Kinetic and mechanistic study of the reaction of atomic chlorine with

J. M. Nicovich, S. Parthasarathy, F. D. Pope, A. T. Pegus, M. L. McKee, and P. H. Wine. The Journal of Physical Chemistry A 2006 110 (21), 6874-6885...
0 downloads 0 Views 2MB Size
9875

J. Phys. Chem. 1992,96,9875-9883 concentration of bromate and R discussed in section 3: S = 2 k ~ l / k ~ 6 / & ~=39.52 & ~ x 10"

b = 2kB6kB9[R]/k~lkW[BrOS-] = 0.545 K 3

k ~ z / k ~=l 1.6

6 = k ~ l / 2 k ~=6 3.33

x

r = k ~ i k ~ a / k ~= k0.101 ~s Q

= 2 k ~ s k ~ / k ~=, 0.0101 '

0 = kB,/kB6 = 1.667 x lo3

B = k ~ 3 / 2 k ~=6 5 0 0 T,,,

= kW[Br03-]trea= 3.91t,/s

Repistry No. BrO,-, 15541-45-4; gallic acid, 149-91-7; ferroin, 14708-99-7.

References pnd Notes (1) Liu, J.; Scott, S. K.J . Chem. Soc., Faraday Trans. 1991, 87, 2135. (2) Liu, J.; Scott, S. K.J . Chem. Soc., Faraday Trans. 1992, 88, 909.

(3) Giles, C. C. D.; Ibison, P.; Liu, J.; Scott, S.K.J. Chem. Soc., Faraday Trans. 1992,88, 917. (4) Orban, M.; Kijrb, E. J . Phys. Chem. 1978,82, 1672. ( 5 ) Orban, M.; Kijrb, E. React. Kinet. Coral. Lett. 1978,8, 273. (6) Orban, M.; K6rb, E.; Noyes, R. M. J . Phys. Chem. 1979,83, 3056. (7) Herbine, P.; Field, R. J. J . Phys. Chem. 1980, 84, 1330. (8) Field, R. J.; K b b , E.; Noyes, R. M. J . Am. Chem. Soc. 1972, 94, 8649.(9) Noyes, R. M. J . Am. Chem. Soc. 1980,102,4644. (10) Gyijrgyi, L.; Turanyi, T.; Field, R. J. J. Phys. Chem. 1990,94,7162. (1 1) Gyijrgyi, L.; Rempe, S.;Field, R. J. J . Phys. Chem. 1991,95,3159. (12) GyBrgyi, L.; Field, R. J. J. Phys. Chem. 1991, 95, 6594. (13) Gyijrgyi, L.; Field, R. J. J . Nafure 1992, 355, 808. (14) Field, R. J.; Fijrsterling, H. D. J . Chem. Phys. 1986, 90,5400. (15) Jwo, J.-J.; Chang, E.-F. J . Phys. Chem. 1989, 93, 2388. (16) Halaerin. J.: Taube. H. J . Am. Chem. Soc. 1952.74.375. (17j Gy&gyi,'L.;Varga,'M.; Kijrb, E.; Field, R. J.; Ruoff, P. J . Phys. Chem. 1989,93,2836. (18) Tyson, J. J. J . Chem. Phys. 1984,80, 6079. (19) Grav. P.: Scott. S. K. Chenical Oscillations and Instabilities: Oxford Unhersity &ess: New York, 1990. (20) Scott, S. K.Chemical Chaos; Oxford University Press: New York, 1991. (21) Doedel, E. AUTO: Software for Continuation and Bifurcation Problems in Ordinary Differential Equations; California Institute of Technology: Los Angela, 1986.

Kinetic and Mechanistic Study of the Reaction of Atomic Chlorine with Dimethyl Sulfide R. E. Stickel, J. M. Nicovich, S.Wang: Z. Zhao,' and P. H. Wine**# Georgia Tech Research Institute, Georgia Institute of Technology, Atlanta, Georgia 30332 (Received: July 20, 1992; In Final Form: September 8, 1992)

Timaresolved fluorescmCe detection Of Cl(9j) following 266-nm laser flash photd~isOf Cl~cO/CH$CH~(DMS)/N~ mixtures has been employed to study the kinetics of the title reaction over the temperature and pressure ranges 240-421 K and 3-700 Torr. The reaction is found to be very fast, occumng on essentially every C1(2PJ)+ DMS encounter. The reaction rate increases with decreasing temperature and shows a significant pressure dependence. At 297 K, for example, cm3molecule-' s-' to a value of the rate coefficient increases from a low-pressure limit value of approximately 1.8 X (3.3 f 0.5) X cm3molecule-' s-l at P = 700 Torr. A few experiments were carried out with CD3SCD3or C2H5SC2Hs replacing DMS as the sulfide reactant; within experimental uncertainty, no dependence of the rate coefficient on the identity of the sulfide reactant was observed. In a separate study, time-resolved tunable diode laser spectroscopic detection of HCl has been coupled with 248-nm laser flash photolysis of Cl&O/DMS/COz/Nz mixtures to measure the HCl product yield from the title reaction as a function of pressure at T = 297 K. The HCl yield approaches unity as P 0 but decreases with increasing pressure to a value of -0.5 at P = 203 Torr. The yield experiments demonstrate that hydrogen abstraction is the dominant reaction mechanism in the low-pressure limit. With increasing pressure, stabilization of a (CH3)gCIadduct apparently becomes competitive with the hydrogen abstraction pathway. The fate of the stabilized adduct remains highly uncertain, although it clearly does not dissociate to Cl(9J) or HCl on the time scale of our experiments (several milliseconds). The potential role of the title reaction in marine atmospheric chemistry is discussed.

