Product Formation in the Cl-Initiated Oxidation of Cyclopropane

LiVermore, California 94551-0969. Michael D. Hurley and Timothy J. Wallington. Research Staff, Ford Motor Company, SRL-3083, Dearborn, Michigan 48121-...
0 downloads 0 Views 271KB Size
1992

J. Phys. Chem. A 2003, 107, 1992-2002

Product Formation in the Cl-Initiated Oxidation of Cyclopropane John D. DeSain,† Stephen J. Klippenstein, and Craig A. Taatjes* Combustion Research Facility, Mail Stop 9055, Sandia National Laboratories, LiVermore, California 94551-0969

Michael D. Hurley and Timothy J. Wallington Research Staff, Ford Motor Company, SRL-3083, Dearborn, Michigan 48121-2053 ReceiVed: September 23, 2002; In Final Form: December 31, 2002

The production of HO2 and OH in the reaction of c-C3H5 + O2 is investigated as a function of temperature (296-700 K) using laser photolysis/CW infrared frequency modulation spectroscopy. The cyclopropyl radical is generated by the Cl + cyclopropane reaction following pulsed laser photolysis of Cl2. Significant OH and HO2 production is observed at 296 K, and both [OH]/[Cl]0 and [HO2]/[Cl]0 increase slowly with increased temperature until ∼600 K, where a sharper increase with temperature is observed. Relative rate and end product measurements are also performed using a smog chamber FTIR apparatus. The relative reactivity of cyclopropyl radicals toward O2 and Cl2 is kc-C3H5+O2/kc-C3H5+Cl2 ) 0.44 ( 0.02 at 700 Torr, 0.44 ( 0.03 at 75 Torr, and 0.24 ( 0.02 at 10 Torr of N2 diluent at 296 K. Ethene and oxirane are identified as end products of the cyclopropane oxidation. Molar yields of oxirane are 0.11 ( 0.03 at 6 Torr, 0.08 ( 0.02 at 10 Torr, and 0.06 ( 0.02 at 50 Torr total pressure of N2/O2 diluent; molar yields of ethene are 0.14 ( 0.02 (6 Torr) and 0.15 ( 0.01 (10 Torr) and 0.30 ( 0.06 (50 Torr). The combined experimental data suggest that HO2 is not a primary product of the cyclopropyl + O2 reaction but arises from secondary reactions of HCO or HCO2 products formed in conjunction with oxirane or ethene. Quantum chemical calculations of stationary points on the cyclopropyl + O2 surface indicate that ring opening and isomerization of c-C3H5O2 to form a dioxirane species is possible, with a calculated transition state energy 0.5 kcal mol-1 above that of the reactants. This dioxirane species is a conceivable precursor to HCO2 + ethene or HCO + oxirane formation; however, the calculations suggest OH + acrolein as the dominant bimolecular products.

1. Introduction Alkyl radical (R) reactions with molecular oxygen are central to understanding hydrocarbon oxidation in combustion systems and in the atmosphere. The most detailed studies of R + O2 reactions have focused on understanding the mechanism of the smaller straight-chain alkyl (ethyl and propyl) reactions with O2.1-8 The R + O2 reactions proceed via a bound alkylperoxy (RO2) radical, which can subsequently eliminate HO2 to form a conjugate alkene or isomerize by intramolecular hydrogen abstraction to form a hydroperoxyalkyl radical (often denoted QOOH). The QOOH radical can in turn react with O2 or dissociate to form HO2 + alkene or OH + cyclic ether products. The branching among these channels is important for chain propagation in low-temperature hydrocarbon oxidation because of the differing reactivities of OH and HO2. Reactions of QOOH are thought to lead to chain branching in low-temperature oxidation, making understanding the formation of QOOH important for modeling autoignition and engine knock. The alkene + HO2 channel largely dominates the reactions of small alkyl radicals with O2 for moderate temperatures.1-3,5-21

alkyl + O2 f alkene + HO2

(1)

For example, Walker and Morley22 report high alkene yields at * To whom correspondence should be addressed. Electronic mail: [email protected]. † Present address: The Aerospace Corporation, 2350 E. El Segundo Blvd., El Segundo, CA 90245-4691.

