Combustion and Flame 158 (2011) 1049–1058

Contents lists available at ScienceDirect

Combustion and Flame

journal homepage: www.elsevier .com/locate /combustflame

Pyrolysis and oxidation of ethyl methyl sulfide in a flow reactor

Xin Zheng a, E.M. Fisher a,⇑, F.C. Gouldin a, J.W. Bozzelli b

a Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, NY 14853, United Statesb Department of Chemistry and Environmental Science, New Jersey Institute of Technology, Newark, NJ 07101, United States

a r t i c l e i n f o a b s t r a c t

Article history:Received 28 May 2009Received in revised form 9 July 2010Accepted 20 October 2010Available online 22 November 2010

Keywords:OrganosulfurReaction mechanismPyrolysisOxidationChemical warfare agentTurbulent flow reactor

0010-2180/$ - see front matter � 2010 The Combustdoi:10.1016/j.combustflame.2010.10.018

⇑ Corresponding author. Fax: +1 607 255 1222.E-mail address: emf4@cornell.edu (E.M. Fisher).

The reactions and kinetics of ethyl methyl sulfide (CH3CH2ASACH3, abbreviation CCSC), a simulant forthe chemical warfare agent sulfur mustard, were studied at temperatures of 630–740 �C, under highlydiluted pyrolysis and oxidation conditions at one atmosphere in a turbulent flow reactor. The loss ofthe ethyl methyl sulfide and the formation of intermediates and products were correlated with timeand temperature. Destruction efficiencies of 50% and 99% were observed for pyrolysis and oxidation,respectively, at 740 �C with a residence time of 0.06 s. For pyrolysis, ethylene, ethane, and methane weredetected at significant levels. In addition to these species, carbon monoxide, carbon dioxide, sulfur diox-ide, and formaldehyde were detected for oxidation. Conversions of ethyl methyl sulfide were observed tobe significantly slower than observed previously for diethyl sulfide; explanations for this observation arepostulated, based on: (1) lower hydrogen abstraction rates or on (2) lower hydrogen atom production as aresult of thermal decomposition pathways. Initial decomposition reactions and production pathways forimportant species observed in the experiments are discussed on a basis of thermochemistry.

� 2010 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1. Introduction

Sulfur mustard, Bis(2-chloroethyl) sulfide (ClCH2CH2ASACH2-CH2Cl), also known as mustard gas, is a cytotoxic chemical withthe ability to form blisters on exposed skin. As a principal compo-nent of the chemical warfare agent (CWA) HD, sulfur mustard wasstockpiled in the form of HD by major countries and distributedover the world during the cold war [1,2]. As part of the 1993 Chem-ical Weapons Convention (CWC) [3], HD stockpiles must bedestroyed by April 29, 2012. A significant number of studies onthe destruction of mustard gas via different methods have beenperformed over the past 15 years with a focus on efficiency andcompleteness of destruction, and minimal production of by-prod-ucts [4–10]. Incineration has been chosen as a method with highdestruction efficiency, high throughput, and high cost effectiveness[11]. Analysis of a report published in 2006 [12] shows that at thattime approximately 13,000 tons of HD, more than 60% of stockpilesleft in the US, were to be disposed by incineration by 2012. The USgovernment has sponsored studies to monitor incinerator effluentand to build detailed kinetic mechanisms to model the incinerationof mustard agent and to predict the safety of their incinerators[13,14]. Nonetheless, there is still public concern over the potentialformation of toxic emissions.

Although experiments on actual chemical warfare agents arenot practical in university laboratories because of safety consider-

ion Institute. Published by Elsevier

ations, studies with simulants provide added validation, and thisserves to improve confidence in model predictions and thus allevi-ate some public concern over the incineration process. The presentwork extended our previous study of diethyl sulfide [15] by addingexperimental data on pyrolysis and oxidation of ethyl methyl sul-fide. The study allows us to compare the effect of methyl group onthe destruction efficiency and product distribution with that ofethyl group from the study on diethyl sulfide [15]. The data willbe used to validate the methyl sulfur reactions under development.In the forthcoming sections of this paper, we qualitatively discussproduct formation and destruction efficiency for ethyl methylsulfide in light of the kinetic mechanism previously developedfor diethyl sulfide and the available thermochemistry [15].

2. Experimental apparatus and methodology

The present work was performed in a 4.5-cm-ID tubular flowreactor, which is fully documented in a previous study [15], andis briefly described here. Two flow streams were mixed in the reac-tor: a main flow and a secondary simulant carrier flow. The mainflow was approximately 95% by mass of the total reactor flow; itwas a turbulent preheated nitrogen flow which became fullydeveloped as it flowed through a 1 m long development section.The secondary carrier flow varied with experimental conditionsas follows. In the pyrolysis experiments, the secondary flow wasnitrogen, while in the oxidation experiments, it was laboratorycompressed air purified by using a Whatman Model 75-52 purgegas generator to remove carbon dioxide and moisture. Ethyl

Inc. All rights reserved.