-

+

Introduction Dimethyl sulfide (CH3SCH3,DMS) is a key atmospheric sulfur species. Roughly half the global flux of sulfur into the atmosphere is thought to be natural in and about half of all natural sulfur entering the atmosphere does so as DMS volatilized from the 0ct8ns.~9~ S ~ m e ,though ~ ? ~ not all,' field studies of marine sulfur chemistry have been interpreted as indicating that the DMS removal rate is too rapid to be accounted for entirely by known sink reactions with OH and NO3 radicals. To explain the apparently short lifetime of DMS in marine air, it was proposed that IO radicals, formed via the reaction of photolytically generated iodine atoms with ozone, could destroy DMS via a rapid reaction Author to whom correspondence should be addressed. Presmt address: Dalian Institute of Chemical Physics, C h i n a Academy of Sciences, P.O. Box 110, Dalian, Peoples Republic of China. *Alsoaffiliated with the Georgia Tech School of Earth and Atmospheric Sciences. 'Present address: School of Chemistry and Biochemistry, School of Earth and Atmospheric Sciences, and Georgia Tech Research Institute, Georgia Institute of Technology, Atlanta, GA 30332.

0022-3654/92/2096-9875$03.00/0

producing I DMSO (dimethyl sulfoxide)! Subsequent laboratory studies of IO DMS reaction kineticsgJOand field observations of significant concentrations of DMSO in the marine boundary layer" seemed to confirm the importance of IO as an initiator of DMS oxidation. However, a direct study of IO DMS kinetics carried out in our laboratory,12the reuslts of which have now been c o 1 1 f i i e d , ~demonstrates ~3~~ that coupling of the marine iodine and sulfw cycles via the IO DMS reaction is negligible. The possibility that reactions occumng on surfaces of sea-salt aerosol particles can lead to significant production of atomic chlorine in the marine boundary layer has received considerable attention in recent years. Production of photochemically labile Cl,(g) via heterogeneous degradation of O3(possibly involving free radical inkrmdites) is one suggested pathway for generation of gas-phase chlorine at"^;'^'' however, recent laboratory'*and modelingIgstudies suggest that this pathway is not important in the atmosphere. On the other hand, it appears that C1NO2, generated via the heterogeneous reaction of N205vapor with moist NaCl(s), may represent a photolytic precursor for atmospherically significant levels of atomic chlorine, even in the remote marine

+

+

+

Q 1992 American Chemical Society

9876 The Journal of Physical Chemistry, Vol. 96, No. 24, 1992

boundary layer where NO, levels are very low.'8~20~21 A recent competitive kinetics study of reaction 1 at 298 K in 1 atm of N2,, suggests that this reaction is extremely fast, Le., kl(298 K) = 3.2 X 1O-Iocm3molecule-' s-I. Combining the above CI(,PJ) + CH3SCH3-3 products result with the fmdings of our study of the kinetics and mechanism of the OH DMS reactionz3suggests that k l / k 2 50 at 298 K and 1 atm of air. Reaction 2 is the dominant known atmosOH + CH3SCH3 products (2) pheric sink for DMS under low NO, conditions, and mean surface-level OH concentrations are thought to be around lo6 molecules per cm3.24 Hence, if atomic chlorine concentrations in the marine boundary layer are as large as lo4 atoms per cm3, and if reaction 1 is as fast as the available kinetic datazzsuggest, then reaction 1 could play an important role in marine sulfur chemistry. In addition to their potential importance in atmospheric chemistry, the kinetics and mechanisms of radical DMS reactions are of considerable fundamental chemical interest. These reactions can often proceed via either C-H or C-S bond cleavage and, at relatively high pressure, can result in formation of bound radi~al-S(CH,)~adducts. The adduct structures, bamers toward possible decomposition pathways, and barriers toward bimolecular reaction pathways with key atmospheric species such as O,, 03, and NO, now appear amenable to evaluation via ab initio quantum chemical methods.25 Hence, detailed kinetic and mechanistic information can be employed to refine theoretical procedures. To illustrate the varying mechanistic behavior which can be observed, consider DMS reactions with O(3P),NO3,and OH. The O(3P) DMS reaction proceeds predominantly via C-S bond cleavage26v27while the NO3 + DMS reaction proceeds predominantly via C-H bond cleavage.28 In the presence of 02,the OH DMS reaction proceeds via both "02-dependent" and "0,-independent" pathways.23 The 0,-dependent pathway clearly involves adduct formation followed by adduct decomposition to OH DMS in competition with an adduct + 0,reaction.23 However, the mechanism for the 0,-independent pathway remains a controversial issue, with some experimental data supporting C-H bond cleavagez3and other data suggesting the probable importance of C-S bond cleavage.29 In this paper we report the results of a kinetic and mechanistic study of the C1 + DMS reaction, Le., reaction 1. Our results confirm that reaction 1 is extremely fast and also provide some interesting new mechanistic insights concerning this complex chemical reaction.

+

-

-

+

+

+

+

Experimental Section Two different types of experiments were performed. Laser flash photolysis (LFP) of C1,CO (phosgene) at 266 nm was combined with detection of chlorine atoms by time-resolved resonance fluorescence (RF) spectroscopy to investigate the kinetics of reaction 1 as a function of temperature and pressure. To gain insight into the mechanism for reaction 1, a separate set of experiments was camed out where laser flash photolysis of ClzCO at 248 nm was combined with detection of HCl by time-resolved tunable diode laser absorption spectroscopy (TDLAS) to measure the HCI product yield at 297 K as a function of pressure. The LFP-RF and LFP-TDLAS experiments are discussed separately below. 'Ibe LFFRF AppMhra The apparatus employed in this study was similar to those employed in our laboratory in several previous studies of chlorine atom kinetics in the absence of 02.3*37 Important features of the apparatus and technique are described below. A jacketed, Pyrex reaction cell with an internal volume of 150 cm3 was used in all experiments; a schematic diagram of the reaction cell is given elsewhere.38 The cell was maintained at a constant temperature by circulating ethylene glycol or a 1:l methanol-ethanol mixture from a thermostated bath through the outer jacket. A copper constantan thermocouple could be injected into the reaction zone through a vacuum seal, thus allowing measurement of the gas temperature under the precise pressure