753 K and 70 Torr from C2H5 + O2 (99%), t-C4H9 + O2 (99%), and c-C5H9 + O2 (90%). The principal other products are cyclic ethers arising from isomerization to QOOH. The elimination of HO2 and the various isomerizations to form QOOH species occur through ring transition states, and the facility of these reactions may be affected by cyclization of the alkyl species. Several recent studies have investigated the oxidation of cyclic alkanes such as cyclopentane12,15,23 and cyclohexane.24 Timeresolved probing of HO2 product formation in Cl-initiated cyclopentane oxidation suggests that, while the energetics of the HO2 elimination are similar to that in ethyl and propyl + O2, the pre-exponential factor is larger for the cyclic alkyl radical.12 Because in cyclic alkyl RO2 species rotation around C-C bonds is already hindered, the entropy loss between reactant and the ring transition state for elimination is smaller than that for acyclic RO2 species, which lose an internal rotation in reaching the transition state. Little is known about the oxidation of the smallest ring alkane, cyclopropane, although Falconer, Knox, and Trotman-Dickenson25 measured the total cyclopropane oxidation rate relative to that for ethane oxidation between 601 and 689 K. If the reaction of the cyclopropyl radical with O2 is similar to other R + O2 reactions, then the major product channels would be M

c-C3H5 + O2 798 c-C3H5O2

(2a)

f c-C3H4 + HO2

(2b)

f c-C3H4O + OH

(2c)

10.1021/jp022120u CCC: $25.00 © 2003 American Chemical Society Published on Web 03/04/2003

Cl-Initiated Oxidation of Cyclopropane

J. Phys. Chem. A, Vol. 107, No. 12, 2003 1993

TABLE 1: Estimated ∆H°r (298) for Several Possible c-C3H5 + O2 Reaction Pathwaysa

a

reaction products

∆H°r (kcal mol-1)

allyl (CH2CHCH2) + O2 cyclopropene (c-C3H4) + HO2 propyne (CHCCH3) + HO2 allene (CH2CCH2) + HO2 acrolein (CH2CHCHO + OH) cyclopropanone (c-C3H4O) + OH acetaldehyde (CH3CHO) + HCO oxirane (C2H4O) + HCO ethene (C2H4) + HOCO cyclopropoxy (c-C3H5O) + O formaldehyde (H2CO) + H2CCHO

-29.0 0.5 -22.8 -21.8 -76.6 -49.1 -100.4 -73.6 -104.9 6.2 -93.0

Thermochemistry data from HL2 calculations.

However, production of cyclopropene + HO2 from cyclopropyl + O2 is predicted to be slightly endothermic. Also, the cyclopropyl radical contains a large amount of potential energy in the form of ring strain, and ring-opening may make other products possible. Without a low-energy path to HO2 + alkene products, cyclopropyl + O2 may follow a significantly different reaction mechanism than other alkyl + O2 reactions. The present work uses a combination of experimental and theoretical methods to investigate the mechanism of the cyclopropyl + O2 reaction. The time behavior of HO2 and OH formation in the Cl-initiated oxidation of c-C3H6 is measured between 296 and 700 K by using infrared spectroscopy following pulsed photolytic initiation. The reaction is observed to produce a significant amount of both OH and HO2 even at 296 K. The formation of HO2 has some qualitative similarities to that observed in previous experiments on C2H5 + O2,2 C3H7 + O2,1 c-C5H9 + O2,12 and C4H9 + O2.13 However, the observation of significant OH formation at such a low temperature is peculiar to cyclopropyl + O2. Smog chamber/FTIR measurements show that ethene and oxirane are products of cyclopropane oxidation at 296 K. Ab initio characterization of stationary points on the cyclopropyl-O2 potential energy surface provides an explanation for the observed products. It is proposed that HO2 is not a primary product of the cyclopropyl + O2 reaction but results from secondary reactions of HCO or HCO2 products formed in coincidence with oxirane or ethene. 2. Experimental Section 2.1. Laser Absorption Measurements at Sandia National Laboratories. The reaction of c-C3H5 + O2 is investigated by Cl-initiated oxidation of cyclopropane, using a laser photolysis/ CW infrared frequency modulation method similar to that employed previously.1,2,12,13 The Cl is generated by 355 nm photolysis of Cl2, and c-C3H5 is formed by the reaction of Cl with cyclopropane. The c-C3H5 radical then reacts with O2: hν (355 nm)