Table 1BFlow parameters of oxidation experiments.

NominalT (�C)

Mainflow(slpm)

Secondaryflow(slpm)

Velocity(m/s)

ReD O2

loading(ppm)

CCSCloading(ppm)

630 330 14.1 11.12 5020 9321 150670 341 15.3 12.02 5060 9270 150700 347 16.0 12.64 5060 9366 150740 350 16.8 13.30 4990 9472 150

Note: The loading uncertainties are approximately 4% and 3.3% for the CCSC and O2

respectively.

1050 X. Zheng et al. / Combustion and Flame 158 (2011) 1049–1058

methyl sulfide was injected into the secondary flow through aninjection port via a syringe pump (Harvard Apparatus modelPHD2000 with 1% uncertainty). Carbon monoxide (CO) was alsoadded to the secondary flow as an inert tracer [15] to determinemixing completeness in the reactor and dilution in the samplingsystem during the pyrolysis experiments. No tracer was used inthe oxidation experiments, in which mixing conditions were as-sumed to be the same as those in the pyrolysis experiments havingthe same heater settings and flow rates. The secondary flow wasinjected into the main carrier flow through four open-ended quartztube injectors. GC/MS analysis of gas samples withdrawn from thesecondary flow showed no evidence of chemical reactions up-stream of the injection point. The main and secondary flows subse-quently mixed, and reactions occurred along the length of a tubularquartz reaction section. This reaction section, 4.5 cm in diameterand 1 m long, was sitting inside a stainless steel pipe which wascovered by electric resistance heaters and insulation material toeffectively maintain the gas temperature at a fixed value.

The experiments were performed at conditions very similar tothose used in a previous study of diethyl sulfide [15], at four differ-ent operating temperatures between 630 �C and 740 �C and at1 atm pressure. The detailed experimental flow parameters andchemical loadings for each operating temperature are shown inTables 1A and 1B, and the uncertainties of the CCSC and O2 load-ings are provided in the table note. As shown in the tables, the flowof the reaction section was turbulent with a Reynolds number ofaround 5000. The initial loading of ethyl methyl sulfide in thereaction section was 150 ppm for all operating conditions. In thisdocument, ppm refers to parts per million on a volume (or molar)basis. In the oxidation experiments, the initial loading of oxygen(O2) was approximately 9000 ppm in the reaction section, whichproduced a fuel-lean condition with an equivalence ratio ofapproximately 0.1 based on SO2, H2O, and CO2 as final products.Gas temperature profiles were adopted from the previous study[15] as the heater settings and flowrates were the same as thoseof the current experiments. The temperature measurements wereperformed without the injection of the CO tracer and sulfur com-pound flows. However, any error in temperature readings due toenthalpies of reaction was negligible under the highly dilute condi-tions of the experiments. Each experiment is identified with a‘‘nominal temperature,’’ the average centerline temperature inthe sampling region, rounded to the nearest 10 �C. In this region,temperatures vary by approximately ±10 �C. Upstream of the firstsampling port, however, there is a region of low temperature(typically reaching 80 �C below the nominal temperature) wherethe cold secondary flow mixes into the preheated main flow. Tem-perature profiles are given in reference [15], and were repeatable.Repeated measurements were performed several months afterthose reported, at two temperature conditions; temperatures werewithin the measurement uncertainty (1%) of the original ones.

The intent of choosing such experimental conditions was two-fold, (1) understanding the initiation of the destruction of sulfurmustard stimulants and (2) exploring routes relevant to emissionsunder off-design incineration modes, such as low temperaturequenching, and fuel rich conditions due to inhomogeneous mixing.

Table 1AFlow parameters of pyrolysis experiments.

NominalT (�C)

Mainflow(slpm)

Secondaryflow(slpm)

Velocity(m/s)

ReD COloading(ppm)

CCSCloading(ppm)

630 330 14.1 11.12 5020 68.6 150670 341 15.3 12.02 5060 66.3 150700 347 16.0 12.64 5060 65.1 150740 350 16.8 13.30 4990 64.4 150

In the experiments, all of the independent flow parameters de-scribed above, including main flow rate, secondary flow rate, walltemperatures of the reaction section, tracer flow rate, infusion rateof liquid ethyl methyl sulfide, as well as the sample dilution flowrate and sampling pressure, were electronically monitored andindependently controlled. The flow reactor system was given suffi-cient time to reach steady conditions prior to sampling.

