JOURNAL OF CHEMICAL PHYSICS

VOLUME 119, NUMBER 16

22 OCTOBER 2003

Selective detection of isomers with photoionization mass spectrometry for studies of hydrocarbon flame chemistry Terrill A. Cool,a) Koichi Nakajima, and Toufik A. Mostefaoui School of Applied and Engineering Physics, Cornell University, Ithaca, New York 14886

Fei Qi and Andrew McIlroy Combustion Research Facility, Sandia National Laboratories, Livermore, California 94551

Phillip R. Westmoreland and Matthew E. Law Department of Chemical Engineering, University of Massachusetts, Amherst, Massachusetts 01003

Lionel Poisson, Darcy S. Peterka, and Musahid Ahmed Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720

共Received 6 May 2003; accepted 28 July 2003兲 We report the first use of synchrotron radiation, continuously tunable from 8 to 15 eV, for flame-sampling photoionization mass spectrometry 共PIMS兲. Synchrotron radiation offers important advantages over the use of pulsed vacuum ultraviolet lasers for PIMS; these include superior signal-to-noise, soft ionization, and access to photon energies outside the limited tuning ranges of current VUV laser sources. Near-threshold photoionization efficiency measurements were used to determine the absolute concentrations of the allene and propyne isomers of C3 H4 in low-pressure laminar ethylene–oxygen and benzene–oxygen flames. Similar measurements of the isomeric composition of C2 H4 O species in a fuel-rich ethylene–oxygen flame revealed the presence of substantial concentrations of ethenol 共vinyl alcohol兲 and acetaldehyde. Ethenol has not been previously detected in hydrocarbon flames. Absolute photoionization cross sections were measured for ethylene, allene, propyne, and acetaldehyde, using propene as a calibration standard. PIE curves are presented for several additional reaction intermediates prominent in hydrocarbon flames. © 2003 American Institute of Physics. 关DOI: 10.1063/1.1611173兴

cilities that have employed tunable VUV laser sources.1–3 These include superior signal-to-noise, low light intensity ‘‘soft ionization’’ without the complications of multiphoton absorption and parent ion fragmentation, and photon energies readily tunable over the 8 –15 eV range required for comprehensive flame species concentration measurements. The new instrument is used for measurements of concentration profiles and photoionization efficiency 共PIE兲 curves for species sampled from well-characterized low-pressure laminar flames. Both stable and radical intermediates are detectable at mole fractions as low as 10⫺6 . Individual isomeric species may often be selectively detected and quantified; this is one of the most important advantages of VUV PIMS. We report here measurements of the absolute concentrations of the allene and propyne isomers of C3 H4 (m/e 40) in premixed laminar flames of ethylene–oxygen–argon and benzene–oxygen–argon. The separate contributions of ethenol and acetaldehyde isomers of C2 H4 O (m/e 44) are also distinguishable in fuel-rich ethylene–oxygen–argon flames. This is the first identification of ethenol 共vinyl alcohol兲 in a hydrocarbon flame and supports the hypothesis that vinyl alcohol may be a highly stable product of the decomposition of the primary HOCH2 CH2 adduct formed by the addition of OH to ethylene at the CvC double bond.6 Direct measurements of the absolute concentrations of flame species are possible with VUV PIMS when species photoionization cross sections are known.7 In our work we

I. INTRODUCTION

Progress in minimizing environmental pollution associated with hydrocarbon combustion requires the continuing development of kinetic models for the combustion of ethylene, ethane, propene, propane, propyne, and higher hydrocarbons including 1,3-butadiene and benzene. Further kinetic model development is also needed for oxygenated hydrocarbon fuels 共e.g., dimethyl ether, methanol, ethanol, propanol兲 with clean-burning characteristics 共low NOx , low soot兲. Improved descriptions of solid propellant chemistry and the chemistry of the incineration of hazardous materials are also of current interest. Kinetic model development in all of these reaction systems requires direct measurements of the absolute concentrations of combustion intermediates in laboratory flames under carefully controlled and documented flame conditions. Vacuum ultraviolet 共VUV兲 photoionization mass spectrometry 共PIMS兲, applied to the selective detection of flame species, is a powerful new approach for studies of flame chemistry,1– 4 which provides a valuable supplement to the traditional use of electron-impact mass spectrometry 共EIMS兲 for such studies. A flame-sampling molecular beam time-offlight mass spectrometer 共TOFMS兲, recently designed and constructed for use with a synchrotron radiation light source,5 provides significant improvements over previous faa兲

Electronic mail: [email protected]

0021-9606/2003/119(16)/8356/10/$20.00

8356

© 2003 American Institute of Physics

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 119, No. 16, 22 October 2003

Hydrocarbon flame chemistry

8357

measure photoionization cross sections for ‘‘target’’ species by comparing ion signals recorded from a binary mixture of the target with a ‘‘standard’’ species of known photoionization cross section. Photoionization cross sections for allene, propyne, ethylene and acetaldehyde are presented here, measured with propene as a standard, for photon energies from 9.7 to 11.75 eV. We have measured cross sections for several additional reaction intermediates observed in ethylene– oxygen–argon, benzene–oxygen–argon, and 1,3-butadiene– hydrogen–oxygen–argon flames.8 II. EXPERIMENT