Cl2 98 2Cl

(3)

Cl + c-C3H6 f c-C3H5 + HCl

(4)

c-C3H5 + O2 f products

(2)

Experimental heats of formation are unavailable for many possible products of reaction 2; estimated thermochemistries for several conceivable channels are listed in Table 1. To minimize the effects of the competing reaction of Cl2 with c-C3H5, the O2 concentration is kept at least 30 times greater than the Cl2 concentration. The rate constant for c-C3H5 + Cl2 is larger than that of c-C3H5 + O2 at 298 K (see below), but

the Cl product of the c-C3H5 + Cl2 reaction regenerates c-C3H5 radicals, and the chain chlorination is soon quenched by the oxidation. Because of the relatively low rate coefficient for the reaction of cyclopropane with Cl (k4(298 K) ) (1.15 ( 0.17) × 10-13 cm3 molecule-1 s-1;26 k ) (8.96 × 10-11)e(-2080K/T) cm3 molecule-1 s-1),27 a large excess of cyclopropane ((0.85) × 1016 cm-3) is used. Nevertheless, at room temperature the contribution of the reaction of Cl with HO2 is not completely negligible. This reaction may reduce the apparent HO2 yield from the c-C3H5 + O2 reaction slightly, although no significant difference in apparent yield is noted over the range of [c-C3H6] employed in the present experiments. The formation of HO2 is monitored by infrared absorption of the overtone of the O-H stretch in HO2 near 1.5 µm using a tunable diode laser, and the formation of OH is monitored by direct absorption on the P(2.5)1- line of the vibrational fundamental at 3484.6 cm-1 using an F-center laser. Two-tone frequency modulation of the diode laser probe is employed to increase the signal-to-noise ratio of the HO2 measurement. For the OH measurement a balanced detector method is employed, where the first detector (reference) monitors a portion of the laser output prior to entering the cell and the second (signal) monitors the infrared beam after passing through the reactor. The average DC power on the two detectors is equalized and the signals from the two detectors are subtracted to reduce the contribution of laser amplitude noise. The experiments are performed in a resistively heated quartz slow-flow reactor. The IR probes are placed on the same path through the reactor by using polarizing prisms to combine and separate the beams. The IR probes are passed multiple times through the reactor by using a Herriott-type multipass cell, in which the probe beams intercept the UV photolysis beam only in the center of the flow cell, where the temperature is more readily controlled. Typical gas concentrations are as follows: O2, 6.4 × 1016 cm-3; Cl2, 2.0 × 1015 cm-3; cyclopropane, 8 × 1015 to 5 × 1016 cm-3. Helium is added to reach a total density of 8.5 × 1017 cm-3. Gases are obtained from commercial sources at the following stated purities: O2, >99.998%; Cl2, >99.99%; He, >99.9999%; cyclopropane, >99.9%. Major impurities in the cyclopropane sample, as characterized by GC/MS, are 1,1and 1,3-dichloropropane, with smaller amounts of propene and n-propanol. The contribution of Cl reaction with n-propanol impurity is of most concern for the HO2 production measurements, since its rate constant is ∼1250 times that of Cl + cyclopropane28,29 and because the product hydroxypropyl radical could, by analogy with CH2OH, rapidly produce HO2 by reaction with oxygen.30 Reactions of Cl with the other impurities will produce substituted propyl radicals, which are not expected to produce significant HO2 at low temperature.1,4 The relative importance of Cl reactions with impurities will be greater at lower temperature, since the Cl + cyclopropane reaction has a significant activation energy. Comparison of the apparent room temperature rate coefficient measured by following HCl appearance with that determined by relative rate methods directly monitoring cyclopropane disappearance, detailed in the accompanying paper, suggests an upper limit for the contributions of reactions with impurities of ∼20%. The consequences of these reactions are relatively minor and are described in the Discussion section below. The HO2 signal produced by the c-C3H5 + O2 reaction is scaled to the initial Cl concentration by comparison with the HO2 signal from the Cl2/CH3OH/O2 system under identical photolysis conditions. This reaction system is assumed to convert