Stickel et al. and flow rate conditions of the experiment. Temperature variation within the reaction volume, i.e., the volume from which fluorescence could be detected, was less than 1 K at both the high and low temperature extremes of the study. Chlorine atoms were produced by 266-nm laser flash photolysis of C1,CO. Fourth-harmonic radiation from a Quanta Ray Model DCR-2 Nd:YAG laser provided the photolytic light source. The photolysis laser could deliver up to 4 X 10I6photons per pulse at a repetition rate of up to 10 Hz; the pulse width was 6 ns. Fluences employed in this study ranged from 0.3 to 11 mJ pulse-'. An atomic resonance lamp situated perpendicular to the photolysis laser excited resonance fluorescence in the photolytically produd atoms. The mnance lamp consisted of an electrodelea9 microwave discharge through about 1 Torr of a flowing mixture containing a trace of Clz in He. The flows of a 0.1% C12 in He mixture and pure He into the lamp were controlled by separate needle valves, thus allowing the total pressure and Clz concentration to be independently adjusted for optimum signal-tenoise. Radiation was coupled out of the lamp through a magnesium fluoride window and into the reaction cell through a magnesium fluoride lens. Before entering the reaction cell, the lamp output passed through a flowing gas filter containing 3 Torr-cm NzO in Nz; this filter blocked virtually all 0 atom impurity emissions at 130-131 nm while transmitting the chlorine lines in the 135140-nm wavelength region. Fluorescence intensities were found to vary linearly with atom concentration up to levels several times higher than any employed in kinetics experiments ([Cl], I 3 X 1O"/cm3 in all experiments). Fluorescence was collected by a magnesium fluoride lens on an axis orthogonal to both the photolysis laser beam and the resonance lamp beam and imaged onto the photocathode of a solar blind photomultiplier. The region between the reaction cell and the photomultiplier was purged with N,. Also, a calcium fluoride window was inserted into this region to prevent detection of emissions at wavelengths shorter than 125 nm (Lyman-aemission, for example). Signals were processed using photon counting techniques in conjunction with multichannel scaling. For each chlorine atom decay measured, signals from a large number of laser shots were averaged in order to obtain a welldfmed temporal profile over (typically) three l/e lifetimes of decay. The multichannel scaler sweep was triggered prior to the photolysis laser in order to allow a pretrigger baseline to be obtained. In order to avoid accumulation of photolysis or reaction products, all experiments were carried out under "slow flow" conditions. The linear flow rate through the reactor was 2-4 cm s-l while the laser repetition rate was varied over the range 1-10 Hz (it was 10 Hz in most experiments). Since the direction of flow was perpendicular to the photolysis laser beam, no volume element of the reaction mixture was subjected to more than a few laser shots. Phosgene (C1,CO) and the sulfide reactant were flowed into the reaction cell from 12-L Pyrex bulbs containing dilute mixtures in nitrogen buffer gas, while nitrogen was flowed directly from its high-pressure storage tank. The sulfide/Nz mixture and additional N2 were premixed before entering the reaction cell. The ClZCO/Nzmixture was injected into the reaction cell through a '/&. 0.d. Teflon tube positioned such that mixing of ClzCO into the main flow occurred 1-5-cm upstream from the detection zone; variation of the position of the injector tip had no effect on observed kinetics. Concentrations of each component in the reaction mixture were determined from measurements of the appropriate mass flow rata and the totalp ~ ~ ~ u r e . The sulfide concentration was also measured in situ in the slow flow system by UV photometry at 202.6 nm using a zinc hollow cathode lamp as the light source. Sulfide absorption crm sections at 202.6 nm were measured during the course of this study; in units of 1O-I' cm2 they were found to be 1.36 for (CH&3, 1.31 for (CD3)2S,and 1.59 for (C2H&S. In most experiments the 2-m-long absorption cell was positioned upstream from the reaction cell; in this configuration it was necessary to correct measured concentrations downward by up to 18% to account for dilution of the sulfide upon addition of the ClZCO/Nz mixture (the dilution

The Journal of Physical Chemistry, Vol. 94, No. 24, 1992 9877

Reaction of Atomic Chlorine with Dimethyl Sulfide

< MIXINQ CELL

N, or COP+

MFM

LOCK IN

1

FUNCTION GENERATOR

CONTROL

-

MONOCHROMATOR

T

9

L

6TM

PRE AMP

-r

n

REF. CELL I

TM

I

I

f I

I

Figure 1. Schematic of the LFP-TDLAS apparatus. DM = dichroic mirror; IRD = infrared detector; MFM = mass flow meter; PR = parabolic reflector; Pres. = pressure gauge; Ref. = reference; TM = telescope mirror.

correction was negligible except at total pressures below 25 Torr). To check for possible systematic errors due to the Occurrence of dark reactions in the slow flow system, some experiments were carried out with the absorption cell positioned downstream from the reaction cell; this variation in experimental technique had no effect on measured rate coefficients. In all photometric measurements (including the cross section measurements) the absorption cell temperature was 297 f 2 K. The LFP-TDLAS Apparatus. A schematic diagram of the LFP-TDLAS apparatus is shown in Figure 1. All experiments were carried out at room temperature, 297 f 2 K. The reaction cell was a Pyrex cylinder 25 mm in diameter and 118 cm in length with angled calcium fluoride windows attached to the ends using high vacuum epoxy (Torr Seal). Calcium fluoride flats coated for IR transmission and UV reflection combined and separated the 248-nm photolysis beam and the 2821.57-cm-’ probe beam (HCl was probed using the u’ = 1, J’ = 2 U” = 0, J” = 3 transition). Both single- and two-pass probe configurations were employed; the two-pass configuration is shown in Figure 1. Care was taken to ensure that the UV beam completely enclosed the probe beam throughout the length of the reaction cell. Chlorine atoms were produced by 248-nm laser flash photolysis of Cl2C0. A Lambda Physik Model 200 EMG KrF excimer laser served as the photolytic light source. The photolysis laser could deliver up to 8 X 10’’ photons per pulse at a repetition rate of up to 10 Hz; the pulse width was 20 ns. A 20-mm-diameter aperture selected the central, most intense, and most spatially uniform portion of the photolysis beam. The fluence in the reaction cell, typically 30 mJ/cm2, was monitored by a silicon photodiode which sampled a fraction of the input beam (reflected off an uncoated quartz plate). A Corning 7-54 filter and a Teflon diffuser were employed to obtain a photodiode signal which varied linearly as a function of laser power. Linearity was confirmed through

