred
H2S + S + M = HSSH + M (R24) degrees. The dierent scaling factors for similar
H + O2 + M = HO2 + M (R221)
reaction conditions (curves in Figs. 3b and 5)
(R29) H2S + O = SH + OH
(R39) SH + O2 = HSO + O
are attributed to the fact that we are adjusting
(-R31) H2S + HO2 = H2O2 + SH only a minimal subset of the reactions and that
(R174) S2 + O = SO + S there are other highly sensitive reactions whose
(-R25) H2S + O2 = SH + HO2
impact would have to be considered in a full
(-R3) SH + SH = H2S + S
(R116) SO + O2 = SO2 + O
model validation. Notwithstanding the variability
(R59) S + O2 = SO + O in the adjustment factors needed under dierent
(-R4) SH + SH = H2S + S conditions, their magnitude is always within the
experimental/computational uncertainty for these
-2 -1 0 1 2
channels [4].
H2S Sensitivity

5.3. H2 selectivity
Fig. 7. Sensitivity coecients with respect to H2S
calculated using the base model at the temperature
corresponding to a predicted 50% consumption of H2S. In general, while the adjusted model is able to
Grey bars: [H2S] = 100 ppm, [O2] = 1000 ppm, T = predict the H2S and SO2 proles, only qualitative
1005 K; black bars: [H2S] = 325 ppm, [O2] = 600 ppm, agreement is achieved for the extent of H2 forma-
T = 1025 K. Reactions are written in the direction in tion which is overpredicted, especially at high tem-
which they proceed. peratures under conditions with low O2 excess
(Fig. 3).
The formation of H2 occurs predominantly in
nations for most of the other reactions, and the the reactions of H with H2S (R2) and, at lower
model employs theoretical estimates. Overall, oxygen concentrations, with SH ( R5); forma-
there are too many reactions with signicant rate tion of H2O is almost entirely via H2S + OH
uncertainty and high sensitivity coecients to (R27/R28), with OH being formed from
enable comprehensive model validation. There- H2S + O (R29). The high selectivity to H2 under
fore we have chosen to investigate how well the oxygen-rich conditions arises because most H is
model can simulate the detailed experimental itself generated by reaction of O with SH (R44)
observations through changes in rate parameters as shown in Fig. 3a, the model captures the
of limited set of reactions. For this purpose, we competition between H2 and H2O formation
consider the SH self-reactions forming (i) under these conditions. When the oxygen excess
H2S + S ( R3 and R4) which, together, have is reduced and the reactions of S2 species become
the greatest sensitivity coecients in the mecha- more important, more reactions contribute to the
nism under all conditions; and (ii) HSSH (R10) production of H but the experimental H2 selectiv-
which has a more pronounced eect when the ity declines slightly, an eect that is not captured
availability of oxygen is reduced. in the model - as with the overall consumption of
As expected, there is no pair of scaling factors H2S, the system is sensitive to a large number of
that enabled the model to reproduce the H2S con- reactions and there is no obvious adjustment that
sumption proles under all the experimental con- can address this decit. As is evident in Fig. 5,
ditions. However, the very high sensitivity of the some of the discrepancy at higher temperatures
system to the reactions chosen for adjustment arises from a failure to predict H2 consumption
means that only modest adjustments were in the quench zone.
required in all cases. The reactions R3/ R4
and R10 have opposite temperature dependence 5.4. Comparisons with other published data
so some ne-tuning of the conversion curves
(e.g. in Fig. 3) can be achieved by adjusting the A number of sulfur channels (R34, R6, R9
balance between these pathways. For the adjusted 10, R12R15) belonging to the H/S subset of
simulation results (solid lines) in Figs. 3 and 4, the Sendt et al. [8] have been updated. We therefore
scaling factor on R10 is always 3; on R3/ R4, checked the revised model against the validation
the factor is 1.5, 1.05 and 0.8 for curves (a), (b) targets used in that work. While the revised model
and (c), respectively. In Fig. 5, these factors are slightly underpredicts the overall reaction rates
1 and 2.8 for R10 and R3/ R4, respectively, under all conditions, the predictions were particu-
these values giving rise to more rapid ignition than larly sensitive only the rates of R7 (S2 + H +
predicted by the base case parameters. Here it M
HSS + M) and R19 (HSSH + H

should be noted that the overall consumption of H2S + SH). Doubling of each of these rate con-
H2S is completed in the isothermal zone of the stants was sucient to obtain satisfactory ts to
furnace in Fig. 5, the reaction proceeds strongly the data, as shown in the Supplementary data.
C. Zhou et al. / Proceedings of the Combustion Institute 34 (2013) 625632 631

