Direct observation and kinetics of a hydroperoxyalkyl radical (QOOH)

See allHide authors and affiliations

Science  06 Feb 2015:
Vol. 347, Issue 6222, pp. 643-646
DOI: 10.1126/science.aaa1495

Catching a glimpse of the elusive QOOH

It's straightforward to write down the net combustion reaction: Oxygen reacts with hydrocarbons to form water and carbon dioxide. The details of how all the bonds break and form in succession are a great deal more complicated. Savee et al. now report direct detection of a long-postulated piece of the puzzle, a so-called QOOH intermediate. This structure results from bound oxygen stripping a hydrogen atom from carbon, leaving a carbon-centered radical behind. The study explores the influence of the hydrocarbon's unsaturation on the stability of QOOH, which has implications for both combustion and tropospheric oxidation chemistry.

Science, this issue p. 643


Oxidation of organic compounds in combustion and in Earth’s troposphere is mediated by reactive species formed by the addition of molecular oxygen (O2) to organic radicals. Among the most crucial and elusive of these intermediates are hydroperoxyalkyl radicals, often denoted “QOOH.” These species and their reactions with O2 are responsible for the radical chain branching that sustains autoignition and are implicated in tropospheric autoxidation that can form low-volatility, highly oxygenated organic aerosol precursors. We report direct observation and kinetics measurements of a QOOH intermediate in the oxidation of 1,3-cycloheptadiene, a molecule that offers insight into both resonance-stabilized and nonstabilized radical intermediates. The results establish that resonance stabilization dramatically changes QOOH reactivity and, hence, that oxidation of unsaturated organics can produce exceptionally long-lived QOOH intermediates.

Oxidation of organic compounds in such disparate processes as secondary organic aerosol (SOA) formation in Earth’s troposphere (13) and fuel autoignition in internal combustion engines (4) is governed by a surprisingly similar set of reactive intermediates. At the heart of low-temperature oxidation are chain reactions initiated by peroxy radicals, ROO (4, 5). These oxygen-centered radicals are formed when O2 adds to an organic radical, R, initially formed by either hydrogen abstraction from RH or addition of radicals (primarily OH) to π bonds between carbon atoms. It is well established that isomerization of ROO to a carbon-centered hydroperoxyalkyl radical (typically denoted as QOOH) can occur via intramolecular hydrogen abstraction, and in many cases this isomerization step profoundly influences the rate and effect of the oxidation reaction (4, 68).

Unimolecular decomposition of QOOH can produce a reactive OH radical and therefore plays an important role in radical chain propagation. More important, reaction of QOOH with O2 can form OOQOOH intermediates that may subsequently decompose, yielding two OH radicals. This mechanism is regarded as the most important radical chain-branching step in hydrocarbon oxidation below ~900 K (4). In combustion, understanding low-temperature chain branching is critical for improving efficiency through new engine designs in which ignition is controlled by fuel oxidation chemistry (autoignition) (9). In the tropospheric context, recent studies suggest that facile isomerization of ROO to QOOH in the oxidation of isoprene (2, 1012) and other biogenic hydrocarbons (1315) may participate both in the generation of low-volatility SOA precursors (1, 8, 16) and in HOx regeneration over remote (low-NOx) areas (11, 1720).

Despite this awareness of the importance of QOOH intermediates, they have eluded direct experimental detection. With one exception (21) studies of QOOH reactivity rely on measurements that necessarily involve complex assumptions about the reaction mechanism. Routine theoretical calculations of QOOH + O2 kinetics are hindered by the large size of these systems and the absence of a saddle point separating QOOH + O2 from OOQOOH, necessitating extensive and computationally demanding sampling of the potential energy surface (PES). Theoretical predictions of QOOH + O2 rate coefficients are thus scarce (2124), and direct experimental measurements combined with theory and modeling are critical for a detailed understanding of this class of chemical reactions.

The PESs for hydrocarbon oxidation reactions illustrate why direct observation of QOOH is fundamentally difficult. For reactions governing alkane oxidation, exemplified by the 2-butyl + O2 reaction shown in Fig. 1 (red path), QOOH intermediates generally lie higher in energy than ROO, with equilibrium heavily favoring the latter (25). Here, barriers to unimolecular decomposition of QOOH generally lie at or even below the reactant (R + O2) energy. At atmospheric conditions (i.e., 300 K and 1 atm), much of the flux from R + O2 is stabilized in the deep ROO well. In the low-temperature oxidation regime in combustion (i.e., 500 to 900 K), isomerization of ROO becomes rapid, resulting in substantial flux through QOOH. However, rapid decomposition, isomerization, and reaction with O2 typically remove QOOH faster than it is formed.