-

comparison of photodiode signals with laser powers measwed using a Scientech disk calorimeter. Because the excimer laser would not run reliably at the low repetition rate (-0.05 Hz) required to ensure a fresh reactant mixture for each flash, the laser was pulsed at a rate of 0.5 Hz and a mechanical shutter was employed to allow only every tenth flash to traverse the reaction cell. The pulse immediately preceding each photolysis flash was used for background subtraction. The infrared probe beam was generated by a lead salt diode laser (Laser Analytics) housed in a liquid helium cooled cryostat. The diode output was collimated by a gold-coated off-axis paraboloid and merged with a helium-neon alignment laser using a glass flat coated for infrared reflection and visible transmission. The infrared beam diameter was reduced to 2 mm by two concave gold-coated reflectors configured as a reducing telescope. The “tightened” beam passed through the reaction cell, entered a 0.5-m grating monochromator where a single mode was selected, then impinged on a HgCdTe infrared detector (the signal detector) which was cooled to 77 K. As a spectral reference, a small fraction of the probe beam was directed through a sealed 10-cm cell containing about 0.5 Torr of HCl and then monitored by a second HgCdTe infrared detector (the reference detector). The output from the signal detector was amplified and applied to a 8-bit analog-to-digital transient recorder card in an MS-DOS compatible microcomputer. For each laser flash, 4096 samples were recorded. The sampling frequency was adjustable by factors of 2 from 20 MHz down to 156.25 kHz. Results from repeated flashes were summed in the computer to obtain an acceptable signal-to-noise ratio; typically, 32 laser shots were averaged. The diode laser output frequency was varied up and down through the HC1 absorption feature via 40-kHz sinusoidal modulation of the drive current. The sample absorption signal was recovered from the digitized intensity data through a cycle-by-cycle computation

!I878

The Journal of Physical Chemistry, Vol. 96, No. 24, 1992

of the second harmonic component. The reference cell signal was applied to a lock-in amplifer adjusted to provide a feedback signal which kept the laser tuned to the reference absorption. One major source of noise in TDLAS systems results from mechanical vibrations induced by the compressor in the closed cycle refrigeration system; this mechanical perturbation results in noise spikes in the diode laser output signal which occur a p proximately every 0.3 s. To alleviate the mechanical noise problem, the approach of Sams and Fried39was employed. A microphone was attached to the cryostat housing, and the resulting signal was processed to produce a logic-level indication of the compressor phase, Le., quiet or noisy. A variable delay was adjusted to produce a gate correlated with quiet phases of the refrigerator cycle. This “acoustic”gate was electronically AND-ed with the diode laser modulation signal to provide a trigger for the excimer laser and the transient recorder. The laser trigger was delayed relative to the start of the transient record to provide a preflash baseline. As in the LFP-RF experiments, all LFFTDLAS experiments were carried out under “slow flow” conditions. The linear flow rate through the reactor was typically 10-15 cm s-I, and as mentioned above, the repetition rate was typically 0.05 Hz; hence, the contents of the reactor were nearly completely replaced between laser shots. Phosgene, DMS, and ethane were flowed into the reaction cell from 12-L Pyrex bulbs containing dilute mixtures in COz, while nitrogen was flowed directly from its high-pressure storage tank. All components in the reaction mixture were premixed before entering the reaction cell. Concentrationsof each component in the reaction mixture were determined from measurements of the appropriate mass flow rates and the total pressure. The concentrations of C12CO and DMS in the storage bulbs were checked frequently by UV photometry at 228.8 nm using a cadmium penray lamp as the light source. Absorption cross sections used to compute concentrations were 1.34 X cm2 for ClZCO(measured during the course of this study) and 1.16 X lo-’* cmz for DMS.4O Chemicals. The pure gases used in this study had the following stated minimum purities: N2, 99.999%; He, 99.999%; C02, 99.99% C& 99.99% Clz, 99.99%; HCl, 99.99% ClzCO, 99.0%. Nitrogen, helium, carbon dioxide, and chlorine were used as supplied, while ethane and phosgene were degassed at 77 K before use. Liquid samples of the sulfides were obtained from Aldrich and had the following stated minimum purities: (CH3)2S, 99+%, anhydrous, packaged under Nz; (CzH,)S, 98%; (CD3)#, 99.9 atom % D, unspecified chemical purity. The sulfides were transferred in a nitrogen-filled glovebox into vials fitted with high-vacuum stopcocks and then degassed repeatedly at 77 K before use.

Results a d Discussion Kioetica Studies. The LFP-RF kinetics experiments were carried out under pseudo-first-order conditions with the sulfide in hundred-to-thousand-fold excess over chlorine atoms. The N2 levels were sufficiently high that relaxation of atoms in the 2PI/z spin-orbit excited state was rapid compared to the rate of chemical removal of c1(2PJ).4”42Hence, all measured chlorine atom temporal profiles should be considered as representative of removal of an equilibrium mixture of C1(2P3/2)and C1(2P1/2). To study the kinetics of reaction 1 it is desirable to establish experimental conditions where the C1(zPJ) temporal profile is governed entirely by the following processes: ClZCO+ hv(266 nm) 2C1(2P~)+ CO (3)

-

Cl(’Pj)

--

+ CH3SCH3

products

(1)

C1(2PJ) first-order loss by diffusion from the detector field of view and/or reaction with background impurities (4) Then, since [CH3SCH3J>> [Cl(zPJ)J,simple first-order kinetics are obeyed: In ([Cl],/[Cl],] = (kl[CH3SCH3]+ k4)r = k’r (I) In eq I, CI C1(2PJ)and k’is the pseudo-fmt-order chlorine atom

Stickel et al.

t4t

4.51 0

I

50

.