Our flame-sampling photoionization time-of-flight mass spectrometer is described in detail elsewhere.5 The apparatus consists of a low-pressure flame chamber, a differentially pumped flame-sampling system, and a time-of-flight mass spectrometer 共TOFMS兲. It is coupled to a 3 m Eagle monochromator used to disperse synchrotron radiation at terminal 3 of the Chemical Dynamics Beamline of the Advanced Light Source 共ALS兲 at the Lawrence Berkeley National Laboratory. Premixed fuel–oxygen–argon flames are stabilized on a flat flame burner 共McKenna兲 at pressures ranging from 20 to 40 Torr. Molecular-beam flame sampling is accomplished with a quartz sampling cone of 0.2 mm diam orifice mounted on a water-cooled flange, which serves as the end wall of the flame chamber. A nickel skimmer, with its 2 mm diam aperture located 23 mm downstream from the sampling cone orifice, forms a molecular beam, which passes horizontally through the 1.27 cm gap between the repeller and extractor plates of a conventional 共Wiley–McLaren9兲 TOFMS. A grounded accelerator plate is located 1.27 cm above the extractor plate; the extractor and accelerator plates have 1.27 cm diam apertures, covered with wire mesh, centered on the vertical flight tube axis of the TOFMS. Pulse gating of the voltage of the repeller plate is used to propel ions up the 1.3 m flight tube to a multichannel plate 共MCP兲 detector. Optimal plate voltages were found to be: Repeller, 2997 V 共ungated兲; extractor, 3005 V, accelerator, 0 V. The 8 V difference in potential between extractor and repeller plates, maintained between gating pulses, acts as a discriminator to reduce a small random background signal caused by ions escaping the source region. Pulsed gating voltages of 325 V amplitude and 15 kHz frequency, with 30 ns rise and fall times and 1100 ns pulse duration, are applied to the repeller plate. The gate pulse duration is sufficient to draw out most of the available ions located below the aperture of the extractor plate. The TOFMS mass resolution is typically m/⌬m⫽400. A multiscaler 共FAST Comtec P7886兲 records TOFMS mass spectra in 15 008 channels of 2 ns width. Useful spectra with a dynamic range of 105 are obtained with 5⫻105 to 4 ⫻106 sweeps. The 3 m off-plane Eagle monochromator, with a 600 lines/mm tungsten grating, delivers photon fluxes of about 1014 photons/s with an energy resolution E/⌬E⭓400 under usual operating conditions in the 9–12 eV photon energy range. The monochromator has an ultimate resolution of 2600 for these energies. A silicon photodiode 共International Radiation Detectors, Inc. SXUV-100兲 records the variation in photon flux with photon energy. Flame temperatures were

FIG. 1. Photoionization cross section for propene measured by Person and Nicole 共Ref. 11兲 共left-hand scale, 1 Mb⫽10⫺18 cm2 ).

measured with a platinum–platinum-rhodium thermocouple that had been given a beryllium oxide–yttrium oxide anticatalytic coating and had been carefully calibrated against flame temperatures measured with the sodium D-line reversal method.10 III. PHOTOIONIZATION CROSS SECTIONS

The absolute concentrations of reaction intermediates are quite sensitive to the reaction mechanisms and rate constants that determine their rates of formation and consumption. Accurate kinetic modeling, therefore, requires measurements of the spatial profiles of the absolute concentrations of as many flame species as possible. Current kinetic models for flame chemistry are largely based on concentration measurements performed by conventional electron-impact mass spectrometry 共EIMS兲, which is universally applicable to flame species because it does not require the detailed spectroscopic information needed for application of less intrusive optical methods such as LIF 共laser induced fluorescence兲, REMPI 共resonance-enhanced multiphoton ionization兲, CRDS 共cavity ring-down spectroscopy兲 or CARS 共coherent anti-Stokes Raman spectroscopy兲. The universal use of EIMS or VUV PIMS for absolute concentration measurements requires knowledge of either electron-impact ionization or photoionization cross sections, respectively, for each species of interest. In this section we describe photoionization cross-section measurements for allene, propyne, ethylene, and acetaldehyde with an accuracy 共⫾25% uncertainty兲 suitable for kinetic modeling. Person and Nicole11 have carefully measured photoionization cross sections for propene, propyne, ethylene, and several other molecules. The cross section for propene over the 9.7–11.75 eV range, shown in Fig. 1, is the standard we use to calibrate the response of our system for the measurement of cross sections for various target molecules.7 Binary mixtures of nominally 10 Torr propene with 10 Torr of a given target were prepared in a 3.8-liter stainless steel sample cylinder to which an additional 2300 Torr of argon

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

8358

J. Chem. Phys., Vol. 119, No. 16, 22 October 2003

Cool et al.

FIG. 2. Photoionization cross section for allene determined from measurements of ion signals at m/e⫽40 and 42 from binary mixtures of allene and propene.

FIG. 3. Photoionization cross section for propyne determined from measurements of ion signals at m/e⫽40 and 42 from binary mixtures of propyne and propene. The dashed curve refers to the cross-section measurements of Person and Nicole 共Ref. 11兲.