Gas samples were extracted at four different positions on thecenterline of the reaction section and analyzed by GC/MS and FTIR.Each sampling probe (Fig. 1) consisted of two radially orientedconcentric quartz tubes with respective 4 and 1 mm ID. A nitrogenflow with approximately the same flow rate as the gas sample flowwas introduced through the inner tube, to quench reactions. Gassampled from the flow reactor mixed with the quenching nitrogenflow and was drawn out of the flow reactor through the annularregion in the probe. The probe design and testing are described in[16]. The diluted sample flow passed through Teflon� transfer linesto a Nicolet Model 6300 FTIR or an Ultra Trace/DSQII GC/MS for anal-ysis. Details regarding analytical methods are available in [15].Detection limits were 0.3 ppm for ethylene, 0.26 ppm for methane,0.25 ppm for ethane, 0.23 ppm for carbon dioxide, 0.20 ppm forcarbon monoxide, 0.25 ppm for formaldehyde, 0.6 ppm for sulfurdioxide and 1.3 ppm for ethyl methyl sulfide.

Species concentrations in the gas samples were affected by themixing of the secondary flow into the main flow and by dilution inthe sampling system, as well as by the chemical reactions in thereaction section. Measurements of tracer concentrations made itpossible to account for the mixing and dilution effects [15].

3. Results

Destruction efficiencies at the last sampling port for pyrolysisand oxidation of ethyl methyl sulfide are presented in Fig. 2, alongwith those reported previously for pyrolysis of diethyl sulfide [15].Destruction efficiency is defined as the fraction of the parent com-pound destroyed at a given position. Species concentration profilesfrom both the pyrolysis and oxidation experiments of ethyl methylsulfide are reported in Figs. 3–10, for all species detected at levelsof 1 ppm or higher. Figures 3 and 9 show profiles of species thatwere detected by both FTIR and GC/MS. In these figures, hollowsymbols represent FTIR data, while filled symbols represent GC/MS data.

As shown in Fig. 2, for pyrolysis of ethyl methyl sulfide, destruc-tion efficiency at the last sampling port of the reaction section(residence times between 0.06 s and 0.072 s) increased with theoperating temperature, from approximately 11% at the 630 �C oper-ating temperature, to 16% at 670 �C, 25% at 700 �C, and 50% at 740 �C.Compared to the pyrolysis of ethyl methyl sulfide, destruction effi-ciency was enhanced by more than a factor of two in the oxidationof ethyl methyl sulfide, shown in Fig. 2. Destruction efficiency wasapproximately 25% at the 630 �C operating temperature, 41% at670 �C, and 67% at 700 �C. Nearly complete destruction was ob-served at the 740 �C operating temperature.

Fig. 1. Sampling probe.

Fig. 2. Destruction efficiencies of ethyl methyl sulfide by pyrolysis and oxidation and destruction efficiencies of diethyl sulfide by pyrolysis, measured at the last samplingport of the flow reactor.

X. Zheng et al. / Combustion and Flame 158 (2011) 1049–1058 1051

The primary products observed in both pyrolysis and oxidationexperiments of ethyl methyl sulfide were ethylene (C2H4) (Fig. 4),ethane (C2H6) (Fig. 5), and methane (CH4) (Fig. 6). Carbon monox-ide (CO) (Fig. 7), carbon dioxide (CO2) (Fig. 8), sulfur dioxide (SO2)(Fig. 9), and formaldehyde (CH2�O) (Fig. 10) were detected in theoxidation experiments only. The following species: acetylene(C2H2), carbon disulfide (CS2), thiophene (CycAC4H4S), dimethyldisulfide (CH3SSCH3) and ethyl methyl disulfide (C2H5SSCH3), weredetected but at levels below 1 ppm. H2O was detected, but was dif-ficult to quantify because of fluctuating background levels in theFTIR instrument. In addition, condensates were observed on theinternal surface of the sampling probes during both pyrolysis andoxidation experiments. A condensate sample was obtained byflushing the sampling probes with methylene chloride after theexperiments. Cyclooctasulfur (S8) was identified by GC/MS with aDB5 column, and a small unidentified peak was also present.Thioformaldehyde trimers (CAS No. 291-21-4), an anticipatedcomponent of the condensates [17], were detectable by this GC/MS method, but not observed in the solution. Elemental balanceswere calculated based on the species concentrations as presentin Figs. 3–10. For the pyrolysis conditions, the sulfur balance de-creased linearly with the destruction efficiency down to 50%, asno sulfur-containing products were detected. Carbon and hydrogenbalances were approximately 80% at the maximum destructionefficiency of 50%. In contrast, lower elemental balances, 35% forsulfur, 70% for carbon and 40% for hydrogen were found at themaximum destruction efficiency of 100% for the oxidation condi-tions. Poor elemental balances are probably due to a few undetect-able species as discussed in the following section.