4 150

100

time us) Figure 2. Typical CI(zPJ)temporal profile observed in the LFF-RF studies. Experimental conditions: T = 297 K P = 50 Torr N2;con-

centrations of CI,CO, DMS, and C1(2P~)r-~ in units of 10” molecules (atoms) per an3 = 4100,860, and 1.1, respectively;number of laser shots averaged = 1OOOO. The solid line is obtained from a least squares analysis; its slope, 23 300 s-I, is the pscuddirst-order C1(2PJ)decay rate. The inset shows the same data plotted on a linear rather than a logarithmic scale. 4c

0

1.4

0.1

[DMS] (ld‘prr cm3)

Figure 3. Plot of k‘versus [DMS] for data obtained at T = 297 K and P = 50 Torr N2.Circles represent data obtained with DMS monitored (by UV photometry) upstream from the reaction cell while triangles repreacnt data obtained with DMS monitored downstream from the reaction cell. The open circle is the data point obtained from the temporal profile shown in Figure 2. The solid line is obtained from a linear least squares analysis; its slope gives k , ( P , n f 24 = (2.71 & 0.09) X cm3 molecule-’ 8.

decay rate. The bimolecular rate coefficients of interest, k1(P,T) are evaluated from the slopes of k’ versus [CH3SCH3] plots. Observations of Cl(’PJ) temporal profiles which are exponential, i.e., obey eq I, a linear dependence of k’ on [CH3SCH3],and invariance of k’ to variations in laser photon fluence and C12C0 concentration strongly suggests that reactions 1,3, and 4 are the only proceascs which affect the Cl(’PJ) time history. The above observations do not rule out possible interferences from impurity reactions. However, the data presented below confirm that reaction 1 is 80 fast that minor impurities could not possibly contribute significantly to observed Cl(’PJ) decay rates. A typical c1(2PJ) temporal profile is shown in Figure 2, and a typical &’versus [CH3SCH3Jplot is shown in Figure 3. Well over 200 CI(zPJ)temporal profiles were measured under a wide variety of experimental conditions. The results used to obtain values for RI(P,T)are summarized in Table I. As typified by the data in Figures 2 and 3, all results are consistent with eq I, Le., all decays arc exponential and, in a given set of experiments at constant temperature and pressure, &’increases linearly as a function of DMS concentration. In addition to the results presented in Table I, a limited number of experiments were carried out to examine the kinetics of C1(zPJ)reactions with CD3SCD3 (DMS-d6) and C2H5SC2H5(DES). The results of our studies Cl(’Pj) CD3SCD3 PrOdUCts (5) C1(IPJ) CZHSSC2Hs products (6) of reactions 5 and 6 are summarized in Table 11. Error bars on values for k l , k5,and k6 reported in Tables I and I1 are 20 and represent precision only. Considering possible systematic errors, primarily in the concentration measurements, we estimate the

+ +

-

+

The Journal of Physical Chemistry, Vol. 96, No.24, 1992 9879

Reaction of Atomic Chlorine with Dimethyl Sulfide

-

TABLE I: S ~ m m r of v Klaeti~ht.for the RmctiW CI('Pj) CH#cH, Products"

T 240

267 297

356

385 421

P

Nc

3.1 6 10 6 25 6 25 5 7 50 6 100 150 4 6 150 6 500 8 50 3.1 6 3.1 6 6.0 6 25 6 26 6 50 5 50 12 150 5 300 6 500 7 700 5 700 6 3.1 6 4 25 11 25 7 50 150 7 6 500 50 6 3.1 6 25 4 5 25 50 11 8 150 150 6 8 700

concentrations Cl2C0 Cb 6900 1.0 7100 1.1 1.o 6300 4600 1.3 6900 1.6 6700 1.o 1.9 8400 5200 1.2 9200 2.8 6100 1.4 5100 0.8 1.0 3300 3000 1.4 4600 0.8 1.1 3000 1900 1.6 4100 1.0 1.3 6600 2100 1.8 2.5 11000 12000 2.6 1500-3000 1.0-2.0 6300 0.9 loo00 3.0 2800 0.5 3500 0.8 5800 1.1 11000 1.9 0.7 2900 5800 0.7 2500 0.4 3000 0.7 1600-38000 0.4-2.2 3500 0.9 1.1 4700 7800 2.7

range of k' 232-20400 107-26000 74-29600 144-29700 273-28200 72-33100 183-24100 38-17900 25-22300 196-23400 368-22400 204-20600 237-19100 362-21100 116-22100 87-29130 231-36100 137-32900 287-31600 203-31200 158-21800 400-28000 442-9910 337-11300 258-16000 173-26100 108-23600 133-23000 182-24100 641-8320 194-6630 412-14800 186-21300 62-9000 147-22400 623-15100

+

kl f 2 8 248 f 9 270 f 25 297 f 16 327 i 16 338 f 9 332 i 15 365 f 48 364 f 41 378 f 28 299 f 18 214 f 7 181 f 29 195 f 18 248 f 7 227 f 25 256 f 10 271 i 9 309 f 11 303 f 11 347 f 17 3 3 4 i 12 305 f 6 183 A 6 210f 4 201 f 18 235 f 7 258 f 13 356f51 233 f 20 183 f 7 168 h 32 177 f 9 193 f 18 167 i 14 182 f 8 231 f 10

"Units are T (K); P (Torr); concentrations (10" cm-)); k' (s-l); kl (10-l2 cm3 molecule-I d). N2buffer gas. N number of experiments; experiment = measurement of a single Cl('PJ) pseudo-first-order decay rate. dErrors represent precision only. TABLE II: SUBUIU~ Of KiaetiC h C for the R-C~~OMO f CI(*Pj) with CD&ED, (DMS-1') .ad C2H&C2Hs(DES) at T = 297 I(" concentrations sulfide pb Nc CllCO C1, range of k' k f 2 8 DMS-ds 50 9 3600 0.8 138-22200 275 f 9 500 6 7800 2.4 195-24200 336 f 24 DES 500 6 7800 2.4 195-24200 352 f 28 "Units are P (Torr); concentrations (10" cm-"); k' (d); k cm3 molecule-' 5-I). b N zbuffer gas. C N = number of experiments; experiment = measurement of a single C1(2PJ)pseudo-first-order decay rate. dErrors represent precision only.