was added. The combined gases were allowed to mix for at least eight hours and then were introduced as a room temperature flow into the flame chamber at a flow rate of 0.1 slm 共standard liters per minute兲, along with a second flow of argon 共as a shroud gas兲 of 0.15 slm. The flame chamber pressure was kept constant at 9 Torr with a servo-controlled throttle value on the chamber exhaust. This arrangement enabled run times of about 1 hour, during which the ratio of target ion signal to propene ion signal was recorded over the 9.7–11.75 eV photon energy range. Multiplication of this ratio by the ratio of propene to target species partial pressures yields the ratio of the photoionization cross section for the target to that of propene as a function of photon energy. In this paper ‘‘ion signal’’ refers to the ion count at a given mass/charge ratio, obtained by integration of the accumulated ion counts per channel over 25 multiscaler channels 共50 ns兲. This permitted integration over the entire temporal profile of each mass peak, while avoiding overlapping contributions from adjacent mass peaks. The ratio of cross sections, multiplied by the propene cross section of Fig. 1, gives the desired cross section for the target species. Although the photon flux was routinely monitored during these measurements, the ratio of recorded ion signals for the target species and propene is independent of variations in photon flux with photon energy. Photoionization cross sections measured for allene, propyne, ethylene, and acetaldehyde are presented in Figs. 2–5. Tabulated values for these cross sections and the propene cross sections of Person and Nicole11 are available as supplementary material.12 The dashed lines shown with the propyne and ethylene data are the cross sections measured by Person and Nicole.11 The comparison between the solid and dashed lines for these two molecules provides a self-consistency check between our measurements and those of Person and Nicole. The cross sections for allene were obtained by averaging the results for three separately prepared samples; two samples each were used for the propyne, ethylene, and ac-

etaldehyde measurements. The error bars shown with the data of Figs. 2–5 indicate sample standard deviations (⫾ ␴ s) at several photon energies. The probable error in the propene cross section data of Person and Nicole is unlikely to exceed ⫾20%.11 This value, when combined with our sample standard deviations, suggests an overall uncertainty of ⫾25% for our cross section measurements. IV. CONCENTRATION MEASUREMENTS OF ISOMERS OF C3 H4 AND C2 H4 O

Because the mechanisms for formation and destruction may be greatly influenced by the molecular structure of a given isomer, knowledge of the isomeric composition of flame species is needed for accurate kinetic modeling. In this

FIG. 4. Photoionization cross section for ethylene determined from measurements of ion signals at m/e⫽28 and 42 from binary mixtures of ethylene and propene. The dashed curve refers to the cross-section measurements of Person and Nicole 共Ref. 11兲.

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 119, No. 16, 22 October 2003

FIG. 5. Photoionization cross section for acetaldehyde determined from measurements of ion signals at m/e⫽44 and 42 from binary mixtures of acetaldehyde and propene.

section we present flame concentration measurements of the allene and propyne isomers of C3 H4 and the ethenol and acetaldehyde isomers of C2 H4 O. A series of measurements of the spatial profiles of flame species were performed for a fuel-rich 共⌽⫽1.9兲 ethylene– oxygen–argon flame at a pressure of 30 Torr at photon energies of 10.3, 11.0, 12.0, 13.0, and 14.0 eV. The flame temperature profile and spatial profiles 共concentrations versus distance from the burner face兲13 for the m/e⫽28, 40, and 44 species, recorded at 11.0 eV, are presented in Fig. 6. The m/e⫽28 data are for C2 H4 ; m/e⫽40 is the sum of the allene and propyne isomers of C3 H4 ; m/e⫽44 is the sum of the ethenol and acetaldehyde isomers of C2 H4 O. The species concentrations 共mol/cm3兲 are directly proportional to the recorded species ion signals divided by the species photoion-

Hydrocarbon flame chemistry

8359

FIG. 7. A comparison of the m/e⫽40 ion signal 共symbols兲, recorded from a binary mixture of allene 共48.6%兲 and propyne 共51.4%兲, with the photoionization cross section 共solid curve兲 calculated with Eq. 共1兲. The arbitrary scale of the ion signal has been adjusted to match the cross section computed for the mixture.

ization cross sections at 11.0 eV. Because the initial concentration of C2 H4 is known, the absolute concentration profile for C2 H4 is directly calculated from the profile of C2 H⫹ 4 ion signal. Determination of the concentrations for m/e⫽40 and 44 of Fig. 6 required measurements of ion signals for all three masses at 11.0 eV, knowledge of the photoionization cross sections at 11.0 eV for C2 H4 and each of the isomers at m/e⫽40 and 44, and the isomeric compositions for m/e⫽40 and 44. The 11.0 eV cross section for ethenol was estimated at 10.9 Mb (1 Mb⫽10⫺18 cm2 ) 共cf. Fig. 11兲; 11.0 eV cross sections for the other species, taken from the data of Figs. 2–5, are 7.6, 8.0, 27.2, and 43.5 Mb for ethylene, acetaldehyde, allene, and propyne, respectively. The remainder of this section is devoted to an explanation of the procedures used for the determination of the isomeric compositions for m/e⫽40 and 44, at a 5 mm distance from the burner face, of the 共⌽⫽1.9兲 ethylene–oxygen–argon flame of Fig. 6. A. Allene and propyne

FIG. 6. Temperature profile and concentration profiles for C2 H4 , the ethenol/acetaldehyde isomers of C2 H4 O, and the allene–propyne isomers of C3 H4 . For a fuel-rich (⌽⫽1.9) C2 H4 – O2 – Ar flame at 30 Torr with flow rates: C2 H4 ⫽0.493 slm, O2 ⫽0.778 slm, and Ar⫽1.27 slm.