Fig. 1 Energy landscape for QOOH formation.

Computed 0 K stationary point energies relevant to the production of QOOH from R + O2 for R = 2-butyl (red, derived from n-butane) (25), 3-oxopent-2-yl (blue, derived from 3-pentanone) (27), and 3,5-cycloheptadienyl (gray, Rα) and 2,4-cycloheptadienyl (black, Rβ) (both derived from c-hpd and calculated in the present work). The different PESs are offset to match at the energy of QOOH. For clarity, slight barriers that may exist in R + O2 entrance channels are not shown.

Molecular structure can substantially affect QOOH reactivity. In particular, pathways in which R and/or QOOH exhibit resonance stabilization are common in oxidation of carbonyl-containing molecules and unsaturated hydrocarbons (8, 11, 26). For example, for 3-pentanone, a model system for autoxidation in the troposphere (8) and a potential biofuel (27), vinoxylic resonance stabilization in R has two significant effects on the R + O2 PES (Fig. 1, blue path) when compared with alkyl + O2 systems. First, resonance stabilization in R results in a smaller well depth for ROO, which cannot be resonance stabilized. Second, resonance stabilization of QOOH makes it nearly isoenergetic with ROO and increases barrier heights for unimolecular decomposition.

Here, we report direct detection of a resonance-stabilized QOOH intermediate formed during oxidation of the cyclic unsaturated hydrocarbon 1,3-cycloheptadiene (c-C7H10, c-hpd). As shown in Fig. 2, abstraction of one of the alkylic (α) or allylic (β) H atoms from c-hpd leads to two starting points on the R + O2 (c-C7H9 + O2) PES: the non–resonance stabilized Rα or the doubly allylic resonance-stabilized Rβ radicals. Although numerous pathways are possible from these two entry points [figs. S14 and S15], electronic structure calculations suggest that the kinetically most favored pathways from both entrance channels lead to the same doubly allylic, resonance-stabilized QOOH intermediate (2-hydroperoxy-4,6-cycloheptadienyl, QOOHαβ), as shown in Fig. 1 (black and gray paths). The double resonance stabilization of QOOHαβ lowers its energy below that of RαOO and RβOO. Our experiments produce both Rα and Rβ, enabling investigation of both nonstabilized and resonance-stabilized R in this system.

Fig. 2 Initial c-C7H9 radicals (Rα, Rβ) formed by H-abstraction from c-hpd.

We characterized QOOHαβ spectroscopically and performed direct kinetics measurements of its reaction with O2 at 400 K and 10 Torr using photoionization mass spectrometry with synchrotron-generated tunable vacuum ultraviolet ionizing radiation. This method, described in the supplementary materials (SM), provides simultaneous kinetic and spectroscopic identification of multiple species in a reacting gas mixture. The experiments are complemented by kinetic modeling using quantum chemical calculations and time-dependent master equation rate coefficient calculations. In experiments, Rα and Rβ were generated in roughly similar yields, based on predictions from structure-activity relationships [see (28) and the SM], via abstraction of H atoms from c-hpd by Cl radicals. The reactions are initiated by pulsed 351-nm laser photolysis of a gas mixture containing c-hpd, Cl2, varying amounts of O2 (in sufficient excess to ensure pseudo–first-order conditions), and He as a buffer gas. In the presence of O2, we observed a transient reaction product at mass-to-charge ratio (m/z) = 125.06, corresponding to the sum formula C7H9O2, consistent with ROO and/or QOOH (Fig. 3 inset and SM).

Fig. 3 Mass and photoionization spectra of C7H9O2.

The inset shows a c-hpd oxidation mass spectrum near m/z = 125, indicating the sum formula C7H9O2. The main figure presents superimposed photoionization spectra for C7H9O2 over two different time intervals, with 2-SD uncertainties derived from Poisson counting statistics for the 0- to 40- ms interval shown in gray. Above 8.5 eV, the spectrum depends on its observation time interval, suggesting the presence of at least two isomers consistent with QOOHαβ (AIE = 7.21 eV) and RαOO (8.70 eV) and/or RβOO (8.78 eV). Calculated AIEs are marked by arrows.

Different C7H9O2 isomers (e.g., ROO and QOOH) could be identified by their characteristic photoionization spectra. A photoionization spectrum of C7H9O2 (Fig. 3) shows a clear onset near 7.2 eV. This value is consistent with the calculated adiabatic ionization energy (AIE) of QOOHαβ (7.21 eV) but not with the calculated AIEs for the RαOO and RβOO isomers (8.70 and 8.78 eV, respectively). Photoionization spectra obtained at a variety of [O2], temperatures, and time intervals after photolysis have identical shapes in the 7.1- to 8.5-eV range, evidence that signal in this spectral region arises solely from QOOHαβ (SM). The scaled photoionization spectrum for C7H9O2 at a late reaction time (Fig. 3) shows increased ion signal above ∼8.5 eV, consistent with contributions from ROO species, which are less reactive than QOOH and therefore persist to longer times. Other C7H9O2 isomers exhibit considerable barriers to their formation or have AIEs inconsistent with the observed C7H9O2 signal onset. Quantum chemical calculations suggest that contributions from dissociative ionization of likely QOOHαβ + O2 products can also be excluded.