absolute accuracy of our reported rate coefficients to be f l 5 % . Measured values for kl at three temperatures are plotted as a function of pressure in Figure 4. Our results demonstrate that reaction 1 is very fast, occurring on virtually every Cl-DMS encounter. The reaction rate is found to increase with decreasing temperature as would be expected for a very fast reaction whose rate is determined by the magnitude of long-range attractive forces between the reactants. The surprisiing aspect of the data in Figure 4 is our observation of a clear pressure dependence for k l , at least at temperatures of 297 K and below. Reaction 1 appears to occur via both pressureindependentand pressure-dependent pathways; the pressure-dependent pathway must involve collisional stabilization of a (CH3),S-C1 adduct. The only published kinetics study of reaction 1 is that of Nielsen et aL2' These authors employed a competitive kinetics technique where the rates of removal of DMS and the reference reactant cyclohexane were followed by gas chromatography under ex-

240

"

I

2

P

f

1

,,,,.,,,I

,,,,,,,,,

-1

K

, , , , , ,

j

0

1

100

10

1000

P (Torr)

Figure 4. Rate coefficients for the Cl('PJ) + DMS reaction at three temperatures plotted as a function of pressure. The solid lines are 'eyeball" fits to the data; their significance is simply as an aid in visualizing the observed pressure dependences.

perimental conditions where loss of both compounds was expected to be dominated by reaction with chlorine atoms, which were generated by cw 253.7-nmphotolysis of C12CO. Assuming a value of 3.1 I x 10-l~cm3molecule-' s-I for the c1(2PJ)+ cyclohexane rate coeffi~ient?~ Nielsen et al. report kl(740 Torr N2, 295 K) = (3.22 0.30) X lo-' an3molecule-' PI,in excellent agreement with our results (Table 1). Nielsen et al. also report k6(740 Torr N2, 295 K) = (4.41 f 0.40) X cm3molecule-' s-I, Le., about 25% faster than the value ka(500 Torr N2, 297 K) = (3.52f 0.53) X cm3 molecule-' s-I obtained as part of this study (Table 11). The difference in reported values for k6 is just about equal to the combined error limits. A small increase in k6 with pressure (unlikely to be more than 5-1096 over the pressure range SOU-740 Torr) cannot, of course, be ruled out. In addition to the work of Nielsen et al.,' one other value for kl appears in the literature. Barnes et a1.,44in a paper describing competitive kinetics studies of the reactions of OH, NO3, and C1 with dimethyl sulfoxide (CH3S(0)CH3,DMSO), report the "unpublished result" k'(760 Torr air, 298 K) = (2.0 f 0.3) X lo-'' cm3 molecule-' s-I; no details of their experimental procedure were presented. HCI Yield Studies Like the L F P R F kinetics experiments, the LFP-TDLAS studies of HCl production from reaction 1 were camed out under (near) pseuda-fmt-order conditions with DMS in exceas over chlorine atoms. However, because the tunable diode laser absorption detection technique is less sensitive than the resonance fluorescence detection technique, [DMS] to [cl(2PJ)], ratios were in the range 4-40 rather than 100-1OOO as in the L F P R F studies. To obtain an HCl yield, we camed out backto-back experiments where the photolytically produced c1(2PJ) reacted with DMS and then with ethane (C,&); the yield of HCl from the C1 C2H6 reaction is known to be unity. Reaction

*

+

Cl('Pj) C2H6 C ~ H + S HC1 (7) mixtures employed in the HCl yield studies contained ClzCO photolyte, DMS or ethane, CO,, and (in some cases) N,. Addition of CO, to the reaction mixture served two purposes. Reaction 1 is sufficiently exothermic that HCl could be produced with one quantum of vibrational excitation. The rate coefficient for vibrational energy transfer from HCl(u = 1) to CO2(0,O,0) is known to be about 2.5 X lo-'' cm3 molecule-' s-1.45-48 Hence, CO, promotes rapid deactivation of any vibrationally hot HCl to the (monitored) u = 0 level. In addition to its role as a vibrational relaxer for HCL, carbon dioxide is known to be very efficient at i.e., k8 1.5 X lo-" cm3 molecule-' deactivating C1(2P1/2), +

-

c1(2Plp) + c02

-

c1(2P3/2)+ coz

(8)

Hence, yield measurements could be carried out at total prrssures as low as 0.6 Torr CO,. Typical HCl yield data are shown in Figures 5 and 6. The results of the second harmonic analysis (seeFigure 6,for example) were fit using a nonlinear least-squares procedure to the sum of an exponential rise and an exponential decay: [HCl], k,C(kd - k,)-'[eXp(-kJ) - eXp(-kdt)] (11)

9880 The Journal of Physical Chemistry, Vol. 96, No. 24, 1992

L I

r

C

I

Stickel et al.

.

0 2 0 0 4 0 0 6 0 0

time (ps) Figure 5. Typical “raw data” intensity versus time plot observed in the HCI yield studies; note that intensity increasestoward the bottom of the plot. The diode laser frequency is modulated by a 40-kHz sine wave, passing through the HCI absorption feature twice each cycle. The width of the sweep is adjusted to maximize the second harmonic amplitude. Consequently the laser frequency never moves far into the wings of the absorption feature, and the baseline shifts slightly as a function of the HCI concentration. Note that a low background level of HCl is present prior to the photolysis flash. Reaction: C1(2PJ)+ DMS. Experimental conditions: T = 297 K, P = 26.2 Torr C02, 1.26 X 10’’ C12CO/cm3, 7.00 X lo” DMS/cm3, [c1(2PJ)]o= 5.5 X lO’*/Cm’, 32 laser shots averaged.