1994 J. Phys. Chem. A, Vol. 107, No. 12, 2003

DeSain et al. TABLE 2: Peak Amplitude of the Scaled HO2 Signal from c-C3H5 + O2

Figure 1. Scaled HO2 signals at 296 K (magenta), 573 K (green), 623 K (cyan), and 683 K (red) from the reaction of c-C3H5 + O2 with [c-C3H6] of 8 × 1015 cm-3 and a total density of 8.5 × 1017 cm-3. The scaled HO2 signal from the reaction of CH2OH + O2 at 296 K and a total density of 8.5 × 1017 cm-3 is shown in black. The signals are scaled to the initial Cl atom concentration as described in the text.

100% of the initial Cl atoms (≡ [Cl]0) to HO2 over the temperature range of concern:30 hν (355 nm)

Cl2 98 2Cl

(3)

CH3OH + Cl f CH2OH + HCl

100%

(5)

CH2OH + O2 f CH2O + HO2

100%

(6)

temp (K)

peak [HO2]/[Cl]0

296 373 473 573 623 638 653 663 668 683 698

0.12 ( 0.03 0.17( 0.04 0.20 ( 0.04 0.19 ( 0.04 0.26 ( 0.05 0.36 ( 0.07 0.42 ( 0.08 0.38 ( 0.08 0.48 ( 0.08 0.55 ( 0.09 0.44 ( 0.09

Cl2/c-C3H6/O2 system also complicates the interpretation of the HO2 amplitude. In previous investigations of R + O2 reactions, it has been possible to simply model the RO2 and HO2 reactions to extract an overall HO2 yield.1,2,12 In the cyclopropane system the production of significant OH requires a more complex system of reactions and (unknown) rate coefficients. Because of these complications, in the present system we simply list the peak [HO2]/[Cl]0 (corrected only for the self-reaction) in Table 2, which is a lower limit on the overall HO2 yield. It should be emphasized that the HO2 yield in the present case is a measure of the fraction of the initial Cl concentration that is converted to HO2 in the course of the overall oxidation, and cannot be simply interpreted as the branching fraction of an elementary reaction. The observed OH signal from c-C3H5 + O2 can be scaled by using the same reference reaction, by reacting the HO2 formed from the reference reaction with NO to form OH radicals.30 100%

Using k(CH2OH + Cl2)/k(CH2OH + O2) ) 2.831 and [O2]/[Cl2] ) 30, it follows that approximately 90% of CH2OH radicals react with O2 while 10% react with Cl2. However, the CH2OH + Cl2 reaction regenerates Cl atoms which react with methanol to regenerate CH2OH radicals. Therefore, the CH2OH + Cl2 reaction slightly decreases (by ∼ 10%) the effective rate of the CH2OH + O2 reaction, which is of no consequence to the present analysis. Dividing the HO2 signal from c-C3H5 + O2, I(t), by the peak amplitude of the reference HO2 signal, AHO2, expresses the amplitude of the HO2 signal from c-C3H5 + O2 in terms of [HO2]/[Cl]0:

R[HO2]t [HO2]t I(t) ) ) AHO2 R[Cl]0 [Cl]0

(7)

where R is a proportionality constant relating the observed FM signal to the concentration of HO2. Representative time-resolved FM signals are shown in Figure 1. If the amplitude of the scaled HO2 signal from c-C3H5 + O2 is to be related to a yield of HO2 in reaction 2, corrections must be made for consumption reactions of HO2 that occur on the time scale of the HO2 formation. Correction for the HO2 selfreaction is straightforwardly accomplished by using the effective rate constant provided by the reference signal, whose decay is dominated by the HO2 + HO2 reaction, as shown in previous R + O2 investigations.1,2,12,13 However, the HO2 formed in the c-C3H5 + O2 reaction suffers additional HO2 removal reactions not present in the reference system, such as HO2 + OH and HO2 + c-C3H5O2. Therefore, the peak amplitude of the scaled HO2 signal is a lower limit to the total amount of HO2 formed during the experiment. The production of significant OH in the

