J. Phys. Chem. A 2010, 114, 12437–12446
12437
A˜-X˜ Absorption of Propargyl Peroxy Radical (H-C≡C-CH2OO · ): A Cavity Ring-Down Spectroscopic and Computational Study Phillip S. Thomas, Neal D. Kline, and Terry A. Miller* Department of Chemistry, The Ohio State UniVersity, 120 West 18th AVenue, Columbus, Ohio 43210, United States ReceiVed: August 27, 2010; ReVised Manuscript ReceiVed: October 13, 2010
The A˜-X˜ electronic absorption spectrum of propargyl peroxy radical has been recorded at room temperature by cavity ring-down spectroscopy. Electronic structure calculations predict two isomeric forms, acetylenic and allenic, with two stable conformers for each. The acetylenic trans conformer, with a band origin at 7631.8 ( 0.1 cm-1, is definitively assigned on the basis of ab initio calculations and rotational simulations, and possible assignments for the acetylenic gauche and allenic trans forms are given. A fourth form, allenic cis, is not observed. Simulations based on calculated torsional potentials predict that the allenic trans form will have a long, poorly resolved progression in the OOCC torsional vibration, consistent with experimental observations. Introduction Hydrocarbon oxidation is a complex process which is central to the understanding of atmospheric1-3 and combustion4-6 chemistry. Oxidation mechanisms are exceedingly complex as they involve numerous transient chemical intermediates and reaction steps and are sensitive to temperature, pressure, and composition of the initial reaction mixtures. Organic peroxy radicals (RO2 · ) are an important class of transients that are formed early in many such mechanisms and are deterministic to reaction outcomes.5 In the atmosphere organic peroxy radicals are implicated in cycles with nitrogen oxides (NOx), leading to accumulation of tropospheric ozone;4,7 in combustion they are involved in the pivotal branching step, resulting in further oxidation ultimately to CO2 and breakdown of organic species, preventing buildup of larger organics and soot.2,8,9 The propargyl radical (C3H3) is an important intermediate in the combustion of hydrocarbons, particularly in a fuel-rich environment. Propargyl is a resonantly stabilized radical which can reach relatively high concentrations in flames.10-12 The bimolecular self-reaction of propargyl and cross-reactions with other resonance-stabilized free radicals13,14 lead to larger unsaturated hydrocarbons, benzene, and ultimately results in soot formation.6,15-18 The kinetic behavior for the C3H3 + C3H3 recombination has been measured for a variety of experimental conditions13 and modeled over wide temperature and pressure ranges via master equation simulations.11,13,19 At 295 K the highpressure limiting rate constants (k∞) for the C3H3 + C3H3 and C3H3 + O2 reactions have been measured at 4.3 × 10-11 and 2.3 × 10-13 cm3 molecule-1 s-1, respectively, indicating a strong preference for the propargyl self-reaction over reaction with O2 at low temperature.20 Propargyl radical has two resonance forms, · CH2-CtCH and H2CdCdCH · , which are predicted to have weights of 65% and 35%, respectively, according to computed spin density fractions on the · CH2- and dCH · carbon atoms, respectively.21 The radical is stabilized by 11 kcal/mol of “resonance energy”;21 as a result, it is more stable under combustion conditions than * To whom correspondence should be addressed. Phone: 614-292-2569. Fax: 614-292-1948. E-mail:
[email protected].
nonconjugated hydrocarbon radicals. The C3H3 + O2 f C3H3OO · reaction yields a peroxy adduct that is not resonance stabilized; as a consequence, the peroxy is more stable than reactants by only 18.9 kcal/mol.22 This value, equivalent to the R-O2 bond energy, is comparable with that of allyl peroxy (∼18 kcal/mol),23 another species derived from a resonance-stabilized hydrocarbon radical, but is much lower than those for peroxy radicals derived from nonresonance-stabilized alkyl radicals, such as methyl (30-35 kcal/mol),24-26 ethyl (35-36 kcal/mol),25-27 vinyl (40-46 kcal/mol),23,28,29 and ethynyl (∼49 kcal/mol)30 peroxies. Two cation states of propargyl peroxy are also reported to be stable.31 Addition of O2 to C3H3 occurs at either of two sites to yield acetylenic ( · OO-CH2-CtCH) or allenic (H2CdCdCH-OO · ) peroxy radical isomers,22 denoted ‘ace’ and ‘all’, respectively. In an early study by Slagle and Gutman22 the kinetics of the C3H3 + O2 reaction were measured over the temperature range 293-900 K by photoionization mass spectrometry. Timedependent yields of propargyl radical, formaldehyde, and ketene were monitored, the latter two species being the anticipated decomposition products of acetylenic and allenic propargyl peroxy isomers, respectively. Below 350 K only C3H3 depletion was observed, whereas reactant yields reached equilibrium over the range 380-430 K; these results were interpreted in terms of the reversible reaction C3H3 + O2 h C3H3O2 where the peroxy well lies 18.9 kcal/mol below reactants. Slagle and Gutman did not observe formaldehyde, indicating that acetylenic propargyl peroxy was not reacting further. Ketene was observed with increasing yields as the temperature was increased beyond 350 K, suggesting activation of a high-temperature mechanism involving decomposition of allenic propargyl peroxy.22 Hahn et al.32 performed a theoretical analysis of the formation and unimolecular decay channels of propargyl peroxy using a combination of electronic structure calculations and master equation simulations. The latter, based on potential energy surfaces calculated at the B3LYP/6-311+G(d,p) and approximate QCISD(T,full)/6-311++G(3df,2pd) levels, provided quantitative agreement with the experimental results of Slagle and Gutman22 and Atkinson and Hudgens20 when slight modifications were made to the surfaces. A important modification was