When applied to the low-pressure fuel-lean (/ whereas on the other, 2SH


HSSH (R10)
 0.5) and stoichiometric (/ = 1) H2S ame data HSSH + SH
H2S + HSS (R20), HSS +
of Merryman and Levy [26,27], the base model SH
H2S + S2 ( R11) is a powerful chain ter-
captures the experimental proles of the all major mination sequence, especially when the oxygen
species except for H2 selectivity which is signi- excess is reduced. All the ow reactor data could
cantly overpredicted see Supplementary data. be modelled with changes to the rate constants
It is noteworthy that the experimental data show for R3/ R4 and R10 less than a factor of 3.
greater H2 selectivity in the lean ame. The factors required for this sub-optimal tting
Flame speeds estimated using the model are in are dierent for the dierent cases but the point
reasonable agreement with measurements of remains that the model does reproduce the exper-
Chamberlin and Clark [28], peak values being imental results for H2S when changes within the
58 cm s 1 versus 50 cm s 1, respectively. likely uncertainty bounds for the rate parameters
are made.
The selectivity of H2S oxidation to H2 in the
6. Conclusions ow reactor and in ames is overpredicted by the
model. This selectivity is controlled by the relative
Recent research into the fundamental kinetics concentrations of radicals, especially H, OH, O
of sulfur reactions has highlighted the need for and S, and is therefore sensitive also to many the
inclusion of new and revised pathways in the relative rates of the many reactions that determine
mechanism of H2S oxidation, particularly in rela- these concentrations. The apparent discrepancy in
tion to disulfur species such as S2, HSS, and the prediction of H2 proles in H2S oxidation sug-
HSSH. We have developed a comprehensive gests the need for further improvement in both the
kinetic model that addresses this need. experiments and the sulfur mechanism.
We have shown that H2S oxidation is strongly The importance of disulfur species in H2S oxi-
catalysed by silica surface but that this eect is dation even under fuel lean conditions is clearly
apparently negated by the application of a B2O3 demonstrated in this work.
coating. While it is not possible to prove conclu-
sively that this coating has no residual eect on
the course of the reaction, the characteristic tem- Acknowledgements
perature for reaction rises by some hundreds of
degrees when the coating is applied, making it The authors acknowledge the support of the
much more likely that gas-phase reactions domi- Australian Research Council for part of this
nate. We therefore report data for the atmo- work. Chenlai (Ryan) Zhou thanks the School
spheric pressure fuel-lean H2S oxidation (100 of Chemical and Biomolecular Engineering of
520 ppm H2S in 2001000 ppm O2; N2 ballast) the University of Sydney for the award of the
in a B2O3-coated reactor in the temperature range FH Loxton Postgraduate Scholarship.
9501150 K. Studies under stoichiometric and
fuel-rich conditions were precluded because of
the excessive build-up of elemental sulfur under Appendix A. Supplementary data
these conditions. The reaction shows up to 40%
selectivity to H2, even under the most fuel lean Supplementary data associated with this article
conditions, [O2]0/[H2S]0 = 10. can be found, in the online version, at http://
Predictions of the base model are in qualita- dx.doi.org/10.1016/j.proci.2012.05.083.
tively good agreement with the experiments, with
product proles shifted by no more than 50 K.
A striking feature of the mechanism is the large References
number of reactions to whose rate parameters
the overall oxidation is very sensitive. These reac- [1] C.L. Rasmussen, P. Glarborg, P. Marshall, Proc.
tions are those that are responsible for the forma- Combust. Inst. 31 (2007) 339347.
tion and destruction of radicals in the system and [2] I.A. Gargurevich, Ind. Eng. Chem. Res. 44 (2005)
increasingly include reactions of disulfur species 77067729.
[3] F.G. Cerru, A. Kronenburg, R.P. Lindstedt, Com-
as the ratio [O2]0/[H2S]0 is reduced. bust. Flame 146 (2006) 437455.
With a large choice of highly sensitive reac- [4] Y. Gao, C. Zhou, K. Sendt, B.S. Haynes, P.
tions, the model could be adjusted in many ways Marshall, Proc. Combust. Inst. 33 (2011) 459465.
to achieve a better quantitative t to the data. [5] Leeds University Sulphur Mechanism, version 5.2,
We have shown how minor variations in the rates available at http://www.chem.leeds.ac.uk/Combus-
of just two of the SH self-reaction channels allows tion/sox.htm/.
accurate description of H2S consumption proles. [6] A. Burcat, B. Ruscic, Ideal Gas Thermochemical
On the one hand, the sequence 2SH
H2S + S Database with Updates from Active Thermochemical
( R3/ R4), S + O2
SO + O (R60), SO + Tables, available at http://gareld.chem.elte.hu/
Burcat/burcat.html/.
O2
SO2 + O (R117) is strongly branching
632 C. Zhou et al. / Proceedings of the Combustion Institute 34 (2013) 625632