4. Discussion

4.1. Mechanism of pyrolysis of ethyl methyl sulfide

Because the molecular structure and bond dissociation energies(BDEs) of ethyl methyl sulfide, (shown in Fig. 11) resemble those ofdiethyl sulfide (CH3CH2ASACH2CH3, abbreviated CCSCC), it is notsurprising that during the pyrolysis experiments the same prod-ucts were observed for these two chemicals [15]. First we quanti-tatively explain the formation of products from ethyl methylsulfide with the reaction scheme discussed below, developed byanalogy to the detailed mechanism for pyrolysis of diethyl sulfide[15]. Then later in this section, we focus on the destruction effi-ciency of CCSC. We compare it to the destruction efficiency ofCCSCC, and use the reaction scheme to provide possible explana-tions for the differences.

4.1.1. Decomposition pathwaysInitial decomposition of either CCSCC or CCSC occurs via CAS

bond cleavage. According to our previous study [15], CCSCC pyro-lysis is initiated through C-S bond cleavage.

C2H5ASAC2H5 $ C2H5 þ �SAC2H5 ð1Þ

It is important to note that Reaction (1) has two identical pathsor a reaction degeneracy of two. In contrast, due to the asymmetricstructure of ethyl methyl sulfide, CAS bond cleavage can occurthrough the two different unimolecular decomposition stepsshown below.

Fig. 3. Mole fraction of ethyl methyl sulfide in the pyrolysis and oxidation experiments. Hollow symbols represent FTIR data while filled symbols represent GC/MS data.Residence times are mean values computed from mass flow rates and gas densities.

1052 X. Zheng et al. / Combustion and Flame 158 (2011) 1049–1058

C2H5ASACH3 $ C2H5 þ �SACH3 ð2Þ

C2H5ASACH3 $ �SAC2H5 þ CH3 ð3Þ

As shown in Fig. 11, BDEs for the CAS bonds broken in Reactions(1)–(3) are the same within uncertainty: The BDEs are 72.4 ± 0.1and 73.4 ± 1.5 kcal/mol for the bonds broken in Reactions (2) and(3), compared to a BDE of 72.5 ± 1.5 cal/mol for the bond brokenin Reaction (1). In addition, initiation can also occur via CAC orCAH bond cleavage. However these reactions are relatively unim-portant as the CAC bond energies are �82.0 ± 2.0 kcal/mol and theCAH bond energies are �100 kcal/mol, considerably higher thanthose of the CAS bonds.

The ethyl (C2H5), ethylthio (�SAC2H5), and methylthio (�SACH3)radicals formed in these initiation steps (Reactions (2) and (3))can undergo b-scission to produce ethylene and thioformaldehyde(CH2�S) along with the methyl radical and H atom respectively.See Reactions (4)–(6).

C2H5 $ C2H4 þH ð4Þ

�SAC2H5 $ CH2 � Sþ CH3 ð5Þ

�SACH3 $ CH2 � SþH ð6Þ

The resulting ethylene was a major product observed in theexperiments (Fig. 4). The thioformaldehyde product was notdetected in the experiments, and its fate is discussed below

As the radical pool becomes more established from the initia-tion and initial beta scission reactions, H abstraction (Reactions

(7)–(9)) is expected to become an important route for the destruc-tion of ethyl methyl sulfide.

C2H5ASACH3 þ R$ RHþ C2H5ASA�CH2 ð7Þ

C2H5ASACH3 þ R$ RHþ �CH2CH2ASACH3 ð8Þ

C2H5ASACH3 þ R$ RHþ CH3�CHASACH3 ð9Þ

R is H atom or CH3 radical in Reactions (7)–(9). In addition, weexpect H abstraction from the a carbon (carbon bonded to sulfuratom) to be the fastest abstraction reaction. This expectation isbased on the thermochemistry of CCSCC developed in our previousstudy [15] which gives a lower BDE of CAH on the a carbon thanthat on the b carbon, in agreement with hydrocarbon thermochem-istry [18]. However, no kinetic data are currently available forReactions (7)–(9). Two stable products, H2 and CH4, are predicted.CH4 is detected at levels up to 10 ppm (Fig. 6). Simulations ofdiethyl sulfide pyrolysis [15] indicate that hydrogen (H2) yieldfrom the reactions above should be higher than methane yield, be-cause abstraction rates by hydrogen atoms are approximately oneorder of magnitude higher than those by methyl radicals. However,H2 is not detectable by our experimental techniques.