In eq 11, k, and kd are the pseudo-first-order rate coefficients for HCl appearance and disappearance, and C is the HCI concentration which would be reached as t m if kd = 0. Since experimental conditions were such that nearly all photolytically generated chlorine atoms reacted with DMS or ethane, bimolecular rate coefficients could be obtained with good accuracy from the simple relationships kl N ka[CH3SCH3] (111) k7 ka[C2H61 (IV) where eq I11 applies to experiments with DMS in the reaction mixture and eq IV applies to experiments with ethane in the reaction mixture. The HCl yield at pressure P, @(P), was evaluated by comparing values for C (normalized for any variation in laser power and/or phosgene concentration) obtained in back-to-back experiments with all experimental parameters held as constant as possible except that chlorine atoms reacted in one case with DMS and in the other case with ethane. W )= C D M S ( P ) / C C ~ H ~ ( P ) (VI

-

The results of the HCl yield experiments are summarized in Table 111. Bimolecular rate coefficients for reactions 1 and 7, evaluated from measured HCl appearance rates using eqs 111and IV, are tabulated in Table 111. Derived values for k7 show no pressure dependence, as one would expect for a direct H-abstraction reaction. An unweighted average of the 43 determinations reported in Table I11 gives k7 f 2a = (5.9 f 0.6) X lo-’’ cm3 molecule-’ s-I, in excellent agreement with the literature value49of (5.7 f 0.6) X lo-’’ cm3 molecule-’ s-I. Derived values for kl increase with increasing pressure and, within experimental uncertainty, are equal to the values obtained from the LFP-RF experiments. The measured HCl appearance rates strongly suggest that the dominant sources of observed HCl are the reactions of interest, Le., reactions 1 and 7. The results in Table I11 demonstrate that the HCl yield approaches unity as P 0 but decreases with increasing pressure. R e ” Mecbraism Possible channels for reaction (1) include the following: CI(’Pj) CH3SCH3 CH3SCH2 + HCl AH = -8 f 2 kcal mol-’ (la)

-

-

+

CH3SCl+ CH3 AH = 8 f 4 CH3S + CH3C1 AH -9.8 f 0.5

+

-.c

-% CH3S(Cl)CH3

AH = -14

f3

(1b) (IC) (14

Thermochemistry for the above reaction channels is estimated

4

0 2 0 0 4 0 0 6 0 0

time (ps) Figure 6. Results of the “sliding cycle” second harmonic analysis for the C1(2P~)+ DMS data shown in Figure 5 and the back-to-back experiment with ethane (at a concentration of 2.99 X 10i4/cm3)replacing DMS as the reactant. The infrared signal was sampled 64 times per modulation cycle. The amplitude of the second harmonic Fourier component was then computed digitally for several hundred 64 points subsets centered at different times. The resulting temporal profiles, after subtraction of the preflash baseline, were fit to eq I1 using a nonlinear least squares procedure; the solid lines represent the best fits. Best fit parameters for the c1(2PJ) DMS experiment were k, = 18 900 s-’, kd = 17 s-I, C 3.21 X 1OI2an-’.Best fit parameters for the CI(pJ) + C2& experiment were k, = 18 OOO s-I, kd = 28 s-I, C = 5.39 X 10l2cm-’. The HCl yield is obtained from C(DMS)/C(C2&) after normalization for any variation in laser power and/or C12C0concentrationbetween the two experiments.

-

+

basedon heats of formation obtained as follows: heats of formation for Cl(’PJ), CH3SCH3, HCl, CH3, and CH3Cl are taken from ref 49; the CH3SClheat of formation is an estimate by Bens~n;~O the CH3Sheat of formation is based on recent experimental work in our laboratory?’ the CH3SCH2 heat of formation is based on the published value of Shum and Bensod2 and on recent (unpublished) work in our laboratory” which suggests that the C-H bond strength in DMS is about 3 kwl mol-’ weaker than the published value;52the adduct bond strength is b a d on recent (unpublished) studies in our laborat0ry~~9~~ where CH3S(Br)CH3 and CH3S(OH)CH3were both found to be bound by 14 kcal mol-’. Uncertainties in the above heats of reaction are our estimates. Although the CH3SC1heat of formation remains highly uncertain, it appears unlikely that reaction 1b is an energetically feasible channel. While quantitative extrapolation of kinetic and HCl yield data to zero prespure is nontrivial, examination of the results in Figure 4 and Table I11 strongly suggests that the following relationship is obeyed (at T = 297 K): @(P) = kl(P O ) / k ’ ( P ) (VI) Clearly, hydrogen abstraction, i.e., channel la, is the dominant reaction pathway in the limit of low total pressure. However, channel Id becomes competitive at higher pressure. Our experiments do not provide definitive information concerning the fate of the collisionally stabilized adduct. The only information we obtain is that adduct decomposition to CI(*PJ)and/or HCl does not occur on the time scale of the experiments, Le., several milliseconds. Recent (unpublished) work in our laboratory has established the foliowing Arrhcnius parameters for Cl(*PJ)reactions with H#, D2S, CH3SH, and CD3SD (units are lo-” cm3 molecule-’ CI(’PJ) + H2S: (3.60 f 0.26)exd(209 f 2 0 ) / n

-

Cl(*Pj) + D2S: (1.85 f 0.41) exd(186 f 5 8 ) / q C1(’PJ)

+ CH3SH, CD’SD:

(11.9 f 1.7) expl(l51 f 38)/7)

The Journal of Physical Chemistry, Vol. 96, No. 24, 1992 9881

Reaction of Atomic Chlorine with Dimethyl Sulfide TABLE JIk Summrry of HCI Yield Dah at T = 297 K“

0.63 0.63 2.01 2.01 2.01 2.01 2.01 2.01 2.02 2.02 2.02 2.02 2.02 2.03 2.04 2.04

5.00

5.00 10.0 10.0 25.2 25.2

I

0 27.3 0 27.7 0 56.3 0 56.6 0 55.4 0 54.7 0 55.0 0 54.2 0 108 0 57.8 0 27.5 0 107 0 55.7 0 109 0 55.0 0 27.6 0 50.6 0 50.0 0 51.4 0 51.6 0 51.8 0 51.7