10.1021/jp108158a 2010 American Chemical Society Published on Web 11/05/2010
12438
J. Phys. Chem. A, Vol. 114, No. 47, 2010
that the barriers of the C3H3 + O2 f C3H3OO · addition were reduced from the calculated values of 3.7 and 7.1 kcal/mol (approximate QCISD(T)) to -0.2 and 3 kcal/mol for the acetylenic and allenic peroxy channels, respectively. In agreement with Slagle and Gutman, Hahn et al. concluded that at low temperatures the only reaction channel available to the acetylenic peroxy isomer is dissociation back to C3H3 + O2. The allenic peroxy isomer, which becomes accessible at T g 350 K, is able to undergo further intramolecular reactions via several channels involving 3-, 4-, or 5-membered ring intermediates, each of which Hahn et al. considered in detail. The 3-membered ring channel predominates up to 2000 K and leads to the ketene product32 observed by Slagle and Gutman. Dong et al.33 measured yields of vibrationally excited CO and CO2 produced from reaction of nascent propargyl radical with O2 via time-resolved FT-IR spectroscopy. Upon the basis of these results, in conjunction with electronic structure calculations, Dong et al. proposed two additional decomposition pathways of allenic propargyl peroxy. One of these occurs at intermediate temperature and leads to CH3CO + CO as products; the other occurs only at high temperatures and leads to CO2 + C2H3. An alternative to the high-temperature pathway of Dong et al. has recently been proposed.34 Jochnowitz et al.34 recently reported the first direct spectral observation of propargyl peroxy radical via a matrix-isolation experiment. Dilute mixtures of propargyl bromide or 1-butyn4-nitrite in Ar were passed through a hyperthermal nozzle to produce propargyl radical, and alternating layers of propargyl and O2 were deposited onto a spectroscopic window at 20 K. Subsequent annealing of the matrix yielded propargyl peroxy radical, which was characterized by FT-IR and linear dichroism spectroscopy. Only the trans conformer of acetylenic propargyl peroxy was observed, demonstrating that the formation of acetylenic propargyl peroxy is barrierless (or nearly so), as suggested previously by Hahn et al.32 In this work we report the observation of the ambient temperature A˜-X˜ electronic spectrum of propargyl peroxy radical via cavity ring-down spectroscopy (CRDS). This work is a continuation of recent spectroscopic studies by our group on A˜-X˜ transitions of organic peroxy radicals at room temperature,35-37 which also include the unsaturated phenyl38 and allyl39 peroxy transitions. The A˜-X˜ system is weak (σ ≈ 10-21 cm2)35 but has the advantage of being species selective unlike the much stronger B˜-X˜ transition. The latter occurs in the UV and has been used extensively in kinetics studies.7 As will be discussed shortly, the A˜-X˜ spectrum of propargyl peroxy will be presented along with an analysis of its vibrational and rotational structure. Experimental Section Our CRDS apparatus used for detection of NIR peroxy radical transitions is similar to the one described previously.40 A 20 Hz pulsed Nd:YAG (Spectra Physics, Quanta-Ray Pro 270) operating at 532 nm was used to pump a dye laser (Sirah, PrecisionScan), yielding 50-90 mJ/pulse over the wavelength range 643-586.5 nm. The laser dyes utilized were DCM, Rhodamine 101, and Rhodamine B (Exciton). Output from the dye laser was focused into a 70 cm single-pass Raman cell charged with 325-350 psi of H2. The desired second Stokes component (7235-8735 cm-1; 1-2 mJ/pulse) of the stimulated Raman scattered radiation was isolated via long-pass filters (Corion, LL-1000-F; Newport, 1000 nm LP) and focused into the ring-down cell. The ring-down cell is 55 cm in length and terminated by two highly reflective mirrors (Los Gatos Research;
Thomas et al. 1.3 µm, g99.995%; 1.2 µm, g99.995%). The inner 20 cm portion of the cell is outfitted with sample gas inlets on either end, a central vacuum port, and two rectangular UV-grade quartz photolysis windows. The cell has a depth of 15.2 cm along the direction of the photolysis beam (compared to 2.3 cm in an earlier version40) so that the quartz windows are spatially removed from the reactant flow zone. The ring-down mirrors and photolysis windows are protected from corrosion and soot deposits by flowing nitrogen over them at 10 and 250 cm3/min, respectively. The NIR light exiting the cell is detected by an amplified InGaAs photodiode (Thorlabs, PDA400), and the signal is recorded via a 12-bit digitizing card (Measurement Computing). Photolysis of precursor molecules was attained by excimer laser (Lambda-Physik, LPX120i) pulses of 120-150 mJ (ArF, 193 nm) or 240-300 