The radicals produced by these reactions undergo beta scission,producing ethyl, methyl, and methylthio radicals along with thio-formaldehyde, thioacetaldehyde (CH3CH�S), and ethylene:

C2H5ASA�CH2 $ C2H5 þ CH2 � S ð10Þ

�CH2CH2ASACH3 $ C2H4 þ �SACH3 ð11Þ

Fig. 4. Mole fraction of ethylene in the pyrolysis and oxidation experiments. Residence times are mean values computed from mass flow rates and gas densities.

X. Zheng et al. / Combustion and Flame 158 (2011) 1049–1058 1053

CH3�CHASACH3 $ CH3CH � Sþ CH3 ð12Þ

Further beta scission of the methylthio and ethyl radicals yieldsadditional thioformaldehyde and ethylene. Recombination of CH3

radicals can form ethane [19], a species observed during the exper-iments at levels up to 10 ppm.

Two thioaldehydes, thioformaldehyde and thioacetaldehyde,are expected to be major products on the basis of the reactionslisted above and subsequent beta scission. Because thioaldehydeswere not detected, it appears that their destruction reactions areimportant. Possible routes for thioaldehyde destruction are specu-lative. Under pyrolytic conditions, these routes include heteroge-neous reactions, either in the flow reactor or in the samplingsystem, hydrogen abstraction or addition reactions, and dispropor-tionation reactions. With a dearth of detected sulfur-containingproducts, little evidence exists to distinguish between the impor-tance of these routes. The importance of disproportionation reac-tions, for instance Reaction (13):

CH2 � Sþ CH2 � S$ C2H4 þ S2 ð13Þis supported by the fact that S2 is a precursor [20] for the solidcyclooctasulfur observed in the probe rinsate. However, kinetic sim-ulations in our previous study of CCSCC pyrolysis [15] indicate thatdisproportionation reactions do not contribute significantly to thio-aldehyde destruction under the conditions studied.

5. Destruction efficiency

The destruction efficiency of ethyl methyl sulfide depends onthe rates of the reactions above and the reactivity of their products.

Of particular importance are the reactions resulting in branching,(net positive production of radicals), e.g. unimolecular decomposi-tion. Since the BDE’s of the initiation steps of ethyl methyl sulfidedecomposition are almost the same as that of diethyl sulfide (Reac-tions (1)–(3); Fig. 11), destruction efficiencies of diethyl sulfide areexpected to be similar to those of ethyl methyl sulfide. In contrastto this expectation, the destruction efficiency of diethyl sulfide wassignificantly higher than that of ethyl methyl sulfide except at the630 �C condition, as can be seen in Fig. 2. Destruction efficiencies atthe last sampling port were 15% at the 630 �C condition, 30% at670 �C, 50% at 700 �C, and 80% at 740 �C for diethyl sulfide, and11%, 16%, 25% and 50% for ethyl methyl sulfide at the correspond-ing temperature and residence time conditions.

In the absence of a detailed mechanism for ethyl methyl sulfide,we can propose two possible explanations for the relatively slowthermal destruction of ethyl methyl sulfide, one involving hydro-gen abstraction and one involving initial unimolecular dissociationreactions.

CCSCC and CCSC’s hydrogen abstraction rates are expected todiffer as that CCSCC has a larger number of weaker CAH bondsas well as a larger total number of hydrogen atoms available toabstract. Thus, at higher temperatures, once the radical pool isestablished, the observed higher destruction efficiencies for CCSCCare supported by the expected higher hydrogen abstraction rates.

The effect of differences in initial unimolecular decompositionssteps is more complex and depends on the relative importance ofthe two different initiation steps for CCSC, as described below.Differences in the number of relatively reactive hydrogen atomsvs relatively nonreactive methyl radicals ultimately produced

Fig. 5. Mole fraction of ethane in the pyrolysis and oxidation experiments. Residence times are mean values computed from mass flow rates and gas densities.

1054 X. Zheng et al. / Combustion and Flame 158 (2011) 1049–1058

through the different initiation steps should influence destructionefficiencies. The first path (Reactions (2), (4), and (6)) is

C2H5ASACH3 $ C2H5 þ �SACH3 ð2Þ

followed by

C2H5 $ C2H4 þH ð4Þ

and by�SACH3 $ CH2 � SþH ð6Þ

This path produces two H atoms and no CH3 radical per CCSC de-stroyed. The second path (Reactions (3) and (5)) is

C2H5ASACH3 $ �SAC2H5 þ CH3 ð3Þ

followed by�SAC2H5 $ CH2 � Sþ CH3 ð5Þ

This path produces no H atoms and two CH3 radicals per CCSCdestroyed.