Figure 7 displays the results of an experiment conducted to test our procedure for determinations of isomeric composition from photoionization efficiency measurements. A mixture of 10.8 Torr allene and 11.4 Torr propyne with 2300 Torr argon was prepared in a 3.8 liter stainless steel sample cylinder and allowed to mix for 13 hours. This mixture was then used to maintain a room temperature flow through the flame chamber as described in Sec. III. The symbols of Fig. 7 are the ion signals at m/e⫽40 recorded for the mixture as the photon energy was varied over the range from 9.6 to 11.05 eV. The ion signals have been normalized point-by-point by the photon flux to remove the influence of flux variations with photon energy. The ion signals of Fig. 7 are compared with the photoionization cross section for a mixture of allene and propyne 共solid curve兲 constructed with the cross section data of Figs. 2 and 3 according to the relationships:

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

8360

Cool et al.

J. Chem. Phys., Vol. 119, No. 16, 22 October 2003

FIG. 8. A comparison of the m/e⫽40 ion signal 共symbols兲, recorded for the fuel-rich (⌽⫽1.9) C2 H4 – O2 – Ar flame of Fig. 6 with the photoionization cross section 共solid curve兲 calculated with Eq. 共1兲. The computed cross section is for a best-fit isomeric composition of 42% allene and 58% propyne. Consideration of the uncertainties in the curve fitting yields mole fractions x a ⫽0.42⫾0.04 and x p ⫽0.58⫾0.04. These data correspond to conditions 5 mm from the burner face 共Ref. 13兲.

␴ ⫽x a ␴ a ⫹x p ␴ p ,

共1兲

x a ⫹x p ⫽1.

共2兲

Here the photoionization cross section ␴ is expressed as a combination of cross sections ␴ a and ␴ p for allene and propyne weighted by their respective mole fractions x a and x p . The comparison was facilitated by adjusting the arbitrary scale for the ion signals as required to match the calculated photoionization cross section. The agreement between the flux-normalized ion signals and photoionization cross sections for the allene–propyne mixture of known composition, demonstrated in Fig. 7, validates our method for determinations of isomeric composition. Application of this approach to flame species requires photoionization cross sections for all isomers of a given m/e. Occasionally it may also be necessary to assign cross sections for the potential formation of photofragment ions with this m/e from precursor species of greater mass. Figure 8 displays the ion signal 共symbols兲 for the C3 H4 species at m/e⫽40 for a fuel-rich 共⌽⫽1.9兲 30 Torr ethylene– oxygen flame recorded for photon energies from 9.7 to 10.6 eV at a position 5 mm 共after correction for probe-sampling effects13兲 from the burner face. At this position the concentration of C3 H4 species approaches its maximum value 共cf. Fig. 6兲 reached at 6.5 mm. The values for allene and propyne mole fractions were varied by trial and error to best-fit the observed flux-normalized ion signals with the calculated 关cf. Eqs. 共1兲 and 共2兲兴 photoionization cross sections for the assumed mixture composition. Here again the arbitrary scale of the ion signals was adjusted to assist in the curve fitting. This procedure has also been applied to benzene– oxygen–argon flames; the results are shown in Figs. 9共a兲 and 9共b兲. The ratios of propyne to allene concentrations, 关 CH3 C

FIG. 9. A comparison of the m/e⫽40 ion signal 共symbols兲 with the photoionization cross section 共solid curve兲 calculated with Eq. 共1兲. 共a兲 Recorded for a fuel-rich (⌽⫽1.4) C6 H6 – O2 – Ar flame 共30 Torr, C6 H6 ⫽0.149 slm, O2 ⫽0.80 slm, Ar⫽0.57 slm兲. The computed cross section is for a best-fit isomeric composition of 45% allene and 55% propyne. Consideration of the uncertainties in curve fitting yields mole fractions X a ⫽0.45⫾0.04 and X p ⫽0.55⫾0.04. The burner position was fixed at 1.75 mm from the sampling position 共Ref. 13兲. 共b兲 For a fuel-lean (⌽⫽0.7) C6 H6 – O2 – Ar flame 共20 Torr, C6 H6 ⫽0.093 slm, O2 ⫽1.00 slm, Ar⫽0.60 slm兲. The best-fit composition is 50% allene and 50% propyne, X a ⫽0.50⫾0.04 and X p ⫽0.50 ⫾0.04. The burner was 1.5 mm from the sampling position 共Ref. 13兲.

wCH兴/关CH2 vCvCH2 兴 , obtained with the fits of Figs. 8 and 9 are quoted, with estimated uncertainties, in Table I. The very rapid forward and reverse rates for the allenepropyne isomerization reactions14 –20 CH2 vCvCH2 ↔CH3 CwCH,

共3兲

drive the isomeric composition toward thermal equilibrium in competition with isomer-specific chemical reactions that consume allene and propyne. The cyclopropene isomer is an intermediate in the allene-propyne isomerization reactions 共3兲.17–20 The ionization energies21 of cyclopropene 共9.67 eV兲 and allene 共9.69 eV兲 are too similar to distinguish between allene and cyclopropene in our experiments. A contribution to the ion signal from cyclopropene seems unlikely, however,

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 119, No. 16, 22 October 2003