HO2 + NO 98 OH + NO2

(8)

CH3OH is in significant excess over NO, so there is no complication from the relatively slow Cl + NO reaction.30 The peak of the OH signal from the reference reaction cannot be used directly to scale the OH signal from c-C3H5 + O2. The OH signal obtained from the reference system must be modeled to account for removal reactions, which reduce the peak amplitude of the signal. The OH signal from the reference system is modeled using the reactions listed in Table 3. Figure 2 shows observed and modeled OH signals from the reference reaction system at 296, 373, and 573 K. The radical density was calculated from the second-order decay of the HO2 signal observed without added NO, using the literature rate constant value for HO2 + HO2 (listed in Table 3). Typical radical densities in the OH/HO2 IR experiments are between 3 × 1013 and 8 × 1013 cm-3. As seen in Figure 2, the model accurately predicts the observed formation rate of the OH signal but overestimates the rate of decay. A better fit can be obtained by adjusting the rate coefficient for the NO + OH + M reaction, which is the major loss mechanism of OH at long times. However, since modeling the peak height and hence the [OH]/ [Cl]0 ratio is the major concern, the literature reaction rate constants are used without alteration. The fitted OH signals in Figure 2 correspond to a peak OH concentration of from 0.32 to 0.50 × [Cl]0 in the reference reaction system, depending on the temperature and concentrations used in each individual trial. The ratio of the peak [OH] concentration to the initial [Cl]0 predicted by the model ([OH]pk/[Cl]0)model is then used to scale the OH signal (IOH(t) ≡ R′[OH]t) observed from the Cl2/c-C3H6/ O2 system under identical photolysis conditions (and hence

Cl-Initiated Oxidation of Cyclopropane

J. Phys. Chem. A, Vol. 107, No. 12, 2003 1995

TABLE 3: Reactions and Rate Constants Used To Model the OH Signal Generated from the Cl2/CH3OH/O2/NO Systema reaction

Ab

CH3OH + Cl f HCl + CH2OH CH2OH + O2 f HO2 + CH2O HO2 + NO f OH + NO2 HO2 + HO2 f O2 + H2O2c OH + HO2 f H2O + O2 OH + CH3OH f CH2OH + H2O OH + NO + M f HNO2 + M OH + OH f O + H2O OH + OH + M f M + H2O2 OH + CH2O f HCO + H2O OH + HCO f CO + H2O OH + NO2 + M f HNO3 + M OH + HNO2 f H2O + NO2 HCO + O2 f HO2 + CO NO + CH2OH f CH2OH(NO)

5.4 × 10 3.77 × 10-15 3.5 × 10-12 2.2 × 10-13 4.8 × 10-11 2.12 × 10-13 2.79 × 10-32 7.89 × 10-14 6.89 × 10-31 4.73 × 10-12 1.70 × 10-10 2.60 × 10-30 6.24 × 10-12 5.60 × 10-12 2.50 × 10-11

n

Ea/R (K)

-11

5.94

2.65 2.60 -0.80 1.18

2284 250 599 250 444 -806 945 225

-2.90 1

68

k (298 K)b

ref

5.4 × 10-11 8.13 × 10-12 8.10 × 10-12 1.66 × 10-12 1.11 × 10-10 9.40 × 10-13 4.3 × 10-31 1.88 × 10-12 6.89 × 10-31 1.01 × 10-11 1.70 × 10-10 2.60 × 10-30 4.97 × 10-12 5.60 × 10-12 2.50 × 10-11

30 54 30 28 28 55 56 28 28 57 57 28 58 59 60

a The rate constants are written in the form A(T/296)ne(-Ea/RT). b Units are (cm3 molecule-1 s-1) for second-order reactions and (cm6 molecule-2 s-1) for third-order reactions. c The rate constant has a pressure dependent term; k ) (4.5 × 10-32)[M] + (2.2 × 10-13)e(599K/T).