In contrast, CCSCC has two degenerate paths, i.e. two identicalC-S cleavage reactions:

C2H5ASAC2H5 $ C2H5 þ �SAC2H5 ð1Þ

and each path ultimately produces one H atom and one CH3 radicalper CCSCC destroyed.

C2H5 $ C2H4 þH ð4Þ

�SAC2H5 $ CH2 � Sþ CH3 ð5Þ

Overall the two identical paths produce two H atoms and two CH3

radicals. Thus, the observed lower destruction efficiency for CCSCrelative to CCSCC could be explained if Reaction (3), ultimately pro-ducing CH3 radicals, is faster than Reaction (2), ultimately producingH atoms. Kinetic mechanism development is under way to estimatethe rates of these reactions, and currently available data does notclearly indicate which rate is higher. Figure 11 shows that the CH3ASbond is �1 kcal mole stronger than the C2H5AS bond, thus the acti-vation energy term, in the dissociation rate constants, would actu-ally be higher for Reaction (2), which would not explain thedifferent destruction efficiencies observed. However, the CH3ASbond energy has a relatively large reported uncertainty as shownin Fig. 11. Also, it is established that dissociation reactions producinga methyl radical (e.g. Reaction (3)) have about a two to three timeshigher pre-exponential factor than those producing larger hydrocar-bon fragments (e.g. Reaction (2)) [21].

6. Mechanism of oxidation of ethyl methyl sulfide

Initiation of the destruction of ethyl methyl sulfide in the oxida-tion experiments includes CAS bond cleavage (Reactions (2) and(3)), as in the pyrolysis regime. An additional path occurs in thepresence of O2, which is the abstraction of a H atom from thehydrocarbon backbone:

O2 þ CCSC$ HO2 þ CC�SC ð14Þ

Which, on the basis of hydrocarbon kinetics [22], is approxi-mately 50 kcal/mol endothermic, with only a 1 kcal/mol barrierover the endothermicity of the reaction.

Fig. 6. Mole fraction of methane in the pyrolysis and oxidation experiments. Residence times are mean values computed from mass flow rates and gas densities.

Fig. 7. Mole fraction of carbon monoxide in the oxidation experiments. Residence times are mean values computed from mass flow rates and gas densities.

X. Zheng et al. / Combustion and Flame 158 (2011) 1049–1058 1055

Once the radical pool is established, abstraction becomes asimportant as these initiation steps. The main difference betweenoxidation and pyrolysis, in the abstraction reactions, is the addi-tional contribution to hydrogen abstraction by hydroxyl (OH) rad-icals and oxygen (O) atoms. Other differences involve terminationreactions:

HO2 þHO2 $ H2O2 þ O2 ð15Þ

Under fuel-lean conditions, the abstraction reactions becomeespecially important as there are high concentrations of hydroxylradicals and oxygen atoms. Among these reactions, abstractionby hydroxyl radical is more important because of its known higherreactivity. Formation of H2O from the abstraction by OH radicals isexpected, but attempts to quantify H2O in the gas samples have notbeen successful to date. b-scission of the radicals resulting from theinitiation and abstraction reactions produces thioacetaldehyde,

Fig. 8. Mole fraction of carbon dioxide in the oxidation experiments. Residence times are mean values computed from mass flow rates and gas densities.

Fig. 9. Mole fraction of sulfur dioxide in the oxidation experiments. Residence times are mean values computed from mass flow rates and gas densities.

Fig. 10. Mole fraction of formaldehyde in the oxidation experiments. Residence times are mean values computed from mass flow rates and gas densities.

1056 X. Zheng et al. / Combustion and Flame 158 (2011) 1049–1058

thioformaldehyde and ethylene (Reactions (10)–(12)). Alterna-tively, in the oxidative environment, the radicals can combine withO2 to produce alkylthioperoxy, alkylperoxy or alkylthioalkylperoxy

radicals, which can isomerize to form new products. Some of thesenew products can lead to chain branching. However, data fromstudies of the oxidation of a similar molecule, dimethyl ether, in

Fig. 11. Bond dissociation energies (BDEs) (cal/mole) of CAS and CAC bonds inethyl methyl sulfide and diethyl sulfide, the BDE of CAC bond of diethyl sulfide,which is marked with �, is not available and adopted from ethylthiol (CH3ACH2SH),all values are from Luo [18].

X. Zheng et al. / Combustion and Flame 158 (2011) 1049–1058 1057

a flow reactor [23] suggests that this route is not important in thetemperature range studied here.