Hydrocarbon flame chemistry

8361

TABLE I. Propyne–allene concentration ratios. Flame

关 CH3 CwCH兴/关CH2 vCvCH2 兴

ethylene–oxygen 共⌽⫽1.9兲a benzene–oxygen 共⌽⫽1.4兲b benzene–oxygen 共⌽⫽0.7兲c

1.38⫾0.25 1.22⫾0.22 1.0⫾0.2

Ethylene–oxygen–argon 共0.493/0.778/1.27兲 关Respective flow rates in standard 共1 atmosphere, 273 K兲 liters per minute 共slm兲兴, 30 Torr, 5.0 关Position after correction for probe sampling effects 共Ref. 13兲兴 mm from burner face. b Benzene–oxygen–argon 共0.015/0.80/0.57兲 关Respective flow rates in standard 共1 atmosphere, 273 K兲 liters per minute 共slm兲兴, 30 Torr, 1.75 关Position after correction for probe sampling effects 共Ref. 13兲兴 mm from burner face. c Benzene–oxygen–argon 共0.093/1.00/0.60兲 关Respective flow rates in standard 共1 atmosphere, 273 K兲 liters per minute 共slm兲兴, 20 Torr, 1.5 关Position after correction for probe sampling effects 共Ref. 13兲兴 mm from burner face. a

because the concentration of this transient intermediate at 1520 K, the flame temperature 5 mm from the burner surface 共cf. Fig. 6兲, is probably negligible compared with those of the allene and propyne isomers of much greater thermochemical stability.17–20 The equilibrium constant for the isomerization reaction 共3兲, calculated with current thermochemical properties21,22 is K eq⫽ 关 CH3 CwCH兴/关CH2 vCvCH2 兴 ⫽1.9,

共4兲

at 1520 K. The smaller ratios given in Table I for the ethylene and benzene flames signify departures from thermal equilibrium. These results are qualitatively similar to the findings of Davis et al.,23 who used a water-cooled gas sampling probe and gas chromatograph to study the oxidation of propyne in a turbulent flow reactor and a premixed laminar flame. They cited the reactions CH3 CwCH⫹H→C2 H2 ⫹CH3 ,

共5兲

CH3 CwCH⫹OH→C3 H3 ⫹H2 O,

共6兲

FIG. 10. A comparison of the m/e⫽44 ion signal 共symbols兲, recorded for the fuel-rich (⌽⫽1.9) C2 H4 – O2 – Ar flame of Fig. 6, with the photoionization cross section 共solid curve兲 calculated for an best-fit isomeric composition of 37⫾5% ethenol and 63⫾5% acetaldehyde. This calculation is based on the cross section for acetaldehyde of Fig. 5 and the cross section 共solid curve兲 for ethenol of Fig. 11; the error limits reflect the uncertainty 共⫾25%兲 of the estimated cross section for ethenol.

CvC ␲-orbitals of alkenes 共e.g., ethylene, propene, butene isomers兲 typically reach values 1 eV above threshold of about 8 –10 Mb.25 A reasonable fit of the solid curve to the m/e⫽44 ion signals, displayed in Fig. 10, is obtained for an ethenol– acetaldehyde concentration ratio 关 CH2 CHOH兴/关CH3 CHO兴 ⫽0.6. Similar fits using the upper and lower dashed curves

as possible sources for selective depletion of propyne in competition with the isomerization reactions 共3兲. B. Ethenol and acetaldehyde

An interesting result of this study is the observation of ethenol, a species never before identified in a hydrocarbon flame, although Ruscic and Berkowitz have observed ethenol and acetaldehyde as products of reactions between atomic fluorine and ethanol.24 The ion signals for m/e⫽44 displayed in Fig. 10 exhibit a dependence on photon energy qualitatively similar to that shown for the allene–propyne data of Figs. 8 and 9. The onset of ionization of acetaldehyde near 10.2 eV is clearly visible, superimposed on the ion signal for a second species with apparent ionization energy near 9.3 eV. The only plausible assignment for this species is ethenol with an IE of 9.33 eV.24 The data show no clear evidence of a potential contribution to the ion signal from ethylene oxide above its threshold 共IE⫽10.56 eV21兲. The solid curve shown with the data of Fig. 10 is based on the cross sections for acetaldehyde of Fig. 5 and estimated cross sections for ethenol presented in Fig. 11. Dashed curves shown in Fig. 11 bracket the range 共⫾25%兲 of probable uncertainty associated with the estimated cross section 共solid curve兲. This estimate is based on the observation that cross sections for ionization from the

FIG. 11. Estimated photoionization cross sections for ethenol. The photoionization efficiency of ethenol for photon energies ranging from threshold near 9.3 eV to the ionization energy of acetaldehyde at 10.23 eV 共Ref. 21兲 is assumed to be proportional to the ion signals of Fig. 10 over this range 共hence the wiggles兲. We estimate a value of 8 Mb for the cross section 共solid curve兲 at 10.2 eV, based on the cross sections observed for photoionization from the CvC ␲-bonds of ethylene, propene, and isomers of butene 共Ref. 25兲. The dashed curves indicate the probable range of uncertainty 共⫾25%兲 in the estimated cross section 共solid curve兲.

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

8362

Cool et al.

J. Chem. Phys., Vol. 119, No. 16, 22 October 2003

FIG. 12. PIE curves measured in benzene–oxygen flames. Solid circles: fuel-rich 共⌽⫽1.4兲; open circles: fuel-lean 共⌽⫽0.7兲. 共a兲 m/e⫽42, 共b兲 m/e⫽50, 共c兲 m/e⫽52, 共d兲 m/e⫽66. The labels give the accepted ionization energies 共Ref. 21兲 for species identified by the observed ionization thresholds.