[7] K. Sendt, B.S. Haynes, J. Phys. Chem. A 109 (2005) [18] A. Goumri, J.D.R. Rocha, D. Laakso, C.E. Smith,
81808186. P. Marshall, J. Phys. Chem. A 103 (1999) 11328
[8] K. Sendt, M. Jazbec, B.S. Haynes, Proc. Combust. 11335.
Inst. 29 (2003) 24392446. [19] J.D. Garrido, M.Y. Ballester, Y. Orozco-Gonzalez,
[9] R.C. Shiell, X.K. Hu, Q.J. Hu, J.W. Hepburn, J. S. Canuto, J. Phys. Chem. A 115 (2011) 14531461.
Phys. Chem. A 104 (2000) 43394342. [20] M.A. Blitz, K.J. Hughes, M.J. Pilling, S.H. Rob-
[10] P.A. Denis, Chem. Phys. Lett. 402 (2005) ertson, J. Phys. Chem. A 110 (2006) 29963009.
289293. [21] R.J. Kee, F.M. Rupley, J.A. Miller, et al., CHEM-
[11] C. Zhou, K. Sendt, B.S. Haynes, J. Phys. Chem. A KIN: Release 4.1.1, Reaction Design, San Diego,
113 (2009) 29752981. CA, 2007.
[12] C. Zhou, Kinetic Study of The Oxidation of Hydro- [22] A. Egerton, D.R. Warren, Proc. R. Soc. London.
gen Sulde, Ph.D. thesis, The University of Sydney, Ser. A 204 (1951) 465476.
2009. [23] R.R. Baldwin, L. Mayor, Trans. Faraday Soc. 56
[13] L.A. Curtiss, K. Raghavachari, P.C. Redfern, V. (1960) 103114.
Rassolov, J.A. Pople, J. Chem. Phys. 109 (1998) [24] R.R. Baldwin, R.W. Walker, D.H. Langford,
77647776. Trans. Faraday Soc. 65 (1969) 792805.
[14] J. Li, Z.W. Zhao, A. Kazakov, F.L. Dryer, Int. J. [25] Z.H. Lodhl, R.W. Walker, J. Chem. Soc. Faraday
Chem. Kinet. 36 (2004) 566575. Trans. 87 (1991) 681689.
[15] N.L. Garland, Chem. Phys. Lett. 290 (1998) 385 [26] E.L. Merryman, A. Levy, J. Air Pollut. Control
390. Assoc. 17 (1967) 800806.
[16] L. Hindiyarti, P. Glarborg, P. Marshall, J. Phys. [27] E.L. Merryman, A. Levy, J. Phys. Chem. 76 (1972)
Chem. A 111 (2007) 39843991. 19251931.
[17] K. Sendt, B.S. Haynes, Proc. Combust. Inst. 31 [28] D.S. Chamberlin, D.R. Clarke, Ind. Eng. Chem. 20
(2007) 257265. (1928) 10161018.

Você também pode gostar