As in the pyrolytic case, the fate of the thioaldehydes involves anumber of possible reactions. Under oxidative conditions, there isevidence of the importance of H abstraction: We can postulate aseries of reactions, starting with H abstraction, that explain theapproximately equal quantities of SO2 and CO observed experi-mentally in all but the highest-temperature oxidative experiments.Specifically, as shown in Reaction (16) below [24] for thioformal-dehyde, H abstraction forms the thioformyl radical (H�CS). The sub-sequent reaction of the resulting thioformyl radical with molecularoxygen produces the formyl radical (H�CO) along with sulfur mon-oxide (SO) (Reaction (17)), and rapid oxidation of unstable SOforms sulfur dioxide (SO2). SO2 is observed experimentally (Fig. 9).

CH2 � Sþ R$ H�CSþ RH ð16Þ

H�CSþ O2 $ H�COþ SO ð17Þ

SOþ O2 $ SO2 þ O ð18Þ

The formyl radical is an unstable species. It undergoes unimo-lecular dissociation (Reaction (19)) and bimolecular reactions(Reactions (20)–(22)) [22].

H�CO$ Hþ CO ð19Þ

H�COþ O2 $ COþHO2 ð20Þ

H�COþ OH$ H2Oþ CO ð21Þ

H�COþ O$ OHþ CO ð22Þ

Carbon monoxide was an important species observed at highlevels during the oxidation experiments (Fig. 7).

From the reaction routes above (Reactions (16)–(22)), approxi-mately equal yields of carbon monoxide and sulfur dioxide are ex-pected, which is consistent with experimental observations exceptat the highest temperature, as shown in Figs. 7 and 9. At the high-est temperature studied, CO is more abundant than SO2, probablybecause the creation of CO by other routes, for example fromC2H4, C2H6, and CH2�O, becomes important. Some oxidation ofCO to form CO2 is observed, but this process is slow under the pres-ent conditions, as can be seen from the low levels of CO2 profiles inFig. 8.

Ethylene is the most abundant product observed in the oxida-tion experiments at the 630 �C, 670 �C and 700 �C operating tem-peratures, where ethylene concentration increased approximatelylinearly with the destruction efficiency. As shown in Fig. 4, thepresence of oxygen doubled ethylene levels from the values inthe pyrolysis experiments at those temperatures, just as it doubledthe destruction efficiency for ethyl methyl sulfide. In contrast, dueto ethylene destruction reactions that become important at the740 �C condition, ethylene levels no longer increase with thedestruction of ethyl methyl sulfide at that temperature (Fig. 4).

Destruction of ethylene and subsequent oxidation reactions fol-low well established pathways found in the hydrocarbon combus-

tion kinetics literature. An important reaction route for ethylene isthe conversion to carbon monoxide via the reactions below [25]

C2H4 þ O$ CH3 þH�CO ð23Þ

H�CO$ COþH ð24Þ

Another destruction route for ethylene is H abstraction leadingto vinyl radical, and subsequent b scission producing acetylene[24], which was a minor species observed in both oxidative andpyrolytic experiments.

C2H4 þ R $ C2H3 þ RH ð25Þ

�Cþ 2H3 $ C2H2 þH ð26Þ

The last important species observed during the oxidation exper-iments is formaldehyde (CH2�O), Fig. 10. Like ethylene, its level in-creases with CCSC destruction efficiency at all but the highesttemperature, where its profile exhibits a maximum. Well knowncombustion reactions [26] can explain CH2�O formation fromCH3, via CH3O, as well as the destruction of CH2�O to form CO.

7. Conclusions

The pyrolysis and oxidation of ethyl methyl sulfide were stud-ied in a turbulent flow reactor under dilute conditions at four dif-ferent operating temperatures. The primary stable productsobserved during both pyrolysis and oxidation experiments wereethylene, methane, and ethane. In addition to these species, carbonmonoxide, carbon dioxide, sulfur dioxide, and formaldehyde weredetected in the oxidation experiments. By analogy to pyrolysis ofdiethyl sulfide, initiation of destruction of ethyl methyl sulfide oc-curs via CAS cleavage, and abstraction reactions become importantonce the radical pool is established. b-scission of the resulting rad-icals forms ethylene, thioformaldehyde and thioacetaldehyde. Sub-sequent abstraction and b-scission reactions convert these speciesto formaldehyde, sulfur dioxide, carbon monoxide and carbondioxide. Significantly slower rates of pyrolysis of ethyl methyl sul-fide were observed than previously observed for diethyl sulfide;possible explanations for the different destruction efficiencies in-clude lower hydrogen abstraction rates, and lower hydrogen atomproduction as a result of thermal decomposition pathways. Furtherinvestigations are needed to build kinetic data for the reactions, inparticular the initiation and abstraction steps associated with thedestruction of ethyl methyl sulfide, and the fate of thioaldehydeintermediates.