of Fig. 11 give values of 0.47 and 0.73, respectively, for the ethenol–acetaldehyde ratio. If we estimate the difference in Gibbs free energy between ethenol and acetaldehyde at 1520 K to be 8.5 kcal/mol,21,26 then the ethenol–acetaldehyde ratio 关 CH2 CHOH兴/关CH3 CHO兴 for thermal equilibrium at 1520 K is only 0.06. Despite the thermochemical stability of acetaldehyde relative to ethenol, a very high barrier 共⬇56 kcal/ mol兲 to the unimolecular tautomerization of ethenol to acetaldehyde exists.27 Indeed, ethenol is known to be highly stable in the gas phase.24,28 –30 How is ethenol formed in the ethylene–oxygen flame? Further experimental and modeling efforts are needed to answer this question. Re-examination of formation mechanisms for acetaldehyde may reveal previously unidentified product branching channels leading to ethenol. Another hypothesis is that vinyl alcohol may be a highly stable product of the decomposition of the primary HOCH2 CH2 adduct formed by the addition of OH to ethylene at the CvC double bond.6 Indeed, Kirchner et al.31 and Howard32 have studied the reaction of chloroethylenes with the OH radical and conclude that OH addition at the carbon–carbon ␲-bond results in Cl atom elimination, e.g.,

OH⫹CHCIvCCI2 →HO–CHCl–CCI2 →CCI2 vCHOH⫹Cl.

共7兲

Evidence for reaction 共7兲 in a methane–oxygen flame doped with trichloroethylene has been presented.33 An analogous reaction of the chemically activated adduct of the OH attack on ethylene may also occur34 OH⫹C2 H4 →HO–CH2 – CH2 →CH2 CHOH⫹H,

共8兲

despite the stronger C–H bond and a higher barrier to H-atom elimination. Howard32 has suggested that the adduct may also react with O2 to form ethenol and HO2 : OH⫹C2 H4 →HO–CH2 – CH2 , HO–CH2 – CH2 ⫹O2 →CH2 CHOH⫹HO2 .

共9兲 共10兲

At elevated flame temperatures (T⬎800 K), however, the HO–CH2 – CH2 adduct may not survive rapid unimolecular decomposition6 to react with O2 by reaction 共10兲. Moreover, at flame temperatures above 650 K, the H-atom abstraction OH⫹C2 H4 →H2 O⫹C2 H3 ,

共11兲

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 119, No. 16, 22 October 2003

Hydrocarbon flame chemistry

8363

FIG. 13. 共a兲–共c兲 PIE curves measured in benzene/oxygen flames. Solid circles: Fuel-rich 共⌽⫽1.4兲; open circles: Fuel-lean 共⌽⫽0.7兲. 共a兲 m/e⫽78, 共b兲 m/e⫽92, 共c兲 m/e⫽94. The labels give the accepted ionization energies 共Ref. 21兲 for species identified by the observed ionization thresholds. 共d兲 PIE curve for 1,3-butadiene measured in 30 Torr. 1,3-butadiene–hydrogen–oxygen–argon flames. Solid circles: Respective flow rates⫽共0.17/1.00/1.00/3.00兲 共slm兲; open circles: Respective flow rates⫽共0.05/1.00/0.65/2.00兲 共slm兲.

dominates OH addition,6 and the production of ethenol by either 共8兲 or 共10兲 will be inhibited. Nevertheless, reactions 共8兲 and possibly 共10兲 provide plausible sources for the production of ethenol in low-temperature regions of ethylene– oxygen flames. The high barrier for tautomerization suggests that once ethenol is formed, it is quite stable; the decompositions of the two C2 H4 O isomers are uncoupled, and may therefore occur at different rates and by different flame reaction mechanisms. V. FLAME SPECIES IDENTIFICATIONS WITH NEAR-THRESHOLD PIMS

The ease with which the photon energy may be precisely tuned near the ionization thresholds for flame species with the synchrotron light source is a feature unmatched by laser sources. The available energy resolution (E/⌬E⬇400), far superior to that of conventional EIMS, facilitates species identification with observations of apparent ionization thresholds. Nevertheless there are questions that arise regarding the influence that rotational and vibrational excitation of flame-

sampled molecules may have on observed ionization thresholds. Moreover, field ionization from high-lying Rydberg states, induced by the 250 V/cm fields used to extract ions from the source region, must be considered. Rotational excitation ensures that Rydberg levels lying within 共3/2兲 kT 共⭐0.04 eV for our apparatus35兲 of the ionization limit contribute to the ion signal.36,37 Similarly, field ionization of Rydberg states lying within ⬃6.1冑250⫽96 cm⫺1 共0.012 eV兲 of the ionization limit38,39 may also contribute to an apparent threshold below the true IE. These effects are comparable to the energy spread 共⬃0.025 eV at 10 eV兲 of the dispersed synchrotron radiation used in these studies. A more serious complication may be the influence of hot bands associated with incomplete vibrational relaxation during the gas dynamic expansion of the molecules sampled by the quartz cone. We originally feared that such effects might place severe limitations on the ability to use apparent ionization thresholds for species identification. As it happens, in practice these problems do not seriously compromise species discrimination based on apparent ionization thresholds.4 In the remainder of this section we present a few examples in support of this conclusion.

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

8364

TABLE II. Hydrocarbon flame species for which PIE curves have been recorded. Species methyl radical water acetylene ethylene formaldehyde oxygen hydrogen peroxide propargyl radical allene propyne ketene ethenol acetaldehyde dimethyl ether formic acid diacetylene vinylacetylene 1,3-butadiene glyoxal 1,3-cyclopentadiene diethyl ether benzene toluene phenol phenylacetylene a

Cool et al.