We also probed the time dependence of QOOHαβ as a function of oxygen concentration ([O2] = 1.9 × 1016 to 1.5 × 1017 cm−3) at a photoionization energy of 8.3 eV. Time-dependent QOOHαβ signals (Fig. 4A and fig. S17) are well fit by a double exponential kinetic model, yielding time constants for both the rise (τrise) and decay (τdecay) that decrease with increasing [O2]. Oxygen-centered ROO radicals are not expected to react with O2, and the presence of two time constants that depend on [O2] provides further evidence that C7H9O2 signal probed at 8.3 eV arises from the carbon-centered QOOHαβ radical. The observed linear dependence of 1/τrise and 1/τdecay on [O2] (Fig. 4B) yields second-order rate coefficients at 10 Torr and 400 K of krise = (2.9 ± 1.0) × 10−15 cm3 s−1 and kdecay = (3.2 ± 0.5) × 10−16 cm3 s−1 (uncertainties 2 SD precision). The y intercepts likely result from reactions with Cl2 molecules and do not affect rate coefficient determinations.

Fig. 4 QOOH kinetics.

(A) Time dependence of m/z = 125.06 signal obtained at 8.3 eV (i.e., probing only QOOHαβ) at several [O2] (open symbols) and the corresponding double exponential fits (solid lines). (B) Second-order plots showing the linearity of 1/τrise and 1/τdecay versus [O2] and accompanying linear fits [symbols and colors correspond to (A)]. Their slopes determine krise and kdecay, respectively, which we assign as rate coefficients for QOOHαβ + O2 → products and Rβ + O2 → QOOHαβ reactions, respectively.

However, detecting the intermediate species in sequential kinetic schemes is insufficient to assign the measured time constants (τrise or τdecay) to formation and depletion. To this end, in separate experiments at a radical concentration lower by a factor of 50, designed to minimize self- and cross-reactions of Rα and Rβ, we measured loss of c-C7H9 as a function of [O2] and independently extracted the rate coefficient for reaction of Rβ with O2, k(Rβ + O2) = (3.3 ± 0.5) × 10−16 cm3 s−1. The equivalence of this value with kdecay obtained from the QOOHαβ time profile is experimental evidence that kdecay is the rate coefficient for QOOHαβ formation (predominantly via Rβ + O2 reactions); therefore, krise describes QOOHαβ consumption via the reaction QOOHαβ + O2 → products.

Examining the underlying C7H9O2 PES and comparing experimental results to a kinetic model of this reaction system (SM) confirms this interpretation and affords additional mechanistic details. This model combines master-equation calculated rate coefficients of reactions on the C7H9O2 PES with kinetic parameters for R + R, RH + Cl, R + Cl2, and others taken from literature measurements of analogous reactions. Because the experiment provides time profiles and photoionization spectra for many species, comparison with the detailed model is a rigorous test of the experimental reaction system. Although we observe several [O2]-dependent signals that likely reflect products formed from QOOHαβ consumption, they cannot be definitively assigned.

The R + O2 PES shown in Fig. 1 and master equation rate coefficients reported in table S3 provide the necessary framework for understanding the observed QOOHαβ kinetics. Although the Rα + O2 ⇌ RαOO addition step is barrierless and the RαOO ⇌ QOOHαβ isomerization barrier is lower in energy than Rα + O2, kinetic modeling suggests that pathways from the more stable Rβ radical are the major source of QOOHαβ production (see fig. S21). The RαOO potential well is rather deep (–33.9 kcal mol−1 relative to Rα + O2) so that thermal equilibrium at 400 K strongly favors the stabilized RαOO species ([Rα]/[RαOO] ~ 10−9 at [O2] ~ 1017 cm−3 and 400 K) and effectively traps almost all Rα as RαOO. The kinetic model (SM) predicts that Rα + O2 generates only 5% of all QOOHαβ, with equal contributions from the chemically activated pathway Rα + O2 → QOOHαβ (i.e., without stabilization in the RαOO well) and from slow, tunneling-mediated isomerization of RαOO → QOOHαβ. Under our conditions, Rα + O2 → RαOO reaches equilibrium in <1 ms and does not affect measurement of the slower Rβ reactions.