Acknowledgements

The authors acknowledge the support of the US Army ResearchOffice under Contract W911NF0410120 and through instrumentGrant DURIP W911NF0610142.

References

[1] Y.C. Yang, Chem. Ind. 1 (1995) 334–337.[2] L.R. Ember, Chem. Eng. News 68 (33) (1990) 9–19.[3] Text of Chemical Weapons Convention, <http://www.opcw.org/chemical-

weapons-convention/>.[4] Y.C. Yang, F.J. Berg, L.L. Szafraniec, W.T. Beaudry, C.A. Bunton, A. Kumar,

J. Chem. Soc. Perkin Trans. 2 (1997) 607–613.[5] Y.C. Yang, J.A. Baker, J. Richard Ward, Chem. Rev. 92 (1992) 1729–1743.[6] A.V. Vorontsov, L. Davydov, E.P. Reddy, C. Lion, E.N. Savinov, P.G. Smirniotis,

New J. Chem. 26 (2002) 732–744.[7] R.S. Davidson, J.E. Pratt, Tetrahedron Lett. 52 (1983) 5903–5906.[8] M.A. Y.S. Kim, A.A. Abdel-Wahab, M. Dulay, Catal. Lett. 5 (1990) 369–376.[9] D.V. Kozlov, A.V. Voronsov, P.G. Smirniotis, E.N. Savinov, Appl. Catal.,

B: Environ. 42 (2003) 77–87.[10] A.V. Vorontsov, C. Lion, E.N. Savinov, P.G. Smirniotis, J. Catal. 220 (2003) 414–

423.

1058 X. Zheng et al. / Combustion and Flame 158 (2011) 1049–1058

[11] J.F. Bunnett, Pure Appl. Chem. 67 (1995) 841–858.[12] J. Siegel, Bull. At. Sci. 62 (2006) 24–25.[13] M.J. Bockelie, M.K. Denison, Engineering Design Software for Military

Incinerators, Report No. DAAD19-01-C0050, US Army Research Office.[14] C.J. Montgomery, M.J. Bockelie, A.F. Sarofim, J. Lee, J.W. Bozzelli,

Thermochemical properties, reaction paths and kinetic mechanism forsulfur-chloro hydrocarbon combustion: part I: thermochemistry andpyrolysis of chlorosulfides, in: American Flame Research CommitteeInternational Symposium on Combustion, Livermore, CA, October 16–17, 2003

[15] X. Zheng, E.M. Fisher, F.C. Gouldin, L. Zhu, J.W. Bozzelli, Proc. Combust. Inst. 32(2009) 469–476.

[16] E.J.P. Zegers, Flow Reactor Pyrolysis of Alkyl Phosphates and Phosphonates,Ph.D. Thesis, Cornell University, 1997.

[17] W.S. Chin, B.W. Ek, C.Y. Mok, H.H. Huang, J. Chem. Soc. Perkin Trans. 2 (1994)883–889.

[18] Y.R. Luo, Handbook of Bond Dissociation Energies in Organic Compounds, CRCPress, Boca Raton, Florida, USA, 2003.

[19] B.S. Wang, H. Hou, L.M. Yoder, J.T. Muckerman, C. Fockenberg, J. Phys. Chem. A107 (2003) 11414–11426.

[20] I.I. Zakharov, A.N. Startcov, O.V. Vorochina, A.V. Pachigreva, N.A. Chashkova,V.N. Parmon, Russ. J. Phys. Chem. 80 (2006) 1403–1410.

[21] A.M. Dean, J. Phys. Chem. 89 (1985) 4600–4608.[22] D.L. Baulch, C.J. Cobos, R. A Cox, C. Esser, P. Frank, Th. Just, J.A. Kerr, M.J. Pilling,

J. Troe, R.W. Walker, J. Warnatz, J. Phys. Chem. Ref. Data 21 (1992) 411–429.[23] S.L. Fischer, F.L. Dryer, H.J. Curran, Int. J. Chem. Kinet. 32 (2000) 713–740.[24] I. Barnes, K.H. Becker, I. Patroescu, Atmos. Environ. 30 (1996) 1805–1814.[25] Y. Hidaka, T. Nishimori, K. Sato, Y. Henmi, R. Okuda, K. Inami, T. Higashihara,

Combust. Flame 117 (1999) 755–776.[26] W. Tsang, R.F. Hampson, J. Phys. Chem. Ref. Data 15 (1986) 1087–1279.