244 0 233 0 228 0 228 0 229 0 227 0 460 0 244 0 115 0 454 0 232 0 454 0 234 0 115 0

450 440 440 440 1010 1010 480 480 2000 2030 2030 2020 500 510 2010 2030 1020 1030 1010 990 1040 1040 1040 1030 1000 990 1040 1040 1020 1010 1020 1010

1.9 1.8 1.9 1.9 8.0 8.4 3.4 3.2 13.5 14.4 9.4 9.8 1.4 1.3 5.3 5.2 4.7 4.9 8.5 8.2 7.2 7.2 7.4 7.4 2.7 2.5 2.8 2.7 4.7 4.7 4.6 4.7

19900 2 m 19900 2 m 64100 64200 64600 64800 63100 63200 63100 63250 64600 64800 63100 63200 64200 64500 64400 64600 64500 64600 64200 64500 64400 64600 64200 64500 65000 65200 65100 65200

61.6 203 60.3 194 58.2 215 57.1 215 61.5 224 59.6 229 56.6 207 60.4 214 57.0 206 59.0 213 64.0 229 57.7 207 58.2 206 54.8 205 58.5 216 59.0 225

54.1 0 27.8 0 54.9 0 58.0 0 58.4 0 61.5 0 61.6 0 55.5 0 55.8 0 56.3 0 56.4 0 60.8 0 60.4 0 69.5 0 70.0

229 0 115 0 233 0 247 0 246 0 256 0 258 0 235 0 233 0 233 0 234 0 237 0 236 0 296 0 299 0

540 540 1050 1050 1050 1060 1020 1010 1040 1010 980 980 980 980 1000 1000 1000 1000 lo00 1000 1000 1000 940 940 1030 1030 1250 1260 1260 1260

0.880

208 0 207 0 212 0 210 0 212 0 211 0

890 890 890 890 890 910 890 890 900 900 900 900

4.4 4.5 4.5 4.5 4.2 4.3 4.3 4.4 4.4 4.4 4.3 4.4

37800 37400 37800 38200 38200 39200 38000 38800 38600 39200 38400 39100

53.8 202 53.6 197 52.8 202 53.8 202 59.0 241 57.3 244

B. N2 Buffer Gas 50.2 0 0.863 66.7 50.2 0 0.833 51.9 99.9 0 0.791 86.5 100 0 0.811 87.2 203 0 0.672 117 203 0 0.679 116

277 0 280 0 353 0 352 0 492 0 484 0

1250 1250 1280 1280 1440 1470 1500 1500 2150 2140 21 10 2080

116 0 116 0

2.04 0.987 2.05 0.970 2.05 0.883 2.05 0.887 2.07 0.881 4.99 0.866 5.00 0.865 5.01 0.883 5.02 0.871 10.1 0.899 10.1 0.898 10.1 0.887 10.1 0.95s 26.1 0.913 26.2 0.912

“Units are P (Torr); concentration (loL2molecules (atoms) cm-’); k , / [ X ] nonzero concentration.

Due to the large electron affinity of Cl(’PJ) (3.617 eV56)and the relatively low ionization potentials of RSR’ species,56one might expect reactivity in C1(2PJ) RSR’ reactions to correlate with the ionization potential of RSR’, since this physical property should reflect the strength of long-range attractive forces associated with stabilization of transition states via charge separation. Such a reactivity trend is indeed observed. The ionization potentials of H2S, CH3SH, DMS, and DES are 10.38 eV, 9.37 eV, 8.69 eV, and 8.43 eV respectively;s6 the order of reactivity with Cl(2PJ) is DES 2 DMS > CH3SH > HIS. In contrast to the excellent correlation of reactivity with RSR’ ionization potential, the correlation between reactivity and exothermicity is rather poor; for example, the very fast Cl(*PJ) 4- DMS reaction is the least exothermic in the series. The H-D kinetic isotope effects observed in c1(2PJ)reactions with HIS, CHSSH, and CH3SCH3provide an interesting comparison. No kinetic isotope effects are observed in the case of the faster reactions of C1(2PJ)with CH3SH and CH3SCH3,but D S is found to react with chlorine atoms about a factor of 2 more slowly than does H& This suggests that formation of an energized RS(C1)R’ adduct is rate limiting in the C1(2PJ) CH3SH, CH3SCH3 reactions, Le., once Cl(2P,) and RSR’ come into close

+

+

0

2.5 65500 2.6 65700 2.8 65500 2.8 65500 7.1 65300 7.3 65500 4.4 65700 4.4 65900 4.6 66000 4.4 66200 4.1 161000 4.2 161000 4.2 161000 4.0 161000 8.3 162000 8.1 162000 8.0 162000 8.1 162000 8.5 327000 8.6 327000 8.4 327000 8.5 327000 3.8 327000 3.9 327000 4.2 327000 4.3 327000 5.5 847000 5.6 847000 5.4 850000 5.5 85oooO

5.9 5.9 6.3 6.2 6.9 6.7 7.1 7.0 10.2 9.8 10.0 9.9

59.3 213 59.8 225 62.7 215 56.3 205 56.1 207 60.9 237 60.1 237 62.6 243 61.8 25 1 65.2 279 65.0 270 63.3 265 63.6 260 62.2 276 60.2 270

51700 58.1 52200 274 52500 58.6 46900 268 62800 59.5 64600 316 63900 56.0 65500 279 90300 64.0 90700 315 89000 60.3 88800 362

0.908 0.871 0.886 0.886 0.886 0.789 0.789 0.797 0.784 0.748 0.743 0.727 0.730 0.644 0.595

0.586 0.584 0.535 0.541 0.512 0.498

cm3 molecule-’ s-l). b X = C2Hs or DMS,whichever has a

enough proximity to interact, a chemical reaction occurs. On the other hand, it appears that dissociation of the energized H2S