TABLE 4: Individual Determinations of the Peak Amplitude of the Scaled OH Signal from c-C3H5 + O2

Figure 2. Time-resolved infrared OH signals from CH2OH/NO/O2 taken at 296, 373, and 573 K at a total density of 8.5 × 1017 cm-3. The dashed lines represent the OH signals predicted by the model described in the text. The peaks of the modeled curves represent the peak [OH]/[Cl]0 ) 0.45 (296 K), 0.40 (373 K), and 0.35 (573 K).

temp (K)

peak ratioa

modeled [OH]pk,ref/[Cl]0

peak [OH]/[Cl]0

296 296 296 373 373 473 473 473 573 573 598 623 648 663 678 698

0.09 0.04 0.08 0.10 0.05 0.13 0.11 0.14 0.15 0.13 0.18 0.17 0.19 0.23 0.23 0.28

0.45 0.45 0.45 0.40 0.47 0.43 0.40 0.35 0.32 0.41 0.39 0.39 0.39 0.38 0.40 0.40

0.04 0.02 0.04 0.04 0.03 0.05 0.04 0.05 0.05 0.05 0.07 0.07 0.07 0.09 0.09 0.11

a Ratio of the peak OH signal from Cl2/c-C3H6/O2 to the peak OH signal from the reference Cl2/CH3OH/O2/NO system.

identical initial Cl concentrations):

[OH]t [Cl]0

=

( )

IOH(t) [OH]pk AOH [Cl]0

(9)

ref,model

Here AOH (≡ R′[OH]pk,ref) is the observed peak amplitude of the OH signal from the reference reaction, and R′ is the constant of proportionality between OH concentration and the observed absorption signal, which includes absorption strength, line shape, and instrument functions. Table 4 lists the peak amplitude of the scaled OH signal from c-C3H5 + O2 for several temperatures. The peak of the scaled OH signal should be significantly smaller than the OH branching fraction, as the OH radicals are removed at a significant rate compared to that of their formation. Figure 3 compares the absorption signal for OH in Cl-initiated oxidation of cyclopropane with the infrared FM signal for HO2 under the same conditions. The peak amplitude of the OH signal occurs at an earlier time than the peak amplitude of the HO2 signal because of the more rapid removal of OH. The ratio of peak concentration to actual branching fraction will also be smaller for OH than for HO2. 2.2. FTIR Smog Chamber System at Ford Motor Company. Experiments are performed in a 140-L Pyrex reactor interfaced to a Mattson Sirus 100 FTIR spectrometer described elsewhere.32 The reactor is surrounded by 22 fluorescent black lamps (GE F15T8-BL), which are used to photochemically

Figure 3. OH absorption signal compared to the HO2 FM signal from the reaction of c-C3H5 + O2 at 296 K and a total density of 8.5 × 1017 cm-3. The signals are scaled to the initial Cl atom concentration as described in the text.

initiate the experiments. Cl atoms were generated by photolysis of molecular chlorine in 700 Torr total pressure of N2 diluent at 295 ( 2 K.

Cl2 + hν f 2Cl

(3)

Loss of c-C3H6 and formation of products were monitored by

1996 J. Phys. Chem. A, Vol. 107, No. 12, 2003

DeSain et al.

Fourier transform infrared spectroscopy using an infrared path length of 27.5 m and a spectral resolution of 0.25 cm-1. Infrared spectra were derived from 32 coadded interferograms. Reagents were obtained from commercial sources at the following stated purities (cyclopropane (>99.9%); chlorine (>99.99%); N2 (UHP); O2 (UHP)) and were used as received. In smog chamber experiments, unwanted loss of reactants and products by photolysis, dark chemistry, and wall reaction have to be considered. Control experiments were performed to check for these losses. Mixtures of c-C3H6 and air were subjected to UV irradiation for 5 min and then left in the dark for 30 min. There was no observable loss (