J. Chem. Phys., Vol. 119, No. 16, 22 October 2003

Formula

m/e

IE 共eV兲a

CH3 H2 O C2 H2 C2 H4 H2 CO O2 H2 O2 C3 H3 C3 H4 C3 H4 CH2 CO CH2 CHOH CH3 HCO (CH3 ) 2 O HCOOH C4 H2 C4 H4 C4 H6 (CHO) 2 C5 H6 (C2 H5 ) 2 O C6 H6 C7 H8 C6 H6 O C8 H6

15 18 26 28 30 32 34 39 40 40 42 44 44 46 46 50 52 54 58 66 74 78 92 94 102

9.84 12.62 11.40 10.51 10.88 12.07 10.58 8.67 9.69 10.36 9.62 9.33 10.23 10.03 11.33 10.17 9.58 9.07 10.2 8.57 9.51 9.24 8.83 8.49 8.82

Reference 21.

Benzene–oxygen–argon flames provide a rich source of reaction intermediates that may be unambiguously identified with near-threshold VUV PIMS. Photoionization efficiency 共PIE兲 curves 共near threshold variation in ion yield with photon energy兲 are presented in Figs. 12共a兲–12共d兲 and 13共a兲– 13共c兲 for several prominent intermediates observed in both fuel-rich 共⌽⫽1.4兲 and fuel-lean 共⌽⫽0.7兲 flames. Figure 13共d兲 presents a PIE curve for 1,3-butadiene, a species observed for both ethylene and benzene flames, measured with a 1,3-butadiene–hydrogen–oxygen–argon flame. The apparent ionization thresholds of the PIE curves of Figs. 12 and 13 are seen to closely match the accepted ionization energy values21 for the species labeled on the figures. With conventional EIMS, some of these species would be indistinguishable from several possible isomers. Similar PIE curves have been recorded for a number of stable and radical intermediates in ethylene–oxygen, benzene–oxygen, and 1,3-butadiene–hydrogen–oxygen flames; a complete tabulation of these species is given in Table II. In every case the apparent ionization thresholds are observed to closely match the known ionization energies21 tabulated in the last column of Table II. VI. CONCLUSION

In this paper we have presented two interesting examples of the use of flame-sampling VUV PIMS for quantitative measurements of the concentrations of isomer pairs of importance in hydrocarbon combustion. The concentrations of the ethenol–acetaldehyde isomers at m/e⫽44 shown in Fig. 6 are comparable to those of many key intermediates in the fuel-rich ⌽⫽1.9 ethylene–oxygen flame and exceed those of

the allene–propyne m/e⫽40 isomers. These data suggest that the mechanisms of formation and consumption of ethenol need to be incorporated in continuing development40 of models for the combustion of C2 hydrocarbons. The ability to supplement measurements of spatial profiles of photoions, created at a fixed photon energy, with measurements of species PIE curves, over a range of photon energies at fixedflame positions, adds an important second dimension to the use of flame-sampling mass spectrometry for the development of kinetic models of flame chemistry. While the focus of this paper is on the isomeric selectivity of VUV PIMS, another significant advantage of this approach is the ability to precisely tailor the photon energy to minimize parent ion fragmentation. An extensive database21 giving the ionization energies of parent species along with the appearance energies for photofragment ions is a valuable resource available to the experimenter. Spatial profiles of the absolute concentrations of flame species are directly measurable when photoionization cross sections are known over an appropriate range of photon energies. The data of Fig. 6 exemplify this approach. Photoionization cross sections for many of the stable intermediates likely to be observed in hydrocarbon flames can be measured with an accuracy 共⫾25% uncertainty兲 suitable for combustion modeling, as discussed in Sec. III, with the use of binary mixtures of a target species with propene. In related work, Neumark and co-workers41 are exploring the use of synchrotron radiation for the measurement of photoionization cross sections of common combustion radicals, e.g., C2 H3 and C3 H3 . Theoretical and semiempirical methods may be very useful for cross section estimates when reliable experimental measurements are unavailable. ACKNOWLEDGMENTS

This research was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences, and Biosciences, and by the U.S. Army Research Office, Chemical Sciences Division. 1

J. H. Werner and T. A. Cool, Twenty-Seventh Symposium (International) on Combustion 共The Combustion Institute, Pittsburgh, PA, 1998兲, pp. 413– 423. 2 J. H. Werner and T. A. Cool, Combust. Flame 117, 78 共1999兲; 120, 125 共2000兲. 3 A. Mcllroy, T. D. Hain, H. A. Michelsen, and T. A. Cool, Twenty-Eighth Symposium (International) on Combustion 共The Combustion Institute, Pittsburgh, PA, 2000兲, pp. 1647–1653. 4 F. Qi and A. Mcllroy, ‘‘Identifying combustion intermediates via tunable vacuum ultraviolet photoionization mass spectrometry’’ 共unpublished兲. 5 T. A. Cool, A. Mcllroy, F. Qi, P. R. Westmoreland, L. Poisson, D. S. Peterka, and M. Ahmed, ‘‘A photoionization mass spectrometer for studies of flame chemistry with a synchrotron light source’’ 共unpublished兲. 6 F. P. Tully, Chem. Phys. Lett. 143, 510 共1988兲; 96, 148 共1983兲. 7 J. H. Werner and T. A. Cool, Combust. Flame 120, 125 共2000兲. 8 F. Qi, T. A. Cool, K. Nakajima, C. A. Taatjes, and A. Mcllroy, ‘‘Photoionization cross sections for reaction intermediates in hydrocarbon combustion’’ 共unpublished兲. 9 W. C. Wiley and H. I. McLaren, Rev. Sci. Instrum. 26, 1150 共1955兲. 10 J. S. Bernstein, A. Fein, J. B. Choi, T. A. Cool, R. C. Sausa, S. L. Howard, R. J. Locke, and A. W. Miziolek, Combust. Flame 92, 85 共1993兲; J. S. Bernstein, Ph.D. thesis, Cornell University, 1991; J. H. Werner, Ph.D. thesis, Cornell University, 1998. 11 J. C. Person and P. P. Nicole, J. Chem. Phys. 53, 1767 共1970兲; 49, 5421 共1968兲. 12 See EPAPS Document No. E-JCPSA6-003340 for photoionization cross