mJ/pulse (KrF, 248 nm), which were focused into a rectangular shape 13 × 0.5 cm in order to maximize overlap with the NIR probe beam. Spectra were acquired by averaging 20-40 measurements at each wavelength, and instrumental step sizes were on the order of 0.1-0.5 cm-1. Spectral artifacts arising from water, precursor absorptions, and cavity structure were removed by subtracting a background trace (without photolysis) which was acquired concurrently with the trace resulting from photolysis. Spectra were calibrated via observed water or O2 signals using frequencies reported in the HITRAN 2004 database.41 Uncertainties in absolute frequencies are estimated at (2 cm-1; band positions are quoted to within (10 cm-1 due to possible overlap of multiple transitions. A more accurate calibration was obtained over 7600-7710 cm-1 using a set of 42 water lines lying in this region, from which an interpolated calibration curve was constructed. This accounted for small instrumental nonlinearities arising from the dye laser grating drive and allowed us to achieve an absolute precision of (0.1 cm-1. Laser operation and data acquisition were achieved by PC-based Labview software. Photolysis of propargyl chloride, propargyl bromide, or propyne was used to generate the propargyl radical; O2 adds to this species in a 3-body collision involving the carrier gas to produce the peroxy. When propargyl chloride (Aldrich, 98%) was used as a precursor, a stream of N2 with a backing pressure of 15-25 psi(g) was seeded through the neat liquid and 11-13 Torr of the resultant mixture was mixed with O2 prior to entering the ring-down cell. A typical gas mixture in the cell consisted of [N2] ) 40-55 Torr (including purge), [O2] ) 35-40 Torr, and [HCtCCH2-Cl] ≈ 1 Torr. For blow-up scans of the 7550-7710 cm-1 region, higher signal-to-noise was obtained by using gas mixtures containing 80-90/78/∼1.5-2 Torr N2/ O2/C3H3Cl. When propargyl bromide (Aldrich, 80% in toluene) was used, the gas mixture contained 77/79/∼1 Torr N2/O2/ C3H3Br, prepared in a manner similar to that used for propargyl chloride. When propyne (Aldrich, 98%) was used, a gas mixture containing 76/87/5.0 Torr N2/O2/propyne (including purge N2) was used in the 7594-7754 cm-1 range to maximize the signal of the sharp propargyl peroxy bands. Since propyne absorbs strongly below 7600 cm-1 at our level of sensitivity, we used a mixture containing 32/38/2.1 Torr N2/O2/propyne (e.g., similar gas ratios but lower total pressure) for the spectrum obtained over the range 7411-7637 cm-1. The trace in Figure 3b was obtained by scaling the intensities of the high-pressure data in order to overlap the 7594-7637 cm-1 portion, which was covered in both scans. Propargyl chloride and propyne were photolyzed using 193 nm (ArF) excimer pulses; propargyl bromide was photolyzed at 248 nm (KrF). For all sets of conditions an excimer-probe pulse delay of 1 µs was found to give the optimal peroxy signal.
A˜-X˜ Absorption of Propargyl Peroxy Radical
J. Phys. Chem. A, Vol. 114, No. 47, 2010 12439
TABLE 1: Calculated Degeneracies (g), X˜ and A˜ State Energies, A˜-X˜ T00 Values (cm-1), Boltzmann Weights (w), Oscillator Strengths (f), Isomer Fractions (fxn), and Relative Intensities (I) for Isomers and Conformers of Propargyl Peroxya
ace-T ace-G all-Tf all-Gg all-C
g
X˜b
1 2 1 2 1
0 123 0 523
UB3LYP A˜ T00 7833 7589 8608 8052 7755
7833 7466 8608h 8052 7232
G2
UCIS
exp
w
X˜c
A˜
T00
w
105 f
fxnd
I
T00
1.00 1.10 1.00
0 107 0
7636 7480 7576h
7636 7373 7576g
1.00 1.19 1.00
1.94 0.68 4.16
0.65 0.65 0.35
1.26 0.53 1.46
7631.8e 7381 (?) 7482h (?)
0.08
374
7236
6862
0.16
3.64
0.35
0.21
All energies are ZPVE corrected. For each isomer the X˜ state energy of the T conformer is chosen to be the zero of energy (but see notes below). Relative intensities are given as products of Boltzmann weights (G2), oscillator strengths, and isomer fractions. b X˜ state relative energies (UB3LYP, cm-1): ace-T, +2269; ace-G, +2392; all-T, 0; all-C, +523. c X˜ state relative energies (G2, cm-1): ace-T, 0; ace-G, +107; all-T, +120; all-C, +494. d Isomer fraction based on statistical addition of O2 to propargyl radical based upon computed spin densities.21 e This value is blue shifted 0.3 cm-1 relative to the observed transition to account for the rotational contour. See text for details. f Minimum in X˜ state only; reported A˜ state energies are for the transition-state structure. g Minimum in A˜ state only. h Frequency (Tmax) of most intense transition. a
Quantum Chemical Calculations Equilibrium geometries and harmonic and fundamental vibrational frequencies were computed at the UB3LYP/cc-pVDZ level of theory. Relative energies, Boltzmann weights, and T00 values were calculated at the G2 level. For A˜ state calculations using B3LYP or G2, the appropriate electron configuration was obtained by permuting the HOMO and SOMO orbitals in an initial ROHF step; the converged A˜ state electronic wave function was used as an initial guess for a single-point calculation at the UHF level; the converged UHF wave function was then used as an initial guess for subsequent calculations at the G2 or B3LYP levels. A˜-X˜ electronic transition moments were computed at the UCIS/6-31G(d) level using UMP2/631G(d) equilibrium geometries obtained within the G2 routine. Potential curve scans were performed at the UB3LYP/cc-pVDZ level (X˜ state) or TD-UB3LYP/cc-pVDZ level (A˜ state) by varying the OOCC dihedral angle in 5° intervals and relaxing all remaining degrees of freedom at each step. Electronic structure calculations were performed using Gaussian 03W42 or 09.43 Rotational simulations utilized rotational constants derived from UMP2/6-31G(d) equilibrium geometries and UCIS/631G(d) transition moment components and were performed at 298 K using our SpecView software. Franck-Condon (FC) simulations were performed using the MolFC package,44 which calculates Franck-Condon factors in the harmonic approximation with full inclusion of Duschinsky rotation; these used UB3LYP/cc-pVDZ harmonic frequencies (scaled ×0.970) and normal modes as input. Excitations are considered from the vibrationless level of the X˜ state to A˜ state levels with ν ) 0, 1, or 2 for any vibration, including combination levels. Since the Franck-Condon simulations only consider cold (0 K) absorption, sequence band (SB)36 profiles were calculated based upon differences in computed X˜ and A˜ state unscaled UB3LYP/ccpVDZ harmonic frequencies, where the weight of each transition is governed solely by a Boltzmann factor for the X˜ state level. All sequence transitions out of levels for which the Boltzmann weight is g0.1% that of the vibrationless level are included, including combination levels. Nonsequence hot transitions are expected to be weak and are not included in the simulations. The highly anharmonic OOCC torsion vibration was not included in the FC or SB simulations but was considered in a separate calculation. Here, the time-independent Schro¨dinger equation was solved numerically for the computed torsional potentials using a basis of 73 plane waves; the resultant energy levels and wave functions are used to calculate Boltzmann weights and Franck-Condon factors for all possible torsional transitions within the range of the basis set. For the energy level
calculations we used a single value of the reduced mass for each surface, chosen to fit the vibrationless levels on that surface to computed harmonic values (UB3LYP/cc-pVDZ, unscaled). For the acetylenic isomer, the vibrationless levels of both conformers were fit in order to minimize the root-mean-squared error; for the allenic isomer only the vibrationless levels of the T conformer (X˜ state) and G conformer (A˜ state) were fit to computed values. Results Computational Results. Table 1 presents a summary of computed relative energies, A˜-X˜ band origin frequencies, Boltzmann weights, oscillator strengths, and predicted relative intensities for the isomers and conformers of propargyl peroxy. In agreement with previous calculations, we find that groundstate propargyl peroxy possesses acetylenic and allenic isomers,32,33 each with two conformers34 related by the OOCC torsion (Figure 1). The all-T species is unusual in that the X˜ state equilibrium geometry does not correspond to a minimum on the A˜ state surface but rather a transition state separating two gauche minima with ∠OOCC ≈ (128° (UB3LYP/cc-pVDZ level). We were unable to obtain the G2 energy for the A˜ state all-G structure due to failure of the MP2 geometry search to locate a minimum geometry; this failure persisted even when the Hessian was computed at every step of the optimization. In Table 1 the relative energies of X˜ state structures of conformers of a given isomer show good agreement between the two levels of theory. However, when comparing different isomers the B3LYP level predicts that the allenic isomer is the more stable by ca. 2000 cm-1 while G2 places the acetylenic isomer slightly lower in energy (footnotes b and c in Table 1). In accordance with previous work, we expect that G2 values are much more accurate than those computed at the B3LYP level, both for isomer relative energies32 and for A˜-X˜ transition frequencies,35 the latter of which are systematically overestimated at the B3LYP level. Indeed, it would be very difficult to reconcile the experimental observations (vide infra) with the results from the B3LYP calculations. Computed vibrational frequencies for stationary points on the propargyl peroxy potential energy surface are shown in Table 2. The OOCC torsion (ω18) is frequently observed in organic peroxy radical spectra35 where the energy level structure of this mode has important consequences for the band shape in the electronic origin region.39,45 We therefore performed potential energy surface scans along the OOCC dihedral angle for both isomers of propargyl peroxy (Figure 2). The potential curves of the acetylenic isomer bear some resemblance to those calculated for methyl45 and allyl39 peroxies,
12440
J. Phys. Chem. A, Vol. 114, No. 47, 2010
Thomas et al.
Figure 1. Isomers and conformers of propargyl peroxy. For the acetylenic isomer, T/G labels correspond to OOCC dihedral angles of 180°/(80°, respectively; for the allenic isomer, C/T/G labels correspond to OOCC dihedral angles of 0°/180°/(128°, respectively. Each gauche conformer exists as mirror image pairs.
TABLE 2: Unscaled UB3LYP/cc-pVDZ Computed Frequencies (cm-1) for the Conformers of Propargyl Peroxya ace-T harm.
c
harm.
all-C
c
harm.
all-T c
all-G
X˜
A˜
X˜
A˜
X˜
A˜
X˜
A˜
X˜
harm. A˜
45 166 318 427 487 604 675 867 949 989 1161 1162 1310 1416 2203 2935 2940 3330
80 164 327 344 493 602 677 919 1005 972 968 1181 1342 1467 2206 2865 2888 3331
62 168 326 431 500 659 717 917 968 1000 1182 1197 1342 1443 2242 3052 3106 3478
96 170 331 348 503 652 714 941 1028 988 1007 1205 1362 1477 2244 3006 3047 3478
70 201 313 421 586 656 710 843 970 1021 1147 1252 1339 1429 2238 3062 3123 3478
118 210 323 349 573 656 710 888 955 1010 999 1251 1354 1433 2234 3029 3104 3478
157 160 310 382 593 737 847 901 908 999 1113 1168 1345 1453 2050 3114 3199 3210
187 144 271 335 597 687 840 900 920 1017 1006 1181 1341 1471 2082 3103 3184 3206
97 176 312 465 564 594 859 912 993 1001 1120 1169 1315 1454 2060 3113 3198 3201
135i 173 248 354 569 582 825 919 1044 1015 1166 945 1340 1468 2081 3102 3183 3204
fund. ω18 ω17 ω16 ω15 ω14 ω13 ω12 ω11 ω10 ω9 ω8 ω7 ω6 ω5 ω4 ω3 ω2 ω1
ace-G
b
c
A˜ 85 198 355 296 571 639 855 902 993 1011 1158 906 1327 1452 2060 3117 3202 3163
a The OOCC torsion vibration corresponds to ω18 for all species. b Computed frequency of the 0-1 vibrational transition with inclusion of anharmonicity. c Computed frequency in the harmonic approximation.
with barriers of 100-400 and 900-1200 cm-1 separating the conformers in the X˜ and A˜ states, respectively. The groundstate barriers are so low that the T and G conformers are only predicted to have four and two bound levels, respectively, before the “quasi-free-rotor” region is reached. X˜ state levels immediately above these barriers, a number of which will be populated at room temperature, will have the best Franck-Condon factors for excitations to levels near the tops of the corresponding barriers on the A˜ state surface; these transitions will appear several hundred wavenumbers to the blue of the ace-T and ace-G band origins and with moderate intensity. The scans for the allenic isomer show large barriers separating the C conformer from the T/G conformer in both electronic states. In the X˜ state, the T conformer carries the bulk of the Boltzmann population at room temperature. In accord with the imaginary frequency computed for the T conformer in the A˜ state (Table 2), the allenic A˜ state OOCC potential exhibits a transition state at the trans geometry 330 cm-1 above the all-G minima. As a consequence, we expect to observe a long torsional progression arising from excitations out of the vibrationless level of the T conformer, where the most intense transitions terminate at levels near the top of the A˜ state barrier. In addition to these,
there are five hot levels of the T conformer with Boltzmann weights g 15% that of the vibrationless level; progressions out of these levels will also contribute to the spectrum. Although the all-T/all-G spectrum is expected to be quite complicated, the majority of the transitions become unresolvable once a rotational profile is added (vide infra). Cavity Ring-Down Spectroscopy of Propargyl Peroxy. We generated propargyl peroxy from three different precursors, from propargyl chloride and propyne photolyzed at 193 nm, and from propargyl bromide photolyzed at 248 nm, in all cases followed by reaction with O2. Figure 3 shows the spectra in the 7250-7750 cm-1 region where the most intense transitions of the three strongest conformers of propargyl peroxy are expected (Table 1). For all three precursors we observe a series of sharp lines in the 7600-7700 cm-1 region superimposed on what first appears to be a broad background absorption. The intensity of this “background” is somewhat variable, but it is present in all of our scans regardless of precursor and typically appears with intensities 40-60% that of the strongest sharp line. The spectrum resulting from propyne photolysis exhibits strong interferences from methyl peroxy (000, 7383 cm-1 (truncated);
A˜-X˜ Absorption of Propargyl Peroxy Radical
J. Phys. Chem. A, Vol. 114, No. 47, 2010 12441
Figure 2. B3LYP potential energy surfaces of propargyl peroxy radical calculated as a function of the OOCC dihedral angle, along with calculated OOCC torsion vibrational levels. The A˜ state potentials have been shifted in frequency by a constant in order to match the G2 prediction for the ace-T conformer.
Figure 3. Spectra of the origin region of propargyl peroxy, obtained via photolysis of (a) propargyl chloride, (b) propyne, and (c) propargyl bromide (inset), in all cases followed by O2 addition. The left and right vertical axes correspond to traces a and b, respectively. Traces b and c in the inset have been vertically offset for clarity.
Figure 4. Chemical tests performed on the propargyl peroxy ace-T conformer origin. In the upper six traces signal intensity is shown as a function of excimer-probe delay time. In the bottom trace the requirement of O2 for peroxy formation is demonstrated.
1211 band, 7488 cm-1) which do not appear in the spectra obtained from the halide precursors. Soot buildup on the photolysis windows presented a serious problem for all of the propargyl peroxy experiments. We attempted to remedy this by adding an N2 purge to the photolysis windows and by lengthening our cell along the photolysis beam direction so that the windows are spatially removed from the probe region. These modifications were partially successful, allowing us to obtain clean scans from propargyl chloride and propyne but not for propargyl bromide, for which soot buildup is much more rapid. Even short scans with this latter precursor (e.g., Figure 3c) exhibited notably worse signal to noise than those from the other precursors. Since the propargyl chloride photolysis spectrum is free of methyl peroxy interferences and has the best signal to noise, the remainder of our discussion will focus on spectra generated from this precursor. However, as Figure 3 shows, important spectral features attributed to C3H3O2 are present in traces from all three precursors. Figure 4 shows the results of chemical tests performed on the features observed in the 7580-7760 cm-1 region. In the
first test the peroxy decay rate is monitored by measuring the signal intensity as a function of excimer-probe delay time. In Figure 4 we observe that the sharp features and the “background” decay at different rates, with the sharp lines disappearing within a few hundred microseconds, while the “background” persists for as long as 2 ms. In the second test O2 is removed from the reaction mixture; the result is that both the sharp signals and the background vanish completely. In the top portion of Figure 5 we present the full A˜-X˜ CRDS spectrum of propargyl peroxy from 193 nm photolysis of propargyl chloride in the presence of O2. From this survey spectrum we note that the broad absorption described previously as “background” appears to be part of a series of broad bands with widths of ∼120 cm-1. This spectrum is a difference trace (“with photolysis” minus “without photolysis”), where the negative absorptions in the 7200-7400 and 8500-8800 cm-1 regions correspond to propargyl chloride bands depleted by photolysis. The authentic propargyl chloride spectrum, shown for comparison in the bottom of Figure 5, was obtained by subtracting empty-cavity mirror absorption profiles from the same “without photolysis” traces used to construct the peroxy
12442
J. Phys. Chem. A, Vol. 114, No. 47, 2010
Figure 5. Survey spectrum of propargyl peroxy (top), obtained via 193 nm photolysis of propargyl chloride in the presence of O2, and authentic spectra of HO2 (center) and propargyl chloride (bottom). The spectrum of HO2 was obtained by 193 nm photolysis of a gas mixture containing 121.7/23.5/0.3 Torr N2/O2/vinyl-Br.39
spectrum at the top of Figure 5. Presumably the propargyl chloride features correspond to ground-state overtone and combination bands which, although weak, absorb strongly at our level of sensitivity. HO2 radical is also evident in the propargyl peroxy spectrum, where its A˜-X˜ 301 band46 appears in the 7800-8200 cm-1 region as a large number of sharp lines of low-to-moderate intensity. We observed HO2 as a photolysis product in mixtures of O2 with alkyl halides and allyl and vinyl bromides.39 Our experiment does not allow us to distinguish whether HO2 is generated from H atoms produced directly by propargyl chloride photolysis or as a later product of a peroxy unimolecular decay channel. Discussion Before attempting a detailed assignment of the propargyl peroxy spectrum, let us consider some of our experimental observations more closely. The appearance of soot from photolysis of all three precursors, but especially from propargyl bromide, is consistent with previous observations.20 We note that soot is not the source of either broad or sharp spectral features since both require O2. This is also consistent with the possibility that a peroxy radical may be the carrier for both types of signals. However, the difference of lifetimes for the two transition types requires that at least two carriers be present. Previous lifetime measurements using experimental conditions similar to ours have yielded values ranging from 1 to 10 ms for saturated organic peroxy radicals35 to tens of microseconds for acetyl peroxy.47 The lifetime of the broad features is similar to that of a saturated peroxy radical, while that of the sharp structure is significantly shorter but not unprecedented. Peroxy radical decay under conditions typical of our experiments is governed primarily by the rate of bimolecular self-reaction;35 rate constants for these reactions depend strongly on the molecular structure of the peroxy radical. Isomers of butyl peroxy, for instance, possess self-reaction rate constants which differ by as much as an order of magnitude.40 All of these observations are consistent with formation of two different isomers of propargyl peroxy. One might question whether we should even expect to observe both isomers. By generating acetylenic propargyl peroxy under matrix isolation conditions, Jochnowitz et al. showed that the formation of this isomer from propargyl + O2 is barrierless. In
Thomas et al. contrast, the allenic peroxy isomer, which was not observed by Jochnowitz et al.,34 has a barrier to formation which Hahn et al. estimated at ∼3 kcal/mol32 based upon the agreement of their theoretical model with the high-temperature results of Slagle and Gutman.22 Since only 0.6 kcal/mol of thermal energy is available at the temperature of our experiments (298 K) we cannot expect to efficiently surmount the barrier connecting reactants to the allenic peroxy well, nor did Slagle and Gutman, who observed products of the allenic isomer decomposition only at T > 350 K. However, the allenic peroxy isomer may be formed in our experiments before the nascent propargyl radical has reached thermal equilibrium. Fan and Pratt reported that propargyl radicals generated from 193 nm photodissociation of propargyl chloride are vibrationally excited by ∼18 kcal/mol at the peak of the observed translational energy probability distribution curve.48 Under our experimental conditions a propargyl radical generated by 193 nm photolysis of C3H3Cl will undergo ∼900 collisions within 1 µs following the excimer pulse, approximately 40% of which will involve O2. We therefore expect that both peroxy radical isomers may be formed, and based upon the potential diagrams of previous authors,32,33 the low-temperature reaction pathways of allenic propargyl peroxy33 will also be accessible. None of the species produced along any of these channels (CO, ketene, CH3CO, and HCO) possesses an electronic transition in our spectral region; however, secondary dissociation of HCO f H + CO may very well provide a source of H atoms which gives rise to the observed HO2 signals. We now turn to results of electronic structure calculations in order to assign the experimental spectrum. As described previously, computed rotational constants and electronic transition moment components (Table 3), vibrational frequencies, and T00 values were used to generate a series of simulations; these are shown in comparison with the origin region of the experimental spectrum in Figure 6. In Figure 6a the pure rotational profiles for three of the conformers of propargyl peroxy are shown, shifted in frequency to the G2-predicted values of the band origins and intensity weighted by calculated relative intensities. (The fourth conformer, all-C, is expected to be weak and outside of our scanning range). For the simulations of the all-T conformer we used the G2 relative energy and MP2 rotational constants for the transition-state structure on the A˜ state surface rather than for the all-G minimum (which we were unable to obtain); the simulated rotational profile therefore corresponds to the transition closest to the top of the barrier, which would be most Franck-Condon allowed, rather than to the band origin. The simulated rotational profiles of the ace and all isomers exhibit striking differences. The ace-T conformer is expected to have distinct, resolved rotational structure, while the all-T conformer is predicted to exhibit a broad profile with no distinguishing characteristics. In Figure 6b vibrational structure has been added to the simulations. For each conformer we convolved the simulated SB structure with progressions from the Franck-Condon factor calculations and with the calculated torsional transitions; the resulting series of transitions are plotted using the rotational profiles shown in Figure 6a. Simulated torsional transitions, which are derived from potential energy curves containing multiple conformer minima (Figure 2), were each given a conformer designation according to the most probable position of the calculated X˜ state eigenfunction. It was necessary to use fundamental rather than harmonic frequencies (Table 2) for the SB simulation of the ace-T conformer in order to correctly reproduce the sequence in ω17, which appears to the red of the
A˜-X˜ Absorption of Propargyl Peroxy Radical
J. Phys. Chem. A, Vol. 114, No. 47, 2010 12443
TABLE 3: Computed Rotational Constants (cm-1) and Transition Dipole Moment Ratios (see footnotes) Used for Rotational Simulations of Propargyl Peroxy Isomers and Conformersa X˜ state
A˜ state
A
B
C
A
B
C
ace- T ace- Gc all- Tb,d all- Cb
1.105 0.4743 1.426 0.4702
0.08468 0.1084 0.08199 0.1178
0.07985 0.09490 0.07878 0.09603
1.110 0.4391 1.413 0.4524
0.08450 0.1126 0.07907 0.1181
0.07972 0.09623 0.07604 0.09550
vibrationless level ω18, V ) 1 ω18, V ) 2
1.086 1.050 1.001
0.08390 0.08402 0.08411
0.07914 0.07942 0.07967
b
individual vibrational levels of the ace-T conformer 0.08401 0.07924 1.094 0.08419 0.07962 1.059 0.08441 0.08002 1.010
a Rotational constants were computed at the UMP2/6-31G(d) equilibrium geometries except for those of individual vibrational levels of the ace-T conformer, which were manually adjusted to provide better agreement with experimental rotational contours. These empirically adjusted values are given at the bottom of the table and are used for the simulations in Figure 7b and 7c. TDM ratios were evaluated at the UCIS/ 6-31G(d) level using UMP2/6-31G(d) equilibrium geometries and have been normalized to unity. b Relative TDM ratios: 0/0/1 a/b/c-type. c Relative TDM ratios: 0.071/0.449/0.480 a/b/c-type. d Planar transition-state geometry in the A˜ state.
Figure 6. Comparison of simulations with the experimental spectrum of propargyl peroxy in the origin region: (a) pure rotational profiles; (b) rotational profiles convolved with sequence band, Franck-Condon, and OOCC torsional structure; (c) sum of simulations in b, where the ace-T and all-T simulations have been shifted in frequency by -4 and -80 cm-1, respectively; (d) experimental spectrum. Simulations in a and b have been shifted in frequency to match the G2 band origins except for the all-T conformer (see text for details). All simulated intensities are scaled by I values from Table 1.
Figure 7. Comparison of simulations with the experimental spectrum of propargyl peroxy in the 7550-7710 cm-1 blowup region: (a) composite simulation from Figure 6c; (b) optimized rotational profiles for band origin (blue), 1811 (red), and 1822 (violet) transitions, empirically shifted in frequency and weighted in intensity to match individual experimental bands; (c) sum of simulations in b, where additional sequence bands have been added; (d) experimental spectrum. Assignments for vibrational transitions of the ace-T conformer of propargyl peroxy are shown above the experimental spectrum.
origin experimentally but is predicted to the blue of the origin by the harmonic calculation. The relative intensities of the simulations shown in Figure 6 correspond to the I values in Table 1 which are given as products of the Boltzmann weight of the conformer, the A˜-X˜ oscillator strength, and the isomer fraction. The latter quantity was calculated assuming that the isomers are not in thermal equilibrium with one another but are formed in proportion to the computed spin density fractions21 at C(1) and C(3) of propargyl radical where O2 is trapped. Alternatively, if the isomers are assumed to be in thermal equilibrium, two modifications must be made to the I values of the conformers of the allenic isomer. First, the Boltzmann factors are multiplied by 0.56/1.00 ) 0.56 (all-T) and 0.09/0.16 ) 0.56 (all-C) since these species lie +120 and +494 cm-1 relative to the ace-T conformer, respectively (G2 level); second, the allenic “isomer fraction” is scaled by 0.65/0.35 ) 1.86 since the two isomers are allowed to reach thermal equilibrium regardless of the proportion in which they are initially formed. These two factors multiplied together give 0.56 × 1.86 ) 1.04, so the I values of the allenic species are nearly the same whether or not the isomers are considered to be formed in equilibrium. Figure
6c shows a “composite” simulation which is a weighted sum of the simulations for all four species, where the ace-T and all-T simulations have been red shifted by 4 and 80 cm-1, respectively, in order to best fit the experimental spectrum. Although the shift used for the all-T conformer simulation is larger than one might expect for an electronic transition energy based on G2 calculations, in this case the transitions with the largest Franck-Condon factors terminate in the vicinity of the barrier in the A˜ state, where the energy level structure is less well defined. The overall agreement between experiment and simulation in Figure 6c is very good, and most of the spectral features can be assigned from the comparison. The ace-T conformer is assigned unambiguously as the carrier of the sharp features in the 7600-7700 cm-1 region; a list of transitions and their assignments is given in Table 4. Assignments are listed according to a prolate symmetric top convention49 as the ace-T conformer is nearly prolate (a > b ≈ c). We will first discuss the spectrum of the ace-T conformer in detail before considering the remainder of the spectrum.
12444
J. Phys. Chem. A, Vol. 114, No. 47, 2010
Thomas et al.
TABLE 4: Observed Transitions and Peak Spacings (cm-1) and Assignments for Propargyl Peroxya band
spacing
7555 7606.1 7607.9 7609.6 7611.4 7613.3 7615.0 7617.0 7618.8 7620.8 7622.7 7624.7 7626.6 7629.0 7630.5 7631.4 7634.4 7636.7 7638.9 7641.1 7641.5 7643.1 7645.1 7647.4 7649.5
-76 -25.3 -23.5 -21.8 -20.1 -18.1 -16.4 -14.4 -12.6 -10.7 -8.7 -6.8 -4.8 -2.5 -0.9 0 3.0 5.3 7.5 9.6 10.0 11.6 13.7 16.0 18.1
A B C D E F G H
7381 7482 7807 7931 8032 8156 8269 8440
assignment 1511 000 000 000 000 000 000 000 000 000 000 000 000 1722 1711 000 000 000 000 000 1611 000 000 000 000
band
p(14)
Q p(13) Q p(12) Q p(11) Q p(10) Q p(9) Q p(8) Q p(7) Q p(6) Q p(5) Q p(4) Q p(3) Q p(1)
Q, R r(2) Q r(3) Q r(4) Q p(1)
r(5)
Q r(6) Q r(7) Q r(8) Q
r(0)
Q
7650.6 7651.9 7653.9 7656.4 7657.9 7660.7 7667.8 7669.5 7671.7 7673.6 7675.6 7677.4 7679.6 7681.0 7682.4 7683.9 7686.0 7688.5 7691.0 7693.6 7694.5 7699.7 7701.4 7703.3 7704.8
spacing ace-T bands 19.1 20.4 22.5 25.0 26.5 29.3 36.4 38.1 40.3 42.2 44.2 46.0 48.2 49.6 51.0 52.5 54.5 57.0 59.5 62.1 63.1 68.3 69.9 71.9 73.4
assignment 000 000 000 000 000
r
1811 1811 1811 1811 1811 1811 1811 1811 1811 1811 1811 1811 1811 1811 1811 1811 1811 1811 1811
Q Q r Q r Q r Q r
16111811
1822 1822 1822 1822 1822 1822 1822 1822 1822 1822 1822 1822
p
Q Q p Q p Q p Q p Q p
p
Q Q p Q p Q p Q p
p(1)
Q, Q r(4) Q r(5) Q r(6) Q r(7) Q r(8) Q r(9) Q r Q r Q r Q r Q r Q r Q
r(0)
Q
r(3)
p(2)
Q Q, r(2) Q r(3) Q r(4) Q r(5) Q p(1)
r(0)
Q 16111822
other bands tentatively assigned to C3H3O2 +175 +299 +400 +524 +888 +922
ace-G 000 all-T OOCC Tmax ace-T OOCC torsion progression ace-T OOCC torsion progression ace-T OOCC torsion progression ace-T OOCC torsion progression ace-G 810 all-T OOCC Tmax + 710
a Multiple assignments are given for bands in regions where rotational contours of different vibrational bands interfere. Rotational assignments are listed according to the prolate symmetric top notation ∆Ka(K′′)∆J (J′′) where ∆K ) -1, 0, 1 are represented by p, q, and r and ∆J ) -1, 0, 1 are represented by P, Q, and R, respectively; here, transitions from individual J′′ levels are not resolved experimentally, so only the dominant ∆J structure is listed.
An expansion of the 7550-7710 cm-1 region is presented in Figure 7, where a detailed comparison of the “composite” simulation (Figure 7a) and the experimental spectrum (Figure 7d) can be made. The simulated spectrum qualitatively accounts for a number of the experimental features. First, the ace-T spectrum consists of the band origin rotational contour and several sequence bands which are identifiable by their strong band heads. Sequences in ω18-ω15 can be tentatively assigned based upon differences in the computed fundamental frequencies listed in Table 2. The lines in the 7600-7626 cm-1 range are assigned unambiguously to the pQ branch of the band origin as there are no intense sequence bands predicted in this region. The 000 rQ branch, however, overlaps sequences in ω18 and ω16, the latter of which appears much more strongly experimentally than in the simulation. In order to provide more detailed assignments in the regions where multiple rotational contours overlap, we performed another simulation in which we manually adjusted the frequencies and intensities of several copies of the ace-T simulated rotational profile. In this simulation, profiles for only eight bands were included: 000, 1811, 1822, 1711, 1722, 1611, 16111811, and 16111822. The frequency adjustments were minor: all but two of the bands (1722, shifted +4.8 cm-1; 16111822, shifted -3.3 cm-1) were shifted by