By contrast, a small barrier (+1.4 kcal mol−1, not shown in Fig. 1) separates the resonance-stabilized Rβ radical from the shallow RβOO well (–13.9 kcal mol−1 relative to Rβ + O2), and equilibrium favors Rβ + O2 under our conditions ([Rβ]/[RβOO] ~ 102 at [O2] ~ 1017 cm−3 and 400 K). Moreover, the RβOO ⇌ QOOHαβ isomerization barrier is higher in energy than Rβ + O2. Nevertheless, the model predicts that Rβ pathways account for 95% of total QOOHαβ formation, primarily through two tunneling-mediated processes: isomerization of RβOO (66%) and chemically activated reaction directly from Rβ + O2 (29%). Based on master equation predictions for Rβ + O2 (table S3) and the large size of the QOOHαβ + O2 system, we anticipate that the present conditions (400 K and 10 Torr) probe kinetics near the high-pressure limit.

Only one direct experimental rate coefficient for any QOOH + O2 reaction exists in the literature (21), along with a few calculated rate coefficients (2124, 29), in all cases for saturated parent hydrocarbons. These rate coefficients are ~10−11 to ~10−12 cm3 s−1 near 400 K and in the high pressure limit (2123, 29), three to four orders of magnitude larger than the (2.9 ± 1.0) × 10−15 cm3 s−1 value determined here for QOOHαβ + O2. This dramatic reduction is likely due to the doubly allylic resonance stabilization of QOOHαβ. Comparing R + O2 rate coefficients for saturated hydrocarbon radicals to analogous systems where R has doubly allylic resonance stabilization (e.g., cyclohexyl versus 2,4-cyclohexadienyl and neo-pentyl versus 2,4-pentadienyl; see table S1) suggests that a decrease of ~103 in the rate coefficient for reaction with O2 is reasonable. Resonance-stabilized QOOH intermediates will occur in many important systems, such as the oxidation of biogenically derived hydrocarbons [e.g., isoprene (11, 12) and α/β-pinene (13, 14)]. This small rate coefficient for reaction with O2 implies a long lifetime of the resonance-stabilized QOOHαβ, suggesting that QOOH photochemistry and bimolecular reactions with minor species such as HOx and NOx may need to be considered. In the general case, both R and QOOH are carbon-centered radicals, and it is often assumed that these species exhibit similar reactivity toward O2 (30). However, we find that the rate coefficient for QOOHαβ + O2 is ~10 times as large as in the analogous Rβ + O2 reaction. This finding suggests reexamination of assumed equivalences between rate coefficients for R + O2 and QOOH + O2 reactions used in many organic oxidation models (31, 32).

In addition to a tunneling-mediated ROO → QOOH pathway (12), we have revealed a chemically activated mechanism as an important source of QOOH production in c-hpd oxidation with resonance-stabilized R and QOOH species. Although chemical activation is more prominent at low pressure, there is evidence for this type of mechanism in the troposphere (33). Current tropospheric models underpredict HOx regeneration by isoprene oxidation in pristine environments (18), and a recently proposed mechanism suggests that ROO ⇌ QOOH isomerization pathways are a substantial source of this missing HOx (11, 12). Like the c-hpd system investigated here, oxidation of isoprene proceeds through resonance-stabilized initial radicals (isoprene-OH adducts) that undergo peroxy chemistry, with the lowest-energy pathways leading to resonance-stabilized QOOH intermediates. The direct detection and kinetic determination provided here give experimental benchmarks for reactivity of resonance-stabilized QOOH and suggest that such radicals, including those proposed in formation of SOA precursors (8), may be relatively long-lived in the troposphere and in combustion systems.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S21

Tables S1 to S4

References (3451)

References and Notes

  1. Gaussian 09, Revision B.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria et al. (Gaussian, Inc., Wallingford, CT, 2009).
  2. MOLPRO, version 2012.1, H. J. Werner, P. J. Knowles, G. Knizia, F. R. Manby et al. (2012).
  3. Acknowledgments: Additional details of the experiments, calculations, and kinetic model are available in the supplementary materials. We thank H. Johnsen and the staff at the Chemical Dynamics Beamline of the Advanced Light Source (ALS) for technical support. This material is based on work supported by the U.S. Department of Energy (DOE), Office of Science, Office of Basic Energy Sciences (BES). Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the National Nuclear Security Administration under contract DE-AC04-94AL85000. This research used resources of the ALS, which is a DOE Office of Science User Facility. The ALS is supported by the Director, Office of Science, BES/DOE, under contract DE-AC02-05CH11231 between Lawrence Berkeley National Laboratory and the DOE.
View Abstract

Navigate This Article