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 119, No. 16, 22 October 2003 sections for propene, propyne, ethylene, and acetaldehyde. A direct link to this document may be found in the online article’s HTML reference section. The document may also be reached via the EPAPS homepage 共http:// www.aip.org/pubservs/epaps.html兲 or from ftp.aip.org in the directory /epaps/. See the EPAPS homepage for more information. 13 These concentration profiles have been shifted toward the burner by 1 mm to compensate for the fact that the sampling cone draws from a flame region extending several nozzle diameters ahead of the cone: O. I. Smith and D. W. Chandler, Combust. Flame 63, 19 共1986兲; D. Stepowski, D. Puechberty, and M. J. Cottereau, Eighteenth Symposium (International) on Combustion 共The Combustion Institute, Pittsburgh, PA, 1981兲, pp. 1567– 1573. 14 A. Lifshitz, M. Frenklach, and A. Burcat, J. Phys. Chem. 79, 1148 共1975兲; 80, 2437 共1976兲. 15 C. H. Wu and R. D. Kern, J. Phys. Chem. 91, 6291 共1987兲. 16 Y. Hidaka, T. Nakamura, A. Miyauchi, T. Shiraishi, and H. Kawano, Int. J. Chem. Kinet. 21, 643 共1989兲. 17 M. Karni, I. Oref, S. Barzilai-Gilboa, and A. Lifshitz, J. Phys. Chem. 92, 6924 共1988兲. 18 C. F. Melius, J. A. Miller, and E. M. Evleth, Twenty-Fourth Symposium (International) on Combustion 共The Combustion Institute, Pittsburgh, PA, 1992兲, pp. 621– 628. 19 J.-F. Pauwels, J. V. Volponi, and J. A. Miller, Combust. Sci. Technol. 110–111, 249 共1995兲. 20 J. H. Keifer, S. S. Kumaran, and P. S. Mudipalli, Chem. Phys. Lett. 224, 51 共1994兲. 21 NIST Standard Reference Database Number 69, March 2003, NIST Chemistry WebBook, http://webbook.nist.gov/ 22 M. S. Robinson, M. L. Polak, V. M. Bierbaum, C. H. DePuy, and W. C. Lineberger, J. Am. Chem. Soc. 117, 6766 共1995兲.

Hydrocarbon flame chemistry

8365

S. G. Davis, C. K. Law, and H. Wang, J. Phys. Chem. A 103, 5889 共1999兲; Twenty-Seventh Symposium (International) on Combustion 共The Combustion Institute, Pittsburgh, PA, 1998兲, pp. 305–312. 24 B. Ruscic and J. Berkowitz, J. Chem. Phys. 101, 10936 共1994兲. 25 H. Koizumi, J. Chem. Phys. 95, 5846 共1991兲. 26 L. Zhu, J. W. Bozzelli, and W.-P. Ho, J. Phys. Chem. A 103, 7800 共1999兲. 27 E. Lewars and I. Bonnycastle, J. Mol. Struct.: THEOCHEM 418, 17 共1997兲. 28 H. Hart, Chem. Rev. 79, 515 共1979兲. 29 F. Turecek and Z. Havlas, J. Org. Chem. 51, 4066 共1986兲. 30 R. Haag, J. Wirz, and P. J. Wagner, Helv. Chim. Acta 60, 2595 共1977兲. 31 K. Kirchner, D. Helf, P. Ott, and S. Vogt, Ber. Bunsenges. Phys. Chem. 94, 77 共1990兲. 32 C. J. Howard, J. Chem. Phys. 65, 4771 共1976兲. 33 J. H. Werner and T. A. Cool, Chem. Phys. Lett. 290, 81 共1998兲. 34 C. Sosa and H. B. Schlegel, J. Am. Chem. Soc. 109, 7007 共1987兲. 35 M. Kamphus, N.-N. Liu, B. Atakan, F. Qi, and A. Mcllroy, ‘‘REMPI Temperature Measurements in Molecular Beams Sampled from LowPressure Flames,’’ Twenty-Ninth Symposium (International) on Combustion 共The Combustion Institute, Pittsburgh, PA, 2003兲, p. 2627. 36 P. M. Guyon and J. Berkowitz, J. Chem. Phys. 54, 1814 共1971兲. 37 S. H. Lin, H. L. Selzle, K. O. Bo¨rnsen, and E. W. Schlag, J. Phys. Chem. 92, 1469 共1988兲. 38 W. A. Chupka, J. Chem. Phys. 98, 4520 共1993兲. 39 S. T. Pratt, J. Chem. Phys. 98, 9241 共1993兲. 40 A. Bhargava and P. R. Westmoreland, Combust. Flame 113, 333 共1998兲; 115, 456 共1998兲. 41 J. C. Robinson, N. Sveum, and D. M. Neumark 共unpublished兲. 23

Downloaded 12 Jan 2004 to 132.166.16.210. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp